{ "best_metric": null, "best_model_checkpoint": null, "epoch": 3.0, "eval_steps": 500, "global_step": 9966, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 6.688963210702341e-07, "loss": 2.7717, "step": 1 }, { "epoch": 0.0, "learning_rate": 1.3377926421404683e-06, "loss": 2.7806, "step": 2 }, { "epoch": 0.0, "learning_rate": 2.0066889632107025e-06, "loss": 2.7754, "step": 3 }, { "epoch": 0.0, "learning_rate": 2.6755852842809365e-06, "loss": 2.7326, "step": 4 }, { "epoch": 0.0, "learning_rate": 3.3444816053511705e-06, "loss": 2.7518, "step": 5 }, { "epoch": 0.0, "learning_rate": 4.013377926421405e-06, "loss": 2.6833, "step": 6 }, { "epoch": 0.0, "learning_rate": 4.682274247491639e-06, "loss": 2.7282, "step": 7 }, { "epoch": 0.0, "learning_rate": 5.351170568561873e-06, "loss": 2.6084, "step": 8 }, { "epoch": 0.0, "learning_rate": 6.0200668896321075e-06, "loss": 2.6692, "step": 9 }, { "epoch": 0.0, "learning_rate": 6.688963210702341e-06, "loss": 2.7197, "step": 10 }, { "epoch": 0.0, "learning_rate": 7.357859531772576e-06, "loss": 2.7415, "step": 11 }, { "epoch": 0.0, "learning_rate": 8.02675585284281e-06, "loss": 2.6479, "step": 12 }, { "epoch": 0.0, "learning_rate": 8.695652173913044e-06, "loss": 2.6248, "step": 13 }, { "epoch": 0.0, "learning_rate": 9.364548494983277e-06, "loss": 2.7256, "step": 14 }, { "epoch": 0.0, "learning_rate": 1.0033444816053512e-05, "loss": 2.6833, "step": 15 }, { "epoch": 0.0, "learning_rate": 1.0702341137123746e-05, "loss": 2.6209, "step": 16 }, { "epoch": 0.01, "learning_rate": 1.137123745819398e-05, "loss": 2.6836, "step": 17 }, { "epoch": 0.01, "learning_rate": 1.2040133779264215e-05, "loss": 2.6521, "step": 18 }, { "epoch": 0.01, "learning_rate": 1.270903010033445e-05, "loss": 2.6308, "step": 19 }, { "epoch": 0.01, "learning_rate": 1.3377926421404682e-05, "loss": 2.569, "step": 20 }, { "epoch": 0.01, "learning_rate": 1.4046822742474917e-05, "loss": 2.604, "step": 21 }, { "epoch": 0.01, "learning_rate": 1.4715719063545153e-05, "loss": 2.5002, "step": 22 }, { "epoch": 0.01, "learning_rate": 1.5384615384615387e-05, "loss": 2.5499, "step": 23 }, { "epoch": 0.01, "learning_rate": 1.605351170568562e-05, "loss": 2.4954, "step": 24 }, { "epoch": 0.01, "learning_rate": 1.6722408026755853e-05, "loss": 2.4983, "step": 25 }, { "epoch": 0.01, "learning_rate": 1.739130434782609e-05, "loss": 2.4204, "step": 26 }, { "epoch": 0.01, "learning_rate": 1.806020066889632e-05, "loss": 2.5112, "step": 27 }, { "epoch": 0.01, "learning_rate": 1.8729096989966554e-05, "loss": 2.5218, "step": 28 }, { "epoch": 0.01, "learning_rate": 1.939799331103679e-05, "loss": 2.4341, "step": 29 }, { "epoch": 0.01, "learning_rate": 2.0066889632107023e-05, "loss": 2.4564, "step": 30 }, { "epoch": 0.01, "learning_rate": 2.073578595317726e-05, "loss": 2.4212, "step": 31 }, { "epoch": 0.01, "learning_rate": 2.1404682274247492e-05, "loss": 2.4082, "step": 32 }, { "epoch": 0.01, "learning_rate": 2.2073578595317725e-05, "loss": 2.3593, "step": 33 }, { "epoch": 0.01, "learning_rate": 2.274247491638796e-05, "loss": 2.4087, "step": 34 }, { "epoch": 0.01, "learning_rate": 2.3411371237458197e-05, "loss": 2.3329, "step": 35 }, { "epoch": 0.01, "learning_rate": 2.408026755852843e-05, "loss": 2.3117, "step": 36 }, { "epoch": 0.01, "learning_rate": 2.4749163879598663e-05, "loss": 2.2906, "step": 37 }, { "epoch": 0.01, "learning_rate": 2.54180602006689e-05, "loss": 2.3773, "step": 38 }, { "epoch": 0.01, "learning_rate": 2.608695652173913e-05, "loss": 2.3387, "step": 39 }, { "epoch": 0.01, "learning_rate": 2.6755852842809364e-05, "loss": 2.2443, "step": 40 }, { "epoch": 0.01, "learning_rate": 2.7424749163879597e-05, "loss": 2.2235, "step": 41 }, { "epoch": 0.01, "learning_rate": 2.8093645484949833e-05, "loss": 2.2124, "step": 42 }, { "epoch": 0.01, "learning_rate": 2.8762541806020066e-05, "loss": 2.1678, "step": 43 }, { "epoch": 0.01, "learning_rate": 2.9431438127090305e-05, "loss": 2.2736, "step": 44 }, { "epoch": 0.01, "learning_rate": 3.0100334448160538e-05, "loss": 2.1292, "step": 45 }, { "epoch": 0.01, "learning_rate": 3.0769230769230774e-05, "loss": 2.1611, "step": 46 }, { "epoch": 0.01, "learning_rate": 3.1438127090301004e-05, "loss": 2.1232, "step": 47 }, { "epoch": 0.01, "learning_rate": 3.210702341137124e-05, "loss": 2.1983, "step": 48 }, { "epoch": 0.01, "learning_rate": 3.2775919732441476e-05, "loss": 2.2164, "step": 49 }, { "epoch": 0.02, "learning_rate": 3.3444816053511705e-05, "loss": 2.181, "step": 50 }, { "epoch": 0.02, "learning_rate": 3.411371237458194e-05, "loss": 2.1533, "step": 51 }, { "epoch": 0.02, "learning_rate": 3.478260869565218e-05, "loss": 2.1247, "step": 52 }, { "epoch": 0.02, "learning_rate": 3.545150501672241e-05, "loss": 2.154, "step": 53 }, { "epoch": 0.02, "learning_rate": 3.612040133779264e-05, "loss": 2.1309, "step": 54 }, { "epoch": 0.02, "learning_rate": 3.678929765886287e-05, "loss": 2.0479, "step": 55 }, { "epoch": 0.02, "learning_rate": 3.745819397993311e-05, "loss": 2.1387, "step": 56 }, { "epoch": 0.02, "learning_rate": 3.812709030100335e-05, "loss": 2.0913, "step": 57 }, { "epoch": 0.02, "learning_rate": 3.879598662207358e-05, "loss": 2.1295, "step": 58 }, { "epoch": 0.02, "learning_rate": 3.946488294314382e-05, "loss": 2.0104, "step": 59 }, { "epoch": 0.02, "learning_rate": 4.0133779264214046e-05, "loss": 2.1353, "step": 60 }, { "epoch": 0.02, "learning_rate": 4.080267558528428e-05, "loss": 2.0791, "step": 61 }, { "epoch": 0.02, "learning_rate": 4.147157190635452e-05, "loss": 2.1027, "step": 62 }, { "epoch": 0.02, "learning_rate": 4.214046822742475e-05, "loss": 2.0683, "step": 63 }, { "epoch": 0.02, "learning_rate": 4.2809364548494984e-05, "loss": 2.0497, "step": 64 }, { "epoch": 0.02, "learning_rate": 4.347826086956522e-05, "loss": 1.9567, "step": 65 }, { "epoch": 0.02, "learning_rate": 4.414715719063545e-05, "loss": 2.1161, "step": 66 }, { "epoch": 0.02, "learning_rate": 4.4816053511705686e-05, "loss": 2.1021, "step": 67 }, { "epoch": 0.02, "learning_rate": 4.548494983277592e-05, "loss": 2.0267, "step": 68 }, { "epoch": 0.02, "learning_rate": 4.615384615384616e-05, "loss": 2.2327, "step": 69 }, { "epoch": 0.02, "learning_rate": 4.6822742474916394e-05, "loss": 1.9866, "step": 70 }, { "epoch": 0.02, "learning_rate": 4.7491638795986624e-05, "loss": 2.1607, "step": 71 }, { "epoch": 0.02, "learning_rate": 4.816053511705686e-05, "loss": 2.1618, "step": 72 }, { "epoch": 0.02, "learning_rate": 4.8829431438127096e-05, "loss": 2.0607, "step": 73 }, { "epoch": 0.02, "learning_rate": 4.9498327759197325e-05, "loss": 2.1151, "step": 74 }, { "epoch": 0.02, "learning_rate": 5.016722408026756e-05, "loss": 2.1111, "step": 75 }, { "epoch": 0.02, "learning_rate": 5.08361204013378e-05, "loss": 2.072, "step": 76 }, { "epoch": 0.02, "learning_rate": 5.150501672240803e-05, "loss": 2.1212, "step": 77 }, { "epoch": 0.02, "learning_rate": 5.217391304347826e-05, "loss": 2.0498, "step": 78 }, { "epoch": 0.02, "learning_rate": 5.284280936454849e-05, "loss": 1.9944, "step": 79 }, { "epoch": 0.02, "learning_rate": 5.351170568561873e-05, "loss": 1.9643, "step": 80 }, { "epoch": 0.02, "learning_rate": 5.4180602006688965e-05, "loss": 1.9677, "step": 81 }, { "epoch": 0.02, "learning_rate": 5.4849498327759194e-05, "loss": 2.0538, "step": 82 }, { "epoch": 0.02, "learning_rate": 5.551839464882943e-05, "loss": 2.0376, "step": 83 }, { "epoch": 0.03, "learning_rate": 5.6187290969899666e-05, "loss": 1.9257, "step": 84 }, { "epoch": 0.03, "learning_rate": 5.6856187290969896e-05, "loss": 1.9896, "step": 85 }, { "epoch": 0.03, "learning_rate": 5.752508361204013e-05, "loss": 2.0393, "step": 86 }, { "epoch": 0.03, "learning_rate": 5.819397993311037e-05, "loss": 2.0825, "step": 87 }, { "epoch": 0.03, "learning_rate": 5.886287625418061e-05, "loss": 2.0168, "step": 88 }, { "epoch": 0.03, "learning_rate": 5.953177257525085e-05, "loss": 2.0621, "step": 89 }, { "epoch": 0.03, "learning_rate": 6.0200668896321076e-05, "loss": 2.0505, "step": 90 }, { "epoch": 0.03, "learning_rate": 6.086956521739131e-05, "loss": 2.1334, "step": 91 }, { "epoch": 0.03, "learning_rate": 6.153846153846155e-05, "loss": 2.0687, "step": 92 }, { "epoch": 0.03, "learning_rate": 6.220735785953178e-05, "loss": 1.9548, "step": 93 }, { "epoch": 0.03, "learning_rate": 6.287625418060201e-05, "loss": 2.0641, "step": 94 }, { "epoch": 0.03, "learning_rate": 6.354515050167224e-05, "loss": 2.0589, "step": 95 }, { "epoch": 0.03, "learning_rate": 6.421404682274248e-05, "loss": 2.0658, "step": 96 }, { "epoch": 0.03, "learning_rate": 6.488294314381272e-05, "loss": 1.9783, "step": 97 }, { "epoch": 0.03, "learning_rate": 6.555183946488295e-05, "loss": 2.094, "step": 98 }, { "epoch": 0.03, "learning_rate": 6.622073578595317e-05, "loss": 2.0401, "step": 99 }, { "epoch": 0.03, "learning_rate": 6.688963210702341e-05, "loss": 2.0209, "step": 100 }, { "epoch": 0.03, "learning_rate": 6.755852842809365e-05, "loss": 1.9129, "step": 101 }, { "epoch": 0.03, "learning_rate": 6.822742474916388e-05, "loss": 1.9835, "step": 102 }, { "epoch": 0.03, "learning_rate": 6.889632107023412e-05, "loss": 2.0763, "step": 103 }, { "epoch": 0.03, "learning_rate": 6.956521739130436e-05, "loss": 2.0257, "step": 104 }, { "epoch": 0.03, "learning_rate": 7.023411371237458e-05, "loss": 1.9921, "step": 105 }, { "epoch": 0.03, "learning_rate": 7.090301003344481e-05, "loss": 2.0306, "step": 106 }, { "epoch": 0.03, "learning_rate": 7.157190635451505e-05, "loss": 2.1127, "step": 107 }, { "epoch": 0.03, "learning_rate": 7.224080267558529e-05, "loss": 1.9955, "step": 108 }, { "epoch": 0.03, "learning_rate": 7.290969899665552e-05, "loss": 1.9717, "step": 109 }, { "epoch": 0.03, "learning_rate": 7.357859531772575e-05, "loss": 1.8824, "step": 110 }, { "epoch": 0.03, "learning_rate": 7.424749163879598e-05, "loss": 1.9742, "step": 111 }, { "epoch": 0.03, "learning_rate": 7.491638795986622e-05, "loss": 1.968, "step": 112 }, { "epoch": 0.03, "learning_rate": 7.558528428093647e-05, "loss": 1.9704, "step": 113 }, { "epoch": 0.03, "learning_rate": 7.62541806020067e-05, "loss": 1.9441, "step": 114 }, { "epoch": 0.03, "learning_rate": 7.692307692307693e-05, "loss": 2.0722, "step": 115 }, { "epoch": 0.03, "learning_rate": 7.759197324414716e-05, "loss": 1.9734, "step": 116 }, { "epoch": 0.04, "learning_rate": 7.82608695652174e-05, "loss": 2.0511, "step": 117 }, { "epoch": 0.04, "learning_rate": 7.892976588628763e-05, "loss": 1.9815, "step": 118 }, { "epoch": 0.04, "learning_rate": 7.959866220735787e-05, "loss": 2.1036, "step": 119 }, { "epoch": 0.04, "learning_rate": 8.026755852842809e-05, "loss": 1.9156, "step": 120 }, { "epoch": 0.04, "learning_rate": 8.093645484949833e-05, "loss": 2.0231, "step": 121 }, { "epoch": 0.04, "learning_rate": 8.160535117056857e-05, "loss": 2.0662, "step": 122 }, { "epoch": 0.04, "learning_rate": 8.22742474916388e-05, "loss": 2.031, "step": 123 }, { "epoch": 0.04, "learning_rate": 8.294314381270904e-05, "loss": 2.1019, "step": 124 }, { "epoch": 0.04, "learning_rate": 8.361204013377927e-05, "loss": 1.9204, "step": 125 }, { "epoch": 0.04, "learning_rate": 8.42809364548495e-05, "loss": 2.0206, "step": 126 }, { "epoch": 0.04, "learning_rate": 8.494983277591973e-05, "loss": 2.0153, "step": 127 }, { "epoch": 0.04, "learning_rate": 8.561872909698997e-05, "loss": 2.0249, "step": 128 }, { "epoch": 0.04, "learning_rate": 8.62876254180602e-05, "loss": 2.0431, "step": 129 }, { "epoch": 0.04, "learning_rate": 8.695652173913044e-05, "loss": 1.9397, "step": 130 }, { "epoch": 0.04, "learning_rate": 8.762541806020068e-05, "loss": 1.9419, "step": 131 }, { "epoch": 0.04, "learning_rate": 8.82943143812709e-05, "loss": 1.9919, "step": 132 }, { "epoch": 0.04, "learning_rate": 8.896321070234114e-05, "loss": 2.004, "step": 133 }, { "epoch": 0.04, "learning_rate": 8.963210702341137e-05, "loss": 1.9698, "step": 134 }, { "epoch": 0.04, "learning_rate": 9.030100334448161e-05, "loss": 1.9663, "step": 135 }, { "epoch": 0.04, "learning_rate": 9.096989966555184e-05, "loss": 1.8822, "step": 136 }, { "epoch": 0.04, "learning_rate": 9.163879598662207e-05, "loss": 1.9337, "step": 137 }, { "epoch": 0.04, "learning_rate": 9.230769230769232e-05, "loss": 2.0501, "step": 138 }, { "epoch": 0.04, "learning_rate": 9.297658862876255e-05, "loss": 2.0527, "step": 139 }, { "epoch": 0.04, "learning_rate": 9.364548494983279e-05, "loss": 1.948, "step": 140 }, { "epoch": 0.04, "learning_rate": 9.431438127090302e-05, "loss": 1.9762, "step": 141 }, { "epoch": 0.04, "learning_rate": 9.498327759197325e-05, "loss": 1.9445, "step": 142 }, { "epoch": 0.04, "learning_rate": 9.565217391304348e-05, "loss": 1.9609, "step": 143 }, { "epoch": 0.04, "learning_rate": 9.632107023411372e-05, "loss": 1.9105, "step": 144 }, { "epoch": 0.04, "learning_rate": 9.698996655518396e-05, "loss": 1.9513, "step": 145 }, { "epoch": 0.04, "learning_rate": 9.765886287625419e-05, "loss": 1.931, "step": 146 }, { "epoch": 0.04, "learning_rate": 9.832775919732441e-05, "loss": 1.8839, "step": 147 }, { "epoch": 0.04, "learning_rate": 9.899665551839465e-05, "loss": 2.0032, "step": 148 }, { "epoch": 0.04, "learning_rate": 9.966555183946489e-05, "loss": 2.0684, "step": 149 }, { "epoch": 0.05, "learning_rate": 0.00010033444816053512, "loss": 1.9002, "step": 150 }, { "epoch": 0.05, "learning_rate": 0.00010100334448160536, "loss": 1.9617, "step": 151 }, { "epoch": 0.05, "learning_rate": 0.0001016722408026756, "loss": 1.9398, "step": 152 }, { "epoch": 0.05, "learning_rate": 0.00010234113712374582, "loss": 1.8857, "step": 153 }, { "epoch": 0.05, "learning_rate": 0.00010301003344481605, "loss": 1.8337, "step": 154 }, { "epoch": 0.05, "learning_rate": 0.00010367892976588629, "loss": 2.0902, "step": 155 }, { "epoch": 0.05, "learning_rate": 0.00010434782608695653, "loss": 2.0209, "step": 156 }, { "epoch": 0.05, "learning_rate": 0.00010501672240802676, "loss": 2.0389, "step": 157 }, { "epoch": 0.05, "learning_rate": 0.00010568561872909698, "loss": 1.9384, "step": 158 }, { "epoch": 0.05, "learning_rate": 0.00010635451505016722, "loss": 1.9734, "step": 159 }, { "epoch": 0.05, "learning_rate": 0.00010702341137123746, "loss": 1.976, "step": 160 }, { "epoch": 0.05, "learning_rate": 0.0001076923076923077, "loss": 1.9876, "step": 161 }, { "epoch": 0.05, "learning_rate": 0.00010836120401337793, "loss": 1.9856, "step": 162 }, { "epoch": 0.05, "learning_rate": 0.00010903010033444817, "loss": 1.9676, "step": 163 }, { "epoch": 0.05, "learning_rate": 0.00010969899665551839, "loss": 2.0002, "step": 164 }, { "epoch": 0.05, "learning_rate": 0.00011036789297658862, "loss": 1.9554, "step": 165 }, { "epoch": 0.05, "learning_rate": 0.00011103678929765886, "loss": 1.9225, "step": 166 }, { "epoch": 0.05, "learning_rate": 0.0001117056856187291, "loss": 1.9847, "step": 167 }, { "epoch": 0.05, "learning_rate": 0.00011237458193979933, "loss": 1.95, "step": 168 }, { "epoch": 0.05, "learning_rate": 0.00011304347826086956, "loss": 1.9535, "step": 169 }, { "epoch": 0.05, "learning_rate": 0.00011371237458193979, "loss": 1.9343, "step": 170 }, { "epoch": 0.05, "learning_rate": 0.00011438127090301003, "loss": 1.9272, "step": 171 }, { "epoch": 0.05, "learning_rate": 0.00011505016722408026, "loss": 1.8708, "step": 172 }, { "epoch": 0.05, "learning_rate": 0.0001157190635451505, "loss": 1.9839, "step": 173 }, { "epoch": 0.05, "learning_rate": 0.00011638795986622074, "loss": 1.9027, "step": 174 }, { "epoch": 0.05, "learning_rate": 0.00011705685618729099, "loss": 1.9005, "step": 175 }, { "epoch": 0.05, "learning_rate": 0.00011772575250836122, "loss": 1.8832, "step": 176 }, { "epoch": 0.05, "learning_rate": 0.00011839464882943146, "loss": 1.9423, "step": 177 }, { "epoch": 0.05, "learning_rate": 0.0001190635451505017, "loss": 1.9057, "step": 178 }, { "epoch": 0.05, "learning_rate": 0.00011973244147157192, "loss": 1.855, "step": 179 }, { "epoch": 0.05, "learning_rate": 0.00012040133779264215, "loss": 1.874, "step": 180 }, { "epoch": 0.05, "learning_rate": 0.00012107023411371239, "loss": 1.9553, "step": 181 }, { "epoch": 0.05, "learning_rate": 0.00012173913043478263, "loss": 1.9491, "step": 182 }, { "epoch": 0.06, "learning_rate": 0.00012240802675585286, "loss": 1.9554, "step": 183 }, { "epoch": 0.06, "learning_rate": 0.0001230769230769231, "loss": 1.8921, "step": 184 }, { "epoch": 0.06, "learning_rate": 0.00012374581939799333, "loss": 1.9264, "step": 185 }, { "epoch": 0.06, "learning_rate": 0.00012441471571906357, "loss": 2.0067, "step": 186 }, { "epoch": 0.06, "learning_rate": 0.00012508361204013378, "loss": 1.8599, "step": 187 }, { "epoch": 0.06, "learning_rate": 0.00012575250836120401, "loss": 1.8353, "step": 188 }, { "epoch": 0.06, "learning_rate": 0.00012642140468227425, "loss": 1.858, "step": 189 }, { "epoch": 0.06, "learning_rate": 0.0001270903010033445, "loss": 2.1525, "step": 190 }, { "epoch": 0.06, "learning_rate": 0.00012775919732441472, "loss": 1.9037, "step": 191 }, { "epoch": 0.06, "learning_rate": 0.00012842809364548496, "loss": 1.8671, "step": 192 }, { "epoch": 0.06, "learning_rate": 0.0001290969899665552, "loss": 1.9515, "step": 193 }, { "epoch": 0.06, "learning_rate": 0.00012976588628762543, "loss": 1.874, "step": 194 }, { "epoch": 0.06, "learning_rate": 0.00013043478260869567, "loss": 1.9059, "step": 195 }, { "epoch": 0.06, "learning_rate": 0.0001311036789297659, "loss": 1.9727, "step": 196 }, { "epoch": 0.06, "learning_rate": 0.00013177257525083614, "loss": 1.9204, "step": 197 }, { "epoch": 0.06, "learning_rate": 0.00013244147157190635, "loss": 1.8656, "step": 198 }, { "epoch": 0.06, "learning_rate": 0.00013311036789297659, "loss": 1.9663, "step": 199 }, { "epoch": 0.06, "learning_rate": 0.00013377926421404682, "loss": 1.9245, "step": 200 }, { "epoch": 0.06, "learning_rate": 0.00013444816053511706, "loss": 1.8957, "step": 201 }, { "epoch": 0.06, "learning_rate": 0.0001351170568561873, "loss": 1.9099, "step": 202 }, { "epoch": 0.06, "learning_rate": 0.00013578595317725753, "loss": 1.8986, "step": 203 }, { "epoch": 0.06, "learning_rate": 0.00013645484949832777, "loss": 1.867, "step": 204 }, { "epoch": 0.06, "learning_rate": 0.000137123745819398, "loss": 1.9533, "step": 205 }, { "epoch": 0.06, "learning_rate": 0.00013779264214046824, "loss": 1.83, "step": 206 }, { "epoch": 0.06, "learning_rate": 0.00013846153846153847, "loss": 1.8684, "step": 207 }, { "epoch": 0.06, "learning_rate": 0.0001391304347826087, "loss": 1.9158, "step": 208 }, { "epoch": 0.06, "learning_rate": 0.00013979933110367892, "loss": 1.9332, "step": 209 }, { "epoch": 0.06, "learning_rate": 0.00014046822742474916, "loss": 2.0097, "step": 210 }, { "epoch": 0.06, "learning_rate": 0.0001411371237458194, "loss": 1.8898, "step": 211 }, { "epoch": 0.06, "learning_rate": 0.00014180602006688963, "loss": 1.9054, "step": 212 }, { "epoch": 0.06, "learning_rate": 0.00014247491638795986, "loss": 1.8904, "step": 213 }, { "epoch": 0.06, "learning_rate": 0.0001431438127090301, "loss": 1.9489, "step": 214 }, { "epoch": 0.06, "learning_rate": 0.00014381270903010034, "loss": 1.8308, "step": 215 }, { "epoch": 0.07, "learning_rate": 0.00014448160535117057, "loss": 1.9254, "step": 216 }, { "epoch": 0.07, "learning_rate": 0.0001451505016722408, "loss": 1.8675, "step": 217 }, { "epoch": 0.07, "learning_rate": 0.00014581939799331104, "loss": 1.772, "step": 218 }, { "epoch": 0.07, "learning_rate": 0.00014648829431438128, "loss": 1.8421, "step": 219 }, { "epoch": 0.07, "learning_rate": 0.0001471571906354515, "loss": 2.1112, "step": 220 }, { "epoch": 0.07, "learning_rate": 0.00014782608695652173, "loss": 1.9413, "step": 221 }, { "epoch": 0.07, "learning_rate": 0.00014849498327759196, "loss": 1.9224, "step": 222 }, { "epoch": 0.07, "learning_rate": 0.0001491638795986622, "loss": 1.8588, "step": 223 }, { "epoch": 0.07, "learning_rate": 0.00014983277591973243, "loss": 2.1073, "step": 224 }, { "epoch": 0.07, "learning_rate": 0.0001505016722408027, "loss": 1.9704, "step": 225 }, { "epoch": 0.07, "learning_rate": 0.00015117056856187293, "loss": 2.0127, "step": 226 }, { "epoch": 0.07, "learning_rate": 0.00015183946488294317, "loss": 1.9209, "step": 227 }, { "epoch": 0.07, "learning_rate": 0.0001525083612040134, "loss": 1.8604, "step": 228 }, { "epoch": 0.07, "learning_rate": 0.00015317725752508362, "loss": 1.8494, "step": 229 }, { "epoch": 0.07, "learning_rate": 0.00015384615384615385, "loss": 1.8621, "step": 230 }, { "epoch": 0.07, "learning_rate": 0.0001545150501672241, "loss": 1.8487, "step": 231 }, { "epoch": 0.07, "learning_rate": 0.00015518394648829432, "loss": 1.7957, "step": 232 }, { "epoch": 0.07, "learning_rate": 0.00015585284280936456, "loss": 1.9955, "step": 233 }, { "epoch": 0.07, "learning_rate": 0.0001565217391304348, "loss": 1.884, "step": 234 }, { "epoch": 0.07, "learning_rate": 0.00015719063545150503, "loss": 2.0258, "step": 235 }, { "epoch": 0.07, "learning_rate": 0.00015785953177257527, "loss": 1.8765, "step": 236 }, { "epoch": 0.07, "learning_rate": 0.0001585284280936455, "loss": 1.9803, "step": 237 }, { "epoch": 0.07, "learning_rate": 0.00015919732441471574, "loss": 1.8401, "step": 238 }, { "epoch": 0.07, "learning_rate": 0.00015986622073578598, "loss": 1.998, "step": 239 }, { "epoch": 0.07, "learning_rate": 0.00016053511705685619, "loss": 1.9732, "step": 240 }, { "epoch": 0.07, "learning_rate": 0.00016120401337792642, "loss": 1.9129, "step": 241 }, { "epoch": 0.07, "learning_rate": 0.00016187290969899666, "loss": 1.8746, "step": 242 }, { "epoch": 0.07, "learning_rate": 0.0001625418060200669, "loss": 1.8504, "step": 243 }, { "epoch": 0.07, "learning_rate": 0.00016321070234113713, "loss": 1.8989, "step": 244 }, { "epoch": 0.07, "learning_rate": 0.00016387959866220737, "loss": 1.8309, "step": 245 }, { "epoch": 0.07, "learning_rate": 0.0001645484949832776, "loss": 1.7838, "step": 246 }, { "epoch": 0.07, "learning_rate": 0.00016521739130434784, "loss": 1.9508, "step": 247 }, { "epoch": 0.07, "learning_rate": 0.00016588628762541807, "loss": 1.9011, "step": 248 }, { "epoch": 0.07, "learning_rate": 0.0001665551839464883, "loss": 1.8541, "step": 249 }, { "epoch": 0.08, "learning_rate": 0.00016722408026755855, "loss": 1.8806, "step": 250 }, { "epoch": 0.08, "learning_rate": 0.00016789297658862878, "loss": 1.8439, "step": 251 }, { "epoch": 0.08, "learning_rate": 0.000168561872909699, "loss": 1.8945, "step": 252 }, { "epoch": 0.08, "learning_rate": 0.00016923076923076923, "loss": 1.9686, "step": 253 }, { "epoch": 0.08, "learning_rate": 0.00016989966555183946, "loss": 1.773, "step": 254 }, { "epoch": 0.08, "learning_rate": 0.0001705685618729097, "loss": 1.7553, "step": 255 }, { "epoch": 0.08, "learning_rate": 0.00017123745819397994, "loss": 1.8907, "step": 256 }, { "epoch": 0.08, "learning_rate": 0.00017190635451505017, "loss": 1.8701, "step": 257 }, { "epoch": 0.08, "learning_rate": 0.0001725752508361204, "loss": 1.9593, "step": 258 }, { "epoch": 0.08, "learning_rate": 0.00017324414715719065, "loss": 2.1013, "step": 259 }, { "epoch": 0.08, "learning_rate": 0.00017391304347826088, "loss": 1.8868, "step": 260 }, { "epoch": 0.08, "learning_rate": 0.00017458193979933112, "loss": 1.7625, "step": 261 }, { "epoch": 0.08, "learning_rate": 0.00017525083612040135, "loss": 1.7528, "step": 262 }, { "epoch": 0.08, "learning_rate": 0.00017591973244147156, "loss": 1.8266, "step": 263 }, { "epoch": 0.08, "learning_rate": 0.0001765886287625418, "loss": 1.8106, "step": 264 }, { "epoch": 0.08, "learning_rate": 0.00017725752508361204, "loss": 1.8959, "step": 265 }, { "epoch": 0.08, "learning_rate": 0.00017792642140468227, "loss": 1.9129, "step": 266 }, { "epoch": 0.08, "learning_rate": 0.0001785953177257525, "loss": 1.8969, "step": 267 }, { "epoch": 0.08, "learning_rate": 0.00017926421404682274, "loss": 1.9479, "step": 268 }, { "epoch": 0.08, "learning_rate": 0.00017993311036789298, "loss": 1.9501, "step": 269 }, { "epoch": 0.08, "learning_rate": 0.00018060200668896322, "loss": 1.9245, "step": 270 }, { "epoch": 0.08, "learning_rate": 0.00018127090301003345, "loss": 1.8242, "step": 271 }, { "epoch": 0.08, "learning_rate": 0.0001819397993311037, "loss": 1.8555, "step": 272 }, { "epoch": 0.08, "learning_rate": 0.00018260869565217392, "loss": 1.7998, "step": 273 }, { "epoch": 0.08, "learning_rate": 0.00018327759197324413, "loss": 1.8935, "step": 274 }, { "epoch": 0.08, "learning_rate": 0.0001839464882943144, "loss": 1.9184, "step": 275 }, { "epoch": 0.08, "learning_rate": 0.00018461538461538463, "loss": 2.0166, "step": 276 }, { "epoch": 0.08, "learning_rate": 0.00018528428093645487, "loss": 1.8483, "step": 277 }, { "epoch": 0.08, "learning_rate": 0.0001859531772575251, "loss": 1.8835, "step": 278 }, { "epoch": 0.08, "learning_rate": 0.00018662207357859534, "loss": 1.8247, "step": 279 }, { "epoch": 0.08, "learning_rate": 0.00018729096989966558, "loss": 1.7858, "step": 280 }, { "epoch": 0.08, "learning_rate": 0.0001879598662207358, "loss": 1.8788, "step": 281 }, { "epoch": 0.08, "learning_rate": 0.00018862876254180605, "loss": 1.9421, "step": 282 }, { "epoch": 0.09, "learning_rate": 0.00018929765886287626, "loss": 1.9172, "step": 283 }, { "epoch": 0.09, "learning_rate": 0.0001899665551839465, "loss": 1.9021, "step": 284 }, { "epoch": 0.09, "learning_rate": 0.00019063545150501673, "loss": 1.7992, "step": 285 }, { "epoch": 0.09, "learning_rate": 0.00019130434782608697, "loss": 1.8729, "step": 286 }, { "epoch": 0.09, "learning_rate": 0.0001919732441471572, "loss": 1.8179, "step": 287 }, { "epoch": 0.09, "learning_rate": 0.00019264214046822744, "loss": 1.7985, "step": 288 }, { "epoch": 0.09, "learning_rate": 0.00019331103678929768, "loss": 1.8525, "step": 289 }, { "epoch": 0.09, "learning_rate": 0.0001939799331103679, "loss": 1.8157, "step": 290 }, { "epoch": 0.09, "learning_rate": 0.00019464882943143815, "loss": 1.8259, "step": 291 }, { "epoch": 0.09, "learning_rate": 0.00019531772575250838, "loss": 1.8777, "step": 292 }, { "epoch": 0.09, "learning_rate": 0.00019598662207357862, "loss": 1.7881, "step": 293 }, { "epoch": 0.09, "learning_rate": 0.00019665551839464883, "loss": 1.7182, "step": 294 }, { "epoch": 0.09, "learning_rate": 0.00019732441471571907, "loss": 1.8926, "step": 295 }, { "epoch": 0.09, "learning_rate": 0.0001979933110367893, "loss": 1.784, "step": 296 }, { "epoch": 0.09, "learning_rate": 0.00019866220735785954, "loss": 1.8236, "step": 297 }, { "epoch": 0.09, "learning_rate": 0.00019933110367892977, "loss": 1.788, "step": 298 }, { "epoch": 0.09, "learning_rate": 0.0002, "loss": 1.8186, "step": 299 }, { "epoch": 0.09, "learning_rate": 0.00019999999471936308, "loss": 1.8485, "step": 300 }, { "epoch": 0.09, "learning_rate": 0.00019999997887745283, "loss": 1.921, "step": 301 }, { "epoch": 0.09, "learning_rate": 0.00019999995247427097, "loss": 1.7631, "step": 302 }, { "epoch": 0.09, "learning_rate": 0.00019999991550982026, "loss": 1.8798, "step": 303 }, { "epoch": 0.09, "learning_rate": 0.00019999986798410462, "loss": 1.7392, "step": 304 }, { "epoch": 0.09, "learning_rate": 0.00019999980989712906, "loss": 1.8124, "step": 305 }, { "epoch": 0.09, "learning_rate": 0.00019999974124889968, "loss": 1.9652, "step": 306 }, { "epoch": 0.09, "learning_rate": 0.0001999996620394238, "loss": 1.8006, "step": 307 }, { "epoch": 0.09, "learning_rate": 0.00019999957226870974, "loss": 1.7956, "step": 308 }, { "epoch": 0.09, "learning_rate": 0.000199999471936767, "loss": 1.9129, "step": 309 }, { "epoch": 0.09, "learning_rate": 0.00019999936104360614, "loss": 1.8728, "step": 310 }, { "epoch": 0.09, "learning_rate": 0.00019999923958923893, "loss": 1.7498, "step": 311 }, { "epoch": 0.09, "learning_rate": 0.00019999910757367815, "loss": 1.8929, "step": 312 }, { "epoch": 0.09, "learning_rate": 0.00019999896499693778, "loss": 1.9899, "step": 313 }, { "epoch": 0.09, "learning_rate": 0.00019999881185903283, "loss": 1.8994, "step": 314 }, { "epoch": 0.09, "learning_rate": 0.0001999986481599795, "loss": 1.8262, "step": 315 }, { "epoch": 0.1, "learning_rate": 0.0001999984738997951, "loss": 1.8977, "step": 316 }, { "epoch": 0.1, "learning_rate": 0.00019999828907849802, "loss": 1.8829, "step": 317 }, { "epoch": 0.1, "learning_rate": 0.00019999809369610774, "loss": 1.8424, "step": 318 }, { "epoch": 0.1, "learning_rate": 0.00019999788775264493, "loss": 1.8548, "step": 319 }, { "epoch": 0.1, "learning_rate": 0.00019999767124813137, "loss": 1.8807, "step": 320 }, { "epoch": 0.1, "learning_rate": 0.00019999744418258984, "loss": 1.9146, "step": 321 }, { "epoch": 0.1, "learning_rate": 0.00019999720655604445, "loss": 1.9298, "step": 322 }, { "epoch": 0.1, "learning_rate": 0.00019999695836852018, "loss": 1.9153, "step": 323 }, { "epoch": 0.1, "learning_rate": 0.00019999669962004327, "loss": 1.9103, "step": 324 }, { "epoch": 0.1, "learning_rate": 0.00019999643031064107, "loss": 1.9376, "step": 325 }, { "epoch": 0.1, "learning_rate": 0.00019999615044034204, "loss": 1.8413, "step": 326 }, { "epoch": 0.1, "learning_rate": 0.00019999586000917568, "loss": 1.8711, "step": 327 }, { "epoch": 0.1, "learning_rate": 0.00019999555901717268, "loss": 1.8639, "step": 328 }, { "epoch": 0.1, "learning_rate": 0.00019999524746436488, "loss": 1.9061, "step": 329 }, { "epoch": 0.1, "learning_rate": 0.00019999492535078512, "loss": 1.8214, "step": 330 }, { "epoch": 0.1, "learning_rate": 0.00019999459267646742, "loss": 1.8504, "step": 331 }, { "epoch": 0.1, "learning_rate": 0.00019999424944144699, "loss": 1.9894, "step": 332 }, { "epoch": 0.1, "learning_rate": 0.00019999389564576, "loss": 1.8854, "step": 333 }, { "epoch": 0.1, "learning_rate": 0.00019999353128944387, "loss": 1.9072, "step": 334 }, { "epoch": 0.1, "learning_rate": 0.00019999315637253705, "loss": 1.8645, "step": 335 }, { "epoch": 0.1, "learning_rate": 0.0001999927708950791, "loss": 1.9217, "step": 336 }, { "epoch": 0.1, "learning_rate": 0.0001999923748571108, "loss": 1.8967, "step": 337 }, { "epoch": 0.1, "learning_rate": 0.00019999196825867398, "loss": 1.8893, "step": 338 }, { "epoch": 0.1, "learning_rate": 0.00019999155109981151, "loss": 1.907, "step": 339 }, { "epoch": 0.1, "learning_rate": 0.00019999112338056754, "loss": 1.9115, "step": 340 }, { "epoch": 0.1, "learning_rate": 0.00019999068510098713, "loss": 1.8622, "step": 341 }, { "epoch": 0.1, "learning_rate": 0.00019999023626111667, "loss": 1.7241, "step": 342 }, { "epoch": 0.1, "learning_rate": 0.0001999897768610035, "loss": 1.912, "step": 343 }, { "epoch": 0.1, "learning_rate": 0.00019998930690069618, "loss": 1.8272, "step": 344 }, { "epoch": 0.1, "learning_rate": 0.0001999888263802443, "loss": 1.7599, "step": 345 }, { "epoch": 0.1, "learning_rate": 0.00019998833529969866, "loss": 1.9239, "step": 346 }, { "epoch": 0.1, "learning_rate": 0.00019998783365911108, "loss": 2.0327, "step": 347 }, { "epoch": 0.1, "learning_rate": 0.0001999873214585346, "loss": 1.8381, "step": 348 }, { "epoch": 0.11, "learning_rate": 0.00019998679869802323, "loss": 1.8353, "step": 349 }, { "epoch": 0.11, "learning_rate": 0.00019998626537763223, "loss": 1.8839, "step": 350 }, { "epoch": 0.11, "learning_rate": 0.0001999857214974179, "loss": 1.7993, "step": 351 }, { "epoch": 0.11, "learning_rate": 0.00019998516705743772, "loss": 1.8097, "step": 352 }, { "epoch": 0.11, "learning_rate": 0.00019998460205775023, "loss": 1.9031, "step": 353 }, { "epoch": 0.11, "learning_rate": 0.0001999840264984151, "loss": 1.861, "step": 354 }, { "epoch": 0.11, "learning_rate": 0.0001999834403794931, "loss": 1.8501, "step": 355 }, { "epoch": 0.11, "learning_rate": 0.00019998284370104617, "loss": 1.8143, "step": 356 }, { "epoch": 0.11, "learning_rate": 0.00019998223646313726, "loss": 1.8088, "step": 357 }, { "epoch": 0.11, "learning_rate": 0.0001999816186658306, "loss": 1.858, "step": 358 }, { "epoch": 0.11, "learning_rate": 0.00019998099030919133, "loss": 1.9811, "step": 359 }, { "epoch": 0.11, "learning_rate": 0.00019998035139328585, "loss": 1.8008, "step": 360 }, { "epoch": 0.11, "learning_rate": 0.0001999797019181817, "loss": 1.7789, "step": 361 }, { "epoch": 0.11, "learning_rate": 0.00019997904188394738, "loss": 1.8904, "step": 362 }, { "epoch": 0.11, "learning_rate": 0.00019997837129065267, "loss": 1.9052, "step": 363 }, { "epoch": 0.11, "learning_rate": 0.00019997769013836836, "loss": 1.7334, "step": 364 }, { "epoch": 0.11, "learning_rate": 0.0001999769984271664, "loss": 1.8185, "step": 365 }, { "epoch": 0.11, "learning_rate": 0.00019997629615711984, "loss": 1.802, "step": 366 }, { "epoch": 0.11, "learning_rate": 0.0001999755833283028, "loss": 1.827, "step": 367 }, { "epoch": 0.11, "learning_rate": 0.00019997485994079065, "loss": 1.8606, "step": 368 }, { "epoch": 0.11, "learning_rate": 0.00019997412599465975, "loss": 1.8757, "step": 369 }, { "epoch": 0.11, "learning_rate": 0.00019997338148998762, "loss": 1.8707, "step": 370 }, { "epoch": 0.11, "learning_rate": 0.00019997262642685283, "loss": 1.8895, "step": 371 }, { "epoch": 0.11, "learning_rate": 0.00019997186080533523, "loss": 1.8825, "step": 372 }, { "epoch": 0.11, "learning_rate": 0.0001999710846255156, "loss": 1.976, "step": 373 }, { "epoch": 0.11, "learning_rate": 0.00019997029788747591, "loss": 1.7429, "step": 374 }, { "epoch": 0.11, "learning_rate": 0.00019996950059129932, "loss": 1.8089, "step": 375 }, { "epoch": 0.11, "learning_rate": 0.00019996869273706998, "loss": 1.8994, "step": 376 }, { "epoch": 0.11, "learning_rate": 0.00019996787432487325, "loss": 1.7067, "step": 377 }, { "epoch": 0.11, "learning_rate": 0.00019996704535479553, "loss": 1.7992, "step": 378 }, { "epoch": 0.11, "learning_rate": 0.00019996620582692437, "loss": 1.8963, "step": 379 }, { "epoch": 0.11, "learning_rate": 0.00019996535574134843, "loss": 1.9056, "step": 380 }, { "epoch": 0.11, "learning_rate": 0.0001999644950981575, "loss": 1.8315, "step": 381 }, { "epoch": 0.11, "learning_rate": 0.0001999636238974425, "loss": 1.8467, "step": 382 }, { "epoch": 0.12, "learning_rate": 0.0001999627421392954, "loss": 1.9356, "step": 383 }, { "epoch": 0.12, "learning_rate": 0.00019996184982380935, "loss": 1.9129, "step": 384 }, { "epoch": 0.12, "learning_rate": 0.0001999609469510786, "loss": 1.9939, "step": 385 }, { "epoch": 0.12, "learning_rate": 0.00019996003352119846, "loss": 1.8366, "step": 386 }, { "epoch": 0.12, "learning_rate": 0.0001999591095342654, "loss": 1.895, "step": 387 }, { "epoch": 0.12, "learning_rate": 0.0001999581749903771, "loss": 1.8437, "step": 388 }, { "epoch": 0.12, "learning_rate": 0.00019995722988963213, "loss": 1.871, "step": 389 }, { "epoch": 0.12, "learning_rate": 0.00019995627423213038, "loss": 1.8205, "step": 390 }, { "epoch": 0.12, "learning_rate": 0.0001999553080179728, "loss": 1.8399, "step": 391 }, { "epoch": 0.12, "learning_rate": 0.00019995433124726138, "loss": 1.9753, "step": 392 }, { "epoch": 0.12, "learning_rate": 0.0001999533439200993, "loss": 1.89, "step": 393 }, { "epoch": 0.12, "learning_rate": 0.0001999523460365908, "loss": 1.8196, "step": 394 }, { "epoch": 0.12, "learning_rate": 0.00019995133759684135, "loss": 1.8006, "step": 395 }, { "epoch": 0.12, "learning_rate": 0.0001999503186009574, "loss": 1.7668, "step": 396 }, { "epoch": 0.12, "learning_rate": 0.00019994928904904654, "loss": 1.894, "step": 397 }, { "epoch": 0.12, "learning_rate": 0.00019994824894121758, "loss": 1.9678, "step": 398 }, { "epoch": 0.12, "learning_rate": 0.00019994719827758032, "loss": 1.8127, "step": 399 }, { "epoch": 0.12, "learning_rate": 0.00019994613705824575, "loss": 1.8995, "step": 400 }, { "epoch": 0.12, "learning_rate": 0.00019994506528332588, "loss": 1.7571, "step": 401 }, { "epoch": 0.12, "learning_rate": 0.00019994398295293398, "loss": 1.8374, "step": 402 }, { "epoch": 0.12, "learning_rate": 0.00019994289006718433, "loss": 1.753, "step": 403 }, { "epoch": 0.12, "learning_rate": 0.00019994178662619235, "loss": 1.8761, "step": 404 }, { "epoch": 0.12, "learning_rate": 0.00019994067263007462, "loss": 1.893, "step": 405 }, { "epoch": 0.12, "learning_rate": 0.0001999395480789487, "loss": 1.9113, "step": 406 }, { "epoch": 0.12, "learning_rate": 0.00019993841297293345, "loss": 1.8335, "step": 407 }, { "epoch": 0.12, "learning_rate": 0.00019993726731214867, "loss": 1.9374, "step": 408 }, { "epoch": 0.12, "learning_rate": 0.00019993611109671544, "loss": 1.8959, "step": 409 }, { "epoch": 0.12, "learning_rate": 0.0001999349443267558, "loss": 1.7892, "step": 410 }, { "epoch": 0.12, "learning_rate": 0.00019993376700239303, "loss": 1.7988, "step": 411 }, { "epoch": 0.12, "learning_rate": 0.00019993257912375142, "loss": 1.8319, "step": 412 }, { "epoch": 0.12, "learning_rate": 0.00019993138069095647, "loss": 1.8612, "step": 413 }, { "epoch": 0.12, "learning_rate": 0.00019993017170413473, "loss": 1.9018, "step": 414 }, { "epoch": 0.12, "learning_rate": 0.00019992895216341388, "loss": 1.7959, "step": 415 }, { "epoch": 0.13, "learning_rate": 0.00019992772206892272, "loss": 1.8159, "step": 416 }, { "epoch": 0.13, "learning_rate": 0.00019992648142079115, "loss": 1.7893, "step": 417 }, { "epoch": 0.13, "learning_rate": 0.00019992523021915027, "loss": 1.9472, "step": 418 }, { "epoch": 0.13, "learning_rate": 0.00019992396846413212, "loss": 1.8016, "step": 419 }, { "epoch": 0.13, "learning_rate": 0.00019992269615587004, "loss": 1.854, "step": 420 }, { "epoch": 0.13, "learning_rate": 0.00019992141329449833, "loss": 1.8203, "step": 421 }, { "epoch": 0.13, "learning_rate": 0.00019992011988015254, "loss": 1.9128, "step": 422 }, { "epoch": 0.13, "learning_rate": 0.00019991881591296923, "loss": 1.8545, "step": 423 }, { "epoch": 0.13, "learning_rate": 0.00019991750139308615, "loss": 1.8747, "step": 424 }, { "epoch": 0.13, "learning_rate": 0.00019991617632064213, "loss": 1.8645, "step": 425 }, { "epoch": 0.13, "learning_rate": 0.00019991484069577706, "loss": 1.7714, "step": 426 }, { "epoch": 0.13, "learning_rate": 0.00019991349451863204, "loss": 1.8165, "step": 427 }, { "epoch": 0.13, "learning_rate": 0.00019991213778934922, "loss": 1.8748, "step": 428 }, { "epoch": 0.13, "learning_rate": 0.00019991077050807195, "loss": 1.887, "step": 429 }, { "epoch": 0.13, "learning_rate": 0.00019990939267494455, "loss": 1.9149, "step": 430 }, { "epoch": 0.13, "learning_rate": 0.00019990800429011261, "loss": 1.8649, "step": 431 }, { "epoch": 0.13, "learning_rate": 0.0001999066053537227, "loss": 1.8247, "step": 432 }, { "epoch": 0.13, "learning_rate": 0.0001999051958659226, "loss": 1.8751, "step": 433 }, { "epoch": 0.13, "learning_rate": 0.00019990377582686117, "loss": 1.8572, "step": 434 }, { "epoch": 0.13, "learning_rate": 0.00019990234523668837, "loss": 1.8271, "step": 435 }, { "epoch": 0.13, "learning_rate": 0.0001999009040955553, "loss": 1.744, "step": 436 }, { "epoch": 0.13, "learning_rate": 0.00019989945240361413, "loss": 1.8599, "step": 437 }, { "epoch": 0.13, "learning_rate": 0.00019989799016101824, "loss": 1.8314, "step": 438 }, { "epoch": 0.13, "learning_rate": 0.00019989651736792202, "loss": 1.9194, "step": 439 }, { "epoch": 0.13, "learning_rate": 0.000199895034024481, "loss": 1.7826, "step": 440 }, { "epoch": 0.13, "learning_rate": 0.0001998935401308519, "loss": 1.8065, "step": 441 }, { "epoch": 0.13, "learning_rate": 0.00019989203568719242, "loss": 1.8777, "step": 442 }, { "epoch": 0.13, "learning_rate": 0.00019989052069366149, "loss": 2.0042, "step": 443 }, { "epoch": 0.13, "learning_rate": 0.0001998889951504191, "loss": 1.8523, "step": 444 }, { "epoch": 0.13, "learning_rate": 0.00019988745905762638, "loss": 1.8761, "step": 445 }, { "epoch": 0.13, "learning_rate": 0.00019988591241544557, "loss": 1.8546, "step": 446 }, { "epoch": 0.13, "learning_rate": 0.00019988435522404, "loss": 1.8503, "step": 447 }, { "epoch": 0.13, "learning_rate": 0.00019988278748357409, "loss": 1.8316, "step": 448 }, { "epoch": 0.14, "learning_rate": 0.0001998812091942135, "loss": 1.8887, "step": 449 }, { "epoch": 0.14, "learning_rate": 0.00019987962035612484, "loss": 1.7232, "step": 450 }, { "epoch": 0.14, "learning_rate": 0.00019987802096947595, "loss": 1.7475, "step": 451 }, { "epoch": 0.14, "learning_rate": 0.00019987641103443572, "loss": 1.9402, "step": 452 }, { "epoch": 0.14, "learning_rate": 0.00019987479055117425, "loss": 1.9217, "step": 453 }, { "epoch": 0.14, "learning_rate": 0.00019987315951986259, "loss": 1.8137, "step": 454 }, { "epoch": 0.14, "learning_rate": 0.00019987151794067303, "loss": 1.8249, "step": 455 }, { "epoch": 0.14, "learning_rate": 0.00019986986581377897, "loss": 1.8687, "step": 456 }, { "epoch": 0.14, "learning_rate": 0.00019986820313935488, "loss": 1.8847, "step": 457 }, { "epoch": 0.14, "learning_rate": 0.00019986652991757633, "loss": 1.8647, "step": 458 }, { "epoch": 0.14, "learning_rate": 0.00019986484614862007, "loss": 1.9255, "step": 459 }, { "epoch": 0.14, "learning_rate": 0.00019986315183266392, "loss": 1.7703, "step": 460 }, { "epoch": 0.14, "learning_rate": 0.00019986144696988682, "loss": 1.7475, "step": 461 }, { "epoch": 0.14, "learning_rate": 0.0001998597315604688, "loss": 1.9437, "step": 462 }, { "epoch": 0.14, "learning_rate": 0.00019985800560459106, "loss": 1.9294, "step": 463 }, { "epoch": 0.14, "learning_rate": 0.0001998562691024359, "loss": 1.8611, "step": 464 }, { "epoch": 0.14, "learning_rate": 0.00019985452205418668, "loss": 1.7988, "step": 465 }, { "epoch": 0.14, "learning_rate": 0.00019985276446002792, "loss": 1.8461, "step": 466 }, { "epoch": 0.14, "learning_rate": 0.00019985099632014523, "loss": 1.8678, "step": 467 }, { "epoch": 0.14, "learning_rate": 0.0001998492176347254, "loss": 1.7568, "step": 468 }, { "epoch": 0.14, "learning_rate": 0.0001998474284039562, "loss": 1.8428, "step": 469 }, { "epoch": 0.14, "learning_rate": 0.00019984562862802668, "loss": 1.8841, "step": 470 }, { "epoch": 0.14, "learning_rate": 0.00019984381830712687, "loss": 1.9653, "step": 471 }, { "epoch": 0.14, "learning_rate": 0.00019984199744144796, "loss": 1.814, "step": 472 }, { "epoch": 0.14, "learning_rate": 0.0001998401660311823, "loss": 1.7967, "step": 473 }, { "epoch": 0.14, "learning_rate": 0.00019983832407652322, "loss": 1.8362, "step": 474 }, { "epoch": 0.14, "learning_rate": 0.00019983647157766537, "loss": 1.9774, "step": 475 }, { "epoch": 0.14, "learning_rate": 0.00019983460853480433, "loss": 1.762, "step": 476 }, { "epoch": 0.14, "learning_rate": 0.00019983273494813685, "loss": 1.9166, "step": 477 }, { "epoch": 0.14, "learning_rate": 0.00019983085081786087, "loss": 1.8378, "step": 478 }, { "epoch": 0.14, "learning_rate": 0.00019982895614417532, "loss": 1.7213, "step": 479 }, { "epoch": 0.14, "learning_rate": 0.0001998270509272803, "loss": 1.94, "step": 480 }, { "epoch": 0.14, "learning_rate": 0.00019982513516737707, "loss": 1.8934, "step": 481 }, { "epoch": 0.15, "learning_rate": 0.00019982320886466793, "loss": 1.7713, "step": 482 }, { "epoch": 0.15, "learning_rate": 0.00019982127201935627, "loss": 1.8548, "step": 483 }, { "epoch": 0.15, "learning_rate": 0.00019981932463164674, "loss": 1.9677, "step": 484 }, { "epoch": 0.15, "learning_rate": 0.00019981736670174498, "loss": 1.8022, "step": 485 }, { "epoch": 0.15, "learning_rate": 0.00019981539822985775, "loss": 1.8264, "step": 486 }, { "epoch": 0.15, "learning_rate": 0.00019981341921619295, "loss": 1.803, "step": 487 }, { "epoch": 0.15, "learning_rate": 0.00019981142966095958, "loss": 1.9566, "step": 488 }, { "epoch": 0.15, "learning_rate": 0.0001998094295643678, "loss": 1.9379, "step": 489 }, { "epoch": 0.15, "learning_rate": 0.00019980741892662883, "loss": 1.8091, "step": 490 }, { "epoch": 0.15, "learning_rate": 0.000199805397747955, "loss": 1.7669, "step": 491 }, { "epoch": 0.15, "learning_rate": 0.00019980336602855978, "loss": 1.8805, "step": 492 }, { "epoch": 0.15, "learning_rate": 0.00019980132376865774, "loss": 1.9369, "step": 493 }, { "epoch": 0.15, "learning_rate": 0.00019979927096846461, "loss": 1.8003, "step": 494 }, { "epoch": 0.15, "learning_rate": 0.00019979720762819714, "loss": 1.9943, "step": 495 }, { "epoch": 0.15, "learning_rate": 0.00019979513374807325, "loss": 1.9371, "step": 496 }, { "epoch": 0.15, "learning_rate": 0.000199793049328312, "loss": 1.8482, "step": 497 }, { "epoch": 0.15, "learning_rate": 0.00019979095436913354, "loss": 1.8186, "step": 498 }, { "epoch": 0.15, "learning_rate": 0.00019978884887075906, "loss": 1.8169, "step": 499 }, { "epoch": 0.15, "learning_rate": 0.00019978673283341098, "loss": 1.9372, "step": 500 }, { "epoch": 0.15, "learning_rate": 0.00019978460625731277, "loss": 1.792, "step": 501 }, { "epoch": 0.15, "learning_rate": 0.00019978246914268902, "loss": 1.8175, "step": 502 }, { "epoch": 0.15, "learning_rate": 0.00019978032148976543, "loss": 1.9393, "step": 503 }, { "epoch": 0.15, "learning_rate": 0.00019977816329876883, "loss": 1.9399, "step": 504 }, { "epoch": 0.15, "learning_rate": 0.00019977599456992717, "loss": 1.8276, "step": 505 }, { "epoch": 0.15, "learning_rate": 0.00019977381530346944, "loss": 1.8325, "step": 506 }, { "epoch": 0.15, "learning_rate": 0.00019977162549962585, "loss": 1.8357, "step": 507 }, { "epoch": 0.15, "learning_rate": 0.00019976942515862766, "loss": 1.8801, "step": 508 }, { "epoch": 0.15, "learning_rate": 0.00019976721428070725, "loss": 1.7663, "step": 509 }, { "epoch": 0.15, "learning_rate": 0.0001997649928660981, "loss": 1.852, "step": 510 }, { "epoch": 0.15, "learning_rate": 0.00019976276091503484, "loss": 1.8173, "step": 511 }, { "epoch": 0.15, "learning_rate": 0.00019976051842775317, "loss": 1.8626, "step": 512 }, { "epoch": 0.15, "learning_rate": 0.00019975826540448996, "loss": 1.8836, "step": 513 }, { "epoch": 0.15, "learning_rate": 0.00019975600184548315, "loss": 1.8237, "step": 514 }, { "epoch": 0.16, "learning_rate": 0.00019975372775097179, "loss": 1.8176, "step": 515 }, { "epoch": 0.16, "learning_rate": 0.00019975144312119603, "loss": 1.8416, "step": 516 }, { "epoch": 0.16, "learning_rate": 0.00019974914795639718, "loss": 1.8996, "step": 517 }, { "epoch": 0.16, "learning_rate": 0.00019974684225681764, "loss": 1.7254, "step": 518 }, { "epoch": 0.16, "learning_rate": 0.00019974452602270097, "loss": 1.8588, "step": 519 }, { "epoch": 0.16, "learning_rate": 0.0001997421992542917, "loss": 1.8067, "step": 520 }, { "epoch": 0.16, "learning_rate": 0.00019973986195183564, "loss": 1.8395, "step": 521 }, { "epoch": 0.16, "learning_rate": 0.00019973751411557958, "loss": 1.8094, "step": 522 }, { "epoch": 0.16, "learning_rate": 0.00019973515574577152, "loss": 1.834, "step": 523 }, { "epoch": 0.16, "learning_rate": 0.0001997327868426605, "loss": 1.9107, "step": 524 }, { "epoch": 0.16, "learning_rate": 0.00019973040740649676, "loss": 1.897, "step": 525 }, { "epoch": 0.16, "learning_rate": 0.00019972801743753154, "loss": 1.8825, "step": 526 }, { "epoch": 0.16, "learning_rate": 0.00019972561693601732, "loss": 1.8282, "step": 527 }, { "epoch": 0.16, "learning_rate": 0.00019972320590220757, "loss": 1.9347, "step": 528 }, { "epoch": 0.16, "learning_rate": 0.00019972078433635692, "loss": 1.86, "step": 529 }, { "epoch": 0.16, "learning_rate": 0.00019971835223872117, "loss": 1.926, "step": 530 }, { "epoch": 0.16, "learning_rate": 0.00019971590960955716, "loss": 1.9442, "step": 531 }, { "epoch": 0.16, "learning_rate": 0.0001997134564491228, "loss": 1.7551, "step": 532 }, { "epoch": 0.16, "learning_rate": 0.00019971099275767725, "loss": 1.8532, "step": 533 }, { "epoch": 0.16, "learning_rate": 0.0001997085185354807, "loss": 1.8714, "step": 534 }, { "epoch": 0.16, "learning_rate": 0.00019970603378279442, "loss": 1.9936, "step": 535 }, { "epoch": 0.16, "learning_rate": 0.00019970353849988089, "loss": 1.8962, "step": 536 }, { "epoch": 0.16, "learning_rate": 0.00019970103268700354, "loss": 1.9057, "step": 537 }, { "epoch": 0.16, "learning_rate": 0.00019969851634442715, "loss": 1.8226, "step": 538 }, { "epoch": 0.16, "learning_rate": 0.0001996959894724174, "loss": 1.9008, "step": 539 }, { "epoch": 0.16, "learning_rate": 0.00019969345207124115, "loss": 1.7749, "step": 540 }, { "epoch": 0.16, "learning_rate": 0.00019969090414116643, "loss": 1.7573, "step": 541 }, { "epoch": 0.16, "learning_rate": 0.00019968834568246229, "loss": 1.85, "step": 542 }, { "epoch": 0.16, "learning_rate": 0.00019968577669539895, "loss": 1.8468, "step": 543 }, { "epoch": 0.16, "learning_rate": 0.00019968319718024775, "loss": 1.7898, "step": 544 }, { "epoch": 0.16, "learning_rate": 0.00019968060713728108, "loss": 1.819, "step": 545 }, { "epoch": 0.16, "learning_rate": 0.00019967800656677252, "loss": 1.7076, "step": 546 }, { "epoch": 0.16, "learning_rate": 0.00019967539546899668, "loss": 1.8277, "step": 547 }, { "epoch": 0.16, "learning_rate": 0.0001996727738442294, "loss": 1.7693, "step": 548 }, { "epoch": 0.17, "learning_rate": 0.00019967014169274745, "loss": 1.8026, "step": 549 }, { "epoch": 0.17, "learning_rate": 0.00019966749901482891, "loss": 1.6912, "step": 550 }, { "epoch": 0.17, "learning_rate": 0.00019966484581075285, "loss": 1.785, "step": 551 }, { "epoch": 0.17, "learning_rate": 0.00019966218208079947, "loss": 1.8084, "step": 552 }, { "epoch": 0.17, "learning_rate": 0.00019965950782525007, "loss": 1.826, "step": 553 }, { "epoch": 0.17, "learning_rate": 0.00019965682304438716, "loss": 1.787, "step": 554 }, { "epoch": 0.17, "learning_rate": 0.00019965412773849424, "loss": 1.8273, "step": 555 }, { "epoch": 0.17, "learning_rate": 0.00019965142190785597, "loss": 1.8502, "step": 556 }, { "epoch": 0.17, "learning_rate": 0.00019964870555275814, "loss": 1.8294, "step": 557 }, { "epoch": 0.17, "learning_rate": 0.00019964597867348758, "loss": 1.8793, "step": 558 }, { "epoch": 0.17, "learning_rate": 0.00019964324127033235, "loss": 1.8506, "step": 559 }, { "epoch": 0.17, "learning_rate": 0.00019964049334358152, "loss": 1.8349, "step": 560 }, { "epoch": 0.17, "learning_rate": 0.0001996377348935253, "loss": 1.9517, "step": 561 }, { "epoch": 0.17, "learning_rate": 0.00019963496592045502, "loss": 1.8893, "step": 562 }, { "epoch": 0.17, "learning_rate": 0.00019963218642466316, "loss": 1.8513, "step": 563 }, { "epoch": 0.17, "learning_rate": 0.0001996293964064432, "loss": 1.7881, "step": 564 }, { "epoch": 0.17, "learning_rate": 0.00019962659586608984, "loss": 1.8427, "step": 565 }, { "epoch": 0.17, "learning_rate": 0.00019962378480389888, "loss": 1.9228, "step": 566 }, { "epoch": 0.17, "learning_rate": 0.00019962096322016714, "loss": 1.7605, "step": 567 }, { "epoch": 0.17, "learning_rate": 0.0001996181311151927, "loss": 1.8914, "step": 568 }, { "epoch": 0.17, "learning_rate": 0.0001996152884892746, "loss": 1.934, "step": 569 }, { "epoch": 0.17, "learning_rate": 0.00019961243534271302, "loss": 1.8823, "step": 570 }, { "epoch": 0.17, "learning_rate": 0.00019960957167580937, "loss": 1.8747, "step": 571 }, { "epoch": 0.17, "learning_rate": 0.0001996066974888661, "loss": 1.8752, "step": 572 }, { "epoch": 0.17, "learning_rate": 0.0001996038127821867, "loss": 1.7787, "step": 573 }, { "epoch": 0.17, "learning_rate": 0.00019960091755607585, "loss": 1.754, "step": 574 }, { "epoch": 0.17, "learning_rate": 0.00019959801181083933, "loss": 1.8482, "step": 575 }, { "epoch": 0.17, "learning_rate": 0.000199595095546784, "loss": 1.8733, "step": 576 }, { "epoch": 0.17, "learning_rate": 0.0001995921687642179, "loss": 1.7656, "step": 577 }, { "epoch": 0.17, "learning_rate": 0.0001995892314634501, "loss": 1.8519, "step": 578 }, { "epoch": 0.17, "learning_rate": 0.0001995862836447908, "loss": 1.8849, "step": 579 }, { "epoch": 0.17, "learning_rate": 0.00019958332530855142, "loss": 1.8482, "step": 580 }, { "epoch": 0.17, "learning_rate": 0.00019958035645504428, "loss": 1.9121, "step": 581 }, { "epoch": 0.18, "learning_rate": 0.000199577377084583, "loss": 1.8489, "step": 582 }, { "epoch": 0.18, "learning_rate": 0.00019957438719748224, "loss": 1.8733, "step": 583 }, { "epoch": 0.18, "learning_rate": 0.00019957138679405772, "loss": 1.822, "step": 584 }, { "epoch": 0.18, "learning_rate": 0.00019956837587462638, "loss": 1.7017, "step": 585 }, { "epoch": 0.18, "learning_rate": 0.00019956535443950616, "loss": 1.8047, "step": 586 }, { "epoch": 0.18, "learning_rate": 0.0001995623224890162, "loss": 1.7887, "step": 587 }, { "epoch": 0.18, "learning_rate": 0.00019955928002347671, "loss": 1.8364, "step": 588 }, { "epoch": 0.18, "learning_rate": 0.00019955622704320898, "loss": 1.9673, "step": 589 }, { "epoch": 0.18, "learning_rate": 0.00019955316354853546, "loss": 1.8255, "step": 590 }, { "epoch": 0.18, "learning_rate": 0.00019955008953977972, "loss": 1.8573, "step": 591 }, { "epoch": 0.18, "learning_rate": 0.00019954700501726637, "loss": 1.8435, "step": 592 }, { "epoch": 0.18, "learning_rate": 0.00019954390998132122, "loss": 1.8086, "step": 593 }, { "epoch": 0.18, "learning_rate": 0.00019954080443227112, "loss": 1.9212, "step": 594 }, { "epoch": 0.18, "learning_rate": 0.00019953768837044404, "loss": 1.7663, "step": 595 }, { "epoch": 0.18, "learning_rate": 0.00019953456179616912, "loss": 1.7951, "step": 596 }, { "epoch": 0.18, "learning_rate": 0.00019953142470977651, "loss": 1.823, "step": 597 }, { "epoch": 0.18, "learning_rate": 0.00019952827711159756, "loss": 1.828, "step": 598 }, { "epoch": 0.18, "learning_rate": 0.00019952511900196473, "loss": 1.7502, "step": 599 }, { "epoch": 0.18, "learning_rate": 0.00019952195038121148, "loss": 1.7764, "step": 600 }, { "epoch": 0.18, "learning_rate": 0.00019951877124967253, "loss": 1.8328, "step": 601 }, { "epoch": 0.18, "learning_rate": 0.0001995155816076836, "loss": 1.7997, "step": 602 }, { "epoch": 0.18, "learning_rate": 0.00019951238145558153, "loss": 1.8688, "step": 603 }, { "epoch": 0.18, "learning_rate": 0.00019950917079370438, "loss": 1.8742, "step": 604 }, { "epoch": 0.18, "learning_rate": 0.00019950594962239113, "loss": 1.98, "step": 605 }, { "epoch": 0.18, "learning_rate": 0.00019950271794198204, "loss": 1.9651, "step": 606 }, { "epoch": 0.18, "learning_rate": 0.00019949947575281844, "loss": 1.8456, "step": 607 }, { "epoch": 0.18, "learning_rate": 0.00019949622305524267, "loss": 1.827, "step": 608 }, { "epoch": 0.18, "learning_rate": 0.00019949295984959836, "loss": 1.7996, "step": 609 }, { "epoch": 0.18, "learning_rate": 0.00019948968613623, "loss": 1.8491, "step": 610 }, { "epoch": 0.18, "learning_rate": 0.0001994864019154835, "loss": 1.8719, "step": 611 }, { "epoch": 0.18, "learning_rate": 0.0001994831071877056, "loss": 1.7874, "step": 612 }, { "epoch": 0.18, "learning_rate": 0.0001994798019532443, "loss": 1.8198, "step": 613 }, { "epoch": 0.18, "learning_rate": 0.0001994764862124487, "loss": 1.8146, "step": 614 }, { "epoch": 0.19, "learning_rate": 0.00019947315996566895, "loss": 1.7536, "step": 615 }, { "epoch": 0.19, "learning_rate": 0.00019946982321325638, "loss": 1.8907, "step": 616 }, { "epoch": 0.19, "learning_rate": 0.00019946647595556335, "loss": 1.8327, "step": 617 }, { "epoch": 0.19, "learning_rate": 0.0001994631181929434, "loss": 1.818, "step": 618 }, { "epoch": 0.19, "learning_rate": 0.0001994597499257511, "loss": 1.8121, "step": 619 }, { "epoch": 0.19, "learning_rate": 0.00019945637115434227, "loss": 1.8481, "step": 620 }, { "epoch": 0.19, "learning_rate": 0.00019945298187907373, "loss": 1.811, "step": 621 }, { "epoch": 0.19, "learning_rate": 0.00019944958210030338, "loss": 1.841, "step": 622 }, { "epoch": 0.19, "learning_rate": 0.0001994461718183903, "loss": 1.7903, "step": 623 }, { "epoch": 0.19, "learning_rate": 0.00019944275103369469, "loss": 1.8052, "step": 624 }, { "epoch": 0.19, "learning_rate": 0.0001994393197465778, "loss": 1.7595, "step": 625 }, { "epoch": 0.19, "learning_rate": 0.00019943587795740201, "loss": 1.8066, "step": 626 }, { "epoch": 0.19, "learning_rate": 0.00019943242566653087, "loss": 1.8857, "step": 627 }, { "epoch": 0.19, "learning_rate": 0.00019942896287432894, "loss": 1.8124, "step": 628 }, { "epoch": 0.19, "learning_rate": 0.0001994254895811619, "loss": 1.952, "step": 629 }, { "epoch": 0.19, "learning_rate": 0.00019942200578739665, "loss": 1.8552, "step": 630 }, { "epoch": 0.19, "learning_rate": 0.00019941851149340108, "loss": 1.8033, "step": 631 }, { "epoch": 0.19, "learning_rate": 0.00019941500669954423, "loss": 1.8368, "step": 632 }, { "epoch": 0.19, "learning_rate": 0.00019941149140619628, "loss": 1.7786, "step": 633 }, { "epoch": 0.19, "learning_rate": 0.00019940796561372846, "loss": 1.7694, "step": 634 }, { "epoch": 0.19, "learning_rate": 0.00019940442932251315, "loss": 1.8835, "step": 635 }, { "epoch": 0.19, "learning_rate": 0.00019940088253292385, "loss": 1.8078, "step": 636 }, { "epoch": 0.19, "learning_rate": 0.0001993973252453351, "loss": 1.8688, "step": 637 }, { "epoch": 0.19, "learning_rate": 0.0001993937574601226, "loss": 1.7704, "step": 638 }, { "epoch": 0.19, "learning_rate": 0.0001993901791776632, "loss": 1.9628, "step": 639 }, { "epoch": 0.19, "learning_rate": 0.00019938659039833479, "loss": 1.9996, "step": 640 }, { "epoch": 0.19, "learning_rate": 0.00019938299112251637, "loss": 1.9193, "step": 641 }, { "epoch": 0.19, "learning_rate": 0.00019937938135058811, "loss": 1.7992, "step": 642 }, { "epoch": 0.19, "learning_rate": 0.0001993757610829312, "loss": 1.8128, "step": 643 }, { "epoch": 0.19, "learning_rate": 0.00019937213031992801, "loss": 1.8232, "step": 644 }, { "epoch": 0.19, "learning_rate": 0.000199368489061962, "loss": 1.887, "step": 645 }, { "epoch": 0.19, "learning_rate": 0.00019936483730941773, "loss": 1.9015, "step": 646 }, { "epoch": 0.19, "learning_rate": 0.00019936117506268085, "loss": 1.8955, "step": 647 }, { "epoch": 0.2, "learning_rate": 0.00019935750232213817, "loss": 1.8566, "step": 648 }, { "epoch": 0.2, "learning_rate": 0.0001993538190881776, "loss": 1.8486, "step": 649 }, { "epoch": 0.2, "learning_rate": 0.00019935012536118806, "loss": 1.9114, "step": 650 }, { "epoch": 0.2, "learning_rate": 0.0001993464211415597, "loss": 1.8685, "step": 651 }, { "epoch": 0.2, "learning_rate": 0.00019934270642968374, "loss": 1.8495, "step": 652 }, { "epoch": 0.2, "learning_rate": 0.00019933898122595248, "loss": 1.8501, "step": 653 }, { "epoch": 0.2, "learning_rate": 0.00019933524553075938, "loss": 1.8041, "step": 654 }, { "epoch": 0.2, "learning_rate": 0.00019933149934449898, "loss": 1.8761, "step": 655 }, { "epoch": 0.2, "learning_rate": 0.00019932774266756686, "loss": 1.7554, "step": 656 }, { "epoch": 0.2, "learning_rate": 0.00019932397550035983, "loss": 1.9019, "step": 657 }, { "epoch": 0.2, "learning_rate": 0.0001993201978432757, "loss": 1.7225, "step": 658 }, { "epoch": 0.2, "learning_rate": 0.00019931640969671354, "loss": 1.9226, "step": 659 }, { "epoch": 0.2, "learning_rate": 0.00019931261106107334, "loss": 1.8834, "step": 660 }, { "epoch": 0.2, "learning_rate": 0.0001993088019367563, "loss": 1.8691, "step": 661 }, { "epoch": 0.2, "learning_rate": 0.0001993049823241647, "loss": 1.8862, "step": 662 }, { "epoch": 0.2, "learning_rate": 0.00019930115222370197, "loss": 1.8336, "step": 663 }, { "epoch": 0.2, "learning_rate": 0.0001992973116357726, "loss": 1.791, "step": 664 }, { "epoch": 0.2, "learning_rate": 0.0001992934605607822, "loss": 1.8561, "step": 665 }, { "epoch": 0.2, "learning_rate": 0.00019928959899913755, "loss": 1.803, "step": 666 }, { "epoch": 0.2, "learning_rate": 0.00019928572695124638, "loss": 1.8322, "step": 667 }, { "epoch": 0.2, "learning_rate": 0.00019928184441751773, "loss": 1.7454, "step": 668 }, { "epoch": 0.2, "learning_rate": 0.00019927795139836155, "loss": 1.7861, "step": 669 }, { "epoch": 0.2, "learning_rate": 0.00019927404789418907, "loss": 1.7589, "step": 670 }, { "epoch": 0.2, "learning_rate": 0.0001992701339054125, "loss": 1.8237, "step": 671 }, { "epoch": 0.2, "learning_rate": 0.00019926620943244526, "loss": 1.9181, "step": 672 }, { "epoch": 0.2, "learning_rate": 0.00019926227447570173, "loss": 1.7819, "step": 673 }, { "epoch": 0.2, "learning_rate": 0.00019925832903559758, "loss": 1.8684, "step": 674 }, { "epoch": 0.2, "learning_rate": 0.00019925437311254948, "loss": 1.8427, "step": 675 }, { "epoch": 0.2, "learning_rate": 0.0001992504067069752, "loss": 1.91, "step": 676 }, { "epoch": 0.2, "learning_rate": 0.00019924642981929366, "loss": 1.8251, "step": 677 }, { "epoch": 0.2, "learning_rate": 0.00019924244244992484, "loss": 1.8159, "step": 678 }, { "epoch": 0.2, "learning_rate": 0.00019923844459928994, "loss": 1.7935, "step": 679 }, { "epoch": 0.2, "learning_rate": 0.00019923443626781109, "loss": 1.7533, "step": 680 }, { "epoch": 0.2, "learning_rate": 0.0001992304174559117, "loss": 1.8052, "step": 681 }, { "epoch": 0.21, "learning_rate": 0.00019922638816401614, "loss": 1.8808, "step": 682 }, { "epoch": 0.21, "learning_rate": 0.00019922234839254996, "loss": 1.7601, "step": 683 }, { "epoch": 0.21, "learning_rate": 0.00019921829814193985, "loss": 1.7313, "step": 684 }, { "epoch": 0.21, "learning_rate": 0.00019921423741261357, "loss": 1.8541, "step": 685 }, { "epoch": 0.21, "learning_rate": 0.00019921016620499994, "loss": 1.9811, "step": 686 }, { "epoch": 0.21, "learning_rate": 0.000199206084519529, "loss": 2.0339, "step": 687 }, { "epoch": 0.21, "learning_rate": 0.00019920199235663172, "loss": 1.9252, "step": 688 }, { "epoch": 0.21, "learning_rate": 0.0001991978897167404, "loss": 1.7972, "step": 689 }, { "epoch": 0.21, "learning_rate": 0.00019919377660028827, "loss": 1.7501, "step": 690 }, { "epoch": 0.21, "learning_rate": 0.00019918965300770976, "loss": 1.8123, "step": 691 }, { "epoch": 0.21, "learning_rate": 0.0001991855189394403, "loss": 1.9139, "step": 692 }, { "epoch": 0.21, "learning_rate": 0.0001991813743959166, "loss": 1.8341, "step": 693 }, { "epoch": 0.21, "learning_rate": 0.0001991772193775763, "loss": 1.7795, "step": 694 }, { "epoch": 0.21, "learning_rate": 0.00019917305388485827, "loss": 1.8758, "step": 695 }, { "epoch": 0.21, "learning_rate": 0.0001991688779182024, "loss": 1.8712, "step": 696 }, { "epoch": 0.21, "learning_rate": 0.00019916469147804978, "loss": 1.8601, "step": 697 }, { "epoch": 0.21, "learning_rate": 0.00019916049456484249, "loss": 1.9376, "step": 698 }, { "epoch": 0.21, "learning_rate": 0.00019915628717902384, "loss": 1.8149, "step": 699 }, { "epoch": 0.21, "learning_rate": 0.0001991520693210381, "loss": 1.9298, "step": 700 }, { "epoch": 0.21, "learning_rate": 0.00019914784099133082, "loss": 1.8647, "step": 701 }, { "epoch": 0.21, "learning_rate": 0.00019914360219034852, "loss": 1.802, "step": 702 }, { "epoch": 0.21, "learning_rate": 0.00019913935291853885, "loss": 1.7971, "step": 703 }, { "epoch": 0.21, "learning_rate": 0.0001991350931763506, "loss": 1.8368, "step": 704 }, { "epoch": 0.21, "learning_rate": 0.0001991308229642337, "loss": 1.7964, "step": 705 }, { "epoch": 0.21, "learning_rate": 0.0001991265422826391, "loss": 1.9288, "step": 706 }, { "epoch": 0.21, "learning_rate": 0.0001991222511320189, "loss": 1.8843, "step": 707 }, { "epoch": 0.21, "learning_rate": 0.0001991179495128263, "loss": 1.7784, "step": 708 }, { "epoch": 0.21, "learning_rate": 0.00019911363742551559, "loss": 1.8553, "step": 709 }, { "epoch": 0.21, "learning_rate": 0.0001991093148705422, "loss": 1.8148, "step": 710 }, { "epoch": 0.21, "learning_rate": 0.00019910498184836265, "loss": 1.7434, "step": 711 }, { "epoch": 0.21, "learning_rate": 0.00019910063835943458, "loss": 1.8688, "step": 712 }, { "epoch": 0.21, "learning_rate": 0.00019909628440421665, "loss": 1.7945, "step": 713 }, { "epoch": 0.21, "learning_rate": 0.00019909191998316876, "loss": 1.7943, "step": 714 }, { "epoch": 0.22, "learning_rate": 0.00019908754509675183, "loss": 1.9248, "step": 715 }, { "epoch": 0.22, "learning_rate": 0.0001990831597454279, "loss": 1.71, "step": 716 }, { "epoch": 0.22, "learning_rate": 0.00019907876392966008, "loss": 1.7895, "step": 717 }, { "epoch": 0.22, "learning_rate": 0.00019907435764991268, "loss": 1.834, "step": 718 }, { "epoch": 0.22, "learning_rate": 0.00019906994090665108, "loss": 1.7273, "step": 719 }, { "epoch": 0.22, "learning_rate": 0.00019906551370034166, "loss": 1.842, "step": 720 }, { "epoch": 0.22, "learning_rate": 0.00019906107603145205, "loss": 1.7982, "step": 721 }, { "epoch": 0.22, "learning_rate": 0.00019905662790045093, "loss": 1.8042, "step": 722 }, { "epoch": 0.22, "learning_rate": 0.000199052169307808, "loss": 1.8883, "step": 723 }, { "epoch": 0.22, "learning_rate": 0.00019904770025399424, "loss": 1.7672, "step": 724 }, { "epoch": 0.22, "learning_rate": 0.00019904322073948164, "loss": 1.8955, "step": 725 }, { "epoch": 0.22, "learning_rate": 0.0001990387307647432, "loss": 1.7783, "step": 726 }, { "epoch": 0.22, "learning_rate": 0.0001990342303302532, "loss": 1.8695, "step": 727 }, { "epoch": 0.22, "learning_rate": 0.0001990297194364869, "loss": 1.8378, "step": 728 }, { "epoch": 0.22, "learning_rate": 0.00019902519808392074, "loss": 1.8402, "step": 729 }, { "epoch": 0.22, "learning_rate": 0.0001990206662730322, "loss": 1.9324, "step": 730 }, { "epoch": 0.22, "learning_rate": 0.00019901612400429993, "loss": 1.9138, "step": 731 }, { "epoch": 0.22, "learning_rate": 0.0001990115712782036, "loss": 1.7806, "step": 732 }, { "epoch": 0.22, "learning_rate": 0.00019900700809522413, "loss": 1.7867, "step": 733 }, { "epoch": 0.22, "learning_rate": 0.00019900243445584336, "loss": 1.8912, "step": 734 }, { "epoch": 0.22, "learning_rate": 0.00019899785036054437, "loss": 1.895, "step": 735 }, { "epoch": 0.22, "learning_rate": 0.00019899325580981128, "loss": 1.8069, "step": 736 }, { "epoch": 0.22, "learning_rate": 0.00019898865080412935, "loss": 1.7626, "step": 737 }, { "epoch": 0.22, "learning_rate": 0.00019898403534398488, "loss": 1.7561, "step": 738 }, { "epoch": 0.22, "learning_rate": 0.00019897940942986537, "loss": 1.885, "step": 739 }, { "epoch": 0.22, "learning_rate": 0.00019897477306225937, "loss": 1.8051, "step": 740 }, { "epoch": 0.22, "learning_rate": 0.00019897012624165653, "loss": 1.8609, "step": 741 }, { "epoch": 0.22, "learning_rate": 0.00019896546896854762, "loss": 1.9436, "step": 742 }, { "epoch": 0.22, "learning_rate": 0.0001989608012434245, "loss": 1.7554, "step": 743 }, { "epoch": 0.22, "learning_rate": 0.00019895612306678015, "loss": 1.8846, "step": 744 }, { "epoch": 0.22, "learning_rate": 0.00019895143443910863, "loss": 1.7345, "step": 745 }, { "epoch": 0.22, "learning_rate": 0.00019894673536090515, "loss": 1.8106, "step": 746 }, { "epoch": 0.22, "learning_rate": 0.00019894202583266594, "loss": 1.8722, "step": 747 }, { "epoch": 0.23, "learning_rate": 0.00019893730585488845, "loss": 1.7495, "step": 748 }, { "epoch": 0.23, "learning_rate": 0.0001989325754280711, "loss": 1.8324, "step": 749 }, { "epoch": 0.23, "learning_rate": 0.00019892783455271356, "loss": 1.8301, "step": 750 }, { "epoch": 0.23, "learning_rate": 0.00019892308322931644, "loss": 1.7488, "step": 751 }, { "epoch": 0.23, "learning_rate": 0.00019891832145838164, "loss": 1.7783, "step": 752 }, { "epoch": 0.23, "learning_rate": 0.00019891354924041194, "loss": 1.8583, "step": 753 }, { "epoch": 0.23, "learning_rate": 0.00019890876657591147, "loss": 1.8491, "step": 754 }, { "epoch": 0.23, "learning_rate": 0.0001989039734653853, "loss": 1.8023, "step": 755 }, { "epoch": 0.23, "learning_rate": 0.00019889916990933958, "loss": 1.8403, "step": 756 }, { "epoch": 0.23, "learning_rate": 0.0001988943559082817, "loss": 1.8687, "step": 757 }, { "epoch": 0.23, "learning_rate": 0.00019888953146272008, "loss": 1.7874, "step": 758 }, { "epoch": 0.23, "learning_rate": 0.0001988846965731642, "loss": 1.843, "step": 759 }, { "epoch": 0.23, "learning_rate": 0.00019887985124012473, "loss": 1.7282, "step": 760 }, { "epoch": 0.23, "learning_rate": 0.00019887499546411336, "loss": 1.7003, "step": 761 }, { "epoch": 0.23, "learning_rate": 0.00019887012924564291, "loss": 1.8387, "step": 762 }, { "epoch": 0.23, "learning_rate": 0.0001988652525852274, "loss": 1.7537, "step": 763 }, { "epoch": 0.23, "learning_rate": 0.00019886036548338174, "loss": 1.8271, "step": 764 }, { "epoch": 0.23, "learning_rate": 0.00019885546794062218, "loss": 1.9441, "step": 765 }, { "epoch": 0.23, "learning_rate": 0.00019885055995746593, "loss": 1.7441, "step": 766 }, { "epoch": 0.23, "learning_rate": 0.0001988456415344313, "loss": 1.814, "step": 767 }, { "epoch": 0.23, "learning_rate": 0.0001988407126720378, "loss": 1.791, "step": 768 }, { "epoch": 0.23, "learning_rate": 0.00019883577337080594, "loss": 1.8742, "step": 769 }, { "epoch": 0.23, "learning_rate": 0.00019883082363125735, "loss": 1.8233, "step": 770 }, { "epoch": 0.23, "learning_rate": 0.00019882586345391486, "loss": 1.8322, "step": 771 }, { "epoch": 0.23, "learning_rate": 0.00019882089283930224, "loss": 1.8236, "step": 772 }, { "epoch": 0.23, "learning_rate": 0.00019881591178794452, "loss": 1.8213, "step": 773 }, { "epoch": 0.23, "learning_rate": 0.0001988109203003677, "loss": 1.8098, "step": 774 }, { "epoch": 0.23, "learning_rate": 0.00019880591837709904, "loss": 1.9055, "step": 775 }, { "epoch": 0.23, "learning_rate": 0.00019880090601866673, "loss": 1.9563, "step": 776 }, { "epoch": 0.23, "learning_rate": 0.00019879588322560014, "loss": 2.0159, "step": 777 }, { "epoch": 0.23, "learning_rate": 0.00019879084999842975, "loss": 1.8957, "step": 778 }, { "epoch": 0.23, "learning_rate": 0.00019878580633768717, "loss": 1.8648, "step": 779 }, { "epoch": 0.23, "learning_rate": 0.00019878075224390504, "loss": 1.7644, "step": 780 }, { "epoch": 0.24, "learning_rate": 0.00019877568771761711, "loss": 1.8118, "step": 781 }, { "epoch": 0.24, "learning_rate": 0.00019877061275935833, "loss": 1.8144, "step": 782 }, { "epoch": 0.24, "learning_rate": 0.0001987655273696646, "loss": 1.7577, "step": 783 }, { "epoch": 0.24, "learning_rate": 0.0001987604315490731, "loss": 1.7831, "step": 784 }, { "epoch": 0.24, "learning_rate": 0.0001987553252981219, "loss": 1.8303, "step": 785 }, { "epoch": 0.24, "learning_rate": 0.00019875020861735037, "loss": 1.9021, "step": 786 }, { "epoch": 0.24, "learning_rate": 0.00019874508150729886, "loss": 1.7545, "step": 787 }, { "epoch": 0.24, "learning_rate": 0.00019873994396850888, "loss": 1.8221, "step": 788 }, { "epoch": 0.24, "learning_rate": 0.000198734796001523, "loss": 1.9445, "step": 789 }, { "epoch": 0.24, "learning_rate": 0.0001987296376068849, "loss": 1.8458, "step": 790 }, { "epoch": 0.24, "learning_rate": 0.0001987244687851394, "loss": 1.8728, "step": 791 }, { "epoch": 0.24, "learning_rate": 0.0001987192895368324, "loss": 1.7931, "step": 792 }, { "epoch": 0.24, "learning_rate": 0.00019871409986251085, "loss": 1.8384, "step": 793 }, { "epoch": 0.24, "learning_rate": 0.0001987088997627229, "loss": 1.814, "step": 794 }, { "epoch": 0.24, "learning_rate": 0.00019870368923801768, "loss": 1.9119, "step": 795 }, { "epoch": 0.24, "learning_rate": 0.00019869846828894554, "loss": 1.8783, "step": 796 }, { "epoch": 0.24, "learning_rate": 0.00019869323691605787, "loss": 1.8841, "step": 797 }, { "epoch": 0.24, "learning_rate": 0.00019868799511990718, "loss": 1.7807, "step": 798 }, { "epoch": 0.24, "learning_rate": 0.00019868274290104701, "loss": 1.9587, "step": 799 }, { "epoch": 0.24, "learning_rate": 0.00019867748026003218, "loss": 1.7877, "step": 800 }, { "epoch": 0.24, "learning_rate": 0.0001986722071974184, "loss": 1.8304, "step": 801 }, { "epoch": 0.24, "learning_rate": 0.00019866692371376256, "loss": 1.7357, "step": 802 }, { "epoch": 0.24, "learning_rate": 0.0001986616298096227, "loss": 1.8624, "step": 803 }, { "epoch": 0.24, "learning_rate": 0.00019865632548555793, "loss": 1.8807, "step": 804 }, { "epoch": 0.24, "learning_rate": 0.00019865101074212845, "loss": 1.9188, "step": 805 }, { "epoch": 0.24, "learning_rate": 0.00019864568557989554, "loss": 1.9265, "step": 806 }, { "epoch": 0.24, "learning_rate": 0.00019864034999942163, "loss": 1.8319, "step": 807 }, { "epoch": 0.24, "learning_rate": 0.0001986350040012702, "loss": 1.8086, "step": 808 }, { "epoch": 0.24, "learning_rate": 0.0001986296475860059, "loss": 1.8085, "step": 809 }, { "epoch": 0.24, "learning_rate": 0.0001986242807541944, "loss": 1.8223, "step": 810 }, { "epoch": 0.24, "learning_rate": 0.00019861890350640252, "loss": 1.7653, "step": 811 }, { "epoch": 0.24, "learning_rate": 0.00019861351584319814, "loss": 1.8314, "step": 812 }, { "epoch": 0.24, "learning_rate": 0.00019860811776515033, "loss": 1.7971, "step": 813 }, { "epoch": 0.25, "learning_rate": 0.0001986027092728291, "loss": 1.8532, "step": 814 }, { "epoch": 0.25, "learning_rate": 0.00019859729036680572, "loss": 1.9675, "step": 815 }, { "epoch": 0.25, "learning_rate": 0.0001985918610476525, "loss": 1.8849, "step": 816 }, { "epoch": 0.25, "learning_rate": 0.00019858642131594283, "loss": 1.8501, "step": 817 }, { "epoch": 0.25, "learning_rate": 0.00019858097117225122, "loss": 1.6848, "step": 818 }, { "epoch": 0.25, "learning_rate": 0.00019857551061715326, "loss": 1.8675, "step": 819 }, { "epoch": 0.25, "learning_rate": 0.00019857003965122565, "loss": 1.7603, "step": 820 }, { "epoch": 0.25, "learning_rate": 0.0001985645582750462, "loss": 1.7384, "step": 821 }, { "epoch": 0.25, "learning_rate": 0.00019855906648919382, "loss": 1.8629, "step": 822 }, { "epoch": 0.25, "learning_rate": 0.00019855356429424853, "loss": 1.792, "step": 823 }, { "epoch": 0.25, "learning_rate": 0.00019854805169079143, "loss": 1.8729, "step": 824 }, { "epoch": 0.25, "learning_rate": 0.00019854252867940465, "loss": 1.7376, "step": 825 }, { "epoch": 0.25, "learning_rate": 0.0001985369952606716, "loss": 1.8253, "step": 826 }, { "epoch": 0.25, "learning_rate": 0.0001985314514351766, "loss": 1.9395, "step": 827 }, { "epoch": 0.25, "learning_rate": 0.00019852589720350516, "loss": 1.9216, "step": 828 }, { "epoch": 0.25, "learning_rate": 0.00019852033256624393, "loss": 1.9148, "step": 829 }, { "epoch": 0.25, "learning_rate": 0.00019851475752398053, "loss": 1.7641, "step": 830 }, { "epoch": 0.25, "learning_rate": 0.00019850917207730383, "loss": 1.7958, "step": 831 }, { "epoch": 0.25, "learning_rate": 0.00019850357622680365, "loss": 1.8547, "step": 832 }, { "epoch": 0.25, "learning_rate": 0.00019849796997307103, "loss": 1.7768, "step": 833 }, { "epoch": 0.25, "learning_rate": 0.00019849235331669807, "loss": 1.9942, "step": 834 }, { "epoch": 0.25, "learning_rate": 0.0001984867262582779, "loss": 1.6718, "step": 835 }, { "epoch": 0.25, "learning_rate": 0.0001984810887984049, "loss": 1.9298, "step": 836 }, { "epoch": 0.25, "learning_rate": 0.00019847544093767438, "loss": 1.8349, "step": 837 }, { "epoch": 0.25, "learning_rate": 0.00019846978267668285, "loss": 1.8368, "step": 838 }, { "epoch": 0.25, "learning_rate": 0.00019846411401602793, "loss": 1.845, "step": 839 }, { "epoch": 0.25, "learning_rate": 0.00019845843495630826, "loss": 1.8275, "step": 840 }, { "epoch": 0.25, "learning_rate": 0.00019845274549812363, "loss": 2.007, "step": 841 }, { "epoch": 0.25, "learning_rate": 0.00019844704564207492, "loss": 1.8235, "step": 842 }, { "epoch": 0.25, "learning_rate": 0.00019844133538876412, "loss": 1.8037, "step": 843 }, { "epoch": 0.25, "learning_rate": 0.00019843561473879427, "loss": 1.8564, "step": 844 }, { "epoch": 0.25, "learning_rate": 0.00019842988369276962, "loss": 1.8006, "step": 845 }, { "epoch": 0.25, "learning_rate": 0.00019842414225129535, "loss": 1.9377, "step": 846 }, { "epoch": 0.25, "learning_rate": 0.0001984183904149779, "loss": 1.9535, "step": 847 }, { "epoch": 0.26, "learning_rate": 0.00019841262818442464, "loss": 1.7829, "step": 848 }, { "epoch": 0.26, "learning_rate": 0.00019840685556024427, "loss": 1.825, "step": 849 }, { "epoch": 0.26, "learning_rate": 0.00019840107254304637, "loss": 1.82, "step": 850 }, { "epoch": 0.26, "learning_rate": 0.0001983952791334417, "loss": 1.7204, "step": 851 }, { "epoch": 0.26, "learning_rate": 0.00019838947533204216, "loss": 1.8411, "step": 852 }, { "epoch": 0.26, "learning_rate": 0.00019838366113946064, "loss": 1.7896, "step": 853 }, { "epoch": 0.26, "learning_rate": 0.00019837783655631127, "loss": 1.818, "step": 854 }, { "epoch": 0.26, "learning_rate": 0.0001983720015832091, "loss": 1.8371, "step": 855 }, { "epoch": 0.26, "learning_rate": 0.0001983661562207705, "loss": 1.7687, "step": 856 }, { "epoch": 0.26, "learning_rate": 0.00019836030046961275, "loss": 1.8726, "step": 857 }, { "epoch": 0.26, "learning_rate": 0.00019835443433035427, "loss": 1.7622, "step": 858 }, { "epoch": 0.26, "learning_rate": 0.00019834855780361465, "loss": 1.8902, "step": 859 }, { "epoch": 0.26, "learning_rate": 0.00019834267089001447, "loss": 1.7916, "step": 860 }, { "epoch": 0.26, "learning_rate": 0.0001983367735901755, "loss": 1.7558, "step": 861 }, { "epoch": 0.26, "learning_rate": 0.00019833086590472057, "loss": 1.893, "step": 862 }, { "epoch": 0.26, "learning_rate": 0.00019832494783427365, "loss": 1.8183, "step": 863 }, { "epoch": 0.26, "learning_rate": 0.00019831901937945965, "loss": 1.8816, "step": 864 }, { "epoch": 0.26, "learning_rate": 0.00019831308054090477, "loss": 1.7919, "step": 865 }, { "epoch": 0.26, "learning_rate": 0.00019830713131923622, "loss": 1.867, "step": 866 }, { "epoch": 0.26, "learning_rate": 0.0001983011717150823, "loss": 1.8023, "step": 867 }, { "epoch": 0.26, "learning_rate": 0.0001982952017290724, "loss": 1.761, "step": 868 }, { "epoch": 0.26, "learning_rate": 0.00019828922136183708, "loss": 1.7543, "step": 869 }, { "epoch": 0.26, "learning_rate": 0.00019828323061400792, "loss": 1.8515, "step": 870 }, { "epoch": 0.26, "learning_rate": 0.00019827722948621759, "loss": 1.8595, "step": 871 }, { "epoch": 0.26, "learning_rate": 0.00019827121797909994, "loss": 1.8662, "step": 872 }, { "epoch": 0.26, "learning_rate": 0.00019826519609328979, "loss": 1.7866, "step": 873 }, { "epoch": 0.26, "learning_rate": 0.0001982591638294232, "loss": 1.8466, "step": 874 }, { "epoch": 0.26, "learning_rate": 0.00019825312118813723, "loss": 1.8294, "step": 875 }, { "epoch": 0.26, "learning_rate": 0.00019824706817007004, "loss": 2.0107, "step": 876 }, { "epoch": 0.26, "learning_rate": 0.0001982410047758609, "loss": 1.7941, "step": 877 }, { "epoch": 0.26, "learning_rate": 0.00019823493100615022, "loss": 1.7673, "step": 878 }, { "epoch": 0.26, "learning_rate": 0.00019822884686157945, "loss": 1.9115, "step": 879 }, { "epoch": 0.26, "learning_rate": 0.00019822275234279116, "loss": 1.9452, "step": 880 }, { "epoch": 0.27, "learning_rate": 0.000198216647450429, "loss": 1.7674, "step": 881 }, { "epoch": 0.27, "learning_rate": 0.0001982105321851377, "loss": 1.9462, "step": 882 }, { "epoch": 0.27, "learning_rate": 0.00019820440654756314, "loss": 1.7971, "step": 883 }, { "epoch": 0.27, "learning_rate": 0.00019819827053835226, "loss": 1.8525, "step": 884 }, { "epoch": 0.27, "learning_rate": 0.0001981921241581531, "loss": 1.8433, "step": 885 }, { "epoch": 0.27, "learning_rate": 0.0001981859674076148, "loss": 1.9846, "step": 886 }, { "epoch": 0.27, "learning_rate": 0.0001981798002873876, "loss": 1.882, "step": 887 }, { "epoch": 0.27, "learning_rate": 0.00019817362279812278, "loss": 2.0419, "step": 888 }, { "epoch": 0.27, "learning_rate": 0.00019816743494047284, "loss": 2.0131, "step": 889 }, { "epoch": 0.27, "learning_rate": 0.00019816123671509122, "loss": 1.8259, "step": 890 }, { "epoch": 0.27, "learning_rate": 0.00019815502812263258, "loss": 1.8553, "step": 891 }, { "epoch": 0.27, "learning_rate": 0.0001981488091637526, "loss": 1.874, "step": 892 }, { "epoch": 0.27, "learning_rate": 0.0001981425798391081, "loss": 1.866, "step": 893 }, { "epoch": 0.27, "learning_rate": 0.00019813634014935697, "loss": 1.8438, "step": 894 }, { "epoch": 0.27, "learning_rate": 0.0001981300900951582, "loss": 1.9885, "step": 895 }, { "epoch": 0.27, "learning_rate": 0.00019812382967717185, "loss": 1.9083, "step": 896 }, { "epoch": 0.27, "learning_rate": 0.00019811755889605914, "loss": 1.7932, "step": 897 }, { "epoch": 0.27, "learning_rate": 0.00019811127775248234, "loss": 1.7632, "step": 898 }, { "epoch": 0.27, "learning_rate": 0.00019810498624710476, "loss": 1.8367, "step": 899 }, { "epoch": 0.27, "learning_rate": 0.00019809868438059095, "loss": 1.8161, "step": 900 }, { "epoch": 0.27, "learning_rate": 0.0001980923721536064, "loss": 1.8451, "step": 901 }, { "epoch": 0.27, "learning_rate": 0.00019808604956681784, "loss": 1.8463, "step": 902 }, { "epoch": 0.27, "learning_rate": 0.00019807971662089293, "loss": 1.8999, "step": 903 }, { "epoch": 0.27, "learning_rate": 0.00019807337331650054, "loss": 1.8215, "step": 904 }, { "epoch": 0.27, "learning_rate": 0.00019806701965431064, "loss": 1.8384, "step": 905 }, { "epoch": 0.27, "learning_rate": 0.00019806065563499417, "loss": 1.8482, "step": 906 }, { "epoch": 0.27, "learning_rate": 0.00019805428125922335, "loss": 1.7946, "step": 907 }, { "epoch": 0.27, "learning_rate": 0.00019804789652767133, "loss": 1.8517, "step": 908 }, { "epoch": 0.27, "learning_rate": 0.00019804150144101245, "loss": 1.7755, "step": 909 }, { "epoch": 0.27, "learning_rate": 0.00019803509599992212, "loss": 1.9497, "step": 910 }, { "epoch": 0.27, "learning_rate": 0.00019802868020507678, "loss": 1.8028, "step": 911 }, { "epoch": 0.27, "learning_rate": 0.0001980222540571541, "loss": 1.7467, "step": 912 }, { "epoch": 0.27, "learning_rate": 0.0001980158175568327, "loss": 1.9568, "step": 913 }, { "epoch": 0.28, "learning_rate": 0.0001980093707047924, "loss": 1.892, "step": 914 }, { "epoch": 0.28, "learning_rate": 0.00019800291350171402, "loss": 1.9232, "step": 915 }, { "epoch": 0.28, "learning_rate": 0.00019799644594827959, "loss": 1.7255, "step": 916 }, { "epoch": 0.28, "learning_rate": 0.0001979899680451721, "loss": 1.8416, "step": 917 }, { "epoch": 0.28, "learning_rate": 0.0001979834797930757, "loss": 1.7653, "step": 918 }, { "epoch": 0.28, "learning_rate": 0.0001979769811926757, "loss": 1.955, "step": 919 }, { "epoch": 0.28, "learning_rate": 0.0001979704722446584, "loss": 1.8065, "step": 920 }, { "epoch": 0.28, "learning_rate": 0.0001979639529497112, "loss": 1.7784, "step": 921 }, { "epoch": 0.28, "learning_rate": 0.00019795742330852264, "loss": 1.7779, "step": 922 }, { "epoch": 0.28, "learning_rate": 0.00019795088332178232, "loss": 1.7903, "step": 923 }, { "epoch": 0.28, "learning_rate": 0.00019794433299018103, "loss": 1.8915, "step": 924 }, { "epoch": 0.28, "learning_rate": 0.00019793777231441045, "loss": 1.9224, "step": 925 }, { "epoch": 0.28, "learning_rate": 0.00019793120129516352, "loss": 1.8174, "step": 926 }, { "epoch": 0.28, "learning_rate": 0.00019792461993313422, "loss": 1.8885, "step": 927 }, { "epoch": 0.28, "learning_rate": 0.00019791802822901767, "loss": 1.885, "step": 928 }, { "epoch": 0.28, "learning_rate": 0.00019791142618350995, "loss": 1.872, "step": 929 }, { "epoch": 0.28, "learning_rate": 0.0001979048137973084, "loss": 1.7446, "step": 930 }, { "epoch": 0.28, "learning_rate": 0.0001978981910711113, "loss": 1.8166, "step": 931 }, { "epoch": 0.28, "learning_rate": 0.00019789155800561817, "loss": 1.8923, "step": 932 }, { "epoch": 0.28, "learning_rate": 0.00019788491460152953, "loss": 1.8873, "step": 933 }, { "epoch": 0.28, "learning_rate": 0.00019787826085954695, "loss": 1.765, "step": 934 }, { "epoch": 0.28, "learning_rate": 0.0001978715967803732, "loss": 1.8735, "step": 935 }, { "epoch": 0.28, "learning_rate": 0.00019786492236471208, "loss": 1.8816, "step": 936 }, { "epoch": 0.28, "learning_rate": 0.0001978582376132685, "loss": 1.8491, "step": 937 }, { "epoch": 0.28, "learning_rate": 0.00019785154252674843, "loss": 1.7925, "step": 938 }, { "epoch": 0.28, "learning_rate": 0.000197844837105859, "loss": 1.788, "step": 939 }, { "epoch": 0.28, "learning_rate": 0.00019783812135130835, "loss": 1.9327, "step": 940 }, { "epoch": 0.28, "learning_rate": 0.00019783139526380577, "loss": 1.7696, "step": 941 }, { "epoch": 0.28, "learning_rate": 0.00019782465884406162, "loss": 1.7996, "step": 942 }, { "epoch": 0.28, "learning_rate": 0.00019781791209278732, "loss": 1.8387, "step": 943 }, { "epoch": 0.28, "learning_rate": 0.00019781115501069544, "loss": 1.819, "step": 944 }, { "epoch": 0.28, "learning_rate": 0.00019780438759849962, "loss": 1.9302, "step": 945 }, { "epoch": 0.28, "learning_rate": 0.00019779760985691458, "loss": 1.7858, "step": 946 }, { "epoch": 0.29, "learning_rate": 0.00019779082178665613, "loss": 1.8839, "step": 947 }, { "epoch": 0.29, "learning_rate": 0.00019778402338844117, "loss": 1.7518, "step": 948 }, { "epoch": 0.29, "learning_rate": 0.00019777721466298772, "loss": 1.815, "step": 949 }, { "epoch": 0.29, "learning_rate": 0.00019777039561101484, "loss": 1.7994, "step": 950 }, { "epoch": 0.29, "learning_rate": 0.00019776356623324275, "loss": 1.7638, "step": 951 }, { "epoch": 0.29, "learning_rate": 0.00019775672653039265, "loss": 1.879, "step": 952 }, { "epoch": 0.29, "learning_rate": 0.00019774987650318697, "loss": 1.8159, "step": 953 }, { "epoch": 0.29, "learning_rate": 0.00019774301615234913, "loss": 1.8474, "step": 954 }, { "epoch": 0.29, "learning_rate": 0.00019773614547860367, "loss": 1.8766, "step": 955 }, { "epoch": 0.29, "learning_rate": 0.00019772926448267622, "loss": 1.8026, "step": 956 }, { "epoch": 0.29, "learning_rate": 0.0001977223731652935, "loss": 1.7934, "step": 957 }, { "epoch": 0.29, "learning_rate": 0.00019771547152718335, "loss": 1.835, "step": 958 }, { "epoch": 0.29, "learning_rate": 0.00019770855956907464, "loss": 1.8516, "step": 959 }, { "epoch": 0.29, "learning_rate": 0.00019770163729169733, "loss": 1.8532, "step": 960 }, { "epoch": 0.29, "learning_rate": 0.00019769470469578257, "loss": 1.8324, "step": 961 }, { "epoch": 0.29, "learning_rate": 0.0001976877617820625, "loss": 1.7201, "step": 962 }, { "epoch": 0.29, "learning_rate": 0.0001976808085512704, "loss": 1.8798, "step": 963 }, { "epoch": 0.29, "learning_rate": 0.00019767384500414051, "loss": 1.8888, "step": 964 }, { "epoch": 0.29, "learning_rate": 0.00019766687114140843, "loss": 1.8618, "step": 965 }, { "epoch": 0.29, "learning_rate": 0.00019765988696381062, "loss": 1.8068, "step": 966 }, { "epoch": 0.29, "learning_rate": 0.00019765289247208466, "loss": 1.8174, "step": 967 }, { "epoch": 0.29, "learning_rate": 0.0001976458876669693, "loss": 1.9252, "step": 968 }, { "epoch": 0.29, "learning_rate": 0.00019763887254920434, "loss": 1.9152, "step": 969 }, { "epoch": 0.29, "learning_rate": 0.00019763184711953064, "loss": 1.8446, "step": 970 }, { "epoch": 0.29, "learning_rate": 0.0001976248113786902, "loss": 1.7944, "step": 971 }, { "epoch": 0.29, "learning_rate": 0.00019761776532742606, "loss": 1.902, "step": 972 }, { "epoch": 0.29, "learning_rate": 0.0001976107089664824, "loss": 1.9156, "step": 973 }, { "epoch": 0.29, "learning_rate": 0.00019760364229660444, "loss": 1.8061, "step": 974 }, { "epoch": 0.29, "learning_rate": 0.0001975965653185385, "loss": 1.8324, "step": 975 }, { "epoch": 0.29, "learning_rate": 0.00019758947803303204, "loss": 1.8119, "step": 976 }, { "epoch": 0.29, "learning_rate": 0.00019758238044083356, "loss": 1.8685, "step": 977 }, { "epoch": 0.29, "learning_rate": 0.00019757527254269262, "loss": 1.8619, "step": 978 }, { "epoch": 0.29, "learning_rate": 0.0001975681543393599, "loss": 1.8517, "step": 979 }, { "epoch": 0.3, "learning_rate": 0.00019756102583158724, "loss": 1.838, "step": 980 }, { "epoch": 0.3, "learning_rate": 0.00019755388702012743, "loss": 1.9169, "step": 981 }, { "epoch": 0.3, "learning_rate": 0.00019754673790573445, "loss": 1.82, "step": 982 }, { "epoch": 0.3, "learning_rate": 0.00019753957848916336, "loss": 1.9478, "step": 983 }, { "epoch": 0.3, "learning_rate": 0.00019753240877117025, "loss": 1.8417, "step": 984 }, { "epoch": 0.3, "learning_rate": 0.0001975252287525123, "loss": 1.9343, "step": 985 }, { "epoch": 0.3, "learning_rate": 0.00019751803843394793, "loss": 1.8674, "step": 986 }, { "epoch": 0.3, "learning_rate": 0.00019751083781623642, "loss": 1.9035, "step": 987 }, { "epoch": 0.3, "learning_rate": 0.0001975036269001383, "loss": 1.9534, "step": 988 }, { "epoch": 0.3, "learning_rate": 0.00019749640568641507, "loss": 1.9795, "step": 989 }, { "epoch": 0.3, "learning_rate": 0.00019748917417582949, "loss": 1.7596, "step": 990 }, { "epoch": 0.3, "learning_rate": 0.00019748193236914522, "loss": 1.8763, "step": 991 }, { "epoch": 0.3, "learning_rate": 0.00019747468026712705, "loss": 1.7822, "step": 992 }, { "epoch": 0.3, "learning_rate": 0.000197467417870541, "loss": 1.8279, "step": 993 }, { "epoch": 0.3, "learning_rate": 0.00019746014518015403, "loss": 1.8327, "step": 994 }, { "epoch": 0.3, "learning_rate": 0.0001974528621967342, "loss": 1.7789, "step": 995 }, { "epoch": 0.3, "learning_rate": 0.00019744556892105074, "loss": 1.8136, "step": 996 }, { "epoch": 0.3, "learning_rate": 0.00019743826535387383, "loss": 1.8923, "step": 997 }, { "epoch": 0.3, "learning_rate": 0.0001974309514959749, "loss": 1.8121, "step": 998 }, { "epoch": 0.3, "learning_rate": 0.00019742362734812637, "loss": 1.8512, "step": 999 }, { "epoch": 0.3, "learning_rate": 0.00019741629291110174, "loss": 1.8836, "step": 1000 }, { "epoch": 0.3, "learning_rate": 0.00019740894818567558, "loss": 1.849, "step": 1001 }, { "epoch": 0.3, "learning_rate": 0.00019740159317262367, "loss": 1.8193, "step": 1002 }, { "epoch": 0.3, "learning_rate": 0.00019739422787272277, "loss": 1.7996, "step": 1003 }, { "epoch": 0.3, "learning_rate": 0.00019738685228675075, "loss": 1.8368, "step": 1004 }, { "epoch": 0.3, "learning_rate": 0.00019737946641548652, "loss": 1.8481, "step": 1005 }, { "epoch": 0.3, "learning_rate": 0.00019737207025971015, "loss": 1.8837, "step": 1006 }, { "epoch": 0.3, "learning_rate": 0.00019736466382020283, "loss": 1.8392, "step": 1007 }, { "epoch": 0.3, "learning_rate": 0.00019735724709774667, "loss": 1.8455, "step": 1008 }, { "epoch": 0.3, "learning_rate": 0.00019734982009312504, "loss": 1.7565, "step": 1009 }, { "epoch": 0.3, "learning_rate": 0.0001973423828071223, "loss": 1.8352, "step": 1010 }, { "epoch": 0.3, "learning_rate": 0.0001973349352405239, "loss": 1.8386, "step": 1011 }, { "epoch": 0.3, "learning_rate": 0.00019732747739411644, "loss": 1.8915, "step": 1012 }, { "epoch": 0.3, "learning_rate": 0.00019732000926868758, "loss": 1.9009, "step": 1013 }, { "epoch": 0.31, "learning_rate": 0.000197312530865026, "loss": 1.8879, "step": 1014 }, { "epoch": 0.31, "learning_rate": 0.00019730504218392155, "loss": 1.9176, "step": 1015 }, { "epoch": 0.31, "learning_rate": 0.00019729754322616506, "loss": 1.9858, "step": 1016 }, { "epoch": 0.31, "learning_rate": 0.0001972900339925486, "loss": 1.8796, "step": 1017 }, { "epoch": 0.31, "learning_rate": 0.00019728251448386525, "loss": 1.9197, "step": 1018 }, { "epoch": 0.31, "learning_rate": 0.00019727498470090905, "loss": 1.8377, "step": 1019 }, { "epoch": 0.31, "learning_rate": 0.00019726744464447537, "loss": 1.848, "step": 1020 }, { "epoch": 0.31, "learning_rate": 0.00019725989431536049, "loss": 1.8193, "step": 1021 }, { "epoch": 0.31, "learning_rate": 0.00019725233371436176, "loss": 1.9491, "step": 1022 }, { "epoch": 0.31, "learning_rate": 0.00019724476284227775, "loss": 1.8728, "step": 1023 }, { "epoch": 0.31, "learning_rate": 0.00019723718169990803, "loss": 1.9022, "step": 1024 }, { "epoch": 0.31, "learning_rate": 0.00019722959028805326, "loss": 1.829, "step": 1025 }, { "epoch": 0.31, "learning_rate": 0.00019722198860751516, "loss": 1.8047, "step": 1026 }, { "epoch": 0.31, "learning_rate": 0.0001972143766590966, "loss": 1.8804, "step": 1027 }, { "epoch": 0.31, "learning_rate": 0.0001972067544436015, "loss": 1.7991, "step": 1028 }, { "epoch": 0.31, "learning_rate": 0.00019719912196183484, "loss": 1.8189, "step": 1029 }, { "epoch": 0.31, "learning_rate": 0.0001971914792146027, "loss": 1.8306, "step": 1030 }, { "epoch": 0.31, "learning_rate": 0.00019718382620271228, "loss": 1.951, "step": 1031 }, { "epoch": 0.31, "learning_rate": 0.00019717616292697185, "loss": 1.8483, "step": 1032 }, { "epoch": 0.31, "learning_rate": 0.0001971684893881907, "loss": 1.7319, "step": 1033 }, { "epoch": 0.31, "learning_rate": 0.00019716080558717928, "loss": 1.7382, "step": 1034 }, { "epoch": 0.31, "learning_rate": 0.00019715311152474909, "loss": 1.8145, "step": 1035 }, { "epoch": 0.31, "learning_rate": 0.00019714540720171275, "loss": 1.8237, "step": 1036 }, { "epoch": 0.31, "learning_rate": 0.00019713769261888388, "loss": 1.8482, "step": 1037 }, { "epoch": 0.31, "learning_rate": 0.0001971299677770773, "loss": 1.884, "step": 1038 }, { "epoch": 0.31, "learning_rate": 0.0001971222326771088, "loss": 1.9409, "step": 1039 }, { "epoch": 0.31, "learning_rate": 0.00019711448731979533, "loss": 1.8708, "step": 1040 }, { "epoch": 0.31, "learning_rate": 0.0001971067317059549, "loss": 1.8434, "step": 1041 }, { "epoch": 0.31, "learning_rate": 0.00019709896583640658, "loss": 1.7976, "step": 1042 }, { "epoch": 0.31, "learning_rate": 0.0001970911897119706, "loss": 1.8304, "step": 1043 }, { "epoch": 0.31, "learning_rate": 0.00019708340333346813, "loss": 1.7727, "step": 1044 }, { "epoch": 0.31, "learning_rate": 0.00019707560670172156, "loss": 1.8947, "step": 1045 }, { "epoch": 0.31, "learning_rate": 0.00019706779981755435, "loss": 1.8737, "step": 1046 }, { "epoch": 0.32, "learning_rate": 0.00019705998268179095, "loss": 1.8887, "step": 1047 }, { "epoch": 0.32, "learning_rate": 0.00019705215529525693, "loss": 1.8488, "step": 1048 }, { "epoch": 0.32, "learning_rate": 0.00019704431765877906, "loss": 1.7653, "step": 1049 }, { "epoch": 0.32, "learning_rate": 0.00019703646977318502, "loss": 1.8629, "step": 1050 }, { "epoch": 0.32, "learning_rate": 0.00019702861163930363, "loss": 1.9628, "step": 1051 }, { "epoch": 0.32, "learning_rate": 0.00019702074325796487, "loss": 1.9057, "step": 1052 }, { "epoch": 0.32, "learning_rate": 0.0001970128646299997, "loss": 1.8824, "step": 1053 }, { "epoch": 0.32, "learning_rate": 0.0001970049757562402, "loss": 1.8548, "step": 1054 }, { "epoch": 0.32, "learning_rate": 0.0001969970766375196, "loss": 1.7995, "step": 1055 }, { "epoch": 0.32, "learning_rate": 0.00019698916727467204, "loss": 1.7643, "step": 1056 }, { "epoch": 0.32, "learning_rate": 0.00019698124766853292, "loss": 1.8752, "step": 1057 }, { "epoch": 0.32, "learning_rate": 0.00019697331781993866, "loss": 1.8479, "step": 1058 }, { "epoch": 0.32, "learning_rate": 0.0001969653777297267, "loss": 1.9367, "step": 1059 }, { "epoch": 0.32, "learning_rate": 0.0001969574273987357, "loss": 1.7961, "step": 1060 }, { "epoch": 0.32, "learning_rate": 0.00019694946682780523, "loss": 1.9976, "step": 1061 }, { "epoch": 0.32, "learning_rate": 0.00019694149601777604, "loss": 1.8748, "step": 1062 }, { "epoch": 0.32, "learning_rate": 0.00019693351496949003, "loss": 1.8901, "step": 1063 }, { "epoch": 0.32, "learning_rate": 0.00019692552368378997, "loss": 1.9113, "step": 1064 }, { "epoch": 0.32, "learning_rate": 0.00019691752216151995, "loss": 1.8356, "step": 1065 }, { "epoch": 0.32, "learning_rate": 0.00019690951040352497, "loss": 1.8297, "step": 1066 }, { "epoch": 0.32, "learning_rate": 0.0001969014884106512, "loss": 1.8704, "step": 1067 }, { "epoch": 0.32, "learning_rate": 0.00019689345618374588, "loss": 1.774, "step": 1068 }, { "epoch": 0.32, "learning_rate": 0.0001968854137236573, "loss": 1.8428, "step": 1069 }, { "epoch": 0.32, "learning_rate": 0.00019687736103123483, "loss": 1.7512, "step": 1070 }, { "epoch": 0.32, "learning_rate": 0.00019686929810732896, "loss": 1.948, "step": 1071 }, { "epoch": 0.32, "learning_rate": 0.00019686122495279123, "loss": 1.7219, "step": 1072 }, { "epoch": 0.32, "learning_rate": 0.00019685314156847426, "loss": 1.9199, "step": 1073 }, { "epoch": 0.32, "learning_rate": 0.00019684504795523176, "loss": 1.8835, "step": 1074 }, { "epoch": 0.32, "learning_rate": 0.00019683694411391854, "loss": 1.9409, "step": 1075 }, { "epoch": 0.32, "learning_rate": 0.00019682883004539043, "loss": 1.7415, "step": 1076 }, { "epoch": 0.32, "learning_rate": 0.0001968207057505044, "loss": 1.7361, "step": 1077 }, { "epoch": 0.32, "learning_rate": 0.0001968125712301185, "loss": 1.7418, "step": 1078 }, { "epoch": 0.32, "learning_rate": 0.0001968044264850918, "loss": 1.7362, "step": 1079 }, { "epoch": 0.33, "learning_rate": 0.00019679627151628453, "loss": 1.8064, "step": 1080 }, { "epoch": 0.33, "learning_rate": 0.00019678810632455792, "loss": 1.9221, "step": 1081 }, { "epoch": 0.33, "learning_rate": 0.00019677993091077434, "loss": 1.8357, "step": 1082 }, { "epoch": 0.33, "learning_rate": 0.0001967717452757972, "loss": 1.8416, "step": 1083 }, { "epoch": 0.33, "learning_rate": 0.00019676354942049104, "loss": 1.8301, "step": 1084 }, { "epoch": 0.33, "learning_rate": 0.00019675534334572144, "loss": 1.7411, "step": 1085 }, { "epoch": 0.33, "learning_rate": 0.000196747127052355, "loss": 1.8146, "step": 1086 }, { "epoch": 0.33, "learning_rate": 0.00019673890054125955, "loss": 1.8195, "step": 1087 }, { "epoch": 0.33, "learning_rate": 0.00019673066381330388, "loss": 1.7747, "step": 1088 }, { "epoch": 0.33, "learning_rate": 0.0001967224168693579, "loss": 1.8178, "step": 1089 }, { "epoch": 0.33, "learning_rate": 0.00019671415971029259, "loss": 1.6635, "step": 1090 }, { "epoch": 0.33, "learning_rate": 0.00019670589233698, "loss": 1.8083, "step": 1091 }, { "epoch": 0.33, "learning_rate": 0.00019669761475029328, "loss": 1.8421, "step": 1092 }, { "epoch": 0.33, "learning_rate": 0.00019668932695110668, "loss": 1.7589, "step": 1093 }, { "epoch": 0.33, "learning_rate": 0.0001966810289402954, "loss": 1.7408, "step": 1094 }, { "epoch": 0.33, "learning_rate": 0.00019667272071873592, "loss": 1.7098, "step": 1095 }, { "epoch": 0.33, "learning_rate": 0.00019666440228730564, "loss": 1.8253, "step": 1096 }, { "epoch": 0.33, "learning_rate": 0.00019665607364688313, "loss": 1.8387, "step": 1097 }, { "epoch": 0.33, "learning_rate": 0.00019664773479834797, "loss": 1.8513, "step": 1098 }, { "epoch": 0.33, "learning_rate": 0.00019663938574258086, "loss": 1.907, "step": 1099 }, { "epoch": 0.33, "learning_rate": 0.00019663102648046356, "loss": 1.8187, "step": 1100 }, { "epoch": 0.33, "learning_rate": 0.0001966226570128789, "loss": 1.8345, "step": 1101 }, { "epoch": 0.33, "learning_rate": 0.00019661427734071082, "loss": 1.8271, "step": 1102 }, { "epoch": 0.33, "learning_rate": 0.00019660588746484435, "loss": 1.9427, "step": 1103 }, { "epoch": 0.33, "learning_rate": 0.0001965974873861655, "loss": 1.8994, "step": 1104 }, { "epoch": 0.33, "learning_rate": 0.0001965890771055615, "loss": 1.7158, "step": 1105 }, { "epoch": 0.33, "learning_rate": 0.0001965806566239205, "loss": 1.8404, "step": 1106 }, { "epoch": 0.33, "learning_rate": 0.0001965722259421319, "loss": 1.813, "step": 1107 }, { "epoch": 0.33, "learning_rate": 0.000196563785061086, "loss": 1.8668, "step": 1108 }, { "epoch": 0.33, "learning_rate": 0.00019655533398167435, "loss": 1.8839, "step": 1109 }, { "epoch": 0.33, "learning_rate": 0.00019654687270478945, "loss": 1.8921, "step": 1110 }, { "epoch": 0.33, "learning_rate": 0.00019653840123132487, "loss": 1.8153, "step": 1111 }, { "epoch": 0.33, "learning_rate": 0.0001965299195621754, "loss": 1.791, "step": 1112 }, { "epoch": 0.34, "learning_rate": 0.00019652142769823673, "loss": 1.8675, "step": 1113 }, { "epoch": 0.34, "learning_rate": 0.00019651292564040577, "loss": 1.7348, "step": 1114 }, { "epoch": 0.34, "learning_rate": 0.00019650441338958042, "loss": 1.8187, "step": 1115 }, { "epoch": 0.34, "learning_rate": 0.00019649589094665966, "loss": 1.7676, "step": 1116 }, { "epoch": 0.34, "learning_rate": 0.00019648735831254363, "loss": 1.7422, "step": 1117 }, { "epoch": 0.34, "learning_rate": 0.0001964788154881334, "loss": 1.8198, "step": 1118 }, { "epoch": 0.34, "learning_rate": 0.0001964702624743313, "loss": 1.8758, "step": 1119 }, { "epoch": 0.34, "learning_rate": 0.00019646169927204054, "loss": 1.8486, "step": 1120 }, { "epoch": 0.34, "learning_rate": 0.00019645312588216558, "loss": 1.8533, "step": 1121 }, { "epoch": 0.34, "learning_rate": 0.0001964445423056118, "loss": 1.8887, "step": 1122 }, { "epoch": 0.34, "learning_rate": 0.00019643594854328584, "loss": 1.8559, "step": 1123 }, { "epoch": 0.34, "learning_rate": 0.00019642734459609522, "loss": 1.8466, "step": 1124 }, { "epoch": 0.34, "learning_rate": 0.00019641873046494866, "loss": 1.9283, "step": 1125 }, { "epoch": 0.34, "learning_rate": 0.0001964101061507559, "loss": 1.8278, "step": 1126 }, { "epoch": 0.34, "learning_rate": 0.00019640147165442783, "loss": 1.879, "step": 1127 }, { "epoch": 0.34, "learning_rate": 0.00019639282697687634, "loss": 1.8359, "step": 1128 }, { "epoch": 0.34, "learning_rate": 0.00019638417211901438, "loss": 1.904, "step": 1129 }, { "epoch": 0.34, "learning_rate": 0.00019637550708175608, "loss": 1.8457, "step": 1130 }, { "epoch": 0.34, "learning_rate": 0.00019636683186601648, "loss": 1.8295, "step": 1131 }, { "epoch": 0.34, "learning_rate": 0.0001963581464727119, "loss": 1.8957, "step": 1132 }, { "epoch": 0.34, "learning_rate": 0.00019634945090275956, "loss": 1.7494, "step": 1133 }, { "epoch": 0.34, "learning_rate": 0.00019634074515707787, "loss": 1.9214, "step": 1134 }, { "epoch": 0.34, "learning_rate": 0.00019633202923658623, "loss": 1.886, "step": 1135 }, { "epoch": 0.34, "learning_rate": 0.00019632330314220514, "loss": 1.8542, "step": 1136 }, { "epoch": 0.34, "learning_rate": 0.00019631456687485625, "loss": 1.8228, "step": 1137 }, { "epoch": 0.34, "learning_rate": 0.00019630582043546216, "loss": 1.9238, "step": 1138 }, { "epoch": 0.34, "learning_rate": 0.00019629706382494666, "loss": 1.8921, "step": 1139 }, { "epoch": 0.34, "learning_rate": 0.0001962882970442345, "loss": 1.7749, "step": 1140 }, { "epoch": 0.34, "learning_rate": 0.0001962795200942516, "loss": 1.9584, "step": 1141 }, { "epoch": 0.34, "learning_rate": 0.0001962707329759249, "loss": 1.8544, "step": 1142 }, { "epoch": 0.34, "learning_rate": 0.00019626193569018246, "loss": 1.808, "step": 1143 }, { "epoch": 0.34, "learning_rate": 0.00019625312823795334, "loss": 1.8478, "step": 1144 }, { "epoch": 0.34, "learning_rate": 0.00019624431062016776, "loss": 1.8911, "step": 1145 }, { "epoch": 0.34, "learning_rate": 0.00019623548283775698, "loss": 1.9703, "step": 1146 }, { "epoch": 0.35, "learning_rate": 0.00019622664489165326, "loss": 1.7779, "step": 1147 }, { "epoch": 0.35, "learning_rate": 0.00019621779678279005, "loss": 1.9172, "step": 1148 }, { "epoch": 0.35, "learning_rate": 0.00019620893851210184, "loss": 1.9177, "step": 1149 }, { "epoch": 0.35, "learning_rate": 0.00019620007008052417, "loss": 1.875, "step": 1150 }, { "epoch": 0.35, "learning_rate": 0.0001961911914889936, "loss": 1.7441, "step": 1151 }, { "epoch": 0.35, "learning_rate": 0.00019618230273844788, "loss": 1.8532, "step": 1152 }, { "epoch": 0.35, "learning_rate": 0.0001961734038298258, "loss": 1.8136, "step": 1153 }, { "epoch": 0.35, "learning_rate": 0.00019616449476406712, "loss": 1.7502, "step": 1154 }, { "epoch": 0.35, "learning_rate": 0.00019615557554211277, "loss": 1.9379, "step": 1155 }, { "epoch": 0.35, "learning_rate": 0.00019614664616490477, "loss": 1.8762, "step": 1156 }, { "epoch": 0.35, "learning_rate": 0.00019613770663338617, "loss": 1.8346, "step": 1157 }, { "epoch": 0.35, "learning_rate": 0.0001961287569485011, "loss": 1.862, "step": 1158 }, { "epoch": 0.35, "learning_rate": 0.00019611979711119473, "loss": 1.8429, "step": 1159 }, { "epoch": 0.35, "learning_rate": 0.00019611082712241333, "loss": 1.7949, "step": 1160 }, { "epoch": 0.35, "learning_rate": 0.0001961018469831043, "loss": 1.9558, "step": 1161 }, { "epoch": 0.35, "learning_rate": 0.000196092856694216, "loss": 1.8534, "step": 1162 }, { "epoch": 0.35, "learning_rate": 0.00019608385625669795, "loss": 1.8596, "step": 1163 }, { "epoch": 0.35, "learning_rate": 0.00019607484567150073, "loss": 1.9051, "step": 1164 }, { "epoch": 0.35, "learning_rate": 0.00019606582493957592, "loss": 1.8978, "step": 1165 }, { "epoch": 0.35, "learning_rate": 0.00019605679406187624, "loss": 1.8458, "step": 1166 }, { "epoch": 0.35, "learning_rate": 0.0001960477530393555, "loss": 1.8797, "step": 1167 }, { "epoch": 0.35, "learning_rate": 0.0001960387018729685, "loss": 1.7901, "step": 1168 }, { "epoch": 0.35, "learning_rate": 0.0001960296405636712, "loss": 1.8478, "step": 1169 }, { "epoch": 0.35, "learning_rate": 0.00019602056911242058, "loss": 1.7454, "step": 1170 }, { "epoch": 0.35, "learning_rate": 0.00019601148752017465, "loss": 1.785, "step": 1171 }, { "epoch": 0.35, "learning_rate": 0.0001960023957878926, "loss": 1.9778, "step": 1172 }, { "epoch": 0.35, "learning_rate": 0.00019599329391653463, "loss": 1.7788, "step": 1173 }, { "epoch": 0.35, "learning_rate": 0.000195984181907062, "loss": 1.9489, "step": 1174 }, { "epoch": 0.35, "learning_rate": 0.00019597505976043706, "loss": 1.8752, "step": 1175 }, { "epoch": 0.35, "learning_rate": 0.0001959659274776232, "loss": 1.918, "step": 1176 }, { "epoch": 0.35, "learning_rate": 0.00019595678505958492, "loss": 1.8735, "step": 1177 }, { "epoch": 0.35, "learning_rate": 0.0001959476325072878, "loss": 1.8635, "step": 1178 }, { "epoch": 0.35, "learning_rate": 0.00019593846982169844, "loss": 1.8398, "step": 1179 }, { "epoch": 0.36, "learning_rate": 0.00019592929700378454, "loss": 1.9814, "step": 1180 }, { "epoch": 0.36, "learning_rate": 0.00019592011405451486, "loss": 1.9297, "step": 1181 }, { "epoch": 0.36, "learning_rate": 0.00019591092097485927, "loss": 1.8131, "step": 1182 }, { "epoch": 0.36, "learning_rate": 0.00019590171776578862, "loss": 1.901, "step": 1183 }, { "epoch": 0.36, "learning_rate": 0.00019589250442827495, "loss": 1.8296, "step": 1184 }, { "epoch": 0.36, "learning_rate": 0.00019588328096329126, "loss": 1.8655, "step": 1185 }, { "epoch": 0.36, "learning_rate": 0.00019587404737181168, "loss": 1.9811, "step": 1186 }, { "epoch": 0.36, "learning_rate": 0.00019586480365481137, "loss": 1.9025, "step": 1187 }, { "epoch": 0.36, "learning_rate": 0.00019585554981326663, "loss": 1.8154, "step": 1188 }, { "epoch": 0.36, "learning_rate": 0.00019584628584815475, "loss": 1.7765, "step": 1189 }, { "epoch": 0.36, "learning_rate": 0.00019583701176045416, "loss": 1.88, "step": 1190 }, { "epoch": 0.36, "learning_rate": 0.00019582772755114428, "loss": 1.8235, "step": 1191 }, { "epoch": 0.36, "learning_rate": 0.00019581843322120568, "loss": 1.8602, "step": 1192 }, { "epoch": 0.36, "learning_rate": 0.0001958091287716199, "loss": 1.7892, "step": 1193 }, { "epoch": 0.36, "learning_rate": 0.00019579981420336968, "loss": 1.7941, "step": 1194 }, { "epoch": 0.36, "learning_rate": 0.0001957904895174387, "loss": 1.9148, "step": 1195 }, { "epoch": 0.36, "learning_rate": 0.0001957811547148118, "loss": 1.8124, "step": 1196 }, { "epoch": 0.36, "learning_rate": 0.00019577180979647482, "loss": 1.8444, "step": 1197 }, { "epoch": 0.36, "learning_rate": 0.00019576245476341474, "loss": 1.8945, "step": 1198 }, { "epoch": 0.36, "learning_rate": 0.00019575308961661955, "loss": 1.9158, "step": 1199 }, { "epoch": 0.36, "learning_rate": 0.00019574371435707833, "loss": 1.8887, "step": 1200 }, { "epoch": 0.36, "learning_rate": 0.00019573432898578122, "loss": 1.9305, "step": 1201 }, { "epoch": 0.36, "learning_rate": 0.00019572493350371949, "loss": 1.844, "step": 1202 }, { "epoch": 0.36, "learning_rate": 0.00019571552791188535, "loss": 1.9371, "step": 1203 }, { "epoch": 0.36, "learning_rate": 0.00019570611221127217, "loss": 1.8787, "step": 1204 }, { "epoch": 0.36, "learning_rate": 0.0001956966864028744, "loss": 1.7354, "step": 1205 }, { "epoch": 0.36, "learning_rate": 0.00019568725048768748, "loss": 1.8591, "step": 1206 }, { "epoch": 0.36, "learning_rate": 0.000195677804466708, "loss": 1.8715, "step": 1207 }, { "epoch": 0.36, "learning_rate": 0.00019566834834093354, "loss": 1.8517, "step": 1208 }, { "epoch": 0.36, "learning_rate": 0.00019565888211136283, "loss": 1.8607, "step": 1209 }, { "epoch": 0.36, "learning_rate": 0.00019564940577899563, "loss": 1.9241, "step": 1210 }, { "epoch": 0.36, "learning_rate": 0.0001956399193448327, "loss": 1.9024, "step": 1211 }, { "epoch": 0.36, "learning_rate": 0.00019563042280987597, "loss": 1.9278, "step": 1212 }, { "epoch": 0.37, "learning_rate": 0.0001956209161751284, "loss": 1.81, "step": 1213 }, { "epoch": 0.37, "learning_rate": 0.000195611399441594, "loss": 1.9995, "step": 1214 }, { "epoch": 0.37, "learning_rate": 0.00019560187261027787, "loss": 1.9119, "step": 1215 }, { "epoch": 0.37, "learning_rate": 0.00019559233568218614, "loss": 2.1156, "step": 1216 }, { "epoch": 0.37, "learning_rate": 0.00019558278865832606, "loss": 1.87, "step": 1217 }, { "epoch": 0.37, "learning_rate": 0.0001955732315397059, "loss": 1.858, "step": 1218 }, { "epoch": 0.37, "learning_rate": 0.00019556366432733503, "loss": 1.7808, "step": 1219 }, { "epoch": 0.37, "learning_rate": 0.00019555408702222384, "loss": 1.9113, "step": 1220 }, { "epoch": 0.37, "learning_rate": 0.00019554449962538387, "loss": 2.0471, "step": 1221 }, { "epoch": 0.37, "learning_rate": 0.00019553490213782758, "loss": 1.8677, "step": 1222 }, { "epoch": 0.37, "learning_rate": 0.00019552529456056864, "loss": 1.9018, "step": 1223 }, { "epoch": 0.37, "learning_rate": 0.0001955156768946218, "loss": 1.9089, "step": 1224 }, { "epoch": 0.37, "learning_rate": 0.00019550604914100268, "loss": 1.8544, "step": 1225 }, { "epoch": 0.37, "learning_rate": 0.00019549641130072818, "loss": 1.9792, "step": 1226 }, { "epoch": 0.37, "learning_rate": 0.00019548676337481614, "loss": 1.9795, "step": 1227 }, { "epoch": 0.37, "learning_rate": 0.00019547710536428552, "loss": 2.0855, "step": 1228 }, { "epoch": 0.37, "learning_rate": 0.00019546743727015634, "loss": 1.9032, "step": 1229 }, { "epoch": 0.37, "learning_rate": 0.00019545775909344965, "loss": 1.8446, "step": 1230 }, { "epoch": 0.37, "learning_rate": 0.00019544807083518763, "loss": 1.8547, "step": 1231 }, { "epoch": 0.37, "learning_rate": 0.00019543837249639338, "loss": 1.9354, "step": 1232 }, { "epoch": 0.37, "learning_rate": 0.0001954286640780913, "loss": 1.8189, "step": 1233 }, { "epoch": 0.37, "learning_rate": 0.00019541894558130663, "loss": 1.9135, "step": 1234 }, { "epoch": 0.37, "learning_rate": 0.00019540921700706584, "loss": 1.7943, "step": 1235 }, { "epoch": 0.37, "learning_rate": 0.00019539947835639634, "loss": 1.8683, "step": 1236 }, { "epoch": 0.37, "learning_rate": 0.00019538972963032664, "loss": 1.9308, "step": 1237 }, { "epoch": 0.37, "learning_rate": 0.00019537997082988637, "loss": 1.7391, "step": 1238 }, { "epoch": 0.37, "learning_rate": 0.0001953702019561062, "loss": 1.8331, "step": 1239 }, { "epoch": 0.37, "learning_rate": 0.00019536042301001777, "loss": 1.7844, "step": 1240 }, { "epoch": 0.37, "learning_rate": 0.00019535063399265395, "loss": 1.8423, "step": 1241 }, { "epoch": 0.37, "learning_rate": 0.00019534083490504854, "loss": 1.8979, "step": 1242 }, { "epoch": 0.37, "learning_rate": 0.00019533102574823642, "loss": 1.8463, "step": 1243 }, { "epoch": 0.37, "learning_rate": 0.0001953212065232536, "loss": 1.8432, "step": 1244 }, { "epoch": 0.37, "learning_rate": 0.00019531137723113716, "loss": 1.9924, "step": 1245 }, { "epoch": 0.38, "learning_rate": 0.0001953015378729251, "loss": 1.8806, "step": 1246 }, { "epoch": 0.38, "learning_rate": 0.00019529168844965663, "loss": 1.9149, "step": 1247 }, { "epoch": 0.38, "learning_rate": 0.000195281828962372, "loss": 1.8541, "step": 1248 }, { "epoch": 0.38, "learning_rate": 0.00019527195941211248, "loss": 1.942, "step": 1249 }, { "epoch": 0.38, "learning_rate": 0.00019526207979992036, "loss": 2.0249, "step": 1250 }, { "epoch": 0.38, "learning_rate": 0.00019525219012683914, "loss": 1.9407, "step": 1251 }, { "epoch": 0.38, "learning_rate": 0.00019524229039391327, "loss": 2.0269, "step": 1252 }, { "epoch": 0.38, "learning_rate": 0.00019523238060218827, "loss": 1.9757, "step": 1253 }, { "epoch": 0.38, "learning_rate": 0.00019522246075271075, "loss": 1.9381, "step": 1254 }, { "epoch": 0.38, "learning_rate": 0.00019521253084652838, "loss": 1.8501, "step": 1255 }, { "epoch": 0.38, "learning_rate": 0.00019520259088468987, "loss": 1.8268, "step": 1256 }, { "epoch": 0.38, "learning_rate": 0.000195192640868245, "loss": 1.9052, "step": 1257 }, { "epoch": 0.38, "learning_rate": 0.00019518268079824462, "loss": 2.0176, "step": 1258 }, { "epoch": 0.38, "learning_rate": 0.0001951727106757407, "loss": 2.0581, "step": 1259 }, { "epoch": 0.38, "learning_rate": 0.0001951627305017861, "loss": 1.9281, "step": 1260 }, { "epoch": 0.38, "learning_rate": 0.00019515274027743495, "loss": 1.9604, "step": 1261 }, { "epoch": 0.38, "learning_rate": 0.00019514274000374232, "loss": 1.8892, "step": 1262 }, { "epoch": 0.38, "learning_rate": 0.00019513272968176435, "loss": 1.9977, "step": 1263 }, { "epoch": 0.38, "learning_rate": 0.00019512270931255825, "loss": 1.8464, "step": 1264 }, { "epoch": 0.38, "learning_rate": 0.00019511267889718234, "loss": 1.9418, "step": 1265 }, { "epoch": 0.38, "learning_rate": 0.00019510263843669593, "loss": 1.8855, "step": 1266 }, { "epoch": 0.38, "learning_rate": 0.0001950925879321594, "loss": 1.9239, "step": 1267 }, { "epoch": 0.38, "learning_rate": 0.00019508252738463427, "loss": 1.9289, "step": 1268 }, { "epoch": 0.38, "learning_rate": 0.00019507245679518297, "loss": 1.8954, "step": 1269 }, { "epoch": 0.38, "learning_rate": 0.0001950623761648692, "loss": 1.8346, "step": 1270 }, { "epoch": 0.38, "learning_rate": 0.0001950522854947575, "loss": 1.8744, "step": 1271 }, { "epoch": 0.38, "learning_rate": 0.00019504218478591366, "loss": 1.8843, "step": 1272 }, { "epoch": 0.38, "learning_rate": 0.00019503207403940437, "loss": 1.9115, "step": 1273 }, { "epoch": 0.38, "learning_rate": 0.0001950219532562975, "loss": 1.8286, "step": 1274 }, { "epoch": 0.38, "learning_rate": 0.0001950118224376619, "loss": 1.7999, "step": 1275 }, { "epoch": 0.38, "learning_rate": 0.00019500168158456753, "loss": 1.8588, "step": 1276 }, { "epoch": 0.38, "learning_rate": 0.0001949915306980854, "loss": 1.8408, "step": 1277 }, { "epoch": 0.38, "learning_rate": 0.0001949813697792876, "loss": 1.9434, "step": 1278 }, { "epoch": 0.39, "learning_rate": 0.00019497119882924718, "loss": 1.9117, "step": 1279 }, { "epoch": 0.39, "learning_rate": 0.00019496101784903836, "loss": 1.9959, "step": 1280 }, { "epoch": 0.39, "learning_rate": 0.00019495082683973638, "loss": 1.9553, "step": 1281 }, { "epoch": 0.39, "learning_rate": 0.00019494062580241756, "loss": 1.8895, "step": 1282 }, { "epoch": 0.39, "learning_rate": 0.00019493041473815924, "loss": 1.9059, "step": 1283 }, { "epoch": 0.39, "learning_rate": 0.00019492019364803983, "loss": 1.8164, "step": 1284 }, { "epoch": 0.39, "learning_rate": 0.00019490996253313883, "loss": 1.787, "step": 1285 }, { "epoch": 0.39, "learning_rate": 0.00019489972139453675, "loss": 1.9278, "step": 1286 }, { "epoch": 0.39, "learning_rate": 0.00019488947023331524, "loss": 1.8619, "step": 1287 }, { "epoch": 0.39, "learning_rate": 0.00019487920905055686, "loss": 1.931, "step": 1288 }, { "epoch": 0.39, "learning_rate": 0.0001948689378473454, "loss": 2.0306, "step": 1289 }, { "epoch": 0.39, "learning_rate": 0.0001948586566247656, "loss": 1.8671, "step": 1290 }, { "epoch": 0.39, "learning_rate": 0.00019484836538390327, "loss": 1.7883, "step": 1291 }, { "epoch": 0.39, "learning_rate": 0.00019483806412584536, "loss": 1.8275, "step": 1292 }, { "epoch": 0.39, "learning_rate": 0.00019482775285167974, "loss": 1.9691, "step": 1293 }, { "epoch": 0.39, "learning_rate": 0.00019481743156249544, "loss": 1.7663, "step": 1294 }, { "epoch": 0.39, "learning_rate": 0.00019480710025938255, "loss": 1.8922, "step": 1295 }, { "epoch": 0.39, "learning_rate": 0.00019479675894343217, "loss": 1.8607, "step": 1296 }, { "epoch": 0.39, "learning_rate": 0.0001947864076157364, "loss": 1.9696, "step": 1297 }, { "epoch": 0.39, "learning_rate": 0.00019477604627738863, "loss": 1.7463, "step": 1298 }, { "epoch": 0.39, "learning_rate": 0.00019476567492948297, "loss": 1.8803, "step": 1299 }, { "epoch": 0.39, "learning_rate": 0.00019475529357311492, "loss": 1.8241, "step": 1300 }, { "epoch": 0.39, "learning_rate": 0.00019474490220938075, "loss": 1.8509, "step": 1301 }, { "epoch": 0.39, "learning_rate": 0.00019473450083937804, "loss": 1.8344, "step": 1302 }, { "epoch": 0.39, "learning_rate": 0.00019472408946420524, "loss": 1.8396, "step": 1303 }, { "epoch": 0.39, "learning_rate": 0.00019471366808496192, "loss": 1.8152, "step": 1304 }, { "epoch": 0.39, "learning_rate": 0.00019470323670274875, "loss": 1.9305, "step": 1305 }, { "epoch": 0.39, "learning_rate": 0.00019469279531866737, "loss": 2.0047, "step": 1306 }, { "epoch": 0.39, "learning_rate": 0.00019468234393382056, "loss": 1.9128, "step": 1307 }, { "epoch": 0.39, "learning_rate": 0.00019467188254931213, "loss": 1.7892, "step": 1308 }, { "epoch": 0.39, "learning_rate": 0.0001946614111662469, "loss": 1.9523, "step": 1309 }, { "epoch": 0.39, "learning_rate": 0.00019465092978573082, "loss": 1.8311, "step": 1310 }, { "epoch": 0.39, "learning_rate": 0.00019464043840887078, "loss": 1.8448, "step": 1311 }, { "epoch": 0.39, "learning_rate": 0.00019462993703677487, "loss": 1.8518, "step": 1312 }, { "epoch": 0.4, "learning_rate": 0.00019461942567055215, "loss": 1.7941, "step": 1313 }, { "epoch": 0.4, "learning_rate": 0.0001946089043113128, "loss": 1.8789, "step": 1314 }, { "epoch": 0.4, "learning_rate": 0.0001945983729601679, "loss": 1.8689, "step": 1315 }, { "epoch": 0.4, "learning_rate": 0.00019458783161822982, "loss": 1.8123, "step": 1316 }, { "epoch": 0.4, "learning_rate": 0.0001945772802866118, "loss": 1.9618, "step": 1317 }, { "epoch": 0.4, "learning_rate": 0.00019456671896642816, "loss": 1.752, "step": 1318 }, { "epoch": 0.4, "learning_rate": 0.00019455614765879434, "loss": 1.8304, "step": 1319 }, { "epoch": 0.4, "learning_rate": 0.00019454556636482688, "loss": 1.8618, "step": 1320 }, { "epoch": 0.4, "learning_rate": 0.00019453497508564317, "loss": 1.7872, "step": 1321 }, { "epoch": 0.4, "learning_rate": 0.00019452437382236188, "loss": 1.8705, "step": 1322 }, { "epoch": 0.4, "learning_rate": 0.00019451376257610258, "loss": 1.8531, "step": 1323 }, { "epoch": 0.4, "learning_rate": 0.000194503141347986, "loss": 1.7301, "step": 1324 }, { "epoch": 0.4, "learning_rate": 0.0001944925101391338, "loss": 1.8827, "step": 1325 }, { "epoch": 0.4, "learning_rate": 0.0001944818689506689, "loss": 1.8937, "step": 1326 }, { "epoch": 0.4, "learning_rate": 0.00019447121778371501, "loss": 1.8192, "step": 1327 }, { "epoch": 0.4, "learning_rate": 0.0001944605566393971, "loss": 1.8381, "step": 1328 }, { "epoch": 0.4, "learning_rate": 0.00019444988551884116, "loss": 1.8588, "step": 1329 }, { "epoch": 0.4, "learning_rate": 0.00019443920442317408, "loss": 1.7889, "step": 1330 }, { "epoch": 0.4, "learning_rate": 0.00019442851335352404, "loss": 1.7997, "step": 1331 }, { "epoch": 0.4, "learning_rate": 0.00019441781231102008, "loss": 1.7823, "step": 1332 }, { "epoch": 0.4, "learning_rate": 0.0001944071012967924, "loss": 1.8154, "step": 1333 }, { "epoch": 0.4, "learning_rate": 0.00019439638031197215, "loss": 1.821, "step": 1334 }, { "epoch": 0.4, "learning_rate": 0.0001943856493576917, "loss": 1.8592, "step": 1335 }, { "epoch": 0.4, "learning_rate": 0.00019437490843508435, "loss": 1.8171, "step": 1336 }, { "epoch": 0.4, "learning_rate": 0.00019436415754528445, "loss": 1.7911, "step": 1337 }, { "epoch": 0.4, "learning_rate": 0.00019435339668942746, "loss": 1.7615, "step": 1338 }, { "epoch": 0.4, "learning_rate": 0.0001943426258686498, "loss": 1.8421, "step": 1339 }, { "epoch": 0.4, "learning_rate": 0.0001943318450840891, "loss": 1.8658, "step": 1340 }, { "epoch": 0.4, "learning_rate": 0.00019432105433688385, "loss": 1.8646, "step": 1341 }, { "epoch": 0.4, "learning_rate": 0.00019431025362817378, "loss": 1.7392, "step": 1342 }, { "epoch": 0.4, "learning_rate": 0.00019429944295909954, "loss": 1.7469, "step": 1343 }, { "epoch": 0.4, "learning_rate": 0.0001942886223308029, "loss": 1.7834, "step": 1344 }, { "epoch": 0.4, "learning_rate": 0.0001942777917444266, "loss": 1.8603, "step": 1345 }, { "epoch": 0.41, "learning_rate": 0.00019426695120111452, "loss": 1.871, "step": 1346 }, { "epoch": 0.41, "learning_rate": 0.00019425610070201158, "loss": 1.9114, "step": 1347 }, { "epoch": 0.41, "learning_rate": 0.00019424524024826373, "loss": 1.8023, "step": 1348 }, { "epoch": 0.41, "learning_rate": 0.00019423436984101792, "loss": 1.9097, "step": 1349 }, { "epoch": 0.41, "learning_rate": 0.00019422348948142223, "loss": 1.7821, "step": 1350 }, { "epoch": 0.41, "learning_rate": 0.0001942125991706258, "loss": 1.8386, "step": 1351 }, { "epoch": 0.41, "learning_rate": 0.00019420169890977876, "loss": 1.9576, "step": 1352 }, { "epoch": 0.41, "learning_rate": 0.00019419078870003232, "loss": 1.859, "step": 1353 }, { "epoch": 0.41, "learning_rate": 0.00019417986854253872, "loss": 1.8064, "step": 1354 }, { "epoch": 0.41, "learning_rate": 0.00019416893843845128, "loss": 1.8415, "step": 1355 }, { "epoch": 0.41, "learning_rate": 0.00019415799838892435, "loss": 1.7897, "step": 1356 }, { "epoch": 0.41, "learning_rate": 0.00019414704839511334, "loss": 1.9176, "step": 1357 }, { "epoch": 0.41, "learning_rate": 0.00019413608845817476, "loss": 1.8255, "step": 1358 }, { "epoch": 0.41, "learning_rate": 0.00019412511857926602, "loss": 1.8859, "step": 1359 }, { "epoch": 0.41, "learning_rate": 0.00019411413875954577, "loss": 1.8242, "step": 1360 }, { "epoch": 0.41, "learning_rate": 0.00019410314900017358, "loss": 1.7803, "step": 1361 }, { "epoch": 0.41, "learning_rate": 0.0001940921493023101, "loss": 1.7794, "step": 1362 }, { "epoch": 0.41, "learning_rate": 0.00019408113966711704, "loss": 1.7689, "step": 1363 }, { "epoch": 0.41, "learning_rate": 0.00019407012009575717, "loss": 1.954, "step": 1364 }, { "epoch": 0.41, "learning_rate": 0.0001940590905893943, "loss": 1.9328, "step": 1365 }, { "epoch": 0.41, "learning_rate": 0.00019404805114919328, "loss": 1.8402, "step": 1366 }, { "epoch": 0.41, "learning_rate": 0.00019403700177632002, "loss": 1.8512, "step": 1367 }, { "epoch": 0.41, "learning_rate": 0.00019402594247194146, "loss": 1.9371, "step": 1368 }, { "epoch": 0.41, "learning_rate": 0.00019401487323722564, "loss": 1.9745, "step": 1369 }, { "epoch": 0.41, "learning_rate": 0.00019400379407334155, "loss": 1.8566, "step": 1370 }, { "epoch": 0.41, "learning_rate": 0.00019399270498145934, "loss": 1.7736, "step": 1371 }, { "epoch": 0.41, "learning_rate": 0.00019398160596275016, "loss": 1.8724, "step": 1372 }, { "epoch": 0.41, "learning_rate": 0.00019397049701838618, "loss": 1.9148, "step": 1373 }, { "epoch": 0.41, "learning_rate": 0.00019395937814954066, "loss": 1.9084, "step": 1374 }, { "epoch": 0.41, "learning_rate": 0.00019394824935738792, "loss": 1.811, "step": 1375 }, { "epoch": 0.41, "learning_rate": 0.00019393711064310325, "loss": 1.7732, "step": 1376 }, { "epoch": 0.41, "learning_rate": 0.00019392596200786306, "loss": 1.8278, "step": 1377 }, { "epoch": 0.41, "learning_rate": 0.0001939148034528448, "loss": 1.8535, "step": 1378 }, { "epoch": 0.42, "learning_rate": 0.00019390363497922696, "loss": 1.8617, "step": 1379 }, { "epoch": 0.42, "learning_rate": 0.00019389245658818906, "loss": 1.7857, "step": 1380 }, { "epoch": 0.42, "learning_rate": 0.0001938812682809117, "loss": 1.8212, "step": 1381 }, { "epoch": 0.42, "learning_rate": 0.00019387007005857644, "loss": 1.8993, "step": 1382 }, { "epoch": 0.42, "learning_rate": 0.00019385886192236607, "loss": 1.8609, "step": 1383 }, { "epoch": 0.42, "learning_rate": 0.0001938476438734642, "loss": 1.838, "step": 1384 }, { "epoch": 0.42, "learning_rate": 0.00019383641591305567, "loss": 1.8251, "step": 1385 }, { "epoch": 0.42, "learning_rate": 0.00019382517804232628, "loss": 1.9035, "step": 1386 }, { "epoch": 0.42, "learning_rate": 0.00019381393026246286, "loss": 1.8654, "step": 1387 }, { "epoch": 0.42, "learning_rate": 0.00019380267257465333, "loss": 1.8659, "step": 1388 }, { "epoch": 0.42, "learning_rate": 0.0001937914049800867, "loss": 1.7973, "step": 1389 }, { "epoch": 0.42, "learning_rate": 0.00019378012747995292, "loss": 1.9608, "step": 1390 }, { "epoch": 0.42, "learning_rate": 0.00019376884007544305, "loss": 1.8458, "step": 1391 }, { "epoch": 0.42, "learning_rate": 0.00019375754276774914, "loss": 1.8208, "step": 1392 }, { "epoch": 0.42, "learning_rate": 0.00019374623555806444, "loss": 1.9062, "step": 1393 }, { "epoch": 0.42, "learning_rate": 0.00019373491844758303, "loss": 1.8587, "step": 1394 }, { "epoch": 0.42, "learning_rate": 0.00019372359143750019, "loss": 1.7626, "step": 1395 }, { "epoch": 0.42, "learning_rate": 0.00019371225452901217, "loss": 1.9418, "step": 1396 }, { "epoch": 0.42, "learning_rate": 0.00019370090772331632, "loss": 1.9756, "step": 1397 }, { "epoch": 0.42, "learning_rate": 0.00019368955102161097, "loss": 1.8321, "step": 1398 }, { "epoch": 0.42, "learning_rate": 0.00019367818442509558, "loss": 1.7757, "step": 1399 }, { "epoch": 0.42, "learning_rate": 0.00019366680793497055, "loss": 1.8794, "step": 1400 }, { "epoch": 0.42, "learning_rate": 0.00019365542155243745, "loss": 1.8615, "step": 1401 }, { "epoch": 0.42, "learning_rate": 0.00019364402527869882, "loss": 1.8608, "step": 1402 }, { "epoch": 0.42, "learning_rate": 0.00019363261911495817, "loss": 1.9307, "step": 1403 }, { "epoch": 0.42, "learning_rate": 0.00019362120306242023, "loss": 1.7744, "step": 1404 }, { "epoch": 0.42, "learning_rate": 0.0001936097771222906, "loss": 1.7984, "step": 1405 }, { "epoch": 0.42, "learning_rate": 0.00019359834129577608, "loss": 1.9243, "step": 1406 }, { "epoch": 0.42, "learning_rate": 0.0001935868955840844, "loss": 1.7713, "step": 1407 }, { "epoch": 0.42, "learning_rate": 0.00019357543998842433, "loss": 1.935, "step": 1408 }, { "epoch": 0.42, "learning_rate": 0.00019356397451000584, "loss": 1.8481, "step": 1409 }, { "epoch": 0.42, "learning_rate": 0.00019355249915003973, "loss": 1.8186, "step": 1410 }, { "epoch": 0.42, "learning_rate": 0.000193541013909738, "loss": 1.8826, "step": 1411 }, { "epoch": 0.43, "learning_rate": 0.00019352951879031362, "loss": 1.8615, "step": 1412 }, { "epoch": 0.43, "learning_rate": 0.0001935180137929806, "loss": 1.8496, "step": 1413 }, { "epoch": 0.43, "learning_rate": 0.00019350649891895403, "loss": 1.851, "step": 1414 }, { "epoch": 0.43, "learning_rate": 0.00019349497416945002, "loss": 1.8355, "step": 1415 }, { "epoch": 0.43, "learning_rate": 0.00019348343954568575, "loss": 1.8151, "step": 1416 }, { "epoch": 0.43, "learning_rate": 0.00019347189504887943, "loss": 1.7843, "step": 1417 }, { "epoch": 0.43, "learning_rate": 0.0001934603406802503, "loss": 1.7704, "step": 1418 }, { "epoch": 0.43, "learning_rate": 0.0001934487764410186, "loss": 1.8469, "step": 1419 }, { "epoch": 0.43, "learning_rate": 0.0001934372023324057, "loss": 1.7747, "step": 1420 }, { "epoch": 0.43, "learning_rate": 0.00019342561835563403, "loss": 1.8581, "step": 1421 }, { "epoch": 0.43, "learning_rate": 0.0001934140245119269, "loss": 1.8926, "step": 1422 }, { "epoch": 0.43, "learning_rate": 0.00019340242080250882, "loss": 1.8447, "step": 1423 }, { "epoch": 0.43, "learning_rate": 0.0001933908072286053, "loss": 1.9126, "step": 1424 }, { "epoch": 0.43, "learning_rate": 0.00019337918379144285, "loss": 1.7707, "step": 1425 }, { "epoch": 0.43, "learning_rate": 0.00019336755049224907, "loss": 1.8869, "step": 1426 }, { "epoch": 0.43, "learning_rate": 0.00019335590733225258, "loss": 1.8311, "step": 1427 }, { "epoch": 0.43, "learning_rate": 0.00019334425431268305, "loss": 1.7814, "step": 1428 }, { "epoch": 0.43, "learning_rate": 0.0001933325914347712, "loss": 1.7136, "step": 1429 }, { "epoch": 0.43, "learning_rate": 0.00019332091869974876, "loss": 1.9631, "step": 1430 }, { "epoch": 0.43, "learning_rate": 0.00019330923610884853, "loss": 1.8306, "step": 1431 }, { "epoch": 0.43, "learning_rate": 0.00019329754366330433, "loss": 1.9029, "step": 1432 }, { "epoch": 0.43, "learning_rate": 0.000193285841364351, "loss": 1.7841, "step": 1433 }, { "epoch": 0.43, "learning_rate": 0.00019327412921322457, "loss": 1.7488, "step": 1434 }, { "epoch": 0.43, "learning_rate": 0.00019326240721116188, "loss": 2.0197, "step": 1435 }, { "epoch": 0.43, "learning_rate": 0.00019325067535940093, "loss": 1.8475, "step": 1436 }, { "epoch": 0.43, "learning_rate": 0.00019323893365918077, "loss": 1.8072, "step": 1437 }, { "epoch": 0.43, "learning_rate": 0.0001932271821117415, "loss": 1.8866, "step": 1438 }, { "epoch": 0.43, "learning_rate": 0.0001932154207183242, "loss": 1.7442, "step": 1439 }, { "epoch": 0.43, "learning_rate": 0.00019320364948017103, "loss": 1.8843, "step": 1440 }, { "epoch": 0.43, "learning_rate": 0.0001931918683985252, "loss": 1.7322, "step": 1441 }, { "epoch": 0.43, "learning_rate": 0.00019318007747463095, "loss": 1.9008, "step": 1442 }, { "epoch": 0.43, "learning_rate": 0.00019316827670973354, "loss": 1.8138, "step": 1443 }, { "epoch": 0.43, "learning_rate": 0.00019315646610507922, "loss": 1.8368, "step": 1444 }, { "epoch": 0.43, "learning_rate": 0.00019314464566191545, "loss": 1.8129, "step": 1445 }, { "epoch": 0.44, "learning_rate": 0.00019313281538149054, "loss": 1.8431, "step": 1446 }, { "epoch": 0.44, "learning_rate": 0.00019312097526505393, "loss": 1.9271, "step": 1447 }, { "epoch": 0.44, "learning_rate": 0.00019310912531385612, "loss": 1.873, "step": 1448 }, { "epoch": 0.44, "learning_rate": 0.00019309726552914857, "loss": 1.961, "step": 1449 }, { "epoch": 0.44, "learning_rate": 0.00019308539591218385, "loss": 1.7777, "step": 1450 }, { "epoch": 0.44, "learning_rate": 0.0001930735164642156, "loss": 1.9115, "step": 1451 }, { "epoch": 0.44, "learning_rate": 0.0001930616271864983, "loss": 1.8283, "step": 1452 }, { "epoch": 0.44, "learning_rate": 0.00019304972808028775, "loss": 1.8265, "step": 1453 }, { "epoch": 0.44, "learning_rate": 0.00019303781914684057, "loss": 1.7957, "step": 1454 }, { "epoch": 0.44, "learning_rate": 0.00019302590038741452, "loss": 1.916, "step": 1455 }, { "epoch": 0.44, "learning_rate": 0.00019301397180326833, "loss": 1.7617, "step": 1456 }, { "epoch": 0.44, "learning_rate": 0.00019300203339566187, "loss": 1.7418, "step": 1457 }, { "epoch": 0.44, "learning_rate": 0.00019299008516585598, "loss": 1.8383, "step": 1458 }, { "epoch": 0.44, "learning_rate": 0.00019297812711511255, "loss": 1.8162, "step": 1459 }, { "epoch": 0.44, "learning_rate": 0.00019296615924469443, "loss": 1.8374, "step": 1460 }, { "epoch": 0.44, "learning_rate": 0.00019295418155586567, "loss": 1.7882, "step": 1461 }, { "epoch": 0.44, "learning_rate": 0.0001929421940498912, "loss": 1.8059, "step": 1462 }, { "epoch": 0.44, "learning_rate": 0.0001929301967280371, "loss": 1.9647, "step": 1463 }, { "epoch": 0.44, "learning_rate": 0.00019291818959157046, "loss": 1.8054, "step": 1464 }, { "epoch": 0.44, "learning_rate": 0.00019290617264175927, "loss": 1.9881, "step": 1465 }, { "epoch": 0.44, "learning_rate": 0.0001928941458798728, "loss": 1.8076, "step": 1466 }, { "epoch": 0.44, "learning_rate": 0.0001928821093071812, "loss": 1.9557, "step": 1467 }, { "epoch": 0.44, "learning_rate": 0.00019287006292495563, "loss": 1.7542, "step": 1468 }, { "epoch": 0.44, "learning_rate": 0.00019285800673446839, "loss": 1.7689, "step": 1469 }, { "epoch": 0.44, "learning_rate": 0.00019284594073699275, "loss": 1.8544, "step": 1470 }, { "epoch": 0.44, "learning_rate": 0.00019283386493380308, "loss": 1.81, "step": 1471 }, { "epoch": 0.44, "learning_rate": 0.00019282177932617461, "loss": 1.8202, "step": 1472 }, { "epoch": 0.44, "learning_rate": 0.0001928096839153839, "loss": 1.9154, "step": 1473 }, { "epoch": 0.44, "learning_rate": 0.00019279757870270826, "loss": 1.9299, "step": 1474 }, { "epoch": 0.44, "learning_rate": 0.00019278546368942622, "loss": 1.9368, "step": 1475 }, { "epoch": 0.44, "learning_rate": 0.00019277333887681727, "loss": 1.9406, "step": 1476 }, { "epoch": 0.44, "learning_rate": 0.00019276120426616188, "loss": 1.809, "step": 1477 }, { "epoch": 0.44, "learning_rate": 0.0001927490598587417, "loss": 1.9028, "step": 1478 }, { "epoch": 0.45, "learning_rate": 0.0001927369056558393, "loss": 1.9194, "step": 1479 }, { "epoch": 0.45, "learning_rate": 0.00019272474165873832, "loss": 1.8415, "step": 1480 }, { "epoch": 0.45, "learning_rate": 0.00019271256786872342, "loss": 1.9672, "step": 1481 }, { "epoch": 0.45, "learning_rate": 0.00019270038428708034, "loss": 1.8543, "step": 1482 }, { "epoch": 0.45, "learning_rate": 0.0001926881909150958, "loss": 1.8532, "step": 1483 }, { "epoch": 0.45, "learning_rate": 0.0001926759877540576, "loss": 1.8836, "step": 1484 }, { "epoch": 0.45, "learning_rate": 0.0001926637748052545, "loss": 1.8179, "step": 1485 }, { "epoch": 0.45, "learning_rate": 0.00019265155206997639, "loss": 1.842, "step": 1486 }, { "epoch": 0.45, "learning_rate": 0.0001926393195495141, "loss": 1.8914, "step": 1487 }, { "epoch": 0.45, "learning_rate": 0.00019262707724515958, "loss": 1.7263, "step": 1488 }, { "epoch": 0.45, "learning_rate": 0.00019261482515820578, "loss": 1.9171, "step": 1489 }, { "epoch": 0.45, "learning_rate": 0.00019260256328994664, "loss": 1.8736, "step": 1490 }, { "epoch": 0.45, "learning_rate": 0.00019259029164167722, "loss": 1.8975, "step": 1491 }, { "epoch": 0.45, "learning_rate": 0.00019257801021469348, "loss": 1.8771, "step": 1492 }, { "epoch": 0.45, "learning_rate": 0.00019256571901029256, "loss": 1.8919, "step": 1493 }, { "epoch": 0.45, "learning_rate": 0.00019255341802977257, "loss": 1.8599, "step": 1494 }, { "epoch": 0.45, "learning_rate": 0.00019254110727443262, "loss": 1.814, "step": 1495 }, { "epoch": 0.45, "learning_rate": 0.0001925287867455729, "loss": 1.8504, "step": 1496 }, { "epoch": 0.45, "learning_rate": 0.0001925164564444946, "loss": 1.8369, "step": 1497 }, { "epoch": 0.45, "learning_rate": 0.0001925041163725, "loss": 1.8109, "step": 1498 }, { "epoch": 0.45, "learning_rate": 0.0001924917665308923, "loss": 1.8791, "step": 1499 }, { "epoch": 0.45, "learning_rate": 0.00019247940692097582, "loss": 1.7349, "step": 1500 }, { "epoch": 0.45, "learning_rate": 0.00019246703754405596, "loss": 1.7718, "step": 1501 }, { "epoch": 0.45, "learning_rate": 0.00019245465840143902, "loss": 1.8959, "step": 1502 }, { "epoch": 0.45, "learning_rate": 0.00019244226949443238, "loss": 1.8425, "step": 1503 }, { "epoch": 0.45, "learning_rate": 0.0001924298708243445, "loss": 1.8627, "step": 1504 }, { "epoch": 0.45, "learning_rate": 0.00019241746239248485, "loss": 1.8698, "step": 1505 }, { "epoch": 0.45, "learning_rate": 0.00019240504420016389, "loss": 1.7653, "step": 1506 }, { "epoch": 0.45, "learning_rate": 0.00019239261624869316, "loss": 1.7632, "step": 1507 }, { "epoch": 0.45, "learning_rate": 0.0001923801785393852, "loss": 1.7495, "step": 1508 }, { "epoch": 0.45, "learning_rate": 0.00019236773107355357, "loss": 1.8959, "step": 1509 }, { "epoch": 0.45, "learning_rate": 0.00019235527385251292, "loss": 1.8017, "step": 1510 }, { "epoch": 0.45, "learning_rate": 0.00019234280687757885, "loss": 1.7899, "step": 1511 }, { "epoch": 0.46, "learning_rate": 0.00019233033015006806, "loss": 1.7401, "step": 1512 }, { "epoch": 0.46, "learning_rate": 0.00019231784367129822, "loss": 1.7897, "step": 1513 }, { "epoch": 0.46, "learning_rate": 0.00019230534744258816, "loss": 1.7643, "step": 1514 }, { "epoch": 0.46, "learning_rate": 0.00019229284146525752, "loss": 1.8687, "step": 1515 }, { "epoch": 0.46, "learning_rate": 0.00019228032574062712, "loss": 1.9156, "step": 1516 }, { "epoch": 0.46, "learning_rate": 0.00019226780027001885, "loss": 1.8007, "step": 1517 }, { "epoch": 0.46, "learning_rate": 0.00019225526505475547, "loss": 1.7967, "step": 1518 }, { "epoch": 0.46, "learning_rate": 0.0001922427200961609, "loss": 1.903, "step": 1519 }, { "epoch": 0.46, "learning_rate": 0.00019223016539556003, "loss": 1.8507, "step": 1520 }, { "epoch": 0.46, "learning_rate": 0.00019221760095427883, "loss": 1.8513, "step": 1521 }, { "epoch": 0.46, "learning_rate": 0.00019220502677364424, "loss": 1.8492, "step": 1522 }, { "epoch": 0.46, "learning_rate": 0.00019219244285498423, "loss": 1.8185, "step": 1523 }, { "epoch": 0.46, "learning_rate": 0.00019217984919962788, "loss": 1.7951, "step": 1524 }, { "epoch": 0.46, "learning_rate": 0.0001921672458089052, "loss": 1.7753, "step": 1525 }, { "epoch": 0.46, "learning_rate": 0.00019215463268414728, "loss": 1.8279, "step": 1526 }, { "epoch": 0.46, "learning_rate": 0.00019214200982668622, "loss": 1.8169, "step": 1527 }, { "epoch": 0.46, "learning_rate": 0.0001921293772378552, "loss": 1.7937, "step": 1528 }, { "epoch": 0.46, "learning_rate": 0.00019211673491898828, "loss": 1.8507, "step": 1529 }, { "epoch": 0.46, "learning_rate": 0.00019210408287142073, "loss": 1.9792, "step": 1530 }, { "epoch": 0.46, "learning_rate": 0.00019209142109648876, "loss": 1.9012, "step": 1531 }, { "epoch": 0.46, "learning_rate": 0.0001920787495955296, "loss": 1.8502, "step": 1532 }, { "epoch": 0.46, "learning_rate": 0.00019206606836988152, "loss": 1.8913, "step": 1533 }, { "epoch": 0.46, "learning_rate": 0.00019205337742088385, "loss": 1.8987, "step": 1534 }, { "epoch": 0.46, "learning_rate": 0.00019204067674987688, "loss": 1.9133, "step": 1535 }, { "epoch": 0.46, "learning_rate": 0.00019202796635820195, "loss": 1.8114, "step": 1536 }, { "epoch": 0.46, "learning_rate": 0.00019201524624720147, "loss": 1.927, "step": 1537 }, { "epoch": 0.46, "learning_rate": 0.00019200251641821888, "loss": 1.9029, "step": 1538 }, { "epoch": 0.46, "learning_rate": 0.00019198977687259853, "loss": 1.7948, "step": 1539 }, { "epoch": 0.46, "learning_rate": 0.00019197702761168594, "loss": 1.8638, "step": 1540 }, { "epoch": 0.46, "learning_rate": 0.00019196426863682755, "loss": 1.8563, "step": 1541 }, { "epoch": 0.46, "learning_rate": 0.00019195149994937093, "loss": 1.8808, "step": 1542 }, { "epoch": 0.46, "learning_rate": 0.00019193872155066455, "loss": 1.885, "step": 1543 }, { "epoch": 0.46, "learning_rate": 0.000191925933442058, "loss": 1.8815, "step": 1544 }, { "epoch": 0.47, "learning_rate": 0.00019191313562490187, "loss": 1.8594, "step": 1545 }, { "epoch": 0.47, "learning_rate": 0.00019190032810054776, "loss": 1.8939, "step": 1546 }, { "epoch": 0.47, "learning_rate": 0.00019188751087034835, "loss": 1.8366, "step": 1547 }, { "epoch": 0.47, "learning_rate": 0.00019187468393565727, "loss": 1.8234, "step": 1548 }, { "epoch": 0.47, "learning_rate": 0.0001918618472978292, "loss": 1.8417, "step": 1549 }, { "epoch": 0.47, "learning_rate": 0.00019184900095821984, "loss": 1.6991, "step": 1550 }, { "epoch": 0.47, "learning_rate": 0.00019183614491818598, "loss": 1.8084, "step": 1551 }, { "epoch": 0.47, "learning_rate": 0.00019182327917908532, "loss": 1.9035, "step": 1552 }, { "epoch": 0.47, "learning_rate": 0.0001918104037422767, "loss": 1.9125, "step": 1553 }, { "epoch": 0.47, "learning_rate": 0.00019179751860911993, "loss": 1.8207, "step": 1554 }, { "epoch": 0.47, "learning_rate": 0.00019178462378097576, "loss": 1.8415, "step": 1555 }, { "epoch": 0.47, "learning_rate": 0.00019177171925920615, "loss": 1.9879, "step": 1556 }, { "epoch": 0.47, "learning_rate": 0.0001917588050451739, "loss": 1.8107, "step": 1557 }, { "epoch": 0.47, "learning_rate": 0.00019174588114024302, "loss": 1.8363, "step": 1558 }, { "epoch": 0.47, "learning_rate": 0.00019173294754577834, "loss": 1.8056, "step": 1559 }, { "epoch": 0.47, "learning_rate": 0.00019172000426314584, "loss": 2.0842, "step": 1560 }, { "epoch": 0.47, "learning_rate": 0.00019170705129371252, "loss": 1.8608, "step": 1561 }, { "epoch": 0.47, "learning_rate": 0.00019169408863884636, "loss": 1.8404, "step": 1562 }, { "epoch": 0.47, "learning_rate": 0.00019168111629991636, "loss": 1.8781, "step": 1563 }, { "epoch": 0.47, "learning_rate": 0.00019166813427829262, "loss": 1.922, "step": 1564 }, { "epoch": 0.47, "learning_rate": 0.00019165514257534616, "loss": 1.7602, "step": 1565 }, { "epoch": 0.47, "learning_rate": 0.00019164214119244907, "loss": 1.9548, "step": 1566 }, { "epoch": 0.47, "learning_rate": 0.0001916291301309745, "loss": 1.8707, "step": 1567 }, { "epoch": 0.47, "learning_rate": 0.00019161610939229652, "loss": 1.8464, "step": 1568 }, { "epoch": 0.47, "learning_rate": 0.00019160307897779036, "loss": 1.8188, "step": 1569 }, { "epoch": 0.47, "learning_rate": 0.00019159003888883218, "loss": 1.8486, "step": 1570 }, { "epoch": 0.47, "learning_rate": 0.0001915769891267991, "loss": 1.7218, "step": 1571 }, { "epoch": 0.47, "learning_rate": 0.00019156392969306947, "loss": 1.8403, "step": 1572 }, { "epoch": 0.47, "learning_rate": 0.00019155086058902243, "loss": 1.8205, "step": 1573 }, { "epoch": 0.47, "learning_rate": 0.0001915377818160383, "loss": 1.8339, "step": 1574 }, { "epoch": 0.47, "learning_rate": 0.0001915246933754983, "loss": 2.0297, "step": 1575 }, { "epoch": 0.47, "learning_rate": 0.00019151159526878484, "loss": 1.7772, "step": 1576 }, { "epoch": 0.47, "learning_rate": 0.00019149848749728117, "loss": 1.8415, "step": 1577 }, { "epoch": 0.48, "learning_rate": 0.00019148537006237164, "loss": 1.7915, "step": 1578 }, { "epoch": 0.48, "learning_rate": 0.00019147224296544164, "loss": 1.7792, "step": 1579 }, { "epoch": 0.48, "learning_rate": 0.00019145910620787753, "loss": 1.838, "step": 1580 }, { "epoch": 0.48, "learning_rate": 0.00019144595979106677, "loss": 1.8233, "step": 1581 }, { "epoch": 0.48, "learning_rate": 0.00019143280371639775, "loss": 1.8315, "step": 1582 }, { "epoch": 0.48, "learning_rate": 0.00019141963798525994, "loss": 1.8404, "step": 1583 }, { "epoch": 0.48, "learning_rate": 0.0001914064625990438, "loss": 1.954, "step": 1584 }, { "epoch": 0.48, "learning_rate": 0.00019139327755914078, "loss": 1.8384, "step": 1585 }, { "epoch": 0.48, "learning_rate": 0.00019138008286694348, "loss": 1.7977, "step": 1586 }, { "epoch": 0.48, "learning_rate": 0.00019136687852384533, "loss": 1.9601, "step": 1587 }, { "epoch": 0.48, "learning_rate": 0.00019135366453124092, "loss": 1.8521, "step": 1588 }, { "epoch": 0.48, "learning_rate": 0.00019134044089052582, "loss": 1.7947, "step": 1589 }, { "epoch": 0.48, "learning_rate": 0.0001913272076030966, "loss": 1.8347, "step": 1590 }, { "epoch": 0.48, "learning_rate": 0.00019131396467035093, "loss": 1.9431, "step": 1591 }, { "epoch": 0.48, "learning_rate": 0.00019130071209368734, "loss": 1.8848, "step": 1592 }, { "epoch": 0.48, "learning_rate": 0.00019128744987450551, "loss": 1.7947, "step": 1593 }, { "epoch": 0.48, "learning_rate": 0.0001912741780142061, "loss": 1.8111, "step": 1594 }, { "epoch": 0.48, "learning_rate": 0.00019126089651419075, "loss": 1.8286, "step": 1595 }, { "epoch": 0.48, "learning_rate": 0.00019124760537586223, "loss": 1.7792, "step": 1596 }, { "epoch": 0.48, "learning_rate": 0.0001912343046006242, "loss": 1.9138, "step": 1597 }, { "epoch": 0.48, "learning_rate": 0.0001912209941898814, "loss": 1.8464, "step": 1598 }, { "epoch": 0.48, "learning_rate": 0.00019120767414503958, "loss": 1.7522, "step": 1599 }, { "epoch": 0.48, "learning_rate": 0.00019119434446750552, "loss": 1.8359, "step": 1600 }, { "epoch": 0.48, "learning_rate": 0.000191181005158687, "loss": 1.8365, "step": 1601 }, { "epoch": 0.48, "learning_rate": 0.0001911676562199928, "loss": 1.904, "step": 1602 }, { "epoch": 0.48, "learning_rate": 0.00019115429765283277, "loss": 1.8642, "step": 1603 }, { "epoch": 0.48, "learning_rate": 0.00019114092945861773, "loss": 1.8761, "step": 1604 }, { "epoch": 0.48, "learning_rate": 0.0001911275516387595, "loss": 1.9002, "step": 1605 }, { "epoch": 0.48, "learning_rate": 0.00019111416419467098, "loss": 1.9559, "step": 1606 }, { "epoch": 0.48, "learning_rate": 0.00019110076712776608, "loss": 1.8793, "step": 1607 }, { "epoch": 0.48, "learning_rate": 0.00019108736043945966, "loss": 1.8186, "step": 1608 }, { "epoch": 0.48, "learning_rate": 0.00019107394413116766, "loss": 1.8779, "step": 1609 }, { "epoch": 0.48, "learning_rate": 0.000191060518204307, "loss": 1.8961, "step": 1610 }, { "epoch": 0.48, "learning_rate": 0.00019104708266029562, "loss": 1.8792, "step": 1611 }, { "epoch": 0.49, "learning_rate": 0.0001910336375005525, "loss": 1.7667, "step": 1612 }, { "epoch": 0.49, "learning_rate": 0.00019102018272649759, "loss": 1.8944, "step": 1613 }, { "epoch": 0.49, "learning_rate": 0.00019100671833955195, "loss": 1.7931, "step": 1614 }, { "epoch": 0.49, "learning_rate": 0.00019099324434113755, "loss": 1.9301, "step": 1615 }, { "epoch": 0.49, "learning_rate": 0.0001909797607326774, "loss": 1.8294, "step": 1616 }, { "epoch": 0.49, "learning_rate": 0.00019096626751559557, "loss": 1.9323, "step": 1617 }, { "epoch": 0.49, "learning_rate": 0.0001909527646913171, "loss": 1.9766, "step": 1618 }, { "epoch": 0.49, "learning_rate": 0.00019093925226126808, "loss": 1.8118, "step": 1619 }, { "epoch": 0.49, "learning_rate": 0.00019092573022687557, "loss": 1.7857, "step": 1620 }, { "epoch": 0.49, "learning_rate": 0.00019091219858956765, "loss": 1.8931, "step": 1621 }, { "epoch": 0.49, "learning_rate": 0.00019089865735077353, "loss": 1.8341, "step": 1622 }, { "epoch": 0.49, "learning_rate": 0.0001908851065119232, "loss": 1.8522, "step": 1623 }, { "epoch": 0.49, "learning_rate": 0.0001908715460744479, "loss": 1.8065, "step": 1624 }, { "epoch": 0.49, "learning_rate": 0.00019085797603977977, "loss": 1.9176, "step": 1625 }, { "epoch": 0.49, "learning_rate": 0.00019084439640935195, "loss": 1.8228, "step": 1626 }, { "epoch": 0.49, "learning_rate": 0.00019083080718459864, "loss": 2.0063, "step": 1627 }, { "epoch": 0.49, "learning_rate": 0.00019081720836695504, "loss": 1.8555, "step": 1628 }, { "epoch": 0.49, "learning_rate": 0.00019080359995785734, "loss": 1.8379, "step": 1629 }, { "epoch": 0.49, "learning_rate": 0.00019078998195874277, "loss": 1.8555, "step": 1630 }, { "epoch": 0.49, "learning_rate": 0.00019077635437104957, "loss": 2.0178, "step": 1631 }, { "epoch": 0.49, "learning_rate": 0.000190762717196217, "loss": 2.0489, "step": 1632 }, { "epoch": 0.49, "learning_rate": 0.00019074907043568526, "loss": 1.929, "step": 1633 }, { "epoch": 0.49, "learning_rate": 0.00019073541409089567, "loss": 2.0344, "step": 1634 }, { "epoch": 0.49, "learning_rate": 0.00019072174816329055, "loss": 1.9434, "step": 1635 }, { "epoch": 0.49, "learning_rate": 0.0001907080726543131, "loss": 1.944, "step": 1636 }, { "epoch": 0.49, "learning_rate": 0.00019069438756540773, "loss": 1.9565, "step": 1637 }, { "epoch": 0.49, "learning_rate": 0.00019068069289801968, "loss": 1.8584, "step": 1638 }, { "epoch": 0.49, "learning_rate": 0.00019066698865359533, "loss": 2.0082, "step": 1639 }, { "epoch": 0.49, "learning_rate": 0.000190653274833582, "loss": 1.8921, "step": 1640 }, { "epoch": 0.49, "learning_rate": 0.00019063955143942804, "loss": 1.8321, "step": 1641 }, { "epoch": 0.49, "learning_rate": 0.00019062581847258285, "loss": 1.7651, "step": 1642 }, { "epoch": 0.49, "learning_rate": 0.00019061207593449675, "loss": 1.8712, "step": 1643 }, { "epoch": 0.49, "learning_rate": 0.00019059832382662116, "loss": 1.8575, "step": 1644 }, { "epoch": 0.5, "learning_rate": 0.00019058456215040852, "loss": 1.9277, "step": 1645 }, { "epoch": 0.5, "learning_rate": 0.00019057079090731215, "loss": 1.8923, "step": 1646 }, { "epoch": 0.5, "learning_rate": 0.00019055701009878652, "loss": 1.9333, "step": 1647 }, { "epoch": 0.5, "learning_rate": 0.00019054321972628707, "loss": 1.8333, "step": 1648 }, { "epoch": 0.5, "learning_rate": 0.0001905294197912702, "loss": 1.8237, "step": 1649 }, { "epoch": 0.5, "learning_rate": 0.0001905156102951934, "loss": 1.7779, "step": 1650 }, { "epoch": 0.5, "learning_rate": 0.0001905017912395151, "loss": 1.7848, "step": 1651 }, { "epoch": 0.5, "learning_rate": 0.00019048796262569474, "loss": 1.8912, "step": 1652 }, { "epoch": 0.5, "learning_rate": 0.00019047412445519292, "loss": 1.8597, "step": 1653 }, { "epoch": 0.5, "learning_rate": 0.00019046027672947095, "loss": 1.8259, "step": 1654 }, { "epoch": 0.5, "learning_rate": 0.00019044641944999146, "loss": 1.7864, "step": 1655 }, { "epoch": 0.5, "learning_rate": 0.0001904325526182179, "loss": 1.7719, "step": 1656 }, { "epoch": 0.5, "learning_rate": 0.00019041867623561482, "loss": 1.7868, "step": 1657 }, { "epoch": 0.5, "learning_rate": 0.0001904047903036477, "loss": 1.88, "step": 1658 }, { "epoch": 0.5, "learning_rate": 0.0001903908948237831, "loss": 1.8723, "step": 1659 }, { "epoch": 0.5, "learning_rate": 0.00019037698979748854, "loss": 1.7484, "step": 1660 }, { "epoch": 0.5, "learning_rate": 0.0001903630752262326, "loss": 1.8679, "step": 1661 }, { "epoch": 0.5, "learning_rate": 0.00019034915111148482, "loss": 1.8025, "step": 1662 }, { "epoch": 0.5, "learning_rate": 0.00019033521745471572, "loss": 1.6999, "step": 1663 }, { "epoch": 0.5, "learning_rate": 0.00019032127425739695, "loss": 1.7546, "step": 1664 }, { "epoch": 0.5, "learning_rate": 0.000190307321521001, "loss": 1.7531, "step": 1665 }, { "epoch": 0.5, "learning_rate": 0.00019029335924700155, "loss": 1.8281, "step": 1666 }, { "epoch": 0.5, "learning_rate": 0.00019027938743687312, "loss": 1.8492, "step": 1667 }, { "epoch": 0.5, "learning_rate": 0.00019026540609209139, "loss": 1.923, "step": 1668 }, { "epoch": 0.5, "learning_rate": 0.00019025141521413288, "loss": 1.9932, "step": 1669 }, { "epoch": 0.5, "learning_rate": 0.00019023741480447525, "loss": 1.7831, "step": 1670 }, { "epoch": 0.5, "learning_rate": 0.00019022340486459714, "loss": 1.8328, "step": 1671 }, { "epoch": 0.5, "learning_rate": 0.0001902093853959781, "loss": 1.9235, "step": 1672 }, { "epoch": 0.5, "learning_rate": 0.00019019535640009885, "loss": 1.8701, "step": 1673 }, { "epoch": 0.5, "learning_rate": 0.00019018131787844103, "loss": 1.8621, "step": 1674 }, { "epoch": 0.5, "learning_rate": 0.0001901672698324872, "loss": 1.8654, "step": 1675 }, { "epoch": 0.5, "learning_rate": 0.00019015321226372112, "loss": 1.8825, "step": 1676 }, { "epoch": 0.5, "learning_rate": 0.00019013914517362735, "loss": 1.8518, "step": 1677 }, { "epoch": 0.51, "learning_rate": 0.00019012506856369164, "loss": 1.7592, "step": 1678 }, { "epoch": 0.51, "learning_rate": 0.0001901109824354006, "loss": 1.7872, "step": 1679 }, { "epoch": 0.51, "learning_rate": 0.0001900968867902419, "loss": 1.8039, "step": 1680 }, { "epoch": 0.51, "learning_rate": 0.0001900827816297043, "loss": 1.9706, "step": 1681 }, { "epoch": 0.51, "learning_rate": 0.0001900686669552774, "loss": 2.0024, "step": 1682 }, { "epoch": 0.51, "learning_rate": 0.00019005454276845192, "loss": 1.8613, "step": 1683 }, { "epoch": 0.51, "learning_rate": 0.00019004040907071956, "loss": 1.8122, "step": 1684 }, { "epoch": 0.51, "learning_rate": 0.000190026265863573, "loss": 1.8789, "step": 1685 }, { "epoch": 0.51, "learning_rate": 0.00019001211314850596, "loss": 1.8943, "step": 1686 }, { "epoch": 0.51, "learning_rate": 0.00018999795092701315, "loss": 1.8077, "step": 1687 }, { "epoch": 0.51, "learning_rate": 0.00018998377920059024, "loss": 1.8156, "step": 1688 }, { "epoch": 0.51, "learning_rate": 0.000189969597970734, "loss": 1.7963, "step": 1689 }, { "epoch": 0.51, "learning_rate": 0.00018995540723894215, "loss": 1.8583, "step": 1690 }, { "epoch": 0.51, "learning_rate": 0.00018994120700671334, "loss": 1.8142, "step": 1691 }, { "epoch": 0.51, "learning_rate": 0.00018992699727554734, "loss": 1.8706, "step": 1692 }, { "epoch": 0.51, "learning_rate": 0.0001899127780469449, "loss": 1.8214, "step": 1693 }, { "epoch": 0.51, "learning_rate": 0.00018989854932240777, "loss": 1.8572, "step": 1694 }, { "epoch": 0.51, "learning_rate": 0.0001898843111034386, "loss": 1.8419, "step": 1695 }, { "epoch": 0.51, "learning_rate": 0.0001898700633915412, "loss": 1.8151, "step": 1696 }, { "epoch": 0.51, "learning_rate": 0.00018985580618822027, "loss": 1.8695, "step": 1697 }, { "epoch": 0.51, "learning_rate": 0.00018984153949498157, "loss": 1.7688, "step": 1698 }, { "epoch": 0.51, "learning_rate": 0.00018982726331333186, "loss": 1.7814, "step": 1699 }, { "epoch": 0.51, "learning_rate": 0.00018981297764477883, "loss": 1.8299, "step": 1700 }, { "epoch": 0.51, "learning_rate": 0.0001897986824908313, "loss": 1.9364, "step": 1701 }, { "epoch": 0.51, "learning_rate": 0.000189784377852999, "loss": 1.7798, "step": 1702 }, { "epoch": 0.51, "learning_rate": 0.00018977006373279264, "loss": 2.0306, "step": 1703 }, { "epoch": 0.51, "learning_rate": 0.000189755740131724, "loss": 1.8487, "step": 1704 }, { "epoch": 0.51, "learning_rate": 0.00018974140705130585, "loss": 1.8003, "step": 1705 }, { "epoch": 0.51, "learning_rate": 0.0001897270644930519, "loss": 1.8179, "step": 1706 }, { "epoch": 0.51, "learning_rate": 0.00018971271245847698, "loss": 1.8248, "step": 1707 }, { "epoch": 0.51, "learning_rate": 0.0001896983509490968, "loss": 1.7663, "step": 1708 }, { "epoch": 0.51, "learning_rate": 0.00018968397996642816, "loss": 1.8147, "step": 1709 }, { "epoch": 0.51, "learning_rate": 0.00018966959951198872, "loss": 1.8606, "step": 1710 }, { "epoch": 0.52, "learning_rate": 0.00018965520958729733, "loss": 1.7941, "step": 1711 }, { "epoch": 0.52, "learning_rate": 0.00018964081019387372, "loss": 1.9286, "step": 1712 }, { "epoch": 0.52, "learning_rate": 0.00018962640133323867, "loss": 1.9253, "step": 1713 }, { "epoch": 0.52, "learning_rate": 0.0001896119830069139, "loss": 1.8046, "step": 1714 }, { "epoch": 0.52, "learning_rate": 0.0001895975552164222, "loss": 1.7779, "step": 1715 }, { "epoch": 0.52, "learning_rate": 0.00018958311796328734, "loss": 1.8871, "step": 1716 }, { "epoch": 0.52, "learning_rate": 0.00018956867124903402, "loss": 1.8432, "step": 1717 }, { "epoch": 0.52, "learning_rate": 0.00018955421507518808, "loss": 1.8243, "step": 1718 }, { "epoch": 0.52, "learning_rate": 0.0001895397494432762, "loss": 1.8389, "step": 1719 }, { "epoch": 0.52, "learning_rate": 0.00018952527435482614, "loss": 1.7903, "step": 1720 }, { "epoch": 0.52, "learning_rate": 0.00018951078981136672, "loss": 1.8773, "step": 1721 }, { "epoch": 0.52, "learning_rate": 0.00018949629581442766, "loss": 1.7889, "step": 1722 }, { "epoch": 0.52, "learning_rate": 0.00018948179236553966, "loss": 1.8264, "step": 1723 }, { "epoch": 0.52, "learning_rate": 0.00018946727946623452, "loss": 1.8953, "step": 1724 }, { "epoch": 0.52, "learning_rate": 0.00018945275711804502, "loss": 1.7653, "step": 1725 }, { "epoch": 0.52, "learning_rate": 0.00018943822532250485, "loss": 1.8209, "step": 1726 }, { "epoch": 0.52, "learning_rate": 0.00018942368408114872, "loss": 1.8714, "step": 1727 }, { "epoch": 0.52, "learning_rate": 0.00018940913339551244, "loss": 1.9483, "step": 1728 }, { "epoch": 0.52, "learning_rate": 0.00018939457326713275, "loss": 1.8579, "step": 1729 }, { "epoch": 0.52, "learning_rate": 0.00018938000369754735, "loss": 1.8156, "step": 1730 }, { "epoch": 0.52, "learning_rate": 0.00018936542468829493, "loss": 1.7536, "step": 1731 }, { "epoch": 0.52, "learning_rate": 0.00018935083624091532, "loss": 1.8336, "step": 1732 }, { "epoch": 0.52, "learning_rate": 0.00018933623835694921, "loss": 1.9679, "step": 1733 }, { "epoch": 0.52, "learning_rate": 0.0001893216310379383, "loss": 1.787, "step": 1734 }, { "epoch": 0.52, "learning_rate": 0.00018930701428542527, "loss": 1.7671, "step": 1735 }, { "epoch": 0.52, "learning_rate": 0.00018929238810095394, "loss": 1.7712, "step": 1736 }, { "epoch": 0.52, "learning_rate": 0.00018927775248606894, "loss": 1.9238, "step": 1737 }, { "epoch": 0.52, "learning_rate": 0.00018926310744231598, "loss": 1.9331, "step": 1738 }, { "epoch": 0.52, "learning_rate": 0.0001892484529712418, "loss": 1.8276, "step": 1739 }, { "epoch": 0.52, "learning_rate": 0.00018923378907439412, "loss": 1.8019, "step": 1740 }, { "epoch": 0.52, "learning_rate": 0.00018921911575332155, "loss": 1.9215, "step": 1741 }, { "epoch": 0.52, "learning_rate": 0.00018920443300957386, "loss": 1.901, "step": 1742 }, { "epoch": 0.52, "learning_rate": 0.0001891897408447017, "loss": 1.9093, "step": 1743 }, { "epoch": 0.52, "learning_rate": 0.00018917503926025672, "loss": 1.7924, "step": 1744 }, { "epoch": 0.53, "learning_rate": 0.00018916032825779167, "loss": 1.8544, "step": 1745 }, { "epoch": 0.53, "learning_rate": 0.00018914560783886016, "loss": 1.8799, "step": 1746 }, { "epoch": 0.53, "learning_rate": 0.00018913087800501686, "loss": 1.7185, "step": 1747 }, { "epoch": 0.53, "learning_rate": 0.00018911613875781743, "loss": 1.9403, "step": 1748 }, { "epoch": 0.53, "learning_rate": 0.00018910139009881854, "loss": 1.8181, "step": 1749 }, { "epoch": 0.53, "learning_rate": 0.00018908663202957783, "loss": 1.8675, "step": 1750 }, { "epoch": 0.53, "learning_rate": 0.00018907186455165392, "loss": 2.057, "step": 1751 }, { "epoch": 0.53, "learning_rate": 0.0001890570876666065, "loss": 1.9061, "step": 1752 }, { "epoch": 0.53, "learning_rate": 0.00018904230137599612, "loss": 1.8911, "step": 1753 }, { "epoch": 0.53, "learning_rate": 0.00018902750568138447, "loss": 1.8542, "step": 1754 }, { "epoch": 0.53, "learning_rate": 0.0001890127005843341, "loss": 1.962, "step": 1755 }, { "epoch": 0.53, "learning_rate": 0.00018899788608640864, "loss": 1.9079, "step": 1756 }, { "epoch": 0.53, "learning_rate": 0.00018898306218917275, "loss": 2.0277, "step": 1757 }, { "epoch": 0.53, "learning_rate": 0.00018896822889419193, "loss": 1.9045, "step": 1758 }, { "epoch": 0.53, "learning_rate": 0.00018895338620303283, "loss": 1.8472, "step": 1759 }, { "epoch": 0.53, "learning_rate": 0.00018893853411726296, "loss": 1.824, "step": 1760 }, { "epoch": 0.53, "learning_rate": 0.00018892367263845098, "loss": 1.8974, "step": 1761 }, { "epoch": 0.53, "learning_rate": 0.0001889088017681664, "loss": 1.9359, "step": 1762 }, { "epoch": 0.53, "learning_rate": 0.00018889392150797974, "loss": 1.8852, "step": 1763 }, { "epoch": 0.53, "learning_rate": 0.0001888790318594626, "loss": 1.8394, "step": 1764 }, { "epoch": 0.53, "learning_rate": 0.00018886413282418753, "loss": 1.904, "step": 1765 }, { "epoch": 0.53, "learning_rate": 0.00018884922440372802, "loss": 1.9333, "step": 1766 }, { "epoch": 0.53, "learning_rate": 0.00018883430659965856, "loss": 2.0928, "step": 1767 }, { "epoch": 0.53, "learning_rate": 0.0001888193794135547, "loss": 1.8942, "step": 1768 }, { "epoch": 0.53, "learning_rate": 0.00018880444284699298, "loss": 1.8815, "step": 1769 }, { "epoch": 0.53, "learning_rate": 0.0001887894969015508, "loss": 1.809, "step": 1770 }, { "epoch": 0.53, "learning_rate": 0.0001887745415788067, "loss": 1.9178, "step": 1771 }, { "epoch": 0.53, "learning_rate": 0.00018875957688034017, "loss": 1.8714, "step": 1772 }, { "epoch": 0.53, "learning_rate": 0.00018874460280773162, "loss": 1.9856, "step": 1773 }, { "epoch": 0.53, "learning_rate": 0.00018872961936256254, "loss": 1.8133, "step": 1774 }, { "epoch": 0.53, "learning_rate": 0.00018871462654641537, "loss": 1.8763, "step": 1775 }, { "epoch": 0.53, "learning_rate": 0.00018869962436087354, "loss": 1.7779, "step": 1776 }, { "epoch": 0.53, "learning_rate": 0.00018868461280752144, "loss": 1.796, "step": 1777 }, { "epoch": 0.54, "learning_rate": 0.00018866959188794453, "loss": 1.8879, "step": 1778 }, { "epoch": 0.54, "learning_rate": 0.0001886545616037292, "loss": 1.7837, "step": 1779 }, { "epoch": 0.54, "learning_rate": 0.00018863952195646277, "loss": 1.9073, "step": 1780 }, { "epoch": 0.54, "learning_rate": 0.00018862447294773374, "loss": 1.8617, "step": 1781 }, { "epoch": 0.54, "learning_rate": 0.00018860941457913137, "loss": 1.9263, "step": 1782 }, { "epoch": 0.54, "learning_rate": 0.00018859434685224605, "loss": 1.903, "step": 1783 }, { "epoch": 0.54, "learning_rate": 0.0001885792697686692, "loss": 1.872, "step": 1784 }, { "epoch": 0.54, "learning_rate": 0.000188564183329993, "loss": 1.8261, "step": 1785 }, { "epoch": 0.54, "learning_rate": 0.00018854908753781092, "loss": 1.9934, "step": 1786 }, { "epoch": 0.54, "learning_rate": 0.0001885339823937172, "loss": 2.02, "step": 1787 }, { "epoch": 0.54, "learning_rate": 0.0001885188678993071, "loss": 1.852, "step": 1788 }, { "epoch": 0.54, "learning_rate": 0.00018850374405617694, "loss": 1.9028, "step": 1789 }, { "epoch": 0.54, "learning_rate": 0.000188488610865924, "loss": 1.8188, "step": 1790 }, { "epoch": 0.54, "learning_rate": 0.00018847346833014656, "loss": 1.9136, "step": 1791 }, { "epoch": 0.54, "learning_rate": 0.0001884583164504438, "loss": 1.9572, "step": 1792 }, { "epoch": 0.54, "learning_rate": 0.000188443155228416, "loss": 1.8964, "step": 1793 }, { "epoch": 0.54, "learning_rate": 0.00018842798466566437, "loss": 1.8917, "step": 1794 }, { "epoch": 0.54, "learning_rate": 0.0001884128047637911, "loss": 1.878, "step": 1795 }, { "epoch": 0.54, "learning_rate": 0.00018839761552439939, "loss": 1.8764, "step": 1796 }, { "epoch": 0.54, "learning_rate": 0.0001883824169490934, "loss": 1.8151, "step": 1797 }, { "epoch": 0.54, "learning_rate": 0.00018836720903947832, "loss": 1.8599, "step": 1798 }, { "epoch": 0.54, "learning_rate": 0.0001883519917971603, "loss": 1.7819, "step": 1799 }, { "epoch": 0.54, "learning_rate": 0.00018833676522374645, "loss": 1.891, "step": 1800 }, { "epoch": 0.54, "learning_rate": 0.00018832152932084492, "loss": 2.0407, "step": 1801 }, { "epoch": 0.54, "learning_rate": 0.0001883062840900648, "loss": 1.8906, "step": 1802 }, { "epoch": 0.54, "learning_rate": 0.00018829102953301616, "loss": 1.8575, "step": 1803 }, { "epoch": 0.54, "learning_rate": 0.0001882757656513101, "loss": 1.9369, "step": 1804 }, { "epoch": 0.54, "learning_rate": 0.00018826049244655867, "loss": 1.911, "step": 1805 }, { "epoch": 0.54, "learning_rate": 0.00018824520992037494, "loss": 1.814, "step": 1806 }, { "epoch": 0.54, "learning_rate": 0.00018822991807437292, "loss": 1.7911, "step": 1807 }, { "epoch": 0.54, "learning_rate": 0.0001882146169101676, "loss": 1.9042, "step": 1808 }, { "epoch": 0.54, "learning_rate": 0.000188199306429375, "loss": 1.8968, "step": 1809 }, { "epoch": 0.54, "learning_rate": 0.00018818398663361213, "loss": 1.7862, "step": 1810 }, { "epoch": 0.55, "learning_rate": 0.00018816865752449693, "loss": 1.7666, "step": 1811 }, { "epoch": 0.55, "learning_rate": 0.00018815331910364833, "loss": 1.7516, "step": 1812 }, { "epoch": 0.55, "learning_rate": 0.00018813797137268628, "loss": 1.829, "step": 1813 }, { "epoch": 0.55, "learning_rate": 0.0001881226143332317, "loss": 1.9482, "step": 1814 }, { "epoch": 0.55, "learning_rate": 0.00018810724798690647, "loss": 1.9379, "step": 1815 }, { "epoch": 0.55, "learning_rate": 0.00018809187233533348, "loss": 1.8817, "step": 1816 }, { "epoch": 0.55, "learning_rate": 0.00018807648738013665, "loss": 1.8566, "step": 1817 }, { "epoch": 0.55, "learning_rate": 0.00018806109312294076, "loss": 1.7999, "step": 1818 }, { "epoch": 0.55, "learning_rate": 0.00018804568956537165, "loss": 1.9303, "step": 1819 }, { "epoch": 0.55, "learning_rate": 0.00018803027670905614, "loss": 1.8848, "step": 1820 }, { "epoch": 0.55, "learning_rate": 0.00018801485455562203, "loss": 1.9238, "step": 1821 }, { "epoch": 0.55, "learning_rate": 0.00018799942310669812, "loss": 1.8437, "step": 1822 }, { "epoch": 0.55, "learning_rate": 0.0001879839823639141, "loss": 1.749, "step": 1823 }, { "epoch": 0.55, "learning_rate": 0.00018796853232890074, "loss": 1.781, "step": 1824 }, { "epoch": 0.55, "learning_rate": 0.00018795307300328978, "loss": 1.8459, "step": 1825 }, { "epoch": 0.55, "learning_rate": 0.00018793760438871394, "loss": 1.7788, "step": 1826 }, { "epoch": 0.55, "learning_rate": 0.00018792212648680682, "loss": 1.7803, "step": 1827 }, { "epoch": 0.55, "learning_rate": 0.00018790663929920317, "loss": 1.8109, "step": 1828 }, { "epoch": 0.55, "learning_rate": 0.0001878911428275386, "loss": 1.8352, "step": 1829 }, { "epoch": 0.55, "learning_rate": 0.0001878756370734497, "loss": 1.7993, "step": 1830 }, { "epoch": 0.55, "learning_rate": 0.00018786012203857416, "loss": 1.8239, "step": 1831 }, { "epoch": 0.55, "learning_rate": 0.0001878445977245505, "loss": 1.9187, "step": 1832 }, { "epoch": 0.55, "learning_rate": 0.0001878290641330183, "loss": 1.8598, "step": 1833 }, { "epoch": 0.55, "learning_rate": 0.0001878135212656181, "loss": 1.9734, "step": 1834 }, { "epoch": 0.55, "learning_rate": 0.00018779796912399142, "loss": 1.8123, "step": 1835 }, { "epoch": 0.55, "learning_rate": 0.00018778240770978078, "loss": 1.8649, "step": 1836 }, { "epoch": 0.55, "learning_rate": 0.00018776683702462967, "loss": 1.8312, "step": 1837 }, { "epoch": 0.55, "learning_rate": 0.00018775125707018252, "loss": 1.8689, "step": 1838 }, { "epoch": 0.55, "learning_rate": 0.00018773566784808482, "loss": 1.8137, "step": 1839 }, { "epoch": 0.55, "learning_rate": 0.00018772006935998295, "loss": 1.786, "step": 1840 }, { "epoch": 0.55, "learning_rate": 0.00018770446160752433, "loss": 1.7658, "step": 1841 }, { "epoch": 0.55, "learning_rate": 0.00018768884459235732, "loss": 1.7365, "step": 1842 }, { "epoch": 0.55, "learning_rate": 0.0001876732183161313, "loss": 1.8872, "step": 1843 }, { "epoch": 0.56, "learning_rate": 0.00018765758278049659, "loss": 1.8068, "step": 1844 }, { "epoch": 0.56, "learning_rate": 0.00018764193798710447, "loss": 1.8689, "step": 1845 }, { "epoch": 0.56, "learning_rate": 0.0001876262839376073, "loss": 1.8595, "step": 1846 }, { "epoch": 0.56, "learning_rate": 0.0001876106206336583, "loss": 2.0084, "step": 1847 }, { "epoch": 0.56, "learning_rate": 0.0001875949480769117, "loss": 1.8829, "step": 1848 }, { "epoch": 0.56, "learning_rate": 0.00018757926626902275, "loss": 1.8777, "step": 1849 }, { "epoch": 0.56, "learning_rate": 0.00018756357521164767, "loss": 1.8809, "step": 1850 }, { "epoch": 0.56, "learning_rate": 0.00018754787490644356, "loss": 1.9073, "step": 1851 }, { "epoch": 0.56, "learning_rate": 0.00018753216535506866, "loss": 1.8605, "step": 1852 }, { "epoch": 0.56, "learning_rate": 0.00018751644655918206, "loss": 1.9434, "step": 1853 }, { "epoch": 0.56, "learning_rate": 0.00018750071852044383, "loss": 1.7358, "step": 1854 }, { "epoch": 0.56, "learning_rate": 0.0001874849812405151, "loss": 1.8228, "step": 1855 }, { "epoch": 0.56, "learning_rate": 0.00018746923472105794, "loss": 1.8338, "step": 1856 }, { "epoch": 0.56, "learning_rate": 0.00018745347896373531, "loss": 1.6811, "step": 1857 }, { "epoch": 0.56, "learning_rate": 0.0001874377139702113, "loss": 1.8149, "step": 1858 }, { "epoch": 0.56, "learning_rate": 0.00018742193974215086, "loss": 1.7734, "step": 1859 }, { "epoch": 0.56, "learning_rate": 0.00018740615628121994, "loss": 1.9525, "step": 1860 }, { "epoch": 0.56, "learning_rate": 0.00018739036358908547, "loss": 1.8655, "step": 1861 }, { "epoch": 0.56, "learning_rate": 0.00018737456166741535, "loss": 1.9252, "step": 1862 }, { "epoch": 0.56, "learning_rate": 0.00018735875051787852, "loss": 1.8027, "step": 1863 }, { "epoch": 0.56, "learning_rate": 0.00018734293014214482, "loss": 1.8185, "step": 1864 }, { "epoch": 0.56, "learning_rate": 0.00018732710054188506, "loss": 1.8976, "step": 1865 }, { "epoch": 0.56, "learning_rate": 0.00018731126171877101, "loss": 1.8549, "step": 1866 }, { "epoch": 0.56, "learning_rate": 0.00018729541367447556, "loss": 1.7633, "step": 1867 }, { "epoch": 0.56, "learning_rate": 0.00018727955641067236, "loss": 1.8594, "step": 1868 }, { "epoch": 0.56, "learning_rate": 0.0001872636899290362, "loss": 1.772, "step": 1869 }, { "epoch": 0.56, "learning_rate": 0.0001872478142312428, "loss": 1.9073, "step": 1870 }, { "epoch": 0.56, "learning_rate": 0.00018723192931896873, "loss": 2.0359, "step": 1871 }, { "epoch": 0.56, "learning_rate": 0.00018721603519389172, "loss": 1.7198, "step": 1872 }, { "epoch": 0.56, "learning_rate": 0.0001872001318576904, "loss": 1.7859, "step": 1873 }, { "epoch": 0.56, "learning_rate": 0.00018718421931204437, "loss": 1.8653, "step": 1874 }, { "epoch": 0.56, "learning_rate": 0.00018716829755863415, "loss": 1.8709, "step": 1875 }, { "epoch": 0.56, "learning_rate": 0.00018715236659914132, "loss": 1.9391, "step": 1876 }, { "epoch": 0.57, "learning_rate": 0.00018713642643524832, "loss": 1.8018, "step": 1877 }, { "epoch": 0.57, "learning_rate": 0.00018712047706863873, "loss": 1.7676, "step": 1878 }, { "epoch": 0.57, "learning_rate": 0.00018710451850099697, "loss": 1.8324, "step": 1879 }, { "epoch": 0.57, "learning_rate": 0.00018708855073400847, "loss": 1.8365, "step": 1880 }, { "epoch": 0.57, "learning_rate": 0.0001870725737693596, "loss": 1.9214, "step": 1881 }, { "epoch": 0.57, "learning_rate": 0.0001870565876087378, "loss": 1.7113, "step": 1882 }, { "epoch": 0.57, "learning_rate": 0.00018704059225383132, "loss": 1.7797, "step": 1883 }, { "epoch": 0.57, "learning_rate": 0.0001870245877063295, "loss": 1.7899, "step": 1884 }, { "epoch": 0.57, "learning_rate": 0.0001870085739679227, "loss": 1.822, "step": 1885 }, { "epoch": 0.57, "learning_rate": 0.00018699255104030212, "loss": 1.9291, "step": 1886 }, { "epoch": 0.57, "learning_rate": 0.00018697651892515994, "loss": 1.8648, "step": 1887 }, { "epoch": 0.57, "learning_rate": 0.00018696047762418945, "loss": 1.8474, "step": 1888 }, { "epoch": 0.57, "learning_rate": 0.00018694442713908474, "loss": 1.8378, "step": 1889 }, { "epoch": 0.57, "learning_rate": 0.00018692836747154096, "loss": 1.8236, "step": 1890 }, { "epoch": 0.57, "learning_rate": 0.00018691229862325426, "loss": 1.8105, "step": 1891 }, { "epoch": 0.57, "learning_rate": 0.00018689622059592164, "loss": 1.8469, "step": 1892 }, { "epoch": 0.57, "learning_rate": 0.00018688013339124125, "loss": 1.8452, "step": 1893 }, { "epoch": 0.57, "learning_rate": 0.00018686403701091199, "loss": 1.7472, "step": 1894 }, { "epoch": 0.57, "learning_rate": 0.00018684793145663388, "loss": 1.8237, "step": 1895 }, { "epoch": 0.57, "learning_rate": 0.00018683181673010793, "loss": 1.7897, "step": 1896 }, { "epoch": 0.57, "learning_rate": 0.00018681569283303598, "loss": 1.8381, "step": 1897 }, { "epoch": 0.57, "learning_rate": 0.00018679955976712095, "loss": 1.7532, "step": 1898 }, { "epoch": 0.57, "learning_rate": 0.00018678341753406675, "loss": 1.804, "step": 1899 }, { "epoch": 0.57, "learning_rate": 0.0001867672661355781, "loss": 1.8067, "step": 1900 }, { "epoch": 0.57, "learning_rate": 0.00018675110557336087, "loss": 1.8826, "step": 1901 }, { "epoch": 0.57, "learning_rate": 0.00018673493584912183, "loss": 1.8324, "step": 1902 }, { "epoch": 0.57, "learning_rate": 0.00018671875696456866, "loss": 1.7797, "step": 1903 }, { "epoch": 0.57, "learning_rate": 0.00018670256892141008, "loss": 2.0067, "step": 1904 }, { "epoch": 0.57, "learning_rate": 0.00018668637172135574, "loss": 1.9743, "step": 1905 }, { "epoch": 0.57, "learning_rate": 0.0001866701653661163, "loss": 1.7849, "step": 1906 }, { "epoch": 0.57, "learning_rate": 0.00018665394985740332, "loss": 1.7761, "step": 1907 }, { "epoch": 0.57, "learning_rate": 0.00018663772519692939, "loss": 1.983, "step": 1908 }, { "epoch": 0.57, "learning_rate": 0.00018662149138640803, "loss": 1.8518, "step": 1909 }, { "epoch": 0.57, "learning_rate": 0.00018660524842755374, "loss": 1.8735, "step": 1910 }, { "epoch": 0.58, "learning_rate": 0.00018658899632208198, "loss": 1.8242, "step": 1911 }, { "epoch": 0.58, "learning_rate": 0.0001865727350717092, "loss": 1.7775, "step": 1912 }, { "epoch": 0.58, "learning_rate": 0.00018655646467815275, "loss": 1.8894, "step": 1913 }, { "epoch": 0.58, "learning_rate": 0.00018654018514313106, "loss": 1.7644, "step": 1914 }, { "epoch": 0.58, "learning_rate": 0.0001865238964683634, "loss": 1.8245, "step": 1915 }, { "epoch": 0.58, "learning_rate": 0.00018650759865557008, "loss": 1.8027, "step": 1916 }, { "epoch": 0.58, "learning_rate": 0.00018649129170647232, "loss": 1.7803, "step": 1917 }, { "epoch": 0.58, "learning_rate": 0.0001864749756227924, "loss": 1.8209, "step": 1918 }, { "epoch": 0.58, "learning_rate": 0.0001864586504062535, "loss": 1.8179, "step": 1919 }, { "epoch": 0.58, "learning_rate": 0.00018644231605857977, "loss": 1.8053, "step": 1920 }, { "epoch": 0.58, "learning_rate": 0.00018642597258149626, "loss": 1.9129, "step": 1921 }, { "epoch": 0.58, "learning_rate": 0.0001864096199767291, "loss": 1.9443, "step": 1922 }, { "epoch": 0.58, "learning_rate": 0.00018639325824600534, "loss": 1.8059, "step": 1923 }, { "epoch": 0.58, "learning_rate": 0.00018637688739105298, "loss": 1.8977, "step": 1924 }, { "epoch": 0.58, "learning_rate": 0.00018636050741360102, "loss": 1.9663, "step": 1925 }, { "epoch": 0.58, "learning_rate": 0.0001863441183153793, "loss": 1.8649, "step": 1926 }, { "epoch": 0.58, "learning_rate": 0.00018632772009811882, "loss": 1.9222, "step": 1927 }, { "epoch": 0.58, "learning_rate": 0.0001863113127635514, "loss": 1.8277, "step": 1928 }, { "epoch": 0.58, "learning_rate": 0.00018629489631340987, "loss": 1.994, "step": 1929 }, { "epoch": 0.58, "learning_rate": 0.000186278470749428, "loss": 1.8144, "step": 1930 }, { "epoch": 0.58, "learning_rate": 0.00018626203607334055, "loss": 1.9529, "step": 1931 }, { "epoch": 0.58, "learning_rate": 0.00018624559228688322, "loss": 1.8224, "step": 1932 }, { "epoch": 0.58, "learning_rate": 0.00018622913939179273, "loss": 1.8073, "step": 1933 }, { "epoch": 0.58, "learning_rate": 0.00018621267738980663, "loss": 1.7147, "step": 1934 }, { "epoch": 0.58, "learning_rate": 0.0001861962062826636, "loss": 1.8694, "step": 1935 }, { "epoch": 0.58, "learning_rate": 0.00018617972607210318, "loss": 1.8755, "step": 1936 }, { "epoch": 0.58, "learning_rate": 0.00018616323675986582, "loss": 1.9292, "step": 1937 }, { "epoch": 0.58, "learning_rate": 0.0001861467383476931, "loss": 1.9914, "step": 1938 }, { "epoch": 0.58, "learning_rate": 0.00018613023083732742, "loss": 1.8753, "step": 1939 }, { "epoch": 0.58, "learning_rate": 0.00018611371423051215, "loss": 1.8551, "step": 1940 }, { "epoch": 0.58, "learning_rate": 0.0001860971885289917, "loss": 1.7502, "step": 1941 }, { "epoch": 0.58, "learning_rate": 0.0001860806537345114, "loss": 1.9084, "step": 1942 }, { "epoch": 0.58, "learning_rate": 0.00018606410984881751, "loss": 1.7886, "step": 1943 }, { "epoch": 0.59, "learning_rate": 0.00018604755687365727, "loss": 1.8279, "step": 1944 }, { "epoch": 0.59, "learning_rate": 0.00018603099481077893, "loss": 1.8587, "step": 1945 }, { "epoch": 0.59, "learning_rate": 0.00018601442366193157, "loss": 1.7736, "step": 1946 }, { "epoch": 0.59, "learning_rate": 0.0001859978434288654, "loss": 1.7647, "step": 1947 }, { "epoch": 0.59, "learning_rate": 0.0001859812541133315, "loss": 1.8117, "step": 1948 }, { "epoch": 0.59, "learning_rate": 0.0001859646557170818, "loss": 1.8847, "step": 1949 }, { "epoch": 0.59, "learning_rate": 0.00018594804824186945, "loss": 1.8065, "step": 1950 }, { "epoch": 0.59, "learning_rate": 0.0001859314316894483, "loss": 1.7848, "step": 1951 }, { "epoch": 0.59, "learning_rate": 0.00018591480606157336, "loss": 1.8196, "step": 1952 }, { "epoch": 0.59, "learning_rate": 0.00018589817136000045, "loss": 1.876, "step": 1953 }, { "epoch": 0.59, "learning_rate": 0.00018588152758648643, "loss": 1.8844, "step": 1954 }, { "epoch": 0.59, "learning_rate": 0.00018586487474278905, "loss": 1.896, "step": 1955 }, { "epoch": 0.59, "learning_rate": 0.0001858482128306671, "loss": 1.7703, "step": 1956 }, { "epoch": 0.59, "learning_rate": 0.00018583154185188033, "loss": 1.7844, "step": 1957 }, { "epoch": 0.59, "learning_rate": 0.0001858148618081893, "loss": 1.907, "step": 1958 }, { "epoch": 0.59, "learning_rate": 0.00018579817270135575, "loss": 1.72, "step": 1959 }, { "epoch": 0.59, "learning_rate": 0.00018578147453314223, "loss": 1.8266, "step": 1960 }, { "epoch": 0.59, "learning_rate": 0.00018576476730531222, "loss": 1.8459, "step": 1961 }, { "epoch": 0.59, "learning_rate": 0.00018574805101963026, "loss": 1.9048, "step": 1962 }, { "epoch": 0.59, "learning_rate": 0.0001857313256778618, "loss": 1.92, "step": 1963 }, { "epoch": 0.59, "learning_rate": 0.00018571459128177326, "loss": 1.9817, "step": 1964 }, { "epoch": 0.59, "learning_rate": 0.000185697847833132, "loss": 1.78, "step": 1965 }, { "epoch": 0.59, "learning_rate": 0.00018568109533370633, "loss": 1.8526, "step": 1966 }, { "epoch": 0.59, "learning_rate": 0.00018566433378526553, "loss": 1.8958, "step": 1967 }, { "epoch": 0.59, "learning_rate": 0.00018564756318957984, "loss": 1.8984, "step": 1968 }, { "epoch": 0.59, "learning_rate": 0.00018563078354842042, "loss": 1.877, "step": 1969 }, { "epoch": 0.59, "learning_rate": 0.00018561399486355948, "loss": 1.7208, "step": 1970 }, { "epoch": 0.59, "learning_rate": 0.00018559719713677003, "loss": 1.774, "step": 1971 }, { "epoch": 0.59, "learning_rate": 0.00018558039036982625, "loss": 1.831, "step": 1972 }, { "epoch": 0.59, "learning_rate": 0.000185563574564503, "loss": 1.7796, "step": 1973 }, { "epoch": 0.59, "learning_rate": 0.00018554674972257634, "loss": 1.9271, "step": 1974 }, { "epoch": 0.59, "learning_rate": 0.00018552991584582315, "loss": 2.0338, "step": 1975 }, { "epoch": 0.59, "learning_rate": 0.00018551307293602131, "loss": 1.8516, "step": 1976 }, { "epoch": 0.6, "learning_rate": 0.00018549622099494966, "loss": 1.8573, "step": 1977 }, { "epoch": 0.6, "learning_rate": 0.00018547936002438796, "loss": 1.8634, "step": 1978 }, { "epoch": 0.6, "learning_rate": 0.00018546249002611694, "loss": 1.7919, "step": 1979 }, { "epoch": 0.6, "learning_rate": 0.00018544561100191834, "loss": 1.7628, "step": 1980 }, { "epoch": 0.6, "learning_rate": 0.00018542872295357473, "loss": 1.9249, "step": 1981 }, { "epoch": 0.6, "learning_rate": 0.00018541182588286973, "loss": 1.8133, "step": 1982 }, { "epoch": 0.6, "learning_rate": 0.00018539491979158789, "loss": 1.952, "step": 1983 }, { "epoch": 0.6, "learning_rate": 0.00018537800468151474, "loss": 1.8465, "step": 1984 }, { "epoch": 0.6, "learning_rate": 0.00018536108055443667, "loss": 1.8371, "step": 1985 }, { "epoch": 0.6, "learning_rate": 0.0001853441474121411, "loss": 1.9247, "step": 1986 }, { "epoch": 0.6, "learning_rate": 0.0001853272052564164, "loss": 1.9356, "step": 1987 }, { "epoch": 0.6, "learning_rate": 0.00018531025408905188, "loss": 1.8916, "step": 1988 }, { "epoch": 0.6, "learning_rate": 0.0001852932939118378, "loss": 1.8636, "step": 1989 }, { "epoch": 0.6, "learning_rate": 0.00018527632472656536, "loss": 1.7114, "step": 1990 }, { "epoch": 0.6, "learning_rate": 0.0001852593465350267, "loss": 1.7608, "step": 1991 }, { "epoch": 0.6, "learning_rate": 0.000185242359339015, "loss": 1.8318, "step": 1992 }, { "epoch": 0.6, "learning_rate": 0.00018522536314032422, "loss": 1.8097, "step": 1993 }, { "epoch": 0.6, "learning_rate": 0.00018520835794074952, "loss": 1.89, "step": 1994 }, { "epoch": 0.6, "learning_rate": 0.00018519134374208674, "loss": 1.8952, "step": 1995 }, { "epoch": 0.6, "learning_rate": 0.00018517432054613285, "loss": 1.8439, "step": 1996 }, { "epoch": 0.6, "learning_rate": 0.0001851572883546857, "loss": 1.8106, "step": 1997 }, { "epoch": 0.6, "learning_rate": 0.00018514024716954412, "loss": 1.8626, "step": 1998 }, { "epoch": 0.6, "learning_rate": 0.00018512319699250787, "loss": 1.903, "step": 1999 }, { "epoch": 0.6, "learning_rate": 0.00018510613782537764, "loss": 1.89, "step": 2000 }, { "epoch": 0.6, "learning_rate": 0.00018508906966995515, "loss": 1.9009, "step": 2001 }, { "epoch": 0.6, "learning_rate": 0.000185071992528043, "loss": 1.8856, "step": 2002 }, { "epoch": 0.6, "learning_rate": 0.00018505490640144468, "loss": 1.8851, "step": 2003 }, { "epoch": 0.6, "learning_rate": 0.0001850378112919648, "loss": 1.8742, "step": 2004 }, { "epoch": 0.6, "learning_rate": 0.00018502070720140878, "loss": 1.9471, "step": 2005 }, { "epoch": 0.6, "learning_rate": 0.00018500359413158303, "loss": 1.8559, "step": 2006 }, { "epoch": 0.6, "learning_rate": 0.0001849864720842949, "loss": 1.8054, "step": 2007 }, { "epoch": 0.6, "learning_rate": 0.00018496934106135274, "loss": 1.7234, "step": 2008 }, { "epoch": 0.6, "learning_rate": 0.00018495220106456577, "loss": 1.7381, "step": 2009 }, { "epoch": 0.61, "learning_rate": 0.00018493505209574414, "loss": 1.9354, "step": 2010 }, { "epoch": 0.61, "learning_rate": 0.0001849178941566991, "loss": 1.8872, "step": 2011 }, { "epoch": 0.61, "learning_rate": 0.00018490072724924267, "loss": 1.9443, "step": 2012 }, { "epoch": 0.61, "learning_rate": 0.00018488355137518796, "loss": 1.7474, "step": 2013 }, { "epoch": 0.61, "learning_rate": 0.0001848663665363489, "loss": 1.9264, "step": 2014 }, { "epoch": 0.61, "learning_rate": 0.00018484917273454046, "loss": 1.8856, "step": 2015 }, { "epoch": 0.61, "learning_rate": 0.00018483196997157853, "loss": 1.8355, "step": 2016 }, { "epoch": 0.61, "learning_rate": 0.0001848147582492799, "loss": 1.8301, "step": 2017 }, { "epoch": 0.61, "learning_rate": 0.0001847975375694624, "loss": 1.8582, "step": 2018 }, { "epoch": 0.61, "learning_rate": 0.00018478030793394473, "loss": 1.8868, "step": 2019 }, { "epoch": 0.61, "learning_rate": 0.00018476306934454652, "loss": 1.8531, "step": 2020 }, { "epoch": 0.61, "learning_rate": 0.00018474582180308846, "loss": 1.8964, "step": 2021 }, { "epoch": 0.61, "learning_rate": 0.00018472856531139207, "loss": 1.9164, "step": 2022 }, { "epoch": 0.61, "learning_rate": 0.00018471129987127985, "loss": 1.9545, "step": 2023 }, { "epoch": 0.61, "learning_rate": 0.00018469402548457526, "loss": 1.8683, "step": 2024 }, { "epoch": 0.61, "learning_rate": 0.0001846767421531027, "loss": 1.8449, "step": 2025 }, { "epoch": 0.61, "learning_rate": 0.00018465944987868747, "loss": 1.8223, "step": 2026 }, { "epoch": 0.61, "learning_rate": 0.0001846421486631559, "loss": 1.8818, "step": 2027 }, { "epoch": 0.61, "learning_rate": 0.00018462483850833523, "loss": 1.8577, "step": 2028 }, { "epoch": 0.61, "learning_rate": 0.00018460751941605361, "loss": 1.9441, "step": 2029 }, { "epoch": 0.61, "learning_rate": 0.00018459019138814014, "loss": 1.8507, "step": 2030 }, { "epoch": 0.61, "learning_rate": 0.00018457285442642488, "loss": 1.8989, "step": 2031 }, { "epoch": 0.61, "learning_rate": 0.00018455550853273888, "loss": 1.8487, "step": 2032 }, { "epoch": 0.61, "learning_rate": 0.00018453815370891405, "loss": 1.8104, "step": 2033 }, { "epoch": 0.61, "learning_rate": 0.0001845207899567833, "loss": 1.7926, "step": 2034 }, { "epoch": 0.61, "learning_rate": 0.00018450341727818044, "loss": 1.7436, "step": 2035 }, { "epoch": 0.61, "learning_rate": 0.00018448603567494025, "loss": 1.7979, "step": 2036 }, { "epoch": 0.61, "learning_rate": 0.00018446864514889847, "loss": 1.9445, "step": 2037 }, { "epoch": 0.61, "learning_rate": 0.00018445124570189175, "loss": 1.8665, "step": 2038 }, { "epoch": 0.61, "learning_rate": 0.00018443383733575766, "loss": 1.7783, "step": 2039 }, { "epoch": 0.61, "learning_rate": 0.00018441642005233478, "loss": 1.9332, "step": 2040 }, { "epoch": 0.61, "learning_rate": 0.00018439899385346264, "loss": 1.7209, "step": 2041 }, { "epoch": 0.61, "learning_rate": 0.0001843815587409816, "loss": 1.9202, "step": 2042 }, { "epoch": 0.61, "learning_rate": 0.00018436411471673306, "loss": 1.7945, "step": 2043 }, { "epoch": 0.62, "learning_rate": 0.00018434666178255928, "loss": 1.9197, "step": 2044 }, { "epoch": 0.62, "learning_rate": 0.0001843291999403036, "loss": 1.6882, "step": 2045 }, { "epoch": 0.62, "learning_rate": 0.00018431172919181017, "loss": 1.8061, "step": 2046 }, { "epoch": 0.62, "learning_rate": 0.00018429424953892408, "loss": 1.7609, "step": 2047 }, { "epoch": 0.62, "learning_rate": 0.00018427676098349148, "loss": 1.8238, "step": 2048 }, { "epoch": 0.62, "learning_rate": 0.00018425926352735937, "loss": 1.8255, "step": 2049 }, { "epoch": 0.62, "learning_rate": 0.0001842417571723756, "loss": 1.8492, "step": 2050 }, { "epoch": 0.62, "learning_rate": 0.00018422424192038924, "loss": 1.8423, "step": 2051 }, { "epoch": 0.62, "learning_rate": 0.00018420671777325, "loss": 1.8244, "step": 2052 }, { "epoch": 0.62, "learning_rate": 0.0001841891847328087, "loss": 1.7361, "step": 2053 }, { "epoch": 0.62, "learning_rate": 0.000184171642800917, "loss": 1.8581, "step": 2054 }, { "epoch": 0.62, "learning_rate": 0.0001841540919794276, "loss": 1.7971, "step": 2055 }, { "epoch": 0.62, "learning_rate": 0.0001841365322701941, "loss": 1.767, "step": 2056 }, { "epoch": 0.62, "learning_rate": 0.000184118963675071, "loss": 1.7598, "step": 2057 }, { "epoch": 0.62, "learning_rate": 0.00018410138619591378, "loss": 1.8576, "step": 2058 }, { "epoch": 0.62, "learning_rate": 0.00018408379983457878, "loss": 1.8529, "step": 2059 }, { "epoch": 0.62, "learning_rate": 0.0001840662045929235, "loss": 1.7433, "step": 2060 }, { "epoch": 0.62, "learning_rate": 0.0001840486004728061, "loss": 1.8469, "step": 2061 }, { "epoch": 0.62, "learning_rate": 0.0001840309874760858, "loss": 1.7898, "step": 2062 }, { "epoch": 0.62, "learning_rate": 0.0001840133656046228, "loss": 1.8254, "step": 2063 }, { "epoch": 0.62, "learning_rate": 0.00018399573486027818, "loss": 1.7923, "step": 2064 }, { "epoch": 0.62, "learning_rate": 0.00018397809524491393, "loss": 1.778, "step": 2065 }, { "epoch": 0.62, "learning_rate": 0.0001839604467603931, "loss": 1.9774, "step": 2066 }, { "epoch": 0.62, "learning_rate": 0.00018394278940857953, "loss": 1.8065, "step": 2067 }, { "epoch": 0.62, "learning_rate": 0.0001839251231913381, "loss": 1.8276, "step": 2068 }, { "epoch": 0.62, "learning_rate": 0.00018390744811053457, "loss": 1.7927, "step": 2069 }, { "epoch": 0.62, "learning_rate": 0.00018388976416803567, "loss": 1.8196, "step": 2070 }, { "epoch": 0.62, "learning_rate": 0.000183872071365709, "loss": 1.7421, "step": 2071 }, { "epoch": 0.62, "learning_rate": 0.00018385436970542318, "loss": 2.0023, "step": 2072 }, { "epoch": 0.62, "learning_rate": 0.00018383665918904774, "loss": 1.8064, "step": 2073 }, { "epoch": 0.62, "learning_rate": 0.00018381893981845313, "loss": 1.7154, "step": 2074 }, { "epoch": 0.62, "learning_rate": 0.00018380121159551074, "loss": 1.8659, "step": 2075 }, { "epoch": 0.62, "learning_rate": 0.0001837834745220929, "loss": 1.8486, "step": 2076 }, { "epoch": 0.63, "learning_rate": 0.0001837657286000728, "loss": 1.8456, "step": 2077 }, { "epoch": 0.63, "learning_rate": 0.00018374797383132477, "loss": 1.8865, "step": 2078 }, { "epoch": 0.63, "learning_rate": 0.00018373021021772385, "loss": 1.9537, "step": 2079 }, { "epoch": 0.63, "learning_rate": 0.00018371243776114615, "loss": 1.8903, "step": 2080 }, { "epoch": 0.63, "learning_rate": 0.0001836946564634686, "loss": 1.807, "step": 2081 }, { "epoch": 0.63, "learning_rate": 0.0001836768663265692, "loss": 1.881, "step": 2082 }, { "epoch": 0.63, "learning_rate": 0.00018365906735232676, "loss": 1.8394, "step": 2083 }, { "epoch": 0.63, "learning_rate": 0.00018364125954262113, "loss": 1.8053, "step": 2084 }, { "epoch": 0.63, "learning_rate": 0.00018362344289933299, "loss": 1.8651, "step": 2085 }, { "epoch": 0.63, "learning_rate": 0.0001836056174243441, "loss": 1.8526, "step": 2086 }, { "epoch": 0.63, "learning_rate": 0.0001835877831195369, "loss": 1.8168, "step": 2087 }, { "epoch": 0.63, "learning_rate": 0.00018356993998679507, "loss": 1.8547, "step": 2088 }, { "epoch": 0.63, "learning_rate": 0.00018355208802800296, "loss": 1.8938, "step": 2089 }, { "epoch": 0.63, "learning_rate": 0.00018353422724504606, "loss": 1.7305, "step": 2090 }, { "epoch": 0.63, "learning_rate": 0.00018351635763981064, "loss": 1.9034, "step": 2091 }, { "epoch": 0.63, "learning_rate": 0.00018349847921418397, "loss": 1.7632, "step": 2092 }, { "epoch": 0.63, "learning_rate": 0.00018348059197005422, "loss": 1.8071, "step": 2093 }, { "epoch": 0.63, "learning_rate": 0.00018346269590931055, "loss": 1.8458, "step": 2094 }, { "epoch": 0.63, "learning_rate": 0.000183444791033843, "loss": 1.8696, "step": 2095 }, { "epoch": 0.63, "learning_rate": 0.00018342687734554253, "loss": 1.7897, "step": 2096 }, { "epoch": 0.63, "learning_rate": 0.00018340895484630107, "loss": 1.9737, "step": 2097 }, { "epoch": 0.63, "learning_rate": 0.00018339102353801144, "loss": 1.8755, "step": 2098 }, { "epoch": 0.63, "learning_rate": 0.00018337308342256749, "loss": 1.8585, "step": 2099 }, { "epoch": 0.63, "learning_rate": 0.00018335513450186387, "loss": 1.7306, "step": 2100 }, { "epoch": 0.63, "learning_rate": 0.00018333717677779615, "loss": 1.7714, "step": 2101 }, { "epoch": 0.63, "learning_rate": 0.000183319210252261, "loss": 1.9358, "step": 2102 }, { "epoch": 0.63, "learning_rate": 0.0001833012349271559, "loss": 1.828, "step": 2103 }, { "epoch": 0.63, "learning_rate": 0.00018328325080437923, "loss": 1.8292, "step": 2104 }, { "epoch": 0.63, "learning_rate": 0.00018326525788583037, "loss": 1.9092, "step": 2105 }, { "epoch": 0.63, "learning_rate": 0.00018324725617340954, "loss": 1.8768, "step": 2106 }, { "epoch": 0.63, "learning_rate": 0.00018322924566901806, "loss": 1.8627, "step": 2107 }, { "epoch": 0.63, "learning_rate": 0.000183211226374558, "loss": 1.8778, "step": 2108 }, { "epoch": 0.63, "learning_rate": 0.0001831931982919324, "loss": 1.8649, "step": 2109 }, { "epoch": 0.64, "learning_rate": 0.00018317516142304533, "loss": 1.868, "step": 2110 }, { "epoch": 0.64, "learning_rate": 0.00018315711576980169, "loss": 1.8157, "step": 2111 }, { "epoch": 0.64, "learning_rate": 0.0001831390613341073, "loss": 1.8044, "step": 2112 }, { "epoch": 0.64, "learning_rate": 0.00018312099811786895, "loss": 1.8605, "step": 2113 }, { "epoch": 0.64, "learning_rate": 0.00018310292612299436, "loss": 1.9083, "step": 2114 }, { "epoch": 0.64, "learning_rate": 0.00018308484535139216, "loss": 1.9008, "step": 2115 }, { "epoch": 0.64, "learning_rate": 0.00018306675580497187, "loss": 1.8782, "step": 2116 }, { "epoch": 0.64, "learning_rate": 0.00018304865748564403, "loss": 1.8139, "step": 2117 }, { "epoch": 0.64, "learning_rate": 0.00018303055039532007, "loss": 1.857, "step": 2118 }, { "epoch": 0.64, "learning_rate": 0.00018301243453591225, "loss": 1.9246, "step": 2119 }, { "epoch": 0.64, "learning_rate": 0.00018299430990933392, "loss": 1.819, "step": 2120 }, { "epoch": 0.64, "learning_rate": 0.0001829761765174992, "loss": 1.8423, "step": 2121 }, { "epoch": 0.64, "learning_rate": 0.00018295803436232323, "loss": 1.8569, "step": 2122 }, { "epoch": 0.64, "learning_rate": 0.00018293988344572206, "loss": 1.96, "step": 2123 }, { "epoch": 0.64, "learning_rate": 0.00018292172376961266, "loss": 1.7264, "step": 2124 }, { "epoch": 0.64, "learning_rate": 0.00018290355533591294, "loss": 1.8967, "step": 2125 }, { "epoch": 0.64, "learning_rate": 0.00018288537814654168, "loss": 1.9662, "step": 2126 }, { "epoch": 0.64, "learning_rate": 0.00018286719220341862, "loss": 1.7397, "step": 2127 }, { "epoch": 0.64, "learning_rate": 0.0001828489975084645, "loss": 1.8282, "step": 2128 }, { "epoch": 0.64, "learning_rate": 0.00018283079406360085, "loss": 1.9815, "step": 2129 }, { "epoch": 0.64, "learning_rate": 0.00018281258187075017, "loss": 1.7766, "step": 2130 }, { "epoch": 0.64, "learning_rate": 0.00018279436093183594, "loss": 1.8881, "step": 2131 }, { "epoch": 0.64, "learning_rate": 0.0001827761312487825, "loss": 1.9597, "step": 2132 }, { "epoch": 0.64, "learning_rate": 0.00018275789282351518, "loss": 1.8053, "step": 2133 }, { "epoch": 0.64, "learning_rate": 0.00018273964565796013, "loss": 1.9301, "step": 2134 }, { "epoch": 0.64, "learning_rate": 0.00018272138975404452, "loss": 1.8286, "step": 2135 }, { "epoch": 0.64, "learning_rate": 0.00018270312511369636, "loss": 1.8489, "step": 2136 }, { "epoch": 0.64, "learning_rate": 0.00018268485173884472, "loss": 1.7604, "step": 2137 }, { "epoch": 0.64, "learning_rate": 0.00018266656963141943, "loss": 1.7752, "step": 2138 }, { "epoch": 0.64, "learning_rate": 0.0001826482787933513, "loss": 1.7215, "step": 2139 }, { "epoch": 0.64, "learning_rate": 0.00018262997922657213, "loss": 1.7635, "step": 2140 }, { "epoch": 0.64, "learning_rate": 0.00018261167093301453, "loss": 1.8196, "step": 2141 }, { "epoch": 0.64, "learning_rate": 0.00018259335391461215, "loss": 1.843, "step": 2142 }, { "epoch": 0.65, "learning_rate": 0.00018257502817329945, "loss": 1.8769, "step": 2143 }, { "epoch": 0.65, "learning_rate": 0.00018255669371101193, "loss": 1.8926, "step": 2144 }, { "epoch": 0.65, "learning_rate": 0.0001825383505296858, "loss": 1.8752, "step": 2145 }, { "epoch": 0.65, "learning_rate": 0.00018251999863125852, "loss": 1.846, "step": 2146 }, { "epoch": 0.65, "learning_rate": 0.00018250163801766816, "loss": 1.856, "step": 2147 }, { "epoch": 0.65, "learning_rate": 0.00018248326869085388, "loss": 1.8787, "step": 2148 }, { "epoch": 0.65, "learning_rate": 0.0001824648906527557, "loss": 1.9305, "step": 2149 }, { "epoch": 0.65, "learning_rate": 0.00018244650390531454, "loss": 1.7832, "step": 2150 }, { "epoch": 0.65, "learning_rate": 0.00018242810845047237, "loss": 1.9413, "step": 2151 }, { "epoch": 0.65, "learning_rate": 0.00018240970429017189, "loss": 1.98, "step": 2152 }, { "epoch": 0.65, "learning_rate": 0.00018239129142635685, "loss": 1.9155, "step": 2153 }, { "epoch": 0.65, "learning_rate": 0.00018237286986097192, "loss": 1.8029, "step": 2154 }, { "epoch": 0.65, "learning_rate": 0.00018235443959596257, "loss": 2.1308, "step": 2155 }, { "epoch": 0.65, "learning_rate": 0.00018233600063327534, "loss": 2.2258, "step": 2156 }, { "epoch": 0.65, "learning_rate": 0.00018231755297485756, "loss": 1.9994, "step": 2157 }, { "epoch": 0.65, "learning_rate": 0.00018229909662265763, "loss": 2.0668, "step": 2158 }, { "epoch": 0.65, "learning_rate": 0.00018228063157862468, "loss": 1.9899, "step": 2159 }, { "epoch": 0.65, "learning_rate": 0.0001822621578447089, "loss": 1.9106, "step": 2160 }, { "epoch": 0.65, "learning_rate": 0.00018224367542286137, "loss": 1.9305, "step": 2161 }, { "epoch": 0.65, "learning_rate": 0.00018222518431503404, "loss": 1.8782, "step": 2162 }, { "epoch": 0.65, "learning_rate": 0.00018220668452317978, "loss": 1.8545, "step": 2163 }, { "epoch": 0.65, "learning_rate": 0.00018218817604925244, "loss": 1.9681, "step": 2164 }, { "epoch": 0.65, "learning_rate": 0.0001821696588952068, "loss": 1.9216, "step": 2165 }, { "epoch": 0.65, "learning_rate": 0.0001821511330629984, "loss": 1.8377, "step": 2166 }, { "epoch": 0.65, "learning_rate": 0.00018213259855458385, "loss": 1.8644, "step": 2167 }, { "epoch": 0.65, "learning_rate": 0.00018211405537192063, "loss": 1.8496, "step": 2168 }, { "epoch": 0.65, "learning_rate": 0.00018209550351696717, "loss": 1.8256, "step": 2169 }, { "epoch": 0.65, "learning_rate": 0.00018207694299168277, "loss": 1.8538, "step": 2170 }, { "epoch": 0.65, "learning_rate": 0.0001820583737980276, "loss": 1.8594, "step": 2171 }, { "epoch": 0.65, "learning_rate": 0.0001820397959379629, "loss": 1.9313, "step": 2172 }, { "epoch": 0.65, "learning_rate": 0.00018202120941345062, "loss": 1.8682, "step": 2173 }, { "epoch": 0.65, "learning_rate": 0.00018200261422645378, "loss": 1.8714, "step": 2174 }, { "epoch": 0.65, "learning_rate": 0.00018198401037893634, "loss": 1.883, "step": 2175 }, { "epoch": 0.66, "learning_rate": 0.000181965397872863, "loss": 1.9158, "step": 2176 }, { "epoch": 0.66, "learning_rate": 0.00018194677671019952, "loss": 1.9635, "step": 2177 }, { "epoch": 0.66, "learning_rate": 0.00018192814689291256, "loss": 1.7308, "step": 2178 }, { "epoch": 0.66, "learning_rate": 0.00018190950842296961, "loss": 1.7553, "step": 2179 }, { "epoch": 0.66, "learning_rate": 0.00018189086130233914, "loss": 1.8884, "step": 2180 }, { "epoch": 0.66, "learning_rate": 0.00018187220553299055, "loss": 1.7218, "step": 2181 }, { "epoch": 0.66, "learning_rate": 0.00018185354111689415, "loss": 1.7994, "step": 2182 }, { "epoch": 0.66, "learning_rate": 0.00018183486805602107, "loss": 1.8613, "step": 2183 }, { "epoch": 0.66, "learning_rate": 0.00018181618635234345, "loss": 1.7876, "step": 2184 }, { "epoch": 0.66, "learning_rate": 0.00018179749600783437, "loss": 1.7707, "step": 2185 }, { "epoch": 0.66, "learning_rate": 0.0001817787970244677, "loss": 1.8462, "step": 2186 }, { "epoch": 0.66, "learning_rate": 0.00018176008940421832, "loss": 1.8746, "step": 2187 }, { "epoch": 0.66, "learning_rate": 0.00018174137314906196, "loss": 1.8154, "step": 2188 }, { "epoch": 0.66, "learning_rate": 0.00018172264826097536, "loss": 1.8513, "step": 2189 }, { "epoch": 0.66, "learning_rate": 0.00018170391474193605, "loss": 1.8442, "step": 2190 }, { "epoch": 0.66, "learning_rate": 0.00018168517259392256, "loss": 1.8479, "step": 2191 }, { "epoch": 0.66, "learning_rate": 0.00018166642181891428, "loss": 1.7348, "step": 2192 }, { "epoch": 0.66, "learning_rate": 0.0001816476624188915, "loss": 1.8232, "step": 2193 }, { "epoch": 0.66, "learning_rate": 0.00018162889439583555, "loss": 1.8884, "step": 2194 }, { "epoch": 0.66, "learning_rate": 0.0001816101177517285, "loss": 1.8758, "step": 2195 }, { "epoch": 0.66, "learning_rate": 0.0001815913324885534, "loss": 1.8932, "step": 2196 }, { "epoch": 0.66, "learning_rate": 0.0001815725386082942, "loss": 1.8706, "step": 2197 }, { "epoch": 0.66, "learning_rate": 0.00018155373611293584, "loss": 2.0179, "step": 2198 }, { "epoch": 0.66, "learning_rate": 0.00018153492500446407, "loss": 1.8459, "step": 2199 }, { "epoch": 0.66, "learning_rate": 0.00018151610528486554, "loss": 1.7155, "step": 2200 }, { "epoch": 0.66, "learning_rate": 0.00018149727695612793, "loss": 1.7808, "step": 2201 }, { "epoch": 0.66, "learning_rate": 0.0001814784400202397, "loss": 1.7919, "step": 2202 }, { "epoch": 0.66, "learning_rate": 0.00018145959447919027, "loss": 1.779, "step": 2203 }, { "epoch": 0.66, "learning_rate": 0.00018144074033497, "loss": 1.8123, "step": 2204 }, { "epoch": 0.66, "learning_rate": 0.00018142187758957008, "loss": 1.7689, "step": 2205 }, { "epoch": 0.66, "learning_rate": 0.0001814030062449827, "loss": 1.8729, "step": 2206 }, { "epoch": 0.66, "learning_rate": 0.00018138412630320087, "loss": 1.7748, "step": 2207 }, { "epoch": 0.66, "learning_rate": 0.00018136523776621864, "loss": 1.9317, "step": 2208 }, { "epoch": 0.66, "learning_rate": 0.00018134634063603078, "loss": 1.8983, "step": 2209 }, { "epoch": 0.67, "learning_rate": 0.0001813274349146331, "loss": 1.8312, "step": 2210 }, { "epoch": 0.67, "learning_rate": 0.00018130852060402227, "loss": 1.7844, "step": 2211 }, { "epoch": 0.67, "learning_rate": 0.00018128959770619596, "loss": 1.7984, "step": 2212 }, { "epoch": 0.67, "learning_rate": 0.0001812706662231526, "loss": 1.7875, "step": 2213 }, { "epoch": 0.67, "learning_rate": 0.00018125172615689162, "loss": 1.9022, "step": 2214 }, { "epoch": 0.67, "learning_rate": 0.0001812327775094133, "loss": 1.8375, "step": 2215 }, { "epoch": 0.67, "learning_rate": 0.0001812138202827189, "loss": 1.7904, "step": 2216 }, { "epoch": 0.67, "learning_rate": 0.00018119485447881052, "loss": 1.8688, "step": 2217 }, { "epoch": 0.67, "learning_rate": 0.00018117588009969118, "loss": 1.7316, "step": 2218 }, { "epoch": 0.67, "learning_rate": 0.00018115689714736483, "loss": 1.7923, "step": 2219 }, { "epoch": 0.67, "learning_rate": 0.00018113790562383634, "loss": 1.8509, "step": 2220 }, { "epoch": 0.67, "learning_rate": 0.0001811189055311114, "loss": 1.9835, "step": 2221 }, { "epoch": 0.67, "learning_rate": 0.00018109989687119672, "loss": 1.8247, "step": 2222 }, { "epoch": 0.67, "learning_rate": 0.0001810808796460998, "loss": 1.8646, "step": 2223 }, { "epoch": 0.67, "learning_rate": 0.00018106185385782916, "loss": 1.8638, "step": 2224 }, { "epoch": 0.67, "learning_rate": 0.00018104281950839413, "loss": 1.9433, "step": 2225 }, { "epoch": 0.67, "learning_rate": 0.000181023776599805, "loss": 1.9251, "step": 2226 }, { "epoch": 0.67, "learning_rate": 0.0001810047251340729, "loss": 1.763, "step": 2227 }, { "epoch": 0.67, "learning_rate": 0.00018098566511320994, "loss": 1.8794, "step": 2228 }, { "epoch": 0.67, "learning_rate": 0.00018096659653922908, "loss": 1.9181, "step": 2229 }, { "epoch": 0.67, "learning_rate": 0.0001809475194141443, "loss": 1.9085, "step": 2230 }, { "epoch": 0.67, "learning_rate": 0.00018092843373997025, "loss": 1.8806, "step": 2231 }, { "epoch": 0.67, "learning_rate": 0.0001809093395187227, "loss": 1.9526, "step": 2232 }, { "epoch": 0.67, "learning_rate": 0.0001808902367524182, "loss": 1.8872, "step": 2233 }, { "epoch": 0.67, "learning_rate": 0.00018087112544307426, "loss": 1.8103, "step": 2234 }, { "epoch": 0.67, "learning_rate": 0.00018085200559270933, "loss": 1.8503, "step": 2235 }, { "epoch": 0.67, "learning_rate": 0.00018083287720334264, "loss": 1.7889, "step": 2236 }, { "epoch": 0.67, "learning_rate": 0.0001808137402769944, "loss": 1.74, "step": 2237 }, { "epoch": 0.67, "learning_rate": 0.00018079459481568576, "loss": 1.7635, "step": 2238 }, { "epoch": 0.67, "learning_rate": 0.0001807754408214387, "loss": 1.85, "step": 2239 }, { "epoch": 0.67, "learning_rate": 0.0001807562782962761, "loss": 1.9777, "step": 2240 }, { "epoch": 0.67, "learning_rate": 0.0001807371072422218, "loss": 1.7076, "step": 2241 }, { "epoch": 0.67, "learning_rate": 0.00018071792766130048, "loss": 1.8584, "step": 2242 }, { "epoch": 0.68, "learning_rate": 0.00018069873955553776, "loss": 1.8579, "step": 2243 }, { "epoch": 0.68, "learning_rate": 0.0001806795429269602, "loss": 1.8588, "step": 2244 }, { "epoch": 0.68, "learning_rate": 0.00018066033777759511, "loss": 1.8534, "step": 2245 }, { "epoch": 0.68, "learning_rate": 0.00018064112410947088, "loss": 1.8227, "step": 2246 }, { "epoch": 0.68, "learning_rate": 0.00018062190192461665, "loss": 1.8452, "step": 2247 }, { "epoch": 0.68, "learning_rate": 0.0001806026712250626, "loss": 1.8454, "step": 2248 }, { "epoch": 0.68, "learning_rate": 0.00018058343201283967, "loss": 1.8187, "step": 2249 }, { "epoch": 0.68, "learning_rate": 0.00018056418428997984, "loss": 1.82, "step": 2250 }, { "epoch": 0.68, "learning_rate": 0.00018054492805851584, "loss": 1.9353, "step": 2251 }, { "epoch": 0.68, "learning_rate": 0.00018052566332048142, "loss": 1.8939, "step": 2252 }, { "epoch": 0.68, "learning_rate": 0.00018050639007791114, "loss": 1.8267, "step": 2253 }, { "epoch": 0.68, "learning_rate": 0.00018048710833284056, "loss": 1.8834, "step": 2254 }, { "epoch": 0.68, "learning_rate": 0.00018046781808730603, "loss": 1.8976, "step": 2255 }, { "epoch": 0.68, "learning_rate": 0.00018044851934334485, "loss": 1.7309, "step": 2256 }, { "epoch": 0.68, "learning_rate": 0.00018042921210299526, "loss": 1.9542, "step": 2257 }, { "epoch": 0.68, "learning_rate": 0.0001804098963682963, "loss": 1.9111, "step": 2258 }, { "epoch": 0.68, "learning_rate": 0.00018039057214128796, "loss": 1.8544, "step": 2259 }, { "epoch": 0.68, "learning_rate": 0.00018037123942401113, "loss": 1.7025, "step": 2260 }, { "epoch": 0.68, "learning_rate": 0.0001803518982185076, "loss": 1.8763, "step": 2261 }, { "epoch": 0.68, "learning_rate": 0.00018033254852682006, "loss": 1.7757, "step": 2262 }, { "epoch": 0.68, "learning_rate": 0.00018031319035099208, "loss": 1.7838, "step": 2263 }, { "epoch": 0.68, "learning_rate": 0.0001802938236930681, "loss": 1.8205, "step": 2264 }, { "epoch": 0.68, "learning_rate": 0.0001802744485550935, "loss": 1.7989, "step": 2265 }, { "epoch": 0.68, "learning_rate": 0.00018025506493911454, "loss": 1.779, "step": 2266 }, { "epoch": 0.68, "learning_rate": 0.00018023567284717842, "loss": 1.8239, "step": 2267 }, { "epoch": 0.68, "learning_rate": 0.0001802162722813331, "loss": 1.8679, "step": 2268 }, { "epoch": 0.68, "learning_rate": 0.00018019686324362763, "loss": 1.8957, "step": 2269 }, { "epoch": 0.68, "learning_rate": 0.0001801774457361118, "loss": 1.7743, "step": 2270 }, { "epoch": 0.68, "learning_rate": 0.00018015801976083634, "loss": 1.8886, "step": 2271 }, { "epoch": 0.68, "learning_rate": 0.0001801385853198529, "loss": 1.8003, "step": 2272 }, { "epoch": 0.68, "learning_rate": 0.00018011914241521397, "loss": 1.9223, "step": 2273 }, { "epoch": 0.68, "learning_rate": 0.000180099691048973, "loss": 1.7895, "step": 2274 }, { "epoch": 0.68, "learning_rate": 0.00018008023122318431, "loss": 1.863, "step": 2275 }, { "epoch": 0.69, "learning_rate": 0.00018006076293990305, "loss": 1.84, "step": 2276 }, { "epoch": 0.69, "learning_rate": 0.0001800412862011854, "loss": 1.9011, "step": 2277 }, { "epoch": 0.69, "learning_rate": 0.0001800218010090883, "loss": 1.8658, "step": 2278 }, { "epoch": 0.69, "learning_rate": 0.0001800023073656697, "loss": 1.8737, "step": 2279 }, { "epoch": 0.69, "learning_rate": 0.00017998280527298826, "loss": 1.8792, "step": 2280 }, { "epoch": 0.69, "learning_rate": 0.00017996329473310372, "loss": 1.7906, "step": 2281 }, { "epoch": 0.69, "learning_rate": 0.00017994377574807662, "loss": 1.8435, "step": 2282 }, { "epoch": 0.69, "learning_rate": 0.00017992424831996844, "loss": 1.7368, "step": 2283 }, { "epoch": 0.69, "learning_rate": 0.00017990471245084153, "loss": 1.7793, "step": 2284 }, { "epoch": 0.69, "learning_rate": 0.0001798851681427591, "loss": 1.791, "step": 2285 }, { "epoch": 0.69, "learning_rate": 0.00017986561539778528, "loss": 1.7508, "step": 2286 }, { "epoch": 0.69, "learning_rate": 0.0001798460542179851, "loss": 1.8073, "step": 2287 }, { "epoch": 0.69, "learning_rate": 0.00017982648460542446, "loss": 1.8013, "step": 2288 }, { "epoch": 0.69, "learning_rate": 0.0001798069065621702, "loss": 1.7836, "step": 2289 }, { "epoch": 0.69, "learning_rate": 0.00017978732009028993, "loss": 1.7685, "step": 2290 }, { "epoch": 0.69, "learning_rate": 0.00017976772519185226, "loss": 1.8375, "step": 2291 }, { "epoch": 0.69, "learning_rate": 0.00017974812186892676, "loss": 1.8766, "step": 2292 }, { "epoch": 0.69, "learning_rate": 0.00017972851012358366, "loss": 1.7826, "step": 2293 }, { "epoch": 0.69, "learning_rate": 0.00017970888995789427, "loss": 1.7731, "step": 2294 }, { "epoch": 0.69, "learning_rate": 0.00017968926137393068, "loss": 1.859, "step": 2295 }, { "epoch": 0.69, "learning_rate": 0.000179669624373766, "loss": 1.8433, "step": 2296 }, { "epoch": 0.69, "learning_rate": 0.0001796499789594741, "loss": 1.9264, "step": 2297 }, { "epoch": 0.69, "learning_rate": 0.00017963032513312977, "loss": 1.8896, "step": 2298 }, { "epoch": 0.69, "learning_rate": 0.00017961066289680876, "loss": 1.8397, "step": 2299 }, { "epoch": 0.69, "learning_rate": 0.00017959099225258758, "loss": 1.8883, "step": 2300 }, { "epoch": 0.69, "learning_rate": 0.00017957131320254374, "loss": 1.821, "step": 2301 }, { "epoch": 0.69, "learning_rate": 0.00017955162574875564, "loss": 1.8549, "step": 2302 }, { "epoch": 0.69, "learning_rate": 0.00017953192989330244, "loss": 1.8513, "step": 2303 }, { "epoch": 0.69, "learning_rate": 0.00017951222563826434, "loss": 1.9299, "step": 2304 }, { "epoch": 0.69, "learning_rate": 0.00017949251298572235, "loss": 1.7494, "step": 2305 }, { "epoch": 0.69, "learning_rate": 0.0001794727919377583, "loss": 1.9002, "step": 2306 }, { "epoch": 0.69, "learning_rate": 0.00017945306249645508, "loss": 1.9068, "step": 2307 }, { "epoch": 0.69, "learning_rate": 0.00017943332466389635, "loss": 1.8266, "step": 2308 }, { "epoch": 0.7, "learning_rate": 0.00017941357844216665, "loss": 1.9632, "step": 2309 }, { "epoch": 0.7, "learning_rate": 0.00017939382383335147, "loss": 1.849, "step": 2310 }, { "epoch": 0.7, "learning_rate": 0.00017937406083953712, "loss": 1.8343, "step": 2311 }, { "epoch": 0.7, "learning_rate": 0.00017935428946281082, "loss": 1.9675, "step": 2312 }, { "epoch": 0.7, "learning_rate": 0.00017933450970526068, "loss": 1.7525, "step": 2313 }, { "epoch": 0.7, "learning_rate": 0.0001793147215689757, "loss": 1.8357, "step": 2314 }, { "epoch": 0.7, "learning_rate": 0.0001792949250560458, "loss": 1.7861, "step": 2315 }, { "epoch": 0.7, "learning_rate": 0.00017927512016856166, "loss": 1.7863, "step": 2316 }, { "epoch": 0.7, "learning_rate": 0.000179255306908615, "loss": 1.8984, "step": 2317 }, { "epoch": 0.7, "learning_rate": 0.00017923548527829831, "loss": 1.904, "step": 2318 }, { "epoch": 0.7, "learning_rate": 0.00017921565527970502, "loss": 1.9791, "step": 2319 }, { "epoch": 0.7, "learning_rate": 0.00017919581691492947, "loss": 1.7743, "step": 2320 }, { "epoch": 0.7, "learning_rate": 0.00017917597018606677, "loss": 1.8997, "step": 2321 }, { "epoch": 0.7, "learning_rate": 0.00017915611509521307, "loss": 1.8023, "step": 2322 }, { "epoch": 0.7, "learning_rate": 0.00017913625164446524, "loss": 1.8445, "step": 2323 }, { "epoch": 0.7, "learning_rate": 0.00017911637983592112, "loss": 1.829, "step": 2324 }, { "epoch": 0.7, "learning_rate": 0.00017909649967167953, "loss": 1.8078, "step": 2325 }, { "epoch": 0.7, "learning_rate": 0.00017907661115383993, "loss": 1.9162, "step": 2326 }, { "epoch": 0.7, "learning_rate": 0.0001790567142845029, "loss": 1.8094, "step": 2327 }, { "epoch": 0.7, "learning_rate": 0.00017903680906576974, "loss": 1.7906, "step": 2328 }, { "epoch": 0.7, "learning_rate": 0.00017901689549974274, "loss": 1.8344, "step": 2329 }, { "epoch": 0.7, "learning_rate": 0.00017899697358852502, "loss": 1.8092, "step": 2330 }, { "epoch": 0.7, "learning_rate": 0.00017897704333422057, "loss": 1.937, "step": 2331 }, { "epoch": 0.7, "learning_rate": 0.00017895710473893428, "loss": 2.0637, "step": 2332 }, { "epoch": 0.7, "learning_rate": 0.0001789371578047719, "loss": 1.818, "step": 2333 }, { "epoch": 0.7, "learning_rate": 0.00017891720253384016, "loss": 1.8097, "step": 2334 }, { "epoch": 0.7, "learning_rate": 0.00017889723892824648, "loss": 1.8143, "step": 2335 }, { "epoch": 0.7, "learning_rate": 0.00017887726699009936, "loss": 1.821, "step": 2336 }, { "epoch": 0.7, "learning_rate": 0.00017885728672150807, "loss": 1.8712, "step": 2337 }, { "epoch": 0.7, "learning_rate": 0.00017883729812458271, "loss": 1.8095, "step": 2338 }, { "epoch": 0.7, "learning_rate": 0.00017881730120143446, "loss": 1.7422, "step": 2339 }, { "epoch": 0.7, "learning_rate": 0.00017879729595417514, "loss": 1.9105, "step": 2340 }, { "epoch": 0.7, "learning_rate": 0.00017877728238491759, "loss": 1.876, "step": 2341 }, { "epoch": 0.7, "learning_rate": 0.00017875726049577554, "loss": 1.8433, "step": 2342 }, { "epoch": 0.71, "learning_rate": 0.00017873723028886352, "loss": 1.9009, "step": 2343 }, { "epoch": 0.71, "learning_rate": 0.00017871719176629694, "loss": 1.8339, "step": 2344 }, { "epoch": 0.71, "learning_rate": 0.00017869714493019218, "loss": 1.8839, "step": 2345 }, { "epoch": 0.71, "learning_rate": 0.0001786770897826664, "loss": 1.7361, "step": 2346 }, { "epoch": 0.71, "learning_rate": 0.00017865702632583773, "loss": 1.9231, "step": 2347 }, { "epoch": 0.71, "learning_rate": 0.0001786369545618251, "loss": 1.8474, "step": 2348 }, { "epoch": 0.71, "learning_rate": 0.00017861687449274833, "loss": 2.0268, "step": 2349 }, { "epoch": 0.71, "learning_rate": 0.00017859678612072813, "loss": 1.8093, "step": 2350 }, { "epoch": 0.71, "learning_rate": 0.00017857668944788616, "loss": 1.7916, "step": 2351 }, { "epoch": 0.71, "learning_rate": 0.00017855658447634475, "loss": 1.808, "step": 2352 }, { "epoch": 0.71, "learning_rate": 0.00017853647120822736, "loss": 1.8394, "step": 2353 }, { "epoch": 0.71, "learning_rate": 0.00017851634964565817, "loss": 1.801, "step": 2354 }, { "epoch": 0.71, "learning_rate": 0.00017849621979076224, "loss": 1.9296, "step": 2355 }, { "epoch": 0.71, "learning_rate": 0.00017847608164566557, "loss": 1.7907, "step": 2356 }, { "epoch": 0.71, "learning_rate": 0.000178455935212495, "loss": 1.7088, "step": 2357 }, { "epoch": 0.71, "learning_rate": 0.00017843578049337827, "loss": 1.8621, "step": 2358 }, { "epoch": 0.71, "learning_rate": 0.00017841561749044394, "loss": 1.926, "step": 2359 }, { "epoch": 0.71, "learning_rate": 0.00017839544620582148, "loss": 1.8846, "step": 2360 }, { "epoch": 0.71, "learning_rate": 0.0001783752666416413, "loss": 1.7431, "step": 2361 }, { "epoch": 0.71, "learning_rate": 0.0001783550788000345, "loss": 1.8308, "step": 2362 }, { "epoch": 0.71, "learning_rate": 0.0001783348826831333, "loss": 1.8532, "step": 2363 }, { "epoch": 0.71, "learning_rate": 0.0001783146782930706, "loss": 1.789, "step": 2364 }, { "epoch": 0.71, "learning_rate": 0.00017829446563198022, "loss": 1.8015, "step": 2365 }, { "epoch": 0.71, "learning_rate": 0.00017827424470199694, "loss": 1.7878, "step": 2366 }, { "epoch": 0.71, "learning_rate": 0.0001782540155052563, "loss": 1.8414, "step": 2367 }, { "epoch": 0.71, "learning_rate": 0.00017823377804389477, "loss": 1.9309, "step": 2368 }, { "epoch": 0.71, "learning_rate": 0.0001782135323200497, "loss": 1.8904, "step": 2369 }, { "epoch": 0.71, "learning_rate": 0.0001781932783358593, "loss": 1.9239, "step": 2370 }, { "epoch": 0.71, "learning_rate": 0.00017817301609346258, "loss": 1.8542, "step": 2371 }, { "epoch": 0.71, "learning_rate": 0.00017815274559499958, "loss": 1.8392, "step": 2372 }, { "epoch": 0.71, "learning_rate": 0.00017813246684261107, "loss": 1.7804, "step": 2373 }, { "epoch": 0.71, "learning_rate": 0.00017811217983843878, "loss": 1.7691, "step": 2374 }, { "epoch": 0.71, "learning_rate": 0.00017809188458462523, "loss": 1.9356, "step": 2375 }, { "epoch": 0.72, "learning_rate": 0.00017807158108331392, "loss": 1.8288, "step": 2376 }, { "epoch": 0.72, "learning_rate": 0.00017805126933664913, "loss": 1.9003, "step": 2377 }, { "epoch": 0.72, "learning_rate": 0.000178030949346776, "loss": 1.7331, "step": 2378 }, { "epoch": 0.72, "learning_rate": 0.0001780106211158406, "loss": 1.8868, "step": 2379 }, { "epoch": 0.72, "learning_rate": 0.0001779902846459899, "loss": 1.7932, "step": 2380 }, { "epoch": 0.72, "learning_rate": 0.00017796993993937163, "loss": 1.8407, "step": 2381 }, { "epoch": 0.72, "learning_rate": 0.00017794958699813448, "loss": 1.8259, "step": 2382 }, { "epoch": 0.72, "learning_rate": 0.00017792922582442796, "loss": 1.8037, "step": 2383 }, { "epoch": 0.72, "learning_rate": 0.0001779088564204025, "loss": 1.7807, "step": 2384 }, { "epoch": 0.72, "learning_rate": 0.00017788847878820934, "loss": 1.8346, "step": 2385 }, { "epoch": 0.72, "learning_rate": 0.00017786809293000063, "loss": 1.773, "step": 2386 }, { "epoch": 0.72, "learning_rate": 0.0001778476988479294, "loss": 1.8015, "step": 2387 }, { "epoch": 0.72, "learning_rate": 0.00017782729654414946, "loss": 1.8833, "step": 2388 }, { "epoch": 0.72, "learning_rate": 0.0001778068860208156, "loss": 1.7368, "step": 2389 }, { "epoch": 0.72, "learning_rate": 0.00017778646728008342, "loss": 1.9055, "step": 2390 }, { "epoch": 0.72, "learning_rate": 0.00017776604032410944, "loss": 1.8802, "step": 2391 }, { "epoch": 0.72, "learning_rate": 0.00017774560515505092, "loss": 1.8839, "step": 2392 }, { "epoch": 0.72, "learning_rate": 0.00017772516177506616, "loss": 1.9086, "step": 2393 }, { "epoch": 0.72, "learning_rate": 0.00017770471018631418, "loss": 1.867, "step": 2394 }, { "epoch": 0.72, "learning_rate": 0.00017768425039095496, "loss": 1.7831, "step": 2395 }, { "epoch": 0.72, "learning_rate": 0.00017766378239114931, "loss": 1.9167, "step": 2396 }, { "epoch": 0.72, "learning_rate": 0.00017764330618905892, "loss": 1.8335, "step": 2397 }, { "epoch": 0.72, "learning_rate": 0.0001776228217868463, "loss": 1.8137, "step": 2398 }, { "epoch": 0.72, "learning_rate": 0.00017760232918667492, "loss": 1.8138, "step": 2399 }, { "epoch": 0.72, "learning_rate": 0.000177581828390709, "loss": 1.7886, "step": 2400 }, { "epoch": 0.72, "learning_rate": 0.00017756131940111376, "loss": 1.8058, "step": 2401 }, { "epoch": 0.72, "learning_rate": 0.00017754080222005513, "loss": 1.8755, "step": 2402 }, { "epoch": 0.72, "learning_rate": 0.00017752027684970006, "loss": 1.8121, "step": 2403 }, { "epoch": 0.72, "learning_rate": 0.0001774997432922162, "loss": 1.8144, "step": 2404 }, { "epoch": 0.72, "learning_rate": 0.00017747920154977223, "loss": 1.6592, "step": 2405 }, { "epoch": 0.72, "learning_rate": 0.0001774586516245376, "loss": 1.7668, "step": 2406 }, { "epoch": 0.72, "learning_rate": 0.00017743809351868266, "loss": 1.788, "step": 2407 }, { "epoch": 0.72, "learning_rate": 0.00017741752723437858, "loss": 1.8902, "step": 2408 }, { "epoch": 0.73, "learning_rate": 0.00017739695277379743, "loss": 1.8324, "step": 2409 }, { "epoch": 0.73, "learning_rate": 0.00017737637013911215, "loss": 1.8573, "step": 2410 }, { "epoch": 0.73, "learning_rate": 0.00017735577933249648, "loss": 1.8538, "step": 2411 }, { "epoch": 0.73, "learning_rate": 0.00017733518035612512, "loss": 1.8594, "step": 2412 }, { "epoch": 0.73, "learning_rate": 0.00017731457321217359, "loss": 1.7993, "step": 2413 }, { "epoch": 0.73, "learning_rate": 0.00017729395790281823, "loss": 1.8609, "step": 2414 }, { "epoch": 0.73, "learning_rate": 0.0001772733344302363, "loss": 1.8177, "step": 2415 }, { "epoch": 0.73, "learning_rate": 0.00017725270279660588, "loss": 1.7562, "step": 2416 }, { "epoch": 0.73, "learning_rate": 0.000177232063004106, "loss": 1.9496, "step": 2417 }, { "epoch": 0.73, "learning_rate": 0.0001772114150549164, "loss": 1.781, "step": 2418 }, { "epoch": 0.73, "learning_rate": 0.0001771907589512178, "loss": 1.7932, "step": 2419 }, { "epoch": 0.73, "learning_rate": 0.00017717009469519175, "loss": 1.8319, "step": 2420 }, { "epoch": 0.73, "learning_rate": 0.00017714942228902068, "loss": 1.9334, "step": 2421 }, { "epoch": 0.73, "learning_rate": 0.00017712874173488781, "loss": 1.8588, "step": 2422 }, { "epoch": 0.73, "learning_rate": 0.0001771080530349773, "loss": 1.7497, "step": 2423 }, { "epoch": 0.73, "learning_rate": 0.0001770873561914742, "loss": 1.859, "step": 2424 }, { "epoch": 0.73, "learning_rate": 0.00017706665120656423, "loss": 1.8179, "step": 2425 }, { "epoch": 0.73, "learning_rate": 0.00017704593808243418, "loss": 1.8986, "step": 2426 }, { "epoch": 0.73, "learning_rate": 0.00017702521682127164, "loss": 1.9175, "step": 2427 }, { "epoch": 0.73, "learning_rate": 0.00017700448742526497, "loss": 1.7327, "step": 2428 }, { "epoch": 0.73, "learning_rate": 0.00017698374989660352, "loss": 1.795, "step": 2429 }, { "epoch": 0.73, "learning_rate": 0.00017696300423747743, "loss": 1.754, "step": 2430 }, { "epoch": 0.73, "learning_rate": 0.00017694225045007767, "loss": 1.6744, "step": 2431 }, { "epoch": 0.73, "learning_rate": 0.0001769214885365961, "loss": 1.8348, "step": 2432 }, { "epoch": 0.73, "learning_rate": 0.00017690071849922552, "loss": 1.9561, "step": 2433 }, { "epoch": 0.73, "learning_rate": 0.00017687994034015943, "loss": 1.8889, "step": 2434 }, { "epoch": 0.73, "learning_rate": 0.00017685915406159228, "loss": 1.8175, "step": 2435 }, { "epoch": 0.73, "learning_rate": 0.0001768383596657194, "loss": 1.8716, "step": 2436 }, { "epoch": 0.73, "learning_rate": 0.0001768175571547369, "loss": 1.8542, "step": 2437 }, { "epoch": 0.73, "learning_rate": 0.00017679674653084187, "loss": 1.7255, "step": 2438 }, { "epoch": 0.73, "learning_rate": 0.00017677592779623207, "loss": 1.8126, "step": 2439 }, { "epoch": 0.73, "learning_rate": 0.0001767551009531063, "loss": 1.9464, "step": 2440 }, { "epoch": 0.73, "learning_rate": 0.0001767342660036641, "loss": 1.7593, "step": 2441 }, { "epoch": 0.74, "learning_rate": 0.0001767134229501059, "loss": 1.7859, "step": 2442 }, { "epoch": 0.74, "learning_rate": 0.00017669257179463307, "loss": 1.7669, "step": 2443 }, { "epoch": 0.74, "learning_rate": 0.00017667171253944765, "loss": 1.8431, "step": 2444 }, { "epoch": 0.74, "learning_rate": 0.0001766508451867527, "loss": 1.8619, "step": 2445 }, { "epoch": 0.74, "learning_rate": 0.0001766299697387521, "loss": 1.9136, "step": 2446 }, { "epoch": 0.74, "learning_rate": 0.00017660908619765051, "loss": 1.8766, "step": 2447 }, { "epoch": 0.74, "learning_rate": 0.00017658819456565353, "loss": 1.8659, "step": 2448 }, { "epoch": 0.74, "learning_rate": 0.0001765672948449676, "loss": 1.835, "step": 2449 }, { "epoch": 0.74, "learning_rate": 0.00017654638703779995, "loss": 1.892, "step": 2450 }, { "epoch": 0.74, "learning_rate": 0.00017652547114635874, "loss": 1.8607, "step": 2451 }, { "epoch": 0.74, "learning_rate": 0.00017650454717285294, "loss": 1.849, "step": 2452 }, { "epoch": 0.74, "learning_rate": 0.00017648361511949238, "loss": 1.7822, "step": 2453 }, { "epoch": 0.74, "learning_rate": 0.0001764626749884878, "loss": 1.7885, "step": 2454 }, { "epoch": 0.74, "learning_rate": 0.0001764417267820507, "loss": 1.9575, "step": 2455 }, { "epoch": 0.74, "learning_rate": 0.00017642077050239346, "loss": 1.8535, "step": 2456 }, { "epoch": 0.74, "learning_rate": 0.0001763998061517294, "loss": 1.9278, "step": 2457 }, { "epoch": 0.74, "learning_rate": 0.00017637883373227258, "loss": 1.7819, "step": 2458 }, { "epoch": 0.74, "learning_rate": 0.00017635785324623794, "loss": 1.8571, "step": 2459 }, { "epoch": 0.74, "learning_rate": 0.0001763368646958413, "loss": 1.7316, "step": 2460 }, { "epoch": 0.74, "learning_rate": 0.00017631586808329933, "loss": 1.8447, "step": 2461 }, { "epoch": 0.74, "learning_rate": 0.00017629486341082953, "loss": 1.7398, "step": 2462 }, { "epoch": 0.74, "learning_rate": 0.00017627385068065023, "loss": 1.872, "step": 2463 }, { "epoch": 0.74, "learning_rate": 0.00017625282989498073, "loss": 1.766, "step": 2464 }, { "epoch": 0.74, "learning_rate": 0.000176231801056041, "loss": 1.8212, "step": 2465 }, { "epoch": 0.74, "learning_rate": 0.00017621076416605202, "loss": 1.7917, "step": 2466 }, { "epoch": 0.74, "learning_rate": 0.0001761897192272355, "loss": 1.8938, "step": 2467 }, { "epoch": 0.74, "learning_rate": 0.0001761686662418141, "loss": 1.8336, "step": 2468 }, { "epoch": 0.74, "learning_rate": 0.00017614760521201124, "loss": 1.8224, "step": 2469 }, { "epoch": 0.74, "learning_rate": 0.00017612653614005125, "loss": 1.783, "step": 2470 }, { "epoch": 0.74, "learning_rate": 0.0001761054590281593, "loss": 1.7614, "step": 2471 }, { "epoch": 0.74, "learning_rate": 0.0001760843738785614, "loss": 1.9209, "step": 2472 }, { "epoch": 0.74, "learning_rate": 0.00017606328069348442, "loss": 1.8743, "step": 2473 }, { "epoch": 0.74, "learning_rate": 0.00017604217947515604, "loss": 1.7917, "step": 2474 }, { "epoch": 0.75, "learning_rate": 0.00017602107022580485, "loss": 1.804, "step": 2475 }, { "epoch": 0.75, "learning_rate": 0.00017599995294766024, "loss": 1.8181, "step": 2476 }, { "epoch": 0.75, "learning_rate": 0.00017597882764295246, "loss": 1.8626, "step": 2477 }, { "epoch": 0.75, "learning_rate": 0.0001759576943139126, "loss": 1.8423, "step": 2478 }, { "epoch": 0.75, "learning_rate": 0.00017593655296277265, "loss": 1.756, "step": 2479 }, { "epoch": 0.75, "learning_rate": 0.00017591540359176535, "loss": 1.9954, "step": 2480 }, { "epoch": 0.75, "learning_rate": 0.00017589424620312438, "loss": 1.8257, "step": 2481 }, { "epoch": 0.75, "learning_rate": 0.00017587308079908422, "loss": 1.9147, "step": 2482 }, { "epoch": 0.75, "learning_rate": 0.00017585190738188025, "loss": 1.8201, "step": 2483 }, { "epoch": 0.75, "learning_rate": 0.00017583072595374856, "loss": 1.7561, "step": 2484 }, { "epoch": 0.75, "learning_rate": 0.00017580953651692626, "loss": 1.8188, "step": 2485 }, { "epoch": 0.75, "learning_rate": 0.00017578833907365117, "loss": 1.9072, "step": 2486 }, { "epoch": 0.75, "learning_rate": 0.00017576713362616207, "loss": 1.8209, "step": 2487 }, { "epoch": 0.75, "learning_rate": 0.00017574592017669847, "loss": 1.7971, "step": 2488 }, { "epoch": 0.75, "learning_rate": 0.00017572469872750079, "loss": 1.9871, "step": 2489 }, { "epoch": 0.75, "learning_rate": 0.00017570346928081032, "loss": 1.8, "step": 2490 }, { "epoch": 0.75, "learning_rate": 0.00017568223183886912, "loss": 1.809, "step": 2491 }, { "epoch": 0.75, "learning_rate": 0.00017566098640392017, "loss": 1.855, "step": 2492 }, { "epoch": 0.75, "learning_rate": 0.00017563973297820719, "loss": 1.7731, "step": 2493 }, { "epoch": 0.75, "learning_rate": 0.0001756184715639749, "loss": 1.91, "step": 2494 }, { "epoch": 0.75, "learning_rate": 0.00017559720216346874, "loss": 1.846, "step": 2495 }, { "epoch": 0.75, "learning_rate": 0.00017557592477893504, "loss": 1.7755, "step": 2496 }, { "epoch": 0.75, "learning_rate": 0.00017555463941262088, "loss": 1.8079, "step": 2497 }, { "epoch": 0.75, "learning_rate": 0.0001755333460667744, "loss": 1.8204, "step": 2498 }, { "epoch": 0.75, "learning_rate": 0.00017551204474364437, "loss": 1.8712, "step": 2499 }, { "epoch": 0.75, "learning_rate": 0.0001754907354454805, "loss": 1.8224, "step": 2500 }, { "epoch": 0.75, "learning_rate": 0.00017546941817453331, "loss": 1.7614, "step": 2501 }, { "epoch": 0.75, "learning_rate": 0.0001754480929330542, "loss": 1.7259, "step": 2502 }, { "epoch": 0.75, "learning_rate": 0.00017542675972329535, "loss": 1.7844, "step": 2503 }, { "epoch": 0.75, "learning_rate": 0.00017540541854750986, "loss": 1.809, "step": 2504 }, { "epoch": 0.75, "learning_rate": 0.00017538406940795158, "loss": 1.6952, "step": 2505 }, { "epoch": 0.75, "learning_rate": 0.0001753627123068753, "loss": 1.8176, "step": 2506 }, { "epoch": 0.75, "learning_rate": 0.00017534134724653657, "loss": 1.8305, "step": 2507 }, { "epoch": 0.75, "learning_rate": 0.00017531997422919185, "loss": 1.9522, "step": 2508 }, { "epoch": 0.76, "learning_rate": 0.00017529859325709837, "loss": 1.9366, "step": 2509 }, { "epoch": 0.76, "learning_rate": 0.00017527720433251426, "loss": 1.85, "step": 2510 }, { "epoch": 0.76, "learning_rate": 0.0001752558074576984, "loss": 1.9793, "step": 2511 }, { "epoch": 0.76, "learning_rate": 0.00017523440263491065, "loss": 1.8418, "step": 2512 }, { "epoch": 0.76, "learning_rate": 0.00017521298986641158, "loss": 1.9001, "step": 2513 }, { "epoch": 0.76, "learning_rate": 0.00017519156915446265, "loss": 1.9716, "step": 2514 }, { "epoch": 0.76, "learning_rate": 0.00017517014050132622, "loss": 1.8754, "step": 2515 }, { "epoch": 0.76, "learning_rate": 0.00017514870390926537, "loss": 1.8243, "step": 2516 }, { "epoch": 0.76, "learning_rate": 0.0001751272593805441, "loss": 1.8083, "step": 2517 }, { "epoch": 0.76, "learning_rate": 0.00017510580691742724, "loss": 1.846, "step": 2518 }, { "epoch": 0.76, "learning_rate": 0.0001750843465221804, "loss": 1.8387, "step": 2519 }, { "epoch": 0.76, "learning_rate": 0.0001750628781970701, "loss": 1.8126, "step": 2520 }, { "epoch": 0.76, "learning_rate": 0.00017504140194436366, "loss": 1.8805, "step": 2521 }, { "epoch": 0.76, "learning_rate": 0.0001750199177663293, "loss": 1.9159, "step": 2522 }, { "epoch": 0.76, "learning_rate": 0.0001749984256652359, "loss": 1.8771, "step": 2523 }, { "epoch": 0.76, "learning_rate": 0.00017497692564335342, "loss": 1.7528, "step": 2524 }, { "epoch": 0.76, "learning_rate": 0.00017495541770295246, "loss": 1.8432, "step": 2525 }, { "epoch": 0.76, "learning_rate": 0.00017493390184630459, "loss": 1.9077, "step": 2526 }, { "epoch": 0.76, "learning_rate": 0.0001749123780756821, "loss": 1.8713, "step": 2527 }, { "epoch": 0.76, "learning_rate": 0.00017489084639335823, "loss": 1.8014, "step": 2528 }, { "epoch": 0.76, "learning_rate": 0.00017486930680160696, "loss": 1.848, "step": 2529 }, { "epoch": 0.76, "learning_rate": 0.00017484775930270318, "loss": 1.9052, "step": 2530 }, { "epoch": 0.76, "learning_rate": 0.00017482620389892254, "loss": 1.6766, "step": 2531 }, { "epoch": 0.76, "learning_rate": 0.0001748046405925416, "loss": 1.8584, "step": 2532 }, { "epoch": 0.76, "learning_rate": 0.00017478306938583768, "loss": 1.8635, "step": 2533 }, { "epoch": 0.76, "learning_rate": 0.000174761490281089, "loss": 1.7421, "step": 2534 }, { "epoch": 0.76, "learning_rate": 0.00017473990328057464, "loss": 1.8182, "step": 2535 }, { "epoch": 0.76, "learning_rate": 0.00017471830838657434, "loss": 1.8686, "step": 2536 }, { "epoch": 0.76, "learning_rate": 0.00017469670560136893, "loss": 1.8143, "step": 2537 }, { "epoch": 0.76, "learning_rate": 0.00017467509492723985, "loss": 1.8961, "step": 2538 }, { "epoch": 0.76, "learning_rate": 0.00017465347636646947, "loss": 1.7658, "step": 2539 }, { "epoch": 0.76, "learning_rate": 0.00017463184992134103, "loss": 2.0107, "step": 2540 }, { "epoch": 0.76, "learning_rate": 0.0001746102155941385, "loss": 1.8306, "step": 2541 }, { "epoch": 0.77, "learning_rate": 0.0001745885733871468, "loss": 1.8877, "step": 2542 }, { "epoch": 0.77, "learning_rate": 0.0001745669233026516, "loss": 1.8369, "step": 2543 }, { "epoch": 0.77, "learning_rate": 0.00017454526534293937, "loss": 1.7765, "step": 2544 }, { "epoch": 0.77, "learning_rate": 0.00017452359951029752, "loss": 1.7395, "step": 2545 }, { "epoch": 0.77, "learning_rate": 0.00017450192580701428, "loss": 1.8614, "step": 2546 }, { "epoch": 0.77, "learning_rate": 0.0001744802442353786, "loss": 1.7904, "step": 2547 }, { "epoch": 0.77, "learning_rate": 0.00017445855479768037, "loss": 1.8815, "step": 2548 }, { "epoch": 0.77, "learning_rate": 0.00017443685749621025, "loss": 1.824, "step": 2549 }, { "epoch": 0.77, "learning_rate": 0.00017441515233325972, "loss": 1.8926, "step": 2550 }, { "epoch": 0.77, "learning_rate": 0.0001743934393111212, "loss": 1.9122, "step": 2551 }, { "epoch": 0.77, "learning_rate": 0.0001743717184320878, "loss": 1.7887, "step": 2552 }, { "epoch": 0.77, "learning_rate": 0.00017434998969845353, "loss": 1.913, "step": 2553 }, { "epoch": 0.77, "learning_rate": 0.00017432825311251325, "loss": 1.9541, "step": 2554 }, { "epoch": 0.77, "learning_rate": 0.0001743065086765626, "loss": 1.8795, "step": 2555 }, { "epoch": 0.77, "learning_rate": 0.00017428475639289805, "loss": 1.8453, "step": 2556 }, { "epoch": 0.77, "learning_rate": 0.00017426299626381697, "loss": 2.0257, "step": 2557 }, { "epoch": 0.77, "learning_rate": 0.00017424122829161744, "loss": 1.9182, "step": 2558 }, { "epoch": 0.77, "learning_rate": 0.0001742194524785985, "loss": 1.9072, "step": 2559 }, { "epoch": 0.77, "learning_rate": 0.0001741976688270599, "loss": 1.8894, "step": 2560 }, { "epoch": 0.77, "learning_rate": 0.00017417587733930233, "loss": 1.8248, "step": 2561 }, { "epoch": 0.77, "learning_rate": 0.0001741540780176272, "loss": 1.856, "step": 2562 }, { "epoch": 0.77, "learning_rate": 0.00017413227086433678, "loss": 1.8593, "step": 2563 }, { "epoch": 0.77, "learning_rate": 0.00017411045588173427, "loss": 1.8481, "step": 2564 }, { "epoch": 0.77, "learning_rate": 0.0001740886330721235, "loss": 1.9204, "step": 2565 }, { "epoch": 0.77, "learning_rate": 0.00017406680243780933, "loss": 1.9164, "step": 2566 }, { "epoch": 0.77, "learning_rate": 0.00017404496398109727, "loss": 1.817, "step": 2567 }, { "epoch": 0.77, "learning_rate": 0.00017402311770429383, "loss": 1.8679, "step": 2568 }, { "epoch": 0.77, "learning_rate": 0.00017400126360970618, "loss": 1.8626, "step": 2569 }, { "epoch": 0.77, "learning_rate": 0.00017397940169964243, "loss": 1.8168, "step": 2570 }, { "epoch": 0.77, "learning_rate": 0.00017395753197641146, "loss": 1.8179, "step": 2571 }, { "epoch": 0.77, "learning_rate": 0.00017393565444232301, "loss": 1.9066, "step": 2572 }, { "epoch": 0.77, "learning_rate": 0.00017391376909968757, "loss": 1.8727, "step": 2573 }, { "epoch": 0.77, "learning_rate": 0.0001738918759508166, "loss": 1.8386, "step": 2574 }, { "epoch": 0.78, "learning_rate": 0.00017386997499802226, "loss": 1.7932, "step": 2575 }, { "epoch": 0.78, "learning_rate": 0.0001738480662436175, "loss": 1.8598, "step": 2576 }, { "epoch": 0.78, "learning_rate": 0.00017382614968991625, "loss": 1.9009, "step": 2577 }, { "epoch": 0.78, "learning_rate": 0.00017380422533923314, "loss": 1.748, "step": 2578 }, { "epoch": 0.78, "learning_rate": 0.00017378229319388368, "loss": 1.8854, "step": 2579 }, { "epoch": 0.78, "learning_rate": 0.0001737603532561842, "loss": 1.9839, "step": 2580 }, { "epoch": 0.78, "learning_rate": 0.00017373840552845174, "loss": 1.9353, "step": 2581 }, { "epoch": 0.78, "learning_rate": 0.0001737164500130044, "loss": 1.8802, "step": 2582 }, { "epoch": 0.78, "learning_rate": 0.00017369448671216087, "loss": 1.8468, "step": 2583 }, { "epoch": 0.78, "learning_rate": 0.00017367251562824077, "loss": 1.843, "step": 2584 }, { "epoch": 0.78, "learning_rate": 0.00017365053676356452, "loss": 1.8617, "step": 2585 }, { "epoch": 0.78, "learning_rate": 0.0001736285501204534, "loss": 1.9452, "step": 2586 }, { "epoch": 0.78, "learning_rate": 0.00017360655570122948, "loss": 1.9003, "step": 2587 }, { "epoch": 0.78, "learning_rate": 0.0001735845535082156, "loss": 1.8338, "step": 2588 }, { "epoch": 0.78, "learning_rate": 0.0001735625435437355, "loss": 1.861, "step": 2589 }, { "epoch": 0.78, "learning_rate": 0.00017354052581011376, "loss": 1.901, "step": 2590 }, { "epoch": 0.78, "learning_rate": 0.00017351850030967563, "loss": 1.9044, "step": 2591 }, { "epoch": 0.78, "learning_rate": 0.00017349646704474737, "loss": 1.8293, "step": 2592 }, { "epoch": 0.78, "learning_rate": 0.00017347442601765595, "loss": 1.9359, "step": 2593 }, { "epoch": 0.78, "learning_rate": 0.00017345237723072918, "loss": 1.8751, "step": 2594 }, { "epoch": 0.78, "learning_rate": 0.0001734303206862957, "loss": 1.8259, "step": 2595 }, { "epoch": 0.78, "learning_rate": 0.00017340825638668494, "loss": 1.9038, "step": 2596 }, { "epoch": 0.78, "learning_rate": 0.00017338618433422717, "loss": 1.8636, "step": 2597 }, { "epoch": 0.78, "learning_rate": 0.00017336410453125354, "loss": 1.8442, "step": 2598 }, { "epoch": 0.78, "learning_rate": 0.0001733420169800959, "loss": 1.8768, "step": 2599 }, { "epoch": 0.78, "learning_rate": 0.00017331992168308696, "loss": 1.9013, "step": 2600 }, { "epoch": 0.78, "learning_rate": 0.0001732978186425603, "loss": 1.8446, "step": 2601 }, { "epoch": 0.78, "learning_rate": 0.00017327570786085034, "loss": 1.8344, "step": 2602 }, { "epoch": 0.78, "learning_rate": 0.0001732535893402921, "loss": 1.8707, "step": 2603 }, { "epoch": 0.78, "learning_rate": 0.00017323146308322173, "loss": 1.8148, "step": 2604 }, { "epoch": 0.78, "learning_rate": 0.000173209329091976, "loss": 1.8944, "step": 2605 }, { "epoch": 0.78, "learning_rate": 0.00017318718736889253, "loss": 1.7577, "step": 2606 }, { "epoch": 0.78, "learning_rate": 0.00017316503791630977, "loss": 1.9606, "step": 2607 }, { "epoch": 0.79, "learning_rate": 0.00017314288073656695, "loss": 1.8458, "step": 2608 }, { "epoch": 0.79, "learning_rate": 0.0001731207158320042, "loss": 1.8101, "step": 2609 }, { "epoch": 0.79, "learning_rate": 0.0001730985432049624, "loss": 1.9099, "step": 2610 }, { "epoch": 0.79, "learning_rate": 0.00017307636285778327, "loss": 1.7686, "step": 2611 }, { "epoch": 0.79, "learning_rate": 0.00017305417479280934, "loss": 1.8426, "step": 2612 }, { "epoch": 0.79, "learning_rate": 0.00017303197901238392, "loss": 1.7403, "step": 2613 }, { "epoch": 0.79, "learning_rate": 0.00017300977551885121, "loss": 1.779, "step": 2614 }, { "epoch": 0.79, "learning_rate": 0.00017298756431455615, "loss": 1.8771, "step": 2615 }, { "epoch": 0.79, "learning_rate": 0.00017296534540184454, "loss": 1.8684, "step": 2616 }, { "epoch": 0.79, "learning_rate": 0.00017294311878306298, "loss": 1.9139, "step": 2617 }, { "epoch": 0.79, "learning_rate": 0.00017292088446055885, "loss": 1.9153, "step": 2618 }, { "epoch": 0.79, "learning_rate": 0.00017289864243668046, "loss": 1.8367, "step": 2619 }, { "epoch": 0.79, "learning_rate": 0.00017287639271377678, "loss": 1.907, "step": 2620 }, { "epoch": 0.79, "learning_rate": 0.00017285413529419767, "loss": 1.8925, "step": 2621 }, { "epoch": 0.79, "learning_rate": 0.00017283187018029384, "loss": 1.8334, "step": 2622 }, { "epoch": 0.79, "learning_rate": 0.0001728095973744167, "loss": 1.8637, "step": 2623 }, { "epoch": 0.79, "learning_rate": 0.0001727873168789186, "loss": 1.8591, "step": 2624 }, { "epoch": 0.79, "learning_rate": 0.00017276502869615265, "loss": 1.9134, "step": 2625 }, { "epoch": 0.79, "learning_rate": 0.00017274273282847274, "loss": 1.8942, "step": 2626 }, { "epoch": 0.79, "learning_rate": 0.00017272042927823358, "loss": 1.9189, "step": 2627 }, { "epoch": 0.79, "learning_rate": 0.00017269811804779072, "loss": 1.8414, "step": 2628 }, { "epoch": 0.79, "learning_rate": 0.00017267579913950054, "loss": 1.8317, "step": 2629 }, { "epoch": 0.79, "learning_rate": 0.0001726534725557202, "loss": 1.9096, "step": 2630 }, { "epoch": 0.79, "learning_rate": 0.00017263113829880765, "loss": 1.8888, "step": 2631 }, { "epoch": 0.79, "learning_rate": 0.00017260879637112165, "loss": 1.8016, "step": 2632 }, { "epoch": 0.79, "learning_rate": 0.0001725864467750218, "loss": 1.8812, "step": 2633 }, { "epoch": 0.79, "learning_rate": 0.00017256408951286852, "loss": 1.9318, "step": 2634 }, { "epoch": 0.79, "learning_rate": 0.00017254172458702307, "loss": 1.8197, "step": 2635 }, { "epoch": 0.79, "learning_rate": 0.0001725193519998474, "loss": 1.8849, "step": 2636 }, { "epoch": 0.79, "learning_rate": 0.00017249697175370434, "loss": 1.8337, "step": 2637 }, { "epoch": 0.79, "learning_rate": 0.0001724745838509576, "loss": 1.7135, "step": 2638 }, { "epoch": 0.79, "learning_rate": 0.00017245218829397153, "loss": 1.8131, "step": 2639 }, { "epoch": 0.79, "learning_rate": 0.00017242978508511146, "loss": 1.8213, "step": 2640 }, { "epoch": 0.8, "learning_rate": 0.0001724073742267434, "loss": 1.8705, "step": 2641 }, { "epoch": 0.8, "learning_rate": 0.00017238495572123428, "loss": 1.8592, "step": 2642 }, { "epoch": 0.8, "learning_rate": 0.00017236252957095175, "loss": 1.809, "step": 2643 }, { "epoch": 0.8, "learning_rate": 0.0001723400957782643, "loss": 1.8897, "step": 2644 }, { "epoch": 0.8, "learning_rate": 0.0001723176543455412, "loss": 1.8894, "step": 2645 }, { "epoch": 0.8, "learning_rate": 0.0001722952052751526, "loss": 1.794, "step": 2646 }, { "epoch": 0.8, "learning_rate": 0.00017227274856946938, "loss": 1.8345, "step": 2647 }, { "epoch": 0.8, "learning_rate": 0.00017225028423086322, "loss": 1.8676, "step": 2648 }, { "epoch": 0.8, "learning_rate": 0.0001722278122617067, "loss": 1.7691, "step": 2649 }, { "epoch": 0.8, "learning_rate": 0.00017220533266437314, "loss": 1.9031, "step": 2650 }, { "epoch": 0.8, "learning_rate": 0.00017218284544123663, "loss": 1.8861, "step": 2651 }, { "epoch": 0.8, "learning_rate": 0.00017216035059467214, "loss": 1.8746, "step": 2652 }, { "epoch": 0.8, "learning_rate": 0.0001721378481270554, "loss": 1.776, "step": 2653 }, { "epoch": 0.8, "learning_rate": 0.00017211533804076297, "loss": 1.7132, "step": 2654 }, { "epoch": 0.8, "learning_rate": 0.00017209282033817217, "loss": 1.8764, "step": 2655 }, { "epoch": 0.8, "learning_rate": 0.00017207029502166118, "loss": 1.8638, "step": 2656 }, { "epoch": 0.8, "learning_rate": 0.000172047762093609, "loss": 1.9291, "step": 2657 }, { "epoch": 0.8, "learning_rate": 0.00017202522155639532, "loss": 1.9547, "step": 2658 }, { "epoch": 0.8, "learning_rate": 0.00017200267341240075, "loss": 1.8234, "step": 2659 }, { "epoch": 0.8, "learning_rate": 0.00017198011766400665, "loss": 1.8795, "step": 2660 }, { "epoch": 0.8, "learning_rate": 0.0001719575543135952, "loss": 1.8285, "step": 2661 }, { "epoch": 0.8, "learning_rate": 0.0001719349833635494, "loss": 1.8186, "step": 2662 }, { "epoch": 0.8, "learning_rate": 0.00017191240481625295, "loss": 1.846, "step": 2663 }, { "epoch": 0.8, "learning_rate": 0.0001718898186740905, "loss": 1.804, "step": 2664 }, { "epoch": 0.8, "learning_rate": 0.00017186722493944744, "loss": 1.7716, "step": 2665 }, { "epoch": 0.8, "learning_rate": 0.00017184462361470994, "loss": 1.7641, "step": 2666 }, { "epoch": 0.8, "learning_rate": 0.00017182201470226498, "loss": 1.8846, "step": 2667 }, { "epoch": 0.8, "learning_rate": 0.00017179939820450034, "loss": 1.9486, "step": 2668 }, { "epoch": 0.8, "learning_rate": 0.00017177677412380464, "loss": 1.962, "step": 2669 }, { "epoch": 0.8, "learning_rate": 0.00017175414246256722, "loss": 1.9241, "step": 2670 }, { "epoch": 0.8, "learning_rate": 0.00017173150322317835, "loss": 1.7877, "step": 2671 }, { "epoch": 0.8, "learning_rate": 0.00017170885640802896, "loss": 1.8946, "step": 2672 }, { "epoch": 0.8, "learning_rate": 0.00017168620201951087, "loss": 1.7765, "step": 2673 }, { "epoch": 0.8, "learning_rate": 0.00017166354006001666, "loss": 1.9373, "step": 2674 }, { "epoch": 0.81, "learning_rate": 0.00017164087053193972, "loss": 1.8304, "step": 2675 }, { "epoch": 0.81, "learning_rate": 0.00017161819343767425, "loss": 1.9234, "step": 2676 }, { "epoch": 0.81, "learning_rate": 0.00017159550877961523, "loss": 1.8443, "step": 2677 }, { "epoch": 0.81, "learning_rate": 0.00017157281656015846, "loss": 1.8489, "step": 2678 }, { "epoch": 0.81, "learning_rate": 0.0001715501167817005, "loss": 1.7013, "step": 2679 }, { "epoch": 0.81, "learning_rate": 0.0001715274094466388, "loss": 1.8016, "step": 2680 }, { "epoch": 0.81, "learning_rate": 0.00017150469455737147, "loss": 1.6703, "step": 2681 }, { "epoch": 0.81, "learning_rate": 0.00017148197211629753, "loss": 1.8543, "step": 2682 }, { "epoch": 0.81, "learning_rate": 0.00017145924212581676, "loss": 1.811, "step": 2683 }, { "epoch": 0.81, "learning_rate": 0.00017143650458832971, "loss": 1.9219, "step": 2684 }, { "epoch": 0.81, "learning_rate": 0.0001714137595062378, "loss": 1.8412, "step": 2685 }, { "epoch": 0.81, "learning_rate": 0.00017139100688194313, "loss": 1.8681, "step": 2686 }, { "epoch": 0.81, "learning_rate": 0.00017136824671784874, "loss": 1.9752, "step": 2687 }, { "epoch": 0.81, "learning_rate": 0.0001713454790163584, "loss": 2.0121, "step": 2688 }, { "epoch": 0.81, "learning_rate": 0.0001713227037798766, "loss": 1.9884, "step": 2689 }, { "epoch": 0.81, "learning_rate": 0.0001712999210108087, "loss": 1.7959, "step": 2690 }, { "epoch": 0.81, "learning_rate": 0.0001712771307115609, "loss": 1.8669, "step": 2691 }, { "epoch": 0.81, "learning_rate": 0.00017125433288454013, "loss": 1.7998, "step": 2692 }, { "epoch": 0.81, "learning_rate": 0.0001712315275321541, "loss": 1.7524, "step": 2693 }, { "epoch": 0.81, "learning_rate": 0.0001712087146568114, "loss": 1.8258, "step": 2694 }, { "epoch": 0.81, "learning_rate": 0.00017118589426092133, "loss": 1.8304, "step": 2695 }, { "epoch": 0.81, "learning_rate": 0.00017116306634689402, "loss": 1.9548, "step": 2696 }, { "epoch": 0.81, "learning_rate": 0.00017114023091714038, "loss": 1.8918, "step": 2697 }, { "epoch": 0.81, "learning_rate": 0.0001711173879740721, "loss": 1.9462, "step": 2698 }, { "epoch": 0.81, "learning_rate": 0.00017109453752010172, "loss": 2.036, "step": 2699 }, { "epoch": 0.81, "learning_rate": 0.00017107167955764254, "loss": 1.8582, "step": 2700 }, { "epoch": 0.81, "learning_rate": 0.00017104881408910866, "loss": 1.7689, "step": 2701 }, { "epoch": 0.81, "learning_rate": 0.00017102594111691492, "loss": 1.9072, "step": 2702 }, { "epoch": 0.81, "learning_rate": 0.00017100306064347706, "loss": 1.8778, "step": 2703 }, { "epoch": 0.81, "learning_rate": 0.00017098017267121147, "loss": 1.9334, "step": 2704 }, { "epoch": 0.81, "learning_rate": 0.00017095727720253549, "loss": 1.8754, "step": 2705 }, { "epoch": 0.81, "learning_rate": 0.00017093437423986712, "loss": 1.9191, "step": 2706 }, { "epoch": 0.81, "learning_rate": 0.00017091146378562524, "loss": 1.9309, "step": 2707 }, { "epoch": 0.82, "learning_rate": 0.0001708885458422295, "loss": 2.0004, "step": 2708 }, { "epoch": 0.82, "learning_rate": 0.00017086562041210025, "loss": 1.9803, "step": 2709 }, { "epoch": 0.82, "learning_rate": 0.00017084268749765875, "loss": 1.8618, "step": 2710 }, { "epoch": 0.82, "learning_rate": 0.000170819747101327, "loss": 1.8995, "step": 2711 }, { "epoch": 0.82, "learning_rate": 0.00017079679922552785, "loss": 1.9323, "step": 2712 }, { "epoch": 0.82, "learning_rate": 0.00017077384387268482, "loss": 1.9158, "step": 2713 }, { "epoch": 0.82, "learning_rate": 0.00017075088104522235, "loss": 1.7973, "step": 2714 }, { "epoch": 0.82, "learning_rate": 0.00017072791074556555, "loss": 1.8971, "step": 2715 }, { "epoch": 0.82, "learning_rate": 0.00017070493297614042, "loss": 1.9048, "step": 2716 }, { "epoch": 0.82, "learning_rate": 0.00017068194773937366, "loss": 1.8367, "step": 2717 }, { "epoch": 0.82, "learning_rate": 0.0001706589550376928, "loss": 1.8338, "step": 2718 }, { "epoch": 0.82, "learning_rate": 0.0001706359548735262, "loss": 1.7512, "step": 2719 }, { "epoch": 0.82, "learning_rate": 0.000170612947249303, "loss": 1.8404, "step": 2720 }, { "epoch": 0.82, "learning_rate": 0.00017058993216745308, "loss": 1.922, "step": 2721 }, { "epoch": 0.82, "learning_rate": 0.00017056690963040706, "loss": 1.9152, "step": 2722 }, { "epoch": 0.82, "learning_rate": 0.00017054387964059644, "loss": 1.8591, "step": 2723 }, { "epoch": 0.82, "learning_rate": 0.00017052084220045352, "loss": 1.863, "step": 2724 }, { "epoch": 0.82, "learning_rate": 0.00017049779731241133, "loss": 1.7856, "step": 2725 }, { "epoch": 0.82, "learning_rate": 0.0001704747449789037, "loss": 1.7945, "step": 2726 }, { "epoch": 0.82, "learning_rate": 0.00017045168520236524, "loss": 1.9265, "step": 2727 }, { "epoch": 0.82, "learning_rate": 0.00017042861798523138, "loss": 1.8799, "step": 2728 }, { "epoch": 0.82, "learning_rate": 0.00017040554332993827, "loss": 1.8423, "step": 2729 }, { "epoch": 0.82, "learning_rate": 0.0001703824612389229, "loss": 1.8718, "step": 2730 }, { "epoch": 0.82, "learning_rate": 0.0001703593717146231, "loss": 1.83, "step": 2731 }, { "epoch": 0.82, "learning_rate": 0.00017033627475947737, "loss": 1.8923, "step": 2732 }, { "epoch": 0.82, "learning_rate": 0.000170313170375925, "loss": 1.8125, "step": 2733 }, { "epoch": 0.82, "learning_rate": 0.00017029005856640614, "loss": 1.8924, "step": 2734 }, { "epoch": 0.82, "learning_rate": 0.00017026693933336172, "loss": 1.9608, "step": 2735 }, { "epoch": 0.82, "learning_rate": 0.0001702438126792334, "loss": 1.8256, "step": 2736 }, { "epoch": 0.82, "learning_rate": 0.00017022067860646364, "loss": 1.7847, "step": 2737 }, { "epoch": 0.82, "learning_rate": 0.00017019753711749571, "loss": 1.8251, "step": 2738 }, { "epoch": 0.82, "learning_rate": 0.00017017438821477363, "loss": 1.928, "step": 2739 }, { "epoch": 0.82, "learning_rate": 0.00017015123190074224, "loss": 1.8457, "step": 2740 }, { "epoch": 0.83, "learning_rate": 0.00017012806817784715, "loss": 1.7519, "step": 2741 }, { "epoch": 0.83, "learning_rate": 0.0001701048970485347, "loss": 1.8693, "step": 2742 }, { "epoch": 0.83, "learning_rate": 0.00017008171851525208, "loss": 1.8258, "step": 2743 }, { "epoch": 0.83, "learning_rate": 0.00017005853258044725, "loss": 1.8448, "step": 2744 }, { "epoch": 0.83, "learning_rate": 0.00017003533924656893, "loss": 1.827, "step": 2745 }, { "epoch": 0.83, "learning_rate": 0.00017001213851606663, "loss": 1.8012, "step": 2746 }, { "epoch": 0.83, "learning_rate": 0.00016998893039139062, "loss": 1.8695, "step": 2747 }, { "epoch": 0.83, "learning_rate": 0.00016996571487499205, "loss": 1.9461, "step": 2748 }, { "epoch": 0.83, "learning_rate": 0.0001699424919693227, "loss": 1.7804, "step": 2749 }, { "epoch": 0.83, "learning_rate": 0.0001699192616768352, "loss": 1.9137, "step": 2750 }, { "epoch": 0.83, "learning_rate": 0.00016989602399998305, "loss": 1.8335, "step": 2751 }, { "epoch": 0.83, "learning_rate": 0.00016987277894122034, "loss": 1.8819, "step": 2752 }, { "epoch": 0.83, "learning_rate": 0.0001698495265030021, "loss": 1.8265, "step": 2753 }, { "epoch": 0.83, "learning_rate": 0.00016982626668778406, "loss": 1.9613, "step": 2754 }, { "epoch": 0.83, "learning_rate": 0.0001698029994980228, "loss": 1.8387, "step": 2755 }, { "epoch": 0.83, "learning_rate": 0.00016977972493617557, "loss": 1.944, "step": 2756 }, { "epoch": 0.83, "learning_rate": 0.00016975644300470052, "loss": 1.885, "step": 2757 }, { "epoch": 0.83, "learning_rate": 0.00016973315370605644, "loss": 1.8196, "step": 2758 }, { "epoch": 0.83, "learning_rate": 0.00016970985704270307, "loss": 1.8242, "step": 2759 }, { "epoch": 0.83, "learning_rate": 0.00016968655301710076, "loss": 1.8695, "step": 2760 }, { "epoch": 0.83, "learning_rate": 0.00016966324163171075, "loss": 1.9309, "step": 2761 }, { "epoch": 0.83, "learning_rate": 0.000169639922888995, "loss": 1.9114, "step": 2762 }, { "epoch": 0.83, "learning_rate": 0.0001696165967914163, "loss": 1.9141, "step": 2763 }, { "epoch": 0.83, "learning_rate": 0.00016959326334143815, "loss": 1.8252, "step": 2764 }, { "epoch": 0.83, "learning_rate": 0.00016956992254152486, "loss": 1.8607, "step": 2765 }, { "epoch": 0.83, "learning_rate": 0.0001695465743941415, "loss": 1.8395, "step": 2766 }, { "epoch": 0.83, "learning_rate": 0.000169523218901754, "loss": 1.7103, "step": 2767 }, { "epoch": 0.83, "learning_rate": 0.00016949985606682895, "loss": 1.9529, "step": 2768 }, { "epoch": 0.83, "learning_rate": 0.00016947648589183377, "loss": 1.8256, "step": 2769 }, { "epoch": 0.83, "learning_rate": 0.0001694531083792366, "loss": 1.8145, "step": 2770 }, { "epoch": 0.83, "learning_rate": 0.00016942972353150648, "loss": 1.7855, "step": 2771 }, { "epoch": 0.83, "learning_rate": 0.00016940633135111312, "loss": 1.8357, "step": 2772 }, { "epoch": 0.83, "learning_rate": 0.00016938293184052704, "loss": 1.7977, "step": 2773 }, { "epoch": 0.84, "learning_rate": 0.0001693595250022195, "loss": 1.7313, "step": 2774 }, { "epoch": 0.84, "learning_rate": 0.00016933611083866254, "loss": 1.8564, "step": 2775 }, { "epoch": 0.84, "learning_rate": 0.00016931268935232907, "loss": 1.867, "step": 2776 }, { "epoch": 0.84, "learning_rate": 0.00016928926054569262, "loss": 1.771, "step": 2777 }, { "epoch": 0.84, "learning_rate": 0.00016926582442122764, "loss": 1.9773, "step": 2778 }, { "epoch": 0.84, "learning_rate": 0.00016924238098140927, "loss": 1.845, "step": 2779 }, { "epoch": 0.84, "learning_rate": 0.00016921893022871337, "loss": 1.8162, "step": 2780 }, { "epoch": 0.84, "learning_rate": 0.0001691954721656167, "loss": 1.8536, "step": 2781 }, { "epoch": 0.84, "learning_rate": 0.00016917200679459672, "loss": 1.8854, "step": 2782 }, { "epoch": 0.84, "learning_rate": 0.00016914853411813165, "loss": 1.8978, "step": 2783 }, { "epoch": 0.84, "learning_rate": 0.00016912505413870055, "loss": 1.8851, "step": 2784 }, { "epoch": 0.84, "learning_rate": 0.00016910156685878315, "loss": 1.9287, "step": 2785 }, { "epoch": 0.84, "learning_rate": 0.00016907807228086003, "loss": 1.7751, "step": 2786 }, { "epoch": 0.84, "learning_rate": 0.00016905457040741254, "loss": 1.8532, "step": 2787 }, { "epoch": 0.84, "learning_rate": 0.00016903106124092275, "loss": 1.779, "step": 2788 }, { "epoch": 0.84, "learning_rate": 0.0001690075447838735, "loss": 1.8202, "step": 2789 }, { "epoch": 0.84, "learning_rate": 0.0001689840210387485, "loss": 1.9283, "step": 2790 }, { "epoch": 0.84, "learning_rate": 0.0001689604900080321, "loss": 1.9254, "step": 2791 }, { "epoch": 0.84, "learning_rate": 0.00016893695169420948, "loss": 1.8165, "step": 2792 }, { "epoch": 0.84, "learning_rate": 0.00016891340609976663, "loss": 1.8237, "step": 2793 }, { "epoch": 0.84, "learning_rate": 0.0001688898532271902, "loss": 1.8781, "step": 2794 }, { "epoch": 0.84, "learning_rate": 0.00016886629307896772, "loss": 1.7863, "step": 2795 }, { "epoch": 0.84, "learning_rate": 0.00016884272565758743, "loss": 1.8527, "step": 2796 }, { "epoch": 0.84, "learning_rate": 0.00016881915096553834, "loss": 1.829, "step": 2797 }, { "epoch": 0.84, "learning_rate": 0.00016879556900531028, "loss": 1.7381, "step": 2798 }, { "epoch": 0.84, "learning_rate": 0.00016877197977939375, "loss": 1.7654, "step": 2799 }, { "epoch": 0.84, "learning_rate": 0.00016874838329028008, "loss": 1.823, "step": 2800 }, { "epoch": 0.84, "learning_rate": 0.00016872477954046138, "loss": 1.8064, "step": 2801 }, { "epoch": 0.84, "learning_rate": 0.0001687011685324305, "loss": 1.9289, "step": 2802 }, { "epoch": 0.84, "learning_rate": 0.0001686775502686811, "loss": 1.8903, "step": 2803 }, { "epoch": 0.84, "learning_rate": 0.0001686539247517075, "loss": 1.8052, "step": 2804 }, { "epoch": 0.84, "learning_rate": 0.00016863029198400492, "loss": 1.7765, "step": 2805 }, { "epoch": 0.84, "learning_rate": 0.00016860665196806923, "loss": 1.9262, "step": 2806 }, { "epoch": 0.84, "learning_rate": 0.0001685830047063971, "loss": 1.8654, "step": 2807 }, { "epoch": 0.85, "learning_rate": 0.00016855935020148607, "loss": 1.9078, "step": 2808 }, { "epoch": 0.85, "learning_rate": 0.00016853568845583426, "loss": 1.7909, "step": 2809 }, { "epoch": 0.85, "learning_rate": 0.00016851201947194074, "loss": 1.8282, "step": 2810 }, { "epoch": 0.85, "learning_rate": 0.00016848834325230522, "loss": 1.8182, "step": 2811 }, { "epoch": 0.85, "learning_rate": 0.0001684646597994282, "loss": 1.8557, "step": 2812 }, { "epoch": 0.85, "learning_rate": 0.00016844096911581096, "loss": 1.8197, "step": 2813 }, { "epoch": 0.85, "learning_rate": 0.00016841727120395553, "loss": 1.7319, "step": 2814 }, { "epoch": 0.85, "learning_rate": 0.0001683935660663647, "loss": 1.8762, "step": 2815 }, { "epoch": 0.85, "learning_rate": 0.00016836985370554208, "loss": 1.7522, "step": 2816 }, { "epoch": 0.85, "learning_rate": 0.00016834613412399193, "loss": 1.8292, "step": 2817 }, { "epoch": 0.85, "learning_rate": 0.00016832240732421946, "loss": 1.859, "step": 2818 }, { "epoch": 0.85, "learning_rate": 0.00016829867330873038, "loss": 1.785, "step": 2819 }, { "epoch": 0.85, "learning_rate": 0.00016827493208003137, "loss": 1.8879, "step": 2820 }, { "epoch": 0.85, "learning_rate": 0.00016825118364062981, "loss": 1.8163, "step": 2821 }, { "epoch": 0.85, "learning_rate": 0.00016822742799303385, "loss": 1.7483, "step": 2822 }, { "epoch": 0.85, "learning_rate": 0.00016820366513975237, "loss": 1.7803, "step": 2823 }, { "epoch": 0.85, "learning_rate": 0.00016817989508329502, "loss": 1.7881, "step": 2824 }, { "epoch": 0.85, "learning_rate": 0.00016815611782617222, "loss": 1.8411, "step": 2825 }, { "epoch": 0.85, "learning_rate": 0.00016813233337089516, "loss": 1.8938, "step": 2826 }, { "epoch": 0.85, "learning_rate": 0.0001681085417199758, "loss": 1.9098, "step": 2827 }, { "epoch": 0.85, "learning_rate": 0.00016808474287592683, "loss": 1.8781, "step": 2828 }, { "epoch": 0.85, "learning_rate": 0.0001680609368412617, "loss": 1.8529, "step": 2829 }, { "epoch": 0.85, "learning_rate": 0.00016803712361849465, "loss": 1.9206, "step": 2830 }, { "epoch": 0.85, "learning_rate": 0.00016801330321014062, "loss": 1.8733, "step": 2831 }, { "epoch": 0.85, "learning_rate": 0.0001679894756187154, "loss": 1.8176, "step": 2832 }, { "epoch": 0.85, "learning_rate": 0.00016796564084673544, "loss": 1.8066, "step": 2833 }, { "epoch": 0.85, "learning_rate": 0.00016794179889671802, "loss": 1.8764, "step": 2834 }, { "epoch": 0.85, "learning_rate": 0.00016791794977118115, "loss": 1.8322, "step": 2835 }, { "epoch": 0.85, "learning_rate": 0.0001678940934726436, "loss": 1.9033, "step": 2836 }, { "epoch": 0.85, "learning_rate": 0.00016787023000362493, "loss": 1.823, "step": 2837 }, { "epoch": 0.85, "learning_rate": 0.00016784635936664536, "loss": 1.9154, "step": 2838 }, { "epoch": 0.85, "learning_rate": 0.000167822481564226, "loss": 1.8078, "step": 2839 }, { "epoch": 0.85, "learning_rate": 0.00016779859659888863, "loss": 1.7901, "step": 2840 }, { "epoch": 0.86, "learning_rate": 0.00016777470447315576, "loss": 1.8586, "step": 2841 }, { "epoch": 0.86, "learning_rate": 0.00016775080518955076, "loss": 1.8451, "step": 2842 }, { "epoch": 0.86, "learning_rate": 0.0001677268987505977, "loss": 1.875, "step": 2843 }, { "epoch": 0.86, "learning_rate": 0.00016770298515882134, "loss": 1.8406, "step": 2844 }, { "epoch": 0.86, "learning_rate": 0.00016767906441674733, "loss": 1.7625, "step": 2845 }, { "epoch": 0.86, "learning_rate": 0.000167655136526902, "loss": 1.8501, "step": 2846 }, { "epoch": 0.86, "learning_rate": 0.0001676312014918124, "loss": 1.9249, "step": 2847 }, { "epoch": 0.86, "learning_rate": 0.0001676072593140064, "loss": 1.9367, "step": 2848 }, { "epoch": 0.86, "learning_rate": 0.00016758330999601257, "loss": 1.8953, "step": 2849 }, { "epoch": 0.86, "learning_rate": 0.0001675593535403603, "loss": 1.7569, "step": 2850 }, { "epoch": 0.86, "learning_rate": 0.0001675353899495797, "loss": 1.8796, "step": 2851 }, { "epoch": 0.86, "learning_rate": 0.0001675114192262016, "loss": 1.9335, "step": 2852 }, { "epoch": 0.86, "learning_rate": 0.00016748744137275761, "loss": 1.8757, "step": 2853 }, { "epoch": 0.86, "learning_rate": 0.00016746345639178013, "loss": 1.8511, "step": 2854 }, { "epoch": 0.86, "learning_rate": 0.0001674394642858023, "loss": 1.7491, "step": 2855 }, { "epoch": 0.86, "learning_rate": 0.00016741546505735788, "loss": 1.8234, "step": 2856 }, { "epoch": 0.86, "learning_rate": 0.00016739145870898163, "loss": 1.8913, "step": 2857 }, { "epoch": 0.86, "learning_rate": 0.00016736744524320884, "loss": 1.9767, "step": 2858 }, { "epoch": 0.86, "learning_rate": 0.00016734342466257567, "loss": 1.6626, "step": 2859 }, { "epoch": 0.86, "learning_rate": 0.000167319396969619, "loss": 1.8329, "step": 2860 }, { "epoch": 0.86, "learning_rate": 0.00016729536216687643, "loss": 1.8346, "step": 2861 }, { "epoch": 0.86, "learning_rate": 0.0001672713202568864, "loss": 1.7987, "step": 2862 }, { "epoch": 0.86, "learning_rate": 0.00016724727124218798, "loss": 1.8692, "step": 2863 }, { "epoch": 0.86, "learning_rate": 0.00016722321512532108, "loss": 1.8279, "step": 2864 }, { "epoch": 0.86, "learning_rate": 0.00016719915190882635, "loss": 1.8002, "step": 2865 }, { "epoch": 0.86, "learning_rate": 0.00016717508159524508, "loss": 1.7794, "step": 2866 }, { "epoch": 0.86, "learning_rate": 0.00016715100418711955, "loss": 1.7686, "step": 2867 }, { "epoch": 0.86, "learning_rate": 0.00016712691968699258, "loss": 1.863, "step": 2868 }, { "epoch": 0.86, "learning_rate": 0.00016710282809740776, "loss": 1.8457, "step": 2869 }, { "epoch": 0.86, "learning_rate": 0.00016707872942090947, "loss": 1.8084, "step": 2870 }, { "epoch": 0.86, "learning_rate": 0.0001670546236600429, "loss": 1.7028, "step": 2871 }, { "epoch": 0.86, "learning_rate": 0.00016703051081735386, "loss": 1.8497, "step": 2872 }, { "epoch": 0.86, "learning_rate": 0.000167006390895389, "loss": 1.7399, "step": 2873 }, { "epoch": 0.87, "learning_rate": 0.0001669822638966957, "loss": 1.8568, "step": 2874 }, { "epoch": 0.87, "learning_rate": 0.0001669581298238221, "loss": 1.8126, "step": 2875 }, { "epoch": 0.87, "learning_rate": 0.00016693398867931701, "loss": 1.7994, "step": 2876 }, { "epoch": 0.87, "learning_rate": 0.00016690984046573008, "loss": 1.8352, "step": 2877 }, { "epoch": 0.87, "learning_rate": 0.00016688568518561162, "loss": 1.8149, "step": 2878 }, { "epoch": 0.87, "learning_rate": 0.00016686152284151281, "loss": 1.7097, "step": 2879 }, { "epoch": 0.87, "learning_rate": 0.00016683735343598545, "loss": 1.7524, "step": 2880 }, { "epoch": 0.87, "learning_rate": 0.00016681317697158217, "loss": 1.7798, "step": 2881 }, { "epoch": 0.87, "learning_rate": 0.0001667889934508563, "loss": 1.779, "step": 2882 }, { "epoch": 0.87, "learning_rate": 0.0001667648028763619, "loss": 1.8614, "step": 2883 }, { "epoch": 0.87, "learning_rate": 0.00016674060525065381, "loss": 1.8233, "step": 2884 }, { "epoch": 0.87, "learning_rate": 0.00016671640057628766, "loss": 1.9064, "step": 2885 }, { "epoch": 0.87, "learning_rate": 0.0001666921888558197, "loss": 1.7189, "step": 2886 }, { "epoch": 0.87, "learning_rate": 0.00016666797009180705, "loss": 1.7994, "step": 2887 }, { "epoch": 0.87, "learning_rate": 0.0001666437442868075, "loss": 1.7472, "step": 2888 }, { "epoch": 0.87, "learning_rate": 0.0001666195114433796, "loss": 1.819, "step": 2889 }, { "epoch": 0.87, "learning_rate": 0.00016659527156408262, "loss": 1.8652, "step": 2890 }, { "epoch": 0.87, "learning_rate": 0.00016657102465147668, "loss": 1.8454, "step": 2891 }, { "epoch": 0.87, "learning_rate": 0.00016654677070812249, "loss": 1.9238, "step": 2892 }, { "epoch": 0.87, "learning_rate": 0.0001665225097365816, "loss": 1.8446, "step": 2893 }, { "epoch": 0.87, "learning_rate": 0.00016649824173941625, "loss": 1.8884, "step": 2894 }, { "epoch": 0.87, "learning_rate": 0.00016647396671918953, "loss": 1.8155, "step": 2895 }, { "epoch": 0.87, "learning_rate": 0.00016644968467846512, "loss": 1.8411, "step": 2896 }, { "epoch": 0.87, "learning_rate": 0.0001664253956198075, "loss": 1.8971, "step": 2897 }, { "epoch": 0.87, "learning_rate": 0.00016640109954578196, "loss": 1.9244, "step": 2898 }, { "epoch": 0.87, "learning_rate": 0.00016637679645895443, "loss": 1.8217, "step": 2899 }, { "epoch": 0.87, "learning_rate": 0.00016635248636189166, "loss": 1.8282, "step": 2900 }, { "epoch": 0.87, "learning_rate": 0.0001663281692571611, "loss": 1.7193, "step": 2901 }, { "epoch": 0.87, "learning_rate": 0.00016630384514733094, "loss": 1.8008, "step": 2902 }, { "epoch": 0.87, "learning_rate": 0.00016627951403497008, "loss": 1.89, "step": 2903 }, { "epoch": 0.87, "learning_rate": 0.00016625517592264825, "loss": 2.0, "step": 2904 }, { "epoch": 0.87, "learning_rate": 0.00016623083081293585, "loss": 1.9359, "step": 2905 }, { "epoch": 0.87, "learning_rate": 0.000166206478708404, "loss": 1.9654, "step": 2906 }, { "epoch": 0.88, "learning_rate": 0.00016618211961162466, "loss": 1.8881, "step": 2907 }, { "epoch": 0.88, "learning_rate": 0.00016615775352517038, "loss": 1.8973, "step": 2908 }, { "epoch": 0.88, "learning_rate": 0.0001661333804516146, "loss": 1.8541, "step": 2909 }, { "epoch": 0.88, "learning_rate": 0.00016610900039353138, "loss": 1.9076, "step": 2910 }, { "epoch": 0.88, "learning_rate": 0.0001660846133534956, "loss": 1.8617, "step": 2911 }, { "epoch": 0.88, "learning_rate": 0.0001660602193340828, "loss": 1.8987, "step": 2912 }, { "epoch": 0.88, "learning_rate": 0.00016603581833786932, "loss": 1.8955, "step": 2913 }, { "epoch": 0.88, "learning_rate": 0.0001660114103674322, "loss": 1.8481, "step": 2914 }, { "epoch": 0.88, "learning_rate": 0.0001659869954253493, "loss": 1.7982, "step": 2915 }, { "epoch": 0.88, "learning_rate": 0.00016596257351419907, "loss": 1.7792, "step": 2916 }, { "epoch": 0.88, "learning_rate": 0.00016593814463656078, "loss": 1.8482, "step": 2917 }, { "epoch": 0.88, "learning_rate": 0.0001659137087950145, "loss": 1.8213, "step": 2918 }, { "epoch": 0.88, "learning_rate": 0.0001658892659921409, "loss": 1.8373, "step": 2919 }, { "epoch": 0.88, "learning_rate": 0.00016586481623052148, "loss": 1.7653, "step": 2920 }, { "epoch": 0.88, "learning_rate": 0.00016584035951273842, "loss": 1.7714, "step": 2921 }, { "epoch": 0.88, "learning_rate": 0.00016581589584137473, "loss": 1.8031, "step": 2922 }, { "epoch": 0.88, "learning_rate": 0.00016579142521901398, "loss": 1.7981, "step": 2923 }, { "epoch": 0.88, "learning_rate": 0.00016576694764824067, "loss": 1.8528, "step": 2924 }, { "epoch": 0.88, "learning_rate": 0.0001657424631316399, "loss": 1.8525, "step": 2925 }, { "epoch": 0.88, "learning_rate": 0.00016571797167179753, "loss": 1.9269, "step": 2926 }, { "epoch": 0.88, "learning_rate": 0.0001656934732713002, "loss": 1.8662, "step": 2927 }, { "epoch": 0.88, "learning_rate": 0.00016566896793273527, "loss": 1.7843, "step": 2928 }, { "epoch": 0.88, "learning_rate": 0.00016564445565869077, "loss": 1.8615, "step": 2929 }, { "epoch": 0.88, "learning_rate": 0.00016561993645175556, "loss": 1.7241, "step": 2930 }, { "epoch": 0.88, "learning_rate": 0.00016559541031451914, "loss": 1.8653, "step": 2931 }, { "epoch": 0.88, "learning_rate": 0.0001655708772495718, "loss": 1.8737, "step": 2932 }, { "epoch": 0.88, "learning_rate": 0.00016554633725950453, "loss": 1.8932, "step": 2933 }, { "epoch": 0.88, "learning_rate": 0.00016552179034690908, "loss": 1.8934, "step": 2934 }, { "epoch": 0.88, "learning_rate": 0.00016549723651437793, "loss": 1.7068, "step": 2935 }, { "epoch": 0.88, "learning_rate": 0.00016547267576450425, "loss": 1.8126, "step": 2936 }, { "epoch": 0.88, "learning_rate": 0.00016544810809988197, "loss": 1.8411, "step": 2937 }, { "epoch": 0.88, "learning_rate": 0.00016542353352310573, "loss": 1.8671, "step": 2938 }, { "epoch": 0.88, "learning_rate": 0.00016539895203677096, "loss": 1.9036, "step": 2939 }, { "epoch": 0.89, "learning_rate": 0.0001653743636434738, "loss": 1.7933, "step": 2940 }, { "epoch": 0.89, "learning_rate": 0.000165349768345811, "loss": 1.8226, "step": 2941 }, { "epoch": 0.89, "learning_rate": 0.00016532516614638026, "loss": 1.8525, "step": 2942 }, { "epoch": 0.89, "learning_rate": 0.0001653005570477798, "loss": 1.769, "step": 2943 }, { "epoch": 0.89, "learning_rate": 0.0001652759410526086, "loss": 1.8705, "step": 2944 }, { "epoch": 0.89, "learning_rate": 0.0001652513181634666, "loss": 1.8388, "step": 2945 }, { "epoch": 0.89, "learning_rate": 0.00016522668838295416, "loss": 1.8758, "step": 2946 }, { "epoch": 0.89, "learning_rate": 0.00016520205171367252, "loss": 1.7886, "step": 2947 }, { "epoch": 0.89, "learning_rate": 0.00016517740815822366, "loss": 1.8546, "step": 2948 }, { "epoch": 0.89, "learning_rate": 0.0001651527577192102, "loss": 1.8526, "step": 2949 }, { "epoch": 0.89, "learning_rate": 0.00016512810039923554, "loss": 1.8664, "step": 2950 }, { "epoch": 0.89, "learning_rate": 0.0001651034362009039, "loss": 1.8261, "step": 2951 }, { "epoch": 0.89, "learning_rate": 0.00016507876512682004, "loss": 1.8129, "step": 2952 }, { "epoch": 0.89, "learning_rate": 0.00016505408717958957, "loss": 1.8875, "step": 2953 }, { "epoch": 0.89, "learning_rate": 0.0001650294023618188, "loss": 1.8163, "step": 2954 }, { "epoch": 0.89, "learning_rate": 0.00016500471067611473, "loss": 1.8717, "step": 2955 }, { "epoch": 0.89, "learning_rate": 0.00016498001212508515, "loss": 1.9022, "step": 2956 }, { "epoch": 0.89, "learning_rate": 0.00016495530671133853, "loss": 1.8505, "step": 2957 }, { "epoch": 0.89, "learning_rate": 0.0001649305944374841, "loss": 1.8031, "step": 2958 }, { "epoch": 0.89, "learning_rate": 0.00016490587530613177, "loss": 1.8521, "step": 2959 }, { "epoch": 0.89, "learning_rate": 0.0001648811493198922, "loss": 1.9258, "step": 2960 }, { "epoch": 0.89, "learning_rate": 0.00016485641648137677, "loss": 1.7675, "step": 2961 }, { "epoch": 0.89, "learning_rate": 0.00016483167679319754, "loss": 1.7911, "step": 2962 }, { "epoch": 0.89, "learning_rate": 0.00016480693025796737, "loss": 1.9692, "step": 2963 }, { "epoch": 0.89, "learning_rate": 0.00016478217687829986, "loss": 1.7706, "step": 2964 }, { "epoch": 0.89, "learning_rate": 0.00016475741665680923, "loss": 1.9866, "step": 2965 }, { "epoch": 0.89, "learning_rate": 0.00016473264959611047, "loss": 1.9089, "step": 2966 }, { "epoch": 0.89, "learning_rate": 0.0001647078756988193, "loss": 1.8857, "step": 2967 }, { "epoch": 0.89, "learning_rate": 0.00016468309496755216, "loss": 1.8713, "step": 2968 }, { "epoch": 0.89, "learning_rate": 0.00016465830740492623, "loss": 1.7366, "step": 2969 }, { "epoch": 0.89, "learning_rate": 0.0001646335130135594, "loss": 1.8582, "step": 2970 }, { "epoch": 0.89, "learning_rate": 0.00016460871179607023, "loss": 1.8839, "step": 2971 }, { "epoch": 0.89, "learning_rate": 0.00016458390375507806, "loss": 1.8542, "step": 2972 }, { "epoch": 0.89, "learning_rate": 0.00016455908889320296, "loss": 1.8498, "step": 2973 }, { "epoch": 0.9, "learning_rate": 0.00016453426721306567, "loss": 1.8678, "step": 2974 }, { "epoch": 0.9, "learning_rate": 0.00016450943871728767, "loss": 1.884, "step": 2975 }, { "epoch": 0.9, "learning_rate": 0.0001644846034084912, "loss": 1.846, "step": 2976 }, { "epoch": 0.9, "learning_rate": 0.00016445976128929917, "loss": 1.843, "step": 2977 }, { "epoch": 0.9, "learning_rate": 0.0001644349123623352, "loss": 1.8846, "step": 2978 }, { "epoch": 0.9, "learning_rate": 0.0001644100566302237, "loss": 1.8178, "step": 2979 }, { "epoch": 0.9, "learning_rate": 0.0001643851940955897, "loss": 1.8673, "step": 2980 }, { "epoch": 0.9, "learning_rate": 0.00016436032476105906, "loss": 1.8704, "step": 2981 }, { "epoch": 0.9, "learning_rate": 0.00016433544862925822, "loss": 1.7397, "step": 2982 }, { "epoch": 0.9, "learning_rate": 0.00016431056570281446, "loss": 1.9086, "step": 2983 }, { "epoch": 0.9, "learning_rate": 0.00016428567598435575, "loss": 1.8929, "step": 2984 }, { "epoch": 0.9, "learning_rate": 0.00016426077947651074, "loss": 1.7828, "step": 2985 }, { "epoch": 0.9, "learning_rate": 0.00016423587618190886, "loss": 1.824, "step": 2986 }, { "epoch": 0.9, "learning_rate": 0.00016421096610318015, "loss": 1.734, "step": 2987 }, { "epoch": 0.9, "learning_rate": 0.00016418604924295545, "loss": 1.8445, "step": 2988 }, { "epoch": 0.9, "learning_rate": 0.0001641611256038663, "loss": 1.8195, "step": 2989 }, { "epoch": 0.9, "learning_rate": 0.00016413619518854496, "loss": 1.8735, "step": 2990 }, { "epoch": 0.9, "learning_rate": 0.00016411125799962442, "loss": 1.9693, "step": 2991 }, { "epoch": 0.9, "learning_rate": 0.00016408631403973836, "loss": 1.813, "step": 2992 }, { "epoch": 0.9, "learning_rate": 0.00016406136331152115, "loss": 1.8382, "step": 2993 }, { "epoch": 0.9, "learning_rate": 0.0001640364058176079, "loss": 1.9245, "step": 2994 }, { "epoch": 0.9, "learning_rate": 0.0001640114415606345, "loss": 1.8302, "step": 2995 }, { "epoch": 0.9, "learning_rate": 0.0001639864705432374, "loss": 1.7593, "step": 2996 }, { "epoch": 0.9, "learning_rate": 0.00016396149276805397, "loss": 1.9271, "step": 2997 }, { "epoch": 0.9, "learning_rate": 0.0001639365082377221, "loss": 1.9398, "step": 2998 }, { "epoch": 0.9, "learning_rate": 0.00016391151695488047, "loss": 1.8942, "step": 2999 }, { "epoch": 0.9, "learning_rate": 0.00016388651892216855, "loss": 1.8617, "step": 3000 }, { "epoch": 0.9, "learning_rate": 0.0001638615141422264, "loss": 1.8365, "step": 3001 }, { "epoch": 0.9, "learning_rate": 0.0001638365026176948, "loss": 1.8683, "step": 3002 }, { "epoch": 0.9, "learning_rate": 0.0001638114843512154, "loss": 1.9595, "step": 3003 }, { "epoch": 0.9, "learning_rate": 0.0001637864593454303, "loss": 1.9314, "step": 3004 }, { "epoch": 0.9, "learning_rate": 0.0001637614276029826, "loss": 1.8542, "step": 3005 }, { "epoch": 0.9, "learning_rate": 0.00016373638912651591, "loss": 1.9567, "step": 3006 }, { "epoch": 0.91, "learning_rate": 0.0001637113439186746, "loss": 1.8783, "step": 3007 }, { "epoch": 0.91, "learning_rate": 0.0001636862919821038, "loss": 1.8704, "step": 3008 }, { "epoch": 0.91, "learning_rate": 0.00016366123331944927, "loss": 1.8553, "step": 3009 }, { "epoch": 0.91, "learning_rate": 0.00016363616793335753, "loss": 1.8274, "step": 3010 }, { "epoch": 0.91, "learning_rate": 0.00016361109582647588, "loss": 1.8361, "step": 3011 }, { "epoch": 0.91, "learning_rate": 0.00016358601700145213, "loss": 1.8953, "step": 3012 }, { "epoch": 0.91, "learning_rate": 0.000163560931460935, "loss": 1.8251, "step": 3013 }, { "epoch": 0.91, "learning_rate": 0.00016353583920757382, "loss": 1.923, "step": 3014 }, { "epoch": 0.91, "learning_rate": 0.00016351074024401868, "loss": 1.8488, "step": 3015 }, { "epoch": 0.91, "learning_rate": 0.00016348563457292033, "loss": 1.8206, "step": 3016 }, { "epoch": 0.91, "learning_rate": 0.00016346052219693027, "loss": 1.8438, "step": 3017 }, { "epoch": 0.91, "learning_rate": 0.00016343540311870063, "loss": 1.7909, "step": 3018 }, { "epoch": 0.91, "learning_rate": 0.00016341027734088437, "loss": 1.9459, "step": 3019 }, { "epoch": 0.91, "learning_rate": 0.00016338514486613503, "loss": 1.8062, "step": 3020 }, { "epoch": 0.91, "learning_rate": 0.00016336000569710697, "loss": 1.8378, "step": 3021 }, { "epoch": 0.91, "learning_rate": 0.0001633348598364552, "loss": 1.8636, "step": 3022 }, { "epoch": 0.91, "learning_rate": 0.0001633097072868354, "loss": 1.7267, "step": 3023 }, { "epoch": 0.91, "learning_rate": 0.00016328454805090406, "loss": 1.834, "step": 3024 }, { "epoch": 0.91, "learning_rate": 0.00016325938213131828, "loss": 1.9369, "step": 3025 }, { "epoch": 0.91, "learning_rate": 0.00016323420953073592, "loss": 1.8421, "step": 3026 }, { "epoch": 0.91, "learning_rate": 0.0001632090302518155, "loss": 1.842, "step": 3027 }, { "epoch": 0.91, "learning_rate": 0.0001631838442972163, "loss": 1.9383, "step": 3028 }, { "epoch": 0.91, "learning_rate": 0.00016315865166959825, "loss": 1.841, "step": 3029 }, { "epoch": 0.91, "learning_rate": 0.00016313345237162205, "loss": 1.9085, "step": 3030 }, { "epoch": 0.91, "learning_rate": 0.00016310824640594903, "loss": 2.8968, "step": 3031 }, { "epoch": 0.91, "learning_rate": 0.00016308303377524127, "loss": 3.1491, "step": 3032 }, { "epoch": 0.91, "learning_rate": 0.00016305781448216157, "loss": 2.4499, "step": 3033 }, { "epoch": 0.91, "learning_rate": 0.0001630325885293734, "loss": 2.07, "step": 3034 }, { "epoch": 0.91, "learning_rate": 0.00016300735591954092, "loss": 2.1461, "step": 3035 }, { "epoch": 0.91, "learning_rate": 0.00016298211665532902, "loss": 2.054, "step": 3036 }, { "epoch": 0.91, "learning_rate": 0.0001629568707394033, "loss": 1.9834, "step": 3037 }, { "epoch": 0.91, "learning_rate": 0.00016293161817443008, "loss": 1.9773, "step": 3038 }, { "epoch": 0.91, "learning_rate": 0.00016290635896307627, "loss": 1.999, "step": 3039 }, { "epoch": 0.92, "learning_rate": 0.00016288109310800964, "loss": 1.8701, "step": 3040 }, { "epoch": 0.92, "learning_rate": 0.00016285582061189855, "loss": 1.8197, "step": 3041 }, { "epoch": 0.92, "learning_rate": 0.0001628305414774121, "loss": 1.8635, "step": 3042 }, { "epoch": 0.92, "learning_rate": 0.00016280525570722013, "loss": 1.8379, "step": 3043 }, { "epoch": 0.92, "learning_rate": 0.00016277996330399307, "loss": 1.9064, "step": 3044 }, { "epoch": 0.92, "learning_rate": 0.00016275466427040215, "loss": 1.8504, "step": 3045 }, { "epoch": 0.92, "learning_rate": 0.00016272935860911928, "loss": 1.82, "step": 3046 }, { "epoch": 0.92, "learning_rate": 0.00016270404632281706, "loss": 1.8039, "step": 3047 }, { "epoch": 0.92, "learning_rate": 0.00016267872741416878, "loss": 1.8207, "step": 3048 }, { "epoch": 0.92, "learning_rate": 0.00016265340188584846, "loss": 1.838, "step": 3049 }, { "epoch": 0.92, "learning_rate": 0.00016262806974053076, "loss": 1.8541, "step": 3050 }, { "epoch": 0.92, "learning_rate": 0.0001626027309808911, "loss": 1.8691, "step": 3051 }, { "epoch": 0.92, "learning_rate": 0.0001625773856096056, "loss": 1.868, "step": 3052 }, { "epoch": 0.92, "learning_rate": 0.00016255203362935104, "loss": 1.8087, "step": 3053 }, { "epoch": 0.92, "learning_rate": 0.00016252667504280488, "loss": 1.9953, "step": 3054 }, { "epoch": 0.92, "learning_rate": 0.00016250130985264534, "loss": 1.8257, "step": 3055 }, { "epoch": 0.92, "learning_rate": 0.0001624759380615513, "loss": 1.9155, "step": 3056 }, { "epoch": 0.92, "learning_rate": 0.00016245055967220232, "loss": 1.8168, "step": 3057 }, { "epoch": 0.92, "learning_rate": 0.00016242517468727873, "loss": 1.855, "step": 3058 }, { "epoch": 0.92, "learning_rate": 0.00016239978310946147, "loss": 1.8527, "step": 3059 }, { "epoch": 0.92, "learning_rate": 0.00016237438494143223, "loss": 1.7803, "step": 3060 }, { "epoch": 0.92, "learning_rate": 0.00016234898018587337, "loss": 1.7315, "step": 3061 }, { "epoch": 0.92, "learning_rate": 0.000162323568845468, "loss": 1.7573, "step": 3062 }, { "epoch": 0.92, "learning_rate": 0.00016229815092289976, "loss": 1.8198, "step": 3063 }, { "epoch": 0.92, "learning_rate": 0.00016227272642085325, "loss": 1.9307, "step": 3064 }, { "epoch": 0.92, "learning_rate": 0.00016224729534201356, "loss": 1.9341, "step": 3065 }, { "epoch": 0.92, "learning_rate": 0.00016222185768906652, "loss": 1.8631, "step": 3066 }, { "epoch": 0.92, "learning_rate": 0.00016219641346469868, "loss": 1.8752, "step": 3067 }, { "epoch": 0.92, "learning_rate": 0.0001621709626715973, "loss": 1.8946, "step": 3068 }, { "epoch": 0.92, "learning_rate": 0.00016214550531245026, "loss": 1.8676, "step": 3069 }, { "epoch": 0.92, "learning_rate": 0.00016212004138994627, "loss": 1.8741, "step": 3070 }, { "epoch": 0.92, "learning_rate": 0.00016209457090677454, "loss": 1.8071, "step": 3071 }, { "epoch": 0.92, "learning_rate": 0.00016206909386562513, "loss": 1.8379, "step": 3072 }, { "epoch": 0.93, "learning_rate": 0.00016204361026918876, "loss": 1.8513, "step": 3073 }, { "epoch": 0.93, "learning_rate": 0.00016201812012015676, "loss": 1.8786, "step": 3074 }, { "epoch": 0.93, "learning_rate": 0.00016199262342122127, "loss": 1.9399, "step": 3075 }, { "epoch": 0.93, "learning_rate": 0.00016196712017507503, "loss": 1.9384, "step": 3076 }, { "epoch": 0.93, "learning_rate": 0.00016194161038441153, "loss": 1.8298, "step": 3077 }, { "epoch": 0.93, "learning_rate": 0.00016191609405192493, "loss": 1.903, "step": 3078 }, { "epoch": 0.93, "learning_rate": 0.00016189057118031006, "loss": 1.8409, "step": 3079 }, { "epoch": 0.93, "learning_rate": 0.00016186504177226248, "loss": 1.8886, "step": 3080 }, { "epoch": 0.93, "learning_rate": 0.0001618395058304784, "loss": 1.8451, "step": 3081 }, { "epoch": 0.93, "learning_rate": 0.00016181396335765476, "loss": 1.8945, "step": 3082 }, { "epoch": 0.93, "learning_rate": 0.00016178841435648917, "loss": 1.8568, "step": 3083 }, { "epoch": 0.93, "learning_rate": 0.00016176285882967995, "loss": 1.7709, "step": 3084 }, { "epoch": 0.93, "learning_rate": 0.00016173729677992603, "loss": 1.8838, "step": 3085 }, { "epoch": 0.93, "learning_rate": 0.00016171172820992713, "loss": 1.8039, "step": 3086 }, { "epoch": 0.93, "learning_rate": 0.0001616861531223836, "loss": 1.8102, "step": 3087 }, { "epoch": 0.93, "learning_rate": 0.00016166057151999652, "loss": 1.7633, "step": 3088 }, { "epoch": 0.93, "learning_rate": 0.00016163498340546762, "loss": 1.8612, "step": 3089 }, { "epoch": 0.93, "learning_rate": 0.0001616093887814993, "loss": 1.8331, "step": 3090 }, { "epoch": 0.93, "learning_rate": 0.0001615837876507947, "loss": 1.7772, "step": 3091 }, { "epoch": 0.93, "learning_rate": 0.00016155818001605766, "loss": 1.7917, "step": 3092 }, { "epoch": 0.93, "learning_rate": 0.00016153256587999267, "loss": 1.8021, "step": 3093 }, { "epoch": 0.93, "learning_rate": 0.00016150694524530483, "loss": 1.8717, "step": 3094 }, { "epoch": 0.93, "learning_rate": 0.00016148131811470008, "loss": 1.9218, "step": 3095 }, { "epoch": 0.93, "learning_rate": 0.00016145568449088494, "loss": 1.9628, "step": 3096 }, { "epoch": 0.93, "learning_rate": 0.0001614300443765667, "loss": 1.8091, "step": 3097 }, { "epoch": 0.93, "learning_rate": 0.0001614043977744532, "loss": 1.7975, "step": 3098 }, { "epoch": 0.93, "learning_rate": 0.00016137874468725308, "loss": 1.7598, "step": 3099 }, { "epoch": 0.93, "learning_rate": 0.00016135308511767564, "loss": 1.7876, "step": 3100 }, { "epoch": 0.93, "learning_rate": 0.00016132741906843087, "loss": 1.8526, "step": 3101 }, { "epoch": 0.93, "learning_rate": 0.00016130174654222944, "loss": 1.8599, "step": 3102 }, { "epoch": 0.93, "learning_rate": 0.00016127606754178265, "loss": 1.7946, "step": 3103 }, { "epoch": 0.93, "learning_rate": 0.00016125038206980254, "loss": 1.8911, "step": 3104 }, { "epoch": 0.93, "learning_rate": 0.00016122469012900187, "loss": 1.8772, "step": 3105 }, { "epoch": 0.93, "learning_rate": 0.000161198991722094, "loss": 1.9037, "step": 3106 }, { "epoch": 0.94, "learning_rate": 0.000161173286851793, "loss": 1.834, "step": 3107 }, { "epoch": 0.94, "learning_rate": 0.00016114757552081364, "loss": 1.8003, "step": 3108 }, { "epoch": 0.94, "learning_rate": 0.0001611218577318714, "loss": 1.7144, "step": 3109 }, { "epoch": 0.94, "learning_rate": 0.0001610961334876823, "loss": 1.8994, "step": 3110 }, { "epoch": 0.94, "learning_rate": 0.00016107040279096332, "loss": 1.8208, "step": 3111 }, { "epoch": 0.94, "learning_rate": 0.0001610446656444318, "loss": 1.8086, "step": 3112 }, { "epoch": 0.94, "learning_rate": 0.00016101892205080596, "loss": 1.8981, "step": 3113 }, { "epoch": 0.94, "learning_rate": 0.00016099317201280467, "loss": 1.8085, "step": 3114 }, { "epoch": 0.94, "learning_rate": 0.00016096741553314745, "loss": 2.0625, "step": 3115 }, { "epoch": 0.94, "learning_rate": 0.00016094165261455447, "loss": 1.7614, "step": 3116 }, { "epoch": 0.94, "learning_rate": 0.0001609158832597467, "loss": 1.8602, "step": 3117 }, { "epoch": 0.94, "learning_rate": 0.00016089010747144568, "loss": 1.901, "step": 3118 }, { "epoch": 0.94, "learning_rate": 0.00016086432525237363, "loss": 1.8124, "step": 3119 }, { "epoch": 0.94, "learning_rate": 0.0001608385366052535, "loss": 1.8449, "step": 3120 }, { "epoch": 0.94, "learning_rate": 0.00016081274153280893, "loss": 1.87, "step": 3121 }, { "epoch": 0.94, "learning_rate": 0.00016078694003776418, "loss": 1.9838, "step": 3122 }, { "epoch": 0.94, "learning_rate": 0.0001607611321228442, "loss": 1.9617, "step": 3123 }, { "epoch": 0.94, "learning_rate": 0.00016073531779077466, "loss": 1.8764, "step": 3124 }, { "epoch": 0.94, "learning_rate": 0.0001607094970442819, "loss": 1.8743, "step": 3125 }, { "epoch": 0.94, "learning_rate": 0.0001606836698860929, "loss": 1.8726, "step": 3126 }, { "epoch": 0.94, "learning_rate": 0.0001606578363189353, "loss": 1.858, "step": 3127 }, { "epoch": 0.94, "learning_rate": 0.00016063199634553746, "loss": 1.8699, "step": 3128 }, { "epoch": 0.94, "learning_rate": 0.0001606061499686285, "loss": 1.9403, "step": 3129 }, { "epoch": 0.94, "learning_rate": 0.00016058029719093805, "loss": 1.8026, "step": 3130 }, { "epoch": 0.94, "learning_rate": 0.00016055443801519652, "loss": 1.7967, "step": 3131 }, { "epoch": 0.94, "learning_rate": 0.00016052857244413493, "loss": 1.9022, "step": 3132 }, { "epoch": 0.94, "learning_rate": 0.00016050270048048506, "loss": 1.8369, "step": 3133 }, { "epoch": 0.94, "learning_rate": 0.00016047682212697926, "loss": 1.8547, "step": 3134 }, { "epoch": 0.94, "learning_rate": 0.00016045093738635067, "loss": 1.8584, "step": 3135 }, { "epoch": 0.94, "learning_rate": 0.00016042504626133304, "loss": 1.8192, "step": 3136 }, { "epoch": 0.94, "learning_rate": 0.00016039914875466078, "loss": 1.7805, "step": 3137 }, { "epoch": 0.94, "learning_rate": 0.00016037324486906904, "loss": 1.6941, "step": 3138 }, { "epoch": 0.94, "learning_rate": 0.00016034733460729353, "loss": 1.9114, "step": 3139 }, { "epoch": 0.95, "learning_rate": 0.00016032141797207077, "loss": 1.8479, "step": 3140 }, { "epoch": 0.95, "learning_rate": 0.00016029549496613784, "loss": 1.8148, "step": 3141 }, { "epoch": 0.95, "learning_rate": 0.00016026956559223257, "loss": 1.7617, "step": 3142 }, { "epoch": 0.95, "learning_rate": 0.00016024362985309342, "loss": 1.932, "step": 3143 }, { "epoch": 0.95, "learning_rate": 0.00016021768775145956, "loss": 1.7471, "step": 3144 }, { "epoch": 0.95, "learning_rate": 0.00016019173929007074, "loss": 1.8547, "step": 3145 }, { "epoch": 0.95, "learning_rate": 0.0001601657844716675, "loss": 1.829, "step": 3146 }, { "epoch": 0.95, "learning_rate": 0.00016013982329899103, "loss": 1.8171, "step": 3147 }, { "epoch": 0.95, "learning_rate": 0.00016011385577478307, "loss": 1.8303, "step": 3148 }, { "epoch": 0.95, "learning_rate": 0.00016008788190178622, "loss": 1.7295, "step": 3149 }, { "epoch": 0.95, "learning_rate": 0.0001600619016827436, "loss": 1.8936, "step": 3150 }, { "epoch": 0.95, "learning_rate": 0.00016003591512039905, "loss": 1.7824, "step": 3151 }, { "epoch": 0.95, "learning_rate": 0.00016000992221749705, "loss": 1.759, "step": 3152 }, { "epoch": 0.95, "learning_rate": 0.00015998392297678283, "loss": 1.8591, "step": 3153 }, { "epoch": 0.95, "learning_rate": 0.00015995791740100227, "loss": 1.8453, "step": 3154 }, { "epoch": 0.95, "learning_rate": 0.00015993190549290183, "loss": 1.7539, "step": 3155 }, { "epoch": 0.95, "learning_rate": 0.00015990588725522871, "loss": 1.7377, "step": 3156 }, { "epoch": 0.95, "learning_rate": 0.0001598798626907308, "loss": 1.8033, "step": 3157 }, { "epoch": 0.95, "learning_rate": 0.00015985383180215656, "loss": 1.8121, "step": 3158 }, { "epoch": 0.95, "learning_rate": 0.00015982779459225525, "loss": 1.9331, "step": 3159 }, { "epoch": 0.95, "learning_rate": 0.00015980175106377673, "loss": 1.8591, "step": 3160 }, { "epoch": 0.95, "learning_rate": 0.00015977570121947148, "loss": 1.8019, "step": 3161 }, { "epoch": 0.95, "learning_rate": 0.00015974964506209073, "loss": 1.8254, "step": 3162 }, { "epoch": 0.95, "learning_rate": 0.00015972358259438632, "loss": 1.8315, "step": 3163 }, { "epoch": 0.95, "learning_rate": 0.0001596975138191108, "loss": 1.7821, "step": 3164 }, { "epoch": 0.95, "learning_rate": 0.00015967143873901738, "loss": 1.8343, "step": 3165 }, { "epoch": 0.95, "learning_rate": 0.00015964535735685985, "loss": 1.8703, "step": 3166 }, { "epoch": 0.95, "learning_rate": 0.00015961926967539283, "loss": 1.8825, "step": 3167 }, { "epoch": 0.95, "learning_rate": 0.00015959317569737142, "loss": 1.7986, "step": 3168 }, { "epoch": 0.95, "learning_rate": 0.00015956707542555157, "loss": 1.7168, "step": 3169 }, { "epoch": 0.95, "learning_rate": 0.0001595409688626897, "loss": 1.8006, "step": 3170 }, { "epoch": 0.95, "learning_rate": 0.0001595148560115431, "loss": 1.8673, "step": 3171 }, { "epoch": 0.95, "learning_rate": 0.00015948873687486953, "loss": 1.878, "step": 3172 }, { "epoch": 0.96, "learning_rate": 0.00015946261145542755, "loss": 1.9493, "step": 3173 }, { "epoch": 0.96, "learning_rate": 0.00015943647975597634, "loss": 1.7972, "step": 3174 }, { "epoch": 0.96, "learning_rate": 0.00015941034177927575, "loss": 1.7892, "step": 3175 }, { "epoch": 0.96, "learning_rate": 0.00015938419752808622, "loss": 1.7731, "step": 3176 }, { "epoch": 0.96, "learning_rate": 0.00015935804700516895, "loss": 1.8303, "step": 3177 }, { "epoch": 0.96, "learning_rate": 0.00015933189021328581, "loss": 1.9018, "step": 3178 }, { "epoch": 0.96, "learning_rate": 0.00015930572715519925, "loss": 1.8779, "step": 3179 }, { "epoch": 0.96, "learning_rate": 0.00015927955783367242, "loss": 1.9139, "step": 3180 }, { "epoch": 0.96, "learning_rate": 0.00015925338225146915, "loss": 1.7629, "step": 3181 }, { "epoch": 0.96, "learning_rate": 0.0001592272004113539, "loss": 1.7661, "step": 3182 }, { "epoch": 0.96, "learning_rate": 0.00015920101231609181, "loss": 1.7799, "step": 3183 }, { "epoch": 0.96, "learning_rate": 0.0001591748179684487, "loss": 1.925, "step": 3184 }, { "epoch": 0.96, "learning_rate": 0.000159148617371191, "loss": 1.8168, "step": 3185 }, { "epoch": 0.96, "learning_rate": 0.00015912241052708584, "loss": 1.86, "step": 3186 }, { "epoch": 0.96, "learning_rate": 0.00015909619743890096, "loss": 1.8263, "step": 3187 }, { "epoch": 0.96, "learning_rate": 0.00015906997810940485, "loss": 1.7995, "step": 3188 }, { "epoch": 0.96, "learning_rate": 0.00015904375254136656, "loss": 1.796, "step": 3189 }, { "epoch": 0.96, "learning_rate": 0.0001590175207375559, "loss": 1.8447, "step": 3190 }, { "epoch": 0.96, "learning_rate": 0.00015899128270074322, "loss": 1.8013, "step": 3191 }, { "epoch": 0.96, "learning_rate": 0.00015896503843369964, "loss": 1.7636, "step": 3192 }, { "epoch": 0.96, "learning_rate": 0.00015893878793919684, "loss": 1.9009, "step": 3193 }, { "epoch": 0.96, "learning_rate": 0.00015891253122000723, "loss": 1.8458, "step": 3194 }, { "epoch": 0.96, "learning_rate": 0.0001588862682789039, "loss": 1.7725, "step": 3195 }, { "epoch": 0.96, "learning_rate": 0.00015885999911866046, "loss": 1.8098, "step": 3196 }, { "epoch": 0.96, "learning_rate": 0.00015883372374205136, "loss": 1.8515, "step": 3197 }, { "epoch": 0.96, "learning_rate": 0.00015880744215185158, "loss": 1.83, "step": 3198 }, { "epoch": 0.96, "learning_rate": 0.00015878115435083676, "loss": 1.8116, "step": 3199 }, { "epoch": 0.96, "learning_rate": 0.0001587548603417832, "loss": 1.9147, "step": 3200 }, { "epoch": 0.96, "learning_rate": 0.00015872856012746802, "loss": 1.712, "step": 3201 }, { "epoch": 0.96, "learning_rate": 0.00015870225371066872, "loss": 1.8236, "step": 3202 }, { "epoch": 0.96, "learning_rate": 0.00015867594109416367, "loss": 1.8347, "step": 3203 }, { "epoch": 0.96, "learning_rate": 0.0001586496222807318, "loss": 1.8698, "step": 3204 }, { "epoch": 0.96, "learning_rate": 0.00015862329727315264, "loss": 1.8243, "step": 3205 }, { "epoch": 0.97, "learning_rate": 0.00015859696607420655, "loss": 1.8205, "step": 3206 }, { "epoch": 0.97, "learning_rate": 0.00015857062868667438, "loss": 1.8146, "step": 3207 }, { "epoch": 0.97, "learning_rate": 0.00015854428511333772, "loss": 1.8469, "step": 3208 }, { "epoch": 0.97, "learning_rate": 0.00015851793535697877, "loss": 1.9338, "step": 3209 }, { "epoch": 0.97, "learning_rate": 0.00015849157942038046, "loss": 1.7976, "step": 3210 }, { "epoch": 0.97, "learning_rate": 0.0001584652173063262, "loss": 1.8015, "step": 3211 }, { "epoch": 0.97, "learning_rate": 0.00015843884901760023, "loss": 1.8125, "step": 3212 }, { "epoch": 0.97, "learning_rate": 0.0001584124745569874, "loss": 1.7159, "step": 3213 }, { "epoch": 0.97, "learning_rate": 0.00015838609392727316, "loss": 1.9182, "step": 3214 }, { "epoch": 0.97, "learning_rate": 0.00015835970713124365, "loss": 1.8131, "step": 3215 }, { "epoch": 0.97, "learning_rate": 0.00015833331417168563, "loss": 1.7679, "step": 3216 }, { "epoch": 0.97, "learning_rate": 0.0001583069150513865, "loss": 1.7809, "step": 3217 }, { "epoch": 0.97, "learning_rate": 0.00015828050977313444, "loss": 1.6786, "step": 3218 }, { "epoch": 0.97, "learning_rate": 0.00015825409833971813, "loss": 1.9172, "step": 3219 }, { "epoch": 0.97, "learning_rate": 0.00015822768075392699, "loss": 1.9544, "step": 3220 }, { "epoch": 0.97, "learning_rate": 0.00015820125701855099, "loss": 1.8707, "step": 3221 }, { "epoch": 0.97, "learning_rate": 0.00015817482713638087, "loss": 1.8973, "step": 3222 }, { "epoch": 0.97, "learning_rate": 0.00015814839111020788, "loss": 1.9525, "step": 3223 }, { "epoch": 0.97, "learning_rate": 0.00015812194894282409, "loss": 1.8068, "step": 3224 }, { "epoch": 0.97, "learning_rate": 0.00015809550063702207, "loss": 1.8024, "step": 3225 }, { "epoch": 0.97, "learning_rate": 0.0001580690461955952, "loss": 1.8969, "step": 3226 }, { "epoch": 0.97, "learning_rate": 0.00015804258562133726, "loss": 1.9031, "step": 3227 }, { "epoch": 0.97, "learning_rate": 0.00015801611891704293, "loss": 1.8453, "step": 3228 }, { "epoch": 0.97, "learning_rate": 0.0001579896460855074, "loss": 1.8036, "step": 3229 }, { "epoch": 0.97, "learning_rate": 0.00015796316712952653, "loss": 1.7804, "step": 3230 }, { "epoch": 0.97, "learning_rate": 0.00015793668205189683, "loss": 1.8105, "step": 3231 }, { "epoch": 0.97, "learning_rate": 0.00015791019085541547, "loss": 1.7492, "step": 3232 }, { "epoch": 0.97, "learning_rate": 0.00015788369354288026, "loss": 1.9959, "step": 3233 }, { "epoch": 0.97, "learning_rate": 0.00015785719011708965, "loss": 1.8212, "step": 3234 }, { "epoch": 0.97, "learning_rate": 0.00015783068058084277, "loss": 1.8369, "step": 3235 }, { "epoch": 0.97, "learning_rate": 0.0001578041649369393, "loss": 1.8548, "step": 3236 }, { "epoch": 0.97, "learning_rate": 0.0001577776431881797, "loss": 1.8631, "step": 3237 }, { "epoch": 0.97, "learning_rate": 0.00015775111533736493, "loss": 1.8775, "step": 3238 }, { "epoch": 0.98, "learning_rate": 0.00015772458138729677, "loss": 1.8908, "step": 3239 }, { "epoch": 0.98, "learning_rate": 0.00015769804134077743, "loss": 1.8509, "step": 3240 }, { "epoch": 0.98, "learning_rate": 0.00015767149520060994, "loss": 1.8266, "step": 3241 }, { "epoch": 0.98, "learning_rate": 0.00015764494296959794, "loss": 1.8534, "step": 3242 }, { "epoch": 0.98, "learning_rate": 0.0001576183846505456, "loss": 1.8106, "step": 3243 }, { "epoch": 0.98, "learning_rate": 0.00015759182024625788, "loss": 1.9062, "step": 3244 }, { "epoch": 0.98, "learning_rate": 0.0001575652497595403, "loss": 1.8658, "step": 3245 }, { "epoch": 0.98, "learning_rate": 0.00015753867319319906, "loss": 1.8343, "step": 3246 }, { "epoch": 0.98, "learning_rate": 0.00015751209055004094, "loss": 1.8493, "step": 3247 }, { "epoch": 0.98, "learning_rate": 0.00015748550183287345, "loss": 1.7962, "step": 3248 }, { "epoch": 0.98, "learning_rate": 0.0001574589070445047, "loss": 1.7207, "step": 3249 }, { "epoch": 0.98, "learning_rate": 0.00015743230618774337, "loss": 1.8516, "step": 3250 }, { "epoch": 0.98, "learning_rate": 0.00015740569926539895, "loss": 1.7938, "step": 3251 }, { "epoch": 0.98, "learning_rate": 0.00015737908628028134, "loss": 1.7974, "step": 3252 }, { "epoch": 0.98, "learning_rate": 0.00015735246723520136, "loss": 1.9629, "step": 3253 }, { "epoch": 0.98, "learning_rate": 0.00015732584213297023, "loss": 1.7874, "step": 3254 }, { "epoch": 0.98, "learning_rate": 0.0001572992109763999, "loss": 1.8434, "step": 3255 }, { "epoch": 0.98, "learning_rate": 0.000157272573768303, "loss": 1.9609, "step": 3256 }, { "epoch": 0.98, "learning_rate": 0.00015724593051149276, "loss": 1.8293, "step": 3257 }, { "epoch": 0.98, "learning_rate": 0.00015721928120878299, "loss": 1.8447, "step": 3258 }, { "epoch": 0.98, "learning_rate": 0.0001571926258629882, "loss": 1.8179, "step": 3259 }, { "epoch": 0.98, "learning_rate": 0.0001571659644769236, "loss": 1.9468, "step": 3260 }, { "epoch": 0.98, "learning_rate": 0.00015713929705340492, "loss": 1.8676, "step": 3261 }, { "epoch": 0.98, "learning_rate": 0.0001571126235952486, "loss": 1.8514, "step": 3262 }, { "epoch": 0.98, "learning_rate": 0.00015708594410527172, "loss": 1.7872, "step": 3263 }, { "epoch": 0.98, "learning_rate": 0.00015705925858629192, "loss": 1.8361, "step": 3264 }, { "epoch": 0.98, "learning_rate": 0.00015703256704112755, "loss": 1.842, "step": 3265 }, { "epoch": 0.98, "learning_rate": 0.0001570058694725976, "loss": 1.835, "step": 3266 }, { "epoch": 0.98, "learning_rate": 0.00015697916588352167, "loss": 1.8448, "step": 3267 }, { "epoch": 0.98, "learning_rate": 0.00015695245627671996, "loss": 1.715, "step": 3268 }, { "epoch": 0.98, "learning_rate": 0.00015692574065501337, "loss": 1.8065, "step": 3269 }, { "epoch": 0.98, "learning_rate": 0.00015689901902122343, "loss": 1.9418, "step": 3270 }, { "epoch": 0.98, "learning_rate": 0.00015687229137817225, "loss": 1.8295, "step": 3271 }, { "epoch": 0.98, "learning_rate": 0.00015684555772868265, "loss": 1.9485, "step": 3272 }, { "epoch": 0.99, "learning_rate": 0.00015681881807557798, "loss": 1.8835, "step": 3273 }, { "epoch": 0.99, "learning_rate": 0.00015679207242168234, "loss": 1.7604, "step": 3274 }, { "epoch": 0.99, "learning_rate": 0.00015676532076982044, "loss": 1.8607, "step": 3275 }, { "epoch": 0.99, "learning_rate": 0.00015673856312281752, "loss": 1.862, "step": 3276 }, { "epoch": 0.99, "learning_rate": 0.00015671179948349956, "loss": 1.7943, "step": 3277 }, { "epoch": 0.99, "learning_rate": 0.00015668502985469312, "loss": 1.8591, "step": 3278 }, { "epoch": 0.99, "learning_rate": 0.00015665825423922551, "loss": 1.8345, "step": 3279 }, { "epoch": 0.99, "learning_rate": 0.00015663147263992447, "loss": 1.9048, "step": 3280 }, { "epoch": 0.99, "learning_rate": 0.00015660468505961848, "loss": 1.884, "step": 3281 }, { "epoch": 0.99, "learning_rate": 0.0001565778915011367, "loss": 1.82, "step": 3282 }, { "epoch": 0.99, "learning_rate": 0.00015655109196730885, "loss": 1.8433, "step": 3283 }, { "epoch": 0.99, "learning_rate": 0.00015652428646096534, "loss": 1.8952, "step": 3284 }, { "epoch": 0.99, "learning_rate": 0.0001564974749849371, "loss": 1.8013, "step": 3285 }, { "epoch": 0.99, "learning_rate": 0.00015647065754205582, "loss": 1.7636, "step": 3286 }, { "epoch": 0.99, "learning_rate": 0.00015644383413515372, "loss": 1.8214, "step": 3287 }, { "epoch": 0.99, "learning_rate": 0.00015641700476706375, "loss": 1.8826, "step": 3288 }, { "epoch": 0.99, "learning_rate": 0.0001563901694406194, "loss": 1.8531, "step": 3289 }, { "epoch": 0.99, "learning_rate": 0.0001563633281586548, "loss": 1.8874, "step": 3290 }, { "epoch": 0.99, "learning_rate": 0.00015633648092400478, "loss": 1.7502, "step": 3291 }, { "epoch": 0.99, "learning_rate": 0.00015630962773950473, "loss": 1.918, "step": 3292 }, { "epoch": 0.99, "learning_rate": 0.00015628276860799067, "loss": 1.9092, "step": 3293 }, { "epoch": 0.99, "learning_rate": 0.00015625590353229927, "loss": 1.8027, "step": 3294 }, { "epoch": 0.99, "learning_rate": 0.00015622903251526782, "loss": 1.8291, "step": 3295 }, { "epoch": 0.99, "learning_rate": 0.00015620215555973426, "loss": 1.8281, "step": 3296 }, { "epoch": 0.99, "learning_rate": 0.00015617527266853718, "loss": 1.803, "step": 3297 }, { "epoch": 0.99, "learning_rate": 0.00015614838384451568, "loss": 1.9138, "step": 3298 }, { "epoch": 0.99, "learning_rate": 0.0001561214890905096, "loss": 1.9663, "step": 3299 }, { "epoch": 0.99, "learning_rate": 0.00015609458840935936, "loss": 1.7962, "step": 3300 }, { "epoch": 0.99, "learning_rate": 0.000156067681803906, "loss": 1.8196, "step": 3301 }, { "epoch": 0.99, "learning_rate": 0.00015604076927699123, "loss": 1.8642, "step": 3302 }, { "epoch": 0.99, "learning_rate": 0.00015601385083145733, "loss": 1.8802, "step": 3303 }, { "epoch": 0.99, "learning_rate": 0.00015598692647014726, "loss": 1.7582, "step": 3304 }, { "epoch": 0.99, "learning_rate": 0.00015595999619590455, "loss": 1.7819, "step": 3305 }, { "epoch": 1.0, "learning_rate": 0.00015593306001157343, "loss": 1.9399, "step": 3306 }, { "epoch": 1.0, "learning_rate": 0.0001559061179199986, "loss": 1.8546, "step": 3307 }, { "epoch": 1.0, "learning_rate": 0.00015587916992402555, "loss": 1.9343, "step": 3308 }, { "epoch": 1.0, "learning_rate": 0.00015585221602650034, "loss": 1.829, "step": 3309 }, { "epoch": 1.0, "learning_rate": 0.0001558252562302697, "loss": 1.8764, "step": 3310 }, { "epoch": 1.0, "learning_rate": 0.00015579829053818077, "loss": 1.9167, "step": 3311 }, { "epoch": 1.0, "learning_rate": 0.00015577131895308163, "loss": 1.753, "step": 3312 }, { "epoch": 1.0, "learning_rate": 0.00015574434147782074, "loss": 1.8377, "step": 3313 }, { "epoch": 1.0, "learning_rate": 0.0001557173581152473, "loss": 1.8894, "step": 3314 }, { "epoch": 1.0, "learning_rate": 0.00015569036886821106, "loss": 1.8445, "step": 3315 }, { "epoch": 1.0, "learning_rate": 0.00015566337373956247, "loss": 1.9463, "step": 3316 }, { "epoch": 1.0, "learning_rate": 0.0001556363727321525, "loss": 1.7762, "step": 3317 }, { "epoch": 1.0, "learning_rate": 0.00015560936584883288, "loss": 1.8327, "step": 3318 }, { "epoch": 1.0, "learning_rate": 0.0001555823530924558, "loss": 1.867, "step": 3319 }, { "epoch": 1.0, "learning_rate": 0.00015555533446587422, "loss": 1.9325, "step": 3320 }, { "epoch": 1.0, "learning_rate": 0.0001555283099719416, "loss": 1.8447, "step": 3321 }, { "epoch": 1.0, "learning_rate": 0.00015550127961351211, "loss": 1.7588, "step": 3322 }, { "epoch": 1.0, "learning_rate": 0.0001554742433934405, "loss": 1.8444, "step": 3323 }, { "epoch": 1.0, "learning_rate": 0.0001554472013145821, "loss": 1.8347, "step": 3324 }, { "epoch": 1.0, "learning_rate": 0.0001554201533797929, "loss": 1.8447, "step": 3325 }, { "epoch": 1.0, "learning_rate": 0.00015539309959192955, "loss": 1.8854, "step": 3326 }, { "epoch": 1.0, "learning_rate": 0.00015536603995384927, "loss": 1.9015, "step": 3327 }, { "epoch": 1.0, "learning_rate": 0.00015533897446840986, "loss": 2.4767, "step": 3328 }, { "epoch": 1.0, "learning_rate": 0.00015531190313846982, "loss": 2.0816, "step": 3329 }, { "epoch": 1.0, "learning_rate": 0.00015528482596688815, "loss": 2.0571, "step": 3330 }, { "epoch": 1.0, "learning_rate": 0.00015525774295652468, "loss": 1.9688, "step": 3331 }, { "epoch": 1.0, "learning_rate": 0.0001552306541102396, "loss": 2.0136, "step": 3332 }, { "epoch": 1.0, "learning_rate": 0.00015520355943089391, "loss": 2.0483, "step": 3333 }, { "epoch": 1.0, "learning_rate": 0.00015517645892134914, "loss": 2.0212, "step": 3334 }, { "epoch": 1.0, "learning_rate": 0.00015514935258446742, "loss": 1.9438, "step": 3335 }, { "epoch": 1.0, "learning_rate": 0.0001551222404231115, "loss": 2.0658, "step": 3336 }, { "epoch": 1.0, "learning_rate": 0.00015509512244014486, "loss": 2.0298, "step": 3337 }, { "epoch": 1.0, "learning_rate": 0.00015506799863843144, "loss": 2.2236, "step": 3338 }, { "epoch": 1.01, "learning_rate": 0.00015504086902083588, "loss": 1.993, "step": 3339 }, { "epoch": 1.01, "learning_rate": 0.00015501373359022339, "loss": 2.1309, "step": 3340 }, { "epoch": 1.01, "learning_rate": 0.00015498659234945984, "loss": 2.0347, "step": 3341 }, { "epoch": 1.01, "learning_rate": 0.0001549594453014117, "loss": 1.9855, "step": 3342 }, { "epoch": 1.01, "learning_rate": 0.00015493229244894601, "loss": 2.2996, "step": 3343 }, { "epoch": 1.01, "learning_rate": 0.00015490513379493048, "loss": 2.1426, "step": 3344 }, { "epoch": 1.01, "learning_rate": 0.0001548779693422334, "loss": 1.9112, "step": 3345 }, { "epoch": 1.01, "learning_rate": 0.0001548507990937237, "loss": 1.9312, "step": 3346 }, { "epoch": 1.01, "learning_rate": 0.00015482362305227086, "loss": 2.0358, "step": 3347 }, { "epoch": 1.01, "learning_rate": 0.0001547964412207451, "loss": 2.0286, "step": 3348 }, { "epoch": 1.01, "learning_rate": 0.0001547692536020171, "loss": 1.8382, "step": 3349 }, { "epoch": 1.01, "learning_rate": 0.0001547420601989582, "loss": 1.8975, "step": 3350 }, { "epoch": 1.01, "learning_rate": 0.00015471486101444043, "loss": 1.9998, "step": 3351 }, { "epoch": 1.01, "learning_rate": 0.00015468765605133636, "loss": 1.9937, "step": 3352 }, { "epoch": 1.01, "learning_rate": 0.00015466044531251916, "loss": 1.9095, "step": 3353 }, { "epoch": 1.01, "learning_rate": 0.00015463322880086258, "loss": 1.9949, "step": 3354 }, { "epoch": 1.01, "learning_rate": 0.00015460600651924115, "loss": 1.9271, "step": 3355 }, { "epoch": 1.01, "learning_rate": 0.0001545787784705298, "loss": 1.9636, "step": 3356 }, { "epoch": 1.01, "learning_rate": 0.0001545515446576042, "loss": 1.9705, "step": 3357 }, { "epoch": 1.01, "learning_rate": 0.00015452430508334058, "loss": 1.8916, "step": 3358 }, { "epoch": 1.01, "learning_rate": 0.00015449705975061575, "loss": 1.9458, "step": 3359 }, { "epoch": 1.01, "learning_rate": 0.0001544698086623072, "loss": 1.9738, "step": 3360 }, { "epoch": 1.01, "learning_rate": 0.000154442551821293, "loss": 1.9465, "step": 3361 }, { "epoch": 1.01, "learning_rate": 0.0001544152892304518, "loss": 1.9844, "step": 3362 }, { "epoch": 1.01, "learning_rate": 0.00015438802089266286, "loss": 1.858, "step": 3363 }, { "epoch": 1.01, "learning_rate": 0.0001543607468108061, "loss": 1.9206, "step": 3364 }, { "epoch": 1.01, "learning_rate": 0.000154333466987762, "loss": 1.8876, "step": 3365 }, { "epoch": 1.01, "learning_rate": 0.00015430618142641165, "loss": 1.9047, "step": 3366 }, { "epoch": 1.01, "learning_rate": 0.00015427889012963674, "loss": 1.8264, "step": 3367 }, { "epoch": 1.01, "learning_rate": 0.00015425159310031961, "loss": 1.9189, "step": 3368 }, { "epoch": 1.01, "learning_rate": 0.00015422429034134315, "loss": 1.8728, "step": 3369 }, { "epoch": 1.01, "learning_rate": 0.00015419698185559092, "loss": 2.027, "step": 3370 }, { "epoch": 1.01, "learning_rate": 0.00015416966764594697, "loss": 1.8914, "step": 3371 }, { "epoch": 1.02, "learning_rate": 0.00015414234771529608, "loss": 1.9108, "step": 3372 }, { "epoch": 1.02, "learning_rate": 0.00015411502206652357, "loss": 1.8988, "step": 3373 }, { "epoch": 1.02, "learning_rate": 0.0001540876907025154, "loss": 1.9166, "step": 3374 }, { "epoch": 1.02, "learning_rate": 0.00015406035362615806, "loss": 1.8941, "step": 3375 }, { "epoch": 1.02, "learning_rate": 0.00015403301084033875, "loss": 1.9721, "step": 3376 }, { "epoch": 1.02, "learning_rate": 0.00015400566234794518, "loss": 1.9639, "step": 3377 }, { "epoch": 1.02, "learning_rate": 0.00015397830815186571, "loss": 1.8845, "step": 3378 }, { "epoch": 1.02, "learning_rate": 0.0001539509482549893, "loss": 2.5239, "step": 3379 }, { "epoch": 1.02, "learning_rate": 0.00015392358266020547, "loss": 1.9056, "step": 3380 }, { "epoch": 1.02, "learning_rate": 0.00015389621137040438, "loss": 1.8971, "step": 3381 }, { "epoch": 1.02, "learning_rate": 0.00015386883438847683, "loss": 1.8051, "step": 3382 }, { "epoch": 1.02, "learning_rate": 0.00015384145171731415, "loss": 1.8419, "step": 3383 }, { "epoch": 1.02, "learning_rate": 0.00015381406335980831, "loss": 1.902, "step": 3384 }, { "epoch": 1.02, "learning_rate": 0.00015378666931885188, "loss": 1.9007, "step": 3385 }, { "epoch": 1.02, "learning_rate": 0.00015375926959733795, "loss": 1.8841, "step": 3386 }, { "epoch": 1.02, "learning_rate": 0.0001537318641981604, "loss": 1.9529, "step": 3387 }, { "epoch": 1.02, "learning_rate": 0.00015370445312421348, "loss": 1.8234, "step": 3388 }, { "epoch": 1.02, "learning_rate": 0.0001536770363783922, "loss": 1.919, "step": 3389 }, { "epoch": 1.02, "learning_rate": 0.00015364961396359208, "loss": 1.8092, "step": 3390 }, { "epoch": 1.02, "learning_rate": 0.0001536221858827093, "loss": 2.0518, "step": 3391 }, { "epoch": 1.02, "learning_rate": 0.00015359475213864068, "loss": 1.9086, "step": 3392 }, { "epoch": 1.02, "learning_rate": 0.00015356731273428345, "loss": 2.0216, "step": 3393 }, { "epoch": 1.02, "learning_rate": 0.00015353986767253566, "loss": 1.8129, "step": 3394 }, { "epoch": 1.02, "learning_rate": 0.00015351241695629576, "loss": 1.8337, "step": 3395 }, { "epoch": 1.02, "learning_rate": 0.000153484960588463, "loss": 1.8661, "step": 3396 }, { "epoch": 1.02, "learning_rate": 0.0001534574985719371, "loss": 1.8784, "step": 3397 }, { "epoch": 1.02, "learning_rate": 0.00015343003090961834, "loss": 1.9551, "step": 3398 }, { "epoch": 1.02, "learning_rate": 0.0001534025576044077, "loss": 1.8499, "step": 3399 }, { "epoch": 1.02, "learning_rate": 0.0001533750786592067, "loss": 2.0286, "step": 3400 }, { "epoch": 1.02, "learning_rate": 0.00015334759407691748, "loss": 1.8355, "step": 3401 }, { "epoch": 1.02, "learning_rate": 0.00015332010386044274, "loss": 1.9119, "step": 3402 }, { "epoch": 1.02, "learning_rate": 0.0001532926080126858, "loss": 1.8569, "step": 3403 }, { "epoch": 1.02, "learning_rate": 0.0001532651065365506, "loss": 1.8398, "step": 3404 }, { "epoch": 1.02, "learning_rate": 0.0001532375994349416, "loss": 1.9239, "step": 3405 }, { "epoch": 1.03, "learning_rate": 0.00015321008671076395, "loss": 1.8324, "step": 3406 }, { "epoch": 1.03, "learning_rate": 0.0001531825683669233, "loss": 1.8559, "step": 3407 }, { "epoch": 1.03, "learning_rate": 0.00015315504440632598, "loss": 1.888, "step": 3408 }, { "epoch": 1.03, "learning_rate": 0.0001531275148318788, "loss": 1.8949, "step": 3409 }, { "epoch": 1.03, "learning_rate": 0.00015309997964648933, "loss": 2.0077, "step": 3410 }, { "epoch": 1.03, "learning_rate": 0.0001530724388530656, "loss": 1.8769, "step": 3411 }, { "epoch": 1.03, "learning_rate": 0.0001530448924545162, "loss": 2.0997, "step": 3412 }, { "epoch": 1.03, "learning_rate": 0.00015301734045375048, "loss": 1.8664, "step": 3413 }, { "epoch": 1.03, "learning_rate": 0.00015298978285367825, "loss": 1.7489, "step": 3414 }, { "epoch": 1.03, "learning_rate": 0.0001529622196572099, "loss": 1.8406, "step": 3415 }, { "epoch": 1.03, "learning_rate": 0.0001529346508672565, "loss": 1.8136, "step": 3416 }, { "epoch": 1.03, "learning_rate": 0.00015290707648672966, "loss": 2.0365, "step": 3417 }, { "epoch": 1.03, "learning_rate": 0.0001528794965185416, "loss": 1.9548, "step": 3418 }, { "epoch": 1.03, "learning_rate": 0.00015285191096560508, "loss": 1.7668, "step": 3419 }, { "epoch": 1.03, "learning_rate": 0.00015282431983083345, "loss": 1.895, "step": 3420 }, { "epoch": 1.03, "learning_rate": 0.00015279672311714078, "loss": 1.7711, "step": 3421 }, { "epoch": 1.03, "learning_rate": 0.0001527691208274416, "loss": 1.8558, "step": 3422 }, { "epoch": 1.03, "learning_rate": 0.00015274151296465106, "loss": 1.9425, "step": 3423 }, { "epoch": 1.03, "learning_rate": 0.00015271389953168488, "loss": 1.9913, "step": 3424 }, { "epoch": 1.03, "learning_rate": 0.0001526862805314594, "loss": 1.99, "step": 3425 }, { "epoch": 1.03, "learning_rate": 0.00015265865596689155, "loss": 1.814, "step": 3426 }, { "epoch": 1.03, "learning_rate": 0.00015263102584089886, "loss": 1.8978, "step": 3427 }, { "epoch": 1.03, "learning_rate": 0.00015260339015639934, "loss": 1.8634, "step": 3428 }, { "epoch": 1.03, "learning_rate": 0.00015257574891631176, "loss": 1.8709, "step": 3429 }, { "epoch": 1.03, "learning_rate": 0.00015254810212355534, "loss": 1.8099, "step": 3430 }, { "epoch": 1.03, "learning_rate": 0.00015252044978104998, "loss": 1.8308, "step": 3431 }, { "epoch": 1.03, "learning_rate": 0.00015249279189171605, "loss": 1.9329, "step": 3432 }, { "epoch": 1.03, "learning_rate": 0.0001524651284584746, "loss": 1.8421, "step": 3433 }, { "epoch": 1.03, "learning_rate": 0.00015243745948424729, "loss": 1.8731, "step": 3434 }, { "epoch": 1.03, "learning_rate": 0.00015240978497195625, "loss": 1.9991, "step": 3435 }, { "epoch": 1.03, "learning_rate": 0.0001523821049245243, "loss": 1.8509, "step": 3436 }, { "epoch": 1.03, "learning_rate": 0.00015235441934487477, "loss": 1.9784, "step": 3437 }, { "epoch": 1.03, "learning_rate": 0.00015232672823593167, "loss": 1.9032, "step": 3438 }, { "epoch": 1.04, "learning_rate": 0.00015229903160061947, "loss": 1.8249, "step": 3439 }, { "epoch": 1.04, "learning_rate": 0.00015227132944186332, "loss": 1.8325, "step": 3440 }, { "epoch": 1.04, "learning_rate": 0.00015224362176258894, "loss": 1.9329, "step": 3441 }, { "epoch": 1.04, "learning_rate": 0.00015221590856572255, "loss": 1.8635, "step": 3442 }, { "epoch": 1.04, "learning_rate": 0.00015218818985419108, "loss": 1.8178, "step": 3443 }, { "epoch": 1.04, "learning_rate": 0.00015216046563092196, "loss": 1.919, "step": 3444 }, { "epoch": 1.04, "learning_rate": 0.0001521327358988432, "loss": 1.833, "step": 3445 }, { "epoch": 1.04, "learning_rate": 0.0001521050006608834, "loss": 1.8131, "step": 3446 }, { "epoch": 1.04, "learning_rate": 0.00015207725991997183, "loss": 1.7899, "step": 3447 }, { "epoch": 1.04, "learning_rate": 0.00015204951367903826, "loss": 1.7549, "step": 3448 }, { "epoch": 1.04, "learning_rate": 0.00015202176194101293, "loss": 1.8629, "step": 3449 }, { "epoch": 1.04, "learning_rate": 0.0001519940047088269, "loss": 1.7822, "step": 3450 }, { "epoch": 1.04, "learning_rate": 0.00015196624198541162, "loss": 1.8334, "step": 3451 }, { "epoch": 1.04, "learning_rate": 0.0001519384737736992, "loss": 1.8762, "step": 3452 }, { "epoch": 1.04, "learning_rate": 0.00015191070007662235, "loss": 1.8328, "step": 3453 }, { "epoch": 1.04, "learning_rate": 0.00015188292089711428, "loss": 1.8421, "step": 3454 }, { "epoch": 1.04, "learning_rate": 0.00015185513623810882, "loss": 1.8474, "step": 3455 }, { "epoch": 1.04, "learning_rate": 0.00015182734610254048, "loss": 1.861, "step": 3456 }, { "epoch": 1.04, "learning_rate": 0.00015179955049334412, "loss": 1.8916, "step": 3457 }, { "epoch": 1.04, "learning_rate": 0.00015177174941345543, "loss": 1.7979, "step": 3458 }, { "epoch": 1.04, "learning_rate": 0.00015174394286581047, "loss": 1.8658, "step": 3459 }, { "epoch": 1.04, "learning_rate": 0.000151716130853346, "loss": 1.8983, "step": 3460 }, { "epoch": 1.04, "learning_rate": 0.0001516883133789993, "loss": 1.9005, "step": 3461 }, { "epoch": 1.04, "learning_rate": 0.00015166049044570826, "loss": 1.9799, "step": 3462 }, { "epoch": 1.04, "learning_rate": 0.0001516326620564114, "loss": 1.7965, "step": 3463 }, { "epoch": 1.04, "learning_rate": 0.00015160482821404764, "loss": 1.8757, "step": 3464 }, { "epoch": 1.04, "learning_rate": 0.0001515769889215567, "loss": 1.8688, "step": 3465 }, { "epoch": 1.04, "learning_rate": 0.00015154914418187868, "loss": 1.8006, "step": 3466 }, { "epoch": 1.04, "learning_rate": 0.00015152129399795437, "loss": 1.8119, "step": 3467 }, { "epoch": 1.04, "learning_rate": 0.0001514934383727251, "loss": 1.9254, "step": 3468 }, { "epoch": 1.04, "learning_rate": 0.00015146557730913278, "loss": 1.8901, "step": 3469 }, { "epoch": 1.04, "learning_rate": 0.00015143771081011992, "loss": 1.8229, "step": 3470 }, { "epoch": 1.04, "learning_rate": 0.00015140983887862954, "loss": 2.0681, "step": 3471 }, { "epoch": 1.05, "learning_rate": 0.00015138196151760527, "loss": 1.959, "step": 3472 }, { "epoch": 1.05, "learning_rate": 0.00015135407872999133, "loss": 1.9657, "step": 3473 }, { "epoch": 1.05, "learning_rate": 0.0001513261905187325, "loss": 1.8947, "step": 3474 }, { "epoch": 1.05, "learning_rate": 0.00015129829688677416, "loss": 1.9677, "step": 3475 }, { "epoch": 1.05, "learning_rate": 0.00015127039783706216, "loss": 1.828, "step": 3476 }, { "epoch": 1.05, "learning_rate": 0.00015124249337254307, "loss": 1.8718, "step": 3477 }, { "epoch": 1.05, "learning_rate": 0.00015121458349616387, "loss": 1.8249, "step": 3478 }, { "epoch": 1.05, "learning_rate": 0.00015118666821087225, "loss": 2.0317, "step": 3479 }, { "epoch": 1.05, "learning_rate": 0.00015115874751961647, "loss": 1.823, "step": 3480 }, { "epoch": 1.05, "learning_rate": 0.00015113082142534522, "loss": 1.9404, "step": 3481 }, { "epoch": 1.05, "learning_rate": 0.00015110288993100792, "loss": 1.8133, "step": 3482 }, { "epoch": 1.05, "learning_rate": 0.00015107495303955443, "loss": 1.8371, "step": 3483 }, { "epoch": 1.05, "learning_rate": 0.0001510470107539353, "loss": 1.86, "step": 3484 }, { "epoch": 1.05, "learning_rate": 0.00015101906307710155, "loss": 1.8971, "step": 3485 }, { "epoch": 1.05, "learning_rate": 0.00015099111001200483, "loss": 1.9141, "step": 3486 }, { "epoch": 1.05, "learning_rate": 0.00015096315156159734, "loss": 1.9586, "step": 3487 }, { "epoch": 1.05, "learning_rate": 0.00015093518772883187, "loss": 1.7692, "step": 3488 }, { "epoch": 1.05, "learning_rate": 0.0001509072185166617, "loss": 1.9042, "step": 3489 }, { "epoch": 1.05, "learning_rate": 0.00015087924392804075, "loss": 1.8957, "step": 3490 }, { "epoch": 1.05, "learning_rate": 0.00015085126396592352, "loss": 1.8444, "step": 3491 }, { "epoch": 1.05, "learning_rate": 0.00015082327863326508, "loss": 1.8935, "step": 3492 }, { "epoch": 1.05, "learning_rate": 0.00015079528793302097, "loss": 1.8388, "step": 3493 }, { "epoch": 1.05, "learning_rate": 0.0001507672918681474, "loss": 1.942, "step": 3494 }, { "epoch": 1.05, "learning_rate": 0.00015073929044160108, "loss": 1.9343, "step": 3495 }, { "epoch": 1.05, "learning_rate": 0.00015071128365633934, "loss": 2.0203, "step": 3496 }, { "epoch": 1.05, "learning_rate": 0.00015068327151532008, "loss": 1.8309, "step": 3497 }, { "epoch": 1.05, "learning_rate": 0.0001506552540215017, "loss": 1.8537, "step": 3498 }, { "epoch": 1.05, "learning_rate": 0.00015062723117784322, "loss": 1.8978, "step": 3499 }, { "epoch": 1.05, "learning_rate": 0.00015059920298730422, "loss": 1.8791, "step": 3500 }, { "epoch": 1.05, "learning_rate": 0.0001505711694528448, "loss": 1.8152, "step": 3501 }, { "epoch": 1.05, "learning_rate": 0.0001505431305774257, "loss": 1.7926, "step": 3502 }, { "epoch": 1.05, "learning_rate": 0.00015051508636400814, "loss": 1.9529, "step": 3503 }, { "epoch": 1.05, "learning_rate": 0.00015048703681555393, "loss": 1.8608, "step": 3504 }, { "epoch": 1.06, "learning_rate": 0.00015045898193502555, "loss": 1.877, "step": 3505 }, { "epoch": 1.06, "learning_rate": 0.00015043092172538588, "loss": 1.8947, "step": 3506 }, { "epoch": 1.06, "learning_rate": 0.0001504028561895985, "loss": 1.8868, "step": 3507 }, { "epoch": 1.06, "learning_rate": 0.0001503747853306274, "loss": 2.0545, "step": 3508 }, { "epoch": 1.06, "learning_rate": 0.00015034670915143724, "loss": 1.9487, "step": 3509 }, { "epoch": 1.06, "learning_rate": 0.0001503186276549933, "loss": 1.8668, "step": 3510 }, { "epoch": 1.06, "learning_rate": 0.00015029054084426125, "loss": 1.9166, "step": 3511 }, { "epoch": 1.06, "learning_rate": 0.0001502624487222075, "loss": 1.8016, "step": 3512 }, { "epoch": 1.06, "learning_rate": 0.00015023435129179883, "loss": 1.9238, "step": 3513 }, { "epoch": 1.06, "learning_rate": 0.0001502062485560028, "loss": 1.9189, "step": 3514 }, { "epoch": 1.06, "learning_rate": 0.00015017814051778735, "loss": 1.9724, "step": 3515 }, { "epoch": 1.06, "learning_rate": 0.00015015002718012105, "loss": 1.7999, "step": 3516 }, { "epoch": 1.06, "learning_rate": 0.00015012190854597304, "loss": 1.9497, "step": 3517 }, { "epoch": 1.06, "learning_rate": 0.000150093784618313, "loss": 1.8752, "step": 3518 }, { "epoch": 1.06, "learning_rate": 0.0001500656554001112, "loss": 1.9174, "step": 3519 }, { "epoch": 1.06, "learning_rate": 0.00015003752089433838, "loss": 1.866, "step": 3520 }, { "epoch": 1.06, "learning_rate": 0.00015000938110396595, "loss": 1.834, "step": 3521 }, { "epoch": 1.06, "learning_rate": 0.00014998123603196585, "loss": 1.7695, "step": 3522 }, { "epoch": 1.06, "learning_rate": 0.00014995308568131052, "loss": 1.8544, "step": 3523 }, { "epoch": 1.06, "learning_rate": 0.00014992493005497304, "loss": 1.8747, "step": 3524 }, { "epoch": 1.06, "learning_rate": 0.00014989676915592694, "loss": 1.8909, "step": 3525 }, { "epoch": 1.06, "learning_rate": 0.00014986860298714643, "loss": 1.9497, "step": 3526 }, { "epoch": 1.06, "learning_rate": 0.00014984043155160615, "loss": 1.7947, "step": 3527 }, { "epoch": 1.06, "learning_rate": 0.00014981225485228138, "loss": 1.8338, "step": 3528 }, { "epoch": 1.06, "learning_rate": 0.00014978407289214802, "loss": 1.9699, "step": 3529 }, { "epoch": 1.06, "learning_rate": 0.00014975588567418236, "loss": 1.8215, "step": 3530 }, { "epoch": 1.06, "learning_rate": 0.00014972769320136133, "loss": 1.9299, "step": 3531 }, { "epoch": 1.06, "learning_rate": 0.00014969949547666246, "loss": 1.9048, "step": 3532 }, { "epoch": 1.06, "learning_rate": 0.00014967129250306372, "loss": 1.8397, "step": 3533 }, { "epoch": 1.06, "learning_rate": 0.0001496430842835438, "loss": 1.9237, "step": 3534 }, { "epoch": 1.06, "learning_rate": 0.00014961487082108178, "loss": 1.9664, "step": 3535 }, { "epoch": 1.06, "learning_rate": 0.0001495866521186574, "loss": 1.9748, "step": 3536 }, { "epoch": 1.06, "learning_rate": 0.00014955842817925084, "loss": 1.8575, "step": 3537 }, { "epoch": 1.07, "learning_rate": 0.000149530199005843, "loss": 1.8767, "step": 3538 }, { "epoch": 1.07, "learning_rate": 0.00014950196460141517, "loss": 1.823, "step": 3539 }, { "epoch": 1.07, "learning_rate": 0.0001494737249689493, "loss": 1.9705, "step": 3540 }, { "epoch": 1.07, "learning_rate": 0.00014944548011142786, "loss": 1.9179, "step": 3541 }, { "epoch": 1.07, "learning_rate": 0.0001494172300318338, "loss": 1.9031, "step": 3542 }, { "epoch": 1.07, "learning_rate": 0.0001493889747331508, "loss": 1.7875, "step": 3543 }, { "epoch": 1.07, "learning_rate": 0.0001493607142183629, "loss": 1.766, "step": 3544 }, { "epoch": 1.07, "learning_rate": 0.0001493324484904548, "loss": 1.8055, "step": 3545 }, { "epoch": 1.07, "learning_rate": 0.00014930417755241172, "loss": 1.8183, "step": 3546 }, { "epoch": 1.07, "learning_rate": 0.00014927590140721938, "loss": 1.9165, "step": 3547 }, { "epoch": 1.07, "learning_rate": 0.0001492476200578642, "loss": 1.8535, "step": 3548 }, { "epoch": 1.07, "learning_rate": 0.00014921933350733294, "loss": 1.8735, "step": 3549 }, { "epoch": 1.07, "learning_rate": 0.0001491910417586131, "loss": 1.7772, "step": 3550 }, { "epoch": 1.07, "learning_rate": 0.00014916274481469263, "loss": 1.8327, "step": 3551 }, { "epoch": 1.07, "learning_rate": 0.00014913444267856005, "loss": 1.8293, "step": 3552 }, { "epoch": 1.07, "learning_rate": 0.00014910613535320443, "loss": 1.8618, "step": 3553 }, { "epoch": 1.07, "learning_rate": 0.0001490778228416153, "loss": 1.9201, "step": 3554 }, { "epoch": 1.07, "learning_rate": 0.00014904950514678297, "loss": 1.7921, "step": 3555 }, { "epoch": 1.07, "learning_rate": 0.00014902118227169803, "loss": 1.9281, "step": 3556 }, { "epoch": 1.07, "learning_rate": 0.00014899285421935177, "loss": 1.8986, "step": 3557 }, { "epoch": 1.07, "learning_rate": 0.00014896452099273606, "loss": 1.9602, "step": 3558 }, { "epoch": 1.07, "learning_rate": 0.00014893618259484315, "loss": 1.8663, "step": 3559 }, { "epoch": 1.07, "learning_rate": 0.000148907839028666, "loss": 1.7929, "step": 3560 }, { "epoch": 1.07, "learning_rate": 0.000148879490297198, "loss": 1.9362, "step": 3561 }, { "epoch": 1.07, "learning_rate": 0.0001488511364034332, "loss": 1.8531, "step": 3562 }, { "epoch": 1.07, "learning_rate": 0.00014882277735036605, "loss": 1.7935, "step": 3563 }, { "epoch": 1.07, "learning_rate": 0.00014879441314099174, "loss": 1.8031, "step": 3564 }, { "epoch": 1.07, "learning_rate": 0.00014876604377830581, "loss": 1.7802, "step": 3565 }, { "epoch": 1.07, "learning_rate": 0.0001487376692653044, "loss": 1.757, "step": 3566 }, { "epoch": 1.07, "learning_rate": 0.00014870928960498432, "loss": 1.81, "step": 3567 }, { "epoch": 1.07, "learning_rate": 0.00014868090480034274, "loss": 1.7933, "step": 3568 }, { "epoch": 1.07, "learning_rate": 0.0001486525148543775, "loss": 1.8621, "step": 3569 }, { "epoch": 1.07, "learning_rate": 0.0001486241197700869, "loss": 1.8711, "step": 3570 }, { "epoch": 1.07, "learning_rate": 0.0001485957195504699, "loss": 1.9032, "step": 3571 }, { "epoch": 1.08, "learning_rate": 0.00014856731419852584, "loss": 1.8553, "step": 3572 }, { "epoch": 1.08, "learning_rate": 0.0001485389037172547, "loss": 1.8616, "step": 3573 }, { "epoch": 1.08, "learning_rate": 0.00014851048810965702, "loss": 1.9646, "step": 3574 }, { "epoch": 1.08, "learning_rate": 0.00014848206737873385, "loss": 1.7888, "step": 3575 }, { "epoch": 1.08, "learning_rate": 0.00014845364152748678, "loss": 1.8499, "step": 3576 }, { "epoch": 1.08, "learning_rate": 0.00014842521055891792, "loss": 1.8424, "step": 3577 }, { "epoch": 1.08, "learning_rate": 0.00014839677447602996, "loss": 1.8173, "step": 3578 }, { "epoch": 1.08, "learning_rate": 0.0001483683332818261, "loss": 1.8692, "step": 3579 }, { "epoch": 1.08, "learning_rate": 0.00014833988697931009, "loss": 1.8891, "step": 3580 }, { "epoch": 1.08, "learning_rate": 0.00014831143557148622, "loss": 1.8554, "step": 3581 }, { "epoch": 1.08, "learning_rate": 0.00014828297906135934, "loss": 1.9859, "step": 3582 }, { "epoch": 1.08, "learning_rate": 0.00014825451745193485, "loss": 1.8981, "step": 3583 }, { "epoch": 1.08, "learning_rate": 0.00014822605074621862, "loss": 1.9277, "step": 3584 }, { "epoch": 1.08, "learning_rate": 0.00014819757894721705, "loss": 1.9074, "step": 3585 }, { "epoch": 1.08, "learning_rate": 0.00014816910205793719, "loss": 1.9861, "step": 3586 }, { "epoch": 1.08, "learning_rate": 0.00014814062008138655, "loss": 1.8247, "step": 3587 }, { "epoch": 1.08, "learning_rate": 0.00014811213302057317, "loss": 1.8693, "step": 3588 }, { "epoch": 1.08, "learning_rate": 0.00014808364087850567, "loss": 1.8191, "step": 3589 }, { "epoch": 1.08, "learning_rate": 0.0001480551436581932, "loss": 1.878, "step": 3590 }, { "epoch": 1.08, "learning_rate": 0.0001480266413626454, "loss": 1.8436, "step": 3591 }, { "epoch": 1.08, "learning_rate": 0.00014799813399487242, "loss": 1.9264, "step": 3592 }, { "epoch": 1.08, "learning_rate": 0.00014796962155788512, "loss": 1.8923, "step": 3593 }, { "epoch": 1.08, "learning_rate": 0.00014794110405469467, "loss": 1.9129, "step": 3594 }, { "epoch": 1.08, "learning_rate": 0.00014791258148831297, "loss": 1.9717, "step": 3595 }, { "epoch": 1.08, "learning_rate": 0.0001478840538617523, "loss": 1.9202, "step": 3596 }, { "epoch": 1.08, "learning_rate": 0.0001478555211780256, "loss": 1.8829, "step": 3597 }, { "epoch": 1.08, "learning_rate": 0.00014782698344014623, "loss": 1.8436, "step": 3598 }, { "epoch": 1.08, "learning_rate": 0.00014779844065112815, "loss": 1.8491, "step": 3599 }, { "epoch": 1.08, "learning_rate": 0.00014776989281398584, "loss": 1.9284, "step": 3600 }, { "epoch": 1.08, "learning_rate": 0.00014774133993173433, "loss": 1.8867, "step": 3601 }, { "epoch": 1.08, "learning_rate": 0.0001477127820073892, "loss": 1.8108, "step": 3602 }, { "epoch": 1.08, "learning_rate": 0.00014768421904396645, "loss": 1.9058, "step": 3603 }, { "epoch": 1.08, "learning_rate": 0.00014765565104448276, "loss": 1.9473, "step": 3604 }, { "epoch": 1.09, "learning_rate": 0.00014762707801195527, "loss": 1.9801, "step": 3605 }, { "epoch": 1.09, "learning_rate": 0.0001475984999494016, "loss": 1.7458, "step": 3606 }, { "epoch": 1.09, "learning_rate": 0.00014756991685984004, "loss": 1.9, "step": 3607 }, { "epoch": 1.09, "learning_rate": 0.00014754132874628923, "loss": 1.8538, "step": 3608 }, { "epoch": 1.09, "learning_rate": 0.00014751273561176853, "loss": 1.7655, "step": 3609 }, { "epoch": 1.09, "learning_rate": 0.0001474841374592977, "loss": 1.7483, "step": 3610 }, { "epoch": 1.09, "learning_rate": 0.00014745553429189705, "loss": 1.9265, "step": 3611 }, { "epoch": 1.09, "learning_rate": 0.00014742692611258748, "loss": 1.7699, "step": 3612 }, { "epoch": 1.09, "learning_rate": 0.00014739831292439034, "loss": 1.865, "step": 3613 }, { "epoch": 1.09, "learning_rate": 0.0001473696947303276, "loss": 1.9418, "step": 3614 }, { "epoch": 1.09, "learning_rate": 0.00014734107153342165, "loss": 1.8733, "step": 3615 }, { "epoch": 1.09, "learning_rate": 0.00014731244333669548, "loss": 1.8003, "step": 3616 }, { "epoch": 1.09, "learning_rate": 0.0001472838101431726, "loss": 1.8437, "step": 3617 }, { "epoch": 1.09, "learning_rate": 0.00014725517195587703, "loss": 1.8396, "step": 3618 }, { "epoch": 1.09, "learning_rate": 0.00014722652877783337, "loss": 1.9023, "step": 3619 }, { "epoch": 1.09, "learning_rate": 0.00014719788061206665, "loss": 1.8807, "step": 3620 }, { "epoch": 1.09, "learning_rate": 0.0001471692274616025, "loss": 1.8229, "step": 3621 }, { "epoch": 1.09, "learning_rate": 0.00014714056932946704, "loss": 1.8865, "step": 3622 }, { "epoch": 1.09, "learning_rate": 0.000147111906218687, "loss": 1.7989, "step": 3623 }, { "epoch": 1.09, "learning_rate": 0.00014708323813228947, "loss": 1.8702, "step": 3624 }, { "epoch": 1.09, "learning_rate": 0.00014705456507330226, "loss": 1.8284, "step": 3625 }, { "epoch": 1.09, "learning_rate": 0.00014702588704475355, "loss": 1.864, "step": 3626 }, { "epoch": 1.09, "learning_rate": 0.00014699720404967213, "loss": 1.8235, "step": 3627 }, { "epoch": 1.09, "learning_rate": 0.00014696851609108727, "loss": 1.8599, "step": 3628 }, { "epoch": 1.09, "learning_rate": 0.00014693982317202882, "loss": 1.8477, "step": 3629 }, { "epoch": 1.09, "learning_rate": 0.00014691112529552705, "loss": 1.8784, "step": 3630 }, { "epoch": 1.09, "learning_rate": 0.0001468824224646129, "loss": 1.7733, "step": 3631 }, { "epoch": 1.09, "learning_rate": 0.00014685371468231773, "loss": 1.8676, "step": 3632 }, { "epoch": 1.09, "learning_rate": 0.00014682500195167342, "loss": 1.8504, "step": 3633 }, { "epoch": 1.09, "learning_rate": 0.0001467962842757124, "loss": 1.7621, "step": 3634 }, { "epoch": 1.09, "learning_rate": 0.00014676756165746765, "loss": 1.7729, "step": 3635 }, { "epoch": 1.09, "learning_rate": 0.00014673883409997265, "loss": 1.7884, "step": 3636 }, { "epoch": 1.09, "learning_rate": 0.00014671010160626136, "loss": 1.8837, "step": 3637 }, { "epoch": 1.1, "learning_rate": 0.00014668136417936834, "loss": 1.8873, "step": 3638 }, { "epoch": 1.1, "learning_rate": 0.0001466526218223286, "loss": 1.872, "step": 3639 }, { "epoch": 1.1, "learning_rate": 0.00014662387453817768, "loss": 1.9227, "step": 3640 }, { "epoch": 1.1, "learning_rate": 0.00014659512232995174, "loss": 1.8015, "step": 3641 }, { "epoch": 1.1, "learning_rate": 0.00014656636520068728, "loss": 1.8757, "step": 3642 }, { "epoch": 1.1, "learning_rate": 0.0001465376031534215, "loss": 1.8369, "step": 3643 }, { "epoch": 1.1, "learning_rate": 0.00014650883619119195, "loss": 1.8639, "step": 3644 }, { "epoch": 1.1, "learning_rate": 0.00014648006431703688, "loss": 1.8333, "step": 3645 }, { "epoch": 1.1, "learning_rate": 0.00014645128753399488, "loss": 1.8297, "step": 3646 }, { "epoch": 1.1, "learning_rate": 0.00014642250584510525, "loss": 1.809, "step": 3647 }, { "epoch": 1.1, "learning_rate": 0.00014639371925340763, "loss": 1.8562, "step": 3648 }, { "epoch": 1.1, "learning_rate": 0.00014636492776194223, "loss": 1.7518, "step": 3649 }, { "epoch": 1.1, "learning_rate": 0.00014633613137374985, "loss": 1.7933, "step": 3650 }, { "epoch": 1.1, "learning_rate": 0.00014630733009187176, "loss": 1.7693, "step": 3651 }, { "epoch": 1.1, "learning_rate": 0.00014627852391934967, "loss": 1.8207, "step": 3652 }, { "epoch": 1.1, "learning_rate": 0.00014624971285922598, "loss": 1.781, "step": 3653 }, { "epoch": 1.1, "learning_rate": 0.00014622089691454344, "loss": 1.8761, "step": 3654 }, { "epoch": 1.1, "learning_rate": 0.0001461920760883454, "loss": 1.8871, "step": 3655 }, { "epoch": 1.1, "learning_rate": 0.0001461632503836757, "loss": 1.7788, "step": 3656 }, { "epoch": 1.1, "learning_rate": 0.0001461344198035787, "loss": 1.8041, "step": 3657 }, { "epoch": 1.1, "learning_rate": 0.00014610558435109924, "loss": 1.8315, "step": 3658 }, { "epoch": 1.1, "learning_rate": 0.0001460767440292828, "loss": 1.8869, "step": 3659 }, { "epoch": 1.1, "learning_rate": 0.00014604789884117522, "loss": 1.8474, "step": 3660 }, { "epoch": 1.1, "learning_rate": 0.00014601904878982296, "loss": 1.8433, "step": 3661 }, { "epoch": 1.1, "learning_rate": 0.0001459901938782729, "loss": 1.9019, "step": 3662 }, { "epoch": 1.1, "learning_rate": 0.0001459613341095725, "loss": 1.9187, "step": 3663 }, { "epoch": 1.1, "learning_rate": 0.00014593246948676978, "loss": 1.8181, "step": 3664 }, { "epoch": 1.1, "learning_rate": 0.00014590360001291312, "loss": 1.8557, "step": 3665 }, { "epoch": 1.1, "learning_rate": 0.0001458747256910516, "loss": 1.7691, "step": 3666 }, { "epoch": 1.1, "learning_rate": 0.0001458458465242346, "loss": 1.9147, "step": 3667 }, { "epoch": 1.1, "learning_rate": 0.00014581696251551228, "loss": 1.8435, "step": 3668 }, { "epoch": 1.1, "learning_rate": 0.000145788073667935, "loss": 1.7758, "step": 3669 }, { "epoch": 1.1, "learning_rate": 0.0001457591799845539, "loss": 1.7686, "step": 3670 }, { "epoch": 1.11, "learning_rate": 0.00014573028146842046, "loss": 1.8345, "step": 3671 }, { "epoch": 1.11, "learning_rate": 0.00014570137812258674, "loss": 1.8887, "step": 3672 }, { "epoch": 1.11, "learning_rate": 0.00014567246995010537, "loss": 1.8978, "step": 3673 }, { "epoch": 1.11, "learning_rate": 0.00014564355695402935, "loss": 1.806, "step": 3674 }, { "epoch": 1.11, "learning_rate": 0.00014561463913741228, "loss": 1.7989, "step": 3675 }, { "epoch": 1.11, "learning_rate": 0.00014558571650330823, "loss": 1.9408, "step": 3676 }, { "epoch": 1.11, "learning_rate": 0.00014555678905477183, "loss": 1.8217, "step": 3677 }, { "epoch": 1.11, "learning_rate": 0.00014552785679485818, "loss": 1.7684, "step": 3678 }, { "epoch": 1.11, "learning_rate": 0.00014549891972662289, "loss": 1.8201, "step": 3679 }, { "epoch": 1.11, "learning_rate": 0.00014546997785312208, "loss": 1.8203, "step": 3680 }, { "epoch": 1.11, "learning_rate": 0.00014544103117741238, "loss": 1.7989, "step": 3681 }, { "epoch": 1.11, "learning_rate": 0.00014541207970255094, "loss": 1.8376, "step": 3682 }, { "epoch": 1.11, "learning_rate": 0.0001453831234315954, "loss": 1.8141, "step": 3683 }, { "epoch": 1.11, "learning_rate": 0.0001453541623676039, "loss": 1.785, "step": 3684 }, { "epoch": 1.11, "learning_rate": 0.00014532519651363509, "loss": 1.9173, "step": 3685 }, { "epoch": 1.11, "learning_rate": 0.00014529622587274816, "loss": 1.8949, "step": 3686 }, { "epoch": 1.11, "learning_rate": 0.00014526725044800275, "loss": 1.8249, "step": 3687 }, { "epoch": 1.11, "learning_rate": 0.0001452382702424591, "loss": 1.8636, "step": 3688 }, { "epoch": 1.11, "learning_rate": 0.00014520928525917778, "loss": 1.8577, "step": 3689 }, { "epoch": 1.11, "learning_rate": 0.00014518029550122008, "loss": 1.7504, "step": 3690 }, { "epoch": 1.11, "learning_rate": 0.0001451513009716476, "loss": 1.7817, "step": 3691 }, { "epoch": 1.11, "learning_rate": 0.0001451223016735226, "loss": 1.8533, "step": 3692 }, { "epoch": 1.11, "learning_rate": 0.00014509329760990775, "loss": 1.8182, "step": 3693 }, { "epoch": 1.11, "learning_rate": 0.0001450642887838662, "loss": 1.7835, "step": 3694 }, { "epoch": 1.11, "learning_rate": 0.00014503527519846177, "loss": 1.9063, "step": 3695 }, { "epoch": 1.11, "learning_rate": 0.00014500625685675853, "loss": 1.7971, "step": 3696 }, { "epoch": 1.11, "learning_rate": 0.00014497723376182128, "loss": 1.9004, "step": 3697 }, { "epoch": 1.11, "learning_rate": 0.00014494820591671518, "loss": 1.8214, "step": 3698 }, { "epoch": 1.11, "learning_rate": 0.00014491917332450595, "loss": 1.7588, "step": 3699 }, { "epoch": 1.11, "learning_rate": 0.00014489013598825985, "loss": 1.8756, "step": 3700 }, { "epoch": 1.11, "learning_rate": 0.0001448610939110435, "loss": 1.9349, "step": 3701 }, { "epoch": 1.11, "learning_rate": 0.0001448320470959242, "loss": 1.9047, "step": 3702 }, { "epoch": 1.11, "learning_rate": 0.0001448029955459696, "loss": 1.8215, "step": 3703 }, { "epoch": 1.11, "learning_rate": 0.00014477393926424798, "loss": 1.9188, "step": 3704 }, { "epoch": 1.12, "learning_rate": 0.000144744878253828, "loss": 1.8214, "step": 3705 }, { "epoch": 1.12, "learning_rate": 0.00014471581251777885, "loss": 1.7352, "step": 3706 }, { "epoch": 1.12, "learning_rate": 0.00014468674205917034, "loss": 1.8696, "step": 3707 }, { "epoch": 1.12, "learning_rate": 0.00014465766688107257, "loss": 1.7453, "step": 3708 }, { "epoch": 1.12, "learning_rate": 0.00014462858698655634, "loss": 1.8254, "step": 3709 }, { "epoch": 1.12, "learning_rate": 0.0001445995023786928, "loss": 1.8494, "step": 3710 }, { "epoch": 1.12, "learning_rate": 0.00014457041306055365, "loss": 1.9308, "step": 3711 }, { "epoch": 1.12, "learning_rate": 0.00014454131903521116, "loss": 1.9104, "step": 3712 }, { "epoch": 1.12, "learning_rate": 0.00014451222030573794, "loss": 1.8717, "step": 3713 }, { "epoch": 1.12, "learning_rate": 0.00014448311687520726, "loss": 1.8725, "step": 3714 }, { "epoch": 1.12, "learning_rate": 0.0001444540087466928, "loss": 1.8642, "step": 3715 }, { "epoch": 1.12, "learning_rate": 0.00014442489592326872, "loss": 1.8371, "step": 3716 }, { "epoch": 1.12, "learning_rate": 0.0001443957784080097, "loss": 1.8336, "step": 3717 }, { "epoch": 1.12, "learning_rate": 0.00014436665620399093, "loss": 1.8958, "step": 3718 }, { "epoch": 1.12, "learning_rate": 0.00014433752931428812, "loss": 1.9236, "step": 3719 }, { "epoch": 1.12, "learning_rate": 0.0001443083977419774, "loss": 1.7978, "step": 3720 }, { "epoch": 1.12, "learning_rate": 0.00014427926149013546, "loss": 1.828, "step": 3721 }, { "epoch": 1.12, "learning_rate": 0.00014425012056183944, "loss": 1.9465, "step": 3722 }, { "epoch": 1.12, "learning_rate": 0.00014422097496016701, "loss": 1.8916, "step": 3723 }, { "epoch": 1.12, "learning_rate": 0.00014419182468819632, "loss": 1.9895, "step": 3724 }, { "epoch": 1.12, "learning_rate": 0.000144162669749006, "loss": 1.8535, "step": 3725 }, { "epoch": 1.12, "learning_rate": 0.00014413351014567514, "loss": 1.8614, "step": 3726 }, { "epoch": 1.12, "learning_rate": 0.00014410434588128345, "loss": 1.865, "step": 3727 }, { "epoch": 1.12, "learning_rate": 0.00014407517695891103, "loss": 1.8288, "step": 3728 }, { "epoch": 1.12, "learning_rate": 0.0001440460033816384, "loss": 1.7805, "step": 3729 }, { "epoch": 1.12, "learning_rate": 0.00014401682515254678, "loss": 1.8225, "step": 3730 }, { "epoch": 1.12, "learning_rate": 0.0001439876422747177, "loss": 1.8396, "step": 3731 }, { "epoch": 1.12, "learning_rate": 0.00014395845475123324, "loss": 1.9723, "step": 3732 }, { "epoch": 1.12, "learning_rate": 0.00014392926258517599, "loss": 1.8802, "step": 3733 }, { "epoch": 1.12, "learning_rate": 0.000143900065779629, "loss": 1.854, "step": 3734 }, { "epoch": 1.12, "learning_rate": 0.00014387086433767586, "loss": 1.9323, "step": 3735 }, { "epoch": 1.12, "learning_rate": 0.00014384165826240058, "loss": 1.8984, "step": 3736 }, { "epoch": 1.12, "learning_rate": 0.0001438124475568877, "loss": 1.812, "step": 3737 }, { "epoch": 1.13, "learning_rate": 0.00014378323222422228, "loss": 1.8698, "step": 3738 }, { "epoch": 1.13, "learning_rate": 0.00014375401226748976, "loss": 1.9077, "step": 3739 }, { "epoch": 1.13, "learning_rate": 0.00014372478768977622, "loss": 1.9218, "step": 3740 }, { "epoch": 1.13, "learning_rate": 0.00014369555849416804, "loss": 1.8312, "step": 3741 }, { "epoch": 1.13, "learning_rate": 0.00014366632468375233, "loss": 1.9337, "step": 3742 }, { "epoch": 1.13, "learning_rate": 0.00014363708626161646, "loss": 1.8197, "step": 3743 }, { "epoch": 1.13, "learning_rate": 0.0001436078432308484, "loss": 1.8164, "step": 3744 }, { "epoch": 1.13, "learning_rate": 0.00014357859559453658, "loss": 1.8709, "step": 3745 }, { "epoch": 1.13, "learning_rate": 0.00014354934335576995, "loss": 1.8521, "step": 3746 }, { "epoch": 1.13, "learning_rate": 0.0001435200865176379, "loss": 1.8653, "step": 3747 }, { "epoch": 1.13, "learning_rate": 0.0001434908250832303, "loss": 1.9431, "step": 3748 }, { "epoch": 1.13, "learning_rate": 0.0001434615590556376, "loss": 1.7935, "step": 3749 }, { "epoch": 1.13, "learning_rate": 0.0001434322884379506, "loss": 1.9383, "step": 3750 }, { "epoch": 1.13, "learning_rate": 0.00014340301323326071, "loss": 1.9211, "step": 3751 }, { "epoch": 1.13, "learning_rate": 0.00014337373344465969, "loss": 1.7772, "step": 3752 }, { "epoch": 1.13, "learning_rate": 0.0001433444490752399, "loss": 1.8486, "step": 3753 }, { "epoch": 1.13, "learning_rate": 0.00014331516012809417, "loss": 1.8021, "step": 3754 }, { "epoch": 1.13, "learning_rate": 0.0001432858666063157, "loss": 1.8129, "step": 3755 }, { "epoch": 1.13, "learning_rate": 0.00014325656851299837, "loss": 1.7946, "step": 3756 }, { "epoch": 1.13, "learning_rate": 0.00014322726585123632, "loss": 1.7954, "step": 3757 }, { "epoch": 1.13, "learning_rate": 0.00014319795862412438, "loss": 1.865, "step": 3758 }, { "epoch": 1.13, "learning_rate": 0.00014316864683475773, "loss": 1.9003, "step": 3759 }, { "epoch": 1.13, "learning_rate": 0.00014313933048623202, "loss": 1.7463, "step": 3760 }, { "epoch": 1.13, "learning_rate": 0.0001431100095816435, "loss": 1.8145, "step": 3761 }, { "epoch": 1.13, "learning_rate": 0.0001430806841240888, "loss": 1.9186, "step": 3762 }, { "epoch": 1.13, "learning_rate": 0.00014305135411666513, "loss": 1.7643, "step": 3763 }, { "epoch": 1.13, "learning_rate": 0.00014302201956247, "loss": 1.8286, "step": 3764 }, { "epoch": 1.13, "learning_rate": 0.00014299268046460154, "loss": 1.866, "step": 3765 }, { "epoch": 1.13, "learning_rate": 0.0001429633368261584, "loss": 1.7957, "step": 3766 }, { "epoch": 1.13, "learning_rate": 0.00014293398865023954, "loss": 1.8009, "step": 3767 }, { "epoch": 1.13, "learning_rate": 0.00014290463593994462, "loss": 1.7171, "step": 3768 }, { "epoch": 1.13, "learning_rate": 0.00014287527869837356, "loss": 1.7909, "step": 3769 }, { "epoch": 1.13, "learning_rate": 0.00014284591692862692, "loss": 1.7034, "step": 3770 }, { "epoch": 1.14, "learning_rate": 0.00014281655063380562, "loss": 1.749, "step": 3771 }, { "epoch": 1.14, "learning_rate": 0.0001427871798170112, "loss": 1.7986, "step": 3772 }, { "epoch": 1.14, "learning_rate": 0.0001427578044813455, "loss": 1.7933, "step": 3773 }, { "epoch": 1.14, "learning_rate": 0.000142728424629911, "loss": 1.8417, "step": 3774 }, { "epoch": 1.14, "learning_rate": 0.00014269904026581055, "loss": 1.8434, "step": 3775 }, { "epoch": 1.14, "learning_rate": 0.0001426696513921475, "loss": 1.8063, "step": 3776 }, { "epoch": 1.14, "learning_rate": 0.00014264025801202573, "loss": 1.8498, "step": 3777 }, { "epoch": 1.14, "learning_rate": 0.00014261086012854954, "loss": 1.8379, "step": 3778 }, { "epoch": 1.14, "learning_rate": 0.00014258145774482368, "loss": 1.8246, "step": 3779 }, { "epoch": 1.14, "learning_rate": 0.0001425520508639535, "loss": 1.9002, "step": 3780 }, { "epoch": 1.14, "learning_rate": 0.0001425226394890447, "loss": 1.8085, "step": 3781 }, { "epoch": 1.14, "learning_rate": 0.00014249322362320344, "loss": 1.7454, "step": 3782 }, { "epoch": 1.14, "learning_rate": 0.00014246380326953648, "loss": 1.8483, "step": 3783 }, { "epoch": 1.14, "learning_rate": 0.00014243437843115096, "loss": 1.929, "step": 3784 }, { "epoch": 1.14, "learning_rate": 0.00014240494911115455, "loss": 1.7559, "step": 3785 }, { "epoch": 1.14, "learning_rate": 0.0001423755153126553, "loss": 1.8064, "step": 3786 }, { "epoch": 1.14, "learning_rate": 0.00014234607703876185, "loss": 1.6417, "step": 3787 }, { "epoch": 1.14, "learning_rate": 0.00014231663429258322, "loss": 1.7502, "step": 3788 }, { "epoch": 1.14, "learning_rate": 0.00014228718707722895, "loss": 1.7753, "step": 3789 }, { "epoch": 1.14, "learning_rate": 0.00014225773539580907, "loss": 1.8465, "step": 3790 }, { "epoch": 1.14, "learning_rate": 0.00014222827925143398, "loss": 1.857, "step": 3791 }, { "epoch": 1.14, "learning_rate": 0.00014219881864721472, "loss": 1.7689, "step": 3792 }, { "epoch": 1.14, "learning_rate": 0.00014216935358626263, "loss": 1.8777, "step": 3793 }, { "epoch": 1.14, "learning_rate": 0.0001421398840716896, "loss": 1.8682, "step": 3794 }, { "epoch": 1.14, "learning_rate": 0.00014211041010660805, "loss": 1.8425, "step": 3795 }, { "epoch": 1.14, "learning_rate": 0.00014208093169413075, "loss": 1.8462, "step": 3796 }, { "epoch": 1.14, "learning_rate": 0.00014205144883737102, "loss": 1.9637, "step": 3797 }, { "epoch": 1.14, "learning_rate": 0.0001420219615394426, "loss": 1.7446, "step": 3798 }, { "epoch": 1.14, "learning_rate": 0.00014199246980345974, "loss": 1.81, "step": 3799 }, { "epoch": 1.14, "learning_rate": 0.00014196297363253715, "loss": 1.9032, "step": 3800 }, { "epoch": 1.14, "learning_rate": 0.00014193347302979, "loss": 1.8506, "step": 3801 }, { "epoch": 1.14, "learning_rate": 0.00014190396799833391, "loss": 2.0226, "step": 3802 }, { "epoch": 1.14, "learning_rate": 0.000141874458541285, "loss": 1.7711, "step": 3803 }, { "epoch": 1.15, "learning_rate": 0.00014184494466175986, "loss": 1.7656, "step": 3804 }, { "epoch": 1.15, "learning_rate": 0.00014181542636287548, "loss": 1.8059, "step": 3805 }, { "epoch": 1.15, "learning_rate": 0.00014178590364774944, "loss": 1.9177, "step": 3806 }, { "epoch": 1.15, "learning_rate": 0.00014175637651949968, "loss": 1.8026, "step": 3807 }, { "epoch": 1.15, "learning_rate": 0.00014172684498124462, "loss": 1.7487, "step": 3808 }, { "epoch": 1.15, "learning_rate": 0.00014169730903610322, "loss": 1.9144, "step": 3809 }, { "epoch": 1.15, "learning_rate": 0.00014166776868719479, "loss": 1.6893, "step": 3810 }, { "epoch": 1.15, "learning_rate": 0.0001416382239376392, "loss": 1.791, "step": 3811 }, { "epoch": 1.15, "learning_rate": 0.00014160867479055673, "loss": 1.7748, "step": 3812 }, { "epoch": 1.15, "learning_rate": 0.0001415791212490682, "loss": 1.8059, "step": 3813 }, { "epoch": 1.15, "learning_rate": 0.0001415495633162948, "loss": 1.7841, "step": 3814 }, { "epoch": 1.15, "learning_rate": 0.00014152000099535822, "loss": 1.894, "step": 3815 }, { "epoch": 1.15, "learning_rate": 0.00014149043428938064, "loss": 1.8967, "step": 3816 }, { "epoch": 1.15, "learning_rate": 0.00014146086320148465, "loss": 1.7956, "step": 3817 }, { "epoch": 1.15, "learning_rate": 0.00014143128773479334, "loss": 1.8057, "step": 3818 }, { "epoch": 1.15, "learning_rate": 0.00014140170789243029, "loss": 1.891, "step": 3819 }, { "epoch": 1.15, "learning_rate": 0.0001413721236775195, "loss": 1.8122, "step": 3820 }, { "epoch": 1.15, "learning_rate": 0.0001413425350931854, "loss": 1.8371, "step": 3821 }, { "epoch": 1.15, "learning_rate": 0.00014131294214255296, "loss": 1.9721, "step": 3822 }, { "epoch": 1.15, "learning_rate": 0.00014128334482874756, "loss": 1.8277, "step": 3823 }, { "epoch": 1.15, "learning_rate": 0.000141253743154895, "loss": 1.9512, "step": 3824 }, { "epoch": 1.15, "learning_rate": 0.00014122413712412172, "loss": 1.8718, "step": 3825 }, { "epoch": 1.15, "learning_rate": 0.00014119452673955438, "loss": 1.9084, "step": 3826 }, { "epoch": 1.15, "learning_rate": 0.00014116491200432028, "loss": 1.8779, "step": 3827 }, { "epoch": 1.15, "learning_rate": 0.0001411352929215471, "loss": 1.8292, "step": 3828 }, { "epoch": 1.15, "learning_rate": 0.00014110566949436295, "loss": 1.7998, "step": 3829 }, { "epoch": 1.15, "learning_rate": 0.0001410760417258965, "loss": 1.7746, "step": 3830 }, { "epoch": 1.15, "learning_rate": 0.00014104640961927673, "loss": 1.8374, "step": 3831 }, { "epoch": 1.15, "learning_rate": 0.0001410167731776333, "loss": 1.9531, "step": 3832 }, { "epoch": 1.15, "learning_rate": 0.00014098713240409607, "loss": 1.7797, "step": 3833 }, { "epoch": 1.15, "learning_rate": 0.00014095748730179558, "loss": 1.7942, "step": 3834 }, { "epoch": 1.15, "learning_rate": 0.00014092783787386269, "loss": 1.7084, "step": 3835 }, { "epoch": 1.15, "learning_rate": 0.0001408981841234287, "loss": 1.9359, "step": 3836 }, { "epoch": 1.16, "learning_rate": 0.00014086852605362554, "loss": 1.8218, "step": 3837 }, { "epoch": 1.16, "learning_rate": 0.00014083886366758537, "loss": 1.8174, "step": 3838 }, { "epoch": 1.16, "learning_rate": 0.00014080919696844103, "loss": 1.8433, "step": 3839 }, { "epoch": 1.16, "learning_rate": 0.0001407795259593256, "loss": 1.9858, "step": 3840 }, { "epoch": 1.16, "learning_rate": 0.00014074985064337276, "loss": 1.9025, "step": 3841 }, { "epoch": 1.16, "learning_rate": 0.00014072017102371659, "loss": 1.9499, "step": 3842 }, { "epoch": 1.16, "learning_rate": 0.0001406904871034916, "loss": 1.9193, "step": 3843 }, { "epoch": 1.16, "learning_rate": 0.0001406607988858329, "loss": 1.8019, "step": 3844 }, { "epoch": 1.16, "learning_rate": 0.00014063110637387586, "loss": 1.8168, "step": 3845 }, { "epoch": 1.16, "learning_rate": 0.00014060140957075637, "loss": 1.9909, "step": 3846 }, { "epoch": 1.16, "learning_rate": 0.00014057170847961084, "loss": 1.7463, "step": 3847 }, { "epoch": 1.16, "learning_rate": 0.00014054200310357608, "loss": 1.8137, "step": 3848 }, { "epoch": 1.16, "learning_rate": 0.00014051229344578934, "loss": 1.8748, "step": 3849 }, { "epoch": 1.16, "learning_rate": 0.00014048257950938833, "loss": 1.8163, "step": 3850 }, { "epoch": 1.16, "learning_rate": 0.00014045286129751126, "loss": 1.8358, "step": 3851 }, { "epoch": 1.16, "learning_rate": 0.00014042313881329667, "loss": 1.8266, "step": 3852 }, { "epoch": 1.16, "learning_rate": 0.00014039341205988371, "loss": 1.8409, "step": 3853 }, { "epoch": 1.16, "learning_rate": 0.0001403636810404119, "loss": 1.8455, "step": 3854 }, { "epoch": 1.16, "learning_rate": 0.00014033394575802116, "loss": 1.7894, "step": 3855 }, { "epoch": 1.16, "learning_rate": 0.000140304206215852, "loss": 1.9206, "step": 3856 }, { "epoch": 1.16, "learning_rate": 0.0001402744624170452, "loss": 1.7748, "step": 3857 }, { "epoch": 1.16, "learning_rate": 0.00014024471436474215, "loss": 1.8625, "step": 3858 }, { "epoch": 1.16, "learning_rate": 0.0001402149620620846, "loss": 1.9171, "step": 3859 }, { "epoch": 1.16, "learning_rate": 0.00014018520551221472, "loss": 1.8354, "step": 3860 }, { "epoch": 1.16, "learning_rate": 0.0001401554447182753, "loss": 1.8831, "step": 3861 }, { "epoch": 1.16, "learning_rate": 0.00014012567968340936, "loss": 1.943, "step": 3862 }, { "epoch": 1.16, "learning_rate": 0.0001400959104107605, "loss": 1.9522, "step": 3863 }, { "epoch": 1.16, "learning_rate": 0.00014006613690347274, "loss": 1.8319, "step": 3864 }, { "epoch": 1.16, "learning_rate": 0.00014003635916469057, "loss": 1.822, "step": 3865 }, { "epoch": 1.16, "learning_rate": 0.00014000657719755882, "loss": 1.8744, "step": 3866 }, { "epoch": 1.16, "learning_rate": 0.0001399767910052229, "loss": 1.821, "step": 3867 }, { "epoch": 1.16, "learning_rate": 0.0001399470005908286, "loss": 1.8156, "step": 3868 }, { "epoch": 1.16, "learning_rate": 0.00013991720595752215, "loss": 1.7916, "step": 3869 }, { "epoch": 1.16, "learning_rate": 0.0001398874071084503, "loss": 1.8469, "step": 3870 }, { "epoch": 1.17, "learning_rate": 0.0001398576040467601, "loss": 1.8307, "step": 3871 }, { "epoch": 1.17, "learning_rate": 0.00013982779677559922, "loss": 1.7938, "step": 3872 }, { "epoch": 1.17, "learning_rate": 0.00013979798529811566, "loss": 1.7579, "step": 3873 }, { "epoch": 1.17, "learning_rate": 0.00013976816961745785, "loss": 1.9003, "step": 3874 }, { "epoch": 1.17, "learning_rate": 0.00013973834973677476, "loss": 1.8505, "step": 3875 }, { "epoch": 1.17, "learning_rate": 0.0001397085256592157, "loss": 1.9162, "step": 3876 }, { "epoch": 1.17, "learning_rate": 0.00013967869738793054, "loss": 1.7851, "step": 3877 }, { "epoch": 1.17, "learning_rate": 0.00013964886492606943, "loss": 1.7476, "step": 3878 }, { "epoch": 1.17, "learning_rate": 0.00013961902827678313, "loss": 1.851, "step": 3879 }, { "epoch": 1.17, "learning_rate": 0.00013958918744322278, "loss": 1.7138, "step": 3880 }, { "epoch": 1.17, "learning_rate": 0.00013955934242853988, "loss": 1.801, "step": 3881 }, { "epoch": 1.17, "learning_rate": 0.00013952949323588654, "loss": 1.7835, "step": 3882 }, { "epoch": 1.17, "learning_rate": 0.0001394996398684151, "loss": 1.7811, "step": 3883 }, { "epoch": 1.17, "learning_rate": 0.00013946978232927854, "loss": 1.874, "step": 3884 }, { "epoch": 1.17, "learning_rate": 0.00013943992062163022, "loss": 1.9302, "step": 3885 }, { "epoch": 1.17, "learning_rate": 0.00013941005474862384, "loss": 1.7438, "step": 3886 }, { "epoch": 1.17, "learning_rate": 0.00013938018471341366, "loss": 1.9407, "step": 3887 }, { "epoch": 1.17, "learning_rate": 0.00013935031051915428, "loss": 1.7604, "step": 3888 }, { "epoch": 1.17, "learning_rate": 0.00013932043216900088, "loss": 1.8011, "step": 3889 }, { "epoch": 1.17, "learning_rate": 0.00013929054966610894, "loss": 1.9326, "step": 3890 }, { "epoch": 1.17, "learning_rate": 0.00013926066301363446, "loss": 1.8437, "step": 3891 }, { "epoch": 1.17, "learning_rate": 0.00013923077221473384, "loss": 1.7742, "step": 3892 }, { "epoch": 1.17, "learning_rate": 0.00013920087727256391, "loss": 1.7722, "step": 3893 }, { "epoch": 1.17, "learning_rate": 0.00013917097819028201, "loss": 1.8737, "step": 3894 }, { "epoch": 1.17, "learning_rate": 0.0001391410749710458, "loss": 1.9788, "step": 3895 }, { "epoch": 1.17, "learning_rate": 0.00013911116761801347, "loss": 1.7885, "step": 3896 }, { "epoch": 1.17, "learning_rate": 0.00013908125613434363, "loss": 1.8477, "step": 3897 }, { "epoch": 1.17, "learning_rate": 0.0001390513405231953, "loss": 1.8037, "step": 3898 }, { "epoch": 1.17, "learning_rate": 0.00013902142078772793, "loss": 1.9128, "step": 3899 }, { "epoch": 1.17, "learning_rate": 0.00013899149693110148, "loss": 1.8329, "step": 3900 }, { "epoch": 1.17, "learning_rate": 0.00013896156895647622, "loss": 1.8151, "step": 3901 }, { "epoch": 1.17, "learning_rate": 0.00013893163686701297, "loss": 1.7994, "step": 3902 }, { "epoch": 1.17, "learning_rate": 0.00013890170066587293, "loss": 1.7585, "step": 3903 }, { "epoch": 1.18, "learning_rate": 0.00013887176035621776, "loss": 1.7779, "step": 3904 }, { "epoch": 1.18, "learning_rate": 0.00013884181594120952, "loss": 1.7679, "step": 3905 }, { "epoch": 1.18, "learning_rate": 0.00013881186742401074, "loss": 1.858, "step": 3906 }, { "epoch": 1.18, "learning_rate": 0.00013878191480778433, "loss": 1.9091, "step": 3907 }, { "epoch": 1.18, "learning_rate": 0.00013875195809569368, "loss": 1.8414, "step": 3908 }, { "epoch": 1.18, "learning_rate": 0.0001387219972909026, "loss": 1.8742, "step": 3909 }, { "epoch": 1.18, "learning_rate": 0.00013869203239657538, "loss": 1.8262, "step": 3910 }, { "epoch": 1.18, "learning_rate": 0.00013866206341587664, "loss": 1.7281, "step": 3911 }, { "epoch": 1.18, "learning_rate": 0.00013863209035197148, "loss": 1.8661, "step": 3912 }, { "epoch": 1.18, "learning_rate": 0.00013860211320802549, "loss": 1.8028, "step": 3913 }, { "epoch": 1.18, "learning_rate": 0.0001385721319872046, "loss": 1.8081, "step": 3914 }, { "epoch": 1.18, "learning_rate": 0.0001385421466926752, "loss": 1.8379, "step": 3915 }, { "epoch": 1.18, "learning_rate": 0.00013851215732760413, "loss": 1.7758, "step": 3916 }, { "epoch": 1.18, "learning_rate": 0.0001384821638951587, "loss": 1.9354, "step": 3917 }, { "epoch": 1.18, "learning_rate": 0.00013845216639850653, "loss": 1.9002, "step": 3918 }, { "epoch": 1.18, "learning_rate": 0.00013842216484081572, "loss": 1.6971, "step": 3919 }, { "epoch": 1.18, "learning_rate": 0.0001383921592252549, "loss": 1.7795, "step": 3920 }, { "epoch": 1.18, "learning_rate": 0.00013836214955499298, "loss": 1.8481, "step": 3921 }, { "epoch": 1.18, "learning_rate": 0.00013833213583319941, "loss": 1.8762, "step": 3922 }, { "epoch": 1.18, "learning_rate": 0.000138302118063044, "loss": 1.7839, "step": 3923 }, { "epoch": 1.18, "learning_rate": 0.00013827209624769698, "loss": 1.8712, "step": 3924 }, { "epoch": 1.18, "learning_rate": 0.00013824207039032907, "loss": 1.8153, "step": 3925 }, { "epoch": 1.18, "learning_rate": 0.00013821204049411137, "loss": 1.8723, "step": 3926 }, { "epoch": 1.18, "learning_rate": 0.00013818200656221546, "loss": 1.8687, "step": 3927 }, { "epoch": 1.18, "learning_rate": 0.00013815196859781324, "loss": 1.797, "step": 3928 }, { "epoch": 1.18, "learning_rate": 0.00013812192660407713, "loss": 1.793, "step": 3929 }, { "epoch": 1.18, "learning_rate": 0.00013809188058417997, "loss": 1.8095, "step": 3930 }, { "epoch": 1.18, "learning_rate": 0.00013806183054129495, "loss": 1.8098, "step": 3931 }, { "epoch": 1.18, "learning_rate": 0.0001380317764785958, "loss": 1.8658, "step": 3932 }, { "epoch": 1.18, "learning_rate": 0.00013800171839925657, "loss": 1.7807, "step": 3933 }, { "epoch": 1.18, "learning_rate": 0.0001379716563064518, "loss": 1.8288, "step": 3934 }, { "epoch": 1.18, "learning_rate": 0.0001379415902033564, "loss": 1.8166, "step": 3935 }, { "epoch": 1.18, "learning_rate": 0.0001379115200931458, "loss": 1.8168, "step": 3936 }, { "epoch": 1.19, "learning_rate": 0.00013788144597899567, "loss": 1.7974, "step": 3937 }, { "epoch": 1.19, "learning_rate": 0.0001378513678640823, "loss": 1.798, "step": 3938 }, { "epoch": 1.19, "learning_rate": 0.00013782128575158233, "loss": 1.8448, "step": 3939 }, { "epoch": 1.19, "learning_rate": 0.00013779119964467276, "loss": 1.7427, "step": 3940 }, { "epoch": 1.19, "learning_rate": 0.00013776110954653113, "loss": 1.7928, "step": 3941 }, { "epoch": 1.19, "learning_rate": 0.0001377310154603353, "loss": 1.9717, "step": 3942 }, { "epoch": 1.19, "learning_rate": 0.0001377009173892636, "loss": 1.8132, "step": 3943 }, { "epoch": 1.19, "learning_rate": 0.00013767081533649477, "loss": 1.7649, "step": 3944 }, { "epoch": 1.19, "learning_rate": 0.00013764070930520793, "loss": 1.8878, "step": 3945 }, { "epoch": 1.19, "learning_rate": 0.00013761059929858274, "loss": 1.8207, "step": 3946 }, { "epoch": 1.19, "learning_rate": 0.00013758048531979915, "loss": 1.8935, "step": 3947 }, { "epoch": 1.19, "learning_rate": 0.00013755036737203759, "loss": 1.8377, "step": 3948 }, { "epoch": 1.19, "learning_rate": 0.00013752024545847887, "loss": 1.8267, "step": 3949 }, { "epoch": 1.19, "learning_rate": 0.00013749011958230428, "loss": 1.8667, "step": 3950 }, { "epoch": 1.19, "learning_rate": 0.00013745998974669553, "loss": 1.9424, "step": 3951 }, { "epoch": 1.19, "learning_rate": 0.00013742985595483463, "loss": 1.8557, "step": 3952 }, { "epoch": 1.19, "learning_rate": 0.00013739971820990414, "loss": 1.8452, "step": 3953 }, { "epoch": 1.19, "learning_rate": 0.000137369576515087, "loss": 1.8732, "step": 3954 }, { "epoch": 1.19, "learning_rate": 0.00013733943087356653, "loss": 1.9132, "step": 3955 }, { "epoch": 1.19, "learning_rate": 0.00013730928128852652, "loss": 1.8322, "step": 3956 }, { "epoch": 1.19, "learning_rate": 0.0001372791277631511, "loss": 1.943, "step": 3957 }, { "epoch": 1.19, "learning_rate": 0.00013724897030062495, "loss": 1.78, "step": 3958 }, { "epoch": 1.19, "learning_rate": 0.00013721880890413303, "loss": 1.9219, "step": 3959 }, { "epoch": 1.19, "learning_rate": 0.00013718864357686078, "loss": 1.7697, "step": 3960 }, { "epoch": 1.19, "learning_rate": 0.000137158474321994, "loss": 1.8204, "step": 3961 }, { "epoch": 1.19, "learning_rate": 0.000137128301142719, "loss": 1.8908, "step": 3962 }, { "epoch": 1.19, "learning_rate": 0.00013709812404222245, "loss": 1.8004, "step": 3963 }, { "epoch": 1.19, "learning_rate": 0.00013706794302369138, "loss": 1.732, "step": 3964 }, { "epoch": 1.19, "learning_rate": 0.00013703775809031338, "loss": 1.7853, "step": 3965 }, { "epoch": 1.19, "learning_rate": 0.0001370075692452763, "loss": 1.8404, "step": 3966 }, { "epoch": 1.19, "learning_rate": 0.00013697737649176847, "loss": 1.6927, "step": 3967 }, { "epoch": 1.19, "learning_rate": 0.00013694717983297867, "loss": 1.8137, "step": 3968 }, { "epoch": 1.19, "learning_rate": 0.00013691697927209598, "loss": 1.8317, "step": 3969 }, { "epoch": 1.2, "learning_rate": 0.00013688677481231006, "loss": 1.9058, "step": 3970 }, { "epoch": 1.2, "learning_rate": 0.0001368565664568108, "loss": 1.8678, "step": 3971 }, { "epoch": 1.2, "learning_rate": 0.00013682635420878862, "loss": 1.805, "step": 3972 }, { "epoch": 1.2, "learning_rate": 0.0001367961380714343, "loss": 1.7617, "step": 3973 }, { "epoch": 1.2, "learning_rate": 0.0001367659180479391, "loss": 1.8011, "step": 3974 }, { "epoch": 1.2, "learning_rate": 0.0001367356941414946, "loss": 1.8243, "step": 3975 }, { "epoch": 1.2, "learning_rate": 0.00013670546635529286, "loss": 1.7864, "step": 3976 }, { "epoch": 1.2, "learning_rate": 0.00013667523469252627, "loss": 1.8929, "step": 3977 }, { "epoch": 1.2, "learning_rate": 0.0001366449991563877, "loss": 1.8279, "step": 3978 }, { "epoch": 1.2, "learning_rate": 0.00013661475975007038, "loss": 1.8116, "step": 3979 }, { "epoch": 1.2, "learning_rate": 0.00013658451647676807, "loss": 1.8047, "step": 3980 }, { "epoch": 1.2, "learning_rate": 0.00013655426933967477, "loss": 1.7246, "step": 3981 }, { "epoch": 1.2, "learning_rate": 0.00013652401834198495, "loss": 1.8402, "step": 3982 }, { "epoch": 1.2, "learning_rate": 0.00013649376348689356, "loss": 1.8505, "step": 3983 }, { "epoch": 1.2, "learning_rate": 0.00013646350477759583, "loss": 1.8649, "step": 3984 }, { "epoch": 1.2, "learning_rate": 0.0001364332422172875, "loss": 1.791, "step": 3985 }, { "epoch": 1.2, "learning_rate": 0.0001364029758091647, "loss": 1.8201, "step": 3986 }, { "epoch": 1.2, "learning_rate": 0.00013637270555642394, "loss": 1.7115, "step": 3987 }, { "epoch": 1.2, "learning_rate": 0.00013634243146226215, "loss": 1.8706, "step": 3988 }, { "epoch": 1.2, "learning_rate": 0.00013631215352987664, "loss": 1.8997, "step": 3989 }, { "epoch": 1.2, "learning_rate": 0.00013628187176246512, "loss": 1.8151, "step": 3990 }, { "epoch": 1.2, "learning_rate": 0.00013625158616322575, "loss": 1.7533, "step": 3991 }, { "epoch": 1.2, "learning_rate": 0.0001362212967353571, "loss": 1.8424, "step": 3992 }, { "epoch": 1.2, "learning_rate": 0.00013619100348205815, "loss": 1.8894, "step": 3993 }, { "epoch": 1.2, "learning_rate": 0.00013616070640652813, "loss": 1.8953, "step": 3994 }, { "epoch": 1.2, "learning_rate": 0.00013613040551196693, "loss": 1.8734, "step": 3995 }, { "epoch": 1.2, "learning_rate": 0.00013610010080157466, "loss": 1.8154, "step": 3996 }, { "epoch": 1.2, "learning_rate": 0.00013606979227855185, "loss": 1.8293, "step": 3997 }, { "epoch": 1.2, "learning_rate": 0.0001360394799460995, "loss": 1.8641, "step": 3998 }, { "epoch": 1.2, "learning_rate": 0.000136009163807419, "loss": 1.7618, "step": 3999 }, { "epoch": 1.2, "learning_rate": 0.0001359788438657121, "loss": 1.8072, "step": 4000 }, { "epoch": 1.2, "learning_rate": 0.00013594852012418094, "loss": 1.8967, "step": 4001 }, { "epoch": 1.2, "learning_rate": 0.00013591819258602813, "loss": 1.8351, "step": 4002 }, { "epoch": 1.2, "learning_rate": 0.00013588786125445662, "loss": 1.9176, "step": 4003 }, { "epoch": 1.21, "learning_rate": 0.0001358575261326698, "loss": 1.7965, "step": 4004 }, { "epoch": 1.21, "learning_rate": 0.00013582718722387146, "loss": 1.8599, "step": 4005 }, { "epoch": 1.21, "learning_rate": 0.00013579684453126576, "loss": 1.8696, "step": 4006 }, { "epoch": 1.21, "learning_rate": 0.00013576649805805728, "loss": 1.7239, "step": 4007 }, { "epoch": 1.21, "learning_rate": 0.00013573614780745097, "loss": 1.8678, "step": 4008 }, { "epoch": 1.21, "learning_rate": 0.00013570579378265223, "loss": 1.8426, "step": 4009 }, { "epoch": 1.21, "learning_rate": 0.00013567543598686684, "loss": 1.7324, "step": 4010 }, { "epoch": 1.21, "learning_rate": 0.00013564507442330093, "loss": 1.7681, "step": 4011 }, { "epoch": 1.21, "learning_rate": 0.0001356147090951611, "loss": 1.818, "step": 4012 }, { "epoch": 1.21, "learning_rate": 0.0001355843400056543, "loss": 1.8978, "step": 4013 }, { "epoch": 1.21, "learning_rate": 0.00013555396715798793, "loss": 1.7158, "step": 4014 }, { "epoch": 1.21, "learning_rate": 0.00013552359055536969, "loss": 1.8745, "step": 4015 }, { "epoch": 1.21, "learning_rate": 0.00013549321020100776, "loss": 1.8212, "step": 4016 }, { "epoch": 1.21, "learning_rate": 0.00013546282609811072, "loss": 1.8093, "step": 4017 }, { "epoch": 1.21, "learning_rate": 0.00013543243824988747, "loss": 1.7578, "step": 4018 }, { "epoch": 1.21, "learning_rate": 0.00013540204665954742, "loss": 1.8229, "step": 4019 }, { "epoch": 1.21, "learning_rate": 0.00013537165133030023, "loss": 1.7733, "step": 4020 }, { "epoch": 1.21, "learning_rate": 0.0001353412522653561, "loss": 1.7214, "step": 4021 }, { "epoch": 1.21, "learning_rate": 0.0001353108494679255, "loss": 1.7824, "step": 4022 }, { "epoch": 1.21, "learning_rate": 0.00013528044294121942, "loss": 1.8405, "step": 4023 }, { "epoch": 1.21, "learning_rate": 0.00013525003268844912, "loss": 1.7807, "step": 4024 }, { "epoch": 1.21, "learning_rate": 0.00013521961871282637, "loss": 1.7758, "step": 4025 }, { "epoch": 1.21, "learning_rate": 0.0001351892010175632, "loss": 1.8379, "step": 4026 }, { "epoch": 1.21, "learning_rate": 0.00013515877960587212, "loss": 1.7518, "step": 4027 }, { "epoch": 1.21, "learning_rate": 0.00013512835448096605, "loss": 1.7754, "step": 4028 }, { "epoch": 1.21, "learning_rate": 0.0001350979256460583, "loss": 1.7906, "step": 4029 }, { "epoch": 1.21, "learning_rate": 0.00013506749310436246, "loss": 1.8559, "step": 4030 }, { "epoch": 1.21, "learning_rate": 0.00013503705685909267, "loss": 1.8612, "step": 4031 }, { "epoch": 1.21, "learning_rate": 0.00013500661691346334, "loss": 1.8895, "step": 4032 }, { "epoch": 1.21, "learning_rate": 0.0001349761732706893, "loss": 1.7905, "step": 4033 }, { "epoch": 1.21, "learning_rate": 0.00013494572593398583, "loss": 1.9652, "step": 4034 }, { "epoch": 1.21, "learning_rate": 0.00013491527490656856, "loss": 1.8165, "step": 4035 }, { "epoch": 1.21, "learning_rate": 0.00013488482019165346, "loss": 1.8061, "step": 4036 }, { "epoch": 1.22, "learning_rate": 0.00013485436179245698, "loss": 1.7787, "step": 4037 }, { "epoch": 1.22, "learning_rate": 0.0001348238997121959, "loss": 1.8638, "step": 4038 }, { "epoch": 1.22, "learning_rate": 0.00013479343395408742, "loss": 1.7564, "step": 4039 }, { "epoch": 1.22, "learning_rate": 0.00013476296452134904, "loss": 1.7611, "step": 4040 }, { "epoch": 1.22, "learning_rate": 0.0001347324914171988, "loss": 1.7953, "step": 4041 }, { "epoch": 1.22, "learning_rate": 0.00013470201464485504, "loss": 1.6384, "step": 4042 }, { "epoch": 1.22, "learning_rate": 0.00013467153420753647, "loss": 1.8986, "step": 4043 }, { "epoch": 1.22, "learning_rate": 0.0001346410501084622, "loss": 1.8903, "step": 4044 }, { "epoch": 1.22, "learning_rate": 0.0001346105623508518, "loss": 1.8574, "step": 4045 }, { "epoch": 1.22, "learning_rate": 0.00013458007093792508, "loss": 1.8797, "step": 4046 }, { "epoch": 1.22, "learning_rate": 0.00013454957587290237, "loss": 1.9314, "step": 4047 }, { "epoch": 1.22, "learning_rate": 0.00013451907715900438, "loss": 1.8648, "step": 4048 }, { "epoch": 1.22, "learning_rate": 0.00013448857479945209, "loss": 1.8092, "step": 4049 }, { "epoch": 1.22, "learning_rate": 0.00013445806879746694, "loss": 1.8226, "step": 4050 }, { "epoch": 1.22, "learning_rate": 0.0001344275591562708, "loss": 1.7481, "step": 4051 }, { "epoch": 1.22, "learning_rate": 0.00013439704587908584, "loss": 1.7925, "step": 4052 }, { "epoch": 1.22, "learning_rate": 0.00013436652896913468, "loss": 1.8743, "step": 4053 }, { "epoch": 1.22, "learning_rate": 0.00013433600842964027, "loss": 1.815, "step": 4054 }, { "epoch": 1.22, "learning_rate": 0.000134305484263826, "loss": 1.7015, "step": 4055 }, { "epoch": 1.22, "learning_rate": 0.00013427495647491553, "loss": 1.8423, "step": 4056 }, { "epoch": 1.22, "learning_rate": 0.0001342444250661331, "loss": 1.7544, "step": 4057 }, { "epoch": 1.22, "learning_rate": 0.00013421389004070314, "loss": 1.9419, "step": 4058 }, { "epoch": 1.22, "learning_rate": 0.00013418335140185053, "loss": 1.8748, "step": 4059 }, { "epoch": 1.22, "learning_rate": 0.0001341528091528006, "loss": 1.9672, "step": 4060 }, { "epoch": 1.22, "learning_rate": 0.00013412226329677894, "loss": 1.7727, "step": 4061 }, { "epoch": 1.22, "learning_rate": 0.00013409171383701158, "loss": 1.9093, "step": 4062 }, { "epoch": 1.22, "learning_rate": 0.00013406116077672496, "loss": 1.8374, "step": 4063 }, { "epoch": 1.22, "learning_rate": 0.00013403060411914592, "loss": 1.8277, "step": 4064 }, { "epoch": 1.22, "learning_rate": 0.00013400004386750157, "loss": 1.8794, "step": 4065 }, { "epoch": 1.22, "learning_rate": 0.00013396948002501947, "loss": 1.8629, "step": 4066 }, { "epoch": 1.22, "learning_rate": 0.00013393891259492753, "loss": 1.8789, "step": 4067 }, { "epoch": 1.22, "learning_rate": 0.00013390834158045407, "loss": 1.7697, "step": 4068 }, { "epoch": 1.22, "learning_rate": 0.00013387776698482783, "loss": 1.8331, "step": 4069 }, { "epoch": 1.23, "learning_rate": 0.00013384718881127782, "loss": 1.7726, "step": 4070 }, { "epoch": 1.23, "learning_rate": 0.00013381660706303356, "loss": 1.8973, "step": 4071 }, { "epoch": 1.23, "learning_rate": 0.00013378602174332477, "loss": 1.8136, "step": 4072 }, { "epoch": 1.23, "learning_rate": 0.00013375543285538171, "loss": 1.8604, "step": 4073 }, { "epoch": 1.23, "learning_rate": 0.00013372484040243492, "loss": 1.9118, "step": 4074 }, { "epoch": 1.23, "learning_rate": 0.00013369424438771537, "loss": 1.8101, "step": 4075 }, { "epoch": 1.23, "learning_rate": 0.00013366364481445445, "loss": 1.8558, "step": 4076 }, { "epoch": 1.23, "learning_rate": 0.00013363304168588375, "loss": 1.9362, "step": 4077 }, { "epoch": 1.23, "learning_rate": 0.00013360243500523545, "loss": 1.7602, "step": 4078 }, { "epoch": 1.23, "learning_rate": 0.00013357182477574195, "loss": 1.7724, "step": 4079 }, { "epoch": 1.23, "learning_rate": 0.00013354121100063606, "loss": 1.8486, "step": 4080 }, { "epoch": 1.23, "learning_rate": 0.00013351059368315108, "loss": 1.7594, "step": 4081 }, { "epoch": 1.23, "learning_rate": 0.00013347997282652047, "loss": 1.847, "step": 4082 }, { "epoch": 1.23, "learning_rate": 0.0001334493484339783, "loss": 1.8581, "step": 4083 }, { "epoch": 1.23, "learning_rate": 0.00013341872050875878, "loss": 1.8378, "step": 4084 }, { "epoch": 1.23, "learning_rate": 0.00013338808905409672, "loss": 1.7743, "step": 4085 }, { "epoch": 1.23, "learning_rate": 0.00013335745407322712, "loss": 1.7276, "step": 4086 }, { "epoch": 1.23, "learning_rate": 0.0001333268155693854, "loss": 1.7751, "step": 4087 }, { "epoch": 1.23, "learning_rate": 0.00013329617354580745, "loss": 1.8571, "step": 4088 }, { "epoch": 1.23, "learning_rate": 0.00013326552800572942, "loss": 1.7784, "step": 4089 }, { "epoch": 1.23, "learning_rate": 0.0001332348789523879, "loss": 1.8095, "step": 4090 }, { "epoch": 1.23, "learning_rate": 0.00013320422638901977, "loss": 1.9154, "step": 4091 }, { "epoch": 1.23, "learning_rate": 0.00013317357031886236, "loss": 1.8022, "step": 4092 }, { "epoch": 1.23, "learning_rate": 0.00013314291074515336, "loss": 1.712, "step": 4093 }, { "epoch": 1.23, "learning_rate": 0.00013311224767113076, "loss": 1.8354, "step": 4094 }, { "epoch": 1.23, "learning_rate": 0.00013308158110003304, "loss": 1.8945, "step": 4095 }, { "epoch": 1.23, "learning_rate": 0.0001330509110350989, "loss": 1.8127, "step": 4096 }, { "epoch": 1.23, "learning_rate": 0.00013302023747956756, "loss": 1.7786, "step": 4097 }, { "epoch": 1.23, "learning_rate": 0.00013298956043667848, "loss": 1.8322, "step": 4098 }, { "epoch": 1.23, "learning_rate": 0.0001329588799096716, "loss": 1.7062, "step": 4099 }, { "epoch": 1.23, "learning_rate": 0.00013292819590178713, "loss": 1.967, "step": 4100 }, { "epoch": 1.23, "learning_rate": 0.00013289750841626572, "loss": 1.7901, "step": 4101 }, { "epoch": 1.23, "learning_rate": 0.0001328668174563484, "loss": 1.7951, "step": 4102 }, { "epoch": 1.24, "learning_rate": 0.0001328361230252764, "loss": 1.8088, "step": 4103 }, { "epoch": 1.24, "learning_rate": 0.00013280542512629156, "loss": 1.7598, "step": 4104 }, { "epoch": 1.24, "learning_rate": 0.00013277472376263595, "loss": 1.6656, "step": 4105 }, { "epoch": 1.24, "learning_rate": 0.000132744018937552, "loss": 1.8577, "step": 4106 }, { "epoch": 1.24, "learning_rate": 0.00013271331065428253, "loss": 1.8999, "step": 4107 }, { "epoch": 1.24, "learning_rate": 0.00013268259891607072, "loss": 1.7737, "step": 4108 }, { "epoch": 1.24, "learning_rate": 0.00013265188372616014, "loss": 1.8368, "step": 4109 }, { "epoch": 1.24, "learning_rate": 0.0001326211650877947, "loss": 1.8725, "step": 4110 }, { "epoch": 1.24, "learning_rate": 0.00013259044300421865, "loss": 1.7832, "step": 4111 }, { "epoch": 1.24, "learning_rate": 0.0001325597174786767, "loss": 1.8593, "step": 4112 }, { "epoch": 1.24, "learning_rate": 0.0001325289885144138, "loss": 1.7861, "step": 4113 }, { "epoch": 1.24, "learning_rate": 0.00013249825611467536, "loss": 1.8115, "step": 4114 }, { "epoch": 1.24, "learning_rate": 0.00013246752028270707, "loss": 1.8435, "step": 4115 }, { "epoch": 1.24, "learning_rate": 0.00013243678102175502, "loss": 1.7856, "step": 4116 }, { "epoch": 1.24, "learning_rate": 0.00013240603833506573, "loss": 1.7608, "step": 4117 }, { "epoch": 1.24, "learning_rate": 0.000132375292225886, "loss": 1.8671, "step": 4118 }, { "epoch": 1.24, "learning_rate": 0.00013234454269746298, "loss": 1.6952, "step": 4119 }, { "epoch": 1.24, "learning_rate": 0.00013231378975304422, "loss": 1.8557, "step": 4120 }, { "epoch": 1.24, "learning_rate": 0.0001322830333958776, "loss": 1.7201, "step": 4121 }, { "epoch": 1.24, "learning_rate": 0.00013225227362921143, "loss": 1.8314, "step": 4122 }, { "epoch": 1.24, "learning_rate": 0.00013222151045629433, "loss": 1.8874, "step": 4123 }, { "epoch": 1.24, "learning_rate": 0.0001321907438803753, "loss": 1.8579, "step": 4124 }, { "epoch": 1.24, "learning_rate": 0.0001321599739047036, "loss": 1.8582, "step": 4125 }, { "epoch": 1.24, "learning_rate": 0.00013212920053252899, "loss": 1.8782, "step": 4126 }, { "epoch": 1.24, "learning_rate": 0.0001320984237671015, "loss": 1.8399, "step": 4127 }, { "epoch": 1.24, "learning_rate": 0.00013206764361167162, "loss": 1.9394, "step": 4128 }, { "epoch": 1.24, "learning_rate": 0.00013203686006949005, "loss": 1.8336, "step": 4129 }, { "epoch": 1.24, "learning_rate": 0.00013200607314380796, "loss": 1.8634, "step": 4130 }, { "epoch": 1.24, "learning_rate": 0.00013197528283787684, "loss": 1.7494, "step": 4131 }, { "epoch": 1.24, "learning_rate": 0.0001319444891549485, "loss": 1.8039, "step": 4132 }, { "epoch": 1.24, "learning_rate": 0.0001319136920982752, "loss": 1.7997, "step": 4133 }, { "epoch": 1.24, "learning_rate": 0.00013188289167110945, "loss": 1.7603, "step": 4134 }, { "epoch": 1.24, "learning_rate": 0.00013185208787670422, "loss": 1.8172, "step": 4135 }, { "epoch": 1.25, "learning_rate": 0.00013182128071831278, "loss": 1.866, "step": 4136 }, { "epoch": 1.25, "learning_rate": 0.0001317904701991887, "loss": 1.7853, "step": 4137 }, { "epoch": 1.25, "learning_rate": 0.00013175965632258602, "loss": 1.8052, "step": 4138 }, { "epoch": 1.25, "learning_rate": 0.00013172883909175906, "loss": 1.8097, "step": 4139 }, { "epoch": 1.25, "learning_rate": 0.0001316980185099625, "loss": 1.8848, "step": 4140 }, { "epoch": 1.25, "learning_rate": 0.0001316671945804514, "loss": 1.8487, "step": 4141 }, { "epoch": 1.25, "learning_rate": 0.00013163636730648115, "loss": 1.8398, "step": 4142 }, { "epoch": 1.25, "learning_rate": 0.0001316055366913075, "loss": 1.8078, "step": 4143 }, { "epoch": 1.25, "learning_rate": 0.00013157470273818657, "loss": 1.8294, "step": 4144 }, { "epoch": 1.25, "learning_rate": 0.00013154386545037483, "loss": 1.8281, "step": 4145 }, { "epoch": 1.25, "learning_rate": 0.00013151302483112905, "loss": 1.7639, "step": 4146 }, { "epoch": 1.25, "learning_rate": 0.00013148218088370644, "loss": 1.8991, "step": 4147 }, { "epoch": 1.25, "learning_rate": 0.00013145133361136448, "loss": 1.9059, "step": 4148 }, { "epoch": 1.25, "learning_rate": 0.00013142048301736104, "loss": 1.905, "step": 4149 }, { "epoch": 1.25, "learning_rate": 0.00013138962910495437, "loss": 1.7902, "step": 4150 }, { "epoch": 1.25, "learning_rate": 0.00013135877187740293, "loss": 1.762, "step": 4151 }, { "epoch": 1.25, "learning_rate": 0.00013132791133796577, "loss": 1.8168, "step": 4152 }, { "epoch": 1.25, "learning_rate": 0.0001312970474899021, "loss": 1.8399, "step": 4153 }, { "epoch": 1.25, "learning_rate": 0.0001312661803364715, "loss": 1.9155, "step": 4154 }, { "epoch": 1.25, "learning_rate": 0.000131235309880934, "loss": 1.7919, "step": 4155 }, { "epoch": 1.25, "learning_rate": 0.00013120443612654986, "loss": 1.7745, "step": 4156 }, { "epoch": 1.25, "learning_rate": 0.0001311735590765798, "loss": 1.7746, "step": 4157 }, { "epoch": 1.25, "learning_rate": 0.00013114267873428473, "loss": 1.7718, "step": 4158 }, { "epoch": 1.25, "learning_rate": 0.00013111179510292612, "loss": 1.8342, "step": 4159 }, { "epoch": 1.25, "learning_rate": 0.0001310809081857656, "loss": 1.8324, "step": 4160 }, { "epoch": 1.25, "learning_rate": 0.0001310500179860653, "loss": 1.8088, "step": 4161 }, { "epoch": 1.25, "learning_rate": 0.00013101912450708754, "loss": 1.7945, "step": 4162 }, { "epoch": 1.25, "learning_rate": 0.00013098822775209503, "loss": 1.8838, "step": 4163 }, { "epoch": 1.25, "learning_rate": 0.00013095732772435098, "loss": 1.8407, "step": 4164 }, { "epoch": 1.25, "learning_rate": 0.00013092642442711876, "loss": 1.8189, "step": 4165 }, { "epoch": 1.25, "learning_rate": 0.00013089551786366218, "loss": 1.8141, "step": 4166 }, { "epoch": 1.25, "learning_rate": 0.0001308646080372453, "loss": 1.6985, "step": 4167 }, { "epoch": 1.25, "learning_rate": 0.00013083369495113268, "loss": 1.7893, "step": 4168 }, { "epoch": 1.25, "learning_rate": 0.00013080277860858907, "loss": 1.817, "step": 4169 }, { "epoch": 1.26, "learning_rate": 0.00013077185901287964, "loss": 1.8387, "step": 4170 }, { "epoch": 1.26, "learning_rate": 0.00013074093616726994, "loss": 1.7757, "step": 4171 }, { "epoch": 1.26, "learning_rate": 0.00013071001007502578, "loss": 1.7848, "step": 4172 }, { "epoch": 1.26, "learning_rate": 0.00013067908073941334, "loss": 1.7798, "step": 4173 }, { "epoch": 1.26, "learning_rate": 0.00013064814816369916, "loss": 1.8274, "step": 4174 }, { "epoch": 1.26, "learning_rate": 0.00013061721235115012, "loss": 1.8926, "step": 4175 }, { "epoch": 1.26, "learning_rate": 0.00013058627330503347, "loss": 1.7703, "step": 4176 }, { "epoch": 1.26, "learning_rate": 0.0001305553310286167, "loss": 1.8133, "step": 4177 }, { "epoch": 1.26, "learning_rate": 0.00013052438552516775, "loss": 1.9399, "step": 4178 }, { "epoch": 1.26, "learning_rate": 0.00013049343679795485, "loss": 1.9266, "step": 4179 }, { "epoch": 1.26, "learning_rate": 0.00013046248485024657, "loss": 1.7698, "step": 4180 }, { "epoch": 1.26, "learning_rate": 0.00013043152968531186, "loss": 1.8087, "step": 4181 }, { "epoch": 1.26, "learning_rate": 0.00013040057130641992, "loss": 1.8323, "step": 4182 }, { "epoch": 1.26, "learning_rate": 0.00013036960971684044, "loss": 1.8527, "step": 4183 }, { "epoch": 1.26, "learning_rate": 0.00013033864491984325, "loss": 1.8194, "step": 4184 }, { "epoch": 1.26, "learning_rate": 0.00013030767691869875, "loss": 1.7679, "step": 4185 }, { "epoch": 1.26, "learning_rate": 0.00013027670571667742, "loss": 1.8288, "step": 4186 }, { "epoch": 1.26, "learning_rate": 0.00013024573131705032, "loss": 1.7778, "step": 4187 }, { "epoch": 1.26, "learning_rate": 0.00013021475372308871, "loss": 1.8135, "step": 4188 }, { "epoch": 1.26, "learning_rate": 0.0001301837729380642, "loss": 1.8466, "step": 4189 }, { "epoch": 1.26, "learning_rate": 0.0001301527889652488, "loss": 1.8337, "step": 4190 }, { "epoch": 1.26, "learning_rate": 0.0001301218018079147, "loss": 1.7583, "step": 4191 }, { "epoch": 1.26, "learning_rate": 0.0001300908114693347, "loss": 1.9586, "step": 4192 }, { "epoch": 1.26, "learning_rate": 0.00013005981795278165, "loss": 1.7664, "step": 4193 }, { "epoch": 1.26, "learning_rate": 0.0001300288212615289, "loss": 1.8965, "step": 4194 }, { "epoch": 1.26, "learning_rate": 0.00012999782139885014, "loss": 1.7603, "step": 4195 }, { "epoch": 1.26, "learning_rate": 0.00012996681836801926, "loss": 1.7274, "step": 4196 }, { "epoch": 1.26, "learning_rate": 0.00012993581217231066, "loss": 1.8062, "step": 4197 }, { "epoch": 1.26, "learning_rate": 0.00012990480281499894, "loss": 1.755, "step": 4198 }, { "epoch": 1.26, "learning_rate": 0.0001298737902993591, "loss": 1.8412, "step": 4199 }, { "epoch": 1.26, "learning_rate": 0.00012984277462866646, "loss": 1.6848, "step": 4200 }, { "epoch": 1.26, "learning_rate": 0.00012981175580619665, "loss": 1.9034, "step": 4201 }, { "epoch": 1.26, "learning_rate": 0.00012978073383522565, "loss": 1.8366, "step": 4202 }, { "epoch": 1.27, "learning_rate": 0.00012974970871902983, "loss": 1.7346, "step": 4203 }, { "epoch": 1.27, "learning_rate": 0.0001297186804608858, "loss": 1.7737, "step": 4204 }, { "epoch": 1.27, "learning_rate": 0.00012968764906407052, "loss": 1.7791, "step": 4205 }, { "epoch": 1.27, "learning_rate": 0.0001296566145318613, "loss": 1.8207, "step": 4206 }, { "epoch": 1.27, "learning_rate": 0.00012962557686753583, "loss": 1.8091, "step": 4207 }, { "epoch": 1.27, "learning_rate": 0.00012959453607437203, "loss": 1.8132, "step": 4208 }, { "epoch": 1.27, "learning_rate": 0.00012956349215564826, "loss": 1.8171, "step": 4209 }, { "epoch": 1.27, "learning_rate": 0.00012953244511464308, "loss": 1.7921, "step": 4210 }, { "epoch": 1.27, "learning_rate": 0.0001295013949546355, "loss": 1.7995, "step": 4211 }, { "epoch": 1.27, "learning_rate": 0.00012947034167890483, "loss": 1.8153, "step": 4212 }, { "epoch": 1.27, "learning_rate": 0.00012943928529073067, "loss": 1.836, "step": 4213 }, { "epoch": 1.27, "learning_rate": 0.00012940822579339295, "loss": 1.7653, "step": 4214 }, { "epoch": 1.27, "learning_rate": 0.00012937716319017198, "loss": 1.775, "step": 4215 }, { "epoch": 1.27, "learning_rate": 0.00012934609748434833, "loss": 1.8985, "step": 4216 }, { "epoch": 1.27, "learning_rate": 0.00012931502867920296, "loss": 1.8952, "step": 4217 }, { "epoch": 1.27, "learning_rate": 0.0001292839567780171, "loss": 1.8598, "step": 4218 }, { "epoch": 1.27, "learning_rate": 0.00012925288178407244, "loss": 1.7888, "step": 4219 }, { "epoch": 1.27, "learning_rate": 0.0001292218037006508, "loss": 1.8891, "step": 4220 }, { "epoch": 1.27, "learning_rate": 0.00012919072253103443, "loss": 1.739, "step": 4221 }, { "epoch": 1.27, "learning_rate": 0.0001291596382785059, "loss": 1.833, "step": 4222 }, { "epoch": 1.27, "learning_rate": 0.00012912855094634811, "loss": 1.8857, "step": 4223 }, { "epoch": 1.27, "learning_rate": 0.00012909746053784427, "loss": 1.8775, "step": 4224 }, { "epoch": 1.27, "learning_rate": 0.000129066367056278, "loss": 1.8532, "step": 4225 }, { "epoch": 1.27, "learning_rate": 0.0001290352705049331, "loss": 1.787, "step": 4226 }, { "epoch": 1.27, "learning_rate": 0.0001290041708870937, "loss": 1.8489, "step": 4227 }, { "epoch": 1.27, "learning_rate": 0.0001289730682060444, "loss": 1.8329, "step": 4228 }, { "epoch": 1.27, "learning_rate": 0.00012894196246507003, "loss": 1.7477, "step": 4229 }, { "epoch": 1.27, "learning_rate": 0.00012891085366745575, "loss": 1.7602, "step": 4230 }, { "epoch": 1.27, "learning_rate": 0.00012887974181648702, "loss": 1.788, "step": 4231 }, { "epoch": 1.27, "learning_rate": 0.0001288486269154497, "loss": 1.7547, "step": 4232 }, { "epoch": 1.27, "learning_rate": 0.00012881750896762988, "loss": 1.8325, "step": 4233 }, { "epoch": 1.27, "learning_rate": 0.00012878638797631398, "loss": 1.8484, "step": 4234 }, { "epoch": 1.27, "learning_rate": 0.00012875526394478883, "loss": 1.7631, "step": 4235 }, { "epoch": 1.28, "learning_rate": 0.00012872413687634152, "loss": 1.8124, "step": 4236 }, { "epoch": 1.28, "learning_rate": 0.00012869300677425944, "loss": 1.9715, "step": 4237 }, { "epoch": 1.28, "learning_rate": 0.00012866187364183034, "loss": 1.883, "step": 4238 }, { "epoch": 1.28, "learning_rate": 0.00012863073748234228, "loss": 1.8424, "step": 4239 }, { "epoch": 1.28, "learning_rate": 0.0001285995982990836, "loss": 1.9207, "step": 4240 }, { "epoch": 1.28, "learning_rate": 0.000128568456095343, "loss": 1.8437, "step": 4241 }, { "epoch": 1.28, "learning_rate": 0.00012853731087440957, "loss": 1.8806, "step": 4242 }, { "epoch": 1.28, "learning_rate": 0.00012850616263957254, "loss": 1.8402, "step": 4243 }, { "epoch": 1.28, "learning_rate": 0.00012847501139412164, "loss": 1.7771, "step": 4244 }, { "epoch": 1.28, "learning_rate": 0.00012844385714134678, "loss": 1.9133, "step": 4245 }, { "epoch": 1.28, "learning_rate": 0.0001284126998845383, "loss": 1.8318, "step": 4246 }, { "epoch": 1.28, "learning_rate": 0.00012838153962698673, "loss": 1.8404, "step": 4247 }, { "epoch": 1.28, "learning_rate": 0.00012835037637198306, "loss": 1.823, "step": 4248 }, { "epoch": 1.28, "learning_rate": 0.0001283192101228185, "loss": 1.8811, "step": 4249 }, { "epoch": 1.28, "learning_rate": 0.0001282880408827846, "loss": 1.8374, "step": 4250 }, { "epoch": 1.28, "learning_rate": 0.00012825686865517322, "loss": 1.7838, "step": 4251 }, { "epoch": 1.28, "learning_rate": 0.00012822569344327658, "loss": 1.8743, "step": 4252 }, { "epoch": 1.28, "learning_rate": 0.00012819451525038713, "loss": 1.8317, "step": 4253 }, { "epoch": 1.28, "learning_rate": 0.00012816333407979775, "loss": 1.8056, "step": 4254 }, { "epoch": 1.28, "learning_rate": 0.00012813214993480146, "loss": 1.7683, "step": 4255 }, { "epoch": 1.28, "learning_rate": 0.00012810096281869185, "loss": 1.7035, "step": 4256 }, { "epoch": 1.28, "learning_rate": 0.00012806977273476254, "loss": 1.914, "step": 4257 }, { "epoch": 1.28, "learning_rate": 0.00012803857968630771, "loss": 1.8019, "step": 4258 }, { "epoch": 1.28, "learning_rate": 0.00012800738367662167, "loss": 1.8831, "step": 4259 }, { "epoch": 1.28, "learning_rate": 0.00012797618470899912, "loss": 1.8029, "step": 4260 }, { "epoch": 1.28, "learning_rate": 0.0001279449827867351, "loss": 1.8155, "step": 4261 }, { "epoch": 1.28, "learning_rate": 0.0001279137779131249, "loss": 1.8122, "step": 4262 }, { "epoch": 1.28, "learning_rate": 0.00012788257009146422, "loss": 1.854, "step": 4263 }, { "epoch": 1.28, "learning_rate": 0.00012785135932504892, "loss": 1.8159, "step": 4264 }, { "epoch": 1.28, "learning_rate": 0.00012782014561717525, "loss": 1.777, "step": 4265 }, { "epoch": 1.28, "learning_rate": 0.00012778892897113988, "loss": 1.7798, "step": 4266 }, { "epoch": 1.28, "learning_rate": 0.00012775770939023957, "loss": 1.7918, "step": 4267 }, { "epoch": 1.28, "learning_rate": 0.00012772648687777157, "loss": 1.8015, "step": 4268 }, { "epoch": 1.29, "learning_rate": 0.00012769526143703336, "loss": 1.7455, "step": 4269 }, { "epoch": 1.29, "learning_rate": 0.00012766403307132274, "loss": 1.8058, "step": 4270 }, { "epoch": 1.29, "learning_rate": 0.00012763280178393783, "loss": 1.7956, "step": 4271 }, { "epoch": 1.29, "learning_rate": 0.00012760156757817703, "loss": 1.7678, "step": 4272 }, { "epoch": 1.29, "learning_rate": 0.00012757033045733912, "loss": 1.7859, "step": 4273 }, { "epoch": 1.29, "learning_rate": 0.00012753909042472307, "loss": 1.9185, "step": 4274 }, { "epoch": 1.29, "learning_rate": 0.0001275078474836283, "loss": 1.8697, "step": 4275 }, { "epoch": 1.29, "learning_rate": 0.0001274766016373544, "loss": 1.8577, "step": 4276 }, { "epoch": 1.29, "learning_rate": 0.00012744535288920135, "loss": 1.8165, "step": 4277 }, { "epoch": 1.29, "learning_rate": 0.00012741410124246944, "loss": 1.7666, "step": 4278 }, { "epoch": 1.29, "learning_rate": 0.0001273828467004592, "loss": 1.8597, "step": 4279 }, { "epoch": 1.29, "learning_rate": 0.00012735158926647152, "loss": 1.856, "step": 4280 }, { "epoch": 1.29, "learning_rate": 0.0001273203289438076, "loss": 1.78, "step": 4281 }, { "epoch": 1.29, "learning_rate": 0.00012728906573576893, "loss": 1.7543, "step": 4282 }, { "epoch": 1.29, "learning_rate": 0.00012725779964565729, "loss": 1.8869, "step": 4283 }, { "epoch": 1.29, "learning_rate": 0.0001272265306767748, "loss": 1.7618, "step": 4284 }, { "epoch": 1.29, "learning_rate": 0.00012719525883242381, "loss": 1.8015, "step": 4285 }, { "epoch": 1.29, "learning_rate": 0.00012716398411590705, "loss": 1.7055, "step": 4286 }, { "epoch": 1.29, "learning_rate": 0.00012713270653052757, "loss": 1.9699, "step": 4287 }, { "epoch": 1.29, "learning_rate": 0.00012710142607958865, "loss": 2.0024, "step": 4288 }, { "epoch": 1.29, "learning_rate": 0.00012707014276639386, "loss": 1.9824, "step": 4289 }, { "epoch": 1.29, "learning_rate": 0.0001270388565942472, "loss": 1.7364, "step": 4290 }, { "epoch": 1.29, "learning_rate": 0.00012700756756645282, "loss": 1.734, "step": 4291 }, { "epoch": 1.29, "learning_rate": 0.00012697627568631527, "loss": 1.9642, "step": 4292 }, { "epoch": 1.29, "learning_rate": 0.00012694498095713937, "loss": 1.845, "step": 4293 }, { "epoch": 1.29, "learning_rate": 0.0001269136833822302, "loss": 1.8383, "step": 4294 }, { "epoch": 1.29, "learning_rate": 0.0001268823829648933, "loss": 1.8268, "step": 4295 }, { "epoch": 1.29, "learning_rate": 0.00012685107970843425, "loss": 1.8797, "step": 4296 }, { "epoch": 1.29, "learning_rate": 0.00012681977361615916, "loss": 1.8182, "step": 4297 }, { "epoch": 1.29, "learning_rate": 0.00012678846469137432, "loss": 1.9272, "step": 4298 }, { "epoch": 1.29, "learning_rate": 0.0001267571529373864, "loss": 1.8153, "step": 4299 }, { "epoch": 1.29, "learning_rate": 0.0001267258383575022, "loss": 1.9011, "step": 4300 }, { "epoch": 1.29, "learning_rate": 0.0001266945209550291, "loss": 1.953, "step": 4301 }, { "epoch": 1.3, "learning_rate": 0.00012666320073327452, "loss": 1.8107, "step": 4302 }, { "epoch": 1.3, "learning_rate": 0.00012663187769554627, "loss": 1.9924, "step": 4303 }, { "epoch": 1.3, "learning_rate": 0.0001266005518451525, "loss": 1.8549, "step": 4304 }, { "epoch": 1.3, "learning_rate": 0.00012656922318540157, "loss": 1.7844, "step": 4305 }, { "epoch": 1.3, "learning_rate": 0.0001265378917196022, "loss": 1.8731, "step": 4306 }, { "epoch": 1.3, "learning_rate": 0.00012650655745106344, "loss": 1.8374, "step": 4307 }, { "epoch": 1.3, "learning_rate": 0.00012647522038309458, "loss": 1.8895, "step": 4308 }, { "epoch": 1.3, "learning_rate": 0.00012644388051900518, "loss": 1.7966, "step": 4309 }, { "epoch": 1.3, "learning_rate": 0.00012641253786210512, "loss": 1.7943, "step": 4310 }, { "epoch": 1.3, "learning_rate": 0.00012638119241570463, "loss": 1.7556, "step": 4311 }, { "epoch": 1.3, "learning_rate": 0.0001263498441831141, "loss": 1.8714, "step": 4312 }, { "epoch": 1.3, "learning_rate": 0.0001263184931676444, "loss": 1.7611, "step": 4313 }, { "epoch": 1.3, "learning_rate": 0.00012628713937260657, "loss": 1.8148, "step": 4314 }, { "epoch": 1.3, "learning_rate": 0.00012625578280131197, "loss": 1.8275, "step": 4315 }, { "epoch": 1.3, "learning_rate": 0.00012622442345707223, "loss": 1.8568, "step": 4316 }, { "epoch": 1.3, "learning_rate": 0.0001261930613431993, "loss": 1.8559, "step": 4317 }, { "epoch": 1.3, "learning_rate": 0.00012616169646300542, "loss": 1.8615, "step": 4318 }, { "epoch": 1.3, "learning_rate": 0.00012613032881980316, "loss": 1.8512, "step": 4319 }, { "epoch": 1.3, "learning_rate": 0.0001260989584169053, "loss": 1.7353, "step": 4320 }, { "epoch": 1.3, "learning_rate": 0.00012606758525762494, "loss": 1.806, "step": 4321 }, { "epoch": 1.3, "learning_rate": 0.00012603620934527557, "loss": 1.8252, "step": 4322 }, { "epoch": 1.3, "learning_rate": 0.0001260048306831708, "loss": 1.8141, "step": 4323 }, { "epoch": 1.3, "learning_rate": 0.0001259734492746246, "loss": 1.7506, "step": 4324 }, { "epoch": 1.3, "learning_rate": 0.00012594206512295134, "loss": 1.786, "step": 4325 }, { "epoch": 1.3, "learning_rate": 0.0001259106782314655, "loss": 1.8701, "step": 4326 }, { "epoch": 1.3, "learning_rate": 0.000125879288603482, "loss": 1.7785, "step": 4327 }, { "epoch": 1.3, "learning_rate": 0.00012584789624231596, "loss": 1.7375, "step": 4328 }, { "epoch": 1.3, "learning_rate": 0.0001258165011512828, "loss": 1.7688, "step": 4329 }, { "epoch": 1.3, "learning_rate": 0.00012578510333369826, "loss": 1.8387, "step": 4330 }, { "epoch": 1.3, "learning_rate": 0.00012575370279287832, "loss": 1.9366, "step": 4331 }, { "epoch": 1.3, "learning_rate": 0.00012572229953213934, "loss": 1.8432, "step": 4332 }, { "epoch": 1.3, "learning_rate": 0.0001256908935547978, "loss": 1.8572, "step": 4333 }, { "epoch": 1.3, "learning_rate": 0.0001256594848641707, "loss": 1.7918, "step": 4334 }, { "epoch": 1.3, "learning_rate": 0.0001256280734635751, "loss": 1.824, "step": 4335 }, { "epoch": 1.31, "learning_rate": 0.00012559665935632847, "loss": 1.7522, "step": 4336 }, { "epoch": 1.31, "learning_rate": 0.00012556524254574858, "loss": 1.7377, "step": 4337 }, { "epoch": 1.31, "learning_rate": 0.00012553382303515334, "loss": 1.6974, "step": 4338 }, { "epoch": 1.31, "learning_rate": 0.0001255024008278612, "loss": 1.8917, "step": 4339 }, { "epoch": 1.31, "learning_rate": 0.00012547097592719063, "loss": 1.7816, "step": 4340 }, { "epoch": 1.31, "learning_rate": 0.00012543954833646055, "loss": 1.8198, "step": 4341 }, { "epoch": 1.31, "learning_rate": 0.00012540811805899008, "loss": 1.7364, "step": 4342 }, { "epoch": 1.31, "learning_rate": 0.00012537668509809868, "loss": 1.9121, "step": 4343 }, { "epoch": 1.31, "learning_rate": 0.00012534524945710608, "loss": 1.8731, "step": 4344 }, { "epoch": 1.31, "learning_rate": 0.00012531381113933227, "loss": 1.7989, "step": 4345 }, { "epoch": 1.31, "learning_rate": 0.00012528237014809755, "loss": 1.8755, "step": 4346 }, { "epoch": 1.31, "learning_rate": 0.00012525092648672247, "loss": 1.9353, "step": 4347 }, { "epoch": 1.31, "learning_rate": 0.00012521948015852788, "loss": 1.7089, "step": 4348 }, { "epoch": 1.31, "learning_rate": 0.00012518803116683496, "loss": 1.8129, "step": 4349 }, { "epoch": 1.31, "learning_rate": 0.00012515657951496506, "loss": 1.871, "step": 4350 }, { "epoch": 1.31, "learning_rate": 0.00012512512520623993, "loss": 1.7623, "step": 4351 }, { "epoch": 1.31, "learning_rate": 0.00012509366824398147, "loss": 1.8098, "step": 4352 }, { "epoch": 1.31, "learning_rate": 0.00012506220863151203, "loss": 1.8719, "step": 4353 }, { "epoch": 1.31, "learning_rate": 0.0001250307463721541, "loss": 1.8669, "step": 4354 }, { "epoch": 1.31, "learning_rate": 0.00012499928146923045, "loss": 1.9837, "step": 4355 }, { "epoch": 1.31, "learning_rate": 0.00012496781392606427, "loss": 2.0502, "step": 4356 }, { "epoch": 1.31, "learning_rate": 0.00012493634374597884, "loss": 1.9834, "step": 4357 }, { "epoch": 1.31, "learning_rate": 0.00012490487093229791, "loss": 1.9755, "step": 4358 }, { "epoch": 1.31, "learning_rate": 0.00012487339548834532, "loss": 1.996, "step": 4359 }, { "epoch": 1.31, "learning_rate": 0.00012484191741744528, "loss": 1.8192, "step": 4360 }, { "epoch": 1.31, "learning_rate": 0.00012481043672292236, "loss": 1.9177, "step": 4361 }, { "epoch": 1.31, "learning_rate": 0.00012477895340810125, "loss": 1.9218, "step": 4362 }, { "epoch": 1.31, "learning_rate": 0.00012474746747630704, "loss": 1.928, "step": 4363 }, { "epoch": 1.31, "learning_rate": 0.000124715978930865, "loss": 1.8582, "step": 4364 }, { "epoch": 1.31, "learning_rate": 0.00012468448777510073, "loss": 2.0024, "step": 4365 }, { "epoch": 1.31, "learning_rate": 0.0001246529940123401, "loss": 1.858, "step": 4366 }, { "epoch": 1.31, "learning_rate": 0.00012462149764590923, "loss": 1.9521, "step": 4367 }, { "epoch": 1.31, "learning_rate": 0.00012458999867913462, "loss": 1.8045, "step": 4368 }, { "epoch": 1.32, "learning_rate": 0.00012455849711534288, "loss": 1.8435, "step": 4369 }, { "epoch": 1.32, "learning_rate": 0.00012452699295786103, "loss": 1.8924, "step": 4370 }, { "epoch": 1.32, "learning_rate": 0.00012449548621001625, "loss": 1.8536, "step": 4371 }, { "epoch": 1.32, "learning_rate": 0.0001244639768751361, "loss": 1.9043, "step": 4372 }, { "epoch": 1.32, "learning_rate": 0.00012443246495654834, "loss": 1.882, "step": 4373 }, { "epoch": 1.32, "learning_rate": 0.00012440095045758106, "loss": 1.8379, "step": 4374 }, { "epoch": 1.32, "learning_rate": 0.00012436943338156259, "loss": 1.8958, "step": 4375 }, { "epoch": 1.32, "learning_rate": 0.0001243379137318215, "loss": 1.9021, "step": 4376 }, { "epoch": 1.32, "learning_rate": 0.00012430639151168668, "loss": 1.8677, "step": 4377 }, { "epoch": 1.32, "learning_rate": 0.0001242748667244873, "loss": 1.7856, "step": 4378 }, { "epoch": 1.32, "learning_rate": 0.00012424333937355276, "loss": 1.814, "step": 4379 }, { "epoch": 1.32, "learning_rate": 0.00012421180946221277, "loss": 1.9103, "step": 4380 }, { "epoch": 1.32, "learning_rate": 0.00012418027699379728, "loss": 1.8206, "step": 4381 }, { "epoch": 1.32, "learning_rate": 0.0001241487419716365, "loss": 1.8464, "step": 4382 }, { "epoch": 1.32, "learning_rate": 0.00012411720439906094, "loss": 1.77, "step": 4383 }, { "epoch": 1.32, "learning_rate": 0.00012408566427940139, "loss": 1.9402, "step": 4384 }, { "epoch": 1.32, "learning_rate": 0.0001240541216159889, "loss": 1.8298, "step": 4385 }, { "epoch": 1.32, "learning_rate": 0.0001240225764121547, "loss": 1.8898, "step": 4386 }, { "epoch": 1.32, "learning_rate": 0.00012399102867123046, "loss": 1.9053, "step": 4387 }, { "epoch": 1.32, "learning_rate": 0.00012395947839654795, "loss": 1.7372, "step": 4388 }, { "epoch": 1.32, "learning_rate": 0.0001239279255914393, "loss": 1.8394, "step": 4389 }, { "epoch": 1.32, "learning_rate": 0.0001238963702592369, "loss": 1.908, "step": 4390 }, { "epoch": 1.32, "learning_rate": 0.0001238648124032734, "loss": 1.837, "step": 4391 }, { "epoch": 1.32, "learning_rate": 0.00012383325202688174, "loss": 1.7836, "step": 4392 }, { "epoch": 1.32, "learning_rate": 0.00012380168913339503, "loss": 1.8472, "step": 4393 }, { "epoch": 1.32, "learning_rate": 0.00012377012372614675, "loss": 1.7859, "step": 4394 }, { "epoch": 1.32, "learning_rate": 0.0001237385558084706, "loss": 1.7878, "step": 4395 }, { "epoch": 1.32, "learning_rate": 0.0001237069853837006, "loss": 1.8952, "step": 4396 }, { "epoch": 1.32, "learning_rate": 0.00012367541245517089, "loss": 1.94, "step": 4397 }, { "epoch": 1.32, "learning_rate": 0.00012364383702621607, "loss": 1.7714, "step": 4398 }, { "epoch": 1.32, "learning_rate": 0.00012361225910017087, "loss": 1.8382, "step": 4399 }, { "epoch": 1.32, "learning_rate": 0.00012358067868037032, "loss": 1.8155, "step": 4400 }, { "epoch": 1.32, "learning_rate": 0.00012354909577014968, "loss": 1.8194, "step": 4401 }, { "epoch": 1.33, "learning_rate": 0.00012351751037284456, "loss": 1.815, "step": 4402 }, { "epoch": 1.33, "learning_rate": 0.00012348592249179077, "loss": 1.8324, "step": 4403 }, { "epoch": 1.33, "learning_rate": 0.0001234543321303244, "loss": 1.776, "step": 4404 }, { "epoch": 1.33, "learning_rate": 0.00012342273929178178, "loss": 1.8222, "step": 4405 }, { "epoch": 1.33, "learning_rate": 0.00012339114397949953, "loss": 1.83, "step": 4406 }, { "epoch": 1.33, "learning_rate": 0.00012335954619681447, "loss": 1.8667, "step": 4407 }, { "epoch": 1.33, "learning_rate": 0.0001233279459470638, "loss": 1.778, "step": 4408 }, { "epoch": 1.33, "learning_rate": 0.00012329634323358484, "loss": 1.891, "step": 4409 }, { "epoch": 1.33, "learning_rate": 0.00012326473805971528, "loss": 1.9603, "step": 4410 }, { "epoch": 1.33, "learning_rate": 0.000123233130428793, "loss": 1.806, "step": 4411 }, { "epoch": 1.33, "learning_rate": 0.00012320152034415625, "loss": 1.779, "step": 4412 }, { "epoch": 1.33, "learning_rate": 0.00012316990780914333, "loss": 1.8049, "step": 4413 }, { "epoch": 1.33, "learning_rate": 0.00012313829282709302, "loss": 1.739, "step": 4414 }, { "epoch": 1.33, "learning_rate": 0.00012310667540134425, "loss": 1.8256, "step": 4415 }, { "epoch": 1.33, "learning_rate": 0.00012307505553523618, "loss": 1.7755, "step": 4416 }, { "epoch": 1.33, "learning_rate": 0.00012304343323210834, "loss": 1.8353, "step": 4417 }, { "epoch": 1.33, "learning_rate": 0.00012301180849530035, "loss": 1.7543, "step": 4418 }, { "epoch": 1.33, "learning_rate": 0.0001229801813281523, "loss": 1.8142, "step": 4419 }, { "epoch": 1.33, "learning_rate": 0.00012294855173400435, "loss": 1.8196, "step": 4420 }, { "epoch": 1.33, "learning_rate": 0.00012291691971619699, "loss": 1.9344, "step": 4421 }, { "epoch": 1.33, "learning_rate": 0.00012288528527807098, "loss": 1.7753, "step": 4422 }, { "epoch": 1.33, "learning_rate": 0.00012285364842296734, "loss": 1.8778, "step": 4423 }, { "epoch": 1.33, "learning_rate": 0.00012282200915422731, "loss": 1.9224, "step": 4424 }, { "epoch": 1.33, "learning_rate": 0.00012279036747519235, "loss": 1.7425, "step": 4425 }, { "epoch": 1.33, "learning_rate": 0.0001227587233892043, "loss": 1.8594, "step": 4426 }, { "epoch": 1.33, "learning_rate": 0.0001227270768996051, "loss": 1.814, "step": 4427 }, { "epoch": 1.33, "learning_rate": 0.0001226954280097371, "loss": 1.9147, "step": 4428 }, { "epoch": 1.33, "learning_rate": 0.00012266377672294281, "loss": 1.8516, "step": 4429 }, { "epoch": 1.33, "learning_rate": 0.000122632123042565, "loss": 1.8685, "step": 4430 }, { "epoch": 1.33, "learning_rate": 0.00012260046697194665, "loss": 1.7122, "step": 4431 }, { "epoch": 1.33, "learning_rate": 0.00012256880851443114, "loss": 1.8733, "step": 4432 }, { "epoch": 1.33, "learning_rate": 0.0001225371476733619, "loss": 1.7543, "step": 4433 }, { "epoch": 1.33, "learning_rate": 0.0001225054844520828, "loss": 1.8723, "step": 4434 }, { "epoch": 1.34, "learning_rate": 0.00012247381885393784, "loss": 1.9511, "step": 4435 }, { "epoch": 1.34, "learning_rate": 0.00012244215088227136, "loss": 1.7798, "step": 4436 }, { "epoch": 1.34, "learning_rate": 0.00012241048054042783, "loss": 1.7784, "step": 4437 }, { "epoch": 1.34, "learning_rate": 0.0001223788078317521, "loss": 1.9673, "step": 4438 }, { "epoch": 1.34, "learning_rate": 0.00012234713275958918, "loss": 1.8914, "step": 4439 }, { "epoch": 1.34, "learning_rate": 0.00012231545532728435, "loss": 1.8392, "step": 4440 }, { "epoch": 1.34, "learning_rate": 0.0001222837755381832, "loss": 1.8549, "step": 4441 }, { "epoch": 1.34, "learning_rate": 0.00012225209339563145, "loss": 1.8171, "step": 4442 }, { "epoch": 1.34, "learning_rate": 0.00012222040890297518, "loss": 1.798, "step": 4443 }, { "epoch": 1.34, "learning_rate": 0.0001221887220635607, "loss": 1.9146, "step": 4444 }, { "epoch": 1.34, "learning_rate": 0.0001221570328807345, "loss": 1.781, "step": 4445 }, { "epoch": 1.34, "learning_rate": 0.00012212534135784341, "loss": 1.8193, "step": 4446 }, { "epoch": 1.34, "learning_rate": 0.00012209364749823438, "loss": 1.8374, "step": 4447 }, { "epoch": 1.34, "learning_rate": 0.00012206195130525477, "loss": 1.8225, "step": 4448 }, { "epoch": 1.34, "learning_rate": 0.00012203025278225204, "loss": 1.8309, "step": 4449 }, { "epoch": 1.34, "learning_rate": 0.00012199855193257396, "loss": 1.8993, "step": 4450 }, { "epoch": 1.34, "learning_rate": 0.00012196684875956861, "loss": 1.768, "step": 4451 }, { "epoch": 1.34, "learning_rate": 0.00012193514326658417, "loss": 1.8372, "step": 4452 }, { "epoch": 1.34, "learning_rate": 0.0001219034354569692, "loss": 1.8388, "step": 4453 }, { "epoch": 1.34, "learning_rate": 0.00012187172533407243, "loss": 1.9021, "step": 4454 }, { "epoch": 1.34, "learning_rate": 0.00012184001290124283, "loss": 1.8411, "step": 4455 }, { "epoch": 1.34, "learning_rate": 0.00012180829816182965, "loss": 1.8143, "step": 4456 }, { "epoch": 1.34, "learning_rate": 0.0001217765811191824, "loss": 1.8116, "step": 4457 }, { "epoch": 1.34, "learning_rate": 0.00012174486177665079, "loss": 1.8025, "step": 4458 }, { "epoch": 1.34, "learning_rate": 0.00012171314013758473, "loss": 1.8058, "step": 4459 }, { "epoch": 1.34, "learning_rate": 0.00012168141620533451, "loss": 1.7681, "step": 4460 }, { "epoch": 1.34, "learning_rate": 0.00012164968998325052, "loss": 1.8133, "step": 4461 }, { "epoch": 1.34, "learning_rate": 0.00012161796147468352, "loss": 1.8573, "step": 4462 }, { "epoch": 1.34, "learning_rate": 0.00012158623068298438, "loss": 1.9589, "step": 4463 }, { "epoch": 1.34, "learning_rate": 0.0001215544976115043, "loss": 1.7596, "step": 4464 }, { "epoch": 1.34, "learning_rate": 0.00012152276226359471, "loss": 1.7026, "step": 4465 }, { "epoch": 1.34, "learning_rate": 0.00012149102464260724, "loss": 1.7864, "step": 4466 }, { "epoch": 1.34, "learning_rate": 0.00012145928475189381, "loss": 1.8148, "step": 4467 }, { "epoch": 1.34, "learning_rate": 0.00012142754259480653, "loss": 1.7513, "step": 4468 }, { "epoch": 1.35, "learning_rate": 0.00012139579817469782, "loss": 1.844, "step": 4469 }, { "epoch": 1.35, "learning_rate": 0.00012136405149492028, "loss": 1.8412, "step": 4470 }, { "epoch": 1.35, "learning_rate": 0.00012133230255882673, "loss": 1.8589, "step": 4471 }, { "epoch": 1.35, "learning_rate": 0.0001213005513697703, "loss": 1.8077, "step": 4472 }, { "epoch": 1.35, "learning_rate": 0.00012126879793110427, "loss": 1.8484, "step": 4473 }, { "epoch": 1.35, "learning_rate": 0.00012123704224618229, "loss": 1.863, "step": 4474 }, { "epoch": 1.35, "learning_rate": 0.0001212052843183581, "loss": 1.8775, "step": 4475 }, { "epoch": 1.35, "learning_rate": 0.00012117352415098576, "loss": 2.035, "step": 4476 }, { "epoch": 1.35, "learning_rate": 0.00012114176174741956, "loss": 1.8853, "step": 4477 }, { "epoch": 1.35, "learning_rate": 0.00012110999711101396, "loss": 1.8243, "step": 4478 }, { "epoch": 1.35, "learning_rate": 0.0001210782302451238, "loss": 1.7955, "step": 4479 }, { "epoch": 1.35, "learning_rate": 0.00012104646115310396, "loss": 1.8104, "step": 4480 }, { "epoch": 1.35, "learning_rate": 0.0001210146898383098, "loss": 1.8404, "step": 4481 }, { "epoch": 1.35, "learning_rate": 0.00012098291630409666, "loss": 1.8838, "step": 4482 }, { "epoch": 1.35, "learning_rate": 0.00012095114055382024, "loss": 1.9434, "step": 4483 }, { "epoch": 1.35, "learning_rate": 0.00012091936259083652, "loss": 1.8116, "step": 4484 }, { "epoch": 1.35, "learning_rate": 0.00012088758241850158, "loss": 1.8304, "step": 4485 }, { "epoch": 1.35, "learning_rate": 0.0001208558000401719, "loss": 1.8352, "step": 4486 }, { "epoch": 1.35, "learning_rate": 0.00012082401545920406, "loss": 1.8168, "step": 4487 }, { "epoch": 1.35, "learning_rate": 0.0001207922286789549, "loss": 1.8801, "step": 4488 }, { "epoch": 1.35, "learning_rate": 0.00012076043970278157, "loss": 1.8493, "step": 4489 }, { "epoch": 1.35, "learning_rate": 0.00012072864853404132, "loss": 1.7359, "step": 4490 }, { "epoch": 1.35, "learning_rate": 0.00012069685517609171, "loss": 1.7076, "step": 4491 }, { "epoch": 1.35, "learning_rate": 0.00012066505963229053, "loss": 1.8046, "step": 4492 }, { "epoch": 1.35, "learning_rate": 0.00012063326190599587, "loss": 1.7564, "step": 4493 }, { "epoch": 1.35, "learning_rate": 0.00012060146200056586, "loss": 1.8856, "step": 4494 }, { "epoch": 1.35, "learning_rate": 0.00012056965991935907, "loss": 1.7548, "step": 4495 }, { "epoch": 1.35, "learning_rate": 0.00012053785566573412, "loss": 1.7984, "step": 4496 }, { "epoch": 1.35, "learning_rate": 0.00012050604924305, "loss": 1.9794, "step": 4497 }, { "epoch": 1.35, "learning_rate": 0.0001204742406546659, "loss": 1.8224, "step": 4498 }, { "epoch": 1.35, "learning_rate": 0.00012044242990394115, "loss": 1.7513, "step": 4499 }, { "epoch": 1.35, "learning_rate": 0.00012041061699423539, "loss": 1.8281, "step": 4500 }, { "epoch": 1.35, "learning_rate": 0.00012037880192890844, "loss": 1.8963, "step": 4501 }, { "epoch": 1.36, "learning_rate": 0.00012034698471132045, "loss": 1.8246, "step": 4502 }, { "epoch": 1.36, "learning_rate": 0.00012031516534483169, "loss": 1.8789, "step": 4503 }, { "epoch": 1.36, "learning_rate": 0.00012028334383280264, "loss": 1.8698, "step": 4504 }, { "epoch": 1.36, "learning_rate": 0.00012025152017859414, "loss": 1.7304, "step": 4505 }, { "epoch": 1.36, "learning_rate": 0.00012021969438556711, "loss": 1.79, "step": 4506 }, { "epoch": 1.36, "learning_rate": 0.00012018786645708279, "loss": 1.8041, "step": 4507 }, { "epoch": 1.36, "learning_rate": 0.00012015603639650261, "loss": 1.7293, "step": 4508 }, { "epoch": 1.36, "learning_rate": 0.00012012420420718822, "loss": 1.7596, "step": 4509 }, { "epoch": 1.36, "learning_rate": 0.00012009236989250152, "loss": 1.8376, "step": 4510 }, { "epoch": 1.36, "learning_rate": 0.0001200605334558046, "loss": 1.7182, "step": 4511 }, { "epoch": 1.36, "learning_rate": 0.0001200286949004598, "loss": 1.8651, "step": 4512 }, { "epoch": 1.36, "learning_rate": 0.00011999685422982968, "loss": 1.6428, "step": 4513 }, { "epoch": 1.36, "learning_rate": 0.00011996501144727701, "loss": 1.753, "step": 4514 }, { "epoch": 1.36, "learning_rate": 0.00011993316655616485, "loss": 1.8766, "step": 4515 }, { "epoch": 1.36, "learning_rate": 0.00011990131955985634, "loss": 1.7923, "step": 4516 }, { "epoch": 1.36, "learning_rate": 0.000119869470461715, "loss": 1.7266, "step": 4517 }, { "epoch": 1.36, "learning_rate": 0.00011983761926510442, "loss": 1.7977, "step": 4518 }, { "epoch": 1.36, "learning_rate": 0.0001198057659733886, "loss": 1.8975, "step": 4519 }, { "epoch": 1.36, "learning_rate": 0.00011977391058993155, "loss": 1.7944, "step": 4520 }, { "epoch": 1.36, "learning_rate": 0.00011974205311809765, "loss": 1.7479, "step": 4521 }, { "epoch": 1.36, "learning_rate": 0.00011971019356125148, "loss": 1.8227, "step": 4522 }, { "epoch": 1.36, "learning_rate": 0.00011967833192275777, "loss": 1.7999, "step": 4523 }, { "epoch": 1.36, "learning_rate": 0.00011964646820598155, "loss": 1.7952, "step": 4524 }, { "epoch": 1.36, "learning_rate": 0.00011961460241428802, "loss": 1.8051, "step": 4525 }, { "epoch": 1.36, "learning_rate": 0.00011958273455104259, "loss": 1.7683, "step": 4526 }, { "epoch": 1.36, "learning_rate": 0.00011955086461961095, "loss": 1.7408, "step": 4527 }, { "epoch": 1.36, "learning_rate": 0.00011951899262335895, "loss": 1.7788, "step": 4528 }, { "epoch": 1.36, "learning_rate": 0.00011948711856565271, "loss": 1.8363, "step": 4529 }, { "epoch": 1.36, "learning_rate": 0.00011945524244985849, "loss": 1.8435, "step": 4530 }, { "epoch": 1.36, "learning_rate": 0.00011942336427934284, "loss": 1.9374, "step": 4531 }, { "epoch": 1.36, "learning_rate": 0.00011939148405747247, "loss": 1.9566, "step": 4532 }, { "epoch": 1.36, "learning_rate": 0.00011935960178761436, "loss": 1.7879, "step": 4533 }, { "epoch": 1.36, "learning_rate": 0.00011932771747313573, "loss": 1.7999, "step": 4534 }, { "epoch": 1.37, "learning_rate": 0.0001192958311174039, "loss": 1.7033, "step": 4535 }, { "epoch": 1.37, "learning_rate": 0.00011926394272378651, "loss": 1.7576, "step": 4536 }, { "epoch": 1.37, "learning_rate": 0.00011923205229565136, "loss": 1.7797, "step": 4537 }, { "epoch": 1.37, "learning_rate": 0.0001192001598363665, "loss": 1.9672, "step": 4538 }, { "epoch": 1.37, "learning_rate": 0.00011916826534930019, "loss": 1.8088, "step": 4539 }, { "epoch": 1.37, "learning_rate": 0.00011913636883782086, "loss": 1.8425, "step": 4540 }, { "epoch": 1.37, "learning_rate": 0.00011910447030529723, "loss": 1.8474, "step": 4541 }, { "epoch": 1.37, "learning_rate": 0.00011907256975509815, "loss": 1.8905, "step": 4542 }, { "epoch": 1.37, "learning_rate": 0.00011904066719059277, "loss": 1.7926, "step": 4543 }, { "epoch": 1.37, "learning_rate": 0.00011900876261515035, "loss": 1.7926, "step": 4544 }, { "epoch": 1.37, "learning_rate": 0.00011897685603214048, "loss": 1.8108, "step": 4545 }, { "epoch": 1.37, "learning_rate": 0.0001189449474449329, "loss": 1.8082, "step": 4546 }, { "epoch": 1.37, "learning_rate": 0.0001189130368568975, "loss": 1.8275, "step": 4547 }, { "epoch": 1.37, "learning_rate": 0.0001188811242714045, "loss": 1.8248, "step": 4548 }, { "epoch": 1.37, "learning_rate": 0.00011884920969182422, "loss": 1.8527, "step": 4549 }, { "epoch": 1.37, "learning_rate": 0.00011881729312152729, "loss": 1.7455, "step": 4550 }, { "epoch": 1.37, "learning_rate": 0.00011878537456388452, "loss": 1.8268, "step": 4551 }, { "epoch": 1.37, "learning_rate": 0.00011875345402226692, "loss": 1.915, "step": 4552 }, { "epoch": 1.37, "learning_rate": 0.00011872153150004565, "loss": 1.7864, "step": 4553 }, { "epoch": 1.37, "learning_rate": 0.00011868960700059218, "loss": 1.9054, "step": 4554 }, { "epoch": 1.37, "learning_rate": 0.00011865768052727812, "loss": 1.7351, "step": 4555 }, { "epoch": 1.37, "learning_rate": 0.0001186257520834753, "loss": 1.8914, "step": 4556 }, { "epoch": 1.37, "learning_rate": 0.00011859382167255584, "loss": 1.8071, "step": 4557 }, { "epoch": 1.37, "learning_rate": 0.00011856188929789194, "loss": 1.6918, "step": 4558 }, { "epoch": 1.37, "learning_rate": 0.00011852995496285606, "loss": 1.9082, "step": 4559 }, { "epoch": 1.37, "learning_rate": 0.00011849801867082093, "loss": 1.7874, "step": 4560 }, { "epoch": 1.37, "learning_rate": 0.00011846608042515934, "loss": 1.8142, "step": 4561 }, { "epoch": 1.37, "learning_rate": 0.00011843414022924443, "loss": 1.7997, "step": 4562 }, { "epoch": 1.37, "learning_rate": 0.0001184021980864495, "loss": 1.7967, "step": 4563 }, { "epoch": 1.37, "learning_rate": 0.00011837025400014806, "loss": 1.8416, "step": 4564 }, { "epoch": 1.37, "learning_rate": 0.00011833830797371374, "loss": 1.8786, "step": 4565 }, { "epoch": 1.37, "learning_rate": 0.00011830636001052052, "loss": 1.7969, "step": 4566 }, { "epoch": 1.37, "learning_rate": 0.0001182744101139425, "loss": 1.9167, "step": 4567 }, { "epoch": 1.38, "learning_rate": 0.00011824245828735392, "loss": 1.8085, "step": 4568 }, { "epoch": 1.38, "learning_rate": 0.00011821050453412941, "loss": 1.754, "step": 4569 }, { "epoch": 1.38, "learning_rate": 0.00011817854885764364, "loss": 1.7981, "step": 4570 }, { "epoch": 1.38, "learning_rate": 0.00011814659126127153, "loss": 1.7364, "step": 4571 }, { "epoch": 1.38, "learning_rate": 0.00011811463174838824, "loss": 1.7249, "step": 4572 }, { "epoch": 1.38, "learning_rate": 0.00011808267032236906, "loss": 1.8108, "step": 4573 }, { "epoch": 1.38, "learning_rate": 0.00011805070698658953, "loss": 1.7938, "step": 4574 }, { "epoch": 1.38, "learning_rate": 0.00011801874174442542, "loss": 1.7712, "step": 4575 }, { "epoch": 1.38, "learning_rate": 0.00011798677459925266, "loss": 1.82, "step": 4576 }, { "epoch": 1.38, "learning_rate": 0.00011795480555444736, "loss": 1.7978, "step": 4577 }, { "epoch": 1.38, "learning_rate": 0.00011792283461338589, "loss": 1.7689, "step": 4578 }, { "epoch": 1.38, "learning_rate": 0.0001178908617794448, "loss": 1.8645, "step": 4579 }, { "epoch": 1.38, "learning_rate": 0.00011785888705600072, "loss": 1.9692, "step": 4580 }, { "epoch": 1.38, "learning_rate": 0.00011782691044643072, "loss": 1.912, "step": 4581 }, { "epoch": 1.38, "learning_rate": 0.00011779493195411187, "loss": 1.8614, "step": 4582 }, { "epoch": 1.38, "learning_rate": 0.00011776295158242155, "loss": 1.9468, "step": 4583 }, { "epoch": 1.38, "learning_rate": 0.00011773096933473725, "loss": 1.9237, "step": 4584 }, { "epoch": 1.38, "learning_rate": 0.00011769898521443672, "loss": 1.838, "step": 4585 }, { "epoch": 1.38, "learning_rate": 0.00011766699922489787, "loss": 1.9395, "step": 4586 }, { "epoch": 1.38, "learning_rate": 0.00011763501136949888, "loss": 1.9045, "step": 4587 }, { "epoch": 1.38, "learning_rate": 0.00011760302165161804, "loss": 1.862, "step": 4588 }, { "epoch": 1.38, "learning_rate": 0.00011757103007463385, "loss": 1.8205, "step": 4589 }, { "epoch": 1.38, "learning_rate": 0.00011753903664192509, "loss": 1.8465, "step": 4590 }, { "epoch": 1.38, "learning_rate": 0.0001175070413568706, "loss": 1.8392, "step": 4591 }, { "epoch": 1.38, "learning_rate": 0.00011747504422284957, "loss": 1.7135, "step": 4592 }, { "epoch": 1.38, "learning_rate": 0.00011744304524324124, "loss": 1.8355, "step": 4593 }, { "epoch": 1.38, "learning_rate": 0.00011741104442142512, "loss": 1.7719, "step": 4594 }, { "epoch": 1.38, "learning_rate": 0.00011737904176078094, "loss": 1.7112, "step": 4595 }, { "epoch": 1.38, "learning_rate": 0.00011734703726468854, "loss": 1.9053, "step": 4596 }, { "epoch": 1.38, "learning_rate": 0.00011731503093652805, "loss": 1.8197, "step": 4597 }, { "epoch": 1.38, "learning_rate": 0.00011728302277967968, "loss": 1.8138, "step": 4598 }, { "epoch": 1.38, "learning_rate": 0.00011725101279752398, "loss": 1.8293, "step": 4599 }, { "epoch": 1.38, "learning_rate": 0.00011721900099344158, "loss": 1.8046, "step": 4600 }, { "epoch": 1.39, "learning_rate": 0.00011718698737081331, "loss": 1.8946, "step": 4601 }, { "epoch": 1.39, "learning_rate": 0.00011715497193302026, "loss": 1.7115, "step": 4602 }, { "epoch": 1.39, "learning_rate": 0.00011712295468344361, "loss": 1.7666, "step": 4603 }, { "epoch": 1.39, "learning_rate": 0.00011709093562546483, "loss": 1.8617, "step": 4604 }, { "epoch": 1.39, "learning_rate": 0.00011705891476246555, "loss": 1.8397, "step": 4605 }, { "epoch": 1.39, "learning_rate": 0.00011702689209782756, "loss": 1.8567, "step": 4606 }, { "epoch": 1.39, "learning_rate": 0.00011699486763493286, "loss": 1.8292, "step": 4607 }, { "epoch": 1.39, "learning_rate": 0.00011696284137716364, "loss": 1.9498, "step": 4608 }, { "epoch": 1.39, "learning_rate": 0.0001169308133279023, "loss": 1.8032, "step": 4609 }, { "epoch": 1.39, "learning_rate": 0.00011689878349053141, "loss": 1.6557, "step": 4610 }, { "epoch": 1.39, "learning_rate": 0.00011686675186843367, "loss": 1.8209, "step": 4611 }, { "epoch": 1.39, "learning_rate": 0.00011683471846499212, "loss": 1.8368, "step": 4612 }, { "epoch": 1.39, "learning_rate": 0.00011680268328358984, "loss": 1.8222, "step": 4613 }, { "epoch": 1.39, "learning_rate": 0.00011677064632761015, "loss": 1.6355, "step": 4614 }, { "epoch": 1.39, "learning_rate": 0.00011673860760043655, "loss": 1.739, "step": 4615 }, { "epoch": 1.39, "learning_rate": 0.00011670656710545278, "loss": 1.7898, "step": 4616 }, { "epoch": 1.39, "learning_rate": 0.00011667452484604273, "loss": 1.926, "step": 4617 }, { "epoch": 1.39, "learning_rate": 0.00011664248082559042, "loss": 1.742, "step": 4618 }, { "epoch": 1.39, "learning_rate": 0.00011661043504748015, "loss": 1.9216, "step": 4619 }, { "epoch": 1.39, "learning_rate": 0.00011657838751509633, "loss": 1.8096, "step": 4620 }, { "epoch": 1.39, "learning_rate": 0.00011654633823182359, "loss": 1.8433, "step": 4621 }, { "epoch": 1.39, "learning_rate": 0.00011651428720104677, "loss": 1.8016, "step": 4622 }, { "epoch": 1.39, "learning_rate": 0.00011648223442615087, "loss": 1.7613, "step": 4623 }, { "epoch": 1.39, "learning_rate": 0.00011645017991052105, "loss": 1.7176, "step": 4624 }, { "epoch": 1.39, "learning_rate": 0.00011641812365754266, "loss": 1.8369, "step": 4625 }, { "epoch": 1.39, "learning_rate": 0.00011638606567060129, "loss": 1.716, "step": 4626 }, { "epoch": 1.39, "learning_rate": 0.0001163540059530826, "loss": 1.7886, "step": 4627 }, { "epoch": 1.39, "learning_rate": 0.0001163219445083726, "loss": 1.8198, "step": 4628 }, { "epoch": 1.39, "learning_rate": 0.00011628988133985732, "loss": 1.8861, "step": 4629 }, { "epoch": 1.39, "learning_rate": 0.00011625781645092306, "loss": 1.7743, "step": 4630 }, { "epoch": 1.39, "learning_rate": 0.00011622574984495632, "loss": 1.6934, "step": 4631 }, { "epoch": 1.39, "learning_rate": 0.00011619368152534365, "loss": 1.8199, "step": 4632 }, { "epoch": 1.39, "learning_rate": 0.00011616161149547194, "loss": 1.9085, "step": 4633 }, { "epoch": 1.39, "learning_rate": 0.00011612953975872818, "loss": 1.8049, "step": 4634 }, { "epoch": 1.4, "learning_rate": 0.00011609746631849956, "loss": 1.8692, "step": 4635 }, { "epoch": 1.4, "learning_rate": 0.00011606539117817345, "loss": 1.7675, "step": 4636 }, { "epoch": 1.4, "learning_rate": 0.00011603331434113735, "loss": 1.8259, "step": 4637 }, { "epoch": 1.4, "learning_rate": 0.00011600123581077904, "loss": 1.8311, "step": 4638 }, { "epoch": 1.4, "learning_rate": 0.00011596915559048635, "loss": 1.8111, "step": 4639 }, { "epoch": 1.4, "learning_rate": 0.00011593707368364745, "loss": 1.7439, "step": 4640 }, { "epoch": 1.4, "learning_rate": 0.00011590499009365055, "loss": 1.8195, "step": 4641 }, { "epoch": 1.4, "learning_rate": 0.00011587290482388409, "loss": 1.8041, "step": 4642 }, { "epoch": 1.4, "learning_rate": 0.00011584081787773669, "loss": 1.7095, "step": 4643 }, { "epoch": 1.4, "learning_rate": 0.0001158087292585971, "loss": 1.8872, "step": 4644 }, { "epoch": 1.4, "learning_rate": 0.00011577663896985433, "loss": 1.7774, "step": 4645 }, { "epoch": 1.4, "learning_rate": 0.0001157445470148975, "loss": 1.8544, "step": 4646 }, { "epoch": 1.4, "learning_rate": 0.00011571245339711596, "loss": 1.789, "step": 4647 }, { "epoch": 1.4, "learning_rate": 0.00011568035811989916, "loss": 1.7123, "step": 4648 }, { "epoch": 1.4, "learning_rate": 0.00011564826118663685, "loss": 1.8398, "step": 4649 }, { "epoch": 1.4, "learning_rate": 0.00011561616260071879, "loss": 1.7545, "step": 4650 }, { "epoch": 1.4, "learning_rate": 0.000115584062365535, "loss": 1.8688, "step": 4651 }, { "epoch": 1.4, "learning_rate": 0.00011555196048447576, "loss": 1.8689, "step": 4652 }, { "epoch": 1.4, "learning_rate": 0.00011551985696093135, "loss": 1.8109, "step": 4653 }, { "epoch": 1.4, "learning_rate": 0.00011548775179829236, "loss": 1.8292, "step": 4654 }, { "epoch": 1.4, "learning_rate": 0.00011545564499994947, "loss": 1.6975, "step": 4655 }, { "epoch": 1.4, "learning_rate": 0.00011542353656929362, "loss": 1.8162, "step": 4656 }, { "epoch": 1.4, "learning_rate": 0.00011539142650971579, "loss": 1.8624, "step": 4657 }, { "epoch": 1.4, "learning_rate": 0.00011535931482460727, "loss": 1.8247, "step": 4658 }, { "epoch": 1.4, "learning_rate": 0.00011532720151735946, "loss": 1.787, "step": 4659 }, { "epoch": 1.4, "learning_rate": 0.0001152950865913639, "loss": 1.8588, "step": 4660 }, { "epoch": 1.4, "learning_rate": 0.00011526297005001236, "loss": 1.7802, "step": 4661 }, { "epoch": 1.4, "learning_rate": 0.00011523085189669676, "loss": 1.785, "step": 4662 }, { "epoch": 1.4, "learning_rate": 0.00011519873213480919, "loss": 1.8664, "step": 4663 }, { "epoch": 1.4, "learning_rate": 0.00011516661076774188, "loss": 1.8423, "step": 4664 }, { "epoch": 1.4, "learning_rate": 0.00011513448779888727, "loss": 1.842, "step": 4665 }, { "epoch": 1.4, "learning_rate": 0.00011510236323163799, "loss": 1.7373, "step": 4666 }, { "epoch": 1.4, "learning_rate": 0.00011507023706938673, "loss": 1.8877, "step": 4667 }, { "epoch": 1.41, "learning_rate": 0.00011503810931552647, "loss": 1.8199, "step": 4668 }, { "epoch": 1.41, "learning_rate": 0.00011500597997345033, "loss": 1.7367, "step": 4669 }, { "epoch": 1.41, "learning_rate": 0.0001149738490465515, "loss": 1.9248, "step": 4670 }, { "epoch": 1.41, "learning_rate": 0.0001149417165382235, "loss": 1.8205, "step": 4671 }, { "epoch": 1.41, "learning_rate": 0.00011490958245185988, "loss": 1.826, "step": 4672 }, { "epoch": 1.41, "learning_rate": 0.00011487744679085445, "loss": 1.8677, "step": 4673 }, { "epoch": 1.41, "learning_rate": 0.00011484530955860109, "loss": 1.8, "step": 4674 }, { "epoch": 1.41, "learning_rate": 0.00011481317075849393, "loss": 1.7762, "step": 4675 }, { "epoch": 1.41, "learning_rate": 0.00011478103039392727, "loss": 1.8524, "step": 4676 }, { "epoch": 1.41, "learning_rate": 0.0001147488884682955, "loss": 1.7875, "step": 4677 }, { "epoch": 1.41, "learning_rate": 0.00011471674498499323, "loss": 1.7639, "step": 4678 }, { "epoch": 1.41, "learning_rate": 0.00011468459994741521, "loss": 1.897, "step": 4679 }, { "epoch": 1.41, "learning_rate": 0.00011465245335895637, "loss": 1.939, "step": 4680 }, { "epoch": 1.41, "learning_rate": 0.00011462030522301179, "loss": 1.9687, "step": 4681 }, { "epoch": 1.41, "learning_rate": 0.00011458815554297675, "loss": 1.851, "step": 4682 }, { "epoch": 1.41, "learning_rate": 0.00011455600432224667, "loss": 1.8486, "step": 4683 }, { "epoch": 1.41, "learning_rate": 0.00011452385156421708, "loss": 1.785, "step": 4684 }, { "epoch": 1.41, "learning_rate": 0.00011449169727228379, "loss": 1.8174, "step": 4685 }, { "epoch": 1.41, "learning_rate": 0.00011445954144984263, "loss": 1.8168, "step": 4686 }, { "epoch": 1.41, "learning_rate": 0.00011442738410028971, "loss": 1.8199, "step": 4687 }, { "epoch": 1.41, "learning_rate": 0.00011439522522702124, "loss": 1.8488, "step": 4688 }, { "epoch": 1.41, "learning_rate": 0.0001143630648334336, "loss": 1.8275, "step": 4689 }, { "epoch": 1.41, "learning_rate": 0.00011433090292292337, "loss": 1.7903, "step": 4690 }, { "epoch": 1.41, "learning_rate": 0.00011429873949888721, "loss": 1.8474, "step": 4691 }, { "epoch": 1.41, "learning_rate": 0.00011426657456472205, "loss": 1.8243, "step": 4692 }, { "epoch": 1.41, "learning_rate": 0.00011423440812382485, "loss": 1.8386, "step": 4693 }, { "epoch": 1.41, "learning_rate": 0.00011420224017959282, "loss": 1.9252, "step": 4694 }, { "epoch": 1.41, "learning_rate": 0.00011417007073542334, "loss": 1.7169, "step": 4695 }, { "epoch": 1.41, "learning_rate": 0.00011413789979471386, "loss": 1.8371, "step": 4696 }, { "epoch": 1.41, "learning_rate": 0.00011410572736086207, "loss": 1.7671, "step": 4697 }, { "epoch": 1.41, "learning_rate": 0.00011407355343726577, "loss": 1.6847, "step": 4698 }, { "epoch": 1.41, "learning_rate": 0.00011404137802732297, "loss": 1.8642, "step": 4699 }, { "epoch": 1.41, "learning_rate": 0.00011400920113443178, "loss": 1.8824, "step": 4700 }, { "epoch": 1.42, "learning_rate": 0.00011397702276199048, "loss": 1.98, "step": 4701 }, { "epoch": 1.42, "learning_rate": 0.00011394484291339755, "loss": 1.847, "step": 4702 }, { "epoch": 1.42, "learning_rate": 0.00011391266159205155, "loss": 1.7506, "step": 4703 }, { "epoch": 1.42, "learning_rate": 0.00011388047880135127, "loss": 1.8918, "step": 4704 }, { "epoch": 1.42, "learning_rate": 0.00011384829454469558, "loss": 1.853, "step": 4705 }, { "epoch": 1.42, "learning_rate": 0.00011381610882548361, "loss": 1.8994, "step": 4706 }, { "epoch": 1.42, "learning_rate": 0.00011378392164711456, "loss": 1.7968, "step": 4707 }, { "epoch": 1.42, "learning_rate": 0.00011375173301298775, "loss": 1.8584, "step": 4708 }, { "epoch": 1.42, "learning_rate": 0.00011371954292650281, "loss": 1.9081, "step": 4709 }, { "epoch": 1.42, "learning_rate": 0.00011368735139105933, "loss": 1.7803, "step": 4710 }, { "epoch": 1.42, "learning_rate": 0.00011365515841005719, "loss": 1.7167, "step": 4711 }, { "epoch": 1.42, "learning_rate": 0.00011362296398689637, "loss": 1.6992, "step": 4712 }, { "epoch": 1.42, "learning_rate": 0.00011359076812497704, "loss": 1.8371, "step": 4713 }, { "epoch": 1.42, "learning_rate": 0.00011355857082769948, "loss": 1.8188, "step": 4714 }, { "epoch": 1.42, "learning_rate": 0.00011352637209846406, "loss": 1.7748, "step": 4715 }, { "epoch": 1.42, "learning_rate": 0.00011349417194067151, "loss": 1.9778, "step": 4716 }, { "epoch": 1.42, "learning_rate": 0.00011346197035772244, "loss": 1.78, "step": 4717 }, { "epoch": 1.42, "learning_rate": 0.00011342976735301785, "loss": 1.7491, "step": 4718 }, { "epoch": 1.42, "learning_rate": 0.00011339756292995875, "loss": 1.932, "step": 4719 }, { "epoch": 1.42, "learning_rate": 0.00011336535709194633, "loss": 1.8127, "step": 4720 }, { "epoch": 1.42, "learning_rate": 0.00011333314984238197, "loss": 1.8834, "step": 4721 }, { "epoch": 1.42, "learning_rate": 0.0001133009411846671, "loss": 1.9634, "step": 4722 }, { "epoch": 1.42, "learning_rate": 0.00011326873112220342, "loss": 1.9007, "step": 4723 }, { "epoch": 1.42, "learning_rate": 0.0001132365196583927, "loss": 1.8084, "step": 4724 }, { "epoch": 1.42, "learning_rate": 0.00011320430679663692, "loss": 1.7847, "step": 4725 }, { "epoch": 1.42, "learning_rate": 0.00011317209254033814, "loss": 1.8741, "step": 4726 }, { "epoch": 1.42, "learning_rate": 0.00011313987689289856, "loss": 1.8447, "step": 4727 }, { "epoch": 1.42, "learning_rate": 0.0001131076598577206, "loss": 1.8532, "step": 4728 }, { "epoch": 1.42, "learning_rate": 0.00011307544143820679, "loss": 1.8088, "step": 4729 }, { "epoch": 1.42, "learning_rate": 0.00011304322163775982, "loss": 1.787, "step": 4730 }, { "epoch": 1.42, "learning_rate": 0.00011301100045978246, "loss": 1.755, "step": 4731 }, { "epoch": 1.42, "learning_rate": 0.00011297877790767774, "loss": 1.8462, "step": 4732 }, { "epoch": 1.42, "learning_rate": 0.00011294655398484874, "loss": 1.7834, "step": 4733 }, { "epoch": 1.43, "learning_rate": 0.00011291432869469867, "loss": 1.8236, "step": 4734 }, { "epoch": 1.43, "learning_rate": 0.00011288210204063104, "loss": 1.8964, "step": 4735 }, { "epoch": 1.43, "learning_rate": 0.00011284987402604929, "loss": 1.8894, "step": 4736 }, { "epoch": 1.43, "learning_rate": 0.0001128176446543572, "loss": 1.7845, "step": 4737 }, { "epoch": 1.43, "learning_rate": 0.00011278541392895852, "loss": 1.7596, "step": 4738 }, { "epoch": 1.43, "learning_rate": 0.00011275318185325728, "loss": 1.8188, "step": 4739 }, { "epoch": 1.43, "learning_rate": 0.00011272094843065754, "loss": 1.8022, "step": 4740 }, { "epoch": 1.43, "learning_rate": 0.00011268871366456362, "loss": 1.639, "step": 4741 }, { "epoch": 1.43, "learning_rate": 0.00011265647755837991, "loss": 1.8869, "step": 4742 }, { "epoch": 1.43, "learning_rate": 0.00011262424011551094, "loss": 1.9525, "step": 4743 }, { "epoch": 1.43, "learning_rate": 0.00011259200133936141, "loss": 1.8643, "step": 4744 }, { "epoch": 1.43, "learning_rate": 0.00011255976123333611, "loss": 1.804, "step": 4745 }, { "epoch": 1.43, "learning_rate": 0.00011252751980084004, "loss": 1.8253, "step": 4746 }, { "epoch": 1.43, "learning_rate": 0.00011249527704527832, "loss": 1.8356, "step": 4747 }, { "epoch": 1.43, "learning_rate": 0.00011246303297005614, "loss": 1.8147, "step": 4748 }, { "epoch": 1.43, "learning_rate": 0.00011243078757857892, "loss": 1.8021, "step": 4749 }, { "epoch": 1.43, "learning_rate": 0.0001123985408742522, "loss": 1.8434, "step": 4750 }, { "epoch": 1.43, "learning_rate": 0.0001123662928604816, "loss": 1.7884, "step": 4751 }, { "epoch": 1.43, "learning_rate": 0.00011233404354067294, "loss": 1.775, "step": 4752 }, { "epoch": 1.43, "learning_rate": 0.00011230179291823218, "loss": 1.8206, "step": 4753 }, { "epoch": 1.43, "learning_rate": 0.00011226954099656537, "loss": 1.9233, "step": 4754 }, { "epoch": 1.43, "learning_rate": 0.00011223728777907874, "loss": 1.951, "step": 4755 }, { "epoch": 1.43, "learning_rate": 0.00011220503326917865, "loss": 1.7176, "step": 4756 }, { "epoch": 1.43, "learning_rate": 0.00011217277747027155, "loss": 1.7421, "step": 4757 }, { "epoch": 1.43, "learning_rate": 0.00011214052038576408, "loss": 1.9448, "step": 4758 }, { "epoch": 1.43, "learning_rate": 0.00011210826201906301, "loss": 1.9177, "step": 4759 }, { "epoch": 1.43, "learning_rate": 0.00011207600237357523, "loss": 1.8022, "step": 4760 }, { "epoch": 1.43, "learning_rate": 0.0001120437414527078, "loss": 1.9202, "step": 4761 }, { "epoch": 1.43, "learning_rate": 0.00011201147925986779, "loss": 1.8138, "step": 4762 }, { "epoch": 1.43, "learning_rate": 0.00011197921579846262, "loss": 1.788, "step": 4763 }, { "epoch": 1.43, "learning_rate": 0.00011194695107189961, "loss": 1.7861, "step": 4764 }, { "epoch": 1.43, "learning_rate": 0.0001119146850835864, "loss": 1.8305, "step": 4765 }, { "epoch": 1.43, "learning_rate": 0.0001118824178369307, "loss": 1.7203, "step": 4766 }, { "epoch": 1.43, "learning_rate": 0.00011185014933534026, "loss": 1.8432, "step": 4767 }, { "epoch": 1.44, "learning_rate": 0.00011181787958222314, "loss": 1.805, "step": 4768 }, { "epoch": 1.44, "learning_rate": 0.00011178560858098735, "loss": 1.7701, "step": 4769 }, { "epoch": 1.44, "learning_rate": 0.00011175333633504117, "loss": 1.8455, "step": 4770 }, { "epoch": 1.44, "learning_rate": 0.00011172106284779298, "loss": 1.7709, "step": 4771 }, { "epoch": 1.44, "learning_rate": 0.00011168878812265123, "loss": 1.7788, "step": 4772 }, { "epoch": 1.44, "learning_rate": 0.00011165651216302455, "loss": 1.988, "step": 4773 }, { "epoch": 1.44, "learning_rate": 0.00011162423497232171, "loss": 1.7796, "step": 4774 }, { "epoch": 1.44, "learning_rate": 0.00011159195655395158, "loss": 1.7637, "step": 4775 }, { "epoch": 1.44, "learning_rate": 0.00011155967691132314, "loss": 1.8201, "step": 4776 }, { "epoch": 1.44, "learning_rate": 0.00011152739604784559, "loss": 1.864, "step": 4777 }, { "epoch": 1.44, "learning_rate": 0.00011149511396692818, "loss": 1.7793, "step": 4778 }, { "epoch": 1.44, "learning_rate": 0.0001114628306719803, "loss": 1.8317, "step": 4779 }, { "epoch": 1.44, "learning_rate": 0.00011143054616641146, "loss": 1.9529, "step": 4780 }, { "epoch": 1.44, "learning_rate": 0.00011139826045363135, "loss": 1.8455, "step": 4781 }, { "epoch": 1.44, "learning_rate": 0.00011136597353704971, "loss": 1.8484, "step": 4782 }, { "epoch": 1.44, "learning_rate": 0.00011133368542007654, "loss": 1.7674, "step": 4783 }, { "epoch": 1.44, "learning_rate": 0.00011130139610612176, "loss": 1.8276, "step": 4784 }, { "epoch": 1.44, "learning_rate": 0.00011126910559859562, "loss": 1.837, "step": 4785 }, { "epoch": 1.44, "learning_rate": 0.00011123681390090837, "loss": 1.8526, "step": 4786 }, { "epoch": 1.44, "learning_rate": 0.00011120452101647043, "loss": 1.7268, "step": 4787 }, { "epoch": 1.44, "learning_rate": 0.00011117222694869229, "loss": 1.8785, "step": 4788 }, { "epoch": 1.44, "learning_rate": 0.00011113993170098474, "loss": 1.7103, "step": 4789 }, { "epoch": 1.44, "learning_rate": 0.00011110763527675848, "loss": 1.8331, "step": 4790 }, { "epoch": 1.44, "learning_rate": 0.0001110753376794244, "loss": 1.8031, "step": 4791 }, { "epoch": 1.44, "learning_rate": 0.0001110430389123936, "loss": 1.8128, "step": 4792 }, { "epoch": 1.44, "learning_rate": 0.00011101073897907721, "loss": 1.8072, "step": 4793 }, { "epoch": 1.44, "learning_rate": 0.00011097843788288653, "loss": 1.8359, "step": 4794 }, { "epoch": 1.44, "learning_rate": 0.00011094613562723295, "loss": 1.9339, "step": 4795 }, { "epoch": 1.44, "learning_rate": 0.00011091383221552803, "loss": 1.7893, "step": 4796 }, { "epoch": 1.44, "learning_rate": 0.00011088152765118338, "loss": 1.812, "step": 4797 }, { "epoch": 1.44, "learning_rate": 0.00011084922193761081, "loss": 1.8422, "step": 4798 }, { "epoch": 1.44, "learning_rate": 0.0001108169150782222, "loss": 1.7876, "step": 4799 }, { "epoch": 1.44, "learning_rate": 0.00011078460707642952, "loss": 1.8328, "step": 4800 }, { "epoch": 1.45, "learning_rate": 0.00011075229793564501, "loss": 1.8278, "step": 4801 }, { "epoch": 1.45, "learning_rate": 0.00011071998765928084, "loss": 1.9302, "step": 4802 }, { "epoch": 1.45, "learning_rate": 0.00011068767625074943, "loss": 1.7511, "step": 4803 }, { "epoch": 1.45, "learning_rate": 0.00011065536371346325, "loss": 1.7332, "step": 4804 }, { "epoch": 1.45, "learning_rate": 0.00011062305005083494, "loss": 1.7581, "step": 4805 }, { "epoch": 1.45, "learning_rate": 0.0001105907352662772, "loss": 1.8124, "step": 4806 }, { "epoch": 1.45, "learning_rate": 0.0001105584193632029, "loss": 1.8208, "step": 4807 }, { "epoch": 1.45, "learning_rate": 0.00011052610234502505, "loss": 1.87, "step": 4808 }, { "epoch": 1.45, "learning_rate": 0.00011049378421515671, "loss": 1.8757, "step": 4809 }, { "epoch": 1.45, "learning_rate": 0.00011046146497701105, "loss": 1.9077, "step": 4810 }, { "epoch": 1.45, "learning_rate": 0.00011042914463400143, "loss": 1.7041, "step": 4811 }, { "epoch": 1.45, "learning_rate": 0.00011039682318954125, "loss": 1.7536, "step": 4812 }, { "epoch": 1.45, "learning_rate": 0.00011036450064704416, "loss": 1.9555, "step": 4813 }, { "epoch": 1.45, "learning_rate": 0.00011033217700992373, "loss": 1.8318, "step": 4814 }, { "epoch": 1.45, "learning_rate": 0.0001102998522815938, "loss": 1.8029, "step": 4815 }, { "epoch": 1.45, "learning_rate": 0.0001102675264654683, "loss": 1.8875, "step": 4816 }, { "epoch": 1.45, "learning_rate": 0.00011023519956496117, "loss": 1.7551, "step": 4817 }, { "epoch": 1.45, "learning_rate": 0.00011020287158348659, "loss": 1.8697, "step": 4818 }, { "epoch": 1.45, "learning_rate": 0.0001101705425244588, "loss": 1.8537, "step": 4819 }, { "epoch": 1.45, "learning_rate": 0.00011013821239129216, "loss": 1.7862, "step": 4820 }, { "epoch": 1.45, "learning_rate": 0.00011010588118740115, "loss": 1.79, "step": 4821 }, { "epoch": 1.45, "learning_rate": 0.00011007354891620036, "loss": 1.8735, "step": 4822 }, { "epoch": 1.45, "learning_rate": 0.0001100412155811045, "loss": 1.8544, "step": 4823 }, { "epoch": 1.45, "learning_rate": 0.00011000888118552831, "loss": 1.8293, "step": 4824 }, { "epoch": 1.45, "learning_rate": 0.00010997654573288678, "loss": 1.7351, "step": 4825 }, { "epoch": 1.45, "learning_rate": 0.00010994420922659494, "loss": 1.8058, "step": 4826 }, { "epoch": 1.45, "learning_rate": 0.00010991187167006795, "loss": 1.9452, "step": 4827 }, { "epoch": 1.45, "learning_rate": 0.00010987953306672102, "loss": 1.9166, "step": 4828 }, { "epoch": 1.45, "learning_rate": 0.00010984719341996958, "loss": 1.7347, "step": 4829 }, { "epoch": 1.45, "learning_rate": 0.00010981485273322904, "loss": 1.7944, "step": 4830 }, { "epoch": 1.45, "learning_rate": 0.00010978251100991503, "loss": 1.8244, "step": 4831 }, { "epoch": 1.45, "learning_rate": 0.00010975016825344328, "loss": 1.866, "step": 4832 }, { "epoch": 1.45, "learning_rate": 0.00010971782446722951, "loss": 1.8136, "step": 4833 }, { "epoch": 1.46, "learning_rate": 0.00010968547965468974, "loss": 1.817, "step": 4834 }, { "epoch": 1.46, "learning_rate": 0.00010965313381923987, "loss": 1.8992, "step": 4835 }, { "epoch": 1.46, "learning_rate": 0.00010962078696429614, "loss": 1.8902, "step": 4836 }, { "epoch": 1.46, "learning_rate": 0.00010958843909327477, "loss": 1.8667, "step": 4837 }, { "epoch": 1.46, "learning_rate": 0.00010955609020959203, "loss": 1.9056, "step": 4838 }, { "epoch": 1.46, "learning_rate": 0.00010952374031666449, "loss": 1.837, "step": 4839 }, { "epoch": 1.46, "learning_rate": 0.00010949138941790862, "loss": 1.7712, "step": 4840 }, { "epoch": 1.46, "learning_rate": 0.00010945903751674113, "loss": 1.9373, "step": 4841 }, { "epoch": 1.46, "learning_rate": 0.00010942668461657879, "loss": 1.8265, "step": 4842 }, { "epoch": 1.46, "learning_rate": 0.00010939433072083844, "loss": 1.816, "step": 4843 }, { "epoch": 1.46, "learning_rate": 0.00010936197583293715, "loss": 1.8018, "step": 4844 }, { "epoch": 1.46, "learning_rate": 0.00010932961995629192, "loss": 1.8166, "step": 4845 }, { "epoch": 1.46, "learning_rate": 0.00010929726309431998, "loss": 1.7373, "step": 4846 }, { "epoch": 1.46, "learning_rate": 0.00010926490525043862, "loss": 1.7887, "step": 4847 }, { "epoch": 1.46, "learning_rate": 0.00010923254642806522, "loss": 1.8364, "step": 4848 }, { "epoch": 1.46, "learning_rate": 0.00010920018663061736, "loss": 1.7839, "step": 4849 }, { "epoch": 1.46, "learning_rate": 0.00010916782586151255, "loss": 1.8799, "step": 4850 }, { "epoch": 1.46, "learning_rate": 0.00010913546412416857, "loss": 1.7415, "step": 4851 }, { "epoch": 1.46, "learning_rate": 0.00010910310142200319, "loss": 1.7392, "step": 4852 }, { "epoch": 1.46, "learning_rate": 0.00010907073775843432, "loss": 1.7579, "step": 4853 }, { "epoch": 1.46, "learning_rate": 0.00010903837313688003, "loss": 1.8468, "step": 4854 }, { "epoch": 1.46, "learning_rate": 0.00010900600756075838, "loss": 1.776, "step": 4855 }, { "epoch": 1.46, "learning_rate": 0.00010897364103348763, "loss": 1.8473, "step": 4856 }, { "epoch": 1.46, "learning_rate": 0.00010894127355848607, "loss": 1.8579, "step": 4857 }, { "epoch": 1.46, "learning_rate": 0.00010890890513917212, "loss": 1.8391, "step": 4858 }, { "epoch": 1.46, "learning_rate": 0.00010887653577896428, "loss": 1.7927, "step": 4859 }, { "epoch": 1.46, "learning_rate": 0.0001088441654812812, "loss": 1.7964, "step": 4860 }, { "epoch": 1.46, "learning_rate": 0.0001088117942495416, "loss": 1.8064, "step": 4861 }, { "epoch": 1.46, "learning_rate": 0.00010877942208716425, "loss": 1.8439, "step": 4862 }, { "epoch": 1.46, "learning_rate": 0.00010874704899756811, "loss": 1.829, "step": 4863 }, { "epoch": 1.46, "learning_rate": 0.00010871467498417214, "loss": 1.8023, "step": 4864 }, { "epoch": 1.46, "learning_rate": 0.0001086823000503955, "loss": 1.9069, "step": 4865 }, { "epoch": 1.46, "learning_rate": 0.00010864992419965738, "loss": 1.7948, "step": 4866 }, { "epoch": 1.47, "learning_rate": 0.00010861754743537707, "loss": 1.9116, "step": 4867 }, { "epoch": 1.47, "learning_rate": 0.00010858516976097397, "loss": 1.8322, "step": 4868 }, { "epoch": 1.47, "learning_rate": 0.00010855279117986758, "loss": 1.839, "step": 4869 }, { "epoch": 1.47, "learning_rate": 0.00010852041169547751, "loss": 1.8248, "step": 4870 }, { "epoch": 1.47, "learning_rate": 0.00010848803131122338, "loss": 1.7283, "step": 4871 }, { "epoch": 1.47, "learning_rate": 0.00010845565003052507, "loss": 1.8441, "step": 4872 }, { "epoch": 1.47, "learning_rate": 0.00010842326785680239, "loss": 1.8414, "step": 4873 }, { "epoch": 1.47, "learning_rate": 0.0001083908847934753, "loss": 1.8513, "step": 4874 }, { "epoch": 1.47, "learning_rate": 0.00010835850084396393, "loss": 1.8094, "step": 4875 }, { "epoch": 1.47, "learning_rate": 0.00010832611601168837, "loss": 1.8178, "step": 4876 }, { "epoch": 1.47, "learning_rate": 0.00010829373030006886, "loss": 1.8312, "step": 4877 }, { "epoch": 1.47, "learning_rate": 0.0001082613437125258, "loss": 1.8632, "step": 4878 }, { "epoch": 1.47, "learning_rate": 0.00010822895625247965, "loss": 1.9285, "step": 4879 }, { "epoch": 1.47, "learning_rate": 0.00010819656792335089, "loss": 1.7913, "step": 4880 }, { "epoch": 1.47, "learning_rate": 0.00010816417872856011, "loss": 1.8591, "step": 4881 }, { "epoch": 1.47, "learning_rate": 0.00010813178867152807, "loss": 4.3594, "step": 4882 }, { "epoch": 1.47, "learning_rate": 0.00010809939775567552, "loss": 2.8055, "step": 4883 }, { "epoch": 1.47, "learning_rate": 0.00010806700598442343, "loss": 1.879, "step": 4884 }, { "epoch": 1.47, "learning_rate": 0.00010803461336119274, "loss": 2.0127, "step": 4885 }, { "epoch": 1.47, "learning_rate": 0.00010800221988940454, "loss": 1.9412, "step": 4886 }, { "epoch": 1.47, "learning_rate": 0.00010796982557247999, "loss": 1.9098, "step": 4887 }, { "epoch": 1.47, "learning_rate": 0.0001079374304138403, "loss": 1.8477, "step": 4888 }, { "epoch": 1.47, "learning_rate": 0.00010790503441690684, "loss": 1.9775, "step": 4889 }, { "epoch": 1.47, "learning_rate": 0.00010787263758510106, "loss": 1.8085, "step": 4890 }, { "epoch": 1.47, "learning_rate": 0.00010784023992184451, "loss": 1.8062, "step": 4891 }, { "epoch": 1.47, "learning_rate": 0.00010780784143055869, "loss": 1.8871, "step": 4892 }, { "epoch": 1.47, "learning_rate": 0.0001077754421146654, "loss": 1.7934, "step": 4893 }, { "epoch": 1.47, "learning_rate": 0.00010774304197758635, "loss": 1.8707, "step": 4894 }, { "epoch": 1.47, "learning_rate": 0.00010771064102274341, "loss": 1.8678, "step": 4895 }, { "epoch": 1.47, "learning_rate": 0.0001076782392535586, "loss": 1.8642, "step": 4896 }, { "epoch": 1.47, "learning_rate": 0.0001076458366734539, "loss": 1.78, "step": 4897 }, { "epoch": 1.47, "learning_rate": 0.00010761343328585143, "loss": 1.8843, "step": 4898 }, { "epoch": 1.47, "learning_rate": 0.00010758102909417343, "loss": 1.8431, "step": 4899 }, { "epoch": 1.48, "learning_rate": 0.00010754862410184221, "loss": 1.8362, "step": 4900 }, { "epoch": 1.48, "learning_rate": 0.0001075162183122801, "loss": 1.8228, "step": 4901 }, { "epoch": 1.48, "learning_rate": 0.00010748381172890958, "loss": 1.7947, "step": 4902 }, { "epoch": 1.48, "learning_rate": 0.00010745140435515324, "loss": 1.8499, "step": 4903 }, { "epoch": 1.48, "learning_rate": 0.00010741899619443365, "loss": 1.8663, "step": 4904 }, { "epoch": 1.48, "learning_rate": 0.00010738658725017358, "loss": 1.8414, "step": 4905 }, { "epoch": 1.48, "learning_rate": 0.0001073541775257958, "loss": 1.8804, "step": 4906 }, { "epoch": 1.48, "learning_rate": 0.00010732176702472314, "loss": 1.863, "step": 4907 }, { "epoch": 1.48, "learning_rate": 0.00010728935575037867, "loss": 1.7835, "step": 4908 }, { "epoch": 1.48, "learning_rate": 0.00010725694370618536, "loss": 1.836, "step": 4909 }, { "epoch": 1.48, "learning_rate": 0.00010722453089556635, "loss": 1.8541, "step": 4910 }, { "epoch": 1.48, "learning_rate": 0.00010719211732194483, "loss": 1.8938, "step": 4911 }, { "epoch": 1.48, "learning_rate": 0.00010715970298874414, "loss": 1.7297, "step": 4912 }, { "epoch": 1.48, "learning_rate": 0.00010712728789938758, "loss": 1.8832, "step": 4913 }, { "epoch": 1.48, "learning_rate": 0.00010709487205729862, "loss": 1.7774, "step": 4914 }, { "epoch": 1.48, "learning_rate": 0.00010706245546590083, "loss": 1.8322, "step": 4915 }, { "epoch": 1.48, "learning_rate": 0.00010703003812861773, "loss": 1.9141, "step": 4916 }, { "epoch": 1.48, "learning_rate": 0.0001069976200488731, "loss": 1.8614, "step": 4917 }, { "epoch": 1.48, "learning_rate": 0.00010696520123009062, "loss": 1.8033, "step": 4918 }, { "epoch": 1.48, "learning_rate": 0.00010693278167569416, "loss": 1.8819, "step": 4919 }, { "epoch": 1.48, "learning_rate": 0.00010690036138910767, "loss": 1.7966, "step": 4920 }, { "epoch": 1.48, "learning_rate": 0.00010686794037375507, "loss": 1.875, "step": 4921 }, { "epoch": 1.48, "learning_rate": 0.00010683551863306053, "loss": 1.7936, "step": 4922 }, { "epoch": 1.48, "learning_rate": 0.00010680309617044812, "loss": 1.7796, "step": 4923 }, { "epoch": 1.48, "learning_rate": 0.0001067706729893421, "loss": 1.9363, "step": 4924 }, { "epoch": 1.48, "learning_rate": 0.00010673824909316674, "loss": 1.9235, "step": 4925 }, { "epoch": 1.48, "learning_rate": 0.00010670582448534645, "loss": 1.7967, "step": 4926 }, { "epoch": 1.48, "learning_rate": 0.0001066733991693057, "loss": 1.9531, "step": 4927 }, { "epoch": 1.48, "learning_rate": 0.00010664097314846898, "loss": 1.7933, "step": 4928 }, { "epoch": 1.48, "learning_rate": 0.00010660854642626088, "loss": 1.8461, "step": 4929 }, { "epoch": 1.48, "learning_rate": 0.0001065761190061061, "loss": 1.8116, "step": 4930 }, { "epoch": 1.48, "learning_rate": 0.00010654369089142937, "loss": 1.8737, "step": 4931 }, { "epoch": 1.48, "learning_rate": 0.00010651126208565554, "loss": 1.82, "step": 4932 }, { "epoch": 1.48, "learning_rate": 0.00010647883259220947, "loss": 1.8353, "step": 4933 }, { "epoch": 1.49, "learning_rate": 0.00010644640241451618, "loss": 1.8518, "step": 4934 }, { "epoch": 1.49, "learning_rate": 0.00010641397155600066, "loss": 1.857, "step": 4935 }, { "epoch": 1.49, "learning_rate": 0.00010638154002008804, "loss": 1.8814, "step": 4936 }, { "epoch": 1.49, "learning_rate": 0.00010634910781020348, "loss": 1.8064, "step": 4937 }, { "epoch": 1.49, "learning_rate": 0.00010631667492977228, "loss": 1.7733, "step": 4938 }, { "epoch": 1.49, "learning_rate": 0.00010628424138221976, "loss": 1.7937, "step": 4939 }, { "epoch": 1.49, "learning_rate": 0.00010625180717097125, "loss": 1.7861, "step": 4940 }, { "epoch": 1.49, "learning_rate": 0.00010621937229945231, "loss": 1.8468, "step": 4941 }, { "epoch": 1.49, "learning_rate": 0.0001061869367710884, "loss": 1.823, "step": 4942 }, { "epoch": 1.49, "learning_rate": 0.00010615450058930514, "loss": 1.6528, "step": 4943 }, { "epoch": 1.49, "learning_rate": 0.00010612206375752827, "loss": 1.8265, "step": 4944 }, { "epoch": 1.49, "learning_rate": 0.00010608962627918344, "loss": 1.9117, "step": 4945 }, { "epoch": 1.49, "learning_rate": 0.00010605718815769651, "loss": 1.7942, "step": 4946 }, { "epoch": 1.49, "learning_rate": 0.00010602474939649333, "loss": 1.8781, "step": 4947 }, { "epoch": 1.49, "learning_rate": 0.00010599230999899991, "loss": 1.8415, "step": 4948 }, { "epoch": 1.49, "learning_rate": 0.0001059598699686422, "loss": 1.7658, "step": 4949 }, { "epoch": 1.49, "learning_rate": 0.00010592742930884629, "loss": 1.7817, "step": 4950 }, { "epoch": 1.49, "learning_rate": 0.00010589498802303834, "loss": 1.9067, "step": 4951 }, { "epoch": 1.49, "learning_rate": 0.00010586254611464457, "loss": 1.8019, "step": 4952 }, { "epoch": 1.49, "learning_rate": 0.00010583010358709126, "loss": 1.8698, "step": 4953 }, { "epoch": 1.49, "learning_rate": 0.00010579766044380472, "loss": 1.8325, "step": 4954 }, { "epoch": 1.49, "learning_rate": 0.0001057652166882114, "loss": 1.9634, "step": 4955 }, { "epoch": 1.49, "learning_rate": 0.00010573277232373776, "loss": 1.9605, "step": 4956 }, { "epoch": 1.49, "learning_rate": 0.0001057003273538103, "loss": 1.6779, "step": 4957 }, { "epoch": 1.49, "learning_rate": 0.00010566788178185569, "loss": 1.9121, "step": 4958 }, { "epoch": 1.49, "learning_rate": 0.00010563543561130056, "loss": 1.7963, "step": 4959 }, { "epoch": 1.49, "learning_rate": 0.00010560298884557162, "loss": 1.7732, "step": 4960 }, { "epoch": 1.49, "learning_rate": 0.0001055705414880957, "loss": 1.7683, "step": 4961 }, { "epoch": 1.49, "learning_rate": 0.00010553809354229964, "loss": 1.9448, "step": 4962 }, { "epoch": 1.49, "learning_rate": 0.00010550564501161038, "loss": 1.7148, "step": 4963 }, { "epoch": 1.49, "learning_rate": 0.00010547319589945483, "loss": 1.9343, "step": 4964 }, { "epoch": 1.49, "learning_rate": 0.0001054407462092601, "loss": 1.8887, "step": 4965 }, { "epoch": 1.49, "learning_rate": 0.00010540829594445323, "loss": 1.8495, "step": 4966 }, { "epoch": 1.5, "learning_rate": 0.00010537584510846143, "loss": 1.78, "step": 4967 }, { "epoch": 1.5, "learning_rate": 0.00010534339370471191, "loss": 1.8764, "step": 4968 }, { "epoch": 1.5, "learning_rate": 0.00010531094173663196, "loss": 1.8803, "step": 4969 }, { "epoch": 1.5, "learning_rate": 0.00010527848920764891, "loss": 1.896, "step": 4970 }, { "epoch": 1.5, "learning_rate": 0.00010524603612119013, "loss": 1.7939, "step": 4971 }, { "epoch": 1.5, "learning_rate": 0.00010521358248068311, "loss": 1.873, "step": 4972 }, { "epoch": 1.5, "learning_rate": 0.00010518112828955538, "loss": 1.7916, "step": 4973 }, { "epoch": 1.5, "learning_rate": 0.0001051486735512345, "loss": 1.7225, "step": 4974 }, { "epoch": 1.5, "learning_rate": 0.0001051162182691481, "loss": 1.8839, "step": 4975 }, { "epoch": 1.5, "learning_rate": 0.00010508376244672389, "loss": 1.8371, "step": 4976 }, { "epoch": 1.5, "learning_rate": 0.00010505130608738963, "loss": 1.7261, "step": 4977 }, { "epoch": 1.5, "learning_rate": 0.00010501884919457303, "loss": 1.9472, "step": 4978 }, { "epoch": 1.5, "learning_rate": 0.00010498639177170206, "loss": 1.8114, "step": 4979 }, { "epoch": 1.5, "learning_rate": 0.0001049539338222046, "loss": 1.8099, "step": 4980 }, { "epoch": 1.5, "learning_rate": 0.00010492147534950865, "loss": 1.8766, "step": 4981 }, { "epoch": 1.5, "learning_rate": 0.00010488901635704216, "loss": 1.8534, "step": 4982 }, { "epoch": 1.5, "learning_rate": 0.0001048565568482333, "loss": 1.8218, "step": 4983 }, { "epoch": 1.5, "learning_rate": 0.00010482409682651016, "loss": 1.9819, "step": 4984 }, { "epoch": 1.5, "learning_rate": 0.00010479163629530093, "loss": 1.8755, "step": 4985 }, { "epoch": 1.5, "learning_rate": 0.00010475917525803389, "loss": 1.7751, "step": 4986 }, { "epoch": 1.5, "learning_rate": 0.0001047267137181373, "loss": 1.7996, "step": 4987 }, { "epoch": 1.5, "learning_rate": 0.00010469425167903954, "loss": 1.7474, "step": 4988 }, { "epoch": 1.5, "learning_rate": 0.000104661789144169, "loss": 1.8862, "step": 4989 }, { "epoch": 1.5, "learning_rate": 0.00010462932611695412, "loss": 1.7491, "step": 4990 }, { "epoch": 1.5, "learning_rate": 0.00010459686260082347, "loss": 1.8154, "step": 4991 }, { "epoch": 1.5, "learning_rate": 0.00010456439859920555, "loss": 1.7092, "step": 4992 }, { "epoch": 1.5, "learning_rate": 0.00010453193411552899, "loss": 1.7915, "step": 4993 }, { "epoch": 1.5, "learning_rate": 0.00010449946915322246, "loss": 1.835, "step": 4994 }, { "epoch": 1.5, "learning_rate": 0.0001044670037157147, "loss": 1.7295, "step": 4995 }, { "epoch": 1.5, "learning_rate": 0.00010443453780643442, "loss": 1.7791, "step": 4996 }, { "epoch": 1.5, "learning_rate": 0.00010440207142881044, "loss": 1.8955, "step": 4997 }, { "epoch": 1.5, "learning_rate": 0.00010436960458627167, "loss": 1.8896, "step": 4998 }, { "epoch": 1.5, "learning_rate": 0.00010433713728224697, "loss": 1.8261, "step": 4999 }, { "epoch": 1.51, "learning_rate": 0.00010430466952016535, "loss": 1.7708, "step": 5000 }, { "epoch": 1.51, "learning_rate": 0.00010427220130345576, "loss": 1.7673, "step": 5001 }, { "epoch": 1.51, "learning_rate": 0.0001042397326355473, "loss": 1.8044, "step": 5002 }, { "epoch": 1.51, "learning_rate": 0.0001042072635198691, "loss": 1.7889, "step": 5003 }, { "epoch": 1.51, "learning_rate": 0.00010417479395985024, "loss": 1.8171, "step": 5004 }, { "epoch": 1.51, "learning_rate": 0.00010414232395891996, "loss": 1.8305, "step": 5005 }, { "epoch": 1.51, "learning_rate": 0.0001041098535205075, "loss": 1.8164, "step": 5006 }, { "epoch": 1.51, "learning_rate": 0.00010407738264804219, "loss": 1.8121, "step": 5007 }, { "epoch": 1.51, "learning_rate": 0.00010404491134495329, "loss": 1.7461, "step": 5008 }, { "epoch": 1.51, "learning_rate": 0.00010401243961467025, "loss": 1.7688, "step": 5009 }, { "epoch": 1.51, "learning_rate": 0.00010397996746062247, "loss": 1.8188, "step": 5010 }, { "epoch": 1.51, "learning_rate": 0.00010394749488623941, "loss": 1.7963, "step": 5011 }, { "epoch": 1.51, "learning_rate": 0.00010391502189495064, "loss": 1.7906, "step": 5012 }, { "epoch": 1.51, "learning_rate": 0.00010388254849018564, "loss": 1.8148, "step": 5013 }, { "epoch": 1.51, "learning_rate": 0.00010385007467537407, "loss": 1.7726, "step": 5014 }, { "epoch": 1.51, "learning_rate": 0.0001038176004539456, "loss": 1.8558, "step": 5015 }, { "epoch": 1.51, "learning_rate": 0.00010378512582932986, "loss": 1.8404, "step": 5016 }, { "epoch": 1.51, "learning_rate": 0.00010375265080495664, "loss": 1.8466, "step": 5017 }, { "epoch": 1.51, "learning_rate": 0.00010372017538425569, "loss": 1.8311, "step": 5018 }, { "epoch": 1.51, "learning_rate": 0.00010368769957065683, "loss": 1.8042, "step": 5019 }, { "epoch": 1.51, "learning_rate": 0.00010365522336758991, "loss": 1.7945, "step": 5020 }, { "epoch": 1.51, "learning_rate": 0.0001036227467784848, "loss": 1.8288, "step": 5021 }, { "epoch": 1.51, "learning_rate": 0.00010359026980677156, "loss": 1.7763, "step": 5022 }, { "epoch": 1.51, "learning_rate": 0.00010355779245588002, "loss": 1.8465, "step": 5023 }, { "epoch": 1.51, "learning_rate": 0.00010352531472924035, "loss": 1.7595, "step": 5024 }, { "epoch": 1.51, "learning_rate": 0.00010349283663028246, "loss": 1.788, "step": 5025 }, { "epoch": 1.51, "learning_rate": 0.00010346035816243657, "loss": 1.7938, "step": 5026 }, { "epoch": 1.51, "learning_rate": 0.00010342787932913278, "loss": 1.8483, "step": 5027 }, { "epoch": 1.51, "learning_rate": 0.00010339540013380125, "loss": 1.8772, "step": 5028 }, { "epoch": 1.51, "learning_rate": 0.00010336292057987222, "loss": 1.8659, "step": 5029 }, { "epoch": 1.51, "learning_rate": 0.00010333044067077595, "loss": 1.8631, "step": 5030 }, { "epoch": 1.51, "learning_rate": 0.0001032979604099427, "loss": 1.7338, "step": 5031 }, { "epoch": 1.51, "learning_rate": 0.00010326547980080283, "loss": 1.8924, "step": 5032 }, { "epoch": 1.52, "learning_rate": 0.0001032329988467867, "loss": 1.8752, "step": 5033 }, { "epoch": 1.52, "learning_rate": 0.00010320051755132469, "loss": 1.8414, "step": 5034 }, { "epoch": 1.52, "learning_rate": 0.00010316803591784727, "loss": 1.7151, "step": 5035 }, { "epoch": 1.52, "learning_rate": 0.00010313555394978491, "loss": 1.8812, "step": 5036 }, { "epoch": 1.52, "learning_rate": 0.00010310307165056809, "loss": 1.6997, "step": 5037 }, { "epoch": 1.52, "learning_rate": 0.00010307058902362739, "loss": 1.8372, "step": 5038 }, { "epoch": 1.52, "learning_rate": 0.00010303810607239337, "loss": 1.8792, "step": 5039 }, { "epoch": 1.52, "learning_rate": 0.00010300562280029664, "loss": 1.7458, "step": 5040 }, { "epoch": 1.52, "learning_rate": 0.00010297313921076788, "loss": 1.9163, "step": 5041 }, { "epoch": 1.52, "learning_rate": 0.00010294065530723771, "loss": 1.8917, "step": 5042 }, { "epoch": 1.52, "learning_rate": 0.0001029081710931369, "loss": 1.8521, "step": 5043 }, { "epoch": 1.52, "learning_rate": 0.00010287568657189614, "loss": 1.8036, "step": 5044 }, { "epoch": 1.52, "learning_rate": 0.00010284320174694629, "loss": 1.8264, "step": 5045 }, { "epoch": 1.52, "learning_rate": 0.00010281071662171813, "loss": 1.858, "step": 5046 }, { "epoch": 1.52, "learning_rate": 0.00010277823119964245, "loss": 1.8985, "step": 5047 }, { "epoch": 1.52, "learning_rate": 0.00010274574548415019, "loss": 1.8791, "step": 5048 }, { "epoch": 1.52, "learning_rate": 0.00010271325947867221, "loss": 1.7674, "step": 5049 }, { "epoch": 1.52, "learning_rate": 0.00010268077318663947, "loss": 1.7744, "step": 5050 }, { "epoch": 1.52, "learning_rate": 0.00010264828661148292, "loss": 1.798, "step": 5051 }, { "epoch": 1.52, "learning_rate": 0.0001026157997566336, "loss": 1.7939, "step": 5052 }, { "epoch": 1.52, "learning_rate": 0.00010258331262552249, "loss": 1.8328, "step": 5053 }, { "epoch": 1.52, "learning_rate": 0.00010255082522158066, "loss": 1.762, "step": 5054 }, { "epoch": 1.52, "learning_rate": 0.00010251833754823921, "loss": 1.8453, "step": 5055 }, { "epoch": 1.52, "learning_rate": 0.00010248584960892917, "loss": 1.7553, "step": 5056 }, { "epoch": 1.52, "learning_rate": 0.00010245336140708182, "loss": 1.829, "step": 5057 }, { "epoch": 1.52, "learning_rate": 0.00010242087294612821, "loss": 1.7986, "step": 5058 }, { "epoch": 1.52, "learning_rate": 0.00010238838422949959, "loss": 1.888, "step": 5059 }, { "epoch": 1.52, "learning_rate": 0.00010235589526062719, "loss": 1.9337, "step": 5060 }, { "epoch": 1.52, "learning_rate": 0.00010232340604294222, "loss": 1.8518, "step": 5061 }, { "epoch": 1.52, "learning_rate": 0.00010229091657987597, "loss": 1.6809, "step": 5062 }, { "epoch": 1.52, "learning_rate": 0.00010225842687485974, "loss": 1.9038, "step": 5063 }, { "epoch": 1.52, "learning_rate": 0.0001022259369313249, "loss": 1.8635, "step": 5064 }, { "epoch": 1.52, "learning_rate": 0.00010219344675270273, "loss": 1.81, "step": 5065 }, { "epoch": 1.52, "learning_rate": 0.00010216095634242464, "loss": 1.834, "step": 5066 }, { "epoch": 1.53, "learning_rate": 0.00010212846570392208, "loss": 1.8167, "step": 5067 }, { "epoch": 1.53, "learning_rate": 0.00010209597484062634, "loss": 1.7962, "step": 5068 }, { "epoch": 1.53, "learning_rate": 0.00010206348375596904, "loss": 1.9189, "step": 5069 }, { "epoch": 1.53, "learning_rate": 0.00010203099245338152, "loss": 1.9728, "step": 5070 }, { "epoch": 1.53, "learning_rate": 0.00010199850093629536, "loss": 1.8919, "step": 5071 }, { "epoch": 1.53, "learning_rate": 0.00010196600920814201, "loss": 1.863, "step": 5072 }, { "epoch": 1.53, "learning_rate": 0.00010193351727235307, "loss": 1.8432, "step": 5073 }, { "epoch": 1.53, "learning_rate": 0.00010190102513236006, "loss": 1.784, "step": 5074 }, { "epoch": 1.53, "learning_rate": 0.00010186853279159458, "loss": 1.7568, "step": 5075 }, { "epoch": 1.53, "learning_rate": 0.00010183604025348825, "loss": 1.9161, "step": 5076 }, { "epoch": 1.53, "learning_rate": 0.00010180354752147265, "loss": 1.8019, "step": 5077 }, { "epoch": 1.53, "learning_rate": 0.00010177105459897951, "loss": 1.8418, "step": 5078 }, { "epoch": 1.53, "learning_rate": 0.00010173856148944039, "loss": 1.7451, "step": 5079 }, { "epoch": 1.53, "learning_rate": 0.00010170606819628706, "loss": 1.8324, "step": 5080 }, { "epoch": 1.53, "learning_rate": 0.00010167357472295118, "loss": 1.8004, "step": 5081 }, { "epoch": 1.53, "learning_rate": 0.0001016410810728645, "loss": 1.8696, "step": 5082 }, { "epoch": 1.53, "learning_rate": 0.00010160858724945876, "loss": 1.7564, "step": 5083 }, { "epoch": 1.53, "learning_rate": 0.00010157609325616568, "loss": 1.8096, "step": 5084 }, { "epoch": 1.53, "learning_rate": 0.0001015435990964171, "loss": 1.8706, "step": 5085 }, { "epoch": 1.53, "learning_rate": 0.00010151110477364481, "loss": 1.7671, "step": 5086 }, { "epoch": 1.53, "learning_rate": 0.00010147861029128059, "loss": 1.8276, "step": 5087 }, { "epoch": 1.53, "learning_rate": 0.00010144611565275628, "loss": 1.863, "step": 5088 }, { "epoch": 1.53, "learning_rate": 0.00010141362086150374, "loss": 1.7859, "step": 5089 }, { "epoch": 1.53, "learning_rate": 0.00010138112592095484, "loss": 1.7934, "step": 5090 }, { "epoch": 1.53, "learning_rate": 0.00010134863083454143, "loss": 1.8644, "step": 5091 }, { "epoch": 1.53, "learning_rate": 0.00010131613560569545, "loss": 1.8276, "step": 5092 }, { "epoch": 1.53, "learning_rate": 0.00010128364023784875, "loss": 1.776, "step": 5093 }, { "epoch": 1.53, "learning_rate": 0.00010125114473443331, "loss": 1.8899, "step": 5094 }, { "epoch": 1.53, "learning_rate": 0.00010121864909888106, "loss": 1.9012, "step": 5095 }, { "epoch": 1.53, "learning_rate": 0.00010118615333462392, "loss": 1.9606, "step": 5096 }, { "epoch": 1.53, "learning_rate": 0.00010115365744509389, "loss": 1.8574, "step": 5097 }, { "epoch": 1.53, "learning_rate": 0.00010112116143372293, "loss": 1.8047, "step": 5098 }, { "epoch": 1.53, "learning_rate": 0.00010108866530394303, "loss": 1.8077, "step": 5099 }, { "epoch": 1.54, "learning_rate": 0.00010105616905918625, "loss": 1.7648, "step": 5100 }, { "epoch": 1.54, "learning_rate": 0.00010102367270288451, "loss": 1.8874, "step": 5101 }, { "epoch": 1.54, "learning_rate": 0.00010099117623846994, "loss": 1.837, "step": 5102 }, { "epoch": 1.54, "learning_rate": 0.00010095867966937448, "loss": 1.8459, "step": 5103 }, { "epoch": 1.54, "learning_rate": 0.00010092618299903027, "loss": 1.7989, "step": 5104 }, { "epoch": 1.54, "learning_rate": 0.00010089368623086934, "loss": 1.8104, "step": 5105 }, { "epoch": 1.54, "learning_rate": 0.00010086118936832374, "loss": 1.7434, "step": 5106 }, { "epoch": 1.54, "learning_rate": 0.0001008286924148256, "loss": 1.7539, "step": 5107 }, { "epoch": 1.54, "learning_rate": 0.00010079619537380695, "loss": 1.8197, "step": 5108 }, { "epoch": 1.54, "learning_rate": 0.00010076369824869993, "loss": 1.8298, "step": 5109 }, { "epoch": 1.54, "learning_rate": 0.00010073120104293667, "loss": 1.8447, "step": 5110 }, { "epoch": 1.54, "learning_rate": 0.00010069870375994926, "loss": 1.824, "step": 5111 }, { "epoch": 1.54, "learning_rate": 0.00010066620640316984, "loss": 1.8181, "step": 5112 }, { "epoch": 1.54, "learning_rate": 0.00010063370897603051, "loss": 1.906, "step": 5113 }, { "epoch": 1.54, "learning_rate": 0.00010060121148196346, "loss": 1.8373, "step": 5114 }, { "epoch": 1.54, "learning_rate": 0.00010056871392440079, "loss": 1.8285, "step": 5115 }, { "epoch": 1.54, "learning_rate": 0.00010053621630677473, "loss": 1.9558, "step": 5116 }, { "epoch": 1.54, "learning_rate": 0.00010050371863251737, "loss": 1.7469, "step": 5117 }, { "epoch": 1.54, "learning_rate": 0.0001004712209050609, "loss": 1.8314, "step": 5118 }, { "epoch": 1.54, "learning_rate": 0.00010043872312783753, "loss": 1.8645, "step": 5119 }, { "epoch": 1.54, "learning_rate": 0.00010040622530427937, "loss": 1.9042, "step": 5120 }, { "epoch": 1.54, "learning_rate": 0.00010037372743781865, "loss": 1.8474, "step": 5121 }, { "epoch": 1.54, "learning_rate": 0.00010034122953188756, "loss": 1.9621, "step": 5122 }, { "epoch": 1.54, "learning_rate": 0.00010030873158991831, "loss": 1.8233, "step": 5123 }, { "epoch": 1.54, "learning_rate": 0.00010027623361534307, "loss": 1.8784, "step": 5124 }, { "epoch": 1.54, "learning_rate": 0.00010024373561159399, "loss": 1.8177, "step": 5125 }, { "epoch": 1.54, "learning_rate": 0.00010021123758210338, "loss": 1.7935, "step": 5126 }, { "epoch": 1.54, "learning_rate": 0.00010017873953030334, "loss": 1.8708, "step": 5127 }, { "epoch": 1.54, "learning_rate": 0.00010014624145962616, "loss": 1.769, "step": 5128 }, { "epoch": 1.54, "learning_rate": 0.00010011374337350401, "loss": 1.8386, "step": 5129 }, { "epoch": 1.54, "learning_rate": 0.00010008124527536913, "loss": 1.961, "step": 5130 }, { "epoch": 1.54, "learning_rate": 0.00010004874716865372, "loss": 1.8113, "step": 5131 }, { "epoch": 1.54, "learning_rate": 0.00010001624905678995, "loss": 1.8172, "step": 5132 }, { "epoch": 1.55, "learning_rate": 9.99837509432101e-05, "loss": 1.8755, "step": 5133 }, { "epoch": 1.55, "learning_rate": 9.99512528313463e-05, "loss": 1.8283, "step": 5134 }, { "epoch": 1.55, "learning_rate": 9.991875472463091e-05, "loss": 1.8741, "step": 5135 }, { "epoch": 1.55, "learning_rate": 9.988625662649601e-05, "loss": 1.8003, "step": 5136 }, { "epoch": 1.55, "learning_rate": 9.985375854037383e-05, "loss": 1.9088, "step": 5137 }, { "epoch": 1.55, "learning_rate": 9.982126046969669e-05, "loss": 1.869, "step": 5138 }, { "epoch": 1.55, "learning_rate": 9.978876241789664e-05, "loss": 1.8733, "step": 5139 }, { "epoch": 1.55, "learning_rate": 9.975626438840604e-05, "loss": 1.8848, "step": 5140 }, { "epoch": 1.55, "learning_rate": 9.972376638465696e-05, "loss": 1.8138, "step": 5141 }, { "epoch": 1.55, "learning_rate": 9.96912684100817e-05, "loss": 1.7871, "step": 5142 }, { "epoch": 1.55, "learning_rate": 9.965877046811246e-05, "loss": 1.8049, "step": 5143 }, { "epoch": 1.55, "learning_rate": 9.962627256218135e-05, "loss": 1.8614, "step": 5144 }, { "epoch": 1.55, "learning_rate": 9.959377469572067e-05, "loss": 1.8953, "step": 5145 }, { "epoch": 1.55, "learning_rate": 9.95612768721625e-05, "loss": 1.7855, "step": 5146 }, { "epoch": 1.55, "learning_rate": 9.952877909493914e-05, "loss": 1.9166, "step": 5147 }, { "epoch": 1.55, "learning_rate": 9.949628136748264e-05, "loss": 1.8793, "step": 5148 }, { "epoch": 1.55, "learning_rate": 9.946378369322528e-05, "loss": 1.6843, "step": 5149 }, { "epoch": 1.55, "learning_rate": 9.943128607559924e-05, "loss": 1.8795, "step": 5150 }, { "epoch": 1.55, "learning_rate": 9.939878851803655e-05, "loss": 1.8135, "step": 5151 }, { "epoch": 1.55, "learning_rate": 9.936629102396954e-05, "loss": 1.7712, "step": 5152 }, { "epoch": 1.55, "learning_rate": 9.93337935968302e-05, "loss": 1.8018, "step": 5153 }, { "epoch": 1.55, "learning_rate": 9.930129624005075e-05, "loss": 1.7328, "step": 5154 }, { "epoch": 1.55, "learning_rate": 9.926879895706335e-05, "loss": 1.8317, "step": 5155 }, { "epoch": 1.55, "learning_rate": 9.923630175130006e-05, "loss": 1.8421, "step": 5156 }, { "epoch": 1.55, "learning_rate": 9.92038046261931e-05, "loss": 1.8801, "step": 5157 }, { "epoch": 1.55, "learning_rate": 9.917130758517443e-05, "loss": 1.8421, "step": 5158 }, { "epoch": 1.55, "learning_rate": 9.91388106316763e-05, "loss": 1.8104, "step": 5159 }, { "epoch": 1.55, "learning_rate": 9.910631376913068e-05, "loss": 1.7568, "step": 5160 }, { "epoch": 1.55, "learning_rate": 9.907381700096974e-05, "loss": 1.7589, "step": 5161 }, { "epoch": 1.55, "learning_rate": 9.904132033062553e-05, "loss": 1.9008, "step": 5162 }, { "epoch": 1.55, "learning_rate": 9.900882376153008e-05, "loss": 1.7957, "step": 5163 }, { "epoch": 1.55, "learning_rate": 9.897632729711554e-05, "loss": 1.856, "step": 5164 }, { "epoch": 1.55, "learning_rate": 9.894383094081379e-05, "loss": 1.7902, "step": 5165 }, { "epoch": 1.56, "learning_rate": 9.891133469605696e-05, "loss": 1.8202, "step": 5166 }, { "epoch": 1.56, "learning_rate": 9.887883856627708e-05, "loss": 1.7661, "step": 5167 }, { "epoch": 1.56, "learning_rate": 9.884634255490612e-05, "loss": 1.7806, "step": 5168 }, { "epoch": 1.56, "learning_rate": 9.881384666537609e-05, "loss": 1.7786, "step": 5169 }, { "epoch": 1.56, "learning_rate": 9.878135090111897e-05, "loss": 1.8707, "step": 5170 }, { "epoch": 1.56, "learning_rate": 9.874885526556673e-05, "loss": 1.8293, "step": 5171 }, { "epoch": 1.56, "learning_rate": 9.871635976215126e-05, "loss": 1.8359, "step": 5172 }, { "epoch": 1.56, "learning_rate": 9.868386439430456e-05, "loss": 1.7481, "step": 5173 }, { "epoch": 1.56, "learning_rate": 9.865136916545858e-05, "loss": 1.8533, "step": 5174 }, { "epoch": 1.56, "learning_rate": 9.861887407904517e-05, "loss": 1.7751, "step": 5175 }, { "epoch": 1.56, "learning_rate": 9.858637913849628e-05, "loss": 1.7708, "step": 5176 }, { "epoch": 1.56, "learning_rate": 9.855388434724373e-05, "loss": 1.8869, "step": 5177 }, { "epoch": 1.56, "learning_rate": 9.852138970871942e-05, "loss": 1.6812, "step": 5178 }, { "epoch": 1.56, "learning_rate": 9.848889522635521e-05, "loss": 1.7826, "step": 5179 }, { "epoch": 1.56, "learning_rate": 9.845640090358291e-05, "loss": 1.7819, "step": 5180 }, { "epoch": 1.56, "learning_rate": 9.842390674383433e-05, "loss": 1.8069, "step": 5181 }, { "epoch": 1.56, "learning_rate": 9.839141275054127e-05, "loss": 1.8337, "step": 5182 }, { "epoch": 1.56, "learning_rate": 9.835891892713554e-05, "loss": 1.7459, "step": 5183 }, { "epoch": 1.56, "learning_rate": 9.832642527704884e-05, "loss": 1.7918, "step": 5184 }, { "epoch": 1.56, "learning_rate": 9.829393180371295e-05, "loss": 1.8943, "step": 5185 }, { "epoch": 1.56, "learning_rate": 9.826143851055962e-05, "loss": 1.7148, "step": 5186 }, { "epoch": 1.56, "learning_rate": 9.822894540102053e-05, "loss": 1.8281, "step": 5187 }, { "epoch": 1.56, "learning_rate": 9.819645247852736e-05, "loss": 1.7399, "step": 5188 }, { "epoch": 1.56, "learning_rate": 9.816395974651177e-05, "loss": 1.8373, "step": 5189 }, { "epoch": 1.56, "learning_rate": 9.813146720840543e-05, "loss": 1.8582, "step": 5190 }, { "epoch": 1.56, "learning_rate": 9.809897486763996e-05, "loss": 1.7969, "step": 5191 }, { "epoch": 1.56, "learning_rate": 9.806648272764695e-05, "loss": 1.7155, "step": 5192 }, { "epoch": 1.56, "learning_rate": 9.803399079185801e-05, "loss": 1.8286, "step": 5193 }, { "epoch": 1.56, "learning_rate": 9.800149906370465e-05, "loss": 1.8989, "step": 5194 }, { "epoch": 1.56, "learning_rate": 9.796900754661848e-05, "loss": 1.9032, "step": 5195 }, { "epoch": 1.56, "learning_rate": 9.7936516244031e-05, "loss": 1.7653, "step": 5196 }, { "epoch": 1.56, "learning_rate": 9.790402515937365e-05, "loss": 1.7986, "step": 5197 }, { "epoch": 1.56, "learning_rate": 9.787153429607797e-05, "loss": 1.7984, "step": 5198 }, { "epoch": 1.57, "learning_rate": 9.783904365757537e-05, "loss": 1.7994, "step": 5199 }, { "epoch": 1.57, "learning_rate": 9.78065532472973e-05, "loss": 1.7756, "step": 5200 }, { "epoch": 1.57, "learning_rate": 9.777406306867514e-05, "loss": 1.8682, "step": 5201 }, { "epoch": 1.57, "learning_rate": 9.774157312514027e-05, "loss": 1.7892, "step": 5202 }, { "epoch": 1.57, "learning_rate": 9.770908342012406e-05, "loss": 1.8302, "step": 5203 }, { "epoch": 1.57, "learning_rate": 9.76765939570578e-05, "loss": 1.9468, "step": 5204 }, { "epoch": 1.57, "learning_rate": 9.764410473937286e-05, "loss": 1.8139, "step": 5205 }, { "epoch": 1.57, "learning_rate": 9.761161577050042e-05, "loss": 1.7852, "step": 5206 }, { "epoch": 1.57, "learning_rate": 9.75791270538718e-05, "loss": 1.7991, "step": 5207 }, { "epoch": 1.57, "learning_rate": 9.754663859291821e-05, "loss": 1.9332, "step": 5208 }, { "epoch": 1.57, "learning_rate": 9.751415039107083e-05, "loss": 1.8836, "step": 5209 }, { "epoch": 1.57, "learning_rate": 9.748166245176084e-05, "loss": 1.7935, "step": 5210 }, { "epoch": 1.57, "learning_rate": 9.744917477841937e-05, "loss": 1.7144, "step": 5211 }, { "epoch": 1.57, "learning_rate": 9.741668737447754e-05, "loss": 1.9222, "step": 5212 }, { "epoch": 1.57, "learning_rate": 9.738420024336642e-05, "loss": 1.7683, "step": 5213 }, { "epoch": 1.57, "learning_rate": 9.735171338851708e-05, "loss": 1.8861, "step": 5214 }, { "epoch": 1.57, "learning_rate": 9.731922681336057e-05, "loss": 1.793, "step": 5215 }, { "epoch": 1.57, "learning_rate": 9.728674052132781e-05, "loss": 1.8482, "step": 5216 }, { "epoch": 1.57, "learning_rate": 9.725425451584986e-05, "loss": 1.8455, "step": 5217 }, { "epoch": 1.57, "learning_rate": 9.722176880035759e-05, "loss": 1.8416, "step": 5218 }, { "epoch": 1.57, "learning_rate": 9.718928337828191e-05, "loss": 1.793, "step": 5219 }, { "epoch": 1.57, "learning_rate": 9.715679825305372e-05, "loss": 1.7643, "step": 5220 }, { "epoch": 1.57, "learning_rate": 9.712431342810385e-05, "loss": 1.7352, "step": 5221 }, { "epoch": 1.57, "learning_rate": 9.709182890686315e-05, "loss": 1.8591, "step": 5222 }, { "epoch": 1.57, "learning_rate": 9.705934469276232e-05, "loss": 1.7932, "step": 5223 }, { "epoch": 1.57, "learning_rate": 9.702686078923217e-05, "loss": 2.0162, "step": 5224 }, { "epoch": 1.57, "learning_rate": 9.699437719970338e-05, "loss": 1.751, "step": 5225 }, { "epoch": 1.57, "learning_rate": 9.696189392760666e-05, "loss": 1.8357, "step": 5226 }, { "epoch": 1.57, "learning_rate": 9.692941097637265e-05, "loss": 1.8771, "step": 5227 }, { "epoch": 1.57, "learning_rate": 9.689692834943193e-05, "loss": 1.7702, "step": 5228 }, { "epoch": 1.57, "learning_rate": 9.686444605021513e-05, "loss": 1.7996, "step": 5229 }, { "epoch": 1.57, "learning_rate": 9.683196408215276e-05, "loss": 1.8109, "step": 5230 }, { "epoch": 1.57, "learning_rate": 9.679948244867532e-05, "loss": 1.8786, "step": 5231 }, { "epoch": 1.57, "learning_rate": 9.676700115321334e-05, "loss": 1.9372, "step": 5232 }, { "epoch": 1.58, "learning_rate": 9.67345201991972e-05, "loss": 1.8315, "step": 5233 }, { "epoch": 1.58, "learning_rate": 9.670203959005733e-05, "loss": 1.7619, "step": 5234 }, { "epoch": 1.58, "learning_rate": 9.666955932922408e-05, "loss": 1.9568, "step": 5235 }, { "epoch": 1.58, "learning_rate": 9.663707942012781e-05, "loss": 1.7134, "step": 5236 }, { "epoch": 1.58, "learning_rate": 9.660459986619877e-05, "loss": 1.8014, "step": 5237 }, { "epoch": 1.58, "learning_rate": 9.657212067086724e-05, "loss": 1.897, "step": 5238 }, { "epoch": 1.58, "learning_rate": 9.653964183756345e-05, "loss": 1.7682, "step": 5239 }, { "epoch": 1.58, "learning_rate": 9.650716336971755e-05, "loss": 1.8042, "step": 5240 }, { "epoch": 1.58, "learning_rate": 9.647468527075972e-05, "loss": 1.7623, "step": 5241 }, { "epoch": 1.58, "learning_rate": 9.644220754411999e-05, "loss": 1.8094, "step": 5242 }, { "epoch": 1.58, "learning_rate": 9.640973019322848e-05, "loss": 1.631, "step": 5243 }, { "epoch": 1.58, "learning_rate": 9.637725322151521e-05, "loss": 1.9381, "step": 5244 }, { "epoch": 1.58, "learning_rate": 9.634477663241013e-05, "loss": 1.9221, "step": 5245 }, { "epoch": 1.58, "learning_rate": 9.631230042934322e-05, "loss": 1.9159, "step": 5246 }, { "epoch": 1.58, "learning_rate": 9.627982461574434e-05, "loss": 1.8342, "step": 5247 }, { "epoch": 1.58, "learning_rate": 9.62473491950434e-05, "loss": 1.8111, "step": 5248 }, { "epoch": 1.58, "learning_rate": 9.621487417067016e-05, "loss": 1.6973, "step": 5249 }, { "epoch": 1.58, "learning_rate": 9.618239954605443e-05, "loss": 1.9379, "step": 5250 }, { "epoch": 1.58, "learning_rate": 9.614992532462594e-05, "loss": 1.8626, "step": 5251 }, { "epoch": 1.58, "learning_rate": 9.611745150981439e-05, "loss": 1.8999, "step": 5252 }, { "epoch": 1.58, "learning_rate": 9.608497810504941e-05, "loss": 1.8368, "step": 5253 }, { "epoch": 1.58, "learning_rate": 9.605250511376061e-05, "loss": 1.8057, "step": 5254 }, { "epoch": 1.58, "learning_rate": 9.602003253937753e-05, "loss": 1.8236, "step": 5255 }, { "epoch": 1.58, "learning_rate": 9.598756038532979e-05, "loss": 1.8315, "step": 5256 }, { "epoch": 1.58, "learning_rate": 9.595508865504674e-05, "loss": 1.7434, "step": 5257 }, { "epoch": 1.58, "learning_rate": 9.592261735195785e-05, "loss": 1.8545, "step": 5258 }, { "epoch": 1.58, "learning_rate": 9.589014647949252e-05, "loss": 1.811, "step": 5259 }, { "epoch": 1.58, "learning_rate": 9.585767604108006e-05, "loss": 1.8576, "step": 5260 }, { "epoch": 1.58, "learning_rate": 9.58252060401498e-05, "loss": 1.864, "step": 5261 }, { "epoch": 1.58, "learning_rate": 9.579273648013093e-05, "loss": 1.909, "step": 5262 }, { "epoch": 1.58, "learning_rate": 9.576026736445271e-05, "loss": 1.8611, "step": 5263 }, { "epoch": 1.58, "learning_rate": 9.572779869654426e-05, "loss": 1.7912, "step": 5264 }, { "epoch": 1.58, "learning_rate": 9.56953304798347e-05, "loss": 1.8382, "step": 5265 }, { "epoch": 1.59, "learning_rate": 9.566286271775305e-05, "loss": 1.7223, "step": 5266 }, { "epoch": 1.59, "learning_rate": 9.563039541372834e-05, "loss": 1.8074, "step": 5267 }, { "epoch": 1.59, "learning_rate": 9.559792857118958e-05, "loss": 1.8276, "step": 5268 }, { "epoch": 1.59, "learning_rate": 9.556546219356562e-05, "loss": 1.9028, "step": 5269 }, { "epoch": 1.59, "learning_rate": 9.553299628428534e-05, "loss": 1.8994, "step": 5270 }, { "epoch": 1.59, "learning_rate": 9.550053084677755e-05, "loss": 1.8764, "step": 5271 }, { "epoch": 1.59, "learning_rate": 9.546806588447103e-05, "loss": 1.749, "step": 5272 }, { "epoch": 1.59, "learning_rate": 9.543560140079448e-05, "loss": 1.8737, "step": 5273 }, { "epoch": 1.59, "learning_rate": 9.540313739917653e-05, "loss": 1.7853, "step": 5274 }, { "epoch": 1.59, "learning_rate": 9.53706738830459e-05, "loss": 1.7603, "step": 5275 }, { "epoch": 1.59, "learning_rate": 9.533821085583103e-05, "loss": 1.8234, "step": 5276 }, { "epoch": 1.59, "learning_rate": 9.530574832096049e-05, "loss": 1.8584, "step": 5277 }, { "epoch": 1.59, "learning_rate": 9.527328628186273e-05, "loss": 1.7924, "step": 5278 }, { "epoch": 1.59, "learning_rate": 9.524082474196612e-05, "loss": 1.8406, "step": 5279 }, { "epoch": 1.59, "learning_rate": 9.52083637046991e-05, "loss": 1.7604, "step": 5280 }, { "epoch": 1.59, "learning_rate": 9.517590317348985e-05, "loss": 1.9006, "step": 5281 }, { "epoch": 1.59, "learning_rate": 9.514344315176674e-05, "loss": 1.8489, "step": 5282 }, { "epoch": 1.59, "learning_rate": 9.511098364295785e-05, "loss": 1.9016, "step": 5283 }, { "epoch": 1.59, "learning_rate": 9.507852465049141e-05, "loss": 1.8678, "step": 5284 }, { "epoch": 1.59, "learning_rate": 9.50460661777954e-05, "loss": 1.84, "step": 5285 }, { "epoch": 1.59, "learning_rate": 9.501360822829792e-05, "loss": 1.8793, "step": 5286 }, { "epoch": 1.59, "learning_rate": 9.498115080542699e-05, "loss": 1.8389, "step": 5287 }, { "epoch": 1.59, "learning_rate": 9.49486939126104e-05, "loss": 1.9935, "step": 5288 }, { "epoch": 1.59, "learning_rate": 9.491623755327615e-05, "loss": 1.805, "step": 5289 }, { "epoch": 1.59, "learning_rate": 9.488378173085192e-05, "loss": 1.8691, "step": 5290 }, { "epoch": 1.59, "learning_rate": 9.48513264487655e-05, "loss": 1.9251, "step": 5291 }, { "epoch": 1.59, "learning_rate": 9.481887171044466e-05, "loss": 1.8762, "step": 5292 }, { "epoch": 1.59, "learning_rate": 9.478641751931689e-05, "loss": 1.8001, "step": 5293 }, { "epoch": 1.59, "learning_rate": 9.475396387880992e-05, "loss": 1.7752, "step": 5294 }, { "epoch": 1.59, "learning_rate": 9.472151079235111e-05, "loss": 1.8997, "step": 5295 }, { "epoch": 1.59, "learning_rate": 9.468905826336809e-05, "loss": 1.7871, "step": 5296 }, { "epoch": 1.59, "learning_rate": 9.465660629528811e-05, "loss": 1.8398, "step": 5297 }, { "epoch": 1.59, "learning_rate": 9.462415489153857e-05, "loss": 1.8738, "step": 5298 }, { "epoch": 1.6, "learning_rate": 9.45917040555468e-05, "loss": 1.74, "step": 5299 }, { "epoch": 1.6, "learning_rate": 9.455925379073992e-05, "loss": 1.7949, "step": 5300 }, { "epoch": 1.6, "learning_rate": 9.452680410054522e-05, "loss": 1.8271, "step": 5301 }, { "epoch": 1.6, "learning_rate": 9.449435498838964e-05, "loss": 1.9158, "step": 5302 }, { "epoch": 1.6, "learning_rate": 9.446190645770034e-05, "loss": 1.8892, "step": 5303 }, { "epoch": 1.6, "learning_rate": 9.442945851190432e-05, "loss": 1.8481, "step": 5304 }, { "epoch": 1.6, "learning_rate": 9.439701115442837e-05, "loss": 1.7767, "step": 5305 }, { "epoch": 1.6, "learning_rate": 9.436456438869948e-05, "loss": 1.7368, "step": 5306 }, { "epoch": 1.6, "learning_rate": 9.433211821814432e-05, "loss": 1.7646, "step": 5307 }, { "epoch": 1.6, "learning_rate": 9.42996726461897e-05, "loss": 1.8673, "step": 5308 }, { "epoch": 1.6, "learning_rate": 9.426722767626227e-05, "loss": 1.7401, "step": 5309 }, { "epoch": 1.6, "learning_rate": 9.423478331178861e-05, "loss": 1.803, "step": 5310 }, { "epoch": 1.6, "learning_rate": 9.420233955619531e-05, "loss": 1.812, "step": 5311 }, { "epoch": 1.6, "learning_rate": 9.416989641290876e-05, "loss": 1.8625, "step": 5312 }, { "epoch": 1.6, "learning_rate": 9.413745388535547e-05, "loss": 1.8939, "step": 5313 }, { "epoch": 1.6, "learning_rate": 9.410501197696167e-05, "loss": 1.6689, "step": 5314 }, { "epoch": 1.6, "learning_rate": 9.407257069115372e-05, "loss": 1.9471, "step": 5315 }, { "epoch": 1.6, "learning_rate": 9.404013003135783e-05, "loss": 1.7651, "step": 5316 }, { "epoch": 1.6, "learning_rate": 9.40076900010001e-05, "loss": 1.757, "step": 5317 }, { "epoch": 1.6, "learning_rate": 9.397525060350669e-05, "loss": 1.7958, "step": 5318 }, { "epoch": 1.6, "learning_rate": 9.394281184230351e-05, "loss": 1.8687, "step": 5319 }, { "epoch": 1.6, "learning_rate": 9.391037372081657e-05, "loss": 1.8692, "step": 5320 }, { "epoch": 1.6, "learning_rate": 9.387793624247177e-05, "loss": 1.9592, "step": 5321 }, { "epoch": 1.6, "learning_rate": 9.384549941069485e-05, "loss": 1.7832, "step": 5322 }, { "epoch": 1.6, "learning_rate": 9.381306322891163e-05, "loss": 1.8284, "step": 5323 }, { "epoch": 1.6, "learning_rate": 9.378062770054772e-05, "loss": 1.964, "step": 5324 }, { "epoch": 1.6, "learning_rate": 9.374819282902878e-05, "loss": 1.8146, "step": 5325 }, { "epoch": 1.6, "learning_rate": 9.371575861778027e-05, "loss": 1.758, "step": 5326 }, { "epoch": 1.6, "learning_rate": 9.368332507022773e-05, "loss": 1.7309, "step": 5327 }, { "epoch": 1.6, "learning_rate": 9.365089218979653e-05, "loss": 1.8195, "step": 5328 }, { "epoch": 1.6, "learning_rate": 9.361845997991197e-05, "loss": 1.7928, "step": 5329 }, { "epoch": 1.6, "learning_rate": 9.358602844399936e-05, "loss": 1.686, "step": 5330 }, { "epoch": 1.6, "learning_rate": 9.355359758548383e-05, "loss": 1.8553, "step": 5331 }, { "epoch": 1.61, "learning_rate": 9.352116740779053e-05, "loss": 1.8197, "step": 5332 }, { "epoch": 1.61, "learning_rate": 9.348873791434448e-05, "loss": 1.841, "step": 5333 }, { "epoch": 1.61, "learning_rate": 9.345630910857064e-05, "loss": 1.8836, "step": 5334 }, { "epoch": 1.61, "learning_rate": 9.342388099389393e-05, "loss": 1.832, "step": 5335 }, { "epoch": 1.61, "learning_rate": 9.339145357373914e-05, "loss": 1.7929, "step": 5336 }, { "epoch": 1.61, "learning_rate": 9.335902685153109e-05, "loss": 1.9858, "step": 5337 }, { "epoch": 1.61, "learning_rate": 9.332660083069433e-05, "loss": 1.8297, "step": 5338 }, { "epoch": 1.61, "learning_rate": 9.329417551465354e-05, "loss": 1.6956, "step": 5339 }, { "epoch": 1.61, "learning_rate": 9.326175090683327e-05, "loss": 1.8665, "step": 5340 }, { "epoch": 1.61, "learning_rate": 9.322932701065792e-05, "loss": 1.8304, "step": 5341 }, { "epoch": 1.61, "learning_rate": 9.319690382955191e-05, "loss": 1.7811, "step": 5342 }, { "epoch": 1.61, "learning_rate": 9.316448136693949e-05, "loss": 1.8119, "step": 5343 }, { "epoch": 1.61, "learning_rate": 9.313205962624492e-05, "loss": 1.7433, "step": 5344 }, { "epoch": 1.61, "learning_rate": 9.309963861089235e-05, "loss": 1.6988, "step": 5345 }, { "epoch": 1.61, "learning_rate": 9.306721832430584e-05, "loss": 1.936, "step": 5346 }, { "epoch": 1.61, "learning_rate": 9.30347987699094e-05, "loss": 1.8199, "step": 5347 }, { "epoch": 1.61, "learning_rate": 9.300237995112693e-05, "loss": 1.7898, "step": 5348 }, { "epoch": 1.61, "learning_rate": 9.296996187138228e-05, "loss": 1.8524, "step": 5349 }, { "epoch": 1.61, "learning_rate": 9.29375445340992e-05, "loss": 1.8333, "step": 5350 }, { "epoch": 1.61, "learning_rate": 9.290512794270137e-05, "loss": 1.8676, "step": 5351 }, { "epoch": 1.61, "learning_rate": 9.287271210061243e-05, "loss": 1.8223, "step": 5352 }, { "epoch": 1.61, "learning_rate": 9.284029701125588e-05, "loss": 1.8392, "step": 5353 }, { "epoch": 1.61, "learning_rate": 9.280788267805518e-05, "loss": 1.7431, "step": 5354 }, { "epoch": 1.61, "learning_rate": 9.277546910443366e-05, "loss": 1.7497, "step": 5355 }, { "epoch": 1.61, "learning_rate": 9.274305629381465e-05, "loss": 1.8482, "step": 5356 }, { "epoch": 1.61, "learning_rate": 9.271064424962134e-05, "loss": 1.7702, "step": 5357 }, { "epoch": 1.61, "learning_rate": 9.267823297527686e-05, "loss": 1.7471, "step": 5358 }, { "epoch": 1.61, "learning_rate": 9.264582247420425e-05, "loss": 1.8465, "step": 5359 }, { "epoch": 1.61, "learning_rate": 9.261341274982644e-05, "loss": 1.7275, "step": 5360 }, { "epoch": 1.61, "learning_rate": 9.258100380556638e-05, "loss": 1.7871, "step": 5361 }, { "epoch": 1.61, "learning_rate": 9.25485956448468e-05, "loss": 1.8819, "step": 5362 }, { "epoch": 1.61, "learning_rate": 9.251618827109042e-05, "loss": 1.7967, "step": 5363 }, { "epoch": 1.61, "learning_rate": 9.248378168771993e-05, "loss": 1.8417, "step": 5364 }, { "epoch": 1.61, "learning_rate": 9.245137589815782e-05, "loss": 1.8456, "step": 5365 }, { "epoch": 1.62, "learning_rate": 9.241897090582658e-05, "loss": 1.8271, "step": 5366 }, { "epoch": 1.62, "learning_rate": 9.238656671414858e-05, "loss": 1.8577, "step": 5367 }, { "epoch": 1.62, "learning_rate": 9.235416332654611e-05, "loss": 1.8289, "step": 5368 }, { "epoch": 1.62, "learning_rate": 9.232176074644143e-05, "loss": 1.7839, "step": 5369 }, { "epoch": 1.62, "learning_rate": 9.228935897725658e-05, "loss": 1.7528, "step": 5370 }, { "epoch": 1.62, "learning_rate": 9.225695802241367e-05, "loss": 1.7928, "step": 5371 }, { "epoch": 1.62, "learning_rate": 9.222455788533463e-05, "loss": 1.9137, "step": 5372 }, { "epoch": 1.62, "learning_rate": 9.219215856944132e-05, "loss": 1.7691, "step": 5373 }, { "epoch": 1.62, "learning_rate": 9.215976007815552e-05, "loss": 1.7839, "step": 5374 }, { "epoch": 1.62, "learning_rate": 9.212736241489892e-05, "loss": 1.7671, "step": 5375 }, { "epoch": 1.62, "learning_rate": 9.209496558309317e-05, "loss": 1.9317, "step": 5376 }, { "epoch": 1.62, "learning_rate": 9.206256958615972e-05, "loss": 1.7651, "step": 5377 }, { "epoch": 1.62, "learning_rate": 9.203017442752006e-05, "loss": 1.9675, "step": 5378 }, { "epoch": 1.62, "learning_rate": 9.199778011059547e-05, "loss": 1.8067, "step": 5379 }, { "epoch": 1.62, "learning_rate": 9.196538663880727e-05, "loss": 1.8245, "step": 5380 }, { "epoch": 1.62, "learning_rate": 9.19329940155766e-05, "loss": 1.887, "step": 5381 }, { "epoch": 1.62, "learning_rate": 9.190060224432449e-05, "loss": 1.8194, "step": 5382 }, { "epoch": 1.62, "learning_rate": 9.186821132847197e-05, "loss": 1.7625, "step": 5383 }, { "epoch": 1.62, "learning_rate": 9.183582127143992e-05, "loss": 1.8517, "step": 5384 }, { "epoch": 1.62, "learning_rate": 9.180343207664916e-05, "loss": 1.7163, "step": 5385 }, { "epoch": 1.62, "learning_rate": 9.177104374752036e-05, "loss": 1.8583, "step": 5386 }, { "epoch": 1.62, "learning_rate": 9.173865628747418e-05, "loss": 1.8397, "step": 5387 }, { "epoch": 1.62, "learning_rate": 9.170626969993115e-05, "loss": 1.8432, "step": 5388 }, { "epoch": 1.62, "learning_rate": 9.167388398831167e-05, "loss": 1.8261, "step": 5389 }, { "epoch": 1.62, "learning_rate": 9.164149915603611e-05, "loss": 1.8106, "step": 5390 }, { "epoch": 1.62, "learning_rate": 9.160911520652471e-05, "loss": 1.8967, "step": 5391 }, { "epoch": 1.62, "learning_rate": 9.157673214319764e-05, "loss": 1.87, "step": 5392 }, { "epoch": 1.62, "learning_rate": 9.154434996947495e-05, "loss": 1.8271, "step": 5393 }, { "epoch": 1.62, "learning_rate": 9.151196868877661e-05, "loss": 1.8158, "step": 5394 }, { "epoch": 1.62, "learning_rate": 9.147958830452254e-05, "loss": 1.7084, "step": 5395 }, { "epoch": 1.62, "learning_rate": 9.144720882013243e-05, "loss": 1.8186, "step": 5396 }, { "epoch": 1.62, "learning_rate": 9.141483023902606e-05, "loss": 1.8519, "step": 5397 }, { "epoch": 1.62, "learning_rate": 9.138245256462297e-05, "loss": 1.8265, "step": 5398 }, { "epoch": 1.63, "learning_rate": 9.135007580034264e-05, "loss": 1.8434, "step": 5399 }, { "epoch": 1.63, "learning_rate": 9.131769994960452e-05, "loss": 1.8156, "step": 5400 }, { "epoch": 1.63, "learning_rate": 9.128532501582787e-05, "loss": 1.8367, "step": 5401 }, { "epoch": 1.63, "learning_rate": 9.125295100243194e-05, "loss": 1.9892, "step": 5402 }, { "epoch": 1.63, "learning_rate": 9.122057791283577e-05, "loss": 1.7971, "step": 5403 }, { "epoch": 1.63, "learning_rate": 9.118820575045843e-05, "loss": 1.8326, "step": 5404 }, { "epoch": 1.63, "learning_rate": 9.115583451871884e-05, "loss": 1.7985, "step": 5405 }, { "epoch": 1.63, "learning_rate": 9.112346422103573e-05, "loss": 1.9131, "step": 5406 }, { "epoch": 1.63, "learning_rate": 9.109109486082793e-05, "loss": 1.8053, "step": 5407 }, { "epoch": 1.63, "learning_rate": 9.105872644151397e-05, "loss": 1.7941, "step": 5408 }, { "epoch": 1.63, "learning_rate": 9.10263589665124e-05, "loss": 1.8849, "step": 5409 }, { "epoch": 1.63, "learning_rate": 9.099399243924163e-05, "loss": 1.7796, "step": 5410 }, { "epoch": 1.63, "learning_rate": 9.096162686311998e-05, "loss": 1.8062, "step": 5411 }, { "epoch": 1.63, "learning_rate": 9.09292622415657e-05, "loss": 1.8426, "step": 5412 }, { "epoch": 1.63, "learning_rate": 9.089689857799683e-05, "loss": 1.8837, "step": 5413 }, { "epoch": 1.63, "learning_rate": 9.086453587583146e-05, "loss": 1.8289, "step": 5414 }, { "epoch": 1.63, "learning_rate": 9.083217413848747e-05, "loss": 1.7774, "step": 5415 }, { "epoch": 1.63, "learning_rate": 9.079981336938267e-05, "loss": 1.7327, "step": 5416 }, { "epoch": 1.63, "learning_rate": 9.076745357193479e-05, "loss": 1.7927, "step": 5417 }, { "epoch": 1.63, "learning_rate": 9.07350947495614e-05, "loss": 1.803, "step": 5418 }, { "epoch": 1.63, "learning_rate": 9.070273690568006e-05, "loss": 1.8273, "step": 5419 }, { "epoch": 1.63, "learning_rate": 9.067038004370812e-05, "loss": 1.8613, "step": 5420 }, { "epoch": 1.63, "learning_rate": 9.063802416706289e-05, "loss": 1.9667, "step": 5421 }, { "epoch": 1.63, "learning_rate": 9.060566927916157e-05, "loss": 1.8979, "step": 5422 }, { "epoch": 1.63, "learning_rate": 9.057331538342123e-05, "loss": 1.8359, "step": 5423 }, { "epoch": 1.63, "learning_rate": 9.05409624832589e-05, "loss": 1.7826, "step": 5424 }, { "epoch": 1.63, "learning_rate": 9.05086105820914e-05, "loss": 1.8953, "step": 5425 }, { "epoch": 1.63, "learning_rate": 9.047625968333555e-05, "loss": 1.9307, "step": 5426 }, { "epoch": 1.63, "learning_rate": 9.044390979040798e-05, "loss": 1.9151, "step": 5427 }, { "epoch": 1.63, "learning_rate": 9.041156090672524e-05, "loss": 1.8079, "step": 5428 }, { "epoch": 1.63, "learning_rate": 9.037921303570387e-05, "loss": 1.778, "step": 5429 }, { "epoch": 1.63, "learning_rate": 9.034686618076014e-05, "loss": 1.8393, "step": 5430 }, { "epoch": 1.63, "learning_rate": 9.031452034531031e-05, "loss": 1.9246, "step": 5431 }, { "epoch": 1.64, "learning_rate": 9.02821755327705e-05, "loss": 1.7969, "step": 5432 }, { "epoch": 1.64, "learning_rate": 9.024983174655673e-05, "loss": 1.8099, "step": 5433 }, { "epoch": 1.64, "learning_rate": 9.021748899008498e-05, "loss": 1.7665, "step": 5434 }, { "epoch": 1.64, "learning_rate": 9.018514726677094e-05, "loss": 1.8753, "step": 5435 }, { "epoch": 1.64, "learning_rate": 9.015280658003045e-05, "loss": 1.8124, "step": 5436 }, { "epoch": 1.64, "learning_rate": 9.012046693327899e-05, "loss": 1.8876, "step": 5437 }, { "epoch": 1.64, "learning_rate": 9.008812832993208e-05, "loss": 1.8796, "step": 5438 }, { "epoch": 1.64, "learning_rate": 9.005579077340507e-05, "loss": 1.7616, "step": 5439 }, { "epoch": 1.64, "learning_rate": 9.00234542671132e-05, "loss": 1.7801, "step": 5440 }, { "epoch": 1.64, "learning_rate": 8.999111881447173e-05, "loss": 1.8564, "step": 5441 }, { "epoch": 1.64, "learning_rate": 8.995878441889554e-05, "loss": 1.7808, "step": 5442 }, { "epoch": 1.64, "learning_rate": 8.992645108379967e-05, "loss": 1.8727, "step": 5443 }, { "epoch": 1.64, "learning_rate": 8.989411881259887e-05, "loss": 1.8804, "step": 5444 }, { "epoch": 1.64, "learning_rate": 8.986178760870783e-05, "loss": 1.8522, "step": 5445 }, { "epoch": 1.64, "learning_rate": 8.982945747554124e-05, "loss": 1.8059, "step": 5446 }, { "epoch": 1.64, "learning_rate": 8.979712841651342e-05, "loss": 1.838, "step": 5447 }, { "epoch": 1.64, "learning_rate": 8.976480043503888e-05, "loss": 1.9144, "step": 5448 }, { "epoch": 1.64, "learning_rate": 8.973247353453172e-05, "loss": 1.8736, "step": 5449 }, { "epoch": 1.64, "learning_rate": 8.970014771840623e-05, "loss": 1.8454, "step": 5450 }, { "epoch": 1.64, "learning_rate": 8.96678229900763e-05, "loss": 1.8688, "step": 5451 }, { "epoch": 1.64, "learning_rate": 8.963549935295585e-05, "loss": 1.8677, "step": 5452 }, { "epoch": 1.64, "learning_rate": 8.960317681045877e-05, "loss": 1.7622, "step": 5453 }, { "epoch": 1.64, "learning_rate": 8.95708553659986e-05, "loss": 1.855, "step": 5454 }, { "epoch": 1.64, "learning_rate": 8.953853502298902e-05, "loss": 1.8679, "step": 5455 }, { "epoch": 1.64, "learning_rate": 8.950621578484333e-05, "loss": 1.7759, "step": 5456 }, { "epoch": 1.64, "learning_rate": 8.947389765497494e-05, "loss": 1.8768, "step": 5457 }, { "epoch": 1.64, "learning_rate": 8.944158063679712e-05, "loss": 1.8486, "step": 5458 }, { "epoch": 1.64, "learning_rate": 8.940926473372282e-05, "loss": 1.8836, "step": 5459 }, { "epoch": 1.64, "learning_rate": 8.937694994916511e-05, "loss": 1.9554, "step": 5460 }, { "epoch": 1.64, "learning_rate": 8.934463628653677e-05, "loss": 1.8262, "step": 5461 }, { "epoch": 1.64, "learning_rate": 8.931232374925061e-05, "loss": 1.7795, "step": 5462 }, { "epoch": 1.64, "learning_rate": 8.928001234071917e-05, "loss": 1.7521, "step": 5463 }, { "epoch": 1.64, "learning_rate": 8.9247702064355e-05, "loss": 1.8935, "step": 5464 }, { "epoch": 1.65, "learning_rate": 8.921539292357049e-05, "loss": 1.8815, "step": 5465 }, { "epoch": 1.65, "learning_rate": 8.918308492177782e-05, "loss": 1.8663, "step": 5466 }, { "epoch": 1.65, "learning_rate": 8.915077806238924e-05, "loss": 1.6902, "step": 5467 }, { "epoch": 1.65, "learning_rate": 8.911847234881663e-05, "loss": 1.7734, "step": 5468 }, { "epoch": 1.65, "learning_rate": 8.908616778447198e-05, "loss": 1.8491, "step": 5469 }, { "epoch": 1.65, "learning_rate": 8.905386437276708e-05, "loss": 1.7742, "step": 5470 }, { "epoch": 1.65, "learning_rate": 8.902156211711348e-05, "loss": 1.7966, "step": 5471 }, { "epoch": 1.65, "learning_rate": 8.898926102092284e-05, "loss": 1.8287, "step": 5472 }, { "epoch": 1.65, "learning_rate": 8.895696108760641e-05, "loss": 1.7362, "step": 5473 }, { "epoch": 1.65, "learning_rate": 8.892466232057565e-05, "loss": 1.8157, "step": 5474 }, { "epoch": 1.65, "learning_rate": 8.889236472324156e-05, "loss": 1.8305, "step": 5475 }, { "epoch": 1.65, "learning_rate": 8.886006829901527e-05, "loss": 1.8234, "step": 5476 }, { "epoch": 1.65, "learning_rate": 8.882777305130772e-05, "loss": 1.8135, "step": 5477 }, { "epoch": 1.65, "learning_rate": 8.87954789835296e-05, "loss": 1.9217, "step": 5478 }, { "epoch": 1.65, "learning_rate": 8.87631860990917e-05, "loss": 1.8368, "step": 5479 }, { "epoch": 1.65, "learning_rate": 8.873089440140439e-05, "loss": 1.8292, "step": 5480 }, { "epoch": 1.65, "learning_rate": 8.869860389387823e-05, "loss": 1.8333, "step": 5481 }, { "epoch": 1.65, "learning_rate": 8.866631457992348e-05, "loss": 1.7967, "step": 5482 }, { "epoch": 1.65, "learning_rate": 8.863402646295027e-05, "loss": 1.9438, "step": 5483 }, { "epoch": 1.65, "learning_rate": 8.86017395463687e-05, "loss": 1.7768, "step": 5484 }, { "epoch": 1.65, "learning_rate": 8.856945383358856e-05, "loss": 1.7685, "step": 5485 }, { "epoch": 1.65, "learning_rate": 8.853716932801977e-05, "loss": 1.9145, "step": 5486 }, { "epoch": 1.65, "learning_rate": 8.850488603307184e-05, "loss": 1.8422, "step": 5487 }, { "epoch": 1.65, "learning_rate": 8.847260395215442e-05, "loss": 1.804, "step": 5488 }, { "epoch": 1.65, "learning_rate": 8.844032308867688e-05, "loss": 1.8684, "step": 5489 }, { "epoch": 1.65, "learning_rate": 8.840804344604845e-05, "loss": 1.874, "step": 5490 }, { "epoch": 1.65, "learning_rate": 8.837576502767835e-05, "loss": 1.8718, "step": 5491 }, { "epoch": 1.65, "learning_rate": 8.834348783697546e-05, "loss": 1.8602, "step": 5492 }, { "epoch": 1.65, "learning_rate": 8.831121187734877e-05, "loss": 1.8924, "step": 5493 }, { "epoch": 1.65, "learning_rate": 8.827893715220703e-05, "loss": 1.7897, "step": 5494 }, { "epoch": 1.65, "learning_rate": 8.824666366495881e-05, "loss": 1.7407, "step": 5495 }, { "epoch": 1.65, "learning_rate": 8.821439141901266e-05, "loss": 1.7445, "step": 5496 }, { "epoch": 1.65, "learning_rate": 8.81821204177769e-05, "loss": 1.7465, "step": 5497 }, { "epoch": 1.66, "learning_rate": 8.814985066465978e-05, "loss": 1.8233, "step": 5498 }, { "epoch": 1.66, "learning_rate": 8.811758216306934e-05, "loss": 1.8671, "step": 5499 }, { "epoch": 1.66, "learning_rate": 8.808531491641359e-05, "loss": 1.9127, "step": 5500 }, { "epoch": 1.66, "learning_rate": 8.80530489281004e-05, "loss": 1.9219, "step": 5501 }, { "epoch": 1.66, "learning_rate": 8.802078420153741e-05, "loss": 1.8432, "step": 5502 }, { "epoch": 1.66, "learning_rate": 8.798852074013223e-05, "loss": 1.8859, "step": 5503 }, { "epoch": 1.66, "learning_rate": 8.795625854729224e-05, "loss": 1.8236, "step": 5504 }, { "epoch": 1.66, "learning_rate": 8.792399762642477e-05, "loss": 1.8582, "step": 5505 }, { "epoch": 1.66, "learning_rate": 8.789173798093701e-05, "loss": 1.7615, "step": 5506 }, { "epoch": 1.66, "learning_rate": 8.785947961423593e-05, "loss": 1.7518, "step": 5507 }, { "epoch": 1.66, "learning_rate": 8.782722252972847e-05, "loss": 1.8455, "step": 5508 }, { "epoch": 1.66, "learning_rate": 8.779496673082136e-05, "loss": 1.8281, "step": 5509 }, { "epoch": 1.66, "learning_rate": 8.776271222092128e-05, "loss": 1.7863, "step": 5510 }, { "epoch": 1.66, "learning_rate": 8.773045900343464e-05, "loss": 1.7748, "step": 5511 }, { "epoch": 1.66, "learning_rate": 8.769820708176783e-05, "loss": 1.88, "step": 5512 }, { "epoch": 1.66, "learning_rate": 8.766595645932709e-05, "loss": 1.7615, "step": 5513 }, { "epoch": 1.66, "learning_rate": 8.763370713951842e-05, "loss": 1.8028, "step": 5514 }, { "epoch": 1.66, "learning_rate": 8.760145912574784e-05, "loss": 1.8439, "step": 5515 }, { "epoch": 1.66, "learning_rate": 8.75692124214211e-05, "loss": 1.7854, "step": 5516 }, { "epoch": 1.66, "learning_rate": 8.753696702994387e-05, "loss": 1.746, "step": 5517 }, { "epoch": 1.66, "learning_rate": 8.750472295472172e-05, "loss": 1.7936, "step": 5518 }, { "epoch": 1.66, "learning_rate": 8.747248019915997e-05, "loss": 1.7967, "step": 5519 }, { "epoch": 1.66, "learning_rate": 8.74402387666639e-05, "loss": 1.8917, "step": 5520 }, { "epoch": 1.66, "learning_rate": 8.74079986606386e-05, "loss": 1.863, "step": 5521 }, { "epoch": 1.66, "learning_rate": 8.737575988448907e-05, "loss": 1.7908, "step": 5522 }, { "epoch": 1.66, "learning_rate": 8.73435224416201e-05, "loss": 1.7773, "step": 5523 }, { "epoch": 1.66, "learning_rate": 8.731128633543638e-05, "loss": 1.7477, "step": 5524 }, { "epoch": 1.66, "learning_rate": 8.727905156934247e-05, "loss": 1.8912, "step": 5525 }, { "epoch": 1.66, "learning_rate": 8.724681814674275e-05, "loss": 1.7324, "step": 5526 }, { "epoch": 1.66, "learning_rate": 8.721458607104152e-05, "loss": 1.7986, "step": 5527 }, { "epoch": 1.66, "learning_rate": 8.718235534564282e-05, "loss": 1.8558, "step": 5528 }, { "epoch": 1.66, "learning_rate": 8.71501259739507e-05, "loss": 1.8098, "step": 5529 }, { "epoch": 1.66, "learning_rate": 8.711789795936898e-05, "loss": 1.7748, "step": 5530 }, { "epoch": 1.66, "learning_rate": 8.708567130530132e-05, "loss": 1.7771, "step": 5531 }, { "epoch": 1.67, "learning_rate": 8.70534460151513e-05, "loss": 1.8427, "step": 5532 }, { "epoch": 1.67, "learning_rate": 8.702122209232228e-05, "loss": 1.8615, "step": 5533 }, { "epoch": 1.67, "learning_rate": 8.698899954021755e-05, "loss": 1.8568, "step": 5534 }, { "epoch": 1.67, "learning_rate": 8.69567783622402e-05, "loss": 1.8231, "step": 5535 }, { "epoch": 1.67, "learning_rate": 8.692455856179321e-05, "loss": 1.8501, "step": 5536 }, { "epoch": 1.67, "learning_rate": 8.689234014227942e-05, "loss": 1.844, "step": 5537 }, { "epoch": 1.67, "learning_rate": 8.686012310710146e-05, "loss": 1.8653, "step": 5538 }, { "epoch": 1.67, "learning_rate": 8.682790745966191e-05, "loss": 1.8309, "step": 5539 }, { "epoch": 1.67, "learning_rate": 8.679569320336309e-05, "loss": 1.8115, "step": 5540 }, { "epoch": 1.67, "learning_rate": 8.676348034160729e-05, "loss": 1.8188, "step": 5541 }, { "epoch": 1.67, "learning_rate": 8.67312688777966e-05, "loss": 1.829, "step": 5542 }, { "epoch": 1.67, "learning_rate": 8.669905881533291e-05, "loss": 1.7568, "step": 5543 }, { "epoch": 1.67, "learning_rate": 8.666685015761808e-05, "loss": 1.7808, "step": 5544 }, { "epoch": 1.67, "learning_rate": 8.663464290805368e-05, "loss": 1.693, "step": 5545 }, { "epoch": 1.67, "learning_rate": 8.660243707004126e-05, "loss": 1.8255, "step": 5546 }, { "epoch": 1.67, "learning_rate": 8.657023264698216e-05, "loss": 1.8162, "step": 5547 }, { "epoch": 1.67, "learning_rate": 8.653802964227757e-05, "loss": 1.8558, "step": 5548 }, { "epoch": 1.67, "learning_rate": 8.650582805932854e-05, "loss": 1.8299, "step": 5549 }, { "epoch": 1.67, "learning_rate": 8.647362790153595e-05, "loss": 1.8035, "step": 5550 }, { "epoch": 1.67, "learning_rate": 8.644142917230058e-05, "loss": 1.8352, "step": 5551 }, { "epoch": 1.67, "learning_rate": 8.640923187502299e-05, "loss": 1.8475, "step": 5552 }, { "epoch": 1.67, "learning_rate": 8.637703601310364e-05, "loss": 1.8222, "step": 5553 }, { "epoch": 1.67, "learning_rate": 8.634484158994283e-05, "loss": 1.8273, "step": 5554 }, { "epoch": 1.67, "learning_rate": 8.63126486089407e-05, "loss": 1.7937, "step": 5555 }, { "epoch": 1.67, "learning_rate": 8.628045707349724e-05, "loss": 1.9071, "step": 5556 }, { "epoch": 1.67, "learning_rate": 8.624826698701226e-05, "loss": 1.8619, "step": 5557 }, { "epoch": 1.67, "learning_rate": 8.621607835288546e-05, "loss": 1.8106, "step": 5558 }, { "epoch": 1.67, "learning_rate": 8.618389117451642e-05, "loss": 1.8018, "step": 5559 }, { "epoch": 1.67, "learning_rate": 8.615170545530442e-05, "loss": 1.774, "step": 5560 }, { "epoch": 1.67, "learning_rate": 8.611952119864877e-05, "loss": 1.8426, "step": 5561 }, { "epoch": 1.67, "learning_rate": 8.608733840794846e-05, "loss": 1.8055, "step": 5562 }, { "epoch": 1.67, "learning_rate": 8.60551570866025e-05, "loss": 1.7978, "step": 5563 }, { "epoch": 1.67, "learning_rate": 8.602297723800953e-05, "loss": 1.8818, "step": 5564 }, { "epoch": 1.68, "learning_rate": 8.599079886556824e-05, "loss": 1.7066, "step": 5565 }, { "epoch": 1.68, "learning_rate": 8.595862197267706e-05, "loss": 1.8608, "step": 5566 }, { "epoch": 1.68, "learning_rate": 8.592644656273424e-05, "loss": 1.7217, "step": 5567 }, { "epoch": 1.68, "learning_rate": 8.589427263913797e-05, "loss": 1.9122, "step": 5568 }, { "epoch": 1.68, "learning_rate": 8.586210020528617e-05, "loss": 1.8402, "step": 5569 }, { "epoch": 1.68, "learning_rate": 8.582992926457669e-05, "loss": 1.8991, "step": 5570 }, { "epoch": 1.68, "learning_rate": 8.57977598204072e-05, "loss": 1.795, "step": 5571 }, { "epoch": 1.68, "learning_rate": 8.576559187617516e-05, "loss": 1.8675, "step": 5572 }, { "epoch": 1.68, "learning_rate": 8.573342543527799e-05, "loss": 1.8738, "step": 5573 }, { "epoch": 1.68, "learning_rate": 8.57012605011128e-05, "loss": 1.7991, "step": 5574 }, { "epoch": 1.68, "learning_rate": 8.566909707707667e-05, "loss": 1.8705, "step": 5575 }, { "epoch": 1.68, "learning_rate": 8.563693516656642e-05, "loss": 1.8556, "step": 5576 }, { "epoch": 1.68, "learning_rate": 8.560477477297878e-05, "loss": 1.8543, "step": 5577 }, { "epoch": 1.68, "learning_rate": 8.557261589971033e-05, "loss": 1.8834, "step": 5578 }, { "epoch": 1.68, "learning_rate": 8.55404585501574e-05, "loss": 1.8336, "step": 5579 }, { "epoch": 1.68, "learning_rate": 8.550830272771626e-05, "loss": 1.7158, "step": 5580 }, { "epoch": 1.68, "learning_rate": 8.547614843578293e-05, "loss": 1.7719, "step": 5581 }, { "epoch": 1.68, "learning_rate": 8.544399567775333e-05, "loss": 1.9382, "step": 5582 }, { "epoch": 1.68, "learning_rate": 8.541184445702326e-05, "loss": 1.7881, "step": 5583 }, { "epoch": 1.68, "learning_rate": 8.537969477698822e-05, "loss": 1.8072, "step": 5584 }, { "epoch": 1.68, "learning_rate": 8.534754664104368e-05, "loss": 1.7893, "step": 5585 }, { "epoch": 1.68, "learning_rate": 8.531540005258483e-05, "loss": 1.7767, "step": 5586 }, { "epoch": 1.68, "learning_rate": 8.528325501500681e-05, "loss": 1.9405, "step": 5587 }, { "epoch": 1.68, "learning_rate": 8.525111153170453e-05, "loss": 1.8147, "step": 5588 }, { "epoch": 1.68, "learning_rate": 8.521896960607271e-05, "loss": 1.8066, "step": 5589 }, { "epoch": 1.68, "learning_rate": 8.518682924150609e-05, "loss": 1.8354, "step": 5590 }, { "epoch": 1.68, "learning_rate": 8.515469044139892e-05, "loss": 1.9143, "step": 5591 }, { "epoch": 1.68, "learning_rate": 8.51225532091456e-05, "loss": 1.8092, "step": 5592 }, { "epoch": 1.68, "learning_rate": 8.509041754814013e-05, "loss": 1.8136, "step": 5593 }, { "epoch": 1.68, "learning_rate": 8.505828346177648e-05, "loss": 1.9358, "step": 5594 }, { "epoch": 1.68, "learning_rate": 8.502615095344852e-05, "loss": 1.8223, "step": 5595 }, { "epoch": 1.68, "learning_rate": 8.49940200265497e-05, "loss": 1.7526, "step": 5596 }, { "epoch": 1.68, "learning_rate": 8.496189068447355e-05, "loss": 1.8337, "step": 5597 }, { "epoch": 1.69, "learning_rate": 8.492976293061328e-05, "loss": 1.7426, "step": 5598 }, { "epoch": 1.69, "learning_rate": 8.489763676836205e-05, "loss": 1.8247, "step": 5599 }, { "epoch": 1.69, "learning_rate": 8.486551220111274e-05, "loss": 1.8568, "step": 5600 }, { "epoch": 1.69, "learning_rate": 8.483338923225811e-05, "loss": 1.8324, "step": 5601 }, { "epoch": 1.69, "learning_rate": 8.480126786519085e-05, "loss": 1.9213, "step": 5602 }, { "epoch": 1.69, "learning_rate": 8.476914810330325e-05, "loss": 1.7301, "step": 5603 }, { "epoch": 1.69, "learning_rate": 8.473702994998767e-05, "loss": 1.806, "step": 5604 }, { "epoch": 1.69, "learning_rate": 8.470491340863613e-05, "loss": 1.8956, "step": 5605 }, { "epoch": 1.69, "learning_rate": 8.467279848264055e-05, "loss": 1.8591, "step": 5606 }, { "epoch": 1.69, "learning_rate": 8.464068517539277e-05, "loss": 1.7623, "step": 5607 }, { "epoch": 1.69, "learning_rate": 8.460857349028422e-05, "loss": 1.7885, "step": 5608 }, { "epoch": 1.69, "learning_rate": 8.457646343070645e-05, "loss": 1.7774, "step": 5609 }, { "epoch": 1.69, "learning_rate": 8.454435500005055e-05, "loss": 1.8972, "step": 5610 }, { "epoch": 1.69, "learning_rate": 8.451224820170767e-05, "loss": 1.8052, "step": 5611 }, { "epoch": 1.69, "learning_rate": 8.448014303906867e-05, "loss": 1.9041, "step": 5612 }, { "epoch": 1.69, "learning_rate": 8.444803951552425e-05, "loss": 1.8033, "step": 5613 }, { "epoch": 1.69, "learning_rate": 8.441593763446501e-05, "loss": 1.8942, "step": 5614 }, { "epoch": 1.69, "learning_rate": 8.438383739928122e-05, "loss": 1.8309, "step": 5615 }, { "epoch": 1.69, "learning_rate": 8.43517388133632e-05, "loss": 1.7338, "step": 5616 }, { "epoch": 1.69, "learning_rate": 8.431964188010085e-05, "loss": 1.8076, "step": 5617 }, { "epoch": 1.69, "learning_rate": 8.428754660288404e-05, "loss": 1.9187, "step": 5618 }, { "epoch": 1.69, "learning_rate": 8.425545298510252e-05, "loss": 1.9777, "step": 5619 }, { "epoch": 1.69, "learning_rate": 8.422336103014568e-05, "loss": 1.8336, "step": 5620 }, { "epoch": 1.69, "learning_rate": 8.419127074140294e-05, "loss": 1.8811, "step": 5621 }, { "epoch": 1.69, "learning_rate": 8.415918212226333e-05, "loss": 1.8917, "step": 5622 }, { "epoch": 1.69, "learning_rate": 8.412709517611596e-05, "loss": 1.8258, "step": 5623 }, { "epoch": 1.69, "learning_rate": 8.409500990634948e-05, "loss": 1.7655, "step": 5624 }, { "epoch": 1.69, "learning_rate": 8.406292631635253e-05, "loss": 1.781, "step": 5625 }, { "epoch": 1.69, "learning_rate": 8.403084440951366e-05, "loss": 1.854, "step": 5626 }, { "epoch": 1.69, "learning_rate": 8.399876418922097e-05, "loss": 1.7669, "step": 5627 }, { "epoch": 1.69, "learning_rate": 8.396668565886268e-05, "loss": 1.8248, "step": 5628 }, { "epoch": 1.69, "learning_rate": 8.393460882182657e-05, "loss": 1.7297, "step": 5629 }, { "epoch": 1.69, "learning_rate": 8.390253368150044e-05, "loss": 2.0186, "step": 5630 }, { "epoch": 1.7, "learning_rate": 8.387046024127186e-05, "loss": 1.8626, "step": 5631 }, { "epoch": 1.7, "learning_rate": 8.383838850452806e-05, "loss": 1.8104, "step": 5632 }, { "epoch": 1.7, "learning_rate": 8.38063184746564e-05, "loss": 1.7093, "step": 5633 }, { "epoch": 1.7, "learning_rate": 8.377425015504372e-05, "loss": 1.8289, "step": 5634 }, { "epoch": 1.7, "learning_rate": 8.374218354907698e-05, "loss": 1.8378, "step": 5635 }, { "epoch": 1.7, "learning_rate": 8.371011866014269e-05, "loss": 1.7149, "step": 5636 }, { "epoch": 1.7, "learning_rate": 8.367805549162741e-05, "loss": 1.8933, "step": 5637 }, { "epoch": 1.7, "learning_rate": 8.364599404691744e-05, "loss": 1.8327, "step": 5638 }, { "epoch": 1.7, "learning_rate": 8.361393432939875e-05, "loss": 1.8984, "step": 5639 }, { "epoch": 1.7, "learning_rate": 8.35818763424574e-05, "loss": 1.8724, "step": 5640 }, { "epoch": 1.7, "learning_rate": 8.354982008947899e-05, "loss": 1.7994, "step": 5641 }, { "epoch": 1.7, "learning_rate": 8.351776557384914e-05, "loss": 1.9056, "step": 5642 }, { "epoch": 1.7, "learning_rate": 8.348571279895323e-05, "loss": 1.7826, "step": 5643 }, { "epoch": 1.7, "learning_rate": 8.34536617681764e-05, "loss": 1.7994, "step": 5644 }, { "epoch": 1.7, "learning_rate": 8.342161248490372e-05, "loss": 1.9127, "step": 5645 }, { "epoch": 1.7, "learning_rate": 8.338956495251988e-05, "loss": 1.8852, "step": 5646 }, { "epoch": 1.7, "learning_rate": 8.335751917440962e-05, "loss": 1.8697, "step": 5647 }, { "epoch": 1.7, "learning_rate": 8.332547515395728e-05, "loss": 1.7853, "step": 5648 }, { "epoch": 1.7, "learning_rate": 8.329343289454721e-05, "loss": 1.8739, "step": 5649 }, { "epoch": 1.7, "learning_rate": 8.326139239956346e-05, "loss": 1.7986, "step": 5650 }, { "epoch": 1.7, "learning_rate": 8.322935367238988e-05, "loss": 1.8866, "step": 5651 }, { "epoch": 1.7, "learning_rate": 8.319731671641023e-05, "loss": 1.8257, "step": 5652 }, { "epoch": 1.7, "learning_rate": 8.316528153500789e-05, "loss": 1.8689, "step": 5653 }, { "epoch": 1.7, "learning_rate": 8.313324813156632e-05, "loss": 1.8577, "step": 5654 }, { "epoch": 1.7, "learning_rate": 8.310121650946863e-05, "loss": 1.8183, "step": 5655 }, { "epoch": 1.7, "learning_rate": 8.306918667209772e-05, "loss": 1.8765, "step": 5656 }, { "epoch": 1.7, "learning_rate": 8.303715862283638e-05, "loss": 1.7896, "step": 5657 }, { "epoch": 1.7, "learning_rate": 8.300513236506714e-05, "loss": 1.8571, "step": 5658 }, { "epoch": 1.7, "learning_rate": 8.297310790217245e-05, "loss": 1.7944, "step": 5659 }, { "epoch": 1.7, "learning_rate": 8.294108523753447e-05, "loss": 1.8654, "step": 5660 }, { "epoch": 1.7, "learning_rate": 8.290906437453516e-05, "loss": 1.8505, "step": 5661 }, { "epoch": 1.7, "learning_rate": 8.28770453165564e-05, "loss": 1.7858, "step": 5662 }, { "epoch": 1.7, "learning_rate": 8.284502806697975e-05, "loss": 1.7935, "step": 5663 }, { "epoch": 1.7, "learning_rate": 8.28130126291867e-05, "loss": 1.7984, "step": 5664 }, { "epoch": 1.71, "learning_rate": 8.278099900655843e-05, "loss": 1.8083, "step": 5665 }, { "epoch": 1.71, "learning_rate": 8.274898720247602e-05, "loss": 1.85, "step": 5666 }, { "epoch": 1.71, "learning_rate": 8.271697722032033e-05, "loss": 1.8003, "step": 5667 }, { "epoch": 1.71, "learning_rate": 8.268496906347199e-05, "loss": 1.8139, "step": 5668 }, { "epoch": 1.71, "learning_rate": 8.26529627353115e-05, "loss": 1.8205, "step": 5669 }, { "epoch": 1.71, "learning_rate": 8.26209582392191e-05, "loss": 1.8204, "step": 5670 }, { "epoch": 1.71, "learning_rate": 8.258895557857489e-05, "loss": 1.8265, "step": 5671 }, { "epoch": 1.71, "learning_rate": 8.25569547567588e-05, "loss": 1.8269, "step": 5672 }, { "epoch": 1.71, "learning_rate": 8.252495577715044e-05, "loss": 1.8668, "step": 5673 }, { "epoch": 1.71, "learning_rate": 8.24929586431294e-05, "loss": 1.6841, "step": 5674 }, { "epoch": 1.71, "learning_rate": 8.246096335807493e-05, "loss": 1.8298, "step": 5675 }, { "epoch": 1.71, "learning_rate": 8.242896992536616e-05, "loss": 1.8819, "step": 5676 }, { "epoch": 1.71, "learning_rate": 8.239697834838198e-05, "loss": 1.9328, "step": 5677 }, { "epoch": 1.71, "learning_rate": 8.236498863050113e-05, "loss": 1.8733, "step": 5678 }, { "epoch": 1.71, "learning_rate": 8.233300077510214e-05, "loss": 1.8494, "step": 5679 }, { "epoch": 1.71, "learning_rate": 8.23010147855633e-05, "loss": 1.8011, "step": 5680 }, { "epoch": 1.71, "learning_rate": 8.226903066526278e-05, "loss": 1.7734, "step": 5681 }, { "epoch": 1.71, "learning_rate": 8.223704841757846e-05, "loss": 1.7616, "step": 5682 }, { "epoch": 1.71, "learning_rate": 8.220506804588813e-05, "loss": 1.726, "step": 5683 }, { "epoch": 1.71, "learning_rate": 8.21730895535693e-05, "loss": 1.7672, "step": 5684 }, { "epoch": 1.71, "learning_rate": 8.214111294399928e-05, "loss": 1.687, "step": 5685 }, { "epoch": 1.71, "learning_rate": 8.210913822055526e-05, "loss": 1.8431, "step": 5686 }, { "epoch": 1.71, "learning_rate": 8.207716538661412e-05, "loss": 1.7317, "step": 5687 }, { "epoch": 1.71, "learning_rate": 8.204519444555265e-05, "loss": 1.88, "step": 5688 }, { "epoch": 1.71, "learning_rate": 8.201322540074734e-05, "loss": 1.7916, "step": 5689 }, { "epoch": 1.71, "learning_rate": 8.198125825557457e-05, "loss": 1.8082, "step": 5690 }, { "epoch": 1.71, "learning_rate": 8.194929301341048e-05, "loss": 1.7834, "step": 5691 }, { "epoch": 1.71, "learning_rate": 8.191732967763097e-05, "loss": 1.8636, "step": 5692 }, { "epoch": 1.71, "learning_rate": 8.18853682516118e-05, "loss": 1.7332, "step": 5693 }, { "epoch": 1.71, "learning_rate": 8.185340873872848e-05, "loss": 1.8607, "step": 5694 }, { "epoch": 1.71, "learning_rate": 8.182145114235637e-05, "loss": 1.8045, "step": 5695 }, { "epoch": 1.71, "learning_rate": 8.178949546587061e-05, "loss": 1.9247, "step": 5696 }, { "epoch": 1.71, "learning_rate": 8.175754171264608e-05, "loss": 1.9474, "step": 5697 }, { "epoch": 1.72, "learning_rate": 8.172558988605756e-05, "loss": 1.8083, "step": 5698 }, { "epoch": 1.72, "learning_rate": 8.16936399894795e-05, "loss": 1.8051, "step": 5699 }, { "epoch": 1.72, "learning_rate": 8.166169202628628e-05, "loss": 1.8468, "step": 5700 }, { "epoch": 1.72, "learning_rate": 8.162974599985198e-05, "loss": 1.8231, "step": 5701 }, { "epoch": 1.72, "learning_rate": 8.15978019135505e-05, "loss": 1.8429, "step": 5702 }, { "epoch": 1.72, "learning_rate": 8.156585977075558e-05, "loss": 1.7551, "step": 5703 }, { "epoch": 1.72, "learning_rate": 8.153391957484068e-05, "loss": 1.835, "step": 5704 }, { "epoch": 1.72, "learning_rate": 8.150198132917912e-05, "loss": 1.7737, "step": 5705 }, { "epoch": 1.72, "learning_rate": 8.147004503714395e-05, "loss": 1.8271, "step": 5706 }, { "epoch": 1.72, "learning_rate": 8.143811070210807e-05, "loss": 1.7364, "step": 5707 }, { "epoch": 1.72, "learning_rate": 8.140617832744417e-05, "loss": 1.8268, "step": 5708 }, { "epoch": 1.72, "learning_rate": 8.13742479165247e-05, "loss": 1.8439, "step": 5709 }, { "epoch": 1.72, "learning_rate": 8.134231947272191e-05, "loss": 1.7959, "step": 5710 }, { "epoch": 1.72, "learning_rate": 8.131039299940784e-05, "loss": 1.9601, "step": 5711 }, { "epoch": 1.72, "learning_rate": 8.127846849995438e-05, "loss": 1.8206, "step": 5712 }, { "epoch": 1.72, "learning_rate": 8.124654597773312e-05, "loss": 1.894, "step": 5713 }, { "epoch": 1.72, "learning_rate": 8.121462543611548e-05, "loss": 1.7673, "step": 5714 }, { "epoch": 1.72, "learning_rate": 8.118270687847272e-05, "loss": 1.7744, "step": 5715 }, { "epoch": 1.72, "learning_rate": 8.115079030817579e-05, "loss": 1.8354, "step": 5716 }, { "epoch": 1.72, "learning_rate": 8.111887572859556e-05, "loss": 1.7943, "step": 5717 }, { "epoch": 1.72, "learning_rate": 8.108696314310253e-05, "loss": 1.8667, "step": 5718 }, { "epoch": 1.72, "learning_rate": 8.105505255506714e-05, "loss": 1.86, "step": 5719 }, { "epoch": 1.72, "learning_rate": 8.102314396785953e-05, "loss": 1.9271, "step": 5720 }, { "epoch": 1.72, "learning_rate": 8.099123738484965e-05, "loss": 1.8617, "step": 5721 }, { "epoch": 1.72, "learning_rate": 8.095933280940727e-05, "loss": 1.9487, "step": 5722 }, { "epoch": 1.72, "learning_rate": 8.092743024490186e-05, "loss": 1.8284, "step": 5723 }, { "epoch": 1.72, "learning_rate": 8.08955296947028e-05, "loss": 1.7893, "step": 5724 }, { "epoch": 1.72, "learning_rate": 8.086363116217916e-05, "loss": 1.8474, "step": 5725 }, { "epoch": 1.72, "learning_rate": 8.083173465069984e-05, "loss": 1.7411, "step": 5726 }, { "epoch": 1.72, "learning_rate": 8.079984016363354e-05, "loss": 1.7731, "step": 5727 }, { "epoch": 1.72, "learning_rate": 8.076794770434867e-05, "loss": 1.8281, "step": 5728 }, { "epoch": 1.72, "learning_rate": 8.073605727621354e-05, "loss": 1.7978, "step": 5729 }, { "epoch": 1.72, "learning_rate": 8.070416888259613e-05, "loss": 1.7756, "step": 5730 }, { "epoch": 1.73, "learning_rate": 8.067228252686429e-05, "loss": 1.9696, "step": 5731 }, { "epoch": 1.73, "learning_rate": 8.064039821238565e-05, "loss": 1.763, "step": 5732 }, { "epoch": 1.73, "learning_rate": 8.060851594252755e-05, "loss": 1.8907, "step": 5733 }, { "epoch": 1.73, "learning_rate": 8.057663572065721e-05, "loss": 1.9294, "step": 5734 }, { "epoch": 1.73, "learning_rate": 8.054475755014155e-05, "loss": 1.8911, "step": 5735 }, { "epoch": 1.73, "learning_rate": 8.051288143434734e-05, "loss": 1.8368, "step": 5736 }, { "epoch": 1.73, "learning_rate": 8.048100737664106e-05, "loss": 1.8146, "step": 5737 }, { "epoch": 1.73, "learning_rate": 8.044913538038906e-05, "loss": 1.8437, "step": 5738 }, { "epoch": 1.73, "learning_rate": 8.041726544895742e-05, "loss": 1.8005, "step": 5739 }, { "epoch": 1.73, "learning_rate": 8.038539758571199e-05, "loss": 1.8343, "step": 5740 }, { "epoch": 1.73, "learning_rate": 8.035353179401848e-05, "loss": 1.8831, "step": 5741 }, { "epoch": 1.73, "learning_rate": 8.032166807724224e-05, "loss": 1.8225, "step": 5742 }, { "epoch": 1.73, "learning_rate": 8.028980643874853e-05, "loss": 1.7807, "step": 5743 }, { "epoch": 1.73, "learning_rate": 8.025794688190235e-05, "loss": 1.8394, "step": 5744 }, { "epoch": 1.73, "learning_rate": 8.022608941006847e-05, "loss": 1.8601, "step": 5745 }, { "epoch": 1.73, "learning_rate": 8.019423402661144e-05, "loss": 1.8742, "step": 5746 }, { "epoch": 1.73, "learning_rate": 8.016238073489559e-05, "loss": 1.7894, "step": 5747 }, { "epoch": 1.73, "learning_rate": 8.013052953828505e-05, "loss": 1.8671, "step": 5748 }, { "epoch": 1.73, "learning_rate": 8.009868044014367e-05, "loss": 1.8519, "step": 5749 }, { "epoch": 1.73, "learning_rate": 8.006683344383518e-05, "loss": 1.7625, "step": 5750 }, { "epoch": 1.73, "learning_rate": 8.0034988552723e-05, "loss": 1.8669, "step": 5751 }, { "epoch": 1.73, "learning_rate": 8.000314577017033e-05, "loss": 1.7669, "step": 5752 }, { "epoch": 1.73, "learning_rate": 7.997130509954023e-05, "loss": 1.8615, "step": 5753 }, { "epoch": 1.73, "learning_rate": 7.993946654419544e-05, "loss": 2.0145, "step": 5754 }, { "epoch": 1.73, "learning_rate": 7.990763010749849e-05, "loss": 2.3341, "step": 5755 }, { "epoch": 1.73, "learning_rate": 7.987579579281182e-05, "loss": 2.2531, "step": 5756 }, { "epoch": 1.73, "learning_rate": 7.984396360349742e-05, "loss": 1.8755, "step": 5757 }, { "epoch": 1.73, "learning_rate": 7.981213354291724e-05, "loss": 1.8455, "step": 5758 }, { "epoch": 1.73, "learning_rate": 7.97803056144329e-05, "loss": 1.8815, "step": 5759 }, { "epoch": 1.73, "learning_rate": 7.97484798214059e-05, "loss": 1.8315, "step": 5760 }, { "epoch": 1.73, "learning_rate": 7.971665616719737e-05, "loss": 1.8837, "step": 5761 }, { "epoch": 1.73, "learning_rate": 7.968483465516832e-05, "loss": 1.9016, "step": 5762 }, { "epoch": 1.73, "learning_rate": 7.965301528867956e-05, "loss": 1.8515, "step": 5763 }, { "epoch": 1.74, "learning_rate": 7.962119807109157e-05, "loss": 1.8836, "step": 5764 }, { "epoch": 1.74, "learning_rate": 7.958938300576466e-05, "loss": 1.8292, "step": 5765 }, { "epoch": 1.74, "learning_rate": 7.955757009605889e-05, "loss": 1.9115, "step": 5766 }, { "epoch": 1.74, "learning_rate": 7.952575934533411e-05, "loss": 1.9656, "step": 5767 }, { "epoch": 1.74, "learning_rate": 7.949395075695001e-05, "loss": 1.8676, "step": 5768 }, { "epoch": 1.74, "learning_rate": 7.946214433426586e-05, "loss": 1.9111, "step": 5769 }, { "epoch": 1.74, "learning_rate": 7.943034008064098e-05, "loss": 1.8696, "step": 5770 }, { "epoch": 1.74, "learning_rate": 7.939853799943416e-05, "loss": 1.9209, "step": 5771 }, { "epoch": 1.74, "learning_rate": 7.93667380940042e-05, "loss": 1.8689, "step": 5772 }, { "epoch": 1.74, "learning_rate": 7.933494036770949e-05, "loss": 1.8983, "step": 5773 }, { "epoch": 1.74, "learning_rate": 7.93031448239083e-05, "loss": 1.9516, "step": 5774 }, { "epoch": 1.74, "learning_rate": 7.927135146595875e-05, "loss": 1.8923, "step": 5775 }, { "epoch": 1.74, "learning_rate": 7.923956029721846e-05, "loss": 1.8301, "step": 5776 }, { "epoch": 1.74, "learning_rate": 7.920777132104514e-05, "loss": 1.8168, "step": 5777 }, { "epoch": 1.74, "learning_rate": 7.917598454079598e-05, "loss": 1.8593, "step": 5778 }, { "epoch": 1.74, "learning_rate": 7.914419995982809e-05, "loss": 1.8334, "step": 5779 }, { "epoch": 1.74, "learning_rate": 7.911241758149844e-05, "loss": 1.8393, "step": 5780 }, { "epoch": 1.74, "learning_rate": 7.90806374091635e-05, "loss": 1.8777, "step": 5781 }, { "epoch": 1.74, "learning_rate": 7.90488594461798e-05, "loss": 1.8441, "step": 5782 }, { "epoch": 1.74, "learning_rate": 7.901708369590338e-05, "loss": 1.8898, "step": 5783 }, { "epoch": 1.74, "learning_rate": 7.898531016169021e-05, "loss": 1.8305, "step": 5784 }, { "epoch": 1.74, "learning_rate": 7.895353884689605e-05, "loss": 1.7675, "step": 5785 }, { "epoch": 1.74, "learning_rate": 7.892176975487621e-05, "loss": 1.8162, "step": 5786 }, { "epoch": 1.74, "learning_rate": 7.889000288898607e-05, "loss": 1.9165, "step": 5787 }, { "epoch": 1.74, "learning_rate": 7.885823825258048e-05, "loss": 1.7581, "step": 5788 }, { "epoch": 1.74, "learning_rate": 7.882647584901429e-05, "loss": 1.8618, "step": 5789 }, { "epoch": 1.74, "learning_rate": 7.879471568164191e-05, "loss": 1.7961, "step": 5790 }, { "epoch": 1.74, "learning_rate": 7.876295775381772e-05, "loss": 1.8802, "step": 5791 }, { "epoch": 1.74, "learning_rate": 7.873120206889574e-05, "loss": 1.8468, "step": 5792 }, { "epoch": 1.74, "learning_rate": 7.869944863022973e-05, "loss": 1.9219, "step": 5793 }, { "epoch": 1.74, "learning_rate": 7.866769744117332e-05, "loss": 1.7713, "step": 5794 }, { "epoch": 1.74, "learning_rate": 7.863594850507973e-05, "loss": 1.8362, "step": 5795 }, { "epoch": 1.74, "learning_rate": 7.860420182530217e-05, "loss": 1.8963, "step": 5796 }, { "epoch": 1.75, "learning_rate": 7.85724574051935e-05, "loss": 1.7866, "step": 5797 }, { "epoch": 1.75, "learning_rate": 7.85407152481062e-05, "loss": 1.9303, "step": 5798 }, { "epoch": 1.75, "learning_rate": 7.85089753573928e-05, "loss": 1.7642, "step": 5799 }, { "epoch": 1.75, "learning_rate": 7.847723773640531e-05, "loss": 1.7626, "step": 5800 }, { "epoch": 1.75, "learning_rate": 7.844550238849574e-05, "loss": 1.7478, "step": 5801 }, { "epoch": 1.75, "learning_rate": 7.841376931701563e-05, "loss": 1.7379, "step": 5802 }, { "epoch": 1.75, "learning_rate": 7.838203852531649e-05, "loss": 1.8339, "step": 5803 }, { "epoch": 1.75, "learning_rate": 7.83503100167495e-05, "loss": 1.8735, "step": 5804 }, { "epoch": 1.75, "learning_rate": 7.83185837946655e-05, "loss": 1.8382, "step": 5805 }, { "epoch": 1.75, "learning_rate": 7.828685986241531e-05, "loss": 1.7599, "step": 5806 }, { "epoch": 1.75, "learning_rate": 7.825513822334925e-05, "loss": 1.7819, "step": 5807 }, { "epoch": 1.75, "learning_rate": 7.82234188808176e-05, "loss": 1.7437, "step": 5808 }, { "epoch": 1.75, "learning_rate": 7.819170183817036e-05, "loss": 1.8932, "step": 5809 }, { "epoch": 1.75, "learning_rate": 7.815998709875718e-05, "loss": 1.7424, "step": 5810 }, { "epoch": 1.75, "learning_rate": 7.812827466592762e-05, "loss": 1.8856, "step": 5811 }, { "epoch": 1.75, "learning_rate": 7.80965645430308e-05, "loss": 1.8498, "step": 5812 }, { "epoch": 1.75, "learning_rate": 7.806485673341586e-05, "loss": 1.7077, "step": 5813 }, { "epoch": 1.75, "learning_rate": 7.80331512404314e-05, "loss": 1.8466, "step": 5814 }, { "epoch": 1.75, "learning_rate": 7.800144806742603e-05, "loss": 1.7587, "step": 5815 }, { "epoch": 1.75, "learning_rate": 7.796974721774799e-05, "loss": 2.0167, "step": 5816 }, { "epoch": 1.75, "learning_rate": 7.793804869474524e-05, "loss": 1.6977, "step": 5817 }, { "epoch": 1.75, "learning_rate": 7.790635250176566e-05, "loss": 1.9016, "step": 5818 }, { "epoch": 1.75, "learning_rate": 7.787465864215662e-05, "loss": 1.8369, "step": 5819 }, { "epoch": 1.75, "learning_rate": 7.784296711926548e-05, "loss": 1.8922, "step": 5820 }, { "epoch": 1.75, "learning_rate": 7.78112779364393e-05, "loss": 1.8401, "step": 5821 }, { "epoch": 1.75, "learning_rate": 7.77795910970248e-05, "loss": 1.7175, "step": 5822 }, { "epoch": 1.75, "learning_rate": 7.774790660436858e-05, "loss": 1.9162, "step": 5823 }, { "epoch": 1.75, "learning_rate": 7.771622446181684e-05, "loss": 1.7683, "step": 5824 }, { "epoch": 1.75, "learning_rate": 7.76845446727157e-05, "loss": 1.8795, "step": 5825 }, { "epoch": 1.75, "learning_rate": 7.765286724041085e-05, "loss": 1.8285, "step": 5826 }, { "epoch": 1.75, "learning_rate": 7.762119216824792e-05, "loss": 1.8453, "step": 5827 }, { "epoch": 1.75, "learning_rate": 7.758951945957218e-05, "loss": 1.8301, "step": 5828 }, { "epoch": 1.75, "learning_rate": 7.755784911772865e-05, "loss": 1.8151, "step": 5829 }, { "epoch": 1.75, "learning_rate": 7.752618114606216e-05, "loss": 1.7889, "step": 5830 }, { "epoch": 1.76, "learning_rate": 7.749451554791721e-05, "loss": 1.928, "step": 5831 }, { "epoch": 1.76, "learning_rate": 7.74628523266381e-05, "loss": 1.8356, "step": 5832 }, { "epoch": 1.76, "learning_rate": 7.74311914855689e-05, "loss": 1.9519, "step": 5833 }, { "epoch": 1.76, "learning_rate": 7.739953302805335e-05, "loss": 1.831, "step": 5834 }, { "epoch": 1.76, "learning_rate": 7.736787695743503e-05, "loss": 1.8028, "step": 5835 }, { "epoch": 1.76, "learning_rate": 7.73362232770572e-05, "loss": 1.8035, "step": 5836 }, { "epoch": 1.76, "learning_rate": 7.73045719902629e-05, "loss": 1.8482, "step": 5837 }, { "epoch": 1.76, "learning_rate": 7.727292310039488e-05, "loss": 1.8571, "step": 5838 }, { "epoch": 1.76, "learning_rate": 7.724127661079573e-05, "loss": 1.8632, "step": 5839 }, { "epoch": 1.76, "learning_rate": 7.720963252480766e-05, "loss": 1.7443, "step": 5840 }, { "epoch": 1.76, "learning_rate": 7.717799084577272e-05, "loss": 1.8712, "step": 5841 }, { "epoch": 1.76, "learning_rate": 7.714635157703269e-05, "loss": 1.8354, "step": 5842 }, { "epoch": 1.76, "learning_rate": 7.711471472192901e-05, "loss": 1.7614, "step": 5843 }, { "epoch": 1.76, "learning_rate": 7.708308028380301e-05, "loss": 1.8387, "step": 5844 }, { "epoch": 1.76, "learning_rate": 7.705144826599567e-05, "loss": 1.8187, "step": 5845 }, { "epoch": 1.76, "learning_rate": 7.701981867184771e-05, "loss": 1.7987, "step": 5846 }, { "epoch": 1.76, "learning_rate": 7.698819150469965e-05, "loss": 1.9265, "step": 5847 }, { "epoch": 1.76, "learning_rate": 7.695656676789169e-05, "loss": 1.8989, "step": 5848 }, { "epoch": 1.76, "learning_rate": 7.692494446476384e-05, "loss": 1.8079, "step": 5849 }, { "epoch": 1.76, "learning_rate": 7.689332459865579e-05, "loss": 1.8377, "step": 5850 }, { "epoch": 1.76, "learning_rate": 7.686170717290699e-05, "loss": 1.926, "step": 5851 }, { "epoch": 1.76, "learning_rate": 7.683009219085668e-05, "loss": 1.9873, "step": 5852 }, { "epoch": 1.76, "learning_rate": 7.679847965584379e-05, "loss": 1.8685, "step": 5853 }, { "epoch": 1.76, "learning_rate": 7.676686957120701e-05, "loss": 1.869, "step": 5854 }, { "epoch": 1.76, "learning_rate": 7.673526194028475e-05, "loss": 1.7667, "step": 5855 }, { "epoch": 1.76, "learning_rate": 7.670365676641518e-05, "loss": 1.9407, "step": 5856 }, { "epoch": 1.76, "learning_rate": 7.667205405293624e-05, "loss": 1.744, "step": 5857 }, { "epoch": 1.76, "learning_rate": 7.664045380318554e-05, "loss": 1.7527, "step": 5858 }, { "epoch": 1.76, "learning_rate": 7.66088560205005e-05, "loss": 1.9015, "step": 5859 }, { "epoch": 1.76, "learning_rate": 7.657726070821822e-05, "loss": 1.8044, "step": 5860 }, { "epoch": 1.76, "learning_rate": 7.654566786967561e-05, "loss": 1.8702, "step": 5861 }, { "epoch": 1.76, "learning_rate": 7.651407750820922e-05, "loss": 1.8804, "step": 5862 }, { "epoch": 1.76, "learning_rate": 7.648248962715543e-05, "loss": 1.8024, "step": 5863 }, { "epoch": 1.77, "learning_rate": 7.645090422985034e-05, "loss": 1.9288, "step": 5864 }, { "epoch": 1.77, "learning_rate": 7.641932131962972e-05, "loss": 1.8197, "step": 5865 }, { "epoch": 1.77, "learning_rate": 7.638774089982917e-05, "loss": 1.8348, "step": 5866 }, { "epoch": 1.77, "learning_rate": 7.635616297378396e-05, "loss": 1.9097, "step": 5867 }, { "epoch": 1.77, "learning_rate": 7.632458754482912e-05, "loss": 1.8998, "step": 5868 }, { "epoch": 1.77, "learning_rate": 7.629301461629945e-05, "loss": 1.9817, "step": 5869 }, { "epoch": 1.77, "learning_rate": 7.62614441915294e-05, "loss": 1.8736, "step": 5870 }, { "epoch": 1.77, "learning_rate": 7.622987627385329e-05, "loss": 1.8684, "step": 5871 }, { "epoch": 1.77, "learning_rate": 7.6198310866605e-05, "loss": 1.7168, "step": 5872 }, { "epoch": 1.77, "learning_rate": 7.616674797311831e-05, "loss": 1.8911, "step": 5873 }, { "epoch": 1.77, "learning_rate": 7.613518759672661e-05, "loss": 1.8733, "step": 5874 }, { "epoch": 1.77, "learning_rate": 7.610362974076312e-05, "loss": 1.7941, "step": 5875 }, { "epoch": 1.77, "learning_rate": 7.607207440856075e-05, "loss": 1.8791, "step": 5876 }, { "epoch": 1.77, "learning_rate": 7.604052160345209e-05, "loss": 1.839, "step": 5877 }, { "epoch": 1.77, "learning_rate": 7.600897132876961e-05, "loss": 1.8178, "step": 5878 }, { "epoch": 1.77, "learning_rate": 7.597742358784534e-05, "loss": 1.8676, "step": 5879 }, { "epoch": 1.77, "learning_rate": 7.594587838401115e-05, "loss": 1.8443, "step": 5880 }, { "epoch": 1.77, "learning_rate": 7.591433572059862e-05, "loss": 1.7551, "step": 5881 }, { "epoch": 1.77, "learning_rate": 7.588279560093907e-05, "loss": 1.7958, "step": 5882 }, { "epoch": 1.77, "learning_rate": 7.585125802836353e-05, "loss": 1.8504, "step": 5883 }, { "epoch": 1.77, "learning_rate": 7.581972300620275e-05, "loss": 1.8495, "step": 5884 }, { "epoch": 1.77, "learning_rate": 7.578819053778725e-05, "loss": 1.8295, "step": 5885 }, { "epoch": 1.77, "learning_rate": 7.575666062644725e-05, "loss": 1.8261, "step": 5886 }, { "epoch": 1.77, "learning_rate": 7.572513327551271e-05, "loss": 1.8154, "step": 5887 }, { "epoch": 1.77, "learning_rate": 7.569360848831333e-05, "loss": 1.6855, "step": 5888 }, { "epoch": 1.77, "learning_rate": 7.566208626817852e-05, "loss": 1.8491, "step": 5889 }, { "epoch": 1.77, "learning_rate": 7.563056661843746e-05, "loss": 1.8238, "step": 5890 }, { "epoch": 1.77, "learning_rate": 7.559904954241896e-05, "loss": 1.785, "step": 5891 }, { "epoch": 1.77, "learning_rate": 7.556753504345166e-05, "loss": 1.8505, "step": 5892 }, { "epoch": 1.77, "learning_rate": 7.553602312486392e-05, "loss": 1.7634, "step": 5893 }, { "epoch": 1.77, "learning_rate": 7.550451378998377e-05, "loss": 1.7524, "step": 5894 }, { "epoch": 1.77, "learning_rate": 7.547300704213902e-05, "loss": 1.888, "step": 5895 }, { "epoch": 1.77, "learning_rate": 7.544150288465713e-05, "loss": 1.7919, "step": 5896 }, { "epoch": 1.78, "learning_rate": 7.541000132086538e-05, "loss": 1.9303, "step": 5897 }, { "epoch": 1.78, "learning_rate": 7.537850235409076e-05, "loss": 1.8651, "step": 5898 }, { "epoch": 1.78, "learning_rate": 7.534700598765992e-05, "loss": 1.8947, "step": 5899 }, { "epoch": 1.78, "learning_rate": 7.53155122248993e-05, "loss": 1.9497, "step": 5900 }, { "epoch": 1.78, "learning_rate": 7.528402106913504e-05, "loss": 1.816, "step": 5901 }, { "epoch": 1.78, "learning_rate": 7.525253252369299e-05, "loss": 1.8918, "step": 5902 }, { "epoch": 1.78, "learning_rate": 7.522104659189876e-05, "loss": 1.8149, "step": 5903 }, { "epoch": 1.78, "learning_rate": 7.518956327707765e-05, "loss": 1.8553, "step": 5904 }, { "epoch": 1.78, "learning_rate": 7.515808258255473e-05, "loss": 1.78, "step": 5905 }, { "epoch": 1.78, "learning_rate": 7.512660451165472e-05, "loss": 1.7462, "step": 5906 }, { "epoch": 1.78, "learning_rate": 7.509512906770214e-05, "loss": 1.9212, "step": 5907 }, { "epoch": 1.78, "learning_rate": 7.506365625402117e-05, "loss": 1.8007, "step": 5908 }, { "epoch": 1.78, "learning_rate": 7.503218607393574e-05, "loss": 1.8749, "step": 5909 }, { "epoch": 1.78, "learning_rate": 7.500071853076957e-05, "loss": 1.9817, "step": 5910 }, { "epoch": 1.78, "learning_rate": 7.496925362784593e-05, "loss": 1.7918, "step": 5911 }, { "epoch": 1.78, "learning_rate": 7.4937791368488e-05, "loss": 1.8322, "step": 5912 }, { "epoch": 1.78, "learning_rate": 7.490633175601853e-05, "loss": 1.9623, "step": 5913 }, { "epoch": 1.78, "learning_rate": 7.487487479376012e-05, "loss": 1.8253, "step": 5914 }, { "epoch": 1.78, "learning_rate": 7.484342048503496e-05, "loss": 1.7838, "step": 5915 }, { "epoch": 1.78, "learning_rate": 7.481196883316503e-05, "loss": 1.8333, "step": 5916 }, { "epoch": 1.78, "learning_rate": 7.478051984147211e-05, "loss": 1.789, "step": 5917 }, { "epoch": 1.78, "learning_rate": 7.474907351327754e-05, "loss": 1.8029, "step": 5918 }, { "epoch": 1.78, "learning_rate": 7.471762985190247e-05, "loss": 1.6683, "step": 5919 }, { "epoch": 1.78, "learning_rate": 7.468618886066774e-05, "loss": 1.8891, "step": 5920 }, { "epoch": 1.78, "learning_rate": 7.46547505428939e-05, "loss": 1.7919, "step": 5921 }, { "epoch": 1.78, "learning_rate": 7.462331490190133e-05, "loss": 1.9927, "step": 5922 }, { "epoch": 1.78, "learning_rate": 7.459188194100992e-05, "loss": 1.7424, "step": 5923 }, { "epoch": 1.78, "learning_rate": 7.456045166353949e-05, "loss": 1.9302, "step": 5924 }, { "epoch": 1.78, "learning_rate": 7.45290240728094e-05, "loss": 1.8018, "step": 5925 }, { "epoch": 1.78, "learning_rate": 7.449759917213884e-05, "loss": 1.909, "step": 5926 }, { "epoch": 1.78, "learning_rate": 7.446617696484665e-05, "loss": 1.8181, "step": 5927 }, { "epoch": 1.78, "learning_rate": 7.443475745425143e-05, "loss": 1.8157, "step": 5928 }, { "epoch": 1.78, "learning_rate": 7.440334064367155e-05, "loss": 1.7422, "step": 5929 }, { "epoch": 1.79, "learning_rate": 7.437192653642493e-05, "loss": 1.8008, "step": 5930 }, { "epoch": 1.79, "learning_rate": 7.434051513582933e-05, "loss": 1.7903, "step": 5931 }, { "epoch": 1.79, "learning_rate": 7.430910644520219e-05, "loss": 1.7793, "step": 5932 }, { "epoch": 1.79, "learning_rate": 7.427770046786067e-05, "loss": 1.8335, "step": 5933 }, { "epoch": 1.79, "learning_rate": 7.42462972071217e-05, "loss": 1.7958, "step": 5934 }, { "epoch": 1.79, "learning_rate": 7.421489666630173e-05, "loss": 1.7677, "step": 5935 }, { "epoch": 1.79, "learning_rate": 7.418349884871722e-05, "loss": 1.7053, "step": 5936 }, { "epoch": 1.79, "learning_rate": 7.415210375768406e-05, "loss": 1.7579, "step": 5937 }, { "epoch": 1.79, "learning_rate": 7.412071139651803e-05, "loss": 1.9611, "step": 5938 }, { "epoch": 1.79, "learning_rate": 7.408932176853452e-05, "loss": 1.9955, "step": 5939 }, { "epoch": 1.79, "learning_rate": 7.405793487704866e-05, "loss": 1.8157, "step": 5940 }, { "epoch": 1.79, "learning_rate": 7.402655072537543e-05, "loss": 1.8463, "step": 5941 }, { "epoch": 1.79, "learning_rate": 7.399516931682923e-05, "loss": 1.7401, "step": 5942 }, { "epoch": 1.79, "learning_rate": 7.396379065472448e-05, "loss": 1.82, "step": 5943 }, { "epoch": 1.79, "learning_rate": 7.393241474237507e-05, "loss": 1.8705, "step": 5944 }, { "epoch": 1.79, "learning_rate": 7.39010415830947e-05, "loss": 1.9121, "step": 5945 }, { "epoch": 1.79, "learning_rate": 7.386967118019688e-05, "loss": 1.8932, "step": 5946 }, { "epoch": 1.79, "learning_rate": 7.383830353699456e-05, "loss": 1.8586, "step": 5947 }, { "epoch": 1.79, "learning_rate": 7.380693865680074e-05, "loss": 1.8164, "step": 5948 }, { "epoch": 1.79, "learning_rate": 7.377557654292778e-05, "loss": 1.8444, "step": 5949 }, { "epoch": 1.79, "learning_rate": 7.374421719868808e-05, "loss": 1.7958, "step": 5950 }, { "epoch": 1.79, "learning_rate": 7.371286062739345e-05, "loss": 1.8349, "step": 5951 }, { "epoch": 1.79, "learning_rate": 7.368150683235557e-05, "loss": 1.8022, "step": 5952 }, { "epoch": 1.79, "learning_rate": 7.365015581688593e-05, "loss": 1.8101, "step": 5953 }, { "epoch": 1.79, "learning_rate": 7.36188075842954e-05, "loss": 1.9415, "step": 5954 }, { "epoch": 1.79, "learning_rate": 7.358746213789491e-05, "loss": 1.9647, "step": 5955 }, { "epoch": 1.79, "learning_rate": 7.355611948099484e-05, "loss": 1.9821, "step": 5956 }, { "epoch": 1.79, "learning_rate": 7.35247796169054e-05, "loss": 2.0188, "step": 5957 }, { "epoch": 1.79, "learning_rate": 7.349344254893657e-05, "loss": 1.7781, "step": 5958 }, { "epoch": 1.79, "learning_rate": 7.346210828039777e-05, "loss": 1.716, "step": 5959 }, { "epoch": 1.79, "learning_rate": 7.343077681459847e-05, "loss": 1.7999, "step": 5960 }, { "epoch": 1.79, "learning_rate": 7.339944815484752e-05, "loss": 1.7929, "step": 5961 }, { "epoch": 1.79, "learning_rate": 7.336812230445378e-05, "loss": 1.8439, "step": 5962 }, { "epoch": 1.8, "learning_rate": 7.33367992667255e-05, "loss": 1.8174, "step": 5963 }, { "epoch": 1.8, "learning_rate": 7.330547904497089e-05, "loss": 1.8623, "step": 5964 }, { "epoch": 1.8, "learning_rate": 7.327416164249781e-05, "loss": 1.8072, "step": 5965 }, { "epoch": 1.8, "learning_rate": 7.324284706261362e-05, "loss": 1.8103, "step": 5966 }, { "epoch": 1.8, "learning_rate": 7.32115353086257e-05, "loss": 1.7992, "step": 5967 }, { "epoch": 1.8, "learning_rate": 7.318022638384085e-05, "loss": 1.7291, "step": 5968 }, { "epoch": 1.8, "learning_rate": 7.314892029156574e-05, "loss": 1.7542, "step": 5969 }, { "epoch": 1.8, "learning_rate": 7.311761703510674e-05, "loss": 1.7701, "step": 5970 }, { "epoch": 1.8, "learning_rate": 7.308631661776977e-05, "loss": 1.8236, "step": 5971 }, { "epoch": 1.8, "learning_rate": 7.305501904286067e-05, "loss": 1.8775, "step": 5972 }, { "epoch": 1.8, "learning_rate": 7.302372431368475e-05, "loss": 1.8544, "step": 5973 }, { "epoch": 1.8, "learning_rate": 7.299243243354723e-05, "loss": 1.7768, "step": 5974 }, { "epoch": 1.8, "learning_rate": 7.296114340575283e-05, "loss": 1.7631, "step": 5975 }, { "epoch": 1.8, "learning_rate": 7.292985723360614e-05, "loss": 1.8947, "step": 5976 }, { "epoch": 1.8, "learning_rate": 7.289857392041138e-05, "loss": 1.8194, "step": 5977 }, { "epoch": 1.8, "learning_rate": 7.286729346947243e-05, "loss": 1.7858, "step": 5978 }, { "epoch": 1.8, "learning_rate": 7.283601588409297e-05, "loss": 1.9762, "step": 5979 }, { "epoch": 1.8, "learning_rate": 7.280474116757621e-05, "loss": 1.852, "step": 5980 }, { "epoch": 1.8, "learning_rate": 7.277346932322522e-05, "loss": 1.7905, "step": 5981 }, { "epoch": 1.8, "learning_rate": 7.274220035434271e-05, "loss": 1.849, "step": 5982 }, { "epoch": 1.8, "learning_rate": 7.271093426423107e-05, "loss": 1.8084, "step": 5983 }, { "epoch": 1.8, "learning_rate": 7.26796710561924e-05, "loss": 1.7503, "step": 5984 }, { "epoch": 1.8, "learning_rate": 7.264841073352849e-05, "loss": 1.7383, "step": 5985 }, { "epoch": 1.8, "learning_rate": 7.261715329954085e-05, "loss": 1.8723, "step": 5986 }, { "epoch": 1.8, "learning_rate": 7.25858987575306e-05, "loss": 1.892, "step": 5987 }, { "epoch": 1.8, "learning_rate": 7.255464711079865e-05, "loss": 1.875, "step": 5988 }, { "epoch": 1.8, "learning_rate": 7.252339836264563e-05, "loss": 1.7239, "step": 5989 }, { "epoch": 1.8, "learning_rate": 7.249215251637172e-05, "loss": 1.8351, "step": 5990 }, { "epoch": 1.8, "learning_rate": 7.246090957527693e-05, "loss": 1.8356, "step": 5991 }, { "epoch": 1.8, "learning_rate": 7.242966954266089e-05, "loss": 1.8459, "step": 5992 }, { "epoch": 1.8, "learning_rate": 7.239843242182297e-05, "loss": 1.8969, "step": 5993 }, { "epoch": 1.8, "learning_rate": 7.23671982160622e-05, "loss": 1.8801, "step": 5994 }, { "epoch": 1.8, "learning_rate": 7.233596692867725e-05, "loss": 1.688, "step": 5995 }, { "epoch": 1.8, "learning_rate": 7.230473856296666e-05, "loss": 1.9242, "step": 5996 }, { "epoch": 1.81, "learning_rate": 7.227351312222843e-05, "loss": 1.9096, "step": 5997 }, { "epoch": 1.81, "learning_rate": 7.224229060976045e-05, "loss": 1.8287, "step": 5998 }, { "epoch": 1.81, "learning_rate": 7.221107102886015e-05, "loss": 1.901, "step": 5999 }, { "epoch": 1.81, "learning_rate": 7.217985438282473e-05, "loss": 1.8078, "step": 6000 }, { "epoch": 1.81, "learning_rate": 7.214864067495112e-05, "loss": 1.8897, "step": 6001 }, { "epoch": 1.81, "learning_rate": 7.21174299085358e-05, "loss": 1.8364, "step": 6002 }, { "epoch": 1.81, "learning_rate": 7.20862220868751e-05, "loss": 1.9376, "step": 6003 }, { "epoch": 1.81, "learning_rate": 7.205501721326491e-05, "loss": 1.8285, "step": 6004 }, { "epoch": 1.81, "learning_rate": 7.202381529100089e-05, "loss": 1.8595, "step": 6005 }, { "epoch": 1.81, "learning_rate": 7.199261632337836e-05, "loss": 1.8649, "step": 6006 }, { "epoch": 1.81, "learning_rate": 7.196142031369231e-05, "loss": 1.7712, "step": 6007 }, { "epoch": 1.81, "learning_rate": 7.193022726523746e-05, "loss": 1.7776, "step": 6008 }, { "epoch": 1.81, "learning_rate": 7.189903718130817e-05, "loss": 1.7504, "step": 6009 }, { "epoch": 1.81, "learning_rate": 7.186785006519851e-05, "loss": 1.9166, "step": 6010 }, { "epoch": 1.81, "learning_rate": 7.183666592020229e-05, "loss": 1.8571, "step": 6011 }, { "epoch": 1.81, "learning_rate": 7.180548474961287e-05, "loss": 1.8626, "step": 6012 }, { "epoch": 1.81, "learning_rate": 7.177430655672344e-05, "loss": 1.9551, "step": 6013 }, { "epoch": 1.81, "learning_rate": 7.174313134482679e-05, "loss": 1.779, "step": 6014 }, { "epoch": 1.81, "learning_rate": 7.171195911721542e-05, "loss": 1.8666, "step": 6015 }, { "epoch": 1.81, "learning_rate": 7.168078987718152e-05, "loss": 1.7952, "step": 6016 }, { "epoch": 1.81, "learning_rate": 7.164962362801695e-05, "loss": 1.7247, "step": 6017 }, { "epoch": 1.81, "learning_rate": 7.161846037301329e-05, "loss": 1.7807, "step": 6018 }, { "epoch": 1.81, "learning_rate": 7.158730011546173e-05, "loss": 1.8559, "step": 6019 }, { "epoch": 1.81, "learning_rate": 7.155614285865324e-05, "loss": 1.8329, "step": 6020 }, { "epoch": 1.81, "learning_rate": 7.152498860587838e-05, "loss": 1.9271, "step": 6021 }, { "epoch": 1.81, "learning_rate": 7.149383736042744e-05, "loss": 1.7887, "step": 6022 }, { "epoch": 1.81, "learning_rate": 7.146268912559047e-05, "loss": 2.0069, "step": 6023 }, { "epoch": 1.81, "learning_rate": 7.143154390465699e-05, "loss": 1.7689, "step": 6024 }, { "epoch": 1.81, "learning_rate": 7.140040170091644e-05, "loss": 1.8575, "step": 6025 }, { "epoch": 1.81, "learning_rate": 7.136926251765776e-05, "loss": 1.6891, "step": 6026 }, { "epoch": 1.81, "learning_rate": 7.13381263581697e-05, "loss": 1.7703, "step": 6027 }, { "epoch": 1.81, "learning_rate": 7.130699322574058e-05, "loss": 1.8116, "step": 6028 }, { "epoch": 1.81, "learning_rate": 7.127586312365848e-05, "loss": 1.8284, "step": 6029 }, { "epoch": 1.82, "learning_rate": 7.124473605521118e-05, "loss": 1.845, "step": 6030 }, { "epoch": 1.82, "learning_rate": 7.121361202368603e-05, "loss": 1.8814, "step": 6031 }, { "epoch": 1.82, "learning_rate": 7.118249103237016e-05, "loss": 1.7492, "step": 6032 }, { "epoch": 1.82, "learning_rate": 7.115137308455031e-05, "loss": 1.8748, "step": 6033 }, { "epoch": 1.82, "learning_rate": 7.112025818351296e-05, "loss": 1.768, "step": 6034 }, { "epoch": 1.82, "learning_rate": 7.108914633254426e-05, "loss": 1.8296, "step": 6035 }, { "epoch": 1.82, "learning_rate": 7.105803753492997e-05, "loss": 1.7314, "step": 6036 }, { "epoch": 1.82, "learning_rate": 7.102693179395561e-05, "loss": 1.8364, "step": 6037 }, { "epoch": 1.82, "learning_rate": 7.099582911290632e-05, "loss": 1.8712, "step": 6038 }, { "epoch": 1.82, "learning_rate": 7.096472949506697e-05, "loss": 1.8269, "step": 6039 }, { "epoch": 1.82, "learning_rate": 7.093363294372202e-05, "loss": 1.8428, "step": 6040 }, { "epoch": 1.82, "learning_rate": 7.09025394621557e-05, "loss": 1.8594, "step": 6041 }, { "epoch": 1.82, "learning_rate": 7.087144905365191e-05, "loss": 1.8756, "step": 6042 }, { "epoch": 1.82, "learning_rate": 7.084036172149411e-05, "loss": 1.854, "step": 6043 }, { "epoch": 1.82, "learning_rate": 7.080927746896562e-05, "loss": 1.8279, "step": 6044 }, { "epoch": 1.82, "learning_rate": 7.077819629934923e-05, "loss": 1.8704, "step": 6045 }, { "epoch": 1.82, "learning_rate": 7.074711821592757e-05, "loss": 1.8208, "step": 6046 }, { "epoch": 1.82, "learning_rate": 7.071604322198288e-05, "loss": 1.8344, "step": 6047 }, { "epoch": 1.82, "learning_rate": 7.068497132079706e-05, "loss": 1.8642, "step": 6048 }, { "epoch": 1.82, "learning_rate": 7.06539025156517e-05, "loss": 1.7851, "step": 6049 }, { "epoch": 1.82, "learning_rate": 7.062283680982806e-05, "loss": 1.8703, "step": 6050 }, { "epoch": 1.82, "learning_rate": 7.059177420660708e-05, "loss": 1.869, "step": 6051 }, { "epoch": 1.82, "learning_rate": 7.056071470926936e-05, "loss": 1.7101, "step": 6052 }, { "epoch": 1.82, "learning_rate": 7.052965832109518e-05, "loss": 1.9117, "step": 6053 }, { "epoch": 1.82, "learning_rate": 7.049860504536452e-05, "loss": 1.8358, "step": 6054 }, { "epoch": 1.82, "learning_rate": 7.046755488535693e-05, "loss": 1.8101, "step": 6055 }, { "epoch": 1.82, "learning_rate": 7.043650784435179e-05, "loss": 1.8784, "step": 6056 }, { "epoch": 1.82, "learning_rate": 7.040546392562799e-05, "loss": 1.8289, "step": 6057 }, { "epoch": 1.82, "learning_rate": 7.03744231324642e-05, "loss": 1.8592, "step": 6058 }, { "epoch": 1.82, "learning_rate": 7.034338546813872e-05, "loss": 1.8051, "step": 6059 }, { "epoch": 1.82, "learning_rate": 7.031235093592951e-05, "loss": 1.8203, "step": 6060 }, { "epoch": 1.82, "learning_rate": 7.028131953911426e-05, "loss": 1.8542, "step": 6061 }, { "epoch": 1.82, "learning_rate": 7.02502912809702e-05, "loss": 1.836, "step": 6062 }, { "epoch": 1.83, "learning_rate": 7.021926616477437e-05, "loss": 1.7519, "step": 6063 }, { "epoch": 1.83, "learning_rate": 7.018824419380337e-05, "loss": 1.8324, "step": 6064 }, { "epoch": 1.83, "learning_rate": 7.015722537133356e-05, "loss": 1.8787, "step": 6065 }, { "epoch": 1.83, "learning_rate": 7.012620970064092e-05, "loss": 1.8056, "step": 6066 }, { "epoch": 1.83, "learning_rate": 7.009519718500109e-05, "loss": 1.7416, "step": 6067 }, { "epoch": 1.83, "learning_rate": 7.006418782768938e-05, "loss": 1.823, "step": 6068 }, { "epoch": 1.83, "learning_rate": 7.003318163198076e-05, "loss": 1.835, "step": 6069 }, { "epoch": 1.83, "learning_rate": 7.000217860114989e-05, "loss": 1.7566, "step": 6070 }, { "epoch": 1.83, "learning_rate": 6.997117873847112e-05, "loss": 1.7846, "step": 6071 }, { "epoch": 1.83, "learning_rate": 6.994018204721838e-05, "loss": 1.8931, "step": 6072 }, { "epoch": 1.83, "learning_rate": 6.990918853066535e-05, "loss": 1.8486, "step": 6073 }, { "epoch": 1.83, "learning_rate": 6.987819819208531e-05, "loss": 1.8805, "step": 6074 }, { "epoch": 1.83, "learning_rate": 6.984721103475128e-05, "loss": 1.7046, "step": 6075 }, { "epoch": 1.83, "learning_rate": 6.981622706193582e-05, "loss": 1.8443, "step": 6076 }, { "epoch": 1.83, "learning_rate": 6.978524627691132e-05, "loss": 1.8468, "step": 6077 }, { "epoch": 1.83, "learning_rate": 6.97542686829497e-05, "loss": 1.7712, "step": 6078 }, { "epoch": 1.83, "learning_rate": 6.972329428332259e-05, "loss": 1.8692, "step": 6079 }, { "epoch": 1.83, "learning_rate": 6.969232308130132e-05, "loss": 1.8669, "step": 6080 }, { "epoch": 1.83, "learning_rate": 6.966135508015676e-05, "loss": 1.8639, "step": 6081 }, { "epoch": 1.83, "learning_rate": 6.963039028315957e-05, "loss": 1.9041, "step": 6082 }, { "epoch": 1.83, "learning_rate": 6.959942869358009e-05, "loss": 1.9138, "step": 6083 }, { "epoch": 1.83, "learning_rate": 6.956847031468818e-05, "loss": 1.833, "step": 6084 }, { "epoch": 1.83, "learning_rate": 6.953751514975346e-05, "loss": 1.8365, "step": 6085 }, { "epoch": 1.83, "learning_rate": 6.950656320204518e-05, "loss": 1.8427, "step": 6086 }, { "epoch": 1.83, "learning_rate": 6.947561447483229e-05, "loss": 1.8281, "step": 6087 }, { "epoch": 1.83, "learning_rate": 6.944466897138333e-05, "loss": 1.8129, "step": 6088 }, { "epoch": 1.83, "learning_rate": 6.941372669496653e-05, "loss": 1.8498, "step": 6089 }, { "epoch": 1.83, "learning_rate": 6.938278764884989e-05, "loss": 1.8357, "step": 6090 }, { "epoch": 1.83, "learning_rate": 6.935185183630085e-05, "loss": 1.9497, "step": 6091 }, { "epoch": 1.83, "learning_rate": 6.932091926058669e-05, "loss": 1.7727, "step": 6092 }, { "epoch": 1.83, "learning_rate": 6.928998992497423e-05, "loss": 1.7855, "step": 6093 }, { "epoch": 1.83, "learning_rate": 6.925906383273006e-05, "loss": 1.7535, "step": 6094 }, { "epoch": 1.83, "learning_rate": 6.922814098712038e-05, "loss": 1.8756, "step": 6095 }, { "epoch": 1.84, "learning_rate": 6.919722139141094e-05, "loss": 1.835, "step": 6096 }, { "epoch": 1.84, "learning_rate": 6.916630504886736e-05, "loss": 1.7263, "step": 6097 }, { "epoch": 1.84, "learning_rate": 6.913539196275473e-05, "loss": 2.0758, "step": 6098 }, { "epoch": 1.84, "learning_rate": 6.910448213633789e-05, "loss": 1.8955, "step": 6099 }, { "epoch": 1.84, "learning_rate": 6.907357557288127e-05, "loss": 1.8802, "step": 6100 }, { "epoch": 1.84, "learning_rate": 6.904267227564903e-05, "loss": 1.8215, "step": 6101 }, { "epoch": 1.84, "learning_rate": 6.9011772247905e-05, "loss": 1.8677, "step": 6102 }, { "epoch": 1.84, "learning_rate": 6.89808754929125e-05, "loss": 1.8935, "step": 6103 }, { "epoch": 1.84, "learning_rate": 6.894998201393476e-05, "loss": 1.7887, "step": 6104 }, { "epoch": 1.84, "learning_rate": 6.891909181423442e-05, "loss": 1.8097, "step": 6105 }, { "epoch": 1.84, "learning_rate": 6.888820489707388e-05, "loss": 1.8558, "step": 6106 }, { "epoch": 1.84, "learning_rate": 6.885732126571529e-05, "loss": 1.8189, "step": 6107 }, { "epoch": 1.84, "learning_rate": 6.882644092342023e-05, "loss": 1.6537, "step": 6108 }, { "epoch": 1.84, "learning_rate": 6.879556387345018e-05, "loss": 1.8052, "step": 6109 }, { "epoch": 1.84, "learning_rate": 6.876469011906601e-05, "loss": 1.7135, "step": 6110 }, { "epoch": 1.84, "learning_rate": 6.873381966352853e-05, "loss": 1.8588, "step": 6111 }, { "epoch": 1.84, "learning_rate": 6.870295251009796e-05, "loss": 1.8474, "step": 6112 }, { "epoch": 1.84, "learning_rate": 6.867208866203423e-05, "loss": 1.7836, "step": 6113 }, { "epoch": 1.84, "learning_rate": 6.86412281225971e-05, "loss": 1.8801, "step": 6114 }, { "epoch": 1.84, "learning_rate": 6.861037089504567e-05, "loss": 1.8783, "step": 6115 }, { "epoch": 1.84, "learning_rate": 6.8579516982639e-05, "loss": 1.7766, "step": 6116 }, { "epoch": 1.84, "learning_rate": 6.854866638863552e-05, "loss": 1.7742, "step": 6117 }, { "epoch": 1.84, "learning_rate": 6.851781911629356e-05, "loss": 1.78, "step": 6118 }, { "epoch": 1.84, "learning_rate": 6.848697516887097e-05, "loss": 1.7621, "step": 6119 }, { "epoch": 1.84, "learning_rate": 6.845613454962517e-05, "loss": 1.8077, "step": 6120 }, { "epoch": 1.84, "learning_rate": 6.842529726181345e-05, "loss": 1.8146, "step": 6121 }, { "epoch": 1.84, "learning_rate": 6.83944633086925e-05, "loss": 1.7449, "step": 6122 }, { "epoch": 1.84, "learning_rate": 6.836363269351887e-05, "loss": 1.9225, "step": 6123 }, { "epoch": 1.84, "learning_rate": 6.833280541954862e-05, "loss": 1.8311, "step": 6124 }, { "epoch": 1.84, "learning_rate": 6.830198149003751e-05, "loss": 1.8623, "step": 6125 }, { "epoch": 1.84, "learning_rate": 6.827116090824099e-05, "loss": 1.8219, "step": 6126 }, { "epoch": 1.84, "learning_rate": 6.824034367741399e-05, "loss": 1.8386, "step": 6127 }, { "epoch": 1.84, "learning_rate": 6.820952980081135e-05, "loss": 1.8335, "step": 6128 }, { "epoch": 1.84, "learning_rate": 6.817871928168724e-05, "loss": 1.8614, "step": 6129 }, { "epoch": 1.85, "learning_rate": 6.814791212329577e-05, "loss": 1.9907, "step": 6130 }, { "epoch": 1.85, "learning_rate": 6.811710832889057e-05, "loss": 1.9181, "step": 6131 }, { "epoch": 1.85, "learning_rate": 6.808630790172481e-05, "loss": 1.8625, "step": 6132 }, { "epoch": 1.85, "learning_rate": 6.805551084505155e-05, "loss": 1.8169, "step": 6133 }, { "epoch": 1.85, "learning_rate": 6.80247171621232e-05, "loss": 1.7787, "step": 6134 }, { "epoch": 1.85, "learning_rate": 6.799392685619205e-05, "loss": 1.8696, "step": 6135 }, { "epoch": 1.85, "learning_rate": 6.796313993050997e-05, "loss": 1.7627, "step": 6136 }, { "epoch": 1.85, "learning_rate": 6.79323563883284e-05, "loss": 1.8214, "step": 6137 }, { "epoch": 1.85, "learning_rate": 6.790157623289853e-05, "loss": 1.7959, "step": 6138 }, { "epoch": 1.85, "learning_rate": 6.787079946747104e-05, "loss": 1.9294, "step": 6139 }, { "epoch": 1.85, "learning_rate": 6.784002609529646e-05, "loss": 1.7868, "step": 6140 }, { "epoch": 1.85, "learning_rate": 6.780925611962474e-05, "loss": 1.9035, "step": 6141 }, { "epoch": 1.85, "learning_rate": 6.777848954370568e-05, "loss": 1.9614, "step": 6142 }, { "epoch": 1.85, "learning_rate": 6.774772637078856e-05, "loss": 1.8016, "step": 6143 }, { "epoch": 1.85, "learning_rate": 6.771696660412241e-05, "loss": 1.8659, "step": 6144 }, { "epoch": 1.85, "learning_rate": 6.768621024695585e-05, "loss": 1.8465, "step": 6145 }, { "epoch": 1.85, "learning_rate": 6.765545730253707e-05, "loss": 1.9291, "step": 6146 }, { "epoch": 1.85, "learning_rate": 6.762470777411402e-05, "loss": 1.6862, "step": 6147 }, { "epoch": 1.85, "learning_rate": 6.759396166493428e-05, "loss": 1.8565, "step": 6148 }, { "epoch": 1.85, "learning_rate": 6.756321897824498e-05, "loss": 1.844, "step": 6149 }, { "epoch": 1.85, "learning_rate": 6.753247971729298e-05, "loss": 1.8901, "step": 6150 }, { "epoch": 1.85, "learning_rate": 6.750174388532467e-05, "loss": 1.8698, "step": 6151 }, { "epoch": 1.85, "learning_rate": 6.747101148558625e-05, "loss": 1.8193, "step": 6152 }, { "epoch": 1.85, "learning_rate": 6.744028252132333e-05, "loss": 1.7636, "step": 6153 }, { "epoch": 1.85, "learning_rate": 6.740955699578136e-05, "loss": 1.8321, "step": 6154 }, { "epoch": 1.85, "learning_rate": 6.737883491220534e-05, "loss": 1.7276, "step": 6155 }, { "epoch": 1.85, "learning_rate": 6.734811627383987e-05, "loss": 1.9615, "step": 6156 }, { "epoch": 1.85, "learning_rate": 6.731740108392933e-05, "loss": 1.8532, "step": 6157 }, { "epoch": 1.85, "learning_rate": 6.72866893457175e-05, "loss": 1.8828, "step": 6158 }, { "epoch": 1.85, "learning_rate": 6.725598106244801e-05, "loss": 1.8531, "step": 6159 }, { "epoch": 1.85, "learning_rate": 6.722527623736406e-05, "loss": 1.7946, "step": 6160 }, { "epoch": 1.85, "learning_rate": 6.719457487370843e-05, "loss": 1.7898, "step": 6161 }, { "epoch": 1.85, "learning_rate": 6.716387697472359e-05, "loss": 1.7771, "step": 6162 }, { "epoch": 1.86, "learning_rate": 6.713318254365163e-05, "loss": 1.8782, "step": 6163 }, { "epoch": 1.86, "learning_rate": 6.710249158373429e-05, "loss": 1.7492, "step": 6164 }, { "epoch": 1.86, "learning_rate": 6.707180409821288e-05, "loss": 1.8806, "step": 6165 }, { "epoch": 1.86, "learning_rate": 6.704112009032841e-05, "loss": 1.8553, "step": 6166 }, { "epoch": 1.86, "learning_rate": 6.701043956332154e-05, "loss": 1.7673, "step": 6167 }, { "epoch": 1.86, "learning_rate": 6.697976252043247e-05, "loss": 1.8174, "step": 6168 }, { "epoch": 1.86, "learning_rate": 6.694908896490113e-05, "loss": 1.7982, "step": 6169 }, { "epoch": 1.86, "learning_rate": 6.6918418899967e-05, "loss": 1.9047, "step": 6170 }, { "epoch": 1.86, "learning_rate": 6.688775232886923e-05, "loss": 1.9459, "step": 6171 }, { "epoch": 1.86, "learning_rate": 6.685708925484666e-05, "loss": 1.8996, "step": 6172 }, { "epoch": 1.86, "learning_rate": 6.682642968113764e-05, "loss": 1.8034, "step": 6173 }, { "epoch": 1.86, "learning_rate": 6.679577361098025e-05, "loss": 1.861, "step": 6174 }, { "epoch": 1.86, "learning_rate": 6.676512104761212e-05, "loss": 1.792, "step": 6175 }, { "epoch": 1.86, "learning_rate": 6.673447199427059e-05, "loss": 1.8118, "step": 6176 }, { "epoch": 1.86, "learning_rate": 6.670382645419256e-05, "loss": 1.8837, "step": 6177 }, { "epoch": 1.86, "learning_rate": 6.66731844306146e-05, "loss": 1.8749, "step": 6178 }, { "epoch": 1.86, "learning_rate": 6.664254592677292e-05, "loss": 1.7279, "step": 6179 }, { "epoch": 1.86, "learning_rate": 6.661191094590331e-05, "loss": 1.8321, "step": 6180 }, { "epoch": 1.86, "learning_rate": 6.658127949124123e-05, "loss": 1.9087, "step": 6181 }, { "epoch": 1.86, "learning_rate": 6.655065156602174e-05, "loss": 1.8551, "step": 6182 }, { "epoch": 1.86, "learning_rate": 6.652002717347951e-05, "loss": 1.8011, "step": 6183 }, { "epoch": 1.86, "learning_rate": 6.648940631684895e-05, "loss": 1.8356, "step": 6184 }, { "epoch": 1.86, "learning_rate": 6.645878899936392e-05, "loss": 1.9602, "step": 6185 }, { "epoch": 1.86, "learning_rate": 6.64281752242581e-05, "loss": 1.7904, "step": 6186 }, { "epoch": 1.86, "learning_rate": 6.639756499476458e-05, "loss": 1.7468, "step": 6187 }, { "epoch": 1.86, "learning_rate": 6.636695831411627e-05, "loss": 1.7886, "step": 6188 }, { "epoch": 1.86, "learning_rate": 6.633635518554559e-05, "loss": 1.8176, "step": 6189 }, { "epoch": 1.86, "learning_rate": 6.630575561228463e-05, "loss": 1.9437, "step": 6190 }, { "epoch": 1.86, "learning_rate": 6.62751595975651e-05, "loss": 1.8786, "step": 6191 }, { "epoch": 1.86, "learning_rate": 6.624456714461834e-05, "loss": 1.8311, "step": 6192 }, { "epoch": 1.86, "learning_rate": 6.621397825667526e-05, "loss": 1.8182, "step": 6193 }, { "epoch": 1.86, "learning_rate": 6.618339293696647e-05, "loss": 1.8358, "step": 6194 }, { "epoch": 1.86, "learning_rate": 6.615281118872216e-05, "loss": 1.6563, "step": 6195 }, { "epoch": 1.87, "learning_rate": 6.612223301517217e-05, "loss": 1.7864, "step": 6196 }, { "epoch": 1.87, "learning_rate": 6.609165841954592e-05, "loss": 1.818, "step": 6197 }, { "epoch": 1.87, "learning_rate": 6.606108740507251e-05, "loss": 1.7804, "step": 6198 }, { "epoch": 1.87, "learning_rate": 6.603051997498057e-05, "loss": 1.8586, "step": 6199 }, { "epoch": 1.87, "learning_rate": 6.599995613249847e-05, "loss": 1.8107, "step": 6200 }, { "epoch": 1.87, "learning_rate": 6.59693958808541e-05, "loss": 1.8256, "step": 6201 }, { "epoch": 1.87, "learning_rate": 6.593883922327502e-05, "loss": 1.9766, "step": 6202 }, { "epoch": 1.87, "learning_rate": 6.590828616298844e-05, "loss": 1.9348, "step": 6203 }, { "epoch": 1.87, "learning_rate": 6.587773670322111e-05, "loss": 1.8483, "step": 6204 }, { "epoch": 1.87, "learning_rate": 6.584719084719946e-05, "loss": 1.7648, "step": 6205 }, { "epoch": 1.87, "learning_rate": 6.58166485981495e-05, "loss": 1.8246, "step": 6206 }, { "epoch": 1.87, "learning_rate": 6.57861099592969e-05, "loss": 1.9685, "step": 6207 }, { "epoch": 1.87, "learning_rate": 6.575557493386694e-05, "loss": 1.8302, "step": 6208 }, { "epoch": 1.87, "learning_rate": 6.572504352508447e-05, "loss": 1.7994, "step": 6209 }, { "epoch": 1.87, "learning_rate": 6.569451573617405e-05, "loss": 1.853, "step": 6210 }, { "epoch": 1.87, "learning_rate": 6.566399157035975e-05, "loss": 1.8613, "step": 6211 }, { "epoch": 1.87, "learning_rate": 6.563347103086535e-05, "loss": 1.7527, "step": 6212 }, { "epoch": 1.87, "learning_rate": 6.560295412091418e-05, "loss": 1.8691, "step": 6213 }, { "epoch": 1.87, "learning_rate": 6.557244084372923e-05, "loss": 1.9065, "step": 6214 }, { "epoch": 1.87, "learning_rate": 6.55419312025331e-05, "loss": 1.7338, "step": 6215 }, { "epoch": 1.87, "learning_rate": 6.551142520054795e-05, "loss": 1.7616, "step": 6216 }, { "epoch": 1.87, "learning_rate": 6.548092284099567e-05, "loss": 1.6607, "step": 6217 }, { "epoch": 1.87, "learning_rate": 6.545042412709764e-05, "loss": 1.8344, "step": 6218 }, { "epoch": 1.87, "learning_rate": 6.541992906207493e-05, "loss": 1.8331, "step": 6219 }, { "epoch": 1.87, "learning_rate": 6.538943764914824e-05, "loss": 1.7247, "step": 6220 }, { "epoch": 1.87, "learning_rate": 6.535894989153782e-05, "loss": 1.8763, "step": 6221 }, { "epoch": 1.87, "learning_rate": 6.532846579246357e-05, "loss": 1.8292, "step": 6222 }, { "epoch": 1.87, "learning_rate": 6.529798535514498e-05, "loss": 1.7344, "step": 6223 }, { "epoch": 1.87, "learning_rate": 6.526750858280122e-05, "loss": 1.8041, "step": 6224 }, { "epoch": 1.87, "learning_rate": 6.523703547865098e-05, "loss": 1.8139, "step": 6225 }, { "epoch": 1.87, "learning_rate": 6.520656604591261e-05, "loss": 1.8895, "step": 6226 }, { "epoch": 1.87, "learning_rate": 6.517610028780412e-05, "loss": 1.734, "step": 6227 }, { "epoch": 1.87, "learning_rate": 6.514563820754302e-05, "loss": 1.78, "step": 6228 }, { "epoch": 1.88, "learning_rate": 6.511517980834656e-05, "loss": 1.8192, "step": 6229 }, { "epoch": 1.88, "learning_rate": 6.508472509343147e-05, "loss": 1.8026, "step": 6230 }, { "epoch": 1.88, "learning_rate": 6.505427406601418e-05, "loss": 1.8134, "step": 6231 }, { "epoch": 1.88, "learning_rate": 6.502382672931072e-05, "loss": 1.8856, "step": 6232 }, { "epoch": 1.88, "learning_rate": 6.49933830865367e-05, "loss": 1.8586, "step": 6233 }, { "epoch": 1.88, "learning_rate": 6.496294314090738e-05, "loss": 1.8191, "step": 6234 }, { "epoch": 1.88, "learning_rate": 6.493250689563756e-05, "loss": 1.8096, "step": 6235 }, { "epoch": 1.88, "learning_rate": 6.490207435394174e-05, "loss": 1.8444, "step": 6236 }, { "epoch": 1.88, "learning_rate": 6.487164551903395e-05, "loss": 1.888, "step": 6237 }, { "epoch": 1.88, "learning_rate": 6.484122039412789e-05, "loss": 1.8855, "step": 6238 }, { "epoch": 1.88, "learning_rate": 6.481079898243685e-05, "loss": 1.8382, "step": 6239 }, { "epoch": 1.88, "learning_rate": 6.478038128717367e-05, "loss": 1.8658, "step": 6240 }, { "epoch": 1.88, "learning_rate": 6.47499673115509e-05, "loss": 1.9074, "step": 6241 }, { "epoch": 1.88, "learning_rate": 6.47195570587806e-05, "loss": 1.842, "step": 6242 }, { "epoch": 1.88, "learning_rate": 6.468915053207447e-05, "loss": 1.8262, "step": 6243 }, { "epoch": 1.88, "learning_rate": 6.465874773464393e-05, "loss": 1.8823, "step": 6244 }, { "epoch": 1.88, "learning_rate": 6.462834866969978e-05, "loss": 1.827, "step": 6245 }, { "epoch": 1.88, "learning_rate": 6.459795334045262e-05, "loss": 1.8467, "step": 6246 }, { "epoch": 1.88, "learning_rate": 6.456756175011254e-05, "loss": 1.8013, "step": 6247 }, { "epoch": 1.88, "learning_rate": 6.453717390188928e-05, "loss": 1.8608, "step": 6248 }, { "epoch": 1.88, "learning_rate": 6.450678979899227e-05, "loss": 1.7789, "step": 6249 }, { "epoch": 1.88, "learning_rate": 6.447640944463031e-05, "loss": 1.778, "step": 6250 }, { "epoch": 1.88, "learning_rate": 6.444603284201212e-05, "loss": 1.8068, "step": 6251 }, { "epoch": 1.88, "learning_rate": 6.441565999434571e-05, "loss": 1.7604, "step": 6252 }, { "epoch": 1.88, "learning_rate": 6.438529090483893e-05, "loss": 1.8574, "step": 6253 }, { "epoch": 1.88, "learning_rate": 6.43549255766991e-05, "loss": 1.8429, "step": 6254 }, { "epoch": 1.88, "learning_rate": 6.432456401313317e-05, "loss": 1.9573, "step": 6255 }, { "epoch": 1.88, "learning_rate": 6.42942062173478e-05, "loss": 1.7666, "step": 6256 }, { "epoch": 1.88, "learning_rate": 6.426385219254905e-05, "loss": 1.7189, "step": 6257 }, { "epoch": 1.88, "learning_rate": 6.423350194194277e-05, "loss": 1.8373, "step": 6258 }, { "epoch": 1.88, "learning_rate": 6.420315546873426e-05, "loss": 1.7962, "step": 6259 }, { "epoch": 1.88, "learning_rate": 6.417281277612853e-05, "loss": 1.7285, "step": 6260 }, { "epoch": 1.88, "learning_rate": 6.41424738673302e-05, "loss": 1.8346, "step": 6261 }, { "epoch": 1.89, "learning_rate": 6.411213874554339e-05, "loss": 1.7449, "step": 6262 }, { "epoch": 1.89, "learning_rate": 6.408180741397192e-05, "loss": 1.8096, "step": 6263 }, { "epoch": 1.89, "learning_rate": 6.40514798758191e-05, "loss": 1.8912, "step": 6264 }, { "epoch": 1.89, "learning_rate": 6.402115613428796e-05, "loss": 1.9575, "step": 6265 }, { "epoch": 1.89, "learning_rate": 6.399083619258101e-05, "loss": 1.7975, "step": 6266 }, { "epoch": 1.89, "learning_rate": 6.396052005390047e-05, "loss": 1.7271, "step": 6267 }, { "epoch": 1.89, "learning_rate": 6.393020772144818e-05, "loss": 1.7677, "step": 6268 }, { "epoch": 1.89, "learning_rate": 6.389989919842533e-05, "loss": 1.7706, "step": 6269 }, { "epoch": 1.89, "learning_rate": 6.38695944880331e-05, "loss": 1.8232, "step": 6270 }, { "epoch": 1.89, "learning_rate": 6.383929359347188e-05, "loss": 1.8717, "step": 6271 }, { "epoch": 1.89, "learning_rate": 6.380899651794186e-05, "loss": 1.9223, "step": 6272 }, { "epoch": 1.89, "learning_rate": 6.377870326464292e-05, "loss": 1.7844, "step": 6273 }, { "epoch": 1.89, "learning_rate": 6.374841383677424e-05, "loss": 1.7468, "step": 6274 }, { "epoch": 1.89, "learning_rate": 6.371812823753493e-05, "loss": 1.8945, "step": 6275 }, { "epoch": 1.89, "learning_rate": 6.36878464701234e-05, "loss": 1.866, "step": 6276 }, { "epoch": 1.89, "learning_rate": 6.36575685377379e-05, "loss": 1.8562, "step": 6277 }, { "epoch": 1.89, "learning_rate": 6.362729444357607e-05, "loss": 1.7845, "step": 6278 }, { "epoch": 1.89, "learning_rate": 6.359702419083528e-05, "loss": 1.8416, "step": 6279 }, { "epoch": 1.89, "learning_rate": 6.356675778271251e-05, "loss": 1.7748, "step": 6280 }, { "epoch": 1.89, "learning_rate": 6.353649522240416e-05, "loss": 1.8186, "step": 6281 }, { "epoch": 1.89, "learning_rate": 6.35062365131065e-05, "loss": 1.7797, "step": 6282 }, { "epoch": 1.89, "learning_rate": 6.347598165801506e-05, "loss": 1.8874, "step": 6283 }, { "epoch": 1.89, "learning_rate": 6.344573066032524e-05, "loss": 1.8527, "step": 6284 }, { "epoch": 1.89, "learning_rate": 6.341548352323197e-05, "loss": 1.8622, "step": 6285 }, { "epoch": 1.89, "learning_rate": 6.33852402499296e-05, "loss": 1.8474, "step": 6286 }, { "epoch": 1.89, "learning_rate": 6.335500084361235e-05, "loss": 1.8076, "step": 6287 }, { "epoch": 1.89, "learning_rate": 6.332476530747376e-05, "loss": 1.7999, "step": 6288 }, { "epoch": 1.89, "learning_rate": 6.32945336447072e-05, "loss": 1.8896, "step": 6289 }, { "epoch": 1.89, "learning_rate": 6.32643058585054e-05, "loss": 1.9093, "step": 6290 }, { "epoch": 1.89, "learning_rate": 6.32340819520609e-05, "loss": 1.7603, "step": 6291 }, { "epoch": 1.89, "learning_rate": 6.320386192856572e-05, "loss": 1.7956, "step": 6292 }, { "epoch": 1.89, "learning_rate": 6.317364579121139e-05, "loss": 1.7179, "step": 6293 }, { "epoch": 1.89, "learning_rate": 6.314343354318925e-05, "loss": 1.7548, "step": 6294 }, { "epoch": 1.89, "learning_rate": 6.311322518768996e-05, "loss": 1.8734, "step": 6295 }, { "epoch": 1.9, "learning_rate": 6.308302072790402e-05, "loss": 1.797, "step": 6296 }, { "epoch": 1.9, "learning_rate": 6.305282016702137e-05, "loss": 1.8205, "step": 6297 }, { "epoch": 1.9, "learning_rate": 6.302262350823154e-05, "loss": 1.9977, "step": 6298 }, { "epoch": 1.9, "learning_rate": 6.299243075472374e-05, "loss": 1.7701, "step": 6299 }, { "epoch": 1.9, "learning_rate": 6.296224190968663e-05, "loss": 1.8099, "step": 6300 }, { "epoch": 1.9, "learning_rate": 6.293205697630864e-05, "loss": 1.7527, "step": 6301 }, { "epoch": 1.9, "learning_rate": 6.290187595777757e-05, "loss": 1.7976, "step": 6302 }, { "epoch": 1.9, "learning_rate": 6.287169885728102e-05, "loss": 1.8056, "step": 6303 }, { "epoch": 1.9, "learning_rate": 6.284152567800602e-05, "loss": 1.7938, "step": 6304 }, { "epoch": 1.9, "learning_rate": 6.281135642313925e-05, "loss": 1.8445, "step": 6305 }, { "epoch": 1.9, "learning_rate": 6.278119109586702e-05, "loss": 1.772, "step": 6306 }, { "epoch": 1.9, "learning_rate": 6.275102969937505e-05, "loss": 1.8122, "step": 6307 }, { "epoch": 1.9, "learning_rate": 6.272087223684888e-05, "loss": 1.8404, "step": 6308 }, { "epoch": 1.9, "learning_rate": 6.269071871147351e-05, "loss": 1.897, "step": 6309 }, { "epoch": 1.9, "learning_rate": 6.266056912643348e-05, "loss": 1.7253, "step": 6310 }, { "epoch": 1.9, "learning_rate": 6.263042348491303e-05, "loss": 1.7615, "step": 6311 }, { "epoch": 1.9, "learning_rate": 6.260028179009587e-05, "loss": 1.7384, "step": 6312 }, { "epoch": 1.9, "learning_rate": 6.257014404516542e-05, "loss": 1.739, "step": 6313 }, { "epoch": 1.9, "learning_rate": 6.25400102533045e-05, "loss": 1.7828, "step": 6314 }, { "epoch": 1.9, "learning_rate": 6.25098804176957e-05, "loss": 1.8683, "step": 6315 }, { "epoch": 1.9, "learning_rate": 6.247975454152114e-05, "loss": 1.8841, "step": 6316 }, { "epoch": 1.9, "learning_rate": 6.244963262796242e-05, "loss": 1.9258, "step": 6317 }, { "epoch": 1.9, "learning_rate": 6.241951468020087e-05, "loss": 1.8936, "step": 6318 }, { "epoch": 1.9, "learning_rate": 6.238940070141725e-05, "loss": 1.8059, "step": 6319 }, { "epoch": 1.9, "learning_rate": 6.235929069479205e-05, "loss": 1.7819, "step": 6320 }, { "epoch": 1.9, "learning_rate": 6.232918466350526e-05, "loss": 1.763, "step": 6321 }, { "epoch": 1.9, "learning_rate": 6.22990826107364e-05, "loss": 1.8242, "step": 6322 }, { "epoch": 1.9, "learning_rate": 6.226898453966473e-05, "loss": 1.8198, "step": 6323 }, { "epoch": 1.9, "learning_rate": 6.223889045346888e-05, "loss": 1.8919, "step": 6324 }, { "epoch": 1.9, "learning_rate": 6.220880035532726e-05, "loss": 1.7941, "step": 6325 }, { "epoch": 1.9, "learning_rate": 6.217871424841771e-05, "loss": 1.9564, "step": 6326 }, { "epoch": 1.9, "learning_rate": 6.214863213591771e-05, "loss": 1.8318, "step": 6327 }, { "epoch": 1.9, "learning_rate": 6.211855402100437e-05, "loss": 1.8838, "step": 6328 }, { "epoch": 1.91, "learning_rate": 6.208847990685425e-05, "loss": 1.8862, "step": 6329 }, { "epoch": 1.91, "learning_rate": 6.205840979664361e-05, "loss": 1.8104, "step": 6330 }, { "epoch": 1.91, "learning_rate": 6.202834369354821e-05, "loss": 1.7663, "step": 6331 }, { "epoch": 1.91, "learning_rate": 6.199828160074343e-05, "loss": 1.8157, "step": 6332 }, { "epoch": 1.91, "learning_rate": 6.196822352140421e-05, "loss": 1.8024, "step": 6333 }, { "epoch": 1.91, "learning_rate": 6.193816945870505e-05, "loss": 1.8355, "step": 6334 }, { "epoch": 1.91, "learning_rate": 6.190811941582005e-05, "loss": 1.9187, "step": 6335 }, { "epoch": 1.91, "learning_rate": 6.187807339592288e-05, "loss": 1.8801, "step": 6336 }, { "epoch": 1.91, "learning_rate": 6.184803140218678e-05, "loss": 1.9425, "step": 6337 }, { "epoch": 1.91, "learning_rate": 6.181799343778456e-05, "loss": 1.7971, "step": 6338 }, { "epoch": 1.91, "learning_rate": 6.178795950588863e-05, "loss": 1.8028, "step": 6339 }, { "epoch": 1.91, "learning_rate": 6.175792960967094e-05, "loss": 1.9339, "step": 6340 }, { "epoch": 1.91, "learning_rate": 6.172790375230303e-05, "loss": 1.7482, "step": 6341 }, { "epoch": 1.91, "learning_rate": 6.169788193695603e-05, "loss": 1.8329, "step": 6342 }, { "epoch": 1.91, "learning_rate": 6.16678641668006e-05, "loss": 1.83, "step": 6343 }, { "epoch": 1.91, "learning_rate": 6.163785044500701e-05, "loss": 1.8483, "step": 6344 }, { "epoch": 1.91, "learning_rate": 6.160784077474512e-05, "loss": 1.7896, "step": 6345 }, { "epoch": 1.91, "learning_rate": 6.157783515918428e-05, "loss": 1.8766, "step": 6346 }, { "epoch": 1.91, "learning_rate": 6.154783360149352e-05, "loss": 1.8292, "step": 6347 }, { "epoch": 1.91, "learning_rate": 6.151783610484133e-05, "loss": 1.852, "step": 6348 }, { "epoch": 1.91, "learning_rate": 6.148784267239588e-05, "loss": 1.7326, "step": 6349 }, { "epoch": 1.91, "learning_rate": 6.145785330732482e-05, "loss": 1.9056, "step": 6350 }, { "epoch": 1.91, "learning_rate": 6.142786801279542e-05, "loss": 1.7605, "step": 6351 }, { "epoch": 1.91, "learning_rate": 6.139788679197454e-05, "loss": 1.8124, "step": 6352 }, { "epoch": 1.91, "learning_rate": 6.136790964802852e-05, "loss": 1.7942, "step": 6353 }, { "epoch": 1.91, "learning_rate": 6.133793658412339e-05, "loss": 1.7081, "step": 6354 }, { "epoch": 1.91, "learning_rate": 6.130796760342464e-05, "loss": 1.7708, "step": 6355 }, { "epoch": 1.91, "learning_rate": 6.12780027090974e-05, "loss": 1.9722, "step": 6356 }, { "epoch": 1.91, "learning_rate": 6.124804190430635e-05, "loss": 1.919, "step": 6357 }, { "epoch": 1.91, "learning_rate": 6.12180851922157e-05, "loss": 1.7929, "step": 6358 }, { "epoch": 1.91, "learning_rate": 6.11881325759893e-05, "loss": 1.884, "step": 6359 }, { "epoch": 1.91, "learning_rate": 6.11581840587905e-05, "loss": 1.6846, "step": 6360 }, { "epoch": 1.91, "learning_rate": 6.112823964378225e-05, "loss": 1.8229, "step": 6361 }, { "epoch": 1.92, "learning_rate": 6.109829933412707e-05, "loss": 1.8707, "step": 6362 }, { "epoch": 1.92, "learning_rate": 6.106836313298704e-05, "loss": 1.8561, "step": 6363 }, { "epoch": 1.92, "learning_rate": 6.10384310435238e-05, "loss": 1.9133, "step": 6364 }, { "epoch": 1.92, "learning_rate": 6.100850306889856e-05, "loss": 1.8351, "step": 6365 }, { "epoch": 1.92, "learning_rate": 6.09785792122721e-05, "loss": 1.8179, "step": 6366 }, { "epoch": 1.92, "learning_rate": 6.094865947680473e-05, "loss": 1.8218, "step": 6367 }, { "epoch": 1.92, "learning_rate": 6.091874386565639e-05, "loss": 1.8122, "step": 6368 }, { "epoch": 1.92, "learning_rate": 6.088883238198655e-05, "loss": 1.8886, "step": 6369 }, { "epoch": 1.92, "learning_rate": 6.0858925028954225e-05, "loss": 1.9403, "step": 6370 }, { "epoch": 1.92, "learning_rate": 6.0829021809718036e-05, "loss": 1.8084, "step": 6371 }, { "epoch": 1.92, "learning_rate": 6.0799122727436095e-05, "loss": 1.8991, "step": 6372 }, { "epoch": 1.92, "learning_rate": 6.076922778526617e-05, "loss": 1.9086, "step": 6373 }, { "epoch": 1.92, "learning_rate": 6.073933698636556e-05, "loss": 1.7484, "step": 6374 }, { "epoch": 1.92, "learning_rate": 6.070945033389106e-05, "loss": 1.8015, "step": 6375 }, { "epoch": 1.92, "learning_rate": 6.067956783099915e-05, "loss": 1.7351, "step": 6376 }, { "epoch": 1.92, "learning_rate": 6.064968948084573e-05, "loss": 1.8244, "step": 6377 }, { "epoch": 1.92, "learning_rate": 6.06198152865864e-05, "loss": 1.8121, "step": 6378 }, { "epoch": 1.92, "learning_rate": 6.05899452513762e-05, "loss": 1.7931, "step": 6379 }, { "epoch": 1.92, "learning_rate": 6.0560079378369806e-05, "loss": 1.8568, "step": 6380 }, { "epoch": 1.92, "learning_rate": 6.0530217670721466e-05, "loss": 1.8416, "step": 6381 }, { "epoch": 1.92, "learning_rate": 6.050036013158491e-05, "loss": 1.7187, "step": 6382 }, { "epoch": 1.92, "learning_rate": 6.0470506764113524e-05, "loss": 1.8275, "step": 6383 }, { "epoch": 1.92, "learning_rate": 6.044065757146014e-05, "loss": 2.0434, "step": 6384 }, { "epoch": 1.92, "learning_rate": 6.0410812556777254e-05, "loss": 1.7973, "step": 6385 }, { "epoch": 1.92, "learning_rate": 6.0380971723216894e-05, "loss": 1.9444, "step": 6386 }, { "epoch": 1.92, "learning_rate": 6.0351135073930596e-05, "loss": 1.8182, "step": 6387 }, { "epoch": 1.92, "learning_rate": 6.032130261206952e-05, "loss": 1.8541, "step": 6388 }, { "epoch": 1.92, "learning_rate": 6.0291474340784325e-05, "loss": 1.9043, "step": 6389 }, { "epoch": 1.92, "learning_rate": 6.0261650263225276e-05, "loss": 1.8597, "step": 6390 }, { "epoch": 1.92, "learning_rate": 6.023183038254217e-05, "loss": 1.8222, "step": 6391 }, { "epoch": 1.92, "learning_rate": 6.020201470188436e-05, "loss": 1.893, "step": 6392 }, { "epoch": 1.92, "learning_rate": 6.01722032244008e-05, "loss": 1.7689, "step": 6393 }, { "epoch": 1.92, "learning_rate": 6.0142395953239896e-05, "loss": 1.8436, "step": 6394 }, { "epoch": 1.93, "learning_rate": 6.0112592891549735e-05, "loss": 1.7819, "step": 6395 }, { "epoch": 1.93, "learning_rate": 6.008279404247785e-05, "loss": 1.7721, "step": 6396 }, { "epoch": 1.93, "learning_rate": 6.0052999409171415e-05, "loss": 1.8283, "step": 6397 }, { "epoch": 1.93, "learning_rate": 6.002320899477714e-05, "loss": 1.8184, "step": 6398 }, { "epoch": 1.93, "learning_rate": 5.9993422802441203e-05, "loss": 1.7873, "step": 6399 }, { "epoch": 1.93, "learning_rate": 5.9963640835309475e-05, "loss": 1.738, "step": 6400 }, { "epoch": 1.93, "learning_rate": 5.993386309652726e-05, "loss": 1.7876, "step": 6401 }, { "epoch": 1.93, "learning_rate": 5.990408958923952e-05, "loss": 1.7576, "step": 6402 }, { "epoch": 1.93, "learning_rate": 5.987432031659066e-05, "loss": 1.8269, "step": 6403 }, { "epoch": 1.93, "learning_rate": 5.984455528172473e-05, "loss": 1.8548, "step": 6404 }, { "epoch": 1.93, "learning_rate": 5.981479448778529e-05, "loss": 1.75, "step": 6405 }, { "epoch": 1.93, "learning_rate": 5.978503793791545e-05, "loss": 1.7753, "step": 6406 }, { "epoch": 1.93, "learning_rate": 5.97552856352579e-05, "loss": 1.8746, "step": 6407 }, { "epoch": 1.93, "learning_rate": 5.9725537582954824e-05, "loss": 1.951, "step": 6408 }, { "epoch": 1.93, "learning_rate": 5.9695793784148015e-05, "loss": 1.817, "step": 6409 }, { "epoch": 1.93, "learning_rate": 5.966605424197885e-05, "loss": 1.8739, "step": 6410 }, { "epoch": 1.93, "learning_rate": 5.963631895958812e-05, "loss": 1.8694, "step": 6411 }, { "epoch": 1.93, "learning_rate": 5.960658794011631e-05, "loss": 1.7787, "step": 6412 }, { "epoch": 1.93, "learning_rate": 5.957686118670335e-05, "loss": 1.7958, "step": 6413 }, { "epoch": 1.93, "learning_rate": 5.95471387024888e-05, "loss": 1.8256, "step": 6414 }, { "epoch": 1.93, "learning_rate": 5.951742049061168e-05, "loss": 1.8091, "step": 6415 }, { "epoch": 1.93, "learning_rate": 5.9487706554210655e-05, "loss": 1.7931, "step": 6416 }, { "epoch": 1.93, "learning_rate": 5.945799689642395e-05, "loss": 1.8207, "step": 6417 }, { "epoch": 1.93, "learning_rate": 5.942829152038917e-05, "loss": 1.8543, "step": 6418 }, { "epoch": 1.93, "learning_rate": 5.939859042924365e-05, "loss": 1.8583, "step": 6419 }, { "epoch": 1.93, "learning_rate": 5.936889362612417e-05, "loss": 1.8352, "step": 6420 }, { "epoch": 1.93, "learning_rate": 5.9339201114167084e-05, "loss": 1.91, "step": 6421 }, { "epoch": 1.93, "learning_rate": 5.930951289650839e-05, "loss": 1.808, "step": 6422 }, { "epoch": 1.93, "learning_rate": 5.927982897628342e-05, "loss": 1.8634, "step": 6423 }, { "epoch": 1.93, "learning_rate": 5.925014935662728e-05, "loss": 1.8848, "step": 6424 }, { "epoch": 1.93, "learning_rate": 5.9220474040674436e-05, "loss": 1.7575, "step": 6425 }, { "epoch": 1.93, "learning_rate": 5.9190803031559015e-05, "loss": 1.895, "step": 6426 }, { "epoch": 1.93, "learning_rate": 5.916113633241462e-05, "loss": 1.7368, "step": 6427 }, { "epoch": 1.93, "learning_rate": 5.913147394637446e-05, "loss": 1.7006, "step": 6428 }, { "epoch": 1.94, "learning_rate": 5.910181587657131e-05, "loss": 1.9112, "step": 6429 }, { "epoch": 1.94, "learning_rate": 5.907216212613733e-05, "loss": 1.9446, "step": 6430 }, { "epoch": 1.94, "learning_rate": 5.904251269820446e-05, "loss": 1.8908, "step": 6431 }, { "epoch": 1.94, "learning_rate": 5.901286759590394e-05, "loss": 1.7374, "step": 6432 }, { "epoch": 1.94, "learning_rate": 5.898322682236671e-05, "loss": 1.7575, "step": 6433 }, { "epoch": 1.94, "learning_rate": 5.8953590380723275e-05, "loss": 1.8836, "step": 6434 }, { "epoch": 1.94, "learning_rate": 5.892395827410353e-05, "loss": 1.7586, "step": 6435 }, { "epoch": 1.94, "learning_rate": 5.889433050563709e-05, "loss": 1.8942, "step": 6436 }, { "epoch": 1.94, "learning_rate": 5.886470707845292e-05, "loss": 1.8344, "step": 6437 }, { "epoch": 1.94, "learning_rate": 5.883508799567975e-05, "loss": 1.7699, "step": 6438 }, { "epoch": 1.94, "learning_rate": 5.8805473260445634e-05, "loss": 1.8067, "step": 6439 }, { "epoch": 1.94, "learning_rate": 5.8775862875878285e-05, "loss": 1.9147, "step": 6440 }, { "epoch": 1.94, "learning_rate": 5.874625684510501e-05, "loss": 1.8766, "step": 6441 }, { "epoch": 1.94, "learning_rate": 5.871665517125247e-05, "loss": 1.7857, "step": 6442 }, { "epoch": 1.94, "learning_rate": 5.8687057857447094e-05, "loss": 1.7509, "step": 6443 }, { "epoch": 1.94, "learning_rate": 5.865746490681462e-05, "loss": 1.8585, "step": 6444 }, { "epoch": 1.94, "learning_rate": 5.862787632248052e-05, "loss": 1.835, "step": 6445 }, { "epoch": 1.94, "learning_rate": 5.8598292107569744e-05, "loss": 1.8392, "step": 6446 }, { "epoch": 1.94, "learning_rate": 5.856871226520666e-05, "loss": 1.9447, "step": 6447 }, { "epoch": 1.94, "learning_rate": 5.853913679851539e-05, "loss": 1.8026, "step": 6448 }, { "epoch": 1.94, "learning_rate": 5.8509565710619386e-05, "loss": 1.7431, "step": 6449 }, { "epoch": 1.94, "learning_rate": 5.847999900464183e-05, "loss": 1.866, "step": 6450 }, { "epoch": 1.94, "learning_rate": 5.8450436683705226e-05, "loss": 1.7397, "step": 6451 }, { "epoch": 1.94, "learning_rate": 5.842087875093181e-05, "loss": 1.8387, "step": 6452 }, { "epoch": 1.94, "learning_rate": 5.83913252094433e-05, "loss": 1.7785, "step": 6453 }, { "epoch": 1.94, "learning_rate": 5.836177606236082e-05, "loss": 1.799, "step": 6454 }, { "epoch": 1.94, "learning_rate": 5.8332231312805264e-05, "loss": 1.8036, "step": 6455 }, { "epoch": 1.94, "learning_rate": 5.8302690963896835e-05, "loss": 1.7437, "step": 6456 }, { "epoch": 1.94, "learning_rate": 5.827315501875541e-05, "loss": 1.8495, "step": 6457 }, { "epoch": 1.94, "learning_rate": 5.824362348050034e-05, "loss": 1.8326, "step": 6458 }, { "epoch": 1.94, "learning_rate": 5.821409635225057e-05, "loss": 1.7892, "step": 6459 }, { "epoch": 1.94, "learning_rate": 5.8184573637124554e-05, "loss": 1.8587, "step": 6460 }, { "epoch": 1.94, "learning_rate": 5.81550553382402e-05, "loss": 1.8454, "step": 6461 }, { "epoch": 1.95, "learning_rate": 5.8125541458715026e-05, "loss": 1.7601, "step": 6462 }, { "epoch": 1.95, "learning_rate": 5.809603200166611e-05, "loss": 1.7885, "step": 6463 }, { "epoch": 1.95, "learning_rate": 5.8066526970210025e-05, "loss": 1.9397, "step": 6464 }, { "epoch": 1.95, "learning_rate": 5.803702636746289e-05, "loss": 1.8337, "step": 6465 }, { "epoch": 1.95, "learning_rate": 5.8007530196540286e-05, "loss": 1.8911, "step": 6466 }, { "epoch": 1.95, "learning_rate": 5.7978038460557424e-05, "loss": 1.8017, "step": 6467 }, { "epoch": 1.95, "learning_rate": 5.794855116262899e-05, "loss": 1.8096, "step": 6468 }, { "epoch": 1.95, "learning_rate": 5.791906830586924e-05, "loss": 1.8588, "step": 6469 }, { "epoch": 1.95, "learning_rate": 5.7889589893391974e-05, "loss": 1.8237, "step": 6470 }, { "epoch": 1.95, "learning_rate": 5.78601159283104e-05, "loss": 1.8247, "step": 6471 }, { "epoch": 1.95, "learning_rate": 5.783064641373739e-05, "loss": 1.6898, "step": 6472 }, { "epoch": 1.95, "learning_rate": 5.780118135278529e-05, "loss": 1.8314, "step": 6473 }, { "epoch": 1.95, "learning_rate": 5.777172074856601e-05, "loss": 1.9396, "step": 6474 }, { "epoch": 1.95, "learning_rate": 5.7742264604190986e-05, "loss": 1.785, "step": 6475 }, { "epoch": 1.95, "learning_rate": 5.7712812922771076e-05, "loss": 1.8662, "step": 6476 }, { "epoch": 1.95, "learning_rate": 5.7683365707416806e-05, "loss": 1.8146, "step": 6477 }, { "epoch": 1.95, "learning_rate": 5.765392296123817e-05, "loss": 1.8055, "step": 6478 }, { "epoch": 1.95, "learning_rate": 5.762448468734474e-05, "loss": 1.8161, "step": 6479 }, { "epoch": 1.95, "learning_rate": 5.7595050888845506e-05, "loss": 1.9006, "step": 6480 }, { "epoch": 1.95, "learning_rate": 5.756562156884906e-05, "loss": 1.8392, "step": 6481 }, { "epoch": 1.95, "learning_rate": 5.7536196730463543e-05, "loss": 1.9085, "step": 6482 }, { "epoch": 1.95, "learning_rate": 5.750677637679658e-05, "loss": 1.8694, "step": 6483 }, { "epoch": 1.95, "learning_rate": 5.7477360510955383e-05, "loss": 1.7731, "step": 6484 }, { "epoch": 1.95, "learning_rate": 5.744794913604654e-05, "loss": 1.7877, "step": 6485 }, { "epoch": 1.95, "learning_rate": 5.7418542255176286e-05, "loss": 1.9047, "step": 6486 }, { "epoch": 1.95, "learning_rate": 5.738913987145049e-05, "loss": 1.9055, "step": 6487 }, { "epoch": 1.95, "learning_rate": 5.7359741987974294e-05, "loss": 1.7735, "step": 6488 }, { "epoch": 1.95, "learning_rate": 5.7330348607852534e-05, "loss": 1.7572, "step": 6489 }, { "epoch": 1.95, "learning_rate": 5.73009597341895e-05, "loss": 1.7381, "step": 6490 }, { "epoch": 1.95, "learning_rate": 5.7271575370089024e-05, "loss": 1.8248, "step": 6491 }, { "epoch": 1.95, "learning_rate": 5.724219551865451e-05, "loss": 1.8297, "step": 6492 }, { "epoch": 1.95, "learning_rate": 5.7212820182988814e-05, "loss": 1.8062, "step": 6493 }, { "epoch": 1.95, "learning_rate": 5.71834493661944e-05, "loss": 1.8002, "step": 6494 }, { "epoch": 1.96, "learning_rate": 5.715408307137311e-05, "loss": 1.7623, "step": 6495 }, { "epoch": 1.96, "learning_rate": 5.7124721301626464e-05, "loss": 1.8009, "step": 6496 }, { "epoch": 1.96, "learning_rate": 5.70953640600554e-05, "loss": 1.7609, "step": 6497 }, { "epoch": 1.96, "learning_rate": 5.706601134976045e-05, "loss": 1.7716, "step": 6498 }, { "epoch": 1.96, "learning_rate": 5.7036663173841655e-05, "loss": 1.8927, "step": 6499 }, { "epoch": 1.96, "learning_rate": 5.700731953539844e-05, "loss": 1.8269, "step": 6500 }, { "epoch": 1.96, "learning_rate": 5.6977980437530045e-05, "loss": 1.9563, "step": 6501 }, { "epoch": 1.96, "learning_rate": 5.6948645883334904e-05, "loss": 1.8924, "step": 6502 }, { "epoch": 1.96, "learning_rate": 5.691931587591122e-05, "loss": 1.8934, "step": 6503 }, { "epoch": 1.96, "learning_rate": 5.6889990418356524e-05, "loss": 1.848, "step": 6504 }, { "epoch": 1.96, "learning_rate": 5.686066951376796e-05, "loss": 1.8235, "step": 6505 }, { "epoch": 1.96, "learning_rate": 5.683135316524231e-05, "loss": 1.8421, "step": 6506 }, { "epoch": 1.96, "learning_rate": 5.6802041375875636e-05, "loss": 1.8876, "step": 6507 }, { "epoch": 1.96, "learning_rate": 5.677273414876372e-05, "loss": 1.7622, "step": 6508 }, { "epoch": 1.96, "learning_rate": 5.674343148700169e-05, "loss": 1.7737, "step": 6509 }, { "epoch": 1.96, "learning_rate": 5.6714133393684275e-05, "loss": 1.8282, "step": 6510 }, { "epoch": 1.96, "learning_rate": 5.668483987190587e-05, "loss": 1.8819, "step": 6511 }, { "epoch": 1.96, "learning_rate": 5.6655550924760116e-05, "loss": 1.7127, "step": 6512 }, { "epoch": 1.96, "learning_rate": 5.662626655534036e-05, "loss": 1.935, "step": 6513 }, { "epoch": 1.96, "learning_rate": 5.65969867667393e-05, "loss": 1.8989, "step": 6514 }, { "epoch": 1.96, "learning_rate": 5.656771156204941e-05, "loss": 1.8815, "step": 6515 }, { "epoch": 1.96, "learning_rate": 5.65384409443624e-05, "loss": 1.9358, "step": 6516 }, { "epoch": 1.96, "learning_rate": 5.650917491676968e-05, "loss": 1.7239, "step": 6517 }, { "epoch": 1.96, "learning_rate": 5.6479913482362146e-05, "loss": 1.8918, "step": 6518 }, { "epoch": 1.96, "learning_rate": 5.6450656644230035e-05, "loss": 1.8419, "step": 6519 }, { "epoch": 1.96, "learning_rate": 5.642140440546344e-05, "loss": 1.8621, "step": 6520 }, { "epoch": 1.96, "learning_rate": 5.639215676915163e-05, "loss": 1.8351, "step": 6521 }, { "epoch": 1.96, "learning_rate": 5.636291373838356e-05, "loss": 1.6588, "step": 6522 }, { "epoch": 1.96, "learning_rate": 5.633367531624772e-05, "loss": 1.7188, "step": 6523 }, { "epoch": 1.96, "learning_rate": 5.630444150583193e-05, "loss": 1.8548, "step": 6524 }, { "epoch": 1.96, "learning_rate": 5.6275212310223824e-05, "loss": 1.7621, "step": 6525 }, { "epoch": 1.96, "learning_rate": 5.624598773251025e-05, "loss": 1.8767, "step": 6526 }, { "epoch": 1.96, "learning_rate": 5.621676777577778e-05, "loss": 1.7892, "step": 6527 }, { "epoch": 1.97, "learning_rate": 5.6187552443112336e-05, "loss": 1.9322, "step": 6528 }, { "epoch": 1.97, "learning_rate": 5.615834173759941e-05, "loss": 1.864, "step": 6529 }, { "epoch": 1.97, "learning_rate": 5.612913566232416e-05, "loss": 1.8258, "step": 6530 }, { "epoch": 1.97, "learning_rate": 5.609993422037101e-05, "loss": 1.7295, "step": 6531 }, { "epoch": 1.97, "learning_rate": 5.607073741482407e-05, "loss": 1.948, "step": 6532 }, { "epoch": 1.97, "learning_rate": 5.604154524876677e-05, "loss": 1.8089, "step": 6533 }, { "epoch": 1.97, "learning_rate": 5.60123577252823e-05, "loss": 1.8903, "step": 6534 }, { "epoch": 1.97, "learning_rate": 5.598317484745326e-05, "loss": 1.8365, "step": 6535 }, { "epoch": 1.97, "learning_rate": 5.5953996618361606e-05, "loss": 1.8419, "step": 6536 }, { "epoch": 1.97, "learning_rate": 5.592482304108903e-05, "loss": 1.7222, "step": 6537 }, { "epoch": 1.97, "learning_rate": 5.5895654118716535e-05, "loss": 1.8063, "step": 6538 }, { "epoch": 1.97, "learning_rate": 5.586648985432488e-05, "loss": 1.8449, "step": 6539 }, { "epoch": 1.97, "learning_rate": 5.583733025099403e-05, "loss": 1.813, "step": 6540 }, { "epoch": 1.97, "learning_rate": 5.58081753118037e-05, "loss": 1.9011, "step": 6541 }, { "epoch": 1.97, "learning_rate": 5.577902503983303e-05, "loss": 1.7688, "step": 6542 }, { "epoch": 1.97, "learning_rate": 5.574987943816056e-05, "loss": 1.7914, "step": 6543 }, { "epoch": 1.97, "learning_rate": 5.5720738509864565e-05, "loss": 1.8842, "step": 6544 }, { "epoch": 1.97, "learning_rate": 5.569160225802263e-05, "loss": 1.7494, "step": 6545 }, { "epoch": 1.97, "learning_rate": 5.5662470685711906e-05, "loss": 1.766, "step": 6546 }, { "epoch": 1.97, "learning_rate": 5.5633343796009084e-05, "loss": 1.7171, "step": 6547 }, { "epoch": 1.97, "learning_rate": 5.560422159199033e-05, "loss": 1.857, "step": 6548 }, { "epoch": 1.97, "learning_rate": 5.5575104076731346e-05, "loss": 1.9882, "step": 6549 }, { "epoch": 1.97, "learning_rate": 5.554599125330724e-05, "loss": 1.7472, "step": 6550 }, { "epoch": 1.97, "learning_rate": 5.55168831247928e-05, "loss": 1.7712, "step": 6551 }, { "epoch": 1.97, "learning_rate": 5.5487779694262055e-05, "loss": 1.7829, "step": 6552 }, { "epoch": 1.97, "learning_rate": 5.545868096478885e-05, "loss": 1.7141, "step": 6553 }, { "epoch": 1.97, "learning_rate": 5.5429586939446367e-05, "loss": 1.883, "step": 6554 }, { "epoch": 1.97, "learning_rate": 5.5400497621307236e-05, "loss": 1.7001, "step": 6555 }, { "epoch": 1.97, "learning_rate": 5.537141301344372e-05, "loss": 1.8974, "step": 6556 }, { "epoch": 1.97, "learning_rate": 5.534233311892743e-05, "loss": 1.8487, "step": 6557 }, { "epoch": 1.97, "learning_rate": 5.531325794082966e-05, "loss": 1.8766, "step": 6558 }, { "epoch": 1.97, "learning_rate": 5.528418748222116e-05, "loss": 1.7586, "step": 6559 }, { "epoch": 1.97, "learning_rate": 5.5255121746172035e-05, "loss": 1.8414, "step": 6560 }, { "epoch": 1.98, "learning_rate": 5.5226060735752036e-05, "loss": 1.9093, "step": 6561 }, { "epoch": 1.98, "learning_rate": 5.5197004454030386e-05, "loss": 1.8297, "step": 6562 }, { "epoch": 1.98, "learning_rate": 5.516795290407584e-05, "loss": 1.761, "step": 6563 }, { "epoch": 1.98, "learning_rate": 5.513890608895651e-05, "loss": 1.8139, "step": 6564 }, { "epoch": 1.98, "learning_rate": 5.510986401174019e-05, "loss": 1.8374, "step": 6565 }, { "epoch": 1.98, "learning_rate": 5.508082667549404e-05, "loss": 1.8848, "step": 6566 }, { "epoch": 1.98, "learning_rate": 5.5051794083284823e-05, "loss": 1.9098, "step": 6567 }, { "epoch": 1.98, "learning_rate": 5.502276623817876e-05, "loss": 1.8522, "step": 6568 }, { "epoch": 1.98, "learning_rate": 5.4993743143241483e-05, "loss": 1.9225, "step": 6569 }, { "epoch": 1.98, "learning_rate": 5.496472480153827e-05, "loss": 1.7798, "step": 6570 }, { "epoch": 1.98, "learning_rate": 5.493571121613379e-05, "loss": 1.776, "step": 6571 }, { "epoch": 1.98, "learning_rate": 5.4906702390092255e-05, "loss": 1.8077, "step": 6572 }, { "epoch": 1.98, "learning_rate": 5.4877698326477424e-05, "loss": 1.852, "step": 6573 }, { "epoch": 1.98, "learning_rate": 5.4848699028352415e-05, "loss": 1.7552, "step": 6574 }, { "epoch": 1.98, "learning_rate": 5.4819704498779936e-05, "loss": 1.7898, "step": 6575 }, { "epoch": 1.98, "learning_rate": 5.4790714740822214e-05, "loss": 1.6843, "step": 6576 }, { "epoch": 1.98, "learning_rate": 5.47617297575409e-05, "loss": 1.7585, "step": 6577 }, { "epoch": 1.98, "learning_rate": 5.4732749551997256e-05, "loss": 1.8223, "step": 6578 }, { "epoch": 1.98, "learning_rate": 5.470377412725185e-05, "loss": 1.8378, "step": 6579 }, { "epoch": 1.98, "learning_rate": 5.467480348636491e-05, "loss": 1.9528, "step": 6580 }, { "epoch": 1.98, "learning_rate": 5.464583763239611e-05, "loss": 1.866, "step": 6581 }, { "epoch": 1.98, "learning_rate": 5.461687656840459e-05, "loss": 1.8568, "step": 6582 }, { "epoch": 1.98, "learning_rate": 5.458792029744908e-05, "loss": 1.7352, "step": 6583 }, { "epoch": 1.98, "learning_rate": 5.4558968822587634e-05, "loss": 1.7076, "step": 6584 }, { "epoch": 1.98, "learning_rate": 5.453002214687793e-05, "loss": 1.8134, "step": 6585 }, { "epoch": 1.98, "learning_rate": 5.45010802733771e-05, "loss": 1.7943, "step": 6586 }, { "epoch": 1.98, "learning_rate": 5.4472143205141846e-05, "loss": 1.8937, "step": 6587 }, { "epoch": 1.98, "learning_rate": 5.44432109452282e-05, "loss": 1.938, "step": 6588 }, { "epoch": 1.98, "learning_rate": 5.441428349669178e-05, "loss": 1.6921, "step": 6589 }, { "epoch": 1.98, "learning_rate": 5.4385360862587745e-05, "loss": 1.7277, "step": 6590 }, { "epoch": 1.98, "learning_rate": 5.435644304597065e-05, "loss": 1.7623, "step": 6591 }, { "epoch": 1.98, "learning_rate": 5.432753004989466e-05, "loss": 1.7618, "step": 6592 }, { "epoch": 1.98, "learning_rate": 5.429862187741326e-05, "loss": 1.9011, "step": 6593 }, { "epoch": 1.98, "learning_rate": 5.4269718531579564e-05, "loss": 1.894, "step": 6594 }, { "epoch": 1.99, "learning_rate": 5.4240820015446126e-05, "loss": 1.8324, "step": 6595 }, { "epoch": 1.99, "learning_rate": 5.4211926332065e-05, "loss": 1.8087, "step": 6596 }, { "epoch": 1.99, "learning_rate": 5.418303748448779e-05, "loss": 1.8374, "step": 6597 }, { "epoch": 1.99, "learning_rate": 5.41541534757654e-05, "loss": 1.7887, "step": 6598 }, { "epoch": 1.99, "learning_rate": 5.412527430894843e-05, "loss": 1.9257, "step": 6599 }, { "epoch": 1.99, "learning_rate": 5.4096399987086874e-05, "loss": 1.8343, "step": 6600 }, { "epoch": 1.99, "learning_rate": 5.406753051323024e-05, "loss": 1.6775, "step": 6601 }, { "epoch": 1.99, "learning_rate": 5.403866589042752e-05, "loss": 1.8265, "step": 6602 }, { "epoch": 1.99, "learning_rate": 5.400980612172713e-05, "loss": 1.8666, "step": 6603 }, { "epoch": 1.99, "learning_rate": 5.398095121017708e-05, "loss": 1.7791, "step": 6604 }, { "epoch": 1.99, "learning_rate": 5.395210115882479e-05, "loss": 1.8012, "step": 6605 }, { "epoch": 1.99, "learning_rate": 5.392325597071719e-05, "loss": 1.8397, "step": 6606 }, { "epoch": 1.99, "learning_rate": 5.389441564890078e-05, "loss": 1.8398, "step": 6607 }, { "epoch": 1.99, "learning_rate": 5.3865580196421337e-05, "loss": 1.7821, "step": 6608 }, { "epoch": 1.99, "learning_rate": 5.3836749616324333e-05, "loss": 1.7747, "step": 6609 }, { "epoch": 1.99, "learning_rate": 5.380792391165461e-05, "loss": 1.8221, "step": 6610 }, { "epoch": 1.99, "learning_rate": 5.377910308545656e-05, "loss": 1.9177, "step": 6611 }, { "epoch": 1.99, "learning_rate": 5.375028714077406e-05, "loss": 1.8717, "step": 6612 }, { "epoch": 1.99, "learning_rate": 5.3721476080650335e-05, "loss": 1.8551, "step": 6613 }, { "epoch": 1.99, "learning_rate": 5.3692669908128265e-05, "loss": 1.8847, "step": 6614 }, { "epoch": 1.99, "learning_rate": 5.366386862625015e-05, "loss": 2.0077, "step": 6615 }, { "epoch": 1.99, "learning_rate": 5.363507223805782e-05, "loss": 1.8469, "step": 6616 }, { "epoch": 1.99, "learning_rate": 5.360628074659242e-05, "loss": 1.9431, "step": 6617 }, { "epoch": 1.99, "learning_rate": 5.357749415489478e-05, "loss": 1.749, "step": 6618 }, { "epoch": 1.99, "learning_rate": 5.354871246600513e-05, "loss": 1.7817, "step": 6619 }, { "epoch": 1.99, "learning_rate": 5.351993568296314e-05, "loss": 1.9309, "step": 6620 }, { "epoch": 1.99, "learning_rate": 5.3491163808808084e-05, "loss": 1.7857, "step": 6621 }, { "epoch": 1.99, "learning_rate": 5.346239684657855e-05, "loss": 1.839, "step": 6622 }, { "epoch": 1.99, "learning_rate": 5.343363479931274e-05, "loss": 1.9497, "step": 6623 }, { "epoch": 1.99, "learning_rate": 5.340487767004829e-05, "loss": 1.8302, "step": 6624 }, { "epoch": 1.99, "learning_rate": 5.337612546182231e-05, "loss": 1.9181, "step": 6625 }, { "epoch": 1.99, "learning_rate": 5.334737817767144e-05, "loss": 1.8134, "step": 6626 }, { "epoch": 1.99, "learning_rate": 5.331863582063169e-05, "loss": 1.7455, "step": 6627 }, { "epoch": 2.0, "learning_rate": 5.328989839373866e-05, "loss": 1.8117, "step": 6628 }, { "epoch": 2.0, "learning_rate": 5.326116590002736e-05, "loss": 1.7742, "step": 6629 }, { "epoch": 2.0, "learning_rate": 5.323243834253235e-05, "loss": 1.7374, "step": 6630 }, { "epoch": 2.0, "learning_rate": 5.320371572428764e-05, "loss": 1.8832, "step": 6631 }, { "epoch": 2.0, "learning_rate": 5.317499804832663e-05, "loss": 1.957, "step": 6632 }, { "epoch": 2.0, "learning_rate": 5.314628531768231e-05, "loss": 1.7764, "step": 6633 }, { "epoch": 2.0, "learning_rate": 5.311757753538711e-05, "loss": 1.8218, "step": 6634 }, { "epoch": 2.0, "learning_rate": 5.3088874704472947e-05, "loss": 1.8837, "step": 6635 }, { "epoch": 2.0, "learning_rate": 5.306017682797124e-05, "loss": 1.8035, "step": 6636 }, { "epoch": 2.0, "learning_rate": 5.303148390891276e-05, "loss": 1.8758, "step": 6637 }, { "epoch": 2.0, "learning_rate": 5.30027959503279e-05, "loss": 1.8115, "step": 6638 }, { "epoch": 2.0, "learning_rate": 5.297411295524646e-05, "loss": 1.7133, "step": 6639 }, { "epoch": 2.0, "learning_rate": 5.294543492669779e-05, "loss": 1.9404, "step": 6640 }, { "epoch": 2.0, "learning_rate": 5.2916761867710554e-05, "loss": 1.7543, "step": 6641 }, { "epoch": 2.0, "learning_rate": 5.288809378131303e-05, "loss": 1.85, "step": 6642 }, { "epoch": 2.0, "learning_rate": 5.285943067053296e-05, "loss": 1.8542, "step": 6643 }, { "epoch": 2.0, "learning_rate": 5.283077253839751e-05, "loss": 1.8984, "step": 6644 }, { "epoch": 2.0, "learning_rate": 5.280211938793339e-05, "loss": 1.8029, "step": 6645 }, { "epoch": 2.0, "learning_rate": 5.277347122216667e-05, "loss": 1.8128, "step": 6646 }, { "epoch": 2.0, "learning_rate": 5.274482804412293e-05, "loss": 1.7486, "step": 6647 }, { "epoch": 2.0, "learning_rate": 5.271618985682741e-05, "loss": 1.7859, "step": 6648 }, { "epoch": 2.0, "learning_rate": 5.2687556663304535e-05, "loss": 1.7281, "step": 6649 }, { "epoch": 2.0, "learning_rate": 5.2658928466578404e-05, "loss": 1.8021, "step": 6650 }, { "epoch": 2.0, "learning_rate": 5.2630305269672444e-05, "loss": 1.866, "step": 6651 }, { "epoch": 2.0, "learning_rate": 5.260168707560967e-05, "loss": 1.7657, "step": 6652 }, { "epoch": 2.0, "learning_rate": 5.257307388741254e-05, "loss": 1.7282, "step": 6653 }, { "epoch": 2.0, "learning_rate": 5.2544465708102965e-05, "loss": 1.8158, "step": 6654 }, { "epoch": 2.0, "learning_rate": 5.2515862540702356e-05, "loss": 1.7998, "step": 6655 }, { "epoch": 2.0, "learning_rate": 5.24872643882315e-05, "loss": 1.8869, "step": 6656 }, { "epoch": 2.0, "learning_rate": 5.245867125371079e-05, "loss": 1.807, "step": 6657 }, { "epoch": 2.0, "learning_rate": 5.2430083140160004e-05, "loss": 1.8213, "step": 6658 }, { "epoch": 2.0, "learning_rate": 5.2401500050598404e-05, "loss": 1.6913, "step": 6659 }, { "epoch": 2.0, "learning_rate": 5.2372921988044775e-05, "loss": 1.9058, "step": 6660 }, { "epoch": 2.01, "learning_rate": 5.2344348955517265e-05, "loss": 1.8731, "step": 6661 }, { "epoch": 2.01, "learning_rate": 5.231578095603357e-05, "loss": 1.9005, "step": 6662 }, { "epoch": 2.01, "learning_rate": 5.228721799261083e-05, "loss": 1.9915, "step": 6663 }, { "epoch": 2.01, "learning_rate": 5.22586600682657e-05, "loss": 1.8873, "step": 6664 }, { "epoch": 2.01, "learning_rate": 5.22301071860142e-05, "loss": 1.8878, "step": 6665 }, { "epoch": 2.01, "learning_rate": 5.220155934887185e-05, "loss": 1.9336, "step": 6666 }, { "epoch": 2.01, "learning_rate": 5.2173016559853806e-05, "loss": 1.8278, "step": 6667 }, { "epoch": 2.01, "learning_rate": 5.214447882197442e-05, "loss": 1.8353, "step": 6668 }, { "epoch": 2.01, "learning_rate": 5.211594613824773e-05, "loss": 1.7675, "step": 6669 }, { "epoch": 2.01, "learning_rate": 5.2087418511687055e-05, "loss": 1.7721, "step": 6670 }, { "epoch": 2.01, "learning_rate": 5.20588959453053e-05, "loss": 1.7348, "step": 6671 }, { "epoch": 2.01, "learning_rate": 5.20303784421149e-05, "loss": 1.8082, "step": 6672 }, { "epoch": 2.01, "learning_rate": 5.200186600512757e-05, "loss": 1.6507, "step": 6673 }, { "epoch": 2.01, "learning_rate": 5.197335863735466e-05, "loss": 1.8131, "step": 6674 }, { "epoch": 2.01, "learning_rate": 5.1944856341806835e-05, "loss": 1.8989, "step": 6675 }, { "epoch": 2.01, "learning_rate": 5.191635912149434e-05, "loss": 1.7813, "step": 6676 }, { "epoch": 2.01, "learning_rate": 5.1887866979426835e-05, "loss": 1.854, "step": 6677 }, { "epoch": 2.01, "learning_rate": 5.185937991861346e-05, "loss": 1.7324, "step": 6678 }, { "epoch": 2.01, "learning_rate": 5.183089794206285e-05, "loss": 1.7492, "step": 6679 }, { "epoch": 2.01, "learning_rate": 5.180242105278295e-05, "loss": 1.8772, "step": 6680 }, { "epoch": 2.01, "learning_rate": 5.177394925378143e-05, "loss": 2.001, "step": 6681 }, { "epoch": 2.01, "learning_rate": 5.174548254806517e-05, "loss": 2.0092, "step": 6682 }, { "epoch": 2.01, "learning_rate": 5.1717020938640646e-05, "loss": 1.8625, "step": 6683 }, { "epoch": 2.01, "learning_rate": 5.168856442851382e-05, "loss": 1.8035, "step": 6684 }, { "epoch": 2.01, "learning_rate": 5.166011302068992e-05, "loss": 1.7256, "step": 6685 }, { "epoch": 2.01, "learning_rate": 5.163166671817394e-05, "loss": 1.8458, "step": 6686 }, { "epoch": 2.01, "learning_rate": 5.160322552397008e-05, "loss": 1.9238, "step": 6687 }, { "epoch": 2.01, "learning_rate": 5.157478944108214e-05, "loss": 1.7625, "step": 6688 }, { "epoch": 2.01, "learning_rate": 5.154635847251327e-05, "loss": 1.7799, "step": 6689 }, { "epoch": 2.01, "learning_rate": 5.151793262126614e-05, "loss": 1.877, "step": 6690 }, { "epoch": 2.01, "learning_rate": 5.148951189034299e-05, "loss": 1.9919, "step": 6691 }, { "epoch": 2.01, "learning_rate": 5.1461096282745314e-05, "loss": 1.9469, "step": 6692 }, { "epoch": 2.01, "learning_rate": 5.143268580147422e-05, "loss": 1.7158, "step": 6693 }, { "epoch": 2.02, "learning_rate": 5.140428044953012e-05, "loss": 1.749, "step": 6694 }, { "epoch": 2.02, "learning_rate": 5.137588022991306e-05, "loss": 1.8703, "step": 6695 }, { "epoch": 2.02, "learning_rate": 5.134748514562252e-05, "loss": 1.8287, "step": 6696 }, { "epoch": 2.02, "learning_rate": 5.1319095199657274e-05, "loss": 1.8452, "step": 6697 }, { "epoch": 2.02, "learning_rate": 5.129071039501572e-05, "loss": 1.7608, "step": 6698 }, { "epoch": 2.02, "learning_rate": 5.1262330734695576e-05, "loss": 1.7601, "step": 6699 }, { "epoch": 2.02, "learning_rate": 5.1233956221694236e-05, "loss": 1.8327, "step": 6700 }, { "epoch": 2.02, "learning_rate": 5.120558685900829e-05, "loss": 1.8175, "step": 6701 }, { "epoch": 2.02, "learning_rate": 5.117722264963394e-05, "loss": 1.8545, "step": 6702 }, { "epoch": 2.02, "learning_rate": 5.1148863596566856e-05, "loss": 1.7405, "step": 6703 }, { "epoch": 2.02, "learning_rate": 5.1120509702802e-05, "loss": 1.8385, "step": 6704 }, { "epoch": 2.02, "learning_rate": 5.1092160971334045e-05, "loss": 1.7395, "step": 6705 }, { "epoch": 2.02, "learning_rate": 5.106381740515687e-05, "loss": 1.8873, "step": 6706 }, { "epoch": 2.02, "learning_rate": 5.1035479007263964e-05, "loss": 1.8493, "step": 6707 }, { "epoch": 2.02, "learning_rate": 5.1007145780648215e-05, "loss": 1.7472, "step": 6708 }, { "epoch": 2.02, "learning_rate": 5.0978817728301975e-05, "loss": 1.857, "step": 6709 }, { "epoch": 2.02, "learning_rate": 5.095049485321708e-05, "loss": 1.8202, "step": 6710 }, { "epoch": 2.02, "learning_rate": 5.09221771583847e-05, "loss": 1.8011, "step": 6711 }, { "epoch": 2.02, "learning_rate": 5.089386464679562e-05, "loss": 1.7233, "step": 6712 }, { "epoch": 2.02, "learning_rate": 5.086555732143996e-05, "loss": 1.9557, "step": 6713 }, { "epoch": 2.02, "learning_rate": 5.083725518530736e-05, "loss": 1.8449, "step": 6714 }, { "epoch": 2.02, "learning_rate": 5.080895824138692e-05, "loss": 1.8862, "step": 6715 }, { "epoch": 2.02, "learning_rate": 5.078066649266707e-05, "loss": 1.8213, "step": 6716 }, { "epoch": 2.02, "learning_rate": 5.075237994213586e-05, "loss": 1.7823, "step": 6717 }, { "epoch": 2.02, "learning_rate": 5.0724098592780625e-05, "loss": 1.838, "step": 6718 }, { "epoch": 2.02, "learning_rate": 5.069582244758829e-05, "loss": 1.8014, "step": 6719 }, { "epoch": 2.02, "learning_rate": 5.066755150954523e-05, "loss": 1.8015, "step": 6720 }, { "epoch": 2.02, "learning_rate": 5.063928578163711e-05, "loss": 1.7859, "step": 6721 }, { "epoch": 2.02, "learning_rate": 5.061102526684921e-05, "loss": 1.8174, "step": 6722 }, { "epoch": 2.02, "learning_rate": 5.0582769968166175e-05, "loss": 1.788, "step": 6723 }, { "epoch": 2.02, "learning_rate": 5.055451988857216e-05, "loss": 1.76, "step": 6724 }, { "epoch": 2.02, "learning_rate": 5.0526275031050726e-05, "loss": 1.8414, "step": 6725 }, { "epoch": 2.02, "learning_rate": 5.049803539858485e-05, "loss": 1.7932, "step": 6726 }, { "epoch": 2.02, "learning_rate": 5.0469800994157037e-05, "loss": 1.7532, "step": 6727 }, { "epoch": 2.03, "learning_rate": 5.044157182074916e-05, "loss": 1.8196, "step": 6728 }, { "epoch": 2.03, "learning_rate": 5.0413347881342656e-05, "loss": 1.8467, "step": 6729 }, { "epoch": 2.03, "learning_rate": 5.038512917891825e-05, "loss": 1.661, "step": 6730 }, { "epoch": 2.03, "learning_rate": 5.035691571645621e-05, "loss": 1.7318, "step": 6731 }, { "epoch": 2.03, "learning_rate": 5.032870749693627e-05, "loss": 1.7625, "step": 6732 }, { "epoch": 2.03, "learning_rate": 5.0300504523337546e-05, "loss": 1.7735, "step": 6733 }, { "epoch": 2.03, "learning_rate": 5.02723067986387e-05, "loss": 1.8246, "step": 6734 }, { "epoch": 2.03, "learning_rate": 5.024411432581768e-05, "loss": 1.9259, "step": 6735 }, { "epoch": 2.03, "learning_rate": 5.0215927107852e-05, "loss": 1.889, "step": 6736 }, { "epoch": 2.03, "learning_rate": 5.01877451477186e-05, "loss": 1.7263, "step": 6737 }, { "epoch": 2.03, "learning_rate": 5.015956844839387e-05, "loss": 1.8004, "step": 6738 }, { "epoch": 2.03, "learning_rate": 5.0131397012853634e-05, "loss": 1.7233, "step": 6739 }, { "epoch": 2.03, "learning_rate": 5.010323084407309e-05, "loss": 1.8093, "step": 6740 }, { "epoch": 2.03, "learning_rate": 5.0075069945026975e-05, "loss": 1.9226, "step": 6741 }, { "epoch": 2.03, "learning_rate": 5.004691431868947e-05, "loss": 1.6436, "step": 6742 }, { "epoch": 2.03, "learning_rate": 5.001876396803413e-05, "loss": 1.8219, "step": 6743 }, { "epoch": 2.03, "learning_rate": 4.999061889603406e-05, "loss": 1.7583, "step": 6744 }, { "epoch": 2.03, "learning_rate": 4.996247910566164e-05, "loss": 1.7679, "step": 6745 }, { "epoch": 2.03, "learning_rate": 4.993434459988884e-05, "loss": 1.8618, "step": 6746 }, { "epoch": 2.03, "learning_rate": 4.990621538168701e-05, "loss": 1.7902, "step": 6747 }, { "epoch": 2.03, "learning_rate": 4.987809145402697e-05, "loss": 1.8233, "step": 6748 }, { "epoch": 2.03, "learning_rate": 4.9849972819878984e-05, "loss": 1.7908, "step": 6749 }, { "epoch": 2.03, "learning_rate": 4.9821859482212685e-05, "loss": 1.9378, "step": 6750 }, { "epoch": 2.03, "learning_rate": 4.979375144399722e-05, "loss": 1.8872, "step": 6751 }, { "epoch": 2.03, "learning_rate": 4.9765648708201165e-05, "loss": 1.8107, "step": 6752 }, { "epoch": 2.03, "learning_rate": 4.973755127779255e-05, "loss": 1.823, "step": 6753 }, { "epoch": 2.03, "learning_rate": 4.970945915573877e-05, "loss": 1.7137, "step": 6754 }, { "epoch": 2.03, "learning_rate": 4.968137234500673e-05, "loss": 1.7919, "step": 6755 }, { "epoch": 2.03, "learning_rate": 4.965329084856275e-05, "loss": 1.8447, "step": 6756 }, { "epoch": 2.03, "learning_rate": 4.962521466937261e-05, "loss": 1.8347, "step": 6757 }, { "epoch": 2.03, "learning_rate": 4.959714381040155e-05, "loss": 1.7186, "step": 6758 }, { "epoch": 2.03, "learning_rate": 4.9569078274614125e-05, "loss": 1.8042, "step": 6759 }, { "epoch": 2.03, "learning_rate": 4.9541018064974455e-05, "loss": 1.7467, "step": 6760 }, { "epoch": 2.04, "learning_rate": 4.9512963184446056e-05, "loss": 1.7367, "step": 6761 }, { "epoch": 2.04, "learning_rate": 4.948491363599188e-05, "loss": 1.832, "step": 6762 }, { "epoch": 2.04, "learning_rate": 4.945686942257436e-05, "loss": 1.8397, "step": 6763 }, { "epoch": 2.04, "learning_rate": 4.9428830547155236e-05, "loss": 1.8535, "step": 6764 }, { "epoch": 2.04, "learning_rate": 4.940079701269581e-05, "loss": 1.8679, "step": 6765 }, { "epoch": 2.04, "learning_rate": 4.937276882215679e-05, "loss": 1.7415, "step": 6766 }, { "epoch": 2.04, "learning_rate": 4.934474597849831e-05, "loss": 1.8436, "step": 6767 }, { "epoch": 2.04, "learning_rate": 4.9316728484679955e-05, "loss": 1.8491, "step": 6768 }, { "epoch": 2.04, "learning_rate": 4.928871634366068e-05, "loss": 1.842, "step": 6769 }, { "epoch": 2.04, "learning_rate": 4.926070955839894e-05, "loss": 1.8701, "step": 6770 }, { "epoch": 2.04, "learning_rate": 4.9232708131852634e-05, "loss": 1.9296, "step": 6771 }, { "epoch": 2.04, "learning_rate": 4.920471206697904e-05, "loss": 1.7254, "step": 6772 }, { "epoch": 2.04, "learning_rate": 4.9176721366734967e-05, "loss": 1.8686, "step": 6773 }, { "epoch": 2.04, "learning_rate": 4.914873603407649e-05, "loss": 1.8722, "step": 6774 }, { "epoch": 2.04, "learning_rate": 4.912075607195926e-05, "loss": 1.8647, "step": 6775 }, { "epoch": 2.04, "learning_rate": 4.9092781483338325e-05, "loss": 1.8419, "step": 6776 }, { "epoch": 2.04, "learning_rate": 4.906481227116819e-05, "loss": 1.8864, "step": 6777 }, { "epoch": 2.04, "learning_rate": 4.903684843840268e-05, "loss": 1.7235, "step": 6778 }, { "epoch": 2.04, "learning_rate": 4.900888998799519e-05, "loss": 1.7389, "step": 6779 }, { "epoch": 2.04, "learning_rate": 4.8980936922898465e-05, "loss": 1.8053, "step": 6780 }, { "epoch": 2.04, "learning_rate": 4.895298924606472e-05, "loss": 1.793, "step": 6781 }, { "epoch": 2.04, "learning_rate": 4.8925046960445606e-05, "loss": 1.7794, "step": 6782 }, { "epoch": 2.04, "learning_rate": 4.8897110068992125e-05, "loss": 1.8496, "step": 6783 }, { "epoch": 2.04, "learning_rate": 4.8869178574654795e-05, "loss": 1.8419, "step": 6784 }, { "epoch": 2.04, "learning_rate": 4.884125248038356e-05, "loss": 1.8133, "step": 6785 }, { "epoch": 2.04, "learning_rate": 4.8813331789127736e-05, "loss": 1.7291, "step": 6786 }, { "epoch": 2.04, "learning_rate": 4.8785416503836176e-05, "loss": 1.8342, "step": 6787 }, { "epoch": 2.04, "learning_rate": 4.8757506627456995e-05, "loss": 1.8441, "step": 6788 }, { "epoch": 2.04, "learning_rate": 4.8729602162937874e-05, "loss": 1.7861, "step": 6789 }, { "epoch": 2.04, "learning_rate": 4.870170311322587e-05, "loss": 1.7505, "step": 6790 }, { "epoch": 2.04, "learning_rate": 4.8673809481267484e-05, "loss": 1.8297, "step": 6791 }, { "epoch": 2.04, "learning_rate": 4.86459212700087e-05, "loss": 1.7941, "step": 6792 }, { "epoch": 2.04, "learning_rate": 4.8618038482394755e-05, "loss": 1.8294, "step": 6793 }, { "epoch": 2.05, "learning_rate": 4.859016112137049e-05, "loss": 1.7763, "step": 6794 }, { "epoch": 2.05, "learning_rate": 4.856228918988009e-05, "loss": 1.8282, "step": 6795 }, { "epoch": 2.05, "learning_rate": 4.853442269086721e-05, "loss": 1.8626, "step": 6796 }, { "epoch": 2.05, "learning_rate": 4.8506561627274925e-05, "loss": 1.842, "step": 6797 }, { "epoch": 2.05, "learning_rate": 4.847870600204566e-05, "loss": 1.9552, "step": 6798 }, { "epoch": 2.05, "learning_rate": 4.8450855818121345e-05, "loss": 1.8567, "step": 6799 }, { "epoch": 2.05, "learning_rate": 4.8423011078443325e-05, "loss": 1.7735, "step": 6800 }, { "epoch": 2.05, "learning_rate": 4.8395171785952386e-05, "loss": 1.775, "step": 6801 }, { "epoch": 2.05, "learning_rate": 4.836733794358864e-05, "loss": 1.832, "step": 6802 }, { "epoch": 2.05, "learning_rate": 4.833950955429175e-05, "loss": 1.7382, "step": 6803 }, { "epoch": 2.05, "learning_rate": 4.8311686621000717e-05, "loss": 1.8319, "step": 6804 }, { "epoch": 2.05, "learning_rate": 4.8283869146654026e-05, "loss": 1.6695, "step": 6805 }, { "epoch": 2.05, "learning_rate": 4.825605713418958e-05, "loss": 1.8126, "step": 6806 }, { "epoch": 2.05, "learning_rate": 4.8228250586544624e-05, "loss": 1.8997, "step": 6807 }, { "epoch": 2.05, "learning_rate": 4.820044950665589e-05, "loss": 1.7734, "step": 6808 }, { "epoch": 2.05, "learning_rate": 4.817265389745954e-05, "loss": 1.7376, "step": 6809 }, { "epoch": 2.05, "learning_rate": 4.814486376189116e-05, "loss": 1.712, "step": 6810 }, { "epoch": 2.05, "learning_rate": 4.811707910288577e-05, "loss": 1.7964, "step": 6811 }, { "epoch": 2.05, "learning_rate": 4.80892999233777e-05, "loss": 1.7869, "step": 6812 }, { "epoch": 2.05, "learning_rate": 4.806152622630082e-05, "loss": 1.7338, "step": 6813 }, { "epoch": 2.05, "learning_rate": 4.803375801458841e-05, "loss": 1.8551, "step": 6814 }, { "epoch": 2.05, "learning_rate": 4.8005995291173126e-05, "loss": 1.8438, "step": 6815 }, { "epoch": 2.05, "learning_rate": 4.79782380589871e-05, "loss": 1.7667, "step": 6816 }, { "epoch": 2.05, "learning_rate": 4.795048632096179e-05, "loss": 1.8049, "step": 6817 }, { "epoch": 2.05, "learning_rate": 4.792274008002816e-05, "loss": 1.7991, "step": 6818 }, { "epoch": 2.05, "learning_rate": 4.7894999339116585e-05, "loss": 1.7843, "step": 6819 }, { "epoch": 2.05, "learning_rate": 4.786726410115682e-05, "loss": 1.7698, "step": 6820 }, { "epoch": 2.05, "learning_rate": 4.783953436907809e-05, "loss": 1.7626, "step": 6821 }, { "epoch": 2.05, "learning_rate": 4.781181014580895e-05, "loss": 1.7112, "step": 6822 }, { "epoch": 2.05, "learning_rate": 4.7784091434277476e-05, "loss": 1.7869, "step": 6823 }, { "epoch": 2.05, "learning_rate": 4.775637823741109e-05, "loss": 1.7525, "step": 6824 }, { "epoch": 2.05, "learning_rate": 4.772867055813669e-05, "loss": 1.7841, "step": 6825 }, { "epoch": 2.05, "learning_rate": 4.770096839938058e-05, "loss": 1.7972, "step": 6826 }, { "epoch": 2.06, "learning_rate": 4.767327176406834e-05, "loss": 1.7743, "step": 6827 }, { "epoch": 2.06, "learning_rate": 4.764558065512524e-05, "loss": 1.8545, "step": 6828 }, { "epoch": 2.06, "learning_rate": 4.761789507547573e-05, "loss": 1.858, "step": 6829 }, { "epoch": 2.06, "learning_rate": 4.75902150280438e-05, "loss": 1.8247, "step": 6830 }, { "epoch": 2.06, "learning_rate": 4.7562540515752765e-05, "loss": 1.846, "step": 6831 }, { "epoch": 2.06, "learning_rate": 4.753487154152538e-05, "loss": 1.8848, "step": 6832 }, { "epoch": 2.06, "learning_rate": 4.7507208108283974e-05, "loss": 1.6942, "step": 6833 }, { "epoch": 2.06, "learning_rate": 4.747955021895004e-05, "loss": 1.7975, "step": 6834 }, { "epoch": 2.06, "learning_rate": 4.7451897876444686e-05, "loss": 1.7985, "step": 6835 }, { "epoch": 2.06, "learning_rate": 4.742425108368826e-05, "loss": 1.7592, "step": 6836 }, { "epoch": 2.06, "learning_rate": 4.739660984360064e-05, "loss": 1.8186, "step": 6837 }, { "epoch": 2.06, "learning_rate": 4.7368974159101175e-05, "loss": 1.8076, "step": 6838 }, { "epoch": 2.06, "learning_rate": 4.7341344033108445e-05, "loss": 1.7971, "step": 6839 }, { "epoch": 2.06, "learning_rate": 4.731371946854064e-05, "loss": 1.7459, "step": 6840 }, { "epoch": 2.06, "learning_rate": 4.728610046831512e-05, "loss": 1.8739, "step": 6841 }, { "epoch": 2.06, "learning_rate": 4.7258487035348975e-05, "loss": 1.7906, "step": 6842 }, { "epoch": 2.06, "learning_rate": 4.7230879172558404e-05, "loss": 1.9254, "step": 6843 }, { "epoch": 2.06, "learning_rate": 4.720327688285921e-05, "loss": 1.7809, "step": 6844 }, { "epoch": 2.06, "learning_rate": 4.717568016916657e-05, "loss": 1.8096, "step": 6845 }, { "epoch": 2.06, "learning_rate": 4.714808903439494e-05, "loss": 1.8947, "step": 6846 }, { "epoch": 2.06, "learning_rate": 4.7120503481458446e-05, "loss": 1.7063, "step": 6847 }, { "epoch": 2.06, "learning_rate": 4.709292351327035e-05, "loss": 1.8621, "step": 6848 }, { "epoch": 2.06, "learning_rate": 4.70653491327435e-05, "loss": 1.9406, "step": 6849 }, { "epoch": 2.06, "learning_rate": 4.703778034279014e-05, "loss": 1.7628, "step": 6850 }, { "epoch": 2.06, "learning_rate": 4.7010217146321756e-05, "loss": 1.763, "step": 6851 }, { "epoch": 2.06, "learning_rate": 4.698265954624954e-05, "loss": 1.7904, "step": 6852 }, { "epoch": 2.06, "learning_rate": 4.69551075454838e-05, "loss": 1.8049, "step": 6853 }, { "epoch": 2.06, "learning_rate": 4.692756114693446e-05, "loss": 1.8197, "step": 6854 }, { "epoch": 2.06, "learning_rate": 4.69000203535107e-05, "loss": 1.8328, "step": 6855 }, { "epoch": 2.06, "learning_rate": 4.6872485168121174e-05, "loss": 1.7779, "step": 6856 }, { "epoch": 2.06, "learning_rate": 4.6844955593674056e-05, "loss": 1.7752, "step": 6857 }, { "epoch": 2.06, "learning_rate": 4.6817431633076716e-05, "loss": 1.873, "step": 6858 }, { "epoch": 2.06, "learning_rate": 4.678991328923611e-05, "loss": 1.7622, "step": 6859 }, { "epoch": 2.07, "learning_rate": 4.6762400565058394e-05, "loss": 1.8178, "step": 6860 }, { "epoch": 2.07, "learning_rate": 4.67348934634494e-05, "loss": 1.8694, "step": 6861 }, { "epoch": 2.07, "learning_rate": 4.670739198731422e-05, "loss": 1.778, "step": 6862 }, { "epoch": 2.07, "learning_rate": 4.6679896139557286e-05, "loss": 1.9697, "step": 6863 }, { "epoch": 2.07, "learning_rate": 4.6652405923082575e-05, "loss": 1.8532, "step": 6864 }, { "epoch": 2.07, "learning_rate": 4.6624921340793304e-05, "loss": 1.7441, "step": 6865 }, { "epoch": 2.07, "learning_rate": 4.659744239559233e-05, "loss": 1.86, "step": 6866 }, { "epoch": 2.07, "learning_rate": 4.656996909038168e-05, "loss": 1.7625, "step": 6867 }, { "epoch": 2.07, "learning_rate": 4.6542501428062926e-05, "loss": 1.6994, "step": 6868 }, { "epoch": 2.07, "learning_rate": 4.6515039411537033e-05, "loss": 1.7616, "step": 6869 }, { "epoch": 2.07, "learning_rate": 4.648758304370422e-05, "loss": 1.7585, "step": 6870 }, { "epoch": 2.07, "learning_rate": 4.646013232746439e-05, "loss": 1.8258, "step": 6871 }, { "epoch": 2.07, "learning_rate": 4.643268726571658e-05, "loss": 1.8025, "step": 6872 }, { "epoch": 2.07, "learning_rate": 4.640524786135936e-05, "loss": 1.8326, "step": 6873 }, { "epoch": 2.07, "learning_rate": 4.63778141172907e-05, "loss": 1.7809, "step": 6874 }, { "epoch": 2.07, "learning_rate": 4.635038603640792e-05, "loss": 1.7743, "step": 6875 }, { "epoch": 2.07, "learning_rate": 4.632296362160785e-05, "loss": 1.7453, "step": 6876 }, { "epoch": 2.07, "learning_rate": 4.6295546875786555e-05, "loss": 1.9143, "step": 6877 }, { "epoch": 2.07, "learning_rate": 4.6268135801839675e-05, "loss": 1.8317, "step": 6878 }, { "epoch": 2.07, "learning_rate": 4.6240730402662036e-05, "loss": 1.8326, "step": 6879 }, { "epoch": 2.07, "learning_rate": 4.621333068114814e-05, "loss": 1.8495, "step": 6880 }, { "epoch": 2.07, "learning_rate": 4.618593664019171e-05, "loss": 1.8963, "step": 6881 }, { "epoch": 2.07, "learning_rate": 4.615854828268586e-05, "loss": 1.7838, "step": 6882 }, { "epoch": 2.07, "learning_rate": 4.6131165611523216e-05, "loss": 1.9041, "step": 6883 }, { "epoch": 2.07, "learning_rate": 4.610378862959562e-05, "loss": 1.9277, "step": 6884 }, { "epoch": 2.07, "learning_rate": 4.6076417339794556e-05, "loss": 1.8247, "step": 6885 }, { "epoch": 2.07, "learning_rate": 4.6049051745010755e-05, "loss": 1.817, "step": 6886 }, { "epoch": 2.07, "learning_rate": 4.6021691848134316e-05, "loss": 1.8138, "step": 6887 }, { "epoch": 2.07, "learning_rate": 4.599433765205484e-05, "loss": 1.7483, "step": 6888 }, { "epoch": 2.07, "learning_rate": 4.596698915966126e-05, "loss": 1.9233, "step": 6889 }, { "epoch": 2.07, "learning_rate": 4.593964637384196e-05, "loss": 1.8329, "step": 6890 }, { "epoch": 2.07, "learning_rate": 4.591230929748462e-05, "loss": 1.9287, "step": 6891 }, { "epoch": 2.07, "learning_rate": 4.5884977933476436e-05, "loss": 1.8277, "step": 6892 }, { "epoch": 2.07, "learning_rate": 4.585765228470392e-05, "loss": 1.8387, "step": 6893 }, { "epoch": 2.08, "learning_rate": 4.583033235405303e-05, "loss": 1.7829, "step": 6894 }, { "epoch": 2.08, "learning_rate": 4.580301814440913e-05, "loss": 1.7798, "step": 6895 }, { "epoch": 2.08, "learning_rate": 4.577570965865686e-05, "loss": 1.7996, "step": 6896 }, { "epoch": 2.08, "learning_rate": 4.5748406899680396e-05, "loss": 1.7802, "step": 6897 }, { "epoch": 2.08, "learning_rate": 4.572110987036325e-05, "loss": 1.7424, "step": 6898 }, { "epoch": 2.08, "learning_rate": 4.569381857358835e-05, "loss": 1.8864, "step": 6899 }, { "epoch": 2.08, "learning_rate": 4.566653301223803e-05, "loss": 1.8255, "step": 6900 }, { "epoch": 2.08, "learning_rate": 4.563925318919392e-05, "loss": 1.6819, "step": 6901 }, { "epoch": 2.08, "learning_rate": 4.561197910733715e-05, "loss": 1.8344, "step": 6902 }, { "epoch": 2.08, "learning_rate": 4.558471076954822e-05, "loss": 1.7903, "step": 6903 }, { "epoch": 2.08, "learning_rate": 4.555744817870701e-05, "loss": 1.7155, "step": 6904 }, { "epoch": 2.08, "learning_rate": 4.553019133769283e-05, "loss": 1.7873, "step": 6905 }, { "epoch": 2.08, "learning_rate": 4.550294024938427e-05, "loss": 1.7675, "step": 6906 }, { "epoch": 2.08, "learning_rate": 4.547569491665945e-05, "loss": 1.8247, "step": 6907 }, { "epoch": 2.08, "learning_rate": 4.544845534239581e-05, "loss": 1.7527, "step": 6908 }, { "epoch": 2.08, "learning_rate": 4.5421221529470194e-05, "loss": 1.7323, "step": 6909 }, { "epoch": 2.08, "learning_rate": 4.539399348075888e-05, "loss": 1.7488, "step": 6910 }, { "epoch": 2.08, "learning_rate": 4.536677119913743e-05, "loss": 1.7162, "step": 6911 }, { "epoch": 2.08, "learning_rate": 4.533955468748088e-05, "loss": 1.7659, "step": 6912 }, { "epoch": 2.08, "learning_rate": 4.531234394866366e-05, "loss": 1.8096, "step": 6913 }, { "epoch": 2.08, "learning_rate": 4.5285138985559604e-05, "loss": 1.8194, "step": 6914 }, { "epoch": 2.08, "learning_rate": 4.525793980104183e-05, "loss": 1.936, "step": 6915 }, { "epoch": 2.08, "learning_rate": 4.5230746397982936e-05, "loss": 1.8588, "step": 6916 }, { "epoch": 2.08, "learning_rate": 4.5203558779254916e-05, "loss": 1.8392, "step": 6917 }, { "epoch": 2.08, "learning_rate": 4.517637694772913e-05, "loss": 1.8393, "step": 6918 }, { "epoch": 2.08, "learning_rate": 4.514920090627635e-05, "loss": 1.8695, "step": 6919 }, { "epoch": 2.08, "learning_rate": 4.512203065776662e-05, "loss": 1.8515, "step": 6920 }, { "epoch": 2.08, "learning_rate": 4.5094866205069544e-05, "loss": 1.7593, "step": 6921 }, { "epoch": 2.08, "learning_rate": 4.506770755105401e-05, "loss": 1.8017, "step": 6922 }, { "epoch": 2.08, "learning_rate": 4.5040554698588314e-05, "loss": 1.839, "step": 6923 }, { "epoch": 2.08, "learning_rate": 4.5013407650540186e-05, "loss": 1.7335, "step": 6924 }, { "epoch": 2.08, "learning_rate": 4.498626640977664e-05, "loss": 1.7845, "step": 6925 }, { "epoch": 2.08, "learning_rate": 4.495913097916414e-05, "loss": 1.7454, "step": 6926 }, { "epoch": 2.09, "learning_rate": 4.493200136156856e-05, "loss": 1.794, "step": 6927 }, { "epoch": 2.09, "learning_rate": 4.490487755985514e-05, "loss": 1.8193, "step": 6928 }, { "epoch": 2.09, "learning_rate": 4.487775957688851e-05, "loss": 1.7203, "step": 6929 }, { "epoch": 2.09, "learning_rate": 4.485064741553262e-05, "loss": 1.753, "step": 6930 }, { "epoch": 2.09, "learning_rate": 4.4823541078650886e-05, "loss": 1.8201, "step": 6931 }, { "epoch": 2.09, "learning_rate": 4.479644056910609e-05, "loss": 1.7845, "step": 6932 }, { "epoch": 2.09, "learning_rate": 4.476934588976038e-05, "loss": 1.8185, "step": 6933 }, { "epoch": 2.09, "learning_rate": 4.474225704347536e-05, "loss": 1.7801, "step": 6934 }, { "epoch": 2.09, "learning_rate": 4.471517403311185e-05, "loss": 1.8791, "step": 6935 }, { "epoch": 2.09, "learning_rate": 4.468809686153021e-05, "loss": 1.7909, "step": 6936 }, { "epoch": 2.09, "learning_rate": 4.466102553159015e-05, "loss": 1.7798, "step": 6937 }, { "epoch": 2.09, "learning_rate": 4.4633960046150736e-05, "loss": 1.8192, "step": 6938 }, { "epoch": 2.09, "learning_rate": 4.460690040807046e-05, "loss": 1.8484, "step": 6939 }, { "epoch": 2.09, "learning_rate": 4.457984662020711e-05, "loss": 1.8527, "step": 6940 }, { "epoch": 2.09, "learning_rate": 4.4552798685417915e-05, "loss": 1.7841, "step": 6941 }, { "epoch": 2.09, "learning_rate": 4.452575660655952e-05, "loss": 1.8298, "step": 6942 }, { "epoch": 2.09, "learning_rate": 4.4498720386487916e-05, "loss": 1.8503, "step": 6943 }, { "epoch": 2.09, "learning_rate": 4.447169002805841e-05, "loss": 1.7862, "step": 6944 }, { "epoch": 2.09, "learning_rate": 4.444466553412579e-05, "loss": 1.823, "step": 6945 }, { "epoch": 2.09, "learning_rate": 4.441764690754421e-05, "loss": 1.7379, "step": 6946 }, { "epoch": 2.09, "learning_rate": 4.439063415116713e-05, "loss": 1.7636, "step": 6947 }, { "epoch": 2.09, "learning_rate": 4.436362726784753e-05, "loss": 1.8778, "step": 6948 }, { "epoch": 2.09, "learning_rate": 4.4336626260437575e-05, "loss": 1.7053, "step": 6949 }, { "epoch": 2.09, "learning_rate": 4.430963113178897e-05, "loss": 1.8051, "step": 6950 }, { "epoch": 2.09, "learning_rate": 4.428264188475272e-05, "loss": 1.7181, "step": 6951 }, { "epoch": 2.09, "learning_rate": 4.425565852217925e-05, "loss": 1.8704, "step": 6952 }, { "epoch": 2.09, "learning_rate": 4.4228681046918395e-05, "loss": 1.7332, "step": 6953 }, { "epoch": 2.09, "learning_rate": 4.420170946181923e-05, "loss": 1.8258, "step": 6954 }, { "epoch": 2.09, "learning_rate": 4.417474376973034e-05, "loss": 1.8185, "step": 6955 }, { "epoch": 2.09, "learning_rate": 4.4147783973499647e-05, "loss": 1.7884, "step": 6956 }, { "epoch": 2.09, "learning_rate": 4.412083007597445e-05, "loss": 1.7777, "step": 6957 }, { "epoch": 2.09, "learning_rate": 4.409388208000145e-05, "loss": 1.7418, "step": 6958 }, { "epoch": 2.09, "learning_rate": 4.4066939988426636e-05, "loss": 1.902, "step": 6959 }, { "epoch": 2.1, "learning_rate": 4.4040003804095465e-05, "loss": 2.1022, "step": 6960 }, { "epoch": 2.1, "learning_rate": 4.401307352985275e-05, "loss": 1.9759, "step": 6961 }, { "epoch": 2.1, "learning_rate": 4.3986149168542666e-05, "loss": 1.8438, "step": 6962 }, { "epoch": 2.1, "learning_rate": 4.3959230723008804e-05, "loss": 1.8353, "step": 6963 }, { "epoch": 2.1, "learning_rate": 4.3932318196094025e-05, "loss": 1.8237, "step": 6964 }, { "epoch": 2.1, "learning_rate": 4.390541159064068e-05, "loss": 1.8311, "step": 6965 }, { "epoch": 2.1, "learning_rate": 4.387851090949042e-05, "loss": 1.8221, "step": 6966 }, { "epoch": 2.1, "learning_rate": 4.3851616155484365e-05, "loss": 1.7576, "step": 6967 }, { "epoch": 2.1, "learning_rate": 4.382472733146286e-05, "loss": 1.7536, "step": 6968 }, { "epoch": 2.1, "learning_rate": 4.379784444026575e-05, "loss": 1.9111, "step": 6969 }, { "epoch": 2.1, "learning_rate": 4.3770967484732197e-05, "loss": 1.8376, "step": 6970 }, { "epoch": 2.1, "learning_rate": 4.374409646770076e-05, "loss": 1.9086, "step": 6971 }, { "epoch": 2.1, "learning_rate": 4.3717231392009396e-05, "loss": 1.8291, "step": 6972 }, { "epoch": 2.1, "learning_rate": 4.369037226049533e-05, "loss": 1.8798, "step": 6973 }, { "epoch": 2.1, "learning_rate": 4.3663519075995205e-05, "loss": 1.8605, "step": 6974 }, { "epoch": 2.1, "learning_rate": 4.363667184134521e-05, "loss": 1.823, "step": 6975 }, { "epoch": 2.1, "learning_rate": 4.360983055938062e-05, "loss": 1.8509, "step": 6976 }, { "epoch": 2.1, "learning_rate": 4.358299523293629e-05, "loss": 1.7894, "step": 6977 }, { "epoch": 2.1, "learning_rate": 4.3556165864846286e-05, "loss": 1.8587, "step": 6978 }, { "epoch": 2.1, "learning_rate": 4.35293424579442e-05, "loss": 1.9149, "step": 6979 }, { "epoch": 2.1, "learning_rate": 4.350252501506291e-05, "loss": 1.8257, "step": 6980 }, { "epoch": 2.1, "learning_rate": 4.347571353903468e-05, "loss": 1.821, "step": 6981 }, { "epoch": 2.1, "learning_rate": 4.3448908032691174e-05, "loss": 1.844, "step": 6982 }, { "epoch": 2.1, "learning_rate": 4.342210849886332e-05, "loss": 1.7544, "step": 6983 }, { "epoch": 2.1, "learning_rate": 4.339531494038154e-05, "loss": 1.7386, "step": 6984 }, { "epoch": 2.1, "learning_rate": 4.336852736007557e-05, "loss": 1.7525, "step": 6985 }, { "epoch": 2.1, "learning_rate": 4.334174576077451e-05, "loss": 1.7892, "step": 6986 }, { "epoch": 2.1, "learning_rate": 4.331497014530689e-05, "loss": 1.7841, "step": 6987 }, { "epoch": 2.1, "learning_rate": 4.3288200516500474e-05, "loss": 1.863, "step": 6988 }, { "epoch": 2.1, "learning_rate": 4.326143687718251e-05, "loss": 1.8352, "step": 6989 }, { "epoch": 2.1, "learning_rate": 4.323467923017959e-05, "loss": 1.871, "step": 6990 }, { "epoch": 2.1, "learning_rate": 4.3207927578317685e-05, "loss": 1.8723, "step": 6991 }, { "epoch": 2.1, "learning_rate": 4.318118192442205e-05, "loss": 1.7212, "step": 6992 }, { "epoch": 2.11, "learning_rate": 4.315444227131736e-05, "loss": 1.8407, "step": 6993 }, { "epoch": 2.11, "learning_rate": 4.312770862182777e-05, "loss": 1.7775, "step": 6994 }, { "epoch": 2.11, "learning_rate": 4.310098097877658e-05, "loss": 1.7958, "step": 6995 }, { "epoch": 2.11, "learning_rate": 4.307425934498667e-05, "loss": 1.904, "step": 6996 }, { "epoch": 2.11, "learning_rate": 4.304754372328008e-05, "loss": 1.7574, "step": 6997 }, { "epoch": 2.11, "learning_rate": 4.302083411647834e-05, "loss": 1.8637, "step": 6998 }, { "epoch": 2.11, "learning_rate": 4.299413052740241e-05, "loss": 1.9169, "step": 6999 }, { "epoch": 2.11, "learning_rate": 4.296743295887245e-05, "loss": 1.8213, "step": 7000 }, { "epoch": 2.11, "learning_rate": 4.2940741413708116e-05, "loss": 1.8159, "step": 7001 }, { "epoch": 2.11, "learning_rate": 4.291405589472832e-05, "loss": 1.8256, "step": 7002 }, { "epoch": 2.11, "learning_rate": 4.2887376404751414e-05, "loss": 1.8509, "step": 7003 }, { "epoch": 2.11, "learning_rate": 4.286070294659509e-05, "loss": 1.8708, "step": 7004 }, { "epoch": 2.11, "learning_rate": 4.283403552307641e-05, "loss": 1.8204, "step": 7005 }, { "epoch": 2.11, "learning_rate": 4.2807374137011836e-05, "loss": 1.8189, "step": 7006 }, { "epoch": 2.11, "learning_rate": 4.278071879121703e-05, "loss": 1.8424, "step": 7007 }, { "epoch": 2.11, "learning_rate": 4.2754069488507285e-05, "loss": 1.8754, "step": 7008 }, { "epoch": 2.11, "learning_rate": 4.272742623169701e-05, "loss": 1.8522, "step": 7009 }, { "epoch": 2.11, "learning_rate": 4.27007890236001e-05, "loss": 1.8045, "step": 7010 }, { "epoch": 2.11, "learning_rate": 4.2674157867029804e-05, "loss": 1.6972, "step": 7011 }, { "epoch": 2.11, "learning_rate": 4.2647532764798634e-05, "loss": 1.7636, "step": 7012 }, { "epoch": 2.11, "learning_rate": 4.2620913719718657e-05, "loss": 1.737, "step": 7013 }, { "epoch": 2.11, "learning_rate": 4.259430073460109e-05, "loss": 1.8744, "step": 7014 }, { "epoch": 2.11, "learning_rate": 4.256769381225667e-05, "loss": 1.7926, "step": 7015 }, { "epoch": 2.11, "learning_rate": 4.2541092955495363e-05, "loss": 1.8523, "step": 7016 }, { "epoch": 2.11, "learning_rate": 4.2514498167126535e-05, "loss": 1.7883, "step": 7017 }, { "epoch": 2.11, "learning_rate": 4.2487909449959075e-05, "loss": 1.7252, "step": 7018 }, { "epoch": 2.11, "learning_rate": 4.246132680680096e-05, "loss": 1.7514, "step": 7019 }, { "epoch": 2.11, "learning_rate": 4.243475024045973e-05, "loss": 1.7545, "step": 7020 }, { "epoch": 2.11, "learning_rate": 4.2408179753742115e-05, "loss": 1.735, "step": 7021 }, { "epoch": 2.11, "learning_rate": 4.238161534945438e-05, "loss": 1.8012, "step": 7022 }, { "epoch": 2.11, "learning_rate": 4.235505703040209e-05, "loss": 1.806, "step": 7023 }, { "epoch": 2.11, "learning_rate": 4.232850479939006e-05, "loss": 1.8807, "step": 7024 }, { "epoch": 2.11, "learning_rate": 4.230195865922261e-05, "loss": 1.7668, "step": 7025 }, { "epoch": 2.11, "learning_rate": 4.227541861270324e-05, "loss": 1.8195, "step": 7026 }, { "epoch": 2.12, "learning_rate": 4.224888466263508e-05, "loss": 1.7932, "step": 7027 }, { "epoch": 2.12, "learning_rate": 4.222235681182033e-05, "loss": 1.7861, "step": 7028 }, { "epoch": 2.12, "learning_rate": 4.2195835063060704e-05, "loss": 1.7483, "step": 7029 }, { "epoch": 2.12, "learning_rate": 4.2169319419157285e-05, "loss": 1.6964, "step": 7030 }, { "epoch": 2.12, "learning_rate": 4.214280988291034e-05, "loss": 1.8103, "step": 7031 }, { "epoch": 2.12, "learning_rate": 4.2116306457119767e-05, "loss": 1.8208, "step": 7032 }, { "epoch": 2.12, "learning_rate": 4.2089809144584556e-05, "loss": 1.7798, "step": 7033 }, { "epoch": 2.12, "learning_rate": 4.2063317948103196e-05, "loss": 1.8922, "step": 7034 }, { "epoch": 2.12, "learning_rate": 4.20368328704735e-05, "loss": 1.7501, "step": 7035 }, { "epoch": 2.12, "learning_rate": 4.201035391449261e-05, "loss": 1.8737, "step": 7036 }, { "epoch": 2.12, "learning_rate": 4.19838810829571e-05, "loss": 1.7769, "step": 7037 }, { "epoch": 2.12, "learning_rate": 4.195741437866275e-05, "loss": 1.8232, "step": 7038 }, { "epoch": 2.12, "learning_rate": 4.193095380440487e-05, "loss": 1.7261, "step": 7039 }, { "epoch": 2.12, "learning_rate": 4.190449936297791e-05, "loss": 1.8783, "step": 7040 }, { "epoch": 2.12, "learning_rate": 4.187805105717592e-05, "loss": 1.7799, "step": 7041 }, { "epoch": 2.12, "learning_rate": 4.1851608889792146e-05, "loss": 1.8159, "step": 7042 }, { "epoch": 2.12, "learning_rate": 4.182517286361919e-05, "loss": 1.7952, "step": 7043 }, { "epoch": 2.12, "learning_rate": 4.179874298144907e-05, "loss": 1.844, "step": 7044 }, { "epoch": 2.12, "learning_rate": 4.1772319246073024e-05, "loss": 1.8007, "step": 7045 }, { "epoch": 2.12, "learning_rate": 4.174590166028185e-05, "loss": 1.8641, "step": 7046 }, { "epoch": 2.12, "learning_rate": 4.171949022686556e-05, "loss": 1.863, "step": 7047 }, { "epoch": 2.12, "learning_rate": 4.1693084948613494e-05, "loss": 1.8692, "step": 7048 }, { "epoch": 2.12, "learning_rate": 4.16666858283144e-05, "loss": 1.8051, "step": 7049 }, { "epoch": 2.12, "learning_rate": 4.164029286875637e-05, "loss": 1.7792, "step": 7050 }, { "epoch": 2.12, "learning_rate": 4.1613906072726874e-05, "loss": 1.821, "step": 7051 }, { "epoch": 2.12, "learning_rate": 4.158752544301262e-05, "loss": 1.7141, "step": 7052 }, { "epoch": 2.12, "learning_rate": 4.156115098239978e-05, "loss": 1.8213, "step": 7053 }, { "epoch": 2.12, "learning_rate": 4.1534782693673815e-05, "loss": 1.7802, "step": 7054 }, { "epoch": 2.12, "learning_rate": 4.150842057961957e-05, "loss": 1.8233, "step": 7055 }, { "epoch": 2.12, "learning_rate": 4.1482064643021235e-05, "loss": 1.7494, "step": 7056 }, { "epoch": 2.12, "learning_rate": 4.145571488666229e-05, "loss": 1.9493, "step": 7057 }, { "epoch": 2.12, "learning_rate": 4.1429371313325626e-05, "loss": 1.8289, "step": 7058 }, { "epoch": 2.12, "learning_rate": 4.140303392579345e-05, "loss": 1.6628, "step": 7059 }, { "epoch": 2.13, "learning_rate": 4.137670272684736e-05, "loss": 1.7174, "step": 7060 }, { "epoch": 2.13, "learning_rate": 4.1350377719268266e-05, "loss": 1.8129, "step": 7061 }, { "epoch": 2.13, "learning_rate": 4.132405890583636e-05, "loss": 1.8194, "step": 7062 }, { "epoch": 2.13, "learning_rate": 4.1297746289331285e-05, "loss": 1.703, "step": 7063 }, { "epoch": 2.13, "learning_rate": 4.127143987253199e-05, "loss": 1.8223, "step": 7064 }, { "epoch": 2.13, "learning_rate": 4.1245139658216756e-05, "loss": 1.7732, "step": 7065 }, { "epoch": 2.13, "learning_rate": 4.121884564916328e-05, "loss": 1.816, "step": 7066 }, { "epoch": 2.13, "learning_rate": 4.119255784814845e-05, "loss": 1.6924, "step": 7067 }, { "epoch": 2.13, "learning_rate": 4.1166276257948645e-05, "loss": 1.8179, "step": 7068 }, { "epoch": 2.13, "learning_rate": 4.1140000881339526e-05, "loss": 1.8433, "step": 7069 }, { "epoch": 2.13, "learning_rate": 4.1113731721096106e-05, "loss": 1.7831, "step": 7070 }, { "epoch": 2.13, "learning_rate": 4.1087468779992785e-05, "loss": 1.8398, "step": 7071 }, { "epoch": 2.13, "learning_rate": 4.106121206080319e-05, "loss": 1.8008, "step": 7072 }, { "epoch": 2.13, "learning_rate": 4.1034961566300387e-05, "loss": 1.7035, "step": 7073 }, { "epoch": 2.13, "learning_rate": 4.1008717299256794e-05, "loss": 1.9062, "step": 7074 }, { "epoch": 2.13, "learning_rate": 4.098247926244412e-05, "loss": 1.8035, "step": 7075 }, { "epoch": 2.13, "learning_rate": 4.0956247458633465e-05, "loss": 1.8204, "step": 7076 }, { "epoch": 2.13, "learning_rate": 4.093002189059518e-05, "loss": 1.9261, "step": 7077 }, { "epoch": 2.13, "learning_rate": 4.090380256109906e-05, "loss": 1.8296, "step": 7078 }, { "epoch": 2.13, "learning_rate": 4.087758947291419e-05, "loss": 1.8152, "step": 7079 }, { "epoch": 2.13, "learning_rate": 4.085138262880904e-05, "loss": 1.7601, "step": 7080 }, { "epoch": 2.13, "learning_rate": 4.0825182031551334e-05, "loss": 1.7356, "step": 7081 }, { "epoch": 2.13, "learning_rate": 4.079898768390821e-05, "loss": 1.8604, "step": 7082 }, { "epoch": 2.13, "learning_rate": 4.077279958864612e-05, "loss": 1.9369, "step": 7083 }, { "epoch": 2.13, "learning_rate": 4.0746617748530856e-05, "loss": 1.7548, "step": 7084 }, { "epoch": 2.13, "learning_rate": 4.0720442166327624e-05, "loss": 1.8345, "step": 7085 }, { "epoch": 2.13, "learning_rate": 4.069427284480078e-05, "loss": 1.804, "step": 7086 }, { "epoch": 2.13, "learning_rate": 4.0668109786714216e-05, "loss": 1.8497, "step": 7087 }, { "epoch": 2.13, "learning_rate": 4.064195299483106e-05, "loss": 1.9431, "step": 7088 }, { "epoch": 2.13, "learning_rate": 4.06158024719138e-05, "loss": 1.9498, "step": 7089 }, { "epoch": 2.13, "learning_rate": 4.058965822072431e-05, "loss": 1.7399, "step": 7090 }, { "epoch": 2.13, "learning_rate": 4.056352024402368e-05, "loss": 1.8133, "step": 7091 }, { "epoch": 2.13, "learning_rate": 4.053738854457246e-05, "loss": 1.8466, "step": 7092 }, { "epoch": 2.14, "learning_rate": 4.0511263125130475e-05, "loss": 1.8666, "step": 7093 }, { "epoch": 2.14, "learning_rate": 4.048514398845691e-05, "loss": 1.8457, "step": 7094 }, { "epoch": 2.14, "learning_rate": 4.0459031137310314e-05, "loss": 1.8109, "step": 7095 }, { "epoch": 2.14, "learning_rate": 4.0432924574448464e-05, "loss": 1.7296, "step": 7096 }, { "epoch": 2.14, "learning_rate": 4.040682430262859e-05, "loss": 1.8227, "step": 7097 }, { "epoch": 2.14, "learning_rate": 4.038073032460718e-05, "loss": 1.8731, "step": 7098 }, { "epoch": 2.14, "learning_rate": 4.035464264314014e-05, "loss": 1.8568, "step": 7099 }, { "epoch": 2.14, "learning_rate": 4.032856126098267e-05, "loss": 1.9402, "step": 7100 }, { "epoch": 2.14, "learning_rate": 4.030248618088922e-05, "loss": 1.7827, "step": 7101 }, { "epoch": 2.14, "learning_rate": 4.0276417405613686e-05, "loss": 1.7726, "step": 7102 }, { "epoch": 2.14, "learning_rate": 4.025035493790929e-05, "loss": 1.8473, "step": 7103 }, { "epoch": 2.14, "learning_rate": 4.022429878052856e-05, "loss": 1.7332, "step": 7104 }, { "epoch": 2.14, "learning_rate": 4.0198248936223316e-05, "loss": 1.7156, "step": 7105 }, { "epoch": 2.14, "learning_rate": 4.017220540774477e-05, "loss": 1.8473, "step": 7106 }, { "epoch": 2.14, "learning_rate": 4.014616819784345e-05, "loss": 1.7408, "step": 7107 }, { "epoch": 2.14, "learning_rate": 4.0120137309269236e-05, "loss": 1.8336, "step": 7108 }, { "epoch": 2.14, "learning_rate": 4.009411274477133e-05, "loss": 1.7474, "step": 7109 }, { "epoch": 2.14, "learning_rate": 4.006809450709821e-05, "loss": 1.7802, "step": 7110 }, { "epoch": 2.14, "learning_rate": 4.0042082598997765e-05, "loss": 1.7586, "step": 7111 }, { "epoch": 2.14, "learning_rate": 4.001607702321717e-05, "loss": 1.9175, "step": 7112 }, { "epoch": 2.14, "learning_rate": 3.9990077782502954e-05, "loss": 1.7763, "step": 7113 }, { "epoch": 2.14, "learning_rate": 3.9964084879601014e-05, "loss": 1.7945, "step": 7114 }, { "epoch": 2.14, "learning_rate": 3.9938098317256445e-05, "loss": 1.6979, "step": 7115 }, { "epoch": 2.14, "learning_rate": 3.9912118098213805e-05, "loss": 1.8932, "step": 7116 }, { "epoch": 2.14, "learning_rate": 3.988614422521692e-05, "loss": 1.7982, "step": 7117 }, { "epoch": 2.14, "learning_rate": 3.9860176701008976e-05, "loss": 1.7973, "step": 7118 }, { "epoch": 2.14, "learning_rate": 3.983421552833251e-05, "loss": 1.7612, "step": 7119 }, { "epoch": 2.14, "learning_rate": 3.980826070992927e-05, "loss": 1.7886, "step": 7120 }, { "epoch": 2.14, "learning_rate": 3.9782312248540466e-05, "loss": 1.8297, "step": 7121 }, { "epoch": 2.14, "learning_rate": 3.975637014690657e-05, "loss": 1.8313, "step": 7122 }, { "epoch": 2.14, "learning_rate": 3.973043440776742e-05, "loss": 1.7835, "step": 7123 }, { "epoch": 2.14, "learning_rate": 3.970450503386218e-05, "loss": 1.8412, "step": 7124 }, { "epoch": 2.14, "learning_rate": 3.967858202792926e-05, "loss": 1.738, "step": 7125 }, { "epoch": 2.15, "learning_rate": 3.9652665392706475e-05, "loss": 1.801, "step": 7126 }, { "epoch": 2.15, "learning_rate": 3.962675513093098e-05, "loss": 1.7692, "step": 7127 }, { "epoch": 2.15, "learning_rate": 3.9600851245339246e-05, "loss": 1.7349, "step": 7128 }, { "epoch": 2.15, "learning_rate": 3.9574953738666985e-05, "loss": 1.8247, "step": 7129 }, { "epoch": 2.15, "learning_rate": 3.954906261364934e-05, "loss": 1.7895, "step": 7130 }, { "epoch": 2.15, "learning_rate": 3.952317787302076e-05, "loss": 1.7975, "step": 7131 }, { "epoch": 2.15, "learning_rate": 3.9497299519514976e-05, "loss": 1.8292, "step": 7132 }, { "epoch": 2.15, "learning_rate": 3.947142755586512e-05, "loss": 1.7569, "step": 7133 }, { "epoch": 2.15, "learning_rate": 3.944556198480353e-05, "loss": 1.8902, "step": 7134 }, { "epoch": 2.15, "learning_rate": 3.941970280906196e-05, "loss": 1.8277, "step": 7135 }, { "epoch": 2.15, "learning_rate": 3.9393850031371504e-05, "loss": 1.779, "step": 7136 }, { "epoch": 2.15, "learning_rate": 3.9368003654462515e-05, "loss": 1.8459, "step": 7137 }, { "epoch": 2.15, "learning_rate": 3.9342163681064745e-05, "loss": 1.6479, "step": 7138 }, { "epoch": 2.15, "learning_rate": 3.931633011390715e-05, "loss": 1.8244, "step": 7139 }, { "epoch": 2.15, "learning_rate": 3.929050295571812e-05, "loss": 1.7953, "step": 7140 }, { "epoch": 2.15, "learning_rate": 3.9264682209225345e-05, "loss": 1.8562, "step": 7141 }, { "epoch": 2.15, "learning_rate": 3.923886787715581e-05, "loss": 1.8344, "step": 7142 }, { "epoch": 2.15, "learning_rate": 3.9213059962235866e-05, "loss": 1.7875, "step": 7143 }, { "epoch": 2.15, "learning_rate": 3.91872584671911e-05, "loss": 1.9249, "step": 7144 }, { "epoch": 2.15, "learning_rate": 3.916146339474651e-05, "loss": 1.8078, "step": 7145 }, { "epoch": 2.15, "learning_rate": 3.9135674747626395e-05, "loss": 1.7564, "step": 7146 }, { "epoch": 2.15, "learning_rate": 3.910989252855434e-05, "loss": 1.7631, "step": 7147 }, { "epoch": 2.15, "learning_rate": 3.908411674025333e-05, "loss": 1.7937, "step": 7148 }, { "epoch": 2.15, "learning_rate": 3.905834738544555e-05, "loss": 1.8219, "step": 7149 }, { "epoch": 2.15, "learning_rate": 3.903258446685258e-05, "loss": 1.9178, "step": 7150 }, { "epoch": 2.15, "learning_rate": 3.9006827987195346e-05, "loss": 1.6938, "step": 7151 }, { "epoch": 2.15, "learning_rate": 3.898107794919409e-05, "loss": 1.847, "step": 7152 }, { "epoch": 2.15, "learning_rate": 3.895533435556825e-05, "loss": 1.6537, "step": 7153 }, { "epoch": 2.15, "learning_rate": 3.892959720903669e-05, "loss": 1.9524, "step": 7154 }, { "epoch": 2.15, "learning_rate": 3.89038665123177e-05, "loss": 1.8354, "step": 7155 }, { "epoch": 2.15, "learning_rate": 3.887814226812865e-05, "loss": 1.7229, "step": 7156 }, { "epoch": 2.15, "learning_rate": 3.885242447918641e-05, "loss": 1.7688, "step": 7157 }, { "epoch": 2.15, "learning_rate": 3.882671314820705e-05, "loss": 1.7788, "step": 7158 }, { "epoch": 2.16, "learning_rate": 3.880100827790601e-05, "loss": 1.9216, "step": 7159 }, { "epoch": 2.16, "learning_rate": 3.877530987099816e-05, "loss": 1.9133, "step": 7160 }, { "epoch": 2.16, "learning_rate": 3.874961793019747e-05, "loss": 1.8186, "step": 7161 }, { "epoch": 2.16, "learning_rate": 3.8723932458217406e-05, "loss": 1.982, "step": 7162 }, { "epoch": 2.16, "learning_rate": 3.8698253457770615e-05, "loss": 1.7996, "step": 7163 }, { "epoch": 2.16, "learning_rate": 3.867258093156916e-05, "loss": 1.8048, "step": 7164 }, { "epoch": 2.16, "learning_rate": 3.864691488232438e-05, "loss": 1.7686, "step": 7165 }, { "epoch": 2.16, "learning_rate": 3.862125531274694e-05, "loss": 1.8323, "step": 7166 }, { "epoch": 2.16, "learning_rate": 3.8595602225546855e-05, "loss": 1.8109, "step": 7167 }, { "epoch": 2.16, "learning_rate": 3.856995562343332e-05, "loss": 1.8764, "step": 7168 }, { "epoch": 2.16, "learning_rate": 3.854431550911507e-05, "loss": 1.8746, "step": 7169 }, { "epoch": 2.16, "learning_rate": 3.851868188529993e-05, "loss": 1.9433, "step": 7170 }, { "epoch": 2.16, "learning_rate": 3.849305475469518e-05, "loss": 1.9261, "step": 7171 }, { "epoch": 2.16, "learning_rate": 3.846743412000738e-05, "loss": 1.9106, "step": 7172 }, { "epoch": 2.16, "learning_rate": 3.844181998394232e-05, "loss": 1.9585, "step": 7173 }, { "epoch": 2.16, "learning_rate": 3.84162123492053e-05, "loss": 2.0072, "step": 7174 }, { "epoch": 2.16, "learning_rate": 3.8390611218500715e-05, "loss": 1.991, "step": 7175 }, { "epoch": 2.16, "learning_rate": 3.83650165945324e-05, "loss": 2.0409, "step": 7176 }, { "epoch": 2.16, "learning_rate": 3.833942848000351e-05, "loss": 1.9727, "step": 7177 }, { "epoch": 2.16, "learning_rate": 3.831384687761639e-05, "loss": 1.915, "step": 7178 }, { "epoch": 2.16, "learning_rate": 3.828827179007289e-05, "loss": 1.8622, "step": 7179 }, { "epoch": 2.16, "learning_rate": 3.826270322007399e-05, "loss": 1.7119, "step": 7180 }, { "epoch": 2.16, "learning_rate": 3.82371411703201e-05, "loss": 1.8549, "step": 7181 }, { "epoch": 2.16, "learning_rate": 3.8211585643510816e-05, "loss": 1.8906, "step": 7182 }, { "epoch": 2.16, "learning_rate": 3.818603664234522e-05, "loss": 1.7636, "step": 7183 }, { "epoch": 2.16, "learning_rate": 3.816049416952161e-05, "loss": 1.8126, "step": 7184 }, { "epoch": 2.16, "learning_rate": 3.813495822773754e-05, "loss": 1.7546, "step": 7185 }, { "epoch": 2.16, "learning_rate": 3.8109428819689986e-05, "loss": 1.6988, "step": 7186 }, { "epoch": 2.16, "learning_rate": 3.8083905948075084e-05, "loss": 1.8076, "step": 7187 }, { "epoch": 2.16, "learning_rate": 3.805838961558846e-05, "loss": 1.7823, "step": 7188 }, { "epoch": 2.16, "learning_rate": 3.8032879824924997e-05, "loss": 1.751, "step": 7189 }, { "epoch": 2.16, "learning_rate": 3.800737657877875e-05, "loss": 1.7976, "step": 7190 }, { "epoch": 2.16, "learning_rate": 3.798187987984328e-05, "loss": 1.9331, "step": 7191 }, { "epoch": 2.16, "learning_rate": 3.7956389730811256e-05, "loss": 1.7685, "step": 7192 }, { "epoch": 2.17, "learning_rate": 3.793090613437489e-05, "loss": 1.8107, "step": 7193 }, { "epoch": 2.17, "learning_rate": 3.7905429093225473e-05, "loss": 1.8509, "step": 7194 }, { "epoch": 2.17, "learning_rate": 3.7879958610053744e-05, "loss": 1.7391, "step": 7195 }, { "epoch": 2.17, "learning_rate": 3.785449468754975e-05, "loss": 1.8706, "step": 7196 }, { "epoch": 2.17, "learning_rate": 3.78290373284027e-05, "loss": 1.7263, "step": 7197 }, { "epoch": 2.17, "learning_rate": 3.7803586535301336e-05, "loss": 1.6734, "step": 7198 }, { "epoch": 2.17, "learning_rate": 3.777814231093351e-05, "loss": 1.9171, "step": 7199 }, { "epoch": 2.17, "learning_rate": 3.775270465798646e-05, "loss": 1.838, "step": 7200 }, { "epoch": 2.17, "learning_rate": 3.772727357914676e-05, "loss": 1.7386, "step": 7201 }, { "epoch": 2.17, "learning_rate": 3.7701849077100225e-05, "loss": 1.8588, "step": 7202 }, { "epoch": 2.17, "learning_rate": 3.767643115453206e-05, "loss": 1.8091, "step": 7203 }, { "epoch": 2.17, "learning_rate": 3.7651019814126654e-05, "loss": 1.689, "step": 7204 }, { "epoch": 2.17, "learning_rate": 3.762561505856783e-05, "loss": 1.7308, "step": 7205 }, { "epoch": 2.17, "learning_rate": 3.760021689053854e-05, "loss": 1.8593, "step": 7206 }, { "epoch": 2.17, "learning_rate": 3.757482531272127e-05, "loss": 1.8592, "step": 7207 }, { "epoch": 2.17, "learning_rate": 3.7549440327797705e-05, "loss": 1.8106, "step": 7208 }, { "epoch": 2.17, "learning_rate": 3.752406193844873e-05, "loss": 1.8179, "step": 7209 }, { "epoch": 2.17, "learning_rate": 3.749869014735471e-05, "loss": 1.8067, "step": 7210 }, { "epoch": 2.17, "learning_rate": 3.747332495719512e-05, "loss": 1.789, "step": 7211 }, { "epoch": 2.17, "learning_rate": 3.744796637064896e-05, "loss": 1.8563, "step": 7212 }, { "epoch": 2.17, "learning_rate": 3.7422614390394404e-05, "loss": 1.8557, "step": 7213 }, { "epoch": 2.17, "learning_rate": 3.7397269019108884e-05, "loss": 1.803, "step": 7214 }, { "epoch": 2.17, "learning_rate": 3.7371930259469234e-05, "loss": 1.7783, "step": 7215 }, { "epoch": 2.17, "learning_rate": 3.734659811415154e-05, "loss": 1.7994, "step": 7216 }, { "epoch": 2.17, "learning_rate": 3.732127258583124e-05, "loss": 1.7769, "step": 7217 }, { "epoch": 2.17, "learning_rate": 3.729595367718296e-05, "loss": 1.8019, "step": 7218 }, { "epoch": 2.17, "learning_rate": 3.7270641390880735e-05, "loss": 1.8218, "step": 7219 }, { "epoch": 2.17, "learning_rate": 3.724533572959786e-05, "loss": 1.8599, "step": 7220 }, { "epoch": 2.17, "learning_rate": 3.7220036696006944e-05, "loss": 1.7691, "step": 7221 }, { "epoch": 2.17, "learning_rate": 3.7194744292779925e-05, "loss": 1.7982, "step": 7222 }, { "epoch": 2.17, "learning_rate": 3.7169458522587916e-05, "loss": 1.8025, "step": 7223 }, { "epoch": 2.17, "learning_rate": 3.7144179388101476e-05, "loss": 1.767, "step": 7224 }, { "epoch": 2.17, "learning_rate": 3.711890689199037e-05, "loss": 1.8313, "step": 7225 }, { "epoch": 2.18, "learning_rate": 3.709364103692373e-05, "loss": 1.8075, "step": 7226 }, { "epoch": 2.18, "learning_rate": 3.7068381825569966e-05, "loss": 1.8537, "step": 7227 }, { "epoch": 2.18, "learning_rate": 3.704312926059671e-05, "loss": 1.7284, "step": 7228 }, { "epoch": 2.18, "learning_rate": 3.701788334467099e-05, "loss": 1.8589, "step": 7229 }, { "epoch": 2.18, "learning_rate": 3.699264408045908e-05, "loss": 1.8674, "step": 7230 }, { "epoch": 2.18, "learning_rate": 3.69674114706266e-05, "loss": 1.8334, "step": 7231 }, { "epoch": 2.18, "learning_rate": 3.694218551783844e-05, "loss": 1.7516, "step": 7232 }, { "epoch": 2.18, "learning_rate": 3.691696622475873e-05, "loss": 1.8579, "step": 7233 }, { "epoch": 2.18, "learning_rate": 3.689175359405098e-05, "loss": 1.8203, "step": 7234 }, { "epoch": 2.18, "learning_rate": 3.686654762837797e-05, "loss": 1.7309, "step": 7235 }, { "epoch": 2.18, "learning_rate": 3.684134833040175e-05, "loss": 1.8251, "step": 7236 }, { "epoch": 2.18, "learning_rate": 3.6816155702783736e-05, "loss": 1.8187, "step": 7237 }, { "epoch": 2.18, "learning_rate": 3.679096974818452e-05, "loss": 1.7168, "step": 7238 }, { "epoch": 2.18, "learning_rate": 3.6765790469264105e-05, "loss": 1.7475, "step": 7239 }, { "epoch": 2.18, "learning_rate": 3.6740617868681725e-05, "loss": 1.8171, "step": 7240 }, { "epoch": 2.18, "learning_rate": 3.671545194909597e-05, "loss": 1.7977, "step": 7241 }, { "epoch": 2.18, "learning_rate": 3.669029271316462e-05, "loss": 1.8055, "step": 7242 }, { "epoch": 2.18, "learning_rate": 3.666514016354483e-05, "loss": 1.7997, "step": 7243 }, { "epoch": 2.18, "learning_rate": 3.6639994302893044e-05, "loss": 1.8022, "step": 7244 }, { "epoch": 2.18, "learning_rate": 3.6614855133864976e-05, "loss": 1.7237, "step": 7245 }, { "epoch": 2.18, "learning_rate": 3.658972265911568e-05, "loss": 1.7137, "step": 7246 }, { "epoch": 2.18, "learning_rate": 3.6564596881299395e-05, "loss": 1.8847, "step": 7247 }, { "epoch": 2.18, "learning_rate": 3.6539477803069764e-05, "loss": 1.8035, "step": 7248 }, { "epoch": 2.18, "learning_rate": 3.6514365427079665e-05, "loss": 1.8106, "step": 7249 }, { "epoch": 2.18, "learning_rate": 3.648925975598131e-05, "loss": 1.7836, "step": 7250 }, { "epoch": 2.18, "learning_rate": 3.6464160792426196e-05, "loss": 1.7249, "step": 7251 }, { "epoch": 2.18, "learning_rate": 3.643906853906502e-05, "loss": 1.84, "step": 7252 }, { "epoch": 2.18, "learning_rate": 3.641398299854789e-05, "loss": 1.7955, "step": 7253 }, { "epoch": 2.18, "learning_rate": 3.638890417352415e-05, "loss": 1.7662, "step": 7254 }, { "epoch": 2.18, "learning_rate": 3.636383206664246e-05, "loss": 1.7344, "step": 7255 }, { "epoch": 2.18, "learning_rate": 3.633876668055076e-05, "loss": 1.7906, "step": 7256 }, { "epoch": 2.18, "learning_rate": 3.6313708017896234e-05, "loss": 1.7124, "step": 7257 }, { "epoch": 2.18, "learning_rate": 3.62886560813254e-05, "loss": 1.8453, "step": 7258 }, { "epoch": 2.19, "learning_rate": 3.62636108734841e-05, "loss": 1.7687, "step": 7259 }, { "epoch": 2.19, "learning_rate": 3.623857239701739e-05, "loss": 1.8058, "step": 7260 }, { "epoch": 2.19, "learning_rate": 3.62135406545697e-05, "loss": 1.8364, "step": 7261 }, { "epoch": 2.19, "learning_rate": 3.618851564878465e-05, "loss": 1.7676, "step": 7262 }, { "epoch": 2.19, "learning_rate": 3.61634973823052e-05, "loss": 1.8367, "step": 7263 }, { "epoch": 2.19, "learning_rate": 3.613848585777363e-05, "loss": 1.8675, "step": 7264 }, { "epoch": 2.19, "learning_rate": 3.611348107783149e-05, "loss": 1.7799, "step": 7265 }, { "epoch": 2.19, "learning_rate": 3.608848304511954e-05, "loss": 1.8563, "step": 7266 }, { "epoch": 2.19, "learning_rate": 3.606349176227792e-05, "loss": 1.8415, "step": 7267 }, { "epoch": 2.19, "learning_rate": 3.603850723194605e-05, "loss": 1.8236, "step": 7268 }, { "epoch": 2.19, "learning_rate": 3.601352945676259e-05, "loss": 1.7641, "step": 7269 }, { "epoch": 2.19, "learning_rate": 3.598855843936554e-05, "loss": 1.7806, "step": 7270 }, { "epoch": 2.19, "learning_rate": 3.596359418239211e-05, "loss": 1.7395, "step": 7271 }, { "epoch": 2.19, "learning_rate": 3.5938636688478876e-05, "loss": 1.8451, "step": 7272 }, { "epoch": 2.19, "learning_rate": 3.5913685960261656e-05, "loss": 1.8601, "step": 7273 }, { "epoch": 2.19, "learning_rate": 3.588874200037558e-05, "loss": 1.8169, "step": 7274 }, { "epoch": 2.19, "learning_rate": 3.586380481145506e-05, "loss": 1.834, "step": 7275 }, { "epoch": 2.19, "learning_rate": 3.5838874396133724e-05, "loss": 1.899, "step": 7276 }, { "epoch": 2.19, "learning_rate": 3.581395075704459e-05, "loss": 1.7104, "step": 7277 }, { "epoch": 2.19, "learning_rate": 3.578903389681988e-05, "loss": 1.752, "step": 7278 }, { "epoch": 2.19, "learning_rate": 3.576412381809117e-05, "loss": 1.8231, "step": 7279 }, { "epoch": 2.19, "learning_rate": 3.573922052348928e-05, "loss": 1.7494, "step": 7280 }, { "epoch": 2.19, "learning_rate": 3.571432401564426e-05, "loss": 1.816, "step": 7281 }, { "epoch": 2.19, "learning_rate": 3.5689434297185545e-05, "loss": 1.8104, "step": 7282 }, { "epoch": 2.19, "learning_rate": 3.566455137074179e-05, "loss": 1.8178, "step": 7283 }, { "epoch": 2.19, "learning_rate": 3.563967523894096e-05, "loss": 1.8306, "step": 7284 }, { "epoch": 2.19, "learning_rate": 3.561480590441032e-05, "loss": 1.7803, "step": 7285 }, { "epoch": 2.19, "learning_rate": 3.5589943369776326e-05, "loss": 1.7519, "step": 7286 }, { "epoch": 2.19, "learning_rate": 3.556508763766481e-05, "loss": 1.9068, "step": 7287 }, { "epoch": 2.19, "learning_rate": 3.554023871070083e-05, "loss": 1.8415, "step": 7288 }, { "epoch": 2.19, "learning_rate": 3.55153965915088e-05, "loss": 1.8345, "step": 7289 }, { "epoch": 2.19, "learning_rate": 3.549056128271235e-05, "loss": 1.8631, "step": 7290 }, { "epoch": 2.19, "learning_rate": 3.5465732786934356e-05, "loss": 1.8066, "step": 7291 }, { "epoch": 2.2, "learning_rate": 3.544091110679707e-05, "loss": 1.8037, "step": 7292 }, { "epoch": 2.2, "learning_rate": 3.5416096244921945e-05, "loss": 1.7855, "step": 7293 }, { "epoch": 2.2, "learning_rate": 3.539128820392982e-05, "loss": 1.8215, "step": 7294 }, { "epoch": 2.2, "learning_rate": 3.536648698644065e-05, "loss": 1.8836, "step": 7295 }, { "epoch": 2.2, "learning_rate": 3.5341692595073794e-05, "loss": 1.7937, "step": 7296 }, { "epoch": 2.2, "learning_rate": 3.5316905032447855e-05, "loss": 1.8336, "step": 7297 }, { "epoch": 2.2, "learning_rate": 3.529212430118072e-05, "loss": 1.8175, "step": 7298 }, { "epoch": 2.2, "learning_rate": 3.526735040388958e-05, "loss": 1.6677, "step": 7299 }, { "epoch": 2.2, "learning_rate": 3.52425833431908e-05, "loss": 1.8426, "step": 7300 }, { "epoch": 2.2, "learning_rate": 3.5217823121700125e-05, "loss": 1.7801, "step": 7301 }, { "epoch": 2.2, "learning_rate": 3.519306974203261e-05, "loss": 1.7339, "step": 7302 }, { "epoch": 2.2, "learning_rate": 3.516832320680247e-05, "loss": 1.7616, "step": 7303 }, { "epoch": 2.2, "learning_rate": 3.514358351862329e-05, "loss": 1.8433, "step": 7304 }, { "epoch": 2.2, "learning_rate": 3.511885068010784e-05, "loss": 1.8367, "step": 7305 }, { "epoch": 2.2, "learning_rate": 3.509412469386825e-05, "loss": 1.7377, "step": 7306 }, { "epoch": 2.2, "learning_rate": 3.50694055625159e-05, "loss": 1.8047, "step": 7307 }, { "epoch": 2.2, "learning_rate": 3.504469328866146e-05, "loss": 1.8212, "step": 7308 }, { "epoch": 2.2, "learning_rate": 3.5019987874914884e-05, "loss": 1.815, "step": 7309 }, { "epoch": 2.2, "learning_rate": 3.4995289323885304e-05, "loss": 1.7, "step": 7310 }, { "epoch": 2.2, "learning_rate": 3.4970597638181245e-05, "loss": 1.8306, "step": 7311 }, { "epoch": 2.2, "learning_rate": 3.494591282041045e-05, "loss": 1.8695, "step": 7312 }, { "epoch": 2.2, "learning_rate": 3.4921234873179973e-05, "loss": 1.8232, "step": 7313 }, { "epoch": 2.2, "learning_rate": 3.489656379909614e-05, "loss": 1.7574, "step": 7314 }, { "epoch": 2.2, "learning_rate": 3.4871899600764465e-05, "loss": 1.7962, "step": 7315 }, { "epoch": 2.2, "learning_rate": 3.4847242280789825e-05, "loss": 1.9089, "step": 7316 }, { "epoch": 2.2, "learning_rate": 3.482259184177637e-05, "loss": 1.8074, "step": 7317 }, { "epoch": 2.2, "learning_rate": 3.479794828632752e-05, "loss": 1.8952, "step": 7318 }, { "epoch": 2.2, "learning_rate": 3.477331161704587e-05, "loss": 1.8645, "step": 7319 }, { "epoch": 2.2, "learning_rate": 3.474868183653338e-05, "loss": 1.8119, "step": 7320 }, { "epoch": 2.2, "learning_rate": 3.472405894739138e-05, "loss": 1.746, "step": 7321 }, { "epoch": 2.2, "learning_rate": 3.469944295222024e-05, "loss": 1.7924, "step": 7322 }, { "epoch": 2.2, "learning_rate": 3.467483385361979e-05, "loss": 1.7626, "step": 7323 }, { "epoch": 2.2, "learning_rate": 3.465023165418901e-05, "loss": 1.8465, "step": 7324 }, { "epoch": 2.2, "learning_rate": 3.4625636356526204e-05, "loss": 1.8016, "step": 7325 }, { "epoch": 2.21, "learning_rate": 3.460104796322904e-05, "loss": 1.7723, "step": 7326 }, { "epoch": 2.21, "learning_rate": 3.457646647689428e-05, "loss": 1.7733, "step": 7327 }, { "epoch": 2.21, "learning_rate": 3.455189190011808e-05, "loss": 1.8292, "step": 7328 }, { "epoch": 2.21, "learning_rate": 3.4527324235495796e-05, "loss": 1.8561, "step": 7329 }, { "epoch": 2.21, "learning_rate": 3.4502763485622105e-05, "loss": 1.7754, "step": 7330 }, { "epoch": 2.21, "learning_rate": 3.447820965309093e-05, "loss": 1.8318, "step": 7331 }, { "epoch": 2.21, "learning_rate": 3.4453662740495475e-05, "loss": 1.7684, "step": 7332 }, { "epoch": 2.21, "learning_rate": 3.4429122750428235e-05, "loss": 1.7564, "step": 7333 }, { "epoch": 2.21, "learning_rate": 3.440458968548085e-05, "loss": 1.8559, "step": 7334 }, { "epoch": 2.21, "learning_rate": 3.438006354824446e-05, "loss": 1.7237, "step": 7335 }, { "epoch": 2.21, "learning_rate": 3.435554434130923e-05, "loss": 1.8302, "step": 7336 }, { "epoch": 2.21, "learning_rate": 3.433103206726474e-05, "loss": 1.7566, "step": 7337 }, { "epoch": 2.21, "learning_rate": 3.430652672869983e-05, "loss": 1.8524, "step": 7338 }, { "epoch": 2.21, "learning_rate": 3.428202832820246e-05, "loss": 1.8739, "step": 7339 }, { "epoch": 2.21, "learning_rate": 3.425753686836014e-05, "loss": 1.8316, "step": 7340 }, { "epoch": 2.21, "learning_rate": 3.423305235175936e-05, "loss": 1.7544, "step": 7341 }, { "epoch": 2.21, "learning_rate": 3.420857478098606e-05, "loss": 1.9333, "step": 7342 }, { "epoch": 2.21, "learning_rate": 3.418410415862532e-05, "loss": 1.8248, "step": 7343 }, { "epoch": 2.21, "learning_rate": 3.415964048726156e-05, "loss": 1.7695, "step": 7344 }, { "epoch": 2.21, "learning_rate": 3.413518376947854e-05, "loss": 1.7764, "step": 7345 }, { "epoch": 2.21, "learning_rate": 3.4110734007859115e-05, "loss": 1.9041, "step": 7346 }, { "epoch": 2.21, "learning_rate": 3.4086291204985546e-05, "loss": 1.7548, "step": 7347 }, { "epoch": 2.21, "learning_rate": 3.406185536343921e-05, "loss": 1.8266, "step": 7348 }, { "epoch": 2.21, "learning_rate": 3.403742648580094e-05, "loss": 1.9054, "step": 7349 }, { "epoch": 2.21, "learning_rate": 3.4013004574650745e-05, "loss": 1.7931, "step": 7350 }, { "epoch": 2.21, "learning_rate": 3.39885896325678e-05, "loss": 1.8042, "step": 7351 }, { "epoch": 2.21, "learning_rate": 3.3964181662130734e-05, "loss": 1.7512, "step": 7352 }, { "epoch": 2.21, "learning_rate": 3.393978066591722e-05, "loss": 1.7353, "step": 7353 }, { "epoch": 2.21, "learning_rate": 3.3915386646504455e-05, "loss": 1.787, "step": 7354 }, { "epoch": 2.21, "learning_rate": 3.389099960646864e-05, "loss": 1.8315, "step": 7355 }, { "epoch": 2.21, "learning_rate": 3.386661954838542e-05, "loss": 1.9067, "step": 7356 }, { "epoch": 2.21, "learning_rate": 3.3842246474829653e-05, "loss": 1.8346, "step": 7357 }, { "epoch": 2.21, "learning_rate": 3.3817880388375346e-05, "loss": 1.7382, "step": 7358 }, { "epoch": 2.22, "learning_rate": 3.379352129159601e-05, "loss": 1.746, "step": 7359 }, { "epoch": 2.22, "learning_rate": 3.376916918706417e-05, "loss": 1.8188, "step": 7360 }, { "epoch": 2.22, "learning_rate": 3.3744824077351755e-05, "loss": 1.8608, "step": 7361 }, { "epoch": 2.22, "learning_rate": 3.372048596502991e-05, "loss": 1.9038, "step": 7362 }, { "epoch": 2.22, "learning_rate": 3.369615485266906e-05, "loss": 1.8359, "step": 7363 }, { "epoch": 2.22, "learning_rate": 3.367183074283892e-05, "loss": 1.792, "step": 7364 }, { "epoch": 2.22, "learning_rate": 3.364751363810834e-05, "loss": 1.9849, "step": 7365 }, { "epoch": 2.22, "learning_rate": 3.3623203541045604e-05, "loss": 1.749, "step": 7366 }, { "epoch": 2.22, "learning_rate": 3.359890045421804e-05, "loss": 1.7981, "step": 7367 }, { "epoch": 2.22, "learning_rate": 3.3574604380192486e-05, "loss": 1.742, "step": 7368 }, { "epoch": 2.22, "learning_rate": 3.355031532153492e-05, "loss": 1.8556, "step": 7369 }, { "epoch": 2.22, "learning_rate": 3.3526033280810484e-05, "loss": 1.7178, "step": 7370 }, { "epoch": 2.22, "learning_rate": 3.3501758260583774e-05, "loss": 1.69, "step": 7371 }, { "epoch": 2.22, "learning_rate": 3.347749026341841e-05, "loss": 1.7921, "step": 7372 }, { "epoch": 2.22, "learning_rate": 3.345322929187751e-05, "loss": 1.7554, "step": 7373 }, { "epoch": 2.22, "learning_rate": 3.342897534852335e-05, "loss": 1.7781, "step": 7374 }, { "epoch": 2.22, "learning_rate": 3.3404728435917374e-05, "loss": 1.9522, "step": 7375 }, { "epoch": 2.22, "learning_rate": 3.338048855662043e-05, "loss": 1.7839, "step": 7376 }, { "epoch": 2.22, "learning_rate": 3.335625571319251e-05, "loss": 1.7236, "step": 7377 }, { "epoch": 2.22, "learning_rate": 3.3332029908192986e-05, "loss": 1.8955, "step": 7378 }, { "epoch": 2.22, "learning_rate": 3.330781114418031e-05, "loss": 1.8271, "step": 7379 }, { "epoch": 2.22, "learning_rate": 3.328359942371237e-05, "loss": 2.0075, "step": 7380 }, { "epoch": 2.22, "learning_rate": 3.325939474934618e-05, "loss": 1.9137, "step": 7381 }, { "epoch": 2.22, "learning_rate": 3.323519712363811e-05, "loss": 1.7916, "step": 7382 }, { "epoch": 2.22, "learning_rate": 3.321100654914373e-05, "loss": 1.9058, "step": 7383 }, { "epoch": 2.22, "learning_rate": 3.3186823028417835e-05, "loss": 1.8396, "step": 7384 }, { "epoch": 2.22, "learning_rate": 3.3162646564014546e-05, "loss": 1.7345, "step": 7385 }, { "epoch": 2.22, "learning_rate": 3.313847715848718e-05, "loss": 1.8087, "step": 7386 }, { "epoch": 2.22, "learning_rate": 3.311431481438836e-05, "loss": 1.7918, "step": 7387 }, { "epoch": 2.22, "learning_rate": 3.3090159534269946e-05, "loss": 1.8027, "step": 7388 }, { "epoch": 2.22, "learning_rate": 3.3066011320682996e-05, "loss": 1.8366, "step": 7389 }, { "epoch": 2.22, "learning_rate": 3.3041870176177944e-05, "loss": 1.9282, "step": 7390 }, { "epoch": 2.22, "learning_rate": 3.3017736103304286e-05, "loss": 1.8702, "step": 7391 }, { "epoch": 2.23, "learning_rate": 3.299360910461098e-05, "loss": 1.7171, "step": 7392 }, { "epoch": 2.23, "learning_rate": 3.296948918264616e-05, "loss": 1.8019, "step": 7393 }, { "epoch": 2.23, "learning_rate": 3.2945376339957126e-05, "loss": 1.8074, "step": 7394 }, { "epoch": 2.23, "learning_rate": 3.292127057909054e-05, "loss": 1.7888, "step": 7395 }, { "epoch": 2.23, "learning_rate": 3.289717190259226e-05, "loss": 1.8354, "step": 7396 }, { "epoch": 2.23, "learning_rate": 3.287308031300743e-05, "loss": 1.8612, "step": 7397 }, { "epoch": 2.23, "learning_rate": 3.284899581288047e-05, "loss": 1.7896, "step": 7398 }, { "epoch": 2.23, "learning_rate": 3.282491840475491e-05, "loss": 1.8349, "step": 7399 }, { "epoch": 2.23, "learning_rate": 3.2800848091173695e-05, "loss": 1.8399, "step": 7400 }, { "epoch": 2.23, "learning_rate": 3.277678487467895e-05, "loss": 1.7794, "step": 7401 }, { "epoch": 2.23, "learning_rate": 3.275272875781208e-05, "loss": 1.7563, "step": 7402 }, { "epoch": 2.23, "learning_rate": 3.272867974311365e-05, "loss": 1.6981, "step": 7403 }, { "epoch": 2.23, "learning_rate": 3.2704637833123595e-05, "loss": 1.7833, "step": 7404 }, { "epoch": 2.23, "learning_rate": 3.268060303038103e-05, "loss": 1.7885, "step": 7405 }, { "epoch": 2.23, "learning_rate": 3.265657533742434e-05, "loss": 1.8105, "step": 7406 }, { "epoch": 2.23, "learning_rate": 3.26325547567912e-05, "loss": 1.7721, "step": 7407 }, { "epoch": 2.23, "learning_rate": 3.260854129101841e-05, "loss": 1.8702, "step": 7408 }, { "epoch": 2.23, "learning_rate": 3.258453494264213e-05, "loss": 1.8878, "step": 7409 }, { "epoch": 2.23, "learning_rate": 3.256053571419774e-05, "loss": 1.7976, "step": 7410 }, { "epoch": 2.23, "learning_rate": 3.253654360821987e-05, "loss": 1.9426, "step": 7411 }, { "epoch": 2.23, "learning_rate": 3.2512558627242415e-05, "loss": 1.8002, "step": 7412 }, { "epoch": 2.23, "learning_rate": 3.248858077379843e-05, "loss": 1.8855, "step": 7413 }, { "epoch": 2.23, "learning_rate": 3.246461005042033e-05, "loss": 1.7945, "step": 7414 }, { "epoch": 2.23, "learning_rate": 3.24406464596397e-05, "loss": 1.8011, "step": 7415 }, { "epoch": 2.23, "learning_rate": 3.241669000398743e-05, "loss": 1.8019, "step": 7416 }, { "epoch": 2.23, "learning_rate": 3.2392740685993636e-05, "loss": 1.971, "step": 7417 }, { "epoch": 2.23, "learning_rate": 3.2368798508187624e-05, "loss": 1.9105, "step": 7418 }, { "epoch": 2.23, "learning_rate": 3.2344863473098016e-05, "loss": 1.8307, "step": 7419 }, { "epoch": 2.23, "learning_rate": 3.2320935583252664e-05, "loss": 1.8798, "step": 7420 }, { "epoch": 2.23, "learning_rate": 3.229701484117865e-05, "loss": 1.8955, "step": 7421 }, { "epoch": 2.23, "learning_rate": 3.227310124940235e-05, "loss": 1.7587, "step": 7422 }, { "epoch": 2.23, "learning_rate": 3.2249194810449255e-05, "loss": 1.7714, "step": 7423 }, { "epoch": 2.23, "learning_rate": 3.222529552684426e-05, "loss": 1.7869, "step": 7424 }, { "epoch": 2.24, "learning_rate": 3.2201403401111396e-05, "loss": 1.7899, "step": 7425 }, { "epoch": 2.24, "learning_rate": 3.2177518435774014e-05, "loss": 1.7243, "step": 7426 }, { "epoch": 2.24, "learning_rate": 3.215364063335466e-05, "loss": 1.7132, "step": 7427 }, { "epoch": 2.24, "learning_rate": 3.21297699963751e-05, "loss": 1.8268, "step": 7428 }, { "epoch": 2.24, "learning_rate": 3.210590652735641e-05, "loss": 1.7667, "step": 7429 }, { "epoch": 2.24, "learning_rate": 3.208205022881886e-05, "loss": 1.7111, "step": 7430 }, { "epoch": 2.24, "learning_rate": 3.205820110328203e-05, "loss": 1.84, "step": 7431 }, { "epoch": 2.24, "learning_rate": 3.20343591532646e-05, "loss": 1.8984, "step": 7432 }, { "epoch": 2.24, "learning_rate": 3.2010524381284644e-05, "loss": 1.8039, "step": 7433 }, { "epoch": 2.24, "learning_rate": 3.198669678985941e-05, "loss": 1.8299, "step": 7434 }, { "epoch": 2.24, "learning_rate": 3.196287638150538e-05, "loss": 1.73, "step": 7435 }, { "epoch": 2.24, "learning_rate": 3.1939063158738344e-05, "loss": 1.8219, "step": 7436 }, { "epoch": 2.24, "learning_rate": 3.191525712407321e-05, "loss": 1.7821, "step": 7437 }, { "epoch": 2.24, "learning_rate": 3.189145828002421e-05, "loss": 1.8811, "step": 7438 }, { "epoch": 2.24, "learning_rate": 3.186766662910484e-05, "loss": 1.7083, "step": 7439 }, { "epoch": 2.24, "learning_rate": 3.184388217382779e-05, "loss": 1.78, "step": 7440 }, { "epoch": 2.24, "learning_rate": 3.182010491670503e-05, "loss": 1.9197, "step": 7441 }, { "epoch": 2.24, "learning_rate": 3.179633486024767e-05, "loss": 1.8058, "step": 7442 }, { "epoch": 2.24, "learning_rate": 3.177257200696617e-05, "loss": 1.9094, "step": 7443 }, { "epoch": 2.24, "learning_rate": 3.1748816359370195e-05, "loss": 1.8267, "step": 7444 }, { "epoch": 2.24, "learning_rate": 3.172506791996863e-05, "loss": 1.849, "step": 7445 }, { "epoch": 2.24, "learning_rate": 3.170132669126966e-05, "loss": 1.7167, "step": 7446 }, { "epoch": 2.24, "learning_rate": 3.1677592675780586e-05, "loss": 1.7942, "step": 7447 }, { "epoch": 2.24, "learning_rate": 3.165386587600806e-05, "loss": 1.8257, "step": 7448 }, { "epoch": 2.24, "learning_rate": 3.163014629445794e-05, "loss": 1.8671, "step": 7449 }, { "epoch": 2.24, "learning_rate": 3.16064339336353e-05, "loss": 1.7635, "step": 7450 }, { "epoch": 2.24, "learning_rate": 3.158272879604451e-05, "loss": 1.7712, "step": 7451 }, { "epoch": 2.24, "learning_rate": 3.155903088418908e-05, "loss": 1.9464, "step": 7452 }, { "epoch": 2.24, "learning_rate": 3.153534020057183e-05, "loss": 1.8787, "step": 7453 }, { "epoch": 2.24, "learning_rate": 3.1511656747694785e-05, "loss": 1.8594, "step": 7454 }, { "epoch": 2.24, "learning_rate": 3.148798052805928e-05, "loss": 1.7965, "step": 7455 }, { "epoch": 2.24, "learning_rate": 3.1464311544165746e-05, "loss": 1.8338, "step": 7456 }, { "epoch": 2.24, "learning_rate": 3.1440649798513964e-05, "loss": 1.9274, "step": 7457 }, { "epoch": 2.25, "learning_rate": 3.141699529360291e-05, "loss": 1.7423, "step": 7458 }, { "epoch": 2.25, "learning_rate": 3.139334803193081e-05, "loss": 1.8113, "step": 7459 }, { "epoch": 2.25, "learning_rate": 3.136970801599515e-05, "loss": 1.7625, "step": 7460 }, { "epoch": 2.25, "learning_rate": 3.1346075248292536e-05, "loss": 1.8325, "step": 7461 }, { "epoch": 2.25, "learning_rate": 3.1322449731318934e-05, "loss": 1.8813, "step": 7462 }, { "epoch": 2.25, "learning_rate": 3.12988314675695e-05, "loss": 1.8086, "step": 7463 }, { "epoch": 2.25, "learning_rate": 3.1275220459538626e-05, "loss": 1.8951, "step": 7464 }, { "epoch": 2.25, "learning_rate": 3.125161670971996e-05, "loss": 1.784, "step": 7465 }, { "epoch": 2.25, "learning_rate": 3.1228020220606304e-05, "loss": 1.8914, "step": 7466 }, { "epoch": 2.25, "learning_rate": 3.120443099468976e-05, "loss": 1.7407, "step": 7467 }, { "epoch": 2.25, "learning_rate": 3.118084903446167e-05, "loss": 1.7924, "step": 7468 }, { "epoch": 2.25, "learning_rate": 3.115727434241258e-05, "loss": 1.8049, "step": 7469 }, { "epoch": 2.25, "learning_rate": 3.113370692103233e-05, "loss": 1.8437, "step": 7470 }, { "epoch": 2.25, "learning_rate": 3.111014677280983e-05, "loss": 1.7692, "step": 7471 }, { "epoch": 2.25, "learning_rate": 3.108659390023341e-05, "loss": 1.8463, "step": 7472 }, { "epoch": 2.25, "learning_rate": 3.106304830579053e-05, "loss": 1.798, "step": 7473 }, { "epoch": 2.25, "learning_rate": 3.103950999196792e-05, "loss": 1.8043, "step": 7474 }, { "epoch": 2.25, "learning_rate": 3.101597896125155e-05, "loss": 1.7614, "step": 7475 }, { "epoch": 2.25, "learning_rate": 3.0992455216126517e-05, "loss": 1.8736, "step": 7476 }, { "epoch": 2.25, "learning_rate": 3.096893875907728e-05, "loss": 1.7645, "step": 7477 }, { "epoch": 2.25, "learning_rate": 3.0945429592587485e-05, "loss": 1.7594, "step": 7478 }, { "epoch": 2.25, "learning_rate": 3.092192771914e-05, "loss": 1.8007, "step": 7479 }, { "epoch": 2.25, "learning_rate": 3.089843314121688e-05, "loss": 1.8505, "step": 7480 }, { "epoch": 2.25, "learning_rate": 3.087494586129944e-05, "loss": 1.8241, "step": 7481 }, { "epoch": 2.25, "learning_rate": 3.0851465881868355e-05, "loss": 1.7338, "step": 7482 }, { "epoch": 2.25, "learning_rate": 3.082799320540329e-05, "loss": 1.7486, "step": 7483 }, { "epoch": 2.25, "learning_rate": 3.080452783438334e-05, "loss": 1.879, "step": 7484 }, { "epoch": 2.25, "learning_rate": 3.078106977128665e-05, "loss": 1.8487, "step": 7485 }, { "epoch": 2.25, "learning_rate": 3.075761901859073e-05, "loss": 1.893, "step": 7486 }, { "epoch": 2.25, "learning_rate": 3.073417557877236e-05, "loss": 1.7363, "step": 7487 }, { "epoch": 2.25, "learning_rate": 3.071073945430737e-05, "loss": 1.7785, "step": 7488 }, { "epoch": 2.25, "learning_rate": 3.068731064767098e-05, "loss": 1.8257, "step": 7489 }, { "epoch": 2.25, "learning_rate": 3.066388916133749e-05, "loss": 1.8195, "step": 7490 }, { "epoch": 2.25, "learning_rate": 3.0640474997780554e-05, "loss": 1.8311, "step": 7491 }, { "epoch": 2.26, "learning_rate": 3.0617068159472995e-05, "loss": 1.8332, "step": 7492 }, { "epoch": 2.26, "learning_rate": 3.05936686488869e-05, "loss": 1.7569, "step": 7493 }, { "epoch": 2.26, "learning_rate": 3.057027646849355e-05, "loss": 1.8424, "step": 7494 }, { "epoch": 2.26, "learning_rate": 3.0546891620763395e-05, "loss": 1.8085, "step": 7495 }, { "epoch": 2.26, "learning_rate": 3.052351410816628e-05, "loss": 1.918, "step": 7496 }, { "epoch": 2.26, "learning_rate": 3.0500143933171076e-05, "loss": 1.9852, "step": 7497 }, { "epoch": 2.26, "learning_rate": 3.047678109824601e-05, "loss": 1.7716, "step": 7498 }, { "epoch": 2.26, "learning_rate": 3.045342560585852e-05, "loss": 1.9132, "step": 7499 }, { "epoch": 2.26, "learning_rate": 3.043007745847515e-05, "loss": 1.798, "step": 7500 }, { "epoch": 2.26, "learning_rate": 3.0406736658561886e-05, "loss": 1.7895, "step": 7501 }, { "epoch": 2.26, "learning_rate": 3.0383403208583726e-05, "loss": 1.8303, "step": 7502 }, { "epoch": 2.26, "learning_rate": 3.0360077111005038e-05, "loss": 1.8571, "step": 7503 }, { "epoch": 2.26, "learning_rate": 3.0336758368289288e-05, "loss": 1.9099, "step": 7504 }, { "epoch": 2.26, "learning_rate": 3.031344698289924e-05, "loss": 1.8016, "step": 7505 }, { "epoch": 2.26, "learning_rate": 3.0290142957296974e-05, "loss": 1.8286, "step": 7506 }, { "epoch": 2.26, "learning_rate": 3.0266846293943572e-05, "loss": 1.7091, "step": 7507 }, { "epoch": 2.26, "learning_rate": 3.024355699529955e-05, "loss": 1.8303, "step": 7508 }, { "epoch": 2.26, "learning_rate": 3.022027506382443e-05, "loss": 1.9364, "step": 7509 }, { "epoch": 2.26, "learning_rate": 3.0197000501977213e-05, "loss": 1.7796, "step": 7510 }, { "epoch": 2.26, "learning_rate": 3.0173733312215957e-05, "loss": 1.865, "step": 7511 }, { "epoch": 2.26, "learning_rate": 3.0150473496997922e-05, "loss": 1.7538, "step": 7512 }, { "epoch": 2.26, "learning_rate": 3.012722105877971e-05, "loss": 1.7427, "step": 7513 }, { "epoch": 2.26, "learning_rate": 3.0103976000016977e-05, "loss": 1.7789, "step": 7514 }, { "epoch": 2.26, "learning_rate": 3.0080738323164813e-05, "loss": 1.8021, "step": 7515 }, { "epoch": 2.26, "learning_rate": 3.0057508030677328e-05, "loss": 1.8428, "step": 7516 }, { "epoch": 2.26, "learning_rate": 3.0034285125007965e-05, "loss": 1.7961, "step": 7517 }, { "epoch": 2.26, "learning_rate": 3.0011069608609388e-05, "loss": 1.7702, "step": 7518 }, { "epoch": 2.26, "learning_rate": 2.9987861483933367e-05, "loss": 1.816, "step": 7519 }, { "epoch": 2.26, "learning_rate": 2.9964660753431096e-05, "loss": 1.8488, "step": 7520 }, { "epoch": 2.26, "learning_rate": 2.994146741955276e-05, "loss": 1.7939, "step": 7521 }, { "epoch": 2.26, "learning_rate": 2.9918281484747922e-05, "loss": 1.8284, "step": 7522 }, { "epoch": 2.26, "learning_rate": 2.9895102951465337e-05, "loss": 1.7557, "step": 7523 }, { "epoch": 2.26, "learning_rate": 2.9871931822152855e-05, "loss": 1.8169, "step": 7524 }, { "epoch": 2.27, "learning_rate": 2.9848768099257774e-05, "loss": 1.8324, "step": 7525 }, { "epoch": 2.27, "learning_rate": 2.982561178522638e-05, "loss": 1.8471, "step": 7526 }, { "epoch": 2.27, "learning_rate": 2.98024628825043e-05, "loss": 1.8989, "step": 7527 }, { "epoch": 2.27, "learning_rate": 2.9779321393536365e-05, "loss": 1.8446, "step": 7528 }, { "epoch": 2.27, "learning_rate": 2.9756187320766606e-05, "loss": 1.7869, "step": 7529 }, { "epoch": 2.27, "learning_rate": 2.9733060666638303e-05, "loss": 1.822, "step": 7530 }, { "epoch": 2.27, "learning_rate": 2.970994143359387e-05, "loss": 1.7825, "step": 7531 }, { "epoch": 2.27, "learning_rate": 2.9686829624075052e-05, "loss": 1.8843, "step": 7532 }, { "epoch": 2.27, "learning_rate": 2.9663725240522656e-05, "loss": 1.8094, "step": 7533 }, { "epoch": 2.27, "learning_rate": 2.964062828537689e-05, "loss": 1.8448, "step": 7534 }, { "epoch": 2.27, "learning_rate": 2.961753876107709e-05, "loss": 1.8512, "step": 7535 }, { "epoch": 2.27, "learning_rate": 2.9594456670061753e-05, "loss": 1.8432, "step": 7536 }, { "epoch": 2.27, "learning_rate": 2.9571382014768677e-05, "loss": 1.6887, "step": 7537 }, { "epoch": 2.27, "learning_rate": 2.954831479763477e-05, "loss": 1.8589, "step": 7538 }, { "epoch": 2.27, "learning_rate": 2.952525502109631e-05, "loss": 1.8809, "step": 7539 }, { "epoch": 2.27, "learning_rate": 2.9502202687588698e-05, "loss": 1.8388, "step": 7540 }, { "epoch": 2.27, "learning_rate": 2.9479157799546498e-05, "loss": 1.8556, "step": 7541 }, { "epoch": 2.27, "learning_rate": 2.9456120359403573e-05, "loss": 1.8551, "step": 7542 }, { "epoch": 2.27, "learning_rate": 2.9433090369592964e-05, "loss": 1.7068, "step": 7543 }, { "epoch": 2.27, "learning_rate": 2.9410067832546974e-05, "loss": 1.7448, "step": 7544 }, { "epoch": 2.27, "learning_rate": 2.9387052750697007e-05, "loss": 1.8386, "step": 7545 }, { "epoch": 2.27, "learning_rate": 2.936404512647378e-05, "loss": 1.9053, "step": 7546 }, { "epoch": 2.27, "learning_rate": 2.9341044962307206e-05, "loss": 1.7672, "step": 7547 }, { "epoch": 2.27, "learning_rate": 2.931805226062637e-05, "loss": 1.8634, "step": 7548 }, { "epoch": 2.27, "learning_rate": 2.9295067023859634e-05, "loss": 1.8444, "step": 7549 }, { "epoch": 2.27, "learning_rate": 2.9272089254434474e-05, "loss": 1.7162, "step": 7550 }, { "epoch": 2.27, "learning_rate": 2.9249118954777676e-05, "loss": 1.9693, "step": 7551 }, { "epoch": 2.27, "learning_rate": 2.9226156127315175e-05, "loss": 1.8463, "step": 7552 }, { "epoch": 2.27, "learning_rate": 2.920320077447215e-05, "loss": 1.7598, "step": 7553 }, { "epoch": 2.27, "learning_rate": 2.918025289867301e-05, "loss": 1.8546, "step": 7554 }, { "epoch": 2.27, "learning_rate": 2.9157312502341284e-05, "loss": 1.8409, "step": 7555 }, { "epoch": 2.27, "learning_rate": 2.9134379587899784e-05, "loss": 1.7922, "step": 7556 }, { "epoch": 2.27, "learning_rate": 2.9111454157770533e-05, "loss": 1.8817, "step": 7557 }, { "epoch": 2.28, "learning_rate": 2.908853621437475e-05, "loss": 1.9253, "step": 7558 }, { "epoch": 2.28, "learning_rate": 2.9065625760132886e-05, "loss": 1.857, "step": 7559 }, { "epoch": 2.28, "learning_rate": 2.9042722797464527e-05, "loss": 1.7502, "step": 7560 }, { "epoch": 2.28, "learning_rate": 2.901982732878853e-05, "loss": 1.8224, "step": 7561 }, { "epoch": 2.28, "learning_rate": 2.8996939356522957e-05, "loss": 1.7851, "step": 7562 }, { "epoch": 2.28, "learning_rate": 2.8974058883085064e-05, "loss": 1.8549, "step": 7563 }, { "epoch": 2.28, "learning_rate": 2.8951185910891365e-05, "loss": 1.799, "step": 7564 }, { "epoch": 2.28, "learning_rate": 2.8928320442357458e-05, "loss": 1.8744, "step": 7565 }, { "epoch": 2.28, "learning_rate": 2.8905462479898282e-05, "loss": 1.7, "step": 7566 }, { "epoch": 2.28, "learning_rate": 2.8882612025927902e-05, "loss": 1.9741, "step": 7567 }, { "epoch": 2.28, "learning_rate": 2.8859769082859666e-05, "loss": 1.7508, "step": 7568 }, { "epoch": 2.28, "learning_rate": 2.8836933653106013e-05, "loss": 1.746, "step": 7569 }, { "epoch": 2.28, "learning_rate": 2.881410573907869e-05, "loss": 1.7701, "step": 7570 }, { "epoch": 2.28, "learning_rate": 2.87912853431886e-05, "loss": 1.7252, "step": 7571 }, { "epoch": 2.28, "learning_rate": 2.8768472467845886e-05, "loss": 1.8523, "step": 7572 }, { "epoch": 2.28, "learning_rate": 2.8745667115459906e-05, "loss": 1.8433, "step": 7573 }, { "epoch": 2.28, "learning_rate": 2.8722869288439124e-05, "loss": 1.8524, "step": 7574 }, { "epoch": 2.28, "learning_rate": 2.8700078989191313e-05, "loss": 1.7829, "step": 7575 }, { "epoch": 2.28, "learning_rate": 2.867729622012343e-05, "loss": 1.8957, "step": 7576 }, { "epoch": 2.28, "learning_rate": 2.865452098364162e-05, "loss": 1.7852, "step": 7577 }, { "epoch": 2.28, "learning_rate": 2.8631753282151263e-05, "loss": 1.7758, "step": 7578 }, { "epoch": 2.28, "learning_rate": 2.860899311805687e-05, "loss": 1.746, "step": 7579 }, { "epoch": 2.28, "learning_rate": 2.8586240493762227e-05, "loss": 1.7228, "step": 7580 }, { "epoch": 2.28, "learning_rate": 2.8563495411670292e-05, "loss": 1.7376, "step": 7581 }, { "epoch": 2.28, "learning_rate": 2.8540757874183255e-05, "loss": 1.7744, "step": 7582 }, { "epoch": 2.28, "learning_rate": 2.8518027883702502e-05, "loss": 1.7041, "step": 7583 }, { "epoch": 2.28, "learning_rate": 2.849530544262856e-05, "loss": 1.8856, "step": 7584 }, { "epoch": 2.28, "learning_rate": 2.8472590553361232e-05, "loss": 1.7931, "step": 7585 }, { "epoch": 2.28, "learning_rate": 2.84498832182995e-05, "loss": 1.8593, "step": 7586 }, { "epoch": 2.28, "learning_rate": 2.8427183439841554e-05, "loss": 1.8431, "step": 7587 }, { "epoch": 2.28, "learning_rate": 2.840449122038481e-05, "loss": 2.096, "step": 7588 }, { "epoch": 2.28, "learning_rate": 2.8381806562325786e-05, "loss": 1.7422, "step": 7589 }, { "epoch": 2.28, "learning_rate": 2.8359129468060307e-05, "loss": 1.75, "step": 7590 }, { "epoch": 2.29, "learning_rate": 2.833645993998336e-05, "loss": 1.8173, "step": 7591 }, { "epoch": 2.29, "learning_rate": 2.8313797980489177e-05, "loss": 1.8456, "step": 7592 }, { "epoch": 2.29, "learning_rate": 2.829114359197107e-05, "loss": 1.7969, "step": 7593 }, { "epoch": 2.29, "learning_rate": 2.8268496776821673e-05, "loss": 1.8675, "step": 7594 }, { "epoch": 2.29, "learning_rate": 2.8245857537432785e-05, "loss": 1.7552, "step": 7595 }, { "epoch": 2.29, "learning_rate": 2.822322587619538e-05, "loss": 1.7968, "step": 7596 }, { "epoch": 2.29, "learning_rate": 2.8200601795499705e-05, "loss": 1.9136, "step": 7597 }, { "epoch": 2.29, "learning_rate": 2.8177985297735065e-05, "loss": 1.8536, "step": 7598 }, { "epoch": 2.29, "learning_rate": 2.815537638529009e-05, "loss": 1.7409, "step": 7599 }, { "epoch": 2.29, "learning_rate": 2.8132775060552574e-05, "loss": 1.7373, "step": 7600 }, { "epoch": 2.29, "learning_rate": 2.81101813259095e-05, "loss": 1.747, "step": 7601 }, { "epoch": 2.29, "learning_rate": 2.8087595183747082e-05, "loss": 1.7211, "step": 7602 }, { "epoch": 2.29, "learning_rate": 2.8065016636450657e-05, "loss": 1.7499, "step": 7603 }, { "epoch": 2.29, "learning_rate": 2.804244568640483e-05, "loss": 1.8366, "step": 7604 }, { "epoch": 2.29, "learning_rate": 2.8019882335993362e-05, "loss": 1.7324, "step": 7605 }, { "epoch": 2.29, "learning_rate": 2.799732658759926e-05, "loss": 1.7798, "step": 7606 }, { "epoch": 2.29, "learning_rate": 2.797477844360471e-05, "loss": 1.7767, "step": 7607 }, { "epoch": 2.29, "learning_rate": 2.7952237906391032e-05, "loss": 1.7857, "step": 7608 }, { "epoch": 2.29, "learning_rate": 2.792970497833882e-05, "loss": 1.9293, "step": 7609 }, { "epoch": 2.29, "learning_rate": 2.7907179661827843e-05, "loss": 1.7791, "step": 7610 }, { "epoch": 2.29, "learning_rate": 2.7884661959237044e-05, "loss": 1.855, "step": 7611 }, { "epoch": 2.29, "learning_rate": 2.7862151872944632e-05, "loss": 1.8366, "step": 7612 }, { "epoch": 2.29, "learning_rate": 2.7839649405327882e-05, "loss": 1.8517, "step": 7613 }, { "epoch": 2.29, "learning_rate": 2.7817154558763393e-05, "loss": 1.8502, "step": 7614 }, { "epoch": 2.29, "learning_rate": 2.7794667335626878e-05, "loss": 1.8856, "step": 7615 }, { "epoch": 2.29, "learning_rate": 2.777218773829333e-05, "loss": 1.7639, "step": 7616 }, { "epoch": 2.29, "learning_rate": 2.774971576913681e-05, "loss": 1.7385, "step": 7617 }, { "epoch": 2.29, "learning_rate": 2.7727251430530666e-05, "loss": 1.821, "step": 7618 }, { "epoch": 2.29, "learning_rate": 2.7704794724847428e-05, "loss": 1.9037, "step": 7619 }, { "epoch": 2.29, "learning_rate": 2.7682345654458808e-05, "loss": 1.7433, "step": 7620 }, { "epoch": 2.29, "learning_rate": 2.7659904221735743e-05, "loss": 1.7508, "step": 7621 }, { "epoch": 2.29, "learning_rate": 2.7637470429048285e-05, "loss": 1.8018, "step": 7622 }, { "epoch": 2.29, "learning_rate": 2.7615044278765735e-05, "loss": 1.8466, "step": 7623 }, { "epoch": 2.3, "learning_rate": 2.7592625773256607e-05, "loss": 1.8604, "step": 7624 }, { "epoch": 2.3, "learning_rate": 2.7570214914888557e-05, "loss": 1.7252, "step": 7625 }, { "epoch": 2.3, "learning_rate": 2.7547811706028505e-05, "loss": 1.8788, "step": 7626 }, { "epoch": 2.3, "learning_rate": 2.752541614904245e-05, "loss": 1.7531, "step": 7627 }, { "epoch": 2.3, "learning_rate": 2.750302824629567e-05, "loss": 1.7417, "step": 7628 }, { "epoch": 2.3, "learning_rate": 2.748064800015262e-05, "loss": 1.793, "step": 7629 }, { "epoch": 2.3, "learning_rate": 2.745827541297694e-05, "loss": 1.8182, "step": 7630 }, { "epoch": 2.3, "learning_rate": 2.7435910487131487e-05, "loss": 1.7596, "step": 7631 }, { "epoch": 2.3, "learning_rate": 2.741355322497823e-05, "loss": 1.8745, "step": 7632 }, { "epoch": 2.3, "learning_rate": 2.7391203628878393e-05, "loss": 1.7083, "step": 7633 }, { "epoch": 2.3, "learning_rate": 2.7368861701192395e-05, "loss": 1.8539, "step": 7634 }, { "epoch": 2.3, "learning_rate": 2.7346527444279823e-05, "loss": 1.8598, "step": 7635 }, { "epoch": 2.3, "learning_rate": 2.7324200860499482e-05, "loss": 1.8462, "step": 7636 }, { "epoch": 2.3, "learning_rate": 2.7301881952209298e-05, "loss": 1.8536, "step": 7637 }, { "epoch": 2.3, "learning_rate": 2.7279570721766444e-05, "loss": 1.7904, "step": 7638 }, { "epoch": 2.3, "learning_rate": 2.725726717152729e-05, "loss": 1.8734, "step": 7639 }, { "epoch": 2.3, "learning_rate": 2.7234971303847357e-05, "loss": 1.7772, "step": 7640 }, { "epoch": 2.3, "learning_rate": 2.7212683121081416e-05, "loss": 1.7201, "step": 7641 }, { "epoch": 2.3, "learning_rate": 2.719040262558329e-05, "loss": 1.7859, "step": 7642 }, { "epoch": 2.3, "learning_rate": 2.71681298197062e-05, "loss": 1.7937, "step": 7643 }, { "epoch": 2.3, "learning_rate": 2.7145864705802338e-05, "loss": 1.8447, "step": 7644 }, { "epoch": 2.3, "learning_rate": 2.7123607286223263e-05, "loss": 1.8019, "step": 7645 }, { "epoch": 2.3, "learning_rate": 2.7101357563319573e-05, "loss": 1.8481, "step": 7646 }, { "epoch": 2.3, "learning_rate": 2.7079115539441126e-05, "loss": 1.8065, "step": 7647 }, { "epoch": 2.3, "learning_rate": 2.7056881216937047e-05, "loss": 1.6965, "step": 7648 }, { "epoch": 2.3, "learning_rate": 2.7034654598155475e-05, "loss": 1.817, "step": 7649 }, { "epoch": 2.3, "learning_rate": 2.7012435685443894e-05, "loss": 1.9233, "step": 7650 }, { "epoch": 2.3, "learning_rate": 2.699022448114883e-05, "loss": 1.8505, "step": 7651 }, { "epoch": 2.3, "learning_rate": 2.696802098761607e-05, "loss": 1.7688, "step": 7652 }, { "epoch": 2.3, "learning_rate": 2.694582520719069e-05, "loss": 1.8491, "step": 7653 }, { "epoch": 2.3, "learning_rate": 2.692363714221674e-05, "loss": 1.7903, "step": 7654 }, { "epoch": 2.3, "learning_rate": 2.6901456795037626e-05, "loss": 1.8171, "step": 7655 }, { "epoch": 2.3, "learning_rate": 2.6879284167995834e-05, "loss": 1.7942, "step": 7656 }, { "epoch": 2.3, "learning_rate": 2.6857119263433084e-05, "loss": 1.7748, "step": 7657 }, { "epoch": 2.31, "learning_rate": 2.6834962083690273e-05, "loss": 1.826, "step": 7658 }, { "epoch": 2.31, "learning_rate": 2.68128126311075e-05, "loss": 1.8665, "step": 7659 }, { "epoch": 2.31, "learning_rate": 2.679067090802404e-05, "loss": 1.7401, "step": 7660 }, { "epoch": 2.31, "learning_rate": 2.676853691677825e-05, "loss": 1.8529, "step": 7661 }, { "epoch": 2.31, "learning_rate": 2.674641065970791e-05, "loss": 1.7938, "step": 7662 }, { "epoch": 2.31, "learning_rate": 2.6724292139149708e-05, "loss": 1.8287, "step": 7663 }, { "epoch": 2.31, "learning_rate": 2.6702181357439692e-05, "loss": 1.7326, "step": 7664 }, { "epoch": 2.31, "learning_rate": 2.668007831691307e-05, "loss": 1.7714, "step": 7665 }, { "epoch": 2.31, "learning_rate": 2.665798301990411e-05, "loss": 1.8067, "step": 7666 }, { "epoch": 2.31, "learning_rate": 2.6635895468746485e-05, "loss": 2.0127, "step": 7667 }, { "epoch": 2.31, "learning_rate": 2.6613815665772823e-05, "loss": 1.7914, "step": 7668 }, { "epoch": 2.31, "learning_rate": 2.6591743613315103e-05, "loss": 1.7504, "step": 7669 }, { "epoch": 2.31, "learning_rate": 2.656967931370433e-05, "loss": 1.7613, "step": 7670 }, { "epoch": 2.31, "learning_rate": 2.654762276927081e-05, "loss": 1.8854, "step": 7671 }, { "epoch": 2.31, "learning_rate": 2.6525573982344053e-05, "loss": 1.7423, "step": 7672 }, { "epoch": 2.31, "learning_rate": 2.650353295525263e-05, "loss": 1.8085, "step": 7673 }, { "epoch": 2.31, "learning_rate": 2.6481499690324396e-05, "loss": 1.7451, "step": 7674 }, { "epoch": 2.31, "learning_rate": 2.645947418988626e-05, "loss": 1.8, "step": 7675 }, { "epoch": 2.31, "learning_rate": 2.6437456456264487e-05, "loss": 1.8285, "step": 7676 }, { "epoch": 2.31, "learning_rate": 2.6415446491784423e-05, "loss": 1.7286, "step": 7677 }, { "epoch": 2.31, "learning_rate": 2.639344429877054e-05, "loss": 1.79, "step": 7678 }, { "epoch": 2.31, "learning_rate": 2.637144987954663e-05, "loss": 1.8127, "step": 7679 }, { "epoch": 2.31, "learning_rate": 2.634946323643547e-05, "loss": 1.8885, "step": 7680 }, { "epoch": 2.31, "learning_rate": 2.632748437175926e-05, "loss": 1.6448, "step": 7681 }, { "epoch": 2.31, "learning_rate": 2.6305513287839156e-05, "loss": 1.7472, "step": 7682 }, { "epoch": 2.31, "learning_rate": 2.6283549986995615e-05, "loss": 1.8004, "step": 7683 }, { "epoch": 2.31, "learning_rate": 2.626159447154828e-05, "loss": 1.8735, "step": 7684 }, { "epoch": 2.31, "learning_rate": 2.6239646743815814e-05, "loss": 1.7913, "step": 7685 }, { "epoch": 2.31, "learning_rate": 2.6217706806116337e-05, "loss": 1.8875, "step": 7686 }, { "epoch": 2.31, "learning_rate": 2.6195774660766868e-05, "loss": 1.8037, "step": 7687 }, { "epoch": 2.31, "learning_rate": 2.617385031008377e-05, "loss": 1.7363, "step": 7688 }, { "epoch": 2.31, "learning_rate": 2.615193375638251e-05, "loss": 1.8939, "step": 7689 }, { "epoch": 2.31, "learning_rate": 2.6130025001977775e-05, "loss": 1.817, "step": 7690 }, { "epoch": 2.32, "learning_rate": 2.6108124049183426e-05, "loss": 1.7404, "step": 7691 }, { "epoch": 2.32, "learning_rate": 2.6086230900312437e-05, "loss": 1.7048, "step": 7692 }, { "epoch": 2.32, "learning_rate": 2.606434555767704e-05, "loss": 1.8353, "step": 7693 }, { "epoch": 2.32, "learning_rate": 2.604246802358854e-05, "loss": 1.8555, "step": 7694 }, { "epoch": 2.32, "learning_rate": 2.602059830035757e-05, "loss": 1.7538, "step": 7695 }, { "epoch": 2.32, "learning_rate": 2.5998736390293833e-05, "loss": 1.7342, "step": 7696 }, { "epoch": 2.32, "learning_rate": 2.5976882295706184e-05, "loss": 1.9244, "step": 7697 }, { "epoch": 2.32, "learning_rate": 2.5955036018902746e-05, "loss": 1.7478, "step": 7698 }, { "epoch": 2.32, "learning_rate": 2.593319756219068e-05, "loss": 1.7949, "step": 7699 }, { "epoch": 2.32, "learning_rate": 2.591136692787649e-05, "loss": 1.6756, "step": 7700 }, { "epoch": 2.32, "learning_rate": 2.5889544118265763e-05, "loss": 1.7124, "step": 7701 }, { "epoch": 2.32, "learning_rate": 2.5867729135663222e-05, "loss": 1.8727, "step": 7702 }, { "epoch": 2.32, "learning_rate": 2.5845921982372824e-05, "loss": 1.7306, "step": 7703 }, { "epoch": 2.32, "learning_rate": 2.5824122660697682e-05, "loss": 1.8691, "step": 7704 }, { "epoch": 2.32, "learning_rate": 2.580233117294011e-05, "loss": 1.7888, "step": 7705 }, { "epoch": 2.32, "learning_rate": 2.5780547521401522e-05, "loss": 1.7546, "step": 7706 }, { "epoch": 2.32, "learning_rate": 2.5758771708382567e-05, "loss": 1.8312, "step": 7707 }, { "epoch": 2.32, "learning_rate": 2.5737003736183053e-05, "loss": 1.7744, "step": 7708 }, { "epoch": 2.32, "learning_rate": 2.5715243607101946e-05, "loss": 1.6889, "step": 7709 }, { "epoch": 2.32, "learning_rate": 2.5693491323437436e-05, "loss": 1.6837, "step": 7710 }, { "epoch": 2.32, "learning_rate": 2.567174688748677e-05, "loss": 1.8819, "step": 7711 }, { "epoch": 2.32, "learning_rate": 2.5650010301546478e-05, "loss": 1.7584, "step": 7712 }, { "epoch": 2.32, "learning_rate": 2.5628281567912214e-05, "loss": 1.759, "step": 7713 }, { "epoch": 2.32, "learning_rate": 2.560656068887881e-05, "loss": 1.8257, "step": 7714 }, { "epoch": 2.32, "learning_rate": 2.55848476667403e-05, "loss": 1.8395, "step": 7715 }, { "epoch": 2.32, "learning_rate": 2.5563142503789783e-05, "loss": 1.9168, "step": 7716 }, { "epoch": 2.32, "learning_rate": 2.5541445202319646e-05, "loss": 1.7472, "step": 7717 }, { "epoch": 2.32, "learning_rate": 2.55197557646214e-05, "loss": 1.7518, "step": 7718 }, { "epoch": 2.32, "learning_rate": 2.549807419298571e-05, "loss": 1.7815, "step": 7719 }, { "epoch": 2.32, "learning_rate": 2.547640048970248e-05, "loss": 1.693, "step": 7720 }, { "epoch": 2.32, "learning_rate": 2.5454734657060654e-05, "loss": 1.8028, "step": 7721 }, { "epoch": 2.32, "learning_rate": 2.5433076697348446e-05, "loss": 1.861, "step": 7722 }, { "epoch": 2.32, "learning_rate": 2.5411426612853216e-05, "loss": 1.8483, "step": 7723 }, { "epoch": 2.33, "learning_rate": 2.5389784405861507e-05, "loss": 1.8785, "step": 7724 }, { "epoch": 2.33, "learning_rate": 2.5368150078659015e-05, "loss": 1.8067, "step": 7725 }, { "epoch": 2.33, "learning_rate": 2.534652363353055e-05, "loss": 1.8681, "step": 7726 }, { "epoch": 2.33, "learning_rate": 2.5324905072760173e-05, "loss": 1.7718, "step": 7727 }, { "epoch": 2.33, "learning_rate": 2.5303294398631082e-05, "loss": 1.746, "step": 7728 }, { "epoch": 2.33, "learning_rate": 2.5281691613425673e-05, "loss": 1.817, "step": 7729 }, { "epoch": 2.33, "learning_rate": 2.5260096719425395e-05, "loss": 1.8132, "step": 7730 }, { "epoch": 2.33, "learning_rate": 2.5238509718910998e-05, "loss": 1.828, "step": 7731 }, { "epoch": 2.33, "learning_rate": 2.5216930614162327e-05, "loss": 1.8478, "step": 7732 }, { "epoch": 2.33, "learning_rate": 2.5195359407458407e-05, "loss": 1.7099, "step": 7733 }, { "epoch": 2.33, "learning_rate": 2.5173796101077484e-05, "loss": 1.9071, "step": 7734 }, { "epoch": 2.33, "learning_rate": 2.5152240697296847e-05, "loss": 1.8525, "step": 7735 }, { "epoch": 2.33, "learning_rate": 2.513069319839305e-05, "loss": 1.7935, "step": 7736 }, { "epoch": 2.33, "learning_rate": 2.5109153606641776e-05, "loss": 1.8029, "step": 7737 }, { "epoch": 2.33, "learning_rate": 2.5087621924317894e-05, "loss": 1.8079, "step": 7738 }, { "epoch": 2.33, "learning_rate": 2.5066098153695438e-05, "loss": 1.7772, "step": 7739 }, { "epoch": 2.33, "learning_rate": 2.5044582297047548e-05, "loss": 1.8418, "step": 7740 }, { "epoch": 2.33, "learning_rate": 2.5023074356646604e-05, "loss": 1.7818, "step": 7741 }, { "epoch": 2.33, "learning_rate": 2.50015743347641e-05, "loss": 1.8548, "step": 7742 }, { "epoch": 2.33, "learning_rate": 2.498008223367073e-05, "loss": 1.8673, "step": 7743 }, { "epoch": 2.33, "learning_rate": 2.495859805563635e-05, "loss": 1.8164, "step": 7744 }, { "epoch": 2.33, "learning_rate": 2.493712180292992e-05, "loss": 1.8696, "step": 7745 }, { "epoch": 2.33, "learning_rate": 2.491565347781961e-05, "loss": 1.8172, "step": 7746 }, { "epoch": 2.33, "learning_rate": 2.489419308257278e-05, "loss": 1.8287, "step": 7747 }, { "epoch": 2.33, "learning_rate": 2.4872740619455892e-05, "loss": 1.8836, "step": 7748 }, { "epoch": 2.33, "learning_rate": 2.4851296090734653e-05, "loss": 1.7217, "step": 7749 }, { "epoch": 2.33, "learning_rate": 2.48298594986738e-05, "loss": 1.7064, "step": 7750 }, { "epoch": 2.33, "learning_rate": 2.4808430845537355e-05, "loss": 1.7774, "step": 7751 }, { "epoch": 2.33, "learning_rate": 2.4787010133588438e-05, "loss": 1.7808, "step": 7752 }, { "epoch": 2.33, "learning_rate": 2.476559736508941e-05, "loss": 1.7797, "step": 7753 }, { "epoch": 2.33, "learning_rate": 2.4744192542301637e-05, "loss": 1.9032, "step": 7754 }, { "epoch": 2.33, "learning_rate": 2.4722795667485797e-05, "loss": 1.8265, "step": 7755 }, { "epoch": 2.33, "learning_rate": 2.4701406742901654e-05, "loss": 1.8783, "step": 7756 }, { "epoch": 2.34, "learning_rate": 2.4680025770808167e-05, "loss": 1.8326, "step": 7757 }, { "epoch": 2.34, "learning_rate": 2.465865275346345e-05, "loss": 1.8392, "step": 7758 }, { "epoch": 2.34, "learning_rate": 2.4637287693124733e-05, "loss": 1.8649, "step": 7759 }, { "epoch": 2.34, "learning_rate": 2.4615930592048442e-05, "loss": 1.8997, "step": 7760 }, { "epoch": 2.34, "learning_rate": 2.4594581452490172e-05, "loss": 1.8208, "step": 7761 }, { "epoch": 2.34, "learning_rate": 2.4573240276704667e-05, "loss": 1.8572, "step": 7762 }, { "epoch": 2.34, "learning_rate": 2.4551907066945846e-05, "loss": 1.6943, "step": 7763 }, { "epoch": 2.34, "learning_rate": 2.4530581825466712e-05, "loss": 1.8519, "step": 7764 }, { "epoch": 2.34, "learning_rate": 2.450926455451953e-05, "loss": 1.8256, "step": 7765 }, { "epoch": 2.34, "learning_rate": 2.4487955256355643e-05, "loss": 1.6945, "step": 7766 }, { "epoch": 2.34, "learning_rate": 2.4466653933225603e-05, "loss": 1.7776, "step": 7767 }, { "epoch": 2.34, "learning_rate": 2.444536058737913e-05, "loss": 1.7125, "step": 7768 }, { "epoch": 2.34, "learning_rate": 2.4424075221065012e-05, "loss": 1.7719, "step": 7769 }, { "epoch": 2.34, "learning_rate": 2.4402797836531278e-05, "loss": 1.7947, "step": 7770 }, { "epoch": 2.34, "learning_rate": 2.4381528436025104e-05, "loss": 1.6827, "step": 7771 }, { "epoch": 2.34, "learning_rate": 2.4360267021792803e-05, "loss": 1.9328, "step": 7772 }, { "epoch": 2.34, "learning_rate": 2.4339013596079874e-05, "loss": 1.8189, "step": 7773 }, { "epoch": 2.34, "learning_rate": 2.4317768161130903e-05, "loss": 1.7641, "step": 7774 }, { "epoch": 2.34, "learning_rate": 2.4296530719189704e-05, "loss": 1.8194, "step": 7775 }, { "epoch": 2.34, "learning_rate": 2.4275301272499217e-05, "loss": 1.7271, "step": 7776 }, { "epoch": 2.34, "learning_rate": 2.4254079823301535e-05, "loss": 1.6837, "step": 7777 }, { "epoch": 2.34, "learning_rate": 2.423286637383796e-05, "loss": 1.8184, "step": 7778 }, { "epoch": 2.34, "learning_rate": 2.4211660926348832e-05, "loss": 1.7592, "step": 7779 }, { "epoch": 2.34, "learning_rate": 2.4190463483073756e-05, "loss": 1.8576, "step": 7780 }, { "epoch": 2.34, "learning_rate": 2.4169274046251446e-05, "loss": 1.7641, "step": 7781 }, { "epoch": 2.34, "learning_rate": 2.41480926181198e-05, "loss": 1.7678, "step": 7782 }, { "epoch": 2.34, "learning_rate": 2.4126919200915788e-05, "loss": 1.8112, "step": 7783 }, { "epoch": 2.34, "learning_rate": 2.410575379687563e-05, "loss": 1.8822, "step": 7784 }, { "epoch": 2.34, "learning_rate": 2.408459640823466e-05, "loss": 1.9093, "step": 7785 }, { "epoch": 2.34, "learning_rate": 2.4063447037227372e-05, "loss": 1.7348, "step": 7786 }, { "epoch": 2.34, "learning_rate": 2.404230568608743e-05, "loss": 1.7359, "step": 7787 }, { "epoch": 2.34, "learning_rate": 2.4021172357047582e-05, "loss": 1.6976, "step": 7788 }, { "epoch": 2.34, "learning_rate": 2.400004705233979e-05, "loss": 1.8376, "step": 7789 }, { "epoch": 2.34, "learning_rate": 2.397892977419517e-05, "loss": 1.768, "step": 7790 }, { "epoch": 2.35, "learning_rate": 2.3957820524843967e-05, "loss": 1.7816, "step": 7791 }, { "epoch": 2.35, "learning_rate": 2.393671930651561e-05, "loss": 1.7882, "step": 7792 }, { "epoch": 2.35, "learning_rate": 2.3915626121438615e-05, "loss": 1.7998, "step": 7793 }, { "epoch": 2.35, "learning_rate": 2.3894540971840717e-05, "loss": 1.7962, "step": 7794 }, { "epoch": 2.35, "learning_rate": 2.3873463859948763e-05, "loss": 1.7628, "step": 7795 }, { "epoch": 2.35, "learning_rate": 2.3852394787988773e-05, "loss": 1.7573, "step": 7796 }, { "epoch": 2.35, "learning_rate": 2.3831333758185947e-05, "loss": 1.7424, "step": 7797 }, { "epoch": 2.35, "learning_rate": 2.3810280772764526e-05, "loss": 1.8088, "step": 7798 }, { "epoch": 2.35, "learning_rate": 2.3789235833948008e-05, "loss": 1.7928, "step": 7799 }, { "epoch": 2.35, "learning_rate": 2.3768198943958997e-05, "loss": 1.7932, "step": 7800 }, { "epoch": 2.35, "learning_rate": 2.3747170105019278e-05, "loss": 1.8007, "step": 7801 }, { "epoch": 2.35, "learning_rate": 2.372614931934978e-05, "loss": 1.7395, "step": 7802 }, { "epoch": 2.35, "learning_rate": 2.3705136589170507e-05, "loss": 1.9972, "step": 7803 }, { "epoch": 2.35, "learning_rate": 2.3684131916700693e-05, "loss": 1.7678, "step": 7804 }, { "epoch": 2.35, "learning_rate": 2.3663135304158723e-05, "loss": 1.8669, "step": 7805 }, { "epoch": 2.35, "learning_rate": 2.3642146753762105e-05, "loss": 1.6916, "step": 7806 }, { "epoch": 2.35, "learning_rate": 2.3621166267727467e-05, "loss": 1.8718, "step": 7807 }, { "epoch": 2.35, "learning_rate": 2.3600193848270604e-05, "loss": 1.8828, "step": 7808 }, { "epoch": 2.35, "learning_rate": 2.3579229497606546e-05, "loss": 1.7741, "step": 7809 }, { "epoch": 2.35, "learning_rate": 2.3558273217949323e-05, "loss": 1.7413, "step": 7810 }, { "epoch": 2.35, "learning_rate": 2.3537325011512246e-05, "loss": 1.732, "step": 7811 }, { "epoch": 2.35, "learning_rate": 2.3516384880507646e-05, "loss": 1.8231, "step": 7812 }, { "epoch": 2.35, "learning_rate": 2.3495452827147068e-05, "loss": 1.7709, "step": 7813 }, { "epoch": 2.35, "learning_rate": 2.34745288536413e-05, "loss": 1.8357, "step": 7814 }, { "epoch": 2.35, "learning_rate": 2.3453612962200068e-05, "loss": 1.8542, "step": 7815 }, { "epoch": 2.35, "learning_rate": 2.343270515503244e-05, "loss": 1.7698, "step": 7816 }, { "epoch": 2.35, "learning_rate": 2.341180543434649e-05, "loss": 1.7703, "step": 7817 }, { "epoch": 2.35, "learning_rate": 2.3390913802349502e-05, "loss": 1.7563, "step": 7818 }, { "epoch": 2.35, "learning_rate": 2.3370030261247923e-05, "loss": 1.8106, "step": 7819 }, { "epoch": 2.35, "learning_rate": 2.3349154813247297e-05, "loss": 1.8078, "step": 7820 }, { "epoch": 2.35, "learning_rate": 2.332828746055238e-05, "loss": 1.8157, "step": 7821 }, { "epoch": 2.35, "learning_rate": 2.3307428205366943e-05, "loss": 1.825, "step": 7822 }, { "epoch": 2.35, "learning_rate": 2.328657704989411e-05, "loss": 1.7595, "step": 7823 }, { "epoch": 2.36, "learning_rate": 2.3265733996335925e-05, "loss": 1.8009, "step": 7824 }, { "epoch": 2.36, "learning_rate": 2.3244899046893732e-05, "loss": 1.651, "step": 7825 }, { "epoch": 2.36, "learning_rate": 2.3224072203767965e-05, "loss": 1.8075, "step": 7826 }, { "epoch": 2.36, "learning_rate": 2.320325346915815e-05, "loss": 1.77, "step": 7827 }, { "epoch": 2.36, "learning_rate": 2.3182442845263108e-05, "loss": 1.8518, "step": 7828 }, { "epoch": 2.36, "learning_rate": 2.3161640334280622e-05, "loss": 1.822, "step": 7829 }, { "epoch": 2.36, "learning_rate": 2.314084593840775e-05, "loss": 1.8188, "step": 7830 }, { "epoch": 2.36, "learning_rate": 2.312005965984061e-05, "loss": 1.8167, "step": 7831 }, { "epoch": 2.36, "learning_rate": 2.3099281500774483e-05, "loss": 1.8055, "step": 7832 }, { "epoch": 2.36, "learning_rate": 2.3078511463403895e-05, "loss": 1.7786, "step": 7833 }, { "epoch": 2.36, "learning_rate": 2.3057749549922346e-05, "loss": 1.955, "step": 7834 }, { "epoch": 2.36, "learning_rate": 2.3036995762522616e-05, "loss": 1.8743, "step": 7835 }, { "epoch": 2.36, "learning_rate": 2.3016250103396473e-05, "loss": 1.8097, "step": 7836 }, { "epoch": 2.36, "learning_rate": 2.2995512574735012e-05, "loss": 1.8319, "step": 7837 }, { "epoch": 2.36, "learning_rate": 2.2974783178728386e-05, "loss": 1.8844, "step": 7838 }, { "epoch": 2.36, "learning_rate": 2.2954061917565815e-05, "loss": 1.7813, "step": 7839 }, { "epoch": 2.36, "learning_rate": 2.2933348793435803e-05, "loss": 1.7243, "step": 7840 }, { "epoch": 2.36, "learning_rate": 2.2912643808525826e-05, "loss": 1.7514, "step": 7841 }, { "epoch": 2.36, "learning_rate": 2.2891946965022705e-05, "loss": 1.7856, "step": 7842 }, { "epoch": 2.36, "learning_rate": 2.2871258265112206e-05, "loss": 1.7593, "step": 7843 }, { "epoch": 2.36, "learning_rate": 2.285057771097934e-05, "loss": 1.8078, "step": 7844 }, { "epoch": 2.36, "learning_rate": 2.2829905304808286e-05, "loss": 1.8191, "step": 7845 }, { "epoch": 2.36, "learning_rate": 2.2809241048782203e-05, "loss": 1.8851, "step": 7846 }, { "epoch": 2.36, "learning_rate": 2.2788584945083635e-05, "loss": 1.6932, "step": 7847 }, { "epoch": 2.36, "learning_rate": 2.2767936995894035e-05, "loss": 1.7858, "step": 7848 }, { "epoch": 2.36, "learning_rate": 2.2747297203394125e-05, "loss": 1.7852, "step": 7849 }, { "epoch": 2.36, "learning_rate": 2.272666556976374e-05, "loss": 1.8226, "step": 7850 }, { "epoch": 2.36, "learning_rate": 2.2706042097181778e-05, "loss": 1.7743, "step": 7851 }, { "epoch": 2.36, "learning_rate": 2.2685426787826434e-05, "loss": 1.8589, "step": 7852 }, { "epoch": 2.36, "learning_rate": 2.266481964387489e-05, "loss": 1.8471, "step": 7853 }, { "epoch": 2.36, "learning_rate": 2.2644220667503556e-05, "loss": 1.8418, "step": 7854 }, { "epoch": 2.36, "learning_rate": 2.2623629860887873e-05, "loss": 1.8901, "step": 7855 }, { "epoch": 2.36, "learning_rate": 2.2603047226202566e-05, "loss": 1.7166, "step": 7856 }, { "epoch": 2.37, "learning_rate": 2.2582472765621432e-05, "loss": 1.8061, "step": 7857 }, { "epoch": 2.37, "learning_rate": 2.256190648131734e-05, "loss": 1.6744, "step": 7858 }, { "epoch": 2.37, "learning_rate": 2.2541348375462413e-05, "loss": 1.6937, "step": 7859 }, { "epoch": 2.37, "learning_rate": 2.2520798450227753e-05, "loss": 1.8003, "step": 7860 }, { "epoch": 2.37, "learning_rate": 2.2500256707783785e-05, "loss": 1.8086, "step": 7861 }, { "epoch": 2.37, "learning_rate": 2.247972315029998e-05, "loss": 1.7698, "step": 7862 }, { "epoch": 2.37, "learning_rate": 2.2459197779944873e-05, "loss": 1.7703, "step": 7863 }, { "epoch": 2.37, "learning_rate": 2.2438680598886285e-05, "loss": 1.8245, "step": 7864 }, { "epoch": 2.37, "learning_rate": 2.2418171609290984e-05, "loss": 1.8604, "step": 7865 }, { "epoch": 2.37, "learning_rate": 2.239767081332511e-05, "loss": 1.7338, "step": 7866 }, { "epoch": 2.37, "learning_rate": 2.2377178213153716e-05, "loss": 1.7366, "step": 7867 }, { "epoch": 2.37, "learning_rate": 2.2356693810941097e-05, "loss": 1.7656, "step": 7868 }, { "epoch": 2.37, "learning_rate": 2.2336217608850696e-05, "loss": 1.8871, "step": 7869 }, { "epoch": 2.37, "learning_rate": 2.2315749609045046e-05, "loss": 1.8174, "step": 7870 }, { "epoch": 2.37, "learning_rate": 2.229528981368585e-05, "loss": 1.8135, "step": 7871 }, { "epoch": 2.37, "learning_rate": 2.2274838224933868e-05, "loss": 1.7361, "step": 7872 }, { "epoch": 2.37, "learning_rate": 2.225439484494909e-05, "loss": 1.7936, "step": 7873 }, { "epoch": 2.37, "learning_rate": 2.2233959675890593e-05, "loss": 1.6999, "step": 7874 }, { "epoch": 2.37, "learning_rate": 2.2213532719916576e-05, "loss": 1.7541, "step": 7875 }, { "epoch": 2.37, "learning_rate": 2.219311397918442e-05, "loss": 1.7508, "step": 7876 }, { "epoch": 2.37, "learning_rate": 2.2172703455850564e-05, "loss": 1.8174, "step": 7877 }, { "epoch": 2.37, "learning_rate": 2.215230115207063e-05, "loss": 1.8111, "step": 7878 }, { "epoch": 2.37, "learning_rate": 2.213190706999937e-05, "loss": 1.8105, "step": 7879 }, { "epoch": 2.37, "learning_rate": 2.2111521211790652e-05, "loss": 1.7932, "step": 7880 }, { "epoch": 2.37, "learning_rate": 2.2091143579597517e-05, "loss": 1.7539, "step": 7881 }, { "epoch": 2.37, "learning_rate": 2.2070774175572052e-05, "loss": 1.8192, "step": 7882 }, { "epoch": 2.37, "learning_rate": 2.2050413001865533e-05, "loss": 1.7287, "step": 7883 }, { "epoch": 2.37, "learning_rate": 2.2030060060628377e-05, "loss": 1.9037, "step": 7884 }, { "epoch": 2.37, "learning_rate": 2.2009715354010107e-05, "loss": 1.779, "step": 7885 }, { "epoch": 2.37, "learning_rate": 2.1989378884159407e-05, "loss": 1.6522, "step": 7886 }, { "epoch": 2.37, "learning_rate": 2.1969050653224022e-05, "loss": 1.9011, "step": 7887 }, { "epoch": 2.37, "learning_rate": 2.1948730663350902e-05, "loss": 1.7906, "step": 7888 }, { "epoch": 2.37, "learning_rate": 2.1928418916686077e-05, "loss": 1.6893, "step": 7889 }, { "epoch": 2.38, "learning_rate": 2.190811541537474e-05, "loss": 1.8199, "step": 7890 }, { "epoch": 2.38, "learning_rate": 2.1887820161561234e-05, "loss": 1.7864, "step": 7891 }, { "epoch": 2.38, "learning_rate": 2.186753315738893e-05, "loss": 1.7878, "step": 7892 }, { "epoch": 2.38, "learning_rate": 2.184725440500043e-05, "loss": 1.7708, "step": 7893 }, { "epoch": 2.38, "learning_rate": 2.182698390653741e-05, "loss": 1.8153, "step": 7894 }, { "epoch": 2.38, "learning_rate": 2.180672166414075e-05, "loss": 1.8041, "step": 7895 }, { "epoch": 2.38, "learning_rate": 2.1786467679950317e-05, "loss": 1.9281, "step": 7896 }, { "epoch": 2.38, "learning_rate": 2.1766221956105237e-05, "loss": 1.8341, "step": 7897 }, { "epoch": 2.38, "learning_rate": 2.174598449474371e-05, "loss": 1.7465, "step": 7898 }, { "epoch": 2.38, "learning_rate": 2.172575529800306e-05, "loss": 1.8737, "step": 7899 }, { "epoch": 2.38, "learning_rate": 2.1705534368019797e-05, "loss": 1.8477, "step": 7900 }, { "epoch": 2.38, "learning_rate": 2.168532170692943e-05, "loss": 1.7996, "step": 7901 }, { "epoch": 2.38, "learning_rate": 2.166511731686671e-05, "loss": 1.7896, "step": 7902 }, { "epoch": 2.38, "learning_rate": 2.1644921199965496e-05, "loss": 1.819, "step": 7903 }, { "epoch": 2.38, "learning_rate": 2.162473335835873e-05, "loss": 1.7558, "step": 7904 }, { "epoch": 2.38, "learning_rate": 2.160455379417854e-05, "loss": 1.774, "step": 7905 }, { "epoch": 2.38, "learning_rate": 2.158438250955609e-05, "loss": 1.8425, "step": 7906 }, { "epoch": 2.38, "learning_rate": 2.1564219506621754e-05, "loss": 1.8358, "step": 7907 }, { "epoch": 2.38, "learning_rate": 2.1544064787505014e-05, "loss": 1.9494, "step": 7908 }, { "epoch": 2.38, "learning_rate": 2.152391835433444e-05, "loss": 1.8444, "step": 7909 }, { "epoch": 2.38, "learning_rate": 2.15037802092378e-05, "loss": 1.8898, "step": 7910 }, { "epoch": 2.38, "learning_rate": 2.1483650354341867e-05, "loss": 1.8134, "step": 7911 }, { "epoch": 2.38, "learning_rate": 2.146352879177266e-05, "loss": 1.7421, "step": 7912 }, { "epoch": 2.38, "learning_rate": 2.1443415523655253e-05, "loss": 1.7608, "step": 7913 }, { "epoch": 2.38, "learning_rate": 2.1423310552113862e-05, "loss": 1.9376, "step": 7914 }, { "epoch": 2.38, "learning_rate": 2.1403213879271876e-05, "loss": 1.7828, "step": 7915 }, { "epoch": 2.38, "learning_rate": 2.1383125507251688e-05, "loss": 1.8437, "step": 7916 }, { "epoch": 2.38, "learning_rate": 2.1363045438174912e-05, "loss": 1.8799, "step": 7917 }, { "epoch": 2.38, "learning_rate": 2.134297367416227e-05, "loss": 1.758, "step": 7918 }, { "epoch": 2.38, "learning_rate": 2.1322910217333614e-05, "loss": 1.8248, "step": 7919 }, { "epoch": 2.38, "learning_rate": 2.1302855069807847e-05, "loss": 1.8932, "step": 7920 }, { "epoch": 2.38, "learning_rate": 2.128280823370309e-05, "loss": 1.7609, "step": 7921 }, { "epoch": 2.38, "learning_rate": 2.126276971113652e-05, "loss": 1.745, "step": 7922 }, { "epoch": 2.39, "learning_rate": 2.1242739504224486e-05, "loss": 1.8956, "step": 7923 }, { "epoch": 2.39, "learning_rate": 2.1222717615082433e-05, "loss": 1.8533, "step": 7924 }, { "epoch": 2.39, "learning_rate": 2.1202704045824896e-05, "loss": 1.9185, "step": 7925 }, { "epoch": 2.39, "learning_rate": 2.118269879856557e-05, "loss": 1.8762, "step": 7926 }, { "epoch": 2.39, "learning_rate": 2.1162701875417278e-05, "loss": 1.8563, "step": 7927 }, { "epoch": 2.39, "learning_rate": 2.1142713278491953e-05, "loss": 1.8533, "step": 7928 }, { "epoch": 2.39, "learning_rate": 2.1122733009900664e-05, "loss": 1.7682, "step": 7929 }, { "epoch": 2.39, "learning_rate": 2.110276107175353e-05, "loss": 1.8794, "step": 7930 }, { "epoch": 2.39, "learning_rate": 2.1082797466159876e-05, "loss": 1.8469, "step": 7931 }, { "epoch": 2.39, "learning_rate": 2.1062842195228093e-05, "loss": 1.8334, "step": 7932 }, { "epoch": 2.39, "learning_rate": 2.104289526106573e-05, "loss": 1.7876, "step": 7933 }, { "epoch": 2.39, "learning_rate": 2.102295666577947e-05, "loss": 1.9472, "step": 7934 }, { "epoch": 2.39, "learning_rate": 2.1003026411475002e-05, "loss": 1.8899, "step": 7935 }, { "epoch": 2.39, "learning_rate": 2.0983104500257266e-05, "loss": 1.7942, "step": 7936 }, { "epoch": 2.39, "learning_rate": 2.0963190934230258e-05, "loss": 1.8304, "step": 7937 }, { "epoch": 2.39, "learning_rate": 2.0943285715497108e-05, "loss": 1.7938, "step": 7938 }, { "epoch": 2.39, "learning_rate": 2.0923388846160098e-05, "loss": 1.7972, "step": 7939 }, { "epoch": 2.39, "learning_rate": 2.0903500328320515e-05, "loss": 1.6361, "step": 7940 }, { "epoch": 2.39, "learning_rate": 2.0883620164078877e-05, "loss": 1.8382, "step": 7941 }, { "epoch": 2.39, "learning_rate": 2.0863748355534784e-05, "loss": 1.7764, "step": 7942 }, { "epoch": 2.39, "learning_rate": 2.0843884904786982e-05, "loss": 1.9489, "step": 7943 }, { "epoch": 2.39, "learning_rate": 2.0824029813933243e-05, "loss": 1.8162, "step": 7944 }, { "epoch": 2.39, "learning_rate": 2.080418308507056e-05, "loss": 1.8984, "step": 7945 }, { "epoch": 2.39, "learning_rate": 2.078434472029498e-05, "loss": 1.7552, "step": 7946 }, { "epoch": 2.39, "learning_rate": 2.076451472170169e-05, "loss": 1.7364, "step": 7947 }, { "epoch": 2.39, "learning_rate": 2.0744693091385035e-05, "loss": 1.7429, "step": 7948 }, { "epoch": 2.39, "learning_rate": 2.072487983143836e-05, "loss": 1.813, "step": 7949 }, { "epoch": 2.39, "learning_rate": 2.070507494395423e-05, "loss": 1.8136, "step": 7950 }, { "epoch": 2.39, "learning_rate": 2.0685278431024302e-05, "loss": 1.7973, "step": 7951 }, { "epoch": 2.39, "learning_rate": 2.066549029473932e-05, "loss": 1.7699, "step": 7952 }, { "epoch": 2.39, "learning_rate": 2.0645710537189212e-05, "loss": 1.8889, "step": 7953 }, { "epoch": 2.39, "learning_rate": 2.0625939160462903e-05, "loss": 1.8819, "step": 7954 }, { "epoch": 2.39, "learning_rate": 2.060617616664854e-05, "loss": 1.834, "step": 7955 }, { "epoch": 2.39, "learning_rate": 2.0586421557833345e-05, "loss": 1.7111, "step": 7956 }, { "epoch": 2.4, "learning_rate": 2.0566675336103648e-05, "loss": 1.8075, "step": 7957 }, { "epoch": 2.4, "learning_rate": 2.054693750354494e-05, "loss": 1.778, "step": 7958 }, { "epoch": 2.4, "learning_rate": 2.052720806224171e-05, "loss": 1.8161, "step": 7959 }, { "epoch": 2.4, "learning_rate": 2.0507487014277703e-05, "loss": 1.7137, "step": 7960 }, { "epoch": 2.4, "learning_rate": 2.048777436173568e-05, "loss": 1.8069, "step": 7961 }, { "epoch": 2.4, "learning_rate": 2.046807010669757e-05, "loss": 1.8964, "step": 7962 }, { "epoch": 2.4, "learning_rate": 2.0448374251244407e-05, "loss": 1.8897, "step": 7963 }, { "epoch": 2.4, "learning_rate": 2.0428686797456266e-05, "loss": 1.8178, "step": 7964 }, { "epoch": 2.4, "learning_rate": 2.0409007747412433e-05, "loss": 1.7726, "step": 7965 }, { "epoch": 2.4, "learning_rate": 2.0389337103191265e-05, "loss": 1.8325, "step": 7966 }, { "epoch": 2.4, "learning_rate": 2.0369674866870257e-05, "loss": 1.7644, "step": 7967 }, { "epoch": 2.4, "learning_rate": 2.0350021040525936e-05, "loss": 1.8077, "step": 7968 }, { "epoch": 2.4, "learning_rate": 2.0330375626234e-05, "loss": 1.7648, "step": 7969 }, { "epoch": 2.4, "learning_rate": 2.0310738626069325e-05, "loss": 1.8419, "step": 7970 }, { "epoch": 2.4, "learning_rate": 2.0291110042105764e-05, "loss": 1.8036, "step": 7971 }, { "epoch": 2.4, "learning_rate": 2.0271489876416382e-05, "loss": 1.728, "step": 7972 }, { "epoch": 2.4, "learning_rate": 2.0251878131073288e-05, "loss": 1.7853, "step": 7973 }, { "epoch": 2.4, "learning_rate": 2.023227480814771e-05, "loss": 1.9287, "step": 7974 }, { "epoch": 2.4, "learning_rate": 2.02126799097101e-05, "loss": 1.9813, "step": 7975 }, { "epoch": 2.4, "learning_rate": 2.0193093437829835e-05, "loss": 1.9185, "step": 7976 }, { "epoch": 2.4, "learning_rate": 2.017351539457557e-05, "loss": 1.8789, "step": 7977 }, { "epoch": 2.4, "learning_rate": 2.0153945782014925e-05, "loss": 1.8826, "step": 7978 }, { "epoch": 2.4, "learning_rate": 2.013438460221474e-05, "loss": 1.8072, "step": 7979 }, { "epoch": 2.4, "learning_rate": 2.0114831857240925e-05, "loss": 1.8037, "step": 7980 }, { "epoch": 2.4, "learning_rate": 2.0095287549158482e-05, "loss": 1.7869, "step": 7981 }, { "epoch": 2.4, "learning_rate": 2.0075751680031584e-05, "loss": 1.8741, "step": 7982 }, { "epoch": 2.4, "learning_rate": 2.0056224251923396e-05, "loss": 1.8972, "step": 7983 }, { "epoch": 2.4, "learning_rate": 2.0036705266896315e-05, "loss": 1.91, "step": 7984 }, { "epoch": 2.4, "learning_rate": 2.001719472701177e-05, "loss": 1.9009, "step": 7985 }, { "epoch": 2.4, "learning_rate": 1.9997692634330335e-05, "loss": 1.8799, "step": 7986 }, { "epoch": 2.4, "learning_rate": 1.997819899091171e-05, "loss": 1.7906, "step": 7987 }, { "epoch": 2.4, "learning_rate": 1.995871379881458e-05, "loss": 1.7173, "step": 7988 }, { "epoch": 2.4, "learning_rate": 1.993923706009694e-05, "loss": 1.7956, "step": 7989 }, { "epoch": 2.41, "learning_rate": 1.9919768776815706e-05, "loss": 1.8135, "step": 7990 }, { "epoch": 2.41, "learning_rate": 1.9900308951027002e-05, "loss": 1.8333, "step": 7991 }, { "epoch": 2.41, "learning_rate": 1.9880857584786062e-05, "loss": 1.8132, "step": 7992 }, { "epoch": 2.41, "learning_rate": 1.9861414680147115e-05, "loss": 1.6838, "step": 7993 }, { "epoch": 2.41, "learning_rate": 1.9841980239163682e-05, "loss": 1.8304, "step": 7994 }, { "epoch": 2.41, "learning_rate": 1.9822554263888215e-05, "loss": 1.7459, "step": 7995 }, { "epoch": 2.41, "learning_rate": 1.9803136756372397e-05, "loss": 1.7851, "step": 7996 }, { "epoch": 2.41, "learning_rate": 1.97837277186669e-05, "loss": 1.7403, "step": 7997 }, { "epoch": 2.41, "learning_rate": 1.9764327152821583e-05, "loss": 1.9276, "step": 7998 }, { "epoch": 2.41, "learning_rate": 1.974493506088546e-05, "loss": 1.8536, "step": 7999 }, { "epoch": 2.41, "learning_rate": 1.972555144490651e-05, "loss": 1.8652, "step": 8000 }, { "epoch": 2.41, "learning_rate": 1.9706176306931945e-05, "loss": 1.8813, "step": 8001 }, { "epoch": 2.41, "learning_rate": 1.968680964900793e-05, "loss": 1.8023, "step": 8002 }, { "epoch": 2.41, "learning_rate": 1.9667451473179933e-05, "loss": 1.6748, "step": 8003 }, { "epoch": 2.41, "learning_rate": 1.96481017814924e-05, "loss": 1.944, "step": 8004 }, { "epoch": 2.41, "learning_rate": 1.962876057598888e-05, "loss": 1.7805, "step": 8005 }, { "epoch": 2.41, "learning_rate": 1.960942785871208e-05, "loss": 1.7191, "step": 8006 }, { "epoch": 2.41, "learning_rate": 1.9590103631703703e-05, "loss": 1.8301, "step": 8007 }, { "epoch": 2.41, "learning_rate": 1.957078789700476e-05, "loss": 1.8078, "step": 8008 }, { "epoch": 2.41, "learning_rate": 1.9551480656655143e-05, "loss": 1.7896, "step": 8009 }, { "epoch": 2.41, "learning_rate": 1.953218191269397e-05, "loss": 1.8446, "step": 8010 }, { "epoch": 2.41, "learning_rate": 1.9512891667159473e-05, "loss": 1.8065, "step": 8011 }, { "epoch": 2.41, "learning_rate": 1.949360992208884e-05, "loss": 1.8124, "step": 8012 }, { "epoch": 2.41, "learning_rate": 1.94743366795186e-05, "loss": 1.8291, "step": 8013 }, { "epoch": 2.41, "learning_rate": 1.9455071941484172e-05, "loss": 1.792, "step": 8014 }, { "epoch": 2.41, "learning_rate": 1.9435815710020176e-05, "loss": 1.8303, "step": 8015 }, { "epoch": 2.41, "learning_rate": 1.9416567987160318e-05, "loss": 1.8839, "step": 8016 }, { "epoch": 2.41, "learning_rate": 1.9397328774937407e-05, "loss": 1.8645, "step": 8017 }, { "epoch": 2.41, "learning_rate": 1.9378098075383365e-05, "loss": 1.8428, "step": 8018 }, { "epoch": 2.41, "learning_rate": 1.935887589052915e-05, "loss": 1.7865, "step": 8019 }, { "epoch": 2.41, "learning_rate": 1.933966222240493e-05, "loss": 1.8831, "step": 8020 }, { "epoch": 2.41, "learning_rate": 1.9320457073039822e-05, "loss": 1.8156, "step": 8021 }, { "epoch": 2.41, "learning_rate": 1.930126044446222e-05, "loss": 1.8682, "step": 8022 }, { "epoch": 2.42, "learning_rate": 1.928207233869954e-05, "loss": 1.9047, "step": 8023 }, { "epoch": 2.42, "learning_rate": 1.926289275777823e-05, "loss": 1.8663, "step": 8024 }, { "epoch": 2.42, "learning_rate": 1.924372170372394e-05, "loss": 1.8708, "step": 8025 }, { "epoch": 2.42, "learning_rate": 1.9224559178561318e-05, "loss": 1.834, "step": 8026 }, { "epoch": 2.42, "learning_rate": 1.9205405184314242e-05, "loss": 1.663, "step": 8027 }, { "epoch": 2.42, "learning_rate": 1.9186259723005617e-05, "loss": 1.8075, "step": 8028 }, { "epoch": 2.42, "learning_rate": 1.916712279665739e-05, "loss": 1.8695, "step": 8029 }, { "epoch": 2.42, "learning_rate": 1.9147994407290694e-05, "loss": 1.7428, "step": 8030 }, { "epoch": 2.42, "learning_rate": 1.9128874556925736e-05, "loss": 1.8872, "step": 8031 }, { "epoch": 2.42, "learning_rate": 1.9109763247581835e-05, "loss": 1.87, "step": 8032 }, { "epoch": 2.42, "learning_rate": 1.9090660481277345e-05, "loss": 1.9076, "step": 8033 }, { "epoch": 2.42, "learning_rate": 1.9071566260029773e-05, "loss": 1.8485, "step": 8034 }, { "epoch": 2.42, "learning_rate": 1.9052480585855725e-05, "loss": 1.7425, "step": 8035 }, { "epoch": 2.42, "learning_rate": 1.9033403460770895e-05, "loss": 1.7638, "step": 8036 }, { "epoch": 2.42, "learning_rate": 1.9014334886790087e-05, "loss": 1.7358, "step": 8037 }, { "epoch": 2.42, "learning_rate": 1.8995274865927125e-05, "loss": 1.8059, "step": 8038 }, { "epoch": 2.42, "learning_rate": 1.8976223400195037e-05, "loss": 1.8995, "step": 8039 }, { "epoch": 2.42, "learning_rate": 1.8957180491605875e-05, "loss": 1.8971, "step": 8040 }, { "epoch": 2.42, "learning_rate": 1.893814614217084e-05, "loss": 1.8925, "step": 8041 }, { "epoch": 2.42, "learning_rate": 1.891912035390021e-05, "loss": 1.8375, "step": 8042 }, { "epoch": 2.42, "learning_rate": 1.8900103128803303e-05, "loss": 1.8044, "step": 8043 }, { "epoch": 2.42, "learning_rate": 1.8881094468888605e-05, "loss": 1.82, "step": 8044 }, { "epoch": 2.42, "learning_rate": 1.8862094376163687e-05, "loss": 1.8003, "step": 8045 }, { "epoch": 2.42, "learning_rate": 1.8843102852635174e-05, "loss": 1.839, "step": 8046 }, { "epoch": 2.42, "learning_rate": 1.8824119900308857e-05, "loss": 1.8937, "step": 8047 }, { "epoch": 2.42, "learning_rate": 1.8805145521189527e-05, "loss": 1.8417, "step": 8048 }, { "epoch": 2.42, "learning_rate": 1.8786179717281126e-05, "loss": 1.7665, "step": 8049 }, { "epoch": 2.42, "learning_rate": 1.876722249058671e-05, "loss": 1.8332, "step": 8050 }, { "epoch": 2.42, "learning_rate": 1.87482738431084e-05, "loss": 1.7756, "step": 8051 }, { "epoch": 2.42, "learning_rate": 1.8729333776847425e-05, "loss": 1.8029, "step": 8052 }, { "epoch": 2.42, "learning_rate": 1.8710402293804052e-05, "loss": 1.7739, "step": 8053 }, { "epoch": 2.42, "learning_rate": 1.869147939597772e-05, "loss": 1.8642, "step": 8054 }, { "epoch": 2.42, "learning_rate": 1.8672565085366924e-05, "loss": 1.8484, "step": 8055 }, { "epoch": 2.43, "learning_rate": 1.8653659363969266e-05, "loss": 1.7452, "step": 8056 }, { "epoch": 2.43, "learning_rate": 1.863476223378141e-05, "loss": 1.7616, "step": 8057 }, { "epoch": 2.43, "learning_rate": 1.8615873696799137e-05, "loss": 1.8964, "step": 8058 }, { "epoch": 2.43, "learning_rate": 1.859699375501732e-05, "loss": 1.7644, "step": 8059 }, { "epoch": 2.43, "learning_rate": 1.857812241042993e-05, "loss": 1.8199, "step": 8060 }, { "epoch": 2.43, "learning_rate": 1.8559259665030038e-05, "loss": 1.7844, "step": 8061 }, { "epoch": 2.43, "learning_rate": 1.8540405520809744e-05, "loss": 1.8402, "step": 8062 }, { "epoch": 2.43, "learning_rate": 1.8521559979760317e-05, "loss": 1.798, "step": 8063 }, { "epoch": 2.43, "learning_rate": 1.8502723043872073e-05, "loss": 1.9228, "step": 8064 }, { "epoch": 2.43, "learning_rate": 1.848389471513444e-05, "loss": 1.7643, "step": 8065 }, { "epoch": 2.43, "learning_rate": 1.8465074995535945e-05, "loss": 1.8836, "step": 8066 }, { "epoch": 2.43, "learning_rate": 1.8446263887064163e-05, "loss": 1.825, "step": 8067 }, { "epoch": 2.43, "learning_rate": 1.842746139170579e-05, "loss": 1.8407, "step": 8068 }, { "epoch": 2.43, "learning_rate": 1.840866751144662e-05, "loss": 1.8978, "step": 8069 }, { "epoch": 2.43, "learning_rate": 1.838988224827152e-05, "loss": 1.9036, "step": 8070 }, { "epoch": 2.43, "learning_rate": 1.8371105604164472e-05, "loss": 1.8616, "step": 8071 }, { "epoch": 2.43, "learning_rate": 1.8352337581108493e-05, "loss": 1.7706, "step": 8072 }, { "epoch": 2.43, "learning_rate": 1.833357818108574e-05, "loss": 1.8229, "step": 8073 }, { "epoch": 2.43, "learning_rate": 1.831482740607746e-05, "loss": 1.8334, "step": 8074 }, { "epoch": 2.43, "learning_rate": 1.8296085258063954e-05, "loss": 1.7783, "step": 8075 }, { "epoch": 2.43, "learning_rate": 1.8277351739024674e-05, "loss": 1.8218, "step": 8076 }, { "epoch": 2.43, "learning_rate": 1.8258626850938053e-05, "loss": 1.7912, "step": 8077 }, { "epoch": 2.43, "learning_rate": 1.8239910595781706e-05, "loss": 1.8045, "step": 8078 }, { "epoch": 2.43, "learning_rate": 1.8221202975532314e-05, "loss": 1.7732, "step": 8079 }, { "epoch": 2.43, "learning_rate": 1.8202503992165665e-05, "loss": 1.8266, "step": 8080 }, { "epoch": 2.43, "learning_rate": 1.8183813647656556e-05, "loss": 1.8342, "step": 8081 }, { "epoch": 2.43, "learning_rate": 1.8165131943978953e-05, "loss": 1.8956, "step": 8082 }, { "epoch": 2.43, "learning_rate": 1.8146458883105876e-05, "loss": 1.7767, "step": 8083 }, { "epoch": 2.43, "learning_rate": 1.812779446700944e-05, "loss": 1.7547, "step": 8084 }, { "epoch": 2.43, "learning_rate": 1.810913869766089e-05, "loss": 1.9028, "step": 8085 }, { "epoch": 2.43, "learning_rate": 1.8090491577030422e-05, "loss": 1.8209, "step": 8086 }, { "epoch": 2.43, "learning_rate": 1.807185310708748e-05, "loss": 1.8735, "step": 8087 }, { "epoch": 2.43, "learning_rate": 1.805322328980048e-05, "loss": 1.8158, "step": 8088 }, { "epoch": 2.43, "learning_rate": 1.8034602127137e-05, "loss": 1.833, "step": 8089 }, { "epoch": 2.44, "learning_rate": 1.8015989621063688e-05, "loss": 1.8201, "step": 8090 }, { "epoch": 2.44, "learning_rate": 1.7997385773546216e-05, "loss": 1.8904, "step": 8091 }, { "epoch": 2.44, "learning_rate": 1.79787905865494e-05, "loss": 1.7584, "step": 8092 }, { "epoch": 2.44, "learning_rate": 1.7960204062037132e-05, "loss": 1.8332, "step": 8093 }, { "epoch": 2.44, "learning_rate": 1.7941626201972396e-05, "loss": 1.7844, "step": 8094 }, { "epoch": 2.44, "learning_rate": 1.7923057008317267e-05, "loss": 1.7808, "step": 8095 }, { "epoch": 2.44, "learning_rate": 1.790449648303284e-05, "loss": 1.7488, "step": 8096 }, { "epoch": 2.44, "learning_rate": 1.7885944628079367e-05, "loss": 1.8047, "step": 8097 }, { "epoch": 2.44, "learning_rate": 1.7867401445416165e-05, "loss": 1.8838, "step": 8098 }, { "epoch": 2.44, "learning_rate": 1.7848866937001617e-05, "loss": 1.7584, "step": 8099 }, { "epoch": 2.44, "learning_rate": 1.7830341104793246e-05, "loss": 1.7334, "step": 8100 }, { "epoch": 2.44, "learning_rate": 1.7811823950747565e-05, "loss": 1.9015, "step": 8101 }, { "epoch": 2.44, "learning_rate": 1.7793315476820225e-05, "loss": 1.8422, "step": 8102 }, { "epoch": 2.44, "learning_rate": 1.7774815684965972e-05, "loss": 1.8342, "step": 8103 }, { "epoch": 2.44, "learning_rate": 1.7756324577138628e-05, "loss": 1.9175, "step": 8104 }, { "epoch": 2.44, "learning_rate": 1.7737842155291107e-05, "loss": 1.822, "step": 8105 }, { "epoch": 2.44, "learning_rate": 1.7719368421375328e-05, "loss": 1.8502, "step": 8106 }, { "epoch": 2.44, "learning_rate": 1.770090337734238e-05, "loss": 1.8177, "step": 8107 }, { "epoch": 2.44, "learning_rate": 1.7682447025142422e-05, "loss": 1.7271, "step": 8108 }, { "epoch": 2.44, "learning_rate": 1.7663999366724693e-05, "loss": 1.6697, "step": 8109 }, { "epoch": 2.44, "learning_rate": 1.7645560404037453e-05, "loss": 1.8649, "step": 8110 }, { "epoch": 2.44, "learning_rate": 1.7627130139028124e-05, "loss": 1.7649, "step": 8111 }, { "epoch": 2.44, "learning_rate": 1.760870857364316e-05, "loss": 1.7687, "step": 8112 }, { "epoch": 2.44, "learning_rate": 1.759029570982812e-05, "loss": 1.8278, "step": 8113 }, { "epoch": 2.44, "learning_rate": 1.7571891549527674e-05, "loss": 1.7503, "step": 8114 }, { "epoch": 2.44, "learning_rate": 1.7553496094685473e-05, "loss": 1.92, "step": 8115 }, { "epoch": 2.44, "learning_rate": 1.7535109347244337e-05, "loss": 1.8572, "step": 8116 }, { "epoch": 2.44, "learning_rate": 1.751673130914614e-05, "loss": 1.8798, "step": 8117 }, { "epoch": 2.44, "learning_rate": 1.7498361982331846e-05, "loss": 1.8153, "step": 8118 }, { "epoch": 2.44, "learning_rate": 1.7480001368741504e-05, "loss": 1.7757, "step": 8119 }, { "epoch": 2.44, "learning_rate": 1.746164947031419e-05, "loss": 1.7315, "step": 8120 }, { "epoch": 2.44, "learning_rate": 1.7443306288988115e-05, "loss": 1.9194, "step": 8121 }, { "epoch": 2.44, "learning_rate": 1.7424971826700552e-05, "loss": 1.8298, "step": 8122 }, { "epoch": 2.45, "learning_rate": 1.7406646085387866e-05, "loss": 1.7981, "step": 8123 }, { "epoch": 2.45, "learning_rate": 1.7388329066985497e-05, "loss": 1.8981, "step": 8124 }, { "epoch": 2.45, "learning_rate": 1.7370020773427908e-05, "loss": 1.7447, "step": 8125 }, { "epoch": 2.45, "learning_rate": 1.735172120664872e-05, "loss": 1.8269, "step": 8126 }, { "epoch": 2.45, "learning_rate": 1.7333430368580605e-05, "loss": 1.8535, "step": 8127 }, { "epoch": 2.45, "learning_rate": 1.7315148261155302e-05, "loss": 1.8283, "step": 8128 }, { "epoch": 2.45, "learning_rate": 1.729687488630365e-05, "loss": 1.7962, "step": 8129 }, { "epoch": 2.45, "learning_rate": 1.7278610245955506e-05, "loss": 1.7087, "step": 8130 }, { "epoch": 2.45, "learning_rate": 1.7260354342039886e-05, "loss": 1.8298, "step": 8131 }, { "epoch": 2.45, "learning_rate": 1.7242107176484835e-05, "loss": 1.7794, "step": 8132 }, { "epoch": 2.45, "learning_rate": 1.722386875121751e-05, "loss": 1.7727, "step": 8133 }, { "epoch": 2.45, "learning_rate": 1.7205639068164082e-05, "loss": 1.8923, "step": 8134 }, { "epoch": 2.45, "learning_rate": 1.718741812924982e-05, "loss": 1.8089, "step": 8135 }, { "epoch": 2.45, "learning_rate": 1.716920593639918e-05, "loss": 1.7135, "step": 8136 }, { "epoch": 2.45, "learning_rate": 1.7151002491535507e-05, "loss": 1.8336, "step": 8137 }, { "epoch": 2.45, "learning_rate": 1.7132807796581384e-05, "loss": 1.8003, "step": 8138 }, { "epoch": 2.45, "learning_rate": 1.7114621853458357e-05, "loss": 1.8661, "step": 8139 }, { "epoch": 2.45, "learning_rate": 1.7096444664087073e-05, "loss": 1.822, "step": 8140 }, { "epoch": 2.45, "learning_rate": 1.7078276230387357e-05, "loss": 1.7876, "step": 8141 }, { "epoch": 2.45, "learning_rate": 1.706011655427796e-05, "loss": 1.8652, "step": 8142 }, { "epoch": 2.45, "learning_rate": 1.7041965637676816e-05, "loss": 1.7463, "step": 8143 }, { "epoch": 2.45, "learning_rate": 1.7023823482500857e-05, "loss": 1.9626, "step": 8144 }, { "epoch": 2.45, "learning_rate": 1.7005690090666127e-05, "loss": 1.8054, "step": 8145 }, { "epoch": 2.45, "learning_rate": 1.6987565464087763e-05, "loss": 1.7553, "step": 8146 }, { "epoch": 2.45, "learning_rate": 1.6969449604679956e-05, "loss": 1.8635, "step": 8147 }, { "epoch": 2.45, "learning_rate": 1.6951342514355983e-05, "loss": 1.8378, "step": 8148 }, { "epoch": 2.45, "learning_rate": 1.693324419502812e-05, "loss": 1.8258, "step": 8149 }, { "epoch": 2.45, "learning_rate": 1.6915154648607867e-05, "loss": 1.7837, "step": 8150 }, { "epoch": 2.45, "learning_rate": 1.6897073877005653e-05, "loss": 1.7898, "step": 8151 }, { "epoch": 2.45, "learning_rate": 1.687900188213105e-05, "loss": 1.8503, "step": 8152 }, { "epoch": 2.45, "learning_rate": 1.6860938665892733e-05, "loss": 1.8452, "step": 8153 }, { "epoch": 2.45, "learning_rate": 1.6842884230198307e-05, "loss": 1.7611, "step": 8154 }, { "epoch": 2.45, "learning_rate": 1.6824838576954672e-05, "loss": 1.7475, "step": 8155 }, { "epoch": 2.46, "learning_rate": 1.6806801708067586e-05, "loss": 1.9294, "step": 8156 }, { "epoch": 2.46, "learning_rate": 1.6788773625442044e-05, "loss": 1.8071, "step": 8157 }, { "epoch": 2.46, "learning_rate": 1.6770754330981976e-05, "loss": 1.8435, "step": 8158 }, { "epoch": 2.46, "learning_rate": 1.6752743826590445e-05, "loss": 1.8156, "step": 8159 }, { "epoch": 2.46, "learning_rate": 1.6734742114169676e-05, "loss": 1.9064, "step": 8160 }, { "epoch": 2.46, "learning_rate": 1.6716749195620795e-05, "loss": 1.8166, "step": 8161 }, { "epoch": 2.46, "learning_rate": 1.6698765072844137e-05, "loss": 1.879, "step": 8162 }, { "epoch": 2.46, "learning_rate": 1.668078974773899e-05, "loss": 1.795, "step": 8163 }, { "epoch": 2.46, "learning_rate": 1.666282322220384e-05, "loss": 1.87, "step": 8164 }, { "epoch": 2.46, "learning_rate": 1.6644865498136175e-05, "loss": 1.7778, "step": 8165 }, { "epoch": 2.46, "learning_rate": 1.6626916577432527e-05, "loss": 1.8216, "step": 8166 }, { "epoch": 2.46, "learning_rate": 1.660897646198857e-05, "loss": 1.7571, "step": 8167 }, { "epoch": 2.46, "learning_rate": 1.6591045153698937e-05, "loss": 1.7592, "step": 8168 }, { "epoch": 2.46, "learning_rate": 1.6573122654457495e-05, "loss": 1.7703, "step": 8169 }, { "epoch": 2.46, "learning_rate": 1.655520896615702e-05, "loss": 1.9066, "step": 8170 }, { "epoch": 2.46, "learning_rate": 1.6537304090689455e-05, "loss": 1.7671, "step": 8171 }, { "epoch": 2.46, "learning_rate": 1.6519408029945805e-05, "loss": 1.8467, "step": 8172 }, { "epoch": 2.46, "learning_rate": 1.650152078581604e-05, "loss": 1.7303, "step": 8173 }, { "epoch": 2.46, "learning_rate": 1.6483642360189388e-05, "loss": 1.8718, "step": 8174 }, { "epoch": 2.46, "learning_rate": 1.6465772754953955e-05, "loss": 1.7764, "step": 8175 }, { "epoch": 2.46, "learning_rate": 1.644791197199704e-05, "loss": 1.8836, "step": 8176 }, { "epoch": 2.46, "learning_rate": 1.643006001320495e-05, "loss": 1.8701, "step": 8177 }, { "epoch": 2.46, "learning_rate": 1.6412216880463094e-05, "loss": 1.7739, "step": 8178 }, { "epoch": 2.46, "learning_rate": 1.6394382575655952e-05, "loss": 1.8006, "step": 8179 }, { "epoch": 2.46, "learning_rate": 1.637655710066701e-05, "loss": 1.7745, "step": 8180 }, { "epoch": 2.46, "learning_rate": 1.635874045737891e-05, "loss": 1.8046, "step": 8181 }, { "epoch": 2.46, "learning_rate": 1.6340932647673246e-05, "loss": 1.8472, "step": 8182 }, { "epoch": 2.46, "learning_rate": 1.6323133673430814e-05, "loss": 1.8728, "step": 8183 }, { "epoch": 2.46, "learning_rate": 1.6305343536531413e-05, "loss": 1.8981, "step": 8184 }, { "epoch": 2.46, "learning_rate": 1.628756223885388e-05, "loss": 1.8592, "step": 8185 }, { "epoch": 2.46, "learning_rate": 1.626978978227617e-05, "loss": 1.7436, "step": 8186 }, { "epoch": 2.46, "learning_rate": 1.6252026168675218e-05, "loss": 1.7086, "step": 8187 }, { "epoch": 2.46, "learning_rate": 1.623427139992717e-05, "loss": 1.9209, "step": 8188 }, { "epoch": 2.47, "learning_rate": 1.621652547790714e-05, "loss": 1.8242, "step": 8189 }, { "epoch": 2.47, "learning_rate": 1.6198788404489273e-05, "loss": 1.7232, "step": 8190 }, { "epoch": 2.47, "learning_rate": 1.6181060181546903e-05, "loss": 1.8003, "step": 8191 }, { "epoch": 2.47, "learning_rate": 1.6163340810952255e-05, "loss": 1.8447, "step": 8192 }, { "epoch": 2.47, "learning_rate": 1.614563029457684e-05, "loss": 1.819, "step": 8193 }, { "epoch": 2.47, "learning_rate": 1.6127928634291022e-05, "loss": 1.8222, "step": 8194 }, { "epoch": 2.47, "learning_rate": 1.611023583196436e-05, "loss": 1.8501, "step": 8195 }, { "epoch": 2.47, "learning_rate": 1.6092551889465434e-05, "loss": 1.8478, "step": 8196 }, { "epoch": 2.47, "learning_rate": 1.6074876808661886e-05, "loss": 1.8178, "step": 8197 }, { "epoch": 2.47, "learning_rate": 1.605721059142048e-05, "loss": 1.815, "step": 8198 }, { "epoch": 2.47, "learning_rate": 1.6039553239606908e-05, "loss": 1.7048, "step": 8199 }, { "epoch": 2.47, "learning_rate": 1.602190475508607e-05, "loss": 1.8441, "step": 8200 }, { "epoch": 2.47, "learning_rate": 1.600426513972184e-05, "loss": 1.8145, "step": 8201 }, { "epoch": 2.47, "learning_rate": 1.598663439537721e-05, "loss": 1.778, "step": 8202 }, { "epoch": 2.47, "learning_rate": 1.5969012523914227e-05, "loss": 1.833, "step": 8203 }, { "epoch": 2.47, "learning_rate": 1.5951399527193945e-05, "loss": 1.7755, "step": 8204 }, { "epoch": 2.47, "learning_rate": 1.5933795407076546e-05, "loss": 1.8058, "step": 8205 }, { "epoch": 2.47, "learning_rate": 1.591620016542119e-05, "loss": 1.6877, "step": 8206 }, { "epoch": 2.47, "learning_rate": 1.589861380408625e-05, "loss": 1.9138, "step": 8207 }, { "epoch": 2.47, "learning_rate": 1.5881036324929033e-05, "loss": 1.8766, "step": 8208 }, { "epoch": 2.47, "learning_rate": 1.5863467729805926e-05, "loss": 1.8431, "step": 8209 }, { "epoch": 2.47, "learning_rate": 1.584590802057241e-05, "loss": 1.8267, "step": 8210 }, { "epoch": 2.47, "learning_rate": 1.582835719908302e-05, "loss": 1.8739, "step": 8211 }, { "epoch": 2.47, "learning_rate": 1.5810815267191338e-05, "loss": 1.8545, "step": 8212 }, { "epoch": 2.47, "learning_rate": 1.579328222675003e-05, "loss": 2.0315, "step": 8213 }, { "epoch": 2.47, "learning_rate": 1.577575807961078e-05, "loss": 1.7768, "step": 8214 }, { "epoch": 2.47, "learning_rate": 1.5758242827624382e-05, "loss": 2.0826, "step": 8215 }, { "epoch": 2.47, "learning_rate": 1.574073647264066e-05, "loss": 1.884, "step": 8216 }, { "epoch": 2.47, "learning_rate": 1.572323901650854e-05, "loss": 1.7644, "step": 8217 }, { "epoch": 2.47, "learning_rate": 1.5705750461075928e-05, "loss": 1.9051, "step": 8218 }, { "epoch": 2.47, "learning_rate": 1.5688270808189864e-05, "loss": 1.7602, "step": 8219 }, { "epoch": 2.47, "learning_rate": 1.567080005969641e-05, "loss": 1.71, "step": 8220 }, { "epoch": 2.47, "learning_rate": 1.5653338217440715e-05, "loss": 1.7626, "step": 8221 }, { "epoch": 2.48, "learning_rate": 1.5635885283266984e-05, "loss": 1.7975, "step": 8222 }, { "epoch": 2.48, "learning_rate": 1.5618441259018425e-05, "loss": 1.8797, "step": 8223 }, { "epoch": 2.48, "learning_rate": 1.560100614653738e-05, "loss": 1.8378, "step": 8224 }, { "epoch": 2.48, "learning_rate": 1.5583579947665205e-05, "loss": 1.8807, "step": 8225 }, { "epoch": 2.48, "learning_rate": 1.5566162664242336e-05, "loss": 1.8285, "step": 8226 }, { "epoch": 2.48, "learning_rate": 1.5548754298108292e-05, "loss": 1.805, "step": 8227 }, { "epoch": 2.48, "learning_rate": 1.5531354851101554e-05, "loss": 1.9301, "step": 8228 }, { "epoch": 2.48, "learning_rate": 1.5513964325059772e-05, "loss": 1.84, "step": 8229 }, { "epoch": 2.48, "learning_rate": 1.5496582721819574e-05, "loss": 1.7603, "step": 8230 }, { "epoch": 2.48, "learning_rate": 1.547921004321672e-05, "loss": 1.8654, "step": 8231 }, { "epoch": 2.48, "learning_rate": 1.546184629108597e-05, "loss": 1.8797, "step": 8232 }, { "epoch": 2.48, "learning_rate": 1.5444491467261134e-05, "loss": 1.8354, "step": 8233 }, { "epoch": 2.48, "learning_rate": 1.542714557357512e-05, "loss": 1.7255, "step": 8234 }, { "epoch": 2.48, "learning_rate": 1.540980861185988e-05, "loss": 1.8521, "step": 8235 }, { "epoch": 2.48, "learning_rate": 1.5392480583946402e-05, "loss": 1.7172, "step": 8236 }, { "epoch": 2.48, "learning_rate": 1.5375161491664792e-05, "loss": 1.8961, "step": 8237 }, { "epoch": 2.48, "learning_rate": 1.535785133684411e-05, "loss": 1.7295, "step": 8238 }, { "epoch": 2.48, "learning_rate": 1.5340550121312536e-05, "loss": 1.7567, "step": 8239 }, { "epoch": 2.48, "learning_rate": 1.5323257846897332e-05, "loss": 1.8639, "step": 8240 }, { "epoch": 2.48, "learning_rate": 1.530597451542476e-05, "loss": 1.8281, "step": 8241 }, { "epoch": 2.48, "learning_rate": 1.528870012872018e-05, "loss": 1.8362, "step": 8242 }, { "epoch": 2.48, "learning_rate": 1.5271434688607954e-05, "loss": 1.7389, "step": 8243 }, { "epoch": 2.48, "learning_rate": 1.525417819691155e-05, "loss": 1.7971, "step": 8244 }, { "epoch": 2.48, "learning_rate": 1.5236930655453475e-05, "loss": 1.7745, "step": 8245 }, { "epoch": 2.48, "learning_rate": 1.5219692066055313e-05, "loss": 1.7946, "step": 8246 }, { "epoch": 2.48, "learning_rate": 1.5202462430537623e-05, "loss": 1.7567, "step": 8247 }, { "epoch": 2.48, "learning_rate": 1.5185241750720114e-05, "loss": 1.8182, "step": 8248 }, { "epoch": 2.48, "learning_rate": 1.5168030028421487e-05, "loss": 1.9119, "step": 8249 }, { "epoch": 2.48, "learning_rate": 1.5150827265459544e-05, "loss": 1.7459, "step": 8250 }, { "epoch": 2.48, "learning_rate": 1.513363346365112e-05, "loss": 1.8318, "step": 8251 }, { "epoch": 2.48, "learning_rate": 1.5116448624812062e-05, "loss": 1.7861, "step": 8252 }, { "epoch": 2.48, "learning_rate": 1.509927275075732e-05, "loss": 1.8067, "step": 8253 }, { "epoch": 2.48, "learning_rate": 1.508210584330091e-05, "loss": 1.8591, "step": 8254 }, { "epoch": 2.48, "learning_rate": 1.5064947904255854e-05, "loss": 1.8624, "step": 8255 }, { "epoch": 2.49, "learning_rate": 1.5047798935434276e-05, "loss": 1.7986, "step": 8256 }, { "epoch": 2.49, "learning_rate": 1.503065893864728e-05, "loss": 1.7825, "step": 8257 }, { "epoch": 2.49, "learning_rate": 1.5013527915705095e-05, "loss": 1.7768, "step": 8258 }, { "epoch": 2.49, "learning_rate": 1.4996405868416975e-05, "loss": 1.7999, "step": 8259 }, { "epoch": 2.49, "learning_rate": 1.4979292798591217e-05, "loss": 1.7815, "step": 8260 }, { "epoch": 2.49, "learning_rate": 1.4962188708035218e-05, "loss": 1.8717, "step": 8261 }, { "epoch": 2.49, "learning_rate": 1.4945093598555327e-05, "loss": 1.7632, "step": 8262 }, { "epoch": 2.49, "learning_rate": 1.492800747195704e-05, "loss": 1.7071, "step": 8263 }, { "epoch": 2.49, "learning_rate": 1.4910930330044858e-05, "loss": 1.8055, "step": 8264 }, { "epoch": 2.49, "learning_rate": 1.4893862174622353e-05, "loss": 1.7799, "step": 8265 }, { "epoch": 2.49, "learning_rate": 1.4876803007492157e-05, "loss": 1.7169, "step": 8266 }, { "epoch": 2.49, "learning_rate": 1.4859752830455897e-05, "loss": 1.7728, "step": 8267 }, { "epoch": 2.49, "learning_rate": 1.4842711645314312e-05, "loss": 1.8605, "step": 8268 }, { "epoch": 2.49, "learning_rate": 1.4825679453867159e-05, "loss": 1.7453, "step": 8269 }, { "epoch": 2.49, "learning_rate": 1.480865625791329e-05, "loss": 1.8389, "step": 8270 }, { "epoch": 2.49, "learning_rate": 1.479164205925051e-05, "loss": 1.8186, "step": 8271 }, { "epoch": 2.49, "learning_rate": 1.4774636859675762e-05, "loss": 1.8195, "step": 8272 }, { "epoch": 2.49, "learning_rate": 1.4757640660985029e-05, "loss": 1.8402, "step": 8273 }, { "epoch": 2.49, "learning_rate": 1.4740653464973298e-05, "loss": 1.8182, "step": 8274 }, { "epoch": 2.49, "learning_rate": 1.4723675273434678e-05, "loss": 1.8921, "step": 8275 }, { "epoch": 2.49, "learning_rate": 1.4706706088162226e-05, "loss": 1.8837, "step": 8276 }, { "epoch": 2.49, "learning_rate": 1.4689745910948138e-05, "loss": 1.8791, "step": 8277 }, { "epoch": 2.49, "learning_rate": 1.4672794743583607e-05, "loss": 1.7422, "step": 8278 }, { "epoch": 2.49, "learning_rate": 1.4655852587858909e-05, "loss": 1.927, "step": 8279 }, { "epoch": 2.49, "learning_rate": 1.463891944556337e-05, "loss": 1.8236, "step": 8280 }, { "epoch": 2.49, "learning_rate": 1.4621995318485292e-05, "loss": 1.8596, "step": 8281 }, { "epoch": 2.49, "learning_rate": 1.4605080208412115e-05, "loss": 1.9547, "step": 8282 }, { "epoch": 2.49, "learning_rate": 1.4588174117130272e-05, "loss": 1.7275, "step": 8283 }, { "epoch": 2.49, "learning_rate": 1.4571277046425269e-05, "loss": 1.9267, "step": 8284 }, { "epoch": 2.49, "learning_rate": 1.4554388998081691e-05, "loss": 1.9288, "step": 8285 }, { "epoch": 2.49, "learning_rate": 1.4537509973883057e-05, "loss": 1.8385, "step": 8286 }, { "epoch": 2.49, "learning_rate": 1.4520639975612048e-05, "loss": 1.8077, "step": 8287 }, { "epoch": 2.49, "learning_rate": 1.4503779005050355e-05, "loss": 1.9279, "step": 8288 }, { "epoch": 2.5, "learning_rate": 1.4486927063978695e-05, "loss": 1.7446, "step": 8289 }, { "epoch": 2.5, "learning_rate": 1.4470084154176888e-05, "loss": 1.876, "step": 8290 }, { "epoch": 2.5, "learning_rate": 1.4453250277423691e-05, "loss": 1.8535, "step": 8291 }, { "epoch": 2.5, "learning_rate": 1.4436425435497026e-05, "loss": 1.6918, "step": 8292 }, { "epoch": 2.5, "learning_rate": 1.4419609630173791e-05, "loss": 1.8638, "step": 8293 }, { "epoch": 2.5, "learning_rate": 1.440280286322998e-05, "loss": 1.6979, "step": 8294 }, { "epoch": 2.5, "learning_rate": 1.438600513644055e-05, "loss": 1.8155, "step": 8295 }, { "epoch": 2.5, "learning_rate": 1.436921645157957e-05, "loss": 1.8687, "step": 8296 }, { "epoch": 2.5, "learning_rate": 1.4352436810420189e-05, "loss": 1.8695, "step": 8297 }, { "epoch": 2.5, "learning_rate": 1.433566621473449e-05, "loss": 1.8711, "step": 8298 }, { "epoch": 2.5, "learning_rate": 1.431890466629371e-05, "loss": 1.8175, "step": 8299 }, { "epoch": 2.5, "learning_rate": 1.4302152166868032e-05, "loss": 1.8275, "step": 8300 }, { "epoch": 2.5, "learning_rate": 1.4285408718226733e-05, "loss": 1.8681, "step": 8301 }, { "epoch": 2.5, "learning_rate": 1.4268674322138209e-05, "loss": 1.8906, "step": 8302 }, { "epoch": 2.5, "learning_rate": 1.4251948980369756e-05, "loss": 1.8815, "step": 8303 }, { "epoch": 2.5, "learning_rate": 1.4235232694687817e-05, "loss": 1.7278, "step": 8304 }, { "epoch": 2.5, "learning_rate": 1.4218525466857812e-05, "loss": 1.8135, "step": 8305 }, { "epoch": 2.5, "learning_rate": 1.4201827298644255e-05, "loss": 1.8136, "step": 8306 }, { "epoch": 2.5, "learning_rate": 1.4185138191810687e-05, "loss": 1.7846, "step": 8307 }, { "epoch": 2.5, "learning_rate": 1.416845814811969e-05, "loss": 1.7779, "step": 8308 }, { "epoch": 2.5, "learning_rate": 1.4151787169332908e-05, "loss": 1.7365, "step": 8309 }, { "epoch": 2.5, "learning_rate": 1.4135125257210968e-05, "loss": 1.8287, "step": 8310 }, { "epoch": 2.5, "learning_rate": 1.4118472413513606e-05, "loss": 1.744, "step": 8311 }, { "epoch": 2.5, "learning_rate": 1.410182863999956e-05, "loss": 1.8185, "step": 8312 }, { "epoch": 2.5, "learning_rate": 1.4085193938426645e-05, "loss": 1.8038, "step": 8313 }, { "epoch": 2.5, "learning_rate": 1.40685683105517e-05, "loss": 1.8155, "step": 8314 }, { "epoch": 2.5, "learning_rate": 1.4051951758130554e-05, "loss": 1.8411, "step": 8315 }, { "epoch": 2.5, "learning_rate": 1.4035344282918194e-05, "loss": 1.7333, "step": 8316 }, { "epoch": 2.5, "learning_rate": 1.4018745886668538e-05, "loss": 1.8554, "step": 8317 }, { "epoch": 2.5, "learning_rate": 1.4002156571134617e-05, "loss": 1.7383, "step": 8318 }, { "epoch": 2.5, "learning_rate": 1.398557633806844e-05, "loss": 1.8297, "step": 8319 }, { "epoch": 2.5, "learning_rate": 1.3969005189221074e-05, "loss": 1.7095, "step": 8320 }, { "epoch": 2.5, "learning_rate": 1.395244312634273e-05, "loss": 1.7267, "step": 8321 }, { "epoch": 2.51, "learning_rate": 1.3935890151182495e-05, "loss": 1.8312, "step": 8322 }, { "epoch": 2.51, "learning_rate": 1.391934626548862e-05, "loss": 1.8217, "step": 8323 }, { "epoch": 2.51, "learning_rate": 1.3902811471008303e-05, "loss": 1.8762, "step": 8324 }, { "epoch": 2.51, "learning_rate": 1.3886285769487838e-05, "loss": 1.8584, "step": 8325 }, { "epoch": 2.51, "learning_rate": 1.3869769162672607e-05, "loss": 1.8256, "step": 8326 }, { "epoch": 2.51, "learning_rate": 1.3853261652306915e-05, "loss": 1.8284, "step": 8327 }, { "epoch": 2.51, "learning_rate": 1.3836763240134198e-05, "loss": 1.7937, "step": 8328 }, { "epoch": 2.51, "learning_rate": 1.3820273927896843e-05, "loss": 1.8337, "step": 8329 }, { "epoch": 2.51, "learning_rate": 1.3803793717336422e-05, "loss": 1.805, "step": 8330 }, { "epoch": 2.51, "learning_rate": 1.3787322610193377e-05, "loss": 1.7772, "step": 8331 }, { "epoch": 2.51, "learning_rate": 1.3770860608207303e-05, "loss": 1.8043, "step": 8332 }, { "epoch": 2.51, "learning_rate": 1.3754407713116802e-05, "loss": 1.8452, "step": 8333 }, { "epoch": 2.51, "learning_rate": 1.3737963926659459e-05, "loss": 1.8363, "step": 8334 }, { "epoch": 2.51, "learning_rate": 1.3721529250572029e-05, "loss": 1.814, "step": 8335 }, { "epoch": 2.51, "learning_rate": 1.3705103686590159e-05, "loss": 1.842, "step": 8336 }, { "epoch": 2.51, "learning_rate": 1.3688687236448616e-05, "loss": 1.7467, "step": 8337 }, { "epoch": 2.51, "learning_rate": 1.3672279901881202e-05, "loss": 1.7493, "step": 8338 }, { "epoch": 2.51, "learning_rate": 1.3655881684620685e-05, "loss": 1.7355, "step": 8339 }, { "epoch": 2.51, "learning_rate": 1.3639492586399028e-05, "loss": 1.8714, "step": 8340 }, { "epoch": 2.51, "learning_rate": 1.362311260894703e-05, "loss": 1.813, "step": 8341 }, { "epoch": 2.51, "learning_rate": 1.3606741753994667e-05, "loss": 1.8984, "step": 8342 }, { "epoch": 2.51, "learning_rate": 1.359038002327091e-05, "loss": 1.7468, "step": 8343 }, { "epoch": 2.51, "learning_rate": 1.3574027418503754e-05, "loss": 1.8233, "step": 8344 }, { "epoch": 2.51, "learning_rate": 1.3557683941420274e-05, "loss": 1.7699, "step": 8345 }, { "epoch": 2.51, "learning_rate": 1.354134959374651e-05, "loss": 1.843, "step": 8346 }, { "epoch": 2.51, "learning_rate": 1.3525024377207618e-05, "loss": 1.7718, "step": 8347 }, { "epoch": 2.51, "learning_rate": 1.3508708293527672e-05, "loss": 1.7634, "step": 8348 }, { "epoch": 2.51, "learning_rate": 1.3492401344429928e-05, "loss": 1.7971, "step": 8349 }, { "epoch": 2.51, "learning_rate": 1.3476103531636619e-05, "loss": 1.8606, "step": 8350 }, { "epoch": 2.51, "learning_rate": 1.3459814856868958e-05, "loss": 1.7863, "step": 8351 }, { "epoch": 2.51, "learning_rate": 1.3443535321847267e-05, "loss": 1.8629, "step": 8352 }, { "epoch": 2.51, "learning_rate": 1.3427264928290806e-05, "loss": 1.8961, "step": 8353 }, { "epoch": 2.51, "learning_rate": 1.341100367791801e-05, "loss": 1.8023, "step": 8354 }, { "epoch": 2.52, "learning_rate": 1.3394751572446274e-05, "loss": 1.7543, "step": 8355 }, { "epoch": 2.52, "learning_rate": 1.337850861359199e-05, "loss": 1.7855, "step": 8356 }, { "epoch": 2.52, "learning_rate": 1.3362274803070629e-05, "loss": 1.8389, "step": 8357 }, { "epoch": 2.52, "learning_rate": 1.334605014259669e-05, "loss": 1.8104, "step": 8358 }, { "epoch": 2.52, "learning_rate": 1.332983463388373e-05, "loss": 1.7604, "step": 8359 }, { "epoch": 2.52, "learning_rate": 1.3313628278644274e-05, "loss": 1.9181, "step": 8360 }, { "epoch": 2.52, "learning_rate": 1.3297431078589938e-05, "loss": 1.8504, "step": 8361 }, { "epoch": 2.52, "learning_rate": 1.3281243035431346e-05, "loss": 1.7838, "step": 8362 }, { "epoch": 2.52, "learning_rate": 1.326506415087817e-05, "loss": 1.8005, "step": 8363 }, { "epoch": 2.52, "learning_rate": 1.3248894426639137e-05, "loss": 1.9976, "step": 8364 }, { "epoch": 2.52, "learning_rate": 1.3232733864421898e-05, "loss": 1.8282, "step": 8365 }, { "epoch": 2.52, "learning_rate": 1.321658246593328e-05, "loss": 1.8453, "step": 8366 }, { "epoch": 2.52, "learning_rate": 1.3200440232879042e-05, "loss": 1.7654, "step": 8367 }, { "epoch": 2.52, "learning_rate": 1.3184307166964027e-05, "loss": 1.8361, "step": 8368 }, { "epoch": 2.52, "learning_rate": 1.316818326989211e-05, "loss": 1.7686, "step": 8369 }, { "epoch": 2.52, "learning_rate": 1.3152068543366125e-05, "loss": 1.7819, "step": 8370 }, { "epoch": 2.52, "learning_rate": 1.3135962989088035e-05, "loss": 1.8362, "step": 8371 }, { "epoch": 2.52, "learning_rate": 1.311986660875878e-05, "loss": 1.7545, "step": 8372 }, { "epoch": 2.52, "learning_rate": 1.3103779404078354e-05, "loss": 1.8609, "step": 8373 }, { "epoch": 2.52, "learning_rate": 1.3087701376745765e-05, "loss": 1.7166, "step": 8374 }, { "epoch": 2.52, "learning_rate": 1.3071632528459044e-05, "loss": 1.7193, "step": 8375 }, { "epoch": 2.52, "learning_rate": 1.3055572860915278e-05, "loss": 1.8478, "step": 8376 }, { "epoch": 2.52, "learning_rate": 1.3039522375810565e-05, "loss": 1.8418, "step": 8377 }, { "epoch": 2.52, "learning_rate": 1.302348107484005e-05, "loss": 1.7593, "step": 8378 }, { "epoch": 2.52, "learning_rate": 1.300744895969791e-05, "loss": 1.7411, "step": 8379 }, { "epoch": 2.52, "learning_rate": 1.2991426032077302e-05, "loss": 1.7873, "step": 8380 }, { "epoch": 2.52, "learning_rate": 1.2975412293670486e-05, "loss": 1.8312, "step": 8381 }, { "epoch": 2.52, "learning_rate": 1.2959407746168695e-05, "loss": 1.8078, "step": 8382 }, { "epoch": 2.52, "learning_rate": 1.2943412391262243e-05, "loss": 1.7498, "step": 8383 }, { "epoch": 2.52, "learning_rate": 1.2927426230640416e-05, "loss": 1.7651, "step": 8384 }, { "epoch": 2.52, "learning_rate": 1.291144926599155e-05, "loss": 1.7997, "step": 8385 }, { "epoch": 2.52, "learning_rate": 1.2895481499003037e-05, "loss": 1.8484, "step": 8386 }, { "epoch": 2.52, "learning_rate": 1.2879522931361265e-05, "loss": 1.7493, "step": 8387 }, { "epoch": 2.52, "learning_rate": 1.2863573564751697e-05, "loss": 1.7505, "step": 8388 }, { "epoch": 2.53, "learning_rate": 1.2847633400858728e-05, "loss": 1.7582, "step": 8389 }, { "epoch": 2.53, "learning_rate": 1.2831702441365879e-05, "loss": 1.79, "step": 8390 }, { "epoch": 2.53, "learning_rate": 1.2815780687955659e-05, "loss": 1.8721, "step": 8391 }, { "epoch": 2.53, "learning_rate": 1.279986814230959e-05, "loss": 1.8676, "step": 8392 }, { "epoch": 2.53, "learning_rate": 1.2783964806108294e-05, "loss": 1.8356, "step": 8393 }, { "epoch": 2.53, "learning_rate": 1.2768070681031297e-05, "loss": 1.8583, "step": 8394 }, { "epoch": 2.53, "learning_rate": 1.2752185768757252e-05, "loss": 1.9044, "step": 8395 }, { "epoch": 2.53, "learning_rate": 1.2736310070963808e-05, "loss": 1.8048, "step": 8396 }, { "epoch": 2.53, "learning_rate": 1.2720443589327636e-05, "loss": 1.9256, "step": 8397 }, { "epoch": 2.53, "learning_rate": 1.270458632552447e-05, "loss": 1.6733, "step": 8398 }, { "epoch": 2.53, "learning_rate": 1.2688738281228984e-05, "loss": 1.7328, "step": 8399 }, { "epoch": 2.53, "learning_rate": 1.2672899458114962e-05, "loss": 1.7928, "step": 8400 }, { "epoch": 2.53, "learning_rate": 1.2657069857855185e-05, "loss": 1.7588, "step": 8401 }, { "epoch": 2.53, "learning_rate": 1.2641249482121464e-05, "loss": 1.8376, "step": 8402 }, { "epoch": 2.53, "learning_rate": 1.262543833258465e-05, "loss": 1.828, "step": 8403 }, { "epoch": 2.53, "learning_rate": 1.260963641091456e-05, "loss": 1.7549, "step": 8404 }, { "epoch": 2.53, "learning_rate": 1.2593843718780084e-05, "loss": 1.8077, "step": 8405 }, { "epoch": 2.53, "learning_rate": 1.2578060257849157e-05, "loss": 1.8396, "step": 8406 }, { "epoch": 2.53, "learning_rate": 1.2562286029788727e-05, "loss": 1.7798, "step": 8407 }, { "epoch": 2.53, "learning_rate": 1.2546521036264703e-05, "loss": 1.7653, "step": 8408 }, { "epoch": 2.53, "learning_rate": 1.2530765278942091e-05, "loss": 1.9051, "step": 8409 }, { "epoch": 2.53, "learning_rate": 1.2515018759484908e-05, "loss": 1.8457, "step": 8410 }, { "epoch": 2.53, "learning_rate": 1.2499281479556179e-05, "loss": 1.7563, "step": 8411 }, { "epoch": 2.53, "learning_rate": 1.2483553440817986e-05, "loss": 1.8181, "step": 8412 }, { "epoch": 2.53, "learning_rate": 1.2467834644931365e-05, "loss": 1.7314, "step": 8413 }, { "epoch": 2.53, "learning_rate": 1.2452125093556455e-05, "loss": 1.7977, "step": 8414 }, { "epoch": 2.53, "learning_rate": 1.2436424788352364e-05, "loss": 1.8026, "step": 8415 }, { "epoch": 2.53, "learning_rate": 1.2420733730977251e-05, "loss": 1.8431, "step": 8416 }, { "epoch": 2.53, "learning_rate": 1.2405051923088329e-05, "loss": 1.8713, "step": 8417 }, { "epoch": 2.53, "learning_rate": 1.2389379366341736e-05, "loss": 1.9231, "step": 8418 }, { "epoch": 2.53, "learning_rate": 1.2373716062392727e-05, "loss": 1.8335, "step": 8419 }, { "epoch": 2.53, "learning_rate": 1.2358062012895533e-05, "loss": 1.7754, "step": 8420 }, { "epoch": 2.53, "learning_rate": 1.2342417219503432e-05, "loss": 1.8429, "step": 8421 }, { "epoch": 2.54, "learning_rate": 1.2326781683868726e-05, "loss": 1.8256, "step": 8422 }, { "epoch": 2.54, "learning_rate": 1.2311155407642694e-05, "loss": 1.8275, "step": 8423 }, { "epoch": 2.54, "learning_rate": 1.229553839247568e-05, "loss": 1.8616, "step": 8424 }, { "epoch": 2.54, "learning_rate": 1.2279930640017056e-05, "loss": 1.786, "step": 8425 }, { "epoch": 2.54, "learning_rate": 1.2264332151915181e-05, "loss": 1.804, "step": 8426 }, { "epoch": 2.54, "learning_rate": 1.2248742929817492e-05, "loss": 1.9045, "step": 8427 }, { "epoch": 2.54, "learning_rate": 1.2233162975370349e-05, "loss": 1.8955, "step": 8428 }, { "epoch": 2.54, "learning_rate": 1.2217592290219237e-05, "loss": 1.8784, "step": 8429 }, { "epoch": 2.54, "learning_rate": 1.2202030876008597e-05, "loss": 1.8265, "step": 8430 }, { "epoch": 2.54, "learning_rate": 1.2186478734381945e-05, "loss": 1.8718, "step": 8431 }, { "epoch": 2.54, "learning_rate": 1.2170935866981737e-05, "loss": 1.8395, "step": 8432 }, { "epoch": 2.54, "learning_rate": 1.2155402275449523e-05, "loss": 1.838, "step": 8433 }, { "epoch": 2.54, "learning_rate": 1.213987796142586e-05, "loss": 1.7149, "step": 8434 }, { "epoch": 2.54, "learning_rate": 1.2124362926550293e-05, "loss": 1.8171, "step": 8435 }, { "epoch": 2.54, "learning_rate": 1.210885717246144e-05, "loss": 1.8803, "step": 8436 }, { "epoch": 2.54, "learning_rate": 1.2093360700796852e-05, "loss": 1.8369, "step": 8437 }, { "epoch": 2.54, "learning_rate": 1.2077873513193195e-05, "loss": 1.8752, "step": 8438 }, { "epoch": 2.54, "learning_rate": 1.2062395611286093e-05, "loss": 1.7397, "step": 8439 }, { "epoch": 2.54, "learning_rate": 1.2046926996710218e-05, "loss": 1.8053, "step": 8440 }, { "epoch": 2.54, "learning_rate": 1.2031467671099284e-05, "loss": 1.8721, "step": 8441 }, { "epoch": 2.54, "learning_rate": 1.201601763608593e-05, "loss": 1.8824, "step": 8442 }, { "epoch": 2.54, "learning_rate": 1.2000576893301918e-05, "loss": 1.9262, "step": 8443 }, { "epoch": 2.54, "learning_rate": 1.1985145444377977e-05, "loss": 1.8842, "step": 8444 }, { "epoch": 2.54, "learning_rate": 1.1969723290943857e-05, "loss": 1.8275, "step": 8445 }, { "epoch": 2.54, "learning_rate": 1.195431043462837e-05, "loss": 1.8965, "step": 8446 }, { "epoch": 2.54, "learning_rate": 1.1938906877059264e-05, "loss": 1.7876, "step": 8447 }, { "epoch": 2.54, "learning_rate": 1.1923512619863364e-05, "loss": 1.853, "step": 8448 }, { "epoch": 2.54, "learning_rate": 1.1908127664666502e-05, "loss": 1.775, "step": 8449 }, { "epoch": 2.54, "learning_rate": 1.1892752013093533e-05, "loss": 1.8821, "step": 8450 }, { "epoch": 2.54, "learning_rate": 1.1877385666768337e-05, "loss": 1.8319, "step": 8451 }, { "epoch": 2.54, "learning_rate": 1.1862028627313748e-05, "loss": 1.8684, "step": 8452 }, { "epoch": 2.54, "learning_rate": 1.1846680896351703e-05, "loss": 1.8035, "step": 8453 }, { "epoch": 2.54, "learning_rate": 1.1831342475503105e-05, "loss": 1.8097, "step": 8454 }, { "epoch": 2.55, "learning_rate": 1.181601336638788e-05, "loss": 1.8545, "step": 8455 }, { "epoch": 2.55, "learning_rate": 1.180069357062502e-05, "loss": 1.7986, "step": 8456 }, { "epoch": 2.55, "learning_rate": 1.1785383089832424e-05, "loss": 1.8396, "step": 8457 }, { "epoch": 2.55, "learning_rate": 1.1770081925627107e-05, "loss": 1.9243, "step": 8458 }, { "epoch": 2.55, "learning_rate": 1.1754790079625077e-05, "loss": 1.8709, "step": 8459 }, { "epoch": 2.55, "learning_rate": 1.1739507553441354e-05, "loss": 1.9079, "step": 8460 }, { "epoch": 2.55, "learning_rate": 1.1724234348689934e-05, "loss": 1.8478, "step": 8461 }, { "epoch": 2.55, "learning_rate": 1.1708970466983849e-05, "loss": 1.8474, "step": 8462 }, { "epoch": 2.55, "learning_rate": 1.1693715909935232e-05, "loss": 1.8475, "step": 8463 }, { "epoch": 2.55, "learning_rate": 1.1678470679155096e-05, "loss": 1.7814, "step": 8464 }, { "epoch": 2.55, "learning_rate": 1.1663234776253573e-05, "loss": 1.8852, "step": 8465 }, { "epoch": 2.55, "learning_rate": 1.1648008202839721e-05, "loss": 1.8431, "step": 8466 }, { "epoch": 2.55, "learning_rate": 1.1632790960521678e-05, "loss": 1.7799, "step": 8467 }, { "epoch": 2.55, "learning_rate": 1.1617583050906611e-05, "loss": 1.7626, "step": 8468 }, { "epoch": 2.55, "learning_rate": 1.1602384475600637e-05, "loss": 1.7813, "step": 8469 }, { "epoch": 2.55, "learning_rate": 1.1587195236208937e-05, "loss": 1.8095, "step": 8470 }, { "epoch": 2.55, "learning_rate": 1.1572015334335662e-05, "loss": 1.8496, "step": 8471 }, { "epoch": 2.55, "learning_rate": 1.1556844771584019e-05, "loss": 1.764, "step": 8472 }, { "epoch": 2.55, "learning_rate": 1.1541683549556215e-05, "loss": 1.8053, "step": 8473 }, { "epoch": 2.55, "learning_rate": 1.1526531669853457e-05, "loss": 1.8528, "step": 8474 }, { "epoch": 2.55, "learning_rate": 1.1511389134076011e-05, "loss": 1.899, "step": 8475 }, { "epoch": 2.55, "learning_rate": 1.1496255943823054e-05, "loss": 1.7646, "step": 8476 }, { "epoch": 2.55, "learning_rate": 1.1481132100692926e-05, "loss": 1.8168, "step": 8477 }, { "epoch": 2.55, "learning_rate": 1.1466017606282841e-05, "loss": 1.9337, "step": 8478 }, { "epoch": 2.55, "learning_rate": 1.145091246218909e-05, "loss": 1.866, "step": 8479 }, { "epoch": 2.55, "learning_rate": 1.1435816670007005e-05, "loss": 1.8074, "step": 8480 }, { "epoch": 2.55, "learning_rate": 1.1420730231330823e-05, "loss": 1.7786, "step": 8481 }, { "epoch": 2.55, "learning_rate": 1.1405653147753947e-05, "loss": 1.8936, "step": 8482 }, { "epoch": 2.55, "learning_rate": 1.1390585420868638e-05, "loss": 1.866, "step": 8483 }, { "epoch": 2.55, "learning_rate": 1.1375527052266299e-05, "loss": 1.7425, "step": 8484 }, { "epoch": 2.55, "learning_rate": 1.1360478043537237e-05, "loss": 1.7364, "step": 8485 }, { "epoch": 2.55, "learning_rate": 1.1345438396270824e-05, "loss": 1.8867, "step": 8486 }, { "epoch": 2.55, "learning_rate": 1.133040811205549e-05, "loss": 1.8287, "step": 8487 }, { "epoch": 2.56, "learning_rate": 1.1315387192478565e-05, "loss": 1.7381, "step": 8488 }, { "epoch": 2.56, "learning_rate": 1.1300375639126492e-05, "loss": 1.8561, "step": 8489 }, { "epoch": 2.56, "learning_rate": 1.1285373453584625e-05, "loss": 1.7834, "step": 8490 }, { "epoch": 2.56, "learning_rate": 1.127038063743745e-05, "loss": 1.7963, "step": 8491 }, { "epoch": 2.56, "learning_rate": 1.125539719226839e-05, "loss": 1.8718, "step": 8492 }, { "epoch": 2.56, "learning_rate": 1.124042311965985e-05, "loss": 1.7469, "step": 8493 }, { "epoch": 2.56, "learning_rate": 1.1225458421193313e-05, "loss": 1.8248, "step": 8494 }, { "epoch": 2.56, "learning_rate": 1.1210503098449198e-05, "loss": 1.8163, "step": 8495 }, { "epoch": 2.56, "learning_rate": 1.1195557153007063e-05, "loss": 1.8248, "step": 8496 }, { "epoch": 2.56, "learning_rate": 1.1180620586445312e-05, "loss": 1.8501, "step": 8497 }, { "epoch": 2.56, "learning_rate": 1.1165693400341459e-05, "loss": 1.8514, "step": 8498 }, { "epoch": 2.56, "learning_rate": 1.1150775596272033e-05, "loss": 1.7783, "step": 8499 }, { "epoch": 2.56, "learning_rate": 1.1135867175812487e-05, "loss": 1.9022, "step": 8500 }, { "epoch": 2.56, "learning_rate": 1.1120968140537403e-05, "loss": 1.8361, "step": 8501 }, { "epoch": 2.56, "learning_rate": 1.1106078492020267e-05, "loss": 1.8013, "step": 8502 }, { "epoch": 2.56, "learning_rate": 1.1091198231833622e-05, "loss": 1.7796, "step": 8503 }, { "epoch": 2.56, "learning_rate": 1.1076327361549032e-05, "loss": 1.7079, "step": 8504 }, { "epoch": 2.56, "learning_rate": 1.1061465882737033e-05, "loss": 1.7914, "step": 8505 }, { "epoch": 2.56, "learning_rate": 1.1046613796967209e-05, "loss": 1.8243, "step": 8506 }, { "epoch": 2.56, "learning_rate": 1.1031771105808087e-05, "loss": 1.7329, "step": 8507 }, { "epoch": 2.56, "learning_rate": 1.1016937810827299e-05, "loss": 1.7958, "step": 8508 }, { "epoch": 2.56, "learning_rate": 1.100211391359135e-05, "loss": 1.7323, "step": 8509 }, { "epoch": 2.56, "learning_rate": 1.0987299415665908e-05, "loss": 1.8835, "step": 8510 }, { "epoch": 2.56, "learning_rate": 1.0972494318615567e-05, "loss": 1.8142, "step": 8511 }, { "epoch": 2.56, "learning_rate": 1.0957698624003898e-05, "loss": 1.9024, "step": 8512 }, { "epoch": 2.56, "learning_rate": 1.094291233339354e-05, "loss": 1.7933, "step": 8513 }, { "epoch": 2.56, "learning_rate": 1.0928135448346078e-05, "loss": 1.7448, "step": 8514 }, { "epoch": 2.56, "learning_rate": 1.0913367970422172e-05, "loss": 1.8935, "step": 8515 }, { "epoch": 2.56, "learning_rate": 1.0898609901181477e-05, "loss": 1.8211, "step": 8516 }, { "epoch": 2.56, "learning_rate": 1.088386124218258e-05, "loss": 1.8807, "step": 8517 }, { "epoch": 2.56, "learning_rate": 1.0869121994983177e-05, "loss": 1.8215, "step": 8518 }, { "epoch": 2.56, "learning_rate": 1.0854392161139858e-05, "loss": 1.8084, "step": 8519 }, { "epoch": 2.56, "learning_rate": 1.0839671742208357e-05, "loss": 1.731, "step": 8520 }, { "epoch": 2.57, "learning_rate": 1.0824960739743285e-05, "loss": 1.8162, "step": 8521 }, { "epoch": 2.57, "learning_rate": 1.0810259155298319e-05, "loss": 1.8722, "step": 8522 }, { "epoch": 2.57, "learning_rate": 1.0795566990426153e-05, "loss": 1.8107, "step": 8523 }, { "epoch": 2.57, "learning_rate": 1.0780884246678447e-05, "loss": 1.8552, "step": 8524 }, { "epoch": 2.57, "learning_rate": 1.0766210925605901e-05, "loss": 1.8514, "step": 8525 }, { "epoch": 2.57, "learning_rate": 1.075154702875819e-05, "loss": 1.8054, "step": 8526 }, { "epoch": 2.57, "learning_rate": 1.073689255768402e-05, "loss": 1.8374, "step": 8527 }, { "epoch": 2.57, "learning_rate": 1.0722247513931072e-05, "loss": 1.7581, "step": 8528 }, { "epoch": 2.57, "learning_rate": 1.0707611899046077e-05, "loss": 1.8389, "step": 8529 }, { "epoch": 2.57, "learning_rate": 1.0692985714574733e-05, "loss": 1.8326, "step": 8530 }, { "epoch": 2.57, "learning_rate": 1.0678368962061736e-05, "loss": 1.8381, "step": 8531 }, { "epoch": 2.57, "learning_rate": 1.066376164305083e-05, "loss": 1.873, "step": 8532 }, { "epoch": 2.57, "learning_rate": 1.0649163759084668e-05, "loss": 1.7867, "step": 8533 }, { "epoch": 2.57, "learning_rate": 1.0634575311705053e-05, "loss": 1.7791, "step": 8534 }, { "epoch": 2.57, "learning_rate": 1.0619996302452684e-05, "loss": 1.8053, "step": 8535 }, { "epoch": 2.57, "learning_rate": 1.0605426732867275e-05, "loss": 1.8484, "step": 8536 }, { "epoch": 2.57, "learning_rate": 1.0590866604487559e-05, "loss": 1.8067, "step": 8537 }, { "epoch": 2.57, "learning_rate": 1.0576315918851287e-05, "loss": 1.7118, "step": 8538 }, { "epoch": 2.57, "learning_rate": 1.0561774677495173e-05, "loss": 1.7704, "step": 8539 }, { "epoch": 2.57, "learning_rate": 1.054724288195501e-05, "loss": 1.7207, "step": 8540 }, { "epoch": 2.57, "learning_rate": 1.0532720533765472e-05, "loss": 1.9451, "step": 8541 }, { "epoch": 2.57, "learning_rate": 1.0518207634460342e-05, "loss": 1.8592, "step": 8542 }, { "epoch": 2.57, "learning_rate": 1.050370418557236e-05, "loss": 1.7739, "step": 8543 }, { "epoch": 2.57, "learning_rate": 1.0489210188633302e-05, "loss": 1.8026, "step": 8544 }, { "epoch": 2.57, "learning_rate": 1.0474725645173866e-05, "loss": 1.8188, "step": 8545 }, { "epoch": 2.57, "learning_rate": 1.0460250556723828e-05, "loss": 1.8862, "step": 8546 }, { "epoch": 2.57, "learning_rate": 1.0445784924811951e-05, "loss": 1.7714, "step": 8547 }, { "epoch": 2.57, "learning_rate": 1.0431328750965975e-05, "loss": 1.7935, "step": 8548 }, { "epoch": 2.57, "learning_rate": 1.0416882036712694e-05, "loss": 1.8092, "step": 8549 }, { "epoch": 2.57, "learning_rate": 1.0402444783577814e-05, "loss": 1.8144, "step": 8550 }, { "epoch": 2.57, "learning_rate": 1.0388016993086103e-05, "loss": 1.766, "step": 8551 }, { "epoch": 2.57, "learning_rate": 1.0373598666761342e-05, "loss": 1.746, "step": 8552 }, { "epoch": 2.57, "learning_rate": 1.035918980612628e-05, "loss": 1.8723, "step": 8553 }, { "epoch": 2.57, "learning_rate": 1.0344790412702688e-05, "loss": 1.8528, "step": 8554 }, { "epoch": 2.58, "learning_rate": 1.0330400488011294e-05, "loss": 1.6977, "step": 8555 }, { "epoch": 2.58, "learning_rate": 1.0316020033571882e-05, "loss": 1.8708, "step": 8556 }, { "epoch": 2.58, "learning_rate": 1.0301649050903206e-05, "loss": 1.8677, "step": 8557 }, { "epoch": 2.58, "learning_rate": 1.0287287541523017e-05, "loss": 1.8241, "step": 8558 }, { "epoch": 2.58, "learning_rate": 1.0272935506948101e-05, "loss": 1.8152, "step": 8559 }, { "epoch": 2.58, "learning_rate": 1.0258592948694179e-05, "loss": 1.8355, "step": 8560 }, { "epoch": 2.58, "learning_rate": 1.0244259868276018e-05, "loss": 1.6722, "step": 8561 }, { "epoch": 2.58, "learning_rate": 1.0229936267207385e-05, "loss": 1.8924, "step": 8562 }, { "epoch": 2.58, "learning_rate": 1.0215622147001024e-05, "loss": 1.7808, "step": 8563 }, { "epoch": 2.58, "learning_rate": 1.0201317509168718e-05, "loss": 1.8564, "step": 8564 }, { "epoch": 2.58, "learning_rate": 1.0187022355221165e-05, "loss": 1.7806, "step": 8565 }, { "epoch": 2.58, "learning_rate": 1.017273668666816e-05, "loss": 1.8478, "step": 8566 }, { "epoch": 2.58, "learning_rate": 1.015846050501843e-05, "loss": 1.8821, "step": 8567 }, { "epoch": 2.58, "learning_rate": 1.0144193811779745e-05, "loss": 1.7736, "step": 8568 }, { "epoch": 2.58, "learning_rate": 1.0129936608458823e-05, "loss": 1.8282, "step": 8569 }, { "epoch": 2.58, "learning_rate": 1.0115688896561404e-05, "loss": 1.8647, "step": 8570 }, { "epoch": 2.58, "learning_rate": 1.010145067759225e-05, "loss": 1.8809, "step": 8571 }, { "epoch": 2.58, "learning_rate": 1.0087221953055082e-05, "loss": 1.8789, "step": 8572 }, { "epoch": 2.58, "learning_rate": 1.0073002724452662e-05, "loss": 1.7706, "step": 8573 }, { "epoch": 2.58, "learning_rate": 1.0058792993286681e-05, "loss": 1.8869, "step": 8574 }, { "epoch": 2.58, "learning_rate": 1.0044592761057892e-05, "loss": 1.7899, "step": 8575 }, { "epoch": 2.58, "learning_rate": 1.0030402029266007e-05, "loss": 1.872, "step": 8576 }, { "epoch": 2.58, "learning_rate": 1.0016220799409759e-05, "loss": 1.7876, "step": 8577 }, { "epoch": 2.58, "learning_rate": 1.0002049072986885e-05, "loss": 1.7203, "step": 8578 }, { "epoch": 2.58, "learning_rate": 9.987886851494066e-06, "loss": 1.9114, "step": 8579 }, { "epoch": 2.58, "learning_rate": 9.973734136427015e-06, "loss": 1.8059, "step": 8580 }, { "epoch": 2.58, "learning_rate": 9.95959092928046e-06, "loss": 1.873, "step": 8581 }, { "epoch": 2.58, "learning_rate": 9.945457231548095e-06, "loss": 1.694, "step": 8582 }, { "epoch": 2.58, "learning_rate": 9.931333044722624e-06, "loss": 1.8146, "step": 8583 }, { "epoch": 2.58, "learning_rate": 9.917218370295723e-06, "loss": 1.8142, "step": 8584 }, { "epoch": 2.58, "learning_rate": 9.903113209758096e-06, "loss": 1.8502, "step": 8585 }, { "epoch": 2.58, "learning_rate": 9.889017564599412e-06, "loss": 1.7739, "step": 8586 }, { "epoch": 2.58, "learning_rate": 9.874931436308376e-06, "loss": 1.7846, "step": 8587 }, { "epoch": 2.59, "learning_rate": 9.860854826372667e-06, "loss": 1.8274, "step": 8588 }, { "epoch": 2.59, "learning_rate": 9.846787736278917e-06, "loss": 1.7214, "step": 8589 }, { "epoch": 2.59, "learning_rate": 9.832730167512805e-06, "loss": 1.7929, "step": 8590 }, { "epoch": 2.59, "learning_rate": 9.818682121558998e-06, "loss": 1.8141, "step": 8591 }, { "epoch": 2.59, "learning_rate": 9.804643599901143e-06, "loss": 1.8493, "step": 8592 }, { "epoch": 2.59, "learning_rate": 9.790614604021908e-06, "loss": 1.8143, "step": 8593 }, { "epoch": 2.59, "learning_rate": 9.7765951354029e-06, "loss": 1.7803, "step": 8594 }, { "epoch": 2.59, "learning_rate": 9.762585195524765e-06, "loss": 1.8642, "step": 8595 }, { "epoch": 2.59, "learning_rate": 9.748584785867133e-06, "loss": 1.852, "step": 8596 }, { "epoch": 2.59, "learning_rate": 9.734593907908652e-06, "loss": 1.8319, "step": 8597 }, { "epoch": 2.59, "learning_rate": 9.720612563126884e-06, "loss": 1.8096, "step": 8598 }, { "epoch": 2.59, "learning_rate": 9.706640752998464e-06, "loss": 1.7735, "step": 8599 }, { "epoch": 2.59, "learning_rate": 9.692678478998996e-06, "loss": 1.6963, "step": 8600 }, { "epoch": 2.59, "learning_rate": 9.678725742603078e-06, "loss": 1.8637, "step": 8601 }, { "epoch": 2.59, "learning_rate": 9.6647825452843e-06, "loss": 1.9428, "step": 8602 }, { "epoch": 2.59, "learning_rate": 9.650848888515217e-06, "loss": 1.8619, "step": 8603 }, { "epoch": 2.59, "learning_rate": 9.63692477376742e-06, "loss": 1.7727, "step": 8604 }, { "epoch": 2.59, "learning_rate": 9.623010202511463e-06, "loss": 1.8546, "step": 8605 }, { "epoch": 2.59, "learning_rate": 9.609105176216903e-06, "loss": 1.8486, "step": 8606 }, { "epoch": 2.59, "learning_rate": 9.595209696352314e-06, "loss": 1.8139, "step": 8607 }, { "epoch": 2.59, "learning_rate": 9.581323764385197e-06, "loss": 1.8701, "step": 8608 }, { "epoch": 2.59, "learning_rate": 9.567447381782103e-06, "loss": 1.9028, "step": 8609 }, { "epoch": 2.59, "learning_rate": 9.553580550008545e-06, "loss": 1.8367, "step": 8610 }, { "epoch": 2.59, "learning_rate": 9.539723270529054e-06, "loss": 1.7211, "step": 8611 }, { "epoch": 2.59, "learning_rate": 9.52587554480714e-06, "loss": 1.8559, "step": 8612 }, { "epoch": 2.59, "learning_rate": 9.51203737430526e-06, "loss": 1.8195, "step": 8613 }, { "epoch": 2.59, "learning_rate": 9.498208760484928e-06, "loss": 1.9419, "step": 8614 }, { "epoch": 2.59, "learning_rate": 9.484389704806617e-06, "loss": 1.8007, "step": 8615 }, { "epoch": 2.59, "learning_rate": 9.470580208729807e-06, "loss": 1.8102, "step": 8616 }, { "epoch": 2.59, "learning_rate": 9.45678027371295e-06, "loss": 1.8706, "step": 8617 }, { "epoch": 2.59, "learning_rate": 9.442989901213483e-06, "loss": 1.8788, "step": 8618 }, { "epoch": 2.59, "learning_rate": 9.429209092687863e-06, "loss": 1.9392, "step": 8619 }, { "epoch": 2.59, "learning_rate": 9.415437849591503e-06, "loss": 1.7768, "step": 8620 }, { "epoch": 2.6, "learning_rate": 9.401676173378838e-06, "loss": 1.7639, "step": 8621 }, { "epoch": 2.6, "learning_rate": 9.387924065503261e-06, "loss": 1.8596, "step": 8622 }, { "epoch": 2.6, "learning_rate": 9.374181527417147e-06, "loss": 1.9105, "step": 8623 }, { "epoch": 2.6, "learning_rate": 9.360448560571955e-06, "loss": 1.8479, "step": 8624 }, { "epoch": 2.6, "learning_rate": 9.346725166418003e-06, "loss": 1.868, "step": 8625 }, { "epoch": 2.6, "learning_rate": 9.333011346404685e-06, "loss": 1.7824, "step": 8626 }, { "epoch": 2.6, "learning_rate": 9.319307101980324e-06, "loss": 1.8142, "step": 8627 }, { "epoch": 2.6, "learning_rate": 9.305612434592258e-06, "loss": 1.7475, "step": 8628 }, { "epoch": 2.6, "learning_rate": 9.29192734568689e-06, "loss": 1.7648, "step": 8629 }, { "epoch": 2.6, "learning_rate": 9.278251836709462e-06, "loss": 1.8577, "step": 8630 }, { "epoch": 2.6, "learning_rate": 9.264585909104328e-06, "loss": 1.7632, "step": 8631 }, { "epoch": 2.6, "learning_rate": 9.250929564314758e-06, "loss": 1.8717, "step": 8632 }, { "epoch": 2.6, "learning_rate": 9.237282803783031e-06, "loss": 1.8106, "step": 8633 }, { "epoch": 2.6, "learning_rate": 9.22364562895044e-06, "loss": 1.9278, "step": 8634 }, { "epoch": 2.6, "learning_rate": 9.21001804125724e-06, "loss": 1.8771, "step": 8635 }, { "epoch": 2.6, "learning_rate": 9.196400042142683e-06, "loss": 1.8686, "step": 8636 }, { "epoch": 2.6, "learning_rate": 9.182791633044962e-06, "loss": 1.8466, "step": 8637 }, { "epoch": 2.6, "learning_rate": 9.169192815401373e-06, "loss": 1.8562, "step": 8638 }, { "epoch": 2.6, "learning_rate": 9.155603590648054e-06, "loss": 1.8004, "step": 8639 }, { "epoch": 2.6, "learning_rate": 9.142023960220237e-06, "loss": 1.8637, "step": 8640 }, { "epoch": 2.6, "learning_rate": 9.128453925552106e-06, "loss": 1.8027, "step": 8641 }, { "epoch": 2.6, "learning_rate": 9.114893488076792e-06, "loss": 1.8607, "step": 8642 }, { "epoch": 2.6, "learning_rate": 9.101342649226507e-06, "loss": 1.7421, "step": 8643 }, { "epoch": 2.6, "learning_rate": 9.08780141043234e-06, "loss": 1.874, "step": 8644 }, { "epoch": 2.6, "learning_rate": 9.074269773124466e-06, "loss": 1.7885, "step": 8645 }, { "epoch": 2.6, "learning_rate": 9.060747738731956e-06, "loss": 1.8513, "step": 8646 }, { "epoch": 2.6, "learning_rate": 9.0472353086829e-06, "loss": 1.708, "step": 8647 }, { "epoch": 2.6, "learning_rate": 9.033732484404445e-06, "loss": 1.7759, "step": 8648 }, { "epoch": 2.6, "learning_rate": 9.020239267322616e-06, "loss": 1.8221, "step": 8649 }, { "epoch": 2.6, "learning_rate": 9.006755658862488e-06, "loss": 1.7723, "step": 8650 }, { "epoch": 2.6, "learning_rate": 8.993281660448071e-06, "loss": 1.9218, "step": 8651 }, { "epoch": 2.6, "learning_rate": 8.9798172735024e-06, "loss": 1.7575, "step": 8652 }, { "epoch": 2.6, "learning_rate": 8.966362499447533e-06, "loss": 1.9298, "step": 8653 }, { "epoch": 2.61, "learning_rate": 8.952917339704403e-06, "loss": 1.7976, "step": 8654 }, { "epoch": 2.61, "learning_rate": 8.93948179569305e-06, "loss": 1.8063, "step": 8655 }, { "epoch": 2.61, "learning_rate": 8.926055868832351e-06, "loss": 1.7978, "step": 8656 }, { "epoch": 2.61, "learning_rate": 8.912639560540358e-06, "loss": 1.7378, "step": 8657 }, { "epoch": 2.61, "learning_rate": 8.899232872233931e-06, "loss": 1.8965, "step": 8658 }, { "epoch": 2.61, "learning_rate": 8.885835805329012e-06, "loss": 1.7899, "step": 8659 }, { "epoch": 2.61, "learning_rate": 8.872448361240526e-06, "loss": 1.8542, "step": 8660 }, { "epoch": 2.61, "learning_rate": 8.859070541382286e-06, "loss": 1.7764, "step": 8661 }, { "epoch": 2.61, "learning_rate": 8.845702347167251e-06, "loss": 1.9006, "step": 8662 }, { "epoch": 2.61, "learning_rate": 8.832343780007202e-06, "loss": 1.8765, "step": 8663 }, { "epoch": 2.61, "learning_rate": 8.81899484131301e-06, "loss": 1.9464, "step": 8664 }, { "epoch": 2.61, "learning_rate": 8.805655532494494e-06, "loss": 1.9824, "step": 8665 }, { "epoch": 2.61, "learning_rate": 8.792325854960416e-06, "loss": 1.9304, "step": 8666 }, { "epoch": 2.61, "learning_rate": 8.779005810118612e-06, "loss": 1.8943, "step": 8667 }, { "epoch": 2.61, "learning_rate": 8.765695399375817e-06, "loss": 1.8784, "step": 8668 }, { "epoch": 2.61, "learning_rate": 8.752394624137806e-06, "loss": 1.9263, "step": 8669 }, { "epoch": 2.61, "learning_rate": 8.739103485809252e-06, "loss": 1.8293, "step": 8670 }, { "epoch": 2.61, "learning_rate": 8.725821985793925e-06, "loss": 1.8215, "step": 8671 }, { "epoch": 2.61, "learning_rate": 8.712550125494523e-06, "loss": 1.7542, "step": 8672 }, { "epoch": 2.61, "learning_rate": 8.699287906312681e-06, "loss": 1.8433, "step": 8673 }, { "epoch": 2.61, "learning_rate": 8.686035329649111e-06, "loss": 1.9179, "step": 8674 }, { "epoch": 2.61, "learning_rate": 8.672792396903384e-06, "loss": 1.7293, "step": 8675 }, { "epoch": 2.61, "learning_rate": 8.659559109474169e-06, "loss": 1.8078, "step": 8676 }, { "epoch": 2.61, "learning_rate": 8.646335468759093e-06, "loss": 1.7934, "step": 8677 }, { "epoch": 2.61, "learning_rate": 8.633121476154694e-06, "loss": 1.9217, "step": 8678 }, { "epoch": 2.61, "learning_rate": 8.619917133056577e-06, "loss": 1.8567, "step": 8679 }, { "epoch": 2.61, "learning_rate": 8.606722440859228e-06, "loss": 1.8689, "step": 8680 }, { "epoch": 2.61, "learning_rate": 8.593537400956242e-06, "loss": 1.8419, "step": 8681 }, { "epoch": 2.61, "learning_rate": 8.580362014740096e-06, "loss": 1.8581, "step": 8682 }, { "epoch": 2.61, "learning_rate": 8.567196283602264e-06, "loss": 1.7486, "step": 8683 }, { "epoch": 2.61, "learning_rate": 8.554040208933246e-06, "loss": 1.8129, "step": 8684 }, { "epoch": 2.61, "learning_rate": 8.540893792122474e-06, "loss": 1.8271, "step": 8685 }, { "epoch": 2.61, "learning_rate": 8.527757034558403e-06, "loss": 1.8014, "step": 8686 }, { "epoch": 2.61, "learning_rate": 8.514629937628394e-06, "loss": 1.8409, "step": 8687 }, { "epoch": 2.62, "learning_rate": 8.501512502718866e-06, "loss": 1.8433, "step": 8688 }, { "epoch": 2.62, "learning_rate": 8.488404731215182e-06, "loss": 1.837, "step": 8689 }, { "epoch": 2.62, "learning_rate": 8.47530662450169e-06, "loss": 1.7792, "step": 8690 }, { "epoch": 2.62, "learning_rate": 8.462218183961734e-06, "loss": 1.8268, "step": 8691 }, { "epoch": 2.62, "learning_rate": 8.44913941097758e-06, "loss": 1.8934, "step": 8692 }, { "epoch": 2.62, "learning_rate": 8.436070306930544e-06, "loss": 1.8091, "step": 8693 }, { "epoch": 2.62, "learning_rate": 8.423010873200876e-06, "loss": 1.8291, "step": 8694 }, { "epoch": 2.62, "learning_rate": 8.40996111116783e-06, "loss": 1.8239, "step": 8695 }, { "epoch": 2.62, "learning_rate": 8.396921022209648e-06, "loss": 1.783, "step": 8696 }, { "epoch": 2.62, "learning_rate": 8.38389060770347e-06, "loss": 1.8345, "step": 8697 }, { "epoch": 2.62, "learning_rate": 8.37086986902551e-06, "loss": 1.8042, "step": 8698 }, { "epoch": 2.62, "learning_rate": 8.357858807550934e-06, "loss": 1.7943, "step": 8699 }, { "epoch": 2.62, "learning_rate": 8.344857424653852e-06, "loss": 1.818, "step": 8700 }, { "epoch": 2.62, "learning_rate": 8.331865721707399e-06, "loss": 1.8147, "step": 8701 }, { "epoch": 2.62, "learning_rate": 8.318883700083647e-06, "loss": 1.796, "step": 8702 }, { "epoch": 2.62, "learning_rate": 8.305911361153662e-06, "loss": 1.8217, "step": 8703 }, { "epoch": 2.62, "learning_rate": 8.292948706287484e-06, "loss": 1.7382, "step": 8704 }, { "epoch": 2.62, "learning_rate": 8.279995736854163e-06, "loss": 1.8028, "step": 8705 }, { "epoch": 2.62, "learning_rate": 8.26705245422168e-06, "loss": 1.7947, "step": 8706 }, { "epoch": 2.62, "learning_rate": 8.254118859756999e-06, "loss": 1.7287, "step": 8707 }, { "epoch": 2.62, "learning_rate": 8.24119495482608e-06, "loss": 1.8075, "step": 8708 }, { "epoch": 2.62, "learning_rate": 8.22828074079386e-06, "loss": 1.7599, "step": 8709 }, { "epoch": 2.62, "learning_rate": 8.215376219024252e-06, "loss": 1.7936, "step": 8710 }, { "epoch": 2.62, "learning_rate": 8.202481390880112e-06, "loss": 1.7486, "step": 8711 }, { "epoch": 2.62, "learning_rate": 8.18959625772331e-06, "loss": 1.8323, "step": 8712 }, { "epoch": 2.62, "learning_rate": 8.176720820914673e-06, "loss": 1.89, "step": 8713 }, { "epoch": 2.62, "learning_rate": 8.163855081814032e-06, "loss": 1.8144, "step": 8714 }, { "epoch": 2.62, "learning_rate": 8.150999041780172e-06, "loss": 1.8169, "step": 8715 }, { "epoch": 2.62, "learning_rate": 8.138152702170831e-06, "loss": 1.7743, "step": 8716 }, { "epoch": 2.62, "learning_rate": 8.125316064342747e-06, "loss": 1.8032, "step": 8717 }, { "epoch": 2.62, "learning_rate": 8.112489129651658e-06, "loss": 1.9108, "step": 8718 }, { "epoch": 2.62, "learning_rate": 8.099671899452232e-06, "loss": 1.8472, "step": 8719 }, { "epoch": 2.62, "learning_rate": 8.086864375098146e-06, "loss": 1.6286, "step": 8720 }, { "epoch": 2.63, "learning_rate": 8.074066557942017e-06, "loss": 1.8528, "step": 8721 }, { "epoch": 2.63, "learning_rate": 8.061278449335474e-06, "loss": 1.6882, "step": 8722 }, { "epoch": 2.63, "learning_rate": 8.048500050629093e-06, "loss": 1.6655, "step": 8723 }, { "epoch": 2.63, "learning_rate": 8.03573136317245e-06, "loss": 1.8126, "step": 8724 }, { "epoch": 2.63, "learning_rate": 8.022972388314087e-06, "loss": 1.7899, "step": 8725 }, { "epoch": 2.63, "learning_rate": 8.010223127401483e-06, "loss": 1.843, "step": 8726 }, { "epoch": 2.63, "learning_rate": 7.99748358178114e-06, "loss": 1.8951, "step": 8727 }, { "epoch": 2.63, "learning_rate": 7.984753752798512e-06, "loss": 1.8674, "step": 8728 }, { "epoch": 2.63, "learning_rate": 7.972033641798038e-06, "loss": 1.7796, "step": 8729 }, { "epoch": 2.63, "learning_rate": 7.95932325012314e-06, "loss": 1.7609, "step": 8730 }, { "epoch": 2.63, "learning_rate": 7.94662257911617e-06, "loss": 1.9196, "step": 8731 }, { "epoch": 2.63, "learning_rate": 7.933931630118474e-06, "loss": 1.89, "step": 8732 }, { "epoch": 2.63, "learning_rate": 7.921250404470404e-06, "loss": 1.8035, "step": 8733 }, { "epoch": 2.63, "learning_rate": 7.908578903511255e-06, "loss": 1.7997, "step": 8734 }, { "epoch": 2.63, "learning_rate": 7.89591712857929e-06, "loss": 1.81, "step": 8735 }, { "epoch": 2.63, "learning_rate": 7.883265081011738e-06, "loss": 1.7771, "step": 8736 }, { "epoch": 2.63, "learning_rate": 7.87062276214484e-06, "loss": 1.9289, "step": 8737 }, { "epoch": 2.63, "learning_rate": 7.857990173313789e-06, "loss": 1.8267, "step": 8738 }, { "epoch": 2.63, "learning_rate": 7.845367315852747e-06, "loss": 1.7584, "step": 8739 }, { "epoch": 2.63, "learning_rate": 7.832754191094815e-06, "loss": 1.8413, "step": 8740 }, { "epoch": 2.63, "learning_rate": 7.820150800372138e-06, "loss": 1.8283, "step": 8741 }, { "epoch": 2.63, "learning_rate": 7.807557145015776e-06, "loss": 1.7196, "step": 8742 }, { "epoch": 2.63, "learning_rate": 7.794973226355784e-06, "loss": 1.8832, "step": 8743 }, { "epoch": 2.63, "learning_rate": 7.782399045721201e-06, "loss": 1.8375, "step": 8744 }, { "epoch": 2.63, "learning_rate": 7.769834604439985e-06, "loss": 1.8515, "step": 8745 }, { "epoch": 2.63, "learning_rate": 7.757279903839133e-06, "loss": 1.876, "step": 8746 }, { "epoch": 2.63, "learning_rate": 7.744734945244558e-06, "loss": 1.8199, "step": 8747 }, { "epoch": 2.63, "learning_rate": 7.73219972998117e-06, "loss": 1.8472, "step": 8748 }, { "epoch": 2.63, "learning_rate": 7.719674259372888e-06, "loss": 1.7516, "step": 8749 }, { "epoch": 2.63, "learning_rate": 7.70715853474251e-06, "loss": 1.8689, "step": 8750 }, { "epoch": 2.63, "learning_rate": 7.694652557411864e-06, "loss": 1.9379, "step": 8751 }, { "epoch": 2.63, "learning_rate": 7.682156328701762e-06, "loss": 1.9044, "step": 8752 }, { "epoch": 2.63, "learning_rate": 7.669669849931949e-06, "loss": 1.7862, "step": 8753 }, { "epoch": 2.64, "learning_rate": 7.65719312242117e-06, "loss": 1.8164, "step": 8754 }, { "epoch": 2.64, "learning_rate": 7.644726147487114e-06, "loss": 1.7586, "step": 8755 }, { "epoch": 2.64, "learning_rate": 7.63226892644645e-06, "loss": 1.7085, "step": 8756 }, { "epoch": 2.64, "learning_rate": 7.619821460614829e-06, "loss": 1.845, "step": 8757 }, { "epoch": 2.64, "learning_rate": 7.6073837513068735e-06, "loss": 1.8398, "step": 8758 }, { "epoch": 2.64, "learning_rate": 7.5949557998361236e-06, "loss": 1.8539, "step": 8759 }, { "epoch": 2.64, "learning_rate": 7.582537607515161e-06, "loss": 1.7615, "step": 8760 }, { "epoch": 2.64, "learning_rate": 7.570129175655505e-06, "loss": 1.6624, "step": 8761 }, { "epoch": 2.64, "learning_rate": 7.557730505567629e-06, "loss": 1.8069, "step": 8762 }, { "epoch": 2.64, "learning_rate": 7.545341598561018e-06, "loss": 1.696, "step": 8763 }, { "epoch": 2.64, "learning_rate": 7.532962455944059e-06, "loss": 1.7963, "step": 8764 }, { "epoch": 2.64, "learning_rate": 7.520593079024174e-06, "loss": 1.9001, "step": 8765 }, { "epoch": 2.64, "learning_rate": 7.508233469107717e-06, "loss": 1.8932, "step": 8766 }, { "epoch": 2.64, "learning_rate": 7.495883627500022e-06, "loss": 1.8078, "step": 8767 }, { "epoch": 2.64, "learning_rate": 7.483543555505412e-06, "loss": 1.7958, "step": 8768 }, { "epoch": 2.64, "learning_rate": 7.471213254427123e-06, "loss": 1.8026, "step": 8769 }, { "epoch": 2.64, "learning_rate": 7.4588927255673906e-06, "loss": 1.7835, "step": 8770 }, { "epoch": 2.64, "learning_rate": 7.44658197022744e-06, "loss": 1.9102, "step": 8771 }, { "epoch": 2.64, "learning_rate": 7.4342809897074424e-06, "loss": 1.8151, "step": 8772 }, { "epoch": 2.64, "learning_rate": 7.421989785306538e-06, "loss": 1.8242, "step": 8773 }, { "epoch": 2.64, "learning_rate": 7.40970835832282e-06, "loss": 1.7537, "step": 8774 }, { "epoch": 2.64, "learning_rate": 7.397436710053362e-06, "loss": 1.7779, "step": 8775 }, { "epoch": 2.64, "learning_rate": 7.385174841794229e-06, "loss": 1.8017, "step": 8776 }, { "epoch": 2.64, "learning_rate": 7.372922754840417e-06, "loss": 1.7399, "step": 8777 }, { "epoch": 2.64, "learning_rate": 7.360680450485913e-06, "loss": 1.8153, "step": 8778 }, { "epoch": 2.64, "learning_rate": 7.348447930023639e-06, "loss": 1.8661, "step": 8779 }, { "epoch": 2.64, "learning_rate": 7.336225194745516e-06, "loss": 1.7544, "step": 8780 }, { "epoch": 2.64, "learning_rate": 7.324012245942424e-06, "loss": 1.809, "step": 8781 }, { "epoch": 2.64, "learning_rate": 7.31180908490422e-06, "loss": 1.8205, "step": 8782 }, { "epoch": 2.64, "learning_rate": 7.299615712919672e-06, "loss": 1.8494, "step": 8783 }, { "epoch": 2.64, "learning_rate": 7.287432131276584e-06, "loss": 1.7467, "step": 8784 }, { "epoch": 2.64, "learning_rate": 7.2752583412616925e-06, "loss": 1.7829, "step": 8785 }, { "epoch": 2.64, "learning_rate": 7.263094344160715e-06, "loss": 1.7867, "step": 8786 }, { "epoch": 2.65, "learning_rate": 7.250940141258322e-06, "loss": 1.8556, "step": 8787 }, { "epoch": 2.65, "learning_rate": 7.238795733838133e-06, "loss": 1.8187, "step": 8788 }, { "epoch": 2.65, "learning_rate": 7.2266611231827544e-06, "loss": 1.7666, "step": 8789 }, { "epoch": 2.65, "learning_rate": 7.214536310573783e-06, "loss": 1.8453, "step": 8790 }, { "epoch": 2.65, "learning_rate": 7.20242129729174e-06, "loss": 1.8111, "step": 8791 }, { "epoch": 2.65, "learning_rate": 7.190316084616122e-06, "loss": 1.6946, "step": 8792 }, { "epoch": 2.65, "learning_rate": 7.178220673825386e-06, "loss": 1.8076, "step": 8793 }, { "epoch": 2.65, "learning_rate": 7.166135066196966e-06, "loss": 1.8792, "step": 8794 }, { "epoch": 2.65, "learning_rate": 7.154059263007251e-06, "loss": 1.7969, "step": 8795 }, { "epoch": 2.65, "learning_rate": 7.141993265531622e-06, "loss": 1.7745, "step": 8796 }, { "epoch": 2.65, "learning_rate": 7.1299370750443905e-06, "loss": 1.7173, "step": 8797 }, { "epoch": 2.65, "learning_rate": 7.117890692818829e-06, "loss": 1.8243, "step": 8798 }, { "epoch": 2.65, "learning_rate": 7.105854120127198e-06, "loss": 1.8215, "step": 8799 }, { "epoch": 2.65, "learning_rate": 7.0938273582407235e-06, "loss": 1.812, "step": 8800 }, { "epoch": 2.65, "learning_rate": 7.081810408429568e-06, "loss": 1.8519, "step": 8801 }, { "epoch": 2.65, "learning_rate": 7.069803271962905e-06, "loss": 1.8285, "step": 8802 }, { "epoch": 2.65, "learning_rate": 7.057805950108787e-06, "loss": 1.7903, "step": 8803 }, { "epoch": 2.65, "learning_rate": 7.0458184441343555e-06, "loss": 1.8671, "step": 8804 }, { "epoch": 2.65, "learning_rate": 7.0338407553055765e-06, "loss": 1.7949, "step": 8805 }, { "epoch": 2.65, "learning_rate": 7.021872884887481e-06, "loss": 1.8647, "step": 8806 }, { "epoch": 2.65, "learning_rate": 7.009914834144038e-06, "loss": 1.7852, "step": 8807 }, { "epoch": 2.65, "learning_rate": 6.997966604338113e-06, "loss": 1.7996, "step": 8808 }, { "epoch": 2.65, "learning_rate": 6.9860281967316754e-06, "loss": 1.7506, "step": 8809 }, { "epoch": 2.65, "learning_rate": 6.974099612585516e-06, "loss": 1.8552, "step": 8810 }, { "epoch": 2.65, "learning_rate": 6.9621808531594725e-06, "loss": 1.8396, "step": 8811 }, { "epoch": 2.65, "learning_rate": 6.9502719197122815e-06, "loss": 1.7456, "step": 8812 }, { "epoch": 2.65, "learning_rate": 6.938372813501692e-06, "loss": 1.843, "step": 8813 }, { "epoch": 2.65, "learning_rate": 6.926483535784445e-06, "loss": 1.775, "step": 8814 }, { "epoch": 2.65, "learning_rate": 6.914604087816146e-06, "loss": 1.7238, "step": 8815 }, { "epoch": 2.65, "learning_rate": 6.902734470851447e-06, "loss": 1.8252, "step": 8816 }, { "epoch": 2.65, "learning_rate": 6.890874686143889e-06, "loss": 1.9186, "step": 8817 }, { "epoch": 2.65, "learning_rate": 6.879024734946071e-06, "loss": 1.842, "step": 8818 }, { "epoch": 2.65, "learning_rate": 6.867184618509481e-06, "loss": 1.8623, "step": 8819 }, { "epoch": 2.66, "learning_rate": 6.855354338084563e-06, "loss": 1.7126, "step": 8820 }, { "epoch": 2.66, "learning_rate": 6.843533894920784e-06, "loss": 1.7717, "step": 8821 }, { "epoch": 2.66, "learning_rate": 6.831723290266479e-06, "loss": 1.8453, "step": 8822 }, { "epoch": 2.66, "learning_rate": 6.81992252536906e-06, "loss": 1.8332, "step": 8823 }, { "epoch": 2.66, "learning_rate": 6.808131601474787e-06, "loss": 1.7493, "step": 8824 }, { "epoch": 2.66, "learning_rate": 6.7963505198289645e-06, "loss": 1.8175, "step": 8825 }, { "epoch": 2.66, "learning_rate": 6.784579281675818e-06, "loss": 1.6727, "step": 8826 }, { "epoch": 2.66, "learning_rate": 6.772817888258509e-06, "loss": 1.8658, "step": 8827 }, { "epoch": 2.66, "learning_rate": 6.761066340819244e-06, "loss": 1.8396, "step": 8828 }, { "epoch": 2.66, "learning_rate": 6.7493246405990974e-06, "loss": 1.8044, "step": 8829 }, { "epoch": 2.66, "learning_rate": 6.737592788838154e-06, "loss": 1.7651, "step": 8830 }, { "epoch": 2.66, "learning_rate": 6.725870786775446e-06, "loss": 1.7925, "step": 8831 }, { "epoch": 2.66, "learning_rate": 6.71415863564897e-06, "loss": 1.8836, "step": 8832 }, { "epoch": 2.66, "learning_rate": 6.702456336695695e-06, "loss": 1.8059, "step": 8833 }, { "epoch": 2.66, "learning_rate": 6.690763891151486e-06, "loss": 1.7685, "step": 8834 }, { "epoch": 2.66, "learning_rate": 6.679081300251266e-06, "loss": 1.8373, "step": 8835 }, { "epoch": 2.66, "learning_rate": 6.667408565228805e-06, "loss": 1.9147, "step": 8836 }, { "epoch": 2.66, "learning_rate": 6.65574568731695e-06, "loss": 1.818, "step": 8837 }, { "epoch": 2.66, "learning_rate": 6.6440926677474365e-06, "loss": 1.7389, "step": 8838 }, { "epoch": 2.66, "learning_rate": 6.632449507750948e-06, "loss": 1.9497, "step": 8839 }, { "epoch": 2.66, "learning_rate": 6.62081620855719e-06, "loss": 1.8617, "step": 8840 }, { "epoch": 2.66, "learning_rate": 6.609192771394724e-06, "loss": 1.8787, "step": 8841 }, { "epoch": 2.66, "learning_rate": 6.5975791974911885e-06, "loss": 1.7405, "step": 8842 }, { "epoch": 2.66, "learning_rate": 6.585975488073126e-06, "loss": 1.7563, "step": 8843 }, { "epoch": 2.66, "learning_rate": 6.574381644366001e-06, "loss": 1.7816, "step": 8844 }, { "epoch": 2.66, "learning_rate": 6.5627976675943005e-06, "loss": 1.7861, "step": 8845 }, { "epoch": 2.66, "learning_rate": 6.5512235589814005e-06, "loss": 1.859, "step": 8846 }, { "epoch": 2.66, "learning_rate": 6.539659319749725e-06, "loss": 1.8272, "step": 8847 }, { "epoch": 2.66, "learning_rate": 6.528104951120573e-06, "loss": 1.8181, "step": 8848 }, { "epoch": 2.66, "learning_rate": 6.516560454314247e-06, "loss": 1.8915, "step": 8849 }, { "epoch": 2.66, "learning_rate": 6.505025830549982e-06, "loss": 1.8247, "step": 8850 }, { "epoch": 2.66, "learning_rate": 6.493501081045983e-06, "loss": 1.8568, "step": 8851 }, { "epoch": 2.66, "learning_rate": 6.4819862070194305e-06, "loss": 1.7891, "step": 8852 }, { "epoch": 2.66, "learning_rate": 6.470481209686419e-06, "loss": 1.8569, "step": 8853 }, { "epoch": 2.67, "learning_rate": 6.45898609026202e-06, "loss": 1.9473, "step": 8854 }, { "epoch": 2.67, "learning_rate": 6.447500849960275e-06, "loss": 1.8132, "step": 8855 }, { "epoch": 2.67, "learning_rate": 6.436025489994169e-06, "loss": 1.7636, "step": 8856 }, { "epoch": 2.67, "learning_rate": 6.424560011575664e-06, "loss": 1.8496, "step": 8857 }, { "epoch": 2.67, "learning_rate": 6.4131044159156385e-06, "loss": 1.9221, "step": 8858 }, { "epoch": 2.67, "learning_rate": 6.4016587042239565e-06, "loss": 1.8343, "step": 8859 }, { "epoch": 2.67, "learning_rate": 6.390222877709406e-06, "loss": 1.8498, "step": 8860 }, { "epoch": 2.67, "learning_rate": 6.378796937579801e-06, "loss": 1.9184, "step": 8861 }, { "epoch": 2.67, "learning_rate": 6.367380885041851e-06, "loss": 1.6366, "step": 8862 }, { "epoch": 2.67, "learning_rate": 6.355974721301217e-06, "loss": 1.8196, "step": 8863 }, { "epoch": 2.67, "learning_rate": 6.344578447562544e-06, "loss": 1.8473, "step": 8864 }, { "epoch": 2.67, "learning_rate": 6.333192065029436e-06, "loss": 1.8358, "step": 8865 }, { "epoch": 2.67, "learning_rate": 6.321815574904421e-06, "loss": 1.8869, "step": 8866 }, { "epoch": 2.67, "learning_rate": 6.310448978389038e-06, "loss": 1.8157, "step": 8867 }, { "epoch": 2.67, "learning_rate": 6.2990922766837046e-06, "loss": 1.8682, "step": 8868 }, { "epoch": 2.67, "learning_rate": 6.2877454709878496e-06, "loss": 1.8614, "step": 8869 }, { "epoch": 2.67, "learning_rate": 6.276408562499825e-06, "loss": 1.841, "step": 8870 }, { "epoch": 2.67, "learning_rate": 6.2650815524169955e-06, "loss": 1.9489, "step": 8871 }, { "epoch": 2.67, "learning_rate": 6.253764441935584e-06, "loss": 1.7628, "step": 8872 }, { "epoch": 2.67, "learning_rate": 6.242457232250843e-06, "loss": 1.8523, "step": 8873 }, { "epoch": 2.67, "learning_rate": 6.231159924556973e-06, "loss": 1.7399, "step": 8874 }, { "epoch": 2.67, "learning_rate": 6.219872520047087e-06, "loss": 1.7987, "step": 8875 }, { "epoch": 2.67, "learning_rate": 6.20859501991331e-06, "loss": 1.7624, "step": 8876 }, { "epoch": 2.67, "learning_rate": 6.197327425346666e-06, "loss": 1.8774, "step": 8877 }, { "epoch": 2.67, "learning_rate": 6.186069737537159e-06, "loss": 1.8288, "step": 8878 }, { "epoch": 2.67, "learning_rate": 6.174821957673748e-06, "loss": 1.8288, "step": 8879 }, { "epoch": 2.67, "learning_rate": 6.163584086944341e-06, "loss": 1.794, "step": 8880 }, { "epoch": 2.67, "learning_rate": 6.152356126535819e-06, "loss": 1.7316, "step": 8881 }, { "epoch": 2.67, "learning_rate": 6.141138077633957e-06, "loss": 1.6815, "step": 8882 }, { "epoch": 2.67, "learning_rate": 6.129929941423562e-06, "loss": 1.792, "step": 8883 }, { "epoch": 2.67, "learning_rate": 6.118731719088322e-06, "loss": 1.825, "step": 8884 }, { "epoch": 2.67, "learning_rate": 6.1075434118109455e-06, "loss": 1.9266, "step": 8885 }, { "epoch": 2.67, "learning_rate": 6.0963650207730534e-06, "loss": 1.7647, "step": 8886 }, { "epoch": 2.68, "learning_rate": 6.085196547155203e-06, "loss": 1.9304, "step": 8887 }, { "epoch": 2.68, "learning_rate": 6.0740379921369495e-06, "loss": 1.846, "step": 8888 }, { "epoch": 2.68, "learning_rate": 6.062889356896761e-06, "loss": 1.745, "step": 8889 }, { "epoch": 2.68, "learning_rate": 6.051750642612097e-06, "loss": 1.8836, "step": 8890 }, { "epoch": 2.68, "learning_rate": 6.040621850459349e-06, "loss": 1.8471, "step": 8891 }, { "epoch": 2.68, "learning_rate": 6.029502981613832e-06, "loss": 1.8728, "step": 8892 }, { "epoch": 2.68, "learning_rate": 6.018394037249853e-06, "loss": 1.75, "step": 8893 }, { "epoch": 2.68, "learning_rate": 6.00729501854066e-06, "loss": 1.7618, "step": 8894 }, { "epoch": 2.68, "learning_rate": 5.9962059266584605e-06, "loss": 1.8287, "step": 8895 }, { "epoch": 2.68, "learning_rate": 5.985126762774385e-06, "loss": 1.7839, "step": 8896 }, { "epoch": 2.68, "learning_rate": 5.974057528058541e-06, "loss": 1.8605, "step": 8897 }, { "epoch": 2.68, "learning_rate": 5.962998223679983e-06, "loss": 1.7217, "step": 8898 }, { "epoch": 2.68, "learning_rate": 5.9519488508067215e-06, "loss": 1.7641, "step": 8899 }, { "epoch": 2.68, "learning_rate": 5.940909410605711e-06, "loss": 1.8725, "step": 8900 }, { "epoch": 2.68, "learning_rate": 5.929879904242841e-06, "loss": 1.9831, "step": 8901 }, { "epoch": 2.68, "learning_rate": 5.918860332882981e-06, "loss": 1.8195, "step": 8902 }, { "epoch": 2.68, "learning_rate": 5.907850697689921e-06, "loss": 1.6817, "step": 8903 }, { "epoch": 2.68, "learning_rate": 5.8968509998264425e-06, "loss": 1.8065, "step": 8904 }, { "epoch": 2.68, "learning_rate": 5.885861240454249e-06, "loss": 1.9364, "step": 8905 }, { "epoch": 2.68, "learning_rate": 5.874881420733991e-06, "loss": 1.8, "step": 8906 }, { "epoch": 2.68, "learning_rate": 5.863911541825274e-06, "loss": 1.7953, "step": 8907 }, { "epoch": 2.68, "learning_rate": 5.852951604886659e-06, "loss": 1.7577, "step": 8908 }, { "epoch": 2.68, "learning_rate": 5.842001611075654e-06, "loss": 1.8624, "step": 8909 }, { "epoch": 2.68, "learning_rate": 5.831061561548745e-06, "loss": 1.7872, "step": 8910 }, { "epoch": 2.68, "learning_rate": 5.820131457461309e-06, "loss": 1.8627, "step": 8911 }, { "epoch": 2.68, "learning_rate": 5.8092112999676985e-06, "loss": 1.8391, "step": 8912 }, { "epoch": 2.68, "learning_rate": 5.798301090221248e-06, "loss": 1.7152, "step": 8913 }, { "epoch": 2.68, "learning_rate": 5.78740082937419e-06, "loss": 1.8628, "step": 8914 }, { "epoch": 2.68, "learning_rate": 5.776510518577771e-06, "loss": 1.994, "step": 8915 }, { "epoch": 2.68, "learning_rate": 5.765630158982094e-06, "loss": 1.7173, "step": 8916 }, { "epoch": 2.68, "learning_rate": 5.754759751736294e-06, "loss": 1.9354, "step": 8917 }, { "epoch": 2.68, "learning_rate": 5.74389929798842e-06, "loss": 1.7968, "step": 8918 }, { "epoch": 2.68, "learning_rate": 5.733048798885477e-06, "loss": 1.8591, "step": 8919 }, { "epoch": 2.69, "learning_rate": 5.722208255573413e-06, "loss": 1.8799, "step": 8920 }, { "epoch": 2.69, "learning_rate": 5.711377669197127e-06, "loss": 1.9276, "step": 8921 }, { "epoch": 2.69, "learning_rate": 5.700557040900456e-06, "loss": 1.7973, "step": 8922 }, { "epoch": 2.69, "learning_rate": 5.68974637182621e-06, "loss": 1.8566, "step": 8923 }, { "epoch": 2.69, "learning_rate": 5.678945663116153e-06, "loss": 1.8686, "step": 8924 }, { "epoch": 2.69, "learning_rate": 5.668154915910928e-06, "loss": 1.8565, "step": 8925 }, { "epoch": 2.69, "learning_rate": 5.657374131350201e-06, "loss": 1.7446, "step": 8926 }, { "epoch": 2.69, "learning_rate": 5.646603310572574e-06, "loss": 1.815, "step": 8927 }, { "epoch": 2.69, "learning_rate": 5.635842454715556e-06, "loss": 1.8633, "step": 8928 }, { "epoch": 2.69, "learning_rate": 5.625091564915674e-06, "loss": 1.898, "step": 8929 }, { "epoch": 2.69, "learning_rate": 5.6143506423082966e-06, "loss": 1.8408, "step": 8930 }, { "epoch": 2.69, "learning_rate": 5.60361968802785e-06, "loss": 1.7982, "step": 8931 }, { "epoch": 2.69, "learning_rate": 5.592898703207639e-06, "loss": 1.7766, "step": 8932 }, { "epoch": 2.69, "learning_rate": 5.582187688979934e-06, "loss": 1.8291, "step": 8933 }, { "epoch": 2.69, "learning_rate": 5.571486646475988e-06, "loss": 1.8659, "step": 8934 }, { "epoch": 2.69, "learning_rate": 5.560795576825928e-06, "loss": 1.7768, "step": 8935 }, { "epoch": 2.69, "learning_rate": 5.5501144811588745e-06, "loss": 1.85, "step": 8936 }, { "epoch": 2.69, "learning_rate": 5.539443360602891e-06, "loss": 1.8525, "step": 8937 }, { "epoch": 2.69, "learning_rate": 5.528782216284989e-06, "loss": 1.8054, "step": 8938 }, { "epoch": 2.69, "learning_rate": 5.518131049331132e-06, "loss": 1.8364, "step": 8939 }, { "epoch": 2.69, "learning_rate": 5.5074898608662e-06, "loss": 1.8799, "step": 8940 }, { "epoch": 2.69, "learning_rate": 5.496858652014025e-06, "loss": 1.9083, "step": 8941 }, { "epoch": 2.69, "learning_rate": 5.486237423897433e-06, "loss": 1.8486, "step": 8942 }, { "epoch": 2.69, "learning_rate": 5.475626177638138e-06, "loss": 1.8006, "step": 8943 }, { "epoch": 2.69, "learning_rate": 5.465024914356842e-06, "loss": 1.8007, "step": 8944 }, { "epoch": 2.69, "learning_rate": 5.4544336351731504e-06, "loss": 1.7677, "step": 8945 }, { "epoch": 2.69, "learning_rate": 5.443852341205646e-06, "loss": 1.9341, "step": 8946 }, { "epoch": 2.69, "learning_rate": 5.433281033571847e-06, "loss": 1.8556, "step": 8947 }, { "epoch": 2.69, "learning_rate": 5.4227197133882355e-06, "loss": 1.7702, "step": 8948 }, { "epoch": 2.69, "learning_rate": 5.4121683817701994e-06, "loss": 1.8648, "step": 8949 }, { "epoch": 2.69, "learning_rate": 5.40162703983208e-06, "loss": 1.8797, "step": 8950 }, { "epoch": 2.69, "learning_rate": 5.391095688687231e-06, "loss": 1.847, "step": 8951 }, { "epoch": 2.69, "learning_rate": 5.380574329447841e-06, "loss": 1.9423, "step": 8952 }, { "epoch": 2.7, "learning_rate": 5.370062963225153e-06, "loss": 1.9149, "step": 8953 }, { "epoch": 2.7, "learning_rate": 5.3595615911292475e-06, "loss": 1.892, "step": 8954 }, { "epoch": 2.7, "learning_rate": 5.349070214269214e-06, "loss": 1.7327, "step": 8955 }, { "epoch": 2.7, "learning_rate": 5.338588833753111e-06, "loss": 1.7973, "step": 8956 }, { "epoch": 2.7, "learning_rate": 5.328117450687875e-06, "loss": 1.9722, "step": 8957 }, { "epoch": 2.7, "learning_rate": 5.3176560661794435e-06, "loss": 1.8548, "step": 8958 }, { "epoch": 2.7, "learning_rate": 5.307204681332633e-06, "loss": 1.7756, "step": 8959 }, { "epoch": 2.7, "learning_rate": 5.296763297251272e-06, "loss": 1.7807, "step": 8960 }, { "epoch": 2.7, "learning_rate": 5.28633191503809e-06, "loss": 1.7995, "step": 8961 }, { "epoch": 2.7, "learning_rate": 5.2759105357947816e-06, "loss": 1.8129, "step": 8962 }, { "epoch": 2.7, "learning_rate": 5.265499160621989e-06, "loss": 1.7641, "step": 8963 }, { "epoch": 2.7, "learning_rate": 5.255097790619245e-06, "loss": 1.8333, "step": 8964 }, { "epoch": 2.7, "learning_rate": 5.244706426885116e-06, "loss": 1.6862, "step": 8965 }, { "epoch": 2.7, "learning_rate": 5.234325070517032e-06, "loss": 1.8157, "step": 8966 }, { "epoch": 2.7, "learning_rate": 5.223953722611408e-06, "loss": 1.8035, "step": 8967 }, { "epoch": 2.7, "learning_rate": 5.213592384263599e-06, "loss": 1.8194, "step": 8968 }, { "epoch": 2.7, "learning_rate": 5.203241056567854e-06, "loss": 1.7812, "step": 8969 }, { "epoch": 2.7, "learning_rate": 5.192899740617463e-06, "loss": 1.8921, "step": 8970 }, { "epoch": 2.7, "learning_rate": 5.182568437504554e-06, "loss": 1.9281, "step": 8971 }, { "epoch": 2.7, "learning_rate": 5.172247148320275e-06, "loss": 1.9627, "step": 8972 }, { "epoch": 2.7, "learning_rate": 5.1619358741546665e-06, "loss": 1.8867, "step": 8973 }, { "epoch": 2.7, "learning_rate": 5.151634616096712e-06, "loss": 1.7487, "step": 8974 }, { "epoch": 2.7, "learning_rate": 5.1413433752344084e-06, "loss": 1.7982, "step": 8975 }, { "epoch": 2.7, "learning_rate": 5.131062152654609e-06, "loss": 1.8318, "step": 8976 }, { "epoch": 2.7, "learning_rate": 5.120790949443155e-06, "loss": 1.9037, "step": 8977 }, { "epoch": 2.7, "learning_rate": 5.110529766684791e-06, "loss": 1.7375, "step": 8978 }, { "epoch": 2.7, "learning_rate": 5.1002786054632376e-06, "loss": 1.8127, "step": 8979 }, { "epoch": 2.7, "learning_rate": 5.090037466861175e-06, "loss": 1.8042, "step": 8980 }, { "epoch": 2.7, "learning_rate": 5.07980635196017e-06, "loss": 1.7229, "step": 8981 }, { "epoch": 2.7, "learning_rate": 5.069585261840781e-06, "loss": 1.7275, "step": 8982 }, { "epoch": 2.7, "learning_rate": 5.059374197582434e-06, "loss": 1.7081, "step": 8983 }, { "epoch": 2.7, "learning_rate": 5.04917316026362e-06, "loss": 1.927, "step": 8984 }, { "epoch": 2.7, "learning_rate": 5.038982150961658e-06, "loss": 1.7312, "step": 8985 }, { "epoch": 2.7, "learning_rate": 5.028801170752839e-06, "loss": 1.8938, "step": 8986 }, { "epoch": 2.71, "learning_rate": 5.018630220712439e-06, "loss": 1.8534, "step": 8987 }, { "epoch": 2.71, "learning_rate": 5.008469301914598e-06, "loss": 1.8645, "step": 8988 }, { "epoch": 2.71, "learning_rate": 4.998318415432479e-06, "loss": 1.8545, "step": 8989 }, { "epoch": 2.71, "learning_rate": 4.988177562338114e-06, "loss": 1.9029, "step": 8990 }, { "epoch": 2.71, "learning_rate": 4.978046743702513e-06, "loss": 1.9129, "step": 8991 }, { "epoch": 2.71, "learning_rate": 4.967925960595654e-06, "loss": 1.8126, "step": 8992 }, { "epoch": 2.71, "learning_rate": 4.957815214086348e-06, "loss": 1.7885, "step": 8993 }, { "epoch": 2.71, "learning_rate": 4.947714505242496e-06, "loss": 1.9082, "step": 8994 }, { "epoch": 2.71, "learning_rate": 4.937623835130823e-06, "loss": 1.8482, "step": 8995 }, { "epoch": 2.71, "learning_rate": 4.927543204817042e-06, "loss": 1.8897, "step": 8996 }, { "epoch": 2.71, "learning_rate": 4.917472615365759e-06, "loss": 1.7289, "step": 8997 }, { "epoch": 2.71, "learning_rate": 4.907412067840611e-06, "loss": 1.7862, "step": 8998 }, { "epoch": 2.71, "learning_rate": 4.897361563304104e-06, "loss": 1.8034, "step": 8999 }, { "epoch": 2.71, "learning_rate": 4.887321102817677e-06, "loss": 1.755, "step": 9000 }, { "epoch": 2.71, "learning_rate": 4.877290687441771e-06, "loss": 1.873, "step": 9001 }, { "epoch": 2.71, "learning_rate": 4.867270318235662e-06, "loss": 1.7205, "step": 9002 }, { "epoch": 2.71, "learning_rate": 4.8572599962576905e-06, "loss": 1.7776, "step": 9003 }, { "epoch": 2.71, "learning_rate": 4.847259722565056e-06, "loss": 1.7219, "step": 9004 }, { "epoch": 2.71, "learning_rate": 4.837269498213903e-06, "loss": 1.8415, "step": 9005 }, { "epoch": 2.71, "learning_rate": 4.827289324259354e-06, "loss": 1.7921, "step": 9006 }, { "epoch": 2.71, "learning_rate": 4.817319201755377e-06, "loss": 1.8275, "step": 9007 }, { "epoch": 2.71, "learning_rate": 4.807359131755029e-06, "loss": 1.8323, "step": 9008 }, { "epoch": 2.71, "learning_rate": 4.797409115310158e-06, "loss": 1.8222, "step": 9009 }, { "epoch": 2.71, "learning_rate": 4.787469153471646e-06, "loss": 1.8548, "step": 9010 }, { "epoch": 2.71, "learning_rate": 4.7775392472892535e-06, "loss": 1.8365, "step": 9011 }, { "epoch": 2.71, "learning_rate": 4.767619397811729e-06, "loss": 1.8901, "step": 9012 }, { "epoch": 2.71, "learning_rate": 4.757709606086736e-06, "loss": 1.8831, "step": 9013 }, { "epoch": 2.71, "learning_rate": 4.747809873160858e-06, "loss": 1.8692, "step": 9014 }, { "epoch": 2.71, "learning_rate": 4.737920200079638e-06, "loss": 1.8241, "step": 9015 }, { "epoch": 2.71, "learning_rate": 4.728040587887539e-06, "loss": 1.8012, "step": 9016 }, { "epoch": 2.71, "learning_rate": 4.718171037628005e-06, "loss": 1.8728, "step": 9017 }, { "epoch": 2.71, "learning_rate": 4.7083115503433694e-06, "loss": 1.8745, "step": 9018 }, { "epoch": 2.71, "learning_rate": 4.698462127074909e-06, "loss": 1.8328, "step": 9019 }, { "epoch": 2.72, "learning_rate": 4.6886227688628715e-06, "loss": 1.8347, "step": 9020 }, { "epoch": 2.72, "learning_rate": 4.678793476746391e-06, "loss": 1.7594, "step": 9021 }, { "epoch": 2.72, "learning_rate": 4.66897425176358e-06, "loss": 1.8798, "step": 9022 }, { "epoch": 2.72, "learning_rate": 4.65916509495149e-06, "loss": 1.8532, "step": 9023 }, { "epoch": 2.72, "learning_rate": 4.649366007346068e-06, "loss": 1.9005, "step": 9024 }, { "epoch": 2.72, "learning_rate": 4.63957698998223e-06, "loss": 1.8262, "step": 9025 }, { "epoch": 2.72, "learning_rate": 4.629798043893818e-06, "loss": 1.8269, "step": 9026 }, { "epoch": 2.72, "learning_rate": 4.6200291701136155e-06, "loss": 1.7518, "step": 9027 }, { "epoch": 2.72, "learning_rate": 4.610270369673364e-06, "loss": 1.8211, "step": 9028 }, { "epoch": 2.72, "learning_rate": 4.600521643603684e-06, "loss": 1.7976, "step": 9029 }, { "epoch": 2.72, "learning_rate": 4.590782992934173e-06, "loss": 1.9302, "step": 9030 }, { "epoch": 2.72, "learning_rate": 4.581054418693354e-06, "loss": 1.7837, "step": 9031 }, { "epoch": 2.72, "learning_rate": 4.571335921908715e-06, "loss": 1.7694, "step": 9032 }, { "epoch": 2.72, "learning_rate": 4.561627503606614e-06, "loss": 1.9665, "step": 9033 }, { "epoch": 2.72, "learning_rate": 4.551929164812407e-06, "loss": 1.8714, "step": 9034 }, { "epoch": 2.72, "learning_rate": 4.542240906550355e-06, "loss": 1.889, "step": 9035 }, { "epoch": 2.72, "learning_rate": 4.5325627298436705e-06, "loss": 1.8972, "step": 9036 }, { "epoch": 2.72, "learning_rate": 4.522894635714492e-06, "loss": 1.8451, "step": 9037 }, { "epoch": 2.72, "learning_rate": 4.51323662518387e-06, "loss": 1.8853, "step": 9038 }, { "epoch": 2.72, "learning_rate": 4.5035886992718325e-06, "loss": 1.8881, "step": 9039 }, { "epoch": 2.72, "learning_rate": 4.493950858997331e-06, "loss": 1.8038, "step": 9040 }, { "epoch": 2.72, "learning_rate": 4.48432310537823e-06, "loss": 1.7861, "step": 9041 }, { "epoch": 2.72, "learning_rate": 4.474705439431348e-06, "loss": 1.7929, "step": 9042 }, { "epoch": 2.72, "learning_rate": 4.4650978621724296e-06, "loss": 1.8864, "step": 9043 }, { "epoch": 2.72, "learning_rate": 4.455500374616161e-06, "loss": 1.7983, "step": 9044 }, { "epoch": 2.72, "learning_rate": 4.445912977776157e-06, "loss": 1.8109, "step": 9045 }, { "epoch": 2.72, "learning_rate": 4.436335672664971e-06, "loss": 1.7867, "step": 9046 }, { "epoch": 2.72, "learning_rate": 4.426768460294106e-06, "loss": 1.8107, "step": 9047 }, { "epoch": 2.72, "learning_rate": 4.417211341673944e-06, "loss": 1.9281, "step": 9048 }, { "epoch": 2.72, "learning_rate": 4.407664317813853e-06, "loss": 1.8438, "step": 9049 }, { "epoch": 2.72, "learning_rate": 4.39812738972214e-06, "loss": 2.0594, "step": 9050 }, { "epoch": 2.72, "learning_rate": 4.388600558405998e-06, "loss": 1.8062, "step": 9051 }, { "epoch": 2.72, "learning_rate": 4.37908382487161e-06, "loss": 1.8239, "step": 9052 }, { "epoch": 2.73, "learning_rate": 4.369577190124041e-06, "loss": 1.7413, "step": 9053 }, { "epoch": 2.73, "learning_rate": 4.360080655167309e-06, "loss": 1.8106, "step": 9054 }, { "epoch": 2.73, "learning_rate": 4.350594221004389e-06, "loss": 1.7826, "step": 9055 }, { "epoch": 2.73, "learning_rate": 4.34111788863717e-06, "loss": 1.7267, "step": 9056 }, { "epoch": 2.73, "learning_rate": 4.33165165906646e-06, "loss": 1.8807, "step": 9057 }, { "epoch": 2.73, "learning_rate": 4.322195533292017e-06, "loss": 1.7976, "step": 9058 }, { "epoch": 2.73, "learning_rate": 4.312749512312531e-06, "loss": 1.9472, "step": 9059 }, { "epoch": 2.73, "learning_rate": 4.303313597125613e-06, "loss": 1.8131, "step": 9060 }, { "epoch": 2.73, "learning_rate": 4.293887788727835e-06, "loss": 1.825, "step": 9061 }, { "epoch": 2.73, "learning_rate": 4.284472088114666e-06, "loss": 1.7811, "step": 9062 }, { "epoch": 2.73, "learning_rate": 4.2750664962805224e-06, "loss": 1.7256, "step": 9063 }, { "epoch": 2.73, "learning_rate": 4.265671014218764e-06, "loss": 1.9378, "step": 9064 }, { "epoch": 2.73, "learning_rate": 4.256285642921665e-06, "loss": 1.8123, "step": 9065 }, { "epoch": 2.73, "learning_rate": 4.246910383380464e-06, "loss": 1.8951, "step": 9066 }, { "epoch": 2.73, "learning_rate": 4.237545236585272e-06, "loss": 1.893, "step": 9067 }, { "epoch": 2.73, "learning_rate": 4.228190203525184e-06, "loss": 1.8025, "step": 9068 }, { "epoch": 2.73, "learning_rate": 4.218845285188222e-06, "loss": 1.8986, "step": 9069 }, { "epoch": 2.73, "learning_rate": 4.209510482561307e-06, "loss": 1.7712, "step": 9070 }, { "epoch": 2.73, "learning_rate": 4.2001857966303405e-06, "loss": 1.7848, "step": 9071 }, { "epoch": 2.73, "learning_rate": 4.1908712283801e-06, "loss": 1.7948, "step": 9072 }, { "epoch": 2.73, "learning_rate": 4.1815667787943326e-06, "loss": 1.7446, "step": 9073 }, { "epoch": 2.73, "learning_rate": 4.172272448855718e-06, "loss": 1.8446, "step": 9074 }, { "epoch": 2.73, "learning_rate": 4.16298823954584e-06, "loss": 1.8182, "step": 9075 }, { "epoch": 2.73, "learning_rate": 4.153714151845245e-06, "loss": 1.8736, "step": 9076 }, { "epoch": 2.73, "learning_rate": 4.144450186733373e-06, "loss": 1.8603, "step": 9077 }, { "epoch": 2.73, "learning_rate": 4.13519634518863e-06, "loss": 1.7822, "step": 9078 }, { "epoch": 2.73, "learning_rate": 4.125952628188334e-06, "loss": 1.8265, "step": 9079 }, { "epoch": 2.73, "learning_rate": 4.11671903670875e-06, "loss": 1.9063, "step": 9080 }, { "epoch": 2.73, "learning_rate": 4.107495571725073e-06, "loss": 1.7385, "step": 9081 }, { "epoch": 2.73, "learning_rate": 4.098282234211381e-06, "loss": 1.9617, "step": 9082 }, { "epoch": 2.73, "learning_rate": 4.0890790251407515e-06, "loss": 1.8369, "step": 9083 }, { "epoch": 2.73, "learning_rate": 4.0798859454851405e-06, "loss": 2.0289, "step": 9084 }, { "epoch": 2.73, "learning_rate": 4.070702996215481e-06, "loss": 1.7713, "step": 9085 }, { "epoch": 2.74, "learning_rate": 4.061530178301576e-06, "loss": 1.8851, "step": 9086 }, { "epoch": 2.74, "learning_rate": 4.052367492712217e-06, "loss": 1.8564, "step": 9087 }, { "epoch": 2.74, "learning_rate": 4.043214940415085e-06, "loss": 1.7402, "step": 9088 }, { "epoch": 2.74, "learning_rate": 4.034072522376809e-06, "loss": 1.8497, "step": 9089 }, { "epoch": 2.74, "learning_rate": 4.024940239562969e-06, "loss": 1.7996, "step": 9090 }, { "epoch": 2.74, "learning_rate": 4.0158180929380176e-06, "loss": 1.8253, "step": 9091 }, { "epoch": 2.74, "learning_rate": 4.006706083465373e-06, "loss": 1.8253, "step": 9092 }, { "epoch": 2.74, "learning_rate": 3.997604212107398e-06, "loss": 1.8454, "step": 9093 }, { "epoch": 2.74, "learning_rate": 3.9885124798253574e-06, "loss": 1.7441, "step": 9094 }, { "epoch": 2.74, "learning_rate": 3.979430887579461e-06, "loss": 1.8324, "step": 9095 }, { "epoch": 2.74, "learning_rate": 3.970359436328819e-06, "loss": 1.8693, "step": 9096 }, { "epoch": 2.74, "learning_rate": 3.96129812703151e-06, "loss": 1.8318, "step": 9097 }, { "epoch": 2.74, "learning_rate": 3.952246960644513e-06, "loss": 1.8031, "step": 9098 }, { "epoch": 2.74, "learning_rate": 3.943205938123762e-06, "loss": 1.766, "step": 9099 }, { "epoch": 2.74, "learning_rate": 3.934175060424105e-06, "loss": 1.7921, "step": 9100 }, { "epoch": 2.74, "learning_rate": 3.925154328499292e-06, "loss": 1.8166, "step": 9101 }, { "epoch": 2.74, "learning_rate": 3.916143743302047e-06, "loss": 1.8159, "step": 9102 }, { "epoch": 2.74, "learning_rate": 3.907143305784e-06, "loss": 1.8408, "step": 9103 }, { "epoch": 2.74, "learning_rate": 3.898153016895711e-06, "loss": 1.7987, "step": 9104 }, { "epoch": 2.74, "learning_rate": 3.889172877586678e-06, "loss": 1.8852, "step": 9105 }, { "epoch": 2.74, "learning_rate": 3.880202888805295e-06, "loss": 1.7222, "step": 9106 }, { "epoch": 2.74, "learning_rate": 3.871243051498918e-06, "loss": 1.8259, "step": 9107 }, { "epoch": 2.74, "learning_rate": 3.862293366613834e-06, "loss": 1.8275, "step": 9108 }, { "epoch": 2.74, "learning_rate": 3.853353835095241e-06, "loss": 1.7468, "step": 9109 }, { "epoch": 2.74, "learning_rate": 3.844424457887241e-06, "loss": 1.7209, "step": 9110 }, { "epoch": 2.74, "learning_rate": 3.835505235932912e-06, "loss": 1.7534, "step": 9111 }, { "epoch": 2.74, "learning_rate": 3.826596170174235e-06, "loss": 1.9299, "step": 9112 }, { "epoch": 2.74, "learning_rate": 3.817697261552122e-06, "loss": 1.8518, "step": 9113 }, { "epoch": 2.74, "learning_rate": 3.808808511006412e-06, "loss": 1.842, "step": 9114 }, { "epoch": 2.74, "learning_rate": 3.799929919475864e-06, "loss": 1.7965, "step": 9115 }, { "epoch": 2.74, "learning_rate": 3.791061487898151e-06, "loss": 1.7544, "step": 9116 }, { "epoch": 2.74, "learning_rate": 3.7822032172099453e-06, "loss": 1.81, "step": 9117 }, { "epoch": 2.74, "learning_rate": 3.7733551083467545e-06, "loss": 1.8664, "step": 9118 }, { "epoch": 2.75, "learning_rate": 3.764517162243064e-06, "loss": 1.7581, "step": 9119 }, { "epoch": 2.75, "learning_rate": 3.755689379832261e-06, "loss": 1.7734, "step": 9120 }, { "epoch": 2.75, "learning_rate": 3.7468717620466774e-06, "loss": 1.8755, "step": 9121 }, { "epoch": 2.75, "learning_rate": 3.7380643098175683e-06, "loss": 1.7797, "step": 9122 }, { "epoch": 2.75, "learning_rate": 3.7292670240751113e-06, "loss": 1.8638, "step": 9123 }, { "epoch": 2.75, "learning_rate": 3.7204799057484197e-06, "loss": 1.8197, "step": 9124 }, { "epoch": 2.75, "learning_rate": 3.711702955765517e-06, "loss": 1.8455, "step": 9125 }, { "epoch": 2.75, "learning_rate": 3.702936175053362e-06, "loss": 1.8499, "step": 9126 }, { "epoch": 2.75, "learning_rate": 3.6941795645378364e-06, "loss": 1.8191, "step": 9127 }, { "epoch": 2.75, "learning_rate": 3.6854331251437556e-06, "loss": 1.8336, "step": 9128 }, { "epoch": 2.75, "learning_rate": 3.676696857794859e-06, "loss": 1.8599, "step": 9129 }, { "epoch": 2.75, "learning_rate": 3.667970763413775e-06, "loss": 1.8577, "step": 9130 }, { "epoch": 2.75, "learning_rate": 3.659254842922144e-06, "loss": 1.912, "step": 9131 }, { "epoch": 2.75, "learning_rate": 3.650549097240441e-06, "loss": 1.8943, "step": 9132 }, { "epoch": 2.75, "learning_rate": 3.6418535272881194e-06, "loss": 1.8454, "step": 9133 }, { "epoch": 2.75, "learning_rate": 3.6331681339835223e-06, "loss": 1.7765, "step": 9134 }, { "epoch": 2.75, "learning_rate": 3.624492918243949e-06, "loss": 1.7436, "step": 9135 }, { "epoch": 2.75, "learning_rate": 3.615827880985634e-06, "loss": 1.9098, "step": 9136 }, { "epoch": 2.75, "learning_rate": 3.607173023123678e-06, "loss": 1.8392, "step": 9137 }, { "epoch": 2.75, "learning_rate": 3.598528345572183e-06, "loss": 1.7572, "step": 9138 }, { "epoch": 2.75, "learning_rate": 3.589893849244097e-06, "loss": 1.8296, "step": 9139 }, { "epoch": 2.75, "learning_rate": 3.581269535051346e-06, "loss": 1.8763, "step": 9140 }, { "epoch": 2.75, "learning_rate": 3.572655403904801e-06, "loss": 1.8374, "step": 9141 }, { "epoch": 2.75, "learning_rate": 3.564051456714179e-06, "loss": 1.9448, "step": 9142 }, { "epoch": 2.75, "learning_rate": 3.555457694388198e-06, "loss": 1.9112, "step": 9143 }, { "epoch": 2.75, "learning_rate": 3.546874117834442e-06, "loss": 1.8416, "step": 9144 }, { "epoch": 2.75, "learning_rate": 3.538300727959476e-06, "loss": 1.8089, "step": 9145 }, { "epoch": 2.75, "learning_rate": 3.5297375256687193e-06, "loss": 1.7022, "step": 9146 }, { "epoch": 2.75, "learning_rate": 3.521184511866593e-06, "loss": 1.8413, "step": 9147 }, { "epoch": 2.75, "learning_rate": 3.512641687456397e-06, "loss": 1.861, "step": 9148 }, { "epoch": 2.75, "learning_rate": 3.504109053340332e-06, "loss": 1.8864, "step": 9149 }, { "epoch": 2.75, "learning_rate": 3.495586610419599e-06, "loss": 1.7801, "step": 9150 }, { "epoch": 2.75, "learning_rate": 3.4870743595942333e-06, "loss": 1.823, "step": 9151 }, { "epoch": 2.75, "learning_rate": 3.4785723017632722e-06, "loss": 1.7942, "step": 9152 }, { "epoch": 2.76, "learning_rate": 3.47008043782463e-06, "loss": 1.7546, "step": 9153 }, { "epoch": 2.76, "learning_rate": 3.4615987686751227e-06, "loss": 1.848, "step": 9154 }, { "epoch": 2.76, "learning_rate": 3.45312729521059e-06, "loss": 1.8107, "step": 9155 }, { "epoch": 2.76, "learning_rate": 3.4446660183256706e-06, "loss": 1.8607, "step": 9156 }, { "epoch": 2.76, "learning_rate": 3.436214938913995e-06, "loss": 1.888, "step": 9157 }, { "epoch": 2.76, "learning_rate": 3.4277740578681163e-06, "loss": 1.8529, "step": 9158 }, { "epoch": 2.76, "learning_rate": 3.419343376079498e-06, "loss": 1.8459, "step": 9159 }, { "epoch": 2.76, "learning_rate": 3.4109228944385395e-06, "loss": 1.7566, "step": 9160 }, { "epoch": 2.76, "learning_rate": 3.402512613834519e-06, "loss": 1.8127, "step": 9161 }, { "epoch": 2.76, "learning_rate": 3.394112535155691e-06, "loss": 1.863, "step": 9162 }, { "epoch": 2.76, "learning_rate": 3.385722659289181e-06, "loss": 1.8502, "step": 9163 }, { "epoch": 2.76, "learning_rate": 3.3773429871211126e-06, "loss": 1.8605, "step": 9164 }, { "epoch": 2.76, "learning_rate": 3.368973519536478e-06, "loss": 1.7978, "step": 9165 }, { "epoch": 2.76, "learning_rate": 3.36061425741917e-06, "loss": 1.8041, "step": 9166 }, { "epoch": 2.76, "learning_rate": 3.3522652016520493e-06, "loss": 1.9211, "step": 9167 }, { "epoch": 2.76, "learning_rate": 3.3439263531168775e-06, "loss": 1.8543, "step": 9168 }, { "epoch": 2.76, "learning_rate": 3.3355977126943494e-06, "loss": 1.7982, "step": 9169 }, { "epoch": 2.76, "learning_rate": 3.327279281264095e-06, "loss": 1.8102, "step": 9170 }, { "epoch": 2.76, "learning_rate": 3.318971059704601e-06, "loss": 1.7846, "step": 9171 }, { "epoch": 2.76, "learning_rate": 3.3106730488933533e-06, "loss": 1.92, "step": 9172 }, { "epoch": 2.76, "learning_rate": 3.3023852497067297e-06, "loss": 1.7921, "step": 9173 }, { "epoch": 2.76, "learning_rate": 3.2941076630200187e-06, "loss": 1.8688, "step": 9174 }, { "epoch": 2.76, "learning_rate": 3.285840289707431e-06, "loss": 1.7833, "step": 9175 }, { "epoch": 2.76, "learning_rate": 3.2775831306421147e-06, "loss": 1.9097, "step": 9176 }, { "epoch": 2.76, "learning_rate": 3.269336186696126e-06, "loss": 1.8196, "step": 9177 }, { "epoch": 2.76, "learning_rate": 3.2610994587404575e-06, "loss": 1.7765, "step": 9178 }, { "epoch": 2.76, "learning_rate": 3.2528729476450135e-06, "loss": 1.8819, "step": 9179 }, { "epoch": 2.76, "learning_rate": 3.2446566542785994e-06, "loss": 1.7952, "step": 9180 }, { "epoch": 2.76, "learning_rate": 3.236450579508976e-06, "loss": 1.8334, "step": 9181 }, { "epoch": 2.76, "learning_rate": 3.228254724202806e-06, "loss": 1.7606, "step": 9182 }, { "epoch": 2.76, "learning_rate": 3.2200690892256744e-06, "loss": 1.8137, "step": 9183 }, { "epoch": 2.76, "learning_rate": 3.2118936754421014e-06, "loss": 1.8922, "step": 9184 }, { "epoch": 2.76, "learning_rate": 3.2037284837154955e-06, "loss": 1.9296, "step": 9185 }, { "epoch": 2.77, "learning_rate": 3.1955735149082122e-06, "loss": 1.8319, "step": 9186 }, { "epoch": 2.77, "learning_rate": 3.1874287698815176e-06, "loss": 1.9158, "step": 9187 }, { "epoch": 2.77, "learning_rate": 3.179294249495601e-06, "loss": 1.9212, "step": 9188 }, { "epoch": 2.77, "learning_rate": 3.1711699546095875e-06, "loss": 1.8507, "step": 9189 }, { "epoch": 2.77, "learning_rate": 3.1630558860814898e-06, "loss": 1.8203, "step": 9190 }, { "epoch": 2.77, "learning_rate": 3.154952044768256e-06, "loss": 1.802, "step": 9191 }, { "epoch": 2.77, "learning_rate": 3.1468584315257567e-06, "loss": 1.8143, "step": 9192 }, { "epoch": 2.77, "learning_rate": 3.138775047208786e-06, "loss": 1.8268, "step": 9193 }, { "epoch": 2.77, "learning_rate": 3.130701892671051e-06, "loss": 1.8336, "step": 9194 }, { "epoch": 2.77, "learning_rate": 3.122638968765179e-06, "loss": 1.8497, "step": 9195 }, { "epoch": 2.77, "learning_rate": 3.1145862763427125e-06, "loss": 1.7049, "step": 9196 }, { "epoch": 2.77, "learning_rate": 3.106543816254115e-06, "loss": 1.7879, "step": 9197 }, { "epoch": 2.77, "learning_rate": 3.098511589348796e-06, "loss": 1.9043, "step": 9198 }, { "epoch": 2.77, "learning_rate": 3.0904895964750434e-06, "loss": 1.8157, "step": 9199 }, { "epoch": 2.77, "learning_rate": 3.082477838480069e-06, "loss": 1.882, "step": 9200 }, { "epoch": 2.77, "learning_rate": 3.07447631621004e-06, "loss": 1.8323, "step": 9201 }, { "epoch": 2.77, "learning_rate": 3.0664850305100022e-06, "loss": 1.7893, "step": 9202 }, { "epoch": 2.77, "learning_rate": 3.058503982223959e-06, "loss": 1.8985, "step": 9203 }, { "epoch": 2.77, "learning_rate": 3.0505331721947916e-06, "loss": 1.7748, "step": 9204 }, { "epoch": 2.77, "learning_rate": 3.0425726012643153e-06, "loss": 1.9469, "step": 9205 }, { "epoch": 2.77, "learning_rate": 3.0346222702732797e-06, "loss": 1.76, "step": 9206 }, { "epoch": 2.77, "learning_rate": 3.0266821800613353e-06, "loss": 1.6878, "step": 9207 }, { "epoch": 2.77, "learning_rate": 3.0187523314670783e-06, "loss": 1.8034, "step": 9208 }, { "epoch": 2.77, "learning_rate": 3.010832725327972e-06, "loss": 1.8054, "step": 9209 }, { "epoch": 2.77, "learning_rate": 3.0029233624804256e-06, "loss": 1.7194, "step": 9210 }, { "epoch": 2.77, "learning_rate": 2.995024243759792e-06, "loss": 1.6891, "step": 9211 }, { "epoch": 2.77, "learning_rate": 2.9871353700003044e-06, "loss": 1.8411, "step": 9212 }, { "epoch": 2.77, "learning_rate": 2.9792567420351523e-06, "loss": 1.7807, "step": 9213 }, { "epoch": 2.77, "learning_rate": 2.971388360696381e-06, "loss": 1.7998, "step": 9214 }, { "epoch": 2.77, "learning_rate": 2.9635302268150035e-06, "loss": 1.7988, "step": 9215 }, { "epoch": 2.77, "learning_rate": 2.9556823412209555e-06, "loss": 1.8119, "step": 9216 }, { "epoch": 2.77, "learning_rate": 2.9478447047430634e-06, "loss": 1.7831, "step": 9217 }, { "epoch": 2.77, "learning_rate": 2.9400173182090873e-06, "loss": 1.8047, "step": 9218 }, { "epoch": 2.78, "learning_rate": 2.9322001824456766e-06, "loss": 1.9029, "step": 9219 }, { "epoch": 2.78, "learning_rate": 2.9243932982784493e-06, "loss": 1.8187, "step": 9220 }, { "epoch": 2.78, "learning_rate": 2.91659666653189e-06, "loss": 1.8403, "step": 9221 }, { "epoch": 2.78, "learning_rate": 2.908810288029451e-06, "loss": 1.9671, "step": 9222 }, { "epoch": 2.78, "learning_rate": 2.9010341635934302e-06, "loss": 1.7932, "step": 9223 }, { "epoch": 2.78, "learning_rate": 2.893268294045126e-06, "loss": 1.809, "step": 9224 }, { "epoch": 2.78, "learning_rate": 2.885512680204683e-06, "loss": 1.8745, "step": 9225 }, { "epoch": 2.78, "learning_rate": 2.8777673228912117e-06, "loss": 1.9209, "step": 9226 }, { "epoch": 2.78, "learning_rate": 2.870032222922725e-06, "loss": 1.9401, "step": 9227 }, { "epoch": 2.78, "learning_rate": 2.862307381116125e-06, "loss": 1.7842, "step": 9228 }, { "epoch": 2.78, "learning_rate": 2.85459279828727e-06, "loss": 1.867, "step": 9229 }, { "epoch": 2.78, "learning_rate": 2.8468884752509085e-06, "loss": 1.9376, "step": 9230 }, { "epoch": 2.78, "learning_rate": 2.8391944128207227e-06, "loss": 1.7737, "step": 9231 }, { "epoch": 2.78, "learning_rate": 2.8315106118093183e-06, "loss": 1.7908, "step": 9232 }, { "epoch": 2.78, "learning_rate": 2.823837073028168e-06, "loss": 1.8184, "step": 9233 }, { "epoch": 2.78, "learning_rate": 2.8161737972877133e-06, "loss": 1.8987, "step": 9234 }, { "epoch": 2.78, "learning_rate": 2.808520785397306e-06, "loss": 1.8686, "step": 9235 }, { "epoch": 2.78, "learning_rate": 2.8008780381651777e-06, "loss": 1.8309, "step": 9236 }, { "epoch": 2.78, "learning_rate": 2.793245556398527e-06, "loss": 1.7309, "step": 9237 }, { "epoch": 2.78, "learning_rate": 2.7856233409034094e-06, "loss": 1.6734, "step": 9238 }, { "epoch": 2.78, "learning_rate": 2.7780113924848583e-06, "loss": 1.8461, "step": 9239 }, { "epoch": 2.78, "learning_rate": 2.7704097119467642e-06, "loss": 1.8127, "step": 9240 }, { "epoch": 2.78, "learning_rate": 2.7628183000919847e-06, "loss": 1.7302, "step": 9241 }, { "epoch": 2.78, "learning_rate": 2.755237157722268e-06, "loss": 1.7787, "step": 9242 }, { "epoch": 2.78, "learning_rate": 2.7476662856382617e-06, "loss": 1.9732, "step": 9243 }, { "epoch": 2.78, "learning_rate": 2.740105684639549e-06, "loss": 1.8725, "step": 9244 }, { "epoch": 2.78, "learning_rate": 2.7325553555246465e-06, "loss": 1.862, "step": 9245 }, { "epoch": 2.78, "learning_rate": 2.7250152990909496e-06, "loss": 1.6969, "step": 9246 }, { "epoch": 2.78, "learning_rate": 2.7174855161347877e-06, "loss": 1.8915, "step": 9247 }, { "epoch": 2.78, "learning_rate": 2.7099660074513923e-06, "loss": 1.848, "step": 9248 }, { "epoch": 2.78, "learning_rate": 2.7024567738349383e-06, "loss": 1.7567, "step": 9249 }, { "epoch": 2.78, "learning_rate": 2.6949578160784805e-06, "loss": 1.7745, "step": 9250 }, { "epoch": 2.78, "learning_rate": 2.687469134974019e-06, "loss": 1.8101, "step": 9251 }, { "epoch": 2.79, "learning_rate": 2.6799907313124316e-06, "loss": 1.8752, "step": 9252 }, { "epoch": 2.79, "learning_rate": 2.672522605883543e-06, "loss": 1.8339, "step": 9253 }, { "epoch": 2.79, "learning_rate": 2.6650647594760993e-06, "loss": 1.9054, "step": 9254 }, { "epoch": 2.79, "learning_rate": 2.657617192877715e-06, "loss": 1.7797, "step": 9255 }, { "epoch": 2.79, "learning_rate": 2.650179906874983e-06, "loss": 1.8139, "step": 9256 }, { "epoch": 2.79, "learning_rate": 2.6427529022533424e-06, "loss": 1.7787, "step": 9257 }, { "epoch": 2.79, "learning_rate": 2.6353361797971987e-06, "loss": 1.8538, "step": 9258 }, { "epoch": 2.79, "learning_rate": 2.627929740289836e-06, "loss": 1.7753, "step": 9259 }, { "epoch": 2.79, "learning_rate": 2.620533584513485e-06, "loss": 1.8453, "step": 9260 }, { "epoch": 2.79, "learning_rate": 2.6131477132492756e-06, "loss": 1.875, "step": 9261 }, { "epoch": 2.79, "learning_rate": 2.6057721272772397e-06, "loss": 1.7816, "step": 9262 }, { "epoch": 2.79, "learning_rate": 2.5984068273763205e-06, "loss": 1.9005, "step": 9263 }, { "epoch": 2.79, "learning_rate": 2.591051814324419e-06, "loss": 1.7318, "step": 9264 }, { "epoch": 2.79, "learning_rate": 2.5837070888982905e-06, "loss": 1.8021, "step": 9265 }, { "epoch": 2.79, "learning_rate": 2.57637265187366e-06, "loss": 1.9388, "step": 9266 }, { "epoch": 2.79, "learning_rate": 2.5690485040251066e-06, "loss": 1.9102, "step": 9267 }, { "epoch": 2.79, "learning_rate": 2.561734646126168e-06, "loss": 1.8234, "step": 9268 }, { "epoch": 2.79, "learning_rate": 2.554431078949282e-06, "loss": 1.747, "step": 9269 }, { "epoch": 2.79, "learning_rate": 2.547137803265798e-06, "loss": 1.8729, "step": 9270 }, { "epoch": 2.79, "learning_rate": 2.5398548198459773e-06, "loss": 1.8587, "step": 9271 }, { "epoch": 2.79, "learning_rate": 2.5325821294589936e-06, "loss": 1.8079, "step": 9272 }, { "epoch": 2.79, "learning_rate": 2.525319732872944e-06, "loss": 1.7285, "step": 9273 }, { "epoch": 2.79, "learning_rate": 2.5180676308548144e-06, "loss": 1.7523, "step": 9274 }, { "epoch": 2.79, "learning_rate": 2.5108258241705486e-06, "loss": 1.8076, "step": 9275 }, { "epoch": 2.79, "learning_rate": 2.503594313584934e-06, "loss": 1.9407, "step": 9276 }, { "epoch": 2.79, "learning_rate": 2.4963730998617263e-06, "loss": 1.8284, "step": 9277 }, { "epoch": 2.79, "learning_rate": 2.489162183763594e-06, "loss": 1.7906, "step": 9278 }, { "epoch": 2.79, "learning_rate": 2.481961566052082e-06, "loss": 1.7433, "step": 9279 }, { "epoch": 2.79, "learning_rate": 2.474771247487695e-06, "loss": 1.8259, "step": 9280 }, { "epoch": 2.79, "learning_rate": 2.46759122882978e-06, "loss": 1.8129, "step": 9281 }, { "epoch": 2.79, "learning_rate": 2.4604215108366522e-06, "loss": 1.8892, "step": 9282 }, { "epoch": 2.79, "learning_rate": 2.4532620942655514e-06, "loss": 1.7884, "step": 9283 }, { "epoch": 2.79, "learning_rate": 2.446112979872572e-06, "loss": 1.7503, "step": 9284 }, { "epoch": 2.8, "learning_rate": 2.4389741684127885e-06, "loss": 1.7594, "step": 9285 }, { "epoch": 2.8, "learning_rate": 2.4318456606401085e-06, "loss": 1.8972, "step": 9286 }, { "epoch": 2.8, "learning_rate": 2.4247274573074075e-06, "loss": 1.7858, "step": 9287 }, { "epoch": 2.8, "learning_rate": 2.4176195591664617e-06, "loss": 1.8515, "step": 9288 }, { "epoch": 2.8, "learning_rate": 2.4105219669679603e-06, "loss": 1.8167, "step": 9289 }, { "epoch": 2.8, "learning_rate": 2.4034346814615026e-06, "loss": 1.8331, "step": 9290 }, { "epoch": 2.8, "learning_rate": 2.3963577033955687e-06, "loss": 1.8054, "step": 9291 }, { "epoch": 2.8, "learning_rate": 2.3892910335176156e-06, "loss": 1.8179, "step": 9292 }, { "epoch": 2.8, "learning_rate": 2.3822346725739464e-06, "loss": 1.8441, "step": 9293 }, { "epoch": 2.8, "learning_rate": 2.3751886213098098e-06, "loss": 1.88, "step": 9294 }, { "epoch": 2.8, "learning_rate": 2.3681528804693763e-06, "loss": 2.0052, "step": 9295 }, { "epoch": 2.8, "learning_rate": 2.3611274507956638e-06, "loss": 1.7371, "step": 9296 }, { "epoch": 2.8, "learning_rate": 2.3541123330307115e-06, "loss": 1.8697, "step": 9297 }, { "epoch": 2.8, "learning_rate": 2.3471075279153485e-06, "loss": 1.8959, "step": 9298 }, { "epoch": 2.8, "learning_rate": 2.3401130361894174e-06, "loss": 1.911, "step": 9299 }, { "epoch": 2.8, "learning_rate": 2.3331288585915822e-06, "loss": 1.8363, "step": 9300 }, { "epoch": 2.8, "learning_rate": 2.326154995859475e-06, "loss": 1.7723, "step": 9301 }, { "epoch": 2.8, "learning_rate": 2.3191914487296516e-06, "loss": 1.8274, "step": 9302 }, { "epoch": 2.8, "learning_rate": 2.3122382179375123e-06, "loss": 1.8339, "step": 9303 }, { "epoch": 2.8, "learning_rate": 2.3052953042174474e-06, "loss": 1.8099, "step": 9304 }, { "epoch": 2.8, "learning_rate": 2.2983627083026704e-06, "loss": 1.8194, "step": 9305 }, { "epoch": 2.8, "learning_rate": 2.2914404309253734e-06, "loss": 1.9327, "step": 9306 }, { "epoch": 2.8, "learning_rate": 2.2845284728166603e-06, "loss": 1.8308, "step": 9307 }, { "epoch": 2.8, "learning_rate": 2.277626834706492e-06, "loss": 1.7207, "step": 9308 }, { "epoch": 2.8, "learning_rate": 2.2707355173237852e-06, "loss": 1.8623, "step": 9309 }, { "epoch": 2.8, "learning_rate": 2.2638545213963357e-06, "loss": 1.7721, "step": 9310 }, { "epoch": 2.8, "learning_rate": 2.2569838476508844e-06, "loss": 1.763, "step": 9311 }, { "epoch": 2.8, "learning_rate": 2.250123496813039e-06, "loss": 1.9043, "step": 9312 }, { "epoch": 2.8, "learning_rate": 2.2432734696073434e-06, "loss": 1.8973, "step": 9313 }, { "epoch": 2.8, "learning_rate": 2.236433766757273e-06, "loss": 1.7927, "step": 9314 }, { "epoch": 2.8, "learning_rate": 2.2296043889851625e-06, "loss": 1.8508, "step": 9315 }, { "epoch": 2.8, "learning_rate": 2.22278533701229e-06, "loss": 1.8274, "step": 9316 }, { "epoch": 2.8, "learning_rate": 2.2159766115588342e-06, "loss": 1.8333, "step": 9317 }, { "epoch": 2.8, "learning_rate": 2.2091782133438763e-06, "loss": 1.9676, "step": 9318 }, { "epoch": 2.81, "learning_rate": 2.202390143085431e-06, "loss": 1.7827, "step": 9319 }, { "epoch": 2.81, "learning_rate": 2.19561240150038e-06, "loss": 1.8263, "step": 9320 }, { "epoch": 2.81, "learning_rate": 2.1888449893045725e-06, "loss": 1.9557, "step": 9321 }, { "epoch": 2.81, "learning_rate": 2.182087907212693e-06, "loss": 1.7862, "step": 9322 }, { "epoch": 2.81, "learning_rate": 2.175341155938415e-06, "loss": 1.7975, "step": 9323 }, { "epoch": 2.81, "learning_rate": 2.1686047361942354e-06, "loss": 1.9419, "step": 9324 }, { "epoch": 2.81, "learning_rate": 2.1618786486916507e-06, "loss": 1.7951, "step": 9325 }, { "epoch": 2.81, "learning_rate": 2.155162894141016e-06, "loss": 1.856, "step": 9326 }, { "epoch": 2.81, "learning_rate": 2.148457473251575e-06, "loss": 1.8103, "step": 9327 }, { "epoch": 2.81, "learning_rate": 2.141762386731516e-06, "loss": 1.7715, "step": 9328 }, { "epoch": 2.81, "learning_rate": 2.1350776352879298e-06, "loss": 1.758, "step": 9329 }, { "epoch": 2.81, "learning_rate": 2.1284032196268067e-06, "loss": 1.757, "step": 9330 }, { "epoch": 2.81, "learning_rate": 2.1217391404530606e-06, "loss": 1.7013, "step": 9331 }, { "epoch": 2.81, "learning_rate": 2.1150853984704955e-06, "loss": 1.7292, "step": 9332 }, { "epoch": 2.81, "learning_rate": 2.108441994381838e-06, "loss": 1.7892, "step": 9333 }, { "epoch": 2.81, "learning_rate": 2.101808928888693e-06, "loss": 1.8508, "step": 9334 }, { "epoch": 2.81, "learning_rate": 2.0951862026916236e-06, "loss": 1.9464, "step": 9335 }, { "epoch": 2.81, "learning_rate": 2.088573816490058e-06, "loss": 1.736, "step": 9336 }, { "epoch": 2.81, "learning_rate": 2.0819717709823605e-06, "loss": 1.9008, "step": 9337 }, { "epoch": 2.81, "learning_rate": 2.075380066865784e-06, "loss": 1.7396, "step": 9338 }, { "epoch": 2.81, "learning_rate": 2.068798704836494e-06, "loss": 1.8531, "step": 9339 }, { "epoch": 2.81, "learning_rate": 2.0622276855895793e-06, "loss": 1.9237, "step": 9340 }, { "epoch": 2.81, "learning_rate": 2.0556670098189955e-06, "loss": 1.7877, "step": 9341 }, { "epoch": 2.81, "learning_rate": 2.0491166782176664e-06, "loss": 1.8174, "step": 9342 }, { "epoch": 2.81, "learning_rate": 2.0425766914773712e-06, "loss": 1.7862, "step": 9343 }, { "epoch": 2.81, "learning_rate": 2.036047050288814e-06, "loss": 1.8558, "step": 9344 }, { "epoch": 2.81, "learning_rate": 2.0295277553416315e-06, "loss": 1.9216, "step": 9345 }, { "epoch": 2.81, "learning_rate": 2.023018807324306e-06, "loss": 1.9184, "step": 9346 }, { "epoch": 2.81, "learning_rate": 2.0165202069243106e-06, "loss": 1.9214, "step": 9347 }, { "epoch": 2.81, "learning_rate": 2.010031954827929e-06, "loss": 1.9604, "step": 9348 }, { "epoch": 2.81, "learning_rate": 2.0035540517204465e-06, "loss": 1.9187, "step": 9349 }, { "epoch": 2.81, "learning_rate": 1.9970864982859937e-06, "loss": 1.7009, "step": 9350 }, { "epoch": 2.81, "learning_rate": 1.9906292952076354e-06, "loss": 1.8376, "step": 9351 }, { "epoch": 2.82, "learning_rate": 1.984182443167315e-06, "loss": 1.8822, "step": 9352 }, { "epoch": 2.82, "learning_rate": 1.977745942845921e-06, "loss": 1.8132, "step": 9353 }, { "epoch": 2.82, "learning_rate": 1.971319794923221e-06, "loss": 1.7038, "step": 9354 }, { "epoch": 2.82, "learning_rate": 1.964904000077916e-06, "loss": 1.8362, "step": 9355 }, { "epoch": 2.82, "learning_rate": 1.9584985589875646e-06, "loss": 1.7919, "step": 9356 }, { "epoch": 2.82, "learning_rate": 1.952103472328681e-06, "loss": 1.8953, "step": 9357 }, { "epoch": 2.82, "learning_rate": 1.9457187407766696e-06, "loss": 1.867, "step": 9358 }, { "epoch": 2.82, "learning_rate": 1.9393443650058463e-06, "loss": 1.7914, "step": 9359 }, { "epoch": 2.82, "learning_rate": 1.9329803456894057e-06, "loss": 1.7331, "step": 9360 }, { "epoch": 2.82, "learning_rate": 1.9266266834994774e-06, "loss": 1.7807, "step": 9361 }, { "epoch": 2.82, "learning_rate": 1.920283379107091e-06, "loss": 1.7293, "step": 9362 }, { "epoch": 2.82, "learning_rate": 1.9139504331821768e-06, "loss": 1.774, "step": 9363 }, { "epoch": 2.82, "learning_rate": 1.9076278463935893e-06, "loss": 1.8313, "step": 9364 }, { "epoch": 2.82, "learning_rate": 1.901315619409061e-06, "loss": 1.9002, "step": 9365 }, { "epoch": 2.82, "learning_rate": 1.895013752895236e-06, "loss": 1.7727, "step": 9366 }, { "epoch": 2.82, "learning_rate": 1.8887222475176936e-06, "loss": 1.8201, "step": 9367 }, { "epoch": 2.82, "learning_rate": 1.8824411039408796e-06, "loss": 1.7108, "step": 9368 }, { "epoch": 2.82, "learning_rate": 1.8761703228281747e-06, "loss": 1.8511, "step": 9369 }, { "epoch": 2.82, "learning_rate": 1.869909904841838e-06, "loss": 1.8093, "step": 9370 }, { "epoch": 2.82, "learning_rate": 1.8636598506430514e-06, "loss": 1.7384, "step": 9371 }, { "epoch": 2.82, "learning_rate": 1.8574201608919095e-06, "loss": 1.8747, "step": 9372 }, { "epoch": 2.82, "learning_rate": 1.8511908362474073e-06, "loss": 1.8454, "step": 9373 }, { "epoch": 2.82, "learning_rate": 1.8449718773674406e-06, "loss": 1.9009, "step": 9374 }, { "epoch": 2.82, "learning_rate": 1.8387632849087955e-06, "loss": 1.789, "step": 9375 }, { "epoch": 2.82, "learning_rate": 1.8325650595271804e-06, "loss": 1.9237, "step": 9376 }, { "epoch": 2.82, "learning_rate": 1.8263772018772163e-06, "loss": 1.8952, "step": 9377 }, { "epoch": 2.82, "learning_rate": 1.8201997126124137e-06, "loss": 1.847, "step": 9378 }, { "epoch": 2.82, "learning_rate": 1.8140325923852064e-06, "loss": 1.8622, "step": 9379 }, { "epoch": 2.82, "learning_rate": 1.8078758418469065e-06, "loss": 1.843, "step": 9380 }, { "epoch": 2.82, "learning_rate": 1.80172946164775e-06, "loss": 1.8849, "step": 9381 }, { "epoch": 2.82, "learning_rate": 1.7955934524368723e-06, "loss": 1.8469, "step": 9382 }, { "epoch": 2.82, "learning_rate": 1.7894678148623222e-06, "loss": 1.918, "step": 9383 }, { "epoch": 2.82, "learning_rate": 1.783352549571038e-06, "loss": 1.7138, "step": 9384 }, { "epoch": 2.83, "learning_rate": 1.777247657208858e-06, "loss": 1.9176, "step": 9385 }, { "epoch": 2.83, "learning_rate": 1.771153138420556e-06, "loss": 1.9153, "step": 9386 }, { "epoch": 2.83, "learning_rate": 1.7650689938497834e-06, "loss": 1.751, "step": 9387 }, { "epoch": 2.83, "learning_rate": 1.7589952241391038e-06, "loss": 1.8347, "step": 9388 }, { "epoch": 2.83, "learning_rate": 1.7529318299299824e-06, "loss": 1.8684, "step": 9389 }, { "epoch": 2.83, "learning_rate": 1.7468788118627844e-06, "loss": 1.7565, "step": 9390 }, { "epoch": 2.83, "learning_rate": 1.7408361705767984e-06, "loss": 1.8136, "step": 9391 }, { "epoch": 2.83, "learning_rate": 1.734803906710203e-06, "loss": 1.8397, "step": 9392 }, { "epoch": 2.83, "learning_rate": 1.7287820209000882e-06, "loss": 1.8075, "step": 9393 }, { "epoch": 2.83, "learning_rate": 1.7227705137824123e-06, "loss": 1.826, "step": 9394 }, { "epoch": 2.83, "learning_rate": 1.7167693859921007e-06, "loss": 1.772, "step": 9395 }, { "epoch": 2.83, "learning_rate": 1.7107786381629243e-06, "loss": 1.8598, "step": 9396 }, { "epoch": 2.83, "learning_rate": 1.704798270927599e-06, "loss": 1.7844, "step": 9397 }, { "epoch": 2.83, "learning_rate": 1.6988282849177306e-06, "loss": 1.8896, "step": 9398 }, { "epoch": 2.83, "learning_rate": 1.692868680763804e-06, "loss": 1.8544, "step": 9399 }, { "epoch": 2.83, "learning_rate": 1.6869194590952487e-06, "loss": 1.8732, "step": 9400 }, { "epoch": 2.83, "learning_rate": 1.6809806205403622e-06, "loss": 1.8739, "step": 9401 }, { "epoch": 2.83, "learning_rate": 1.6750521657263761e-06, "loss": 1.9594, "step": 9402 }, { "epoch": 2.83, "learning_rate": 1.6691340952794232e-06, "loss": 1.8515, "step": 9403 }, { "epoch": 2.83, "learning_rate": 1.6632264098244921e-06, "loss": 1.8872, "step": 9404 }, { "epoch": 2.83, "learning_rate": 1.6573291099855282e-06, "loss": 1.717, "step": 9405 }, { "epoch": 2.83, "learning_rate": 1.6514421963853555e-06, "loss": 1.8556, "step": 9406 }, { "epoch": 2.83, "learning_rate": 1.6455656696457212e-06, "loss": 1.7618, "step": 9407 }, { "epoch": 2.83, "learning_rate": 1.639699530387262e-06, "loss": 1.8414, "step": 9408 }, { "epoch": 2.83, "learning_rate": 1.6338437792294936e-06, "loss": 1.7169, "step": 9409 }, { "epoch": 2.83, "learning_rate": 1.6279984167908769e-06, "loss": 1.8853, "step": 9410 }, { "epoch": 2.83, "learning_rate": 1.62216344368874e-06, "loss": 1.8674, "step": 9411 }, { "epoch": 2.83, "learning_rate": 1.616338860539357e-06, "loss": 1.9241, "step": 9412 }, { "epoch": 2.83, "learning_rate": 1.6105246679578578e-06, "loss": 1.7975, "step": 9413 }, { "epoch": 2.83, "learning_rate": 1.6047208665582959e-06, "loss": 1.7586, "step": 9414 }, { "epoch": 2.83, "learning_rate": 1.5989274569536362e-06, "loss": 1.7728, "step": 9415 }, { "epoch": 2.83, "learning_rate": 1.5931444397557338e-06, "loss": 1.8085, "step": 9416 }, { "epoch": 2.83, "learning_rate": 1.5873718155753448e-06, "loss": 1.8143, "step": 9417 }, { "epoch": 2.84, "learning_rate": 1.5816095850221257e-06, "loss": 1.8534, "step": 9418 }, { "epoch": 2.84, "learning_rate": 1.5758577487046567e-06, "loss": 1.8198, "step": 9419 }, { "epoch": 2.84, "learning_rate": 1.570116307230396e-06, "loss": 1.942, "step": 9420 }, { "epoch": 2.84, "learning_rate": 1.5643852612057142e-06, "loss": 1.8227, "step": 9421 }, { "epoch": 2.84, "learning_rate": 1.5586646112358939e-06, "loss": 1.8686, "step": 9422 }, { "epoch": 2.84, "learning_rate": 1.5529543579250848e-06, "loss": 1.7821, "step": 9423 }, { "epoch": 2.84, "learning_rate": 1.5472545018763718e-06, "loss": 1.7478, "step": 9424 }, { "epoch": 2.84, "learning_rate": 1.5415650436917395e-06, "loss": 1.8469, "step": 9425 }, { "epoch": 2.84, "learning_rate": 1.5358859839720741e-06, "loss": 1.9656, "step": 9426 }, { "epoch": 2.84, "learning_rate": 1.5302173233171402e-06, "loss": 1.8491, "step": 9427 }, { "epoch": 2.84, "learning_rate": 1.5245590623256256e-06, "loss": 1.7498, "step": 9428 }, { "epoch": 2.84, "learning_rate": 1.518911201595119e-06, "loss": 1.8583, "step": 9429 }, { "epoch": 2.84, "learning_rate": 1.5132737417220877e-06, "loss": 1.7564, "step": 9430 }, { "epoch": 2.84, "learning_rate": 1.5076466833019554e-06, "loss": 1.8094, "step": 9431 }, { "epoch": 2.84, "learning_rate": 1.50203002692898e-06, "loss": 1.8724, "step": 9432 }, { "epoch": 2.84, "learning_rate": 1.4964237731963648e-06, "loss": 1.8475, "step": 9433 }, { "epoch": 2.84, "learning_rate": 1.490827922696203e-06, "loss": 1.8815, "step": 9434 }, { "epoch": 2.84, "learning_rate": 1.4852424760194772e-06, "loss": 1.7688, "step": 9435 }, { "epoch": 2.84, "learning_rate": 1.4796674337560935e-06, "loss": 1.89, "step": 9436 }, { "epoch": 2.84, "learning_rate": 1.4741027964948362e-06, "loss": 1.8224, "step": 9437 }, { "epoch": 2.84, "learning_rate": 1.4685485648234132e-06, "loss": 1.7766, "step": 9438 }, { "epoch": 2.84, "learning_rate": 1.4630047393284107e-06, "loss": 1.8683, "step": 9439 }, { "epoch": 2.84, "learning_rate": 1.4574713205953383e-06, "loss": 1.9188, "step": 9440 }, { "epoch": 2.84, "learning_rate": 1.451948309208595e-06, "loss": 1.86, "step": 9441 }, { "epoch": 2.84, "learning_rate": 1.44643570575147e-06, "loss": 1.8404, "step": 9442 }, { "epoch": 2.84, "learning_rate": 1.4409335108061639e-06, "loss": 1.8698, "step": 9443 }, { "epoch": 2.84, "learning_rate": 1.435441724953812e-06, "loss": 1.8534, "step": 9444 }, { "epoch": 2.84, "learning_rate": 1.429960348774373e-06, "loss": 1.7279, "step": 9445 }, { "epoch": 2.84, "learning_rate": 1.424489382846772e-06, "loss": 1.8055, "step": 9446 }, { "epoch": 2.84, "learning_rate": 1.419028827748803e-06, "loss": 1.7625, "step": 9447 }, { "epoch": 2.84, "learning_rate": 1.4135786840571818e-06, "loss": 1.8152, "step": 9448 }, { "epoch": 2.84, "learning_rate": 1.4081389523475041e-06, "loss": 1.7654, "step": 9449 }, { "epoch": 2.84, "learning_rate": 1.4027096331942768e-06, "loss": 1.8481, "step": 9450 }, { "epoch": 2.84, "learning_rate": 1.3972907271709078e-06, "loss": 1.8741, "step": 9451 }, { "epoch": 2.85, "learning_rate": 1.3918822348497063e-06, "loss": 1.8727, "step": 9452 }, { "epoch": 2.85, "learning_rate": 1.3864841568018593e-06, "loss": 1.8562, "step": 9453 }, { "epoch": 2.85, "learning_rate": 1.3810964935975e-06, "loss": 1.8718, "step": 9454 }, { "epoch": 2.85, "learning_rate": 1.3757192458056067e-06, "loss": 1.9337, "step": 9455 }, { "epoch": 2.85, "learning_rate": 1.3703524139941138e-06, "loss": 1.8224, "step": 9456 }, { "epoch": 2.85, "learning_rate": 1.36499599872979e-06, "loss": 1.8213, "step": 9457 }, { "epoch": 2.85, "learning_rate": 1.359650000578394e-06, "loss": 1.7881, "step": 9458 }, { "epoch": 2.85, "learning_rate": 1.3543144201044744e-06, "loss": 1.923, "step": 9459 }, { "epoch": 2.85, "learning_rate": 1.3489892578715802e-06, "loss": 1.9199, "step": 9460 }, { "epoch": 2.85, "learning_rate": 1.3436745144420837e-06, "loss": 1.812, "step": 9461 }, { "epoch": 2.85, "learning_rate": 1.3383701903773026e-06, "loss": 1.874, "step": 9462 }, { "epoch": 2.85, "learning_rate": 1.333076286237467e-06, "loss": 1.865, "step": 9463 }, { "epoch": 2.85, "learning_rate": 1.3277928025816289e-06, "loss": 1.8025, "step": 9464 }, { "epoch": 2.85, "learning_rate": 1.3225197399678425e-06, "loss": 1.9115, "step": 9465 }, { "epoch": 2.85, "learning_rate": 1.3172570989529731e-06, "loss": 1.8145, "step": 9466 }, { "epoch": 2.85, "learning_rate": 1.3120048800928319e-06, "loss": 1.7782, "step": 9467 }, { "epoch": 2.85, "learning_rate": 1.3067630839421308e-06, "loss": 1.7254, "step": 9468 }, { "epoch": 2.85, "learning_rate": 1.301531711054471e-06, "loss": 1.8363, "step": 9469 }, { "epoch": 2.85, "learning_rate": 1.2963107619823445e-06, "loss": 1.7667, "step": 9470 }, { "epoch": 2.85, "learning_rate": 1.291100237277132e-06, "loss": 1.817, "step": 9471 }, { "epoch": 2.85, "learning_rate": 1.285900137489171e-06, "loss": 1.7693, "step": 9472 }, { "epoch": 2.85, "learning_rate": 1.2807104631676336e-06, "loss": 1.7416, "step": 9473 }, { "epoch": 2.85, "learning_rate": 1.2755312148606147e-06, "loss": 1.8025, "step": 9474 }, { "epoch": 2.85, "learning_rate": 1.2703623931151209e-06, "loss": 1.8005, "step": 9475 }, { "epoch": 2.85, "learning_rate": 1.2652039984770158e-06, "loss": 1.8302, "step": 9476 }, { "epoch": 2.85, "learning_rate": 1.2600560314911413e-06, "loss": 1.7969, "step": 9477 }, { "epoch": 2.85, "learning_rate": 1.2549184927011404e-06, "loss": 1.8931, "step": 9478 }, { "epoch": 2.85, "learning_rate": 1.2497913826496344e-06, "loss": 1.8939, "step": 9479 }, { "epoch": 2.85, "learning_rate": 1.2446747018781012e-06, "loss": 1.7922, "step": 9480 }, { "epoch": 2.85, "learning_rate": 1.23956845092692e-06, "loss": 1.8921, "step": 9481 }, { "epoch": 2.85, "learning_rate": 1.2344726303353927e-06, "loss": 1.7881, "step": 9482 }, { "epoch": 2.85, "learning_rate": 1.2293872406416884e-06, "loss": 1.8625, "step": 9483 }, { "epoch": 2.85, "learning_rate": 1.224312282382889e-06, "loss": 1.8305, "step": 9484 }, { "epoch": 2.86, "learning_rate": 1.2192477560949767e-06, "loss": 1.8463, "step": 9485 }, { "epoch": 2.86, "learning_rate": 1.2141936623128347e-06, "loss": 1.8429, "step": 9486 }, { "epoch": 2.86, "learning_rate": 1.2091500015702474e-06, "loss": 1.836, "step": 9487 }, { "epoch": 2.86, "learning_rate": 1.204116774399866e-06, "loss": 1.774, "step": 9488 }, { "epoch": 2.86, "learning_rate": 1.199093981333288e-06, "loss": 1.7483, "step": 9489 }, { "epoch": 2.86, "learning_rate": 1.1940816229009555e-06, "loss": 1.8346, "step": 9490 }, { "epoch": 2.86, "learning_rate": 1.1890796996322672e-06, "loss": 1.9226, "step": 9491 }, { "epoch": 2.86, "learning_rate": 1.1840882120554898e-06, "loss": 1.8153, "step": 9492 }, { "epoch": 2.86, "learning_rate": 1.1791071606977565e-06, "loss": 1.912, "step": 9493 }, { "epoch": 2.86, "learning_rate": 1.1741365460851695e-06, "loss": 1.7772, "step": 9494 }, { "epoch": 2.86, "learning_rate": 1.1691763687426416e-06, "loss": 1.933, "step": 9495 }, { "epoch": 2.86, "learning_rate": 1.1642266291940762e-06, "loss": 1.844, "step": 9496 }, { "epoch": 2.86, "learning_rate": 1.1592873279622107e-06, "loss": 1.94, "step": 9497 }, { "epoch": 2.86, "learning_rate": 1.1543584655686944e-06, "loss": 1.8383, "step": 9498 }, { "epoch": 2.86, "learning_rate": 1.149440042534078e-06, "loss": 1.8003, "step": 9499 }, { "epoch": 2.86, "learning_rate": 1.144532059377823e-06, "loss": 1.8075, "step": 9500 }, { "epoch": 2.86, "learning_rate": 1.1396345166182598e-06, "loss": 1.849, "step": 9501 }, { "epoch": 2.86, "learning_rate": 1.13474741477263e-06, "loss": 1.7577, "step": 9502 }, { "epoch": 2.86, "learning_rate": 1.1298707543570875e-06, "loss": 1.7063, "step": 9503 }, { "epoch": 2.86, "learning_rate": 1.1250045358866646e-06, "loss": 1.7959, "step": 9504 }, { "epoch": 2.86, "learning_rate": 1.1201487598752947e-06, "loss": 1.8962, "step": 9505 }, { "epoch": 2.86, "learning_rate": 1.115303426835812e-06, "loss": 1.7448, "step": 9506 }, { "epoch": 2.86, "learning_rate": 1.1104685372799295e-06, "loss": 1.728, "step": 9507 }, { "epoch": 2.86, "learning_rate": 1.1056440917182941e-06, "loss": 1.8567, "step": 9508 }, { "epoch": 2.86, "learning_rate": 1.1008300906604208e-06, "loss": 1.7773, "step": 9509 }, { "epoch": 2.86, "learning_rate": 1.0960265346147247e-06, "loss": 1.781, "step": 9510 }, { "epoch": 2.86, "learning_rate": 1.0912334240885335e-06, "loss": 1.8635, "step": 9511 }, { "epoch": 2.86, "learning_rate": 1.0864507595880536e-06, "loss": 1.7991, "step": 9512 }, { "epoch": 2.86, "learning_rate": 1.0816785416184028e-06, "loss": 1.7962, "step": 9513 }, { "epoch": 2.86, "learning_rate": 1.0769167706835559e-06, "loss": 1.8381, "step": 9514 }, { "epoch": 2.86, "learning_rate": 1.0721654472864661e-06, "loss": 1.8713, "step": 9515 }, { "epoch": 2.86, "learning_rate": 1.06742457192891e-06, "loss": 1.7979, "step": 9516 }, { "epoch": 2.86, "learning_rate": 1.0626941451115757e-06, "loss": 1.7721, "step": 9517 }, { "epoch": 2.87, "learning_rate": 1.0579741673340637e-06, "loss": 1.8067, "step": 9518 }, { "epoch": 2.87, "learning_rate": 1.0532646390948642e-06, "loss": 1.7516, "step": 9519 }, { "epoch": 2.87, "learning_rate": 1.048565560891368e-06, "loss": 1.695, "step": 9520 }, { "epoch": 2.87, "learning_rate": 1.0438769332198562e-06, "loss": 1.8307, "step": 9521 }, { "epoch": 2.87, "learning_rate": 1.039198756575499e-06, "loss": 1.82, "step": 9522 }, { "epoch": 2.87, "learning_rate": 1.0345310314523794e-06, "loss": 1.8041, "step": 9523 }, { "epoch": 2.87, "learning_rate": 1.0298737583434693e-06, "loss": 1.8266, "step": 9524 }, { "epoch": 2.87, "learning_rate": 1.025226937740631e-06, "loss": 1.9186, "step": 9525 }, { "epoch": 2.87, "learning_rate": 1.0205905701346385e-06, "loss": 1.8644, "step": 9526 }, { "epoch": 2.87, "learning_rate": 1.0159646560151337e-06, "loss": 1.9025, "step": 9527 }, { "epoch": 2.87, "learning_rate": 1.011349195870681e-06, "loss": 1.7732, "step": 9528 }, { "epoch": 2.87, "learning_rate": 1.0067441901887353e-06, "loss": 1.8657, "step": 9529 }, { "epoch": 2.87, "learning_rate": 1.0021496394556517e-06, "loss": 1.8394, "step": 9530 }, { "epoch": 2.87, "learning_rate": 9.975655441566534e-07, "loss": 1.7424, "step": 9531 }, { "epoch": 2.87, "learning_rate": 9.929919047758861e-07, "loss": 1.7336, "step": 9532 }, { "epoch": 2.87, "learning_rate": 9.884287217963861e-07, "loss": 1.9187, "step": 9533 }, { "epoch": 2.87, "learning_rate": 9.838759957000787e-07, "loss": 1.8757, "step": 9534 }, { "epoch": 2.87, "learning_rate": 9.793337269678127e-07, "loss": 1.7746, "step": 9535 }, { "epoch": 2.87, "learning_rate": 9.74801916079282e-07, "loss": 1.8008, "step": 9536 }, { "epoch": 2.87, "learning_rate": 9.702805635131152e-07, "loss": 1.8407, "step": 9537 }, { "epoch": 2.87, "learning_rate": 9.65769669746819e-07, "loss": 1.8516, "step": 9538 }, { "epoch": 2.87, "learning_rate": 9.61269235256812e-07, "loss": 1.82, "step": 9539 }, { "epoch": 2.87, "learning_rate": 9.567792605183922e-07, "loss": 1.8503, "step": 9540 }, { "epoch": 2.87, "learning_rate": 9.522997460057581e-07, "loss": 1.7404, "step": 9541 }, { "epoch": 2.87, "learning_rate": 9.478306921919977e-07, "loss": 1.8597, "step": 9542 }, { "epoch": 2.87, "learning_rate": 9.433720995491002e-07, "loss": 1.7897, "step": 9543 }, { "epoch": 2.87, "learning_rate": 9.389239685479556e-07, "loss": 1.9017, "step": 9544 }, { "epoch": 2.87, "learning_rate": 9.344862996583547e-07, "loss": 1.7679, "step": 9545 }, { "epoch": 2.87, "learning_rate": 9.30059093348945e-07, "loss": 1.7817, "step": 9546 }, { "epoch": 2.87, "learning_rate": 9.25642350087319e-07, "loss": 1.6946, "step": 9547 }, { "epoch": 2.87, "learning_rate": 9.212360703399259e-07, "loss": 1.8242, "step": 9548 }, { "epoch": 2.87, "learning_rate": 9.168402545721377e-07, "loss": 1.7711, "step": 9549 }, { "epoch": 2.87, "learning_rate": 9.124549032481944e-07, "loss": 1.8153, "step": 9550 }, { "epoch": 2.88, "learning_rate": 9.080800168312475e-07, "loss": 1.9057, "step": 9551 }, { "epoch": 2.88, "learning_rate": 9.037155957833499e-07, "loss": 1.7867, "step": 9552 }, { "epoch": 2.88, "learning_rate": 8.993616405654437e-07, "loss": 1.8546, "step": 9553 }, { "epoch": 2.88, "learning_rate": 8.950181516373501e-07, "loss": 1.7933, "step": 9554 }, { "epoch": 2.88, "learning_rate": 8.906851294578022e-07, "loss": 1.814, "step": 9555 }, { "epoch": 2.88, "learning_rate": 8.863625744844118e-07, "loss": 1.8426, "step": 9556 }, { "epoch": 2.88, "learning_rate": 8.820504871737134e-07, "loss": 1.7207, "step": 9557 }, { "epoch": 2.88, "learning_rate": 8.777488679811097e-07, "loss": 1.8159, "step": 9558 }, { "epoch": 2.88, "learning_rate": 8.734577173609037e-07, "loss": 1.9857, "step": 9559 }, { "epoch": 2.88, "learning_rate": 8.691770357662998e-07, "loss": 1.8652, "step": 9560 }, { "epoch": 2.88, "learning_rate": 8.649068236493918e-07, "loss": 1.867, "step": 9561 }, { "epoch": 2.88, "learning_rate": 8.606470814611745e-07, "loss": 1.9023, "step": 9562 }, { "epoch": 2.88, "learning_rate": 8.563978096515102e-07, "loss": 1.8206, "step": 9563 }, { "epoch": 2.88, "learning_rate": 8.521590086692066e-07, "loss": 1.8514, "step": 9564 }, { "epoch": 2.88, "learning_rate": 8.479306789619057e-07, "loss": 1.7539, "step": 9565 }, { "epoch": 2.88, "learning_rate": 8.437128209761947e-07, "loss": 1.8776, "step": 9566 }, { "epoch": 2.88, "learning_rate": 8.395054351575171e-07, "loss": 2.0111, "step": 9567 }, { "epoch": 2.88, "learning_rate": 8.353085219502288e-07, "loss": 1.8727, "step": 9568 }, { "epoch": 2.88, "learning_rate": 8.311220817975973e-07, "loss": 1.8544, "step": 9569 }, { "epoch": 2.88, "learning_rate": 8.269461151417468e-07, "loss": 1.9143, "step": 9570 }, { "epoch": 2.88, "learning_rate": 8.227806224237133e-07, "loss": 1.8823, "step": 9571 }, { "epoch": 2.88, "learning_rate": 8.186256040834228e-07, "loss": 1.732, "step": 9572 }, { "epoch": 2.88, "learning_rate": 8.14481060559702e-07, "loss": 1.7728, "step": 9573 }, { "epoch": 2.88, "learning_rate": 8.103469922902784e-07, "loss": 1.8606, "step": 9574 }, { "epoch": 2.88, "learning_rate": 8.062233997117363e-07, "loss": 1.8588, "step": 9575 }, { "epoch": 2.88, "learning_rate": 8.021102832596051e-07, "loss": 1.8676, "step": 9576 }, { "epoch": 2.88, "learning_rate": 7.980076433682703e-07, "loss": 1.8089, "step": 9577 }, { "epoch": 2.88, "learning_rate": 7.939154804710302e-07, "loss": 1.7967, "step": 9578 }, { "epoch": 2.88, "learning_rate": 7.898337950000611e-07, "loss": 1.8427, "step": 9579 }, { "epoch": 2.88, "learning_rate": 7.857625873864404e-07, "loss": 1.8229, "step": 9580 }, { "epoch": 2.88, "learning_rate": 7.817018580601465e-07, "loss": 1.8325, "step": 9581 }, { "epoch": 2.88, "learning_rate": 7.776516074500473e-07, "loss": 1.8355, "step": 9582 }, { "epoch": 2.88, "learning_rate": 7.736118359838896e-07, "loss": 1.8424, "step": 9583 }, { "epoch": 2.89, "learning_rate": 7.69582544088332e-07, "loss": 1.862, "step": 9584 }, { "epoch": 2.89, "learning_rate": 7.655637321889119e-07, "loss": 1.9167, "step": 9585 }, { "epoch": 2.89, "learning_rate": 7.615554007100789e-07, "loss": 1.9324, "step": 9586 }, { "epoch": 2.89, "learning_rate": 7.575575500751497e-07, "loss": 1.8861, "step": 9587 }, { "epoch": 2.89, "learning_rate": 7.535701807063644e-07, "loss": 1.8433, "step": 9588 }, { "epoch": 2.89, "learning_rate": 7.495932930248196e-07, "loss": 1.8219, "step": 9589 }, { "epoch": 2.89, "learning_rate": 7.456268874505456e-07, "loss": 1.7734, "step": 9590 }, { "epoch": 2.89, "learning_rate": 7.4167096440243e-07, "loss": 1.7357, "step": 9591 }, { "epoch": 2.89, "learning_rate": 7.377255242982717e-07, "loss": 1.8443, "step": 9592 }, { "epoch": 2.89, "learning_rate": 7.337905675547819e-07, "loss": 1.8243, "step": 9593 }, { "epoch": 2.89, "learning_rate": 7.298660945875058e-07, "loss": 1.8369, "step": 9594 }, { "epoch": 2.89, "learning_rate": 7.259521058109342e-07, "loss": 1.858, "step": 9595 }, { "epoch": 2.89, "learning_rate": 7.220486016384475e-07, "loss": 1.7971, "step": 9596 }, { "epoch": 2.89, "learning_rate": 7.181555824822828e-07, "loss": 1.874, "step": 9597 }, { "epoch": 2.89, "learning_rate": 7.142730487536109e-07, "loss": 1.8143, "step": 9598 }, { "epoch": 2.89, "learning_rate": 7.104010008624596e-07, "loss": 1.74, "step": 9599 }, { "epoch": 2.89, "learning_rate": 7.065394392177793e-07, "loss": 1.7672, "step": 9600 }, { "epoch": 2.89, "learning_rate": 7.026883642273995e-07, "loss": 1.8258, "step": 9601 }, { "epoch": 2.89, "learning_rate": 6.988477762980394e-07, "loss": 1.8132, "step": 9602 }, { "epoch": 2.89, "learning_rate": 6.950176758353078e-07, "loss": 1.8131, "step": 9603 }, { "epoch": 2.89, "learning_rate": 6.911980632437254e-07, "loss": 1.8554, "step": 9604 }, { "epoch": 2.89, "learning_rate": 6.873889389266808e-07, "loss": 1.785, "step": 9605 }, { "epoch": 2.89, "learning_rate": 6.835903032864744e-07, "loss": 1.8171, "step": 9606 }, { "epoch": 2.89, "learning_rate": 6.798021567242852e-07, "loss": 1.7819, "step": 9607 }, { "epoch": 2.89, "learning_rate": 6.760244996401932e-07, "loss": 1.7952, "step": 9608 }, { "epoch": 2.89, "learning_rate": 6.722573324331571e-07, "loss": 1.8198, "step": 9609 }, { "epoch": 2.89, "learning_rate": 6.685006555010476e-07, "loss": 1.8714, "step": 9610 }, { "epoch": 2.89, "learning_rate": 6.647544692406249e-07, "loss": 1.6713, "step": 9611 }, { "epoch": 2.89, "learning_rate": 6.610187740475171e-07, "loss": 1.8831, "step": 9612 }, { "epoch": 2.89, "learning_rate": 6.572935703162752e-07, "loss": 1.7943, "step": 9613 }, { "epoch": 2.89, "learning_rate": 6.535788584403068e-07, "loss": 1.7216, "step": 9614 }, { "epoch": 2.89, "learning_rate": 6.498746388119647e-07, "loss": 1.896, "step": 9615 }, { "epoch": 2.89, "learning_rate": 6.46180911822436e-07, "loss": 1.7767, "step": 9616 }, { "epoch": 2.89, "learning_rate": 6.424976778618308e-07, "loss": 1.7861, "step": 9617 }, { "epoch": 2.9, "learning_rate": 6.388249373191491e-07, "loss": 1.7988, "step": 9618 }, { "epoch": 2.9, "learning_rate": 6.351626905822916e-07, "loss": 1.8808, "step": 9619 }, { "epoch": 2.9, "learning_rate": 6.315109380380158e-07, "loss": 1.8469, "step": 9620 }, { "epoch": 2.9, "learning_rate": 6.278696800720019e-07, "loss": 1.9117, "step": 9621 }, { "epoch": 2.9, "learning_rate": 6.242389170688201e-07, "loss": 1.8354, "step": 9622 }, { "epoch": 2.9, "learning_rate": 6.206186494119081e-07, "loss": 1.9962, "step": 9623 }, { "epoch": 2.9, "learning_rate": 6.17008877483638e-07, "loss": 1.8136, "step": 9624 }, { "epoch": 2.9, "learning_rate": 6.134096016652158e-07, "loss": 1.8997, "step": 9625 }, { "epoch": 2.9, "learning_rate": 6.09820822336804e-07, "loss": 1.9503, "step": 9626 }, { "epoch": 2.9, "learning_rate": 6.062425398773997e-07, "loss": 1.763, "step": 9627 }, { "epoch": 2.9, "learning_rate": 6.026747546649225e-07, "loss": 1.7668, "step": 9628 }, { "epoch": 2.9, "learning_rate": 5.991174670761823e-07, "loss": 1.8674, "step": 9629 }, { "epoch": 2.9, "learning_rate": 5.955706774868674e-07, "loss": 1.8136, "step": 9630 }, { "epoch": 2.9, "learning_rate": 5.920343862715671e-07, "loss": 1.899, "step": 9631 }, { "epoch": 2.9, "learning_rate": 5.885085938037382e-07, "loss": 1.8338, "step": 9632 }, { "epoch": 2.9, "learning_rate": 5.849933004557828e-07, "loss": 1.8701, "step": 9633 }, { "epoch": 2.9, "learning_rate": 5.814885065989484e-07, "loss": 1.9229, "step": 9634 }, { "epoch": 2.9, "learning_rate": 5.779942126033721e-07, "loss": 1.8257, "step": 9635 }, { "epoch": 2.9, "learning_rate": 5.745104188381145e-07, "loss": 1.791, "step": 9636 }, { "epoch": 2.9, "learning_rate": 5.710371256710922e-07, "loss": 1.7721, "step": 9637 }, { "epoch": 2.9, "learning_rate": 5.675743334691452e-07, "loss": 1.8102, "step": 9638 }, { "epoch": 2.9, "learning_rate": 5.64122042597981e-07, "loss": 1.7918, "step": 9639 }, { "epoch": 2.9, "learning_rate": 5.606802534222077e-07, "loss": 1.8821, "step": 9640 }, { "epoch": 2.9, "learning_rate": 5.572489663053126e-07, "loss": 1.8506, "step": 9641 }, { "epoch": 2.9, "learning_rate": 5.538281816096947e-07, "loss": 1.7894, "step": 9642 }, { "epoch": 2.9, "learning_rate": 5.504178996966314e-07, "loss": 1.8262, "step": 9643 }, { "epoch": 2.9, "learning_rate": 5.470181209262792e-07, "loss": 1.8822, "step": 9644 }, { "epoch": 2.9, "learning_rate": 5.436288456577177e-07, "loss": 1.7531, "step": 9645 }, { "epoch": 2.9, "learning_rate": 5.402500742488825e-07, "loss": 1.7897, "step": 9646 }, { "epoch": 2.9, "learning_rate": 5.368818070566217e-07, "loss": 1.7362, "step": 9647 }, { "epoch": 2.9, "learning_rate": 5.335240444366618e-07, "loss": 2.0262, "step": 9648 }, { "epoch": 2.9, "learning_rate": 5.301767867436302e-07, "loss": 1.7756, "step": 9649 }, { "epoch": 2.9, "learning_rate": 5.268400343310442e-07, "loss": 1.7715, "step": 9650 }, { "epoch": 2.91, "learning_rate": 5.235137875512996e-07, "loss": 1.7607, "step": 9651 }, { "epoch": 2.91, "learning_rate": 5.201980467556822e-07, "loss": 1.8157, "step": 9652 }, { "epoch": 2.91, "learning_rate": 5.168928122944005e-07, "loss": 1.8524, "step": 9653 }, { "epoch": 2.91, "learning_rate": 5.135980845165089e-07, "loss": 1.7744, "step": 9654 }, { "epoch": 2.91, "learning_rate": 5.103138637699845e-07, "loss": 1.8143, "step": 9655 }, { "epoch": 2.91, "learning_rate": 5.07040150401672e-07, "loss": 1.7643, "step": 9656 }, { "epoch": 2.91, "learning_rate": 5.037769447573282e-07, "loss": 1.866, "step": 9657 }, { "epoch": 2.91, "learning_rate": 5.005242471815885e-07, "loss": 1.8103, "step": 9658 }, { "epoch": 2.91, "learning_rate": 4.972820580179672e-07, "loss": 1.7718, "step": 9659 }, { "epoch": 2.91, "learning_rate": 4.940503776088901e-07, "loss": 1.8227, "step": 9660 }, { "epoch": 2.91, "learning_rate": 4.90829206295651e-07, "loss": 1.9174, "step": 9661 }, { "epoch": 2.91, "learning_rate": 4.876185444184777e-07, "loss": 1.9025, "step": 9662 }, { "epoch": 2.91, "learning_rate": 4.844183923164214e-07, "loss": 1.6625, "step": 9663 }, { "epoch": 2.91, "learning_rate": 4.812287503274781e-07, "loss": 1.8576, "step": 9664 }, { "epoch": 2.91, "learning_rate": 4.78049618788512e-07, "loss": 1.7784, "step": 9665 }, { "epoch": 2.91, "learning_rate": 4.748809980352875e-07, "loss": 1.804, "step": 9666 }, { "epoch": 2.91, "learning_rate": 4.7172288840243714e-07, "loss": 1.9709, "step": 9667 }, { "epoch": 2.91, "learning_rate": 4.6857529022349413e-07, "loss": 1.8652, "step": 9668 }, { "epoch": 2.91, "learning_rate": 4.6543820383090354e-07, "loss": 1.8598, "step": 9669 }, { "epoch": 2.91, "learning_rate": 4.6231162955596706e-07, "loss": 1.8165, "step": 9670 }, { "epoch": 2.91, "learning_rate": 4.5919556772889837e-07, "loss": 1.772, "step": 9671 }, { "epoch": 2.91, "learning_rate": 4.5609001867880084e-07, "loss": 1.8682, "step": 9672 }, { "epoch": 2.91, "learning_rate": 4.529949827336344e-07, "loss": 1.7434, "step": 9673 }, { "epoch": 2.91, "learning_rate": 4.499104602203041e-07, "loss": 1.7757, "step": 9674 }, { "epoch": 2.91, "learning_rate": 4.4683645146454954e-07, "loss": 1.878, "step": 9675 }, { "epoch": 2.91, "learning_rate": 4.4377295679103315e-07, "loss": 1.9241, "step": 9676 }, { "epoch": 2.91, "learning_rate": 4.4071997652330724e-07, "loss": 1.8009, "step": 9677 }, { "epoch": 2.91, "learning_rate": 4.3767751098380273e-07, "loss": 1.8705, "step": 9678 }, { "epoch": 2.91, "learning_rate": 4.346455604938404e-07, "loss": 1.8323, "step": 9679 }, { "epoch": 2.91, "learning_rate": 4.3162412537361975e-07, "loss": 1.8158, "step": 9680 }, { "epoch": 2.91, "learning_rate": 4.286132059422743e-07, "loss": 1.857, "step": 9681 }, { "epoch": 2.91, "learning_rate": 4.2561280251777193e-07, "loss": 1.9298, "step": 9682 }, { "epoch": 2.91, "learning_rate": 4.226229154169925e-07, "loss": 1.9059, "step": 9683 }, { "epoch": 2.92, "learning_rate": 4.196435449557057e-07, "loss": 1.8684, "step": 9684 }, { "epoch": 2.92, "learning_rate": 4.1667469144859305e-07, "loss": 1.7588, "step": 9685 }, { "epoch": 2.92, "learning_rate": 4.137163552091816e-07, "loss": 1.8454, "step": 9686 }, { "epoch": 2.92, "learning_rate": 4.107685365499103e-07, "loss": 1.9381, "step": 9687 }, { "epoch": 2.92, "learning_rate": 4.07831235782119e-07, "loss": 1.8221, "step": 9688 }, { "epoch": 2.92, "learning_rate": 4.049044532160151e-07, "loss": 1.8225, "step": 9689 }, { "epoch": 2.92, "learning_rate": 4.019881891606958e-07, "loss": 1.9497, "step": 9690 }, { "epoch": 2.92, "learning_rate": 3.9908244392418136e-07, "loss": 1.9186, "step": 9691 }, { "epoch": 2.92, "learning_rate": 3.9618721781332635e-07, "loss": 1.8572, "step": 9692 }, { "epoch": 2.92, "learning_rate": 3.9330251113391947e-07, "loss": 1.79, "step": 9693 }, { "epoch": 2.92, "learning_rate": 3.9042832419062814e-07, "loss": 1.8332, "step": 9694 }, { "epoch": 2.92, "learning_rate": 3.875646572869873e-07, "loss": 1.8374, "step": 9695 }, { "epoch": 2.92, "learning_rate": 3.84711510725444e-07, "loss": 1.8273, "step": 9696 }, { "epoch": 2.92, "learning_rate": 3.8186888480732377e-07, "loss": 1.8457, "step": 9697 }, { "epoch": 2.92, "learning_rate": 3.7903677983285314e-07, "loss": 1.7529, "step": 9698 }, { "epoch": 2.92, "learning_rate": 3.762151961011262e-07, "loss": 1.8741, "step": 9699 }, { "epoch": 2.92, "learning_rate": 3.7340413391014903e-07, "loss": 1.9027, "step": 9700 }, { "epoch": 2.92, "learning_rate": 3.706035935568064e-07, "loss": 1.8365, "step": 9701 }, { "epoch": 2.92, "learning_rate": 3.678135753368617e-07, "loss": 1.8388, "step": 9702 }, { "epoch": 2.92, "learning_rate": 3.6503407954497915e-07, "loss": 1.7542, "step": 9703 }, { "epoch": 2.92, "learning_rate": 3.622651064747129e-07, "loss": 1.723, "step": 9704 }, { "epoch": 2.92, "learning_rate": 3.595066564184957e-07, "loss": 1.8422, "step": 9705 }, { "epoch": 2.92, "learning_rate": 3.567587296676611e-07, "loss": 1.8162, "step": 9706 }, { "epoch": 2.92, "learning_rate": 3.5402132651242146e-07, "loss": 1.758, "step": 9707 }, { "epoch": 2.92, "learning_rate": 3.512944472418789e-07, "loss": 1.8578, "step": 9708 }, { "epoch": 2.92, "learning_rate": 3.4857809214403623e-07, "loss": 1.8364, "step": 9709 }, { "epoch": 2.92, "learning_rate": 3.4587226150577524e-07, "loss": 1.7189, "step": 9710 }, { "epoch": 2.92, "learning_rate": 3.43176955612845e-07, "loss": 1.9107, "step": 9711 }, { "epoch": 2.92, "learning_rate": 3.40492174749929e-07, "loss": 1.8281, "step": 9712 }, { "epoch": 2.92, "learning_rate": 3.378179192005559e-07, "loss": 1.7541, "step": 9713 }, { "epoch": 2.92, "learning_rate": 3.351541892471777e-07, "loss": 1.7308, "step": 9714 }, { "epoch": 2.92, "learning_rate": 3.325009851711136e-07, "loss": 1.8097, "step": 9715 }, { "epoch": 2.92, "learning_rate": 3.2985830725256184e-07, "loss": 1.9095, "step": 9716 }, { "epoch": 2.93, "learning_rate": 3.272261557706324e-07, "loss": 1.8389, "step": 9717 }, { "epoch": 2.93, "learning_rate": 3.246045310033252e-07, "loss": 1.857, "step": 9718 }, { "epoch": 2.93, "learning_rate": 3.2199343322749653e-07, "loss": 1.8971, "step": 9719 }, { "epoch": 2.93, "learning_rate": 3.1939286271893685e-07, "loss": 1.8048, "step": 9720 }, { "epoch": 2.93, "learning_rate": 3.16802819752271e-07, "loss": 1.7286, "step": 9721 }, { "epoch": 2.93, "learning_rate": 3.1422330460105785e-07, "loss": 1.8833, "step": 9722 }, { "epoch": 2.93, "learning_rate": 3.11654317537724e-07, "loss": 1.8411, "step": 9723 }, { "epoch": 2.93, "learning_rate": 3.090958588335857e-07, "loss": 1.8141, "step": 9724 }, { "epoch": 2.93, "learning_rate": 3.0654792875886016e-07, "loss": 1.7379, "step": 9725 }, { "epoch": 2.93, "learning_rate": 3.0401052758262106e-07, "loss": 1.8824, "step": 9726 }, { "epoch": 2.93, "learning_rate": 3.01483655572854e-07, "loss": 1.866, "step": 9727 }, { "epoch": 2.93, "learning_rate": 2.989673129964454e-07, "loss": 1.7706, "step": 9728 }, { "epoch": 2.93, "learning_rate": 2.9646150011913845e-07, "loss": 1.875, "step": 9729 }, { "epoch": 2.93, "learning_rate": 2.9396621720558794e-07, "loss": 1.9152, "step": 9730 }, { "epoch": 2.93, "learning_rate": 2.9148146451931646e-07, "loss": 1.8482, "step": 9731 }, { "epoch": 2.93, "learning_rate": 2.8900724232274747e-07, "loss": 1.8864, "step": 9732 }, { "epoch": 2.93, "learning_rate": 2.865435508772052e-07, "loss": 1.8411, "step": 9733 }, { "epoch": 2.93, "learning_rate": 2.8409039044287047e-07, "loss": 1.7561, "step": 9734 }, { "epoch": 2.93, "learning_rate": 2.8164776127883596e-07, "loss": 1.8133, "step": 9735 }, { "epoch": 2.93, "learning_rate": 2.7921566364307316e-07, "loss": 1.7782, "step": 9736 }, { "epoch": 2.93, "learning_rate": 2.7679409779243216e-07, "loss": 1.7643, "step": 9737 }, { "epoch": 2.93, "learning_rate": 2.743830639826861e-07, "loss": 1.8286, "step": 9738 }, { "epoch": 2.93, "learning_rate": 2.7198256246845354e-07, "loss": 1.759, "step": 9739 }, { "epoch": 2.93, "learning_rate": 2.6959259350325394e-07, "loss": 1.8562, "step": 9740 }, { "epoch": 2.93, "learning_rate": 2.6721315733950757e-07, "loss": 1.7805, "step": 9741 }, { "epoch": 2.93, "learning_rate": 2.6484425422850234e-07, "loss": 1.7779, "step": 9742 }, { "epoch": 2.93, "learning_rate": 2.624858844204381e-07, "loss": 1.7358, "step": 9743 }, { "epoch": 2.93, "learning_rate": 2.6013804816439333e-07, "loss": 1.8894, "step": 9744 }, { "epoch": 2.93, "learning_rate": 2.5780074570830313e-07, "loss": 1.8688, "step": 9745 }, { "epoch": 2.93, "learning_rate": 2.5547397729904776e-07, "loss": 1.831, "step": 9746 }, { "epoch": 2.93, "learning_rate": 2.531577431823418e-07, "loss": 1.8182, "step": 9747 }, { "epoch": 2.93, "learning_rate": 2.508520436028117e-07, "loss": 1.7746, "step": 9748 }, { "epoch": 2.93, "learning_rate": 2.485568788039738e-07, "loss": 1.9185, "step": 9749 }, { "epoch": 2.93, "learning_rate": 2.462722490282343e-07, "loss": 1.8445, "step": 9750 }, { "epoch": 2.94, "learning_rate": 2.439981545168557e-07, "loss": 1.836, "step": 9751 }, { "epoch": 2.94, "learning_rate": 2.417345955100347e-07, "loss": 1.7796, "step": 9752 }, { "epoch": 2.94, "learning_rate": 2.3948157224682467e-07, "loss": 1.7759, "step": 9753 }, { "epoch": 2.94, "learning_rate": 2.3723908496517955e-07, "loss": 1.7884, "step": 9754 }, { "epoch": 2.94, "learning_rate": 2.3500713390190998e-07, "loss": 1.7691, "step": 9755 }, { "epoch": 2.94, "learning_rate": 2.3278571929277182e-07, "loss": 1.8688, "step": 9756 }, { "epoch": 2.94, "learning_rate": 2.305748413723552e-07, "loss": 1.7865, "step": 9757 }, { "epoch": 2.94, "learning_rate": 2.283745003741511e-07, "loss": 1.811, "step": 9758 }, { "epoch": 2.94, "learning_rate": 2.261846965305625e-07, "loss": 1.8474, "step": 9759 }, { "epoch": 2.94, "learning_rate": 2.240054300728489e-07, "loss": 1.7723, "step": 9760 }, { "epoch": 2.94, "learning_rate": 2.2183670123117062e-07, "loss": 1.7916, "step": 9761 }, { "epoch": 2.94, "learning_rate": 2.1967851023456665e-07, "loss": 1.7802, "step": 9762 }, { "epoch": 2.94, "learning_rate": 2.17530857310988e-07, "loss": 1.823, "step": 9763 }, { "epoch": 2.94, "learning_rate": 2.153937426872421e-07, "loss": 1.9179, "step": 9764 }, { "epoch": 2.94, "learning_rate": 2.1326716658902623e-07, "loss": 1.8542, "step": 9765 }, { "epoch": 2.94, "learning_rate": 2.111511292409496e-07, "loss": 1.8519, "step": 9766 }, { "epoch": 2.94, "learning_rate": 2.0904563086647787e-07, "loss": 1.7854, "step": 9767 }, { "epoch": 2.94, "learning_rate": 2.0695067168799987e-07, "loss": 1.8663, "step": 9768 }, { "epoch": 2.94, "learning_rate": 2.048662519267497e-07, "loss": 1.8335, "step": 9769 }, { "epoch": 2.94, "learning_rate": 2.0279237180287347e-07, "loss": 1.8585, "step": 9770 }, { "epoch": 2.94, "learning_rate": 2.0072903153540712e-07, "loss": 1.7677, "step": 9771 }, { "epoch": 2.94, "learning_rate": 1.986762313422541e-07, "loss": 1.7934, "step": 9772 }, { "epoch": 2.94, "learning_rate": 1.9663397144022988e-07, "loss": 1.9387, "step": 9773 }, { "epoch": 2.94, "learning_rate": 1.9460225204501747e-07, "loss": 1.9262, "step": 9774 }, { "epoch": 2.94, "learning_rate": 1.9258107337118969e-07, "loss": 1.7991, "step": 9775 }, { "epoch": 2.94, "learning_rate": 1.9057043563220912e-07, "loss": 1.882, "step": 9776 }, { "epoch": 2.94, "learning_rate": 1.8857033904041698e-07, "loss": 1.8461, "step": 9777 }, { "epoch": 2.94, "learning_rate": 1.8658078380706657e-07, "loss": 1.8764, "step": 9778 }, { "epoch": 2.94, "learning_rate": 1.8460177014226754e-07, "loss": 1.8078, "step": 9779 }, { "epoch": 2.94, "learning_rate": 1.8263329825503052e-07, "loss": 1.9336, "step": 9780 }, { "epoch": 2.94, "learning_rate": 1.8067536835325584e-07, "loss": 1.6693, "step": 9781 }, { "epoch": 2.94, "learning_rate": 1.7872798064372253e-07, "loss": 1.8127, "step": 9782 }, { "epoch": 2.94, "learning_rate": 1.7679113533211055e-07, "loss": 1.7968, "step": 9783 }, { "epoch": 2.95, "learning_rate": 1.7486483262294516e-07, "loss": 1.8333, "step": 9784 }, { "epoch": 2.95, "learning_rate": 1.7294907271970806e-07, "loss": 1.8333, "step": 9785 }, { "epoch": 2.95, "learning_rate": 1.7104385582469297e-07, "loss": 1.8355, "step": 9786 }, { "epoch": 2.95, "learning_rate": 1.6914918213913887e-07, "loss": 1.7654, "step": 9787 }, { "epoch": 2.95, "learning_rate": 1.6726505186314135e-07, "loss": 1.8909, "step": 9788 }, { "epoch": 2.95, "learning_rate": 1.653914651956745e-07, "loss": 1.7877, "step": 9789 }, { "epoch": 2.95, "learning_rate": 1.6352842233463561e-07, "loss": 1.844, "step": 9790 }, { "epoch": 2.95, "learning_rate": 1.616759234767673e-07, "loss": 1.7479, "step": 9791 }, { "epoch": 2.95, "learning_rate": 1.5983396881773526e-07, "loss": 1.8115, "step": 9792 }, { "epoch": 2.95, "learning_rate": 1.580025585520506e-07, "loss": 1.7819, "step": 9793 }, { "epoch": 2.95, "learning_rate": 1.5618169287314743e-07, "loss": 1.8878, "step": 9794 }, { "epoch": 2.95, "learning_rate": 1.5437137197333863e-07, "loss": 1.8101, "step": 9795 }, { "epoch": 2.95, "learning_rate": 1.5257159604379344e-07, "loss": 1.8509, "step": 9796 }, { "epoch": 2.95, "learning_rate": 1.5078236527462654e-07, "loss": 1.8151, "step": 9797 }, { "epoch": 2.95, "learning_rate": 1.4900367985476448e-07, "loss": 1.8625, "step": 9798 }, { "epoch": 2.95, "learning_rate": 1.4723553997209038e-07, "loss": 1.9126, "step": 9799 }, { "epoch": 2.95, "learning_rate": 1.4547794581333262e-07, "loss": 1.8075, "step": 9800 }, { "epoch": 2.95, "learning_rate": 1.437308975640983e-07, "loss": 1.8117, "step": 9801 }, { "epoch": 2.95, "learning_rate": 1.4199439540892866e-07, "loss": 1.8, "step": 9802 }, { "epoch": 2.95, "learning_rate": 1.4026843953118818e-07, "loss": 1.8794, "step": 9803 }, { "epoch": 2.95, "learning_rate": 1.3855303011319765e-07, "loss": 1.8909, "step": 9804 }, { "epoch": 2.95, "learning_rate": 1.3684816733608997e-07, "loss": 1.8031, "step": 9805 }, { "epoch": 2.95, "learning_rate": 1.351538513799322e-07, "loss": 1.7955, "step": 9806 }, { "epoch": 2.95, "learning_rate": 1.334700824236812e-07, "loss": 1.8336, "step": 9807 }, { "epoch": 2.95, "learning_rate": 1.317968606451392e-07, "loss": 1.7785, "step": 9808 }, { "epoch": 2.95, "learning_rate": 1.3013418622104257e-07, "loss": 1.8214, "step": 9809 }, { "epoch": 2.95, "learning_rate": 1.2848205932697312e-07, "loss": 1.7879, "step": 9810 }, { "epoch": 2.95, "learning_rate": 1.2684048013742457e-07, "loss": 1.9356, "step": 9811 }, { "epoch": 2.95, "learning_rate": 1.2520944882576935e-07, "loss": 1.9422, "step": 9812 }, { "epoch": 2.95, "learning_rate": 1.235889655642697e-07, "loss": 1.8659, "step": 9813 }, { "epoch": 2.95, "learning_rate": 1.2197903052405535e-07, "loss": 1.8703, "step": 9814 }, { "epoch": 2.95, "learning_rate": 1.203796438751681e-07, "loss": 1.8524, "step": 9815 }, { "epoch": 2.95, "learning_rate": 1.1879080578651725e-07, "loss": 1.7917, "step": 9816 }, { "epoch": 2.96, "learning_rate": 1.1721251642591302e-07, "loss": 1.8138, "step": 9817 }, { "epoch": 2.96, "learning_rate": 1.156447759600221e-07, "loss": 1.8541, "step": 9818 }, { "epoch": 2.96, "learning_rate": 1.1408758455444534e-07, "loss": 1.8147, "step": 9819 }, { "epoch": 2.96, "learning_rate": 1.12540942373629e-07, "loss": 1.8053, "step": 9820 }, { "epoch": 2.96, "learning_rate": 1.1100484958090907e-07, "loss": 1.9186, "step": 9821 }, { "epoch": 2.96, "learning_rate": 1.0947930633853353e-07, "loss": 1.8907, "step": 9822 }, { "epoch": 2.96, "learning_rate": 1.0796431280760688e-07, "loss": 1.8417, "step": 9823 }, { "epoch": 2.96, "learning_rate": 1.0645986914813445e-07, "loss": 1.8226, "step": 9824 }, { "epoch": 2.96, "learning_rate": 1.0496597551901133e-07, "loss": 1.7578, "step": 9825 }, { "epoch": 2.96, "learning_rate": 1.0348263207800024e-07, "loss": 1.7935, "step": 9826 }, { "epoch": 2.96, "learning_rate": 1.0200983898177586e-07, "loss": 1.7895, "step": 9827 }, { "epoch": 2.96, "learning_rate": 1.005475963858693e-07, "loss": 1.79, "step": 9828 }, { "epoch": 2.96, "learning_rate": 9.909590444472372e-08, "loss": 1.8517, "step": 9829 }, { "epoch": 2.96, "learning_rate": 9.765476331163869e-08, "loss": 1.7582, "step": 9830 }, { "epoch": 2.96, "learning_rate": 9.622417313883691e-08, "loss": 1.7898, "step": 9831 }, { "epoch": 2.96, "learning_rate": 9.480413407740862e-08, "loss": 1.8934, "step": 9832 }, { "epoch": 2.96, "learning_rate": 9.339464627730054e-08, "loss": 1.8124, "step": 9833 }, { "epoch": 2.96, "learning_rate": 9.19957098874047e-08, "loss": 1.9523, "step": 9834 }, { "epoch": 2.96, "learning_rate": 9.060732505544733e-08, "loss": 1.8542, "step": 9835 }, { "epoch": 2.96, "learning_rate": 8.922949192806673e-08, "loss": 1.7842, "step": 9836 }, { "epoch": 2.96, "learning_rate": 8.786221065076872e-08, "loss": 1.789, "step": 9837 }, { "epoch": 2.96, "learning_rate": 8.65054813679822e-08, "loss": 1.8023, "step": 9838 }, { "epoch": 2.96, "learning_rate": 8.515930422295926e-08, "loss": 1.7868, "step": 9839 }, { "epoch": 2.96, "learning_rate": 8.382367935789725e-08, "loss": 1.8815, "step": 9840 }, { "epoch": 2.96, "learning_rate": 8.249860691385003e-08, "loss": 1.7944, "step": 9841 }, { "epoch": 2.96, "learning_rate": 8.118408703076118e-08, "loss": 1.848, "step": 9842 }, { "epoch": 2.96, "learning_rate": 7.988011984746413e-08, "loss": 1.8597, "step": 9843 }, { "epoch": 2.96, "learning_rate": 7.858670550167091e-08, "loss": 1.8426, "step": 9844 }, { "epoch": 2.96, "learning_rate": 7.730384412998337e-08, "loss": 1.7919, "step": 9845 }, { "epoch": 2.96, "learning_rate": 7.603153586788204e-08, "loss": 1.7358, "step": 9846 }, { "epoch": 2.96, "learning_rate": 7.476978084974828e-08, "loss": 1.8045, "step": 9847 }, { "epoch": 2.96, "learning_rate": 7.35185792088422e-08, "loss": 1.7115, "step": 9848 }, { "epoch": 2.96, "learning_rate": 7.227793107729142e-08, "loss": 1.8753, "step": 9849 }, { "epoch": 2.97, "learning_rate": 7.104783658613557e-08, "loss": 1.9267, "step": 9850 }, { "epoch": 2.97, "learning_rate": 6.982829586528183e-08, "loss": 1.7962, "step": 9851 }, { "epoch": 2.97, "learning_rate": 6.861930904353831e-08, "loss": 1.8873, "step": 9852 }, { "epoch": 2.97, "learning_rate": 6.742087624858062e-08, "loss": 1.8516, "step": 9853 }, { "epoch": 2.97, "learning_rate": 6.623299760698532e-08, "loss": 1.8511, "step": 9854 }, { "epoch": 2.97, "learning_rate": 6.505567324420758e-08, "loss": 1.8016, "step": 9855 }, { "epoch": 2.97, "learning_rate": 6.38889032845702e-08, "loss": 1.8129, "step": 9856 }, { "epoch": 2.97, "learning_rate": 6.273268785133013e-08, "loss": 1.8173, "step": 9857 }, { "epoch": 2.97, "learning_rate": 6.15870270665675e-08, "loss": 1.8254, "step": 9858 }, { "epoch": 2.97, "learning_rate": 6.04519210513077e-08, "loss": 1.856, "step": 9859 }, { "epoch": 2.97, "learning_rate": 5.932736992539933e-08, "loss": 1.9512, "step": 9860 }, { "epoch": 2.97, "learning_rate": 5.8213373807647354e-08, "loss": 1.7756, "step": 9861 }, { "epoch": 2.97, "learning_rate": 5.7109932815668784e-08, "loss": 1.8627, "step": 9862 }, { "epoch": 2.97, "learning_rate": 5.601704706602595e-08, "loss": 1.8453, "step": 9863 }, { "epoch": 2.97, "learning_rate": 5.493471667412653e-08, "loss": 1.849, "step": 9864 }, { "epoch": 2.97, "learning_rate": 5.386294175427908e-08, "loss": 1.7969, "step": 9865 }, { "epoch": 2.97, "learning_rate": 5.2801722419693056e-08, "loss": 1.8435, "step": 9866 }, { "epoch": 2.97, "learning_rate": 5.1751058782423256e-08, "loss": 1.8087, "step": 9867 }, { "epoch": 2.97, "learning_rate": 5.0710950953447576e-08, "loss": 1.8851, "step": 9868 }, { "epoch": 2.97, "learning_rate": 4.968139904262259e-08, "loss": 1.8691, "step": 9869 }, { "epoch": 2.97, "learning_rate": 4.866240315866133e-08, "loss": 1.8379, "step": 9870 }, { "epoch": 2.97, "learning_rate": 4.7653963409188815e-08, "loss": 1.8375, "step": 9871 }, { "epoch": 2.97, "learning_rate": 4.665607990073095e-08, "loss": 1.8262, "step": 9872 }, { "epoch": 2.97, "learning_rate": 4.56687527386368e-08, "loss": 1.8669, "step": 9873 }, { "epoch": 2.97, "learning_rate": 4.469198202722291e-08, "loss": 1.8874, "step": 9874 }, { "epoch": 2.97, "learning_rate": 4.372576786961791e-08, "loss": 1.8999, "step": 9875 }, { "epoch": 2.97, "learning_rate": 4.277011036788459e-08, "loss": 1.834, "step": 9876 }, { "epoch": 2.97, "learning_rate": 4.182500962293112e-08, "loss": 1.7446, "step": 9877 }, { "epoch": 2.97, "learning_rate": 4.089046573459987e-08, "loss": 1.8222, "step": 9878 }, { "epoch": 2.97, "learning_rate": 3.996647880156745e-08, "loss": 1.8352, "step": 9879 }, { "epoch": 2.97, "learning_rate": 3.9053048921433576e-08, "loss": 1.8574, "step": 9880 }, { "epoch": 2.97, "learning_rate": 3.815017619066552e-08, "loss": 1.8067, "step": 9881 }, { "epoch": 2.97, "learning_rate": 3.725786070460924e-08, "loss": 1.9079, "step": 9882 }, { "epoch": 2.98, "learning_rate": 3.6376102557522665e-08, "loss": 1.8863, "step": 9883 }, { "epoch": 2.98, "learning_rate": 3.55049018425091e-08, "loss": 1.852, "step": 9884 }, { "epoch": 2.98, "learning_rate": 3.4644258651583826e-08, "loss": 1.7802, "step": 9885 }, { "epoch": 2.98, "learning_rate": 3.379417307565191e-08, "loss": 1.8649, "step": 9886 }, { "epoch": 2.98, "learning_rate": 3.295464520448599e-08, "loss": 1.8602, "step": 9887 }, { "epoch": 2.98, "learning_rate": 3.212567512675957e-08, "loss": 1.8722, "step": 9888 }, { "epoch": 2.98, "learning_rate": 3.130726293000263e-08, "loss": 1.8576, "step": 9889 }, { "epoch": 2.98, "learning_rate": 3.049940870066825e-08, "loss": 1.8101, "step": 9890 }, { "epoch": 2.98, "learning_rate": 2.9702112524077065e-08, "loss": 1.9605, "step": 9891 }, { "epoch": 2.98, "learning_rate": 2.891537448441728e-08, "loss": 1.8016, "step": 9892 }, { "epoch": 2.98, "learning_rate": 2.8139194664789094e-08, "loss": 1.818, "step": 9893 }, { "epoch": 2.98, "learning_rate": 2.7373573147171373e-08, "loss": 1.8224, "step": 9894 }, { "epoch": 2.98, "learning_rate": 2.661851001241056e-08, "loss": 1.8106, "step": 9895 }, { "epoch": 2.98, "learning_rate": 2.5874005340265072e-08, "loss": 1.8788, "step": 9896 }, { "epoch": 2.98, "learning_rate": 2.514005920934981e-08, "loss": 1.8404, "step": 9897 }, { "epoch": 2.98, "learning_rate": 2.4416671697191638e-08, "loss": 1.8658, "step": 9898 }, { "epoch": 2.98, "learning_rate": 2.3703842880185014e-08, "loss": 1.8993, "step": 9899 }, { "epoch": 2.98, "learning_rate": 2.3001572833614148e-08, "loss": 1.741, "step": 9900 }, { "epoch": 2.98, "learning_rate": 2.230986163164195e-08, "loss": 2.073, "step": 9901 }, { "epoch": 2.98, "learning_rate": 2.1628709347332187e-08, "loss": 1.8612, "step": 9902 }, { "epoch": 2.98, "learning_rate": 2.0958116052616216e-08, "loss": 1.7985, "step": 9903 }, { "epoch": 2.98, "learning_rate": 2.0298081818315162e-08, "loss": 1.8497, "step": 9904 }, { "epoch": 2.98, "learning_rate": 1.964860671415103e-08, "loss": 1.923, "step": 9905 }, { "epoch": 2.98, "learning_rate": 1.9009690808691194e-08, "loss": 1.8307, "step": 9906 }, { "epoch": 2.98, "learning_rate": 1.8381334169437213e-08, "loss": 1.8814, "step": 9907 }, { "epoch": 2.98, "learning_rate": 1.776353686274712e-08, "loss": 1.8308, "step": 9908 }, { "epoch": 2.98, "learning_rate": 1.715629895385762e-08, "loss": 1.7923, "step": 9909 }, { "epoch": 2.98, "learning_rate": 1.6559620506906294e-08, "loss": 1.8541, "step": 9910 }, { "epoch": 2.98, "learning_rate": 1.59735015849094e-08, "loss": 1.7232, "step": 9911 }, { "epoch": 2.98, "learning_rate": 1.5397942249772978e-08, "loss": 1.8734, "step": 9912 }, { "epoch": 2.98, "learning_rate": 1.4832942562281738e-08, "loss": 1.7522, "step": 9913 }, { "epoch": 2.98, "learning_rate": 1.4278502582099062e-08, "loss": 1.8389, "step": 9914 }, { "epoch": 2.98, "learning_rate": 1.3734622367789219e-08, "loss": 1.8508, "step": 9915 }, { "epoch": 2.98, "learning_rate": 1.3201301976795144e-08, "loss": 1.9639, "step": 9916 }, { "epoch": 2.99, "learning_rate": 1.2678541465427351e-08, "loss": 1.8816, "step": 9917 }, { "epoch": 2.99, "learning_rate": 1.2166340888919437e-08, "loss": 1.8118, "step": 9918 }, { "epoch": 2.99, "learning_rate": 1.1664700301350362e-08, "loss": 1.7374, "step": 9919 }, { "epoch": 2.99, "learning_rate": 1.1173619755699971e-08, "loss": 1.8093, "step": 9920 }, { "epoch": 2.99, "learning_rate": 1.0693099303837884e-08, "loss": 1.7745, "step": 9921 }, { "epoch": 2.99, "learning_rate": 1.0223138996501292e-08, "loss": 1.8032, "step": 9922 }, { "epoch": 2.99, "learning_rate": 9.763738883339368e-09, "loss": 1.8912, "step": 9923 }, { "epoch": 2.99, "learning_rate": 9.314899012868861e-09, "loss": 1.8372, "step": 9924 }, { "epoch": 2.99, "learning_rate": 8.87661943248519e-09, "loss": 1.7682, "step": 9925 }, { "epoch": 2.99, "learning_rate": 8.448900188484655e-09, "loss": 1.8057, "step": 9926 }, { "epoch": 2.99, "learning_rate": 8.031741326031128e-09, "loss": 1.7026, "step": 9927 }, { "epoch": 2.99, "learning_rate": 7.625142889189363e-09, "loss": 1.8368, "step": 9928 }, { "epoch": 2.99, "learning_rate": 7.229104920891683e-09, "loss": 1.7834, "step": 9929 }, { "epoch": 2.99, "learning_rate": 6.843627462971292e-09, "loss": 1.8805, "step": 9930 }, { "epoch": 2.99, "learning_rate": 6.468710556151169e-09, "loss": 1.8519, "step": 9931 }, { "epoch": 2.99, "learning_rate": 6.104354239999666e-09, "loss": 1.8326, "step": 9932 }, { "epoch": 2.99, "learning_rate": 5.750558553019314e-09, "loss": 1.8204, "step": 9933 }, { "epoch": 2.99, "learning_rate": 5.40732353256912e-09, "loss": 1.7964, "step": 9934 }, { "epoch": 2.99, "learning_rate": 5.074649214897864e-09, "loss": 1.8302, "step": 9935 }, { "epoch": 2.99, "learning_rate": 4.752535635144106e-09, "loss": 1.7815, "step": 9936 }, { "epoch": 2.99, "learning_rate": 4.4409828273250796e-09, "loss": 1.9257, "step": 9937 }, { "epoch": 2.99, "learning_rate": 4.139990824347795e-09, "loss": 1.8458, "step": 9938 }, { "epoch": 2.99, "learning_rate": 3.849559657986834e-09, "loss": 1.8255, "step": 9939 }, { "epoch": 2.99, "learning_rate": 3.569689358939865e-09, "loss": 1.8548, "step": 9940 }, { "epoch": 2.99, "learning_rate": 3.3003799567388193e-09, "loss": 1.9171, "step": 9941 }, { "epoch": 2.99, "learning_rate": 3.0416314798387134e-09, "loss": 1.7863, "step": 9942 }, { "epoch": 2.99, "learning_rate": 2.7934439555732384e-09, "loss": 1.7546, "step": 9943 }, { "epoch": 2.99, "learning_rate": 2.555817410143657e-09, "loss": 1.9304, "step": 9944 }, { "epoch": 2.99, "learning_rate": 2.32875186864101e-09, "loss": 1.8951, "step": 9945 }, { "epoch": 2.99, "learning_rate": 2.1122473550683197e-09, "loss": 1.8387, "step": 9946 }, { "epoch": 2.99, "learning_rate": 1.906303892273975e-09, "loss": 1.8489, "step": 9947 }, { "epoch": 2.99, "learning_rate": 1.7109215020072456e-09, "loss": 1.8839, "step": 9948 }, { "epoch": 2.99, "learning_rate": 1.526100204907177e-09, "loss": 1.8956, "step": 9949 }, { "epoch": 3.0, "learning_rate": 1.3518400205025927e-09, "loss": 1.7959, "step": 9950 }, { "epoch": 3.0, "learning_rate": 1.1881409671787857e-09, "loss": 1.7912, "step": 9951 }, { "epoch": 3.0, "learning_rate": 1.0350030622441332e-09, "loss": 1.8484, "step": 9952 }, { "epoch": 3.0, "learning_rate": 8.924263218523798e-10, "loss": 1.8535, "step": 9953 }, { "epoch": 3.0, "learning_rate": 7.604107610803546e-10, "loss": 1.9722, "step": 9954 }, { "epoch": 3.0, "learning_rate": 6.389563938613563e-10, "loss": 1.7528, "step": 9955 }, { "epoch": 3.0, "learning_rate": 5.280632330184609e-10, "loss": 1.773, "step": 9956 }, { "epoch": 3.0, "learning_rate": 4.2773129027562365e-10, "loss": 1.8404, "step": 9957 }, { "epoch": 3.0, "learning_rate": 3.379605762132698e-10, "loss": 1.9314, "step": 9958 }, { "epoch": 3.0, "learning_rate": 2.5875110032380636e-10, "loss": 1.9387, "step": 9959 }, { "epoch": 3.0, "learning_rate": 1.9010287096721258e-10, "loss": 1.7518, "step": 9960 }, { "epoch": 3.0, "learning_rate": 1.3201589539324488e-10, "loss": 1.8125, "step": 9961 }, { "epoch": 3.0, "learning_rate": 8.449017974143659e-11, "loss": 1.7958, "step": 9962 }, { "epoch": 3.0, "learning_rate": 4.7525729029995746e-11, "loss": 1.8502, "step": 9963 }, { "epoch": 3.0, "learning_rate": 2.1122547166907425e-11, "loss": 1.858, "step": 9964 }, { "epoch": 3.0, "learning_rate": 5.280636927729177e-12, "loss": 1.8124, "step": 9965 }, { "epoch": 3.0, "learning_rate": 0.0, "loss": 1.8167, "step": 9966 }, { "epoch": 3.0, "step": 9966, "total_flos": 4.0741891871890637e+18, "train_loss": 1.8420501036646852, "train_runtime": 95143.9297, "train_samples_per_second": 6.703, "train_steps_per_second": 0.105 } ], "logging_steps": 1.0, "max_steps": 9966, "num_input_tokens_seen": 0, "num_train_epochs": 3, "save_steps": 921, "total_flos": 4.0741891871890637e+18, "train_batch_size": 64, "trial_name": null, "trial_params": null }