{ "best_metric": null, "best_model_checkpoint": null, "epoch": 14.992389649923897, "eval_steps": 500, "global_step": 14775, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 4.504504504504505e-08, "loss": 1.8493, "step": 1 }, { "epoch": 0.0, "learning_rate": 9.00900900900901e-08, "loss": 1.7602, "step": 2 }, { "epoch": 0.0, "learning_rate": 1.3513513513513515e-07, "loss": 1.8546, "step": 3 }, { "epoch": 0.0, "learning_rate": 1.801801801801802e-07, "loss": 1.722, "step": 4 }, { "epoch": 0.01, "learning_rate": 2.2522522522522524e-07, "loss": 1.7697, "step": 5 }, { "epoch": 0.01, "learning_rate": 2.702702702702703e-07, "loss": 1.7926, "step": 6 }, { "epoch": 0.01, "learning_rate": 3.153153153153153e-07, "loss": 1.7981, "step": 7 }, { "epoch": 0.01, "learning_rate": 3.603603603603604e-07, "loss": 1.7953, "step": 8 }, { "epoch": 0.01, "learning_rate": 4.0540540540540546e-07, "loss": 1.6978, "step": 9 }, { "epoch": 0.01, "learning_rate": 4.504504504504505e-07, "loss": 1.7729, "step": 10 }, { "epoch": 0.01, "learning_rate": 4.954954954954956e-07, "loss": 1.7418, "step": 11 }, { "epoch": 0.01, "learning_rate": 5.405405405405406e-07, "loss": 1.6804, "step": 12 }, { "epoch": 0.01, "learning_rate": 5.855855855855856e-07, "loss": 1.7507, "step": 13 }, { "epoch": 0.01, "learning_rate": 6.306306306306306e-07, "loss": 1.7399, "step": 14 }, { "epoch": 0.02, "learning_rate": 6.756756756756758e-07, "loss": 1.7591, "step": 15 }, { "epoch": 0.02, "learning_rate": 7.207207207207208e-07, "loss": 1.769, "step": 16 }, { "epoch": 0.02, "learning_rate": 7.657657657657658e-07, "loss": 1.6872, "step": 17 }, { "epoch": 0.02, "learning_rate": 8.108108108108109e-07, "loss": 1.7154, "step": 18 }, { "epoch": 0.02, "learning_rate": 8.55855855855856e-07, "loss": 1.6699, "step": 19 }, { "epoch": 0.02, "learning_rate": 9.00900900900901e-07, "loss": 1.639, "step": 20 }, { "epoch": 0.02, "learning_rate": 9.459459459459461e-07, "loss": 1.5066, "step": 21 }, { "epoch": 0.02, "learning_rate": 9.909909909909911e-07, "loss": 1.6117, "step": 22 }, { "epoch": 0.02, "learning_rate": 1.0360360360360361e-06, "loss": 1.4837, "step": 23 }, { "epoch": 0.02, "learning_rate": 1.0810810810810812e-06, "loss": 1.5661, "step": 24 }, { "epoch": 0.03, "learning_rate": 1.1261261261261262e-06, "loss": 1.4398, "step": 25 }, { "epoch": 0.03, "learning_rate": 1.1711711711711712e-06, "loss": 1.4637, "step": 26 }, { "epoch": 0.03, "learning_rate": 1.2162162162162164e-06, "loss": 1.411, "step": 27 }, { "epoch": 0.03, "learning_rate": 1.2612612612612613e-06, "loss": 1.4116, "step": 28 }, { "epoch": 0.03, "learning_rate": 1.3063063063063065e-06, "loss": 1.445, "step": 29 }, { "epoch": 0.03, "learning_rate": 1.3513513513513515e-06, "loss": 1.2336, "step": 30 }, { "epoch": 0.03, "learning_rate": 1.3963963963963963e-06, "loss": 1.2575, "step": 31 }, { "epoch": 0.03, "learning_rate": 1.4414414414414416e-06, "loss": 1.1755, "step": 32 }, { "epoch": 0.03, "learning_rate": 1.4864864864864868e-06, "loss": 1.2352, "step": 33 }, { "epoch": 0.03, "learning_rate": 1.5315315315315316e-06, "loss": 1.2128, "step": 34 }, { "epoch": 0.04, "learning_rate": 1.5765765765765766e-06, "loss": 1.2145, "step": 35 }, { "epoch": 0.04, "learning_rate": 1.6216216216216219e-06, "loss": 1.1741, "step": 36 }, { "epoch": 0.04, "learning_rate": 1.6666666666666667e-06, "loss": 1.1209, "step": 37 }, { "epoch": 0.04, "learning_rate": 1.711711711711712e-06, "loss": 1.1483, "step": 38 }, { "epoch": 0.04, "learning_rate": 1.756756756756757e-06, "loss": 0.9746, "step": 39 }, { "epoch": 0.04, "learning_rate": 1.801801801801802e-06, "loss": 1.0274, "step": 40 }, { "epoch": 0.04, "learning_rate": 1.846846846846847e-06, "loss": 0.9828, "step": 41 }, { "epoch": 0.04, "learning_rate": 1.8918918918918922e-06, "loss": 1.0295, "step": 42 }, { "epoch": 0.04, "learning_rate": 1.9369369369369372e-06, "loss": 0.9123, "step": 43 }, { "epoch": 0.04, "learning_rate": 1.9819819819819822e-06, "loss": 1.0684, "step": 44 }, { "epoch": 0.05, "learning_rate": 2.0270270270270273e-06, "loss": 0.9467, "step": 45 }, { "epoch": 0.05, "learning_rate": 2.0720720720720723e-06, "loss": 0.9427, "step": 46 }, { "epoch": 0.05, "learning_rate": 2.1171171171171173e-06, "loss": 0.9261, "step": 47 }, { "epoch": 0.05, "learning_rate": 2.1621621621621623e-06, "loss": 0.9759, "step": 48 }, { "epoch": 0.05, "learning_rate": 2.2072072072072073e-06, "loss": 0.9649, "step": 49 }, { "epoch": 0.05, "learning_rate": 2.2522522522522524e-06, "loss": 0.919, "step": 50 }, { "epoch": 0.05, "learning_rate": 2.297297297297298e-06, "loss": 0.8621, "step": 51 }, { "epoch": 0.05, "learning_rate": 2.3423423423423424e-06, "loss": 0.8712, "step": 52 }, { "epoch": 0.05, "learning_rate": 2.3873873873873874e-06, "loss": 0.8667, "step": 53 }, { "epoch": 0.05, "learning_rate": 2.432432432432433e-06, "loss": 0.8509, "step": 54 }, { "epoch": 0.06, "learning_rate": 2.4774774774774775e-06, "loss": 0.8881, "step": 55 }, { "epoch": 0.06, "learning_rate": 2.5225225225225225e-06, "loss": 0.8258, "step": 56 }, { "epoch": 0.06, "learning_rate": 2.5675675675675675e-06, "loss": 0.8978, "step": 57 }, { "epoch": 0.06, "learning_rate": 2.612612612612613e-06, "loss": 0.8448, "step": 58 }, { "epoch": 0.06, "learning_rate": 2.657657657657658e-06, "loss": 0.799, "step": 59 }, { "epoch": 0.06, "learning_rate": 2.702702702702703e-06, "loss": 0.8464, "step": 60 }, { "epoch": 0.06, "learning_rate": 2.747747747747748e-06, "loss": 0.8702, "step": 61 }, { "epoch": 0.06, "learning_rate": 2.7927927927927926e-06, "loss": 0.8303, "step": 62 }, { "epoch": 0.06, "learning_rate": 2.837837837837838e-06, "loss": 0.8522, "step": 63 }, { "epoch": 0.06, "learning_rate": 2.882882882882883e-06, "loss": 0.8668, "step": 64 }, { "epoch": 0.07, "learning_rate": 2.927927927927928e-06, "loss": 0.7868, "step": 65 }, { "epoch": 0.07, "learning_rate": 2.9729729729729736e-06, "loss": 0.8067, "step": 66 }, { "epoch": 0.07, "learning_rate": 3.0180180180180186e-06, "loss": 0.8666, "step": 67 }, { "epoch": 0.07, "learning_rate": 3.063063063063063e-06, "loss": 0.8129, "step": 68 }, { "epoch": 0.07, "learning_rate": 3.1081081081081082e-06, "loss": 0.7559, "step": 69 }, { "epoch": 0.07, "learning_rate": 3.1531531531531532e-06, "loss": 0.7798, "step": 70 }, { "epoch": 0.07, "learning_rate": 3.1981981981981987e-06, "loss": 0.8153, "step": 71 }, { "epoch": 0.07, "learning_rate": 3.2432432432432437e-06, "loss": 0.7256, "step": 72 }, { "epoch": 0.07, "learning_rate": 3.2882882882882887e-06, "loss": 0.7061, "step": 73 }, { "epoch": 0.08, "learning_rate": 3.3333333333333333e-06, "loss": 0.832, "step": 74 }, { "epoch": 0.08, "learning_rate": 3.3783783783783788e-06, "loss": 0.8024, "step": 75 }, { "epoch": 0.08, "learning_rate": 3.423423423423424e-06, "loss": 0.7477, "step": 76 }, { "epoch": 0.08, "learning_rate": 3.468468468468469e-06, "loss": 0.7788, "step": 77 }, { "epoch": 0.08, "learning_rate": 3.513513513513514e-06, "loss": 0.7874, "step": 78 }, { "epoch": 0.08, "learning_rate": 3.5585585585585584e-06, "loss": 0.706, "step": 79 }, { "epoch": 0.08, "learning_rate": 3.603603603603604e-06, "loss": 0.8001, "step": 80 }, { "epoch": 0.08, "learning_rate": 3.648648648648649e-06, "loss": 0.7643, "step": 81 }, { "epoch": 0.08, "learning_rate": 3.693693693693694e-06, "loss": 0.7632, "step": 82 }, { "epoch": 0.08, "learning_rate": 3.7387387387387394e-06, "loss": 0.7437, "step": 83 }, { "epoch": 0.09, "learning_rate": 3.7837837837837844e-06, "loss": 0.7479, "step": 84 }, { "epoch": 0.09, "learning_rate": 3.828828828828829e-06, "loss": 0.6203, "step": 85 }, { "epoch": 0.09, "learning_rate": 3.8738738738738744e-06, "loss": 0.7012, "step": 86 }, { "epoch": 0.09, "learning_rate": 3.918918918918919e-06, "loss": 0.6985, "step": 87 }, { "epoch": 0.09, "learning_rate": 3.9639639639639645e-06, "loss": 0.7835, "step": 88 }, { "epoch": 0.09, "learning_rate": 4.009009009009009e-06, "loss": 0.7699, "step": 89 }, { "epoch": 0.09, "learning_rate": 4.0540540540540545e-06, "loss": 0.7653, "step": 90 }, { "epoch": 0.09, "learning_rate": 4.099099099099099e-06, "loss": 0.8082, "step": 91 }, { "epoch": 0.09, "learning_rate": 4.1441441441441446e-06, "loss": 0.7454, "step": 92 }, { "epoch": 0.09, "learning_rate": 4.189189189189189e-06, "loss": 0.6484, "step": 93 }, { "epoch": 0.1, "learning_rate": 4.234234234234235e-06, "loss": 0.659, "step": 94 }, { "epoch": 0.1, "learning_rate": 4.27927927927928e-06, "loss": 0.616, "step": 95 }, { "epoch": 0.1, "learning_rate": 4.324324324324325e-06, "loss": 0.7171, "step": 96 }, { "epoch": 0.1, "learning_rate": 4.369369369369369e-06, "loss": 0.7232, "step": 97 }, { "epoch": 0.1, "learning_rate": 4.414414414414415e-06, "loss": 0.6405, "step": 98 }, { "epoch": 0.1, "learning_rate": 4.45945945945946e-06, "loss": 0.7366, "step": 99 }, { "epoch": 0.1, "learning_rate": 4.504504504504505e-06, "loss": 0.7848, "step": 100 }, { "epoch": 0.1, "learning_rate": 4.54954954954955e-06, "loss": 0.6533, "step": 101 }, { "epoch": 0.1, "learning_rate": 4.594594594594596e-06, "loss": 0.6594, "step": 102 }, { "epoch": 0.1, "learning_rate": 4.63963963963964e-06, "loss": 0.6419, "step": 103 }, { "epoch": 0.11, "learning_rate": 4.684684684684685e-06, "loss": 0.6999, "step": 104 }, { "epoch": 0.11, "learning_rate": 4.72972972972973e-06, "loss": 0.6612, "step": 105 }, { "epoch": 0.11, "learning_rate": 4.774774774774775e-06, "loss": 0.7587, "step": 106 }, { "epoch": 0.11, "learning_rate": 4.81981981981982e-06, "loss": 0.6832, "step": 107 }, { "epoch": 0.11, "learning_rate": 4.864864864864866e-06, "loss": 0.707, "step": 108 }, { "epoch": 0.11, "learning_rate": 4.90990990990991e-06, "loss": 0.6147, "step": 109 }, { "epoch": 0.11, "learning_rate": 4.954954954954955e-06, "loss": 0.7146, "step": 110 }, { "epoch": 0.11, "learning_rate": 5e-06, "loss": 0.769, "step": 111 }, { "epoch": 0.11, "learning_rate": 5.045045045045045e-06, "loss": 0.6912, "step": 112 }, { "epoch": 0.11, "learning_rate": 5.0900900900900905e-06, "loss": 0.5884, "step": 113 }, { "epoch": 0.12, "learning_rate": 5.135135135135135e-06, "loss": 0.7078, "step": 114 }, { "epoch": 0.12, "learning_rate": 5.180180180180181e-06, "loss": 0.7048, "step": 115 }, { "epoch": 0.12, "learning_rate": 5.225225225225226e-06, "loss": 0.6255, "step": 116 }, { "epoch": 0.12, "learning_rate": 5.2702702702702705e-06, "loss": 0.7397, "step": 117 }, { "epoch": 0.12, "learning_rate": 5.315315315315316e-06, "loss": 0.5719, "step": 118 }, { "epoch": 0.12, "learning_rate": 5.360360360360361e-06, "loss": 0.6441, "step": 119 }, { "epoch": 0.12, "learning_rate": 5.405405405405406e-06, "loss": 0.6967, "step": 120 }, { "epoch": 0.12, "learning_rate": 5.450450450450451e-06, "loss": 0.7553, "step": 121 }, { "epoch": 0.12, "learning_rate": 5.495495495495496e-06, "loss": 0.7038, "step": 122 }, { "epoch": 0.12, "learning_rate": 5.540540540540541e-06, "loss": 0.6571, "step": 123 }, { "epoch": 0.13, "learning_rate": 5.585585585585585e-06, "loss": 0.6691, "step": 124 }, { "epoch": 0.13, "learning_rate": 5.6306306306306316e-06, "loss": 0.7455, "step": 125 }, { "epoch": 0.13, "learning_rate": 5.675675675675676e-06, "loss": 0.6737, "step": 126 }, { "epoch": 0.13, "learning_rate": 5.720720720720722e-06, "loss": 0.663, "step": 127 }, { "epoch": 0.13, "learning_rate": 5.765765765765766e-06, "loss": 0.6704, "step": 128 }, { "epoch": 0.13, "learning_rate": 5.810810810810811e-06, "loss": 0.6904, "step": 129 }, { "epoch": 0.13, "learning_rate": 5.855855855855856e-06, "loss": 0.7067, "step": 130 }, { "epoch": 0.13, "learning_rate": 5.900900900900901e-06, "loss": 0.6274, "step": 131 }, { "epoch": 0.13, "learning_rate": 5.945945945945947e-06, "loss": 0.6585, "step": 132 }, { "epoch": 0.13, "learning_rate": 5.990990990990992e-06, "loss": 0.6732, "step": 133 }, { "epoch": 0.14, "learning_rate": 6.036036036036037e-06, "loss": 0.6415, "step": 134 }, { "epoch": 0.14, "learning_rate": 6.081081081081082e-06, "loss": 0.6267, "step": 135 }, { "epoch": 0.14, "learning_rate": 6.126126126126126e-06, "loss": 0.7198, "step": 136 }, { "epoch": 0.14, "learning_rate": 6.171171171171172e-06, "loss": 0.6293, "step": 137 }, { "epoch": 0.14, "learning_rate": 6.2162162162162164e-06, "loss": 0.6281, "step": 138 }, { "epoch": 0.14, "learning_rate": 6.261261261261262e-06, "loss": 0.6519, "step": 139 }, { "epoch": 0.14, "learning_rate": 6.3063063063063065e-06, "loss": 0.6241, "step": 140 }, { "epoch": 0.14, "learning_rate": 6.351351351351351e-06, "loss": 0.5871, "step": 141 }, { "epoch": 0.14, "learning_rate": 6.396396396396397e-06, "loss": 0.5593, "step": 142 }, { "epoch": 0.15, "learning_rate": 6.441441441441442e-06, "loss": 0.6558, "step": 143 }, { "epoch": 0.15, "learning_rate": 6.486486486486487e-06, "loss": 0.6537, "step": 144 }, { "epoch": 0.15, "learning_rate": 6.531531531531532e-06, "loss": 0.5958, "step": 145 }, { "epoch": 0.15, "learning_rate": 6.5765765765765775e-06, "loss": 0.6313, "step": 146 }, { "epoch": 0.15, "learning_rate": 6.621621621621622e-06, "loss": 0.6429, "step": 147 }, { "epoch": 0.15, "learning_rate": 6.666666666666667e-06, "loss": 0.5696, "step": 148 }, { "epoch": 0.15, "learning_rate": 6.711711711711713e-06, "loss": 0.6584, "step": 149 }, { "epoch": 0.15, "learning_rate": 6.7567567567567575e-06, "loss": 0.6166, "step": 150 }, { "epoch": 0.15, "learning_rate": 6.801801801801803e-06, "loss": 0.6278, "step": 151 }, { "epoch": 0.15, "learning_rate": 6.846846846846848e-06, "loss": 0.6545, "step": 152 }, { "epoch": 0.16, "learning_rate": 6.891891891891892e-06, "loss": 0.6552, "step": 153 }, { "epoch": 0.16, "learning_rate": 6.936936936936938e-06, "loss": 0.6959, "step": 154 }, { "epoch": 0.16, "learning_rate": 6.981981981981982e-06, "loss": 0.631, "step": 155 }, { "epoch": 0.16, "learning_rate": 7.027027027027028e-06, "loss": 0.613, "step": 156 }, { "epoch": 0.16, "learning_rate": 7.072072072072072e-06, "loss": 0.6229, "step": 157 }, { "epoch": 0.16, "learning_rate": 7.117117117117117e-06, "loss": 0.6529, "step": 158 }, { "epoch": 0.16, "learning_rate": 7.162162162162163e-06, "loss": 0.6443, "step": 159 }, { "epoch": 0.16, "learning_rate": 7.207207207207208e-06, "loss": 0.6947, "step": 160 }, { "epoch": 0.16, "learning_rate": 7.252252252252253e-06, "loss": 0.6872, "step": 161 }, { "epoch": 0.16, "learning_rate": 7.297297297297298e-06, "loss": 0.6619, "step": 162 }, { "epoch": 0.17, "learning_rate": 7.342342342342343e-06, "loss": 0.567, "step": 163 }, { "epoch": 0.17, "learning_rate": 7.387387387387388e-06, "loss": 0.6313, "step": 164 }, { "epoch": 0.17, "learning_rate": 7.4324324324324324e-06, "loss": 0.691, "step": 165 }, { "epoch": 0.17, "learning_rate": 7.477477477477479e-06, "loss": 0.5837, "step": 166 }, { "epoch": 0.17, "learning_rate": 7.5225225225225225e-06, "loss": 0.6106, "step": 167 }, { "epoch": 0.17, "learning_rate": 7.567567567567569e-06, "loss": 0.6413, "step": 168 }, { "epoch": 0.17, "learning_rate": 7.612612612612613e-06, "loss": 0.5383, "step": 169 }, { "epoch": 0.17, "learning_rate": 7.657657657657658e-06, "loss": 0.7011, "step": 170 }, { "epoch": 0.17, "learning_rate": 7.702702702702704e-06, "loss": 0.7578, "step": 171 }, { "epoch": 0.17, "learning_rate": 7.747747747747749e-06, "loss": 0.5794, "step": 172 }, { "epoch": 0.18, "learning_rate": 7.792792792792793e-06, "loss": 0.5652, "step": 173 }, { "epoch": 0.18, "learning_rate": 7.837837837837838e-06, "loss": 0.6383, "step": 174 }, { "epoch": 0.18, "learning_rate": 7.882882882882884e-06, "loss": 0.5821, "step": 175 }, { "epoch": 0.18, "learning_rate": 7.927927927927929e-06, "loss": 0.6151, "step": 176 }, { "epoch": 0.18, "learning_rate": 7.972972972972974e-06, "loss": 0.6028, "step": 177 }, { "epoch": 0.18, "learning_rate": 8.018018018018018e-06, "loss": 0.6266, "step": 178 }, { "epoch": 0.18, "learning_rate": 8.063063063063063e-06, "loss": 0.5739, "step": 179 }, { "epoch": 0.18, "learning_rate": 8.108108108108109e-06, "loss": 0.7055, "step": 180 }, { "epoch": 0.18, "learning_rate": 8.153153153153154e-06, "loss": 0.6076, "step": 181 }, { "epoch": 0.18, "learning_rate": 8.198198198198198e-06, "loss": 0.6481, "step": 182 }, { "epoch": 0.19, "learning_rate": 8.243243243243245e-06, "loss": 0.6165, "step": 183 }, { "epoch": 0.19, "learning_rate": 8.288288288288289e-06, "loss": 0.5923, "step": 184 }, { "epoch": 0.19, "learning_rate": 8.333333333333334e-06, "loss": 0.6189, "step": 185 }, { "epoch": 0.19, "learning_rate": 8.378378378378378e-06, "loss": 0.5698, "step": 186 }, { "epoch": 0.19, "learning_rate": 8.423423423423423e-06, "loss": 0.6975, "step": 187 }, { "epoch": 0.19, "learning_rate": 8.46846846846847e-06, "loss": 0.6372, "step": 188 }, { "epoch": 0.19, "learning_rate": 8.513513513513514e-06, "loss": 0.7062, "step": 189 }, { "epoch": 0.19, "learning_rate": 8.55855855855856e-06, "loss": 0.6191, "step": 190 }, { "epoch": 0.19, "learning_rate": 8.603603603603605e-06, "loss": 0.6412, "step": 191 }, { "epoch": 0.19, "learning_rate": 8.64864864864865e-06, "loss": 0.5911, "step": 192 }, { "epoch": 0.2, "learning_rate": 8.693693693693694e-06, "loss": 0.5392, "step": 193 }, { "epoch": 0.2, "learning_rate": 8.738738738738739e-06, "loss": 0.5775, "step": 194 }, { "epoch": 0.2, "learning_rate": 8.783783783783785e-06, "loss": 0.6752, "step": 195 }, { "epoch": 0.2, "learning_rate": 8.82882882882883e-06, "loss": 0.6693, "step": 196 }, { "epoch": 0.2, "learning_rate": 8.873873873873876e-06, "loss": 0.6977, "step": 197 }, { "epoch": 0.2, "learning_rate": 8.91891891891892e-06, "loss": 0.6253, "step": 198 }, { "epoch": 0.2, "learning_rate": 8.963963963963965e-06, "loss": 0.6523, "step": 199 }, { "epoch": 0.2, "learning_rate": 9.00900900900901e-06, "loss": 0.5519, "step": 200 }, { "epoch": 0.2, "learning_rate": 9.054054054054054e-06, "loss": 0.5601, "step": 201 }, { "epoch": 0.2, "learning_rate": 9.0990990990991e-06, "loss": 0.6516, "step": 202 }, { "epoch": 0.21, "learning_rate": 9.144144144144145e-06, "loss": 0.6619, "step": 203 }, { "epoch": 0.21, "learning_rate": 9.189189189189191e-06, "loss": 0.659, "step": 204 }, { "epoch": 0.21, "learning_rate": 9.234234234234236e-06, "loss": 0.618, "step": 205 }, { "epoch": 0.21, "learning_rate": 9.27927927927928e-06, "loss": 0.6292, "step": 206 }, { "epoch": 0.21, "learning_rate": 9.324324324324325e-06, "loss": 0.5677, "step": 207 }, { "epoch": 0.21, "learning_rate": 9.36936936936937e-06, "loss": 0.5945, "step": 208 }, { "epoch": 0.21, "learning_rate": 9.414414414414416e-06, "loss": 0.6781, "step": 209 }, { "epoch": 0.21, "learning_rate": 9.45945945945946e-06, "loss": 0.6144, "step": 210 }, { "epoch": 0.21, "learning_rate": 9.504504504504505e-06, "loss": 0.5094, "step": 211 }, { "epoch": 0.22, "learning_rate": 9.54954954954955e-06, "loss": 0.6466, "step": 212 }, { "epoch": 0.22, "learning_rate": 9.594594594594594e-06, "loss": 0.6025, "step": 213 }, { "epoch": 0.22, "learning_rate": 9.63963963963964e-06, "loss": 0.5852, "step": 214 }, { "epoch": 0.22, "learning_rate": 9.684684684684685e-06, "loss": 0.6984, "step": 215 }, { "epoch": 0.22, "learning_rate": 9.729729729729732e-06, "loss": 0.5656, "step": 216 }, { "epoch": 0.22, "learning_rate": 9.774774774774776e-06, "loss": 0.6619, "step": 217 }, { "epoch": 0.22, "learning_rate": 9.81981981981982e-06, "loss": 0.6307, "step": 218 }, { "epoch": 0.22, "learning_rate": 9.864864864864865e-06, "loss": 0.6592, "step": 219 }, { "epoch": 0.22, "learning_rate": 9.90990990990991e-06, "loss": 0.6066, "step": 220 }, { "epoch": 0.22, "learning_rate": 9.954954954954956e-06, "loss": 0.6747, "step": 221 }, { "epoch": 0.23, "learning_rate": 1e-05, "loss": 0.6495, "step": 222 }, { "epoch": 0.23, "learning_rate": 1.0045045045045045e-05, "loss": 0.6062, "step": 223 }, { "epoch": 0.23, "learning_rate": 1.009009009009009e-05, "loss": 0.6322, "step": 224 }, { "epoch": 0.23, "learning_rate": 1.0135135135135136e-05, "loss": 0.5622, "step": 225 }, { "epoch": 0.23, "learning_rate": 1.0180180180180181e-05, "loss": 0.5406, "step": 226 }, { "epoch": 0.23, "learning_rate": 1.0225225225225226e-05, "loss": 0.5479, "step": 227 }, { "epoch": 0.23, "learning_rate": 1.027027027027027e-05, "loss": 0.5668, "step": 228 }, { "epoch": 0.23, "learning_rate": 1.0315315315315315e-05, "loss": 0.7043, "step": 229 }, { "epoch": 0.23, "learning_rate": 1.0360360360360363e-05, "loss": 0.5753, "step": 230 }, { "epoch": 0.23, "learning_rate": 1.0405405405405407e-05, "loss": 0.6463, "step": 231 }, { "epoch": 0.24, "learning_rate": 1.0450450450450452e-05, "loss": 0.6281, "step": 232 }, { "epoch": 0.24, "learning_rate": 1.0495495495495496e-05, "loss": 0.6238, "step": 233 }, { "epoch": 0.24, "learning_rate": 1.0540540540540541e-05, "loss": 0.5574, "step": 234 }, { "epoch": 0.24, "learning_rate": 1.0585585585585587e-05, "loss": 0.6076, "step": 235 }, { "epoch": 0.24, "learning_rate": 1.0630630630630632e-05, "loss": 0.5551, "step": 236 }, { "epoch": 0.24, "learning_rate": 1.0675675675675677e-05, "loss": 0.5296, "step": 237 }, { "epoch": 0.24, "learning_rate": 1.0720720720720721e-05, "loss": 0.6479, "step": 238 }, { "epoch": 0.24, "learning_rate": 1.0765765765765767e-05, "loss": 0.6112, "step": 239 }, { "epoch": 0.24, "learning_rate": 1.0810810810810812e-05, "loss": 0.5312, "step": 240 }, { "epoch": 0.24, "learning_rate": 1.0855855855855857e-05, "loss": 0.6728, "step": 241 }, { "epoch": 0.25, "learning_rate": 1.0900900900900901e-05, "loss": 0.5791, "step": 242 }, { "epoch": 0.25, "learning_rate": 1.0945945945945946e-05, "loss": 0.7036, "step": 243 }, { "epoch": 0.25, "learning_rate": 1.0990990990990992e-05, "loss": 0.6575, "step": 244 }, { "epoch": 0.25, "learning_rate": 1.1036036036036037e-05, "loss": 0.6305, "step": 245 }, { "epoch": 0.25, "learning_rate": 1.1081081081081081e-05, "loss": 0.6023, "step": 246 }, { "epoch": 0.25, "learning_rate": 1.1126126126126126e-05, "loss": 0.648, "step": 247 }, { "epoch": 0.25, "learning_rate": 1.117117117117117e-05, "loss": 0.6388, "step": 248 }, { "epoch": 0.25, "learning_rate": 1.1216216216216219e-05, "loss": 0.5415, "step": 249 }, { "epoch": 0.25, "learning_rate": 1.1261261261261263e-05, "loss": 0.5951, "step": 250 }, { "epoch": 0.25, "learning_rate": 1.1306306306306308e-05, "loss": 0.6176, "step": 251 }, { "epoch": 0.26, "learning_rate": 1.1351351351351352e-05, "loss": 0.6217, "step": 252 }, { "epoch": 0.26, "learning_rate": 1.1396396396396397e-05, "loss": 0.5765, "step": 253 }, { "epoch": 0.26, "learning_rate": 1.1441441441441443e-05, "loss": 0.61, "step": 254 }, { "epoch": 0.26, "learning_rate": 1.1486486486486488e-05, "loss": 0.596, "step": 255 }, { "epoch": 0.26, "learning_rate": 1.1531531531531532e-05, "loss": 0.6579, "step": 256 }, { "epoch": 0.26, "learning_rate": 1.1576576576576577e-05, "loss": 0.6967, "step": 257 }, { "epoch": 0.26, "learning_rate": 1.1621621621621622e-05, "loss": 0.6673, "step": 258 }, { "epoch": 0.26, "learning_rate": 1.1666666666666668e-05, "loss": 0.6341, "step": 259 }, { "epoch": 0.26, "learning_rate": 1.1711711711711713e-05, "loss": 0.5815, "step": 260 }, { "epoch": 0.26, "learning_rate": 1.1756756756756757e-05, "loss": 0.6228, "step": 261 }, { "epoch": 0.27, "learning_rate": 1.1801801801801802e-05, "loss": 0.6621, "step": 262 }, { "epoch": 0.27, "learning_rate": 1.1846846846846846e-05, "loss": 0.564, "step": 263 }, { "epoch": 0.27, "learning_rate": 1.1891891891891894e-05, "loss": 0.5758, "step": 264 }, { "epoch": 0.27, "learning_rate": 1.1936936936936939e-05, "loss": 0.6269, "step": 265 }, { "epoch": 0.27, "learning_rate": 1.1981981981981983e-05, "loss": 0.5859, "step": 266 }, { "epoch": 0.27, "learning_rate": 1.2027027027027028e-05, "loss": 0.5936, "step": 267 }, { "epoch": 0.27, "learning_rate": 1.2072072072072074e-05, "loss": 0.5904, "step": 268 }, { "epoch": 0.27, "learning_rate": 1.2117117117117119e-05, "loss": 0.5774, "step": 269 }, { "epoch": 0.27, "learning_rate": 1.2162162162162164e-05, "loss": 0.5949, "step": 270 }, { "epoch": 0.27, "learning_rate": 1.2207207207207208e-05, "loss": 0.5935, "step": 271 }, { "epoch": 0.28, "learning_rate": 1.2252252252252253e-05, "loss": 0.5946, "step": 272 }, { "epoch": 0.28, "learning_rate": 1.2297297297297299e-05, "loss": 0.5866, "step": 273 }, { "epoch": 0.28, "learning_rate": 1.2342342342342344e-05, "loss": 0.6008, "step": 274 }, { "epoch": 0.28, "learning_rate": 1.2387387387387388e-05, "loss": 0.6262, "step": 275 }, { "epoch": 0.28, "learning_rate": 1.2432432432432433e-05, "loss": 0.6583, "step": 276 }, { "epoch": 0.28, "learning_rate": 1.2477477477477477e-05, "loss": 0.6057, "step": 277 }, { "epoch": 0.28, "learning_rate": 1.2522522522522524e-05, "loss": 0.5339, "step": 278 }, { "epoch": 0.28, "learning_rate": 1.2567567567567568e-05, "loss": 0.5553, "step": 279 }, { "epoch": 0.28, "learning_rate": 1.2612612612612613e-05, "loss": 0.5303, "step": 280 }, { "epoch": 0.29, "learning_rate": 1.2657657657657658e-05, "loss": 0.5518, "step": 281 }, { "epoch": 0.29, "learning_rate": 1.2702702702702702e-05, "loss": 0.5924, "step": 282 }, { "epoch": 0.29, "learning_rate": 1.274774774774775e-05, "loss": 0.6376, "step": 283 }, { "epoch": 0.29, "learning_rate": 1.2792792792792795e-05, "loss": 0.5246, "step": 284 }, { "epoch": 0.29, "learning_rate": 1.283783783783784e-05, "loss": 0.6782, "step": 285 }, { "epoch": 0.29, "learning_rate": 1.2882882882882884e-05, "loss": 0.6225, "step": 286 }, { "epoch": 0.29, "learning_rate": 1.2927927927927929e-05, "loss": 0.6421, "step": 287 }, { "epoch": 0.29, "learning_rate": 1.2972972972972975e-05, "loss": 0.5976, "step": 288 }, { "epoch": 0.29, "learning_rate": 1.301801801801802e-05, "loss": 0.5869, "step": 289 }, { "epoch": 0.29, "learning_rate": 1.3063063063063064e-05, "loss": 0.6019, "step": 290 }, { "epoch": 0.3, "learning_rate": 1.3108108108108109e-05, "loss": 0.6597, "step": 291 }, { "epoch": 0.3, "learning_rate": 1.3153153153153155e-05, "loss": 0.6247, "step": 292 }, { "epoch": 0.3, "learning_rate": 1.31981981981982e-05, "loss": 0.5828, "step": 293 }, { "epoch": 0.3, "learning_rate": 1.3243243243243244e-05, "loss": 0.5798, "step": 294 }, { "epoch": 0.3, "learning_rate": 1.3288288288288289e-05, "loss": 0.5574, "step": 295 }, { "epoch": 0.3, "learning_rate": 1.3333333333333333e-05, "loss": 0.6334, "step": 296 }, { "epoch": 0.3, "learning_rate": 1.3378378378378381e-05, "loss": 0.6632, "step": 297 }, { "epoch": 0.3, "learning_rate": 1.3423423423423426e-05, "loss": 0.6392, "step": 298 }, { "epoch": 0.3, "learning_rate": 1.346846846846847e-05, "loss": 0.5954, "step": 299 }, { "epoch": 0.3, "learning_rate": 1.3513513513513515e-05, "loss": 0.6046, "step": 300 }, { "epoch": 0.31, "learning_rate": 1.355855855855856e-05, "loss": 0.7199, "step": 301 }, { "epoch": 0.31, "learning_rate": 1.3603603603603606e-05, "loss": 0.5831, "step": 302 }, { "epoch": 0.31, "learning_rate": 1.364864864864865e-05, "loss": 0.6111, "step": 303 }, { "epoch": 0.31, "learning_rate": 1.3693693693693695e-05, "loss": 0.6625, "step": 304 }, { "epoch": 0.31, "learning_rate": 1.373873873873874e-05, "loss": 0.5973, "step": 305 }, { "epoch": 0.31, "learning_rate": 1.3783783783783784e-05, "loss": 0.519, "step": 306 }, { "epoch": 0.31, "learning_rate": 1.382882882882883e-05, "loss": 0.6452, "step": 307 }, { "epoch": 0.31, "learning_rate": 1.3873873873873875e-05, "loss": 0.5764, "step": 308 }, { "epoch": 0.31, "learning_rate": 1.391891891891892e-05, "loss": 0.6376, "step": 309 }, { "epoch": 0.31, "learning_rate": 1.3963963963963964e-05, "loss": 0.6052, "step": 310 }, { "epoch": 0.32, "learning_rate": 1.4009009009009009e-05, "loss": 0.6169, "step": 311 }, { "epoch": 0.32, "learning_rate": 1.4054054054054055e-05, "loss": 0.5703, "step": 312 }, { "epoch": 0.32, "learning_rate": 1.40990990990991e-05, "loss": 0.5805, "step": 313 }, { "epoch": 0.32, "learning_rate": 1.4144144144144145e-05, "loss": 0.4879, "step": 314 }, { "epoch": 0.32, "learning_rate": 1.4189189189189189e-05, "loss": 0.6093, "step": 315 }, { "epoch": 0.32, "learning_rate": 1.4234234234234234e-05, "loss": 0.5264, "step": 316 }, { "epoch": 0.32, "learning_rate": 1.4279279279279282e-05, "loss": 0.5411, "step": 317 }, { "epoch": 0.32, "learning_rate": 1.4324324324324326e-05, "loss": 0.6065, "step": 318 }, { "epoch": 0.32, "learning_rate": 1.4369369369369371e-05, "loss": 0.5885, "step": 319 }, { "epoch": 0.32, "learning_rate": 1.4414414414414416e-05, "loss": 0.6249, "step": 320 }, { "epoch": 0.33, "learning_rate": 1.4459459459459462e-05, "loss": 0.6009, "step": 321 }, { "epoch": 0.33, "learning_rate": 1.4504504504504506e-05, "loss": 0.5557, "step": 322 }, { "epoch": 0.33, "learning_rate": 1.4549549549549551e-05, "loss": 0.6448, "step": 323 }, { "epoch": 0.33, "learning_rate": 1.4594594594594596e-05, "loss": 0.5773, "step": 324 }, { "epoch": 0.33, "learning_rate": 1.463963963963964e-05, "loss": 0.5565, "step": 325 }, { "epoch": 0.33, "learning_rate": 1.4684684684684686e-05, "loss": 0.6216, "step": 326 }, { "epoch": 0.33, "learning_rate": 1.4729729729729731e-05, "loss": 0.55, "step": 327 }, { "epoch": 0.33, "learning_rate": 1.4774774774774776e-05, "loss": 0.5338, "step": 328 }, { "epoch": 0.33, "learning_rate": 1.481981981981982e-05, "loss": 0.532, "step": 329 }, { "epoch": 0.33, "learning_rate": 1.4864864864864865e-05, "loss": 0.5559, "step": 330 }, { "epoch": 0.34, "learning_rate": 1.4909909909909913e-05, "loss": 0.559, "step": 331 }, { "epoch": 0.34, "learning_rate": 1.4954954954954957e-05, "loss": 0.5772, "step": 332 }, { "epoch": 0.34, "learning_rate": 1.5000000000000002e-05, "loss": 0.5673, "step": 333 }, { "epoch": 0.34, "learning_rate": 1.5045045045045045e-05, "loss": 0.5791, "step": 334 }, { "epoch": 0.34, "learning_rate": 1.509009009009009e-05, "loss": 0.5567, "step": 335 }, { "epoch": 0.34, "learning_rate": 1.5135135135135138e-05, "loss": 0.5357, "step": 336 }, { "epoch": 0.34, "learning_rate": 1.5180180180180182e-05, "loss": 0.5429, "step": 337 }, { "epoch": 0.34, "learning_rate": 1.5225225225225227e-05, "loss": 0.552, "step": 338 }, { "epoch": 0.34, "learning_rate": 1.527027027027027e-05, "loss": 0.617, "step": 339 }, { "epoch": 0.35, "learning_rate": 1.5315315315315316e-05, "loss": 0.5487, "step": 340 }, { "epoch": 0.35, "learning_rate": 1.5360360360360362e-05, "loss": 0.5526, "step": 341 }, { "epoch": 0.35, "learning_rate": 1.540540540540541e-05, "loss": 0.5679, "step": 342 }, { "epoch": 0.35, "learning_rate": 1.545045045045045e-05, "loss": 0.5855, "step": 343 }, { "epoch": 0.35, "learning_rate": 1.5495495495495498e-05, "loss": 0.6932, "step": 344 }, { "epoch": 0.35, "learning_rate": 1.554054054054054e-05, "loss": 0.6451, "step": 345 }, { "epoch": 0.35, "learning_rate": 1.5585585585585587e-05, "loss": 0.6112, "step": 346 }, { "epoch": 0.35, "learning_rate": 1.5630630630630633e-05, "loss": 0.5492, "step": 347 }, { "epoch": 0.35, "learning_rate": 1.5675675675675676e-05, "loss": 0.5996, "step": 348 }, { "epoch": 0.35, "learning_rate": 1.5720720720720722e-05, "loss": 0.521, "step": 349 }, { "epoch": 0.36, "learning_rate": 1.576576576576577e-05, "loss": 0.6212, "step": 350 }, { "epoch": 0.36, "learning_rate": 1.581081081081081e-05, "loss": 0.5517, "step": 351 }, { "epoch": 0.36, "learning_rate": 1.5855855855855858e-05, "loss": 0.5545, "step": 352 }, { "epoch": 0.36, "learning_rate": 1.59009009009009e-05, "loss": 0.5622, "step": 353 }, { "epoch": 0.36, "learning_rate": 1.5945945945945947e-05, "loss": 0.577, "step": 354 }, { "epoch": 0.36, "learning_rate": 1.5990990990990993e-05, "loss": 0.6152, "step": 355 }, { "epoch": 0.36, "learning_rate": 1.6036036036036036e-05, "loss": 0.6111, "step": 356 }, { "epoch": 0.36, "learning_rate": 1.6081081081081083e-05, "loss": 0.6042, "step": 357 }, { "epoch": 0.36, "learning_rate": 1.6126126126126126e-05, "loss": 0.5836, "step": 358 }, { "epoch": 0.36, "learning_rate": 1.6171171171171172e-05, "loss": 0.5947, "step": 359 }, { "epoch": 0.37, "learning_rate": 1.6216216216216218e-05, "loss": 0.6127, "step": 360 }, { "epoch": 0.37, "learning_rate": 1.6261261261261264e-05, "loss": 0.6552, "step": 361 }, { "epoch": 0.37, "learning_rate": 1.6306306306306307e-05, "loss": 0.5746, "step": 362 }, { "epoch": 0.37, "learning_rate": 1.6351351351351354e-05, "loss": 0.5397, "step": 363 }, { "epoch": 0.37, "learning_rate": 1.6396396396396396e-05, "loss": 0.543, "step": 364 }, { "epoch": 0.37, "learning_rate": 1.6441441441441443e-05, "loss": 0.6187, "step": 365 }, { "epoch": 0.37, "learning_rate": 1.648648648648649e-05, "loss": 0.5607, "step": 366 }, { "epoch": 0.37, "learning_rate": 1.6531531531531532e-05, "loss": 0.5144, "step": 367 }, { "epoch": 0.37, "learning_rate": 1.6576576576576578e-05, "loss": 0.6671, "step": 368 }, { "epoch": 0.37, "learning_rate": 1.662162162162162e-05, "loss": 0.5572, "step": 369 }, { "epoch": 0.38, "learning_rate": 1.6666666666666667e-05, "loss": 0.5547, "step": 370 }, { "epoch": 0.38, "learning_rate": 1.6711711711711714e-05, "loss": 0.6171, "step": 371 }, { "epoch": 0.38, "learning_rate": 1.6756756756756757e-05, "loss": 0.5488, "step": 372 }, { "epoch": 0.38, "learning_rate": 1.6801801801801803e-05, "loss": 0.5265, "step": 373 }, { "epoch": 0.38, "learning_rate": 1.6846846846846846e-05, "loss": 0.5699, "step": 374 }, { "epoch": 0.38, "learning_rate": 1.6891891891891896e-05, "loss": 0.5605, "step": 375 }, { "epoch": 0.38, "learning_rate": 1.693693693693694e-05, "loss": 0.5677, "step": 376 }, { "epoch": 0.38, "learning_rate": 1.6981981981981985e-05, "loss": 0.6297, "step": 377 }, { "epoch": 0.38, "learning_rate": 1.7027027027027028e-05, "loss": 0.54, "step": 378 }, { "epoch": 0.38, "learning_rate": 1.7072072072072074e-05, "loss": 0.6082, "step": 379 }, { "epoch": 0.39, "learning_rate": 1.711711711711712e-05, "loss": 0.6107, "step": 380 }, { "epoch": 0.39, "learning_rate": 1.7162162162162163e-05, "loss": 0.4891, "step": 381 }, { "epoch": 0.39, "learning_rate": 1.720720720720721e-05, "loss": 0.5493, "step": 382 }, { "epoch": 0.39, "learning_rate": 1.7252252252252252e-05, "loss": 0.5271, "step": 383 }, { "epoch": 0.39, "learning_rate": 1.72972972972973e-05, "loss": 0.5614, "step": 384 }, { "epoch": 0.39, "learning_rate": 1.7342342342342345e-05, "loss": 0.5685, "step": 385 }, { "epoch": 0.39, "learning_rate": 1.7387387387387388e-05, "loss": 0.5908, "step": 386 }, { "epoch": 0.39, "learning_rate": 1.7432432432432434e-05, "loss": 0.5739, "step": 387 }, { "epoch": 0.39, "learning_rate": 1.7477477477477477e-05, "loss": 0.5784, "step": 388 }, { "epoch": 0.39, "learning_rate": 1.7522522522522523e-05, "loss": 0.5274, "step": 389 }, { "epoch": 0.4, "learning_rate": 1.756756756756757e-05, "loss": 0.5256, "step": 390 }, { "epoch": 0.4, "learning_rate": 1.7612612612612613e-05, "loss": 0.6323, "step": 391 }, { "epoch": 0.4, "learning_rate": 1.765765765765766e-05, "loss": 0.6278, "step": 392 }, { "epoch": 0.4, "learning_rate": 1.7702702702702702e-05, "loss": 0.5382, "step": 393 }, { "epoch": 0.4, "learning_rate": 1.774774774774775e-05, "loss": 0.6631, "step": 394 }, { "epoch": 0.4, "learning_rate": 1.7792792792792794e-05, "loss": 0.5606, "step": 395 }, { "epoch": 0.4, "learning_rate": 1.783783783783784e-05, "loss": 0.5105, "step": 396 }, { "epoch": 0.4, "learning_rate": 1.7882882882882883e-05, "loss": 0.5984, "step": 397 }, { "epoch": 0.4, "learning_rate": 1.792792792792793e-05, "loss": 0.5543, "step": 398 }, { "epoch": 0.4, "learning_rate": 1.7972972972972976e-05, "loss": 0.517, "step": 399 }, { "epoch": 0.41, "learning_rate": 1.801801801801802e-05, "loss": 0.6643, "step": 400 }, { "epoch": 0.41, "learning_rate": 1.8063063063063065e-05, "loss": 0.6515, "step": 401 }, { "epoch": 0.41, "learning_rate": 1.8108108108108108e-05, "loss": 0.5475, "step": 402 }, { "epoch": 0.41, "learning_rate": 1.8153153153153154e-05, "loss": 0.5457, "step": 403 }, { "epoch": 0.41, "learning_rate": 1.81981981981982e-05, "loss": 0.6033, "step": 404 }, { "epoch": 0.41, "learning_rate": 1.8243243243243244e-05, "loss": 0.5285, "step": 405 }, { "epoch": 0.41, "learning_rate": 1.828828828828829e-05, "loss": 0.5709, "step": 406 }, { "epoch": 0.41, "learning_rate": 1.8333333333333333e-05, "loss": 0.6133, "step": 407 }, { "epoch": 0.41, "learning_rate": 1.8378378378378383e-05, "loss": 0.5215, "step": 408 }, { "epoch": 0.42, "learning_rate": 1.8423423423423425e-05, "loss": 0.5722, "step": 409 }, { "epoch": 0.42, "learning_rate": 1.8468468468468472e-05, "loss": 0.5247, "step": 410 }, { "epoch": 0.42, "learning_rate": 1.8513513513513515e-05, "loss": 0.582, "step": 411 }, { "epoch": 0.42, "learning_rate": 1.855855855855856e-05, "loss": 0.5851, "step": 412 }, { "epoch": 0.42, "learning_rate": 1.8603603603603607e-05, "loss": 0.6099, "step": 413 }, { "epoch": 0.42, "learning_rate": 1.864864864864865e-05, "loss": 0.6221, "step": 414 }, { "epoch": 0.42, "learning_rate": 1.8693693693693696e-05, "loss": 0.5985, "step": 415 }, { "epoch": 0.42, "learning_rate": 1.873873873873874e-05, "loss": 0.5521, "step": 416 }, { "epoch": 0.42, "learning_rate": 1.8783783783783786e-05, "loss": 0.5962, "step": 417 }, { "epoch": 0.42, "learning_rate": 1.8828828828828832e-05, "loss": 0.5196, "step": 418 }, { "epoch": 0.43, "learning_rate": 1.8873873873873875e-05, "loss": 0.6123, "step": 419 }, { "epoch": 0.43, "learning_rate": 1.891891891891892e-05, "loss": 0.5191, "step": 420 }, { "epoch": 0.43, "learning_rate": 1.8963963963963964e-05, "loss": 0.6239, "step": 421 }, { "epoch": 0.43, "learning_rate": 1.900900900900901e-05, "loss": 0.5223, "step": 422 }, { "epoch": 0.43, "learning_rate": 1.9054054054054057e-05, "loss": 0.5492, "step": 423 }, { "epoch": 0.43, "learning_rate": 1.90990990990991e-05, "loss": 0.6846, "step": 424 }, { "epoch": 0.43, "learning_rate": 1.9144144144144146e-05, "loss": 0.5626, "step": 425 }, { "epoch": 0.43, "learning_rate": 1.918918918918919e-05, "loss": 0.5029, "step": 426 }, { "epoch": 0.43, "learning_rate": 1.9234234234234235e-05, "loss": 0.5833, "step": 427 }, { "epoch": 0.43, "learning_rate": 1.927927927927928e-05, "loss": 0.5668, "step": 428 }, { "epoch": 0.44, "learning_rate": 1.9324324324324328e-05, "loss": 0.5992, "step": 429 }, { "epoch": 0.44, "learning_rate": 1.936936936936937e-05, "loss": 0.5569, "step": 430 }, { "epoch": 0.44, "learning_rate": 1.9414414414414417e-05, "loss": 0.5725, "step": 431 }, { "epoch": 0.44, "learning_rate": 1.9459459459459463e-05, "loss": 0.5657, "step": 432 }, { "epoch": 0.44, "learning_rate": 1.9504504504504506e-05, "loss": 0.5468, "step": 433 }, { "epoch": 0.44, "learning_rate": 1.9549549549549552e-05, "loss": 0.4937, "step": 434 }, { "epoch": 0.44, "learning_rate": 1.9594594594594595e-05, "loss": 0.5099, "step": 435 }, { "epoch": 0.44, "learning_rate": 1.963963963963964e-05, "loss": 0.6248, "step": 436 }, { "epoch": 0.44, "learning_rate": 1.9684684684684688e-05, "loss": 0.4376, "step": 437 }, { "epoch": 0.44, "learning_rate": 1.972972972972973e-05, "loss": 0.4925, "step": 438 }, { "epoch": 0.45, "learning_rate": 1.9774774774774777e-05, "loss": 0.5803, "step": 439 }, { "epoch": 0.45, "learning_rate": 1.981981981981982e-05, "loss": 0.4569, "step": 440 }, { "epoch": 0.45, "learning_rate": 1.9864864864864866e-05, "loss": 0.5568, "step": 441 }, { "epoch": 0.45, "learning_rate": 1.9909909909909912e-05, "loss": 0.5917, "step": 442 }, { "epoch": 0.45, "learning_rate": 1.995495495495496e-05, "loss": 0.5608, "step": 443 }, { "epoch": 0.45, "learning_rate": 2e-05, "loss": 0.533, "step": 444 }, { "epoch": 0.45, "learning_rate": 1.999999975972048e-05, "loss": 0.5105, "step": 445 }, { "epoch": 0.45, "learning_rate": 1.999999903888193e-05, "loss": 0.5999, "step": 446 }, { "epoch": 0.45, "learning_rate": 1.999999783748439e-05, "loss": 0.5604, "step": 447 }, { "epoch": 0.45, "learning_rate": 1.999999615552791e-05, "loss": 0.6036, "step": 448 }, { "epoch": 0.46, "learning_rate": 1.9999993993012573e-05, "loss": 0.5823, "step": 449 }, { "epoch": 0.46, "learning_rate": 1.999999134993849e-05, "loss": 0.5409, "step": 450 }, { "epoch": 0.46, "learning_rate": 1.999998822630578e-05, "loss": 0.6017, "step": 451 }, { "epoch": 0.46, "learning_rate": 1.999998462211459e-05, "loss": 0.562, "step": 452 }, { "epoch": 0.46, "learning_rate": 1.9999980537365105e-05, "loss": 0.5493, "step": 453 }, { "epoch": 0.46, "learning_rate": 1.9999975972057514e-05, "loss": 0.6253, "step": 454 }, { "epoch": 0.46, "learning_rate": 1.9999970926192035e-05, "loss": 0.5499, "step": 455 }, { "epoch": 0.46, "learning_rate": 1.999996539976891e-05, "loss": 0.6191, "step": 456 }, { "epoch": 0.46, "learning_rate": 1.9999959392788415e-05, "loss": 0.4954, "step": 457 }, { "epoch": 0.46, "learning_rate": 1.9999952905250826e-05, "loss": 0.5708, "step": 458 }, { "epoch": 0.47, "learning_rate": 1.999994593715646e-05, "loss": 0.5225, "step": 459 }, { "epoch": 0.47, "learning_rate": 1.9999938488505658e-05, "loss": 0.5636, "step": 460 }, { "epoch": 0.47, "learning_rate": 1.9999930559298766e-05, "loss": 0.517, "step": 461 }, { "epoch": 0.47, "learning_rate": 1.999992214953617e-05, "loss": 0.5682, "step": 462 }, { "epoch": 0.47, "learning_rate": 1.999991325921828e-05, "loss": 0.5895, "step": 463 }, { "epoch": 0.47, "learning_rate": 1.9999903888345512e-05, "loss": 0.4935, "step": 464 }, { "epoch": 0.47, "learning_rate": 1.9999894036918326e-05, "loss": 0.5536, "step": 465 }, { "epoch": 0.47, "learning_rate": 1.9999883704937192e-05, "loss": 0.5904, "step": 466 }, { "epoch": 0.47, "learning_rate": 1.99998728924026e-05, "loss": 0.635, "step": 467 }, { "epoch": 0.47, "learning_rate": 1.9999861599315083e-05, "loss": 0.5099, "step": 468 }, { "epoch": 0.48, "learning_rate": 1.9999849825675176e-05, "loss": 0.544, "step": 469 }, { "epoch": 0.48, "learning_rate": 1.9999837571483445e-05, "loss": 0.5933, "step": 470 }, { "epoch": 0.48, "learning_rate": 1.9999824836740477e-05, "loss": 0.6294, "step": 471 }, { "epoch": 0.48, "learning_rate": 1.999981162144689e-05, "loss": 0.6933, "step": 472 }, { "epoch": 0.48, "learning_rate": 1.9999797925603313e-05, "loss": 0.5805, "step": 473 }, { "epoch": 0.48, "learning_rate": 1.9999783749210408e-05, "loss": 0.5996, "step": 474 }, { "epoch": 0.48, "learning_rate": 1.9999769092268854e-05, "loss": 0.6599, "step": 475 }, { "epoch": 0.48, "learning_rate": 1.9999753954779356e-05, "loss": 0.5585, "step": 476 }, { "epoch": 0.48, "learning_rate": 1.999973833674264e-05, "loss": 0.5921, "step": 477 }, { "epoch": 0.49, "learning_rate": 1.999972223815946e-05, "loss": 0.507, "step": 478 }, { "epoch": 0.49, "learning_rate": 1.9999705659030588e-05, "loss": 0.5119, "step": 479 }, { "epoch": 0.49, "learning_rate": 1.999968859935682e-05, "loss": 0.5948, "step": 480 }, { "epoch": 0.49, "learning_rate": 1.9999671059138978e-05, "loss": 0.51, "step": 481 }, { "epoch": 0.49, "learning_rate": 1.9999653038377896e-05, "loss": 0.6508, "step": 482 }, { "epoch": 0.49, "learning_rate": 1.9999634537074458e-05, "loss": 0.5588, "step": 483 }, { "epoch": 0.49, "learning_rate": 1.9999615555229533e-05, "loss": 0.4581, "step": 484 }, { "epoch": 0.49, "learning_rate": 1.999959609284405e-05, "loss": 0.6428, "step": 485 }, { "epoch": 0.49, "learning_rate": 1.999957614991893e-05, "loss": 0.6184, "step": 486 }, { "epoch": 0.49, "learning_rate": 1.9999555726455142e-05, "loss": 0.5443, "step": 487 }, { "epoch": 0.5, "learning_rate": 1.9999534822453664e-05, "loss": 0.4905, "step": 488 }, { "epoch": 0.5, "learning_rate": 1.9999513437915498e-05, "loss": 0.618, "step": 489 }, { "epoch": 0.5, "learning_rate": 1.9999491572841676e-05, "loss": 0.5708, "step": 490 }, { "epoch": 0.5, "learning_rate": 1.9999469227233247e-05, "loss": 0.5753, "step": 491 }, { "epoch": 0.5, "learning_rate": 1.9999446401091283e-05, "loss": 0.5665, "step": 492 }, { "epoch": 0.5, "learning_rate": 1.9999423094416882e-05, "loss": 0.5801, "step": 493 }, { "epoch": 0.5, "learning_rate": 1.9999399307211167e-05, "loss": 0.4612, "step": 494 }, { "epoch": 0.5, "learning_rate": 1.9999375039475278e-05, "loss": 0.5157, "step": 495 }, { "epoch": 0.5, "learning_rate": 1.999935029121038e-05, "loss": 0.552, "step": 496 }, { "epoch": 0.5, "learning_rate": 1.9999325062417663e-05, "loss": 0.5412, "step": 497 }, { "epoch": 0.51, "learning_rate": 1.9999299353098343e-05, "loss": 0.5486, "step": 498 }, { "epoch": 0.51, "learning_rate": 1.999927316325365e-05, "loss": 0.5385, "step": 499 }, { "epoch": 0.51, "learning_rate": 1.999924649288485e-05, "loss": 0.5818, "step": 500 }, { "epoch": 0.51, "learning_rate": 1.9999219341993217e-05, "loss": 0.5107, "step": 501 }, { "epoch": 0.51, "learning_rate": 1.9999191710580058e-05, "loss": 0.5666, "step": 502 }, { "epoch": 0.51, "learning_rate": 1.9999163598646704e-05, "loss": 0.5532, "step": 503 }, { "epoch": 0.51, "learning_rate": 1.9999135006194503e-05, "loss": 0.5667, "step": 504 }, { "epoch": 0.51, "learning_rate": 1.9999105933224828e-05, "loss": 0.5418, "step": 505 }, { "epoch": 0.51, "learning_rate": 1.999907637973908e-05, "loss": 0.6281, "step": 506 }, { "epoch": 0.51, "learning_rate": 1.9999046345738678e-05, "loss": 0.5673, "step": 507 }, { "epoch": 0.52, "learning_rate": 1.9999015831225063e-05, "loss": 0.6278, "step": 508 }, { "epoch": 0.52, "learning_rate": 1.9998984836199703e-05, "loss": 0.609, "step": 509 }, { "epoch": 0.52, "learning_rate": 1.9998953360664085e-05, "loss": 0.5981, "step": 510 }, { "epoch": 0.52, "learning_rate": 1.9998921404619728e-05, "loss": 0.4766, "step": 511 }, { "epoch": 0.52, "learning_rate": 1.9998888968068158e-05, "loss": 0.5267, "step": 512 }, { "epoch": 0.52, "learning_rate": 1.9998856051010945e-05, "loss": 0.5765, "step": 513 }, { "epoch": 0.52, "learning_rate": 1.9998822653449665e-05, "loss": 0.5419, "step": 514 }, { "epoch": 0.52, "learning_rate": 1.999878877538592e-05, "loss": 0.5351, "step": 515 }, { "epoch": 0.52, "learning_rate": 1.9998754416821345e-05, "loss": 0.5858, "step": 516 }, { "epoch": 0.52, "learning_rate": 1.9998719577757583e-05, "loss": 0.5502, "step": 517 }, { "epoch": 0.53, "learning_rate": 1.9998684258196315e-05, "loss": 0.5083, "step": 518 }, { "epoch": 0.53, "learning_rate": 1.9998648458139233e-05, "loss": 0.6201, "step": 519 }, { "epoch": 0.53, "learning_rate": 1.9998612177588065e-05, "loss": 0.4752, "step": 520 }, { "epoch": 0.53, "learning_rate": 1.9998575416544546e-05, "loss": 0.4888, "step": 521 }, { "epoch": 0.53, "learning_rate": 1.9998538175010447e-05, "loss": 0.5964, "step": 522 }, { "epoch": 0.53, "learning_rate": 1.9998500452987557e-05, "loss": 0.5689, "step": 523 }, { "epoch": 0.53, "learning_rate": 1.999846225047769e-05, "loss": 0.5423, "step": 524 }, { "epoch": 0.53, "learning_rate": 1.999842356748268e-05, "loss": 0.5328, "step": 525 }, { "epoch": 0.53, "learning_rate": 1.9998384404004386e-05, "loss": 0.6094, "step": 526 }, { "epoch": 0.53, "learning_rate": 1.999834476004469e-05, "loss": 0.5573, "step": 527 }, { "epoch": 0.54, "learning_rate": 1.99983046356055e-05, "loss": 0.5388, "step": 528 }, { "epoch": 0.54, "learning_rate": 1.9998264030688742e-05, "loss": 0.559, "step": 529 }, { "epoch": 0.54, "learning_rate": 1.9998222945296363e-05, "loss": 0.4489, "step": 530 }, { "epoch": 0.54, "learning_rate": 1.999818137943035e-05, "loss": 0.5836, "step": 531 }, { "epoch": 0.54, "learning_rate": 1.9998139333092682e-05, "loss": 0.5569, "step": 532 }, { "epoch": 0.54, "learning_rate": 1.9998096806285397e-05, "loss": 0.5978, "step": 533 }, { "epoch": 0.54, "learning_rate": 1.999805379901053e-05, "loss": 0.5164, "step": 534 }, { "epoch": 0.54, "learning_rate": 1.9998010311270146e-05, "loss": 0.5198, "step": 535 }, { "epoch": 0.54, "learning_rate": 1.999796634306634e-05, "loss": 0.5705, "step": 536 }, { "epoch": 0.54, "learning_rate": 1.9997921894401223e-05, "loss": 0.5427, "step": 537 }, { "epoch": 0.55, "learning_rate": 1.999787696527693e-05, "loss": 0.5489, "step": 538 }, { "epoch": 0.55, "learning_rate": 1.999783155569562e-05, "loss": 0.5606, "step": 539 }, { "epoch": 0.55, "learning_rate": 1.9997785665659482e-05, "loss": 0.5945, "step": 540 }, { "epoch": 0.55, "learning_rate": 1.999773929517071e-05, "loss": 0.5831, "step": 541 }, { "epoch": 0.55, "learning_rate": 1.999769244423154e-05, "loss": 0.4796, "step": 542 }, { "epoch": 0.55, "learning_rate": 1.9997645112844222e-05, "loss": 0.5667, "step": 543 }, { "epoch": 0.55, "learning_rate": 1.999759730101103e-05, "loss": 0.5883, "step": 544 }, { "epoch": 0.55, "learning_rate": 1.999754900873426e-05, "loss": 0.5767, "step": 545 }, { "epoch": 0.55, "learning_rate": 1.9997500236016233e-05, "loss": 0.6138, "step": 546 }, { "epoch": 0.56, "learning_rate": 1.9997450982859298e-05, "loss": 0.5099, "step": 547 }, { "epoch": 0.56, "learning_rate": 1.9997401249265813e-05, "loss": 0.5515, "step": 548 }, { "epoch": 0.56, "learning_rate": 1.999735103523818e-05, "loss": 0.4892, "step": 549 }, { "epoch": 0.56, "learning_rate": 1.99973003407788e-05, "loss": 0.5873, "step": 550 }, { "epoch": 0.56, "learning_rate": 1.9997249165890115e-05, "loss": 0.5165, "step": 551 }, { "epoch": 0.56, "learning_rate": 1.999719751057458e-05, "loss": 0.531, "step": 552 }, { "epoch": 0.56, "learning_rate": 1.9997145374834687e-05, "loss": 0.65, "step": 553 }, { "epoch": 0.56, "learning_rate": 1.9997092758672934e-05, "loss": 0.5533, "step": 554 }, { "epoch": 0.56, "learning_rate": 1.999703966209185e-05, "loss": 0.5425, "step": 555 }, { "epoch": 0.56, "learning_rate": 1.9996986085093987e-05, "loss": 0.5788, "step": 556 }, { "epoch": 0.57, "learning_rate": 1.999693202768192e-05, "loss": 0.5777, "step": 557 }, { "epoch": 0.57, "learning_rate": 1.9996877489858244e-05, "loss": 0.6183, "step": 558 }, { "epoch": 0.57, "learning_rate": 1.9996822471625585e-05, "loss": 0.5466, "step": 559 }, { "epoch": 0.57, "learning_rate": 1.999676697298659e-05, "loss": 0.6205, "step": 560 }, { "epoch": 0.57, "learning_rate": 1.9996710993943912e-05, "loss": 0.532, "step": 561 }, { "epoch": 0.57, "learning_rate": 1.9996654534500255e-05, "loss": 0.587, "step": 562 }, { "epoch": 0.57, "learning_rate": 1.9996597594658328e-05, "loss": 0.6216, "step": 563 }, { "epoch": 0.57, "learning_rate": 1.9996540174420862e-05, "loss": 0.6049, "step": 564 }, { "epoch": 0.57, "learning_rate": 1.9996482273790625e-05, "loss": 0.5218, "step": 565 }, { "epoch": 0.57, "learning_rate": 1.9996423892770392e-05, "loss": 0.5, "step": 566 }, { "epoch": 0.58, "learning_rate": 1.9996365031362975e-05, "loss": 0.5601, "step": 567 }, { "epoch": 0.58, "learning_rate": 1.9996305689571197e-05, "loss": 0.5079, "step": 568 }, { "epoch": 0.58, "learning_rate": 1.9996245867397913e-05, "loss": 0.5922, "step": 569 }, { "epoch": 0.58, "learning_rate": 1.9996185564845997e-05, "loss": 0.6159, "step": 570 }, { "epoch": 0.58, "learning_rate": 1.9996124781918345e-05, "loss": 0.6238, "step": 571 }, { "epoch": 0.58, "learning_rate": 1.9996063518617882e-05, "loss": 0.5487, "step": 572 }, { "epoch": 0.58, "learning_rate": 1.9996001774947546e-05, "loss": 0.5739, "step": 573 }, { "epoch": 0.58, "learning_rate": 1.9995939550910313e-05, "loss": 0.5664, "step": 574 }, { "epoch": 0.58, "learning_rate": 1.9995876846509166e-05, "loss": 0.5498, "step": 575 }, { "epoch": 0.58, "learning_rate": 1.999581366174712e-05, "loss": 0.5522, "step": 576 }, { "epoch": 0.59, "learning_rate": 1.9995749996627216e-05, "loss": 0.5134, "step": 577 }, { "epoch": 0.59, "learning_rate": 1.9995685851152503e-05, "loss": 0.5282, "step": 578 }, { "epoch": 0.59, "learning_rate": 1.9995621225326074e-05, "loss": 0.5149, "step": 579 }, { "epoch": 0.59, "learning_rate": 1.9995556119151032e-05, "loss": 0.5169, "step": 580 }, { "epoch": 0.59, "learning_rate": 1.99954905326305e-05, "loss": 0.5362, "step": 581 }, { "epoch": 0.59, "learning_rate": 1.9995424465767636e-05, "loss": 0.5738, "step": 582 }, { "epoch": 0.59, "learning_rate": 1.9995357918565616e-05, "loss": 0.5397, "step": 583 }, { "epoch": 0.59, "learning_rate": 1.999529089102763e-05, "loss": 0.5168, "step": 584 }, { "epoch": 0.59, "learning_rate": 1.999522338315691e-05, "loss": 0.5319, "step": 585 }, { "epoch": 0.59, "learning_rate": 1.999515539495669e-05, "loss": 0.5118, "step": 586 }, { "epoch": 0.6, "learning_rate": 1.9995086926430244e-05, "loss": 0.5642, "step": 587 }, { "epoch": 0.6, "learning_rate": 1.9995017977580857e-05, "loss": 0.5695, "step": 588 }, { "epoch": 0.6, "learning_rate": 1.9994948548411846e-05, "loss": 0.4722, "step": 589 }, { "epoch": 0.6, "learning_rate": 1.9994878638926548e-05, "loss": 0.4927, "step": 590 }, { "epoch": 0.6, "learning_rate": 1.999480824912832e-05, "loss": 0.4985, "step": 591 }, { "epoch": 0.6, "learning_rate": 1.9994737379020547e-05, "loss": 0.6063, "step": 592 }, { "epoch": 0.6, "learning_rate": 1.9994666028606638e-05, "loss": 0.5274, "step": 593 }, { "epoch": 0.6, "learning_rate": 1.999459419789001e-05, "loss": 0.5701, "step": 594 }, { "epoch": 0.6, "learning_rate": 1.999452188687413e-05, "loss": 0.5512, "step": 595 }, { "epoch": 0.6, "learning_rate": 1.9994449095562457e-05, "loss": 0.6143, "step": 596 }, { "epoch": 0.61, "learning_rate": 1.9994375823958504e-05, "loss": 0.6248, "step": 597 }, { "epoch": 0.61, "learning_rate": 1.999430207206578e-05, "loss": 0.5436, "step": 598 }, { "epoch": 0.61, "learning_rate": 1.9994227839887838e-05, "loss": 0.5896, "step": 599 }, { "epoch": 0.61, "learning_rate": 1.999415312742824e-05, "loss": 0.5129, "step": 600 }, { "epoch": 0.61, "learning_rate": 1.9994077934690584e-05, "loss": 0.5533, "step": 601 }, { "epoch": 0.61, "learning_rate": 1.9994002261678473e-05, "loss": 0.5539, "step": 602 }, { "epoch": 0.61, "learning_rate": 1.9993926108395548e-05, "loss": 0.5917, "step": 603 }, { "epoch": 0.61, "learning_rate": 1.9993849474845473e-05, "loss": 0.4805, "step": 604 }, { "epoch": 0.61, "learning_rate": 1.9993772361031922e-05, "loss": 0.5549, "step": 605 }, { "epoch": 0.61, "learning_rate": 1.9993694766958608e-05, "loss": 0.5321, "step": 606 }, { "epoch": 0.62, "learning_rate": 1.9993616692629258e-05, "loss": 0.5121, "step": 607 }, { "epoch": 0.62, "learning_rate": 1.9993538138047622e-05, "loss": 0.5305, "step": 608 }, { "epoch": 0.62, "learning_rate": 1.999345910321748e-05, "loss": 0.4669, "step": 609 }, { "epoch": 0.62, "learning_rate": 1.999337958814262e-05, "loss": 0.4991, "step": 610 }, { "epoch": 0.62, "learning_rate": 1.9993299592826875e-05, "loss": 0.6743, "step": 611 }, { "epoch": 0.62, "learning_rate": 1.9993219117274082e-05, "loss": 0.552, "step": 612 }, { "epoch": 0.62, "learning_rate": 1.999313816148811e-05, "loss": 0.5349, "step": 613 }, { "epoch": 0.62, "learning_rate": 1.9993056725472846e-05, "loss": 0.4783, "step": 614 }, { "epoch": 0.62, "learning_rate": 1.9992974809232213e-05, "loss": 0.5008, "step": 615 }, { "epoch": 0.63, "learning_rate": 1.9992892412770137e-05, "loss": 0.5855, "step": 616 }, { "epoch": 0.63, "learning_rate": 1.9992809536090585e-05, "loss": 0.5015, "step": 617 }, { "epoch": 0.63, "learning_rate": 1.999272617919754e-05, "loss": 0.5635, "step": 618 }, { "epoch": 0.63, "learning_rate": 1.9992642342094995e-05, "loss": 0.5904, "step": 619 }, { "epoch": 0.63, "learning_rate": 1.9992558024786997e-05, "loss": 0.5197, "step": 620 }, { "epoch": 0.63, "learning_rate": 1.9992473227277586e-05, "loss": 0.4886, "step": 621 }, { "epoch": 0.63, "learning_rate": 1.9992387949570842e-05, "loss": 0.4825, "step": 622 }, { "epoch": 0.63, "learning_rate": 1.999230219167086e-05, "loss": 0.4157, "step": 623 }, { "epoch": 0.63, "learning_rate": 1.9992215953581766e-05, "loss": 0.4964, "step": 624 }, { "epoch": 0.63, "learning_rate": 1.9992129235307698e-05, "loss": 0.5453, "step": 625 }, { "epoch": 0.64, "learning_rate": 1.9992042036852827e-05, "loss": 0.5233, "step": 626 }, { "epoch": 0.64, "learning_rate": 1.9991954358221345e-05, "loss": 0.5648, "step": 627 }, { "epoch": 0.64, "learning_rate": 1.999186619941746e-05, "loss": 0.4499, "step": 628 }, { "epoch": 0.64, "learning_rate": 1.9991777560445412e-05, "loss": 0.5158, "step": 629 }, { "epoch": 0.64, "learning_rate": 1.9991688441309462e-05, "loss": 0.6237, "step": 630 }, { "epoch": 0.64, "learning_rate": 1.9991598842013892e-05, "loss": 0.5123, "step": 631 }, { "epoch": 0.64, "learning_rate": 1.9991508762563005e-05, "loss": 0.5375, "step": 632 }, { "epoch": 0.64, "learning_rate": 1.999141820296113e-05, "loss": 0.5422, "step": 633 }, { "epoch": 0.64, "learning_rate": 1.9991327163212627e-05, "loss": 0.5056, "step": 634 }, { "epoch": 0.64, "learning_rate": 1.999123564332186e-05, "loss": 0.5898, "step": 635 }, { "epoch": 0.65, "learning_rate": 1.9991143643293233e-05, "loss": 0.5648, "step": 636 }, { "epoch": 0.65, "learning_rate": 1.9991051163131165e-05, "loss": 0.5487, "step": 637 }, { "epoch": 0.65, "learning_rate": 1.9990958202840102e-05, "loss": 0.5499, "step": 638 }, { "epoch": 0.65, "learning_rate": 1.999086476242451e-05, "loss": 0.501, "step": 639 }, { "epoch": 0.65, "learning_rate": 1.9990770841888882e-05, "loss": 0.5634, "step": 640 }, { "epoch": 0.65, "learning_rate": 1.9990676441237727e-05, "loss": 0.4858, "step": 641 }, { "epoch": 0.65, "learning_rate": 1.9990581560475585e-05, "loss": 0.5449, "step": 642 }, { "epoch": 0.65, "learning_rate": 1.9990486199607013e-05, "loss": 0.5558, "step": 643 }, { "epoch": 0.65, "learning_rate": 1.9990390358636596e-05, "loss": 0.4579, "step": 644 }, { "epoch": 0.65, "learning_rate": 1.9990294037568934e-05, "loss": 0.5503, "step": 645 }, { "epoch": 0.66, "learning_rate": 1.9990197236408666e-05, "loss": 0.5247, "step": 646 }, { "epoch": 0.66, "learning_rate": 1.9990099955160434e-05, "loss": 0.544, "step": 647 }, { "epoch": 0.66, "learning_rate": 1.9990002193828923e-05, "loss": 0.525, "step": 648 }, { "epoch": 0.66, "learning_rate": 1.9989903952418823e-05, "loss": 0.5358, "step": 649 }, { "epoch": 0.66, "learning_rate": 1.9989805230934856e-05, "loss": 0.5009, "step": 650 }, { "epoch": 0.66, "learning_rate": 1.998970602938177e-05, "loss": 0.6341, "step": 651 }, { "epoch": 0.66, "learning_rate": 1.9989606347764327e-05, "loss": 0.4803, "step": 652 }, { "epoch": 0.66, "learning_rate": 1.9989506186087323e-05, "loss": 0.4911, "step": 653 }, { "epoch": 0.66, "learning_rate": 1.9989405544355565e-05, "loss": 0.5773, "step": 654 }, { "epoch": 0.66, "learning_rate": 1.9989304422573895e-05, "loss": 0.5387, "step": 655 }, { "epoch": 0.67, "learning_rate": 1.9989202820747167e-05, "loss": 0.5855, "step": 656 }, { "epoch": 0.67, "learning_rate": 1.9989100738880272e-05, "loss": 0.4894, "step": 657 }, { "epoch": 0.67, "learning_rate": 1.9988998176978108e-05, "loss": 0.5961, "step": 658 }, { "epoch": 0.67, "learning_rate": 1.9988895135045604e-05, "loss": 0.506, "step": 659 }, { "epoch": 0.67, "learning_rate": 1.998879161308772e-05, "loss": 0.5509, "step": 660 }, { "epoch": 0.67, "learning_rate": 1.9988687611109418e-05, "loss": 0.5158, "step": 661 }, { "epoch": 0.67, "learning_rate": 1.998858312911571e-05, "loss": 0.5009, "step": 662 }, { "epoch": 0.67, "learning_rate": 1.99884781671116e-05, "loss": 0.483, "step": 663 }, { "epoch": 0.67, "learning_rate": 1.998837272510215e-05, "loss": 0.5319, "step": 664 }, { "epoch": 0.67, "learning_rate": 1.998826680309242e-05, "loss": 0.5363, "step": 665 }, { "epoch": 0.68, "learning_rate": 1.9988160401087492e-05, "loss": 0.5196, "step": 666 }, { "epoch": 0.68, "learning_rate": 1.998805351909249e-05, "loss": 0.5677, "step": 667 }, { "epoch": 0.68, "learning_rate": 1.9987946157112548e-05, "loss": 0.488, "step": 668 }, { "epoch": 0.68, "learning_rate": 1.998783831515282e-05, "loss": 0.4956, "step": 669 }, { "epoch": 0.68, "learning_rate": 1.9987729993218495e-05, "loss": 0.5384, "step": 670 }, { "epoch": 0.68, "learning_rate": 1.998762119131478e-05, "loss": 0.5185, "step": 671 }, { "epoch": 0.68, "learning_rate": 1.9987511909446893e-05, "loss": 0.5287, "step": 672 }, { "epoch": 0.68, "learning_rate": 1.9987402147620096e-05, "loss": 0.4564, "step": 673 }, { "epoch": 0.68, "learning_rate": 1.9987291905839656e-05, "loss": 0.628, "step": 674 }, { "epoch": 0.68, "learning_rate": 1.9987181184110877e-05, "loss": 0.607, "step": 675 }, { "epoch": 0.69, "learning_rate": 1.998706998243908e-05, "loss": 0.5418, "step": 676 }, { "epoch": 0.69, "learning_rate": 1.9986958300829605e-05, "loss": 0.5361, "step": 677 }, { "epoch": 0.69, "learning_rate": 1.9986846139287818e-05, "loss": 0.5194, "step": 678 }, { "epoch": 0.69, "learning_rate": 1.998673349781911e-05, "loss": 0.483, "step": 679 }, { "epoch": 0.69, "learning_rate": 1.9986620376428896e-05, "loss": 0.4545, "step": 680 }, { "epoch": 0.69, "learning_rate": 1.9986506775122617e-05, "loss": 0.59, "step": 681 }, { "epoch": 0.69, "learning_rate": 1.9986392693905725e-05, "loss": 0.5269, "step": 682 }, { "epoch": 0.69, "learning_rate": 1.9986278132783702e-05, "loss": 0.4578, "step": 683 }, { "epoch": 0.69, "learning_rate": 1.9986163091762055e-05, "loss": 0.5486, "step": 684 }, { "epoch": 0.7, "learning_rate": 1.9986047570846315e-05, "loss": 0.5384, "step": 685 }, { "epoch": 0.7, "learning_rate": 1.9985931570042026e-05, "loss": 0.4681, "step": 686 }, { "epoch": 0.7, "learning_rate": 1.9985815089354775e-05, "loss": 0.5299, "step": 687 }, { "epoch": 0.7, "learning_rate": 1.998569812879015e-05, "loss": 0.5071, "step": 688 }, { "epoch": 0.7, "learning_rate": 1.9985580688353772e-05, "loss": 0.5354, "step": 689 }, { "epoch": 0.7, "learning_rate": 1.998546276805129e-05, "loss": 0.5224, "step": 690 }, { "epoch": 0.7, "learning_rate": 1.9985344367888364e-05, "loss": 0.5141, "step": 691 }, { "epoch": 0.7, "learning_rate": 1.9985225487870688e-05, "loss": 0.5351, "step": 692 }, { "epoch": 0.7, "learning_rate": 1.9985106128003977e-05, "loss": 0.4773, "step": 693 }, { "epoch": 0.7, "learning_rate": 1.9984986288293963e-05, "loss": 0.5189, "step": 694 }, { "epoch": 0.71, "learning_rate": 1.9984865968746405e-05, "loss": 0.5362, "step": 695 }, { "epoch": 0.71, "learning_rate": 1.9984745169367086e-05, "loss": 0.5527, "step": 696 }, { "epoch": 0.71, "learning_rate": 1.9984623890161813e-05, "loss": 0.4921, "step": 697 }, { "epoch": 0.71, "learning_rate": 1.998450213113641e-05, "loss": 0.534, "step": 698 }, { "epoch": 0.71, "learning_rate": 1.998437989229673e-05, "loss": 0.5174, "step": 699 }, { "epoch": 0.71, "learning_rate": 1.9984257173648653e-05, "loss": 0.5227, "step": 700 }, { "epoch": 0.71, "learning_rate": 1.998413397519807e-05, "loss": 0.5254, "step": 701 }, { "epoch": 0.71, "learning_rate": 1.99840102969509e-05, "loss": 0.4995, "step": 702 }, { "epoch": 0.71, "learning_rate": 1.9983886138913094e-05, "loss": 0.5255, "step": 703 }, { "epoch": 0.71, "learning_rate": 1.9983761501090607e-05, "loss": 0.507, "step": 704 }, { "epoch": 0.72, "learning_rate": 1.9983636383489442e-05, "loss": 0.4737, "step": 705 }, { "epoch": 0.72, "learning_rate": 1.99835107861156e-05, "loss": 0.4467, "step": 706 }, { "epoch": 0.72, "learning_rate": 1.9983384708975127e-05, "loss": 0.5136, "step": 707 }, { "epoch": 0.72, "learning_rate": 1.9983258152074073e-05, "loss": 0.51, "step": 708 }, { "epoch": 0.72, "learning_rate": 1.9983131115418522e-05, "loss": 0.5536, "step": 709 }, { "epoch": 0.72, "learning_rate": 1.9983003599014585e-05, "loss": 0.487, "step": 710 }, { "epoch": 0.72, "learning_rate": 1.9982875602868385e-05, "loss": 0.4809, "step": 711 }, { "epoch": 0.72, "learning_rate": 1.9982747126986068e-05, "loss": 0.5245, "step": 712 }, { "epoch": 0.72, "learning_rate": 1.9982618171373817e-05, "loss": 0.4862, "step": 713 }, { "epoch": 0.72, "learning_rate": 1.9982488736037822e-05, "loss": 0.5478, "step": 714 }, { "epoch": 0.73, "learning_rate": 1.9982358820984308e-05, "loss": 0.5382, "step": 715 }, { "epoch": 0.73, "learning_rate": 1.9982228426219518e-05, "loss": 0.4829, "step": 716 }, { "epoch": 0.73, "learning_rate": 1.9982097551749715e-05, "loss": 0.4513, "step": 717 }, { "epoch": 0.73, "learning_rate": 1.998196619758119e-05, "loss": 0.5882, "step": 718 }, { "epoch": 0.73, "learning_rate": 1.9981834363720253e-05, "loss": 0.5692, "step": 719 }, { "epoch": 0.73, "learning_rate": 1.9981702050173245e-05, "loss": 0.5053, "step": 720 }, { "epoch": 0.73, "learning_rate": 1.9981569256946523e-05, "loss": 0.5367, "step": 721 }, { "epoch": 0.73, "learning_rate": 1.9981435984046463e-05, "loss": 0.5124, "step": 722 }, { "epoch": 0.73, "learning_rate": 1.9981302231479475e-05, "loss": 0.5718, "step": 723 }, { "epoch": 0.73, "learning_rate": 1.9981167999251985e-05, "loss": 0.4051, "step": 724 }, { "epoch": 0.74, "learning_rate": 1.9981033287370443e-05, "loss": 0.542, "step": 725 }, { "epoch": 0.74, "learning_rate": 1.9980898095841322e-05, "loss": 0.5054, "step": 726 }, { "epoch": 0.74, "learning_rate": 1.998076242467112e-05, "loss": 0.5416, "step": 727 }, { "epoch": 0.74, "learning_rate": 1.998062627386636e-05, "loss": 0.516, "step": 728 }, { "epoch": 0.74, "learning_rate": 1.9980489643433582e-05, "loss": 0.4591, "step": 729 }, { "epoch": 0.74, "learning_rate": 1.9980352533379345e-05, "loss": 0.43, "step": 730 }, { "epoch": 0.74, "learning_rate": 1.998021494371025e-05, "loss": 0.5411, "step": 731 }, { "epoch": 0.74, "learning_rate": 1.9980076874432904e-05, "loss": 0.5312, "step": 732 }, { "epoch": 0.74, "learning_rate": 1.9979938325553937e-05, "loss": 0.5545, "step": 733 }, { "epoch": 0.74, "learning_rate": 1.9979799297080013e-05, "loss": 0.5433, "step": 734 }, { "epoch": 0.75, "learning_rate": 1.9979659789017813e-05, "loss": 0.5076, "step": 735 }, { "epoch": 0.75, "learning_rate": 1.9979519801374042e-05, "loss": 0.4909, "step": 736 }, { "epoch": 0.75, "learning_rate": 1.9979379334155422e-05, "loss": 0.4958, "step": 737 }, { "epoch": 0.75, "learning_rate": 1.9979238387368706e-05, "loss": 0.6346, "step": 738 }, { "epoch": 0.75, "learning_rate": 1.997909696102067e-05, "loss": 0.492, "step": 739 }, { "epoch": 0.75, "learning_rate": 1.9978955055118104e-05, "loss": 0.6304, "step": 740 }, { "epoch": 0.75, "learning_rate": 1.9978812669667838e-05, "loss": 0.5094, "step": 741 }, { "epoch": 0.75, "learning_rate": 1.9978669804676703e-05, "loss": 0.4586, "step": 742 }, { "epoch": 0.75, "learning_rate": 1.9978526460151573e-05, "loss": 0.5272, "step": 743 }, { "epoch": 0.75, "learning_rate": 1.997838263609933e-05, "loss": 0.472, "step": 744 }, { "epoch": 0.76, "learning_rate": 1.9978238332526892e-05, "loss": 0.5407, "step": 745 }, { "epoch": 0.76, "learning_rate": 1.9978093549441188e-05, "loss": 0.5759, "step": 746 }, { "epoch": 0.76, "learning_rate": 1.9977948286849177e-05, "loss": 0.5154, "step": 747 }, { "epoch": 0.76, "learning_rate": 1.997780254475784e-05, "loss": 0.5266, "step": 748 }, { "epoch": 0.76, "learning_rate": 1.9977656323174182e-05, "loss": 0.5115, "step": 749 }, { "epoch": 0.76, "learning_rate": 1.9977509622105233e-05, "loss": 0.4509, "step": 750 }, { "epoch": 0.76, "learning_rate": 1.997736244155804e-05, "loss": 0.5048, "step": 751 }, { "epoch": 0.76, "learning_rate": 1.997721478153967e-05, "loss": 0.4585, "step": 752 }, { "epoch": 0.76, "learning_rate": 1.9977066642057224e-05, "loss": 0.4864, "step": 753 }, { "epoch": 0.77, "learning_rate": 1.9976918023117824e-05, "loss": 0.4666, "step": 754 }, { "epoch": 0.77, "learning_rate": 1.997676892472861e-05, "loss": 0.4729, "step": 755 }, { "epoch": 0.77, "learning_rate": 1.9976619346896743e-05, "loss": 0.554, "step": 756 }, { "epoch": 0.77, "learning_rate": 1.9976469289629414e-05, "loss": 0.526, "step": 757 }, { "epoch": 0.77, "learning_rate": 1.9976318752933834e-05, "loss": 0.5425, "step": 758 }, { "epoch": 0.77, "learning_rate": 1.9976167736817237e-05, "loss": 0.5436, "step": 759 }, { "epoch": 0.77, "learning_rate": 1.9976016241286883e-05, "loss": 0.5188, "step": 760 }, { "epoch": 0.77, "learning_rate": 1.997586426635005e-05, "loss": 0.5281, "step": 761 }, { "epoch": 0.77, "learning_rate": 1.997571181201404e-05, "loss": 0.4989, "step": 762 }, { "epoch": 0.77, "learning_rate": 1.997555887828618e-05, "loss": 0.4788, "step": 763 }, { "epoch": 0.78, "learning_rate": 1.997540546517382e-05, "loss": 0.4647, "step": 764 }, { "epoch": 0.78, "learning_rate": 1.9975251572684333e-05, "loss": 0.5028, "step": 765 }, { "epoch": 0.78, "learning_rate": 1.9975097200825115e-05, "loss": 0.4682, "step": 766 }, { "epoch": 0.78, "learning_rate": 1.997494234960358e-05, "loss": 0.5572, "step": 767 }, { "epoch": 0.78, "learning_rate": 1.9974787019027174e-05, "loss": 0.5199, "step": 768 }, { "epoch": 0.78, "learning_rate": 1.9974631209103363e-05, "loss": 0.5612, "step": 769 }, { "epoch": 0.78, "learning_rate": 1.997447491983963e-05, "loss": 0.4855, "step": 770 }, { "epoch": 0.78, "learning_rate": 1.9974318151243484e-05, "loss": 0.523, "step": 771 }, { "epoch": 0.78, "learning_rate": 1.9974160903322467e-05, "loss": 0.4794, "step": 772 }, { "epoch": 0.78, "learning_rate": 1.997400317608413e-05, "loss": 0.4688, "step": 773 }, { "epoch": 0.79, "learning_rate": 1.9973844969536055e-05, "loss": 0.583, "step": 774 }, { "epoch": 0.79, "learning_rate": 1.997368628368584e-05, "loss": 0.5174, "step": 775 }, { "epoch": 0.79, "learning_rate": 1.9973527118541114e-05, "loss": 0.4927, "step": 776 }, { "epoch": 0.79, "learning_rate": 1.9973367474109525e-05, "loss": 0.5127, "step": 777 }, { "epoch": 0.79, "learning_rate": 1.997320735039875e-05, "loss": 0.5089, "step": 778 }, { "epoch": 0.79, "learning_rate": 1.9973046747416478e-05, "loss": 0.4937, "step": 779 }, { "epoch": 0.79, "learning_rate": 1.997288566517043e-05, "loss": 0.5556, "step": 780 }, { "epoch": 0.79, "learning_rate": 1.997272410366834e-05, "loss": 0.4468, "step": 781 }, { "epoch": 0.79, "learning_rate": 1.9972562062917982e-05, "loss": 0.5562, "step": 782 }, { "epoch": 0.79, "learning_rate": 1.997239954292714e-05, "loss": 0.5441, "step": 783 }, { "epoch": 0.8, "learning_rate": 1.997223654370362e-05, "loss": 0.5118, "step": 784 }, { "epoch": 0.8, "learning_rate": 1.9972073065255257e-05, "loss": 0.5458, "step": 785 }, { "epoch": 0.8, "learning_rate": 1.9971909107589907e-05, "loss": 0.5169, "step": 786 }, { "epoch": 0.8, "learning_rate": 1.9971744670715454e-05, "loss": 0.4845, "step": 787 }, { "epoch": 0.8, "learning_rate": 1.9971579754639792e-05, "loss": 0.4913, "step": 788 }, { "epoch": 0.8, "learning_rate": 1.9971414359370853e-05, "loss": 0.4701, "step": 789 }, { "epoch": 0.8, "learning_rate": 1.9971248484916585e-05, "loss": 0.437, "step": 790 }, { "epoch": 0.8, "learning_rate": 1.997108213128495e-05, "loss": 0.4839, "step": 791 }, { "epoch": 0.8, "learning_rate": 1.997091529848395e-05, "loss": 0.4461, "step": 792 }, { "epoch": 0.8, "learning_rate": 1.9970747986521605e-05, "loss": 0.5608, "step": 793 }, { "epoch": 0.81, "learning_rate": 1.997058019540595e-05, "loss": 0.5217, "step": 794 }, { "epoch": 0.81, "learning_rate": 1.9970411925145053e-05, "loss": 0.4701, "step": 795 }, { "epoch": 0.81, "learning_rate": 1.9970243175746992e-05, "loss": 0.6007, "step": 796 }, { "epoch": 0.81, "learning_rate": 1.9970073947219886e-05, "loss": 0.4601, "step": 797 }, { "epoch": 0.81, "learning_rate": 1.9969904239571862e-05, "loss": 0.5945, "step": 798 }, { "epoch": 0.81, "learning_rate": 1.9969734052811077e-05, "loss": 0.4407, "step": 799 }, { "epoch": 0.81, "learning_rate": 1.9969563386945706e-05, "loss": 0.433, "step": 800 }, { "epoch": 0.81, "learning_rate": 1.9969392241983957e-05, "loss": 0.5534, "step": 801 }, { "epoch": 0.81, "learning_rate": 1.9969220617934052e-05, "loss": 0.5266, "step": 802 }, { "epoch": 0.81, "learning_rate": 1.9969048514804236e-05, "loss": 0.4724, "step": 803 }, { "epoch": 0.82, "learning_rate": 1.996887593260278e-05, "loss": 0.5318, "step": 804 }, { "epoch": 0.82, "learning_rate": 1.996870287133798e-05, "loss": 0.506, "step": 805 }, { "epoch": 0.82, "learning_rate": 1.9968529331018155e-05, "loss": 0.5225, "step": 806 }, { "epoch": 0.82, "learning_rate": 1.9968355311651638e-05, "loss": 0.4472, "step": 807 }, { "epoch": 0.82, "learning_rate": 1.9968180813246793e-05, "loss": 0.4544, "step": 808 }, { "epoch": 0.82, "learning_rate": 1.996800583581201e-05, "loss": 0.5466, "step": 809 }, { "epoch": 0.82, "learning_rate": 1.9967830379355694e-05, "loss": 0.4766, "step": 810 }, { "epoch": 0.82, "learning_rate": 1.9967654443886277e-05, "loss": 0.5608, "step": 811 }, { "epoch": 0.82, "learning_rate": 1.9967478029412218e-05, "loss": 0.5465, "step": 812 }, { "epoch": 0.82, "learning_rate": 1.996730113594199e-05, "loss": 0.4541, "step": 813 }, { "epoch": 0.83, "learning_rate": 1.9967123763484094e-05, "loss": 0.5093, "step": 814 }, { "epoch": 0.83, "learning_rate": 1.9966945912047056e-05, "loss": 0.5432, "step": 815 }, { "epoch": 0.83, "learning_rate": 1.996676758163942e-05, "loss": 0.4883, "step": 816 }, { "epoch": 0.83, "learning_rate": 1.9966588772269757e-05, "loss": 0.4599, "step": 817 }, { "epoch": 0.83, "learning_rate": 1.9966409483946664e-05, "loss": 0.4629, "step": 818 }, { "epoch": 0.83, "learning_rate": 1.9966229716678752e-05, "loss": 0.5185, "step": 819 }, { "epoch": 0.83, "learning_rate": 1.996604947047466e-05, "loss": 0.4767, "step": 820 }, { "epoch": 0.83, "learning_rate": 1.996586874534305e-05, "loss": 0.5499, "step": 821 }, { "epoch": 0.83, "learning_rate": 1.9965687541292612e-05, "loss": 0.4909, "step": 822 }, { "epoch": 0.84, "learning_rate": 1.9965505858332046e-05, "loss": 0.5306, "step": 823 }, { "epoch": 0.84, "learning_rate": 1.996532369647009e-05, "loss": 0.5127, "step": 824 }, { "epoch": 0.84, "learning_rate": 1.9965141055715495e-05, "loss": 0.4124, "step": 825 }, { "epoch": 0.84, "learning_rate": 1.996495793607703e-05, "loss": 0.4881, "step": 826 }, { "epoch": 0.84, "learning_rate": 1.9964774337563512e-05, "loss": 0.5496, "step": 827 }, { "epoch": 0.84, "learning_rate": 1.9964590260183752e-05, "loss": 0.5574, "step": 828 }, { "epoch": 0.84, "learning_rate": 1.9964405703946597e-05, "loss": 0.528, "step": 829 }, { "epoch": 0.84, "learning_rate": 1.9964220668860918e-05, "loss": 0.4545, "step": 830 }, { "epoch": 0.84, "learning_rate": 1.9964035154935607e-05, "loss": 0.4795, "step": 831 }, { "epoch": 0.84, "learning_rate": 1.9963849162179583e-05, "loss": 0.4903, "step": 832 }, { "epoch": 0.85, "learning_rate": 1.9963662690601774e-05, "loss": 0.4883, "step": 833 }, { "epoch": 0.85, "learning_rate": 1.996347574021115e-05, "loss": 0.504, "step": 834 }, { "epoch": 0.85, "learning_rate": 1.9963288311016695e-05, "loss": 0.4018, "step": 835 }, { "epoch": 0.85, "learning_rate": 1.9963100403027406e-05, "loss": 0.5156, "step": 836 }, { "epoch": 0.85, "learning_rate": 1.9962912016252324e-05, "loss": 0.458, "step": 837 }, { "epoch": 0.85, "learning_rate": 1.99627231507005e-05, "loss": 0.4034, "step": 838 }, { "epoch": 0.85, "learning_rate": 1.9962533806381005e-05, "loss": 0.5132, "step": 839 }, { "epoch": 0.85, "learning_rate": 1.9962343983302945e-05, "loss": 0.5192, "step": 840 }, { "epoch": 0.85, "learning_rate": 1.9962153681475436e-05, "loss": 0.473, "step": 841 }, { "epoch": 0.85, "learning_rate": 1.9961962900907626e-05, "loss": 0.5225, "step": 842 }, { "epoch": 0.86, "learning_rate": 1.9961771641608684e-05, "loss": 0.5394, "step": 843 }, { "epoch": 0.86, "learning_rate": 1.9961579903587797e-05, "loss": 0.478, "step": 844 }, { "epoch": 0.86, "learning_rate": 1.9961387686854185e-05, "loss": 0.4783, "step": 845 }, { "epoch": 0.86, "learning_rate": 1.9961194991417084e-05, "loss": 0.5173, "step": 846 }, { "epoch": 0.86, "learning_rate": 1.9961001817285747e-05, "loss": 0.4723, "step": 847 }, { "epoch": 0.86, "learning_rate": 1.9960808164469467e-05, "loss": 0.4832, "step": 848 }, { "epoch": 0.86, "learning_rate": 1.9960614032977546e-05, "loss": 0.5555, "step": 849 }, { "epoch": 0.86, "learning_rate": 1.996041942281931e-05, "loss": 0.4772, "step": 850 }, { "epoch": 0.86, "learning_rate": 1.9960224334004117e-05, "loss": 0.4404, "step": 851 }, { "epoch": 0.86, "learning_rate": 1.9960028766541336e-05, "loss": 0.484, "step": 852 }, { "epoch": 0.87, "learning_rate": 1.9959832720440372e-05, "loss": 0.4332, "step": 853 }, { "epoch": 0.87, "learning_rate": 1.9959636195710638e-05, "loss": 0.4364, "step": 854 }, { "epoch": 0.87, "learning_rate": 1.9959439192361586e-05, "loss": 0.4631, "step": 855 }, { "epoch": 0.87, "learning_rate": 1.995924171040268e-05, "loss": 0.5179, "step": 856 }, { "epoch": 0.87, "learning_rate": 1.995904374984341e-05, "loss": 0.4718, "step": 857 }, { "epoch": 0.87, "learning_rate": 1.995884531069329e-05, "loss": 0.4466, "step": 858 }, { "epoch": 0.87, "learning_rate": 1.9958646392961855e-05, "loss": 0.4838, "step": 859 }, { "epoch": 0.87, "learning_rate": 1.9958446996658662e-05, "loss": 0.5229, "step": 860 }, { "epoch": 0.87, "learning_rate": 1.99582471217933e-05, "loss": 0.5125, "step": 861 }, { "epoch": 0.87, "learning_rate": 1.9958046768375367e-05, "loss": 0.4789, "step": 862 }, { "epoch": 0.88, "learning_rate": 1.9957845936414493e-05, "loss": 0.5669, "step": 863 }, { "epoch": 0.88, "learning_rate": 1.9957644625920337e-05, "loss": 0.5056, "step": 864 }, { "epoch": 0.88, "learning_rate": 1.995744283690256e-05, "loss": 0.4846, "step": 865 }, { "epoch": 0.88, "learning_rate": 1.9957240569370867e-05, "loss": 0.527, "step": 866 }, { "epoch": 0.88, "learning_rate": 1.9957037823334978e-05, "loss": 0.5198, "step": 867 }, { "epoch": 0.88, "learning_rate": 1.9956834598804634e-05, "loss": 0.5238, "step": 868 }, { "epoch": 0.88, "learning_rate": 1.9956630895789604e-05, "loss": 0.5194, "step": 869 }, { "epoch": 0.88, "learning_rate": 1.995642671429967e-05, "loss": 0.5228, "step": 870 }, { "epoch": 0.88, "learning_rate": 1.9956222054344655e-05, "loss": 0.503, "step": 871 }, { "epoch": 0.88, "learning_rate": 1.995601691593439e-05, "loss": 0.5302, "step": 872 }, { "epoch": 0.89, "learning_rate": 1.9955811299078727e-05, "loss": 0.5294, "step": 873 }, { "epoch": 0.89, "learning_rate": 1.995560520378755e-05, "loss": 0.5084, "step": 874 }, { "epoch": 0.89, "learning_rate": 1.9955398630070767e-05, "loss": 0.5267, "step": 875 }, { "epoch": 0.89, "learning_rate": 1.9955191577938305e-05, "loss": 0.488, "step": 876 }, { "epoch": 0.89, "learning_rate": 1.995498404740011e-05, "loss": 0.5587, "step": 877 }, { "epoch": 0.89, "learning_rate": 1.9954776038466157e-05, "loss": 0.5065, "step": 878 }, { "epoch": 0.89, "learning_rate": 1.9954567551146443e-05, "loss": 0.5164, "step": 879 }, { "epoch": 0.89, "learning_rate": 1.9954358585450984e-05, "loss": 0.5209, "step": 880 }, { "epoch": 0.89, "learning_rate": 1.9954149141389826e-05, "loss": 0.4862, "step": 881 }, { "epoch": 0.89, "learning_rate": 1.9953939218973032e-05, "loss": 0.4015, "step": 882 }, { "epoch": 0.9, "learning_rate": 1.9953728818210688e-05, "loss": 0.4658, "step": 883 }, { "epoch": 0.9, "learning_rate": 1.995351793911291e-05, "loss": 0.4693, "step": 884 }, { "epoch": 0.9, "learning_rate": 1.9953306581689828e-05, "loss": 0.4414, "step": 885 }, { "epoch": 0.9, "learning_rate": 1.99530947459516e-05, "loss": 0.5079, "step": 886 }, { "epoch": 0.9, "learning_rate": 1.9952882431908406e-05, "loss": 0.4833, "step": 887 }, { "epoch": 0.9, "learning_rate": 1.995266963957045e-05, "loss": 0.4527, "step": 888 }, { "epoch": 0.9, "learning_rate": 1.9952456368947954e-05, "loss": 0.5437, "step": 889 }, { "epoch": 0.9, "learning_rate": 1.9952242620051174e-05, "loss": 0.4586, "step": 890 }, { "epoch": 0.9, "learning_rate": 1.9952028392890376e-05, "loss": 0.5049, "step": 891 }, { "epoch": 0.91, "learning_rate": 1.9951813687475857e-05, "loss": 0.5482, "step": 892 }, { "epoch": 0.91, "learning_rate": 1.9951598503817933e-05, "loss": 0.4725, "step": 893 }, { "epoch": 0.91, "learning_rate": 1.995138284192695e-05, "loss": 0.6136, "step": 894 }, { "epoch": 0.91, "learning_rate": 1.995116670181326e-05, "loss": 0.508, "step": 895 }, { "epoch": 0.91, "learning_rate": 1.9950950083487266e-05, "loss": 0.5107, "step": 896 }, { "epoch": 0.91, "learning_rate": 1.9950732986959368e-05, "loss": 0.5046, "step": 897 }, { "epoch": 0.91, "learning_rate": 1.9950515412239998e-05, "loss": 0.4792, "step": 898 }, { "epoch": 0.91, "learning_rate": 1.9950297359339616e-05, "loss": 0.4964, "step": 899 }, { "epoch": 0.91, "learning_rate": 1.99500788282687e-05, "loss": 0.506, "step": 900 }, { "epoch": 0.91, "learning_rate": 1.9949859819037747e-05, "loss": 0.5973, "step": 901 }, { "epoch": 0.92, "learning_rate": 1.994964033165729e-05, "loss": 0.5066, "step": 902 }, { "epoch": 0.92, "learning_rate": 1.994942036613787e-05, "loss": 0.4375, "step": 903 }, { "epoch": 0.92, "learning_rate": 1.994919992249006e-05, "loss": 0.4855, "step": 904 }, { "epoch": 0.92, "learning_rate": 1.9948979000724455e-05, "loss": 0.4355, "step": 905 }, { "epoch": 0.92, "learning_rate": 1.9948757600851668e-05, "loss": 0.4323, "step": 906 }, { "epoch": 0.92, "learning_rate": 1.994853572288234e-05, "loss": 0.485, "step": 907 }, { "epoch": 0.92, "learning_rate": 1.9948313366827136e-05, "loss": 0.5163, "step": 908 }, { "epoch": 0.92, "learning_rate": 1.994809053269674e-05, "loss": 0.4896, "step": 909 }, { "epoch": 0.92, "learning_rate": 1.9947867220501855e-05, "loss": 0.5091, "step": 910 }, { "epoch": 0.92, "learning_rate": 1.994764343025322e-05, "loss": 0.5495, "step": 911 }, { "epoch": 0.93, "learning_rate": 1.994741916196159e-05, "loss": 0.4655, "step": 912 }, { "epoch": 0.93, "learning_rate": 1.9947194415637737e-05, "loss": 0.4468, "step": 913 }, { "epoch": 0.93, "learning_rate": 1.9946969191292463e-05, "loss": 0.5484, "step": 914 }, { "epoch": 0.93, "learning_rate": 1.9946743488936593e-05, "loss": 0.4529, "step": 915 }, { "epoch": 0.93, "learning_rate": 1.9946517308580977e-05, "loss": 0.4394, "step": 916 }, { "epoch": 0.93, "learning_rate": 1.9946290650236472e-05, "loss": 0.4076, "step": 917 }, { "epoch": 0.93, "learning_rate": 1.9946063513913982e-05, "loss": 0.4854, "step": 918 }, { "epoch": 0.93, "learning_rate": 1.9945835899624417e-05, "loss": 0.4859, "step": 919 }, { "epoch": 0.93, "learning_rate": 1.9945607807378717e-05, "loss": 0.4841, "step": 920 }, { "epoch": 0.93, "learning_rate": 1.9945379237187846e-05, "loss": 0.4431, "step": 921 }, { "epoch": 0.94, "learning_rate": 1.9945150189062782e-05, "loss": 0.4763, "step": 922 }, { "epoch": 0.94, "learning_rate": 1.9944920663014532e-05, "loss": 0.5376, "step": 923 }, { "epoch": 0.94, "learning_rate": 1.9944690659054132e-05, "loss": 0.4518, "step": 924 }, { "epoch": 0.94, "learning_rate": 1.9944460177192634e-05, "loss": 0.4333, "step": 925 }, { "epoch": 0.94, "learning_rate": 1.994422921744111e-05, "loss": 0.5678, "step": 926 }, { "epoch": 0.94, "learning_rate": 1.994399777981066e-05, "loss": 0.4591, "step": 927 }, { "epoch": 0.94, "learning_rate": 1.9943765864312407e-05, "loss": 0.4841, "step": 928 }, { "epoch": 0.94, "learning_rate": 1.9943533470957498e-05, "loss": 0.5406, "step": 929 }, { "epoch": 0.94, "learning_rate": 1.9943300599757097e-05, "loss": 0.4699, "step": 930 }, { "epoch": 0.94, "learning_rate": 1.9943067250722398e-05, "loss": 0.4539, "step": 931 }, { "epoch": 0.95, "learning_rate": 1.9942833423864613e-05, "loss": 0.4354, "step": 932 }, { "epoch": 0.95, "learning_rate": 1.994259911919498e-05, "loss": 0.4386, "step": 933 }, { "epoch": 0.95, "learning_rate": 1.9942364336724757e-05, "loss": 0.4586, "step": 934 }, { "epoch": 0.95, "learning_rate": 1.9942129076465228e-05, "loss": 0.4883, "step": 935 }, { "epoch": 0.95, "learning_rate": 1.9941893338427695e-05, "loss": 0.4606, "step": 936 }, { "epoch": 0.95, "learning_rate": 1.9941657122623494e-05, "loss": 0.4594, "step": 937 }, { "epoch": 0.95, "learning_rate": 1.994142042906397e-05, "loss": 0.4714, "step": 938 }, { "epoch": 0.95, "learning_rate": 1.9941183257760503e-05, "loss": 0.412, "step": 939 }, { "epoch": 0.95, "learning_rate": 1.9940945608724484e-05, "loss": 0.5019, "step": 940 }, { "epoch": 0.95, "learning_rate": 1.9940707481967337e-05, "loss": 0.4746, "step": 941 }, { "epoch": 0.96, "learning_rate": 1.9940468877500508e-05, "loss": 0.5181, "step": 942 }, { "epoch": 0.96, "learning_rate": 1.994022979533546e-05, "loss": 0.4673, "step": 943 }, { "epoch": 0.96, "learning_rate": 1.993999023548368e-05, "loss": 0.4826, "step": 944 }, { "epoch": 0.96, "learning_rate": 1.9939750197956687e-05, "loss": 0.4608, "step": 945 }, { "epoch": 0.96, "learning_rate": 1.9939509682766007e-05, "loss": 0.5144, "step": 946 }, { "epoch": 0.96, "learning_rate": 1.993926868992321e-05, "loss": 0.4688, "step": 947 }, { "epoch": 0.96, "learning_rate": 1.9939027219439867e-05, "loss": 0.4839, "step": 948 }, { "epoch": 0.96, "learning_rate": 1.9938785271327586e-05, "loss": 0.4739, "step": 949 }, { "epoch": 0.96, "learning_rate": 1.9938542845597993e-05, "loss": 0.4924, "step": 950 }, { "epoch": 0.96, "learning_rate": 1.9938299942262743e-05, "loss": 0.4871, "step": 951 }, { "epoch": 0.97, "learning_rate": 1.9938056561333502e-05, "loss": 0.5166, "step": 952 }, { "epoch": 0.97, "learning_rate": 1.9937812702821966e-05, "loss": 0.4659, "step": 953 }, { "epoch": 0.97, "learning_rate": 1.9937568366739858e-05, "loss": 0.4694, "step": 954 }, { "epoch": 0.97, "learning_rate": 1.993732355309892e-05, "loss": 0.4504, "step": 955 }, { "epoch": 0.97, "learning_rate": 1.9937078261910914e-05, "loss": 0.4842, "step": 956 }, { "epoch": 0.97, "learning_rate": 1.993683249318763e-05, "loss": 0.5579, "step": 957 }, { "epoch": 0.97, "learning_rate": 1.9936586246940875e-05, "loss": 0.6141, "step": 958 }, { "epoch": 0.97, "learning_rate": 1.9936339523182488e-05, "loss": 0.5338, "step": 959 }, { "epoch": 0.97, "learning_rate": 1.993609232192432e-05, "loss": 0.4551, "step": 960 }, { "epoch": 0.98, "learning_rate": 1.9935844643178256e-05, "loss": 0.3826, "step": 961 }, { "epoch": 0.98, "learning_rate": 1.993559648695619e-05, "loss": 0.4304, "step": 962 }, { "epoch": 0.98, "learning_rate": 1.9935347853270058e-05, "loss": 0.422, "step": 963 }, { "epoch": 0.98, "learning_rate": 1.99350987421318e-05, "loss": 0.4714, "step": 964 }, { "epoch": 0.98, "learning_rate": 1.9934849153553387e-05, "loss": 0.5062, "step": 965 }, { "epoch": 0.98, "learning_rate": 1.9934599087546823e-05, "loss": 0.5131, "step": 966 }, { "epoch": 0.98, "learning_rate": 1.9934348544124117e-05, "loss": 0.4535, "step": 967 }, { "epoch": 0.98, "learning_rate": 1.993409752329731e-05, "loss": 0.468, "step": 968 }, { "epoch": 0.98, "learning_rate": 1.9933846025078465e-05, "loss": 0.4373, "step": 969 }, { "epoch": 0.98, "learning_rate": 1.993359404947967e-05, "loss": 0.4531, "step": 970 }, { "epoch": 0.99, "learning_rate": 1.9933341596513033e-05, "loss": 0.4683, "step": 971 }, { "epoch": 0.99, "learning_rate": 1.9933088666190686e-05, "loss": 0.4163, "step": 972 }, { "epoch": 0.99, "learning_rate": 1.993283525852478e-05, "loss": 0.5572, "step": 973 }, { "epoch": 0.99, "learning_rate": 1.9932581373527497e-05, "loss": 0.5376, "step": 974 }, { "epoch": 0.99, "learning_rate": 1.993232701121104e-05, "loss": 0.4828, "step": 975 }, { "epoch": 0.99, "learning_rate": 1.9932072171587627e-05, "loss": 0.5167, "step": 976 }, { "epoch": 0.99, "learning_rate": 1.9931816854669507e-05, "loss": 0.4691, "step": 977 }, { "epoch": 0.99, "learning_rate": 1.9931561060468948e-05, "loss": 0.5, "step": 978 }, { "epoch": 0.99, "learning_rate": 1.9931304788998245e-05, "loss": 0.4913, "step": 979 }, { "epoch": 0.99, "learning_rate": 1.9931048040269712e-05, "loss": 0.5542, "step": 980 }, { "epoch": 1.0, "learning_rate": 1.9930790814295688e-05, "loss": 0.4788, "step": 981 }, { "epoch": 1.0, "learning_rate": 1.9930533111088532e-05, "loss": 0.4603, "step": 982 }, { "epoch": 1.0, "learning_rate": 1.9930274930660632e-05, "loss": 0.4661, "step": 983 }, { "epoch": 1.0, "learning_rate": 1.9930016273024395e-05, "loss": 0.4906, "step": 984 }, { "epoch": 1.0, "learning_rate": 1.9929757138192245e-05, "loss": 0.4602, "step": 985 }, { "epoch": 1.0, "learning_rate": 1.992949752617664e-05, "loss": 0.4803, "step": 986 }, { "epoch": 1.0, "learning_rate": 1.9929237436990054e-05, "loss": 0.3677, "step": 987 }, { "epoch": 1.0, "learning_rate": 1.992897687064499e-05, "loss": 0.3927, "step": 988 }, { "epoch": 1.0, "learning_rate": 1.992871582715396e-05, "loss": 0.3445, "step": 989 }, { "epoch": 1.0, "learning_rate": 1.9928454306529523e-05, "loss": 0.3326, "step": 990 }, { "epoch": 1.01, "learning_rate": 1.9928192308784233e-05, "loss": 0.3771, "step": 991 }, { "epoch": 1.01, "learning_rate": 1.9927929833930688e-05, "loss": 0.3939, "step": 992 }, { "epoch": 1.01, "learning_rate": 1.99276668819815e-05, "loss": 0.3788, "step": 993 }, { "epoch": 1.01, "learning_rate": 1.9927403452949306e-05, "loss": 0.3912, "step": 994 }, { "epoch": 1.01, "learning_rate": 1.9927139546846762e-05, "loss": 0.4114, "step": 995 }, { "epoch": 1.01, "learning_rate": 1.992687516368655e-05, "loss": 0.3667, "step": 996 }, { "epoch": 1.01, "learning_rate": 1.9926610303481385e-05, "loss": 0.3235, "step": 997 }, { "epoch": 1.01, "learning_rate": 1.9926344966243984e-05, "loss": 0.4716, "step": 998 }, { "epoch": 1.01, "learning_rate": 1.9926079151987104e-05, "loss": 0.3508, "step": 999 }, { "epoch": 1.01, "learning_rate": 1.992581286072351e-05, "loss": 0.41, "step": 1000 }, { "epoch": 1.02, "learning_rate": 1.9925546092466016e-05, "loss": 0.3407, "step": 1001 }, { "epoch": 1.02, "learning_rate": 1.9925278847227423e-05, "loss": 0.3586, "step": 1002 }, { "epoch": 1.02, "learning_rate": 1.992501112502059e-05, "loss": 0.3166, "step": 1003 }, { "epoch": 1.02, "learning_rate": 1.9924742925858366e-05, "loss": 0.3673, "step": 1004 }, { "epoch": 1.02, "learning_rate": 1.9924474249753656e-05, "loss": 0.3995, "step": 1005 }, { "epoch": 1.02, "learning_rate": 1.992420509671936e-05, "loss": 0.3475, "step": 1006 }, { "epoch": 1.02, "learning_rate": 1.9923935466768415e-05, "loss": 0.4345, "step": 1007 }, { "epoch": 1.02, "learning_rate": 1.992366535991378e-05, "loss": 0.3751, "step": 1008 }, { "epoch": 1.02, "learning_rate": 1.992339477616844e-05, "loss": 0.4218, "step": 1009 }, { "epoch": 1.02, "learning_rate": 1.992312371554539e-05, "loss": 0.3565, "step": 1010 }, { "epoch": 1.03, "learning_rate": 1.992285217805766e-05, "loss": 0.4029, "step": 1011 }, { "epoch": 1.03, "learning_rate": 1.9922580163718296e-05, "loss": 0.3771, "step": 1012 }, { "epoch": 1.03, "learning_rate": 1.9922307672540372e-05, "loss": 0.3169, "step": 1013 }, { "epoch": 1.03, "learning_rate": 1.9922034704536984e-05, "loss": 0.3518, "step": 1014 }, { "epoch": 1.03, "learning_rate": 1.9921761259721247e-05, "loss": 0.411, "step": 1015 }, { "epoch": 1.03, "learning_rate": 1.9921487338106303e-05, "loss": 0.4055, "step": 1016 }, { "epoch": 1.03, "learning_rate": 1.9921212939705315e-05, "loss": 0.3632, "step": 1017 }, { "epoch": 1.03, "learning_rate": 1.992093806453147e-05, "loss": 0.3632, "step": 1018 }, { "epoch": 1.03, "learning_rate": 1.9920662712597983e-05, "loss": 0.3755, "step": 1019 }, { "epoch": 1.04, "learning_rate": 1.9920386883918076e-05, "loss": 0.3833, "step": 1020 }, { "epoch": 1.04, "learning_rate": 1.9920110578505007e-05, "loss": 0.3631, "step": 1021 }, { "epoch": 1.04, "learning_rate": 1.991983379637206e-05, "loss": 0.3549, "step": 1022 }, { "epoch": 1.04, "learning_rate": 1.991955653753253e-05, "loss": 0.3363, "step": 1023 }, { "epoch": 1.04, "learning_rate": 1.9919278801999743e-05, "loss": 0.3274, "step": 1024 }, { "epoch": 1.04, "learning_rate": 1.991900058978705e-05, "loss": 0.3358, "step": 1025 }, { "epoch": 1.04, "learning_rate": 1.991872190090781e-05, "loss": 0.3784, "step": 1026 }, { "epoch": 1.04, "learning_rate": 1.9918442735375425e-05, "loss": 0.3985, "step": 1027 }, { "epoch": 1.04, "learning_rate": 1.9918163093203308e-05, "loss": 0.301, "step": 1028 }, { "epoch": 1.04, "learning_rate": 1.99178829744049e-05, "loss": 0.3364, "step": 1029 }, { "epoch": 1.05, "learning_rate": 1.9917602378993653e-05, "loss": 0.4175, "step": 1030 }, { "epoch": 1.05, "learning_rate": 1.9917321306983064e-05, "loss": 0.3785, "step": 1031 }, { "epoch": 1.05, "learning_rate": 1.9917039758386632e-05, "loss": 0.3364, "step": 1032 }, { "epoch": 1.05, "learning_rate": 1.9916757733217887e-05, "loss": 0.3472, "step": 1033 }, { "epoch": 1.05, "learning_rate": 1.9916475231490387e-05, "loss": 0.3382, "step": 1034 }, { "epoch": 1.05, "learning_rate": 1.9916192253217707e-05, "loss": 0.366, "step": 1035 }, { "epoch": 1.05, "learning_rate": 1.9915908798413438e-05, "loss": 0.3933, "step": 1036 }, { "epoch": 1.05, "learning_rate": 1.991562486709121e-05, "loss": 0.3669, "step": 1037 }, { "epoch": 1.05, "learning_rate": 1.9915340459264664e-05, "loss": 0.3377, "step": 1038 }, { "epoch": 1.05, "learning_rate": 1.991505557494747e-05, "loss": 0.3499, "step": 1039 }, { "epoch": 1.06, "learning_rate": 1.991477021415332e-05, "loss": 0.3719, "step": 1040 }, { "epoch": 1.06, "learning_rate": 1.991448437689592e-05, "loss": 0.4067, "step": 1041 }, { "epoch": 1.06, "learning_rate": 1.991419806318901e-05, "loss": 0.3791, "step": 1042 }, { "epoch": 1.06, "learning_rate": 1.991391127304635e-05, "loss": 0.3675, "step": 1043 }, { "epoch": 1.06, "learning_rate": 1.991362400648172e-05, "loss": 0.3524, "step": 1044 }, { "epoch": 1.06, "learning_rate": 1.9913336263508927e-05, "loss": 0.3475, "step": 1045 }, { "epoch": 1.06, "learning_rate": 1.99130480441418e-05, "loss": 0.3584, "step": 1046 }, { "epoch": 1.06, "learning_rate": 1.9912759348394187e-05, "loss": 0.435, "step": 1047 }, { "epoch": 1.06, "learning_rate": 1.991247017627996e-05, "loss": 0.3833, "step": 1048 }, { "epoch": 1.06, "learning_rate": 1.991218052781302e-05, "loss": 0.426, "step": 1049 }, { "epoch": 1.07, "learning_rate": 1.9911890403007283e-05, "loss": 0.3482, "step": 1050 }, { "epoch": 1.07, "learning_rate": 1.9911599801876687e-05, "loss": 0.3823, "step": 1051 }, { "epoch": 1.07, "learning_rate": 1.9911308724435207e-05, "loss": 0.3757, "step": 1052 }, { "epoch": 1.07, "learning_rate": 1.9911017170696827e-05, "loss": 0.3887, "step": 1053 }, { "epoch": 1.07, "learning_rate": 1.9910725140675554e-05, "loss": 0.3357, "step": 1054 }, { "epoch": 1.07, "learning_rate": 1.9910432634385427e-05, "loss": 0.3989, "step": 1055 }, { "epoch": 1.07, "learning_rate": 1.9910139651840497e-05, "loss": 0.3574, "step": 1056 }, { "epoch": 1.07, "learning_rate": 1.9909846193054854e-05, "loss": 0.3828, "step": 1057 }, { "epoch": 1.07, "learning_rate": 1.990955225804259e-05, "loss": 0.411, "step": 1058 }, { "epoch": 1.07, "learning_rate": 1.990925784681783e-05, "loss": 0.3908, "step": 1059 }, { "epoch": 1.08, "learning_rate": 1.990896295939473e-05, "loss": 0.3893, "step": 1060 }, { "epoch": 1.08, "learning_rate": 1.9908667595787455e-05, "loss": 0.3371, "step": 1061 }, { "epoch": 1.08, "learning_rate": 1.9908371756010202e-05, "loss": 0.3422, "step": 1062 }, { "epoch": 1.08, "learning_rate": 1.9908075440077186e-05, "loss": 0.3549, "step": 1063 }, { "epoch": 1.08, "learning_rate": 1.990777864800265e-05, "loss": 0.353, "step": 1064 }, { "epoch": 1.08, "learning_rate": 1.990748137980085e-05, "loss": 0.3991, "step": 1065 }, { "epoch": 1.08, "learning_rate": 1.9907183635486077e-05, "loss": 0.4014, "step": 1066 }, { "epoch": 1.08, "learning_rate": 1.9906885415072642e-05, "loss": 0.3603, "step": 1067 }, { "epoch": 1.08, "learning_rate": 1.9906586718574866e-05, "loss": 0.3892, "step": 1068 }, { "epoch": 1.08, "learning_rate": 1.990628754600711e-05, "loss": 0.3662, "step": 1069 }, { "epoch": 1.09, "learning_rate": 1.9905987897383757e-05, "loss": 0.3611, "step": 1070 }, { "epoch": 1.09, "learning_rate": 1.9905687772719195e-05, "loss": 0.3796, "step": 1071 }, { "epoch": 1.09, "learning_rate": 1.9905387172027853e-05, "loss": 0.3612, "step": 1072 }, { "epoch": 1.09, "learning_rate": 1.990508609532417e-05, "loss": 0.4106, "step": 1073 }, { "epoch": 1.09, "learning_rate": 1.990478454262263e-05, "loss": 0.3954, "step": 1074 }, { "epoch": 1.09, "learning_rate": 1.990448251393771e-05, "loss": 0.3746, "step": 1075 }, { "epoch": 1.09, "learning_rate": 1.9904180009283932e-05, "loss": 0.4224, "step": 1076 }, { "epoch": 1.09, "learning_rate": 1.9903877028675827e-05, "loss": 0.413, "step": 1077 }, { "epoch": 1.09, "learning_rate": 1.9903573572127957e-05, "loss": 0.3494, "step": 1078 }, { "epoch": 1.09, "learning_rate": 1.990326963965491e-05, "loss": 0.3609, "step": 1079 }, { "epoch": 1.1, "learning_rate": 1.990296523127128e-05, "loss": 0.4377, "step": 1080 }, { "epoch": 1.1, "learning_rate": 1.9902660346991708e-05, "loss": 0.3763, "step": 1081 }, { "epoch": 1.1, "learning_rate": 1.990235498683084e-05, "loss": 0.3374, "step": 1082 }, { "epoch": 1.1, "learning_rate": 1.990204915080335e-05, "loss": 0.3677, "step": 1083 }, { "epoch": 1.1, "learning_rate": 1.990174283892394e-05, "loss": 0.3651, "step": 1084 }, { "epoch": 1.1, "learning_rate": 1.9901436051207327e-05, "loss": 0.4349, "step": 1085 }, { "epoch": 1.1, "learning_rate": 1.9901128787668248e-05, "loss": 0.4402, "step": 1086 }, { "epoch": 1.1, "learning_rate": 1.9900821048321476e-05, "loss": 0.3948, "step": 1087 }, { "epoch": 1.1, "learning_rate": 1.99005128331818e-05, "loss": 0.3329, "step": 1088 }, { "epoch": 1.11, "learning_rate": 1.990020414226403e-05, "loss": 0.3565, "step": 1089 }, { "epoch": 1.11, "learning_rate": 1.9899894975582997e-05, "loss": 0.4154, "step": 1090 }, { "epoch": 1.11, "learning_rate": 1.9899585333153563e-05, "loss": 0.396, "step": 1091 }, { "epoch": 1.11, "learning_rate": 1.9899275214990602e-05, "loss": 0.3879, "step": 1092 }, { "epoch": 1.11, "learning_rate": 1.9898964621109025e-05, "loss": 0.3746, "step": 1093 }, { "epoch": 1.11, "learning_rate": 1.9898653551523756e-05, "loss": 0.343, "step": 1094 }, { "epoch": 1.11, "learning_rate": 1.9898342006249738e-05, "loss": 0.3692, "step": 1095 }, { "epoch": 1.11, "learning_rate": 1.989802998530195e-05, "loss": 0.3252, "step": 1096 }, { "epoch": 1.11, "learning_rate": 1.9897717488695382e-05, "loss": 0.2916, "step": 1097 }, { "epoch": 1.11, "learning_rate": 1.989740451644505e-05, "loss": 0.3846, "step": 1098 }, { "epoch": 1.12, "learning_rate": 1.9897091068565997e-05, "loss": 0.3491, "step": 1099 }, { "epoch": 1.12, "learning_rate": 1.9896777145073284e-05, "loss": 0.428, "step": 1100 }, { "epoch": 1.12, "learning_rate": 1.9896462745982002e-05, "loss": 0.3357, "step": 1101 }, { "epoch": 1.12, "learning_rate": 1.9896147871307257e-05, "loss": 0.3651, "step": 1102 }, { "epoch": 1.12, "learning_rate": 1.9895832521064177e-05, "loss": 0.3442, "step": 1103 }, { "epoch": 1.12, "learning_rate": 1.989551669526792e-05, "loss": 0.3456, "step": 1104 }, { "epoch": 1.12, "learning_rate": 1.9895200393933657e-05, "loss": 0.3635, "step": 1105 }, { "epoch": 1.12, "learning_rate": 1.9894883617076602e-05, "loss": 0.3665, "step": 1106 }, { "epoch": 1.12, "learning_rate": 1.9894566364711965e-05, "loss": 0.3493, "step": 1107 }, { "epoch": 1.12, "learning_rate": 1.9894248636854997e-05, "loss": 0.382, "step": 1108 }, { "epoch": 1.13, "learning_rate": 1.9893930433520964e-05, "loss": 0.3947, "step": 1109 }, { "epoch": 1.13, "learning_rate": 1.9893611754725162e-05, "loss": 0.3451, "step": 1110 }, { "epoch": 1.13, "learning_rate": 1.9893292600482903e-05, "loss": 0.3991, "step": 1111 }, { "epoch": 1.13, "learning_rate": 1.9892972970809524e-05, "loss": 0.3528, "step": 1112 }, { "epoch": 1.13, "learning_rate": 1.9892652865720387e-05, "loss": 0.3787, "step": 1113 }, { "epoch": 1.13, "learning_rate": 1.9892332285230868e-05, "loss": 0.3712, "step": 1114 }, { "epoch": 1.13, "learning_rate": 1.9892011229356382e-05, "loss": 0.3695, "step": 1115 }, { "epoch": 1.13, "learning_rate": 1.9891689698112352e-05, "loss": 0.3049, "step": 1116 }, { "epoch": 1.13, "learning_rate": 1.9891367691514232e-05, "loss": 0.4116, "step": 1117 }, { "epoch": 1.13, "learning_rate": 1.9891045209577496e-05, "loss": 0.3559, "step": 1118 }, { "epoch": 1.14, "learning_rate": 1.9890722252317643e-05, "loss": 0.3172, "step": 1119 }, { "epoch": 1.14, "learning_rate": 1.989039881975019e-05, "loss": 0.3431, "step": 1120 }, { "epoch": 1.14, "learning_rate": 1.9890074911890677e-05, "loss": 0.3892, "step": 1121 }, { "epoch": 1.14, "learning_rate": 1.9889750528754673e-05, "loss": 0.3706, "step": 1122 }, { "epoch": 1.14, "learning_rate": 1.988942567035777e-05, "loss": 0.3574, "step": 1123 }, { "epoch": 1.14, "learning_rate": 1.9889100336715575e-05, "loss": 0.4016, "step": 1124 }, { "epoch": 1.14, "learning_rate": 1.9888774527843723e-05, "loss": 0.2946, "step": 1125 }, { "epoch": 1.14, "learning_rate": 1.988844824375787e-05, "loss": 0.4201, "step": 1126 }, { "epoch": 1.14, "learning_rate": 1.98881214844737e-05, "loss": 0.4082, "step": 1127 }, { "epoch": 1.14, "learning_rate": 1.9887794250006907e-05, "loss": 0.4189, "step": 1128 }, { "epoch": 1.15, "learning_rate": 1.988746654037323e-05, "loss": 0.3736, "step": 1129 }, { "epoch": 1.15, "learning_rate": 1.9887138355588404e-05, "loss": 0.3901, "step": 1130 }, { "epoch": 1.15, "learning_rate": 1.988680969566821e-05, "loss": 0.3982, "step": 1131 }, { "epoch": 1.15, "learning_rate": 1.9886480560628433e-05, "loss": 0.3782, "step": 1132 }, { "epoch": 1.15, "learning_rate": 1.9886150950484902e-05, "loss": 0.3538, "step": 1133 }, { "epoch": 1.15, "learning_rate": 1.9885820865253446e-05, "loss": 0.4039, "step": 1134 }, { "epoch": 1.15, "learning_rate": 1.988549030494993e-05, "loss": 0.3912, "step": 1135 }, { "epoch": 1.15, "learning_rate": 1.9885159269590243e-05, "loss": 0.3082, "step": 1136 }, { "epoch": 1.15, "learning_rate": 1.988482775919029e-05, "loss": 0.3756, "step": 1137 }, { "epoch": 1.15, "learning_rate": 1.9884495773766002e-05, "loss": 0.3818, "step": 1138 }, { "epoch": 1.16, "learning_rate": 1.9884163313333334e-05, "loss": 0.3788, "step": 1139 }, { "epoch": 1.16, "learning_rate": 1.9883830377908265e-05, "loss": 0.3523, "step": 1140 }, { "epoch": 1.16, "learning_rate": 1.9883496967506788e-05, "loss": 0.3461, "step": 1141 }, { "epoch": 1.16, "learning_rate": 1.988316308214493e-05, "loss": 0.348, "step": 1142 }, { "epoch": 1.16, "learning_rate": 1.9882828721838735e-05, "loss": 0.3698, "step": 1143 }, { "epoch": 1.16, "learning_rate": 1.9882493886604276e-05, "loss": 0.4253, "step": 1144 }, { "epoch": 1.16, "learning_rate": 1.9882158576457633e-05, "loss": 0.3746, "step": 1145 }, { "epoch": 1.16, "learning_rate": 1.9881822791414928e-05, "loss": 0.4023, "step": 1146 }, { "epoch": 1.16, "learning_rate": 1.9881486531492298e-05, "loss": 0.4164, "step": 1147 }, { "epoch": 1.16, "learning_rate": 1.9881149796705898e-05, "loss": 0.354, "step": 1148 }, { "epoch": 1.17, "learning_rate": 1.9880812587071913e-05, "loss": 0.4205, "step": 1149 }, { "epoch": 1.17, "learning_rate": 1.9880474902606544e-05, "loss": 0.4091, "step": 1150 }, { "epoch": 1.17, "learning_rate": 1.988013674332602e-05, "loss": 0.4482, "step": 1151 }, { "epoch": 1.17, "learning_rate": 1.9879798109246596e-05, "loss": 0.3823, "step": 1152 }, { "epoch": 1.17, "learning_rate": 1.987945900038454e-05, "loss": 0.3845, "step": 1153 }, { "epoch": 1.17, "learning_rate": 1.987911941675615e-05, "loss": 0.3682, "step": 1154 }, { "epoch": 1.17, "learning_rate": 1.987877935837775e-05, "loss": 0.324, "step": 1155 }, { "epoch": 1.17, "learning_rate": 1.9878438825265674e-05, "loss": 0.3306, "step": 1156 }, { "epoch": 1.17, "learning_rate": 1.9878097817436292e-05, "loss": 0.4152, "step": 1157 }, { "epoch": 1.18, "learning_rate": 1.9877756334905983e-05, "loss": 0.3279, "step": 1158 }, { "epoch": 1.18, "learning_rate": 1.987741437769117e-05, "loss": 0.3993, "step": 1159 }, { "epoch": 1.18, "learning_rate": 1.9877071945808275e-05, "loss": 0.3735, "step": 1160 }, { "epoch": 1.18, "learning_rate": 1.987672903927376e-05, "loss": 0.3585, "step": 1161 }, { "epoch": 1.18, "learning_rate": 1.9876385658104106e-05, "loss": 0.3801, "step": 1162 }, { "epoch": 1.18, "learning_rate": 1.9876041802315808e-05, "loss": 0.3245, "step": 1163 }, { "epoch": 1.18, "learning_rate": 1.987569747192539e-05, "loss": 0.3946, "step": 1164 }, { "epoch": 1.18, "learning_rate": 1.9875352666949404e-05, "loss": 0.4053, "step": 1165 }, { "epoch": 1.18, "learning_rate": 1.987500738740442e-05, "loss": 0.3663, "step": 1166 }, { "epoch": 1.18, "learning_rate": 1.9874661633307027e-05, "loss": 0.3624, "step": 1167 }, { "epoch": 1.19, "learning_rate": 1.9874315404673844e-05, "loss": 0.3796, "step": 1168 }, { "epoch": 1.19, "learning_rate": 1.9873968701521505e-05, "loss": 0.3609, "step": 1169 }, { "epoch": 1.19, "learning_rate": 1.9873621523866676e-05, "loss": 0.3672, "step": 1170 }, { "epoch": 1.19, "learning_rate": 1.9873273871726036e-05, "loss": 0.361, "step": 1171 }, { "epoch": 1.19, "learning_rate": 1.9872925745116296e-05, "loss": 0.3826, "step": 1172 }, { "epoch": 1.19, "learning_rate": 1.9872577144054182e-05, "loss": 0.4073, "step": 1173 }, { "epoch": 1.19, "learning_rate": 1.987222806855645e-05, "loss": 0.3677, "step": 1174 }, { "epoch": 1.19, "learning_rate": 1.9871878518639875e-05, "loss": 0.3647, "step": 1175 }, { "epoch": 1.19, "learning_rate": 1.987152849432125e-05, "loss": 0.3623, "step": 1176 }, { "epoch": 1.19, "learning_rate": 1.98711779956174e-05, "loss": 0.3403, "step": 1177 }, { "epoch": 1.2, "learning_rate": 1.9870827022545168e-05, "loss": 0.3469, "step": 1178 }, { "epoch": 1.2, "learning_rate": 1.987047557512142e-05, "loss": 0.3811, "step": 1179 }, { "epoch": 1.2, "learning_rate": 1.9870123653363046e-05, "loss": 0.4047, "step": 1180 }, { "epoch": 1.2, "learning_rate": 1.9869771257286957e-05, "loss": 0.3607, "step": 1181 }, { "epoch": 1.2, "learning_rate": 1.9869418386910087e-05, "loss": 0.4029, "step": 1182 }, { "epoch": 1.2, "learning_rate": 1.986906504224939e-05, "loss": 0.3522, "step": 1183 }, { "epoch": 1.2, "learning_rate": 1.9868711223321858e-05, "loss": 0.3956, "step": 1184 }, { "epoch": 1.2, "learning_rate": 1.9868356930144478e-05, "loss": 0.3693, "step": 1185 }, { "epoch": 1.2, "learning_rate": 1.9868002162734293e-05, "loss": 0.3456, "step": 1186 }, { "epoch": 1.2, "learning_rate": 1.9867646921108335e-05, "loss": 0.3564, "step": 1187 }, { "epoch": 1.21, "learning_rate": 1.986729120528369e-05, "loss": 0.3183, "step": 1188 }, { "epoch": 1.21, "learning_rate": 1.9866935015277446e-05, "loss": 0.33, "step": 1189 }, { "epoch": 1.21, "learning_rate": 1.9866578351106717e-05, "loss": 0.3431, "step": 1190 }, { "epoch": 1.21, "learning_rate": 1.9866221212788646e-05, "loss": 0.3581, "step": 1191 }, { "epoch": 1.21, "learning_rate": 1.9865863600340393e-05, "loss": 0.384, "step": 1192 }, { "epoch": 1.21, "learning_rate": 1.986550551377915e-05, "loss": 0.3528, "step": 1193 }, { "epoch": 1.21, "learning_rate": 1.9865146953122116e-05, "loss": 0.3989, "step": 1194 }, { "epoch": 1.21, "learning_rate": 1.986478791838653e-05, "loss": 0.3601, "step": 1195 }, { "epoch": 1.21, "learning_rate": 1.986442840958964e-05, "loss": 0.3644, "step": 1196 }, { "epoch": 1.21, "learning_rate": 1.986406842674873e-05, "loss": 0.3986, "step": 1197 }, { "epoch": 1.22, "learning_rate": 1.986370796988109e-05, "loss": 0.4017, "step": 1198 }, { "epoch": 1.22, "learning_rate": 1.986334703900405e-05, "loss": 0.4098, "step": 1199 }, { "epoch": 1.22, "learning_rate": 1.9862985634134948e-05, "loss": 0.4041, "step": 1200 }, { "epoch": 1.22, "learning_rate": 1.9862623755291156e-05, "loss": 0.4303, "step": 1201 }, { "epoch": 1.22, "learning_rate": 1.986226140249006e-05, "loss": 0.3944, "step": 1202 }, { "epoch": 1.22, "learning_rate": 1.9861898575749082e-05, "loss": 0.3557, "step": 1203 }, { "epoch": 1.22, "learning_rate": 1.986153527508565e-05, "loss": 0.3558, "step": 1204 }, { "epoch": 1.22, "learning_rate": 1.9861171500517223e-05, "loss": 0.3742, "step": 1205 }, { "epoch": 1.22, "learning_rate": 1.9860807252061288e-05, "loss": 0.3742, "step": 1206 }, { "epoch": 1.22, "learning_rate": 1.9860442529735346e-05, "loss": 0.372, "step": 1207 }, { "epoch": 1.23, "learning_rate": 1.9860077333556923e-05, "loss": 0.4299, "step": 1208 }, { "epoch": 1.23, "learning_rate": 1.9859711663543573e-05, "loss": 0.3885, "step": 1209 }, { "epoch": 1.23, "learning_rate": 1.985934551971286e-05, "loss": 0.3938, "step": 1210 }, { "epoch": 1.23, "learning_rate": 1.9858978902082388e-05, "loss": 0.4026, "step": 1211 }, { "epoch": 1.23, "learning_rate": 1.9858611810669772e-05, "loss": 0.3528, "step": 1212 }, { "epoch": 1.23, "learning_rate": 1.985824424549265e-05, "loss": 0.3568, "step": 1213 }, { "epoch": 1.23, "learning_rate": 1.9857876206568692e-05, "loss": 0.365, "step": 1214 }, { "epoch": 1.23, "learning_rate": 1.985750769391558e-05, "loss": 0.4239, "step": 1215 }, { "epoch": 1.23, "learning_rate": 1.9857138707551026e-05, "loss": 0.3549, "step": 1216 }, { "epoch": 1.23, "learning_rate": 1.985676924749276e-05, "loss": 0.3737, "step": 1217 }, { "epoch": 1.24, "learning_rate": 1.9856399313758533e-05, "loss": 0.4169, "step": 1218 }, { "epoch": 1.24, "learning_rate": 1.9856028906366132e-05, "loss": 0.4048, "step": 1219 }, { "epoch": 1.24, "learning_rate": 1.9855658025333348e-05, "loss": 0.3196, "step": 1220 }, { "epoch": 1.24, "learning_rate": 1.985528667067801e-05, "loss": 0.4524, "step": 1221 }, { "epoch": 1.24, "learning_rate": 1.985491484241796e-05, "loss": 0.3296, "step": 1222 }, { "epoch": 1.24, "learning_rate": 1.985454254057107e-05, "loss": 0.3823, "step": 1223 }, { "epoch": 1.24, "learning_rate": 1.9854169765155228e-05, "loss": 0.4248, "step": 1224 }, { "epoch": 1.24, "learning_rate": 1.9853796516188345e-05, "loss": 0.3374, "step": 1225 }, { "epoch": 1.24, "learning_rate": 1.985342279368837e-05, "loss": 0.3359, "step": 1226 }, { "epoch": 1.25, "learning_rate": 1.985304859767325e-05, "loss": 0.4005, "step": 1227 }, { "epoch": 1.25, "learning_rate": 1.9852673928160973e-05, "loss": 0.3565, "step": 1228 }, { "epoch": 1.25, "learning_rate": 1.9852298785169544e-05, "loss": 0.4074, "step": 1229 }, { "epoch": 1.25, "learning_rate": 1.985192316871699e-05, "loss": 0.4234, "step": 1230 }, { "epoch": 1.25, "learning_rate": 1.9851547078821362e-05, "loss": 0.3799, "step": 1231 }, { "epoch": 1.25, "learning_rate": 1.9851170515500733e-05, "loss": 0.4267, "step": 1232 }, { "epoch": 1.25, "learning_rate": 1.9850793478773197e-05, "loss": 0.4249, "step": 1233 }, { "epoch": 1.25, "learning_rate": 1.985041596865688e-05, "loss": 0.3891, "step": 1234 }, { "epoch": 1.25, "learning_rate": 1.9850037985169912e-05, "loss": 0.3645, "step": 1235 }, { "epoch": 1.25, "learning_rate": 1.9849659528330467e-05, "loss": 0.3719, "step": 1236 }, { "epoch": 1.26, "learning_rate": 1.984928059815673e-05, "loss": 0.3534, "step": 1237 }, { "epoch": 1.26, "learning_rate": 1.984890119466691e-05, "loss": 0.3653, "step": 1238 }, { "epoch": 1.26, "learning_rate": 1.984852131787924e-05, "loss": 0.3526, "step": 1239 }, { "epoch": 1.26, "learning_rate": 1.984814096781197e-05, "loss": 0.3801, "step": 1240 }, { "epoch": 1.26, "learning_rate": 1.9847760144483388e-05, "loss": 0.4712, "step": 1241 }, { "epoch": 1.26, "learning_rate": 1.9847378847911787e-05, "loss": 0.3322, "step": 1242 }, { "epoch": 1.26, "learning_rate": 1.9846997078115493e-05, "loss": 0.401, "step": 1243 }, { "epoch": 1.26, "learning_rate": 1.984661483511285e-05, "loss": 0.3768, "step": 1244 }, { "epoch": 1.26, "learning_rate": 1.9846232118922234e-05, "loss": 0.3921, "step": 1245 }, { "epoch": 1.26, "learning_rate": 1.9845848929562034e-05, "loss": 0.3537, "step": 1246 }, { "epoch": 1.27, "learning_rate": 1.9845465267050656e-05, "loss": 0.3177, "step": 1247 }, { "epoch": 1.27, "learning_rate": 1.984508113140655e-05, "loss": 0.3899, "step": 1248 }, { "epoch": 1.27, "learning_rate": 1.9844696522648164e-05, "loss": 0.3715, "step": 1249 }, { "epoch": 1.27, "learning_rate": 1.984431144079399e-05, "loss": 0.3636, "step": 1250 }, { "epoch": 1.27, "learning_rate": 1.984392588586253e-05, "loss": 0.3644, "step": 1251 }, { "epoch": 1.27, "learning_rate": 1.9843539857872313e-05, "loss": 0.4593, "step": 1252 }, { "epoch": 1.27, "learning_rate": 1.984315335684189e-05, "loss": 0.3759, "step": 1253 }, { "epoch": 1.27, "learning_rate": 1.984276638278983e-05, "loss": 0.4046, "step": 1254 }, { "epoch": 1.27, "learning_rate": 1.9842378935734737e-05, "loss": 0.3645, "step": 1255 }, { "epoch": 1.27, "learning_rate": 1.9841991015695225e-05, "loss": 0.3579, "step": 1256 }, { "epoch": 1.28, "learning_rate": 1.984160262268994e-05, "loss": 0.3337, "step": 1257 }, { "epoch": 1.28, "learning_rate": 1.9841213756737542e-05, "loss": 0.3648, "step": 1258 }, { "epoch": 1.28, "learning_rate": 1.9840824417856715e-05, "loss": 0.345, "step": 1259 }, { "epoch": 1.28, "learning_rate": 1.9840434606066182e-05, "loss": 0.3867, "step": 1260 }, { "epoch": 1.28, "learning_rate": 1.9840044321384663e-05, "loss": 0.3591, "step": 1261 }, { "epoch": 1.28, "learning_rate": 1.983965356383092e-05, "loss": 0.4043, "step": 1262 }, { "epoch": 1.28, "learning_rate": 1.983926233342373e-05, "loss": 0.425, "step": 1263 }, { "epoch": 1.28, "learning_rate": 1.9838870630181893e-05, "loss": 0.3495, "step": 1264 }, { "epoch": 1.28, "learning_rate": 1.9838478454124236e-05, "loss": 0.3741, "step": 1265 }, { "epoch": 1.28, "learning_rate": 1.98380858052696e-05, "loss": 0.3786, "step": 1266 }, { "epoch": 1.29, "learning_rate": 1.9837692683636856e-05, "loss": 0.3799, "step": 1267 }, { "epoch": 1.29, "learning_rate": 1.98372990892449e-05, "loss": 0.3519, "step": 1268 }, { "epoch": 1.29, "learning_rate": 1.983690502211264e-05, "loss": 0.3705, "step": 1269 }, { "epoch": 1.29, "learning_rate": 1.9836510482259014e-05, "loss": 0.4912, "step": 1270 }, { "epoch": 1.29, "learning_rate": 1.9836115469702987e-05, "loss": 0.4049, "step": 1271 }, { "epoch": 1.29, "learning_rate": 1.983571998446354e-05, "loss": 0.3771, "step": 1272 }, { "epoch": 1.29, "learning_rate": 1.9835324026559676e-05, "loss": 0.391, "step": 1273 }, { "epoch": 1.29, "learning_rate": 1.9834927596010426e-05, "loss": 0.3486, "step": 1274 }, { "epoch": 1.29, "learning_rate": 1.9834530692834838e-05, "loss": 0.428, "step": 1275 }, { "epoch": 1.29, "learning_rate": 1.983413331705199e-05, "loss": 0.3531, "step": 1276 }, { "epoch": 1.3, "learning_rate": 1.9833735468680967e-05, "loss": 0.3615, "step": 1277 }, { "epoch": 1.3, "learning_rate": 1.9833337147740903e-05, "loss": 0.3875, "step": 1278 }, { "epoch": 1.3, "learning_rate": 1.983293835425093e-05, "loss": 0.4195, "step": 1279 }, { "epoch": 1.3, "learning_rate": 1.9832539088230215e-05, "loss": 0.4113, "step": 1280 }, { "epoch": 1.3, "learning_rate": 1.9832139349697948e-05, "loss": 0.4224, "step": 1281 }, { "epoch": 1.3, "learning_rate": 1.9831739138673332e-05, "loss": 0.3561, "step": 1282 }, { "epoch": 1.3, "learning_rate": 1.9831338455175606e-05, "loss": 0.3043, "step": 1283 }, { "epoch": 1.3, "learning_rate": 1.9830937299224022e-05, "loss": 0.3979, "step": 1284 }, { "epoch": 1.3, "learning_rate": 1.9830535670837858e-05, "loss": 0.3503, "step": 1285 }, { "epoch": 1.3, "learning_rate": 1.9830133570036414e-05, "loss": 0.3871, "step": 1286 }, { "epoch": 1.31, "learning_rate": 1.982973099683902e-05, "loss": 0.4488, "step": 1287 }, { "epoch": 1.31, "learning_rate": 1.982932795126501e-05, "loss": 0.3594, "step": 1288 }, { "epoch": 1.31, "learning_rate": 1.9828924433333767e-05, "loss": 0.383, "step": 1289 }, { "epoch": 1.31, "learning_rate": 1.982852044306467e-05, "loss": 0.4037, "step": 1290 }, { "epoch": 1.31, "learning_rate": 1.9828115980477135e-05, "loss": 0.3836, "step": 1291 }, { "epoch": 1.31, "learning_rate": 1.9827711045590606e-05, "loss": 0.4366, "step": 1292 }, { "epoch": 1.31, "learning_rate": 1.9827305638424537e-05, "loss": 0.3536, "step": 1293 }, { "epoch": 1.31, "learning_rate": 1.982689975899841e-05, "loss": 0.381, "step": 1294 }, { "epoch": 1.31, "learning_rate": 1.9826493407331732e-05, "loss": 0.3325, "step": 1295 }, { "epoch": 1.32, "learning_rate": 1.982608658344403e-05, "loss": 0.4018, "step": 1296 }, { "epoch": 1.32, "learning_rate": 1.9825679287354855e-05, "loss": 0.3997, "step": 1297 }, { "epoch": 1.32, "learning_rate": 1.9825271519083778e-05, "loss": 0.3611, "step": 1298 }, { "epoch": 1.32, "learning_rate": 1.9824863278650396e-05, "loss": 0.3843, "step": 1299 }, { "epoch": 1.32, "learning_rate": 1.9824454566074328e-05, "loss": 0.3688, "step": 1300 }, { "epoch": 1.32, "learning_rate": 1.9824045381375214e-05, "loss": 0.3436, "step": 1301 }, { "epoch": 1.32, "learning_rate": 1.982363572457272e-05, "loss": 0.3996, "step": 1302 }, { "epoch": 1.32, "learning_rate": 1.9823225595686527e-05, "loss": 0.3544, "step": 1303 }, { "epoch": 1.32, "learning_rate": 1.9822814994736348e-05, "loss": 0.3766, "step": 1304 }, { "epoch": 1.32, "learning_rate": 1.9822403921741914e-05, "loss": 0.4176, "step": 1305 }, { "epoch": 1.33, "learning_rate": 1.982199237672298e-05, "loss": 0.3517, "step": 1306 }, { "epoch": 1.33, "learning_rate": 1.9821580359699326e-05, "loss": 0.3613, "step": 1307 }, { "epoch": 1.33, "learning_rate": 1.9821167870690746e-05, "loss": 0.4179, "step": 1308 }, { "epoch": 1.33, "learning_rate": 1.9820754909717067e-05, "loss": 0.4066, "step": 1309 }, { "epoch": 1.33, "learning_rate": 1.982034147679813e-05, "loss": 0.361, "step": 1310 }, { "epoch": 1.33, "learning_rate": 1.9819927571953804e-05, "loss": 0.3315, "step": 1311 }, { "epoch": 1.33, "learning_rate": 1.9819513195203985e-05, "loss": 0.3684, "step": 1312 }, { "epoch": 1.33, "learning_rate": 1.981909834656858e-05, "loss": 0.3725, "step": 1313 }, { "epoch": 1.33, "learning_rate": 1.9818683026067528e-05, "loss": 0.4212, "step": 1314 }, { "epoch": 1.33, "learning_rate": 1.9818267233720787e-05, "loss": 0.4007, "step": 1315 }, { "epoch": 1.34, "learning_rate": 1.981785096954834e-05, "loss": 0.4139, "step": 1316 }, { "epoch": 1.34, "learning_rate": 1.9817434233570186e-05, "loss": 0.3906, "step": 1317 }, { "epoch": 1.34, "learning_rate": 1.9817017025806356e-05, "loss": 0.3554, "step": 1318 }, { "epoch": 1.34, "learning_rate": 1.9816599346276898e-05, "loss": 0.3666, "step": 1319 }, { "epoch": 1.34, "learning_rate": 1.9816181195001882e-05, "loss": 0.3288, "step": 1320 }, { "epoch": 1.34, "learning_rate": 1.9815762572001407e-05, "loss": 0.3593, "step": 1321 }, { "epoch": 1.34, "learning_rate": 1.9815343477295586e-05, "loss": 0.4089, "step": 1322 }, { "epoch": 1.34, "learning_rate": 1.9814923910904563e-05, "loss": 0.3658, "step": 1323 }, { "epoch": 1.34, "learning_rate": 1.9814503872848495e-05, "loss": 0.355, "step": 1324 }, { "epoch": 1.34, "learning_rate": 1.9814083363147572e-05, "loss": 0.3602, "step": 1325 }, { "epoch": 1.35, "learning_rate": 1.9813662381822004e-05, "loss": 0.4104, "step": 1326 }, { "epoch": 1.35, "learning_rate": 1.981324092889202e-05, "loss": 0.3134, "step": 1327 }, { "epoch": 1.35, "learning_rate": 1.9812819004377862e-05, "loss": 0.3809, "step": 1328 }, { "epoch": 1.35, "learning_rate": 1.9812396608299824e-05, "loss": 0.364, "step": 1329 }, { "epoch": 1.35, "learning_rate": 1.981197374067819e-05, "loss": 0.4095, "step": 1330 }, { "epoch": 1.35, "learning_rate": 1.9811550401533294e-05, "loss": 0.3674, "step": 1331 }, { "epoch": 1.35, "learning_rate": 1.981112659088547e-05, "loss": 0.3634, "step": 1332 }, { "epoch": 1.35, "learning_rate": 1.981070230875509e-05, "loss": 0.4268, "step": 1333 }, { "epoch": 1.35, "learning_rate": 1.9810277555162537e-05, "loss": 0.4151, "step": 1334 }, { "epoch": 1.35, "learning_rate": 1.9809852330128235e-05, "loss": 0.3623, "step": 1335 }, { "epoch": 1.36, "learning_rate": 1.9809426633672607e-05, "loss": 0.3773, "step": 1336 }, { "epoch": 1.36, "learning_rate": 1.980900046581611e-05, "loss": 0.3765, "step": 1337 }, { "epoch": 1.36, "learning_rate": 1.980857382657923e-05, "loss": 0.3957, "step": 1338 }, { "epoch": 1.36, "learning_rate": 1.980814671598247e-05, "loss": 0.3813, "step": 1339 }, { "epoch": 1.36, "learning_rate": 1.9807719134046354e-05, "loss": 0.3803, "step": 1340 }, { "epoch": 1.36, "learning_rate": 1.9807291080791423e-05, "loss": 0.3914, "step": 1341 }, { "epoch": 1.36, "learning_rate": 1.9806862556238257e-05, "loss": 0.3948, "step": 1342 }, { "epoch": 1.36, "learning_rate": 1.9806433560407444e-05, "loss": 0.3518, "step": 1343 }, { "epoch": 1.36, "learning_rate": 1.9806004093319603e-05, "loss": 0.4303, "step": 1344 }, { "epoch": 1.36, "learning_rate": 1.980557415499537e-05, "loss": 0.317, "step": 1345 }, { "epoch": 1.37, "learning_rate": 1.9805143745455403e-05, "loss": 0.3441, "step": 1346 }, { "epoch": 1.37, "learning_rate": 1.980471286472039e-05, "loss": 0.363, "step": 1347 }, { "epoch": 1.37, "learning_rate": 1.980428151281104e-05, "loss": 0.3376, "step": 1348 }, { "epoch": 1.37, "learning_rate": 1.9803849689748076e-05, "loss": 0.4082, "step": 1349 }, { "epoch": 1.37, "learning_rate": 1.9803417395552254e-05, "loss": 0.3503, "step": 1350 }, { "epoch": 1.37, "learning_rate": 1.9802984630244344e-05, "loss": 0.3461, "step": 1351 }, { "epoch": 1.37, "learning_rate": 1.9802551393845145e-05, "loss": 0.3705, "step": 1352 }, { "epoch": 1.37, "learning_rate": 1.9802117686375478e-05, "loss": 0.3819, "step": 1353 }, { "epoch": 1.37, "learning_rate": 1.9801683507856185e-05, "loss": 0.3515, "step": 1354 }, { "epoch": 1.37, "learning_rate": 1.980124885830813e-05, "loss": 0.3367, "step": 1355 }, { "epoch": 1.38, "learning_rate": 1.98008137377522e-05, "loss": 0.5064, "step": 1356 }, { "epoch": 1.38, "learning_rate": 1.9800378146209306e-05, "loss": 0.3852, "step": 1357 }, { "epoch": 1.38, "learning_rate": 1.979994208370038e-05, "loss": 0.3491, "step": 1358 }, { "epoch": 1.38, "learning_rate": 1.9799505550246375e-05, "loss": 0.3548, "step": 1359 }, { "epoch": 1.38, "learning_rate": 1.9799068545868273e-05, "loss": 0.4728, "step": 1360 }, { "epoch": 1.38, "learning_rate": 1.979863107058708e-05, "loss": 0.3814, "step": 1361 }, { "epoch": 1.38, "learning_rate": 1.9798193124423804e-05, "loss": 0.3836, "step": 1362 }, { "epoch": 1.38, "learning_rate": 1.9797754707399504e-05, "loss": 0.3847, "step": 1363 }, { "epoch": 1.38, "learning_rate": 1.979731581953524e-05, "loss": 0.358, "step": 1364 }, { "epoch": 1.39, "learning_rate": 1.979687646085211e-05, "loss": 0.42, "step": 1365 }, { "epoch": 1.39, "learning_rate": 1.9796436631371224e-05, "loss": 0.3593, "step": 1366 }, { "epoch": 1.39, "learning_rate": 1.979599633111372e-05, "loss": 0.4402, "step": 1367 }, { "epoch": 1.39, "learning_rate": 1.9795555560100758e-05, "loss": 0.3445, "step": 1368 }, { "epoch": 1.39, "learning_rate": 1.979511431835352e-05, "loss": 0.328, "step": 1369 }, { "epoch": 1.39, "learning_rate": 1.9794672605893203e-05, "loss": 0.3505, "step": 1370 }, { "epoch": 1.39, "learning_rate": 1.9794230422741042e-05, "loss": 0.3928, "step": 1371 }, { "epoch": 1.39, "learning_rate": 1.9793787768918286e-05, "loss": 0.3653, "step": 1372 }, { "epoch": 1.39, "learning_rate": 1.97933446444462e-05, "loss": 0.325, "step": 1373 }, { "epoch": 1.39, "learning_rate": 1.9792901049346084e-05, "loss": 0.3482, "step": 1374 }, { "epoch": 1.4, "learning_rate": 1.979245698363926e-05, "loss": 0.2978, "step": 1375 }, { "epoch": 1.4, "learning_rate": 1.9792012447347057e-05, "loss": 0.398, "step": 1376 }, { "epoch": 1.4, "learning_rate": 1.979156744049085e-05, "loss": 0.3976, "step": 1377 }, { "epoch": 1.4, "learning_rate": 1.979112196309201e-05, "loss": 0.4705, "step": 1378 }, { "epoch": 1.4, "learning_rate": 1.979067601517196e-05, "loss": 0.4327, "step": 1379 }, { "epoch": 1.4, "learning_rate": 1.9790229596752118e-05, "loss": 0.3812, "step": 1380 }, { "epoch": 1.4, "learning_rate": 1.9789782707853944e-05, "loss": 0.3553, "step": 1381 }, { "epoch": 1.4, "learning_rate": 1.9789335348498912e-05, "loss": 0.3386, "step": 1382 }, { "epoch": 1.4, "learning_rate": 1.978888751870852e-05, "loss": 0.3776, "step": 1383 }, { "epoch": 1.4, "learning_rate": 1.9788439218504283e-05, "loss": 0.383, "step": 1384 }, { "epoch": 1.41, "learning_rate": 1.978799044790776e-05, "loss": 0.4247, "step": 1385 }, { "epoch": 1.41, "learning_rate": 1.9787541206940502e-05, "loss": 0.386, "step": 1386 }, { "epoch": 1.41, "learning_rate": 1.9787091495624105e-05, "loss": 0.3436, "step": 1387 }, { "epoch": 1.41, "learning_rate": 1.978664131398018e-05, "loss": 0.3677, "step": 1388 }, { "epoch": 1.41, "learning_rate": 1.9786190662030353e-05, "loss": 0.353, "step": 1389 }, { "epoch": 1.41, "learning_rate": 1.978573953979629e-05, "loss": 0.4406, "step": 1390 }, { "epoch": 1.41, "learning_rate": 1.978528794729967e-05, "loss": 0.3281, "step": 1391 }, { "epoch": 1.41, "learning_rate": 1.9784835884562185e-05, "loss": 0.2964, "step": 1392 }, { "epoch": 1.41, "learning_rate": 1.978438335160557e-05, "loss": 0.3579, "step": 1393 }, { "epoch": 1.41, "learning_rate": 1.9783930348451568e-05, "loss": 0.3548, "step": 1394 }, { "epoch": 1.42, "learning_rate": 1.9783476875121944e-05, "loss": 0.3554, "step": 1395 }, { "epoch": 1.42, "learning_rate": 1.97830229316385e-05, "loss": 0.3943, "step": 1396 }, { "epoch": 1.42, "learning_rate": 1.978256851802304e-05, "loss": 0.3955, "step": 1397 }, { "epoch": 1.42, "learning_rate": 1.978211363429741e-05, "loss": 0.3734, "step": 1398 }, { "epoch": 1.42, "learning_rate": 1.9781658280483464e-05, "loss": 0.3465, "step": 1399 }, { "epoch": 1.42, "learning_rate": 1.9781202456603087e-05, "loss": 0.361, "step": 1400 }, { "epoch": 1.42, "learning_rate": 1.978074616267818e-05, "loss": 0.3957, "step": 1401 }, { "epoch": 1.42, "learning_rate": 1.9780289398730678e-05, "loss": 0.3504, "step": 1402 }, { "epoch": 1.42, "learning_rate": 1.977983216478253e-05, "loss": 0.3541, "step": 1403 }, { "epoch": 1.42, "learning_rate": 1.97793744608557e-05, "loss": 0.3992, "step": 1404 }, { "epoch": 1.43, "learning_rate": 1.9778916286972193e-05, "loss": 0.4028, "step": 1405 }, { "epoch": 1.43, "learning_rate": 1.977845764315402e-05, "loss": 0.3752, "step": 1406 }, { "epoch": 1.43, "learning_rate": 1.977799852942323e-05, "loss": 0.3435, "step": 1407 }, { "epoch": 1.43, "learning_rate": 1.9777538945801877e-05, "loss": 0.3629, "step": 1408 }, { "epoch": 1.43, "learning_rate": 1.9777078892312053e-05, "loss": 0.3712, "step": 1409 }, { "epoch": 1.43, "learning_rate": 1.9776618368975862e-05, "loss": 0.3405, "step": 1410 }, { "epoch": 1.43, "learning_rate": 1.977615737581544e-05, "loss": 0.3599, "step": 1411 }, { "epoch": 1.43, "learning_rate": 1.9775695912852934e-05, "loss": 0.3931, "step": 1412 }, { "epoch": 1.43, "learning_rate": 1.9775233980110524e-05, "loss": 0.4143, "step": 1413 }, { "epoch": 1.43, "learning_rate": 1.9774771577610408e-05, "loss": 0.3785, "step": 1414 }, { "epoch": 1.44, "learning_rate": 1.9774308705374808e-05, "loss": 0.3977, "step": 1415 }, { "epoch": 1.44, "learning_rate": 1.9773845363425964e-05, "loss": 0.3662, "step": 1416 }, { "epoch": 1.44, "learning_rate": 1.977338155178615e-05, "loss": 0.4193, "step": 1417 }, { "epoch": 1.44, "learning_rate": 1.977291727047765e-05, "loss": 0.4216, "step": 1418 }, { "epoch": 1.44, "learning_rate": 1.9772452519522772e-05, "loss": 0.3341, "step": 1419 }, { "epoch": 1.44, "learning_rate": 1.9771987298943854e-05, "loss": 0.3255, "step": 1420 }, { "epoch": 1.44, "learning_rate": 1.9771521608763253e-05, "loss": 0.3863, "step": 1421 }, { "epoch": 1.44, "learning_rate": 1.9771055449003347e-05, "loss": 0.4323, "step": 1422 }, { "epoch": 1.44, "learning_rate": 1.9770588819686537e-05, "loss": 0.3771, "step": 1423 }, { "epoch": 1.44, "learning_rate": 1.977012172083525e-05, "loss": 0.3726, "step": 1424 }, { "epoch": 1.45, "learning_rate": 1.9769654152471937e-05, "loss": 0.3823, "step": 1425 }, { "epoch": 1.45, "learning_rate": 1.9769186114619054e-05, "loss": 0.3609, "step": 1426 }, { "epoch": 1.45, "learning_rate": 1.97687176072991e-05, "loss": 0.3879, "step": 1427 }, { "epoch": 1.45, "learning_rate": 1.976824863053459e-05, "loss": 0.3492, "step": 1428 }, { "epoch": 1.45, "learning_rate": 1.9767779184348068e-05, "loss": 0.3915, "step": 1429 }, { "epoch": 1.45, "learning_rate": 1.976730926876208e-05, "loss": 0.3655, "step": 1430 }, { "epoch": 1.45, "learning_rate": 1.976683888379922e-05, "loss": 0.3463, "step": 1431 }, { "epoch": 1.45, "learning_rate": 1.9766368029482086e-05, "loss": 0.365, "step": 1432 }, { "epoch": 1.45, "learning_rate": 1.9765896705833306e-05, "loss": 0.3813, "step": 1433 }, { "epoch": 1.46, "learning_rate": 1.9765424912875533e-05, "loss": 0.3559, "step": 1434 }, { "epoch": 1.46, "learning_rate": 1.9764952650631438e-05, "loss": 0.3334, "step": 1435 }, { "epoch": 1.46, "learning_rate": 1.9764479919123714e-05, "loss": 0.3572, "step": 1436 }, { "epoch": 1.46, "learning_rate": 1.976400671837508e-05, "loss": 0.3154, "step": 1437 }, { "epoch": 1.46, "learning_rate": 1.9763533048408273e-05, "loss": 0.3958, "step": 1438 }, { "epoch": 1.46, "learning_rate": 1.9763058909246066e-05, "loss": 0.3295, "step": 1439 }, { "epoch": 1.46, "learning_rate": 1.9762584300911235e-05, "loss": 0.3516, "step": 1440 }, { "epoch": 1.46, "learning_rate": 1.9762109223426583e-05, "loss": 0.3628, "step": 1441 }, { "epoch": 1.46, "learning_rate": 1.9761633676814955e-05, "loss": 0.3933, "step": 1442 }, { "epoch": 1.46, "learning_rate": 1.9761157661099192e-05, "loss": 0.3804, "step": 1443 }, { "epoch": 1.47, "learning_rate": 1.9760681176302173e-05, "loss": 0.4192, "step": 1444 }, { "epoch": 1.47, "learning_rate": 1.97602042224468e-05, "loss": 0.4081, "step": 1445 }, { "epoch": 1.47, "learning_rate": 1.9759726799555986e-05, "loss": 0.3085, "step": 1446 }, { "epoch": 1.47, "learning_rate": 1.975924890765268e-05, "loss": 0.4006, "step": 1447 }, { "epoch": 1.47, "learning_rate": 1.975877054675984e-05, "loss": 0.4309, "step": 1448 }, { "epoch": 1.47, "learning_rate": 1.9758291716900462e-05, "loss": 0.3891, "step": 1449 }, { "epoch": 1.47, "learning_rate": 1.975781241809756e-05, "loss": 0.3722, "step": 1450 }, { "epoch": 1.47, "learning_rate": 1.9757332650374152e-05, "loss": 0.339, "step": 1451 }, { "epoch": 1.47, "learning_rate": 1.9756852413753307e-05, "loss": 0.3981, "step": 1452 }, { "epoch": 1.47, "learning_rate": 1.9756371708258098e-05, "loss": 0.3615, "step": 1453 }, { "epoch": 1.48, "learning_rate": 1.9755890533911627e-05, "loss": 0.3779, "step": 1454 }, { "epoch": 1.48, "learning_rate": 1.9755408890737014e-05, "loss": 0.3635, "step": 1455 }, { "epoch": 1.48, "learning_rate": 1.975492677875741e-05, "loss": 0.4325, "step": 1456 }, { "epoch": 1.48, "learning_rate": 1.975444419799598e-05, "loss": 0.3587, "step": 1457 }, { "epoch": 1.48, "learning_rate": 1.9753961148475915e-05, "loss": 0.3684, "step": 1458 }, { "epoch": 1.48, "learning_rate": 1.975347763022043e-05, "loss": 0.365, "step": 1459 }, { "epoch": 1.48, "learning_rate": 1.975299364325276e-05, "loss": 0.3739, "step": 1460 }, { "epoch": 1.48, "learning_rate": 1.9752509187596164e-05, "loss": 0.3576, "step": 1461 }, { "epoch": 1.48, "learning_rate": 1.9752024263273922e-05, "loss": 0.3778, "step": 1462 }, { "epoch": 1.48, "learning_rate": 1.9751538870309338e-05, "loss": 0.3623, "step": 1463 }, { "epoch": 1.49, "learning_rate": 1.9751053008725736e-05, "loss": 0.3495, "step": 1464 }, { "epoch": 1.49, "learning_rate": 1.9750566678546473e-05, "loss": 0.3774, "step": 1465 }, { "epoch": 1.49, "learning_rate": 1.9750079879794906e-05, "loss": 0.3915, "step": 1466 }, { "epoch": 1.49, "learning_rate": 1.974959261249444e-05, "loss": 0.3909, "step": 1467 }, { "epoch": 1.49, "learning_rate": 1.9749104876668483e-05, "loss": 0.3655, "step": 1468 }, { "epoch": 1.49, "learning_rate": 1.9748616672340482e-05, "loss": 0.3921, "step": 1469 }, { "epoch": 1.49, "learning_rate": 1.9748127999533893e-05, "loss": 0.3901, "step": 1470 }, { "epoch": 1.49, "learning_rate": 1.97476388582722e-05, "loss": 0.3609, "step": 1471 }, { "epoch": 1.49, "learning_rate": 1.9747149248578908e-05, "loss": 0.3874, "step": 1472 }, { "epoch": 1.49, "learning_rate": 1.9746659170477547e-05, "loss": 0.375, "step": 1473 }, { "epoch": 1.5, "learning_rate": 1.974616862399167e-05, "loss": 0.3614, "step": 1474 }, { "epoch": 1.5, "learning_rate": 1.974567760914485e-05, "loss": 0.3919, "step": 1475 }, { "epoch": 1.5, "learning_rate": 1.974518612596068e-05, "loss": 0.3799, "step": 1476 }, { "epoch": 1.5, "learning_rate": 1.974469417446278e-05, "loss": 0.3687, "step": 1477 }, { "epoch": 1.5, "learning_rate": 1.9744201754674795e-05, "loss": 0.3531, "step": 1478 }, { "epoch": 1.5, "learning_rate": 1.9743708866620385e-05, "loss": 0.3727, "step": 1479 }, { "epoch": 1.5, "learning_rate": 1.974321551032323e-05, "loss": 0.3391, "step": 1480 }, { "epoch": 1.5, "learning_rate": 1.9742721685807057e-05, "loss": 0.4337, "step": 1481 }, { "epoch": 1.5, "learning_rate": 1.974222739309558e-05, "loss": 0.3838, "step": 1482 }, { "epoch": 1.5, "learning_rate": 1.9741732632212556e-05, "loss": 0.337, "step": 1483 }, { "epoch": 1.51, "learning_rate": 1.9741237403181765e-05, "loss": 0.3713, "step": 1484 }, { "epoch": 1.51, "learning_rate": 1.9740741706027004e-05, "loss": 0.4051, "step": 1485 }, { "epoch": 1.51, "learning_rate": 1.9740245540772096e-05, "loss": 0.3792, "step": 1486 }, { "epoch": 1.51, "learning_rate": 1.9739748907440882e-05, "loss": 0.3498, "step": 1487 }, { "epoch": 1.51, "learning_rate": 1.973925180605723e-05, "loss": 0.3734, "step": 1488 }, { "epoch": 1.51, "learning_rate": 1.973875423664503e-05, "loss": 0.4013, "step": 1489 }, { "epoch": 1.51, "learning_rate": 1.9738256199228187e-05, "loss": 0.3509, "step": 1490 }, { "epoch": 1.51, "learning_rate": 1.973775769383064e-05, "loss": 0.3542, "step": 1491 }, { "epoch": 1.51, "learning_rate": 1.9737258720476342e-05, "loss": 0.3498, "step": 1492 }, { "epoch": 1.51, "learning_rate": 1.9736759279189278e-05, "loss": 0.3971, "step": 1493 }, { "epoch": 1.52, "learning_rate": 1.973625936999344e-05, "loss": 0.3918, "step": 1494 }, { "epoch": 1.52, "learning_rate": 1.9735758992912858e-05, "loss": 0.3702, "step": 1495 }, { "epoch": 1.52, "learning_rate": 1.9735258147971573e-05, "loss": 0.3586, "step": 1496 }, { "epoch": 1.52, "learning_rate": 1.9734756835193662e-05, "loss": 0.3818, "step": 1497 }, { "epoch": 1.52, "learning_rate": 1.9734255054603212e-05, "loss": 0.3535, "step": 1498 }, { "epoch": 1.52, "learning_rate": 1.973375280622433e-05, "loss": 0.3799, "step": 1499 }, { "epoch": 1.52, "learning_rate": 1.9733250090081157e-05, "loss": 0.4023, "step": 1500 }, { "epoch": 1.52, "learning_rate": 1.9732746906197854e-05, "loss": 0.3538, "step": 1501 }, { "epoch": 1.52, "learning_rate": 1.9732243254598603e-05, "loss": 0.3669, "step": 1502 }, { "epoch": 1.53, "learning_rate": 1.9731739135307597e-05, "loss": 0.3828, "step": 1503 }, { "epoch": 1.53, "learning_rate": 1.9731234548349074e-05, "loss": 0.3503, "step": 1504 }, { "epoch": 1.53, "learning_rate": 1.9730729493747275e-05, "loss": 0.3549, "step": 1505 }, { "epoch": 1.53, "learning_rate": 1.9730223971526478e-05, "loss": 0.3472, "step": 1506 }, { "epoch": 1.53, "learning_rate": 1.9729717981710964e-05, "loss": 0.3623, "step": 1507 }, { "epoch": 1.53, "learning_rate": 1.9729211524325064e-05, "loss": 0.3633, "step": 1508 }, { "epoch": 1.53, "learning_rate": 1.97287045993931e-05, "loss": 0.3788, "step": 1509 }, { "epoch": 1.53, "learning_rate": 1.972819720693945e-05, "loss": 0.3845, "step": 1510 }, { "epoch": 1.53, "learning_rate": 1.9727689346988484e-05, "loss": 0.3635, "step": 1511 }, { "epoch": 1.53, "learning_rate": 1.9727181019564616e-05, "loss": 0.4043, "step": 1512 }, { "epoch": 1.54, "learning_rate": 1.9726672224692266e-05, "loss": 0.3436, "step": 1513 }, { "epoch": 1.54, "learning_rate": 1.972616296239589e-05, "loss": 0.381, "step": 1514 }, { "epoch": 1.54, "learning_rate": 1.9725653232699962e-05, "loss": 0.3232, "step": 1515 }, { "epoch": 1.54, "learning_rate": 1.9725143035628977e-05, "loss": 0.4574, "step": 1516 }, { "epoch": 1.54, "learning_rate": 1.9724632371207447e-05, "loss": 0.3607, "step": 1517 }, { "epoch": 1.54, "learning_rate": 1.972412123945992e-05, "loss": 0.3773, "step": 1518 }, { "epoch": 1.54, "learning_rate": 1.9723609640410956e-05, "loss": 0.3588, "step": 1519 }, { "epoch": 1.54, "learning_rate": 1.9723097574085138e-05, "loss": 0.3935, "step": 1520 }, { "epoch": 1.54, "learning_rate": 1.972258504050708e-05, "loss": 0.4453, "step": 1521 }, { "epoch": 1.54, "learning_rate": 1.9722072039701405e-05, "loss": 0.4191, "step": 1522 }, { "epoch": 1.55, "learning_rate": 1.9721558571692768e-05, "loss": 0.3898, "step": 1523 }, { "epoch": 1.55, "learning_rate": 1.972104463650585e-05, "loss": 0.3528, "step": 1524 }, { "epoch": 1.55, "learning_rate": 1.972053023416534e-05, "loss": 0.3856, "step": 1525 }, { "epoch": 1.55, "learning_rate": 1.9720015364695963e-05, "loss": 0.3332, "step": 1526 }, { "epoch": 1.55, "learning_rate": 1.9719500028122463e-05, "loss": 0.3872, "step": 1527 }, { "epoch": 1.55, "learning_rate": 1.9718984224469598e-05, "loss": 0.364, "step": 1528 }, { "epoch": 1.55, "learning_rate": 1.9718467953762162e-05, "loss": 0.3653, "step": 1529 }, { "epoch": 1.55, "learning_rate": 1.9717951216024965e-05, "loss": 0.3755, "step": 1530 }, { "epoch": 1.55, "learning_rate": 1.9717434011282833e-05, "loss": 0.3522, "step": 1531 }, { "epoch": 1.55, "learning_rate": 1.9716916339560626e-05, "loss": 0.3753, "step": 1532 }, { "epoch": 1.56, "learning_rate": 1.971639820088322e-05, "loss": 0.3751, "step": 1533 }, { "epoch": 1.56, "learning_rate": 1.9715879595275512e-05, "loss": 0.377, "step": 1534 }, { "epoch": 1.56, "learning_rate": 1.971536052276243e-05, "loss": 0.3649, "step": 1535 }, { "epoch": 1.56, "learning_rate": 1.9714840983368913e-05, "loss": 0.4018, "step": 1536 }, { "epoch": 1.56, "learning_rate": 1.9714320977119935e-05, "loss": 0.3694, "step": 1537 }, { "epoch": 1.56, "learning_rate": 1.9713800504040478e-05, "loss": 0.3685, "step": 1538 }, { "epoch": 1.56, "learning_rate": 1.9713279564155553e-05, "loss": 0.2946, "step": 1539 }, { "epoch": 1.56, "learning_rate": 1.97127581574902e-05, "loss": 0.3963, "step": 1540 }, { "epoch": 1.56, "learning_rate": 1.9712236284069473e-05, "loss": 0.3438, "step": 1541 }, { "epoch": 1.56, "learning_rate": 1.9711713943918448e-05, "loss": 0.3733, "step": 1542 }, { "epoch": 1.57, "learning_rate": 1.9711191137062237e-05, "loss": 0.3784, "step": 1543 }, { "epoch": 1.57, "learning_rate": 1.971066786352595e-05, "loss": 0.3437, "step": 1544 }, { "epoch": 1.57, "learning_rate": 1.971014412333474e-05, "loss": 0.4407, "step": 1545 }, { "epoch": 1.57, "learning_rate": 1.970961991651378e-05, "loss": 0.4324, "step": 1546 }, { "epoch": 1.57, "learning_rate": 1.9709095243088255e-05, "loss": 0.3796, "step": 1547 }, { "epoch": 1.57, "learning_rate": 1.970857010308338e-05, "loss": 0.3727, "step": 1548 }, { "epoch": 1.57, "learning_rate": 1.9708044496524394e-05, "loss": 0.4083, "step": 1549 }, { "epoch": 1.57, "learning_rate": 1.970751842343655e-05, "loss": 0.3373, "step": 1550 }, { "epoch": 1.57, "learning_rate": 1.9706991883845136e-05, "loss": 0.4022, "step": 1551 }, { "epoch": 1.57, "learning_rate": 1.9706464877775452e-05, "loss": 0.3802, "step": 1552 }, { "epoch": 1.58, "learning_rate": 1.970593740525282e-05, "loss": 0.3627, "step": 1553 }, { "epoch": 1.58, "learning_rate": 1.9705409466302592e-05, "loss": 0.4069, "step": 1554 }, { "epoch": 1.58, "learning_rate": 1.970488106095014e-05, "loss": 0.3712, "step": 1555 }, { "epoch": 1.58, "learning_rate": 1.970435218922085e-05, "loss": 0.3811, "step": 1556 }, { "epoch": 1.58, "learning_rate": 1.9703822851140152e-05, "loss": 0.4063, "step": 1557 }, { "epoch": 1.58, "learning_rate": 1.9703293046733466e-05, "loss": 0.3863, "step": 1558 }, { "epoch": 1.58, "learning_rate": 1.9702762776026267e-05, "loss": 0.3957, "step": 1559 }, { "epoch": 1.58, "learning_rate": 1.9702232039044024e-05, "loss": 0.4493, "step": 1560 }, { "epoch": 1.58, "learning_rate": 1.9701700835812252e-05, "loss": 0.3294, "step": 1561 }, { "epoch": 1.58, "learning_rate": 1.970116916635648e-05, "loss": 0.4028, "step": 1562 }, { "epoch": 1.59, "learning_rate": 1.970063703070225e-05, "loss": 0.3358, "step": 1563 }, { "epoch": 1.59, "learning_rate": 1.970010442887514e-05, "loss": 0.3821, "step": 1564 }, { "epoch": 1.59, "learning_rate": 1.9699571360900736e-05, "loss": 0.3894, "step": 1565 }, { "epoch": 1.59, "learning_rate": 1.969903782680467e-05, "loss": 0.3588, "step": 1566 }, { "epoch": 1.59, "learning_rate": 1.9698503826612568e-05, "loss": 0.3951, "step": 1567 }, { "epoch": 1.59, "learning_rate": 1.9697969360350098e-05, "loss": 0.4159, "step": 1568 }, { "epoch": 1.59, "learning_rate": 1.969743442804294e-05, "loss": 0.3588, "step": 1569 }, { "epoch": 1.59, "learning_rate": 1.969689902971681e-05, "loss": 0.3293, "step": 1570 }, { "epoch": 1.59, "learning_rate": 1.9696363165397425e-05, "loss": 0.3384, "step": 1571 }, { "epoch": 1.6, "learning_rate": 1.9695826835110545e-05, "loss": 0.3298, "step": 1572 }, { "epoch": 1.6, "learning_rate": 1.969529003888194e-05, "loss": 0.3544, "step": 1573 }, { "epoch": 1.6, "learning_rate": 1.969475277673741e-05, "loss": 0.389, "step": 1574 }, { "epoch": 1.6, "learning_rate": 1.9694215048702768e-05, "loss": 0.3338, "step": 1575 }, { "epoch": 1.6, "learning_rate": 1.969367685480386e-05, "loss": 0.3797, "step": 1576 }, { "epoch": 1.6, "learning_rate": 1.9693138195066548e-05, "loss": 0.409, "step": 1577 }, { "epoch": 1.6, "learning_rate": 1.9692599069516715e-05, "loss": 0.4053, "step": 1578 }, { "epoch": 1.6, "learning_rate": 1.9692059478180272e-05, "loss": 0.3966, "step": 1579 }, { "epoch": 1.6, "learning_rate": 1.9691519421083146e-05, "loss": 0.3618, "step": 1580 }, { "epoch": 1.6, "learning_rate": 1.96909788982513e-05, "loss": 0.3623, "step": 1581 }, { "epoch": 1.61, "learning_rate": 1.9690437909710697e-05, "loss": 0.3539, "step": 1582 }, { "epoch": 1.61, "learning_rate": 1.9689896455487338e-05, "loss": 0.3807, "step": 1583 }, { "epoch": 1.61, "learning_rate": 1.968935453560725e-05, "loss": 0.3628, "step": 1584 }, { "epoch": 1.61, "learning_rate": 1.9688812150096467e-05, "loss": 0.3338, "step": 1585 }, { "epoch": 1.61, "learning_rate": 1.968826929898106e-05, "loss": 0.3208, "step": 1586 }, { "epoch": 1.61, "learning_rate": 1.968772598228711e-05, "loss": 0.3603, "step": 1587 }, { "epoch": 1.61, "learning_rate": 1.968718220004073e-05, "loss": 0.3427, "step": 1588 }, { "epoch": 1.61, "learning_rate": 1.9686637952268053e-05, "loss": 0.3387, "step": 1589 }, { "epoch": 1.61, "learning_rate": 1.968609323899523e-05, "loss": 0.3471, "step": 1590 }, { "epoch": 1.61, "learning_rate": 1.9685548060248445e-05, "loss": 0.4405, "step": 1591 }, { "epoch": 1.62, "learning_rate": 1.9685002416053887e-05, "loss": 0.366, "step": 1592 }, { "epoch": 1.62, "learning_rate": 1.9684456306437788e-05, "loss": 0.3673, "step": 1593 }, { "epoch": 1.62, "learning_rate": 1.968390973142638e-05, "loss": 0.373, "step": 1594 }, { "epoch": 1.62, "learning_rate": 1.9683362691045936e-05, "loss": 0.3671, "step": 1595 }, { "epoch": 1.62, "learning_rate": 1.9682815185322747e-05, "loss": 0.3952, "step": 1596 }, { "epoch": 1.62, "learning_rate": 1.968226721428312e-05, "loss": 0.3624, "step": 1597 }, { "epoch": 1.62, "learning_rate": 1.9681718777953386e-05, "loss": 0.3957, "step": 1598 }, { "epoch": 1.62, "learning_rate": 1.9681169876359907e-05, "loss": 0.3484, "step": 1599 }, { "epoch": 1.62, "learning_rate": 1.968062050952906e-05, "loss": 0.3874, "step": 1600 }, { "epoch": 1.62, "learning_rate": 1.9680070677487237e-05, "loss": 0.4, "step": 1601 }, { "epoch": 1.63, "learning_rate": 1.9679520380260868e-05, "loss": 0.3658, "step": 1602 }, { "epoch": 1.63, "learning_rate": 1.9678969617876397e-05, "loss": 0.319, "step": 1603 }, { "epoch": 1.63, "learning_rate": 1.967841839036029e-05, "loss": 0.356, "step": 1604 }, { "epoch": 1.63, "learning_rate": 1.967786669773904e-05, "loss": 0.3544, "step": 1605 }, { "epoch": 1.63, "learning_rate": 1.9677314540039158e-05, "loss": 0.4166, "step": 1606 }, { "epoch": 1.63, "learning_rate": 1.9676761917287172e-05, "loss": 0.3384, "step": 1607 }, { "epoch": 1.63, "learning_rate": 1.9676208829509647e-05, "loss": 0.3641, "step": 1608 }, { "epoch": 1.63, "learning_rate": 1.9675655276733157e-05, "loss": 0.3972, "step": 1609 }, { "epoch": 1.63, "learning_rate": 1.9675101258984308e-05, "loss": 0.3704, "step": 1610 }, { "epoch": 1.63, "learning_rate": 1.9674546776289723e-05, "loss": 0.3695, "step": 1611 }, { "epoch": 1.64, "learning_rate": 1.9673991828676043e-05, "loss": 0.3452, "step": 1612 }, { "epoch": 1.64, "learning_rate": 1.9673436416169942e-05, "loss": 0.4009, "step": 1613 }, { "epoch": 1.64, "learning_rate": 1.967288053879811e-05, "loss": 0.3195, "step": 1614 }, { "epoch": 1.64, "learning_rate": 1.9672324196587257e-05, "loss": 0.3921, "step": 1615 }, { "epoch": 1.64, "learning_rate": 1.967176738956412e-05, "loss": 0.3691, "step": 1616 }, { "epoch": 1.64, "learning_rate": 1.967121011775546e-05, "loss": 0.3694, "step": 1617 }, { "epoch": 1.64, "learning_rate": 1.9670652381188052e-05, "loss": 0.3244, "step": 1618 }, { "epoch": 1.64, "learning_rate": 1.9670094179888704e-05, "loss": 0.4034, "step": 1619 }, { "epoch": 1.64, "learning_rate": 1.9669535513884234e-05, "loss": 0.3752, "step": 1620 }, { "epoch": 1.64, "learning_rate": 1.9668976383201495e-05, "loss": 0.329, "step": 1621 }, { "epoch": 1.65, "learning_rate": 1.966841678786736e-05, "loss": 0.3643, "step": 1622 }, { "epoch": 1.65, "learning_rate": 1.966785672790871e-05, "loss": 0.3971, "step": 1623 }, { "epoch": 1.65, "learning_rate": 1.9667296203352468e-05, "loss": 0.37, "step": 1624 }, { "epoch": 1.65, "learning_rate": 1.9666735214225564e-05, "loss": 0.3451, "step": 1625 }, { "epoch": 1.65, "learning_rate": 1.966617376055496e-05, "loss": 0.4033, "step": 1626 }, { "epoch": 1.65, "learning_rate": 1.9665611842367643e-05, "loss": 0.4056, "step": 1627 }, { "epoch": 1.65, "learning_rate": 1.9665049459690608e-05, "loss": 0.3053, "step": 1628 }, { "epoch": 1.65, "learning_rate": 1.9664486612550883e-05, "loss": 0.3347, "step": 1629 }, { "epoch": 1.65, "learning_rate": 1.9663923300975515e-05, "loss": 0.3675, "step": 1630 }, { "epoch": 1.65, "learning_rate": 1.966335952499158e-05, "loss": 0.3773, "step": 1631 }, { "epoch": 1.66, "learning_rate": 1.9662795284626168e-05, "loss": 0.3956, "step": 1632 }, { "epoch": 1.66, "learning_rate": 1.9662230579906388e-05, "loss": 0.3378, "step": 1633 }, { "epoch": 1.66, "learning_rate": 1.966166541085938e-05, "loss": 0.419, "step": 1634 }, { "epoch": 1.66, "learning_rate": 1.9661099777512316e-05, "loss": 0.4406, "step": 1635 }, { "epoch": 1.66, "learning_rate": 1.966053367989236e-05, "loss": 0.394, "step": 1636 }, { "epoch": 1.66, "learning_rate": 1.965996711802673e-05, "loss": 0.41, "step": 1637 }, { "epoch": 1.66, "learning_rate": 1.9659400091942645e-05, "loss": 0.3965, "step": 1638 }, { "epoch": 1.66, "learning_rate": 1.9658832601667357e-05, "loss": 0.3638, "step": 1639 }, { "epoch": 1.66, "learning_rate": 1.9658264647228137e-05, "loss": 0.3635, "step": 1640 }, { "epoch": 1.67, "learning_rate": 1.9657696228652277e-05, "loss": 0.3402, "step": 1641 }, { "epoch": 1.67, "learning_rate": 1.9657127345967093e-05, "loss": 0.3804, "step": 1642 }, { "epoch": 1.67, "learning_rate": 1.9656557999199926e-05, "loss": 0.3943, "step": 1643 }, { "epoch": 1.67, "learning_rate": 1.9655988188378132e-05, "loss": 0.3992, "step": 1644 }, { "epoch": 1.67, "learning_rate": 1.9655417913529097e-05, "loss": 0.3307, "step": 1645 }, { "epoch": 1.67, "learning_rate": 1.9654847174680226e-05, "loss": 0.3027, "step": 1646 }, { "epoch": 1.67, "learning_rate": 1.9654275971858947e-05, "loss": 0.3558, "step": 1647 }, { "epoch": 1.67, "learning_rate": 1.9653704305092708e-05, "loss": 0.3401, "step": 1648 }, { "epoch": 1.67, "learning_rate": 1.965313217440898e-05, "loss": 0.3714, "step": 1649 }, { "epoch": 1.67, "learning_rate": 1.965255957983526e-05, "loss": 0.3814, "step": 1650 }, { "epoch": 1.68, "learning_rate": 1.965198652139906e-05, "loss": 0.3944, "step": 1651 }, { "epoch": 1.68, "learning_rate": 1.9651412999127927e-05, "loss": 0.4188, "step": 1652 }, { "epoch": 1.68, "learning_rate": 1.9650839013049415e-05, "loss": 0.3488, "step": 1653 }, { "epoch": 1.68, "learning_rate": 1.9650264563191108e-05, "loss": 0.3964, "step": 1654 }, { "epoch": 1.68, "learning_rate": 1.9649689649580615e-05, "loss": 0.3716, "step": 1655 }, { "epoch": 1.68, "learning_rate": 1.964911427224556e-05, "loss": 0.3956, "step": 1656 }, { "epoch": 1.68, "learning_rate": 1.9648538431213598e-05, "loss": 0.3309, "step": 1657 }, { "epoch": 1.68, "learning_rate": 1.96479621265124e-05, "loss": 0.417, "step": 1658 }, { "epoch": 1.68, "learning_rate": 1.9647385358169657e-05, "loss": 0.3354, "step": 1659 }, { "epoch": 1.68, "learning_rate": 1.964680812621309e-05, "loss": 0.3323, "step": 1660 }, { "epoch": 1.69, "learning_rate": 1.964623043067044e-05, "loss": 0.4154, "step": 1661 }, { "epoch": 1.69, "learning_rate": 1.9645652271569462e-05, "loss": 0.3494, "step": 1662 }, { "epoch": 1.69, "learning_rate": 1.9645073648937947e-05, "loss": 0.3708, "step": 1663 }, { "epoch": 1.69, "learning_rate": 1.9644494562803698e-05, "loss": 0.3595, "step": 1664 }, { "epoch": 1.69, "learning_rate": 1.964391501319454e-05, "loss": 0.3606, "step": 1665 }, { "epoch": 1.69, "learning_rate": 1.9643335000138332e-05, "loss": 0.383, "step": 1666 }, { "epoch": 1.69, "learning_rate": 1.9642754523662943e-05, "loss": 0.3292, "step": 1667 }, { "epoch": 1.69, "learning_rate": 1.9642173583796265e-05, "loss": 0.45, "step": 1668 }, { "epoch": 1.69, "learning_rate": 1.964159218056622e-05, "loss": 0.3774, "step": 1669 }, { "epoch": 1.69, "learning_rate": 1.964101031400075e-05, "loss": 0.3863, "step": 1670 }, { "epoch": 1.7, "learning_rate": 1.964042798412781e-05, "loss": 0.3655, "step": 1671 }, { "epoch": 1.7, "learning_rate": 1.9639845190975387e-05, "loss": 0.3288, "step": 1672 }, { "epoch": 1.7, "learning_rate": 1.963926193457149e-05, "loss": 0.3946, "step": 1673 }, { "epoch": 1.7, "learning_rate": 1.9638678214944143e-05, "loss": 0.3296, "step": 1674 }, { "epoch": 1.7, "learning_rate": 1.9638094032121405e-05, "loss": 0.4467, "step": 1675 }, { "epoch": 1.7, "learning_rate": 1.9637509386131347e-05, "loss": 0.375, "step": 1676 }, { "epoch": 1.7, "learning_rate": 1.963692427700206e-05, "loss": 0.3186, "step": 1677 }, { "epoch": 1.7, "learning_rate": 1.9636338704761666e-05, "loss": 0.3803, "step": 1678 }, { "epoch": 1.7, "learning_rate": 1.96357526694383e-05, "loss": 0.355, "step": 1679 }, { "epoch": 1.7, "learning_rate": 1.9635166171060135e-05, "loss": 0.3363, "step": 1680 }, { "epoch": 1.71, "learning_rate": 1.963457920965534e-05, "loss": 0.3863, "step": 1681 }, { "epoch": 1.71, "learning_rate": 1.963399178525214e-05, "loss": 0.3758, "step": 1682 }, { "epoch": 1.71, "learning_rate": 1.9633403897878752e-05, "loss": 0.4247, "step": 1683 }, { "epoch": 1.71, "learning_rate": 1.963281554756343e-05, "loss": 0.3502, "step": 1684 }, { "epoch": 1.71, "learning_rate": 1.963222673433445e-05, "loss": 0.3575, "step": 1685 }, { "epoch": 1.71, "learning_rate": 1.9631637458220106e-05, "loss": 0.3937, "step": 1686 }, { "epoch": 1.71, "learning_rate": 1.9631047719248717e-05, "loss": 0.3486, "step": 1687 }, { "epoch": 1.71, "learning_rate": 1.9630457517448623e-05, "loss": 0.3522, "step": 1688 }, { "epoch": 1.71, "learning_rate": 1.9629866852848184e-05, "loss": 0.401, "step": 1689 }, { "epoch": 1.71, "learning_rate": 1.962927572547579e-05, "loss": 0.3745, "step": 1690 }, { "epoch": 1.72, "learning_rate": 1.9628684135359846e-05, "loss": 0.3756, "step": 1691 }, { "epoch": 1.72, "learning_rate": 1.9628092082528782e-05, "loss": 0.4274, "step": 1692 }, { "epoch": 1.72, "learning_rate": 1.9627499567011047e-05, "loss": 0.3471, "step": 1693 }, { "epoch": 1.72, "learning_rate": 1.9626906588835116e-05, "loss": 0.4074, "step": 1694 }, { "epoch": 1.72, "learning_rate": 1.962631314802949e-05, "loss": 0.3465, "step": 1695 }, { "epoch": 1.72, "learning_rate": 1.962571924462268e-05, "loss": 0.3676, "step": 1696 }, { "epoch": 1.72, "learning_rate": 1.9625124878643233e-05, "loss": 0.3593, "step": 1697 }, { "epoch": 1.72, "learning_rate": 1.9624530050119706e-05, "loss": 0.4124, "step": 1698 }, { "epoch": 1.72, "learning_rate": 1.9623934759080685e-05, "loss": 0.4103, "step": 1699 }, { "epoch": 1.73, "learning_rate": 1.962333900555478e-05, "loss": 0.4043, "step": 1700 }, { "epoch": 1.73, "learning_rate": 1.9622742789570618e-05, "loss": 0.389, "step": 1701 }, { "epoch": 1.73, "learning_rate": 1.9622146111156858e-05, "loss": 0.3249, "step": 1702 }, { "epoch": 1.73, "learning_rate": 1.962154897034216e-05, "loss": 0.3029, "step": 1703 }, { "epoch": 1.73, "learning_rate": 1.9620951367155233e-05, "loss": 0.4225, "step": 1704 }, { "epoch": 1.73, "learning_rate": 1.962035330162479e-05, "loss": 0.367, "step": 1705 }, { "epoch": 1.73, "learning_rate": 1.9619754773779572e-05, "loss": 0.3751, "step": 1706 }, { "epoch": 1.73, "learning_rate": 1.9619155783648337e-05, "loss": 0.354, "step": 1707 }, { "epoch": 1.73, "learning_rate": 1.9618556331259883e-05, "loss": 0.3974, "step": 1708 }, { "epoch": 1.73, "learning_rate": 1.9617956416643e-05, "loss": 0.3256, "step": 1709 }, { "epoch": 1.74, "learning_rate": 1.9617356039826533e-05, "loss": 0.3901, "step": 1710 }, { "epoch": 1.74, "learning_rate": 1.9616755200839327e-05, "loss": 0.406, "step": 1711 }, { "epoch": 1.74, "learning_rate": 1.9616153899710253e-05, "loss": 0.3692, "step": 1712 }, { "epoch": 1.74, "learning_rate": 1.961555213646821e-05, "loss": 0.3872, "step": 1713 }, { "epoch": 1.74, "learning_rate": 1.9614949911142118e-05, "loss": 0.3952, "step": 1714 }, { "epoch": 1.74, "learning_rate": 1.9614347223760915e-05, "loss": 0.3638, "step": 1715 }, { "epoch": 1.74, "learning_rate": 1.9613744074353564e-05, "loss": 0.3556, "step": 1716 }, { "epoch": 1.74, "learning_rate": 1.9613140462949052e-05, "loss": 0.3695, "step": 1717 }, { "epoch": 1.74, "learning_rate": 1.9612536389576383e-05, "loss": 0.385, "step": 1718 }, { "epoch": 1.74, "learning_rate": 1.961193185426459e-05, "loss": 0.4128, "step": 1719 }, { "epoch": 1.75, "learning_rate": 1.961132685704272e-05, "loss": 0.3351, "step": 1720 }, { "epoch": 1.75, "learning_rate": 1.9610721397939852e-05, "loss": 0.3065, "step": 1721 }, { "epoch": 1.75, "learning_rate": 1.9610115476985074e-05, "loss": 0.3611, "step": 1722 }, { "epoch": 1.75, "learning_rate": 1.9609509094207513e-05, "loss": 0.4011, "step": 1723 }, { "epoch": 1.75, "learning_rate": 1.96089022496363e-05, "loss": 0.3733, "step": 1724 }, { "epoch": 1.75, "learning_rate": 1.960829494330061e-05, "loss": 0.4146, "step": 1725 }, { "epoch": 1.75, "learning_rate": 1.9607687175229615e-05, "loss": 0.3807, "step": 1726 }, { "epoch": 1.75, "learning_rate": 1.9607078945452528e-05, "loss": 0.3358, "step": 1727 }, { "epoch": 1.75, "learning_rate": 1.9606470253998578e-05, "loss": 0.3475, "step": 1728 }, { "epoch": 1.75, "learning_rate": 1.9605861100897018e-05, "loss": 0.3547, "step": 1729 }, { "epoch": 1.76, "learning_rate": 1.9605251486177114e-05, "loss": 0.3448, "step": 1730 }, { "epoch": 1.76, "learning_rate": 1.960464140986817e-05, "loss": 0.4219, "step": 1731 }, { "epoch": 1.76, "learning_rate": 1.9604030871999497e-05, "loss": 0.3143, "step": 1732 }, { "epoch": 1.76, "learning_rate": 1.960341987260044e-05, "loss": 0.3887, "step": 1733 }, { "epoch": 1.76, "learning_rate": 1.960280841170036e-05, "loss": 0.4144, "step": 1734 }, { "epoch": 1.76, "learning_rate": 1.960219648932864e-05, "loss": 0.374, "step": 1735 }, { "epoch": 1.76, "learning_rate": 1.9601584105514683e-05, "loss": 0.3385, "step": 1736 }, { "epoch": 1.76, "learning_rate": 1.9600971260287927e-05, "loss": 0.3635, "step": 1737 }, { "epoch": 1.76, "learning_rate": 1.9600357953677816e-05, "loss": 0.3095, "step": 1738 }, { "epoch": 1.76, "learning_rate": 1.9599744185713825e-05, "loss": 0.3291, "step": 1739 }, { "epoch": 1.77, "learning_rate": 1.9599129956425446e-05, "loss": 0.3977, "step": 1740 }, { "epoch": 1.77, "learning_rate": 1.9598515265842202e-05, "loss": 0.416, "step": 1741 }, { "epoch": 1.77, "learning_rate": 1.9597900113993627e-05, "loss": 0.3452, "step": 1742 }, { "epoch": 1.77, "learning_rate": 1.9597284500909287e-05, "loss": 0.4154, "step": 1743 }, { "epoch": 1.77, "learning_rate": 1.9596668426618766e-05, "loss": 0.3929, "step": 1744 }, { "epoch": 1.77, "learning_rate": 1.9596051891151668e-05, "loss": 0.3397, "step": 1745 }, { "epoch": 1.77, "learning_rate": 1.9595434894537616e-05, "loss": 0.3373, "step": 1746 }, { "epoch": 1.77, "learning_rate": 1.959481743680627e-05, "loss": 0.3104, "step": 1747 }, { "epoch": 1.77, "learning_rate": 1.9594199517987295e-05, "loss": 0.3327, "step": 1748 }, { "epoch": 1.77, "learning_rate": 1.9593581138110392e-05, "loss": 0.3492, "step": 1749 }, { "epoch": 1.78, "learning_rate": 1.9592962297205274e-05, "loss": 0.4026, "step": 1750 }, { "epoch": 1.78, "learning_rate": 1.9592342995301677e-05, "loss": 0.3516, "step": 1751 }, { "epoch": 1.78, "learning_rate": 1.9591723232429368e-05, "loss": 0.3339, "step": 1752 }, { "epoch": 1.78, "learning_rate": 1.9591103008618128e-05, "loss": 0.418, "step": 1753 }, { "epoch": 1.78, "learning_rate": 1.959048232389776e-05, "loss": 0.3951, "step": 1754 }, { "epoch": 1.78, "learning_rate": 1.95898611782981e-05, "loss": 0.3835, "step": 1755 }, { "epoch": 1.78, "learning_rate": 1.9589239571848987e-05, "loss": 0.3907, "step": 1756 }, { "epoch": 1.78, "learning_rate": 1.9588617504580295e-05, "loss": 0.3769, "step": 1757 }, { "epoch": 1.78, "learning_rate": 1.9587994976521923e-05, "loss": 0.3409, "step": 1758 }, { "epoch": 1.78, "learning_rate": 1.9587371987703787e-05, "loss": 0.3973, "step": 1759 }, { "epoch": 1.79, "learning_rate": 1.9586748538155818e-05, "loss": 0.3544, "step": 1760 }, { "epoch": 1.79, "learning_rate": 1.9586124627907988e-05, "loss": 0.4034, "step": 1761 }, { "epoch": 1.79, "learning_rate": 1.9585500256990268e-05, "loss": 0.3987, "step": 1762 }, { "epoch": 1.79, "learning_rate": 1.9584875425432667e-05, "loss": 0.3481, "step": 1763 }, { "epoch": 1.79, "learning_rate": 1.9584250133265216e-05, "loss": 0.464, "step": 1764 }, { "epoch": 1.79, "learning_rate": 1.9583624380517956e-05, "loss": 0.3875, "step": 1765 }, { "epoch": 1.79, "learning_rate": 1.9582998167220967e-05, "loss": 0.3502, "step": 1766 }, { "epoch": 1.79, "learning_rate": 1.9582371493404333e-05, "loss": 0.3591, "step": 1767 }, { "epoch": 1.79, "learning_rate": 1.9581744359098178e-05, "loss": 0.3933, "step": 1768 }, { "epoch": 1.8, "learning_rate": 1.9581116764332633e-05, "loss": 0.3577, "step": 1769 }, { "epoch": 1.8, "learning_rate": 1.958048870913786e-05, "loss": 0.3707, "step": 1770 }, { "epoch": 1.8, "learning_rate": 1.9579860193544046e-05, "loss": 0.3243, "step": 1771 }, { "epoch": 1.8, "learning_rate": 1.9579231217581384e-05, "loss": 0.3173, "step": 1772 }, { "epoch": 1.8, "learning_rate": 1.957860178128011e-05, "loss": 0.3868, "step": 1773 }, { "epoch": 1.8, "learning_rate": 1.9577971884670465e-05, "loss": 0.4181, "step": 1774 }, { "epoch": 1.8, "learning_rate": 1.957734152778272e-05, "loss": 0.3493, "step": 1775 }, { "epoch": 1.8, "learning_rate": 1.9576710710647177e-05, "loss": 0.3312, "step": 1776 }, { "epoch": 1.8, "learning_rate": 1.957607943329414e-05, "loss": 0.3355, "step": 1777 }, { "epoch": 1.8, "learning_rate": 1.9575447695753948e-05, "loss": 0.3498, "step": 1778 }, { "epoch": 1.81, "learning_rate": 1.957481549805696e-05, "loss": 0.337, "step": 1779 }, { "epoch": 1.81, "learning_rate": 1.9574182840233557e-05, "loss": 0.3285, "step": 1780 }, { "epoch": 1.81, "learning_rate": 1.9573549722314145e-05, "loss": 0.3392, "step": 1781 }, { "epoch": 1.81, "learning_rate": 1.9572916144329143e-05, "loss": 0.3588, "step": 1782 }, { "epoch": 1.81, "learning_rate": 1.9572282106309004e-05, "loss": 0.3887, "step": 1783 }, { "epoch": 1.81, "learning_rate": 1.957164760828419e-05, "loss": 0.3476, "step": 1784 }, { "epoch": 1.81, "learning_rate": 1.9571012650285204e-05, "loss": 0.3306, "step": 1785 }, { "epoch": 1.81, "learning_rate": 1.957037723234255e-05, "loss": 0.3053, "step": 1786 }, { "epoch": 1.81, "learning_rate": 1.9569741354486767e-05, "loss": 0.3453, "step": 1787 }, { "epoch": 1.81, "learning_rate": 1.956910501674841e-05, "loss": 0.3683, "step": 1788 }, { "epoch": 1.82, "learning_rate": 1.956846821915806e-05, "loss": 0.3754, "step": 1789 }, { "epoch": 1.82, "learning_rate": 1.9567830961746322e-05, "loss": 0.3591, "step": 1790 }, { "epoch": 1.82, "learning_rate": 1.9567193244543815e-05, "loss": 0.3699, "step": 1791 }, { "epoch": 1.82, "learning_rate": 1.9566555067581188e-05, "loss": 0.3128, "step": 1792 }, { "epoch": 1.82, "learning_rate": 1.956591643088911e-05, "loss": 0.3932, "step": 1793 }, { "epoch": 1.82, "learning_rate": 1.9565277334498272e-05, "loss": 0.3646, "step": 1794 }, { "epoch": 1.82, "learning_rate": 1.9564637778439383e-05, "loss": 0.3597, "step": 1795 }, { "epoch": 1.82, "learning_rate": 1.9563997762743177e-05, "loss": 0.3727, "step": 1796 }, { "epoch": 1.82, "learning_rate": 1.956335728744041e-05, "loss": 0.4029, "step": 1797 }, { "epoch": 1.82, "learning_rate": 1.9562716352561867e-05, "loss": 0.4006, "step": 1798 }, { "epoch": 1.83, "learning_rate": 1.9562074958138343e-05, "loss": 0.3679, "step": 1799 }, { "epoch": 1.83, "learning_rate": 1.9561433104200664e-05, "loss": 0.3716, "step": 1800 }, { "epoch": 1.83, "learning_rate": 1.9560790790779673e-05, "loss": 0.3668, "step": 1801 }, { "epoch": 1.83, "learning_rate": 1.9560148017906236e-05, "loss": 0.362, "step": 1802 }, { "epoch": 1.83, "learning_rate": 1.9559504785611246e-05, "loss": 0.3732, "step": 1803 }, { "epoch": 1.83, "learning_rate": 1.955886109392561e-05, "loss": 0.3445, "step": 1804 }, { "epoch": 1.83, "learning_rate": 1.955821694288026e-05, "loss": 0.3903, "step": 1805 }, { "epoch": 1.83, "learning_rate": 1.9557572332506155e-05, "loss": 0.3731, "step": 1806 }, { "epoch": 1.83, "learning_rate": 1.9556927262834276e-05, "loss": 0.3991, "step": 1807 }, { "epoch": 1.83, "learning_rate": 1.9556281733895616e-05, "loss": 0.3025, "step": 1808 }, { "epoch": 1.84, "learning_rate": 1.9555635745721192e-05, "loss": 0.3902, "step": 1809 }, { "epoch": 1.84, "learning_rate": 1.9554989298342058e-05, "loss": 0.3597, "step": 1810 }, { "epoch": 1.84, "learning_rate": 1.9554342391789277e-05, "loss": 0.3512, "step": 1811 }, { "epoch": 1.84, "learning_rate": 1.955369502609393e-05, "loss": 0.365, "step": 1812 }, { "epoch": 1.84, "learning_rate": 1.9553047201287137e-05, "loss": 0.3419, "step": 1813 }, { "epoch": 1.84, "learning_rate": 1.955239891740002e-05, "loss": 0.3171, "step": 1814 }, { "epoch": 1.84, "learning_rate": 1.9551750174463743e-05, "loss": 0.4085, "step": 1815 }, { "epoch": 1.84, "learning_rate": 1.955110097250947e-05, "loss": 0.3634, "step": 1816 }, { "epoch": 1.84, "learning_rate": 1.955045131156841e-05, "loss": 0.3845, "step": 1817 }, { "epoch": 1.84, "learning_rate": 1.9549801191671775e-05, "loss": 0.3506, "step": 1818 }, { "epoch": 1.85, "learning_rate": 1.9549150612850813e-05, "loss": 0.4101, "step": 1819 }, { "epoch": 1.85, "learning_rate": 1.9548499575136786e-05, "loss": 0.362, "step": 1820 }, { "epoch": 1.85, "learning_rate": 1.9547848078560975e-05, "loss": 0.3817, "step": 1821 }, { "epoch": 1.85, "learning_rate": 1.9547196123154698e-05, "loss": 0.3687, "step": 1822 }, { "epoch": 1.85, "learning_rate": 1.9546543708949278e-05, "loss": 0.3393, "step": 1823 }, { "epoch": 1.85, "learning_rate": 1.9545890835976072e-05, "loss": 0.3365, "step": 1824 }, { "epoch": 1.85, "learning_rate": 1.954523750426645e-05, "loss": 0.3674, "step": 1825 }, { "epoch": 1.85, "learning_rate": 1.9544583713851808e-05, "loss": 0.3658, "step": 1826 }, { "epoch": 1.85, "learning_rate": 1.954392946476357e-05, "loss": 0.3373, "step": 1827 }, { "epoch": 1.85, "learning_rate": 1.9543274757033174e-05, "loss": 0.3939, "step": 1828 }, { "epoch": 1.86, "learning_rate": 1.9542619590692084e-05, "loss": 0.3471, "step": 1829 }, { "epoch": 1.86, "learning_rate": 1.954196396577178e-05, "loss": 0.3817, "step": 1830 }, { "epoch": 1.86, "learning_rate": 1.954130788230377e-05, "loss": 0.3816, "step": 1831 }, { "epoch": 1.86, "learning_rate": 1.9540651340319586e-05, "loss": 0.5132, "step": 1832 }, { "epoch": 1.86, "learning_rate": 1.9539994339850777e-05, "loss": 0.3788, "step": 1833 }, { "epoch": 1.86, "learning_rate": 1.9539336880928915e-05, "loss": 0.4264, "step": 1834 }, { "epoch": 1.86, "learning_rate": 1.9538678963585595e-05, "loss": 0.3305, "step": 1835 }, { "epoch": 1.86, "learning_rate": 1.9538020587852436e-05, "loss": 0.3935, "step": 1836 }, { "epoch": 1.86, "learning_rate": 1.9537361753761074e-05, "loss": 0.4283, "step": 1837 }, { "epoch": 1.87, "learning_rate": 1.953670246134317e-05, "loss": 0.3558, "step": 1838 }, { "epoch": 1.87, "learning_rate": 1.9536042710630406e-05, "loss": 0.3544, "step": 1839 }, { "epoch": 1.87, "learning_rate": 1.9535382501654494e-05, "loss": 0.336, "step": 1840 }, { "epoch": 1.87, "learning_rate": 1.953472183444715e-05, "loss": 0.3167, "step": 1841 }, { "epoch": 1.87, "learning_rate": 1.9534060709040134e-05, "loss": 0.4137, "step": 1842 }, { "epoch": 1.87, "learning_rate": 1.9533399125465206e-05, "loss": 0.3952, "step": 1843 }, { "epoch": 1.87, "learning_rate": 1.953273708375417e-05, "loss": 0.3689, "step": 1844 }, { "epoch": 1.87, "learning_rate": 1.953207458393883e-05, "loss": 0.3924, "step": 1845 }, { "epoch": 1.87, "learning_rate": 1.953141162605103e-05, "loss": 0.3975, "step": 1846 }, { "epoch": 1.87, "learning_rate": 1.9530748210122634e-05, "loss": 0.3807, "step": 1847 }, { "epoch": 1.88, "learning_rate": 1.9530084336185507e-05, "loss": 0.3332, "step": 1848 }, { "epoch": 1.88, "learning_rate": 1.9529420004271568e-05, "loss": 0.3742, "step": 1849 }, { "epoch": 1.88, "learning_rate": 1.9528755214412735e-05, "loss": 0.3665, "step": 1850 }, { "epoch": 1.88, "learning_rate": 1.9528089966640952e-05, "loss": 0.372, "step": 1851 }, { "epoch": 1.88, "learning_rate": 1.9527424260988196e-05, "loss": 0.3511, "step": 1852 }, { "epoch": 1.88, "learning_rate": 1.9526758097486454e-05, "loss": 0.3864, "step": 1853 }, { "epoch": 1.88, "learning_rate": 1.9526091476167738e-05, "loss": 0.338, "step": 1854 }, { "epoch": 1.88, "learning_rate": 1.9525424397064082e-05, "loss": 0.3878, "step": 1855 }, { "epoch": 1.88, "learning_rate": 1.9524756860207548e-05, "loss": 0.354, "step": 1856 }, { "epoch": 1.88, "learning_rate": 1.9524088865630212e-05, "loss": 0.3567, "step": 1857 }, { "epoch": 1.89, "learning_rate": 1.952342041336418e-05, "loss": 0.3334, "step": 1858 }, { "epoch": 1.89, "learning_rate": 1.9522751503441562e-05, "loss": 0.3914, "step": 1859 }, { "epoch": 1.89, "learning_rate": 1.952208213589452e-05, "loss": 0.3696, "step": 1860 }, { "epoch": 1.89, "learning_rate": 1.9521412310755204e-05, "loss": 0.381, "step": 1861 }, { "epoch": 1.89, "learning_rate": 1.9520742028055817e-05, "loss": 0.3775, "step": 1862 }, { "epoch": 1.89, "learning_rate": 1.9520071287828565e-05, "loss": 0.3542, "step": 1863 }, { "epoch": 1.89, "learning_rate": 1.951940009010568e-05, "loss": 0.3606, "step": 1864 }, { "epoch": 1.89, "learning_rate": 1.9518728434919417e-05, "loss": 0.381, "step": 1865 }, { "epoch": 1.89, "learning_rate": 1.9518056322302055e-05, "loss": 0.377, "step": 1866 }, { "epoch": 1.89, "learning_rate": 1.951738375228589e-05, "loss": 0.3554, "step": 1867 }, { "epoch": 1.9, "learning_rate": 1.9516710724903247e-05, "loss": 0.3627, "step": 1868 }, { "epoch": 1.9, "learning_rate": 1.9516037240186464e-05, "loss": 0.3605, "step": 1869 }, { "epoch": 1.9, "learning_rate": 1.951536329816791e-05, "loss": 0.3241, "step": 1870 }, { "epoch": 1.9, "learning_rate": 1.951468889887997e-05, "loss": 0.3404, "step": 1871 }, { "epoch": 1.9, "learning_rate": 1.9514014042355057e-05, "loss": 0.3623, "step": 1872 }, { "epoch": 1.9, "learning_rate": 1.9513338728625594e-05, "loss": 0.3394, "step": 1873 }, { "epoch": 1.9, "learning_rate": 1.951266295772404e-05, "loss": 0.3491, "step": 1874 }, { "epoch": 1.9, "learning_rate": 1.951198672968287e-05, "loss": 0.3372, "step": 1875 }, { "epoch": 1.9, "learning_rate": 1.9511310044534575e-05, "loss": 0.367, "step": 1876 }, { "epoch": 1.9, "learning_rate": 1.9510632902311676e-05, "loss": 0.3034, "step": 1877 }, { "epoch": 1.91, "learning_rate": 1.950995530304672e-05, "loss": 0.3518, "step": 1878 }, { "epoch": 1.91, "learning_rate": 1.950927724677226e-05, "loss": 0.3533, "step": 1879 }, { "epoch": 1.91, "learning_rate": 1.950859873352089e-05, "loss": 0.3952, "step": 1880 }, { "epoch": 1.91, "learning_rate": 1.9507919763325206e-05, "loss": 0.3812, "step": 1881 }, { "epoch": 1.91, "learning_rate": 1.9507240336217844e-05, "loss": 0.3524, "step": 1882 }, { "epoch": 1.91, "learning_rate": 1.9506560452231454e-05, "loss": 0.3273, "step": 1883 }, { "epoch": 1.91, "learning_rate": 1.9505880111398708e-05, "loss": 0.3575, "step": 1884 }, { "epoch": 1.91, "learning_rate": 1.9505199313752298e-05, "loss": 0.3555, "step": 1885 }, { "epoch": 1.91, "learning_rate": 1.9504518059324943e-05, "loss": 0.3377, "step": 1886 }, { "epoch": 1.91, "learning_rate": 1.9503836348149382e-05, "loss": 0.3262, "step": 1887 }, { "epoch": 1.92, "learning_rate": 1.950315418025837e-05, "loss": 0.3572, "step": 1888 }, { "epoch": 1.92, "learning_rate": 1.9502471555684695e-05, "loss": 0.386, "step": 1889 }, { "epoch": 1.92, "learning_rate": 1.950178847446116e-05, "loss": 0.3712, "step": 1890 }, { "epoch": 1.92, "learning_rate": 1.9501104936620586e-05, "loss": 0.3577, "step": 1891 }, { "epoch": 1.92, "learning_rate": 1.9500420942195827e-05, "loss": 0.3381, "step": 1892 }, { "epoch": 1.92, "learning_rate": 1.9499736491219754e-05, "loss": 0.3383, "step": 1893 }, { "epoch": 1.92, "learning_rate": 1.949905158372525e-05, "loss": 0.3464, "step": 1894 }, { "epoch": 1.92, "learning_rate": 1.949836621974524e-05, "loss": 0.4198, "step": 1895 }, { "epoch": 1.92, "learning_rate": 1.9497680399312653e-05, "loss": 0.4077, "step": 1896 }, { "epoch": 1.92, "learning_rate": 1.9496994122460446e-05, "loss": 0.4007, "step": 1897 }, { "epoch": 1.93, "learning_rate": 1.94963073892216e-05, "loss": 0.3242, "step": 1898 }, { "epoch": 1.93, "learning_rate": 1.9495620199629123e-05, "loss": 0.4141, "step": 1899 }, { "epoch": 1.93, "learning_rate": 1.949493255371603e-05, "loss": 0.3587, "step": 1900 }, { "epoch": 1.93, "learning_rate": 1.949424445151537e-05, "loss": 0.3769, "step": 1901 }, { "epoch": 1.93, "learning_rate": 1.949355589306021e-05, "loss": 0.3431, "step": 1902 }, { "epoch": 1.93, "learning_rate": 1.9492866878383634e-05, "loss": 0.345, "step": 1903 }, { "epoch": 1.93, "learning_rate": 1.9492177407518762e-05, "loss": 0.3255, "step": 1904 }, { "epoch": 1.93, "learning_rate": 1.9491487480498724e-05, "loss": 0.3567, "step": 1905 }, { "epoch": 1.93, "learning_rate": 1.9490797097356674e-05, "loss": 0.3754, "step": 1906 }, { "epoch": 1.94, "learning_rate": 1.949010625812579e-05, "loss": 0.4463, "step": 1907 }, { "epoch": 1.94, "learning_rate": 1.948941496283927e-05, "loss": 0.379, "step": 1908 }, { "epoch": 1.94, "learning_rate": 1.9488723211530336e-05, "loss": 0.3536, "step": 1909 }, { "epoch": 1.94, "learning_rate": 1.948803100423223e-05, "loss": 0.3741, "step": 1910 }, { "epoch": 1.94, "learning_rate": 1.9487338340978215e-05, "loss": 0.373, "step": 1911 }, { "epoch": 1.94, "learning_rate": 1.9486645221801578e-05, "loss": 0.3626, "step": 1912 }, { "epoch": 1.94, "learning_rate": 1.9485951646735634e-05, "loss": 0.2974, "step": 1913 }, { "epoch": 1.94, "learning_rate": 1.9485257615813705e-05, "loss": 0.3781, "step": 1914 }, { "epoch": 1.94, "learning_rate": 1.9484563129069144e-05, "loss": 0.3603, "step": 1915 }, { "epoch": 1.94, "learning_rate": 1.948386818653533e-05, "loss": 0.3578, "step": 1916 }, { "epoch": 1.95, "learning_rate": 1.9483172788245656e-05, "loss": 0.3346, "step": 1917 }, { "epoch": 1.95, "learning_rate": 1.9482476934233542e-05, "loss": 0.4086, "step": 1918 }, { "epoch": 1.95, "learning_rate": 1.9481780624532424e-05, "loss": 0.4009, "step": 1919 }, { "epoch": 1.95, "learning_rate": 1.948108385917577e-05, "loss": 0.3663, "step": 1920 }, { "epoch": 1.95, "learning_rate": 1.9480386638197054e-05, "loss": 0.3048, "step": 1921 }, { "epoch": 1.95, "learning_rate": 1.9479688961629796e-05, "loss": 0.3715, "step": 1922 }, { "epoch": 1.95, "learning_rate": 1.9478990829507507e-05, "loss": 0.3585, "step": 1923 }, { "epoch": 1.95, "learning_rate": 1.9478292241863747e-05, "loss": 0.3973, "step": 1924 }, { "epoch": 1.95, "learning_rate": 1.9477593198732085e-05, "loss": 0.3584, "step": 1925 }, { "epoch": 1.95, "learning_rate": 1.9476893700146115e-05, "loss": 0.3636, "step": 1926 }, { "epoch": 1.96, "learning_rate": 1.947619374613945e-05, "loss": 0.3493, "step": 1927 }, { "epoch": 1.96, "learning_rate": 1.9475493336745725e-05, "loss": 0.4086, "step": 1928 }, { "epoch": 1.96, "learning_rate": 1.9474792471998607e-05, "loss": 0.3787, "step": 1929 }, { "epoch": 1.96, "learning_rate": 1.9474091151931767e-05, "loss": 0.2954, "step": 1930 }, { "epoch": 1.96, "learning_rate": 1.9473389376578914e-05, "loss": 0.3696, "step": 1931 }, { "epoch": 1.96, "learning_rate": 1.9472687145973767e-05, "loss": 0.3437, "step": 1932 }, { "epoch": 1.96, "learning_rate": 1.9471984460150078e-05, "loss": 0.3456, "step": 1933 }, { "epoch": 1.96, "learning_rate": 1.9471281319141614e-05, "loss": 0.4271, "step": 1934 }, { "epoch": 1.96, "learning_rate": 1.9470577722982163e-05, "loss": 0.3606, "step": 1935 }, { "epoch": 1.96, "learning_rate": 1.9469873671705537e-05, "loss": 0.3667, "step": 1936 }, { "epoch": 1.97, "learning_rate": 1.9469169165345573e-05, "loss": 0.3792, "step": 1937 }, { "epoch": 1.97, "learning_rate": 1.9468464203936118e-05, "loss": 0.2963, "step": 1938 }, { "epoch": 1.97, "learning_rate": 1.9467758787511063e-05, "loss": 0.3369, "step": 1939 }, { "epoch": 1.97, "learning_rate": 1.9467052916104297e-05, "loss": 0.4186, "step": 1940 }, { "epoch": 1.97, "learning_rate": 1.946634658974975e-05, "loss": 0.3784, "step": 1941 }, { "epoch": 1.97, "learning_rate": 1.9465639808481353e-05, "loss": 0.3774, "step": 1942 }, { "epoch": 1.97, "learning_rate": 1.9464932572333076e-05, "loss": 0.3694, "step": 1943 }, { "epoch": 1.97, "learning_rate": 1.9464224881338915e-05, "loss": 0.3718, "step": 1944 }, { "epoch": 1.97, "learning_rate": 1.9463516735532865e-05, "loss": 0.3795, "step": 1945 }, { "epoch": 1.97, "learning_rate": 1.946280813494897e-05, "loss": 0.3477, "step": 1946 }, { "epoch": 1.98, "learning_rate": 1.9462099079621265e-05, "loss": 0.3354, "step": 1947 }, { "epoch": 1.98, "learning_rate": 1.9461389569583842e-05, "loss": 0.3785, "step": 1948 }, { "epoch": 1.98, "learning_rate": 1.9460679604870792e-05, "loss": 0.3821, "step": 1949 }, { "epoch": 1.98, "learning_rate": 1.9459969185516226e-05, "loss": 0.3642, "step": 1950 }, { "epoch": 1.98, "learning_rate": 1.945925831155429e-05, "loss": 0.408, "step": 1951 }, { "epoch": 1.98, "learning_rate": 1.9458546983019143e-05, "loss": 0.3622, "step": 1952 }, { "epoch": 1.98, "learning_rate": 1.9457835199944974e-05, "loss": 0.3245, "step": 1953 }, { "epoch": 1.98, "learning_rate": 1.945712296236598e-05, "loss": 0.3424, "step": 1954 }, { "epoch": 1.98, "learning_rate": 1.9456410270316396e-05, "loss": 0.3167, "step": 1955 }, { "epoch": 1.98, "learning_rate": 1.9455697123830463e-05, "loss": 0.4151, "step": 1956 }, { "epoch": 1.99, "learning_rate": 1.945498352294246e-05, "loss": 0.3479, "step": 1957 }, { "epoch": 1.99, "learning_rate": 1.9454269467686677e-05, "loss": 0.3315, "step": 1958 }, { "epoch": 1.99, "learning_rate": 1.9453554958097426e-05, "loss": 0.3744, "step": 1959 }, { "epoch": 1.99, "learning_rate": 1.9452839994209046e-05, "loss": 0.4002, "step": 1960 }, { "epoch": 1.99, "learning_rate": 1.945212457605589e-05, "loss": 0.4272, "step": 1961 }, { "epoch": 1.99, "learning_rate": 1.945140870367235e-05, "loss": 0.3301, "step": 1962 }, { "epoch": 1.99, "learning_rate": 1.945069237709282e-05, "loss": 0.3884, "step": 1963 }, { "epoch": 1.99, "learning_rate": 1.9449975596351717e-05, "loss": 0.3172, "step": 1964 }, { "epoch": 1.99, "learning_rate": 1.94492583614835e-05, "loss": 0.3661, "step": 1965 }, { "epoch": 1.99, "learning_rate": 1.9448540672522623e-05, "loss": 0.3255, "step": 1966 }, { "epoch": 2.0, "learning_rate": 1.944782252950359e-05, "loss": 0.3074, "step": 1967 }, { "epoch": 2.0, "learning_rate": 1.94471039324609e-05, "loss": 0.3536, "step": 1968 }, { "epoch": 2.0, "learning_rate": 1.9446384881429092e-05, "loss": 0.3523, "step": 1969 }, { "epoch": 2.0, "learning_rate": 1.9445665376442716e-05, "loss": 0.3403, "step": 1970 }, { "epoch": 2.0, "learning_rate": 1.9444945417536354e-05, "loss": 0.347, "step": 1971 }, { "epoch": 2.0, "learning_rate": 1.9444225004744602e-05, "loss": 0.2313, "step": 1972 }, { "epoch": 2.0, "learning_rate": 1.9443504138102077e-05, "loss": 0.246, "step": 1973 }, { "epoch": 2.0, "learning_rate": 1.9442782817643425e-05, "loss": 0.215, "step": 1974 }, { "epoch": 2.0, "learning_rate": 1.9442061043403308e-05, "loss": 0.2166, "step": 1975 }, { "epoch": 2.01, "learning_rate": 1.944133881541641e-05, "loss": 0.2422, "step": 1976 }, { "epoch": 2.01, "learning_rate": 1.9440616133717438e-05, "loss": 0.2227, "step": 1977 }, { "epoch": 2.01, "learning_rate": 1.9439892998341127e-05, "loss": 0.2435, "step": 1978 }, { "epoch": 2.01, "learning_rate": 1.943916940932222e-05, "loss": 0.2264, "step": 1979 }, { "epoch": 2.01, "learning_rate": 1.94384453666955e-05, "loss": 0.236, "step": 1980 }, { "epoch": 2.01, "learning_rate": 1.9437720870495753e-05, "loss": 0.2539, "step": 1981 }, { "epoch": 2.01, "learning_rate": 1.9436995920757797e-05, "loss": 0.2227, "step": 1982 }, { "epoch": 2.01, "learning_rate": 1.943627051751647e-05, "loss": 0.2341, "step": 1983 }, { "epoch": 2.01, "learning_rate": 1.9435544660806632e-05, "loss": 0.2243, "step": 1984 }, { "epoch": 2.01, "learning_rate": 1.9434818350663168e-05, "loss": 0.2217, "step": 1985 }, { "epoch": 2.02, "learning_rate": 1.9434091587120975e-05, "loss": 0.2092, "step": 1986 }, { "epoch": 2.02, "learning_rate": 1.9433364370214985e-05, "loss": 0.223, "step": 1987 }, { "epoch": 2.02, "learning_rate": 1.943263669998014e-05, "loss": 0.1876, "step": 1988 }, { "epoch": 2.02, "learning_rate": 1.943190857645141e-05, "loss": 0.226, "step": 1989 }, { "epoch": 2.02, "learning_rate": 1.9431179999663792e-05, "loss": 0.2586, "step": 1990 }, { "epoch": 2.02, "learning_rate": 1.943045096965229e-05, "loss": 0.2234, "step": 1991 }, { "epoch": 2.02, "learning_rate": 1.9429721486451942e-05, "loss": 0.2123, "step": 1992 }, { "epoch": 2.02, "learning_rate": 1.94289915500978e-05, "loss": 0.2407, "step": 1993 }, { "epoch": 2.02, "learning_rate": 1.942826116062495e-05, "loss": 0.1978, "step": 1994 }, { "epoch": 2.02, "learning_rate": 1.942753031806848e-05, "loss": 0.2701, "step": 1995 }, { "epoch": 2.03, "learning_rate": 1.9426799022463526e-05, "loss": 0.2673, "step": 1996 }, { "epoch": 2.03, "learning_rate": 1.9426067273845216e-05, "loss": 0.1955, "step": 1997 }, { "epoch": 2.03, "learning_rate": 1.9425335072248723e-05, "loss": 0.2238, "step": 1998 }, { "epoch": 2.03, "learning_rate": 1.9424602417709235e-05, "loss": 0.2514, "step": 1999 }, { "epoch": 2.03, "learning_rate": 1.9423869310261955e-05, "loss": 0.2213, "step": 2000 }, { "epoch": 2.03, "learning_rate": 1.9423135749942117e-05, "loss": 0.2484, "step": 2001 }, { "epoch": 2.03, "learning_rate": 1.942240173678497e-05, "loss": 0.2003, "step": 2002 }, { "epoch": 2.03, "learning_rate": 1.9421667270825792e-05, "loss": 0.2277, "step": 2003 }, { "epoch": 2.03, "learning_rate": 1.9420932352099877e-05, "loss": 0.2452, "step": 2004 }, { "epoch": 2.03, "learning_rate": 1.942019698064254e-05, "loss": 0.237, "step": 2005 }, { "epoch": 2.04, "learning_rate": 1.9419461156489116e-05, "loss": 0.2488, "step": 2006 }, { "epoch": 2.04, "learning_rate": 1.9418724879674978e-05, "loss": 0.2272, "step": 2007 }, { "epoch": 2.04, "learning_rate": 1.9417988150235496e-05, "loss": 0.2324, "step": 2008 }, { "epoch": 2.04, "learning_rate": 1.941725096820608e-05, "loss": 0.2477, "step": 2009 }, { "epoch": 2.04, "learning_rate": 1.9416513333622156e-05, "loss": 0.3172, "step": 2010 }, { "epoch": 2.04, "learning_rate": 1.941577524651917e-05, "loss": 0.203, "step": 2011 }, { "epoch": 2.04, "learning_rate": 1.941503670693259e-05, "loss": 0.2226, "step": 2012 }, { "epoch": 2.04, "learning_rate": 1.9414297714897915e-05, "loss": 0.2338, "step": 2013 }, { "epoch": 2.04, "learning_rate": 1.9413558270450652e-05, "loss": 0.2337, "step": 2014 }, { "epoch": 2.04, "learning_rate": 1.9412818373626335e-05, "loss": 0.2231, "step": 2015 }, { "epoch": 2.05, "learning_rate": 1.9412078024460522e-05, "loss": 0.2023, "step": 2016 }, { "epoch": 2.05, "learning_rate": 1.941133722298879e-05, "loss": 0.2092, "step": 2017 }, { "epoch": 2.05, "learning_rate": 1.941059596924674e-05, "loss": 0.2207, "step": 2018 }, { "epoch": 2.05, "learning_rate": 1.9409854263269993e-05, "loss": 0.2259, "step": 2019 }, { "epoch": 2.05, "learning_rate": 1.940911210509419e-05, "loss": 0.2, "step": 2020 }, { "epoch": 2.05, "learning_rate": 1.9408369494755e-05, "loss": 0.1915, "step": 2021 }, { "epoch": 2.05, "learning_rate": 1.940762643228811e-05, "loss": 0.202, "step": 2022 }, { "epoch": 2.05, "learning_rate": 1.9406882917729227e-05, "loss": 0.2151, "step": 2023 }, { "epoch": 2.05, "learning_rate": 1.9406138951114083e-05, "loss": 0.2614, "step": 2024 }, { "epoch": 2.05, "learning_rate": 1.9405394532478422e-05, "loss": 0.2482, "step": 2025 }, { "epoch": 2.06, "learning_rate": 1.940464966185803e-05, "loss": 0.2295, "step": 2026 }, { "epoch": 2.06, "learning_rate": 1.9403904339288693e-05, "loss": 0.2664, "step": 2027 }, { "epoch": 2.06, "learning_rate": 1.9403158564806234e-05, "loss": 0.234, "step": 2028 }, { "epoch": 2.06, "learning_rate": 1.9402412338446487e-05, "loss": 0.2409, "step": 2029 }, { "epoch": 2.06, "learning_rate": 1.9401665660245318e-05, "loss": 0.2346, "step": 2030 }, { "epoch": 2.06, "learning_rate": 1.9400918530238605e-05, "loss": 0.2048, "step": 2031 }, { "epoch": 2.06, "learning_rate": 1.940017094846225e-05, "loss": 0.2532, "step": 2032 }, { "epoch": 2.06, "learning_rate": 1.9399422914952185e-05, "loss": 0.2287, "step": 2033 }, { "epoch": 2.06, "learning_rate": 1.9398674429744355e-05, "loss": 0.2586, "step": 2034 }, { "epoch": 2.06, "learning_rate": 1.9397925492874727e-05, "loss": 0.2257, "step": 2035 }, { "epoch": 2.07, "learning_rate": 1.9397176104379296e-05, "loss": 0.2322, "step": 2036 }, { "epoch": 2.07, "learning_rate": 1.939642626429407e-05, "loss": 0.2405, "step": 2037 }, { "epoch": 2.07, "learning_rate": 1.9395675972655085e-05, "loss": 0.2353, "step": 2038 }, { "epoch": 2.07, "learning_rate": 1.93949252294984e-05, "loss": 0.2385, "step": 2039 }, { "epoch": 2.07, "learning_rate": 1.9394174034860085e-05, "loss": 0.262, "step": 2040 }, { "epoch": 2.07, "learning_rate": 1.9393422388776247e-05, "loss": 0.2385, "step": 2041 }, { "epoch": 2.07, "learning_rate": 1.9392670291283005e-05, "loss": 0.2363, "step": 2042 }, { "epoch": 2.07, "learning_rate": 1.93919177424165e-05, "loss": 0.2594, "step": 2043 }, { "epoch": 2.07, "learning_rate": 1.9391164742212894e-05, "loss": 0.2432, "step": 2044 }, { "epoch": 2.08, "learning_rate": 1.9390411290708382e-05, "loss": 0.2552, "step": 2045 }, { "epoch": 2.08, "learning_rate": 1.9389657387939163e-05, "loss": 0.1951, "step": 2046 }, { "epoch": 2.08, "learning_rate": 1.9388903033941472e-05, "loss": 0.2678, "step": 2047 }, { "epoch": 2.08, "learning_rate": 1.9388148228751557e-05, "loss": 0.2555, "step": 2048 }, { "epoch": 2.08, "learning_rate": 1.9387392972405687e-05, "loss": 0.2503, "step": 2049 }, { "epoch": 2.08, "learning_rate": 1.9386637264940166e-05, "loss": 0.2473, "step": 2050 }, { "epoch": 2.08, "learning_rate": 1.9385881106391304e-05, "loss": 0.2511, "step": 2051 }, { "epoch": 2.08, "learning_rate": 1.9385124496795443e-05, "loss": 0.2588, "step": 2052 }, { "epoch": 2.08, "learning_rate": 1.9384367436188937e-05, "loss": 0.2363, "step": 2053 }, { "epoch": 2.08, "learning_rate": 1.938360992460817e-05, "loss": 0.2225, "step": 2054 }, { "epoch": 2.09, "learning_rate": 1.9382851962089545e-05, "loss": 0.1976, "step": 2055 }, { "epoch": 2.09, "learning_rate": 1.9382093548669487e-05, "loss": 0.2618, "step": 2056 }, { "epoch": 2.09, "learning_rate": 1.9381334684384444e-05, "loss": 0.2645, "step": 2057 }, { "epoch": 2.09, "learning_rate": 1.9380575369270878e-05, "loss": 0.2238, "step": 2058 }, { "epoch": 2.09, "learning_rate": 1.9379815603365283e-05, "loss": 0.2434, "step": 2059 }, { "epoch": 2.09, "learning_rate": 1.937905538670417e-05, "loss": 0.2295, "step": 2060 }, { "epoch": 2.09, "learning_rate": 1.9378294719324073e-05, "loss": 0.2113, "step": 2061 }, { "epoch": 2.09, "learning_rate": 1.9377533601261546e-05, "loss": 0.234, "step": 2062 }, { "epoch": 2.09, "learning_rate": 1.937677203255316e-05, "loss": 0.221, "step": 2063 }, { "epoch": 2.09, "learning_rate": 1.937601001323552e-05, "loss": 0.2609, "step": 2064 }, { "epoch": 2.1, "learning_rate": 1.937524754334524e-05, "loss": 0.2164, "step": 2065 }, { "epoch": 2.1, "learning_rate": 1.9374484622918968e-05, "loss": 0.2469, "step": 2066 }, { "epoch": 2.1, "learning_rate": 1.937372125199336e-05, "loss": 0.221, "step": 2067 }, { "epoch": 2.1, "learning_rate": 1.9372957430605104e-05, "loss": 0.2432, "step": 2068 }, { "epoch": 2.1, "learning_rate": 1.9372193158790903e-05, "loss": 0.2102, "step": 2069 }, { "epoch": 2.1, "learning_rate": 1.937142843658749e-05, "loss": 0.2401, "step": 2070 }, { "epoch": 2.1, "learning_rate": 1.937066326403161e-05, "loss": 0.2596, "step": 2071 }, { "epoch": 2.1, "learning_rate": 1.9369897641160035e-05, "loss": 0.2535, "step": 2072 }, { "epoch": 2.1, "learning_rate": 1.936913156800956e-05, "loss": 0.2341, "step": 2073 }, { "epoch": 2.1, "learning_rate": 1.9368365044616996e-05, "loss": 0.2412, "step": 2074 }, { "epoch": 2.11, "learning_rate": 1.936759807101918e-05, "loss": 0.2268, "step": 2075 }, { "epoch": 2.11, "learning_rate": 1.9366830647252974e-05, "loss": 0.23, "step": 2076 }, { "epoch": 2.11, "learning_rate": 1.9366062773355247e-05, "loss": 0.2492, "step": 2077 }, { "epoch": 2.11, "learning_rate": 1.936529444936291e-05, "loss": 0.2711, "step": 2078 }, { "epoch": 2.11, "learning_rate": 1.936452567531288e-05, "loss": 0.2262, "step": 2079 }, { "epoch": 2.11, "learning_rate": 1.9363756451242107e-05, "loss": 0.2163, "step": 2080 }, { "epoch": 2.11, "learning_rate": 1.9362986777187548e-05, "loss": 0.2246, "step": 2081 }, { "epoch": 2.11, "learning_rate": 1.9362216653186197e-05, "loss": 0.253, "step": 2082 }, { "epoch": 2.11, "learning_rate": 1.936144607927506e-05, "loss": 0.2426, "step": 2083 }, { "epoch": 2.11, "learning_rate": 1.936067505549117e-05, "loss": 0.2478, "step": 2084 }, { "epoch": 2.12, "learning_rate": 1.9359903581871576e-05, "loss": 0.2633, "step": 2085 }, { "epoch": 2.12, "learning_rate": 1.9359131658453358e-05, "loss": 0.2261, "step": 2086 }, { "epoch": 2.12, "learning_rate": 1.93583592852736e-05, "loss": 0.2515, "step": 2087 }, { "epoch": 2.12, "learning_rate": 1.9357586462369432e-05, "loss": 0.218, "step": 2088 }, { "epoch": 2.12, "learning_rate": 1.9356813189777982e-05, "loss": 0.2668, "step": 2089 }, { "epoch": 2.12, "learning_rate": 1.935603946753642e-05, "loss": 0.219, "step": 2090 }, { "epoch": 2.12, "learning_rate": 1.935526529568192e-05, "loss": 0.266, "step": 2091 }, { "epoch": 2.12, "learning_rate": 1.9354490674251692e-05, "loss": 0.2204, "step": 2092 }, { "epoch": 2.12, "learning_rate": 1.935371560328296e-05, "loss": 0.2258, "step": 2093 }, { "epoch": 2.12, "learning_rate": 1.935294008281296e-05, "loss": 0.2429, "step": 2094 }, { "epoch": 2.13, "learning_rate": 1.9352164112878977e-05, "loss": 0.2148, "step": 2095 }, { "epoch": 2.13, "learning_rate": 1.935138769351829e-05, "loss": 0.2278, "step": 2096 }, { "epoch": 2.13, "learning_rate": 1.9350610824768213e-05, "loss": 0.2164, "step": 2097 }, { "epoch": 2.13, "learning_rate": 1.934983350666608e-05, "loss": 0.2438, "step": 2098 }, { "epoch": 2.13, "learning_rate": 1.9349055739249246e-05, "loss": 0.2481, "step": 2099 }, { "epoch": 2.13, "learning_rate": 1.9348277522555086e-05, "loss": 0.2024, "step": 2100 }, { "epoch": 2.13, "learning_rate": 1.9347498856621e-05, "loss": 0.2308, "step": 2101 }, { "epoch": 2.13, "learning_rate": 1.93467197414844e-05, "loss": 0.2182, "step": 2102 }, { "epoch": 2.13, "learning_rate": 1.934594017718274e-05, "loss": 0.2391, "step": 2103 }, { "epoch": 2.13, "learning_rate": 1.9345160163753474e-05, "loss": 0.2416, "step": 2104 }, { "epoch": 2.14, "learning_rate": 1.9344379701234087e-05, "loss": 0.2284, "step": 2105 }, { "epoch": 2.14, "learning_rate": 1.9343598789662084e-05, "loss": 0.2452, "step": 2106 }, { "epoch": 2.14, "learning_rate": 1.9342817429075e-05, "loss": 0.2295, "step": 2107 }, { "epoch": 2.14, "learning_rate": 1.934203561951037e-05, "loss": 0.2376, "step": 2108 }, { "epoch": 2.14, "learning_rate": 1.934125336100578e-05, "loss": 0.2295, "step": 2109 }, { "epoch": 2.14, "learning_rate": 1.934047065359881e-05, "loss": 0.2532, "step": 2110 }, { "epoch": 2.14, "learning_rate": 1.9339687497327086e-05, "loss": 0.2261, "step": 2111 }, { "epoch": 2.14, "learning_rate": 1.933890389222823e-05, "loss": 0.2275, "step": 2112 }, { "epoch": 2.14, "learning_rate": 1.9338119838339905e-05, "loss": 0.2202, "step": 2113 }, { "epoch": 2.15, "learning_rate": 1.933733533569979e-05, "loss": 0.2456, "step": 2114 }, { "epoch": 2.15, "learning_rate": 1.933655038434558e-05, "loss": 0.2487, "step": 2115 }, { "epoch": 2.15, "learning_rate": 1.933576498431501e-05, "loss": 0.2395, "step": 2116 }, { "epoch": 2.15, "learning_rate": 1.9334979135645807e-05, "loss": 0.2352, "step": 2117 }, { "epoch": 2.15, "learning_rate": 1.9334192838375745e-05, "loss": 0.2429, "step": 2118 }, { "epoch": 2.15, "learning_rate": 1.933340609254261e-05, "loss": 0.2154, "step": 2119 }, { "epoch": 2.15, "learning_rate": 1.9332618898184204e-05, "loss": 0.2454, "step": 2120 }, { "epoch": 2.15, "learning_rate": 1.933183125533836e-05, "loss": 0.2455, "step": 2121 }, { "epoch": 2.15, "learning_rate": 1.933104316404293e-05, "loss": 0.2404, "step": 2122 }, { "epoch": 2.15, "learning_rate": 1.9330254624335782e-05, "loss": 0.2526, "step": 2123 }, { "epoch": 2.16, "learning_rate": 1.9329465636254815e-05, "loss": 0.2693, "step": 2124 }, { "epoch": 2.16, "learning_rate": 1.9328676199837947e-05, "loss": 0.2001, "step": 2125 }, { "epoch": 2.16, "learning_rate": 1.9327886315123104e-05, "loss": 0.2694, "step": 2126 }, { "epoch": 2.16, "learning_rate": 1.9327095982148258e-05, "loss": 0.2445, "step": 2127 }, { "epoch": 2.16, "learning_rate": 1.932630520095138e-05, "loss": 0.2155, "step": 2128 }, { "epoch": 2.16, "learning_rate": 1.9325513971570472e-05, "loss": 0.1961, "step": 2129 }, { "epoch": 2.16, "learning_rate": 1.932472229404356e-05, "loss": 0.2456, "step": 2130 }, { "epoch": 2.16, "learning_rate": 1.9323930168408688e-05, "loss": 0.2586, "step": 2131 }, { "epoch": 2.16, "learning_rate": 1.9323137594703925e-05, "loss": 0.2531, "step": 2132 }, { "epoch": 2.16, "learning_rate": 1.9322344572967356e-05, "loss": 0.2654, "step": 2133 }, { "epoch": 2.17, "learning_rate": 1.9321551103237088e-05, "loss": 0.2119, "step": 2134 }, { "epoch": 2.17, "learning_rate": 1.932075718555126e-05, "loss": 0.2348, "step": 2135 }, { "epoch": 2.17, "learning_rate": 1.9319962819948016e-05, "loss": 0.2228, "step": 2136 }, { "epoch": 2.17, "learning_rate": 1.9319168006465533e-05, "loss": 0.2253, "step": 2137 }, { "epoch": 2.17, "learning_rate": 1.9318372745142007e-05, "loss": 0.2144, "step": 2138 }, { "epoch": 2.17, "learning_rate": 1.9317577036015654e-05, "loss": 0.2341, "step": 2139 }, { "epoch": 2.17, "learning_rate": 1.9316780879124715e-05, "loss": 0.2855, "step": 2140 }, { "epoch": 2.17, "learning_rate": 1.9315984274507446e-05, "loss": 0.2329, "step": 2141 }, { "epoch": 2.17, "learning_rate": 1.931518722220213e-05, "loss": 0.2043, "step": 2142 }, { "epoch": 2.17, "learning_rate": 1.9314389722247077e-05, "loss": 0.2289, "step": 2143 }, { "epoch": 2.18, "learning_rate": 1.93135917746806e-05, "loss": 0.2247, "step": 2144 }, { "epoch": 2.18, "learning_rate": 1.9312793379541054e-05, "loss": 0.2221, "step": 2145 }, { "epoch": 2.18, "learning_rate": 1.93119945368668e-05, "loss": 0.2418, "step": 2146 }, { "epoch": 2.18, "learning_rate": 1.9311195246696227e-05, "loss": 0.2744, "step": 2147 }, { "epoch": 2.18, "learning_rate": 1.9310395509067754e-05, "loss": 0.2198, "step": 2148 }, { "epoch": 2.18, "learning_rate": 1.9309595324019805e-05, "loss": 0.2537, "step": 2149 }, { "epoch": 2.18, "learning_rate": 1.930879469159084e-05, "loss": 0.2421, "step": 2150 }, { "epoch": 2.18, "learning_rate": 1.9307993611819328e-05, "loss": 0.2219, "step": 2151 }, { "epoch": 2.18, "learning_rate": 1.9307192084743768e-05, "loss": 0.267, "step": 2152 }, { "epoch": 2.18, "learning_rate": 1.9306390110402682e-05, "loss": 0.2623, "step": 2153 }, { "epoch": 2.19, "learning_rate": 1.93055876888346e-05, "loss": 0.2701, "step": 2154 }, { "epoch": 2.19, "learning_rate": 1.9304784820078094e-05, "loss": 0.254, "step": 2155 }, { "epoch": 2.19, "learning_rate": 1.9303981504171742e-05, "loss": 0.2176, "step": 2156 }, { "epoch": 2.19, "learning_rate": 1.9303177741154146e-05, "loss": 0.2448, "step": 2157 }, { "epoch": 2.19, "learning_rate": 1.930237353106393e-05, "loss": 0.2579, "step": 2158 }, { "epoch": 2.19, "learning_rate": 1.9301568873939753e-05, "loss": 0.2416, "step": 2159 }, { "epoch": 2.19, "learning_rate": 1.9300763769820265e-05, "loss": 0.2225, "step": 2160 }, { "epoch": 2.19, "learning_rate": 1.929995821874417e-05, "loss": 0.2543, "step": 2161 }, { "epoch": 2.19, "learning_rate": 1.9299152220750178e-05, "loss": 0.2389, "step": 2162 }, { "epoch": 2.19, "learning_rate": 1.9298345775877014e-05, "loss": 0.2258, "step": 2163 }, { "epoch": 2.2, "learning_rate": 1.9297538884163443e-05, "loss": 0.2702, "step": 2164 }, { "epoch": 2.2, "learning_rate": 1.929673154564823e-05, "loss": 0.2574, "step": 2165 }, { "epoch": 2.2, "learning_rate": 1.9295923760370183e-05, "loss": 0.2402, "step": 2166 }, { "epoch": 2.2, "learning_rate": 1.9295115528368113e-05, "loss": 0.2422, "step": 2167 }, { "epoch": 2.2, "learning_rate": 1.9294306849680866e-05, "loss": 0.228, "step": 2168 }, { "epoch": 2.2, "learning_rate": 1.92934977243473e-05, "loss": 0.229, "step": 2169 }, { "epoch": 2.2, "learning_rate": 1.92926881524063e-05, "loss": 0.25, "step": 2170 }, { "epoch": 2.2, "learning_rate": 1.9291878133896767e-05, "loss": 0.2595, "step": 2171 }, { "epoch": 2.2, "learning_rate": 1.929106766885763e-05, "loss": 0.2251, "step": 2172 }, { "epoch": 2.2, "learning_rate": 1.929025675732784e-05, "loss": 0.2413, "step": 2173 }, { "epoch": 2.21, "learning_rate": 1.9289445399346363e-05, "loss": 0.2512, "step": 2174 }, { "epoch": 2.21, "learning_rate": 1.9288633594952185e-05, "loss": 0.2244, "step": 2175 }, { "epoch": 2.21, "learning_rate": 1.9287821344184326e-05, "loss": 0.2485, "step": 2176 }, { "epoch": 2.21, "learning_rate": 1.9287008647081816e-05, "loss": 0.2613, "step": 2177 }, { "epoch": 2.21, "learning_rate": 1.928619550368371e-05, "loss": 0.2774, "step": 2178 }, { "epoch": 2.21, "learning_rate": 1.928538191402908e-05, "loss": 0.2484, "step": 2179 }, { "epoch": 2.21, "learning_rate": 1.9284567878157032e-05, "loss": 0.239, "step": 2180 }, { "epoch": 2.21, "learning_rate": 1.928375339610668e-05, "loss": 0.2359, "step": 2181 }, { "epoch": 2.21, "learning_rate": 1.9282938467917163e-05, "loss": 0.2633, "step": 2182 }, { "epoch": 2.22, "learning_rate": 1.928212309362765e-05, "loss": 0.2498, "step": 2183 }, { "epoch": 2.22, "learning_rate": 1.928130727327732e-05, "loss": 0.253, "step": 2184 }, { "epoch": 2.22, "learning_rate": 1.9280491006905377e-05, "loss": 0.2372, "step": 2185 }, { "epoch": 2.22, "learning_rate": 1.927967429455105e-05, "loss": 0.2694, "step": 2186 }, { "epoch": 2.22, "learning_rate": 1.927885713625358e-05, "loss": 0.2043, "step": 2187 }, { "epoch": 2.22, "learning_rate": 1.9278039532052247e-05, "loss": 0.2382, "step": 2188 }, { "epoch": 2.22, "learning_rate": 1.9277221481986338e-05, "loss": 0.2472, "step": 2189 }, { "epoch": 2.22, "learning_rate": 1.927640298609516e-05, "loss": 0.2362, "step": 2190 }, { "epoch": 2.22, "learning_rate": 1.9275584044418054e-05, "loss": 0.2342, "step": 2191 }, { "epoch": 2.22, "learning_rate": 1.927476465699437e-05, "loss": 0.2234, "step": 2192 }, { "epoch": 2.23, "learning_rate": 1.9273944823863485e-05, "loss": 0.2625, "step": 2193 }, { "epoch": 2.23, "learning_rate": 1.9273124545064796e-05, "loss": 0.2512, "step": 2194 }, { "epoch": 2.23, "learning_rate": 1.9272303820637727e-05, "loss": 0.2458, "step": 2195 }, { "epoch": 2.23, "learning_rate": 1.9271482650621715e-05, "loss": 0.2702, "step": 2196 }, { "epoch": 2.23, "learning_rate": 1.927066103505622e-05, "loss": 0.2323, "step": 2197 }, { "epoch": 2.23, "learning_rate": 1.9269838973980726e-05, "loss": 0.2656, "step": 2198 }, { "epoch": 2.23, "learning_rate": 1.9269016467434747e-05, "loss": 0.2235, "step": 2199 }, { "epoch": 2.23, "learning_rate": 1.9268193515457797e-05, "loss": 0.2378, "step": 2200 }, { "epoch": 2.23, "learning_rate": 1.926737011808943e-05, "loss": 0.2576, "step": 2201 }, { "epoch": 2.23, "learning_rate": 1.9266546275369217e-05, "loss": 0.2326, "step": 2202 }, { "epoch": 2.24, "learning_rate": 1.9265721987336743e-05, "loss": 0.2649, "step": 2203 }, { "epoch": 2.24, "learning_rate": 1.9264897254031623e-05, "loss": 0.2787, "step": 2204 }, { "epoch": 2.24, "learning_rate": 1.9264072075493486e-05, "loss": 0.2425, "step": 2205 }, { "epoch": 2.24, "learning_rate": 1.9263246451761995e-05, "loss": 0.2296, "step": 2206 }, { "epoch": 2.24, "learning_rate": 1.9262420382876817e-05, "loss": 0.2504, "step": 2207 }, { "epoch": 2.24, "learning_rate": 1.9261593868877662e-05, "loss": 0.252, "step": 2208 }, { "epoch": 2.24, "learning_rate": 1.9260766909804235e-05, "loss": 0.2465, "step": 2209 }, { "epoch": 2.24, "learning_rate": 1.9259939505696283e-05, "loss": 0.2285, "step": 2210 }, { "epoch": 2.24, "learning_rate": 1.9259111656593564e-05, "loss": 0.2048, "step": 2211 }, { "epoch": 2.24, "learning_rate": 1.9258283362535866e-05, "loss": 0.2519, "step": 2212 }, { "epoch": 2.25, "learning_rate": 1.9257454623562993e-05, "loss": 0.241, "step": 2213 }, { "epoch": 2.25, "learning_rate": 1.925662543971477e-05, "loss": 0.2544, "step": 2214 }, { "epoch": 2.25, "learning_rate": 1.925579581103104e-05, "loss": 0.2453, "step": 2215 }, { "epoch": 2.25, "learning_rate": 1.9254965737551678e-05, "loss": 0.2379, "step": 2216 }, { "epoch": 2.25, "learning_rate": 1.9254135219316564e-05, "loss": 0.2209, "step": 2217 }, { "epoch": 2.25, "learning_rate": 1.9253304256365623e-05, "loss": 0.2296, "step": 2218 }, { "epoch": 2.25, "learning_rate": 1.9252472848738776e-05, "loss": 0.1955, "step": 2219 }, { "epoch": 2.25, "learning_rate": 1.9251640996475986e-05, "loss": 0.2514, "step": 2220 }, { "epoch": 2.25, "learning_rate": 1.9250808699617224e-05, "loss": 0.2296, "step": 2221 }, { "epoch": 2.25, "learning_rate": 1.9249975958202483e-05, "loss": 0.2336, "step": 2222 }, { "epoch": 2.26, "learning_rate": 1.9249142772271786e-05, "loss": 0.2569, "step": 2223 }, { "epoch": 2.26, "learning_rate": 1.9248309141865172e-05, "loss": 0.2747, "step": 2224 }, { "epoch": 2.26, "learning_rate": 1.9247475067022705e-05, "loss": 0.2554, "step": 2225 }, { "epoch": 2.26, "learning_rate": 1.924664054778446e-05, "loss": 0.197, "step": 2226 }, { "epoch": 2.26, "learning_rate": 1.9245805584190545e-05, "loss": 0.2492, "step": 2227 }, { "epoch": 2.26, "learning_rate": 1.9244970176281083e-05, "loss": 0.2463, "step": 2228 }, { "epoch": 2.26, "learning_rate": 1.9244134324096223e-05, "loss": 0.2354, "step": 2229 }, { "epoch": 2.26, "learning_rate": 1.924329802767613e-05, "loss": 0.2459, "step": 2230 }, { "epoch": 2.26, "learning_rate": 1.9242461287060995e-05, "loss": 0.2538, "step": 2231 }, { "epoch": 2.26, "learning_rate": 1.9241624102291027e-05, "loss": 0.2668, "step": 2232 }, { "epoch": 2.27, "learning_rate": 1.924078647340646e-05, "loss": 0.2912, "step": 2233 }, { "epoch": 2.27, "learning_rate": 1.9239948400447545e-05, "loss": 0.2322, "step": 2234 }, { "epoch": 2.27, "learning_rate": 1.9239109883454552e-05, "loss": 0.2708, "step": 2235 }, { "epoch": 2.27, "learning_rate": 1.9238270922467784e-05, "loss": 0.2364, "step": 2236 }, { "epoch": 2.27, "learning_rate": 1.9237431517527554e-05, "loss": 0.2423, "step": 2237 }, { "epoch": 2.27, "learning_rate": 1.9236591668674202e-05, "loss": 0.2477, "step": 2238 }, { "epoch": 2.27, "learning_rate": 1.9235751375948085e-05, "loss": 0.2387, "step": 2239 }, { "epoch": 2.27, "learning_rate": 1.923491063938959e-05, "loss": 0.2165, "step": 2240 }, { "epoch": 2.27, "learning_rate": 1.9234069459039116e-05, "loss": 0.243, "step": 2241 }, { "epoch": 2.27, "learning_rate": 1.9233227834937084e-05, "loss": 0.2488, "step": 2242 }, { "epoch": 2.28, "learning_rate": 1.923238576712394e-05, "loss": 0.2676, "step": 2243 }, { "epoch": 2.28, "learning_rate": 1.9231543255640156e-05, "loss": 0.2528, "step": 2244 }, { "epoch": 2.28, "learning_rate": 1.9230700300526214e-05, "loss": 0.236, "step": 2245 }, { "epoch": 2.28, "learning_rate": 1.922985690182262e-05, "loss": 0.2661, "step": 2246 }, { "epoch": 2.28, "learning_rate": 1.9229013059569913e-05, "loss": 0.242, "step": 2247 }, { "epoch": 2.28, "learning_rate": 1.922816877380864e-05, "loss": 0.247, "step": 2248 }, { "epoch": 2.28, "learning_rate": 1.9227324044579368e-05, "loss": 0.2412, "step": 2249 }, { "epoch": 2.28, "learning_rate": 1.9226478871922705e-05, "loss": 0.2585, "step": 2250 }, { "epoch": 2.28, "learning_rate": 1.9225633255879256e-05, "loss": 0.2222, "step": 2251 }, { "epoch": 2.29, "learning_rate": 1.922478719648966e-05, "loss": 0.2696, "step": 2252 }, { "epoch": 2.29, "learning_rate": 1.9223940693794577e-05, "loss": 0.2412, "step": 2253 }, { "epoch": 2.29, "learning_rate": 1.9223093747834682e-05, "loss": 0.2401, "step": 2254 }, { "epoch": 2.29, "learning_rate": 1.9222246358650683e-05, "loss": 0.2204, "step": 2255 }, { "epoch": 2.29, "learning_rate": 1.9221398526283295e-05, "loss": 0.2559, "step": 2256 }, { "epoch": 2.29, "learning_rate": 1.9220550250773266e-05, "loss": 0.2369, "step": 2257 }, { "epoch": 2.29, "learning_rate": 1.921970153216136e-05, "loss": 0.2639, "step": 2258 }, { "epoch": 2.29, "learning_rate": 1.9218852370488357e-05, "loss": 0.2403, "step": 2259 }, { "epoch": 2.29, "learning_rate": 1.9218002765795078e-05, "loss": 0.2395, "step": 2260 }, { "epoch": 2.29, "learning_rate": 1.9217152718122334e-05, "loss": 0.27, "step": 2261 }, { "epoch": 2.3, "learning_rate": 1.9216302227510986e-05, "loss": 0.2341, "step": 2262 }, { "epoch": 2.3, "learning_rate": 1.9215451294001902e-05, "loss": 0.2447, "step": 2263 }, { "epoch": 2.3, "learning_rate": 1.921459991763598e-05, "loss": 0.239, "step": 2264 }, { "epoch": 2.3, "learning_rate": 1.9213748098454122e-05, "loss": 0.2229, "step": 2265 }, { "epoch": 2.3, "learning_rate": 1.9212895836497277e-05, "loss": 0.2198, "step": 2266 }, { "epoch": 2.3, "learning_rate": 1.921204313180639e-05, "loss": 0.2595, "step": 2267 }, { "epoch": 2.3, "learning_rate": 1.921118998442244e-05, "loss": 0.2576, "step": 2268 }, { "epoch": 2.3, "learning_rate": 1.921033639438643e-05, "loss": 0.2561, "step": 2269 }, { "epoch": 2.3, "learning_rate": 1.920948236173938e-05, "loss": 0.2676, "step": 2270 }, { "epoch": 2.3, "learning_rate": 1.9208627886522326e-05, "loss": 0.2874, "step": 2271 }, { "epoch": 2.31, "learning_rate": 1.9207772968776338e-05, "loss": 0.2456, "step": 2272 }, { "epoch": 2.31, "learning_rate": 1.9206917608542493e-05, "loss": 0.2379, "step": 2273 }, { "epoch": 2.31, "learning_rate": 1.92060618058619e-05, "loss": 0.2622, "step": 2274 }, { "epoch": 2.31, "learning_rate": 1.9205205560775683e-05, "loss": 0.2533, "step": 2275 }, { "epoch": 2.31, "learning_rate": 1.9204348873324995e-05, "loss": 0.2614, "step": 2276 }, { "epoch": 2.31, "learning_rate": 1.9203491743551e-05, "loss": 0.2208, "step": 2277 }, { "epoch": 2.31, "learning_rate": 1.9202634171494883e-05, "loss": 0.24, "step": 2278 }, { "epoch": 2.31, "learning_rate": 1.9201776157197867e-05, "loss": 0.2253, "step": 2279 }, { "epoch": 2.31, "learning_rate": 1.9200917700701176e-05, "loss": 0.2461, "step": 2280 }, { "epoch": 2.31, "learning_rate": 1.920005880204607e-05, "loss": 0.2698, "step": 2281 }, { "epoch": 2.32, "learning_rate": 1.9199199461273816e-05, "loss": 0.2394, "step": 2282 }, { "epoch": 2.32, "learning_rate": 1.9198339678425722e-05, "loss": 0.2544, "step": 2283 }, { "epoch": 2.32, "learning_rate": 1.9197479453543097e-05, "loss": 0.2485, "step": 2284 }, { "epoch": 2.32, "learning_rate": 1.9196618786667282e-05, "loss": 0.2719, "step": 2285 }, { "epoch": 2.32, "learning_rate": 1.9195757677839633e-05, "loss": 0.2434, "step": 2286 }, { "epoch": 2.32, "learning_rate": 1.919489612710154e-05, "loss": 0.2423, "step": 2287 }, { "epoch": 2.32, "learning_rate": 1.91940341344944e-05, "loss": 0.2697, "step": 2288 }, { "epoch": 2.32, "learning_rate": 1.9193171700059638e-05, "loss": 0.2259, "step": 2289 }, { "epoch": 2.32, "learning_rate": 1.91923088238387e-05, "loss": 0.2607, "step": 2290 }, { "epoch": 2.32, "learning_rate": 1.919144550587305e-05, "loss": 0.2758, "step": 2291 }, { "epoch": 2.33, "learning_rate": 1.919058174620418e-05, "loss": 0.2675, "step": 2292 }, { "epoch": 2.33, "learning_rate": 1.918971754487359e-05, "loss": 0.2445, "step": 2293 }, { "epoch": 2.33, "learning_rate": 1.9188852901922817e-05, "loss": 0.2499, "step": 2294 }, { "epoch": 2.33, "learning_rate": 1.9187987817393416e-05, "loss": 0.2808, "step": 2295 }, { "epoch": 2.33, "learning_rate": 1.9187122291326948e-05, "loss": 0.2331, "step": 2296 }, { "epoch": 2.33, "learning_rate": 1.9186256323765014e-05, "loss": 0.2734, "step": 2297 }, { "epoch": 2.33, "learning_rate": 1.918538991474923e-05, "loss": 0.244, "step": 2298 }, { "epoch": 2.33, "learning_rate": 1.918452306432123e-05, "loss": 0.2572, "step": 2299 }, { "epoch": 2.33, "learning_rate": 1.9183655772522672e-05, "loss": 0.2417, "step": 2300 }, { "epoch": 2.33, "learning_rate": 1.918278803939523e-05, "loss": 0.2222, "step": 2301 }, { "epoch": 2.34, "learning_rate": 1.918191986498061e-05, "loss": 0.2226, "step": 2302 }, { "epoch": 2.34, "learning_rate": 1.9181051249320527e-05, "loss": 0.2465, "step": 2303 }, { "epoch": 2.34, "learning_rate": 1.9180182192456728e-05, "loss": 0.2846, "step": 2304 }, { "epoch": 2.34, "learning_rate": 1.9179312694430977e-05, "loss": 0.2241, "step": 2305 }, { "epoch": 2.34, "learning_rate": 1.9178442755285053e-05, "loss": 0.2683, "step": 2306 }, { "epoch": 2.34, "learning_rate": 1.9177572375060766e-05, "loss": 0.2486, "step": 2307 }, { "epoch": 2.34, "learning_rate": 1.917670155379994e-05, "loss": 0.2699, "step": 2308 }, { "epoch": 2.34, "learning_rate": 1.9175830291544426e-05, "loss": 0.2408, "step": 2309 }, { "epoch": 2.34, "learning_rate": 1.9174958588336095e-05, "loss": 0.2593, "step": 2310 }, { "epoch": 2.35, "learning_rate": 1.917408644421683e-05, "loss": 0.2425, "step": 2311 }, { "epoch": 2.35, "learning_rate": 1.917321385922855e-05, "loss": 0.2607, "step": 2312 }, { "epoch": 2.35, "learning_rate": 1.9172340833413185e-05, "loss": 0.2738, "step": 2313 }, { "epoch": 2.35, "learning_rate": 1.9171467366812687e-05, "loss": 0.2498, "step": 2314 }, { "epoch": 2.35, "learning_rate": 1.9170593459469034e-05, "loss": 0.2539, "step": 2315 }, { "epoch": 2.35, "learning_rate": 1.9169719111424222e-05, "loss": 0.2701, "step": 2316 }, { "epoch": 2.35, "learning_rate": 1.9168844322720267e-05, "loss": 0.233, "step": 2317 }, { "epoch": 2.35, "learning_rate": 1.9167969093399215e-05, "loss": 0.2752, "step": 2318 }, { "epoch": 2.35, "learning_rate": 1.9167093423503113e-05, "loss": 0.2512, "step": 2319 }, { "epoch": 2.35, "learning_rate": 1.9166217313074056e-05, "loss": 0.2705, "step": 2320 }, { "epoch": 2.36, "learning_rate": 1.9165340762154134e-05, "loss": 0.2535, "step": 2321 }, { "epoch": 2.36, "learning_rate": 1.916446377078548e-05, "loss": 0.2404, "step": 2322 }, { "epoch": 2.36, "learning_rate": 1.916358633901023e-05, "loss": 0.2565, "step": 2323 }, { "epoch": 2.36, "learning_rate": 1.9162708466870558e-05, "loss": 0.2498, "step": 2324 }, { "epoch": 2.36, "learning_rate": 1.9161830154408643e-05, "loss": 0.257, "step": 2325 }, { "epoch": 2.36, "learning_rate": 1.9160951401666704e-05, "loss": 0.2164, "step": 2326 }, { "epoch": 2.36, "learning_rate": 1.9160072208686956e-05, "loss": 0.2488, "step": 2327 }, { "epoch": 2.36, "learning_rate": 1.9159192575511663e-05, "loss": 0.2475, "step": 2328 }, { "epoch": 2.36, "learning_rate": 1.915831250218309e-05, "loss": 0.2795, "step": 2329 }, { "epoch": 2.36, "learning_rate": 1.9157431988743527e-05, "loss": 0.2642, "step": 2330 }, { "epoch": 2.37, "learning_rate": 1.915655103523529e-05, "loss": 0.2397, "step": 2331 }, { "epoch": 2.37, "learning_rate": 1.915566964170072e-05, "loss": 0.2286, "step": 2332 }, { "epoch": 2.37, "learning_rate": 1.9154787808182168e-05, "loss": 0.2471, "step": 2333 }, { "epoch": 2.37, "learning_rate": 1.915390553472201e-05, "loss": 0.239, "step": 2334 }, { "epoch": 2.37, "learning_rate": 1.9153022821362646e-05, "loss": 0.2338, "step": 2335 }, { "epoch": 2.37, "learning_rate": 1.9152139668146493e-05, "loss": 0.2182, "step": 2336 }, { "epoch": 2.37, "learning_rate": 1.9151256075116e-05, "loss": 0.2561, "step": 2337 }, { "epoch": 2.37, "learning_rate": 1.915037204231362e-05, "loss": 0.259, "step": 2338 }, { "epoch": 2.37, "learning_rate": 1.9149487569781843e-05, "loss": 0.2469, "step": 2339 }, { "epoch": 2.37, "learning_rate": 1.9148602657563163e-05, "loss": 0.2532, "step": 2340 }, { "epoch": 2.38, "learning_rate": 1.9147717305700116e-05, "loss": 0.2431, "step": 2341 }, { "epoch": 2.38, "learning_rate": 1.9146831514235243e-05, "loss": 0.2719, "step": 2342 }, { "epoch": 2.38, "learning_rate": 1.9145945283211113e-05, "loss": 0.2405, "step": 2343 }, { "epoch": 2.38, "learning_rate": 1.9145058612670314e-05, "loss": 0.2389, "step": 2344 }, { "epoch": 2.38, "learning_rate": 1.9144171502655456e-05, "loss": 0.2459, "step": 2345 }, { "epoch": 2.38, "learning_rate": 1.9143283953209168e-05, "loss": 0.2571, "step": 2346 }, { "epoch": 2.38, "learning_rate": 1.9142395964374107e-05, "loss": 0.2412, "step": 2347 }, { "epoch": 2.38, "learning_rate": 1.914150753619294e-05, "loss": 0.2225, "step": 2348 }, { "epoch": 2.38, "learning_rate": 1.9140618668708365e-05, "loss": 0.2404, "step": 2349 }, { "epoch": 2.38, "learning_rate": 1.9139729361963096e-05, "loss": 0.22, "step": 2350 }, { "epoch": 2.39, "learning_rate": 1.9138839615999873e-05, "loss": 0.2419, "step": 2351 }, { "epoch": 2.39, "learning_rate": 1.9137949430861448e-05, "loss": 0.263, "step": 2352 }, { "epoch": 2.39, "learning_rate": 1.9137058806590602e-05, "loss": 0.2423, "step": 2353 }, { "epoch": 2.39, "learning_rate": 1.9136167743230135e-05, "loss": 0.238, "step": 2354 }, { "epoch": 2.39, "learning_rate": 1.9135276240822872e-05, "loss": 0.2747, "step": 2355 }, { "epoch": 2.39, "learning_rate": 1.9134384299411643e-05, "loss": 0.2515, "step": 2356 }, { "epoch": 2.39, "learning_rate": 1.913349191903932e-05, "loss": 0.2562, "step": 2357 }, { "epoch": 2.39, "learning_rate": 1.9132599099748793e-05, "loss": 0.261, "step": 2358 }, { "epoch": 2.39, "learning_rate": 1.9131705841582953e-05, "loss": 0.3108, "step": 2359 }, { "epoch": 2.39, "learning_rate": 1.913081214458474e-05, "loss": 0.2468, "step": 2360 }, { "epoch": 2.4, "learning_rate": 1.9129918008797088e-05, "loss": 0.2523, "step": 2361 }, { "epoch": 2.4, "learning_rate": 1.9129023434262976e-05, "loss": 0.2524, "step": 2362 }, { "epoch": 2.4, "learning_rate": 1.9128128421025388e-05, "loss": 0.2194, "step": 2363 }, { "epoch": 2.4, "learning_rate": 1.912723296912734e-05, "loss": 0.2865, "step": 2364 }, { "epoch": 2.4, "learning_rate": 1.9126337078611855e-05, "loss": 0.2554, "step": 2365 }, { "epoch": 2.4, "learning_rate": 1.9125440749521994e-05, "loss": 0.2709, "step": 2366 }, { "epoch": 2.4, "learning_rate": 1.9124543981900827e-05, "loss": 0.3, "step": 2367 }, { "epoch": 2.4, "learning_rate": 1.9123646775791452e-05, "loss": 0.2617, "step": 2368 }, { "epoch": 2.4, "learning_rate": 1.9122749131236984e-05, "loss": 0.2908, "step": 2369 }, { "epoch": 2.4, "learning_rate": 1.9121851048280556e-05, "loss": 0.2557, "step": 2370 }, { "epoch": 2.41, "learning_rate": 1.9120952526965333e-05, "loss": 0.2461, "step": 2371 }, { "epoch": 2.41, "learning_rate": 1.9120053567334487e-05, "loss": 0.2775, "step": 2372 }, { "epoch": 2.41, "learning_rate": 1.9119154169431224e-05, "loss": 0.2378, "step": 2373 }, { "epoch": 2.41, "learning_rate": 1.9118254333298764e-05, "loss": 0.2521, "step": 2374 }, { "epoch": 2.41, "learning_rate": 1.911735405898035e-05, "loss": 0.2597, "step": 2375 }, { "epoch": 2.41, "learning_rate": 1.911645334651924e-05, "loss": 0.2419, "step": 2376 }, { "epoch": 2.41, "learning_rate": 1.9115552195958727e-05, "loss": 0.2755, "step": 2377 }, { "epoch": 2.41, "learning_rate": 1.911465060734211e-05, "loss": 0.2434, "step": 2378 }, { "epoch": 2.41, "learning_rate": 1.9113748580712718e-05, "loss": 0.2621, "step": 2379 }, { "epoch": 2.42, "learning_rate": 1.9112846116113902e-05, "loss": 0.2836, "step": 2380 }, { "epoch": 2.42, "learning_rate": 1.9111943213589024e-05, "loss": 0.2858, "step": 2381 }, { "epoch": 2.42, "learning_rate": 1.9111039873181478e-05, "loss": 0.2733, "step": 2382 }, { "epoch": 2.42, "learning_rate": 1.9110136094934674e-05, "loss": 0.2664, "step": 2383 }, { "epoch": 2.42, "learning_rate": 1.9109231878892047e-05, "loss": 0.2589, "step": 2384 }, { "epoch": 2.42, "learning_rate": 1.9108327225097046e-05, "loss": 0.2658, "step": 2385 }, { "epoch": 2.42, "learning_rate": 1.910742213359314e-05, "loss": 0.2559, "step": 2386 }, { "epoch": 2.42, "learning_rate": 1.910651660442384e-05, "loss": 0.2713, "step": 2387 }, { "epoch": 2.42, "learning_rate": 1.9105610637632646e-05, "loss": 0.2592, "step": 2388 }, { "epoch": 2.42, "learning_rate": 1.9104704233263103e-05, "loss": 0.248, "step": 2389 }, { "epoch": 2.43, "learning_rate": 1.9103797391358766e-05, "loss": 0.2746, "step": 2390 }, { "epoch": 2.43, "learning_rate": 1.9102890111963216e-05, "loss": 0.25, "step": 2391 }, { "epoch": 2.43, "learning_rate": 1.910198239512005e-05, "loss": 0.2688, "step": 2392 }, { "epoch": 2.43, "learning_rate": 1.9101074240872895e-05, "loss": 0.2317, "step": 2393 }, { "epoch": 2.43, "learning_rate": 1.910016564926539e-05, "loss": 0.2491, "step": 2394 }, { "epoch": 2.43, "learning_rate": 1.9099256620341193e-05, "loss": 0.2176, "step": 2395 }, { "epoch": 2.43, "learning_rate": 1.9098347154143996e-05, "loss": 0.2416, "step": 2396 }, { "epoch": 2.43, "learning_rate": 1.9097437250717503e-05, "loss": 0.295, "step": 2397 }, { "epoch": 2.43, "learning_rate": 1.9096526910105437e-05, "loss": 0.2344, "step": 2398 }, { "epoch": 2.43, "learning_rate": 1.9095616132351545e-05, "loss": 0.2384, "step": 2399 }, { "epoch": 2.44, "learning_rate": 1.9094704917499596e-05, "loss": 0.2471, "step": 2400 }, { "epoch": 2.44, "learning_rate": 1.909379326559339e-05, "loss": 0.2285, "step": 2401 }, { "epoch": 2.44, "learning_rate": 1.909288117667672e-05, "loss": 0.2608, "step": 2402 }, { "epoch": 2.44, "learning_rate": 1.9091968650793423e-05, "loss": 0.2284, "step": 2403 }, { "epoch": 2.44, "learning_rate": 1.9091055687987357e-05, "loss": 0.2433, "step": 2404 }, { "epoch": 2.44, "learning_rate": 1.909014228830239e-05, "loss": 0.2656, "step": 2405 }, { "epoch": 2.44, "learning_rate": 1.9089228451782417e-05, "loss": 0.272, "step": 2406 }, { "epoch": 2.44, "learning_rate": 1.9088314178471353e-05, "loss": 0.2304, "step": 2407 }, { "epoch": 2.44, "learning_rate": 1.9087399468413137e-05, "loss": 0.2732, "step": 2408 }, { "epoch": 2.44, "learning_rate": 1.9086484321651723e-05, "loss": 0.2614, "step": 2409 }, { "epoch": 2.45, "learning_rate": 1.908556873823109e-05, "loss": 0.2265, "step": 2410 }, { "epoch": 2.45, "learning_rate": 1.9084652718195237e-05, "loss": 0.2636, "step": 2411 }, { "epoch": 2.45, "learning_rate": 1.908373626158819e-05, "loss": 0.3056, "step": 2412 }, { "epoch": 2.45, "learning_rate": 1.908281936845398e-05, "loss": 0.23, "step": 2413 }, { "epoch": 2.45, "learning_rate": 1.9081902038836674e-05, "loss": 0.2222, "step": 2414 }, { "epoch": 2.45, "learning_rate": 1.9080984272780356e-05, "loss": 0.2393, "step": 2415 }, { "epoch": 2.45, "learning_rate": 1.9080066070329133e-05, "loss": 0.2498, "step": 2416 }, { "epoch": 2.45, "learning_rate": 1.9079147431527122e-05, "loss": 0.2237, "step": 2417 }, { "epoch": 2.45, "learning_rate": 1.9078228356418474e-05, "loss": 0.2398, "step": 2418 }, { "epoch": 2.45, "learning_rate": 1.9077308845047356e-05, "loss": 0.2388, "step": 2419 }, { "epoch": 2.46, "learning_rate": 1.9076388897457957e-05, "loss": 0.234, "step": 2420 }, { "epoch": 2.46, "learning_rate": 1.907546851369448e-05, "loss": 0.2277, "step": 2421 }, { "epoch": 2.46, "learning_rate": 1.907454769380116e-05, "loss": 0.2442, "step": 2422 }, { "epoch": 2.46, "learning_rate": 1.907362643782225e-05, "loss": 0.2611, "step": 2423 }, { "epoch": 2.46, "learning_rate": 1.907270474580202e-05, "loss": 0.2584, "step": 2424 }, { "epoch": 2.46, "learning_rate": 1.9071782617784754e-05, "loss": 0.2528, "step": 2425 }, { "epoch": 2.46, "learning_rate": 1.907086005381478e-05, "loss": 0.2589, "step": 2426 }, { "epoch": 2.46, "learning_rate": 1.9069937053936422e-05, "loss": 0.2642, "step": 2427 }, { "epoch": 2.46, "learning_rate": 1.9069013618194042e-05, "loss": 0.2394, "step": 2428 }, { "epoch": 2.46, "learning_rate": 1.9068089746632012e-05, "loss": 0.2623, "step": 2429 }, { "epoch": 2.47, "learning_rate": 1.9067165439294733e-05, "loss": 0.2271, "step": 2430 }, { "epoch": 2.47, "learning_rate": 1.9066240696226622e-05, "loss": 0.2606, "step": 2431 }, { "epoch": 2.47, "learning_rate": 1.9065315517472116e-05, "loss": 0.2336, "step": 2432 }, { "epoch": 2.47, "learning_rate": 1.9064389903075676e-05, "loss": 0.244, "step": 2433 }, { "epoch": 2.47, "learning_rate": 1.906346385308179e-05, "loss": 0.2468, "step": 2434 }, { "epoch": 2.47, "learning_rate": 1.9062537367534952e-05, "loss": 0.2578, "step": 2435 }, { "epoch": 2.47, "learning_rate": 1.9061610446479687e-05, "loss": 0.267, "step": 2436 }, { "epoch": 2.47, "learning_rate": 1.9060683089960542e-05, "loss": 0.2269, "step": 2437 }, { "epoch": 2.47, "learning_rate": 1.9059755298022078e-05, "loss": 0.2646, "step": 2438 }, { "epoch": 2.47, "learning_rate": 1.9058827070708886e-05, "loss": 0.2341, "step": 2439 }, { "epoch": 2.48, "learning_rate": 1.9057898408065567e-05, "loss": 0.2565, "step": 2440 }, { "epoch": 2.48, "learning_rate": 1.905696931013675e-05, "loss": 0.2636, "step": 2441 }, { "epoch": 2.48, "learning_rate": 1.905603977696709e-05, "loss": 0.2864, "step": 2442 }, { "epoch": 2.48, "learning_rate": 1.9055109808601248e-05, "loss": 0.2768, "step": 2443 }, { "epoch": 2.48, "learning_rate": 1.9054179405083916e-05, "loss": 0.2748, "step": 2444 }, { "epoch": 2.48, "learning_rate": 1.9053248566459813e-05, "loss": 0.2323, "step": 2445 }, { "epoch": 2.48, "learning_rate": 1.9052317292773662e-05, "loss": 0.2411, "step": 2446 }, { "epoch": 2.48, "learning_rate": 1.905138558407022e-05, "loss": 0.2432, "step": 2447 }, { "epoch": 2.48, "learning_rate": 1.9050453440394264e-05, "loss": 0.2444, "step": 2448 }, { "epoch": 2.49, "learning_rate": 1.904952086179058e-05, "loss": 0.2417, "step": 2449 }, { "epoch": 2.49, "learning_rate": 1.9048587848303995e-05, "loss": 0.2221, "step": 2450 }, { "epoch": 2.49, "learning_rate": 1.904765439997934e-05, "loss": 0.2543, "step": 2451 }, { "epoch": 2.49, "learning_rate": 1.9046720516861473e-05, "loss": 0.2574, "step": 2452 }, { "epoch": 2.49, "learning_rate": 1.9045786198995275e-05, "loss": 0.2722, "step": 2453 }, { "epoch": 2.49, "learning_rate": 1.9044851446425642e-05, "loss": 0.2311, "step": 2454 }, { "epoch": 2.49, "learning_rate": 1.9043916259197496e-05, "loss": 0.2727, "step": 2455 }, { "epoch": 2.49, "learning_rate": 1.9042980637355775e-05, "loss": 0.2562, "step": 2456 }, { "epoch": 2.49, "learning_rate": 1.9042044580945447e-05, "loss": 0.245, "step": 2457 }, { "epoch": 2.49, "learning_rate": 1.904110809001149e-05, "loss": 0.2379, "step": 2458 }, { "epoch": 2.5, "learning_rate": 1.9040171164598914e-05, "loss": 0.2568, "step": 2459 }, { "epoch": 2.5, "learning_rate": 1.9039233804752738e-05, "loss": 0.2811, "step": 2460 }, { "epoch": 2.5, "learning_rate": 1.903829601051801e-05, "loss": 0.2561, "step": 2461 }, { "epoch": 2.5, "learning_rate": 1.9037357781939798e-05, "loss": 0.2495, "step": 2462 }, { "epoch": 2.5, "learning_rate": 1.9036419119063188e-05, "loss": 0.2775, "step": 2463 }, { "epoch": 2.5, "learning_rate": 1.9035480021933284e-05, "loss": 0.2642, "step": 2464 }, { "epoch": 2.5, "learning_rate": 1.9034540490595223e-05, "loss": 0.2335, "step": 2465 }, { "epoch": 2.5, "learning_rate": 1.903360052509415e-05, "loss": 0.2701, "step": 2466 }, { "epoch": 2.5, "learning_rate": 1.9032660125475238e-05, "loss": 0.2816, "step": 2467 }, { "epoch": 2.5, "learning_rate": 1.9031719291783678e-05, "loss": 0.2399, "step": 2468 }, { "epoch": 2.51, "learning_rate": 1.903077802406468e-05, "loss": 0.2292, "step": 2469 }, { "epoch": 2.51, "learning_rate": 1.902983632236348e-05, "loss": 0.2529, "step": 2470 }, { "epoch": 2.51, "learning_rate": 1.9028894186725334e-05, "loss": 0.2467, "step": 2471 }, { "epoch": 2.51, "learning_rate": 1.9027951617195515e-05, "loss": 0.2693, "step": 2472 }, { "epoch": 2.51, "learning_rate": 1.902700861381932e-05, "loss": 0.2597, "step": 2473 }, { "epoch": 2.51, "learning_rate": 1.9026065176642065e-05, "loss": 0.2392, "step": 2474 }, { "epoch": 2.51, "learning_rate": 1.9025121305709088e-05, "loss": 0.2463, "step": 2475 }, { "epoch": 2.51, "learning_rate": 1.9024177001065747e-05, "loss": 0.2505, "step": 2476 }, { "epoch": 2.51, "learning_rate": 1.9023232262757422e-05, "loss": 0.267, "step": 2477 }, { "epoch": 2.51, "learning_rate": 1.9022287090829515e-05, "loss": 0.2538, "step": 2478 }, { "epoch": 2.52, "learning_rate": 1.9021341485327445e-05, "loss": 0.2614, "step": 2479 }, { "epoch": 2.52, "learning_rate": 1.9020395446296654e-05, "loss": 0.2708, "step": 2480 }, { "epoch": 2.52, "learning_rate": 1.9019448973782606e-05, "loss": 0.2539, "step": 2481 }, { "epoch": 2.52, "learning_rate": 1.901850206783078e-05, "loss": 0.2715, "step": 2482 }, { "epoch": 2.52, "learning_rate": 1.9017554728486688e-05, "loss": 0.2399, "step": 2483 }, { "epoch": 2.52, "learning_rate": 1.901660695579585e-05, "loss": 0.2579, "step": 2484 }, { "epoch": 2.52, "learning_rate": 1.9015658749803814e-05, "loss": 0.2404, "step": 2485 }, { "epoch": 2.52, "learning_rate": 1.9014710110556146e-05, "loss": 0.2538, "step": 2486 }, { "epoch": 2.52, "learning_rate": 1.9013761038098435e-05, "loss": 0.2788, "step": 2487 }, { "epoch": 2.52, "learning_rate": 1.9012811532476287e-05, "loss": 0.2336, "step": 2488 }, { "epoch": 2.53, "learning_rate": 1.9011861593735334e-05, "loss": 0.2863, "step": 2489 }, { "epoch": 2.53, "learning_rate": 1.9010911221921224e-05, "loss": 0.2417, "step": 2490 }, { "epoch": 2.53, "learning_rate": 1.900996041707963e-05, "loss": 0.2699, "step": 2491 }, { "epoch": 2.53, "learning_rate": 1.9009009179256243e-05, "loss": 0.2381, "step": 2492 }, { "epoch": 2.53, "learning_rate": 1.9008057508496773e-05, "loss": 0.2346, "step": 2493 }, { "epoch": 2.53, "learning_rate": 1.900710540484696e-05, "loss": 0.2809, "step": 2494 }, { "epoch": 2.53, "learning_rate": 1.900615286835255e-05, "loss": 0.2309, "step": 2495 }, { "epoch": 2.53, "learning_rate": 1.9005199899059327e-05, "loss": 0.254, "step": 2496 }, { "epoch": 2.53, "learning_rate": 1.9004246497013078e-05, "loss": 0.2556, "step": 2497 }, { "epoch": 2.53, "learning_rate": 1.9003292662259625e-05, "loss": 0.255, "step": 2498 }, { "epoch": 2.54, "learning_rate": 1.90023383948448e-05, "loss": 0.2573, "step": 2499 }, { "epoch": 2.54, "learning_rate": 1.900138369481447e-05, "loss": 0.2311, "step": 2500 }, { "epoch": 2.54, "learning_rate": 1.900042856221451e-05, "loss": 0.2445, "step": 2501 }, { "epoch": 2.54, "learning_rate": 1.8999472997090814e-05, "loss": 0.2617, "step": 2502 }, { "epoch": 2.54, "learning_rate": 1.8998516999489308e-05, "loss": 0.2951, "step": 2503 }, { "epoch": 2.54, "learning_rate": 1.8997560569455934e-05, "loss": 0.2821, "step": 2504 }, { "epoch": 2.54, "learning_rate": 1.8996603707036655e-05, "loss": 0.2733, "step": 2505 }, { "epoch": 2.54, "learning_rate": 1.8995646412277448e-05, "loss": 0.2665, "step": 2506 }, { "epoch": 2.54, "learning_rate": 1.899468868522432e-05, "loss": 0.2844, "step": 2507 }, { "epoch": 2.54, "learning_rate": 1.89937305259233e-05, "loss": 0.2712, "step": 2508 }, { "epoch": 2.55, "learning_rate": 1.8992771934420428e-05, "loss": 0.25, "step": 2509 }, { "epoch": 2.55, "learning_rate": 1.8991812910761768e-05, "loss": 0.2403, "step": 2510 }, { "epoch": 2.55, "learning_rate": 1.8990853454993413e-05, "loss": 0.2534, "step": 2511 }, { "epoch": 2.55, "learning_rate": 1.8989893567161464e-05, "loss": 0.2695, "step": 2512 }, { "epoch": 2.55, "learning_rate": 1.8988933247312052e-05, "loss": 0.2407, "step": 2513 }, { "epoch": 2.55, "learning_rate": 1.8987972495491328e-05, "loss": 0.2327, "step": 2514 }, { "epoch": 2.55, "learning_rate": 1.898701131174546e-05, "loss": 0.295, "step": 2515 }, { "epoch": 2.55, "learning_rate": 1.8986049696120643e-05, "loss": 0.2411, "step": 2516 }, { "epoch": 2.55, "learning_rate": 1.898508764866308e-05, "loss": 0.262, "step": 2517 }, { "epoch": 2.56, "learning_rate": 1.898412516941901e-05, "loss": 0.277, "step": 2518 }, { "epoch": 2.56, "learning_rate": 1.898316225843468e-05, "loss": 0.2599, "step": 2519 }, { "epoch": 2.56, "learning_rate": 1.8982198915756366e-05, "loss": 0.2894, "step": 2520 }, { "epoch": 2.56, "learning_rate": 1.8981235141430365e-05, "loss": 0.2686, "step": 2521 }, { "epoch": 2.56, "learning_rate": 1.898027093550299e-05, "loss": 0.2676, "step": 2522 }, { "epoch": 2.56, "learning_rate": 1.8979306298020573e-05, "loss": 0.2345, "step": 2523 }, { "epoch": 2.56, "learning_rate": 1.897834122902948e-05, "loss": 0.2771, "step": 2524 }, { "epoch": 2.56, "learning_rate": 1.8977375728576078e-05, "loss": 0.258, "step": 2525 }, { "epoch": 2.56, "learning_rate": 1.8976409796706774e-05, "loss": 0.2294, "step": 2526 }, { "epoch": 2.56, "learning_rate": 1.897544343346798e-05, "loss": 0.2476, "step": 2527 }, { "epoch": 2.57, "learning_rate": 1.8974476638906138e-05, "loss": 0.2761, "step": 2528 }, { "epoch": 2.57, "learning_rate": 1.8973509413067707e-05, "loss": 0.2258, "step": 2529 }, { "epoch": 2.57, "learning_rate": 1.8972541755999166e-05, "loss": 0.2374, "step": 2530 }, { "epoch": 2.57, "learning_rate": 1.897157366774703e-05, "loss": 0.297, "step": 2531 }, { "epoch": 2.57, "learning_rate": 1.89706051483578e-05, "loss": 0.2165, "step": 2532 }, { "epoch": 2.57, "learning_rate": 1.8969636197878034e-05, "loss": 0.25, "step": 2533 }, { "epoch": 2.57, "learning_rate": 1.8968666816354292e-05, "loss": 0.2677, "step": 2534 }, { "epoch": 2.57, "learning_rate": 1.8967697003833156e-05, "loss": 0.2894, "step": 2535 }, { "epoch": 2.57, "learning_rate": 1.8966726760361233e-05, "loss": 0.2328, "step": 2536 }, { "epoch": 2.57, "learning_rate": 1.896575608598515e-05, "loss": 0.2459, "step": 2537 }, { "epoch": 2.58, "learning_rate": 1.8964784980751556e-05, "loss": 0.2321, "step": 2538 }, { "epoch": 2.58, "learning_rate": 1.8963813444707115e-05, "loss": 0.2647, "step": 2539 }, { "epoch": 2.58, "learning_rate": 1.8962841477898513e-05, "loss": 0.2587, "step": 2540 }, { "epoch": 2.58, "learning_rate": 1.896186908037246e-05, "loss": 0.2037, "step": 2541 }, { "epoch": 2.58, "learning_rate": 1.896089625217569e-05, "loss": 0.271, "step": 2542 }, { "epoch": 2.58, "learning_rate": 1.895992299335495e-05, "loss": 0.2661, "step": 2543 }, { "epoch": 2.58, "learning_rate": 1.8958949303957007e-05, "loss": 0.2323, "step": 2544 }, { "epoch": 2.58, "learning_rate": 1.8957975184028657e-05, "loss": 0.2902, "step": 2545 }, { "epoch": 2.58, "learning_rate": 1.8957000633616713e-05, "loss": 0.2512, "step": 2546 }, { "epoch": 2.58, "learning_rate": 1.8956025652768005e-05, "loss": 0.2747, "step": 2547 }, { "epoch": 2.59, "learning_rate": 1.8955050241529386e-05, "loss": 0.2347, "step": 2548 }, { "epoch": 2.59, "learning_rate": 1.8954074399947735e-05, "loss": 0.2488, "step": 2549 }, { "epoch": 2.59, "learning_rate": 1.8953098128069943e-05, "loss": 0.2716, "step": 2550 }, { "epoch": 2.59, "learning_rate": 1.8952121425942923e-05, "loss": 0.2407, "step": 2551 }, { "epoch": 2.59, "learning_rate": 1.8951144293613623e-05, "loss": 0.2426, "step": 2552 }, { "epoch": 2.59, "learning_rate": 1.8950166731128984e-05, "loss": 0.2598, "step": 2553 }, { "epoch": 2.59, "learning_rate": 1.8949188738535994e-05, "loss": 0.2464, "step": 2554 }, { "epoch": 2.59, "learning_rate": 1.894821031588165e-05, "loss": 0.2472, "step": 2555 }, { "epoch": 2.59, "learning_rate": 1.894723146321297e-05, "loss": 0.217, "step": 2556 }, { "epoch": 2.59, "learning_rate": 1.8946252180576993e-05, "loss": 0.2258, "step": 2557 }, { "epoch": 2.6, "learning_rate": 1.8945272468020777e-05, "loss": 0.2542, "step": 2558 }, { "epoch": 2.6, "learning_rate": 1.894429232559141e-05, "loss": 0.2674, "step": 2559 }, { "epoch": 2.6, "learning_rate": 1.8943311753335988e-05, "loss": 0.2615, "step": 2560 }, { "epoch": 2.6, "learning_rate": 1.8942330751301635e-05, "loss": 0.2638, "step": 2561 }, { "epoch": 2.6, "learning_rate": 1.894134931953549e-05, "loss": 0.2885, "step": 2562 }, { "epoch": 2.6, "learning_rate": 1.8940367458084724e-05, "loss": 0.2648, "step": 2563 }, { "epoch": 2.6, "learning_rate": 1.8939385166996516e-05, "loss": 0.2326, "step": 2564 }, { "epoch": 2.6, "learning_rate": 1.8938402446318072e-05, "loss": 0.2971, "step": 2565 }, { "epoch": 2.6, "learning_rate": 1.893741929609662e-05, "loss": 0.2446, "step": 2566 }, { "epoch": 2.6, "learning_rate": 1.8936435716379402e-05, "loss": 0.2837, "step": 2567 }, { "epoch": 2.61, "learning_rate": 1.8935451707213687e-05, "loss": 0.2831, "step": 2568 }, { "epoch": 2.61, "learning_rate": 1.893446726864676e-05, "loss": 0.2275, "step": 2569 }, { "epoch": 2.61, "learning_rate": 1.8933482400725936e-05, "loss": 0.2314, "step": 2570 }, { "epoch": 2.61, "learning_rate": 1.8932497103498536e-05, "loss": 0.2475, "step": 2571 }, { "epoch": 2.61, "learning_rate": 1.8931511377011915e-05, "loss": 0.2299, "step": 2572 }, { "epoch": 2.61, "learning_rate": 1.8930525221313442e-05, "loss": 0.2805, "step": 2573 }, { "epoch": 2.61, "learning_rate": 1.89295386364505e-05, "loss": 0.2672, "step": 2574 }, { "epoch": 2.61, "learning_rate": 1.892855162247051e-05, "loss": 0.2313, "step": 2575 }, { "epoch": 2.61, "learning_rate": 1.8927564179420897e-05, "loss": 0.2722, "step": 2576 }, { "epoch": 2.61, "learning_rate": 1.892657630734912e-05, "loss": 0.2396, "step": 2577 }, { "epoch": 2.62, "learning_rate": 1.8925588006302648e-05, "loss": 0.2525, "step": 2578 }, { "epoch": 2.62, "learning_rate": 1.8924599276328974e-05, "loss": 0.2835, "step": 2579 }, { "epoch": 2.62, "learning_rate": 1.8923610117475616e-05, "loss": 0.2307, "step": 2580 }, { "epoch": 2.62, "learning_rate": 1.8922620529790105e-05, "loss": 0.2316, "step": 2581 }, { "epoch": 2.62, "learning_rate": 1.892163051332e-05, "loss": 0.2754, "step": 2582 }, { "epoch": 2.62, "learning_rate": 1.8920640068112873e-05, "loss": 0.2648, "step": 2583 }, { "epoch": 2.62, "learning_rate": 1.891964919421632e-05, "loss": 0.2386, "step": 2584 }, { "epoch": 2.62, "learning_rate": 1.8918657891677968e-05, "loss": 0.2515, "step": 2585 }, { "epoch": 2.62, "learning_rate": 1.8917666160545446e-05, "loss": 0.2397, "step": 2586 }, { "epoch": 2.63, "learning_rate": 1.8916674000866413e-05, "loss": 0.2459, "step": 2587 }, { "epoch": 2.63, "learning_rate": 1.891568141268855e-05, "loss": 0.2426, "step": 2588 }, { "epoch": 2.63, "learning_rate": 1.8914688396059556e-05, "loss": 0.2598, "step": 2589 }, { "epoch": 2.63, "learning_rate": 1.8913694951027155e-05, "loss": 0.252, "step": 2590 }, { "epoch": 2.63, "learning_rate": 1.8912701077639083e-05, "loss": 0.2473, "step": 2591 }, { "epoch": 2.63, "learning_rate": 1.8911706775943105e-05, "loss": 0.2451, "step": 2592 }, { "epoch": 2.63, "learning_rate": 1.8910712045986995e-05, "loss": 0.2868, "step": 2593 }, { "epoch": 2.63, "learning_rate": 1.890971688781857e-05, "loss": 0.267, "step": 2594 }, { "epoch": 2.63, "learning_rate": 1.8908721301485645e-05, "loss": 0.2822, "step": 2595 }, { "epoch": 2.63, "learning_rate": 1.890772528703606e-05, "loss": 0.279, "step": 2596 }, { "epoch": 2.64, "learning_rate": 1.8906728844517683e-05, "loss": 0.2783, "step": 2597 }, { "epoch": 2.64, "learning_rate": 1.8905731973978404e-05, "loss": 0.2229, "step": 2598 }, { "epoch": 2.64, "learning_rate": 1.8904734675466117e-05, "loss": 0.2863, "step": 2599 }, { "epoch": 2.64, "learning_rate": 1.890373694902876e-05, "loss": 0.2295, "step": 2600 }, { "epoch": 2.64, "learning_rate": 1.8902738794714275e-05, "loss": 0.2316, "step": 2601 }, { "epoch": 2.64, "learning_rate": 1.8901740212570628e-05, "loss": 0.2424, "step": 2602 }, { "epoch": 2.64, "learning_rate": 1.8900741202645808e-05, "loss": 0.2584, "step": 2603 }, { "epoch": 2.64, "learning_rate": 1.8899741764987825e-05, "loss": 0.2638, "step": 2604 }, { "epoch": 2.64, "learning_rate": 1.8898741899644703e-05, "loss": 0.2325, "step": 2605 }, { "epoch": 2.64, "learning_rate": 1.889774160666449e-05, "loss": 0.2434, "step": 2606 }, { "epoch": 2.65, "learning_rate": 1.889674088609527e-05, "loss": 0.2535, "step": 2607 }, { "epoch": 2.65, "learning_rate": 1.889573973798512e-05, "loss": 0.2608, "step": 2608 }, { "epoch": 2.65, "learning_rate": 1.889473816238215e-05, "loss": 0.237, "step": 2609 }, { "epoch": 2.65, "learning_rate": 1.88937361593345e-05, "loss": 0.2579, "step": 2610 }, { "epoch": 2.65, "learning_rate": 1.889273372889032e-05, "loss": 0.2646, "step": 2611 }, { "epoch": 2.65, "learning_rate": 1.8891730871097776e-05, "loss": 0.235, "step": 2612 }, { "epoch": 2.65, "learning_rate": 1.8890727586005073e-05, "loss": 0.2188, "step": 2613 }, { "epoch": 2.65, "learning_rate": 1.8889723873660417e-05, "loss": 0.2277, "step": 2614 }, { "epoch": 2.65, "learning_rate": 1.8888719734112044e-05, "loss": 0.294, "step": 2615 }, { "epoch": 2.65, "learning_rate": 1.8887715167408205e-05, "loss": 0.2569, "step": 2616 }, { "epoch": 2.66, "learning_rate": 1.8886710173597184e-05, "loss": 0.2488, "step": 2617 }, { "epoch": 2.66, "learning_rate": 1.8885704752727266e-05, "loss": 0.2492, "step": 2618 }, { "epoch": 2.66, "learning_rate": 1.8884698904846775e-05, "loss": 0.2622, "step": 2619 }, { "epoch": 2.66, "learning_rate": 1.8883692630004046e-05, "loss": 0.2522, "step": 2620 }, { "epoch": 2.66, "learning_rate": 1.888268592824744e-05, "loss": 0.2925, "step": 2621 }, { "epoch": 2.66, "learning_rate": 1.888167879962533e-05, "loss": 0.2464, "step": 2622 }, { "epoch": 2.66, "learning_rate": 1.888067124418611e-05, "loss": 0.2435, "step": 2623 }, { "epoch": 2.66, "learning_rate": 1.8879663261978214e-05, "loss": 0.2683, "step": 2624 }, { "epoch": 2.66, "learning_rate": 1.8878654853050066e-05, "loss": 0.2476, "step": 2625 }, { "epoch": 2.66, "learning_rate": 1.887764601745014e-05, "loss": 0.2697, "step": 2626 }, { "epoch": 2.67, "learning_rate": 1.8876636755226902e-05, "loss": 0.2673, "step": 2627 }, { "epoch": 2.67, "learning_rate": 1.887562706642886e-05, "loss": 0.294, "step": 2628 }, { "epoch": 2.67, "learning_rate": 1.8874616951104536e-05, "loss": 0.2641, "step": 2629 }, { "epoch": 2.67, "learning_rate": 1.8873606409302473e-05, "loss": 0.2326, "step": 2630 }, { "epoch": 2.67, "learning_rate": 1.8872595441071234e-05, "loss": 0.2543, "step": 2631 }, { "epoch": 2.67, "learning_rate": 1.8871584046459395e-05, "loss": 0.2313, "step": 2632 }, { "epoch": 2.67, "learning_rate": 1.887057222551557e-05, "loss": 0.2676, "step": 2633 }, { "epoch": 2.67, "learning_rate": 1.8869559978288374e-05, "loss": 0.2643, "step": 2634 }, { "epoch": 2.67, "learning_rate": 1.8868547304826455e-05, "loss": 0.2448, "step": 2635 }, { "epoch": 2.67, "learning_rate": 1.8867534205178476e-05, "loss": 0.2414, "step": 2636 }, { "epoch": 2.68, "learning_rate": 1.8866520679393127e-05, "loss": 0.241, "step": 2637 }, { "epoch": 2.68, "learning_rate": 1.8865506727519113e-05, "loss": 0.2785, "step": 2638 }, { "epoch": 2.68, "learning_rate": 1.8864492349605155e-05, "loss": 0.2566, "step": 2639 }, { "epoch": 2.68, "learning_rate": 1.8863477545700004e-05, "loss": 0.2519, "step": 2640 }, { "epoch": 2.68, "learning_rate": 1.8862462315852427e-05, "loss": 0.2796, "step": 2641 }, { "epoch": 2.68, "learning_rate": 1.8861446660111212e-05, "loss": 0.2804, "step": 2642 }, { "epoch": 2.68, "learning_rate": 1.8860430578525165e-05, "loss": 0.2906, "step": 2643 }, { "epoch": 2.68, "learning_rate": 1.885941407114312e-05, "loss": 0.2633, "step": 2644 }, { "epoch": 2.68, "learning_rate": 1.885839713801392e-05, "loss": 0.259, "step": 2645 }, { "epoch": 2.68, "learning_rate": 1.8857379779186438e-05, "loss": 0.2416, "step": 2646 }, { "epoch": 2.69, "learning_rate": 1.885636199470956e-05, "loss": 0.2466, "step": 2647 }, { "epoch": 2.69, "learning_rate": 1.8855343784632206e-05, "loss": 0.2572, "step": 2648 }, { "epoch": 2.69, "learning_rate": 1.88543251490033e-05, "loss": 0.2509, "step": 2649 }, { "epoch": 2.69, "learning_rate": 1.885330608787179e-05, "loss": 0.2588, "step": 2650 }, { "epoch": 2.69, "learning_rate": 1.8852286601286654e-05, "loss": 0.2665, "step": 2651 }, { "epoch": 2.69, "learning_rate": 1.8851266689296883e-05, "loss": 0.2415, "step": 2652 }, { "epoch": 2.69, "learning_rate": 1.885024635195149e-05, "loss": 0.2565, "step": 2653 }, { "epoch": 2.69, "learning_rate": 1.8849225589299508e-05, "loss": 0.2874, "step": 2654 }, { "epoch": 2.69, "learning_rate": 1.8848204401389992e-05, "loss": 0.2446, "step": 2655 }, { "epoch": 2.7, "learning_rate": 1.884718278827201e-05, "loss": 0.244, "step": 2656 }, { "epoch": 2.7, "learning_rate": 1.884616074999466e-05, "loss": 0.2728, "step": 2657 }, { "epoch": 2.7, "learning_rate": 1.8845138286607063e-05, "loss": 0.2391, "step": 2658 }, { "epoch": 2.7, "learning_rate": 1.884411539815835e-05, "loss": 0.251, "step": 2659 }, { "epoch": 2.7, "learning_rate": 1.884309208469767e-05, "loss": 0.2894, "step": 2660 }, { "epoch": 2.7, "learning_rate": 1.884206834627421e-05, "loss": 0.2561, "step": 2661 }, { "epoch": 2.7, "learning_rate": 1.884104418293716e-05, "loss": 0.2931, "step": 2662 }, { "epoch": 2.7, "learning_rate": 1.8840019594735742e-05, "loss": 0.2321, "step": 2663 }, { "epoch": 2.7, "learning_rate": 1.8838994581719186e-05, "loss": 0.2416, "step": 2664 }, { "epoch": 2.7, "learning_rate": 1.883796914393676e-05, "loss": 0.2629, "step": 2665 }, { "epoch": 2.71, "learning_rate": 1.8836943281437735e-05, "loss": 0.2166, "step": 2666 }, { "epoch": 2.71, "learning_rate": 1.883591699427141e-05, "loss": 0.2412, "step": 2667 }, { "epoch": 2.71, "learning_rate": 1.883489028248711e-05, "loss": 0.2409, "step": 2668 }, { "epoch": 2.71, "learning_rate": 1.8833863146134164e-05, "loss": 0.2397, "step": 2669 }, { "epoch": 2.71, "learning_rate": 1.8832835585261944e-05, "loss": 0.2407, "step": 2670 }, { "epoch": 2.71, "learning_rate": 1.8831807599919824e-05, "loss": 0.2536, "step": 2671 }, { "epoch": 2.71, "learning_rate": 1.8830779190157205e-05, "loss": 0.2814, "step": 2672 }, { "epoch": 2.71, "learning_rate": 1.882975035602351e-05, "loss": 0.2614, "step": 2673 }, { "epoch": 2.71, "learning_rate": 1.8828721097568176e-05, "loss": 0.2982, "step": 2674 }, { "epoch": 2.71, "learning_rate": 1.882769141484067e-05, "loss": 0.2538, "step": 2675 }, { "epoch": 2.72, "learning_rate": 1.8826661307890473e-05, "loss": 0.3337, "step": 2676 }, { "epoch": 2.72, "learning_rate": 1.8825630776767085e-05, "loss": 0.2337, "step": 2677 }, { "epoch": 2.72, "learning_rate": 1.8824599821520034e-05, "loss": 0.2757, "step": 2678 }, { "epoch": 2.72, "learning_rate": 1.882356844219886e-05, "loss": 0.2569, "step": 2679 }, { "epoch": 2.72, "learning_rate": 1.8822536638853126e-05, "loss": 0.2378, "step": 2680 }, { "epoch": 2.72, "learning_rate": 1.8821504411532418e-05, "loss": 0.2103, "step": 2681 }, { "epoch": 2.72, "learning_rate": 1.8820471760286346e-05, "loss": 0.2613, "step": 2682 }, { "epoch": 2.72, "learning_rate": 1.8819438685164523e-05, "loss": 0.2396, "step": 2683 }, { "epoch": 2.72, "learning_rate": 1.8818405186216605e-05, "loss": 0.2973, "step": 2684 }, { "epoch": 2.72, "learning_rate": 1.8817371263492252e-05, "loss": 0.2598, "step": 2685 }, { "epoch": 2.73, "learning_rate": 1.8816336917041152e-05, "loss": 0.2509, "step": 2686 }, { "epoch": 2.73, "learning_rate": 1.881530214691301e-05, "loss": 0.252, "step": 2687 }, { "epoch": 2.73, "learning_rate": 1.8814266953157557e-05, "loss": 0.2796, "step": 2688 }, { "epoch": 2.73, "learning_rate": 1.8813231335824534e-05, "loss": 0.2446, "step": 2689 }, { "epoch": 2.73, "learning_rate": 1.8812195294963715e-05, "loss": 0.2521, "step": 2690 }, { "epoch": 2.73, "learning_rate": 1.881115883062488e-05, "loss": 0.2561, "step": 2691 }, { "epoch": 2.73, "learning_rate": 1.8810121942857848e-05, "loss": 0.2608, "step": 2692 }, { "epoch": 2.73, "learning_rate": 1.8809084631712434e-05, "loss": 0.2029, "step": 2693 }, { "epoch": 2.73, "learning_rate": 1.88080468972385e-05, "loss": 0.2914, "step": 2694 }, { "epoch": 2.73, "learning_rate": 1.880700873948591e-05, "loss": 0.2599, "step": 2695 }, { "epoch": 2.74, "learning_rate": 1.880597015850455e-05, "loss": 0.246, "step": 2696 }, { "epoch": 2.74, "learning_rate": 1.8804931154344336e-05, "loss": 0.257, "step": 2697 }, { "epoch": 2.74, "learning_rate": 1.8803891727055192e-05, "loss": 0.2941, "step": 2698 }, { "epoch": 2.74, "learning_rate": 1.8802851876687075e-05, "loss": 0.2612, "step": 2699 }, { "epoch": 2.74, "learning_rate": 1.880181160328995e-05, "loss": 0.256, "step": 2700 }, { "epoch": 2.74, "learning_rate": 1.8800770906913815e-05, "loss": 0.2462, "step": 2701 }, { "epoch": 2.74, "learning_rate": 1.8799729787608675e-05, "loss": 0.2619, "step": 2702 }, { "epoch": 2.74, "learning_rate": 1.8798688245424568e-05, "loss": 0.2363, "step": 2703 }, { "epoch": 2.74, "learning_rate": 1.8797646280411538e-05, "loss": 0.2348, "step": 2704 }, { "epoch": 2.74, "learning_rate": 1.8796603892619668e-05, "loss": 0.2468, "step": 2705 }, { "epoch": 2.75, "learning_rate": 1.8795561082099045e-05, "loss": 0.2477, "step": 2706 }, { "epoch": 2.75, "learning_rate": 1.8794517848899778e-05, "loss": 0.2609, "step": 2707 }, { "epoch": 2.75, "learning_rate": 1.8793474193072007e-05, "loss": 0.2732, "step": 2708 }, { "epoch": 2.75, "learning_rate": 1.8792430114665888e-05, "loss": 0.262, "step": 2709 }, { "epoch": 2.75, "learning_rate": 1.8791385613731586e-05, "loss": 0.2739, "step": 2710 }, { "epoch": 2.75, "learning_rate": 1.8790340690319304e-05, "loss": 0.2681, "step": 2711 }, { "epoch": 2.75, "learning_rate": 1.8789295344479253e-05, "loss": 0.2702, "step": 2712 }, { "epoch": 2.75, "learning_rate": 1.878824957626167e-05, "loss": 0.271, "step": 2713 }, { "epoch": 2.75, "learning_rate": 1.8787203385716802e-05, "loss": 0.2733, "step": 2714 }, { "epoch": 2.75, "learning_rate": 1.8786156772894938e-05, "loss": 0.267, "step": 2715 }, { "epoch": 2.76, "learning_rate": 1.8785109737846364e-05, "loss": 0.2638, "step": 2716 }, { "epoch": 2.76, "learning_rate": 1.8784062280621404e-05, "loss": 0.2563, "step": 2717 }, { "epoch": 2.76, "learning_rate": 1.8783014401270384e-05, "loss": 0.2795, "step": 2718 }, { "epoch": 2.76, "learning_rate": 1.878196609984367e-05, "loss": 0.2562, "step": 2719 }, { "epoch": 2.76, "learning_rate": 1.8780917376391638e-05, "loss": 0.2396, "step": 2720 }, { "epoch": 2.76, "learning_rate": 1.877986823096468e-05, "loss": 0.2528, "step": 2721 }, { "epoch": 2.76, "learning_rate": 1.8778818663613217e-05, "loss": 0.269, "step": 2722 }, { "epoch": 2.76, "learning_rate": 1.8777768674387688e-05, "loss": 0.2831, "step": 2723 }, { "epoch": 2.76, "learning_rate": 1.877671826333855e-05, "loss": 0.2443, "step": 2724 }, { "epoch": 2.77, "learning_rate": 1.8775667430516283e-05, "loss": 0.2456, "step": 2725 }, { "epoch": 2.77, "learning_rate": 1.8774616175971378e-05, "loss": 0.2333, "step": 2726 }, { "epoch": 2.77, "learning_rate": 1.8773564499754366e-05, "loss": 0.2542, "step": 2727 }, { "epoch": 2.77, "learning_rate": 1.8772512401915778e-05, "loss": 0.2682, "step": 2728 }, { "epoch": 2.77, "learning_rate": 1.8771459882506178e-05, "loss": 0.269, "step": 2729 }, { "epoch": 2.77, "learning_rate": 1.877040694157614e-05, "loss": 0.2496, "step": 2730 }, { "epoch": 2.77, "learning_rate": 1.8769353579176272e-05, "loss": 0.2608, "step": 2731 }, { "epoch": 2.77, "learning_rate": 1.8768299795357187e-05, "loss": 0.2588, "step": 2732 }, { "epoch": 2.77, "learning_rate": 1.876724559016953e-05, "loss": 0.2542, "step": 2733 }, { "epoch": 2.77, "learning_rate": 1.876619096366396e-05, "loss": 0.2716, "step": 2734 }, { "epoch": 2.78, "learning_rate": 1.8765135915891156e-05, "loss": 0.2741, "step": 2735 }, { "epoch": 2.78, "learning_rate": 1.8764080446901824e-05, "loss": 0.2416, "step": 2736 }, { "epoch": 2.78, "learning_rate": 1.8763024556746684e-05, "loss": 0.2591, "step": 2737 }, { "epoch": 2.78, "learning_rate": 1.8761968245476477e-05, "loss": 0.2405, "step": 2738 }, { "epoch": 2.78, "learning_rate": 1.876091151314196e-05, "loss": 0.2507, "step": 2739 }, { "epoch": 2.78, "learning_rate": 1.8759854359793928e-05, "loss": 0.2799, "step": 2740 }, { "epoch": 2.78, "learning_rate": 1.8758796785483173e-05, "loss": 0.259, "step": 2741 }, { "epoch": 2.78, "learning_rate": 1.8757738790260515e-05, "loss": 0.2453, "step": 2742 }, { "epoch": 2.78, "learning_rate": 1.875668037417681e-05, "loss": 0.2608, "step": 2743 }, { "epoch": 2.78, "learning_rate": 1.8755621537282907e-05, "loss": 0.2796, "step": 2744 }, { "epoch": 2.79, "learning_rate": 1.8754562279629697e-05, "loss": 0.2951, "step": 2745 }, { "epoch": 2.79, "learning_rate": 1.8753502601268084e-05, "loss": 0.2417, "step": 2746 }, { "epoch": 2.79, "learning_rate": 1.8752442502248987e-05, "loss": 0.2537, "step": 2747 }, { "epoch": 2.79, "learning_rate": 1.8751381982623355e-05, "loss": 0.2429, "step": 2748 }, { "epoch": 2.79, "learning_rate": 1.875032104244215e-05, "loss": 0.238, "step": 2749 }, { "epoch": 2.79, "learning_rate": 1.8749259681756356e-05, "loss": 0.2542, "step": 2750 }, { "epoch": 2.79, "learning_rate": 1.874819790061698e-05, "loss": 0.251, "step": 2751 }, { "epoch": 2.79, "learning_rate": 1.8747135699075046e-05, "loss": 0.2607, "step": 2752 }, { "epoch": 2.79, "learning_rate": 1.8746073077181598e-05, "loss": 0.2459, "step": 2753 }, { "epoch": 2.79, "learning_rate": 1.8745010034987697e-05, "loss": 0.233, "step": 2754 }, { "epoch": 2.8, "learning_rate": 1.8743946572544437e-05, "loss": 0.2368, "step": 2755 }, { "epoch": 2.8, "learning_rate": 1.874288268990292e-05, "loss": 0.2436, "step": 2756 }, { "epoch": 2.8, "learning_rate": 1.8741818387114274e-05, "loss": 0.2523, "step": 2757 }, { "epoch": 2.8, "learning_rate": 1.8740753664229637e-05, "loss": 0.2671, "step": 2758 }, { "epoch": 2.8, "learning_rate": 1.8739688521300183e-05, "loss": 0.2373, "step": 2759 }, { "epoch": 2.8, "learning_rate": 1.8738622958377096e-05, "loss": 0.2608, "step": 2760 }, { "epoch": 2.8, "learning_rate": 1.8737556975511584e-05, "loss": 0.31, "step": 2761 }, { "epoch": 2.8, "learning_rate": 1.8736490572754872e-05, "loss": 0.2701, "step": 2762 }, { "epoch": 2.8, "learning_rate": 1.8735423750158202e-05, "loss": 0.2434, "step": 2763 }, { "epoch": 2.8, "learning_rate": 1.8734356507772854e-05, "loss": 0.2449, "step": 2764 }, { "epoch": 2.81, "learning_rate": 1.8733288845650105e-05, "loss": 0.2616, "step": 2765 }, { "epoch": 2.81, "learning_rate": 1.8732220763841266e-05, "loss": 0.2815, "step": 2766 }, { "epoch": 2.81, "learning_rate": 1.8731152262397664e-05, "loss": 0.2605, "step": 2767 }, { "epoch": 2.81, "learning_rate": 1.8730083341370648e-05, "loss": 0.2949, "step": 2768 }, { "epoch": 2.81, "learning_rate": 1.8729014000811584e-05, "loss": 0.2323, "step": 2769 }, { "epoch": 2.81, "learning_rate": 1.8727944240771857e-05, "loss": 0.2498, "step": 2770 }, { "epoch": 2.81, "learning_rate": 1.8726874061302882e-05, "loss": 0.2573, "step": 2771 }, { "epoch": 2.81, "learning_rate": 1.8725803462456087e-05, "loss": 0.2548, "step": 2772 }, { "epoch": 2.81, "learning_rate": 1.8724732444282917e-05, "loss": 0.2438, "step": 2773 }, { "epoch": 2.81, "learning_rate": 1.872366100683484e-05, "loss": 0.2883, "step": 2774 }, { "epoch": 2.82, "learning_rate": 1.872258915016335e-05, "loss": 0.243, "step": 2775 }, { "epoch": 2.82, "learning_rate": 1.872151687431995e-05, "loss": 0.2582, "step": 2776 }, { "epoch": 2.82, "learning_rate": 1.8720444179356172e-05, "loss": 0.2798, "step": 2777 }, { "epoch": 2.82, "learning_rate": 1.8719371065323567e-05, "loss": 0.251, "step": 2778 }, { "epoch": 2.82, "learning_rate": 1.87182975322737e-05, "loss": 0.2919, "step": 2779 }, { "epoch": 2.82, "learning_rate": 1.871722358025817e-05, "loss": 0.2431, "step": 2780 }, { "epoch": 2.82, "learning_rate": 1.8716149209328573e-05, "loss": 0.2558, "step": 2781 }, { "epoch": 2.82, "learning_rate": 1.871507441953655e-05, "loss": 0.2877, "step": 2782 }, { "epoch": 2.82, "learning_rate": 1.8713999210933743e-05, "loss": 0.2465, "step": 2783 }, { "epoch": 2.82, "learning_rate": 1.871292358357183e-05, "loss": 0.2562, "step": 2784 }, { "epoch": 2.83, "learning_rate": 1.8711847537502498e-05, "loss": 0.2814, "step": 2785 }, { "epoch": 2.83, "learning_rate": 1.8710771072777455e-05, "loss": 0.2757, "step": 2786 }, { "epoch": 2.83, "learning_rate": 1.8709694189448434e-05, "loss": 0.2494, "step": 2787 }, { "epoch": 2.83, "learning_rate": 1.8708616887567187e-05, "loss": 0.2463, "step": 2788 }, { "epoch": 2.83, "learning_rate": 1.870753916718548e-05, "loss": 0.2454, "step": 2789 }, { "epoch": 2.83, "learning_rate": 1.8706461028355107e-05, "loss": 0.2975, "step": 2790 }, { "epoch": 2.83, "learning_rate": 1.8705382471127877e-05, "loss": 0.2566, "step": 2791 }, { "epoch": 2.83, "learning_rate": 1.870430349555562e-05, "loss": 0.2861, "step": 2792 }, { "epoch": 2.83, "learning_rate": 1.8703224101690195e-05, "loss": 0.2468, "step": 2793 }, { "epoch": 2.84, "learning_rate": 1.8702144289583467e-05, "loss": 0.2638, "step": 2794 }, { "epoch": 2.84, "learning_rate": 1.8701064059287327e-05, "loss": 0.2554, "step": 2795 }, { "epoch": 2.84, "learning_rate": 1.8699983410853688e-05, "loss": 0.2922, "step": 2796 }, { "epoch": 2.84, "learning_rate": 1.8698902344334482e-05, "loss": 0.234, "step": 2797 }, { "epoch": 2.84, "learning_rate": 1.8697820859781658e-05, "loss": 0.2829, "step": 2798 }, { "epoch": 2.84, "learning_rate": 1.8696738957247188e-05, "loss": 0.2683, "step": 2799 }, { "epoch": 2.84, "learning_rate": 1.8695656636783067e-05, "loss": 0.2611, "step": 2800 }, { "epoch": 2.84, "learning_rate": 1.8694573898441306e-05, "loss": 0.2565, "step": 2801 }, { "epoch": 2.84, "learning_rate": 1.8693490742273938e-05, "loss": 0.2727, "step": 2802 }, { "epoch": 2.84, "learning_rate": 1.869240716833301e-05, "loss": 0.2745, "step": 2803 }, { "epoch": 2.85, "learning_rate": 1.8691323176670596e-05, "loss": 0.2509, "step": 2804 }, { "epoch": 2.85, "learning_rate": 1.8690238767338792e-05, "loss": 0.253, "step": 2805 }, { "epoch": 2.85, "learning_rate": 1.868915394038971e-05, "loss": 0.2391, "step": 2806 }, { "epoch": 2.85, "learning_rate": 1.8688068695875477e-05, "loss": 0.2626, "step": 2807 }, { "epoch": 2.85, "learning_rate": 1.8686983033848246e-05, "loss": 0.2841, "step": 2808 }, { "epoch": 2.85, "learning_rate": 1.8685896954360196e-05, "loss": 0.2601, "step": 2809 }, { "epoch": 2.85, "learning_rate": 1.8684810457463516e-05, "loss": 0.2764, "step": 2810 }, { "epoch": 2.85, "learning_rate": 1.8683723543210416e-05, "loss": 0.2452, "step": 2811 }, { "epoch": 2.85, "learning_rate": 1.868263621165313e-05, "loss": 0.2458, "step": 2812 }, { "epoch": 2.85, "learning_rate": 1.8681548462843914e-05, "loss": 0.2915, "step": 2813 }, { "epoch": 2.86, "learning_rate": 1.8680460296835038e-05, "loss": 0.265, "step": 2814 }, { "epoch": 2.86, "learning_rate": 1.8679371713678794e-05, "loss": 0.2761, "step": 2815 }, { "epoch": 2.86, "learning_rate": 1.8678282713427497e-05, "loss": 0.2689, "step": 2816 }, { "epoch": 2.86, "learning_rate": 1.8677193296133478e-05, "loss": 0.2598, "step": 2817 }, { "epoch": 2.86, "learning_rate": 1.8676103461849092e-05, "loss": 0.2371, "step": 2818 }, { "epoch": 2.86, "learning_rate": 1.8675013210626708e-05, "loss": 0.2691, "step": 2819 }, { "epoch": 2.86, "learning_rate": 1.8673922542518722e-05, "loss": 0.2338, "step": 2820 }, { "epoch": 2.86, "learning_rate": 1.867283145757755e-05, "loss": 0.2811, "step": 2821 }, { "epoch": 2.86, "learning_rate": 1.867173995585562e-05, "loss": 0.2611, "step": 2822 }, { "epoch": 2.86, "learning_rate": 1.8670648037405383e-05, "loss": 0.2697, "step": 2823 }, { "epoch": 2.87, "learning_rate": 1.866955570227932e-05, "loss": 0.2893, "step": 2824 }, { "epoch": 2.87, "learning_rate": 1.8668462950529923e-05, "loss": 0.2335, "step": 2825 }, { "epoch": 2.87, "learning_rate": 1.8667369782209697e-05, "loss": 0.2543, "step": 2826 }, { "epoch": 2.87, "learning_rate": 1.866627619737118e-05, "loss": 0.2481, "step": 2827 }, { "epoch": 2.87, "learning_rate": 1.866518219606693e-05, "loss": 0.2594, "step": 2828 }, { "epoch": 2.87, "learning_rate": 1.8664087778349516e-05, "loss": 0.2558, "step": 2829 }, { "epoch": 2.87, "learning_rate": 1.866299294427153e-05, "loss": 0.267, "step": 2830 }, { "epoch": 2.87, "learning_rate": 1.8661897693885586e-05, "loss": 0.2496, "step": 2831 }, { "epoch": 2.87, "learning_rate": 1.8660802027244315e-05, "loss": 0.2962, "step": 2832 }, { "epoch": 2.87, "learning_rate": 1.8659705944400378e-05, "loss": 0.2614, "step": 2833 }, { "epoch": 2.88, "learning_rate": 1.8658609445406437e-05, "loss": 0.2542, "step": 2834 }, { "epoch": 2.88, "learning_rate": 1.8657512530315197e-05, "loss": 0.2647, "step": 2835 }, { "epoch": 2.88, "learning_rate": 1.8656415199179365e-05, "loss": 0.2701, "step": 2836 }, { "epoch": 2.88, "learning_rate": 1.8655317452051673e-05, "loss": 0.2907, "step": 2837 }, { "epoch": 2.88, "learning_rate": 1.8654219288984877e-05, "loss": 0.268, "step": 2838 }, { "epoch": 2.88, "learning_rate": 1.865312071003175e-05, "loss": 0.2436, "step": 2839 }, { "epoch": 2.88, "learning_rate": 1.8652021715245084e-05, "loss": 0.2603, "step": 2840 }, { "epoch": 2.88, "learning_rate": 1.865092230467769e-05, "loss": 0.2633, "step": 2841 }, { "epoch": 2.88, "learning_rate": 1.864982247838241e-05, "loss": 0.261, "step": 2842 }, { "epoch": 2.88, "learning_rate": 1.8648722236412086e-05, "loss": 0.2918, "step": 2843 }, { "epoch": 2.89, "learning_rate": 1.86476215788196e-05, "loss": 0.2437, "step": 2844 }, { "epoch": 2.89, "learning_rate": 1.8646520505657836e-05, "loss": 0.2696, "step": 2845 }, { "epoch": 2.89, "learning_rate": 1.8645419016979715e-05, "loss": 0.2437, "step": 2846 }, { "epoch": 2.89, "learning_rate": 1.8644317112838172e-05, "loss": 0.2881, "step": 2847 }, { "epoch": 2.89, "learning_rate": 1.8643214793286152e-05, "loss": 0.2877, "step": 2848 }, { "epoch": 2.89, "learning_rate": 1.8642112058376628e-05, "loss": 0.2888, "step": 2849 }, { "epoch": 2.89, "learning_rate": 1.8641008908162602e-05, "loss": 0.2633, "step": 2850 }, { "epoch": 2.89, "learning_rate": 1.8639905342697077e-05, "loss": 0.2715, "step": 2851 }, { "epoch": 2.89, "learning_rate": 1.863880136203309e-05, "loss": 0.2635, "step": 2852 }, { "epoch": 2.89, "learning_rate": 1.86376969662237e-05, "loss": 0.2505, "step": 2853 }, { "epoch": 2.9, "learning_rate": 1.8636592155321965e-05, "loss": 0.242, "step": 2854 }, { "epoch": 2.9, "learning_rate": 1.863548692938099e-05, "loss": 0.2622, "step": 2855 }, { "epoch": 2.9, "learning_rate": 1.8634381288453882e-05, "loss": 0.2639, "step": 2856 }, { "epoch": 2.9, "learning_rate": 1.863327523259378e-05, "loss": 0.2808, "step": 2857 }, { "epoch": 2.9, "learning_rate": 1.8632168761853827e-05, "loss": 0.2848, "step": 2858 }, { "epoch": 2.9, "learning_rate": 1.8631061876287204e-05, "loss": 0.2701, "step": 2859 }, { "epoch": 2.9, "learning_rate": 1.8629954575947097e-05, "loss": 0.2571, "step": 2860 }, { "epoch": 2.9, "learning_rate": 1.8628846860886723e-05, "loss": 0.2851, "step": 2861 }, { "epoch": 2.9, "learning_rate": 1.862773873115931e-05, "loss": 0.2494, "step": 2862 }, { "epoch": 2.91, "learning_rate": 1.8626630186818115e-05, "loss": 0.2695, "step": 2863 }, { "epoch": 2.91, "learning_rate": 1.8625521227916407e-05, "loss": 0.2503, "step": 2864 }, { "epoch": 2.91, "learning_rate": 1.862441185450748e-05, "loss": 0.281, "step": 2865 }, { "epoch": 2.91, "learning_rate": 1.8623302066644644e-05, "loss": 0.2345, "step": 2866 }, { "epoch": 2.91, "learning_rate": 1.862219186438123e-05, "loss": 0.2704, "step": 2867 }, { "epoch": 2.91, "learning_rate": 1.8621081247770593e-05, "loss": 0.2539, "step": 2868 }, { "epoch": 2.91, "learning_rate": 1.8619970216866102e-05, "loss": 0.2513, "step": 2869 }, { "epoch": 2.91, "learning_rate": 1.8618858771721148e-05, "loss": 0.2813, "step": 2870 }, { "epoch": 2.91, "learning_rate": 1.8617746912389144e-05, "loss": 0.2415, "step": 2871 }, { "epoch": 2.91, "learning_rate": 1.8616634638923522e-05, "loss": 0.2742, "step": 2872 }, { "epoch": 2.92, "learning_rate": 1.8615521951377732e-05, "loss": 0.2574, "step": 2873 }, { "epoch": 2.92, "learning_rate": 1.861440884980525e-05, "loss": 0.2378, "step": 2874 }, { "epoch": 2.92, "learning_rate": 1.861329533425956e-05, "loss": 0.2661, "step": 2875 }, { "epoch": 2.92, "learning_rate": 1.8612181404794175e-05, "loss": 0.2255, "step": 2876 }, { "epoch": 2.92, "learning_rate": 1.861106706146263e-05, "loss": 0.2733, "step": 2877 }, { "epoch": 2.92, "learning_rate": 1.8609952304318472e-05, "loss": 0.2402, "step": 2878 }, { "epoch": 2.92, "learning_rate": 1.860883713341527e-05, "loss": 0.2795, "step": 2879 }, { "epoch": 2.92, "learning_rate": 1.8607721548806618e-05, "loss": 0.2599, "step": 2880 }, { "epoch": 2.92, "learning_rate": 1.8606605550546126e-05, "loss": 0.2585, "step": 2881 }, { "epoch": 2.92, "learning_rate": 1.860548913868742e-05, "loss": 0.2399, "step": 2882 }, { "epoch": 2.93, "learning_rate": 1.8604372313284163e-05, "loss": 0.2255, "step": 2883 }, { "epoch": 2.93, "learning_rate": 1.8603255074390007e-05, "loss": 0.2386, "step": 2884 }, { "epoch": 2.93, "learning_rate": 1.8602137422058656e-05, "loss": 0.2717, "step": 2885 }, { "epoch": 2.93, "learning_rate": 1.860101935634381e-05, "loss": 0.2696, "step": 2886 }, { "epoch": 2.93, "learning_rate": 1.8599900877299208e-05, "loss": 0.26, "step": 2887 }, { "epoch": 2.93, "learning_rate": 1.8598781984978592e-05, "loss": 0.2447, "step": 2888 }, { "epoch": 2.93, "learning_rate": 1.8597662679435735e-05, "loss": 0.2946, "step": 2889 }, { "epoch": 2.93, "learning_rate": 1.8596542960724422e-05, "loss": 0.2905, "step": 2890 }, { "epoch": 2.93, "learning_rate": 1.859542282889847e-05, "loss": 0.2845, "step": 2891 }, { "epoch": 2.93, "learning_rate": 1.8594302284011704e-05, "loss": 0.2574, "step": 2892 }, { "epoch": 2.94, "learning_rate": 1.859318132611797e-05, "loss": 0.289, "step": 2893 }, { "epoch": 2.94, "learning_rate": 1.8592059955271136e-05, "loss": 0.2659, "step": 2894 }, { "epoch": 2.94, "learning_rate": 1.8590938171525096e-05, "loss": 0.2675, "step": 2895 }, { "epoch": 2.94, "learning_rate": 1.8589815974933757e-05, "loss": 0.2557, "step": 2896 }, { "epoch": 2.94, "learning_rate": 1.8588693365551042e-05, "loss": 0.2599, "step": 2897 }, { "epoch": 2.94, "learning_rate": 1.8587570343430907e-05, "loss": 0.2345, "step": 2898 }, { "epoch": 2.94, "learning_rate": 1.8586446908627314e-05, "loss": 0.293, "step": 2899 }, { "epoch": 2.94, "learning_rate": 1.8585323061194254e-05, "loss": 0.2945, "step": 2900 }, { "epoch": 2.94, "learning_rate": 1.8584198801185732e-05, "loss": 0.249, "step": 2901 }, { "epoch": 2.94, "learning_rate": 1.8583074128655774e-05, "loss": 0.2727, "step": 2902 }, { "epoch": 2.95, "learning_rate": 1.8581949043658433e-05, "loss": 0.2744, "step": 2903 }, { "epoch": 2.95, "learning_rate": 1.858082354624777e-05, "loss": 0.2778, "step": 2904 }, { "epoch": 2.95, "learning_rate": 1.8579697636477876e-05, "loss": 0.2353, "step": 2905 }, { "epoch": 2.95, "learning_rate": 1.857857131440285e-05, "loss": 0.2684, "step": 2906 }, { "epoch": 2.95, "learning_rate": 1.8577444580076832e-05, "loss": 0.239, "step": 2907 }, { "epoch": 2.95, "learning_rate": 1.8576317433553957e-05, "loss": 0.246, "step": 2908 }, { "epoch": 2.95, "learning_rate": 1.8575189874888396e-05, "loss": 0.2652, "step": 2909 }, { "epoch": 2.95, "learning_rate": 1.857406190413433e-05, "loss": 0.2823, "step": 2910 }, { "epoch": 2.95, "learning_rate": 1.857293352134597e-05, "loss": 0.2547, "step": 2911 }, { "epoch": 2.95, "learning_rate": 1.857180472657754e-05, "loss": 0.2451, "step": 2912 }, { "epoch": 2.96, "learning_rate": 1.8570675519883283e-05, "loss": 0.2665, "step": 2913 }, { "epoch": 2.96, "learning_rate": 1.8569545901317468e-05, "loss": 0.2703, "step": 2914 }, { "epoch": 2.96, "learning_rate": 1.8568415870934375e-05, "loss": 0.2795, "step": 2915 }, { "epoch": 2.96, "learning_rate": 1.8567285428788312e-05, "loss": 0.233, "step": 2916 }, { "epoch": 2.96, "learning_rate": 1.8566154574933604e-05, "loss": 0.2575, "step": 2917 }, { "epoch": 2.96, "learning_rate": 1.8565023309424594e-05, "loss": 0.2823, "step": 2918 }, { "epoch": 2.96, "learning_rate": 1.8563891632315644e-05, "loss": 0.2625, "step": 2919 }, { "epoch": 2.96, "learning_rate": 1.856275954366114e-05, "loss": 0.2841, "step": 2920 }, { "epoch": 2.96, "learning_rate": 1.8561627043515487e-05, "loss": 0.269, "step": 2921 }, { "epoch": 2.96, "learning_rate": 1.8560494131933103e-05, "loss": 0.2493, "step": 2922 }, { "epoch": 2.97, "learning_rate": 1.855936080896844e-05, "loss": 0.257, "step": 2923 }, { "epoch": 2.97, "learning_rate": 1.8558227074675948e-05, "loss": 0.2937, "step": 2924 }, { "epoch": 2.97, "learning_rate": 1.8557092929110122e-05, "loss": 0.2603, "step": 2925 }, { "epoch": 2.97, "learning_rate": 1.8555958372325456e-05, "loss": 0.2693, "step": 2926 }, { "epoch": 2.97, "learning_rate": 1.8554823404376477e-05, "loss": 0.2326, "step": 2927 }, { "epoch": 2.97, "learning_rate": 1.8553688025317732e-05, "loss": 0.2624, "step": 2928 }, { "epoch": 2.97, "learning_rate": 1.8552552235203768e-05, "loss": 0.2268, "step": 2929 }, { "epoch": 2.97, "learning_rate": 1.8551416034089177e-05, "loss": 0.2719, "step": 2930 }, { "epoch": 2.97, "learning_rate": 1.8550279422028555e-05, "loss": 0.2588, "step": 2931 }, { "epoch": 2.98, "learning_rate": 1.854914239907653e-05, "loss": 0.2224, "step": 2932 }, { "epoch": 2.98, "learning_rate": 1.854800496528774e-05, "loss": 0.244, "step": 2933 }, { "epoch": 2.98, "learning_rate": 1.8546867120716842e-05, "loss": 0.2432, "step": 2934 }, { "epoch": 2.98, "learning_rate": 1.8545728865418513e-05, "loss": 0.2581, "step": 2935 }, { "epoch": 2.98, "learning_rate": 1.854459019944746e-05, "loss": 0.2394, "step": 2936 }, { "epoch": 2.98, "learning_rate": 1.85434511228584e-05, "loss": 0.2892, "step": 2937 }, { "epoch": 2.98, "learning_rate": 1.8542311635706078e-05, "loss": 0.2756, "step": 2938 }, { "epoch": 2.98, "learning_rate": 1.8541171738045244e-05, "loss": 0.2709, "step": 2939 }, { "epoch": 2.98, "learning_rate": 1.8540031429930678e-05, "loss": 0.2356, "step": 2940 }, { "epoch": 2.98, "learning_rate": 1.8538890711417184e-05, "loss": 0.2634, "step": 2941 }, { "epoch": 2.99, "learning_rate": 1.853774958255958e-05, "loss": 0.2713, "step": 2942 }, { "epoch": 2.99, "learning_rate": 1.85366080434127e-05, "loss": 0.2574, "step": 2943 }, { "epoch": 2.99, "learning_rate": 1.85354660940314e-05, "loss": 0.2585, "step": 2944 }, { "epoch": 2.99, "learning_rate": 1.8534323734470562e-05, "loss": 0.2694, "step": 2945 }, { "epoch": 2.99, "learning_rate": 1.853318096478508e-05, "loss": 0.2467, "step": 2946 }, { "epoch": 2.99, "learning_rate": 1.8532037785029877e-05, "loss": 0.2735, "step": 2947 }, { "epoch": 2.99, "learning_rate": 1.8530894195259884e-05, "loss": 0.2922, "step": 2948 }, { "epoch": 2.99, "learning_rate": 1.8529750195530054e-05, "loss": 0.2989, "step": 2949 }, { "epoch": 2.99, "learning_rate": 1.852860578589537e-05, "loss": 0.2764, "step": 2950 }, { "epoch": 2.99, "learning_rate": 1.8527460966410826e-05, "loss": 0.2615, "step": 2951 }, { "epoch": 3.0, "learning_rate": 1.8526315737131436e-05, "loss": 0.2825, "step": 2952 }, { "epoch": 3.0, "learning_rate": 1.8525170098112234e-05, "loss": 0.286, "step": 2953 }, { "epoch": 3.0, "learning_rate": 1.8524024049408275e-05, "loss": 0.2571, "step": 2954 }, { "epoch": 3.0, "learning_rate": 1.8522877591074636e-05, "loss": 0.2537, "step": 2955 }, { "epoch": 3.0, "learning_rate": 1.8521730723166408e-05, "loss": 0.2729, "step": 2956 }, { "epoch": 3.0, "learning_rate": 1.852058344573871e-05, "loss": 0.2093, "step": 2957 }, { "epoch": 3.0, "learning_rate": 1.8519435758846667e-05, "loss": 0.1777, "step": 2958 }, { "epoch": 3.0, "learning_rate": 1.8518287662545437e-05, "loss": 0.1685, "step": 2959 }, { "epoch": 3.0, "learning_rate": 1.8517139156890196e-05, "loss": 0.146, "step": 2960 }, { "epoch": 3.0, "learning_rate": 1.851599024193613e-05, "loss": 0.1743, "step": 2961 }, { "epoch": 3.01, "learning_rate": 1.8514840917738457e-05, "loss": 0.1524, "step": 2962 }, { "epoch": 3.01, "learning_rate": 1.8513691184352405e-05, "loss": 0.1613, "step": 2963 }, { "epoch": 3.01, "learning_rate": 1.8512541041833224e-05, "loss": 0.1974, "step": 2964 }, { "epoch": 3.01, "learning_rate": 1.851139049023619e-05, "loss": 0.1747, "step": 2965 }, { "epoch": 3.01, "learning_rate": 1.851023952961659e-05, "loss": 0.1582, "step": 2966 }, { "epoch": 3.01, "learning_rate": 1.8509088160029737e-05, "loss": 0.1363, "step": 2967 }, { "epoch": 3.01, "learning_rate": 1.850793638153096e-05, "loss": 0.1467, "step": 2968 }, { "epoch": 3.01, "learning_rate": 1.8506784194175608e-05, "loss": 0.1682, "step": 2969 }, { "epoch": 3.01, "learning_rate": 1.8505631598019053e-05, "loss": 0.1747, "step": 2970 }, { "epoch": 3.01, "learning_rate": 1.850447859311668e-05, "loss": 0.1811, "step": 2971 }, { "epoch": 3.02, "learning_rate": 1.85033251795239e-05, "loss": 0.1213, "step": 2972 }, { "epoch": 3.02, "learning_rate": 1.8502171357296144e-05, "loss": 0.1601, "step": 2973 }, { "epoch": 3.02, "learning_rate": 1.8501017126488854e-05, "loss": 0.1598, "step": 2974 }, { "epoch": 3.02, "learning_rate": 1.8499862487157503e-05, "loss": 0.1655, "step": 2975 }, { "epoch": 3.02, "learning_rate": 1.8498707439357574e-05, "loss": 0.143, "step": 2976 }, { "epoch": 3.02, "learning_rate": 1.8497551983144576e-05, "loss": 0.1418, "step": 2977 }, { "epoch": 3.02, "learning_rate": 1.849639611857404e-05, "loss": 0.1655, "step": 2978 }, { "epoch": 3.02, "learning_rate": 1.84952398457015e-05, "loss": 0.1541, "step": 2979 }, { "epoch": 3.02, "learning_rate": 1.8494083164582533e-05, "loss": 0.1337, "step": 2980 }, { "epoch": 3.02, "learning_rate": 1.8492926075272723e-05, "loss": 0.1773, "step": 2981 }, { "epoch": 3.03, "learning_rate": 1.849176857782767e-05, "loss": 0.1619, "step": 2982 }, { "epoch": 3.03, "learning_rate": 1.8490610672303e-05, "loss": 0.1361, "step": 2983 }, { "epoch": 3.03, "learning_rate": 1.848945235875436e-05, "loss": 0.1754, "step": 2984 }, { "epoch": 3.03, "learning_rate": 1.8488293637237413e-05, "loss": 0.1676, "step": 2985 }, { "epoch": 3.03, "learning_rate": 1.848713450780784e-05, "loss": 0.1543, "step": 2986 }, { "epoch": 3.03, "learning_rate": 1.8485974970521346e-05, "loss": 0.1382, "step": 2987 }, { "epoch": 3.03, "learning_rate": 1.8484815025433652e-05, "loss": 0.1636, "step": 2988 }, { "epoch": 3.03, "learning_rate": 1.8483654672600506e-05, "loss": 0.1599, "step": 2989 }, { "epoch": 3.03, "learning_rate": 1.848249391207766e-05, "loss": 0.133, "step": 2990 }, { "epoch": 3.04, "learning_rate": 1.8481332743920902e-05, "loss": 0.1566, "step": 2991 }, { "epoch": 3.04, "learning_rate": 1.8480171168186036e-05, "loss": 0.1559, "step": 2992 }, { "epoch": 3.04, "learning_rate": 1.8479009184928873e-05, "loss": 0.1382, "step": 2993 }, { "epoch": 3.04, "learning_rate": 1.8477846794205258e-05, "loss": 0.1665, "step": 2994 }, { "epoch": 3.04, "learning_rate": 1.8476683996071053e-05, "loss": 0.1519, "step": 2995 }, { "epoch": 3.04, "learning_rate": 1.8475520790582136e-05, "loss": 0.164, "step": 2996 }, { "epoch": 3.04, "learning_rate": 1.8474357177794404e-05, "loss": 0.1655, "step": 2997 }, { "epoch": 3.04, "learning_rate": 1.8473193157763778e-05, "loss": 0.1186, "step": 2998 }, { "epoch": 3.04, "learning_rate": 1.8472028730546193e-05, "loss": 0.1683, "step": 2999 }, { "epoch": 3.04, "learning_rate": 1.847086389619761e-05, "loss": 0.1472, "step": 3000 }, { "epoch": 3.05, "learning_rate": 1.8469698654774003e-05, "loss": 0.164, "step": 3001 }, { "epoch": 3.05, "learning_rate": 1.8468533006331373e-05, "loss": 0.1779, "step": 3002 }, { "epoch": 3.05, "learning_rate": 1.8467366950925732e-05, "loss": 0.134, "step": 3003 }, { "epoch": 3.05, "learning_rate": 1.8466200488613117e-05, "loss": 0.1427, "step": 3004 }, { "epoch": 3.05, "learning_rate": 1.8465033619449584e-05, "loss": 0.1476, "step": 3005 }, { "epoch": 3.05, "learning_rate": 1.8463866343491208e-05, "loss": 0.1783, "step": 3006 }, { "epoch": 3.05, "learning_rate": 1.846269866079408e-05, "loss": 0.1606, "step": 3007 }, { "epoch": 3.05, "learning_rate": 1.8461530571414323e-05, "loss": 0.1459, "step": 3008 }, { "epoch": 3.05, "learning_rate": 1.8460362075408063e-05, "loss": 0.1693, "step": 3009 }, { "epoch": 3.05, "learning_rate": 1.845919317283145e-05, "loss": 0.1987, "step": 3010 }, { "epoch": 3.06, "learning_rate": 1.8458023863740667e-05, "loss": 0.1396, "step": 3011 }, { "epoch": 3.06, "learning_rate": 1.84568541481919e-05, "loss": 0.1444, "step": 3012 }, { "epoch": 3.06, "learning_rate": 1.845568402624136e-05, "loss": 0.1628, "step": 3013 }, { "epoch": 3.06, "learning_rate": 1.8454513497945282e-05, "loss": 0.1622, "step": 3014 }, { "epoch": 3.06, "learning_rate": 1.8453342563359916e-05, "loss": 0.1705, "step": 3015 }, { "epoch": 3.06, "learning_rate": 1.8452171222541532e-05, "loss": 0.1626, "step": 3016 }, { "epoch": 3.06, "learning_rate": 1.8450999475546415e-05, "loss": 0.1368, "step": 3017 }, { "epoch": 3.06, "learning_rate": 1.844982732243088e-05, "loss": 0.153, "step": 3018 }, { "epoch": 3.06, "learning_rate": 1.8448654763251254e-05, "loss": 0.1551, "step": 3019 }, { "epoch": 3.06, "learning_rate": 1.8447481798063883e-05, "loss": 0.1499, "step": 3020 }, { "epoch": 3.07, "learning_rate": 1.8446308426925136e-05, "loss": 0.1377, "step": 3021 }, { "epoch": 3.07, "learning_rate": 1.8445134649891406e-05, "loss": 0.1569, "step": 3022 }, { "epoch": 3.07, "learning_rate": 1.84439604670191e-05, "loss": 0.1633, "step": 3023 }, { "epoch": 3.07, "learning_rate": 1.844278587836463e-05, "loss": 0.1668, "step": 3024 }, { "epoch": 3.07, "learning_rate": 1.844161088398446e-05, "loss": 0.1411, "step": 3025 }, { "epoch": 3.07, "learning_rate": 1.8440435483935042e-05, "loss": 0.1601, "step": 3026 }, { "epoch": 3.07, "learning_rate": 1.843925967827287e-05, "loss": 0.1346, "step": 3027 }, { "epoch": 3.07, "learning_rate": 1.8438083467054444e-05, "loss": 0.1583, "step": 3028 }, { "epoch": 3.07, "learning_rate": 1.8436906850336287e-05, "loss": 0.1702, "step": 3029 }, { "epoch": 3.07, "learning_rate": 1.8435729828174948e-05, "loss": 0.1508, "step": 3030 }, { "epoch": 3.08, "learning_rate": 1.8434552400626984e-05, "loss": 0.1761, "step": 3031 }, { "epoch": 3.08, "learning_rate": 1.8433374567748977e-05, "loss": 0.1411, "step": 3032 }, { "epoch": 3.08, "learning_rate": 1.8432196329597533e-05, "loss": 0.1413, "step": 3033 }, { "epoch": 3.08, "learning_rate": 1.8431017686229272e-05, "loss": 0.1592, "step": 3034 }, { "epoch": 3.08, "learning_rate": 1.8429838637700835e-05, "loss": 0.1621, "step": 3035 }, { "epoch": 3.08, "learning_rate": 1.842865918406888e-05, "loss": 0.1685, "step": 3036 }, { "epoch": 3.08, "learning_rate": 1.8427479325390084e-05, "loss": 0.1389, "step": 3037 }, { "epoch": 3.08, "learning_rate": 1.8426299061721157e-05, "loss": 0.1654, "step": 3038 }, { "epoch": 3.08, "learning_rate": 1.842511839311881e-05, "loss": 0.1589, "step": 3039 }, { "epoch": 3.08, "learning_rate": 1.8423937319639778e-05, "loss": 0.135, "step": 3040 }, { "epoch": 3.09, "learning_rate": 1.8422755841340824e-05, "loss": 0.1675, "step": 3041 }, { "epoch": 3.09, "learning_rate": 1.8421573958278727e-05, "loss": 0.1457, "step": 3042 }, { "epoch": 3.09, "learning_rate": 1.8420391670510278e-05, "loss": 0.1582, "step": 3043 }, { "epoch": 3.09, "learning_rate": 1.8419208978092292e-05, "loss": 0.1739, "step": 3044 }, { "epoch": 3.09, "learning_rate": 1.8418025881081612e-05, "loss": 0.1535, "step": 3045 }, { "epoch": 3.09, "learning_rate": 1.8416842379535085e-05, "loss": 0.1791, "step": 3046 }, { "epoch": 3.09, "learning_rate": 1.841565847350959e-05, "loss": 0.1889, "step": 3047 }, { "epoch": 3.09, "learning_rate": 1.8414474163062018e-05, "loss": 0.1763, "step": 3048 }, { "epoch": 3.09, "learning_rate": 1.8413289448249282e-05, "loss": 0.1553, "step": 3049 }, { "epoch": 3.09, "learning_rate": 1.841210432912832e-05, "loss": 0.1347, "step": 3050 }, { "epoch": 3.1, "learning_rate": 1.8410918805756076e-05, "loss": 0.1672, "step": 3051 }, { "epoch": 3.1, "learning_rate": 1.8409732878189528e-05, "loss": 0.1493, "step": 3052 }, { "epoch": 3.1, "learning_rate": 1.8408546546485658e-05, "loss": 0.168, "step": 3053 }, { "epoch": 3.1, "learning_rate": 1.840735981070149e-05, "loss": 0.1739, "step": 3054 }, { "epoch": 3.1, "learning_rate": 1.8406172670894042e-05, "loss": 0.1709, "step": 3055 }, { "epoch": 3.1, "learning_rate": 1.8404985127120367e-05, "loss": 0.1728, "step": 3056 }, { "epoch": 3.1, "learning_rate": 1.840379717943753e-05, "loss": 0.1756, "step": 3057 }, { "epoch": 3.1, "learning_rate": 1.8402608827902628e-05, "loss": 0.1673, "step": 3058 }, { "epoch": 3.1, "learning_rate": 1.8401420072572757e-05, "loss": 0.1446, "step": 3059 }, { "epoch": 3.11, "learning_rate": 1.8400230913505053e-05, "loss": 0.1454, "step": 3060 }, { "epoch": 3.11, "learning_rate": 1.8399041350756655e-05, "loss": 0.162, "step": 3061 }, { "epoch": 3.11, "learning_rate": 1.839785138438474e-05, "loss": 0.1432, "step": 3062 }, { "epoch": 3.11, "learning_rate": 1.839666101444648e-05, "loss": 0.1426, "step": 3063 }, { "epoch": 3.11, "learning_rate": 1.839547024099908e-05, "loss": 0.1806, "step": 3064 }, { "epoch": 3.11, "learning_rate": 1.839427906409977e-05, "loss": 0.1643, "step": 3065 }, { "epoch": 3.11, "learning_rate": 1.8393087483805795e-05, "loss": 0.177, "step": 3066 }, { "epoch": 3.11, "learning_rate": 1.8391895500174408e-05, "loss": 0.1494, "step": 3067 }, { "epoch": 3.11, "learning_rate": 1.83907031132629e-05, "loss": 0.1755, "step": 3068 }, { "epoch": 3.11, "learning_rate": 1.8389510323128566e-05, "loss": 0.1541, "step": 3069 }, { "epoch": 3.12, "learning_rate": 1.838831712982873e-05, "loss": 0.1463, "step": 3070 }, { "epoch": 3.12, "learning_rate": 1.838712353342073e-05, "loss": 0.1563, "step": 3071 }, { "epoch": 3.12, "learning_rate": 1.838592953396193e-05, "loss": 0.1679, "step": 3072 }, { "epoch": 3.12, "learning_rate": 1.8384735131509698e-05, "loss": 0.1754, "step": 3073 }, { "epoch": 3.12, "learning_rate": 1.8383540326121444e-05, "loss": 0.1376, "step": 3074 }, { "epoch": 3.12, "learning_rate": 1.8382345117854584e-05, "loss": 0.1688, "step": 3075 }, { "epoch": 3.12, "learning_rate": 1.8381149506766544e-05, "loss": 0.1624, "step": 3076 }, { "epoch": 3.12, "learning_rate": 1.8379953492914793e-05, "loss": 0.1552, "step": 3077 }, { "epoch": 3.12, "learning_rate": 1.83787570763568e-05, "loss": 0.1565, "step": 3078 }, { "epoch": 3.12, "learning_rate": 1.837756025715006e-05, "loss": 0.1699, "step": 3079 }, { "epoch": 3.13, "learning_rate": 1.837636303535209e-05, "loss": 0.1532, "step": 3080 }, { "epoch": 3.13, "learning_rate": 1.837516541102042e-05, "loss": 0.1803, "step": 3081 }, { "epoch": 3.13, "learning_rate": 1.8373967384212607e-05, "loss": 0.1503, "step": 3082 }, { "epoch": 3.13, "learning_rate": 1.837276895498622e-05, "loss": 0.1539, "step": 3083 }, { "epoch": 3.13, "learning_rate": 1.8371570123398853e-05, "loss": 0.1561, "step": 3084 }, { "epoch": 3.13, "learning_rate": 1.837037088950811e-05, "loss": 0.1751, "step": 3085 }, { "epoch": 3.13, "learning_rate": 1.836917125337163e-05, "loss": 0.1677, "step": 3086 }, { "epoch": 3.13, "learning_rate": 1.836797121504706e-05, "loss": 0.16, "step": 3087 }, { "epoch": 3.13, "learning_rate": 1.836677077459207e-05, "loss": 0.1746, "step": 3088 }, { "epoch": 3.13, "learning_rate": 1.8365569932064343e-05, "loss": 0.1697, "step": 3089 }, { "epoch": 3.14, "learning_rate": 1.8364368687521596e-05, "loss": 0.1878, "step": 3090 }, { "epoch": 3.14, "learning_rate": 1.8363167041021545e-05, "loss": 0.1785, "step": 3091 }, { "epoch": 3.14, "learning_rate": 1.8361964992621942e-05, "loss": 0.1641, "step": 3092 }, { "epoch": 3.14, "learning_rate": 1.836076254238055e-05, "loss": 0.1515, "step": 3093 }, { "epoch": 3.14, "learning_rate": 1.835955969035516e-05, "loss": 0.1544, "step": 3094 }, { "epoch": 3.14, "learning_rate": 1.835835643660357e-05, "loss": 0.1684, "step": 3095 }, { "epoch": 3.14, "learning_rate": 1.8357152781183606e-05, "loss": 0.1925, "step": 3096 }, { "epoch": 3.14, "learning_rate": 1.8355948724153108e-05, "loss": 0.1736, "step": 3097 }, { "epoch": 3.14, "learning_rate": 1.8354744265569938e-05, "loss": 0.1472, "step": 3098 }, { "epoch": 3.14, "learning_rate": 1.8353539405491985e-05, "loss": 0.1696, "step": 3099 }, { "epoch": 3.15, "learning_rate": 1.8352334143977143e-05, "loss": 0.161, "step": 3100 }, { "epoch": 3.15, "learning_rate": 1.8351128481083328e-05, "loss": 0.1706, "step": 3101 }, { "epoch": 3.15, "learning_rate": 1.834992241686849e-05, "loss": 0.1494, "step": 3102 }, { "epoch": 3.15, "learning_rate": 1.8348715951390578e-05, "loss": 0.1539, "step": 3103 }, { "epoch": 3.15, "learning_rate": 1.8347509084707575e-05, "loss": 0.1816, "step": 3104 }, { "epoch": 3.15, "learning_rate": 1.8346301816877473e-05, "loss": 0.1639, "step": 3105 }, { "epoch": 3.15, "learning_rate": 1.8345094147958297e-05, "loss": 0.1519, "step": 3106 }, { "epoch": 3.15, "learning_rate": 1.8343886078008076e-05, "loss": 0.1585, "step": 3107 }, { "epoch": 3.15, "learning_rate": 1.8342677607084868e-05, "loss": 0.1655, "step": 3108 }, { "epoch": 3.15, "learning_rate": 1.8341468735246743e-05, "loss": 0.1591, "step": 3109 }, { "epoch": 3.16, "learning_rate": 1.8340259462551796e-05, "loss": 0.197, "step": 3110 }, { "epoch": 3.16, "learning_rate": 1.8339049789058142e-05, "loss": 0.1787, "step": 3111 }, { "epoch": 3.16, "learning_rate": 1.833783971482391e-05, "loss": 0.1689, "step": 3112 }, { "epoch": 3.16, "learning_rate": 1.8336629239907255e-05, "loss": 0.1788, "step": 3113 }, { "epoch": 3.16, "learning_rate": 1.8335418364366347e-05, "loss": 0.1673, "step": 3114 }, { "epoch": 3.16, "learning_rate": 1.833420708825937e-05, "loss": 0.2047, "step": 3115 }, { "epoch": 3.16, "learning_rate": 1.833299541164454e-05, "loss": 0.2041, "step": 3116 }, { "epoch": 3.16, "learning_rate": 1.833178333458008e-05, "loss": 0.167, "step": 3117 }, { "epoch": 3.16, "learning_rate": 1.8330570857124242e-05, "loss": 0.1508, "step": 3118 }, { "epoch": 3.16, "learning_rate": 1.8329357979335286e-05, "loss": 0.1366, "step": 3119 }, { "epoch": 3.17, "learning_rate": 1.8328144701271505e-05, "loss": 0.1885, "step": 3120 }, { "epoch": 3.17, "learning_rate": 1.8326931022991202e-05, "loss": 0.1621, "step": 3121 }, { "epoch": 3.17, "learning_rate": 1.8325716944552704e-05, "loss": 0.1626, "step": 3122 }, { "epoch": 3.17, "learning_rate": 1.8324502466014345e-05, "loss": 0.1553, "step": 3123 }, { "epoch": 3.17, "learning_rate": 1.83232875874345e-05, "loss": 0.1588, "step": 3124 }, { "epoch": 3.17, "learning_rate": 1.8322072308871542e-05, "loss": 0.1514, "step": 3125 }, { "epoch": 3.17, "learning_rate": 1.8320856630383877e-05, "loss": 0.1679, "step": 3126 }, { "epoch": 3.17, "learning_rate": 1.8319640552029928e-05, "loss": 0.1706, "step": 3127 }, { "epoch": 3.17, "learning_rate": 1.8318424073868123e-05, "loss": 0.1509, "step": 3128 }, { "epoch": 3.18, "learning_rate": 1.8317207195956936e-05, "loss": 0.1936, "step": 3129 }, { "epoch": 3.18, "learning_rate": 1.8315989918354835e-05, "loss": 0.1535, "step": 3130 }, { "epoch": 3.18, "learning_rate": 1.8314772241120322e-05, "loss": 0.1462, "step": 3131 }, { "epoch": 3.18, "learning_rate": 1.8313554164311908e-05, "loss": 0.1995, "step": 3132 }, { "epoch": 3.18, "learning_rate": 1.8312335687988135e-05, "loss": 0.1779, "step": 3133 }, { "epoch": 3.18, "learning_rate": 1.831111681220756e-05, "loss": 0.1455, "step": 3134 }, { "epoch": 3.18, "learning_rate": 1.8309897537028752e-05, "loss": 0.1395, "step": 3135 }, { "epoch": 3.18, "learning_rate": 1.8308677862510303e-05, "loss": 0.1713, "step": 3136 }, { "epoch": 3.18, "learning_rate": 1.830745778871083e-05, "loss": 0.1716, "step": 3137 }, { "epoch": 3.18, "learning_rate": 1.8306237315688958e-05, "loss": 0.1743, "step": 3138 }, { "epoch": 3.19, "learning_rate": 1.8305016443503346e-05, "loss": 0.1551, "step": 3139 }, { "epoch": 3.19, "learning_rate": 1.8303795172212663e-05, "loss": 0.1675, "step": 3140 }, { "epoch": 3.19, "learning_rate": 1.8302573501875593e-05, "loss": 0.1521, "step": 3141 }, { "epoch": 3.19, "learning_rate": 1.8301351432550846e-05, "loss": 0.1657, "step": 3142 }, { "epoch": 3.19, "learning_rate": 1.8300128964297155e-05, "loss": 0.1639, "step": 3143 }, { "epoch": 3.19, "learning_rate": 1.8298906097173265e-05, "loss": 0.1559, "step": 3144 }, { "epoch": 3.19, "learning_rate": 1.8297682831237934e-05, "loss": 0.1554, "step": 3145 }, { "epoch": 3.19, "learning_rate": 1.8296459166549955e-05, "loss": 0.1648, "step": 3146 }, { "epoch": 3.19, "learning_rate": 1.829523510316813e-05, "loss": 0.1937, "step": 3147 }, { "epoch": 3.19, "learning_rate": 1.8294010641151288e-05, "loss": 0.1795, "step": 3148 }, { "epoch": 3.2, "learning_rate": 1.829278578055826e-05, "loss": 0.1898, "step": 3149 }, { "epoch": 3.2, "learning_rate": 1.829156052144792e-05, "loss": 0.143, "step": 3150 }, { "epoch": 3.2, "learning_rate": 1.829033486387914e-05, "loss": 0.1577, "step": 3151 }, { "epoch": 3.2, "learning_rate": 1.8289108807910826e-05, "loss": 0.171, "step": 3152 }, { "epoch": 3.2, "learning_rate": 1.8287882353601892e-05, "loss": 0.1808, "step": 3153 }, { "epoch": 3.2, "learning_rate": 1.828665550101128e-05, "loss": 0.1505, "step": 3154 }, { "epoch": 3.2, "learning_rate": 1.8285428250197947e-05, "loss": 0.171, "step": 3155 }, { "epoch": 3.2, "learning_rate": 1.8284200601220866e-05, "loss": 0.1735, "step": 3156 }, { "epoch": 3.2, "learning_rate": 1.8282972554139038e-05, "loss": 0.1511, "step": 3157 }, { "epoch": 3.2, "learning_rate": 1.8281744109011475e-05, "loss": 0.1707, "step": 3158 }, { "epoch": 3.21, "learning_rate": 1.8280515265897215e-05, "loss": 0.1516, "step": 3159 }, { "epoch": 3.21, "learning_rate": 1.8279286024855306e-05, "loss": 0.1602, "step": 3160 }, { "epoch": 3.21, "learning_rate": 1.8278056385944823e-05, "loss": 0.1614, "step": 3161 }, { "epoch": 3.21, "learning_rate": 1.8276826349224856e-05, "loss": 0.177, "step": 3162 }, { "epoch": 3.21, "learning_rate": 1.8275595914754517e-05, "loss": 0.1722, "step": 3163 }, { "epoch": 3.21, "learning_rate": 1.8274365082592932e-05, "loss": 0.1874, "step": 3164 }, { "epoch": 3.21, "learning_rate": 1.8273133852799256e-05, "loss": 0.1735, "step": 3165 }, { "epoch": 3.21, "learning_rate": 1.827190222543265e-05, "loss": 0.1722, "step": 3166 }, { "epoch": 3.21, "learning_rate": 1.827067020055231e-05, "loss": 0.164, "step": 3167 }, { "epoch": 3.21, "learning_rate": 1.8269437778217432e-05, "loss": 0.1717, "step": 3168 }, { "epoch": 3.22, "learning_rate": 1.826820495848725e-05, "loss": 0.1842, "step": 3169 }, { "epoch": 3.22, "learning_rate": 1.8266971741421004e-05, "loss": 0.1432, "step": 3170 }, { "epoch": 3.22, "learning_rate": 1.8265738127077952e-05, "loss": 0.1793, "step": 3171 }, { "epoch": 3.22, "learning_rate": 1.8264504115517386e-05, "loss": 0.1391, "step": 3172 }, { "epoch": 3.22, "learning_rate": 1.82632697067986e-05, "loss": 0.1687, "step": 3173 }, { "epoch": 3.22, "learning_rate": 1.826203490098092e-05, "loss": 0.1542, "step": 3174 }, { "epoch": 3.22, "learning_rate": 1.8260799698123685e-05, "loss": 0.1646, "step": 3175 }, { "epoch": 3.22, "learning_rate": 1.825956409828625e-05, "loss": 0.1486, "step": 3176 }, { "epoch": 3.22, "learning_rate": 1.8258328101528e-05, "loss": 0.1662, "step": 3177 }, { "epoch": 3.22, "learning_rate": 1.825709170790832e-05, "loss": 0.1719, "step": 3178 }, { "epoch": 3.23, "learning_rate": 1.8255854917486637e-05, "loss": 0.1441, "step": 3179 }, { "epoch": 3.23, "learning_rate": 1.825461773032238e-05, "loss": 0.1799, "step": 3180 }, { "epoch": 3.23, "learning_rate": 1.825338014647501e-05, "loss": 0.16, "step": 3181 }, { "epoch": 3.23, "learning_rate": 1.8252142166003993e-05, "loss": 0.1483, "step": 3182 }, { "epoch": 3.23, "learning_rate": 1.825090378896883e-05, "loss": 0.1572, "step": 3183 }, { "epoch": 3.23, "learning_rate": 1.824966501542902e-05, "loss": 0.1506, "step": 3184 }, { "epoch": 3.23, "learning_rate": 1.8248425845444098e-05, "loss": 0.1681, "step": 3185 }, { "epoch": 3.23, "learning_rate": 1.8247186279073618e-05, "loss": 0.1875, "step": 3186 }, { "epoch": 3.23, "learning_rate": 1.8245946316377146e-05, "loss": 0.1687, "step": 3187 }, { "epoch": 3.23, "learning_rate": 1.8244705957414268e-05, "loss": 0.1606, "step": 3188 }, { "epoch": 3.24, "learning_rate": 1.8243465202244593e-05, "loss": 0.1584, "step": 3189 }, { "epoch": 3.24, "learning_rate": 1.8242224050927746e-05, "loss": 0.152, "step": 3190 }, { "epoch": 3.24, "learning_rate": 1.824098250352337e-05, "loss": 0.1726, "step": 3191 }, { "epoch": 3.24, "learning_rate": 1.823974056009113e-05, "loss": 0.1556, "step": 3192 }, { "epoch": 3.24, "learning_rate": 1.8238498220690707e-05, "loss": 0.1831, "step": 3193 }, { "epoch": 3.24, "learning_rate": 1.8237255485381808e-05, "loss": 0.1574, "step": 3194 }, { "epoch": 3.24, "learning_rate": 1.8236012354224146e-05, "loss": 0.1615, "step": 3195 }, { "epoch": 3.24, "learning_rate": 1.8234768827277465e-05, "loss": 0.1817, "step": 3196 }, { "epoch": 3.24, "learning_rate": 1.8233524904601524e-05, "loss": 0.1855, "step": 3197 }, { "epoch": 3.25, "learning_rate": 1.82322805862561e-05, "loss": 0.1706, "step": 3198 }, { "epoch": 3.25, "learning_rate": 1.823103587230099e-05, "loss": 0.1497, "step": 3199 }, { "epoch": 3.25, "learning_rate": 1.822979076279601e-05, "loss": 0.1966, "step": 3200 }, { "epoch": 3.25, "learning_rate": 1.8228545257801e-05, "loss": 0.1633, "step": 3201 }, { "epoch": 3.25, "learning_rate": 1.8227299357375802e-05, "loss": 0.1625, "step": 3202 }, { "epoch": 3.25, "learning_rate": 1.82260530615803e-05, "loss": 0.1706, "step": 3203 }, { "epoch": 3.25, "learning_rate": 1.8224806370474378e-05, "loss": 0.1487, "step": 3204 }, { "epoch": 3.25, "learning_rate": 1.8223559284117955e-05, "loss": 0.1659, "step": 3205 }, { "epoch": 3.25, "learning_rate": 1.822231180257095e-05, "loss": 0.1875, "step": 3206 }, { "epoch": 3.25, "learning_rate": 1.8221063925893324e-05, "loss": 0.1598, "step": 3207 }, { "epoch": 3.26, "learning_rate": 1.8219815654145035e-05, "loss": 0.1881, "step": 3208 }, { "epoch": 3.26, "learning_rate": 1.8218566987386075e-05, "loss": 0.1729, "step": 3209 }, { "epoch": 3.26, "learning_rate": 1.821731792567645e-05, "loss": 0.1593, "step": 3210 }, { "epoch": 3.26, "learning_rate": 1.821606846907618e-05, "loss": 0.1595, "step": 3211 }, { "epoch": 3.26, "learning_rate": 1.821481861764532e-05, "loss": 0.1841, "step": 3212 }, { "epoch": 3.26, "learning_rate": 1.8213568371443914e-05, "loss": 0.1705, "step": 3213 }, { "epoch": 3.26, "learning_rate": 1.821231773053206e-05, "loss": 0.1605, "step": 3214 }, { "epoch": 3.26, "learning_rate": 1.8211066694969855e-05, "loss": 0.1738, "step": 3215 }, { "epoch": 3.26, "learning_rate": 1.8209815264817415e-05, "loss": 0.1533, "step": 3216 }, { "epoch": 3.26, "learning_rate": 1.820856344013488e-05, "loss": 0.1507, "step": 3217 }, { "epoch": 3.27, "learning_rate": 1.820731122098241e-05, "loss": 0.1658, "step": 3218 }, { "epoch": 3.27, "learning_rate": 1.8206058607420178e-05, "loss": 0.1644, "step": 3219 }, { "epoch": 3.27, "learning_rate": 1.820480559950838e-05, "loss": 0.1739, "step": 3220 }, { "epoch": 3.27, "learning_rate": 1.8203552197307233e-05, "loss": 0.1612, "step": 3221 }, { "epoch": 3.27, "learning_rate": 1.8202298400876968e-05, "loss": 0.1479, "step": 3222 }, { "epoch": 3.27, "learning_rate": 1.820104421027784e-05, "loss": 0.1736, "step": 3223 }, { "epoch": 3.27, "learning_rate": 1.8199789625570113e-05, "loss": 0.1751, "step": 3224 }, { "epoch": 3.27, "learning_rate": 1.8198534646814093e-05, "loss": 0.1624, "step": 3225 }, { "epoch": 3.27, "learning_rate": 1.819727927407007e-05, "loss": 0.1664, "step": 3226 }, { "epoch": 3.27, "learning_rate": 1.8196023507398383e-05, "loss": 0.1673, "step": 3227 }, { "epoch": 3.28, "learning_rate": 1.819476734685938e-05, "loss": 0.163, "step": 3228 }, { "epoch": 3.28, "learning_rate": 1.819351079251342e-05, "loss": 0.1837, "step": 3229 }, { "epoch": 3.28, "learning_rate": 1.8192253844420894e-05, "loss": 0.1644, "step": 3230 }, { "epoch": 3.28, "learning_rate": 1.8190996502642204e-05, "loss": 0.1763, "step": 3231 }, { "epoch": 3.28, "learning_rate": 1.8189738767237768e-05, "loss": 0.1875, "step": 3232 }, { "epoch": 3.28, "learning_rate": 1.8188480638268036e-05, "loss": 0.2022, "step": 3233 }, { "epoch": 3.28, "learning_rate": 1.818722211579346e-05, "loss": 0.1668, "step": 3234 }, { "epoch": 3.28, "learning_rate": 1.8185963199874528e-05, "loss": 0.1596, "step": 3235 }, { "epoch": 3.28, "learning_rate": 1.8184703890571732e-05, "loss": 0.1655, "step": 3236 }, { "epoch": 3.28, "learning_rate": 1.8183444187945593e-05, "loss": 0.1846, "step": 3237 }, { "epoch": 3.29, "learning_rate": 1.8182184092056645e-05, "loss": 0.1752, "step": 3238 }, { "epoch": 3.29, "learning_rate": 1.8180923602965442e-05, "loss": 0.1689, "step": 3239 }, { "epoch": 3.29, "learning_rate": 1.817966272073256e-05, "loss": 0.1927, "step": 3240 }, { "epoch": 3.29, "learning_rate": 1.8178401445418592e-05, "loss": 0.1711, "step": 3241 }, { "epoch": 3.29, "learning_rate": 1.8177139777084146e-05, "loss": 0.1599, "step": 3242 }, { "epoch": 3.29, "learning_rate": 1.817587771578986e-05, "loss": 0.1916, "step": 3243 }, { "epoch": 3.29, "learning_rate": 1.8174615261596376e-05, "loss": 0.1756, "step": 3244 }, { "epoch": 3.29, "learning_rate": 1.817335241456437e-05, "loss": 0.1814, "step": 3245 }, { "epoch": 3.29, "learning_rate": 1.817208917475452e-05, "loss": 0.1667, "step": 3246 }, { "epoch": 3.29, "learning_rate": 1.817082554222754e-05, "loss": 0.133, "step": 3247 }, { "epoch": 3.3, "learning_rate": 1.816956151704415e-05, "loss": 0.1626, "step": 3248 }, { "epoch": 3.3, "learning_rate": 1.8168297099265094e-05, "loss": 0.1644, "step": 3249 }, { "epoch": 3.3, "learning_rate": 1.816703228895114e-05, "loss": 0.167, "step": 3250 }, { "epoch": 3.3, "learning_rate": 1.8165767086163066e-05, "loss": 0.1616, "step": 3251 }, { "epoch": 3.3, "learning_rate": 1.816450149096167e-05, "loss": 0.1827, "step": 3252 }, { "epoch": 3.3, "learning_rate": 1.8163235503407776e-05, "loss": 0.1635, "step": 3253 }, { "epoch": 3.3, "learning_rate": 1.816196912356222e-05, "loss": 0.174, "step": 3254 }, { "epoch": 3.3, "learning_rate": 1.8160702351485855e-05, "loss": 0.1799, "step": 3255 }, { "epoch": 3.3, "learning_rate": 1.8159435187239567e-05, "loss": 0.1598, "step": 3256 }, { "epoch": 3.3, "learning_rate": 1.815816763088424e-05, "loss": 0.1754, "step": 3257 }, { "epoch": 3.31, "learning_rate": 1.8156899682480795e-05, "loss": 0.2035, "step": 3258 }, { "epoch": 3.31, "learning_rate": 1.815563134209016e-05, "loss": 0.1885, "step": 3259 }, { "epoch": 3.31, "learning_rate": 1.815436260977329e-05, "loss": 0.2026, "step": 3260 }, { "epoch": 3.31, "learning_rate": 1.8153093485591153e-05, "loss": 0.1801, "step": 3261 }, { "epoch": 3.31, "learning_rate": 1.8151823969604735e-05, "loss": 0.1782, "step": 3262 }, { "epoch": 3.31, "learning_rate": 1.8150554061875048e-05, "loss": 0.176, "step": 3263 }, { "epoch": 3.31, "learning_rate": 1.814928376246312e-05, "loss": 0.1697, "step": 3264 }, { "epoch": 3.31, "learning_rate": 1.814801307142999e-05, "loss": 0.1712, "step": 3265 }, { "epoch": 3.31, "learning_rate": 1.8146741988836728e-05, "loss": 0.1714, "step": 3266 }, { "epoch": 3.32, "learning_rate": 1.8145470514744415e-05, "loss": 0.1617, "step": 3267 }, { "epoch": 3.32, "learning_rate": 1.8144198649214148e-05, "loss": 0.1587, "step": 3268 }, { "epoch": 3.32, "learning_rate": 1.8142926392307057e-05, "loss": 0.1888, "step": 3269 }, { "epoch": 3.32, "learning_rate": 1.8141653744084276e-05, "loss": 0.1721, "step": 3270 }, { "epoch": 3.32, "learning_rate": 1.814038070460696e-05, "loss": 0.1647, "step": 3271 }, { "epoch": 3.32, "learning_rate": 1.8139107273936296e-05, "loss": 0.1688, "step": 3272 }, { "epoch": 3.32, "learning_rate": 1.8137833452133473e-05, "loss": 0.1484, "step": 3273 }, { "epoch": 3.32, "learning_rate": 1.8136559239259702e-05, "loss": 0.1837, "step": 3274 }, { "epoch": 3.32, "learning_rate": 1.8135284635376228e-05, "loss": 0.1765, "step": 3275 }, { "epoch": 3.32, "learning_rate": 1.813400964054429e-05, "loss": 0.2168, "step": 3276 }, { "epoch": 3.33, "learning_rate": 1.813273425482517e-05, "loss": 0.164, "step": 3277 }, { "epoch": 3.33, "learning_rate": 1.813145847828015e-05, "loss": 0.1586, "step": 3278 }, { "epoch": 3.33, "learning_rate": 1.813018231097054e-05, "loss": 0.1691, "step": 3279 }, { "epoch": 3.33, "learning_rate": 1.8128905752957676e-05, "loss": 0.1855, "step": 3280 }, { "epoch": 3.33, "learning_rate": 1.8127628804302892e-05, "loss": 0.1668, "step": 3281 }, { "epoch": 3.33, "learning_rate": 1.812635146506756e-05, "loss": 0.1593, "step": 3282 }, { "epoch": 3.33, "learning_rate": 1.812507373531306e-05, "loss": 0.1786, "step": 3283 }, { "epoch": 3.33, "learning_rate": 1.81237956151008e-05, "loss": 0.1562, "step": 3284 }, { "epoch": 3.33, "learning_rate": 1.812251710449219e-05, "loss": 0.1854, "step": 3285 }, { "epoch": 3.33, "learning_rate": 1.8121238203548684e-05, "loss": 0.1693, "step": 3286 }, { "epoch": 3.34, "learning_rate": 1.811995891233173e-05, "loss": 0.1806, "step": 3287 }, { "epoch": 3.34, "learning_rate": 1.8118679230902817e-05, "loss": 0.1612, "step": 3288 }, { "epoch": 3.34, "learning_rate": 1.811739915932343e-05, "loss": 0.1653, "step": 3289 }, { "epoch": 3.34, "learning_rate": 1.811611869765509e-05, "loss": 0.1771, "step": 3290 }, { "epoch": 3.34, "learning_rate": 1.8114837845959322e-05, "loss": 0.1787, "step": 3291 }, { "epoch": 3.34, "learning_rate": 1.811355660429769e-05, "loss": 0.1718, "step": 3292 }, { "epoch": 3.34, "learning_rate": 1.8112274972731757e-05, "loss": 0.1482, "step": 3293 }, { "epoch": 3.34, "learning_rate": 1.8110992951323125e-05, "loss": 0.1621, "step": 3294 }, { "epoch": 3.34, "learning_rate": 1.8109710540133386e-05, "loss": 0.2045, "step": 3295 }, { "epoch": 3.34, "learning_rate": 1.8108427739224177e-05, "loss": 0.1941, "step": 3296 }, { "epoch": 3.35, "learning_rate": 1.8107144548657143e-05, "loss": 0.1615, "step": 3297 }, { "epoch": 3.35, "learning_rate": 1.810586096849395e-05, "loss": 0.1688, "step": 3298 }, { "epoch": 3.35, "learning_rate": 1.810457699879628e-05, "loss": 0.1792, "step": 3299 }, { "epoch": 3.35, "learning_rate": 1.810329263962584e-05, "loss": 0.1668, "step": 3300 }, { "epoch": 3.35, "learning_rate": 1.810200789104434e-05, "loss": 0.1588, "step": 3301 }, { "epoch": 3.35, "learning_rate": 1.8100722753113525e-05, "loss": 0.1655, "step": 3302 }, { "epoch": 3.35, "learning_rate": 1.809943722589516e-05, "loss": 0.1451, "step": 3303 }, { "epoch": 3.35, "learning_rate": 1.8098151309451014e-05, "loss": 0.1639, "step": 3304 }, { "epoch": 3.35, "learning_rate": 1.8096865003842887e-05, "loss": 0.1614, "step": 3305 }, { "epoch": 3.35, "learning_rate": 1.8095578309132592e-05, "loss": 0.2044, "step": 3306 }, { "epoch": 3.36, "learning_rate": 1.8094291225381965e-05, "loss": 0.157, "step": 3307 }, { "epoch": 3.36, "learning_rate": 1.8093003752652853e-05, "loss": 0.2018, "step": 3308 }, { "epoch": 3.36, "learning_rate": 1.809171589100713e-05, "loss": 0.1648, "step": 3309 }, { "epoch": 3.36, "learning_rate": 1.8090427640506683e-05, "loss": 0.1852, "step": 3310 }, { "epoch": 3.36, "learning_rate": 1.8089139001213425e-05, "loss": 0.1709, "step": 3311 }, { "epoch": 3.36, "learning_rate": 1.8087849973189276e-05, "loss": 0.1756, "step": 3312 }, { "epoch": 3.36, "learning_rate": 1.808656055649619e-05, "loss": 0.1956, "step": 3313 }, { "epoch": 3.36, "learning_rate": 1.8085270751196122e-05, "loss": 0.1783, "step": 3314 }, { "epoch": 3.36, "learning_rate": 1.8083980557351058e-05, "loss": 0.1649, "step": 3315 }, { "epoch": 3.36, "learning_rate": 1.8082689975023e-05, "loss": 0.1747, "step": 3316 }, { "epoch": 3.37, "learning_rate": 1.8081399004273972e-05, "loss": 0.1825, "step": 3317 }, { "epoch": 3.37, "learning_rate": 1.8080107645166004e-05, "loss": 0.1516, "step": 3318 }, { "epoch": 3.37, "learning_rate": 1.8078815897761162e-05, "loss": 0.1464, "step": 3319 }, { "epoch": 3.37, "learning_rate": 1.807752376212152e-05, "loss": 0.149, "step": 3320 }, { "epoch": 3.37, "learning_rate": 1.807623123830917e-05, "loss": 0.1966, "step": 3321 }, { "epoch": 3.37, "learning_rate": 1.8074938326386222e-05, "loss": 0.1741, "step": 3322 }, { "epoch": 3.37, "learning_rate": 1.807364502641482e-05, "loss": 0.1775, "step": 3323 }, { "epoch": 3.37, "learning_rate": 1.8072351338457102e-05, "loss": 0.1665, "step": 3324 }, { "epoch": 3.37, "learning_rate": 1.8071057262575246e-05, "loss": 0.1557, "step": 3325 }, { "epoch": 3.37, "learning_rate": 1.8069762798831433e-05, "loss": 0.1693, "step": 3326 }, { "epoch": 3.38, "learning_rate": 1.8068467947287877e-05, "loss": 0.1611, "step": 3327 }, { "epoch": 3.38, "learning_rate": 1.8067172708006796e-05, "loss": 0.1655, "step": 3328 }, { "epoch": 3.38, "learning_rate": 1.806587708105044e-05, "loss": 0.1686, "step": 3329 }, { "epoch": 3.38, "learning_rate": 1.806458106648107e-05, "loss": 0.185, "step": 3330 }, { "epoch": 3.38, "learning_rate": 1.8063284664360968e-05, "loss": 0.1802, "step": 3331 }, { "epoch": 3.38, "learning_rate": 1.8061987874752428e-05, "loss": 0.1827, "step": 3332 }, { "epoch": 3.38, "learning_rate": 1.8060690697717773e-05, "loss": 0.1696, "step": 3333 }, { "epoch": 3.38, "learning_rate": 1.805939313331934e-05, "loss": 0.1906, "step": 3334 }, { "epoch": 3.38, "learning_rate": 1.8058095181619485e-05, "loss": 0.169, "step": 3335 }, { "epoch": 3.39, "learning_rate": 1.805679684268058e-05, "loss": 0.1787, "step": 3336 }, { "epoch": 3.39, "learning_rate": 1.8055498116565016e-05, "loss": 0.1674, "step": 3337 }, { "epoch": 3.39, "learning_rate": 1.8054199003335214e-05, "loss": 0.1922, "step": 3338 }, { "epoch": 3.39, "learning_rate": 1.8052899503053594e-05, "loss": 0.1793, "step": 3339 }, { "epoch": 3.39, "learning_rate": 1.8051599615782608e-05, "loss": 0.1677, "step": 3340 }, { "epoch": 3.39, "learning_rate": 1.8050299341584728e-05, "loss": 0.1866, "step": 3341 }, { "epoch": 3.39, "learning_rate": 1.804899868052243e-05, "loss": 0.1459, "step": 3342 }, { "epoch": 3.39, "learning_rate": 1.8047697632658224e-05, "loss": 0.1506, "step": 3343 }, { "epoch": 3.39, "learning_rate": 1.804639619805464e-05, "loss": 0.1747, "step": 3344 }, { "epoch": 3.39, "learning_rate": 1.8045094376774205e-05, "loss": 0.1568, "step": 3345 }, { "epoch": 3.4, "learning_rate": 1.8043792168879487e-05, "loss": 0.1775, "step": 3346 }, { "epoch": 3.4, "learning_rate": 1.8042489574433068e-05, "loss": 0.1707, "step": 3347 }, { "epoch": 3.4, "learning_rate": 1.804118659349754e-05, "loss": 0.156, "step": 3348 }, { "epoch": 3.4, "learning_rate": 1.803988322613552e-05, "loss": 0.173, "step": 3349 }, { "epoch": 3.4, "learning_rate": 1.8038579472409648e-05, "loss": 0.1791, "step": 3350 }, { "epoch": 3.4, "learning_rate": 1.803727533238257e-05, "loss": 0.1748, "step": 3351 }, { "epoch": 3.4, "learning_rate": 1.8035970806116956e-05, "loss": 0.1713, "step": 3352 }, { "epoch": 3.4, "learning_rate": 1.8034665893675504e-05, "loss": 0.191, "step": 3353 }, { "epoch": 3.4, "learning_rate": 1.8033360595120917e-05, "loss": 0.1658, "step": 3354 }, { "epoch": 3.4, "learning_rate": 1.8032054910515923e-05, "loss": 0.1523, "step": 3355 }, { "epoch": 3.41, "learning_rate": 1.8030748839923273e-05, "loss": 0.1739, "step": 3356 }, { "epoch": 3.41, "learning_rate": 1.8029442383405725e-05, "loss": 0.1668, "step": 3357 }, { "epoch": 3.41, "learning_rate": 1.8028135541026063e-05, "loss": 0.1869, "step": 3358 }, { "epoch": 3.41, "learning_rate": 1.802682831284709e-05, "loss": 0.1635, "step": 3359 }, { "epoch": 3.41, "learning_rate": 1.802552069893163e-05, "loss": 0.1884, "step": 3360 }, { "epoch": 3.41, "learning_rate": 1.8024212699342513e-05, "loss": 0.1978, "step": 3361 }, { "epoch": 3.41, "learning_rate": 1.8022904314142603e-05, "loss": 0.145, "step": 3362 }, { "epoch": 3.41, "learning_rate": 1.8021595543394773e-05, "loss": 0.1558, "step": 3363 }, { "epoch": 3.41, "learning_rate": 1.8020286387161917e-05, "loss": 0.1746, "step": 3364 }, { "epoch": 3.41, "learning_rate": 1.801897684550695e-05, "loss": 0.1669, "step": 3365 }, { "epoch": 3.42, "learning_rate": 1.80176669184928e-05, "loss": 0.1817, "step": 3366 }, { "epoch": 3.42, "learning_rate": 1.8016356606182417e-05, "loss": 0.1926, "step": 3367 }, { "epoch": 3.42, "learning_rate": 1.801504590863877e-05, "loss": 0.1755, "step": 3368 }, { "epoch": 3.42, "learning_rate": 1.8013734825924845e-05, "loss": 0.1812, "step": 3369 }, { "epoch": 3.42, "learning_rate": 1.8012423358103653e-05, "loss": 0.1656, "step": 3370 }, { "epoch": 3.42, "learning_rate": 1.801111150523821e-05, "loss": 0.1674, "step": 3371 }, { "epoch": 3.42, "learning_rate": 1.800979926739156e-05, "loss": 0.1611, "step": 3372 }, { "epoch": 3.42, "learning_rate": 1.8008486644626768e-05, "loss": 0.1667, "step": 3373 }, { "epoch": 3.42, "learning_rate": 1.800717363700691e-05, "loss": 0.1727, "step": 3374 }, { "epoch": 3.42, "learning_rate": 1.8005860244595082e-05, "loss": 0.167, "step": 3375 }, { "epoch": 3.43, "learning_rate": 1.8004546467454403e-05, "loss": 0.1639, "step": 3376 }, { "epoch": 3.43, "learning_rate": 1.800323230564801e-05, "loss": 0.1721, "step": 3377 }, { "epoch": 3.43, "learning_rate": 1.800191775923905e-05, "loss": 0.1905, "step": 3378 }, { "epoch": 3.43, "learning_rate": 1.8000602828290697e-05, "loss": 0.1634, "step": 3379 }, { "epoch": 3.43, "learning_rate": 1.7999287512866147e-05, "loss": 0.1814, "step": 3380 }, { "epoch": 3.43, "learning_rate": 1.79979718130286e-05, "loss": 0.1882, "step": 3381 }, { "epoch": 3.43, "learning_rate": 1.799665572884129e-05, "loss": 0.1962, "step": 3382 }, { "epoch": 3.43, "learning_rate": 1.799533926036746e-05, "loss": 0.2039, "step": 3383 }, { "epoch": 3.43, "learning_rate": 1.7994022407670372e-05, "loss": 0.1673, "step": 3384 }, { "epoch": 3.43, "learning_rate": 1.7992705170813306e-05, "loss": 0.1609, "step": 3385 }, { "epoch": 3.44, "learning_rate": 1.7991387549859576e-05, "loss": 0.1744, "step": 3386 }, { "epoch": 3.44, "learning_rate": 1.7990069544872488e-05, "loss": 0.1902, "step": 3387 }, { "epoch": 3.44, "learning_rate": 1.7988751155915386e-05, "loss": 0.1658, "step": 3388 }, { "epoch": 3.44, "learning_rate": 1.7987432383051623e-05, "loss": 0.1616, "step": 3389 }, { "epoch": 3.44, "learning_rate": 1.7986113226344577e-05, "loss": 0.1808, "step": 3390 }, { "epoch": 3.44, "learning_rate": 1.7984793685857643e-05, "loss": 0.1873, "step": 3391 }, { "epoch": 3.44, "learning_rate": 1.7983473761654226e-05, "loss": 0.1782, "step": 3392 }, { "epoch": 3.44, "learning_rate": 1.7982153453797763e-05, "loss": 0.169, "step": 3393 }, { "epoch": 3.44, "learning_rate": 1.7980832762351696e-05, "loss": 0.1941, "step": 3394 }, { "epoch": 3.44, "learning_rate": 1.7979511687379498e-05, "loss": 0.1775, "step": 3395 }, { "epoch": 3.45, "learning_rate": 1.7978190228944653e-05, "loss": 0.1802, "step": 3396 }, { "epoch": 3.45, "learning_rate": 1.7976868387110662e-05, "loss": 0.1861, "step": 3397 }, { "epoch": 3.45, "learning_rate": 1.797554616194105e-05, "loss": 0.1766, "step": 3398 }, { "epoch": 3.45, "learning_rate": 1.797422355349936e-05, "loss": 0.1853, "step": 3399 }, { "epoch": 3.45, "learning_rate": 1.7972900561849146e-05, "loss": 0.2084, "step": 3400 }, { "epoch": 3.45, "learning_rate": 1.7971577187053985e-05, "loss": 0.1834, "step": 3401 }, { "epoch": 3.45, "learning_rate": 1.7970253429177477e-05, "loss": 0.1565, "step": 3402 }, { "epoch": 3.45, "learning_rate": 1.7968929288283238e-05, "loss": 0.1722, "step": 3403 }, { "epoch": 3.45, "learning_rate": 1.7967604764434897e-05, "loss": 0.1633, "step": 3404 }, { "epoch": 3.46, "learning_rate": 1.7966279857696106e-05, "loss": 0.1785, "step": 3405 }, { "epoch": 3.46, "learning_rate": 1.7964954568130533e-05, "loss": 0.1628, "step": 3406 }, { "epoch": 3.46, "learning_rate": 1.7963628895801867e-05, "loss": 0.1851, "step": 3407 }, { "epoch": 3.46, "learning_rate": 1.796230284077382e-05, "loss": 0.1729, "step": 3408 }, { "epoch": 3.46, "learning_rate": 1.7960976403110106e-05, "loss": 0.18, "step": 3409 }, { "epoch": 3.46, "learning_rate": 1.7959649582874478e-05, "loss": 0.1535, "step": 3410 }, { "epoch": 3.46, "learning_rate": 1.795832238013069e-05, "loss": 0.1817, "step": 3411 }, { "epoch": 3.46, "learning_rate": 1.795699479494253e-05, "loss": 0.1562, "step": 3412 }, { "epoch": 3.46, "learning_rate": 1.795566682737379e-05, "loss": 0.1564, "step": 3413 }, { "epoch": 3.46, "learning_rate": 1.7954338477488284e-05, "loss": 0.1937, "step": 3414 }, { "epoch": 3.47, "learning_rate": 1.7953009745349854e-05, "loss": 0.1883, "step": 3415 }, { "epoch": 3.47, "learning_rate": 1.7951680631022356e-05, "loss": 0.1806, "step": 3416 }, { "epoch": 3.47, "learning_rate": 1.7950351134569652e-05, "loss": 0.1873, "step": 3417 }, { "epoch": 3.47, "learning_rate": 1.7949021256055635e-05, "loss": 0.1857, "step": 3418 }, { "epoch": 3.47, "learning_rate": 1.7947690995544217e-05, "loss": 0.1718, "step": 3419 }, { "epoch": 3.47, "learning_rate": 1.7946360353099327e-05, "loss": 0.1718, "step": 3420 }, { "epoch": 3.47, "learning_rate": 1.7945029328784903e-05, "loss": 0.1795, "step": 3421 }, { "epoch": 3.47, "learning_rate": 1.7943697922664913e-05, "loss": 0.1735, "step": 3422 }, { "epoch": 3.47, "learning_rate": 1.794236613480334e-05, "loss": 0.1829, "step": 3423 }, { "epoch": 3.47, "learning_rate": 1.7941033965264178e-05, "loss": 0.1691, "step": 3424 }, { "epoch": 3.48, "learning_rate": 1.793970141411145e-05, "loss": 0.1859, "step": 3425 }, { "epoch": 3.48, "learning_rate": 1.7938368481409197e-05, "loss": 0.1913, "step": 3426 }, { "epoch": 3.48, "learning_rate": 1.793703516722147e-05, "loss": 0.2105, "step": 3427 }, { "epoch": 3.48, "learning_rate": 1.793570147161234e-05, "loss": 0.178, "step": 3428 }, { "epoch": 3.48, "learning_rate": 1.79343673946459e-05, "loss": 0.1705, "step": 3429 }, { "epoch": 3.48, "learning_rate": 1.7933032936386263e-05, "loss": 0.1783, "step": 3430 }, { "epoch": 3.48, "learning_rate": 1.7931698096897558e-05, "loss": 0.1877, "step": 3431 }, { "epoch": 3.48, "learning_rate": 1.793036287624393e-05, "loss": 0.1761, "step": 3432 }, { "epoch": 3.48, "learning_rate": 1.7929027274489542e-05, "loss": 0.1683, "step": 3433 }, { "epoch": 3.48, "learning_rate": 1.7927691291698587e-05, "loss": 0.1865, "step": 3434 }, { "epoch": 3.49, "learning_rate": 1.7926354927935255e-05, "loss": 0.1924, "step": 3435 }, { "epoch": 3.49, "learning_rate": 1.7925018183263772e-05, "loss": 0.1774, "step": 3436 }, { "epoch": 3.49, "learning_rate": 1.7923681057748376e-05, "loss": 0.1809, "step": 3437 }, { "epoch": 3.49, "learning_rate": 1.7922343551453327e-05, "loss": 0.162, "step": 3438 }, { "epoch": 3.49, "learning_rate": 1.792100566444289e-05, "loss": 0.1666, "step": 3439 }, { "epoch": 3.49, "learning_rate": 1.7919667396781366e-05, "loss": 0.1926, "step": 3440 }, { "epoch": 3.49, "learning_rate": 1.791832874853307e-05, "loss": 0.1672, "step": 3441 }, { "epoch": 3.49, "learning_rate": 1.7916989719762325e-05, "loss": 0.1956, "step": 3442 }, { "epoch": 3.49, "learning_rate": 1.7915650310533485e-05, "loss": 0.185, "step": 3443 }, { "epoch": 3.49, "learning_rate": 1.7914310520910907e-05, "loss": 0.1646, "step": 3444 }, { "epoch": 3.5, "learning_rate": 1.7912970350958985e-05, "loss": 0.1911, "step": 3445 }, { "epoch": 3.5, "learning_rate": 1.791162980074212e-05, "loss": 0.1684, "step": 3446 }, { "epoch": 3.5, "learning_rate": 1.791028887032473e-05, "loss": 0.1717, "step": 3447 }, { "epoch": 3.5, "learning_rate": 1.790894755977126e-05, "loss": 0.1962, "step": 3448 }, { "epoch": 3.5, "learning_rate": 1.7907605869146167e-05, "loss": 0.1774, "step": 3449 }, { "epoch": 3.5, "learning_rate": 1.7906263798513925e-05, "loss": 0.1724, "step": 3450 }, { "epoch": 3.5, "learning_rate": 1.7904921347939023e-05, "loss": 0.1835, "step": 3451 }, { "epoch": 3.5, "learning_rate": 1.7903578517485986e-05, "loss": 0.2172, "step": 3452 }, { "epoch": 3.5, "learning_rate": 1.7902235307219333e-05, "loss": 0.1996, "step": 3453 }, { "epoch": 3.5, "learning_rate": 1.7900891717203622e-05, "loss": 0.1733, "step": 3454 }, { "epoch": 3.51, "learning_rate": 1.7899547747503417e-05, "loss": 0.1794, "step": 3455 }, { "epoch": 3.51, "learning_rate": 1.78982033981833e-05, "loss": 0.202, "step": 3456 }, { "epoch": 3.51, "learning_rate": 1.789685866930788e-05, "loss": 0.1614, "step": 3457 }, { "epoch": 3.51, "learning_rate": 1.7895513560941776e-05, "loss": 0.1789, "step": 3458 }, { "epoch": 3.51, "learning_rate": 1.7894168073149635e-05, "loss": 0.1923, "step": 3459 }, { "epoch": 3.51, "learning_rate": 1.7892822205996108e-05, "loss": 0.1614, "step": 3460 }, { "epoch": 3.51, "learning_rate": 1.7891475959545873e-05, "loss": 0.1732, "step": 3461 }, { "epoch": 3.51, "learning_rate": 1.789012933386363e-05, "loss": 0.1629, "step": 3462 }, { "epoch": 3.51, "learning_rate": 1.7888782329014086e-05, "loss": 0.1846, "step": 3463 }, { "epoch": 3.51, "learning_rate": 1.788743494506198e-05, "loss": 0.2063, "step": 3464 }, { "epoch": 3.52, "learning_rate": 1.788608718207205e-05, "loss": 0.1792, "step": 3465 }, { "epoch": 3.52, "learning_rate": 1.7884739040109075e-05, "loss": 0.1976, "step": 3466 }, { "epoch": 3.52, "learning_rate": 1.788339051923784e-05, "loss": 0.1657, "step": 3467 }, { "epoch": 3.52, "learning_rate": 1.788204161952314e-05, "loss": 0.2138, "step": 3468 }, { "epoch": 3.52, "learning_rate": 1.7880692341029808e-05, "loss": 0.194, "step": 3469 }, { "epoch": 3.52, "learning_rate": 1.7879342683822683e-05, "loss": 0.1656, "step": 3470 }, { "epoch": 3.52, "learning_rate": 1.787799264796662e-05, "loss": 0.1627, "step": 3471 }, { "epoch": 3.52, "learning_rate": 1.7876642233526498e-05, "loss": 0.173, "step": 3472 }, { "epoch": 3.52, "learning_rate": 1.7875291440567217e-05, "loss": 0.174, "step": 3473 }, { "epoch": 3.53, "learning_rate": 1.7873940269153682e-05, "loss": 0.1839, "step": 3474 }, { "epoch": 3.53, "learning_rate": 1.787258871935083e-05, "loss": 0.1739, "step": 3475 }, { "epoch": 3.53, "learning_rate": 1.787123679122361e-05, "loss": 0.1639, "step": 3476 }, { "epoch": 3.53, "learning_rate": 1.786988448483699e-05, "loss": 0.1933, "step": 3477 }, { "epoch": 3.53, "learning_rate": 1.7868531800255958e-05, "loss": 0.186, "step": 3478 }, { "epoch": 3.53, "learning_rate": 1.7867178737545513e-05, "loss": 0.1924, "step": 3479 }, { "epoch": 3.53, "learning_rate": 1.7865825296770684e-05, "loss": 0.1835, "step": 3480 }, { "epoch": 3.53, "learning_rate": 1.7864471477996514e-05, "loss": 0.1641, "step": 3481 }, { "epoch": 3.53, "learning_rate": 1.786311728128805e-05, "loss": 0.17, "step": 3482 }, { "epoch": 3.53, "learning_rate": 1.786176270671038e-05, "loss": 0.1977, "step": 3483 }, { "epoch": 3.54, "learning_rate": 1.7860407754328598e-05, "loss": 0.1715, "step": 3484 }, { "epoch": 3.54, "learning_rate": 1.785905242420781e-05, "loss": 0.189, "step": 3485 }, { "epoch": 3.54, "learning_rate": 1.7857696716413154e-05, "loss": 0.1852, "step": 3486 }, { "epoch": 3.54, "learning_rate": 1.7856340631009784e-05, "loss": 0.1846, "step": 3487 }, { "epoch": 3.54, "learning_rate": 1.7854984168062857e-05, "loss": 0.1764, "step": 3488 }, { "epoch": 3.54, "learning_rate": 1.7853627327637567e-05, "loss": 0.1831, "step": 3489 }, { "epoch": 3.54, "learning_rate": 1.7852270109799115e-05, "loss": 0.1768, "step": 3490 }, { "epoch": 3.54, "learning_rate": 1.7850912514612724e-05, "loss": 0.1836, "step": 3491 }, { "epoch": 3.54, "learning_rate": 1.7849554542143638e-05, "loss": 0.1709, "step": 3492 }, { "epoch": 3.54, "learning_rate": 1.784819619245711e-05, "loss": 0.1739, "step": 3493 }, { "epoch": 3.55, "learning_rate": 1.7846837465618417e-05, "loss": 0.1783, "step": 3494 }, { "epoch": 3.55, "learning_rate": 1.7845478361692857e-05, "loss": 0.1758, "step": 3495 }, { "epoch": 3.55, "learning_rate": 1.7844118880745743e-05, "loss": 0.207, "step": 3496 }, { "epoch": 3.55, "learning_rate": 1.7842759022842404e-05, "loss": 0.1773, "step": 3497 }, { "epoch": 3.55, "learning_rate": 1.784139878804819e-05, "loss": 0.1714, "step": 3498 }, { "epoch": 3.55, "learning_rate": 1.7840038176428468e-05, "loss": 0.1864, "step": 3499 }, { "epoch": 3.55, "learning_rate": 1.7838677188048622e-05, "loss": 0.1858, "step": 3500 }, { "epoch": 3.55, "learning_rate": 1.783731582297406e-05, "loss": 0.1849, "step": 3501 }, { "epoch": 3.55, "learning_rate": 1.78359540812702e-05, "loss": 0.1671, "step": 3502 }, { "epoch": 3.55, "learning_rate": 1.7834591963002487e-05, "loss": 0.1866, "step": 3503 }, { "epoch": 3.56, "learning_rate": 1.7833229468236367e-05, "loss": 0.2, "step": 3504 }, { "epoch": 3.56, "learning_rate": 1.7831866597037327e-05, "loss": 0.1968, "step": 3505 }, { "epoch": 3.56, "learning_rate": 1.783050334947086e-05, "loss": 0.1939, "step": 3506 }, { "epoch": 3.56, "learning_rate": 1.7829139725602472e-05, "loss": 0.1683, "step": 3507 }, { "epoch": 3.56, "learning_rate": 1.78277757254977e-05, "loss": 0.1816, "step": 3508 }, { "epoch": 3.56, "learning_rate": 1.7826411349222085e-05, "loss": 0.184, "step": 3509 }, { "epoch": 3.56, "learning_rate": 1.78250465968412e-05, "loss": 0.1806, "step": 3510 }, { "epoch": 3.56, "learning_rate": 1.7823681468420622e-05, "loss": 0.1736, "step": 3511 }, { "epoch": 3.56, "learning_rate": 1.7822315964025963e-05, "loss": 0.1788, "step": 3512 }, { "epoch": 3.56, "learning_rate": 1.7820950083722838e-05, "loss": 0.1885, "step": 3513 }, { "epoch": 3.57, "learning_rate": 1.7819583827576887e-05, "loss": 0.1599, "step": 3514 }, { "epoch": 3.57, "learning_rate": 1.7818217195653767e-05, "loss": 0.1787, "step": 3515 }, { "epoch": 3.57, "learning_rate": 1.7816850188019148e-05, "loss": 0.1889, "step": 3516 }, { "epoch": 3.57, "learning_rate": 1.7815482804738725e-05, "loss": 0.1615, "step": 3517 }, { "epoch": 3.57, "learning_rate": 1.781411504587822e-05, "loss": 0.1833, "step": 3518 }, { "epoch": 3.57, "learning_rate": 1.781274691150334e-05, "loss": 0.1581, "step": 3519 }, { "epoch": 3.57, "learning_rate": 1.7811378401679853e-05, "loss": 0.1973, "step": 3520 }, { "epoch": 3.57, "learning_rate": 1.781000951647351e-05, "loss": 0.1808, "step": 3521 }, { "epoch": 3.57, "learning_rate": 1.7808640255950104e-05, "loss": 0.1883, "step": 3522 }, { "epoch": 3.57, "learning_rate": 1.7807270620175426e-05, "loss": 0.1767, "step": 3523 }, { "epoch": 3.58, "learning_rate": 1.7805900609215303e-05, "loss": 0.1784, "step": 3524 }, { "epoch": 3.58, "learning_rate": 1.780453022313557e-05, "loss": 0.1695, "step": 3525 }, { "epoch": 3.58, "learning_rate": 1.7803159462002083e-05, "loss": 0.1684, "step": 3526 }, { "epoch": 3.58, "learning_rate": 1.780178832588071e-05, "loss": 0.1925, "step": 3527 }, { "epoch": 3.58, "learning_rate": 1.780041681483735e-05, "loss": 0.18, "step": 3528 }, { "epoch": 3.58, "learning_rate": 1.7799044928937907e-05, "loss": 0.1727, "step": 3529 }, { "epoch": 3.58, "learning_rate": 1.7797672668248305e-05, "loss": 0.186, "step": 3530 }, { "epoch": 3.58, "learning_rate": 1.7796300032834498e-05, "loss": 0.1875, "step": 3531 }, { "epoch": 3.58, "learning_rate": 1.7794927022762444e-05, "loss": 0.1887, "step": 3532 }, { "epoch": 3.58, "learning_rate": 1.7793553638098125e-05, "loss": 0.1858, "step": 3533 }, { "epoch": 3.59, "learning_rate": 1.7792179878907537e-05, "loss": 0.1742, "step": 3534 }, { "epoch": 3.59, "learning_rate": 1.7790805745256703e-05, "loss": 0.1649, "step": 3535 }, { "epoch": 3.59, "learning_rate": 1.778943123721166e-05, "loss": 0.1724, "step": 3536 }, { "epoch": 3.59, "learning_rate": 1.7788056354838452e-05, "loss": 0.1704, "step": 3537 }, { "epoch": 3.59, "learning_rate": 1.778668109820316e-05, "loss": 0.2131, "step": 3538 }, { "epoch": 3.59, "learning_rate": 1.7785305467371863e-05, "loss": 0.2013, "step": 3539 }, { "epoch": 3.59, "learning_rate": 1.7783929462410675e-05, "loss": 0.1942, "step": 3540 }, { "epoch": 3.59, "learning_rate": 1.778255308338572e-05, "loss": 0.1689, "step": 3541 }, { "epoch": 3.59, "learning_rate": 1.778117633036314e-05, "loss": 0.1806, "step": 3542 }, { "epoch": 3.6, "learning_rate": 1.77797992034091e-05, "loss": 0.1642, "step": 3543 }, { "epoch": 3.6, "learning_rate": 1.7778421702589774e-05, "loss": 0.1915, "step": 3544 }, { "epoch": 3.6, "learning_rate": 1.7777043827971362e-05, "loss": 0.1832, "step": 3545 }, { "epoch": 3.6, "learning_rate": 1.7775665579620076e-05, "loss": 0.1748, "step": 3546 }, { "epoch": 3.6, "learning_rate": 1.7774286957602153e-05, "loss": 0.1999, "step": 3547 }, { "epoch": 3.6, "learning_rate": 1.7772907961983846e-05, "loss": 0.1861, "step": 3548 }, { "epoch": 3.6, "learning_rate": 1.777152859283141e-05, "loss": 0.1825, "step": 3549 }, { "epoch": 3.6, "learning_rate": 1.777014885021115e-05, "loss": 0.183, "step": 3550 }, { "epoch": 3.6, "learning_rate": 1.7768768734189362e-05, "loss": 0.1824, "step": 3551 }, { "epoch": 3.6, "learning_rate": 1.7767388244832365e-05, "loss": 0.2196, "step": 3552 }, { "epoch": 3.61, "learning_rate": 1.7766007382206505e-05, "loss": 0.1838, "step": 3553 }, { "epoch": 3.61, "learning_rate": 1.7764626146378146e-05, "loss": 0.1679, "step": 3554 }, { "epoch": 3.61, "learning_rate": 1.776324453741365e-05, "loss": 0.1849, "step": 3555 }, { "epoch": 3.61, "learning_rate": 1.7761862555379425e-05, "loss": 0.183, "step": 3556 }, { "epoch": 3.61, "learning_rate": 1.7760480200341875e-05, "loss": 0.1797, "step": 3557 }, { "epoch": 3.61, "learning_rate": 1.775909747236743e-05, "loss": 0.1894, "step": 3558 }, { "epoch": 3.61, "learning_rate": 1.7757714371522548e-05, "loss": 0.1861, "step": 3559 }, { "epoch": 3.61, "learning_rate": 1.7756330897873684e-05, "loss": 0.1636, "step": 3560 }, { "epoch": 3.61, "learning_rate": 1.775494705148733e-05, "loss": 0.1799, "step": 3561 }, { "epoch": 3.61, "learning_rate": 1.7753562832429986e-05, "loss": 0.1703, "step": 3562 }, { "epoch": 3.62, "learning_rate": 1.7752178240768166e-05, "loss": 0.1863, "step": 3563 }, { "epoch": 3.62, "learning_rate": 1.7750793276568414e-05, "loss": 0.1927, "step": 3564 }, { "epoch": 3.62, "learning_rate": 1.7749407939897285e-05, "loss": 0.1891, "step": 3565 }, { "epoch": 3.62, "learning_rate": 1.774802223082135e-05, "loss": 0.1949, "step": 3566 }, { "epoch": 3.62, "learning_rate": 1.7746636149407205e-05, "loss": 0.1811, "step": 3567 }, { "epoch": 3.62, "learning_rate": 1.7745249695721453e-05, "loss": 0.2032, "step": 3568 }, { "epoch": 3.62, "learning_rate": 1.7743862869830728e-05, "loss": 0.2015, "step": 3569 }, { "epoch": 3.62, "learning_rate": 1.7742475671801675e-05, "loss": 0.1733, "step": 3570 }, { "epoch": 3.62, "learning_rate": 1.7741088101700948e-05, "loss": 0.2012, "step": 3571 }, { "epoch": 3.62, "learning_rate": 1.7739700159595236e-05, "loss": 0.1948, "step": 3572 }, { "epoch": 3.63, "learning_rate": 1.773831184555124e-05, "loss": 0.2033, "step": 3573 }, { "epoch": 3.63, "learning_rate": 1.7736923159635667e-05, "loss": 0.1791, "step": 3574 }, { "epoch": 3.63, "learning_rate": 1.7735534101915258e-05, "loss": 0.1603, "step": 3575 }, { "epoch": 3.63, "learning_rate": 1.7734144672456767e-05, "loss": 0.1824, "step": 3576 }, { "epoch": 3.63, "learning_rate": 1.7732754871326958e-05, "loss": 0.1838, "step": 3577 }, { "epoch": 3.63, "learning_rate": 1.7731364698592625e-05, "loss": 0.2079, "step": 3578 }, { "epoch": 3.63, "learning_rate": 1.7729974154320574e-05, "loss": 0.184, "step": 3579 }, { "epoch": 3.63, "learning_rate": 1.7728583238577623e-05, "loss": 0.2064, "step": 3580 }, { "epoch": 3.63, "learning_rate": 1.7727191951430623e-05, "loss": 0.1757, "step": 3581 }, { "epoch": 3.63, "learning_rate": 1.7725800292946425e-05, "loss": 0.1728, "step": 3582 }, { "epoch": 3.64, "learning_rate": 1.7724408263191908e-05, "loss": 0.1797, "step": 3583 }, { "epoch": 3.64, "learning_rate": 1.772301586223397e-05, "loss": 0.1699, "step": 3584 }, { "epoch": 3.64, "learning_rate": 1.7721623090139525e-05, "loss": 0.1971, "step": 3585 }, { "epoch": 3.64, "learning_rate": 1.77202299469755e-05, "loss": 0.163, "step": 3586 }, { "epoch": 3.64, "learning_rate": 1.7718836432808844e-05, "loss": 0.1965, "step": 3587 }, { "epoch": 3.64, "learning_rate": 1.7717442547706526e-05, "loss": 0.2088, "step": 3588 }, { "epoch": 3.64, "learning_rate": 1.7716048291735534e-05, "loss": 0.1743, "step": 3589 }, { "epoch": 3.64, "learning_rate": 1.7714653664962863e-05, "loss": 0.1655, "step": 3590 }, { "epoch": 3.64, "learning_rate": 1.7713258667455532e-05, "loss": 0.1755, "step": 3591 }, { "epoch": 3.64, "learning_rate": 1.7711863299280587e-05, "loss": 0.1784, "step": 3592 }, { "epoch": 3.65, "learning_rate": 1.7710467560505078e-05, "loss": 0.1895, "step": 3593 }, { "epoch": 3.65, "learning_rate": 1.770907145119608e-05, "loss": 0.1911, "step": 3594 }, { "epoch": 3.65, "learning_rate": 1.7707674971420686e-05, "loss": 0.1603, "step": 3595 }, { "epoch": 3.65, "learning_rate": 1.7706278121246e-05, "loss": 0.1976, "step": 3596 }, { "epoch": 3.65, "learning_rate": 1.7704880900739155e-05, "loss": 0.1958, "step": 3597 }, { "epoch": 3.65, "learning_rate": 1.770348330996729e-05, "loss": 0.1667, "step": 3598 }, { "epoch": 3.65, "learning_rate": 1.7702085348997575e-05, "loss": 0.1625, "step": 3599 }, { "epoch": 3.65, "learning_rate": 1.770068701789718e-05, "loss": 0.1861, "step": 3600 }, { "epoch": 3.65, "learning_rate": 1.7699288316733312e-05, "loss": 0.1712, "step": 3601 }, { "epoch": 3.65, "learning_rate": 1.769788924557318e-05, "loss": 0.1897, "step": 3602 }, { "epoch": 3.66, "learning_rate": 1.7696489804484028e-05, "loss": 0.1898, "step": 3603 }, { "epoch": 3.66, "learning_rate": 1.7695089993533095e-05, "loss": 0.1662, "step": 3604 }, { "epoch": 3.66, "learning_rate": 1.7693689812787656e-05, "loss": 0.1716, "step": 3605 }, { "epoch": 3.66, "learning_rate": 1.7692289262315e-05, "loss": 0.2052, "step": 3606 }, { "epoch": 3.66, "learning_rate": 1.769088834218243e-05, "loss": 0.1941, "step": 3607 }, { "epoch": 3.66, "learning_rate": 1.7689487052457264e-05, "loss": 0.1905, "step": 3608 }, { "epoch": 3.66, "learning_rate": 1.768808539320685e-05, "loss": 0.1718, "step": 3609 }, { "epoch": 3.66, "learning_rate": 1.768668336449854e-05, "loss": 0.2141, "step": 3610 }, { "epoch": 3.66, "learning_rate": 1.768528096639971e-05, "loss": 0.1893, "step": 3611 }, { "epoch": 3.67, "learning_rate": 1.768387819897776e-05, "loss": 0.1984, "step": 3612 }, { "epoch": 3.67, "learning_rate": 1.7682475062300094e-05, "loss": 0.1708, "step": 3613 }, { "epoch": 3.67, "learning_rate": 1.7681071556434147e-05, "loss": 0.191, "step": 3614 }, { "epoch": 3.67, "learning_rate": 1.767966768144736e-05, "loss": 0.1784, "step": 3615 }, { "epoch": 3.67, "learning_rate": 1.76782634374072e-05, "loss": 0.1767, "step": 3616 }, { "epoch": 3.67, "learning_rate": 1.7676858824381147e-05, "loss": 0.1982, "step": 3617 }, { "epoch": 3.67, "learning_rate": 1.7675453842436704e-05, "loss": 0.1881, "step": 3618 }, { "epoch": 3.67, "learning_rate": 1.767404849164139e-05, "loss": 0.2121, "step": 3619 }, { "epoch": 3.67, "learning_rate": 1.7672642772062734e-05, "loss": 0.1741, "step": 3620 }, { "epoch": 3.67, "learning_rate": 1.76712366837683e-05, "loss": 0.179, "step": 3621 }, { "epoch": 3.68, "learning_rate": 1.7669830226825645e-05, "loss": 0.1831, "step": 3622 }, { "epoch": 3.68, "learning_rate": 1.7668423401302372e-05, "loss": 0.2044, "step": 3623 }, { "epoch": 3.68, "learning_rate": 1.7667016207266075e-05, "loss": 0.184, "step": 3624 }, { "epoch": 3.68, "learning_rate": 1.766560864478438e-05, "loss": 0.1857, "step": 3625 }, { "epoch": 3.68, "learning_rate": 1.766420071392494e-05, "loss": 0.1758, "step": 3626 }, { "epoch": 3.68, "learning_rate": 1.76627924147554e-05, "loss": 0.178, "step": 3627 }, { "epoch": 3.68, "learning_rate": 1.7661383747343444e-05, "loss": 0.1812, "step": 3628 }, { "epoch": 3.68, "learning_rate": 1.7659974711756767e-05, "loss": 0.1864, "step": 3629 }, { "epoch": 3.68, "learning_rate": 1.765856530806308e-05, "loss": 0.1963, "step": 3630 }, { "epoch": 3.68, "learning_rate": 1.7657155536330113e-05, "loss": 0.1858, "step": 3631 }, { "epoch": 3.69, "learning_rate": 1.7655745396625614e-05, "loss": 0.1955, "step": 3632 }, { "epoch": 3.69, "learning_rate": 1.765433488901735e-05, "loss": 0.1977, "step": 3633 }, { "epoch": 3.69, "learning_rate": 1.7652924013573104e-05, "loss": 0.1838, "step": 3634 }, { "epoch": 3.69, "learning_rate": 1.7651512770360675e-05, "loss": 0.183, "step": 3635 }, { "epoch": 3.69, "learning_rate": 1.7650101159447883e-05, "loss": 0.1824, "step": 3636 }, { "epoch": 3.69, "learning_rate": 1.7648689180902564e-05, "loss": 0.1915, "step": 3637 }, { "epoch": 3.69, "learning_rate": 1.764727683479257e-05, "loss": 0.1647, "step": 3638 }, { "epoch": 3.69, "learning_rate": 1.7645864121185778e-05, "loss": 0.2113, "step": 3639 }, { "epoch": 3.69, "learning_rate": 1.764445104015007e-05, "loss": 0.1602, "step": 3640 }, { "epoch": 3.69, "learning_rate": 1.7643037591753356e-05, "loss": 0.1725, "step": 3641 }, { "epoch": 3.7, "learning_rate": 1.764162377606356e-05, "loss": 0.1688, "step": 3642 }, { "epoch": 3.7, "learning_rate": 1.7640209593148627e-05, "loss": 0.1889, "step": 3643 }, { "epoch": 3.7, "learning_rate": 1.763879504307652e-05, "loss": 0.1524, "step": 3644 }, { "epoch": 3.7, "learning_rate": 1.7637380125915205e-05, "loss": 0.1896, "step": 3645 }, { "epoch": 3.7, "learning_rate": 1.7635964841732686e-05, "loss": 0.1749, "step": 3646 }, { "epoch": 3.7, "learning_rate": 1.763454919059697e-05, "loss": 0.185, "step": 3647 }, { "epoch": 3.7, "learning_rate": 1.7633133172576098e-05, "loss": 0.1846, "step": 3648 }, { "epoch": 3.7, "learning_rate": 1.7631716787738103e-05, "loss": 0.1685, "step": 3649 }, { "epoch": 3.7, "learning_rate": 1.763030003615106e-05, "loss": 0.1816, "step": 3650 }, { "epoch": 3.7, "learning_rate": 1.762888291788306e-05, "loss": 0.1702, "step": 3651 }, { "epoch": 3.71, "learning_rate": 1.7627465433002183e-05, "loss": 0.1862, "step": 3652 }, { "epoch": 3.71, "learning_rate": 1.7626047581576564e-05, "loss": 0.1727, "step": 3653 }, { "epoch": 3.71, "learning_rate": 1.7624629363674334e-05, "loss": 0.1756, "step": 3654 }, { "epoch": 3.71, "learning_rate": 1.7623210779363646e-05, "loss": 0.1421, "step": 3655 }, { "epoch": 3.71, "learning_rate": 1.7621791828712675e-05, "loss": 0.1938, "step": 3656 }, { "epoch": 3.71, "learning_rate": 1.7620372511789607e-05, "loss": 0.19, "step": 3657 }, { "epoch": 3.71, "learning_rate": 1.7618952828662646e-05, "loss": 0.2083, "step": 3658 }, { "epoch": 3.71, "learning_rate": 1.7617532779400022e-05, "loss": 0.1765, "step": 3659 }, { "epoch": 3.71, "learning_rate": 1.7616112364069972e-05, "loss": 0.1846, "step": 3660 }, { "epoch": 3.71, "learning_rate": 1.7614691582740756e-05, "loss": 0.1946, "step": 3661 }, { "epoch": 3.72, "learning_rate": 1.7613270435480656e-05, "loss": 0.2189, "step": 3662 }, { "epoch": 3.72, "learning_rate": 1.761184892235796e-05, "loss": 0.1738, "step": 3663 }, { "epoch": 3.72, "learning_rate": 1.7610427043440983e-05, "loss": 0.1609, "step": 3664 }, { "epoch": 3.72, "learning_rate": 1.7609004798798054e-05, "loss": 0.1774, "step": 3665 }, { "epoch": 3.72, "learning_rate": 1.760758218849752e-05, "loss": 0.2089, "step": 3666 }, { "epoch": 3.72, "learning_rate": 1.760615921260775e-05, "loss": 0.1867, "step": 3667 }, { "epoch": 3.72, "learning_rate": 1.7604735871197117e-05, "loss": 0.177, "step": 3668 }, { "epoch": 3.72, "learning_rate": 1.7603312164334032e-05, "loss": 0.1826, "step": 3669 }, { "epoch": 3.72, "learning_rate": 1.7601888092086903e-05, "loss": 0.1627, "step": 3670 }, { "epoch": 3.73, "learning_rate": 1.7600463654524175e-05, "loss": 0.1624, "step": 3671 }, { "epoch": 3.73, "learning_rate": 1.759903885171429e-05, "loss": 0.182, "step": 3672 }, { "epoch": 3.73, "learning_rate": 1.7597613683725725e-05, "loss": 0.1914, "step": 3673 }, { "epoch": 3.73, "learning_rate": 1.7596188150626964e-05, "loss": 0.1673, "step": 3674 }, { "epoch": 3.73, "learning_rate": 1.7594762252486517e-05, "loss": 0.1717, "step": 3675 }, { "epoch": 3.73, "learning_rate": 1.7593335989372903e-05, "loss": 0.1782, "step": 3676 }, { "epoch": 3.73, "learning_rate": 1.7591909361354665e-05, "loss": 0.209, "step": 3677 }, { "epoch": 3.73, "learning_rate": 1.7590482368500357e-05, "loss": 0.1863, "step": 3678 }, { "epoch": 3.73, "learning_rate": 1.7589055010878557e-05, "loss": 0.1713, "step": 3679 }, { "epoch": 3.73, "learning_rate": 1.758762728855786e-05, "loss": 0.1971, "step": 3680 }, { "epoch": 3.74, "learning_rate": 1.7586199201606874e-05, "loss": 0.1886, "step": 3681 }, { "epoch": 3.74, "learning_rate": 1.7584770750094224e-05, "loss": 0.1819, "step": 3682 }, { "epoch": 3.74, "learning_rate": 1.758334193408856e-05, "loss": 0.1873, "step": 3683 }, { "epoch": 3.74, "learning_rate": 1.7581912753658544e-05, "loss": 0.1581, "step": 3684 }, { "epoch": 3.74, "learning_rate": 1.758048320887286e-05, "loss": 0.1862, "step": 3685 }, { "epoch": 3.74, "learning_rate": 1.7579053299800194e-05, "loss": 0.1977, "step": 3686 }, { "epoch": 3.74, "learning_rate": 1.7577623026509277e-05, "loss": 0.1954, "step": 3687 }, { "epoch": 3.74, "learning_rate": 1.7576192389068834e-05, "loss": 0.1936, "step": 3688 }, { "epoch": 3.74, "learning_rate": 1.7574761387547616e-05, "loss": 0.1763, "step": 3689 }, { "epoch": 3.74, "learning_rate": 1.757333002201439e-05, "loss": 0.1765, "step": 3690 }, { "epoch": 3.75, "learning_rate": 1.7571898292537944e-05, "loss": 0.1566, "step": 3691 }, { "epoch": 3.75, "learning_rate": 1.7570466199187083e-05, "loss": 0.1726, "step": 3692 }, { "epoch": 3.75, "learning_rate": 1.756903374203062e-05, "loss": 0.1921, "step": 3693 }, { "epoch": 3.75, "learning_rate": 1.75676009211374e-05, "loss": 0.162, "step": 3694 }, { "epoch": 3.75, "learning_rate": 1.756616773657628e-05, "loss": 0.1908, "step": 3695 }, { "epoch": 3.75, "learning_rate": 1.7564734188416127e-05, "loss": 0.1771, "step": 3696 }, { "epoch": 3.75, "learning_rate": 1.756330027672583e-05, "loss": 0.1912, "step": 3697 }, { "epoch": 3.75, "learning_rate": 1.7561866001574305e-05, "loss": 0.1725, "step": 3698 }, { "epoch": 3.75, "learning_rate": 1.756043136303047e-05, "loss": 0.1777, "step": 3699 }, { "epoch": 3.75, "learning_rate": 1.7558996361163276e-05, "loss": 0.172, "step": 3700 }, { "epoch": 3.76, "learning_rate": 1.7557560996041676e-05, "loss": 0.1659, "step": 3701 }, { "epoch": 3.76, "learning_rate": 1.7556125267734648e-05, "loss": 0.1781, "step": 3702 }, { "epoch": 3.76, "learning_rate": 1.7554689176311192e-05, "loss": 0.1599, "step": 3703 }, { "epoch": 3.76, "learning_rate": 1.7553252721840314e-05, "loss": 0.1738, "step": 3704 }, { "epoch": 3.76, "learning_rate": 1.7551815904391052e-05, "loss": 0.1688, "step": 3705 }, { "epoch": 3.76, "learning_rate": 1.755037872403245e-05, "loss": 0.2018, "step": 3706 }, { "epoch": 3.76, "learning_rate": 1.754894118083357e-05, "loss": 0.1942, "step": 3707 }, { "epoch": 3.76, "learning_rate": 1.75475032748635e-05, "loss": 0.174, "step": 3708 }, { "epoch": 3.76, "learning_rate": 1.7546065006191332e-05, "loss": 0.1787, "step": 3709 }, { "epoch": 3.76, "learning_rate": 1.7544626374886194e-05, "loss": 0.2024, "step": 3710 }, { "epoch": 3.77, "learning_rate": 1.754318738101721e-05, "loss": 0.1951, "step": 3711 }, { "epoch": 3.77, "learning_rate": 1.754174802465354e-05, "loss": 0.1754, "step": 3712 }, { "epoch": 3.77, "learning_rate": 1.754030830586435e-05, "loss": 0.1725, "step": 3713 }, { "epoch": 3.77, "learning_rate": 1.753886822471883e-05, "loss": 0.1822, "step": 3714 }, { "epoch": 3.77, "learning_rate": 1.7537427781286182e-05, "loss": 0.1823, "step": 3715 }, { "epoch": 3.77, "learning_rate": 1.7535986975635623e-05, "loss": 0.168, "step": 3716 }, { "epoch": 3.77, "learning_rate": 1.7534545807836402e-05, "loss": 0.1814, "step": 3717 }, { "epoch": 3.77, "learning_rate": 1.7533104277957766e-05, "loss": 0.1753, "step": 3718 }, { "epoch": 3.77, "learning_rate": 1.7531662386068998e-05, "loss": 0.1839, "step": 3719 }, { "epoch": 3.77, "learning_rate": 1.7530220132239382e-05, "loss": 0.1869, "step": 3720 }, { "epoch": 3.78, "learning_rate": 1.7528777516538233e-05, "loss": 0.1909, "step": 3721 }, { "epoch": 3.78, "learning_rate": 1.752733453903487e-05, "loss": 0.1776, "step": 3722 }, { "epoch": 3.78, "learning_rate": 1.7525891199798646e-05, "loss": 0.1885, "step": 3723 }, { "epoch": 3.78, "learning_rate": 1.7524447498898912e-05, "loss": 0.1949, "step": 3724 }, { "epoch": 3.78, "learning_rate": 1.7523003436405055e-05, "loss": 0.1833, "step": 3725 }, { "epoch": 3.78, "learning_rate": 1.7521559012386467e-05, "loss": 0.2081, "step": 3726 }, { "epoch": 3.78, "learning_rate": 1.7520114226912558e-05, "loss": 0.1952, "step": 3727 }, { "epoch": 3.78, "learning_rate": 1.7518669080052763e-05, "loss": 0.1893, "step": 3728 }, { "epoch": 3.78, "learning_rate": 1.751722357187653e-05, "loss": 0.178, "step": 3729 }, { "epoch": 3.78, "learning_rate": 1.7515777702453317e-05, "loss": 0.1997, "step": 3730 }, { "epoch": 3.79, "learning_rate": 1.7514331471852616e-05, "loss": 0.1868, "step": 3731 }, { "epoch": 3.79, "learning_rate": 1.7512884880143922e-05, "loss": 0.1893, "step": 3732 }, { "epoch": 3.79, "learning_rate": 1.7511437927396754e-05, "loss": 0.1717, "step": 3733 }, { "epoch": 3.79, "learning_rate": 1.750999061368065e-05, "loss": 0.1701, "step": 3734 }, { "epoch": 3.79, "learning_rate": 1.750854293906515e-05, "loss": 0.1995, "step": 3735 }, { "epoch": 3.79, "learning_rate": 1.750709490361983e-05, "loss": 0.161, "step": 3736 }, { "epoch": 3.79, "learning_rate": 1.7505646507414287e-05, "loss": 0.1958, "step": 3737 }, { "epoch": 3.79, "learning_rate": 1.750419775051811e-05, "loss": 0.1877, "step": 3738 }, { "epoch": 3.79, "learning_rate": 1.7502748633000924e-05, "loss": 0.1764, "step": 3739 }, { "epoch": 3.8, "learning_rate": 1.7501299154932374e-05, "loss": 0.1742, "step": 3740 }, { "epoch": 3.8, "learning_rate": 1.7499849316382112e-05, "loss": 0.1838, "step": 3741 }, { "epoch": 3.8, "learning_rate": 1.749839911741981e-05, "loss": 0.1993, "step": 3742 }, { "epoch": 3.8, "learning_rate": 1.749694855811516e-05, "loss": 0.2028, "step": 3743 }, { "epoch": 3.8, "learning_rate": 1.7495497638537865e-05, "loss": 0.2118, "step": 3744 }, { "epoch": 3.8, "learning_rate": 1.749404635875766e-05, "loss": 0.1793, "step": 3745 }, { "epoch": 3.8, "learning_rate": 1.749259471884428e-05, "loss": 0.1974, "step": 3746 }, { "epoch": 3.8, "learning_rate": 1.7491142718867485e-05, "loss": 0.1729, "step": 3747 }, { "epoch": 3.8, "learning_rate": 1.7489690358897057e-05, "loss": 0.1751, "step": 3748 }, { "epoch": 3.8, "learning_rate": 1.748823763900279e-05, "loss": 0.1653, "step": 3749 }, { "epoch": 3.81, "learning_rate": 1.748678455925449e-05, "loss": 0.1649, "step": 3750 }, { "epoch": 3.81, "learning_rate": 1.7485331119721993e-05, "loss": 0.1956, "step": 3751 }, { "epoch": 3.81, "learning_rate": 1.748387732047514e-05, "loss": 0.1744, "step": 3752 }, { "epoch": 3.81, "learning_rate": 1.74824231615838e-05, "loss": 0.1788, "step": 3753 }, { "epoch": 3.81, "learning_rate": 1.7480968643117846e-05, "loss": 0.179, "step": 3754 }, { "epoch": 3.81, "learning_rate": 1.7479513765147182e-05, "loss": 0.1779, "step": 3755 }, { "epoch": 3.81, "learning_rate": 1.7478058527741723e-05, "loss": 0.1784, "step": 3756 }, { "epoch": 3.81, "learning_rate": 1.7476602930971402e-05, "loss": 0.1752, "step": 3757 }, { "epoch": 3.81, "learning_rate": 1.7475146974906168e-05, "loss": 0.2045, "step": 3758 }, { "epoch": 3.81, "learning_rate": 1.747369065961599e-05, "loss": 0.1772, "step": 3759 }, { "epoch": 3.82, "learning_rate": 1.7472233985170847e-05, "loss": 0.2024, "step": 3760 }, { "epoch": 3.82, "learning_rate": 1.7470776951640744e-05, "loss": 0.1632, "step": 3761 }, { "epoch": 3.82, "learning_rate": 1.7469319559095702e-05, "loss": 0.1544, "step": 3762 }, { "epoch": 3.82, "learning_rate": 1.7467861807605758e-05, "loss": 0.1728, "step": 3763 }, { "epoch": 3.82, "learning_rate": 1.7466403697240962e-05, "loss": 0.1783, "step": 3764 }, { "epoch": 3.82, "learning_rate": 1.746494522807139e-05, "loss": 0.1814, "step": 3765 }, { "epoch": 3.82, "learning_rate": 1.746348640016712e-05, "loss": 0.1908, "step": 3766 }, { "epoch": 3.82, "learning_rate": 1.746202721359827e-05, "loss": 0.1988, "step": 3767 }, { "epoch": 3.82, "learning_rate": 1.7460567668434952e-05, "loss": 0.1924, "step": 3768 }, { "epoch": 3.82, "learning_rate": 1.7459107764747314e-05, "loss": 0.1897, "step": 3769 }, { "epoch": 3.83, "learning_rate": 1.7457647502605508e-05, "loss": 0.1956, "step": 3770 }, { "epoch": 3.83, "learning_rate": 1.7456186882079707e-05, "loss": 0.1844, "step": 3771 }, { "epoch": 3.83, "learning_rate": 1.7454725903240105e-05, "loss": 0.1694, "step": 3772 }, { "epoch": 3.83, "learning_rate": 1.7453264566156913e-05, "loss": 0.1795, "step": 3773 }, { "epoch": 3.83, "learning_rate": 1.7451802870900354e-05, "loss": 0.1834, "step": 3774 }, { "epoch": 3.83, "learning_rate": 1.745034081754067e-05, "loss": 0.1691, "step": 3775 }, { "epoch": 3.83, "learning_rate": 1.7448878406148122e-05, "loss": 0.1925, "step": 3776 }, { "epoch": 3.83, "learning_rate": 1.744741563679299e-05, "loss": 0.191, "step": 3777 }, { "epoch": 3.83, "learning_rate": 1.7445952509545565e-05, "loss": 0.1983, "step": 3778 }, { "epoch": 3.83, "learning_rate": 1.744448902447616e-05, "loss": 0.1792, "step": 3779 }, { "epoch": 3.84, "learning_rate": 1.744302518165511e-05, "loss": 0.1752, "step": 3780 }, { "epoch": 3.84, "learning_rate": 1.744156098115275e-05, "loss": 0.1584, "step": 3781 }, { "epoch": 3.84, "learning_rate": 1.7440096423039453e-05, "loss": 0.1761, "step": 3782 }, { "epoch": 3.84, "learning_rate": 1.7438631507385598e-05, "loss": 0.1813, "step": 3783 }, { "epoch": 3.84, "learning_rate": 1.7437166234261575e-05, "loss": 0.1986, "step": 3784 }, { "epoch": 3.84, "learning_rate": 1.743570060373781e-05, "loss": 0.1801, "step": 3785 }, { "epoch": 3.84, "learning_rate": 1.743423461588473e-05, "loss": 0.1881, "step": 3786 }, { "epoch": 3.84, "learning_rate": 1.7432768270772787e-05, "loss": 0.1997, "step": 3787 }, { "epoch": 3.84, "learning_rate": 1.743130156847244e-05, "loss": 0.1845, "step": 3788 }, { "epoch": 3.84, "learning_rate": 1.742983450905418e-05, "loss": 0.1757, "step": 3789 }, { "epoch": 3.85, "learning_rate": 1.7428367092588508e-05, "loss": 0.1809, "step": 3790 }, { "epoch": 3.85, "learning_rate": 1.7426899319145936e-05, "loss": 0.1825, "step": 3791 }, { "epoch": 3.85, "learning_rate": 1.7425431188797006e-05, "loss": 0.1791, "step": 3792 }, { "epoch": 3.85, "learning_rate": 1.7423962701612268e-05, "loss": 0.1796, "step": 3793 }, { "epoch": 3.85, "learning_rate": 1.7422493857662288e-05, "loss": 0.1888, "step": 3794 }, { "epoch": 3.85, "learning_rate": 1.7421024657017657e-05, "loss": 0.1799, "step": 3795 }, { "epoch": 3.85, "learning_rate": 1.741955509974898e-05, "loss": 0.179, "step": 3796 }, { "epoch": 3.85, "learning_rate": 1.741808518592687e-05, "loss": 0.1661, "step": 3797 }, { "epoch": 3.85, "learning_rate": 1.7416614915621976e-05, "loss": 0.1828, "step": 3798 }, { "epoch": 3.85, "learning_rate": 1.7415144288904943e-05, "loss": 0.209, "step": 3799 }, { "epoch": 3.86, "learning_rate": 1.741367330584645e-05, "loss": 0.1859, "step": 3800 }, { "epoch": 3.86, "learning_rate": 1.741220196651718e-05, "loss": 0.1874, "step": 3801 }, { "epoch": 3.86, "learning_rate": 1.7410730270987847e-05, "loss": 0.1936, "step": 3802 }, { "epoch": 3.86, "learning_rate": 1.740925821932917e-05, "loss": 0.1815, "step": 3803 }, { "epoch": 3.86, "learning_rate": 1.740778581161189e-05, "loss": 0.2075, "step": 3804 }, { "epoch": 3.86, "learning_rate": 1.7406313047906773e-05, "loss": 0.1896, "step": 3805 }, { "epoch": 3.86, "learning_rate": 1.740483992828458e-05, "loss": 0.1952, "step": 3806 }, { "epoch": 3.86, "learning_rate": 1.740336645281611e-05, "loss": 0.1745, "step": 3807 }, { "epoch": 3.86, "learning_rate": 1.7401892621572175e-05, "loss": 0.1877, "step": 3808 }, { "epoch": 3.87, "learning_rate": 1.74004184346236e-05, "loss": 0.1832, "step": 3809 }, { "epoch": 3.87, "learning_rate": 1.7398943892041223e-05, "loss": 0.175, "step": 3810 }, { "epoch": 3.87, "learning_rate": 1.739746899389591e-05, "loss": 0.1861, "step": 3811 }, { "epoch": 3.87, "learning_rate": 1.7395993740258537e-05, "loss": 0.1566, "step": 3812 }, { "epoch": 3.87, "learning_rate": 1.7394518131199996e-05, "loss": 0.1983, "step": 3813 }, { "epoch": 3.87, "learning_rate": 1.739304216679121e-05, "loss": 0.1753, "step": 3814 }, { "epoch": 3.87, "learning_rate": 1.739156584710309e-05, "loss": 0.1829, "step": 3815 }, { "epoch": 3.87, "learning_rate": 1.7390089172206594e-05, "loss": 0.1804, "step": 3816 }, { "epoch": 3.87, "learning_rate": 1.7388612142172682e-05, "loss": 0.1901, "step": 3817 }, { "epoch": 3.87, "learning_rate": 1.7387134757072334e-05, "loss": 0.1763, "step": 3818 }, { "epoch": 3.88, "learning_rate": 1.7385657016976544e-05, "loss": 0.2089, "step": 3819 }, { "epoch": 3.88, "learning_rate": 1.7384178921956334e-05, "loss": 0.1965, "step": 3820 }, { "epoch": 3.88, "learning_rate": 1.738270047208273e-05, "loss": 0.1527, "step": 3821 }, { "epoch": 3.88, "learning_rate": 1.7381221667426776e-05, "loss": 0.2235, "step": 3822 }, { "epoch": 3.88, "learning_rate": 1.7379742508059548e-05, "loss": 0.1836, "step": 3823 }, { "epoch": 3.88, "learning_rate": 1.737826299405212e-05, "loss": 0.1611, "step": 3824 }, { "epoch": 3.88, "learning_rate": 1.737678312547559e-05, "loss": 0.1752, "step": 3825 }, { "epoch": 3.88, "learning_rate": 1.737530290240108e-05, "loss": 0.1917, "step": 3826 }, { "epoch": 3.88, "learning_rate": 1.7373822324899724e-05, "loss": 0.1953, "step": 3827 }, { "epoch": 3.88, "learning_rate": 1.737234139304267e-05, "loss": 0.1711, "step": 3828 }, { "epoch": 3.89, "learning_rate": 1.737086010690108e-05, "loss": 0.1868, "step": 3829 }, { "epoch": 3.89, "learning_rate": 1.736937846654615e-05, "loss": 0.191, "step": 3830 }, { "epoch": 3.89, "learning_rate": 1.7367896472049075e-05, "loss": 0.2142, "step": 3831 }, { "epoch": 3.89, "learning_rate": 1.7366414123481075e-05, "loss": 0.1972, "step": 3832 }, { "epoch": 3.89, "learning_rate": 1.736493142091338e-05, "loss": 0.2126, "step": 3833 }, { "epoch": 3.89, "learning_rate": 1.7363448364417252e-05, "loss": 0.1795, "step": 3834 }, { "epoch": 3.89, "learning_rate": 1.736196495406395e-05, "loss": 0.1977, "step": 3835 }, { "epoch": 3.89, "learning_rate": 1.7360481189924772e-05, "loss": 0.1875, "step": 3836 }, { "epoch": 3.89, "learning_rate": 1.7358997072071017e-05, "loss": 0.2133, "step": 3837 }, { "epoch": 3.89, "learning_rate": 1.7357512600574004e-05, "loss": 0.1828, "step": 3838 }, { "epoch": 3.9, "learning_rate": 1.7356027775505072e-05, "loss": 0.1755, "step": 3839 }, { "epoch": 3.9, "learning_rate": 1.7354542596935573e-05, "loss": 0.1802, "step": 3840 }, { "epoch": 3.9, "learning_rate": 1.735305706493688e-05, "loss": 0.1873, "step": 3841 }, { "epoch": 3.9, "learning_rate": 1.7351571179580386e-05, "loss": 0.1698, "step": 3842 }, { "epoch": 3.9, "learning_rate": 1.735008494093749e-05, "loss": 0.1911, "step": 3843 }, { "epoch": 3.9, "learning_rate": 1.7348598349079618e-05, "loss": 0.1765, "step": 3844 }, { "epoch": 3.9, "learning_rate": 1.7347111404078208e-05, "loss": 0.1825, "step": 3845 }, { "epoch": 3.9, "learning_rate": 1.734562410600472e-05, "loss": 0.1835, "step": 3846 }, { "epoch": 3.9, "learning_rate": 1.7344136454930626e-05, "loss": 0.1845, "step": 3847 }, { "epoch": 3.9, "learning_rate": 1.7342648450927412e-05, "loss": 0.1922, "step": 3848 }, { "epoch": 3.91, "learning_rate": 1.734116009406659e-05, "loss": 0.206, "step": 3849 }, { "epoch": 3.91, "learning_rate": 1.7339671384419686e-05, "loss": 0.1978, "step": 3850 }, { "epoch": 3.91, "learning_rate": 1.7338182322058234e-05, "loss": 0.1914, "step": 3851 }, { "epoch": 3.91, "learning_rate": 1.7336692907053798e-05, "loss": 0.1851, "step": 3852 }, { "epoch": 3.91, "learning_rate": 1.733520313947795e-05, "loss": 0.1896, "step": 3853 }, { "epoch": 3.91, "learning_rate": 1.7333713019402284e-05, "loss": 0.1796, "step": 3854 }, { "epoch": 3.91, "learning_rate": 1.7332222546898413e-05, "loss": 0.1836, "step": 3855 }, { "epoch": 3.91, "learning_rate": 1.7330731722037953e-05, "loss": 0.1933, "step": 3856 }, { "epoch": 3.91, "learning_rate": 1.732924054489256e-05, "loss": 0.1738, "step": 3857 }, { "epoch": 3.91, "learning_rate": 1.732774901553388e-05, "loss": 0.1849, "step": 3858 }, { "epoch": 3.92, "learning_rate": 1.7326257134033602e-05, "loss": 0.1819, "step": 3859 }, { "epoch": 3.92, "learning_rate": 1.732476490046341e-05, "loss": 0.1885, "step": 3860 }, { "epoch": 3.92, "learning_rate": 1.7323272314895022e-05, "loss": 0.2146, "step": 3861 }, { "epoch": 3.92, "learning_rate": 1.732177937740016e-05, "loss": 0.1846, "step": 3862 }, { "epoch": 3.92, "learning_rate": 1.732028608805057e-05, "loss": 0.1916, "step": 3863 }, { "epoch": 3.92, "learning_rate": 1.7318792446918018e-05, "loss": 0.2063, "step": 3864 }, { "epoch": 3.92, "learning_rate": 1.7317298454074277e-05, "loss": 0.1923, "step": 3865 }, { "epoch": 3.92, "learning_rate": 1.7315804109591144e-05, "loss": 0.167, "step": 3866 }, { "epoch": 3.92, "learning_rate": 1.731430941354043e-05, "loss": 0.1908, "step": 3867 }, { "epoch": 3.92, "learning_rate": 1.7312814365993968e-05, "loss": 0.1756, "step": 3868 }, { "epoch": 3.93, "learning_rate": 1.7311318967023595e-05, "loss": 0.1752, "step": 3869 }, { "epoch": 3.93, "learning_rate": 1.7309823216701183e-05, "loss": 0.2111, "step": 3870 }, { "epoch": 3.93, "learning_rate": 1.730832711509861e-05, "loss": 0.1861, "step": 3871 }, { "epoch": 3.93, "learning_rate": 1.7306830662287767e-05, "loss": 0.1864, "step": 3872 }, { "epoch": 3.93, "learning_rate": 1.730533385834057e-05, "loss": 0.1959, "step": 3873 }, { "epoch": 3.93, "learning_rate": 1.730383670332895e-05, "loss": 0.1905, "step": 3874 }, { "epoch": 3.93, "learning_rate": 1.730233919732486e-05, "loss": 0.2066, "step": 3875 }, { "epoch": 3.93, "learning_rate": 1.7300841340400252e-05, "loss": 0.186, "step": 3876 }, { "epoch": 3.93, "learning_rate": 1.7299343132627114e-05, "loss": 0.1738, "step": 3877 }, { "epoch": 3.94, "learning_rate": 1.7297844574077447e-05, "loss": 0.172, "step": 3878 }, { "epoch": 3.94, "learning_rate": 1.729634566482326e-05, "loss": 0.1866, "step": 3879 }, { "epoch": 3.94, "learning_rate": 1.7294846404936584e-05, "loss": 0.1999, "step": 3880 }, { "epoch": 3.94, "learning_rate": 1.7293346794489473e-05, "loss": 0.1995, "step": 3881 }, { "epoch": 3.94, "learning_rate": 1.7291846833553982e-05, "loss": 0.1658, "step": 3882 }, { "epoch": 3.94, "learning_rate": 1.7290346522202204e-05, "loss": 0.1924, "step": 3883 }, { "epoch": 3.94, "learning_rate": 1.7288845860506232e-05, "loss": 0.1737, "step": 3884 }, { "epoch": 3.94, "learning_rate": 1.7287344848538182e-05, "loss": 0.1785, "step": 3885 }, { "epoch": 3.94, "learning_rate": 1.728584348637019e-05, "loss": 0.2016, "step": 3886 }, { "epoch": 3.94, "learning_rate": 1.72843417740744e-05, "loss": 0.1868, "step": 3887 }, { "epoch": 3.95, "learning_rate": 1.728283971172298e-05, "loss": 0.1937, "step": 3888 }, { "epoch": 3.95, "learning_rate": 1.728133729938812e-05, "loss": 0.2037, "step": 3889 }, { "epoch": 3.95, "learning_rate": 1.7279834537142006e-05, "loss": 0.1894, "step": 3890 }, { "epoch": 3.95, "learning_rate": 1.7278331425056866e-05, "loss": 0.1819, "step": 3891 }, { "epoch": 3.95, "learning_rate": 1.727682796320493e-05, "loss": 0.1822, "step": 3892 }, { "epoch": 3.95, "learning_rate": 1.7275324151658447e-05, "loss": 0.1762, "step": 3893 }, { "epoch": 3.95, "learning_rate": 1.7273819990489685e-05, "loss": 0.1926, "step": 3894 }, { "epoch": 3.95, "learning_rate": 1.7272315479770925e-05, "loss": 0.1999, "step": 3895 }, { "epoch": 3.95, "learning_rate": 1.727081061957447e-05, "loss": 0.2012, "step": 3896 }, { "epoch": 3.95, "learning_rate": 1.726930540997264e-05, "loss": 0.1898, "step": 3897 }, { "epoch": 3.96, "learning_rate": 1.726779985103777e-05, "loss": 0.1708, "step": 3898 }, { "epoch": 3.96, "learning_rate": 1.72662939428422e-05, "loss": 0.1928, "step": 3899 }, { "epoch": 3.96, "learning_rate": 1.726478768545831e-05, "loss": 0.2068, "step": 3900 }, { "epoch": 3.96, "learning_rate": 1.7263281078958483e-05, "loss": 0.1807, "step": 3901 }, { "epoch": 3.96, "learning_rate": 1.7261774123415113e-05, "loss": 0.1928, "step": 3902 }, { "epoch": 3.96, "learning_rate": 1.7260266818900623e-05, "loss": 0.1617, "step": 3903 }, { "epoch": 3.96, "learning_rate": 1.7258759165487452e-05, "loss": 0.1899, "step": 3904 }, { "epoch": 3.96, "learning_rate": 1.7257251163248044e-05, "loss": 0.1842, "step": 3905 }, { "epoch": 3.96, "learning_rate": 1.725574281225487e-05, "loss": 0.1918, "step": 3906 }, { "epoch": 3.96, "learning_rate": 1.7254234112580417e-05, "loss": 0.185, "step": 3907 }, { "epoch": 3.97, "learning_rate": 1.7252725064297183e-05, "loss": 0.1784, "step": 3908 }, { "epoch": 3.97, "learning_rate": 1.7251215667477693e-05, "loss": 0.1825, "step": 3909 }, { "epoch": 3.97, "learning_rate": 1.724970592219448e-05, "loss": 0.1672, "step": 3910 }, { "epoch": 3.97, "learning_rate": 1.724819582852009e-05, "loss": 0.1734, "step": 3911 }, { "epoch": 3.97, "learning_rate": 1.7246685386527098e-05, "loss": 0.1709, "step": 3912 }, { "epoch": 3.97, "learning_rate": 1.724517459628809e-05, "loss": 0.1903, "step": 3913 }, { "epoch": 3.97, "learning_rate": 1.7243663457875665e-05, "loss": 0.1565, "step": 3914 }, { "epoch": 3.97, "learning_rate": 1.7242151971362444e-05, "loss": 0.1818, "step": 3915 }, { "epoch": 3.97, "learning_rate": 1.7240640136821064e-05, "loss": 0.1961, "step": 3916 }, { "epoch": 3.97, "learning_rate": 1.7239127954324178e-05, "loss": 0.1894, "step": 3917 }, { "epoch": 3.98, "learning_rate": 1.7237615423944452e-05, "loss": 0.1725, "step": 3918 }, { "epoch": 3.98, "learning_rate": 1.7236102545754574e-05, "loss": 0.199, "step": 3919 }, { "epoch": 3.98, "learning_rate": 1.723458931982725e-05, "loss": 0.1633, "step": 3920 }, { "epoch": 3.98, "learning_rate": 1.723307574623519e-05, "loss": 0.1919, "step": 3921 }, { "epoch": 3.98, "learning_rate": 1.7231561825051137e-05, "loss": 0.1757, "step": 3922 }, { "epoch": 3.98, "learning_rate": 1.7230047556347846e-05, "loss": 0.2288, "step": 3923 }, { "epoch": 3.98, "learning_rate": 1.7228532940198085e-05, "loss": 0.1923, "step": 3924 }, { "epoch": 3.98, "learning_rate": 1.7227017976674632e-05, "loss": 0.1833, "step": 3925 }, { "epoch": 3.98, "learning_rate": 1.72255026658503e-05, "loss": 0.1914, "step": 3926 }, { "epoch": 3.98, "learning_rate": 1.7223987007797908e-05, "loss": 0.1832, "step": 3927 }, { "epoch": 3.99, "learning_rate": 1.7222471002590285e-05, "loss": 0.1696, "step": 3928 }, { "epoch": 3.99, "learning_rate": 1.7220954650300293e-05, "loss": 0.1708, "step": 3929 }, { "epoch": 3.99, "learning_rate": 1.7219437951000798e-05, "loss": 0.1982, "step": 3930 }, { "epoch": 3.99, "learning_rate": 1.721792090476468e-05, "loss": 0.1899, "step": 3931 }, { "epoch": 3.99, "learning_rate": 1.7216403511664853e-05, "loss": 0.1829, "step": 3932 }, { "epoch": 3.99, "learning_rate": 1.7214885771774228e-05, "loss": 0.1657, "step": 3933 }, { "epoch": 3.99, "learning_rate": 1.7213367685165747e-05, "loss": 0.1896, "step": 3934 }, { "epoch": 3.99, "learning_rate": 1.721184925191236e-05, "loss": 0.1617, "step": 3935 }, { "epoch": 3.99, "learning_rate": 1.7210330472087036e-05, "loss": 0.1844, "step": 3936 }, { "epoch": 3.99, "learning_rate": 1.7208811345762766e-05, "loss": 0.1887, "step": 3937 }, { "epoch": 4.0, "learning_rate": 1.7207291873012545e-05, "loss": 0.1807, "step": 3938 }, { "epoch": 4.0, "learning_rate": 1.7205772053909402e-05, "loss": 0.184, "step": 3939 }, { "epoch": 4.0, "learning_rate": 1.7204251888526367e-05, "loss": 0.1751, "step": 3940 }, { "epoch": 4.0, "learning_rate": 1.7202731376936494e-05, "loss": 0.1776, "step": 3941 }, { "epoch": 4.0, "learning_rate": 1.7201210519212855e-05, "loss": 0.1747, "step": 3942 }, { "epoch": 4.0, "learning_rate": 1.7199689315428534e-05, "loss": 0.0972, "step": 3943 }, { "epoch": 4.0, "learning_rate": 1.7198167765656632e-05, "loss": 0.1051, "step": 3944 }, { "epoch": 4.0, "learning_rate": 1.719664586997027e-05, "loss": 0.1054, "step": 3945 }, { "epoch": 4.0, "learning_rate": 1.719512362844259e-05, "loss": 0.1142, "step": 3946 }, { "epoch": 4.01, "learning_rate": 1.7193601041146733e-05, "loss": 0.1028, "step": 3947 }, { "epoch": 4.01, "learning_rate": 1.719207810815588e-05, "loss": 0.1025, "step": 3948 }, { "epoch": 4.01, "learning_rate": 1.719055482954321e-05, "loss": 0.1068, "step": 3949 }, { "epoch": 4.01, "learning_rate": 1.7189031205381924e-05, "loss": 0.1078, "step": 3950 }, { "epoch": 4.01, "learning_rate": 1.7187507235745248e-05, "loss": 0.1276, "step": 3951 }, { "epoch": 4.01, "learning_rate": 1.7185982920706415e-05, "loss": 0.0994, "step": 3952 }, { "epoch": 4.01, "learning_rate": 1.7184458260338676e-05, "loss": 0.1287, "step": 3953 }, { "epoch": 4.01, "learning_rate": 1.7182933254715297e-05, "loss": 0.1093, "step": 3954 }, { "epoch": 4.01, "learning_rate": 1.7181407903909567e-05, "loss": 0.1163, "step": 3955 }, { "epoch": 4.01, "learning_rate": 1.717988220799479e-05, "loss": 0.0973, "step": 3956 }, { "epoch": 4.02, "learning_rate": 1.7178356167044283e-05, "loss": 0.0955, "step": 3957 }, { "epoch": 4.02, "learning_rate": 1.7176829781131382e-05, "loss": 0.1146, "step": 3958 }, { "epoch": 4.02, "learning_rate": 1.7175303050329434e-05, "loss": 0.1123, "step": 3959 }, { "epoch": 4.02, "learning_rate": 1.7173775974711814e-05, "loss": 0.1114, "step": 3960 }, { "epoch": 4.02, "learning_rate": 1.7172248554351903e-05, "loss": 0.1233, "step": 3961 }, { "epoch": 4.02, "learning_rate": 1.7170720789323105e-05, "loss": 0.1029, "step": 3962 }, { "epoch": 4.02, "learning_rate": 1.7169192679698837e-05, "loss": 0.0898, "step": 3963 }, { "epoch": 4.02, "learning_rate": 1.716766422555253e-05, "loss": 0.1065, "step": 3964 }, { "epoch": 4.02, "learning_rate": 1.7166135426957644e-05, "loss": 0.1255, "step": 3965 }, { "epoch": 4.02, "learning_rate": 1.7164606283987638e-05, "loss": 0.1133, "step": 3966 }, { "epoch": 4.03, "learning_rate": 1.7163076796716e-05, "loss": 0.1277, "step": 3967 }, { "epoch": 4.03, "learning_rate": 1.7161546965216236e-05, "loss": 0.1236, "step": 3968 }, { "epoch": 4.03, "learning_rate": 1.7160016789561852e-05, "loss": 0.0959, "step": 3969 }, { "epoch": 4.03, "learning_rate": 1.715848626982639e-05, "loss": 0.1323, "step": 3970 }, { "epoch": 4.03, "learning_rate": 1.7156955406083403e-05, "loss": 0.114, "step": 3971 }, { "epoch": 4.03, "learning_rate": 1.7155424198406447e-05, "loss": 0.1072, "step": 3972 }, { "epoch": 4.03, "learning_rate": 1.7153892646869118e-05, "loss": 0.0992, "step": 3973 }, { "epoch": 4.03, "learning_rate": 1.7152360751545008e-05, "loss": 0.1057, "step": 3974 }, { "epoch": 4.03, "learning_rate": 1.715082851250774e-05, "loss": 0.1103, "step": 3975 }, { "epoch": 4.03, "learning_rate": 1.7149295929830934e-05, "loss": 0.1091, "step": 3976 }, { "epoch": 4.04, "learning_rate": 1.7147763003588256e-05, "loss": 0.1226, "step": 3977 }, { "epoch": 4.04, "learning_rate": 1.7146229733853365e-05, "loss": 0.1071, "step": 3978 }, { "epoch": 4.04, "learning_rate": 1.7144696120699943e-05, "loss": 0.1091, "step": 3979 }, { "epoch": 4.04, "learning_rate": 1.714316216420169e-05, "loss": 0.1079, "step": 3980 }, { "epoch": 4.04, "learning_rate": 1.714162786443232e-05, "loss": 0.1074, "step": 3981 }, { "epoch": 4.04, "learning_rate": 1.7140093221465566e-05, "loss": 0.1022, "step": 3982 }, { "epoch": 4.04, "learning_rate": 1.713855823537518e-05, "loss": 0.113, "step": 3983 }, { "epoch": 4.04, "learning_rate": 1.7137022906234925e-05, "loss": 0.0941, "step": 3984 }, { "epoch": 4.04, "learning_rate": 1.713548723411858e-05, "loss": 0.1049, "step": 3985 }, { "epoch": 4.04, "learning_rate": 1.7133951219099944e-05, "loss": 0.1051, "step": 3986 }, { "epoch": 4.05, "learning_rate": 1.713241486125284e-05, "loss": 0.108, "step": 3987 }, { "epoch": 4.05, "learning_rate": 1.7130878160651083e-05, "loss": 0.107, "step": 3988 }, { "epoch": 4.05, "learning_rate": 1.7129341117368532e-05, "loss": 0.1021, "step": 3989 }, { "epoch": 4.05, "learning_rate": 1.7127803731479053e-05, "loss": 0.093, "step": 3990 }, { "epoch": 4.05, "learning_rate": 1.7126266003056514e-05, "loss": 0.0983, "step": 3991 }, { "epoch": 4.05, "learning_rate": 1.7124727932174825e-05, "loss": 0.1148, "step": 3992 }, { "epoch": 4.05, "learning_rate": 1.7123189518907893e-05, "loss": 0.0864, "step": 3993 }, { "epoch": 4.05, "learning_rate": 1.7121650763329652e-05, "loss": 0.089, "step": 3994 }, { "epoch": 4.05, "learning_rate": 1.7120111665514043e-05, "loss": 0.1107, "step": 3995 }, { "epoch": 4.05, "learning_rate": 1.711857222553503e-05, "loss": 0.1141, "step": 3996 }, { "epoch": 4.06, "learning_rate": 1.7117032443466593e-05, "loss": 0.1147, "step": 3997 }, { "epoch": 4.06, "learning_rate": 1.711549231938273e-05, "loss": 0.0957, "step": 3998 }, { "epoch": 4.06, "learning_rate": 1.7113951853357453e-05, "loss": 0.0995, "step": 3999 }, { "epoch": 4.06, "learning_rate": 1.7112411045464784e-05, "loss": 0.1051, "step": 4000 }, { "epoch": 4.06, "learning_rate": 1.7110869895778773e-05, "loss": 0.1094, "step": 4001 }, { "epoch": 4.06, "learning_rate": 1.7109328404373484e-05, "loss": 0.0975, "step": 4002 }, { "epoch": 4.06, "learning_rate": 1.710778657132299e-05, "loss": 0.1126, "step": 4003 }, { "epoch": 4.06, "learning_rate": 1.7106244396701385e-05, "loss": 0.1063, "step": 4004 }, { "epoch": 4.06, "learning_rate": 1.7104701880582782e-05, "loss": 0.113, "step": 4005 }, { "epoch": 4.06, "learning_rate": 1.7103159023041306e-05, "loss": 0.1166, "step": 4006 }, { "epoch": 4.07, "learning_rate": 1.7101615824151104e-05, "loss": 0.0945, "step": 4007 }, { "epoch": 4.07, "learning_rate": 1.7100072283986335e-05, "loss": 0.0987, "step": 4008 }, { "epoch": 4.07, "learning_rate": 1.709852840262117e-05, "loss": 0.1039, "step": 4009 }, { "epoch": 4.07, "learning_rate": 1.709698418012981e-05, "loss": 0.1075, "step": 4010 }, { "epoch": 4.07, "learning_rate": 1.7095439616586452e-05, "loss": 0.1213, "step": 4011 }, { "epoch": 4.07, "learning_rate": 1.7093894712065334e-05, "loss": 0.098, "step": 4012 }, { "epoch": 4.07, "learning_rate": 1.7092349466640693e-05, "loss": 0.1034, "step": 4013 }, { "epoch": 4.07, "learning_rate": 1.7090803880386784e-05, "loss": 0.0937, "step": 4014 }, { "epoch": 4.07, "learning_rate": 1.7089257953377885e-05, "loss": 0.1093, "step": 4015 }, { "epoch": 4.08, "learning_rate": 1.7087711685688284e-05, "loss": 0.101, "step": 4016 }, { "epoch": 4.08, "learning_rate": 1.7086165077392296e-05, "loss": 0.0966, "step": 4017 }, { "epoch": 4.08, "learning_rate": 1.7084618128564234e-05, "loss": 0.1118, "step": 4018 }, { "epoch": 4.08, "learning_rate": 1.7083070839278446e-05, "loss": 0.1058, "step": 4019 }, { "epoch": 4.08, "learning_rate": 1.7081523209609282e-05, "loss": 0.1289, "step": 4020 }, { "epoch": 4.08, "learning_rate": 1.7079975239631124e-05, "loss": 0.0982, "step": 4021 }, { "epoch": 4.08, "learning_rate": 1.707842692941835e-05, "loss": 0.1167, "step": 4022 }, { "epoch": 4.08, "learning_rate": 1.7076878279045374e-05, "loss": 0.0933, "step": 4023 }, { "epoch": 4.08, "learning_rate": 1.7075329288586614e-05, "loss": 0.1061, "step": 4024 }, { "epoch": 4.08, "learning_rate": 1.707377995811651e-05, "loss": 0.1215, "step": 4025 }, { "epoch": 4.09, "learning_rate": 1.707223028770951e-05, "loss": 0.1087, "step": 4026 }, { "epoch": 4.09, "learning_rate": 1.7070680277440093e-05, "loss": 0.1081, "step": 4027 }, { "epoch": 4.09, "learning_rate": 1.7069129927382745e-05, "loss": 0.1157, "step": 4028 }, { "epoch": 4.09, "learning_rate": 1.7067579237611968e-05, "loss": 0.1403, "step": 4029 }, { "epoch": 4.09, "learning_rate": 1.7066028208202276e-05, "loss": 0.1042, "step": 4030 }, { "epoch": 4.09, "learning_rate": 1.7064476839228213e-05, "loss": 0.1242, "step": 4031 }, { "epoch": 4.09, "learning_rate": 1.7062925130764334e-05, "loss": 0.1134, "step": 4032 }, { "epoch": 4.09, "learning_rate": 1.7061373082885198e-05, "loss": 0.1009, "step": 4033 }, { "epoch": 4.09, "learning_rate": 1.7059820695665395e-05, "loss": 0.1259, "step": 4034 }, { "epoch": 4.09, "learning_rate": 1.705826796917953e-05, "loss": 0.0999, "step": 4035 }, { "epoch": 4.1, "learning_rate": 1.705671490350221e-05, "loss": 0.12, "step": 4036 }, { "epoch": 4.1, "learning_rate": 1.705516149870808e-05, "loss": 0.1116, "step": 4037 }, { "epoch": 4.1, "learning_rate": 1.7053607754871783e-05, "loss": 0.111, "step": 4038 }, { "epoch": 4.1, "learning_rate": 1.7052053672067992e-05, "loss": 0.127, "step": 4039 }, { "epoch": 4.1, "learning_rate": 1.705049925037139e-05, "loss": 0.1048, "step": 4040 }, { "epoch": 4.1, "learning_rate": 1.7048944489856666e-05, "loss": 0.1013, "step": 4041 }, { "epoch": 4.1, "learning_rate": 1.7047389390598543e-05, "loss": 0.0998, "step": 4042 }, { "epoch": 4.1, "learning_rate": 1.7045833952671755e-05, "loss": 0.1095, "step": 4043 }, { "epoch": 4.1, "learning_rate": 1.7044278176151044e-05, "loss": 0.1106, "step": 4044 }, { "epoch": 4.1, "learning_rate": 1.704272206111118e-05, "loss": 0.1169, "step": 4045 }, { "epoch": 4.11, "learning_rate": 1.7041165607626938e-05, "loss": 0.1098, "step": 4046 }, { "epoch": 4.11, "learning_rate": 1.7039608815773116e-05, "loss": 0.1045, "step": 4047 }, { "epoch": 4.11, "learning_rate": 1.703805168562453e-05, "loss": 0.1401, "step": 4048 }, { "epoch": 4.11, "learning_rate": 1.7036494217256008e-05, "loss": 0.0925, "step": 4049 }, { "epoch": 4.11, "learning_rate": 1.7034936410742394e-05, "loss": 0.1265, "step": 4050 }, { "epoch": 4.11, "learning_rate": 1.7033378266158554e-05, "loss": 0.1233, "step": 4051 }, { "epoch": 4.11, "learning_rate": 1.7031819783579358e-05, "loss": 0.1117, "step": 4052 }, { "epoch": 4.11, "learning_rate": 1.703026096307971e-05, "loss": 0.1181, "step": 4053 }, { "epoch": 4.11, "learning_rate": 1.7028701804734515e-05, "loss": 0.1011, "step": 4054 }, { "epoch": 4.11, "learning_rate": 1.70271423086187e-05, "loss": 0.1217, "step": 4055 }, { "epoch": 4.12, "learning_rate": 1.7025582474807208e-05, "loss": 0.1131, "step": 4056 }, { "epoch": 4.12, "learning_rate": 1.7024022303375e-05, "loss": 0.1085, "step": 4057 }, { "epoch": 4.12, "learning_rate": 1.7022461794397048e-05, "loss": 0.1055, "step": 4058 }, { "epoch": 4.12, "learning_rate": 1.702090094794835e-05, "loss": 0.1173, "step": 4059 }, { "epoch": 4.12, "learning_rate": 1.7019339764103903e-05, "loss": 0.1135, "step": 4060 }, { "epoch": 4.12, "learning_rate": 1.7017778242938744e-05, "loss": 0.1288, "step": 4061 }, { "epoch": 4.12, "learning_rate": 1.7016216384527905e-05, "loss": 0.117, "step": 4062 }, { "epoch": 4.12, "learning_rate": 1.7014654188946446e-05, "loss": 0.1159, "step": 4063 }, { "epoch": 4.12, "learning_rate": 1.701309165626944e-05, "loss": 0.0935, "step": 4064 }, { "epoch": 4.12, "learning_rate": 1.701152878657197e-05, "loss": 0.0931, "step": 4065 }, { "epoch": 4.13, "learning_rate": 1.7009965579929146e-05, "loss": 0.1186, "step": 4066 }, { "epoch": 4.13, "learning_rate": 1.700840203641609e-05, "loss": 0.0975, "step": 4067 }, { "epoch": 4.13, "learning_rate": 1.700683815610794e-05, "loss": 0.1088, "step": 4068 }, { "epoch": 4.13, "learning_rate": 1.7005273939079846e-05, "loss": 0.0831, "step": 4069 }, { "epoch": 4.13, "learning_rate": 1.700370938540698e-05, "loss": 0.1066, "step": 4070 }, { "epoch": 4.13, "learning_rate": 1.700214449516453e-05, "loss": 0.097, "step": 4071 }, { "epoch": 4.13, "learning_rate": 1.7000579268427696e-05, "loss": 0.1158, "step": 4072 }, { "epoch": 4.13, "learning_rate": 1.699901370527169e-05, "loss": 0.1256, "step": 4073 }, { "epoch": 4.13, "learning_rate": 1.6997447805771763e-05, "loss": 0.0984, "step": 4074 }, { "epoch": 4.13, "learning_rate": 1.699588157000315e-05, "loss": 0.1209, "step": 4075 }, { "epoch": 4.14, "learning_rate": 1.6994314998041126e-05, "loss": 0.1028, "step": 4076 }, { "epoch": 4.14, "learning_rate": 1.699274808996097e-05, "loss": 0.1171, "step": 4077 }, { "epoch": 4.14, "learning_rate": 1.6991180845837983e-05, "loss": 0.1148, "step": 4078 }, { "epoch": 4.14, "learning_rate": 1.698961326574748e-05, "loss": 0.1164, "step": 4079 }, { "epoch": 4.14, "learning_rate": 1.6988045349764794e-05, "loss": 0.1114, "step": 4080 }, { "epoch": 4.14, "learning_rate": 1.6986477097965267e-05, "loss": 0.1242, "step": 4081 }, { "epoch": 4.14, "learning_rate": 1.6984908510424272e-05, "loss": 0.1101, "step": 4082 }, { "epoch": 4.14, "learning_rate": 1.6983339587217182e-05, "loss": 0.1103, "step": 4083 }, { "epoch": 4.14, "learning_rate": 1.69817703284194e-05, "loss": 0.1082, "step": 4084 }, { "epoch": 4.15, "learning_rate": 1.6980200734106325e-05, "loss": 0.1291, "step": 4085 }, { "epoch": 4.15, "learning_rate": 1.6978630804353396e-05, "loss": 0.1407, "step": 4086 }, { "epoch": 4.15, "learning_rate": 1.6977060539236056e-05, "loss": 0.1062, "step": 4087 }, { "epoch": 4.15, "learning_rate": 1.6975489938829764e-05, "loss": 0.1052, "step": 4088 }, { "epoch": 4.15, "learning_rate": 1.6973919003209996e-05, "loss": 0.1066, "step": 4089 }, { "epoch": 4.15, "learning_rate": 1.6972347732452248e-05, "loss": 0.092, "step": 4090 }, { "epoch": 4.15, "learning_rate": 1.6970776126632023e-05, "loss": 0.1179, "step": 4091 }, { "epoch": 4.15, "learning_rate": 1.6969204185824854e-05, "loss": 0.1201, "step": 4092 }, { "epoch": 4.15, "learning_rate": 1.6967631910106278e-05, "loss": 0.1085, "step": 4093 }, { "epoch": 4.15, "learning_rate": 1.6966059299551844e-05, "loss": 0.1165, "step": 4094 }, { "epoch": 4.16, "learning_rate": 1.696448635423714e-05, "loss": 0.1128, "step": 4095 }, { "epoch": 4.16, "learning_rate": 1.6962913074237747e-05, "loss": 0.114, "step": 4096 }, { "epoch": 4.16, "learning_rate": 1.696133945962927e-05, "loss": 0.1092, "step": 4097 }, { "epoch": 4.16, "learning_rate": 1.695976551048733e-05, "loss": 0.1222, "step": 4098 }, { "epoch": 4.16, "learning_rate": 1.695819122688757e-05, "loss": 0.1109, "step": 4099 }, { "epoch": 4.16, "learning_rate": 1.6956616608905636e-05, "loss": 0.1248, "step": 4100 }, { "epoch": 4.16, "learning_rate": 1.6955041656617205e-05, "loss": 0.1057, "step": 4101 }, { "epoch": 4.16, "learning_rate": 1.695346637009796e-05, "loss": 0.1247, "step": 4102 }, { "epoch": 4.16, "learning_rate": 1.69518907494236e-05, "loss": 0.1233, "step": 4103 }, { "epoch": 4.16, "learning_rate": 1.6950314794669853e-05, "loss": 0.1077, "step": 4104 }, { "epoch": 4.17, "learning_rate": 1.6948738505912437e-05, "loss": 0.1177, "step": 4105 }, { "epoch": 4.17, "learning_rate": 1.694716188322711e-05, "loss": 0.1307, "step": 4106 }, { "epoch": 4.17, "learning_rate": 1.6945584926689642e-05, "loss": 0.1128, "step": 4107 }, { "epoch": 4.17, "learning_rate": 1.694400763637581e-05, "loss": 0.1142, "step": 4108 }, { "epoch": 4.17, "learning_rate": 1.6942430012361414e-05, "loss": 0.1108, "step": 4109 }, { "epoch": 4.17, "learning_rate": 1.6940852054722266e-05, "loss": 0.1227, "step": 4110 }, { "epoch": 4.17, "learning_rate": 1.69392737635342e-05, "loss": 0.1224, "step": 4111 }, { "epoch": 4.17, "learning_rate": 1.693769513887306e-05, "loss": 0.1194, "step": 4112 }, { "epoch": 4.17, "learning_rate": 1.6936116180814706e-05, "loss": 0.1189, "step": 4113 }, { "epoch": 4.17, "learning_rate": 1.6934536889435017e-05, "loss": 0.118, "step": 4114 }, { "epoch": 4.18, "learning_rate": 1.6932957264809893e-05, "loss": 0.1311, "step": 4115 }, { "epoch": 4.18, "learning_rate": 1.693137730701524e-05, "loss": 0.1216, "step": 4116 }, { "epoch": 4.18, "learning_rate": 1.6929797016126982e-05, "loss": 0.1162, "step": 4117 }, { "epoch": 4.18, "learning_rate": 1.6928216392221065e-05, "loss": 0.1327, "step": 4118 }, { "epoch": 4.18, "learning_rate": 1.6926635435373445e-05, "loss": 0.0965, "step": 4119 }, { "epoch": 4.18, "learning_rate": 1.6925054145660094e-05, "loss": 0.1059, "step": 4120 }, { "epoch": 4.18, "learning_rate": 1.6923472523157013e-05, "loss": 0.1172, "step": 4121 }, { "epoch": 4.18, "learning_rate": 1.6921890567940196e-05, "loss": 0.1184, "step": 4122 }, { "epoch": 4.18, "learning_rate": 1.692030828008567e-05, "loss": 0.0932, "step": 4123 }, { "epoch": 4.18, "learning_rate": 1.6918725659669476e-05, "loss": 0.1095, "step": 4124 }, { "epoch": 4.19, "learning_rate": 1.691714270676766e-05, "loss": 0.1292, "step": 4125 }, { "epoch": 4.19, "learning_rate": 1.6915559421456305e-05, "loss": 0.0877, "step": 4126 }, { "epoch": 4.19, "learning_rate": 1.6913975803811488e-05, "loss": 0.0954, "step": 4127 }, { "epoch": 4.19, "learning_rate": 1.6912391853909313e-05, "loss": 0.1013, "step": 4128 }, { "epoch": 4.19, "learning_rate": 1.6910807571825895e-05, "loss": 0.1019, "step": 4129 }, { "epoch": 4.19, "learning_rate": 1.6909222957637375e-05, "loss": 0.1198, "step": 4130 }, { "epoch": 4.19, "learning_rate": 1.6907638011419903e-05, "loss": 0.0979, "step": 4131 }, { "epoch": 4.19, "learning_rate": 1.6906052733249635e-05, "loss": 0.1178, "step": 4132 }, { "epoch": 4.19, "learning_rate": 1.6904467123202762e-05, "loss": 0.1079, "step": 4133 }, { "epoch": 4.19, "learning_rate": 1.6902881181355482e-05, "loss": 0.1069, "step": 4134 }, { "epoch": 4.2, "learning_rate": 1.6901294907784004e-05, "loss": 0.1348, "step": 4135 }, { "epoch": 4.2, "learning_rate": 1.689970830256456e-05, "loss": 0.126, "step": 4136 }, { "epoch": 4.2, "learning_rate": 1.6898121365773398e-05, "loss": 0.1102, "step": 4137 }, { "epoch": 4.2, "learning_rate": 1.6896534097486773e-05, "loss": 0.1089, "step": 4138 }, { "epoch": 4.2, "learning_rate": 1.6894946497780974e-05, "loss": 0.1269, "step": 4139 }, { "epoch": 4.2, "learning_rate": 1.6893358566732284e-05, "loss": 0.127, "step": 4140 }, { "epoch": 4.2, "learning_rate": 1.6891770304417015e-05, "loss": 0.104, "step": 4141 }, { "epoch": 4.2, "learning_rate": 1.6890181710911494e-05, "loss": 0.112, "step": 4142 }, { "epoch": 4.2, "learning_rate": 1.6888592786292063e-05, "loss": 0.0996, "step": 4143 }, { "epoch": 4.2, "learning_rate": 1.6887003530635077e-05, "loss": 0.1171, "step": 4144 }, { "epoch": 4.21, "learning_rate": 1.6885413944016913e-05, "loss": 0.1102, "step": 4145 }, { "epoch": 4.21, "learning_rate": 1.688382402651396e-05, "loss": 0.114, "step": 4146 }, { "epoch": 4.21, "learning_rate": 1.6882233778202612e-05, "loss": 0.1067, "step": 4147 }, { "epoch": 4.21, "learning_rate": 1.6880643199159303e-05, "loss": 0.1203, "step": 4148 }, { "epoch": 4.21, "learning_rate": 1.6879052289460463e-05, "loss": 0.1252, "step": 4149 }, { "epoch": 4.21, "learning_rate": 1.687746104918255e-05, "loss": 0.1209, "step": 4150 }, { "epoch": 4.21, "learning_rate": 1.6875869478402024e-05, "loss": 0.1009, "step": 4151 }, { "epoch": 4.21, "learning_rate": 1.6874277577195375e-05, "loss": 0.1206, "step": 4152 }, { "epoch": 4.21, "learning_rate": 1.6872685345639103e-05, "loss": 0.1207, "step": 4153 }, { "epoch": 4.22, "learning_rate": 1.6871092783809727e-05, "loss": 0.1088, "step": 4154 }, { "epoch": 4.22, "learning_rate": 1.686949989178377e-05, "loss": 0.1184, "step": 4155 }, { "epoch": 4.22, "learning_rate": 1.686790666963779e-05, "loss": 0.1117, "step": 4156 }, { "epoch": 4.22, "learning_rate": 1.6866313117448342e-05, "loss": 0.1167, "step": 4157 }, { "epoch": 4.22, "learning_rate": 1.6864719235292014e-05, "loss": 0.1091, "step": 4158 }, { "epoch": 4.22, "learning_rate": 1.6863125023245396e-05, "loss": 0.1207, "step": 4159 }, { "epoch": 4.22, "learning_rate": 1.6861530481385104e-05, "loss": 0.1143, "step": 4160 }, { "epoch": 4.22, "learning_rate": 1.6859935609787753e-05, "loss": 0.1261, "step": 4161 }, { "epoch": 4.22, "learning_rate": 1.685834040853e-05, "loss": 0.1346, "step": 4162 }, { "epoch": 4.22, "learning_rate": 1.68567448776885e-05, "loss": 0.1214, "step": 4163 }, { "epoch": 4.23, "learning_rate": 1.6855149017339925e-05, "loss": 0.1316, "step": 4164 }, { "epoch": 4.23, "learning_rate": 1.6853552827560964e-05, "loss": 0.1008, "step": 4165 }, { "epoch": 4.23, "learning_rate": 1.6851956308428325e-05, "loss": 0.1279, "step": 4166 }, { "epoch": 4.23, "learning_rate": 1.6850359460018737e-05, "loss": 0.1168, "step": 4167 }, { "epoch": 4.23, "learning_rate": 1.684876228240893e-05, "loss": 0.1195, "step": 4168 }, { "epoch": 4.23, "learning_rate": 1.6847164775675658e-05, "loss": 0.1224, "step": 4169 }, { "epoch": 4.23, "learning_rate": 1.6845566939895694e-05, "loss": 0.1097, "step": 4170 }, { "epoch": 4.23, "learning_rate": 1.6843968775145823e-05, "loss": 0.0944, "step": 4171 }, { "epoch": 4.23, "learning_rate": 1.6842370281502846e-05, "loss": 0.12, "step": 4172 }, { "epoch": 4.23, "learning_rate": 1.6840771459043574e-05, "loss": 0.1126, "step": 4173 }, { "epoch": 4.24, "learning_rate": 1.683917230784485e-05, "loss": 0.1079, "step": 4174 }, { "epoch": 4.24, "learning_rate": 1.6837572827983518e-05, "loss": 0.1256, "step": 4175 }, { "epoch": 4.24, "learning_rate": 1.683597301953644e-05, "loss": 0.1195, "step": 4176 }, { "epoch": 4.24, "learning_rate": 1.68343728825805e-05, "loss": 0.1106, "step": 4177 }, { "epoch": 4.24, "learning_rate": 1.6832772417192593e-05, "loss": 0.1115, "step": 4178 }, { "epoch": 4.24, "learning_rate": 1.6831171623449632e-05, "loss": 0.1133, "step": 4179 }, { "epoch": 4.24, "learning_rate": 1.682957050142854e-05, "loss": 0.1268, "step": 4180 }, { "epoch": 4.24, "learning_rate": 1.6827969051206266e-05, "loss": 0.1014, "step": 4181 }, { "epoch": 4.24, "learning_rate": 1.6826367272859763e-05, "loss": 0.1084, "step": 4182 }, { "epoch": 4.24, "learning_rate": 1.682476516646601e-05, "loss": 0.1112, "step": 4183 }, { "epoch": 4.25, "learning_rate": 1.6823162732102002e-05, "loss": 0.1163, "step": 4184 }, { "epoch": 4.25, "learning_rate": 1.6821559969844735e-05, "loss": 0.1057, "step": 4185 }, { "epoch": 4.25, "learning_rate": 1.681995687977124e-05, "loss": 0.1053, "step": 4186 }, { "epoch": 4.25, "learning_rate": 1.6818353461958552e-05, "loss": 0.1231, "step": 4187 }, { "epoch": 4.25, "learning_rate": 1.681674971648372e-05, "loss": 0.0973, "step": 4188 }, { "epoch": 4.25, "learning_rate": 1.6815145643423822e-05, "loss": 0.1066, "step": 4189 }, { "epoch": 4.25, "learning_rate": 1.6813541242855936e-05, "loss": 0.0874, "step": 4190 }, { "epoch": 4.25, "learning_rate": 1.681193651485717e-05, "loss": 0.1206, "step": 4191 }, { "epoch": 4.25, "learning_rate": 1.6810331459504636e-05, "loss": 0.1054, "step": 4192 }, { "epoch": 4.25, "learning_rate": 1.6808726076875465e-05, "loss": 0.1087, "step": 4193 }, { "epoch": 4.26, "learning_rate": 1.6807120367046805e-05, "loss": 0.1192, "step": 4194 }, { "epoch": 4.26, "learning_rate": 1.6805514330095826e-05, "loss": 0.1134, "step": 4195 }, { "epoch": 4.26, "learning_rate": 1.68039079660997e-05, "loss": 0.1121, "step": 4196 }, { "epoch": 4.26, "learning_rate": 1.680230127513563e-05, "loss": 0.1065, "step": 4197 }, { "epoch": 4.26, "learning_rate": 1.680069425728082e-05, "loss": 0.1086, "step": 4198 }, { "epoch": 4.26, "learning_rate": 1.67990869126125e-05, "loss": 0.1249, "step": 4199 }, { "epoch": 4.26, "learning_rate": 1.679747924120791e-05, "loss": 0.1143, "step": 4200 }, { "epoch": 4.26, "learning_rate": 1.679587124314431e-05, "loss": 0.0992, "step": 4201 }, { "epoch": 4.26, "learning_rate": 1.679426291849898e-05, "loss": 0.1329, "step": 4202 }, { "epoch": 4.26, "learning_rate": 1.6792654267349196e-05, "loss": 0.1128, "step": 4203 }, { "epoch": 4.27, "learning_rate": 1.679104528977227e-05, "loss": 0.0955, "step": 4204 }, { "epoch": 4.27, "learning_rate": 1.678943598584553e-05, "loss": 0.1316, "step": 4205 }, { "epoch": 4.27, "learning_rate": 1.6787826355646302e-05, "loss": 0.115, "step": 4206 }, { "epoch": 4.27, "learning_rate": 1.678621639925194e-05, "loss": 0.1304, "step": 4207 }, { "epoch": 4.27, "learning_rate": 1.6784606116739817e-05, "loss": 0.1196, "step": 4208 }, { "epoch": 4.27, "learning_rate": 1.678299550818731e-05, "loss": 0.1034, "step": 4209 }, { "epoch": 4.27, "learning_rate": 1.6781384573671825e-05, "loss": 0.1303, "step": 4210 }, { "epoch": 4.27, "learning_rate": 1.6779773313270773e-05, "loss": 0.1259, "step": 4211 }, { "epoch": 4.27, "learning_rate": 1.6778161727061587e-05, "loss": 0.1286, "step": 4212 }, { "epoch": 4.27, "learning_rate": 1.677654981512171e-05, "loss": 0.118, "step": 4213 }, { "epoch": 4.28, "learning_rate": 1.6774937577528607e-05, "loss": 0.1011, "step": 4214 }, { "epoch": 4.28, "learning_rate": 1.677332501435975e-05, "loss": 0.1252, "step": 4215 }, { "epoch": 4.28, "learning_rate": 1.6771712125692642e-05, "loss": 0.1226, "step": 4216 }, { "epoch": 4.28, "learning_rate": 1.677009891160478e-05, "loss": 0.1233, "step": 4217 }, { "epoch": 4.28, "learning_rate": 1.6768485372173696e-05, "loss": 0.1162, "step": 4218 }, { "epoch": 4.28, "learning_rate": 1.6766871507476928e-05, "loss": 0.1123, "step": 4219 }, { "epoch": 4.28, "learning_rate": 1.6765257317592035e-05, "loss": 0.0892, "step": 4220 }, { "epoch": 4.28, "learning_rate": 1.6763642802596587e-05, "loss": 0.1068, "step": 4221 }, { "epoch": 4.28, "learning_rate": 1.6762027962568167e-05, "loss": 0.1201, "step": 4222 }, { "epoch": 4.29, "learning_rate": 1.676041279758438e-05, "loss": 0.1144, "step": 4223 }, { "epoch": 4.29, "learning_rate": 1.6758797307722843e-05, "loss": 0.1263, "step": 4224 }, { "epoch": 4.29, "learning_rate": 1.6757181493061193e-05, "loss": 0.1011, "step": 4225 }, { "epoch": 4.29, "learning_rate": 1.6755565353677076e-05, "loss": 0.1156, "step": 4226 }, { "epoch": 4.29, "learning_rate": 1.675394888964816e-05, "loss": 0.1104, "step": 4227 }, { "epoch": 4.29, "learning_rate": 1.6752332101052125e-05, "loss": 0.1173, "step": 4228 }, { "epoch": 4.29, "learning_rate": 1.6750714987966667e-05, "loss": 0.1201, "step": 4229 }, { "epoch": 4.29, "learning_rate": 1.6749097550469494e-05, "loss": 0.1487, "step": 4230 }, { "epoch": 4.29, "learning_rate": 1.674747978863834e-05, "loss": 0.1137, "step": 4231 }, { "epoch": 4.29, "learning_rate": 1.6745861702550944e-05, "loss": 0.1336, "step": 4232 }, { "epoch": 4.3, "learning_rate": 1.6744243292285065e-05, "loss": 0.122, "step": 4233 }, { "epoch": 4.3, "learning_rate": 1.6742624557918478e-05, "loss": 0.1132, "step": 4234 }, { "epoch": 4.3, "learning_rate": 1.674100549952897e-05, "loss": 0.1298, "step": 4235 }, { "epoch": 4.3, "learning_rate": 1.6739386117194353e-05, "loss": 0.1233, "step": 4236 }, { "epoch": 4.3, "learning_rate": 1.6737766410992442e-05, "loss": 0.1052, "step": 4237 }, { "epoch": 4.3, "learning_rate": 1.6736146381001074e-05, "loss": 0.1153, "step": 4238 }, { "epoch": 4.3, "learning_rate": 1.67345260272981e-05, "loss": 0.1191, "step": 4239 }, { "epoch": 4.3, "learning_rate": 1.6732905349961394e-05, "loss": 0.1146, "step": 4240 }, { "epoch": 4.3, "learning_rate": 1.6731284349068833e-05, "loss": 0.115, "step": 4241 }, { "epoch": 4.3, "learning_rate": 1.6729663024698317e-05, "loss": 0.1289, "step": 4242 }, { "epoch": 4.31, "learning_rate": 1.672804137692776e-05, "loss": 0.1194, "step": 4243 }, { "epoch": 4.31, "learning_rate": 1.6726419405835093e-05, "loss": 0.1174, "step": 4244 }, { "epoch": 4.31, "learning_rate": 1.672479711149826e-05, "loss": 0.1145, "step": 4245 }, { "epoch": 4.31, "learning_rate": 1.6723174493995224e-05, "loss": 0.132, "step": 4246 }, { "epoch": 4.31, "learning_rate": 1.6721551553403953e-05, "loss": 0.123, "step": 4247 }, { "epoch": 4.31, "learning_rate": 1.6719928289802456e-05, "loss": 0.1245, "step": 4248 }, { "epoch": 4.31, "learning_rate": 1.671830470326872e-05, "loss": 0.1256, "step": 4249 }, { "epoch": 4.31, "learning_rate": 1.6716680793880783e-05, "loss": 0.1114, "step": 4250 }, { "epoch": 4.31, "learning_rate": 1.671505656171668e-05, "loss": 0.1093, "step": 4251 }, { "epoch": 4.31, "learning_rate": 1.671343200685446e-05, "loss": 0.1135, "step": 4252 }, { "epoch": 4.32, "learning_rate": 1.6711807129372198e-05, "loss": 0.1233, "step": 4253 }, { "epoch": 4.32, "learning_rate": 1.6710181929347978e-05, "loss": 0.1092, "step": 4254 }, { "epoch": 4.32, "learning_rate": 1.6708556406859897e-05, "loss": 0.1122, "step": 4255 }, { "epoch": 4.32, "learning_rate": 1.6706930561986072e-05, "loss": 0.1147, "step": 4256 }, { "epoch": 4.32, "learning_rate": 1.6705304394804637e-05, "loss": 0.0955, "step": 4257 }, { "epoch": 4.32, "learning_rate": 1.670367790539374e-05, "loss": 0.1303, "step": 4258 }, { "epoch": 4.32, "learning_rate": 1.6702051093831535e-05, "loss": 0.1159, "step": 4259 }, { "epoch": 4.32, "learning_rate": 1.6700423960196207e-05, "loss": 0.1341, "step": 4260 }, { "epoch": 4.32, "learning_rate": 1.6698796504565952e-05, "loss": 0.1161, "step": 4261 }, { "epoch": 4.32, "learning_rate": 1.6697168727018972e-05, "loss": 0.1054, "step": 4262 }, { "epoch": 4.33, "learning_rate": 1.66955406276335e-05, "loss": 0.12, "step": 4263 }, { "epoch": 4.33, "learning_rate": 1.6693912206487763e-05, "loss": 0.1211, "step": 4264 }, { "epoch": 4.33, "learning_rate": 1.6692283463660028e-05, "loss": 0.1209, "step": 4265 }, { "epoch": 4.33, "learning_rate": 1.669065439922856e-05, "loss": 0.1098, "step": 4266 }, { "epoch": 4.33, "learning_rate": 1.6689025013271645e-05, "loss": 0.1086, "step": 4267 }, { "epoch": 4.33, "learning_rate": 1.6687395305867587e-05, "loss": 0.1152, "step": 4268 }, { "epoch": 4.33, "learning_rate": 1.6685765277094702e-05, "loss": 0.1119, "step": 4269 }, { "epoch": 4.33, "learning_rate": 1.6684134927031318e-05, "loss": 0.121, "step": 4270 }, { "epoch": 4.33, "learning_rate": 1.6682504255755794e-05, "loss": 0.128, "step": 4271 }, { "epoch": 4.33, "learning_rate": 1.6680873263346485e-05, "loss": 0.126, "step": 4272 }, { "epoch": 4.34, "learning_rate": 1.667924194988177e-05, "loss": 0.1114, "step": 4273 }, { "epoch": 4.34, "learning_rate": 1.6677610315440045e-05, "loss": 0.1051, "step": 4274 }, { "epoch": 4.34, "learning_rate": 1.667597836009972e-05, "loss": 0.1177, "step": 4275 }, { "epoch": 4.34, "learning_rate": 1.667434608393922e-05, "loss": 0.1337, "step": 4276 }, { "epoch": 4.34, "learning_rate": 1.6672713487036985e-05, "loss": 0.1393, "step": 4277 }, { "epoch": 4.34, "learning_rate": 1.667108056947147e-05, "loss": 0.1185, "step": 4278 }, { "epoch": 4.34, "learning_rate": 1.666944733132115e-05, "loss": 0.1232, "step": 4279 }, { "epoch": 4.34, "learning_rate": 1.6667813772664507e-05, "loss": 0.1277, "step": 4280 }, { "epoch": 4.34, "learning_rate": 1.6666179893580046e-05, "loss": 0.1317, "step": 4281 }, { "epoch": 4.35, "learning_rate": 1.6664545694146285e-05, "loss": 0.1368, "step": 4282 }, { "epoch": 4.35, "learning_rate": 1.6662911174441752e-05, "loss": 0.1652, "step": 4283 }, { "epoch": 4.35, "learning_rate": 1.6661276334545004e-05, "loss": 0.119, "step": 4284 }, { "epoch": 4.35, "learning_rate": 1.6659641174534592e-05, "loss": 0.1119, "step": 4285 }, { "epoch": 4.35, "learning_rate": 1.665800569448911e-05, "loss": 0.1162, "step": 4286 }, { "epoch": 4.35, "learning_rate": 1.6656369894487142e-05, "loss": 0.1218, "step": 4287 }, { "epoch": 4.35, "learning_rate": 1.6654733774607302e-05, "loss": 0.1185, "step": 4288 }, { "epoch": 4.35, "learning_rate": 1.6653097334928218e-05, "loss": 0.1203, "step": 4289 }, { "epoch": 4.35, "learning_rate": 1.6651460575528522e-05, "loss": 0.1235, "step": 4290 }, { "epoch": 4.35, "learning_rate": 1.6649823496486875e-05, "loss": 0.1168, "step": 4291 }, { "epoch": 4.36, "learning_rate": 1.664818609788195e-05, "loss": 0.1242, "step": 4292 }, { "epoch": 4.36, "learning_rate": 1.664654837979243e-05, "loss": 0.1213, "step": 4293 }, { "epoch": 4.36, "learning_rate": 1.6644910342297023e-05, "loss": 0.1115, "step": 4294 }, { "epoch": 4.36, "learning_rate": 1.664327198547444e-05, "loss": 0.1201, "step": 4295 }, { "epoch": 4.36, "learning_rate": 1.6641633309403416e-05, "loss": 0.133, "step": 4296 }, { "epoch": 4.36, "learning_rate": 1.66399943141627e-05, "loss": 0.1303, "step": 4297 }, { "epoch": 4.36, "learning_rate": 1.6638354999831054e-05, "loss": 0.1148, "step": 4298 }, { "epoch": 4.36, "learning_rate": 1.6636715366487254e-05, "loss": 0.0935, "step": 4299 }, { "epoch": 4.36, "learning_rate": 1.66350754142101e-05, "loss": 0.1103, "step": 4300 }, { "epoch": 4.36, "learning_rate": 1.66334351430784e-05, "loss": 0.1129, "step": 4301 }, { "epoch": 4.37, "learning_rate": 1.6631794553170976e-05, "loss": 0.1426, "step": 4302 }, { "epoch": 4.37, "learning_rate": 1.663015364456667e-05, "loss": 0.1354, "step": 4303 }, { "epoch": 4.37, "learning_rate": 1.6628512417344332e-05, "loss": 0.1111, "step": 4304 }, { "epoch": 4.37, "learning_rate": 1.662687087158284e-05, "loss": 0.1403, "step": 4305 }, { "epoch": 4.37, "learning_rate": 1.6625229007361083e-05, "loss": 0.1232, "step": 4306 }, { "epoch": 4.37, "learning_rate": 1.6623586824757952e-05, "loss": 0.1192, "step": 4307 }, { "epoch": 4.37, "learning_rate": 1.662194432385237e-05, "loss": 0.1055, "step": 4308 }, { "epoch": 4.37, "learning_rate": 1.6620301504723264e-05, "loss": 0.1271, "step": 4309 }, { "epoch": 4.37, "learning_rate": 1.661865836744959e-05, "loss": 0.1098, "step": 4310 }, { "epoch": 4.37, "learning_rate": 1.6617014912110297e-05, "loss": 0.1152, "step": 4311 }, { "epoch": 4.38, "learning_rate": 1.6615371138784374e-05, "loss": 0.1322, "step": 4312 }, { "epoch": 4.38, "learning_rate": 1.661372704755081e-05, "loss": 0.1031, "step": 4313 }, { "epoch": 4.38, "learning_rate": 1.6612082638488615e-05, "loss": 0.1204, "step": 4314 }, { "epoch": 4.38, "learning_rate": 1.6610437911676812e-05, "loss": 0.1198, "step": 4315 }, { "epoch": 4.38, "learning_rate": 1.6608792867194436e-05, "loss": 0.1279, "step": 4316 }, { "epoch": 4.38, "learning_rate": 1.6607147505120547e-05, "loss": 0.1179, "step": 4317 }, { "epoch": 4.38, "learning_rate": 1.6605501825534212e-05, "loss": 0.1326, "step": 4318 }, { "epoch": 4.38, "learning_rate": 1.6603855828514512e-05, "loss": 0.1135, "step": 4319 }, { "epoch": 4.38, "learning_rate": 1.6602209514140552e-05, "loss": 0.1308, "step": 4320 }, { "epoch": 4.38, "learning_rate": 1.6600562882491445e-05, "loss": 0.1028, "step": 4321 }, { "epoch": 4.39, "learning_rate": 1.659891593364632e-05, "loss": 0.1123, "step": 4322 }, { "epoch": 4.39, "learning_rate": 1.659726866768433e-05, "loss": 0.138, "step": 4323 }, { "epoch": 4.39, "learning_rate": 1.659562108468462e-05, "loss": 0.1168, "step": 4324 }, { "epoch": 4.39, "learning_rate": 1.6593973184726383e-05, "loss": 0.1196, "step": 4325 }, { "epoch": 4.39, "learning_rate": 1.6592324967888804e-05, "loss": 0.1252, "step": 4326 }, { "epoch": 4.39, "learning_rate": 1.6590676434251086e-05, "loss": 0.1197, "step": 4327 }, { "epoch": 4.39, "learning_rate": 1.6589027583892456e-05, "loss": 0.1229, "step": 4328 }, { "epoch": 4.39, "learning_rate": 1.6587378416892147e-05, "loss": 0.1046, "step": 4329 }, { "epoch": 4.39, "learning_rate": 1.6585728933329412e-05, "loss": 0.1223, "step": 4330 }, { "epoch": 4.39, "learning_rate": 1.6584079133283523e-05, "loss": 0.1223, "step": 4331 }, { "epoch": 4.4, "learning_rate": 1.6582429016833755e-05, "loss": 0.1231, "step": 4332 }, { "epoch": 4.4, "learning_rate": 1.658077858405941e-05, "loss": 0.1354, "step": 4333 }, { "epoch": 4.4, "learning_rate": 1.6579127835039807e-05, "loss": 0.119, "step": 4334 }, { "epoch": 4.4, "learning_rate": 1.6577476769854266e-05, "loss": 0.1266, "step": 4335 }, { "epoch": 4.4, "learning_rate": 1.6575825388582128e-05, "loss": 0.1192, "step": 4336 }, { "epoch": 4.4, "learning_rate": 1.657417369130276e-05, "loss": 0.1221, "step": 4337 }, { "epoch": 4.4, "learning_rate": 1.657252167809553e-05, "loss": 0.1214, "step": 4338 }, { "epoch": 4.4, "learning_rate": 1.657086934903983e-05, "loss": 0.141, "step": 4339 }, { "epoch": 4.4, "learning_rate": 1.6569216704215065e-05, "loss": 0.1256, "step": 4340 }, { "epoch": 4.4, "learning_rate": 1.656756374370065e-05, "loss": 0.1124, "step": 4341 }, { "epoch": 4.41, "learning_rate": 1.656591046757602e-05, "loss": 0.1237, "step": 4342 }, { "epoch": 4.41, "learning_rate": 1.656425687592063e-05, "loss": 0.1179, "step": 4343 }, { "epoch": 4.41, "learning_rate": 1.6562602968813942e-05, "loss": 0.1249, "step": 4344 }, { "epoch": 4.41, "learning_rate": 1.6560948746335435e-05, "loss": 0.1143, "step": 4345 }, { "epoch": 4.41, "learning_rate": 1.6559294208564604e-05, "loss": 0.0988, "step": 4346 }, { "epoch": 4.41, "learning_rate": 1.6557639355580957e-05, "loss": 0.11, "step": 4347 }, { "epoch": 4.41, "learning_rate": 1.655598418746403e-05, "loss": 0.1152, "step": 4348 }, { "epoch": 4.41, "learning_rate": 1.655432870429335e-05, "loss": 0.1461, "step": 4349 }, { "epoch": 4.41, "learning_rate": 1.655267290614848e-05, "loss": 0.1005, "step": 4350 }, { "epoch": 4.42, "learning_rate": 1.655101679310899e-05, "loss": 0.1406, "step": 4351 }, { "epoch": 4.42, "learning_rate": 1.6549360365254466e-05, "loss": 0.1352, "step": 4352 }, { "epoch": 4.42, "learning_rate": 1.654770362266451e-05, "loss": 0.133, "step": 4353 }, { "epoch": 4.42, "learning_rate": 1.6546046565418734e-05, "loss": 0.1259, "step": 4354 }, { "epoch": 4.42, "learning_rate": 1.6544389193596773e-05, "loss": 0.0919, "step": 4355 }, { "epoch": 4.42, "learning_rate": 1.6542731507278276e-05, "loss": 0.1092, "step": 4356 }, { "epoch": 4.42, "learning_rate": 1.6541073506542897e-05, "loss": 0.1077, "step": 4357 }, { "epoch": 4.42, "learning_rate": 1.653941519147032e-05, "loss": 0.1392, "step": 4358 }, { "epoch": 4.42, "learning_rate": 1.6537756562140235e-05, "loss": 0.129, "step": 4359 }, { "epoch": 4.42, "learning_rate": 1.6536097618632347e-05, "loss": 0.1109, "step": 4360 }, { "epoch": 4.43, "learning_rate": 1.653443836102638e-05, "loss": 0.1293, "step": 4361 }, { "epoch": 4.43, "learning_rate": 1.6532778789402067e-05, "loss": 0.1168, "step": 4362 }, { "epoch": 4.43, "learning_rate": 1.653111890383917e-05, "loss": 0.1166, "step": 4363 }, { "epoch": 4.43, "learning_rate": 1.6529458704417444e-05, "loss": 0.1088, "step": 4364 }, { "epoch": 4.43, "learning_rate": 1.652779819121668e-05, "loss": 0.1187, "step": 4365 }, { "epoch": 4.43, "learning_rate": 1.652613736431667e-05, "loss": 0.1103, "step": 4366 }, { "epoch": 4.43, "learning_rate": 1.652447622379723e-05, "loss": 0.1262, "step": 4367 }, { "epoch": 4.43, "learning_rate": 1.6522814769738188e-05, "loss": 0.1092, "step": 4368 }, { "epoch": 4.43, "learning_rate": 1.6521153002219384e-05, "loss": 0.1295, "step": 4369 }, { "epoch": 4.43, "learning_rate": 1.6519490921320677e-05, "loss": 0.1041, "step": 4370 }, { "epoch": 4.44, "learning_rate": 1.6517828527121942e-05, "loss": 0.114, "step": 4371 }, { "epoch": 4.44, "learning_rate": 1.6516165819703064e-05, "loss": 0.1252, "step": 4372 }, { "epoch": 4.44, "learning_rate": 1.6514502799143945e-05, "loss": 0.1416, "step": 4373 }, { "epoch": 4.44, "learning_rate": 1.6512839465524506e-05, "loss": 0.1391, "step": 4374 }, { "epoch": 4.44, "learning_rate": 1.6511175818924678e-05, "loss": 0.122, "step": 4375 }, { "epoch": 4.44, "learning_rate": 1.650951185942441e-05, "loss": 0.1417, "step": 4376 }, { "epoch": 4.44, "learning_rate": 1.6507847587103665e-05, "loss": 0.1283, "step": 4377 }, { "epoch": 4.44, "learning_rate": 1.650618300204242e-05, "loss": 0.1305, "step": 4378 }, { "epoch": 4.44, "learning_rate": 1.6504518104320673e-05, "loss": 0.1174, "step": 4379 }, { "epoch": 4.44, "learning_rate": 1.6502852894018423e-05, "loss": 0.1333, "step": 4380 }, { "epoch": 4.45, "learning_rate": 1.6501187371215704e-05, "loss": 0.1039, "step": 4381 }, { "epoch": 4.45, "learning_rate": 1.6499521535992543e-05, "loss": 0.1147, "step": 4382 }, { "epoch": 4.45, "learning_rate": 1.6497855388429e-05, "loss": 0.1272, "step": 4383 }, { "epoch": 4.45, "learning_rate": 1.6496188928605144e-05, "loss": 0.1274, "step": 4384 }, { "epoch": 4.45, "learning_rate": 1.6494522156601057e-05, "loss": 0.1256, "step": 4385 }, { "epoch": 4.45, "learning_rate": 1.6492855072496835e-05, "loss": 0.1038, "step": 4386 }, { "epoch": 4.45, "learning_rate": 1.6491187676372593e-05, "loss": 0.1169, "step": 4387 }, { "epoch": 4.45, "learning_rate": 1.648951996830846e-05, "loss": 0.123, "step": 4388 }, { "epoch": 4.45, "learning_rate": 1.6487851948384577e-05, "loss": 0.1, "step": 4389 }, { "epoch": 4.45, "learning_rate": 1.6486183616681102e-05, "loss": 0.1181, "step": 4390 }, { "epoch": 4.46, "learning_rate": 1.648451497327821e-05, "loss": 0.1349, "step": 4391 }, { "epoch": 4.46, "learning_rate": 1.6482846018256088e-05, "loss": 0.1136, "step": 4392 }, { "epoch": 4.46, "learning_rate": 1.6481176751694942e-05, "loss": 0.1279, "step": 4393 }, { "epoch": 4.46, "learning_rate": 1.6479507173674988e-05, "loss": 0.1161, "step": 4394 }, { "epoch": 4.46, "learning_rate": 1.647783728427646e-05, "loss": 0.1145, "step": 4395 }, { "epoch": 4.46, "learning_rate": 1.6476167083579604e-05, "loss": 0.1314, "step": 4396 }, { "epoch": 4.46, "learning_rate": 1.6474496571664683e-05, "loss": 0.1208, "step": 4397 }, { "epoch": 4.46, "learning_rate": 1.647282574861198e-05, "loss": 0.1174, "step": 4398 }, { "epoch": 4.46, "learning_rate": 1.6471154614501776e-05, "loss": 0.1251, "step": 4399 }, { "epoch": 4.46, "learning_rate": 1.6469483169414395e-05, "loss": 0.1353, "step": 4400 }, { "epoch": 4.47, "learning_rate": 1.646781141343015e-05, "loss": 0.1373, "step": 4401 }, { "epoch": 4.47, "learning_rate": 1.6466139346629377e-05, "loss": 0.1388, "step": 4402 }, { "epoch": 4.47, "learning_rate": 1.6464466969092433e-05, "loss": 0.1516, "step": 4403 }, { "epoch": 4.47, "learning_rate": 1.6462794280899688e-05, "loss": 0.1291, "step": 4404 }, { "epoch": 4.47, "learning_rate": 1.6461121282131518e-05, "loss": 0.1345, "step": 4405 }, { "epoch": 4.47, "learning_rate": 1.6459447972868327e-05, "loss": 0.1053, "step": 4406 }, { "epoch": 4.47, "learning_rate": 1.645777435319052e-05, "loss": 0.1157, "step": 4407 }, { "epoch": 4.47, "learning_rate": 1.6456100423178528e-05, "loss": 0.1118, "step": 4408 }, { "epoch": 4.47, "learning_rate": 1.6454426182912798e-05, "loss": 0.1156, "step": 4409 }, { "epoch": 4.47, "learning_rate": 1.645275163247378e-05, "loss": 0.1084, "step": 4410 }, { "epoch": 4.48, "learning_rate": 1.645107677194195e-05, "loss": 0.1499, "step": 4411 }, { "epoch": 4.48, "learning_rate": 1.6449401601397795e-05, "loss": 0.1256, "step": 4412 }, { "epoch": 4.48, "learning_rate": 1.6447726120921813e-05, "loss": 0.1121, "step": 4413 }, { "epoch": 4.48, "learning_rate": 1.6446050330594525e-05, "loss": 0.1263, "step": 4414 }, { "epoch": 4.48, "learning_rate": 1.644437423049646e-05, "loss": 0.1189, "step": 4415 }, { "epoch": 4.48, "learning_rate": 1.6442697820708166e-05, "loss": 0.1215, "step": 4416 }, { "epoch": 4.48, "learning_rate": 1.64410211013102e-05, "loss": 0.1045, "step": 4417 }, { "epoch": 4.48, "learning_rate": 1.643934407238315e-05, "loss": 0.1259, "step": 4418 }, { "epoch": 4.48, "learning_rate": 1.6437666734007592e-05, "loss": 0.1499, "step": 4419 }, { "epoch": 4.49, "learning_rate": 1.6435989086264142e-05, "loss": 0.1169, "step": 4420 }, { "epoch": 4.49, "learning_rate": 1.643431112923342e-05, "loss": 0.1278, "step": 4421 }, { "epoch": 4.49, "learning_rate": 1.6432632862996056e-05, "loss": 0.1101, "step": 4422 }, { "epoch": 4.49, "learning_rate": 1.6430954287632707e-05, "loss": 0.1286, "step": 4423 }, { "epoch": 4.49, "learning_rate": 1.6429275403224038e-05, "loss": 0.1097, "step": 4424 }, { "epoch": 4.49, "learning_rate": 1.6427596209850723e-05, "loss": 0.1058, "step": 4425 }, { "epoch": 4.49, "learning_rate": 1.6425916707593464e-05, "loss": 0.1118, "step": 4426 }, { "epoch": 4.49, "learning_rate": 1.6424236896532966e-05, "loss": 0.112, "step": 4427 }, { "epoch": 4.49, "learning_rate": 1.6422556776749955e-05, "loss": 0.1342, "step": 4428 }, { "epoch": 4.49, "learning_rate": 1.6420876348325178e-05, "loss": 0.1181, "step": 4429 }, { "epoch": 4.5, "learning_rate": 1.641919561133938e-05, "loss": 0.1136, "step": 4430 }, { "epoch": 4.5, "learning_rate": 1.6417514565873335e-05, "loss": 0.1212, "step": 4431 }, { "epoch": 4.5, "learning_rate": 1.6415833212007823e-05, "loss": 0.1307, "step": 4432 }, { "epoch": 4.5, "learning_rate": 1.6414151549823647e-05, "loss": 0.1019, "step": 4433 }, { "epoch": 4.5, "learning_rate": 1.641246957940162e-05, "loss": 0.1426, "step": 4434 }, { "epoch": 4.5, "learning_rate": 1.6410787300822572e-05, "loss": 0.1246, "step": 4435 }, { "epoch": 4.5, "learning_rate": 1.640910471416734e-05, "loss": 0.1322, "step": 4436 }, { "epoch": 4.5, "learning_rate": 1.640742181951679e-05, "loss": 0.1082, "step": 4437 }, { "epoch": 4.5, "learning_rate": 1.6405738616951794e-05, "loss": 0.1237, "step": 4438 }, { "epoch": 4.5, "learning_rate": 1.6404055106553236e-05, "loss": 0.1173, "step": 4439 }, { "epoch": 4.51, "learning_rate": 1.640237128840202e-05, "loss": 0.156, "step": 4440 }, { "epoch": 4.51, "learning_rate": 1.640068716257906e-05, "loss": 0.1309, "step": 4441 }, { "epoch": 4.51, "learning_rate": 1.63990027291653e-05, "loss": 0.1341, "step": 4442 }, { "epoch": 4.51, "learning_rate": 1.6397317988241677e-05, "loss": 0.103, "step": 4443 }, { "epoch": 4.51, "learning_rate": 1.639563293988915e-05, "loss": 0.1213, "step": 4444 }, { "epoch": 4.51, "learning_rate": 1.6393947584188704e-05, "loss": 0.1269, "step": 4445 }, { "epoch": 4.51, "learning_rate": 1.6392261921221323e-05, "loss": 0.1286, "step": 4446 }, { "epoch": 4.51, "learning_rate": 1.639057595106802e-05, "loss": 0.1273, "step": 4447 }, { "epoch": 4.51, "learning_rate": 1.6388889673809815e-05, "loss": 0.1244, "step": 4448 }, { "epoch": 4.51, "learning_rate": 1.6387203089527733e-05, "loss": 0.1295, "step": 4449 }, { "epoch": 4.52, "learning_rate": 1.638551619830284e-05, "loss": 0.1207, "step": 4450 }, { "epoch": 4.52, "learning_rate": 1.6383829000216186e-05, "loss": 0.1201, "step": 4451 }, { "epoch": 4.52, "learning_rate": 1.6382141495348865e-05, "loss": 0.1279, "step": 4452 }, { "epoch": 4.52, "learning_rate": 1.638045368378196e-05, "loss": 0.1355, "step": 4453 }, { "epoch": 4.52, "learning_rate": 1.6378765565596588e-05, "loss": 0.137, "step": 4454 }, { "epoch": 4.52, "learning_rate": 1.6377077140873868e-05, "loss": 0.1166, "step": 4455 }, { "epoch": 4.52, "learning_rate": 1.6375388409694942e-05, "loss": 0.1202, "step": 4456 }, { "epoch": 4.52, "learning_rate": 1.6373699372140967e-05, "loss": 0.1318, "step": 4457 }, { "epoch": 4.52, "learning_rate": 1.63720100282931e-05, "loss": 0.1319, "step": 4458 }, { "epoch": 4.52, "learning_rate": 1.637032037823253e-05, "loss": 0.1166, "step": 4459 }, { "epoch": 4.53, "learning_rate": 1.636863042204046e-05, "loss": 0.1081, "step": 4460 }, { "epoch": 4.53, "learning_rate": 1.63669401597981e-05, "loss": 0.1325, "step": 4461 }, { "epoch": 4.53, "learning_rate": 1.6365249591586666e-05, "loss": 0.112, "step": 4462 }, { "epoch": 4.53, "learning_rate": 1.6363558717487412e-05, "loss": 0.1564, "step": 4463 }, { "epoch": 4.53, "learning_rate": 1.6361867537581596e-05, "loss": 0.1257, "step": 4464 }, { "epoch": 4.53, "learning_rate": 1.636017605195048e-05, "loss": 0.1311, "step": 4465 }, { "epoch": 4.53, "learning_rate": 1.635848426067535e-05, "loss": 0.1306, "step": 4466 }, { "epoch": 4.53, "learning_rate": 1.6356792163837518e-05, "loss": 0.1314, "step": 4467 }, { "epoch": 4.53, "learning_rate": 1.6355099761518286e-05, "loss": 0.1206, "step": 4468 }, { "epoch": 4.53, "learning_rate": 1.6353407053798994e-05, "loss": 0.1281, "step": 4469 }, { "epoch": 4.54, "learning_rate": 1.6351714040760977e-05, "loss": 0.1151, "step": 4470 }, { "epoch": 4.54, "learning_rate": 1.6350020722485608e-05, "loss": 0.1416, "step": 4471 }, { "epoch": 4.54, "learning_rate": 1.6348327099054247e-05, "loss": 0.1192, "step": 4472 }, { "epoch": 4.54, "learning_rate": 1.634663317054829e-05, "loss": 0.1346, "step": 4473 }, { "epoch": 4.54, "learning_rate": 1.6344938937049135e-05, "loss": 0.1096, "step": 4474 }, { "epoch": 4.54, "learning_rate": 1.6343244398638205e-05, "loss": 0.1187, "step": 4475 }, { "epoch": 4.54, "learning_rate": 1.6341549555396935e-05, "loss": 0.1038, "step": 4476 }, { "epoch": 4.54, "learning_rate": 1.633985440740676e-05, "loss": 0.1218, "step": 4477 }, { "epoch": 4.54, "learning_rate": 1.6338158954749155e-05, "loss": 0.1224, "step": 4478 }, { "epoch": 4.54, "learning_rate": 1.6336463197505594e-05, "loss": 0.1343, "step": 4479 }, { "epoch": 4.55, "learning_rate": 1.6334767135757563e-05, "loss": 0.1525, "step": 4480 }, { "epoch": 4.55, "learning_rate": 1.6333070769586573e-05, "loss": 0.1177, "step": 4481 }, { "epoch": 4.55, "learning_rate": 1.6331374099074137e-05, "loss": 0.1067, "step": 4482 }, { "epoch": 4.55, "learning_rate": 1.6329677124301798e-05, "loss": 0.1266, "step": 4483 }, { "epoch": 4.55, "learning_rate": 1.6327979845351102e-05, "loss": 0.1281, "step": 4484 }, { "epoch": 4.55, "learning_rate": 1.632628226230361e-05, "loss": 0.1328, "step": 4485 }, { "epoch": 4.55, "learning_rate": 1.632458437524091e-05, "loss": 0.1353, "step": 4486 }, { "epoch": 4.55, "learning_rate": 1.6322886184244588e-05, "loss": 0.0984, "step": 4487 }, { "epoch": 4.55, "learning_rate": 1.6321187689396257e-05, "loss": 0.1206, "step": 4488 }, { "epoch": 4.56, "learning_rate": 1.6319488890777536e-05, "loss": 0.1248, "step": 4489 }, { "epoch": 4.56, "learning_rate": 1.6317789788470063e-05, "loss": 0.1176, "step": 4490 }, { "epoch": 4.56, "learning_rate": 1.631609038255549e-05, "loss": 0.1138, "step": 4491 }, { "epoch": 4.56, "learning_rate": 1.631439067311548e-05, "loss": 0.1549, "step": 4492 }, { "epoch": 4.56, "learning_rate": 1.6312690660231723e-05, "loss": 0.135, "step": 4493 }, { "epoch": 4.56, "learning_rate": 1.6310990343985907e-05, "loss": 0.1139, "step": 4494 }, { "epoch": 4.56, "learning_rate": 1.6309289724459744e-05, "loss": 0.1091, "step": 4495 }, { "epoch": 4.56, "learning_rate": 1.630758880173496e-05, "loss": 0.1329, "step": 4496 }, { "epoch": 4.56, "learning_rate": 1.6305887575893297e-05, "loss": 0.1144, "step": 4497 }, { "epoch": 4.56, "learning_rate": 1.6304186047016502e-05, "loss": 0.1078, "step": 4498 }, { "epoch": 4.57, "learning_rate": 1.630248421518635e-05, "loss": 0.1346, "step": 4499 }, { "epoch": 4.57, "learning_rate": 1.630078208048462e-05, "loss": 0.1214, "step": 4500 }, { "epoch": 4.57, "learning_rate": 1.6299079642993112e-05, "loss": 0.1368, "step": 4501 }, { "epoch": 4.57, "learning_rate": 1.6297376902793636e-05, "loss": 0.1349, "step": 4502 }, { "epoch": 4.57, "learning_rate": 1.629567385996802e-05, "loss": 0.1185, "step": 4503 }, { "epoch": 4.57, "learning_rate": 1.6293970514598107e-05, "loss": 0.1207, "step": 4504 }, { "epoch": 4.57, "learning_rate": 1.629226686676575e-05, "loss": 0.1334, "step": 4505 }, { "epoch": 4.57, "learning_rate": 1.629056291655282e-05, "loss": 0.1258, "step": 4506 }, { "epoch": 4.57, "learning_rate": 1.62888586640412e-05, "loss": 0.1132, "step": 4507 }, { "epoch": 4.57, "learning_rate": 1.6287154109312797e-05, "loss": 0.1245, "step": 4508 }, { "epoch": 4.58, "learning_rate": 1.6285449252449512e-05, "loss": 0.1192, "step": 4509 }, { "epoch": 4.58, "learning_rate": 1.6283744093533286e-05, "loss": 0.1283, "step": 4510 }, { "epoch": 4.58, "learning_rate": 1.6282038632646057e-05, "loss": 0.1201, "step": 4511 }, { "epoch": 4.58, "learning_rate": 1.628033286986978e-05, "loss": 0.1242, "step": 4512 }, { "epoch": 4.58, "learning_rate": 1.6278626805286432e-05, "loss": 0.1168, "step": 4513 }, { "epoch": 4.58, "learning_rate": 1.6276920438977994e-05, "loss": 0.1241, "step": 4514 }, { "epoch": 4.58, "learning_rate": 1.6275213771026474e-05, "loss": 0.1339, "step": 4515 }, { "epoch": 4.58, "learning_rate": 1.6273506801513877e-05, "loss": 0.1315, "step": 4516 }, { "epoch": 4.58, "learning_rate": 1.6271799530522244e-05, "loss": 0.123, "step": 4517 }, { "epoch": 4.58, "learning_rate": 1.6270091958133615e-05, "loss": 0.121, "step": 4518 }, { "epoch": 4.59, "learning_rate": 1.6268384084430043e-05, "loss": 0.1308, "step": 4519 }, { "epoch": 4.59, "learning_rate": 1.626667590949361e-05, "loss": 0.1351, "step": 4520 }, { "epoch": 4.59, "learning_rate": 1.6264967433406404e-05, "loss": 0.117, "step": 4521 }, { "epoch": 4.59, "learning_rate": 1.626325865625052e-05, "loss": 0.1182, "step": 4522 }, { "epoch": 4.59, "learning_rate": 1.626154957810808e-05, "loss": 0.1377, "step": 4523 }, { "epoch": 4.59, "learning_rate": 1.6259840199061215e-05, "loss": 0.1239, "step": 4524 }, { "epoch": 4.59, "learning_rate": 1.6258130519192072e-05, "loss": 0.1199, "step": 4525 }, { "epoch": 4.59, "learning_rate": 1.625642053858281e-05, "loss": 0.1326, "step": 4526 }, { "epoch": 4.59, "learning_rate": 1.6254710257315602e-05, "loss": 0.131, "step": 4527 }, { "epoch": 4.59, "learning_rate": 1.6252999675472637e-05, "loss": 0.1458, "step": 4528 }, { "epoch": 4.6, "learning_rate": 1.625128879313612e-05, "loss": 0.1451, "step": 4529 }, { "epoch": 4.6, "learning_rate": 1.6249577610388267e-05, "loss": 0.1137, "step": 4530 }, { "epoch": 4.6, "learning_rate": 1.6247866127311316e-05, "loss": 0.1359, "step": 4531 }, { "epoch": 4.6, "learning_rate": 1.6246154343987508e-05, "loss": 0.1218, "step": 4532 }, { "epoch": 4.6, "learning_rate": 1.6244442260499106e-05, "loss": 0.112, "step": 4533 }, { "epoch": 4.6, "learning_rate": 1.624272987692838e-05, "loss": 0.117, "step": 4534 }, { "epoch": 4.6, "learning_rate": 1.6241017193357634e-05, "loss": 0.1062, "step": 4535 }, { "epoch": 4.6, "learning_rate": 1.6239304209869163e-05, "loss": 0.1328, "step": 4536 }, { "epoch": 4.6, "learning_rate": 1.6237590926545286e-05, "loss": 0.1235, "step": 4537 }, { "epoch": 4.6, "learning_rate": 1.6235877343468342e-05, "loss": 0.1282, "step": 4538 }, { "epoch": 4.61, "learning_rate": 1.6234163460720675e-05, "loss": 0.1307, "step": 4539 }, { "epoch": 4.61, "learning_rate": 1.623244927838464e-05, "loss": 0.1175, "step": 4540 }, { "epoch": 4.61, "learning_rate": 1.6230734796542628e-05, "loss": 0.1183, "step": 4541 }, { "epoch": 4.61, "learning_rate": 1.6229020015277016e-05, "loss": 0.1209, "step": 4542 }, { "epoch": 4.61, "learning_rate": 1.6227304934670222e-05, "loss": 0.1311, "step": 4543 }, { "epoch": 4.61, "learning_rate": 1.622558955480466e-05, "loss": 0.1163, "step": 4544 }, { "epoch": 4.61, "learning_rate": 1.6223873875762758e-05, "loss": 0.1276, "step": 4545 }, { "epoch": 4.61, "learning_rate": 1.6222157897626975e-05, "loss": 0.091, "step": 4546 }, { "epoch": 4.61, "learning_rate": 1.6220441620479762e-05, "loss": 0.1312, "step": 4547 }, { "epoch": 4.61, "learning_rate": 1.6218725044403615e-05, "loss": 0.1203, "step": 4548 }, { "epoch": 4.62, "learning_rate": 1.6217008169481006e-05, "loss": 0.1169, "step": 4549 }, { "epoch": 4.62, "learning_rate": 1.621529099579445e-05, "loss": 0.1179, "step": 4550 }, { "epoch": 4.62, "learning_rate": 1.6213573523426466e-05, "loss": 0.1423, "step": 4551 }, { "epoch": 4.62, "learning_rate": 1.621185575245959e-05, "loss": 0.1143, "step": 4552 }, { "epoch": 4.62, "learning_rate": 1.621013768297637e-05, "loss": 0.0963, "step": 4553 }, { "epoch": 4.62, "learning_rate": 1.620841931505937e-05, "loss": 0.1387, "step": 4554 }, { "epoch": 4.62, "learning_rate": 1.620670064879117e-05, "loss": 0.1198, "step": 4555 }, { "epoch": 4.62, "learning_rate": 1.6204981684254356e-05, "loss": 0.1147, "step": 4556 }, { "epoch": 4.62, "learning_rate": 1.620326242153154e-05, "loss": 0.123, "step": 4557 }, { "epoch": 4.63, "learning_rate": 1.620154286070534e-05, "loss": 0.1393, "step": 4558 }, { "epoch": 4.63, "learning_rate": 1.6199823001858387e-05, "loss": 0.1165, "step": 4559 }, { "epoch": 4.63, "learning_rate": 1.619810284507334e-05, "loss": 0.1241, "step": 4560 }, { "epoch": 4.63, "learning_rate": 1.6196382390432855e-05, "loss": 0.114, "step": 4561 }, { "epoch": 4.63, "learning_rate": 1.619466163801961e-05, "loss": 0.1288, "step": 4562 }, { "epoch": 4.63, "learning_rate": 1.6192940587916304e-05, "loss": 0.1234, "step": 4563 }, { "epoch": 4.63, "learning_rate": 1.619121924020564e-05, "loss": 0.113, "step": 4564 }, { "epoch": 4.63, "learning_rate": 1.6189497594970335e-05, "loss": 0.138, "step": 4565 }, { "epoch": 4.63, "learning_rate": 1.618777565229313e-05, "loss": 0.1, "step": 4566 }, { "epoch": 4.63, "learning_rate": 1.618605341225677e-05, "loss": 0.125, "step": 4567 }, { "epoch": 4.64, "learning_rate": 1.618433087494402e-05, "loss": 0.1392, "step": 4568 }, { "epoch": 4.64, "learning_rate": 1.6182608040437663e-05, "loss": 0.1329, "step": 4569 }, { "epoch": 4.64, "learning_rate": 1.6180884908820483e-05, "loss": 0.1342, "step": 4570 }, { "epoch": 4.64, "learning_rate": 1.6179161480175296e-05, "loss": 0.1237, "step": 4571 }, { "epoch": 4.64, "learning_rate": 1.617743775458491e-05, "loss": 0.1267, "step": 4572 }, { "epoch": 4.64, "learning_rate": 1.6175713732132176e-05, "loss": 0.1213, "step": 4573 }, { "epoch": 4.64, "learning_rate": 1.6173989412899937e-05, "loss": 0.1294, "step": 4574 }, { "epoch": 4.64, "learning_rate": 1.617226479697105e-05, "loss": 0.1151, "step": 4575 }, { "epoch": 4.64, "learning_rate": 1.61705398844284e-05, "loss": 0.1384, "step": 4576 }, { "epoch": 4.64, "learning_rate": 1.6168814675354875e-05, "loss": 0.1362, "step": 4577 }, { "epoch": 4.65, "learning_rate": 1.616708916983339e-05, "loss": 0.1466, "step": 4578 }, { "epoch": 4.65, "learning_rate": 1.616536336794685e-05, "loss": 0.1101, "step": 4579 }, { "epoch": 4.65, "learning_rate": 1.6163637269778208e-05, "loss": 0.1272, "step": 4580 }, { "epoch": 4.65, "learning_rate": 1.61619108754104e-05, "loss": 0.1163, "step": 4581 }, { "epoch": 4.65, "learning_rate": 1.6160184184926396e-05, "loss": 0.1233, "step": 4582 }, { "epoch": 4.65, "learning_rate": 1.6158457198409174e-05, "loss": 0.1479, "step": 4583 }, { "epoch": 4.65, "learning_rate": 1.6156729915941724e-05, "loss": 0.137, "step": 4584 }, { "epoch": 4.65, "learning_rate": 1.615500233760705e-05, "loss": 0.1327, "step": 4585 }, { "epoch": 4.65, "learning_rate": 1.6153274463488174e-05, "loss": 0.1209, "step": 4586 }, { "epoch": 4.65, "learning_rate": 1.615154629366813e-05, "loss": 0.11, "step": 4587 }, { "epoch": 4.66, "learning_rate": 1.614981782822997e-05, "loss": 0.1115, "step": 4588 }, { "epoch": 4.66, "learning_rate": 1.6148089067256756e-05, "loss": 0.1461, "step": 4589 }, { "epoch": 4.66, "learning_rate": 1.614636001083156e-05, "loss": 0.1478, "step": 4590 }, { "epoch": 4.66, "learning_rate": 1.6144630659037478e-05, "loss": 0.1201, "step": 4591 }, { "epoch": 4.66, "learning_rate": 1.6142901011957615e-05, "loss": 0.121, "step": 4592 }, { "epoch": 4.66, "learning_rate": 1.6141171069675095e-05, "loss": 0.1242, "step": 4593 }, { "epoch": 4.66, "learning_rate": 1.6139440832273042e-05, "loss": 0.1401, "step": 4594 }, { "epoch": 4.66, "learning_rate": 1.613771029983461e-05, "loss": 0.1406, "step": 4595 }, { "epoch": 4.66, "learning_rate": 1.6135979472442965e-05, "loss": 0.1283, "step": 4596 }, { "epoch": 4.66, "learning_rate": 1.6134248350181273e-05, "loss": 0.129, "step": 4597 }, { "epoch": 4.67, "learning_rate": 1.6132516933132737e-05, "loss": 0.1289, "step": 4598 }, { "epoch": 4.67, "learning_rate": 1.6130785221380555e-05, "loss": 0.1248, "step": 4599 }, { "epoch": 4.67, "learning_rate": 1.6129053215007942e-05, "loss": 0.1081, "step": 4600 }, { "epoch": 4.67, "learning_rate": 1.6127320914098138e-05, "loss": 0.142, "step": 4601 }, { "epoch": 4.67, "learning_rate": 1.6125588318734392e-05, "loss": 0.1477, "step": 4602 }, { "epoch": 4.67, "learning_rate": 1.612385542899996e-05, "loss": 0.1483, "step": 4603 }, { "epoch": 4.67, "learning_rate": 1.6122122244978116e-05, "loss": 0.142, "step": 4604 }, { "epoch": 4.67, "learning_rate": 1.6120388766752153e-05, "loss": 0.1273, "step": 4605 }, { "epoch": 4.67, "learning_rate": 1.6118654994405377e-05, "loss": 0.1107, "step": 4606 }, { "epoch": 4.67, "learning_rate": 1.6116920928021104e-05, "loss": 0.117, "step": 4607 }, { "epoch": 4.68, "learning_rate": 1.6115186567682663e-05, "loss": 0.119, "step": 4608 }, { "epoch": 4.68, "learning_rate": 1.611345191347341e-05, "loss": 0.1187, "step": 4609 }, { "epoch": 4.68, "learning_rate": 1.6111716965476692e-05, "loss": 0.1177, "step": 4610 }, { "epoch": 4.68, "learning_rate": 1.6109981723775892e-05, "loss": 0.1215, "step": 4611 }, { "epoch": 4.68, "learning_rate": 1.6108246188454394e-05, "loss": 0.1487, "step": 4612 }, { "epoch": 4.68, "learning_rate": 1.610651035959561e-05, "loss": 0.1252, "step": 4613 }, { "epoch": 4.68, "learning_rate": 1.6104774237282942e-05, "loss": 0.1214, "step": 4614 }, { "epoch": 4.68, "learning_rate": 1.6103037821599837e-05, "loss": 0.1267, "step": 4615 }, { "epoch": 4.68, "learning_rate": 1.6101301112629724e-05, "loss": 0.1242, "step": 4616 }, { "epoch": 4.68, "learning_rate": 1.6099564110456077e-05, "loss": 0.1285, "step": 4617 }, { "epoch": 4.69, "learning_rate": 1.609782681516236e-05, "loss": 0.1361, "step": 4618 }, { "epoch": 4.69, "learning_rate": 1.6096089226832065e-05, "loss": 0.1381, "step": 4619 }, { "epoch": 4.69, "learning_rate": 1.609435134554869e-05, "loss": 0.1096, "step": 4620 }, { "epoch": 4.69, "learning_rate": 1.609261317139575e-05, "loss": 0.1261, "step": 4621 }, { "epoch": 4.69, "learning_rate": 1.6090874704456774e-05, "loss": 0.1109, "step": 4622 }, { "epoch": 4.69, "learning_rate": 1.6089135944815312e-05, "loss": 0.1388, "step": 4623 }, { "epoch": 4.69, "learning_rate": 1.6087396892554917e-05, "loss": 0.1173, "step": 4624 }, { "epoch": 4.69, "learning_rate": 1.6085657547759158e-05, "loss": 0.1263, "step": 4625 }, { "epoch": 4.69, "learning_rate": 1.608391791051163e-05, "loss": 0.1009, "step": 4626 }, { "epoch": 4.7, "learning_rate": 1.608217798089592e-05, "loss": 0.1391, "step": 4627 }, { "epoch": 4.7, "learning_rate": 1.6080437758995654e-05, "loss": 0.1332, "step": 4628 }, { "epoch": 4.7, "learning_rate": 1.607869724489445e-05, "loss": 0.1376, "step": 4629 }, { "epoch": 4.7, "learning_rate": 1.607695643867596e-05, "loss": 0.1401, "step": 4630 }, { "epoch": 4.7, "learning_rate": 1.6075215340423828e-05, "loss": 0.1212, "step": 4631 }, { "epoch": 4.7, "learning_rate": 1.6073473950221737e-05, "loss": 0.1232, "step": 4632 }, { "epoch": 4.7, "learning_rate": 1.607173226815336e-05, "loss": 0.1339, "step": 4633 }, { "epoch": 4.7, "learning_rate": 1.60699902943024e-05, "loss": 0.1309, "step": 4634 }, { "epoch": 4.7, "learning_rate": 1.6068248028752572e-05, "loss": 0.108, "step": 4635 }, { "epoch": 4.7, "learning_rate": 1.6066505471587596e-05, "loss": 0.1176, "step": 4636 }, { "epoch": 4.71, "learning_rate": 1.6064762622891216e-05, "loss": 0.1384, "step": 4637 }, { "epoch": 4.71, "learning_rate": 1.6063019482747188e-05, "loss": 0.1285, "step": 4638 }, { "epoch": 4.71, "learning_rate": 1.6061276051239275e-05, "loss": 0.1327, "step": 4639 }, { "epoch": 4.71, "learning_rate": 1.6059532328451264e-05, "loss": 0.1283, "step": 4640 }, { "epoch": 4.71, "learning_rate": 1.6057788314466946e-05, "loss": 0.141, "step": 4641 }, { "epoch": 4.71, "learning_rate": 1.6056044009370136e-05, "loss": 0.1368, "step": 4642 }, { "epoch": 4.71, "learning_rate": 1.6054299413244655e-05, "loss": 0.1343, "step": 4643 }, { "epoch": 4.71, "learning_rate": 1.6052554526174343e-05, "loss": 0.1302, "step": 4644 }, { "epoch": 4.71, "learning_rate": 1.605080934824305e-05, "loss": 0.1262, "step": 4645 }, { "epoch": 4.71, "learning_rate": 1.604906387953465e-05, "loss": 0.1191, "step": 4646 }, { "epoch": 4.72, "learning_rate": 1.604731812013301e-05, "loss": 0.1396, "step": 4647 }, { "epoch": 4.72, "learning_rate": 1.6045572070122034e-05, "loss": 0.123, "step": 4648 }, { "epoch": 4.72, "learning_rate": 1.6043825729585628e-05, "loss": 0.1229, "step": 4649 }, { "epoch": 4.72, "learning_rate": 1.6042079098607706e-05, "loss": 0.1284, "step": 4650 }, { "epoch": 4.72, "learning_rate": 1.6040332177272216e-05, "loss": 0.1211, "step": 4651 }, { "epoch": 4.72, "learning_rate": 1.6038584965663102e-05, "loss": 0.1414, "step": 4652 }, { "epoch": 4.72, "learning_rate": 1.603683746386433e-05, "loss": 0.1449, "step": 4653 }, { "epoch": 4.72, "learning_rate": 1.603508967195987e-05, "loss": 0.1114, "step": 4654 }, { "epoch": 4.72, "learning_rate": 1.6033341590033725e-05, "loss": 0.1416, "step": 4655 }, { "epoch": 4.72, "learning_rate": 1.60315932181699e-05, "loss": 0.1257, "step": 4656 }, { "epoch": 4.73, "learning_rate": 1.6029844556452403e-05, "loss": 0.1204, "step": 4657 }, { "epoch": 4.73, "learning_rate": 1.6028095604965276e-05, "loss": 0.1285, "step": 4658 }, { "epoch": 4.73, "learning_rate": 1.6026346363792565e-05, "loss": 0.127, "step": 4659 }, { "epoch": 4.73, "learning_rate": 1.6024596833018335e-05, "loss": 0.1298, "step": 4660 }, { "epoch": 4.73, "learning_rate": 1.6022847012726653e-05, "loss": 0.124, "step": 4661 }, { "epoch": 4.73, "learning_rate": 1.6021096903001615e-05, "loss": 0.1419, "step": 4662 }, { "epoch": 4.73, "learning_rate": 1.601934650392732e-05, "loss": 0.1316, "step": 4663 }, { "epoch": 4.73, "learning_rate": 1.6017595815587893e-05, "loss": 0.1118, "step": 4664 }, { "epoch": 4.73, "learning_rate": 1.6015844838067453e-05, "loss": 0.1377, "step": 4665 }, { "epoch": 4.73, "learning_rate": 1.601409357145015e-05, "loss": 0.117, "step": 4666 }, { "epoch": 4.74, "learning_rate": 1.601234201582015e-05, "loss": 0.1404, "step": 4667 }, { "epoch": 4.74, "learning_rate": 1.6010590171261617e-05, "loss": 0.1359, "step": 4668 }, { "epoch": 4.74, "learning_rate": 1.6008838037858736e-05, "loss": 0.1153, "step": 4669 }, { "epoch": 4.74, "learning_rate": 1.6007085615695714e-05, "loss": 0.1152, "step": 4670 }, { "epoch": 4.74, "learning_rate": 1.600533290485676e-05, "loss": 0.124, "step": 4671 }, { "epoch": 4.74, "learning_rate": 1.6003579905426104e-05, "loss": 0.1327, "step": 4672 }, { "epoch": 4.74, "learning_rate": 1.6001826617487992e-05, "loss": 0.1284, "step": 4673 }, { "epoch": 4.74, "learning_rate": 1.6000073041126674e-05, "loss": 0.1175, "step": 4674 }, { "epoch": 4.74, "learning_rate": 1.599831917642642e-05, "loss": 0.1181, "step": 4675 }, { "epoch": 4.74, "learning_rate": 1.5996565023471516e-05, "loss": 0.1371, "step": 4676 }, { "epoch": 4.75, "learning_rate": 1.599481058234626e-05, "loss": 0.1231, "step": 4677 }, { "epoch": 4.75, "learning_rate": 1.5993055853134964e-05, "loss": 0.1339, "step": 4678 }, { "epoch": 4.75, "learning_rate": 1.5991300835921946e-05, "loss": 0.1145, "step": 4679 }, { "epoch": 4.75, "learning_rate": 1.5989545530791553e-05, "loss": 0.1356, "step": 4680 }, { "epoch": 4.75, "learning_rate": 1.5987789937828133e-05, "loss": 0.1215, "step": 4681 }, { "epoch": 4.75, "learning_rate": 1.5986034057116057e-05, "loss": 0.151, "step": 4682 }, { "epoch": 4.75, "learning_rate": 1.59842778887397e-05, "loss": 0.12, "step": 4683 }, { "epoch": 4.75, "learning_rate": 1.5982521432783464e-05, "loss": 0.1167, "step": 4684 }, { "epoch": 4.75, "learning_rate": 1.5980764689331746e-05, "loss": 0.1367, "step": 4685 }, { "epoch": 4.75, "learning_rate": 1.5979007658468976e-05, "loss": 0.1436, "step": 4686 }, { "epoch": 4.76, "learning_rate": 1.597725034027959e-05, "loss": 0.134, "step": 4687 }, { "epoch": 4.76, "learning_rate": 1.5975492734848032e-05, "loss": 0.1252, "step": 4688 }, { "epoch": 4.76, "learning_rate": 1.597373484225877e-05, "loss": 0.1273, "step": 4689 }, { "epoch": 4.76, "learning_rate": 1.5971976662596283e-05, "loss": 0.1353, "step": 4690 }, { "epoch": 4.76, "learning_rate": 1.5970218195945054e-05, "loss": 0.1298, "step": 4691 }, { "epoch": 4.76, "learning_rate": 1.5968459442389593e-05, "loss": 0.142, "step": 4692 }, { "epoch": 4.76, "learning_rate": 1.5966700402014418e-05, "loss": 0.1418, "step": 4693 }, { "epoch": 4.76, "learning_rate": 1.5964941074904064e-05, "loss": 0.1265, "step": 4694 }, { "epoch": 4.76, "learning_rate": 1.5963181461143074e-05, "loss": 0.1265, "step": 4695 }, { "epoch": 4.77, "learning_rate": 1.5961421560816005e-05, "loss": 0.1415, "step": 4696 }, { "epoch": 4.77, "learning_rate": 1.5959661374007434e-05, "loss": 0.1226, "step": 4697 }, { "epoch": 4.77, "learning_rate": 1.595790090080195e-05, "loss": 0.1264, "step": 4698 }, { "epoch": 4.77, "learning_rate": 1.595614014128415e-05, "loss": 0.1343, "step": 4699 }, { "epoch": 4.77, "learning_rate": 1.5954379095538652e-05, "loss": 0.1354, "step": 4700 }, { "epoch": 4.77, "learning_rate": 1.5952617763650082e-05, "loss": 0.1247, "step": 4701 }, { "epoch": 4.77, "learning_rate": 1.5950856145703087e-05, "loss": 0.1337, "step": 4702 }, { "epoch": 4.77, "learning_rate": 1.5949094241782315e-05, "loss": 0.1141, "step": 4703 }, { "epoch": 4.77, "learning_rate": 1.5947332051972447e-05, "loss": 0.1241, "step": 4704 }, { "epoch": 4.77, "learning_rate": 1.594556957635816e-05, "loss": 0.1441, "step": 4705 }, { "epoch": 4.78, "learning_rate": 1.594380681502415e-05, "loss": 0.1113, "step": 4706 }, { "epoch": 4.78, "learning_rate": 1.594204376805513e-05, "loss": 0.1191, "step": 4707 }, { "epoch": 4.78, "learning_rate": 1.5940280435535823e-05, "loss": 0.1539, "step": 4708 }, { "epoch": 4.78, "learning_rate": 1.5938516817550973e-05, "loss": 0.1167, "step": 4709 }, { "epoch": 4.78, "learning_rate": 1.5936752914185328e-05, "loss": 0.1279, "step": 4710 }, { "epoch": 4.78, "learning_rate": 1.5934988725523655e-05, "loss": 0.1173, "step": 4711 }, { "epoch": 4.78, "learning_rate": 1.593322425165073e-05, "loss": 0.1255, "step": 4712 }, { "epoch": 4.78, "learning_rate": 1.5931459492651353e-05, "loss": 0.1263, "step": 4713 }, { "epoch": 4.78, "learning_rate": 1.5929694448610327e-05, "loss": 0.1214, "step": 4714 }, { "epoch": 4.78, "learning_rate": 1.5927929119612477e-05, "loss": 0.1264, "step": 4715 }, { "epoch": 4.79, "learning_rate": 1.592616350574263e-05, "loss": 0.1418, "step": 4716 }, { "epoch": 4.79, "learning_rate": 1.5924397607085638e-05, "loss": 0.1452, "step": 4717 }, { "epoch": 4.79, "learning_rate": 1.5922631423726365e-05, "loss": 0.1393, "step": 4718 }, { "epoch": 4.79, "learning_rate": 1.5920864955749684e-05, "loss": 0.1545, "step": 4719 }, { "epoch": 4.79, "learning_rate": 1.5919098203240486e-05, "loss": 0.1186, "step": 4720 }, { "epoch": 4.79, "learning_rate": 1.5917331166283673e-05, "loss": 0.1249, "step": 4721 }, { "epoch": 4.79, "learning_rate": 1.591556384496416e-05, "loss": 0.137, "step": 4722 }, { "epoch": 4.79, "learning_rate": 1.5913796239366875e-05, "loss": 0.1238, "step": 4723 }, { "epoch": 4.79, "learning_rate": 1.5912028349576775e-05, "loss": 0.1307, "step": 4724 }, { "epoch": 4.79, "learning_rate": 1.5910260175678797e-05, "loss": 0.1127, "step": 4725 }, { "epoch": 4.8, "learning_rate": 1.590849171775793e-05, "loss": 0.111, "step": 4726 }, { "epoch": 4.8, "learning_rate": 1.5906722975899152e-05, "loss": 0.1318, "step": 4727 }, { "epoch": 4.8, "learning_rate": 1.5904953950187458e-05, "loss": 0.1348, "step": 4728 }, { "epoch": 4.8, "learning_rate": 1.5903184640707864e-05, "loss": 0.1329, "step": 4729 }, { "epoch": 4.8, "learning_rate": 1.59014150475454e-05, "loss": 0.1436, "step": 4730 }, { "epoch": 4.8, "learning_rate": 1.5899645170785097e-05, "loss": 0.1186, "step": 4731 }, { "epoch": 4.8, "learning_rate": 1.5897875010512015e-05, "loss": 0.1153, "step": 4732 }, { "epoch": 4.8, "learning_rate": 1.5896104566811212e-05, "loss": 0.1266, "step": 4733 }, { "epoch": 4.8, "learning_rate": 1.5894333839767778e-05, "loss": 0.1521, "step": 4734 }, { "epoch": 4.8, "learning_rate": 1.5892562829466802e-05, "loss": 0.1134, "step": 4735 }, { "epoch": 4.81, "learning_rate": 1.589079153599339e-05, "loss": 0.1229, "step": 4736 }, { "epoch": 4.81, "learning_rate": 1.588901995943267e-05, "loss": 0.1469, "step": 4737 }, { "epoch": 4.81, "learning_rate": 1.588724809986977e-05, "loss": 0.1355, "step": 4738 }, { "epoch": 4.81, "learning_rate": 1.5885475957389838e-05, "loss": 0.1289, "step": 4739 }, { "epoch": 4.81, "learning_rate": 1.588370353207804e-05, "loss": 0.141, "step": 4740 }, { "epoch": 4.81, "learning_rate": 1.588193082401955e-05, "loss": 0.1464, "step": 4741 }, { "epoch": 4.81, "learning_rate": 1.5880157833299555e-05, "loss": 0.1469, "step": 4742 }, { "epoch": 4.81, "learning_rate": 1.5878384560003263e-05, "loss": 0.1182, "step": 4743 }, { "epoch": 4.81, "learning_rate": 1.5876611004215883e-05, "loss": 0.114, "step": 4744 }, { "epoch": 4.81, "learning_rate": 1.587483716602265e-05, "loss": 0.1407, "step": 4745 }, { "epoch": 4.82, "learning_rate": 1.5873063045508804e-05, "loss": 0.1455, "step": 4746 }, { "epoch": 4.82, "learning_rate": 1.5871288642759607e-05, "loss": 0.1301, "step": 4747 }, { "epoch": 4.82, "learning_rate": 1.5869513957860324e-05, "loss": 0.1253, "step": 4748 }, { "epoch": 4.82, "learning_rate": 1.5867738990896243e-05, "loss": 0.1131, "step": 4749 }, { "epoch": 4.82, "learning_rate": 1.586596374195266e-05, "loss": 0.1276, "step": 4750 }, { "epoch": 4.82, "learning_rate": 1.5864188211114885e-05, "loss": 0.1177, "step": 4751 }, { "epoch": 4.82, "learning_rate": 1.5862412398468248e-05, "loss": 0.1273, "step": 4752 }, { "epoch": 4.82, "learning_rate": 1.586063630409808e-05, "loss": 0.1157, "step": 4753 }, { "epoch": 4.82, "learning_rate": 1.5858859928089737e-05, "loss": 0.1546, "step": 4754 }, { "epoch": 4.82, "learning_rate": 1.5857083270528582e-05, "loss": 0.1182, "step": 4755 }, { "epoch": 4.83, "learning_rate": 1.5855306331499998e-05, "loss": 0.1403, "step": 4756 }, { "epoch": 4.83, "learning_rate": 1.5853529111089377e-05, "loss": 0.1199, "step": 4757 }, { "epoch": 4.83, "learning_rate": 1.5851751609382118e-05, "loss": 0.1244, "step": 4758 }, { "epoch": 4.83, "learning_rate": 1.5849973826463645e-05, "loss": 0.1198, "step": 4759 }, { "epoch": 4.83, "learning_rate": 1.5848195762419395e-05, "loss": 0.1264, "step": 4760 }, { "epoch": 4.83, "learning_rate": 1.5846417417334808e-05, "loss": 0.1274, "step": 4761 }, { "epoch": 4.83, "learning_rate": 1.5844638791295348e-05, "loss": 0.1312, "step": 4762 }, { "epoch": 4.83, "learning_rate": 1.584285988438649e-05, "loss": 0.1331, "step": 4763 }, { "epoch": 4.83, "learning_rate": 1.584108069669371e-05, "loss": 0.1338, "step": 4764 }, { "epoch": 4.84, "learning_rate": 1.5839301228302524e-05, "loss": 0.132, "step": 4765 }, { "epoch": 4.84, "learning_rate": 1.5837521479298434e-05, "loss": 0.1392, "step": 4766 }, { "epoch": 4.84, "learning_rate": 1.5835741449766974e-05, "loss": 0.126, "step": 4767 }, { "epoch": 4.84, "learning_rate": 1.5833961139793682e-05, "loss": 0.136, "step": 4768 }, { "epoch": 4.84, "learning_rate": 1.5832180549464116e-05, "loss": 0.1218, "step": 4769 }, { "epoch": 4.84, "learning_rate": 1.583039967886384e-05, "loss": 0.1092, "step": 4770 }, { "epoch": 4.84, "learning_rate": 1.5828618528078434e-05, "loss": 0.1251, "step": 4771 }, { "epoch": 4.84, "learning_rate": 1.58268370971935e-05, "loss": 0.1242, "step": 4772 }, { "epoch": 4.84, "learning_rate": 1.5825055386294636e-05, "loss": 0.1239, "step": 4773 }, { "epoch": 4.84, "learning_rate": 1.5823273395467477e-05, "loss": 0.127, "step": 4774 }, { "epoch": 4.85, "learning_rate": 1.5821491124797643e-05, "loss": 0.1324, "step": 4775 }, { "epoch": 4.85, "learning_rate": 1.5819708574370794e-05, "loss": 0.1565, "step": 4776 }, { "epoch": 4.85, "learning_rate": 1.581792574427259e-05, "loss": 0.1226, "step": 4777 }, { "epoch": 4.85, "learning_rate": 1.58161426345887e-05, "loss": 0.1158, "step": 4778 }, { "epoch": 4.85, "learning_rate": 1.5814359245404818e-05, "loss": 0.1298, "step": 4779 }, { "epoch": 4.85, "learning_rate": 1.581257557680665e-05, "loss": 0.1159, "step": 4780 }, { "epoch": 4.85, "learning_rate": 1.5810791628879903e-05, "loss": 0.1357, "step": 4781 }, { "epoch": 4.85, "learning_rate": 1.5809007401710318e-05, "loss": 0.1424, "step": 4782 }, { "epoch": 4.85, "learning_rate": 1.5807222895383624e-05, "loss": 0.1332, "step": 4783 }, { "epoch": 4.85, "learning_rate": 1.580543810998559e-05, "loss": 0.1454, "step": 4784 }, { "epoch": 4.86, "learning_rate": 1.5803653045601974e-05, "loss": 0.1235, "step": 4785 }, { "epoch": 4.86, "learning_rate": 1.5801867702318566e-05, "loss": 0.1336, "step": 4786 }, { "epoch": 4.86, "learning_rate": 1.580008208022116e-05, "loss": 0.1374, "step": 4787 }, { "epoch": 4.86, "learning_rate": 1.5798296179395568e-05, "loss": 0.1299, "step": 4788 }, { "epoch": 4.86, "learning_rate": 1.579650999992761e-05, "loss": 0.1502, "step": 4789 }, { "epoch": 4.86, "learning_rate": 1.5794723541903124e-05, "loss": 0.1452, "step": 4790 }, { "epoch": 4.86, "learning_rate": 1.579293680540796e-05, "loss": 0.1418, "step": 4791 }, { "epoch": 4.86, "learning_rate": 1.579114979052798e-05, "loss": 0.144, "step": 4792 }, { "epoch": 4.86, "learning_rate": 1.5789362497349064e-05, "loss": 0.1422, "step": 4793 }, { "epoch": 4.86, "learning_rate": 1.5787574925957095e-05, "loss": 0.1378, "step": 4794 }, { "epoch": 4.87, "learning_rate": 1.5785787076437983e-05, "loss": 0.1264, "step": 4795 }, { "epoch": 4.87, "learning_rate": 1.5783998948877642e-05, "loss": 0.1427, "step": 4796 }, { "epoch": 4.87, "learning_rate": 1.5782210543362003e-05, "loss": 0.1448, "step": 4797 }, { "epoch": 4.87, "learning_rate": 1.578042185997701e-05, "loss": 0.1328, "step": 4798 }, { "epoch": 4.87, "learning_rate": 1.5778632898808618e-05, "loss": 0.1262, "step": 4799 }, { "epoch": 4.87, "learning_rate": 1.57768436599428e-05, "loss": 0.123, "step": 4800 }, { "epoch": 4.87, "learning_rate": 1.5775054143465533e-05, "loss": 0.1299, "step": 4801 }, { "epoch": 4.87, "learning_rate": 1.577326434946282e-05, "loss": 0.1241, "step": 4802 }, { "epoch": 4.87, "learning_rate": 1.577147427802067e-05, "loss": 0.1201, "step": 4803 }, { "epoch": 4.87, "learning_rate": 1.5769683929225107e-05, "loss": 0.1289, "step": 4804 }, { "epoch": 4.88, "learning_rate": 1.5767893303162165e-05, "loss": 0.1197, "step": 4805 }, { "epoch": 4.88, "learning_rate": 1.57661023999179e-05, "loss": 0.129, "step": 4806 }, { "epoch": 4.88, "learning_rate": 1.5764311219578368e-05, "loss": 0.1186, "step": 4807 }, { "epoch": 4.88, "learning_rate": 1.5762519762229647e-05, "loss": 0.1171, "step": 4808 }, { "epoch": 4.88, "learning_rate": 1.5760728027957833e-05, "loss": 0.1186, "step": 4809 }, { "epoch": 4.88, "learning_rate": 1.5758936016849025e-05, "loss": 0.1191, "step": 4810 }, { "epoch": 4.88, "learning_rate": 1.575714372898934e-05, "loss": 0.14, "step": 4811 }, { "epoch": 4.88, "learning_rate": 1.575535116446491e-05, "loss": 0.1292, "step": 4812 }, { "epoch": 4.88, "learning_rate": 1.575355832336187e-05, "loss": 0.1353, "step": 4813 }, { "epoch": 4.88, "learning_rate": 1.575176520576639e-05, "loss": 0.1279, "step": 4814 }, { "epoch": 4.89, "learning_rate": 1.5749971811764634e-05, "loss": 0.124, "step": 4815 }, { "epoch": 4.89, "learning_rate": 1.5748178141442782e-05, "loss": 0.1207, "step": 4816 }, { "epoch": 4.89, "learning_rate": 1.5746384194887033e-05, "loss": 0.1449, "step": 4817 }, { "epoch": 4.89, "learning_rate": 1.5744589972183598e-05, "loss": 0.1232, "step": 4818 }, { "epoch": 4.89, "learning_rate": 1.57427954734187e-05, "loss": 0.1308, "step": 4819 }, { "epoch": 4.89, "learning_rate": 1.574100069867857e-05, "loss": 0.1211, "step": 4820 }, { "epoch": 4.89, "learning_rate": 1.5739205648049462e-05, "loss": 0.1389, "step": 4821 }, { "epoch": 4.89, "learning_rate": 1.573741032161764e-05, "loss": 0.1113, "step": 4822 }, { "epoch": 4.89, "learning_rate": 1.5735614719469376e-05, "loss": 0.1414, "step": 4823 }, { "epoch": 4.89, "learning_rate": 1.5733818841690966e-05, "loss": 0.1168, "step": 4824 }, { "epoch": 4.9, "learning_rate": 1.57320226883687e-05, "loss": 0.1256, "step": 4825 }, { "epoch": 4.9, "learning_rate": 1.573022625958891e-05, "loss": 0.121, "step": 4826 }, { "epoch": 4.9, "learning_rate": 1.572842955543791e-05, "loss": 0.1287, "step": 4827 }, { "epoch": 4.9, "learning_rate": 1.5726632576002056e-05, "loss": 0.1377, "step": 4828 }, { "epoch": 4.9, "learning_rate": 1.572483532136769e-05, "loss": 0.143, "step": 4829 }, { "epoch": 4.9, "learning_rate": 1.5723037791621193e-05, "loss": 0.1339, "step": 4830 }, { "epoch": 4.9, "learning_rate": 1.5721239986848937e-05, "loss": 0.1301, "step": 4831 }, { "epoch": 4.9, "learning_rate": 1.5719441907137324e-05, "loss": 0.1233, "step": 4832 }, { "epoch": 4.9, "learning_rate": 1.5717643552572756e-05, "loss": 0.1167, "step": 4833 }, { "epoch": 4.91, "learning_rate": 1.571584492324166e-05, "loss": 0.1086, "step": 4834 }, { "epoch": 4.91, "learning_rate": 1.571404601923047e-05, "loss": 0.1396, "step": 4835 }, { "epoch": 4.91, "learning_rate": 1.5712246840625632e-05, "loss": 0.1475, "step": 4836 }, { "epoch": 4.91, "learning_rate": 1.571044738751361e-05, "loss": 0.1224, "step": 4837 }, { "epoch": 4.91, "learning_rate": 1.570864765998087e-05, "loss": 0.1165, "step": 4838 }, { "epoch": 4.91, "learning_rate": 1.570684765811391e-05, "loss": 0.1448, "step": 4839 }, { "epoch": 4.91, "learning_rate": 1.5705047381999226e-05, "loss": 0.122, "step": 4840 }, { "epoch": 4.91, "learning_rate": 1.5703246831723333e-05, "loss": 0.1259, "step": 4841 }, { "epoch": 4.91, "learning_rate": 1.570144600737276e-05, "loss": 0.1217, "step": 4842 }, { "epoch": 4.91, "learning_rate": 1.5699644909034043e-05, "loss": 0.1358, "step": 4843 }, { "epoch": 4.92, "learning_rate": 1.569784353679374e-05, "loss": 0.1297, "step": 4844 }, { "epoch": 4.92, "learning_rate": 1.5696041890738406e-05, "loss": 0.1254, "step": 4845 }, { "epoch": 4.92, "learning_rate": 1.5694239970954636e-05, "loss": 0.1252, "step": 4846 }, { "epoch": 4.92, "learning_rate": 1.569243777752902e-05, "loss": 0.157, "step": 4847 }, { "epoch": 4.92, "learning_rate": 1.569063531054815e-05, "loss": 0.1279, "step": 4848 }, { "epoch": 4.92, "learning_rate": 1.5688832570098665e-05, "loss": 0.1512, "step": 4849 }, { "epoch": 4.92, "learning_rate": 1.5687029556267184e-05, "loss": 0.1271, "step": 4850 }, { "epoch": 4.92, "learning_rate": 1.5685226269140356e-05, "loss": 0.138, "step": 4851 }, { "epoch": 4.92, "learning_rate": 1.5683422708804844e-05, "loss": 0.1592, "step": 4852 }, { "epoch": 4.92, "learning_rate": 1.5681618875347312e-05, "loss": 0.1246, "step": 4853 }, { "epoch": 4.93, "learning_rate": 1.5679814768854453e-05, "loss": 0.1257, "step": 4854 }, { "epoch": 4.93, "learning_rate": 1.5678010389412954e-05, "loss": 0.1263, "step": 4855 }, { "epoch": 4.93, "learning_rate": 1.5676205737109536e-05, "loss": 0.1414, "step": 4856 }, { "epoch": 4.93, "learning_rate": 1.567440081203092e-05, "loss": 0.1319, "step": 4857 }, { "epoch": 4.93, "learning_rate": 1.567259561426384e-05, "loss": 0.1192, "step": 4858 }, { "epoch": 4.93, "learning_rate": 1.5670790143895053e-05, "loss": 0.1262, "step": 4859 }, { "epoch": 4.93, "learning_rate": 1.5668984401011318e-05, "loss": 0.1187, "step": 4860 }, { "epoch": 4.93, "learning_rate": 1.566717838569941e-05, "loss": 0.1213, "step": 4861 }, { "epoch": 4.93, "learning_rate": 1.566537209804613e-05, "loss": 0.141, "step": 4862 }, { "epoch": 4.93, "learning_rate": 1.5663565538138264e-05, "loss": 0.1309, "step": 4863 }, { "epoch": 4.94, "learning_rate": 1.566175870606264e-05, "loss": 0.1231, "step": 4864 }, { "epoch": 4.94, "learning_rate": 1.565995160190608e-05, "loss": 0.1234, "step": 4865 }, { "epoch": 4.94, "learning_rate": 1.565814422575543e-05, "loss": 0.1197, "step": 4866 }, { "epoch": 4.94, "learning_rate": 1.5656336577697544e-05, "loss": 0.1465, "step": 4867 }, { "epoch": 4.94, "learning_rate": 1.5654528657819296e-05, "loss": 0.129, "step": 4868 }, { "epoch": 4.94, "learning_rate": 1.5652720466207556e-05, "loss": 0.1273, "step": 4869 }, { "epoch": 4.94, "learning_rate": 1.5650912002949222e-05, "loss": 0.1206, "step": 4870 }, { "epoch": 4.94, "learning_rate": 1.5649103268131204e-05, "loss": 0.1318, "step": 4871 }, { "epoch": 4.94, "learning_rate": 1.5647294261840424e-05, "loss": 0.136, "step": 4872 }, { "epoch": 4.94, "learning_rate": 1.564548498416381e-05, "loss": 0.121, "step": 4873 }, { "epoch": 4.95, "learning_rate": 1.5643675435188316e-05, "loss": 0.1294, "step": 4874 }, { "epoch": 4.95, "learning_rate": 1.564186561500089e-05, "loss": 0.1157, "step": 4875 }, { "epoch": 4.95, "learning_rate": 1.5640055523688518e-05, "loss": 0.1429, "step": 4876 }, { "epoch": 4.95, "learning_rate": 1.5638245161338176e-05, "loss": 0.127, "step": 4877 }, { "epoch": 4.95, "learning_rate": 1.5636434528036864e-05, "loss": 0.136, "step": 4878 }, { "epoch": 4.95, "learning_rate": 1.56346236238716e-05, "loss": 0.1185, "step": 4879 }, { "epoch": 4.95, "learning_rate": 1.5632812448929403e-05, "loss": 0.1429, "step": 4880 }, { "epoch": 4.95, "learning_rate": 1.563100100329731e-05, "loss": 0.1222, "step": 4881 }, { "epoch": 4.95, "learning_rate": 1.5629189287062372e-05, "loss": 0.1232, "step": 4882 }, { "epoch": 4.95, "learning_rate": 1.562737730031165e-05, "loss": 0.1299, "step": 4883 }, { "epoch": 4.96, "learning_rate": 1.562556504313223e-05, "loss": 0.1321, "step": 4884 }, { "epoch": 4.96, "learning_rate": 1.5623752515611198e-05, "loss": 0.1202, "step": 4885 }, { "epoch": 4.96, "learning_rate": 1.562193971783565e-05, "loss": 0.1264, "step": 4886 }, { "epoch": 4.96, "learning_rate": 1.562012664989271e-05, "loss": 0.1345, "step": 4887 }, { "epoch": 4.96, "learning_rate": 1.5618313311869503e-05, "loss": 0.1325, "step": 4888 }, { "epoch": 4.96, "learning_rate": 1.5616499703853167e-05, "loss": 0.1331, "step": 4889 }, { "epoch": 4.96, "learning_rate": 1.5614685825930863e-05, "loss": 0.1277, "step": 4890 }, { "epoch": 4.96, "learning_rate": 1.5612871678189758e-05, "loss": 0.1163, "step": 4891 }, { "epoch": 4.96, "learning_rate": 1.5611057260717025e-05, "loss": 0.1297, "step": 4892 }, { "epoch": 4.96, "learning_rate": 1.5609242573599868e-05, "loss": 0.1123, "step": 4893 }, { "epoch": 4.97, "learning_rate": 1.5607427616925486e-05, "loss": 0.1184, "step": 4894 }, { "epoch": 4.97, "learning_rate": 1.5605612390781098e-05, "loss": 0.1231, "step": 4895 }, { "epoch": 4.97, "learning_rate": 1.560379689525394e-05, "loss": 0.1245, "step": 4896 }, { "epoch": 4.97, "learning_rate": 1.5601981130431258e-05, "loss": 0.1374, "step": 4897 }, { "epoch": 4.97, "learning_rate": 1.560016509640031e-05, "loss": 0.1218, "step": 4898 }, { "epoch": 4.97, "learning_rate": 1.5598348793248365e-05, "loss": 0.1353, "step": 4899 }, { "epoch": 4.97, "learning_rate": 1.5596532221062704e-05, "loss": 0.1212, "step": 4900 }, { "epoch": 4.97, "learning_rate": 1.559471537993063e-05, "loss": 0.1245, "step": 4901 }, { "epoch": 4.97, "learning_rate": 1.5592898269939448e-05, "loss": 0.1385, "step": 4902 }, { "epoch": 4.98, "learning_rate": 1.5591080891176485e-05, "loss": 0.1418, "step": 4903 }, { "epoch": 4.98, "learning_rate": 1.5589263243729077e-05, "loss": 0.1178, "step": 4904 }, { "epoch": 4.98, "learning_rate": 1.558744532768457e-05, "loss": 0.1418, "step": 4905 }, { "epoch": 4.98, "learning_rate": 1.5585627143130326e-05, "loss": 0.1407, "step": 4906 }, { "epoch": 4.98, "learning_rate": 1.5583808690153724e-05, "loss": 0.1377, "step": 4907 }, { "epoch": 4.98, "learning_rate": 1.5581989968842144e-05, "loss": 0.1402, "step": 4908 }, { "epoch": 4.98, "learning_rate": 1.5580170979282988e-05, "loss": 0.1429, "step": 4909 }, { "epoch": 4.98, "learning_rate": 1.5578351721563674e-05, "loss": 0.1423, "step": 4910 }, { "epoch": 4.98, "learning_rate": 1.557653219577162e-05, "loss": 0.1255, "step": 4911 }, { "epoch": 4.98, "learning_rate": 1.5574712401994276e-05, "loss": 0.1398, "step": 4912 }, { "epoch": 4.99, "learning_rate": 1.5572892340319084e-05, "loss": 0.1328, "step": 4913 }, { "epoch": 4.99, "learning_rate": 1.557107201083352e-05, "loss": 0.1195, "step": 4914 }, { "epoch": 4.99, "learning_rate": 1.5569251413625043e-05, "loss": 0.1333, "step": 4915 }, { "epoch": 4.99, "learning_rate": 1.5567430548781162e-05, "loss": 0.1467, "step": 4916 }, { "epoch": 4.99, "learning_rate": 1.556560941638937e-05, "loss": 0.1197, "step": 4917 }, { "epoch": 4.99, "learning_rate": 1.5563788016537185e-05, "loss": 0.1212, "step": 4918 }, { "epoch": 4.99, "learning_rate": 1.556196634931214e-05, "loss": 0.1522, "step": 4919 }, { "epoch": 4.99, "learning_rate": 1.5560144414801772e-05, "loss": 0.1199, "step": 4920 }, { "epoch": 4.99, "learning_rate": 1.555832221309364e-05, "loss": 0.1251, "step": 4921 }, { "epoch": 4.99, "learning_rate": 1.5556499744275308e-05, "loss": 0.1289, "step": 4922 }, { "epoch": 5.0, "learning_rate": 1.5554677008434357e-05, "loss": 0.1598, "step": 4923 }, { "epoch": 5.0, "learning_rate": 1.555285400565838e-05, "loss": 0.1327, "step": 4924 }, { "epoch": 5.0, "learning_rate": 1.555103073603498e-05, "loss": 0.1375, "step": 4925 }, { "epoch": 5.0, "learning_rate": 1.5549207199651787e-05, "loss": 0.1487, "step": 4926 }, { "epoch": 5.0, "learning_rate": 1.554738339659642e-05, "loss": 0.1396, "step": 4927 }, { "epoch": 5.0, "learning_rate": 1.5545559326956534e-05, "loss": 0.122, "step": 4928 }, { "epoch": 5.0, "learning_rate": 1.5543734990819776e-05, "loss": 0.0722, "step": 4929 }, { "epoch": 5.0, "learning_rate": 1.5541910388273823e-05, "loss": 0.0681, "step": 4930 }, { "epoch": 5.0, "learning_rate": 1.554008551940636e-05, "loss": 0.0791, "step": 4931 }, { "epoch": 5.0, "learning_rate": 1.5538260384305076e-05, "loss": 0.0845, "step": 4932 }, { "epoch": 5.01, "learning_rate": 1.553643498305768e-05, "loss": 0.085, "step": 4933 }, { "epoch": 5.01, "learning_rate": 1.5534609315751897e-05, "loss": 0.0743, "step": 4934 }, { "epoch": 5.01, "learning_rate": 1.5532783382475462e-05, "loss": 0.0719, "step": 4935 }, { "epoch": 5.01, "learning_rate": 1.553095718331612e-05, "loss": 0.0691, "step": 4936 }, { "epoch": 5.01, "learning_rate": 1.552913071836163e-05, "loss": 0.0612, "step": 4937 }, { "epoch": 5.01, "learning_rate": 1.552730398769976e-05, "loss": 0.0823, "step": 4938 }, { "epoch": 5.01, "learning_rate": 1.5525476991418305e-05, "loss": 0.0738, "step": 4939 }, { "epoch": 5.01, "learning_rate": 1.552364972960506e-05, "loss": 0.0718, "step": 4940 }, { "epoch": 5.01, "learning_rate": 1.552182220234783e-05, "loss": 0.0784, "step": 4941 }, { "epoch": 5.01, "learning_rate": 1.5519994409734445e-05, "loss": 0.0737, "step": 4942 }, { "epoch": 5.02, "learning_rate": 1.5518166351852736e-05, "loss": 0.06, "step": 4943 }, { "epoch": 5.02, "learning_rate": 1.5516338028790555e-05, "loss": 0.0981, "step": 4944 }, { "epoch": 5.02, "learning_rate": 1.551450944063576e-05, "loss": 0.0649, "step": 4945 }, { "epoch": 5.02, "learning_rate": 1.551268058747623e-05, "loss": 0.0586, "step": 4946 }, { "epoch": 5.02, "learning_rate": 1.5510851469399855e-05, "loss": 0.0801, "step": 4947 }, { "epoch": 5.02, "learning_rate": 1.5509022086494526e-05, "loss": 0.0867, "step": 4948 }, { "epoch": 5.02, "learning_rate": 1.5507192438848164e-05, "loss": 0.066, "step": 4949 }, { "epoch": 5.02, "learning_rate": 1.550536252654869e-05, "loss": 0.0676, "step": 4950 }, { "epoch": 5.02, "learning_rate": 1.550353234968404e-05, "loss": 0.0789, "step": 4951 }, { "epoch": 5.02, "learning_rate": 1.5501701908342168e-05, "loss": 0.0682, "step": 4952 }, { "epoch": 5.03, "learning_rate": 1.5499871202611038e-05, "loss": 0.0949, "step": 4953 }, { "epoch": 5.03, "learning_rate": 1.5498040232578628e-05, "loss": 0.071, "step": 4954 }, { "epoch": 5.03, "learning_rate": 1.549620899833292e-05, "loss": 0.0781, "step": 4955 }, { "epoch": 5.03, "learning_rate": 1.549437749996192e-05, "loss": 0.0746, "step": 4956 }, { "epoch": 5.03, "learning_rate": 1.5492545737553645e-05, "loss": 0.0682, "step": 4957 }, { "epoch": 5.03, "learning_rate": 1.5490713711196116e-05, "loss": 0.0836, "step": 4958 }, { "epoch": 5.03, "learning_rate": 1.548888142097738e-05, "loss": 0.0773, "step": 4959 }, { "epoch": 5.03, "learning_rate": 1.548704886698548e-05, "loss": 0.0651, "step": 4960 }, { "epoch": 5.03, "learning_rate": 1.548521604930849e-05, "loss": 0.0601, "step": 4961 }, { "epoch": 5.04, "learning_rate": 1.5483382968034483e-05, "loss": 0.0752, "step": 4962 }, { "epoch": 5.04, "learning_rate": 1.548154962325155e-05, "loss": 0.0842, "step": 4963 }, { "epoch": 5.04, "learning_rate": 1.5479716015047795e-05, "loss": 0.0777, "step": 4964 }, { "epoch": 5.04, "learning_rate": 1.5477882143511332e-05, "loss": 0.0686, "step": 4965 }, { "epoch": 5.04, "learning_rate": 1.547604800873029e-05, "loss": 0.0767, "step": 4966 }, { "epoch": 5.04, "learning_rate": 1.5474213610792813e-05, "loss": 0.07, "step": 4967 }, { "epoch": 5.04, "learning_rate": 1.547237894978705e-05, "loss": 0.0766, "step": 4968 }, { "epoch": 5.04, "learning_rate": 1.547054402580117e-05, "loss": 0.0647, "step": 4969 }, { "epoch": 5.04, "learning_rate": 1.546870883892335e-05, "loss": 0.0826, "step": 4970 }, { "epoch": 5.04, "learning_rate": 1.5466873389241786e-05, "loss": 0.0734, "step": 4971 }, { "epoch": 5.05, "learning_rate": 1.5465037676844675e-05, "loss": 0.0812, "step": 4972 }, { "epoch": 5.05, "learning_rate": 1.5463201701820242e-05, "loss": 0.0674, "step": 4973 }, { "epoch": 5.05, "learning_rate": 1.5461365464256714e-05, "loss": 0.0894, "step": 4974 }, { "epoch": 5.05, "learning_rate": 1.5459528964242324e-05, "loss": 0.0754, "step": 4975 }, { "epoch": 5.05, "learning_rate": 1.5457692201865337e-05, "loss": 0.0641, "step": 4976 }, { "epoch": 5.05, "learning_rate": 1.545585517721402e-05, "loss": 0.076, "step": 4977 }, { "epoch": 5.05, "learning_rate": 1.5454017890376648e-05, "loss": 0.0809, "step": 4978 }, { "epoch": 5.05, "learning_rate": 1.5452180341441515e-05, "loss": 0.0734, "step": 4979 }, { "epoch": 5.05, "learning_rate": 1.5450342530496926e-05, "loss": 0.0797, "step": 4980 }, { "epoch": 5.05, "learning_rate": 1.54485044576312e-05, "loss": 0.0869, "step": 4981 }, { "epoch": 5.06, "learning_rate": 1.5446666122932673e-05, "loss": 0.0818, "step": 4982 }, { "epoch": 5.06, "learning_rate": 1.5444827526489675e-05, "loss": 0.0628, "step": 4983 }, { "epoch": 5.06, "learning_rate": 1.544298866839057e-05, "loss": 0.0762, "step": 4984 }, { "epoch": 5.06, "learning_rate": 1.5441149548723723e-05, "loss": 0.0825, "step": 4985 }, { "epoch": 5.06, "learning_rate": 1.5439310167577514e-05, "loss": 0.059, "step": 4986 }, { "epoch": 5.06, "learning_rate": 1.543747052504034e-05, "loss": 0.0774, "step": 4987 }, { "epoch": 5.06, "learning_rate": 1.54356306212006e-05, "loss": 0.0709, "step": 4988 }, { "epoch": 5.06, "learning_rate": 1.5433790456146724e-05, "loss": 0.0773, "step": 4989 }, { "epoch": 5.06, "learning_rate": 1.5431950029967132e-05, "loss": 0.0759, "step": 4990 }, { "epoch": 5.06, "learning_rate": 1.543010934275027e-05, "loss": 0.0961, "step": 4991 }, { "epoch": 5.07, "learning_rate": 1.5428268394584595e-05, "loss": 0.0768, "step": 4992 }, { "epoch": 5.07, "learning_rate": 1.5426427185558572e-05, "loss": 0.0894, "step": 4993 }, { "epoch": 5.07, "learning_rate": 1.5424585715760686e-05, "loss": 0.0796, "step": 4994 }, { "epoch": 5.07, "learning_rate": 1.5422743985279432e-05, "loss": 0.0615, "step": 4995 }, { "epoch": 5.07, "learning_rate": 1.5420901994203315e-05, "loss": 0.0761, "step": 4996 }, { "epoch": 5.07, "learning_rate": 1.5419059742620847e-05, "loss": 0.0718, "step": 4997 }, { "epoch": 5.07, "learning_rate": 1.5417217230620568e-05, "loss": 0.0865, "step": 4998 }, { "epoch": 5.07, "learning_rate": 1.541537445829102e-05, "loss": 0.0678, "step": 4999 }, { "epoch": 5.07, "learning_rate": 1.5413531425720747e-05, "loss": 0.073, "step": 5000 }, { "epoch": 5.07, "learning_rate": 1.5411688132998332e-05, "loss": 0.0985, "step": 5001 }, { "epoch": 5.08, "learning_rate": 1.5409844580212354e-05, "loss": 0.0827, "step": 5002 }, { "epoch": 5.08, "learning_rate": 1.5408000767451402e-05, "loss": 0.0866, "step": 5003 }, { "epoch": 5.08, "learning_rate": 1.5406156694804086e-05, "loss": 0.059, "step": 5004 }, { "epoch": 5.08, "learning_rate": 1.5404312362359018e-05, "loss": 0.076, "step": 5005 }, { "epoch": 5.08, "learning_rate": 1.540246777020484e-05, "loss": 0.0718, "step": 5006 }, { "epoch": 5.08, "learning_rate": 1.5400622918430184e-05, "loss": 0.0863, "step": 5007 }, { "epoch": 5.08, "learning_rate": 1.5398777807123715e-05, "loss": 0.0805, "step": 5008 }, { "epoch": 5.08, "learning_rate": 1.5396932436374097e-05, "loss": 0.0867, "step": 5009 }, { "epoch": 5.08, "learning_rate": 1.539508680627001e-05, "loss": 0.0683, "step": 5010 }, { "epoch": 5.08, "learning_rate": 1.5393240916900155e-05, "loss": 0.0776, "step": 5011 }, { "epoch": 5.09, "learning_rate": 1.5391394768353226e-05, "loss": 0.088, "step": 5012 }, { "epoch": 5.09, "learning_rate": 1.5389548360717946e-05, "loss": 0.0658, "step": 5013 }, { "epoch": 5.09, "learning_rate": 1.5387701694083054e-05, "loss": 0.0715, "step": 5014 }, { "epoch": 5.09, "learning_rate": 1.538585476853728e-05, "loss": 0.069, "step": 5015 }, { "epoch": 5.09, "learning_rate": 1.538400758416939e-05, "loss": 0.0722, "step": 5016 }, { "epoch": 5.09, "learning_rate": 1.5382160141068147e-05, "loss": 0.0726, "step": 5017 }, { "epoch": 5.09, "learning_rate": 1.5380312439322334e-05, "loss": 0.0861, "step": 5018 }, { "epoch": 5.09, "learning_rate": 1.537846447902074e-05, "loss": 0.0639, "step": 5019 }, { "epoch": 5.09, "learning_rate": 1.5376616260252176e-05, "loss": 0.0804, "step": 5020 }, { "epoch": 5.09, "learning_rate": 1.5374767783105457e-05, "loss": 0.0879, "step": 5021 }, { "epoch": 5.1, "learning_rate": 1.5372919047669414e-05, "loss": 0.0715, "step": 5022 }, { "epoch": 5.1, "learning_rate": 1.5371070054032883e-05, "loss": 0.0714, "step": 5023 }, { "epoch": 5.1, "learning_rate": 1.536922080228473e-05, "loss": 0.072, "step": 5024 }, { "epoch": 5.1, "learning_rate": 1.5367371292513818e-05, "loss": 0.0622, "step": 5025 }, { "epoch": 5.1, "learning_rate": 1.5365521524809022e-05, "loss": 0.0696, "step": 5026 }, { "epoch": 5.1, "learning_rate": 1.536367149925924e-05, "loss": 0.0814, "step": 5027 }, { "epoch": 5.1, "learning_rate": 1.5361821215953378e-05, "loss": 0.0855, "step": 5028 }, { "epoch": 5.1, "learning_rate": 1.535997067498035e-05, "loss": 0.0674, "step": 5029 }, { "epoch": 5.1, "learning_rate": 1.535811987642908e-05, "loss": 0.0751, "step": 5030 }, { "epoch": 5.11, "learning_rate": 1.535626882038852e-05, "loss": 0.0792, "step": 5031 }, { "epoch": 5.11, "learning_rate": 1.535441750694762e-05, "loss": 0.0901, "step": 5032 }, { "epoch": 5.11, "learning_rate": 1.5352565936195346e-05, "loss": 0.0746, "step": 5033 }, { "epoch": 5.11, "learning_rate": 1.5350714108220673e-05, "loss": 0.075, "step": 5034 }, { "epoch": 5.11, "learning_rate": 1.53488620231126e-05, "loss": 0.0841, "step": 5035 }, { "epoch": 5.11, "learning_rate": 1.534700968096013e-05, "loss": 0.0748, "step": 5036 }, { "epoch": 5.11, "learning_rate": 1.5345157081852272e-05, "loss": 0.0845, "step": 5037 }, { "epoch": 5.11, "learning_rate": 1.5343304225878056e-05, "loss": 0.078, "step": 5038 }, { "epoch": 5.11, "learning_rate": 1.534145111312653e-05, "loss": 0.0738, "step": 5039 }, { "epoch": 5.11, "learning_rate": 1.5339597743686742e-05, "loss": 0.0938, "step": 5040 }, { "epoch": 5.12, "learning_rate": 1.533774411764775e-05, "loss": 0.075, "step": 5041 }, { "epoch": 5.12, "learning_rate": 1.5335890235098646e-05, "loss": 0.085, "step": 5042 }, { "epoch": 5.12, "learning_rate": 1.533403609612851e-05, "loss": 0.0866, "step": 5043 }, { "epoch": 5.12, "learning_rate": 1.5332181700826448e-05, "loss": 0.0672, "step": 5044 }, { "epoch": 5.12, "learning_rate": 1.5330327049281575e-05, "loss": 0.0876, "step": 5045 }, { "epoch": 5.12, "learning_rate": 1.5328472141583015e-05, "loss": 0.0781, "step": 5046 }, { "epoch": 5.12, "learning_rate": 1.532661697781991e-05, "loss": 0.0561, "step": 5047 }, { "epoch": 5.12, "learning_rate": 1.5324761558081412e-05, "loss": 0.066, "step": 5048 }, { "epoch": 5.12, "learning_rate": 1.532290588245668e-05, "loss": 0.0767, "step": 5049 }, { "epoch": 5.12, "learning_rate": 1.53210499510349e-05, "loss": 0.0884, "step": 5050 }, { "epoch": 5.13, "learning_rate": 1.5319193763905247e-05, "loss": 0.0817, "step": 5051 }, { "epoch": 5.13, "learning_rate": 1.531733732115693e-05, "loss": 0.0784, "step": 5052 }, { "epoch": 5.13, "learning_rate": 1.5315480622879164e-05, "loss": 0.0729, "step": 5053 }, { "epoch": 5.13, "learning_rate": 1.5313623669161173e-05, "loss": 0.0823, "step": 5054 }, { "epoch": 5.13, "learning_rate": 1.5311766460092188e-05, "loss": 0.0867, "step": 5055 }, { "epoch": 5.13, "learning_rate": 1.5309908995761466e-05, "loss": 0.0841, "step": 5056 }, { "epoch": 5.13, "learning_rate": 1.5308051276258264e-05, "loss": 0.0897, "step": 5057 }, { "epoch": 5.13, "learning_rate": 1.5306193301671862e-05, "loss": 0.077, "step": 5058 }, { "epoch": 5.13, "learning_rate": 1.5304335072091548e-05, "loss": 0.0748, "step": 5059 }, { "epoch": 5.13, "learning_rate": 1.5302476587606612e-05, "loss": 0.0779, "step": 5060 }, { "epoch": 5.14, "learning_rate": 1.530061784830637e-05, "loss": 0.0842, "step": 5061 }, { "epoch": 5.14, "learning_rate": 1.5298758854280146e-05, "loss": 0.0822, "step": 5062 }, { "epoch": 5.14, "learning_rate": 1.5296899605617278e-05, "loss": 0.0792, "step": 5063 }, { "epoch": 5.14, "learning_rate": 1.5295040102407104e-05, "loss": 0.0772, "step": 5064 }, { "epoch": 5.14, "learning_rate": 1.5293180344738998e-05, "loss": 0.0748, "step": 5065 }, { "epoch": 5.14, "learning_rate": 1.5291320332702323e-05, "loss": 0.0824, "step": 5066 }, { "epoch": 5.14, "learning_rate": 1.5289460066386465e-05, "loss": 0.071, "step": 5067 }, { "epoch": 5.14, "learning_rate": 1.528759954588082e-05, "loss": 0.0729, "step": 5068 }, { "epoch": 5.14, "learning_rate": 1.5285738771274804e-05, "loss": 0.0726, "step": 5069 }, { "epoch": 5.14, "learning_rate": 1.5283877742657826e-05, "loss": 0.0816, "step": 5070 }, { "epoch": 5.15, "learning_rate": 1.528201646011933e-05, "loss": 0.065, "step": 5071 }, { "epoch": 5.15, "learning_rate": 1.528015492374876e-05, "loss": 0.0774, "step": 5072 }, { "epoch": 5.15, "learning_rate": 1.5278293133635568e-05, "loss": 0.0912, "step": 5073 }, { "epoch": 5.15, "learning_rate": 1.527643108986923e-05, "loss": 0.0676, "step": 5074 }, { "epoch": 5.15, "learning_rate": 1.5274568792539225e-05, "loss": 0.0832, "step": 5075 }, { "epoch": 5.15, "learning_rate": 1.5272706241735048e-05, "loss": 0.0797, "step": 5076 }, { "epoch": 5.15, "learning_rate": 1.5270843437546206e-05, "loss": 0.0594, "step": 5077 }, { "epoch": 5.15, "learning_rate": 1.526898038006222e-05, "loss": 0.0754, "step": 5078 }, { "epoch": 5.15, "learning_rate": 1.5267117069372613e-05, "loss": 0.079, "step": 5079 }, { "epoch": 5.15, "learning_rate": 1.5265253505566938e-05, "loss": 0.0893, "step": 5080 }, { "epoch": 5.16, "learning_rate": 1.5263389688734744e-05, "loss": 0.0859, "step": 5081 }, { "epoch": 5.16, "learning_rate": 1.5261525618965604e-05, "loss": 0.0856, "step": 5082 }, { "epoch": 5.16, "learning_rate": 1.5259661296349092e-05, "loss": 0.0858, "step": 5083 }, { "epoch": 5.16, "learning_rate": 1.5257796720974798e-05, "loss": 0.0831, "step": 5084 }, { "epoch": 5.16, "learning_rate": 1.5255931892932333e-05, "loss": 0.0664, "step": 5085 }, { "epoch": 5.16, "learning_rate": 1.5254066812311305e-05, "loss": 0.0886, "step": 5086 }, { "epoch": 5.16, "learning_rate": 1.5252201479201352e-05, "loss": 0.0708, "step": 5087 }, { "epoch": 5.16, "learning_rate": 1.5250335893692106e-05, "loss": 0.0765, "step": 5088 }, { "epoch": 5.16, "learning_rate": 1.5248470055873222e-05, "loss": 0.0723, "step": 5089 }, { "epoch": 5.16, "learning_rate": 1.5246603965834369e-05, "loss": 0.0905, "step": 5090 }, { "epoch": 5.17, "learning_rate": 1.5244737623665212e-05, "loss": 0.0738, "step": 5091 }, { "epoch": 5.17, "learning_rate": 1.5242871029455452e-05, "loss": 0.083, "step": 5092 }, { "epoch": 5.17, "learning_rate": 1.524100418329478e-05, "loss": 0.0691, "step": 5093 }, { "epoch": 5.17, "learning_rate": 1.523913708527292e-05, "loss": 0.0743, "step": 5094 }, { "epoch": 5.17, "learning_rate": 1.523726973547959e-05, "loss": 0.0798, "step": 5095 }, { "epoch": 5.17, "learning_rate": 1.5235402134004526e-05, "loss": 0.0552, "step": 5096 }, { "epoch": 5.17, "learning_rate": 1.5233534280937478e-05, "loss": 0.0792, "step": 5097 }, { "epoch": 5.17, "learning_rate": 1.5231666176368211e-05, "loss": 0.0749, "step": 5098 }, { "epoch": 5.17, "learning_rate": 1.5229797820386498e-05, "loss": 0.0792, "step": 5099 }, { "epoch": 5.18, "learning_rate": 1.5227929213082122e-05, "loss": 0.0913, "step": 5100 }, { "epoch": 5.18, "learning_rate": 1.522606035454488e-05, "loss": 0.0885, "step": 5101 }, { "epoch": 5.18, "learning_rate": 1.5224191244864583e-05, "loss": 0.0927, "step": 5102 }, { "epoch": 5.18, "learning_rate": 1.5222321884131055e-05, "loss": 0.0898, "step": 5103 }, { "epoch": 5.18, "learning_rate": 1.5220452272434125e-05, "loss": 0.0825, "step": 5104 }, { "epoch": 5.18, "learning_rate": 1.5218582409863642e-05, "loss": 0.0679, "step": 5105 }, { "epoch": 5.18, "learning_rate": 1.5216712296509468e-05, "loss": 0.078, "step": 5106 }, { "epoch": 5.18, "learning_rate": 1.5214841932461464e-05, "loss": 0.0758, "step": 5107 }, { "epoch": 5.18, "learning_rate": 1.5212971317809517e-05, "loss": 0.0943, "step": 5108 }, { "epoch": 5.18, "learning_rate": 1.5211100452643522e-05, "loss": 0.0797, "step": 5109 }, { "epoch": 5.19, "learning_rate": 1.5209229337053381e-05, "loss": 0.0885, "step": 5110 }, { "epoch": 5.19, "learning_rate": 1.520735797112902e-05, "loss": 0.0621, "step": 5111 }, { "epoch": 5.19, "learning_rate": 1.5205486354960361e-05, "loss": 0.0818, "step": 5112 }, { "epoch": 5.19, "learning_rate": 1.5203614488637349e-05, "loss": 0.08, "step": 5113 }, { "epoch": 5.19, "learning_rate": 1.5201742372249937e-05, "loss": 0.0766, "step": 5114 }, { "epoch": 5.19, "learning_rate": 1.5199870005888095e-05, "loss": 0.0932, "step": 5115 }, { "epoch": 5.19, "learning_rate": 1.5197997389641801e-05, "loss": 0.0684, "step": 5116 }, { "epoch": 5.19, "learning_rate": 1.5196124523601042e-05, "loss": 0.0787, "step": 5117 }, { "epoch": 5.19, "learning_rate": 1.5194251407855817e-05, "loss": 0.1019, "step": 5118 }, { "epoch": 5.19, "learning_rate": 1.5192378042496148e-05, "loss": 0.0661, "step": 5119 }, { "epoch": 5.2, "learning_rate": 1.519050442761206e-05, "loss": 0.0624, "step": 5120 }, { "epoch": 5.2, "learning_rate": 1.5188630563293587e-05, "loss": 0.0756, "step": 5121 }, { "epoch": 5.2, "learning_rate": 1.5186756449630782e-05, "loss": 0.094, "step": 5122 }, { "epoch": 5.2, "learning_rate": 1.5184882086713709e-05, "loss": 0.0734, "step": 5123 }, { "epoch": 5.2, "learning_rate": 1.5183007474632436e-05, "loss": 0.0811, "step": 5124 }, { "epoch": 5.2, "learning_rate": 1.5181132613477056e-05, "loss": 0.0872, "step": 5125 }, { "epoch": 5.2, "learning_rate": 1.5179257503337663e-05, "loss": 0.0746, "step": 5126 }, { "epoch": 5.2, "learning_rate": 1.517738214430437e-05, "loss": 0.0756, "step": 5127 }, { "epoch": 5.2, "learning_rate": 1.5175506536467298e-05, "loss": 0.0855, "step": 5128 }, { "epoch": 5.2, "learning_rate": 1.5173630679916578e-05, "loss": 0.0843, "step": 5129 }, { "epoch": 5.21, "learning_rate": 1.5171754574742358e-05, "loss": 0.0836, "step": 5130 }, { "epoch": 5.21, "learning_rate": 1.51698782210348e-05, "loss": 0.0744, "step": 5131 }, { "epoch": 5.21, "learning_rate": 1.516800161888407e-05, "loss": 0.0752, "step": 5132 }, { "epoch": 5.21, "learning_rate": 1.5166124768380349e-05, "loss": 0.0933, "step": 5133 }, { "epoch": 5.21, "learning_rate": 1.5164247669613833e-05, "loss": 0.1174, "step": 5134 }, { "epoch": 5.21, "learning_rate": 1.5162370322674725e-05, "loss": 0.0885, "step": 5135 }, { "epoch": 5.21, "learning_rate": 1.5160492727653241e-05, "loss": 0.112, "step": 5136 }, { "epoch": 5.21, "learning_rate": 1.515861488463962e-05, "loss": 0.0733, "step": 5137 }, { "epoch": 5.21, "learning_rate": 1.5156736793724095e-05, "loss": 0.065, "step": 5138 }, { "epoch": 5.21, "learning_rate": 1.5154858454996922e-05, "loss": 0.094, "step": 5139 }, { "epoch": 5.22, "learning_rate": 1.5152979868548363e-05, "loss": 0.0745, "step": 5140 }, { "epoch": 5.22, "learning_rate": 1.5151101034468699e-05, "loss": 0.0923, "step": 5141 }, { "epoch": 5.22, "learning_rate": 1.5149221952848221e-05, "loss": 0.0731, "step": 5142 }, { "epoch": 5.22, "learning_rate": 1.5147342623777222e-05, "loss": 0.0821, "step": 5143 }, { "epoch": 5.22, "learning_rate": 1.5145463047346023e-05, "loss": 0.0747, "step": 5144 }, { "epoch": 5.22, "learning_rate": 1.5143583223644949e-05, "loss": 0.0778, "step": 5145 }, { "epoch": 5.22, "learning_rate": 1.5141703152764326e-05, "loss": 0.0643, "step": 5146 }, { "epoch": 5.22, "learning_rate": 1.5139822834794513e-05, "loss": 0.0849, "step": 5147 }, { "epoch": 5.22, "learning_rate": 1.513794226982587e-05, "loss": 0.0771, "step": 5148 }, { "epoch": 5.22, "learning_rate": 1.513606145794876e-05, "loss": 0.1004, "step": 5149 }, { "epoch": 5.23, "learning_rate": 1.513418039925358e-05, "loss": 0.0845, "step": 5150 }, { "epoch": 5.23, "learning_rate": 1.5132299093830716e-05, "loss": 0.0737, "step": 5151 }, { "epoch": 5.23, "learning_rate": 1.5130417541770579e-05, "loss": 0.0685, "step": 5152 }, { "epoch": 5.23, "learning_rate": 1.512853574316359e-05, "loss": 0.0805, "step": 5153 }, { "epoch": 5.23, "learning_rate": 1.512665369810018e-05, "loss": 0.0926, "step": 5154 }, { "epoch": 5.23, "learning_rate": 1.5124771406670791e-05, "loss": 0.0743, "step": 5155 }, { "epoch": 5.23, "learning_rate": 1.5122888868965882e-05, "loss": 0.0699, "step": 5156 }, { "epoch": 5.23, "learning_rate": 1.5121006085075912e-05, "loss": 0.0761, "step": 5157 }, { "epoch": 5.23, "learning_rate": 1.5119123055091369e-05, "loss": 0.083, "step": 5158 }, { "epoch": 5.23, "learning_rate": 1.5117239779102738e-05, "loss": 0.0723, "step": 5159 }, { "epoch": 5.24, "learning_rate": 1.5115356257200522e-05, "loss": 0.0658, "step": 5160 }, { "epoch": 5.24, "learning_rate": 1.5113472489475238e-05, "loss": 0.0802, "step": 5161 }, { "epoch": 5.24, "learning_rate": 1.5111588476017414e-05, "loss": 0.0862, "step": 5162 }, { "epoch": 5.24, "learning_rate": 1.510970421691758e-05, "loss": 0.0811, "step": 5163 }, { "epoch": 5.24, "learning_rate": 1.5107819712266292e-05, "loss": 0.0897, "step": 5164 }, { "epoch": 5.24, "learning_rate": 1.5105934962154112e-05, "loss": 0.0778, "step": 5165 }, { "epoch": 5.24, "learning_rate": 1.5104049966671608e-05, "loss": 0.0767, "step": 5166 }, { "epoch": 5.24, "learning_rate": 1.5102164725909374e-05, "loss": 0.1096, "step": 5167 }, { "epoch": 5.24, "learning_rate": 1.5100279239957996e-05, "loss": 0.0856, "step": 5168 }, { "epoch": 5.25, "learning_rate": 1.509839350890809e-05, "loss": 0.0807, "step": 5169 }, { "epoch": 5.25, "learning_rate": 1.5096507532850275e-05, "loss": 0.0782, "step": 5170 }, { "epoch": 5.25, "learning_rate": 1.5094621311875183e-05, "loss": 0.0824, "step": 5171 }, { "epoch": 5.25, "learning_rate": 1.5092734846073457e-05, "loss": 0.0781, "step": 5172 }, { "epoch": 5.25, "learning_rate": 1.5090848135535759e-05, "loss": 0.1016, "step": 5173 }, { "epoch": 5.25, "learning_rate": 1.5088961180352746e-05, "loss": 0.0758, "step": 5174 }, { "epoch": 5.25, "learning_rate": 1.5087073980615103e-05, "loss": 0.0835, "step": 5175 }, { "epoch": 5.25, "learning_rate": 1.5085186536413523e-05, "loss": 0.0734, "step": 5176 }, { "epoch": 5.25, "learning_rate": 1.508329884783871e-05, "loss": 0.0853, "step": 5177 }, { "epoch": 5.25, "learning_rate": 1.5081410914981371e-05, "loss": 0.0765, "step": 5178 }, { "epoch": 5.26, "learning_rate": 1.5079522737932238e-05, "loss": 0.0723, "step": 5179 }, { "epoch": 5.26, "learning_rate": 1.5077634316782046e-05, "loss": 0.076, "step": 5180 }, { "epoch": 5.26, "learning_rate": 1.507574565162155e-05, "loss": 0.0836, "step": 5181 }, { "epoch": 5.26, "learning_rate": 1.5073856742541505e-05, "loss": 0.0703, "step": 5182 }, { "epoch": 5.26, "learning_rate": 1.507196758963269e-05, "loss": 0.077, "step": 5183 }, { "epoch": 5.26, "learning_rate": 1.5070078192985886e-05, "loss": 0.0761, "step": 5184 }, { "epoch": 5.26, "learning_rate": 1.506818855269189e-05, "loss": 0.0835, "step": 5185 }, { "epoch": 5.26, "learning_rate": 1.5066298668841511e-05, "loss": 0.0832, "step": 5186 }, { "epoch": 5.26, "learning_rate": 1.5064408541525573e-05, "loss": 0.0701, "step": 5187 }, { "epoch": 5.26, "learning_rate": 1.5062518170834902e-05, "loss": 0.0715, "step": 5188 }, { "epoch": 5.27, "learning_rate": 1.506062755686034e-05, "loss": 0.078, "step": 5189 }, { "epoch": 5.27, "learning_rate": 1.505873669969275e-05, "loss": 0.0816, "step": 5190 }, { "epoch": 5.27, "learning_rate": 1.5056845599422993e-05, "loss": 0.0854, "step": 5191 }, { "epoch": 5.27, "learning_rate": 1.5054954256141952e-05, "loss": 0.0793, "step": 5192 }, { "epoch": 5.27, "learning_rate": 1.505306266994051e-05, "loss": 0.0919, "step": 5193 }, { "epoch": 5.27, "learning_rate": 1.5051170840909577e-05, "loss": 0.0693, "step": 5194 }, { "epoch": 5.27, "learning_rate": 1.5049278769140063e-05, "loss": 0.0643, "step": 5195 }, { "epoch": 5.27, "learning_rate": 1.5047386454722888e-05, "loss": 0.0885, "step": 5196 }, { "epoch": 5.27, "learning_rate": 1.5045493897748997e-05, "loss": 0.0909, "step": 5197 }, { "epoch": 5.27, "learning_rate": 1.5043601098309338e-05, "loss": 0.0796, "step": 5198 }, { "epoch": 5.28, "learning_rate": 1.5041708056494865e-05, "loss": 0.0763, "step": 5199 }, { "epoch": 5.28, "learning_rate": 1.5039814772396557e-05, "loss": 0.0693, "step": 5200 }, { "epoch": 5.28, "learning_rate": 1.5037921246105392e-05, "loss": 0.0746, "step": 5201 }, { "epoch": 5.28, "learning_rate": 1.5036027477712364e-05, "loss": 0.0854, "step": 5202 }, { "epoch": 5.28, "learning_rate": 1.5034133467308488e-05, "loss": 0.0812, "step": 5203 }, { "epoch": 5.28, "learning_rate": 1.5032239214984773e-05, "loss": 0.0895, "step": 5204 }, { "epoch": 5.28, "learning_rate": 1.5030344720832254e-05, "loss": 0.0714, "step": 5205 }, { "epoch": 5.28, "learning_rate": 1.5028449984941975e-05, "loss": 0.0771, "step": 5206 }, { "epoch": 5.28, "learning_rate": 1.5026555007404984e-05, "loss": 0.0794, "step": 5207 }, { "epoch": 5.28, "learning_rate": 1.5024659788312347e-05, "loss": 0.0753, "step": 5208 }, { "epoch": 5.29, "learning_rate": 1.5022764327755145e-05, "loss": 0.0821, "step": 5209 }, { "epoch": 5.29, "learning_rate": 1.5020868625824458e-05, "loss": 0.0775, "step": 5210 }, { "epoch": 5.29, "learning_rate": 1.5018972682611392e-05, "loss": 0.0752, "step": 5211 }, { "epoch": 5.29, "learning_rate": 1.501707649820706e-05, "loss": 0.0733, "step": 5212 }, { "epoch": 5.29, "learning_rate": 1.5015180072702576e-05, "loss": 0.0606, "step": 5213 }, { "epoch": 5.29, "learning_rate": 1.501328340618908e-05, "loss": 0.0716, "step": 5214 }, { "epoch": 5.29, "learning_rate": 1.5011386498757721e-05, "loss": 0.0885, "step": 5215 }, { "epoch": 5.29, "learning_rate": 1.5009489350499653e-05, "loss": 0.085, "step": 5216 }, { "epoch": 5.29, "learning_rate": 1.5007591961506042e-05, "loss": 0.0969, "step": 5217 }, { "epoch": 5.29, "learning_rate": 1.5005694331868074e-05, "loss": 0.0898, "step": 5218 }, { "epoch": 5.3, "learning_rate": 1.500379646167694e-05, "loss": 0.0846, "step": 5219 }, { "epoch": 5.3, "learning_rate": 1.5001898351023844e-05, "loss": 0.0634, "step": 5220 }, { "epoch": 5.3, "learning_rate": 1.5000000000000002e-05, "loss": 0.0835, "step": 5221 }, { "epoch": 5.3, "learning_rate": 1.4998101408696637e-05, "loss": 0.0737, "step": 5222 }, { "epoch": 5.3, "learning_rate": 1.4996202577204992e-05, "loss": 0.0772, "step": 5223 }, { "epoch": 5.3, "learning_rate": 1.4994303505616314e-05, "loss": 0.1012, "step": 5224 }, { "epoch": 5.3, "learning_rate": 1.4992404194021868e-05, "loss": 0.0811, "step": 5225 }, { "epoch": 5.3, "learning_rate": 1.4990504642512925e-05, "loss": 0.0931, "step": 5226 }, { "epoch": 5.3, "learning_rate": 1.4988604851180768e-05, "loss": 0.0686, "step": 5227 }, { "epoch": 5.3, "learning_rate": 1.4986704820116698e-05, "loss": 0.0843, "step": 5228 }, { "epoch": 5.31, "learning_rate": 1.4984804549412018e-05, "loss": 0.0853, "step": 5229 }, { "epoch": 5.31, "learning_rate": 1.4982904039158048e-05, "loss": 0.0754, "step": 5230 }, { "epoch": 5.31, "learning_rate": 1.498100328944612e-05, "loss": 0.0812, "step": 5231 }, { "epoch": 5.31, "learning_rate": 1.4979102300367576e-05, "loss": 0.0666, "step": 5232 }, { "epoch": 5.31, "learning_rate": 1.4977201072013771e-05, "loss": 0.086, "step": 5233 }, { "epoch": 5.31, "learning_rate": 1.497529960447607e-05, "loss": 0.0798, "step": 5234 }, { "epoch": 5.31, "learning_rate": 1.4973397897845845e-05, "loss": 0.0742, "step": 5235 }, { "epoch": 5.31, "learning_rate": 1.497149595221449e-05, "loss": 0.0704, "step": 5236 }, { "epoch": 5.31, "learning_rate": 1.4969593767673403e-05, "loss": 0.0898, "step": 5237 }, { "epoch": 5.32, "learning_rate": 1.4967691344313995e-05, "loss": 0.0807, "step": 5238 }, { "epoch": 5.32, "learning_rate": 1.4965788682227685e-05, "loss": 0.0889, "step": 5239 }, { "epoch": 5.32, "learning_rate": 1.4963885781505916e-05, "loss": 0.086, "step": 5240 }, { "epoch": 5.32, "learning_rate": 1.4961982642240122e-05, "loss": 0.0772, "step": 5241 }, { "epoch": 5.32, "learning_rate": 1.4960079264521773e-05, "loss": 0.0771, "step": 5242 }, { "epoch": 5.32, "learning_rate": 1.4958175648442326e-05, "loss": 0.0997, "step": 5243 }, { "epoch": 5.32, "learning_rate": 1.495627179409327e-05, "loss": 0.0818, "step": 5244 }, { "epoch": 5.32, "learning_rate": 1.4954367701566089e-05, "loss": 0.0688, "step": 5245 }, { "epoch": 5.32, "learning_rate": 1.4952463370952289e-05, "loss": 0.0848, "step": 5246 }, { "epoch": 5.32, "learning_rate": 1.4950558802343385e-05, "loss": 0.0774, "step": 5247 }, { "epoch": 5.33, "learning_rate": 1.4948653995830901e-05, "loss": 0.0862, "step": 5248 }, { "epoch": 5.33, "learning_rate": 1.494674895150638e-05, "loss": 0.0964, "step": 5249 }, { "epoch": 5.33, "learning_rate": 1.4944843669461364e-05, "loss": 0.058, "step": 5250 }, { "epoch": 5.33, "learning_rate": 1.4942938149787414e-05, "loss": 0.1048, "step": 5251 }, { "epoch": 5.33, "learning_rate": 1.4941032392576102e-05, "loss": 0.0887, "step": 5252 }, { "epoch": 5.33, "learning_rate": 1.4939126397919014e-05, "loss": 0.0844, "step": 5253 }, { "epoch": 5.33, "learning_rate": 1.4937220165907742e-05, "loss": 0.0728, "step": 5254 }, { "epoch": 5.33, "learning_rate": 1.493531369663389e-05, "loss": 0.0838, "step": 5255 }, { "epoch": 5.33, "learning_rate": 1.4933406990189076e-05, "loss": 0.0764, "step": 5256 }, { "epoch": 5.33, "learning_rate": 1.4931500046664932e-05, "loss": 0.0846, "step": 5257 }, { "epoch": 5.34, "learning_rate": 1.4929592866153091e-05, "loss": 0.0787, "step": 5258 }, { "epoch": 5.34, "learning_rate": 1.4927685448745211e-05, "loss": 0.0853, "step": 5259 }, { "epoch": 5.34, "learning_rate": 1.4925777794532951e-05, "loss": 0.0703, "step": 5260 }, { "epoch": 5.34, "learning_rate": 1.4923869903607985e-05, "loss": 0.1051, "step": 5261 }, { "epoch": 5.34, "learning_rate": 1.4921961776062004e-05, "loss": 0.0793, "step": 5262 }, { "epoch": 5.34, "learning_rate": 1.4920053411986696e-05, "loss": 0.0842, "step": 5263 }, { "epoch": 5.34, "learning_rate": 1.4918144811473773e-05, "loss": 0.1059, "step": 5264 }, { "epoch": 5.34, "learning_rate": 1.4916235974614956e-05, "loss": 0.0814, "step": 5265 }, { "epoch": 5.34, "learning_rate": 1.4914326901501977e-05, "loss": 0.0928, "step": 5266 }, { "epoch": 5.34, "learning_rate": 1.4912417592226576e-05, "loss": 0.0916, "step": 5267 }, { "epoch": 5.35, "learning_rate": 1.4910508046880503e-05, "loss": 0.1037, "step": 5268 }, { "epoch": 5.35, "learning_rate": 1.4908598265555528e-05, "loss": 0.0729, "step": 5269 }, { "epoch": 5.35, "learning_rate": 1.4906688248343426e-05, "loss": 0.0807, "step": 5270 }, { "epoch": 5.35, "learning_rate": 1.4904777995335986e-05, "loss": 0.0833, "step": 5271 }, { "epoch": 5.35, "learning_rate": 1.4902867506625002e-05, "loss": 0.1283, "step": 5272 }, { "epoch": 5.35, "learning_rate": 1.4900956782302291e-05, "loss": 0.0784, "step": 5273 }, { "epoch": 5.35, "learning_rate": 1.4899045822459668e-05, "loss": 0.0923, "step": 5274 }, { "epoch": 5.35, "learning_rate": 1.4897134627188974e-05, "loss": 0.0962, "step": 5275 }, { "epoch": 5.35, "learning_rate": 1.4895223196582045e-05, "loss": 0.0889, "step": 5276 }, { "epoch": 5.35, "learning_rate": 1.4893311530730738e-05, "loss": 0.0797, "step": 5277 }, { "epoch": 5.36, "learning_rate": 1.4891399629726926e-05, "loss": 0.1015, "step": 5278 }, { "epoch": 5.36, "learning_rate": 1.4889487493662482e-05, "loss": 0.0865, "step": 5279 }, { "epoch": 5.36, "learning_rate": 1.4887575122629293e-05, "loss": 0.0619, "step": 5280 }, { "epoch": 5.36, "learning_rate": 1.4885662516719264e-05, "loss": 0.0882, "step": 5281 }, { "epoch": 5.36, "learning_rate": 1.4883749676024309e-05, "loss": 0.0849, "step": 5282 }, { "epoch": 5.36, "learning_rate": 1.4881836600636346e-05, "loss": 0.0985, "step": 5283 }, { "epoch": 5.36, "learning_rate": 1.4879923290647314e-05, "loss": 0.0713, "step": 5284 }, { "epoch": 5.36, "learning_rate": 1.4878009746149152e-05, "loss": 0.0934, "step": 5285 }, { "epoch": 5.36, "learning_rate": 1.4876095967233823e-05, "loss": 0.0681, "step": 5286 }, { "epoch": 5.36, "learning_rate": 1.4874181953993297e-05, "loss": 0.0929, "step": 5287 }, { "epoch": 5.37, "learning_rate": 1.4872267706519549e-05, "loss": 0.0834, "step": 5288 }, { "epoch": 5.37, "learning_rate": 1.4870353224904572e-05, "loss": 0.0835, "step": 5289 }, { "epoch": 5.37, "learning_rate": 1.4868438509240369e-05, "loss": 0.076, "step": 5290 }, { "epoch": 5.37, "learning_rate": 1.4866523559618948e-05, "loss": 0.1049, "step": 5291 }, { "epoch": 5.37, "learning_rate": 1.4864608376132343e-05, "loss": 0.0772, "step": 5292 }, { "epoch": 5.37, "learning_rate": 1.4862692958872583e-05, "loss": 0.0817, "step": 5293 }, { "epoch": 5.37, "learning_rate": 1.4860777307931715e-05, "loss": 0.0724, "step": 5294 }, { "epoch": 5.37, "learning_rate": 1.4858861423401803e-05, "loss": 0.0854, "step": 5295 }, { "epoch": 5.37, "learning_rate": 1.4856945305374908e-05, "loss": 0.0875, "step": 5296 }, { "epoch": 5.37, "learning_rate": 1.4855028953943116e-05, "loss": 0.0704, "step": 5297 }, { "epoch": 5.38, "learning_rate": 1.4853112369198522e-05, "loss": 0.0842, "step": 5298 }, { "epoch": 5.38, "learning_rate": 1.4851195551233227e-05, "loss": 0.0991, "step": 5299 }, { "epoch": 5.38, "learning_rate": 1.4849278500139338e-05, "loss": 0.0828, "step": 5300 }, { "epoch": 5.38, "learning_rate": 1.4847361216008993e-05, "loss": 0.0912, "step": 5301 }, { "epoch": 5.38, "learning_rate": 1.4845443698934316e-05, "loss": 0.0672, "step": 5302 }, { "epoch": 5.38, "learning_rate": 1.4843525949007467e-05, "loss": 0.0757, "step": 5303 }, { "epoch": 5.38, "learning_rate": 1.4841607966320598e-05, "loss": 0.0754, "step": 5304 }, { "epoch": 5.38, "learning_rate": 1.4839689750965885e-05, "loss": 0.0892, "step": 5305 }, { "epoch": 5.38, "learning_rate": 1.48377713030355e-05, "loss": 0.097, "step": 5306 }, { "epoch": 5.39, "learning_rate": 1.4835852622621643e-05, "loss": 0.0878, "step": 5307 }, { "epoch": 5.39, "learning_rate": 1.4833933709816517e-05, "loss": 0.0977, "step": 5308 }, { "epoch": 5.39, "learning_rate": 1.4832014564712336e-05, "loss": 0.0831, "step": 5309 }, { "epoch": 5.39, "learning_rate": 1.4830095187401327e-05, "loss": 0.0868, "step": 5310 }, { "epoch": 5.39, "learning_rate": 1.482817557797573e-05, "loss": 0.0794, "step": 5311 }, { "epoch": 5.39, "learning_rate": 1.4826255736527788e-05, "loss": 0.0779, "step": 5312 }, { "epoch": 5.39, "learning_rate": 1.4824335663149762e-05, "loss": 0.0788, "step": 5313 }, { "epoch": 5.39, "learning_rate": 1.4822415357933926e-05, "loss": 0.0816, "step": 5314 }, { "epoch": 5.39, "learning_rate": 1.4820494820972562e-05, "loss": 0.0743, "step": 5315 }, { "epoch": 5.39, "learning_rate": 1.481857405235796e-05, "loss": 0.0881, "step": 5316 }, { "epoch": 5.4, "learning_rate": 1.4816653052182428e-05, "loss": 0.0907, "step": 5317 }, { "epoch": 5.4, "learning_rate": 1.4814731820538275e-05, "loss": 0.078, "step": 5318 }, { "epoch": 5.4, "learning_rate": 1.4812810357517832e-05, "loss": 0.0682, "step": 5319 }, { "epoch": 5.4, "learning_rate": 1.4810888663213441e-05, "loss": 0.0797, "step": 5320 }, { "epoch": 5.4, "learning_rate": 1.4808966737717444e-05, "loss": 0.0902, "step": 5321 }, { "epoch": 5.4, "learning_rate": 1.4807044581122201e-05, "loss": 0.0868, "step": 5322 }, { "epoch": 5.4, "learning_rate": 1.4805122193520088e-05, "loss": 0.0862, "step": 5323 }, { "epoch": 5.4, "learning_rate": 1.480319957500348e-05, "loss": 0.0886, "step": 5324 }, { "epoch": 5.4, "learning_rate": 1.4801276725664777e-05, "loss": 0.0821, "step": 5325 }, { "epoch": 5.4, "learning_rate": 1.4799353645596379e-05, "loss": 0.0903, "step": 5326 }, { "epoch": 5.41, "learning_rate": 1.4797430334890704e-05, "loss": 0.0873, "step": 5327 }, { "epoch": 5.41, "learning_rate": 1.4795506793640177e-05, "loss": 0.1047, "step": 5328 }, { "epoch": 5.41, "learning_rate": 1.4793583021937237e-05, "loss": 0.095, "step": 5329 }, { "epoch": 5.41, "learning_rate": 1.479165901987433e-05, "loss": 0.0749, "step": 5330 }, { "epoch": 5.41, "learning_rate": 1.4789734787543914e-05, "loss": 0.0619, "step": 5331 }, { "epoch": 5.41, "learning_rate": 1.4787810325038465e-05, "loss": 0.0852, "step": 5332 }, { "epoch": 5.41, "learning_rate": 1.4785885632450465e-05, "loss": 0.0825, "step": 5333 }, { "epoch": 5.41, "learning_rate": 1.4783960709872402e-05, "loss": 0.0796, "step": 5334 }, { "epoch": 5.41, "learning_rate": 1.4782035557396783e-05, "loss": 0.07, "step": 5335 }, { "epoch": 5.41, "learning_rate": 1.478011017511612e-05, "loss": 0.0909, "step": 5336 }, { "epoch": 5.42, "learning_rate": 1.4778184563122945e-05, "loss": 0.082, "step": 5337 }, { "epoch": 5.42, "learning_rate": 1.477625872150979e-05, "loss": 0.0873, "step": 5338 }, { "epoch": 5.42, "learning_rate": 1.4774332650369201e-05, "loss": 0.0711, "step": 5339 }, { "epoch": 5.42, "learning_rate": 1.4772406349793744e-05, "loss": 0.0852, "step": 5340 }, { "epoch": 5.42, "learning_rate": 1.4770479819875983e-05, "loss": 0.0891, "step": 5341 }, { "epoch": 5.42, "learning_rate": 1.4768553060708504e-05, "loss": 0.0852, "step": 5342 }, { "epoch": 5.42, "learning_rate": 1.4766626072383896e-05, "loss": 0.0925, "step": 5343 }, { "epoch": 5.42, "learning_rate": 1.4764698854994763e-05, "loss": 0.0871, "step": 5344 }, { "epoch": 5.42, "learning_rate": 1.4762771408633718e-05, "loss": 0.0979, "step": 5345 }, { "epoch": 5.42, "learning_rate": 1.4760843733393388e-05, "loss": 0.0725, "step": 5346 }, { "epoch": 5.43, "learning_rate": 1.475891582936641e-05, "loss": 0.087, "step": 5347 }, { "epoch": 5.43, "learning_rate": 1.4756987696645424e-05, "loss": 0.0703, "step": 5348 }, { "epoch": 5.43, "learning_rate": 1.4755059335323098e-05, "loss": 0.0935, "step": 5349 }, { "epoch": 5.43, "learning_rate": 1.47531307454921e-05, "loss": 0.097, "step": 5350 }, { "epoch": 5.43, "learning_rate": 1.4751201927245105e-05, "loss": 0.0735, "step": 5351 }, { "epoch": 5.43, "learning_rate": 1.4749272880674803e-05, "loss": 0.1022, "step": 5352 }, { "epoch": 5.43, "learning_rate": 1.47473436058739e-05, "loss": 0.0941, "step": 5353 }, { "epoch": 5.43, "learning_rate": 1.4745414102935109e-05, "loss": 0.0809, "step": 5354 }, { "epoch": 5.43, "learning_rate": 1.4743484371951156e-05, "loss": 0.0755, "step": 5355 }, { "epoch": 5.43, "learning_rate": 1.474155441301477e-05, "loss": 0.0989, "step": 5356 }, { "epoch": 5.44, "learning_rate": 1.4739624226218701e-05, "loss": 0.0891, "step": 5357 }, { "epoch": 5.44, "learning_rate": 1.4737693811655702e-05, "loss": 0.0661, "step": 5358 }, { "epoch": 5.44, "learning_rate": 1.4735763169418549e-05, "loss": 0.0792, "step": 5359 }, { "epoch": 5.44, "learning_rate": 1.4733832299600013e-05, "loss": 0.0715, "step": 5360 }, { "epoch": 5.44, "learning_rate": 1.4731901202292886e-05, "loss": 0.0911, "step": 5361 }, { "epoch": 5.44, "learning_rate": 1.472996987758997e-05, "loss": 0.0864, "step": 5362 }, { "epoch": 5.44, "learning_rate": 1.4728038325584075e-05, "loss": 0.0923, "step": 5363 }, { "epoch": 5.44, "learning_rate": 1.4726106546368024e-05, "loss": 0.0806, "step": 5364 }, { "epoch": 5.44, "learning_rate": 1.4724174540034652e-05, "loss": 0.0779, "step": 5365 }, { "epoch": 5.44, "learning_rate": 1.4722242306676799e-05, "loss": 0.078, "step": 5366 }, { "epoch": 5.45, "learning_rate": 1.4720309846387327e-05, "loss": 0.0863, "step": 5367 }, { "epoch": 5.45, "learning_rate": 1.4718377159259094e-05, "loss": 0.0839, "step": 5368 }, { "epoch": 5.45, "learning_rate": 1.4716444245384981e-05, "loss": 0.0968, "step": 5369 }, { "epoch": 5.45, "learning_rate": 1.4714511104857877e-05, "loss": 0.0761, "step": 5370 }, { "epoch": 5.45, "learning_rate": 1.4712577737770682e-05, "loss": 0.0902, "step": 5371 }, { "epoch": 5.45, "learning_rate": 1.4710644144216302e-05, "loss": 0.0952, "step": 5372 }, { "epoch": 5.45, "learning_rate": 1.4708710324287658e-05, "loss": 0.0856, "step": 5373 }, { "epoch": 5.45, "learning_rate": 1.4706776278077681e-05, "loss": 0.0724, "step": 5374 }, { "epoch": 5.45, "learning_rate": 1.4704842005679319e-05, "loss": 0.0869, "step": 5375 }, { "epoch": 5.46, "learning_rate": 1.470290750718552e-05, "loss": 0.088, "step": 5376 }, { "epoch": 5.46, "learning_rate": 1.4700972782689247e-05, "loss": 0.0706, "step": 5377 }, { "epoch": 5.46, "learning_rate": 1.4699037832283478e-05, "loss": 0.0751, "step": 5378 }, { "epoch": 5.46, "learning_rate": 1.46971026560612e-05, "loss": 0.0764, "step": 5379 }, { "epoch": 5.46, "learning_rate": 1.4695167254115403e-05, "loss": 0.0881, "step": 5380 }, { "epoch": 5.46, "learning_rate": 1.4693231626539102e-05, "loss": 0.0862, "step": 5381 }, { "epoch": 5.46, "learning_rate": 1.4691295773425314e-05, "loss": 0.0799, "step": 5382 }, { "epoch": 5.46, "learning_rate": 1.4689359694867064e-05, "loss": 0.0891, "step": 5383 }, { "epoch": 5.46, "learning_rate": 1.4687423390957398e-05, "loss": 0.0814, "step": 5384 }, { "epoch": 5.46, "learning_rate": 1.4685486861789358e-05, "loss": 0.09, "step": 5385 }, { "epoch": 5.47, "learning_rate": 1.4683550107456012e-05, "loss": 0.1023, "step": 5386 }, { "epoch": 5.47, "learning_rate": 1.4681613128050435e-05, "loss": 0.0938, "step": 5387 }, { "epoch": 5.47, "learning_rate": 1.4679675923665706e-05, "loss": 0.0908, "step": 5388 }, { "epoch": 5.47, "learning_rate": 1.4677738494394917e-05, "loss": 0.0733, "step": 5389 }, { "epoch": 5.47, "learning_rate": 1.4675800840331178e-05, "loss": 0.0706, "step": 5390 }, { "epoch": 5.47, "learning_rate": 1.4673862961567602e-05, "loss": 0.0926, "step": 5391 }, { "epoch": 5.47, "learning_rate": 1.4671924858197317e-05, "loss": 0.0903, "step": 5392 }, { "epoch": 5.47, "learning_rate": 1.4669986530313457e-05, "loss": 0.1041, "step": 5393 }, { "epoch": 5.47, "learning_rate": 1.4668047978009177e-05, "loss": 0.0803, "step": 5394 }, { "epoch": 5.47, "learning_rate": 1.4666109201377627e-05, "loss": 0.0953, "step": 5395 }, { "epoch": 5.48, "learning_rate": 1.4664170200511983e-05, "loss": 0.079, "step": 5396 }, { "epoch": 5.48, "learning_rate": 1.4662230975505423e-05, "loss": 0.0872, "step": 5397 }, { "epoch": 5.48, "learning_rate": 1.4660291526451137e-05, "loss": 0.0883, "step": 5398 }, { "epoch": 5.48, "learning_rate": 1.4658351853442331e-05, "loss": 0.0905, "step": 5399 }, { "epoch": 5.48, "learning_rate": 1.4656411956572216e-05, "loss": 0.0922, "step": 5400 }, { "epoch": 5.48, "learning_rate": 1.4654471835934016e-05, "loss": 0.0887, "step": 5401 }, { "epoch": 5.48, "learning_rate": 1.465253149162096e-05, "loss": 0.0769, "step": 5402 }, { "epoch": 5.48, "learning_rate": 1.4650590923726298e-05, "loss": 0.102, "step": 5403 }, { "epoch": 5.48, "learning_rate": 1.4648650132343288e-05, "loss": 0.0935, "step": 5404 }, { "epoch": 5.48, "learning_rate": 1.464670911756519e-05, "loss": 0.0953, "step": 5405 }, { "epoch": 5.49, "learning_rate": 1.4644767879485284e-05, "loss": 0.066, "step": 5406 }, { "epoch": 5.49, "learning_rate": 1.4642826418196862e-05, "loss": 0.0821, "step": 5407 }, { "epoch": 5.49, "learning_rate": 1.4640884733793216e-05, "loss": 0.0779, "step": 5408 }, { "epoch": 5.49, "learning_rate": 1.4638942826367662e-05, "loss": 0.0786, "step": 5409 }, { "epoch": 5.49, "learning_rate": 1.4637000696013512e-05, "loss": 0.0758, "step": 5410 }, { "epoch": 5.49, "learning_rate": 1.4635058342824107e-05, "loss": 0.0805, "step": 5411 }, { "epoch": 5.49, "learning_rate": 1.463311576689278e-05, "loss": 0.0864, "step": 5412 }, { "epoch": 5.49, "learning_rate": 1.4631172968312886e-05, "loss": 0.0787, "step": 5413 }, { "epoch": 5.49, "learning_rate": 1.4629229947177788e-05, "loss": 0.1039, "step": 5414 }, { "epoch": 5.49, "learning_rate": 1.4627286703580863e-05, "loss": 0.0808, "step": 5415 }, { "epoch": 5.5, "learning_rate": 1.4625343237615491e-05, "loss": 0.0647, "step": 5416 }, { "epoch": 5.5, "learning_rate": 1.4623399549375068e-05, "loss": 0.0751, "step": 5417 }, { "epoch": 5.5, "learning_rate": 1.4621455638953e-05, "loss": 0.0768, "step": 5418 }, { "epoch": 5.5, "learning_rate": 1.4619511506442701e-05, "loss": 0.0643, "step": 5419 }, { "epoch": 5.5, "learning_rate": 1.4617567151937602e-05, "loss": 0.0909, "step": 5420 }, { "epoch": 5.5, "learning_rate": 1.4615622575531138e-05, "loss": 0.0853, "step": 5421 }, { "epoch": 5.5, "learning_rate": 1.4613677777316761e-05, "loss": 0.0953, "step": 5422 }, { "epoch": 5.5, "learning_rate": 1.4611732757387927e-05, "loss": 0.0748, "step": 5423 }, { "epoch": 5.5, "learning_rate": 1.4609787515838106e-05, "loss": 0.0741, "step": 5424 }, { "epoch": 5.5, "learning_rate": 1.4607842052760777e-05, "loss": 0.0757, "step": 5425 }, { "epoch": 5.51, "learning_rate": 1.4605896368249435e-05, "loss": 0.0906, "step": 5426 }, { "epoch": 5.51, "learning_rate": 1.4603950462397576e-05, "loss": 0.0839, "step": 5427 }, { "epoch": 5.51, "learning_rate": 1.4602004335298717e-05, "loss": 0.1035, "step": 5428 }, { "epoch": 5.51, "learning_rate": 1.460005798704638e-05, "loss": 0.1021, "step": 5429 }, { "epoch": 5.51, "learning_rate": 1.4598111417734098e-05, "loss": 0.0737, "step": 5430 }, { "epoch": 5.51, "learning_rate": 1.4596164627455411e-05, "loss": 0.1022, "step": 5431 }, { "epoch": 5.51, "learning_rate": 1.4594217616303883e-05, "loss": 0.0854, "step": 5432 }, { "epoch": 5.51, "learning_rate": 1.4592270384373072e-05, "loss": 0.0956, "step": 5433 }, { "epoch": 5.51, "learning_rate": 1.4590322931756557e-05, "loss": 0.0835, "step": 5434 }, { "epoch": 5.51, "learning_rate": 1.4588375258547923e-05, "loss": 0.088, "step": 5435 }, { "epoch": 5.52, "learning_rate": 1.4586427364840767e-05, "loss": 0.106, "step": 5436 }, { "epoch": 5.52, "learning_rate": 1.4584479250728698e-05, "loss": 0.0888, "step": 5437 }, { "epoch": 5.52, "learning_rate": 1.4582530916305335e-05, "loss": 0.0965, "step": 5438 }, { "epoch": 5.52, "learning_rate": 1.4580582361664306e-05, "loss": 0.0896, "step": 5439 }, { "epoch": 5.52, "learning_rate": 1.4578633586899251e-05, "loss": 0.0886, "step": 5440 }, { "epoch": 5.52, "learning_rate": 1.4576684592103818e-05, "loss": 0.0989, "step": 5441 }, { "epoch": 5.52, "learning_rate": 1.457473537737167e-05, "loss": 0.108, "step": 5442 }, { "epoch": 5.52, "learning_rate": 1.4572785942796479e-05, "loss": 0.0936, "step": 5443 }, { "epoch": 5.52, "learning_rate": 1.4570836288471925e-05, "loss": 0.1019, "step": 5444 }, { "epoch": 5.53, "learning_rate": 1.45688864144917e-05, "loss": 0.1002, "step": 5445 }, { "epoch": 5.53, "learning_rate": 1.4566936320949511e-05, "loss": 0.0913, "step": 5446 }, { "epoch": 5.53, "learning_rate": 1.4564986007939066e-05, "loss": 0.0865, "step": 5447 }, { "epoch": 5.53, "learning_rate": 1.456303547555409e-05, "loss": 0.0967, "step": 5448 }, { "epoch": 5.53, "learning_rate": 1.456108472388832e-05, "loss": 0.0967, "step": 5449 }, { "epoch": 5.53, "learning_rate": 1.4559133753035503e-05, "loss": 0.0908, "step": 5450 }, { "epoch": 5.53, "learning_rate": 1.4557182563089391e-05, "loss": 0.0993, "step": 5451 }, { "epoch": 5.53, "learning_rate": 1.455523115414375e-05, "loss": 0.0936, "step": 5452 }, { "epoch": 5.53, "learning_rate": 1.4553279526292357e-05, "loss": 0.086, "step": 5453 }, { "epoch": 5.53, "learning_rate": 1.4551327679629e-05, "loss": 0.0957, "step": 5454 }, { "epoch": 5.54, "learning_rate": 1.4549375614247478e-05, "loss": 0.0919, "step": 5455 }, { "epoch": 5.54, "learning_rate": 1.4547423330241599e-05, "loss": 0.0813, "step": 5456 }, { "epoch": 5.54, "learning_rate": 1.454547082770518e-05, "loss": 0.0813, "step": 5457 }, { "epoch": 5.54, "learning_rate": 1.4543518106732048e-05, "loss": 0.0767, "step": 5458 }, { "epoch": 5.54, "learning_rate": 1.454156516741605e-05, "loss": 0.0837, "step": 5459 }, { "epoch": 5.54, "learning_rate": 1.453961200985103e-05, "loss": 0.1012, "step": 5460 }, { "epoch": 5.54, "learning_rate": 1.4537658634130853e-05, "loss": 0.0832, "step": 5461 }, { "epoch": 5.54, "learning_rate": 1.4535705040349388e-05, "loss": 0.0813, "step": 5462 }, { "epoch": 5.54, "learning_rate": 1.4533751228600515e-05, "loss": 0.1035, "step": 5463 }, { "epoch": 5.54, "learning_rate": 1.453179719897813e-05, "loss": 0.0992, "step": 5464 }, { "epoch": 5.55, "learning_rate": 1.452984295157613e-05, "loss": 0.0761, "step": 5465 }, { "epoch": 5.55, "learning_rate": 1.4527888486488436e-05, "loss": 0.0712, "step": 5466 }, { "epoch": 5.55, "learning_rate": 1.4525933803808967e-05, "loss": 0.07, "step": 5467 }, { "epoch": 5.55, "learning_rate": 1.4523978903631655e-05, "loss": 0.078, "step": 5468 }, { "epoch": 5.55, "learning_rate": 1.4522023786050448e-05, "loss": 0.1002, "step": 5469 }, { "epoch": 5.55, "learning_rate": 1.45200684511593e-05, "loss": 0.0851, "step": 5470 }, { "epoch": 5.55, "learning_rate": 1.4518112899052177e-05, "loss": 0.0894, "step": 5471 }, { "epoch": 5.55, "learning_rate": 1.4516157129823055e-05, "loss": 0.0754, "step": 5472 }, { "epoch": 5.55, "learning_rate": 1.4514201143565917e-05, "loss": 0.0718, "step": 5473 }, { "epoch": 5.55, "learning_rate": 1.4512244940374762e-05, "loss": 0.0781, "step": 5474 }, { "epoch": 5.56, "learning_rate": 1.4510288520343598e-05, "loss": 0.097, "step": 5475 }, { "epoch": 5.56, "learning_rate": 1.4508331883566445e-05, "loss": 0.0767, "step": 5476 }, { "epoch": 5.56, "learning_rate": 1.4506375030137321e-05, "loss": 0.0753, "step": 5477 }, { "epoch": 5.56, "learning_rate": 1.4504417960150277e-05, "loss": 0.1078, "step": 5478 }, { "epoch": 5.56, "learning_rate": 1.4502460673699355e-05, "loss": 0.0773, "step": 5479 }, { "epoch": 5.56, "learning_rate": 1.450050317087861e-05, "loss": 0.092, "step": 5480 }, { "epoch": 5.56, "learning_rate": 1.449854545178212e-05, "loss": 0.0911, "step": 5481 }, { "epoch": 5.56, "learning_rate": 1.4496587516503961e-05, "loss": 0.0824, "step": 5482 }, { "epoch": 5.56, "learning_rate": 1.4494629365138227e-05, "loss": 0.085, "step": 5483 }, { "epoch": 5.56, "learning_rate": 1.4492670997779012e-05, "loss": 0.0915, "step": 5484 }, { "epoch": 5.57, "learning_rate": 1.4490712414520431e-05, "loss": 0.0757, "step": 5485 }, { "epoch": 5.57, "learning_rate": 1.4488753615456603e-05, "loss": 0.11, "step": 5486 }, { "epoch": 5.57, "learning_rate": 1.4486794600681667e-05, "loss": 0.105, "step": 5487 }, { "epoch": 5.57, "learning_rate": 1.4484835370289758e-05, "loss": 0.1062, "step": 5488 }, { "epoch": 5.57, "learning_rate": 1.4482875924375032e-05, "loss": 0.0884, "step": 5489 }, { "epoch": 5.57, "learning_rate": 1.4480916263031648e-05, "loss": 0.0868, "step": 5490 }, { "epoch": 5.57, "learning_rate": 1.4478956386353783e-05, "loss": 0.1043, "step": 5491 }, { "epoch": 5.57, "learning_rate": 1.4476996294435621e-05, "loss": 0.0781, "step": 5492 }, { "epoch": 5.57, "learning_rate": 1.4475035987371355e-05, "loss": 0.0716, "step": 5493 }, { "epoch": 5.57, "learning_rate": 1.4473075465255188e-05, "loss": 0.092, "step": 5494 }, { "epoch": 5.58, "learning_rate": 1.4471114728181337e-05, "loss": 0.0864, "step": 5495 }, { "epoch": 5.58, "learning_rate": 1.4469153776244025e-05, "loss": 0.0768, "step": 5496 }, { "epoch": 5.58, "learning_rate": 1.4467192609537488e-05, "loss": 0.0837, "step": 5497 }, { "epoch": 5.58, "learning_rate": 1.446523122815597e-05, "loss": 0.0845, "step": 5498 }, { "epoch": 5.58, "learning_rate": 1.4463269632193731e-05, "loss": 0.1085, "step": 5499 }, { "epoch": 5.58, "learning_rate": 1.4461307821745033e-05, "loss": 0.1006, "step": 5500 }, { "epoch": 5.58, "learning_rate": 1.4459345796904155e-05, "loss": 0.0876, "step": 5501 }, { "epoch": 5.58, "learning_rate": 1.4457383557765385e-05, "loss": 0.0825, "step": 5502 }, { "epoch": 5.58, "learning_rate": 1.4455421104423016e-05, "loss": 0.0869, "step": 5503 }, { "epoch": 5.58, "learning_rate": 1.445345843697136e-05, "loss": 0.0846, "step": 5504 }, { "epoch": 5.59, "learning_rate": 1.4451495555504734e-05, "loss": 0.0864, "step": 5505 }, { "epoch": 5.59, "learning_rate": 1.4449532460117459e-05, "loss": 0.0929, "step": 5506 }, { "epoch": 5.59, "learning_rate": 1.4447569150903884e-05, "loss": 0.1043, "step": 5507 }, { "epoch": 5.59, "learning_rate": 1.444560562795835e-05, "loss": 0.0847, "step": 5508 }, { "epoch": 5.59, "learning_rate": 1.4443641891375223e-05, "loss": 0.0908, "step": 5509 }, { "epoch": 5.59, "learning_rate": 1.4441677941248862e-05, "loss": 0.0793, "step": 5510 }, { "epoch": 5.59, "learning_rate": 1.4439713777673657e-05, "loss": 0.0793, "step": 5511 }, { "epoch": 5.59, "learning_rate": 1.4437749400743993e-05, "loss": 0.0948, "step": 5512 }, { "epoch": 5.59, "learning_rate": 1.4435784810554263e-05, "loss": 0.0757, "step": 5513 }, { "epoch": 5.6, "learning_rate": 1.4433820007198891e-05, "loss": 0.0855, "step": 5514 }, { "epoch": 5.6, "learning_rate": 1.4431854990772288e-05, "loss": 0.0953, "step": 5515 }, { "epoch": 5.6, "learning_rate": 1.4429889761368886e-05, "loss": 0.0951, "step": 5516 }, { "epoch": 5.6, "learning_rate": 1.4427924319083127e-05, "loss": 0.0886, "step": 5517 }, { "epoch": 5.6, "learning_rate": 1.4425958664009463e-05, "loss": 0.0907, "step": 5518 }, { "epoch": 5.6, "learning_rate": 1.442399279624235e-05, "loss": 0.0861, "step": 5519 }, { "epoch": 5.6, "learning_rate": 1.4422026715876268e-05, "loss": 0.0878, "step": 5520 }, { "epoch": 5.6, "learning_rate": 1.4420060423005696e-05, "loss": 0.0968, "step": 5521 }, { "epoch": 5.6, "learning_rate": 1.441809391772512e-05, "loss": 0.0788, "step": 5522 }, { "epoch": 5.6, "learning_rate": 1.4416127200129048e-05, "loss": 0.0882, "step": 5523 }, { "epoch": 5.61, "learning_rate": 1.4414160270311994e-05, "loss": 0.094, "step": 5524 }, { "epoch": 5.61, "learning_rate": 1.4412193128368475e-05, "loss": 0.08, "step": 5525 }, { "epoch": 5.61, "learning_rate": 1.4410225774393024e-05, "loss": 0.0854, "step": 5526 }, { "epoch": 5.61, "learning_rate": 1.440825820848019e-05, "loss": 0.0883, "step": 5527 }, { "epoch": 5.61, "learning_rate": 1.4406290430724523e-05, "loss": 0.0788, "step": 5528 }, { "epoch": 5.61, "learning_rate": 1.4404322441220586e-05, "loss": 0.0922, "step": 5529 }, { "epoch": 5.61, "learning_rate": 1.440235424006295e-05, "loss": 0.1043, "step": 5530 }, { "epoch": 5.61, "learning_rate": 1.4400385827346202e-05, "loss": 0.0748, "step": 5531 }, { "epoch": 5.61, "learning_rate": 1.4398417203164936e-05, "loss": 0.0931, "step": 5532 }, { "epoch": 5.61, "learning_rate": 1.4396448367613754e-05, "loss": 0.0854, "step": 5533 }, { "epoch": 5.62, "learning_rate": 1.4394479320787274e-05, "loss": 0.0825, "step": 5534 }, { "epoch": 5.62, "learning_rate": 1.4392510062780112e-05, "loss": 0.0875, "step": 5535 }, { "epoch": 5.62, "learning_rate": 1.439054059368691e-05, "loss": 0.0797, "step": 5536 }, { "epoch": 5.62, "learning_rate": 1.4388570913602313e-05, "loss": 0.095, "step": 5537 }, { "epoch": 5.62, "learning_rate": 1.4386601022620972e-05, "loss": 0.084, "step": 5538 }, { "epoch": 5.62, "learning_rate": 1.4384630920837555e-05, "loss": 0.0963, "step": 5539 }, { "epoch": 5.62, "learning_rate": 1.4382660608346733e-05, "loss": 0.1039, "step": 5540 }, { "epoch": 5.62, "learning_rate": 1.4380690085243192e-05, "loss": 0.0904, "step": 5541 }, { "epoch": 5.62, "learning_rate": 1.437871935162163e-05, "loss": 0.0853, "step": 5542 }, { "epoch": 5.62, "learning_rate": 1.4376748407576756e-05, "loss": 0.0962, "step": 5543 }, { "epoch": 5.63, "learning_rate": 1.4374777253203273e-05, "loss": 0.0886, "step": 5544 }, { "epoch": 5.63, "learning_rate": 1.4372805888595918e-05, "loss": 0.0809, "step": 5545 }, { "epoch": 5.63, "learning_rate": 1.4370834313849422e-05, "loss": 0.0974, "step": 5546 }, { "epoch": 5.63, "learning_rate": 1.4368862529058532e-05, "loss": 0.099, "step": 5547 }, { "epoch": 5.63, "learning_rate": 1.4366890534318003e-05, "loss": 0.0887, "step": 5548 }, { "epoch": 5.63, "learning_rate": 1.43649183297226e-05, "loss": 0.0828, "step": 5549 }, { "epoch": 5.63, "learning_rate": 1.4362945915367106e-05, "loss": 0.0909, "step": 5550 }, { "epoch": 5.63, "learning_rate": 1.43609732913463e-05, "loss": 0.1047, "step": 5551 }, { "epoch": 5.63, "learning_rate": 1.4359000457754974e-05, "loss": 0.0868, "step": 5552 }, { "epoch": 5.63, "learning_rate": 1.4357027414687945e-05, "loss": 0.1018, "step": 5553 }, { "epoch": 5.64, "learning_rate": 1.4355054162240025e-05, "loss": 0.0791, "step": 5554 }, { "epoch": 5.64, "learning_rate": 1.4353080700506041e-05, "loss": 0.0852, "step": 5555 }, { "epoch": 5.64, "learning_rate": 1.4351107029580826e-05, "loss": 0.0999, "step": 5556 }, { "epoch": 5.64, "learning_rate": 1.4349133149559232e-05, "loss": 0.0933, "step": 5557 }, { "epoch": 5.64, "learning_rate": 1.4347159060536108e-05, "loss": 0.0977, "step": 5558 }, { "epoch": 5.64, "learning_rate": 1.434518476260633e-05, "loss": 0.0855, "step": 5559 }, { "epoch": 5.64, "learning_rate": 1.434321025586477e-05, "loss": 0.1083, "step": 5560 }, { "epoch": 5.64, "learning_rate": 1.4341235540406314e-05, "loss": 0.0726, "step": 5561 }, { "epoch": 5.64, "learning_rate": 1.4339260616325859e-05, "loss": 0.0971, "step": 5562 }, { "epoch": 5.64, "learning_rate": 1.433728548371831e-05, "loss": 0.1008, "step": 5563 }, { "epoch": 5.65, "learning_rate": 1.4335310142678589e-05, "loss": 0.0966, "step": 5564 }, { "epoch": 5.65, "learning_rate": 1.4333334593301617e-05, "loss": 0.1003, "step": 5565 }, { "epoch": 5.65, "learning_rate": 1.4331358835682336e-05, "loss": 0.0877, "step": 5566 }, { "epoch": 5.65, "learning_rate": 1.4329382869915691e-05, "loss": 0.0781, "step": 5567 }, { "epoch": 5.65, "learning_rate": 1.4327406696096639e-05, "loss": 0.0837, "step": 5568 }, { "epoch": 5.65, "learning_rate": 1.4325430314320143e-05, "loss": 0.088, "step": 5569 }, { "epoch": 5.65, "learning_rate": 1.4323453724681185e-05, "loss": 0.0973, "step": 5570 }, { "epoch": 5.65, "learning_rate": 1.432147692727475e-05, "loss": 0.0867, "step": 5571 }, { "epoch": 5.65, "learning_rate": 1.4319499922195833e-05, "loss": 0.0945, "step": 5572 }, { "epoch": 5.65, "learning_rate": 1.4317522709539442e-05, "loss": 0.0915, "step": 5573 }, { "epoch": 5.66, "learning_rate": 1.4315545289400596e-05, "loss": 0.0952, "step": 5574 }, { "epoch": 5.66, "learning_rate": 1.4313567661874318e-05, "loss": 0.094, "step": 5575 }, { "epoch": 5.66, "learning_rate": 1.4311589827055647e-05, "loss": 0.0715, "step": 5576 }, { "epoch": 5.66, "learning_rate": 1.4309611785039628e-05, "loss": 0.0783, "step": 5577 }, { "epoch": 5.66, "learning_rate": 1.4307633535921319e-05, "loss": 0.0965, "step": 5578 }, { "epoch": 5.66, "learning_rate": 1.430565507979579e-05, "loss": 0.0965, "step": 5579 }, { "epoch": 5.66, "learning_rate": 1.4303676416758107e-05, "loss": 0.1059, "step": 5580 }, { "epoch": 5.66, "learning_rate": 1.4301697546903367e-05, "loss": 0.0863, "step": 5581 }, { "epoch": 5.66, "learning_rate": 1.4299718470326664e-05, "loss": 0.0811, "step": 5582 }, { "epoch": 5.67, "learning_rate": 1.42977391871231e-05, "loss": 0.0866, "step": 5583 }, { "epoch": 5.67, "learning_rate": 1.4295759697387798e-05, "loss": 0.096, "step": 5584 }, { "epoch": 5.67, "learning_rate": 1.4293780001215876e-05, "loss": 0.075, "step": 5585 }, { "epoch": 5.67, "learning_rate": 1.4291800098702475e-05, "loss": 0.0935, "step": 5586 }, { "epoch": 5.67, "learning_rate": 1.4289819989942742e-05, "loss": 0.0835, "step": 5587 }, { "epoch": 5.67, "learning_rate": 1.4287839675031832e-05, "loss": 0.0916, "step": 5588 }, { "epoch": 5.67, "learning_rate": 1.4285859154064908e-05, "loss": 0.0786, "step": 5589 }, { "epoch": 5.67, "learning_rate": 1.4283878427137149e-05, "loss": 0.0927, "step": 5590 }, { "epoch": 5.67, "learning_rate": 1.4281897494343735e-05, "loss": 0.0876, "step": 5591 }, { "epoch": 5.67, "learning_rate": 1.427991635577987e-05, "loss": 0.0844, "step": 5592 }, { "epoch": 5.68, "learning_rate": 1.4277935011540756e-05, "loss": 0.0861, "step": 5593 }, { "epoch": 5.68, "learning_rate": 1.4275953461721605e-05, "loss": 0.0938, "step": 5594 }, { "epoch": 5.68, "learning_rate": 1.4273971706417648e-05, "loss": 0.095, "step": 5595 }, { "epoch": 5.68, "learning_rate": 1.4271989745724115e-05, "loss": 0.084, "step": 5596 }, { "epoch": 5.68, "learning_rate": 1.4270007579736254e-05, "loss": 0.0783, "step": 5597 }, { "epoch": 5.68, "learning_rate": 1.4268025208549317e-05, "loss": 0.0712, "step": 5598 }, { "epoch": 5.68, "learning_rate": 1.426604263225857e-05, "loss": 0.0801, "step": 5599 }, { "epoch": 5.68, "learning_rate": 1.4264059850959288e-05, "loss": 0.0741, "step": 5600 }, { "epoch": 5.68, "learning_rate": 1.4262076864746757e-05, "loss": 0.0906, "step": 5601 }, { "epoch": 5.68, "learning_rate": 1.4260093673716266e-05, "loss": 0.0995, "step": 5602 }, { "epoch": 5.69, "learning_rate": 1.4258110277963124e-05, "loss": 0.0891, "step": 5603 }, { "epoch": 5.69, "learning_rate": 1.4256126677582644e-05, "loss": 0.083, "step": 5604 }, { "epoch": 5.69, "learning_rate": 1.4254142872670147e-05, "loss": 0.0849, "step": 5605 }, { "epoch": 5.69, "learning_rate": 1.4252158863320971e-05, "loss": 0.0825, "step": 5606 }, { "epoch": 5.69, "learning_rate": 1.4250174649630455e-05, "loss": 0.0801, "step": 5607 }, { "epoch": 5.69, "learning_rate": 1.4248190231693953e-05, "loss": 0.0825, "step": 5608 }, { "epoch": 5.69, "learning_rate": 1.4246205609606828e-05, "loss": 0.0869, "step": 5609 }, { "epoch": 5.69, "learning_rate": 1.4244220783464457e-05, "loss": 0.0808, "step": 5610 }, { "epoch": 5.69, "learning_rate": 1.4242235753362216e-05, "loss": 0.1074, "step": 5611 }, { "epoch": 5.69, "learning_rate": 1.4240250519395504e-05, "loss": 0.0868, "step": 5612 }, { "epoch": 5.7, "learning_rate": 1.4238265081659717e-05, "loss": 0.0783, "step": 5613 }, { "epoch": 5.7, "learning_rate": 1.4236279440250274e-05, "loss": 0.0792, "step": 5614 }, { "epoch": 5.7, "learning_rate": 1.4234293595262588e-05, "loss": 0.0983, "step": 5615 }, { "epoch": 5.7, "learning_rate": 1.4232307546792099e-05, "loss": 0.1028, "step": 5616 }, { "epoch": 5.7, "learning_rate": 1.4230321294934244e-05, "loss": 0.0841, "step": 5617 }, { "epoch": 5.7, "learning_rate": 1.4228334839784475e-05, "loss": 0.0947, "step": 5618 }, { "epoch": 5.7, "learning_rate": 1.4226348181438251e-05, "loss": 0.0888, "step": 5619 }, { "epoch": 5.7, "learning_rate": 1.4224361319991045e-05, "loss": 0.0682, "step": 5620 }, { "epoch": 5.7, "learning_rate": 1.4222374255538336e-05, "loss": 0.0824, "step": 5621 }, { "epoch": 5.7, "learning_rate": 1.4220386988175618e-05, "loss": 0.0925, "step": 5622 }, { "epoch": 5.71, "learning_rate": 1.4218399517998384e-05, "loss": 0.0998, "step": 5623 }, { "epoch": 5.71, "learning_rate": 1.4216411845102151e-05, "loss": 0.0855, "step": 5624 }, { "epoch": 5.71, "learning_rate": 1.4214423969582433e-05, "loss": 0.0889, "step": 5625 }, { "epoch": 5.71, "learning_rate": 1.4212435891534764e-05, "loss": 0.0941, "step": 5626 }, { "epoch": 5.71, "learning_rate": 1.4210447611054675e-05, "loss": 0.0868, "step": 5627 }, { "epoch": 5.71, "learning_rate": 1.4208459128237724e-05, "loss": 0.0846, "step": 5628 }, { "epoch": 5.71, "learning_rate": 1.4206470443179466e-05, "loss": 0.1034, "step": 5629 }, { "epoch": 5.71, "learning_rate": 1.4204481555975463e-05, "loss": 0.0918, "step": 5630 }, { "epoch": 5.71, "learning_rate": 1.42024924667213e-05, "loss": 0.0942, "step": 5631 }, { "epoch": 5.71, "learning_rate": 1.4200503175512563e-05, "loss": 0.0991, "step": 5632 }, { "epoch": 5.72, "learning_rate": 1.4198513682444847e-05, "loss": 0.0942, "step": 5633 }, { "epoch": 5.72, "learning_rate": 1.4196523987613764e-05, "loss": 0.097, "step": 5634 }, { "epoch": 5.72, "learning_rate": 1.4194534091114925e-05, "loss": 0.0837, "step": 5635 }, { "epoch": 5.72, "learning_rate": 1.4192543993043954e-05, "loss": 0.0762, "step": 5636 }, { "epoch": 5.72, "learning_rate": 1.4190553693496498e-05, "loss": 0.0772, "step": 5637 }, { "epoch": 5.72, "learning_rate": 1.4188563192568194e-05, "loss": 0.085, "step": 5638 }, { "epoch": 5.72, "learning_rate": 1.4186572490354699e-05, "loss": 0.094, "step": 5639 }, { "epoch": 5.72, "learning_rate": 1.4184581586951675e-05, "loss": 0.0888, "step": 5640 }, { "epoch": 5.72, "learning_rate": 1.4182590482454806e-05, "loss": 0.0718, "step": 5641 }, { "epoch": 5.73, "learning_rate": 1.4180599176959763e-05, "loss": 0.0904, "step": 5642 }, { "epoch": 5.73, "learning_rate": 1.4178607670562253e-05, "loss": 0.098, "step": 5643 }, { "epoch": 5.73, "learning_rate": 1.417661596335797e-05, "loss": 0.0875, "step": 5644 }, { "epoch": 5.73, "learning_rate": 1.4174624055442632e-05, "loss": 0.0918, "step": 5645 }, { "epoch": 5.73, "learning_rate": 1.4172631946911964e-05, "loss": 0.0966, "step": 5646 }, { "epoch": 5.73, "learning_rate": 1.4170639637861693e-05, "loss": 0.0976, "step": 5647 }, { "epoch": 5.73, "learning_rate": 1.416864712838756e-05, "loss": 0.0823, "step": 5648 }, { "epoch": 5.73, "learning_rate": 1.4166654418585326e-05, "loss": 0.0853, "step": 5649 }, { "epoch": 5.73, "learning_rate": 1.4164661508550748e-05, "loss": 0.1169, "step": 5650 }, { "epoch": 5.73, "learning_rate": 1.4162668398379593e-05, "loss": 0.0916, "step": 5651 }, { "epoch": 5.74, "learning_rate": 1.4160675088167646e-05, "loss": 0.0916, "step": 5652 }, { "epoch": 5.74, "learning_rate": 1.4158681578010694e-05, "loss": 0.0946, "step": 5653 }, { "epoch": 5.74, "learning_rate": 1.4156687868004543e-05, "loss": 0.0948, "step": 5654 }, { "epoch": 5.74, "learning_rate": 1.4154693958244998e-05, "loss": 0.0872, "step": 5655 }, { "epoch": 5.74, "learning_rate": 1.4152699848827878e-05, "loss": 0.0802, "step": 5656 }, { "epoch": 5.74, "learning_rate": 1.4150705539849012e-05, "loss": 0.0752, "step": 5657 }, { "epoch": 5.74, "learning_rate": 1.4148711031404238e-05, "loss": 0.0723, "step": 5658 }, { "epoch": 5.74, "learning_rate": 1.4146716323589407e-05, "loss": 0.0736, "step": 5659 }, { "epoch": 5.74, "learning_rate": 1.4144721416500375e-05, "loss": 0.082, "step": 5660 }, { "epoch": 5.74, "learning_rate": 1.4142726310233005e-05, "loss": 0.0998, "step": 5661 }, { "epoch": 5.75, "learning_rate": 1.4140731004883179e-05, "loss": 0.1197, "step": 5662 }, { "epoch": 5.75, "learning_rate": 1.4138735500546781e-05, "loss": 0.0817, "step": 5663 }, { "epoch": 5.75, "learning_rate": 1.4136739797319709e-05, "loss": 0.0958, "step": 5664 }, { "epoch": 5.75, "learning_rate": 1.4134743895297862e-05, "loss": 0.0847, "step": 5665 }, { "epoch": 5.75, "learning_rate": 1.4132747794577164e-05, "loss": 0.0851, "step": 5666 }, { "epoch": 5.75, "learning_rate": 1.4130751495253532e-05, "loss": 0.1065, "step": 5667 }, { "epoch": 5.75, "learning_rate": 1.4128754997422905e-05, "loss": 0.0831, "step": 5668 }, { "epoch": 5.75, "learning_rate": 1.412675830118122e-05, "loss": 0.0893, "step": 5669 }, { "epoch": 5.75, "learning_rate": 1.4124761406624435e-05, "loss": 0.0892, "step": 5670 }, { "epoch": 5.75, "learning_rate": 1.4122764313848514e-05, "loss": 0.1019, "step": 5671 }, { "epoch": 5.76, "learning_rate": 1.4120767022949426e-05, "loss": 0.0867, "step": 5672 }, { "epoch": 5.76, "learning_rate": 1.4118769534023151e-05, "loss": 0.0855, "step": 5673 }, { "epoch": 5.76, "learning_rate": 1.4116771847165685e-05, "loss": 0.1003, "step": 5674 }, { "epoch": 5.76, "learning_rate": 1.4114773962473025e-05, "loss": 0.0932, "step": 5675 }, { "epoch": 5.76, "learning_rate": 1.4112775880041182e-05, "loss": 0.0846, "step": 5676 }, { "epoch": 5.76, "learning_rate": 1.4110777599966179e-05, "loss": 0.0847, "step": 5677 }, { "epoch": 5.76, "learning_rate": 1.4108779122344038e-05, "loss": 0.1006, "step": 5678 }, { "epoch": 5.76, "learning_rate": 1.4106780447270807e-05, "loss": 0.0833, "step": 5679 }, { "epoch": 5.76, "learning_rate": 1.4104781574842524e-05, "loss": 0.0922, "step": 5680 }, { "epoch": 5.76, "learning_rate": 1.4102782505155255e-05, "loss": 0.0978, "step": 5681 }, { "epoch": 5.77, "learning_rate": 1.410078323830506e-05, "loss": 0.0929, "step": 5682 }, { "epoch": 5.77, "learning_rate": 1.4098783774388023e-05, "loss": 0.0795, "step": 5683 }, { "epoch": 5.77, "learning_rate": 1.4096784113500226e-05, "loss": 0.08, "step": 5684 }, { "epoch": 5.77, "learning_rate": 1.4094784255737762e-05, "loss": 0.0842, "step": 5685 }, { "epoch": 5.77, "learning_rate": 1.4092784201196738e-05, "loss": 0.1053, "step": 5686 }, { "epoch": 5.77, "learning_rate": 1.4090783949973272e-05, "loss": 0.0969, "step": 5687 }, { "epoch": 5.77, "learning_rate": 1.4088783502163486e-05, "loss": 0.1006, "step": 5688 }, { "epoch": 5.77, "learning_rate": 1.4086782857863513e-05, "loss": 0.0966, "step": 5689 }, { "epoch": 5.77, "learning_rate": 1.4084782017169492e-05, "loss": 0.1139, "step": 5690 }, { "epoch": 5.77, "learning_rate": 1.4082780980177579e-05, "loss": 0.081, "step": 5691 }, { "epoch": 5.78, "learning_rate": 1.4080779746983935e-05, "loss": 0.0765, "step": 5692 }, { "epoch": 5.78, "learning_rate": 1.4078778317684734e-05, "loss": 0.0814, "step": 5693 }, { "epoch": 5.78, "learning_rate": 1.4076776692376149e-05, "loss": 0.0933, "step": 5694 }, { "epoch": 5.78, "learning_rate": 1.4074774871154377e-05, "loss": 0.1022, "step": 5695 }, { "epoch": 5.78, "learning_rate": 1.4072772854115617e-05, "loss": 0.0779, "step": 5696 }, { "epoch": 5.78, "learning_rate": 1.407077064135607e-05, "loss": 0.0861, "step": 5697 }, { "epoch": 5.78, "learning_rate": 1.4068768232971966e-05, "loss": 0.0843, "step": 5698 }, { "epoch": 5.78, "learning_rate": 1.406676562905952e-05, "loss": 0.0925, "step": 5699 }, { "epoch": 5.78, "learning_rate": 1.406476282971498e-05, "loss": 0.0994, "step": 5700 }, { "epoch": 5.78, "learning_rate": 1.4062759835034588e-05, "loss": 0.0903, "step": 5701 }, { "epoch": 5.79, "learning_rate": 1.4060756645114595e-05, "loss": 0.0732, "step": 5702 }, { "epoch": 5.79, "learning_rate": 1.4058753260051273e-05, "loss": 0.1048, "step": 5703 }, { "epoch": 5.79, "learning_rate": 1.4056749679940893e-05, "loss": 0.0986, "step": 5704 }, { "epoch": 5.79, "learning_rate": 1.405474590487974e-05, "loss": 0.1008, "step": 5705 }, { "epoch": 5.79, "learning_rate": 1.4052741934964109e-05, "loss": 0.0704, "step": 5706 }, { "epoch": 5.79, "learning_rate": 1.4050737770290297e-05, "loss": 0.1034, "step": 5707 }, { "epoch": 5.79, "learning_rate": 1.4048733410954621e-05, "loss": 0.0872, "step": 5708 }, { "epoch": 5.79, "learning_rate": 1.40467288570534e-05, "loss": 0.0913, "step": 5709 }, { "epoch": 5.79, "learning_rate": 1.4044724108682967e-05, "loss": 0.1009, "step": 5710 }, { "epoch": 5.8, "learning_rate": 1.404271916593966e-05, "loss": 0.0948, "step": 5711 }, { "epoch": 5.8, "learning_rate": 1.4040714028919828e-05, "loss": 0.0987, "step": 5712 }, { "epoch": 5.8, "learning_rate": 1.4038708697719831e-05, "loss": 0.1077, "step": 5713 }, { "epoch": 5.8, "learning_rate": 1.4036703172436037e-05, "loss": 0.0907, "step": 5714 }, { "epoch": 5.8, "learning_rate": 1.403469745316482e-05, "loss": 0.0902, "step": 5715 }, { "epoch": 5.8, "learning_rate": 1.4032691540002571e-05, "loss": 0.1063, "step": 5716 }, { "epoch": 5.8, "learning_rate": 1.4030685433045687e-05, "loss": 0.0879, "step": 5717 }, { "epoch": 5.8, "learning_rate": 1.402867913239057e-05, "loss": 0.0899, "step": 5718 }, { "epoch": 5.8, "learning_rate": 1.4026672638133632e-05, "loss": 0.0984, "step": 5719 }, { "epoch": 5.8, "learning_rate": 1.40246659503713e-05, "loss": 0.0912, "step": 5720 }, { "epoch": 5.81, "learning_rate": 1.4022659069200012e-05, "loss": 0.0735, "step": 5721 }, { "epoch": 5.81, "learning_rate": 1.4020651994716202e-05, "loss": 0.0929, "step": 5722 }, { "epoch": 5.81, "learning_rate": 1.401864472701633e-05, "loss": 0.0821, "step": 5723 }, { "epoch": 5.81, "learning_rate": 1.4016637266196846e-05, "loss": 0.1074, "step": 5724 }, { "epoch": 5.81, "learning_rate": 1.401462961235423e-05, "loss": 0.0947, "step": 5725 }, { "epoch": 5.81, "learning_rate": 1.4012621765584957e-05, "loss": 0.1032, "step": 5726 }, { "epoch": 5.81, "learning_rate": 1.401061372598552e-05, "loss": 0.0789, "step": 5727 }, { "epoch": 5.81, "learning_rate": 1.4008605493652415e-05, "loss": 0.08, "step": 5728 }, { "epoch": 5.81, "learning_rate": 1.4006597068682148e-05, "loss": 0.0855, "step": 5729 }, { "epoch": 5.81, "learning_rate": 1.4004588451171236e-05, "loss": 0.0956, "step": 5730 }, { "epoch": 5.82, "learning_rate": 1.4002579641216207e-05, "loss": 0.089, "step": 5731 }, { "epoch": 5.82, "learning_rate": 1.4000570638913593e-05, "loss": 0.0972, "step": 5732 }, { "epoch": 5.82, "learning_rate": 1.3998561444359942e-05, "loss": 0.087, "step": 5733 }, { "epoch": 5.82, "learning_rate": 1.3996552057651806e-05, "loss": 0.1052, "step": 5734 }, { "epoch": 5.82, "learning_rate": 1.3994542478885746e-05, "loss": 0.101, "step": 5735 }, { "epoch": 5.82, "learning_rate": 1.3992532708158337e-05, "loss": 0.0902, "step": 5736 }, { "epoch": 5.82, "learning_rate": 1.3990522745566162e-05, "loss": 0.0822, "step": 5737 }, { "epoch": 5.82, "learning_rate": 1.3988512591205807e-05, "loss": 0.0847, "step": 5738 }, { "epoch": 5.82, "learning_rate": 1.3986502245173874e-05, "loss": 0.0837, "step": 5739 }, { "epoch": 5.82, "learning_rate": 1.3984491707566972e-05, "loss": 0.0767, "step": 5740 }, { "epoch": 5.83, "learning_rate": 1.3982480978481718e-05, "loss": 0.0905, "step": 5741 }, { "epoch": 5.83, "learning_rate": 1.3980470058014739e-05, "loss": 0.0791, "step": 5742 }, { "epoch": 5.83, "learning_rate": 1.3978458946262676e-05, "loss": 0.0899, "step": 5743 }, { "epoch": 5.83, "learning_rate": 1.3976447643322168e-05, "loss": 0.0975, "step": 5744 }, { "epoch": 5.83, "learning_rate": 1.3974436149289877e-05, "loss": 0.0861, "step": 5745 }, { "epoch": 5.83, "learning_rate": 1.3972424464262463e-05, "loss": 0.0716, "step": 5746 }, { "epoch": 5.83, "learning_rate": 1.3970412588336597e-05, "loss": 0.1005, "step": 5747 }, { "epoch": 5.83, "learning_rate": 1.3968400521608969e-05, "loss": 0.0984, "step": 5748 }, { "epoch": 5.83, "learning_rate": 1.3966388264176262e-05, "loss": 0.0859, "step": 5749 }, { "epoch": 5.83, "learning_rate": 1.3964375816135187e-05, "loss": 0.0927, "step": 5750 }, { "epoch": 5.84, "learning_rate": 1.3962363177582445e-05, "loss": 0.084, "step": 5751 }, { "epoch": 5.84, "learning_rate": 1.3960350348614757e-05, "loss": 0.0846, "step": 5752 }, { "epoch": 5.84, "learning_rate": 1.3958337329328855e-05, "loss": 0.0943, "step": 5753 }, { "epoch": 5.84, "learning_rate": 1.3956324119821473e-05, "loss": 0.0823, "step": 5754 }, { "epoch": 5.84, "learning_rate": 1.395431072018936e-05, "loss": 0.0878, "step": 5755 }, { "epoch": 5.84, "learning_rate": 1.3952297130529272e-05, "loss": 0.0845, "step": 5756 }, { "epoch": 5.84, "learning_rate": 1.3950283350937967e-05, "loss": 0.0853, "step": 5757 }, { "epoch": 5.84, "learning_rate": 1.3948269381512227e-05, "loss": 0.1019, "step": 5758 }, { "epoch": 5.84, "learning_rate": 1.3946255222348835e-05, "loss": 0.0731, "step": 5759 }, { "epoch": 5.84, "learning_rate": 1.394424087354458e-05, "loss": 0.0892, "step": 5760 }, { "epoch": 5.85, "learning_rate": 1.3942226335196261e-05, "loss": 0.088, "step": 5761 }, { "epoch": 5.85, "learning_rate": 1.3940211607400693e-05, "loss": 0.0899, "step": 5762 }, { "epoch": 5.85, "learning_rate": 1.3938196690254698e-05, "loss": 0.0932, "step": 5763 }, { "epoch": 5.85, "learning_rate": 1.3936181583855094e-05, "loss": 0.1078, "step": 5764 }, { "epoch": 5.85, "learning_rate": 1.3934166288298729e-05, "loss": 0.116, "step": 5765 }, { "epoch": 5.85, "learning_rate": 1.3932150803682447e-05, "loss": 0.0992, "step": 5766 }, { "epoch": 5.85, "learning_rate": 1.3930135130103103e-05, "loss": 0.085, "step": 5767 }, { "epoch": 5.85, "learning_rate": 1.3928119267657563e-05, "loss": 0.0757, "step": 5768 }, { "epoch": 5.85, "learning_rate": 1.3926103216442699e-05, "loss": 0.0864, "step": 5769 }, { "epoch": 5.85, "learning_rate": 1.3924086976555396e-05, "loss": 0.0875, "step": 5770 }, { "epoch": 5.86, "learning_rate": 1.3922070548092546e-05, "loss": 0.0935, "step": 5771 }, { "epoch": 5.86, "learning_rate": 1.392005393115105e-05, "loss": 0.0917, "step": 5772 }, { "epoch": 5.86, "learning_rate": 1.391803712582782e-05, "loss": 0.0878, "step": 5773 }, { "epoch": 5.86, "learning_rate": 1.3916020132219771e-05, "loss": 0.0793, "step": 5774 }, { "epoch": 5.86, "learning_rate": 1.3914002950423835e-05, "loss": 0.0705, "step": 5775 }, { "epoch": 5.86, "learning_rate": 1.391198558053695e-05, "loss": 0.0883, "step": 5776 }, { "epoch": 5.86, "learning_rate": 1.3909968022656063e-05, "loss": 0.0831, "step": 5777 }, { "epoch": 5.86, "learning_rate": 1.3907950276878124e-05, "loss": 0.0949, "step": 5778 }, { "epoch": 5.86, "learning_rate": 1.3905932343300107e-05, "loss": 0.0868, "step": 5779 }, { "epoch": 5.87, "learning_rate": 1.3903914222018973e-05, "loss": 0.0936, "step": 5780 }, { "epoch": 5.87, "learning_rate": 1.3901895913131718e-05, "loss": 0.0872, "step": 5781 }, { "epoch": 5.87, "learning_rate": 1.3899877416735322e-05, "loss": 0.0844, "step": 5782 }, { "epoch": 5.87, "learning_rate": 1.3897858732926794e-05, "loss": 0.0886, "step": 5783 }, { "epoch": 5.87, "learning_rate": 1.3895839861803142e-05, "loss": 0.0803, "step": 5784 }, { "epoch": 5.87, "learning_rate": 1.3893820803461385e-05, "loss": 0.0897, "step": 5785 }, { "epoch": 5.87, "learning_rate": 1.389180155799854e-05, "loss": 0.0823, "step": 5786 }, { "epoch": 5.87, "learning_rate": 1.3889782125511664e-05, "loss": 0.0802, "step": 5787 }, { "epoch": 5.87, "learning_rate": 1.3887762506097785e-05, "loss": 0.0959, "step": 5788 }, { "epoch": 5.87, "learning_rate": 1.3885742699853967e-05, "loss": 0.0962, "step": 5789 }, { "epoch": 5.88, "learning_rate": 1.388372270687727e-05, "loss": 0.0826, "step": 5790 }, { "epoch": 5.88, "learning_rate": 1.3881702527264767e-05, "loss": 0.0967, "step": 5791 }, { "epoch": 5.88, "learning_rate": 1.3879682161113538e-05, "loss": 0.0667, "step": 5792 }, { "epoch": 5.88, "learning_rate": 1.3877661608520678e-05, "loss": 0.097, "step": 5793 }, { "epoch": 5.88, "learning_rate": 1.3875640869583284e-05, "loss": 0.1036, "step": 5794 }, { "epoch": 5.88, "learning_rate": 1.3873619944398464e-05, "loss": 0.1064, "step": 5795 }, { "epoch": 5.88, "learning_rate": 1.3871598833063335e-05, "loss": 0.1096, "step": 5796 }, { "epoch": 5.88, "learning_rate": 1.3869577535675022e-05, "loss": 0.0986, "step": 5797 }, { "epoch": 5.88, "learning_rate": 1.3867556052330667e-05, "loss": 0.102, "step": 5798 }, { "epoch": 5.88, "learning_rate": 1.3865534383127406e-05, "loss": 0.1036, "step": 5799 }, { "epoch": 5.89, "learning_rate": 1.3863512528162397e-05, "loss": 0.0749, "step": 5800 }, { "epoch": 5.89, "learning_rate": 1.38614904875328e-05, "loss": 0.0896, "step": 5801 }, { "epoch": 5.89, "learning_rate": 1.3859468261335786e-05, "loss": 0.0836, "step": 5802 }, { "epoch": 5.89, "learning_rate": 1.3857445849668536e-05, "loss": 0.0866, "step": 5803 }, { "epoch": 5.89, "learning_rate": 1.385542325262824e-05, "loss": 0.1017, "step": 5804 }, { "epoch": 5.89, "learning_rate": 1.3853400470312095e-05, "loss": 0.0861, "step": 5805 }, { "epoch": 5.89, "learning_rate": 1.3851377502817305e-05, "loss": 0.1177, "step": 5806 }, { "epoch": 5.89, "learning_rate": 1.3849354350241087e-05, "loss": 0.0842, "step": 5807 }, { "epoch": 5.89, "learning_rate": 1.3847331012680664e-05, "loss": 0.0945, "step": 5808 }, { "epoch": 5.89, "learning_rate": 1.3845307490233275e-05, "loss": 0.0936, "step": 5809 }, { "epoch": 5.9, "learning_rate": 1.3843283782996157e-05, "loss": 0.0986, "step": 5810 }, { "epoch": 5.9, "learning_rate": 1.384125989106656e-05, "loss": 0.0835, "step": 5811 }, { "epoch": 5.9, "learning_rate": 1.383923581454175e-05, "loss": 0.0855, "step": 5812 }, { "epoch": 5.9, "learning_rate": 1.3837211553518989e-05, "loss": 0.1012, "step": 5813 }, { "epoch": 5.9, "learning_rate": 1.383518710809556e-05, "loss": 0.0945, "step": 5814 }, { "epoch": 5.9, "learning_rate": 1.3833162478368744e-05, "loss": 0.0899, "step": 5815 }, { "epoch": 5.9, "learning_rate": 1.3831137664435846e-05, "loss": 0.0919, "step": 5816 }, { "epoch": 5.9, "learning_rate": 1.382911266639416e-05, "loss": 0.1027, "step": 5817 }, { "epoch": 5.9, "learning_rate": 1.3827087484341005e-05, "loss": 0.0934, "step": 5818 }, { "epoch": 5.9, "learning_rate": 1.3825062118373698e-05, "loss": 0.0899, "step": 5819 }, { "epoch": 5.91, "learning_rate": 1.3823036568589573e-05, "loss": 0.0765, "step": 5820 }, { "epoch": 5.91, "learning_rate": 1.3821010835085972e-05, "loss": 0.0783, "step": 5821 }, { "epoch": 5.91, "learning_rate": 1.3818984917960241e-05, "loss": 0.0895, "step": 5822 }, { "epoch": 5.91, "learning_rate": 1.3816958817309738e-05, "loss": 0.1007, "step": 5823 }, { "epoch": 5.91, "learning_rate": 1.3814932533231826e-05, "loss": 0.1065, "step": 5824 }, { "epoch": 5.91, "learning_rate": 1.3812906065823883e-05, "loss": 0.1006, "step": 5825 }, { "epoch": 5.91, "learning_rate": 1.3810879415183292e-05, "loss": 0.0859, "step": 5826 }, { "epoch": 5.91, "learning_rate": 1.3808852581407446e-05, "loss": 0.0911, "step": 5827 }, { "epoch": 5.91, "learning_rate": 1.3806825564593744e-05, "loss": 0.0848, "step": 5828 }, { "epoch": 5.91, "learning_rate": 1.3804798364839603e-05, "loss": 0.081, "step": 5829 }, { "epoch": 5.92, "learning_rate": 1.380277098224243e-05, "loss": 0.1087, "step": 5830 }, { "epoch": 5.92, "learning_rate": 1.3800743416899667e-05, "loss": 0.1029, "step": 5831 }, { "epoch": 5.92, "learning_rate": 1.3798715668908735e-05, "loss": 0.0924, "step": 5832 }, { "epoch": 5.92, "learning_rate": 1.3796687738367095e-05, "loss": 0.1019, "step": 5833 }, { "epoch": 5.92, "learning_rate": 1.379465962537219e-05, "loss": 0.0842, "step": 5834 }, { "epoch": 5.92, "learning_rate": 1.3792631330021489e-05, "loss": 0.0882, "step": 5835 }, { "epoch": 5.92, "learning_rate": 1.3790602852412455e-05, "loss": 0.0817, "step": 5836 }, { "epoch": 5.92, "learning_rate": 1.3788574192642576e-05, "loss": 0.08, "step": 5837 }, { "epoch": 5.92, "learning_rate": 1.3786545350809341e-05, "loss": 0.0949, "step": 5838 }, { "epoch": 5.92, "learning_rate": 1.3784516327010246e-05, "loss": 0.0757, "step": 5839 }, { "epoch": 5.93, "learning_rate": 1.3782487121342798e-05, "loss": 0.0918, "step": 5840 }, { "epoch": 5.93, "learning_rate": 1.3780457733904511e-05, "loss": 0.0867, "step": 5841 }, { "epoch": 5.93, "learning_rate": 1.3778428164792908e-05, "loss": 0.078, "step": 5842 }, { "epoch": 5.93, "learning_rate": 1.3776398414105528e-05, "loss": 0.0789, "step": 5843 }, { "epoch": 5.93, "learning_rate": 1.3774368481939906e-05, "loss": 0.0956, "step": 5844 }, { "epoch": 5.93, "learning_rate": 1.3772338368393591e-05, "loss": 0.0878, "step": 5845 }, { "epoch": 5.93, "learning_rate": 1.3770308073564152e-05, "loss": 0.086, "step": 5846 }, { "epoch": 5.93, "learning_rate": 1.3768277597549144e-05, "loss": 0.0854, "step": 5847 }, { "epoch": 5.93, "learning_rate": 1.3766246940446153e-05, "loss": 0.091, "step": 5848 }, { "epoch": 5.94, "learning_rate": 1.376421610235276e-05, "loss": 0.0863, "step": 5849 }, { "epoch": 5.94, "learning_rate": 1.3762185083366557e-05, "loss": 0.079, "step": 5850 }, { "epoch": 5.94, "learning_rate": 1.3760153883585152e-05, "loss": 0.0849, "step": 5851 }, { "epoch": 5.94, "learning_rate": 1.375812250310615e-05, "loss": 0.0926, "step": 5852 }, { "epoch": 5.94, "learning_rate": 1.3756090942027175e-05, "loss": 0.0938, "step": 5853 }, { "epoch": 5.94, "learning_rate": 1.3754059200445855e-05, "loss": 0.0964, "step": 5854 }, { "epoch": 5.94, "learning_rate": 1.3752027278459825e-05, "loss": 0.1062, "step": 5855 }, { "epoch": 5.94, "learning_rate": 1.3749995176166734e-05, "loss": 0.072, "step": 5856 }, { "epoch": 5.94, "learning_rate": 1.3747962893664233e-05, "loss": 0.082, "step": 5857 }, { "epoch": 5.94, "learning_rate": 1.3745930431049985e-05, "loss": 0.0916, "step": 5858 }, { "epoch": 5.95, "learning_rate": 1.3743897788421668e-05, "loss": 0.0989, "step": 5859 }, { "epoch": 5.95, "learning_rate": 1.3741864965876956e-05, "loss": 0.0907, "step": 5860 }, { "epoch": 5.95, "learning_rate": 1.3739831963513538e-05, "loss": 0.0809, "step": 5861 }, { "epoch": 5.95, "learning_rate": 1.373779878142912e-05, "loss": 0.078, "step": 5862 }, { "epoch": 5.95, "learning_rate": 1.37357654197214e-05, "loss": 0.0925, "step": 5863 }, { "epoch": 5.95, "learning_rate": 1.3733731878488092e-05, "loss": 0.1005, "step": 5864 }, { "epoch": 5.95, "learning_rate": 1.3731698157826926e-05, "loss": 0.0829, "step": 5865 }, { "epoch": 5.95, "learning_rate": 1.3729664257835631e-05, "loss": 0.0932, "step": 5866 }, { "epoch": 5.95, "learning_rate": 1.3727630178611948e-05, "loss": 0.0856, "step": 5867 }, { "epoch": 5.95, "learning_rate": 1.3725595920253629e-05, "loss": 0.0887, "step": 5868 }, { "epoch": 5.96, "learning_rate": 1.3723561482858428e-05, "loss": 0.0963, "step": 5869 }, { "epoch": 5.96, "learning_rate": 1.3721526866524112e-05, "loss": 0.0772, "step": 5870 }, { "epoch": 5.96, "learning_rate": 1.3719492071348461e-05, "loss": 0.0789, "step": 5871 }, { "epoch": 5.96, "learning_rate": 1.3717457097429256e-05, "loss": 0.0875, "step": 5872 }, { "epoch": 5.96, "learning_rate": 1.3715421944864289e-05, "loss": 0.0814, "step": 5873 }, { "epoch": 5.96, "learning_rate": 1.3713386613751361e-05, "loss": 0.0884, "step": 5874 }, { "epoch": 5.96, "learning_rate": 1.3711351104188281e-05, "loss": 0.0965, "step": 5875 }, { "epoch": 5.96, "learning_rate": 1.370931541627287e-05, "loss": 0.1099, "step": 5876 }, { "epoch": 5.96, "learning_rate": 1.3707279550102955e-05, "loss": 0.1019, "step": 5877 }, { "epoch": 5.96, "learning_rate": 1.3705243505776366e-05, "loss": 0.0887, "step": 5878 }, { "epoch": 5.97, "learning_rate": 1.3703207283390953e-05, "loss": 0.0936, "step": 5879 }, { "epoch": 5.97, "learning_rate": 1.3701170883044566e-05, "loss": 0.0877, "step": 5880 }, { "epoch": 5.97, "learning_rate": 1.3699134304835066e-05, "loss": 0.0799, "step": 5881 }, { "epoch": 5.97, "learning_rate": 1.3697097548860323e-05, "loss": 0.0928, "step": 5882 }, { "epoch": 5.97, "learning_rate": 1.3695060615218215e-05, "loss": 0.0906, "step": 5883 }, { "epoch": 5.97, "learning_rate": 1.3693023504006629e-05, "loss": 0.0871, "step": 5884 }, { "epoch": 5.97, "learning_rate": 1.3690986215323461e-05, "loss": 0.0831, "step": 5885 }, { "epoch": 5.97, "learning_rate": 1.368894874926661e-05, "loss": 0.0815, "step": 5886 }, { "epoch": 5.97, "learning_rate": 1.3686911105933994e-05, "loss": 0.096, "step": 5887 }, { "epoch": 5.97, "learning_rate": 1.3684873285423533e-05, "loss": 0.0929, "step": 5888 }, { "epoch": 5.98, "learning_rate": 1.3682835287833154e-05, "loss": 0.0861, "step": 5889 }, { "epoch": 5.98, "learning_rate": 1.3680797113260797e-05, "loss": 0.0885, "step": 5890 }, { "epoch": 5.98, "learning_rate": 1.3678758761804404e-05, "loss": 0.1075, "step": 5891 }, { "epoch": 5.98, "learning_rate": 1.3676720233561933e-05, "loss": 0.0793, "step": 5892 }, { "epoch": 5.98, "learning_rate": 1.3674681528631351e-05, "loss": 0.0934, "step": 5893 }, { "epoch": 5.98, "learning_rate": 1.3672642647110625e-05, "loss": 0.085, "step": 5894 }, { "epoch": 5.98, "learning_rate": 1.3670603589097734e-05, "loss": 0.077, "step": 5895 }, { "epoch": 5.98, "learning_rate": 1.3668564354690672e-05, "loss": 0.0807, "step": 5896 }, { "epoch": 5.98, "learning_rate": 1.3666524943987431e-05, "loss": 0.0886, "step": 5897 }, { "epoch": 5.98, "learning_rate": 1.366448535708602e-05, "loss": 0.0995, "step": 5898 }, { "epoch": 5.99, "learning_rate": 1.3662445594084453e-05, "loss": 0.0873, "step": 5899 }, { "epoch": 5.99, "learning_rate": 1.3660405655080752e-05, "loss": 0.0869, "step": 5900 }, { "epoch": 5.99, "learning_rate": 1.3658365540172948e-05, "loss": 0.09, "step": 5901 }, { "epoch": 5.99, "learning_rate": 1.3656325249459082e-05, "loss": 0.0837, "step": 5902 }, { "epoch": 5.99, "learning_rate": 1.3654284783037198e-05, "loss": 0.0853, "step": 5903 }, { "epoch": 5.99, "learning_rate": 1.3652244141005358e-05, "loss": 0.0984, "step": 5904 }, { "epoch": 5.99, "learning_rate": 1.3650203323461622e-05, "loss": 0.0894, "step": 5905 }, { "epoch": 5.99, "learning_rate": 1.3648162330504067e-05, "loss": 0.0993, "step": 5906 }, { "epoch": 5.99, "learning_rate": 1.3646121162230769e-05, "loss": 0.0999, "step": 5907 }, { "epoch": 5.99, "learning_rate": 1.3644079818739829e-05, "loss": 0.0942, "step": 5908 }, { "epoch": 6.0, "learning_rate": 1.3642038300129331e-05, "loss": 0.1026, "step": 5909 }, { "epoch": 6.0, "learning_rate": 1.3639996606497397e-05, "loss": 0.0891, "step": 5910 }, { "epoch": 6.0, "learning_rate": 1.3637954737942134e-05, "loss": 0.0991, "step": 5911 }, { "epoch": 6.0, "learning_rate": 1.3635912694561664e-05, "loss": 0.1004, "step": 5912 }, { "epoch": 6.0, "learning_rate": 1.3633870476454125e-05, "loss": 0.1031, "step": 5913 }, { "epoch": 6.0, "learning_rate": 1.3631828083717655e-05, "loss": 0.045, "step": 5914 }, { "epoch": 6.0, "learning_rate": 1.3629785516450405e-05, "loss": 0.0529, "step": 5915 }, { "epoch": 6.0, "learning_rate": 1.3627742774750525e-05, "loss": 0.0526, "step": 5916 }, { "epoch": 6.0, "learning_rate": 1.3625699858716191e-05, "loss": 0.0509, "step": 5917 }, { "epoch": 6.01, "learning_rate": 1.3623656768445572e-05, "loss": 0.0547, "step": 5918 }, { "epoch": 6.01, "learning_rate": 1.3621613504036848e-05, "loss": 0.0484, "step": 5919 }, { "epoch": 6.01, "learning_rate": 1.3619570065588215e-05, "loss": 0.0583, "step": 5920 }, { "epoch": 6.01, "learning_rate": 1.3617526453197871e-05, "loss": 0.0534, "step": 5921 }, { "epoch": 6.01, "learning_rate": 1.3615482666964023e-05, "loss": 0.0509, "step": 5922 }, { "epoch": 6.01, "learning_rate": 1.3613438706984889e-05, "loss": 0.048, "step": 5923 }, { "epoch": 6.01, "learning_rate": 1.3611394573358687e-05, "loss": 0.0467, "step": 5924 }, { "epoch": 6.01, "learning_rate": 1.3609350266183655e-05, "loss": 0.0471, "step": 5925 }, { "epoch": 6.01, "learning_rate": 1.3607305785558033e-05, "loss": 0.0529, "step": 5926 }, { "epoch": 6.01, "learning_rate": 1.3605261131580075e-05, "loss": 0.0529, "step": 5927 }, { "epoch": 6.02, "learning_rate": 1.3603216304348028e-05, "loss": 0.048, "step": 5928 }, { "epoch": 6.02, "learning_rate": 1.3601171303960165e-05, "loss": 0.0716, "step": 5929 }, { "epoch": 6.02, "learning_rate": 1.3599126130514764e-05, "loss": 0.0671, "step": 5930 }, { "epoch": 6.02, "learning_rate": 1.3597080784110098e-05, "loss": 0.06, "step": 5931 }, { "epoch": 6.02, "learning_rate": 1.3595035264844463e-05, "loss": 0.0597, "step": 5932 }, { "epoch": 6.02, "learning_rate": 1.3592989572816163e-05, "loss": 0.074, "step": 5933 }, { "epoch": 6.02, "learning_rate": 1.3590943708123496e-05, "loss": 0.0425, "step": 5934 }, { "epoch": 6.02, "learning_rate": 1.3588897670864787e-05, "loss": 0.0548, "step": 5935 }, { "epoch": 6.02, "learning_rate": 1.3586851461138353e-05, "loss": 0.0537, "step": 5936 }, { "epoch": 6.02, "learning_rate": 1.3584805079042529e-05, "loss": 0.0719, "step": 5937 }, { "epoch": 6.03, "learning_rate": 1.3582758524675658e-05, "loss": 0.0653, "step": 5938 }, { "epoch": 6.03, "learning_rate": 1.3580711798136089e-05, "loss": 0.0556, "step": 5939 }, { "epoch": 6.03, "learning_rate": 1.3578664899522174e-05, "loss": 0.0417, "step": 5940 }, { "epoch": 6.03, "learning_rate": 1.3576617828932281e-05, "loss": 0.065, "step": 5941 }, { "epoch": 6.03, "learning_rate": 1.3574570586464786e-05, "loss": 0.0564, "step": 5942 }, { "epoch": 6.03, "learning_rate": 1.357252317221807e-05, "loss": 0.0613, "step": 5943 }, { "epoch": 6.03, "learning_rate": 1.3570475586290526e-05, "loss": 0.0659, "step": 5944 }, { "epoch": 6.03, "learning_rate": 1.3568427828780549e-05, "loss": 0.0492, "step": 5945 }, { "epoch": 6.03, "learning_rate": 1.3566379899786544e-05, "loss": 0.0409, "step": 5946 }, { "epoch": 6.03, "learning_rate": 1.356433179940693e-05, "loss": 0.0447, "step": 5947 }, { "epoch": 6.04, "learning_rate": 1.356228352774013e-05, "loss": 0.0536, "step": 5948 }, { "epoch": 6.04, "learning_rate": 1.3560235084884575e-05, "loss": 0.0651, "step": 5949 }, { "epoch": 6.04, "learning_rate": 1.3558186470938704e-05, "loss": 0.0545, "step": 5950 }, { "epoch": 6.04, "learning_rate": 1.3556137686000966e-05, "loss": 0.05, "step": 5951 }, { "epoch": 6.04, "learning_rate": 1.3554088730169814e-05, "loss": 0.0667, "step": 5952 }, { "epoch": 6.04, "learning_rate": 1.3552039603543716e-05, "loss": 0.0477, "step": 5953 }, { "epoch": 6.04, "learning_rate": 1.3549990306221145e-05, "loss": 0.0477, "step": 5954 }, { "epoch": 6.04, "learning_rate": 1.3547940838300582e-05, "loss": 0.0532, "step": 5955 }, { "epoch": 6.04, "learning_rate": 1.3545891199880511e-05, "loss": 0.0568, "step": 5956 }, { "epoch": 6.04, "learning_rate": 1.3543841391059436e-05, "loss": 0.0584, "step": 5957 }, { "epoch": 6.05, "learning_rate": 1.3541791411935856e-05, "loss": 0.0515, "step": 5958 }, { "epoch": 6.05, "learning_rate": 1.3539741262608288e-05, "loss": 0.0554, "step": 5959 }, { "epoch": 6.05, "learning_rate": 1.3537690943175256e-05, "loss": 0.0669, "step": 5960 }, { "epoch": 6.05, "learning_rate": 1.3535640453735288e-05, "loss": 0.055, "step": 5961 }, { "epoch": 6.05, "learning_rate": 1.3533589794386919e-05, "loss": 0.0546, "step": 5962 }, { "epoch": 6.05, "learning_rate": 1.35315389652287e-05, "loss": 0.0649, "step": 5963 }, { "epoch": 6.05, "learning_rate": 1.3529487966359182e-05, "loss": 0.0644, "step": 5964 }, { "epoch": 6.05, "learning_rate": 1.3527436797876929e-05, "loss": 0.0627, "step": 5965 }, { "epoch": 6.05, "learning_rate": 1.3525385459880511e-05, "loss": 0.0628, "step": 5966 }, { "epoch": 6.05, "learning_rate": 1.3523333952468512e-05, "loss": 0.0648, "step": 5967 }, { "epoch": 6.06, "learning_rate": 1.3521282275739512e-05, "loss": 0.0554, "step": 5968 }, { "epoch": 6.06, "learning_rate": 1.3519230429792107e-05, "loss": 0.0453, "step": 5969 }, { "epoch": 6.06, "learning_rate": 1.3517178414724903e-05, "loss": 0.0722, "step": 5970 }, { "epoch": 6.06, "learning_rate": 1.351512623063651e-05, "loss": 0.0496, "step": 5971 }, { "epoch": 6.06, "learning_rate": 1.3513073877625554e-05, "loss": 0.0521, "step": 5972 }, { "epoch": 6.06, "learning_rate": 1.3511021355790652e-05, "loss": 0.0712, "step": 5973 }, { "epoch": 6.06, "learning_rate": 1.3508968665230444e-05, "loss": 0.0614, "step": 5974 }, { "epoch": 6.06, "learning_rate": 1.3506915806043575e-05, "loss": 0.063, "step": 5975 }, { "epoch": 6.06, "learning_rate": 1.3504862778328698e-05, "loss": 0.0476, "step": 5976 }, { "epoch": 6.06, "learning_rate": 1.3502809582184471e-05, "loss": 0.0586, "step": 5977 }, { "epoch": 6.07, "learning_rate": 1.350075621770956e-05, "loss": 0.0603, "step": 5978 }, { "epoch": 6.07, "learning_rate": 1.349870268500265e-05, "loss": 0.0578, "step": 5979 }, { "epoch": 6.07, "learning_rate": 1.3496648984162416e-05, "loss": 0.052, "step": 5980 }, { "epoch": 6.07, "learning_rate": 1.3494595115287555e-05, "loss": 0.051, "step": 5981 }, { "epoch": 6.07, "learning_rate": 1.3492541078476766e-05, "loss": 0.0591, "step": 5982 }, { "epoch": 6.07, "learning_rate": 1.3490486873828759e-05, "loss": 0.0592, "step": 5983 }, { "epoch": 6.07, "learning_rate": 1.348843250144225e-05, "loss": 0.083, "step": 5984 }, { "epoch": 6.07, "learning_rate": 1.3486377961415962e-05, "loss": 0.0555, "step": 5985 }, { "epoch": 6.07, "learning_rate": 1.348432325384863e-05, "loss": 0.0661, "step": 5986 }, { "epoch": 6.08, "learning_rate": 1.3482268378838991e-05, "loss": 0.0601, "step": 5987 }, { "epoch": 6.08, "learning_rate": 1.34802133364858e-05, "loss": 0.0712, "step": 5988 }, { "epoch": 6.08, "learning_rate": 1.3478158126887813e-05, "loss": 0.0489, "step": 5989 }, { "epoch": 6.08, "learning_rate": 1.3476102750143791e-05, "loss": 0.0513, "step": 5990 }, { "epoch": 6.08, "learning_rate": 1.3474047206352507e-05, "loss": 0.0606, "step": 5991 }, { "epoch": 6.08, "learning_rate": 1.3471991495612745e-05, "loss": 0.0551, "step": 5992 }, { "epoch": 6.08, "learning_rate": 1.3469935618023292e-05, "loss": 0.0726, "step": 5993 }, { "epoch": 6.08, "learning_rate": 1.346787957368295e-05, "loss": 0.0682, "step": 5994 }, { "epoch": 6.08, "learning_rate": 1.3465823362690516e-05, "loss": 0.0501, "step": 5995 }, { "epoch": 6.08, "learning_rate": 1.3463766985144808e-05, "loss": 0.0514, "step": 5996 }, { "epoch": 6.09, "learning_rate": 1.3461710441144646e-05, "loss": 0.0556, "step": 5997 }, { "epoch": 6.09, "learning_rate": 1.345965373078886e-05, "loss": 0.0598, "step": 5998 }, { "epoch": 6.09, "learning_rate": 1.3457596854176285e-05, "loss": 0.0589, "step": 5999 }, { "epoch": 6.09, "learning_rate": 1.345553981140577e-05, "loss": 0.041, "step": 6000 }, { "epoch": 6.09, "learning_rate": 1.3453482602576164e-05, "loss": 0.0505, "step": 6001 }, { "epoch": 6.09, "learning_rate": 1.345142522778633e-05, "loss": 0.0538, "step": 6002 }, { "epoch": 6.09, "learning_rate": 1.3449367687135134e-05, "loss": 0.0524, "step": 6003 }, { "epoch": 6.09, "learning_rate": 1.3447309980721457e-05, "loss": 0.0494, "step": 6004 }, { "epoch": 6.09, "learning_rate": 1.3445252108644185e-05, "loss": 0.0579, "step": 6005 }, { "epoch": 6.09, "learning_rate": 1.3443194071002205e-05, "loss": 0.0574, "step": 6006 }, { "epoch": 6.1, "learning_rate": 1.3441135867894423e-05, "loss": 0.0539, "step": 6007 }, { "epoch": 6.1, "learning_rate": 1.3439077499419743e-05, "loss": 0.0633, "step": 6008 }, { "epoch": 6.1, "learning_rate": 1.3437018965677087e-05, "loss": 0.0509, "step": 6009 }, { "epoch": 6.1, "learning_rate": 1.3434960266765378e-05, "loss": 0.056, "step": 6010 }, { "epoch": 6.1, "learning_rate": 1.3432901402783549e-05, "loss": 0.0584, "step": 6011 }, { "epoch": 6.1, "learning_rate": 1.3430842373830534e-05, "loss": 0.0494, "step": 6012 }, { "epoch": 6.1, "learning_rate": 1.3428783180005294e-05, "loss": 0.0481, "step": 6013 }, { "epoch": 6.1, "learning_rate": 1.3426723821406773e-05, "loss": 0.0564, "step": 6014 }, { "epoch": 6.1, "learning_rate": 1.3424664298133948e-05, "loss": 0.06, "step": 6015 }, { "epoch": 6.1, "learning_rate": 1.3422604610285777e-05, "loss": 0.0666, "step": 6016 }, { "epoch": 6.11, "learning_rate": 1.342054475796125e-05, "loss": 0.0467, "step": 6017 }, { "epoch": 6.11, "learning_rate": 1.3418484741259352e-05, "loss": 0.071, "step": 6018 }, { "epoch": 6.11, "learning_rate": 1.341642456027908e-05, "loss": 0.0429, "step": 6019 }, { "epoch": 6.11, "learning_rate": 1.3414364215119436e-05, "loss": 0.0568, "step": 6020 }, { "epoch": 6.11, "learning_rate": 1.3412303705879435e-05, "loss": 0.0521, "step": 6021 }, { "epoch": 6.11, "learning_rate": 1.3410243032658094e-05, "loss": 0.0566, "step": 6022 }, { "epoch": 6.11, "learning_rate": 1.3408182195554443e-05, "loss": 0.0514, "step": 6023 }, { "epoch": 6.11, "learning_rate": 1.340612119466751e-05, "loss": 0.0494, "step": 6024 }, { "epoch": 6.11, "learning_rate": 1.3404060030096346e-05, "loss": 0.0483, "step": 6025 }, { "epoch": 6.11, "learning_rate": 1.3401998701940002e-05, "loss": 0.0503, "step": 6026 }, { "epoch": 6.12, "learning_rate": 1.3399937210297533e-05, "loss": 0.056, "step": 6027 }, { "epoch": 6.12, "learning_rate": 1.339787555526801e-05, "loss": 0.0589, "step": 6028 }, { "epoch": 6.12, "learning_rate": 1.3395813736950503e-05, "loss": 0.0484, "step": 6029 }, { "epoch": 6.12, "learning_rate": 1.3393751755444098e-05, "loss": 0.0666, "step": 6030 }, { "epoch": 6.12, "learning_rate": 1.3391689610847882e-05, "loss": 0.0505, "step": 6031 }, { "epoch": 6.12, "learning_rate": 1.338962730326096e-05, "loss": 0.0519, "step": 6032 }, { "epoch": 6.12, "learning_rate": 1.338756483278243e-05, "loss": 0.0505, "step": 6033 }, { "epoch": 6.12, "learning_rate": 1.338550219951141e-05, "loss": 0.057, "step": 6034 }, { "epoch": 6.12, "learning_rate": 1.3383439403547023e-05, "loss": 0.066, "step": 6035 }, { "epoch": 6.12, "learning_rate": 1.3381376444988394e-05, "loss": 0.0506, "step": 6036 }, { "epoch": 6.13, "learning_rate": 1.3379313323934664e-05, "loss": 0.0531, "step": 6037 }, { "epoch": 6.13, "learning_rate": 1.337725004048498e-05, "loss": 0.0461, "step": 6038 }, { "epoch": 6.13, "learning_rate": 1.3375186594738489e-05, "loss": 0.0607, "step": 6039 }, { "epoch": 6.13, "learning_rate": 1.3373122986794359e-05, "loss": 0.0489, "step": 6040 }, { "epoch": 6.13, "learning_rate": 1.3371059216751748e-05, "loss": 0.0627, "step": 6041 }, { "epoch": 6.13, "learning_rate": 1.3368995284709842e-05, "loss": 0.0527, "step": 6042 }, { "epoch": 6.13, "learning_rate": 1.3366931190767822e-05, "loss": 0.0529, "step": 6043 }, { "epoch": 6.13, "learning_rate": 1.3364866935024881e-05, "loss": 0.056, "step": 6044 }, { "epoch": 6.13, "learning_rate": 1.3362802517580215e-05, "loss": 0.0422, "step": 6045 }, { "epoch": 6.13, "learning_rate": 1.3360737938533038e-05, "loss": 0.0432, "step": 6046 }, { "epoch": 6.14, "learning_rate": 1.3358673197982555e-05, "loss": 0.0572, "step": 6047 }, { "epoch": 6.14, "learning_rate": 1.3356608296028e-05, "loss": 0.053, "step": 6048 }, { "epoch": 6.14, "learning_rate": 1.3354543232768596e-05, "loss": 0.0494, "step": 6049 }, { "epoch": 6.14, "learning_rate": 1.3352478008303587e-05, "loss": 0.069, "step": 6050 }, { "epoch": 6.14, "learning_rate": 1.3350412622732215e-05, "loss": 0.0567, "step": 6051 }, { "epoch": 6.14, "learning_rate": 1.3348347076153738e-05, "loss": 0.0525, "step": 6052 }, { "epoch": 6.14, "learning_rate": 1.334628136866741e-05, "loss": 0.0668, "step": 6053 }, { "epoch": 6.14, "learning_rate": 1.334421550037251e-05, "loss": 0.0701, "step": 6054 }, { "epoch": 6.14, "learning_rate": 1.334214947136831e-05, "loss": 0.0556, "step": 6055 }, { "epoch": 6.15, "learning_rate": 1.3340083281754097e-05, "loss": 0.0489, "step": 6056 }, { "epoch": 6.15, "learning_rate": 1.333801693162916e-05, "loss": 0.063, "step": 6057 }, { "epoch": 6.15, "learning_rate": 1.3335950421092803e-05, "loss": 0.0632, "step": 6058 }, { "epoch": 6.15, "learning_rate": 1.3333883750244332e-05, "loss": 0.0448, "step": 6059 }, { "epoch": 6.15, "learning_rate": 1.3331816919183063e-05, "loss": 0.0609, "step": 6060 }, { "epoch": 6.15, "learning_rate": 1.3329749928008322e-05, "loss": 0.0713, "step": 6061 }, { "epoch": 6.15, "learning_rate": 1.3327682776819435e-05, "loss": 0.0507, "step": 6062 }, { "epoch": 6.15, "learning_rate": 1.3325615465715746e-05, "loss": 0.0561, "step": 6063 }, { "epoch": 6.15, "learning_rate": 1.3323547994796597e-05, "loss": 0.053, "step": 6064 }, { "epoch": 6.15, "learning_rate": 1.3321480364161348e-05, "loss": 0.0444, "step": 6065 }, { "epoch": 6.16, "learning_rate": 1.3319412573909353e-05, "loss": 0.0567, "step": 6066 }, { "epoch": 6.16, "learning_rate": 1.3317344624139992e-05, "loss": 0.0534, "step": 6067 }, { "epoch": 6.16, "learning_rate": 1.331527651495263e-05, "loss": 0.0797, "step": 6068 }, { "epoch": 6.16, "learning_rate": 1.331320824644666e-05, "loss": 0.0522, "step": 6069 }, { "epoch": 6.16, "learning_rate": 1.3311139818721473e-05, "loss": 0.043, "step": 6070 }, { "epoch": 6.16, "learning_rate": 1.3309071231876469e-05, "loss": 0.0604, "step": 6071 }, { "epoch": 6.16, "learning_rate": 1.3307002486011058e-05, "loss": 0.0491, "step": 6072 }, { "epoch": 6.16, "learning_rate": 1.3304933581224648e-05, "loss": 0.0606, "step": 6073 }, { "epoch": 6.16, "learning_rate": 1.3302864517616671e-05, "loss": 0.0618, "step": 6074 }, { "epoch": 6.16, "learning_rate": 1.3300795295286549e-05, "loss": 0.0722, "step": 6075 }, { "epoch": 6.17, "learning_rate": 1.329872591433373e-05, "loss": 0.0634, "step": 6076 }, { "epoch": 6.17, "learning_rate": 1.3296656374857654e-05, "loss": 0.0598, "step": 6077 }, { "epoch": 6.17, "learning_rate": 1.3294586676957772e-05, "loss": 0.0602, "step": 6078 }, { "epoch": 6.17, "learning_rate": 1.3292516820733552e-05, "loss": 0.0506, "step": 6079 }, { "epoch": 6.17, "learning_rate": 1.3290446806284459e-05, "loss": 0.0379, "step": 6080 }, { "epoch": 6.17, "learning_rate": 1.328837663370997e-05, "loss": 0.0474, "step": 6081 }, { "epoch": 6.17, "learning_rate": 1.3286306303109568e-05, "loss": 0.0516, "step": 6082 }, { "epoch": 6.17, "learning_rate": 1.3284235814582747e-05, "loss": 0.0629, "step": 6083 }, { "epoch": 6.17, "learning_rate": 1.3282165168229004e-05, "loss": 0.0613, "step": 6084 }, { "epoch": 6.17, "learning_rate": 1.328009436414785e-05, "loss": 0.0652, "step": 6085 }, { "epoch": 6.18, "learning_rate": 1.327802340243879e-05, "loss": 0.0535, "step": 6086 }, { "epoch": 6.18, "learning_rate": 1.3275952283201354e-05, "loss": 0.0602, "step": 6087 }, { "epoch": 6.18, "learning_rate": 1.3273881006535074e-05, "loss": 0.0561, "step": 6088 }, { "epoch": 6.18, "learning_rate": 1.327180957253948e-05, "loss": 0.0525, "step": 6089 }, { "epoch": 6.18, "learning_rate": 1.326973798131412e-05, "loss": 0.059, "step": 6090 }, { "epoch": 6.18, "learning_rate": 1.3267666232958542e-05, "loss": 0.0582, "step": 6091 }, { "epoch": 6.18, "learning_rate": 1.3265594327572309e-05, "loss": 0.0589, "step": 6092 }, { "epoch": 6.18, "learning_rate": 1.326352226525499e-05, "loss": 0.0628, "step": 6093 }, { "epoch": 6.18, "learning_rate": 1.3261450046106161e-05, "loss": 0.0563, "step": 6094 }, { "epoch": 6.18, "learning_rate": 1.3259377670225398e-05, "loss": 0.0541, "step": 6095 }, { "epoch": 6.19, "learning_rate": 1.3257305137712297e-05, "loss": 0.0612, "step": 6096 }, { "epoch": 6.19, "learning_rate": 1.3255232448666448e-05, "loss": 0.0571, "step": 6097 }, { "epoch": 6.19, "learning_rate": 1.3253159603187466e-05, "loss": 0.06, "step": 6098 }, { "epoch": 6.19, "learning_rate": 1.3251086601374959e-05, "loss": 0.06, "step": 6099 }, { "epoch": 6.19, "learning_rate": 1.3249013443328542e-05, "loss": 0.057, "step": 6100 }, { "epoch": 6.19, "learning_rate": 1.324694012914785e-05, "loss": 0.0771, "step": 6101 }, { "epoch": 6.19, "learning_rate": 1.3244866658932513e-05, "loss": 0.0551, "step": 6102 }, { "epoch": 6.19, "learning_rate": 1.3242793032782175e-05, "loss": 0.0624, "step": 6103 }, { "epoch": 6.19, "learning_rate": 1.3240719250796485e-05, "loss": 0.0578, "step": 6104 }, { "epoch": 6.19, "learning_rate": 1.3238645313075104e-05, "loss": 0.0583, "step": 6105 }, { "epoch": 6.2, "learning_rate": 1.3236571219717696e-05, "loss": 0.0616, "step": 6106 }, { "epoch": 6.2, "learning_rate": 1.3234496970823932e-05, "loss": 0.0594, "step": 6107 }, { "epoch": 6.2, "learning_rate": 1.323242256649349e-05, "loss": 0.0587, "step": 6108 }, { "epoch": 6.2, "learning_rate": 1.3230348006826056e-05, "loss": 0.0603, "step": 6109 }, { "epoch": 6.2, "learning_rate": 1.3228273291921335e-05, "loss": 0.0577, "step": 6110 }, { "epoch": 6.2, "learning_rate": 1.322619842187902e-05, "loss": 0.0575, "step": 6111 }, { "epoch": 6.2, "learning_rate": 1.322412339679882e-05, "loss": 0.0699, "step": 6112 }, { "epoch": 6.2, "learning_rate": 1.3222048216780459e-05, "loss": 0.0527, "step": 6113 }, { "epoch": 6.2, "learning_rate": 1.3219972881923658e-05, "loss": 0.0621, "step": 6114 }, { "epoch": 6.2, "learning_rate": 1.3217897392328147e-05, "loss": 0.0796, "step": 6115 }, { "epoch": 6.21, "learning_rate": 1.3215821748093669e-05, "loss": 0.0681, "step": 6116 }, { "epoch": 6.21, "learning_rate": 1.3213745949319972e-05, "loss": 0.0703, "step": 6117 }, { "epoch": 6.21, "learning_rate": 1.3211669996106807e-05, "loss": 0.0731, "step": 6118 }, { "epoch": 6.21, "learning_rate": 1.3209593888553934e-05, "loss": 0.0591, "step": 6119 }, { "epoch": 6.21, "learning_rate": 1.3207517626761126e-05, "loss": 0.0539, "step": 6120 }, { "epoch": 6.21, "learning_rate": 1.3205441210828159e-05, "loss": 0.0723, "step": 6121 }, { "epoch": 6.21, "learning_rate": 1.3203364640854819e-05, "loss": 0.0524, "step": 6122 }, { "epoch": 6.21, "learning_rate": 1.3201287916940895e-05, "loss": 0.0566, "step": 6123 }, { "epoch": 6.21, "learning_rate": 1.3199211039186186e-05, "loss": 0.0509, "step": 6124 }, { "epoch": 6.22, "learning_rate": 1.3197134007690496e-05, "loss": 0.0562, "step": 6125 }, { "epoch": 6.22, "learning_rate": 1.319505682255364e-05, "loss": 0.0576, "step": 6126 }, { "epoch": 6.22, "learning_rate": 1.3192979483875443e-05, "loss": 0.0535, "step": 6127 }, { "epoch": 6.22, "learning_rate": 1.3190901991755728e-05, "loss": 0.0661, "step": 6128 }, { "epoch": 6.22, "learning_rate": 1.3188824346294335e-05, "loss": 0.0568, "step": 6129 }, { "epoch": 6.22, "learning_rate": 1.3186746547591104e-05, "loss": 0.0583, "step": 6130 }, { "epoch": 6.22, "learning_rate": 1.3184668595745889e-05, "loss": 0.0524, "step": 6131 }, { "epoch": 6.22, "learning_rate": 1.3182590490858545e-05, "loss": 0.0571, "step": 6132 }, { "epoch": 6.22, "learning_rate": 1.3180512233028935e-05, "loss": 0.051, "step": 6133 }, { "epoch": 6.22, "learning_rate": 1.3178433822356937e-05, "loss": 0.0512, "step": 6134 }, { "epoch": 6.23, "learning_rate": 1.3176355258942427e-05, "loss": 0.043, "step": 6135 }, { "epoch": 6.23, "learning_rate": 1.3174276542885294e-05, "loss": 0.0457, "step": 6136 }, { "epoch": 6.23, "learning_rate": 1.3172197674285432e-05, "loss": 0.0653, "step": 6137 }, { "epoch": 6.23, "learning_rate": 1.3170118653242743e-05, "loss": 0.0453, "step": 6138 }, { "epoch": 6.23, "learning_rate": 1.3168039479857138e-05, "loss": 0.0629, "step": 6139 }, { "epoch": 6.23, "learning_rate": 1.316596015422853e-05, "loss": 0.0534, "step": 6140 }, { "epoch": 6.23, "learning_rate": 1.3163880676456844e-05, "loss": 0.0704, "step": 6141 }, { "epoch": 6.23, "learning_rate": 1.3161801046642011e-05, "loss": 0.0728, "step": 6142 }, { "epoch": 6.23, "learning_rate": 1.3159721264883972e-05, "loss": 0.0516, "step": 6143 }, { "epoch": 6.23, "learning_rate": 1.3157641331282673e-05, "loss": 0.0578, "step": 6144 }, { "epoch": 6.24, "learning_rate": 1.3155561245938064e-05, "loss": 0.0747, "step": 6145 }, { "epoch": 6.24, "learning_rate": 1.3153481008950104e-05, "loss": 0.0627, "step": 6146 }, { "epoch": 6.24, "learning_rate": 1.3151400620418763e-05, "loss": 0.0564, "step": 6147 }, { "epoch": 6.24, "learning_rate": 1.314932008044402e-05, "loss": 0.0495, "step": 6148 }, { "epoch": 6.24, "learning_rate": 1.3147239389125854e-05, "loss": 0.0703, "step": 6149 }, { "epoch": 6.24, "learning_rate": 1.314515854656425e-05, "loss": 0.0674, "step": 6150 }, { "epoch": 6.24, "learning_rate": 1.3143077552859211e-05, "loss": 0.0721, "step": 6151 }, { "epoch": 6.24, "learning_rate": 1.3140996408110739e-05, "loss": 0.0559, "step": 6152 }, { "epoch": 6.24, "learning_rate": 1.3138915112418845e-05, "loss": 0.0397, "step": 6153 }, { "epoch": 6.24, "learning_rate": 1.3136833665883543e-05, "loss": 0.0446, "step": 6154 }, { "epoch": 6.25, "learning_rate": 1.313475206860487e-05, "loss": 0.0601, "step": 6155 }, { "epoch": 6.25, "learning_rate": 1.313267032068285e-05, "loss": 0.0626, "step": 6156 }, { "epoch": 6.25, "learning_rate": 1.3130588422217528e-05, "loss": 0.0521, "step": 6157 }, { "epoch": 6.25, "learning_rate": 1.3128506373308944e-05, "loss": 0.0516, "step": 6158 }, { "epoch": 6.25, "learning_rate": 1.312642417405716e-05, "loss": 0.0548, "step": 6159 }, { "epoch": 6.25, "learning_rate": 1.3124341824562237e-05, "loss": 0.0586, "step": 6160 }, { "epoch": 6.25, "learning_rate": 1.3122259324924242e-05, "loss": 0.0522, "step": 6161 }, { "epoch": 6.25, "learning_rate": 1.3120176675243252e-05, "loss": 0.062, "step": 6162 }, { "epoch": 6.25, "learning_rate": 1.3118093875619354e-05, "loss": 0.0578, "step": 6163 }, { "epoch": 6.25, "learning_rate": 1.3116010926152633e-05, "loss": 0.0553, "step": 6164 }, { "epoch": 6.26, "learning_rate": 1.3113927826943193e-05, "loss": 0.0605, "step": 6165 }, { "epoch": 6.26, "learning_rate": 1.3111844578091132e-05, "loss": 0.0499, "step": 6166 }, { "epoch": 6.26, "learning_rate": 1.310976117969657e-05, "loss": 0.0537, "step": 6167 }, { "epoch": 6.26, "learning_rate": 1.3107677631859623e-05, "loss": 0.0774, "step": 6168 }, { "epoch": 6.26, "learning_rate": 1.3105593934680416e-05, "loss": 0.0551, "step": 6169 }, { "epoch": 6.26, "learning_rate": 1.3103510088259088e-05, "loss": 0.0527, "step": 6170 }, { "epoch": 6.26, "learning_rate": 1.3101426092695773e-05, "loss": 0.0535, "step": 6171 }, { "epoch": 6.26, "learning_rate": 1.3099341948090628e-05, "loss": 0.0604, "step": 6172 }, { "epoch": 6.26, "learning_rate": 1.30972576545438e-05, "loss": 0.0504, "step": 6173 }, { "epoch": 6.26, "learning_rate": 1.3095173212155459e-05, "loss": 0.0557, "step": 6174 }, { "epoch": 6.27, "learning_rate": 1.3093088621025767e-05, "loss": 0.0548, "step": 6175 }, { "epoch": 6.27, "learning_rate": 1.3091003881254905e-05, "loss": 0.0623, "step": 6176 }, { "epoch": 6.27, "learning_rate": 1.308891899294306e-05, "loss": 0.0512, "step": 6177 }, { "epoch": 6.27, "learning_rate": 1.308683395619042e-05, "loss": 0.0689, "step": 6178 }, { "epoch": 6.27, "learning_rate": 1.3084748771097182e-05, "loss": 0.0737, "step": 6179 }, { "epoch": 6.27, "learning_rate": 1.3082663437763556e-05, "loss": 0.0603, "step": 6180 }, { "epoch": 6.27, "learning_rate": 1.3080577956289749e-05, "loss": 0.0473, "step": 6181 }, { "epoch": 6.27, "learning_rate": 1.3078492326775985e-05, "loss": 0.0517, "step": 6182 }, { "epoch": 6.27, "learning_rate": 1.3076406549322486e-05, "loss": 0.0748, "step": 6183 }, { "epoch": 6.27, "learning_rate": 1.3074320624029493e-05, "loss": 0.0695, "step": 6184 }, { "epoch": 6.28, "learning_rate": 1.3072234550997243e-05, "loss": 0.0503, "step": 6185 }, { "epoch": 6.28, "learning_rate": 1.307014833032598e-05, "loss": 0.0535, "step": 6186 }, { "epoch": 6.28, "learning_rate": 1.3068061962115966e-05, "loss": 0.0791, "step": 6187 }, { "epoch": 6.28, "learning_rate": 1.306597544646746e-05, "loss": 0.054, "step": 6188 }, { "epoch": 6.28, "learning_rate": 1.3063888783480735e-05, "loss": 0.0605, "step": 6189 }, { "epoch": 6.28, "learning_rate": 1.3061801973256066e-05, "loss": 0.061, "step": 6190 }, { "epoch": 6.28, "learning_rate": 1.3059715015893729e-05, "loss": 0.0614, "step": 6191 }, { "epoch": 6.28, "learning_rate": 1.3057627911494022e-05, "loss": 0.0642, "step": 6192 }, { "epoch": 6.28, "learning_rate": 1.3055540660157243e-05, "loss": 0.0585, "step": 6193 }, { "epoch": 6.29, "learning_rate": 1.3053453261983697e-05, "loss": 0.0662, "step": 6194 }, { "epoch": 6.29, "learning_rate": 1.3051365717073696e-05, "loss": 0.0555, "step": 6195 }, { "epoch": 6.29, "learning_rate": 1.3049278025527552e-05, "loss": 0.0765, "step": 6196 }, { "epoch": 6.29, "learning_rate": 1.3047190187445598e-05, "loss": 0.0699, "step": 6197 }, { "epoch": 6.29, "learning_rate": 1.3045102202928164e-05, "loss": 0.0553, "step": 6198 }, { "epoch": 6.29, "learning_rate": 1.304301407207559e-05, "loss": 0.0534, "step": 6199 }, { "epoch": 6.29, "learning_rate": 1.3040925794988227e-05, "loss": 0.0589, "step": 6200 }, { "epoch": 6.29, "learning_rate": 1.3038837371766422e-05, "loss": 0.059, "step": 6201 }, { "epoch": 6.29, "learning_rate": 1.3036748802510544e-05, "loss": 0.0577, "step": 6202 }, { "epoch": 6.29, "learning_rate": 1.3034660087320951e-05, "loss": 0.0624, "step": 6203 }, { "epoch": 6.3, "learning_rate": 1.3032571226298027e-05, "loss": 0.0729, "step": 6204 }, { "epoch": 6.3, "learning_rate": 1.3030482219542154e-05, "loss": 0.0525, "step": 6205 }, { "epoch": 6.3, "learning_rate": 1.3028393067153717e-05, "loss": 0.0508, "step": 6206 }, { "epoch": 6.3, "learning_rate": 1.3026303769233112e-05, "loss": 0.0534, "step": 6207 }, { "epoch": 6.3, "learning_rate": 1.3024214325880744e-05, "loss": 0.0624, "step": 6208 }, { "epoch": 6.3, "learning_rate": 1.3022124737197022e-05, "loss": 0.0644, "step": 6209 }, { "epoch": 6.3, "learning_rate": 1.3020035003282366e-05, "loss": 0.0578, "step": 6210 }, { "epoch": 6.3, "learning_rate": 1.3017945124237196e-05, "loss": 0.0575, "step": 6211 }, { "epoch": 6.3, "learning_rate": 1.3015855100161943e-05, "loss": 0.0583, "step": 6212 }, { "epoch": 6.3, "learning_rate": 1.301376493115705e-05, "loss": 0.0719, "step": 6213 }, { "epoch": 6.31, "learning_rate": 1.3011674617322955e-05, "loss": 0.0549, "step": 6214 }, { "epoch": 6.31, "learning_rate": 1.3009584158760118e-05, "loss": 0.0405, "step": 6215 }, { "epoch": 6.31, "learning_rate": 1.3007493555568991e-05, "loss": 0.0544, "step": 6216 }, { "epoch": 6.31, "learning_rate": 1.3005402807850041e-05, "loss": 0.0654, "step": 6217 }, { "epoch": 6.31, "learning_rate": 1.3003311915703744e-05, "loss": 0.0598, "step": 6218 }, { "epoch": 6.31, "learning_rate": 1.3001220879230579e-05, "loss": 0.054, "step": 6219 }, { "epoch": 6.31, "learning_rate": 1.299912969853103e-05, "loss": 0.0618, "step": 6220 }, { "epoch": 6.31, "learning_rate": 1.299703837370559e-05, "loss": 0.0541, "step": 6221 }, { "epoch": 6.31, "learning_rate": 1.2994946904854764e-05, "loss": 0.0598, "step": 6222 }, { "epoch": 6.31, "learning_rate": 1.2992855292079058e-05, "loss": 0.0616, "step": 6223 }, { "epoch": 6.32, "learning_rate": 1.2990763535478986e-05, "loss": 0.0611, "step": 6224 }, { "epoch": 6.32, "learning_rate": 1.2988671635155064e-05, "loss": 0.0654, "step": 6225 }, { "epoch": 6.32, "learning_rate": 1.2986579591207826e-05, "loss": 0.0651, "step": 6226 }, { "epoch": 6.32, "learning_rate": 1.2984487403737807e-05, "loss": 0.0654, "step": 6227 }, { "epoch": 6.32, "learning_rate": 1.2982395072845548e-05, "loss": 0.0587, "step": 6228 }, { "epoch": 6.32, "learning_rate": 1.2980302598631598e-05, "loss": 0.049, "step": 6229 }, { "epoch": 6.32, "learning_rate": 1.2978209981196512e-05, "loss": 0.0527, "step": 6230 }, { "epoch": 6.32, "learning_rate": 1.2976117220640849e-05, "loss": 0.0557, "step": 6231 }, { "epoch": 6.32, "learning_rate": 1.2974024317065188e-05, "loss": 0.0581, "step": 6232 }, { "epoch": 6.32, "learning_rate": 1.2971931270570098e-05, "loss": 0.0591, "step": 6233 }, { "epoch": 6.33, "learning_rate": 1.2969838081256163e-05, "loss": 0.066, "step": 6234 }, { "epoch": 6.33, "learning_rate": 1.2967744749223976e-05, "loss": 0.0675, "step": 6235 }, { "epoch": 6.33, "learning_rate": 1.2965651274574132e-05, "loss": 0.059, "step": 6236 }, { "epoch": 6.33, "learning_rate": 1.296355765740723e-05, "loss": 0.0555, "step": 6237 }, { "epoch": 6.33, "learning_rate": 1.2961463897823892e-05, "loss": 0.0601, "step": 6238 }, { "epoch": 6.33, "learning_rate": 1.295936999592473e-05, "loss": 0.0531, "step": 6239 }, { "epoch": 6.33, "learning_rate": 1.2957275951810364e-05, "loss": 0.0585, "step": 6240 }, { "epoch": 6.33, "learning_rate": 1.2955181765581428e-05, "loss": 0.056, "step": 6241 }, { "epoch": 6.33, "learning_rate": 1.2953087437338558e-05, "loss": 0.0608, "step": 6242 }, { "epoch": 6.33, "learning_rate": 1.2950992967182407e-05, "loss": 0.0541, "step": 6243 }, { "epoch": 6.34, "learning_rate": 1.2948898355213622e-05, "loss": 0.068, "step": 6244 }, { "epoch": 6.34, "learning_rate": 1.2946803601532858e-05, "loss": 0.0726, "step": 6245 }, { "epoch": 6.34, "learning_rate": 1.2944708706240783e-05, "loss": 0.0575, "step": 6246 }, { "epoch": 6.34, "learning_rate": 1.294261366943807e-05, "loss": 0.0651, "step": 6247 }, { "epoch": 6.34, "learning_rate": 1.2940518491225394e-05, "loss": 0.0569, "step": 6248 }, { "epoch": 6.34, "learning_rate": 1.2938423171703446e-05, "loss": 0.0582, "step": 6249 }, { "epoch": 6.34, "learning_rate": 1.2936327710972914e-05, "loss": 0.0768, "step": 6250 }, { "epoch": 6.34, "learning_rate": 1.2934232109134501e-05, "loss": 0.0566, "step": 6251 }, { "epoch": 6.34, "learning_rate": 1.2932136366288911e-05, "loss": 0.0613, "step": 6252 }, { "epoch": 6.35, "learning_rate": 1.2930040482536858e-05, "loss": 0.0625, "step": 6253 }, { "epoch": 6.35, "learning_rate": 1.292794445797906e-05, "loss": 0.0502, "step": 6254 }, { "epoch": 6.35, "learning_rate": 1.2925848292716245e-05, "loss": 0.0551, "step": 6255 }, { "epoch": 6.35, "learning_rate": 1.2923751986849145e-05, "loss": 0.0645, "step": 6256 }, { "epoch": 6.35, "learning_rate": 1.2921655540478499e-05, "loss": 0.057, "step": 6257 }, { "epoch": 6.35, "learning_rate": 1.2919558953705055e-05, "loss": 0.0689, "step": 6258 }, { "epoch": 6.35, "learning_rate": 1.2917462226629564e-05, "loss": 0.053, "step": 6259 }, { "epoch": 6.35, "learning_rate": 1.2915365359352792e-05, "loss": 0.0494, "step": 6260 }, { "epoch": 6.35, "learning_rate": 1.2913268351975503e-05, "loss": 0.061, "step": 6261 }, { "epoch": 6.35, "learning_rate": 1.2911171204598467e-05, "loss": 0.0573, "step": 6262 }, { "epoch": 6.36, "learning_rate": 1.2909073917322468e-05, "loss": 0.0607, "step": 6263 }, { "epoch": 6.36, "learning_rate": 1.2906976490248289e-05, "loss": 0.0773, "step": 6264 }, { "epoch": 6.36, "learning_rate": 1.290487892347673e-05, "loss": 0.0585, "step": 6265 }, { "epoch": 6.36, "learning_rate": 1.2902781217108589e-05, "loss": 0.0501, "step": 6266 }, { "epoch": 6.36, "learning_rate": 1.290068337124467e-05, "loss": 0.0602, "step": 6267 }, { "epoch": 6.36, "learning_rate": 1.2898585385985788e-05, "loss": 0.0661, "step": 6268 }, { "epoch": 6.36, "learning_rate": 1.289648726143277e-05, "loss": 0.0595, "step": 6269 }, { "epoch": 6.36, "learning_rate": 1.2894388997686434e-05, "loss": 0.0538, "step": 6270 }, { "epoch": 6.36, "learning_rate": 1.2892290594847618e-05, "loss": 0.0628, "step": 6271 }, { "epoch": 6.36, "learning_rate": 1.2890192053017165e-05, "loss": 0.0565, "step": 6272 }, { "epoch": 6.37, "learning_rate": 1.288809337229592e-05, "loss": 0.0599, "step": 6273 }, { "epoch": 6.37, "learning_rate": 1.2885994552784737e-05, "loss": 0.0548, "step": 6274 }, { "epoch": 6.37, "learning_rate": 1.2883895594584474e-05, "loss": 0.0659, "step": 6275 }, { "epoch": 6.37, "learning_rate": 1.2881796497796001e-05, "loss": 0.0692, "step": 6276 }, { "epoch": 6.37, "learning_rate": 1.2879697262520194e-05, "loss": 0.0628, "step": 6277 }, { "epoch": 6.37, "learning_rate": 1.2877597888857935e-05, "loss": 0.0548, "step": 6278 }, { "epoch": 6.37, "learning_rate": 1.2875498376910102e-05, "loss": 0.0526, "step": 6279 }, { "epoch": 6.37, "learning_rate": 1.28733987267776e-05, "loss": 0.0572, "step": 6280 }, { "epoch": 6.37, "learning_rate": 1.2871298938561321e-05, "loss": 0.0606, "step": 6281 }, { "epoch": 6.37, "learning_rate": 1.2869199012362178e-05, "loss": 0.0747, "step": 6282 }, { "epoch": 6.38, "learning_rate": 1.2867098948281078e-05, "loss": 0.0547, "step": 6283 }, { "epoch": 6.38, "learning_rate": 1.2864998746418951e-05, "loss": 0.0755, "step": 6284 }, { "epoch": 6.38, "learning_rate": 1.2862898406876721e-05, "loss": 0.0724, "step": 6285 }, { "epoch": 6.38, "learning_rate": 1.2860797929755316e-05, "loss": 0.0595, "step": 6286 }, { "epoch": 6.38, "learning_rate": 1.2858697315155679e-05, "loss": 0.0587, "step": 6287 }, { "epoch": 6.38, "learning_rate": 1.285659656317876e-05, "loss": 0.0564, "step": 6288 }, { "epoch": 6.38, "learning_rate": 1.2854495673925512e-05, "loss": 0.061, "step": 6289 }, { "epoch": 6.38, "learning_rate": 1.2852394647496894e-05, "loss": 0.0727, "step": 6290 }, { "epoch": 6.38, "learning_rate": 1.2850293483993873e-05, "loss": 0.0597, "step": 6291 }, { "epoch": 6.38, "learning_rate": 1.2848192183517415e-05, "loss": 0.0501, "step": 6292 }, { "epoch": 6.39, "learning_rate": 1.2846090746168513e-05, "loss": 0.0527, "step": 6293 }, { "epoch": 6.39, "learning_rate": 1.2843989172048147e-05, "loss": 0.0623, "step": 6294 }, { "epoch": 6.39, "learning_rate": 1.284188746125731e-05, "loss": 0.0691, "step": 6295 }, { "epoch": 6.39, "learning_rate": 1.2839785613897e-05, "loss": 0.0766, "step": 6296 }, { "epoch": 6.39, "learning_rate": 1.2837683630068227e-05, "loss": 0.0665, "step": 6297 }, { "epoch": 6.39, "learning_rate": 1.2835581509871999e-05, "loss": 0.0696, "step": 6298 }, { "epoch": 6.39, "learning_rate": 1.2833479253409341e-05, "loss": 0.056, "step": 6299 }, { "epoch": 6.39, "learning_rate": 1.2831376860781274e-05, "loss": 0.0501, "step": 6300 }, { "epoch": 6.39, "learning_rate": 1.2829274332088837e-05, "loss": 0.063, "step": 6301 }, { "epoch": 6.39, "learning_rate": 1.282717166743306e-05, "loss": 0.0545, "step": 6302 }, { "epoch": 6.4, "learning_rate": 1.2825068866914994e-05, "loss": 0.0453, "step": 6303 }, { "epoch": 6.4, "learning_rate": 1.2822965930635687e-05, "loss": 0.0733, "step": 6304 }, { "epoch": 6.4, "learning_rate": 1.2820862858696202e-05, "loss": 0.0534, "step": 6305 }, { "epoch": 6.4, "learning_rate": 1.2818759651197605e-05, "loss": 0.06, "step": 6306 }, { "epoch": 6.4, "learning_rate": 1.2816656308240963e-05, "loss": 0.0673, "step": 6307 }, { "epoch": 6.4, "learning_rate": 1.2814552829927354e-05, "loss": 0.0717, "step": 6308 }, { "epoch": 6.4, "learning_rate": 1.2812449216357863e-05, "loss": 0.0574, "step": 6309 }, { "epoch": 6.4, "learning_rate": 1.2810345467633585e-05, "loss": 0.0639, "step": 6310 }, { "epoch": 6.4, "learning_rate": 1.2808241583855617e-05, "loss": 0.0553, "step": 6311 }, { "epoch": 6.4, "learning_rate": 1.2806137565125058e-05, "loss": 0.0646, "step": 6312 }, { "epoch": 6.41, "learning_rate": 1.280403341154302e-05, "loss": 0.0519, "step": 6313 }, { "epoch": 6.41, "learning_rate": 1.2801929123210623e-05, "loss": 0.0574, "step": 6314 }, { "epoch": 6.41, "learning_rate": 1.279982470022899e-05, "loss": 0.0679, "step": 6315 }, { "epoch": 6.41, "learning_rate": 1.279772014269925e-05, "loss": 0.0669, "step": 6316 }, { "epoch": 6.41, "learning_rate": 1.2795615450722536e-05, "loss": 0.0569, "step": 6317 }, { "epoch": 6.41, "learning_rate": 1.2793510624399998e-05, "loss": 0.065, "step": 6318 }, { "epoch": 6.41, "learning_rate": 1.2791405663832782e-05, "loss": 0.0527, "step": 6319 }, { "epoch": 6.41, "learning_rate": 1.278930056912204e-05, "loss": 0.0568, "step": 6320 }, { "epoch": 6.41, "learning_rate": 1.2787195340368936e-05, "loss": 0.0689, "step": 6321 }, { "epoch": 6.42, "learning_rate": 1.2785089977674646e-05, "loss": 0.0618, "step": 6322 }, { "epoch": 6.42, "learning_rate": 1.2782984481140337e-05, "loss": 0.0592, "step": 6323 }, { "epoch": 6.42, "learning_rate": 1.2780878850867194e-05, "loss": 0.0643, "step": 6324 }, { "epoch": 6.42, "learning_rate": 1.27787730869564e-05, "loss": 0.0617, "step": 6325 }, { "epoch": 6.42, "learning_rate": 1.2776667189509155e-05, "loss": 0.0561, "step": 6326 }, { "epoch": 6.42, "learning_rate": 1.277456115862666e-05, "loss": 0.0526, "step": 6327 }, { "epoch": 6.42, "learning_rate": 1.277245499441012e-05, "loss": 0.0562, "step": 6328 }, { "epoch": 6.42, "learning_rate": 1.2770348696960745e-05, "loss": 0.0645, "step": 6329 }, { "epoch": 6.42, "learning_rate": 1.2768242266379762e-05, "loss": 0.0643, "step": 6330 }, { "epoch": 6.42, "learning_rate": 1.2766135702768393e-05, "loss": 0.066, "step": 6331 }, { "epoch": 6.43, "learning_rate": 1.2764029006227874e-05, "loss": 0.0402, "step": 6332 }, { "epoch": 6.43, "learning_rate": 1.2761922176859443e-05, "loss": 0.0636, "step": 6333 }, { "epoch": 6.43, "learning_rate": 1.2759815214764342e-05, "loss": 0.0637, "step": 6334 }, { "epoch": 6.43, "learning_rate": 1.2757708120043825e-05, "loss": 0.0514, "step": 6335 }, { "epoch": 6.43, "learning_rate": 1.2755600892799154e-05, "loss": 0.0481, "step": 6336 }, { "epoch": 6.43, "learning_rate": 1.2753493533131592e-05, "loss": 0.0573, "step": 6337 }, { "epoch": 6.43, "learning_rate": 1.2751386041142406e-05, "loss": 0.0462, "step": 6338 }, { "epoch": 6.43, "learning_rate": 1.2749278416932878e-05, "loss": 0.058, "step": 6339 }, { "epoch": 6.43, "learning_rate": 1.274717066060429e-05, "loss": 0.0538, "step": 6340 }, { "epoch": 6.43, "learning_rate": 1.2745062772257934e-05, "loss": 0.0689, "step": 6341 }, { "epoch": 6.44, "learning_rate": 1.2742954751995102e-05, "loss": 0.0571, "step": 6342 }, { "epoch": 6.44, "learning_rate": 1.27408465999171e-05, "loss": 0.0577, "step": 6343 }, { "epoch": 6.44, "learning_rate": 1.2738738316125238e-05, "loss": 0.0648, "step": 6344 }, { "epoch": 6.44, "learning_rate": 1.2736629900720832e-05, "loss": 0.0617, "step": 6345 }, { "epoch": 6.44, "learning_rate": 1.2734521353805198e-05, "loss": 0.0579, "step": 6346 }, { "epoch": 6.44, "learning_rate": 1.2732412675479671e-05, "loss": 0.0545, "step": 6347 }, { "epoch": 6.44, "learning_rate": 1.2730303865845582e-05, "loss": 0.0606, "step": 6348 }, { "epoch": 6.44, "learning_rate": 1.2728194925004273e-05, "loss": 0.0534, "step": 6349 }, { "epoch": 6.44, "learning_rate": 1.2726085853057089e-05, "loss": 0.0572, "step": 6350 }, { "epoch": 6.44, "learning_rate": 1.2723976650105385e-05, "loss": 0.0559, "step": 6351 }, { "epoch": 6.45, "learning_rate": 1.2721867316250522e-05, "loss": 0.0602, "step": 6352 }, { "epoch": 6.45, "learning_rate": 1.2719757851593864e-05, "loss": 0.0618, "step": 6353 }, { "epoch": 6.45, "learning_rate": 1.2717648256236784e-05, "loss": 0.0593, "step": 6354 }, { "epoch": 6.45, "learning_rate": 1.271553853028066e-05, "loss": 0.0572, "step": 6355 }, { "epoch": 6.45, "learning_rate": 1.2713428673826877e-05, "loss": 0.0548, "step": 6356 }, { "epoch": 6.45, "learning_rate": 1.2711318686976828e-05, "loss": 0.065, "step": 6357 }, { "epoch": 6.45, "learning_rate": 1.2709208569831904e-05, "loss": 0.0543, "step": 6358 }, { "epoch": 6.45, "learning_rate": 1.2707098322493514e-05, "loss": 0.0575, "step": 6359 }, { "epoch": 6.45, "learning_rate": 1.270498794506307e-05, "loss": 0.0486, "step": 6360 }, { "epoch": 6.45, "learning_rate": 1.2702877437641983e-05, "loss": 0.0556, "step": 6361 }, { "epoch": 6.46, "learning_rate": 1.2700766800331676e-05, "loss": 0.0523, "step": 6362 }, { "epoch": 6.46, "learning_rate": 1.269865603323358e-05, "loss": 0.0671, "step": 6363 }, { "epoch": 6.46, "learning_rate": 1.2696545136449128e-05, "loss": 0.0869, "step": 6364 }, { "epoch": 6.46, "learning_rate": 1.2694434110079763e-05, "loss": 0.0594, "step": 6365 }, { "epoch": 6.46, "learning_rate": 1.2692322954226932e-05, "loss": 0.052, "step": 6366 }, { "epoch": 6.46, "learning_rate": 1.2690211668992083e-05, "loss": 0.0776, "step": 6367 }, { "epoch": 6.46, "learning_rate": 1.2688100254476685e-05, "loss": 0.0575, "step": 6368 }, { "epoch": 6.46, "learning_rate": 1.2685988710782198e-05, "loss": 0.0457, "step": 6369 }, { "epoch": 6.46, "learning_rate": 1.2683877038010093e-05, "loss": 0.0483, "step": 6370 }, { "epoch": 6.46, "learning_rate": 1.2681765236261852e-05, "loss": 0.0608, "step": 6371 }, { "epoch": 6.47, "learning_rate": 1.2679653305638957e-05, "loss": 0.0581, "step": 6372 }, { "epoch": 6.47, "learning_rate": 1.2677541246242904e-05, "loss": 0.0741, "step": 6373 }, { "epoch": 6.47, "learning_rate": 1.2675429058175183e-05, "loss": 0.0664, "step": 6374 }, { "epoch": 6.47, "learning_rate": 1.2673316741537297e-05, "loss": 0.0509, "step": 6375 }, { "epoch": 6.47, "learning_rate": 1.2671204296430761e-05, "loss": 0.0566, "step": 6376 }, { "epoch": 6.47, "learning_rate": 1.266909172295709e-05, "loss": 0.0533, "step": 6377 }, { "epoch": 6.47, "learning_rate": 1.2666979021217803e-05, "loss": 0.0662, "step": 6378 }, { "epoch": 6.47, "learning_rate": 1.2664866191314427e-05, "loss": 0.0575, "step": 6379 }, { "epoch": 6.47, "learning_rate": 1.2662753233348495e-05, "loss": 0.0585, "step": 6380 }, { "epoch": 6.47, "learning_rate": 1.2660640147421548e-05, "loss": 0.0606, "step": 6381 }, { "epoch": 6.48, "learning_rate": 1.2658526933635136e-05, "loss": 0.0697, "step": 6382 }, { "epoch": 6.48, "learning_rate": 1.2656413592090811e-05, "loss": 0.0633, "step": 6383 }, { "epoch": 6.48, "learning_rate": 1.2654300122890124e-05, "loss": 0.0554, "step": 6384 }, { "epoch": 6.48, "learning_rate": 1.2652186526134651e-05, "loss": 0.0781, "step": 6385 }, { "epoch": 6.48, "learning_rate": 1.2650072801925952e-05, "loss": 0.0723, "step": 6386 }, { "epoch": 6.48, "learning_rate": 1.264795895036561e-05, "loss": 0.0542, "step": 6387 }, { "epoch": 6.48, "learning_rate": 1.2645844971555207e-05, "loss": 0.0595, "step": 6388 }, { "epoch": 6.48, "learning_rate": 1.2643730865596333e-05, "loss": 0.0605, "step": 6389 }, { "epoch": 6.48, "learning_rate": 1.2641616632590583e-05, "loss": 0.0688, "step": 6390 }, { "epoch": 6.49, "learning_rate": 1.2639502272639557e-05, "loss": 0.0598, "step": 6391 }, { "epoch": 6.49, "learning_rate": 1.263738778584486e-05, "loss": 0.0533, "step": 6392 }, { "epoch": 6.49, "learning_rate": 1.2635273172308111e-05, "loss": 0.0507, "step": 6393 }, { "epoch": 6.49, "learning_rate": 1.263315843213093e-05, "loss": 0.0547, "step": 6394 }, { "epoch": 6.49, "learning_rate": 1.2631043565414937e-05, "loss": 0.0672, "step": 6395 }, { "epoch": 6.49, "learning_rate": 1.2628928572261766e-05, "loss": 0.0535, "step": 6396 }, { "epoch": 6.49, "learning_rate": 1.2626813452773059e-05, "loss": 0.0773, "step": 6397 }, { "epoch": 6.49, "learning_rate": 1.2624698207050454e-05, "loss": 0.0848, "step": 6398 }, { "epoch": 6.49, "learning_rate": 1.2622582835195605e-05, "loss": 0.0494, "step": 6399 }, { "epoch": 6.49, "learning_rate": 1.2620467337310167e-05, "loss": 0.0763, "step": 6400 }, { "epoch": 6.5, "learning_rate": 1.2618351713495803e-05, "loss": 0.0587, "step": 6401 }, { "epoch": 6.5, "learning_rate": 1.261623596385418e-05, "loss": 0.0519, "step": 6402 }, { "epoch": 6.5, "learning_rate": 1.261412008848697e-05, "loss": 0.0535, "step": 6403 }, { "epoch": 6.5, "learning_rate": 1.2612004087495859e-05, "loss": 0.0743, "step": 6404 }, { "epoch": 6.5, "learning_rate": 1.2609887960982528e-05, "loss": 0.0565, "step": 6405 }, { "epoch": 6.5, "learning_rate": 1.2607771709048673e-05, "loss": 0.0585, "step": 6406 }, { "epoch": 6.5, "learning_rate": 1.2605655331795993e-05, "loss": 0.0599, "step": 6407 }, { "epoch": 6.5, "learning_rate": 1.2603538829326185e-05, "loss": 0.057, "step": 6408 }, { "epoch": 6.5, "learning_rate": 1.2601422201740967e-05, "loss": 0.0494, "step": 6409 }, { "epoch": 6.5, "learning_rate": 1.2599305449142056e-05, "loss": 0.069, "step": 6410 }, { "epoch": 6.51, "learning_rate": 1.259718857163117e-05, "loss": 0.0663, "step": 6411 }, { "epoch": 6.51, "learning_rate": 1.259507156931004e-05, "loss": 0.0631, "step": 6412 }, { "epoch": 6.51, "learning_rate": 1.2592954442280398e-05, "loss": 0.0511, "step": 6413 }, { "epoch": 6.51, "learning_rate": 1.2590837190643987e-05, "loss": 0.0578, "step": 6414 }, { "epoch": 6.51, "learning_rate": 1.2588719814502553e-05, "loss": 0.0561, "step": 6415 }, { "epoch": 6.51, "learning_rate": 1.258660231395785e-05, "loss": 0.0671, "step": 6416 }, { "epoch": 6.51, "learning_rate": 1.2584484689111629e-05, "loss": 0.0631, "step": 6417 }, { "epoch": 6.51, "learning_rate": 1.2582366940065664e-05, "loss": 0.0692, "step": 6418 }, { "epoch": 6.51, "learning_rate": 1.2580249066921721e-05, "loss": 0.0622, "step": 6419 }, { "epoch": 6.51, "learning_rate": 1.2578131069781577e-05, "loss": 0.0762, "step": 6420 }, { "epoch": 6.52, "learning_rate": 1.2576012948747009e-05, "loss": 0.072, "step": 6421 }, { "epoch": 6.52, "learning_rate": 1.2573894703919816e-05, "loss": 0.0673, "step": 6422 }, { "epoch": 6.52, "learning_rate": 1.2571776335401784e-05, "loss": 0.0659, "step": 6423 }, { "epoch": 6.52, "learning_rate": 1.2569657843294716e-05, "loss": 0.0658, "step": 6424 }, { "epoch": 6.52, "learning_rate": 1.2567539227700415e-05, "loss": 0.0585, "step": 6425 }, { "epoch": 6.52, "learning_rate": 1.2565420488720694e-05, "loss": 0.0618, "step": 6426 }, { "epoch": 6.52, "learning_rate": 1.2563301626457375e-05, "loss": 0.06, "step": 6427 }, { "epoch": 6.52, "learning_rate": 1.256118264101228e-05, "loss": 0.0768, "step": 6428 }, { "epoch": 6.52, "learning_rate": 1.2559063532487237e-05, "loss": 0.0609, "step": 6429 }, { "epoch": 6.52, "learning_rate": 1.2556944300984081e-05, "loss": 0.0595, "step": 6430 }, { "epoch": 6.53, "learning_rate": 1.2554824946604653e-05, "loss": 0.0529, "step": 6431 }, { "epoch": 6.53, "learning_rate": 1.2552705469450807e-05, "loss": 0.0655, "step": 6432 }, { "epoch": 6.53, "learning_rate": 1.2550585869624392e-05, "loss": 0.0551, "step": 6433 }, { "epoch": 6.53, "learning_rate": 1.2548466147227264e-05, "loss": 0.0673, "step": 6434 }, { "epoch": 6.53, "learning_rate": 1.2546346302361294e-05, "loss": 0.0565, "step": 6435 }, { "epoch": 6.53, "learning_rate": 1.2544226335128349e-05, "loss": 0.0576, "step": 6436 }, { "epoch": 6.53, "learning_rate": 1.2542106245630308e-05, "loss": 0.0632, "step": 6437 }, { "epoch": 6.53, "learning_rate": 1.2539986033969052e-05, "loss": 0.0688, "step": 6438 }, { "epoch": 6.53, "learning_rate": 1.2537865700246474e-05, "loss": 0.0638, "step": 6439 }, { "epoch": 6.53, "learning_rate": 1.2535745244564464e-05, "loss": 0.0684, "step": 6440 }, { "epoch": 6.54, "learning_rate": 1.2533624667024924e-05, "loss": 0.0435, "step": 6441 }, { "epoch": 6.54, "learning_rate": 1.2531503967729759e-05, "loss": 0.0579, "step": 6442 }, { "epoch": 6.54, "learning_rate": 1.252938314678088e-05, "loss": 0.0613, "step": 6443 }, { "epoch": 6.54, "learning_rate": 1.2527262204280212e-05, "loss": 0.0402, "step": 6444 }, { "epoch": 6.54, "learning_rate": 1.2525141140329673e-05, "loss": 0.0577, "step": 6445 }, { "epoch": 6.54, "learning_rate": 1.2523019955031192e-05, "loss": 0.0615, "step": 6446 }, { "epoch": 6.54, "learning_rate": 1.2520898648486709e-05, "loss": 0.0609, "step": 6447 }, { "epoch": 6.54, "learning_rate": 1.2518777220798157e-05, "loss": 0.0526, "step": 6448 }, { "epoch": 6.54, "learning_rate": 1.2516655672067493e-05, "loss": 0.0628, "step": 6449 }, { "epoch": 6.54, "learning_rate": 1.2514534002396666e-05, "loss": 0.0545, "step": 6450 }, { "epoch": 6.55, "learning_rate": 1.2512412211887633e-05, "loss": 0.0721, "step": 6451 }, { "epoch": 6.55, "learning_rate": 1.251029030064236e-05, "loss": 0.0628, "step": 6452 }, { "epoch": 6.55, "learning_rate": 1.2508168268762815e-05, "loss": 0.0574, "step": 6453 }, { "epoch": 6.55, "learning_rate": 1.2506046116350978e-05, "loss": 0.0742, "step": 6454 }, { "epoch": 6.55, "learning_rate": 1.250392384350883e-05, "loss": 0.0527, "step": 6455 }, { "epoch": 6.55, "learning_rate": 1.2501801450338359e-05, "loss": 0.0595, "step": 6456 }, { "epoch": 6.55, "learning_rate": 1.2499678936941557e-05, "loss": 0.064, "step": 6457 }, { "epoch": 6.55, "learning_rate": 1.2497556303420422e-05, "loss": 0.0509, "step": 6458 }, { "epoch": 6.55, "learning_rate": 1.2495433549876961e-05, "loss": 0.0779, "step": 6459 }, { "epoch": 6.56, "learning_rate": 1.2493310676413187e-05, "loss": 0.0563, "step": 6460 }, { "epoch": 6.56, "learning_rate": 1.2491187683131114e-05, "loss": 0.0612, "step": 6461 }, { "epoch": 6.56, "learning_rate": 1.2489064570132764e-05, "loss": 0.0749, "step": 6462 }, { "epoch": 6.56, "learning_rate": 1.2486941337520166e-05, "loss": 0.0431, "step": 6463 }, { "epoch": 6.56, "learning_rate": 1.2484817985395355e-05, "loss": 0.0564, "step": 6464 }, { "epoch": 6.56, "learning_rate": 1.2482694513860368e-05, "loss": 0.0524, "step": 6465 }, { "epoch": 6.56, "learning_rate": 1.2480570923017252e-05, "loss": 0.0627, "step": 6466 }, { "epoch": 6.56, "learning_rate": 1.2478447212968057e-05, "loss": 0.0569, "step": 6467 }, { "epoch": 6.56, "learning_rate": 1.247632338381484e-05, "loss": 0.0535, "step": 6468 }, { "epoch": 6.56, "learning_rate": 1.2474199435659666e-05, "loss": 0.0563, "step": 6469 }, { "epoch": 6.57, "learning_rate": 1.2472075368604602e-05, "loss": 0.0556, "step": 6470 }, { "epoch": 6.57, "learning_rate": 1.2469951182751719e-05, "loss": 0.0563, "step": 6471 }, { "epoch": 6.57, "learning_rate": 1.24678268782031e-05, "loss": 0.0723, "step": 6472 }, { "epoch": 6.57, "learning_rate": 1.2465702455060831e-05, "loss": 0.0696, "step": 6473 }, { "epoch": 6.57, "learning_rate": 1.2463577913427e-05, "loss": 0.0584, "step": 6474 }, { "epoch": 6.57, "learning_rate": 1.2461453253403703e-05, "loss": 0.0606, "step": 6475 }, { "epoch": 6.57, "learning_rate": 1.2459328475093046e-05, "loss": 0.0548, "step": 6476 }, { "epoch": 6.57, "learning_rate": 1.2457203578597139e-05, "loss": 0.0605, "step": 6477 }, { "epoch": 6.57, "learning_rate": 1.2455078564018091e-05, "loss": 0.0554, "step": 6478 }, { "epoch": 6.57, "learning_rate": 1.2452953431458021e-05, "loss": 0.0692, "step": 6479 }, { "epoch": 6.58, "learning_rate": 1.2450828181019058e-05, "loss": 0.0459, "step": 6480 }, { "epoch": 6.58, "learning_rate": 1.244870281280333e-05, "loss": 0.0711, "step": 6481 }, { "epoch": 6.58, "learning_rate": 1.2446577326912977e-05, "loss": 0.0707, "step": 6482 }, { "epoch": 6.58, "learning_rate": 1.2444451723450137e-05, "loss": 0.068, "step": 6483 }, { "epoch": 6.58, "learning_rate": 1.2442326002516957e-05, "loss": 0.0603, "step": 6484 }, { "epoch": 6.58, "learning_rate": 1.2440200164215599e-05, "loss": 0.0583, "step": 6485 }, { "epoch": 6.58, "learning_rate": 1.2438074208648212e-05, "loss": 0.0565, "step": 6486 }, { "epoch": 6.58, "learning_rate": 1.2435948135916967e-05, "loss": 0.0588, "step": 6487 }, { "epoch": 6.58, "learning_rate": 1.2433821946124027e-05, "loss": 0.0598, "step": 6488 }, { "epoch": 6.58, "learning_rate": 1.243169563937158e-05, "loss": 0.0717, "step": 6489 }, { "epoch": 6.59, "learning_rate": 1.2429569215761798e-05, "loss": 0.0623, "step": 6490 }, { "epoch": 6.59, "learning_rate": 1.2427442675396873e-05, "loss": 0.0555, "step": 6491 }, { "epoch": 6.59, "learning_rate": 1.242531601837899e-05, "loss": 0.0635, "step": 6492 }, { "epoch": 6.59, "learning_rate": 1.2423189244810357e-05, "loss": 0.064, "step": 6493 }, { "epoch": 6.59, "learning_rate": 1.2421062354793173e-05, "loss": 0.0572, "step": 6494 }, { "epoch": 6.59, "learning_rate": 1.241893534842965e-05, "loss": 0.0776, "step": 6495 }, { "epoch": 6.59, "learning_rate": 1.2416808225822004e-05, "loss": 0.0525, "step": 6496 }, { "epoch": 6.59, "learning_rate": 1.2414680987072449e-05, "loss": 0.07, "step": 6497 }, { "epoch": 6.59, "learning_rate": 1.2412553632283219e-05, "loss": 0.0685, "step": 6498 }, { "epoch": 6.59, "learning_rate": 1.2410426161556544e-05, "loss": 0.0683, "step": 6499 }, { "epoch": 6.6, "learning_rate": 1.240829857499466e-05, "loss": 0.0645, "step": 6500 }, { "epoch": 6.6, "learning_rate": 1.240617087269981e-05, "loss": 0.062, "step": 6501 }, { "epoch": 6.6, "learning_rate": 1.2404043054774244e-05, "loss": 0.0647, "step": 6502 }, { "epoch": 6.6, "learning_rate": 1.2401915121320214e-05, "loss": 0.0606, "step": 6503 }, { "epoch": 6.6, "learning_rate": 1.2399787072439984e-05, "loss": 0.0691, "step": 6504 }, { "epoch": 6.6, "learning_rate": 1.2397658908235814e-05, "loss": 0.061, "step": 6505 }, { "epoch": 6.6, "learning_rate": 1.2395530628809983e-05, "loss": 0.0795, "step": 6506 }, { "epoch": 6.6, "learning_rate": 1.2393402234264757e-05, "loss": 0.0505, "step": 6507 }, { "epoch": 6.6, "learning_rate": 1.2391273724702423e-05, "loss": 0.0639, "step": 6508 }, { "epoch": 6.6, "learning_rate": 1.2389145100225268e-05, "loss": 0.0677, "step": 6509 }, { "epoch": 6.61, "learning_rate": 1.2387016360935588e-05, "loss": 0.0647, "step": 6510 }, { "epoch": 6.61, "learning_rate": 1.2384887506935678e-05, "loss": 0.051, "step": 6511 }, { "epoch": 6.61, "learning_rate": 1.2382758538327844e-05, "loss": 0.0701, "step": 6512 }, { "epoch": 6.61, "learning_rate": 1.2380629455214392e-05, "loss": 0.0595, "step": 6513 }, { "epoch": 6.61, "learning_rate": 1.2378500257697641e-05, "loss": 0.0627, "step": 6514 }, { "epoch": 6.61, "learning_rate": 1.2376370945879907e-05, "loss": 0.0845, "step": 6515 }, { "epoch": 6.61, "learning_rate": 1.2374241519863523e-05, "loss": 0.0674, "step": 6516 }, { "epoch": 6.61, "learning_rate": 1.2372111979750814e-05, "loss": 0.0759, "step": 6517 }, { "epoch": 6.61, "learning_rate": 1.2369982325644121e-05, "loss": 0.0704, "step": 6518 }, { "epoch": 6.61, "learning_rate": 1.2367852557645785e-05, "loss": 0.066, "step": 6519 }, { "epoch": 6.62, "learning_rate": 1.2365722675858152e-05, "loss": 0.0569, "step": 6520 }, { "epoch": 6.62, "learning_rate": 1.2363592680383577e-05, "loss": 0.0665, "step": 6521 }, { "epoch": 6.62, "learning_rate": 1.2361462571324419e-05, "loss": 0.0559, "step": 6522 }, { "epoch": 6.62, "learning_rate": 1.2359332348783043e-05, "loss": 0.0532, "step": 6523 }, { "epoch": 6.62, "learning_rate": 1.2357202012861817e-05, "loss": 0.0646, "step": 6524 }, { "epoch": 6.62, "learning_rate": 1.2355071563663118e-05, "loss": 0.0557, "step": 6525 }, { "epoch": 6.62, "learning_rate": 1.2352941001289325e-05, "loss": 0.0512, "step": 6526 }, { "epoch": 6.62, "learning_rate": 1.2350810325842829e-05, "loss": 0.058, "step": 6527 }, { "epoch": 6.62, "learning_rate": 1.2348679537426013e-05, "loss": 0.0715, "step": 6528 }, { "epoch": 6.63, "learning_rate": 1.2346548636141282e-05, "loss": 0.0571, "step": 6529 }, { "epoch": 6.63, "learning_rate": 1.2344417622091032e-05, "loss": 0.0647, "step": 6530 }, { "epoch": 6.63, "learning_rate": 1.2342286495377674e-05, "loss": 0.0501, "step": 6531 }, { "epoch": 6.63, "learning_rate": 1.2340155256103623e-05, "loss": 0.0566, "step": 6532 }, { "epoch": 6.63, "learning_rate": 1.2338023904371295e-05, "loss": 0.0725, "step": 6533 }, { "epoch": 6.63, "learning_rate": 1.2335892440283113e-05, "loss": 0.064, "step": 6534 }, { "epoch": 6.63, "learning_rate": 1.2333760863941511e-05, "loss": 0.0644, "step": 6535 }, { "epoch": 6.63, "learning_rate": 1.2331629175448921e-05, "loss": 0.0888, "step": 6536 }, { "epoch": 6.63, "learning_rate": 1.232949737490778e-05, "loss": 0.0588, "step": 6537 }, { "epoch": 6.63, "learning_rate": 1.2327365462420535e-05, "loss": 0.065, "step": 6538 }, { "epoch": 6.64, "learning_rate": 1.2325233438089644e-05, "loss": 0.0708, "step": 6539 }, { "epoch": 6.64, "learning_rate": 1.2323101302017554e-05, "loss": 0.0685, "step": 6540 }, { "epoch": 6.64, "learning_rate": 1.2320969054306734e-05, "loss": 0.0556, "step": 6541 }, { "epoch": 6.64, "learning_rate": 1.2318836695059646e-05, "loss": 0.0647, "step": 6542 }, { "epoch": 6.64, "learning_rate": 1.2316704224378765e-05, "loss": 0.0607, "step": 6543 }, { "epoch": 6.64, "learning_rate": 1.2314571642366566e-05, "loss": 0.0714, "step": 6544 }, { "epoch": 6.64, "learning_rate": 1.2312438949125539e-05, "loss": 0.0643, "step": 6545 }, { "epoch": 6.64, "learning_rate": 1.2310306144758164e-05, "loss": 0.0618, "step": 6546 }, { "epoch": 6.64, "learning_rate": 1.230817322936694e-05, "loss": 0.074, "step": 6547 }, { "epoch": 6.64, "learning_rate": 1.2306040203054362e-05, "loss": 0.0637, "step": 6548 }, { "epoch": 6.65, "learning_rate": 1.230390706592294e-05, "loss": 0.0555, "step": 6549 }, { "epoch": 6.65, "learning_rate": 1.2301773818075182e-05, "loss": 0.0582, "step": 6550 }, { "epoch": 6.65, "learning_rate": 1.22996404596136e-05, "loss": 0.0523, "step": 6551 }, { "epoch": 6.65, "learning_rate": 1.2297506990640717e-05, "loss": 0.0558, "step": 6552 }, { "epoch": 6.65, "learning_rate": 1.229537341125906e-05, "loss": 0.0549, "step": 6553 }, { "epoch": 6.65, "learning_rate": 1.2293239721571157e-05, "loss": 0.0518, "step": 6554 }, { "epoch": 6.65, "learning_rate": 1.2291105921679547e-05, "loss": 0.0616, "step": 6555 }, { "epoch": 6.65, "learning_rate": 1.2288972011686773e-05, "loss": 0.0546, "step": 6556 }, { "epoch": 6.65, "learning_rate": 1.2286837991695377e-05, "loss": 0.0595, "step": 6557 }, { "epoch": 6.65, "learning_rate": 1.2284703861807917e-05, "loss": 0.0649, "step": 6558 }, { "epoch": 6.66, "learning_rate": 1.2282569622126938e-05, "loss": 0.0675, "step": 6559 }, { "epoch": 6.66, "learning_rate": 1.228043527275502e-05, "loss": 0.071, "step": 6560 }, { "epoch": 6.66, "learning_rate": 1.2278300813794726e-05, "loss": 0.0702, "step": 6561 }, { "epoch": 6.66, "learning_rate": 1.2276166245348624e-05, "loss": 0.0606, "step": 6562 }, { "epoch": 6.66, "learning_rate": 1.2274031567519291e-05, "loss": 0.0554, "step": 6563 }, { "epoch": 6.66, "learning_rate": 1.2271896780409321e-05, "loss": 0.0658, "step": 6564 }, { "epoch": 6.66, "learning_rate": 1.2269761884121293e-05, "loss": 0.0747, "step": 6565 }, { "epoch": 6.66, "learning_rate": 1.226762687875781e-05, "loss": 0.0692, "step": 6566 }, { "epoch": 6.66, "learning_rate": 1.2265491764421467e-05, "loss": 0.0689, "step": 6567 }, { "epoch": 6.66, "learning_rate": 1.2263356541214866e-05, "loss": 0.0734, "step": 6568 }, { "epoch": 6.67, "learning_rate": 1.226122120924062e-05, "loss": 0.0737, "step": 6569 }, { "epoch": 6.67, "learning_rate": 1.2259085768601347e-05, "loss": 0.0581, "step": 6570 }, { "epoch": 6.67, "learning_rate": 1.2256950219399663e-05, "loss": 0.0551, "step": 6571 }, { "epoch": 6.67, "learning_rate": 1.2254814561738198e-05, "loss": 0.0653, "step": 6572 }, { "epoch": 6.67, "learning_rate": 1.2252678795719581e-05, "loss": 0.0556, "step": 6573 }, { "epoch": 6.67, "learning_rate": 1.2250542921446447e-05, "loss": 0.0558, "step": 6574 }, { "epoch": 6.67, "learning_rate": 1.2248406939021438e-05, "loss": 0.0537, "step": 6575 }, { "epoch": 6.67, "learning_rate": 1.22462708485472e-05, "loss": 0.0668, "step": 6576 }, { "epoch": 6.67, "learning_rate": 1.2244134650126386e-05, "loss": 0.0694, "step": 6577 }, { "epoch": 6.67, "learning_rate": 1.2241998343861656e-05, "loss": 0.0579, "step": 6578 }, { "epoch": 6.68, "learning_rate": 1.2239861929855667e-05, "loss": 0.055, "step": 6579 }, { "epoch": 6.68, "learning_rate": 1.2237725408211084e-05, "loss": 0.065, "step": 6580 }, { "epoch": 6.68, "learning_rate": 1.2235588779030587e-05, "loss": 0.0582, "step": 6581 }, { "epoch": 6.68, "learning_rate": 1.2233452042416851e-05, "loss": 0.0712, "step": 6582 }, { "epoch": 6.68, "learning_rate": 1.2231315198472558e-05, "loss": 0.0537, "step": 6583 }, { "epoch": 6.68, "learning_rate": 1.2229178247300393e-05, "loss": 0.0621, "step": 6584 }, { "epoch": 6.68, "learning_rate": 1.2227041189003055e-05, "loss": 0.0625, "step": 6585 }, { "epoch": 6.68, "learning_rate": 1.2224904023683239e-05, "loss": 0.0666, "step": 6586 }, { "epoch": 6.68, "learning_rate": 1.2222766751443649e-05, "loss": 0.0513, "step": 6587 }, { "epoch": 6.68, "learning_rate": 1.2220629372386989e-05, "loss": 0.0571, "step": 6588 }, { "epoch": 6.69, "learning_rate": 1.2218491886615981e-05, "loss": 0.0478, "step": 6589 }, { "epoch": 6.69, "learning_rate": 1.2216354294233342e-05, "loss": 0.0499, "step": 6590 }, { "epoch": 6.69, "learning_rate": 1.2214216595341792e-05, "loss": 0.0568, "step": 6591 }, { "epoch": 6.69, "learning_rate": 1.221207879004406e-05, "loss": 0.0606, "step": 6592 }, { "epoch": 6.69, "learning_rate": 1.2209940878442881e-05, "loss": 0.0464, "step": 6593 }, { "epoch": 6.69, "learning_rate": 1.2207802860640998e-05, "loss": 0.0516, "step": 6594 }, { "epoch": 6.69, "learning_rate": 1.2205664736741152e-05, "loss": 0.057, "step": 6595 }, { "epoch": 6.69, "learning_rate": 1.2203526506846095e-05, "loss": 0.0675, "step": 6596 }, { "epoch": 6.69, "learning_rate": 1.2201388171058575e-05, "loss": 0.0672, "step": 6597 }, { "epoch": 6.7, "learning_rate": 1.2199249729481356e-05, "loss": 0.0735, "step": 6598 }, { "epoch": 6.7, "learning_rate": 1.2197111182217208e-05, "loss": 0.0702, "step": 6599 }, { "epoch": 6.7, "learning_rate": 1.2194972529368893e-05, "loss": 0.0596, "step": 6600 }, { "epoch": 6.7, "learning_rate": 1.2192833771039189e-05, "loss": 0.0657, "step": 6601 }, { "epoch": 6.7, "learning_rate": 1.2190694907330878e-05, "loss": 0.0606, "step": 6602 }, { "epoch": 6.7, "learning_rate": 1.218855593834674e-05, "loss": 0.0587, "step": 6603 }, { "epoch": 6.7, "learning_rate": 1.218641686418957e-05, "loss": 0.057, "step": 6604 }, { "epoch": 6.7, "learning_rate": 1.2184277684962158e-05, "loss": 0.0631, "step": 6605 }, { "epoch": 6.7, "learning_rate": 1.2182138400767309e-05, "loss": 0.0535, "step": 6606 }, { "epoch": 6.7, "learning_rate": 1.2179999011707827e-05, "loss": 0.0768, "step": 6607 }, { "epoch": 6.71, "learning_rate": 1.2177859517886524e-05, "loss": 0.0486, "step": 6608 }, { "epoch": 6.71, "learning_rate": 1.2175719919406208e-05, "loss": 0.058, "step": 6609 }, { "epoch": 6.71, "learning_rate": 1.2173580216369704e-05, "loss": 0.0668, "step": 6610 }, { "epoch": 6.71, "learning_rate": 1.2171440408879842e-05, "loss": 0.0615, "step": 6611 }, { "epoch": 6.71, "learning_rate": 1.2169300497039447e-05, "loss": 0.0582, "step": 6612 }, { "epoch": 6.71, "learning_rate": 1.2167160480951357e-05, "loss": 0.0673, "step": 6613 }, { "epoch": 6.71, "learning_rate": 1.2165020360718407e-05, "loss": 0.0559, "step": 6614 }, { "epoch": 6.71, "learning_rate": 1.2162880136443447e-05, "loss": 0.057, "step": 6615 }, { "epoch": 6.71, "learning_rate": 1.2160739808229326e-05, "loss": 0.0542, "step": 6616 }, { "epoch": 6.71, "learning_rate": 1.2158599376178905e-05, "loss": 0.0662, "step": 6617 }, { "epoch": 6.72, "learning_rate": 1.2156458840395033e-05, "loss": 0.0607, "step": 6618 }, { "epoch": 6.72, "learning_rate": 1.2154318200980589e-05, "loss": 0.0562, "step": 6619 }, { "epoch": 6.72, "learning_rate": 1.215217745803843e-05, "loss": 0.0556, "step": 6620 }, { "epoch": 6.72, "learning_rate": 1.215003661167144e-05, "loss": 0.0743, "step": 6621 }, { "epoch": 6.72, "learning_rate": 1.2147895661982499e-05, "loss": 0.0612, "step": 6622 }, { "epoch": 6.72, "learning_rate": 1.2145754609074488e-05, "loss": 0.0587, "step": 6623 }, { "epoch": 6.72, "learning_rate": 1.21436134530503e-05, "loss": 0.057, "step": 6624 }, { "epoch": 6.72, "learning_rate": 1.214147219401283e-05, "loss": 0.0564, "step": 6625 }, { "epoch": 6.72, "learning_rate": 1.2139330832064975e-05, "loss": 0.0711, "step": 6626 }, { "epoch": 6.72, "learning_rate": 1.2137189367309645e-05, "loss": 0.0616, "step": 6627 }, { "epoch": 6.73, "learning_rate": 1.2135047799849748e-05, "loss": 0.0814, "step": 6628 }, { "epoch": 6.73, "learning_rate": 1.21329061297882e-05, "loss": 0.0662, "step": 6629 }, { "epoch": 6.73, "learning_rate": 1.2130764357227918e-05, "loss": 0.0698, "step": 6630 }, { "epoch": 6.73, "learning_rate": 1.2128622482271828e-05, "loss": 0.0593, "step": 6631 }, { "epoch": 6.73, "learning_rate": 1.2126480505022861e-05, "loss": 0.0524, "step": 6632 }, { "epoch": 6.73, "learning_rate": 1.2124338425583953e-05, "loss": 0.0647, "step": 6633 }, { "epoch": 6.73, "learning_rate": 1.2122196244058037e-05, "loss": 0.059, "step": 6634 }, { "epoch": 6.73, "learning_rate": 1.2120053960548067e-05, "loss": 0.0557, "step": 6635 }, { "epoch": 6.73, "learning_rate": 1.2117911575156987e-05, "loss": 0.0521, "step": 6636 }, { "epoch": 6.73, "learning_rate": 1.2115769087987748e-05, "loss": 0.0637, "step": 6637 }, { "epoch": 6.74, "learning_rate": 1.2113626499143314e-05, "loss": 0.0728, "step": 6638 }, { "epoch": 6.74, "learning_rate": 1.2111483808726647e-05, "loss": 0.0628, "step": 6639 }, { "epoch": 6.74, "learning_rate": 1.2109341016840719e-05, "loss": 0.0598, "step": 6640 }, { "epoch": 6.74, "learning_rate": 1.2107198123588502e-05, "loss": 0.0669, "step": 6641 }, { "epoch": 6.74, "learning_rate": 1.2105055129072972e-05, "loss": 0.07, "step": 6642 }, { "epoch": 6.74, "learning_rate": 1.2102912033397112e-05, "loss": 0.0523, "step": 6643 }, { "epoch": 6.74, "learning_rate": 1.2100768836663916e-05, "loss": 0.0607, "step": 6644 }, { "epoch": 6.74, "learning_rate": 1.2098625538976378e-05, "loss": 0.0716, "step": 6645 }, { "epoch": 6.74, "learning_rate": 1.209648214043749e-05, "loss": 0.0616, "step": 6646 }, { "epoch": 6.74, "learning_rate": 1.2094338641150255e-05, "loss": 0.0688, "step": 6647 }, { "epoch": 6.75, "learning_rate": 1.2092195041217684e-05, "loss": 0.0755, "step": 6648 }, { "epoch": 6.75, "learning_rate": 1.209005134074279e-05, "loss": 0.0564, "step": 6649 }, { "epoch": 6.75, "learning_rate": 1.2087907539828592e-05, "loss": 0.0604, "step": 6650 }, { "epoch": 6.75, "learning_rate": 1.2085763638578106e-05, "loss": 0.0652, "step": 6651 }, { "epoch": 6.75, "learning_rate": 1.2083619637094363e-05, "loss": 0.0489, "step": 6652 }, { "epoch": 6.75, "learning_rate": 1.2081475535480393e-05, "loss": 0.0657, "step": 6653 }, { "epoch": 6.75, "learning_rate": 1.207933133383924e-05, "loss": 0.0548, "step": 6654 }, { "epoch": 6.75, "learning_rate": 1.2077187032273934e-05, "loss": 0.0672, "step": 6655 }, { "epoch": 6.75, "learning_rate": 1.2075042630887532e-05, "loss": 0.069, "step": 6656 }, { "epoch": 6.75, "learning_rate": 1.207289812978308e-05, "loss": 0.0661, "step": 6657 }, { "epoch": 6.76, "learning_rate": 1.2070753529063633e-05, "loss": 0.0596, "step": 6658 }, { "epoch": 6.76, "learning_rate": 1.206860882883225e-05, "loss": 0.0655, "step": 6659 }, { "epoch": 6.76, "learning_rate": 1.2066464029192001e-05, "loss": 0.0559, "step": 6660 }, { "epoch": 6.76, "learning_rate": 1.2064319130245958e-05, "loss": 0.0458, "step": 6661 }, { "epoch": 6.76, "learning_rate": 1.2062174132097194e-05, "loss": 0.0637, "step": 6662 }, { "epoch": 6.76, "learning_rate": 1.2060029034848786e-05, "loss": 0.0603, "step": 6663 }, { "epoch": 6.76, "learning_rate": 1.205788383860382e-05, "loss": 0.0636, "step": 6664 }, { "epoch": 6.76, "learning_rate": 1.2055738543465383e-05, "loss": 0.0706, "step": 6665 }, { "epoch": 6.76, "learning_rate": 1.2053593149536576e-05, "loss": 0.0575, "step": 6666 }, { "epoch": 6.77, "learning_rate": 1.2051447656920491e-05, "loss": 0.0719, "step": 6667 }, { "epoch": 6.77, "learning_rate": 1.2049302065720237e-05, "loss": 0.0795, "step": 6668 }, { "epoch": 6.77, "learning_rate": 1.2047156376038918e-05, "loss": 0.0639, "step": 6669 }, { "epoch": 6.77, "learning_rate": 1.2045010587979646e-05, "loss": 0.0606, "step": 6670 }, { "epoch": 6.77, "learning_rate": 1.2042864701645547e-05, "loss": 0.0618, "step": 6671 }, { "epoch": 6.77, "learning_rate": 1.2040718717139732e-05, "loss": 0.0604, "step": 6672 }, { "epoch": 6.77, "learning_rate": 1.203857263456534e-05, "loss": 0.0677, "step": 6673 }, { "epoch": 6.77, "learning_rate": 1.2036426454025494e-05, "loss": 0.0651, "step": 6674 }, { "epoch": 6.77, "learning_rate": 1.2034280175623331e-05, "loss": 0.0831, "step": 6675 }, { "epoch": 6.77, "learning_rate": 1.2032133799461998e-05, "loss": 0.0546, "step": 6676 }, { "epoch": 6.78, "learning_rate": 1.2029987325644642e-05, "loss": 0.0663, "step": 6677 }, { "epoch": 6.78, "learning_rate": 1.2027840754274404e-05, "loss": 0.0567, "step": 6678 }, { "epoch": 6.78, "learning_rate": 1.202569408545445e-05, "loss": 0.0717, "step": 6679 }, { "epoch": 6.78, "learning_rate": 1.2023547319287932e-05, "loss": 0.0557, "step": 6680 }, { "epoch": 6.78, "learning_rate": 1.2021400455878021e-05, "loss": 0.056, "step": 6681 }, { "epoch": 6.78, "learning_rate": 1.2019253495327878e-05, "loss": 0.0593, "step": 6682 }, { "epoch": 6.78, "learning_rate": 1.2017106437740688e-05, "loss": 0.0517, "step": 6683 }, { "epoch": 6.78, "learning_rate": 1.2014959283219625e-05, "loss": 0.0612, "step": 6684 }, { "epoch": 6.78, "learning_rate": 1.2012812031867869e-05, "loss": 0.0581, "step": 6685 }, { "epoch": 6.78, "learning_rate": 1.2010664683788614e-05, "loss": 0.0619, "step": 6686 }, { "epoch": 6.79, "learning_rate": 1.2008517239085046e-05, "loss": 0.0623, "step": 6687 }, { "epoch": 6.79, "learning_rate": 1.2006369697860372e-05, "loss": 0.0629, "step": 6688 }, { "epoch": 6.79, "learning_rate": 1.2004222060217783e-05, "loss": 0.0789, "step": 6689 }, { "epoch": 6.79, "learning_rate": 1.2002074326260493e-05, "loss": 0.0707, "step": 6690 }, { "epoch": 6.79, "learning_rate": 1.1999926496091715e-05, "loss": 0.064, "step": 6691 }, { "epoch": 6.79, "learning_rate": 1.1997778569814656e-05, "loss": 0.0599, "step": 6692 }, { "epoch": 6.79, "learning_rate": 1.1995630547532539e-05, "loss": 0.0512, "step": 6693 }, { "epoch": 6.79, "learning_rate": 1.1993482429348599e-05, "loss": 0.0679, "step": 6694 }, { "epoch": 6.79, "learning_rate": 1.1991334215366058e-05, "loss": 0.0512, "step": 6695 }, { "epoch": 6.79, "learning_rate": 1.1989185905688147e-05, "loss": 0.061, "step": 6696 }, { "epoch": 6.8, "learning_rate": 1.198703750041811e-05, "loss": 0.0746, "step": 6697 }, { "epoch": 6.8, "learning_rate": 1.198488899965919e-05, "loss": 0.0645, "step": 6698 }, { "epoch": 6.8, "learning_rate": 1.1982740403514637e-05, "loss": 0.0641, "step": 6699 }, { "epoch": 6.8, "learning_rate": 1.19805917120877e-05, "loss": 0.0637, "step": 6700 }, { "epoch": 6.8, "learning_rate": 1.1978442925481637e-05, "loss": 0.0632, "step": 6701 }, { "epoch": 6.8, "learning_rate": 1.1976294043799712e-05, "loss": 0.0594, "step": 6702 }, { "epoch": 6.8, "learning_rate": 1.1974145067145189e-05, "loss": 0.0688, "step": 6703 }, { "epoch": 6.8, "learning_rate": 1.1971995995621342e-05, "loss": 0.0601, "step": 6704 }, { "epoch": 6.8, "learning_rate": 1.1969846829331442e-05, "loss": 0.0661, "step": 6705 }, { "epoch": 6.8, "learning_rate": 1.1967697568378773e-05, "loss": 0.0569, "step": 6706 }, { "epoch": 6.81, "learning_rate": 1.196554821286662e-05, "loss": 0.0471, "step": 6707 }, { "epoch": 6.81, "learning_rate": 1.196339876289827e-05, "loss": 0.0617, "step": 6708 }, { "epoch": 6.81, "learning_rate": 1.1961249218577015e-05, "loss": 0.0715, "step": 6709 }, { "epoch": 6.81, "learning_rate": 1.1959099580006157e-05, "loss": 0.0715, "step": 6710 }, { "epoch": 6.81, "learning_rate": 1.1956949847288998e-05, "loss": 0.0762, "step": 6711 }, { "epoch": 6.81, "learning_rate": 1.1954800020528846e-05, "loss": 0.0588, "step": 6712 }, { "epoch": 6.81, "learning_rate": 1.1952650099829013e-05, "loss": 0.0554, "step": 6713 }, { "epoch": 6.81, "learning_rate": 1.195050008529281e-05, "loss": 0.0607, "step": 6714 }, { "epoch": 6.81, "learning_rate": 1.1948349977023563e-05, "loss": 0.0595, "step": 6715 }, { "epoch": 6.81, "learning_rate": 1.1946199775124598e-05, "loss": 0.064, "step": 6716 }, { "epoch": 6.82, "learning_rate": 1.1944049479699244e-05, "loss": 0.0568, "step": 6717 }, { "epoch": 6.82, "learning_rate": 1.1941899090850832e-05, "loss": 0.0508, "step": 6718 }, { "epoch": 6.82, "learning_rate": 1.1939748608682706e-05, "loss": 0.0625, "step": 6719 }, { "epoch": 6.82, "learning_rate": 1.1937598033298205e-05, "loss": 0.0601, "step": 6720 }, { "epoch": 6.82, "learning_rate": 1.1935447364800683e-05, "loss": 0.0695, "step": 6721 }, { "epoch": 6.82, "learning_rate": 1.1933296603293483e-05, "loss": 0.061, "step": 6722 }, { "epoch": 6.82, "learning_rate": 1.193114574887997e-05, "loss": 0.0707, "step": 6723 }, { "epoch": 6.82, "learning_rate": 1.1928994801663503e-05, "loss": 0.0597, "step": 6724 }, { "epoch": 6.82, "learning_rate": 1.1926843761747444e-05, "loss": 0.0787, "step": 6725 }, { "epoch": 6.82, "learning_rate": 1.1924692629235167e-05, "loss": 0.0667, "step": 6726 }, { "epoch": 6.83, "learning_rate": 1.1922541404230045e-05, "loss": 0.0631, "step": 6727 }, { "epoch": 6.83, "learning_rate": 1.1920390086835463e-05, "loss": 0.0695, "step": 6728 }, { "epoch": 6.83, "learning_rate": 1.1918238677154794e-05, "loss": 0.0574, "step": 6729 }, { "epoch": 6.83, "learning_rate": 1.1916087175291436e-05, "loss": 0.0535, "step": 6730 }, { "epoch": 6.83, "learning_rate": 1.191393558134877e-05, "loss": 0.0659, "step": 6731 }, { "epoch": 6.83, "learning_rate": 1.1911783895430203e-05, "loss": 0.0587, "step": 6732 }, { "epoch": 6.83, "learning_rate": 1.1909632117639132e-05, "loss": 0.0669, "step": 6733 }, { "epoch": 6.83, "learning_rate": 1.1907480248078965e-05, "loss": 0.0626, "step": 6734 }, { "epoch": 6.83, "learning_rate": 1.190532828685311e-05, "loss": 0.0729, "step": 6735 }, { "epoch": 6.84, "learning_rate": 1.1903176234064979e-05, "loss": 0.073, "step": 6736 }, { "epoch": 6.84, "learning_rate": 1.1901024089817994e-05, "loss": 0.0723, "step": 6737 }, { "epoch": 6.84, "learning_rate": 1.189887185421558e-05, "loss": 0.0619, "step": 6738 }, { "epoch": 6.84, "learning_rate": 1.1896719527361157e-05, "loss": 0.0563, "step": 6739 }, { "epoch": 6.84, "learning_rate": 1.1894567109358168e-05, "loss": 0.0663, "step": 6740 }, { "epoch": 6.84, "learning_rate": 1.1892414600310046e-05, "loss": 0.079, "step": 6741 }, { "epoch": 6.84, "learning_rate": 1.1890262000320222e-05, "loss": 0.0737, "step": 6742 }, { "epoch": 6.84, "learning_rate": 1.1888109309492153e-05, "loss": 0.0634, "step": 6743 }, { "epoch": 6.84, "learning_rate": 1.1885956527929286e-05, "loss": 0.0584, "step": 6744 }, { "epoch": 6.84, "learning_rate": 1.188380365573507e-05, "loss": 0.0686, "step": 6745 }, { "epoch": 6.85, "learning_rate": 1.1881650693012969e-05, "loss": 0.0672, "step": 6746 }, { "epoch": 6.85, "learning_rate": 1.187949763986644e-05, "loss": 0.0695, "step": 6747 }, { "epoch": 6.85, "learning_rate": 1.1877344496398952e-05, "loss": 0.0707, "step": 6748 }, { "epoch": 6.85, "learning_rate": 1.1875191262713979e-05, "loss": 0.0559, "step": 6749 }, { "epoch": 6.85, "learning_rate": 1.1873037938914998e-05, "loss": 0.0672, "step": 6750 }, { "epoch": 6.85, "learning_rate": 1.187088452510548e-05, "loss": 0.0633, "step": 6751 }, { "epoch": 6.85, "learning_rate": 1.1868731021388917e-05, "loss": 0.059, "step": 6752 }, { "epoch": 6.85, "learning_rate": 1.1866577427868798e-05, "loss": 0.0524, "step": 6753 }, { "epoch": 6.85, "learning_rate": 1.1864423744648608e-05, "loss": 0.058, "step": 6754 }, { "epoch": 6.85, "learning_rate": 1.1862269971831853e-05, "loss": 0.069, "step": 6755 }, { "epoch": 6.86, "learning_rate": 1.1860116109522028e-05, "loss": 0.0659, "step": 6756 }, { "epoch": 6.86, "learning_rate": 1.1857962157822645e-05, "loss": 0.0667, "step": 6757 }, { "epoch": 6.86, "learning_rate": 1.185580811683721e-05, "loss": 0.0533, "step": 6758 }, { "epoch": 6.86, "learning_rate": 1.1853653986669238e-05, "loss": 0.0565, "step": 6759 }, { "epoch": 6.86, "learning_rate": 1.1851499767422246e-05, "loss": 0.0541, "step": 6760 }, { "epoch": 6.86, "learning_rate": 1.1849345459199761e-05, "loss": 0.0648, "step": 6761 }, { "epoch": 6.86, "learning_rate": 1.184719106210531e-05, "loss": 0.0546, "step": 6762 }, { "epoch": 6.86, "learning_rate": 1.184503657624242e-05, "loss": 0.0584, "step": 6763 }, { "epoch": 6.86, "learning_rate": 1.1842882001714629e-05, "loss": 0.0613, "step": 6764 }, { "epoch": 6.86, "learning_rate": 1.1840727338625477e-05, "loss": 0.0636, "step": 6765 }, { "epoch": 6.87, "learning_rate": 1.183857258707851e-05, "loss": 0.0536, "step": 6766 }, { "epoch": 6.87, "learning_rate": 1.1836417747177276e-05, "loss": 0.0768, "step": 6767 }, { "epoch": 6.87, "learning_rate": 1.1834262819025326e-05, "loss": 0.067, "step": 6768 }, { "epoch": 6.87, "learning_rate": 1.1832107802726218e-05, "loss": 0.0616, "step": 6769 }, { "epoch": 6.87, "learning_rate": 1.182995269838351e-05, "loss": 0.0688, "step": 6770 }, { "epoch": 6.87, "learning_rate": 1.1827797506100777e-05, "loss": 0.0576, "step": 6771 }, { "epoch": 6.87, "learning_rate": 1.1825642225981577e-05, "loss": 0.0769, "step": 6772 }, { "epoch": 6.87, "learning_rate": 1.1823486858129493e-05, "loss": 0.0635, "step": 6773 }, { "epoch": 6.87, "learning_rate": 1.18213314026481e-05, "loss": 0.0551, "step": 6774 }, { "epoch": 6.87, "learning_rate": 1.1819175859640979e-05, "loss": 0.0665, "step": 6775 }, { "epoch": 6.88, "learning_rate": 1.1817020229211713e-05, "loss": 0.0733, "step": 6776 }, { "epoch": 6.88, "learning_rate": 1.1814864511463902e-05, "loss": 0.069, "step": 6777 }, { "epoch": 6.88, "learning_rate": 1.1812708706501136e-05, "loss": 0.0699, "step": 6778 }, { "epoch": 6.88, "learning_rate": 1.1810552814427014e-05, "loss": 0.0682, "step": 6779 }, { "epoch": 6.88, "learning_rate": 1.1808396835345142e-05, "loss": 0.049, "step": 6780 }, { "epoch": 6.88, "learning_rate": 1.1806240769359121e-05, "loss": 0.0603, "step": 6781 }, { "epoch": 6.88, "learning_rate": 1.1804084616572566e-05, "loss": 0.0606, "step": 6782 }, { "epoch": 6.88, "learning_rate": 1.1801928377089098e-05, "loss": 0.071, "step": 6783 }, { "epoch": 6.88, "learning_rate": 1.179977205101233e-05, "loss": 0.0585, "step": 6784 }, { "epoch": 6.88, "learning_rate": 1.179761563844589e-05, "loss": 0.0653, "step": 6785 }, { "epoch": 6.89, "learning_rate": 1.1795459139493406e-05, "loss": 0.0691, "step": 6786 }, { "epoch": 6.89, "learning_rate": 1.1793302554258508e-05, "loss": 0.0585, "step": 6787 }, { "epoch": 6.89, "learning_rate": 1.1791145882844837e-05, "loss": 0.0649, "step": 6788 }, { "epoch": 6.89, "learning_rate": 1.178898912535603e-05, "loss": 0.0887, "step": 6789 }, { "epoch": 6.89, "learning_rate": 1.1786832281895734e-05, "loss": 0.0578, "step": 6790 }, { "epoch": 6.89, "learning_rate": 1.1784675352567599e-05, "loss": 0.0723, "step": 6791 }, { "epoch": 6.89, "learning_rate": 1.1782518337475272e-05, "loss": 0.0515, "step": 6792 }, { "epoch": 6.89, "learning_rate": 1.1780361236722415e-05, "loss": 0.0799, "step": 6793 }, { "epoch": 6.89, "learning_rate": 1.1778204050412693e-05, "loss": 0.0532, "step": 6794 }, { "epoch": 6.89, "learning_rate": 1.1776046778649768e-05, "loss": 0.0791, "step": 6795 }, { "epoch": 6.9, "learning_rate": 1.1773889421537307e-05, "loss": 0.0626, "step": 6796 }, { "epoch": 6.9, "learning_rate": 1.1771731979178987e-05, "loss": 0.0605, "step": 6797 }, { "epoch": 6.9, "learning_rate": 1.1769574451678483e-05, "loss": 0.0672, "step": 6798 }, { "epoch": 6.9, "learning_rate": 1.1767416839139483e-05, "loss": 0.0662, "step": 6799 }, { "epoch": 6.9, "learning_rate": 1.1765259141665668e-05, "loss": 0.0692, "step": 6800 }, { "epoch": 6.9, "learning_rate": 1.176310135936073e-05, "loss": 0.0754, "step": 6801 }, { "epoch": 6.9, "learning_rate": 1.1760943492328358e-05, "loss": 0.0657, "step": 6802 }, { "epoch": 6.9, "learning_rate": 1.1758785540672257e-05, "loss": 0.0608, "step": 6803 }, { "epoch": 6.9, "learning_rate": 1.1756627504496126e-05, "loss": 0.0636, "step": 6804 }, { "epoch": 6.91, "learning_rate": 1.1754469383903673e-05, "loss": 0.0702, "step": 6805 }, { "epoch": 6.91, "learning_rate": 1.1752311178998608e-05, "loss": 0.0609, "step": 6806 }, { "epoch": 6.91, "learning_rate": 1.1750152889884646e-05, "loss": 0.063, "step": 6807 }, { "epoch": 6.91, "learning_rate": 1.1747994516665504e-05, "loss": 0.0642, "step": 6808 }, { "epoch": 6.91, "learning_rate": 1.1745836059444903e-05, "loss": 0.0709, "step": 6809 }, { "epoch": 6.91, "learning_rate": 1.1743677518326573e-05, "loss": 0.0526, "step": 6810 }, { "epoch": 6.91, "learning_rate": 1.1741518893414244e-05, "loss": 0.0575, "step": 6811 }, { "epoch": 6.91, "learning_rate": 1.1739360184811653e-05, "loss": 0.0708, "step": 6812 }, { "epoch": 6.91, "learning_rate": 1.1737201392622532e-05, "loss": 0.0631, "step": 6813 }, { "epoch": 6.91, "learning_rate": 1.1735042516950628e-05, "loss": 0.0765, "step": 6814 }, { "epoch": 6.92, "learning_rate": 1.1732883557899687e-05, "loss": 0.0629, "step": 6815 }, { "epoch": 6.92, "learning_rate": 1.1730724515573461e-05, "loss": 0.058, "step": 6816 }, { "epoch": 6.92, "learning_rate": 1.1728565390075705e-05, "loss": 0.0576, "step": 6817 }, { "epoch": 6.92, "learning_rate": 1.1726406181510175e-05, "loss": 0.0618, "step": 6818 }, { "epoch": 6.92, "learning_rate": 1.1724246889980638e-05, "loss": 0.0614, "step": 6819 }, { "epoch": 6.92, "learning_rate": 1.1722087515590853e-05, "loss": 0.0632, "step": 6820 }, { "epoch": 6.92, "learning_rate": 1.1719928058444598e-05, "loss": 0.0706, "step": 6821 }, { "epoch": 6.92, "learning_rate": 1.1717768518645645e-05, "loss": 0.0601, "step": 6822 }, { "epoch": 6.92, "learning_rate": 1.1715608896297775e-05, "loss": 0.0614, "step": 6823 }, { "epoch": 6.92, "learning_rate": 1.1713449191504767e-05, "loss": 0.0609, "step": 6824 }, { "epoch": 6.93, "learning_rate": 1.171128940437041e-05, "loss": 0.0727, "step": 6825 }, { "epoch": 6.93, "learning_rate": 1.1709129534998492e-05, "loss": 0.0653, "step": 6826 }, { "epoch": 6.93, "learning_rate": 1.1706969583492808e-05, "loss": 0.0612, "step": 6827 }, { "epoch": 6.93, "learning_rate": 1.1704809549957159e-05, "loss": 0.0596, "step": 6828 }, { "epoch": 6.93, "learning_rate": 1.1702649434495348e-05, "loss": 0.0679, "step": 6829 }, { "epoch": 6.93, "learning_rate": 1.170048923721118e-05, "loss": 0.0669, "step": 6830 }, { "epoch": 6.93, "learning_rate": 1.1698328958208459e-05, "loss": 0.0683, "step": 6831 }, { "epoch": 6.93, "learning_rate": 1.1696168597591006e-05, "loss": 0.0692, "step": 6832 }, { "epoch": 6.93, "learning_rate": 1.1694008155462639e-05, "loss": 0.0683, "step": 6833 }, { "epoch": 6.93, "learning_rate": 1.169184763192718e-05, "loss": 0.0655, "step": 6834 }, { "epoch": 6.94, "learning_rate": 1.168968702708845e-05, "loss": 0.059, "step": 6835 }, { "epoch": 6.94, "learning_rate": 1.1687526341050287e-05, "loss": 0.0544, "step": 6836 }, { "epoch": 6.94, "learning_rate": 1.1685365573916515e-05, "loss": 0.0636, "step": 6837 }, { "epoch": 6.94, "learning_rate": 1.1683204725790982e-05, "loss": 0.0565, "step": 6838 }, { "epoch": 6.94, "learning_rate": 1.168104379677752e-05, "loss": 0.0749, "step": 6839 }, { "epoch": 6.94, "learning_rate": 1.1678882786979982e-05, "loss": 0.0652, "step": 6840 }, { "epoch": 6.94, "learning_rate": 1.1676721696502211e-05, "loss": 0.0546, "step": 6841 }, { "epoch": 6.94, "learning_rate": 1.1674560525448065e-05, "loss": 0.0602, "step": 6842 }, { "epoch": 6.94, "learning_rate": 1.1672399273921396e-05, "loss": 0.0616, "step": 6843 }, { "epoch": 6.94, "learning_rate": 1.1670237942026071e-05, "loss": 0.0809, "step": 6844 }, { "epoch": 6.95, "learning_rate": 1.166807652986595e-05, "loss": 0.0636, "step": 6845 }, { "epoch": 6.95, "learning_rate": 1.1665915037544905e-05, "loss": 0.0704, "step": 6846 }, { "epoch": 6.95, "learning_rate": 1.1663753465166808e-05, "loss": 0.0633, "step": 6847 }, { "epoch": 6.95, "learning_rate": 1.166159181283553e-05, "loss": 0.0567, "step": 6848 }, { "epoch": 6.95, "learning_rate": 1.165943008065496e-05, "loss": 0.0707, "step": 6849 }, { "epoch": 6.95, "learning_rate": 1.1657268268728973e-05, "loss": 0.0695, "step": 6850 }, { "epoch": 6.95, "learning_rate": 1.1655106377161465e-05, "loss": 0.0691, "step": 6851 }, { "epoch": 6.95, "learning_rate": 1.1652944406056321e-05, "loss": 0.0733, "step": 6852 }, { "epoch": 6.95, "learning_rate": 1.165078235551744e-05, "loss": 0.0765, "step": 6853 }, { "epoch": 6.95, "learning_rate": 1.1648620225648722e-05, "loss": 0.0668, "step": 6854 }, { "epoch": 6.96, "learning_rate": 1.1646458016554068e-05, "loss": 0.0602, "step": 6855 }, { "epoch": 6.96, "learning_rate": 1.1644295728337383e-05, "loss": 0.0635, "step": 6856 }, { "epoch": 6.96, "learning_rate": 1.1642133361102582e-05, "loss": 0.0773, "step": 6857 }, { "epoch": 6.96, "learning_rate": 1.1639970914953578e-05, "loss": 0.0548, "step": 6858 }, { "epoch": 6.96, "learning_rate": 1.163780838999429e-05, "loss": 0.0655, "step": 6859 }, { "epoch": 6.96, "learning_rate": 1.1635645786328636e-05, "loss": 0.066, "step": 6860 }, { "epoch": 6.96, "learning_rate": 1.1633483104060549e-05, "loss": 0.0786, "step": 6861 }, { "epoch": 6.96, "learning_rate": 1.1631320343293952e-05, "loss": 0.0835, "step": 6862 }, { "epoch": 6.96, "learning_rate": 1.1629157504132785e-05, "loss": 0.0536, "step": 6863 }, { "epoch": 6.96, "learning_rate": 1.1626994586680975e-05, "loss": 0.0622, "step": 6864 }, { "epoch": 6.97, "learning_rate": 1.1624831591042471e-05, "loss": 0.0653, "step": 6865 }, { "epoch": 6.97, "learning_rate": 1.1622668517321221e-05, "loss": 0.0856, "step": 6866 }, { "epoch": 6.97, "learning_rate": 1.1620505365621164e-05, "loss": 0.0646, "step": 6867 }, { "epoch": 6.97, "learning_rate": 1.161834213604626e-05, "loss": 0.0519, "step": 6868 }, { "epoch": 6.97, "learning_rate": 1.1616178828700458e-05, "loss": 0.0586, "step": 6869 }, { "epoch": 6.97, "learning_rate": 1.1614015443687723e-05, "loss": 0.0595, "step": 6870 }, { "epoch": 6.97, "learning_rate": 1.1611851981112017e-05, "loss": 0.0609, "step": 6871 }, { "epoch": 6.97, "learning_rate": 1.1609688441077307e-05, "loss": 0.0695, "step": 6872 }, { "epoch": 6.97, "learning_rate": 1.1607524823687561e-05, "loss": 0.0696, "step": 6873 }, { "epoch": 6.98, "learning_rate": 1.160536112904676e-05, "loss": 0.0715, "step": 6874 }, { "epoch": 6.98, "learning_rate": 1.1603197357258878e-05, "loss": 0.0715, "step": 6875 }, { "epoch": 6.98, "learning_rate": 1.1601033508427894e-05, "loss": 0.0669, "step": 6876 }, { "epoch": 6.98, "learning_rate": 1.1598869582657802e-05, "loss": 0.0644, "step": 6877 }, { "epoch": 6.98, "learning_rate": 1.1596705580052584e-05, "loss": 0.0763, "step": 6878 }, { "epoch": 6.98, "learning_rate": 1.1594541500716238e-05, "loss": 0.0926, "step": 6879 }, { "epoch": 6.98, "learning_rate": 1.1592377344752758e-05, "loss": 0.0678, "step": 6880 }, { "epoch": 6.98, "learning_rate": 1.1590213112266145e-05, "loss": 0.0772, "step": 6881 }, { "epoch": 6.98, "learning_rate": 1.15880488033604e-05, "loss": 0.0544, "step": 6882 }, { "epoch": 6.98, "learning_rate": 1.1585884418139539e-05, "loss": 0.0716, "step": 6883 }, { "epoch": 6.99, "learning_rate": 1.1583719956707569e-05, "loss": 0.0809, "step": 6884 }, { "epoch": 6.99, "learning_rate": 1.1581555419168499e-05, "loss": 0.0509, "step": 6885 }, { "epoch": 6.99, "learning_rate": 1.1579390805626359e-05, "loss": 0.0759, "step": 6886 }, { "epoch": 6.99, "learning_rate": 1.1577226116185162e-05, "loss": 0.0674, "step": 6887 }, { "epoch": 6.99, "learning_rate": 1.1575061350948942e-05, "loss": 0.067, "step": 6888 }, { "epoch": 6.99, "learning_rate": 1.157289651002172e-05, "loss": 0.0661, "step": 6889 }, { "epoch": 6.99, "learning_rate": 1.157073159350754e-05, "loss": 0.0639, "step": 6890 }, { "epoch": 6.99, "learning_rate": 1.1568566601510432e-05, "loss": 0.0728, "step": 6891 }, { "epoch": 6.99, "learning_rate": 1.1566401534134432e-05, "loss": 0.0709, "step": 6892 }, { "epoch": 6.99, "learning_rate": 1.1564236391483594e-05, "loss": 0.0552, "step": 6893 }, { "epoch": 7.0, "learning_rate": 1.1562071173661964e-05, "loss": 0.057, "step": 6894 }, { "epoch": 7.0, "learning_rate": 1.1559905880773591e-05, "loss": 0.0556, "step": 6895 }, { "epoch": 7.0, "learning_rate": 1.1557740512922533e-05, "loss": 0.0792, "step": 6896 }, { "epoch": 7.0, "learning_rate": 1.1555575070212843e-05, "loss": 0.0675, "step": 6897 }, { "epoch": 7.0, "learning_rate": 1.1553409552748587e-05, "loss": 0.061, "step": 6898 }, { "epoch": 7.0, "learning_rate": 1.155124396063383e-05, "loss": 0.0471, "step": 6899 }, { "epoch": 7.0, "learning_rate": 1.1549078293972641e-05, "loss": 0.0457, "step": 6900 }, { "epoch": 7.0, "learning_rate": 1.1546912552869098e-05, "loss": 0.0345, "step": 6901 }, { "epoch": 7.0, "learning_rate": 1.154474673742727e-05, "loss": 0.0421, "step": 6902 }, { "epoch": 7.0, "learning_rate": 1.1542580847751245e-05, "loss": 0.043, "step": 6903 }, { "epoch": 7.01, "learning_rate": 1.1540414883945099e-05, "loss": 0.0406, "step": 6904 }, { "epoch": 7.01, "learning_rate": 1.1538248846112928e-05, "loss": 0.038, "step": 6905 }, { "epoch": 7.01, "learning_rate": 1.153608273435881e-05, "loss": 0.0339, "step": 6906 }, { "epoch": 7.01, "learning_rate": 1.1533916548786856e-05, "loss": 0.0382, "step": 6907 }, { "epoch": 7.01, "learning_rate": 1.1531750289501154e-05, "loss": 0.0447, "step": 6908 }, { "epoch": 7.01, "learning_rate": 1.1529583956605803e-05, "loss": 0.045, "step": 6909 }, { "epoch": 7.01, "learning_rate": 1.1527417550204912e-05, "loss": 0.0274, "step": 6910 }, { "epoch": 7.01, "learning_rate": 1.1525251070402592e-05, "loss": 0.0503, "step": 6911 }, { "epoch": 7.01, "learning_rate": 1.1523084517302956e-05, "loss": 0.0498, "step": 6912 }, { "epoch": 7.01, "learning_rate": 1.1520917891010113e-05, "loss": 0.0357, "step": 6913 }, { "epoch": 7.02, "learning_rate": 1.1518751191628185e-05, "loss": 0.0429, "step": 6914 }, { "epoch": 7.02, "learning_rate": 1.1516584419261296e-05, "loss": 0.0333, "step": 6915 }, { "epoch": 7.02, "learning_rate": 1.1514417574013575e-05, "loss": 0.0449, "step": 6916 }, { "epoch": 7.02, "learning_rate": 1.1512250655989145e-05, "loss": 0.0435, "step": 6917 }, { "epoch": 7.02, "learning_rate": 1.1510083665292144e-05, "loss": 0.0498, "step": 6918 }, { "epoch": 7.02, "learning_rate": 1.1507916602026708e-05, "loss": 0.0268, "step": 6919 }, { "epoch": 7.02, "learning_rate": 1.1505749466296973e-05, "loss": 0.0386, "step": 6920 }, { "epoch": 7.02, "learning_rate": 1.150358225820709e-05, "loss": 0.0355, "step": 6921 }, { "epoch": 7.02, "learning_rate": 1.15014149778612e-05, "loss": 0.0526, "step": 6922 }, { "epoch": 7.02, "learning_rate": 1.1499247625363456e-05, "loss": 0.0525, "step": 6923 }, { "epoch": 7.03, "learning_rate": 1.1497080200818014e-05, "loss": 0.0442, "step": 6924 }, { "epoch": 7.03, "learning_rate": 1.149491270432903e-05, "loss": 0.0347, "step": 6925 }, { "epoch": 7.03, "learning_rate": 1.1492745136000661e-05, "loss": 0.0459, "step": 6926 }, { "epoch": 7.03, "learning_rate": 1.1490577495937076e-05, "loss": 0.0382, "step": 6927 }, { "epoch": 7.03, "learning_rate": 1.1488409784242442e-05, "loss": 0.0394, "step": 6928 }, { "epoch": 7.03, "learning_rate": 1.1486242001020932e-05, "loss": 0.0497, "step": 6929 }, { "epoch": 7.03, "learning_rate": 1.148407414637672e-05, "loss": 0.0431, "step": 6930 }, { "epoch": 7.03, "learning_rate": 1.148190622041398e-05, "loss": 0.0429, "step": 6931 }, { "epoch": 7.03, "learning_rate": 1.1479738223236895e-05, "loss": 0.0471, "step": 6932 }, { "epoch": 7.04, "learning_rate": 1.1477570154949658e-05, "loss": 0.0384, "step": 6933 }, { "epoch": 7.04, "learning_rate": 1.147540201565645e-05, "loss": 0.0542, "step": 6934 }, { "epoch": 7.04, "learning_rate": 1.147323380546146e-05, "loss": 0.0396, "step": 6935 }, { "epoch": 7.04, "learning_rate": 1.1471065524468891e-05, "loss": 0.044, "step": 6936 }, { "epoch": 7.04, "learning_rate": 1.1468897172782939e-05, "loss": 0.0345, "step": 6937 }, { "epoch": 7.04, "learning_rate": 1.1466728750507805e-05, "loss": 0.0489, "step": 6938 }, { "epoch": 7.04, "learning_rate": 1.1464560257747692e-05, "loss": 0.0406, "step": 6939 }, { "epoch": 7.04, "learning_rate": 1.1462391694606816e-05, "loss": 0.044, "step": 6940 }, { "epoch": 7.04, "learning_rate": 1.1460223061189384e-05, "loss": 0.0404, "step": 6941 }, { "epoch": 7.04, "learning_rate": 1.1458054357599608e-05, "loss": 0.0393, "step": 6942 }, { "epoch": 7.05, "learning_rate": 1.1455885583941717e-05, "loss": 0.0395, "step": 6943 }, { "epoch": 7.05, "learning_rate": 1.1453716740319927e-05, "loss": 0.0365, "step": 6944 }, { "epoch": 7.05, "learning_rate": 1.1451547826838464e-05, "loss": 0.0537, "step": 6945 }, { "epoch": 7.05, "learning_rate": 1.144937884360156e-05, "loss": 0.0341, "step": 6946 }, { "epoch": 7.05, "learning_rate": 1.1447209790713444e-05, "loss": 0.037, "step": 6947 }, { "epoch": 7.05, "learning_rate": 1.1445040668278354e-05, "loss": 0.0451, "step": 6948 }, { "epoch": 7.05, "learning_rate": 1.1442871476400523e-05, "loss": 0.0349, "step": 6949 }, { "epoch": 7.05, "learning_rate": 1.1440702215184205e-05, "loss": 0.0328, "step": 6950 }, { "epoch": 7.05, "learning_rate": 1.143853288473364e-05, "loss": 0.0463, "step": 6951 }, { "epoch": 7.05, "learning_rate": 1.1436363485153072e-05, "loss": 0.03, "step": 6952 }, { "epoch": 7.06, "learning_rate": 1.1434194016546763e-05, "loss": 0.0458, "step": 6953 }, { "epoch": 7.06, "learning_rate": 1.1432024479018959e-05, "loss": 0.0582, "step": 6954 }, { "epoch": 7.06, "learning_rate": 1.1429854872673927e-05, "loss": 0.0444, "step": 6955 }, { "epoch": 7.06, "learning_rate": 1.1427685197615925e-05, "loss": 0.0372, "step": 6956 }, { "epoch": 7.06, "learning_rate": 1.1425515453949223e-05, "loss": 0.0467, "step": 6957 }, { "epoch": 7.06, "learning_rate": 1.1423345641778085e-05, "loss": 0.0468, "step": 6958 }, { "epoch": 7.06, "learning_rate": 1.1421175761206787e-05, "loss": 0.0354, "step": 6959 }, { "epoch": 7.06, "learning_rate": 1.1419005812339598e-05, "loss": 0.0371, "step": 6960 }, { "epoch": 7.06, "learning_rate": 1.1416835795280808e-05, "loss": 0.0332, "step": 6961 }, { "epoch": 7.06, "learning_rate": 1.1414665710134691e-05, "loss": 0.0404, "step": 6962 }, { "epoch": 7.07, "learning_rate": 1.1412495557005537e-05, "loss": 0.0447, "step": 6963 }, { "epoch": 7.07, "learning_rate": 1.1410325335997628e-05, "loss": 0.0395, "step": 6964 }, { "epoch": 7.07, "learning_rate": 1.1408155047215259e-05, "loss": 0.0413, "step": 6965 }, { "epoch": 7.07, "learning_rate": 1.1405984690762728e-05, "loss": 0.0432, "step": 6966 }, { "epoch": 7.07, "learning_rate": 1.1403814266744333e-05, "loss": 0.0406, "step": 6967 }, { "epoch": 7.07, "learning_rate": 1.1401643775264375e-05, "loss": 0.0388, "step": 6968 }, { "epoch": 7.07, "learning_rate": 1.1399473216427155e-05, "loss": 0.05, "step": 6969 }, { "epoch": 7.07, "learning_rate": 1.1397302590336987e-05, "loss": 0.0467, "step": 6970 }, { "epoch": 7.07, "learning_rate": 1.1395131897098177e-05, "loss": 0.0416, "step": 6971 }, { "epoch": 7.07, "learning_rate": 1.1392961136815046e-05, "loss": 0.0403, "step": 6972 }, { "epoch": 7.08, "learning_rate": 1.1390790309591906e-05, "loss": 0.0318, "step": 6973 }, { "epoch": 7.08, "learning_rate": 1.1388619415533081e-05, "loss": 0.0362, "step": 6974 }, { "epoch": 7.08, "learning_rate": 1.1386448454742895e-05, "loss": 0.049, "step": 6975 }, { "epoch": 7.08, "learning_rate": 1.1384277427325675e-05, "loss": 0.0502, "step": 6976 }, { "epoch": 7.08, "learning_rate": 1.138210633338575e-05, "loss": 0.0366, "step": 6977 }, { "epoch": 7.08, "learning_rate": 1.1379935173027457e-05, "loss": 0.0367, "step": 6978 }, { "epoch": 7.08, "learning_rate": 1.1377763946355133e-05, "loss": 0.0513, "step": 6979 }, { "epoch": 7.08, "learning_rate": 1.1375592653473118e-05, "loss": 0.0467, "step": 6980 }, { "epoch": 7.08, "learning_rate": 1.1373421294485749e-05, "loss": 0.0458, "step": 6981 }, { "epoch": 7.08, "learning_rate": 1.137124986949738e-05, "loss": 0.0411, "step": 6982 }, { "epoch": 7.09, "learning_rate": 1.1369078378612361e-05, "loss": 0.0434, "step": 6983 }, { "epoch": 7.09, "learning_rate": 1.1366906821935044e-05, "loss": 0.0419, "step": 6984 }, { "epoch": 7.09, "learning_rate": 1.136473519956978e-05, "loss": 0.0449, "step": 6985 }, { "epoch": 7.09, "learning_rate": 1.1362563511620937e-05, "loss": 0.0509, "step": 6986 }, { "epoch": 7.09, "learning_rate": 1.1360391758192868e-05, "loss": 0.0335, "step": 6987 }, { "epoch": 7.09, "learning_rate": 1.1358219939389945e-05, "loss": 0.037, "step": 6988 }, { "epoch": 7.09, "learning_rate": 1.1356048055316538e-05, "loss": 0.045, "step": 6989 }, { "epoch": 7.09, "learning_rate": 1.1353876106077012e-05, "loss": 0.0535, "step": 6990 }, { "epoch": 7.09, "learning_rate": 1.1351704091775746e-05, "loss": 0.0381, "step": 6991 }, { "epoch": 7.09, "learning_rate": 1.1349532012517117e-05, "loss": 0.0543, "step": 6992 }, { "epoch": 7.1, "learning_rate": 1.134735986840551e-05, "loss": 0.0332, "step": 6993 }, { "epoch": 7.1, "learning_rate": 1.1345187659545305e-05, "loss": 0.043, "step": 6994 }, { "epoch": 7.1, "learning_rate": 1.1343015386040892e-05, "loss": 0.0451, "step": 6995 }, { "epoch": 7.1, "learning_rate": 1.1340843047996659e-05, "loss": 0.0499, "step": 6996 }, { "epoch": 7.1, "learning_rate": 1.1338670645517002e-05, "loss": 0.0404, "step": 6997 }, { "epoch": 7.1, "learning_rate": 1.1336498178706317e-05, "loss": 0.036, "step": 6998 }, { "epoch": 7.1, "learning_rate": 1.1334325647669002e-05, "loss": 0.0433, "step": 6999 }, { "epoch": 7.1, "learning_rate": 1.1332153052509464e-05, "loss": 0.0386, "step": 7000 }, { "epoch": 7.1, "learning_rate": 1.1329980393332108e-05, "loss": 0.0394, "step": 7001 }, { "epoch": 7.11, "learning_rate": 1.1327807670241338e-05, "loss": 0.0459, "step": 7002 }, { "epoch": 7.11, "learning_rate": 1.1325634883341572e-05, "loss": 0.0455, "step": 7003 }, { "epoch": 7.11, "learning_rate": 1.1323462032737222e-05, "loss": 0.057, "step": 7004 }, { "epoch": 7.11, "learning_rate": 1.132128911853271e-05, "loss": 0.0366, "step": 7005 }, { "epoch": 7.11, "learning_rate": 1.1319116140832452e-05, "loss": 0.0393, "step": 7006 }, { "epoch": 7.11, "learning_rate": 1.1316943099740878e-05, "loss": 0.0323, "step": 7007 }, { "epoch": 7.11, "learning_rate": 1.1314769995362412e-05, "loss": 0.0406, "step": 7008 }, { "epoch": 7.11, "learning_rate": 1.1312596827801482e-05, "loss": 0.0365, "step": 7009 }, { "epoch": 7.11, "learning_rate": 1.1310423597162528e-05, "loss": 0.0448, "step": 7010 }, { "epoch": 7.11, "learning_rate": 1.1308250303549984e-05, "loss": 0.0497, "step": 7011 }, { "epoch": 7.12, "learning_rate": 1.130607694706829e-05, "loss": 0.0417, "step": 7012 }, { "epoch": 7.12, "learning_rate": 1.1303903527821886e-05, "loss": 0.0361, "step": 7013 }, { "epoch": 7.12, "learning_rate": 1.1301730045915218e-05, "loss": 0.0476, "step": 7014 }, { "epoch": 7.12, "learning_rate": 1.1299556501452735e-05, "loss": 0.0395, "step": 7015 }, { "epoch": 7.12, "learning_rate": 1.1297382894538894e-05, "loss": 0.0403, "step": 7016 }, { "epoch": 7.12, "learning_rate": 1.129520922527814e-05, "loss": 0.0385, "step": 7017 }, { "epoch": 7.12, "learning_rate": 1.129303549377494e-05, "loss": 0.0431, "step": 7018 }, { "epoch": 7.12, "learning_rate": 1.1290861700133748e-05, "loss": 0.0401, "step": 7019 }, { "epoch": 7.12, "learning_rate": 1.1288687844459031e-05, "loss": 0.052, "step": 7020 }, { "epoch": 7.12, "learning_rate": 1.1286513926855254e-05, "loss": 0.0332, "step": 7021 }, { "epoch": 7.13, "learning_rate": 1.1284339947426888e-05, "loss": 0.0415, "step": 7022 }, { "epoch": 7.13, "learning_rate": 1.1282165906278402e-05, "loss": 0.0429, "step": 7023 }, { "epoch": 7.13, "learning_rate": 1.1279991803514279e-05, "loss": 0.0433, "step": 7024 }, { "epoch": 7.13, "learning_rate": 1.1277817639238991e-05, "loss": 0.038, "step": 7025 }, { "epoch": 7.13, "learning_rate": 1.1275643413557018e-05, "loss": 0.0396, "step": 7026 }, { "epoch": 7.13, "learning_rate": 1.127346912657285e-05, "loss": 0.0339, "step": 7027 }, { "epoch": 7.13, "learning_rate": 1.1271294778390971e-05, "loss": 0.047, "step": 7028 }, { "epoch": 7.13, "learning_rate": 1.1269120369115874e-05, "loss": 0.0347, "step": 7029 }, { "epoch": 7.13, "learning_rate": 1.1266945898852048e-05, "loss": 0.0395, "step": 7030 }, { "epoch": 7.13, "learning_rate": 1.1264771367703992e-05, "loss": 0.038, "step": 7031 }, { "epoch": 7.14, "learning_rate": 1.1262596775776204e-05, "loss": 0.0546, "step": 7032 }, { "epoch": 7.14, "learning_rate": 1.126042212317319e-05, "loss": 0.0339, "step": 7033 }, { "epoch": 7.14, "learning_rate": 1.125824740999945e-05, "loss": 0.0372, "step": 7034 }, { "epoch": 7.14, "learning_rate": 1.1256072636359493e-05, "loss": 0.0433, "step": 7035 }, { "epoch": 7.14, "learning_rate": 1.1253897802357827e-05, "loss": 0.0303, "step": 7036 }, { "epoch": 7.14, "learning_rate": 1.125172290809897e-05, "loss": 0.0426, "step": 7037 }, { "epoch": 7.14, "learning_rate": 1.1249547953687438e-05, "loss": 0.0458, "step": 7038 }, { "epoch": 7.14, "learning_rate": 1.1247372939227752e-05, "loss": 0.0471, "step": 7039 }, { "epoch": 7.14, "learning_rate": 1.1245197864824426e-05, "loss": 0.041, "step": 7040 }, { "epoch": 7.14, "learning_rate": 1.1243022730581996e-05, "loss": 0.0465, "step": 7041 }, { "epoch": 7.15, "learning_rate": 1.1240847536604984e-05, "loss": 0.0413, "step": 7042 }, { "epoch": 7.15, "learning_rate": 1.1238672282997917e-05, "loss": 0.0529, "step": 7043 }, { "epoch": 7.15, "learning_rate": 1.1236496969865338e-05, "loss": 0.0409, "step": 7044 }, { "epoch": 7.15, "learning_rate": 1.1234321597311779e-05, "loss": 0.0303, "step": 7045 }, { "epoch": 7.15, "learning_rate": 1.1232146165441782e-05, "loss": 0.0421, "step": 7046 }, { "epoch": 7.15, "learning_rate": 1.1229970674359884e-05, "loss": 0.0488, "step": 7047 }, { "epoch": 7.15, "learning_rate": 1.1227795124170633e-05, "loss": 0.0418, "step": 7048 }, { "epoch": 7.15, "learning_rate": 1.1225619514978576e-05, "loss": 0.0479, "step": 7049 }, { "epoch": 7.15, "learning_rate": 1.1223443846888268e-05, "loss": 0.0414, "step": 7050 }, { "epoch": 7.15, "learning_rate": 1.122126812000426e-05, "loss": 0.0462, "step": 7051 }, { "epoch": 7.16, "learning_rate": 1.1219092334431107e-05, "loss": 0.041, "step": 7052 }, { "epoch": 7.16, "learning_rate": 1.1216916490273373e-05, "loss": 0.0512, "step": 7053 }, { "epoch": 7.16, "learning_rate": 1.1214740587635613e-05, "loss": 0.0425, "step": 7054 }, { "epoch": 7.16, "learning_rate": 1.1212564626622398e-05, "loss": 0.0358, "step": 7055 }, { "epoch": 7.16, "learning_rate": 1.1210388607338292e-05, "loss": 0.0466, "step": 7056 }, { "epoch": 7.16, "learning_rate": 1.1208212529887869e-05, "loss": 0.0403, "step": 7057 }, { "epoch": 7.16, "learning_rate": 1.1206036394375702e-05, "loss": 0.0402, "step": 7058 }, { "epoch": 7.16, "learning_rate": 1.1203860200906361e-05, "loss": 0.038, "step": 7059 }, { "epoch": 7.16, "learning_rate": 1.1201683949584434e-05, "loss": 0.0398, "step": 7060 }, { "epoch": 7.16, "learning_rate": 1.1199507640514496e-05, "loss": 0.0565, "step": 7061 }, { "epoch": 7.17, "learning_rate": 1.1197331273801135e-05, "loss": 0.0451, "step": 7062 }, { "epoch": 7.17, "learning_rate": 1.1195154849548937e-05, "loss": 0.037, "step": 7063 }, { "epoch": 7.17, "learning_rate": 1.1192978367862493e-05, "loss": 0.0459, "step": 7064 }, { "epoch": 7.17, "learning_rate": 1.1190801828846392e-05, "loss": 0.0468, "step": 7065 }, { "epoch": 7.17, "learning_rate": 1.1188625232605235e-05, "loss": 0.0497, "step": 7066 }, { "epoch": 7.17, "learning_rate": 1.1186448579243618e-05, "loss": 0.0458, "step": 7067 }, { "epoch": 7.17, "learning_rate": 1.1184271868866143e-05, "loss": 0.0337, "step": 7068 }, { "epoch": 7.17, "learning_rate": 1.118209510157741e-05, "loss": 0.0429, "step": 7069 }, { "epoch": 7.17, "learning_rate": 1.117991827748203e-05, "loss": 0.0421, "step": 7070 }, { "epoch": 7.18, "learning_rate": 1.117774139668461e-05, "loss": 0.0287, "step": 7071 }, { "epoch": 7.18, "learning_rate": 1.1175564459289763e-05, "loss": 0.0427, "step": 7072 }, { "epoch": 7.18, "learning_rate": 1.1173387465402102e-05, "loss": 0.048, "step": 7073 }, { "epoch": 7.18, "learning_rate": 1.1171210415126248e-05, "loss": 0.0419, "step": 7074 }, { "epoch": 7.18, "learning_rate": 1.1169033308566816e-05, "loss": 0.05, "step": 7075 }, { "epoch": 7.18, "learning_rate": 1.1166856145828432e-05, "loss": 0.0464, "step": 7076 }, { "epoch": 7.18, "learning_rate": 1.1164678927015722e-05, "loss": 0.038, "step": 7077 }, { "epoch": 7.18, "learning_rate": 1.1162501652233313e-05, "loss": 0.0389, "step": 7078 }, { "epoch": 7.18, "learning_rate": 1.1160324321585836e-05, "loss": 0.0376, "step": 7079 }, { "epoch": 7.18, "learning_rate": 1.1158146935177927e-05, "loss": 0.047, "step": 7080 }, { "epoch": 7.19, "learning_rate": 1.1155969493114216e-05, "loss": 0.0448, "step": 7081 }, { "epoch": 7.19, "learning_rate": 1.1153791995499346e-05, "loss": 0.0442, "step": 7082 }, { "epoch": 7.19, "learning_rate": 1.1151614442437962e-05, "loss": 0.0395, "step": 7083 }, { "epoch": 7.19, "learning_rate": 1.1149436834034704e-05, "loss": 0.0405, "step": 7084 }, { "epoch": 7.19, "learning_rate": 1.114725917039422e-05, "loss": 0.0416, "step": 7085 }, { "epoch": 7.19, "learning_rate": 1.1145081451621156e-05, "loss": 0.0427, "step": 7086 }, { "epoch": 7.19, "learning_rate": 1.1142903677820169e-05, "loss": 0.0432, "step": 7087 }, { "epoch": 7.19, "learning_rate": 1.1140725849095915e-05, "loss": 0.0396, "step": 7088 }, { "epoch": 7.19, "learning_rate": 1.1138547965553049e-05, "loss": 0.0447, "step": 7089 }, { "epoch": 7.19, "learning_rate": 1.1136370027296228e-05, "loss": 0.0393, "step": 7090 }, { "epoch": 7.2, "learning_rate": 1.1134192034430123e-05, "loss": 0.0372, "step": 7091 }, { "epoch": 7.2, "learning_rate": 1.1132013987059391e-05, "loss": 0.0422, "step": 7092 }, { "epoch": 7.2, "learning_rate": 1.1129835885288703e-05, "loss": 0.0417, "step": 7093 }, { "epoch": 7.2, "learning_rate": 1.1127657729222729e-05, "loss": 0.0497, "step": 7094 }, { "epoch": 7.2, "learning_rate": 1.1125479518966146e-05, "loss": 0.0486, "step": 7095 }, { "epoch": 7.2, "learning_rate": 1.1123301254623628e-05, "loss": 0.037, "step": 7096 }, { "epoch": 7.2, "learning_rate": 1.1121122936299853e-05, "loss": 0.0525, "step": 7097 }, { "epoch": 7.2, "learning_rate": 1.1118944564099499e-05, "loss": 0.0418, "step": 7098 }, { "epoch": 7.2, "learning_rate": 1.111676613812725e-05, "loss": 0.042, "step": 7099 }, { "epoch": 7.2, "learning_rate": 1.11145876584878e-05, "loss": 0.0396, "step": 7100 }, { "epoch": 7.21, "learning_rate": 1.1112409125285833e-05, "loss": 0.0462, "step": 7101 }, { "epoch": 7.21, "learning_rate": 1.1110230538626035e-05, "loss": 0.0347, "step": 7102 }, { "epoch": 7.21, "learning_rate": 1.1108051898613107e-05, "loss": 0.0405, "step": 7103 }, { "epoch": 7.21, "learning_rate": 1.1105873205351745e-05, "loss": 0.0422, "step": 7104 }, { "epoch": 7.21, "learning_rate": 1.1103694458946646e-05, "loss": 0.036, "step": 7105 }, { "epoch": 7.21, "learning_rate": 1.110151565950251e-05, "loss": 0.0372, "step": 7106 }, { "epoch": 7.21, "learning_rate": 1.1099336807124044e-05, "loss": 0.0455, "step": 7107 }, { "epoch": 7.21, "learning_rate": 1.1097157901915954e-05, "loss": 0.0527, "step": 7108 }, { "epoch": 7.21, "learning_rate": 1.1094978943982949e-05, "loss": 0.0506, "step": 7109 }, { "epoch": 7.21, "learning_rate": 1.1092799933429741e-05, "loss": 0.0436, "step": 7110 }, { "epoch": 7.22, "learning_rate": 1.1090620870361044e-05, "loss": 0.0315, "step": 7111 }, { "epoch": 7.22, "learning_rate": 1.1088441754881576e-05, "loss": 0.0543, "step": 7112 }, { "epoch": 7.22, "learning_rate": 1.1086262587096054e-05, "loss": 0.0432, "step": 7113 }, { "epoch": 7.22, "learning_rate": 1.1084083367109204e-05, "loss": 0.0455, "step": 7114 }, { "epoch": 7.22, "learning_rate": 1.1081904095025744e-05, "loss": 0.0493, "step": 7115 }, { "epoch": 7.22, "learning_rate": 1.1079724770950406e-05, "loss": 0.0433, "step": 7116 }, { "epoch": 7.22, "learning_rate": 1.1077545394987921e-05, "loss": 0.0533, "step": 7117 }, { "epoch": 7.22, "learning_rate": 1.1075365967243018e-05, "loss": 0.041, "step": 7118 }, { "epoch": 7.22, "learning_rate": 1.1073186487820426e-05, "loss": 0.0423, "step": 7119 }, { "epoch": 7.22, "learning_rate": 1.107100695682489e-05, "loss": 0.0378, "step": 7120 }, { "epoch": 7.23, "learning_rate": 1.1068827374361146e-05, "loss": 0.0469, "step": 7121 }, { "epoch": 7.23, "learning_rate": 1.1066647740533938e-05, "loss": 0.0486, "step": 7122 }, { "epoch": 7.23, "learning_rate": 1.1064468055448006e-05, "loss": 0.0513, "step": 7123 }, { "epoch": 7.23, "learning_rate": 1.1062288319208101e-05, "loss": 0.0477, "step": 7124 }, { "epoch": 7.23, "learning_rate": 1.1060108531918972e-05, "loss": 0.0463, "step": 7125 }, { "epoch": 7.23, "learning_rate": 1.1057928693685367e-05, "loss": 0.0463, "step": 7126 }, { "epoch": 7.23, "learning_rate": 1.1055748804612041e-05, "loss": 0.0343, "step": 7127 }, { "epoch": 7.23, "learning_rate": 1.1053568864803754e-05, "loss": 0.0397, "step": 7128 }, { "epoch": 7.23, "learning_rate": 1.1051388874365263e-05, "loss": 0.0521, "step": 7129 }, { "epoch": 7.23, "learning_rate": 1.1049208833401329e-05, "loss": 0.042, "step": 7130 }, { "epoch": 7.24, "learning_rate": 1.1047028742016713e-05, "loss": 0.0397, "step": 7131 }, { "epoch": 7.24, "learning_rate": 1.1044848600316184e-05, "loss": 0.0347, "step": 7132 }, { "epoch": 7.24, "learning_rate": 1.1042668408404516e-05, "loss": 0.0438, "step": 7133 }, { "epoch": 7.24, "learning_rate": 1.1040488166386474e-05, "loss": 0.0435, "step": 7134 }, { "epoch": 7.24, "learning_rate": 1.103830787436683e-05, "loss": 0.0329, "step": 7135 }, { "epoch": 7.24, "learning_rate": 1.1036127532450362e-05, "loss": 0.0432, "step": 7136 }, { "epoch": 7.24, "learning_rate": 1.1033947140741849e-05, "loss": 0.04, "step": 7137 }, { "epoch": 7.24, "learning_rate": 1.1031766699346074e-05, "loss": 0.0526, "step": 7138 }, { "epoch": 7.24, "learning_rate": 1.1029586208367817e-05, "loss": 0.0579, "step": 7139 }, { "epoch": 7.25, "learning_rate": 1.1027405667911861e-05, "loss": 0.0415, "step": 7140 }, { "epoch": 7.25, "learning_rate": 1.1025225078083e-05, "loss": 0.0475, "step": 7141 }, { "epoch": 7.25, "learning_rate": 1.102304443898602e-05, "loss": 0.0461, "step": 7142 }, { "epoch": 7.25, "learning_rate": 1.1020863750725715e-05, "loss": 0.0401, "step": 7143 }, { "epoch": 7.25, "learning_rate": 1.1018683013406877e-05, "loss": 0.0448, "step": 7144 }, { "epoch": 7.25, "learning_rate": 1.101650222713431e-05, "loss": 0.0401, "step": 7145 }, { "epoch": 7.25, "learning_rate": 1.1014321392012807e-05, "loss": 0.0547, "step": 7146 }, { "epoch": 7.25, "learning_rate": 1.1012140508147173e-05, "loss": 0.055, "step": 7147 }, { "epoch": 7.25, "learning_rate": 1.1009959575642211e-05, "loss": 0.0428, "step": 7148 }, { "epoch": 7.25, "learning_rate": 1.1007778594602729e-05, "loss": 0.0448, "step": 7149 }, { "epoch": 7.26, "learning_rate": 1.1005597565133537e-05, "loss": 0.0409, "step": 7150 }, { "epoch": 7.26, "learning_rate": 1.1003416487339445e-05, "loss": 0.0404, "step": 7151 }, { "epoch": 7.26, "learning_rate": 1.1001235361325265e-05, "loss": 0.0384, "step": 7152 }, { "epoch": 7.26, "learning_rate": 1.0999054187195815e-05, "loss": 0.0339, "step": 7153 }, { "epoch": 7.26, "learning_rate": 1.099687296505591e-05, "loss": 0.0445, "step": 7154 }, { "epoch": 7.26, "learning_rate": 1.0994691695010378e-05, "loss": 0.0451, "step": 7155 }, { "epoch": 7.26, "learning_rate": 1.0992510377164036e-05, "loss": 0.0485, "step": 7156 }, { "epoch": 7.26, "learning_rate": 1.099032901162171e-05, "loss": 0.0457, "step": 7157 }, { "epoch": 7.26, "learning_rate": 1.098814759848823e-05, "loss": 0.0413, "step": 7158 }, { "epoch": 7.26, "learning_rate": 1.098596613786842e-05, "loss": 0.0418, "step": 7159 }, { "epoch": 7.27, "learning_rate": 1.0983784629867121e-05, "loss": 0.0439, "step": 7160 }, { "epoch": 7.27, "learning_rate": 1.0981603074589155e-05, "loss": 0.0509, "step": 7161 }, { "epoch": 7.27, "learning_rate": 1.0979421472139373e-05, "loss": 0.052, "step": 7162 }, { "epoch": 7.27, "learning_rate": 1.0977239822622603e-05, "loss": 0.0581, "step": 7163 }, { "epoch": 7.27, "learning_rate": 1.0975058126143692e-05, "loss": 0.0534, "step": 7164 }, { "epoch": 7.27, "learning_rate": 1.097287638280748e-05, "loss": 0.0383, "step": 7165 }, { "epoch": 7.27, "learning_rate": 1.0970694592718811e-05, "loss": 0.0429, "step": 7166 }, { "epoch": 7.27, "learning_rate": 1.0968512755982541e-05, "loss": 0.0536, "step": 7167 }, { "epoch": 7.27, "learning_rate": 1.0966330872703513e-05, "loss": 0.0427, "step": 7168 }, { "epoch": 7.27, "learning_rate": 1.0964148942986581e-05, "loss": 0.036, "step": 7169 }, { "epoch": 7.28, "learning_rate": 1.09619669669366e-05, "loss": 0.0425, "step": 7170 }, { "epoch": 7.28, "learning_rate": 1.0959784944658425e-05, "loss": 0.0373, "step": 7171 }, { "epoch": 7.28, "learning_rate": 1.0957602876256921e-05, "loss": 0.0405, "step": 7172 }, { "epoch": 7.28, "learning_rate": 1.0955420761836938e-05, "loss": 0.0465, "step": 7173 }, { "epoch": 7.28, "learning_rate": 1.0953238601503354e-05, "loss": 0.0451, "step": 7174 }, { "epoch": 7.28, "learning_rate": 1.0951056395361026e-05, "loss": 0.0362, "step": 7175 }, { "epoch": 7.28, "learning_rate": 1.094887414351482e-05, "loss": 0.0358, "step": 7176 }, { "epoch": 7.28, "learning_rate": 1.094669184606961e-05, "loss": 0.0523, "step": 7177 }, { "epoch": 7.28, "learning_rate": 1.0944509503130267e-05, "loss": 0.0371, "step": 7178 }, { "epoch": 7.28, "learning_rate": 1.094232711480167e-05, "loss": 0.0496, "step": 7179 }, { "epoch": 7.29, "learning_rate": 1.094014468118869e-05, "loss": 0.0378, "step": 7180 }, { "epoch": 7.29, "learning_rate": 1.0937962202396204e-05, "loss": 0.0487, "step": 7181 }, { "epoch": 7.29, "learning_rate": 1.0935779678529097e-05, "loss": 0.0451, "step": 7182 }, { "epoch": 7.29, "learning_rate": 1.0933597109692256e-05, "loss": 0.0352, "step": 7183 }, { "epoch": 7.29, "learning_rate": 1.0931414495990557e-05, "loss": 0.045, "step": 7184 }, { "epoch": 7.29, "learning_rate": 1.0929231837528897e-05, "loss": 0.0381, "step": 7185 }, { "epoch": 7.29, "learning_rate": 1.0927049134412158e-05, "loss": 0.0377, "step": 7186 }, { "epoch": 7.29, "learning_rate": 1.0924866386745237e-05, "loss": 0.035, "step": 7187 }, { "epoch": 7.29, "learning_rate": 1.092268359463302e-05, "loss": 0.0498, "step": 7188 }, { "epoch": 7.29, "learning_rate": 1.0920500758180414e-05, "loss": 0.0456, "step": 7189 }, { "epoch": 7.3, "learning_rate": 1.091831787749231e-05, "loss": 0.0544, "step": 7190 }, { "epoch": 7.3, "learning_rate": 1.0916134952673612e-05, "loss": 0.0402, "step": 7191 }, { "epoch": 7.3, "learning_rate": 1.091395198382922e-05, "loss": 0.0455, "step": 7192 }, { "epoch": 7.3, "learning_rate": 1.0911768971064036e-05, "loss": 0.0389, "step": 7193 }, { "epoch": 7.3, "learning_rate": 1.0909585914482972e-05, "loss": 0.0634, "step": 7194 }, { "epoch": 7.3, "learning_rate": 1.0907402814190938e-05, "loss": 0.0398, "step": 7195 }, { "epoch": 7.3, "learning_rate": 1.0905219670292837e-05, "loss": 0.0463, "step": 7196 }, { "epoch": 7.3, "learning_rate": 1.090303648289359e-05, "loss": 0.0363, "step": 7197 }, { "epoch": 7.3, "learning_rate": 1.0900853252098105e-05, "loss": 0.04, "step": 7198 }, { "epoch": 7.3, "learning_rate": 1.0898669978011303e-05, "loss": 0.0498, "step": 7199 }, { "epoch": 7.31, "learning_rate": 1.0896486660738103e-05, "loss": 0.0412, "step": 7200 }, { "epoch": 7.31, "learning_rate": 1.0894303300383428e-05, "loss": 0.0398, "step": 7201 }, { "epoch": 7.31, "learning_rate": 1.0892119897052198e-05, "loss": 0.0413, "step": 7202 }, { "epoch": 7.31, "learning_rate": 1.0889936450849338e-05, "loss": 0.044, "step": 7203 }, { "epoch": 7.31, "learning_rate": 1.0887752961879781e-05, "loss": 0.0446, "step": 7204 }, { "epoch": 7.31, "learning_rate": 1.088556943024845e-05, "loss": 0.0473, "step": 7205 }, { "epoch": 7.31, "learning_rate": 1.0883385856060283e-05, "loss": 0.0496, "step": 7206 }, { "epoch": 7.31, "learning_rate": 1.0881202239420206e-05, "loss": 0.0497, "step": 7207 }, { "epoch": 7.31, "learning_rate": 1.087901858043316e-05, "loss": 0.0502, "step": 7208 }, { "epoch": 7.32, "learning_rate": 1.087683487920408e-05, "loss": 0.0409, "step": 7209 }, { "epoch": 7.32, "learning_rate": 1.0874651135837912e-05, "loss": 0.0471, "step": 7210 }, { "epoch": 7.32, "learning_rate": 1.0872467350439588e-05, "loss": 0.0388, "step": 7211 }, { "epoch": 7.32, "learning_rate": 1.087028352311406e-05, "loss": 0.038, "step": 7212 }, { "epoch": 7.32, "learning_rate": 1.086809965396627e-05, "loss": 0.0416, "step": 7213 }, { "epoch": 7.32, "learning_rate": 1.0865915743101168e-05, "loss": 0.0442, "step": 7214 }, { "epoch": 7.32, "learning_rate": 1.0863731790623698e-05, "loss": 0.0317, "step": 7215 }, { "epoch": 7.32, "learning_rate": 1.0861547796638816e-05, "loss": 0.0489, "step": 7216 }, { "epoch": 7.32, "learning_rate": 1.0859363761251478e-05, "loss": 0.0446, "step": 7217 }, { "epoch": 7.32, "learning_rate": 1.0857179684566638e-05, "loss": 0.0437, "step": 7218 }, { "epoch": 7.33, "learning_rate": 1.085499556668925e-05, "loss": 0.0415, "step": 7219 }, { "epoch": 7.33, "learning_rate": 1.0852811407724282e-05, "loss": 0.0418, "step": 7220 }, { "epoch": 7.33, "learning_rate": 1.0850627207776687e-05, "loss": 0.0447, "step": 7221 }, { "epoch": 7.33, "learning_rate": 1.0848442966951433e-05, "loss": 0.048, "step": 7222 }, { "epoch": 7.33, "learning_rate": 1.0846258685353489e-05, "loss": 0.0409, "step": 7223 }, { "epoch": 7.33, "learning_rate": 1.0844074363087814e-05, "loss": 0.0487, "step": 7224 }, { "epoch": 7.33, "learning_rate": 1.0841890000259385e-05, "loss": 0.0415, "step": 7225 }, { "epoch": 7.33, "learning_rate": 1.0839705596973169e-05, "loss": 0.0396, "step": 7226 }, { "epoch": 7.33, "learning_rate": 1.0837521153334143e-05, "loss": 0.0444, "step": 7227 }, { "epoch": 7.33, "learning_rate": 1.083533666944728e-05, "loss": 0.0417, "step": 7228 }, { "epoch": 7.34, "learning_rate": 1.083315214541756e-05, "loss": 0.0389, "step": 7229 }, { "epoch": 7.34, "learning_rate": 1.083096758134996e-05, "loss": 0.0488, "step": 7230 }, { "epoch": 7.34, "learning_rate": 1.0828782977349459e-05, "loss": 0.0399, "step": 7231 }, { "epoch": 7.34, "learning_rate": 1.0826598333521043e-05, "loss": 0.0395, "step": 7232 }, { "epoch": 7.34, "learning_rate": 1.08244136499697e-05, "loss": 0.0422, "step": 7233 }, { "epoch": 7.34, "learning_rate": 1.0822228926800413e-05, "loss": 0.0414, "step": 7234 }, { "epoch": 7.34, "learning_rate": 1.0820044164118172e-05, "loss": 0.0428, "step": 7235 }, { "epoch": 7.34, "learning_rate": 1.0817859362027965e-05, "loss": 0.061, "step": 7236 }, { "epoch": 7.34, "learning_rate": 1.0815674520634789e-05, "loss": 0.0341, "step": 7237 }, { "epoch": 7.34, "learning_rate": 1.0813489640043634e-05, "loss": 0.0433, "step": 7238 }, { "epoch": 7.35, "learning_rate": 1.08113047203595e-05, "loss": 0.0535, "step": 7239 }, { "epoch": 7.35, "learning_rate": 1.0809119761687386e-05, "loss": 0.0471, "step": 7240 }, { "epoch": 7.35, "learning_rate": 1.0806934764132289e-05, "loss": 0.0416, "step": 7241 }, { "epoch": 7.35, "learning_rate": 1.0804749727799213e-05, "loss": 0.0445, "step": 7242 }, { "epoch": 7.35, "learning_rate": 1.080256465279316e-05, "loss": 0.0448, "step": 7243 }, { "epoch": 7.35, "learning_rate": 1.0800379539219136e-05, "loss": 0.0517, "step": 7244 }, { "epoch": 7.35, "learning_rate": 1.0798194387182153e-05, "loss": 0.0476, "step": 7245 }, { "epoch": 7.35, "learning_rate": 1.0796009196787217e-05, "loss": 0.0628, "step": 7246 }, { "epoch": 7.35, "learning_rate": 1.0793823968139339e-05, "loss": 0.0464, "step": 7247 }, { "epoch": 7.35, "learning_rate": 1.079163870134353e-05, "loss": 0.053, "step": 7248 }, { "epoch": 7.36, "learning_rate": 1.0789453396504807e-05, "loss": 0.0382, "step": 7249 }, { "epoch": 7.36, "learning_rate": 1.0787268053728193e-05, "loss": 0.0403, "step": 7250 }, { "epoch": 7.36, "learning_rate": 1.0785082673118696e-05, "loss": 0.0376, "step": 7251 }, { "epoch": 7.36, "learning_rate": 1.0782897254781346e-05, "loss": 0.0448, "step": 7252 }, { "epoch": 7.36, "learning_rate": 1.0780711798821155e-05, "loss": 0.0444, "step": 7253 }, { "epoch": 7.36, "learning_rate": 1.0778526305343156e-05, "loss": 0.041, "step": 7254 }, { "epoch": 7.36, "learning_rate": 1.077634077445237e-05, "loss": 0.0412, "step": 7255 }, { "epoch": 7.36, "learning_rate": 1.0774155206253828e-05, "loss": 0.0438, "step": 7256 }, { "epoch": 7.36, "learning_rate": 1.0771969600852557e-05, "loss": 0.0452, "step": 7257 }, { "epoch": 7.36, "learning_rate": 1.0769783958353588e-05, "loss": 0.0447, "step": 7258 }, { "epoch": 7.37, "learning_rate": 1.0767598278861957e-05, "loss": 0.0532, "step": 7259 }, { "epoch": 7.37, "learning_rate": 1.0765412562482696e-05, "loss": 0.0501, "step": 7260 }, { "epoch": 7.37, "learning_rate": 1.076322680932084e-05, "loss": 0.0323, "step": 7261 }, { "epoch": 7.37, "learning_rate": 1.0761041019481432e-05, "loss": 0.0524, "step": 7262 }, { "epoch": 7.37, "learning_rate": 1.075885519306951e-05, "loss": 0.0448, "step": 7263 }, { "epoch": 7.37, "learning_rate": 1.0756669330190118e-05, "loss": 0.0423, "step": 7264 }, { "epoch": 7.37, "learning_rate": 1.0754483430948292e-05, "loss": 0.0481, "step": 7265 }, { "epoch": 7.37, "learning_rate": 1.0752297495449084e-05, "loss": 0.0561, "step": 7266 }, { "epoch": 7.37, "learning_rate": 1.0750111523797542e-05, "loss": 0.053, "step": 7267 }, { "epoch": 7.37, "learning_rate": 1.0747925516098712e-05, "loss": 0.048, "step": 7268 }, { "epoch": 7.38, "learning_rate": 1.0745739472457644e-05, "loss": 0.049, "step": 7269 }, { "epoch": 7.38, "learning_rate": 1.074355339297939e-05, "loss": 0.0571, "step": 7270 }, { "epoch": 7.38, "learning_rate": 1.0741367277769008e-05, "loss": 0.0401, "step": 7271 }, { "epoch": 7.38, "learning_rate": 1.0739181126931553e-05, "loss": 0.0315, "step": 7272 }, { "epoch": 7.38, "learning_rate": 1.0736994940572079e-05, "loss": 0.0349, "step": 7273 }, { "epoch": 7.38, "learning_rate": 1.0734808718795643e-05, "loss": 0.0517, "step": 7274 }, { "epoch": 7.38, "learning_rate": 1.0732622461707315e-05, "loss": 0.0382, "step": 7275 }, { "epoch": 7.38, "learning_rate": 1.0730436169412148e-05, "loss": 0.0387, "step": 7276 }, { "epoch": 7.38, "learning_rate": 1.0728249842015215e-05, "loss": 0.0503, "step": 7277 }, { "epoch": 7.39, "learning_rate": 1.0726063479621574e-05, "loss": 0.0494, "step": 7278 }, { "epoch": 7.39, "learning_rate": 1.0723877082336298e-05, "loss": 0.0411, "step": 7279 }, { "epoch": 7.39, "learning_rate": 1.0721690650264456e-05, "loss": 0.0451, "step": 7280 }, { "epoch": 7.39, "learning_rate": 1.0719504183511113e-05, "loss": 0.0514, "step": 7281 }, { "epoch": 7.39, "learning_rate": 1.0717317682181347e-05, "loss": 0.0486, "step": 7282 }, { "epoch": 7.39, "learning_rate": 1.0715131146380236e-05, "loss": 0.0397, "step": 7283 }, { "epoch": 7.39, "learning_rate": 1.0712944576212847e-05, "loss": 0.0412, "step": 7284 }, { "epoch": 7.39, "learning_rate": 1.0710757971784265e-05, "loss": 0.0524, "step": 7285 }, { "epoch": 7.39, "learning_rate": 1.0708571333199563e-05, "loss": 0.059, "step": 7286 }, { "epoch": 7.39, "learning_rate": 1.0706384660563824e-05, "loss": 0.0388, "step": 7287 }, { "epoch": 7.4, "learning_rate": 1.0704197953982134e-05, "loss": 0.0587, "step": 7288 }, { "epoch": 7.4, "learning_rate": 1.0702011213559573e-05, "loss": 0.0377, "step": 7289 }, { "epoch": 7.4, "learning_rate": 1.069982443940123e-05, "loss": 0.0409, "step": 7290 }, { "epoch": 7.4, "learning_rate": 1.0697637631612193e-05, "loss": 0.0485, "step": 7291 }, { "epoch": 7.4, "learning_rate": 1.0695450790297547e-05, "loss": 0.0592, "step": 7292 }, { "epoch": 7.4, "learning_rate": 1.0693263915562382e-05, "loss": 0.0457, "step": 7293 }, { "epoch": 7.4, "learning_rate": 1.0691077007511794e-05, "loss": 0.0437, "step": 7294 }, { "epoch": 7.4, "learning_rate": 1.0688890066250876e-05, "loss": 0.0383, "step": 7295 }, { "epoch": 7.4, "learning_rate": 1.0686703091884725e-05, "loss": 0.0328, "step": 7296 }, { "epoch": 7.4, "learning_rate": 1.0684516084518434e-05, "loss": 0.0449, "step": 7297 }, { "epoch": 7.41, "learning_rate": 1.0682329044257104e-05, "loss": 0.039, "step": 7298 }, { "epoch": 7.41, "learning_rate": 1.0680141971205834e-05, "loss": 0.045, "step": 7299 }, { "epoch": 7.41, "learning_rate": 1.0677954865469728e-05, "loss": 0.0445, "step": 7300 }, { "epoch": 7.41, "learning_rate": 1.0675767727153889e-05, "loss": 0.054, "step": 7301 }, { "epoch": 7.41, "learning_rate": 1.0673580556363423e-05, "loss": 0.0456, "step": 7302 }, { "epoch": 7.41, "learning_rate": 1.067139335320343e-05, "loss": 0.0518, "step": 7303 }, { "epoch": 7.41, "learning_rate": 1.0669206117779022e-05, "loss": 0.0491, "step": 7304 }, { "epoch": 7.41, "learning_rate": 1.0667018850195311e-05, "loss": 0.0464, "step": 7305 }, { "epoch": 7.41, "learning_rate": 1.0664831550557409e-05, "loss": 0.0442, "step": 7306 }, { "epoch": 7.41, "learning_rate": 1.0662644218970421e-05, "loss": 0.0432, "step": 7307 }, { "epoch": 7.42, "learning_rate": 1.0660456855539469e-05, "loss": 0.0455, "step": 7308 }, { "epoch": 7.42, "learning_rate": 1.0658269460369665e-05, "loss": 0.0493, "step": 7309 }, { "epoch": 7.42, "learning_rate": 1.0656082033566124e-05, "loss": 0.0577, "step": 7310 }, { "epoch": 7.42, "learning_rate": 1.065389457523397e-05, "loss": 0.0433, "step": 7311 }, { "epoch": 7.42, "learning_rate": 1.0651707085478322e-05, "loss": 0.0493, "step": 7312 }, { "epoch": 7.42, "learning_rate": 1.0649519564404303e-05, "loss": 0.0532, "step": 7313 }, { "epoch": 7.42, "learning_rate": 1.0647332012117031e-05, "loss": 0.049, "step": 7314 }, { "epoch": 7.42, "learning_rate": 1.0645144428721635e-05, "loss": 0.046, "step": 7315 }, { "epoch": 7.42, "learning_rate": 1.0642956814323236e-05, "loss": 0.0356, "step": 7316 }, { "epoch": 7.42, "learning_rate": 1.0640769169026971e-05, "loss": 0.0508, "step": 7317 }, { "epoch": 7.43, "learning_rate": 1.0638581492937966e-05, "loss": 0.0405, "step": 7318 }, { "epoch": 7.43, "learning_rate": 1.0636393786161346e-05, "loss": 0.0422, "step": 7319 }, { "epoch": 7.43, "learning_rate": 1.0634206048802247e-05, "loss": 0.0441, "step": 7320 }, { "epoch": 7.43, "learning_rate": 1.0632018280965804e-05, "loss": 0.0467, "step": 7321 }, { "epoch": 7.43, "learning_rate": 1.0629830482757151e-05, "loss": 0.0444, "step": 7322 }, { "epoch": 7.43, "learning_rate": 1.0627642654281426e-05, "loss": 0.0417, "step": 7323 }, { "epoch": 7.43, "learning_rate": 1.0625454795643764e-05, "loss": 0.0424, "step": 7324 }, { "epoch": 7.43, "learning_rate": 1.0623266906949308e-05, "loss": 0.0434, "step": 7325 }, { "epoch": 7.43, "learning_rate": 1.0621078988303195e-05, "loss": 0.0537, "step": 7326 }, { "epoch": 7.43, "learning_rate": 1.0618891039810572e-05, "loss": 0.0427, "step": 7327 }, { "epoch": 7.44, "learning_rate": 1.061670306157658e-05, "loss": 0.045, "step": 7328 }, { "epoch": 7.44, "learning_rate": 1.0614515053706367e-05, "loss": 0.0524, "step": 7329 }, { "epoch": 7.44, "learning_rate": 1.0612327016305075e-05, "loss": 0.0511, "step": 7330 }, { "epoch": 7.44, "learning_rate": 1.0610138949477858e-05, "loss": 0.0416, "step": 7331 }, { "epoch": 7.44, "learning_rate": 1.0607950853329854e-05, "loss": 0.0464, "step": 7332 }, { "epoch": 7.44, "learning_rate": 1.060576272796623e-05, "loss": 0.0504, "step": 7333 }, { "epoch": 7.44, "learning_rate": 1.0603574573492127e-05, "loss": 0.0686, "step": 7334 }, { "epoch": 7.44, "learning_rate": 1.0601386390012705e-05, "loss": 0.0464, "step": 7335 }, { "epoch": 7.44, "learning_rate": 1.0599198177633112e-05, "loss": 0.0456, "step": 7336 }, { "epoch": 7.44, "learning_rate": 1.0597009936458513e-05, "loss": 0.0431, "step": 7337 }, { "epoch": 7.45, "learning_rate": 1.059482166659406e-05, "loss": 0.0464, "step": 7338 }, { "epoch": 7.45, "learning_rate": 1.0592633368144913e-05, "loss": 0.0475, "step": 7339 }, { "epoch": 7.45, "learning_rate": 1.0590445041216238e-05, "loss": 0.0406, "step": 7340 }, { "epoch": 7.45, "learning_rate": 1.0588256685913187e-05, "loss": 0.0543, "step": 7341 }, { "epoch": 7.45, "learning_rate": 1.0586068302340932e-05, "loss": 0.0491, "step": 7342 }, { "epoch": 7.45, "learning_rate": 1.0583879890604635e-05, "loss": 0.0706, "step": 7343 }, { "epoch": 7.45, "learning_rate": 1.0581691450809464e-05, "loss": 0.0702, "step": 7344 }, { "epoch": 7.45, "learning_rate": 1.0579502983060579e-05, "loss": 0.0507, "step": 7345 }, { "epoch": 7.45, "learning_rate": 1.057731448746316e-05, "loss": 0.0555, "step": 7346 }, { "epoch": 7.46, "learning_rate": 1.057512596412237e-05, "loss": 0.0379, "step": 7347 }, { "epoch": 7.46, "learning_rate": 1.0572937413143377e-05, "loss": 0.0387, "step": 7348 }, { "epoch": 7.46, "learning_rate": 1.0570748834631363e-05, "loss": 0.0504, "step": 7349 }, { "epoch": 7.46, "learning_rate": 1.0568560228691497e-05, "loss": 0.0473, "step": 7350 }, { "epoch": 7.46, "learning_rate": 1.0566371595428958e-05, "loss": 0.0341, "step": 7351 }, { "epoch": 7.46, "learning_rate": 1.056418293494892e-05, "loss": 0.0427, "step": 7352 }, { "epoch": 7.46, "learning_rate": 1.0561994247356555e-05, "loss": 0.0565, "step": 7353 }, { "epoch": 7.46, "learning_rate": 1.055980553275705e-05, "loss": 0.0422, "step": 7354 }, { "epoch": 7.46, "learning_rate": 1.0557616791255588e-05, "loss": 0.0486, "step": 7355 }, { "epoch": 7.46, "learning_rate": 1.0555428022957346e-05, "loss": 0.0402, "step": 7356 }, { "epoch": 7.47, "learning_rate": 1.0553239227967504e-05, "loss": 0.0426, "step": 7357 }, { "epoch": 7.47, "learning_rate": 1.0551050406391256e-05, "loss": 0.0477, "step": 7358 }, { "epoch": 7.47, "learning_rate": 1.0548861558333781e-05, "loss": 0.0475, "step": 7359 }, { "epoch": 7.47, "learning_rate": 1.0546672683900265e-05, "loss": 0.0588, "step": 7360 }, { "epoch": 7.47, "learning_rate": 1.05444837831959e-05, "loss": 0.0475, "step": 7361 }, { "epoch": 7.47, "learning_rate": 1.0542294856325877e-05, "loss": 0.0483, "step": 7362 }, { "epoch": 7.47, "learning_rate": 1.0540105903395383e-05, "loss": 0.0576, "step": 7363 }, { "epoch": 7.47, "learning_rate": 1.0537916924509613e-05, "loss": 0.0329, "step": 7364 }, { "epoch": 7.47, "learning_rate": 1.0535727919773757e-05, "loss": 0.0514, "step": 7365 }, { "epoch": 7.47, "learning_rate": 1.053353888929301e-05, "loss": 0.0516, "step": 7366 }, { "epoch": 7.48, "learning_rate": 1.0531349833172572e-05, "loss": 0.0407, "step": 7367 }, { "epoch": 7.48, "learning_rate": 1.0529160751517637e-05, "loss": 0.0324, "step": 7368 }, { "epoch": 7.48, "learning_rate": 1.0526971644433405e-05, "loss": 0.0513, "step": 7369 }, { "epoch": 7.48, "learning_rate": 1.0524782512025071e-05, "loss": 0.0427, "step": 7370 }, { "epoch": 7.48, "learning_rate": 1.052259335439784e-05, "loss": 0.0481, "step": 7371 }, { "epoch": 7.48, "learning_rate": 1.0520404171656915e-05, "loss": 0.0368, "step": 7372 }, { "epoch": 7.48, "learning_rate": 1.0518214963907496e-05, "loss": 0.039, "step": 7373 }, { "epoch": 7.48, "learning_rate": 1.0516025731254786e-05, "loss": 0.0443, "step": 7374 }, { "epoch": 7.48, "learning_rate": 1.0513836473803997e-05, "loss": 0.0461, "step": 7375 }, { "epoch": 7.48, "learning_rate": 1.0511647191660328e-05, "loss": 0.0488, "step": 7376 }, { "epoch": 7.49, "learning_rate": 1.0509457884928994e-05, "loss": 0.0469, "step": 7377 }, { "epoch": 7.49, "learning_rate": 1.0507268553715198e-05, "loss": 0.057, "step": 7378 }, { "epoch": 7.49, "learning_rate": 1.0505079198124155e-05, "loss": 0.0427, "step": 7379 }, { "epoch": 7.49, "learning_rate": 1.0502889818261075e-05, "loss": 0.0435, "step": 7380 }, { "epoch": 7.49, "learning_rate": 1.0500700414231169e-05, "loss": 0.053, "step": 7381 }, { "epoch": 7.49, "learning_rate": 1.0498510986139649e-05, "loss": 0.039, "step": 7382 }, { "epoch": 7.49, "learning_rate": 1.0496321534091736e-05, "loss": 0.0476, "step": 7383 }, { "epoch": 7.49, "learning_rate": 1.0494132058192643e-05, "loss": 0.0364, "step": 7384 }, { "epoch": 7.49, "learning_rate": 1.0491942558547588e-05, "loss": 0.0535, "step": 7385 }, { "epoch": 7.49, "learning_rate": 1.0489753035261787e-05, "loss": 0.0493, "step": 7386 }, { "epoch": 7.5, "learning_rate": 1.0487563488440461e-05, "loss": 0.0363, "step": 7387 }, { "epoch": 7.5, "learning_rate": 1.0485373918188833e-05, "loss": 0.0406, "step": 7388 }, { "epoch": 7.5, "learning_rate": 1.048318432461212e-05, "loss": 0.0363, "step": 7389 }, { "epoch": 7.5, "learning_rate": 1.0480994707815551e-05, "loss": 0.0448, "step": 7390 }, { "epoch": 7.5, "learning_rate": 1.047880506790434e-05, "loss": 0.0364, "step": 7391 }, { "epoch": 7.5, "learning_rate": 1.0476615404983725e-05, "loss": 0.0445, "step": 7392 }, { "epoch": 7.5, "learning_rate": 1.0474425719158922e-05, "loss": 0.0542, "step": 7393 }, { "epoch": 7.5, "learning_rate": 1.0472236010535167e-05, "loss": 0.0482, "step": 7394 }, { "epoch": 7.5, "learning_rate": 1.0470046279217676e-05, "loss": 0.0439, "step": 7395 }, { "epoch": 7.5, "learning_rate": 1.0467856525311692e-05, "loss": 0.0521, "step": 7396 }, { "epoch": 7.51, "learning_rate": 1.0465666748922438e-05, "loss": 0.0408, "step": 7397 }, { "epoch": 7.51, "learning_rate": 1.0463476950155146e-05, "loss": 0.0495, "step": 7398 }, { "epoch": 7.51, "learning_rate": 1.0461287129115052e-05, "loss": 0.0453, "step": 7399 }, { "epoch": 7.51, "learning_rate": 1.0459097285907389e-05, "loss": 0.0536, "step": 7400 }, { "epoch": 7.51, "learning_rate": 1.045690742063739e-05, "loss": 0.0442, "step": 7401 }, { "epoch": 7.51, "learning_rate": 1.0454717533410293e-05, "loss": 0.0496, "step": 7402 }, { "epoch": 7.51, "learning_rate": 1.045252762433133e-05, "loss": 0.041, "step": 7403 }, { "epoch": 7.51, "learning_rate": 1.0450337693505744e-05, "loss": 0.0435, "step": 7404 }, { "epoch": 7.51, "learning_rate": 1.0448147741038774e-05, "loss": 0.0511, "step": 7405 }, { "epoch": 7.51, "learning_rate": 1.0445957767035661e-05, "loss": 0.0388, "step": 7406 }, { "epoch": 7.52, "learning_rate": 1.044376777160164e-05, "loss": 0.0608, "step": 7407 }, { "epoch": 7.52, "learning_rate": 1.0441577754841963e-05, "loss": 0.0521, "step": 7408 }, { "epoch": 7.52, "learning_rate": 1.0439387716861866e-05, "loss": 0.0426, "step": 7409 }, { "epoch": 7.52, "learning_rate": 1.0437197657766592e-05, "loss": 0.0447, "step": 7410 }, { "epoch": 7.52, "learning_rate": 1.0435007577661393e-05, "loss": 0.0579, "step": 7411 }, { "epoch": 7.52, "learning_rate": 1.0432817476651509e-05, "loss": 0.0429, "step": 7412 }, { "epoch": 7.52, "learning_rate": 1.0430627354842192e-05, "loss": 0.0429, "step": 7413 }, { "epoch": 7.52, "learning_rate": 1.0428437212338686e-05, "loss": 0.0452, "step": 7414 }, { "epoch": 7.52, "learning_rate": 1.0426247049246244e-05, "loss": 0.0443, "step": 7415 }, { "epoch": 7.53, "learning_rate": 1.0424056865670112e-05, "loss": 0.0376, "step": 7416 }, { "epoch": 7.53, "learning_rate": 1.0421866661715547e-05, "loss": 0.0471, "step": 7417 }, { "epoch": 7.53, "learning_rate": 1.0419676437487798e-05, "loss": 0.0501, "step": 7418 }, { "epoch": 7.53, "learning_rate": 1.0417486193092119e-05, "loss": 0.06, "step": 7419 }, { "epoch": 7.53, "learning_rate": 1.041529592863376e-05, "loss": 0.0409, "step": 7420 }, { "epoch": 7.53, "learning_rate": 1.0413105644217983e-05, "loss": 0.0456, "step": 7421 }, { "epoch": 7.53, "learning_rate": 1.0410915339950038e-05, "loss": 0.0396, "step": 7422 }, { "epoch": 7.53, "learning_rate": 1.0408725015935188e-05, "loss": 0.0459, "step": 7423 }, { "epoch": 7.53, "learning_rate": 1.0406534672278685e-05, "loss": 0.0404, "step": 7424 }, { "epoch": 7.53, "learning_rate": 1.0404344309085794e-05, "loss": 0.0511, "step": 7425 }, { "epoch": 7.54, "learning_rate": 1.040215392646177e-05, "loss": 0.0485, "step": 7426 }, { "epoch": 7.54, "learning_rate": 1.0399963524511875e-05, "loss": 0.046, "step": 7427 }, { "epoch": 7.54, "learning_rate": 1.039777310334137e-05, "loss": 0.0427, "step": 7428 }, { "epoch": 7.54, "learning_rate": 1.039558266305552e-05, "loss": 0.0401, "step": 7429 }, { "epoch": 7.54, "learning_rate": 1.0393392203759592e-05, "loss": 0.0588, "step": 7430 }, { "epoch": 7.54, "learning_rate": 1.0391201725558842e-05, "loss": 0.0393, "step": 7431 }, { "epoch": 7.54, "learning_rate": 1.0389011228558539e-05, "loss": 0.0662, "step": 7432 }, { "epoch": 7.54, "learning_rate": 1.038682071286395e-05, "loss": 0.0462, "step": 7433 }, { "epoch": 7.54, "learning_rate": 1.0384630178580342e-05, "loss": 0.0485, "step": 7434 }, { "epoch": 7.54, "learning_rate": 1.0382439625812984e-05, "loss": 0.0461, "step": 7435 }, { "epoch": 7.55, "learning_rate": 1.0380249054667144e-05, "loss": 0.0524, "step": 7436 }, { "epoch": 7.55, "learning_rate": 1.0378058465248091e-05, "loss": 0.0484, "step": 7437 }, { "epoch": 7.55, "learning_rate": 1.0375867857661095e-05, "loss": 0.0435, "step": 7438 }, { "epoch": 7.55, "learning_rate": 1.0373677232011432e-05, "loss": 0.0445, "step": 7439 }, { "epoch": 7.55, "learning_rate": 1.0371486588404372e-05, "loss": 0.0488, "step": 7440 }, { "epoch": 7.55, "learning_rate": 1.0369295926945187e-05, "loss": 0.0428, "step": 7441 }, { "epoch": 7.55, "learning_rate": 1.0367105247739153e-05, "loss": 0.0415, "step": 7442 }, { "epoch": 7.55, "learning_rate": 1.0364914550891543e-05, "loss": 0.0568, "step": 7443 }, { "epoch": 7.55, "learning_rate": 1.0362723836507636e-05, "loss": 0.0396, "step": 7444 }, { "epoch": 7.55, "learning_rate": 1.0360533104692705e-05, "loss": 0.0561, "step": 7445 }, { "epoch": 7.56, "learning_rate": 1.0358342355552034e-05, "loss": 0.0366, "step": 7446 }, { "epoch": 7.56, "learning_rate": 1.0356151589190895e-05, "loss": 0.0391, "step": 7447 }, { "epoch": 7.56, "learning_rate": 1.0353960805714567e-05, "loss": 0.0394, "step": 7448 }, { "epoch": 7.56, "learning_rate": 1.0351770005228332e-05, "loss": 0.034, "step": 7449 }, { "epoch": 7.56, "learning_rate": 1.0349579187837475e-05, "loss": 0.0449, "step": 7450 }, { "epoch": 7.56, "learning_rate": 1.0347388353647274e-05, "loss": 0.0454, "step": 7451 }, { "epoch": 7.56, "learning_rate": 1.0345197502763012e-05, "loss": 0.0448, "step": 7452 }, { "epoch": 7.56, "learning_rate": 1.0343006635289968e-05, "loss": 0.0533, "step": 7453 }, { "epoch": 7.56, "learning_rate": 1.0340815751333432e-05, "loss": 0.0403, "step": 7454 }, { "epoch": 7.56, "learning_rate": 1.0338624850998686e-05, "loss": 0.0514, "step": 7455 }, { "epoch": 7.57, "learning_rate": 1.0336433934391021e-05, "loss": 0.0584, "step": 7456 }, { "epoch": 7.57, "learning_rate": 1.0334243001615713e-05, "loss": 0.0427, "step": 7457 }, { "epoch": 7.57, "learning_rate": 1.0332052052778061e-05, "loss": 0.0516, "step": 7458 }, { "epoch": 7.57, "learning_rate": 1.0329861087983346e-05, "loss": 0.0496, "step": 7459 }, { "epoch": 7.57, "learning_rate": 1.0327670107336856e-05, "loss": 0.0462, "step": 7460 }, { "epoch": 7.57, "learning_rate": 1.0325479110943883e-05, "loss": 0.0496, "step": 7461 }, { "epoch": 7.57, "learning_rate": 1.0323288098909718e-05, "loss": 0.0421, "step": 7462 }, { "epoch": 7.57, "learning_rate": 1.0321097071339651e-05, "loss": 0.0445, "step": 7463 }, { "epoch": 7.57, "learning_rate": 1.0318906028338975e-05, "loss": 0.0395, "step": 7464 }, { "epoch": 7.57, "learning_rate": 1.031671497001298e-05, "loss": 0.0428, "step": 7465 }, { "epoch": 7.58, "learning_rate": 1.0314523896466961e-05, "loss": 0.0447, "step": 7466 }, { "epoch": 7.58, "learning_rate": 1.0312332807806213e-05, "loss": 0.043, "step": 7467 }, { "epoch": 7.58, "learning_rate": 1.031014170413603e-05, "loss": 0.041, "step": 7468 }, { "epoch": 7.58, "learning_rate": 1.0307950585561705e-05, "loss": 0.0439, "step": 7469 }, { "epoch": 7.58, "learning_rate": 1.0305759452188537e-05, "loss": 0.0479, "step": 7470 }, { "epoch": 7.58, "learning_rate": 1.0303568304121821e-05, "loss": 0.0418, "step": 7471 }, { "epoch": 7.58, "learning_rate": 1.0301377141466857e-05, "loss": 0.0482, "step": 7472 }, { "epoch": 7.58, "learning_rate": 1.0299185964328943e-05, "loss": 0.042, "step": 7473 }, { "epoch": 7.58, "learning_rate": 1.0296994772813376e-05, "loss": 0.044, "step": 7474 }, { "epoch": 7.58, "learning_rate": 1.0294803567025458e-05, "loss": 0.0408, "step": 7475 }, { "epoch": 7.59, "learning_rate": 1.0292612347070483e-05, "loss": 0.0501, "step": 7476 }, { "epoch": 7.59, "learning_rate": 1.0290421113053763e-05, "loss": 0.0504, "step": 7477 }, { "epoch": 7.59, "learning_rate": 1.028822986508059e-05, "loss": 0.0591, "step": 7478 }, { "epoch": 7.59, "learning_rate": 1.0286038603256272e-05, "loss": 0.0508, "step": 7479 }, { "epoch": 7.59, "learning_rate": 1.0283847327686112e-05, "loss": 0.0436, "step": 7480 }, { "epoch": 7.59, "learning_rate": 1.0281656038475411e-05, "loss": 0.0474, "step": 7481 }, { "epoch": 7.59, "learning_rate": 1.0279464735729472e-05, "loss": 0.0482, "step": 7482 }, { "epoch": 7.59, "learning_rate": 1.0277273419553605e-05, "loss": 0.0561, "step": 7483 }, { "epoch": 7.59, "learning_rate": 1.0275082090053112e-05, "loss": 0.0485, "step": 7484 }, { "epoch": 7.6, "learning_rate": 1.0272890747333302e-05, "loss": 0.0416, "step": 7485 }, { "epoch": 7.6, "learning_rate": 1.027069939149948e-05, "loss": 0.0401, "step": 7486 }, { "epoch": 7.6, "learning_rate": 1.0268508022656953e-05, "loss": 0.0461, "step": 7487 }, { "epoch": 7.6, "learning_rate": 1.026631664091103e-05, "loss": 0.0432, "step": 7488 }, { "epoch": 7.6, "learning_rate": 1.0264125246367021e-05, "loss": 0.0386, "step": 7489 }, { "epoch": 7.6, "learning_rate": 1.0261933839130234e-05, "loss": 0.0403, "step": 7490 }, { "epoch": 7.6, "learning_rate": 1.0259742419305979e-05, "loss": 0.0464, "step": 7491 }, { "epoch": 7.6, "learning_rate": 1.025755098699957e-05, "loss": 0.0572, "step": 7492 }, { "epoch": 7.6, "learning_rate": 1.025535954231631e-05, "loss": 0.0467, "step": 7493 }, { "epoch": 7.6, "learning_rate": 1.0253168085361522e-05, "loss": 0.0387, "step": 7494 }, { "epoch": 7.61, "learning_rate": 1.025097661624051e-05, "loss": 0.0396, "step": 7495 }, { "epoch": 7.61, "learning_rate": 1.024878513505859e-05, "loss": 0.044, "step": 7496 }, { "epoch": 7.61, "learning_rate": 1.0246593641921077e-05, "loss": 0.0391, "step": 7497 }, { "epoch": 7.61, "learning_rate": 1.0244402136933281e-05, "loss": 0.0585, "step": 7498 }, { "epoch": 7.61, "learning_rate": 1.024221062020052e-05, "loss": 0.044, "step": 7499 }, { "epoch": 7.61, "learning_rate": 1.0240019091828108e-05, "loss": 0.0526, "step": 7500 }, { "epoch": 7.61, "learning_rate": 1.0237827551921364e-05, "loss": 0.0417, "step": 7501 }, { "epoch": 7.61, "learning_rate": 1.0235636000585602e-05, "loss": 0.0389, "step": 7502 }, { "epoch": 7.61, "learning_rate": 1.0233444437926138e-05, "loss": 0.0436, "step": 7503 }, { "epoch": 7.61, "learning_rate": 1.0231252864048291e-05, "loss": 0.0445, "step": 7504 }, { "epoch": 7.62, "learning_rate": 1.0229061279057378e-05, "loss": 0.0479, "step": 7505 }, { "epoch": 7.62, "learning_rate": 1.0226869683058718e-05, "loss": 0.0474, "step": 7506 }, { "epoch": 7.62, "learning_rate": 1.022467807615763e-05, "loss": 0.0402, "step": 7507 }, { "epoch": 7.62, "learning_rate": 1.0222486458459435e-05, "loss": 0.046, "step": 7508 }, { "epoch": 7.62, "learning_rate": 1.0220294830069452e-05, "loss": 0.0556, "step": 7509 }, { "epoch": 7.62, "learning_rate": 1.0218103191093e-05, "loss": 0.0471, "step": 7510 }, { "epoch": 7.62, "learning_rate": 1.0215911541635404e-05, "loss": 0.051, "step": 7511 }, { "epoch": 7.62, "learning_rate": 1.0213719881801982e-05, "loss": 0.0536, "step": 7512 }, { "epoch": 7.62, "learning_rate": 1.0211528211698059e-05, "loss": 0.046, "step": 7513 }, { "epoch": 7.62, "learning_rate": 1.0209336531428956e-05, "loss": 0.0453, "step": 7514 }, { "epoch": 7.63, "learning_rate": 1.0207144841099998e-05, "loss": 0.0348, "step": 7515 }, { "epoch": 7.63, "learning_rate": 1.0204953140816505e-05, "loss": 0.0432, "step": 7516 }, { "epoch": 7.63, "learning_rate": 1.0202761430683805e-05, "loss": 0.0608, "step": 7517 }, { "epoch": 7.63, "learning_rate": 1.0200569710807222e-05, "loss": 0.0538, "step": 7518 }, { "epoch": 7.63, "learning_rate": 1.019837798129208e-05, "loss": 0.0589, "step": 7519 }, { "epoch": 7.63, "learning_rate": 1.0196186242243704e-05, "loss": 0.0479, "step": 7520 }, { "epoch": 7.63, "learning_rate": 1.019399449376742e-05, "loss": 0.0508, "step": 7521 }, { "epoch": 7.63, "learning_rate": 1.0191802735968557e-05, "loss": 0.0468, "step": 7522 }, { "epoch": 7.63, "learning_rate": 1.018961096895244e-05, "loss": 0.0561, "step": 7523 }, { "epoch": 7.63, "learning_rate": 1.0187419192824395e-05, "loss": 0.0357, "step": 7524 }, { "epoch": 7.64, "learning_rate": 1.0185227407689754e-05, "loss": 0.0493, "step": 7525 }, { "epoch": 7.64, "learning_rate": 1.018303561365384e-05, "loss": 0.0425, "step": 7526 }, { "epoch": 7.64, "learning_rate": 1.0180843810821984e-05, "loss": 0.0522, "step": 7527 }, { "epoch": 7.64, "learning_rate": 1.0178651999299516e-05, "loss": 0.0454, "step": 7528 }, { "epoch": 7.64, "learning_rate": 1.0176460179191764e-05, "loss": 0.0412, "step": 7529 }, { "epoch": 7.64, "learning_rate": 1.017426835060406e-05, "loss": 0.0529, "step": 7530 }, { "epoch": 7.64, "learning_rate": 1.0172076513641731e-05, "loss": 0.041, "step": 7531 }, { "epoch": 7.64, "learning_rate": 1.0169884668410111e-05, "loss": 0.0558, "step": 7532 }, { "epoch": 7.64, "learning_rate": 1.0167692815014527e-05, "loss": 0.0498, "step": 7533 }, { "epoch": 7.64, "learning_rate": 1.0165500953560316e-05, "loss": 0.0465, "step": 7534 }, { "epoch": 7.65, "learning_rate": 1.0163309084152807e-05, "loss": 0.0581, "step": 7535 }, { "epoch": 7.65, "learning_rate": 1.0161117206897333e-05, "loss": 0.0437, "step": 7536 }, { "epoch": 7.65, "learning_rate": 1.0158925321899222e-05, "loss": 0.0504, "step": 7537 }, { "epoch": 7.65, "learning_rate": 1.0156733429263815e-05, "loss": 0.0462, "step": 7538 }, { "epoch": 7.65, "learning_rate": 1.015454152909644e-05, "loss": 0.0392, "step": 7539 }, { "epoch": 7.65, "learning_rate": 1.0152349621502433e-05, "loss": 0.0626, "step": 7540 }, { "epoch": 7.65, "learning_rate": 1.0150157706587125e-05, "loss": 0.0539, "step": 7541 }, { "epoch": 7.65, "learning_rate": 1.0147965784455855e-05, "loss": 0.0443, "step": 7542 }, { "epoch": 7.65, "learning_rate": 1.0145773855213952e-05, "loss": 0.0487, "step": 7543 }, { "epoch": 7.65, "learning_rate": 1.0143581918966758e-05, "loss": 0.0387, "step": 7544 }, { "epoch": 7.66, "learning_rate": 1.0141389975819605e-05, "loss": 0.0505, "step": 7545 }, { "epoch": 7.66, "learning_rate": 1.0139198025877828e-05, "loss": 0.0471, "step": 7546 }, { "epoch": 7.66, "learning_rate": 1.0137006069246764e-05, "loss": 0.0602, "step": 7547 }, { "epoch": 7.66, "learning_rate": 1.013481410603175e-05, "loss": 0.0384, "step": 7548 }, { "epoch": 7.66, "learning_rate": 1.0132622136338119e-05, "loss": 0.0593, "step": 7549 }, { "epoch": 7.66, "learning_rate": 1.0130430160271215e-05, "loss": 0.0396, "step": 7550 }, { "epoch": 7.66, "learning_rate": 1.0128238177936372e-05, "loss": 0.0453, "step": 7551 }, { "epoch": 7.66, "learning_rate": 1.0126046189438927e-05, "loss": 0.0516, "step": 7552 }, { "epoch": 7.66, "learning_rate": 1.012385419488422e-05, "loss": 0.0423, "step": 7553 }, { "epoch": 7.67, "learning_rate": 1.0121662194377584e-05, "loss": 0.0438, "step": 7554 }, { "epoch": 7.67, "learning_rate": 1.011947018802436e-05, "loss": 0.0493, "step": 7555 }, { "epoch": 7.67, "learning_rate": 1.0117278175929892e-05, "loss": 0.0427, "step": 7556 }, { "epoch": 7.67, "learning_rate": 1.0115086158199514e-05, "loss": 0.0435, "step": 7557 }, { "epoch": 7.67, "learning_rate": 1.0112894134938567e-05, "loss": 0.0433, "step": 7558 }, { "epoch": 7.67, "learning_rate": 1.011070210625239e-05, "loss": 0.0485, "step": 7559 }, { "epoch": 7.67, "learning_rate": 1.0108510072246322e-05, "loss": 0.0661, "step": 7560 }, { "epoch": 7.67, "learning_rate": 1.0106318033025706e-05, "loss": 0.0429, "step": 7561 }, { "epoch": 7.67, "learning_rate": 1.010412598869588e-05, "loss": 0.0407, "step": 7562 }, { "epoch": 7.67, "learning_rate": 1.0101933939362187e-05, "loss": 0.0427, "step": 7563 }, { "epoch": 7.68, "learning_rate": 1.0099741885129965e-05, "loss": 0.0407, "step": 7564 }, { "epoch": 7.68, "learning_rate": 1.0097549826104556e-05, "loss": 0.0457, "step": 7565 }, { "epoch": 7.68, "learning_rate": 1.00953577623913e-05, "loss": 0.0437, "step": 7566 }, { "epoch": 7.68, "learning_rate": 1.0093165694095546e-05, "loss": 0.039, "step": 7567 }, { "epoch": 7.68, "learning_rate": 1.0090973621322628e-05, "loss": 0.0486, "step": 7568 }, { "epoch": 7.68, "learning_rate": 1.008878154417789e-05, "loss": 0.038, "step": 7569 }, { "epoch": 7.68, "learning_rate": 1.0086589462766673e-05, "loss": 0.0445, "step": 7570 }, { "epoch": 7.68, "learning_rate": 1.0084397377194322e-05, "loss": 0.0425, "step": 7571 }, { "epoch": 7.68, "learning_rate": 1.008220528756618e-05, "loss": 0.0455, "step": 7572 }, { "epoch": 7.68, "learning_rate": 1.008001319398759e-05, "loss": 0.0422, "step": 7573 }, { "epoch": 7.69, "learning_rate": 1.0077821096563888e-05, "loss": 0.0518, "step": 7574 }, { "epoch": 7.69, "learning_rate": 1.0075628995400428e-05, "loss": 0.0486, "step": 7575 }, { "epoch": 7.69, "learning_rate": 1.0073436890602548e-05, "loss": 0.042, "step": 7576 }, { "epoch": 7.69, "learning_rate": 1.0071244782275591e-05, "loss": 0.0479, "step": 7577 }, { "epoch": 7.69, "learning_rate": 1.0069052670524902e-05, "loss": 0.0365, "step": 7578 }, { "epoch": 7.69, "learning_rate": 1.0066860555455822e-05, "loss": 0.0424, "step": 7579 }, { "epoch": 7.69, "learning_rate": 1.0064668437173699e-05, "loss": 0.04, "step": 7580 }, { "epoch": 7.69, "learning_rate": 1.0062476315783878e-05, "loss": 0.0557, "step": 7581 }, { "epoch": 7.69, "learning_rate": 1.0060284191391698e-05, "loss": 0.0413, "step": 7582 }, { "epoch": 7.69, "learning_rate": 1.0058092064102504e-05, "loss": 0.0531, "step": 7583 }, { "epoch": 7.7, "learning_rate": 1.0055899934021649e-05, "loss": 0.0394, "step": 7584 }, { "epoch": 7.7, "learning_rate": 1.0053707801254471e-05, "loss": 0.0487, "step": 7585 }, { "epoch": 7.7, "learning_rate": 1.0051515665906317e-05, "loss": 0.0441, "step": 7586 }, { "epoch": 7.7, "learning_rate": 1.0049323528082528e-05, "loss": 0.039, "step": 7587 }, { "epoch": 7.7, "learning_rate": 1.0047131387888454e-05, "loss": 0.0458, "step": 7588 }, { "epoch": 7.7, "learning_rate": 1.0044939245429437e-05, "loss": 0.05, "step": 7589 }, { "epoch": 7.7, "learning_rate": 1.0042747100810827e-05, "loss": 0.0487, "step": 7590 }, { "epoch": 7.7, "learning_rate": 1.0040554954137964e-05, "loss": 0.0518, "step": 7591 }, { "epoch": 7.7, "learning_rate": 1.0038362805516198e-05, "loss": 0.0525, "step": 7592 }, { "epoch": 7.7, "learning_rate": 1.003617065505087e-05, "loss": 0.0463, "step": 7593 }, { "epoch": 7.71, "learning_rate": 1.003397850284733e-05, "loss": 0.0421, "step": 7594 }, { "epoch": 7.71, "learning_rate": 1.0031786349010921e-05, "loss": 0.0552, "step": 7595 }, { "epoch": 7.71, "learning_rate": 1.0029594193646992e-05, "loss": 0.044, "step": 7596 }, { "epoch": 7.71, "learning_rate": 1.0027402036860887e-05, "loss": 0.0369, "step": 7597 }, { "epoch": 7.71, "learning_rate": 1.0025209878757953e-05, "loss": 0.0438, "step": 7598 }, { "epoch": 7.71, "learning_rate": 1.0023017719443533e-05, "loss": 0.0446, "step": 7599 }, { "epoch": 7.71, "learning_rate": 1.0020825559022977e-05, "loss": 0.042, "step": 7600 }, { "epoch": 7.71, "learning_rate": 1.0018633397601632e-05, "loss": 0.0574, "step": 7601 }, { "epoch": 7.71, "learning_rate": 1.001644123528484e-05, "loss": 0.0379, "step": 7602 }, { "epoch": 7.71, "learning_rate": 1.0014249072177951e-05, "loss": 0.0459, "step": 7603 }, { "epoch": 7.72, "learning_rate": 1.0012056908386307e-05, "loss": 0.0467, "step": 7604 }, { "epoch": 7.72, "learning_rate": 1.0009864744015259e-05, "loss": 0.04, "step": 7605 }, { "epoch": 7.72, "learning_rate": 1.000767257917015e-05, "loss": 0.0416, "step": 7606 }, { "epoch": 7.72, "learning_rate": 1.0005480413956334e-05, "loss": 0.0461, "step": 7607 }, { "epoch": 7.72, "learning_rate": 1.0003288248479146e-05, "loss": 0.0453, "step": 7608 }, { "epoch": 7.72, "learning_rate": 1.0001096082843941e-05, "loss": 0.0483, "step": 7609 }, { "epoch": 7.72, "learning_rate": 9.99890391715606e-06, "loss": 0.0508, "step": 7610 }, { "epoch": 7.72, "learning_rate": 9.996711751520855e-06, "loss": 0.0432, "step": 7611 }, { "epoch": 7.72, "learning_rate": 9.994519586043669e-06, "loss": 0.0467, "step": 7612 }, { "epoch": 7.73, "learning_rate": 9.992327420829851e-06, "loss": 0.035, "step": 7613 }, { "epoch": 7.73, "learning_rate": 9.990135255984744e-06, "loss": 0.0377, "step": 7614 }, { "epoch": 7.73, "learning_rate": 9.987943091613695e-06, "loss": 0.0426, "step": 7615 }, { "epoch": 7.73, "learning_rate": 9.985750927822054e-06, "loss": 0.0499, "step": 7616 }, { "epoch": 7.73, "learning_rate": 9.983558764715166e-06, "loss": 0.0526, "step": 7617 }, { "epoch": 7.73, "learning_rate": 9.98136660239837e-06, "loss": 0.0386, "step": 7618 }, { "epoch": 7.73, "learning_rate": 9.979174440977024e-06, "loss": 0.0477, "step": 7619 }, { "epoch": 7.73, "learning_rate": 9.976982280556469e-06, "loss": 0.0484, "step": 7620 }, { "epoch": 7.73, "learning_rate": 9.974790121242049e-06, "loss": 0.0446, "step": 7621 }, { "epoch": 7.73, "learning_rate": 9.972597963139116e-06, "loss": 0.0398, "step": 7622 }, { "epoch": 7.74, "learning_rate": 9.970405806353012e-06, "loss": 0.0472, "step": 7623 }, { "epoch": 7.74, "learning_rate": 9.968213650989082e-06, "loss": 0.0481, "step": 7624 }, { "epoch": 7.74, "learning_rate": 9.966021497152676e-06, "loss": 0.0436, "step": 7625 }, { "epoch": 7.74, "learning_rate": 9.963829344949133e-06, "loss": 0.0598, "step": 7626 }, { "epoch": 7.74, "learning_rate": 9.961637194483807e-06, "loss": 0.0446, "step": 7627 }, { "epoch": 7.74, "learning_rate": 9.959445045862038e-06, "loss": 0.0449, "step": 7628 }, { "epoch": 7.74, "learning_rate": 9.957252899189176e-06, "loss": 0.0429, "step": 7629 }, { "epoch": 7.74, "learning_rate": 9.955060754570564e-06, "loss": 0.0421, "step": 7630 }, { "epoch": 7.74, "learning_rate": 9.952868612111551e-06, "loss": 0.0389, "step": 7631 }, { "epoch": 7.74, "learning_rate": 9.950676471917475e-06, "loss": 0.0521, "step": 7632 }, { "epoch": 7.75, "learning_rate": 9.948484334093688e-06, "loss": 0.0527, "step": 7633 }, { "epoch": 7.75, "learning_rate": 9.946292198745534e-06, "loss": 0.0443, "step": 7634 }, { "epoch": 7.75, "learning_rate": 9.944100065978351e-06, "loss": 0.0421, "step": 7635 }, { "epoch": 7.75, "learning_rate": 9.941907935897494e-06, "loss": 0.046, "step": 7636 }, { "epoch": 7.75, "learning_rate": 9.939715808608306e-06, "loss": 0.0423, "step": 7637 }, { "epoch": 7.75, "learning_rate": 9.937523684216126e-06, "loss": 0.0477, "step": 7638 }, { "epoch": 7.75, "learning_rate": 9.935331562826303e-06, "loss": 0.052, "step": 7639 }, { "epoch": 7.75, "learning_rate": 9.933139444544182e-06, "loss": 0.0408, "step": 7640 }, { "epoch": 7.75, "learning_rate": 9.930947329475105e-06, "loss": 0.0486, "step": 7641 }, { "epoch": 7.75, "learning_rate": 9.928755217724412e-06, "loss": 0.0473, "step": 7642 }, { "epoch": 7.76, "learning_rate": 9.926563109397454e-06, "loss": 0.045, "step": 7643 }, { "epoch": 7.76, "learning_rate": 9.924371004599572e-06, "loss": 0.0691, "step": 7644 }, { "epoch": 7.76, "learning_rate": 9.92217890343611e-06, "loss": 0.0436, "step": 7645 }, { "epoch": 7.76, "learning_rate": 9.919986806012414e-06, "loss": 0.0471, "step": 7646 }, { "epoch": 7.76, "learning_rate": 9.917794712433822e-06, "loss": 0.0417, "step": 7647 }, { "epoch": 7.76, "learning_rate": 9.91560262280568e-06, "loss": 0.0432, "step": 7648 }, { "epoch": 7.76, "learning_rate": 9.913410537233328e-06, "loss": 0.0423, "step": 7649 }, { "epoch": 7.76, "learning_rate": 9.911218455822116e-06, "loss": 0.045, "step": 7650 }, { "epoch": 7.76, "learning_rate": 9.909026378677377e-06, "loss": 0.0405, "step": 7651 }, { "epoch": 7.76, "learning_rate": 9.906834305904456e-06, "loss": 0.0419, "step": 7652 }, { "epoch": 7.77, "learning_rate": 9.9046422376087e-06, "loss": 0.0475, "step": 7653 }, { "epoch": 7.77, "learning_rate": 9.902450173895448e-06, "loss": 0.0445, "step": 7654 }, { "epoch": 7.77, "learning_rate": 9.900258114870037e-06, "loss": 0.0511, "step": 7655 }, { "epoch": 7.77, "learning_rate": 9.898066060637816e-06, "loss": 0.0457, "step": 7656 }, { "epoch": 7.77, "learning_rate": 9.895874011304122e-06, "loss": 0.0368, "step": 7657 }, { "epoch": 7.77, "learning_rate": 9.8936819669743e-06, "loss": 0.0427, "step": 7658 }, { "epoch": 7.77, "learning_rate": 9.891489927753681e-06, "loss": 0.0412, "step": 7659 }, { "epoch": 7.77, "learning_rate": 9.889297893747613e-06, "loss": 0.0438, "step": 7660 }, { "epoch": 7.77, "learning_rate": 9.887105865061435e-06, "loss": 0.0443, "step": 7661 }, { "epoch": 7.77, "learning_rate": 9.884913841800487e-06, "loss": 0.0537, "step": 7662 }, { "epoch": 7.78, "learning_rate": 9.882721824070111e-06, "loss": 0.0454, "step": 7663 }, { "epoch": 7.78, "learning_rate": 9.880529811975642e-06, "loss": 0.0441, "step": 7664 }, { "epoch": 7.78, "learning_rate": 9.87833780562242e-06, "loss": 0.0535, "step": 7665 }, { "epoch": 7.78, "learning_rate": 9.876145805115787e-06, "loss": 0.0444, "step": 7666 }, { "epoch": 7.78, "learning_rate": 9.873953810561078e-06, "loss": 0.0436, "step": 7667 }, { "epoch": 7.78, "learning_rate": 9.87176182206363e-06, "loss": 0.0596, "step": 7668 }, { "epoch": 7.78, "learning_rate": 9.869569839728785e-06, "loss": 0.0378, "step": 7669 }, { "epoch": 7.78, "learning_rate": 9.867377863661881e-06, "loss": 0.0339, "step": 7670 }, { "epoch": 7.78, "learning_rate": 9.865185893968255e-06, "loss": 0.0443, "step": 7671 }, { "epoch": 7.78, "learning_rate": 9.86299393075324e-06, "loss": 0.0496, "step": 7672 }, { "epoch": 7.79, "learning_rate": 9.860801974122175e-06, "loss": 0.0568, "step": 7673 }, { "epoch": 7.79, "learning_rate": 9.8586100241804e-06, "loss": 0.0558, "step": 7674 }, { "epoch": 7.79, "learning_rate": 9.856418081033247e-06, "loss": 0.0622, "step": 7675 }, { "epoch": 7.79, "learning_rate": 9.85422614478605e-06, "loss": 0.0469, "step": 7676 }, { "epoch": 7.79, "learning_rate": 9.852034215544148e-06, "loss": 0.0445, "step": 7677 }, { "epoch": 7.79, "learning_rate": 9.849842293412875e-06, "loss": 0.0475, "step": 7678 }, { "epoch": 7.79, "learning_rate": 9.84765037849757e-06, "loss": 0.0491, "step": 7679 }, { "epoch": 7.79, "learning_rate": 9.845458470903563e-06, "loss": 0.0487, "step": 7680 }, { "epoch": 7.79, "learning_rate": 9.84326657073619e-06, "loss": 0.0657, "step": 7681 }, { "epoch": 7.8, "learning_rate": 9.84107467810078e-06, "loss": 0.0426, "step": 7682 }, { "epoch": 7.8, "learning_rate": 9.838882793102672e-06, "loss": 0.054, "step": 7683 }, { "epoch": 7.8, "learning_rate": 9.836690915847198e-06, "loss": 0.0433, "step": 7684 }, { "epoch": 7.8, "learning_rate": 9.834499046439685e-06, "loss": 0.0534, "step": 7685 }, { "epoch": 7.8, "learning_rate": 9.832307184985475e-06, "loss": 0.0428, "step": 7686 }, { "epoch": 7.8, "learning_rate": 9.830115331589894e-06, "loss": 0.0393, "step": 7687 }, { "epoch": 7.8, "learning_rate": 9.827923486358272e-06, "loss": 0.0459, "step": 7688 }, { "epoch": 7.8, "learning_rate": 9.825731649395944e-06, "loss": 0.0451, "step": 7689 }, { "epoch": 7.8, "learning_rate": 9.82353982080824e-06, "loss": 0.051, "step": 7690 }, { "epoch": 7.8, "learning_rate": 9.82134800070049e-06, "loss": 0.0434, "step": 7691 }, { "epoch": 7.81, "learning_rate": 9.819156189178021e-06, "loss": 0.0439, "step": 7692 }, { "epoch": 7.81, "learning_rate": 9.816964386346165e-06, "loss": 0.0405, "step": 7693 }, { "epoch": 7.81, "learning_rate": 9.814772592310248e-06, "loss": 0.0459, "step": 7694 }, { "epoch": 7.81, "learning_rate": 9.812580807175607e-06, "loss": 0.0449, "step": 7695 }, { "epoch": 7.81, "learning_rate": 9.810389031047562e-06, "loss": 0.0495, "step": 7696 }, { "epoch": 7.81, "learning_rate": 9.808197264031446e-06, "loss": 0.0563, "step": 7697 }, { "epoch": 7.81, "learning_rate": 9.806005506232583e-06, "loss": 0.0594, "step": 7698 }, { "epoch": 7.81, "learning_rate": 9.803813757756298e-06, "loss": 0.037, "step": 7699 }, { "epoch": 7.81, "learning_rate": 9.801622018707923e-06, "loss": 0.0423, "step": 7700 }, { "epoch": 7.81, "learning_rate": 9.799430289192783e-06, "loss": 0.0422, "step": 7701 }, { "epoch": 7.82, "learning_rate": 9.797238569316196e-06, "loss": 0.0489, "step": 7702 }, { "epoch": 7.82, "learning_rate": 9.795046859183495e-06, "loss": 0.0458, "step": 7703 }, { "epoch": 7.82, "learning_rate": 9.792855158900006e-06, "loss": 0.0552, "step": 7704 }, { "epoch": 7.82, "learning_rate": 9.790663468571045e-06, "loss": 0.048, "step": 7705 }, { "epoch": 7.82, "learning_rate": 9.788471788301943e-06, "loss": 0.0456, "step": 7706 }, { "epoch": 7.82, "learning_rate": 9.786280118198022e-06, "loss": 0.051, "step": 7707 }, { "epoch": 7.82, "learning_rate": 9.7840884583646e-06, "loss": 0.0464, "step": 7708 }, { "epoch": 7.82, "learning_rate": 9.781896808907003e-06, "loss": 0.0538, "step": 7709 }, { "epoch": 7.82, "learning_rate": 9.779705169930551e-06, "loss": 0.0418, "step": 7710 }, { "epoch": 7.82, "learning_rate": 9.777513541540567e-06, "loss": 0.0484, "step": 7711 }, { "epoch": 7.83, "learning_rate": 9.775321923842371e-06, "loss": 0.0483, "step": 7712 }, { "epoch": 7.83, "learning_rate": 9.773130316941286e-06, "loss": 0.0448, "step": 7713 }, { "epoch": 7.83, "learning_rate": 9.770938720942625e-06, "loss": 0.0322, "step": 7714 }, { "epoch": 7.83, "learning_rate": 9.768747135951712e-06, "loss": 0.0488, "step": 7715 }, { "epoch": 7.83, "learning_rate": 9.766555562073865e-06, "loss": 0.0468, "step": 7716 }, { "epoch": 7.83, "learning_rate": 9.764363999414403e-06, "loss": 0.0417, "step": 7717 }, { "epoch": 7.83, "learning_rate": 9.76217244807864e-06, "loss": 0.0514, "step": 7718 }, { "epoch": 7.83, "learning_rate": 9.75998090817189e-06, "loss": 0.0434, "step": 7719 }, { "epoch": 7.83, "learning_rate": 9.757789379799482e-06, "loss": 0.0589, "step": 7720 }, { "epoch": 7.83, "learning_rate": 9.755597863066722e-06, "loss": 0.0453, "step": 7721 }, { "epoch": 7.84, "learning_rate": 9.753406358078927e-06, "loss": 0.0561, "step": 7722 }, { "epoch": 7.84, "learning_rate": 9.751214864941413e-06, "loss": 0.0439, "step": 7723 }, { "epoch": 7.84, "learning_rate": 9.749023383759494e-06, "loss": 0.0455, "step": 7724 }, { "epoch": 7.84, "learning_rate": 9.746831914638483e-06, "loss": 0.0471, "step": 7725 }, { "epoch": 7.84, "learning_rate": 9.744640457683693e-06, "loss": 0.0418, "step": 7726 }, { "epoch": 7.84, "learning_rate": 9.742449013000435e-06, "loss": 0.0461, "step": 7727 }, { "epoch": 7.84, "learning_rate": 9.740257580694023e-06, "loss": 0.0511, "step": 7728 }, { "epoch": 7.84, "learning_rate": 9.738066160869769e-06, "loss": 0.0467, "step": 7729 }, { "epoch": 7.84, "learning_rate": 9.735874753632982e-06, "loss": 0.04, "step": 7730 }, { "epoch": 7.84, "learning_rate": 9.733683359088974e-06, "loss": 0.044, "step": 7731 }, { "epoch": 7.85, "learning_rate": 9.73149197734305e-06, "loss": 0.0562, "step": 7732 }, { "epoch": 7.85, "learning_rate": 9.729300608500524e-06, "loss": 0.0411, "step": 7733 }, { "epoch": 7.85, "learning_rate": 9.727109252666703e-06, "loss": 0.0432, "step": 7734 }, { "epoch": 7.85, "learning_rate": 9.724917909946887e-06, "loss": 0.0489, "step": 7735 }, { "epoch": 7.85, "learning_rate": 9.722726580446397e-06, "loss": 0.0544, "step": 7736 }, { "epoch": 7.85, "learning_rate": 9.720535264270529e-06, "loss": 0.0459, "step": 7737 }, { "epoch": 7.85, "learning_rate": 9.71834396152459e-06, "loss": 0.0501, "step": 7738 }, { "epoch": 7.85, "learning_rate": 9.716152672313891e-06, "loss": 0.0607, "step": 7739 }, { "epoch": 7.85, "learning_rate": 9.71396139674373e-06, "loss": 0.0519, "step": 7740 }, { "epoch": 7.85, "learning_rate": 9.711770134919413e-06, "loss": 0.0471, "step": 7741 }, { "epoch": 7.86, "learning_rate": 9.709578886946243e-06, "loss": 0.0514, "step": 7742 }, { "epoch": 7.86, "learning_rate": 9.707387652929519e-06, "loss": 0.049, "step": 7743 }, { "epoch": 7.86, "learning_rate": 9.705196432974544e-06, "loss": 0.0439, "step": 7744 }, { "epoch": 7.86, "learning_rate": 9.703005227186626e-06, "loss": 0.0658, "step": 7745 }, { "epoch": 7.86, "learning_rate": 9.700814035671058e-06, "loss": 0.0522, "step": 7746 }, { "epoch": 7.86, "learning_rate": 9.698622858533145e-06, "loss": 0.0523, "step": 7747 }, { "epoch": 7.86, "learning_rate": 9.696431695878182e-06, "loss": 0.0456, "step": 7748 }, { "epoch": 7.86, "learning_rate": 9.694240547811466e-06, "loss": 0.0445, "step": 7749 }, { "epoch": 7.86, "learning_rate": 9.692049414438298e-06, "loss": 0.0359, "step": 7750 }, { "epoch": 7.87, "learning_rate": 9.689858295863975e-06, "loss": 0.0409, "step": 7751 }, { "epoch": 7.87, "learning_rate": 9.687667192193788e-06, "loss": 0.0465, "step": 7752 }, { "epoch": 7.87, "learning_rate": 9.68547610353304e-06, "loss": 0.0403, "step": 7753 }, { "epoch": 7.87, "learning_rate": 9.683285029987021e-06, "loss": 0.0578, "step": 7754 }, { "epoch": 7.87, "learning_rate": 9.681093971661027e-06, "loss": 0.0467, "step": 7755 }, { "epoch": 7.87, "learning_rate": 9.67890292866035e-06, "loss": 0.0597, "step": 7756 }, { "epoch": 7.87, "learning_rate": 9.676711901090285e-06, "loss": 0.0465, "step": 7757 }, { "epoch": 7.87, "learning_rate": 9.67452088905612e-06, "loss": 0.0514, "step": 7758 }, { "epoch": 7.87, "learning_rate": 9.672329892663147e-06, "loss": 0.0524, "step": 7759 }, { "epoch": 7.87, "learning_rate": 9.670138912016657e-06, "loss": 0.0503, "step": 7760 }, { "epoch": 7.88, "learning_rate": 9.66794794722194e-06, "loss": 0.0632, "step": 7761 }, { "epoch": 7.88, "learning_rate": 9.665756998384287e-06, "loss": 0.0409, "step": 7762 }, { "epoch": 7.88, "learning_rate": 9.663566065608982e-06, "loss": 0.0402, "step": 7763 }, { "epoch": 7.88, "learning_rate": 9.661375149001315e-06, "loss": 0.048, "step": 7764 }, { "epoch": 7.88, "learning_rate": 9.65918424866657e-06, "loss": 0.0392, "step": 7765 }, { "epoch": 7.88, "learning_rate": 9.656993364710034e-06, "loss": 0.0435, "step": 7766 }, { "epoch": 7.88, "learning_rate": 9.654802497236995e-06, "loss": 0.0473, "step": 7767 }, { "epoch": 7.88, "learning_rate": 9.652611646352732e-06, "loss": 0.0515, "step": 7768 }, { "epoch": 7.88, "learning_rate": 9.650420812162525e-06, "loss": 0.0478, "step": 7769 }, { "epoch": 7.88, "learning_rate": 9.648229994771668e-06, "loss": 0.058, "step": 7770 }, { "epoch": 7.89, "learning_rate": 9.646039194285437e-06, "loss": 0.0445, "step": 7771 }, { "epoch": 7.89, "learning_rate": 9.64384841080911e-06, "loss": 0.0508, "step": 7772 }, { "epoch": 7.89, "learning_rate": 9.64165764444797e-06, "loss": 0.0437, "step": 7773 }, { "epoch": 7.89, "learning_rate": 9.639466895307298e-06, "loss": 0.0529, "step": 7774 }, { "epoch": 7.89, "learning_rate": 9.63727616349237e-06, "loss": 0.0488, "step": 7775 }, { "epoch": 7.89, "learning_rate": 9.63508544910846e-06, "loss": 0.0512, "step": 7776 }, { "epoch": 7.89, "learning_rate": 9.632894752260849e-06, "loss": 0.0492, "step": 7777 }, { "epoch": 7.89, "learning_rate": 9.630704073054815e-06, "loss": 0.0421, "step": 7778 }, { "epoch": 7.89, "learning_rate": 9.62851341159563e-06, "loss": 0.061, "step": 7779 }, { "epoch": 7.89, "learning_rate": 9.626322767988571e-06, "loss": 0.0399, "step": 7780 }, { "epoch": 7.9, "learning_rate": 9.624132142338907e-06, "loss": 0.0569, "step": 7781 }, { "epoch": 7.9, "learning_rate": 9.621941534751912e-06, "loss": 0.0501, "step": 7782 }, { "epoch": 7.9, "learning_rate": 9.619750945332861e-06, "loss": 0.0451, "step": 7783 }, { "epoch": 7.9, "learning_rate": 9.617560374187021e-06, "loss": 0.0417, "step": 7784 }, { "epoch": 7.9, "learning_rate": 9.615369821419658e-06, "loss": 0.0472, "step": 7785 }, { "epoch": 7.9, "learning_rate": 9.613179287136052e-06, "loss": 0.0462, "step": 7786 }, { "epoch": 7.9, "learning_rate": 9.610988771441464e-06, "loss": 0.0452, "step": 7787 }, { "epoch": 7.9, "learning_rate": 9.60879827444116e-06, "loss": 0.0373, "step": 7788 }, { "epoch": 7.9, "learning_rate": 9.606607796240411e-06, "loss": 0.0516, "step": 7789 }, { "epoch": 7.9, "learning_rate": 9.60441733694448e-06, "loss": 0.0516, "step": 7790 }, { "epoch": 7.91, "learning_rate": 9.602226896658631e-06, "loss": 0.0427, "step": 7791 }, { "epoch": 7.91, "learning_rate": 9.60003647548813e-06, "loss": 0.049, "step": 7792 }, { "epoch": 7.91, "learning_rate": 9.597846073538234e-06, "loss": 0.0649, "step": 7793 }, { "epoch": 7.91, "learning_rate": 9.595655690914208e-06, "loss": 0.0499, "step": 7794 }, { "epoch": 7.91, "learning_rate": 9.593465327721314e-06, "loss": 0.0407, "step": 7795 }, { "epoch": 7.91, "learning_rate": 9.591274984064814e-06, "loss": 0.0435, "step": 7796 }, { "epoch": 7.91, "learning_rate": 9.589084660049964e-06, "loss": 0.0522, "step": 7797 }, { "epoch": 7.91, "learning_rate": 9.586894355782022e-06, "loss": 0.0435, "step": 7798 }, { "epoch": 7.91, "learning_rate": 9.584704071366242e-06, "loss": 0.0477, "step": 7799 }, { "epoch": 7.91, "learning_rate": 9.582513806907886e-06, "loss": 0.0466, "step": 7800 }, { "epoch": 7.92, "learning_rate": 9.580323562512207e-06, "loss": 0.0483, "step": 7801 }, { "epoch": 7.92, "learning_rate": 9.578133338284453e-06, "loss": 0.0559, "step": 7802 }, { "epoch": 7.92, "learning_rate": 9.57594313432989e-06, "loss": 0.0586, "step": 7803 }, { "epoch": 7.92, "learning_rate": 9.57375295075376e-06, "loss": 0.0497, "step": 7804 }, { "epoch": 7.92, "learning_rate": 9.571562787661316e-06, "loss": 0.0454, "step": 7805 }, { "epoch": 7.92, "learning_rate": 9.569372645157812e-06, "loss": 0.0527, "step": 7806 }, { "epoch": 7.92, "learning_rate": 9.567182523348495e-06, "loss": 0.0481, "step": 7807 }, { "epoch": 7.92, "learning_rate": 9.564992422338612e-06, "loss": 0.0507, "step": 7808 }, { "epoch": 7.92, "learning_rate": 9.56280234223341e-06, "loss": 0.0489, "step": 7809 }, { "epoch": 7.92, "learning_rate": 9.56061228313814e-06, "loss": 0.0576, "step": 7810 }, { "epoch": 7.93, "learning_rate": 9.558422245158039e-06, "loss": 0.0524, "step": 7811 }, { "epoch": 7.93, "learning_rate": 9.55623222839836e-06, "loss": 0.0483, "step": 7812 }, { "epoch": 7.93, "learning_rate": 9.554042232964342e-06, "loss": 0.0388, "step": 7813 }, { "epoch": 7.93, "learning_rate": 9.551852258961228e-06, "loss": 0.0434, "step": 7814 }, { "epoch": 7.93, "learning_rate": 9.54966230649426e-06, "loss": 0.0475, "step": 7815 }, { "epoch": 7.93, "learning_rate": 9.547472375668673e-06, "loss": 0.0465, "step": 7816 }, { "epoch": 7.93, "learning_rate": 9.545282466589712e-06, "loss": 0.0501, "step": 7817 }, { "epoch": 7.93, "learning_rate": 9.543092579362614e-06, "loss": 0.0454, "step": 7818 }, { "epoch": 7.93, "learning_rate": 9.540902714092613e-06, "loss": 0.0481, "step": 7819 }, { "epoch": 7.94, "learning_rate": 9.538712870884948e-06, "loss": 0.0447, "step": 7820 }, { "epoch": 7.94, "learning_rate": 9.536523049844855e-06, "loss": 0.051, "step": 7821 }, { "epoch": 7.94, "learning_rate": 9.534333251077564e-06, "loss": 0.0393, "step": 7822 }, { "epoch": 7.94, "learning_rate": 9.53214347468831e-06, "loss": 0.0508, "step": 7823 }, { "epoch": 7.94, "learning_rate": 9.529953720782326e-06, "loss": 0.0655, "step": 7824 }, { "epoch": 7.94, "learning_rate": 9.52776398946484e-06, "loss": 0.0411, "step": 7825 }, { "epoch": 7.94, "learning_rate": 9.52557428084108e-06, "loss": 0.0625, "step": 7826 }, { "epoch": 7.94, "learning_rate": 9.523384595016277e-06, "loss": 0.0469, "step": 7827 }, { "epoch": 7.94, "learning_rate": 9.521194932095658e-06, "loss": 0.0624, "step": 7828 }, { "epoch": 7.94, "learning_rate": 9.519005292184454e-06, "loss": 0.0447, "step": 7829 }, { "epoch": 7.95, "learning_rate": 9.516815675387883e-06, "loss": 0.0599, "step": 7830 }, { "epoch": 7.95, "learning_rate": 9.514626081811172e-06, "loss": 0.0458, "step": 7831 }, { "epoch": 7.95, "learning_rate": 9.51243651155954e-06, "loss": 0.054, "step": 7832 }, { "epoch": 7.95, "learning_rate": 9.510246964738216e-06, "loss": 0.046, "step": 7833 }, { "epoch": 7.95, "learning_rate": 9.508057441452417e-06, "loss": 0.046, "step": 7834 }, { "epoch": 7.95, "learning_rate": 9.505867941807362e-06, "loss": 0.0491, "step": 7835 }, { "epoch": 7.95, "learning_rate": 9.503678465908265e-06, "loss": 0.0579, "step": 7836 }, { "epoch": 7.95, "learning_rate": 9.501489013860353e-06, "loss": 0.056, "step": 7837 }, { "epoch": 7.95, "learning_rate": 9.499299585768834e-06, "loss": 0.0479, "step": 7838 }, { "epoch": 7.95, "learning_rate": 9.497110181738928e-06, "loss": 0.0703, "step": 7839 }, { "epoch": 7.96, "learning_rate": 9.494920801875848e-06, "loss": 0.0533, "step": 7840 }, { "epoch": 7.96, "learning_rate": 9.492731446284805e-06, "loss": 0.0483, "step": 7841 }, { "epoch": 7.96, "learning_rate": 9.490542115071011e-06, "loss": 0.0495, "step": 7842 }, { "epoch": 7.96, "learning_rate": 9.488352808339675e-06, "loss": 0.0531, "step": 7843 }, { "epoch": 7.96, "learning_rate": 9.486163526196006e-06, "loss": 0.0554, "step": 7844 }, { "epoch": 7.96, "learning_rate": 9.483974268745214e-06, "loss": 0.0464, "step": 7845 }, { "epoch": 7.96, "learning_rate": 9.481785036092506e-06, "loss": 0.0525, "step": 7846 }, { "epoch": 7.96, "learning_rate": 9.479595828343089e-06, "loss": 0.0424, "step": 7847 }, { "epoch": 7.96, "learning_rate": 9.477406645602163e-06, "loss": 0.0542, "step": 7848 }, { "epoch": 7.96, "learning_rate": 9.47521748797493e-06, "loss": 0.0498, "step": 7849 }, { "epoch": 7.97, "learning_rate": 9.473028355566598e-06, "loss": 0.051, "step": 7850 }, { "epoch": 7.97, "learning_rate": 9.470839248482366e-06, "loss": 0.0431, "step": 7851 }, { "epoch": 7.97, "learning_rate": 9.468650166827428e-06, "loss": 0.0353, "step": 7852 }, { "epoch": 7.97, "learning_rate": 9.466461110706991e-06, "loss": 0.051, "step": 7853 }, { "epoch": 7.97, "learning_rate": 9.464272080226246e-06, "loss": 0.0536, "step": 7854 }, { "epoch": 7.97, "learning_rate": 9.462083075490389e-06, "loss": 0.053, "step": 7855 }, { "epoch": 7.97, "learning_rate": 9.459894096604618e-06, "loss": 0.0433, "step": 7856 }, { "epoch": 7.97, "learning_rate": 9.457705143674126e-06, "loss": 0.0492, "step": 7857 }, { "epoch": 7.97, "learning_rate": 9.455516216804102e-06, "loss": 0.0598, "step": 7858 }, { "epoch": 7.97, "learning_rate": 9.453327316099737e-06, "loss": 0.0576, "step": 7859 }, { "epoch": 7.98, "learning_rate": 9.451138441666224e-06, "loss": 0.0429, "step": 7860 }, { "epoch": 7.98, "learning_rate": 9.448949593608746e-06, "loss": 0.0446, "step": 7861 }, { "epoch": 7.98, "learning_rate": 9.446760772032496e-06, "loss": 0.0567, "step": 7862 }, { "epoch": 7.98, "learning_rate": 9.444571977042659e-06, "loss": 0.0467, "step": 7863 }, { "epoch": 7.98, "learning_rate": 9.442383208744415e-06, "loss": 0.0458, "step": 7864 }, { "epoch": 7.98, "learning_rate": 9.440194467242952e-06, "loss": 0.0413, "step": 7865 }, { "epoch": 7.98, "learning_rate": 9.438005752643448e-06, "loss": 0.0485, "step": 7866 }, { "epoch": 7.98, "learning_rate": 9.435817065051087e-06, "loss": 0.055, "step": 7867 }, { "epoch": 7.98, "learning_rate": 9.433628404571047e-06, "loss": 0.0554, "step": 7868 }, { "epoch": 7.98, "learning_rate": 9.431439771308503e-06, "loss": 0.0472, "step": 7869 }, { "epoch": 7.99, "learning_rate": 9.429251165368638e-06, "loss": 0.0435, "step": 7870 }, { "epoch": 7.99, "learning_rate": 9.427062586856624e-06, "loss": 0.0604, "step": 7871 }, { "epoch": 7.99, "learning_rate": 9.424874035877633e-06, "loss": 0.051, "step": 7872 }, { "epoch": 7.99, "learning_rate": 9.422685512536844e-06, "loss": 0.0398, "step": 7873 }, { "epoch": 7.99, "learning_rate": 9.420497016939423e-06, "loss": 0.0481, "step": 7874 }, { "epoch": 7.99, "learning_rate": 9.418308549190543e-06, "loss": 0.0539, "step": 7875 }, { "epoch": 7.99, "learning_rate": 9.416120109395368e-06, "loss": 0.0503, "step": 7876 }, { "epoch": 7.99, "learning_rate": 9.41393169765907e-06, "loss": 0.0457, "step": 7877 }, { "epoch": 7.99, "learning_rate": 9.411743314086813e-06, "loss": 0.0438, "step": 7878 }, { "epoch": 7.99, "learning_rate": 9.409554958783765e-06, "loss": 0.0477, "step": 7879 }, { "epoch": 8.0, "learning_rate": 9.407366631855089e-06, "loss": 0.0402, "step": 7880 }, { "epoch": 8.0, "learning_rate": 9.405178333405945e-06, "loss": 0.0476, "step": 7881 }, { "epoch": 8.0, "learning_rate": 9.40299006354149e-06, "loss": 0.0494, "step": 7882 }, { "epoch": 8.0, "learning_rate": 9.40080182236689e-06, "loss": 0.0465, "step": 7883 }, { "epoch": 8.0, "learning_rate": 9.3986136099873e-06, "loss": 0.0458, "step": 7884 }, { "epoch": 8.0, "learning_rate": 9.396425426507878e-06, "loss": 0.0321, "step": 7885 }, { "epoch": 8.0, "learning_rate": 9.394237272033772e-06, "loss": 0.0282, "step": 7886 }, { "epoch": 8.0, "learning_rate": 9.392049146670146e-06, "loss": 0.027, "step": 7887 }, { "epoch": 8.0, "learning_rate": 9.389861050522147e-06, "loss": 0.0441, "step": 7888 }, { "epoch": 8.01, "learning_rate": 9.387672983694927e-06, "loss": 0.0393, "step": 7889 }, { "epoch": 8.01, "learning_rate": 9.385484946293636e-06, "loss": 0.0406, "step": 7890 }, { "epoch": 8.01, "learning_rate": 9.383296938423423e-06, "loss": 0.0312, "step": 7891 }, { "epoch": 8.01, "learning_rate": 9.381108960189431e-06, "loss": 0.028, "step": 7892 }, { "epoch": 8.01, "learning_rate": 9.378921011696807e-06, "loss": 0.041, "step": 7893 }, { "epoch": 8.01, "learning_rate": 9.376733093050694e-06, "loss": 0.0358, "step": 7894 }, { "epoch": 8.01, "learning_rate": 9.374545204356237e-06, "loss": 0.0323, "step": 7895 }, { "epoch": 8.01, "learning_rate": 9.372357345718577e-06, "loss": 0.03, "step": 7896 }, { "epoch": 8.01, "learning_rate": 9.370169517242852e-06, "loss": 0.0258, "step": 7897 }, { "epoch": 8.01, "learning_rate": 9.367981719034201e-06, "loss": 0.0307, "step": 7898 }, { "epoch": 8.02, "learning_rate": 9.365793951197756e-06, "loss": 0.0356, "step": 7899 }, { "epoch": 8.02, "learning_rate": 9.36360621383866e-06, "loss": 0.0372, "step": 7900 }, { "epoch": 8.02, "learning_rate": 9.361418507062041e-06, "loss": 0.0399, "step": 7901 }, { "epoch": 8.02, "learning_rate": 9.35923083097303e-06, "loss": 0.0267, "step": 7902 }, { "epoch": 8.02, "learning_rate": 9.357043185676764e-06, "loss": 0.0284, "step": 7903 }, { "epoch": 8.02, "learning_rate": 9.354855571278369e-06, "loss": 0.0209, "step": 7904 }, { "epoch": 8.02, "learning_rate": 9.352667987882972e-06, "loss": 0.0281, "step": 7905 }, { "epoch": 8.02, "learning_rate": 9.350480435595702e-06, "loss": 0.0291, "step": 7906 }, { "epoch": 8.02, "learning_rate": 9.348292914521681e-06, "loss": 0.0392, "step": 7907 }, { "epoch": 8.02, "learning_rate": 9.346105424766032e-06, "loss": 0.0486, "step": 7908 }, { "epoch": 8.03, "learning_rate": 9.343917966433878e-06, "loss": 0.0358, "step": 7909 }, { "epoch": 8.03, "learning_rate": 9.34173053963034e-06, "loss": 0.0246, "step": 7910 }, { "epoch": 8.03, "learning_rate": 9.339543144460533e-06, "loss": 0.0409, "step": 7911 }, { "epoch": 8.03, "learning_rate": 9.33735578102958e-06, "loss": 0.0229, "step": 7912 }, { "epoch": 8.03, "learning_rate": 9.335168449442596e-06, "loss": 0.0258, "step": 7913 }, { "epoch": 8.03, "learning_rate": 9.33298114980469e-06, "loss": 0.0311, "step": 7914 }, { "epoch": 8.03, "learning_rate": 9.330793882220981e-06, "loss": 0.029, "step": 7915 }, { "epoch": 8.03, "learning_rate": 9.328606646796575e-06, "loss": 0.039, "step": 7916 }, { "epoch": 8.03, "learning_rate": 9.326419443636582e-06, "loss": 0.0293, "step": 7917 }, { "epoch": 8.03, "learning_rate": 9.324232272846115e-06, "loss": 0.0398, "step": 7918 }, { "epoch": 8.04, "learning_rate": 9.322045134530272e-06, "loss": 0.0414, "step": 7919 }, { "epoch": 8.04, "learning_rate": 9.319858028794167e-06, "loss": 0.038, "step": 7920 }, { "epoch": 8.04, "learning_rate": 9.317670955742899e-06, "loss": 0.0246, "step": 7921 }, { "epoch": 8.04, "learning_rate": 9.315483915481567e-06, "loss": 0.0356, "step": 7922 }, { "epoch": 8.04, "learning_rate": 9.313296908115278e-06, "loss": 0.0364, "step": 7923 }, { "epoch": 8.04, "learning_rate": 9.311109933749127e-06, "loss": 0.0243, "step": 7924 }, { "epoch": 8.04, "learning_rate": 9.30892299248821e-06, "loss": 0.0318, "step": 7925 }, { "epoch": 8.04, "learning_rate": 9.306736084437621e-06, "loss": 0.0313, "step": 7926 }, { "epoch": 8.04, "learning_rate": 9.304549209702459e-06, "loss": 0.0303, "step": 7927 }, { "epoch": 8.04, "learning_rate": 9.302362368387809e-06, "loss": 0.039, "step": 7928 }, { "epoch": 8.05, "learning_rate": 9.30017556059877e-06, "loss": 0.0374, "step": 7929 }, { "epoch": 8.05, "learning_rate": 9.297988786440429e-06, "loss": 0.0259, "step": 7930 }, { "epoch": 8.05, "learning_rate": 9.29580204601787e-06, "loss": 0.0229, "step": 7931 }, { "epoch": 8.05, "learning_rate": 9.293615339436178e-06, "loss": 0.0341, "step": 7932 }, { "epoch": 8.05, "learning_rate": 9.291428666800442e-06, "loss": 0.0324, "step": 7933 }, { "epoch": 8.05, "learning_rate": 9.289242028215742e-06, "loss": 0.0342, "step": 7934 }, { "epoch": 8.05, "learning_rate": 9.287055423787158e-06, "loss": 0.0369, "step": 7935 }, { "epoch": 8.05, "learning_rate": 9.284868853619766e-06, "loss": 0.0393, "step": 7936 }, { "epoch": 8.05, "learning_rate": 9.282682317818653e-06, "loss": 0.0344, "step": 7937 }, { "epoch": 8.05, "learning_rate": 9.28049581648889e-06, "loss": 0.0325, "step": 7938 }, { "epoch": 8.06, "learning_rate": 9.278309349735547e-06, "loss": 0.0314, "step": 7939 }, { "epoch": 8.06, "learning_rate": 9.276122917663704e-06, "loss": 0.0361, "step": 7940 }, { "epoch": 8.06, "learning_rate": 9.273936520378428e-06, "loss": 0.0343, "step": 7941 }, { "epoch": 8.06, "learning_rate": 9.27175015798479e-06, "loss": 0.0277, "step": 7942 }, { "epoch": 8.06, "learning_rate": 9.269563830587854e-06, "loss": 0.0282, "step": 7943 }, { "epoch": 8.06, "learning_rate": 9.267377538292688e-06, "loss": 0.0287, "step": 7944 }, { "epoch": 8.06, "learning_rate": 9.265191281204357e-06, "loss": 0.026, "step": 7945 }, { "epoch": 8.06, "learning_rate": 9.263005059427924e-06, "loss": 0.0331, "step": 7946 }, { "epoch": 8.06, "learning_rate": 9.260818873068452e-06, "loss": 0.03, "step": 7947 }, { "epoch": 8.06, "learning_rate": 9.258632722230995e-06, "loss": 0.0298, "step": 7948 }, { "epoch": 8.07, "learning_rate": 9.256446607020611e-06, "loss": 0.0227, "step": 7949 }, { "epoch": 8.07, "learning_rate": 9.25426052754236e-06, "loss": 0.0276, "step": 7950 }, { "epoch": 8.07, "learning_rate": 9.252074483901293e-06, "loss": 0.0342, "step": 7951 }, { "epoch": 8.07, "learning_rate": 9.249888476202465e-06, "loss": 0.0249, "step": 7952 }, { "epoch": 8.07, "learning_rate": 9.247702504550918e-06, "loss": 0.0308, "step": 7953 }, { "epoch": 8.07, "learning_rate": 9.245516569051711e-06, "loss": 0.032, "step": 7954 }, { "epoch": 8.07, "learning_rate": 9.243330669809887e-06, "loss": 0.027, "step": 7955 }, { "epoch": 8.07, "learning_rate": 9.241144806930492e-06, "loss": 0.0292, "step": 7956 }, { "epoch": 8.07, "learning_rate": 9.238958980518571e-06, "loss": 0.0226, "step": 7957 }, { "epoch": 8.08, "learning_rate": 9.236773190679163e-06, "loss": 0.0339, "step": 7958 }, { "epoch": 8.08, "learning_rate": 9.23458743751731e-06, "loss": 0.0314, "step": 7959 }, { "epoch": 8.08, "learning_rate": 9.232401721138048e-06, "loss": 0.0324, "step": 7960 }, { "epoch": 8.08, "learning_rate": 9.230216041646412e-06, "loss": 0.038, "step": 7961 }, { "epoch": 8.08, "learning_rate": 9.228030399147443e-06, "loss": 0.0303, "step": 7962 }, { "epoch": 8.08, "learning_rate": 9.225844793746175e-06, "loss": 0.0264, "step": 7963 }, { "epoch": 8.08, "learning_rate": 9.223659225547632e-06, "loss": 0.0383, "step": 7964 }, { "epoch": 8.08, "learning_rate": 9.221473694656849e-06, "loss": 0.0323, "step": 7965 }, { "epoch": 8.08, "learning_rate": 9.219288201178848e-06, "loss": 0.0293, "step": 7966 }, { "epoch": 8.08, "learning_rate": 9.21710274521866e-06, "loss": 0.0356, "step": 7967 }, { "epoch": 8.09, "learning_rate": 9.214917326881307e-06, "loss": 0.0364, "step": 7968 }, { "epoch": 8.09, "learning_rate": 9.212731946271809e-06, "loss": 0.029, "step": 7969 }, { "epoch": 8.09, "learning_rate": 9.210546603495193e-06, "loss": 0.0303, "step": 7970 }, { "epoch": 8.09, "learning_rate": 9.208361298656473e-06, "loss": 0.0286, "step": 7971 }, { "epoch": 8.09, "learning_rate": 9.206176031860664e-06, "loss": 0.0327, "step": 7972 }, { "epoch": 8.09, "learning_rate": 9.203990803212787e-06, "loss": 0.0305, "step": 7973 }, { "epoch": 8.09, "learning_rate": 9.20180561281785e-06, "loss": 0.0364, "step": 7974 }, { "epoch": 8.09, "learning_rate": 9.199620460780867e-06, "loss": 0.0337, "step": 7975 }, { "epoch": 8.09, "learning_rate": 9.197435347206844e-06, "loss": 0.0319, "step": 7976 }, { "epoch": 8.09, "learning_rate": 9.195250272200792e-06, "loss": 0.0371, "step": 7977 }, { "epoch": 8.1, "learning_rate": 9.193065235867713e-06, "loss": 0.0269, "step": 7978 }, { "epoch": 8.1, "learning_rate": 9.190880238312615e-06, "loss": 0.0264, "step": 7979 }, { "epoch": 8.1, "learning_rate": 9.188695279640501e-06, "loss": 0.0266, "step": 7980 }, { "epoch": 8.1, "learning_rate": 9.186510359956368e-06, "loss": 0.0282, "step": 7981 }, { "epoch": 8.1, "learning_rate": 9.184325479365214e-06, "loss": 0.0369, "step": 7982 }, { "epoch": 8.1, "learning_rate": 9.182140637972039e-06, "loss": 0.0328, "step": 7983 }, { "epoch": 8.1, "learning_rate": 9.179955835881833e-06, "loss": 0.0352, "step": 7984 }, { "epoch": 8.1, "learning_rate": 9.177771073199592e-06, "loss": 0.0332, "step": 7985 }, { "epoch": 8.1, "learning_rate": 9.175586350030301e-06, "loss": 0.0276, "step": 7986 }, { "epoch": 8.1, "learning_rate": 9.173401666478957e-06, "loss": 0.0289, "step": 7987 }, { "epoch": 8.11, "learning_rate": 9.171217022650543e-06, "loss": 0.0312, "step": 7988 }, { "epoch": 8.11, "learning_rate": 9.169032418650043e-06, "loss": 0.0438, "step": 7989 }, { "epoch": 8.11, "learning_rate": 9.166847854582444e-06, "loss": 0.0321, "step": 7990 }, { "epoch": 8.11, "learning_rate": 9.164663330552723e-06, "loss": 0.0279, "step": 7991 }, { "epoch": 8.11, "learning_rate": 9.16247884666586e-06, "loss": 0.0356, "step": 7992 }, { "epoch": 8.11, "learning_rate": 9.160294403026834e-06, "loss": 0.0319, "step": 7993 }, { "epoch": 8.11, "learning_rate": 9.158109999740617e-06, "loss": 0.0433, "step": 7994 }, { "epoch": 8.11, "learning_rate": 9.155925636912186e-06, "loss": 0.0402, "step": 7995 }, { "epoch": 8.11, "learning_rate": 9.153741314646515e-06, "loss": 0.0374, "step": 7996 }, { "epoch": 8.11, "learning_rate": 9.151557033048569e-06, "loss": 0.0303, "step": 7997 }, { "epoch": 8.12, "learning_rate": 9.149372792223316e-06, "loss": 0.0259, "step": 7998 }, { "epoch": 8.12, "learning_rate": 9.147188592275721e-06, "loss": 0.0304, "step": 7999 }, { "epoch": 8.12, "learning_rate": 9.145004433310752e-06, "loss": 0.0304, "step": 8000 }, { "epoch": 8.12, "learning_rate": 9.142820315433367e-06, "loss": 0.0349, "step": 8001 }, { "epoch": 8.12, "learning_rate": 9.140636238748527e-06, "loss": 0.0269, "step": 8002 }, { "epoch": 8.12, "learning_rate": 9.138452203361185e-06, "loss": 0.0347, "step": 8003 }, { "epoch": 8.12, "learning_rate": 9.136268209376306e-06, "loss": 0.0368, "step": 8004 }, { "epoch": 8.12, "learning_rate": 9.134084256898836e-06, "loss": 0.0274, "step": 8005 }, { "epoch": 8.12, "learning_rate": 9.131900346033733e-06, "loss": 0.0264, "step": 8006 }, { "epoch": 8.12, "learning_rate": 9.129716476885943e-06, "loss": 0.0331, "step": 8007 }, { "epoch": 8.13, "learning_rate": 9.127532649560415e-06, "loss": 0.0231, "step": 8008 }, { "epoch": 8.13, "learning_rate": 9.125348864162093e-06, "loss": 0.0336, "step": 8009 }, { "epoch": 8.13, "learning_rate": 9.123165120795921e-06, "loss": 0.0416, "step": 8010 }, { "epoch": 8.13, "learning_rate": 9.12098141956684e-06, "loss": 0.0334, "step": 8011 }, { "epoch": 8.13, "learning_rate": 9.118797760579794e-06, "loss": 0.0359, "step": 8012 }, { "epoch": 8.13, "learning_rate": 9.11661414393972e-06, "loss": 0.0438, "step": 8013 }, { "epoch": 8.13, "learning_rate": 9.114430569751553e-06, "loss": 0.0335, "step": 8014 }, { "epoch": 8.13, "learning_rate": 9.112247038120224e-06, "loss": 0.0271, "step": 8015 }, { "epoch": 8.13, "learning_rate": 9.110063549150663e-06, "loss": 0.0327, "step": 8016 }, { "epoch": 8.13, "learning_rate": 9.107880102947807e-06, "loss": 0.0313, "step": 8017 }, { "epoch": 8.14, "learning_rate": 9.105696699616577e-06, "loss": 0.0349, "step": 8018 }, { "epoch": 8.14, "learning_rate": 9.103513339261899e-06, "loss": 0.0286, "step": 8019 }, { "epoch": 8.14, "learning_rate": 9.101330021988698e-06, "loss": 0.0329, "step": 8020 }, { "epoch": 8.14, "learning_rate": 9.0991467479019e-06, "loss": 0.0347, "step": 8021 }, { "epoch": 8.14, "learning_rate": 9.096963517106414e-06, "loss": 0.0359, "step": 8022 }, { "epoch": 8.14, "learning_rate": 9.094780329707166e-06, "loss": 0.0264, "step": 8023 }, { "epoch": 8.14, "learning_rate": 9.092597185809067e-06, "loss": 0.0356, "step": 8024 }, { "epoch": 8.14, "learning_rate": 9.090414085517031e-06, "loss": 0.0278, "step": 8025 }, { "epoch": 8.14, "learning_rate": 9.088231028935965e-06, "loss": 0.0375, "step": 8026 }, { "epoch": 8.15, "learning_rate": 9.086048016170786e-06, "loss": 0.0347, "step": 8027 }, { "epoch": 8.15, "learning_rate": 9.08386504732639e-06, "loss": 0.0365, "step": 8028 }, { "epoch": 8.15, "learning_rate": 9.081682122507692e-06, "loss": 0.0356, "step": 8029 }, { "epoch": 8.15, "learning_rate": 9.079499241819588e-06, "loss": 0.033, "step": 8030 }, { "epoch": 8.15, "learning_rate": 9.07731640536698e-06, "loss": 0.0348, "step": 8031 }, { "epoch": 8.15, "learning_rate": 9.075133613254767e-06, "loss": 0.0484, "step": 8032 }, { "epoch": 8.15, "learning_rate": 9.072950865587847e-06, "loss": 0.0281, "step": 8033 }, { "epoch": 8.15, "learning_rate": 9.070768162471108e-06, "loss": 0.0282, "step": 8034 }, { "epoch": 8.15, "learning_rate": 9.068585504009446e-06, "loss": 0.0315, "step": 8035 }, { "epoch": 8.15, "learning_rate": 9.066402890307746e-06, "loss": 0.0331, "step": 8036 }, { "epoch": 8.16, "learning_rate": 9.064220321470903e-06, "loss": 0.0363, "step": 8037 }, { "epoch": 8.16, "learning_rate": 9.0620377976038e-06, "loss": 0.0338, "step": 8038 }, { "epoch": 8.16, "learning_rate": 9.059855318811314e-06, "loss": 0.0251, "step": 8039 }, { "epoch": 8.16, "learning_rate": 9.057672885198334e-06, "loss": 0.0299, "step": 8040 }, { "epoch": 8.16, "learning_rate": 9.055490496869734e-06, "loss": 0.0332, "step": 8041 }, { "epoch": 8.16, "learning_rate": 9.053308153930393e-06, "loss": 0.0405, "step": 8042 }, { "epoch": 8.16, "learning_rate": 9.051125856485183e-06, "loss": 0.0292, "step": 8043 }, { "epoch": 8.16, "learning_rate": 9.04894360463898e-06, "loss": 0.0404, "step": 8044 }, { "epoch": 8.16, "learning_rate": 9.046761398496646e-06, "loss": 0.0296, "step": 8045 }, { "epoch": 8.16, "learning_rate": 9.04457923816306e-06, "loss": 0.034, "step": 8046 }, { "epoch": 8.17, "learning_rate": 9.042397123743082e-06, "loss": 0.0402, "step": 8047 }, { "epoch": 8.17, "learning_rate": 9.040215055341578e-06, "loss": 0.0378, "step": 8048 }, { "epoch": 8.17, "learning_rate": 9.038033033063402e-06, "loss": 0.0363, "step": 8049 }, { "epoch": 8.17, "learning_rate": 9.035851057013422e-06, "loss": 0.038, "step": 8050 }, { "epoch": 8.17, "learning_rate": 9.033669127296492e-06, "loss": 0.0348, "step": 8051 }, { "epoch": 8.17, "learning_rate": 9.031487244017464e-06, "loss": 0.0368, "step": 8052 }, { "epoch": 8.17, "learning_rate": 9.029305407281189e-06, "loss": 0.0295, "step": 8053 }, { "epoch": 8.17, "learning_rate": 9.027123617192524e-06, "loss": 0.0345, "step": 8054 }, { "epoch": 8.17, "learning_rate": 9.02494187385631e-06, "loss": 0.0476, "step": 8055 }, { "epoch": 8.17, "learning_rate": 9.022760177377398e-06, "loss": 0.0375, "step": 8056 }, { "epoch": 8.18, "learning_rate": 9.02057852786063e-06, "loss": 0.0312, "step": 8057 }, { "epoch": 8.18, "learning_rate": 9.018396925410848e-06, "loss": 0.0339, "step": 8058 }, { "epoch": 8.18, "learning_rate": 9.016215370132886e-06, "loss": 0.0376, "step": 8059 }, { "epoch": 8.18, "learning_rate": 9.014033862131583e-06, "loss": 0.0341, "step": 8060 }, { "epoch": 8.18, "learning_rate": 9.011852401511774e-06, "loss": 0.0343, "step": 8061 }, { "epoch": 8.18, "learning_rate": 9.009670988378291e-06, "loss": 0.0267, "step": 8062 }, { "epoch": 8.18, "learning_rate": 9.007489622835965e-06, "loss": 0.0423, "step": 8063 }, { "epoch": 8.18, "learning_rate": 9.005308304989624e-06, "loss": 0.0281, "step": 8064 }, { "epoch": 8.18, "learning_rate": 9.003127034944091e-06, "loss": 0.0335, "step": 8065 }, { "epoch": 8.18, "learning_rate": 9.000945812804189e-06, "loss": 0.0306, "step": 8066 }, { "epoch": 8.19, "learning_rate": 8.998764638674739e-06, "loss": 0.03, "step": 8067 }, { "epoch": 8.19, "learning_rate": 8.99658351266056e-06, "loss": 0.0388, "step": 8068 }, { "epoch": 8.19, "learning_rate": 8.994402434866468e-06, "loss": 0.0375, "step": 8069 }, { "epoch": 8.19, "learning_rate": 8.992221405397273e-06, "loss": 0.0331, "step": 8070 }, { "epoch": 8.19, "learning_rate": 8.99004042435779e-06, "loss": 0.0272, "step": 8071 }, { "epoch": 8.19, "learning_rate": 8.987859491852829e-06, "loss": 0.0311, "step": 8072 }, { "epoch": 8.19, "learning_rate": 8.985678607987196e-06, "loss": 0.0261, "step": 8073 }, { "epoch": 8.19, "learning_rate": 8.983497772865695e-06, "loss": 0.0276, "step": 8074 }, { "epoch": 8.19, "learning_rate": 8.981316986593127e-06, "loss": 0.0429, "step": 8075 }, { "epoch": 8.19, "learning_rate": 8.979136249274289e-06, "loss": 0.0325, "step": 8076 }, { "epoch": 8.2, "learning_rate": 8.976955561013984e-06, "loss": 0.0399, "step": 8077 }, { "epoch": 8.2, "learning_rate": 8.974774921917e-06, "loss": 0.0385, "step": 8078 }, { "epoch": 8.2, "learning_rate": 8.972594332088138e-06, "loss": 0.0293, "step": 8079 }, { "epoch": 8.2, "learning_rate": 8.970413791632187e-06, "loss": 0.0396, "step": 8080 }, { "epoch": 8.2, "learning_rate": 8.96823330065393e-06, "loss": 0.0377, "step": 8081 }, { "epoch": 8.2, "learning_rate": 8.966052859258153e-06, "loss": 0.0397, "step": 8082 }, { "epoch": 8.2, "learning_rate": 8.96387246754964e-06, "loss": 0.0309, "step": 8083 }, { "epoch": 8.2, "learning_rate": 8.961692125633174e-06, "loss": 0.0392, "step": 8084 }, { "epoch": 8.2, "learning_rate": 8.959511833613533e-06, "loss": 0.0274, "step": 8085 }, { "epoch": 8.2, "learning_rate": 8.957331591595486e-06, "loss": 0.0272, "step": 8086 }, { "epoch": 8.21, "learning_rate": 8.955151399683815e-06, "loss": 0.0289, "step": 8087 }, { "epoch": 8.21, "learning_rate": 8.952971257983289e-06, "loss": 0.0223, "step": 8088 }, { "epoch": 8.21, "learning_rate": 8.950791166598675e-06, "loss": 0.039, "step": 8089 }, { "epoch": 8.21, "learning_rate": 8.94861112563474e-06, "loss": 0.0355, "step": 8090 }, { "epoch": 8.21, "learning_rate": 8.94643113519625e-06, "loss": 0.0355, "step": 8091 }, { "epoch": 8.21, "learning_rate": 8.944251195387962e-06, "loss": 0.0264, "step": 8092 }, { "epoch": 8.21, "learning_rate": 8.942071306314637e-06, "loss": 0.0324, "step": 8093 }, { "epoch": 8.21, "learning_rate": 8.939891468081033e-06, "loss": 0.0287, "step": 8094 }, { "epoch": 8.21, "learning_rate": 8.937711680791899e-06, "loss": 0.036, "step": 8095 }, { "epoch": 8.22, "learning_rate": 8.935531944551994e-06, "loss": 0.0446, "step": 8096 }, { "epoch": 8.22, "learning_rate": 8.933352259466065e-06, "loss": 0.0413, "step": 8097 }, { "epoch": 8.22, "learning_rate": 8.931172625638857e-06, "loss": 0.0447, "step": 8098 }, { "epoch": 8.22, "learning_rate": 8.928993043175112e-06, "loss": 0.0341, "step": 8099 }, { "epoch": 8.22, "learning_rate": 8.926813512179576e-06, "loss": 0.026, "step": 8100 }, { "epoch": 8.22, "learning_rate": 8.924634032756989e-06, "loss": 0.0344, "step": 8101 }, { "epoch": 8.22, "learning_rate": 8.922454605012084e-06, "loss": 0.0381, "step": 8102 }, { "epoch": 8.22, "learning_rate": 8.920275229049593e-06, "loss": 0.0356, "step": 8103 }, { "epoch": 8.22, "learning_rate": 8.918095904974256e-06, "loss": 0.0414, "step": 8104 }, { "epoch": 8.22, "learning_rate": 8.915916632890798e-06, "loss": 0.033, "step": 8105 }, { "epoch": 8.23, "learning_rate": 8.913737412903947e-06, "loss": 0.0412, "step": 8106 }, { "epoch": 8.23, "learning_rate": 8.911558245118427e-06, "loss": 0.0318, "step": 8107 }, { "epoch": 8.23, "learning_rate": 8.909379129638959e-06, "loss": 0.0508, "step": 8108 }, { "epoch": 8.23, "learning_rate": 8.907200066570264e-06, "loss": 0.0378, "step": 8109 }, { "epoch": 8.23, "learning_rate": 8.905021056017055e-06, "loss": 0.0212, "step": 8110 }, { "epoch": 8.23, "learning_rate": 8.902842098084047e-06, "loss": 0.0427, "step": 8111 }, { "epoch": 8.23, "learning_rate": 8.900663192875956e-06, "loss": 0.0349, "step": 8112 }, { "epoch": 8.23, "learning_rate": 8.898484340497491e-06, "loss": 0.0356, "step": 8113 }, { "epoch": 8.23, "learning_rate": 8.896305541053358e-06, "loss": 0.0267, "step": 8114 }, { "epoch": 8.23, "learning_rate": 8.894126794648259e-06, "loss": 0.0427, "step": 8115 }, { "epoch": 8.24, "learning_rate": 8.891948101386894e-06, "loss": 0.0353, "step": 8116 }, { "epoch": 8.24, "learning_rate": 8.889769461373967e-06, "loss": 0.0346, "step": 8117 }, { "epoch": 8.24, "learning_rate": 8.887590874714172e-06, "loss": 0.028, "step": 8118 }, { "epoch": 8.24, "learning_rate": 8.885412341512204e-06, "loss": 0.0291, "step": 8119 }, { "epoch": 8.24, "learning_rate": 8.88323386187275e-06, "loss": 0.0321, "step": 8120 }, { "epoch": 8.24, "learning_rate": 8.881055435900504e-06, "loss": 0.0377, "step": 8121 }, { "epoch": 8.24, "learning_rate": 8.878877063700152e-06, "loss": 0.038, "step": 8122 }, { "epoch": 8.24, "learning_rate": 8.876698745376375e-06, "loss": 0.0375, "step": 8123 }, { "epoch": 8.24, "learning_rate": 8.874520481033855e-06, "loss": 0.0463, "step": 8124 }, { "epoch": 8.24, "learning_rate": 8.872342270777273e-06, "loss": 0.0377, "step": 8125 }, { "epoch": 8.25, "learning_rate": 8.8701641147113e-06, "loss": 0.0378, "step": 8126 }, { "epoch": 8.25, "learning_rate": 8.867986012940614e-06, "loss": 0.0379, "step": 8127 }, { "epoch": 8.25, "learning_rate": 8.865807965569879e-06, "loss": 0.0389, "step": 8128 }, { "epoch": 8.25, "learning_rate": 8.863629972703772e-06, "loss": 0.033, "step": 8129 }, { "epoch": 8.25, "learning_rate": 8.861452034446953e-06, "loss": 0.0235, "step": 8130 }, { "epoch": 8.25, "learning_rate": 8.859274150904087e-06, "loss": 0.0302, "step": 8131 }, { "epoch": 8.25, "learning_rate": 8.857096322179833e-06, "loss": 0.0319, "step": 8132 }, { "epoch": 8.25, "learning_rate": 8.854918548378845e-06, "loss": 0.0435, "step": 8133 }, { "epoch": 8.25, "learning_rate": 8.852740829605785e-06, "loss": 0.0503, "step": 8134 }, { "epoch": 8.25, "learning_rate": 8.850563165965301e-06, "loss": 0.0544, "step": 8135 }, { "epoch": 8.26, "learning_rate": 8.848385557562038e-06, "loss": 0.0321, "step": 8136 }, { "epoch": 8.26, "learning_rate": 8.846208004500654e-06, "loss": 0.0294, "step": 8137 }, { "epoch": 8.26, "learning_rate": 8.844030506885787e-06, "loss": 0.03, "step": 8138 }, { "epoch": 8.26, "learning_rate": 8.841853064822077e-06, "loss": 0.0516, "step": 8139 }, { "epoch": 8.26, "learning_rate": 8.839675678414166e-06, "loss": 0.0335, "step": 8140 }, { "epoch": 8.26, "learning_rate": 8.837498347766689e-06, "loss": 0.0337, "step": 8141 }, { "epoch": 8.26, "learning_rate": 8.835321072984281e-06, "loss": 0.0371, "step": 8142 }, { "epoch": 8.26, "learning_rate": 8.833143854171571e-06, "loss": 0.0299, "step": 8143 }, { "epoch": 8.26, "learning_rate": 8.830966691433189e-06, "loss": 0.0318, "step": 8144 }, { "epoch": 8.26, "learning_rate": 8.828789584873754e-06, "loss": 0.0293, "step": 8145 }, { "epoch": 8.27, "learning_rate": 8.826612534597898e-06, "loss": 0.0426, "step": 8146 }, { "epoch": 8.27, "learning_rate": 8.824435540710238e-06, "loss": 0.0305, "step": 8147 }, { "epoch": 8.27, "learning_rate": 8.822258603315392e-06, "loss": 0.0333, "step": 8148 }, { "epoch": 8.27, "learning_rate": 8.82008172251797e-06, "loss": 0.05, "step": 8149 }, { "epoch": 8.27, "learning_rate": 8.817904898422592e-06, "loss": 0.0332, "step": 8150 }, { "epoch": 8.27, "learning_rate": 8.815728131133862e-06, "loss": 0.04, "step": 8151 }, { "epoch": 8.27, "learning_rate": 8.813551420756387e-06, "loss": 0.0373, "step": 8152 }, { "epoch": 8.27, "learning_rate": 8.811374767394766e-06, "loss": 0.0398, "step": 8153 }, { "epoch": 8.27, "learning_rate": 8.809198171153608e-06, "loss": 0.0305, "step": 8154 }, { "epoch": 8.27, "learning_rate": 8.807021632137512e-06, "loss": 0.0284, "step": 8155 }, { "epoch": 8.28, "learning_rate": 8.804845150451065e-06, "loss": 0.036, "step": 8156 }, { "epoch": 8.28, "learning_rate": 8.802668726198868e-06, "loss": 0.0263, "step": 8157 }, { "epoch": 8.28, "learning_rate": 8.800492359485509e-06, "loss": 0.0433, "step": 8158 }, { "epoch": 8.28, "learning_rate": 8.798316050415571e-06, "loss": 0.0318, "step": 8159 }, { "epoch": 8.28, "learning_rate": 8.79613979909364e-06, "loss": 0.029, "step": 8160 }, { "epoch": 8.28, "learning_rate": 8.793963605624305e-06, "loss": 0.0366, "step": 8161 }, { "epoch": 8.28, "learning_rate": 8.791787470112133e-06, "loss": 0.0412, "step": 8162 }, { "epoch": 8.28, "learning_rate": 8.78961139266171e-06, "loss": 0.0325, "step": 8163 }, { "epoch": 8.28, "learning_rate": 8.787435373377606e-06, "loss": 0.031, "step": 8164 }, { "epoch": 8.29, "learning_rate": 8.78525941236439e-06, "loss": 0.0303, "step": 8165 }, { "epoch": 8.29, "learning_rate": 8.78308350972663e-06, "loss": 0.0363, "step": 8166 }, { "epoch": 8.29, "learning_rate": 8.780907665568895e-06, "loss": 0.0287, "step": 8167 }, { "epoch": 8.29, "learning_rate": 8.778731879995744e-06, "loss": 0.0344, "step": 8168 }, { "epoch": 8.29, "learning_rate": 8.776556153111737e-06, "loss": 0.0332, "step": 8169 }, { "epoch": 8.29, "learning_rate": 8.774380485021424e-06, "loss": 0.0333, "step": 8170 }, { "epoch": 8.29, "learning_rate": 8.77220487582937e-06, "loss": 0.0317, "step": 8171 }, { "epoch": 8.29, "learning_rate": 8.770029325640117e-06, "loss": 0.0285, "step": 8172 }, { "epoch": 8.29, "learning_rate": 8.767853834558221e-06, "loss": 0.0295, "step": 8173 }, { "epoch": 8.29, "learning_rate": 8.765678402688223e-06, "loss": 0.0358, "step": 8174 }, { "epoch": 8.3, "learning_rate": 8.763503030134664e-06, "loss": 0.0318, "step": 8175 }, { "epoch": 8.3, "learning_rate": 8.761327717002086e-06, "loss": 0.0328, "step": 8176 }, { "epoch": 8.3, "learning_rate": 8.759152463395021e-06, "loss": 0.0299, "step": 8177 }, { "epoch": 8.3, "learning_rate": 8.756977269418005e-06, "loss": 0.0314, "step": 8178 }, { "epoch": 8.3, "learning_rate": 8.754802135175574e-06, "loss": 0.0315, "step": 8179 }, { "epoch": 8.3, "learning_rate": 8.752627060772251e-06, "loss": 0.0294, "step": 8180 }, { "epoch": 8.3, "learning_rate": 8.750452046312564e-06, "loss": 0.0336, "step": 8181 }, { "epoch": 8.3, "learning_rate": 8.748277091901033e-06, "loss": 0.0379, "step": 8182 }, { "epoch": 8.3, "learning_rate": 8.746102197642175e-06, "loss": 0.0286, "step": 8183 }, { "epoch": 8.3, "learning_rate": 8.743927363640512e-06, "loss": 0.029, "step": 8184 }, { "epoch": 8.31, "learning_rate": 8.741752590000556e-06, "loss": 0.0365, "step": 8185 }, { "epoch": 8.31, "learning_rate": 8.739577876826815e-06, "loss": 0.036, "step": 8186 }, { "epoch": 8.31, "learning_rate": 8.737403224223796e-06, "loss": 0.0364, "step": 8187 }, { "epoch": 8.31, "learning_rate": 8.73522863229601e-06, "loss": 0.0346, "step": 8188 }, { "epoch": 8.31, "learning_rate": 8.733054101147953e-06, "loss": 0.0386, "step": 8189 }, { "epoch": 8.31, "learning_rate": 8.73087963088413e-06, "loss": 0.0296, "step": 8190 }, { "epoch": 8.31, "learning_rate": 8.728705221609032e-06, "loss": 0.0379, "step": 8191 }, { "epoch": 8.31, "learning_rate": 8.726530873427155e-06, "loss": 0.033, "step": 8192 }, { "epoch": 8.31, "learning_rate": 8.724356586442985e-06, "loss": 0.0457, "step": 8193 }, { "epoch": 8.31, "learning_rate": 8.722182360761016e-06, "loss": 0.0314, "step": 8194 }, { "epoch": 8.32, "learning_rate": 8.720008196485723e-06, "loss": 0.0418, "step": 8195 }, { "epoch": 8.32, "learning_rate": 8.717834093721598e-06, "loss": 0.0409, "step": 8196 }, { "epoch": 8.32, "learning_rate": 8.715660052573115e-06, "loss": 0.0359, "step": 8197 }, { "epoch": 8.32, "learning_rate": 8.71348607314475e-06, "loss": 0.0404, "step": 8198 }, { "epoch": 8.32, "learning_rate": 8.711312155540972e-06, "loss": 0.0353, "step": 8199 }, { "epoch": 8.32, "learning_rate": 8.709138299866255e-06, "loss": 0.0358, "step": 8200 }, { "epoch": 8.32, "learning_rate": 8.706964506225065e-06, "loss": 0.0384, "step": 8201 }, { "epoch": 8.32, "learning_rate": 8.704790774721863e-06, "loss": 0.0387, "step": 8202 }, { "epoch": 8.32, "learning_rate": 8.702617105461108e-06, "loss": 0.039, "step": 8203 }, { "epoch": 8.32, "learning_rate": 8.700443498547265e-06, "loss": 0.0377, "step": 8204 }, { "epoch": 8.33, "learning_rate": 8.698269954084784e-06, "loss": 0.0324, "step": 8205 }, { "epoch": 8.33, "learning_rate": 8.696096472178117e-06, "loss": 0.0318, "step": 8206 }, { "epoch": 8.33, "learning_rate": 8.693923052931713e-06, "loss": 0.033, "step": 8207 }, { "epoch": 8.33, "learning_rate": 8.691749696450018e-06, "loss": 0.033, "step": 8208 }, { "epoch": 8.33, "learning_rate": 8.689576402837474e-06, "loss": 0.0452, "step": 8209 }, { "epoch": 8.33, "learning_rate": 8.68740317219852e-06, "loss": 0.0237, "step": 8210 }, { "epoch": 8.33, "learning_rate": 8.685230004637593e-06, "loss": 0.0327, "step": 8211 }, { "epoch": 8.33, "learning_rate": 8.683056900259123e-06, "loss": 0.0355, "step": 8212 }, { "epoch": 8.33, "learning_rate": 8.680883859167548e-06, "loss": 0.0359, "step": 8213 }, { "epoch": 8.33, "learning_rate": 8.678710881467294e-06, "loss": 0.0345, "step": 8214 }, { "epoch": 8.34, "learning_rate": 8.676537967262781e-06, "loss": 0.033, "step": 8215 }, { "epoch": 8.34, "learning_rate": 8.674365116658431e-06, "loss": 0.0314, "step": 8216 }, { "epoch": 8.34, "learning_rate": 8.672192329758666e-06, "loss": 0.0378, "step": 8217 }, { "epoch": 8.34, "learning_rate": 8.6700196066679e-06, "loss": 0.0397, "step": 8218 }, { "epoch": 8.34, "learning_rate": 8.667846947490541e-06, "loss": 0.0369, "step": 8219 }, { "epoch": 8.34, "learning_rate": 8.665674352331e-06, "loss": 0.0351, "step": 8220 }, { "epoch": 8.34, "learning_rate": 8.663501821293687e-06, "loss": 0.0345, "step": 8221 }, { "epoch": 8.34, "learning_rate": 8.661329354483e-06, "loss": 0.0342, "step": 8222 }, { "epoch": 8.34, "learning_rate": 8.659156952003343e-06, "loss": 0.0281, "step": 8223 }, { "epoch": 8.35, "learning_rate": 8.656984613959113e-06, "loss": 0.0376, "step": 8224 }, { "epoch": 8.35, "learning_rate": 8.654812340454698e-06, "loss": 0.0303, "step": 8225 }, { "epoch": 8.35, "learning_rate": 8.652640131594495e-06, "loss": 0.0334, "step": 8226 }, { "epoch": 8.35, "learning_rate": 8.650467987482886e-06, "loss": 0.0374, "step": 8227 }, { "epoch": 8.35, "learning_rate": 8.648295908224255e-06, "loss": 0.0263, "step": 8228 }, { "epoch": 8.35, "learning_rate": 8.64612389392299e-06, "loss": 0.035, "step": 8229 }, { "epoch": 8.35, "learning_rate": 8.643951944683466e-06, "loss": 0.0324, "step": 8230 }, { "epoch": 8.35, "learning_rate": 8.641780060610057e-06, "loss": 0.0319, "step": 8231 }, { "epoch": 8.35, "learning_rate": 8.639608241807135e-06, "loss": 0.0312, "step": 8232 }, { "epoch": 8.35, "learning_rate": 8.637436488379067e-06, "loss": 0.0415, "step": 8233 }, { "epoch": 8.36, "learning_rate": 8.635264800430223e-06, "loss": 0.0335, "step": 8234 }, { "epoch": 8.36, "learning_rate": 8.633093178064962e-06, "loss": 0.0383, "step": 8235 }, { "epoch": 8.36, "learning_rate": 8.630921621387644e-06, "loss": 0.0325, "step": 8236 }, { "epoch": 8.36, "learning_rate": 8.62875013050262e-06, "loss": 0.0417, "step": 8237 }, { "epoch": 8.36, "learning_rate": 8.626578705514253e-06, "loss": 0.0395, "step": 8238 }, { "epoch": 8.36, "learning_rate": 8.624407346526887e-06, "loss": 0.0282, "step": 8239 }, { "epoch": 8.36, "learning_rate": 8.62223605364487e-06, "loss": 0.034, "step": 8240 }, { "epoch": 8.36, "learning_rate": 8.620064826972546e-06, "loss": 0.0265, "step": 8241 }, { "epoch": 8.36, "learning_rate": 8.617893666614254e-06, "loss": 0.027, "step": 8242 }, { "epoch": 8.36, "learning_rate": 8.61572257267433e-06, "loss": 0.032, "step": 8243 }, { "epoch": 8.37, "learning_rate": 8.61355154525711e-06, "loss": 0.0258, "step": 8244 }, { "epoch": 8.37, "learning_rate": 8.61138058446692e-06, "loss": 0.0295, "step": 8245 }, { "epoch": 8.37, "learning_rate": 8.609209690408095e-06, "loss": 0.0347, "step": 8246 }, { "epoch": 8.37, "learning_rate": 8.607038863184957e-06, "loss": 0.0373, "step": 8247 }, { "epoch": 8.37, "learning_rate": 8.604868102901825e-06, "loss": 0.0341, "step": 8248 }, { "epoch": 8.37, "learning_rate": 8.602697409663016e-06, "loss": 0.0328, "step": 8249 }, { "epoch": 8.37, "learning_rate": 8.600526783572847e-06, "loss": 0.0234, "step": 8250 }, { "epoch": 8.37, "learning_rate": 8.59835622473563e-06, "loss": 0.0335, "step": 8251 }, { "epoch": 8.37, "learning_rate": 8.596185733255672e-06, "loss": 0.0451, "step": 8252 }, { "epoch": 8.37, "learning_rate": 8.594015309237272e-06, "loss": 0.0302, "step": 8253 }, { "epoch": 8.38, "learning_rate": 8.591844952784743e-06, "loss": 0.0354, "step": 8254 }, { "epoch": 8.38, "learning_rate": 8.589674664002377e-06, "loss": 0.0316, "step": 8255 }, { "epoch": 8.38, "learning_rate": 8.587504442994467e-06, "loss": 0.0375, "step": 8256 }, { "epoch": 8.38, "learning_rate": 8.585334289865312e-06, "loss": 0.0341, "step": 8257 }, { "epoch": 8.38, "learning_rate": 8.583164204719196e-06, "loss": 0.0396, "step": 8258 }, { "epoch": 8.38, "learning_rate": 8.580994187660403e-06, "loss": 0.0266, "step": 8259 }, { "epoch": 8.38, "learning_rate": 8.578824238793217e-06, "loss": 0.0358, "step": 8260 }, { "epoch": 8.38, "learning_rate": 8.576654358221919e-06, "loss": 0.0394, "step": 8261 }, { "epoch": 8.38, "learning_rate": 8.574484546050779e-06, "loss": 0.0261, "step": 8262 }, { "epoch": 8.38, "learning_rate": 8.572314802384075e-06, "loss": 0.0308, "step": 8263 }, { "epoch": 8.39, "learning_rate": 8.570145127326075e-06, "loss": 0.0301, "step": 8264 }, { "epoch": 8.39, "learning_rate": 8.567975520981045e-06, "loss": 0.0423, "step": 8265 }, { "epoch": 8.39, "learning_rate": 8.565805983453242e-06, "loss": 0.0389, "step": 8266 }, { "epoch": 8.39, "learning_rate": 8.563636514846931e-06, "loss": 0.0298, "step": 8267 }, { "epoch": 8.39, "learning_rate": 8.561467115266367e-06, "loss": 0.0391, "step": 8268 }, { "epoch": 8.39, "learning_rate": 8.5592977848158e-06, "loss": 0.0423, "step": 8269 }, { "epoch": 8.39, "learning_rate": 8.557128523599476e-06, "loss": 0.034, "step": 8270 }, { "epoch": 8.39, "learning_rate": 8.554959331721651e-06, "loss": 0.0332, "step": 8271 }, { "epoch": 8.39, "learning_rate": 8.552790209286559e-06, "loss": 0.0322, "step": 8272 }, { "epoch": 8.39, "learning_rate": 8.550621156398443e-06, "loss": 0.0397, "step": 8273 }, { "epoch": 8.4, "learning_rate": 8.548452173161539e-06, "loss": 0.031, "step": 8274 }, { "epoch": 8.4, "learning_rate": 8.546283259680078e-06, "loss": 0.0358, "step": 8275 }, { "epoch": 8.4, "learning_rate": 8.544114416058288e-06, "loss": 0.0325, "step": 8276 }, { "epoch": 8.4, "learning_rate": 8.541945642400394e-06, "loss": 0.0402, "step": 8277 }, { "epoch": 8.4, "learning_rate": 8.539776938810622e-06, "loss": 0.0264, "step": 8278 }, { "epoch": 8.4, "learning_rate": 8.537608305393185e-06, "loss": 0.0324, "step": 8279 }, { "epoch": 8.4, "learning_rate": 8.535439742252308e-06, "loss": 0.0366, "step": 8280 }, { "epoch": 8.4, "learning_rate": 8.533271249492198e-06, "loss": 0.0356, "step": 8281 }, { "epoch": 8.4, "learning_rate": 8.531102827217065e-06, "loss": 0.0364, "step": 8282 }, { "epoch": 8.4, "learning_rate": 8.52893447553111e-06, "loss": 0.0323, "step": 8283 }, { "epoch": 8.41, "learning_rate": 8.526766194538541e-06, "loss": 0.0422, "step": 8284 }, { "epoch": 8.41, "learning_rate": 8.524597984343556e-06, "loss": 0.0372, "step": 8285 }, { "epoch": 8.41, "learning_rate": 8.522429845050347e-06, "loss": 0.0409, "step": 8286 }, { "epoch": 8.41, "learning_rate": 8.520261776763103e-06, "loss": 0.0313, "step": 8287 }, { "epoch": 8.41, "learning_rate": 8.518093779586023e-06, "loss": 0.0401, "step": 8288 }, { "epoch": 8.41, "learning_rate": 8.515925853623283e-06, "loss": 0.0312, "step": 8289 }, { "epoch": 8.41, "learning_rate": 8.51375799897907e-06, "loss": 0.0374, "step": 8290 }, { "epoch": 8.41, "learning_rate": 8.51159021575756e-06, "loss": 0.0312, "step": 8291 }, { "epoch": 8.41, "learning_rate": 8.509422504062927e-06, "loss": 0.0396, "step": 8292 }, { "epoch": 8.42, "learning_rate": 8.50725486399934e-06, "loss": 0.0368, "step": 8293 }, { "epoch": 8.42, "learning_rate": 8.505087295670976e-06, "loss": 0.0354, "step": 8294 }, { "epoch": 8.42, "learning_rate": 8.502919799181986e-06, "loss": 0.0331, "step": 8295 }, { "epoch": 8.42, "learning_rate": 8.500752374636544e-06, "loss": 0.033, "step": 8296 }, { "epoch": 8.42, "learning_rate": 8.498585022138801e-06, "loss": 0.0315, "step": 8297 }, { "epoch": 8.42, "learning_rate": 8.496417741792912e-06, "loss": 0.0326, "step": 8298 }, { "epoch": 8.42, "learning_rate": 8.49425053370303e-06, "loss": 0.0362, "step": 8299 }, { "epoch": 8.42, "learning_rate": 8.492083397973297e-06, "loss": 0.0396, "step": 8300 }, { "epoch": 8.42, "learning_rate": 8.48991633470786e-06, "loss": 0.0355, "step": 8301 }, { "epoch": 8.42, "learning_rate": 8.48774934401086e-06, "loss": 0.0351, "step": 8302 }, { "epoch": 8.43, "learning_rate": 8.485582425986431e-06, "loss": 0.0404, "step": 8303 }, { "epoch": 8.43, "learning_rate": 8.483415580738704e-06, "loss": 0.0299, "step": 8304 }, { "epoch": 8.43, "learning_rate": 8.481248808371816e-06, "loss": 0.0331, "step": 8305 }, { "epoch": 8.43, "learning_rate": 8.479082108989888e-06, "loss": 0.0409, "step": 8306 }, { "epoch": 8.43, "learning_rate": 8.476915482697047e-06, "loss": 0.029, "step": 8307 }, { "epoch": 8.43, "learning_rate": 8.47474892959741e-06, "loss": 0.0322, "step": 8308 }, { "epoch": 8.43, "learning_rate": 8.47258244979509e-06, "loss": 0.0357, "step": 8309 }, { "epoch": 8.43, "learning_rate": 8.4704160433942e-06, "loss": 0.0392, "step": 8310 }, { "epoch": 8.43, "learning_rate": 8.468249710498853e-06, "loss": 0.0391, "step": 8311 }, { "epoch": 8.43, "learning_rate": 8.466083451213145e-06, "loss": 0.0526, "step": 8312 }, { "epoch": 8.44, "learning_rate": 8.463917265641188e-06, "loss": 0.0359, "step": 8313 }, { "epoch": 8.44, "learning_rate": 8.461751153887077e-06, "loss": 0.03, "step": 8314 }, { "epoch": 8.44, "learning_rate": 8.459585116054903e-06, "loss": 0.0338, "step": 8315 }, { "epoch": 8.44, "learning_rate": 8.457419152248758e-06, "loss": 0.0333, "step": 8316 }, { "epoch": 8.44, "learning_rate": 8.455253262572731e-06, "loss": 0.0313, "step": 8317 }, { "epoch": 8.44, "learning_rate": 8.453087447130907e-06, "loss": 0.0358, "step": 8318 }, { "epoch": 8.44, "learning_rate": 8.450921706027362e-06, "loss": 0.0286, "step": 8319 }, { "epoch": 8.44, "learning_rate": 8.448756039366172e-06, "loss": 0.0325, "step": 8320 }, { "epoch": 8.44, "learning_rate": 8.446590447251417e-06, "loss": 0.037, "step": 8321 }, { "epoch": 8.44, "learning_rate": 8.444424929787159e-06, "loss": 0.027, "step": 8322 }, { "epoch": 8.45, "learning_rate": 8.44225948707747e-06, "loss": 0.0332, "step": 8323 }, { "epoch": 8.45, "learning_rate": 8.44009411922641e-06, "loss": 0.0263, "step": 8324 }, { "epoch": 8.45, "learning_rate": 8.437928826338038e-06, "loss": 0.0337, "step": 8325 }, { "epoch": 8.45, "learning_rate": 8.435763608516408e-06, "loss": 0.0315, "step": 8326 }, { "epoch": 8.45, "learning_rate": 8.43359846586557e-06, "loss": 0.0311, "step": 8327 }, { "epoch": 8.45, "learning_rate": 8.431433398489575e-06, "loss": 0.0359, "step": 8328 }, { "epoch": 8.45, "learning_rate": 8.429268406492461e-06, "loss": 0.0358, "step": 8329 }, { "epoch": 8.45, "learning_rate": 8.42710348997828e-06, "loss": 0.0374, "step": 8330 }, { "epoch": 8.45, "learning_rate": 8.424938649051063e-06, "loss": 0.0379, "step": 8331 }, { "epoch": 8.45, "learning_rate": 8.422773883814841e-06, "loss": 0.0479, "step": 8332 }, { "epoch": 8.46, "learning_rate": 8.420609194373644e-06, "loss": 0.0375, "step": 8333 }, { "epoch": 8.46, "learning_rate": 8.418444580831503e-06, "loss": 0.0363, "step": 8334 }, { "epoch": 8.46, "learning_rate": 8.416280043292438e-06, "loss": 0.0364, "step": 8335 }, { "epoch": 8.46, "learning_rate": 8.414115581860466e-06, "loss": 0.0284, "step": 8336 }, { "epoch": 8.46, "learning_rate": 8.4119511966396e-06, "loss": 0.034, "step": 8337 }, { "epoch": 8.46, "learning_rate": 8.409786887733858e-06, "loss": 0.0334, "step": 8338 }, { "epoch": 8.46, "learning_rate": 8.407622655247243e-06, "loss": 0.0433, "step": 8339 }, { "epoch": 8.46, "learning_rate": 8.405458499283763e-06, "loss": 0.0382, "step": 8340 }, { "epoch": 8.46, "learning_rate": 8.403294419947419e-06, "loss": 0.026, "step": 8341 }, { "epoch": 8.46, "learning_rate": 8.401130417342202e-06, "loss": 0.0391, "step": 8342 }, { "epoch": 8.47, "learning_rate": 8.398966491572107e-06, "loss": 0.0265, "step": 8343 }, { "epoch": 8.47, "learning_rate": 8.396802642741126e-06, "loss": 0.0322, "step": 8344 }, { "epoch": 8.47, "learning_rate": 8.39463887095324e-06, "loss": 0.0307, "step": 8345 }, { "epoch": 8.47, "learning_rate": 8.392475176312439e-06, "loss": 0.042, "step": 8346 }, { "epoch": 8.47, "learning_rate": 8.390311558922696e-06, "loss": 0.04, "step": 8347 }, { "epoch": 8.47, "learning_rate": 8.388148018887986e-06, "loss": 0.0339, "step": 8348 }, { "epoch": 8.47, "learning_rate": 8.385984556312282e-06, "loss": 0.0319, "step": 8349 }, { "epoch": 8.47, "learning_rate": 8.383821171299544e-06, "loss": 0.0472, "step": 8350 }, { "epoch": 8.47, "learning_rate": 8.381657863953745e-06, "loss": 0.0336, "step": 8351 }, { "epoch": 8.47, "learning_rate": 8.37949463437884e-06, "loss": 0.0414, "step": 8352 }, { "epoch": 8.48, "learning_rate": 8.377331482678784e-06, "loss": 0.0316, "step": 8353 }, { "epoch": 8.48, "learning_rate": 8.375168408957529e-06, "loss": 0.0303, "step": 8354 }, { "epoch": 8.48, "learning_rate": 8.373005413319027e-06, "loss": 0.042, "step": 8355 }, { "epoch": 8.48, "learning_rate": 8.37084249586722e-06, "loss": 0.0344, "step": 8356 }, { "epoch": 8.48, "learning_rate": 8.368679656706051e-06, "loss": 0.0338, "step": 8357 }, { "epoch": 8.48, "learning_rate": 8.366516895939455e-06, "loss": 0.0334, "step": 8358 }, { "epoch": 8.48, "learning_rate": 8.364354213671367e-06, "loss": 0.0432, "step": 8359 }, { "epoch": 8.48, "learning_rate": 8.362191610005715e-06, "loss": 0.031, "step": 8360 }, { "epoch": 8.48, "learning_rate": 8.360029085046427e-06, "loss": 0.0424, "step": 8361 }, { "epoch": 8.49, "learning_rate": 8.357866638897418e-06, "loss": 0.0459, "step": 8362 }, { "epoch": 8.49, "learning_rate": 8.355704271662619e-06, "loss": 0.0368, "step": 8363 }, { "epoch": 8.49, "learning_rate": 8.353541983445936e-06, "loss": 0.0375, "step": 8364 }, { "epoch": 8.49, "learning_rate": 8.351379774351283e-06, "loss": 0.0399, "step": 8365 }, { "epoch": 8.49, "learning_rate": 8.349217644482563e-06, "loss": 0.0338, "step": 8366 }, { "epoch": 8.49, "learning_rate": 8.347055593943682e-06, "loss": 0.034, "step": 8367 }, { "epoch": 8.49, "learning_rate": 8.34489362283854e-06, "loss": 0.0321, "step": 8368 }, { "epoch": 8.49, "learning_rate": 8.342731731271032e-06, "loss": 0.0344, "step": 8369 }, { "epoch": 8.49, "learning_rate": 8.340569919345042e-06, "loss": 0.0345, "step": 8370 }, { "epoch": 8.49, "learning_rate": 8.33840818716447e-06, "loss": 0.0357, "step": 8371 }, { "epoch": 8.5, "learning_rate": 8.336246534833194e-06, "loss": 0.0329, "step": 8372 }, { "epoch": 8.5, "learning_rate": 8.334084962455096e-06, "loss": 0.0342, "step": 8373 }, { "epoch": 8.5, "learning_rate": 8.331923470134052e-06, "loss": 0.0332, "step": 8374 }, { "epoch": 8.5, "learning_rate": 8.329762057973932e-06, "loss": 0.0384, "step": 8375 }, { "epoch": 8.5, "learning_rate": 8.327600726078607e-06, "loss": 0.0356, "step": 8376 }, { "epoch": 8.5, "learning_rate": 8.32543947455194e-06, "loss": 0.0345, "step": 8377 }, { "epoch": 8.5, "learning_rate": 8.323278303497794e-06, "loss": 0.0308, "step": 8378 }, { "epoch": 8.5, "learning_rate": 8.32111721302002e-06, "loss": 0.0292, "step": 8379 }, { "epoch": 8.5, "learning_rate": 8.318956203222482e-06, "loss": 0.0362, "step": 8380 }, { "epoch": 8.5, "learning_rate": 8.316795274209022e-06, "loss": 0.0304, "step": 8381 }, { "epoch": 8.51, "learning_rate": 8.314634426083488e-06, "loss": 0.0421, "step": 8382 }, { "epoch": 8.51, "learning_rate": 8.312473658949717e-06, "loss": 0.042, "step": 8383 }, { "epoch": 8.51, "learning_rate": 8.310312972911552e-06, "loss": 0.0357, "step": 8384 }, { "epoch": 8.51, "learning_rate": 8.308152368072826e-06, "loss": 0.0354, "step": 8385 }, { "epoch": 8.51, "learning_rate": 8.305991844537365e-06, "loss": 0.0339, "step": 8386 }, { "epoch": 8.51, "learning_rate": 8.303831402408995e-06, "loss": 0.0374, "step": 8387 }, { "epoch": 8.51, "learning_rate": 8.301671041791543e-06, "loss": 0.0431, "step": 8388 }, { "epoch": 8.51, "learning_rate": 8.299510762788825e-06, "loss": 0.0331, "step": 8389 }, { "epoch": 8.51, "learning_rate": 8.297350565504655e-06, "loss": 0.037, "step": 8390 }, { "epoch": 8.51, "learning_rate": 8.295190450042843e-06, "loss": 0.0347, "step": 8391 }, { "epoch": 8.52, "learning_rate": 8.293030416507195e-06, "loss": 0.0355, "step": 8392 }, { "epoch": 8.52, "learning_rate": 8.290870465001512e-06, "loss": 0.0249, "step": 8393 }, { "epoch": 8.52, "learning_rate": 8.288710595629594e-06, "loss": 0.0491, "step": 8394 }, { "epoch": 8.52, "learning_rate": 8.286550808495233e-06, "loss": 0.0352, "step": 8395 }, { "epoch": 8.52, "learning_rate": 8.284391103702227e-06, "loss": 0.0266, "step": 8396 }, { "epoch": 8.52, "learning_rate": 8.282231481354356e-06, "loss": 0.0402, "step": 8397 }, { "epoch": 8.52, "learning_rate": 8.280071941555405e-06, "loss": 0.038, "step": 8398 }, { "epoch": 8.52, "learning_rate": 8.277912484409149e-06, "loss": 0.0434, "step": 8399 }, { "epoch": 8.52, "learning_rate": 8.275753110019367e-06, "loss": 0.037, "step": 8400 }, { "epoch": 8.52, "learning_rate": 8.273593818489828e-06, "loss": 0.035, "step": 8401 }, { "epoch": 8.53, "learning_rate": 8.2714346099243e-06, "loss": 0.0281, "step": 8402 }, { "epoch": 8.53, "learning_rate": 8.269275484426544e-06, "loss": 0.0399, "step": 8403 }, { "epoch": 8.53, "learning_rate": 8.267116442100313e-06, "loss": 0.0266, "step": 8404 }, { "epoch": 8.53, "learning_rate": 8.264957483049375e-06, "loss": 0.0345, "step": 8405 }, { "epoch": 8.53, "learning_rate": 8.26279860737747e-06, "loss": 0.0327, "step": 8406 }, { "epoch": 8.53, "learning_rate": 8.26063981518835e-06, "loss": 0.0389, "step": 8407 }, { "epoch": 8.53, "learning_rate": 8.258481106585758e-06, "loss": 0.0287, "step": 8408 }, { "epoch": 8.53, "learning_rate": 8.256322481673432e-06, "loss": 0.0279, "step": 8409 }, { "epoch": 8.53, "learning_rate": 8.2541639405551e-06, "loss": 0.0351, "step": 8410 }, { "epoch": 8.53, "learning_rate": 8.252005483334501e-06, "loss": 0.0371, "step": 8411 }, { "epoch": 8.54, "learning_rate": 8.249847110115356e-06, "loss": 0.0414, "step": 8412 }, { "epoch": 8.54, "learning_rate": 8.247688821001393e-06, "loss": 0.0453, "step": 8413 }, { "epoch": 8.54, "learning_rate": 8.245530616096329e-06, "loss": 0.0384, "step": 8414 }, { "epoch": 8.54, "learning_rate": 8.243372495503876e-06, "loss": 0.0299, "step": 8415 }, { "epoch": 8.54, "learning_rate": 8.241214459327746e-06, "loss": 0.033, "step": 8416 }, { "epoch": 8.54, "learning_rate": 8.239056507671645e-06, "loss": 0.033, "step": 8417 }, { "epoch": 8.54, "learning_rate": 8.236898640639277e-06, "loss": 0.0321, "step": 8418 }, { "epoch": 8.54, "learning_rate": 8.234740858334337e-06, "loss": 0.0352, "step": 8419 }, { "epoch": 8.54, "learning_rate": 8.232583160860522e-06, "loss": 0.0384, "step": 8420 }, { "epoch": 8.54, "learning_rate": 8.230425548321517e-06, "loss": 0.041, "step": 8421 }, { "epoch": 8.55, "learning_rate": 8.228268020821015e-06, "loss": 0.0376, "step": 8422 }, { "epoch": 8.55, "learning_rate": 8.226110578462694e-06, "loss": 0.0344, "step": 8423 }, { "epoch": 8.55, "learning_rate": 8.223953221350235e-06, "loss": 0.0348, "step": 8424 }, { "epoch": 8.55, "learning_rate": 8.22179594958731e-06, "loss": 0.0321, "step": 8425 }, { "epoch": 8.55, "learning_rate": 8.219638763277586e-06, "loss": 0.0396, "step": 8426 }, { "epoch": 8.55, "learning_rate": 8.217481662524731e-06, "loss": 0.0317, "step": 8427 }, { "epoch": 8.55, "learning_rate": 8.215324647432408e-06, "loss": 0.0383, "step": 8428 }, { "epoch": 8.55, "learning_rate": 8.213167718104268e-06, "loss": 0.0286, "step": 8429 }, { "epoch": 8.55, "learning_rate": 8.21101087464397e-06, "loss": 0.0377, "step": 8430 }, { "epoch": 8.56, "learning_rate": 8.208854117155164e-06, "loss": 0.0396, "step": 8431 }, { "epoch": 8.56, "learning_rate": 8.206697445741494e-06, "loss": 0.0351, "step": 8432 }, { "epoch": 8.56, "learning_rate": 8.204540860506596e-06, "loss": 0.0367, "step": 8433 }, { "epoch": 8.56, "learning_rate": 8.202384361554112e-06, "loss": 0.0383, "step": 8434 }, { "epoch": 8.56, "learning_rate": 8.200227948987674e-06, "loss": 0.034, "step": 8435 }, { "epoch": 8.56, "learning_rate": 8.198071622910909e-06, "loss": 0.0353, "step": 8436 }, { "epoch": 8.56, "learning_rate": 8.195915383427434e-06, "loss": 0.0312, "step": 8437 }, { "epoch": 8.56, "learning_rate": 8.193759230640884e-06, "loss": 0.0318, "step": 8438 }, { "epoch": 8.56, "learning_rate": 8.191603164654863e-06, "loss": 0.0348, "step": 8439 }, { "epoch": 8.56, "learning_rate": 8.18944718557299e-06, "loss": 0.0452, "step": 8440 }, { "epoch": 8.57, "learning_rate": 8.187291293498867e-06, "loss": 0.0324, "step": 8441 }, { "epoch": 8.57, "learning_rate": 8.1851354885361e-06, "loss": 0.0341, "step": 8442 }, { "epoch": 8.57, "learning_rate": 8.18297977078829e-06, "loss": 0.0281, "step": 8443 }, { "epoch": 8.57, "learning_rate": 8.180824140359026e-06, "loss": 0.0485, "step": 8444 }, { "epoch": 8.57, "learning_rate": 8.178668597351905e-06, "loss": 0.0346, "step": 8445 }, { "epoch": 8.57, "learning_rate": 8.176513141870506e-06, "loss": 0.0334, "step": 8446 }, { "epoch": 8.57, "learning_rate": 8.174357774018423e-06, "loss": 0.0346, "step": 8447 }, { "epoch": 8.57, "learning_rate": 8.172202493899226e-06, "loss": 0.0379, "step": 8448 }, { "epoch": 8.57, "learning_rate": 8.170047301616491e-06, "loss": 0.0383, "step": 8449 }, { "epoch": 8.57, "learning_rate": 8.167892197273785e-06, "loss": 0.0417, "step": 8450 }, { "epoch": 8.58, "learning_rate": 8.165737180974678e-06, "loss": 0.0299, "step": 8451 }, { "epoch": 8.58, "learning_rate": 8.163582252822728e-06, "loss": 0.0295, "step": 8452 }, { "epoch": 8.58, "learning_rate": 8.161427412921494e-06, "loss": 0.0334, "step": 8453 }, { "epoch": 8.58, "learning_rate": 8.159272661374525e-06, "loss": 0.0394, "step": 8454 }, { "epoch": 8.58, "learning_rate": 8.157117998285373e-06, "loss": 0.041, "step": 8455 }, { "epoch": 8.58, "learning_rate": 8.154963423757583e-06, "loss": 0.0299, "step": 8456 }, { "epoch": 8.58, "learning_rate": 8.152808937894694e-06, "loss": 0.0415, "step": 8457 }, { "epoch": 8.58, "learning_rate": 8.15065454080024e-06, "loss": 0.0364, "step": 8458 }, { "epoch": 8.58, "learning_rate": 8.148500232577756e-06, "loss": 0.0298, "step": 8459 }, { "epoch": 8.58, "learning_rate": 8.146346013330765e-06, "loss": 0.0418, "step": 8460 }, { "epoch": 8.59, "learning_rate": 8.144191883162795e-06, "loss": 0.0345, "step": 8461 }, { "epoch": 8.59, "learning_rate": 8.142037842177355e-06, "loss": 0.0285, "step": 8462 }, { "epoch": 8.59, "learning_rate": 8.139883890477972e-06, "loss": 0.0346, "step": 8463 }, { "epoch": 8.59, "learning_rate": 8.137730028168149e-06, "loss": 0.0352, "step": 8464 }, { "epoch": 8.59, "learning_rate": 8.135576255351395e-06, "loss": 0.0368, "step": 8465 }, { "epoch": 8.59, "learning_rate": 8.133422572131207e-06, "loss": 0.0389, "step": 8466 }, { "epoch": 8.59, "learning_rate": 8.131268978611085e-06, "loss": 0.0333, "step": 8467 }, { "epoch": 8.59, "learning_rate": 8.129115474894523e-06, "loss": 0.0297, "step": 8468 }, { "epoch": 8.59, "learning_rate": 8.126962061085009e-06, "loss": 0.0396, "step": 8469 }, { "epoch": 8.59, "learning_rate": 8.124808737286025e-06, "loss": 0.0372, "step": 8470 }, { "epoch": 8.6, "learning_rate": 8.12265550360105e-06, "loss": 0.0364, "step": 8471 }, { "epoch": 8.6, "learning_rate": 8.120502360133564e-06, "loss": 0.0419, "step": 8472 }, { "epoch": 8.6, "learning_rate": 8.118349306987035e-06, "loss": 0.0384, "step": 8473 }, { "epoch": 8.6, "learning_rate": 8.116196344264933e-06, "loss": 0.0387, "step": 8474 }, { "epoch": 8.6, "learning_rate": 8.114043472070718e-06, "loss": 0.0459, "step": 8475 }, { "epoch": 8.6, "learning_rate": 8.11189069050785e-06, "loss": 0.0341, "step": 8476 }, { "epoch": 8.6, "learning_rate": 8.10973799967978e-06, "loss": 0.0324, "step": 8477 }, { "epoch": 8.6, "learning_rate": 8.107585399689961e-06, "loss": 0.0417, "step": 8478 }, { "epoch": 8.6, "learning_rate": 8.105432890641832e-06, "loss": 0.0376, "step": 8479 }, { "epoch": 8.6, "learning_rate": 8.103280472638841e-06, "loss": 0.0277, "step": 8480 }, { "epoch": 8.61, "learning_rate": 8.101128145784424e-06, "loss": 0.0406, "step": 8481 }, { "epoch": 8.61, "learning_rate": 8.09897591018201e-06, "loss": 0.0341, "step": 8482 }, { "epoch": 8.61, "learning_rate": 8.096823765935024e-06, "loss": 0.0409, "step": 8483 }, { "epoch": 8.61, "learning_rate": 8.094671713146896e-06, "loss": 0.0481, "step": 8484 }, { "epoch": 8.61, "learning_rate": 8.09251975192104e-06, "loss": 0.0449, "step": 8485 }, { "epoch": 8.61, "learning_rate": 8.090367882360871e-06, "loss": 0.0382, "step": 8486 }, { "epoch": 8.61, "learning_rate": 8.088216104569799e-06, "loss": 0.0389, "step": 8487 }, { "epoch": 8.61, "learning_rate": 8.086064418651232e-06, "loss": 0.0356, "step": 8488 }, { "epoch": 8.61, "learning_rate": 8.083912824708569e-06, "loss": 0.0413, "step": 8489 }, { "epoch": 8.61, "learning_rate": 8.081761322845208e-06, "loss": 0.0421, "step": 8490 }, { "epoch": 8.62, "learning_rate": 8.07960991316454e-06, "loss": 0.0419, "step": 8491 }, { "epoch": 8.62, "learning_rate": 8.077458595769956e-06, "loss": 0.0371, "step": 8492 }, { "epoch": 8.62, "learning_rate": 8.075307370764837e-06, "loss": 0.0369, "step": 8493 }, { "epoch": 8.62, "learning_rate": 8.073156238252559e-06, "loss": 0.0342, "step": 8494 }, { "epoch": 8.62, "learning_rate": 8.071005198336502e-06, "loss": 0.0416, "step": 8495 }, { "epoch": 8.62, "learning_rate": 8.068854251120031e-06, "loss": 0.0327, "step": 8496 }, { "epoch": 8.62, "learning_rate": 8.066703396706517e-06, "loss": 0.0325, "step": 8497 }, { "epoch": 8.62, "learning_rate": 8.06455263519932e-06, "loss": 0.0358, "step": 8498 }, { "epoch": 8.62, "learning_rate": 8.062401966701798e-06, "loss": 0.0452, "step": 8499 }, { "epoch": 8.63, "learning_rate": 8.060251391317296e-06, "loss": 0.0332, "step": 8500 }, { "epoch": 8.63, "learning_rate": 8.05810090914917e-06, "loss": 0.041, "step": 8501 }, { "epoch": 8.63, "learning_rate": 8.05595052030076e-06, "loss": 0.0437, "step": 8502 }, { "epoch": 8.63, "learning_rate": 8.053800224875408e-06, "loss": 0.0308, "step": 8503 }, { "epoch": 8.63, "learning_rate": 8.051650022976439e-06, "loss": 0.0322, "step": 8504 }, { "epoch": 8.63, "learning_rate": 8.049499914707192e-06, "loss": 0.0328, "step": 8505 }, { "epoch": 8.63, "learning_rate": 8.04734990017099e-06, "loss": 0.0399, "step": 8506 }, { "epoch": 8.63, "learning_rate": 8.045199979471156e-06, "loss": 0.0423, "step": 8507 }, { "epoch": 8.63, "learning_rate": 8.043050152711005e-06, "loss": 0.0356, "step": 8508 }, { "epoch": 8.63, "learning_rate": 8.040900419993846e-06, "loss": 0.0324, "step": 8509 }, { "epoch": 8.64, "learning_rate": 8.038750781422988e-06, "loss": 0.0354, "step": 8510 }, { "epoch": 8.64, "learning_rate": 8.036601237101735e-06, "loss": 0.0363, "step": 8511 }, { "epoch": 8.64, "learning_rate": 8.034451787133381e-06, "loss": 0.0431, "step": 8512 }, { "epoch": 8.64, "learning_rate": 8.032302431621227e-06, "loss": 0.0373, "step": 8513 }, { "epoch": 8.64, "learning_rate": 8.03015317066856e-06, "loss": 0.0371, "step": 8514 }, { "epoch": 8.64, "learning_rate": 8.028004004378662e-06, "loss": 0.0331, "step": 8515 }, { "epoch": 8.64, "learning_rate": 8.025854932854815e-06, "loss": 0.0248, "step": 8516 }, { "epoch": 8.64, "learning_rate": 8.02370595620029e-06, "loss": 0.0413, "step": 8517 }, { "epoch": 8.64, "learning_rate": 8.021557074518364e-06, "loss": 0.0361, "step": 8518 }, { "epoch": 8.64, "learning_rate": 8.019408287912304e-06, "loss": 0.0342, "step": 8519 }, { "epoch": 8.65, "learning_rate": 8.017259596485367e-06, "loss": 0.0311, "step": 8520 }, { "epoch": 8.65, "learning_rate": 8.01511100034081e-06, "loss": 0.0294, "step": 8521 }, { "epoch": 8.65, "learning_rate": 8.012962499581891e-06, "loss": 0.0361, "step": 8522 }, { "epoch": 8.65, "learning_rate": 8.010814094311855e-06, "loss": 0.0311, "step": 8523 }, { "epoch": 8.65, "learning_rate": 8.008665784633947e-06, "loss": 0.0279, "step": 8524 }, { "epoch": 8.65, "learning_rate": 8.006517570651404e-06, "loss": 0.0392, "step": 8525 }, { "epoch": 8.65, "learning_rate": 8.004369452467463e-06, "loss": 0.0427, "step": 8526 }, { "epoch": 8.65, "learning_rate": 8.002221430185349e-06, "loss": 0.0342, "step": 8527 }, { "epoch": 8.65, "learning_rate": 8.000073503908292e-06, "loss": 0.039, "step": 8528 }, { "epoch": 8.65, "learning_rate": 7.997925673739507e-06, "loss": 0.0423, "step": 8529 }, { "epoch": 8.66, "learning_rate": 7.995777939782219e-06, "loss": 0.0375, "step": 8530 }, { "epoch": 8.66, "learning_rate": 7.993630302139632e-06, "loss": 0.035, "step": 8531 }, { "epoch": 8.66, "learning_rate": 7.991482760914955e-06, "loss": 0.0372, "step": 8532 }, { "epoch": 8.66, "learning_rate": 7.989335316211388e-06, "loss": 0.0336, "step": 8533 }, { "epoch": 8.66, "learning_rate": 7.987187968132133e-06, "loss": 0.04, "step": 8534 }, { "epoch": 8.66, "learning_rate": 7.98504071678038e-06, "loss": 0.037, "step": 8535 }, { "epoch": 8.66, "learning_rate": 7.982893562259317e-06, "loss": 0.0345, "step": 8536 }, { "epoch": 8.66, "learning_rate": 7.980746504672126e-06, "loss": 0.0432, "step": 8537 }, { "epoch": 8.66, "learning_rate": 7.978599544121984e-06, "loss": 0.04, "step": 8538 }, { "epoch": 8.66, "learning_rate": 7.97645268071207e-06, "loss": 0.0345, "step": 8539 }, { "epoch": 8.67, "learning_rate": 7.974305914545552e-06, "loss": 0.0319, "step": 8540 }, { "epoch": 8.67, "learning_rate": 7.972159245725597e-06, "loss": 0.0431, "step": 8541 }, { "epoch": 8.67, "learning_rate": 7.970012674355363e-06, "loss": 0.0338, "step": 8542 }, { "epoch": 8.67, "learning_rate": 7.967866200538004e-06, "loss": 0.0452, "step": 8543 }, { "epoch": 8.67, "learning_rate": 7.96571982437667e-06, "loss": 0.036, "step": 8544 }, { "epoch": 8.67, "learning_rate": 7.963573545974511e-06, "loss": 0.0472, "step": 8545 }, { "epoch": 8.67, "learning_rate": 7.961427365434662e-06, "loss": 0.0349, "step": 8546 }, { "epoch": 8.67, "learning_rate": 7.959281282860268e-06, "loss": 0.0334, "step": 8547 }, { "epoch": 8.67, "learning_rate": 7.957135298354457e-06, "loss": 0.0453, "step": 8548 }, { "epoch": 8.67, "learning_rate": 7.954989412020355e-06, "loss": 0.0388, "step": 8549 }, { "epoch": 8.68, "learning_rate": 7.952843623961085e-06, "loss": 0.0372, "step": 8550 }, { "epoch": 8.68, "learning_rate": 7.950697934279767e-06, "loss": 0.0424, "step": 8551 }, { "epoch": 8.68, "learning_rate": 7.948552343079512e-06, "loss": 0.0307, "step": 8552 }, { "epoch": 8.68, "learning_rate": 7.94640685046343e-06, "loss": 0.0427, "step": 8553 }, { "epoch": 8.68, "learning_rate": 7.944261456534617e-06, "loss": 0.0347, "step": 8554 }, { "epoch": 8.68, "learning_rate": 7.942116161396184e-06, "loss": 0.0371, "step": 8555 }, { "epoch": 8.68, "learning_rate": 7.939970965151217e-06, "loss": 0.0373, "step": 8556 }, { "epoch": 8.68, "learning_rate": 7.937825867902808e-06, "loss": 0.0435, "step": 8557 }, { "epoch": 8.68, "learning_rate": 7.935680869754043e-06, "loss": 0.0396, "step": 8558 }, { "epoch": 8.68, "learning_rate": 7.933535970808e-06, "loss": 0.0366, "step": 8559 }, { "epoch": 8.69, "learning_rate": 7.931391171167751e-06, "loss": 0.0322, "step": 8560 }, { "epoch": 8.69, "learning_rate": 7.929246470936373e-06, "loss": 0.0382, "step": 8561 }, { "epoch": 8.69, "learning_rate": 7.927101870216927e-06, "loss": 0.0419, "step": 8562 }, { "epoch": 8.69, "learning_rate": 7.92495736911247e-06, "loss": 0.0492, "step": 8563 }, { "epoch": 8.69, "learning_rate": 7.922812967726066e-06, "loss": 0.0354, "step": 8564 }, { "epoch": 8.69, "learning_rate": 7.920668666160765e-06, "loss": 0.034, "step": 8565 }, { "epoch": 8.69, "learning_rate": 7.918524464519608e-06, "loss": 0.0329, "step": 8566 }, { "epoch": 8.69, "learning_rate": 7.91638036290564e-06, "loss": 0.0344, "step": 8567 }, { "epoch": 8.69, "learning_rate": 7.914236361421899e-06, "loss": 0.0298, "step": 8568 }, { "epoch": 8.7, "learning_rate": 7.912092460171414e-06, "loss": 0.0384, "step": 8569 }, { "epoch": 8.7, "learning_rate": 7.909948659257213e-06, "loss": 0.0377, "step": 8570 }, { "epoch": 8.7, "learning_rate": 7.907804958782317e-06, "loss": 0.0379, "step": 8571 }, { "epoch": 8.7, "learning_rate": 7.905661358849748e-06, "loss": 0.032, "step": 8572 }, { "epoch": 8.7, "learning_rate": 7.903517859562512e-06, "loss": 0.0307, "step": 8573 }, { "epoch": 8.7, "learning_rate": 7.901374461023625e-06, "loss": 0.0333, "step": 8574 }, { "epoch": 8.7, "learning_rate": 7.899231163336086e-06, "loss": 0.0328, "step": 8575 }, { "epoch": 8.7, "learning_rate": 7.897087966602892e-06, "loss": 0.0371, "step": 8576 }, { "epoch": 8.7, "learning_rate": 7.894944870927033e-06, "loss": 0.0385, "step": 8577 }, { "epoch": 8.7, "learning_rate": 7.892801876411505e-06, "loss": 0.0395, "step": 8578 }, { "epoch": 8.71, "learning_rate": 7.890658983159283e-06, "loss": 0.0426, "step": 8579 }, { "epoch": 8.71, "learning_rate": 7.888516191273354e-06, "loss": 0.0393, "step": 8580 }, { "epoch": 8.71, "learning_rate": 7.88637350085669e-06, "loss": 0.0396, "step": 8581 }, { "epoch": 8.71, "learning_rate": 7.884230912012257e-06, "loss": 0.0374, "step": 8582 }, { "epoch": 8.71, "learning_rate": 7.882088424843018e-06, "loss": 0.043, "step": 8583 }, { "epoch": 8.71, "learning_rate": 7.879946039451935e-06, "loss": 0.0365, "step": 8584 }, { "epoch": 8.71, "learning_rate": 7.877803755941965e-06, "loss": 0.0396, "step": 8585 }, { "epoch": 8.71, "learning_rate": 7.875661574416052e-06, "loss": 0.0403, "step": 8586 }, { "epoch": 8.71, "learning_rate": 7.873519494977144e-06, "loss": 0.0376, "step": 8587 }, { "epoch": 8.71, "learning_rate": 7.871377517728173e-06, "loss": 0.0414, "step": 8588 }, { "epoch": 8.72, "learning_rate": 7.869235642772084e-06, "loss": 0.0339, "step": 8589 }, { "epoch": 8.72, "learning_rate": 7.867093870211801e-06, "loss": 0.0397, "step": 8590 }, { "epoch": 8.72, "learning_rate": 7.864952200150254e-06, "loss": 0.0285, "step": 8591 }, { "epoch": 8.72, "learning_rate": 7.862810632690358e-06, "loss": 0.036, "step": 8592 }, { "epoch": 8.72, "learning_rate": 7.860669167935028e-06, "loss": 0.0432, "step": 8593 }, { "epoch": 8.72, "learning_rate": 7.858527805987176e-06, "loss": 0.0346, "step": 8594 }, { "epoch": 8.72, "learning_rate": 7.856386546949704e-06, "loss": 0.0408, "step": 8595 }, { "epoch": 8.72, "learning_rate": 7.854245390925514e-06, "loss": 0.0322, "step": 8596 }, { "epoch": 8.72, "learning_rate": 7.852104338017503e-06, "loss": 0.026, "step": 8597 }, { "epoch": 8.72, "learning_rate": 7.849963388328562e-06, "loss": 0.0423, "step": 8598 }, { "epoch": 8.73, "learning_rate": 7.847822541961573e-06, "loss": 0.0336, "step": 8599 }, { "epoch": 8.73, "learning_rate": 7.845681799019416e-06, "loss": 0.0247, "step": 8600 }, { "epoch": 8.73, "learning_rate": 7.843541159604969e-06, "loss": 0.0387, "step": 8601 }, { "epoch": 8.73, "learning_rate": 7.841400623821101e-06, "loss": 0.0351, "step": 8602 }, { "epoch": 8.73, "learning_rate": 7.839260191770677e-06, "loss": 0.0362, "step": 8603 }, { "epoch": 8.73, "learning_rate": 7.837119863556554e-06, "loss": 0.0318, "step": 8604 }, { "epoch": 8.73, "learning_rate": 7.834979639281598e-06, "loss": 0.0267, "step": 8605 }, { "epoch": 8.73, "learning_rate": 7.832839519048646e-06, "loss": 0.0345, "step": 8606 }, { "epoch": 8.73, "learning_rate": 7.830699502960556e-06, "loss": 0.0427, "step": 8607 }, { "epoch": 8.73, "learning_rate": 7.828559591120161e-06, "loss": 0.0382, "step": 8608 }, { "epoch": 8.74, "learning_rate": 7.826419783630297e-06, "loss": 0.0405, "step": 8609 }, { "epoch": 8.74, "learning_rate": 7.824280080593795e-06, "loss": 0.0375, "step": 8610 }, { "epoch": 8.74, "learning_rate": 7.822140482113483e-06, "loss": 0.0386, "step": 8611 }, { "epoch": 8.74, "learning_rate": 7.820000988292178e-06, "loss": 0.0349, "step": 8612 }, { "epoch": 8.74, "learning_rate": 7.817861599232691e-06, "loss": 0.0408, "step": 8613 }, { "epoch": 8.74, "learning_rate": 7.815722315037844e-06, "loss": 0.0387, "step": 8614 }, { "epoch": 8.74, "learning_rate": 7.813583135810435e-06, "loss": 0.0306, "step": 8615 }, { "epoch": 8.74, "learning_rate": 7.811444061653263e-06, "loss": 0.0377, "step": 8616 }, { "epoch": 8.74, "learning_rate": 7.809305092669125e-06, "loss": 0.032, "step": 8617 }, { "epoch": 8.74, "learning_rate": 7.807166228960813e-06, "loss": 0.0387, "step": 8618 }, { "epoch": 8.75, "learning_rate": 7.80502747063111e-06, "loss": 0.0285, "step": 8619 }, { "epoch": 8.75, "learning_rate": 7.802888817782797e-06, "loss": 0.0346, "step": 8620 }, { "epoch": 8.75, "learning_rate": 7.800750270518644e-06, "loss": 0.0454, "step": 8621 }, { "epoch": 8.75, "learning_rate": 7.798611828941428e-06, "loss": 0.0351, "step": 8622 }, { "epoch": 8.75, "learning_rate": 7.796473493153909e-06, "loss": 0.0293, "step": 8623 }, { "epoch": 8.75, "learning_rate": 7.79433526325885e-06, "loss": 0.0371, "step": 8624 }, { "epoch": 8.75, "learning_rate": 7.792197139359005e-06, "loss": 0.038, "step": 8625 }, { "epoch": 8.75, "learning_rate": 7.790059121557122e-06, "loss": 0.043, "step": 8626 }, { "epoch": 8.75, "learning_rate": 7.787921209955943e-06, "loss": 0.0347, "step": 8627 }, { "epoch": 8.75, "learning_rate": 7.785783404658214e-06, "loss": 0.0407, "step": 8628 }, { "epoch": 8.76, "learning_rate": 7.783645705766658e-06, "loss": 0.0343, "step": 8629 }, { "epoch": 8.76, "learning_rate": 7.781508113384017e-06, "loss": 0.026, "step": 8630 }, { "epoch": 8.76, "learning_rate": 7.77937062761301e-06, "loss": 0.0412, "step": 8631 }, { "epoch": 8.76, "learning_rate": 7.777233248556354e-06, "loss": 0.0378, "step": 8632 }, { "epoch": 8.76, "learning_rate": 7.775095976316762e-06, "loss": 0.0386, "step": 8633 }, { "epoch": 8.76, "learning_rate": 7.772958810996947e-06, "loss": 0.0394, "step": 8634 }, { "epoch": 8.76, "learning_rate": 7.770821752699609e-06, "loss": 0.0426, "step": 8635 }, { "epoch": 8.76, "learning_rate": 7.768684801527446e-06, "loss": 0.0291, "step": 8636 }, { "epoch": 8.76, "learning_rate": 7.766547957583154e-06, "loss": 0.0336, "step": 8637 }, { "epoch": 8.77, "learning_rate": 7.764411220969413e-06, "loss": 0.0341, "step": 8638 }, { "epoch": 8.77, "learning_rate": 7.762274591788917e-06, "loss": 0.0349, "step": 8639 }, { "epoch": 8.77, "learning_rate": 7.760138070144336e-06, "loss": 0.0312, "step": 8640 }, { "epoch": 8.77, "learning_rate": 7.758001656138347e-06, "loss": 0.037, "step": 8641 }, { "epoch": 8.77, "learning_rate": 7.755865349873616e-06, "loss": 0.0401, "step": 8642 }, { "epoch": 8.77, "learning_rate": 7.753729151452804e-06, "loss": 0.0451, "step": 8643 }, { "epoch": 8.77, "learning_rate": 7.751593060978566e-06, "loss": 0.0359, "step": 8644 }, { "epoch": 8.77, "learning_rate": 7.749457078553558e-06, "loss": 0.0519, "step": 8645 }, { "epoch": 8.77, "learning_rate": 7.74732120428042e-06, "loss": 0.0323, "step": 8646 }, { "epoch": 8.77, "learning_rate": 7.745185438261803e-06, "loss": 0.052, "step": 8647 }, { "epoch": 8.78, "learning_rate": 7.743049780600339e-06, "loss": 0.0384, "step": 8648 }, { "epoch": 8.78, "learning_rate": 7.740914231398656e-06, "loss": 0.0333, "step": 8649 }, { "epoch": 8.78, "learning_rate": 7.738778790759381e-06, "loss": 0.0365, "step": 8650 }, { "epoch": 8.78, "learning_rate": 7.736643458785137e-06, "loss": 0.0425, "step": 8651 }, { "epoch": 8.78, "learning_rate": 7.73450823557854e-06, "loss": 0.0418, "step": 8652 }, { "epoch": 8.78, "learning_rate": 7.732373121242196e-06, "loss": 0.0386, "step": 8653 }, { "epoch": 8.78, "learning_rate": 7.730238115878708e-06, "loss": 0.0312, "step": 8654 }, { "epoch": 8.78, "learning_rate": 7.72810321959068e-06, "loss": 0.0294, "step": 8655 }, { "epoch": 8.78, "learning_rate": 7.725968432480709e-06, "loss": 0.0364, "step": 8656 }, { "epoch": 8.78, "learning_rate": 7.72383375465138e-06, "loss": 0.0429, "step": 8657 }, { "epoch": 8.79, "learning_rate": 7.721699186205279e-06, "loss": 0.0322, "step": 8658 }, { "epoch": 8.79, "learning_rate": 7.719564727244982e-06, "loss": 0.0355, "step": 8659 }, { "epoch": 8.79, "learning_rate": 7.717430377873064e-06, "loss": 0.0423, "step": 8660 }, { "epoch": 8.79, "learning_rate": 7.71529613819209e-06, "loss": 0.0374, "step": 8661 }, { "epoch": 8.79, "learning_rate": 7.713162008304628e-06, "loss": 0.0393, "step": 8662 }, { "epoch": 8.79, "learning_rate": 7.711027988313228e-06, "loss": 0.0383, "step": 8663 }, { "epoch": 8.79, "learning_rate": 7.708894078320453e-06, "loss": 0.0343, "step": 8664 }, { "epoch": 8.79, "learning_rate": 7.706760278428845e-06, "loss": 0.0307, "step": 8665 }, { "epoch": 8.79, "learning_rate": 7.704626588740943e-06, "loss": 0.0412, "step": 8666 }, { "epoch": 8.79, "learning_rate": 7.702493009359284e-06, "loss": 0.0531, "step": 8667 }, { "epoch": 8.8, "learning_rate": 7.700359540386403e-06, "loss": 0.0349, "step": 8668 }, { "epoch": 8.8, "learning_rate": 7.698226181924823e-06, "loss": 0.0401, "step": 8669 }, { "epoch": 8.8, "learning_rate": 7.696092934077064e-06, "loss": 0.0311, "step": 8670 }, { "epoch": 8.8, "learning_rate": 7.693959796945638e-06, "loss": 0.0434, "step": 8671 }, { "epoch": 8.8, "learning_rate": 7.691826770633063e-06, "loss": 0.0371, "step": 8672 }, { "epoch": 8.8, "learning_rate": 7.689693855241838e-06, "loss": 0.0416, "step": 8673 }, { "epoch": 8.8, "learning_rate": 7.687561050874465e-06, "loss": 0.0412, "step": 8674 }, { "epoch": 8.8, "learning_rate": 7.685428357633435e-06, "loss": 0.0364, "step": 8675 }, { "epoch": 8.8, "learning_rate": 7.68329577562124e-06, "loss": 0.0415, "step": 8676 }, { "epoch": 8.8, "learning_rate": 7.681163304940357e-06, "loss": 0.0351, "step": 8677 }, { "epoch": 8.81, "learning_rate": 7.67903094569327e-06, "loss": 0.046, "step": 8678 }, { "epoch": 8.81, "learning_rate": 7.676898697982449e-06, "loss": 0.0371, "step": 8679 }, { "epoch": 8.81, "learning_rate": 7.674766561910358e-06, "loss": 0.0293, "step": 8680 }, { "epoch": 8.81, "learning_rate": 7.672634537579465e-06, "loss": 0.0304, "step": 8681 }, { "epoch": 8.81, "learning_rate": 7.670502625092225e-06, "loss": 0.0415, "step": 8682 }, { "epoch": 8.81, "learning_rate": 7.668370824551084e-06, "loss": 0.0327, "step": 8683 }, { "epoch": 8.81, "learning_rate": 7.666239136058492e-06, "loss": 0.0331, "step": 8684 }, { "epoch": 8.81, "learning_rate": 7.664107559716889e-06, "loss": 0.0429, "step": 8685 }, { "epoch": 8.81, "learning_rate": 7.66197609562871e-06, "loss": 0.0323, "step": 8686 }, { "epoch": 8.81, "learning_rate": 7.659844743896382e-06, "loss": 0.0396, "step": 8687 }, { "epoch": 8.82, "learning_rate": 7.657713504622326e-06, "loss": 0.0363, "step": 8688 }, { "epoch": 8.82, "learning_rate": 7.65558237790897e-06, "loss": 0.038, "step": 8689 }, { "epoch": 8.82, "learning_rate": 7.65345136385872e-06, "loss": 0.0396, "step": 8690 }, { "epoch": 8.82, "learning_rate": 7.651320462573988e-06, "loss": 0.0351, "step": 8691 }, { "epoch": 8.82, "learning_rate": 7.649189674157176e-06, "loss": 0.0366, "step": 8692 }, { "epoch": 8.82, "learning_rate": 7.647058998710676e-06, "loss": 0.0479, "step": 8693 }, { "epoch": 8.82, "learning_rate": 7.644928436336885e-06, "loss": 0.0331, "step": 8694 }, { "epoch": 8.82, "learning_rate": 7.642797987138186e-06, "loss": 0.0358, "step": 8695 }, { "epoch": 8.82, "learning_rate": 7.640667651216957e-06, "loss": 0.0418, "step": 8696 }, { "epoch": 8.82, "learning_rate": 7.638537428675581e-06, "loss": 0.0294, "step": 8697 }, { "epoch": 8.83, "learning_rate": 7.636407319616425e-06, "loss": 0.0395, "step": 8698 }, { "epoch": 8.83, "learning_rate": 7.634277324141853e-06, "loss": 0.0315, "step": 8699 }, { "epoch": 8.83, "learning_rate": 7.632147442354218e-06, "loss": 0.0402, "step": 8700 }, { "epoch": 8.83, "learning_rate": 7.630017674355882e-06, "loss": 0.04, "step": 8701 }, { "epoch": 8.83, "learning_rate": 7.627888020249189e-06, "loss": 0.0325, "step": 8702 }, { "epoch": 8.83, "learning_rate": 7.625758480136482e-06, "loss": 0.0402, "step": 8703 }, { "epoch": 8.83, "learning_rate": 7.623629054120094e-06, "loss": 0.0289, "step": 8704 }, { "epoch": 8.83, "learning_rate": 7.621499742302361e-06, "loss": 0.0311, "step": 8705 }, { "epoch": 8.83, "learning_rate": 7.619370544785608e-06, "loss": 0.047, "step": 8706 }, { "epoch": 8.84, "learning_rate": 7.617241461672158e-06, "loss": 0.0314, "step": 8707 }, { "epoch": 8.84, "learning_rate": 7.615112493064324e-06, "loss": 0.0413, "step": 8708 }, { "epoch": 8.84, "learning_rate": 7.612983639064414e-06, "loss": 0.0321, "step": 8709 }, { "epoch": 8.84, "learning_rate": 7.610854899774735e-06, "loss": 0.0338, "step": 8710 }, { "epoch": 8.84, "learning_rate": 7.60872627529758e-06, "loss": 0.0371, "step": 8711 }, { "epoch": 8.84, "learning_rate": 7.606597765735249e-06, "loss": 0.0376, "step": 8712 }, { "epoch": 8.84, "learning_rate": 7.60446937119002e-06, "loss": 0.0368, "step": 8713 }, { "epoch": 8.84, "learning_rate": 7.602341091764186e-06, "loss": 0.0373, "step": 8714 }, { "epoch": 8.84, "learning_rate": 7.600212927560019e-06, "loss": 0.0352, "step": 8715 }, { "epoch": 8.84, "learning_rate": 7.5980848786797885e-06, "loss": 0.0309, "step": 8716 }, { "epoch": 8.85, "learning_rate": 7.595956945225759e-06, "loss": 0.0422, "step": 8717 }, { "epoch": 8.85, "learning_rate": 7.593829127300194e-06, "loss": 0.0349, "step": 8718 }, { "epoch": 8.85, "learning_rate": 7.591701425005345e-06, "loss": 0.0342, "step": 8719 }, { "epoch": 8.85, "learning_rate": 7.589573838443461e-06, "loss": 0.0442, "step": 8720 }, { "epoch": 8.85, "learning_rate": 7.587446367716782e-06, "loss": 0.0311, "step": 8721 }, { "epoch": 8.85, "learning_rate": 7.585319012927552e-06, "loss": 0.034, "step": 8722 }, { "epoch": 8.85, "learning_rate": 7.583191774177999e-06, "loss": 0.0395, "step": 8723 }, { "epoch": 8.85, "learning_rate": 7.581064651570351e-06, "loss": 0.0327, "step": 8724 }, { "epoch": 8.85, "learning_rate": 7.578937645206829e-06, "loss": 0.0358, "step": 8725 }, { "epoch": 8.85, "learning_rate": 7.576810755189647e-06, "loss": 0.0323, "step": 8726 }, { "epoch": 8.86, "learning_rate": 7.574683981621011e-06, "loss": 0.0379, "step": 8727 }, { "epoch": 8.86, "learning_rate": 7.572557324603134e-06, "loss": 0.0302, "step": 8728 }, { "epoch": 8.86, "learning_rate": 7.570430784238207e-06, "loss": 0.0449, "step": 8729 }, { "epoch": 8.86, "learning_rate": 7.568304360628421e-06, "loss": 0.0383, "step": 8730 }, { "epoch": 8.86, "learning_rate": 7.566178053875972e-06, "loss": 0.0339, "step": 8731 }, { "epoch": 8.86, "learning_rate": 7.564051864083037e-06, "loss": 0.0361, "step": 8732 }, { "epoch": 8.86, "learning_rate": 7.561925791351789e-06, "loss": 0.0367, "step": 8733 }, { "epoch": 8.86, "learning_rate": 7.5597998357844046e-06, "loss": 0.0413, "step": 8734 }, { "epoch": 8.86, "learning_rate": 7.5576739974830435e-06, "loss": 0.0334, "step": 8735 }, { "epoch": 8.86, "learning_rate": 7.555548276549868e-06, "loss": 0.0373, "step": 8736 }, { "epoch": 8.87, "learning_rate": 7.55342267308703e-06, "loss": 0.034, "step": 8737 }, { "epoch": 8.87, "learning_rate": 7.551297187196671e-06, "loss": 0.0323, "step": 8738 }, { "epoch": 8.87, "learning_rate": 7.549171818980945e-06, "loss": 0.0316, "step": 8739 }, { "epoch": 8.87, "learning_rate": 7.54704656854198e-06, "loss": 0.0391, "step": 8740 }, { "epoch": 8.87, "learning_rate": 7.544921435981913e-06, "loss": 0.0421, "step": 8741 }, { "epoch": 8.87, "learning_rate": 7.542796421402865e-06, "loss": 0.0339, "step": 8742 }, { "epoch": 8.87, "learning_rate": 7.540671524906956e-06, "loss": 0.0313, "step": 8743 }, { "epoch": 8.87, "learning_rate": 7.538546746596299e-06, "loss": 0.0381, "step": 8744 }, { "epoch": 8.87, "learning_rate": 7.536422086573005e-06, "loss": 0.0311, "step": 8745 }, { "epoch": 8.87, "learning_rate": 7.534297544939171e-06, "loss": 0.0318, "step": 8746 }, { "epoch": 8.88, "learning_rate": 7.5321731217969e-06, "loss": 0.0461, "step": 8747 }, { "epoch": 8.88, "learning_rate": 7.530048817248283e-06, "loss": 0.0464, "step": 8748 }, { "epoch": 8.88, "learning_rate": 7.527924631395402e-06, "loss": 0.0375, "step": 8749 }, { "epoch": 8.88, "learning_rate": 7.525800564340335e-06, "loss": 0.0363, "step": 8750 }, { "epoch": 8.88, "learning_rate": 7.523676616185162e-06, "loss": 0.0371, "step": 8751 }, { "epoch": 8.88, "learning_rate": 7.521552787031947e-06, "loss": 0.0407, "step": 8752 }, { "epoch": 8.88, "learning_rate": 7.5194290769827534e-06, "loss": 0.0554, "step": 8753 }, { "epoch": 8.88, "learning_rate": 7.5173054861396366e-06, "loss": 0.0319, "step": 8754 }, { "epoch": 8.88, "learning_rate": 7.515182014604648e-06, "loss": 0.0371, "step": 8755 }, { "epoch": 8.88, "learning_rate": 7.513058662479835e-06, "loss": 0.0373, "step": 8756 }, { "epoch": 8.89, "learning_rate": 7.510935429867237e-06, "loss": 0.033, "step": 8757 }, { "epoch": 8.89, "learning_rate": 7.508812316868888e-06, "loss": 0.0488, "step": 8758 }, { "epoch": 8.89, "learning_rate": 7.506689323586817e-06, "loss": 0.0307, "step": 8759 }, { "epoch": 8.89, "learning_rate": 7.504566450123042e-06, "loss": 0.0392, "step": 8760 }, { "epoch": 8.89, "learning_rate": 7.502443696579581e-06, "loss": 0.0365, "step": 8761 }, { "epoch": 8.89, "learning_rate": 7.5003210630584485e-06, "loss": 0.0457, "step": 8762 }, { "epoch": 8.89, "learning_rate": 7.498198549661641e-06, "loss": 0.0358, "step": 8763 }, { "epoch": 8.89, "learning_rate": 7.496076156491171e-06, "loss": 0.0373, "step": 8764 }, { "epoch": 8.89, "learning_rate": 7.493953883649023e-06, "loss": 0.0409, "step": 8765 }, { "epoch": 8.89, "learning_rate": 7.491831731237188e-06, "loss": 0.0459, "step": 8766 }, { "epoch": 8.9, "learning_rate": 7.489709699357644e-06, "loss": 0.0308, "step": 8767 }, { "epoch": 8.9, "learning_rate": 7.487587788112372e-06, "loss": 0.0394, "step": 8768 }, { "epoch": 8.9, "learning_rate": 7.485465997603338e-06, "loss": 0.0392, "step": 8769 }, { "epoch": 8.9, "learning_rate": 7.483344327932512e-06, "loss": 0.0368, "step": 8770 }, { "epoch": 8.9, "learning_rate": 7.481222779201845e-06, "loss": 0.0417, "step": 8771 }, { "epoch": 8.9, "learning_rate": 7.479101351513296e-06, "loss": 0.0359, "step": 8772 }, { "epoch": 8.9, "learning_rate": 7.476980044968809e-06, "loss": 0.0351, "step": 8773 }, { "epoch": 8.9, "learning_rate": 7.47485885967033e-06, "loss": 0.0367, "step": 8774 }, { "epoch": 8.9, "learning_rate": 7.472737795719791e-06, "loss": 0.0389, "step": 8775 }, { "epoch": 8.91, "learning_rate": 7.470616853219122e-06, "loss": 0.0406, "step": 8776 }, { "epoch": 8.91, "learning_rate": 7.468496032270245e-06, "loss": 0.0346, "step": 8777 }, { "epoch": 8.91, "learning_rate": 7.46637533297508e-06, "loss": 0.0396, "step": 8778 }, { "epoch": 8.91, "learning_rate": 7.46425475543554e-06, "loss": 0.0347, "step": 8779 }, { "epoch": 8.91, "learning_rate": 7.462134299753527e-06, "loss": 0.0445, "step": 8780 }, { "epoch": 8.91, "learning_rate": 7.460013966030948e-06, "loss": 0.0394, "step": 8781 }, { "epoch": 8.91, "learning_rate": 7.457893754369694e-06, "loss": 0.0349, "step": 8782 }, { "epoch": 8.91, "learning_rate": 7.455773664871654e-06, "loss": 0.0317, "step": 8783 }, { "epoch": 8.91, "learning_rate": 7.453653697638707e-06, "loss": 0.0437, "step": 8784 }, { "epoch": 8.91, "learning_rate": 7.451533852772739e-06, "loss": 0.044, "step": 8785 }, { "epoch": 8.92, "learning_rate": 7.449414130375612e-06, "loss": 0.0308, "step": 8786 }, { "epoch": 8.92, "learning_rate": 7.447294530549196e-06, "loss": 0.0503, "step": 8787 }, { "epoch": 8.92, "learning_rate": 7.445175053395346e-06, "loss": 0.0335, "step": 8788 }, { "epoch": 8.92, "learning_rate": 7.443055699015922e-06, "loss": 0.0365, "step": 8789 }, { "epoch": 8.92, "learning_rate": 7.4409364675127645e-06, "loss": 0.0401, "step": 8790 }, { "epoch": 8.92, "learning_rate": 7.438817358987722e-06, "loss": 0.0344, "step": 8791 }, { "epoch": 8.92, "learning_rate": 7.4366983735426265e-06, "loss": 0.0336, "step": 8792 }, { "epoch": 8.92, "learning_rate": 7.434579511279308e-06, "loss": 0.0362, "step": 8793 }, { "epoch": 8.92, "learning_rate": 7.432460772299588e-06, "loss": 0.0401, "step": 8794 }, { "epoch": 8.92, "learning_rate": 7.43034215670529e-06, "loss": 0.0356, "step": 8795 }, { "epoch": 8.93, "learning_rate": 7.428223664598222e-06, "loss": 0.0337, "step": 8796 }, { "epoch": 8.93, "learning_rate": 7.426105296080186e-06, "loss": 0.0384, "step": 8797 }, { "epoch": 8.93, "learning_rate": 7.42398705125299e-06, "loss": 0.0447, "step": 8798 }, { "epoch": 8.93, "learning_rate": 7.4218689302184276e-06, "loss": 0.0275, "step": 8799 }, { "epoch": 8.93, "learning_rate": 7.4197509330782805e-06, "loss": 0.0378, "step": 8800 }, { "epoch": 8.93, "learning_rate": 7.417633059934337e-06, "loss": 0.0476, "step": 8801 }, { "epoch": 8.93, "learning_rate": 7.415515310888373e-06, "loss": 0.0364, "step": 8802 }, { "epoch": 8.93, "learning_rate": 7.413397686042156e-06, "loss": 0.0278, "step": 8803 }, { "epoch": 8.93, "learning_rate": 7.4112801854974515e-06, "loss": 0.0362, "step": 8804 }, { "epoch": 8.93, "learning_rate": 7.409162809356014e-06, "loss": 0.0362, "step": 8805 }, { "epoch": 8.94, "learning_rate": 7.407045557719602e-06, "loss": 0.0306, "step": 8806 }, { "epoch": 8.94, "learning_rate": 7.404928430689962e-06, "loss": 0.0301, "step": 8807 }, { "epoch": 8.94, "learning_rate": 7.402811428368832e-06, "loss": 0.0405, "step": 8808 }, { "epoch": 8.94, "learning_rate": 7.400694550857948e-06, "loss": 0.0505, "step": 8809 }, { "epoch": 8.94, "learning_rate": 7.398577798259035e-06, "loss": 0.0388, "step": 8810 }, { "epoch": 8.94, "learning_rate": 7.396461170673818e-06, "loss": 0.0344, "step": 8811 }, { "epoch": 8.94, "learning_rate": 7.394344668204013e-06, "loss": 0.0309, "step": 8812 }, { "epoch": 8.94, "learning_rate": 7.392228290951327e-06, "loss": 0.0451, "step": 8813 }, { "epoch": 8.94, "learning_rate": 7.390112039017474e-06, "loss": 0.0426, "step": 8814 }, { "epoch": 8.94, "learning_rate": 7.3879959125041445e-06, "loss": 0.0331, "step": 8815 }, { "epoch": 8.95, "learning_rate": 7.385879911513033e-06, "loss": 0.0383, "step": 8816 }, { "epoch": 8.95, "learning_rate": 7.383764036145824e-06, "loss": 0.0368, "step": 8817 }, { "epoch": 8.95, "learning_rate": 7.381648286504201e-06, "loss": 0.0355, "step": 8818 }, { "epoch": 8.95, "learning_rate": 7.379532662689837e-06, "loss": 0.0351, "step": 8819 }, { "epoch": 8.95, "learning_rate": 7.377417164804399e-06, "loss": 0.0351, "step": 8820 }, { "epoch": 8.95, "learning_rate": 7.375301792949549e-06, "loss": 0.0377, "step": 8821 }, { "epoch": 8.95, "learning_rate": 7.373186547226943e-06, "loss": 0.0371, "step": 8822 }, { "epoch": 8.95, "learning_rate": 7.371071427738234e-06, "loss": 0.0343, "step": 8823 }, { "epoch": 8.95, "learning_rate": 7.368956434585065e-06, "loss": 0.0413, "step": 8824 }, { "epoch": 8.95, "learning_rate": 7.3668415678690745e-06, "loss": 0.0319, "step": 8825 }, { "epoch": 8.96, "learning_rate": 7.36472682769189e-06, "loss": 0.0323, "step": 8826 }, { "epoch": 8.96, "learning_rate": 7.36261221415514e-06, "loss": 0.0329, "step": 8827 }, { "epoch": 8.96, "learning_rate": 7.360497727360447e-06, "loss": 0.0409, "step": 8828 }, { "epoch": 8.96, "learning_rate": 7.358383367409421e-06, "loss": 0.0324, "step": 8829 }, { "epoch": 8.96, "learning_rate": 7.3562691344036665e-06, "loss": 0.0375, "step": 8830 }, { "epoch": 8.96, "learning_rate": 7.354155028444793e-06, "loss": 0.0328, "step": 8831 }, { "epoch": 8.96, "learning_rate": 7.352041049634391e-06, "loss": 0.0406, "step": 8832 }, { "epoch": 8.96, "learning_rate": 7.349927198074051e-06, "loss": 0.0301, "step": 8833 }, { "epoch": 8.96, "learning_rate": 7.347813473865353e-06, "loss": 0.0304, "step": 8834 }, { "epoch": 8.96, "learning_rate": 7.345699877109877e-06, "loss": 0.0351, "step": 8835 }, { "epoch": 8.97, "learning_rate": 7.343586407909195e-06, "loss": 0.0397, "step": 8836 }, { "epoch": 8.97, "learning_rate": 7.341473066364868e-06, "loss": 0.0421, "step": 8837 }, { "epoch": 8.97, "learning_rate": 7.339359852578452e-06, "loss": 0.0309, "step": 8838 }, { "epoch": 8.97, "learning_rate": 7.337246766651509e-06, "loss": 0.0416, "step": 8839 }, { "epoch": 8.97, "learning_rate": 7.3351338086855776e-06, "loss": 0.044, "step": 8840 }, { "epoch": 8.97, "learning_rate": 7.3330209787822015e-06, "loss": 0.0319, "step": 8841 }, { "epoch": 8.97, "learning_rate": 7.330908277042914e-06, "loss": 0.039, "step": 8842 }, { "epoch": 8.97, "learning_rate": 7.328795703569241e-06, "loss": 0.0351, "step": 8843 }, { "epoch": 8.97, "learning_rate": 7.326683258462703e-06, "loss": 0.0382, "step": 8844 }, { "epoch": 8.98, "learning_rate": 7.3245709418248226e-06, "loss": 0.036, "step": 8845 }, { "epoch": 8.98, "learning_rate": 7.322458753757102e-06, "loss": 0.0467, "step": 8846 }, { "epoch": 8.98, "learning_rate": 7.320346694361042e-06, "loss": 0.037, "step": 8847 }, { "epoch": 8.98, "learning_rate": 7.31823476373815e-06, "loss": 0.0329, "step": 8848 }, { "epoch": 8.98, "learning_rate": 7.3161229619899095e-06, "loss": 0.0268, "step": 8849 }, { "epoch": 8.98, "learning_rate": 7.314011289217805e-06, "loss": 0.0363, "step": 8850 }, { "epoch": 8.98, "learning_rate": 7.311899745523319e-06, "loss": 0.0363, "step": 8851 }, { "epoch": 8.98, "learning_rate": 7.309788331007919e-06, "loss": 0.0293, "step": 8852 }, { "epoch": 8.98, "learning_rate": 7.307677045773074e-06, "loss": 0.033, "step": 8853 }, { "epoch": 8.98, "learning_rate": 7.305565889920242e-06, "loss": 0.0345, "step": 8854 }, { "epoch": 8.99, "learning_rate": 7.303454863550873e-06, "loss": 0.0304, "step": 8855 }, { "epoch": 8.99, "learning_rate": 7.301343966766421e-06, "loss": 0.0409, "step": 8856 }, { "epoch": 8.99, "learning_rate": 7.299233199668325e-06, "loss": 0.0334, "step": 8857 }, { "epoch": 8.99, "learning_rate": 7.297122562358021e-06, "loss": 0.037, "step": 8858 }, { "epoch": 8.99, "learning_rate": 7.295012054936934e-06, "loss": 0.0418, "step": 8859 }, { "epoch": 8.99, "learning_rate": 7.292901677506489e-06, "loss": 0.0363, "step": 8860 }, { "epoch": 8.99, "learning_rate": 7.290791430168099e-06, "loss": 0.044, "step": 8861 }, { "epoch": 8.99, "learning_rate": 7.288681313023178e-06, "loss": 0.0383, "step": 8862 }, { "epoch": 8.99, "learning_rate": 7.2865713261731244e-06, "loss": 0.0439, "step": 8863 }, { "epoch": 8.99, "learning_rate": 7.284461469719342e-06, "loss": 0.0503, "step": 8864 }, { "epoch": 9.0, "learning_rate": 7.282351743763219e-06, "loss": 0.036, "step": 8865 }, { "epoch": 9.0, "learning_rate": 7.280242148406138e-06, "loss": 0.0406, "step": 8866 }, { "epoch": 9.0, "learning_rate": 7.278132683749479e-06, "loss": 0.0294, "step": 8867 }, { "epoch": 9.0, "learning_rate": 7.276023349894616e-06, "loss": 0.0389, "step": 8868 }, { "epoch": 9.0, "learning_rate": 7.273914146942914e-06, "loss": 0.0277, "step": 8869 }, { "epoch": 9.0, "learning_rate": 7.271805074995733e-06, "loss": 0.0405, "step": 8870 }, { "epoch": 9.0, "learning_rate": 7.2696961341544215e-06, "loss": 0.028, "step": 8871 }, { "epoch": 9.0, "learning_rate": 7.267587324520331e-06, "loss": 0.028, "step": 8872 }, { "epoch": 9.0, "learning_rate": 7.265478646194802e-06, "loss": 0.0245, "step": 8873 }, { "epoch": 9.0, "learning_rate": 7.263370099279173e-06, "loss": 0.0211, "step": 8874 }, { "epoch": 9.01, "learning_rate": 7.261261683874765e-06, "loss": 0.0242, "step": 8875 }, { "epoch": 9.01, "learning_rate": 7.259153400082904e-06, "loss": 0.0321, "step": 8876 }, { "epoch": 9.01, "learning_rate": 7.257045248004902e-06, "loss": 0.0257, "step": 8877 }, { "epoch": 9.01, "learning_rate": 7.254937227742071e-06, "loss": 0.0236, "step": 8878 }, { "epoch": 9.01, "learning_rate": 7.252829339395715e-06, "loss": 0.0214, "step": 8879 }, { "epoch": 9.01, "learning_rate": 7.250721583067123e-06, "loss": 0.0209, "step": 8880 }, { "epoch": 9.01, "learning_rate": 7.248613958857596e-06, "loss": 0.0243, "step": 8881 }, { "epoch": 9.01, "learning_rate": 7.246506466868412e-06, "loss": 0.026, "step": 8882 }, { "epoch": 9.01, "learning_rate": 7.244399107200848e-06, "loss": 0.0272, "step": 8883 }, { "epoch": 9.01, "learning_rate": 7.242291879956176e-06, "loss": 0.0314, "step": 8884 }, { "epoch": 9.02, "learning_rate": 7.240184785235662e-06, "loss": 0.0153, "step": 8885 }, { "epoch": 9.02, "learning_rate": 7.238077823140563e-06, "loss": 0.0242, "step": 8886 }, { "epoch": 9.02, "learning_rate": 7.2359709937721315e-06, "loss": 0.0218, "step": 8887 }, { "epoch": 9.02, "learning_rate": 7.23386429723161e-06, "loss": 0.0267, "step": 8888 }, { "epoch": 9.02, "learning_rate": 7.23175773362024e-06, "loss": 0.0294, "step": 8889 }, { "epoch": 9.02, "learning_rate": 7.229651303039255e-06, "loss": 0.0239, "step": 8890 }, { "epoch": 9.02, "learning_rate": 7.227545005589884e-06, "loss": 0.0264, "step": 8891 }, { "epoch": 9.02, "learning_rate": 7.225438841373343e-06, "loss": 0.0254, "step": 8892 }, { "epoch": 9.02, "learning_rate": 7.223332810490848e-06, "loss": 0.0312, "step": 8893 }, { "epoch": 9.02, "learning_rate": 7.221226913043602e-06, "loss": 0.03, "step": 8894 }, { "epoch": 9.03, "learning_rate": 7.219121149132811e-06, "loss": 0.0278, "step": 8895 }, { "epoch": 9.03, "learning_rate": 7.217015518859667e-06, "loss": 0.0206, "step": 8896 }, { "epoch": 9.03, "learning_rate": 7.214910022325354e-06, "loss": 0.0361, "step": 8897 }, { "epoch": 9.03, "learning_rate": 7.212804659631063e-06, "loss": 0.0245, "step": 8898 }, { "epoch": 9.03, "learning_rate": 7.210699430877963e-06, "loss": 0.0271, "step": 8899 }, { "epoch": 9.03, "learning_rate": 7.208594336167221e-06, "loss": 0.0247, "step": 8900 }, { "epoch": 9.03, "learning_rate": 7.206489375600004e-06, "loss": 0.0373, "step": 8901 }, { "epoch": 9.03, "learning_rate": 7.204384549277466e-06, "loss": 0.0275, "step": 8902 }, { "epoch": 9.03, "learning_rate": 7.202279857300756e-06, "loss": 0.0288, "step": 8903 }, { "epoch": 9.04, "learning_rate": 7.200175299771015e-06, "loss": 0.0232, "step": 8904 }, { "epoch": 9.04, "learning_rate": 7.198070876789378e-06, "loss": 0.0233, "step": 8905 }, { "epoch": 9.04, "learning_rate": 7.195966588456982e-06, "loss": 0.0316, "step": 8906 }, { "epoch": 9.04, "learning_rate": 7.193862434874945e-06, "loss": 0.0269, "step": 8907 }, { "epoch": 9.04, "learning_rate": 7.191758416144388e-06, "loss": 0.0217, "step": 8908 }, { "epoch": 9.04, "learning_rate": 7.189654532366418e-06, "loss": 0.0218, "step": 8909 }, { "epoch": 9.04, "learning_rate": 7.187550783642141e-06, "loss": 0.0212, "step": 8910 }, { "epoch": 9.04, "learning_rate": 7.1854471700726505e-06, "loss": 0.0241, "step": 8911 }, { "epoch": 9.04, "learning_rate": 7.183343691759044e-06, "loss": 0.0253, "step": 8912 }, { "epoch": 9.04, "learning_rate": 7.1812403488024004e-06, "loss": 0.0224, "step": 8913 }, { "epoch": 9.05, "learning_rate": 7.179137141303798e-06, "loss": 0.0178, "step": 8914 }, { "epoch": 9.05, "learning_rate": 7.177034069364315e-06, "loss": 0.0257, "step": 8915 }, { "epoch": 9.05, "learning_rate": 7.174931133085011e-06, "loss": 0.0476, "step": 8916 }, { "epoch": 9.05, "learning_rate": 7.172828332566942e-06, "loss": 0.035, "step": 8917 }, { "epoch": 9.05, "learning_rate": 7.170725667911167e-06, "loss": 0.0234, "step": 8918 }, { "epoch": 9.05, "learning_rate": 7.168623139218726e-06, "loss": 0.0253, "step": 8919 }, { "epoch": 9.05, "learning_rate": 7.166520746590662e-06, "loss": 0.0237, "step": 8920 }, { "epoch": 9.05, "learning_rate": 7.164418490128002e-06, "loss": 0.0269, "step": 8921 }, { "epoch": 9.05, "learning_rate": 7.162316369931775e-06, "loss": 0.0256, "step": 8922 }, { "epoch": 9.05, "learning_rate": 7.160214386103001e-06, "loss": 0.0259, "step": 8923 }, { "epoch": 9.06, "learning_rate": 7.158112538742692e-06, "loss": 0.0348, "step": 8924 }, { "epoch": 9.06, "learning_rate": 7.156010827951856e-06, "loss": 0.0203, "step": 8925 }, { "epoch": 9.06, "learning_rate": 7.1539092538314906e-06, "loss": 0.0245, "step": 8926 }, { "epoch": 9.06, "learning_rate": 7.151807816482588e-06, "loss": 0.0272, "step": 8927 }, { "epoch": 9.06, "learning_rate": 7.149706516006134e-06, "loss": 0.0286, "step": 8928 }, { "epoch": 9.06, "learning_rate": 7.147605352503112e-06, "loss": 0.0273, "step": 8929 }, { "epoch": 9.06, "learning_rate": 7.145504326074488e-06, "loss": 0.0209, "step": 8930 }, { "epoch": 9.06, "learning_rate": 7.143403436821241e-06, "loss": 0.028, "step": 8931 }, { "epoch": 9.06, "learning_rate": 7.141302684844322e-06, "loss": 0.0304, "step": 8932 }, { "epoch": 9.06, "learning_rate": 7.139202070244689e-06, "loss": 0.0266, "step": 8933 }, { "epoch": 9.07, "learning_rate": 7.137101593123283e-06, "loss": 0.029, "step": 8934 }, { "epoch": 9.07, "learning_rate": 7.135001253581051e-06, "loss": 0.0281, "step": 8935 }, { "epoch": 9.07, "learning_rate": 7.132901051718923e-06, "loss": 0.0246, "step": 8936 }, { "epoch": 9.07, "learning_rate": 7.130800987637828e-06, "loss": 0.0279, "step": 8937 }, { "epoch": 9.07, "learning_rate": 7.128701061438683e-06, "loss": 0.0254, "step": 8938 }, { "epoch": 9.07, "learning_rate": 7.126601273222404e-06, "loss": 0.0263, "step": 8939 }, { "epoch": 9.07, "learning_rate": 7.124501623089898e-06, "loss": 0.0309, "step": 8940 }, { "epoch": 9.07, "learning_rate": 7.122402111142068e-06, "loss": 0.0257, "step": 8941 }, { "epoch": 9.07, "learning_rate": 7.120302737479807e-06, "loss": 0.0251, "step": 8942 }, { "epoch": 9.07, "learning_rate": 7.118203502204001e-06, "loss": 0.02, "step": 8943 }, { "epoch": 9.08, "learning_rate": 7.116104405415529e-06, "loss": 0.0205, "step": 8944 }, { "epoch": 9.08, "learning_rate": 7.114005447215268e-06, "loss": 0.0269, "step": 8945 }, { "epoch": 9.08, "learning_rate": 7.111906627704086e-06, "loss": 0.0235, "step": 8946 }, { "epoch": 9.08, "learning_rate": 7.109807946982836e-06, "loss": 0.0282, "step": 8947 }, { "epoch": 9.08, "learning_rate": 7.107709405152383e-06, "loss": 0.0382, "step": 8948 }, { "epoch": 9.08, "learning_rate": 7.105611002313568e-06, "loss": 0.0287, "step": 8949 }, { "epoch": 9.08, "learning_rate": 7.1035127385672335e-06, "loss": 0.0241, "step": 8950 }, { "epoch": 9.08, "learning_rate": 7.1014146140142125e-06, "loss": 0.0233, "step": 8951 }, { "epoch": 9.08, "learning_rate": 7.099316628755333e-06, "loss": 0.0335, "step": 8952 }, { "epoch": 9.08, "learning_rate": 7.097218782891416e-06, "loss": 0.0254, "step": 8953 }, { "epoch": 9.09, "learning_rate": 7.095121076523275e-06, "loss": 0.0265, "step": 8954 }, { "epoch": 9.09, "learning_rate": 7.0930235097517106e-06, "loss": 0.0276, "step": 8955 }, { "epoch": 9.09, "learning_rate": 7.090926082677536e-06, "loss": 0.0258, "step": 8956 }, { "epoch": 9.09, "learning_rate": 7.0888287954015344e-06, "loss": 0.0239, "step": 8957 }, { "epoch": 9.09, "learning_rate": 7.086731648024501e-06, "loss": 0.0283, "step": 8958 }, { "epoch": 9.09, "learning_rate": 7.08463464064721e-06, "loss": 0.0225, "step": 8959 }, { "epoch": 9.09, "learning_rate": 7.082537773370437e-06, "loss": 0.0193, "step": 8960 }, { "epoch": 9.09, "learning_rate": 7.080441046294948e-06, "loss": 0.0275, "step": 8961 }, { "epoch": 9.09, "learning_rate": 7.078344459521505e-06, "loss": 0.026, "step": 8962 }, { "epoch": 9.09, "learning_rate": 7.0762480131508605e-06, "loss": 0.0213, "step": 8963 }, { "epoch": 9.1, "learning_rate": 7.074151707283756e-06, "loss": 0.0346, "step": 8964 }, { "epoch": 9.1, "learning_rate": 7.0720555420209415e-06, "loss": 0.0195, "step": 8965 }, { "epoch": 9.1, "learning_rate": 7.069959517463145e-06, "loss": 0.0313, "step": 8966 }, { "epoch": 9.1, "learning_rate": 7.0678636337110895e-06, "loss": 0.0227, "step": 8967 }, { "epoch": 9.1, "learning_rate": 7.0657678908655e-06, "loss": 0.0243, "step": 8968 }, { "epoch": 9.1, "learning_rate": 7.0636722890270885e-06, "loss": 0.0208, "step": 8969 }, { "epoch": 9.1, "learning_rate": 7.0615768282965595e-06, "loss": 0.0222, "step": 8970 }, { "epoch": 9.1, "learning_rate": 7.059481508774609e-06, "loss": 0.0242, "step": 8971 }, { "epoch": 9.1, "learning_rate": 7.057386330561935e-06, "loss": 0.0217, "step": 8972 }, { "epoch": 9.11, "learning_rate": 7.055291293759219e-06, "loss": 0.0243, "step": 8973 }, { "epoch": 9.11, "learning_rate": 7.053196398467145e-06, "loss": 0.028, "step": 8974 }, { "epoch": 9.11, "learning_rate": 7.051101644786383e-06, "loss": 0.0256, "step": 8975 }, { "epoch": 9.11, "learning_rate": 7.049007032817595e-06, "loss": 0.028, "step": 8976 }, { "epoch": 9.11, "learning_rate": 7.046912562661443e-06, "loss": 0.0319, "step": 8977 }, { "epoch": 9.11, "learning_rate": 7.0448182344185755e-06, "loss": 0.0245, "step": 8978 }, { "epoch": 9.11, "learning_rate": 7.042724048189642e-06, "loss": 0.0261, "step": 8979 }, { "epoch": 9.11, "learning_rate": 7.040630004075273e-06, "loss": 0.0253, "step": 8980 }, { "epoch": 9.11, "learning_rate": 7.038536102176108e-06, "loss": 0.0365, "step": 8981 }, { "epoch": 9.11, "learning_rate": 7.036442342592769e-06, "loss": 0.0242, "step": 8982 }, { "epoch": 9.12, "learning_rate": 7.034348725425872e-06, "loss": 0.0297, "step": 8983 }, { "epoch": 9.12, "learning_rate": 7.032255250776026e-06, "loss": 0.0292, "step": 8984 }, { "epoch": 9.12, "learning_rate": 7.030161918743838e-06, "loss": 0.0358, "step": 8985 }, { "epoch": 9.12, "learning_rate": 7.0280687294299066e-06, "loss": 0.023, "step": 8986 }, { "epoch": 9.12, "learning_rate": 7.025975682934816e-06, "loss": 0.0405, "step": 8987 }, { "epoch": 9.12, "learning_rate": 7.023882779359152e-06, "loss": 0.0254, "step": 8988 }, { "epoch": 9.12, "learning_rate": 7.0217900188034915e-06, "loss": 0.0291, "step": 8989 }, { "epoch": 9.12, "learning_rate": 7.0196974013684035e-06, "loss": 0.0259, "step": 8990 }, { "epoch": 9.12, "learning_rate": 7.017604927154453e-06, "loss": 0.0202, "step": 8991 }, { "epoch": 9.12, "learning_rate": 7.015512596262195e-06, "loss": 0.0272, "step": 8992 }, { "epoch": 9.13, "learning_rate": 7.013420408792177e-06, "loss": 0.0225, "step": 8993 }, { "epoch": 9.13, "learning_rate": 7.011328364844938e-06, "loss": 0.032, "step": 8994 }, { "epoch": 9.13, "learning_rate": 7.009236464521021e-06, "loss": 0.025, "step": 8995 }, { "epoch": 9.13, "learning_rate": 7.0071447079209474e-06, "loss": 0.0203, "step": 8996 }, { "epoch": 9.13, "learning_rate": 7.0050530951452355e-06, "loss": 0.0214, "step": 8997 }, { "epoch": 9.13, "learning_rate": 7.002961626294411e-06, "loss": 0.0333, "step": 8998 }, { "epoch": 9.13, "learning_rate": 7.000870301468974e-06, "loss": 0.0295, "step": 8999 }, { "epoch": 9.13, "learning_rate": 6.9987791207694255e-06, "loss": 0.0238, "step": 9000 }, { "epoch": 9.13, "learning_rate": 6.996688084296257e-06, "loss": 0.0206, "step": 9001 }, { "epoch": 9.13, "learning_rate": 6.994597192149961e-06, "loss": 0.0244, "step": 9002 }, { "epoch": 9.14, "learning_rate": 6.992506444431013e-06, "loss": 0.0282, "step": 9003 }, { "epoch": 9.14, "learning_rate": 6.990415841239888e-06, "loss": 0.0321, "step": 9004 }, { "epoch": 9.14, "learning_rate": 6.988325382677046e-06, "loss": 0.0282, "step": 9005 }, { "epoch": 9.14, "learning_rate": 6.986235068842953e-06, "loss": 0.0238, "step": 9006 }, { "epoch": 9.14, "learning_rate": 6.984144899838057e-06, "loss": 0.0244, "step": 9007 }, { "epoch": 9.14, "learning_rate": 6.982054875762807e-06, "loss": 0.0235, "step": 9008 }, { "epoch": 9.14, "learning_rate": 6.979964996717638e-06, "loss": 0.0249, "step": 9009 }, { "epoch": 9.14, "learning_rate": 6.977875262802981e-06, "loss": 0.0256, "step": 9010 }, { "epoch": 9.14, "learning_rate": 6.9757856741192585e-06, "loss": 0.0235, "step": 9011 }, { "epoch": 9.14, "learning_rate": 6.973696230766891e-06, "loss": 0.0273, "step": 9012 }, { "epoch": 9.15, "learning_rate": 6.971606932846289e-06, "loss": 0.03, "step": 9013 }, { "epoch": 9.15, "learning_rate": 6.969517780457846e-06, "loss": 0.0262, "step": 9014 }, { "epoch": 9.15, "learning_rate": 6.967428773701973e-06, "loss": 0.0207, "step": 9015 }, { "epoch": 9.15, "learning_rate": 6.96533991267905e-06, "loss": 0.0245, "step": 9016 }, { "epoch": 9.15, "learning_rate": 6.96325119748946e-06, "loss": 0.0227, "step": 9017 }, { "epoch": 9.15, "learning_rate": 6.961162628233579e-06, "loss": 0.0278, "step": 9018 }, { "epoch": 9.15, "learning_rate": 6.9590742050117775e-06, "loss": 0.0228, "step": 9019 }, { "epoch": 9.15, "learning_rate": 6.9569859279244136e-06, "loss": 0.0271, "step": 9020 }, { "epoch": 9.15, "learning_rate": 6.9548977970718424e-06, "loss": 0.0209, "step": 9021 }, { "epoch": 9.15, "learning_rate": 6.952809812554404e-06, "loss": 0.0253, "step": 9022 }, { "epoch": 9.16, "learning_rate": 6.95072197447245e-06, "loss": 0.0206, "step": 9023 }, { "epoch": 9.16, "learning_rate": 6.948634282926308e-06, "loss": 0.0222, "step": 9024 }, { "epoch": 9.16, "learning_rate": 6.946546738016304e-06, "loss": 0.0273, "step": 9025 }, { "epoch": 9.16, "learning_rate": 6.944459339842757e-06, "loss": 0.025, "step": 9026 }, { "epoch": 9.16, "learning_rate": 6.9423720885059796e-06, "loss": 0.0223, "step": 9027 }, { "epoch": 9.16, "learning_rate": 6.940284984106274e-06, "loss": 0.0316, "step": 9028 }, { "epoch": 9.16, "learning_rate": 6.938198026743941e-06, "loss": 0.0245, "step": 9029 }, { "epoch": 9.16, "learning_rate": 6.93611121651927e-06, "loss": 0.0247, "step": 9030 }, { "epoch": 9.16, "learning_rate": 6.934024553532539e-06, "loss": 0.0248, "step": 9031 }, { "epoch": 9.16, "learning_rate": 6.931938037884036e-06, "loss": 0.0287, "step": 9032 }, { "epoch": 9.17, "learning_rate": 6.9298516696740225e-06, "loss": 0.0251, "step": 9033 }, { "epoch": 9.17, "learning_rate": 6.927765449002762e-06, "loss": 0.0272, "step": 9034 }, { "epoch": 9.17, "learning_rate": 6.925679375970511e-06, "loss": 0.0233, "step": 9035 }, { "epoch": 9.17, "learning_rate": 6.923593450677517e-06, "loss": 0.0326, "step": 9036 }, { "epoch": 9.17, "learning_rate": 6.9215076732240215e-06, "loss": 0.0268, "step": 9037 }, { "epoch": 9.17, "learning_rate": 6.9194220437102556e-06, "loss": 0.0364, "step": 9038 }, { "epoch": 9.17, "learning_rate": 6.917336562236448e-06, "loss": 0.0279, "step": 9039 }, { "epoch": 9.17, "learning_rate": 6.915251228902819e-06, "loss": 0.0217, "step": 9040 }, { "epoch": 9.17, "learning_rate": 6.913166043809583e-06, "loss": 0.0336, "step": 9041 }, { "epoch": 9.18, "learning_rate": 6.911081007056943e-06, "loss": 0.0227, "step": 9042 }, { "epoch": 9.18, "learning_rate": 6.908996118745097e-06, "loss": 0.0271, "step": 9043 }, { "epoch": 9.18, "learning_rate": 6.9069113789742356e-06, "loss": 0.0219, "step": 9044 }, { "epoch": 9.18, "learning_rate": 6.904826787844546e-06, "loss": 0.0347, "step": 9045 }, { "epoch": 9.18, "learning_rate": 6.902742345456204e-06, "loss": 0.027, "step": 9046 }, { "epoch": 9.18, "learning_rate": 6.900658051909375e-06, "loss": 0.0258, "step": 9047 }, { "epoch": 9.18, "learning_rate": 6.898573907304227e-06, "loss": 0.0309, "step": 9048 }, { "epoch": 9.18, "learning_rate": 6.896489911740915e-06, "loss": 0.0283, "step": 9049 }, { "epoch": 9.18, "learning_rate": 6.894406065319586e-06, "loss": 0.0249, "step": 9050 }, { "epoch": 9.18, "learning_rate": 6.8923223681403795e-06, "loss": 0.0218, "step": 9051 }, { "epoch": 9.19, "learning_rate": 6.890238820303432e-06, "loss": 0.0305, "step": 9052 }, { "epoch": 9.19, "learning_rate": 6.888155421908871e-06, "loss": 0.0303, "step": 9053 }, { "epoch": 9.19, "learning_rate": 6.886072173056813e-06, "loss": 0.0279, "step": 9054 }, { "epoch": 9.19, "learning_rate": 6.88398907384737e-06, "loss": 0.0215, "step": 9055 }, { "epoch": 9.19, "learning_rate": 6.881906124380648e-06, "loss": 0.0307, "step": 9056 }, { "epoch": 9.19, "learning_rate": 6.879823324756748e-06, "loss": 0.0464, "step": 9057 }, { "epoch": 9.19, "learning_rate": 6.877740675075759e-06, "loss": 0.0236, "step": 9058 }, { "epoch": 9.19, "learning_rate": 6.8756581754377675e-06, "loss": 0.0284, "step": 9059 }, { "epoch": 9.19, "learning_rate": 6.873575825942844e-06, "loss": 0.0293, "step": 9060 }, { "epoch": 9.19, "learning_rate": 6.871493626691059e-06, "loss": 0.031, "step": 9061 }, { "epoch": 9.2, "learning_rate": 6.869411577782479e-06, "loss": 0.0238, "step": 9062 }, { "epoch": 9.2, "learning_rate": 6.8673296793171555e-06, "loss": 0.0244, "step": 9063 }, { "epoch": 9.2, "learning_rate": 6.865247931395132e-06, "loss": 0.033, "step": 9064 }, { "epoch": 9.2, "learning_rate": 6.863166334116457e-06, "loss": 0.0297, "step": 9065 }, { "epoch": 9.2, "learning_rate": 6.86108488758116e-06, "loss": 0.0269, "step": 9066 }, { "epoch": 9.2, "learning_rate": 6.859003591889264e-06, "loss": 0.0341, "step": 9067 }, { "epoch": 9.2, "learning_rate": 6.856922447140792e-06, "loss": 0.0242, "step": 9068 }, { "epoch": 9.2, "learning_rate": 6.854841453435753e-06, "loss": 0.0239, "step": 9069 }, { "epoch": 9.2, "learning_rate": 6.8527606108741515e-06, "loss": 0.0246, "step": 9070 }, { "epoch": 9.2, "learning_rate": 6.850679919555985e-06, "loss": 0.0294, "step": 9071 }, { "epoch": 9.21, "learning_rate": 6.848599379581237e-06, "loss": 0.0243, "step": 9072 }, { "epoch": 9.21, "learning_rate": 6.8465189910498966e-06, "loss": 0.0234, "step": 9073 }, { "epoch": 9.21, "learning_rate": 6.844438754061939e-06, "loss": 0.0246, "step": 9074 }, { "epoch": 9.21, "learning_rate": 6.8423586687173305e-06, "loss": 0.025, "step": 9075 }, { "epoch": 9.21, "learning_rate": 6.840278735116031e-06, "loss": 0.0326, "step": 9076 }, { "epoch": 9.21, "learning_rate": 6.8381989533579925e-06, "loss": 0.0274, "step": 9077 }, { "epoch": 9.21, "learning_rate": 6.83611932354316e-06, "loss": 0.0317, "step": 9078 }, { "epoch": 9.21, "learning_rate": 6.834039845771475e-06, "loss": 0.0224, "step": 9079 }, { "epoch": 9.21, "learning_rate": 6.831960520142868e-06, "loss": 0.0342, "step": 9080 }, { "epoch": 9.21, "learning_rate": 6.829881346757259e-06, "loss": 0.0314, "step": 9081 }, { "epoch": 9.22, "learning_rate": 6.82780232571457e-06, "loss": 0.0404, "step": 9082 }, { "epoch": 9.22, "learning_rate": 6.825723457114709e-06, "loss": 0.0265, "step": 9083 }, { "epoch": 9.22, "learning_rate": 6.823644741057576e-06, "loss": 0.027, "step": 9084 }, { "epoch": 9.22, "learning_rate": 6.821566177643066e-06, "loss": 0.0279, "step": 9085 }, { "epoch": 9.22, "learning_rate": 6.8194877669710695e-06, "loss": 0.0255, "step": 9086 }, { "epoch": 9.22, "learning_rate": 6.817409509141462e-06, "loss": 0.0293, "step": 9087 }, { "epoch": 9.22, "learning_rate": 6.815331404254116e-06, "loss": 0.0261, "step": 9088 }, { "epoch": 9.22, "learning_rate": 6.8132534524088964e-06, "loss": 0.0254, "step": 9089 }, { "epoch": 9.22, "learning_rate": 6.811175653705666e-06, "loss": 0.026, "step": 9090 }, { "epoch": 9.22, "learning_rate": 6.809098008244273e-06, "loss": 0.0225, "step": 9091 }, { "epoch": 9.23, "learning_rate": 6.807020516124561e-06, "loss": 0.0292, "step": 9092 }, { "epoch": 9.23, "learning_rate": 6.804943177446364e-06, "loss": 0.0248, "step": 9093 }, { "epoch": 9.23, "learning_rate": 6.802865992309508e-06, "loss": 0.0348, "step": 9094 }, { "epoch": 9.23, "learning_rate": 6.80078896081382e-06, "loss": 0.0297, "step": 9095 }, { "epoch": 9.23, "learning_rate": 6.798712083059112e-06, "loss": 0.0271, "step": 9096 }, { "epoch": 9.23, "learning_rate": 6.796635359145183e-06, "loss": 0.0268, "step": 9097 }, { "epoch": 9.23, "learning_rate": 6.794558789171841e-06, "loss": 0.0282, "step": 9098 }, { "epoch": 9.23, "learning_rate": 6.792482373238875e-06, "loss": 0.0297, "step": 9099 }, { "epoch": 9.23, "learning_rate": 6.79040611144607e-06, "loss": 0.0268, "step": 9100 }, { "epoch": 9.23, "learning_rate": 6.7883300038931975e-06, "loss": 0.0276, "step": 9101 }, { "epoch": 9.24, "learning_rate": 6.7862540506800325e-06, "loss": 0.0248, "step": 9102 }, { "epoch": 9.24, "learning_rate": 6.784178251906334e-06, "loss": 0.0309, "step": 9103 }, { "epoch": 9.24, "learning_rate": 6.782102607671857e-06, "loss": 0.0192, "step": 9104 }, { "epoch": 9.24, "learning_rate": 6.780027118076346e-06, "loss": 0.0221, "step": 9105 }, { "epoch": 9.24, "learning_rate": 6.777951783219543e-06, "loss": 0.0253, "step": 9106 }, { "epoch": 9.24, "learning_rate": 6.775876603201179e-06, "loss": 0.0253, "step": 9107 }, { "epoch": 9.24, "learning_rate": 6.773801578120983e-06, "loss": 0.0268, "step": 9108 }, { "epoch": 9.24, "learning_rate": 6.771726708078669e-06, "loss": 0.0199, "step": 9109 }, { "epoch": 9.24, "learning_rate": 6.7696519931739446e-06, "loss": 0.0237, "step": 9110 }, { "epoch": 9.25, "learning_rate": 6.767577433506514e-06, "loss": 0.0267, "step": 9111 }, { "epoch": 9.25, "learning_rate": 6.7655030291760725e-06, "loss": 0.0282, "step": 9112 }, { "epoch": 9.25, "learning_rate": 6.763428780282309e-06, "loss": 0.0254, "step": 9113 }, { "epoch": 9.25, "learning_rate": 6.761354686924895e-06, "loss": 0.0286, "step": 9114 }, { "epoch": 9.25, "learning_rate": 6.759280749203514e-06, "loss": 0.0321, "step": 9115 }, { "epoch": 9.25, "learning_rate": 6.757206967217828e-06, "loss": 0.0204, "step": 9116 }, { "epoch": 9.25, "learning_rate": 6.755133341067489e-06, "loss": 0.0252, "step": 9117 }, { "epoch": 9.25, "learning_rate": 6.753059870852153e-06, "loss": 0.0221, "step": 9118 }, { "epoch": 9.25, "learning_rate": 6.750986556671462e-06, "loss": 0.0264, "step": 9119 }, { "epoch": 9.25, "learning_rate": 6.748913398625046e-06, "loss": 0.0278, "step": 9120 }, { "epoch": 9.26, "learning_rate": 6.746840396812538e-06, "loss": 0.0227, "step": 9121 }, { "epoch": 9.26, "learning_rate": 6.7447675513335505e-06, "loss": 0.0295, "step": 9122 }, { "epoch": 9.26, "learning_rate": 6.742694862287707e-06, "loss": 0.0191, "step": 9123 }, { "epoch": 9.26, "learning_rate": 6.7406223297746035e-06, "loss": 0.024, "step": 9124 }, { "epoch": 9.26, "learning_rate": 6.738549953893843e-06, "loss": 0.0314, "step": 9125 }, { "epoch": 9.26, "learning_rate": 6.736477734745012e-06, "loss": 0.0241, "step": 9126 }, { "epoch": 9.26, "learning_rate": 6.734405672427693e-06, "loss": 0.0301, "step": 9127 }, { "epoch": 9.26, "learning_rate": 6.7323337670414615e-06, "loss": 0.0249, "step": 9128 }, { "epoch": 9.26, "learning_rate": 6.730262018685887e-06, "loss": 0.0257, "step": 9129 }, { "epoch": 9.26, "learning_rate": 6.728190427460526e-06, "loss": 0.0246, "step": 9130 }, { "epoch": 9.27, "learning_rate": 6.7261189934649275e-06, "loss": 0.0297, "step": 9131 }, { "epoch": 9.27, "learning_rate": 6.724047716798645e-06, "loss": 0.0253, "step": 9132 }, { "epoch": 9.27, "learning_rate": 6.721976597561211e-06, "loss": 0.0272, "step": 9133 }, { "epoch": 9.27, "learning_rate": 6.719905635852153e-06, "loss": 0.0352, "step": 9134 }, { "epoch": 9.27, "learning_rate": 6.7178348317709975e-06, "loss": 0.0265, "step": 9135 }, { "epoch": 9.27, "learning_rate": 6.715764185417256e-06, "loss": 0.0212, "step": 9136 }, { "epoch": 9.27, "learning_rate": 6.7136936968904355e-06, "loss": 0.0279, "step": 9137 }, { "epoch": 9.27, "learning_rate": 6.711623366290033e-06, "loss": 0.0237, "step": 9138 }, { "epoch": 9.27, "learning_rate": 6.709553193715543e-06, "loss": 0.0281, "step": 9139 }, { "epoch": 9.27, "learning_rate": 6.707483179266449e-06, "loss": 0.0225, "step": 9140 }, { "epoch": 9.28, "learning_rate": 6.705413323042229e-06, "loss": 0.0233, "step": 9141 }, { "epoch": 9.28, "learning_rate": 6.7033436251423515e-06, "loss": 0.0233, "step": 9142 }, { "epoch": 9.28, "learning_rate": 6.701274085666274e-06, "loss": 0.0325, "step": 9143 }, { "epoch": 9.28, "learning_rate": 6.699204704713454e-06, "loss": 0.024, "step": 9144 }, { "epoch": 9.28, "learning_rate": 6.697135482383334e-06, "loss": 0.0266, "step": 9145 }, { "epoch": 9.28, "learning_rate": 6.695066418775356e-06, "loss": 0.0299, "step": 9146 }, { "epoch": 9.28, "learning_rate": 6.692997513988948e-06, "loss": 0.0226, "step": 9147 }, { "epoch": 9.28, "learning_rate": 6.69092876812353e-06, "loss": 0.024, "step": 9148 }, { "epoch": 9.28, "learning_rate": 6.688860181278527e-06, "loss": 0.0284, "step": 9149 }, { "epoch": 9.28, "learning_rate": 6.686791753553341e-06, "loss": 0.0245, "step": 9150 }, { "epoch": 9.29, "learning_rate": 6.68472348504737e-06, "loss": 0.0242, "step": 9151 }, { "epoch": 9.29, "learning_rate": 6.682655375860012e-06, "loss": 0.0188, "step": 9152 }, { "epoch": 9.29, "learning_rate": 6.6805874260906486e-06, "loss": 0.0295, "step": 9153 }, { "epoch": 9.29, "learning_rate": 6.678519635838657e-06, "loss": 0.0247, "step": 9154 }, { "epoch": 9.29, "learning_rate": 6.6764520052034054e-06, "loss": 0.0227, "step": 9155 }, { "epoch": 9.29, "learning_rate": 6.6743845342842575e-06, "loss": 0.021, "step": 9156 }, { "epoch": 9.29, "learning_rate": 6.672317223180567e-06, "loss": 0.0314, "step": 9157 }, { "epoch": 9.29, "learning_rate": 6.670250071991682e-06, "loss": 0.0228, "step": 9158 }, { "epoch": 9.29, "learning_rate": 6.668183080816939e-06, "loss": 0.0307, "step": 9159 }, { "epoch": 9.29, "learning_rate": 6.666116249755672e-06, "loss": 0.0251, "step": 9160 }, { "epoch": 9.3, "learning_rate": 6.6640495789072e-06, "loss": 0.0211, "step": 9161 }, { "epoch": 9.3, "learning_rate": 6.661983068370843e-06, "loss": 0.0279, "step": 9162 }, { "epoch": 9.3, "learning_rate": 6.659916718245909e-06, "loss": 0.0322, "step": 9163 }, { "epoch": 9.3, "learning_rate": 6.65785052863169e-06, "loss": 0.027, "step": 9164 }, { "epoch": 9.3, "learning_rate": 6.655784499627491e-06, "loss": 0.0324, "step": 9165 }, { "epoch": 9.3, "learning_rate": 6.653718631332591e-06, "loss": 0.0252, "step": 9166 }, { "epoch": 9.3, "learning_rate": 6.6516529238462645e-06, "loss": 0.0271, "step": 9167 }, { "epoch": 9.3, "learning_rate": 6.649587377267786e-06, "loss": 0.0306, "step": 9168 }, { "epoch": 9.3, "learning_rate": 6.647521991696415e-06, "loss": 0.0266, "step": 9169 }, { "epoch": 9.3, "learning_rate": 6.6454567672314066e-06, "loss": 0.0216, "step": 9170 }, { "epoch": 9.31, "learning_rate": 6.643391703972004e-06, "loss": 0.0335, "step": 9171 }, { "epoch": 9.31, "learning_rate": 6.641326802017447e-06, "loss": 0.0228, "step": 9172 }, { "epoch": 9.31, "learning_rate": 6.639262061466966e-06, "loss": 0.0332, "step": 9173 }, { "epoch": 9.31, "learning_rate": 6.637197482419786e-06, "loss": 0.0234, "step": 9174 }, { "epoch": 9.31, "learning_rate": 6.635133064975121e-06, "loss": 0.0374, "step": 9175 }, { "epoch": 9.31, "learning_rate": 6.63306880923218e-06, "loss": 0.0285, "step": 9176 }, { "epoch": 9.31, "learning_rate": 6.631004715290161e-06, "loss": 0.028, "step": 9177 }, { "epoch": 9.31, "learning_rate": 6.628940783248255e-06, "loss": 0.0307, "step": 9178 }, { "epoch": 9.31, "learning_rate": 6.626877013205648e-06, "loss": 0.0326, "step": 9179 }, { "epoch": 9.32, "learning_rate": 6.624813405261515e-06, "loss": 0.0361, "step": 9180 }, { "epoch": 9.32, "learning_rate": 6.622749959515022e-06, "loss": 0.0244, "step": 9181 }, { "epoch": 9.32, "learning_rate": 6.620686676065338e-06, "loss": 0.0215, "step": 9182 }, { "epoch": 9.32, "learning_rate": 6.618623555011608e-06, "loss": 0.03, "step": 9183 }, { "epoch": 9.32, "learning_rate": 6.61656059645298e-06, "loss": 0.0278, "step": 9184 }, { "epoch": 9.32, "learning_rate": 6.614497800488593e-06, "loss": 0.0346, "step": 9185 }, { "epoch": 9.32, "learning_rate": 6.612435167217575e-06, "loss": 0.0213, "step": 9186 }, { "epoch": 9.32, "learning_rate": 6.610372696739047e-06, "loss": 0.0232, "step": 9187 }, { "epoch": 9.32, "learning_rate": 6.6083103891521206e-06, "loss": 0.0261, "step": 9188 }, { "epoch": 9.32, "learning_rate": 6.606248244555905e-06, "loss": 0.0228, "step": 9189 }, { "epoch": 9.33, "learning_rate": 6.604186263049499e-06, "loss": 0.0337, "step": 9190 }, { "epoch": 9.33, "learning_rate": 6.602124444731993e-06, "loss": 0.0312, "step": 9191 }, { "epoch": 9.33, "learning_rate": 6.600062789702469e-06, "loss": 0.0282, "step": 9192 }, { "epoch": 9.33, "learning_rate": 6.598001298060002e-06, "loss": 0.0326, "step": 9193 }, { "epoch": 9.33, "learning_rate": 6.595939969903657e-06, "loss": 0.0292, "step": 9194 }, { "epoch": 9.33, "learning_rate": 6.593878805332492e-06, "loss": 0.0213, "step": 9195 }, { "epoch": 9.33, "learning_rate": 6.591817804445564e-06, "loss": 0.0289, "step": 9196 }, { "epoch": 9.33, "learning_rate": 6.589756967341912e-06, "loss": 0.0332, "step": 9197 }, { "epoch": 9.33, "learning_rate": 6.587696294120567e-06, "loss": 0.028, "step": 9198 }, { "epoch": 9.33, "learning_rate": 6.585635784880566e-06, "loss": 0.0289, "step": 9199 }, { "epoch": 9.34, "learning_rate": 6.583575439720923e-06, "loss": 0.0285, "step": 9200 }, { "epoch": 9.34, "learning_rate": 6.5815152587406495e-06, "loss": 0.0266, "step": 9201 }, { "epoch": 9.34, "learning_rate": 6.579455242038753e-06, "loss": 0.0279, "step": 9202 }, { "epoch": 9.34, "learning_rate": 6.5773953897142275e-06, "loss": 0.0256, "step": 9203 }, { "epoch": 9.34, "learning_rate": 6.575335701866059e-06, "loss": 0.0321, "step": 9204 }, { "epoch": 9.34, "learning_rate": 6.573276178593229e-06, "loss": 0.0254, "step": 9205 }, { "epoch": 9.34, "learning_rate": 6.571216819994709e-06, "loss": 0.0273, "step": 9206 }, { "epoch": 9.34, "learning_rate": 6.569157626169464e-06, "loss": 0.0323, "step": 9207 }, { "epoch": 9.34, "learning_rate": 6.567098597216455e-06, "loss": 0.0284, "step": 9208 }, { "epoch": 9.34, "learning_rate": 6.565039733234625e-06, "loss": 0.0385, "step": 9209 }, { "epoch": 9.35, "learning_rate": 6.562981034322916e-06, "loss": 0.0275, "step": 9210 }, { "epoch": 9.35, "learning_rate": 6.560922500580258e-06, "loss": 0.0296, "step": 9211 }, { "epoch": 9.35, "learning_rate": 6.5588641321055826e-06, "loss": 0.0345, "step": 9212 }, { "epoch": 9.35, "learning_rate": 6.5568059289978005e-06, "loss": 0.0263, "step": 9213 }, { "epoch": 9.35, "learning_rate": 6.554747891355817e-06, "loss": 0.0291, "step": 9214 }, { "epoch": 9.35, "learning_rate": 6.552690019278544e-06, "loss": 0.0252, "step": 9215 }, { "epoch": 9.35, "learning_rate": 6.550632312864869e-06, "loss": 0.0242, "step": 9216 }, { "epoch": 9.35, "learning_rate": 6.548574772213672e-06, "loss": 0.0288, "step": 9217 }, { "epoch": 9.35, "learning_rate": 6.5465173974238394e-06, "loss": 0.0308, "step": 9218 }, { "epoch": 9.35, "learning_rate": 6.544460188594232e-06, "loss": 0.0306, "step": 9219 }, { "epoch": 9.36, "learning_rate": 6.542403145823717e-06, "loss": 0.0278, "step": 9220 }, { "epoch": 9.36, "learning_rate": 6.540346269211144e-06, "loss": 0.0271, "step": 9221 }, { "epoch": 9.36, "learning_rate": 6.538289558855358e-06, "loss": 0.03, "step": 9222 }, { "epoch": 9.36, "learning_rate": 6.536233014855193e-06, "loss": 0.0265, "step": 9223 }, { "epoch": 9.36, "learning_rate": 6.534176637309485e-06, "loss": 0.0228, "step": 9224 }, { "epoch": 9.36, "learning_rate": 6.532120426317052e-06, "loss": 0.0347, "step": 9225 }, { "epoch": 9.36, "learning_rate": 6.5300643819767084e-06, "loss": 0.0264, "step": 9226 }, { "epoch": 9.36, "learning_rate": 6.528008504387258e-06, "loss": 0.0284, "step": 9227 }, { "epoch": 9.36, "learning_rate": 6.525952793647496e-06, "loss": 0.0306, "step": 9228 }, { "epoch": 9.36, "learning_rate": 6.523897249856214e-06, "loss": 0.0265, "step": 9229 }, { "epoch": 9.37, "learning_rate": 6.521841873112191e-06, "loss": 0.0267, "step": 9230 }, { "epoch": 9.37, "learning_rate": 6.519786663514199e-06, "loss": 0.0222, "step": 9231 }, { "epoch": 9.37, "learning_rate": 6.51773162116101e-06, "loss": 0.0317, "step": 9232 }, { "epoch": 9.37, "learning_rate": 6.5156767461513735e-06, "loss": 0.0213, "step": 9233 }, { "epoch": 9.37, "learning_rate": 6.51362203858404e-06, "loss": 0.0264, "step": 9234 }, { "epoch": 9.37, "learning_rate": 6.5115674985577535e-06, "loss": 0.0214, "step": 9235 }, { "epoch": 9.37, "learning_rate": 6.509513126171245e-06, "loss": 0.0293, "step": 9236 }, { "epoch": 9.37, "learning_rate": 6.507458921523237e-06, "loss": 0.0271, "step": 9237 }, { "epoch": 9.37, "learning_rate": 6.505404884712447e-06, "loss": 0.0287, "step": 9238 }, { "epoch": 9.37, "learning_rate": 6.5033510158375844e-06, "loss": 0.0322, "step": 9239 }, { "epoch": 9.38, "learning_rate": 6.50129731499735e-06, "loss": 0.0294, "step": 9240 }, { "epoch": 9.38, "learning_rate": 6.4992437822904385e-06, "loss": 0.0327, "step": 9241 }, { "epoch": 9.38, "learning_rate": 6.497190417815531e-06, "loss": 0.0268, "step": 9242 }, { "epoch": 9.38, "learning_rate": 6.495137221671306e-06, "loss": 0.0283, "step": 9243 }, { "epoch": 9.38, "learning_rate": 6.493084193956429e-06, "loss": 0.0229, "step": 9244 }, { "epoch": 9.38, "learning_rate": 6.49103133476956e-06, "loss": 0.0333, "step": 9245 }, { "epoch": 9.38, "learning_rate": 6.488978644209354e-06, "loss": 0.0231, "step": 9246 }, { "epoch": 9.38, "learning_rate": 6.486926122374453e-06, "loss": 0.0267, "step": 9247 }, { "epoch": 9.38, "learning_rate": 6.4848737693634885e-06, "loss": 0.0262, "step": 9248 }, { "epoch": 9.39, "learning_rate": 6.482821585275098e-06, "loss": 0.029, "step": 9249 }, { "epoch": 9.39, "learning_rate": 6.480769570207897e-06, "loss": 0.0213, "step": 9250 }, { "epoch": 9.39, "learning_rate": 6.478717724260491e-06, "loss": 0.0242, "step": 9251 }, { "epoch": 9.39, "learning_rate": 6.476666047531492e-06, "loss": 0.0368, "step": 9252 }, { "epoch": 9.39, "learning_rate": 6.4746145401194906e-06, "loss": 0.028, "step": 9253 }, { "epoch": 9.39, "learning_rate": 6.4725632021230745e-06, "loss": 0.0397, "step": 9254 }, { "epoch": 9.39, "learning_rate": 6.470512033640821e-06, "loss": 0.0272, "step": 9255 }, { "epoch": 9.39, "learning_rate": 6.468461034771303e-06, "loss": 0.029, "step": 9256 }, { "epoch": 9.39, "learning_rate": 6.466410205613082e-06, "loss": 0.0307, "step": 9257 }, { "epoch": 9.39, "learning_rate": 6.464359546264715e-06, "loss": 0.0265, "step": 9258 }, { "epoch": 9.4, "learning_rate": 6.462309056824746e-06, "loss": 0.0236, "step": 9259 }, { "epoch": 9.4, "learning_rate": 6.460258737391713e-06, "loss": 0.0396, "step": 9260 }, { "epoch": 9.4, "learning_rate": 6.4582085880641455e-06, "loss": 0.0243, "step": 9261 }, { "epoch": 9.4, "learning_rate": 6.456158608940568e-06, "loss": 0.02, "step": 9262 }, { "epoch": 9.4, "learning_rate": 6.454108800119493e-06, "loss": 0.0304, "step": 9263 }, { "epoch": 9.4, "learning_rate": 6.452059161699425e-06, "loss": 0.0216, "step": 9264 }, { "epoch": 9.4, "learning_rate": 6.450009693778856e-06, "loss": 0.0274, "step": 9265 }, { "epoch": 9.4, "learning_rate": 6.4479603964562855e-06, "loss": 0.0305, "step": 9266 }, { "epoch": 9.4, "learning_rate": 6.445911269830189e-06, "loss": 0.0312, "step": 9267 }, { "epoch": 9.4, "learning_rate": 6.4438623139990375e-06, "loss": 0.0311, "step": 9268 }, { "epoch": 9.41, "learning_rate": 6.441813529061301e-06, "loss": 0.0314, "step": 9269 }, { "epoch": 9.41, "learning_rate": 6.43976491511543e-06, "loss": 0.0288, "step": 9270 }, { "epoch": 9.41, "learning_rate": 6.437716472259874e-06, "loss": 0.0254, "step": 9271 }, { "epoch": 9.41, "learning_rate": 6.435668200593073e-06, "loss": 0.032, "step": 9272 }, { "epoch": 9.41, "learning_rate": 6.4336201002134576e-06, "loss": 0.0249, "step": 9273 }, { "epoch": 9.41, "learning_rate": 6.431572171219453e-06, "loss": 0.0271, "step": 9274 }, { "epoch": 9.41, "learning_rate": 6.429524413709476e-06, "loss": 0.0241, "step": 9275 }, { "epoch": 9.41, "learning_rate": 6.4274768277819315e-06, "loss": 0.028, "step": 9276 }, { "epoch": 9.41, "learning_rate": 6.425429413535217e-06, "loss": 0.024, "step": 9277 }, { "epoch": 9.41, "learning_rate": 6.423382171067721e-06, "loss": 0.0366, "step": 9278 }, { "epoch": 9.42, "learning_rate": 6.421335100477831e-06, "loss": 0.026, "step": 9279 }, { "epoch": 9.42, "learning_rate": 6.419288201863918e-06, "loss": 0.0266, "step": 9280 }, { "epoch": 9.42, "learning_rate": 6.417241475324342e-06, "loss": 0.0368, "step": 9281 }, { "epoch": 9.42, "learning_rate": 6.415194920957473e-06, "loss": 0.0244, "step": 9282 }, { "epoch": 9.42, "learning_rate": 6.413148538861651e-06, "loss": 0.0282, "step": 9283 }, { "epoch": 9.42, "learning_rate": 6.411102329135216e-06, "loss": 0.0314, "step": 9284 }, { "epoch": 9.42, "learning_rate": 6.409056291876506e-06, "loss": 0.0255, "step": 9285 }, { "epoch": 9.42, "learning_rate": 6.407010427183842e-06, "loss": 0.0249, "step": 9286 }, { "epoch": 9.42, "learning_rate": 6.404964735155539e-06, "loss": 0.0276, "step": 9287 }, { "epoch": 9.42, "learning_rate": 6.4029192158899064e-06, "loss": 0.0287, "step": 9288 }, { "epoch": 9.43, "learning_rate": 6.400873869485242e-06, "loss": 0.0256, "step": 9289 }, { "epoch": 9.43, "learning_rate": 6.398828696039834e-06, "loss": 0.0276, "step": 9290 }, { "epoch": 9.43, "learning_rate": 6.396783695651973e-06, "loss": 0.0451, "step": 9291 }, { "epoch": 9.43, "learning_rate": 6.39473886841993e-06, "loss": 0.0256, "step": 9292 }, { "epoch": 9.43, "learning_rate": 6.392694214441968e-06, "loss": 0.0294, "step": 9293 }, { "epoch": 9.43, "learning_rate": 6.390649733816348e-06, "loss": 0.0253, "step": 9294 }, { "epoch": 9.43, "learning_rate": 6.388605426641316e-06, "loss": 0.0234, "step": 9295 }, { "epoch": 9.43, "learning_rate": 6.386561293015116e-06, "loss": 0.0246, "step": 9296 }, { "epoch": 9.43, "learning_rate": 6.38451733303598e-06, "loss": 0.0274, "step": 9297 }, { "epoch": 9.43, "learning_rate": 6.382473546802128e-06, "loss": 0.0279, "step": 9298 }, { "epoch": 9.44, "learning_rate": 6.3804299344117845e-06, "loss": 0.0235, "step": 9299 }, { "epoch": 9.44, "learning_rate": 6.378386495963154e-06, "loss": 0.0277, "step": 9300 }, { "epoch": 9.44, "learning_rate": 6.376343231554431e-06, "loss": 0.0242, "step": 9301 }, { "epoch": 9.44, "learning_rate": 6.374300141283812e-06, "loss": 0.0326, "step": 9302 }, { "epoch": 9.44, "learning_rate": 6.372257225249478e-06, "loss": 0.0307, "step": 9303 }, { "epoch": 9.44, "learning_rate": 6.370214483549601e-06, "loss": 0.0261, "step": 9304 }, { "epoch": 9.44, "learning_rate": 6.368171916282349e-06, "loss": 0.0316, "step": 9305 }, { "epoch": 9.44, "learning_rate": 6.366129523545877e-06, "loss": 0.0342, "step": 9306 }, { "epoch": 9.44, "learning_rate": 6.364087305438336e-06, "loss": 0.0258, "step": 9307 }, { "epoch": 9.44, "learning_rate": 6.362045262057869e-06, "loss": 0.03, "step": 9308 }, { "epoch": 9.45, "learning_rate": 6.360003393502605e-06, "loss": 0.0242, "step": 9309 }, { "epoch": 9.45, "learning_rate": 6.3579616998706694e-06, "loss": 0.0265, "step": 9310 }, { "epoch": 9.45, "learning_rate": 6.355920181260175e-06, "loss": 0.0289, "step": 9311 }, { "epoch": 9.45, "learning_rate": 6.353878837769232e-06, "loss": 0.0266, "step": 9312 }, { "epoch": 9.45, "learning_rate": 6.351837669495938e-06, "loss": 0.03, "step": 9313 }, { "epoch": 9.45, "learning_rate": 6.3497966765383834e-06, "loss": 0.025, "step": 9314 }, { "epoch": 9.45, "learning_rate": 6.347755858994644e-06, "loss": 0.0249, "step": 9315 }, { "epoch": 9.45, "learning_rate": 6.345715216962805e-06, "loss": 0.0305, "step": 9316 }, { "epoch": 9.45, "learning_rate": 6.343674750540922e-06, "loss": 0.0296, "step": 9317 }, { "epoch": 9.46, "learning_rate": 6.341634459827053e-06, "loss": 0.032, "step": 9318 }, { "epoch": 9.46, "learning_rate": 6.339594344919251e-06, "loss": 0.0347, "step": 9319 }, { "epoch": 9.46, "learning_rate": 6.337554405915549e-06, "loss": 0.0435, "step": 9320 }, { "epoch": 9.46, "learning_rate": 6.335514642913983e-06, "loss": 0.0263, "step": 9321 }, { "epoch": 9.46, "learning_rate": 6.333475056012572e-06, "loss": 0.0252, "step": 9322 }, { "epoch": 9.46, "learning_rate": 6.331435645309332e-06, "loss": 0.029, "step": 9323 }, { "epoch": 9.46, "learning_rate": 6.3293964109022665e-06, "loss": 0.0286, "step": 9324 }, { "epoch": 9.46, "learning_rate": 6.327357352889379e-06, "loss": 0.0427, "step": 9325 }, { "epoch": 9.46, "learning_rate": 6.3253184713686525e-06, "loss": 0.0371, "step": 9326 }, { "epoch": 9.46, "learning_rate": 6.323279766438068e-06, "loss": 0.0305, "step": 9327 }, { "epoch": 9.47, "learning_rate": 6.321241238195599e-06, "loss": 0.0203, "step": 9328 }, { "epoch": 9.47, "learning_rate": 6.319202886739208e-06, "loss": 0.0241, "step": 9329 }, { "epoch": 9.47, "learning_rate": 6.317164712166851e-06, "loss": 0.0419, "step": 9330 }, { "epoch": 9.47, "learning_rate": 6.3151267145764695e-06, "loss": 0.0336, "step": 9331 }, { "epoch": 9.47, "learning_rate": 6.313088894066007e-06, "loss": 0.0254, "step": 9332 }, { "epoch": 9.47, "learning_rate": 6.311051250733393e-06, "loss": 0.0296, "step": 9333 }, { "epoch": 9.47, "learning_rate": 6.309013784676543e-06, "loss": 0.0315, "step": 9334 }, { "epoch": 9.47, "learning_rate": 6.306976495993373e-06, "loss": 0.0341, "step": 9335 }, { "epoch": 9.47, "learning_rate": 6.304939384781788e-06, "loss": 0.0251, "step": 9336 }, { "epoch": 9.47, "learning_rate": 6.302902451139681e-06, "loss": 0.0281, "step": 9337 }, { "epoch": 9.48, "learning_rate": 6.300865695164939e-06, "loss": 0.0288, "step": 9338 }, { "epoch": 9.48, "learning_rate": 6.298829116955439e-06, "loss": 0.0302, "step": 9339 }, { "epoch": 9.48, "learning_rate": 6.296792716609047e-06, "loss": 0.0383, "step": 9340 }, { "epoch": 9.48, "learning_rate": 6.294756494223635e-06, "loss": 0.026, "step": 9341 }, { "epoch": 9.48, "learning_rate": 6.292720449897048e-06, "loss": 0.0236, "step": 9342 }, { "epoch": 9.48, "learning_rate": 6.290684583727132e-06, "loss": 0.0278, "step": 9343 }, { "epoch": 9.48, "learning_rate": 6.288648895811722e-06, "loss": 0.0261, "step": 9344 }, { "epoch": 9.48, "learning_rate": 6.2866133862486424e-06, "loss": 0.0243, "step": 9345 }, { "epoch": 9.48, "learning_rate": 6.284578055135715e-06, "loss": 0.0341, "step": 9346 }, { "epoch": 9.48, "learning_rate": 6.28254290257075e-06, "loss": 0.036, "step": 9347 }, { "epoch": 9.49, "learning_rate": 6.2805079286515405e-06, "loss": 0.0301, "step": 9348 }, { "epoch": 9.49, "learning_rate": 6.2784731334758885e-06, "loss": 0.039, "step": 9349 }, { "epoch": 9.49, "learning_rate": 6.276438517141575e-06, "loss": 0.0334, "step": 9350 }, { "epoch": 9.49, "learning_rate": 6.274404079746374e-06, "loss": 0.0284, "step": 9351 }, { "epoch": 9.49, "learning_rate": 6.272369821388054e-06, "loss": 0.0304, "step": 9352 }, { "epoch": 9.49, "learning_rate": 6.270335742164373e-06, "loss": 0.0333, "step": 9353 }, { "epoch": 9.49, "learning_rate": 6.268301842173078e-06, "loss": 0.0298, "step": 9354 }, { "epoch": 9.49, "learning_rate": 6.266268121511911e-06, "loss": 0.0271, "step": 9355 }, { "epoch": 9.49, "learning_rate": 6.264234580278603e-06, "loss": 0.0315, "step": 9356 }, { "epoch": 9.49, "learning_rate": 6.2622012185708815e-06, "loss": 0.038, "step": 9357 }, { "epoch": 9.5, "learning_rate": 6.260168036486461e-06, "loss": 0.0263, "step": 9358 }, { "epoch": 9.5, "learning_rate": 6.258135034123046e-06, "loss": 0.0294, "step": 9359 }, { "epoch": 9.5, "learning_rate": 6.256102211578334e-06, "loss": 0.03, "step": 9360 }, { "epoch": 9.5, "learning_rate": 6.254069568950017e-06, "loss": 0.0278, "step": 9361 }, { "epoch": 9.5, "learning_rate": 6.25203710633577e-06, "loss": 0.0279, "step": 9362 }, { "epoch": 9.5, "learning_rate": 6.25000482383327e-06, "loss": 0.0262, "step": 9363 }, { "epoch": 9.5, "learning_rate": 6.247972721540179e-06, "loss": 0.033, "step": 9364 }, { "epoch": 9.5, "learning_rate": 6.245940799554147e-06, "loss": 0.0324, "step": 9365 }, { "epoch": 9.5, "learning_rate": 6.243909057972826e-06, "loss": 0.0266, "step": 9366 }, { "epoch": 9.5, "learning_rate": 6.241877496893851e-06, "loss": 0.0335, "step": 9367 }, { "epoch": 9.51, "learning_rate": 6.23984611641485e-06, "loss": 0.0262, "step": 9368 }, { "epoch": 9.51, "learning_rate": 6.237814916633444e-06, "loss": 0.029, "step": 9369 }, { "epoch": 9.51, "learning_rate": 6.235783897647244e-06, "loss": 0.0355, "step": 9370 }, { "epoch": 9.51, "learning_rate": 6.233753059553851e-06, "loss": 0.0266, "step": 9371 }, { "epoch": 9.51, "learning_rate": 6.231722402450858e-06, "loss": 0.0267, "step": 9372 }, { "epoch": 9.51, "learning_rate": 6.229691926435852e-06, "loss": 0.0244, "step": 9373 }, { "epoch": 9.51, "learning_rate": 6.227661631606408e-06, "loss": 0.0206, "step": 9374 }, { "epoch": 9.51, "learning_rate": 6.225631518060098e-06, "loss": 0.0321, "step": 9375 }, { "epoch": 9.51, "learning_rate": 6.223601585894476e-06, "loss": 0.028, "step": 9376 }, { "epoch": 9.51, "learning_rate": 6.221571835207094e-06, "loss": 0.0225, "step": 9377 }, { "epoch": 9.52, "learning_rate": 6.219542266095493e-06, "loss": 0.0329, "step": 9378 }, { "epoch": 9.52, "learning_rate": 6.217512878657206e-06, "loss": 0.0257, "step": 9379 }, { "epoch": 9.52, "learning_rate": 6.215483672989757e-06, "loss": 0.0269, "step": 9380 }, { "epoch": 9.52, "learning_rate": 6.213454649190664e-06, "loss": 0.0282, "step": 9381 }, { "epoch": 9.52, "learning_rate": 6.211425807357424e-06, "loss": 0.0369, "step": 9382 }, { "epoch": 9.52, "learning_rate": 6.209397147587548e-06, "loss": 0.0312, "step": 9383 }, { "epoch": 9.52, "learning_rate": 6.207368669978515e-06, "loss": 0.0269, "step": 9384 }, { "epoch": 9.52, "learning_rate": 6.205340374627813e-06, "loss": 0.0278, "step": 9385 }, { "epoch": 9.52, "learning_rate": 6.203312261632908e-06, "loss": 0.0307, "step": 9386 }, { "epoch": 9.53, "learning_rate": 6.201284331091266e-06, "loss": 0.0262, "step": 9387 }, { "epoch": 9.53, "learning_rate": 6.199256583100339e-06, "loss": 0.0315, "step": 9388 }, { "epoch": 9.53, "learning_rate": 6.197229017757571e-06, "loss": 0.0316, "step": 9389 }, { "epoch": 9.53, "learning_rate": 6.195201635160401e-06, "loss": 0.0277, "step": 9390 }, { "epoch": 9.53, "learning_rate": 6.193174435406255e-06, "loss": 0.0348, "step": 9391 }, { "epoch": 9.53, "learning_rate": 6.191147418592556e-06, "loss": 0.0234, "step": 9392 }, { "epoch": 9.53, "learning_rate": 6.18912058481671e-06, "loss": 0.0274, "step": 9393 }, { "epoch": 9.53, "learning_rate": 6.187093934176121e-06, "loss": 0.0327, "step": 9394 }, { "epoch": 9.53, "learning_rate": 6.1850674667681765e-06, "loss": 0.0248, "step": 9395 }, { "epoch": 9.53, "learning_rate": 6.183041182690266e-06, "loss": 0.0367, "step": 9396 }, { "epoch": 9.54, "learning_rate": 6.181015082039763e-06, "loss": 0.0268, "step": 9397 }, { "epoch": 9.54, "learning_rate": 6.178989164914028e-06, "loss": 0.0236, "step": 9398 }, { "epoch": 9.54, "learning_rate": 6.176963431410428e-06, "loss": 0.0298, "step": 9399 }, { "epoch": 9.54, "learning_rate": 6.174937881626304e-06, "loss": 0.0265, "step": 9400 }, { "epoch": 9.54, "learning_rate": 6.172912515658997e-06, "loss": 0.032, "step": 9401 }, { "epoch": 9.54, "learning_rate": 6.170887333605843e-06, "loss": 0.0281, "step": 9402 }, { "epoch": 9.54, "learning_rate": 6.1688623355641584e-06, "loss": 0.0311, "step": 9403 }, { "epoch": 9.54, "learning_rate": 6.166837521631257e-06, "loss": 0.026, "step": 9404 }, { "epoch": 9.54, "learning_rate": 6.164812891904442e-06, "loss": 0.0279, "step": 9405 }, { "epoch": 9.54, "learning_rate": 6.162788446481014e-06, "loss": 0.0369, "step": 9406 }, { "epoch": 9.55, "learning_rate": 6.160764185458251e-06, "loss": 0.0361, "step": 9407 }, { "epoch": 9.55, "learning_rate": 6.15874010893344e-06, "loss": 0.0269, "step": 9408 }, { "epoch": 9.55, "learning_rate": 6.156716217003846e-06, "loss": 0.027, "step": 9409 }, { "epoch": 9.55, "learning_rate": 6.154692509766729e-06, "loss": 0.0289, "step": 9410 }, { "epoch": 9.55, "learning_rate": 6.152668987319339e-06, "loss": 0.0377, "step": 9411 }, { "epoch": 9.55, "learning_rate": 6.150645649758916e-06, "loss": 0.0272, "step": 9412 }, { "epoch": 9.55, "learning_rate": 6.148622497182699e-06, "loss": 0.0285, "step": 9413 }, { "epoch": 9.55, "learning_rate": 6.146599529687911e-06, "loss": 0.0278, "step": 9414 }, { "epoch": 9.55, "learning_rate": 6.14457674737176e-06, "loss": 0.0298, "step": 9415 }, { "epoch": 9.55, "learning_rate": 6.1425541503314645e-06, "loss": 0.0316, "step": 9416 }, { "epoch": 9.56, "learning_rate": 6.140531738664216e-06, "loss": 0.0287, "step": 9417 }, { "epoch": 9.56, "learning_rate": 6.138509512467203e-06, "loss": 0.031, "step": 9418 }, { "epoch": 9.56, "learning_rate": 6.136487471837607e-06, "loss": 0.0204, "step": 9419 }, { "epoch": 9.56, "learning_rate": 6.134465616872598e-06, "loss": 0.0254, "step": 9420 }, { "epoch": 9.56, "learning_rate": 6.132443947669338e-06, "loss": 0.0323, "step": 9421 }, { "epoch": 9.56, "learning_rate": 6.13042246432498e-06, "loss": 0.0249, "step": 9422 }, { "epoch": 9.56, "learning_rate": 6.128401166936668e-06, "loss": 0.0236, "step": 9423 }, { "epoch": 9.56, "learning_rate": 6.126380055601537e-06, "loss": 0.0382, "step": 9424 }, { "epoch": 9.56, "learning_rate": 6.124359130416717e-06, "loss": 0.0347, "step": 9425 }, { "epoch": 9.56, "learning_rate": 6.122338391479323e-06, "loss": 0.0278, "step": 9426 }, { "epoch": 9.57, "learning_rate": 6.120317838886463e-06, "loss": 0.0253, "step": 9427 }, { "epoch": 9.57, "learning_rate": 6.118297472735235e-06, "loss": 0.0257, "step": 9428 }, { "epoch": 9.57, "learning_rate": 6.116277293122733e-06, "loss": 0.0307, "step": 9429 }, { "epoch": 9.57, "learning_rate": 6.114257300146037e-06, "loss": 0.0266, "step": 9430 }, { "epoch": 9.57, "learning_rate": 6.112237493902219e-06, "loss": 0.03, "step": 9431 }, { "epoch": 9.57, "learning_rate": 6.110217874488338e-06, "loss": 0.0325, "step": 9432 }, { "epoch": 9.57, "learning_rate": 6.108198442001458e-06, "loss": 0.0362, "step": 9433 }, { "epoch": 9.57, "learning_rate": 6.1061791965386185e-06, "loss": 0.0249, "step": 9434 }, { "epoch": 9.57, "learning_rate": 6.10416013819686e-06, "loss": 0.0282, "step": 9435 }, { "epoch": 9.57, "learning_rate": 6.102141267073207e-06, "loss": 0.0241, "step": 9436 }, { "epoch": 9.58, "learning_rate": 6.1001225832646806e-06, "loss": 0.0269, "step": 9437 }, { "epoch": 9.58, "learning_rate": 6.098104086868288e-06, "loss": 0.0329, "step": 9438 }, { "epoch": 9.58, "learning_rate": 6.096085777981029e-06, "loss": 0.0279, "step": 9439 }, { "epoch": 9.58, "learning_rate": 6.094067656699898e-06, "loss": 0.026, "step": 9440 }, { "epoch": 9.58, "learning_rate": 6.0920497231218765e-06, "loss": 0.0242, "step": 9441 }, { "epoch": 9.58, "learning_rate": 6.09003197734394e-06, "loss": 0.033, "step": 9442 }, { "epoch": 9.58, "learning_rate": 6.088014419463053e-06, "loss": 0.0251, "step": 9443 }, { "epoch": 9.58, "learning_rate": 6.0859970495761675e-06, "loss": 0.0331, "step": 9444 }, { "epoch": 9.58, "learning_rate": 6.083979867780232e-06, "loss": 0.0323, "step": 9445 }, { "epoch": 9.58, "learning_rate": 6.081962874172185e-06, "loss": 0.0281, "step": 9446 }, { "epoch": 9.59, "learning_rate": 6.079946068848954e-06, "loss": 0.0331, "step": 9447 }, { "epoch": 9.59, "learning_rate": 6.077929451907455e-06, "loss": 0.0267, "step": 9448 }, { "epoch": 9.59, "learning_rate": 6.075913023444606e-06, "loss": 0.0314, "step": 9449 }, { "epoch": 9.59, "learning_rate": 6.073896783557305e-06, "loss": 0.0264, "step": 9450 }, { "epoch": 9.59, "learning_rate": 6.071880732342439e-06, "loss": 0.0324, "step": 9451 }, { "epoch": 9.59, "learning_rate": 6.0698648698969e-06, "loss": 0.0252, "step": 9452 }, { "epoch": 9.59, "learning_rate": 6.067849196317556e-06, "loss": 0.0302, "step": 9453 }, { "epoch": 9.59, "learning_rate": 6.065833711701274e-06, "loss": 0.0283, "step": 9454 }, { "epoch": 9.59, "learning_rate": 6.063818416144909e-06, "loss": 0.0334, "step": 9455 }, { "epoch": 9.6, "learning_rate": 6.061803309745309e-06, "loss": 0.0339, "step": 9456 }, { "epoch": 9.6, "learning_rate": 6.0597883925993065e-06, "loss": 0.0265, "step": 9457 }, { "epoch": 9.6, "learning_rate": 6.05777366480374e-06, "loss": 0.0258, "step": 9458 }, { "epoch": 9.6, "learning_rate": 6.055759126455424e-06, "loss": 0.0312, "step": 9459 }, { "epoch": 9.6, "learning_rate": 6.053744777651168e-06, "loss": 0.0258, "step": 9460 }, { "epoch": 9.6, "learning_rate": 6.051730618487776e-06, "loss": 0.03, "step": 9461 }, { "epoch": 9.6, "learning_rate": 6.049716649062034e-06, "loss": 0.0322, "step": 9462 }, { "epoch": 9.6, "learning_rate": 6.047702869470733e-06, "loss": 0.0294, "step": 9463 }, { "epoch": 9.6, "learning_rate": 6.0456892798106446e-06, "loss": 0.03, "step": 9464 }, { "epoch": 9.6, "learning_rate": 6.0436758801785274e-06, "loss": 0.0328, "step": 9465 }, { "epoch": 9.61, "learning_rate": 6.041662670671147e-06, "loss": 0.0318, "step": 9466 }, { "epoch": 9.61, "learning_rate": 6.0396496513852445e-06, "loss": 0.0275, "step": 9467 }, { "epoch": 9.61, "learning_rate": 6.037636822417556e-06, "loss": 0.0273, "step": 9468 }, { "epoch": 9.61, "learning_rate": 6.035624183864817e-06, "loss": 0.0307, "step": 9469 }, { "epoch": 9.61, "learning_rate": 6.033611735823739e-06, "loss": 0.027, "step": 9470 }, { "epoch": 9.61, "learning_rate": 6.0315994783910345e-06, "loss": 0.0318, "step": 9471 }, { "epoch": 9.61, "learning_rate": 6.029587411663405e-06, "loss": 0.0286, "step": 9472 }, { "epoch": 9.61, "learning_rate": 6.02757553573754e-06, "loss": 0.0232, "step": 9473 }, { "epoch": 9.61, "learning_rate": 6.025563850710125e-06, "loss": 0.0236, "step": 9474 }, { "epoch": 9.61, "learning_rate": 6.023552356677833e-06, "loss": 0.0246, "step": 9475 }, { "epoch": 9.62, "learning_rate": 6.021541053737327e-06, "loss": 0.0226, "step": 9476 }, { "epoch": 9.62, "learning_rate": 6.0195299419852634e-06, "loss": 0.0261, "step": 9477 }, { "epoch": 9.62, "learning_rate": 6.017519021518286e-06, "loss": 0.0299, "step": 9478 }, { "epoch": 9.62, "learning_rate": 6.015508292433032e-06, "loss": 0.0263, "step": 9479 }, { "epoch": 9.62, "learning_rate": 6.01349775482613e-06, "loss": 0.0302, "step": 9480 }, { "epoch": 9.62, "learning_rate": 6.011487408794197e-06, "loss": 0.0325, "step": 9481 }, { "epoch": 9.62, "learning_rate": 6.009477254433839e-06, "loss": 0.0306, "step": 9482 }, { "epoch": 9.62, "learning_rate": 6.0074672918416625e-06, "loss": 0.0332, "step": 9483 }, { "epoch": 9.62, "learning_rate": 6.005457521114255e-06, "loss": 0.0243, "step": 9484 }, { "epoch": 9.62, "learning_rate": 6.003447942348195e-06, "loss": 0.0254, "step": 9485 }, { "epoch": 9.63, "learning_rate": 6.00143855564006e-06, "loss": 0.0282, "step": 9486 }, { "epoch": 9.63, "learning_rate": 5.99942936108641e-06, "loss": 0.0257, "step": 9487 }, { "epoch": 9.63, "learning_rate": 5.9974203587837985e-06, "loss": 0.032, "step": 9488 }, { "epoch": 9.63, "learning_rate": 5.995411548828767e-06, "loss": 0.0313, "step": 9489 }, { "epoch": 9.63, "learning_rate": 5.993402931317856e-06, "loss": 0.0246, "step": 9490 }, { "epoch": 9.63, "learning_rate": 5.991394506347588e-06, "loss": 0.0252, "step": 9491 }, { "epoch": 9.63, "learning_rate": 5.9893862740144815e-06, "loss": 0.0255, "step": 9492 }, { "epoch": 9.63, "learning_rate": 5.9873782344150445e-06, "loss": 0.026, "step": 9493 }, { "epoch": 9.63, "learning_rate": 5.985370387645775e-06, "loss": 0.0311, "step": 9494 }, { "epoch": 9.63, "learning_rate": 5.983362733803157e-06, "loss": 0.0318, "step": 9495 }, { "epoch": 9.64, "learning_rate": 5.9813552729836776e-06, "loss": 0.0306, "step": 9496 }, { "epoch": 9.64, "learning_rate": 5.979348005283803e-06, "loss": 0.0278, "step": 9497 }, { "epoch": 9.64, "learning_rate": 5.977340930799995e-06, "loss": 0.0269, "step": 9498 }, { "epoch": 9.64, "learning_rate": 5.9753340496287e-06, "loss": 0.0321, "step": 9499 }, { "epoch": 9.64, "learning_rate": 5.973327361866371e-06, "loss": 0.031, "step": 9500 }, { "epoch": 9.64, "learning_rate": 5.971320867609434e-06, "loss": 0.0267, "step": 9501 }, { "epoch": 9.64, "learning_rate": 5.969314566954316e-06, "loss": 0.0285, "step": 9502 }, { "epoch": 9.64, "learning_rate": 5.9673084599974295e-06, "loss": 0.0351, "step": 9503 }, { "epoch": 9.64, "learning_rate": 5.965302546835183e-06, "loss": 0.0305, "step": 9504 }, { "epoch": 9.64, "learning_rate": 5.963296827563969e-06, "loss": 0.0361, "step": 9505 }, { "epoch": 9.65, "learning_rate": 5.961291302280172e-06, "loss": 0.0315, "step": 9506 }, { "epoch": 9.65, "learning_rate": 5.959285971080172e-06, "loss": 0.0232, "step": 9507 }, { "epoch": 9.65, "learning_rate": 5.957280834060342e-06, "loss": 0.0267, "step": 9508 }, { "epoch": 9.65, "learning_rate": 5.955275891317034e-06, "loss": 0.0268, "step": 9509 }, { "epoch": 9.65, "learning_rate": 5.953271142946602e-06, "loss": 0.0311, "step": 9510 }, { "epoch": 9.65, "learning_rate": 5.951266589045382e-06, "loss": 0.0342, "step": 9511 }, { "epoch": 9.65, "learning_rate": 5.949262229709704e-06, "loss": 0.0242, "step": 9512 }, { "epoch": 9.65, "learning_rate": 5.947258065035896e-06, "loss": 0.0293, "step": 9513 }, { "epoch": 9.65, "learning_rate": 5.945254095120263e-06, "loss": 0.0364, "step": 9514 }, { "epoch": 9.65, "learning_rate": 5.943250320059108e-06, "loss": 0.0302, "step": 9515 }, { "epoch": 9.66, "learning_rate": 5.94124673994873e-06, "loss": 0.0312, "step": 9516 }, { "epoch": 9.66, "learning_rate": 5.939243354885408e-06, "loss": 0.0335, "step": 9517 }, { "epoch": 9.66, "learning_rate": 5.937240164965417e-06, "loss": 0.023, "step": 9518 }, { "epoch": 9.66, "learning_rate": 5.935237170285022e-06, "loss": 0.0303, "step": 9519 }, { "epoch": 9.66, "learning_rate": 5.933234370940482e-06, "loss": 0.0355, "step": 9520 }, { "epoch": 9.66, "learning_rate": 5.931231767028041e-06, "loss": 0.028, "step": 9521 }, { "epoch": 9.66, "learning_rate": 5.929229358643932e-06, "loss": 0.0294, "step": 9522 }, { "epoch": 9.66, "learning_rate": 5.927227145884389e-06, "loss": 0.0257, "step": 9523 }, { "epoch": 9.66, "learning_rate": 5.9252251288456245e-06, "loss": 0.0275, "step": 9524 }, { "epoch": 9.67, "learning_rate": 5.923223307623852e-06, "loss": 0.0297, "step": 9525 }, { "epoch": 9.67, "learning_rate": 5.9212216823152706e-06, "loss": 0.0363, "step": 9526 }, { "epoch": 9.67, "learning_rate": 5.919220253016067e-06, "loss": 0.0261, "step": 9527 }, { "epoch": 9.67, "learning_rate": 5.917219019822424e-06, "loss": 0.03, "step": 9528 }, { "epoch": 9.67, "learning_rate": 5.915217982830512e-06, "loss": 0.0308, "step": 9529 }, { "epoch": 9.67, "learning_rate": 5.913217142136493e-06, "loss": 0.0309, "step": 9530 }, { "epoch": 9.67, "learning_rate": 5.911216497836518e-06, "loss": 0.0271, "step": 9531 }, { "epoch": 9.67, "learning_rate": 5.909216050026727e-06, "loss": 0.0226, "step": 9532 }, { "epoch": 9.67, "learning_rate": 5.907215798803263e-06, "loss": 0.0302, "step": 9533 }, { "epoch": 9.67, "learning_rate": 5.9052157442622416e-06, "loss": 0.0311, "step": 9534 }, { "epoch": 9.68, "learning_rate": 5.903215886499778e-06, "loss": 0.0256, "step": 9535 }, { "epoch": 9.68, "learning_rate": 5.901216225611981e-06, "loss": 0.0301, "step": 9536 }, { "epoch": 9.68, "learning_rate": 5.899216761694943e-06, "loss": 0.0287, "step": 9537 }, { "epoch": 9.68, "learning_rate": 5.89721749484475e-06, "loss": 0.0442, "step": 9538 }, { "epoch": 9.68, "learning_rate": 5.895218425157479e-06, "loss": 0.0346, "step": 9539 }, { "epoch": 9.68, "learning_rate": 5.893219552729197e-06, "loss": 0.0329, "step": 9540 }, { "epoch": 9.68, "learning_rate": 5.891220877655961e-06, "loss": 0.0268, "step": 9541 }, { "epoch": 9.68, "learning_rate": 5.889222400033823e-06, "loss": 0.0236, "step": 9542 }, { "epoch": 9.68, "learning_rate": 5.8872241199588185e-06, "loss": 0.0308, "step": 9543 }, { "epoch": 9.68, "learning_rate": 5.8852260375269785e-06, "loss": 0.0256, "step": 9544 }, { "epoch": 9.69, "learning_rate": 5.883228152834317e-06, "loss": 0.034, "step": 9545 }, { "epoch": 9.69, "learning_rate": 5.881230465976851e-06, "loss": 0.0247, "step": 9546 }, { "epoch": 9.69, "learning_rate": 5.87923297705058e-06, "loss": 0.0341, "step": 9547 }, { "epoch": 9.69, "learning_rate": 5.877235686151492e-06, "loss": 0.0293, "step": 9548 }, { "epoch": 9.69, "learning_rate": 5.875238593375567e-06, "loss": 0.0332, "step": 9549 }, { "epoch": 9.69, "learning_rate": 5.873241698818784e-06, "loss": 0.0259, "step": 9550 }, { "epoch": 9.69, "learning_rate": 5.8712450025771e-06, "loss": 0.0245, "step": 9551 }, { "epoch": 9.69, "learning_rate": 5.86924850474647e-06, "loss": 0.0295, "step": 9552 }, { "epoch": 9.69, "learning_rate": 5.8672522054228396e-06, "loss": 0.027, "step": 9553 }, { "epoch": 9.69, "learning_rate": 5.865256104702139e-06, "loss": 0.0314, "step": 9554 }, { "epoch": 9.7, "learning_rate": 5.863260202680297e-06, "loss": 0.036, "step": 9555 }, { "epoch": 9.7, "learning_rate": 5.861264499453221e-06, "loss": 0.0306, "step": 9556 }, { "epoch": 9.7, "learning_rate": 5.8592689951168205e-06, "loss": 0.0364, "step": 9557 }, { "epoch": 9.7, "learning_rate": 5.8572736897669955e-06, "loss": 0.0314, "step": 9558 }, { "epoch": 9.7, "learning_rate": 5.855278583499629e-06, "loss": 0.0323, "step": 9559 }, { "epoch": 9.7, "learning_rate": 5.853283676410595e-06, "loss": 0.0264, "step": 9560 }, { "epoch": 9.7, "learning_rate": 5.851288968595765e-06, "loss": 0.0284, "step": 9561 }, { "epoch": 9.7, "learning_rate": 5.849294460150992e-06, "loss": 0.025, "step": 9562 }, { "epoch": 9.7, "learning_rate": 5.8473001511721265e-06, "loss": 0.0272, "step": 9563 }, { "epoch": 9.7, "learning_rate": 5.845306041755008e-06, "loss": 0.0305, "step": 9564 }, { "epoch": 9.71, "learning_rate": 5.843312131995457e-06, "loss": 0.0298, "step": 9565 }, { "epoch": 9.71, "learning_rate": 5.841318421989306e-06, "loss": 0.0237, "step": 9566 }, { "epoch": 9.71, "learning_rate": 5.839324911832357e-06, "loss": 0.0245, "step": 9567 }, { "epoch": 9.71, "learning_rate": 5.837331601620411e-06, "loss": 0.0282, "step": 9568 }, { "epoch": 9.71, "learning_rate": 5.835338491449256e-06, "loss": 0.0329, "step": 9569 }, { "epoch": 9.71, "learning_rate": 5.833345581414674e-06, "loss": 0.0317, "step": 9570 }, { "epoch": 9.71, "learning_rate": 5.831352871612441e-06, "loss": 0.0305, "step": 9571 }, { "epoch": 9.71, "learning_rate": 5.8293603621383125e-06, "loss": 0.0302, "step": 9572 }, { "epoch": 9.71, "learning_rate": 5.827368053088043e-06, "loss": 0.0257, "step": 9573 }, { "epoch": 9.71, "learning_rate": 5.825375944557367e-06, "loss": 0.025, "step": 9574 }, { "epoch": 9.72, "learning_rate": 5.823384036642033e-06, "loss": 0.0278, "step": 9575 }, { "epoch": 9.72, "learning_rate": 5.82139232943775e-06, "loss": 0.0248, "step": 9576 }, { "epoch": 9.72, "learning_rate": 5.819400823040237e-06, "loss": 0.0256, "step": 9577 }, { "epoch": 9.72, "learning_rate": 5.8174095175452005e-06, "loss": 0.0343, "step": 9578 }, { "epoch": 9.72, "learning_rate": 5.815418413048326e-06, "loss": 0.027, "step": 9579 }, { "epoch": 9.72, "learning_rate": 5.8134275096453085e-06, "loss": 0.0312, "step": 9580 }, { "epoch": 9.72, "learning_rate": 5.811436807431812e-06, "loss": 0.0318, "step": 9581 }, { "epoch": 9.72, "learning_rate": 5.809446306503501e-06, "loss": 0.0236, "step": 9582 }, { "epoch": 9.72, "learning_rate": 5.807456006956045e-06, "loss": 0.0358, "step": 9583 }, { "epoch": 9.73, "learning_rate": 5.805465908885078e-06, "loss": 0.0308, "step": 9584 }, { "epoch": 9.73, "learning_rate": 5.80347601238624e-06, "loss": 0.0399, "step": 9585 }, { "epoch": 9.73, "learning_rate": 5.8014863175551535e-06, "loss": 0.0273, "step": 9586 }, { "epoch": 9.73, "learning_rate": 5.799496824487438e-06, "loss": 0.0336, "step": 9587 }, { "epoch": 9.73, "learning_rate": 5.7975075332787036e-06, "loss": 0.0249, "step": 9588 }, { "epoch": 9.73, "learning_rate": 5.79551844402454e-06, "loss": 0.0281, "step": 9589 }, { "epoch": 9.73, "learning_rate": 5.793529556820538e-06, "loss": 0.0287, "step": 9590 }, { "epoch": 9.73, "learning_rate": 5.791540871762276e-06, "loss": 0.027, "step": 9591 }, { "epoch": 9.73, "learning_rate": 5.789552388945328e-06, "loss": 0.0305, "step": 9592 }, { "epoch": 9.73, "learning_rate": 5.78756410846524e-06, "loss": 0.0244, "step": 9593 }, { "epoch": 9.74, "learning_rate": 5.7855760304175676e-06, "loss": 0.0289, "step": 9594 }, { "epoch": 9.74, "learning_rate": 5.783588154897853e-06, "loss": 0.0314, "step": 9595 }, { "epoch": 9.74, "learning_rate": 5.781600482001617e-06, "loss": 0.0327, "step": 9596 }, { "epoch": 9.74, "learning_rate": 5.779613011824385e-06, "loss": 0.0239, "step": 9597 }, { "epoch": 9.74, "learning_rate": 5.777625744461668e-06, "loss": 0.0278, "step": 9598 }, { "epoch": 9.74, "learning_rate": 5.775638680008955e-06, "loss": 0.0358, "step": 9599 }, { "epoch": 9.74, "learning_rate": 5.773651818561752e-06, "loss": 0.0244, "step": 9600 }, { "epoch": 9.74, "learning_rate": 5.7716651602155275e-06, "loss": 0.0275, "step": 9601 }, { "epoch": 9.74, "learning_rate": 5.7696787050657596e-06, "loss": 0.0364, "step": 9602 }, { "epoch": 9.74, "learning_rate": 5.767692453207904e-06, "loss": 0.0332, "step": 9603 }, { "epoch": 9.75, "learning_rate": 5.765706404737412e-06, "loss": 0.027, "step": 9604 }, { "epoch": 9.75, "learning_rate": 5.763720559749732e-06, "loss": 0.0341, "step": 9605 }, { "epoch": 9.75, "learning_rate": 5.7617349183402845e-06, "loss": 0.0278, "step": 9606 }, { "epoch": 9.75, "learning_rate": 5.759749480604499e-06, "loss": 0.0268, "step": 9607 }, { "epoch": 9.75, "learning_rate": 5.757764246637784e-06, "loss": 0.0313, "step": 9608 }, { "epoch": 9.75, "learning_rate": 5.755779216535548e-06, "loss": 0.0308, "step": 9609 }, { "epoch": 9.75, "learning_rate": 5.753794390393173e-06, "loss": 0.0336, "step": 9610 }, { "epoch": 9.75, "learning_rate": 5.7518097683060495e-06, "loss": 0.025, "step": 9611 }, { "epoch": 9.75, "learning_rate": 5.749825350369551e-06, "loss": 0.0322, "step": 9612 }, { "epoch": 9.75, "learning_rate": 5.747841136679033e-06, "loss": 0.029, "step": 9613 }, { "epoch": 9.76, "learning_rate": 5.7458571273298545e-06, "loss": 0.0266, "step": 9614 }, { "epoch": 9.76, "learning_rate": 5.743873322417362e-06, "loss": 0.0328, "step": 9615 }, { "epoch": 9.76, "learning_rate": 5.741889722036875e-06, "loss": 0.0308, "step": 9616 }, { "epoch": 9.76, "learning_rate": 5.739906326283735e-06, "loss": 0.0291, "step": 9617 }, { "epoch": 9.76, "learning_rate": 5.7379231352532445e-06, "loss": 0.0304, "step": 9618 }, { "epoch": 9.76, "learning_rate": 5.7359401490407155e-06, "loss": 0.0303, "step": 9619 }, { "epoch": 9.76, "learning_rate": 5.733957367741433e-06, "loss": 0.0333, "step": 9620 }, { "epoch": 9.76, "learning_rate": 5.731974791450685e-06, "loss": 0.0266, "step": 9621 }, { "epoch": 9.76, "learning_rate": 5.729992420263753e-06, "loss": 0.0315, "step": 9622 }, { "epoch": 9.76, "learning_rate": 5.728010254275889e-06, "loss": 0.0273, "step": 9623 }, { "epoch": 9.77, "learning_rate": 5.726028293582355e-06, "loss": 0.0256, "step": 9624 }, { "epoch": 9.77, "learning_rate": 5.724046538278395e-06, "loss": 0.0317, "step": 9625 }, { "epoch": 9.77, "learning_rate": 5.722064988459244e-06, "loss": 0.0303, "step": 9626 }, { "epoch": 9.77, "learning_rate": 5.720083644220131e-06, "loss": 0.0349, "step": 9627 }, { "epoch": 9.77, "learning_rate": 5.7181025056562655e-06, "loss": 0.0287, "step": 9628 }, { "epoch": 9.77, "learning_rate": 5.716121572862858e-06, "loss": 0.0295, "step": 9629 }, { "epoch": 9.77, "learning_rate": 5.714140845935096e-06, "loss": 0.0308, "step": 9630 }, { "epoch": 9.77, "learning_rate": 5.712160324968172e-06, "loss": 0.0267, "step": 9631 }, { "epoch": 9.77, "learning_rate": 5.71018001005726e-06, "loss": 0.0323, "step": 9632 }, { "epoch": 9.77, "learning_rate": 5.708199901297524e-06, "loss": 0.0291, "step": 9633 }, { "epoch": 9.78, "learning_rate": 5.706219998784127e-06, "loss": 0.0273, "step": 9634 }, { "epoch": 9.78, "learning_rate": 5.704240302612205e-06, "loss": 0.0327, "step": 9635 }, { "epoch": 9.78, "learning_rate": 5.702260812876903e-06, "loss": 0.0286, "step": 9636 }, { "epoch": 9.78, "learning_rate": 5.700281529673339e-06, "loss": 0.0258, "step": 9637 }, { "epoch": 9.78, "learning_rate": 5.698302453096633e-06, "loss": 0.0278, "step": 9638 }, { "epoch": 9.78, "learning_rate": 5.696323583241896e-06, "loss": 0.0382, "step": 9639 }, { "epoch": 9.78, "learning_rate": 5.6943449202042156e-06, "loss": 0.0345, "step": 9640 }, { "epoch": 9.78, "learning_rate": 5.692366464078683e-06, "loss": 0.038, "step": 9641 }, { "epoch": 9.78, "learning_rate": 5.690388214960373e-06, "loss": 0.029, "step": 9642 }, { "epoch": 9.78, "learning_rate": 5.688410172944354e-06, "loss": 0.0272, "step": 9643 }, { "epoch": 9.79, "learning_rate": 5.686432338125686e-06, "loss": 0.0345, "step": 9644 }, { "epoch": 9.79, "learning_rate": 5.684454710599408e-06, "loss": 0.0329, "step": 9645 }, { "epoch": 9.79, "learning_rate": 5.682477290460563e-06, "loss": 0.0336, "step": 9646 }, { "epoch": 9.79, "learning_rate": 5.680500077804172e-06, "loss": 0.0318, "step": 9647 }, { "epoch": 9.79, "learning_rate": 5.678523072725254e-06, "loss": 0.0274, "step": 9648 }, { "epoch": 9.79, "learning_rate": 5.676546275318817e-06, "loss": 0.0384, "step": 9649 }, { "epoch": 9.79, "learning_rate": 5.674569685679857e-06, "loss": 0.0348, "step": 9650 }, { "epoch": 9.79, "learning_rate": 5.672593303903365e-06, "loss": 0.0312, "step": 9651 }, { "epoch": 9.79, "learning_rate": 5.670617130084311e-06, "loss": 0.0289, "step": 9652 }, { "epoch": 9.8, "learning_rate": 5.668641164317668e-06, "loss": 0.0301, "step": 9653 }, { "epoch": 9.8, "learning_rate": 5.666665406698385e-06, "loss": 0.0323, "step": 9654 }, { "epoch": 9.8, "learning_rate": 5.6646898573214135e-06, "loss": 0.0272, "step": 9655 }, { "epoch": 9.8, "learning_rate": 5.662714516281696e-06, "loss": 0.0272, "step": 9656 }, { "epoch": 9.8, "learning_rate": 5.6607393836741435e-06, "loss": 0.0314, "step": 9657 }, { "epoch": 9.8, "learning_rate": 5.658764459593691e-06, "loss": 0.0277, "step": 9658 }, { "epoch": 9.8, "learning_rate": 5.656789744135233e-06, "loss": 0.0337, "step": 9659 }, { "epoch": 9.8, "learning_rate": 5.654815237393669e-06, "loss": 0.0367, "step": 9660 }, { "epoch": 9.8, "learning_rate": 5.652840939463893e-06, "loss": 0.0285, "step": 9661 }, { "epoch": 9.8, "learning_rate": 5.650866850440772e-06, "loss": 0.0304, "step": 9662 }, { "epoch": 9.81, "learning_rate": 5.648892970419179e-06, "loss": 0.0318, "step": 9663 }, { "epoch": 9.81, "learning_rate": 5.646919299493963e-06, "loss": 0.0284, "step": 9664 }, { "epoch": 9.81, "learning_rate": 5.644945837759977e-06, "loss": 0.0266, "step": 9665 }, { "epoch": 9.81, "learning_rate": 5.642972585312055e-06, "loss": 0.0326, "step": 9666 }, { "epoch": 9.81, "learning_rate": 5.640999542245025e-06, "loss": 0.0294, "step": 9667 }, { "epoch": 9.81, "learning_rate": 5.639026708653707e-06, "loss": 0.0352, "step": 9668 }, { "epoch": 9.81, "learning_rate": 5.6370540846328984e-06, "loss": 0.0314, "step": 9669 }, { "epoch": 9.81, "learning_rate": 5.635081670277399e-06, "loss": 0.0305, "step": 9670 }, { "epoch": 9.81, "learning_rate": 5.633109465682001e-06, "loss": 0.0271, "step": 9671 }, { "epoch": 9.81, "learning_rate": 5.631137470941471e-06, "loss": 0.0309, "step": 9672 }, { "epoch": 9.82, "learning_rate": 5.6291656861505835e-06, "loss": 0.0339, "step": 9673 }, { "epoch": 9.82, "learning_rate": 5.627194111404083e-06, "loss": 0.0276, "step": 9674 }, { "epoch": 9.82, "learning_rate": 5.62522274679673e-06, "loss": 0.0286, "step": 9675 }, { "epoch": 9.82, "learning_rate": 5.6232515924232486e-06, "loss": 0.0252, "step": 9676 }, { "epoch": 9.82, "learning_rate": 5.621280648378369e-06, "loss": 0.0272, "step": 9677 }, { "epoch": 9.82, "learning_rate": 5.619309914756811e-06, "loss": 0.0308, "step": 9678 }, { "epoch": 9.82, "learning_rate": 5.617339391653271e-06, "loss": 0.0322, "step": 9679 }, { "epoch": 9.82, "learning_rate": 5.615369079162451e-06, "loss": 0.0304, "step": 9680 }, { "epoch": 9.82, "learning_rate": 5.613398977379032e-06, "loss": 0.0248, "step": 9681 }, { "epoch": 9.82, "learning_rate": 5.61142908639769e-06, "loss": 0.0276, "step": 9682 }, { "epoch": 9.83, "learning_rate": 5.609459406313089e-06, "loss": 0.0253, "step": 9683 }, { "epoch": 9.83, "learning_rate": 5.6074899372198875e-06, "loss": 0.0246, "step": 9684 }, { "epoch": 9.83, "learning_rate": 5.605520679212732e-06, "loss": 0.0264, "step": 9685 }, { "epoch": 9.83, "learning_rate": 5.603551632386249e-06, "loss": 0.0329, "step": 9686 }, { "epoch": 9.83, "learning_rate": 5.601582796835066e-06, "loss": 0.0293, "step": 9687 }, { "epoch": 9.83, "learning_rate": 5.599614172653803e-06, "loss": 0.0288, "step": 9688 }, { "epoch": 9.83, "learning_rate": 5.597645759937052e-06, "loss": 0.0347, "step": 9689 }, { "epoch": 9.83, "learning_rate": 5.595677558779421e-06, "loss": 0.027, "step": 9690 }, { "epoch": 9.83, "learning_rate": 5.593709569275479e-06, "loss": 0.0394, "step": 9691 }, { "epoch": 9.83, "learning_rate": 5.591741791519812e-06, "loss": 0.0298, "step": 9692 }, { "epoch": 9.84, "learning_rate": 5.5897742256069764e-06, "loss": 0.0329, "step": 9693 }, { "epoch": 9.84, "learning_rate": 5.587806871631527e-06, "loss": 0.0286, "step": 9694 }, { "epoch": 9.84, "learning_rate": 5.585839729688012e-06, "loss": 0.0274, "step": 9695 }, { "epoch": 9.84, "learning_rate": 5.583872799870954e-06, "loss": 0.0282, "step": 9696 }, { "epoch": 9.84, "learning_rate": 5.581906082274881e-06, "loss": 0.0273, "step": 9697 }, { "epoch": 9.84, "learning_rate": 5.579939576994311e-06, "loss": 0.0292, "step": 9698 }, { "epoch": 9.84, "learning_rate": 5.577973284123731e-06, "loss": 0.0335, "step": 9699 }, { "epoch": 9.84, "learning_rate": 5.576007203757651e-06, "loss": 0.0241, "step": 9700 }, { "epoch": 9.84, "learning_rate": 5.57404133599054e-06, "loss": 0.0316, "step": 9701 }, { "epoch": 9.84, "learning_rate": 5.572075680916878e-06, "loss": 0.0289, "step": 9702 }, { "epoch": 9.85, "learning_rate": 5.570110238631117e-06, "loss": 0.0324, "step": 9703 }, { "epoch": 9.85, "learning_rate": 5.568145009227715e-06, "loss": 0.0262, "step": 9704 }, { "epoch": 9.85, "learning_rate": 5.566179992801114e-06, "loss": 0.0296, "step": 9705 }, { "epoch": 9.85, "learning_rate": 5.564215189445738e-06, "loss": 0.0268, "step": 9706 }, { "epoch": 9.85, "learning_rate": 5.562250599256012e-06, "loss": 0.0374, "step": 9707 }, { "epoch": 9.85, "learning_rate": 5.560286222326344e-06, "loss": 0.0329, "step": 9708 }, { "epoch": 9.85, "learning_rate": 5.55832205875114e-06, "loss": 0.0246, "step": 9709 }, { "epoch": 9.85, "learning_rate": 5.556358108624781e-06, "loss": 0.0284, "step": 9710 }, { "epoch": 9.85, "learning_rate": 5.5543943720416495e-06, "loss": 0.0296, "step": 9711 }, { "epoch": 9.85, "learning_rate": 5.55243084909612e-06, "loss": 0.0288, "step": 9712 }, { "epoch": 9.86, "learning_rate": 5.550467539882543e-06, "loss": 0.0282, "step": 9713 }, { "epoch": 9.86, "learning_rate": 5.548504444495271e-06, "loss": 0.0296, "step": 9714 }, { "epoch": 9.86, "learning_rate": 5.546541563028644e-06, "loss": 0.0292, "step": 9715 }, { "epoch": 9.86, "learning_rate": 5.544578895576983e-06, "loss": 0.0292, "step": 9716 }, { "epoch": 9.86, "learning_rate": 5.542616442234618e-06, "loss": 0.0304, "step": 9717 }, { "epoch": 9.86, "learning_rate": 5.540654203095845e-06, "loss": 0.029, "step": 9718 }, { "epoch": 9.86, "learning_rate": 5.538692178254971e-06, "loss": 0.0284, "step": 9719 }, { "epoch": 9.86, "learning_rate": 5.536730367806273e-06, "loss": 0.0362, "step": 9720 }, { "epoch": 9.86, "learning_rate": 5.534768771844032e-06, "loss": 0.0261, "step": 9721 }, { "epoch": 9.87, "learning_rate": 5.532807390462518e-06, "loss": 0.0358, "step": 9722 }, { "epoch": 9.87, "learning_rate": 5.530846223755979e-06, "loss": 0.0334, "step": 9723 }, { "epoch": 9.87, "learning_rate": 5.528885271818666e-06, "loss": 0.0282, "step": 9724 }, { "epoch": 9.87, "learning_rate": 5.526924534744813e-06, "loss": 0.0264, "step": 9725 }, { "epoch": 9.87, "learning_rate": 5.524964012628648e-06, "loss": 0.0232, "step": 9726 }, { "epoch": 9.87, "learning_rate": 5.52300370556438e-06, "loss": 0.0287, "step": 9727 }, { "epoch": 9.87, "learning_rate": 5.521043613646217e-06, "loss": 0.0304, "step": 9728 }, { "epoch": 9.87, "learning_rate": 5.519083736968356e-06, "loss": 0.0322, "step": 9729 }, { "epoch": 9.87, "learning_rate": 5.5171240756249725e-06, "loss": 0.0359, "step": 9730 }, { "epoch": 9.87, "learning_rate": 5.5151646297102435e-06, "loss": 0.0333, "step": 9731 }, { "epoch": 9.88, "learning_rate": 5.513205399318333e-06, "loss": 0.0255, "step": 9732 }, { "epoch": 9.88, "learning_rate": 5.511246384543394e-06, "loss": 0.0216, "step": 9733 }, { "epoch": 9.88, "learning_rate": 5.509287585479572e-06, "loss": 0.0358, "step": 9734 }, { "epoch": 9.88, "learning_rate": 5.507329002220991e-06, "loss": 0.0264, "step": 9735 }, { "epoch": 9.88, "learning_rate": 5.505370634861779e-06, "loss": 0.0293, "step": 9736 }, { "epoch": 9.88, "learning_rate": 5.5034124834960405e-06, "loss": 0.0315, "step": 9737 }, { "epoch": 9.88, "learning_rate": 5.501454548217882e-06, "loss": 0.0226, "step": 9738 }, { "epoch": 9.88, "learning_rate": 5.499496829121393e-06, "loss": 0.0318, "step": 9739 }, { "epoch": 9.88, "learning_rate": 5.497539326300652e-06, "loss": 0.027, "step": 9740 }, { "epoch": 9.88, "learning_rate": 5.4955820398497275e-06, "loss": 0.0308, "step": 9741 }, { "epoch": 9.89, "learning_rate": 5.493624969862679e-06, "loss": 0.0316, "step": 9742 }, { "epoch": 9.89, "learning_rate": 5.4916681164335575e-06, "loss": 0.0218, "step": 9743 }, { "epoch": 9.89, "learning_rate": 5.489711479656405e-06, "loss": 0.0293, "step": 9744 }, { "epoch": 9.89, "learning_rate": 5.487755059625239e-06, "loss": 0.0276, "step": 9745 }, { "epoch": 9.89, "learning_rate": 5.4857988564340884e-06, "loss": 0.0308, "step": 9746 }, { "epoch": 9.89, "learning_rate": 5.48384287017695e-06, "loss": 0.027, "step": 9747 }, { "epoch": 9.89, "learning_rate": 5.481887100947825e-06, "loss": 0.0327, "step": 9748 }, { "epoch": 9.89, "learning_rate": 5.4799315488407e-06, "loss": 0.0328, "step": 9749 }, { "epoch": 9.89, "learning_rate": 5.477976213949552e-06, "loss": 0.0241, "step": 9750 }, { "epoch": 9.89, "learning_rate": 5.476021096368348e-06, "loss": 0.0263, "step": 9751 }, { "epoch": 9.9, "learning_rate": 5.474066196191037e-06, "loss": 0.0305, "step": 9752 }, { "epoch": 9.9, "learning_rate": 5.472111513511569e-06, "loss": 0.0312, "step": 9753 }, { "epoch": 9.9, "learning_rate": 5.470157048423872e-06, "loss": 0.0329, "step": 9754 }, { "epoch": 9.9, "learning_rate": 5.468202801021874e-06, "loss": 0.0261, "step": 9755 }, { "epoch": 9.9, "learning_rate": 5.466248771399491e-06, "loss": 0.0308, "step": 9756 }, { "epoch": 9.9, "learning_rate": 5.4642949596506166e-06, "loss": 0.0268, "step": 9757 }, { "epoch": 9.9, "learning_rate": 5.46234136586915e-06, "loss": 0.0301, "step": 9758 }, { "epoch": 9.9, "learning_rate": 5.4603879901489695e-06, "loss": 0.0309, "step": 9759 }, { "epoch": 9.9, "learning_rate": 5.45843483258395e-06, "loss": 0.0269, "step": 9760 }, { "epoch": 9.9, "learning_rate": 5.456481893267953e-06, "loss": 0.0286, "step": 9761 }, { "epoch": 9.91, "learning_rate": 5.454529172294824e-06, "loss": 0.0322, "step": 9762 }, { "epoch": 9.91, "learning_rate": 5.452576669758407e-06, "loss": 0.0264, "step": 9763 }, { "epoch": 9.91, "learning_rate": 5.450624385752525e-06, "loss": 0.0345, "step": 9764 }, { "epoch": 9.91, "learning_rate": 5.4486723203710025e-06, "loss": 0.0288, "step": 9765 }, { "epoch": 9.91, "learning_rate": 5.446720473707645e-06, "loss": 0.0211, "step": 9766 }, { "epoch": 9.91, "learning_rate": 5.444768845856252e-06, "loss": 0.0276, "step": 9767 }, { "epoch": 9.91, "learning_rate": 5.442817436910614e-06, "loss": 0.0273, "step": 9768 }, { "epoch": 9.91, "learning_rate": 5.440866246964499e-06, "loss": 0.0198, "step": 9769 }, { "epoch": 9.91, "learning_rate": 5.438915276111679e-06, "loss": 0.0252, "step": 9770 }, { "epoch": 9.91, "learning_rate": 5.436964524445913e-06, "loss": 0.0302, "step": 9771 }, { "epoch": 9.92, "learning_rate": 5.435013992060938e-06, "loss": 0.0317, "step": 9772 }, { "epoch": 9.92, "learning_rate": 5.433063679050496e-06, "loss": 0.0281, "step": 9773 }, { "epoch": 9.92, "learning_rate": 5.4311135855082986e-06, "loss": 0.0298, "step": 9774 }, { "epoch": 9.92, "learning_rate": 5.429163711528078e-06, "loss": 0.026, "step": 9775 }, { "epoch": 9.92, "learning_rate": 5.4272140572035225e-06, "loss": 0.0375, "step": 9776 }, { "epoch": 9.92, "learning_rate": 5.42526462262833e-06, "loss": 0.0349, "step": 9777 }, { "epoch": 9.92, "learning_rate": 5.423315407896186e-06, "loss": 0.0306, "step": 9778 }, { "epoch": 9.92, "learning_rate": 5.421366413100752e-06, "loss": 0.0268, "step": 9779 }, { "epoch": 9.92, "learning_rate": 5.419417638335699e-06, "loss": 0.0276, "step": 9780 }, { "epoch": 9.92, "learning_rate": 5.417469083694669e-06, "loss": 0.0265, "step": 9781 }, { "epoch": 9.93, "learning_rate": 5.415520749271304e-06, "loss": 0.0299, "step": 9782 }, { "epoch": 9.93, "learning_rate": 5.413572635159235e-06, "loss": 0.029, "step": 9783 }, { "epoch": 9.93, "learning_rate": 5.411624741452078e-06, "loss": 0.0321, "step": 9784 }, { "epoch": 9.93, "learning_rate": 5.409677068243446e-06, "loss": 0.0337, "step": 9785 }, { "epoch": 9.93, "learning_rate": 5.40772961562693e-06, "loss": 0.0343, "step": 9786 }, { "epoch": 9.93, "learning_rate": 5.405782383696119e-06, "loss": 0.0273, "step": 9787 }, { "epoch": 9.93, "learning_rate": 5.40383537254459e-06, "loss": 0.0306, "step": 9788 }, { "epoch": 9.93, "learning_rate": 5.401888582265906e-06, "loss": 0.033, "step": 9789 }, { "epoch": 9.93, "learning_rate": 5.399942012953626e-06, "loss": 0.0246, "step": 9790 }, { "epoch": 9.94, "learning_rate": 5.397995664701283e-06, "loss": 0.0293, "step": 9791 }, { "epoch": 9.94, "learning_rate": 5.396049537602427e-06, "loss": 0.0292, "step": 9792 }, { "epoch": 9.94, "learning_rate": 5.394103631750569e-06, "loss": 0.0328, "step": 9793 }, { "epoch": 9.94, "learning_rate": 5.392157947239224e-06, "loss": 0.0322, "step": 9794 }, { "epoch": 9.94, "learning_rate": 5.390212484161899e-06, "loss": 0.0309, "step": 9795 }, { "epoch": 9.94, "learning_rate": 5.388267242612076e-06, "loss": 0.0217, "step": 9796 }, { "epoch": 9.94, "learning_rate": 5.386322222683245e-06, "loss": 0.0307, "step": 9797 }, { "epoch": 9.94, "learning_rate": 5.384377424468865e-06, "loss": 0.0258, "step": 9798 }, { "epoch": 9.94, "learning_rate": 5.382432848062398e-06, "loss": 0.0269, "step": 9799 }, { "epoch": 9.94, "learning_rate": 5.380488493557301e-06, "loss": 0.029, "step": 9800 }, { "epoch": 9.95, "learning_rate": 5.378544361047002e-06, "loss": 0.0336, "step": 9801 }, { "epoch": 9.95, "learning_rate": 5.376600450624937e-06, "loss": 0.026, "step": 9802 }, { "epoch": 9.95, "learning_rate": 5.374656762384512e-06, "loss": 0.0349, "step": 9803 }, { "epoch": 9.95, "learning_rate": 5.372713296419138e-06, "loss": 0.0282, "step": 9804 }, { "epoch": 9.95, "learning_rate": 5.370770052822214e-06, "loss": 0.0264, "step": 9805 }, { "epoch": 9.95, "learning_rate": 5.368827031687116e-06, "loss": 0.0308, "step": 9806 }, { "epoch": 9.95, "learning_rate": 5.366884233107226e-06, "loss": 0.028, "step": 9807 }, { "epoch": 9.95, "learning_rate": 5.364941657175895e-06, "loss": 0.0265, "step": 9808 }, { "epoch": 9.95, "learning_rate": 5.362999303986489e-06, "loss": 0.027, "step": 9809 }, { "epoch": 9.95, "learning_rate": 5.361057173632341e-06, "loss": 0.0323, "step": 9810 }, { "epoch": 9.96, "learning_rate": 5.359115266206785e-06, "loss": 0.0313, "step": 9811 }, { "epoch": 9.96, "learning_rate": 5.357173581803142e-06, "loss": 0.0303, "step": 9812 }, { "epoch": 9.96, "learning_rate": 5.355232120514717e-06, "loss": 0.0294, "step": 9813 }, { "epoch": 9.96, "learning_rate": 5.3532908824348115e-06, "loss": 0.0292, "step": 9814 }, { "epoch": 9.96, "learning_rate": 5.351349867656719e-06, "loss": 0.0262, "step": 9815 }, { "epoch": 9.96, "learning_rate": 5.349409076273702e-06, "loss": 0.0283, "step": 9816 }, { "epoch": 9.96, "learning_rate": 5.347468508379042e-06, "loss": 0.0386, "step": 9817 }, { "epoch": 9.96, "learning_rate": 5.345528164065987e-06, "loss": 0.0289, "step": 9818 }, { "epoch": 9.96, "learning_rate": 5.343588043427788e-06, "loss": 0.028, "step": 9819 }, { "epoch": 9.96, "learning_rate": 5.341648146557669e-06, "loss": 0.0265, "step": 9820 }, { "epoch": 9.97, "learning_rate": 5.3397084735488625e-06, "loss": 0.0261, "step": 9821 }, { "epoch": 9.97, "learning_rate": 5.337769024494581e-06, "loss": 0.0281, "step": 9822 }, { "epoch": 9.97, "learning_rate": 5.335829799488019e-06, "loss": 0.0324, "step": 9823 }, { "epoch": 9.97, "learning_rate": 5.333890798622374e-06, "loss": 0.0383, "step": 9824 }, { "epoch": 9.97, "learning_rate": 5.331952021990825e-06, "loss": 0.0269, "step": 9825 }, { "epoch": 9.97, "learning_rate": 5.330013469686545e-06, "loss": 0.0338, "step": 9826 }, { "epoch": 9.97, "learning_rate": 5.328075141802686e-06, "loss": 0.0244, "step": 9827 }, { "epoch": 9.97, "learning_rate": 5.326137038432399e-06, "loss": 0.0285, "step": 9828 }, { "epoch": 9.97, "learning_rate": 5.3241991596688255e-06, "loss": 0.0321, "step": 9829 }, { "epoch": 9.97, "learning_rate": 5.322261505605085e-06, "loss": 0.0319, "step": 9830 }, { "epoch": 9.98, "learning_rate": 5.320324076334297e-06, "loss": 0.0363, "step": 9831 }, { "epoch": 9.98, "learning_rate": 5.31838687194957e-06, "loss": 0.0336, "step": 9832 }, { "epoch": 9.98, "learning_rate": 5.3164498925439855e-06, "loss": 0.0305, "step": 9833 }, { "epoch": 9.98, "learning_rate": 5.314513138210644e-06, "loss": 0.0346, "step": 9834 }, { "epoch": 9.98, "learning_rate": 5.312576609042606e-06, "loss": 0.031, "step": 9835 }, { "epoch": 9.98, "learning_rate": 5.310640305132939e-06, "loss": 0.03, "step": 9836 }, { "epoch": 9.98, "learning_rate": 5.30870422657469e-06, "loss": 0.0263, "step": 9837 }, { "epoch": 9.98, "learning_rate": 5.306768373460899e-06, "loss": 0.0274, "step": 9838 }, { "epoch": 9.98, "learning_rate": 5.304832745884601e-06, "loss": 0.027, "step": 9839 }, { "epoch": 9.98, "learning_rate": 5.302897343938805e-06, "loss": 0.0265, "step": 9840 }, { "epoch": 9.99, "learning_rate": 5.300962167716523e-06, "loss": 0.0331, "step": 9841 }, { "epoch": 9.99, "learning_rate": 5.299027217310755e-06, "loss": 0.0337, "step": 9842 }, { "epoch": 9.99, "learning_rate": 5.297092492814481e-06, "loss": 0.0264, "step": 9843 }, { "epoch": 9.99, "learning_rate": 5.2951579943206846e-06, "loss": 0.0321, "step": 9844 }, { "epoch": 9.99, "learning_rate": 5.293223721922319e-06, "loss": 0.0327, "step": 9845 }, { "epoch": 9.99, "learning_rate": 5.2912896757123475e-06, "loss": 0.0331, "step": 9846 }, { "epoch": 9.99, "learning_rate": 5.289355855783702e-06, "loss": 0.027, "step": 9847 }, { "epoch": 9.99, "learning_rate": 5.287422262229321e-06, "loss": 0.0255, "step": 9848 }, { "epoch": 9.99, "learning_rate": 5.285488895142123e-06, "loss": 0.0305, "step": 9849 }, { "epoch": 9.99, "learning_rate": 5.283555754615017e-06, "loss": 0.03, "step": 9850 }, { "epoch": 10.0, "learning_rate": 5.28162284074091e-06, "loss": 0.0273, "step": 9851 }, { "epoch": 10.0, "learning_rate": 5.279690153612676e-06, "loss": 0.0301, "step": 9852 }, { "epoch": 10.0, "learning_rate": 5.277757693323204e-06, "loss": 0.0325, "step": 9853 }, { "epoch": 10.0, "learning_rate": 5.275825459965351e-06, "loss": 0.032, "step": 9854 }, { "epoch": 10.0, "learning_rate": 5.2738934536319775e-06, "loss": 0.0343, "step": 9855 }, { "epoch": 10.0, "learning_rate": 5.271961674415929e-06, "loss": 0.0264, "step": 9856 }, { "epoch": 10.0, "learning_rate": 5.270030122410033e-06, "loss": 0.0229, "step": 9857 }, { "epoch": 10.0, "learning_rate": 5.268098797707115e-06, "loss": 0.026, "step": 9858 }, { "epoch": 10.0, "learning_rate": 5.266167700399989e-06, "loss": 0.0227, "step": 9859 }, { "epoch": 10.01, "learning_rate": 5.264236830581452e-06, "loss": 0.0208, "step": 9860 }, { "epoch": 10.01, "learning_rate": 5.262306188344299e-06, "loss": 0.0197, "step": 9861 }, { "epoch": 10.01, "learning_rate": 5.260375773781302e-06, "loss": 0.0195, "step": 9862 }, { "epoch": 10.01, "learning_rate": 5.258445586985236e-06, "loss": 0.0223, "step": 9863 }, { "epoch": 10.01, "learning_rate": 5.256515628048851e-06, "loss": 0.0176, "step": 9864 }, { "epoch": 10.01, "learning_rate": 5.254585897064893e-06, "loss": 0.022, "step": 9865 }, { "epoch": 10.01, "learning_rate": 5.252656394126101e-06, "loss": 0.0217, "step": 9866 }, { "epoch": 10.01, "learning_rate": 5.250727119325198e-06, "loss": 0.0248, "step": 9867 }, { "epoch": 10.01, "learning_rate": 5.248798072754901e-06, "loss": 0.0269, "step": 9868 }, { "epoch": 10.01, "learning_rate": 5.246869254507904e-06, "loss": 0.0228, "step": 9869 }, { "epoch": 10.02, "learning_rate": 5.244940664676905e-06, "loss": 0.0223, "step": 9870 }, { "epoch": 10.02, "learning_rate": 5.243012303354578e-06, "loss": 0.0283, "step": 9871 }, { "epoch": 10.02, "learning_rate": 5.241084170633595e-06, "loss": 0.0244, "step": 9872 }, { "epoch": 10.02, "learning_rate": 5.239156266606617e-06, "loss": 0.0213, "step": 9873 }, { "epoch": 10.02, "learning_rate": 5.2372285913662855e-06, "loss": 0.0187, "step": 9874 }, { "epoch": 10.02, "learning_rate": 5.2353011450052405e-06, "loss": 0.0176, "step": 9875 }, { "epoch": 10.02, "learning_rate": 5.233373927616105e-06, "loss": 0.0253, "step": 9876 }, { "epoch": 10.02, "learning_rate": 5.231446939291496e-06, "loss": 0.0204, "step": 9877 }, { "epoch": 10.02, "learning_rate": 5.229520180124019e-06, "loss": 0.0237, "step": 9878 }, { "epoch": 10.02, "learning_rate": 5.227593650206258e-06, "loss": 0.0143, "step": 9879 }, { "epoch": 10.03, "learning_rate": 5.225667349630803e-06, "loss": 0.0185, "step": 9880 }, { "epoch": 10.03, "learning_rate": 5.223741278490215e-06, "loss": 0.0174, "step": 9881 }, { "epoch": 10.03, "learning_rate": 5.221815436877058e-06, "loss": 0.023, "step": 9882 }, { "epoch": 10.03, "learning_rate": 5.219889824883879e-06, "loss": 0.0242, "step": 9883 }, { "epoch": 10.03, "learning_rate": 5.217964442603218e-06, "loss": 0.023, "step": 9884 }, { "epoch": 10.03, "learning_rate": 5.216039290127601e-06, "loss": 0.02, "step": 9885 }, { "epoch": 10.03, "learning_rate": 5.214114367549538e-06, "loss": 0.0205, "step": 9886 }, { "epoch": 10.03, "learning_rate": 5.212189674961534e-06, "loss": 0.0176, "step": 9887 }, { "epoch": 10.03, "learning_rate": 5.2102652124560895e-06, "loss": 0.0222, "step": 9888 }, { "epoch": 10.03, "learning_rate": 5.208340980125675e-06, "loss": 0.0216, "step": 9889 }, { "epoch": 10.04, "learning_rate": 5.20641697806277e-06, "loss": 0.0251, "step": 9890 }, { "epoch": 10.04, "learning_rate": 5.204493206359824e-06, "loss": 0.022, "step": 9891 }, { "epoch": 10.04, "learning_rate": 5.2025696651093e-06, "loss": 0.0197, "step": 9892 }, { "epoch": 10.04, "learning_rate": 5.200646354403622e-06, "loss": 0.0215, "step": 9893 }, { "epoch": 10.04, "learning_rate": 5.1987232743352245e-06, "loss": 0.0199, "step": 9894 }, { "epoch": 10.04, "learning_rate": 5.196800424996524e-06, "loss": 0.0226, "step": 9895 }, { "epoch": 10.04, "learning_rate": 5.194877806479916e-06, "loss": 0.0227, "step": 9896 }, { "epoch": 10.04, "learning_rate": 5.192955418877804e-06, "loss": 0.0251, "step": 9897 }, { "epoch": 10.04, "learning_rate": 5.191033262282561e-06, "loss": 0.0213, "step": 9898 }, { "epoch": 10.04, "learning_rate": 5.189111336786563e-06, "loss": 0.0254, "step": 9899 }, { "epoch": 10.05, "learning_rate": 5.187189642482168e-06, "loss": 0.03, "step": 9900 }, { "epoch": 10.05, "learning_rate": 5.185268179461725e-06, "loss": 0.0228, "step": 9901 }, { "epoch": 10.05, "learning_rate": 5.183346947817578e-06, "loss": 0.0179, "step": 9902 }, { "epoch": 10.05, "learning_rate": 5.181425947642043e-06, "loss": 0.0193, "step": 9903 }, { "epoch": 10.05, "learning_rate": 5.17950517902744e-06, "loss": 0.0213, "step": 9904 }, { "epoch": 10.05, "learning_rate": 5.1775846420660755e-06, "loss": 0.021, "step": 9905 }, { "epoch": 10.05, "learning_rate": 5.1756643368502395e-06, "loss": 0.0214, "step": 9906 }, { "epoch": 10.05, "learning_rate": 5.173744263472218e-06, "loss": 0.0213, "step": 9907 }, { "epoch": 10.05, "learning_rate": 5.171824422024272e-06, "loss": 0.0227, "step": 9908 }, { "epoch": 10.05, "learning_rate": 5.1699048125986765e-06, "loss": 0.0263, "step": 9909 }, { "epoch": 10.06, "learning_rate": 5.167985435287666e-06, "loss": 0.0235, "step": 9910 }, { "epoch": 10.06, "learning_rate": 5.166066290183485e-06, "loss": 0.0215, "step": 9911 }, { "epoch": 10.06, "learning_rate": 5.164147377378362e-06, "loss": 0.0192, "step": 9912 }, { "epoch": 10.06, "learning_rate": 5.162228696964504e-06, "loss": 0.0196, "step": 9913 }, { "epoch": 10.06, "learning_rate": 5.160310249034119e-06, "loss": 0.0228, "step": 9914 }, { "epoch": 10.06, "learning_rate": 5.158392033679406e-06, "loss": 0.0204, "step": 9915 }, { "epoch": 10.06, "learning_rate": 5.156474050992532e-06, "loss": 0.0212, "step": 9916 }, { "epoch": 10.06, "learning_rate": 5.154556301065684e-06, "loss": 0.0229, "step": 9917 }, { "epoch": 10.06, "learning_rate": 5.15263878399101e-06, "loss": 0.0198, "step": 9918 }, { "epoch": 10.06, "learning_rate": 5.150721499860665e-06, "loss": 0.02, "step": 9919 }, { "epoch": 10.07, "learning_rate": 5.148804448766778e-06, "loss": 0.022, "step": 9920 }, { "epoch": 10.07, "learning_rate": 5.146887630801478e-06, "loss": 0.0216, "step": 9921 }, { "epoch": 10.07, "learning_rate": 5.144971046056886e-06, "loss": 0.0236, "step": 9922 }, { "epoch": 10.07, "learning_rate": 5.1430546946250935e-06, "loss": 0.0216, "step": 9923 }, { "epoch": 10.07, "learning_rate": 5.141138576598205e-06, "loss": 0.0205, "step": 9924 }, { "epoch": 10.07, "learning_rate": 5.139222692068286e-06, "loss": 0.0214, "step": 9925 }, { "epoch": 10.07, "learning_rate": 5.137307041127422e-06, "loss": 0.0228, "step": 9926 }, { "epoch": 10.07, "learning_rate": 5.1353916238676605e-06, "loss": 0.0197, "step": 9927 }, { "epoch": 10.07, "learning_rate": 5.1334764403810515e-06, "loss": 0.0249, "step": 9928 }, { "epoch": 10.08, "learning_rate": 5.131561490759637e-06, "loss": 0.0183, "step": 9929 }, { "epoch": 10.08, "learning_rate": 5.129646775095432e-06, "loss": 0.0199, "step": 9930 }, { "epoch": 10.08, "learning_rate": 5.127732293480453e-06, "loss": 0.0197, "step": 9931 }, { "epoch": 10.08, "learning_rate": 5.125818046006707e-06, "loss": 0.0185, "step": 9932 }, { "epoch": 10.08, "learning_rate": 5.123904032766176e-06, "loss": 0.0203, "step": 9933 }, { "epoch": 10.08, "learning_rate": 5.12199025385085e-06, "loss": 0.027, "step": 9934 }, { "epoch": 10.08, "learning_rate": 5.1200767093526906e-06, "loss": 0.0261, "step": 9935 }, { "epoch": 10.08, "learning_rate": 5.1181633993636595e-06, "loss": 0.0198, "step": 9936 }, { "epoch": 10.08, "learning_rate": 5.1162503239756955e-06, "loss": 0.0195, "step": 9937 }, { "epoch": 10.08, "learning_rate": 5.114337483280737e-06, "loss": 0.0244, "step": 9938 }, { "epoch": 10.09, "learning_rate": 5.112424877370712e-06, "loss": 0.0191, "step": 9939 }, { "epoch": 10.09, "learning_rate": 5.110512506337524e-06, "loss": 0.0221, "step": 9940 }, { "epoch": 10.09, "learning_rate": 5.108600370273077e-06, "loss": 0.0225, "step": 9941 }, { "epoch": 10.09, "learning_rate": 5.106688469269261e-06, "loss": 0.022, "step": 9942 }, { "epoch": 10.09, "learning_rate": 5.104776803417955e-06, "loss": 0.0209, "step": 9943 }, { "epoch": 10.09, "learning_rate": 5.10286537281103e-06, "loss": 0.025, "step": 9944 }, { "epoch": 10.09, "learning_rate": 5.100954177540332e-06, "loss": 0.0255, "step": 9945 }, { "epoch": 10.09, "learning_rate": 5.099043217697713e-06, "loss": 0.0219, "step": 9946 }, { "epoch": 10.09, "learning_rate": 5.097132493375e-06, "loss": 0.0206, "step": 9947 }, { "epoch": 10.09, "learning_rate": 5.095222004664017e-06, "loss": 0.0259, "step": 9948 }, { "epoch": 10.1, "learning_rate": 5.0933117516565776e-06, "loss": 0.0214, "step": 9949 }, { "epoch": 10.1, "learning_rate": 5.091401734444472e-06, "loss": 0.0171, "step": 9950 }, { "epoch": 10.1, "learning_rate": 5.0894919531194995e-06, "loss": 0.021, "step": 9951 }, { "epoch": 10.1, "learning_rate": 5.087582407773428e-06, "loss": 0.0232, "step": 9952 }, { "epoch": 10.1, "learning_rate": 5.085673098498027e-06, "loss": 0.0233, "step": 9953 }, { "epoch": 10.1, "learning_rate": 5.083764025385045e-06, "loss": 0.0227, "step": 9954 }, { "epoch": 10.1, "learning_rate": 5.081855188526228e-06, "loss": 0.0229, "step": 9955 }, { "epoch": 10.1, "learning_rate": 5.079946588013309e-06, "loss": 0.0316, "step": 9956 }, { "epoch": 10.1, "learning_rate": 5.078038223938001e-06, "loss": 0.0189, "step": 9957 }, { "epoch": 10.1, "learning_rate": 5.076130096392016e-06, "loss": 0.0245, "step": 9958 }, { "epoch": 10.11, "learning_rate": 5.07422220546705e-06, "loss": 0.0194, "step": 9959 }, { "epoch": 10.11, "learning_rate": 5.072314551254789e-06, "loss": 0.024, "step": 9960 }, { "epoch": 10.11, "learning_rate": 5.070407133846911e-06, "loss": 0.0217, "step": 9961 }, { "epoch": 10.11, "learning_rate": 5.068499953335071e-06, "loss": 0.0275, "step": 9962 }, { "epoch": 10.11, "learning_rate": 5.066593009810928e-06, "loss": 0.0207, "step": 9963 }, { "epoch": 10.11, "learning_rate": 5.064686303366113e-06, "loss": 0.0233, "step": 9964 }, { "epoch": 10.11, "learning_rate": 5.062779834092261e-06, "loss": 0.0234, "step": 9965 }, { "epoch": 10.11, "learning_rate": 5.060873602080986e-06, "loss": 0.0191, "step": 9966 }, { "epoch": 10.11, "learning_rate": 5.058967607423897e-06, "loss": 0.0245, "step": 9967 }, { "epoch": 10.11, "learning_rate": 5.057061850212588e-06, "loss": 0.0194, "step": 9968 }, { "epoch": 10.12, "learning_rate": 5.055156330538638e-06, "loss": 0.0237, "step": 9969 }, { "epoch": 10.12, "learning_rate": 5.0532510484936235e-06, "loss": 0.0274, "step": 9970 }, { "epoch": 10.12, "learning_rate": 5.0513460041691e-06, "loss": 0.0262, "step": 9971 }, { "epoch": 10.12, "learning_rate": 5.049441197656617e-06, "loss": 0.0211, "step": 9972 }, { "epoch": 10.12, "learning_rate": 5.047536629047716e-06, "loss": 0.02, "step": 9973 }, { "epoch": 10.12, "learning_rate": 5.045632298433916e-06, "loss": 0.0202, "step": 9974 }, { "epoch": 10.12, "learning_rate": 5.043728205906735e-06, "loss": 0.0182, "step": 9975 }, { "epoch": 10.12, "learning_rate": 5.041824351557676e-06, "loss": 0.0217, "step": 9976 }, { "epoch": 10.12, "learning_rate": 5.03992073547823e-06, "loss": 0.0211, "step": 9977 }, { "epoch": 10.12, "learning_rate": 5.038017357759879e-06, "loss": 0.0209, "step": 9978 }, { "epoch": 10.13, "learning_rate": 5.036114218494088e-06, "loss": 0.0239, "step": 9979 }, { "epoch": 10.13, "learning_rate": 5.03421131777232e-06, "loss": 0.0201, "step": 9980 }, { "epoch": 10.13, "learning_rate": 5.032308655686011e-06, "loss": 0.0219, "step": 9981 }, { "epoch": 10.13, "learning_rate": 5.030406232326601e-06, "loss": 0.0214, "step": 9982 }, { "epoch": 10.13, "learning_rate": 5.0285040477855114e-06, "loss": 0.023, "step": 9983 }, { "epoch": 10.13, "learning_rate": 5.026602102154155e-06, "loss": 0.0217, "step": 9984 }, { "epoch": 10.13, "learning_rate": 5.024700395523935e-06, "loss": 0.0173, "step": 9985 }, { "epoch": 10.13, "learning_rate": 5.022798927986231e-06, "loss": 0.0301, "step": 9986 }, { "epoch": 10.13, "learning_rate": 5.0208976996324235e-06, "loss": 0.0241, "step": 9987 }, { "epoch": 10.13, "learning_rate": 5.018996710553883e-06, "loss": 0.0203, "step": 9988 }, { "epoch": 10.14, "learning_rate": 5.0170959608419545e-06, "loss": 0.0223, "step": 9989 }, { "epoch": 10.14, "learning_rate": 5.015195450587988e-06, "loss": 0.0207, "step": 9990 }, { "epoch": 10.14, "learning_rate": 5.013295179883307e-06, "loss": 0.0273, "step": 9991 }, { "epoch": 10.14, "learning_rate": 5.011395148819234e-06, "loss": 0.0223, "step": 9992 }, { "epoch": 10.14, "learning_rate": 5.009495357487078e-06, "loss": 0.0265, "step": 9993 }, { "epoch": 10.14, "learning_rate": 5.007595805978132e-06, "loss": 0.0186, "step": 9994 }, { "epoch": 10.14, "learning_rate": 5.005696494383687e-06, "loss": 0.02, "step": 9995 }, { "epoch": 10.14, "learning_rate": 5.0037974227950095e-06, "loss": 0.0215, "step": 9996 }, { "epoch": 10.14, "learning_rate": 5.0018985913033665e-06, "loss": 0.0217, "step": 9997 }, { "epoch": 10.15, "learning_rate": 5.000000000000003e-06, "loss": 0.0246, "step": 9998 }, { "epoch": 10.15, "learning_rate": 4.998101648976158e-06, "loss": 0.0183, "step": 9999 }, { "epoch": 10.15, "learning_rate": 4.99620353832306e-06, "loss": 0.0258, "step": 10000 }, { "epoch": 10.15, "learning_rate": 4.994305668131925e-06, "loss": 0.0233, "step": 10001 }, { "epoch": 10.15, "learning_rate": 4.9924080384939595e-06, "loss": 0.0266, "step": 10002 }, { "epoch": 10.15, "learning_rate": 4.990510649500351e-06, "loss": 0.0183, "step": 10003 }, { "epoch": 10.15, "learning_rate": 4.98861350124228e-06, "loss": 0.0246, "step": 10004 }, { "epoch": 10.15, "learning_rate": 4.986716593810922e-06, "loss": 0.0244, "step": 10005 }, { "epoch": 10.15, "learning_rate": 4.984819927297427e-06, "loss": 0.0184, "step": 10006 }, { "epoch": 10.15, "learning_rate": 4.982923501792948e-06, "loss": 0.026, "step": 10007 }, { "epoch": 10.16, "learning_rate": 4.981027317388607e-06, "loss": 0.0233, "step": 10008 }, { "epoch": 10.16, "learning_rate": 4.979131374175544e-06, "loss": 0.0195, "step": 10009 }, { "epoch": 10.16, "learning_rate": 4.977235672244858e-06, "loss": 0.0217, "step": 10010 }, { "epoch": 10.16, "learning_rate": 4.975340211687653e-06, "loss": 0.0227, "step": 10011 }, { "epoch": 10.16, "learning_rate": 4.973444992595019e-06, "loss": 0.0201, "step": 10012 }, { "epoch": 10.16, "learning_rate": 4.971550015058027e-06, "loss": 0.022, "step": 10013 }, { "epoch": 10.16, "learning_rate": 4.969655279167749e-06, "loss": 0.0248, "step": 10014 }, { "epoch": 10.16, "learning_rate": 4.96776078501523e-06, "loss": 0.0206, "step": 10015 }, { "epoch": 10.16, "learning_rate": 4.9658665326915155e-06, "loss": 0.025, "step": 10016 }, { "epoch": 10.16, "learning_rate": 4.963972522287637e-06, "loss": 0.0238, "step": 10017 }, { "epoch": 10.17, "learning_rate": 4.962078753894611e-06, "loss": 0.0233, "step": 10018 }, { "epoch": 10.17, "learning_rate": 4.960185227603448e-06, "loss": 0.0222, "step": 10019 }, { "epoch": 10.17, "learning_rate": 4.958291943505137e-06, "loss": 0.0255, "step": 10020 }, { "epoch": 10.17, "learning_rate": 4.9563989016906646e-06, "loss": 0.0206, "step": 10021 }, { "epoch": 10.17, "learning_rate": 4.9545061022510056e-06, "loss": 0.0201, "step": 10022 }, { "epoch": 10.17, "learning_rate": 4.952613545277114e-06, "loss": 0.0245, "step": 10023 }, { "epoch": 10.17, "learning_rate": 4.950721230859945e-06, "loss": 0.0219, "step": 10024 }, { "epoch": 10.17, "learning_rate": 4.9488291590904245e-06, "loss": 0.0275, "step": 10025 }, { "epoch": 10.17, "learning_rate": 4.946937330059493e-06, "loss": 0.0231, "step": 10026 }, { "epoch": 10.17, "learning_rate": 4.945045743858051e-06, "loss": 0.0238, "step": 10027 }, { "epoch": 10.18, "learning_rate": 4.9431544005770074e-06, "loss": 0.0246, "step": 10028 }, { "epoch": 10.18, "learning_rate": 4.941263300307254e-06, "loss": 0.0215, "step": 10029 }, { "epoch": 10.18, "learning_rate": 4.939372443139662e-06, "loss": 0.0192, "step": 10030 }, { "epoch": 10.18, "learning_rate": 4.937481829165102e-06, "loss": 0.0268, "step": 10031 }, { "epoch": 10.18, "learning_rate": 4.935591458474433e-06, "loss": 0.0182, "step": 10032 }, { "epoch": 10.18, "learning_rate": 4.933701331158488e-06, "loss": 0.0275, "step": 10033 }, { "epoch": 10.18, "learning_rate": 4.931811447308112e-06, "loss": 0.02, "step": 10034 }, { "epoch": 10.18, "learning_rate": 4.929921807014117e-06, "loss": 0.0201, "step": 10035 }, { "epoch": 10.18, "learning_rate": 4.928032410367315e-06, "loss": 0.0188, "step": 10036 }, { "epoch": 10.18, "learning_rate": 4.926143257458496e-06, "loss": 0.022, "step": 10037 }, { "epoch": 10.19, "learning_rate": 4.924254348378452e-06, "loss": 0.0199, "step": 10038 }, { "epoch": 10.19, "learning_rate": 4.922365683217957e-06, "loss": 0.0188, "step": 10039 }, { "epoch": 10.19, "learning_rate": 4.920477262067766e-06, "loss": 0.0189, "step": 10040 }, { "epoch": 10.19, "learning_rate": 4.9185890850186355e-06, "loss": 0.0236, "step": 10041 }, { "epoch": 10.19, "learning_rate": 4.916701152161293e-06, "loss": 0.0175, "step": 10042 }, { "epoch": 10.19, "learning_rate": 4.914813463586479e-06, "loss": 0.0262, "step": 10043 }, { "epoch": 10.19, "learning_rate": 4.912926019384898e-06, "loss": 0.0257, "step": 10044 }, { "epoch": 10.19, "learning_rate": 4.9110388196472555e-06, "loss": 0.0227, "step": 10045 }, { "epoch": 10.19, "learning_rate": 4.909151864464248e-06, "loss": 0.0197, "step": 10046 }, { "epoch": 10.19, "learning_rate": 4.9072651539265445e-06, "loss": 0.0292, "step": 10047 }, { "epoch": 10.2, "learning_rate": 4.905378688124819e-06, "loss": 0.0258, "step": 10048 }, { "epoch": 10.2, "learning_rate": 4.90349246714973e-06, "loss": 0.0221, "step": 10049 }, { "epoch": 10.2, "learning_rate": 4.9016064910919104e-06, "loss": 0.0236, "step": 10050 }, { "epoch": 10.2, "learning_rate": 4.8997207600420075e-06, "loss": 0.0191, "step": 10051 }, { "epoch": 10.2, "learning_rate": 4.89783527409063e-06, "loss": 0.0196, "step": 10052 }, { "epoch": 10.2, "learning_rate": 4.895950033328395e-06, "loss": 0.0201, "step": 10053 }, { "epoch": 10.2, "learning_rate": 4.894065037845892e-06, "loss": 0.0207, "step": 10054 }, { "epoch": 10.2, "learning_rate": 4.89218028773371e-06, "loss": 0.0226, "step": 10055 }, { "epoch": 10.2, "learning_rate": 4.890295783082425e-06, "loss": 0.0177, "step": 10056 }, { "epoch": 10.2, "learning_rate": 4.888411523982591e-06, "loss": 0.0252, "step": 10057 }, { "epoch": 10.21, "learning_rate": 4.886527510524764e-06, "loss": 0.0207, "step": 10058 }, { "epoch": 10.21, "learning_rate": 4.884643742799478e-06, "loss": 0.0256, "step": 10059 }, { "epoch": 10.21, "learning_rate": 4.882760220897262e-06, "loss": 0.0223, "step": 10060 }, { "epoch": 10.21, "learning_rate": 4.880876944908635e-06, "loss": 0.0201, "step": 10061 }, { "epoch": 10.21, "learning_rate": 4.878993914924089e-06, "loss": 0.0247, "step": 10062 }, { "epoch": 10.21, "learning_rate": 4.8771111310341245e-06, "loss": 0.0195, "step": 10063 }, { "epoch": 10.21, "learning_rate": 4.875228593329212e-06, "loss": 0.0232, "step": 10064 }, { "epoch": 10.21, "learning_rate": 4.8733463018998214e-06, "loss": 0.0217, "step": 10065 }, { "epoch": 10.21, "learning_rate": 4.871464256836414e-06, "loss": 0.0215, "step": 10066 }, { "epoch": 10.22, "learning_rate": 4.86958245822942e-06, "loss": 0.0234, "step": 10067 }, { "epoch": 10.22, "learning_rate": 4.867700906169287e-06, "loss": 0.0284, "step": 10068 }, { "epoch": 10.22, "learning_rate": 4.865819600746422e-06, "loss": 0.0251, "step": 10069 }, { "epoch": 10.22, "learning_rate": 4.863938542051242e-06, "loss": 0.0239, "step": 10070 }, { "epoch": 10.22, "learning_rate": 4.862057730174135e-06, "loss": 0.0263, "step": 10071 }, { "epoch": 10.22, "learning_rate": 4.860177165205488e-06, "loss": 0.0263, "step": 10072 }, { "epoch": 10.22, "learning_rate": 4.858296847235678e-06, "loss": 0.0233, "step": 10073 }, { "epoch": 10.22, "learning_rate": 4.856416776355057e-06, "loss": 0.0254, "step": 10074 }, { "epoch": 10.22, "learning_rate": 4.854536952653979e-06, "loss": 0.0302, "step": 10075 }, { "epoch": 10.22, "learning_rate": 4.852657376222778e-06, "loss": 0.026, "step": 10076 }, { "epoch": 10.23, "learning_rate": 4.8507780471517806e-06, "loss": 0.0222, "step": 10077 }, { "epoch": 10.23, "learning_rate": 4.848898965531304e-06, "loss": 0.0304, "step": 10078 }, { "epoch": 10.23, "learning_rate": 4.8470201314516394e-06, "loss": 0.0257, "step": 10079 }, { "epoch": 10.23, "learning_rate": 4.845141545003085e-06, "loss": 0.0169, "step": 10080 }, { "epoch": 10.23, "learning_rate": 4.8432632062759096e-06, "loss": 0.0226, "step": 10081 }, { "epoch": 10.23, "learning_rate": 4.841385115360383e-06, "loss": 0.0245, "step": 10082 }, { "epoch": 10.23, "learning_rate": 4.8395072723467585e-06, "loss": 0.0193, "step": 10083 }, { "epoch": 10.23, "learning_rate": 4.837629677325278e-06, "loss": 0.0226, "step": 10084 }, { "epoch": 10.23, "learning_rate": 4.835752330386172e-06, "loss": 0.0221, "step": 10085 }, { "epoch": 10.23, "learning_rate": 4.833875231619654e-06, "loss": 0.0222, "step": 10086 }, { "epoch": 10.24, "learning_rate": 4.8319983811159355e-06, "loss": 0.0175, "step": 10087 }, { "epoch": 10.24, "learning_rate": 4.830121778965203e-06, "loss": 0.0237, "step": 10088 }, { "epoch": 10.24, "learning_rate": 4.828245425257642e-06, "loss": 0.0208, "step": 10089 }, { "epoch": 10.24, "learning_rate": 4.826369320083427e-06, "loss": 0.023, "step": 10090 }, { "epoch": 10.24, "learning_rate": 4.8244934635327065e-06, "loss": 0.0268, "step": 10091 }, { "epoch": 10.24, "learning_rate": 4.822617855695633e-06, "loss": 0.0205, "step": 10092 }, { "epoch": 10.24, "learning_rate": 4.820742496662338e-06, "loss": 0.0229, "step": 10093 }, { "epoch": 10.24, "learning_rate": 4.818867386522944e-06, "loss": 0.0252, "step": 10094 }, { "epoch": 10.24, "learning_rate": 4.816992525367566e-06, "loss": 0.0231, "step": 10095 }, { "epoch": 10.24, "learning_rate": 4.8151179132862945e-06, "loss": 0.0196, "step": 10096 }, { "epoch": 10.25, "learning_rate": 4.813243550369222e-06, "loss": 0.0196, "step": 10097 }, { "epoch": 10.25, "learning_rate": 4.811369436706417e-06, "loss": 0.0204, "step": 10098 }, { "epoch": 10.25, "learning_rate": 4.8094955723879435e-06, "loss": 0.0239, "step": 10099 }, { "epoch": 10.25, "learning_rate": 4.807621957503853e-06, "loss": 0.0235, "step": 10100 }, { "epoch": 10.25, "learning_rate": 4.8057485921441825e-06, "loss": 0.0161, "step": 10101 }, { "epoch": 10.25, "learning_rate": 4.803875476398964e-06, "loss": 0.022, "step": 10102 }, { "epoch": 10.25, "learning_rate": 4.802002610358203e-06, "loss": 0.0216, "step": 10103 }, { "epoch": 10.25, "learning_rate": 4.800129994111905e-06, "loss": 0.0227, "step": 10104 }, { "epoch": 10.25, "learning_rate": 4.7982576277500655e-06, "loss": 0.0202, "step": 10105 }, { "epoch": 10.25, "learning_rate": 4.796385511362655e-06, "loss": 0.021, "step": 10106 }, { "epoch": 10.26, "learning_rate": 4.794513645039646e-06, "loss": 0.0249, "step": 10107 }, { "epoch": 10.26, "learning_rate": 4.792642028870985e-06, "loss": 0.0225, "step": 10108 }, { "epoch": 10.26, "learning_rate": 4.79077066294662e-06, "loss": 0.0193, "step": 10109 }, { "epoch": 10.26, "learning_rate": 4.78889954735648e-06, "loss": 0.0255, "step": 10110 }, { "epoch": 10.26, "learning_rate": 4.787028682190483e-06, "loss": 0.0204, "step": 10111 }, { "epoch": 10.26, "learning_rate": 4.78515806753854e-06, "loss": 0.0233, "step": 10112 }, { "epoch": 10.26, "learning_rate": 4.783287703490536e-06, "loss": 0.0279, "step": 10113 }, { "epoch": 10.26, "learning_rate": 4.7814175901363614e-06, "loss": 0.0207, "step": 10114 }, { "epoch": 10.26, "learning_rate": 4.779547727565878e-06, "loss": 0.027, "step": 10115 }, { "epoch": 10.26, "learning_rate": 4.777678115868948e-06, "loss": 0.0264, "step": 10116 }, { "epoch": 10.27, "learning_rate": 4.775808755135418e-06, "loss": 0.0269, "step": 10117 }, { "epoch": 10.27, "learning_rate": 4.77393964545512e-06, "loss": 0.0211, "step": 10118 }, { "epoch": 10.27, "learning_rate": 4.7720707869178815e-06, "loss": 0.0227, "step": 10119 }, { "epoch": 10.27, "learning_rate": 4.770202179613504e-06, "loss": 0.0221, "step": 10120 }, { "epoch": 10.27, "learning_rate": 4.7683338236317885e-06, "loss": 0.0229, "step": 10121 }, { "epoch": 10.27, "learning_rate": 4.7664657190625244e-06, "loss": 0.0264, "step": 10122 }, { "epoch": 10.27, "learning_rate": 4.7645978659954774e-06, "loss": 0.0182, "step": 10123 }, { "epoch": 10.27, "learning_rate": 4.762730264520417e-06, "loss": 0.0262, "step": 10124 }, { "epoch": 10.27, "learning_rate": 4.76086291472708e-06, "loss": 0.0223, "step": 10125 }, { "epoch": 10.27, "learning_rate": 4.758995816705221e-06, "loss": 0.0208, "step": 10126 }, { "epoch": 10.28, "learning_rate": 4.757128970544551e-06, "loss": 0.0245, "step": 10127 }, { "epoch": 10.28, "learning_rate": 4.755262376334788e-06, "loss": 0.0236, "step": 10128 }, { "epoch": 10.28, "learning_rate": 4.753396034165638e-06, "loss": 0.0272, "step": 10129 }, { "epoch": 10.28, "learning_rate": 4.75152994412678e-06, "loss": 0.0248, "step": 10130 }, { "epoch": 10.28, "learning_rate": 4.749664106307895e-06, "loss": 0.0215, "step": 10131 }, { "epoch": 10.28, "learning_rate": 4.747798520798652e-06, "loss": 0.0161, "step": 10132 }, { "epoch": 10.28, "learning_rate": 4.745933187688696e-06, "loss": 0.0242, "step": 10133 }, { "epoch": 10.28, "learning_rate": 4.74406810706767e-06, "loss": 0.0218, "step": 10134 }, { "epoch": 10.28, "learning_rate": 4.7422032790252024e-06, "loss": 0.022, "step": 10135 }, { "epoch": 10.29, "learning_rate": 4.740338703650914e-06, "loss": 0.019, "step": 10136 }, { "epoch": 10.29, "learning_rate": 4.7384743810343994e-06, "loss": 0.0198, "step": 10137 }, { "epoch": 10.29, "learning_rate": 4.736610311265255e-06, "loss": 0.0262, "step": 10138 }, { "epoch": 10.29, "learning_rate": 4.734746494433064e-06, "loss": 0.0247, "step": 10139 }, { "epoch": 10.29, "learning_rate": 4.7328829306273886e-06, "loss": 0.025, "step": 10140 }, { "epoch": 10.29, "learning_rate": 4.731019619937787e-06, "loss": 0.0238, "step": 10141 }, { "epoch": 10.29, "learning_rate": 4.729156562453794e-06, "loss": 0.0212, "step": 10142 }, { "epoch": 10.29, "learning_rate": 4.7272937582649545e-06, "loss": 0.0224, "step": 10143 }, { "epoch": 10.29, "learning_rate": 4.725431207460777e-06, "loss": 0.0261, "step": 10144 }, { "epoch": 10.29, "learning_rate": 4.723568910130771e-06, "loss": 0.0185, "step": 10145 }, { "epoch": 10.3, "learning_rate": 4.721706866364435e-06, "loss": 0.0184, "step": 10146 }, { "epoch": 10.3, "learning_rate": 4.719845076251244e-06, "loss": 0.0265, "step": 10147 }, { "epoch": 10.3, "learning_rate": 4.71798353988067e-06, "loss": 0.0231, "step": 10148 }, { "epoch": 10.3, "learning_rate": 4.716122257342177e-06, "loss": 0.0216, "step": 10149 }, { "epoch": 10.3, "learning_rate": 4.714261228725198e-06, "loss": 0.027, "step": 10150 }, { "epoch": 10.3, "learning_rate": 4.712400454119181e-06, "loss": 0.0215, "step": 10151 }, { "epoch": 10.3, "learning_rate": 4.710539933613537e-06, "loss": 0.024, "step": 10152 }, { "epoch": 10.3, "learning_rate": 4.708679667297682e-06, "loss": 0.0166, "step": 10153 }, { "epoch": 10.3, "learning_rate": 4.706819655261006e-06, "loss": 0.0223, "step": 10154 }, { "epoch": 10.3, "learning_rate": 4.7049598975928955e-06, "loss": 0.0243, "step": 10155 }, { "epoch": 10.31, "learning_rate": 4.703100394382728e-06, "loss": 0.0218, "step": 10156 }, { "epoch": 10.31, "learning_rate": 4.701241145719856e-06, "loss": 0.0252, "step": 10157 }, { "epoch": 10.31, "learning_rate": 4.699382151693636e-06, "loss": 0.0293, "step": 10158 }, { "epoch": 10.31, "learning_rate": 4.6975234123933885e-06, "loss": 0.0243, "step": 10159 }, { "epoch": 10.31, "learning_rate": 4.695664927908453e-06, "loss": 0.0181, "step": 10160 }, { "epoch": 10.31, "learning_rate": 4.693806698328138e-06, "loss": 0.02, "step": 10161 }, { "epoch": 10.31, "learning_rate": 4.691948723741735e-06, "loss": 0.0227, "step": 10162 }, { "epoch": 10.31, "learning_rate": 4.690091004238538e-06, "loss": 0.0289, "step": 10163 }, { "epoch": 10.31, "learning_rate": 4.688233539907815e-06, "loss": 0.026, "step": 10164 }, { "epoch": 10.31, "learning_rate": 4.686376330838831e-06, "loss": 0.0227, "step": 10165 }, { "epoch": 10.32, "learning_rate": 4.684519377120841e-06, "loss": 0.0271, "step": 10166 }, { "epoch": 10.32, "learning_rate": 4.682662678843068e-06, "loss": 0.0227, "step": 10167 }, { "epoch": 10.32, "learning_rate": 4.680806236094756e-06, "loss": 0.029, "step": 10168 }, { "epoch": 10.32, "learning_rate": 4.678950048965106e-06, "loss": 0.0223, "step": 10169 }, { "epoch": 10.32, "learning_rate": 4.677094117543324e-06, "loss": 0.0243, "step": 10170 }, { "epoch": 10.32, "learning_rate": 4.6752384419185926e-06, "loss": 0.0245, "step": 10171 }, { "epoch": 10.32, "learning_rate": 4.673383022180092e-06, "loss": 0.0197, "step": 10172 }, { "epoch": 10.32, "learning_rate": 4.67152785841699e-06, "loss": 0.0247, "step": 10173 }, { "epoch": 10.32, "learning_rate": 4.6696729507184295e-06, "loss": 0.0226, "step": 10174 }, { "epoch": 10.32, "learning_rate": 4.6678182991735545e-06, "loss": 0.0228, "step": 10175 }, { "epoch": 10.33, "learning_rate": 4.665963903871491e-06, "loss": 0.0248, "step": 10176 }, { "epoch": 10.33, "learning_rate": 4.664109764901354e-06, "loss": 0.0306, "step": 10177 }, { "epoch": 10.33, "learning_rate": 4.6622558823522514e-06, "loss": 0.0219, "step": 10178 }, { "epoch": 10.33, "learning_rate": 4.6604022563132624e-06, "loss": 0.0224, "step": 10179 }, { "epoch": 10.33, "learning_rate": 4.658548886873474e-06, "loss": 0.0224, "step": 10180 }, { "epoch": 10.33, "learning_rate": 4.656695774121946e-06, "loss": 0.0258, "step": 10181 }, { "epoch": 10.33, "learning_rate": 4.654842918147731e-06, "loss": 0.0243, "step": 10182 }, { "epoch": 10.33, "learning_rate": 4.652990319039876e-06, "loss": 0.0232, "step": 10183 }, { "epoch": 10.33, "learning_rate": 4.651137976887397e-06, "loss": 0.0228, "step": 10184 }, { "epoch": 10.33, "learning_rate": 4.649285891779327e-06, "loss": 0.0188, "step": 10185 }, { "epoch": 10.34, "learning_rate": 4.647434063804655e-06, "loss": 0.0268, "step": 10186 }, { "epoch": 10.34, "learning_rate": 4.645582493052384e-06, "loss": 0.0227, "step": 10187 }, { "epoch": 10.34, "learning_rate": 4.643731179611481e-06, "loss": 0.0232, "step": 10188 }, { "epoch": 10.34, "learning_rate": 4.64188012357092e-06, "loss": 0.024, "step": 10189 }, { "epoch": 10.34, "learning_rate": 4.640029325019657e-06, "loss": 0.0257, "step": 10190 }, { "epoch": 10.34, "learning_rate": 4.638178784046627e-06, "loss": 0.0231, "step": 10191 }, { "epoch": 10.34, "learning_rate": 4.636328500740761e-06, "loss": 0.0257, "step": 10192 }, { "epoch": 10.34, "learning_rate": 4.634478475190979e-06, "loss": 0.0214, "step": 10193 }, { "epoch": 10.34, "learning_rate": 4.632628707486185e-06, "loss": 0.0226, "step": 10194 }, { "epoch": 10.35, "learning_rate": 4.630779197715274e-06, "loss": 0.0204, "step": 10195 }, { "epoch": 10.35, "learning_rate": 4.628929945967118e-06, "loss": 0.0284, "step": 10196 }, { "epoch": 10.35, "learning_rate": 4.627080952330594e-06, "loss": 0.0202, "step": 10197 }, { "epoch": 10.35, "learning_rate": 4.6252322168945474e-06, "loss": 0.0295, "step": 10198 }, { "epoch": 10.35, "learning_rate": 4.623383739747827e-06, "loss": 0.0229, "step": 10199 }, { "epoch": 10.35, "learning_rate": 4.62153552097926e-06, "loss": 0.0197, "step": 10200 }, { "epoch": 10.35, "learning_rate": 4.619687560677666e-06, "loss": 0.0251, "step": 10201 }, { "epoch": 10.35, "learning_rate": 4.617839858931855e-06, "loss": 0.0222, "step": 10202 }, { "epoch": 10.35, "learning_rate": 4.615992415830611e-06, "loss": 0.0245, "step": 10203 }, { "epoch": 10.35, "learning_rate": 4.614145231462719e-06, "loss": 0.0221, "step": 10204 }, { "epoch": 10.36, "learning_rate": 4.61229830591695e-06, "loss": 0.0235, "step": 10205 }, { "epoch": 10.36, "learning_rate": 4.610451639282053e-06, "loss": 0.0285, "step": 10206 }, { "epoch": 10.36, "learning_rate": 4.6086052316467785e-06, "loss": 0.0262, "step": 10207 }, { "epoch": 10.36, "learning_rate": 4.606759083099851e-06, "loss": 0.0257, "step": 10208 }, { "epoch": 10.36, "learning_rate": 4.6049131937299896e-06, "loss": 0.0202, "step": 10209 }, { "epoch": 10.36, "learning_rate": 4.603067563625903e-06, "loss": 0.0257, "step": 10210 }, { "epoch": 10.36, "learning_rate": 4.601222192876284e-06, "loss": 0.0227, "step": 10211 }, { "epoch": 10.36, "learning_rate": 4.5993770815698166e-06, "loss": 0.0259, "step": 10212 }, { "epoch": 10.36, "learning_rate": 4.597532229795162e-06, "loss": 0.03, "step": 10213 }, { "epoch": 10.36, "learning_rate": 4.5956876376409835e-06, "loss": 0.0193, "step": 10214 }, { "epoch": 10.37, "learning_rate": 4.593843305195918e-06, "loss": 0.0233, "step": 10215 }, { "epoch": 10.37, "learning_rate": 4.5919992325485995e-06, "loss": 0.0216, "step": 10216 }, { "epoch": 10.37, "learning_rate": 4.590155419787647e-06, "loss": 0.0227, "step": 10217 }, { "epoch": 10.37, "learning_rate": 4.588311867001667e-06, "loss": 0.0271, "step": 10218 }, { "epoch": 10.37, "learning_rate": 4.5864685742792555e-06, "loss": 0.0197, "step": 10219 }, { "epoch": 10.37, "learning_rate": 4.584625541708987e-06, "loss": 0.025, "step": 10220 }, { "epoch": 10.37, "learning_rate": 4.5827827693794335e-06, "loss": 0.0211, "step": 10221 }, { "epoch": 10.37, "learning_rate": 4.580940257379156e-06, "loss": 0.0222, "step": 10222 }, { "epoch": 10.37, "learning_rate": 4.57909800579669e-06, "loss": 0.0203, "step": 10223 }, { "epoch": 10.37, "learning_rate": 4.577256014720573e-06, "loss": 0.0252, "step": 10224 }, { "epoch": 10.38, "learning_rate": 4.575414284239316e-06, "loss": 0.0174, "step": 10225 }, { "epoch": 10.38, "learning_rate": 4.573572814441431e-06, "loss": 0.0246, "step": 10226 }, { "epoch": 10.38, "learning_rate": 4.5717316054154095e-06, "loss": 0.0241, "step": 10227 }, { "epoch": 10.38, "learning_rate": 4.569890657249733e-06, "loss": 0.0273, "step": 10228 }, { "epoch": 10.38, "learning_rate": 4.568049970032873e-06, "loss": 0.0259, "step": 10229 }, { "epoch": 10.38, "learning_rate": 4.56620954385328e-06, "loss": 0.0231, "step": 10230 }, { "epoch": 10.38, "learning_rate": 4.564369378799401e-06, "loss": 0.023, "step": 10231 }, { "epoch": 10.38, "learning_rate": 4.562529474959664e-06, "loss": 0.0218, "step": 10232 }, { "epoch": 10.38, "learning_rate": 4.560689832422488e-06, "loss": 0.0217, "step": 10233 }, { "epoch": 10.38, "learning_rate": 4.55885045127628e-06, "loss": 0.027, "step": 10234 }, { "epoch": 10.39, "learning_rate": 4.557011331609432e-06, "loss": 0.0262, "step": 10235 }, { "epoch": 10.39, "learning_rate": 4.5551724735103285e-06, "loss": 0.0286, "step": 10236 }, { "epoch": 10.39, "learning_rate": 4.553333877067331e-06, "loss": 0.0268, "step": 10237 }, { "epoch": 10.39, "learning_rate": 4.551495542368798e-06, "loss": 0.0283, "step": 10238 }, { "epoch": 10.39, "learning_rate": 4.549657469503076e-06, "loss": 0.0223, "step": 10239 }, { "epoch": 10.39, "learning_rate": 4.547819658558488e-06, "loss": 0.0295, "step": 10240 }, { "epoch": 10.39, "learning_rate": 4.545982109623358e-06, "loss": 0.0273, "step": 10241 }, { "epoch": 10.39, "learning_rate": 4.544144822785981e-06, "loss": 0.0214, "step": 10242 }, { "epoch": 10.39, "learning_rate": 4.542307798134665e-06, "loss": 0.0266, "step": 10243 }, { "epoch": 10.39, "learning_rate": 4.540471035757678e-06, "loss": 0.0222, "step": 10244 }, { "epoch": 10.4, "learning_rate": 4.5386345357432905e-06, "loss": 0.0246, "step": 10245 }, { "epoch": 10.4, "learning_rate": 4.536798298179761e-06, "loss": 0.0253, "step": 10246 }, { "epoch": 10.4, "learning_rate": 4.534962323155325e-06, "loss": 0.0295, "step": 10247 }, { "epoch": 10.4, "learning_rate": 4.533126610758215e-06, "loss": 0.0239, "step": 10248 }, { "epoch": 10.4, "learning_rate": 4.531291161076652e-06, "loss": 0.0249, "step": 10249 }, { "epoch": 10.4, "learning_rate": 4.529455974198832e-06, "loss": 0.0214, "step": 10250 }, { "epoch": 10.4, "learning_rate": 4.527621050212951e-06, "loss": 0.026, "step": 10251 }, { "epoch": 10.4, "learning_rate": 4.525786389207187e-06, "loss": 0.0213, "step": 10252 }, { "epoch": 10.4, "learning_rate": 4.523951991269712e-06, "loss": 0.0202, "step": 10253 }, { "epoch": 10.4, "learning_rate": 4.52211785648867e-06, "loss": 0.0188, "step": 10254 }, { "epoch": 10.41, "learning_rate": 4.5202839849522066e-06, "loss": 0.0251, "step": 10255 }, { "epoch": 10.41, "learning_rate": 4.518450376748455e-06, "loss": 0.0211, "step": 10256 }, { "epoch": 10.41, "learning_rate": 4.51661703196552e-06, "loss": 0.0243, "step": 10257 }, { "epoch": 10.41, "learning_rate": 4.514783950691516e-06, "loss": 0.0227, "step": 10258 }, { "epoch": 10.41, "learning_rate": 4.51295113301452e-06, "loss": 0.0215, "step": 10259 }, { "epoch": 10.41, "learning_rate": 4.5111185790226256e-06, "loss": 0.0201, "step": 10260 }, { "epoch": 10.41, "learning_rate": 4.5092862888038855e-06, "loss": 0.0209, "step": 10261 }, { "epoch": 10.41, "learning_rate": 4.507454262446358e-06, "loss": 0.0244, "step": 10262 }, { "epoch": 10.41, "learning_rate": 4.505622500038084e-06, "loss": 0.0265, "step": 10263 }, { "epoch": 10.42, "learning_rate": 4.503791001667084e-06, "loss": 0.0254, "step": 10264 }, { "epoch": 10.42, "learning_rate": 4.5019597674213766e-06, "loss": 0.0194, "step": 10265 }, { "epoch": 10.42, "learning_rate": 4.5001287973889665e-06, "loss": 0.0299, "step": 10266 }, { "epoch": 10.42, "learning_rate": 4.498298091657833e-06, "loss": 0.0277, "step": 10267 }, { "epoch": 10.42, "learning_rate": 4.496467650315963e-06, "loss": 0.0298, "step": 10268 }, { "epoch": 10.42, "learning_rate": 4.494637473451314e-06, "loss": 0.0227, "step": 10269 }, { "epoch": 10.42, "learning_rate": 4.492807561151841e-06, "loss": 0.0206, "step": 10270 }, { "epoch": 10.42, "learning_rate": 4.490977913505477e-06, "loss": 0.0249, "step": 10271 }, { "epoch": 10.42, "learning_rate": 4.489148530600148e-06, "loss": 0.0256, "step": 10272 }, { "epoch": 10.42, "learning_rate": 4.487319412523772e-06, "loss": 0.0226, "step": 10273 }, { "epoch": 10.43, "learning_rate": 4.485490559364243e-06, "loss": 0.0257, "step": 10274 }, { "epoch": 10.43, "learning_rate": 4.4836619712094485e-06, "loss": 0.0193, "step": 10275 }, { "epoch": 10.43, "learning_rate": 4.481833648147267e-06, "loss": 0.0224, "step": 10276 }, { "epoch": 10.43, "learning_rate": 4.480005590265557e-06, "loss": 0.0201, "step": 10277 }, { "epoch": 10.43, "learning_rate": 4.478177797652173e-06, "loss": 0.0209, "step": 10278 }, { "epoch": 10.43, "learning_rate": 4.476350270394942e-06, "loss": 0.0239, "step": 10279 }, { "epoch": 10.43, "learning_rate": 4.474523008581697e-06, "loss": 0.0268, "step": 10280 }, { "epoch": 10.43, "learning_rate": 4.472696012300239e-06, "loss": 0.0223, "step": 10281 }, { "epoch": 10.43, "learning_rate": 4.470869281638373e-06, "loss": 0.0206, "step": 10282 }, { "epoch": 10.43, "learning_rate": 4.469042816683884e-06, "loss": 0.0239, "step": 10283 }, { "epoch": 10.44, "learning_rate": 4.467216617524537e-06, "loss": 0.0258, "step": 10284 }, { "epoch": 10.44, "learning_rate": 4.465390684248103e-06, "loss": 0.0178, "step": 10285 }, { "epoch": 10.44, "learning_rate": 4.463565016942321e-06, "loss": 0.0205, "step": 10286 }, { "epoch": 10.44, "learning_rate": 4.461739615694929e-06, "loss": 0.0302, "step": 10287 }, { "epoch": 10.44, "learning_rate": 4.459914480593644e-06, "loss": 0.0214, "step": 10288 }, { "epoch": 10.44, "learning_rate": 4.458089611726178e-06, "loss": 0.0223, "step": 10289 }, { "epoch": 10.44, "learning_rate": 4.456265009180227e-06, "loss": 0.0233, "step": 10290 }, { "epoch": 10.44, "learning_rate": 4.454440673043471e-06, "loss": 0.0286, "step": 10291 }, { "epoch": 10.44, "learning_rate": 4.452616603403581e-06, "loss": 0.0271, "step": 10292 }, { "epoch": 10.44, "learning_rate": 4.450792800348215e-06, "loss": 0.0302, "step": 10293 }, { "epoch": 10.45, "learning_rate": 4.448969263965019e-06, "loss": 0.0224, "step": 10294 }, { "epoch": 10.45, "learning_rate": 4.447145994341625e-06, "loss": 0.0232, "step": 10295 }, { "epoch": 10.45, "learning_rate": 4.445322991565648e-06, "loss": 0.0295, "step": 10296 }, { "epoch": 10.45, "learning_rate": 4.443500255724699e-06, "loss": 0.0281, "step": 10297 }, { "epoch": 10.45, "learning_rate": 4.441677786906366e-06, "loss": 0.021, "step": 10298 }, { "epoch": 10.45, "learning_rate": 4.439855585198231e-06, "loss": 0.021, "step": 10299 }, { "epoch": 10.45, "learning_rate": 4.438033650687865e-06, "loss": 0.0185, "step": 10300 }, { "epoch": 10.45, "learning_rate": 4.436211983462814e-06, "loss": 0.0206, "step": 10301 }, { "epoch": 10.45, "learning_rate": 4.434390583610634e-06, "loss": 0.0235, "step": 10302 }, { "epoch": 10.45, "learning_rate": 4.432569451218841e-06, "loss": 0.0221, "step": 10303 }, { "epoch": 10.46, "learning_rate": 4.4307485863749566e-06, "loss": 0.0218, "step": 10304 }, { "epoch": 10.46, "learning_rate": 4.4289279891664876e-06, "loss": 0.0327, "step": 10305 }, { "epoch": 10.46, "learning_rate": 4.427107659680916e-06, "loss": 0.0267, "step": 10306 }, { "epoch": 10.46, "learning_rate": 4.425287598005728e-06, "loss": 0.0229, "step": 10307 }, { "epoch": 10.46, "learning_rate": 4.423467804228381e-06, "loss": 0.0262, "step": 10308 }, { "epoch": 10.46, "learning_rate": 4.4216482784363305e-06, "loss": 0.0236, "step": 10309 }, { "epoch": 10.46, "learning_rate": 4.419829020717014e-06, "loss": 0.0248, "step": 10310 }, { "epoch": 10.46, "learning_rate": 4.418010031157858e-06, "loss": 0.0255, "step": 10311 }, { "epoch": 10.46, "learning_rate": 4.416191309846281e-06, "loss": 0.0285, "step": 10312 }, { "epoch": 10.46, "learning_rate": 4.414372856869674e-06, "loss": 0.0248, "step": 10313 }, { "epoch": 10.47, "learning_rate": 4.412554672315433e-06, "loss": 0.0214, "step": 10314 }, { "epoch": 10.47, "learning_rate": 4.410736756270926e-06, "loss": 0.0258, "step": 10315 }, { "epoch": 10.47, "learning_rate": 4.408919108823515e-06, "loss": 0.0308, "step": 10316 }, { "epoch": 10.47, "learning_rate": 4.4071017300605525e-06, "loss": 0.0205, "step": 10317 }, { "epoch": 10.47, "learning_rate": 4.405284620069371e-06, "loss": 0.0254, "step": 10318 }, { "epoch": 10.47, "learning_rate": 4.4034677789373e-06, "loss": 0.0274, "step": 10319 }, { "epoch": 10.47, "learning_rate": 4.401651206751639e-06, "loss": 0.0181, "step": 10320 }, { "epoch": 10.47, "learning_rate": 4.399834903599692e-06, "loss": 0.0248, "step": 10321 }, { "epoch": 10.47, "learning_rate": 4.398018869568744e-06, "loss": 0.0267, "step": 10322 }, { "epoch": 10.47, "learning_rate": 4.3962031047460605e-06, "loss": 0.0271, "step": 10323 }, { "epoch": 10.48, "learning_rate": 4.394387609218906e-06, "loss": 0.0195, "step": 10324 }, { "epoch": 10.48, "learning_rate": 4.392572383074519e-06, "loss": 0.025, "step": 10325 }, { "epoch": 10.48, "learning_rate": 4.3907574264001366e-06, "loss": 0.0213, "step": 10326 }, { "epoch": 10.48, "learning_rate": 4.388942739282976e-06, "loss": 0.0199, "step": 10327 }, { "epoch": 10.48, "learning_rate": 4.387128321810244e-06, "loss": 0.024, "step": 10328 }, { "epoch": 10.48, "learning_rate": 4.385314174069139e-06, "loss": 0.0237, "step": 10329 }, { "epoch": 10.48, "learning_rate": 4.383500296146834e-06, "loss": 0.0262, "step": 10330 }, { "epoch": 10.48, "learning_rate": 4.381686688130503e-06, "loss": 0.0203, "step": 10331 }, { "epoch": 10.48, "learning_rate": 4.379873350107293e-06, "loss": 0.0252, "step": 10332 }, { "epoch": 10.49, "learning_rate": 4.378060282164352e-06, "loss": 0.0209, "step": 10333 }, { "epoch": 10.49, "learning_rate": 4.376247484388804e-06, "loss": 0.0199, "step": 10334 }, { "epoch": 10.49, "learning_rate": 4.374434956867768e-06, "loss": 0.0257, "step": 10335 }, { "epoch": 10.49, "learning_rate": 4.37262269968835e-06, "loss": 0.0273, "step": 10336 }, { "epoch": 10.49, "learning_rate": 4.370810712937632e-06, "loss": 0.0236, "step": 10337 }, { "epoch": 10.49, "learning_rate": 4.368998996702694e-06, "loss": 0.0218, "step": 10338 }, { "epoch": 10.49, "learning_rate": 4.367187551070603e-06, "loss": 0.0244, "step": 10339 }, { "epoch": 10.49, "learning_rate": 4.365376376128403e-06, "loss": 0.0227, "step": 10340 }, { "epoch": 10.49, "learning_rate": 4.363565471963139e-06, "loss": 0.0224, "step": 10341 }, { "epoch": 10.49, "learning_rate": 4.361754838661824e-06, "loss": 0.0205, "step": 10342 }, { "epoch": 10.5, "learning_rate": 4.359944476311485e-06, "loss": 0.0279, "step": 10343 }, { "epoch": 10.5, "learning_rate": 4.35813438499911e-06, "loss": 0.0233, "step": 10344 }, { "epoch": 10.5, "learning_rate": 4.356324564811686e-06, "loss": 0.0277, "step": 10345 }, { "epoch": 10.5, "learning_rate": 4.354515015836191e-06, "loss": 0.0293, "step": 10346 }, { "epoch": 10.5, "learning_rate": 4.352705738159579e-06, "loss": 0.028, "step": 10347 }, { "epoch": 10.5, "learning_rate": 4.350896731868797e-06, "loss": 0.0215, "step": 10348 }, { "epoch": 10.5, "learning_rate": 4.349087997050781e-06, "loss": 0.0234, "step": 10349 }, { "epoch": 10.5, "learning_rate": 4.347279533792449e-06, "loss": 0.03, "step": 10350 }, { "epoch": 10.5, "learning_rate": 4.345471342180708e-06, "loss": 0.0229, "step": 10351 }, { "epoch": 10.5, "learning_rate": 4.343663422302454e-06, "loss": 0.0289, "step": 10352 }, { "epoch": 10.51, "learning_rate": 4.341855774244571e-06, "loss": 0.0267, "step": 10353 }, { "epoch": 10.51, "learning_rate": 4.340048398093921e-06, "loss": 0.0226, "step": 10354 }, { "epoch": 10.51, "learning_rate": 4.338241293937362e-06, "loss": 0.0259, "step": 10355 }, { "epoch": 10.51, "learning_rate": 4.336434461861738e-06, "loss": 0.0314, "step": 10356 }, { "epoch": 10.51, "learning_rate": 4.334627901953874e-06, "loss": 0.0246, "step": 10357 }, { "epoch": 10.51, "learning_rate": 4.332821614300591e-06, "loss": 0.0275, "step": 10358 }, { "epoch": 10.51, "learning_rate": 4.331015598988681e-06, "loss": 0.0238, "step": 10359 }, { "epoch": 10.51, "learning_rate": 4.329209856104949e-06, "loss": 0.0252, "step": 10360 }, { "epoch": 10.51, "learning_rate": 4.32740438573616e-06, "loss": 0.0264, "step": 10361 }, { "epoch": 10.51, "learning_rate": 4.325599187969083e-06, "loss": 0.0254, "step": 10362 }, { "epoch": 10.52, "learning_rate": 4.3237942628904685e-06, "loss": 0.0197, "step": 10363 }, { "epoch": 10.52, "learning_rate": 4.321989610587049e-06, "loss": 0.0264, "step": 10364 }, { "epoch": 10.52, "learning_rate": 4.3201852311455525e-06, "loss": 0.027, "step": 10365 }, { "epoch": 10.52, "learning_rate": 4.318381124652692e-06, "loss": 0.0225, "step": 10366 }, { "epoch": 10.52, "learning_rate": 4.31657729119516e-06, "loss": 0.032, "step": 10367 }, { "epoch": 10.52, "learning_rate": 4.314773730859645e-06, "loss": 0.0291, "step": 10368 }, { "epoch": 10.52, "learning_rate": 4.312970443732817e-06, "loss": 0.0187, "step": 10369 }, { "epoch": 10.52, "learning_rate": 4.3111674299013394e-06, "loss": 0.0369, "step": 10370 }, { "epoch": 10.52, "learning_rate": 4.3093646894518505e-06, "loss": 0.0233, "step": 10371 }, { "epoch": 10.52, "learning_rate": 4.307562222470985e-06, "loss": 0.0223, "step": 10372 }, { "epoch": 10.53, "learning_rate": 4.305760029045367e-06, "loss": 0.0278, "step": 10373 }, { "epoch": 10.53, "learning_rate": 4.303958109261595e-06, "loss": 0.0263, "step": 10374 }, { "epoch": 10.53, "learning_rate": 4.30215646320627e-06, "loss": 0.0269, "step": 10375 }, { "epoch": 10.53, "learning_rate": 4.30035509096596e-06, "loss": 0.0285, "step": 10376 }, { "epoch": 10.53, "learning_rate": 4.298553992627241e-06, "loss": 0.0229, "step": 10377 }, { "epoch": 10.53, "learning_rate": 4.2967531682766685e-06, "loss": 0.0295, "step": 10378 }, { "epoch": 10.53, "learning_rate": 4.294952618000775e-06, "loss": 0.0215, "step": 10379 }, { "epoch": 10.53, "learning_rate": 4.293152341886094e-06, "loss": 0.0254, "step": 10380 }, { "epoch": 10.53, "learning_rate": 4.291352340019132e-06, "loss": 0.0229, "step": 10381 }, { "epoch": 10.53, "learning_rate": 4.289552612486396e-06, "loss": 0.0242, "step": 10382 }, { "epoch": 10.54, "learning_rate": 4.287753159374375e-06, "loss": 0.024, "step": 10383 }, { "epoch": 10.54, "learning_rate": 4.285953980769532e-06, "loss": 0.0258, "step": 10384 }, { "epoch": 10.54, "learning_rate": 4.284155076758344e-06, "loss": 0.0281, "step": 10385 }, { "epoch": 10.54, "learning_rate": 4.282356447427246e-06, "loss": 0.0247, "step": 10386 }, { "epoch": 10.54, "learning_rate": 4.280558092862683e-06, "loss": 0.0331, "step": 10387 }, { "epoch": 10.54, "learning_rate": 4.278760013151068e-06, "loss": 0.024, "step": 10388 }, { "epoch": 10.54, "learning_rate": 4.276962208378811e-06, "loss": 0.0212, "step": 10389 }, { "epoch": 10.54, "learning_rate": 4.2751646786323145e-06, "loss": 0.0209, "step": 10390 }, { "epoch": 10.54, "learning_rate": 4.273367423997949e-06, "loss": 0.0213, "step": 10391 }, { "epoch": 10.54, "learning_rate": 4.271570444562091e-06, "loss": 0.0234, "step": 10392 }, { "epoch": 10.55, "learning_rate": 4.269773740411092e-06, "loss": 0.0318, "step": 10393 }, { "epoch": 10.55, "learning_rate": 4.267977311631299e-06, "loss": 0.0242, "step": 10394 }, { "epoch": 10.55, "learning_rate": 4.2661811583090395e-06, "loss": 0.0222, "step": 10395 }, { "epoch": 10.55, "learning_rate": 4.264385280530625e-06, "loss": 0.0218, "step": 10396 }, { "epoch": 10.55, "learning_rate": 4.262589678382365e-06, "loss": 0.0244, "step": 10397 }, { "epoch": 10.55, "learning_rate": 4.2607943519505415e-06, "loss": 0.0253, "step": 10398 }, { "epoch": 10.55, "learning_rate": 4.258999301321432e-06, "loss": 0.0297, "step": 10399 }, { "epoch": 10.55, "learning_rate": 4.257204526581307e-06, "loss": 0.0243, "step": 10400 }, { "epoch": 10.55, "learning_rate": 4.255410027816401e-06, "loss": 0.0206, "step": 10401 }, { "epoch": 10.56, "learning_rate": 4.253615805112969e-06, "loss": 0.0239, "step": 10402 }, { "epoch": 10.56, "learning_rate": 4.251821858557218e-06, "loss": 0.027, "step": 10403 }, { "epoch": 10.56, "learning_rate": 4.250028188235369e-06, "loss": 0.0218, "step": 10404 }, { "epoch": 10.56, "learning_rate": 4.24823479423361e-06, "loss": 0.0244, "step": 10405 }, { "epoch": 10.56, "learning_rate": 4.246441676638128e-06, "loss": 0.0227, "step": 10406 }, { "epoch": 10.56, "learning_rate": 4.2446488355350955e-06, "loss": 0.0249, "step": 10407 }, { "epoch": 10.56, "learning_rate": 4.242856271010663e-06, "loss": 0.0214, "step": 10408 }, { "epoch": 10.56, "learning_rate": 4.241063983150977e-06, "loss": 0.0247, "step": 10409 }, { "epoch": 10.56, "learning_rate": 4.2392719720421685e-06, "loss": 0.0271, "step": 10410 }, { "epoch": 10.56, "learning_rate": 4.237480237770352e-06, "loss": 0.0294, "step": 10411 }, { "epoch": 10.57, "learning_rate": 4.235688780421636e-06, "loss": 0.0231, "step": 10412 }, { "epoch": 10.57, "learning_rate": 4.233897600082103e-06, "loss": 0.0243, "step": 10413 }, { "epoch": 10.57, "learning_rate": 4.232106696837838e-06, "loss": 0.0258, "step": 10414 }, { "epoch": 10.57, "learning_rate": 4.230316070774895e-06, "loss": 0.0244, "step": 10415 }, { "epoch": 10.57, "learning_rate": 4.228525721979331e-06, "loss": 0.0247, "step": 10416 }, { "epoch": 10.57, "learning_rate": 4.226735650537184e-06, "loss": 0.0184, "step": 10417 }, { "epoch": 10.57, "learning_rate": 4.224945856534467e-06, "loss": 0.0312, "step": 10418 }, { "epoch": 10.57, "learning_rate": 4.223156340057205e-06, "loss": 0.0238, "step": 10419 }, { "epoch": 10.57, "learning_rate": 4.221367101191384e-06, "loss": 0.0266, "step": 10420 }, { "epoch": 10.57, "learning_rate": 4.21957814002299e-06, "loss": 0.021, "step": 10421 }, { "epoch": 10.58, "learning_rate": 4.217789456637999e-06, "loss": 0.0271, "step": 10422 }, { "epoch": 10.58, "learning_rate": 4.2160010511223595e-06, "loss": 0.0231, "step": 10423 }, { "epoch": 10.58, "learning_rate": 4.214212923562022e-06, "loss": 0.0275, "step": 10424 }, { "epoch": 10.58, "learning_rate": 4.212425074042909e-06, "loss": 0.0253, "step": 10425 }, { "epoch": 10.58, "learning_rate": 4.2106375026509404e-06, "loss": 0.0225, "step": 10426 }, { "epoch": 10.58, "learning_rate": 4.2088502094720216e-06, "loss": 0.0264, "step": 10427 }, { "epoch": 10.58, "learning_rate": 4.207063194592041e-06, "loss": 0.026, "step": 10428 }, { "epoch": 10.58, "learning_rate": 4.205276458096879e-06, "loss": 0.0287, "step": 10429 }, { "epoch": 10.58, "learning_rate": 4.203490000072392e-06, "loss": 0.0243, "step": 10430 }, { "epoch": 10.58, "learning_rate": 4.201703820604437e-06, "loss": 0.0207, "step": 10431 }, { "epoch": 10.59, "learning_rate": 4.199917919778843e-06, "loss": 0.0196, "step": 10432 }, { "epoch": 10.59, "learning_rate": 4.1981322976814366e-06, "loss": 0.0258, "step": 10433 }, { "epoch": 10.59, "learning_rate": 4.196346954398027e-06, "loss": 0.0234, "step": 10434 }, { "epoch": 10.59, "learning_rate": 4.194561890014412e-06, "loss": 0.0194, "step": 10435 }, { "epoch": 10.59, "learning_rate": 4.192777104616378e-06, "loss": 0.0211, "step": 10436 }, { "epoch": 10.59, "learning_rate": 4.190992598289686e-06, "loss": 0.0372, "step": 10437 }, { "epoch": 10.59, "learning_rate": 4.189208371120096e-06, "loss": 0.025, "step": 10438 }, { "epoch": 10.59, "learning_rate": 4.187424423193355e-06, "loss": 0.0254, "step": 10439 }, { "epoch": 10.59, "learning_rate": 4.185640754595183e-06, "loss": 0.0232, "step": 10440 }, { "epoch": 10.59, "learning_rate": 4.183857365411306e-06, "loss": 0.0232, "step": 10441 }, { "epoch": 10.6, "learning_rate": 4.182074255727417e-06, "loss": 0.0272, "step": 10442 }, { "epoch": 10.6, "learning_rate": 4.180291425629209e-06, "loss": 0.0242, "step": 10443 }, { "epoch": 10.6, "learning_rate": 4.178508875202359e-06, "loss": 0.0251, "step": 10444 }, { "epoch": 10.6, "learning_rate": 4.176726604532528e-06, "loss": 0.0247, "step": 10445 }, { "epoch": 10.6, "learning_rate": 4.174944613705366e-06, "loss": 0.0228, "step": 10446 }, { "epoch": 10.6, "learning_rate": 4.173162902806504e-06, "loss": 0.018, "step": 10447 }, { "epoch": 10.6, "learning_rate": 4.171381471921569e-06, "loss": 0.0248, "step": 10448 }, { "epoch": 10.6, "learning_rate": 4.1696003211361644e-06, "loss": 0.0223, "step": 10449 }, { "epoch": 10.6, "learning_rate": 4.1678194505358874e-06, "loss": 0.0252, "step": 10450 }, { "epoch": 10.6, "learning_rate": 4.166038860206318e-06, "loss": 0.0248, "step": 10451 }, { "epoch": 10.61, "learning_rate": 4.164258550233025e-06, "loss": 0.0214, "step": 10452 }, { "epoch": 10.61, "learning_rate": 4.162478520701568e-06, "loss": 0.0219, "step": 10453 }, { "epoch": 10.61, "learning_rate": 4.16069877169748e-06, "loss": 0.0282, "step": 10454 }, { "epoch": 10.61, "learning_rate": 4.15891930330629e-06, "loss": 0.0192, "step": 10455 }, { "epoch": 10.61, "learning_rate": 4.157140115613517e-06, "loss": 0.0288, "step": 10456 }, { "epoch": 10.61, "learning_rate": 4.155361208704654e-06, "loss": 0.0239, "step": 10457 }, { "epoch": 10.61, "learning_rate": 4.153582582665196e-06, "loss": 0.0223, "step": 10458 }, { "epoch": 10.61, "learning_rate": 4.151804237580607e-06, "loss": 0.0297, "step": 10459 }, { "epoch": 10.61, "learning_rate": 4.1500261735363566e-06, "loss": 0.029, "step": 10460 }, { "epoch": 10.61, "learning_rate": 4.148248390617885e-06, "loss": 0.0291, "step": 10461 }, { "epoch": 10.62, "learning_rate": 4.146470888910627e-06, "loss": 0.028, "step": 10462 }, { "epoch": 10.62, "learning_rate": 4.144693668500005e-06, "loss": 0.0209, "step": 10463 }, { "epoch": 10.62, "learning_rate": 4.142916729471419e-06, "loss": 0.03, "step": 10464 }, { "epoch": 10.62, "learning_rate": 4.141140071910264e-06, "loss": 0.0254, "step": 10465 }, { "epoch": 10.62, "learning_rate": 4.139363695901925e-06, "loss": 0.0219, "step": 10466 }, { "epoch": 10.62, "learning_rate": 4.1375876015317565e-06, "loss": 0.022, "step": 10467 }, { "epoch": 10.62, "learning_rate": 4.135811788885116e-06, "loss": 0.0275, "step": 10468 }, { "epoch": 10.62, "learning_rate": 4.13403625804734e-06, "loss": 0.0321, "step": 10469 }, { "epoch": 10.62, "learning_rate": 4.1322610091037596e-06, "loss": 0.0223, "step": 10470 }, { "epoch": 10.63, "learning_rate": 4.130486042139678e-06, "loss": 0.0251, "step": 10471 }, { "epoch": 10.63, "learning_rate": 4.128711357240395e-06, "loss": 0.0211, "step": 10472 }, { "epoch": 10.63, "learning_rate": 4.1269369544912e-06, "loss": 0.0248, "step": 10473 }, { "epoch": 10.63, "learning_rate": 4.125162833977354e-06, "loss": 0.0237, "step": 10474 }, { "epoch": 10.63, "learning_rate": 4.123388995784123e-06, "loss": 0.0289, "step": 10475 }, { "epoch": 10.63, "learning_rate": 4.12161543999674e-06, "loss": 0.0261, "step": 10476 }, { "epoch": 10.63, "learning_rate": 4.119842166700449e-06, "loss": 0.0231, "step": 10477 }, { "epoch": 10.63, "learning_rate": 4.1180691759804534e-06, "loss": 0.0354, "step": 10478 }, { "epoch": 10.63, "learning_rate": 4.116296467921962e-06, "loss": 0.023, "step": 10479 }, { "epoch": 10.63, "learning_rate": 4.114524042610166e-06, "loss": 0.0208, "step": 10480 }, { "epoch": 10.64, "learning_rate": 4.112751900130235e-06, "loss": 0.0254, "step": 10481 }, { "epoch": 10.64, "learning_rate": 4.110980040567333e-06, "loss": 0.0327, "step": 10482 }, { "epoch": 10.64, "learning_rate": 4.109208464006613e-06, "loss": 0.0208, "step": 10483 }, { "epoch": 10.64, "learning_rate": 4.107437170533202e-06, "loss": 0.0278, "step": 10484 }, { "epoch": 10.64, "learning_rate": 4.105666160232225e-06, "loss": 0.0249, "step": 10485 }, { "epoch": 10.64, "learning_rate": 4.103895433188788e-06, "loss": 0.0212, "step": 10486 }, { "epoch": 10.64, "learning_rate": 4.102124989487991e-06, "loss": 0.0263, "step": 10487 }, { "epoch": 10.64, "learning_rate": 4.100354829214905e-06, "loss": 0.0251, "step": 10488 }, { "epoch": 10.64, "learning_rate": 4.098584952454603e-06, "loss": 0.0247, "step": 10489 }, { "epoch": 10.64, "learning_rate": 4.096815359292138e-06, "loss": 0.0268, "step": 10490 }, { "epoch": 10.65, "learning_rate": 4.095046049812545e-06, "loss": 0.023, "step": 10491 }, { "epoch": 10.65, "learning_rate": 4.093277024100853e-06, "loss": 0.0193, "step": 10492 }, { "epoch": 10.65, "learning_rate": 4.091508282242072e-06, "loss": 0.0229, "step": 10493 }, { "epoch": 10.65, "learning_rate": 4.089739824321201e-06, "loss": 0.0249, "step": 10494 }, { "epoch": 10.65, "learning_rate": 4.087971650423232e-06, "loss": 0.0279, "step": 10495 }, { "epoch": 10.65, "learning_rate": 4.086203760633125e-06, "loss": 0.0253, "step": 10496 }, { "epoch": 10.65, "learning_rate": 4.084436155035846e-06, "loss": 0.0203, "step": 10497 }, { "epoch": 10.65, "learning_rate": 4.082668833716332e-06, "loss": 0.0248, "step": 10498 }, { "epoch": 10.65, "learning_rate": 4.080901796759517e-06, "loss": 0.024, "step": 10499 }, { "epoch": 10.65, "learning_rate": 4.079135044250321e-06, "loss": 0.0245, "step": 10500 }, { "epoch": 10.66, "learning_rate": 4.0773685762736345e-06, "loss": 0.0268, "step": 10501 }, { "epoch": 10.66, "learning_rate": 4.075602392914364e-06, "loss": 0.0309, "step": 10502 }, { "epoch": 10.66, "learning_rate": 4.073836494257371e-06, "loss": 0.025, "step": 10503 }, { "epoch": 10.66, "learning_rate": 4.072070880387528e-06, "loss": 0.0254, "step": 10504 }, { "epoch": 10.66, "learning_rate": 4.070305551389675e-06, "loss": 0.0287, "step": 10505 }, { "epoch": 10.66, "learning_rate": 4.068540507348647e-06, "loss": 0.0211, "step": 10506 }, { "epoch": 10.66, "learning_rate": 4.0667757483492735e-06, "loss": 0.0216, "step": 10507 }, { "epoch": 10.66, "learning_rate": 4.065011274476349e-06, "loss": 0.026, "step": 10508 }, { "epoch": 10.66, "learning_rate": 4.0632470858146745e-06, "loss": 0.0223, "step": 10509 }, { "epoch": 10.66, "learning_rate": 4.061483182449028e-06, "loss": 0.0245, "step": 10510 }, { "epoch": 10.67, "learning_rate": 4.0597195644641754e-06, "loss": 0.0235, "step": 10511 }, { "epoch": 10.67, "learning_rate": 4.057956231944873e-06, "loss": 0.0275, "step": 10512 }, { "epoch": 10.67, "learning_rate": 4.056193184975853e-06, "loss": 0.0238, "step": 10513 }, { "epoch": 10.67, "learning_rate": 4.054430423641846e-06, "loss": 0.0196, "step": 10514 }, { "epoch": 10.67, "learning_rate": 4.052667948027555e-06, "loss": 0.0272, "step": 10515 }, { "epoch": 10.67, "learning_rate": 4.0509057582176845e-06, "loss": 0.032, "step": 10516 }, { "epoch": 10.67, "learning_rate": 4.049143854296919e-06, "loss": 0.0216, "step": 10517 }, { "epoch": 10.67, "learning_rate": 4.047382236349917e-06, "loss": 0.0235, "step": 10518 }, { "epoch": 10.67, "learning_rate": 4.045620904461351e-06, "loss": 0.0256, "step": 10519 }, { "epoch": 10.67, "learning_rate": 4.0438598587158515e-06, "loss": 0.0221, "step": 10520 }, { "epoch": 10.68, "learning_rate": 4.042099099198052e-06, "loss": 0.0221, "step": 10521 }, { "epoch": 10.68, "learning_rate": 4.040338625992568e-06, "loss": 0.0224, "step": 10522 }, { "epoch": 10.68, "learning_rate": 4.038578439183998e-06, "loss": 0.0222, "step": 10523 }, { "epoch": 10.68, "learning_rate": 4.0368185388569315e-06, "loss": 0.0272, "step": 10524 }, { "epoch": 10.68, "learning_rate": 4.03505892509594e-06, "loss": 0.0328, "step": 10525 }, { "epoch": 10.68, "learning_rate": 4.033299597985581e-06, "loss": 0.0221, "step": 10526 }, { "epoch": 10.68, "learning_rate": 4.0315405576104074e-06, "loss": 0.0262, "step": 10527 }, { "epoch": 10.68, "learning_rate": 4.029781804054947e-06, "loss": 0.0293, "step": 10528 }, { "epoch": 10.68, "learning_rate": 4.028023337403722e-06, "loss": 0.0225, "step": 10529 }, { "epoch": 10.68, "learning_rate": 4.0262651577412305e-06, "loss": 0.0236, "step": 10530 }, { "epoch": 10.69, "learning_rate": 4.024507265151971e-06, "loss": 0.0289, "step": 10531 }, { "epoch": 10.69, "learning_rate": 4.022749659720414e-06, "loss": 0.0226, "step": 10532 }, { "epoch": 10.69, "learning_rate": 4.020992341531026e-06, "loss": 0.0403, "step": 10533 }, { "epoch": 10.69, "learning_rate": 4.019235310668259e-06, "loss": 0.0258, "step": 10534 }, { "epoch": 10.69, "learning_rate": 4.017478567216538e-06, "loss": 0.0209, "step": 10535 }, { "epoch": 10.69, "learning_rate": 4.0157221112603015e-06, "loss": 0.0244, "step": 10536 }, { "epoch": 10.69, "learning_rate": 4.013965942883945e-06, "loss": 0.0242, "step": 10537 }, { "epoch": 10.69, "learning_rate": 4.012210062171867e-06, "loss": 0.0267, "step": 10538 }, { "epoch": 10.69, "learning_rate": 4.010454469208451e-06, "loss": 0.0241, "step": 10539 }, { "epoch": 10.7, "learning_rate": 4.008699164078056e-06, "loss": 0.0232, "step": 10540 }, { "epoch": 10.7, "learning_rate": 4.006944146865043e-06, "loss": 0.0241, "step": 10541 }, { "epoch": 10.7, "learning_rate": 4.005189417653743e-06, "loss": 0.0228, "step": 10542 }, { "epoch": 10.7, "learning_rate": 4.003434976528486e-06, "loss": 0.0331, "step": 10543 }, { "epoch": 10.7, "learning_rate": 4.001680823573582e-06, "loss": 0.0218, "step": 10544 }, { "epoch": 10.7, "learning_rate": 3.9999269588733284e-06, "loss": 0.0291, "step": 10545 }, { "epoch": 10.7, "learning_rate": 3.998173382512011e-06, "loss": 0.0265, "step": 10546 }, { "epoch": 10.7, "learning_rate": 3.996420094573898e-06, "loss": 0.0297, "step": 10547 }, { "epoch": 10.7, "learning_rate": 3.994667095143244e-06, "loss": 0.0226, "step": 10548 }, { "epoch": 10.7, "learning_rate": 3.992914384304291e-06, "loss": 0.0284, "step": 10549 }, { "epoch": 10.71, "learning_rate": 3.991161962141266e-06, "loss": 0.0221, "step": 10550 }, { "epoch": 10.71, "learning_rate": 3.9894098287383865e-06, "loss": 0.0279, "step": 10551 }, { "epoch": 10.71, "learning_rate": 3.987657984179851e-06, "loss": 0.0235, "step": 10552 }, { "epoch": 10.71, "learning_rate": 3.985906428549851e-06, "loss": 0.0269, "step": 10553 }, { "epoch": 10.71, "learning_rate": 3.984155161932549e-06, "loss": 0.0264, "step": 10554 }, { "epoch": 10.71, "learning_rate": 3.98240418441211e-06, "loss": 0.026, "step": 10555 }, { "epoch": 10.71, "learning_rate": 3.980653496072682e-06, "loss": 0.0242, "step": 10556 }, { "epoch": 10.71, "learning_rate": 3.978903096998388e-06, "loss": 0.024, "step": 10557 }, { "epoch": 10.71, "learning_rate": 3.977152987273352e-06, "loss": 0.0268, "step": 10558 }, { "epoch": 10.71, "learning_rate": 3.975403166981671e-06, "loss": 0.0209, "step": 10559 }, { "epoch": 10.72, "learning_rate": 3.973653636207437e-06, "loss": 0.024, "step": 10560 }, { "epoch": 10.72, "learning_rate": 3.971904395034726e-06, "loss": 0.0251, "step": 10561 }, { "epoch": 10.72, "learning_rate": 3.970155443547598e-06, "loss": 0.0239, "step": 10562 }, { "epoch": 10.72, "learning_rate": 3.968406781830107e-06, "loss": 0.0226, "step": 10563 }, { "epoch": 10.72, "learning_rate": 3.9666584099662756e-06, "loss": 0.0219, "step": 10564 }, { "epoch": 10.72, "learning_rate": 3.964910328040128e-06, "loss": 0.0207, "step": 10565 }, { "epoch": 10.72, "learning_rate": 3.963162536135675e-06, "loss": 0.024, "step": 10566 }, { "epoch": 10.72, "learning_rate": 3.9614150343369006e-06, "loss": 0.0237, "step": 10567 }, { "epoch": 10.72, "learning_rate": 3.959667822727785e-06, "loss": 0.0245, "step": 10568 }, { "epoch": 10.72, "learning_rate": 3.957920901392293e-06, "loss": 0.0271, "step": 10569 }, { "epoch": 10.73, "learning_rate": 3.956174270414378e-06, "loss": 0.0275, "step": 10570 }, { "epoch": 10.73, "learning_rate": 3.95442792987797e-06, "loss": 0.029, "step": 10571 }, { "epoch": 10.73, "learning_rate": 3.952681879866991e-06, "loss": 0.0247, "step": 10572 }, { "epoch": 10.73, "learning_rate": 3.950936120465356e-06, "loss": 0.0294, "step": 10573 }, { "epoch": 10.73, "learning_rate": 3.9491906517569505e-06, "loss": 0.0252, "step": 10574 }, { "epoch": 10.73, "learning_rate": 3.947445473825662e-06, "loss": 0.0316, "step": 10575 }, { "epoch": 10.73, "learning_rate": 3.9457005867553454e-06, "loss": 0.0266, "step": 10576 }, { "epoch": 10.73, "learning_rate": 3.943955990629868e-06, "loss": 0.0253, "step": 10577 }, { "epoch": 10.73, "learning_rate": 3.942211685533056e-06, "loss": 0.0215, "step": 10578 }, { "epoch": 10.73, "learning_rate": 3.940467671548739e-06, "loss": 0.0222, "step": 10579 }, { "epoch": 10.74, "learning_rate": 3.938723948760729e-06, "loss": 0.0225, "step": 10580 }, { "epoch": 10.74, "learning_rate": 3.936980517252815e-06, "loss": 0.026, "step": 10581 }, { "epoch": 10.74, "learning_rate": 3.935237377108785e-06, "loss": 0.029, "step": 10582 }, { "epoch": 10.74, "learning_rate": 3.933494528412408e-06, "loss": 0.0274, "step": 10583 }, { "epoch": 10.74, "learning_rate": 3.931751971247432e-06, "loss": 0.0243, "step": 10584 }, { "epoch": 10.74, "learning_rate": 3.9300097056976015e-06, "loss": 0.0231, "step": 10585 }, { "epoch": 10.74, "learning_rate": 3.9282677318466426e-06, "loss": 0.0257, "step": 10586 }, { "epoch": 10.74, "learning_rate": 3.926526049778269e-06, "loss": 0.0274, "step": 10587 }, { "epoch": 10.74, "learning_rate": 3.924784659576174e-06, "loss": 0.0242, "step": 10588 }, { "epoch": 10.74, "learning_rate": 3.923043561324044e-06, "loss": 0.0238, "step": 10589 }, { "epoch": 10.75, "learning_rate": 3.921302755105554e-06, "loss": 0.0228, "step": 10590 }, { "epoch": 10.75, "learning_rate": 3.91956224100435e-06, "loss": 0.0292, "step": 10591 }, { "epoch": 10.75, "learning_rate": 3.917822019104084e-06, "loss": 0.026, "step": 10592 }, { "epoch": 10.75, "learning_rate": 3.916082089488372e-06, "loss": 0.0247, "step": 10593 }, { "epoch": 10.75, "learning_rate": 3.9143424522408405e-06, "loss": 0.0294, "step": 10594 }, { "epoch": 10.75, "learning_rate": 3.912603107445085e-06, "loss": 0.0228, "step": 10595 }, { "epoch": 10.75, "learning_rate": 3.910864055184689e-06, "loss": 0.023, "step": 10596 }, { "epoch": 10.75, "learning_rate": 3.909125295543228e-06, "loss": 0.0277, "step": 10597 }, { "epoch": 10.75, "learning_rate": 3.907386828604254e-06, "loss": 0.0231, "step": 10598 }, { "epoch": 10.75, "learning_rate": 3.905648654451313e-06, "loss": 0.0293, "step": 10599 }, { "epoch": 10.76, "learning_rate": 3.90391077316794e-06, "loss": 0.0192, "step": 10600 }, { "epoch": 10.76, "learning_rate": 3.902173184837642e-06, "loss": 0.0275, "step": 10601 }, { "epoch": 10.76, "learning_rate": 3.900435889543924e-06, "loss": 0.0207, "step": 10602 }, { "epoch": 10.76, "learning_rate": 3.898698887370273e-06, "loss": 0.0265, "step": 10603 }, { "epoch": 10.76, "learning_rate": 3.896962178400168e-06, "loss": 0.0243, "step": 10604 }, { "epoch": 10.76, "learning_rate": 3.895225762717059e-06, "loss": 0.0215, "step": 10605 }, { "epoch": 10.76, "learning_rate": 3.893489640404393e-06, "loss": 0.0251, "step": 10606 }, { "epoch": 10.76, "learning_rate": 3.891753811545608e-06, "loss": 0.025, "step": 10607 }, { "epoch": 10.76, "learning_rate": 3.890018276224112e-06, "loss": 0.027, "step": 10608 }, { "epoch": 10.77, "learning_rate": 3.888283034523311e-06, "loss": 0.0233, "step": 10609 }, { "epoch": 10.77, "learning_rate": 3.886548086526594e-06, "loss": 0.0273, "step": 10610 }, { "epoch": 10.77, "learning_rate": 3.884813432317335e-06, "loss": 0.0264, "step": 10611 }, { "epoch": 10.77, "learning_rate": 3.883079071978899e-06, "loss": 0.0305, "step": 10612 }, { "epoch": 10.77, "learning_rate": 3.881345005594625e-06, "loss": 0.0212, "step": 10613 }, { "epoch": 10.77, "learning_rate": 3.879611233247851e-06, "loss": 0.0231, "step": 10614 }, { "epoch": 10.77, "learning_rate": 3.877877755021889e-06, "loss": 0.0255, "step": 10615 }, { "epoch": 10.77, "learning_rate": 3.876144571000044e-06, "loss": 0.0238, "step": 10616 }, { "epoch": 10.77, "learning_rate": 3.874411681265614e-06, "loss": 0.0242, "step": 10617 }, { "epoch": 10.77, "learning_rate": 3.87267908590186e-06, "loss": 0.0223, "step": 10618 }, { "epoch": 10.78, "learning_rate": 3.87094678499206e-06, "loss": 0.0263, "step": 10619 }, { "epoch": 10.78, "learning_rate": 3.869214778619449e-06, "loss": 0.0248, "step": 10620 }, { "epoch": 10.78, "learning_rate": 3.867483066867268e-06, "loss": 0.0224, "step": 10621 }, { "epoch": 10.78, "learning_rate": 3.865751649818727e-06, "loss": 0.0248, "step": 10622 }, { "epoch": 10.78, "learning_rate": 3.864020527557039e-06, "loss": 0.0241, "step": 10623 }, { "epoch": 10.78, "learning_rate": 3.862289700165392e-06, "loss": 0.0269, "step": 10624 }, { "epoch": 10.78, "learning_rate": 3.860559167726961e-06, "loss": 0.0246, "step": 10625 }, { "epoch": 10.78, "learning_rate": 3.858828930324908e-06, "loss": 0.0402, "step": 10626 }, { "epoch": 10.78, "learning_rate": 3.857098988042384e-06, "loss": 0.0232, "step": 10627 }, { "epoch": 10.78, "learning_rate": 3.8553693409625205e-06, "loss": 0.0283, "step": 10628 }, { "epoch": 10.79, "learning_rate": 3.853639989168442e-06, "loss": 0.026, "step": 10629 }, { "epoch": 10.79, "learning_rate": 3.851910932743247e-06, "loss": 0.0249, "step": 10630 }, { "epoch": 10.79, "learning_rate": 3.850182171770032e-06, "loss": 0.0236, "step": 10631 }, { "epoch": 10.79, "learning_rate": 3.848453706331872e-06, "loss": 0.0284, "step": 10632 }, { "epoch": 10.79, "learning_rate": 3.846725536511828e-06, "loss": 0.0215, "step": 10633 }, { "epoch": 10.79, "learning_rate": 3.844997662392955e-06, "loss": 0.0244, "step": 10634 }, { "epoch": 10.79, "learning_rate": 3.843270084058277e-06, "loss": 0.0216, "step": 10635 }, { "epoch": 10.79, "learning_rate": 3.8415428015908285e-06, "loss": 0.0267, "step": 10636 }, { "epoch": 10.79, "learning_rate": 3.839815815073604e-06, "loss": 0.0213, "step": 10637 }, { "epoch": 10.79, "learning_rate": 3.838089124589599e-06, "loss": 0.0248, "step": 10638 }, { "epoch": 10.8, "learning_rate": 3.836362730221797e-06, "loss": 0.0249, "step": 10639 }, { "epoch": 10.8, "learning_rate": 3.83463663205315e-06, "loss": 0.0287, "step": 10640 }, { "epoch": 10.8, "learning_rate": 3.832910830166619e-06, "loss": 0.0249, "step": 10641 }, { "epoch": 10.8, "learning_rate": 3.831185324645128e-06, "loss": 0.0267, "step": 10642 }, { "epoch": 10.8, "learning_rate": 3.829460115571604e-06, "loss": 0.022, "step": 10643 }, { "epoch": 10.8, "learning_rate": 3.827735203028953e-06, "loss": 0.0243, "step": 10644 }, { "epoch": 10.8, "learning_rate": 3.826010587100067e-06, "loss": 0.0311, "step": 10645 }, { "epoch": 10.8, "learning_rate": 3.824286267867825e-06, "loss": 0.0222, "step": 10646 }, { "epoch": 10.8, "learning_rate": 3.822562245415088e-06, "loss": 0.0201, "step": 10647 }, { "epoch": 10.8, "learning_rate": 3.820838519824709e-06, "loss": 0.0254, "step": 10648 }, { "epoch": 10.81, "learning_rate": 3.819115091179519e-06, "loss": 0.0285, "step": 10649 }, { "epoch": 10.81, "learning_rate": 3.8173919595623395e-06, "loss": 0.0286, "step": 10650 }, { "epoch": 10.81, "learning_rate": 3.815669125055983e-06, "loss": 0.0213, "step": 10651 }, { "epoch": 10.81, "learning_rate": 3.8139465877432313e-06, "loss": 0.0275, "step": 10652 }, { "epoch": 10.81, "learning_rate": 3.8122243477068744e-06, "loss": 0.0237, "step": 10653 }, { "epoch": 10.81, "learning_rate": 3.810502405029668e-06, "loss": 0.023, "step": 10654 }, { "epoch": 10.81, "learning_rate": 3.8087807597943628e-06, "loss": 0.0228, "step": 10655 }, { "epoch": 10.81, "learning_rate": 3.8070594120837e-06, "loss": 0.0207, "step": 10656 }, { "epoch": 10.81, "learning_rate": 3.805338361980392e-06, "loss": 0.0248, "step": 10657 }, { "epoch": 10.81, "learning_rate": 3.803617609567152e-06, "loss": 0.0234, "step": 10658 }, { "epoch": 10.82, "learning_rate": 3.801897154926666e-06, "loss": 0.0243, "step": 10659 }, { "epoch": 10.82, "learning_rate": 3.8001769981416157e-06, "loss": 0.023, "step": 10660 }, { "epoch": 10.82, "learning_rate": 3.7984571392946643e-06, "loss": 0.0295, "step": 10661 }, { "epoch": 10.82, "learning_rate": 3.7967375784684623e-06, "loss": 0.027, "step": 10662 }, { "epoch": 10.82, "learning_rate": 3.7950183157456466e-06, "loss": 0.0253, "step": 10663 }, { "epoch": 10.82, "learning_rate": 3.793299351208832e-06, "loss": 0.0223, "step": 10664 }, { "epoch": 10.82, "learning_rate": 3.791580684940629e-06, "loss": 0.0232, "step": 10665 }, { "epoch": 10.82, "learning_rate": 3.7898623170236317e-06, "loss": 0.0251, "step": 10666 }, { "epoch": 10.82, "learning_rate": 3.7881442475404107e-06, "loss": 0.0416, "step": 10667 }, { "epoch": 10.82, "learning_rate": 3.7864264765735347e-06, "loss": 0.0225, "step": 10668 }, { "epoch": 10.83, "learning_rate": 3.784709004205551e-06, "loss": 0.0286, "step": 10669 }, { "epoch": 10.83, "learning_rate": 3.7829918305189985e-06, "loss": 0.0232, "step": 10670 }, { "epoch": 10.83, "learning_rate": 3.78127495559639e-06, "loss": 0.0242, "step": 10671 }, { "epoch": 10.83, "learning_rate": 3.7795583795202363e-06, "loss": 0.0208, "step": 10672 }, { "epoch": 10.83, "learning_rate": 3.7778421023730305e-06, "loss": 0.0267, "step": 10673 }, { "epoch": 10.83, "learning_rate": 3.776126124237245e-06, "loss": 0.0241, "step": 10674 }, { "epoch": 10.83, "learning_rate": 3.774410445195349e-06, "loss": 0.0222, "step": 10675 }, { "epoch": 10.83, "learning_rate": 3.7726950653297822e-06, "loss": 0.0274, "step": 10676 }, { "epoch": 10.83, "learning_rate": 3.7709799847229856e-06, "loss": 0.0222, "step": 10677 }, { "epoch": 10.84, "learning_rate": 3.769265203457375e-06, "loss": 0.0285, "step": 10678 }, { "epoch": 10.84, "learning_rate": 3.7675507216153597e-06, "loss": 0.0238, "step": 10679 }, { "epoch": 10.84, "learning_rate": 3.7658365392793317e-06, "loss": 0.0293, "step": 10680 }, { "epoch": 10.84, "learning_rate": 3.7641226565316603e-06, "loss": 0.0241, "step": 10681 }, { "epoch": 10.84, "learning_rate": 3.7624090734547126e-06, "loss": 0.0238, "step": 10682 }, { "epoch": 10.84, "learning_rate": 3.7606957901308395e-06, "loss": 0.0256, "step": 10683 }, { "epoch": 10.84, "learning_rate": 3.7589828066423674e-06, "loss": 0.0256, "step": 10684 }, { "epoch": 10.84, "learning_rate": 3.757270123071619e-06, "loss": 0.0239, "step": 10685 }, { "epoch": 10.84, "learning_rate": 3.755557739500898e-06, "loss": 0.0317, "step": 10686 }, { "epoch": 10.84, "learning_rate": 3.753845656012498e-06, "loss": 0.0274, "step": 10687 }, { "epoch": 10.85, "learning_rate": 3.752133872688688e-06, "loss": 0.0285, "step": 10688 }, { "epoch": 10.85, "learning_rate": 3.7504223896117343e-06, "loss": 0.0325, "step": 10689 }, { "epoch": 10.85, "learning_rate": 3.7487112068638855e-06, "loss": 0.0265, "step": 10690 }, { "epoch": 10.85, "learning_rate": 3.7470003245273667e-06, "loss": 0.0317, "step": 10691 }, { "epoch": 10.85, "learning_rate": 3.745289742684405e-06, "loss": 0.0168, "step": 10692 }, { "epoch": 10.85, "learning_rate": 3.7435794614171904e-06, "loss": 0.0224, "step": 10693 }, { "epoch": 10.85, "learning_rate": 3.741869480807929e-06, "loss": 0.0275, "step": 10694 }, { "epoch": 10.85, "learning_rate": 3.740159800938784e-06, "loss": 0.0265, "step": 10695 }, { "epoch": 10.85, "learning_rate": 3.7384504218919184e-06, "loss": 0.0214, "step": 10696 }, { "epoch": 10.85, "learning_rate": 3.7367413437494813e-06, "loss": 0.0259, "step": 10697 }, { "epoch": 10.86, "learning_rate": 3.7350325665935994e-06, "loss": 0.0284, "step": 10698 }, { "epoch": 10.86, "learning_rate": 3.7333240905063896e-06, "loss": 0.0295, "step": 10699 }, { "epoch": 10.86, "learning_rate": 3.7316159155699594e-06, "loss": 0.0287, "step": 10700 }, { "epoch": 10.86, "learning_rate": 3.72990804186639e-06, "loss": 0.0309, "step": 10701 }, { "epoch": 10.86, "learning_rate": 3.7282004694777584e-06, "loss": 0.0216, "step": 10702 }, { "epoch": 10.86, "learning_rate": 3.726493198486123e-06, "loss": 0.0258, "step": 10703 }, { "epoch": 10.86, "learning_rate": 3.7247862289735326e-06, "loss": 0.0253, "step": 10704 }, { "epoch": 10.86, "learning_rate": 3.7230795610220083e-06, "loss": 0.0254, "step": 10705 }, { "epoch": 10.86, "learning_rate": 3.721373194713571e-06, "loss": 0.0214, "step": 10706 }, { "epoch": 10.86, "learning_rate": 3.7196671301302234e-06, "loss": 0.0233, "step": 10707 }, { "epoch": 10.87, "learning_rate": 3.717961367353947e-06, "loss": 0.0231, "step": 10708 }, { "epoch": 10.87, "learning_rate": 3.716255906466716e-06, "loss": 0.0251, "step": 10709 }, { "epoch": 10.87, "learning_rate": 3.714550747550488e-06, "loss": 0.0254, "step": 10710 }, { "epoch": 10.87, "learning_rate": 3.7128458906872055e-06, "loss": 0.035, "step": 10711 }, { "epoch": 10.87, "learning_rate": 3.7111413359588024e-06, "loss": 0.0262, "step": 10712 }, { "epoch": 10.87, "learning_rate": 3.7094370834471825e-06, "loss": 0.0281, "step": 10713 }, { "epoch": 10.87, "learning_rate": 3.707733133234255e-06, "loss": 0.0276, "step": 10714 }, { "epoch": 10.87, "learning_rate": 3.7060294854018963e-06, "loss": 0.0255, "step": 10715 }, { "epoch": 10.87, "learning_rate": 3.704326140031981e-06, "loss": 0.0249, "step": 10716 }, { "epoch": 10.87, "learning_rate": 3.7026230972063683e-06, "loss": 0.0242, "step": 10717 }, { "epoch": 10.88, "learning_rate": 3.700920357006892e-06, "loss": 0.0253, "step": 10718 }, { "epoch": 10.88, "learning_rate": 3.6992179195153823e-06, "loss": 0.027, "step": 10719 }, { "epoch": 10.88, "learning_rate": 3.697515784813651e-06, "loss": 0.0258, "step": 10720 }, { "epoch": 10.88, "learning_rate": 3.6958139529835e-06, "loss": 0.0224, "step": 10721 }, { "epoch": 10.88, "learning_rate": 3.694112424106706e-06, "loss": 0.0215, "step": 10722 }, { "epoch": 10.88, "learning_rate": 3.692411198265039e-06, "loss": 0.0267, "step": 10723 }, { "epoch": 10.88, "learning_rate": 3.690710275540259e-06, "loss": 0.0217, "step": 10724 }, { "epoch": 10.88, "learning_rate": 3.6890096560140964e-06, "loss": 0.0224, "step": 10725 }, { "epoch": 10.88, "learning_rate": 3.687309339768279e-06, "loss": 0.0239, "step": 10726 }, { "epoch": 10.88, "learning_rate": 3.6856093268845193e-06, "loss": 0.0242, "step": 10727 }, { "epoch": 10.89, "learning_rate": 3.6839096174445124e-06, "loss": 0.0252, "step": 10728 }, { "epoch": 10.89, "learning_rate": 3.6822102115299406e-06, "loss": 0.0217, "step": 10729 }, { "epoch": 10.89, "learning_rate": 3.6805111092224666e-06, "loss": 0.0257, "step": 10730 }, { "epoch": 10.89, "learning_rate": 3.6788123106037467e-06, "loss": 0.0192, "step": 10731 }, { "epoch": 10.89, "learning_rate": 3.6771138157554134e-06, "loss": 0.0322, "step": 10732 }, { "epoch": 10.89, "learning_rate": 3.675415624759091e-06, "loss": 0.0255, "step": 10733 }, { "epoch": 10.89, "learning_rate": 3.673717737696392e-06, "loss": 0.0221, "step": 10734 }, { "epoch": 10.89, "learning_rate": 3.6720201546488986e-06, "loss": 0.0246, "step": 10735 }, { "epoch": 10.89, "learning_rate": 3.6703228756982058e-06, "loss": 0.0266, "step": 10736 }, { "epoch": 10.89, "learning_rate": 3.6686259009258653e-06, "loss": 0.0285, "step": 10737 }, { "epoch": 10.9, "learning_rate": 3.6669292304134295e-06, "loss": 0.028, "step": 10738 }, { "epoch": 10.9, "learning_rate": 3.6652328642424406e-06, "loss": 0.0249, "step": 10739 }, { "epoch": 10.9, "learning_rate": 3.6635368024944085e-06, "loss": 0.0368, "step": 10740 }, { "epoch": 10.9, "learning_rate": 3.6618410452508466e-06, "loss": 0.0247, "step": 10741 }, { "epoch": 10.9, "learning_rate": 3.6601455925932404e-06, "loss": 0.0233, "step": 10742 }, { "epoch": 10.9, "learning_rate": 3.65845044460307e-06, "loss": 0.0227, "step": 10743 }, { "epoch": 10.9, "learning_rate": 3.6567556013617954e-06, "loss": 0.0247, "step": 10744 }, { "epoch": 10.9, "learning_rate": 3.6550610629508656e-06, "loss": 0.0208, "step": 10745 }, { "epoch": 10.9, "learning_rate": 3.6533668294517154e-06, "loss": 0.0299, "step": 10746 }, { "epoch": 10.91, "learning_rate": 3.6516729009457564e-06, "loss": 0.0269, "step": 10747 }, { "epoch": 10.91, "learning_rate": 3.6499792775143983e-06, "loss": 0.026, "step": 10748 }, { "epoch": 10.91, "learning_rate": 3.648285959239023e-06, "loss": 0.0276, "step": 10749 }, { "epoch": 10.91, "learning_rate": 3.6465929462010085e-06, "loss": 0.0277, "step": 10750 }, { "epoch": 10.91, "learning_rate": 3.6449002384817176e-06, "loss": 0.0265, "step": 10751 }, { "epoch": 10.91, "learning_rate": 3.6432078361624835e-06, "loss": 0.0249, "step": 10752 }, { "epoch": 10.91, "learning_rate": 3.6415157393246514e-06, "loss": 0.0297, "step": 10753 }, { "epoch": 10.91, "learning_rate": 3.6398239480495247e-06, "loss": 0.0222, "step": 10754 }, { "epoch": 10.91, "learning_rate": 3.6381324624184065e-06, "loss": 0.0262, "step": 10755 }, { "epoch": 10.91, "learning_rate": 3.6364412825125893e-06, "loss": 0.0244, "step": 10756 }, { "epoch": 10.92, "learning_rate": 3.6347504084133356e-06, "loss": 0.024, "step": 10757 }, { "epoch": 10.92, "learning_rate": 3.633059840201908e-06, "loss": 0.0238, "step": 10758 }, { "epoch": 10.92, "learning_rate": 3.631369577959543e-06, "loss": 0.0279, "step": 10759 }, { "epoch": 10.92, "learning_rate": 3.62967962176747e-06, "loss": 0.0286, "step": 10760 }, { "epoch": 10.92, "learning_rate": 3.6279899717069023e-06, "loss": 0.0266, "step": 10761 }, { "epoch": 10.92, "learning_rate": 3.6263006278590364e-06, "loss": 0.0216, "step": 10762 }, { "epoch": 10.92, "learning_rate": 3.6246115903050592e-06, "loss": 0.0332, "step": 10763 }, { "epoch": 10.92, "learning_rate": 3.622922859126132e-06, "loss": 0.0201, "step": 10764 }, { "epoch": 10.92, "learning_rate": 3.6212344344034155e-06, "loss": 0.0234, "step": 10765 }, { "epoch": 10.92, "learning_rate": 3.619546316218042e-06, "loss": 0.0299, "step": 10766 }, { "epoch": 10.93, "learning_rate": 3.6178585046511383e-06, "loss": 0.0314, "step": 10767 }, { "epoch": 10.93, "learning_rate": 3.616170999783817e-06, "loss": 0.0204, "step": 10768 }, { "epoch": 10.93, "learning_rate": 3.614483801697163e-06, "loss": 0.0215, "step": 10769 }, { "epoch": 10.93, "learning_rate": 3.6127969104722683e-06, "loss": 0.0235, "step": 10770 }, { "epoch": 10.93, "learning_rate": 3.6111103261901902e-06, "loss": 0.0245, "step": 10771 }, { "epoch": 10.93, "learning_rate": 3.60942404893198e-06, "loss": 0.0201, "step": 10772 }, { "epoch": 10.93, "learning_rate": 3.6077380787786787e-06, "loss": 0.0232, "step": 10773 }, { "epoch": 10.93, "learning_rate": 3.6060524158112987e-06, "loss": 0.0243, "step": 10774 }, { "epoch": 10.93, "learning_rate": 3.604367060110854e-06, "loss": 0.0254, "step": 10775 }, { "epoch": 10.93, "learning_rate": 3.6026820117583294e-06, "loss": 0.0198, "step": 10776 }, { "epoch": 10.94, "learning_rate": 3.600997270834703e-06, "loss": 0.0288, "step": 10777 }, { "epoch": 10.94, "learning_rate": 3.5993128374209376e-06, "loss": 0.0243, "step": 10778 }, { "epoch": 10.94, "learning_rate": 3.5976287115979813e-06, "loss": 0.0208, "step": 10779 }, { "epoch": 10.94, "learning_rate": 3.595944893446768e-06, "loss": 0.0229, "step": 10780 }, { "epoch": 10.94, "learning_rate": 3.5942613830482087e-06, "loss": 0.026, "step": 10781 }, { "epoch": 10.94, "learning_rate": 3.592578180483209e-06, "loss": 0.0226, "step": 10782 }, { "epoch": 10.94, "learning_rate": 3.5908952858326617e-06, "loss": 0.0297, "step": 10783 }, { "epoch": 10.94, "learning_rate": 3.589212699177432e-06, "loss": 0.0249, "step": 10784 }, { "epoch": 10.94, "learning_rate": 3.587530420598382e-06, "loss": 0.0194, "step": 10785 }, { "epoch": 10.94, "learning_rate": 3.5858484501763534e-06, "loss": 0.0243, "step": 10786 }, { "epoch": 10.95, "learning_rate": 3.5841667879921803e-06, "loss": 0.0211, "step": 10787 }, { "epoch": 10.95, "learning_rate": 3.582485434126669e-06, "loss": 0.0264, "step": 10788 }, { "epoch": 10.95, "learning_rate": 3.5808043886606225e-06, "loss": 0.0195, "step": 10789 }, { "epoch": 10.95, "learning_rate": 3.579123651674826e-06, "loss": 0.0201, "step": 10790 }, { "epoch": 10.95, "learning_rate": 3.5774432232500454e-06, "loss": 0.0289, "step": 10791 }, { "epoch": 10.95, "learning_rate": 3.5757631034670394e-06, "loss": 0.0277, "step": 10792 }, { "epoch": 10.95, "learning_rate": 3.5740832924065417e-06, "loss": 0.0298, "step": 10793 }, { "epoch": 10.95, "learning_rate": 3.57240379014928e-06, "loss": 0.0278, "step": 10794 }, { "epoch": 10.95, "learning_rate": 3.5707245967759663e-06, "loss": 0.0263, "step": 10795 }, { "epoch": 10.95, "learning_rate": 3.569045712367294e-06, "loss": 0.02, "step": 10796 }, { "epoch": 10.96, "learning_rate": 3.5673671370039464e-06, "loss": 0.0232, "step": 10797 }, { "epoch": 10.96, "learning_rate": 3.565688870766584e-06, "loss": 0.0316, "step": 10798 }, { "epoch": 10.96, "learning_rate": 3.5640109137358592e-06, "loss": 0.019, "step": 10799 }, { "epoch": 10.96, "learning_rate": 3.5623332659924114e-06, "loss": 0.0237, "step": 10800 }, { "epoch": 10.96, "learning_rate": 3.560655927616855e-06, "loss": 0.0222, "step": 10801 }, { "epoch": 10.96, "learning_rate": 3.5589788986898e-06, "loss": 0.0252, "step": 10802 }, { "epoch": 10.96, "learning_rate": 3.557302179291836e-06, "loss": 0.0217, "step": 10803 }, { "epoch": 10.96, "learning_rate": 3.5556257695035447e-06, "loss": 0.0285, "step": 10804 }, { "epoch": 10.96, "learning_rate": 3.5539496694054785e-06, "loss": 0.0265, "step": 10805 }, { "epoch": 10.96, "learning_rate": 3.5522738790781884e-06, "loss": 0.0258, "step": 10806 }, { "epoch": 10.97, "learning_rate": 3.5505983986022098e-06, "loss": 0.0242, "step": 10807 }, { "epoch": 10.97, "learning_rate": 3.5489232280580523e-06, "loss": 0.0244, "step": 10808 }, { "epoch": 10.97, "learning_rate": 3.5472483675262246e-06, "loss": 0.0232, "step": 10809 }, { "epoch": 10.97, "learning_rate": 3.5455738170872035e-06, "loss": 0.0271, "step": 10810 }, { "epoch": 10.97, "learning_rate": 3.54389957682147e-06, "loss": 0.0238, "step": 10811 }, { "epoch": 10.97, "learning_rate": 3.542225646809483e-06, "loss": 0.0248, "step": 10812 }, { "epoch": 10.97, "learning_rate": 3.540552027131676e-06, "loss": 0.0252, "step": 10813 }, { "epoch": 10.97, "learning_rate": 3.5388787178684856e-06, "loss": 0.0242, "step": 10814 }, { "epoch": 10.97, "learning_rate": 3.537205719100316e-06, "loss": 0.0252, "step": 10815 }, { "epoch": 10.98, "learning_rate": 3.5355330309075675e-06, "loss": 0.0242, "step": 10816 }, { "epoch": 10.98, "learning_rate": 3.533860653370628e-06, "loss": 0.0228, "step": 10817 }, { "epoch": 10.98, "learning_rate": 3.532188586569856e-06, "loss": 0.0304, "step": 10818 }, { "epoch": 10.98, "learning_rate": 3.5305168305856087e-06, "loss": 0.0236, "step": 10819 }, { "epoch": 10.98, "learning_rate": 3.528845385498223e-06, "loss": 0.0312, "step": 10820 }, { "epoch": 10.98, "learning_rate": 3.527174251388026e-06, "loss": 0.0248, "step": 10821 }, { "epoch": 10.98, "learning_rate": 3.5255034283353197e-06, "loss": 0.0293, "step": 10822 }, { "epoch": 10.98, "learning_rate": 3.523832916420399e-06, "loss": 0.0267, "step": 10823 }, { "epoch": 10.98, "learning_rate": 3.522162715723545e-06, "loss": 0.0263, "step": 10824 }, { "epoch": 10.98, "learning_rate": 3.5204928263250138e-06, "loss": 0.021, "step": 10825 }, { "epoch": 10.99, "learning_rate": 3.5188232483050586e-06, "loss": 0.0252, "step": 10826 }, { "epoch": 10.99, "learning_rate": 3.5171539817439116e-06, "loss": 0.0246, "step": 10827 }, { "epoch": 10.99, "learning_rate": 3.51548502672179e-06, "loss": 0.0267, "step": 10828 }, { "epoch": 10.99, "learning_rate": 3.513816383318901e-06, "loss": 0.0285, "step": 10829 }, { "epoch": 10.99, "learning_rate": 3.512148051615426e-06, "loss": 0.0229, "step": 10830 }, { "epoch": 10.99, "learning_rate": 3.5104800316915454e-06, "loss": 0.0265, "step": 10831 }, { "epoch": 10.99, "learning_rate": 3.508812323627411e-06, "loss": 0.0243, "step": 10832 }, { "epoch": 10.99, "learning_rate": 3.5071449275031667e-06, "loss": 0.0232, "step": 10833 }, { "epoch": 10.99, "learning_rate": 3.5054778433989478e-06, "loss": 0.0262, "step": 10834 }, { "epoch": 10.99, "learning_rate": 3.5038110713948582e-06, "loss": 0.028, "step": 10835 }, { "epoch": 11.0, "learning_rate": 3.502144611571e-06, "loss": 0.0265, "step": 10836 }, { "epoch": 11.0, "learning_rate": 3.5004784640074573e-06, "loss": 0.0231, "step": 10837 }, { "epoch": 11.0, "learning_rate": 3.4988126287843003e-06, "loss": 0.025, "step": 10838 }, { "epoch": 11.0, "learning_rate": 3.4971471059815774e-06, "loss": 0.0305, "step": 10839 }, { "epoch": 11.0, "learning_rate": 3.4954818956793292e-06, "loss": 0.0275, "step": 10840 }, { "epoch": 11.0, "learning_rate": 3.493816997957582e-06, "loss": 0.019, "step": 10841 }, { "epoch": 11.0, "learning_rate": 3.492152412896337e-06, "loss": 0.0176, "step": 10842 }, { "epoch": 11.0, "learning_rate": 3.4904881405755907e-06, "loss": 0.0211, "step": 10843 }, { "epoch": 11.0, "learning_rate": 3.488824181075322e-06, "loss": 0.0218, "step": 10844 }, { "epoch": 11.0, "learning_rate": 3.4871605344754934e-06, "loss": 0.0175, "step": 10845 }, { "epoch": 11.01, "learning_rate": 3.485497200856057e-06, "loss": 0.0188, "step": 10846 }, { "epoch": 11.01, "learning_rate": 3.4838341802969387e-06, "loss": 0.0199, "step": 10847 }, { "epoch": 11.01, "learning_rate": 3.482171472878062e-06, "loss": 0.0194, "step": 10848 }, { "epoch": 11.01, "learning_rate": 3.480509078679325e-06, "loss": 0.0138, "step": 10849 }, { "epoch": 11.01, "learning_rate": 3.478846997780618e-06, "loss": 0.0201, "step": 10850 }, { "epoch": 11.01, "learning_rate": 3.477185230261817e-06, "loss": 0.0203, "step": 10851 }, { "epoch": 11.01, "learning_rate": 3.4755237762027707e-06, "loss": 0.0194, "step": 10852 }, { "epoch": 11.01, "learning_rate": 3.473862635683333e-06, "loss": 0.0207, "step": 10853 }, { "epoch": 11.01, "learning_rate": 3.472201808783323e-06, "loss": 0.0196, "step": 10854 }, { "epoch": 11.01, "learning_rate": 3.4705412955825567e-06, "loss": 0.0181, "step": 10855 }, { "epoch": 11.02, "learning_rate": 3.4688810961608354e-06, "loss": 0.0169, "step": 10856 }, { "epoch": 11.02, "learning_rate": 3.467221210597933e-06, "loss": 0.0193, "step": 10857 }, { "epoch": 11.02, "learning_rate": 3.4655616389736244e-06, "loss": 0.0202, "step": 10858 }, { "epoch": 11.02, "learning_rate": 3.4639023813676564e-06, "loss": 0.0202, "step": 10859 }, { "epoch": 11.02, "learning_rate": 3.462243437859768e-06, "loss": 0.0223, "step": 10860 }, { "epoch": 11.02, "learning_rate": 3.460584808529681e-06, "loss": 0.0249, "step": 10861 }, { "epoch": 11.02, "learning_rate": 3.458926493457102e-06, "loss": 0.0202, "step": 10862 }, { "epoch": 11.02, "learning_rate": 3.457268492721728e-06, "loss": 0.0141, "step": 10863 }, { "epoch": 11.02, "learning_rate": 3.4556108064032288e-06, "loss": 0.0179, "step": 10864 }, { "epoch": 11.02, "learning_rate": 3.453953434581271e-06, "loss": 0.0189, "step": 10865 }, { "epoch": 11.03, "learning_rate": 3.4522963773354956e-06, "loss": 0.018, "step": 10866 }, { "epoch": 11.03, "learning_rate": 3.4506396347455373e-06, "loss": 0.0186, "step": 10867 }, { "epoch": 11.03, "learning_rate": 3.4489832068910158e-06, "loss": 0.0175, "step": 10868 }, { "epoch": 11.03, "learning_rate": 3.4473270938515213e-06, "loss": 0.0146, "step": 10869 }, { "epoch": 11.03, "learning_rate": 3.4456712957066542e-06, "loss": 0.022, "step": 10870 }, { "epoch": 11.03, "learning_rate": 3.4440158125359747e-06, "loss": 0.0211, "step": 10871 }, { "epoch": 11.03, "learning_rate": 3.4423606444190414e-06, "loss": 0.021, "step": 10872 }, { "epoch": 11.03, "learning_rate": 3.4407057914354002e-06, "loss": 0.0198, "step": 10873 }, { "epoch": 11.03, "learning_rate": 3.439051253664568e-06, "loss": 0.0169, "step": 10874 }, { "epoch": 11.04, "learning_rate": 3.437397031186063e-06, "loss": 0.0188, "step": 10875 }, { "epoch": 11.04, "learning_rate": 3.4357431240793726e-06, "loss": 0.0175, "step": 10876 }, { "epoch": 11.04, "learning_rate": 3.434089532423981e-06, "loss": 0.0216, "step": 10877 }, { "epoch": 11.04, "learning_rate": 3.432436256299353e-06, "loss": 0.0214, "step": 10878 }, { "epoch": 11.04, "learning_rate": 3.4307832957849373e-06, "loss": 0.016, "step": 10879 }, { "epoch": 11.04, "learning_rate": 3.4291306509601728e-06, "loss": 0.0195, "step": 10880 }, { "epoch": 11.04, "learning_rate": 3.4274783219044718e-06, "loss": 0.0168, "step": 10881 }, { "epoch": 11.04, "learning_rate": 3.425826308697242e-06, "loss": 0.0179, "step": 10882 }, { "epoch": 11.04, "learning_rate": 3.4241746114178753e-06, "loss": 0.0151, "step": 10883 }, { "epoch": 11.04, "learning_rate": 3.422523230145739e-06, "loss": 0.0189, "step": 10884 }, { "epoch": 11.05, "learning_rate": 3.420872164960198e-06, "loss": 0.0159, "step": 10885 }, { "epoch": 11.05, "learning_rate": 3.4192214159405866e-06, "loss": 0.0236, "step": 10886 }, { "epoch": 11.05, "learning_rate": 3.4175709831662463e-06, "loss": 0.0195, "step": 10887 }, { "epoch": 11.05, "learning_rate": 3.41592086671648e-06, "loss": 0.0168, "step": 10888 }, { "epoch": 11.05, "learning_rate": 3.414271066670588e-06, "loss": 0.0197, "step": 10889 }, { "epoch": 11.05, "learning_rate": 3.4126215831078567e-06, "loss": 0.0139, "step": 10890 }, { "epoch": 11.05, "learning_rate": 3.4109724161075473e-06, "loss": 0.0179, "step": 10891 }, { "epoch": 11.05, "learning_rate": 3.409323565748919e-06, "loss": 0.0176, "step": 10892 }, { "epoch": 11.05, "learning_rate": 3.407675032111202e-06, "loss": 0.0216, "step": 10893 }, { "epoch": 11.05, "learning_rate": 3.4060268152736196e-06, "loss": 0.0222, "step": 10894 }, { "epoch": 11.06, "learning_rate": 3.4043789153153807e-06, "loss": 0.0175, "step": 10895 }, { "epoch": 11.06, "learning_rate": 3.4027313323156754e-06, "loss": 0.0204, "step": 10896 }, { "epoch": 11.06, "learning_rate": 3.4010840663536828e-06, "loss": 0.0221, "step": 10897 }, { "epoch": 11.06, "learning_rate": 3.3994371175085585e-06, "loss": 0.0144, "step": 10898 }, { "epoch": 11.06, "learning_rate": 3.39779048585945e-06, "loss": 0.0183, "step": 10899 }, { "epoch": 11.06, "learning_rate": 3.3961441714854924e-06, "loss": 0.02, "step": 10900 }, { "epoch": 11.06, "learning_rate": 3.3944981744657934e-06, "loss": 0.0162, "step": 10901 }, { "epoch": 11.06, "learning_rate": 3.392852494879455e-06, "loss": 0.0158, "step": 10902 }, { "epoch": 11.06, "learning_rate": 3.391207132805564e-06, "loss": 0.0223, "step": 10903 }, { "epoch": 11.06, "learning_rate": 3.3895620883231916e-06, "loss": 0.0229, "step": 10904 }, { "epoch": 11.07, "learning_rate": 3.387917361511387e-06, "loss": 0.0196, "step": 10905 }, { "epoch": 11.07, "learning_rate": 3.38627295244919e-06, "loss": 0.0184, "step": 10906 }, { "epoch": 11.07, "learning_rate": 3.3846288612156285e-06, "loss": 0.0197, "step": 10907 }, { "epoch": 11.07, "learning_rate": 3.382985087889705e-06, "loss": 0.0193, "step": 10908 }, { "epoch": 11.07, "learning_rate": 3.3813416325504176e-06, "loss": 0.0192, "step": 10909 }, { "epoch": 11.07, "learning_rate": 3.3796984952767385e-06, "loss": 0.0171, "step": 10910 }, { "epoch": 11.07, "learning_rate": 3.3780556761476303e-06, "loss": 0.0201, "step": 10911 }, { "epoch": 11.07, "learning_rate": 3.3764131752420494e-06, "loss": 0.0185, "step": 10912 }, { "epoch": 11.07, "learning_rate": 3.3747709926389183e-06, "loss": 0.0175, "step": 10913 }, { "epoch": 11.07, "learning_rate": 3.3731291284171584e-06, "loss": 0.0221, "step": 10914 }, { "epoch": 11.08, "learning_rate": 3.371487582655668e-06, "loss": 0.0196, "step": 10915 }, { "epoch": 11.08, "learning_rate": 3.3698463554333328e-06, "loss": 0.0205, "step": 10916 }, { "epoch": 11.08, "learning_rate": 3.3682054468290283e-06, "loss": 0.0199, "step": 10917 }, { "epoch": 11.08, "learning_rate": 3.3665648569216037e-06, "loss": 0.0191, "step": 10918 }, { "epoch": 11.08, "learning_rate": 3.3649245857899005e-06, "loss": 0.0221, "step": 10919 }, { "epoch": 11.08, "learning_rate": 3.3632846335127456e-06, "loss": 0.0153, "step": 10920 }, { "epoch": 11.08, "learning_rate": 3.3616450001689504e-06, "loss": 0.0243, "step": 10921 }, { "epoch": 11.08, "learning_rate": 3.3600056858373033e-06, "loss": 0.0264, "step": 10922 }, { "epoch": 11.08, "learning_rate": 3.358366690596585e-06, "loss": 0.0202, "step": 10923 }, { "epoch": 11.08, "learning_rate": 3.3567280145255634e-06, "loss": 0.0232, "step": 10924 }, { "epoch": 11.09, "learning_rate": 3.355089657702979e-06, "loss": 0.016, "step": 10925 }, { "epoch": 11.09, "learning_rate": 3.3534516202075684e-06, "loss": 0.0194, "step": 10926 }, { "epoch": 11.09, "learning_rate": 3.3518139021180498e-06, "loss": 0.0204, "step": 10927 }, { "epoch": 11.09, "learning_rate": 3.350176503513124e-06, "loss": 0.0202, "step": 10928 }, { "epoch": 11.09, "learning_rate": 3.3485394244714798e-06, "loss": 0.0208, "step": 10929 }, { "epoch": 11.09, "learning_rate": 3.346902665071785e-06, "loss": 0.0223, "step": 10930 }, { "epoch": 11.09, "learning_rate": 3.3452662253927003e-06, "loss": 0.0204, "step": 10931 }, { "epoch": 11.09, "learning_rate": 3.3436301055128597e-06, "loss": 0.0185, "step": 10932 }, { "epoch": 11.09, "learning_rate": 3.3419943055108917e-06, "loss": 0.0236, "step": 10933 }, { "epoch": 11.09, "learning_rate": 3.3403588254654097e-06, "loss": 0.0163, "step": 10934 }, { "epoch": 11.1, "learning_rate": 3.3387236654550016e-06, "loss": 0.0187, "step": 10935 }, { "epoch": 11.1, "learning_rate": 3.33708882555825e-06, "loss": 0.0205, "step": 10936 }, { "epoch": 11.1, "learning_rate": 3.335454305853718e-06, "loss": 0.0212, "step": 10937 }, { "epoch": 11.1, "learning_rate": 3.333820106419958e-06, "loss": 0.014, "step": 10938 }, { "epoch": 11.1, "learning_rate": 3.3321862273354955e-06, "loss": 0.0196, "step": 10939 }, { "epoch": 11.1, "learning_rate": 3.3305526686788526e-06, "loss": 0.0251, "step": 10940 }, { "epoch": 11.1, "learning_rate": 3.3289194305285333e-06, "loss": 0.0175, "step": 10941 }, { "epoch": 11.1, "learning_rate": 3.327286512963018e-06, "loss": 0.0231, "step": 10942 }, { "epoch": 11.1, "learning_rate": 3.3256539160607813e-06, "loss": 0.0163, "step": 10943 }, { "epoch": 11.11, "learning_rate": 3.3240216399002814e-06, "loss": 0.0172, "step": 10944 }, { "epoch": 11.11, "learning_rate": 3.3223896845599544e-06, "loss": 0.0245, "step": 10945 }, { "epoch": 11.11, "learning_rate": 3.3207580501182324e-06, "loss": 0.018, "step": 10946 }, { "epoch": 11.11, "learning_rate": 3.3191267366535174e-06, "loss": 0.0221, "step": 10947 }, { "epoch": 11.11, "learning_rate": 3.31749574424421e-06, "loss": 0.0204, "step": 10948 }, { "epoch": 11.11, "learning_rate": 3.3158650729686824e-06, "loss": 0.0187, "step": 10949 }, { "epoch": 11.11, "learning_rate": 3.314234722905302e-06, "loss": 0.0185, "step": 10950 }, { "epoch": 11.11, "learning_rate": 3.3126046941324185e-06, "loss": 0.0257, "step": 10951 }, { "epoch": 11.11, "learning_rate": 3.3109749867283592e-06, "loss": 0.0206, "step": 10952 }, { "epoch": 11.11, "learning_rate": 3.309345600771444e-06, "loss": 0.0223, "step": 10953 }, { "epoch": 11.12, "learning_rate": 3.3077165363399744e-06, "loss": 0.019, "step": 10954 }, { "epoch": 11.12, "learning_rate": 3.306087793512237e-06, "loss": 0.0198, "step": 10955 }, { "epoch": 11.12, "learning_rate": 3.304459372366505e-06, "loss": 0.0198, "step": 10956 }, { "epoch": 11.12, "learning_rate": 3.3028312729810275e-06, "loss": 0.0242, "step": 10957 }, { "epoch": 11.12, "learning_rate": 3.301203495434051e-06, "loss": 0.0188, "step": 10958 }, { "epoch": 11.12, "learning_rate": 3.2995760398037933e-06, "loss": 0.0176, "step": 10959 }, { "epoch": 11.12, "learning_rate": 3.2979489061684668e-06, "loss": 0.0222, "step": 10960 }, { "epoch": 11.12, "learning_rate": 3.2963220946062634e-06, "loss": 0.0175, "step": 10961 }, { "epoch": 11.12, "learning_rate": 3.294695605195364e-06, "loss": 0.0191, "step": 10962 }, { "epoch": 11.12, "learning_rate": 3.2930694380139305e-06, "loss": 0.0177, "step": 10963 }, { "epoch": 11.13, "learning_rate": 3.2914435931401056e-06, "loss": 0.0215, "step": 10964 }, { "epoch": 11.13, "learning_rate": 3.2898180706520277e-06, "loss": 0.0184, "step": 10965 }, { "epoch": 11.13, "learning_rate": 3.288192870627804e-06, "loss": 0.0197, "step": 10966 }, { "epoch": 11.13, "learning_rate": 3.286567993145541e-06, "loss": 0.0258, "step": 10967 }, { "epoch": 11.13, "learning_rate": 3.2849434382833244e-06, "loss": 0.0222, "step": 10968 }, { "epoch": 11.13, "learning_rate": 3.283319206119215e-06, "loss": 0.0221, "step": 10969 }, { "epoch": 11.13, "learning_rate": 3.281695296731281e-06, "loss": 0.0187, "step": 10970 }, { "epoch": 11.13, "learning_rate": 3.2800717101975486e-06, "loss": 0.0181, "step": 10971 }, { "epoch": 11.13, "learning_rate": 3.2784484465960452e-06, "loss": 0.0176, "step": 10972 }, { "epoch": 11.13, "learning_rate": 3.2768255060047816e-06, "loss": 0.0223, "step": 10973 }, { "epoch": 11.14, "learning_rate": 3.275202888501744e-06, "loss": 0.0239, "step": 10974 }, { "epoch": 11.14, "learning_rate": 3.2735805941649135e-06, "loss": 0.0208, "step": 10975 }, { "epoch": 11.14, "learning_rate": 3.2719586230722444e-06, "loss": 0.0202, "step": 10976 }, { "epoch": 11.14, "learning_rate": 3.2703369753016867e-06, "loss": 0.0232, "step": 10977 }, { "epoch": 11.14, "learning_rate": 3.26871565093117e-06, "loss": 0.0244, "step": 10978 }, { "epoch": 11.14, "learning_rate": 3.267094650038607e-06, "loss": 0.0184, "step": 10979 }, { "epoch": 11.14, "learning_rate": 3.2654739727019014e-06, "loss": 0.0186, "step": 10980 }, { "epoch": 11.14, "learning_rate": 3.263853618998929e-06, "loss": 0.0242, "step": 10981 }, { "epoch": 11.14, "learning_rate": 3.262233589007563e-06, "loss": 0.0225, "step": 10982 }, { "epoch": 11.14, "learning_rate": 3.260613882805651e-06, "loss": 0.0218, "step": 10983 }, { "epoch": 11.15, "learning_rate": 3.2589945004710312e-06, "loss": 0.021, "step": 10984 }, { "epoch": 11.15, "learning_rate": 3.2573754420815273e-06, "loss": 0.0174, "step": 10985 }, { "epoch": 11.15, "learning_rate": 3.2557567077149356e-06, "loss": 0.0212, "step": 10986 }, { "epoch": 11.15, "learning_rate": 3.25413829744906e-06, "loss": 0.0225, "step": 10987 }, { "epoch": 11.15, "learning_rate": 3.2525202113616627e-06, "loss": 0.0186, "step": 10988 }, { "epoch": 11.15, "learning_rate": 3.2509024495305064e-06, "loss": 0.0208, "step": 10989 }, { "epoch": 11.15, "learning_rate": 3.2492850120333373e-06, "loss": 0.0201, "step": 10990 }, { "epoch": 11.15, "learning_rate": 3.2476678989478782e-06, "loss": 0.0207, "step": 10991 }, { "epoch": 11.15, "learning_rate": 3.2460511103518444e-06, "loss": 0.0187, "step": 10992 }, { "epoch": 11.15, "learning_rate": 3.244434646322927e-06, "loss": 0.015, "step": 10993 }, { "epoch": 11.16, "learning_rate": 3.2428185069388108e-06, "loss": 0.0202, "step": 10994 }, { "epoch": 11.16, "learning_rate": 3.241202692277159e-06, "loss": 0.0194, "step": 10995 }, { "epoch": 11.16, "learning_rate": 3.239587202415623e-06, "loss": 0.0231, "step": 10996 }, { "epoch": 11.16, "learning_rate": 3.237972037431838e-06, "loss": 0.0184, "step": 10997 }, { "epoch": 11.16, "learning_rate": 3.236357197403417e-06, "loss": 0.0225, "step": 10998 }, { "epoch": 11.16, "learning_rate": 3.234742682407965e-06, "loss": 0.0208, "step": 10999 }, { "epoch": 11.16, "learning_rate": 3.2331284925230734e-06, "loss": 0.0194, "step": 11000 }, { "epoch": 11.16, "learning_rate": 3.2315146278263053e-06, "loss": 0.0233, "step": 11001 }, { "epoch": 11.16, "learning_rate": 3.229901088395224e-06, "loss": 0.0201, "step": 11002 }, { "epoch": 11.16, "learning_rate": 3.228287874307361e-06, "loss": 0.0197, "step": 11003 }, { "epoch": 11.17, "learning_rate": 3.226674985640251e-06, "loss": 0.02, "step": 11004 }, { "epoch": 11.17, "learning_rate": 3.2250624224713955e-06, "loss": 0.0179, "step": 11005 }, { "epoch": 11.17, "learning_rate": 3.223450184878291e-06, "loss": 0.0164, "step": 11006 }, { "epoch": 11.17, "learning_rate": 3.2218382729384157e-06, "loss": 0.0186, "step": 11007 }, { "epoch": 11.17, "learning_rate": 3.2202266867292277e-06, "loss": 0.0211, "step": 11008 }, { "epoch": 11.17, "learning_rate": 3.2186154263281777e-06, "loss": 0.0189, "step": 11009 }, { "epoch": 11.17, "learning_rate": 3.217004491812691e-06, "loss": 0.021, "step": 11010 }, { "epoch": 11.17, "learning_rate": 3.215393883260186e-06, "loss": 0.0208, "step": 11011 }, { "epoch": 11.17, "learning_rate": 3.2137836007480604e-06, "loss": 0.0218, "step": 11012 }, { "epoch": 11.18, "learning_rate": 3.2121736443537e-06, "loss": 0.0218, "step": 11013 }, { "epoch": 11.18, "learning_rate": 3.2105640141544746e-06, "loss": 0.02, "step": 11014 }, { "epoch": 11.18, "learning_rate": 3.20895471022773e-06, "loss": 0.0197, "step": 11015 }, { "epoch": 11.18, "learning_rate": 3.207345732650805e-06, "loss": 0.0224, "step": 11016 }, { "epoch": 11.18, "learning_rate": 3.2057370815010257e-06, "loss": 0.0185, "step": 11017 }, { "epoch": 11.18, "learning_rate": 3.20412875685569e-06, "loss": 0.0218, "step": 11018 }, { "epoch": 11.18, "learning_rate": 3.2025207587920905e-06, "loss": 0.0248, "step": 11019 }, { "epoch": 11.18, "learning_rate": 3.2009130873875018e-06, "loss": 0.0188, "step": 11020 }, { "epoch": 11.18, "learning_rate": 3.199305742719183e-06, "loss": 0.0253, "step": 11021 }, { "epoch": 11.18, "learning_rate": 3.1976987248643733e-06, "loss": 0.02, "step": 11022 }, { "epoch": 11.19, "learning_rate": 3.1960920339003e-06, "loss": 0.0237, "step": 11023 }, { "epoch": 11.19, "learning_rate": 3.1944856699041773e-06, "loss": 0.0169, "step": 11024 }, { "epoch": 11.19, "learning_rate": 3.1928796329531964e-06, "loss": 0.0173, "step": 11025 }, { "epoch": 11.19, "learning_rate": 3.191273923124537e-06, "loss": 0.0196, "step": 11026 }, { "epoch": 11.19, "learning_rate": 3.1896685404953685e-06, "loss": 0.0197, "step": 11027 }, { "epoch": 11.19, "learning_rate": 3.1880634851428293e-06, "loss": 0.0191, "step": 11028 }, { "epoch": 11.19, "learning_rate": 3.186458757144063e-06, "loss": 0.0209, "step": 11029 }, { "epoch": 11.19, "learning_rate": 3.184854356576178e-06, "loss": 0.0196, "step": 11030 }, { "epoch": 11.19, "learning_rate": 3.1832502835162817e-06, "loss": 0.019, "step": 11031 }, { "epoch": 11.19, "learning_rate": 3.181646538041452e-06, "loss": 0.0217, "step": 11032 }, { "epoch": 11.2, "learning_rate": 3.1800431202287618e-06, "loss": 0.0219, "step": 11033 }, { "epoch": 11.2, "learning_rate": 3.1784400301552676e-06, "loss": 0.0194, "step": 11034 }, { "epoch": 11.2, "learning_rate": 3.176837267898002e-06, "loss": 0.021, "step": 11035 }, { "epoch": 11.2, "learning_rate": 3.17523483353399e-06, "loss": 0.0209, "step": 11036 }, { "epoch": 11.2, "learning_rate": 3.1736327271402367e-06, "loss": 0.0204, "step": 11037 }, { "epoch": 11.2, "learning_rate": 3.1720309487937384e-06, "loss": 0.018, "step": 11038 }, { "epoch": 11.2, "learning_rate": 3.1704294985714625e-06, "loss": 0.021, "step": 11039 }, { "epoch": 11.2, "learning_rate": 3.1688283765503703e-06, "loss": 0.0227, "step": 11040 }, { "epoch": 11.2, "learning_rate": 3.1672275828074094e-06, "loss": 0.019, "step": 11041 }, { "epoch": 11.2, "learning_rate": 3.1656271174195018e-06, "loss": 0.0214, "step": 11042 }, { "epoch": 11.21, "learning_rate": 3.1640269804635605e-06, "loss": 0.0186, "step": 11043 }, { "epoch": 11.21, "learning_rate": 3.1624271720164836e-06, "loss": 0.0205, "step": 11044 }, { "epoch": 11.21, "learning_rate": 3.1608276921551497e-06, "loss": 0.0146, "step": 11045 }, { "epoch": 11.21, "learning_rate": 3.159228540956427e-06, "loss": 0.019, "step": 11046 }, { "epoch": 11.21, "learning_rate": 3.157629718497157e-06, "loss": 0.0248, "step": 11047 }, { "epoch": 11.21, "learning_rate": 3.156031224854181e-06, "loss": 0.0201, "step": 11048 }, { "epoch": 11.21, "learning_rate": 3.154433060104308e-06, "loss": 0.0212, "step": 11049 }, { "epoch": 11.21, "learning_rate": 3.1528352243243433e-06, "loss": 0.0219, "step": 11050 }, { "epoch": 11.21, "learning_rate": 3.1512377175910748e-06, "loss": 0.0206, "step": 11051 }, { "epoch": 11.21, "learning_rate": 3.149640539981267e-06, "loss": 0.0221, "step": 11052 }, { "epoch": 11.22, "learning_rate": 3.148043691571675e-06, "loss": 0.0206, "step": 11053 }, { "epoch": 11.22, "learning_rate": 3.1464471724390376e-06, "loss": 0.0201, "step": 11054 }, { "epoch": 11.22, "learning_rate": 3.1448509826600805e-06, "loss": 0.026, "step": 11055 }, { "epoch": 11.22, "learning_rate": 3.1432551223115037e-06, "loss": 0.0187, "step": 11056 }, { "epoch": 11.22, "learning_rate": 3.141659591470001e-06, "loss": 0.019, "step": 11057 }, { "epoch": 11.22, "learning_rate": 3.1400643902122495e-06, "loss": 0.0248, "step": 11058 }, { "epoch": 11.22, "learning_rate": 3.1384695186149026e-06, "loss": 0.0225, "step": 11059 }, { "epoch": 11.22, "learning_rate": 3.136874976754606e-06, "loss": 0.0229, "step": 11060 }, { "epoch": 11.22, "learning_rate": 3.1352807647079865e-06, "loss": 0.0228, "step": 11061 }, { "epoch": 11.22, "learning_rate": 3.1336868825516554e-06, "loss": 0.0224, "step": 11062 }, { "epoch": 11.23, "learning_rate": 3.1320933303622127e-06, "loss": 0.023, "step": 11063 }, { "epoch": 11.23, "learning_rate": 3.13050010821623e-06, "loss": 0.0176, "step": 11064 }, { "epoch": 11.23, "learning_rate": 3.1289072161902776e-06, "loss": 0.0215, "step": 11065 }, { "epoch": 11.23, "learning_rate": 3.127314654360898e-06, "loss": 0.0181, "step": 11066 }, { "epoch": 11.23, "learning_rate": 3.1257224228046256e-06, "loss": 0.0199, "step": 11067 }, { "epoch": 11.23, "learning_rate": 3.124130521597981e-06, "loss": 0.0216, "step": 11068 }, { "epoch": 11.23, "learning_rate": 3.122538950817452e-06, "loss": 0.0201, "step": 11069 }, { "epoch": 11.23, "learning_rate": 3.1209477105395393e-06, "loss": 0.021, "step": 11070 }, { "epoch": 11.23, "learning_rate": 3.119356800840698e-06, "loss": 0.0196, "step": 11071 }, { "epoch": 11.23, "learning_rate": 3.117766221797388e-06, "loss": 0.0197, "step": 11072 }, { "epoch": 11.24, "learning_rate": 3.1161759734860465e-06, "loss": 0.02, "step": 11073 }, { "epoch": 11.24, "learning_rate": 3.1145860559830877e-06, "loss": 0.0207, "step": 11074 }, { "epoch": 11.24, "learning_rate": 3.112996469364925e-06, "loss": 0.0183, "step": 11075 }, { "epoch": 11.24, "learning_rate": 3.111407213707939e-06, "loss": 0.0243, "step": 11076 }, { "epoch": 11.24, "learning_rate": 3.1098182890885078e-06, "loss": 0.0201, "step": 11077 }, { "epoch": 11.24, "learning_rate": 3.1082296955829874e-06, "loss": 0.0212, "step": 11078 }, { "epoch": 11.24, "learning_rate": 3.1066414332677186e-06, "loss": 0.0256, "step": 11079 }, { "epoch": 11.24, "learning_rate": 3.105053502219031e-06, "loss": 0.0221, "step": 11080 }, { "epoch": 11.24, "learning_rate": 3.1034659025132276e-06, "loss": 0.0213, "step": 11081 }, { "epoch": 11.25, "learning_rate": 3.101878634226608e-06, "loss": 0.0189, "step": 11082 }, { "epoch": 11.25, "learning_rate": 3.1002916974354436e-06, "loss": 0.0233, "step": 11083 }, { "epoch": 11.25, "learning_rate": 3.098705092215999e-06, "loss": 0.0229, "step": 11084 }, { "epoch": 11.25, "learning_rate": 3.0971188186445233e-06, "loss": 0.0216, "step": 11085 }, { "epoch": 11.25, "learning_rate": 3.095532876797237e-06, "loss": 0.0209, "step": 11086 }, { "epoch": 11.25, "learning_rate": 3.093947266750367e-06, "loss": 0.0186, "step": 11087 }, { "epoch": 11.25, "learning_rate": 3.0923619885801015e-06, "loss": 0.0184, "step": 11088 }, { "epoch": 11.25, "learning_rate": 3.0907770423626238e-06, "loss": 0.0245, "step": 11089 }, { "epoch": 11.25, "learning_rate": 3.089192428174106e-06, "loss": 0.0233, "step": 11090 }, { "epoch": 11.25, "learning_rate": 3.0876081460906905e-06, "loss": 0.0236, "step": 11091 }, { "epoch": 11.26, "learning_rate": 3.086024196188517e-06, "loss": 0.0202, "step": 11092 }, { "epoch": 11.26, "learning_rate": 3.0844405785436983e-06, "loss": 0.0265, "step": 11093 }, { "epoch": 11.26, "learning_rate": 3.0828572932323396e-06, "loss": 0.0197, "step": 11094 }, { "epoch": 11.26, "learning_rate": 3.081274340330527e-06, "loss": 0.0214, "step": 11095 }, { "epoch": 11.26, "learning_rate": 3.0796917199143306e-06, "loss": 0.0247, "step": 11096 }, { "epoch": 11.26, "learning_rate": 3.078109432059807e-06, "loss": 0.021, "step": 11097 }, { "epoch": 11.26, "learning_rate": 3.0765274768429907e-06, "loss": 0.0226, "step": 11098 }, { "epoch": 11.26, "learning_rate": 3.0749458543399047e-06, "loss": 0.023, "step": 11099 }, { "epoch": 11.26, "learning_rate": 3.07336456462656e-06, "loss": 0.0238, "step": 11100 }, { "epoch": 11.26, "learning_rate": 3.0717836077789386e-06, "loss": 0.0206, "step": 11101 }, { "epoch": 11.27, "learning_rate": 3.070202983873023e-06, "loss": 0.0223, "step": 11102 }, { "epoch": 11.27, "learning_rate": 3.0686226929847617e-06, "loss": 0.0226, "step": 11103 }, { "epoch": 11.27, "learning_rate": 3.067042735190109e-06, "loss": 0.0151, "step": 11104 }, { "epoch": 11.27, "learning_rate": 3.0654631105649823e-06, "loss": 0.0205, "step": 11105 }, { "epoch": 11.27, "learning_rate": 3.0638838191852947e-06, "loss": 0.0222, "step": 11106 }, { "epoch": 11.27, "learning_rate": 3.0623048611269445e-06, "loss": 0.0169, "step": 11107 }, { "epoch": 11.27, "learning_rate": 3.060726236465803e-06, "loss": 0.0221, "step": 11108 }, { "epoch": 11.27, "learning_rate": 3.0591479452777385e-06, "loss": 0.023, "step": 11109 }, { "epoch": 11.27, "learning_rate": 3.05756998763859e-06, "loss": 0.02, "step": 11110 }, { "epoch": 11.27, "learning_rate": 3.055992363624193e-06, "loss": 0.0228, "step": 11111 }, { "epoch": 11.28, "learning_rate": 3.0544150733103595e-06, "loss": 0.0204, "step": 11112 }, { "epoch": 11.28, "learning_rate": 3.052838116772889e-06, "loss": 0.0234, "step": 11113 }, { "epoch": 11.28, "learning_rate": 3.0512614940875673e-06, "loss": 0.0183, "step": 11114 }, { "epoch": 11.28, "learning_rate": 3.049685205330153e-06, "loss": 0.0177, "step": 11115 }, { "epoch": 11.28, "learning_rate": 3.0481092505763987e-06, "loss": 0.0189, "step": 11116 }, { "epoch": 11.28, "learning_rate": 3.046533629902043e-06, "loss": 0.0214, "step": 11117 }, { "epoch": 11.28, "learning_rate": 3.044958343382797e-06, "loss": 0.0188, "step": 11118 }, { "epoch": 11.28, "learning_rate": 3.043383391094368e-06, "loss": 0.0199, "step": 11119 }, { "epoch": 11.28, "learning_rate": 3.0418087731124324e-06, "loss": 0.02, "step": 11120 }, { "epoch": 11.28, "learning_rate": 3.0402344895126732e-06, "loss": 0.0184, "step": 11121 }, { "epoch": 11.29, "learning_rate": 3.0386605403707347e-06, "loss": 0.023, "step": 11122 }, { "epoch": 11.29, "learning_rate": 3.0370869257622572e-06, "loss": 0.0231, "step": 11123 }, { "epoch": 11.29, "learning_rate": 3.0355136457628644e-06, "loss": 0.018, "step": 11124 }, { "epoch": 11.29, "learning_rate": 3.0339407004481568e-06, "loss": 0.0188, "step": 11125 }, { "epoch": 11.29, "learning_rate": 3.03236808989373e-06, "loss": 0.0199, "step": 11126 }, { "epoch": 11.29, "learning_rate": 3.0307958141751504e-06, "loss": 0.0161, "step": 11127 }, { "epoch": 11.29, "learning_rate": 3.0292238733679747e-06, "loss": 0.0199, "step": 11128 }, { "epoch": 11.29, "learning_rate": 3.0276522675477538e-06, "loss": 0.02, "step": 11129 }, { "epoch": 11.29, "learning_rate": 3.026080996790004e-06, "loss": 0.0151, "step": 11130 }, { "epoch": 11.29, "learning_rate": 3.024510061170239e-06, "loss": 0.0212, "step": 11131 }, { "epoch": 11.3, "learning_rate": 3.022939460763946e-06, "loss": 0.0219, "step": 11132 }, { "epoch": 11.3, "learning_rate": 3.021369195646605e-06, "loss": 0.0212, "step": 11133 }, { "epoch": 11.3, "learning_rate": 3.019799265893679e-06, "loss": 0.0167, "step": 11134 }, { "epoch": 11.3, "learning_rate": 3.018229671580607e-06, "loss": 0.0177, "step": 11135 }, { "epoch": 11.3, "learning_rate": 3.016660412782819e-06, "loss": 0.0184, "step": 11136 }, { "epoch": 11.3, "learning_rate": 3.015091489575729e-06, "loss": 0.0186, "step": 11137 }, { "epoch": 11.3, "learning_rate": 3.0135229020347346e-06, "loss": 0.0209, "step": 11138 }, { "epoch": 11.3, "learning_rate": 3.01195465023521e-06, "loss": 0.0213, "step": 11139 }, { "epoch": 11.3, "learning_rate": 3.0103867342525217e-06, "loss": 0.02, "step": 11140 }, { "epoch": 11.3, "learning_rate": 3.0088191541620214e-06, "loss": 0.0225, "step": 11141 }, { "epoch": 11.31, "learning_rate": 3.007251910039034e-06, "loss": 0.0194, "step": 11142 }, { "epoch": 11.31, "learning_rate": 3.0056850019588777e-06, "loss": 0.0246, "step": 11143 }, { "epoch": 11.31, "learning_rate": 3.004118429996854e-06, "loss": 0.0194, "step": 11144 }, { "epoch": 11.31, "learning_rate": 3.002552194228239e-06, "loss": 0.0229, "step": 11145 }, { "epoch": 11.31, "learning_rate": 3.0009862947283088e-06, "loss": 0.019, "step": 11146 }, { "epoch": 11.31, "learning_rate": 2.9994207315723067e-06, "loss": 0.0192, "step": 11147 }, { "epoch": 11.31, "learning_rate": 2.997855504835474e-06, "loss": 0.0201, "step": 11148 }, { "epoch": 11.31, "learning_rate": 2.9962906145930214e-06, "loss": 0.0193, "step": 11149 }, { "epoch": 11.31, "learning_rate": 2.9947260609201556e-06, "loss": 0.0185, "step": 11150 }, { "epoch": 11.32, "learning_rate": 2.993161843892064e-06, "loss": 0.0203, "step": 11151 }, { "epoch": 11.32, "learning_rate": 2.991597963583912e-06, "loss": 0.0218, "step": 11152 }, { "epoch": 11.32, "learning_rate": 2.9900344200708552e-06, "loss": 0.0216, "step": 11153 }, { "epoch": 11.32, "learning_rate": 2.9884712134280324e-06, "loss": 0.0243, "step": 11154 }, { "epoch": 11.32, "learning_rate": 2.986908343730567e-06, "loss": 0.0187, "step": 11155 }, { "epoch": 11.32, "learning_rate": 2.9853458110535573e-06, "loss": 0.0229, "step": 11156 }, { "epoch": 11.32, "learning_rate": 2.983783615472097e-06, "loss": 0.0199, "step": 11157 }, { "epoch": 11.32, "learning_rate": 2.98222175706126e-06, "loss": 0.0215, "step": 11158 }, { "epoch": 11.32, "learning_rate": 2.9806602358960977e-06, "loss": 0.0264, "step": 11159 }, { "epoch": 11.32, "learning_rate": 2.979099052051655e-06, "loss": 0.0211, "step": 11160 }, { "epoch": 11.33, "learning_rate": 2.9775382056029535e-06, "loss": 0.02, "step": 11161 }, { "epoch": 11.33, "learning_rate": 2.975977696625001e-06, "loss": 0.0198, "step": 11162 }, { "epoch": 11.33, "learning_rate": 2.9744175251927942e-06, "loss": 0.021, "step": 11163 }, { "epoch": 11.33, "learning_rate": 2.9728576913813024e-06, "loss": 0.0211, "step": 11164 }, { "epoch": 11.33, "learning_rate": 2.9712981952654895e-06, "loss": 0.0251, "step": 11165 }, { "epoch": 11.33, "learning_rate": 2.9697390369202927e-06, "loss": 0.021, "step": 11166 }, { "epoch": 11.33, "learning_rate": 2.968180216420642e-06, "loss": 0.0199, "step": 11167 }, { "epoch": 11.33, "learning_rate": 2.9666217338414506e-06, "loss": 0.0155, "step": 11168 }, { "epoch": 11.33, "learning_rate": 2.9650635892576074e-06, "loss": 0.0179, "step": 11169 }, { "epoch": 11.33, "learning_rate": 2.963505782743994e-06, "loss": 0.019, "step": 11170 }, { "epoch": 11.34, "learning_rate": 2.9619483143754703e-06, "loss": 0.0184, "step": 11171 }, { "epoch": 11.34, "learning_rate": 2.960391184226884e-06, "loss": 0.024, "step": 11172 }, { "epoch": 11.34, "learning_rate": 2.9588343923730657e-06, "loss": 0.0196, "step": 11173 }, { "epoch": 11.34, "learning_rate": 2.957277938888824e-06, "loss": 0.0197, "step": 11174 }, { "epoch": 11.34, "learning_rate": 2.9557218238489595e-06, "loss": 0.0213, "step": 11175 }, { "epoch": 11.34, "learning_rate": 2.9541660473282483e-06, "loss": 0.0226, "step": 11176 }, { "epoch": 11.34, "learning_rate": 2.952610609401458e-06, "loss": 0.0228, "step": 11177 }, { "epoch": 11.34, "learning_rate": 2.951055510143336e-06, "loss": 0.0189, "step": 11178 }, { "epoch": 11.34, "learning_rate": 2.949500749628613e-06, "loss": 0.0242, "step": 11179 }, { "epoch": 11.34, "learning_rate": 2.9479463279320088e-06, "loss": 0.0214, "step": 11180 }, { "epoch": 11.35, "learning_rate": 2.9463922451282166e-06, "loss": 0.0222, "step": 11181 }, { "epoch": 11.35, "learning_rate": 2.9448385012919244e-06, "loss": 0.027, "step": 11182 }, { "epoch": 11.35, "learning_rate": 2.943285096497793e-06, "loss": 0.0202, "step": 11183 }, { "epoch": 11.35, "learning_rate": 2.9417320308204756e-06, "loss": 0.0195, "step": 11184 }, { "epoch": 11.35, "learning_rate": 2.9401793043346095e-06, "loss": 0.023, "step": 11185 }, { "epoch": 11.35, "learning_rate": 2.938626917114803e-06, "loss": 0.0166, "step": 11186 }, { "epoch": 11.35, "learning_rate": 2.93707486923567e-06, "loss": 0.0239, "step": 11187 }, { "epoch": 11.35, "learning_rate": 2.935523160771786e-06, "loss": 0.0221, "step": 11188 }, { "epoch": 11.35, "learning_rate": 2.933971791797724e-06, "loss": 0.021, "step": 11189 }, { "epoch": 11.35, "learning_rate": 2.9324207623880374e-06, "loss": 0.0219, "step": 11190 }, { "epoch": 11.36, "learning_rate": 2.930870072617258e-06, "loss": 0.0198, "step": 11191 }, { "epoch": 11.36, "learning_rate": 2.92931972255991e-06, "loss": 0.02, "step": 11192 }, { "epoch": 11.36, "learning_rate": 2.9277697122904925e-06, "loss": 0.0177, "step": 11193 }, { "epoch": 11.36, "learning_rate": 2.9262200418834954e-06, "loss": 0.0213, "step": 11194 }, { "epoch": 11.36, "learning_rate": 2.924670711413389e-06, "loss": 0.0156, "step": 11195 }, { "epoch": 11.36, "learning_rate": 2.923121720954627e-06, "loss": 0.0204, "step": 11196 }, { "epoch": 11.36, "learning_rate": 2.9215730705816525e-06, "loss": 0.0219, "step": 11197 }, { "epoch": 11.36, "learning_rate": 2.920024760368879e-06, "loss": 0.0211, "step": 11198 }, { "epoch": 11.36, "learning_rate": 2.9184767903907197e-06, "loss": 0.0225, "step": 11199 }, { "epoch": 11.36, "learning_rate": 2.9169291607215576e-06, "loss": 0.017, "step": 11200 }, { "epoch": 11.37, "learning_rate": 2.915381871435767e-06, "loss": 0.0242, "step": 11201 }, { "epoch": 11.37, "learning_rate": 2.91383492260771e-06, "loss": 0.0228, "step": 11202 }, { "epoch": 11.37, "learning_rate": 2.9122883143117143e-06, "loss": 0.0183, "step": 11203 }, { "epoch": 11.37, "learning_rate": 2.910742046622118e-06, "loss": 0.0203, "step": 11204 }, { "epoch": 11.37, "learning_rate": 2.909196119613218e-06, "loss": 0.0207, "step": 11205 }, { "epoch": 11.37, "learning_rate": 2.90765053335931e-06, "loss": 0.0276, "step": 11206 }, { "epoch": 11.37, "learning_rate": 2.9061052879346685e-06, "loss": 0.0201, "step": 11207 }, { "epoch": 11.37, "learning_rate": 2.9045603834135482e-06, "loss": 0.0176, "step": 11208 }, { "epoch": 11.37, "learning_rate": 2.9030158198701975e-06, "loss": 0.0199, "step": 11209 }, { "epoch": 11.37, "learning_rate": 2.9014715973788332e-06, "loss": 0.0228, "step": 11210 }, { "epoch": 11.38, "learning_rate": 2.8999277160136687e-06, "loss": 0.0181, "step": 11211 }, { "epoch": 11.38, "learning_rate": 2.8983841758488964e-06, "loss": 0.0189, "step": 11212 }, { "epoch": 11.38, "learning_rate": 2.8968409769586926e-06, "loss": 0.018, "step": 11213 }, { "epoch": 11.38, "learning_rate": 2.89529811941722e-06, "loss": 0.0198, "step": 11214 }, { "epoch": 11.38, "learning_rate": 2.893755603298617e-06, "loss": 0.0167, "step": 11215 }, { "epoch": 11.38, "learning_rate": 2.892213428677012e-06, "loss": 0.0201, "step": 11216 }, { "epoch": 11.38, "learning_rate": 2.8906715956265195e-06, "loss": 0.0244, "step": 11217 }, { "epoch": 11.38, "learning_rate": 2.889130104221227e-06, "loss": 0.0251, "step": 11218 }, { "epoch": 11.38, "learning_rate": 2.88758895453522e-06, "loss": 0.0201, "step": 11219 }, { "epoch": 11.39, "learning_rate": 2.886048146642548e-06, "loss": 0.0224, "step": 11220 }, { "epoch": 11.39, "learning_rate": 2.884507680617271e-06, "loss": 0.0219, "step": 11221 }, { "epoch": 11.39, "learning_rate": 2.882967556533407e-06, "loss": 0.0211, "step": 11222 }, { "epoch": 11.39, "learning_rate": 2.88142777446497e-06, "loss": 0.0197, "step": 11223 }, { "epoch": 11.39, "learning_rate": 2.8798883344859606e-06, "loss": 0.0246, "step": 11224 }, { "epoch": 11.39, "learning_rate": 2.8783492366703515e-06, "loss": 0.0201, "step": 11225 }, { "epoch": 11.39, "learning_rate": 2.8768104810921093e-06, "loss": 0.0217, "step": 11226 }, { "epoch": 11.39, "learning_rate": 2.8752720678251776e-06, "loss": 0.0185, "step": 11227 }, { "epoch": 11.39, "learning_rate": 2.8737339969434873e-06, "loss": 0.0222, "step": 11228 }, { "epoch": 11.39, "learning_rate": 2.8721962685209513e-06, "loss": 0.0223, "step": 11229 }, { "epoch": 11.4, "learning_rate": 2.8706588826314686e-06, "loss": 0.0193, "step": 11230 }, { "epoch": 11.4, "learning_rate": 2.8691218393489207e-06, "loss": 0.0202, "step": 11231 }, { "epoch": 11.4, "learning_rate": 2.8675851387471664e-06, "loss": 0.0253, "step": 11232 }, { "epoch": 11.4, "learning_rate": 2.866048780900057e-06, "loss": 0.0244, "step": 11233 }, { "epoch": 11.4, "learning_rate": 2.864512765881424e-06, "loss": 0.0234, "step": 11234 }, { "epoch": 11.4, "learning_rate": 2.8629770937650793e-06, "loss": 0.0236, "step": 11235 }, { "epoch": 11.4, "learning_rate": 2.861441764624825e-06, "loss": 0.0175, "step": 11236 }, { "epoch": 11.4, "learning_rate": 2.8599067785344335e-06, "loss": 0.0201, "step": 11237 }, { "epoch": 11.4, "learning_rate": 2.8583721355676843e-06, "loss": 0.0215, "step": 11238 }, { "epoch": 11.4, "learning_rate": 2.8568378357983138e-06, "loss": 0.0322, "step": 11239 }, { "epoch": 11.41, "learning_rate": 2.85530387930006e-06, "loss": 0.0236, "step": 11240 }, { "epoch": 11.41, "learning_rate": 2.8537702661466392e-06, "loss": 0.0209, "step": 11241 }, { "epoch": 11.41, "learning_rate": 2.8522369964117456e-06, "loss": 0.0205, "step": 11242 }, { "epoch": 11.41, "learning_rate": 2.850704070169066e-06, "loss": 0.018, "step": 11243 }, { "epoch": 11.41, "learning_rate": 2.849171487492267e-06, "loss": 0.024, "step": 11244 }, { "epoch": 11.41, "learning_rate": 2.847639248454992e-06, "loss": 0.0252, "step": 11245 }, { "epoch": 11.41, "learning_rate": 2.846107353130885e-06, "loss": 0.0216, "step": 11246 }, { "epoch": 11.41, "learning_rate": 2.8445758015935533e-06, "loss": 0.0255, "step": 11247 }, { "epoch": 11.41, "learning_rate": 2.843044593916602e-06, "loss": 0.0224, "step": 11248 }, { "epoch": 11.41, "learning_rate": 2.8415137301736107e-06, "loss": 0.0238, "step": 11249 }, { "epoch": 11.42, "learning_rate": 2.8399832104381488e-06, "loss": 0.0176, "step": 11250 }, { "epoch": 11.42, "learning_rate": 2.838453034783769e-06, "loss": 0.0189, "step": 11251 }, { "epoch": 11.42, "learning_rate": 2.836923203284001e-06, "loss": 0.0222, "step": 11252 }, { "epoch": 11.42, "learning_rate": 2.8353937160123634e-06, "loss": 0.031, "step": 11253 }, { "epoch": 11.42, "learning_rate": 2.8338645730423574e-06, "loss": 0.022, "step": 11254 }, { "epoch": 11.42, "learning_rate": 2.832335774447471e-06, "loss": 0.0154, "step": 11255 }, { "epoch": 11.42, "learning_rate": 2.8308073203011667e-06, "loss": 0.0237, "step": 11256 }, { "epoch": 11.42, "learning_rate": 2.829279210676896e-06, "loss": 0.0204, "step": 11257 }, { "epoch": 11.42, "learning_rate": 2.8277514456481003e-06, "loss": 0.0218, "step": 11258 }, { "epoch": 11.42, "learning_rate": 2.826224025288189e-06, "loss": 0.0277, "step": 11259 }, { "epoch": 11.43, "learning_rate": 2.824696949670567e-06, "loss": 0.0242, "step": 11260 }, { "epoch": 11.43, "learning_rate": 2.8231702188686226e-06, "loss": 0.0193, "step": 11261 }, { "epoch": 11.43, "learning_rate": 2.821643832955716e-06, "loss": 0.0172, "step": 11262 }, { "epoch": 11.43, "learning_rate": 2.8201177920052114e-06, "loss": 0.0243, "step": 11263 }, { "epoch": 11.43, "learning_rate": 2.8185920960904323e-06, "loss": 0.021, "step": 11264 }, { "epoch": 11.43, "learning_rate": 2.817066745284707e-06, "loss": 0.0244, "step": 11265 }, { "epoch": 11.43, "learning_rate": 2.8155417396613285e-06, "loss": 0.0254, "step": 11266 }, { "epoch": 11.43, "learning_rate": 2.8140170792935874e-06, "loss": 0.0213, "step": 11267 }, { "epoch": 11.43, "learning_rate": 2.812492764254754e-06, "loss": 0.0252, "step": 11268 }, { "epoch": 11.43, "learning_rate": 2.8109687946180765e-06, "loss": 0.02, "step": 11269 }, { "epoch": 11.44, "learning_rate": 2.8094451704567925e-06, "loss": 0.0237, "step": 11270 }, { "epoch": 11.44, "learning_rate": 2.807921891844121e-06, "loss": 0.0242, "step": 11271 }, { "epoch": 11.44, "learning_rate": 2.8063989588532647e-06, "loss": 0.0203, "step": 11272 }, { "epoch": 11.44, "learning_rate": 2.8048763715574133e-06, "loss": 0.0243, "step": 11273 }, { "epoch": 11.44, "learning_rate": 2.8033541300297296e-06, "loss": 0.0323, "step": 11274 }, { "epoch": 11.44, "learning_rate": 2.8018322343433713e-06, "loss": 0.0172, "step": 11275 }, { "epoch": 11.44, "learning_rate": 2.80031068457147e-06, "loss": 0.0214, "step": 11276 }, { "epoch": 11.44, "learning_rate": 2.7987894807871475e-06, "loss": 0.0189, "step": 11277 }, { "epoch": 11.44, "learning_rate": 2.797268623063506e-06, "loss": 0.0241, "step": 11278 }, { "epoch": 11.44, "learning_rate": 2.7957481114736328e-06, "loss": 0.018, "step": 11279 }, { "epoch": 11.45, "learning_rate": 2.7942279460905997e-06, "loss": 0.0201, "step": 11280 }, { "epoch": 11.45, "learning_rate": 2.7927081269874545e-06, "loss": 0.0188, "step": 11281 }, { "epoch": 11.45, "learning_rate": 2.791188654237239e-06, "loss": 0.0222, "step": 11282 }, { "epoch": 11.45, "learning_rate": 2.789669527912966e-06, "loss": 0.0223, "step": 11283 }, { "epoch": 11.45, "learning_rate": 2.7881507480876424e-06, "loss": 0.0211, "step": 11284 }, { "epoch": 11.45, "learning_rate": 2.7866323148342566e-06, "loss": 0.018, "step": 11285 }, { "epoch": 11.45, "learning_rate": 2.7851142282257747e-06, "loss": 0.0181, "step": 11286 }, { "epoch": 11.45, "learning_rate": 2.7835964883351496e-06, "loss": 0.0206, "step": 11287 }, { "epoch": 11.45, "learning_rate": 2.7820790952353194e-06, "loss": 0.0202, "step": 11288 }, { "epoch": 11.46, "learning_rate": 2.7805620489992038e-06, "loss": 0.02, "step": 11289 }, { "epoch": 11.46, "learning_rate": 2.779045349699708e-06, "loss": 0.0251, "step": 11290 }, { "epoch": 11.46, "learning_rate": 2.777528997409714e-06, "loss": 0.0227, "step": 11291 }, { "epoch": 11.46, "learning_rate": 2.776012992202096e-06, "loss": 0.0229, "step": 11292 }, { "epoch": 11.46, "learning_rate": 2.7744973341497005e-06, "loss": 0.0211, "step": 11293 }, { "epoch": 11.46, "learning_rate": 2.772982023325368e-06, "loss": 0.0225, "step": 11294 }, { "epoch": 11.46, "learning_rate": 2.771467059801919e-06, "loss": 0.0234, "step": 11295 }, { "epoch": 11.46, "learning_rate": 2.769952443652154e-06, "loss": 0.0214, "step": 11296 }, { "epoch": 11.46, "learning_rate": 2.768438174948863e-06, "loss": 0.0219, "step": 11297 }, { "epoch": 11.46, "learning_rate": 2.766924253764811e-06, "loss": 0.0238, "step": 11298 }, { "epoch": 11.47, "learning_rate": 2.7654106801727566e-06, "loss": 0.0186, "step": 11299 }, { "epoch": 11.47, "learning_rate": 2.7638974542454276e-06, "loss": 0.0234, "step": 11300 }, { "epoch": 11.47, "learning_rate": 2.76238457605555e-06, "loss": 0.0211, "step": 11301 }, { "epoch": 11.47, "learning_rate": 2.7608720456758267e-06, "loss": 0.0169, "step": 11302 }, { "epoch": 11.47, "learning_rate": 2.7593598631789353e-06, "loss": 0.0203, "step": 11303 }, { "epoch": 11.47, "learning_rate": 2.7578480286375577e-06, "loss": 0.024, "step": 11304 }, { "epoch": 11.47, "learning_rate": 2.756336542124337e-06, "loss": 0.0215, "step": 11305 }, { "epoch": 11.47, "learning_rate": 2.7548254037119128e-06, "loss": 0.0197, "step": 11306 }, { "epoch": 11.47, "learning_rate": 2.753314613472906e-06, "loss": 0.0211, "step": 11307 }, { "epoch": 11.47, "learning_rate": 2.7518041714799137e-06, "loss": 0.0168, "step": 11308 }, { "epoch": 11.48, "learning_rate": 2.7502940778055277e-06, "loss": 0.0164, "step": 11309 }, { "epoch": 11.48, "learning_rate": 2.7487843325223116e-06, "loss": 0.0192, "step": 11310 }, { "epoch": 11.48, "learning_rate": 2.747274935702818e-06, "loss": 0.0217, "step": 11311 }, { "epoch": 11.48, "learning_rate": 2.745765887419586e-06, "loss": 0.0173, "step": 11312 }, { "epoch": 11.48, "learning_rate": 2.7442571877451306e-06, "loss": 0.0219, "step": 11313 }, { "epoch": 11.48, "learning_rate": 2.7427488367519596e-06, "loss": 0.0209, "step": 11314 }, { "epoch": 11.48, "learning_rate": 2.741240834512551e-06, "loss": 0.0228, "step": 11315 }, { "epoch": 11.48, "learning_rate": 2.7397331810993754e-06, "loss": 0.0188, "step": 11316 }, { "epoch": 11.48, "learning_rate": 2.738225876584889e-06, "loss": 0.0229, "step": 11317 }, { "epoch": 11.48, "learning_rate": 2.73671892104152e-06, "loss": 0.0206, "step": 11318 }, { "epoch": 11.49, "learning_rate": 2.735212314541692e-06, "loss": 0.0198, "step": 11319 }, { "epoch": 11.49, "learning_rate": 2.733706057157799e-06, "loss": 0.0244, "step": 11320 }, { "epoch": 11.49, "learning_rate": 2.7322001489622353e-06, "loss": 0.0236, "step": 11321 }, { "epoch": 11.49, "learning_rate": 2.730694590027361e-06, "loss": 0.0196, "step": 11322 }, { "epoch": 11.49, "learning_rate": 2.72918938042553e-06, "loss": 0.0232, "step": 11323 }, { "epoch": 11.49, "learning_rate": 2.727684520229079e-06, "loss": 0.0192, "step": 11324 }, { "epoch": 11.49, "learning_rate": 2.7261800095103196e-06, "loss": 0.0204, "step": 11325 }, { "epoch": 11.49, "learning_rate": 2.7246758483415592e-06, "loss": 0.0215, "step": 11326 }, { "epoch": 11.49, "learning_rate": 2.723172036795074e-06, "loss": 0.0199, "step": 11327 }, { "epoch": 11.49, "learning_rate": 2.721668574943136e-06, "loss": 0.0231, "step": 11328 }, { "epoch": 11.5, "learning_rate": 2.7201654628579945e-06, "loss": 0.0221, "step": 11329 }, { "epoch": 11.5, "learning_rate": 2.7186627006118828e-06, "loss": 0.0195, "step": 11330 }, { "epoch": 11.5, "learning_rate": 2.717160288277021e-06, "loss": 0.0237, "step": 11331 }, { "epoch": 11.5, "learning_rate": 2.7156582259256013e-06, "loss": 0.0239, "step": 11332 }, { "epoch": 11.5, "learning_rate": 2.714156513629812e-06, "loss": 0.0177, "step": 11333 }, { "epoch": 11.5, "learning_rate": 2.7126551514618205e-06, "loss": 0.0201, "step": 11334 }, { "epoch": 11.5, "learning_rate": 2.71115413949377e-06, "loss": 0.0224, "step": 11335 }, { "epoch": 11.5, "learning_rate": 2.709653477797801e-06, "loss": 0.0214, "step": 11336 }, { "epoch": 11.5, "learning_rate": 2.7081531664460182e-06, "loss": 0.0204, "step": 11337 }, { "epoch": 11.5, "learning_rate": 2.706653205510532e-06, "loss": 0.0223, "step": 11338 }, { "epoch": 11.51, "learning_rate": 2.7051535950634176e-06, "loss": 0.0197, "step": 11339 }, { "epoch": 11.51, "learning_rate": 2.703654335176742e-06, "loss": 0.0202, "step": 11340 }, { "epoch": 11.51, "learning_rate": 2.7021554259225567e-06, "loss": 0.021, "step": 11341 }, { "epoch": 11.51, "learning_rate": 2.700656867372886e-06, "loss": 0.0231, "step": 11342 }, { "epoch": 11.51, "learning_rate": 2.6991586595997523e-06, "loss": 0.0202, "step": 11343 }, { "epoch": 11.51, "learning_rate": 2.697660802675146e-06, "loss": 0.0165, "step": 11344 }, { "epoch": 11.51, "learning_rate": 2.6961632966710483e-06, "loss": 0.0186, "step": 11345 }, { "epoch": 11.51, "learning_rate": 2.694666141659432e-06, "loss": 0.0241, "step": 11346 }, { "epoch": 11.51, "learning_rate": 2.693169337712236e-06, "loss": 0.0173, "step": 11347 }, { "epoch": 11.51, "learning_rate": 2.691672884901395e-06, "loss": 0.0198, "step": 11348 }, { "epoch": 11.52, "learning_rate": 2.690176783298819e-06, "loss": 0.0255, "step": 11349 }, { "epoch": 11.52, "learning_rate": 2.688681032976406e-06, "loss": 0.0231, "step": 11350 }, { "epoch": 11.52, "learning_rate": 2.6871856340060377e-06, "loss": 0.0275, "step": 11351 }, { "epoch": 11.52, "learning_rate": 2.6856905864595716e-06, "loss": 0.0204, "step": 11352 }, { "epoch": 11.52, "learning_rate": 2.6841958904088596e-06, "loss": 0.0229, "step": 11353 }, { "epoch": 11.52, "learning_rate": 2.682701545925722e-06, "loss": 0.0224, "step": 11354 }, { "epoch": 11.52, "learning_rate": 2.6812075530819835e-06, "loss": 0.0254, "step": 11355 }, { "epoch": 11.52, "learning_rate": 2.6797139119494288e-06, "loss": 0.0294, "step": 11356 }, { "epoch": 11.52, "learning_rate": 2.678220622599841e-06, "loss": 0.0217, "step": 11357 }, { "epoch": 11.53, "learning_rate": 2.6767276851049818e-06, "loss": 0.0216, "step": 11358 }, { "epoch": 11.53, "learning_rate": 2.6752350995365917e-06, "loss": 0.0232, "step": 11359 }, { "epoch": 11.53, "learning_rate": 2.6737428659664e-06, "loss": 0.0194, "step": 11360 }, { "epoch": 11.53, "learning_rate": 2.6722509844661217e-06, "loss": 0.0229, "step": 11361 }, { "epoch": 11.53, "learning_rate": 2.6707594551074414e-06, "loss": 0.0225, "step": 11362 }, { "epoch": 11.53, "learning_rate": 2.6692682779620473e-06, "loss": 0.0202, "step": 11363 }, { "epoch": 11.53, "learning_rate": 2.667777453101589e-06, "loss": 0.0231, "step": 11364 }, { "epoch": 11.53, "learning_rate": 2.6662869805977177e-06, "loss": 0.0165, "step": 11365 }, { "epoch": 11.53, "learning_rate": 2.664796860522052e-06, "loss": 0.0217, "step": 11366 }, { "epoch": 11.53, "learning_rate": 2.6633070929462047e-06, "loss": 0.0211, "step": 11367 }, { "epoch": 11.54, "learning_rate": 2.6618176779417716e-06, "loss": 0.0221, "step": 11368 }, { "epoch": 11.54, "learning_rate": 2.66032861558032e-06, "loss": 0.0208, "step": 11369 }, { "epoch": 11.54, "learning_rate": 2.6588399059334113e-06, "loss": 0.021, "step": 11370 }, { "epoch": 11.54, "learning_rate": 2.6573515490725887e-06, "loss": 0.0215, "step": 11371 }, { "epoch": 11.54, "learning_rate": 2.655863545069378e-06, "loss": 0.0324, "step": 11372 }, { "epoch": 11.54, "learning_rate": 2.654375893995281e-06, "loss": 0.0188, "step": 11373 }, { "epoch": 11.54, "learning_rate": 2.652888595921791e-06, "loss": 0.0245, "step": 11374 }, { "epoch": 11.54, "learning_rate": 2.6514016509203843e-06, "loss": 0.0238, "step": 11375 }, { "epoch": 11.54, "learning_rate": 2.6499150590625122e-06, "loss": 0.0214, "step": 11376 }, { "epoch": 11.54, "learning_rate": 2.648428820419616e-06, "loss": 0.0204, "step": 11377 }, { "epoch": 11.55, "learning_rate": 2.6469429350631226e-06, "loss": 0.0197, "step": 11378 }, { "epoch": 11.55, "learning_rate": 2.6454574030644275e-06, "loss": 0.0227, "step": 11379 }, { "epoch": 11.55, "learning_rate": 2.6439722244949316e-06, "loss": 0.0238, "step": 11380 }, { "epoch": 11.55, "learning_rate": 2.642487399425998e-06, "loss": 0.0194, "step": 11381 }, { "epoch": 11.55, "learning_rate": 2.6410029279289863e-06, "loss": 0.0221, "step": 11382 }, { "epoch": 11.55, "learning_rate": 2.639518810075229e-06, "loss": 0.0229, "step": 11383 }, { "epoch": 11.55, "learning_rate": 2.638035045936048e-06, "loss": 0.0257, "step": 11384 }, { "epoch": 11.55, "learning_rate": 2.6365516355827535e-06, "loss": 0.0195, "step": 11385 }, { "epoch": 11.55, "learning_rate": 2.6350685790866227e-06, "loss": 0.0233, "step": 11386 }, { "epoch": 11.55, "learning_rate": 2.6335858765189293e-06, "loss": 0.0228, "step": 11387 }, { "epoch": 11.56, "learning_rate": 2.6321035279509266e-06, "loss": 0.0207, "step": 11388 }, { "epoch": 11.56, "learning_rate": 2.63062153345385e-06, "loss": 0.015, "step": 11389 }, { "epoch": 11.56, "learning_rate": 2.6291398930989196e-06, "loss": 0.0202, "step": 11390 }, { "epoch": 11.56, "learning_rate": 2.627658606957333e-06, "loss": 0.0202, "step": 11391 }, { "epoch": 11.56, "learning_rate": 2.6261776751002798e-06, "loss": 0.0221, "step": 11392 }, { "epoch": 11.56, "learning_rate": 2.6246970975989217e-06, "loss": 0.0277, "step": 11393 }, { "epoch": 11.56, "learning_rate": 2.6232168745244114e-06, "loss": 0.0181, "step": 11394 }, { "epoch": 11.56, "learning_rate": 2.621737005947884e-06, "loss": 0.0265, "step": 11395 }, { "epoch": 11.56, "learning_rate": 2.6202574919404545e-06, "loss": 0.0191, "step": 11396 }, { "epoch": 11.56, "learning_rate": 2.6187783325732253e-06, "loss": 0.0218, "step": 11397 }, { "epoch": 11.57, "learning_rate": 2.617299527917273e-06, "loss": 0.0222, "step": 11398 }, { "epoch": 11.57, "learning_rate": 2.61582107804367e-06, "loss": 0.021, "step": 11399 }, { "epoch": 11.57, "learning_rate": 2.614342983023458e-06, "loss": 0.0228, "step": 11400 }, { "epoch": 11.57, "learning_rate": 2.612865242927669e-06, "loss": 0.0229, "step": 11401 }, { "epoch": 11.57, "learning_rate": 2.6113878578273233e-06, "loss": 0.0216, "step": 11402 }, { "epoch": 11.57, "learning_rate": 2.6099108277934105e-06, "loss": 0.0242, "step": 11403 }, { "epoch": 11.57, "learning_rate": 2.6084341528969135e-06, "loss": 0.02, "step": 11404 }, { "epoch": 11.57, "learning_rate": 2.6069578332087953e-06, "loss": 0.0211, "step": 11405 }, { "epoch": 11.57, "learning_rate": 2.605481868800003e-06, "loss": 0.0235, "step": 11406 }, { "epoch": 11.57, "learning_rate": 2.6040062597414663e-06, "loss": 0.0248, "step": 11407 }, { "epoch": 11.58, "learning_rate": 2.602531006104092e-06, "loss": 0.0182, "step": 11408 }, { "epoch": 11.58, "learning_rate": 2.6010561079587817e-06, "loss": 0.0209, "step": 11409 }, { "epoch": 11.58, "learning_rate": 2.5995815653764045e-06, "loss": 0.0182, "step": 11410 }, { "epoch": 11.58, "learning_rate": 2.5981073784278266e-06, "loss": 0.0241, "step": 11411 }, { "epoch": 11.58, "learning_rate": 2.59663354718389e-06, "loss": 0.0211, "step": 11412 }, { "epoch": 11.58, "learning_rate": 2.5951600717154212e-06, "loss": 0.0224, "step": 11413 }, { "epoch": 11.58, "learning_rate": 2.5936869520932316e-06, "loss": 0.0279, "step": 11414 }, { "epoch": 11.58, "learning_rate": 2.592214188388109e-06, "loss": 0.0219, "step": 11415 }, { "epoch": 11.58, "learning_rate": 2.5907417806708335e-06, "loss": 0.0232, "step": 11416 }, { "epoch": 11.58, "learning_rate": 2.589269729012156e-06, "loss": 0.0232, "step": 11417 }, { "epoch": 11.59, "learning_rate": 2.587798033482821e-06, "loss": 0.0206, "step": 11418 }, { "epoch": 11.59, "learning_rate": 2.586326694153557e-06, "loss": 0.0229, "step": 11419 }, { "epoch": 11.59, "learning_rate": 2.584855711095059e-06, "loss": 0.0215, "step": 11420 }, { "epoch": 11.59, "learning_rate": 2.5833850843780285e-06, "loss": 0.031, "step": 11421 }, { "epoch": 11.59, "learning_rate": 2.581914814073131e-06, "loss": 0.0208, "step": 11422 }, { "epoch": 11.59, "learning_rate": 2.580444900251022e-06, "loss": 0.0272, "step": 11423 }, { "epoch": 11.59, "learning_rate": 2.578975342982345e-06, "loss": 0.0265, "step": 11424 }, { "epoch": 11.59, "learning_rate": 2.5775061423377137e-06, "loss": 0.0212, "step": 11425 }, { "epoch": 11.59, "learning_rate": 2.576037298387737e-06, "loss": 0.0198, "step": 11426 }, { "epoch": 11.6, "learning_rate": 2.5745688112029965e-06, "loss": 0.0201, "step": 11427 }, { "epoch": 11.6, "learning_rate": 2.573100680854065e-06, "loss": 0.0192, "step": 11428 }, { "epoch": 11.6, "learning_rate": 2.5716329074114942e-06, "loss": 0.0226, "step": 11429 }, { "epoch": 11.6, "learning_rate": 2.5701654909458194e-06, "loss": 0.0173, "step": 11430 }, { "epoch": 11.6, "learning_rate": 2.568698431527563e-06, "loss": 0.0187, "step": 11431 }, { "epoch": 11.6, "learning_rate": 2.5672317292272174e-06, "loss": 0.0231, "step": 11432 }, { "epoch": 11.6, "learning_rate": 2.565765384115271e-06, "loss": 0.0242, "step": 11433 }, { "epoch": 11.6, "learning_rate": 2.564299396262192e-06, "loss": 0.0216, "step": 11434 }, { "epoch": 11.6, "learning_rate": 2.562833765738425e-06, "loss": 0.0213, "step": 11435 }, { "epoch": 11.6, "learning_rate": 2.5613684926144087e-06, "loss": 0.0193, "step": 11436 }, { "epoch": 11.61, "learning_rate": 2.5599035769605473e-06, "loss": 0.0229, "step": 11437 }, { "epoch": 11.61, "learning_rate": 2.5584390188472517e-06, "loss": 0.0185, "step": 11438 }, { "epoch": 11.61, "learning_rate": 2.5569748183448938e-06, "loss": 0.0227, "step": 11439 }, { "epoch": 11.61, "learning_rate": 2.5555109755238395e-06, "loss": 0.0223, "step": 11440 }, { "epoch": 11.61, "learning_rate": 2.554047490454439e-06, "loss": 0.0196, "step": 11441 }, { "epoch": 11.61, "learning_rate": 2.5525843632070136e-06, "loss": 0.0194, "step": 11442 }, { "epoch": 11.61, "learning_rate": 2.551121593851883e-06, "loss": 0.024, "step": 11443 }, { "epoch": 11.61, "learning_rate": 2.5496591824593344e-06, "loss": 0.0189, "step": 11444 }, { "epoch": 11.61, "learning_rate": 2.5481971290996497e-06, "loss": 0.022, "step": 11445 }, { "epoch": 11.61, "learning_rate": 2.5467354338430894e-06, "loss": 0.0219, "step": 11446 }, { "epoch": 11.62, "learning_rate": 2.5452740967598945e-06, "loss": 0.0188, "step": 11447 }, { "epoch": 11.62, "learning_rate": 2.5438131179202964e-06, "loss": 0.0256, "step": 11448 }, { "epoch": 11.62, "learning_rate": 2.542352497394496e-06, "loss": 0.0226, "step": 11449 }, { "epoch": 11.62, "learning_rate": 2.5408922352526877e-06, "loss": 0.0211, "step": 11450 }, { "epoch": 11.62, "learning_rate": 2.539432331565049e-06, "loss": 0.0191, "step": 11451 }, { "epoch": 11.62, "learning_rate": 2.5379727864017323e-06, "loss": 0.0252, "step": 11452 }, { "epoch": 11.62, "learning_rate": 2.5365135998328816e-06, "loss": 0.0223, "step": 11453 }, { "epoch": 11.62, "learning_rate": 2.5350547719286124e-06, "loss": 0.0241, "step": 11454 }, { "epoch": 11.62, "learning_rate": 2.5335963027590395e-06, "loss": 0.0173, "step": 11455 }, { "epoch": 11.62, "learning_rate": 2.5321381923942434e-06, "loss": 0.0219, "step": 11456 }, { "epoch": 11.63, "learning_rate": 2.5306804409042974e-06, "loss": 0.018, "step": 11457 }, { "epoch": 11.63, "learning_rate": 2.5292230483592585e-06, "loss": 0.0172, "step": 11458 }, { "epoch": 11.63, "learning_rate": 2.527766014829157e-06, "loss": 0.024, "step": 11459 }, { "epoch": 11.63, "learning_rate": 2.5263093403840145e-06, "loss": 0.0198, "step": 11460 }, { "epoch": 11.63, "learning_rate": 2.5248530250938363e-06, "loss": 0.0204, "step": 11461 }, { "epoch": 11.63, "learning_rate": 2.523397069028597e-06, "loss": 0.0214, "step": 11462 }, { "epoch": 11.63, "learning_rate": 2.521941472258278e-06, "loss": 0.0192, "step": 11463 }, { "epoch": 11.63, "learning_rate": 2.5204862348528183e-06, "loss": 0.0181, "step": 11464 }, { "epoch": 11.63, "learning_rate": 2.5190313568821566e-06, "loss": 0.0217, "step": 11465 }, { "epoch": 11.63, "learning_rate": 2.5175768384162035e-06, "loss": 0.0232, "step": 11466 }, { "epoch": 11.64, "learning_rate": 2.5161226795248607e-06, "loss": 0.0257, "step": 11467 }, { "epoch": 11.64, "learning_rate": 2.5146688802780094e-06, "loss": 0.0234, "step": 11468 }, { "epoch": 11.64, "learning_rate": 2.513215440745511e-06, "loss": 0.0241, "step": 11469 }, { "epoch": 11.64, "learning_rate": 2.511762360997214e-06, "loss": 0.0211, "step": 11470 }, { "epoch": 11.64, "learning_rate": 2.5103096411029417e-06, "loss": 0.0186, "step": 11471 }, { "epoch": 11.64, "learning_rate": 2.5088572811325164e-06, "loss": 0.017, "step": 11472 }, { "epoch": 11.64, "learning_rate": 2.5074052811557235e-06, "loss": 0.0202, "step": 11473 }, { "epoch": 11.64, "learning_rate": 2.5059536412423434e-06, "loss": 0.0251, "step": 11474 }, { "epoch": 11.64, "learning_rate": 2.504502361462139e-06, "loss": 0.0172, "step": 11475 }, { "epoch": 11.64, "learning_rate": 2.5030514418848463e-06, "loss": 0.0227, "step": 11476 }, { "epoch": 11.65, "learning_rate": 2.501600882580194e-06, "loss": 0.0224, "step": 11477 }, { "epoch": 11.65, "learning_rate": 2.5001506836178933e-06, "loss": 0.0229, "step": 11478 }, { "epoch": 11.65, "learning_rate": 2.4987008450676253e-06, "loss": 0.0303, "step": 11479 }, { "epoch": 11.65, "learning_rate": 2.497251366999076e-06, "loss": 0.0202, "step": 11480 }, { "epoch": 11.65, "learning_rate": 2.495802249481892e-06, "loss": 0.0258, "step": 11481 }, { "epoch": 11.65, "learning_rate": 2.4943534925857183e-06, "loss": 0.0206, "step": 11482 }, { "epoch": 11.65, "learning_rate": 2.4929050963801693e-06, "loss": 0.0201, "step": 11483 }, { "epoch": 11.65, "learning_rate": 2.491457060934853e-06, "loss": 0.0232, "step": 11484 }, { "epoch": 11.65, "learning_rate": 2.4900093863193575e-06, "loss": 0.0183, "step": 11485 }, { "epoch": 11.65, "learning_rate": 2.488562072603249e-06, "loss": 0.0186, "step": 11486 }, { "epoch": 11.66, "learning_rate": 2.4871151198560794e-06, "loss": 0.0216, "step": 11487 }, { "epoch": 11.66, "learning_rate": 2.4856685281473845e-06, "loss": 0.0201, "step": 11488 }, { "epoch": 11.66, "learning_rate": 2.4842222975466824e-06, "loss": 0.0214, "step": 11489 }, { "epoch": 11.66, "learning_rate": 2.4827764281234743e-06, "loss": 0.0186, "step": 11490 }, { "epoch": 11.66, "learning_rate": 2.481330919947238e-06, "loss": 0.0222, "step": 11491 }, { "epoch": 11.66, "learning_rate": 2.479885773087445e-06, "loss": 0.021, "step": 11492 }, { "epoch": 11.66, "learning_rate": 2.4784409876135364e-06, "loss": 0.0228, "step": 11493 }, { "epoch": 11.66, "learning_rate": 2.4769965635949454e-06, "loss": 0.0212, "step": 11494 }, { "epoch": 11.66, "learning_rate": 2.4755525011010896e-06, "loss": 0.0182, "step": 11495 }, { "epoch": 11.67, "learning_rate": 2.474108800201355e-06, "loss": 0.0233, "step": 11496 }, { "epoch": 11.67, "learning_rate": 2.4726654609651302e-06, "loss": 0.0197, "step": 11497 }, { "epoch": 11.67, "learning_rate": 2.4712224834617694e-06, "loss": 0.0238, "step": 11498 }, { "epoch": 11.67, "learning_rate": 2.469779867760621e-06, "loss": 0.024, "step": 11499 }, { "epoch": 11.67, "learning_rate": 2.4683376139310054e-06, "loss": 0.023, "step": 11500 }, { "epoch": 11.67, "learning_rate": 2.466895722042235e-06, "loss": 0.0227, "step": 11501 }, { "epoch": 11.67, "learning_rate": 2.465454192163604e-06, "loss": 0.0179, "step": 11502 }, { "epoch": 11.67, "learning_rate": 2.46401302436438e-06, "loss": 0.0207, "step": 11503 }, { "epoch": 11.67, "learning_rate": 2.4625722187138236e-06, "loss": 0.0229, "step": 11504 }, { "epoch": 11.67, "learning_rate": 2.4611317752811724e-06, "loss": 0.0209, "step": 11505 }, { "epoch": 11.68, "learning_rate": 2.45969169413565e-06, "loss": 0.0232, "step": 11506 }, { "epoch": 11.68, "learning_rate": 2.4582519753464617e-06, "loss": 0.0218, "step": 11507 }, { "epoch": 11.68, "learning_rate": 2.45681261898279e-06, "loss": 0.0212, "step": 11508 }, { "epoch": 11.68, "learning_rate": 2.4553736251138105e-06, "loss": 0.0231, "step": 11509 }, { "epoch": 11.68, "learning_rate": 2.453934993808669e-06, "loss": 0.0238, "step": 11510 }, { "epoch": 11.68, "learning_rate": 2.452496725136503e-06, "loss": 0.0206, "step": 11511 }, { "epoch": 11.68, "learning_rate": 2.451058819166431e-06, "loss": 0.0233, "step": 11512 }, { "epoch": 11.68, "learning_rate": 2.449621275967551e-06, "loss": 0.0213, "step": 11513 }, { "epoch": 11.68, "learning_rate": 2.44818409560895e-06, "loss": 0.0246, "step": 11514 }, { "epoch": 11.68, "learning_rate": 2.4467472781596856e-06, "loss": 0.0213, "step": 11515 }, { "epoch": 11.69, "learning_rate": 2.4453108236888124e-06, "loss": 0.0232, "step": 11516 }, { "epoch": 11.69, "learning_rate": 2.4438747322653545e-06, "loss": 0.0217, "step": 11517 }, { "epoch": 11.69, "learning_rate": 2.442439003958327e-06, "loss": 0.0198, "step": 11518 }, { "epoch": 11.69, "learning_rate": 2.4410036388367287e-06, "loss": 0.0244, "step": 11519 }, { "epoch": 11.69, "learning_rate": 2.4395686369695305e-06, "loss": 0.0209, "step": 11520 }, { "epoch": 11.69, "learning_rate": 2.438133998425697e-06, "loss": 0.023, "step": 11521 }, { "epoch": 11.69, "learning_rate": 2.4366997232741696e-06, "loss": 0.0271, "step": 11522 }, { "epoch": 11.69, "learning_rate": 2.4352658115838754e-06, "loss": 0.0206, "step": 11523 }, { "epoch": 11.69, "learning_rate": 2.4338322634237246e-06, "loss": 0.0304, "step": 11524 }, { "epoch": 11.69, "learning_rate": 2.4323990788626006e-06, "loss": 0.0207, "step": 11525 }, { "epoch": 11.7, "learning_rate": 2.4309662579693826e-06, "loss": 0.0233, "step": 11526 }, { "epoch": 11.7, "learning_rate": 2.429533800812922e-06, "loss": 0.0224, "step": 11527 }, { "epoch": 11.7, "learning_rate": 2.428101707462058e-06, "loss": 0.0227, "step": 11528 }, { "epoch": 11.7, "learning_rate": 2.4266699779856116e-06, "loss": 0.018, "step": 11529 }, { "epoch": 11.7, "learning_rate": 2.4252386124523863e-06, "loss": 0.0195, "step": 11530 }, { "epoch": 11.7, "learning_rate": 2.4238076109311703e-06, "loss": 0.0235, "step": 11531 }, { "epoch": 11.7, "learning_rate": 2.4223769734907253e-06, "loss": 0.0227, "step": 11532 }, { "epoch": 11.7, "learning_rate": 2.4209467001998053e-06, "loss": 0.019, "step": 11533 }, { "epoch": 11.7, "learning_rate": 2.4195167911271467e-06, "loss": 0.0265, "step": 11534 }, { "epoch": 11.7, "learning_rate": 2.4180872463414583e-06, "loss": 0.0235, "step": 11535 }, { "epoch": 11.71, "learning_rate": 2.4166580659114446e-06, "loss": 0.0196, "step": 11536 }, { "epoch": 11.71, "learning_rate": 2.415229249905777e-06, "loss": 0.0195, "step": 11537 }, { "epoch": 11.71, "learning_rate": 2.413800798393131e-06, "loss": 0.0188, "step": 11538 }, { "epoch": 11.71, "learning_rate": 2.4123727114421425e-06, "loss": 0.0357, "step": 11539 }, { "epoch": 11.71, "learning_rate": 2.410944989121443e-06, "loss": 0.023, "step": 11540 }, { "epoch": 11.71, "learning_rate": 2.409517631499646e-06, "loss": 0.0207, "step": 11541 }, { "epoch": 11.71, "learning_rate": 2.4080906386453383e-06, "loss": 0.0219, "step": 11542 }, { "epoch": 11.71, "learning_rate": 2.406664010627101e-06, "loss": 0.0211, "step": 11543 }, { "epoch": 11.71, "learning_rate": 2.405237747513486e-06, "loss": 0.0212, "step": 11544 }, { "epoch": 11.71, "learning_rate": 2.4038118493730366e-06, "loss": 0.0214, "step": 11545 }, { "epoch": 11.72, "learning_rate": 2.4023863162742777e-06, "loss": 0.0225, "step": 11546 }, { "epoch": 11.72, "learning_rate": 2.400961148285711e-06, "loss": 0.0198, "step": 11547 }, { "epoch": 11.72, "learning_rate": 2.39953634547583e-06, "loss": 0.0225, "step": 11548 }, { "epoch": 11.72, "learning_rate": 2.3981119079130978e-06, "loss": 0.0187, "step": 11549 }, { "epoch": 11.72, "learning_rate": 2.3966878356659706e-06, "loss": 0.0274, "step": 11550 }, { "epoch": 11.72, "learning_rate": 2.3952641288028857e-06, "loss": 0.0201, "step": 11551 }, { "epoch": 11.72, "learning_rate": 2.3938407873922543e-06, "loss": 0.0227, "step": 11552 }, { "epoch": 11.72, "learning_rate": 2.392417811502484e-06, "loss": 0.0274, "step": 11553 }, { "epoch": 11.72, "learning_rate": 2.390995201201948e-06, "loss": 0.0217, "step": 11554 }, { "epoch": 11.73, "learning_rate": 2.3895729565590207e-06, "loss": 0.0212, "step": 11555 }, { "epoch": 11.73, "learning_rate": 2.3881510776420427e-06, "loss": 0.0294, "step": 11556 }, { "epoch": 11.73, "learning_rate": 2.3867295645193456e-06, "loss": 0.0186, "step": 11557 }, { "epoch": 11.73, "learning_rate": 2.3853084172592456e-06, "loss": 0.0227, "step": 11558 }, { "epoch": 11.73, "learning_rate": 2.38388763593003e-06, "loss": 0.02, "step": 11559 }, { "epoch": 11.73, "learning_rate": 2.3824672205999832e-06, "loss": 0.0214, "step": 11560 }, { "epoch": 11.73, "learning_rate": 2.3810471713373574e-06, "loss": 0.0238, "step": 11561 }, { "epoch": 11.73, "learning_rate": 2.3796274882103964e-06, "loss": 0.0234, "step": 11562 }, { "epoch": 11.73, "learning_rate": 2.3782081712873263e-06, "loss": 0.0233, "step": 11563 }, { "epoch": 11.73, "learning_rate": 2.376789220636353e-06, "loss": 0.0239, "step": 11564 }, { "epoch": 11.74, "learning_rate": 2.375370636325668e-06, "loss": 0.0249, "step": 11565 }, { "epoch": 11.74, "learning_rate": 2.3739524184234373e-06, "loss": 0.019, "step": 11566 }, { "epoch": 11.74, "learning_rate": 2.372534566997817e-06, "loss": 0.0217, "step": 11567 }, { "epoch": 11.74, "learning_rate": 2.371117082116946e-06, "loss": 0.0191, "step": 11568 }, { "epoch": 11.74, "learning_rate": 2.3696999638489383e-06, "loss": 0.021, "step": 11569 }, { "epoch": 11.74, "learning_rate": 2.368283212261899e-06, "loss": 0.0186, "step": 11570 }, { "epoch": 11.74, "learning_rate": 2.366866827423905e-06, "loss": 0.0182, "step": 11571 }, { "epoch": 11.74, "learning_rate": 2.3654508094030304e-06, "loss": 0.0181, "step": 11572 }, { "epoch": 11.74, "learning_rate": 2.3640351582673172e-06, "loss": 0.0238, "step": 11573 }, { "epoch": 11.74, "learning_rate": 2.3626198740847972e-06, "loss": 0.02, "step": 11574 }, { "epoch": 11.75, "learning_rate": 2.3612049569234863e-06, "loss": 0.0258, "step": 11575 }, { "epoch": 11.75, "learning_rate": 2.359790406851373e-06, "loss": 0.0226, "step": 11576 }, { "epoch": 11.75, "learning_rate": 2.3583762239364406e-06, "loss": 0.0232, "step": 11577 }, { "epoch": 11.75, "learning_rate": 2.3569624082466493e-06, "loss": 0.0192, "step": 11578 }, { "epoch": 11.75, "learning_rate": 2.3555489598499327e-06, "loss": 0.0211, "step": 11579 }, { "epoch": 11.75, "learning_rate": 2.3541358788142277e-06, "loss": 0.0209, "step": 11580 }, { "epoch": 11.75, "learning_rate": 2.3527231652074324e-06, "loss": 0.0212, "step": 11581 }, { "epoch": 11.75, "learning_rate": 2.351310819097442e-06, "loss": 0.0224, "step": 11582 }, { "epoch": 11.75, "learning_rate": 2.3498988405521204e-06, "loss": 0.0229, "step": 11583 }, { "epoch": 11.75, "learning_rate": 2.348487229639328e-06, "loss": 0.0222, "step": 11584 }, { "epoch": 11.76, "learning_rate": 2.347075986426901e-06, "loss": 0.0193, "step": 11585 }, { "epoch": 11.76, "learning_rate": 2.3456651109826524e-06, "loss": 0.0253, "step": 11586 }, { "epoch": 11.76, "learning_rate": 2.3442546033743906e-06, "loss": 0.0225, "step": 11587 }, { "epoch": 11.76, "learning_rate": 2.342844463669889e-06, "loss": 0.024, "step": 11588 }, { "epoch": 11.76, "learning_rate": 2.3414346919369234e-06, "loss": 0.0183, "step": 11589 }, { "epoch": 11.76, "learning_rate": 2.340025288243236e-06, "loss": 0.019, "step": 11590 }, { "epoch": 11.76, "learning_rate": 2.338616252656558e-06, "loss": 0.0164, "step": 11591 }, { "epoch": 11.76, "learning_rate": 2.337207585244604e-06, "loss": 0.0211, "step": 11592 }, { "epoch": 11.76, "learning_rate": 2.3357992860750644e-06, "loss": 0.0228, "step": 11593 }, { "epoch": 11.76, "learning_rate": 2.3343913552156196e-06, "loss": 0.024, "step": 11594 }, { "epoch": 11.77, "learning_rate": 2.3329837927339303e-06, "loss": 0.0236, "step": 11595 }, { "epoch": 11.77, "learning_rate": 2.33157659869763e-06, "loss": 0.0218, "step": 11596 }, { "epoch": 11.77, "learning_rate": 2.3301697731743555e-06, "loss": 0.0193, "step": 11597 }, { "epoch": 11.77, "learning_rate": 2.3287633162317024e-06, "loss": 0.0243, "step": 11598 }, { "epoch": 11.77, "learning_rate": 2.327357227937267e-06, "loss": 0.0242, "step": 11599 }, { "epoch": 11.77, "learning_rate": 2.3259515083586127e-06, "loss": 0.0192, "step": 11600 }, { "epoch": 11.77, "learning_rate": 2.324546157563297e-06, "loss": 0.0148, "step": 11601 }, { "epoch": 11.77, "learning_rate": 2.323141175618857e-06, "loss": 0.0236, "step": 11602 }, { "epoch": 11.77, "learning_rate": 2.3217365625928045e-06, "loss": 0.0203, "step": 11603 }, { "epoch": 11.77, "learning_rate": 2.320332318552644e-06, "loss": 0.0217, "step": 11604 }, { "epoch": 11.78, "learning_rate": 2.3189284435658556e-06, "loss": 0.0257, "step": 11605 }, { "epoch": 11.78, "learning_rate": 2.3175249376999054e-06, "loss": 0.0189, "step": 11606 }, { "epoch": 11.78, "learning_rate": 2.3161218010222417e-06, "loss": 0.0204, "step": 11607 }, { "epoch": 11.78, "learning_rate": 2.314719033600289e-06, "loss": 0.0234, "step": 11608 }, { "epoch": 11.78, "learning_rate": 2.3133166355014637e-06, "loss": 0.022, "step": 11609 }, { "epoch": 11.78, "learning_rate": 2.311914606793154e-06, "loss": 0.022, "step": 11610 }, { "epoch": 11.78, "learning_rate": 2.310512947542737e-06, "loss": 0.0176, "step": 11611 }, { "epoch": 11.78, "learning_rate": 2.3091116578175756e-06, "loss": 0.0242, "step": 11612 }, { "epoch": 11.78, "learning_rate": 2.3077107376850005e-06, "loss": 0.0228, "step": 11613 }, { "epoch": 11.78, "learning_rate": 2.306310187212345e-06, "loss": 0.0299, "step": 11614 }, { "epoch": 11.79, "learning_rate": 2.3049100064669073e-06, "loss": 0.0236, "step": 11615 }, { "epoch": 11.79, "learning_rate": 2.3035101955159776e-06, "loss": 0.0193, "step": 11616 }, { "epoch": 11.79, "learning_rate": 2.3021107544268205e-06, "loss": 0.0221, "step": 11617 }, { "epoch": 11.79, "learning_rate": 2.300711683266691e-06, "loss": 0.0261, "step": 11618 }, { "epoch": 11.79, "learning_rate": 2.299312982102824e-06, "loss": 0.0205, "step": 11619 }, { "epoch": 11.79, "learning_rate": 2.297914651002431e-06, "loss": 0.0241, "step": 11620 }, { "epoch": 11.79, "learning_rate": 2.2965166900327117e-06, "loss": 0.0241, "step": 11621 }, { "epoch": 11.79, "learning_rate": 2.2951190992608474e-06, "loss": 0.021, "step": 11622 }, { "epoch": 11.79, "learning_rate": 2.2937218787540005e-06, "loss": 0.0241, "step": 11623 }, { "epoch": 11.8, "learning_rate": 2.2923250285793174e-06, "loss": 0.0264, "step": 11624 }, { "epoch": 11.8, "learning_rate": 2.2909285488039213e-06, "loss": 0.0243, "step": 11625 }, { "epoch": 11.8, "learning_rate": 2.289532439494926e-06, "loss": 0.0276, "step": 11626 }, { "epoch": 11.8, "learning_rate": 2.2881367007194155e-06, "loss": 0.0222, "step": 11627 }, { "epoch": 11.8, "learning_rate": 2.2867413325444686e-06, "loss": 0.0232, "step": 11628 }, { "epoch": 11.8, "learning_rate": 2.2853463350371407e-06, "loss": 0.0227, "step": 11629 }, { "epoch": 11.8, "learning_rate": 2.283951708264468e-06, "loss": 0.0185, "step": 11630 }, { "epoch": 11.8, "learning_rate": 2.2825574522934747e-06, "loss": 0.0206, "step": 11631 }, { "epoch": 11.8, "learning_rate": 2.281163567191156e-06, "loss": 0.0277, "step": 11632 }, { "epoch": 11.8, "learning_rate": 2.2797700530245016e-06, "loss": 0.02, "step": 11633 }, { "epoch": 11.81, "learning_rate": 2.278376909860478e-06, "loss": 0.024, "step": 11634 }, { "epoch": 11.81, "learning_rate": 2.2769841377660307e-06, "loss": 0.0201, "step": 11635 }, { "epoch": 11.81, "learning_rate": 2.275591736808096e-06, "loss": 0.0218, "step": 11636 }, { "epoch": 11.81, "learning_rate": 2.274199707053579e-06, "loss": 0.0223, "step": 11637 }, { "epoch": 11.81, "learning_rate": 2.2728080485693803e-06, "loss": 0.0178, "step": 11638 }, { "epoch": 11.81, "learning_rate": 2.271416761422377e-06, "loss": 0.0215, "step": 11639 }, { "epoch": 11.81, "learning_rate": 2.270025845679428e-06, "loss": 0.0238, "step": 11640 }, { "epoch": 11.81, "learning_rate": 2.268635301407377e-06, "loss": 0.0191, "step": 11641 }, { "epoch": 11.81, "learning_rate": 2.2672451286730434e-06, "loss": 0.0188, "step": 11642 }, { "epoch": 11.81, "learning_rate": 2.2658553275432395e-06, "loss": 0.0237, "step": 11643 }, { "epoch": 11.82, "learning_rate": 2.2644658980847455e-06, "loss": 0.0212, "step": 11644 }, { "epoch": 11.82, "learning_rate": 2.263076840364338e-06, "loss": 0.0208, "step": 11645 }, { "epoch": 11.82, "learning_rate": 2.2616881544487656e-06, "loss": 0.0249, "step": 11646 }, { "epoch": 11.82, "learning_rate": 2.260299840404765e-06, "loss": 0.0321, "step": 11647 }, { "epoch": 11.82, "learning_rate": 2.2589118982990554e-06, "loss": 0.026, "step": 11648 }, { "epoch": 11.82, "learning_rate": 2.2575243281983295e-06, "loss": 0.0216, "step": 11649 }, { "epoch": 11.82, "learning_rate": 2.2561371301692724e-06, "loss": 0.0249, "step": 11650 }, { "epoch": 11.82, "learning_rate": 2.2547503042785478e-06, "loss": 0.0219, "step": 11651 }, { "epoch": 11.82, "learning_rate": 2.253363850592798e-06, "loss": 0.0252, "step": 11652 }, { "epoch": 11.82, "learning_rate": 2.2519777691786527e-06, "loss": 0.0285, "step": 11653 }, { "epoch": 11.83, "learning_rate": 2.2505920601027155e-06, "loss": 0.0185, "step": 11654 }, { "epoch": 11.83, "learning_rate": 2.2492067234315886e-06, "loss": 0.0257, "step": 11655 }, { "epoch": 11.83, "learning_rate": 2.2478217592318365e-06, "loss": 0.0204, "step": 11656 }, { "epoch": 11.83, "learning_rate": 2.246437167570017e-06, "loss": 0.0271, "step": 11657 }, { "epoch": 11.83, "learning_rate": 2.2450529485126725e-06, "loss": 0.0198, "step": 11658 }, { "epoch": 11.83, "learning_rate": 2.2436691021263156e-06, "loss": 0.0228, "step": 11659 }, { "epoch": 11.83, "learning_rate": 2.2422856284774565e-06, "loss": 0.0313, "step": 11660 }, { "epoch": 11.83, "learning_rate": 2.24090252763257e-06, "loss": 0.0241, "step": 11661 }, { "epoch": 11.83, "learning_rate": 2.2395197996581288e-06, "loss": 0.0192, "step": 11662 }, { "epoch": 11.83, "learning_rate": 2.2381374446205784e-06, "loss": 0.019, "step": 11663 }, { "epoch": 11.84, "learning_rate": 2.2367554625863496e-06, "loss": 0.0292, "step": 11664 }, { "epoch": 11.84, "learning_rate": 2.2353738536218583e-06, "loss": 0.0194, "step": 11665 }, { "epoch": 11.84, "learning_rate": 2.2339926177934936e-06, "loss": 0.0204, "step": 11666 }, { "epoch": 11.84, "learning_rate": 2.232611755167634e-06, "loss": 0.0248, "step": 11667 }, { "epoch": 11.84, "learning_rate": 2.2312312658106416e-06, "loss": 0.0224, "step": 11668 }, { "epoch": 11.84, "learning_rate": 2.229851149788851e-06, "loss": 0.0207, "step": 11669 }, { "epoch": 11.84, "learning_rate": 2.228471407168591e-06, "loss": 0.0215, "step": 11670 }, { "epoch": 11.84, "learning_rate": 2.2270920380161575e-06, "loss": 0.0301, "step": 11671 }, { "epoch": 11.84, "learning_rate": 2.225713042397848e-06, "loss": 0.0194, "step": 11672 }, { "epoch": 11.84, "learning_rate": 2.2243344203799243e-06, "loss": 0.0196, "step": 11673 }, { "epoch": 11.85, "learning_rate": 2.2229561720286386e-06, "loss": 0.0214, "step": 11674 }, { "epoch": 11.85, "learning_rate": 2.2215782974102286e-06, "loss": 0.0201, "step": 11675 }, { "epoch": 11.85, "learning_rate": 2.2202007965909023e-06, "loss": 0.0262, "step": 11676 }, { "epoch": 11.85, "learning_rate": 2.2188236696368604e-06, "loss": 0.0221, "step": 11677 }, { "epoch": 11.85, "learning_rate": 2.217446916614283e-06, "loss": 0.0211, "step": 11678 }, { "epoch": 11.85, "learning_rate": 2.216070537589328e-06, "loss": 0.0233, "step": 11679 }, { "epoch": 11.85, "learning_rate": 2.21469453262814e-06, "loss": 0.0229, "step": 11680 }, { "epoch": 11.85, "learning_rate": 2.2133189017968436e-06, "loss": 0.02, "step": 11681 }, { "epoch": 11.85, "learning_rate": 2.2119436451615506e-06, "loss": 0.0272, "step": 11682 }, { "epoch": 11.85, "learning_rate": 2.2105687627883433e-06, "loss": 0.0192, "step": 11683 }, { "epoch": 11.86, "learning_rate": 2.209194254743295e-06, "loss": 0.0233, "step": 11684 }, { "epoch": 11.86, "learning_rate": 2.2078201210924645e-06, "loss": 0.0202, "step": 11685 }, { "epoch": 11.86, "learning_rate": 2.2064463619018784e-06, "loss": 0.0232, "step": 11686 }, { "epoch": 11.86, "learning_rate": 2.2050729772375614e-06, "loss": 0.0246, "step": 11687 }, { "epoch": 11.86, "learning_rate": 2.2036999671655036e-06, "loss": 0.0189, "step": 11688 }, { "epoch": 11.86, "learning_rate": 2.202327331751697e-06, "loss": 0.0208, "step": 11689 }, { "epoch": 11.86, "learning_rate": 2.2009550710620976e-06, "loss": 0.023, "step": 11690 }, { "epoch": 11.86, "learning_rate": 2.1995831851626527e-06, "loss": 0.0238, "step": 11691 }, { "epoch": 11.86, "learning_rate": 2.1982116741192925e-06, "loss": 0.0269, "step": 11692 }, { "epoch": 11.87, "learning_rate": 2.1968405379979207e-06, "loss": 0.0231, "step": 11693 }, { "epoch": 11.87, "learning_rate": 2.1954697768644306e-06, "loss": 0.0228, "step": 11694 }, { "epoch": 11.87, "learning_rate": 2.1940993907847e-06, "loss": 0.023, "step": 11695 }, { "epoch": 11.87, "learning_rate": 2.192729379824573e-06, "loss": 0.0225, "step": 11696 }, { "epoch": 11.87, "learning_rate": 2.191359744049899e-06, "loss": 0.0212, "step": 11697 }, { "epoch": 11.87, "learning_rate": 2.1899904835264895e-06, "loss": 0.0212, "step": 11698 }, { "epoch": 11.87, "learning_rate": 2.188621598320151e-06, "loss": 0.0202, "step": 11699 }, { "epoch": 11.87, "learning_rate": 2.18725308849666e-06, "loss": 0.0205, "step": 11700 }, { "epoch": 11.87, "learning_rate": 2.185884954121785e-06, "loss": 0.0218, "step": 11701 }, { "epoch": 11.87, "learning_rate": 2.1845171952612764e-06, "loss": 0.0185, "step": 11702 }, { "epoch": 11.88, "learning_rate": 2.183149811980856e-06, "loss": 0.0281, "step": 11703 }, { "epoch": 11.88, "learning_rate": 2.1817828043462407e-06, "loss": 0.0243, "step": 11704 }, { "epoch": 11.88, "learning_rate": 2.1804161724231143e-06, "loss": 0.0234, "step": 11705 }, { "epoch": 11.88, "learning_rate": 2.1790499162771615e-06, "loss": 0.0178, "step": 11706 }, { "epoch": 11.88, "learning_rate": 2.177684035974038e-06, "loss": 0.0242, "step": 11707 }, { "epoch": 11.88, "learning_rate": 2.176318531579378e-06, "loss": 0.0219, "step": 11708 }, { "epoch": 11.88, "learning_rate": 2.1749534031588047e-06, "loss": 0.0213, "step": 11709 }, { "epoch": 11.88, "learning_rate": 2.173588650777918e-06, "loss": 0.0223, "step": 11710 }, { "epoch": 11.88, "learning_rate": 2.1722242745023036e-06, "loss": 0.0218, "step": 11711 }, { "epoch": 11.88, "learning_rate": 2.1708602743975316e-06, "loss": 0.0238, "step": 11712 }, { "epoch": 11.89, "learning_rate": 2.1694966505291416e-06, "loss": 0.0227, "step": 11713 }, { "epoch": 11.89, "learning_rate": 2.168133402962673e-06, "loss": 0.0234, "step": 11714 }, { "epoch": 11.89, "learning_rate": 2.1667705317636333e-06, "loss": 0.0231, "step": 11715 }, { "epoch": 11.89, "learning_rate": 2.1654080369975185e-06, "loss": 0.0205, "step": 11716 }, { "epoch": 11.89, "learning_rate": 2.1640459187298013e-06, "loss": 0.0222, "step": 11717 }, { "epoch": 11.89, "learning_rate": 2.162684177025941e-06, "loss": 0.0222, "step": 11718 }, { "epoch": 11.89, "learning_rate": 2.161322811951381e-06, "loss": 0.0215, "step": 11719 }, { "epoch": 11.89, "learning_rate": 2.1599618235715357e-06, "loss": 0.024, "step": 11720 }, { "epoch": 11.89, "learning_rate": 2.158601211951813e-06, "loss": 0.024, "step": 11721 }, { "epoch": 11.89, "learning_rate": 2.1572409771575987e-06, "loss": 0.021, "step": 11722 }, { "epoch": 11.9, "learning_rate": 2.1558811192542583e-06, "loss": 0.021, "step": 11723 }, { "epoch": 11.9, "learning_rate": 2.1545216383071454e-06, "loss": 0.0199, "step": 11724 }, { "epoch": 11.9, "learning_rate": 2.153162534381585e-06, "loss": 0.0226, "step": 11725 }, { "epoch": 11.9, "learning_rate": 2.151803807542896e-06, "loss": 0.021, "step": 11726 }, { "epoch": 11.9, "learning_rate": 2.1504454578563662e-06, "loss": 0.0187, "step": 11727 }, { "epoch": 11.9, "learning_rate": 2.1490874853872767e-06, "loss": 0.0261, "step": 11728 }, { "epoch": 11.9, "learning_rate": 2.147729890200889e-06, "loss": 0.023, "step": 11729 }, { "epoch": 11.9, "learning_rate": 2.1463726723624334e-06, "loss": 0.0185, "step": 11730 }, { "epoch": 11.9, "learning_rate": 2.145015831937145e-06, "loss": 0.0233, "step": 11731 }, { "epoch": 11.9, "learning_rate": 2.1436593689902195e-06, "loss": 0.0203, "step": 11732 }, { "epoch": 11.91, "learning_rate": 2.1423032835868485e-06, "loss": 0.0228, "step": 11733 }, { "epoch": 11.91, "learning_rate": 2.1409475757921927e-06, "loss": 0.0221, "step": 11734 }, { "epoch": 11.91, "learning_rate": 2.139592245671407e-06, "loss": 0.0198, "step": 11735 }, { "epoch": 11.91, "learning_rate": 2.1382372932896245e-06, "loss": 0.0188, "step": 11736 }, { "epoch": 11.91, "learning_rate": 2.1368827187119532e-06, "loss": 0.0186, "step": 11737 }, { "epoch": 11.91, "learning_rate": 2.1355285220034915e-06, "loss": 0.0233, "step": 11738 }, { "epoch": 11.91, "learning_rate": 2.134174703229317e-06, "loss": 0.0231, "step": 11739 }, { "epoch": 11.91, "learning_rate": 2.1328212624544874e-06, "loss": 0.0218, "step": 11740 }, { "epoch": 11.91, "learning_rate": 2.1314681997440467e-06, "loss": 0.0227, "step": 11741 }, { "epoch": 11.91, "learning_rate": 2.1301155151630126e-06, "loss": 0.0227, "step": 11742 }, { "epoch": 11.92, "learning_rate": 2.1287632087763953e-06, "loss": 0.0246, "step": 11743 }, { "epoch": 11.92, "learning_rate": 2.127411280649174e-06, "loss": 0.023, "step": 11744 }, { "epoch": 11.92, "learning_rate": 2.1260597308463215e-06, "loss": 0.023, "step": 11745 }, { "epoch": 11.92, "learning_rate": 2.1247085594327864e-06, "loss": 0.0234, "step": 11746 }, { "epoch": 11.92, "learning_rate": 2.1233577664735018e-06, "loss": 0.0247, "step": 11747 }, { "epoch": 11.92, "learning_rate": 2.122007352033383e-06, "loss": 0.0285, "step": 11748 }, { "epoch": 11.92, "learning_rate": 2.120657316177319e-06, "loss": 0.0291, "step": 11749 }, { "epoch": 11.92, "learning_rate": 2.119307658970192e-06, "loss": 0.0235, "step": 11750 }, { "epoch": 11.92, "learning_rate": 2.1179583804768623e-06, "loss": 0.0214, "step": 11751 }, { "epoch": 11.92, "learning_rate": 2.116609480762166e-06, "loss": 0.0195, "step": 11752 }, { "epoch": 11.93, "learning_rate": 2.1152609598909303e-06, "loss": 0.0217, "step": 11753 }, { "epoch": 11.93, "learning_rate": 2.1139128179279534e-06, "loss": 0.0213, "step": 11754 }, { "epoch": 11.93, "learning_rate": 2.112565054938027e-06, "loss": 0.0218, "step": 11755 }, { "epoch": 11.93, "learning_rate": 2.111217670985917e-06, "loss": 0.0178, "step": 11756 }, { "epoch": 11.93, "learning_rate": 2.1098706661363723e-06, "loss": 0.0242, "step": 11757 }, { "epoch": 11.93, "learning_rate": 2.1085240404541297e-06, "loss": 0.0208, "step": 11758 }, { "epoch": 11.93, "learning_rate": 2.107177794003895e-06, "loss": 0.0193, "step": 11759 }, { "epoch": 11.93, "learning_rate": 2.10583192685037e-06, "loss": 0.0315, "step": 11760 }, { "epoch": 11.93, "learning_rate": 2.1044864390582254e-06, "loss": 0.0195, "step": 11761 }, { "epoch": 11.94, "learning_rate": 2.103141330692122e-06, "loss": 0.0197, "step": 11762 }, { "epoch": 11.94, "learning_rate": 2.1017966018167026e-06, "loss": 0.0207, "step": 11763 }, { "epoch": 11.94, "learning_rate": 2.1004522524965863e-06, "loss": 0.0249, "step": 11764 }, { "epoch": 11.94, "learning_rate": 2.0991082827963818e-06, "loss": 0.0202, "step": 11765 }, { "epoch": 11.94, "learning_rate": 2.0977646927806682e-06, "loss": 0.0212, "step": 11766 }, { "epoch": 11.94, "learning_rate": 2.096421482514016e-06, "loss": 0.0219, "step": 11767 }, { "epoch": 11.94, "learning_rate": 2.095078652060979e-06, "loss": 0.0216, "step": 11768 }, { "epoch": 11.94, "learning_rate": 2.0937362014860795e-06, "loss": 0.0237, "step": 11769 }, { "epoch": 11.94, "learning_rate": 2.0923941308538365e-06, "loss": 0.0233, "step": 11770 }, { "epoch": 11.94, "learning_rate": 2.091052440228738e-06, "loss": 0.023, "step": 11771 }, { "epoch": 11.95, "learning_rate": 2.089711129675269e-06, "loss": 0.0226, "step": 11772 }, { "epoch": 11.95, "learning_rate": 2.088370199257881e-06, "loss": 0.0218, "step": 11773 }, { "epoch": 11.95, "learning_rate": 2.0870296490410146e-06, "loss": 0.0246, "step": 11774 }, { "epoch": 11.95, "learning_rate": 2.085689479089096e-06, "loss": 0.0198, "step": 11775 }, { "epoch": 11.95, "learning_rate": 2.0843496894665206e-06, "loss": 0.0233, "step": 11776 }, { "epoch": 11.95, "learning_rate": 2.0830102802376794e-06, "loss": 0.024, "step": 11777 }, { "epoch": 11.95, "learning_rate": 2.0816712514669336e-06, "loss": 0.017, "step": 11778 }, { "epoch": 11.95, "learning_rate": 2.0803326032186345e-06, "loss": 0.0186, "step": 11779 }, { "epoch": 11.95, "learning_rate": 2.078994335557112e-06, "loss": 0.0249, "step": 11780 }, { "epoch": 11.95, "learning_rate": 2.0776564485466776e-06, "loss": 0.0226, "step": 11781 }, { "epoch": 11.96, "learning_rate": 2.076318942251627e-06, "loss": 0.0209, "step": 11782 }, { "epoch": 11.96, "learning_rate": 2.07498181673623e-06, "loss": 0.0217, "step": 11783 }, { "epoch": 11.96, "learning_rate": 2.073645072064747e-06, "loss": 0.0171, "step": 11784 }, { "epoch": 11.96, "learning_rate": 2.0723087083014182e-06, "loss": 0.0244, "step": 11785 }, { "epoch": 11.96, "learning_rate": 2.0709727255104583e-06, "loss": 0.0186, "step": 11786 }, { "epoch": 11.96, "learning_rate": 2.069637123756074e-06, "loss": 0.0247, "step": 11787 }, { "epoch": 11.96, "learning_rate": 2.068301903102443e-06, "loss": 0.0193, "step": 11788 }, { "epoch": 11.96, "learning_rate": 2.0669670636137397e-06, "loss": 0.019, "step": 11789 }, { "epoch": 11.96, "learning_rate": 2.065632605354103e-06, "loss": 0.0225, "step": 11790 }, { "epoch": 11.96, "learning_rate": 2.0642985283876637e-06, "loss": 0.0269, "step": 11791 }, { "epoch": 11.97, "learning_rate": 2.0629648327785367e-06, "loss": 0.0222, "step": 11792 }, { "epoch": 11.97, "learning_rate": 2.0616315185908064e-06, "loss": 0.0193, "step": 11793 }, { "epoch": 11.97, "learning_rate": 2.0602985858885505e-06, "loss": 0.0172, "step": 11794 }, { "epoch": 11.97, "learning_rate": 2.0589660347358263e-06, "loss": 0.0166, "step": 11795 }, { "epoch": 11.97, "learning_rate": 2.057633865196662e-06, "loss": 0.0258, "step": 11796 }, { "epoch": 11.97, "learning_rate": 2.0563020773350895e-06, "loss": 0.0208, "step": 11797 }, { "epoch": 11.97, "learning_rate": 2.0549706712150983e-06, "loss": 0.0204, "step": 11798 }, { "epoch": 11.97, "learning_rate": 2.053639646900677e-06, "loss": 0.0213, "step": 11799 }, { "epoch": 11.97, "learning_rate": 2.0523090044557824e-06, "loss": 0.0223, "step": 11800 }, { "epoch": 11.97, "learning_rate": 2.050978743944365e-06, "loss": 0.023, "step": 11801 }, { "epoch": 11.98, "learning_rate": 2.0496488654303524e-06, "loss": 0.0256, "step": 11802 }, { "epoch": 11.98, "learning_rate": 2.0483193689776483e-06, "loss": 0.0238, "step": 11803 }, { "epoch": 11.98, "learning_rate": 2.0469902546501487e-06, "loss": 0.0217, "step": 11804 }, { "epoch": 11.98, "learning_rate": 2.045661522511716e-06, "loss": 0.0234, "step": 11805 }, { "epoch": 11.98, "learning_rate": 2.0443331726262162e-06, "loss": 0.025, "step": 11806 }, { "epoch": 11.98, "learning_rate": 2.0430052050574745e-06, "loss": 0.0252, "step": 11807 }, { "epoch": 11.98, "learning_rate": 2.0416776198693113e-06, "loss": 0.0212, "step": 11808 }, { "epoch": 11.98, "learning_rate": 2.040350417125526e-06, "loss": 0.023, "step": 11809 }, { "epoch": 11.98, "learning_rate": 2.0390235968898965e-06, "loss": 0.0247, "step": 11810 }, { "epoch": 11.98, "learning_rate": 2.037697159226184e-06, "loss": 0.0212, "step": 11811 }, { "epoch": 11.99, "learning_rate": 2.036371104198136e-06, "loss": 0.0253, "step": 11812 }, { "epoch": 11.99, "learning_rate": 2.035045431869468e-06, "loss": 0.0254, "step": 11813 }, { "epoch": 11.99, "learning_rate": 2.0337201423038978e-06, "loss": 0.0216, "step": 11814 }, { "epoch": 11.99, "learning_rate": 2.0323952355651044e-06, "loss": 0.0254, "step": 11815 }, { "epoch": 11.99, "learning_rate": 2.031070711716765e-06, "loss": 0.0258, "step": 11816 }, { "epoch": 11.99, "learning_rate": 2.029746570822524e-06, "loss": 0.0249, "step": 11817 }, { "epoch": 11.99, "learning_rate": 2.028422812946016e-06, "loss": 0.0214, "step": 11818 }, { "epoch": 11.99, "learning_rate": 2.027099438150859e-06, "loss": 0.0232, "step": 11819 }, { "epoch": 11.99, "learning_rate": 2.025776446500645e-06, "loss": 0.0232, "step": 11820 }, { "epoch": 11.99, "learning_rate": 2.024453838058951e-06, "loss": 0.0223, "step": 11821 }, { "epoch": 12.0, "learning_rate": 2.0231316128893385e-06, "loss": 0.0155, "step": 11822 }, { "epoch": 12.0, "learning_rate": 2.0218097710553475e-06, "loss": 0.0305, "step": 11823 }, { "epoch": 12.0, "learning_rate": 2.0204883126205045e-06, "loss": 0.0204, "step": 11824 }, { "epoch": 12.0, "learning_rate": 2.019167237648305e-06, "loss": 0.0173, "step": 11825 }, { "epoch": 12.0, "learning_rate": 2.0178465462022423e-06, "loss": 0.0214, "step": 11826 }, { "epoch": 12.0, "learning_rate": 2.0165262383457772e-06, "loss": 0.0169, "step": 11827 }, { "epoch": 12.0, "learning_rate": 2.0152063141423617e-06, "loss": 0.0193, "step": 11828 }, { "epoch": 12.0, "learning_rate": 2.013886773655427e-06, "loss": 0.0185, "step": 11829 }, { "epoch": 12.0, "learning_rate": 2.0125676169483777e-06, "loss": 0.0176, "step": 11830 }, { "epoch": 12.01, "learning_rate": 2.011248844084618e-06, "loss": 0.0248, "step": 11831 }, { "epoch": 12.01, "learning_rate": 2.0099304551275145e-06, "loss": 0.0174, "step": 11832 }, { "epoch": 12.01, "learning_rate": 2.0086124501404294e-06, "loss": 0.0201, "step": 11833 }, { "epoch": 12.01, "learning_rate": 2.0072948291866945e-06, "loss": 0.0178, "step": 11834 }, { "epoch": 12.01, "learning_rate": 2.005977592329632e-06, "loss": 0.0163, "step": 11835 }, { "epoch": 12.01, "learning_rate": 2.0046607396325456e-06, "loss": 0.0136, "step": 11836 }, { "epoch": 12.01, "learning_rate": 2.0033442711587138e-06, "loss": 0.0177, "step": 11837 }, { "epoch": 12.01, "learning_rate": 2.0020281869714007e-06, "loss": 0.0168, "step": 11838 }, { "epoch": 12.01, "learning_rate": 2.0007124871338548e-06, "loss": 0.0144, "step": 11839 }, { "epoch": 12.01, "learning_rate": 1.999397171709302e-06, "loss": 0.0178, "step": 11840 }, { "epoch": 12.02, "learning_rate": 1.9980822407609523e-06, "loss": 0.0153, "step": 11841 }, { "epoch": 12.02, "learning_rate": 1.996767694351993e-06, "loss": 0.0205, "step": 11842 }, { "epoch": 12.02, "learning_rate": 1.995453532545599e-06, "loss": 0.018, "step": 11843 }, { "epoch": 12.02, "learning_rate": 1.9941397554049203e-06, "loss": 0.0169, "step": 11844 }, { "epoch": 12.02, "learning_rate": 1.9928263629930923e-06, "loss": 0.0161, "step": 11845 }, { "epoch": 12.02, "learning_rate": 1.991513355373236e-06, "loss": 0.0165, "step": 11846 }, { "epoch": 12.02, "learning_rate": 1.9902007326084383e-06, "loss": 0.0211, "step": 11847 }, { "epoch": 12.02, "learning_rate": 1.9888884947617927e-06, "loss": 0.0197, "step": 11848 }, { "epoch": 12.02, "learning_rate": 1.987576641896348e-06, "loss": 0.0148, "step": 11849 }, { "epoch": 12.02, "learning_rate": 1.9862651740751526e-06, "loss": 0.0189, "step": 11850 }, { "epoch": 12.03, "learning_rate": 1.984954091361232e-06, "loss": 0.0174, "step": 11851 }, { "epoch": 12.03, "learning_rate": 1.9836433938175847e-06, "loss": 0.0177, "step": 11852 }, { "epoch": 12.03, "learning_rate": 1.982333081507205e-06, "loss": 0.0185, "step": 11853 }, { "epoch": 12.03, "learning_rate": 1.9810231544930537e-06, "loss": 0.0195, "step": 11854 }, { "epoch": 12.03, "learning_rate": 1.9797136128380846e-06, "loss": 0.0189, "step": 11855 }, { "epoch": 12.03, "learning_rate": 1.9784044566052275e-06, "loss": 0.0185, "step": 11856 }, { "epoch": 12.03, "learning_rate": 1.9770956858573964e-06, "loss": 0.0185, "step": 11857 }, { "epoch": 12.03, "learning_rate": 1.975787300657488e-06, "loss": 0.0171, "step": 11858 }, { "epoch": 12.03, "learning_rate": 1.9744793010683726e-06, "loss": 0.02, "step": 11859 }, { "epoch": 12.03, "learning_rate": 1.973171687152912e-06, "loss": 0.0209, "step": 11860 }, { "epoch": 12.04, "learning_rate": 1.97186445897394e-06, "loss": 0.0139, "step": 11861 }, { "epoch": 12.04, "learning_rate": 1.9705576165942785e-06, "loss": 0.0176, "step": 11862 }, { "epoch": 12.04, "learning_rate": 1.9692511600767304e-06, "loss": 0.0201, "step": 11863 }, { "epoch": 12.04, "learning_rate": 1.967945089484078e-06, "loss": 0.0193, "step": 11864 }, { "epoch": 12.04, "learning_rate": 1.966639404879087e-06, "loss": 0.0187, "step": 11865 }, { "epoch": 12.04, "learning_rate": 1.9653341063245003e-06, "loss": 0.0167, "step": 11866 }, { "epoch": 12.04, "learning_rate": 1.964029193883046e-06, "loss": 0.0221, "step": 11867 }, { "epoch": 12.04, "learning_rate": 1.9627246676174363e-06, "loss": 0.0211, "step": 11868 }, { "epoch": 12.04, "learning_rate": 1.9614205275903566e-06, "loss": 0.0172, "step": 11869 }, { "epoch": 12.04, "learning_rate": 1.960116773864482e-06, "loss": 0.0181, "step": 11870 }, { "epoch": 12.05, "learning_rate": 1.9588134065024632e-06, "loss": 0.0201, "step": 11871 }, { "epoch": 12.05, "learning_rate": 1.9575104255669343e-06, "loss": 0.0176, "step": 11872 }, { "epoch": 12.05, "learning_rate": 1.9562078311205135e-06, "loss": 0.0178, "step": 11873 }, { "epoch": 12.05, "learning_rate": 1.954905623225797e-06, "loss": 0.0166, "step": 11874 }, { "epoch": 12.05, "learning_rate": 1.953603801945365e-06, "loss": 0.0212, "step": 11875 }, { "epoch": 12.05, "learning_rate": 1.952302367341775e-06, "loss": 0.0154, "step": 11876 }, { "epoch": 12.05, "learning_rate": 1.9510013194775735e-06, "loss": 0.0205, "step": 11877 }, { "epoch": 12.05, "learning_rate": 1.9497006584152767e-06, "loss": 0.0188, "step": 11878 }, { "epoch": 12.05, "learning_rate": 1.9484003842173927e-06, "loss": 0.0175, "step": 11879 }, { "epoch": 12.05, "learning_rate": 1.947100496946407e-06, "loss": 0.0191, "step": 11880 }, { "epoch": 12.06, "learning_rate": 1.945800996664786e-06, "loss": 0.017, "step": 11881 }, { "epoch": 12.06, "learning_rate": 1.944501883434984e-06, "loss": 0.0176, "step": 11882 }, { "epoch": 12.06, "learning_rate": 1.9432031573194223e-06, "loss": 0.0163, "step": 11883 }, { "epoch": 12.06, "learning_rate": 1.9419048183805167e-06, "loss": 0.0129, "step": 11884 }, { "epoch": 12.06, "learning_rate": 1.940606866680663e-06, "loss": 0.0191, "step": 11885 }, { "epoch": 12.06, "learning_rate": 1.9393093022822296e-06, "loss": 0.0163, "step": 11886 }, { "epoch": 12.06, "learning_rate": 1.938012125247577e-06, "loss": 0.0181, "step": 11887 }, { "epoch": 12.06, "learning_rate": 1.9367153356390345e-06, "loss": 0.0158, "step": 11888 }, { "epoch": 12.06, "learning_rate": 1.935418933518931e-06, "loss": 0.0182, "step": 11889 }, { "epoch": 12.06, "learning_rate": 1.93412291894956e-06, "loss": 0.0207, "step": 11890 }, { "epoch": 12.07, "learning_rate": 1.9328272919932032e-06, "loss": 0.0164, "step": 11891 }, { "epoch": 12.07, "learning_rate": 1.9315320527121263e-06, "loss": 0.0195, "step": 11892 }, { "epoch": 12.07, "learning_rate": 1.9302372011685688e-06, "loss": 0.0185, "step": 11893 }, { "epoch": 12.07, "learning_rate": 1.928942737424757e-06, "loss": 0.0158, "step": 11894 }, { "epoch": 12.07, "learning_rate": 1.9276486615429013e-06, "loss": 0.0209, "step": 11895 }, { "epoch": 12.07, "learning_rate": 1.926354973585185e-06, "loss": 0.0206, "step": 11896 }, { "epoch": 12.07, "learning_rate": 1.9250616736137785e-06, "loss": 0.0191, "step": 11897 }, { "epoch": 12.07, "learning_rate": 1.923768761690833e-06, "loss": 0.0212, "step": 11898 }, { "epoch": 12.07, "learning_rate": 1.9224762378784844e-06, "loss": 0.0215, "step": 11899 }, { "epoch": 12.08, "learning_rate": 1.9211841022388388e-06, "loss": 0.0181, "step": 11900 }, { "epoch": 12.08, "learning_rate": 1.9198923548339955e-06, "loss": 0.0183, "step": 11901 }, { "epoch": 12.08, "learning_rate": 1.9186009957260322e-06, "loss": 0.0187, "step": 11902 }, { "epoch": 12.08, "learning_rate": 1.9173100249770015e-06, "loss": 0.0176, "step": 11903 }, { "epoch": 12.08, "learning_rate": 1.9160194426489466e-06, "loss": 0.0174, "step": 11904 }, { "epoch": 12.08, "learning_rate": 1.9147292488038803e-06, "loss": 0.0165, "step": 11905 }, { "epoch": 12.08, "learning_rate": 1.913439443503815e-06, "loss": 0.0187, "step": 11906 }, { "epoch": 12.08, "learning_rate": 1.912150026810725e-06, "loss": 0.0228, "step": 11907 }, { "epoch": 12.08, "learning_rate": 1.9108609987865768e-06, "loss": 0.0137, "step": 11908 }, { "epoch": 12.08, "learning_rate": 1.9095723594933193e-06, "loss": 0.016, "step": 11909 }, { "epoch": 12.09, "learning_rate": 1.908284108992873e-06, "loss": 0.0226, "step": 11910 }, { "epoch": 12.09, "learning_rate": 1.9069962473471492e-06, "loss": 0.0193, "step": 11911 }, { "epoch": 12.09, "learning_rate": 1.9057087746180392e-06, "loss": 0.0233, "step": 11912 }, { "epoch": 12.09, "learning_rate": 1.9044216908674073e-06, "loss": 0.0175, "step": 11913 }, { "epoch": 12.09, "learning_rate": 1.9031349961571144e-06, "loss": 0.0175, "step": 11914 }, { "epoch": 12.09, "learning_rate": 1.9018486905489864e-06, "loss": 0.0189, "step": 11915 }, { "epoch": 12.09, "learning_rate": 1.9005627741048427e-06, "loss": 0.0162, "step": 11916 }, { "epoch": 12.09, "learning_rate": 1.899277246886475e-06, "loss": 0.0215, "step": 11917 }, { "epoch": 12.09, "learning_rate": 1.897992108955663e-06, "loss": 0.0185, "step": 11918 }, { "epoch": 12.09, "learning_rate": 1.896707360374167e-06, "loss": 0.0191, "step": 11919 }, { "epoch": 12.1, "learning_rate": 1.8954230012037212e-06, "loss": 0.02, "step": 11920 }, { "epoch": 12.1, "learning_rate": 1.8941390315060527e-06, "loss": 0.0182, "step": 11921 }, { "epoch": 12.1, "learning_rate": 1.8928554513428554e-06, "loss": 0.0149, "step": 11922 }, { "epoch": 12.1, "learning_rate": 1.8915722607758224e-06, "loss": 0.0183, "step": 11923 }, { "epoch": 12.1, "learning_rate": 1.8902894598666167e-06, "loss": 0.0157, "step": 11924 }, { "epoch": 12.1, "learning_rate": 1.8890070486768786e-06, "loss": 0.0199, "step": 11925 }, { "epoch": 12.1, "learning_rate": 1.8877250272682435e-06, "loss": 0.0162, "step": 11926 }, { "epoch": 12.1, "learning_rate": 1.8864433957023132e-06, "loss": 0.0174, "step": 11927 }, { "epoch": 12.1, "learning_rate": 1.8851621540406795e-06, "loss": 0.0173, "step": 11928 }, { "epoch": 12.1, "learning_rate": 1.8838813023449176e-06, "loss": 0.0164, "step": 11929 }, { "epoch": 12.11, "learning_rate": 1.8826008406765717e-06, "loss": 0.0202, "step": 11930 }, { "epoch": 12.11, "learning_rate": 1.8813207690971868e-06, "loss": 0.02, "step": 11931 }, { "epoch": 12.11, "learning_rate": 1.8800410876682684e-06, "loss": 0.0173, "step": 11932 }, { "epoch": 12.11, "learning_rate": 1.8787617964513193e-06, "loss": 0.0189, "step": 11933 }, { "epoch": 12.11, "learning_rate": 1.8774828955078106e-06, "loss": 0.0169, "step": 11934 }, { "epoch": 12.11, "learning_rate": 1.876204384899205e-06, "loss": 0.0191, "step": 11935 }, { "epoch": 12.11, "learning_rate": 1.8749262646869437e-06, "loss": 0.0168, "step": 11936 }, { "epoch": 12.11, "learning_rate": 1.873648534932444e-06, "loss": 0.0183, "step": 11937 }, { "epoch": 12.11, "learning_rate": 1.8723711956971114e-06, "loss": 0.0223, "step": 11938 }, { "epoch": 12.11, "learning_rate": 1.8710942470423277e-06, "loss": 0.0176, "step": 11939 }, { "epoch": 12.12, "learning_rate": 1.8698176890294583e-06, "loss": 0.0201, "step": 11940 }, { "epoch": 12.12, "learning_rate": 1.8685415217198532e-06, "loss": 0.0217, "step": 11941 }, { "epoch": 12.12, "learning_rate": 1.8672657451748334e-06, "loss": 0.019, "step": 11942 }, { "epoch": 12.12, "learning_rate": 1.8659903594557127e-06, "loss": 0.0156, "step": 11943 }, { "epoch": 12.12, "learning_rate": 1.864715364623777e-06, "loss": 0.0203, "step": 11944 }, { "epoch": 12.12, "learning_rate": 1.8634407607402982e-06, "loss": 0.0177, "step": 11945 }, { "epoch": 12.12, "learning_rate": 1.8621665478665319e-06, "loss": 0.0182, "step": 11946 }, { "epoch": 12.12, "learning_rate": 1.8608927260637044e-06, "loss": 0.0199, "step": 11947 }, { "epoch": 12.12, "learning_rate": 1.8596192953930391e-06, "loss": 0.018, "step": 11948 }, { "epoch": 12.12, "learning_rate": 1.858346255915726e-06, "loss": 0.0183, "step": 11949 }, { "epoch": 12.13, "learning_rate": 1.8570736076929463e-06, "loss": 0.018, "step": 11950 }, { "epoch": 12.13, "learning_rate": 1.8558013507858542e-06, "loss": 0.0199, "step": 11951 }, { "epoch": 12.13, "learning_rate": 1.8545294852555896e-06, "loss": 0.016, "step": 11952 }, { "epoch": 12.13, "learning_rate": 1.853258011163277e-06, "loss": 0.019, "step": 11953 }, { "epoch": 12.13, "learning_rate": 1.8519869285700132e-06, "loss": 0.0171, "step": 11954 }, { "epoch": 12.13, "learning_rate": 1.8507162375368836e-06, "loss": 0.0185, "step": 11955 }, { "epoch": 12.13, "learning_rate": 1.8494459381249518e-06, "loss": 0.0191, "step": 11956 }, { "epoch": 12.13, "learning_rate": 1.8481760303952646e-06, "loss": 0.0181, "step": 11957 }, { "epoch": 12.13, "learning_rate": 1.8469065144088505e-06, "loss": 0.0179, "step": 11958 }, { "epoch": 12.13, "learning_rate": 1.8456373902267122e-06, "loss": 0.0213, "step": 11959 }, { "epoch": 12.14, "learning_rate": 1.844368657909842e-06, "loss": 0.017, "step": 11960 }, { "epoch": 12.14, "learning_rate": 1.8431003175192074e-06, "loss": 0.0204, "step": 11961 }, { "epoch": 12.14, "learning_rate": 1.8418323691157613e-06, "loss": 0.018, "step": 11962 }, { "epoch": 12.14, "learning_rate": 1.840564812760438e-06, "loss": 0.0188, "step": 11963 }, { "epoch": 12.14, "learning_rate": 1.8392976485141433e-06, "loss": 0.0195, "step": 11964 }, { "epoch": 12.14, "learning_rate": 1.8380308764377841e-06, "loss": 0.0177, "step": 11965 }, { "epoch": 12.14, "learning_rate": 1.836764496592226e-06, "loss": 0.0164, "step": 11966 }, { "epoch": 12.14, "learning_rate": 1.8354985090383315e-06, "loss": 0.0183, "step": 11967 }, { "epoch": 12.14, "learning_rate": 1.8342329138369387e-06, "loss": 0.018, "step": 11968 }, { "epoch": 12.15, "learning_rate": 1.8329677110488619e-06, "loss": 0.0196, "step": 11969 }, { "epoch": 12.15, "learning_rate": 1.8317029007349086e-06, "loss": 0.0169, "step": 11970 }, { "epoch": 12.15, "learning_rate": 1.8304384829558542e-06, "loss": 0.0215, "step": 11971 }, { "epoch": 12.15, "learning_rate": 1.8291744577724635e-06, "loss": 0.02, "step": 11972 }, { "epoch": 12.15, "learning_rate": 1.8279108252454813e-06, "loss": 0.0229, "step": 11973 }, { "epoch": 12.15, "learning_rate": 1.8266475854356326e-06, "loss": 0.0167, "step": 11974 }, { "epoch": 12.15, "learning_rate": 1.825384738403625e-06, "loss": 0.0196, "step": 11975 }, { "epoch": 12.15, "learning_rate": 1.8241222842101414e-06, "loss": 0.0198, "step": 11976 }, { "epoch": 12.15, "learning_rate": 1.8228602229158555e-06, "loss": 0.0193, "step": 11977 }, { "epoch": 12.15, "learning_rate": 1.8215985545814107e-06, "loss": 0.0208, "step": 11978 }, { "epoch": 12.16, "learning_rate": 1.8203372792674424e-06, "loss": 0.0167, "step": 11979 }, { "epoch": 12.16, "learning_rate": 1.8190763970345593e-06, "loss": 0.0168, "step": 11980 }, { "epoch": 12.16, "learning_rate": 1.8178159079433567e-06, "loss": 0.015, "step": 11981 }, { "epoch": 12.16, "learning_rate": 1.8165558120544101e-06, "loss": 0.0187, "step": 11982 }, { "epoch": 12.16, "learning_rate": 1.8152961094282696e-06, "loss": 0.0187, "step": 11983 }, { "epoch": 12.16, "learning_rate": 1.8140368001254727e-06, "loss": 0.0169, "step": 11984 }, { "epoch": 12.16, "learning_rate": 1.8127778842065414e-06, "loss": 0.0143, "step": 11985 }, { "epoch": 12.16, "learning_rate": 1.811519361731967e-06, "loss": 0.0166, "step": 11986 }, { "epoch": 12.16, "learning_rate": 1.8102612327622359e-06, "loss": 0.0181, "step": 11987 }, { "epoch": 12.16, "learning_rate": 1.8090034973578018e-06, "loss": 0.0207, "step": 11988 }, { "epoch": 12.17, "learning_rate": 1.8077461555791098e-06, "loss": 0.0172, "step": 11989 }, { "epoch": 12.17, "learning_rate": 1.8064892074865814e-06, "loss": 0.0171, "step": 11990 }, { "epoch": 12.17, "learning_rate": 1.8052326531406228e-06, "loss": 0.0194, "step": 11991 }, { "epoch": 12.17, "learning_rate": 1.803976492601619e-06, "loss": 0.0191, "step": 11992 }, { "epoch": 12.17, "learning_rate": 1.8027207259299317e-06, "loss": 0.0222, "step": 11993 }, { "epoch": 12.17, "learning_rate": 1.8014653531859117e-06, "loss": 0.0203, "step": 11994 }, { "epoch": 12.17, "learning_rate": 1.800210374429887e-06, "loss": 0.0189, "step": 11995 }, { "epoch": 12.17, "learning_rate": 1.7989557897221632e-06, "loss": 0.0209, "step": 11996 }, { "epoch": 12.17, "learning_rate": 1.797701599123034e-06, "loss": 0.0176, "step": 11997 }, { "epoch": 12.17, "learning_rate": 1.796447802692769e-06, "loss": 0.0188, "step": 11998 }, { "epoch": 12.18, "learning_rate": 1.7951944004916232e-06, "loss": 0.0211, "step": 11999 }, { "epoch": 12.18, "learning_rate": 1.7939413925798255e-06, "loss": 0.0168, "step": 12000 }, { "epoch": 12.18, "learning_rate": 1.792688779017593e-06, "loss": 0.0181, "step": 12001 }, { "epoch": 12.18, "learning_rate": 1.791436559865124e-06, "loss": 0.0194, "step": 12002 }, { "epoch": 12.18, "learning_rate": 1.7901847351825886e-06, "loss": 0.0189, "step": 12003 }, { "epoch": 12.18, "learning_rate": 1.7889333050301505e-06, "loss": 0.0176, "step": 12004 }, { "epoch": 12.18, "learning_rate": 1.7876822694679398e-06, "loss": 0.0145, "step": 12005 }, { "epoch": 12.18, "learning_rate": 1.7864316285560878e-06, "loss": 0.0175, "step": 12006 }, { "epoch": 12.18, "learning_rate": 1.785181382354686e-06, "loss": 0.0198, "step": 12007 }, { "epoch": 12.18, "learning_rate": 1.7839315309238192e-06, "loss": 0.0196, "step": 12008 }, { "epoch": 12.19, "learning_rate": 1.782682074323554e-06, "loss": 0.0211, "step": 12009 }, { "epoch": 12.19, "learning_rate": 1.7814330126139268e-06, "loss": 0.0163, "step": 12010 }, { "epoch": 12.19, "learning_rate": 1.780184345854966e-06, "loss": 0.0184, "step": 12011 }, { "epoch": 12.19, "learning_rate": 1.7789360741066809e-06, "loss": 0.0154, "step": 12012 }, { "epoch": 12.19, "learning_rate": 1.7776881974290506e-06, "loss": 0.0173, "step": 12013 }, { "epoch": 12.19, "learning_rate": 1.7764407158820485e-06, "loss": 0.0176, "step": 12014 }, { "epoch": 12.19, "learning_rate": 1.7751936295256223e-06, "loss": 0.0187, "step": 12015 }, { "epoch": 12.19, "learning_rate": 1.7739469384197039e-06, "loss": 0.0199, "step": 12016 }, { "epoch": 12.19, "learning_rate": 1.7727006426241989e-06, "loss": 0.0189, "step": 12017 }, { "epoch": 12.19, "learning_rate": 1.7714547421990037e-06, "loss": 0.0184, "step": 12018 }, { "epoch": 12.2, "learning_rate": 1.7702092372039903e-06, "loss": 0.0187, "step": 12019 }, { "epoch": 12.2, "learning_rate": 1.768964127699011e-06, "loss": 0.0191, "step": 12020 }, { "epoch": 12.2, "learning_rate": 1.7677194137439036e-06, "loss": 0.0211, "step": 12021 }, { "epoch": 12.2, "learning_rate": 1.7664750953984766e-06, "loss": 0.0182, "step": 12022 }, { "epoch": 12.2, "learning_rate": 1.765231172722538e-06, "loss": 0.0187, "step": 12023 }, { "epoch": 12.2, "learning_rate": 1.7639876457758576e-06, "loss": 0.0207, "step": 12024 }, { "epoch": 12.2, "learning_rate": 1.7627445146181953e-06, "loss": 0.0166, "step": 12025 }, { "epoch": 12.2, "learning_rate": 1.761501779309296e-06, "loss": 0.0235, "step": 12026 }, { "epoch": 12.2, "learning_rate": 1.7602594399088736e-06, "loss": 0.0184, "step": 12027 }, { "epoch": 12.2, "learning_rate": 1.7590174964766315e-06, "loss": 0.0186, "step": 12028 }, { "epoch": 12.21, "learning_rate": 1.7577759490722579e-06, "loss": 0.0183, "step": 12029 }, { "epoch": 12.21, "learning_rate": 1.756534797755406e-06, "loss": 0.0194, "step": 12030 }, { "epoch": 12.21, "learning_rate": 1.7552940425857323e-06, "loss": 0.017, "step": 12031 }, { "epoch": 12.21, "learning_rate": 1.754053683622855e-06, "loss": 0.0194, "step": 12032 }, { "epoch": 12.21, "learning_rate": 1.752813720926384e-06, "loss": 0.019, "step": 12033 }, { "epoch": 12.21, "learning_rate": 1.751574154555904e-06, "loss": 0.0185, "step": 12034 }, { "epoch": 12.21, "learning_rate": 1.750334984570984e-06, "loss": 0.0191, "step": 12035 }, { "epoch": 12.21, "learning_rate": 1.7490962110311772e-06, "loss": 0.0178, "step": 12036 }, { "epoch": 12.21, "learning_rate": 1.7478578339960083e-06, "loss": 0.016, "step": 12037 }, { "epoch": 12.22, "learning_rate": 1.7466198535249913e-06, "loss": 0.0233, "step": 12038 }, { "epoch": 12.22, "learning_rate": 1.7453822696776179e-06, "loss": 0.021, "step": 12039 }, { "epoch": 12.22, "learning_rate": 1.7441450825133632e-06, "loss": 0.0208, "step": 12040 }, { "epoch": 12.22, "learning_rate": 1.7429082920916819e-06, "loss": 0.021, "step": 12041 }, { "epoch": 12.22, "learning_rate": 1.741671898472005e-06, "loss": 0.0189, "step": 12042 }, { "epoch": 12.22, "learning_rate": 1.740435901713753e-06, "loss": 0.0181, "step": 12043 }, { "epoch": 12.22, "learning_rate": 1.7392003018763193e-06, "loss": 0.0222, "step": 12044 }, { "epoch": 12.22, "learning_rate": 1.7379650990190822e-06, "loss": 0.0236, "step": 12045 }, { "epoch": 12.22, "learning_rate": 1.736730293201403e-06, "loss": 0.0161, "step": 12046 }, { "epoch": 12.22, "learning_rate": 1.7354958844826163e-06, "loss": 0.0213, "step": 12047 }, { "epoch": 12.23, "learning_rate": 1.7342618729220507e-06, "loss": 0.0171, "step": 12048 }, { "epoch": 12.23, "learning_rate": 1.7330282585790004e-06, "loss": 0.0182, "step": 12049 }, { "epoch": 12.23, "learning_rate": 1.7317950415127537e-06, "loss": 0.0189, "step": 12050 }, { "epoch": 12.23, "learning_rate": 1.7305622217825692e-06, "loss": 0.0183, "step": 12051 }, { "epoch": 12.23, "learning_rate": 1.7293297994476922e-06, "loss": 0.0166, "step": 12052 }, { "epoch": 12.23, "learning_rate": 1.7280977745673499e-06, "loss": 0.0191, "step": 12053 }, { "epoch": 12.23, "learning_rate": 1.7268661472007464e-06, "loss": 0.0197, "step": 12054 }, { "epoch": 12.23, "learning_rate": 1.7256349174070685e-06, "loss": 0.0183, "step": 12055 }, { "epoch": 12.23, "learning_rate": 1.7244040852454857e-06, "loss": 0.021, "step": 12056 }, { "epoch": 12.23, "learning_rate": 1.723173650775145e-06, "loss": 0.0191, "step": 12057 }, { "epoch": 12.24, "learning_rate": 1.7219436140551804e-06, "loss": 0.0213, "step": 12058 }, { "epoch": 12.24, "learning_rate": 1.7207139751446955e-06, "loss": 0.0187, "step": 12059 }, { "epoch": 12.24, "learning_rate": 1.7194847341027886e-06, "loss": 0.02, "step": 12060 }, { "epoch": 12.24, "learning_rate": 1.718255890988526e-06, "loss": 0.0183, "step": 12061 }, { "epoch": 12.24, "learning_rate": 1.7170274458609636e-06, "loss": 0.0182, "step": 12062 }, { "epoch": 12.24, "learning_rate": 1.7157993987791376e-06, "loss": 0.0173, "step": 12063 }, { "epoch": 12.24, "learning_rate": 1.7145717498020565e-06, "loss": 0.0197, "step": 12064 }, { "epoch": 12.24, "learning_rate": 1.7133444989887238e-06, "loss": 0.0178, "step": 12065 }, { "epoch": 12.24, "learning_rate": 1.7121176463981116e-06, "loss": 0.0171, "step": 12066 }, { "epoch": 12.24, "learning_rate": 1.710891192089177e-06, "loss": 0.0194, "step": 12067 }, { "epoch": 12.25, "learning_rate": 1.7096651361208628e-06, "loss": 0.0187, "step": 12068 }, { "epoch": 12.25, "learning_rate": 1.708439478552083e-06, "loss": 0.022, "step": 12069 }, { "epoch": 12.25, "learning_rate": 1.7072142194417418e-06, "loss": 0.0174, "step": 12070 }, { "epoch": 12.25, "learning_rate": 1.7059893588487164e-06, "loss": 0.0151, "step": 12071 }, { "epoch": 12.25, "learning_rate": 1.7047648968318697e-06, "loss": 0.0205, "step": 12072 }, { "epoch": 12.25, "learning_rate": 1.7035408334500458e-06, "loss": 0.0163, "step": 12073 }, { "epoch": 12.25, "learning_rate": 1.7023171687620677e-06, "loss": 0.0219, "step": 12074 }, { "epoch": 12.25, "learning_rate": 1.7010939028267404e-06, "loss": 0.0169, "step": 12075 }, { "epoch": 12.25, "learning_rate": 1.6998710357028469e-06, "loss": 0.0188, "step": 12076 }, { "epoch": 12.25, "learning_rate": 1.698648567449156e-06, "loss": 0.0182, "step": 12077 }, { "epoch": 12.26, "learning_rate": 1.6974264981244116e-06, "loss": 0.0211, "step": 12078 }, { "epoch": 12.26, "learning_rate": 1.696204827787341e-06, "loss": 0.0164, "step": 12079 }, { "epoch": 12.26, "learning_rate": 1.6949835564966577e-06, "loss": 0.019, "step": 12080 }, { "epoch": 12.26, "learning_rate": 1.693762684311042e-06, "loss": 0.0169, "step": 12081 }, { "epoch": 12.26, "learning_rate": 1.6925422112891754e-06, "loss": 0.0163, "step": 12082 }, { "epoch": 12.26, "learning_rate": 1.6913221374897004e-06, "loss": 0.0153, "step": 12083 }, { "epoch": 12.26, "learning_rate": 1.6901024629712516e-06, "loss": 0.0229, "step": 12084 }, { "epoch": 12.26, "learning_rate": 1.6888831877924427e-06, "loss": 0.0205, "step": 12085 }, { "epoch": 12.26, "learning_rate": 1.687664312011864e-06, "loss": 0.021, "step": 12086 }, { "epoch": 12.26, "learning_rate": 1.6864458356880942e-06, "loss": 0.0219, "step": 12087 }, { "epoch": 12.27, "learning_rate": 1.6852277588796828e-06, "loss": 0.018, "step": 12088 }, { "epoch": 12.27, "learning_rate": 1.6840100816451676e-06, "loss": 0.0166, "step": 12089 }, { "epoch": 12.27, "learning_rate": 1.6827928040430663e-06, "loss": 0.0177, "step": 12090 }, { "epoch": 12.27, "learning_rate": 1.6815759261318765e-06, "loss": 0.0184, "step": 12091 }, { "epoch": 12.27, "learning_rate": 1.680359447970077e-06, "loss": 0.0218, "step": 12092 }, { "epoch": 12.27, "learning_rate": 1.6791433696161242e-06, "loss": 0.0208, "step": 12093 }, { "epoch": 12.27, "learning_rate": 1.6779276911284604e-06, "loss": 0.0179, "step": 12094 }, { "epoch": 12.27, "learning_rate": 1.6767124125655032e-06, "loss": 0.0197, "step": 12095 }, { "epoch": 12.27, "learning_rate": 1.675497533985655e-06, "loss": 0.0184, "step": 12096 }, { "epoch": 12.27, "learning_rate": 1.6742830554472989e-06, "loss": 0.0204, "step": 12097 }, { "epoch": 12.28, "learning_rate": 1.673068977008796e-06, "loss": 0.0209, "step": 12098 }, { "epoch": 12.28, "learning_rate": 1.6718552987284952e-06, "loss": 0.022, "step": 12099 }, { "epoch": 12.28, "learning_rate": 1.6706420206647135e-06, "loss": 0.0195, "step": 12100 }, { "epoch": 12.28, "learning_rate": 1.6694291428757604e-06, "loss": 0.0187, "step": 12101 }, { "epoch": 12.28, "learning_rate": 1.6682166654199227e-06, "loss": 0.0197, "step": 12102 }, { "epoch": 12.28, "learning_rate": 1.667004588355463e-06, "loss": 0.0196, "step": 12103 }, { "epoch": 12.28, "learning_rate": 1.6657929117406336e-06, "loss": 0.0193, "step": 12104 }, { "epoch": 12.28, "learning_rate": 1.6645816356336574e-06, "loss": 0.0214, "step": 12105 }, { "epoch": 12.28, "learning_rate": 1.6633707600927461e-06, "loss": 0.017, "step": 12106 }, { "epoch": 12.29, "learning_rate": 1.6621602851760899e-06, "loss": 0.0184, "step": 12107 }, { "epoch": 12.29, "learning_rate": 1.6609502109418596e-06, "loss": 0.0221, "step": 12108 }, { "epoch": 12.29, "learning_rate": 1.6597405374482068e-06, "loss": 0.0182, "step": 12109 }, { "epoch": 12.29, "learning_rate": 1.6585312647532603e-06, "loss": 0.0168, "step": 12110 }, { "epoch": 12.29, "learning_rate": 1.657322392915135e-06, "loss": 0.0162, "step": 12111 }, { "epoch": 12.29, "learning_rate": 1.6561139219919265e-06, "loss": 0.0169, "step": 12112 }, { "epoch": 12.29, "learning_rate": 1.654905852041705e-06, "loss": 0.0206, "step": 12113 }, { "epoch": 12.29, "learning_rate": 1.6536981831225263e-06, "loss": 0.019, "step": 12114 }, { "epoch": 12.29, "learning_rate": 1.6524909152924263e-06, "loss": 0.0185, "step": 12115 }, { "epoch": 12.29, "learning_rate": 1.651284048609425e-06, "loss": 0.017, "step": 12116 }, { "epoch": 12.3, "learning_rate": 1.650077583131513e-06, "loss": 0.0195, "step": 12117 }, { "epoch": 12.3, "learning_rate": 1.6488715189166716e-06, "loss": 0.0199, "step": 12118 }, { "epoch": 12.3, "learning_rate": 1.647665856022862e-06, "loss": 0.0167, "step": 12119 }, { "epoch": 12.3, "learning_rate": 1.646460594508017e-06, "loss": 0.0189, "step": 12120 }, { "epoch": 12.3, "learning_rate": 1.6452557344300635e-06, "loss": 0.0181, "step": 12121 }, { "epoch": 12.3, "learning_rate": 1.644051275846893e-06, "loss": 0.0234, "step": 12122 }, { "epoch": 12.3, "learning_rate": 1.642847218816398e-06, "loss": 0.0172, "step": 12123 }, { "epoch": 12.3, "learning_rate": 1.6416435633964322e-06, "loss": 0.0205, "step": 12124 }, { "epoch": 12.3, "learning_rate": 1.6404403096448407e-06, "loss": 0.0181, "step": 12125 }, { "epoch": 12.3, "learning_rate": 1.6392374576194504e-06, "loss": 0.0216, "step": 12126 }, { "epoch": 12.31, "learning_rate": 1.6380350073780605e-06, "loss": 0.0186, "step": 12127 }, { "epoch": 12.31, "learning_rate": 1.6368329589784572e-06, "loss": 0.0214, "step": 12128 }, { "epoch": 12.31, "learning_rate": 1.6356313124784096e-06, "loss": 0.0219, "step": 12129 }, { "epoch": 12.31, "learning_rate": 1.6344300679356572e-06, "loss": 0.0219, "step": 12130 }, { "epoch": 12.31, "learning_rate": 1.6332292254079318e-06, "loss": 0.0165, "step": 12131 }, { "epoch": 12.31, "learning_rate": 1.6320287849529392e-06, "loss": 0.0219, "step": 12132 }, { "epoch": 12.31, "learning_rate": 1.6308287466283712e-06, "loss": 0.0206, "step": 12133 }, { "epoch": 12.31, "learning_rate": 1.6296291104918915e-06, "loss": 0.0181, "step": 12134 }, { "epoch": 12.31, "learning_rate": 1.6284298766011508e-06, "loss": 0.0193, "step": 12135 }, { "epoch": 12.31, "learning_rate": 1.6272310450137841e-06, "loss": 0.0205, "step": 12136 }, { "epoch": 12.32, "learning_rate": 1.6260326157873963e-06, "loss": 0.0236, "step": 12137 }, { "epoch": 12.32, "learning_rate": 1.6248345889795836e-06, "loss": 0.0173, "step": 12138 }, { "epoch": 12.32, "learning_rate": 1.623636964647911e-06, "loss": 0.0188, "step": 12139 }, { "epoch": 12.32, "learning_rate": 1.6224397428499394e-06, "loss": 0.0195, "step": 12140 }, { "epoch": 12.32, "learning_rate": 1.6212429236432014e-06, "loss": 0.0202, "step": 12141 }, { "epoch": 12.32, "learning_rate": 1.620046507085208e-06, "loss": 0.0221, "step": 12142 }, { "epoch": 12.32, "learning_rate": 1.618850493233457e-06, "loss": 0.0188, "step": 12143 }, { "epoch": 12.32, "learning_rate": 1.6176548821454197e-06, "loss": 0.0174, "step": 12144 }, { "epoch": 12.32, "learning_rate": 1.6164596738785553e-06, "loss": 0.021, "step": 12145 }, { "epoch": 12.32, "learning_rate": 1.6152648684903027e-06, "loss": 0.0204, "step": 12146 }, { "epoch": 12.33, "learning_rate": 1.6140704660380724e-06, "loss": 0.0191, "step": 12147 }, { "epoch": 12.33, "learning_rate": 1.6128764665792719e-06, "loss": 0.0177, "step": 12148 }, { "epoch": 12.33, "learning_rate": 1.6116828701712716e-06, "loss": 0.0208, "step": 12149 }, { "epoch": 12.33, "learning_rate": 1.6104896768714373e-06, "loss": 0.0183, "step": 12150 }, { "epoch": 12.33, "learning_rate": 1.6092968867371039e-06, "loss": 0.019, "step": 12151 }, { "epoch": 12.33, "learning_rate": 1.6081044998255935e-06, "loss": 0.0202, "step": 12152 }, { "epoch": 12.33, "learning_rate": 1.6069125161942112e-06, "loss": 0.0212, "step": 12153 }, { "epoch": 12.33, "learning_rate": 1.6057209359002323e-06, "loss": 0.0208, "step": 12154 }, { "epoch": 12.33, "learning_rate": 1.6045297590009223e-06, "loss": 0.016, "step": 12155 }, { "epoch": 12.33, "learning_rate": 1.6033389855535253e-06, "loss": 0.0205, "step": 12156 }, { "epoch": 12.34, "learning_rate": 1.6021486156152633e-06, "loss": 0.0187, "step": 12157 }, { "epoch": 12.34, "learning_rate": 1.600958649243345e-06, "loss": 0.0175, "step": 12158 }, { "epoch": 12.34, "learning_rate": 1.5997690864949489e-06, "loss": 0.0177, "step": 12159 }, { "epoch": 12.34, "learning_rate": 1.598579927427245e-06, "loss": 0.0175, "step": 12160 }, { "epoch": 12.34, "learning_rate": 1.5973911720973767e-06, "loss": 0.0206, "step": 12161 }, { "epoch": 12.34, "learning_rate": 1.5962028205624713e-06, "loss": 0.0173, "step": 12162 }, { "epoch": 12.34, "learning_rate": 1.5950148728796388e-06, "loss": 0.0203, "step": 12163 }, { "epoch": 12.34, "learning_rate": 1.5938273291059602e-06, "loss": 0.0193, "step": 12164 }, { "epoch": 12.34, "learning_rate": 1.5926401892985143e-06, "loss": 0.0199, "step": 12165 }, { "epoch": 12.35, "learning_rate": 1.591453453514341e-06, "loss": 0.0196, "step": 12166 }, { "epoch": 12.35, "learning_rate": 1.590267121810477e-06, "loss": 0.0144, "step": 12167 }, { "epoch": 12.35, "learning_rate": 1.5890811942439266e-06, "loss": 0.0211, "step": 12168 }, { "epoch": 12.35, "learning_rate": 1.5878956708716831e-06, "loss": 0.0181, "step": 12169 }, { "epoch": 12.35, "learning_rate": 1.5867105517507198e-06, "loss": 0.019, "step": 12170 }, { "epoch": 12.35, "learning_rate": 1.5855258369379844e-06, "loss": 0.0186, "step": 12171 }, { "epoch": 12.35, "learning_rate": 1.5843415264904127e-06, "loss": 0.0179, "step": 12172 }, { "epoch": 12.35, "learning_rate": 1.5831576204649168e-06, "loss": 0.0134, "step": 12173 }, { "epoch": 12.35, "learning_rate": 1.5819741189183902e-06, "loss": 0.0231, "step": 12174 }, { "epoch": 12.35, "learning_rate": 1.5807910219077094e-06, "loss": 0.0185, "step": 12175 }, { "epoch": 12.36, "learning_rate": 1.5796083294897258e-06, "loss": 0.0185, "step": 12176 }, { "epoch": 12.36, "learning_rate": 1.5784260417212783e-06, "loss": 0.0157, "step": 12177 }, { "epoch": 12.36, "learning_rate": 1.577244158659178e-06, "loss": 0.0219, "step": 12178 }, { "epoch": 12.36, "learning_rate": 1.5760626803602242e-06, "loss": 0.0209, "step": 12179 }, { "epoch": 12.36, "learning_rate": 1.5748816068811967e-06, "loss": 0.018, "step": 12180 }, { "epoch": 12.36, "learning_rate": 1.5737009382788448e-06, "loss": 0.0207, "step": 12181 }, { "epoch": 12.36, "learning_rate": 1.5725206746099165e-06, "loss": 0.019, "step": 12182 }, { "epoch": 12.36, "learning_rate": 1.5713408159311239e-06, "loss": 0.0182, "step": 12183 }, { "epoch": 12.36, "learning_rate": 1.5701613622991685e-06, "loss": 0.0197, "step": 12184 }, { "epoch": 12.36, "learning_rate": 1.5689823137707316e-06, "loss": 0.0198, "step": 12185 }, { "epoch": 12.37, "learning_rate": 1.567803670402469e-06, "loss": 0.0193, "step": 12186 }, { "epoch": 12.37, "learning_rate": 1.5666254322510266e-06, "loss": 0.0219, "step": 12187 }, { "epoch": 12.37, "learning_rate": 1.565447599373021e-06, "loss": 0.0208, "step": 12188 }, { "epoch": 12.37, "learning_rate": 1.5642701718250553e-06, "loss": 0.0211, "step": 12189 }, { "epoch": 12.37, "learning_rate": 1.5630931496637136e-06, "loss": 0.0159, "step": 12190 }, { "epoch": 12.37, "learning_rate": 1.5619165329455578e-06, "loss": 0.0251, "step": 12191 }, { "epoch": 12.37, "learning_rate": 1.5607403217271333e-06, "loss": 0.0206, "step": 12192 }, { "epoch": 12.37, "learning_rate": 1.5595645160649586e-06, "loss": 0.0202, "step": 12193 }, { "epoch": 12.37, "learning_rate": 1.558389116015545e-06, "loss": 0.0193, "step": 12194 }, { "epoch": 12.37, "learning_rate": 1.557214121635372e-06, "loss": 0.0215, "step": 12195 }, { "epoch": 12.38, "learning_rate": 1.5560395329809063e-06, "loss": 0.0205, "step": 12196 }, { "epoch": 12.38, "learning_rate": 1.5548653501085964e-06, "loss": 0.0177, "step": 12197 }, { "epoch": 12.38, "learning_rate": 1.5536915730748625e-06, "loss": 0.0181, "step": 12198 }, { "epoch": 12.38, "learning_rate": 1.5525182019361207e-06, "loss": 0.0181, "step": 12199 }, { "epoch": 12.38, "learning_rate": 1.5513452367487514e-06, "loss": 0.0186, "step": 12200 }, { "epoch": 12.38, "learning_rate": 1.5501726775691239e-06, "loss": 0.0187, "step": 12201 }, { "epoch": 12.38, "learning_rate": 1.5490005244535899e-06, "loss": 0.0172, "step": 12202 }, { "epoch": 12.38, "learning_rate": 1.5478287774584732e-06, "loss": 0.0168, "step": 12203 }, { "epoch": 12.38, "learning_rate": 1.5466574366400877e-06, "loss": 0.0223, "step": 12204 }, { "epoch": 12.38, "learning_rate": 1.5454865020547194e-06, "loss": 0.0218, "step": 12205 }, { "epoch": 12.39, "learning_rate": 1.54431597375864e-06, "loss": 0.0182, "step": 12206 }, { "epoch": 12.39, "learning_rate": 1.543145851808101e-06, "loss": 0.0201, "step": 12207 }, { "epoch": 12.39, "learning_rate": 1.5419761362593334e-06, "loss": 0.0173, "step": 12208 }, { "epoch": 12.39, "learning_rate": 1.5408068271685505e-06, "loss": 0.0199, "step": 12209 }, { "epoch": 12.39, "learning_rate": 1.5396379245919413e-06, "loss": 0.0183, "step": 12210 }, { "epoch": 12.39, "learning_rate": 1.5384694285856795e-06, "loss": 0.0152, "step": 12211 }, { "epoch": 12.39, "learning_rate": 1.537301339205921e-06, "loss": 0.0169, "step": 12212 }, { "epoch": 12.39, "learning_rate": 1.5361336565087959e-06, "loss": 0.0195, "step": 12213 }, { "epoch": 12.39, "learning_rate": 1.5349663805504189e-06, "loss": 0.0197, "step": 12214 }, { "epoch": 12.39, "learning_rate": 1.5337995113868853e-06, "loss": 0.0182, "step": 12215 }, { "epoch": 12.4, "learning_rate": 1.5326330490742724e-06, "loss": 0.0211, "step": 12216 }, { "epoch": 12.4, "learning_rate": 1.53146699366863e-06, "loss": 0.0177, "step": 12217 }, { "epoch": 12.4, "learning_rate": 1.5303013452259974e-06, "loss": 0.021, "step": 12218 }, { "epoch": 12.4, "learning_rate": 1.5291361038023933e-06, "loss": 0.0193, "step": 12219 }, { "epoch": 12.4, "learning_rate": 1.5279712694538084e-06, "loss": 0.0186, "step": 12220 }, { "epoch": 12.4, "learning_rate": 1.5268068422362258e-06, "loss": 0.0187, "step": 12221 }, { "epoch": 12.4, "learning_rate": 1.5256428222055986e-06, "loss": 0.0225, "step": 12222 }, { "epoch": 12.4, "learning_rate": 1.5244792094178662e-06, "loss": 0.0181, "step": 12223 }, { "epoch": 12.4, "learning_rate": 1.5233160039289485e-06, "loss": 0.0195, "step": 12224 }, { "epoch": 12.4, "learning_rate": 1.522153205794742e-06, "loss": 0.0166, "step": 12225 }, { "epoch": 12.41, "learning_rate": 1.5209908150711306e-06, "loss": 0.021, "step": 12226 }, { "epoch": 12.41, "learning_rate": 1.5198288318139687e-06, "loss": 0.0225, "step": 12227 }, { "epoch": 12.41, "learning_rate": 1.5186672560790982e-06, "loss": 0.0169, "step": 12228 }, { "epoch": 12.41, "learning_rate": 1.5175060879223425e-06, "loss": 0.0197, "step": 12229 }, { "epoch": 12.41, "learning_rate": 1.5163453273994978e-06, "loss": 0.0156, "step": 12230 }, { "epoch": 12.41, "learning_rate": 1.5151849745663483e-06, "loss": 0.0206, "step": 12231 }, { "epoch": 12.41, "learning_rate": 1.514025029478654e-06, "loss": 0.0211, "step": 12232 }, { "epoch": 12.41, "learning_rate": 1.5128654921921627e-06, "loss": 0.0196, "step": 12233 }, { "epoch": 12.41, "learning_rate": 1.5117063627625893e-06, "loss": 0.0179, "step": 12234 }, { "epoch": 12.42, "learning_rate": 1.510547641245641e-06, "loss": 0.0204, "step": 12235 }, { "epoch": 12.42, "learning_rate": 1.5093893276970019e-06, "loss": 0.0209, "step": 12236 }, { "epoch": 12.42, "learning_rate": 1.5082314221723327e-06, "loss": 0.0196, "step": 12237 }, { "epoch": 12.42, "learning_rate": 1.5070739247272813e-06, "loss": 0.0202, "step": 12238 }, { "epoch": 12.42, "learning_rate": 1.505916835417467e-06, "loss": 0.0239, "step": 12239 }, { "epoch": 12.42, "learning_rate": 1.5047601542985013e-06, "loss": 0.0232, "step": 12240 }, { "epoch": 12.42, "learning_rate": 1.5036038814259645e-06, "loss": 0.0182, "step": 12241 }, { "epoch": 12.42, "learning_rate": 1.5024480168554246e-06, "loss": 0.0197, "step": 12242 }, { "epoch": 12.42, "learning_rate": 1.501292560642429e-06, "loss": 0.0183, "step": 12243 }, { "epoch": 12.42, "learning_rate": 1.5001375128425e-06, "loss": 0.0171, "step": 12244 }, { "epoch": 12.43, "learning_rate": 1.4989828735111477e-06, "loss": 0.0197, "step": 12245 }, { "epoch": 12.43, "learning_rate": 1.4978286427038602e-06, "loss": 0.022, "step": 12246 }, { "epoch": 12.43, "learning_rate": 1.4966748204761017e-06, "loss": 0.0146, "step": 12247 }, { "epoch": 12.43, "learning_rate": 1.4955214068833212e-06, "loss": 0.0201, "step": 12248 }, { "epoch": 12.43, "learning_rate": 1.494368401980949e-06, "loss": 0.0196, "step": 12249 }, { "epoch": 12.43, "learning_rate": 1.4932158058243939e-06, "loss": 0.0187, "step": 12250 }, { "epoch": 12.43, "learning_rate": 1.4920636184690418e-06, "loss": 0.0191, "step": 12251 }, { "epoch": 12.43, "learning_rate": 1.4909118399702638e-06, "loss": 0.0207, "step": 12252 }, { "epoch": 12.43, "learning_rate": 1.4897604703834123e-06, "loss": 0.0218, "step": 12253 }, { "epoch": 12.43, "learning_rate": 1.4886095097638131e-06, "loss": 0.0185, "step": 12254 }, { "epoch": 12.44, "learning_rate": 1.4874589581667775e-06, "loss": 0.0206, "step": 12255 }, { "epoch": 12.44, "learning_rate": 1.486308815647598e-06, "loss": 0.0172, "step": 12256 }, { "epoch": 12.44, "learning_rate": 1.4851590822615447e-06, "loss": 0.0182, "step": 12257 }, { "epoch": 12.44, "learning_rate": 1.4840097580638713e-06, "loss": 0.0192, "step": 12258 }, { "epoch": 12.44, "learning_rate": 1.4828608431098057e-06, "loss": 0.0211, "step": 12259 }, { "epoch": 12.44, "learning_rate": 1.481712337454565e-06, "loss": 0.0204, "step": 12260 }, { "epoch": 12.44, "learning_rate": 1.480564241153335e-06, "loss": 0.0222, "step": 12261 }, { "epoch": 12.44, "learning_rate": 1.4794165542612937e-06, "loss": 0.0218, "step": 12262 }, { "epoch": 12.44, "learning_rate": 1.4782692768335937e-06, "loss": 0.0197, "step": 12263 }, { "epoch": 12.44, "learning_rate": 1.4771224089253644e-06, "loss": 0.0195, "step": 12264 }, { "epoch": 12.45, "learning_rate": 1.475975950591726e-06, "loss": 0.0176, "step": 12265 }, { "epoch": 12.45, "learning_rate": 1.4748299018877676e-06, "loss": 0.0202, "step": 12266 }, { "epoch": 12.45, "learning_rate": 1.4736842628685678e-06, "loss": 0.0206, "step": 12267 }, { "epoch": 12.45, "learning_rate": 1.4725390335891755e-06, "loss": 0.0142, "step": 12268 }, { "epoch": 12.45, "learning_rate": 1.47139421410463e-06, "loss": 0.0195, "step": 12269 }, { "epoch": 12.45, "learning_rate": 1.4702498044699476e-06, "loss": 0.0206, "step": 12270 }, { "epoch": 12.45, "learning_rate": 1.4691058047401197e-06, "loss": 0.0196, "step": 12271 }, { "epoch": 12.45, "learning_rate": 1.4679622149701245e-06, "loss": 0.0202, "step": 12272 }, { "epoch": 12.45, "learning_rate": 1.4668190352149191e-06, "loss": 0.0172, "step": 12273 }, { "epoch": 12.45, "learning_rate": 1.4656762655294377e-06, "loss": 0.0241, "step": 12274 }, { "epoch": 12.46, "learning_rate": 1.4645339059686025e-06, "loss": 0.022, "step": 12275 }, { "epoch": 12.46, "learning_rate": 1.4633919565873033e-06, "loss": 0.0219, "step": 12276 }, { "epoch": 12.46, "learning_rate": 1.4622504174404239e-06, "loss": 0.0237, "step": 12277 }, { "epoch": 12.46, "learning_rate": 1.461109288582817e-06, "loss": 0.0214, "step": 12278 }, { "epoch": 12.46, "learning_rate": 1.4599685700693221e-06, "loss": 0.0187, "step": 12279 }, { "epoch": 12.46, "learning_rate": 1.4588282619547601e-06, "loss": 0.0212, "step": 12280 }, { "epoch": 12.46, "learning_rate": 1.4576883642939232e-06, "loss": 0.0201, "step": 12281 }, { "epoch": 12.46, "learning_rate": 1.4565488771415992e-06, "loss": 0.0192, "step": 12282 }, { "epoch": 12.46, "learning_rate": 1.4554098005525408e-06, "loss": 0.0171, "step": 12283 }, { "epoch": 12.46, "learning_rate": 1.4542711345814876e-06, "loss": 0.0169, "step": 12284 }, { "epoch": 12.47, "learning_rate": 1.4531328792831634e-06, "loss": 0.0187, "step": 12285 }, { "epoch": 12.47, "learning_rate": 1.4519950347122635e-06, "loss": 0.0215, "step": 12286 }, { "epoch": 12.47, "learning_rate": 1.450857600923472e-06, "loss": 0.0212, "step": 12287 }, { "epoch": 12.47, "learning_rate": 1.4497205779714452e-06, "loss": 0.0186, "step": 12288 }, { "epoch": 12.47, "learning_rate": 1.4485839659108246e-06, "loss": 0.0192, "step": 12289 }, { "epoch": 12.47, "learning_rate": 1.4474477647962338e-06, "loss": 0.022, "step": 12290 }, { "epoch": 12.47, "learning_rate": 1.4463119746822718e-06, "loss": 0.0167, "step": 12291 }, { "epoch": 12.47, "learning_rate": 1.445176595623522e-06, "loss": 0.0177, "step": 12292 }, { "epoch": 12.47, "learning_rate": 1.444041627674544e-06, "loss": 0.023, "step": 12293 }, { "epoch": 12.47, "learning_rate": 1.4429070708898818e-06, "loss": 0.0199, "step": 12294 }, { "epoch": 12.48, "learning_rate": 1.4417729253240543e-06, "loss": 0.0186, "step": 12295 }, { "epoch": 12.48, "learning_rate": 1.4406391910315653e-06, "loss": 0.0198, "step": 12296 }, { "epoch": 12.48, "learning_rate": 1.4395058680669006e-06, "loss": 0.0209, "step": 12297 }, { "epoch": 12.48, "learning_rate": 1.438372956484515e-06, "loss": 0.0232, "step": 12298 }, { "epoch": 12.48, "learning_rate": 1.437240456338862e-06, "loss": 0.016, "step": 12299 }, { "epoch": 12.48, "learning_rate": 1.436108367684358e-06, "loss": 0.0189, "step": 12300 }, { "epoch": 12.48, "learning_rate": 1.434976690575408e-06, "loss": 0.0228, "step": 12301 }, { "epoch": 12.48, "learning_rate": 1.4338454250663992e-06, "loss": 0.0201, "step": 12302 }, { "epoch": 12.48, "learning_rate": 1.4327145712116897e-06, "loss": 0.022, "step": 12303 }, { "epoch": 12.49, "learning_rate": 1.4315841290656284e-06, "loss": 0.019, "step": 12304 }, { "epoch": 12.49, "learning_rate": 1.4304540986825354e-06, "loss": 0.0199, "step": 12305 }, { "epoch": 12.49, "learning_rate": 1.4293244801167193e-06, "loss": 0.0182, "step": 12306 }, { "epoch": 12.49, "learning_rate": 1.4281952734224624e-06, "loss": 0.0196, "step": 12307 }, { "epoch": 12.49, "learning_rate": 1.4270664786540312e-06, "loss": 0.0196, "step": 12308 }, { "epoch": 12.49, "learning_rate": 1.4259380958656721e-06, "loss": 0.0172, "step": 12309 }, { "epoch": 12.49, "learning_rate": 1.4248101251116075e-06, "loss": 0.0169, "step": 12310 }, { "epoch": 12.49, "learning_rate": 1.4236825664460474e-06, "loss": 0.0185, "step": 12311 }, { "epoch": 12.49, "learning_rate": 1.4225554199231718e-06, "loss": 0.02, "step": 12312 }, { "epoch": 12.49, "learning_rate": 1.4214286855971494e-06, "loss": 0.0218, "step": 12313 }, { "epoch": 12.5, "learning_rate": 1.420302363522129e-06, "loss": 0.0223, "step": 12314 }, { "epoch": 12.5, "learning_rate": 1.4191764537522313e-06, "loss": 0.0144, "step": 12315 }, { "epoch": 12.5, "learning_rate": 1.4180509563415711e-06, "loss": 0.0202, "step": 12316 }, { "epoch": 12.5, "learning_rate": 1.4169258713442269e-06, "loss": 0.0168, "step": 12317 }, { "epoch": 12.5, "learning_rate": 1.4158011988142705e-06, "loss": 0.018, "step": 12318 }, { "epoch": 12.5, "learning_rate": 1.4146769388057492e-06, "loss": 0.0164, "step": 12319 }, { "epoch": 12.5, "learning_rate": 1.413553091372688e-06, "loss": 0.0235, "step": 12320 }, { "epoch": 12.5, "learning_rate": 1.4124296565690965e-06, "loss": 0.0164, "step": 12321 }, { "epoch": 12.5, "learning_rate": 1.4113066344489601e-06, "loss": 0.0181, "step": 12322 }, { "epoch": 12.5, "learning_rate": 1.410184025066247e-06, "loss": 0.0199, "step": 12323 }, { "epoch": 12.51, "learning_rate": 1.4090618284749058e-06, "loss": 0.0228, "step": 12324 }, { "epoch": 12.51, "learning_rate": 1.407940044728865e-06, "loss": 0.0177, "step": 12325 }, { "epoch": 12.51, "learning_rate": 1.4068186738820356e-06, "loss": 0.0229, "step": 12326 }, { "epoch": 12.51, "learning_rate": 1.4056977159883011e-06, "loss": 0.024, "step": 12327 }, { "epoch": 12.51, "learning_rate": 1.404577171101532e-06, "loss": 0.0227, "step": 12328 }, { "epoch": 12.51, "learning_rate": 1.4034570392755797e-06, "loss": 0.0207, "step": 12329 }, { "epoch": 12.51, "learning_rate": 1.4023373205642687e-06, "loss": 0.0186, "step": 12330 }, { "epoch": 12.51, "learning_rate": 1.4012180150214105e-06, "loss": 0.0221, "step": 12331 }, { "epoch": 12.51, "learning_rate": 1.4000991227007943e-06, "loss": 0.0195, "step": 12332 }, { "epoch": 12.51, "learning_rate": 1.3989806436561914e-06, "loss": 0.0198, "step": 12333 }, { "epoch": 12.52, "learning_rate": 1.397862577941348e-06, "loss": 0.021, "step": 12334 }, { "epoch": 12.52, "learning_rate": 1.396744925609993e-06, "loss": 0.0246, "step": 12335 }, { "epoch": 12.52, "learning_rate": 1.3956276867158425e-06, "loss": 0.0214, "step": 12336 }, { "epoch": 12.52, "learning_rate": 1.3945108613125779e-06, "loss": 0.0139, "step": 12337 }, { "epoch": 12.52, "learning_rate": 1.3933944494538777e-06, "loss": 0.0194, "step": 12338 }, { "epoch": 12.52, "learning_rate": 1.392278451193384e-06, "loss": 0.0198, "step": 12339 }, { "epoch": 12.52, "learning_rate": 1.3911628665847321e-06, "loss": 0.0201, "step": 12340 }, { "epoch": 12.52, "learning_rate": 1.39004769568153e-06, "loss": 0.0199, "step": 12341 }, { "epoch": 12.52, "learning_rate": 1.3889329385373706e-06, "loss": 0.0173, "step": 12342 }, { "epoch": 12.52, "learning_rate": 1.3878185952058255e-06, "loss": 0.0215, "step": 12343 }, { "epoch": 12.53, "learning_rate": 1.3867046657404416e-06, "loss": 0.0186, "step": 12344 }, { "epoch": 12.53, "learning_rate": 1.3855911501947517e-06, "loss": 0.0219, "step": 12345 }, { "epoch": 12.53, "learning_rate": 1.3844780486222686e-06, "loss": 0.023, "step": 12346 }, { "epoch": 12.53, "learning_rate": 1.3833653610764796e-06, "loss": 0.0171, "step": 12347 }, { "epoch": 12.53, "learning_rate": 1.3822530876108576e-06, "loss": 0.0209, "step": 12348 }, { "epoch": 12.53, "learning_rate": 1.3811412282788539e-06, "loss": 0.0198, "step": 12349 }, { "epoch": 12.53, "learning_rate": 1.3800297831339026e-06, "loss": 0.0173, "step": 12350 }, { "epoch": 12.53, "learning_rate": 1.3789187522294111e-06, "loss": 0.0244, "step": 12351 }, { "epoch": 12.53, "learning_rate": 1.3778081356187723e-06, "loss": 0.0188, "step": 12352 }, { "epoch": 12.53, "learning_rate": 1.376697933355361e-06, "loss": 0.0158, "step": 12353 }, { "epoch": 12.54, "learning_rate": 1.3755881454925235e-06, "loss": 0.0179, "step": 12354 }, { "epoch": 12.54, "learning_rate": 1.374478772083594e-06, "loss": 0.0223, "step": 12355 }, { "epoch": 12.54, "learning_rate": 1.3733698131818852e-06, "loss": 0.0212, "step": 12356 }, { "epoch": 12.54, "learning_rate": 1.372261268840689e-06, "loss": 0.0209, "step": 12357 }, { "epoch": 12.54, "learning_rate": 1.3711531391132793e-06, "loss": 0.0187, "step": 12358 }, { "epoch": 12.54, "learning_rate": 1.3700454240529038e-06, "loss": 0.0206, "step": 12359 }, { "epoch": 12.54, "learning_rate": 1.3689381237127997e-06, "loss": 0.022, "step": 12360 }, { "epoch": 12.54, "learning_rate": 1.3678312381461755e-06, "loss": 0.0207, "step": 12361 }, { "epoch": 12.54, "learning_rate": 1.3667247674062234e-06, "loss": 0.0179, "step": 12362 }, { "epoch": 12.54, "learning_rate": 1.3656187115461206e-06, "loss": 0.022, "step": 12363 }, { "epoch": 12.55, "learning_rate": 1.3645130706190136e-06, "loss": 0.0175, "step": 12364 }, { "epoch": 12.55, "learning_rate": 1.3634078446780374e-06, "loss": 0.0204, "step": 12365 }, { "epoch": 12.55, "learning_rate": 1.3623030337763054e-06, "loss": 0.0204, "step": 12366 }, { "epoch": 12.55, "learning_rate": 1.361198637966913e-06, "loss": 0.024, "step": 12367 }, { "epoch": 12.55, "learning_rate": 1.3600946573029262e-06, "loss": 0.0214, "step": 12368 }, { "epoch": 12.55, "learning_rate": 1.3589910918374016e-06, "loss": 0.02, "step": 12369 }, { "epoch": 12.55, "learning_rate": 1.3578879416233747e-06, "loss": 0.017, "step": 12370 }, { "epoch": 12.55, "learning_rate": 1.356785206713853e-06, "loss": 0.0227, "step": 12371 }, { "epoch": 12.55, "learning_rate": 1.355682887161832e-06, "loss": 0.0173, "step": 12372 }, { "epoch": 12.56, "learning_rate": 1.3545809830202838e-06, "loss": 0.0176, "step": 12373 }, { "epoch": 12.56, "learning_rate": 1.3534794943421636e-06, "loss": 0.0171, "step": 12374 }, { "epoch": 12.56, "learning_rate": 1.3523784211804035e-06, "loss": 0.0198, "step": 12375 }, { "epoch": 12.56, "learning_rate": 1.351277763587916e-06, "loss": 0.0204, "step": 12376 }, { "epoch": 12.56, "learning_rate": 1.350177521617595e-06, "loss": 0.0175, "step": 12377 }, { "epoch": 12.56, "learning_rate": 1.3490776953223107e-06, "loss": 0.0171, "step": 12378 }, { "epoch": 12.56, "learning_rate": 1.3479782847549194e-06, "loss": 0.0212, "step": 12379 }, { "epoch": 12.56, "learning_rate": 1.3468792899682547e-06, "loss": 0.0214, "step": 12380 }, { "epoch": 12.56, "learning_rate": 1.3457807110151245e-06, "loss": 0.0246, "step": 12381 }, { "epoch": 12.56, "learning_rate": 1.3446825479483293e-06, "loss": 0.0198, "step": 12382 }, { "epoch": 12.57, "learning_rate": 1.3435848008206376e-06, "loss": 0.0213, "step": 12383 }, { "epoch": 12.57, "learning_rate": 1.3424874696848056e-06, "loss": 0.0218, "step": 12384 }, { "epoch": 12.57, "learning_rate": 1.3413905545935634e-06, "loss": 0.02, "step": 12385 }, { "epoch": 12.57, "learning_rate": 1.3402940555996257e-06, "loss": 0.0167, "step": 12386 }, { "epoch": 12.57, "learning_rate": 1.3391979727556881e-06, "loss": 0.0201, "step": 12387 }, { "epoch": 12.57, "learning_rate": 1.3381023061144183e-06, "loss": 0.0204, "step": 12388 }, { "epoch": 12.57, "learning_rate": 1.3370070557284743e-06, "loss": 0.024, "step": 12389 }, { "epoch": 12.57, "learning_rate": 1.3359122216504872e-06, "loss": 0.0176, "step": 12390 }, { "epoch": 12.57, "learning_rate": 1.3348178039330705e-06, "loss": 0.0218, "step": 12391 }, { "epoch": 12.57, "learning_rate": 1.33372380262882e-06, "loss": 0.0237, "step": 12392 }, { "epoch": 12.58, "learning_rate": 1.332630217790305e-06, "loss": 0.0171, "step": 12393 }, { "epoch": 12.58, "learning_rate": 1.331537049470083e-06, "loss": 0.0182, "step": 12394 }, { "epoch": 12.58, "learning_rate": 1.3304442977206822e-06, "loss": 0.0175, "step": 12395 }, { "epoch": 12.58, "learning_rate": 1.3293519625946173e-06, "loss": 0.0187, "step": 12396 }, { "epoch": 12.58, "learning_rate": 1.328260044144385e-06, "loss": 0.0177, "step": 12397 }, { "epoch": 12.58, "learning_rate": 1.3271685424224524e-06, "loss": 0.0176, "step": 12398 }, { "epoch": 12.58, "learning_rate": 1.3260774574812796e-06, "loss": 0.0191, "step": 12399 }, { "epoch": 12.58, "learning_rate": 1.3249867893732936e-06, "loss": 0.018, "step": 12400 }, { "epoch": 12.58, "learning_rate": 1.3238965381509107e-06, "loss": 0.0206, "step": 12401 }, { "epoch": 12.58, "learning_rate": 1.3228067038665248e-06, "loss": 0.0178, "step": 12402 }, { "epoch": 12.59, "learning_rate": 1.3217172865725047e-06, "loss": 0.0192, "step": 12403 }, { "epoch": 12.59, "learning_rate": 1.3206282863212083e-06, "loss": 0.0165, "step": 12404 }, { "epoch": 12.59, "learning_rate": 1.319539703164965e-06, "loss": 0.0233, "step": 12405 }, { "epoch": 12.59, "learning_rate": 1.3184515371560868e-06, "loss": 0.0205, "step": 12406 }, { "epoch": 12.59, "learning_rate": 1.3173637883468693e-06, "loss": 0.0187, "step": 12407 }, { "epoch": 12.59, "learning_rate": 1.3162764567895848e-06, "loss": 0.0195, "step": 12408 }, { "epoch": 12.59, "learning_rate": 1.3151895425364868e-06, "loss": 0.0212, "step": 12409 }, { "epoch": 12.59, "learning_rate": 1.3141030456398053e-06, "loss": 0.0242, "step": 12410 }, { "epoch": 12.59, "learning_rate": 1.3130169661517556e-06, "loss": 0.0225, "step": 12411 }, { "epoch": 12.59, "learning_rate": 1.3119313041245263e-06, "loss": 0.0203, "step": 12412 }, { "epoch": 12.6, "learning_rate": 1.3108460596102935e-06, "loss": 0.0173, "step": 12413 }, { "epoch": 12.6, "learning_rate": 1.3097612326612108e-06, "loss": 0.0177, "step": 12414 }, { "epoch": 12.6, "learning_rate": 1.3086768233294034e-06, "loss": 0.0196, "step": 12415 }, { "epoch": 12.6, "learning_rate": 1.3075928316669928e-06, "loss": 0.0208, "step": 12416 }, { "epoch": 12.6, "learning_rate": 1.3065092577260651e-06, "loss": 0.0213, "step": 12417 }, { "epoch": 12.6, "learning_rate": 1.3054261015586944e-06, "loss": 0.0214, "step": 12418 }, { "epoch": 12.6, "learning_rate": 1.3043433632169345e-06, "loss": 0.0268, "step": 12419 }, { "epoch": 12.6, "learning_rate": 1.3032610427528137e-06, "loss": 0.0218, "step": 12420 }, { "epoch": 12.6, "learning_rate": 1.302179140218347e-06, "loss": 0.0178, "step": 12421 }, { "epoch": 12.6, "learning_rate": 1.301097655665523e-06, "loss": 0.0178, "step": 12422 }, { "epoch": 12.61, "learning_rate": 1.3000165891463145e-06, "loss": 0.0229, "step": 12423 }, { "epoch": 12.61, "learning_rate": 1.2989359407126746e-06, "loss": 0.0226, "step": 12424 }, { "epoch": 12.61, "learning_rate": 1.2978557104165345e-06, "loss": 0.0183, "step": 12425 }, { "epoch": 12.61, "learning_rate": 1.2967758983098066e-06, "loss": 0.022, "step": 12426 }, { "epoch": 12.61, "learning_rate": 1.2956965044443793e-06, "loss": 0.016, "step": 12427 }, { "epoch": 12.61, "learning_rate": 1.2946175288721251e-06, "loss": 0.0237, "step": 12428 }, { "epoch": 12.61, "learning_rate": 1.2935389716448976e-06, "loss": 0.0205, "step": 12429 }, { "epoch": 12.61, "learning_rate": 1.2924608328145237e-06, "loss": 0.0184, "step": 12430 }, { "epoch": 12.61, "learning_rate": 1.2913831124328192e-06, "loss": 0.0188, "step": 12431 }, { "epoch": 12.61, "learning_rate": 1.2903058105515665e-06, "loss": 0.0181, "step": 12432 }, { "epoch": 12.62, "learning_rate": 1.2892289272225466e-06, "loss": 0.0192, "step": 12433 }, { "epoch": 12.62, "learning_rate": 1.2881524624975039e-06, "loss": 0.0167, "step": 12434 }, { "epoch": 12.62, "learning_rate": 1.2870764164281702e-06, "loss": 0.0193, "step": 12435 }, { "epoch": 12.62, "learning_rate": 1.2860007890662573e-06, "loss": 0.0192, "step": 12436 }, { "epoch": 12.62, "learning_rate": 1.284925580463453e-06, "loss": 0.0219, "step": 12437 }, { "epoch": 12.62, "learning_rate": 1.2838507906714304e-06, "loss": 0.0221, "step": 12438 }, { "epoch": 12.62, "learning_rate": 1.2827764197418358e-06, "loss": 0.0206, "step": 12439 }, { "epoch": 12.62, "learning_rate": 1.2817024677263012e-06, "loss": 0.0217, "step": 12440 }, { "epoch": 12.62, "learning_rate": 1.2806289346764356e-06, "loss": 0.022, "step": 12441 }, { "epoch": 12.63, "learning_rate": 1.2795558206438296e-06, "loss": 0.0232, "step": 12442 }, { "epoch": 12.63, "learning_rate": 1.2784831256800546e-06, "loss": 0.0196, "step": 12443 }, { "epoch": 12.63, "learning_rate": 1.2774108498366544e-06, "loss": 0.0223, "step": 12444 }, { "epoch": 12.63, "learning_rate": 1.2763389931651615e-06, "loss": 0.0206, "step": 12445 }, { "epoch": 12.63, "learning_rate": 1.275267555717088e-06, "loss": 0.0191, "step": 12446 }, { "epoch": 12.63, "learning_rate": 1.274196537543916e-06, "loss": 0.0224, "step": 12447 }, { "epoch": 12.63, "learning_rate": 1.2731259386971185e-06, "loss": 0.0176, "step": 12448 }, { "epoch": 12.63, "learning_rate": 1.2720557592281434e-06, "loss": 0.0221, "step": 12449 }, { "epoch": 12.63, "learning_rate": 1.2709859991884211e-06, "loss": 0.0215, "step": 12450 }, { "epoch": 12.63, "learning_rate": 1.2699166586293554e-06, "loss": 0.0189, "step": 12451 }, { "epoch": 12.64, "learning_rate": 1.2688477376023366e-06, "loss": 0.0204, "step": 12452 }, { "epoch": 12.64, "learning_rate": 1.267779236158736e-06, "loss": 0.0156, "step": 12453 }, { "epoch": 12.64, "learning_rate": 1.2667111543498967e-06, "loss": 0.0194, "step": 12454 }, { "epoch": 12.64, "learning_rate": 1.26564349222715e-06, "loss": 0.02, "step": 12455 }, { "epoch": 12.64, "learning_rate": 1.264576249841798e-06, "loss": 0.0188, "step": 12456 }, { "epoch": 12.64, "learning_rate": 1.263509427245131e-06, "loss": 0.0173, "step": 12457 }, { "epoch": 12.64, "learning_rate": 1.2624430244884189e-06, "loss": 0.0207, "step": 12458 }, { "epoch": 12.64, "learning_rate": 1.2613770416229054e-06, "loss": 0.02, "step": 12459 }, { "epoch": 12.64, "learning_rate": 1.2603114786998205e-06, "loss": 0.0199, "step": 12460 }, { "epoch": 12.64, "learning_rate": 1.2592463357703655e-06, "loss": 0.0215, "step": 12461 }, { "epoch": 12.65, "learning_rate": 1.2581816128857306e-06, "loss": 0.0225, "step": 12462 }, { "epoch": 12.65, "learning_rate": 1.2571173100970824e-06, "loss": 0.02, "step": 12463 }, { "epoch": 12.65, "learning_rate": 1.2560534274555647e-06, "loss": 0.0224, "step": 12464 }, { "epoch": 12.65, "learning_rate": 1.2549899650123031e-06, "loss": 0.0225, "step": 12465 }, { "epoch": 12.65, "learning_rate": 1.2539269228184047e-06, "loss": 0.0205, "step": 12466 }, { "epoch": 12.65, "learning_rate": 1.2528643009249574e-06, "loss": 0.0218, "step": 12467 }, { "epoch": 12.65, "learning_rate": 1.2518020993830216e-06, "loss": 0.0184, "step": 12468 }, { "epoch": 12.65, "learning_rate": 1.250740318243645e-06, "loss": 0.0235, "step": 12469 }, { "epoch": 12.65, "learning_rate": 1.2496789575578528e-06, "loss": 0.0141, "step": 12470 }, { "epoch": 12.65, "learning_rate": 1.2486180173766483e-06, "loss": 0.0199, "step": 12471 }, { "epoch": 12.66, "learning_rate": 1.2475574977510152e-06, "loss": 0.0168, "step": 12472 }, { "epoch": 12.66, "learning_rate": 1.2464973987319196e-06, "loss": 0.0184, "step": 12473 }, { "epoch": 12.66, "learning_rate": 1.245437720370305e-06, "loss": 0.0159, "step": 12474 }, { "epoch": 12.66, "learning_rate": 1.2443784627170974e-06, "loss": 0.0208, "step": 12475 }, { "epoch": 12.66, "learning_rate": 1.243319625823195e-06, "loss": 0.0189, "step": 12476 }, { "epoch": 12.66, "learning_rate": 1.242261209739487e-06, "loss": 0.0201, "step": 12477 }, { "epoch": 12.66, "learning_rate": 1.241203214516833e-06, "loss": 0.0161, "step": 12478 }, { "epoch": 12.66, "learning_rate": 1.2401456402060752e-06, "loss": 0.0201, "step": 12479 }, { "epoch": 12.66, "learning_rate": 1.23908848685804e-06, "loss": 0.0217, "step": 12480 }, { "epoch": 12.66, "learning_rate": 1.2380317545235265e-06, "loss": 0.0211, "step": 12481 }, { "epoch": 12.67, "learning_rate": 1.2369754432533176e-06, "loss": 0.0172, "step": 12482 }, { "epoch": 12.67, "learning_rate": 1.235919553098176e-06, "loss": 0.0201, "step": 12483 }, { "epoch": 12.67, "learning_rate": 1.2348640841088443e-06, "loss": 0.0174, "step": 12484 }, { "epoch": 12.67, "learning_rate": 1.2338090363360422e-06, "loss": 0.0208, "step": 12485 }, { "epoch": 12.67, "learning_rate": 1.232754409830471e-06, "loss": 0.0212, "step": 12486 }, { "epoch": 12.67, "learning_rate": 1.231700204642815e-06, "loss": 0.0201, "step": 12487 }, { "epoch": 12.67, "learning_rate": 1.2306464208237312e-06, "loss": 0.0168, "step": 12488 }, { "epoch": 12.67, "learning_rate": 1.2295930584238602e-06, "loss": 0.0232, "step": 12489 }, { "epoch": 12.67, "learning_rate": 1.228540117493824e-06, "loss": 0.019, "step": 12490 }, { "epoch": 12.67, "learning_rate": 1.2274875980842216e-06, "loss": 0.0217, "step": 12491 }, { "epoch": 12.68, "learning_rate": 1.2264355002456362e-06, "loss": 0.0213, "step": 12492 }, { "epoch": 12.68, "learning_rate": 1.225383824028622e-06, "loss": 0.0179, "step": 12493 }, { "epoch": 12.68, "learning_rate": 1.2243325694837227e-06, "loss": 0.024, "step": 12494 }, { "epoch": 12.68, "learning_rate": 1.2232817366614525e-06, "loss": 0.0219, "step": 12495 }, { "epoch": 12.68, "learning_rate": 1.2222313256123141e-06, "loss": 0.0167, "step": 12496 }, { "epoch": 12.68, "learning_rate": 1.2211813363867852e-06, "loss": 0.0214, "step": 12497 }, { "epoch": 12.68, "learning_rate": 1.220131769035321e-06, "loss": 0.0187, "step": 12498 }, { "epoch": 12.68, "learning_rate": 1.2190826236083652e-06, "loss": 0.0223, "step": 12499 }, { "epoch": 12.68, "learning_rate": 1.21803390015633e-06, "loss": 0.0204, "step": 12500 }, { "epoch": 12.68, "learning_rate": 1.2169855987296153e-06, "loss": 0.0218, "step": 12501 }, { "epoch": 12.69, "learning_rate": 1.2159377193786004e-06, "loss": 0.0214, "step": 12502 }, { "epoch": 12.69, "learning_rate": 1.2148902621536362e-06, "loss": 0.0207, "step": 12503 }, { "epoch": 12.69, "learning_rate": 1.2138432271050648e-06, "loss": 0.0182, "step": 12504 }, { "epoch": 12.69, "learning_rate": 1.2127966142831981e-06, "loss": 0.0251, "step": 12505 }, { "epoch": 12.69, "learning_rate": 1.211750423738335e-06, "loss": 0.0223, "step": 12506 }, { "epoch": 12.69, "learning_rate": 1.210704655520749e-06, "loss": 0.0154, "step": 12507 }, { "epoch": 12.69, "learning_rate": 1.2096593096806964e-06, "loss": 0.024, "step": 12508 }, { "epoch": 12.69, "learning_rate": 1.2086143862684153e-06, "loss": 0.021, "step": 12509 }, { "epoch": 12.69, "learning_rate": 1.2075698853341155e-06, "loss": 0.0192, "step": 12510 }, { "epoch": 12.7, "learning_rate": 1.2065258069279951e-06, "loss": 0.014, "step": 12511 }, { "epoch": 12.7, "learning_rate": 1.2054821511002246e-06, "loss": 0.0209, "step": 12512 }, { "epoch": 12.7, "learning_rate": 1.2044389179009596e-06, "loss": 0.0222, "step": 12513 }, { "epoch": 12.7, "learning_rate": 1.2033961073803368e-06, "loss": 0.0219, "step": 12514 }, { "epoch": 12.7, "learning_rate": 1.2023537195884616e-06, "loss": 0.0212, "step": 12515 }, { "epoch": 12.7, "learning_rate": 1.201311754575436e-06, "loss": 0.0195, "step": 12516 }, { "epoch": 12.7, "learning_rate": 1.2002702123913257e-06, "loss": 0.0151, "step": 12517 }, { "epoch": 12.7, "learning_rate": 1.1992290930861872e-06, "loss": 0.0206, "step": 12518 }, { "epoch": 12.7, "learning_rate": 1.1981883967100517e-06, "loss": 0.0186, "step": 12519 }, { "epoch": 12.7, "learning_rate": 1.1971481233129278e-06, "loss": 0.0196, "step": 12520 }, { "epoch": 12.71, "learning_rate": 1.1961082729448115e-06, "loss": 0.0226, "step": 12521 }, { "epoch": 12.71, "learning_rate": 1.195068845655668e-06, "loss": 0.0198, "step": 12522 }, { "epoch": 12.71, "learning_rate": 1.194029841495452e-06, "loss": 0.0184, "step": 12523 }, { "epoch": 12.71, "learning_rate": 1.1929912605140926e-06, "loss": 0.0194, "step": 12524 }, { "epoch": 12.71, "learning_rate": 1.1919531027614995e-06, "loss": 0.0182, "step": 12525 }, { "epoch": 12.71, "learning_rate": 1.1909153682875664e-06, "loss": 0.0195, "step": 12526 }, { "epoch": 12.71, "learning_rate": 1.1898780571421554e-06, "loss": 0.0248, "step": 12527 }, { "epoch": 12.71, "learning_rate": 1.1888411693751211e-06, "loss": 0.0227, "step": 12528 }, { "epoch": 12.71, "learning_rate": 1.1878047050362885e-06, "loss": 0.0224, "step": 12529 }, { "epoch": 12.71, "learning_rate": 1.1867686641754672e-06, "loss": 0.0211, "step": 12530 }, { "epoch": 12.72, "learning_rate": 1.1857330468424466e-06, "loss": 0.0182, "step": 12531 }, { "epoch": 12.72, "learning_rate": 1.18469785308699e-06, "loss": 0.0206, "step": 12532 }, { "epoch": 12.72, "learning_rate": 1.1836630829588503e-06, "loss": 0.0218, "step": 12533 }, { "epoch": 12.72, "learning_rate": 1.1826287365077505e-06, "loss": 0.0194, "step": 12534 }, { "epoch": 12.72, "learning_rate": 1.1815948137833966e-06, "loss": 0.016, "step": 12535 }, { "epoch": 12.72, "learning_rate": 1.1805613148354788e-06, "loss": 0.0229, "step": 12536 }, { "epoch": 12.72, "learning_rate": 1.1795282397136576e-06, "loss": 0.0233, "step": 12537 }, { "epoch": 12.72, "learning_rate": 1.178495588467583e-06, "loss": 0.0169, "step": 12538 }, { "epoch": 12.72, "learning_rate": 1.1774633611468767e-06, "loss": 0.0196, "step": 12539 }, { "epoch": 12.72, "learning_rate": 1.1764315578011431e-06, "loss": 0.0221, "step": 12540 }, { "epoch": 12.73, "learning_rate": 1.1754001784799685e-06, "loss": 0.0188, "step": 12541 }, { "epoch": 12.73, "learning_rate": 1.1743692232329163e-06, "loss": 0.0177, "step": 12542 }, { "epoch": 12.73, "learning_rate": 1.1733386921095313e-06, "loss": 0.0183, "step": 12543 }, { "epoch": 12.73, "learning_rate": 1.172308585159333e-06, "loss": 0.018, "step": 12544 }, { "epoch": 12.73, "learning_rate": 1.1712789024318261e-06, "loss": 0.0153, "step": 12545 }, { "epoch": 12.73, "learning_rate": 1.1702496439764954e-06, "loss": 0.0219, "step": 12546 }, { "epoch": 12.73, "learning_rate": 1.169220809842797e-06, "loss": 0.0202, "step": 12547 }, { "epoch": 12.73, "learning_rate": 1.1681924000801791e-06, "loss": 0.0238, "step": 12548 }, { "epoch": 12.73, "learning_rate": 1.1671644147380557e-06, "loss": 0.0234, "step": 12549 }, { "epoch": 12.73, "learning_rate": 1.1661368538658357e-06, "loss": 0.0183, "step": 12550 }, { "epoch": 12.74, "learning_rate": 1.1651097175128923e-06, "loss": 0.0184, "step": 12551 }, { "epoch": 12.74, "learning_rate": 1.164083005728589e-06, "loss": 0.0199, "step": 12552 }, { "epoch": 12.74, "learning_rate": 1.1630567185622677e-06, "loss": 0.0189, "step": 12553 }, { "epoch": 12.74, "learning_rate": 1.162030856063241e-06, "loss": 0.0189, "step": 12554 }, { "epoch": 12.74, "learning_rate": 1.161005418280815e-06, "loss": 0.0246, "step": 12555 }, { "epoch": 12.74, "learning_rate": 1.1599804052642617e-06, "loss": 0.0195, "step": 12556 }, { "epoch": 12.74, "learning_rate": 1.1589558170628411e-06, "loss": 0.0176, "step": 12557 }, { "epoch": 12.74, "learning_rate": 1.157931653725791e-06, "loss": 0.0191, "step": 12558 }, { "epoch": 12.74, "learning_rate": 1.1569079153023299e-06, "loss": 0.02, "step": 12559 }, { "epoch": 12.74, "learning_rate": 1.1558846018416547e-06, "loss": 0.0188, "step": 12560 }, { "epoch": 12.75, "learning_rate": 1.1548617133929385e-06, "loss": 0.0183, "step": 12561 }, { "epoch": 12.75, "learning_rate": 1.1538392500053386e-06, "loss": 0.0215, "step": 12562 }, { "epoch": 12.75, "learning_rate": 1.1528172117279933e-06, "loss": 0.0195, "step": 12563 }, { "epoch": 12.75, "learning_rate": 1.1517955986100126e-06, "loss": 0.0195, "step": 12564 }, { "epoch": 12.75, "learning_rate": 1.1507744107004937e-06, "loss": 0.0203, "step": 12565 }, { "epoch": 12.75, "learning_rate": 1.1497536480485106e-06, "loss": 0.0241, "step": 12566 }, { "epoch": 12.75, "learning_rate": 1.1487333107031185e-06, "loss": 0.0204, "step": 12567 }, { "epoch": 12.75, "learning_rate": 1.1477133987133472e-06, "loss": 0.0202, "step": 12568 }, { "epoch": 12.75, "learning_rate": 1.1466939121282117e-06, "loss": 0.0198, "step": 12569 }, { "epoch": 12.75, "learning_rate": 1.1456748509967053e-06, "loss": 0.0222, "step": 12570 }, { "epoch": 12.76, "learning_rate": 1.1446562153677964e-06, "loss": 0.0219, "step": 12571 }, { "epoch": 12.76, "learning_rate": 1.1436380052904394e-06, "loss": 0.02, "step": 12572 }, { "epoch": 12.76, "learning_rate": 1.1426202208135662e-06, "loss": 0.0221, "step": 12573 }, { "epoch": 12.76, "learning_rate": 1.141602861986081e-06, "loss": 0.0177, "step": 12574 }, { "epoch": 12.76, "learning_rate": 1.1405859288568832e-06, "loss": 0.02, "step": 12575 }, { "epoch": 12.76, "learning_rate": 1.1395694214748354e-06, "loss": 0.0206, "step": 12576 }, { "epoch": 12.76, "learning_rate": 1.1385533398887904e-06, "loss": 0.0202, "step": 12577 }, { "epoch": 12.76, "learning_rate": 1.137537684147575e-06, "loss": 0.022, "step": 12578 }, { "epoch": 12.76, "learning_rate": 1.1365224542999974e-06, "loss": 0.0184, "step": 12579 }, { "epoch": 12.77, "learning_rate": 1.1355076503948482e-06, "loss": 0.0189, "step": 12580 }, { "epoch": 12.77, "learning_rate": 1.134493272480891e-06, "loss": 0.0218, "step": 12581 }, { "epoch": 12.77, "learning_rate": 1.1334793206068739e-06, "loss": 0.0227, "step": 12582 }, { "epoch": 12.77, "learning_rate": 1.1324657948215245e-06, "loss": 0.0185, "step": 12583 }, { "epoch": 12.77, "learning_rate": 1.1314526951735483e-06, "loss": 0.0222, "step": 12584 }, { "epoch": 12.77, "learning_rate": 1.1304400217116297e-06, "loss": 0.0208, "step": 12585 }, { "epoch": 12.77, "learning_rate": 1.129427774484434e-06, "loss": 0.0215, "step": 12586 }, { "epoch": 12.77, "learning_rate": 1.128415953540607e-06, "loss": 0.0216, "step": 12587 }, { "epoch": 12.77, "learning_rate": 1.1274045589287708e-06, "loss": 0.0188, "step": 12588 }, { "epoch": 12.77, "learning_rate": 1.1263935906975288e-06, "loss": 0.019, "step": 12589 }, { "epoch": 12.78, "learning_rate": 1.1253830488954641e-06, "loss": 0.0202, "step": 12590 }, { "epoch": 12.78, "learning_rate": 1.1243729335711407e-06, "loss": 0.0195, "step": 12591 }, { "epoch": 12.78, "learning_rate": 1.1233632447731024e-06, "loss": 0.0157, "step": 12592 }, { "epoch": 12.78, "learning_rate": 1.1223539825498652e-06, "loss": 0.0211, "step": 12593 }, { "epoch": 12.78, "learning_rate": 1.1213451469499358e-06, "loss": 0.0149, "step": 12594 }, { "epoch": 12.78, "learning_rate": 1.12033673802179e-06, "loss": 0.0199, "step": 12595 }, { "epoch": 12.78, "learning_rate": 1.1193287558138888e-06, "loss": 0.0196, "step": 12596 }, { "epoch": 12.78, "learning_rate": 1.1183212003746746e-06, "loss": 0.0179, "step": 12597 }, { "epoch": 12.78, "learning_rate": 1.1173140717525631e-06, "loss": 0.0176, "step": 12598 }, { "epoch": 12.78, "learning_rate": 1.1163073699959548e-06, "loss": 0.0195, "step": 12599 }, { "epoch": 12.79, "learning_rate": 1.1153010951532261e-06, "loss": 0.0196, "step": 12600 }, { "epoch": 12.79, "learning_rate": 1.1142952472727354e-06, "loss": 0.0183, "step": 12601 }, { "epoch": 12.79, "learning_rate": 1.1132898264028213e-06, "loss": 0.0214, "step": 12602 }, { "epoch": 12.79, "learning_rate": 1.1122848325917967e-06, "loss": 0.021, "step": 12603 }, { "epoch": 12.79, "learning_rate": 1.1112802658879607e-06, "loss": 0.0176, "step": 12604 }, { "epoch": 12.79, "learning_rate": 1.1102761263395857e-06, "loss": 0.0151, "step": 12605 }, { "epoch": 12.79, "learning_rate": 1.1092724139949284e-06, "loss": 0.0212, "step": 12606 }, { "epoch": 12.79, "learning_rate": 1.108269128902223e-06, "loss": 0.0207, "step": 12607 }, { "epoch": 12.79, "learning_rate": 1.1072662711096816e-06, "loss": 0.0181, "step": 12608 }, { "epoch": 12.79, "learning_rate": 1.1062638406655024e-06, "loss": 0.019, "step": 12609 }, { "epoch": 12.8, "learning_rate": 1.1052618376178514e-06, "loss": 0.0205, "step": 12610 }, { "epoch": 12.8, "learning_rate": 1.1042602620148868e-06, "loss": 0.0193, "step": 12611 }, { "epoch": 12.8, "learning_rate": 1.1032591139047344e-06, "loss": 0.0238, "step": 12612 }, { "epoch": 12.8, "learning_rate": 1.102258393335509e-06, "loss": 0.0179, "step": 12613 }, { "epoch": 12.8, "learning_rate": 1.1012581003553013e-06, "loss": 0.0239, "step": 12614 }, { "epoch": 12.8, "learning_rate": 1.100258235012177e-06, "loss": 0.0186, "step": 12615 }, { "epoch": 12.8, "learning_rate": 1.0992587973541935e-06, "loss": 0.016, "step": 12616 }, { "epoch": 12.8, "learning_rate": 1.098259787429372e-06, "loss": 0.0218, "step": 12617 }, { "epoch": 12.8, "learning_rate": 1.0972612052857246e-06, "loss": 0.0209, "step": 12618 }, { "epoch": 12.8, "learning_rate": 1.0962630509712402e-06, "loss": 0.0153, "step": 12619 }, { "epoch": 12.81, "learning_rate": 1.0952653245338818e-06, "loss": 0.0184, "step": 12620 }, { "epoch": 12.81, "learning_rate": 1.094268026021601e-06, "loss": 0.0167, "step": 12621 }, { "epoch": 12.81, "learning_rate": 1.0932711554823184e-06, "loss": 0.0236, "step": 12622 }, { "epoch": 12.81, "learning_rate": 1.0922747129639433e-06, "loss": 0.019, "step": 12623 }, { "epoch": 12.81, "learning_rate": 1.0912786985143587e-06, "loss": 0.0186, "step": 12624 }, { "epoch": 12.81, "learning_rate": 1.0902831121814305e-06, "loss": 0.0184, "step": 12625 }, { "epoch": 12.81, "learning_rate": 1.089287954013004e-06, "loss": 0.0218, "step": 12626 }, { "epoch": 12.81, "learning_rate": 1.0882932240568988e-06, "loss": 0.0201, "step": 12627 }, { "epoch": 12.81, "learning_rate": 1.087298922360921e-06, "loss": 0.0236, "step": 12628 }, { "epoch": 12.81, "learning_rate": 1.0863050489728489e-06, "loss": 0.0222, "step": 12629 }, { "epoch": 12.82, "learning_rate": 1.0853116039404455e-06, "loss": 0.0188, "step": 12630 }, { "epoch": 12.82, "learning_rate": 1.0843185873114548e-06, "loss": 0.0206, "step": 12631 }, { "epoch": 12.82, "learning_rate": 1.0833259991335898e-06, "loss": 0.0186, "step": 12632 }, { "epoch": 12.82, "learning_rate": 1.082333839454559e-06, "loss": 0.0201, "step": 12633 }, { "epoch": 12.82, "learning_rate": 1.0813421083220354e-06, "loss": 0.0216, "step": 12634 }, { "epoch": 12.82, "learning_rate": 1.0803508057836798e-06, "loss": 0.0208, "step": 12635 }, { "epoch": 12.82, "learning_rate": 1.0793599318871317e-06, "loss": 0.024, "step": 12636 }, { "epoch": 12.82, "learning_rate": 1.0783694866800044e-06, "loss": 0.0173, "step": 12637 }, { "epoch": 12.82, "learning_rate": 1.0773794702098993e-06, "loss": 0.0166, "step": 12638 }, { "epoch": 12.82, "learning_rate": 1.0763898825243878e-06, "loss": 0.0228, "step": 12639 }, { "epoch": 12.83, "learning_rate": 1.0754007236710274e-06, "loss": 0.0195, "step": 12640 }, { "epoch": 12.83, "learning_rate": 1.074411993697354e-06, "loss": 0.0216, "step": 12641 }, { "epoch": 12.83, "learning_rate": 1.0734236926508802e-06, "loss": 0.0226, "step": 12642 }, { "epoch": 12.83, "learning_rate": 1.0724358205791042e-06, "loss": 0.0221, "step": 12643 }, { "epoch": 12.83, "learning_rate": 1.0714483775294926e-06, "loss": 0.0211, "step": 12644 }, { "epoch": 12.83, "learning_rate": 1.0704613635495009e-06, "loss": 0.0172, "step": 12645 }, { "epoch": 12.83, "learning_rate": 1.0694747786865633e-06, "loss": 0.0171, "step": 12646 }, { "epoch": 12.83, "learning_rate": 1.0684886229880864e-06, "loss": 0.0145, "step": 12647 }, { "epoch": 12.83, "learning_rate": 1.0675028965014656e-06, "loss": 0.0221, "step": 12648 }, { "epoch": 12.84, "learning_rate": 1.066517599274064e-06, "loss": 0.0229, "step": 12649 }, { "epoch": 12.84, "learning_rate": 1.0655327313532393e-06, "loss": 0.0255, "step": 12650 }, { "epoch": 12.84, "learning_rate": 1.0645482927863149e-06, "loss": 0.0215, "step": 12651 }, { "epoch": 12.84, "learning_rate": 1.0635642836205994e-06, "loss": 0.0194, "step": 12652 }, { "epoch": 12.84, "learning_rate": 1.062580703903383e-06, "loss": 0.0238, "step": 12653 }, { "epoch": 12.84, "learning_rate": 1.061597553681929e-06, "loss": 0.0138, "step": 12654 }, { "epoch": 12.84, "learning_rate": 1.0606148330034872e-06, "loss": 0.0178, "step": 12655 }, { "epoch": 12.84, "learning_rate": 1.0596325419152798e-06, "loss": 0.0195, "step": 12656 }, { "epoch": 12.84, "learning_rate": 1.0586506804645113e-06, "loss": 0.0203, "step": 12657 }, { "epoch": 12.84, "learning_rate": 1.0576692486983686e-06, "loss": 0.0234, "step": 12658 }, { "epoch": 12.85, "learning_rate": 1.0566882466640138e-06, "loss": 0.0209, "step": 12659 }, { "epoch": 12.85, "learning_rate": 1.0557076744085937e-06, "loss": 0.0213, "step": 12660 }, { "epoch": 12.85, "learning_rate": 1.054727531979224e-06, "loss": 0.0206, "step": 12661 }, { "epoch": 12.85, "learning_rate": 1.0537478194230088e-06, "loss": 0.0197, "step": 12662 }, { "epoch": 12.85, "learning_rate": 1.0527685367870333e-06, "loss": 0.0145, "step": 12663 }, { "epoch": 12.85, "learning_rate": 1.0517896841183518e-06, "loss": 0.0204, "step": 12664 }, { "epoch": 12.85, "learning_rate": 1.0508112614640087e-06, "loss": 0.0207, "step": 12665 }, { "epoch": 12.85, "learning_rate": 1.0498332688710168e-06, "loss": 0.0223, "step": 12666 }, { "epoch": 12.85, "learning_rate": 1.0488557063863814e-06, "loss": 0.0194, "step": 12667 }, { "epoch": 12.85, "learning_rate": 1.0478785740570762e-06, "loss": 0.0204, "step": 12668 }, { "epoch": 12.86, "learning_rate": 1.0469018719300595e-06, "loss": 0.0232, "step": 12669 }, { "epoch": 12.86, "learning_rate": 1.0459256000522678e-06, "loss": 0.0196, "step": 12670 }, { "epoch": 12.86, "learning_rate": 1.0449497584706147e-06, "loss": 0.0211, "step": 12671 }, { "epoch": 12.86, "learning_rate": 1.0439743472319997e-06, "loss": 0.0198, "step": 12672 }, { "epoch": 12.86, "learning_rate": 1.0429993663832905e-06, "loss": 0.0207, "step": 12673 }, { "epoch": 12.86, "learning_rate": 1.0420248159713431e-06, "loss": 0.0156, "step": 12674 }, { "epoch": 12.86, "learning_rate": 1.0410506960429945e-06, "loss": 0.0207, "step": 12675 }, { "epoch": 12.86, "learning_rate": 1.0400770066450527e-06, "loss": 0.0189, "step": 12676 }, { "epoch": 12.86, "learning_rate": 1.0391037478243126e-06, "loss": 0.0194, "step": 12677 }, { "epoch": 12.86, "learning_rate": 1.0381309196275401e-06, "loss": 0.0201, "step": 12678 }, { "epoch": 12.87, "learning_rate": 1.037158522101488e-06, "loss": 0.0235, "step": 12679 }, { "epoch": 12.87, "learning_rate": 1.0361865552928885e-06, "loss": 0.0226, "step": 12680 }, { "epoch": 12.87, "learning_rate": 1.0352150192484456e-06, "loss": 0.0161, "step": 12681 }, { "epoch": 12.87, "learning_rate": 1.0342439140148485e-06, "loss": 0.0168, "step": 12682 }, { "epoch": 12.87, "learning_rate": 1.0332732396387667e-06, "loss": 0.0209, "step": 12683 }, { "epoch": 12.87, "learning_rate": 1.0323029961668463e-06, "loss": 0.0209, "step": 12684 }, { "epoch": 12.87, "learning_rate": 1.0313331836457118e-06, "loss": 0.0179, "step": 12685 }, { "epoch": 12.87, "learning_rate": 1.0303638021219686e-06, "loss": 0.0163, "step": 12686 }, { "epoch": 12.87, "learning_rate": 1.0293948516422035e-06, "loss": 0.021, "step": 12687 }, { "epoch": 12.87, "learning_rate": 1.0284263322529763e-06, "loss": 0.0194, "step": 12688 }, { "epoch": 12.88, "learning_rate": 1.027458244000833e-06, "loss": 0.0246, "step": 12689 }, { "epoch": 12.88, "learning_rate": 1.0264905869322961e-06, "loss": 0.0224, "step": 12690 }, { "epoch": 12.88, "learning_rate": 1.025523361093863e-06, "loss": 0.0186, "step": 12691 }, { "epoch": 12.88, "learning_rate": 1.0245565665320223e-06, "loss": 0.021, "step": 12692 }, { "epoch": 12.88, "learning_rate": 1.0235902032932277e-06, "loss": 0.0198, "step": 12693 }, { "epoch": 12.88, "learning_rate": 1.0226242714239233e-06, "loss": 0.0217, "step": 12694 }, { "epoch": 12.88, "learning_rate": 1.0216587709705217e-06, "loss": 0.0212, "step": 12695 }, { "epoch": 12.88, "learning_rate": 1.020693701979426e-06, "loss": 0.0193, "step": 12696 }, { "epoch": 12.88, "learning_rate": 1.0197290644970137e-06, "loss": 0.0194, "step": 12697 }, { "epoch": 12.88, "learning_rate": 1.0187648585696374e-06, "loss": 0.0202, "step": 12698 }, { "epoch": 12.89, "learning_rate": 1.0178010842436358e-06, "loss": 0.02, "step": 12699 }, { "epoch": 12.89, "learning_rate": 1.016837741565323e-06, "loss": 0.02, "step": 12700 }, { "epoch": 12.89, "learning_rate": 1.0158748305809952e-06, "loss": 0.0214, "step": 12701 }, { "epoch": 12.89, "learning_rate": 1.014912351336923e-06, "loss": 0.0223, "step": 12702 }, { "epoch": 12.89, "learning_rate": 1.0139503038793597e-06, "loss": 0.0227, "step": 12703 }, { "epoch": 12.89, "learning_rate": 1.0129886882545414e-06, "loss": 0.0182, "step": 12704 }, { "epoch": 12.89, "learning_rate": 1.0120275045086736e-06, "loss": 0.02, "step": 12705 }, { "epoch": 12.89, "learning_rate": 1.0110667526879492e-06, "loss": 0.0233, "step": 12706 }, { "epoch": 12.89, "learning_rate": 1.010106432838538e-06, "loss": 0.0198, "step": 12707 }, { "epoch": 12.89, "learning_rate": 1.0091465450065897e-06, "loss": 0.0181, "step": 12708 }, { "epoch": 12.9, "learning_rate": 1.008187089238234e-06, "loss": 0.0198, "step": 12709 }, { "epoch": 12.9, "learning_rate": 1.0072280655795752e-06, "loss": 0.024, "step": 12710 }, { "epoch": 12.9, "learning_rate": 1.006269474076703e-06, "loss": 0.0235, "step": 12711 }, { "epoch": 12.9, "learning_rate": 1.0053113147756798e-06, "loss": 0.0226, "step": 12712 }, { "epoch": 12.9, "learning_rate": 1.004353587722553e-06, "loss": 0.0184, "step": 12713 }, { "epoch": 12.9, "learning_rate": 1.0033962929633489e-06, "loss": 0.0196, "step": 12714 }, { "epoch": 12.9, "learning_rate": 1.0024394305440677e-06, "loss": 0.0213, "step": 12715 }, { "epoch": 12.9, "learning_rate": 1.0014830005106935e-06, "loss": 0.0202, "step": 12716 }, { "epoch": 12.9, "learning_rate": 1.0005270029091885e-06, "loss": 0.02, "step": 12717 }, { "epoch": 12.91, "learning_rate": 9.995714377854936e-07, "loss": 0.0192, "step": 12718 }, { "epoch": 12.91, "learning_rate": 9.986163051855324e-07, "loss": 0.0193, "step": 12719 }, { "epoch": 12.91, "learning_rate": 9.976616051552e-07, "loss": 0.0213, "step": 12720 }, { "epoch": 12.91, "learning_rate": 9.967073377403792e-07, "loss": 0.0195, "step": 12721 }, { "epoch": 12.91, "learning_rate": 9.95753502986926e-07, "loss": 0.0188, "step": 12722 }, { "epoch": 12.91, "learning_rate": 9.948001009406772e-07, "loss": 0.0218, "step": 12723 }, { "epoch": 12.91, "learning_rate": 9.938471316474507e-07, "loss": 0.0198, "step": 12724 }, { "epoch": 12.91, "learning_rate": 9.928945951530422e-07, "loss": 0.0199, "step": 12725 }, { "epoch": 12.91, "learning_rate": 9.919424915032283e-07, "loss": 0.0211, "step": 12726 }, { "epoch": 12.91, "learning_rate": 9.909908207437592e-07, "loss": 0.0226, "step": 12727 }, { "epoch": 12.92, "learning_rate": 9.900395829203736e-07, "loss": 0.021, "step": 12728 }, { "epoch": 12.92, "learning_rate": 9.890887780787784e-07, "loss": 0.02, "step": 12729 }, { "epoch": 12.92, "learning_rate": 9.881384062646693e-07, "loss": 0.0204, "step": 12730 }, { "epoch": 12.92, "learning_rate": 9.871884675237165e-07, "loss": 0.0247, "step": 12731 }, { "epoch": 12.92, "learning_rate": 9.862389619015666e-07, "loss": 0.0154, "step": 12732 }, { "epoch": 12.92, "learning_rate": 9.852898894438567e-07, "loss": 0.0209, "step": 12733 }, { "epoch": 12.92, "learning_rate": 9.84341250196188e-07, "loss": 0.0215, "step": 12734 }, { "epoch": 12.92, "learning_rate": 9.833930442041506e-07, "loss": 0.019, "step": 12735 }, { "epoch": 12.92, "learning_rate": 9.824452715133137e-07, "loss": 0.0199, "step": 12736 }, { "epoch": 12.92, "learning_rate": 9.814979321692208e-07, "loss": 0.0212, "step": 12737 }, { "epoch": 12.93, "learning_rate": 9.80551026217399e-07, "loss": 0.0174, "step": 12738 }, { "epoch": 12.93, "learning_rate": 9.796045537033495e-07, "loss": 0.0211, "step": 12739 }, { "epoch": 12.93, "learning_rate": 9.78658514672557e-07, "loss": 0.021, "step": 12740 }, { "epoch": 12.93, "learning_rate": 9.777129091704862e-07, "loss": 0.0214, "step": 12741 }, { "epoch": 12.93, "learning_rate": 9.767677372425776e-07, "loss": 0.022, "step": 12742 }, { "epoch": 12.93, "learning_rate": 9.75822998934255e-07, "loss": 0.0221, "step": 12743 }, { "epoch": 12.93, "learning_rate": 9.748786942909139e-07, "loss": 0.0177, "step": 12744 }, { "epoch": 12.93, "learning_rate": 9.739348233579382e-07, "loss": 0.0188, "step": 12745 }, { "epoch": 12.93, "learning_rate": 9.729913861806829e-07, "loss": 0.0193, "step": 12746 }, { "epoch": 12.93, "learning_rate": 9.720483828044868e-07, "loss": 0.0193, "step": 12747 }, { "epoch": 12.94, "learning_rate": 9.711058132746688e-07, "loss": 0.0192, "step": 12748 }, { "epoch": 12.94, "learning_rate": 9.701636776365197e-07, "loss": 0.0213, "step": 12749 }, { "epoch": 12.94, "learning_rate": 9.692219759353227e-07, "loss": 0.0194, "step": 12750 }, { "epoch": 12.94, "learning_rate": 9.682807082163259e-07, "loss": 0.016, "step": 12751 }, { "epoch": 12.94, "learning_rate": 9.673398745247643e-07, "loss": 0.0216, "step": 12752 }, { "epoch": 12.94, "learning_rate": 9.663994749058526e-07, "loss": 0.0189, "step": 12753 }, { "epoch": 12.94, "learning_rate": 9.654595094047791e-07, "loss": 0.0194, "step": 12754 }, { "epoch": 12.94, "learning_rate": 9.645199780667181e-07, "loss": 0.0234, "step": 12755 }, { "epoch": 12.94, "learning_rate": 9.635808809368153e-07, "loss": 0.023, "step": 12756 }, { "epoch": 12.94, "learning_rate": 9.626422180602034e-07, "loss": 0.0215, "step": 12757 }, { "epoch": 12.95, "learning_rate": 9.617039894819901e-07, "loss": 0.0225, "step": 12758 }, { "epoch": 12.95, "learning_rate": 9.607661952472613e-07, "loss": 0.0226, "step": 12759 }, { "epoch": 12.95, "learning_rate": 9.598288354010877e-07, "loss": 0.0216, "step": 12760 }, { "epoch": 12.95, "learning_rate": 9.588919099885097e-07, "loss": 0.0186, "step": 12761 }, { "epoch": 12.95, "learning_rate": 9.579554190545548e-07, "loss": 0.0186, "step": 12762 }, { "epoch": 12.95, "learning_rate": 9.57019362644228e-07, "loss": 0.0165, "step": 12763 }, { "epoch": 12.95, "learning_rate": 9.560837408025092e-07, "loss": 0.0212, "step": 12764 }, { "epoch": 12.95, "learning_rate": 9.551485535743631e-07, "loss": 0.0202, "step": 12765 }, { "epoch": 12.95, "learning_rate": 9.542138010047274e-07, "loss": 0.0187, "step": 12766 }, { "epoch": 12.95, "learning_rate": 9.532794831385294e-07, "loss": 0.0203, "step": 12767 }, { "epoch": 12.96, "learning_rate": 9.52345600020661e-07, "loss": 0.0221, "step": 12768 }, { "epoch": 12.96, "learning_rate": 9.514121516960051e-07, "loss": 0.0216, "step": 12769 }, { "epoch": 12.96, "learning_rate": 9.504791382094203e-07, "loss": 0.0214, "step": 12770 }, { "epoch": 12.96, "learning_rate": 9.495465596057396e-07, "loss": 0.0243, "step": 12771 }, { "epoch": 12.96, "learning_rate": 9.486144159297828e-07, "loss": 0.0181, "step": 12772 }, { "epoch": 12.96, "learning_rate": 9.476827072263405e-07, "loss": 0.0162, "step": 12773 }, { "epoch": 12.96, "learning_rate": 9.467514335401906e-07, "loss": 0.0168, "step": 12774 }, { "epoch": 12.96, "learning_rate": 9.458205949160837e-07, "loss": 0.0186, "step": 12775 }, { "epoch": 12.96, "learning_rate": 9.44890191398754e-07, "loss": 0.0204, "step": 12776 }, { "epoch": 12.96, "learning_rate": 9.439602230329137e-07, "loss": 0.0201, "step": 12777 }, { "epoch": 12.97, "learning_rate": 9.430306898632502e-07, "loss": 0.0228, "step": 12778 }, { "epoch": 12.97, "learning_rate": 9.421015919344345e-07, "loss": 0.0212, "step": 12779 }, { "epoch": 12.97, "learning_rate": 9.411729292911175e-07, "loss": 0.0213, "step": 12780 }, { "epoch": 12.97, "learning_rate": 9.402447019779226e-07, "loss": 0.0223, "step": 12781 }, { "epoch": 12.97, "learning_rate": 9.393169100394617e-07, "loss": 0.0205, "step": 12782 }, { "epoch": 12.97, "learning_rate": 9.383895535203136e-07, "loss": 0.0245, "step": 12783 }, { "epoch": 12.97, "learning_rate": 9.374626324650504e-07, "loss": 0.0204, "step": 12784 }, { "epoch": 12.97, "learning_rate": 9.365361469182122e-07, "loss": 0.0196, "step": 12785 }, { "epoch": 12.97, "learning_rate": 9.356100969243231e-07, "loss": 0.019, "step": 12786 }, { "epoch": 12.98, "learning_rate": 9.346844825278867e-07, "loss": 0.0214, "step": 12787 }, { "epoch": 12.98, "learning_rate": 9.337593037733817e-07, "loss": 0.0189, "step": 12788 }, { "epoch": 12.98, "learning_rate": 9.328345607052692e-07, "loss": 0.0241, "step": 12789 }, { "epoch": 12.98, "learning_rate": 9.319102533679902e-07, "loss": 0.018, "step": 12790 }, { "epoch": 12.98, "learning_rate": 9.309863818059583e-07, "loss": 0.0204, "step": 12791 }, { "epoch": 12.98, "learning_rate": 9.300629460635791e-07, "loss": 0.0168, "step": 12792 }, { "epoch": 12.98, "learning_rate": 9.291399461852213e-07, "loss": 0.0156, "step": 12793 }, { "epoch": 12.98, "learning_rate": 9.282173822152462e-07, "loss": 0.0244, "step": 12794 }, { "epoch": 12.98, "learning_rate": 9.27295254197984e-07, "loss": 0.0238, "step": 12795 }, { "epoch": 12.98, "learning_rate": 9.263735621777514e-07, "loss": 0.0181, "step": 12796 }, { "epoch": 12.99, "learning_rate": 9.254523061988407e-07, "loss": 0.021, "step": 12797 }, { "epoch": 12.99, "learning_rate": 9.245314863055221e-07, "loss": 0.0204, "step": 12798 }, { "epoch": 12.99, "learning_rate": 9.236111025420468e-07, "loss": 0.024, "step": 12799 }, { "epoch": 12.99, "learning_rate": 9.226911549526452e-07, "loss": 0.0192, "step": 12800 }, { "epoch": 12.99, "learning_rate": 9.217716435815283e-07, "loss": 0.0208, "step": 12801 }, { "epoch": 12.99, "learning_rate": 9.208525684728809e-07, "loss": 0.0185, "step": 12802 }, { "epoch": 12.99, "learning_rate": 9.199339296708698e-07, "loss": 0.0215, "step": 12803 }, { "epoch": 12.99, "learning_rate": 9.190157272196454e-07, "loss": 0.0179, "step": 12804 }, { "epoch": 12.99, "learning_rate": 9.180979611633278e-07, "loss": 0.0207, "step": 12805 }, { "epoch": 12.99, "learning_rate": 9.17180631546023e-07, "loss": 0.0186, "step": 12806 }, { "epoch": 13.0, "learning_rate": 9.162637384118156e-07, "loss": 0.0243, "step": 12807 }, { "epoch": 13.0, "learning_rate": 9.153472818047627e-07, "loss": 0.0207, "step": 12808 }, { "epoch": 13.0, "learning_rate": 9.144312617689121e-07, "loss": 0.0203, "step": 12809 }, { "epoch": 13.0, "learning_rate": 9.13515678348279e-07, "loss": 0.0202, "step": 12810 }, { "epoch": 13.0, "learning_rate": 9.126005315868669e-07, "loss": 0.0222, "step": 12811 }, { "epoch": 13.0, "learning_rate": 9.116858215286495e-07, "loss": 0.0175, "step": 12812 }, { "epoch": 13.0, "learning_rate": 9.107715482175871e-07, "loss": 0.0201, "step": 12813 }, { "epoch": 13.0, "learning_rate": 9.098577116976148e-07, "loss": 0.0195, "step": 12814 }, { "epoch": 13.0, "learning_rate": 9.089443120126473e-07, "loss": 0.0158, "step": 12815 }, { "epoch": 13.0, "learning_rate": 9.080313492065795e-07, "loss": 0.0151, "step": 12816 }, { "epoch": 13.01, "learning_rate": 9.071188233232842e-07, "loss": 0.0171, "step": 12817 }, { "epoch": 13.01, "learning_rate": 9.06206734406615e-07, "loss": 0.018, "step": 12818 }, { "epoch": 13.01, "learning_rate": 9.052950825004026e-07, "loss": 0.0157, "step": 12819 }, { "epoch": 13.01, "learning_rate": 9.043838676484562e-07, "loss": 0.012, "step": 12820 }, { "epoch": 13.01, "learning_rate": 9.034730898945676e-07, "loss": 0.0183, "step": 12821 }, { "epoch": 13.01, "learning_rate": 9.025627492825007e-07, "loss": 0.0184, "step": 12822 }, { "epoch": 13.01, "learning_rate": 9.016528458560059e-07, "loss": 0.0182, "step": 12823 }, { "epoch": 13.01, "learning_rate": 9.007433796588083e-07, "loss": 0.0182, "step": 12824 }, { "epoch": 13.01, "learning_rate": 8.99834350734613e-07, "loss": 0.0201, "step": 12825 }, { "epoch": 13.01, "learning_rate": 8.989257591271083e-07, "loss": 0.0187, "step": 12826 }, { "epoch": 13.02, "learning_rate": 8.980176048799505e-07, "loss": 0.0168, "step": 12827 }, { "epoch": 13.02, "learning_rate": 8.971098880367878e-07, "loss": 0.0175, "step": 12828 }, { "epoch": 13.02, "learning_rate": 8.962026086412368e-07, "loss": 0.0169, "step": 12829 }, { "epoch": 13.02, "learning_rate": 8.952957667368999e-07, "loss": 0.0167, "step": 12830 }, { "epoch": 13.02, "learning_rate": 8.943893623673571e-07, "loss": 0.0179, "step": 12831 }, { "epoch": 13.02, "learning_rate": 8.934833955761635e-07, "loss": 0.0192, "step": 12832 }, { "epoch": 13.02, "learning_rate": 8.925778664068585e-07, "loss": 0.0145, "step": 12833 }, { "epoch": 13.02, "learning_rate": 8.916727749029564e-07, "loss": 0.0161, "step": 12834 }, { "epoch": 13.02, "learning_rate": 8.907681211079544e-07, "loss": 0.0186, "step": 12835 }, { "epoch": 13.02, "learning_rate": 8.898639050653269e-07, "loss": 0.0169, "step": 12836 }, { "epoch": 13.03, "learning_rate": 8.889601268185233e-07, "loss": 0.0144, "step": 12837 }, { "epoch": 13.03, "learning_rate": 8.880567864109791e-07, "loss": 0.0189, "step": 12838 }, { "epoch": 13.03, "learning_rate": 8.871538838861016e-07, "loss": 0.0185, "step": 12839 }, { "epoch": 13.03, "learning_rate": 8.862514192872828e-07, "loss": 0.0171, "step": 12840 }, { "epoch": 13.03, "learning_rate": 8.853493926578916e-07, "loss": 0.0168, "step": 12841 }, { "epoch": 13.03, "learning_rate": 8.844478040412752e-07, "loss": 0.0196, "step": 12842 }, { "epoch": 13.03, "learning_rate": 8.835466534807613e-07, "loss": 0.0214, "step": 12843 }, { "epoch": 13.03, "learning_rate": 8.826459410196531e-07, "loss": 0.0192, "step": 12844 }, { "epoch": 13.03, "learning_rate": 8.817456667012381e-07, "loss": 0.0178, "step": 12845 }, { "epoch": 13.04, "learning_rate": 8.808458305687772e-07, "loss": 0.0193, "step": 12846 }, { "epoch": 13.04, "learning_rate": 8.799464326655138e-07, "loss": 0.0196, "step": 12847 }, { "epoch": 13.04, "learning_rate": 8.79047473034671e-07, "loss": 0.014, "step": 12848 }, { "epoch": 13.04, "learning_rate": 8.78148951719443e-07, "loss": 0.0162, "step": 12849 }, { "epoch": 13.04, "learning_rate": 8.772508687630188e-07, "loss": 0.019, "step": 12850 }, { "epoch": 13.04, "learning_rate": 8.763532242085482e-07, "loss": 0.0178, "step": 12851 }, { "epoch": 13.04, "learning_rate": 8.754560180991722e-07, "loss": 0.0173, "step": 12852 }, { "epoch": 13.04, "learning_rate": 8.745592504780076e-07, "loss": 0.0167, "step": 12853 }, { "epoch": 13.04, "learning_rate": 8.736629213881465e-07, "loss": 0.0186, "step": 12854 }, { "epoch": 13.04, "learning_rate": 8.727670308726654e-07, "loss": 0.0177, "step": 12855 }, { "epoch": 13.05, "learning_rate": 8.718715789746146e-07, "loss": 0.0195, "step": 12856 }, { "epoch": 13.05, "learning_rate": 8.709765657370273e-07, "loss": 0.0158, "step": 12857 }, { "epoch": 13.05, "learning_rate": 8.700819912029135e-07, "loss": 0.0172, "step": 12858 }, { "epoch": 13.05, "learning_rate": 8.691878554152644e-07, "loss": 0.0197, "step": 12859 }, { "epoch": 13.05, "learning_rate": 8.682941584170479e-07, "loss": 0.0192, "step": 12860 }, { "epoch": 13.05, "learning_rate": 8.674009002512107e-07, "loss": 0.0182, "step": 12861 }, { "epoch": 13.05, "learning_rate": 8.665080809606774e-07, "loss": 0.0164, "step": 12862 }, { "epoch": 13.05, "learning_rate": 8.656157005883592e-07, "loss": 0.0185, "step": 12863 }, { "epoch": 13.05, "learning_rate": 8.64723759177133e-07, "loss": 0.0188, "step": 12864 }, { "epoch": 13.05, "learning_rate": 8.638322567698676e-07, "loss": 0.0178, "step": 12865 }, { "epoch": 13.06, "learning_rate": 8.629411934093979e-07, "loss": 0.018, "step": 12866 }, { "epoch": 13.06, "learning_rate": 8.620505691385539e-07, "loss": 0.0165, "step": 12867 }, { "epoch": 13.06, "learning_rate": 8.611603840001293e-07, "loss": 0.0205, "step": 12868 }, { "epoch": 13.06, "learning_rate": 8.602706380369031e-07, "loss": 0.0172, "step": 12869 }, { "epoch": 13.06, "learning_rate": 8.593813312916366e-07, "loss": 0.0178, "step": 12870 }, { "epoch": 13.06, "learning_rate": 8.584924638070614e-07, "loss": 0.021, "step": 12871 }, { "epoch": 13.06, "learning_rate": 8.576040356258974e-07, "loss": 0.0193, "step": 12872 }, { "epoch": 13.06, "learning_rate": 8.567160467908342e-07, "loss": 0.0199, "step": 12873 }, { "epoch": 13.06, "learning_rate": 8.558284973445474e-07, "loss": 0.0185, "step": 12874 }, { "epoch": 13.06, "learning_rate": 8.549413873296886e-07, "loss": 0.0177, "step": 12875 }, { "epoch": 13.07, "learning_rate": 8.540547167888891e-07, "loss": 0.0159, "step": 12876 }, { "epoch": 13.07, "learning_rate": 8.531684857647593e-07, "loss": 0.0182, "step": 12877 }, { "epoch": 13.07, "learning_rate": 8.522826942998863e-07, "loss": 0.0174, "step": 12878 }, { "epoch": 13.07, "learning_rate": 8.513973424368371e-07, "loss": 0.0149, "step": 12879 }, { "epoch": 13.07, "learning_rate": 8.505124302181622e-07, "loss": 0.0161, "step": 12880 }, { "epoch": 13.07, "learning_rate": 8.49627957686382e-07, "loss": 0.0165, "step": 12881 }, { "epoch": 13.07, "learning_rate": 8.487439248840035e-07, "loss": 0.0189, "step": 12882 }, { "epoch": 13.07, "learning_rate": 8.478603318535061e-07, "loss": 0.0158, "step": 12883 }, { "epoch": 13.07, "learning_rate": 8.46977178637357e-07, "loss": 0.0181, "step": 12884 }, { "epoch": 13.07, "learning_rate": 8.460944652779924e-07, "loss": 0.0161, "step": 12885 }, { "epoch": 13.08, "learning_rate": 8.452121918178346e-07, "loss": 0.0177, "step": 12886 }, { "epoch": 13.08, "learning_rate": 8.443303582992824e-07, "loss": 0.0146, "step": 12887 }, { "epoch": 13.08, "learning_rate": 8.434489647647093e-07, "loss": 0.0169, "step": 12888 }, { "epoch": 13.08, "learning_rate": 8.425680112564771e-07, "loss": 0.0178, "step": 12889 }, { "epoch": 13.08, "learning_rate": 8.416874978169154e-07, "loss": 0.0198, "step": 12890 }, { "epoch": 13.08, "learning_rate": 8.40807424488338e-07, "loss": 0.0171, "step": 12891 }, { "epoch": 13.08, "learning_rate": 8.399277913130443e-07, "loss": 0.0191, "step": 12892 }, { "epoch": 13.08, "learning_rate": 8.390485983332997e-07, "loss": 0.0164, "step": 12893 }, { "epoch": 13.08, "learning_rate": 8.381698455913578e-07, "loss": 0.0139, "step": 12894 }, { "epoch": 13.08, "learning_rate": 8.372915331294451e-07, "loss": 0.017, "step": 12895 }, { "epoch": 13.09, "learning_rate": 8.364136609897711e-07, "loss": 0.0182, "step": 12896 }, { "epoch": 13.09, "learning_rate": 8.355362292145253e-07, "loss": 0.0157, "step": 12897 }, { "epoch": 13.09, "learning_rate": 8.346592378458685e-07, "loss": 0.0161, "step": 12898 }, { "epoch": 13.09, "learning_rate": 8.337826869259491e-07, "loss": 0.0155, "step": 12899 }, { "epoch": 13.09, "learning_rate": 8.329065764968869e-07, "loss": 0.0202, "step": 12900 }, { "epoch": 13.09, "learning_rate": 8.320309066007893e-07, "loss": 0.0154, "step": 12901 }, { "epoch": 13.09, "learning_rate": 8.311556772797324e-07, "loss": 0.0187, "step": 12902 }, { "epoch": 13.09, "learning_rate": 8.302808885757785e-07, "loss": 0.0173, "step": 12903 }, { "epoch": 13.09, "learning_rate": 8.294065405309692e-07, "loss": 0.0203, "step": 12904 }, { "epoch": 13.09, "learning_rate": 8.285326331873156e-07, "loss": 0.0163, "step": 12905 }, { "epoch": 13.1, "learning_rate": 8.276591665868195e-07, "loss": 0.0128, "step": 12906 }, { "epoch": 13.1, "learning_rate": 8.267861407714539e-07, "loss": 0.0161, "step": 12907 }, { "epoch": 13.1, "learning_rate": 8.25913555783171e-07, "loss": 0.0185, "step": 12908 }, { "epoch": 13.1, "learning_rate": 8.250414116639094e-07, "loss": 0.02, "step": 12909 }, { "epoch": 13.1, "learning_rate": 8.241697084555744e-07, "loss": 0.0151, "step": 12910 }, { "epoch": 13.1, "learning_rate": 8.232984462000626e-07, "loss": 0.0161, "step": 12911 }, { "epoch": 13.1, "learning_rate": 8.224276249392371e-07, "loss": 0.0178, "step": 12912 }, { "epoch": 13.1, "learning_rate": 8.215572447149489e-07, "loss": 0.0148, "step": 12913 }, { "epoch": 13.1, "learning_rate": 8.206873055690267e-07, "loss": 0.0161, "step": 12914 }, { "epoch": 13.11, "learning_rate": 8.198178075432739e-07, "loss": 0.0164, "step": 12915 }, { "epoch": 13.11, "learning_rate": 8.189487506794736e-07, "loss": 0.0153, "step": 12916 }, { "epoch": 13.11, "learning_rate": 8.180801350193923e-07, "loss": 0.0195, "step": 12917 }, { "epoch": 13.11, "learning_rate": 8.172119606047724e-07, "loss": 0.016, "step": 12918 }, { "epoch": 13.11, "learning_rate": 8.163442274773315e-07, "loss": 0.0183, "step": 12919 }, { "epoch": 13.11, "learning_rate": 8.154769356787706e-07, "loss": 0.0195, "step": 12920 }, { "epoch": 13.11, "learning_rate": 8.14610085250771e-07, "loss": 0.0143, "step": 12921 }, { "epoch": 13.11, "learning_rate": 8.13743676234986e-07, "loss": 0.0207, "step": 12922 }, { "epoch": 13.11, "learning_rate": 8.128777086730522e-07, "loss": 0.0183, "step": 12923 }, { "epoch": 13.11, "learning_rate": 8.120121826065885e-07, "loss": 0.0151, "step": 12924 }, { "epoch": 13.12, "learning_rate": 8.111470980771819e-07, "loss": 0.0179, "step": 12925 }, { "epoch": 13.12, "learning_rate": 8.10282455126411e-07, "loss": 0.0172, "step": 12926 }, { "epoch": 13.12, "learning_rate": 8.09418253795824e-07, "loss": 0.0174, "step": 12927 }, { "epoch": 13.12, "learning_rate": 8.08554494126953e-07, "loss": 0.0165, "step": 12928 }, { "epoch": 13.12, "learning_rate": 8.076911761613027e-07, "loss": 0.0172, "step": 12929 }, { "epoch": 13.12, "learning_rate": 8.068282999403632e-07, "loss": 0.0167, "step": 12930 }, { "epoch": 13.12, "learning_rate": 8.059658655056024e-07, "loss": 0.0181, "step": 12931 }, { "epoch": 13.12, "learning_rate": 8.051038728984617e-07, "loss": 0.0171, "step": 12932 }, { "epoch": 13.12, "learning_rate": 8.042423221603679e-07, "loss": 0.0153, "step": 12933 }, { "epoch": 13.12, "learning_rate": 8.033812133327213e-07, "loss": 0.0194, "step": 12934 }, { "epoch": 13.13, "learning_rate": 8.025205464569053e-07, "loss": 0.0173, "step": 12935 }, { "epoch": 13.13, "learning_rate": 8.016603215742813e-07, "loss": 0.0174, "step": 12936 }, { "epoch": 13.13, "learning_rate": 8.008005387261842e-07, "loss": 0.0173, "step": 12937 }, { "epoch": 13.13, "learning_rate": 7.99941197953934e-07, "loss": 0.021, "step": 12938 }, { "epoch": 13.13, "learning_rate": 7.990822992988267e-07, "loss": 0.0207, "step": 12939 }, { "epoch": 13.13, "learning_rate": 7.982238428021361e-07, "loss": 0.0169, "step": 12940 }, { "epoch": 13.13, "learning_rate": 7.973658285051178e-07, "loss": 0.0187, "step": 12941 }, { "epoch": 13.13, "learning_rate": 7.965082564490045e-07, "loss": 0.0161, "step": 12942 }, { "epoch": 13.13, "learning_rate": 7.956511266750088e-07, "loss": 0.0184, "step": 12943 }, { "epoch": 13.13, "learning_rate": 7.947944392243179e-07, "loss": 0.0154, "step": 12944 }, { "epoch": 13.14, "learning_rate": 7.93938194138103e-07, "loss": 0.0169, "step": 12945 }, { "epoch": 13.14, "learning_rate": 7.930823914575092e-07, "loss": 0.0176, "step": 12946 }, { "epoch": 13.14, "learning_rate": 7.922270312236646e-07, "loss": 0.017, "step": 12947 }, { "epoch": 13.14, "learning_rate": 7.913721134776752e-07, "loss": 0.0198, "step": 12948 }, { "epoch": 13.14, "learning_rate": 7.905176382606227e-07, "loss": 0.0154, "step": 12949 }, { "epoch": 13.14, "learning_rate": 7.896636056135709e-07, "loss": 0.0144, "step": 12950 }, { "epoch": 13.14, "learning_rate": 7.888100155775602e-07, "loss": 0.0158, "step": 12951 }, { "epoch": 13.14, "learning_rate": 7.879568681936123e-07, "loss": 0.0169, "step": 12952 }, { "epoch": 13.14, "learning_rate": 7.871041635027266e-07, "loss": 0.017, "step": 12953 }, { "epoch": 13.14, "learning_rate": 7.86251901545877e-07, "loss": 0.016, "step": 12954 }, { "epoch": 13.15, "learning_rate": 7.85400082364024e-07, "loss": 0.0129, "step": 12955 }, { "epoch": 13.15, "learning_rate": 7.845487059980972e-07, "loss": 0.0135, "step": 12956 }, { "epoch": 13.15, "learning_rate": 7.836977724890149e-07, "loss": 0.0171, "step": 12957 }, { "epoch": 13.15, "learning_rate": 7.828472818776678e-07, "loss": 0.0192, "step": 12958 }, { "epoch": 13.15, "learning_rate": 7.819972342049265e-07, "loss": 0.0193, "step": 12959 }, { "epoch": 13.15, "learning_rate": 7.811476295116438e-07, "loss": 0.0163, "step": 12960 }, { "epoch": 13.15, "learning_rate": 7.802984678386438e-07, "loss": 0.0185, "step": 12961 }, { "epoch": 13.15, "learning_rate": 7.794497492267362e-07, "loss": 0.0197, "step": 12962 }, { "epoch": 13.15, "learning_rate": 7.78601473716708e-07, "loss": 0.0157, "step": 12963 }, { "epoch": 13.15, "learning_rate": 7.777536413493203e-07, "loss": 0.0195, "step": 12964 }, { "epoch": 13.16, "learning_rate": 7.769062521653203e-07, "loss": 0.0171, "step": 12965 }, { "epoch": 13.16, "learning_rate": 7.760593062054256e-07, "loss": 0.0169, "step": 12966 }, { "epoch": 13.16, "learning_rate": 7.752128035103435e-07, "loss": 0.0158, "step": 12967 }, { "epoch": 13.16, "learning_rate": 7.74366744120747e-07, "loss": 0.0207, "step": 12968 }, { "epoch": 13.16, "learning_rate": 7.735211280772969e-07, "loss": 0.0194, "step": 12969 }, { "epoch": 13.16, "learning_rate": 7.72675955420632e-07, "loss": 0.0155, "step": 12970 }, { "epoch": 13.16, "learning_rate": 7.718312261913641e-07, "loss": 0.019, "step": 12971 }, { "epoch": 13.16, "learning_rate": 7.709869404300907e-07, "loss": 0.0157, "step": 12972 }, { "epoch": 13.16, "learning_rate": 7.701430981773816e-07, "loss": 0.0208, "step": 12973 }, { "epoch": 13.16, "learning_rate": 7.692996994737901e-07, "loss": 0.0151, "step": 12974 }, { "epoch": 13.17, "learning_rate": 7.684567443598457e-07, "loss": 0.0185, "step": 12975 }, { "epoch": 13.17, "learning_rate": 7.676142328760594e-07, "loss": 0.0184, "step": 12976 }, { "epoch": 13.17, "learning_rate": 7.667721650629179e-07, "loss": 0.0182, "step": 12977 }, { "epoch": 13.17, "learning_rate": 7.659305409608863e-07, "loss": 0.0201, "step": 12978 }, { "epoch": 13.17, "learning_rate": 7.650893606104104e-07, "loss": 0.016, "step": 12979 }, { "epoch": 13.17, "learning_rate": 7.642486240519154e-07, "loss": 0.0168, "step": 12980 }, { "epoch": 13.17, "learning_rate": 7.634083313258e-07, "loss": 0.0181, "step": 12981 }, { "epoch": 13.17, "learning_rate": 7.625684824724489e-07, "loss": 0.0192, "step": 12982 }, { "epoch": 13.17, "learning_rate": 7.617290775322184e-07, "loss": 0.0179, "step": 12983 }, { "epoch": 13.18, "learning_rate": 7.608901165454507e-07, "loss": 0.0136, "step": 12984 }, { "epoch": 13.18, "learning_rate": 7.600515995524604e-07, "loss": 0.0157, "step": 12985 }, { "epoch": 13.18, "learning_rate": 7.59213526593543e-07, "loss": 0.0206, "step": 12986 }, { "epoch": 13.18, "learning_rate": 7.58375897708975e-07, "loss": 0.0163, "step": 12987 }, { "epoch": 13.18, "learning_rate": 7.575387129390066e-07, "loss": 0.0147, "step": 12988 }, { "epoch": 13.18, "learning_rate": 7.567019723238722e-07, "loss": 0.0185, "step": 12989 }, { "epoch": 13.18, "learning_rate": 7.558656759037796e-07, "loss": 0.0134, "step": 12990 }, { "epoch": 13.18, "learning_rate": 7.55029823718918e-07, "loss": 0.0204, "step": 12991 }, { "epoch": 13.18, "learning_rate": 7.541944158094572e-07, "loss": 0.0195, "step": 12992 }, { "epoch": 13.18, "learning_rate": 7.533594522155419e-07, "loss": 0.0185, "step": 12993 }, { "epoch": 13.19, "learning_rate": 7.525249329772988e-07, "loss": 0.0195, "step": 12994 }, { "epoch": 13.19, "learning_rate": 7.516908581348281e-07, "loss": 0.0196, "step": 12995 }, { "epoch": 13.19, "learning_rate": 7.508572277282145e-07, "loss": 0.017, "step": 12996 }, { "epoch": 13.19, "learning_rate": 7.500240417975191e-07, "loss": 0.0178, "step": 12997 }, { "epoch": 13.19, "learning_rate": 7.4919130038278e-07, "loss": 0.0177, "step": 12998 }, { "epoch": 13.19, "learning_rate": 7.483590035240174e-07, "loss": 0.0166, "step": 12999 }, { "epoch": 13.19, "learning_rate": 7.475271512612236e-07, "loss": 0.016, "step": 13000 }, { "epoch": 13.19, "learning_rate": 7.466957436343792e-07, "loss": 0.0164, "step": 13001 }, { "epoch": 13.19, "learning_rate": 7.458647806834352e-07, "loss": 0.0206, "step": 13002 }, { "epoch": 13.19, "learning_rate": 7.450342624483254e-07, "loss": 0.016, "step": 13003 }, { "epoch": 13.2, "learning_rate": 7.442041889689622e-07, "loss": 0.0193, "step": 13004 }, { "epoch": 13.2, "learning_rate": 7.433745602852338e-07, "loss": 0.0173, "step": 13005 }, { "epoch": 13.2, "learning_rate": 7.425453764370072e-07, "loss": 0.0204, "step": 13006 }, { "epoch": 13.2, "learning_rate": 7.417166374641349e-07, "loss": 0.016, "step": 13007 }, { "epoch": 13.2, "learning_rate": 7.408883434064351e-07, "loss": 0.0175, "step": 13008 }, { "epoch": 13.2, "learning_rate": 7.400604943037204e-07, "loss": 0.0185, "step": 13009 }, { "epoch": 13.2, "learning_rate": 7.392330901957679e-07, "loss": 0.0163, "step": 13010 }, { "epoch": 13.2, "learning_rate": 7.384061311223423e-07, "loss": 0.0185, "step": 13011 }, { "epoch": 13.2, "learning_rate": 7.375796171231819e-07, "loss": 0.0166, "step": 13012 }, { "epoch": 13.2, "learning_rate": 7.367535482380061e-07, "loss": 0.0191, "step": 13013 }, { "epoch": 13.21, "learning_rate": 7.359279245065154e-07, "loss": 0.022, "step": 13014 }, { "epoch": 13.21, "learning_rate": 7.351027459683813e-07, "loss": 0.0159, "step": 13015 }, { "epoch": 13.21, "learning_rate": 7.34278012663261e-07, "loss": 0.0186, "step": 13016 }, { "epoch": 13.21, "learning_rate": 7.334537246307849e-07, "loss": 0.0181, "step": 13017 }, { "epoch": 13.21, "learning_rate": 7.326298819105704e-07, "loss": 0.0174, "step": 13018 }, { "epoch": 13.21, "learning_rate": 7.318064845422035e-07, "loss": 0.0158, "step": 13019 }, { "epoch": 13.21, "learning_rate": 7.309835325652548e-07, "loss": 0.0167, "step": 13020 }, { "epoch": 13.21, "learning_rate": 7.301610260192738e-07, "loss": 0.0189, "step": 13021 }, { "epoch": 13.21, "learning_rate": 7.293389649437832e-07, "loss": 0.0185, "step": 13022 }, { "epoch": 13.21, "learning_rate": 7.285173493782893e-07, "loss": 0.0197, "step": 13023 }, { "epoch": 13.22, "learning_rate": 7.276961793622772e-07, "loss": 0.0136, "step": 13024 }, { "epoch": 13.22, "learning_rate": 7.268754549352042e-07, "loss": 0.0169, "step": 13025 }, { "epoch": 13.22, "learning_rate": 7.260551761365187e-07, "loss": 0.02, "step": 13026 }, { "epoch": 13.22, "learning_rate": 7.252353430056336e-07, "loss": 0.0187, "step": 13027 }, { "epoch": 13.22, "learning_rate": 7.244159555819496e-07, "loss": 0.0197, "step": 13028 }, { "epoch": 13.22, "learning_rate": 7.23597013904842e-07, "loss": 0.0158, "step": 13029 }, { "epoch": 13.22, "learning_rate": 7.227785180136649e-07, "loss": 0.0181, "step": 13030 }, { "epoch": 13.22, "learning_rate": 7.219604679477554e-07, "loss": 0.0175, "step": 13031 }, { "epoch": 13.22, "learning_rate": 7.211428637464213e-07, "loss": 0.0157, "step": 13032 }, { "epoch": 13.22, "learning_rate": 7.203257054489554e-07, "loss": 0.018, "step": 13033 }, { "epoch": 13.23, "learning_rate": 7.195089930946264e-07, "loss": 0.0163, "step": 13034 }, { "epoch": 13.23, "learning_rate": 7.186927267226829e-07, "loss": 0.02, "step": 13035 }, { "epoch": 13.23, "learning_rate": 7.178769063723534e-07, "loss": 0.0182, "step": 13036 }, { "epoch": 13.23, "learning_rate": 7.170615320828378e-07, "loss": 0.0202, "step": 13037 }, { "epoch": 13.23, "learning_rate": 7.162466038933247e-07, "loss": 0.0155, "step": 13038 }, { "epoch": 13.23, "learning_rate": 7.154321218429716e-07, "loss": 0.0192, "step": 13039 }, { "epoch": 13.23, "learning_rate": 7.146180859709218e-07, "loss": 0.0221, "step": 13040 }, { "epoch": 13.23, "learning_rate": 7.13804496316296e-07, "loss": 0.0209, "step": 13041 }, { "epoch": 13.23, "learning_rate": 7.129913529181864e-07, "loss": 0.0187, "step": 13042 }, { "epoch": 13.23, "learning_rate": 7.121786558156762e-07, "loss": 0.0161, "step": 13043 }, { "epoch": 13.24, "learning_rate": 7.113664050478153e-07, "loss": 0.0173, "step": 13044 }, { "epoch": 13.24, "learning_rate": 7.105546006536412e-07, "loss": 0.0189, "step": 13045 }, { "epoch": 13.24, "learning_rate": 7.097432426721629e-07, "loss": 0.0192, "step": 13046 }, { "epoch": 13.24, "learning_rate": 7.089323311423701e-07, "loss": 0.0159, "step": 13047 }, { "epoch": 13.24, "learning_rate": 7.081218661032363e-07, "loss": 0.0164, "step": 13048 }, { "epoch": 13.24, "learning_rate": 7.073118475937046e-07, "loss": 0.0177, "step": 13049 }, { "epoch": 13.24, "learning_rate": 7.065022756527029e-07, "loss": 0.0203, "step": 13050 }, { "epoch": 13.24, "learning_rate": 7.056931503191356e-07, "loss": 0.0182, "step": 13051 }, { "epoch": 13.24, "learning_rate": 7.048844716318859e-07, "loss": 0.0162, "step": 13052 }, { "epoch": 13.25, "learning_rate": 7.040762396298184e-07, "loss": 0.0201, "step": 13053 }, { "epoch": 13.25, "learning_rate": 7.032684543517698e-07, "loss": 0.0186, "step": 13054 }, { "epoch": 13.25, "learning_rate": 7.024611158365613e-07, "loss": 0.0202, "step": 13055 }, { "epoch": 13.25, "learning_rate": 7.016542241229873e-07, "loss": 0.0178, "step": 13056 }, { "epoch": 13.25, "learning_rate": 7.008477792498258e-07, "loss": 0.0194, "step": 13057 }, { "epoch": 13.25, "learning_rate": 7.000417812558302e-07, "loss": 0.0169, "step": 13058 }, { "epoch": 13.25, "learning_rate": 6.99236230179735e-07, "loss": 0.0201, "step": 13059 }, { "epoch": 13.25, "learning_rate": 6.984311260602527e-07, "loss": 0.0184, "step": 13060 }, { "epoch": 13.25, "learning_rate": 6.976264689360701e-07, "loss": 0.0187, "step": 13061 }, { "epoch": 13.25, "learning_rate": 6.968222588458584e-07, "loss": 0.0176, "step": 13062 }, { "epoch": 13.26, "learning_rate": 6.960184958282612e-07, "loss": 0.0186, "step": 13063 }, { "epoch": 13.26, "learning_rate": 6.952151799219065e-07, "loss": 0.0165, "step": 13064 }, { "epoch": 13.26, "learning_rate": 6.944123111654011e-07, "loss": 0.0208, "step": 13065 }, { "epoch": 13.26, "learning_rate": 6.93609889597322e-07, "loss": 0.0157, "step": 13066 }, { "epoch": 13.26, "learning_rate": 6.928079152562328e-07, "loss": 0.0195, "step": 13067 }, { "epoch": 13.26, "learning_rate": 6.920063881806726e-07, "loss": 0.0159, "step": 13068 }, { "epoch": 13.26, "learning_rate": 6.912053084091608e-07, "loss": 0.0183, "step": 13069 }, { "epoch": 13.26, "learning_rate": 6.904046759801952e-07, "loss": 0.017, "step": 13070 }, { "epoch": 13.26, "learning_rate": 6.896044909322474e-07, "loss": 0.0164, "step": 13071 }, { "epoch": 13.26, "learning_rate": 6.888047533037745e-07, "loss": 0.0156, "step": 13072 }, { "epoch": 13.27, "learning_rate": 6.880054631332044e-07, "loss": 0.0181, "step": 13073 }, { "epoch": 13.27, "learning_rate": 6.872066204589512e-07, "loss": 0.0174, "step": 13074 }, { "epoch": 13.27, "learning_rate": 6.864082253194027e-07, "loss": 0.0191, "step": 13075 }, { "epoch": 13.27, "learning_rate": 6.856102777529261e-07, "loss": 0.0188, "step": 13076 }, { "epoch": 13.27, "learning_rate": 6.848127777978697e-07, "loss": 0.0153, "step": 13077 }, { "epoch": 13.27, "learning_rate": 6.840157254925551e-07, "loss": 0.0194, "step": 13078 }, { "epoch": 13.27, "learning_rate": 6.83219120875287e-07, "loss": 0.0173, "step": 13079 }, { "epoch": 13.27, "learning_rate": 6.824229639843471e-07, "loss": 0.016, "step": 13080 }, { "epoch": 13.27, "learning_rate": 6.81627254857995e-07, "loss": 0.0172, "step": 13081 }, { "epoch": 13.27, "learning_rate": 6.808319935344699e-07, "loss": 0.0194, "step": 13082 }, { "epoch": 13.28, "learning_rate": 6.800371800519856e-07, "loss": 0.0199, "step": 13083 }, { "epoch": 13.28, "learning_rate": 6.792428144487429e-07, "loss": 0.0171, "step": 13084 }, { "epoch": 13.28, "learning_rate": 6.784488967629122e-07, "loss": 0.0208, "step": 13085 }, { "epoch": 13.28, "learning_rate": 6.776554270326452e-07, "loss": 0.0171, "step": 13086 }, { "epoch": 13.28, "learning_rate": 6.76862405296077e-07, "loss": 0.0167, "step": 13087 }, { "epoch": 13.28, "learning_rate": 6.760698315913128e-07, "loss": 0.0168, "step": 13088 }, { "epoch": 13.28, "learning_rate": 6.752777059564431e-07, "loss": 0.019, "step": 13089 }, { "epoch": 13.28, "learning_rate": 6.744860284295318e-07, "loss": 0.0166, "step": 13090 }, { "epoch": 13.28, "learning_rate": 6.736947990486253e-07, "loss": 0.0186, "step": 13091 }, { "epoch": 13.28, "learning_rate": 6.729040178517454e-07, "loss": 0.0173, "step": 13092 }, { "epoch": 13.29, "learning_rate": 6.721136848768961e-07, "loss": 0.0154, "step": 13093 }, { "epoch": 13.29, "learning_rate": 6.71323800162057e-07, "loss": 0.0152, "step": 13094 }, { "epoch": 13.29, "learning_rate": 6.705343637451855e-07, "loss": 0.0173, "step": 13095 }, { "epoch": 13.29, "learning_rate": 6.697453756642192e-07, "loss": 0.0198, "step": 13096 }, { "epoch": 13.29, "learning_rate": 6.689568359570742e-07, "loss": 0.0185, "step": 13097 }, { "epoch": 13.29, "learning_rate": 6.681687446616436e-07, "loss": 0.0197, "step": 13098 }, { "epoch": 13.29, "learning_rate": 6.673811018158005e-07, "loss": 0.0201, "step": 13099 }, { "epoch": 13.29, "learning_rate": 6.665939074573935e-07, "loss": 0.017, "step": 13100 }, { "epoch": 13.29, "learning_rate": 6.658071616242567e-07, "loss": 0.0162, "step": 13101 }, { "epoch": 13.29, "learning_rate": 6.650208643541934e-07, "loss": 0.0239, "step": 13102 }, { "epoch": 13.3, "learning_rate": 6.642350156849919e-07, "loss": 0.0194, "step": 13103 }, { "epoch": 13.3, "learning_rate": 6.634496156544179e-07, "loss": 0.0186, "step": 13104 }, { "epoch": 13.3, "learning_rate": 6.626646643002122e-07, "loss": 0.0203, "step": 13105 }, { "epoch": 13.3, "learning_rate": 6.618801616600989e-07, "loss": 0.0206, "step": 13106 }, { "epoch": 13.3, "learning_rate": 6.610961077717737e-07, "loss": 0.0164, "step": 13107 }, { "epoch": 13.3, "learning_rate": 6.603125026729174e-07, "loss": 0.0204, "step": 13108 }, { "epoch": 13.3, "learning_rate": 6.595293464011887e-07, "loss": 0.0178, "step": 13109 }, { "epoch": 13.3, "learning_rate": 6.587466389942209e-07, "loss": 0.0169, "step": 13110 }, { "epoch": 13.3, "learning_rate": 6.579643804896297e-07, "loss": 0.0159, "step": 13111 }, { "epoch": 13.3, "learning_rate": 6.571825709250035e-07, "loss": 0.0184, "step": 13112 }, { "epoch": 13.31, "learning_rate": 6.564012103379158e-07, "loss": 0.0177, "step": 13113 }, { "epoch": 13.31, "learning_rate": 6.556202987659166e-07, "loss": 0.0168, "step": 13114 }, { "epoch": 13.31, "learning_rate": 6.548398362465291e-07, "loss": 0.017, "step": 13115 }, { "epoch": 13.31, "learning_rate": 6.540598228172634e-07, "loss": 0.0146, "step": 13116 }, { "epoch": 13.31, "learning_rate": 6.532802585155995e-07, "loss": 0.0184, "step": 13117 }, { "epoch": 13.31, "learning_rate": 6.525011433790041e-07, "loss": 0.0201, "step": 13118 }, { "epoch": 13.31, "learning_rate": 6.517224774449171e-07, "loss": 0.0182, "step": 13119 }, { "epoch": 13.31, "learning_rate": 6.509442607507566e-07, "loss": 0.0177, "step": 13120 }, { "epoch": 13.31, "learning_rate": 6.501664933339225e-07, "loss": 0.0193, "step": 13121 }, { "epoch": 13.32, "learning_rate": 6.493891752317894e-07, "loss": 0.0172, "step": 13122 }, { "epoch": 13.32, "learning_rate": 6.486123064817119e-07, "loss": 0.0162, "step": 13123 }, { "epoch": 13.32, "learning_rate": 6.478358871210266e-07, "loss": 0.0174, "step": 13124 }, { "epoch": 13.32, "learning_rate": 6.470599171870384e-07, "loss": 0.0179, "step": 13125 }, { "epoch": 13.32, "learning_rate": 6.46284396717044e-07, "loss": 0.0187, "step": 13126 }, { "epoch": 13.32, "learning_rate": 6.45509325748308e-07, "loss": 0.0156, "step": 13127 }, { "epoch": 13.32, "learning_rate": 6.447347043180797e-07, "loss": 0.0177, "step": 13128 }, { "epoch": 13.32, "learning_rate": 6.439605324635811e-07, "loss": 0.0151, "step": 13129 }, { "epoch": 13.32, "learning_rate": 6.431868102220173e-07, "loss": 0.0189, "step": 13130 }, { "epoch": 13.32, "learning_rate": 6.424135376305718e-07, "loss": 0.0155, "step": 13131 }, { "epoch": 13.33, "learning_rate": 6.416407147264014e-07, "loss": 0.0156, "step": 13132 }, { "epoch": 13.33, "learning_rate": 6.408683415466477e-07, "loss": 0.0163, "step": 13133 }, { "epoch": 13.33, "learning_rate": 6.400964181284253e-07, "loss": 0.018, "step": 13134 }, { "epoch": 13.33, "learning_rate": 6.393249445088335e-07, "loss": 0.0178, "step": 13135 }, { "epoch": 13.33, "learning_rate": 6.385539207249425e-07, "loss": 0.02, "step": 13136 }, { "epoch": 13.33, "learning_rate": 6.37783346813805e-07, "loss": 0.0158, "step": 13137 }, { "epoch": 13.33, "learning_rate": 6.370132228124548e-07, "loss": 0.0162, "step": 13138 }, { "epoch": 13.33, "learning_rate": 6.362435487578977e-07, "loss": 0.0194, "step": 13139 }, { "epoch": 13.33, "learning_rate": 6.354743246871209e-07, "loss": 0.0176, "step": 13140 }, { "epoch": 13.33, "learning_rate": 6.347055506370925e-07, "loss": 0.0174, "step": 13141 }, { "epoch": 13.34, "learning_rate": 6.33937226644753e-07, "loss": 0.0179, "step": 13142 }, { "epoch": 13.34, "learning_rate": 6.331693527470306e-07, "loss": 0.0186, "step": 13143 }, { "epoch": 13.34, "learning_rate": 6.324019289808215e-07, "loss": 0.0188, "step": 13144 }, { "epoch": 13.34, "learning_rate": 6.31634955383007e-07, "loss": 0.0204, "step": 13145 }, { "epoch": 13.34, "learning_rate": 6.308684319904423e-07, "loss": 0.0186, "step": 13146 }, { "epoch": 13.34, "learning_rate": 6.301023588399657e-07, "loss": 0.0181, "step": 13147 }, { "epoch": 13.34, "learning_rate": 6.293367359683921e-07, "loss": 0.0181, "step": 13148 }, { "epoch": 13.34, "learning_rate": 6.285715634125122e-07, "loss": 0.019, "step": 13149 }, { "epoch": 13.34, "learning_rate": 6.278068412090976e-07, "loss": 0.0188, "step": 13150 }, { "epoch": 13.34, "learning_rate": 6.270425693948978e-07, "loss": 0.0172, "step": 13151 }, { "epoch": 13.35, "learning_rate": 6.262787480066412e-07, "loss": 0.0163, "step": 13152 }, { "epoch": 13.35, "learning_rate": 6.25515377081034e-07, "loss": 0.0175, "step": 13153 }, { "epoch": 13.35, "learning_rate": 6.247524566547592e-07, "loss": 0.0194, "step": 13154 }, { "epoch": 13.35, "learning_rate": 6.239899867644828e-07, "loss": 0.0199, "step": 13155 }, { "epoch": 13.35, "learning_rate": 6.232279674468412e-07, "loss": 0.0209, "step": 13156 }, { "epoch": 13.35, "learning_rate": 6.224663987384571e-07, "loss": 0.0188, "step": 13157 }, { "epoch": 13.35, "learning_rate": 6.217052806759294e-07, "loss": 0.0175, "step": 13158 }, { "epoch": 13.35, "learning_rate": 6.209446132958296e-07, "loss": 0.0202, "step": 13159 }, { "epoch": 13.35, "learning_rate": 6.201843966347176e-07, "loss": 0.0175, "step": 13160 }, { "epoch": 13.35, "learning_rate": 6.194246307291241e-07, "loss": 0.0183, "step": 13161 }, { "epoch": 13.36, "learning_rate": 6.186653156155608e-07, "loss": 0.0186, "step": 13162 }, { "epoch": 13.36, "learning_rate": 6.179064513305155e-07, "loss": 0.0188, "step": 13163 }, { "epoch": 13.36, "learning_rate": 6.171480379104566e-07, "loss": 0.019, "step": 13164 }, { "epoch": 13.36, "learning_rate": 6.16390075391834e-07, "loss": 0.0186, "step": 13165 }, { "epoch": 13.36, "learning_rate": 6.156325638110671e-07, "loss": 0.0166, "step": 13166 }, { "epoch": 13.36, "learning_rate": 6.148755032045606e-07, "loss": 0.0182, "step": 13167 }, { "epoch": 13.36, "learning_rate": 6.141188936086973e-07, "loss": 0.0171, "step": 13168 }, { "epoch": 13.36, "learning_rate": 6.133627350598348e-07, "loss": 0.0197, "step": 13169 }, { "epoch": 13.36, "learning_rate": 6.126070275943141e-07, "loss": 0.0191, "step": 13170 }, { "epoch": 13.36, "learning_rate": 6.118517712484473e-07, "loss": 0.0195, "step": 13171 }, { "epoch": 13.37, "learning_rate": 6.110969660585331e-07, "loss": 0.019, "step": 13172 }, { "epoch": 13.37, "learning_rate": 6.103426120608391e-07, "loss": 0.0165, "step": 13173 }, { "epoch": 13.37, "learning_rate": 6.095887092916208e-07, "loss": 0.0169, "step": 13174 }, { "epoch": 13.37, "learning_rate": 6.08835257787106e-07, "loss": 0.0169, "step": 13175 }, { "epoch": 13.37, "learning_rate": 6.080822575835021e-07, "loss": 0.017, "step": 13176 }, { "epoch": 13.37, "learning_rate": 6.073297087169982e-07, "loss": 0.0216, "step": 13177 }, { "epoch": 13.37, "learning_rate": 6.065776112237543e-07, "loss": 0.02, "step": 13178 }, { "epoch": 13.37, "learning_rate": 6.058259651399157e-07, "loss": 0.0192, "step": 13179 }, { "epoch": 13.37, "learning_rate": 6.050747705016036e-07, "loss": 0.0175, "step": 13180 }, { "epoch": 13.37, "learning_rate": 6.043240273449158e-07, "loss": 0.0157, "step": 13181 }, { "epoch": 13.38, "learning_rate": 6.035737357059323e-07, "loss": 0.0182, "step": 13182 }, { "epoch": 13.38, "learning_rate": 6.028238956207066e-07, "loss": 0.0185, "step": 13183 }, { "epoch": 13.38, "learning_rate": 6.020745071252732e-07, "loss": 0.0173, "step": 13184 }, { "epoch": 13.38, "learning_rate": 6.013255702556464e-07, "loss": 0.0167, "step": 13185 }, { "epoch": 13.38, "learning_rate": 6.005770850478143e-07, "loss": 0.0178, "step": 13186 }, { "epoch": 13.38, "learning_rate": 5.998290515377503e-07, "loss": 0.0166, "step": 13187 }, { "epoch": 13.38, "learning_rate": 5.990814697613978e-07, "loss": 0.0182, "step": 13188 }, { "epoch": 13.38, "learning_rate": 5.983343397546859e-07, "loss": 0.0174, "step": 13189 }, { "epoch": 13.38, "learning_rate": 5.975876615535148e-07, "loss": 0.017, "step": 13190 }, { "epoch": 13.39, "learning_rate": 5.968414351937691e-07, "loss": 0.0167, "step": 13191 }, { "epoch": 13.39, "learning_rate": 5.960956607113078e-07, "loss": 0.019, "step": 13192 }, { "epoch": 13.39, "learning_rate": 5.953503381419723e-07, "loss": 0.0186, "step": 13193 }, { "epoch": 13.39, "learning_rate": 5.946054675215785e-07, "loss": 0.0193, "step": 13194 }, { "epoch": 13.39, "learning_rate": 5.938610488859209e-07, "loss": 0.0177, "step": 13195 }, { "epoch": 13.39, "learning_rate": 5.931170822707743e-07, "loss": 0.0179, "step": 13196 }, { "epoch": 13.39, "learning_rate": 5.923735677118925e-07, "loss": 0.0202, "step": 13197 }, { "epoch": 13.39, "learning_rate": 5.916305052450011e-07, "loss": 0.0184, "step": 13198 }, { "epoch": 13.39, "learning_rate": 5.908878949058127e-07, "loss": 0.0161, "step": 13199 }, { "epoch": 13.39, "learning_rate": 5.901457367300101e-07, "loss": 0.0194, "step": 13200 }, { "epoch": 13.4, "learning_rate": 5.894040307532634e-07, "loss": 0.0201, "step": 13201 }, { "epoch": 13.4, "learning_rate": 5.88662777011213e-07, "loss": 0.0193, "step": 13202 }, { "epoch": 13.4, "learning_rate": 5.879219755394805e-07, "loss": 0.0186, "step": 13203 }, { "epoch": 13.4, "learning_rate": 5.871816263736673e-07, "loss": 0.0206, "step": 13204 }, { "epoch": 13.4, "learning_rate": 5.864417295493497e-07, "loss": 0.0201, "step": 13205 }, { "epoch": 13.4, "learning_rate": 5.857022851020855e-07, "loss": 0.0175, "step": 13206 }, { "epoch": 13.4, "learning_rate": 5.849632930674087e-07, "loss": 0.0209, "step": 13207 }, { "epoch": 13.4, "learning_rate": 5.842247534808309e-07, "loss": 0.0171, "step": 13208 }, { "epoch": 13.4, "learning_rate": 5.83486666377846e-07, "loss": 0.0185, "step": 13209 }, { "epoch": 13.4, "learning_rate": 5.82749031793921e-07, "loss": 0.0179, "step": 13210 }, { "epoch": 13.41, "learning_rate": 5.820118497645078e-07, "loss": 0.0186, "step": 13211 }, { "epoch": 13.41, "learning_rate": 5.812751203250267e-07, "loss": 0.0144, "step": 13212 }, { "epoch": 13.41, "learning_rate": 5.805388435108839e-07, "loss": 0.0189, "step": 13213 }, { "epoch": 13.41, "learning_rate": 5.798030193574656e-07, "loss": 0.0161, "step": 13214 }, { "epoch": 13.41, "learning_rate": 5.790676479001267e-07, "loss": 0.0218, "step": 13215 }, { "epoch": 13.41, "learning_rate": 5.783327291742103e-07, "loss": 0.019, "step": 13216 }, { "epoch": 13.41, "learning_rate": 5.77598263215029e-07, "loss": 0.0187, "step": 13217 }, { "epoch": 13.41, "learning_rate": 5.768642500578847e-07, "loss": 0.0182, "step": 13218 }, { "epoch": 13.41, "learning_rate": 5.761306897380469e-07, "loss": 0.0174, "step": 13219 }, { "epoch": 13.41, "learning_rate": 5.753975822907676e-07, "loss": 0.0187, "step": 13220 }, { "epoch": 13.42, "learning_rate": 5.746649277512783e-07, "loss": 0.0187, "step": 13221 }, { "epoch": 13.42, "learning_rate": 5.739327261547866e-07, "loss": 0.0175, "step": 13222 }, { "epoch": 13.42, "learning_rate": 5.732009775364778e-07, "loss": 0.02, "step": 13223 }, { "epoch": 13.42, "learning_rate": 5.724696819315211e-07, "loss": 0.0161, "step": 13224 }, { "epoch": 13.42, "learning_rate": 5.717388393750522e-07, "loss": 0.0146, "step": 13225 }, { "epoch": 13.42, "learning_rate": 5.710084499022006e-07, "loss": 0.0182, "step": 13226 }, { "epoch": 13.42, "learning_rate": 5.702785135480604e-07, "loss": 0.0192, "step": 13227 }, { "epoch": 13.42, "learning_rate": 5.695490303477125e-07, "loss": 0.0176, "step": 13228 }, { "epoch": 13.42, "learning_rate": 5.688200003362099e-07, "loss": 0.0189, "step": 13229 }, { "epoch": 13.42, "learning_rate": 5.680914235485891e-07, "loss": 0.02, "step": 13230 }, { "epoch": 13.43, "learning_rate": 5.67363300019862e-07, "loss": 0.0181, "step": 13231 }, { "epoch": 13.43, "learning_rate": 5.666356297850173e-07, "loss": 0.0209, "step": 13232 }, { "epoch": 13.43, "learning_rate": 5.659084128790282e-07, "loss": 0.0117, "step": 13233 }, { "epoch": 13.43, "learning_rate": 5.651816493368345e-07, "loss": 0.0199, "step": 13234 }, { "epoch": 13.43, "learning_rate": 5.644553391933705e-07, "loss": 0.0184, "step": 13235 }, { "epoch": 13.43, "learning_rate": 5.637294824835327e-07, "loss": 0.0176, "step": 13236 }, { "epoch": 13.43, "learning_rate": 5.630040792422054e-07, "loss": 0.0165, "step": 13237 }, { "epoch": 13.43, "learning_rate": 5.622791295042507e-07, "loss": 0.018, "step": 13238 }, { "epoch": 13.43, "learning_rate": 5.615546333045019e-07, "loss": 0.0176, "step": 13239 }, { "epoch": 13.43, "learning_rate": 5.608305906777778e-07, "loss": 0.0189, "step": 13240 }, { "epoch": 13.44, "learning_rate": 5.60107001658875e-07, "loss": 0.0183, "step": 13241 }, { "epoch": 13.44, "learning_rate": 5.593838662825612e-07, "loss": 0.019, "step": 13242 }, { "epoch": 13.44, "learning_rate": 5.586611845835932e-07, "loss": 0.0144, "step": 13243 }, { "epoch": 13.44, "learning_rate": 5.579389565966964e-07, "loss": 0.0143, "step": 13244 }, { "epoch": 13.44, "learning_rate": 5.572171823565797e-07, "loss": 0.0157, "step": 13245 }, { "epoch": 13.44, "learning_rate": 5.564958618979266e-07, "loss": 0.0196, "step": 13246 }, { "epoch": 13.44, "learning_rate": 5.557749952554015e-07, "loss": 0.0172, "step": 13247 }, { "epoch": 13.44, "learning_rate": 5.550545824636489e-07, "loss": 0.0172, "step": 13248 }, { "epoch": 13.44, "learning_rate": 5.543346235572855e-07, "loss": 0.0212, "step": 13249 }, { "epoch": 13.44, "learning_rate": 5.536151185709105e-07, "loss": 0.0179, "step": 13250 }, { "epoch": 13.45, "learning_rate": 5.528960675391015e-07, "loss": 0.0189, "step": 13251 }, { "epoch": 13.45, "learning_rate": 5.521774704964122e-07, "loss": 0.0174, "step": 13252 }, { "epoch": 13.45, "learning_rate": 5.514593274773761e-07, "loss": 0.0153, "step": 13253 }, { "epoch": 13.45, "learning_rate": 5.507416385165031e-07, "loss": 0.0152, "step": 13254 }, { "epoch": 13.45, "learning_rate": 5.50024403648285e-07, "loss": 0.0188, "step": 13255 }, { "epoch": 13.45, "learning_rate": 5.493076229071858e-07, "loss": 0.0193, "step": 13256 }, { "epoch": 13.45, "learning_rate": 5.485912963276518e-07, "loss": 0.0164, "step": 13257 }, { "epoch": 13.45, "learning_rate": 5.478754239441098e-07, "loss": 0.0177, "step": 13258 }, { "epoch": 13.45, "learning_rate": 5.471600057909555e-07, "loss": 0.0221, "step": 13259 }, { "epoch": 13.46, "learning_rate": 5.464450419025757e-07, "loss": 0.0168, "step": 13260 }, { "epoch": 13.46, "learning_rate": 5.457305323133244e-07, "loss": 0.0196, "step": 13261 }, { "epoch": 13.46, "learning_rate": 5.450164770575417e-07, "loss": 0.018, "step": 13262 }, { "epoch": 13.46, "learning_rate": 5.44302876169538e-07, "loss": 0.017, "step": 13263 }, { "epoch": 13.46, "learning_rate": 5.435897296836068e-07, "loss": 0.0151, "step": 13264 }, { "epoch": 13.46, "learning_rate": 5.42877037634022e-07, "loss": 0.0167, "step": 13265 }, { "epoch": 13.46, "learning_rate": 5.421648000550295e-07, "loss": 0.0199, "step": 13266 }, { "epoch": 13.46, "learning_rate": 5.414530169808574e-07, "loss": 0.0193, "step": 13267 }, { "epoch": 13.46, "learning_rate": 5.407416884457117e-07, "loss": 0.0208, "step": 13268 }, { "epoch": 13.46, "learning_rate": 5.400308144837751e-07, "loss": 0.0166, "step": 13269 }, { "epoch": 13.47, "learning_rate": 5.393203951292125e-07, "loss": 0.0153, "step": 13270 }, { "epoch": 13.47, "learning_rate": 5.386104304161588e-07, "loss": 0.0189, "step": 13271 }, { "epoch": 13.47, "learning_rate": 5.379009203787356e-07, "loss": 0.0206, "step": 13272 }, { "epoch": 13.47, "learning_rate": 5.371918650510355e-07, "loss": 0.0203, "step": 13273 }, { "epoch": 13.47, "learning_rate": 5.364832644671358e-07, "loss": 0.0193, "step": 13274 }, { "epoch": 13.47, "learning_rate": 5.357751186610904e-07, "loss": 0.0201, "step": 13275 }, { "epoch": 13.47, "learning_rate": 5.350674276669232e-07, "loss": 0.0211, "step": 13276 }, { "epoch": 13.47, "learning_rate": 5.343601915186513e-07, "loss": 0.0198, "step": 13277 }, { "epoch": 13.47, "learning_rate": 5.336534102502555e-07, "loss": 0.0191, "step": 13278 }, { "epoch": 13.47, "learning_rate": 5.329470838957052e-07, "loss": 0.0176, "step": 13279 }, { "epoch": 13.48, "learning_rate": 5.322412124889398e-07, "loss": 0.0205, "step": 13280 }, { "epoch": 13.48, "learning_rate": 5.315357960638812e-07, "loss": 0.0177, "step": 13281 }, { "epoch": 13.48, "learning_rate": 5.308308346544322e-07, "loss": 0.0204, "step": 13282 }, { "epoch": 13.48, "learning_rate": 5.301263282944657e-07, "loss": 0.0183, "step": 13283 }, { "epoch": 13.48, "learning_rate": 5.294222770178392e-07, "loss": 0.0161, "step": 13284 }, { "epoch": 13.48, "learning_rate": 5.287186808583878e-07, "loss": 0.0176, "step": 13285 }, { "epoch": 13.48, "learning_rate": 5.280155398499221e-07, "loss": 0.0191, "step": 13286 }, { "epoch": 13.48, "learning_rate": 5.27312854026234e-07, "loss": 0.0173, "step": 13287 }, { "epoch": 13.48, "learning_rate": 5.266106234210888e-07, "loss": 0.0206, "step": 13288 }, { "epoch": 13.48, "learning_rate": 5.259088480682362e-07, "loss": 0.0198, "step": 13289 }, { "epoch": 13.49, "learning_rate": 5.252075280013968e-07, "loss": 0.0168, "step": 13290 }, { "epoch": 13.49, "learning_rate": 5.24506663254275e-07, "loss": 0.0206, "step": 13291 }, { "epoch": 13.49, "learning_rate": 5.238062538605526e-07, "loss": 0.0173, "step": 13292 }, { "epoch": 13.49, "learning_rate": 5.231062998538871e-07, "loss": 0.0183, "step": 13293 }, { "epoch": 13.49, "learning_rate": 5.224068012679162e-07, "loss": 0.0174, "step": 13294 }, { "epoch": 13.49, "learning_rate": 5.217077581362539e-07, "loss": 0.0165, "step": 13295 }, { "epoch": 13.49, "learning_rate": 5.210091704924947e-07, "loss": 0.0186, "step": 13296 }, { "epoch": 13.49, "learning_rate": 5.203110383702092e-07, "loss": 0.016, "step": 13297 }, { "epoch": 13.49, "learning_rate": 5.196133618029464e-07, "loss": 0.0203, "step": 13298 }, { "epoch": 13.49, "learning_rate": 5.189161408242338e-07, "loss": 0.0151, "step": 13299 }, { "epoch": 13.5, "learning_rate": 5.182193754675779e-07, "loss": 0.0181, "step": 13300 }, { "epoch": 13.5, "learning_rate": 5.175230657664598e-07, "loss": 0.0176, "step": 13301 }, { "epoch": 13.5, "learning_rate": 5.168272117543449e-07, "loss": 0.0173, "step": 13302 }, { "epoch": 13.5, "learning_rate": 5.161318134646709e-07, "loss": 0.0218, "step": 13303 }, { "epoch": 13.5, "learning_rate": 5.154368709308566e-07, "loss": 0.022, "step": 13304 }, { "epoch": 13.5, "learning_rate": 5.147423841862975e-07, "loss": 0.0199, "step": 13305 }, { "epoch": 13.5, "learning_rate": 5.140483532643692e-07, "loss": 0.0199, "step": 13306 }, { "epoch": 13.5, "learning_rate": 5.133547781984216e-07, "loss": 0.0188, "step": 13307 }, { "epoch": 13.5, "learning_rate": 5.12661659021787e-07, "loss": 0.0163, "step": 13308 }, { "epoch": 13.5, "learning_rate": 5.119689957677721e-07, "loss": 0.0192, "step": 13309 }, { "epoch": 13.51, "learning_rate": 5.112767884696656e-07, "loss": 0.0186, "step": 13310 }, { "epoch": 13.51, "learning_rate": 5.105850371607324e-07, "loss": 0.0213, "step": 13311 }, { "epoch": 13.51, "learning_rate": 5.098937418742111e-07, "loss": 0.0199, "step": 13312 }, { "epoch": 13.51, "learning_rate": 5.092029026433265e-07, "loss": 0.0176, "step": 13313 }, { "epoch": 13.51, "learning_rate": 5.085125195012785e-07, "loss": 0.0187, "step": 13314 }, { "epoch": 13.51, "learning_rate": 5.078225924812397e-07, "loss": 0.0193, "step": 13315 }, { "epoch": 13.51, "learning_rate": 5.071331216163689e-07, "loss": 0.0181, "step": 13316 }, { "epoch": 13.51, "learning_rate": 5.064441069397952e-07, "loss": 0.0173, "step": 13317 }, { "epoch": 13.51, "learning_rate": 5.057555484846343e-07, "loss": 0.0168, "step": 13318 }, { "epoch": 13.51, "learning_rate": 5.050674462839733e-07, "loss": 0.0197, "step": 13319 }, { "epoch": 13.52, "learning_rate": 5.043798003708799e-07, "loss": 0.0169, "step": 13320 }, { "epoch": 13.52, "learning_rate": 5.036926107784001e-07, "loss": 0.014, "step": 13321 }, { "epoch": 13.52, "learning_rate": 5.030058775395563e-07, "loss": 0.0194, "step": 13322 }, { "epoch": 13.52, "learning_rate": 5.023196006873499e-07, "loss": 0.0192, "step": 13323 }, { "epoch": 13.52, "learning_rate": 5.016337802547633e-07, "loss": 0.0171, "step": 13324 }, { "epoch": 13.52, "learning_rate": 5.009484162747502e-07, "loss": 0.0173, "step": 13325 }, { "epoch": 13.52, "learning_rate": 5.002635087802487e-07, "loss": 0.0192, "step": 13326 }, { "epoch": 13.52, "learning_rate": 4.995790578041725e-07, "loss": 0.0188, "step": 13327 }, { "epoch": 13.52, "learning_rate": 4.988950633794154e-07, "loss": 0.02, "step": 13328 }, { "epoch": 13.53, "learning_rate": 4.982115255388431e-07, "loss": 0.0183, "step": 13329 }, { "epoch": 13.53, "learning_rate": 4.975284443153062e-07, "loss": 0.0165, "step": 13330 }, { "epoch": 13.53, "learning_rate": 4.968458197416315e-07, "loss": 0.0162, "step": 13331 }, { "epoch": 13.53, "learning_rate": 4.961636518506207e-07, "loss": 0.0172, "step": 13332 }, { "epoch": 13.53, "learning_rate": 4.954819406750589e-07, "loss": 0.0168, "step": 13333 }, { "epoch": 13.53, "learning_rate": 4.948006862477017e-07, "loss": 0.0161, "step": 13334 }, { "epoch": 13.53, "learning_rate": 4.941198886012932e-07, "loss": 0.0202, "step": 13335 }, { "epoch": 13.53, "learning_rate": 4.93439547768546e-07, "loss": 0.0181, "step": 13336 }, { "epoch": 13.53, "learning_rate": 4.927596637821553e-07, "loss": 0.0174, "step": 13337 }, { "epoch": 13.53, "learning_rate": 4.920802366747957e-07, "loss": 0.0193, "step": 13338 }, { "epoch": 13.54, "learning_rate": 4.914012664791146e-07, "loss": 0.0185, "step": 13339 }, { "epoch": 13.54, "learning_rate": 4.907227532277414e-07, "loss": 0.0169, "step": 13340 }, { "epoch": 13.54, "learning_rate": 4.900446969532835e-07, "loss": 0.0221, "step": 13341 }, { "epoch": 13.54, "learning_rate": 4.893670976883235e-07, "loss": 0.0154, "step": 13342 }, { "epoch": 13.54, "learning_rate": 4.886899554654278e-07, "loss": 0.0183, "step": 13343 }, { "epoch": 13.54, "learning_rate": 4.880132703171337e-07, "loss": 0.0146, "step": 13344 }, { "epoch": 13.54, "learning_rate": 4.873370422759615e-07, "loss": 0.0181, "step": 13345 }, { "epoch": 13.54, "learning_rate": 4.866612713744068e-07, "loss": 0.0184, "step": 13346 }, { "epoch": 13.54, "learning_rate": 4.859859576449444e-07, "loss": 0.0197, "step": 13347 }, { "epoch": 13.54, "learning_rate": 4.853111011200307e-07, "loss": 0.022, "step": 13348 }, { "epoch": 13.55, "learning_rate": 4.846367018320908e-07, "loss": 0.0216, "step": 13349 }, { "epoch": 13.55, "learning_rate": 4.839627598135377e-07, "loss": 0.0193, "step": 13350 }, { "epoch": 13.55, "learning_rate": 4.832892750967555e-07, "loss": 0.0165, "step": 13351 }, { "epoch": 13.55, "learning_rate": 4.826162477141117e-07, "loss": 0.0199, "step": 13352 }, { "epoch": 13.55, "learning_rate": 4.81943677697948e-07, "loss": 0.0175, "step": 13353 }, { "epoch": 13.55, "learning_rate": 4.812715650805855e-07, "loss": 0.0205, "step": 13354 }, { "epoch": 13.55, "learning_rate": 4.805999098943237e-07, "loss": 0.0167, "step": 13355 }, { "epoch": 13.55, "learning_rate": 4.79928712171438e-07, "loss": 0.0235, "step": 13356 }, { "epoch": 13.55, "learning_rate": 4.792579719441848e-07, "loss": 0.0184, "step": 13357 }, { "epoch": 13.55, "learning_rate": 4.785876892447982e-07, "loss": 0.0199, "step": 13358 }, { "epoch": 13.56, "learning_rate": 4.779178641054849e-07, "loss": 0.0194, "step": 13359 }, { "epoch": 13.56, "learning_rate": 4.77248496558439e-07, "loss": 0.0159, "step": 13360 }, { "epoch": 13.56, "learning_rate": 4.7657958663582473e-07, "loss": 0.0181, "step": 13361 }, { "epoch": 13.56, "learning_rate": 4.7591113436978977e-07, "loss": 0.0179, "step": 13362 }, { "epoch": 13.56, "learning_rate": 4.7524313979245286e-07, "loss": 0.0201, "step": 13363 }, { "epoch": 13.56, "learning_rate": 4.745756029359183e-07, "loss": 0.016, "step": 13364 }, { "epoch": 13.56, "learning_rate": 4.7390852383226607e-07, "loss": 0.0176, "step": 13365 }, { "epoch": 13.56, "learning_rate": 4.732419025135493e-07, "loss": 0.0161, "step": 13366 }, { "epoch": 13.56, "learning_rate": 4.7257573901180575e-07, "loss": 0.0194, "step": 13367 }, { "epoch": 13.56, "learning_rate": 4.719100333590487e-07, "loss": 0.0171, "step": 13368 }, { "epoch": 13.57, "learning_rate": 4.7124478558726814e-07, "loss": 0.0159, "step": 13369 }, { "epoch": 13.57, "learning_rate": 4.7057999572843516e-07, "loss": 0.0183, "step": 13370 }, { "epoch": 13.57, "learning_rate": 4.699156638144942e-07, "loss": 0.0213, "step": 13371 }, { "epoch": 13.57, "learning_rate": 4.6925178987737184e-07, "loss": 0.0189, "step": 13372 }, { "epoch": 13.57, "learning_rate": 4.685883739489705e-07, "loss": 0.0193, "step": 13373 }, { "epoch": 13.57, "learning_rate": 4.6792541606117127e-07, "loss": 0.0176, "step": 13374 }, { "epoch": 13.57, "learning_rate": 4.6726291624583533e-07, "loss": 0.0182, "step": 13375 }, { "epoch": 13.57, "learning_rate": 4.6660087453479495e-07, "loss": 0.0196, "step": 13376 }, { "epoch": 13.57, "learning_rate": 4.6593929095987036e-07, "loss": 0.0155, "step": 13377 }, { "epoch": 13.57, "learning_rate": 4.652781655528504e-07, "loss": 0.0214, "step": 13378 }, { "epoch": 13.58, "learning_rate": 4.6461749834550984e-07, "loss": 0.0155, "step": 13379 }, { "epoch": 13.58, "learning_rate": 4.639572893695943e-07, "loss": 0.0174, "step": 13380 }, { "epoch": 13.58, "learning_rate": 4.632975386568328e-07, "loss": 0.02, "step": 13381 }, { "epoch": 13.58, "learning_rate": 4.6263824623893006e-07, "loss": 0.0185, "step": 13382 }, { "epoch": 13.58, "learning_rate": 4.619794121475674e-07, "loss": 0.0225, "step": 13383 }, { "epoch": 13.58, "learning_rate": 4.6132103641440606e-07, "loss": 0.0206, "step": 13384 }, { "epoch": 13.58, "learning_rate": 4.6066311907108643e-07, "loss": 0.0162, "step": 13385 }, { "epoch": 13.58, "learning_rate": 4.600056601492242e-07, "loss": 0.0181, "step": 13386 }, { "epoch": 13.58, "learning_rate": 4.593486596804153e-07, "loss": 0.0175, "step": 13387 }, { "epoch": 13.58, "learning_rate": 4.586921176962311e-07, "loss": 0.0192, "step": 13388 }, { "epoch": 13.59, "learning_rate": 4.580360342282242e-07, "loss": 0.0187, "step": 13389 }, { "epoch": 13.59, "learning_rate": 4.573804093079193e-07, "loss": 0.0213, "step": 13390 }, { "epoch": 13.59, "learning_rate": 4.5672524296682674e-07, "loss": 0.0192, "step": 13391 }, { "epoch": 13.59, "learning_rate": 4.560705352364314e-07, "loss": 0.0162, "step": 13392 }, { "epoch": 13.59, "learning_rate": 4.5541628614819144e-07, "loss": 0.0187, "step": 13393 }, { "epoch": 13.59, "learning_rate": 4.547624957335539e-07, "loss": 0.0186, "step": 13394 }, { "epoch": 13.59, "learning_rate": 4.5410916402393145e-07, "loss": 0.019, "step": 13395 }, { "epoch": 13.59, "learning_rate": 4.5345629105072343e-07, "loss": 0.0197, "step": 13396 }, { "epoch": 13.59, "learning_rate": 4.528038768453047e-07, "loss": 0.0175, "step": 13397 }, { "epoch": 13.6, "learning_rate": 4.5215192143902577e-07, "loss": 0.0192, "step": 13398 }, { "epoch": 13.6, "learning_rate": 4.515004248632182e-07, "loss": 0.0176, "step": 13399 }, { "epoch": 13.6, "learning_rate": 4.5084938714918923e-07, "loss": 0.0199, "step": 13400 }, { "epoch": 13.6, "learning_rate": 4.50198808328226e-07, "loss": 0.0208, "step": 13401 }, { "epoch": 13.6, "learning_rate": 4.495486884315925e-07, "loss": 0.0178, "step": 13402 }, { "epoch": 13.6, "learning_rate": 4.4889902749053025e-07, "loss": 0.0189, "step": 13403 }, { "epoch": 13.6, "learning_rate": 4.4824982553626107e-07, "loss": 0.0183, "step": 13404 }, { "epoch": 13.6, "learning_rate": 4.476010825999799e-07, "loss": 0.0171, "step": 13405 }, { "epoch": 13.6, "learning_rate": 4.469527987128663e-07, "loss": 0.0182, "step": 13406 }, { "epoch": 13.6, "learning_rate": 4.463049739060699e-07, "loss": 0.0177, "step": 13407 }, { "epoch": 13.61, "learning_rate": 4.456576082107256e-07, "loss": 0.0195, "step": 13408 }, { "epoch": 13.61, "learning_rate": 4.4501070165794305e-07, "loss": 0.0212, "step": 13409 }, { "epoch": 13.61, "learning_rate": 4.443642542788085e-07, "loss": 0.0221, "step": 13410 }, { "epoch": 13.61, "learning_rate": 4.437182661043893e-07, "loss": 0.0198, "step": 13411 }, { "epoch": 13.61, "learning_rate": 4.430727371657273e-07, "loss": 0.0157, "step": 13412 }, { "epoch": 13.61, "learning_rate": 4.424276674938444e-07, "loss": 0.0177, "step": 13413 }, { "epoch": 13.61, "learning_rate": 4.4178305711974124e-07, "loss": 0.0187, "step": 13414 }, { "epoch": 13.61, "learning_rate": 4.4113890607439315e-07, "loss": 0.0212, "step": 13415 }, { "epoch": 13.61, "learning_rate": 4.4049521438875753e-07, "loss": 0.0184, "step": 13416 }, { "epoch": 13.61, "learning_rate": 4.398519820937652e-07, "loss": 0.0224, "step": 13417 }, { "epoch": 13.62, "learning_rate": 4.3920920922032927e-07, "loss": 0.0181, "step": 13418 }, { "epoch": 13.62, "learning_rate": 4.385668957993372e-07, "loss": 0.0212, "step": 13419 }, { "epoch": 13.62, "learning_rate": 4.379250418616576e-07, "loss": 0.0214, "step": 13420 }, { "epoch": 13.62, "learning_rate": 4.372836474381348e-07, "loss": 0.0166, "step": 13421 }, { "epoch": 13.62, "learning_rate": 4.366427125595907e-07, "loss": 0.0207, "step": 13422 }, { "epoch": 13.62, "learning_rate": 4.360022372568273e-07, "loss": 0.0168, "step": 13423 }, { "epoch": 13.62, "learning_rate": 4.3536222156062237e-07, "loss": 0.0172, "step": 13424 }, { "epoch": 13.62, "learning_rate": 4.3472266550173223e-07, "loss": 0.0236, "step": 13425 }, { "epoch": 13.62, "learning_rate": 4.3408356911089134e-07, "loss": 0.0195, "step": 13426 }, { "epoch": 13.62, "learning_rate": 4.334449324188128e-07, "loss": 0.0219, "step": 13427 }, { "epoch": 13.63, "learning_rate": 4.3280675545618654e-07, "loss": 0.0183, "step": 13428 }, { "epoch": 13.63, "learning_rate": 4.321690382536814e-07, "loss": 0.0198, "step": 13429 }, { "epoch": 13.63, "learning_rate": 4.315317808419417e-07, "loss": 0.019, "step": 13430 }, { "epoch": 13.63, "learning_rate": 4.3089498325159296e-07, "loss": 0.0172, "step": 13431 }, { "epoch": 13.63, "learning_rate": 4.3025864551323627e-07, "loss": 0.0188, "step": 13432 }, { "epoch": 13.63, "learning_rate": 4.296227676574527e-07, "loss": 0.0188, "step": 13433 }, { "epoch": 13.63, "learning_rate": 4.289873497147967e-07, "loss": 0.0169, "step": 13434 }, { "epoch": 13.63, "learning_rate": 4.283523917158083e-07, "loss": 0.0188, "step": 13435 }, { "epoch": 13.63, "learning_rate": 4.2771789369099757e-07, "loss": 0.0204, "step": 13436 }, { "epoch": 13.63, "learning_rate": 4.2708385567085787e-07, "loss": 0.0182, "step": 13437 }, { "epoch": 13.64, "learning_rate": 4.264502776858581e-07, "loss": 0.0206, "step": 13438 }, { "epoch": 13.64, "learning_rate": 4.25817159766444e-07, "loss": 0.0192, "step": 13439 }, { "epoch": 13.64, "learning_rate": 4.251845019430412e-07, "loss": 0.0143, "step": 13440 }, { "epoch": 13.64, "learning_rate": 4.245523042460553e-07, "loss": 0.02, "step": 13441 }, { "epoch": 13.64, "learning_rate": 4.2392056670586325e-07, "loss": 0.0215, "step": 13442 }, { "epoch": 13.64, "learning_rate": 4.232892893528251e-07, "loss": 0.0201, "step": 13443 }, { "epoch": 13.64, "learning_rate": 4.2265847221727773e-07, "loss": 0.0183, "step": 13444 }, { "epoch": 13.64, "learning_rate": 4.220281153295369e-07, "loss": 0.02, "step": 13445 }, { "epoch": 13.64, "learning_rate": 4.213982187198928e-07, "loss": 0.0147, "step": 13446 }, { "epoch": 13.64, "learning_rate": 4.2076878241861687e-07, "loss": 0.018, "step": 13447 }, { "epoch": 13.65, "learning_rate": 4.201398064559581e-07, "loss": 0.0184, "step": 13448 }, { "epoch": 13.65, "learning_rate": 4.1951129086214015e-07, "loss": 0.0162, "step": 13449 }, { "epoch": 13.65, "learning_rate": 4.1888323566737e-07, "loss": 0.0176, "step": 13450 }, { "epoch": 13.65, "learning_rate": 4.182556409018246e-07, "loss": 0.0166, "step": 13451 }, { "epoch": 13.65, "learning_rate": 4.1762850659566866e-07, "loss": 0.0173, "step": 13452 }, { "epoch": 13.65, "learning_rate": 4.170018327790371e-07, "loss": 0.0202, "step": 13453 }, { "epoch": 13.65, "learning_rate": 4.163756194820456e-07, "loss": 0.0214, "step": 13454 }, { "epoch": 13.65, "learning_rate": 4.157498667347881e-07, "loss": 0.0202, "step": 13455 }, { "epoch": 13.65, "learning_rate": 4.151245745673349e-07, "loss": 0.019, "step": 13456 }, { "epoch": 13.65, "learning_rate": 4.144997430097342e-07, "loss": 0.0187, "step": 13457 }, { "epoch": 13.66, "learning_rate": 4.1387537209201636e-07, "loss": 0.0186, "step": 13458 }, { "epoch": 13.66, "learning_rate": 4.1325146184418075e-07, "loss": 0.0176, "step": 13459 }, { "epoch": 13.66, "learning_rate": 4.126280122962156e-07, "loss": 0.018, "step": 13460 }, { "epoch": 13.66, "learning_rate": 4.1200502347807704e-07, "loss": 0.0134, "step": 13461 }, { "epoch": 13.66, "learning_rate": 4.1138249541970543e-07, "loss": 0.0186, "step": 13462 }, { "epoch": 13.66, "learning_rate": 4.107604281510169e-07, "loss": 0.0169, "step": 13463 }, { "epoch": 13.66, "learning_rate": 4.101388217019031e-07, "loss": 0.0174, "step": 13464 }, { "epoch": 13.66, "learning_rate": 4.095176761022401e-07, "loss": 0.0212, "step": 13465 }, { "epoch": 13.66, "learning_rate": 4.088969913818741e-07, "loss": 0.0191, "step": 13466 }, { "epoch": 13.67, "learning_rate": 4.0827676757063453e-07, "loss": 0.0179, "step": 13467 }, { "epoch": 13.67, "learning_rate": 4.0765700469832416e-07, "loss": 0.0193, "step": 13468 }, { "epoch": 13.67, "learning_rate": 4.070377027947292e-07, "loss": 0.0199, "step": 13469 }, { "epoch": 13.67, "learning_rate": 4.0641886188961145e-07, "loss": 0.018, "step": 13470 }, { "epoch": 13.67, "learning_rate": 4.0580048201270597e-07, "loss": 0.0158, "step": 13471 }, { "epoch": 13.67, "learning_rate": 4.051825631937334e-07, "loss": 0.0176, "step": 13472 }, { "epoch": 13.67, "learning_rate": 4.045651054623856e-07, "loss": 0.0186, "step": 13473 }, { "epoch": 13.67, "learning_rate": 4.0394810884833656e-07, "loss": 0.0196, "step": 13474 }, { "epoch": 13.67, "learning_rate": 4.033315733812371e-07, "loss": 0.0195, "step": 13475 }, { "epoch": 13.67, "learning_rate": 4.027154990907123e-07, "loss": 0.0165, "step": 13476 }, { "epoch": 13.68, "learning_rate": 4.0209988600637296e-07, "loss": 0.0187, "step": 13477 }, { "epoch": 13.68, "learning_rate": 4.014847341577988e-07, "loss": 0.017, "step": 13478 }, { "epoch": 13.68, "learning_rate": 4.0087004357455495e-07, "loss": 0.0181, "step": 13479 }, { "epoch": 13.68, "learning_rate": 4.002558142861779e-07, "loss": 0.0207, "step": 13480 }, { "epoch": 13.68, "learning_rate": 3.996420463221851e-07, "loss": 0.0174, "step": 13481 }, { "epoch": 13.68, "learning_rate": 3.990287397120751e-07, "loss": 0.0203, "step": 13482 }, { "epoch": 13.68, "learning_rate": 3.984158944853167e-07, "loss": 0.0177, "step": 13483 }, { "epoch": 13.68, "learning_rate": 3.978035106713629e-07, "loss": 0.016, "step": 13484 }, { "epoch": 13.68, "learning_rate": 3.971915882996424e-07, "loss": 0.0167, "step": 13485 }, { "epoch": 13.68, "learning_rate": 3.9658012739956064e-07, "loss": 0.0204, "step": 13486 }, { "epoch": 13.69, "learning_rate": 3.95969128000504e-07, "loss": 0.0152, "step": 13487 }, { "epoch": 13.69, "learning_rate": 3.953585901318324e-07, "loss": 0.0191, "step": 13488 }, { "epoch": 13.69, "learning_rate": 3.947485138228879e-07, "loss": 0.0161, "step": 13489 }, { "epoch": 13.69, "learning_rate": 3.9413889910298596e-07, "loss": 0.0208, "step": 13490 }, { "epoch": 13.69, "learning_rate": 3.9352974600142205e-07, "loss": 0.0161, "step": 13491 }, { "epoch": 13.69, "learning_rate": 3.929210545474738e-07, "loss": 0.0201, "step": 13492 }, { "epoch": 13.69, "learning_rate": 3.923128247703856e-07, "loss": 0.0204, "step": 13493 }, { "epoch": 13.69, "learning_rate": 3.9170505669939295e-07, "loss": 0.0201, "step": 13494 }, { "epoch": 13.69, "learning_rate": 3.910977503636992e-07, "loss": 0.0172, "step": 13495 }, { "epoch": 13.69, "learning_rate": 3.9049090579248995e-07, "loss": 0.0164, "step": 13496 }, { "epoch": 13.7, "learning_rate": 3.898845230149273e-07, "loss": 0.0186, "step": 13497 }, { "epoch": 13.7, "learning_rate": 3.892786020601513e-07, "loss": 0.0177, "step": 13498 }, { "epoch": 13.7, "learning_rate": 3.886731429572821e-07, "loss": 0.0188, "step": 13499 }, { "epoch": 13.7, "learning_rate": 3.8806814573541185e-07, "loss": 0.0179, "step": 13500 }, { "epoch": 13.7, "learning_rate": 3.8746361042361735e-07, "loss": 0.0191, "step": 13501 }, { "epoch": 13.7, "learning_rate": 3.868595370509487e-07, "loss": 0.017, "step": 13502 }, { "epoch": 13.7, "learning_rate": 3.86255925646436e-07, "loss": 0.0186, "step": 13503 }, { "epoch": 13.7, "learning_rate": 3.8565277623908603e-07, "loss": 0.0168, "step": 13504 }, { "epoch": 13.7, "learning_rate": 3.850500888578834e-07, "loss": 0.0175, "step": 13505 }, { "epoch": 13.7, "learning_rate": 3.8444786353179164e-07, "loss": 0.0207, "step": 13506 }, { "epoch": 13.71, "learning_rate": 3.838461002897498e-07, "loss": 0.018, "step": 13507 }, { "epoch": 13.71, "learning_rate": 3.8324479916067694e-07, "loss": 0.0207, "step": 13508 }, { "epoch": 13.71, "learning_rate": 3.826439601734699e-07, "loss": 0.0161, "step": 13509 }, { "epoch": 13.71, "learning_rate": 3.820435833569991e-07, "loss": 0.0215, "step": 13510 }, { "epoch": 13.71, "learning_rate": 3.8144366874012127e-07, "loss": 0.015, "step": 13511 }, { "epoch": 13.71, "learning_rate": 3.8084421635166347e-07, "loss": 0.0165, "step": 13512 }, { "epoch": 13.71, "learning_rate": 3.8024522622043146e-07, "loss": 0.0166, "step": 13513 }, { "epoch": 13.71, "learning_rate": 3.796466983752134e-07, "loss": 0.0186, "step": 13514 }, { "epoch": 13.71, "learning_rate": 3.7904863284476847e-07, "loss": 0.0214, "step": 13515 }, { "epoch": 13.71, "learning_rate": 3.784510296578414e-07, "loss": 0.0191, "step": 13516 }, { "epoch": 13.72, "learning_rate": 3.778538888431471e-07, "loss": 0.0194, "step": 13517 }, { "epoch": 13.72, "learning_rate": 3.7725721042938146e-07, "loss": 0.0201, "step": 13518 }, { "epoch": 13.72, "learning_rate": 3.766609944452215e-07, "loss": 0.0214, "step": 13519 }, { "epoch": 13.72, "learning_rate": 3.760652409193166e-07, "loss": 0.019, "step": 13520 }, { "epoch": 13.72, "learning_rate": 3.754699498802972e-07, "loss": 0.0164, "step": 13521 }, { "epoch": 13.72, "learning_rate": 3.748751213567703e-07, "loss": 0.0183, "step": 13522 }, { "epoch": 13.72, "learning_rate": 3.74280755377322e-07, "loss": 0.0188, "step": 13523 }, { "epoch": 13.72, "learning_rate": 3.736868519705117e-07, "loss": 0.0166, "step": 13524 }, { "epoch": 13.72, "learning_rate": 3.7309341116488317e-07, "loss": 0.0195, "step": 13525 }, { "epoch": 13.73, "learning_rate": 3.7250043298895365e-07, "loss": 0.0179, "step": 13526 }, { "epoch": 13.73, "learning_rate": 3.719079174712192e-07, "loss": 0.0154, "step": 13527 }, { "epoch": 13.73, "learning_rate": 3.713158646401549e-07, "loss": 0.0203, "step": 13528 }, { "epoch": 13.73, "learning_rate": 3.7072427452421013e-07, "loss": 0.02, "step": 13529 }, { "epoch": 13.73, "learning_rate": 3.701331471518155e-07, "loss": 0.0183, "step": 13530 }, { "epoch": 13.73, "learning_rate": 3.6954248255138045e-07, "loss": 0.0181, "step": 13531 }, { "epoch": 13.73, "learning_rate": 3.6895228075128465e-07, "loss": 0.0192, "step": 13532 }, { "epoch": 13.73, "learning_rate": 3.683625417798964e-07, "loss": 0.0188, "step": 13533 }, { "epoch": 13.73, "learning_rate": 3.6777326566555196e-07, "loss": 0.0177, "step": 13534 }, { "epoch": 13.73, "learning_rate": 3.671844524365709e-07, "loss": 0.0205, "step": 13535 }, { "epoch": 13.74, "learning_rate": 3.665961021212494e-07, "loss": 0.0178, "step": 13536 }, { "epoch": 13.74, "learning_rate": 3.6600821474786054e-07, "loss": 0.0151, "step": 13537 }, { "epoch": 13.74, "learning_rate": 3.654207903446583e-07, "loss": 0.0183, "step": 13538 }, { "epoch": 13.74, "learning_rate": 3.6483382893986896e-07, "loss": 0.0191, "step": 13539 }, { "epoch": 13.74, "learning_rate": 3.6424733056170005e-07, "loss": 0.0182, "step": 13540 }, { "epoch": 13.74, "learning_rate": 3.636612952383378e-07, "loss": 0.0167, "step": 13541 }, { "epoch": 13.74, "learning_rate": 3.63075722997942e-07, "loss": 0.0158, "step": 13542 }, { "epoch": 13.74, "learning_rate": 3.624906138686546e-07, "loss": 0.0154, "step": 13543 }, { "epoch": 13.74, "learning_rate": 3.619059678785941e-07, "loss": 0.016, "step": 13544 }, { "epoch": 13.74, "learning_rate": 3.613217850558559e-07, "loss": 0.0215, "step": 13545 }, { "epoch": 13.75, "learning_rate": 3.6073806542851196e-07, "loss": 0.0195, "step": 13546 }, { "epoch": 13.75, "learning_rate": 3.601548090246143e-07, "loss": 0.0186, "step": 13547 }, { "epoch": 13.75, "learning_rate": 3.5957201587219384e-07, "loss": 0.0163, "step": 13548 }, { "epoch": 13.75, "learning_rate": 3.5898968599925364e-07, "loss": 0.019, "step": 13549 }, { "epoch": 13.75, "learning_rate": 3.584078194337814e-07, "loss": 0.0207, "step": 13550 }, { "epoch": 13.75, "learning_rate": 3.578264162037348e-07, "loss": 0.0219, "step": 13551 }, { "epoch": 13.75, "learning_rate": 3.572454763370603e-07, "loss": 0.0204, "step": 13552 }, { "epoch": 13.75, "learning_rate": 3.5666499986166893e-07, "loss": 0.0158, "step": 13553 }, { "epoch": 13.75, "learning_rate": 3.5608498680545944e-07, "loss": 0.0173, "step": 13554 }, { "epoch": 13.75, "learning_rate": 3.5550543719630624e-07, "loss": 0.0174, "step": 13555 }, { "epoch": 13.76, "learning_rate": 3.54926351062056e-07, "loss": 0.0174, "step": 13556 }, { "epoch": 13.76, "learning_rate": 3.5434772843053966e-07, "loss": 0.0175, "step": 13557 }, { "epoch": 13.76, "learning_rate": 3.5376956932956506e-07, "loss": 0.0203, "step": 13558 }, { "epoch": 13.76, "learning_rate": 3.531918737869122e-07, "loss": 0.0169, "step": 13559 }, { "epoch": 13.76, "learning_rate": 3.5261464183034553e-07, "loss": 0.0193, "step": 13560 }, { "epoch": 13.76, "learning_rate": 3.5203787348760286e-07, "loss": 0.0194, "step": 13561 }, { "epoch": 13.76, "learning_rate": 3.5146156878640426e-07, "loss": 0.0175, "step": 13562 }, { "epoch": 13.76, "learning_rate": 3.5088572775444086e-07, "loss": 0.0187, "step": 13563 }, { "epoch": 13.76, "learning_rate": 3.503103504193872e-07, "loss": 0.0198, "step": 13564 }, { "epoch": 13.76, "learning_rate": 3.4973543680889457e-07, "loss": 0.0177, "step": 13565 }, { "epoch": 13.77, "learning_rate": 3.491609869505885e-07, "loss": 0.0214, "step": 13566 }, { "epoch": 13.77, "learning_rate": 3.4858700087207706e-07, "loss": 0.0177, "step": 13567 }, { "epoch": 13.77, "learning_rate": 3.4801347860093924e-07, "loss": 0.0178, "step": 13568 }, { "epoch": 13.77, "learning_rate": 3.47440420164743e-07, "loss": 0.0204, "step": 13569 }, { "epoch": 13.77, "learning_rate": 3.468678255910218e-07, "loss": 0.0161, "step": 13570 }, { "epoch": 13.77, "learning_rate": 3.4629569490729375e-07, "loss": 0.019, "step": 13571 }, { "epoch": 13.77, "learning_rate": 3.4572402814105455e-07, "loss": 0.0189, "step": 13572 }, { "epoch": 13.77, "learning_rate": 3.4515282531977444e-07, "loss": 0.0139, "step": 13573 }, { "epoch": 13.77, "learning_rate": 3.4458208647090376e-07, "loss": 0.0195, "step": 13574 }, { "epoch": 13.77, "learning_rate": 3.440118116218694e-07, "loss": 0.0175, "step": 13575 }, { "epoch": 13.78, "learning_rate": 3.4344200080007605e-07, "loss": 0.0193, "step": 13576 }, { "epoch": 13.78, "learning_rate": 3.428726540329086e-07, "loss": 0.0157, "step": 13577 }, { "epoch": 13.78, "learning_rate": 3.423037713477251e-07, "loss": 0.0182, "step": 13578 }, { "epoch": 13.78, "learning_rate": 3.417353527718659e-07, "loss": 0.0179, "step": 13579 }, { "epoch": 13.78, "learning_rate": 3.411673983326447e-07, "loss": 0.0194, "step": 13580 }, { "epoch": 13.78, "learning_rate": 3.405999080573552e-07, "loss": 0.0189, "step": 13581 }, { "epoch": 13.78, "learning_rate": 3.400328819732712e-07, "loss": 0.0192, "step": 13582 }, { "epoch": 13.78, "learning_rate": 3.394663201076398e-07, "loss": 0.0169, "step": 13583 }, { "epoch": 13.78, "learning_rate": 3.3890022248768696e-07, "loss": 0.0182, "step": 13584 }, { "epoch": 13.78, "learning_rate": 3.383345891406176e-07, "loss": 0.0186, "step": 13585 }, { "epoch": 13.79, "learning_rate": 3.3776942009361435e-07, "loss": 0.0189, "step": 13586 }, { "epoch": 13.79, "learning_rate": 3.372047153738378e-07, "loss": 0.0179, "step": 13587 }, { "epoch": 13.79, "learning_rate": 3.366404750084229e-07, "loss": 0.0209, "step": 13588 }, { "epoch": 13.79, "learning_rate": 3.360766990244868e-07, "loss": 0.0204, "step": 13589 }, { "epoch": 13.79, "learning_rate": 3.355133874491201e-07, "loss": 0.0187, "step": 13590 }, { "epoch": 13.79, "learning_rate": 3.3495054030939553e-07, "loss": 0.0162, "step": 13591 }, { "epoch": 13.79, "learning_rate": 3.3438815763236044e-07, "loss": 0.018, "step": 13592 }, { "epoch": 13.79, "learning_rate": 3.3382623944503866e-07, "loss": 0.0171, "step": 13593 }, { "epoch": 13.79, "learning_rate": 3.3326478577443757e-07, "loss": 0.0229, "step": 13594 }, { "epoch": 13.8, "learning_rate": 3.3270379664753547e-07, "loss": 0.0166, "step": 13595 }, { "epoch": 13.8, "learning_rate": 3.321432720912932e-07, "loss": 0.0188, "step": 13596 }, { "epoch": 13.8, "learning_rate": 3.315832121326434e-07, "loss": 0.0147, "step": 13597 }, { "epoch": 13.8, "learning_rate": 3.310236167985048e-07, "loss": 0.0215, "step": 13598 }, { "epoch": 13.8, "learning_rate": 3.304644861157669e-07, "loss": 0.0218, "step": 13599 }, { "epoch": 13.8, "learning_rate": 3.2990582011129923e-07, "loss": 0.0184, "step": 13600 }, { "epoch": 13.8, "learning_rate": 3.2934761881195044e-07, "loss": 0.0197, "step": 13601 }, { "epoch": 13.8, "learning_rate": 3.2878988224454346e-07, "loss": 0.0194, "step": 13602 }, { "epoch": 13.8, "learning_rate": 3.2823261043588126e-07, "loss": 0.0196, "step": 13603 }, { "epoch": 13.8, "learning_rate": 3.2767580341274587e-07, "loss": 0.0205, "step": 13604 }, { "epoch": 13.81, "learning_rate": 3.271194612018935e-07, "loss": 0.0199, "step": 13605 }, { "epoch": 13.81, "learning_rate": 3.265635838300607e-07, "loss": 0.0154, "step": 13606 }, { "epoch": 13.81, "learning_rate": 3.260081713239582e-07, "loss": 0.0185, "step": 13607 }, { "epoch": 13.81, "learning_rate": 3.254532237102803e-07, "loss": 0.0179, "step": 13608 }, { "epoch": 13.81, "learning_rate": 3.248987410156934e-07, "loss": 0.0173, "step": 13609 }, { "epoch": 13.81, "learning_rate": 3.2434472326684173e-07, "loss": 0.0151, "step": 13610 }, { "epoch": 13.81, "learning_rate": 3.2379117049035513e-07, "loss": 0.0181, "step": 13611 }, { "epoch": 13.81, "learning_rate": 3.2323808271282895e-07, "loss": 0.0199, "step": 13612 }, { "epoch": 13.81, "learning_rate": 3.226854599608453e-07, "loss": 0.0174, "step": 13613 }, { "epoch": 13.81, "learning_rate": 3.2213330226096185e-07, "loss": 0.0181, "step": 13614 }, { "epoch": 13.82, "learning_rate": 3.2158160963971063e-07, "loss": 0.0205, "step": 13615 }, { "epoch": 13.82, "learning_rate": 3.2103038212360494e-07, "loss": 0.0186, "step": 13616 }, { "epoch": 13.82, "learning_rate": 3.204796197391347e-07, "loss": 0.0214, "step": 13617 }, { "epoch": 13.82, "learning_rate": 3.1992932251276533e-07, "loss": 0.0202, "step": 13618 }, { "epoch": 13.82, "learning_rate": 3.1937949047094465e-07, "loss": 0.0188, "step": 13619 }, { "epoch": 13.82, "learning_rate": 3.1883012364009367e-07, "loss": 0.0214, "step": 13620 }, { "epoch": 13.82, "learning_rate": 3.182812220466136e-07, "loss": 0.0176, "step": 13621 }, { "epoch": 13.82, "learning_rate": 3.1773278571688217e-07, "loss": 0.0173, "step": 13622 }, { "epoch": 13.82, "learning_rate": 3.171848146772549e-07, "loss": 0.0175, "step": 13623 }, { "epoch": 13.82, "learning_rate": 3.1663730895406417e-07, "loss": 0.0184, "step": 13624 }, { "epoch": 13.83, "learning_rate": 3.1609026857362113e-07, "loss": 0.017, "step": 13625 }, { "epoch": 13.83, "learning_rate": 3.1554369356221694e-07, "loss": 0.0192, "step": 13626 }, { "epoch": 13.83, "learning_rate": 3.149975839461128e-07, "loss": 0.0207, "step": 13627 }, { "epoch": 13.83, "learning_rate": 3.1445193975155773e-07, "loss": 0.018, "step": 13628 }, { "epoch": 13.83, "learning_rate": 3.139067610047697e-07, "loss": 0.0151, "step": 13629 }, { "epoch": 13.83, "learning_rate": 3.1336204773194765e-07, "loss": 0.0211, "step": 13630 }, { "epoch": 13.83, "learning_rate": 3.128177999592719e-07, "loss": 0.0171, "step": 13631 }, { "epoch": 13.83, "learning_rate": 3.1227401771289246e-07, "loss": 0.015, "step": 13632 }, { "epoch": 13.83, "learning_rate": 3.117307010189452e-07, "loss": 0.015, "step": 13633 }, { "epoch": 13.83, "learning_rate": 3.111878499035359e-07, "loss": 0.0199, "step": 13634 }, { "epoch": 13.84, "learning_rate": 3.1064546439275256e-07, "loss": 0.0169, "step": 13635 }, { "epoch": 13.84, "learning_rate": 3.1010354451266213e-07, "loss": 0.0166, "step": 13636 }, { "epoch": 13.84, "learning_rate": 3.0956209028930483e-07, "loss": 0.0213, "step": 13637 }, { "epoch": 13.84, "learning_rate": 3.0902110174870324e-07, "loss": 0.0186, "step": 13638 }, { "epoch": 13.84, "learning_rate": 3.0848057891685324e-07, "loss": 0.0185, "step": 13639 }, { "epoch": 13.84, "learning_rate": 3.0794052181973067e-07, "loss": 0.0159, "step": 13640 }, { "epoch": 13.84, "learning_rate": 3.07400930483287e-07, "loss": 0.0186, "step": 13641 }, { "epoch": 13.84, "learning_rate": 3.0686180493345483e-07, "loss": 0.02, "step": 13642 }, { "epoch": 13.84, "learning_rate": 3.0632314519614126e-07, "loss": 0.0192, "step": 13643 }, { "epoch": 13.84, "learning_rate": 3.0578495129723216e-07, "loss": 0.0193, "step": 13644 }, { "epoch": 13.85, "learning_rate": 3.052472232625925e-07, "loss": 0.0191, "step": 13645 }, { "epoch": 13.85, "learning_rate": 3.047099611180604e-07, "loss": 0.0173, "step": 13646 }, { "epoch": 13.85, "learning_rate": 3.0417316488945637e-07, "loss": 0.0196, "step": 13647 }, { "epoch": 13.85, "learning_rate": 3.036368346025764e-07, "loss": 0.0193, "step": 13648 }, { "epoch": 13.85, "learning_rate": 3.031009702831933e-07, "loss": 0.0188, "step": 13649 }, { "epoch": 13.85, "learning_rate": 3.025655719570608e-07, "loss": 0.0182, "step": 13650 }, { "epoch": 13.85, "learning_rate": 3.020306396499062e-07, "loss": 0.0166, "step": 13651 }, { "epoch": 13.85, "learning_rate": 3.0149617338743555e-07, "loss": 0.0169, "step": 13652 }, { "epoch": 13.85, "learning_rate": 3.0096217319533386e-07, "loss": 0.0178, "step": 13653 }, { "epoch": 13.85, "learning_rate": 3.004286390992639e-07, "loss": 0.021, "step": 13654 }, { "epoch": 13.86, "learning_rate": 2.998955711248652e-07, "loss": 0.0205, "step": 13655 }, { "epoch": 13.86, "learning_rate": 2.9936296929775286e-07, "loss": 0.0178, "step": 13656 }, { "epoch": 13.86, "learning_rate": 2.9883083364352305e-07, "loss": 0.0193, "step": 13657 }, { "epoch": 13.86, "learning_rate": 2.982991641877486e-07, "loss": 0.016, "step": 13658 }, { "epoch": 13.86, "learning_rate": 2.977679609559769e-07, "loss": 0.0217, "step": 13659 }, { "epoch": 13.86, "learning_rate": 2.9723722397373755e-07, "loss": 0.0179, "step": 13660 }, { "epoch": 13.86, "learning_rate": 2.9670695326653456e-07, "loss": 0.0186, "step": 13661 }, { "epoch": 13.86, "learning_rate": 2.961771488598519e-07, "loss": 0.0218, "step": 13662 }, { "epoch": 13.86, "learning_rate": 2.9564781077914826e-07, "loss": 0.0242, "step": 13663 }, { "epoch": 13.87, "learning_rate": 2.951189390498621e-07, "loss": 0.0213, "step": 13664 }, { "epoch": 13.87, "learning_rate": 2.9459053369740973e-07, "loss": 0.0172, "step": 13665 }, { "epoch": 13.87, "learning_rate": 2.9406259474718313e-07, "loss": 0.0182, "step": 13666 }, { "epoch": 13.87, "learning_rate": 2.9353512222455305e-07, "loss": 0.0191, "step": 13667 }, { "epoch": 13.87, "learning_rate": 2.930081161548659e-07, "loss": 0.0182, "step": 13668 }, { "epoch": 13.87, "learning_rate": 2.924815765634503e-07, "loss": 0.019, "step": 13669 }, { "epoch": 13.87, "learning_rate": 2.919555034756083e-07, "loss": 0.0211, "step": 13670 }, { "epoch": 13.87, "learning_rate": 2.9142989691662073e-07, "loss": 0.0188, "step": 13671 }, { "epoch": 13.87, "learning_rate": 2.9090475691174736e-07, "loss": 0.0171, "step": 13672 }, { "epoch": 13.87, "learning_rate": 2.903800834862225e-07, "loss": 0.0165, "step": 13673 }, { "epoch": 13.88, "learning_rate": 2.8985587666526036e-07, "loss": 0.0212, "step": 13674 }, { "epoch": 13.88, "learning_rate": 2.8933213647405314e-07, "loss": 0.018, "step": 13675 }, { "epoch": 13.88, "learning_rate": 2.8880886293776834e-07, "loss": 0.0181, "step": 13676 }, { "epoch": 13.88, "learning_rate": 2.882860560815526e-07, "loss": 0.018, "step": 13677 }, { "epoch": 13.88, "learning_rate": 2.8776371593053023e-07, "loss": 0.0188, "step": 13678 }, { "epoch": 13.88, "learning_rate": 2.872418425098034e-07, "loss": 0.0195, "step": 13679 }, { "epoch": 13.88, "learning_rate": 2.8672043584444866e-07, "loss": 0.0191, "step": 13680 }, { "epoch": 13.88, "learning_rate": 2.86199495959526e-07, "loss": 0.0184, "step": 13681 }, { "epoch": 13.88, "learning_rate": 2.8567902288006877e-07, "loss": 0.0151, "step": 13682 }, { "epoch": 13.88, "learning_rate": 2.851590166310858e-07, "loss": 0.0193, "step": 13683 }, { "epoch": 13.89, "learning_rate": 2.8463947723757045e-07, "loss": 0.0153, "step": 13684 }, { "epoch": 13.89, "learning_rate": 2.841204047244872e-07, "loss": 0.0192, "step": 13685 }, { "epoch": 13.89, "learning_rate": 2.8360179911678164e-07, "loss": 0.0183, "step": 13686 }, { "epoch": 13.89, "learning_rate": 2.8308366043937606e-07, "loss": 0.0206, "step": 13687 }, { "epoch": 13.89, "learning_rate": 2.8256598871716943e-07, "loss": 0.0194, "step": 13688 }, { "epoch": 13.89, "learning_rate": 2.8204878397503963e-07, "loss": 0.0199, "step": 13689 }, { "epoch": 13.89, "learning_rate": 2.8153204623784015e-07, "loss": 0.0174, "step": 13690 }, { "epoch": 13.89, "learning_rate": 2.810157755304033e-07, "loss": 0.0204, "step": 13691 }, { "epoch": 13.89, "learning_rate": 2.804999718775414e-07, "loss": 0.0176, "step": 13692 }, { "epoch": 13.89, "learning_rate": 2.7998463530403695e-07, "loss": 0.0157, "step": 13693 }, { "epoch": 13.9, "learning_rate": 2.7946976583466123e-07, "loss": 0.0206, "step": 13694 }, { "epoch": 13.9, "learning_rate": 2.789553634941522e-07, "loss": 0.0213, "step": 13695 }, { "epoch": 13.9, "learning_rate": 2.784414283072323e-07, "loss": 0.0178, "step": 13696 }, { "epoch": 13.9, "learning_rate": 2.779279602985974e-07, "loss": 0.0212, "step": 13697 }, { "epoch": 13.9, "learning_rate": 2.774149594929232e-07, "loss": 0.0179, "step": 13698 }, { "epoch": 13.9, "learning_rate": 2.769024259148634e-07, "loss": 0.0191, "step": 13699 }, { "epoch": 13.9, "learning_rate": 2.76390359589046e-07, "loss": 0.0188, "step": 13700 }, { "epoch": 13.9, "learning_rate": 2.7587876054008145e-07, "loss": 0.0187, "step": 13701 }, { "epoch": 13.9, "learning_rate": 2.753676287925533e-07, "loss": 0.0191, "step": 13702 }, { "epoch": 13.9, "learning_rate": 2.7485696437102527e-07, "loss": 0.0194, "step": 13703 }, { "epoch": 13.91, "learning_rate": 2.7434676730003886e-07, "loss": 0.019, "step": 13704 }, { "epoch": 13.91, "learning_rate": 2.7383703760410995e-07, "loss": 0.0183, "step": 13705 }, { "epoch": 13.91, "learning_rate": 2.733277753077357e-07, "loss": 0.0146, "step": 13706 }, { "epoch": 13.91, "learning_rate": 2.7281898043538757e-07, "loss": 0.0173, "step": 13707 }, { "epoch": 13.91, "learning_rate": 2.7231065301151607e-07, "loss": 0.0176, "step": 13708 }, { "epoch": 13.91, "learning_rate": 2.718027930605527e-07, "loss": 0.0189, "step": 13709 }, { "epoch": 13.91, "learning_rate": 2.71295400606898e-07, "loss": 0.0198, "step": 13710 }, { "epoch": 13.91, "learning_rate": 2.7078847567494013e-07, "loss": 0.0201, "step": 13711 }, { "epoch": 13.91, "learning_rate": 2.7028201828903646e-07, "loss": 0.0195, "step": 13712 }, { "epoch": 13.91, "learning_rate": 2.697760284735273e-07, "loss": 0.0169, "step": 13713 }, { "epoch": 13.92, "learning_rate": 2.692705062527257e-07, "loss": 0.0203, "step": 13714 }, { "epoch": 13.92, "learning_rate": 2.687654516509275e-07, "loss": 0.0168, "step": 13715 }, { "epoch": 13.92, "learning_rate": 2.6826086469240455e-07, "loss": 0.0149, "step": 13716 }, { "epoch": 13.92, "learning_rate": 2.6775674540140183e-07, "loss": 0.0217, "step": 13717 }, { "epoch": 13.92, "learning_rate": 2.672530938021467e-07, "loss": 0.0195, "step": 13718 }, { "epoch": 13.92, "learning_rate": 2.667499099188431e-07, "loss": 0.0208, "step": 13719 }, { "epoch": 13.92, "learning_rate": 2.662471937756728e-07, "loss": 0.016, "step": 13720 }, { "epoch": 13.92, "learning_rate": 2.657449453967931e-07, "loss": 0.0173, "step": 13721 }, { "epoch": 13.92, "learning_rate": 2.6524316480633916e-07, "loss": 0.0205, "step": 13722 }, { "epoch": 13.92, "learning_rate": 2.647418520284262e-07, "loss": 0.0168, "step": 13723 }, { "epoch": 13.93, "learning_rate": 2.6424100708714374e-07, "loss": 0.0167, "step": 13724 }, { "epoch": 13.93, "learning_rate": 2.6374063000656145e-07, "loss": 0.0178, "step": 13725 }, { "epoch": 13.93, "learning_rate": 2.632407208107257e-07, "loss": 0.0167, "step": 13726 }, { "epoch": 13.93, "learning_rate": 2.6274127952365834e-07, "loss": 0.0193, "step": 13727 }, { "epoch": 13.93, "learning_rate": 2.6224230616936244e-07, "loss": 0.0198, "step": 13728 }, { "epoch": 13.93, "learning_rate": 2.617438007718154e-07, "loss": 0.0192, "step": 13729 }, { "epoch": 13.93, "learning_rate": 2.6124576335497364e-07, "loss": 0.0183, "step": 13730 }, { "epoch": 13.93, "learning_rate": 2.607481939427714e-07, "loss": 0.0201, "step": 13731 }, { "epoch": 13.93, "learning_rate": 2.602510925591195e-07, "loss": 0.0157, "step": 13732 }, { "epoch": 13.94, "learning_rate": 2.5975445922790555e-07, "loss": 0.0176, "step": 13733 }, { "epoch": 13.94, "learning_rate": 2.592582939729971e-07, "loss": 0.0206, "step": 13734 }, { "epoch": 13.94, "learning_rate": 2.587625968182361e-07, "loss": 0.0214, "step": 13735 }, { "epoch": 13.94, "learning_rate": 2.582673677874459e-07, "loss": 0.0174, "step": 13736 }, { "epoch": 13.94, "learning_rate": 2.5777260690442397e-07, "loss": 0.0146, "step": 13737 }, { "epoch": 13.94, "learning_rate": 2.572783141929469e-07, "loss": 0.0175, "step": 13738 }, { "epoch": 13.94, "learning_rate": 2.567844896767679e-07, "loss": 0.0204, "step": 13739 }, { "epoch": 13.94, "learning_rate": 2.562911333796192e-07, "loss": 0.0205, "step": 13740 }, { "epoch": 13.94, "learning_rate": 2.5579824532520833e-07, "loss": 0.0213, "step": 13741 }, { "epoch": 13.94, "learning_rate": 2.5530582553722093e-07, "loss": 0.0169, "step": 13742 }, { "epoch": 13.95, "learning_rate": 2.5481387403932246e-07, "loss": 0.0173, "step": 13743 }, { "epoch": 13.95, "learning_rate": 2.5432239085515176e-07, "loss": 0.0159, "step": 13744 }, { "epoch": 13.95, "learning_rate": 2.5383137600833105e-07, "loss": 0.0172, "step": 13745 }, { "epoch": 13.95, "learning_rate": 2.5334082952245374e-07, "loss": 0.0191, "step": 13746 }, { "epoch": 13.95, "learning_rate": 2.528507514210932e-07, "loss": 0.0173, "step": 13747 }, { "epoch": 13.95, "learning_rate": 2.5236114172780266e-07, "loss": 0.0195, "step": 13748 }, { "epoch": 13.95, "learning_rate": 2.5187200046610904e-07, "loss": 0.0182, "step": 13749 }, { "epoch": 13.95, "learning_rate": 2.513833276595201e-07, "loss": 0.0183, "step": 13750 }, { "epoch": 13.95, "learning_rate": 2.508951233315171e-07, "loss": 0.019, "step": 13751 }, { "epoch": 13.95, "learning_rate": 2.504073875055624e-07, "loss": 0.0161, "step": 13752 }, { "epoch": 13.96, "learning_rate": 2.499201202050949e-07, "loss": 0.0174, "step": 13753 }, { "epoch": 13.96, "learning_rate": 2.494333214535305e-07, "loss": 0.0212, "step": 13754 }, { "epoch": 13.96, "learning_rate": 2.489469912742637e-07, "loss": 0.0174, "step": 13755 }, { "epoch": 13.96, "learning_rate": 2.484611296906636e-07, "loss": 0.0169, "step": 13756 }, { "epoch": 13.96, "learning_rate": 2.479757367260793e-07, "loss": 0.0193, "step": 13757 }, { "epoch": 13.96, "learning_rate": 2.4749081240383776e-07, "loss": 0.0192, "step": 13758 }, { "epoch": 13.96, "learning_rate": 2.470063567472414e-07, "loss": 0.0186, "step": 13759 }, { "epoch": 13.96, "learning_rate": 2.4652236977957046e-07, "loss": 0.0192, "step": 13760 }, { "epoch": 13.96, "learning_rate": 2.4603885152408635e-07, "loss": 0.0188, "step": 13761 }, { "epoch": 13.96, "learning_rate": 2.455558020040227e-07, "loss": 0.0186, "step": 13762 }, { "epoch": 13.97, "learning_rate": 2.4507322124259214e-07, "loss": 0.0175, "step": 13763 }, { "epoch": 13.97, "learning_rate": 2.445911092629882e-07, "loss": 0.02, "step": 13764 }, { "epoch": 13.97, "learning_rate": 2.4410946608837693e-07, "loss": 0.0202, "step": 13765 }, { "epoch": 13.97, "learning_rate": 2.4362829174190527e-07, "loss": 0.0205, "step": 13766 }, { "epoch": 13.97, "learning_rate": 2.431475862466959e-07, "loss": 0.0153, "step": 13767 }, { "epoch": 13.97, "learning_rate": 2.4266734962585026e-07, "loss": 0.0205, "step": 13768 }, { "epoch": 13.97, "learning_rate": 2.421875819024444e-07, "loss": 0.0157, "step": 13769 }, { "epoch": 13.97, "learning_rate": 2.417082830995365e-07, "loss": 0.0158, "step": 13770 }, { "epoch": 13.97, "learning_rate": 2.4122945324015933e-07, "loss": 0.0175, "step": 13771 }, { "epoch": 13.97, "learning_rate": 2.407510923473233e-07, "loss": 0.0214, "step": 13772 }, { "epoch": 13.98, "learning_rate": 2.402732004440156e-07, "loss": 0.0161, "step": 13773 }, { "epoch": 13.98, "learning_rate": 2.3979577755320224e-07, "loss": 0.0177, "step": 13774 }, { "epoch": 13.98, "learning_rate": 2.393188236978272e-07, "loss": 0.0192, "step": 13775 }, { "epoch": 13.98, "learning_rate": 2.388423389008088e-07, "loss": 0.0182, "step": 13776 }, { "epoch": 13.98, "learning_rate": 2.3836632318504638e-07, "loss": 0.0208, "step": 13777 }, { "epoch": 13.98, "learning_rate": 2.3789077657341508e-07, "loss": 0.0203, "step": 13778 }, { "epoch": 13.98, "learning_rate": 2.3741569908876882e-07, "loss": 0.0159, "step": 13779 }, { "epoch": 13.98, "learning_rate": 2.3694109075393602e-07, "loss": 0.0161, "step": 13780 }, { "epoch": 13.98, "learning_rate": 2.3646695159172507e-07, "loss": 0.0199, "step": 13781 }, { "epoch": 13.98, "learning_rate": 2.359932816249233e-07, "loss": 0.0188, "step": 13782 }, { "epoch": 13.99, "learning_rate": 2.3552008087628918e-07, "loss": 0.0189, "step": 13783 }, { "epoch": 13.99, "learning_rate": 2.3504734936856566e-07, "loss": 0.0158, "step": 13784 }, { "epoch": 13.99, "learning_rate": 2.3457508712446786e-07, "loss": 0.0218, "step": 13785 }, { "epoch": 13.99, "learning_rate": 2.341032941666943e-07, "loss": 0.0204, "step": 13786 }, { "epoch": 13.99, "learning_rate": 2.3363197051791576e-07, "loss": 0.0194, "step": 13787 }, { "epoch": 13.99, "learning_rate": 2.3316111620078186e-07, "loss": 0.0194, "step": 13788 }, { "epoch": 13.99, "learning_rate": 2.3269073123792007e-07, "loss": 0.0149, "step": 13789 }, { "epoch": 13.99, "learning_rate": 2.3222081565193454e-07, "loss": 0.0212, "step": 13790 }, { "epoch": 13.99, "learning_rate": 2.3175136946540832e-07, "loss": 0.0157, "step": 13791 }, { "epoch": 13.99, "learning_rate": 2.3128239270090114e-07, "loss": 0.0202, "step": 13792 }, { "epoch": 14.0, "learning_rate": 2.3081388538094939e-07, "loss": 0.0212, "step": 13793 }, { "epoch": 14.0, "learning_rate": 2.3034584752806844e-07, "loss": 0.0174, "step": 13794 }, { "epoch": 14.0, "learning_rate": 2.2987827916474915e-07, "loss": 0.0196, "step": 13795 }, { "epoch": 14.0, "learning_rate": 2.2941118031346244e-07, "loss": 0.0183, "step": 13796 }, { "epoch": 14.0, "learning_rate": 2.289445509966537e-07, "loss": 0.021, "step": 13797 }, { "epoch": 14.0, "learning_rate": 2.284783912367483e-07, "loss": 0.017, "step": 13798 }, { "epoch": 14.0, "learning_rate": 2.2801270105614838e-07, "loss": 0.0167, "step": 13799 }, { "epoch": 14.0, "learning_rate": 2.2754748047723042e-07, "loss": 0.019, "step": 13800 }, { "epoch": 14.0, "learning_rate": 2.2708272952235433e-07, "loss": 0.0174, "step": 13801 }, { "epoch": 14.01, "learning_rate": 2.2661844821385116e-07, "loss": 0.0195, "step": 13802 }, { "epoch": 14.01, "learning_rate": 2.2615463657403524e-07, "loss": 0.0177, "step": 13803 }, { "epoch": 14.01, "learning_rate": 2.2569129462519434e-07, "loss": 0.0169, "step": 13804 }, { "epoch": 14.01, "learning_rate": 2.2522842238959397e-07, "loss": 0.0191, "step": 13805 }, { "epoch": 14.01, "learning_rate": 2.2476601988947965e-07, "loss": 0.0144, "step": 13806 }, { "epoch": 14.01, "learning_rate": 2.2430408714706919e-07, "loss": 0.0178, "step": 13807 }, { "epoch": 14.01, "learning_rate": 2.2384262418456481e-07, "loss": 0.0175, "step": 13808 }, { "epoch": 14.01, "learning_rate": 2.2338163102414101e-07, "loss": 0.0162, "step": 13809 }, { "epoch": 14.01, "learning_rate": 2.2292110768795005e-07, "loss": 0.0188, "step": 13810 }, { "epoch": 14.01, "learning_rate": 2.2246105419812535e-07, "loss": 0.0136, "step": 13811 }, { "epoch": 14.02, "learning_rate": 2.2200147057677368e-07, "loss": 0.0158, "step": 13812 }, { "epoch": 14.02, "learning_rate": 2.2154235684598179e-07, "loss": 0.0185, "step": 13813 }, { "epoch": 14.02, "learning_rate": 2.2108371302780985e-07, "loss": 0.0168, "step": 13814 }, { "epoch": 14.02, "learning_rate": 2.206255391443013e-07, "loss": 0.0152, "step": 13815 }, { "epoch": 14.02, "learning_rate": 2.201678352174752e-07, "loss": 0.016, "step": 13816 }, { "epoch": 14.02, "learning_rate": 2.1971060126932287e-07, "loss": 0.0176, "step": 13817 }, { "epoch": 14.02, "learning_rate": 2.1925383732182003e-07, "loss": 0.0159, "step": 13818 }, { "epoch": 14.02, "learning_rate": 2.1879754339691584e-07, "loss": 0.0163, "step": 13819 }, { "epoch": 14.02, "learning_rate": 2.1834171951653826e-07, "loss": 0.016, "step": 13820 }, { "epoch": 14.02, "learning_rate": 2.1788636570259313e-07, "loss": 0.0203, "step": 13821 }, { "epoch": 14.03, "learning_rate": 2.1743148197696074e-07, "loss": 0.0136, "step": 13822 }, { "epoch": 14.03, "learning_rate": 2.1697706836150356e-07, "loss": 0.0171, "step": 13823 }, { "epoch": 14.03, "learning_rate": 2.165231248780564e-07, "loss": 0.0149, "step": 13824 }, { "epoch": 14.03, "learning_rate": 2.1606965154843507e-07, "loss": 0.0183, "step": 13825 }, { "epoch": 14.03, "learning_rate": 2.156166483944333e-07, "loss": 0.0171, "step": 13826 }, { "epoch": 14.03, "learning_rate": 2.151641154378159e-07, "loss": 0.0176, "step": 13827 }, { "epoch": 14.03, "learning_rate": 2.1471205270033547e-07, "loss": 0.0156, "step": 13828 }, { "epoch": 14.03, "learning_rate": 2.142604602037124e-07, "loss": 0.0166, "step": 13829 }, { "epoch": 14.03, "learning_rate": 2.1380933796964931e-07, "loss": 0.0166, "step": 13830 }, { "epoch": 14.03, "learning_rate": 2.1335868601982557e-07, "loss": 0.0172, "step": 13831 }, { "epoch": 14.04, "learning_rate": 2.1290850437589828e-07, "loss": 0.0176, "step": 13832 }, { "epoch": 14.04, "learning_rate": 2.1245879305950012e-07, "loss": 0.018, "step": 13833 }, { "epoch": 14.04, "learning_rate": 2.1200955209224273e-07, "loss": 0.0177, "step": 13834 }, { "epoch": 14.04, "learning_rate": 2.1156078149571547e-07, "loss": 0.0173, "step": 13835 }, { "epoch": 14.04, "learning_rate": 2.1111248129148332e-07, "loss": 0.0149, "step": 13836 }, { "epoch": 14.04, "learning_rate": 2.1066465150109018e-07, "loss": 0.0188, "step": 13837 }, { "epoch": 14.04, "learning_rate": 2.102172921460588e-07, "loss": 0.0178, "step": 13838 }, { "epoch": 14.04, "learning_rate": 2.0977040324788423e-07, "loss": 0.0194, "step": 13839 }, { "epoch": 14.04, "learning_rate": 2.0932398482804374e-07, "loss": 0.0154, "step": 13840 }, { "epoch": 14.04, "learning_rate": 2.0887803690799012e-07, "loss": 0.0179, "step": 13841 }, { "epoch": 14.05, "learning_rate": 2.0843255950915408e-07, "loss": 0.0149, "step": 13842 }, { "epoch": 14.05, "learning_rate": 2.0798755265294402e-07, "loss": 0.0153, "step": 13843 }, { "epoch": 14.05, "learning_rate": 2.075430163607428e-07, "loss": 0.0136, "step": 13844 }, { "epoch": 14.05, "learning_rate": 2.070989506539156e-07, "loss": 0.0186, "step": 13845 }, { "epoch": 14.05, "learning_rate": 2.06655355553802e-07, "loss": 0.0179, "step": 13846 }, { "epoch": 14.05, "learning_rate": 2.0621223108171717e-07, "loss": 0.0182, "step": 13847 }, { "epoch": 14.05, "learning_rate": 2.0576957725895962e-07, "loss": 0.0184, "step": 13848 }, { "epoch": 14.05, "learning_rate": 2.053273941067979e-07, "loss": 0.0189, "step": 13849 }, { "epoch": 14.05, "learning_rate": 2.0488568164648393e-07, "loss": 0.0172, "step": 13850 }, { "epoch": 14.05, "learning_rate": 2.044444398992418e-07, "loss": 0.0147, "step": 13851 }, { "epoch": 14.06, "learning_rate": 2.04003668886279e-07, "loss": 0.023, "step": 13852 }, { "epoch": 14.06, "learning_rate": 2.0356336862877523e-07, "loss": 0.0179, "step": 13853 }, { "epoch": 14.06, "learning_rate": 2.031235391478903e-07, "loss": 0.0168, "step": 13854 }, { "epoch": 14.06, "learning_rate": 2.0268418046476058e-07, "loss": 0.0182, "step": 13855 }, { "epoch": 14.06, "learning_rate": 2.0224529260049919e-07, "loss": 0.0176, "step": 13856 }, { "epoch": 14.06, "learning_rate": 2.0180687557619816e-07, "loss": 0.0157, "step": 13857 }, { "epoch": 14.06, "learning_rate": 2.0136892941292508e-07, "loss": 0.0172, "step": 13858 }, { "epoch": 14.06, "learning_rate": 2.0093145413172643e-07, "loss": 0.0174, "step": 13859 }, { "epoch": 14.06, "learning_rate": 2.0049444975362542e-07, "loss": 0.0166, "step": 13860 }, { "epoch": 14.06, "learning_rate": 2.0005791629962302e-07, "loss": 0.0168, "step": 13861 }, { "epoch": 14.07, "learning_rate": 1.9962185379069799e-07, "loss": 0.0146, "step": 13862 }, { "epoch": 14.07, "learning_rate": 1.9918626224780247e-07, "loss": 0.0169, "step": 13863 }, { "epoch": 14.07, "learning_rate": 1.9875114169187193e-07, "loss": 0.0185, "step": 13864 }, { "epoch": 14.07, "learning_rate": 1.9831649214381743e-07, "loss": 0.0205, "step": 13865 }, { "epoch": 14.07, "learning_rate": 1.9788231362452337e-07, "loss": 0.0185, "step": 13866 }, { "epoch": 14.07, "learning_rate": 1.974486061548575e-07, "loss": 0.0172, "step": 13867 }, { "epoch": 14.07, "learning_rate": 1.9701536975565983e-07, "loss": 0.0165, "step": 13868 }, { "epoch": 14.07, "learning_rate": 1.9658260444775035e-07, "loss": 0.0153, "step": 13869 }, { "epoch": 14.07, "learning_rate": 1.9615031025192576e-07, "loss": 0.0173, "step": 13870 }, { "epoch": 14.08, "learning_rate": 1.957184871889617e-07, "loss": 0.0181, "step": 13871 }, { "epoch": 14.08, "learning_rate": 1.952871352796104e-07, "loss": 0.017, "step": 13872 }, { "epoch": 14.08, "learning_rate": 1.9485625454459867e-07, "loss": 0.0161, "step": 13873 }, { "epoch": 14.08, "learning_rate": 1.9442584500463323e-07, "loss": 0.017, "step": 13874 }, { "epoch": 14.08, "learning_rate": 1.939959066803998e-07, "loss": 0.0162, "step": 13875 }, { "epoch": 14.08, "learning_rate": 1.9356643959255628e-07, "loss": 0.0156, "step": 13876 }, { "epoch": 14.08, "learning_rate": 1.931374437617439e-07, "loss": 0.0157, "step": 13877 }, { "epoch": 14.08, "learning_rate": 1.9270891920857626e-07, "loss": 0.0198, "step": 13878 }, { "epoch": 14.08, "learning_rate": 1.9228086595364902e-07, "loss": 0.019, "step": 13879 }, { "epoch": 14.08, "learning_rate": 1.9185328401753024e-07, "loss": 0.0183, "step": 13880 }, { "epoch": 14.09, "learning_rate": 1.9142617342076897e-07, "loss": 0.02, "step": 13881 }, { "epoch": 14.09, "learning_rate": 1.9099953418389104e-07, "loss": 0.0152, "step": 13882 }, { "epoch": 14.09, "learning_rate": 1.9057336632739787e-07, "loss": 0.0162, "step": 13883 }, { "epoch": 14.09, "learning_rate": 1.901476698717697e-07, "loss": 0.0182, "step": 13884 }, { "epoch": 14.09, "learning_rate": 1.8972244483746238e-07, "loss": 0.0175, "step": 13885 }, { "epoch": 14.09, "learning_rate": 1.8929769124491293e-07, "loss": 0.0179, "step": 13886 }, { "epoch": 14.09, "learning_rate": 1.8887340911453167e-07, "loss": 0.0152, "step": 13887 }, { "epoch": 14.09, "learning_rate": 1.8844959846670784e-07, "loss": 0.0181, "step": 13888 }, { "epoch": 14.09, "learning_rate": 1.8802625932180963e-07, "loss": 0.019, "step": 13889 }, { "epoch": 14.09, "learning_rate": 1.8760339170017848e-07, "loss": 0.0192, "step": 13890 }, { "epoch": 14.1, "learning_rate": 1.871809956221382e-07, "loss": 0.0147, "step": 13891 }, { "epoch": 14.1, "learning_rate": 1.8675907110798696e-07, "loss": 0.0174, "step": 13892 }, { "epoch": 14.1, "learning_rate": 1.8633761817799855e-07, "loss": 0.0155, "step": 13893 }, { "epoch": 14.1, "learning_rate": 1.859166368524279e-07, "loss": 0.0191, "step": 13894 }, { "epoch": 14.1, "learning_rate": 1.8549612715150544e-07, "loss": 0.0147, "step": 13895 }, { "epoch": 14.1, "learning_rate": 1.850760890954395e-07, "loss": 0.0168, "step": 13896 }, { "epoch": 14.1, "learning_rate": 1.8465652270441504e-07, "loss": 0.0169, "step": 13897 }, { "epoch": 14.1, "learning_rate": 1.8423742799859477e-07, "loss": 0.0143, "step": 13898 }, { "epoch": 14.1, "learning_rate": 1.8381880499811932e-07, "loss": 0.0168, "step": 13899 }, { "epoch": 14.1, "learning_rate": 1.8340065372310478e-07, "loss": 0.0162, "step": 13900 }, { "epoch": 14.11, "learning_rate": 1.829829741936462e-07, "loss": 0.0169, "step": 13901 }, { "epoch": 14.11, "learning_rate": 1.8256576642981527e-07, "loss": 0.0152, "step": 13902 }, { "epoch": 14.11, "learning_rate": 1.8214903045166267e-07, "loss": 0.0163, "step": 13903 }, { "epoch": 14.11, "learning_rate": 1.8173276627921343e-07, "loss": 0.0186, "step": 13904 }, { "epoch": 14.11, "learning_rate": 1.8131697393247273e-07, "loss": 0.0174, "step": 13905 }, { "epoch": 14.11, "learning_rate": 1.809016534314212e-07, "loss": 0.0141, "step": 13906 }, { "epoch": 14.11, "learning_rate": 1.8048680479601731e-07, "loss": 0.0186, "step": 13907 }, { "epoch": 14.11, "learning_rate": 1.800724280461963e-07, "loss": 0.0138, "step": 13908 }, { "epoch": 14.11, "learning_rate": 1.7965852320187328e-07, "loss": 0.0148, "step": 13909 }, { "epoch": 14.11, "learning_rate": 1.792450902829379e-07, "loss": 0.016, "step": 13910 }, { "epoch": 14.12, "learning_rate": 1.7883212930925653e-07, "loss": 0.0181, "step": 13911 }, { "epoch": 14.12, "learning_rate": 1.784196403006766e-07, "loss": 0.0143, "step": 13912 }, { "epoch": 14.12, "learning_rate": 1.7800762327702114e-07, "loss": 0.0181, "step": 13913 }, { "epoch": 14.12, "learning_rate": 1.7759607825808767e-07, "loss": 0.0167, "step": 13914 }, { "epoch": 14.12, "learning_rate": 1.7718500526365367e-07, "loss": 0.0167, "step": 13915 }, { "epoch": 14.12, "learning_rate": 1.767744043134756e-07, "loss": 0.0128, "step": 13916 }, { "epoch": 14.12, "learning_rate": 1.7636427542728428e-07, "loss": 0.0156, "step": 13917 }, { "epoch": 14.12, "learning_rate": 1.7595461862478736e-07, "loss": 0.0179, "step": 13918 }, { "epoch": 14.12, "learning_rate": 1.7554543392567237e-07, "loss": 0.0178, "step": 13919 }, { "epoch": 14.12, "learning_rate": 1.7513672134960357e-07, "loss": 0.0168, "step": 13920 }, { "epoch": 14.13, "learning_rate": 1.7472848091622307e-07, "loss": 0.0183, "step": 13921 }, { "epoch": 14.13, "learning_rate": 1.7432071264514627e-07, "loss": 0.0184, "step": 13922 }, { "epoch": 14.13, "learning_rate": 1.7391341655597083e-07, "loss": 0.0178, "step": 13923 }, { "epoch": 14.13, "learning_rate": 1.7350659266826886e-07, "loss": 0.0189, "step": 13924 }, { "epoch": 14.13, "learning_rate": 1.731002410015914e-07, "loss": 0.0206, "step": 13925 }, { "epoch": 14.13, "learning_rate": 1.7269436157546615e-07, "loss": 0.018, "step": 13926 }, { "epoch": 14.13, "learning_rate": 1.722889544093953e-07, "loss": 0.014, "step": 13927 }, { "epoch": 14.13, "learning_rate": 1.718840195228655e-07, "loss": 0.0174, "step": 13928 }, { "epoch": 14.13, "learning_rate": 1.7147955693533335e-07, "loss": 0.0177, "step": 13929 }, { "epoch": 14.13, "learning_rate": 1.710755666662367e-07, "loss": 0.0178, "step": 13930 }, { "epoch": 14.14, "learning_rate": 1.7067204873498888e-07, "loss": 0.0162, "step": 13931 }, { "epoch": 14.14, "learning_rate": 1.7026900316098217e-07, "loss": 0.0185, "step": 13932 }, { "epoch": 14.14, "learning_rate": 1.698664299635855e-07, "loss": 0.0176, "step": 13933 }, { "epoch": 14.14, "learning_rate": 1.694643291621434e-07, "loss": 0.0168, "step": 13934 }, { "epoch": 14.14, "learning_rate": 1.6906270077598042e-07, "loss": 0.0164, "step": 13935 }, { "epoch": 14.14, "learning_rate": 1.6866154482439556e-07, "loss": 0.0179, "step": 13936 }, { "epoch": 14.14, "learning_rate": 1.6826086132666898e-07, "loss": 0.0188, "step": 13937 }, { "epoch": 14.14, "learning_rate": 1.6786065030205523e-07, "loss": 0.0167, "step": 13938 }, { "epoch": 14.14, "learning_rate": 1.6746091176978673e-07, "loss": 0.019, "step": 13939 }, { "epoch": 14.15, "learning_rate": 1.6706164574907258e-07, "loss": 0.017, "step": 13940 }, { "epoch": 14.15, "learning_rate": 1.6666285225910072e-07, "loss": 0.0194, "step": 13941 }, { "epoch": 14.15, "learning_rate": 1.6626453131903475e-07, "loss": 0.0162, "step": 13942 }, { "epoch": 14.15, "learning_rate": 1.658666829480171e-07, "loss": 0.0163, "step": 13943 }, { "epoch": 14.15, "learning_rate": 1.6546930716516474e-07, "loss": 0.0207, "step": 13944 }, { "epoch": 14.15, "learning_rate": 1.650724039895768e-07, "loss": 0.0142, "step": 13945 }, { "epoch": 14.15, "learning_rate": 1.646759734403258e-07, "loss": 0.0166, "step": 13946 }, { "epoch": 14.15, "learning_rate": 1.642800155364621e-07, "loss": 0.0137, "step": 13947 }, { "epoch": 14.15, "learning_rate": 1.6388453029701489e-07, "loss": 0.0159, "step": 13948 }, { "epoch": 14.15, "learning_rate": 1.6348951774098676e-07, "loss": 0.0169, "step": 13949 }, { "epoch": 14.16, "learning_rate": 1.6309497788736473e-07, "loss": 0.0166, "step": 13950 }, { "epoch": 14.16, "learning_rate": 1.6270091075510474e-07, "loss": 0.0166, "step": 13951 }, { "epoch": 14.16, "learning_rate": 1.623073163631461e-07, "loss": 0.0209, "step": 13952 }, { "epoch": 14.16, "learning_rate": 1.619141947304026e-07, "loss": 0.0202, "step": 13953 }, { "epoch": 14.16, "learning_rate": 1.6152154587576685e-07, "loss": 0.017, "step": 13954 }, { "epoch": 14.16, "learning_rate": 1.6112936981810712e-07, "loss": 0.0186, "step": 13955 }, { "epoch": 14.16, "learning_rate": 1.6073766657627055e-07, "loss": 0.0168, "step": 13956 }, { "epoch": 14.16, "learning_rate": 1.6034643616908098e-07, "loss": 0.02, "step": 13957 }, { "epoch": 14.16, "learning_rate": 1.5995567861533778e-07, "loss": 0.0166, "step": 13958 }, { "epoch": 14.16, "learning_rate": 1.5956539393382043e-07, "loss": 0.0143, "step": 13959 }, { "epoch": 14.17, "learning_rate": 1.5917558214328498e-07, "loss": 0.0149, "step": 13960 }, { "epoch": 14.17, "learning_rate": 1.5878624326246095e-07, "loss": 0.0147, "step": 13961 }, { "epoch": 14.17, "learning_rate": 1.5839737731006332e-07, "loss": 0.0173, "step": 13962 }, { "epoch": 14.17, "learning_rate": 1.5800898430477608e-07, "loss": 0.0182, "step": 13963 }, { "epoch": 14.17, "learning_rate": 1.5762106426526425e-07, "loss": 0.0169, "step": 13964 }, { "epoch": 14.17, "learning_rate": 1.5723361721017072e-07, "loss": 0.0177, "step": 13965 }, { "epoch": 14.17, "learning_rate": 1.568466431581128e-07, "loss": 0.0157, "step": 13966 }, { "epoch": 14.17, "learning_rate": 1.56460142127689e-07, "loss": 0.0187, "step": 13967 }, { "epoch": 14.17, "learning_rate": 1.5607411413747108e-07, "loss": 0.0192, "step": 13968 }, { "epoch": 14.17, "learning_rate": 1.5568855920601088e-07, "loss": 0.0183, "step": 13969 }, { "epoch": 14.18, "learning_rate": 1.5530347735183693e-07, "loss": 0.0148, "step": 13970 }, { "epoch": 14.18, "learning_rate": 1.549188685934533e-07, "loss": 0.0178, "step": 13971 }, { "epoch": 14.18, "learning_rate": 1.545347329493452e-07, "loss": 0.0163, "step": 13972 }, { "epoch": 14.18, "learning_rate": 1.5415107043797006e-07, "loss": 0.0149, "step": 13973 }, { "epoch": 14.18, "learning_rate": 1.5376788107776651e-07, "loss": 0.018, "step": 13974 }, { "epoch": 14.18, "learning_rate": 1.5338516488714984e-07, "loss": 0.0148, "step": 13975 }, { "epoch": 14.18, "learning_rate": 1.5300292188450972e-07, "loss": 0.0171, "step": 13976 }, { "epoch": 14.18, "learning_rate": 1.5262115208821594e-07, "loss": 0.0192, "step": 13977 }, { "epoch": 14.18, "learning_rate": 1.5223985551661492e-07, "loss": 0.0176, "step": 13978 }, { "epoch": 14.18, "learning_rate": 1.5185903218803088e-07, "loss": 0.017, "step": 13979 }, { "epoch": 14.19, "learning_rate": 1.5147868212076367e-07, "loss": 0.0192, "step": 13980 }, { "epoch": 14.19, "learning_rate": 1.5109880533309195e-07, "loss": 0.0186, "step": 13981 }, { "epoch": 14.19, "learning_rate": 1.5071940184327115e-07, "loss": 0.0149, "step": 13982 }, { "epoch": 14.19, "learning_rate": 1.5034047166953335e-07, "loss": 0.0154, "step": 13983 }, { "epoch": 14.19, "learning_rate": 1.4996201483008842e-07, "loss": 0.0178, "step": 13984 }, { "epoch": 14.19, "learning_rate": 1.4958403134312405e-07, "loss": 0.0155, "step": 13985 }, { "epoch": 14.19, "learning_rate": 1.4920652122680345e-07, "loss": 0.0194, "step": 13986 }, { "epoch": 14.19, "learning_rate": 1.4882948449926882e-07, "loss": 0.0181, "step": 13987 }, { "epoch": 14.19, "learning_rate": 1.4845292117863897e-07, "loss": 0.0165, "step": 13988 }, { "epoch": 14.19, "learning_rate": 1.4807683128301163e-07, "loss": 0.0152, "step": 13989 }, { "epoch": 14.2, "learning_rate": 1.4770121483045685e-07, "loss": 0.0178, "step": 13990 }, { "epoch": 14.2, "learning_rate": 1.4732607183902793e-07, "loss": 0.0175, "step": 13991 }, { "epoch": 14.2, "learning_rate": 1.4695140232675154e-07, "loss": 0.0161, "step": 13992 }, { "epoch": 14.2, "learning_rate": 1.465772063116333e-07, "loss": 0.0169, "step": 13993 }, { "epoch": 14.2, "learning_rate": 1.462034838116544e-07, "loss": 0.0164, "step": 13994 }, { "epoch": 14.2, "learning_rate": 1.4583023484477598e-07, "loss": 0.0167, "step": 13995 }, { "epoch": 14.2, "learning_rate": 1.4545745942893375e-07, "loss": 0.0199, "step": 13996 }, { "epoch": 14.2, "learning_rate": 1.4508515758204223e-07, "loss": 0.0162, "step": 13997 }, { "epoch": 14.2, "learning_rate": 1.4471332932199268e-07, "loss": 0.0178, "step": 13998 }, { "epoch": 14.2, "learning_rate": 1.4434197466665301e-07, "loss": 0.0203, "step": 13999 }, { "epoch": 14.21, "learning_rate": 1.4397109363387007e-07, "loss": 0.0152, "step": 14000 }, { "epoch": 14.21, "learning_rate": 1.436006862414674e-07, "loss": 0.0161, "step": 14001 }, { "epoch": 14.21, "learning_rate": 1.4323075250724295e-07, "loss": 0.0171, "step": 14002 }, { "epoch": 14.21, "learning_rate": 1.4286129244897474e-07, "loss": 0.0171, "step": 14003 }, { "epoch": 14.21, "learning_rate": 1.4249230608441967e-07, "loss": 0.0163, "step": 14004 }, { "epoch": 14.21, "learning_rate": 1.42123793431308e-07, "loss": 0.0167, "step": 14005 }, { "epoch": 14.21, "learning_rate": 1.417557545073489e-07, "loss": 0.0157, "step": 14006 }, { "epoch": 14.21, "learning_rate": 1.4138818933022937e-07, "loss": 0.0161, "step": 14007 }, { "epoch": 14.21, "learning_rate": 1.4102109791761298e-07, "loss": 0.0181, "step": 14008 }, { "epoch": 14.22, "learning_rate": 1.4065448028714124e-07, "loss": 0.0194, "step": 14009 }, { "epoch": 14.22, "learning_rate": 1.4028833645643113e-07, "loss": 0.0176, "step": 14010 }, { "epoch": 14.22, "learning_rate": 1.3992266644307752e-07, "loss": 0.0176, "step": 14011 }, { "epoch": 14.22, "learning_rate": 1.3955747026465404e-07, "loss": 0.0153, "step": 14012 }, { "epoch": 14.22, "learning_rate": 1.391927479387123e-07, "loss": 0.0173, "step": 14013 }, { "epoch": 14.22, "learning_rate": 1.38828499482776e-07, "loss": 0.0181, "step": 14014 }, { "epoch": 14.22, "learning_rate": 1.3846472491435115e-07, "loss": 0.0152, "step": 14015 }, { "epoch": 14.22, "learning_rate": 1.3810142425092044e-07, "loss": 0.0163, "step": 14016 }, { "epoch": 14.22, "learning_rate": 1.377385975099399e-07, "loss": 0.017, "step": 14017 }, { "epoch": 14.22, "learning_rate": 1.373762447088467e-07, "loss": 0.0167, "step": 14018 }, { "epoch": 14.23, "learning_rate": 1.3701436586505467e-07, "loss": 0.015, "step": 14019 }, { "epoch": 14.23, "learning_rate": 1.3665296099595326e-07, "loss": 0.0164, "step": 14020 }, { "epoch": 14.23, "learning_rate": 1.362920301189108e-07, "loss": 0.0177, "step": 14021 }, { "epoch": 14.23, "learning_rate": 1.3593157325127226e-07, "loss": 0.0169, "step": 14022 }, { "epoch": 14.23, "learning_rate": 1.355715904103594e-07, "loss": 0.0157, "step": 14023 }, { "epoch": 14.23, "learning_rate": 1.3521208161347167e-07, "loss": 0.016, "step": 14024 }, { "epoch": 14.23, "learning_rate": 1.3485304687788525e-07, "loss": 0.0168, "step": 14025 }, { "epoch": 14.23, "learning_rate": 1.3449448622085416e-07, "loss": 0.0178, "step": 14026 }, { "epoch": 14.23, "learning_rate": 1.34136399659609e-07, "loss": 0.0171, "step": 14027 }, { "epoch": 14.23, "learning_rate": 1.3377878721135718e-07, "loss": 0.0195, "step": 14028 }, { "epoch": 14.24, "learning_rate": 1.33421648893286e-07, "loss": 0.0196, "step": 14029 }, { "epoch": 14.24, "learning_rate": 1.3306498472255847e-07, "loss": 0.0181, "step": 14030 }, { "epoch": 14.24, "learning_rate": 1.3270879471631194e-07, "loss": 0.018, "step": 14031 }, { "epoch": 14.24, "learning_rate": 1.3235307889166381e-07, "loss": 0.0167, "step": 14032 }, { "epoch": 14.24, "learning_rate": 1.3199783726571048e-07, "loss": 0.0186, "step": 14033 }, { "epoch": 14.24, "learning_rate": 1.3164306985552156e-07, "loss": 0.0157, "step": 14034 }, { "epoch": 14.24, "learning_rate": 1.3128877667814566e-07, "loss": 0.0191, "step": 14035 }, { "epoch": 14.24, "learning_rate": 1.3093495775061027e-07, "loss": 0.0197, "step": 14036 }, { "epoch": 14.24, "learning_rate": 1.3058161308991625e-07, "loss": 0.0156, "step": 14037 }, { "epoch": 14.24, "learning_rate": 1.3022874271304664e-07, "loss": 0.0144, "step": 14038 }, { "epoch": 14.25, "learning_rate": 1.298763466369557e-07, "loss": 0.018, "step": 14039 }, { "epoch": 14.25, "learning_rate": 1.2952442487858097e-07, "loss": 0.0138, "step": 14040 }, { "epoch": 14.25, "learning_rate": 1.2917297745483338e-07, "loss": 0.0176, "step": 14041 }, { "epoch": 14.25, "learning_rate": 1.288220043826005e-07, "loss": 0.02, "step": 14042 }, { "epoch": 14.25, "learning_rate": 1.2847150567875222e-07, "loss": 0.0175, "step": 14043 }, { "epoch": 14.25, "learning_rate": 1.2812148136012726e-07, "loss": 0.0186, "step": 14044 }, { "epoch": 14.25, "learning_rate": 1.2777193144355105e-07, "loss": 0.0183, "step": 14045 }, { "epoch": 14.25, "learning_rate": 1.2742285594581795e-07, "loss": 0.0168, "step": 14046 }, { "epoch": 14.25, "learning_rate": 1.2707425488370562e-07, "loss": 0.0174, "step": 14047 }, { "epoch": 14.25, "learning_rate": 1.2672612827396514e-07, "loss": 0.016, "step": 14048 }, { "epoch": 14.26, "learning_rate": 1.2637847613332643e-07, "loss": 0.0175, "step": 14049 }, { "epoch": 14.26, "learning_rate": 1.2603129847849727e-07, "loss": 0.0185, "step": 14050 }, { "epoch": 14.26, "learning_rate": 1.2568459532615874e-07, "loss": 0.0171, "step": 14051 }, { "epoch": 14.26, "learning_rate": 1.2533836669297416e-07, "loss": 0.0162, "step": 14052 }, { "epoch": 14.26, "learning_rate": 1.2499261259558138e-07, "loss": 0.0178, "step": 14053 }, { "epoch": 14.26, "learning_rate": 1.2464733305059485e-07, "loss": 0.0173, "step": 14054 }, { "epoch": 14.26, "learning_rate": 1.243025280746102e-07, "loss": 0.0181, "step": 14055 }, { "epoch": 14.26, "learning_rate": 1.2395819768419526e-07, "loss": 0.0183, "step": 14056 }, { "epoch": 14.26, "learning_rate": 1.2361434189589683e-07, "loss": 0.0165, "step": 14057 }, { "epoch": 14.26, "learning_rate": 1.2327096072623946e-07, "loss": 0.0161, "step": 14058 }, { "epoch": 14.27, "learning_rate": 1.229280541917255e-07, "loss": 0.0184, "step": 14059 }, { "epoch": 14.27, "learning_rate": 1.2258562230883286e-07, "loss": 0.0179, "step": 14060 }, { "epoch": 14.27, "learning_rate": 1.2224366509401732e-07, "loss": 0.0173, "step": 14061 }, { "epoch": 14.27, "learning_rate": 1.2190218256371234e-07, "loss": 0.0184, "step": 14062 }, { "epoch": 14.27, "learning_rate": 1.2156117473432815e-07, "loss": 0.0164, "step": 14063 }, { "epoch": 14.27, "learning_rate": 1.212206416222528e-07, "loss": 0.0187, "step": 14064 }, { "epoch": 14.27, "learning_rate": 1.2088058324384978e-07, "loss": 0.0174, "step": 14065 }, { "epoch": 14.27, "learning_rate": 1.205409996154605e-07, "loss": 0.0179, "step": 14066 }, { "epoch": 14.27, "learning_rate": 1.2020189075340638e-07, "loss": 0.0198, "step": 14067 }, { "epoch": 14.27, "learning_rate": 1.1986325667398101e-07, "loss": 0.0208, "step": 14068 }, { "epoch": 14.28, "learning_rate": 1.1952509739345918e-07, "loss": 0.0193, "step": 14069 }, { "epoch": 14.28, "learning_rate": 1.1918741292809121e-07, "loss": 0.0187, "step": 14070 }, { "epoch": 14.28, "learning_rate": 1.1885020329410412e-07, "loss": 0.0155, "step": 14071 }, { "epoch": 14.28, "learning_rate": 1.1851346850770385e-07, "loss": 0.0142, "step": 14072 }, { "epoch": 14.28, "learning_rate": 1.1817720858507186e-07, "loss": 0.0183, "step": 14073 }, { "epoch": 14.28, "learning_rate": 1.1784142354236749e-07, "loss": 0.0189, "step": 14074 }, { "epoch": 14.28, "learning_rate": 1.1750611339572782e-07, "loss": 0.0179, "step": 14075 }, { "epoch": 14.28, "learning_rate": 1.1717127816126439e-07, "loss": 0.0191, "step": 14076 }, { "epoch": 14.28, "learning_rate": 1.1683691785507212e-07, "loss": 0.0156, "step": 14077 }, { "epoch": 14.29, "learning_rate": 1.1650303249321371e-07, "loss": 0.0183, "step": 14078 }, { "epoch": 14.29, "learning_rate": 1.1616962209173854e-07, "loss": 0.0189, "step": 14079 }, { "epoch": 14.29, "learning_rate": 1.1583668666666715e-07, "loss": 0.0159, "step": 14080 }, { "epoch": 14.29, "learning_rate": 1.1550422623399893e-07, "loss": 0.0171, "step": 14081 }, { "epoch": 14.29, "learning_rate": 1.1517224080971223e-07, "loss": 0.017, "step": 14082 }, { "epoch": 14.29, "learning_rate": 1.1484073040975874e-07, "loss": 0.0169, "step": 14083 }, { "epoch": 14.29, "learning_rate": 1.1450969505007126e-07, "loss": 0.0141, "step": 14084 }, { "epoch": 14.29, "learning_rate": 1.1417913474655707e-07, "loss": 0.02, "step": 14085 }, { "epoch": 14.29, "learning_rate": 1.1384904951510123e-07, "loss": 0.0174, "step": 14086 }, { "epoch": 14.29, "learning_rate": 1.1351943937156662e-07, "loss": 0.0195, "step": 14087 }, { "epoch": 14.3, "learning_rate": 1.1319030433179279e-07, "loss": 0.0186, "step": 14088 }, { "epoch": 14.3, "learning_rate": 1.1286164441159708e-07, "loss": 0.0174, "step": 14089 }, { "epoch": 14.3, "learning_rate": 1.1253345962677353e-07, "loss": 0.016, "step": 14090 }, { "epoch": 14.3, "learning_rate": 1.1220574999309286e-07, "loss": 0.0171, "step": 14091 }, { "epoch": 14.3, "learning_rate": 1.1187851552630468e-07, "loss": 0.017, "step": 14092 }, { "epoch": 14.3, "learning_rate": 1.115517562421331e-07, "loss": 0.0186, "step": 14093 }, { "epoch": 14.3, "learning_rate": 1.1122547215627998e-07, "loss": 0.0181, "step": 14094 }, { "epoch": 14.3, "learning_rate": 1.1089966328442836e-07, "loss": 0.0147, "step": 14095 }, { "epoch": 14.3, "learning_rate": 1.1057432964223347e-07, "loss": 0.0192, "step": 14096 }, { "epoch": 14.3, "learning_rate": 1.1024947124532836e-07, "loss": 0.0201, "step": 14097 }, { "epoch": 14.31, "learning_rate": 1.0992508810932612e-07, "loss": 0.0212, "step": 14098 }, { "epoch": 14.31, "learning_rate": 1.0960118024981426e-07, "loss": 0.0147, "step": 14099 }, { "epoch": 14.31, "learning_rate": 1.0927774768235921e-07, "loss": 0.0202, "step": 14100 }, { "epoch": 14.31, "learning_rate": 1.0895479042250411e-07, "loss": 0.018, "step": 14101 }, { "epoch": 14.31, "learning_rate": 1.0863230848576767e-07, "loss": 0.019, "step": 14102 }, { "epoch": 14.31, "learning_rate": 1.0831030188764858e-07, "loss": 0.0156, "step": 14103 }, { "epoch": 14.31, "learning_rate": 1.079887706436189e-07, "loss": 0.0191, "step": 14104 }, { "epoch": 14.31, "learning_rate": 1.0766771476913185e-07, "loss": 0.0185, "step": 14105 }, { "epoch": 14.31, "learning_rate": 1.0734713427961729e-07, "loss": 0.0196, "step": 14106 }, { "epoch": 14.31, "learning_rate": 1.0702702919047847e-07, "loss": 0.017, "step": 14107 }, { "epoch": 14.32, "learning_rate": 1.0670739951709863e-07, "loss": 0.0172, "step": 14108 }, { "epoch": 14.32, "learning_rate": 1.0638824527483993e-07, "loss": 0.0183, "step": 14109 }, { "epoch": 14.32, "learning_rate": 1.0606956647903677e-07, "loss": 0.0195, "step": 14110 }, { "epoch": 14.32, "learning_rate": 1.0575136314500578e-07, "loss": 0.0181, "step": 14111 }, { "epoch": 14.32, "learning_rate": 1.0543363528803696e-07, "loss": 0.0169, "step": 14112 }, { "epoch": 14.32, "learning_rate": 1.0511638292340143e-07, "loss": 0.018, "step": 14113 }, { "epoch": 14.32, "learning_rate": 1.0479960606634254e-07, "loss": 0.0201, "step": 14114 }, { "epoch": 14.32, "learning_rate": 1.0448330473208368e-07, "loss": 0.0174, "step": 14115 }, { "epoch": 14.32, "learning_rate": 1.0416747893582602e-07, "loss": 0.0197, "step": 14116 }, { "epoch": 14.32, "learning_rate": 1.038521286927463e-07, "loss": 0.0141, "step": 14117 }, { "epoch": 14.33, "learning_rate": 1.0353725401799908e-07, "loss": 0.0188, "step": 14118 }, { "epoch": 14.33, "learning_rate": 1.0322285492671557e-07, "loss": 0.0197, "step": 14119 }, { "epoch": 14.33, "learning_rate": 1.0290893143400371e-07, "loss": 0.0153, "step": 14120 }, { "epoch": 14.33, "learning_rate": 1.0259548355495253e-07, "loss": 0.0133, "step": 14121 }, { "epoch": 14.33, "learning_rate": 1.0228251130462108e-07, "loss": 0.0124, "step": 14122 }, { "epoch": 14.33, "learning_rate": 1.019700146980529e-07, "loss": 0.0149, "step": 14123 }, { "epoch": 14.33, "learning_rate": 1.0165799375026265e-07, "loss": 0.0153, "step": 14124 }, { "epoch": 14.33, "learning_rate": 1.0134644847624608e-07, "loss": 0.0144, "step": 14125 }, { "epoch": 14.33, "learning_rate": 1.010353788909757e-07, "loss": 0.0149, "step": 14126 }, { "epoch": 14.33, "learning_rate": 1.0072478500939842e-07, "loss": 0.0139, "step": 14127 }, { "epoch": 14.34, "learning_rate": 1.0041466684644007e-07, "loss": 0.0179, "step": 14128 }, { "epoch": 14.34, "learning_rate": 1.0010502441700543e-07, "loss": 0.0161, "step": 14129 }, { "epoch": 14.34, "learning_rate": 9.979585773597367e-08, "loss": 0.0139, "step": 14130 }, { "epoch": 14.34, "learning_rate": 9.948716681820181e-08, "loss": 0.0147, "step": 14131 }, { "epoch": 14.34, "learning_rate": 9.917895167852465e-08, "loss": 0.0169, "step": 14132 }, { "epoch": 14.34, "learning_rate": 9.887121233175368e-08, "loss": 0.0176, "step": 14133 }, { "epoch": 14.34, "learning_rate": 9.856394879267706e-08, "loss": 0.0194, "step": 14134 }, { "epoch": 14.34, "learning_rate": 9.825716107606076e-08, "loss": 0.0169, "step": 14135 }, { "epoch": 14.34, "learning_rate": 9.795084919664854e-08, "loss": 0.0162, "step": 14136 }, { "epoch": 14.35, "learning_rate": 9.764501316915975e-08, "loss": 0.0137, "step": 14137 }, { "epoch": 14.35, "learning_rate": 9.733965300829373e-08, "loss": 0.0168, "step": 14138 }, { "epoch": 14.35, "learning_rate": 9.703476872872097e-08, "loss": 0.0153, "step": 14139 }, { "epoch": 14.35, "learning_rate": 9.673036034509642e-08, "loss": 0.0144, "step": 14140 }, { "epoch": 14.35, "learning_rate": 9.642642787204614e-08, "loss": 0.0157, "step": 14141 }, { "epoch": 14.35, "learning_rate": 9.612297132417624e-08, "loss": 0.0176, "step": 14142 }, { "epoch": 14.35, "learning_rate": 9.581999071607174e-08, "loss": 0.0154, "step": 14143 }, { "epoch": 14.35, "learning_rate": 9.551748606229095e-08, "loss": 0.0172, "step": 14144 }, { "epoch": 14.35, "learning_rate": 9.521545737737115e-08, "loss": 0.014, "step": 14145 }, { "epoch": 14.35, "learning_rate": 9.49139046758274e-08, "loss": 0.0157, "step": 14146 }, { "epoch": 14.36, "learning_rate": 9.461282797214921e-08, "loss": 0.0161, "step": 14147 }, { "epoch": 14.36, "learning_rate": 9.431222728080836e-08, "loss": 0.0165, "step": 14148 }, { "epoch": 14.36, "learning_rate": 9.40121026162466e-08, "loss": 0.0162, "step": 14149 }, { "epoch": 14.36, "learning_rate": 9.37124539928902e-08, "loss": 0.0197, "step": 14150 }, { "epoch": 14.36, "learning_rate": 9.34132814251354e-08, "loss": 0.0189, "step": 14151 }, { "epoch": 14.36, "learning_rate": 9.311458492736291e-08, "loss": 0.0186, "step": 14152 }, { "epoch": 14.36, "learning_rate": 9.28163645139235e-08, "loss": 0.0193, "step": 14153 }, { "epoch": 14.36, "learning_rate": 9.251862019915126e-08, "loss": 0.0172, "step": 14154 }, { "epoch": 14.36, "learning_rate": 9.22213519973536e-08, "loss": 0.0187, "step": 14155 }, { "epoch": 14.36, "learning_rate": 9.192455992281579e-08, "loss": 0.0165, "step": 14156 }, { "epoch": 14.37, "learning_rate": 9.162824398979974e-08, "loss": 0.0185, "step": 14157 }, { "epoch": 14.37, "learning_rate": 9.13324042125463e-08, "loss": 0.0172, "step": 14158 }, { "epoch": 14.37, "learning_rate": 9.103704060527186e-08, "loss": 0.0174, "step": 14159 }, { "epoch": 14.37, "learning_rate": 9.074215318217173e-08, "loss": 0.0168, "step": 14160 }, { "epoch": 14.37, "learning_rate": 9.044774195741345e-08, "loss": 0.0154, "step": 14161 }, { "epoch": 14.37, "learning_rate": 9.015380694514909e-08, "loss": 0.0167, "step": 14162 }, { "epoch": 14.37, "learning_rate": 8.986034815950173e-08, "loss": 0.0119, "step": 14163 }, { "epoch": 14.37, "learning_rate": 8.956736561457346e-08, "loss": 0.0174, "step": 14164 }, { "epoch": 14.37, "learning_rate": 8.927485932444635e-08, "loss": 0.0156, "step": 14165 }, { "epoch": 14.37, "learning_rate": 8.89828293031747e-08, "loss": 0.0171, "step": 14166 }, { "epoch": 14.38, "learning_rate": 8.869127556479396e-08, "loss": 0.0163, "step": 14167 }, { "epoch": 14.38, "learning_rate": 8.840019812331291e-08, "loss": 0.0162, "step": 14168 }, { "epoch": 14.38, "learning_rate": 8.810959699272148e-08, "loss": 0.0205, "step": 14169 }, { "epoch": 14.38, "learning_rate": 8.781947218698295e-08, "loss": 0.0164, "step": 14170 }, { "epoch": 14.38, "learning_rate": 8.75298237200417e-08, "loss": 0.0187, "step": 14171 }, { "epoch": 14.38, "learning_rate": 8.724065160581662e-08, "loss": 0.0155, "step": 14172 }, { "epoch": 14.38, "learning_rate": 8.695195585820215e-08, "loss": 0.0128, "step": 14173 }, { "epoch": 14.38, "learning_rate": 8.666373649107385e-08, "loss": 0.014, "step": 14174 }, { "epoch": 14.38, "learning_rate": 8.637599351828063e-08, "loss": 0.0197, "step": 14175 }, { "epoch": 14.38, "learning_rate": 8.608872695365255e-08, "loss": 0.0197, "step": 14176 }, { "epoch": 14.39, "learning_rate": 8.580193681099303e-08, "loss": 0.0187, "step": 14177 }, { "epoch": 14.39, "learning_rate": 8.551562310408213e-08, "loss": 0.0183, "step": 14178 }, { "epoch": 14.39, "learning_rate": 8.522978584668329e-08, "loss": 0.016, "step": 14179 }, { "epoch": 14.39, "learning_rate": 8.494442505252998e-08, "loss": 0.0152, "step": 14180 }, { "epoch": 14.39, "learning_rate": 8.465954073533456e-08, "loss": 0.0175, "step": 14181 }, { "epoch": 14.39, "learning_rate": 8.437513290879052e-08, "loss": 0.0185, "step": 14182 }, { "epoch": 14.39, "learning_rate": 8.409120158656248e-08, "loss": 0.0173, "step": 14183 }, { "epoch": 14.39, "learning_rate": 8.38077467822962e-08, "loss": 0.0161, "step": 14184 }, { "epoch": 14.39, "learning_rate": 8.3524768509613e-08, "loss": 0.0198, "step": 14185 }, { "epoch": 14.39, "learning_rate": 8.3242266782112e-08, "loss": 0.0191, "step": 14186 }, { "epoch": 14.4, "learning_rate": 8.296024161336902e-08, "loss": 0.0164, "step": 14187 }, { "epoch": 14.4, "learning_rate": 8.267869301693654e-08, "loss": 0.0185, "step": 14188 }, { "epoch": 14.4, "learning_rate": 8.239762100634597e-08, "loss": 0.0172, "step": 14189 }, { "epoch": 14.4, "learning_rate": 8.211702559510315e-08, "loss": 0.0172, "step": 14190 }, { "epoch": 14.4, "learning_rate": 8.183690679669176e-08, "loss": 0.0172, "step": 14191 }, { "epoch": 14.4, "learning_rate": 8.155726462457547e-08, "loss": 0.015, "step": 14192 }, { "epoch": 14.4, "learning_rate": 8.127809909219131e-08, "loss": 0.0128, "step": 14193 }, { "epoch": 14.4, "learning_rate": 8.09994102129541e-08, "loss": 0.0189, "step": 14194 }, { "epoch": 14.4, "learning_rate": 8.072119800025757e-08, "loss": 0.018, "step": 14195 }, { "epoch": 14.4, "learning_rate": 8.044346246747103e-08, "loss": 0.015, "step": 14196 }, { "epoch": 14.41, "learning_rate": 8.016620362794159e-08, "loss": 0.0171, "step": 14197 }, { "epoch": 14.41, "learning_rate": 7.988942149499302e-08, "loss": 0.0199, "step": 14198 }, { "epoch": 14.41, "learning_rate": 7.961311608192801e-08, "loss": 0.0162, "step": 14199 }, { "epoch": 14.41, "learning_rate": 7.933728740202152e-08, "loss": 0.0185, "step": 14200 }, { "epoch": 14.41, "learning_rate": 7.906193546853069e-08, "loss": 0.017, "step": 14201 }, { "epoch": 14.41, "learning_rate": 7.878706029468608e-08, "loss": 0.0176, "step": 14202 }, { "epoch": 14.41, "learning_rate": 7.851266189369933e-08, "loss": 0.0177, "step": 14203 }, { "epoch": 14.41, "learning_rate": 7.823874027875656e-08, "loss": 0.0184, "step": 14204 }, { "epoch": 14.41, "learning_rate": 7.796529546301945e-08, "loss": 0.0171, "step": 14205 }, { "epoch": 14.42, "learning_rate": 7.769232745963085e-08, "loss": 0.0182, "step": 14206 }, { "epoch": 14.42, "learning_rate": 7.741983628170802e-08, "loss": 0.0178, "step": 14207 }, { "epoch": 14.42, "learning_rate": 7.714782194234383e-08, "loss": 0.0167, "step": 14208 }, { "epoch": 14.42, "learning_rate": 7.687628445461337e-08, "loss": 0.0135, "step": 14209 }, { "epoch": 14.42, "learning_rate": 7.660522383156178e-08, "loss": 0.0166, "step": 14210 }, { "epoch": 14.42, "learning_rate": 7.63346400862186e-08, "loss": 0.0198, "step": 14211 }, { "epoch": 14.42, "learning_rate": 7.606453323158569e-08, "loss": 0.0157, "step": 14212 }, { "epoch": 14.42, "learning_rate": 7.579490328064265e-08, "loss": 0.0177, "step": 14213 }, { "epoch": 14.42, "learning_rate": 7.55257502463469e-08, "loss": 0.015, "step": 14214 }, { "epoch": 14.42, "learning_rate": 7.525707414163364e-08, "loss": 0.0189, "step": 14215 }, { "epoch": 14.43, "learning_rate": 7.498887497941476e-08, "loss": 0.0205, "step": 14216 }, { "epoch": 14.43, "learning_rate": 7.472115277257774e-08, "loss": 0.021, "step": 14217 }, { "epoch": 14.43, "learning_rate": 7.445390753398784e-08, "loss": 0.0179, "step": 14218 }, { "epoch": 14.43, "learning_rate": 7.418713927648924e-08, "loss": 0.0162, "step": 14219 }, { "epoch": 14.43, "learning_rate": 7.392084801289945e-08, "loss": 0.0173, "step": 14220 }, { "epoch": 14.43, "learning_rate": 7.365503375601823e-08, "loss": 0.0148, "step": 14221 }, { "epoch": 14.43, "learning_rate": 7.338969651861761e-08, "loss": 0.0199, "step": 14222 }, { "epoch": 14.43, "learning_rate": 7.312483631344846e-08, "loss": 0.0166, "step": 14223 }, { "epoch": 14.43, "learning_rate": 7.286045315324064e-08, "loss": 0.0192, "step": 14224 }, { "epoch": 14.43, "learning_rate": 7.25965470506973e-08, "loss": 0.0173, "step": 14225 }, { "epoch": 14.44, "learning_rate": 7.233311801850274e-08, "loss": 0.018, "step": 14226 }, { "epoch": 14.44, "learning_rate": 7.20701660693135e-08, "loss": 0.0162, "step": 14227 }, { "epoch": 14.44, "learning_rate": 7.180769121576836e-08, "loss": 0.0155, "step": 14228 }, { "epoch": 14.44, "learning_rate": 7.154569347047946e-08, "loss": 0.0177, "step": 14229 }, { "epoch": 14.44, "learning_rate": 7.128417284603895e-08, "loss": 0.0178, "step": 14230 }, { "epoch": 14.44, "learning_rate": 7.102312935501343e-08, "loss": 0.02, "step": 14231 }, { "epoch": 14.44, "learning_rate": 7.076256300994622e-08, "loss": 0.0183, "step": 14232 }, { "epoch": 14.44, "learning_rate": 7.050247382336172e-08, "loss": 0.0182, "step": 14233 }, { "epoch": 14.44, "learning_rate": 7.024286180775664e-08, "loss": 0.0161, "step": 14234 }, { "epoch": 14.44, "learning_rate": 6.998372697560762e-08, "loss": 0.0169, "step": 14235 }, { "epoch": 14.45, "learning_rate": 6.972506933936807e-08, "loss": 0.0178, "step": 14236 }, { "epoch": 14.45, "learning_rate": 6.946688891146692e-08, "loss": 0.0165, "step": 14237 }, { "epoch": 14.45, "learning_rate": 6.920918570431311e-08, "loss": 0.0148, "step": 14238 }, { "epoch": 14.45, "learning_rate": 6.895195973028901e-08, "loss": 0.015, "step": 14239 }, { "epoch": 14.45, "learning_rate": 6.869521100175691e-08, "loss": 0.0211, "step": 14240 }, { "epoch": 14.45, "learning_rate": 6.843893953105363e-08, "loss": 0.0162, "step": 14241 }, { "epoch": 14.45, "learning_rate": 6.818314533049597e-08, "loss": 0.0158, "step": 14242 }, { "epoch": 14.45, "learning_rate": 6.792782841237633e-08, "loss": 0.0164, "step": 14243 }, { "epoch": 14.45, "learning_rate": 6.767298878896378e-08, "loss": 0.0175, "step": 14244 }, { "epoch": 14.45, "learning_rate": 6.741862647250408e-08, "loss": 0.0211, "step": 14245 }, { "epoch": 14.46, "learning_rate": 6.71647414752219e-08, "loss": 0.0142, "step": 14246 }, { "epoch": 14.46, "learning_rate": 6.691133380931858e-08, "loss": 0.0198, "step": 14247 }, { "epoch": 14.46, "learning_rate": 6.665840348696994e-08, "loss": 0.0157, "step": 14248 }, { "epoch": 14.46, "learning_rate": 6.640595052033183e-08, "loss": 0.0188, "step": 14249 }, { "epoch": 14.46, "learning_rate": 6.615397492153674e-08, "loss": 0.0167, "step": 14250 }, { "epoch": 14.46, "learning_rate": 6.590247670269279e-08, "loss": 0.0203, "step": 14251 }, { "epoch": 14.46, "learning_rate": 6.565145587588474e-08, "loss": 0.0137, "step": 14252 }, { "epoch": 14.46, "learning_rate": 6.540091245317847e-08, "loss": 0.0154, "step": 14253 }, { "epoch": 14.46, "learning_rate": 6.515084644661107e-08, "loss": 0.0162, "step": 14254 }, { "epoch": 14.46, "learning_rate": 6.49012578682029e-08, "loss": 0.0191, "step": 14255 }, { "epoch": 14.47, "learning_rate": 6.465214672994547e-08, "loss": 0.0169, "step": 14256 }, { "epoch": 14.47, "learning_rate": 6.440351304381142e-08, "loss": 0.0157, "step": 14257 }, { "epoch": 14.47, "learning_rate": 6.415535682174789e-08, "loss": 0.017, "step": 14258 }, { "epoch": 14.47, "learning_rate": 6.390767807568199e-08, "loss": 0.0172, "step": 14259 }, { "epoch": 14.47, "learning_rate": 6.36604768175153e-08, "loss": 0.0185, "step": 14260 }, { "epoch": 14.47, "learning_rate": 6.341375305912612e-08, "loss": 0.0178, "step": 14261 }, { "epoch": 14.47, "learning_rate": 6.31675068123716e-08, "loss": 0.0164, "step": 14262 }, { "epoch": 14.47, "learning_rate": 6.292173808908674e-08, "loss": 0.0165, "step": 14263 }, { "epoch": 14.47, "learning_rate": 6.2676446901081e-08, "loss": 0.019, "step": 14264 }, { "epoch": 14.47, "learning_rate": 6.243163326014268e-08, "loss": 0.0174, "step": 14265 }, { "epoch": 14.48, "learning_rate": 6.218729717803573e-08, "loss": 0.0157, "step": 14266 }, { "epoch": 14.48, "learning_rate": 6.194343866650188e-08, "loss": 0.0156, "step": 14267 }, { "epoch": 14.48, "learning_rate": 6.170005773726063e-08, "loss": 0.0172, "step": 14268 }, { "epoch": 14.48, "learning_rate": 6.145715440200817e-08, "loss": 0.0197, "step": 14269 }, { "epoch": 14.48, "learning_rate": 6.121472867241629e-08, "loss": 0.0189, "step": 14270 }, { "epoch": 14.48, "learning_rate": 6.097278056013567e-08, "loss": 0.0201, "step": 14271 }, { "epoch": 14.48, "learning_rate": 6.073131007679366e-08, "loss": 0.017, "step": 14272 }, { "epoch": 14.48, "learning_rate": 6.049031723399323e-08, "loss": 0.0175, "step": 14273 }, { "epoch": 14.48, "learning_rate": 6.024980204331731e-08, "loss": 0.0205, "step": 14274 }, { "epoch": 14.49, "learning_rate": 6.000976451632223e-08, "loss": 0.0174, "step": 14275 }, { "epoch": 14.49, "learning_rate": 5.97702046645443e-08, "loss": 0.0163, "step": 14276 }, { "epoch": 14.49, "learning_rate": 5.953112249949544e-08, "loss": 0.0187, "step": 14277 }, { "epoch": 14.49, "learning_rate": 5.929251803266422e-08, "loss": 0.019, "step": 14278 }, { "epoch": 14.49, "learning_rate": 5.9054391275518155e-08, "loss": 0.0198, "step": 14279 }, { "epoch": 14.49, "learning_rate": 5.8816742239500293e-08, "loss": 0.0166, "step": 14280 }, { "epoch": 14.49, "learning_rate": 5.857957093603151e-08, "loss": 0.019, "step": 14281 }, { "epoch": 14.49, "learning_rate": 5.8342877376508235e-08, "loss": 0.0138, "step": 14282 }, { "epoch": 14.49, "learning_rate": 5.8106661572305824e-08, "loss": 0.0169, "step": 14283 }, { "epoch": 14.49, "learning_rate": 5.7870923534776304e-08, "loss": 0.0179, "step": 14284 }, { "epoch": 14.5, "learning_rate": 5.7635663275246165e-08, "loss": 0.0148, "step": 14285 }, { "epoch": 14.5, "learning_rate": 5.740088080502304e-08, "loss": 0.0179, "step": 14286 }, { "epoch": 14.5, "learning_rate": 5.716657613538901e-08, "loss": 0.0199, "step": 14287 }, { "epoch": 14.5, "learning_rate": 5.693274927760395e-08, "loss": 0.0194, "step": 14288 }, { "epoch": 14.5, "learning_rate": 5.669940024290443e-08, "loss": 0.0153, "step": 14289 }, { "epoch": 14.5, "learning_rate": 5.6466529042503714e-08, "loss": 0.0138, "step": 14290 }, { "epoch": 14.5, "learning_rate": 5.623413568759395e-08, "loss": 0.0161, "step": 14291 }, { "epoch": 14.5, "learning_rate": 5.600222018934176e-08, "loss": 0.016, "step": 14292 }, { "epoch": 14.5, "learning_rate": 5.577078255889268e-08, "loss": 0.0156, "step": 14293 }, { "epoch": 14.5, "learning_rate": 5.553982280736892e-08, "loss": 0.02, "step": 14294 }, { "epoch": 14.51, "learning_rate": 5.5309340945868264e-08, "loss": 0.0208, "step": 14295 }, { "epoch": 14.51, "learning_rate": 5.5079336985468526e-08, "loss": 0.0174, "step": 14296 }, { "epoch": 14.51, "learning_rate": 5.484981093722086e-08, "loss": 0.0182, "step": 14297 }, { "epoch": 14.51, "learning_rate": 5.462076281215756e-08, "loss": 0.0171, "step": 14298 }, { "epoch": 14.51, "learning_rate": 5.4392192621283146e-08, "loss": 0.0193, "step": 14299 }, { "epoch": 14.51, "learning_rate": 5.416410037558328e-08, "loss": 0.0154, "step": 14300 }, { "epoch": 14.51, "learning_rate": 5.3936486086020315e-08, "loss": 0.0194, "step": 14301 }, { "epoch": 14.51, "learning_rate": 5.3709349763528816e-08, "loss": 0.0176, "step": 14302 }, { "epoch": 14.51, "learning_rate": 5.348269141902784e-08, "loss": 0.0162, "step": 14303 }, { "epoch": 14.51, "learning_rate": 5.325651106340757e-08, "loss": 0.0167, "step": 14304 }, { "epoch": 14.52, "learning_rate": 5.3030808707537075e-08, "loss": 0.0175, "step": 14305 }, { "epoch": 14.52, "learning_rate": 5.280558436226435e-08, "loss": 0.016, "step": 14306 }, { "epoch": 14.52, "learning_rate": 5.258083803841185e-08, "loss": 0.0174, "step": 14307 }, { "epoch": 14.52, "learning_rate": 5.235656974677983e-08, "loss": 0.0157, "step": 14308 }, { "epoch": 14.52, "learning_rate": 5.213277949814632e-08, "loss": 0.0169, "step": 14309 }, { "epoch": 14.52, "learning_rate": 5.190946730326496e-08, "loss": 0.0173, "step": 14310 }, { "epoch": 14.52, "learning_rate": 5.168663317286715e-08, "loss": 0.0202, "step": 14311 }, { "epoch": 14.52, "learning_rate": 5.1464277117661e-08, "loss": 0.0189, "step": 14312 }, { "epoch": 14.52, "learning_rate": 5.124239914833462e-08, "loss": 0.0164, "step": 14313 }, { "epoch": 14.52, "learning_rate": 5.102099927554727e-08, "loss": 0.0166, "step": 14314 }, { "epoch": 14.53, "learning_rate": 5.080007750994043e-08, "loss": 0.0171, "step": 14315 }, { "epoch": 14.53, "learning_rate": 5.057963386213116e-08, "loss": 0.0169, "step": 14316 }, { "epoch": 14.53, "learning_rate": 5.0359668342711e-08, "loss": 0.0167, "step": 14317 }, { "epoch": 14.53, "learning_rate": 5.014018096225259e-08, "loss": 0.0202, "step": 14318 }, { "epoch": 14.53, "learning_rate": 4.9921171731301955e-08, "loss": 0.0165, "step": 14319 }, { "epoch": 14.53, "learning_rate": 4.9702640660385104e-08, "loss": 0.0167, "step": 14320 }, { "epoch": 14.53, "learning_rate": 4.9484587760002535e-08, "loss": 0.0162, "step": 14321 }, { "epoch": 14.53, "learning_rate": 4.9267013040634746e-08, "loss": 0.0166, "step": 14322 }, { "epoch": 14.53, "learning_rate": 4.9049916512735604e-08, "loss": 0.0186, "step": 14323 }, { "epoch": 14.53, "learning_rate": 4.883329818673899e-08, "loss": 0.0219, "step": 14324 }, { "epoch": 14.54, "learning_rate": 4.861715807305434e-08, "loss": 0.017, "step": 14325 }, { "epoch": 14.54, "learning_rate": 4.840149618206891e-08, "loss": 0.0182, "step": 14326 }, { "epoch": 14.54, "learning_rate": 4.818631252414663e-08, "loss": 0.0153, "step": 14327 }, { "epoch": 14.54, "learning_rate": 4.7971607109625894e-08, "loss": 0.0182, "step": 14328 }, { "epoch": 14.54, "learning_rate": 4.7757379948828454e-08, "loss": 0.0169, "step": 14329 }, { "epoch": 14.54, "learning_rate": 4.754363105204607e-08, "loss": 0.0196, "step": 14330 }, { "epoch": 14.54, "learning_rate": 4.7330360429552745e-08, "loss": 0.0167, "step": 14331 }, { "epoch": 14.54, "learning_rate": 4.711756809159585e-08, "loss": 0.0159, "step": 14332 }, { "epoch": 14.54, "learning_rate": 4.690525404840163e-08, "loss": 0.0135, "step": 14333 }, { "epoch": 14.54, "learning_rate": 4.6693418310174155e-08, "loss": 0.0194, "step": 14334 }, { "epoch": 14.55, "learning_rate": 4.648206088709195e-08, "loss": 0.0156, "step": 14335 }, { "epoch": 14.55, "learning_rate": 4.6271181789313554e-08, "loss": 0.0169, "step": 14336 }, { "epoch": 14.55, "learning_rate": 4.606078102696976e-08, "loss": 0.0181, "step": 14337 }, { "epoch": 14.55, "learning_rate": 4.58508586101758e-08, "loss": 0.0202, "step": 14338 }, { "epoch": 14.55, "learning_rate": 4.5641414549016936e-08, "loss": 0.0143, "step": 14339 }, { "epoch": 14.55, "learning_rate": 4.543244885355957e-08, "loss": 0.0158, "step": 14340 }, { "epoch": 14.55, "learning_rate": 4.5223961533844566e-08, "loss": 0.0167, "step": 14341 }, { "epoch": 14.55, "learning_rate": 4.5015952599891665e-08, "loss": 0.0179, "step": 14342 }, { "epoch": 14.55, "learning_rate": 4.480842206169733e-08, "loss": 0.0173, "step": 14343 }, { "epoch": 14.56, "learning_rate": 4.460136992923358e-08, "loss": 0.0189, "step": 14344 }, { "epoch": 14.56, "learning_rate": 4.439479621245135e-08, "loss": 0.0173, "step": 14345 }, { "epoch": 14.56, "learning_rate": 4.418870092127603e-08, "loss": 0.0151, "step": 14346 }, { "epoch": 14.56, "learning_rate": 4.3983084065614135e-08, "loss": 0.0155, "step": 14347 }, { "epoch": 14.56, "learning_rate": 4.377794565534554e-08, "loss": 0.0145, "step": 14348 }, { "epoch": 14.56, "learning_rate": 4.3573285700329035e-08, "loss": 0.0193, "step": 14349 }, { "epoch": 14.56, "learning_rate": 4.336910421039897e-08, "loss": 0.0167, "step": 14350 }, { "epoch": 14.56, "learning_rate": 4.3165401195367496e-08, "loss": 0.0166, "step": 14351 }, { "epoch": 14.56, "learning_rate": 4.296217666502345e-08, "loss": 0.0182, "step": 14352 }, { "epoch": 14.56, "learning_rate": 4.275943062913346e-08, "loss": 0.0178, "step": 14353 }, { "epoch": 14.57, "learning_rate": 4.255716309744085e-08, "loss": 0.0172, "step": 14354 }, { "epoch": 14.57, "learning_rate": 4.235537407966672e-08, "loss": 0.0192, "step": 14355 }, { "epoch": 14.57, "learning_rate": 4.2154063585505554e-08, "loss": 0.0161, "step": 14356 }, { "epoch": 14.57, "learning_rate": 4.195323162463405e-08, "loss": 0.0154, "step": 14357 }, { "epoch": 14.57, "learning_rate": 4.175287820670226e-08, "loss": 0.018, "step": 14358 }, { "epoch": 14.57, "learning_rate": 4.155300334133805e-08, "loss": 0.0168, "step": 14359 }, { "epoch": 14.57, "learning_rate": 4.135360703814817e-08, "loss": 0.0188, "step": 14360 }, { "epoch": 14.57, "learning_rate": 4.115468930671274e-08, "loss": 0.0179, "step": 14361 }, { "epoch": 14.57, "learning_rate": 4.095625015659077e-08, "loss": 0.0164, "step": 14362 }, { "epoch": 14.57, "learning_rate": 4.075828959732131e-08, "loss": 0.0193, "step": 14363 }, { "epoch": 14.58, "learning_rate": 4.0560807638414526e-08, "loss": 0.0158, "step": 14364 }, { "epoch": 14.58, "learning_rate": 4.036380428936282e-08, "loss": 0.0157, "step": 14365 }, { "epoch": 14.58, "learning_rate": 4.016727955963085e-08, "loss": 0.0154, "step": 14366 }, { "epoch": 14.58, "learning_rate": 3.9971233458665495e-08, "loss": 0.0187, "step": 14367 }, { "epoch": 14.58, "learning_rate": 3.977566599588589e-08, "loss": 0.0218, "step": 14368 }, { "epoch": 14.58, "learning_rate": 3.95805771806923e-08, "loss": 0.0198, "step": 14369 }, { "epoch": 14.58, "learning_rate": 3.9385967022457225e-08, "loss": 0.0182, "step": 14370 }, { "epoch": 14.58, "learning_rate": 3.91918355305343e-08, "loss": 0.0191, "step": 14371 }, { "epoch": 14.58, "learning_rate": 3.899818271425382e-08, "loss": 0.0168, "step": 14372 }, { "epoch": 14.58, "learning_rate": 3.880500858291947e-08, "loss": 0.0175, "step": 14373 }, { "epoch": 14.59, "learning_rate": 3.861231314581604e-08, "loss": 0.0178, "step": 14374 }, { "epoch": 14.59, "learning_rate": 3.84200964122039e-08, "loss": 0.0186, "step": 14375 }, { "epoch": 14.59, "learning_rate": 3.8228358391318996e-08, "loss": 0.0204, "step": 14376 }, { "epoch": 14.59, "learning_rate": 3.803709909237618e-08, "loss": 0.0161, "step": 14377 }, { "epoch": 14.59, "learning_rate": 3.7846318524565885e-08, "loss": 0.017, "step": 14378 }, { "epoch": 14.59, "learning_rate": 3.765601669705743e-08, "loss": 0.0191, "step": 14379 }, { "epoch": 14.59, "learning_rate": 3.746619361899573e-08, "loss": 0.0181, "step": 14380 }, { "epoch": 14.59, "learning_rate": 3.727684929950237e-08, "loss": 0.0189, "step": 14381 }, { "epoch": 14.59, "learning_rate": 3.708798374767675e-08, "loss": 0.0146, "step": 14382 }, { "epoch": 14.59, "learning_rate": 3.689959697259382e-08, "loss": 0.0164, "step": 14383 }, { "epoch": 14.6, "learning_rate": 3.671168898330857e-08, "loss": 0.0151, "step": 14384 }, { "epoch": 14.6, "learning_rate": 3.6524259788850436e-08, "loss": 0.0221, "step": 14385 }, { "epoch": 14.6, "learning_rate": 3.633730939822555e-08, "loss": 0.0168, "step": 14386 }, { "epoch": 14.6, "learning_rate": 3.615083782041895e-08, "loss": 0.0161, "step": 14387 }, { "epoch": 14.6, "learning_rate": 3.596484506439124e-08, "loss": 0.0183, "step": 14388 }, { "epoch": 14.6, "learning_rate": 3.577933113908194e-08, "loss": 0.0177, "step": 14389 }, { "epoch": 14.6, "learning_rate": 3.559429605340392e-08, "loss": 0.017, "step": 14390 }, { "epoch": 14.6, "learning_rate": 3.5409739816250066e-08, "loss": 0.0163, "step": 14391 }, { "epoch": 14.6, "learning_rate": 3.5225662436489947e-08, "loss": 0.0143, "step": 14392 }, { "epoch": 14.6, "learning_rate": 3.504206392296872e-08, "loss": 0.0162, "step": 14393 }, { "epoch": 14.61, "learning_rate": 3.485894428450931e-08, "loss": 0.0186, "step": 14394 }, { "epoch": 14.61, "learning_rate": 3.467630352991247e-08, "loss": 0.0169, "step": 14395 }, { "epoch": 14.61, "learning_rate": 3.449414166795562e-08, "loss": 0.0152, "step": 14396 }, { "epoch": 14.61, "learning_rate": 3.4312458707390636e-08, "loss": 0.0173, "step": 14397 }, { "epoch": 14.61, "learning_rate": 3.413125465695055e-08, "loss": 0.0168, "step": 14398 }, { "epoch": 14.61, "learning_rate": 3.3950529525342833e-08, "loss": 0.0146, "step": 14399 }, { "epoch": 14.61, "learning_rate": 3.377028332125165e-08, "loss": 0.0177, "step": 14400 }, { "epoch": 14.61, "learning_rate": 3.359051605333896e-08, "loss": 0.0132, "step": 14401 }, { "epoch": 14.61, "learning_rate": 3.341122773024341e-08, "loss": 0.0168, "step": 14402 }, { "epoch": 14.61, "learning_rate": 3.323241836058255e-08, "loss": 0.0169, "step": 14403 }, { "epoch": 14.62, "learning_rate": 3.305408795294729e-08, "loss": 0.019, "step": 14404 }, { "epoch": 14.62, "learning_rate": 3.287623651590854e-08, "loss": 0.017, "step": 14405 }, { "epoch": 14.62, "learning_rate": 3.2698864058013925e-08, "loss": 0.0159, "step": 14406 }, { "epoch": 14.62, "learning_rate": 3.252197058778439e-08, "loss": 0.0166, "step": 14407 }, { "epoch": 14.62, "learning_rate": 3.234555611372314e-08, "loss": 0.0174, "step": 14408 }, { "epoch": 14.62, "learning_rate": 3.216962064430784e-08, "loss": 0.0148, "step": 14409 }, { "epoch": 14.62, "learning_rate": 3.1994164187991724e-08, "loss": 0.0172, "step": 14410 }, { "epoch": 14.62, "learning_rate": 3.1819186753208056e-08, "loss": 0.0174, "step": 14411 }, { "epoch": 14.62, "learning_rate": 3.164468834836454e-08, "loss": 0.021, "step": 14412 }, { "epoch": 14.63, "learning_rate": 3.147066898184781e-08, "loss": 0.0155, "step": 14413 }, { "epoch": 14.63, "learning_rate": 3.129712866202006e-08, "loss": 0.0134, "step": 14414 }, { "epoch": 14.63, "learning_rate": 3.112406739722018e-08, "loss": 0.0184, "step": 14415 }, { "epoch": 14.63, "learning_rate": 3.095148519576596e-08, "loss": 0.0195, "step": 14416 }, { "epoch": 14.63, "learning_rate": 3.077938206594966e-08, "loss": 0.0197, "step": 14417 }, { "epoch": 14.63, "learning_rate": 3.0607758016043546e-08, "loss": 0.0179, "step": 14418 }, { "epoch": 14.63, "learning_rate": 3.043661305429324e-08, "loss": 0.0154, "step": 14419 }, { "epoch": 14.63, "learning_rate": 3.026594718892439e-08, "loss": 0.0193, "step": 14420 }, { "epoch": 14.63, "learning_rate": 3.009576042813933e-08, "loss": 0.0185, "step": 14421 }, { "epoch": 14.63, "learning_rate": 2.9926052780114846e-08, "loss": 0.0181, "step": 14422 }, { "epoch": 14.64, "learning_rate": 2.9756824253007742e-08, "loss": 0.0168, "step": 14423 }, { "epoch": 14.64, "learning_rate": 2.95880748549493e-08, "loss": 0.0178, "step": 14424 }, { "epoch": 14.64, "learning_rate": 2.9419804594049695e-08, "loss": 0.0183, "step": 14425 }, { "epoch": 14.64, "learning_rate": 2.925201347839468e-08, "loss": 0.0182, "step": 14426 }, { "epoch": 14.64, "learning_rate": 2.908470151604892e-08, "loss": 0.0181, "step": 14427 }, { "epoch": 14.64, "learning_rate": 2.8917868715051535e-08, "loss": 0.0187, "step": 14428 }, { "epoch": 14.64, "learning_rate": 2.8751515083419447e-08, "loss": 0.0163, "step": 14429 }, { "epoch": 14.64, "learning_rate": 2.858564062914848e-08, "loss": 0.0146, "step": 14430 }, { "epoch": 14.64, "learning_rate": 2.842024536020782e-08, "loss": 0.0178, "step": 14431 }, { "epoch": 14.64, "learning_rate": 2.825532928454777e-08, "loss": 0.0167, "step": 14432 }, { "epoch": 14.65, "learning_rate": 2.8090892410093108e-08, "loss": 0.0163, "step": 14433 }, { "epoch": 14.65, "learning_rate": 2.7926934744745283e-08, "loss": 0.0174, "step": 14434 }, { "epoch": 14.65, "learning_rate": 2.776345629638355e-08, "loss": 0.0177, "step": 14435 }, { "epoch": 14.65, "learning_rate": 2.760045707286385e-08, "loss": 0.0159, "step": 14436 }, { "epoch": 14.65, "learning_rate": 2.7437937082018806e-08, "loss": 0.0177, "step": 14437 }, { "epoch": 14.65, "learning_rate": 2.7275896331661056e-08, "loss": 0.0175, "step": 14438 }, { "epoch": 14.65, "learning_rate": 2.7114334829574375e-08, "loss": 0.017, "step": 14439 }, { "epoch": 14.65, "learning_rate": 2.695325258352477e-08, "loss": 0.0117, "step": 14440 }, { "epoch": 14.65, "learning_rate": 2.6792649601252717e-08, "loss": 0.0147, "step": 14441 }, { "epoch": 14.65, "learning_rate": 2.663252589047538e-08, "loss": 0.0186, "step": 14442 }, { "epoch": 14.66, "learning_rate": 2.647288145888882e-08, "loss": 0.0162, "step": 14443 }, { "epoch": 14.66, "learning_rate": 2.6313716314163573e-08, "loss": 0.0191, "step": 14444 }, { "epoch": 14.66, "learning_rate": 2.6155030463949073e-08, "loss": 0.0175, "step": 14445 }, { "epoch": 14.66, "learning_rate": 2.599682391587255e-08, "loss": 0.0177, "step": 14446 }, { "epoch": 14.66, "learning_rate": 2.5839096677534592e-08, "loss": 0.0173, "step": 14447 }, { "epoch": 14.66, "learning_rate": 2.5681848756515803e-08, "loss": 0.0161, "step": 14448 }, { "epoch": 14.66, "learning_rate": 2.5525080160372362e-08, "loss": 0.0179, "step": 14449 }, { "epoch": 14.66, "learning_rate": 2.536879089663935e-08, "loss": 0.0191, "step": 14450 }, { "epoch": 14.66, "learning_rate": 2.521298097282632e-08, "loss": 0.016, "step": 14451 }, { "epoch": 14.66, "learning_rate": 2.5057650396420607e-08, "loss": 0.0201, "step": 14452 }, { "epoch": 14.67, "learning_rate": 2.490279917488736e-08, "loss": 0.0168, "step": 14453 }, { "epoch": 14.67, "learning_rate": 2.47484273156684e-08, "loss": 0.0156, "step": 14454 }, { "epoch": 14.67, "learning_rate": 2.4594534826181127e-08, "loss": 0.0187, "step": 14455 }, { "epoch": 14.67, "learning_rate": 2.4441121713821848e-08, "loss": 0.0158, "step": 14456 }, { "epoch": 14.67, "learning_rate": 2.428818798596244e-08, "loss": 0.0169, "step": 14457 }, { "epoch": 14.67, "learning_rate": 2.4135733649952587e-08, "loss": 0.018, "step": 14458 }, { "epoch": 14.67, "learning_rate": 2.3983758713118644e-08, "loss": 0.017, "step": 14459 }, { "epoch": 14.67, "learning_rate": 2.383226318276366e-08, "loss": 0.0173, "step": 14460 }, { "epoch": 14.67, "learning_rate": 2.3681247066168478e-08, "loss": 0.0167, "step": 14461 }, { "epoch": 14.67, "learning_rate": 2.3530710370588407e-08, "loss": 0.0169, "step": 14462 }, { "epoch": 14.68, "learning_rate": 2.3380653103260987e-08, "loss": 0.0151, "step": 14463 }, { "epoch": 14.68, "learning_rate": 2.3231075271394897e-08, "loss": 0.0182, "step": 14464 }, { "epoch": 14.68, "learning_rate": 2.3081976882177724e-08, "loss": 0.019, "step": 14465 }, { "epoch": 14.68, "learning_rate": 2.2933357942777067e-08, "loss": 0.0164, "step": 14466 }, { "epoch": 14.68, "learning_rate": 2.278521846033277e-08, "loss": 0.0172, "step": 14467 }, { "epoch": 14.68, "learning_rate": 2.263755844196469e-08, "loss": 0.0184, "step": 14468 }, { "epoch": 14.68, "learning_rate": 2.2490377894768266e-08, "loss": 0.0153, "step": 14469 }, { "epoch": 14.68, "learning_rate": 2.2343676825816728e-08, "loss": 0.0175, "step": 14470 }, { "epoch": 14.68, "learning_rate": 2.2197455242159993e-08, "loss": 0.017, "step": 14471 }, { "epoch": 14.68, "learning_rate": 2.205171315082466e-08, "loss": 0.0188, "step": 14472 }, { "epoch": 14.69, "learning_rate": 2.190645055881513e-08, "loss": 0.0187, "step": 14473 }, { "epoch": 14.69, "learning_rate": 2.176166747311137e-08, "loss": 0.018, "step": 14474 }, { "epoch": 14.69, "learning_rate": 2.161736390067115e-08, "loss": 0.016, "step": 14475 }, { "epoch": 14.69, "learning_rate": 2.1473539848428927e-08, "loss": 0.018, "step": 14476 }, { "epoch": 14.69, "learning_rate": 2.1330195323298052e-08, "loss": 0.016, "step": 14477 }, { "epoch": 14.69, "learning_rate": 2.1187330332164135e-08, "loss": 0.018, "step": 14478 }, { "epoch": 14.69, "learning_rate": 2.1044944881895013e-08, "loss": 0.0176, "step": 14479 }, { "epoch": 14.69, "learning_rate": 2.090303897933188e-08, "loss": 0.0194, "step": 14480 }, { "epoch": 14.69, "learning_rate": 2.0761612631294836e-08, "loss": 0.0157, "step": 14481 }, { "epoch": 14.7, "learning_rate": 2.0620665844580667e-08, "loss": 0.0184, "step": 14482 }, { "epoch": 14.7, "learning_rate": 2.0480198625960623e-08, "loss": 0.0176, "step": 14483 }, { "epoch": 14.7, "learning_rate": 2.034021098218819e-08, "loss": 0.0148, "step": 14484 }, { "epoch": 14.7, "learning_rate": 2.0200702919987992e-08, "loss": 0.0177, "step": 14485 }, { "epoch": 14.7, "learning_rate": 2.006167444606466e-08, "loss": 0.0157, "step": 14486 }, { "epoch": 14.7, "learning_rate": 1.9923125567099523e-08, "loss": 0.0152, "step": 14487 }, { "epoch": 14.7, "learning_rate": 1.9785056289751693e-08, "loss": 0.0141, "step": 14488 }, { "epoch": 14.7, "learning_rate": 1.9647466620655865e-08, "loss": 0.0182, "step": 14489 }, { "epoch": 14.7, "learning_rate": 1.9510356566422306e-08, "loss": 0.0173, "step": 14490 }, { "epoch": 14.7, "learning_rate": 1.937372613364241e-08, "loss": 0.018, "step": 14491 }, { "epoch": 14.71, "learning_rate": 1.9237575328879822e-08, "loss": 0.016, "step": 14492 }, { "epoch": 14.71, "learning_rate": 1.9101904158679295e-08, "loss": 0.0168, "step": 14493 }, { "epoch": 14.71, "learning_rate": 1.896671262955896e-08, "loss": 0.0163, "step": 14494 }, { "epoch": 14.71, "learning_rate": 1.8832000748016943e-08, "loss": 0.0162, "step": 14495 }, { "epoch": 14.71, "learning_rate": 1.869776852052696e-08, "loss": 0.0185, "step": 14496 }, { "epoch": 14.71, "learning_rate": 1.8564015953538296e-08, "loss": 0.018, "step": 14497 }, { "epoch": 14.71, "learning_rate": 1.8430743053479138e-08, "loss": 0.0183, "step": 14498 }, { "epoch": 14.71, "learning_rate": 1.8297949826755478e-08, "loss": 0.0167, "step": 14499 }, { "epoch": 14.71, "learning_rate": 1.8165636279746658e-08, "loss": 0.0146, "step": 14500 }, { "epoch": 14.71, "learning_rate": 1.803380241881314e-08, "loss": 0.0154, "step": 14501 }, { "epoch": 14.72, "learning_rate": 1.790244825028764e-08, "loss": 0.018, "step": 14502 }, { "epoch": 14.72, "learning_rate": 1.7771573780485107e-08, "loss": 0.0188, "step": 14503 }, { "epoch": 14.72, "learning_rate": 1.7641179015693843e-08, "loss": 0.0155, "step": 14504 }, { "epoch": 14.72, "learning_rate": 1.7511263962179948e-08, "loss": 0.0134, "step": 14505 }, { "epoch": 14.72, "learning_rate": 1.7381828626186204e-08, "loss": 0.0177, "step": 14506 }, { "epoch": 14.72, "learning_rate": 1.7252873013934302e-08, "loss": 0.0171, "step": 14507 }, { "epoch": 14.72, "learning_rate": 1.7124397131619287e-08, "loss": 0.0192, "step": 14508 }, { "epoch": 14.72, "learning_rate": 1.6996400985417326e-08, "loss": 0.0169, "step": 14509 }, { "epoch": 14.72, "learning_rate": 1.6868884581476842e-08, "loss": 0.018, "step": 14510 }, { "epoch": 14.72, "learning_rate": 1.6741847925928477e-08, "loss": 0.0151, "step": 14511 }, { "epoch": 14.73, "learning_rate": 1.6615291024875135e-08, "loss": 0.0159, "step": 14512 }, { "epoch": 14.73, "learning_rate": 1.6489213884399724e-08, "loss": 0.0173, "step": 14513 }, { "epoch": 14.73, "learning_rate": 1.636361651055962e-08, "loss": 0.018, "step": 14514 }, { "epoch": 14.73, "learning_rate": 1.623849890939222e-08, "loss": 0.0185, "step": 14515 }, { "epoch": 14.73, "learning_rate": 1.6113861086909375e-08, "loss": 0.0173, "step": 14516 }, { "epoch": 14.73, "learning_rate": 1.5989703049100747e-08, "loss": 0.0184, "step": 14517 }, { "epoch": 14.73, "learning_rate": 1.5866024801932667e-08, "loss": 0.0193, "step": 14518 }, { "epoch": 14.73, "learning_rate": 1.5742826351348163e-08, "loss": 0.015, "step": 14519 }, { "epoch": 14.73, "learning_rate": 1.562010770326916e-08, "loss": 0.017, "step": 14520 }, { "epoch": 14.73, "learning_rate": 1.5497868863590948e-08, "loss": 0.0169, "step": 14521 }, { "epoch": 14.74, "learning_rate": 1.5376109838189933e-08, "loss": 0.0162, "step": 14522 }, { "epoch": 14.74, "learning_rate": 1.5254830632915885e-08, "loss": 0.0155, "step": 14523 }, { "epoch": 14.74, "learning_rate": 1.513403125359747e-08, "loss": 0.0193, "step": 14524 }, { "epoch": 14.74, "learning_rate": 1.5013711706040047e-08, "loss": 0.02, "step": 14525 }, { "epoch": 14.74, "learning_rate": 1.4893871996025655e-08, "loss": 0.014, "step": 14526 }, { "epoch": 14.74, "learning_rate": 1.4774512129313023e-08, "loss": 0.0166, "step": 14527 }, { "epoch": 14.74, "learning_rate": 1.4655632111637564e-08, "loss": 0.0172, "step": 14528 }, { "epoch": 14.74, "learning_rate": 1.4537231948712482e-08, "loss": 0.0182, "step": 14529 }, { "epoch": 14.74, "learning_rate": 1.4419311646228784e-08, "loss": 0.0203, "step": 14530 }, { "epoch": 14.74, "learning_rate": 1.4301871209851936e-08, "loss": 0.0164, "step": 14531 }, { "epoch": 14.75, "learning_rate": 1.4184910645226312e-08, "loss": 0.0171, "step": 14532 }, { "epoch": 14.75, "learning_rate": 1.4068429957972973e-08, "loss": 0.0212, "step": 14533 }, { "epoch": 14.75, "learning_rate": 1.395242915368744e-08, "loss": 0.0181, "step": 14534 }, { "epoch": 14.75, "learning_rate": 1.3836908237946367e-08, "loss": 0.0164, "step": 14535 }, { "epoch": 14.75, "learning_rate": 1.3721867216300866e-08, "loss": 0.0172, "step": 14536 }, { "epoch": 14.75, "learning_rate": 1.3607306094277628e-08, "loss": 0.0166, "step": 14537 }, { "epoch": 14.75, "learning_rate": 1.349322487738447e-08, "loss": 0.0172, "step": 14538 }, { "epoch": 14.75, "learning_rate": 1.3379623571102562e-08, "loss": 0.0167, "step": 14539 }, { "epoch": 14.75, "learning_rate": 1.3266502180889763e-08, "loss": 0.0163, "step": 14540 }, { "epoch": 14.75, "learning_rate": 1.3153860712185052e-08, "loss": 0.021, "step": 14541 }, { "epoch": 14.76, "learning_rate": 1.3041699170398548e-08, "loss": 0.0174, "step": 14542 }, { "epoch": 14.76, "learning_rate": 1.2930017560922603e-08, "loss": 0.0173, "step": 14543 }, { "epoch": 14.76, "learning_rate": 1.2818815889124037e-08, "loss": 0.0146, "step": 14544 }, { "epoch": 14.76, "learning_rate": 1.2708094160345242e-08, "loss": 0.0174, "step": 14545 }, { "epoch": 14.76, "learning_rate": 1.2597852379907516e-08, "loss": 0.0192, "step": 14546 }, { "epoch": 14.76, "learning_rate": 1.2488090553108845e-08, "loss": 0.02, "step": 14547 }, { "epoch": 14.76, "learning_rate": 1.2378808685223897e-08, "loss": 0.0173, "step": 14548 }, { "epoch": 14.76, "learning_rate": 1.2270006781505139e-08, "loss": 0.0177, "step": 14549 }, { "epoch": 14.76, "learning_rate": 1.21616848471795e-08, "loss": 0.0154, "step": 14550 }, { "epoch": 14.77, "learning_rate": 1.2053842887453926e-08, "loss": 0.0156, "step": 14551 }, { "epoch": 14.77, "learning_rate": 1.1946480907510938e-08, "loss": 0.0166, "step": 14552 }, { "epoch": 14.77, "learning_rate": 1.1839598912508632e-08, "loss": 0.0191, "step": 14553 }, { "epoch": 14.77, "learning_rate": 1.173319690758401e-08, "loss": 0.0145, "step": 14554 }, { "epoch": 14.77, "learning_rate": 1.162727489785076e-08, "loss": 0.0158, "step": 14555 }, { "epoch": 14.77, "learning_rate": 1.1521832888398144e-08, "loss": 0.0183, "step": 14556 }, { "epoch": 14.77, "learning_rate": 1.1416870884294329e-08, "loss": 0.0199, "step": 14557 }, { "epoch": 14.77, "learning_rate": 1.1312388890581949e-08, "loss": 0.0161, "step": 14558 }, { "epoch": 14.77, "learning_rate": 1.1208386912282543e-08, "loss": 0.0157, "step": 14559 }, { "epoch": 14.77, "learning_rate": 1.1104864954395444e-08, "loss": 0.0184, "step": 14560 }, { "epoch": 14.78, "learning_rate": 1.1001823021894453e-08, "loss": 0.0147, "step": 14561 }, { "epoch": 14.78, "learning_rate": 1.0899261119730054e-08, "loss": 0.0181, "step": 14562 }, { "epoch": 14.78, "learning_rate": 1.0797179252832745e-08, "loss": 0.0162, "step": 14563 }, { "epoch": 14.78, "learning_rate": 1.0695577426107496e-08, "loss": 0.0186, "step": 14564 }, { "epoch": 14.78, "learning_rate": 1.0594455644437062e-08, "loss": 0.0152, "step": 14565 }, { "epoch": 14.78, "learning_rate": 1.0493813912679785e-08, "loss": 0.018, "step": 14566 }, { "epoch": 14.78, "learning_rate": 1.0393652235674012e-08, "loss": 0.0175, "step": 14567 }, { "epoch": 14.78, "learning_rate": 1.0293970618232563e-08, "loss": 0.0142, "step": 14568 }, { "epoch": 14.78, "learning_rate": 1.0194769065144939e-08, "loss": 0.0194, "step": 14569 }, { "epoch": 14.78, "learning_rate": 1.0096047581178437e-08, "loss": 0.0166, "step": 14570 }, { "epoch": 14.79, "learning_rate": 9.99780617107815e-09, "loss": 0.0164, "step": 14571 }, { "epoch": 14.79, "learning_rate": 9.900044839564749e-09, "loss": 0.0174, "step": 14572 }, { "epoch": 14.79, "learning_rate": 9.802763591335585e-09, "loss": 0.0135, "step": 14573 }, { "epoch": 14.79, "learning_rate": 9.705962431065807e-09, "loss": 0.0176, "step": 14574 }, { "epoch": 14.79, "learning_rate": 9.609641363407251e-09, "loss": 0.0181, "step": 14575 }, { "epoch": 14.79, "learning_rate": 9.513800392989548e-09, "loss": 0.0138, "step": 14576 }, { "epoch": 14.79, "learning_rate": 9.4184395244179e-09, "loss": 0.0185, "step": 14577 }, { "epoch": 14.79, "learning_rate": 9.32355876227531e-09, "loss": 0.018, "step": 14578 }, { "epoch": 14.79, "learning_rate": 9.229158111120352e-09, "loss": 0.0158, "step": 14579 }, { "epoch": 14.79, "learning_rate": 9.135237575490508e-09, "loss": 0.0178, "step": 14580 }, { "epoch": 14.8, "learning_rate": 9.041797159898836e-09, "loss": 0.0186, "step": 14581 }, { "epoch": 14.8, "learning_rate": 8.948836868836186e-09, "loss": 0.0201, "step": 14582 }, { "epoch": 14.8, "learning_rate": 8.856356706767877e-09, "loss": 0.0189, "step": 14583 }, { "epoch": 14.8, "learning_rate": 8.76435667814146e-09, "loss": 0.0184, "step": 14584 }, { "epoch": 14.8, "learning_rate": 8.672836787375628e-09, "loss": 0.0185, "step": 14585 }, { "epoch": 14.8, "learning_rate": 8.581797038869078e-09, "loss": 0.0183, "step": 14586 }, { "epoch": 14.8, "learning_rate": 8.491237436996092e-09, "loss": 0.0177, "step": 14587 }, { "epoch": 14.8, "learning_rate": 8.401157986109853e-09, "loss": 0.019, "step": 14588 }, { "epoch": 14.8, "learning_rate": 8.311558690539124e-09, "loss": 0.0139, "step": 14589 }, { "epoch": 14.8, "learning_rate": 8.222439554588235e-09, "loss": 0.0173, "step": 14590 }, { "epoch": 14.81, "learning_rate": 8.13380058254265e-09, "loss": 0.0132, "step": 14591 }, { "epoch": 14.81, "learning_rate": 8.045641778658963e-09, "loss": 0.0172, "step": 14592 }, { "epoch": 14.81, "learning_rate": 7.957963147174897e-09, "loss": 0.0168, "step": 14593 }, { "epoch": 14.81, "learning_rate": 7.870764692303745e-09, "loss": 0.0164, "step": 14594 }, { "epoch": 14.81, "learning_rate": 7.784046418236601e-09, "loss": 0.0182, "step": 14595 }, { "epoch": 14.81, "learning_rate": 7.697808329141243e-09, "loss": 0.0203, "step": 14596 }, { "epoch": 14.81, "learning_rate": 7.612050429159912e-09, "loss": 0.0177, "step": 14597 }, { "epoch": 14.81, "learning_rate": 7.526772722414866e-09, "loss": 0.0184, "step": 14598 }, { "epoch": 14.81, "learning_rate": 7.441975213005048e-09, "loss": 0.0169, "step": 14599 }, { "epoch": 14.81, "learning_rate": 7.357657905004978e-09, "loss": 0.0156, "step": 14600 }, { "epoch": 14.82, "learning_rate": 7.273820802465859e-09, "loss": 0.0203, "step": 14601 }, { "epoch": 14.82, "learning_rate": 7.1904639094166894e-09, "loss": 0.0126, "step": 14602 }, { "epoch": 14.82, "learning_rate": 7.1075872298631555e-09, "loss": 0.0153, "step": 14603 }, { "epoch": 14.82, "learning_rate": 7.025190767789847e-09, "loss": 0.0161, "step": 14604 }, { "epoch": 14.82, "learning_rate": 6.9432745271535986e-09, "loss": 0.0179, "step": 14605 }, { "epoch": 14.82, "learning_rate": 6.861838511893482e-09, "loss": 0.0185, "step": 14606 }, { "epoch": 14.82, "learning_rate": 6.780882725920812e-09, "loss": 0.016, "step": 14607 }, { "epoch": 14.82, "learning_rate": 6.7004071731269214e-09, "loss": 0.0132, "step": 14608 }, { "epoch": 14.82, "learning_rate": 6.620411857379827e-09, "loss": 0.0151, "step": 14609 }, { "epoch": 14.82, "learning_rate": 6.540896782523121e-09, "loss": 0.0153, "step": 14610 }, { "epoch": 14.83, "learning_rate": 6.46186195237819e-09, "loss": 0.0186, "step": 14611 }, { "epoch": 14.83, "learning_rate": 6.3833073707431085e-09, "loss": 0.0141, "step": 14612 }, { "epoch": 14.83, "learning_rate": 6.305233041391523e-09, "loss": 0.0174, "step": 14613 }, { "epoch": 14.83, "learning_rate": 6.227638968078209e-09, "loss": 0.015, "step": 14614 }, { "epoch": 14.83, "learning_rate": 6.150525154529074e-09, "loss": 0.0158, "step": 14615 }, { "epoch": 14.83, "learning_rate": 6.073891604452265e-09, "loss": 0.0153, "step": 14616 }, { "epoch": 14.83, "learning_rate": 5.997738321528168e-09, "loss": 0.0138, "step": 14617 }, { "epoch": 14.83, "learning_rate": 5.922065309418301e-09, "loss": 0.0147, "step": 14618 }, { "epoch": 14.83, "learning_rate": 5.846872571758644e-09, "loss": 0.013, "step": 14619 }, { "epoch": 14.84, "learning_rate": 5.772160112162972e-09, "loss": 0.0168, "step": 14620 }, { "epoch": 14.84, "learning_rate": 5.6979279342206375e-09, "loss": 0.0185, "step": 14621 }, { "epoch": 14.84, "learning_rate": 5.6241760414987856e-09, "loss": 0.0142, "step": 14622 }, { "epoch": 14.84, "learning_rate": 5.5509044375434695e-09, "loss": 0.0159, "step": 14623 }, { "epoch": 14.84, "learning_rate": 5.478113125875206e-09, "loss": 0.0157, "step": 14624 }, { "epoch": 14.84, "learning_rate": 5.405802109990088e-09, "loss": 0.0169, "step": 14625 }, { "epoch": 14.84, "learning_rate": 5.333971393365334e-09, "loss": 0.0208, "step": 14626 }, { "epoch": 14.84, "learning_rate": 5.2626209794526265e-09, "loss": 0.0138, "step": 14627 }, { "epoch": 14.84, "learning_rate": 5.1917508716792246e-09, "loss": 0.0178, "step": 14628 }, { "epoch": 14.84, "learning_rate": 5.121361073452402e-09, "loss": 0.0167, "step": 14629 }, { "epoch": 14.85, "learning_rate": 5.051451588153899e-09, "loss": 0.0164, "step": 14630 }, { "epoch": 14.85, "learning_rate": 4.982022419144361e-09, "loss": 0.0186, "step": 14631 }, { "epoch": 14.85, "learning_rate": 4.913073569758897e-09, "loss": 0.0191, "step": 14632 }, { "epoch": 14.85, "learning_rate": 4.844605043311523e-09, "loss": 0.0165, "step": 14633 }, { "epoch": 14.85, "learning_rate": 4.776616843092941e-09, "loss": 0.0193, "step": 14634 }, { "epoch": 14.85, "learning_rate": 4.709108972370535e-09, "loss": 0.0181, "step": 14635 }, { "epoch": 14.85, "learning_rate": 4.6420814343872685e-09, "loss": 0.0162, "step": 14636 }, { "epoch": 14.85, "learning_rate": 4.575534232363898e-09, "loss": 0.0164, "step": 14637 }, { "epoch": 14.85, "learning_rate": 4.509467369501197e-09, "loss": 0.0183, "step": 14638 }, { "epoch": 14.85, "learning_rate": 4.443880848971072e-09, "loss": 0.0186, "step": 14639 }, { "epoch": 14.86, "learning_rate": 4.378774673927666e-09, "loss": 0.0185, "step": 14640 }, { "epoch": 14.86, "learning_rate": 4.3141488474973685e-09, "loss": 0.0192, "step": 14641 }, { "epoch": 14.86, "learning_rate": 4.250003372787692e-09, "loss": 0.0189, "step": 14642 }, { "epoch": 14.86, "learning_rate": 4.186338252880617e-09, "loss": 0.013, "step": 14643 }, { "epoch": 14.86, "learning_rate": 4.123153490834808e-09, "loss": 0.0178, "step": 14644 }, { "epoch": 14.86, "learning_rate": 4.060449089688945e-09, "loss": 0.0187, "step": 14645 }, { "epoch": 14.86, "learning_rate": 3.998225052453952e-09, "loss": 0.0177, "step": 14646 }, { "epoch": 14.86, "learning_rate": 3.9364813821207715e-09, "loss": 0.0176, "step": 14647 }, { "epoch": 14.86, "learning_rate": 3.875218081657028e-09, "loss": 0.0159, "step": 14648 }, { "epoch": 14.86, "learning_rate": 3.814435154005924e-09, "loss": 0.018, "step": 14649 }, { "epoch": 14.87, "learning_rate": 3.754132602089566e-09, "loss": 0.0161, "step": 14650 }, { "epoch": 14.87, "learning_rate": 3.6943104288056363e-09, "loss": 0.0171, "step": 14651 }, { "epoch": 14.87, "learning_rate": 3.6349686370273917e-09, "loss": 0.0158, "step": 14652 }, { "epoch": 14.87, "learning_rate": 3.5761072296092157e-09, "loss": 0.0161, "step": 14653 }, { "epoch": 14.87, "learning_rate": 3.5177262093777363e-09, "loss": 0.0177, "step": 14654 }, { "epoch": 14.87, "learning_rate": 3.459825579138487e-09, "loss": 0.0196, "step": 14655 }, { "epoch": 14.87, "learning_rate": 3.402405341674797e-09, "loss": 0.0166, "step": 14656 }, { "epoch": 14.87, "learning_rate": 3.3454654997455702e-09, "loss": 0.0194, "step": 14657 }, { "epoch": 14.87, "learning_rate": 3.289006056088617e-09, "loss": 0.0188, "step": 14658 }, { "epoch": 14.87, "learning_rate": 3.233027013413992e-09, "loss": 0.0143, "step": 14659 }, { "epoch": 14.88, "learning_rate": 3.1775283744150954e-09, "loss": 0.0172, "step": 14660 }, { "epoch": 14.88, "learning_rate": 3.122510141756463e-09, "loss": 0.0164, "step": 14661 }, { "epoch": 14.88, "learning_rate": 3.067972318083756e-09, "loss": 0.0161, "step": 14662 }, { "epoch": 14.88, "learning_rate": 3.0139149060171015e-09, "loss": 0.0179, "step": 14663 }, { "epoch": 14.88, "learning_rate": 2.9603379081533103e-09, "loss": 0.0165, "step": 14664 }, { "epoch": 14.88, "learning_rate": 2.90724132706921e-09, "loss": 0.017, "step": 14665 }, { "epoch": 14.88, "learning_rate": 2.8546251653149837e-09, "loss": 0.02, "step": 14666 }, { "epoch": 14.88, "learning_rate": 2.8024894254197186e-09, "loss": 0.0177, "step": 14667 }, { "epoch": 14.88, "learning_rate": 2.7508341098880788e-09, "loss": 0.0187, "step": 14668 }, { "epoch": 14.88, "learning_rate": 2.699659221203632e-09, "loss": 0.0166, "step": 14669 }, { "epoch": 14.89, "learning_rate": 2.648964761824413e-09, "loss": 0.0173, "step": 14670 }, { "epoch": 14.89, "learning_rate": 2.598750734187361e-09, "loss": 0.0182, "step": 14671 }, { "epoch": 14.89, "learning_rate": 2.5490171407049904e-09, "loss": 0.0173, "step": 14672 }, { "epoch": 14.89, "learning_rate": 2.4997639837687217e-09, "loss": 0.0209, "step": 14673 }, { "epoch": 14.89, "learning_rate": 2.45099126574333e-09, "loss": 0.0178, "step": 14674 }, { "epoch": 14.89, "learning_rate": 2.4026989889736064e-09, "loss": 0.0199, "step": 14675 }, { "epoch": 14.89, "learning_rate": 2.354887155779917e-09, "loss": 0.0192, "step": 14676 }, { "epoch": 14.89, "learning_rate": 2.3075557684615336e-09, "loss": 0.0185, "step": 14677 }, { "epoch": 14.89, "learning_rate": 2.2607048292910827e-09, "loss": 0.0155, "step": 14678 }, { "epoch": 14.89, "learning_rate": 2.214334340521207e-09, "loss": 0.018, "step": 14679 }, { "epoch": 14.9, "learning_rate": 2.1684443043790137e-09, "loss": 0.017, "step": 14680 }, { "epoch": 14.9, "learning_rate": 2.123034723070516e-09, "loss": 0.0198, "step": 14681 }, { "epoch": 14.9, "learning_rate": 2.078105598778413e-09, "loss": 0.0172, "step": 14682 }, { "epoch": 14.9, "learning_rate": 2.033656933662087e-09, "loss": 0.0152, "step": 14683 }, { "epoch": 14.9, "learning_rate": 1.989688729855388e-09, "loss": 0.0159, "step": 14684 }, { "epoch": 14.9, "learning_rate": 1.9462009894744006e-09, "loss": 0.0173, "step": 14685 }, { "epoch": 14.9, "learning_rate": 1.903193714605234e-09, "loss": 0.0181, "step": 14686 }, { "epoch": 14.9, "learning_rate": 1.8606669073184536e-09, "loss": 0.0177, "step": 14687 }, { "epoch": 14.9, "learning_rate": 1.81862056965465e-09, "loss": 0.0193, "step": 14688 }, { "epoch": 14.91, "learning_rate": 1.7770547036366491e-09, "loss": 0.0176, "step": 14689 }, { "epoch": 14.91, "learning_rate": 1.7359693112606323e-09, "loss": 0.0178, "step": 14690 }, { "epoch": 14.91, "learning_rate": 1.6953643945016862e-09, "loss": 0.0157, "step": 14691 }, { "epoch": 14.91, "learning_rate": 1.6552399553104726e-09, "loss": 0.0183, "step": 14692 }, { "epoch": 14.91, "learning_rate": 1.6155959956154488e-09, "loss": 0.0178, "step": 14693 }, { "epoch": 14.91, "learning_rate": 1.5764325173217577e-09, "loss": 0.0184, "step": 14694 }, { "epoch": 14.91, "learning_rate": 1.5377495223123374e-09, "loss": 0.0166, "step": 14695 }, { "epoch": 14.91, "learning_rate": 1.499547012444591e-09, "loss": 0.0172, "step": 14696 }, { "epoch": 14.91, "learning_rate": 1.4618249895548274e-09, "loss": 0.0125, "step": 14697 }, { "epoch": 14.91, "learning_rate": 1.424583455456041e-09, "loss": 0.0168, "step": 14698 }, { "epoch": 14.92, "learning_rate": 1.3878224119379114e-09, "loss": 0.0163, "step": 14699 }, { "epoch": 14.92, "learning_rate": 1.3515418607679132e-09, "loss": 0.0168, "step": 14700 }, { "epoch": 14.92, "learning_rate": 1.3157418036879866e-09, "loss": 0.0156, "step": 14701 }, { "epoch": 14.92, "learning_rate": 1.2804222424189773e-09, "loss": 0.0166, "step": 14702 }, { "epoch": 14.92, "learning_rate": 1.245583178658416e-09, "loss": 0.0168, "step": 14703 }, { "epoch": 14.92, "learning_rate": 1.2112246140816298e-09, "loss": 0.0183, "step": 14704 }, { "epoch": 14.92, "learning_rate": 1.1773465503372994e-09, "loss": 0.0172, "step": 14705 }, { "epoch": 14.92, "learning_rate": 1.1439489890552324e-09, "loss": 0.0182, "step": 14706 }, { "epoch": 14.92, "learning_rate": 1.1110319318408114e-09, "loss": 0.0168, "step": 14707 }, { "epoch": 14.92, "learning_rate": 1.0785953802738835e-09, "loss": 0.0175, "step": 14708 }, { "epoch": 14.93, "learning_rate": 1.0466393359154226e-09, "loss": 0.0162, "step": 14709 }, { "epoch": 14.93, "learning_rate": 1.0151638002997566e-09, "loss": 0.0147, "step": 14710 }, { "epoch": 14.93, "learning_rate": 9.841687749390094e-10, "loss": 0.0203, "step": 14711 }, { "epoch": 14.93, "learning_rate": 9.536542613242106e-10, "loss": 0.021, "step": 14712 }, { "epoch": 14.93, "learning_rate": 9.236202609208545e-10, "loss": 0.0161, "step": 14713 }, { "epoch": 14.93, "learning_rate": 8.94066775172231e-10, "loss": 0.0154, "step": 14714 }, { "epoch": 14.93, "learning_rate": 8.649938054983153e-10, "loss": 0.0162, "step": 14715 }, { "epoch": 14.93, "learning_rate": 8.364013532979887e-10, "loss": 0.0167, "step": 14716 }, { "epoch": 14.93, "learning_rate": 8.082894199423763e-10, "loss": 0.0172, "step": 14717 }, { "epoch": 14.93, "learning_rate": 7.806580067848402e-10, "loss": 0.0146, "step": 14718 }, { "epoch": 14.94, "learning_rate": 7.53507115153207e-10, "loss": 0.0167, "step": 14719 }, { "epoch": 14.94, "learning_rate": 7.268367463497683e-10, "loss": 0.017, "step": 14720 }, { "epoch": 14.94, "learning_rate": 7.006469016590523e-10, "loss": 0.0172, "step": 14721 }, { "epoch": 14.94, "learning_rate": 6.749375823378312e-10, "loss": 0.0169, "step": 14722 }, { "epoch": 14.94, "learning_rate": 6.497087896228937e-10, "loss": 0.0186, "step": 14723 }, { "epoch": 14.94, "learning_rate": 6.24960524725493e-10, "loss": 0.0188, "step": 14724 }, { "epoch": 14.94, "learning_rate": 6.00692788834678e-10, "loss": 0.0187, "step": 14725 }, { "epoch": 14.94, "learning_rate": 5.769055831184034e-10, "loss": 0.016, "step": 14726 }, { "epoch": 14.94, "learning_rate": 5.535989087179782e-10, "loss": 0.0178, "step": 14727 }, { "epoch": 14.94, "learning_rate": 5.30772766754728e-10, "loss": 0.0199, "step": 14728 }, { "epoch": 14.95, "learning_rate": 5.084271583244427e-10, "loss": 0.0183, "step": 14729 }, { "epoch": 14.95, "learning_rate": 4.865620845018182e-10, "loss": 0.0185, "step": 14730 }, { "epoch": 14.95, "learning_rate": 4.6517754633712553e-10, "loss": 0.0176, "step": 14731 }, { "epoch": 14.95, "learning_rate": 4.442735448584312e-10, "loss": 0.0182, "step": 14732 }, { "epoch": 14.95, "learning_rate": 4.238500810704871e-10, "loss": 0.0177, "step": 14733 }, { "epoch": 14.95, "learning_rate": 4.039071559536201e-10, "loss": 0.0153, "step": 14734 }, { "epoch": 14.95, "learning_rate": 3.8444477046706285e-10, "loss": 0.0175, "step": 14735 }, { "epoch": 14.95, "learning_rate": 3.6546292554673345e-10, "loss": 0.0168, "step": 14736 }, { "epoch": 14.95, "learning_rate": 3.469616221030148e-10, "loss": 0.0166, "step": 14737 }, { "epoch": 14.95, "learning_rate": 3.2894086102630564e-10, "loss": 0.0143, "step": 14738 }, { "epoch": 14.96, "learning_rate": 3.1140064318258003e-10, "loss": 0.0175, "step": 14739 }, { "epoch": 14.96, "learning_rate": 2.943409694144972e-10, "loss": 0.0155, "step": 14740 }, { "epoch": 14.96, "learning_rate": 2.77761840542512e-10, "loss": 0.0169, "step": 14741 }, { "epoch": 14.96, "learning_rate": 2.6166325736154407e-10, "loss": 0.0169, "step": 14742 }, { "epoch": 14.96, "learning_rate": 2.4604522064763935e-10, "loss": 0.0168, "step": 14743 }, { "epoch": 14.96, "learning_rate": 2.3090773114908816e-10, "loss": 0.0176, "step": 14744 }, { "epoch": 14.96, "learning_rate": 2.1625078959530698e-10, "loss": 0.017, "step": 14745 }, { "epoch": 14.96, "learning_rate": 2.0207439668906703e-10, "loss": 0.0162, "step": 14746 }, { "epoch": 14.96, "learning_rate": 1.8837855311204522e-10, "loss": 0.0212, "step": 14747 }, { "epoch": 14.96, "learning_rate": 1.751632595226038e-10, "loss": 0.017, "step": 14748 }, { "epoch": 14.97, "learning_rate": 1.624285165569006e-10, "loss": 0.0175, "step": 14749 }, { "epoch": 14.97, "learning_rate": 1.5017432482555828e-10, "loss": 0.0155, "step": 14750 }, { "epoch": 14.97, "learning_rate": 1.38400684916995e-10, "loss": 0.0174, "step": 14751 }, { "epoch": 14.97, "learning_rate": 1.2710759739853473e-10, "loss": 0.0182, "step": 14752 }, { "epoch": 14.97, "learning_rate": 1.1629506281085612e-10, "loss": 0.0191, "step": 14753 }, { "epoch": 14.97, "learning_rate": 1.0596308167576396e-10, "loss": 0.0174, "step": 14754 }, { "epoch": 14.97, "learning_rate": 9.611165448952797e-11, "loss": 0.0174, "step": 14755 }, { "epoch": 14.97, "learning_rate": 8.674078172399291e-11, "loss": 0.0156, "step": 14756 }, { "epoch": 14.97, "learning_rate": 7.785046382990935e-11, "loss": 0.0172, "step": 14757 }, { "epoch": 14.98, "learning_rate": 6.944070123582336e-11, "loss": 0.016, "step": 14758 }, { "epoch": 14.98, "learning_rate": 6.15114943447459e-11, "loss": 0.0222, "step": 14759 }, { "epoch": 14.98, "learning_rate": 5.40628435385937e-11, "loss": 0.0176, "step": 14760 }, { "epoch": 14.98, "learning_rate": 4.709474917374834e-11, "loss": 0.0187, "step": 14761 }, { "epoch": 14.98, "learning_rate": 4.060721158660741e-11, "loss": 0.0217, "step": 14762 }, { "epoch": 14.98, "learning_rate": 3.460023108803334e-11, "loss": 0.0171, "step": 14763 }, { "epoch": 14.98, "learning_rate": 2.9073807966684132e-11, "loss": 0.0184, "step": 14764 }, { "epoch": 14.98, "learning_rate": 2.40279424890133e-11, "loss": 0.0211, "step": 14765 }, { "epoch": 14.98, "learning_rate": 1.9462634897049472e-11, "loss": 0.0151, "step": 14766 }, { "epoch": 14.98, "learning_rate": 1.53778854106168e-11, "loss": 0.0176, "step": 14767 }, { "epoch": 14.99, "learning_rate": 1.1773694224004318e-11, "loss": 0.0185, "step": 14768 }, { "epoch": 14.99, "learning_rate": 8.650061513737485e-12, "loss": 0.0185, "step": 14769 }, { "epoch": 14.99, "learning_rate": 6.00698742636574e-12, "loss": 0.0169, "step": 14770 }, { "epoch": 14.99, "learning_rate": 3.8444720917851785e-12, "loss": 0.0188, "step": 14771 }, { "epoch": 14.99, "learning_rate": 2.162515612136318e-12, "loss": 0.0179, "step": 14772 }, { "epoch": 14.99, "learning_rate": 9.611180695756617e-13, "loss": 0.0179, "step": 14773 }, { "epoch": 14.99, "learning_rate": 2.4027951961436147e-13, "loss": 0.0176, "step": 14774 }, { "epoch": 14.99, "learning_rate": 0.0, "loss": 0.0178, "step": 14775 } ], "logging_steps": 1.0, "max_steps": 14775, "num_input_tokens_seen": 0, "num_train_epochs": 15, "save_steps": 3.0, "total_flos": 189404615024640.0, "train_batch_size": 4, "trial_name": null, "trial_params": null }