{ "best_metric": null, "best_model_checkpoint": null, "epoch": 2.0, "eval_steps": 500, "global_step": 3216, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 2.061855670103093e-07, "loss": 1.0357, "step": 1 }, { "epoch": 0.0, "learning_rate": 4.123711340206186e-07, "loss": 1.0128, "step": 2 }, { "epoch": 0.0, "learning_rate": 6.185567010309279e-07, "loss": 1.0223, "step": 3 }, { "epoch": 0.0, "learning_rate": 8.247422680412372e-07, "loss": 1.0056, "step": 4 }, { "epoch": 0.0, "learning_rate": 1.0309278350515464e-06, "loss": 0.9542, "step": 5 }, { "epoch": 0.0, "learning_rate": 1.2371134020618557e-06, "loss": 0.9174, "step": 6 }, { "epoch": 0.0, "learning_rate": 1.4432989690721649e-06, "loss": 0.8934, "step": 7 }, { "epoch": 0.0, "learning_rate": 1.6494845360824744e-06, "loss": 0.6981, "step": 8 }, { "epoch": 0.01, "learning_rate": 1.8556701030927837e-06, "loss": 0.683, "step": 9 }, { "epoch": 0.01, "learning_rate": 2.061855670103093e-06, "loss": 0.6239, "step": 10 }, { "epoch": 0.01, "learning_rate": 2.268041237113402e-06, "loss": 0.3454, "step": 11 }, { "epoch": 0.01, "learning_rate": 2.4742268041237115e-06, "loss": 0.3368, "step": 12 }, { "epoch": 0.01, "learning_rate": 2.680412371134021e-06, "loss": 0.2977, "step": 13 }, { "epoch": 0.01, "learning_rate": 2.8865979381443297e-06, "loss": 0.1829, "step": 14 }, { "epoch": 0.01, "learning_rate": 3.0927835051546395e-06, "loss": 0.1403, "step": 15 }, { "epoch": 0.01, "learning_rate": 3.298969072164949e-06, "loss": 0.143, "step": 16 }, { "epoch": 0.01, "learning_rate": 3.5051546391752577e-06, "loss": 0.1318, "step": 17 }, { "epoch": 0.01, "learning_rate": 3.7113402061855674e-06, "loss": 0.1336, "step": 18 }, { "epoch": 0.01, "learning_rate": 3.917525773195877e-06, "loss": 0.1294, "step": 19 }, { "epoch": 0.01, "learning_rate": 4.123711340206186e-06, "loss": 0.1917, "step": 20 }, { "epoch": 0.01, "learning_rate": 4.329896907216495e-06, "loss": 0.1902, "step": 21 }, { "epoch": 0.01, "learning_rate": 4.536082474226804e-06, "loss": 0.1666, "step": 22 }, { "epoch": 0.01, "learning_rate": 4.742268041237113e-06, "loss": 0.1257, "step": 23 }, { "epoch": 0.01, "learning_rate": 4.948453608247423e-06, "loss": 0.1333, "step": 24 }, { "epoch": 0.02, "learning_rate": 5.154639175257732e-06, "loss": 0.1508, "step": 25 }, { "epoch": 0.02, "learning_rate": 5.360824742268042e-06, "loss": 0.1463, "step": 26 }, { "epoch": 0.02, "learning_rate": 5.567010309278351e-06, "loss": 0.1246, "step": 27 }, { "epoch": 0.02, "learning_rate": 5.7731958762886594e-06, "loss": 0.1235, "step": 28 }, { "epoch": 0.02, "learning_rate": 5.979381443298969e-06, "loss": 0.1304, "step": 29 }, { "epoch": 0.02, "learning_rate": 6.185567010309279e-06, "loss": 0.1176, "step": 30 }, { "epoch": 0.02, "learning_rate": 6.391752577319588e-06, "loss": 0.1216, "step": 31 }, { "epoch": 0.02, "learning_rate": 6.597938144329898e-06, "loss": 0.1341, "step": 32 }, { "epoch": 0.02, "learning_rate": 6.804123711340207e-06, "loss": 0.1461, "step": 33 }, { "epoch": 0.02, "learning_rate": 7.010309278350515e-06, "loss": 0.1247, "step": 34 }, { "epoch": 0.02, "learning_rate": 7.216494845360825e-06, "loss": 0.1212, "step": 35 }, { "epoch": 0.02, "learning_rate": 7.422680412371135e-06, "loss": 0.1163, "step": 36 }, { "epoch": 0.02, "learning_rate": 7.628865979381444e-06, "loss": 0.1306, "step": 37 }, { "epoch": 0.02, "learning_rate": 7.835051546391754e-06, "loss": 0.1217, "step": 38 }, { "epoch": 0.02, "learning_rate": 8.041237113402063e-06, "loss": 0.1319, "step": 39 }, { "epoch": 0.02, "learning_rate": 8.247422680412371e-06, "loss": 0.1184, "step": 40 }, { "epoch": 0.03, "learning_rate": 8.453608247422681e-06, "loss": 0.1182, "step": 41 }, { "epoch": 0.03, "learning_rate": 8.65979381443299e-06, "loss": 0.1196, "step": 42 }, { "epoch": 0.03, "learning_rate": 8.865979381443299e-06, "loss": 0.1182, "step": 43 }, { "epoch": 0.03, "learning_rate": 9.072164948453609e-06, "loss": 0.1193, "step": 44 }, { "epoch": 0.03, "learning_rate": 9.278350515463918e-06, "loss": 0.1161, "step": 45 }, { "epoch": 0.03, "learning_rate": 9.484536082474226e-06, "loss": 0.1171, "step": 46 }, { "epoch": 0.03, "learning_rate": 9.690721649484536e-06, "loss": 0.1193, "step": 47 }, { "epoch": 0.03, "learning_rate": 9.896907216494846e-06, "loss": 0.1198, "step": 48 }, { "epoch": 0.03, "learning_rate": 1.0103092783505156e-05, "loss": 0.118, "step": 49 }, { "epoch": 0.03, "learning_rate": 1.0309278350515464e-05, "loss": 0.1165, "step": 50 }, { "epoch": 0.03, "learning_rate": 1.0515463917525775e-05, "loss": 0.1161, "step": 51 }, { "epoch": 0.03, "learning_rate": 1.0721649484536083e-05, "loss": 0.1168, "step": 52 }, { "epoch": 0.03, "learning_rate": 1.0927835051546391e-05, "loss": 0.1148, "step": 53 }, { "epoch": 0.03, "learning_rate": 1.1134020618556703e-05, "loss": 0.1124, "step": 54 }, { "epoch": 0.03, "learning_rate": 1.134020618556701e-05, "loss": 0.1152, "step": 55 }, { "epoch": 0.03, "learning_rate": 1.1546391752577319e-05, "loss": 0.1127, "step": 56 }, { "epoch": 0.04, "learning_rate": 1.175257731958763e-05, "loss": 0.1263, "step": 57 }, { "epoch": 0.04, "learning_rate": 1.1958762886597938e-05, "loss": 0.121, "step": 58 }, { "epoch": 0.04, "learning_rate": 1.2164948453608248e-05, "loss": 0.115, "step": 59 }, { "epoch": 0.04, "learning_rate": 1.2371134020618558e-05, "loss": 0.1164, "step": 60 }, { "epoch": 0.04, "learning_rate": 1.2577319587628866e-05, "loss": 0.119, "step": 61 }, { "epoch": 0.04, "learning_rate": 1.2783505154639176e-05, "loss": 0.1132, "step": 62 }, { "epoch": 0.04, "learning_rate": 1.2989690721649485e-05, "loss": 0.1131, "step": 63 }, { "epoch": 0.04, "learning_rate": 1.3195876288659795e-05, "loss": 0.1175, "step": 64 }, { "epoch": 0.04, "learning_rate": 1.3402061855670103e-05, "loss": 0.1244, "step": 65 }, { "epoch": 0.04, "learning_rate": 1.3608247422680415e-05, "loss": 0.117, "step": 66 }, { "epoch": 0.04, "learning_rate": 1.3814432989690723e-05, "loss": 0.1161, "step": 67 }, { "epoch": 0.04, "learning_rate": 1.402061855670103e-05, "loss": 0.1204, "step": 68 }, { "epoch": 0.04, "learning_rate": 1.4226804123711342e-05, "loss": 0.1124, "step": 69 }, { "epoch": 0.04, "learning_rate": 1.443298969072165e-05, "loss": 0.11, "step": 70 }, { "epoch": 0.04, "learning_rate": 1.4639175257731958e-05, "loss": 0.1237, "step": 71 }, { "epoch": 0.04, "learning_rate": 1.484536082474227e-05, "loss": 0.1276, "step": 72 }, { "epoch": 0.05, "learning_rate": 1.5051546391752578e-05, "loss": 0.1186, "step": 73 }, { "epoch": 0.05, "learning_rate": 1.5257731958762888e-05, "loss": 0.1213, "step": 74 }, { "epoch": 0.05, "learning_rate": 1.5463917525773197e-05, "loss": 0.1143, "step": 75 }, { "epoch": 0.05, "learning_rate": 1.5670103092783507e-05, "loss": 0.1161, "step": 76 }, { "epoch": 0.05, "learning_rate": 1.5876288659793813e-05, "loss": 0.1122, "step": 77 }, { "epoch": 0.05, "learning_rate": 1.6082474226804127e-05, "loss": 0.1153, "step": 78 }, { "epoch": 0.05, "learning_rate": 1.6288659793814433e-05, "loss": 0.1171, "step": 79 }, { "epoch": 0.05, "learning_rate": 1.6494845360824743e-05, "loss": 0.1049, "step": 80 }, { "epoch": 0.05, "learning_rate": 1.6701030927835052e-05, "loss": 0.1039, "step": 81 }, { "epoch": 0.05, "learning_rate": 1.6907216494845362e-05, "loss": 0.1093, "step": 82 }, { "epoch": 0.05, "learning_rate": 1.7113402061855672e-05, "loss": 0.0987, "step": 83 }, { "epoch": 0.05, "learning_rate": 1.731958762886598e-05, "loss": 0.1004, "step": 84 }, { "epoch": 0.05, "learning_rate": 1.752577319587629e-05, "loss": 0.1037, "step": 85 }, { "epoch": 0.05, "learning_rate": 1.7731958762886598e-05, "loss": 0.1075, "step": 86 }, { "epoch": 0.05, "learning_rate": 1.793814432989691e-05, "loss": 0.1003, "step": 87 }, { "epoch": 0.05, "learning_rate": 1.8144329896907217e-05, "loss": 0.0957, "step": 88 }, { "epoch": 0.06, "learning_rate": 1.8350515463917527e-05, "loss": 0.1033, "step": 89 }, { "epoch": 0.06, "learning_rate": 1.8556701030927837e-05, "loss": 0.0942, "step": 90 }, { "epoch": 0.06, "learning_rate": 1.8762886597938147e-05, "loss": 0.0859, "step": 91 }, { "epoch": 0.06, "learning_rate": 1.8969072164948453e-05, "loss": 0.0876, "step": 92 }, { "epoch": 0.06, "learning_rate": 1.9175257731958766e-05, "loss": 0.0873, "step": 93 }, { "epoch": 0.06, "learning_rate": 1.9381443298969072e-05, "loss": 0.0839, "step": 94 }, { "epoch": 0.06, "learning_rate": 1.9587628865979382e-05, "loss": 0.0808, "step": 95 }, { "epoch": 0.06, "learning_rate": 1.9793814432989692e-05, "loss": 0.0924, "step": 96 }, { "epoch": 0.06, "learning_rate": 2e-05, "loss": 0.0863, "step": 97 }, { "epoch": 0.06, "learning_rate": 1.9999994927302518e-05, "loss": 0.0826, "step": 98 }, { "epoch": 0.06, "learning_rate": 1.9999979709215212e-05, "loss": 0.0902, "step": 99 }, { "epoch": 0.06, "learning_rate": 1.9999954345753522e-05, "loss": 0.0868, "step": 100 }, { "epoch": 0.06, "learning_rate": 1.999991883694318e-05, "loss": 0.0824, "step": 101 }, { "epoch": 0.06, "learning_rate": 1.9999873182820214e-05, "loss": 0.0739, "step": 102 }, { "epoch": 0.06, "learning_rate": 1.999981738343094e-05, "loss": 0.0841, "step": 103 }, { "epoch": 0.06, "learning_rate": 1.9999751438831965e-05, "loss": 0.0785, "step": 104 }, { "epoch": 0.07, "learning_rate": 1.99996753490902e-05, "loss": 0.0768, "step": 105 }, { "epoch": 0.07, "learning_rate": 1.9999589114282834e-05, "loss": 0.081, "step": 106 }, { "epoch": 0.07, "learning_rate": 1.9999492734497365e-05, "loss": 0.0737, "step": 107 }, { "epoch": 0.07, "learning_rate": 1.999938620983156e-05, "loss": 0.0812, "step": 108 }, { "epoch": 0.07, "learning_rate": 1.9999269540393507e-05, "loss": 0.0708, "step": 109 }, { "epoch": 0.07, "learning_rate": 1.999914272630156e-05, "loss": 0.0654, "step": 110 }, { "epoch": 0.07, "learning_rate": 1.9999005767684388e-05, "loss": 0.085, "step": 111 }, { "epoch": 0.07, "learning_rate": 1.999885866468093e-05, "loss": 0.0903, "step": 112 }, { "epoch": 0.07, "learning_rate": 1.9998701417440433e-05, "loss": 0.0698, "step": 113 }, { "epoch": 0.07, "learning_rate": 1.9998534026122433e-05, "loss": 0.072, "step": 114 }, { "epoch": 0.07, "learning_rate": 1.999835649089675e-05, "loss": 0.0708, "step": 115 }, { "epoch": 0.07, "learning_rate": 1.9998168811943506e-05, "loss": 0.0773, "step": 116 }, { "epoch": 0.07, "learning_rate": 1.9997970989453103e-05, "loss": 0.0663, "step": 117 }, { "epoch": 0.07, "learning_rate": 1.9997763023626245e-05, "loss": 0.0836, "step": 118 }, { "epoch": 0.07, "learning_rate": 1.9997544914673915e-05, "loss": 0.0661, "step": 119 }, { "epoch": 0.07, "learning_rate": 1.9997316662817403e-05, "loss": 0.0727, "step": 120 }, { "epoch": 0.08, "learning_rate": 1.999707826828827e-05, "loss": 0.0788, "step": 121 }, { "epoch": 0.08, "learning_rate": 1.9996829731328376e-05, "loss": 0.067, "step": 122 }, { "epoch": 0.08, "learning_rate": 1.9996571052189882e-05, "loss": 0.0841, "step": 123 }, { "epoch": 0.08, "learning_rate": 1.999630223113522e-05, "loss": 0.078, "step": 124 }, { "epoch": 0.08, "learning_rate": 1.999602326843712e-05, "loss": 0.0641, "step": 125 }, { "epoch": 0.08, "learning_rate": 1.9995734164378604e-05, "loss": 0.0738, "step": 126 }, { "epoch": 0.08, "learning_rate": 1.999543491925298e-05, "loss": 0.0623, "step": 127 }, { "epoch": 0.08, "learning_rate": 1.9995125533363836e-05, "loss": 0.0709, "step": 128 }, { "epoch": 0.08, "learning_rate": 1.9994806007025068e-05, "loss": 0.0679, "step": 129 }, { "epoch": 0.08, "learning_rate": 1.999447634056084e-05, "loss": 0.0589, "step": 130 }, { "epoch": 0.08, "learning_rate": 1.999413653430561e-05, "loss": 0.0645, "step": 131 }, { "epoch": 0.08, "learning_rate": 1.999378658860413e-05, "loss": 0.0786, "step": 132 }, { "epoch": 0.08, "learning_rate": 1.9993426503811438e-05, "loss": 0.0662, "step": 133 }, { "epoch": 0.08, "learning_rate": 1.9993056280292845e-05, "loss": 0.0562, "step": 134 }, { "epoch": 0.08, "learning_rate": 1.9992675918423958e-05, "loss": 0.0724, "step": 135 }, { "epoch": 0.08, "learning_rate": 1.9992285418590676e-05, "loss": 0.0657, "step": 136 }, { "epoch": 0.09, "learning_rate": 1.999188478118917e-05, "loss": 0.06, "step": 137 }, { "epoch": 0.09, "learning_rate": 1.9991474006625907e-05, "loss": 0.0667, "step": 138 }, { "epoch": 0.09, "learning_rate": 1.999105309531763e-05, "loss": 0.0686, "step": 139 }, { "epoch": 0.09, "learning_rate": 1.9990622047691372e-05, "loss": 0.0616, "step": 140 }, { "epoch": 0.09, "learning_rate": 1.9990180864184452e-05, "loss": 0.0617, "step": 141 }, { "epoch": 0.09, "learning_rate": 1.998972954524446e-05, "loss": 0.0623, "step": 142 }, { "epoch": 0.09, "learning_rate": 1.9989268091329282e-05, "loss": 0.0591, "step": 143 }, { "epoch": 0.09, "learning_rate": 1.9988796502907083e-05, "loss": 0.0682, "step": 144 }, { "epoch": 0.09, "learning_rate": 1.9988314780456305e-05, "loss": 0.0684, "step": 145 }, { "epoch": 0.09, "learning_rate": 1.998782292446567e-05, "loss": 0.0641, "step": 146 }, { "epoch": 0.09, "learning_rate": 1.9987320935434194e-05, "loss": 0.0613, "step": 147 }, { "epoch": 0.09, "learning_rate": 1.9986808813871162e-05, "loss": 0.0695, "step": 148 }, { "epoch": 0.09, "learning_rate": 1.9986286560296134e-05, "loss": 0.0647, "step": 149 }, { "epoch": 0.09, "learning_rate": 1.9985754175238972e-05, "loss": 0.0651, "step": 150 }, { "epoch": 0.09, "learning_rate": 1.9985211659239786e-05, "loss": 0.0645, "step": 151 }, { "epoch": 0.09, "learning_rate": 1.998465901284899e-05, "loss": 0.0595, "step": 152 }, { "epoch": 0.1, "learning_rate": 1.9984096236627263e-05, "loss": 0.0596, "step": 153 }, { "epoch": 0.1, "learning_rate": 1.998352333114556e-05, "loss": 0.0706, "step": 154 }, { "epoch": 0.1, "learning_rate": 1.9982940296985124e-05, "loss": 0.0563, "step": 155 }, { "epoch": 0.1, "learning_rate": 1.9982347134737455e-05, "loss": 0.0552, "step": 156 }, { "epoch": 0.1, "learning_rate": 1.9981743845004354e-05, "loss": 0.0599, "step": 157 }, { "epoch": 0.1, "learning_rate": 1.998113042839787e-05, "loss": 0.0623, "step": 158 }, { "epoch": 0.1, "learning_rate": 1.998050688554034e-05, "loss": 0.0676, "step": 159 }, { "epoch": 0.1, "learning_rate": 1.997987321706438e-05, "loss": 0.0588, "step": 160 }, { "epoch": 0.1, "learning_rate": 1.9979229423612865e-05, "loss": 0.0625, "step": 161 }, { "epoch": 0.1, "learning_rate": 1.9978575505838952e-05, "loss": 0.0599, "step": 162 }, { "epoch": 0.1, "learning_rate": 1.9977911464406068e-05, "loss": 0.0661, "step": 163 }, { "epoch": 0.1, "learning_rate": 1.9977237299987903e-05, "loss": 0.0609, "step": 164 }, { "epoch": 0.1, "learning_rate": 1.9976553013268428e-05, "loss": 0.0646, "step": 165 }, { "epoch": 0.1, "learning_rate": 1.997585860494188e-05, "loss": 0.0684, "step": 166 }, { "epoch": 0.1, "learning_rate": 1.997515407571276e-05, "loss": 0.0622, "step": 167 }, { "epoch": 0.1, "learning_rate": 1.9974439426295843e-05, "loss": 0.0571, "step": 168 }, { "epoch": 0.11, "learning_rate": 1.997371465741617e-05, "loss": 0.052, "step": 169 }, { "epoch": 0.11, "learning_rate": 1.9972979769809044e-05, "loss": 0.0592, "step": 170 }, { "epoch": 0.11, "learning_rate": 1.997223476422004e-05, "loss": 0.0596, "step": 171 }, { "epoch": 0.11, "learning_rate": 1.9971479641405e-05, "loss": 0.0563, "step": 172 }, { "epoch": 0.11, "learning_rate": 1.9970714402130017e-05, "loss": 0.0604, "step": 173 }, { "epoch": 0.11, "learning_rate": 1.996993904717146e-05, "loss": 0.056, "step": 174 }, { "epoch": 0.11, "learning_rate": 1.996915357731596e-05, "loss": 0.0614, "step": 175 }, { "epoch": 0.11, "learning_rate": 1.9968357993360405e-05, "loss": 0.0538, "step": 176 }, { "epoch": 0.11, "learning_rate": 1.9967552296111945e-05, "loss": 0.0526, "step": 177 }, { "epoch": 0.11, "learning_rate": 1.9966736486387997e-05, "loss": 0.0573, "step": 178 }, { "epoch": 0.11, "learning_rate": 1.9965910565016223e-05, "loss": 0.0706, "step": 179 }, { "epoch": 0.11, "learning_rate": 1.996507453283456e-05, "loss": 0.0556, "step": 180 }, { "epoch": 0.11, "learning_rate": 1.9964228390691197e-05, "loss": 0.0612, "step": 181 }, { "epoch": 0.11, "learning_rate": 1.9963372139444574e-05, "loss": 0.0499, "step": 182 }, { "epoch": 0.11, "learning_rate": 1.9962505779963392e-05, "loss": 0.0493, "step": 183 }, { "epoch": 0.11, "learning_rate": 1.9961629313126608e-05, "loss": 0.0538, "step": 184 }, { "epoch": 0.12, "learning_rate": 1.9960742739823433e-05, "loss": 0.053, "step": 185 }, { "epoch": 0.12, "learning_rate": 1.9959846060953332e-05, "loss": 0.0603, "step": 186 }, { "epoch": 0.12, "learning_rate": 1.9958939277426017e-05, "loss": 0.0629, "step": 187 }, { "epoch": 0.12, "learning_rate": 1.995802239016146e-05, "loss": 0.0505, "step": 188 }, { "epoch": 0.12, "learning_rate": 1.9957095400089875e-05, "loss": 0.0628, "step": 189 }, { "epoch": 0.12, "learning_rate": 1.995615830815173e-05, "loss": 0.0598, "step": 190 }, { "epoch": 0.12, "learning_rate": 1.995521111529775e-05, "loss": 0.059, "step": 191 }, { "epoch": 0.12, "learning_rate": 1.9954253822488888e-05, "loss": 0.0551, "step": 192 }, { "epoch": 0.12, "learning_rate": 1.9953286430696365e-05, "loss": 0.0532, "step": 193 }, { "epoch": 0.12, "learning_rate": 1.9952308940901634e-05, "loss": 0.0557, "step": 194 }, { "epoch": 0.12, "learning_rate": 1.9951321354096395e-05, "loss": 0.0586, "step": 195 }, { "epoch": 0.12, "learning_rate": 1.9950323671282597e-05, "loss": 0.0487, "step": 196 }, { "epoch": 0.12, "learning_rate": 1.9949315893472426e-05, "loss": 0.0597, "step": 197 }, { "epoch": 0.12, "learning_rate": 1.9948298021688314e-05, "loss": 0.0617, "step": 198 }, { "epoch": 0.12, "learning_rate": 1.9947270056962934e-05, "loss": 0.0518, "step": 199 }, { "epoch": 0.12, "learning_rate": 1.9946232000339192e-05, "loss": 0.0623, "step": 200 }, { "epoch": 0.12, "learning_rate": 1.9945183852870245e-05, "loss": 0.0593, "step": 201 }, { "epoch": 0.13, "learning_rate": 1.9944125615619473e-05, "loss": 0.0518, "step": 202 }, { "epoch": 0.13, "learning_rate": 1.9943057289660502e-05, "loss": 0.0541, "step": 203 }, { "epoch": 0.13, "learning_rate": 1.994197887607719e-05, "loss": 0.0607, "step": 204 }, { "epoch": 0.13, "learning_rate": 1.9940890375963634e-05, "loss": 0.0578, "step": 205 }, { "epoch": 0.13, "learning_rate": 1.9939791790424154e-05, "loss": 0.0542, "step": 206 }, { "epoch": 0.13, "learning_rate": 1.9938683120573313e-05, "loss": 0.0607, "step": 207 }, { "epoch": 0.13, "learning_rate": 1.99375643675359e-05, "loss": 0.0572, "step": 208 }, { "epoch": 0.13, "learning_rate": 1.993643553244693e-05, "loss": 0.049, "step": 209 }, { "epoch": 0.13, "learning_rate": 1.9935296616451654e-05, "loss": 0.06, "step": 210 }, { "epoch": 0.13, "learning_rate": 1.993414762070555e-05, "loss": 0.0652, "step": 211 }, { "epoch": 0.13, "learning_rate": 1.9932988546374314e-05, "loss": 0.0483, "step": 212 }, { "epoch": 0.13, "learning_rate": 1.9931819394633876e-05, "loss": 0.0575, "step": 213 }, { "epoch": 0.13, "learning_rate": 1.993064016667039e-05, "loss": 0.057, "step": 214 }, { "epoch": 0.13, "learning_rate": 1.992945086368022e-05, "loss": 0.0626, "step": 215 }, { "epoch": 0.13, "learning_rate": 1.992825148686997e-05, "loss": 0.0503, "step": 216 }, { "epoch": 0.13, "learning_rate": 1.9927042037456453e-05, "loss": 0.0519, "step": 217 }, { "epoch": 0.14, "learning_rate": 1.9925822516666694e-05, "loss": 0.0498, "step": 218 }, { "epoch": 0.14, "learning_rate": 1.992459292573796e-05, "loss": 0.0477, "step": 219 }, { "epoch": 0.14, "learning_rate": 1.992335326591771e-05, "loss": 0.0534, "step": 220 }, { "epoch": 0.14, "learning_rate": 1.992210353846363e-05, "loss": 0.0501, "step": 221 }, { "epoch": 0.14, "learning_rate": 1.992084374464362e-05, "loss": 0.0478, "step": 222 }, { "epoch": 0.14, "learning_rate": 1.9919573885735784e-05, "loss": 0.0498, "step": 223 }, { "epoch": 0.14, "learning_rate": 1.991829396302845e-05, "loss": 0.0516, "step": 224 }, { "epoch": 0.14, "learning_rate": 1.9917003977820154e-05, "loss": 0.0553, "step": 225 }, { "epoch": 0.14, "learning_rate": 1.9915703931419626e-05, "loss": 0.0612, "step": 226 }, { "epoch": 0.14, "learning_rate": 1.991439382514582e-05, "loss": 0.0549, "step": 227 }, { "epoch": 0.14, "learning_rate": 1.9913073660327892e-05, "loss": 0.0514, "step": 228 }, { "epoch": 0.14, "learning_rate": 1.9911743438305203e-05, "loss": 0.0533, "step": 229 }, { "epoch": 0.14, "learning_rate": 1.9910403160427308e-05, "loss": 0.0535, "step": 230 }, { "epoch": 0.14, "learning_rate": 1.9909052828053976e-05, "loss": 0.0481, "step": 231 }, { "epoch": 0.14, "learning_rate": 1.9907692442555176e-05, "loss": 0.0515, "step": 232 }, { "epoch": 0.14, "learning_rate": 1.9906322005311072e-05, "loss": 0.0586, "step": 233 }, { "epoch": 0.15, "learning_rate": 1.990494151771202e-05, "loss": 0.0536, "step": 234 }, { "epoch": 0.15, "learning_rate": 1.990355098115859e-05, "loss": 0.0581, "step": 235 }, { "epoch": 0.15, "learning_rate": 1.9902150397061524e-05, "loss": 0.0529, "step": 236 }, { "epoch": 0.15, "learning_rate": 1.9900739766841777e-05, "loss": 0.055, "step": 237 }, { "epoch": 0.15, "learning_rate": 1.9899319091930488e-05, "loss": 0.0553, "step": 238 }, { "epoch": 0.15, "learning_rate": 1.989788837376899e-05, "loss": 0.0506, "step": 239 }, { "epoch": 0.15, "learning_rate": 1.9896447613808802e-05, "loss": 0.0449, "step": 240 }, { "epoch": 0.15, "learning_rate": 1.9894996813511626e-05, "loss": 0.0613, "step": 241 }, { "epoch": 0.15, "learning_rate": 1.9893535974349363e-05, "loss": 0.0484, "step": 242 }, { "epoch": 0.15, "learning_rate": 1.989206509780409e-05, "loss": 0.0501, "step": 243 }, { "epoch": 0.15, "learning_rate": 1.989058418536807e-05, "loss": 0.0506, "step": 244 }, { "epoch": 0.15, "learning_rate": 1.9889093238543746e-05, "loss": 0.0496, "step": 245 }, { "epoch": 0.15, "learning_rate": 1.9887592258843745e-05, "loss": 0.0494, "step": 246 }, { "epoch": 0.15, "learning_rate": 1.9886081247790867e-05, "loss": 0.0532, "step": 247 }, { "epoch": 0.15, "learning_rate": 1.9884560206918094e-05, "loss": 0.0521, "step": 248 }, { "epoch": 0.15, "learning_rate": 1.988302913776858e-05, "loss": 0.052, "step": 249 }, { "epoch": 0.16, "learning_rate": 1.988148804189566e-05, "loss": 0.0489, "step": 250 }, { "epoch": 0.16, "learning_rate": 1.987993692086283e-05, "loss": 0.0589, "step": 251 }, { "epoch": 0.16, "learning_rate": 1.987837577624377e-05, "loss": 0.0513, "step": 252 }, { "epoch": 0.16, "learning_rate": 1.9876804609622314e-05, "loss": 0.0477, "step": 253 }, { "epoch": 0.16, "learning_rate": 1.9875223422592485e-05, "loss": 0.0523, "step": 254 }, { "epoch": 0.16, "learning_rate": 1.9873632216758448e-05, "loss": 0.0574, "step": 255 }, { "epoch": 0.16, "learning_rate": 1.9872030993734555e-05, "loss": 0.0479, "step": 256 }, { "epoch": 0.16, "learning_rate": 1.98704197551453e-05, "loss": 0.0517, "step": 257 }, { "epoch": 0.16, "learning_rate": 1.9868798502625353e-05, "loss": 0.0493, "step": 258 }, { "epoch": 0.16, "learning_rate": 1.986716723781954e-05, "loss": 0.0492, "step": 259 }, { "epoch": 0.16, "learning_rate": 1.986552596238284e-05, "loss": 0.046, "step": 260 }, { "epoch": 0.16, "learning_rate": 1.9863874677980394e-05, "loss": 0.049, "step": 261 }, { "epoch": 0.16, "learning_rate": 1.9862213386287495e-05, "loss": 0.0566, "step": 262 }, { "epoch": 0.16, "learning_rate": 1.986054208898959e-05, "loss": 0.0492, "step": 263 }, { "epoch": 0.16, "learning_rate": 1.985886078778227e-05, "loss": 0.0522, "step": 264 }, { "epoch": 0.16, "learning_rate": 1.985716948437129e-05, "loss": 0.0406, "step": 265 }, { "epoch": 0.17, "learning_rate": 1.9855468180472542e-05, "loss": 0.0514, "step": 266 }, { "epoch": 0.17, "learning_rate": 1.9853756877812064e-05, "loss": 0.0522, "step": 267 }, { "epoch": 0.17, "learning_rate": 1.9852035578126036e-05, "loss": 0.0485, "step": 268 }, { "epoch": 0.17, "learning_rate": 1.9850304283160793e-05, "loss": 0.0467, "step": 269 }, { "epoch": 0.17, "learning_rate": 1.9848562994672797e-05, "loss": 0.0555, "step": 270 }, { "epoch": 0.17, "learning_rate": 1.984681171442866e-05, "loss": 0.0475, "step": 271 }, { "epoch": 0.17, "learning_rate": 1.9845050444205115e-05, "loss": 0.0522, "step": 272 }, { "epoch": 0.17, "learning_rate": 1.9843279185789046e-05, "loss": 0.056, "step": 273 }, { "epoch": 0.17, "learning_rate": 1.9841497940977464e-05, "loss": 0.0487, "step": 274 }, { "epoch": 0.17, "learning_rate": 1.9839706711577514e-05, "loss": 0.0492, "step": 275 }, { "epoch": 0.17, "learning_rate": 1.9837905499406467e-05, "loss": 0.0396, "step": 276 }, { "epoch": 0.17, "learning_rate": 1.9836094306291723e-05, "loss": 0.0421, "step": 277 }, { "epoch": 0.17, "learning_rate": 1.983427313407081e-05, "loss": 0.0477, "step": 278 }, { "epoch": 0.17, "learning_rate": 1.983244198459138e-05, "loss": 0.0405, "step": 279 }, { "epoch": 0.17, "learning_rate": 1.9830600859711207e-05, "loss": 0.0431, "step": 280 }, { "epoch": 0.17, "learning_rate": 1.982874976129818e-05, "loss": 0.0453, "step": 281 }, { "epoch": 0.18, "learning_rate": 1.982688869123032e-05, "loss": 0.056, "step": 282 }, { "epoch": 0.18, "learning_rate": 1.9825017651395747e-05, "loss": 0.0397, "step": 283 }, { "epoch": 0.18, "learning_rate": 1.982313664369271e-05, "loss": 0.0515, "step": 284 }, { "epoch": 0.18, "learning_rate": 1.9821245670029566e-05, "loss": 0.0476, "step": 285 }, { "epoch": 0.18, "learning_rate": 1.9819344732324776e-05, "loss": 0.0374, "step": 286 }, { "epoch": 0.18, "learning_rate": 1.9817433832506926e-05, "loss": 0.0417, "step": 287 }, { "epoch": 0.18, "learning_rate": 1.9815512972514695e-05, "loss": 0.039, "step": 288 }, { "epoch": 0.18, "learning_rate": 1.981358215429687e-05, "loss": 0.051, "step": 289 }, { "epoch": 0.18, "learning_rate": 1.9811641379812342e-05, "loss": 0.0407, "step": 290 }, { "epoch": 0.18, "learning_rate": 1.9809690651030106e-05, "loss": 0.0533, "step": 291 }, { "epoch": 0.18, "learning_rate": 1.9807729969929255e-05, "loss": 0.058, "step": 292 }, { "epoch": 0.18, "learning_rate": 1.980575933849897e-05, "loss": 0.0438, "step": 293 }, { "epoch": 0.18, "learning_rate": 1.9803778758738543e-05, "loss": 0.0593, "step": 294 }, { "epoch": 0.18, "learning_rate": 1.9801788232657343e-05, "loss": 0.0518, "step": 295 }, { "epoch": 0.18, "learning_rate": 1.9799787762274844e-05, "loss": 0.0461, "step": 296 }, { "epoch": 0.18, "learning_rate": 1.9797777349620597e-05, "loss": 0.0522, "step": 297 }, { "epoch": 0.19, "learning_rate": 1.9795756996734248e-05, "loss": 0.0477, "step": 298 }, { "epoch": 0.19, "learning_rate": 1.9793726705665524e-05, "loss": 0.0483, "step": 299 }, { "epoch": 0.19, "learning_rate": 1.9791686478474235e-05, "loss": 0.0426, "step": 300 }, { "epoch": 0.19, "learning_rate": 1.978963631723027e-05, "loss": 0.0521, "step": 301 }, { "epoch": 0.19, "learning_rate": 1.9787576224013602e-05, "loss": 0.0445, "step": 302 }, { "epoch": 0.19, "learning_rate": 1.9785506200914276e-05, "loss": 0.037, "step": 303 }, { "epoch": 0.19, "learning_rate": 1.9783426250032412e-05, "loss": 0.0429, "step": 304 }, { "epoch": 0.19, "learning_rate": 1.9781336373478207e-05, "loss": 0.0432, "step": 305 }, { "epoch": 0.19, "learning_rate": 1.977923657337191e-05, "loss": 0.0415, "step": 306 }, { "epoch": 0.19, "learning_rate": 1.9777126851843866e-05, "loss": 0.0444, "step": 307 }, { "epoch": 0.19, "learning_rate": 1.9775007211034466e-05, "loss": 0.0531, "step": 308 }, { "epoch": 0.19, "learning_rate": 1.9772877653094165e-05, "loss": 0.0494, "step": 309 }, { "epoch": 0.19, "learning_rate": 1.9770738180183485e-05, "loss": 0.0622, "step": 310 }, { "epoch": 0.19, "learning_rate": 1.976858879447301e-05, "loss": 0.0452, "step": 311 }, { "epoch": 0.19, "learning_rate": 1.976642949814337e-05, "loss": 0.0488, "step": 312 }, { "epoch": 0.19, "learning_rate": 1.9764260293385264e-05, "loss": 0.0476, "step": 313 }, { "epoch": 0.2, "learning_rate": 1.9762081182399434e-05, "loss": 0.0574, "step": 314 }, { "epoch": 0.2, "learning_rate": 1.975989216739667e-05, "loss": 0.043, "step": 315 }, { "epoch": 0.2, "learning_rate": 1.9757693250597818e-05, "loss": 0.0426, "step": 316 }, { "epoch": 0.2, "learning_rate": 1.975548443423376e-05, "loss": 0.0547, "step": 317 }, { "epoch": 0.2, "learning_rate": 1.975326572054544e-05, "loss": 0.0561, "step": 318 }, { "epoch": 0.2, "learning_rate": 1.9751037111783818e-05, "loss": 0.0467, "step": 319 }, { "epoch": 0.2, "learning_rate": 1.9748798610209912e-05, "loss": 0.0523, "step": 320 }, { "epoch": 0.2, "learning_rate": 1.974655021809477e-05, "loss": 0.0526, "step": 321 }, { "epoch": 0.2, "learning_rate": 1.9744291937719468e-05, "loss": 0.0488, "step": 322 }, { "epoch": 0.2, "learning_rate": 1.974202377137513e-05, "loss": 0.0529, "step": 323 }, { "epoch": 0.2, "learning_rate": 1.9739745721362897e-05, "loss": 0.049, "step": 324 }, { "epoch": 0.2, "learning_rate": 1.9737457789993936e-05, "loss": 0.0454, "step": 325 }, { "epoch": 0.2, "learning_rate": 1.973515997958945e-05, "loss": 0.0624, "step": 326 }, { "epoch": 0.2, "learning_rate": 1.9732852292480657e-05, "loss": 0.0466, "step": 327 }, { "epoch": 0.2, "learning_rate": 1.9730534731008793e-05, "loss": 0.0512, "step": 328 }, { "epoch": 0.2, "learning_rate": 1.9728207297525125e-05, "loss": 0.064, "step": 329 }, { "epoch": 0.21, "learning_rate": 1.9725869994390912e-05, "loss": 0.0511, "step": 330 }, { "epoch": 0.21, "learning_rate": 1.972352282397745e-05, "loss": 0.0547, "step": 331 }, { "epoch": 0.21, "learning_rate": 1.9721165788666036e-05, "loss": 0.0519, "step": 332 }, { "epoch": 0.21, "learning_rate": 1.9718798890847976e-05, "loss": 0.0476, "step": 333 }, { "epoch": 0.21, "learning_rate": 1.9716422132924572e-05, "loss": 0.049, "step": 334 }, { "epoch": 0.21, "learning_rate": 1.971403551730715e-05, "loss": 0.0514, "step": 335 }, { "epoch": 0.21, "learning_rate": 1.971163904641702e-05, "loss": 0.047, "step": 336 }, { "epoch": 0.21, "learning_rate": 1.97092327226855e-05, "loss": 0.0446, "step": 337 }, { "epoch": 0.21, "learning_rate": 1.9706816548553895e-05, "loss": 0.044, "step": 338 }, { "epoch": 0.21, "learning_rate": 1.9704390526473515e-05, "loss": 0.0395, "step": 339 }, { "epoch": 0.21, "learning_rate": 1.9701954658905652e-05, "loss": 0.0421, "step": 340 }, { "epoch": 0.21, "learning_rate": 1.969950894832159e-05, "loss": 0.0416, "step": 341 }, { "epoch": 0.21, "learning_rate": 1.96970533972026e-05, "loss": 0.0505, "step": 342 }, { "epoch": 0.21, "learning_rate": 1.9694588008039935e-05, "loss": 0.0474, "step": 343 }, { "epoch": 0.21, "learning_rate": 1.9692112783334826e-05, "loss": 0.0536, "step": 344 }, { "epoch": 0.21, "learning_rate": 1.9689627725598495e-05, "loss": 0.0463, "step": 345 }, { "epoch": 0.22, "learning_rate": 1.9687132837352124e-05, "loss": 0.0523, "step": 346 }, { "epoch": 0.22, "learning_rate": 1.9684628121126878e-05, "loss": 0.0432, "step": 347 }, { "epoch": 0.22, "learning_rate": 1.9682113579463894e-05, "loss": 0.0448, "step": 348 }, { "epoch": 0.22, "learning_rate": 1.967958921491426e-05, "loss": 0.0423, "step": 349 }, { "epoch": 0.22, "learning_rate": 1.9677055030039067e-05, "loss": 0.0394, "step": 350 }, { "epoch": 0.22, "learning_rate": 1.9674511027409325e-05, "loss": 0.0472, "step": 351 }, { "epoch": 0.22, "learning_rate": 1.967195720960603e-05, "loss": 0.0462, "step": 352 }, { "epoch": 0.22, "learning_rate": 1.966939357922014e-05, "loss": 0.0428, "step": 353 }, { "epoch": 0.22, "learning_rate": 1.966682013885255e-05, "loss": 0.0442, "step": 354 }, { "epoch": 0.22, "learning_rate": 1.966423689111412e-05, "loss": 0.0452, "step": 355 }, { "epoch": 0.22, "learning_rate": 1.9661643838625653e-05, "loss": 0.0444, "step": 356 }, { "epoch": 0.22, "learning_rate": 1.9659040984017913e-05, "loss": 0.0452, "step": 357 }, { "epoch": 0.22, "learning_rate": 1.9656428329931588e-05, "loss": 0.0445, "step": 358 }, { "epoch": 0.22, "learning_rate": 1.9653805879017323e-05, "loss": 0.0519, "step": 359 }, { "epoch": 0.22, "learning_rate": 1.9651173633935702e-05, "loss": 0.0423, "step": 360 }, { "epoch": 0.22, "learning_rate": 1.9648531597357238e-05, "loss": 0.0421, "step": 361 }, { "epoch": 0.23, "learning_rate": 1.9645879771962376e-05, "loss": 0.0449, "step": 362 }, { "epoch": 0.23, "learning_rate": 1.964321816044151e-05, "loss": 0.0434, "step": 363 }, { "epoch": 0.23, "learning_rate": 1.964054676549494e-05, "loss": 0.0588, "step": 364 }, { "epoch": 0.23, "learning_rate": 1.9637865589832902e-05, "loss": 0.0475, "step": 365 }, { "epoch": 0.23, "learning_rate": 1.963517463617556e-05, "loss": 0.0438, "step": 366 }, { "epoch": 0.23, "learning_rate": 1.963247390725299e-05, "loss": 0.0525, "step": 367 }, { "epoch": 0.23, "learning_rate": 1.962976340580519e-05, "loss": 0.0531, "step": 368 }, { "epoch": 0.23, "learning_rate": 1.9627043134582068e-05, "loss": 0.0381, "step": 369 }, { "epoch": 0.23, "learning_rate": 1.9624313096343445e-05, "loss": 0.0366, "step": 370 }, { "epoch": 0.23, "learning_rate": 1.9621573293859058e-05, "loss": 0.0447, "step": 371 }, { "epoch": 0.23, "learning_rate": 1.961882372990854e-05, "loss": 0.047, "step": 372 }, { "epoch": 0.23, "learning_rate": 1.9616064407281435e-05, "loss": 0.0509, "step": 373 }, { "epoch": 0.23, "learning_rate": 1.9613295328777187e-05, "loss": 0.0512, "step": 374 }, { "epoch": 0.23, "learning_rate": 1.961051649720513e-05, "loss": 0.0436, "step": 375 }, { "epoch": 0.23, "learning_rate": 1.96077279153845e-05, "loss": 0.0431, "step": 376 }, { "epoch": 0.23, "learning_rate": 1.9604929586144426e-05, "loss": 0.0446, "step": 377 }, { "epoch": 0.24, "learning_rate": 1.960212151232392e-05, "loss": 0.0378, "step": 378 }, { "epoch": 0.24, "learning_rate": 1.959930369677189e-05, "loss": 0.0408, "step": 379 }, { "epoch": 0.24, "learning_rate": 1.9596476142347114e-05, "loss": 0.0438, "step": 380 }, { "epoch": 0.24, "learning_rate": 1.959363885191826e-05, "loss": 0.0443, "step": 381 }, { "epoch": 0.24, "learning_rate": 1.9590791828363874e-05, "loss": 0.0465, "step": 382 }, { "epoch": 0.24, "learning_rate": 1.958793507457237e-05, "loss": 0.0349, "step": 383 }, { "epoch": 0.24, "learning_rate": 1.958506859344204e-05, "loss": 0.0412, "step": 384 }, { "epoch": 0.24, "learning_rate": 1.9582192387881037e-05, "loss": 0.0399, "step": 385 }, { "epoch": 0.24, "learning_rate": 1.9579306460807398e-05, "loss": 0.0425, "step": 386 }, { "epoch": 0.24, "learning_rate": 1.9576410815148996e-05, "loss": 0.0445, "step": 387 }, { "epoch": 0.24, "learning_rate": 1.9573505453843586e-05, "loss": 0.0388, "step": 388 }, { "epoch": 0.24, "learning_rate": 1.9570590379838767e-05, "loss": 0.0386, "step": 389 }, { "epoch": 0.24, "learning_rate": 1.9567665596092e-05, "loss": 0.0391, "step": 390 }, { "epoch": 0.24, "learning_rate": 1.9564731105570592e-05, "loss": 0.0445, "step": 391 }, { "epoch": 0.24, "learning_rate": 1.95617869112517e-05, "loss": 0.0432, "step": 392 }, { "epoch": 0.24, "learning_rate": 1.9558833016122327e-05, "loss": 0.0439, "step": 393 }, { "epoch": 0.25, "learning_rate": 1.9555869423179316e-05, "loss": 0.0402, "step": 394 }, { "epoch": 0.25, "learning_rate": 1.9552896135429346e-05, "loss": 0.0458, "step": 395 }, { "epoch": 0.25, "learning_rate": 1.954991315588894e-05, "loss": 0.0511, "step": 396 }, { "epoch": 0.25, "learning_rate": 1.9546920487584442e-05, "loss": 0.0383, "step": 397 }, { "epoch": 0.25, "learning_rate": 1.9543918133552034e-05, "loss": 0.0462, "step": 398 }, { "epoch": 0.25, "learning_rate": 1.9540906096837727e-05, "loss": 0.0396, "step": 399 }, { "epoch": 0.25, "learning_rate": 1.953788438049735e-05, "loss": 0.0457, "step": 400 }, { "epoch": 0.25, "learning_rate": 1.953485298759655e-05, "loss": 0.0488, "step": 401 }, { "epoch": 0.25, "learning_rate": 1.95318119212108e-05, "loss": 0.0479, "step": 402 }, { "epoch": 0.25, "learning_rate": 1.9528761184425376e-05, "loss": 0.0466, "step": 403 }, { "epoch": 0.25, "learning_rate": 1.9525700780335372e-05, "loss": 0.0429, "step": 404 }, { "epoch": 0.25, "learning_rate": 1.9522630712045695e-05, "loss": 0.0348, "step": 405 }, { "epoch": 0.25, "learning_rate": 1.9519550982671042e-05, "loss": 0.0473, "step": 406 }, { "epoch": 0.25, "learning_rate": 1.951646159533593e-05, "loss": 0.0435, "step": 407 }, { "epoch": 0.25, "learning_rate": 1.9513362553174654e-05, "loss": 0.0389, "step": 408 }, { "epoch": 0.25, "learning_rate": 1.951025385933132e-05, "loss": 0.0434, "step": 409 }, { "epoch": 0.25, "learning_rate": 1.9507135516959824e-05, "loss": 0.0403, "step": 410 }, { "epoch": 0.26, "learning_rate": 1.9504007529223838e-05, "loss": 0.0385, "step": 411 }, { "epoch": 0.26, "learning_rate": 1.950086989929684e-05, "loss": 0.0395, "step": 412 }, { "epoch": 0.26, "learning_rate": 1.949772263036207e-05, "loss": 0.0418, "step": 413 }, { "epoch": 0.26, "learning_rate": 1.9494565725612565e-05, "loss": 0.0356, "step": 414 }, { "epoch": 0.26, "learning_rate": 1.9491399188251123e-05, "loss": 0.0323, "step": 415 }, { "epoch": 0.26, "learning_rate": 1.9488223021490324e-05, "loss": 0.0401, "step": 416 }, { "epoch": 0.26, "learning_rate": 1.9485037228552513e-05, "loss": 0.0422, "step": 417 }, { "epoch": 0.26, "learning_rate": 1.9481841812669806e-05, "loss": 0.0387, "step": 418 }, { "epoch": 0.26, "learning_rate": 1.9478636777084077e-05, "loss": 0.0436, "step": 419 }, { "epoch": 0.26, "learning_rate": 1.9475422125046962e-05, "loss": 0.0451, "step": 420 }, { "epoch": 0.26, "learning_rate": 1.947219785981985e-05, "loss": 0.0412, "step": 421 }, { "epoch": 0.26, "learning_rate": 1.9468963984673885e-05, "loss": 0.0407, "step": 422 }, { "epoch": 0.26, "learning_rate": 1.9465720502889963e-05, "loss": 0.0462, "step": 423 }, { "epoch": 0.26, "learning_rate": 1.946246741775873e-05, "loss": 0.0506, "step": 424 }, { "epoch": 0.26, "learning_rate": 1.9459204732580557e-05, "loss": 0.0588, "step": 425 }, { "epoch": 0.26, "learning_rate": 1.9455932450665576e-05, "loss": 0.037, "step": 426 }, { "epoch": 0.27, "learning_rate": 1.9452650575333646e-05, "loss": 0.0469, "step": 427 }, { "epoch": 0.27, "learning_rate": 1.944935910991435e-05, "loss": 0.0445, "step": 428 }, { "epoch": 0.27, "learning_rate": 1.9446058057747025e-05, "loss": 0.0395, "step": 429 }, { "epoch": 0.27, "learning_rate": 1.9442747422180704e-05, "loss": 0.0385, "step": 430 }, { "epoch": 0.27, "learning_rate": 1.9439427206574168e-05, "loss": 0.037, "step": 431 }, { "epoch": 0.27, "learning_rate": 1.9436097414295904e-05, "loss": 0.0489, "step": 432 }, { "epoch": 0.27, "learning_rate": 1.943275804872411e-05, "loss": 0.0531, "step": 433 }, { "epoch": 0.27, "learning_rate": 1.9429409113246715e-05, "loss": 0.0432, "step": 434 }, { "epoch": 0.27, "learning_rate": 1.9426050611261343e-05, "loss": 0.0552, "step": 435 }, { "epoch": 0.27, "learning_rate": 1.9422682546175325e-05, "loss": 0.0442, "step": 436 }, { "epoch": 0.27, "learning_rate": 1.94193049214057e-05, "loss": 0.0404, "step": 437 }, { "epoch": 0.27, "learning_rate": 1.9415917740379195e-05, "loss": 0.0492, "step": 438 }, { "epoch": 0.27, "learning_rate": 1.9412521006532245e-05, "loss": 0.0567, "step": 439 }, { "epoch": 0.27, "learning_rate": 1.9409114723310967e-05, "loss": 0.042, "step": 440 }, { "epoch": 0.27, "learning_rate": 1.9405698894171172e-05, "loss": 0.036, "step": 441 }, { "epoch": 0.27, "learning_rate": 1.940227352257835e-05, "loss": 0.038, "step": 442 }, { "epoch": 0.28, "learning_rate": 1.9398838612007686e-05, "loss": 0.0444, "step": 443 }, { "epoch": 0.28, "learning_rate": 1.939539416594402e-05, "loss": 0.0398, "step": 444 }, { "epoch": 0.28, "learning_rate": 1.9391940187881885e-05, "loss": 0.0496, "step": 445 }, { "epoch": 0.28, "learning_rate": 1.9388476681325476e-05, "loss": 0.0387, "step": 446 }, { "epoch": 0.28, "learning_rate": 1.9385003649788657e-05, "loss": 0.039, "step": 447 }, { "epoch": 0.28, "learning_rate": 1.9381521096794963e-05, "loss": 0.0444, "step": 448 }, { "epoch": 0.28, "learning_rate": 1.937802902587757e-05, "loss": 0.0402, "step": 449 }, { "epoch": 0.28, "learning_rate": 1.937452744057933e-05, "loss": 0.0359, "step": 450 }, { "epoch": 0.28, "learning_rate": 1.9371016344452733e-05, "loss": 0.0564, "step": 451 }, { "epoch": 0.28, "learning_rate": 1.936749574105993e-05, "loss": 0.0443, "step": 452 }, { "epoch": 0.28, "learning_rate": 1.936396563397271e-05, "loss": 0.0393, "step": 453 }, { "epoch": 0.28, "learning_rate": 1.936042602677251e-05, "loss": 0.0337, "step": 454 }, { "epoch": 0.28, "learning_rate": 1.9356876923050395e-05, "loss": 0.0377, "step": 455 }, { "epoch": 0.28, "learning_rate": 1.9353318326407076e-05, "loss": 0.0489, "step": 456 }, { "epoch": 0.28, "learning_rate": 1.9349750240452886e-05, "loss": 0.0481, "step": 457 }, { "epoch": 0.28, "learning_rate": 1.934617266880779e-05, "loss": 0.0407, "step": 458 }, { "epoch": 0.29, "learning_rate": 1.934258561510138e-05, "loss": 0.0403, "step": 459 }, { "epoch": 0.29, "learning_rate": 1.933898908297286e-05, "loss": 0.0393, "step": 460 }, { "epoch": 0.29, "learning_rate": 1.933538307607105e-05, "loss": 0.0357, "step": 461 }, { "epoch": 0.29, "learning_rate": 1.9331767598054395e-05, "loss": 0.0529, "step": 462 }, { "epoch": 0.29, "learning_rate": 1.9328142652590934e-05, "loss": 0.0366, "step": 463 }, { "epoch": 0.29, "learning_rate": 1.932450824335832e-05, "loss": 0.0332, "step": 464 }, { "epoch": 0.29, "learning_rate": 1.9320864374043805e-05, "loss": 0.0416, "step": 465 }, { "epoch": 0.29, "learning_rate": 1.9317211048344235e-05, "loss": 0.0352, "step": 466 }, { "epoch": 0.29, "learning_rate": 1.931354826996606e-05, "loss": 0.0419, "step": 467 }, { "epoch": 0.29, "learning_rate": 1.9309876042625304e-05, "loss": 0.0372, "step": 468 }, { "epoch": 0.29, "learning_rate": 1.9306194370047592e-05, "loss": 0.0454, "step": 469 }, { "epoch": 0.29, "learning_rate": 1.9302503255968127e-05, "loss": 0.0418, "step": 470 }, { "epoch": 0.29, "learning_rate": 1.929880270413169e-05, "loss": 0.038, "step": 471 }, { "epoch": 0.29, "learning_rate": 1.9295092718292632e-05, "loss": 0.0367, "step": 472 }, { "epoch": 0.29, "learning_rate": 1.929137330221489e-05, "loss": 0.047, "step": 473 }, { "epoch": 0.29, "learning_rate": 1.9287644459671948e-05, "loss": 0.0419, "step": 474 }, { "epoch": 0.3, "learning_rate": 1.928390619444687e-05, "loss": 0.0396, "step": 475 }, { "epoch": 0.3, "learning_rate": 1.928015851033227e-05, "loss": 0.0467, "step": 476 }, { "epoch": 0.3, "learning_rate": 1.9276401411130328e-05, "loss": 0.0411, "step": 477 }, { "epoch": 0.3, "learning_rate": 1.927263490065276e-05, "loss": 0.0441, "step": 478 }, { "epoch": 0.3, "learning_rate": 1.926885898272085e-05, "loss": 0.0379, "step": 479 }, { "epoch": 0.3, "learning_rate": 1.9265073661165407e-05, "loss": 0.0362, "step": 480 }, { "epoch": 0.3, "learning_rate": 1.9261278939826797e-05, "loss": 0.0426, "step": 481 }, { "epoch": 0.3, "learning_rate": 1.9257474822554907e-05, "loss": 0.0373, "step": 482 }, { "epoch": 0.3, "learning_rate": 1.925366131320917e-05, "loss": 0.0396, "step": 483 }, { "epoch": 0.3, "learning_rate": 1.9249838415658543e-05, "loss": 0.0456, "step": 484 }, { "epoch": 0.3, "learning_rate": 1.9246006133781496e-05, "loss": 0.0441, "step": 485 }, { "epoch": 0.3, "learning_rate": 1.924216447146604e-05, "loss": 0.0376, "step": 486 }, { "epoch": 0.3, "learning_rate": 1.9238313432609695e-05, "loss": 0.0401, "step": 487 }, { "epoch": 0.3, "learning_rate": 1.9234453021119484e-05, "loss": 0.0341, "step": 488 }, { "epoch": 0.3, "learning_rate": 1.9230583240911954e-05, "loss": 0.0434, "step": 489 }, { "epoch": 0.3, "learning_rate": 1.9226704095913143e-05, "loss": 0.0321, "step": 490 }, { "epoch": 0.31, "learning_rate": 1.9222815590058604e-05, "loss": 0.0417, "step": 491 }, { "epoch": 0.31, "learning_rate": 1.9218917727293374e-05, "loss": 0.0329, "step": 492 }, { "epoch": 0.31, "learning_rate": 1.921501051157199e-05, "loss": 0.0285, "step": 493 }, { "epoch": 0.31, "learning_rate": 1.9211093946858484e-05, "loss": 0.0367, "step": 494 }, { "epoch": 0.31, "learning_rate": 1.9207168037126352e-05, "loss": 0.0412, "step": 495 }, { "epoch": 0.31, "learning_rate": 1.9203232786358595e-05, "loss": 0.0353, "step": 496 }, { "epoch": 0.31, "learning_rate": 1.919928819854768e-05, "loss": 0.0452, "step": 497 }, { "epoch": 0.31, "learning_rate": 1.9195334277695542e-05, "loss": 0.0429, "step": 498 }, { "epoch": 0.31, "learning_rate": 1.919137102781359e-05, "loss": 0.0367, "step": 499 }, { "epoch": 0.31, "learning_rate": 1.91873984529227e-05, "loss": 0.0408, "step": 500 }, { "epoch": 0.31, "learning_rate": 1.9183416557053207e-05, "loss": 0.0483, "step": 501 }, { "epoch": 0.31, "learning_rate": 1.9179425344244896e-05, "loss": 0.04, "step": 502 }, { "epoch": 0.31, "learning_rate": 1.917542481854702e-05, "loss": 0.0425, "step": 503 }, { "epoch": 0.31, "learning_rate": 1.9171414984018266e-05, "loss": 0.0461, "step": 504 }, { "epoch": 0.31, "learning_rate": 1.9167395844726763e-05, "loss": 0.0321, "step": 505 }, { "epoch": 0.31, "learning_rate": 1.9163367404750096e-05, "loss": 0.0439, "step": 506 }, { "epoch": 0.32, "learning_rate": 1.915932966817527e-05, "loss": 0.0417, "step": 507 }, { "epoch": 0.32, "learning_rate": 1.9155282639098732e-05, "loss": 0.044, "step": 508 }, { "epoch": 0.32, "learning_rate": 1.915122632162635e-05, "loss": 0.0482, "step": 509 }, { "epoch": 0.32, "learning_rate": 1.914716071987342e-05, "loss": 0.0468, "step": 510 }, { "epoch": 0.32, "learning_rate": 1.9143085837964655e-05, "loss": 0.0445, "step": 511 }, { "epoch": 0.32, "learning_rate": 1.9139001680034185e-05, "loss": 0.0382, "step": 512 }, { "epoch": 0.32, "learning_rate": 1.9134908250225548e-05, "loss": 0.0431, "step": 513 }, { "epoch": 0.32, "learning_rate": 1.913080555269169e-05, "loss": 0.0445, "step": 514 }, { "epoch": 0.32, "learning_rate": 1.912669359159496e-05, "loss": 0.0442, "step": 515 }, { "epoch": 0.32, "learning_rate": 1.9122572371107105e-05, "loss": 0.035, "step": 516 }, { "epoch": 0.32, "learning_rate": 1.911844189540927e-05, "loss": 0.048, "step": 517 }, { "epoch": 0.32, "learning_rate": 1.9114302168691978e-05, "loss": 0.0377, "step": 518 }, { "epoch": 0.32, "learning_rate": 1.911015319515515e-05, "loss": 0.0395, "step": 519 }, { "epoch": 0.32, "learning_rate": 1.9105994979008083e-05, "loss": 0.0342, "step": 520 }, { "epoch": 0.32, "learning_rate": 1.9101827524469453e-05, "loss": 0.0453, "step": 521 }, { "epoch": 0.32, "learning_rate": 1.9097650835767302e-05, "loss": 0.037, "step": 522 }, { "epoch": 0.33, "learning_rate": 1.909346491713905e-05, "loss": 0.0407, "step": 523 }, { "epoch": 0.33, "learning_rate": 1.908926977283148e-05, "loss": 0.0379, "step": 524 }, { "epoch": 0.33, "learning_rate": 1.9085065407100723e-05, "loss": 0.0357, "step": 525 }, { "epoch": 0.33, "learning_rate": 1.9080851824212282e-05, "loss": 0.046, "step": 526 }, { "epoch": 0.33, "learning_rate": 1.9076629028440997e-05, "loss": 0.0372, "step": 527 }, { "epoch": 0.33, "learning_rate": 1.9072397024071066e-05, "loss": 0.0356, "step": 528 }, { "epoch": 0.33, "learning_rate": 1.9068155815396018e-05, "loss": 0.0318, "step": 529 }, { "epoch": 0.33, "learning_rate": 1.9063905406718735e-05, "loss": 0.0428, "step": 530 }, { "epoch": 0.33, "learning_rate": 1.9059645802351423e-05, "loss": 0.0563, "step": 531 }, { "epoch": 0.33, "learning_rate": 1.9055377006615613e-05, "loss": 0.0476, "step": 532 }, { "epoch": 0.33, "learning_rate": 1.9051099023842173e-05, "loss": 0.0409, "step": 533 }, { "epoch": 0.33, "learning_rate": 1.904681185837128e-05, "loss": 0.0352, "step": 534 }, { "epoch": 0.33, "learning_rate": 1.9042515514552438e-05, "loss": 0.0525, "step": 535 }, { "epoch": 0.33, "learning_rate": 1.9038209996744453e-05, "loss": 0.0457, "step": 536 }, { "epoch": 0.33, "learning_rate": 1.903389530931545e-05, "loss": 0.0446, "step": 537 }, { "epoch": 0.33, "learning_rate": 1.902957145664284e-05, "loss": 0.0381, "step": 538 }, { "epoch": 0.34, "learning_rate": 1.9025238443113346e-05, "loss": 0.0318, "step": 539 }, { "epoch": 0.34, "learning_rate": 1.9020896273122987e-05, "loss": 0.0444, "step": 540 }, { "epoch": 0.34, "learning_rate": 1.9016544951077057e-05, "loss": 0.0425, "step": 541 }, { "epoch": 0.34, "learning_rate": 1.9012184481390156e-05, "loss": 0.0377, "step": 542 }, { "epoch": 0.34, "learning_rate": 1.900781486848614e-05, "loss": 0.0427, "step": 543 }, { "epoch": 0.34, "learning_rate": 1.9003436116798156e-05, "loss": 0.0434, "step": 544 }, { "epoch": 0.34, "learning_rate": 1.899904823076863e-05, "loss": 0.0463, "step": 545 }, { "epoch": 0.34, "learning_rate": 1.899465121484924e-05, "loss": 0.0409, "step": 546 }, { "epoch": 0.34, "learning_rate": 1.8990245073500925e-05, "loss": 0.0371, "step": 547 }, { "epoch": 0.34, "learning_rate": 1.8985829811193902e-05, "loss": 0.0449, "step": 548 }, { "epoch": 0.34, "learning_rate": 1.898140543240762e-05, "loss": 0.0418, "step": 549 }, { "epoch": 0.34, "learning_rate": 1.8976971941630792e-05, "loss": 0.0341, "step": 550 }, { "epoch": 0.34, "learning_rate": 1.897252934336136e-05, "loss": 0.0366, "step": 551 }, { "epoch": 0.34, "learning_rate": 1.8968077642106532e-05, "loss": 0.0374, "step": 552 }, { "epoch": 0.34, "learning_rate": 1.8963616842382716e-05, "loss": 0.0451, "step": 553 }, { "epoch": 0.34, "learning_rate": 1.8959146948715582e-05, "loss": 0.0313, "step": 554 }, { "epoch": 0.35, "learning_rate": 1.8954667965640013e-05, "loss": 0.0441, "step": 555 }, { "epoch": 0.35, "learning_rate": 1.8950179897700106e-05, "loss": 0.0448, "step": 556 }, { "epoch": 0.35, "learning_rate": 1.8945682749449193e-05, "loss": 0.0322, "step": 557 }, { "epoch": 0.35, "learning_rate": 1.89411765254498e-05, "loss": 0.0453, "step": 558 }, { "epoch": 0.35, "learning_rate": 1.8936661230273677e-05, "loss": 0.0382, "step": 559 }, { "epoch": 0.35, "learning_rate": 1.8932136868501763e-05, "loss": 0.0442, "step": 560 }, { "epoch": 0.35, "learning_rate": 1.8927603444724206e-05, "loss": 0.048, "step": 561 }, { "epoch": 0.35, "learning_rate": 1.892306096354034e-05, "loss": 0.0425, "step": 562 }, { "epoch": 0.35, "learning_rate": 1.8918509429558695e-05, "loss": 0.0433, "step": 563 }, { "epoch": 0.35, "learning_rate": 1.8913948847396978e-05, "loss": 0.0475, "step": 564 }, { "epoch": 0.35, "learning_rate": 1.8909379221682083e-05, "loss": 0.0418, "step": 565 }, { "epoch": 0.35, "learning_rate": 1.8904800557050075e-05, "loss": 0.0444, "step": 566 }, { "epoch": 0.35, "learning_rate": 1.890021285814619e-05, "loss": 0.0406, "step": 567 }, { "epoch": 0.35, "learning_rate": 1.8895616129624826e-05, "loss": 0.0462, "step": 568 }, { "epoch": 0.35, "learning_rate": 1.8891010376149554e-05, "loss": 0.048, "step": 569 }, { "epoch": 0.35, "learning_rate": 1.8886395602393087e-05, "loss": 0.0392, "step": 570 }, { "epoch": 0.36, "learning_rate": 1.8881771813037297e-05, "loss": 0.0405, "step": 571 }, { "epoch": 0.36, "learning_rate": 1.88771390127732e-05, "loss": 0.051, "step": 572 }, { "epoch": 0.36, "learning_rate": 1.8872497206300954e-05, "loss": 0.0459, "step": 573 }, { "epoch": 0.36, "learning_rate": 1.8867846398329856e-05, "loss": 0.0405, "step": 574 }, { "epoch": 0.36, "learning_rate": 1.8863186593578336e-05, "loss": 0.0381, "step": 575 }, { "epoch": 0.36, "learning_rate": 1.8858517796773948e-05, "loss": 0.0322, "step": 576 }, { "epoch": 0.36, "learning_rate": 1.885384001265337e-05, "loss": 0.0359, "step": 577 }, { "epoch": 0.36, "learning_rate": 1.88491532459624e-05, "loss": 0.0382, "step": 578 }, { "epoch": 0.36, "learning_rate": 1.884445750145595e-05, "loss": 0.0462, "step": 579 }, { "epoch": 0.36, "learning_rate": 1.8839752783898033e-05, "loss": 0.0461, "step": 580 }, { "epoch": 0.36, "learning_rate": 1.883503909806178e-05, "loss": 0.0323, "step": 581 }, { "epoch": 0.36, "learning_rate": 1.8830316448729397e-05, "loss": 0.0386, "step": 582 }, { "epoch": 0.36, "learning_rate": 1.8825584840692212e-05, "loss": 0.035, "step": 583 }, { "epoch": 0.36, "learning_rate": 1.882084427875062e-05, "loss": 0.0351, "step": 584 }, { "epoch": 0.36, "learning_rate": 1.8816094767714113e-05, "loss": 0.0416, "step": 585 }, { "epoch": 0.36, "learning_rate": 1.8811336312401258e-05, "loss": 0.0338, "step": 586 }, { "epoch": 0.37, "learning_rate": 1.880656891763969e-05, "loss": 0.0375, "step": 587 }, { "epoch": 0.37, "learning_rate": 1.8801792588266125e-05, "loss": 0.0288, "step": 588 }, { "epoch": 0.37, "learning_rate": 1.8797007329126336e-05, "loss": 0.0432, "step": 589 }, { "epoch": 0.37, "learning_rate": 1.8792213145075156e-05, "loss": 0.0383, "step": 590 }, { "epoch": 0.37, "learning_rate": 1.8787410040976472e-05, "loss": 0.0382, "step": 591 }, { "epoch": 0.37, "learning_rate": 1.878259802170323e-05, "loss": 0.0402, "step": 592 }, { "epoch": 0.37, "learning_rate": 1.8777777092137407e-05, "loss": 0.0282, "step": 593 }, { "epoch": 0.37, "learning_rate": 1.8772947257170034e-05, "loss": 0.0353, "step": 594 }, { "epoch": 0.37, "learning_rate": 1.876810852170116e-05, "loss": 0.0357, "step": 595 }, { "epoch": 0.37, "learning_rate": 1.876326089063988e-05, "loss": 0.0332, "step": 596 }, { "epoch": 0.37, "learning_rate": 1.8758404368904303e-05, "loss": 0.0397, "step": 597 }, { "epoch": 0.37, "learning_rate": 1.8753538961421565e-05, "loss": 0.0447, "step": 598 }, { "epoch": 0.37, "learning_rate": 1.8748664673127814e-05, "loss": 0.0327, "step": 599 }, { "epoch": 0.37, "learning_rate": 1.8743781508968206e-05, "loss": 0.0473, "step": 600 }, { "epoch": 0.37, "learning_rate": 1.8738889473896903e-05, "loss": 0.0422, "step": 601 }, { "epoch": 0.37, "learning_rate": 1.873398857287707e-05, "loss": 0.0429, "step": 602 }, { "epoch": 0.38, "learning_rate": 1.8729078810880866e-05, "loss": 0.0441, "step": 603 }, { "epoch": 0.38, "learning_rate": 1.872416019288944e-05, "loss": 0.0477, "step": 604 }, { "epoch": 0.38, "learning_rate": 1.8719232723892916e-05, "loss": 0.0437, "step": 605 }, { "epoch": 0.38, "learning_rate": 1.8714296408890414e-05, "loss": 0.0354, "step": 606 }, { "epoch": 0.38, "learning_rate": 1.8709351252890015e-05, "loss": 0.0336, "step": 607 }, { "epoch": 0.38, "learning_rate": 1.870439726090878e-05, "loss": 0.0337, "step": 608 }, { "epoch": 0.38, "learning_rate": 1.8699434437972726e-05, "loss": 0.0398, "step": 609 }, { "epoch": 0.38, "learning_rate": 1.869446278911683e-05, "loss": 0.0413, "step": 610 }, { "epoch": 0.38, "learning_rate": 1.8689482319385037e-05, "loss": 0.0391, "step": 611 }, { "epoch": 0.38, "learning_rate": 1.868449303383022e-05, "loss": 0.0425, "step": 612 }, { "epoch": 0.38, "learning_rate": 1.867949493751421e-05, "loss": 0.0369, "step": 613 }, { "epoch": 0.38, "learning_rate": 1.8674488035507776e-05, "loss": 0.0292, "step": 614 }, { "epoch": 0.38, "learning_rate": 1.866947233289061e-05, "loss": 0.0378, "step": 615 }, { "epoch": 0.38, "learning_rate": 1.8664447834751348e-05, "loss": 0.0363, "step": 616 }, { "epoch": 0.38, "learning_rate": 1.8659414546187537e-05, "loss": 0.0385, "step": 617 }, { "epoch": 0.38, "learning_rate": 1.8654372472305648e-05, "loss": 0.0343, "step": 618 }, { "epoch": 0.38, "learning_rate": 1.864932161822107e-05, "loss": 0.0374, "step": 619 }, { "epoch": 0.39, "learning_rate": 1.8644261989058082e-05, "loss": 0.0296, "step": 620 }, { "epoch": 0.39, "learning_rate": 1.863919358994989e-05, "loss": 0.0428, "step": 621 }, { "epoch": 0.39, "learning_rate": 1.8634116426038574e-05, "loss": 0.0343, "step": 622 }, { "epoch": 0.39, "learning_rate": 1.862903050247513e-05, "loss": 0.0346, "step": 623 }, { "epoch": 0.39, "learning_rate": 1.8623935824419416e-05, "loss": 0.0345, "step": 624 }, { "epoch": 0.39, "learning_rate": 1.8618832397040192e-05, "loss": 0.0485, "step": 625 }, { "epoch": 0.39, "learning_rate": 1.8613720225515084e-05, "loss": 0.0355, "step": 626 }, { "epoch": 0.39, "learning_rate": 1.860859931503059e-05, "loss": 0.0404, "step": 627 }, { "epoch": 0.39, "learning_rate": 1.8603469670782082e-05, "loss": 0.0453, "step": 628 }, { "epoch": 0.39, "learning_rate": 1.859833129797378e-05, "loss": 0.0425, "step": 629 }, { "epoch": 0.39, "learning_rate": 1.859318420181877e-05, "loss": 0.0442, "step": 630 }, { "epoch": 0.39, "learning_rate": 1.858802838753898e-05, "loss": 0.0315, "step": 631 }, { "epoch": 0.39, "learning_rate": 1.858286386036519e-05, "loss": 0.0376, "step": 632 }, { "epoch": 0.39, "learning_rate": 1.8577690625537022e-05, "loss": 0.0494, "step": 633 }, { "epoch": 0.39, "learning_rate": 1.857250868830292e-05, "loss": 0.0393, "step": 634 }, { "epoch": 0.39, "learning_rate": 1.8567318053920166e-05, "loss": 0.0397, "step": 635 }, { "epoch": 0.4, "learning_rate": 1.8562118727654864e-05, "loss": 0.0426, "step": 636 }, { "epoch": 0.4, "learning_rate": 1.8556910714781935e-05, "loss": 0.04, "step": 637 }, { "epoch": 0.4, "learning_rate": 1.8551694020585117e-05, "loss": 0.0444, "step": 638 }, { "epoch": 0.4, "learning_rate": 1.8546468650356947e-05, "loss": 0.0272, "step": 639 }, { "epoch": 0.4, "learning_rate": 1.854123460939877e-05, "loss": 0.0374, "step": 640 }, { "epoch": 0.4, "learning_rate": 1.8535991903020733e-05, "loss": 0.0372, "step": 641 }, { "epoch": 0.4, "learning_rate": 1.8530740536541766e-05, "loss": 0.0408, "step": 642 }, { "epoch": 0.4, "learning_rate": 1.8525480515289586e-05, "loss": 0.0414, "step": 643 }, { "epoch": 0.4, "learning_rate": 1.852021184460069e-05, "loss": 0.0421, "step": 644 }, { "epoch": 0.4, "learning_rate": 1.851493452982036e-05, "loss": 0.0373, "step": 645 }, { "epoch": 0.4, "learning_rate": 1.8509648576302635e-05, "loss": 0.0391, "step": 646 }, { "epoch": 0.4, "learning_rate": 1.8504353989410326e-05, "loss": 0.0452, "step": 647 }, { "epoch": 0.4, "learning_rate": 1.8499050774514994e-05, "loss": 0.0475, "step": 648 }, { "epoch": 0.4, "learning_rate": 1.849373893699697e-05, "loss": 0.0372, "step": 649 }, { "epoch": 0.4, "learning_rate": 1.8488418482245317e-05, "loss": 0.0373, "step": 650 }, { "epoch": 0.4, "learning_rate": 1.8483089415657847e-05, "loss": 0.0402, "step": 651 }, { "epoch": 0.41, "learning_rate": 1.8477751742641107e-05, "loss": 0.0352, "step": 652 }, { "epoch": 0.41, "learning_rate": 1.8472405468610383e-05, "loss": 0.0343, "step": 653 }, { "epoch": 0.41, "learning_rate": 1.8467050598989677e-05, "loss": 0.0432, "step": 654 }, { "epoch": 0.41, "learning_rate": 1.8461687139211712e-05, "loss": 0.0445, "step": 655 }, { "epoch": 0.41, "learning_rate": 1.8456315094717937e-05, "loss": 0.047, "step": 656 }, { "epoch": 0.41, "learning_rate": 1.84509344709585e-05, "loss": 0.0287, "step": 657 }, { "epoch": 0.41, "learning_rate": 1.844554527339226e-05, "loss": 0.0367, "step": 658 }, { "epoch": 0.41, "learning_rate": 1.8440147507486765e-05, "loss": 0.0431, "step": 659 }, { "epoch": 0.41, "learning_rate": 1.843474117871826e-05, "loss": 0.0323, "step": 660 }, { "epoch": 0.41, "learning_rate": 1.8429326292571686e-05, "loss": 0.0302, "step": 661 }, { "epoch": 0.41, "learning_rate": 1.8423902854540655e-05, "loss": 0.0407, "step": 662 }, { "epoch": 0.41, "learning_rate": 1.8418470870127464e-05, "loss": 0.0329, "step": 663 }, { "epoch": 0.41, "learning_rate": 1.8413030344843064e-05, "loss": 0.0298, "step": 664 }, { "epoch": 0.41, "learning_rate": 1.84075812842071e-05, "loss": 0.0344, "step": 665 }, { "epoch": 0.41, "learning_rate": 1.8402123693747845e-05, "loss": 0.0311, "step": 666 }, { "epoch": 0.41, "learning_rate": 1.8396657579002245e-05, "loss": 0.0321, "step": 667 }, { "epoch": 0.42, "learning_rate": 1.8391182945515888e-05, "loss": 0.0367, "step": 668 }, { "epoch": 0.42, "learning_rate": 1.838569979884301e-05, "loss": 0.0402, "step": 669 }, { "epoch": 0.42, "learning_rate": 1.8380208144546473e-05, "loss": 0.0338, "step": 670 }, { "epoch": 0.42, "learning_rate": 1.8374707988197786e-05, "loss": 0.0349, "step": 671 }, { "epoch": 0.42, "learning_rate": 1.8369199335377065e-05, "loss": 0.0436, "step": 672 }, { "epoch": 0.42, "learning_rate": 1.8363682191673065e-05, "loss": 0.0364, "step": 673 }, { "epoch": 0.42, "learning_rate": 1.835815656268314e-05, "loss": 0.0422, "step": 674 }, { "epoch": 0.42, "learning_rate": 1.835262245401326e-05, "loss": 0.0411, "step": 675 }, { "epoch": 0.42, "learning_rate": 1.8347079871278e-05, "loss": 0.0453, "step": 676 }, { "epoch": 0.42, "learning_rate": 1.8341528820100525e-05, "loss": 0.0234, "step": 677 }, { "epoch": 0.42, "learning_rate": 1.83359693061126e-05, "loss": 0.0434, "step": 678 }, { "epoch": 0.42, "learning_rate": 1.8330401334954567e-05, "loss": 0.0384, "step": 679 }, { "epoch": 0.42, "learning_rate": 1.8324824912275355e-05, "loss": 0.0377, "step": 680 }, { "epoch": 0.42, "learning_rate": 1.8319240043732465e-05, "loss": 0.0324, "step": 681 }, { "epoch": 0.42, "learning_rate": 1.8313646734991965e-05, "loss": 0.0443, "step": 682 }, { "epoch": 0.42, "learning_rate": 1.830804499172849e-05, "loss": 0.0402, "step": 683 }, { "epoch": 0.43, "learning_rate": 1.8302434819625234e-05, "loss": 0.0315, "step": 684 }, { "epoch": 0.43, "learning_rate": 1.8296816224373926e-05, "loss": 0.0318, "step": 685 }, { "epoch": 0.43, "learning_rate": 1.8291189211674864e-05, "loss": 0.0289, "step": 686 }, { "epoch": 0.43, "learning_rate": 1.8285553787236868e-05, "loss": 0.0509, "step": 687 }, { "epoch": 0.43, "learning_rate": 1.82799099567773e-05, "loss": 0.0335, "step": 688 }, { "epoch": 0.43, "learning_rate": 1.8274257726022054e-05, "loss": 0.029, "step": 689 }, { "epoch": 0.43, "learning_rate": 1.8268597100705534e-05, "loss": 0.0403, "step": 690 }, { "epoch": 0.43, "learning_rate": 1.826292808657067e-05, "loss": 0.0364, "step": 691 }, { "epoch": 0.43, "learning_rate": 1.825725068936891e-05, "loss": 0.0406, "step": 692 }, { "epoch": 0.43, "learning_rate": 1.8251564914860186e-05, "loss": 0.0456, "step": 693 }, { "epoch": 0.43, "learning_rate": 1.824587076881294e-05, "loss": 0.0401, "step": 694 }, { "epoch": 0.43, "learning_rate": 1.824016825700412e-05, "loss": 0.0487, "step": 695 }, { "epoch": 0.43, "learning_rate": 1.8234457385219135e-05, "loss": 0.0359, "step": 696 }, { "epoch": 0.43, "learning_rate": 1.8228738159251902e-05, "loss": 0.0283, "step": 697 }, { "epoch": 0.43, "learning_rate": 1.8223010584904797e-05, "loss": 0.0462, "step": 698 }, { "epoch": 0.43, "learning_rate": 1.821727466798867e-05, "loss": 0.0397, "step": 699 }, { "epoch": 0.44, "learning_rate": 1.8211530414322834e-05, "loss": 0.0375, "step": 700 }, { "epoch": 0.44, "learning_rate": 1.8205777829735062e-05, "loss": 0.041, "step": 701 }, { "epoch": 0.44, "learning_rate": 1.820001692006158e-05, "loss": 0.0398, "step": 702 }, { "epoch": 0.44, "learning_rate": 1.8194247691147055e-05, "loss": 0.0396, "step": 703 }, { "epoch": 0.44, "learning_rate": 1.8188470148844602e-05, "loss": 0.0411, "step": 704 }, { "epoch": 0.44, "learning_rate": 1.8182684299015764e-05, "loss": 0.0381, "step": 705 }, { "epoch": 0.44, "learning_rate": 1.8176890147530518e-05, "loss": 0.0347, "step": 706 }, { "epoch": 0.44, "learning_rate": 1.8171087700267248e-05, "loss": 0.0365, "step": 707 }, { "epoch": 0.44, "learning_rate": 1.816527696311278e-05, "loss": 0.03, "step": 708 }, { "epoch": 0.44, "learning_rate": 1.8159457941962325e-05, "loss": 0.0354, "step": 709 }, { "epoch": 0.44, "learning_rate": 1.815363064271952e-05, "loss": 0.0261, "step": 710 }, { "epoch": 0.44, "learning_rate": 1.814779507129638e-05, "loss": 0.0295, "step": 711 }, { "epoch": 0.44, "learning_rate": 1.814195123361333e-05, "loss": 0.0336, "step": 712 }, { "epoch": 0.44, "learning_rate": 1.813609913559917e-05, "loss": 0.0395, "step": 713 }, { "epoch": 0.44, "learning_rate": 1.8130238783191087e-05, "loss": 0.0391, "step": 714 }, { "epoch": 0.44, "learning_rate": 1.812437018233464e-05, "loss": 0.034, "step": 715 }, { "epoch": 0.45, "learning_rate": 1.8118493338983755e-05, "loss": 0.0369, "step": 716 }, { "epoch": 0.45, "learning_rate": 1.8112608259100723e-05, "loss": 0.0282, "step": 717 }, { "epoch": 0.45, "learning_rate": 1.8106714948656188e-05, "loss": 0.03, "step": 718 }, { "epoch": 0.45, "learning_rate": 1.810081341362915e-05, "loss": 0.0385, "step": 719 }, { "epoch": 0.45, "learning_rate": 1.809490366000694e-05, "loss": 0.0291, "step": 720 }, { "epoch": 0.45, "learning_rate": 1.808898569378525e-05, "loss": 0.0314, "step": 721 }, { "epoch": 0.45, "learning_rate": 1.8083059520968086e-05, "loss": 0.0393, "step": 722 }, { "epoch": 0.45, "learning_rate": 1.8077125147567775e-05, "loss": 0.0372, "step": 723 }, { "epoch": 0.45, "learning_rate": 1.8071182579604986e-05, "loss": 0.0344, "step": 724 }, { "epoch": 0.45, "learning_rate": 1.806523182310868e-05, "loss": 0.0389, "step": 725 }, { "epoch": 0.45, "learning_rate": 1.805927288411614e-05, "loss": 0.0302, "step": 726 }, { "epoch": 0.45, "learning_rate": 1.8053305768672942e-05, "loss": 0.0436, "step": 727 }, { "epoch": 0.45, "learning_rate": 1.804733048283296e-05, "loss": 0.0355, "step": 728 }, { "epoch": 0.45, "learning_rate": 1.804134703265836e-05, "loss": 0.0345, "step": 729 }, { "epoch": 0.45, "learning_rate": 1.8035355424219586e-05, "loss": 0.0289, "step": 730 }, { "epoch": 0.45, "learning_rate": 1.8029355663595363e-05, "loss": 0.0395, "step": 731 }, { "epoch": 0.46, "learning_rate": 1.8023347756872686e-05, "loss": 0.0318, "step": 732 }, { "epoch": 0.46, "learning_rate": 1.801733171014681e-05, "loss": 0.0304, "step": 733 }, { "epoch": 0.46, "learning_rate": 1.8011307529521255e-05, "loss": 0.032, "step": 734 }, { "epoch": 0.46, "learning_rate": 1.800527522110779e-05, "loss": 0.0357, "step": 735 }, { "epoch": 0.46, "learning_rate": 1.799923479102643e-05, "loss": 0.0325, "step": 736 }, { "epoch": 0.46, "learning_rate": 1.799318624540543e-05, "loss": 0.0364, "step": 737 }, { "epoch": 0.46, "learning_rate": 1.7987129590381275e-05, "loss": 0.0403, "step": 738 }, { "epoch": 0.46, "learning_rate": 1.7981064832098687e-05, "loss": 0.0379, "step": 739 }, { "epoch": 0.46, "learning_rate": 1.79749919767106e-05, "loss": 0.0341, "step": 740 }, { "epoch": 0.46, "learning_rate": 1.7968911030378163e-05, "loss": 0.0344, "step": 741 }, { "epoch": 0.46, "learning_rate": 1.796282199927074e-05, "loss": 0.0345, "step": 742 }, { "epoch": 0.46, "learning_rate": 1.795672488956589e-05, "loss": 0.0372, "step": 743 }, { "epoch": 0.46, "learning_rate": 1.7950619707449374e-05, "loss": 0.0488, "step": 744 }, { "epoch": 0.46, "learning_rate": 1.794450645911514e-05, "loss": 0.0307, "step": 745 }, { "epoch": 0.46, "learning_rate": 1.7938385150765322e-05, "loss": 0.0349, "step": 746 }, { "epoch": 0.46, "learning_rate": 1.7932255788610223e-05, "loss": 0.0286, "step": 747 }, { "epoch": 0.47, "learning_rate": 1.792611837886833e-05, "loss": 0.0397, "step": 748 }, { "epoch": 0.47, "learning_rate": 1.7919972927766288e-05, "loss": 0.0391, "step": 749 }, { "epoch": 0.47, "learning_rate": 1.791381944153889e-05, "loss": 0.0412, "step": 750 }, { "epoch": 0.47, "learning_rate": 1.79076579264291e-05, "loss": 0.0467, "step": 751 }, { "epoch": 0.47, "learning_rate": 1.7901488388688015e-05, "loss": 0.042, "step": 752 }, { "epoch": 0.47, "learning_rate": 1.7895310834574873e-05, "loss": 0.036, "step": 753 }, { "epoch": 0.47, "learning_rate": 1.7889125270357053e-05, "loss": 0.0373, "step": 754 }, { "epoch": 0.47, "learning_rate": 1.788293170231005e-05, "loss": 0.0377, "step": 755 }, { "epoch": 0.47, "learning_rate": 1.7876730136717484e-05, "loss": 0.037, "step": 756 }, { "epoch": 0.47, "learning_rate": 1.7870520579871087e-05, "loss": 0.0272, "step": 757 }, { "epoch": 0.47, "learning_rate": 1.78643030380707e-05, "loss": 0.0366, "step": 758 }, { "epoch": 0.47, "learning_rate": 1.7858077517624265e-05, "loss": 0.0442, "step": 759 }, { "epoch": 0.47, "learning_rate": 1.785184402484782e-05, "loss": 0.0353, "step": 760 }, { "epoch": 0.47, "learning_rate": 1.7845602566065488e-05, "loss": 0.0328, "step": 761 }, { "epoch": 0.47, "learning_rate": 1.7839353147609476e-05, "loss": 0.0337, "step": 762 }, { "epoch": 0.47, "learning_rate": 1.7833095775820065e-05, "loss": 0.0367, "step": 763 }, { "epoch": 0.48, "learning_rate": 1.7826830457045608e-05, "loss": 0.0262, "step": 764 }, { "epoch": 0.48, "learning_rate": 1.7820557197642513e-05, "loss": 0.0385, "step": 765 }, { "epoch": 0.48, "learning_rate": 1.7814276003975255e-05, "loss": 0.0452, "step": 766 }, { "epoch": 0.48, "learning_rate": 1.7807986882416354e-05, "loss": 0.0365, "step": 767 }, { "epoch": 0.48, "learning_rate": 1.7801689839346366e-05, "loss": 0.0266, "step": 768 }, { "epoch": 0.48, "learning_rate": 1.7795384881153896e-05, "loss": 0.047, "step": 769 }, { "epoch": 0.48, "learning_rate": 1.778907201423557e-05, "loss": 0.0364, "step": 770 }, { "epoch": 0.48, "learning_rate": 1.7782751244996038e-05, "loss": 0.0301, "step": 771 }, { "epoch": 0.48, "learning_rate": 1.777642257984798e-05, "loss": 0.0346, "step": 772 }, { "epoch": 0.48, "learning_rate": 1.7770086025212064e-05, "loss": 0.0433, "step": 773 }, { "epoch": 0.48, "learning_rate": 1.7763741587516983e-05, "loss": 0.0341, "step": 774 }, { "epoch": 0.48, "learning_rate": 1.775738927319942e-05, "loss": 0.0374, "step": 775 }, { "epoch": 0.48, "learning_rate": 1.7751029088704044e-05, "loss": 0.0448, "step": 776 }, { "epoch": 0.48, "learning_rate": 1.7744661040483516e-05, "loss": 0.0352, "step": 777 }, { "epoch": 0.48, "learning_rate": 1.7738285134998473e-05, "loss": 0.039, "step": 778 }, { "epoch": 0.48, "learning_rate": 1.7731901378717523e-05, "loss": 0.042, "step": 779 }, { "epoch": 0.49, "learning_rate": 1.7725509778117242e-05, "loss": 0.0332, "step": 780 }, { "epoch": 0.49, "learning_rate": 1.771911033968215e-05, "loss": 0.0345, "step": 781 }, { "epoch": 0.49, "learning_rate": 1.771270306990474e-05, "loss": 0.0424, "step": 782 }, { "epoch": 0.49, "learning_rate": 1.770628797528544e-05, "loss": 0.0432, "step": 783 }, { "epoch": 0.49, "learning_rate": 1.769986506233261e-05, "loss": 0.0314, "step": 784 }, { "epoch": 0.49, "learning_rate": 1.7693434337562557e-05, "loss": 0.0335, "step": 785 }, { "epoch": 0.49, "learning_rate": 1.7686995807499502e-05, "loss": 0.0249, "step": 786 }, { "epoch": 0.49, "learning_rate": 1.768054947867559e-05, "loss": 0.0329, "step": 787 }, { "epoch": 0.49, "learning_rate": 1.767409535763087e-05, "loss": 0.0334, "step": 788 }, { "epoch": 0.49, "learning_rate": 1.7667633450913307e-05, "loss": 0.0337, "step": 789 }, { "epoch": 0.49, "learning_rate": 1.766116376507876e-05, "loss": 0.0403, "step": 790 }, { "epoch": 0.49, "learning_rate": 1.7654686306690983e-05, "loss": 0.0331, "step": 791 }, { "epoch": 0.49, "learning_rate": 1.764820108232161e-05, "loss": 0.0389, "step": 792 }, { "epoch": 0.49, "learning_rate": 1.7641708098550162e-05, "loss": 0.0272, "step": 793 }, { "epoch": 0.49, "learning_rate": 1.763520736196402e-05, "loss": 0.0403, "step": 794 }, { "epoch": 0.49, "learning_rate": 1.7628698879158444e-05, "loss": 0.0375, "step": 795 }, { "epoch": 0.5, "learning_rate": 1.7622182656736546e-05, "loss": 0.0426, "step": 796 }, { "epoch": 0.5, "learning_rate": 1.7615658701309287e-05, "loss": 0.0407, "step": 797 }, { "epoch": 0.5, "learning_rate": 1.7609127019495488e-05, "loss": 0.0359, "step": 798 }, { "epoch": 0.5, "learning_rate": 1.7602587617921785e-05, "loss": 0.0336, "step": 799 }, { "epoch": 0.5, "learning_rate": 1.7596040503222665e-05, "loss": 0.0328, "step": 800 }, { "epoch": 0.5, "learning_rate": 1.7589485682040433e-05, "loss": 0.0306, "step": 801 }, { "epoch": 0.5, "learning_rate": 1.758292316102522e-05, "loss": 0.0404, "step": 802 }, { "epoch": 0.5, "learning_rate": 1.7576352946834955e-05, "loss": 0.0294, "step": 803 }, { "epoch": 0.5, "learning_rate": 1.7569775046135388e-05, "loss": 0.037, "step": 804 }, { "epoch": 0.5, "learning_rate": 1.7563189465600047e-05, "loss": 0.034, "step": 805 }, { "epoch": 0.5, "learning_rate": 1.7556596211910275e-05, "loss": 0.03, "step": 806 }, { "epoch": 0.5, "learning_rate": 1.7549995291755182e-05, "loss": 0.0413, "step": 807 }, { "epoch": 0.5, "learning_rate": 1.7543386711831666e-05, "loss": 0.0323, "step": 808 }, { "epoch": 0.5, "learning_rate": 1.753677047884439e-05, "loss": 0.0334, "step": 809 }, { "epoch": 0.5, "learning_rate": 1.7530146599505782e-05, "loss": 0.0414, "step": 810 }, { "epoch": 0.5, "learning_rate": 1.7523515080536032e-05, "loss": 0.033, "step": 811 }, { "epoch": 0.5, "learning_rate": 1.751687592866308e-05, "loss": 0.0409, "step": 812 }, { "epoch": 0.51, "learning_rate": 1.7510229150622604e-05, "loss": 0.0281, "step": 813 }, { "epoch": 0.51, "learning_rate": 1.7503574753158022e-05, "loss": 0.0315, "step": 814 }, { "epoch": 0.51, "learning_rate": 1.7496912743020487e-05, "loss": 0.0349, "step": 815 }, { "epoch": 0.51, "learning_rate": 1.7490243126968868e-05, "loss": 0.039, "step": 816 }, { "epoch": 0.51, "learning_rate": 1.748356591176976e-05, "loss": 0.0369, "step": 817 }, { "epoch": 0.51, "learning_rate": 1.747688110419745e-05, "loss": 0.0378, "step": 818 }, { "epoch": 0.51, "learning_rate": 1.747018871103395e-05, "loss": 0.0365, "step": 819 }, { "epoch": 0.51, "learning_rate": 1.7463488739068952e-05, "loss": 0.0363, "step": 820 }, { "epoch": 0.51, "learning_rate": 1.7456781195099843e-05, "loss": 0.0432, "step": 821 }, { "epoch": 0.51, "learning_rate": 1.7450066085931694e-05, "loss": 0.0317, "step": 822 }, { "epoch": 0.51, "learning_rate": 1.7443343418377244e-05, "loss": 0.0338, "step": 823 }, { "epoch": 0.51, "learning_rate": 1.743661319925691e-05, "loss": 0.0381, "step": 824 }, { "epoch": 0.51, "learning_rate": 1.742987543539876e-05, "loss": 0.0377, "step": 825 }, { "epoch": 0.51, "learning_rate": 1.7423130133638524e-05, "loss": 0.0332, "step": 826 }, { "epoch": 0.51, "learning_rate": 1.741637730081958e-05, "loss": 0.0376, "step": 827 }, { "epoch": 0.51, "learning_rate": 1.7409616943792937e-05, "loss": 0.0402, "step": 828 }, { "epoch": 0.52, "learning_rate": 1.7402849069417246e-05, "loss": 0.0365, "step": 829 }, { "epoch": 0.52, "learning_rate": 1.7396073684558788e-05, "loss": 0.033, "step": 830 }, { "epoch": 0.52, "learning_rate": 1.7389290796091453e-05, "loss": 0.0291, "step": 831 }, { "epoch": 0.52, "learning_rate": 1.738250041089675e-05, "loss": 0.0425, "step": 832 }, { "epoch": 0.52, "learning_rate": 1.73757025358638e-05, "loss": 0.0311, "step": 833 }, { "epoch": 0.52, "learning_rate": 1.7368897177889307e-05, "loss": 0.03, "step": 834 }, { "epoch": 0.52, "learning_rate": 1.7362084343877576e-05, "loss": 0.0438, "step": 835 }, { "epoch": 0.52, "learning_rate": 1.73552640407405e-05, "loss": 0.0348, "step": 836 }, { "epoch": 0.52, "learning_rate": 1.7348436275397547e-05, "loss": 0.0295, "step": 837 }, { "epoch": 0.52, "learning_rate": 1.734160105477575e-05, "loss": 0.0266, "step": 838 }, { "epoch": 0.52, "learning_rate": 1.7334758385809715e-05, "loss": 0.0336, "step": 839 }, { "epoch": 0.52, "learning_rate": 1.732790827544159e-05, "loss": 0.0348, "step": 840 }, { "epoch": 0.52, "learning_rate": 1.7321050730621095e-05, "loss": 0.0307, "step": 841 }, { "epoch": 0.52, "learning_rate": 1.731418575830548e-05, "loss": 0.0329, "step": 842 }, { "epoch": 0.52, "learning_rate": 1.7307313365459516e-05, "loss": 0.0327, "step": 843 }, { "epoch": 0.52, "learning_rate": 1.7300433559055533e-05, "loss": 0.0337, "step": 844 }, { "epoch": 0.53, "learning_rate": 1.729354634607336e-05, "loss": 0.0328, "step": 845 }, { "epoch": 0.53, "learning_rate": 1.7286651733500345e-05, "loss": 0.0364, "step": 846 }, { "epoch": 0.53, "learning_rate": 1.7279749728331346e-05, "loss": 0.0331, "step": 847 }, { "epoch": 0.53, "learning_rate": 1.727284033756872e-05, "loss": 0.0293, "step": 848 }, { "epoch": 0.53, "learning_rate": 1.7265923568222315e-05, "loss": 0.0332, "step": 849 }, { "epoch": 0.53, "learning_rate": 1.725899942730947e-05, "loss": 0.0435, "step": 850 }, { "epoch": 0.53, "learning_rate": 1.7252067921855e-05, "loss": 0.0337, "step": 851 }, { "epoch": 0.53, "learning_rate": 1.7245129058891187e-05, "loss": 0.0393, "step": 852 }, { "epoch": 0.53, "learning_rate": 1.7238182845457786e-05, "loss": 0.0354, "step": 853 }, { "epoch": 0.53, "learning_rate": 1.7231229288602e-05, "loss": 0.0338, "step": 854 }, { "epoch": 0.53, "learning_rate": 1.7224268395378493e-05, "loss": 0.0312, "step": 855 }, { "epoch": 0.53, "learning_rate": 1.7217300172849357e-05, "loss": 0.0322, "step": 856 }, { "epoch": 0.53, "learning_rate": 1.721032462808414e-05, "loss": 0.0434, "step": 857 }, { "epoch": 0.53, "learning_rate": 1.7203341768159798e-05, "loss": 0.0363, "step": 858 }, { "epoch": 0.53, "learning_rate": 1.7196351600160725e-05, "loss": 0.0443, "step": 859 }, { "epoch": 0.53, "learning_rate": 1.718935413117872e-05, "loss": 0.0368, "step": 860 }, { "epoch": 0.54, "learning_rate": 1.7182349368312992e-05, "loss": 0.0344, "step": 861 }, { "epoch": 0.54, "learning_rate": 1.717533731867015e-05, "loss": 0.0387, "step": 862 }, { "epoch": 0.54, "learning_rate": 1.7168317989364192e-05, "loss": 0.0357, "step": 863 }, { "epoch": 0.54, "learning_rate": 1.716129138751651e-05, "loss": 0.04, "step": 864 }, { "epoch": 0.54, "learning_rate": 1.7154257520255864e-05, "loss": 0.047, "step": 865 }, { "epoch": 0.54, "learning_rate": 1.7147216394718395e-05, "loss": 0.0474, "step": 866 }, { "epoch": 0.54, "learning_rate": 1.71401680180476e-05, "loss": 0.0263, "step": 867 }, { "epoch": 0.54, "learning_rate": 1.7133112397394336e-05, "loss": 0.0345, "step": 868 }, { "epoch": 0.54, "learning_rate": 1.712604953991681e-05, "loss": 0.0445, "step": 869 }, { "epoch": 0.54, "learning_rate": 1.711897945278057e-05, "loss": 0.0352, "step": 870 }, { "epoch": 0.54, "learning_rate": 1.7111902143158495e-05, "loss": 0.0304, "step": 871 }, { "epoch": 0.54, "learning_rate": 1.7104817618230797e-05, "loss": 0.0366, "step": 872 }, { "epoch": 0.54, "learning_rate": 1.7097725885185008e-05, "loss": 0.0287, "step": 873 }, { "epoch": 0.54, "learning_rate": 1.709062695121597e-05, "loss": 0.0307, "step": 874 }, { "epoch": 0.54, "learning_rate": 1.7083520823525836e-05, "loss": 0.0291, "step": 875 }, { "epoch": 0.54, "learning_rate": 1.7076407509324046e-05, "loss": 0.0336, "step": 876 }, { "epoch": 0.55, "learning_rate": 1.7069287015827343e-05, "loss": 0.0362, "step": 877 }, { "epoch": 0.55, "learning_rate": 1.7062159350259747e-05, "loss": 0.0371, "step": 878 }, { "epoch": 0.55, "learning_rate": 1.7055024519852554e-05, "loss": 0.033, "step": 879 }, { "epoch": 0.55, "learning_rate": 1.7047882531844338e-05, "loss": 0.0363, "step": 880 }, { "epoch": 0.55, "learning_rate": 1.704073339348092e-05, "loss": 0.0361, "step": 881 }, { "epoch": 0.55, "learning_rate": 1.703357711201539e-05, "loss": 0.0388, "step": 882 }, { "epoch": 0.55, "learning_rate": 1.7026413694708077e-05, "loss": 0.0428, "step": 883 }, { "epoch": 0.55, "learning_rate": 1.7019243148826547e-05, "loss": 0.0357, "step": 884 }, { "epoch": 0.55, "learning_rate": 1.7012065481645604e-05, "loss": 0.026, "step": 885 }, { "epoch": 0.55, "learning_rate": 1.7004880700447278e-05, "loss": 0.03, "step": 886 }, { "epoch": 0.55, "learning_rate": 1.699768881252081e-05, "loss": 0.0355, "step": 887 }, { "epoch": 0.55, "learning_rate": 1.6990489825162647e-05, "loss": 0.0313, "step": 888 }, { "epoch": 0.55, "learning_rate": 1.6983283745676464e-05, "loss": 0.0359, "step": 889 }, { "epoch": 0.55, "learning_rate": 1.6976070581373098e-05, "loss": 0.0334, "step": 890 }, { "epoch": 0.55, "learning_rate": 1.6968850339570595e-05, "loss": 0.0342, "step": 891 }, { "epoch": 0.55, "learning_rate": 1.6961623027594175e-05, "loss": 0.0306, "step": 892 }, { "epoch": 0.56, "learning_rate": 1.695438865277623e-05, "loss": 0.0307, "step": 893 }, { "epoch": 0.56, "learning_rate": 1.6947147222456318e-05, "loss": 0.0217, "step": 894 }, { "epoch": 0.56, "learning_rate": 1.693989874398116e-05, "loss": 0.0309, "step": 895 }, { "epoch": 0.56, "learning_rate": 1.6932643224704623e-05, "loss": 0.0475, "step": 896 }, { "epoch": 0.56, "learning_rate": 1.6925380671987715e-05, "loss": 0.0323, "step": 897 }, { "epoch": 0.56, "learning_rate": 1.6918111093198585e-05, "loss": 0.0241, "step": 898 }, { "epoch": 0.56, "learning_rate": 1.6910834495712504e-05, "loss": 0.041, "step": 899 }, { "epoch": 0.56, "learning_rate": 1.6903550886911873e-05, "loss": 0.0292, "step": 900 }, { "epoch": 0.56, "learning_rate": 1.68962602741862e-05, "loss": 0.0414, "step": 901 }, { "epoch": 0.56, "learning_rate": 1.6888962664932095e-05, "loss": 0.038, "step": 902 }, { "epoch": 0.56, "learning_rate": 1.6881658066553276e-05, "loss": 0.0419, "step": 903 }, { "epoch": 0.56, "learning_rate": 1.6874346486460543e-05, "loss": 0.0468, "step": 904 }, { "epoch": 0.56, "learning_rate": 1.6867027932071786e-05, "loss": 0.0297, "step": 905 }, { "epoch": 0.56, "learning_rate": 1.6859702410811968e-05, "loss": 0.044, "step": 906 }, { "epoch": 0.56, "learning_rate": 1.6852369930113113e-05, "loss": 0.0315, "step": 907 }, { "epoch": 0.56, "learning_rate": 1.6845030497414316e-05, "loss": 0.0359, "step": 908 }, { "epoch": 0.57, "learning_rate": 1.6837684120161723e-05, "loss": 0.0353, "step": 909 }, { "epoch": 0.57, "learning_rate": 1.6830330805808524e-05, "loss": 0.0383, "step": 910 }, { "epoch": 0.57, "learning_rate": 1.6822970561814944e-05, "loss": 0.0339, "step": 911 }, { "epoch": 0.57, "learning_rate": 1.6815603395648243e-05, "loss": 0.0324, "step": 912 }, { "epoch": 0.57, "learning_rate": 1.6808229314782704e-05, "loss": 0.0275, "step": 913 }, { "epoch": 0.57, "learning_rate": 1.680084832669962e-05, "loss": 0.0352, "step": 914 }, { "epoch": 0.57, "learning_rate": 1.6793460438887294e-05, "loss": 0.0344, "step": 915 }, { "epoch": 0.57, "learning_rate": 1.6786065658841037e-05, "loss": 0.0263, "step": 916 }, { "epoch": 0.57, "learning_rate": 1.6778663994063137e-05, "loss": 0.0262, "step": 917 }, { "epoch": 0.57, "learning_rate": 1.677125545206288e-05, "loss": 0.0427, "step": 918 }, { "epoch": 0.57, "learning_rate": 1.6763840040356522e-05, "loss": 0.0335, "step": 919 }, { "epoch": 0.57, "learning_rate": 1.6756417766467294e-05, "loss": 0.0379, "step": 920 }, { "epoch": 0.57, "learning_rate": 1.6748988637925384e-05, "loss": 0.0332, "step": 921 }, { "epoch": 0.57, "learning_rate": 1.6741552662267934e-05, "loss": 0.0369, "step": 922 }, { "epoch": 0.57, "learning_rate": 1.673410984703904e-05, "loss": 0.0378, "step": 923 }, { "epoch": 0.57, "learning_rate": 1.6726660199789733e-05, "loss": 0.0322, "step": 924 }, { "epoch": 0.58, "learning_rate": 1.6719203728077963e-05, "loss": 0.0278, "step": 925 }, { "epoch": 0.58, "learning_rate": 1.6711740439468632e-05, "loss": 0.0356, "step": 926 }, { "epoch": 0.58, "learning_rate": 1.6704270341533527e-05, "loss": 0.0311, "step": 927 }, { "epoch": 0.58, "learning_rate": 1.6696793441851366e-05, "loss": 0.0311, "step": 928 }, { "epoch": 0.58, "learning_rate": 1.6689309748007753e-05, "loss": 0.0353, "step": 929 }, { "epoch": 0.58, "learning_rate": 1.6681819267595193e-05, "loss": 0.038, "step": 930 }, { "epoch": 0.58, "learning_rate": 1.6674322008213076e-05, "loss": 0.0438, "step": 931 }, { "epoch": 0.58, "learning_rate": 1.6666817977467667e-05, "loss": 0.0377, "step": 932 }, { "epoch": 0.58, "learning_rate": 1.66593071829721e-05, "loss": 0.0346, "step": 933 }, { "epoch": 0.58, "learning_rate": 1.6651789632346377e-05, "loss": 0.034, "step": 934 }, { "epoch": 0.58, "learning_rate": 1.6644265333217347e-05, "loss": 0.0337, "step": 935 }, { "epoch": 0.58, "learning_rate": 1.6636734293218702e-05, "loss": 0.0349, "step": 936 }, { "epoch": 0.58, "learning_rate": 1.6629196519990994e-05, "loss": 0.0331, "step": 937 }, { "epoch": 0.58, "learning_rate": 1.662165202118158e-05, "loss": 0.0342, "step": 938 }, { "epoch": 0.58, "learning_rate": 1.6614100804444657e-05, "loss": 0.0263, "step": 939 }, { "epoch": 0.58, "learning_rate": 1.660654287744123e-05, "loss": 0.0359, "step": 940 }, { "epoch": 0.59, "learning_rate": 1.6598978247839118e-05, "loss": 0.0313, "step": 941 }, { "epoch": 0.59, "learning_rate": 1.659140692331293e-05, "loss": 0.0307, "step": 942 }, { "epoch": 0.59, "learning_rate": 1.6583828911544083e-05, "loss": 0.0287, "step": 943 }, { "epoch": 0.59, "learning_rate": 1.6576244220220763e-05, "loss": 0.0326, "step": 944 }, { "epoch": 0.59, "learning_rate": 1.656865285703794e-05, "loss": 0.0351, "step": 945 }, { "epoch": 0.59, "learning_rate": 1.656105482969735e-05, "loss": 0.025, "step": 946 }, { "epoch": 0.59, "learning_rate": 1.6553450145907497e-05, "loss": 0.0291, "step": 947 }, { "epoch": 0.59, "learning_rate": 1.654583881338363e-05, "loss": 0.0285, "step": 948 }, { "epoch": 0.59, "learning_rate": 1.6538220839847745e-05, "loss": 0.0275, "step": 949 }, { "epoch": 0.59, "learning_rate": 1.653059623302858e-05, "loss": 0.0455, "step": 950 }, { "epoch": 0.59, "learning_rate": 1.6522965000661596e-05, "loss": 0.0331, "step": 951 }, { "epoch": 0.59, "learning_rate": 1.6515327150488984e-05, "loss": 0.0314, "step": 952 }, { "epoch": 0.59, "learning_rate": 1.6507682690259643e-05, "loss": 0.0205, "step": 953 }, { "epoch": 0.59, "learning_rate": 1.6500031627729178e-05, "loss": 0.0311, "step": 954 }, { "epoch": 0.59, "learning_rate": 1.6492373970659897e-05, "loss": 0.031, "step": 955 }, { "epoch": 0.59, "learning_rate": 1.648470972682079e-05, "loss": 0.0339, "step": 956 }, { "epoch": 0.6, "learning_rate": 1.6477038903987547e-05, "loss": 0.0415, "step": 957 }, { "epoch": 0.6, "learning_rate": 1.6469361509942507e-05, "loss": 0.0363, "step": 958 }, { "epoch": 0.6, "learning_rate": 1.6461677552474698e-05, "loss": 0.0374, "step": 959 }, { "epoch": 0.6, "learning_rate": 1.645398703937979e-05, "loss": 0.0338, "step": 960 }, { "epoch": 0.6, "learning_rate": 1.6446289978460126e-05, "loss": 0.0341, "step": 961 }, { "epoch": 0.6, "learning_rate": 1.643858637752467e-05, "loss": 0.0226, "step": 962 }, { "epoch": 0.6, "learning_rate": 1.6430876244389026e-05, "loss": 0.0373, "step": 963 }, { "epoch": 0.6, "learning_rate": 1.642315958687543e-05, "loss": 0.0378, "step": 964 }, { "epoch": 0.6, "learning_rate": 1.6415436412812742e-05, "loss": 0.0298, "step": 965 }, { "epoch": 0.6, "learning_rate": 1.6407706730036425e-05, "loss": 0.0427, "step": 966 }, { "epoch": 0.6, "learning_rate": 1.6399970546388544e-05, "loss": 0.0362, "step": 967 }, { "epoch": 0.6, "learning_rate": 1.6392227869717766e-05, "loss": 0.0407, "step": 968 }, { "epoch": 0.6, "learning_rate": 1.6384478707879337e-05, "loss": 0.0349, "step": 969 }, { "epoch": 0.6, "learning_rate": 1.6376723068735096e-05, "loss": 0.0302, "step": 970 }, { "epoch": 0.6, "learning_rate": 1.636896096015344e-05, "loss": 0.0291, "step": 971 }, { "epoch": 0.6, "learning_rate": 1.6361192390009334e-05, "loss": 0.0416, "step": 972 }, { "epoch": 0.61, "learning_rate": 1.6353417366184305e-05, "loss": 0.0428, "step": 973 }, { "epoch": 0.61, "learning_rate": 1.6345635896566415e-05, "loss": 0.03, "step": 974 }, { "epoch": 0.61, "learning_rate": 1.6337847989050276e-05, "loss": 0.0257, "step": 975 }, { "epoch": 0.61, "learning_rate": 1.6330053651537024e-05, "loss": 0.0283, "step": 976 }, { "epoch": 0.61, "learning_rate": 1.6322252891934326e-05, "loss": 0.0291, "step": 977 }, { "epoch": 0.61, "learning_rate": 1.631444571815636e-05, "loss": 0.0318, "step": 978 }, { "epoch": 0.61, "learning_rate": 1.6306632138123814e-05, "loss": 0.0318, "step": 979 }, { "epoch": 0.61, "learning_rate": 1.6298812159763868e-05, "loss": 0.0336, "step": 980 }, { "epoch": 0.61, "learning_rate": 1.6290985791010204e-05, "loss": 0.04, "step": 981 }, { "epoch": 0.61, "learning_rate": 1.628315303980298e-05, "loss": 0.0261, "step": 982 }, { "epoch": 0.61, "learning_rate": 1.6275313914088834e-05, "loss": 0.0398, "step": 983 }, { "epoch": 0.61, "learning_rate": 1.626746842182087e-05, "loss": 0.0333, "step": 984 }, { "epoch": 0.61, "learning_rate": 1.6259616570958638e-05, "loss": 0.0362, "step": 985 }, { "epoch": 0.61, "learning_rate": 1.6251758369468166e-05, "loss": 0.0344, "step": 986 }, { "epoch": 0.61, "learning_rate": 1.62438938253219e-05, "loss": 0.0369, "step": 987 }, { "epoch": 0.61, "learning_rate": 1.6236022946498734e-05, "loss": 0.0275, "step": 988 }, { "epoch": 0.62, "learning_rate": 1.6228145740983986e-05, "loss": 0.0278, "step": 989 }, { "epoch": 0.62, "learning_rate": 1.6220262216769393e-05, "loss": 0.0336, "step": 990 }, { "epoch": 0.62, "learning_rate": 1.6212372381853096e-05, "loss": 0.0301, "step": 991 }, { "epoch": 0.62, "learning_rate": 1.620447624423965e-05, "loss": 0.0262, "step": 992 }, { "epoch": 0.62, "learning_rate": 1.6196573811939996e-05, "loss": 0.0271, "step": 993 }, { "epoch": 0.62, "learning_rate": 1.618866509297147e-05, "loss": 0.042, "step": 994 }, { "epoch": 0.62, "learning_rate": 1.6180750095357766e-05, "loss": 0.034, "step": 995 }, { "epoch": 0.62, "learning_rate": 1.6172828827128976e-05, "loss": 0.0274, "step": 996 }, { "epoch": 0.62, "learning_rate": 1.6164901296321528e-05, "loss": 0.0324, "step": 997 }, { "epoch": 0.62, "learning_rate": 1.6156967510978224e-05, "loss": 0.0326, "step": 998 }, { "epoch": 0.62, "learning_rate": 1.61490274791482e-05, "loss": 0.0351, "step": 999 }, { "epoch": 0.62, "learning_rate": 1.614108120888693e-05, "loss": 0.0394, "step": 1000 }, { "epoch": 0.62, "learning_rate": 1.6133128708256217e-05, "loss": 0.0375, "step": 1001 }, { "epoch": 0.62, "learning_rate": 1.612516998532419e-05, "loss": 0.0363, "step": 1002 }, { "epoch": 0.62, "learning_rate": 1.611720504816529e-05, "loss": 0.0346, "step": 1003 }, { "epoch": 0.62, "learning_rate": 1.6109233904860258e-05, "loss": 0.0434, "step": 1004 }, { "epoch": 0.62, "learning_rate": 1.6101256563496133e-05, "loss": 0.043, "step": 1005 }, { "epoch": 0.63, "learning_rate": 1.6093273032166242e-05, "loss": 0.0289, "step": 1006 }, { "epoch": 0.63, "learning_rate": 1.60852833189702e-05, "loss": 0.0297, "step": 1007 }, { "epoch": 0.63, "learning_rate": 1.6077287432013875e-05, "loss": 0.035, "step": 1008 }, { "epoch": 0.63, "learning_rate": 1.606928537940942e-05, "loss": 0.0346, "step": 1009 }, { "epoch": 0.63, "learning_rate": 1.6061277169275227e-05, "loss": 0.0319, "step": 1010 }, { "epoch": 0.63, "learning_rate": 1.6053262809735946e-05, "loss": 0.0266, "step": 1011 }, { "epoch": 0.63, "learning_rate": 1.604524230892246e-05, "loss": 0.035, "step": 1012 }, { "epoch": 0.63, "learning_rate": 1.6037215674971882e-05, "loss": 0.026, "step": 1013 }, { "epoch": 0.63, "learning_rate": 1.602918291602755e-05, "loss": 0.0338, "step": 1014 }, { "epoch": 0.63, "learning_rate": 1.6021144040239017e-05, "loss": 0.0337, "step": 1015 }, { "epoch": 0.63, "learning_rate": 1.6013099055762037e-05, "loss": 0.0414, "step": 1016 }, { "epoch": 0.63, "learning_rate": 1.6005047970758567e-05, "loss": 0.0252, "step": 1017 }, { "epoch": 0.63, "learning_rate": 1.5996990793396753e-05, "loss": 0.0285, "step": 1018 }, { "epoch": 0.63, "learning_rate": 1.5988927531850913e-05, "loss": 0.0333, "step": 1019 }, { "epoch": 0.63, "learning_rate": 1.598085819430155e-05, "loss": 0.0319, "step": 1020 }, { "epoch": 0.63, "learning_rate": 1.5972782788935318e-05, "loss": 0.0314, "step": 1021 }, { "epoch": 0.64, "learning_rate": 1.5964701323945046e-05, "loss": 0.0256, "step": 1022 }, { "epoch": 0.64, "learning_rate": 1.595661380752969e-05, "loss": 0.0298, "step": 1023 }, { "epoch": 0.64, "learning_rate": 1.5948520247894363e-05, "loss": 0.0328, "step": 1024 }, { "epoch": 0.64, "learning_rate": 1.5940420653250293e-05, "loss": 0.0307, "step": 1025 }, { "epoch": 0.64, "learning_rate": 1.5932315031814843e-05, "loss": 0.028, "step": 1026 }, { "epoch": 0.64, "learning_rate": 1.5924203391811486e-05, "loss": 0.028, "step": 1027 }, { "epoch": 0.64, "learning_rate": 1.5916085741469796e-05, "loss": 0.028, "step": 1028 }, { "epoch": 0.64, "learning_rate": 1.590796208902546e-05, "loss": 0.0365, "step": 1029 }, { "epoch": 0.64, "learning_rate": 1.589983244272024e-05, "loss": 0.0349, "step": 1030 }, { "epoch": 0.64, "learning_rate": 1.589169681080198e-05, "loss": 0.0284, "step": 1031 }, { "epoch": 0.64, "learning_rate": 1.5883555201524597e-05, "loss": 0.0256, "step": 1032 }, { "epoch": 0.64, "learning_rate": 1.587540762314808e-05, "loss": 0.0208, "step": 1033 }, { "epoch": 0.64, "learning_rate": 1.5867254083938472e-05, "loss": 0.0287, "step": 1034 }, { "epoch": 0.64, "learning_rate": 1.5859094592167858e-05, "loss": 0.0261, "step": 1035 }, { "epoch": 0.64, "learning_rate": 1.5850929156114363e-05, "loss": 0.0303, "step": 1036 }, { "epoch": 0.64, "learning_rate": 1.5842757784062145e-05, "loss": 0.026, "step": 1037 }, { "epoch": 0.65, "learning_rate": 1.583458048430139e-05, "loss": 0.0289, "step": 1038 }, { "epoch": 0.65, "learning_rate": 1.582639726512828e-05, "loss": 0.0349, "step": 1039 }, { "epoch": 0.65, "learning_rate": 1.5818208134845022e-05, "loss": 0.0431, "step": 1040 }, { "epoch": 0.65, "learning_rate": 1.5810013101759806e-05, "loss": 0.0298, "step": 1041 }, { "epoch": 0.65, "learning_rate": 1.5801812174186825e-05, "loss": 0.0307, "step": 1042 }, { "epoch": 0.65, "learning_rate": 1.579360536044624e-05, "loss": 0.027, "step": 1043 }, { "epoch": 0.65, "learning_rate": 1.5785392668864186e-05, "loss": 0.0301, "step": 1044 }, { "epoch": 0.65, "learning_rate": 1.577717410777276e-05, "loss": 0.0267, "step": 1045 }, { "epoch": 0.65, "learning_rate": 1.576894968551003e-05, "loss": 0.0281, "step": 1046 }, { "epoch": 0.65, "learning_rate": 1.5760719410419984e-05, "loss": 0.0377, "step": 1047 }, { "epoch": 0.65, "learning_rate": 1.5752483290852566e-05, "loss": 0.0389, "step": 1048 }, { "epoch": 0.65, "learning_rate": 1.5744241335163642e-05, "loss": 0.0295, "step": 1049 }, { "epoch": 0.65, "learning_rate": 1.5735993551715006e-05, "loss": 0.0286, "step": 1050 }, { "epoch": 0.65, "learning_rate": 1.572773994887436e-05, "loss": 0.0286, "step": 1051 }, { "epoch": 0.65, "learning_rate": 1.5719480535015303e-05, "loss": 0.0285, "step": 1052 }, { "epoch": 0.65, "learning_rate": 1.5711215318517345e-05, "loss": 0.0302, "step": 1053 }, { "epoch": 0.66, "learning_rate": 1.570294430776587e-05, "loss": 0.0328, "step": 1054 }, { "epoch": 0.66, "learning_rate": 1.5694667511152143e-05, "loss": 0.0345, "step": 1055 }, { "epoch": 0.66, "learning_rate": 1.5686384937073307e-05, "loss": 0.0392, "step": 1056 }, { "epoch": 0.66, "learning_rate": 1.567809659393236e-05, "loss": 0.0311, "step": 1057 }, { "epoch": 0.66, "learning_rate": 1.5669802490138152e-05, "loss": 0.0413, "step": 1058 }, { "epoch": 0.66, "learning_rate": 1.5661502634105376e-05, "loss": 0.0398, "step": 1059 }, { "epoch": 0.66, "learning_rate": 1.5653197034254564e-05, "loss": 0.0401, "step": 1060 }, { "epoch": 0.66, "learning_rate": 1.564488569901208e-05, "loss": 0.0405, "step": 1061 }, { "epoch": 0.66, "learning_rate": 1.5636568636810094e-05, "loss": 0.0237, "step": 1062 }, { "epoch": 0.66, "learning_rate": 1.5628245856086605e-05, "loss": 0.0293, "step": 1063 }, { "epoch": 0.66, "learning_rate": 1.5619917365285394e-05, "loss": 0.0382, "step": 1064 }, { "epoch": 0.66, "learning_rate": 1.5611583172856045e-05, "loss": 0.0411, "step": 1065 }, { "epoch": 0.66, "learning_rate": 1.560324328725393e-05, "loss": 0.0377, "step": 1066 }, { "epoch": 0.66, "learning_rate": 1.5594897716940186e-05, "loss": 0.0319, "step": 1067 }, { "epoch": 0.66, "learning_rate": 1.5586546470381732e-05, "loss": 0.0345, "step": 1068 }, { "epoch": 0.66, "learning_rate": 1.557818955605123e-05, "loss": 0.0289, "step": 1069 }, { "epoch": 0.67, "learning_rate": 1.55698269824271e-05, "loss": 0.0327, "step": 1070 }, { "epoch": 0.67, "learning_rate": 1.556145875799351e-05, "loss": 0.0276, "step": 1071 }, { "epoch": 0.67, "learning_rate": 1.5553084891240344e-05, "loss": 0.0342, "step": 1072 }, { "epoch": 0.67, "learning_rate": 1.554470539066323e-05, "loss": 0.0386, "step": 1073 }, { "epoch": 0.67, "learning_rate": 1.55363202647635e-05, "loss": 0.0297, "step": 1074 }, { "epoch": 0.67, "learning_rate": 1.552792952204819e-05, "loss": 0.0284, "step": 1075 }, { "epoch": 0.67, "learning_rate": 1.5519533171030046e-05, "loss": 0.0272, "step": 1076 }, { "epoch": 0.67, "learning_rate": 1.5511131220227495e-05, "loss": 0.0359, "step": 1077 }, { "epoch": 0.67, "learning_rate": 1.550272367816465e-05, "loss": 0.033, "step": 1078 }, { "epoch": 0.67, "learning_rate": 1.5494310553371292e-05, "loss": 0.0246, "step": 1079 }, { "epoch": 0.67, "learning_rate": 1.5485891854382868e-05, "loss": 0.0298, "step": 1080 }, { "epoch": 0.67, "learning_rate": 1.547746758974048e-05, "loss": 0.0282, "step": 1081 }, { "epoch": 0.67, "learning_rate": 1.5469037767990887e-05, "loss": 0.0299, "step": 1082 }, { "epoch": 0.67, "learning_rate": 1.5460602397686463e-05, "loss": 0.0283, "step": 1083 }, { "epoch": 0.67, "learning_rate": 1.545216148738523e-05, "loss": 0.0335, "step": 1084 }, { "epoch": 0.67, "learning_rate": 1.544371504565082e-05, "loss": 0.0245, "step": 1085 }, { "epoch": 0.68, "learning_rate": 1.5435263081052495e-05, "loss": 0.0257, "step": 1086 }, { "epoch": 0.68, "learning_rate": 1.5426805602165097e-05, "loss": 0.0334, "step": 1087 }, { "epoch": 0.68, "learning_rate": 1.541834261756907e-05, "loss": 0.0365, "step": 1088 }, { "epoch": 0.68, "learning_rate": 1.5409874135850453e-05, "loss": 0.0334, "step": 1089 }, { "epoch": 0.68, "learning_rate": 1.5401400165600848e-05, "loss": 0.0433, "step": 1090 }, { "epoch": 0.68, "learning_rate": 1.539292071541744e-05, "loss": 0.0268, "step": 1091 }, { "epoch": 0.68, "learning_rate": 1.538443579390296e-05, "loss": 0.026, "step": 1092 }, { "epoch": 0.68, "learning_rate": 1.53759454096657e-05, "loss": 0.0253, "step": 1093 }, { "epoch": 0.68, "learning_rate": 1.5367449571319486e-05, "loss": 0.03, "step": 1094 }, { "epoch": 0.68, "learning_rate": 1.5358948287483688e-05, "loss": 0.0294, "step": 1095 }, { "epoch": 0.68, "learning_rate": 1.535044156678319e-05, "loss": 0.0368, "step": 1096 }, { "epoch": 0.68, "learning_rate": 1.5341929417848392e-05, "loss": 0.0366, "step": 1097 }, { "epoch": 0.68, "learning_rate": 1.5333411849315215e-05, "loss": 0.0416, "step": 1098 }, { "epoch": 0.68, "learning_rate": 1.5324888869825062e-05, "loss": 0.0368, "step": 1099 }, { "epoch": 0.68, "learning_rate": 1.5316360488024838e-05, "loss": 0.0312, "step": 1100 }, { "epoch": 0.68, "learning_rate": 1.530782671256691e-05, "loss": 0.0313, "step": 1101 }, { "epoch": 0.69, "learning_rate": 1.529928755210915e-05, "loss": 0.0393, "step": 1102 }, { "epoch": 0.69, "learning_rate": 1.5290743015314857e-05, "loss": 0.036, "step": 1103 }, { "epoch": 0.69, "learning_rate": 1.5282193110852806e-05, "loss": 0.0303, "step": 1104 }, { "epoch": 0.69, "learning_rate": 1.5273637847397216e-05, "loss": 0.0319, "step": 1105 }, { "epoch": 0.69, "learning_rate": 1.5265077233627737e-05, "loss": 0.0349, "step": 1106 }, { "epoch": 0.69, "learning_rate": 1.5256511278229451e-05, "loss": 0.0306, "step": 1107 }, { "epoch": 0.69, "learning_rate": 1.5247939989892859e-05, "loss": 0.0349, "step": 1108 }, { "epoch": 0.69, "learning_rate": 1.5239363377313864e-05, "loss": 0.0394, "step": 1109 }, { "epoch": 0.69, "learning_rate": 1.5230781449193787e-05, "loss": 0.0308, "step": 1110 }, { "epoch": 0.69, "learning_rate": 1.5222194214239334e-05, "loss": 0.0267, "step": 1111 }, { "epoch": 0.69, "learning_rate": 1.5213601681162587e-05, "loss": 0.0292, "step": 1112 }, { "epoch": 0.69, "learning_rate": 1.5205003858681013e-05, "loss": 0.0338, "step": 1113 }, { "epoch": 0.69, "learning_rate": 1.5196400755517445e-05, "loss": 0.036, "step": 1114 }, { "epoch": 0.69, "learning_rate": 1.5187792380400066e-05, "loss": 0.0284, "step": 1115 }, { "epoch": 0.69, "learning_rate": 1.517917874206242e-05, "loss": 0.0298, "step": 1116 }, { "epoch": 0.69, "learning_rate": 1.5170559849243374e-05, "loss": 0.0306, "step": 1117 }, { "epoch": 0.7, "learning_rate": 1.5161935710687142e-05, "loss": 0.0406, "step": 1118 }, { "epoch": 0.7, "learning_rate": 1.5153306335143247e-05, "loss": 0.029, "step": 1119 }, { "epoch": 0.7, "learning_rate": 1.5144671731366537e-05, "loss": 0.0286, "step": 1120 }, { "epoch": 0.7, "learning_rate": 1.5136031908117162e-05, "loss": 0.0341, "step": 1121 }, { "epoch": 0.7, "learning_rate": 1.5127386874160558e-05, "loss": 0.0302, "step": 1122 }, { "epoch": 0.7, "learning_rate": 1.5118736638267449e-05, "loss": 0.0298, "step": 1123 }, { "epoch": 0.7, "learning_rate": 1.5110081209213849e-05, "loss": 0.0273, "step": 1124 }, { "epoch": 0.7, "learning_rate": 1.510142059578103e-05, "loss": 0.035, "step": 1125 }, { "epoch": 0.7, "learning_rate": 1.5092754806755527e-05, "loss": 0.0308, "step": 1126 }, { "epoch": 0.7, "learning_rate": 1.5084083850929125e-05, "loss": 0.0404, "step": 1127 }, { "epoch": 0.7, "learning_rate": 1.5075407737098849e-05, "loss": 0.0385, "step": 1128 }, { "epoch": 0.7, "learning_rate": 1.5066726474066962e-05, "loss": 0.0276, "step": 1129 }, { "epoch": 0.7, "learning_rate": 1.5058040070640947e-05, "loss": 0.0362, "step": 1130 }, { "epoch": 0.7, "learning_rate": 1.5049348535633503e-05, "loss": 0.0377, "step": 1131 }, { "epoch": 0.7, "learning_rate": 1.5040651877862535e-05, "loss": 0.0365, "step": 1132 }, { "epoch": 0.7, "learning_rate": 1.5031950106151149e-05, "loss": 0.0341, "step": 1133 }, { "epoch": 0.71, "learning_rate": 1.5023243229327631e-05, "loss": 0.0365, "step": 1134 }, { "epoch": 0.71, "learning_rate": 1.5014531256225459e-05, "loss": 0.0349, "step": 1135 }, { "epoch": 0.71, "learning_rate": 1.5005814195683264e-05, "loss": 0.0253, "step": 1136 }, { "epoch": 0.71, "learning_rate": 1.4997092056544856e-05, "loss": 0.0393, "step": 1137 }, { "epoch": 0.71, "learning_rate": 1.4988364847659188e-05, "loss": 0.0283, "step": 1138 }, { "epoch": 0.71, "learning_rate": 1.4979632577880355e-05, "loss": 0.031, "step": 1139 }, { "epoch": 0.71, "learning_rate": 1.4970895256067593e-05, "loss": 0.031, "step": 1140 }, { "epoch": 0.71, "learning_rate": 1.4962152891085258e-05, "loss": 0.0286, "step": 1141 }, { "epoch": 0.71, "learning_rate": 1.4953405491802823e-05, "loss": 0.0293, "step": 1142 }, { "epoch": 0.71, "learning_rate": 1.4944653067094877e-05, "loss": 0.0337, "step": 1143 }, { "epoch": 0.71, "learning_rate": 1.4935895625841095e-05, "loss": 0.0284, "step": 1144 }, { "epoch": 0.71, "learning_rate": 1.4927133176926245e-05, "loss": 0.0253, "step": 1145 }, { "epoch": 0.71, "learning_rate": 1.4918365729240184e-05, "loss": 0.0273, "step": 1146 }, { "epoch": 0.71, "learning_rate": 1.4909593291677828e-05, "loss": 0.0248, "step": 1147 }, { "epoch": 0.71, "learning_rate": 1.4900815873139166e-05, "loss": 0.0255, "step": 1148 }, { "epoch": 0.71, "learning_rate": 1.4892033482529233e-05, "loss": 0.0328, "step": 1149 }, { "epoch": 0.72, "learning_rate": 1.488324612875811e-05, "loss": 0.0226, "step": 1150 }, { "epoch": 0.72, "learning_rate": 1.4874453820740917e-05, "loss": 0.0315, "step": 1151 }, { "epoch": 0.72, "learning_rate": 1.48656565673978e-05, "loss": 0.0303, "step": 1152 }, { "epoch": 0.72, "learning_rate": 1.4856854377653913e-05, "loss": 0.0243, "step": 1153 }, { "epoch": 0.72, "learning_rate": 1.484804726043943e-05, "loss": 0.0305, "step": 1154 }, { "epoch": 0.72, "learning_rate": 1.483923522468952e-05, "loss": 0.0272, "step": 1155 }, { "epoch": 0.72, "learning_rate": 1.4830418279344341e-05, "loss": 0.0358, "step": 1156 }, { "epoch": 0.72, "learning_rate": 1.482159643334903e-05, "loss": 0.0261, "step": 1157 }, { "epoch": 0.72, "learning_rate": 1.4812769695653697e-05, "loss": 0.0248, "step": 1158 }, { "epoch": 0.72, "learning_rate": 1.480393807521342e-05, "loss": 0.0308, "step": 1159 }, { "epoch": 0.72, "learning_rate": 1.4795101580988228e-05, "loss": 0.0294, "step": 1160 }, { "epoch": 0.72, "learning_rate": 1.4786260221943085e-05, "loss": 0.027, "step": 1161 }, { "epoch": 0.72, "learning_rate": 1.4777414007047907e-05, "loss": 0.0336, "step": 1162 }, { "epoch": 0.72, "learning_rate": 1.4768562945277526e-05, "loss": 0.0292, "step": 1163 }, { "epoch": 0.72, "learning_rate": 1.4759707045611694e-05, "loss": 0.0209, "step": 1164 }, { "epoch": 0.72, "learning_rate": 1.4750846317035071e-05, "loss": 0.0254, "step": 1165 }, { "epoch": 0.73, "learning_rate": 1.4741980768537214e-05, "loss": 0.0292, "step": 1166 }, { "epoch": 0.73, "learning_rate": 1.4733110409112574e-05, "loss": 0.0256, "step": 1167 }, { "epoch": 0.73, "learning_rate": 1.472423524776048e-05, "loss": 0.0225, "step": 1168 }, { "epoch": 0.73, "learning_rate": 1.4715355293485134e-05, "loss": 0.0281, "step": 1169 }, { "epoch": 0.73, "learning_rate": 1.4706470555295605e-05, "loss": 0.0272, "step": 1170 }, { "epoch": 0.73, "learning_rate": 1.4697581042205805e-05, "loss": 0.0305, "step": 1171 }, { "epoch": 0.73, "learning_rate": 1.4688686763234494e-05, "loss": 0.0249, "step": 1172 }, { "epoch": 0.73, "learning_rate": 1.4679787727405277e-05, "loss": 0.0319, "step": 1173 }, { "epoch": 0.73, "learning_rate": 1.4670883943746575e-05, "loss": 0.0346, "step": 1174 }, { "epoch": 0.73, "learning_rate": 1.4661975421291623e-05, "loss": 0.0344, "step": 1175 }, { "epoch": 0.73, "learning_rate": 1.4653062169078474e-05, "loss": 0.0227, "step": 1176 }, { "epoch": 0.73, "learning_rate": 1.4644144196149972e-05, "loss": 0.0293, "step": 1177 }, { "epoch": 0.73, "learning_rate": 1.4635221511553754e-05, "loss": 0.021, "step": 1178 }, { "epoch": 0.73, "learning_rate": 1.4626294124342237e-05, "loss": 0.032, "step": 1179 }, { "epoch": 0.73, "learning_rate": 1.4617362043572607e-05, "loss": 0.0298, "step": 1180 }, { "epoch": 0.73, "learning_rate": 1.460842527830681e-05, "loss": 0.0353, "step": 1181 }, { "epoch": 0.74, "learning_rate": 1.4599483837611552e-05, "loss": 0.0396, "step": 1182 }, { "epoch": 0.74, "learning_rate": 1.4590537730558272e-05, "loss": 0.0258, "step": 1183 }, { "epoch": 0.74, "learning_rate": 1.4581586966223156e-05, "loss": 0.038, "step": 1184 }, { "epoch": 0.74, "learning_rate": 1.4572631553687103e-05, "loss": 0.0338, "step": 1185 }, { "epoch": 0.74, "learning_rate": 1.4563671502035731e-05, "loss": 0.044, "step": 1186 }, { "epoch": 0.74, "learning_rate": 1.4554706820359374e-05, "loss": 0.0303, "step": 1187 }, { "epoch": 0.74, "learning_rate": 1.454573751775305e-05, "loss": 0.0214, "step": 1188 }, { "epoch": 0.74, "learning_rate": 1.453676360331647e-05, "loss": 0.0347, "step": 1189 }, { "epoch": 0.74, "learning_rate": 1.4527785086154027e-05, "loss": 0.0299, "step": 1190 }, { "epoch": 0.74, "learning_rate": 1.4518801975374781e-05, "loss": 0.0277, "step": 1191 }, { "epoch": 0.74, "learning_rate": 1.4509814280092453e-05, "loss": 0.029, "step": 1192 }, { "epoch": 0.74, "learning_rate": 1.4500822009425415e-05, "loss": 0.0239, "step": 1193 }, { "epoch": 0.74, "learning_rate": 1.4491825172496675e-05, "loss": 0.0319, "step": 1194 }, { "epoch": 0.74, "learning_rate": 1.448282377843389e-05, "loss": 0.0288, "step": 1195 }, { "epoch": 0.74, "learning_rate": 1.447381783636932e-05, "loss": 0.0226, "step": 1196 }, { "epoch": 0.74, "learning_rate": 1.4464807355439856e-05, "loss": 0.0182, "step": 1197 }, { "epoch": 0.75, "learning_rate": 1.4455792344786982e-05, "loss": 0.026, "step": 1198 }, { "epoch": 0.75, "learning_rate": 1.4446772813556784e-05, "loss": 0.0294, "step": 1199 }, { "epoch": 0.75, "learning_rate": 1.4437748770899936e-05, "loss": 0.0309, "step": 1200 }, { "epoch": 0.75, "learning_rate": 1.4428720225971681e-05, "loss": 0.0279, "step": 1201 }, { "epoch": 0.75, "learning_rate": 1.4419687187931835e-05, "loss": 0.0277, "step": 1202 }, { "epoch": 0.75, "learning_rate": 1.4410649665944772e-05, "loss": 0.0279, "step": 1203 }, { "epoch": 0.75, "learning_rate": 1.4401607669179415e-05, "loss": 0.0275, "step": 1204 }, { "epoch": 0.75, "learning_rate": 1.4392561206809232e-05, "loss": 0.0305, "step": 1205 }, { "epoch": 0.75, "learning_rate": 1.4383510288012213e-05, "loss": 0.0286, "step": 1206 }, { "epoch": 0.75, "learning_rate": 1.4374454921970869e-05, "loss": 0.0354, "step": 1207 }, { "epoch": 0.75, "learning_rate": 1.4365395117872227e-05, "loss": 0.0234, "step": 1208 }, { "epoch": 0.75, "learning_rate": 1.4356330884907823e-05, "loss": 0.0399, "step": 1209 }, { "epoch": 0.75, "learning_rate": 1.4347262232273671e-05, "loss": 0.0435, "step": 1210 }, { "epoch": 0.75, "learning_rate": 1.4338189169170285e-05, "loss": 0.0415, "step": 1211 }, { "epoch": 0.75, "learning_rate": 1.4329111704802636e-05, "loss": 0.028, "step": 1212 }, { "epoch": 0.75, "learning_rate": 1.432002984838018e-05, "loss": 0.0298, "step": 1213 }, { "epoch": 0.75, "learning_rate": 1.4310943609116815e-05, "loss": 0.0269, "step": 1214 }, { "epoch": 0.76, "learning_rate": 1.4301852996230889e-05, "loss": 0.024, "step": 1215 }, { "epoch": 0.76, "learning_rate": 1.4292758018945188e-05, "loss": 0.0318, "step": 1216 }, { "epoch": 0.76, "learning_rate": 1.4283658686486927e-05, "loss": 0.026, "step": 1217 }, { "epoch": 0.76, "learning_rate": 1.4274555008087736e-05, "loss": 0.0296, "step": 1218 }, { "epoch": 0.76, "learning_rate": 1.4265446992983661e-05, "loss": 0.0363, "step": 1219 }, { "epoch": 0.76, "learning_rate": 1.4256334650415135e-05, "loss": 0.0238, "step": 1220 }, { "epoch": 0.76, "learning_rate": 1.4247217989626996e-05, "loss": 0.0302, "step": 1221 }, { "epoch": 0.76, "learning_rate": 1.4238097019868457e-05, "loss": 0.0317, "step": 1222 }, { "epoch": 0.76, "learning_rate": 1.4228971750393098e-05, "loss": 0.0257, "step": 1223 }, { "epoch": 0.76, "learning_rate": 1.4219842190458865e-05, "loss": 0.0305, "step": 1224 }, { "epoch": 0.76, "learning_rate": 1.4210708349328062e-05, "loss": 0.0285, "step": 1225 }, { "epoch": 0.76, "learning_rate": 1.4201570236267326e-05, "loss": 0.0296, "step": 1226 }, { "epoch": 0.76, "learning_rate": 1.419242786054764e-05, "loss": 0.0242, "step": 1227 }, { "epoch": 0.76, "learning_rate": 1.41832812314443e-05, "loss": 0.0297, "step": 1228 }, { "epoch": 0.76, "learning_rate": 1.4174130358236924e-05, "loss": 0.0271, "step": 1229 }, { "epoch": 0.76, "learning_rate": 1.4164975250209432e-05, "loss": 0.0372, "step": 1230 }, { "epoch": 0.77, "learning_rate": 1.4155815916650048e-05, "loss": 0.0291, "step": 1231 }, { "epoch": 0.77, "learning_rate": 1.4146652366851274e-05, "loss": 0.028, "step": 1232 }, { "epoch": 0.77, "learning_rate": 1.4137484610109895e-05, "loss": 0.0395, "step": 1233 }, { "epoch": 0.77, "learning_rate": 1.4128312655726957e-05, "loss": 0.0254, "step": 1234 }, { "epoch": 0.77, "learning_rate": 1.4119136513007776e-05, "loss": 0.0334, "step": 1235 }, { "epoch": 0.77, "learning_rate": 1.4109956191261913e-05, "loss": 0.026, "step": 1236 }, { "epoch": 0.77, "learning_rate": 1.4100771699803159e-05, "loss": 0.0299, "step": 1237 }, { "epoch": 0.77, "learning_rate": 1.409158304794955e-05, "loss": 0.0227, "step": 1238 }, { "epoch": 0.77, "learning_rate": 1.4082390245023337e-05, "loss": 0.0236, "step": 1239 }, { "epoch": 0.77, "learning_rate": 1.4073193300350973e-05, "loss": 0.0241, "step": 1240 }, { "epoch": 0.77, "learning_rate": 1.406399222326313e-05, "loss": 0.0278, "step": 1241 }, { "epoch": 0.77, "learning_rate": 1.4054787023094662e-05, "loss": 0.0243, "step": 1242 }, { "epoch": 0.77, "learning_rate": 1.4045577709184605e-05, "loss": 0.0242, "step": 1243 }, { "epoch": 0.77, "learning_rate": 1.4036364290876176e-05, "loss": 0.031, "step": 1244 }, { "epoch": 0.77, "learning_rate": 1.402714677751675e-05, "loss": 0.0397, "step": 1245 }, { "epoch": 0.77, "learning_rate": 1.401792517845786e-05, "loss": 0.0259, "step": 1246 }, { "epoch": 0.78, "learning_rate": 1.4008699503055179e-05, "loss": 0.0312, "step": 1247 }, { "epoch": 0.78, "learning_rate": 1.399946976066852e-05, "loss": 0.0415, "step": 1248 }, { "epoch": 0.78, "learning_rate": 1.3990235960661824e-05, "loss": 0.0387, "step": 1249 }, { "epoch": 0.78, "learning_rate": 1.3980998112403146e-05, "loss": 0.028, "step": 1250 }, { "epoch": 0.78, "learning_rate": 1.3971756225264643e-05, "loss": 0.0388, "step": 1251 }, { "epoch": 0.78, "learning_rate": 1.3962510308622576e-05, "loss": 0.0315, "step": 1252 }, { "epoch": 0.78, "learning_rate": 1.3953260371857296e-05, "loss": 0.0237, "step": 1253 }, { "epoch": 0.78, "learning_rate": 1.3944006424353229e-05, "loss": 0.0217, "step": 1254 }, { "epoch": 0.78, "learning_rate": 1.3934748475498867e-05, "loss": 0.0309, "step": 1255 }, { "epoch": 0.78, "learning_rate": 1.3925486534686764e-05, "loss": 0.0385, "step": 1256 }, { "epoch": 0.78, "learning_rate": 1.391622061131353e-05, "loss": 0.0326, "step": 1257 }, { "epoch": 0.78, "learning_rate": 1.3906950714779805e-05, "loss": 0.0281, "step": 1258 }, { "epoch": 0.78, "learning_rate": 1.389767685449027e-05, "loss": 0.0246, "step": 1259 }, { "epoch": 0.78, "learning_rate": 1.3888399039853618e-05, "loss": 0.0212, "step": 1260 }, { "epoch": 0.78, "learning_rate": 1.3879117280282559e-05, "loss": 0.0265, "step": 1261 }, { "epoch": 0.78, "learning_rate": 1.386983158519381e-05, "loss": 0.0337, "step": 1262 }, { "epoch": 0.79, "learning_rate": 1.386054196400807e-05, "loss": 0.0324, "step": 1263 }, { "epoch": 0.79, "learning_rate": 1.3851248426150026e-05, "loss": 0.0261, "step": 1264 }, { "epoch": 0.79, "learning_rate": 1.3841950981048342e-05, "loss": 0.0245, "step": 1265 }, { "epoch": 0.79, "learning_rate": 1.3832649638135647e-05, "loss": 0.0355, "step": 1266 }, { "epoch": 0.79, "learning_rate": 1.382334440684851e-05, "loss": 0.0215, "step": 1267 }, { "epoch": 0.79, "learning_rate": 1.3814035296627466e-05, "loss": 0.0313, "step": 1268 }, { "epoch": 0.79, "learning_rate": 1.380472231691697e-05, "loss": 0.0249, "step": 1269 }, { "epoch": 0.79, "learning_rate": 1.3795405477165406e-05, "loss": 0.0295, "step": 1270 }, { "epoch": 0.79, "learning_rate": 1.3786084786825084e-05, "loss": 0.0327, "step": 1271 }, { "epoch": 0.79, "learning_rate": 1.3776760255352204e-05, "loss": 0.0292, "step": 1272 }, { "epoch": 0.79, "learning_rate": 1.3767431892206876e-05, "loss": 0.0296, "step": 1273 }, { "epoch": 0.79, "learning_rate": 1.375809970685309e-05, "loss": 0.0326, "step": 1274 }, { "epoch": 0.79, "learning_rate": 1.3748763708758717e-05, "loss": 0.0224, "step": 1275 }, { "epoch": 0.79, "learning_rate": 1.3739423907395502e-05, "loss": 0.0285, "step": 1276 }, { "epoch": 0.79, "learning_rate": 1.3730080312239034e-05, "loss": 0.03, "step": 1277 }, { "epoch": 0.79, "learning_rate": 1.3720732932768758e-05, "loss": 0.0246, "step": 1278 }, { "epoch": 0.8, "learning_rate": 1.3711381778467972e-05, "loss": 0.024, "step": 1279 }, { "epoch": 0.8, "learning_rate": 1.3702026858823781e-05, "loss": 0.0267, "step": 1280 }, { "epoch": 0.8, "learning_rate": 1.3692668183327123e-05, "loss": 0.0267, "step": 1281 }, { "epoch": 0.8, "learning_rate": 1.3683305761472743e-05, "loss": 0.0291, "step": 1282 }, { "epoch": 0.8, "learning_rate": 1.3673939602759187e-05, "loss": 0.0295, "step": 1283 }, { "epoch": 0.8, "learning_rate": 1.36645697166888e-05, "loss": 0.0239, "step": 1284 }, { "epoch": 0.8, "learning_rate": 1.3655196112767695e-05, "loss": 0.0282, "step": 1285 }, { "epoch": 0.8, "learning_rate": 1.3645818800505763e-05, "loss": 0.0274, "step": 1286 }, { "epoch": 0.8, "learning_rate": 1.3636437789416658e-05, "loss": 0.026, "step": 1287 }, { "epoch": 0.8, "learning_rate": 1.362705308901779e-05, "loss": 0.0393, "step": 1288 }, { "epoch": 0.8, "learning_rate": 1.3617664708830304e-05, "loss": 0.0234, "step": 1289 }, { "epoch": 0.8, "learning_rate": 1.3608272658379087e-05, "loss": 0.0218, "step": 1290 }, { "epoch": 0.8, "learning_rate": 1.359887694719274e-05, "loss": 0.0245, "step": 1291 }, { "epoch": 0.8, "learning_rate": 1.3589477584803588e-05, "loss": 0.0337, "step": 1292 }, { "epoch": 0.8, "learning_rate": 1.3580074580747653e-05, "loss": 0.0295, "step": 1293 }, { "epoch": 0.8, "learning_rate": 1.3570667944564651e-05, "loss": 0.0316, "step": 1294 }, { "epoch": 0.81, "learning_rate": 1.3561257685797992e-05, "loss": 0.0303, "step": 1295 }, { "epoch": 0.81, "learning_rate": 1.355184381399475e-05, "loss": 0.0334, "step": 1296 }, { "epoch": 0.81, "learning_rate": 1.3542426338705672e-05, "loss": 0.0271, "step": 1297 }, { "epoch": 0.81, "learning_rate": 1.353300526948516e-05, "loss": 0.0228, "step": 1298 }, { "epoch": 0.81, "learning_rate": 1.3523580615891258e-05, "loss": 0.0299, "step": 1299 }, { "epoch": 0.81, "learning_rate": 1.3514152387485651e-05, "loss": 0.0319, "step": 1300 }, { "epoch": 0.81, "learning_rate": 1.3504720593833648e-05, "loss": 0.0238, "step": 1301 }, { "epoch": 0.81, "learning_rate": 1.349528524450418e-05, "loss": 0.0346, "step": 1302 }, { "epoch": 0.81, "learning_rate": 1.3485846349069779e-05, "loss": 0.0272, "step": 1303 }, { "epoch": 0.81, "learning_rate": 1.347640391710657e-05, "loss": 0.0249, "step": 1304 }, { "epoch": 0.81, "learning_rate": 1.3466957958194284e-05, "loss": 0.028, "step": 1305 }, { "epoch": 0.81, "learning_rate": 1.3457508481916217e-05, "loss": 0.0295, "step": 1306 }, { "epoch": 0.81, "learning_rate": 1.3448055497859232e-05, "loss": 0.027, "step": 1307 }, { "epoch": 0.81, "learning_rate": 1.343859901561376e-05, "loss": 0.0276, "step": 1308 }, { "epoch": 0.81, "learning_rate": 1.3429139044773768e-05, "loss": 0.0296, "step": 1309 }, { "epoch": 0.81, "learning_rate": 1.3419675594936778e-05, "loss": 0.0268, "step": 1310 }, { "epoch": 0.82, "learning_rate": 1.3410208675703831e-05, "loss": 0.0275, "step": 1311 }, { "epoch": 0.82, "learning_rate": 1.340073829667949e-05, "loss": 0.0222, "step": 1312 }, { "epoch": 0.82, "learning_rate": 1.3391264467471825e-05, "loss": 0.0338, "step": 1313 }, { "epoch": 0.82, "learning_rate": 1.3381787197692413e-05, "loss": 0.023, "step": 1314 }, { "epoch": 0.82, "learning_rate": 1.3372306496956324e-05, "loss": 0.0269, "step": 1315 }, { "epoch": 0.82, "learning_rate": 1.3362822374882094e-05, "loss": 0.0337, "step": 1316 }, { "epoch": 0.82, "learning_rate": 1.3353334841091746e-05, "loss": 0.0272, "step": 1317 }, { "epoch": 0.82, "learning_rate": 1.3343843905210751e-05, "loss": 0.0275, "step": 1318 }, { "epoch": 0.82, "learning_rate": 1.3334349576868046e-05, "loss": 0.0346, "step": 1319 }, { "epoch": 0.82, "learning_rate": 1.3324851865696e-05, "loss": 0.0302, "step": 1320 }, { "epoch": 0.82, "learning_rate": 1.3315350781330414e-05, "loss": 0.0265, "step": 1321 }, { "epoch": 0.82, "learning_rate": 1.3305846333410513e-05, "loss": 0.023, "step": 1322 }, { "epoch": 0.82, "learning_rate": 1.3296338531578939e-05, "loss": 0.0269, "step": 1323 }, { "epoch": 0.82, "learning_rate": 1.3286827385481726e-05, "loss": 0.0201, "step": 1324 }, { "epoch": 0.82, "learning_rate": 1.3277312904768314e-05, "loss": 0.0346, "step": 1325 }, { "epoch": 0.82, "learning_rate": 1.3267795099091515e-05, "loss": 0.0224, "step": 1326 }, { "epoch": 0.83, "learning_rate": 1.3258273978107519e-05, "loss": 0.0252, "step": 1327 }, { "epoch": 0.83, "learning_rate": 1.324874955147588e-05, "loss": 0.0313, "step": 1328 }, { "epoch": 0.83, "learning_rate": 1.3239221828859509e-05, "loss": 0.0312, "step": 1329 }, { "epoch": 0.83, "learning_rate": 1.3229690819924653e-05, "loss": 0.0433, "step": 1330 }, { "epoch": 0.83, "learning_rate": 1.3220156534340896e-05, "loss": 0.0261, "step": 1331 }, { "epoch": 0.83, "learning_rate": 1.3210618981781147e-05, "loss": 0.0344, "step": 1332 }, { "epoch": 0.83, "learning_rate": 1.3201078171921635e-05, "loss": 0.0368, "step": 1333 }, { "epoch": 0.83, "learning_rate": 1.3191534114441883e-05, "loss": 0.0279, "step": 1334 }, { "epoch": 0.83, "learning_rate": 1.3181986819024717e-05, "loss": 0.0252, "step": 1335 }, { "epoch": 0.83, "learning_rate": 1.3172436295356241e-05, "loss": 0.0248, "step": 1336 }, { "epoch": 0.83, "learning_rate": 1.3162882553125845e-05, "loss": 0.029, "step": 1337 }, { "epoch": 0.83, "learning_rate": 1.3153325602026175e-05, "loss": 0.03, "step": 1338 }, { "epoch": 0.83, "learning_rate": 1.3143765451753137e-05, "loss": 0.0257, "step": 1339 }, { "epoch": 0.83, "learning_rate": 1.3134202112005876e-05, "loss": 0.0248, "step": 1340 }, { "epoch": 0.83, "learning_rate": 1.312463559248678e-05, "loss": 0.0307, "step": 1341 }, { "epoch": 0.83, "learning_rate": 1.3115065902901467e-05, "loss": 0.0257, "step": 1342 }, { "epoch": 0.84, "learning_rate": 1.3105493052958756e-05, "loss": 0.0323, "step": 1343 }, { "epoch": 0.84, "learning_rate": 1.3095917052370686e-05, "loss": 0.03, "step": 1344 }, { "epoch": 0.84, "learning_rate": 1.3086337910852487e-05, "loss": 0.0309, "step": 1345 }, { "epoch": 0.84, "learning_rate": 1.3076755638122578e-05, "loss": 0.0325, "step": 1346 }, { "epoch": 0.84, "learning_rate": 1.3067170243902553e-05, "loss": 0.0315, "step": 1347 }, { "epoch": 0.84, "learning_rate": 1.3057581737917172e-05, "loss": 0.0236, "step": 1348 }, { "epoch": 0.84, "learning_rate": 1.3047990129894348e-05, "loss": 0.0411, "step": 1349 }, { "epoch": 0.84, "learning_rate": 1.3038395429565156e-05, "loss": 0.0352, "step": 1350 }, { "epoch": 0.84, "learning_rate": 1.3028797646663792e-05, "loss": 0.0336, "step": 1351 }, { "epoch": 0.84, "learning_rate": 1.3019196790927587e-05, "loss": 0.0247, "step": 1352 }, { "epoch": 0.84, "learning_rate": 1.3009592872096987e-05, "loss": 0.03, "step": 1353 }, { "epoch": 0.84, "learning_rate": 1.299998589991555e-05, "loss": 0.0319, "step": 1354 }, { "epoch": 0.84, "learning_rate": 1.299037588412993e-05, "loss": 0.0267, "step": 1355 }, { "epoch": 0.84, "learning_rate": 1.2980762834489862e-05, "loss": 0.0307, "step": 1356 }, { "epoch": 0.84, "learning_rate": 1.297114676074817e-05, "loss": 0.0311, "step": 1357 }, { "epoch": 0.84, "learning_rate": 1.2961527672660736e-05, "loss": 0.0302, "step": 1358 }, { "epoch": 0.85, "learning_rate": 1.2951905579986506e-05, "loss": 0.022, "step": 1359 }, { "epoch": 0.85, "learning_rate": 1.2942280492487478e-05, "loss": 0.0304, "step": 1360 }, { "epoch": 0.85, "learning_rate": 1.2932652419928677e-05, "loss": 0.0255, "step": 1361 }, { "epoch": 0.85, "learning_rate": 1.2923021372078164e-05, "loss": 0.0345, "step": 1362 }, { "epoch": 0.85, "learning_rate": 1.2913387358707018e-05, "loss": 0.0273, "step": 1363 }, { "epoch": 0.85, "learning_rate": 1.290375038958933e-05, "loss": 0.0272, "step": 1364 }, { "epoch": 0.85, "learning_rate": 1.2894110474502182e-05, "loss": 0.0299, "step": 1365 }, { "epoch": 0.85, "learning_rate": 1.288446762322565e-05, "loss": 0.0292, "step": 1366 }, { "epoch": 0.85, "learning_rate": 1.2874821845542782e-05, "loss": 0.0214, "step": 1367 }, { "epoch": 0.85, "learning_rate": 1.2865173151239611e-05, "loss": 0.0251, "step": 1368 }, { "epoch": 0.85, "learning_rate": 1.285552155010511e-05, "loss": 0.0256, "step": 1369 }, { "epoch": 0.85, "learning_rate": 1.2845867051931214e-05, "loss": 0.0251, "step": 1370 }, { "epoch": 0.85, "learning_rate": 1.2836209666512786e-05, "loss": 0.0214, "step": 1371 }, { "epoch": 0.85, "learning_rate": 1.2826549403647633e-05, "loss": 0.0268, "step": 1372 }, { "epoch": 0.85, "learning_rate": 1.2816886273136471e-05, "loss": 0.02, "step": 1373 }, { "epoch": 0.85, "learning_rate": 1.2807220284782926e-05, "loss": 0.0251, "step": 1374 }, { "epoch": 0.86, "learning_rate": 1.2797551448393527e-05, "loss": 0.0272, "step": 1375 }, { "epoch": 0.86, "learning_rate": 1.2787879773777685e-05, "loss": 0.0323, "step": 1376 }, { "epoch": 0.86, "learning_rate": 1.2778205270747703e-05, "loss": 0.0272, "step": 1377 }, { "epoch": 0.86, "learning_rate": 1.2768527949118742e-05, "loss": 0.03, "step": 1378 }, { "epoch": 0.86, "learning_rate": 1.2758847818708832e-05, "loss": 0.0327, "step": 1379 }, { "epoch": 0.86, "learning_rate": 1.274916488933884e-05, "loss": 0.0238, "step": 1380 }, { "epoch": 0.86, "learning_rate": 1.2739479170832485e-05, "loss": 0.0292, "step": 1381 }, { "epoch": 0.86, "learning_rate": 1.272979067301631e-05, "loss": 0.0268, "step": 1382 }, { "epoch": 0.86, "learning_rate": 1.2720099405719679e-05, "loss": 0.0297, "step": 1383 }, { "epoch": 0.86, "learning_rate": 1.2710405378774768e-05, "loss": 0.0287, "step": 1384 }, { "epoch": 0.86, "learning_rate": 1.2700708602016545e-05, "loss": 0.033, "step": 1385 }, { "epoch": 0.86, "learning_rate": 1.2691009085282774e-05, "loss": 0.0327, "step": 1386 }, { "epoch": 0.86, "learning_rate": 1.2681306838414006e-05, "loss": 0.037, "step": 1387 }, { "epoch": 0.86, "learning_rate": 1.2671601871253543e-05, "loss": 0.0331, "step": 1388 }, { "epoch": 0.86, "learning_rate": 1.2661894193647458e-05, "loss": 0.0229, "step": 1389 }, { "epoch": 0.86, "learning_rate": 1.265218381544458e-05, "loss": 0.0305, "step": 1390 }, { "epoch": 0.87, "learning_rate": 1.264247074649647e-05, "loss": 0.0297, "step": 1391 }, { "epoch": 0.87, "learning_rate": 1.2632754996657417e-05, "loss": 0.0289, "step": 1392 }, { "epoch": 0.87, "learning_rate": 1.2623036575784433e-05, "loss": 0.0319, "step": 1393 }, { "epoch": 0.87, "learning_rate": 1.261331549373724e-05, "loss": 0.0287, "step": 1394 }, { "epoch": 0.87, "learning_rate": 1.2603591760378266e-05, "loss": 0.0205, "step": 1395 }, { "epoch": 0.87, "learning_rate": 1.2593865385572617e-05, "loss": 0.0245, "step": 1396 }, { "epoch": 0.87, "learning_rate": 1.258413637918808e-05, "loss": 0.0273, "step": 1397 }, { "epoch": 0.87, "learning_rate": 1.2574404751095121e-05, "loss": 0.0292, "step": 1398 }, { "epoch": 0.87, "learning_rate": 1.2564670511166865e-05, "loss": 0.0322, "step": 1399 }, { "epoch": 0.87, "learning_rate": 1.2554933669279076e-05, "loss": 0.0238, "step": 1400 }, { "epoch": 0.87, "learning_rate": 1.2545194235310173e-05, "loss": 0.0311, "step": 1401 }, { "epoch": 0.87, "learning_rate": 1.2535452219141182e-05, "loss": 0.0324, "step": 1402 }, { "epoch": 0.87, "learning_rate": 1.2525707630655777e-05, "loss": 0.0336, "step": 1403 }, { "epoch": 0.87, "learning_rate": 1.2515960479740224e-05, "loss": 0.03, "step": 1404 }, { "epoch": 0.87, "learning_rate": 1.250621077628339e-05, "loss": 0.0293, "step": 1405 }, { "epoch": 0.87, "learning_rate": 1.2496458530176736e-05, "loss": 0.0307, "step": 1406 }, { "epoch": 0.88, "learning_rate": 1.2486703751314303e-05, "loss": 0.0296, "step": 1407 }, { "epoch": 0.88, "learning_rate": 1.2476946449592693e-05, "loss": 0.0259, "step": 1408 }, { "epoch": 0.88, "learning_rate": 1.246718663491108e-05, "loss": 0.0233, "step": 1409 }, { "epoch": 0.88, "learning_rate": 1.245742431717118e-05, "loss": 0.026, "step": 1410 }, { "epoch": 0.88, "learning_rate": 1.244765950627725e-05, "loss": 0.0226, "step": 1411 }, { "epoch": 0.88, "learning_rate": 1.2437892212136076e-05, "loss": 0.0245, "step": 1412 }, { "epoch": 0.88, "learning_rate": 1.2428122444656963e-05, "loss": 0.0245, "step": 1413 }, { "epoch": 0.88, "learning_rate": 1.2418350213751728e-05, "loss": 0.0212, "step": 1414 }, { "epoch": 0.88, "learning_rate": 1.240857552933468e-05, "loss": 0.0221, "step": 1415 }, { "epoch": 0.88, "learning_rate": 1.2398798401322632e-05, "loss": 0.0199, "step": 1416 }, { "epoch": 0.88, "learning_rate": 1.2389018839634857e-05, "loss": 0.0422, "step": 1417 }, { "epoch": 0.88, "learning_rate": 1.2379236854193113e-05, "loss": 0.0328, "step": 1418 }, { "epoch": 0.88, "learning_rate": 1.2369452454921604e-05, "loss": 0.0307, "step": 1419 }, { "epoch": 0.88, "learning_rate": 1.2359665651746995e-05, "loss": 0.033, "step": 1420 }, { "epoch": 0.88, "learning_rate": 1.2349876454598384e-05, "loss": 0.0256, "step": 1421 }, { "epoch": 0.88, "learning_rate": 1.2340084873407297e-05, "loss": 0.0243, "step": 1422 }, { "epoch": 0.88, "learning_rate": 1.233029091810768e-05, "loss": 0.0256, "step": 1423 }, { "epoch": 0.89, "learning_rate": 1.2320494598635886e-05, "loss": 0.0273, "step": 1424 }, { "epoch": 0.89, "learning_rate": 1.231069592493067e-05, "loss": 0.0232, "step": 1425 }, { "epoch": 0.89, "learning_rate": 1.2300894906933172e-05, "loss": 0.0289, "step": 1426 }, { "epoch": 0.89, "learning_rate": 1.2291091554586913e-05, "loss": 0.0269, "step": 1427 }, { "epoch": 0.89, "learning_rate": 1.2281285877837783e-05, "loss": 0.0364, "step": 1428 }, { "epoch": 0.89, "learning_rate": 1.2271477886634023e-05, "loss": 0.0255, "step": 1429 }, { "epoch": 0.89, "learning_rate": 1.2261667590926232e-05, "loss": 0.0255, "step": 1430 }, { "epoch": 0.89, "learning_rate": 1.225185500066734e-05, "loss": 0.0248, "step": 1431 }, { "epoch": 0.89, "learning_rate": 1.224204012581261e-05, "loss": 0.0227, "step": 1432 }, { "epoch": 0.89, "learning_rate": 1.223222297631962e-05, "loss": 0.0299, "step": 1433 }, { "epoch": 0.89, "learning_rate": 1.2222403562148252e-05, "loss": 0.031, "step": 1434 }, { "epoch": 0.89, "learning_rate": 1.2212581893260694e-05, "loss": 0.0394, "step": 1435 }, { "epoch": 0.89, "learning_rate": 1.2202757979621415e-05, "loss": 0.0247, "step": 1436 }, { "epoch": 0.89, "learning_rate": 1.2192931831197163e-05, "loss": 0.0257, "step": 1437 }, { "epoch": 0.89, "learning_rate": 1.2183103457956954e-05, "loss": 0.0314, "step": 1438 }, { "epoch": 0.89, "learning_rate": 1.2173272869872062e-05, "loss": 0.0334, "step": 1439 }, { "epoch": 0.9, "learning_rate": 1.2163440076916005e-05, "loss": 0.0292, "step": 1440 }, { "epoch": 0.9, "learning_rate": 1.2153605089064542e-05, "loss": 0.0309, "step": 1441 }, { "epoch": 0.9, "learning_rate": 1.2143767916295655e-05, "loss": 0.0259, "step": 1442 }, { "epoch": 0.9, "learning_rate": 1.2133928568589544e-05, "loss": 0.024, "step": 1443 }, { "epoch": 0.9, "learning_rate": 1.2124087055928617e-05, "loss": 0.0293, "step": 1444 }, { "epoch": 0.9, "learning_rate": 1.2114243388297478e-05, "loss": 0.0315, "step": 1445 }, { "epoch": 0.9, "learning_rate": 1.2104397575682914e-05, "loss": 0.0231, "step": 1446 }, { "epoch": 0.9, "learning_rate": 1.2094549628073893e-05, "loss": 0.0324, "step": 1447 }, { "epoch": 0.9, "learning_rate": 1.2084699555461544e-05, "loss": 0.0247, "step": 1448 }, { "epoch": 0.9, "learning_rate": 1.207484736783916e-05, "loss": 0.0248, "step": 1449 }, { "epoch": 0.9, "learning_rate": 1.2064993075202172e-05, "loss": 0.0303, "step": 1450 }, { "epoch": 0.9, "learning_rate": 1.2055136687548142e-05, "loss": 0.0248, "step": 1451 }, { "epoch": 0.9, "learning_rate": 1.2045278214876776e-05, "loss": 0.0286, "step": 1452 }, { "epoch": 0.9, "learning_rate": 1.2035417667189882e-05, "loss": 0.0318, "step": 1453 }, { "epoch": 0.9, "learning_rate": 1.2025555054491367e-05, "loss": 0.0246, "step": 1454 }, { "epoch": 0.9, "learning_rate": 1.2015690386787245e-05, "loss": 0.0277, "step": 1455 }, { "epoch": 0.91, "learning_rate": 1.2005823674085613e-05, "loss": 0.0237, "step": 1456 }, { "epoch": 0.91, "learning_rate": 1.1995954926396642e-05, "loss": 0.0292, "step": 1457 }, { "epoch": 0.91, "learning_rate": 1.1986084153732567e-05, "loss": 0.0238, "step": 1458 }, { "epoch": 0.91, "learning_rate": 1.1976211366107668e-05, "loss": 0.0214, "step": 1459 }, { "epoch": 0.91, "learning_rate": 1.1966336573538287e-05, "loss": 0.0274, "step": 1460 }, { "epoch": 0.91, "learning_rate": 1.1956459786042786e-05, "loss": 0.0264, "step": 1461 }, { "epoch": 0.91, "learning_rate": 1.1946581013641559e-05, "loss": 0.0295, "step": 1462 }, { "epoch": 0.91, "learning_rate": 1.193670026635701e-05, "loss": 0.0254, "step": 1463 }, { "epoch": 0.91, "learning_rate": 1.1926817554213548e-05, "loss": 0.0341, "step": 1464 }, { "epoch": 0.91, "learning_rate": 1.1916932887237571e-05, "loss": 0.0211, "step": 1465 }, { "epoch": 0.91, "learning_rate": 1.190704627545747e-05, "loss": 0.0197, "step": 1466 }, { "epoch": 0.91, "learning_rate": 1.1897157728903599e-05, "loss": 0.0221, "step": 1467 }, { "epoch": 0.91, "learning_rate": 1.188726725760828e-05, "loss": 0.0288, "step": 1468 }, { "epoch": 0.91, "learning_rate": 1.1877374871605786e-05, "loss": 0.0243, "step": 1469 }, { "epoch": 0.91, "learning_rate": 1.1867480580932336e-05, "loss": 0.0284, "step": 1470 }, { "epoch": 0.91, "learning_rate": 1.1857584395626077e-05, "loss": 0.0353, "step": 1471 }, { "epoch": 0.92, "learning_rate": 1.184768632572708e-05, "loss": 0.0237, "step": 1472 }, { "epoch": 0.92, "learning_rate": 1.1837786381277323e-05, "loss": 0.0312, "step": 1473 }, { "epoch": 0.92, "learning_rate": 1.18278845723207e-05, "loss": 0.0319, "step": 1474 }, { "epoch": 0.92, "learning_rate": 1.1817980908902981e-05, "loss": 0.0221, "step": 1475 }, { "epoch": 0.92, "learning_rate": 1.1808075401071828e-05, "loss": 0.0357, "step": 1476 }, { "epoch": 0.92, "learning_rate": 1.1798168058876765e-05, "loss": 0.0292, "step": 1477 }, { "epoch": 0.92, "learning_rate": 1.1788258892369185e-05, "loss": 0.0285, "step": 1478 }, { "epoch": 0.92, "learning_rate": 1.1778347911602329e-05, "loss": 0.029, "step": 1479 }, { "epoch": 0.92, "learning_rate": 1.1768435126631276e-05, "loss": 0.0257, "step": 1480 }, { "epoch": 0.92, "learning_rate": 1.175852054751294e-05, "loss": 0.0338, "step": 1481 }, { "epoch": 0.92, "learning_rate": 1.1748604184306049e-05, "loss": 0.0341, "step": 1482 }, { "epoch": 0.92, "learning_rate": 1.1738686047071152e-05, "loss": 0.0243, "step": 1483 }, { "epoch": 0.92, "learning_rate": 1.1728766145870587e-05, "loss": 0.0229, "step": 1484 }, { "epoch": 0.92, "learning_rate": 1.1718844490768488e-05, "loss": 0.028, "step": 1485 }, { "epoch": 0.92, "learning_rate": 1.170892109183076e-05, "loss": 0.0293, "step": 1486 }, { "epoch": 0.92, "learning_rate": 1.169899595912509e-05, "loss": 0.0341, "step": 1487 }, { "epoch": 0.93, "learning_rate": 1.1689069102720914e-05, "loss": 0.0289, "step": 1488 }, { "epoch": 0.93, "learning_rate": 1.167914053268942e-05, "loss": 0.0316, "step": 1489 }, { "epoch": 0.93, "learning_rate": 1.1669210259103535e-05, "loss": 0.0251, "step": 1490 }, { "epoch": 0.93, "learning_rate": 1.1659278292037914e-05, "loss": 0.0262, "step": 1491 }, { "epoch": 0.93, "learning_rate": 1.1649344641568927e-05, "loss": 0.0243, "step": 1492 }, { "epoch": 0.93, "learning_rate": 1.1639409317774661e-05, "loss": 0.0371, "step": 1493 }, { "epoch": 0.93, "learning_rate": 1.1629472330734888e-05, "loss": 0.0305, "step": 1494 }, { "epoch": 0.93, "learning_rate": 1.1619533690531076e-05, "loss": 0.0294, "step": 1495 }, { "epoch": 0.93, "learning_rate": 1.1609593407246369e-05, "loss": 0.0232, "step": 1496 }, { "epoch": 0.93, "learning_rate": 1.1599651490965576e-05, "loss": 0.0206, "step": 1497 }, { "epoch": 0.93, "learning_rate": 1.1589707951775167e-05, "loss": 0.0294, "step": 1498 }, { "epoch": 0.93, "learning_rate": 1.1579762799763249e-05, "loss": 0.0313, "step": 1499 }, { "epoch": 0.93, "learning_rate": 1.1569816045019575e-05, "loss": 0.0309, "step": 1500 }, { "epoch": 0.93, "learning_rate": 1.1559867697635523e-05, "loss": 0.0289, "step": 1501 }, { "epoch": 0.93, "learning_rate": 1.1549917767704083e-05, "loss": 0.0303, "step": 1502 }, { "epoch": 0.93, "learning_rate": 1.1539966265319843e-05, "loss": 0.0289, "step": 1503 }, { "epoch": 0.94, "learning_rate": 1.1530013200579008e-05, "loss": 0.0358, "step": 1504 }, { "epoch": 0.94, "learning_rate": 1.152005858357935e-05, "loss": 0.0229, "step": 1505 }, { "epoch": 0.94, "learning_rate": 1.1510102424420225e-05, "loss": 0.0207, "step": 1506 }, { "epoch": 0.94, "learning_rate": 1.1500144733202543e-05, "loss": 0.0264, "step": 1507 }, { "epoch": 0.94, "learning_rate": 1.1490185520028776e-05, "loss": 0.0306, "step": 1508 }, { "epoch": 0.94, "learning_rate": 1.1480224795002943e-05, "loss": 0.0211, "step": 1509 }, { "epoch": 0.94, "learning_rate": 1.1470262568230593e-05, "loss": 0.0289, "step": 1510 }, { "epoch": 0.94, "learning_rate": 1.1460298849818793e-05, "loss": 0.0274, "step": 1511 }, { "epoch": 0.94, "learning_rate": 1.1450333649876135e-05, "loss": 0.0234, "step": 1512 }, { "epoch": 0.94, "learning_rate": 1.1440366978512703e-05, "loss": 0.0277, "step": 1513 }, { "epoch": 0.94, "learning_rate": 1.1430398845840085e-05, "loss": 0.0277, "step": 1514 }, { "epoch": 0.94, "learning_rate": 1.1420429261971342e-05, "loss": 0.0207, "step": 1515 }, { "epoch": 0.94, "learning_rate": 1.1410458237021005e-05, "loss": 0.0235, "step": 1516 }, { "epoch": 0.94, "learning_rate": 1.1400485781105082e-05, "loss": 0.0142, "step": 1517 }, { "epoch": 0.94, "learning_rate": 1.1390511904341017e-05, "loss": 0.025, "step": 1518 }, { "epoch": 0.94, "learning_rate": 1.1380536616847706e-05, "loss": 0.0287, "step": 1519 }, { "epoch": 0.95, "learning_rate": 1.137055992874547e-05, "loss": 0.0223, "step": 1520 }, { "epoch": 0.95, "learning_rate": 1.1360581850156053e-05, "loss": 0.0185, "step": 1521 }, { "epoch": 0.95, "learning_rate": 1.135060239120261e-05, "loss": 0.0309, "step": 1522 }, { "epoch": 0.95, "learning_rate": 1.1340621562009696e-05, "loss": 0.0291, "step": 1523 }, { "epoch": 0.95, "learning_rate": 1.1330639372703258e-05, "loss": 0.0231, "step": 1524 }, { "epoch": 0.95, "learning_rate": 1.1320655833410621e-05, "loss": 0.0285, "step": 1525 }, { "epoch": 0.95, "learning_rate": 1.1310670954260479e-05, "loss": 0.0226, "step": 1526 }, { "epoch": 0.95, "learning_rate": 1.1300684745382884e-05, "loss": 0.0341, "step": 1527 }, { "epoch": 0.95, "learning_rate": 1.1290697216909243e-05, "loss": 0.0279, "step": 1528 }, { "epoch": 0.95, "learning_rate": 1.12807083789723e-05, "loss": 0.0292, "step": 1529 }, { "epoch": 0.95, "learning_rate": 1.1270718241706118e-05, "loss": 0.0341, "step": 1530 }, { "epoch": 0.95, "learning_rate": 1.1260726815246093e-05, "loss": 0.0256, "step": 1531 }, { "epoch": 0.95, "learning_rate": 1.1250734109728916e-05, "loss": 0.0265, "step": 1532 }, { "epoch": 0.95, "learning_rate": 1.124074013529259e-05, "loss": 0.026, "step": 1533 }, { "epoch": 0.95, "learning_rate": 1.123074490207639e-05, "loss": 0.0391, "step": 1534 }, { "epoch": 0.95, "learning_rate": 1.122074842022087e-05, "loss": 0.0261, "step": 1535 }, { "epoch": 0.96, "learning_rate": 1.1210750699867867e-05, "loss": 0.0178, "step": 1536 }, { "epoch": 0.96, "learning_rate": 1.120075175116046e-05, "loss": 0.0307, "step": 1537 }, { "epoch": 0.96, "learning_rate": 1.1190751584242972e-05, "loss": 0.0166, "step": 1538 }, { "epoch": 0.96, "learning_rate": 1.1180750209260972e-05, "loss": 0.0326, "step": 1539 }, { "epoch": 0.96, "learning_rate": 1.1170747636361245e-05, "loss": 0.0299, "step": 1540 }, { "epoch": 0.96, "learning_rate": 1.1160743875691806e-05, "loss": 0.026, "step": 1541 }, { "epoch": 0.96, "learning_rate": 1.1150738937401858e-05, "loss": 0.0195, "step": 1542 }, { "epoch": 0.96, "learning_rate": 1.1140732831641805e-05, "loss": 0.0222, "step": 1543 }, { "epoch": 0.96, "learning_rate": 1.1130725568563241e-05, "loss": 0.0222, "step": 1544 }, { "epoch": 0.96, "learning_rate": 1.1120717158318928e-05, "loss": 0.0254, "step": 1545 }, { "epoch": 0.96, "learning_rate": 1.111070761106279e-05, "loss": 0.0273, "step": 1546 }, { "epoch": 0.96, "learning_rate": 1.1100696936949915e-05, "loss": 0.0248, "step": 1547 }, { "epoch": 0.96, "learning_rate": 1.109068514613652e-05, "loss": 0.0299, "step": 1548 }, { "epoch": 0.96, "learning_rate": 1.1080672248779964e-05, "loss": 0.0289, "step": 1549 }, { "epoch": 0.96, "learning_rate": 1.1070658255038733e-05, "loss": 0.0294, "step": 1550 }, { "epoch": 0.96, "learning_rate": 1.1060643175072411e-05, "loss": 0.0269, "step": 1551 }, { "epoch": 0.97, "learning_rate": 1.10506270190417e-05, "loss": 0.0188, "step": 1552 }, { "epoch": 0.97, "learning_rate": 1.1040609797108378e-05, "loss": 0.0312, "step": 1553 }, { "epoch": 0.97, "learning_rate": 1.1030591519435316e-05, "loss": 0.0313, "step": 1554 }, { "epoch": 0.97, "learning_rate": 1.1020572196186457e-05, "loss": 0.0246, "step": 1555 }, { "epoch": 0.97, "learning_rate": 1.1010551837526793e-05, "loss": 0.0294, "step": 1556 }, { "epoch": 0.97, "learning_rate": 1.1000530453622374e-05, "loss": 0.0273, "step": 1557 }, { "epoch": 0.97, "learning_rate": 1.0990508054640292e-05, "loss": 0.0197, "step": 1558 }, { "epoch": 0.97, "learning_rate": 1.0980484650748666e-05, "loss": 0.0305, "step": 1559 }, { "epoch": 0.97, "learning_rate": 1.097046025211664e-05, "loss": 0.0213, "step": 1560 }, { "epoch": 0.97, "learning_rate": 1.0960434868914355e-05, "loss": 0.025, "step": 1561 }, { "epoch": 0.97, "learning_rate": 1.095040851131296e-05, "loss": 0.0246, "step": 1562 }, { "epoch": 0.97, "learning_rate": 1.0940381189484595e-05, "loss": 0.0261, "step": 1563 }, { "epoch": 0.97, "learning_rate": 1.0930352913602371e-05, "loss": 0.0184, "step": 1564 }, { "epoch": 0.97, "learning_rate": 1.0920323693840368e-05, "loss": 0.0236, "step": 1565 }, { "epoch": 0.97, "learning_rate": 1.0910293540373629e-05, "loss": 0.0263, "step": 1566 }, { "epoch": 0.97, "learning_rate": 1.0900262463378141e-05, "loss": 0.028, "step": 1567 }, { "epoch": 0.98, "learning_rate": 1.0890230473030823e-05, "loss": 0.0282, "step": 1568 }, { "epoch": 0.98, "learning_rate": 1.0880197579509532e-05, "loss": 0.0212, "step": 1569 }, { "epoch": 0.98, "learning_rate": 1.087016379299303e-05, "loss": 0.0226, "step": 1570 }, { "epoch": 0.98, "learning_rate": 1.086012912366099e-05, "loss": 0.0229, "step": 1571 }, { "epoch": 0.98, "learning_rate": 1.0850093581693983e-05, "loss": 0.0236, "step": 1572 }, { "epoch": 0.98, "learning_rate": 1.0840057177273463e-05, "loss": 0.0193, "step": 1573 }, { "epoch": 0.98, "learning_rate": 1.0830019920581753e-05, "loss": 0.0206, "step": 1574 }, { "epoch": 0.98, "learning_rate": 1.0819981821802053e-05, "loss": 0.024, "step": 1575 }, { "epoch": 0.98, "learning_rate": 1.0809942891118408e-05, "loss": 0.0276, "step": 1576 }, { "epoch": 0.98, "learning_rate": 1.079990313871571e-05, "loss": 0.0326, "step": 1577 }, { "epoch": 0.98, "learning_rate": 1.0789862574779681e-05, "loss": 0.0209, "step": 1578 }, { "epoch": 0.98, "learning_rate": 1.0779821209496876e-05, "loss": 0.0238, "step": 1579 }, { "epoch": 0.98, "learning_rate": 1.0769779053054651e-05, "loss": 0.0225, "step": 1580 }, { "epoch": 0.98, "learning_rate": 1.0759736115641175e-05, "loss": 0.026, "step": 1581 }, { "epoch": 0.98, "learning_rate": 1.0749692407445405e-05, "loss": 0.0252, "step": 1582 }, { "epoch": 0.98, "learning_rate": 1.0739647938657074e-05, "loss": 0.0219, "step": 1583 }, { "epoch": 0.99, "learning_rate": 1.0729602719466692e-05, "loss": 0.0217, "step": 1584 }, { "epoch": 0.99, "learning_rate": 1.071955676006554e-05, "loss": 0.0273, "step": 1585 }, { "epoch": 0.99, "learning_rate": 1.0709510070645633e-05, "loss": 0.0231, "step": 1586 }, { "epoch": 0.99, "learning_rate": 1.0699462661399733e-05, "loss": 0.0284, "step": 1587 }, { "epoch": 0.99, "learning_rate": 1.0689414542521336e-05, "loss": 0.026, "step": 1588 }, { "epoch": 0.99, "learning_rate": 1.067936572420466e-05, "loss": 0.028, "step": 1589 }, { "epoch": 0.99, "learning_rate": 1.0669316216644623e-05, "loss": 0.0218, "step": 1590 }, { "epoch": 0.99, "learning_rate": 1.0659266030036847e-05, "loss": 0.0194, "step": 1591 }, { "epoch": 0.99, "learning_rate": 1.0649215174577645e-05, "loss": 0.0262, "step": 1592 }, { "epoch": 0.99, "learning_rate": 1.0639163660464007e-05, "loss": 0.0237, "step": 1593 }, { "epoch": 0.99, "learning_rate": 1.0629111497893591e-05, "loss": 0.0281, "step": 1594 }, { "epoch": 0.99, "learning_rate": 1.061905869706471e-05, "loss": 0.035, "step": 1595 }, { "epoch": 0.99, "learning_rate": 1.0609005268176333e-05, "loss": 0.028, "step": 1596 }, { "epoch": 0.99, "learning_rate": 1.0598951221428055e-05, "loss": 0.0264, "step": 1597 }, { "epoch": 0.99, "learning_rate": 1.0588896567020106e-05, "loss": 0.0278, "step": 1598 }, { "epoch": 0.99, "learning_rate": 1.0578841315153333e-05, "loss": 0.0206, "step": 1599 }, { "epoch": 1.0, "learning_rate": 1.056878547602918e-05, "loss": 0.0269, "step": 1600 }, { "epoch": 1.0, "learning_rate": 1.05587290598497e-05, "loss": 0.0314, "step": 1601 }, { "epoch": 1.0, "learning_rate": 1.0548672076817516e-05, "loss": 0.0219, "step": 1602 }, { "epoch": 1.0, "learning_rate": 1.0538614537135843e-05, "loss": 0.0295, "step": 1603 }, { "epoch": 1.0, "learning_rate": 1.0528556451008447e-05, "loss": 0.0294, "step": 1604 }, { "epoch": 1.0, "learning_rate": 1.0518497828639658e-05, "loss": 0.0173, "step": 1605 }, { "epoch": 1.0, "learning_rate": 1.0508438680234337e-05, "loss": 0.0285, "step": 1606 }, { "epoch": 1.0, "learning_rate": 1.0498379015997898e-05, "loss": 0.0213, "step": 1607 }, { "epoch": 1.0, "learning_rate": 1.0488318846136261e-05, "loss": 0.0228, "step": 1608 }, { "epoch": 1.0, "learning_rate": 1.0478258180855869e-05, "loss": 0.0268, "step": 1609 }, { "epoch": 1.0, "learning_rate": 1.046819703036366e-05, "loss": 0.0252, "step": 1610 }, { "epoch": 1.0, "learning_rate": 1.0458135404867071e-05, "loss": 0.0267, "step": 1611 }, { "epoch": 1.0, "learning_rate": 1.0448073314574024e-05, "loss": 0.0175, "step": 1612 }, { "epoch": 1.0, "learning_rate": 1.04380107696929e-05, "loss": 0.0205, "step": 1613 }, { "epoch": 1.0, "learning_rate": 1.0427947780432547e-05, "loss": 0.0261, "step": 1614 }, { "epoch": 1.0, "learning_rate": 1.0417884357002268e-05, "loss": 0.023, "step": 1615 }, { "epoch": 1.0, "learning_rate": 1.0407820509611806e-05, "loss": 0.0272, "step": 1616 }, { "epoch": 1.01, "learning_rate": 1.0397756248471328e-05, "loss": 0.0204, "step": 1617 }, { "epoch": 1.01, "learning_rate": 1.038769158379143e-05, "loss": 0.0181, "step": 1618 }, { "epoch": 1.01, "learning_rate": 1.0377626525783101e-05, "loss": 0.0184, "step": 1619 }, { "epoch": 1.01, "learning_rate": 1.0367561084657752e-05, "loss": 0.0255, "step": 1620 }, { "epoch": 1.01, "learning_rate": 1.0357495270627162e-05, "loss": 0.0233, "step": 1621 }, { "epoch": 1.01, "learning_rate": 1.03474290939035e-05, "loss": 0.024, "step": 1622 }, { "epoch": 1.01, "learning_rate": 1.0337362564699303e-05, "loss": 0.0262, "step": 1623 }, { "epoch": 1.01, "learning_rate": 1.0327295693227454e-05, "loss": 0.0168, "step": 1624 }, { "epoch": 1.01, "learning_rate": 1.03172284897012e-05, "loss": 0.0273, "step": 1625 }, { "epoch": 1.01, "learning_rate": 1.0307160964334118e-05, "loss": 0.0174, "step": 1626 }, { "epoch": 1.01, "learning_rate": 1.0297093127340098e-05, "loss": 0.0247, "step": 1627 }, { "epoch": 1.01, "learning_rate": 1.0287024988933369e-05, "loss": 0.0226, "step": 1628 }, { "epoch": 1.01, "learning_rate": 1.0276956559328455e-05, "loss": 0.0194, "step": 1629 }, { "epoch": 1.01, "learning_rate": 1.026688784874017e-05, "loss": 0.0169, "step": 1630 }, { "epoch": 1.01, "learning_rate": 1.0256818867383623e-05, "loss": 0.0211, "step": 1631 }, { "epoch": 1.01, "learning_rate": 1.0246749625474189e-05, "loss": 0.0204, "step": 1632 }, { "epoch": 1.02, "learning_rate": 1.0236680133227515e-05, "loss": 0.0186, "step": 1633 }, { "epoch": 1.02, "learning_rate": 1.0226610400859498e-05, "loss": 0.0236, "step": 1634 }, { "epoch": 1.02, "learning_rate": 1.021654043858628e-05, "loss": 0.0264, "step": 1635 }, { "epoch": 1.02, "learning_rate": 1.0206470256624233e-05, "loss": 0.0268, "step": 1636 }, { "epoch": 1.02, "learning_rate": 1.0196399865189956e-05, "loss": 0.0202, "step": 1637 }, { "epoch": 1.02, "learning_rate": 1.0186329274500258e-05, "loss": 0.0175, "step": 1638 }, { "epoch": 1.02, "learning_rate": 1.0176258494772153e-05, "loss": 0.0227, "step": 1639 }, { "epoch": 1.02, "learning_rate": 1.0166187536222844e-05, "loss": 0.02, "step": 1640 }, { "epoch": 1.02, "learning_rate": 1.0156116409069711e-05, "loss": 0.0266, "step": 1641 }, { "epoch": 1.02, "learning_rate": 1.014604512353032e-05, "loss": 0.0189, "step": 1642 }, { "epoch": 1.02, "learning_rate": 1.0135973689822379e-05, "loss": 0.0243, "step": 1643 }, { "epoch": 1.02, "learning_rate": 1.0125902118163762e-05, "loss": 0.0214, "step": 1644 }, { "epoch": 1.02, "learning_rate": 1.0115830418772474e-05, "loss": 0.0313, "step": 1645 }, { "epoch": 1.02, "learning_rate": 1.0105758601866646e-05, "loss": 0.0262, "step": 1646 }, { "epoch": 1.02, "learning_rate": 1.0095686677664544e-05, "loss": 0.0224, "step": 1647 }, { "epoch": 1.02, "learning_rate": 1.0085614656384529e-05, "loss": 0.0201, "step": 1648 }, { "epoch": 1.03, "learning_rate": 1.007554254824506e-05, "loss": 0.0189, "step": 1649 }, { "epoch": 1.03, "learning_rate": 1.0065470363464694e-05, "loss": 0.0143, "step": 1650 }, { "epoch": 1.03, "learning_rate": 1.005539811226206e-05, "loss": 0.0261, "step": 1651 }, { "epoch": 1.03, "learning_rate": 1.004532580485585e-05, "loss": 0.0171, "step": 1652 }, { "epoch": 1.03, "learning_rate": 1.0035253451464824e-05, "loss": 0.0247, "step": 1653 }, { "epoch": 1.03, "learning_rate": 1.0025181062307774e-05, "loss": 0.0278, "step": 1654 }, { "epoch": 1.03, "learning_rate": 1.0015108647603545e-05, "loss": 0.0185, "step": 1655 }, { "epoch": 1.03, "learning_rate": 1.0005036217570999e-05, "loss": 0.026, "step": 1656 }, { "epoch": 1.03, "learning_rate": 9.994963782429006e-06, "loss": 0.0206, "step": 1657 }, { "epoch": 1.03, "learning_rate": 9.984891352396455e-06, "loss": 0.0317, "step": 1658 }, { "epoch": 1.03, "learning_rate": 9.974818937692228e-06, "loss": 0.0282, "step": 1659 }, { "epoch": 1.03, "learning_rate": 9.96474654853518e-06, "loss": 0.0271, "step": 1660 }, { "epoch": 1.03, "learning_rate": 9.95467419514415e-06, "loss": 0.0204, "step": 1661 }, { "epoch": 1.03, "learning_rate": 9.944601887737946e-06, "loss": 0.0274, "step": 1662 }, { "epoch": 1.03, "learning_rate": 9.93452963653531e-06, "loss": 0.0221, "step": 1663 }, { "epoch": 1.03, "learning_rate": 9.92445745175494e-06, "loss": 0.0249, "step": 1664 }, { "epoch": 1.04, "learning_rate": 9.914385343615476e-06, "loss": 0.0175, "step": 1665 }, { "epoch": 1.04, "learning_rate": 9.904313322335459e-06, "loss": 0.0244, "step": 1666 }, { "epoch": 1.04, "learning_rate": 9.894241398133354e-06, "loss": 0.0204, "step": 1667 }, { "epoch": 1.04, "learning_rate": 9.884169581227532e-06, "loss": 0.0309, "step": 1668 }, { "epoch": 1.04, "learning_rate": 9.874097881836241e-06, "loss": 0.0217, "step": 1669 }, { "epoch": 1.04, "learning_rate": 9.864026310177621e-06, "loss": 0.0237, "step": 1670 }, { "epoch": 1.04, "learning_rate": 9.853954876469685e-06, "loss": 0.0213, "step": 1671 }, { "epoch": 1.04, "learning_rate": 9.84388359093029e-06, "loss": 0.0214, "step": 1672 }, { "epoch": 1.04, "learning_rate": 9.83381246377716e-06, "loss": 0.0257, "step": 1673 }, { "epoch": 1.04, "learning_rate": 9.823741505227852e-06, "loss": 0.0222, "step": 1674 }, { "epoch": 1.04, "learning_rate": 9.813670725499745e-06, "loss": 0.0244, "step": 1675 }, { "epoch": 1.04, "learning_rate": 9.803600134810048e-06, "loss": 0.0257, "step": 1676 }, { "epoch": 1.04, "learning_rate": 9.793529743375769e-06, "loss": 0.0228, "step": 1677 }, { "epoch": 1.04, "learning_rate": 9.783459561413724e-06, "loss": 0.0261, "step": 1678 }, { "epoch": 1.04, "learning_rate": 9.773389599140504e-06, "loss": 0.0252, "step": 1679 }, { "epoch": 1.04, "learning_rate": 9.763319866772486e-06, "loss": 0.0216, "step": 1680 }, { "epoch": 1.05, "learning_rate": 9.753250374525813e-06, "loss": 0.025, "step": 1681 }, { "epoch": 1.05, "learning_rate": 9.743181132616379e-06, "loss": 0.0264, "step": 1682 }, { "epoch": 1.05, "learning_rate": 9.73311215125983e-06, "loss": 0.0237, "step": 1683 }, { "epoch": 1.05, "learning_rate": 9.72304344067155e-06, "loss": 0.0195, "step": 1684 }, { "epoch": 1.05, "learning_rate": 9.712975011066633e-06, "loss": 0.0255, "step": 1685 }, { "epoch": 1.05, "learning_rate": 9.7029068726599e-06, "loss": 0.023, "step": 1686 }, { "epoch": 1.05, "learning_rate": 9.692839035665889e-06, "loss": 0.024, "step": 1687 }, { "epoch": 1.05, "learning_rate": 9.682771510298801e-06, "loss": 0.02, "step": 1688 }, { "epoch": 1.05, "learning_rate": 9.672704306772547e-06, "loss": 0.0231, "step": 1689 }, { "epoch": 1.05, "learning_rate": 9.662637435300702e-06, "loss": 0.0179, "step": 1690 }, { "epoch": 1.05, "learning_rate": 9.652570906096501e-06, "loss": 0.0239, "step": 1691 }, { "epoch": 1.05, "learning_rate": 9.64250472937284e-06, "loss": 0.0224, "step": 1692 }, { "epoch": 1.05, "learning_rate": 9.632438915342253e-06, "loss": 0.024, "step": 1693 }, { "epoch": 1.05, "learning_rate": 9.6223734742169e-06, "loss": 0.0218, "step": 1694 }, { "epoch": 1.05, "learning_rate": 9.612308416208573e-06, "loss": 0.0212, "step": 1695 }, { "epoch": 1.05, "learning_rate": 9.602243751528672e-06, "loss": 0.0208, "step": 1696 }, { "epoch": 1.06, "learning_rate": 9.592179490388197e-06, "loss": 0.0199, "step": 1697 }, { "epoch": 1.06, "learning_rate": 9.582115642997735e-06, "loss": 0.0278, "step": 1698 }, { "epoch": 1.06, "learning_rate": 9.572052219567455e-06, "loss": 0.0286, "step": 1699 }, { "epoch": 1.06, "learning_rate": 9.561989230307106e-06, "loss": 0.0203, "step": 1700 }, { "epoch": 1.06, "learning_rate": 9.55192668542598e-06, "loss": 0.0255, "step": 1701 }, { "epoch": 1.06, "learning_rate": 9.541864595132929e-06, "loss": 0.0157, "step": 1702 }, { "epoch": 1.06, "learning_rate": 9.531802969636343e-06, "loss": 0.03, "step": 1703 }, { "epoch": 1.06, "learning_rate": 9.521741819144135e-06, "loss": 0.0266, "step": 1704 }, { "epoch": 1.06, "learning_rate": 9.51168115386374e-06, "loss": 0.0207, "step": 1705 }, { "epoch": 1.06, "learning_rate": 9.501620984002107e-06, "loss": 0.0173, "step": 1706 }, { "epoch": 1.06, "learning_rate": 9.491561319765665e-06, "loss": 0.0244, "step": 1707 }, { "epoch": 1.06, "learning_rate": 9.481502171360345e-06, "loss": 0.025, "step": 1708 }, { "epoch": 1.06, "learning_rate": 9.471443548991557e-06, "loss": 0.0264, "step": 1709 }, { "epoch": 1.06, "learning_rate": 9.46138546286416e-06, "loss": 0.0283, "step": 1710 }, { "epoch": 1.06, "learning_rate": 9.451327923182486e-06, "loss": 0.0208, "step": 1711 }, { "epoch": 1.06, "learning_rate": 9.441270940150305e-06, "loss": 0.019, "step": 1712 }, { "epoch": 1.07, "learning_rate": 9.431214523970822e-06, "loss": 0.0268, "step": 1713 }, { "epoch": 1.07, "learning_rate": 9.421158684846669e-06, "loss": 0.015, "step": 1714 }, { "epoch": 1.07, "learning_rate": 9.411103432979895e-06, "loss": 0.0236, "step": 1715 }, { "epoch": 1.07, "learning_rate": 9.401048778571948e-06, "loss": 0.0156, "step": 1716 }, { "epoch": 1.07, "learning_rate": 9.39099473182367e-06, "loss": 0.016, "step": 1717 }, { "epoch": 1.07, "learning_rate": 9.38094130293529e-06, "loss": 0.028, "step": 1718 }, { "epoch": 1.07, "learning_rate": 9.370888502106414e-06, "loss": 0.0181, "step": 1719 }, { "epoch": 1.07, "learning_rate": 9.360836339535997e-06, "loss": 0.0271, "step": 1720 }, { "epoch": 1.07, "learning_rate": 9.350784825422357e-06, "loss": 0.0245, "step": 1721 }, { "epoch": 1.07, "learning_rate": 9.340733969963156e-06, "loss": 0.023, "step": 1722 }, { "epoch": 1.07, "learning_rate": 9.330683783355379e-06, "loss": 0.031, "step": 1723 }, { "epoch": 1.07, "learning_rate": 9.320634275795342e-06, "loss": 0.0244, "step": 1724 }, { "epoch": 1.07, "learning_rate": 9.310585457478665e-06, "loss": 0.0282, "step": 1725 }, { "epoch": 1.07, "learning_rate": 9.30053733860027e-06, "loss": 0.0297, "step": 1726 }, { "epoch": 1.07, "learning_rate": 9.29048992935437e-06, "loss": 0.024, "step": 1727 }, { "epoch": 1.07, "learning_rate": 9.280443239934467e-06, "loss": 0.0206, "step": 1728 }, { "epoch": 1.08, "learning_rate": 9.270397280533311e-06, "loss": 0.0237, "step": 1729 }, { "epoch": 1.08, "learning_rate": 9.260352061342928e-06, "loss": 0.0205, "step": 1730 }, { "epoch": 1.08, "learning_rate": 9.250307592554601e-06, "loss": 0.0239, "step": 1731 }, { "epoch": 1.08, "learning_rate": 9.240263884358827e-06, "loss": 0.0272, "step": 1732 }, { "epoch": 1.08, "learning_rate": 9.23022094694535e-06, "loss": 0.0249, "step": 1733 }, { "epoch": 1.08, "learning_rate": 9.220178790503125e-06, "loss": 0.0223, "step": 1734 }, { "epoch": 1.08, "learning_rate": 9.21013742522032e-06, "loss": 0.0192, "step": 1735 }, { "epoch": 1.08, "learning_rate": 9.200096861284293e-06, "loss": 0.0294, "step": 1736 }, { "epoch": 1.08, "learning_rate": 9.190057108881594e-06, "loss": 0.021, "step": 1737 }, { "epoch": 1.08, "learning_rate": 9.180018178197949e-06, "loss": 0.033, "step": 1738 }, { "epoch": 1.08, "learning_rate": 9.169980079418248e-06, "loss": 0.0347, "step": 1739 }, { "epoch": 1.08, "learning_rate": 9.159942822726539e-06, "loss": 0.021, "step": 1740 }, { "epoch": 1.08, "learning_rate": 9.149906418306022e-06, "loss": 0.021, "step": 1741 }, { "epoch": 1.08, "learning_rate": 9.139870876339013e-06, "loss": 0.0234, "step": 1742 }, { "epoch": 1.08, "learning_rate": 9.129836207006974e-06, "loss": 0.0282, "step": 1743 }, { "epoch": 1.08, "learning_rate": 9.119802420490473e-06, "loss": 0.0256, "step": 1744 }, { "epoch": 1.09, "learning_rate": 9.109769526969179e-06, "loss": 0.023, "step": 1745 }, { "epoch": 1.09, "learning_rate": 9.099737536621862e-06, "loss": 0.0264, "step": 1746 }, { "epoch": 1.09, "learning_rate": 9.089706459626374e-06, "loss": 0.0205, "step": 1747 }, { "epoch": 1.09, "learning_rate": 9.079676306159633e-06, "loss": 0.0224, "step": 1748 }, { "epoch": 1.09, "learning_rate": 9.06964708639763e-06, "loss": 0.0234, "step": 1749 }, { "epoch": 1.09, "learning_rate": 9.059618810515405e-06, "loss": 0.0206, "step": 1750 }, { "epoch": 1.09, "learning_rate": 9.049591488687042e-06, "loss": 0.0213, "step": 1751 }, { "epoch": 1.09, "learning_rate": 9.039565131085647e-06, "loss": 0.0216, "step": 1752 }, { "epoch": 1.09, "learning_rate": 9.02953974788336e-06, "loss": 0.0186, "step": 1753 }, { "epoch": 1.09, "learning_rate": 9.019515349251337e-06, "loss": 0.0154, "step": 1754 }, { "epoch": 1.09, "learning_rate": 9.009491945359711e-06, "loss": 0.0252, "step": 1755 }, { "epoch": 1.09, "learning_rate": 8.99946954637763e-06, "loss": 0.0207, "step": 1756 }, { "epoch": 1.09, "learning_rate": 8.989448162473213e-06, "loss": 0.0245, "step": 1757 }, { "epoch": 1.09, "learning_rate": 8.979427803813548e-06, "loss": 0.0219, "step": 1758 }, { "epoch": 1.09, "learning_rate": 8.969408480564684e-06, "loss": 0.0289, "step": 1759 }, { "epoch": 1.09, "learning_rate": 8.959390202891625e-06, "loss": 0.0159, "step": 1760 }, { "epoch": 1.1, "learning_rate": 8.949372980958302e-06, "loss": 0.0261, "step": 1761 }, { "epoch": 1.1, "learning_rate": 8.939356824927589e-06, "loss": 0.0215, "step": 1762 }, { "epoch": 1.1, "learning_rate": 8.929341744961272e-06, "loss": 0.0253, "step": 1763 }, { "epoch": 1.1, "learning_rate": 8.919327751220038e-06, "loss": 0.0282, "step": 1764 }, { "epoch": 1.1, "learning_rate": 8.909314853863483e-06, "loss": 0.0242, "step": 1765 }, { "epoch": 1.1, "learning_rate": 8.899303063050091e-06, "loss": 0.0274, "step": 1766 }, { "epoch": 1.1, "learning_rate": 8.889292388937214e-06, "loss": 0.0179, "step": 1767 }, { "epoch": 1.1, "learning_rate": 8.879282841681077e-06, "loss": 0.0185, "step": 1768 }, { "epoch": 1.1, "learning_rate": 8.86927443143676e-06, "loss": 0.0156, "step": 1769 }, { "epoch": 1.1, "learning_rate": 8.859267168358198e-06, "loss": 0.0168, "step": 1770 }, { "epoch": 1.1, "learning_rate": 8.849261062598144e-06, "loss": 0.0194, "step": 1771 }, { "epoch": 1.1, "learning_rate": 8.839256124308194e-06, "loss": 0.0187, "step": 1772 }, { "epoch": 1.1, "learning_rate": 8.829252363638756e-06, "loss": 0.0187, "step": 1773 }, { "epoch": 1.1, "learning_rate": 8.819249790739033e-06, "loss": 0.0198, "step": 1774 }, { "epoch": 1.1, "learning_rate": 8.809248415757028e-06, "loss": 0.0214, "step": 1775 }, { "epoch": 1.1, "learning_rate": 8.799248248839545e-06, "loss": 0.0179, "step": 1776 }, { "epoch": 1.11, "learning_rate": 8.789249300132135e-06, "loss": 0.0217, "step": 1777 }, { "epoch": 1.11, "learning_rate": 8.779251579779131e-06, "loss": 0.0306, "step": 1778 }, { "epoch": 1.11, "learning_rate": 8.769255097923617e-06, "loss": 0.0231, "step": 1779 }, { "epoch": 1.11, "learning_rate": 8.759259864707415e-06, "loss": 0.0258, "step": 1780 }, { "epoch": 1.11, "learning_rate": 8.749265890271084e-06, "loss": 0.0218, "step": 1781 }, { "epoch": 1.11, "learning_rate": 8.73927318475391e-06, "loss": 0.0261, "step": 1782 }, { "epoch": 1.11, "learning_rate": 8.729281758293884e-06, "loss": 0.0238, "step": 1783 }, { "epoch": 1.11, "learning_rate": 8.719291621027703e-06, "loss": 0.0219, "step": 1784 }, { "epoch": 1.11, "learning_rate": 8.70930278309076e-06, "loss": 0.0192, "step": 1785 }, { "epoch": 1.11, "learning_rate": 8.69931525461712e-06, "loss": 0.0197, "step": 1786 }, { "epoch": 1.11, "learning_rate": 8.689329045739526e-06, "loss": 0.0236, "step": 1787 }, { "epoch": 1.11, "learning_rate": 8.679344166589382e-06, "loss": 0.0138, "step": 1788 }, { "epoch": 1.11, "learning_rate": 8.669360627296745e-06, "loss": 0.0189, "step": 1789 }, { "epoch": 1.11, "learning_rate": 8.659378437990305e-06, "loss": 0.0251, "step": 1790 }, { "epoch": 1.11, "learning_rate": 8.649397608797391e-06, "loss": 0.0182, "step": 1791 }, { "epoch": 1.11, "learning_rate": 8.63941814984395e-06, "loss": 0.0242, "step": 1792 }, { "epoch": 1.12, "learning_rate": 8.629440071254533e-06, "loss": 0.0176, "step": 1793 }, { "epoch": 1.12, "learning_rate": 8.619463383152296e-06, "loss": 0.0207, "step": 1794 }, { "epoch": 1.12, "learning_rate": 8.609488095658987e-06, "loss": 0.0116, "step": 1795 }, { "epoch": 1.12, "learning_rate": 8.599514218894923e-06, "loss": 0.0293, "step": 1796 }, { "epoch": 1.12, "learning_rate": 8.589541762978995e-06, "loss": 0.0165, "step": 1797 }, { "epoch": 1.12, "learning_rate": 8.579570738028665e-06, "loss": 0.0168, "step": 1798 }, { "epoch": 1.12, "learning_rate": 8.56960115415992e-06, "loss": 0.0168, "step": 1799 }, { "epoch": 1.12, "learning_rate": 8.559633021487298e-06, "loss": 0.0309, "step": 1800 }, { "epoch": 1.12, "learning_rate": 8.549666350123868e-06, "loss": 0.0226, "step": 1801 }, { "epoch": 1.12, "learning_rate": 8.53970115018121e-06, "loss": 0.0185, "step": 1802 }, { "epoch": 1.12, "learning_rate": 8.52973743176941e-06, "loss": 0.0162, "step": 1803 }, { "epoch": 1.12, "learning_rate": 8.519775204997063e-06, "loss": 0.0267, "step": 1804 }, { "epoch": 1.12, "learning_rate": 8.509814479971226e-06, "loss": 0.0237, "step": 1805 }, { "epoch": 1.12, "learning_rate": 8.49985526679746e-06, "loss": 0.0229, "step": 1806 }, { "epoch": 1.12, "learning_rate": 8.489897575579777e-06, "loss": 0.0179, "step": 1807 }, { "epoch": 1.12, "learning_rate": 8.479941416420652e-06, "loss": 0.0279, "step": 1808 }, { "epoch": 1.12, "learning_rate": 8.469986799420993e-06, "loss": 0.0169, "step": 1809 }, { "epoch": 1.13, "learning_rate": 8.460033734680158e-06, "loss": 0.0221, "step": 1810 }, { "epoch": 1.13, "learning_rate": 8.450082232295924e-06, "loss": 0.0221, "step": 1811 }, { "epoch": 1.13, "learning_rate": 8.44013230236448e-06, "loss": 0.0208, "step": 1812 }, { "epoch": 1.13, "learning_rate": 8.430183954980426e-06, "loss": 0.0166, "step": 1813 }, { "epoch": 1.13, "learning_rate": 8.420237200236753e-06, "loss": 0.0256, "step": 1814 }, { "epoch": 1.13, "learning_rate": 8.410292048224838e-06, "loss": 0.0182, "step": 1815 }, { "epoch": 1.13, "learning_rate": 8.400348509034425e-06, "loss": 0.021, "step": 1816 }, { "epoch": 1.13, "learning_rate": 8.390406592753636e-06, "loss": 0.0227, "step": 1817 }, { "epoch": 1.13, "learning_rate": 8.38046630946893e-06, "loss": 0.0167, "step": 1818 }, { "epoch": 1.13, "learning_rate": 8.370527669265114e-06, "loss": 0.0175, "step": 1819 }, { "epoch": 1.13, "learning_rate": 8.360590682225345e-06, "loss": 0.0225, "step": 1820 }, { "epoch": 1.13, "learning_rate": 8.350655358431075e-06, "loss": 0.0279, "step": 1821 }, { "epoch": 1.13, "learning_rate": 8.34072170796209e-06, "loss": 0.0159, "step": 1822 }, { "epoch": 1.13, "learning_rate": 8.33078974089647e-06, "loss": 0.0214, "step": 1823 }, { "epoch": 1.13, "learning_rate": 8.320859467310582e-06, "loss": 0.0215, "step": 1824 }, { "epoch": 1.13, "learning_rate": 8.310930897279088e-06, "loss": 0.0238, "step": 1825 }, { "epoch": 1.14, "learning_rate": 8.301004040874911e-06, "loss": 0.0214, "step": 1826 }, { "epoch": 1.14, "learning_rate": 8.291078908169242e-06, "loss": 0.0289, "step": 1827 }, { "epoch": 1.14, "learning_rate": 8.281155509231515e-06, "loss": 0.023, "step": 1828 }, { "epoch": 1.14, "learning_rate": 8.271233854129413e-06, "loss": 0.021, "step": 1829 }, { "epoch": 1.14, "learning_rate": 8.261313952928853e-06, "loss": 0.0208, "step": 1830 }, { "epoch": 1.14, "learning_rate": 8.251395815693953e-06, "loss": 0.0166, "step": 1831 }, { "epoch": 1.14, "learning_rate": 8.241479452487065e-06, "loss": 0.0206, "step": 1832 }, { "epoch": 1.14, "learning_rate": 8.231564873368729e-06, "loss": 0.021, "step": 1833 }, { "epoch": 1.14, "learning_rate": 8.221652088397675e-06, "loss": 0.0224, "step": 1834 }, { "epoch": 1.14, "learning_rate": 8.211741107630817e-06, "loss": 0.0184, "step": 1835 }, { "epoch": 1.14, "learning_rate": 8.201831941123237e-06, "loss": 0.0199, "step": 1836 }, { "epoch": 1.14, "learning_rate": 8.191924598928174e-06, "loss": 0.0139, "step": 1837 }, { "epoch": 1.14, "learning_rate": 8.182019091097017e-06, "loss": 0.0217, "step": 1838 }, { "epoch": 1.14, "learning_rate": 8.172115427679304e-06, "loss": 0.0211, "step": 1839 }, { "epoch": 1.14, "learning_rate": 8.162213618722679e-06, "loss": 0.0157, "step": 1840 }, { "epoch": 1.14, "learning_rate": 8.152313674272923e-06, "loss": 0.0148, "step": 1841 }, { "epoch": 1.15, "learning_rate": 8.142415604373928e-06, "loss": 0.0166, "step": 1842 }, { "epoch": 1.15, "learning_rate": 8.132519419067668e-06, "loss": 0.0193, "step": 1843 }, { "epoch": 1.15, "learning_rate": 8.122625128394216e-06, "loss": 0.018, "step": 1844 }, { "epoch": 1.15, "learning_rate": 8.112732742391722e-06, "loss": 0.0207, "step": 1845 }, { "epoch": 1.15, "learning_rate": 8.102842271096405e-06, "loss": 0.0169, "step": 1846 }, { "epoch": 1.15, "learning_rate": 8.092953724542532e-06, "loss": 0.0206, "step": 1847 }, { "epoch": 1.15, "learning_rate": 8.08306711276243e-06, "loss": 0.0278, "step": 1848 }, { "epoch": 1.15, "learning_rate": 8.073182445786455e-06, "loss": 0.0199, "step": 1849 }, { "epoch": 1.15, "learning_rate": 8.063299733642991e-06, "loss": 0.0179, "step": 1850 }, { "epoch": 1.15, "learning_rate": 8.053418986358443e-06, "loss": 0.0197, "step": 1851 }, { "epoch": 1.15, "learning_rate": 8.043540213957219e-06, "loss": 0.0241, "step": 1852 }, { "epoch": 1.15, "learning_rate": 8.033663426461717e-06, "loss": 0.0311, "step": 1853 }, { "epoch": 1.15, "learning_rate": 8.023788633892334e-06, "loss": 0.0153, "step": 1854 }, { "epoch": 1.15, "learning_rate": 8.013915846267439e-06, "loss": 0.0155, "step": 1855 }, { "epoch": 1.15, "learning_rate": 8.00404507360336e-06, "loss": 0.0211, "step": 1856 }, { "epoch": 1.15, "learning_rate": 7.994176325914387e-06, "loss": 0.017, "step": 1857 }, { "epoch": 1.16, "learning_rate": 7.984309613212757e-06, "loss": 0.0228, "step": 1858 }, { "epoch": 1.16, "learning_rate": 7.974444945508637e-06, "loss": 0.0263, "step": 1859 }, { "epoch": 1.16, "learning_rate": 7.964582332810122e-06, "loss": 0.0206, "step": 1860 }, { "epoch": 1.16, "learning_rate": 7.954721785123227e-06, "loss": 0.0204, "step": 1861 }, { "epoch": 1.16, "learning_rate": 7.94486331245186e-06, "loss": 0.0185, "step": 1862 }, { "epoch": 1.16, "learning_rate": 7.935006924797832e-06, "loss": 0.0265, "step": 1863 }, { "epoch": 1.16, "learning_rate": 7.925152632160841e-06, "loss": 0.0176, "step": 1864 }, { "epoch": 1.16, "learning_rate": 7.915300444538458e-06, "loss": 0.0223, "step": 1865 }, { "epoch": 1.16, "learning_rate": 7.905450371926112e-06, "loss": 0.0218, "step": 1866 }, { "epoch": 1.16, "learning_rate": 7.895602424317089e-06, "loss": 0.0231, "step": 1867 }, { "epoch": 1.16, "learning_rate": 7.885756611702527e-06, "loss": 0.0217, "step": 1868 }, { "epoch": 1.16, "learning_rate": 7.875912944071386e-06, "loss": 0.0258, "step": 1869 }, { "epoch": 1.16, "learning_rate": 7.866071431410458e-06, "loss": 0.017, "step": 1870 }, { "epoch": 1.16, "learning_rate": 7.856232083704348e-06, "loss": 0.0214, "step": 1871 }, { "epoch": 1.16, "learning_rate": 7.84639491093546e-06, "loss": 0.0374, "step": 1872 }, { "epoch": 1.16, "learning_rate": 7.836559923083997e-06, "loss": 0.0173, "step": 1873 }, { "epoch": 1.17, "learning_rate": 7.826727130127942e-06, "loss": 0.0205, "step": 1874 }, { "epoch": 1.17, "learning_rate": 7.816896542043048e-06, "loss": 0.0216, "step": 1875 }, { "epoch": 1.17, "learning_rate": 7.80706816880284e-06, "loss": 0.0213, "step": 1876 }, { "epoch": 1.17, "learning_rate": 7.79724202037859e-06, "loss": 0.0201, "step": 1877 }, { "epoch": 1.17, "learning_rate": 7.787418106739308e-06, "loss": 0.0245, "step": 1878 }, { "epoch": 1.17, "learning_rate": 7.77759643785175e-06, "loss": 0.0222, "step": 1879 }, { "epoch": 1.17, "learning_rate": 7.767777023680384e-06, "loss": 0.0193, "step": 1880 }, { "epoch": 1.17, "learning_rate": 7.757959874187392e-06, "loss": 0.0251, "step": 1881 }, { "epoch": 1.17, "learning_rate": 7.748144999332661e-06, "loss": 0.023, "step": 1882 }, { "epoch": 1.17, "learning_rate": 7.738332409073768e-06, "loss": 0.0228, "step": 1883 }, { "epoch": 1.17, "learning_rate": 7.72852211336598e-06, "loss": 0.0247, "step": 1884 }, { "epoch": 1.17, "learning_rate": 7.71871412216222e-06, "loss": 0.0199, "step": 1885 }, { "epoch": 1.17, "learning_rate": 7.708908445413089e-06, "loss": 0.0197, "step": 1886 }, { "epoch": 1.17, "learning_rate": 7.699105093066833e-06, "loss": 0.0207, "step": 1887 }, { "epoch": 1.17, "learning_rate": 7.689304075069335e-06, "loss": 0.0213, "step": 1888 }, { "epoch": 1.17, "learning_rate": 7.679505401364116e-06, "loss": 0.0185, "step": 1889 }, { "epoch": 1.18, "learning_rate": 7.669709081892324e-06, "loss": 0.0187, "step": 1890 }, { "epoch": 1.18, "learning_rate": 7.659915126592706e-06, "loss": 0.0204, "step": 1891 }, { "epoch": 1.18, "learning_rate": 7.650123545401617e-06, "loss": 0.0251, "step": 1892 }, { "epoch": 1.18, "learning_rate": 7.640334348253007e-06, "loss": 0.0283, "step": 1893 }, { "epoch": 1.18, "learning_rate": 7.630547545078398e-06, "loss": 0.0136, "step": 1894 }, { "epoch": 1.18, "learning_rate": 7.62076314580689e-06, "loss": 0.0277, "step": 1895 }, { "epoch": 1.18, "learning_rate": 7.610981160365147e-06, "loss": 0.02, "step": 1896 }, { "epoch": 1.18, "learning_rate": 7.6012015986773725e-06, "loss": 0.0167, "step": 1897 }, { "epoch": 1.18, "learning_rate": 7.5914244706653206e-06, "loss": 0.0261, "step": 1898 }, { "epoch": 1.18, "learning_rate": 7.581649786248276e-06, "loss": 0.0196, "step": 1899 }, { "epoch": 1.18, "learning_rate": 7.571877555343039e-06, "loss": 0.0167, "step": 1900 }, { "epoch": 1.18, "learning_rate": 7.562107787863926e-06, "loss": 0.0204, "step": 1901 }, { "epoch": 1.18, "learning_rate": 7.552340493722751e-06, "loss": 0.0262, "step": 1902 }, { "epoch": 1.18, "learning_rate": 7.542575682828822e-06, "loss": 0.0217, "step": 1903 }, { "epoch": 1.18, "learning_rate": 7.532813365088921e-06, "loss": 0.0214, "step": 1904 }, { "epoch": 1.18, "learning_rate": 7.523053550407308e-06, "loss": 0.0185, "step": 1905 }, { "epoch": 1.19, "learning_rate": 7.513296248685704e-06, "loss": 0.0238, "step": 1906 }, { "epoch": 1.19, "learning_rate": 7.503541469823266e-06, "loss": 0.0212, "step": 1907 }, { "epoch": 1.19, "learning_rate": 7.493789223716611e-06, "loss": 0.0234, "step": 1908 }, { "epoch": 1.19, "learning_rate": 7.484039520259781e-06, "loss": 0.0234, "step": 1909 }, { "epoch": 1.19, "learning_rate": 7.474292369344226e-06, "loss": 0.0202, "step": 1910 }, { "epoch": 1.19, "learning_rate": 7.46454778085882e-06, "loss": 0.0194, "step": 1911 }, { "epoch": 1.19, "learning_rate": 7.454805764689834e-06, "loss": 0.0256, "step": 1912 }, { "epoch": 1.19, "learning_rate": 7.4450663307209255e-06, "loss": 0.0256, "step": 1913 }, { "epoch": 1.19, "learning_rate": 7.435329488833137e-06, "loss": 0.0162, "step": 1914 }, { "epoch": 1.19, "learning_rate": 7.42559524890488e-06, "loss": 0.0151, "step": 1915 }, { "epoch": 1.19, "learning_rate": 7.415863620811923e-06, "loss": 0.0217, "step": 1916 }, { "epoch": 1.19, "learning_rate": 7.406134614427387e-06, "loss": 0.0208, "step": 1917 }, { "epoch": 1.19, "learning_rate": 7.396408239621734e-06, "loss": 0.0194, "step": 1918 }, { "epoch": 1.19, "learning_rate": 7.38668450626276e-06, "loss": 0.025, "step": 1919 }, { "epoch": 1.19, "learning_rate": 7.376963424215568e-06, "loss": 0.0146, "step": 1920 }, { "epoch": 1.19, "learning_rate": 7.367245003342583e-06, "loss": 0.0232, "step": 1921 }, { "epoch": 1.2, "learning_rate": 7.3575292535035335e-06, "loss": 0.0135, "step": 1922 }, { "epoch": 1.2, "learning_rate": 7.347816184555421e-06, "loss": 0.0274, "step": 1923 }, { "epoch": 1.2, "learning_rate": 7.338105806352542e-06, "loss": 0.0262, "step": 1924 }, { "epoch": 1.2, "learning_rate": 7.328398128746463e-06, "loss": 0.0129, "step": 1925 }, { "epoch": 1.2, "learning_rate": 7.318693161585999e-06, "loss": 0.0228, "step": 1926 }, { "epoch": 1.2, "learning_rate": 7.308990914717226e-06, "loss": 0.0196, "step": 1927 }, { "epoch": 1.2, "learning_rate": 7.29929139798346e-06, "loss": 0.0204, "step": 1928 }, { "epoch": 1.2, "learning_rate": 7.289594621225236e-06, "loss": 0.022, "step": 1929 }, { "epoch": 1.2, "learning_rate": 7.27990059428032e-06, "loss": 0.0181, "step": 1930 }, { "epoch": 1.2, "learning_rate": 7.270209326983694e-06, "loss": 0.0208, "step": 1931 }, { "epoch": 1.2, "learning_rate": 7.260520829167519e-06, "loss": 0.0178, "step": 1932 }, { "epoch": 1.2, "learning_rate": 7.250835110661163e-06, "loss": 0.0196, "step": 1933 }, { "epoch": 1.2, "learning_rate": 7.241152181291173e-06, "loss": 0.0186, "step": 1934 }, { "epoch": 1.2, "learning_rate": 7.23147205088126e-06, "loss": 0.0262, "step": 1935 }, { "epoch": 1.2, "learning_rate": 7.221794729252298e-06, "loss": 0.0252, "step": 1936 }, { "epoch": 1.2, "learning_rate": 7.212120226222314e-06, "loss": 0.0218, "step": 1937 }, { "epoch": 1.21, "learning_rate": 7.202448551606477e-06, "loss": 0.0197, "step": 1938 }, { "epoch": 1.21, "learning_rate": 7.192779715217075e-06, "loss": 0.0185, "step": 1939 }, { "epoch": 1.21, "learning_rate": 7.183113726863529e-06, "loss": 0.0169, "step": 1940 }, { "epoch": 1.21, "learning_rate": 7.173450596352369e-06, "loss": 0.0163, "step": 1941 }, { "epoch": 1.21, "learning_rate": 7.163790333487216e-06, "loss": 0.0152, "step": 1942 }, { "epoch": 1.21, "learning_rate": 7.154132948068788e-06, "loss": 0.0142, "step": 1943 }, { "epoch": 1.21, "learning_rate": 7.144478449894894e-06, "loss": 0.0262, "step": 1944 }, { "epoch": 1.21, "learning_rate": 7.134826848760392e-06, "loss": 0.0215, "step": 1945 }, { "epoch": 1.21, "learning_rate": 7.1251781544572175e-06, "loss": 0.0189, "step": 1946 }, { "epoch": 1.21, "learning_rate": 7.1155323767743546e-06, "loss": 0.0246, "step": 1947 }, { "epoch": 1.21, "learning_rate": 7.10588952549782e-06, "loss": 0.0186, "step": 1948 }, { "epoch": 1.21, "learning_rate": 7.096249610410671e-06, "loss": 0.0097, "step": 1949 }, { "epoch": 1.21, "learning_rate": 7.086612641292985e-06, "loss": 0.0275, "step": 1950 }, { "epoch": 1.21, "learning_rate": 7.076978627921838e-06, "loss": 0.016, "step": 1951 }, { "epoch": 1.21, "learning_rate": 7.067347580071326e-06, "loss": 0.0172, "step": 1952 }, { "epoch": 1.21, "learning_rate": 7.0577195075125285e-06, "loss": 0.0182, "step": 1953 }, { "epoch": 1.22, "learning_rate": 7.0480944200134975e-06, "loss": 0.0234, "step": 1954 }, { "epoch": 1.22, "learning_rate": 7.038472327339268e-06, "loss": 0.0216, "step": 1955 }, { "epoch": 1.22, "learning_rate": 7.028853239251834e-06, "loss": 0.0318, "step": 1956 }, { "epoch": 1.22, "learning_rate": 7.019237165510141e-06, "loss": 0.0192, "step": 1957 }, { "epoch": 1.22, "learning_rate": 7.009624115870073e-06, "loss": 0.0226, "step": 1958 }, { "epoch": 1.22, "learning_rate": 7.00001410008445e-06, "loss": 0.0178, "step": 1959 }, { "epoch": 1.22, "learning_rate": 6.990407127903015e-06, "loss": 0.0222, "step": 1960 }, { "epoch": 1.22, "learning_rate": 6.980803209072416e-06, "loss": 0.0161, "step": 1961 }, { "epoch": 1.22, "learning_rate": 6.971202353336211e-06, "loss": 0.0188, "step": 1962 }, { "epoch": 1.22, "learning_rate": 6.961604570434849e-06, "loss": 0.0156, "step": 1963 }, { "epoch": 1.22, "learning_rate": 6.952009870105654e-06, "loss": 0.0218, "step": 1964 }, { "epoch": 1.22, "learning_rate": 6.942418262082833e-06, "loss": 0.0244, "step": 1965 }, { "epoch": 1.22, "learning_rate": 6.932829756097452e-06, "loss": 0.0213, "step": 1966 }, { "epoch": 1.22, "learning_rate": 6.923244361877424e-06, "loss": 0.024, "step": 1967 }, { "epoch": 1.22, "learning_rate": 6.913662089147514e-06, "loss": 0.0199, "step": 1968 }, { "epoch": 1.22, "learning_rate": 6.904082947629317e-06, "loss": 0.0224, "step": 1969 }, { "epoch": 1.23, "learning_rate": 6.894506947041246e-06, "loss": 0.0147, "step": 1970 }, { "epoch": 1.23, "learning_rate": 6.8849340970985365e-06, "loss": 0.016, "step": 1971 }, { "epoch": 1.23, "learning_rate": 6.875364407513223e-06, "loss": 0.0198, "step": 1972 }, { "epoch": 1.23, "learning_rate": 6.865797887994127e-06, "loss": 0.0229, "step": 1973 }, { "epoch": 1.23, "learning_rate": 6.856234548246866e-06, "loss": 0.0219, "step": 1974 }, { "epoch": 1.23, "learning_rate": 6.846674397973825e-06, "loss": 0.018, "step": 1975 }, { "epoch": 1.23, "learning_rate": 6.837117446874158e-06, "loss": 0.0197, "step": 1976 }, { "epoch": 1.23, "learning_rate": 6.82756370464376e-06, "loss": 0.0228, "step": 1977 }, { "epoch": 1.23, "learning_rate": 6.818013180975287e-06, "loss": 0.0275, "step": 1978 }, { "epoch": 1.23, "learning_rate": 6.808465885558122e-06, "loss": 0.0238, "step": 1979 }, { "epoch": 1.23, "learning_rate": 6.7989218280783686e-06, "loss": 0.0153, "step": 1980 }, { "epoch": 1.23, "learning_rate": 6.789381018218855e-06, "loss": 0.024, "step": 1981 }, { "epoch": 1.23, "learning_rate": 6.779843465659108e-06, "loss": 0.0195, "step": 1982 }, { "epoch": 1.23, "learning_rate": 6.77030918007535e-06, "loss": 0.0188, "step": 1983 }, { "epoch": 1.23, "learning_rate": 6.760778171140492e-06, "loss": 0.0185, "step": 1984 }, { "epoch": 1.23, "learning_rate": 6.751250448524123e-06, "loss": 0.0184, "step": 1985 }, { "epoch": 1.24, "learning_rate": 6.741726021892485e-06, "loss": 0.0232, "step": 1986 }, { "epoch": 1.24, "learning_rate": 6.732204900908489e-06, "loss": 0.0204, "step": 1987 }, { "epoch": 1.24, "learning_rate": 6.72268709523169e-06, "loss": 0.0222, "step": 1988 }, { "epoch": 1.24, "learning_rate": 6.713172614518278e-06, "loss": 0.0189, "step": 1989 }, { "epoch": 1.24, "learning_rate": 6.703661468421065e-06, "loss": 0.0305, "step": 1990 }, { "epoch": 1.24, "learning_rate": 6.694153666589489e-06, "loss": 0.0195, "step": 1991 }, { "epoch": 1.24, "learning_rate": 6.684649218669589e-06, "loss": 0.0239, "step": 1992 }, { "epoch": 1.24, "learning_rate": 6.675148134304001e-06, "loss": 0.02, "step": 1993 }, { "epoch": 1.24, "learning_rate": 6.665650423131953e-06, "loss": 0.0251, "step": 1994 }, { "epoch": 1.24, "learning_rate": 6.656156094789253e-06, "loss": 0.0269, "step": 1995 }, { "epoch": 1.24, "learning_rate": 6.646665158908257e-06, "loss": 0.0214, "step": 1996 }, { "epoch": 1.24, "learning_rate": 6.637177625117906e-06, "loss": 0.0164, "step": 1997 }, { "epoch": 1.24, "learning_rate": 6.627693503043681e-06, "loss": 0.0131, "step": 1998 }, { "epoch": 1.24, "learning_rate": 6.618212802307589e-06, "loss": 0.0211, "step": 1999 }, { "epoch": 1.24, "learning_rate": 6.608735532528178e-06, "loss": 0.0171, "step": 2000 }, { "epoch": 1.24, "learning_rate": 6.599261703320517e-06, "loss": 0.0211, "step": 2001 }, { "epoch": 1.25, "learning_rate": 6.589791324296173e-06, "loss": 0.0163, "step": 2002 }, { "epoch": 1.25, "learning_rate": 6.580324405063223e-06, "loss": 0.0244, "step": 2003 }, { "epoch": 1.25, "learning_rate": 6.570860955226234e-06, "loss": 0.0224, "step": 2004 }, { "epoch": 1.25, "learning_rate": 6.561400984386243e-06, "loss": 0.0211, "step": 2005 }, { "epoch": 1.25, "learning_rate": 6.551944502140768e-06, "loss": 0.0166, "step": 2006 }, { "epoch": 1.25, "learning_rate": 6.542491518083786e-06, "loss": 0.0215, "step": 2007 }, { "epoch": 1.25, "learning_rate": 6.5330420418057175e-06, "loss": 0.0239, "step": 2008 }, { "epoch": 1.25, "learning_rate": 6.5235960828934305e-06, "loss": 0.0162, "step": 2009 }, { "epoch": 1.25, "learning_rate": 6.514153650930228e-06, "loss": 0.0173, "step": 2010 }, { "epoch": 1.25, "learning_rate": 6.5047147554958234e-06, "loss": 0.0193, "step": 2011 }, { "epoch": 1.25, "learning_rate": 6.4952794061663525e-06, "loss": 0.0233, "step": 2012 }, { "epoch": 1.25, "learning_rate": 6.48584761251435e-06, "loss": 0.024, "step": 2013 }, { "epoch": 1.25, "learning_rate": 6.476419384108745e-06, "loss": 0.0204, "step": 2014 }, { "epoch": 1.25, "learning_rate": 6.466994730514842e-06, "loss": 0.012, "step": 2015 }, { "epoch": 1.25, "learning_rate": 6.457573661294329e-06, "loss": 0.015, "step": 2016 }, { "epoch": 1.25, "learning_rate": 6.448156186005255e-06, "loss": 0.0214, "step": 2017 }, { "epoch": 1.25, "learning_rate": 6.438742314202011e-06, "loss": 0.0206, "step": 2018 }, { "epoch": 1.26, "learning_rate": 6.429332055435349e-06, "loss": 0.0214, "step": 2019 }, { "epoch": 1.26, "learning_rate": 6.419925419252353e-06, "loss": 0.0178, "step": 2020 }, { "epoch": 1.26, "learning_rate": 6.410522415196416e-06, "loss": 0.0203, "step": 2021 }, { "epoch": 1.26, "learning_rate": 6.401123052807261e-06, "loss": 0.0164, "step": 2022 }, { "epoch": 1.26, "learning_rate": 6.391727341620917e-06, "loss": 0.0246, "step": 2023 }, { "epoch": 1.26, "learning_rate": 6.382335291169698e-06, "loss": 0.0199, "step": 2024 }, { "epoch": 1.26, "learning_rate": 6.3729469109822115e-06, "loss": 0.0166, "step": 2025 }, { "epoch": 1.26, "learning_rate": 6.363562210583345e-06, "loss": 0.0121, "step": 2026 }, { "epoch": 1.26, "learning_rate": 6.3541811994942405e-06, "loss": 0.0213, "step": 2027 }, { "epoch": 1.26, "learning_rate": 6.344803887232308e-06, "loss": 0.0135, "step": 2028 }, { "epoch": 1.26, "learning_rate": 6.335430283311206e-06, "loss": 0.0234, "step": 2029 }, { "epoch": 1.26, "learning_rate": 6.326060397240814e-06, "loss": 0.0181, "step": 2030 }, { "epoch": 1.26, "learning_rate": 6.3166942385272616e-06, "loss": 0.0258, "step": 2031 }, { "epoch": 1.26, "learning_rate": 6.3073318166728795e-06, "loss": 0.0237, "step": 2032 }, { "epoch": 1.26, "learning_rate": 6.297973141176224e-06, "loss": 0.0191, "step": 2033 }, { "epoch": 1.26, "learning_rate": 6.288618221532031e-06, "loss": 0.0199, "step": 2034 }, { "epoch": 1.27, "learning_rate": 6.279267067231242e-06, "loss": 0.0284, "step": 2035 }, { "epoch": 1.27, "learning_rate": 6.269919687760972e-06, "loss": 0.0109, "step": 2036 }, { "epoch": 1.27, "learning_rate": 6.260576092604502e-06, "loss": 0.0279, "step": 2037 }, { "epoch": 1.27, "learning_rate": 6.251236291241282e-06, "loss": 0.0227, "step": 2038 }, { "epoch": 1.27, "learning_rate": 6.241900293146915e-06, "loss": 0.0152, "step": 2039 }, { "epoch": 1.27, "learning_rate": 6.232568107793127e-06, "loss": 0.0211, "step": 2040 }, { "epoch": 1.27, "learning_rate": 6.223239744647797e-06, "loss": 0.0269, "step": 2041 }, { "epoch": 1.27, "learning_rate": 6.2139152131749216e-06, "loss": 0.0228, "step": 2042 }, { "epoch": 1.27, "learning_rate": 6.204594522834595e-06, "loss": 0.0244, "step": 2043 }, { "epoch": 1.27, "learning_rate": 6.195277683083033e-06, "loss": 0.0187, "step": 2044 }, { "epoch": 1.27, "learning_rate": 6.185964703372538e-06, "loss": 0.0142, "step": 2045 }, { "epoch": 1.27, "learning_rate": 6.1766555931514925e-06, "loss": 0.0144, "step": 2046 }, { "epoch": 1.27, "learning_rate": 6.167350361864357e-06, "loss": 0.0273, "step": 2047 }, { "epoch": 1.27, "learning_rate": 6.158049018951659e-06, "loss": 0.0252, "step": 2048 }, { "epoch": 1.27, "learning_rate": 6.148751573849976e-06, "loss": 0.0191, "step": 2049 }, { "epoch": 1.27, "learning_rate": 6.139458035991932e-06, "loss": 0.02, "step": 2050 }, { "epoch": 1.28, "learning_rate": 6.130168414806191e-06, "loss": 0.0172, "step": 2051 }, { "epoch": 1.28, "learning_rate": 6.1208827197174435e-06, "loss": 0.0303, "step": 2052 }, { "epoch": 1.28, "learning_rate": 6.111600960146385e-06, "loss": 0.0227, "step": 2053 }, { "epoch": 1.28, "learning_rate": 6.102323145509732e-06, "loss": 0.0219, "step": 2054 }, { "epoch": 1.28, "learning_rate": 6.093049285220198e-06, "loss": 0.0159, "step": 2055 }, { "epoch": 1.28, "learning_rate": 6.083779388686474e-06, "loss": 0.0159, "step": 2056 }, { "epoch": 1.28, "learning_rate": 6.074513465313238e-06, "loss": 0.0199, "step": 2057 }, { "epoch": 1.28, "learning_rate": 6.065251524501138e-06, "loss": 0.0194, "step": 2058 }, { "epoch": 1.28, "learning_rate": 6.055993575646775e-06, "loss": 0.0181, "step": 2059 }, { "epoch": 1.28, "learning_rate": 6.046739628142705e-06, "loss": 0.0133, "step": 2060 }, { "epoch": 1.28, "learning_rate": 6.037489691377427e-06, "loss": 0.0179, "step": 2061 }, { "epoch": 1.28, "learning_rate": 6.028243774735361e-06, "loss": 0.0154, "step": 2062 }, { "epoch": 1.28, "learning_rate": 6.019001887596856e-06, "loss": 0.0247, "step": 2063 }, { "epoch": 1.28, "learning_rate": 6.00976403933818e-06, "loss": 0.0224, "step": 2064 }, { "epoch": 1.28, "learning_rate": 6.000530239331481e-06, "loss": 0.018, "step": 2065 }, { "epoch": 1.28, "learning_rate": 5.991300496944823e-06, "loss": 0.0156, "step": 2066 }, { "epoch": 1.29, "learning_rate": 5.982074821542144e-06, "loss": 0.0136, "step": 2067 }, { "epoch": 1.29, "learning_rate": 5.972853222483253e-06, "loss": 0.0128, "step": 2068 }, { "epoch": 1.29, "learning_rate": 5.963635709123825e-06, "loss": 0.022, "step": 2069 }, { "epoch": 1.29, "learning_rate": 5.954422290815396e-06, "loss": 0.0168, "step": 2070 }, { "epoch": 1.29, "learning_rate": 5.945212976905341e-06, "loss": 0.014, "step": 2071 }, { "epoch": 1.29, "learning_rate": 5.936007776736872e-06, "loss": 0.0138, "step": 2072 }, { "epoch": 1.29, "learning_rate": 5.926806699649028e-06, "loss": 0.0173, "step": 2073 }, { "epoch": 1.29, "learning_rate": 5.91760975497667e-06, "loss": 0.0154, "step": 2074 }, { "epoch": 1.29, "learning_rate": 5.908416952050453e-06, "loss": 0.0209, "step": 2075 }, { "epoch": 1.29, "learning_rate": 5.899228300196839e-06, "loss": 0.0138, "step": 2076 }, { "epoch": 1.29, "learning_rate": 5.890043808738092e-06, "loss": 0.0137, "step": 2077 }, { "epoch": 1.29, "learning_rate": 5.880863486992225e-06, "loss": 0.0214, "step": 2078 }, { "epoch": 1.29, "learning_rate": 5.871687344273045e-06, "loss": 0.0171, "step": 2079 }, { "epoch": 1.29, "learning_rate": 5.862515389890111e-06, "loss": 0.0172, "step": 2080 }, { "epoch": 1.29, "learning_rate": 5.8533476331487295e-06, "loss": 0.0192, "step": 2081 }, { "epoch": 1.29, "learning_rate": 5.844184083349955e-06, "loss": 0.014, "step": 2082 }, { "epoch": 1.3, "learning_rate": 5.835024749790572e-06, "loss": 0.0173, "step": 2083 }, { "epoch": 1.3, "learning_rate": 5.8258696417630825e-06, "loss": 0.0191, "step": 2084 }, { "epoch": 1.3, "learning_rate": 5.8167187685557045e-06, "loss": 0.018, "step": 2085 }, { "epoch": 1.3, "learning_rate": 5.807572139452363e-06, "loss": 0.0207, "step": 2086 }, { "epoch": 1.3, "learning_rate": 5.798429763732673e-06, "loss": 0.0131, "step": 2087 }, { "epoch": 1.3, "learning_rate": 5.789291650671939e-06, "loss": 0.0243, "step": 2088 }, { "epoch": 1.3, "learning_rate": 5.780157809541134e-06, "loss": 0.0145, "step": 2089 }, { "epoch": 1.3, "learning_rate": 5.771028249606906e-06, "loss": 0.0174, "step": 2090 }, { "epoch": 1.3, "learning_rate": 5.761902980131546e-06, "loss": 0.0205, "step": 2091 }, { "epoch": 1.3, "learning_rate": 5.752782010373004e-06, "loss": 0.0115, "step": 2092 }, { "epoch": 1.3, "learning_rate": 5.743665349584868e-06, "loss": 0.0168, "step": 2093 }, { "epoch": 1.3, "learning_rate": 5.734553007016345e-06, "loss": 0.0162, "step": 2094 }, { "epoch": 1.3, "learning_rate": 5.725444991912266e-06, "loss": 0.0229, "step": 2095 }, { "epoch": 1.3, "learning_rate": 5.716341313513077e-06, "loss": 0.0182, "step": 2096 }, { "epoch": 1.3, "learning_rate": 5.707241981054814e-06, "loss": 0.0168, "step": 2097 }, { "epoch": 1.3, "learning_rate": 5.698147003769112e-06, "loss": 0.0223, "step": 2098 }, { "epoch": 1.31, "learning_rate": 5.68905639088319e-06, "loss": 0.0172, "step": 2099 }, { "epoch": 1.31, "learning_rate": 5.679970151619823e-06, "loss": 0.022, "step": 2100 }, { "epoch": 1.31, "learning_rate": 5.670888295197365e-06, "loss": 0.0251, "step": 2101 }, { "epoch": 1.31, "learning_rate": 5.661810830829724e-06, "loss": 0.0154, "step": 2102 }, { "epoch": 1.31, "learning_rate": 5.652737767726333e-06, "loss": 0.0174, "step": 2103 }, { "epoch": 1.31, "learning_rate": 5.643669115092183e-06, "loss": 0.0165, "step": 2104 }, { "epoch": 1.31, "learning_rate": 5.634604882127771e-06, "loss": 0.0127, "step": 2105 }, { "epoch": 1.31, "learning_rate": 5.625545078029135e-06, "loss": 0.0155, "step": 2106 }, { "epoch": 1.31, "learning_rate": 5.61648971198779e-06, "loss": 0.0279, "step": 2107 }, { "epoch": 1.31, "learning_rate": 5.607438793190767e-06, "loss": 0.0185, "step": 2108 }, { "epoch": 1.31, "learning_rate": 5.598392330820586e-06, "loss": 0.0107, "step": 2109 }, { "epoch": 1.31, "learning_rate": 5.58935033405523e-06, "loss": 0.0259, "step": 2110 }, { "epoch": 1.31, "learning_rate": 5.5803128120681684e-06, "loss": 0.0166, "step": 2111 }, { "epoch": 1.31, "learning_rate": 5.5712797740283265e-06, "loss": 0.0176, "step": 2112 }, { "epoch": 1.31, "learning_rate": 5.562251229100066e-06, "loss": 0.0197, "step": 2113 }, { "epoch": 1.31, "learning_rate": 5.553227186443215e-06, "loss": 0.0154, "step": 2114 }, { "epoch": 1.32, "learning_rate": 5.544207655213021e-06, "loss": 0.0227, "step": 2115 }, { "epoch": 1.32, "learning_rate": 5.535192644560148e-06, "loss": 0.0175, "step": 2116 }, { "epoch": 1.32, "learning_rate": 5.526182163630682e-06, "loss": 0.0195, "step": 2117 }, { "epoch": 1.32, "learning_rate": 5.5171762215661165e-06, "loss": 0.0157, "step": 2118 }, { "epoch": 1.32, "learning_rate": 5.508174827503328e-06, "loss": 0.0223, "step": 2119 }, { "epoch": 1.32, "learning_rate": 5.499177990574591e-06, "loss": 0.0158, "step": 2120 }, { "epoch": 1.32, "learning_rate": 5.49018571990755e-06, "loss": 0.0259, "step": 2121 }, { "epoch": 1.32, "learning_rate": 5.48119802462522e-06, "loss": 0.0155, "step": 2122 }, { "epoch": 1.32, "learning_rate": 5.472214913845973e-06, "loss": 0.0205, "step": 2123 }, { "epoch": 1.32, "learning_rate": 5.46323639668353e-06, "loss": 0.0136, "step": 2124 }, { "epoch": 1.32, "learning_rate": 5.454262482246954e-06, "loss": 0.0226, "step": 2125 }, { "epoch": 1.32, "learning_rate": 5.445293179640628e-06, "loss": 0.0194, "step": 2126 }, { "epoch": 1.32, "learning_rate": 5.4363284979642685e-06, "loss": 0.0153, "step": 2127 }, { "epoch": 1.32, "learning_rate": 5.427368446312902e-06, "loss": 0.0173, "step": 2128 }, { "epoch": 1.32, "learning_rate": 5.4184130337768485e-06, "loss": 0.015, "step": 2129 }, { "epoch": 1.32, "learning_rate": 5.40946226944173e-06, "loss": 0.0186, "step": 2130 }, { "epoch": 1.33, "learning_rate": 5.400516162388452e-06, "loss": 0.0128, "step": 2131 }, { "epoch": 1.33, "learning_rate": 5.391574721693192e-06, "loss": 0.0194, "step": 2132 }, { "epoch": 1.33, "learning_rate": 5.382637956427394e-06, "loss": 0.0162, "step": 2133 }, { "epoch": 1.33, "learning_rate": 5.373705875657766e-06, "loss": 0.0123, "step": 2134 }, { "epoch": 1.33, "learning_rate": 5.364778488446247e-06, "loss": 0.0173, "step": 2135 }, { "epoch": 1.33, "learning_rate": 5.3558558038500295e-06, "loss": 0.013, "step": 2136 }, { "epoch": 1.33, "learning_rate": 5.346937830921531e-06, "loss": 0.023, "step": 2137 }, { "epoch": 1.33, "learning_rate": 5.338024578708381e-06, "loss": 0.0192, "step": 2138 }, { "epoch": 1.33, "learning_rate": 5.329116056253429e-06, "loss": 0.0112, "step": 2139 }, { "epoch": 1.33, "learning_rate": 5.320212272594725e-06, "loss": 0.0153, "step": 2140 }, { "epoch": 1.33, "learning_rate": 5.3113132367655065e-06, "loss": 0.0247, "step": 2141 }, { "epoch": 1.33, "learning_rate": 5.302418957794198e-06, "loss": 0.0224, "step": 2142 }, { "epoch": 1.33, "learning_rate": 5.293529444704395e-06, "loss": 0.0211, "step": 2143 }, { "epoch": 1.33, "learning_rate": 5.284644706514868e-06, "loss": 0.0212, "step": 2144 }, { "epoch": 1.33, "learning_rate": 5.275764752239523e-06, "loss": 0.0217, "step": 2145 }, { "epoch": 1.33, "learning_rate": 5.266889590887429e-06, "loss": 0.0186, "step": 2146 }, { "epoch": 1.34, "learning_rate": 5.258019231462792e-06, "loss": 0.0158, "step": 2147 }, { "epoch": 1.34, "learning_rate": 5.249153682964935e-06, "loss": 0.0172, "step": 2148 }, { "epoch": 1.34, "learning_rate": 5.240292954388306e-06, "loss": 0.0167, "step": 2149 }, { "epoch": 1.34, "learning_rate": 5.231437054722477e-06, "loss": 0.0267, "step": 2150 }, { "epoch": 1.34, "learning_rate": 5.222585992952094e-06, "loss": 0.0228, "step": 2151 }, { "epoch": 1.34, "learning_rate": 5.213739778056915e-06, "loss": 0.021, "step": 2152 }, { "epoch": 1.34, "learning_rate": 5.204898419011778e-06, "loss": 0.0136, "step": 2153 }, { "epoch": 1.34, "learning_rate": 5.1960619247865815e-06, "loss": 0.0173, "step": 2154 }, { "epoch": 1.34, "learning_rate": 5.187230304346304e-06, "loss": 0.027, "step": 2155 }, { "epoch": 1.34, "learning_rate": 5.178403566650976e-06, "loss": 0.0212, "step": 2156 }, { "epoch": 1.34, "learning_rate": 5.169581720655661e-06, "loss": 0.0171, "step": 2157 }, { "epoch": 1.34, "learning_rate": 5.160764775310479e-06, "loss": 0.0184, "step": 2158 }, { "epoch": 1.34, "learning_rate": 5.15195273956057e-06, "loss": 0.0199, "step": 2159 }, { "epoch": 1.34, "learning_rate": 5.143145622346089e-06, "loss": 0.0164, "step": 2160 }, { "epoch": 1.34, "learning_rate": 5.1343434326022024e-06, "loss": 0.0171, "step": 2161 }, { "epoch": 1.34, "learning_rate": 5.125546179259082e-06, "loss": 0.0181, "step": 2162 }, { "epoch": 1.35, "learning_rate": 5.116753871241893e-06, "loss": 0.0172, "step": 2163 }, { "epoch": 1.35, "learning_rate": 5.107966517470771e-06, "loss": 0.014, "step": 2164 }, { "epoch": 1.35, "learning_rate": 5.099184126860837e-06, "loss": 0.0177, "step": 2165 }, { "epoch": 1.35, "learning_rate": 5.090406708322173e-06, "loss": 0.0118, "step": 2166 }, { "epoch": 1.35, "learning_rate": 5.081634270759818e-06, "loss": 0.0119, "step": 2167 }, { "epoch": 1.35, "learning_rate": 5.072866823073754e-06, "loss": 0.0212, "step": 2168 }, { "epoch": 1.35, "learning_rate": 5.064104374158909e-06, "loss": 0.0114, "step": 2169 }, { "epoch": 1.35, "learning_rate": 5.055346932905125e-06, "loss": 0.018, "step": 2170 }, { "epoch": 1.35, "learning_rate": 5.046594508197177e-06, "loss": 0.0129, "step": 2171 }, { "epoch": 1.35, "learning_rate": 5.037847108914748e-06, "loss": 0.0162, "step": 2172 }, { "epoch": 1.35, "learning_rate": 5.029104743932411e-06, "loss": 0.0177, "step": 2173 }, { "epoch": 1.35, "learning_rate": 5.0203674221196485e-06, "loss": 0.0138, "step": 2174 }, { "epoch": 1.35, "learning_rate": 5.011635152340816e-06, "loss": 0.0204, "step": 2175 }, { "epoch": 1.35, "learning_rate": 5.002907943455145e-06, "loss": 0.0171, "step": 2176 }, { "epoch": 1.35, "learning_rate": 4.994185804316735e-06, "loss": 0.0135, "step": 2177 }, { "epoch": 1.35, "learning_rate": 4.985468743774546e-06, "loss": 0.0139, "step": 2178 }, { "epoch": 1.36, "learning_rate": 4.9767567706723706e-06, "loss": 0.0188, "step": 2179 }, { "epoch": 1.36, "learning_rate": 4.968049893848854e-06, "loss": 0.0255, "step": 2180 }, { "epoch": 1.36, "learning_rate": 4.959348122137466e-06, "loss": 0.0127, "step": 2181 }, { "epoch": 1.36, "learning_rate": 4.950651464366502e-06, "loss": 0.0112, "step": 2182 }, { "epoch": 1.36, "learning_rate": 4.941959929359057e-06, "loss": 0.0188, "step": 2183 }, { "epoch": 1.36, "learning_rate": 4.933273525933041e-06, "loss": 0.0228, "step": 2184 }, { "epoch": 1.36, "learning_rate": 4.924592262901154e-06, "loss": 0.0161, "step": 2185 }, { "epoch": 1.36, "learning_rate": 4.915916149070877e-06, "loss": 0.0219, "step": 2186 }, { "epoch": 1.36, "learning_rate": 4.907245193244473e-06, "loss": 0.0189, "step": 2187 }, { "epoch": 1.36, "learning_rate": 4.898579404218973e-06, "loss": 0.0139, "step": 2188 }, { "epoch": 1.36, "learning_rate": 4.889918790786153e-06, "loss": 0.0191, "step": 2189 }, { "epoch": 1.36, "learning_rate": 4.881263361732552e-06, "loss": 0.0145, "step": 2190 }, { "epoch": 1.36, "learning_rate": 4.872613125839449e-06, "loss": 0.0146, "step": 2191 }, { "epoch": 1.36, "learning_rate": 4.863968091882843e-06, "loss": 0.012, "step": 2192 }, { "epoch": 1.36, "learning_rate": 4.855328268633463e-06, "loss": 0.0117, "step": 2193 }, { "epoch": 1.36, "learning_rate": 4.846693664856754e-06, "loss": 0.0169, "step": 2194 }, { "epoch": 1.37, "learning_rate": 4.838064289312862e-06, "loss": 0.0181, "step": 2195 }, { "epoch": 1.37, "learning_rate": 4.829440150756628e-06, "loss": 0.0224, "step": 2196 }, { "epoch": 1.37, "learning_rate": 4.820821257937587e-06, "loss": 0.0185, "step": 2197 }, { "epoch": 1.37, "learning_rate": 4.8122076195999366e-06, "loss": 0.0229, "step": 2198 }, { "epoch": 1.37, "learning_rate": 4.803599244482558e-06, "loss": 0.008, "step": 2199 }, { "epoch": 1.37, "learning_rate": 4.7949961413189885e-06, "loss": 0.0145, "step": 2200 }, { "epoch": 1.37, "learning_rate": 4.786398318837417e-06, "loss": 0.0158, "step": 2201 }, { "epoch": 1.37, "learning_rate": 4.7778057857606664e-06, "loss": 0.0096, "step": 2202 }, { "epoch": 1.37, "learning_rate": 4.769218550806209e-06, "loss": 0.0169, "step": 2203 }, { "epoch": 1.37, "learning_rate": 4.760636622686136e-06, "loss": 0.0192, "step": 2204 }, { "epoch": 1.37, "learning_rate": 4.752060010107145e-06, "loss": 0.0138, "step": 2205 }, { "epoch": 1.37, "learning_rate": 4.74348872177055e-06, "loss": 0.0232, "step": 2206 }, { "epoch": 1.37, "learning_rate": 4.734922766372267e-06, "loss": 0.0197, "step": 2207 }, { "epoch": 1.37, "learning_rate": 4.726362152602787e-06, "loss": 0.0128, "step": 2208 }, { "epoch": 1.37, "learning_rate": 4.717806889147196e-06, "loss": 0.0141, "step": 2209 }, { "epoch": 1.37, "learning_rate": 4.709256984685147e-06, "loss": 0.0185, "step": 2210 }, { "epoch": 1.38, "learning_rate": 4.700712447890853e-06, "loss": 0.0151, "step": 2211 }, { "epoch": 1.38, "learning_rate": 4.692173287433087e-06, "loss": 0.0173, "step": 2212 }, { "epoch": 1.38, "learning_rate": 4.683639511975167e-06, "loss": 0.0166, "step": 2213 }, { "epoch": 1.38, "learning_rate": 4.675111130174939e-06, "loss": 0.0247, "step": 2214 }, { "epoch": 1.38, "learning_rate": 4.666588150684786e-06, "loss": 0.0267, "step": 2215 }, { "epoch": 1.38, "learning_rate": 4.65807058215161e-06, "loss": 0.0222, "step": 2216 }, { "epoch": 1.38, "learning_rate": 4.6495584332168165e-06, "loss": 0.0235, "step": 2217 }, { "epoch": 1.38, "learning_rate": 4.641051712516315e-06, "loss": 0.0147, "step": 2218 }, { "epoch": 1.38, "learning_rate": 4.632550428680515e-06, "loss": 0.0164, "step": 2219 }, { "epoch": 1.38, "learning_rate": 4.624054590334304e-06, "loss": 0.0149, "step": 2220 }, { "epoch": 1.38, "learning_rate": 4.6155642060970415e-06, "loss": 0.0146, "step": 2221 }, { "epoch": 1.38, "learning_rate": 4.607079284582563e-06, "loss": 0.0197, "step": 2222 }, { "epoch": 1.38, "learning_rate": 4.598599834399155e-06, "loss": 0.016, "step": 2223 }, { "epoch": 1.38, "learning_rate": 4.590125864149551e-06, "loss": 0.0151, "step": 2224 }, { "epoch": 1.38, "learning_rate": 4.581657382430932e-06, "loss": 0.016, "step": 2225 }, { "epoch": 1.38, "learning_rate": 4.573194397834909e-06, "loss": 0.0206, "step": 2226 }, { "epoch": 1.38, "learning_rate": 4.564736918947509e-06, "loss": 0.0192, "step": 2227 }, { "epoch": 1.39, "learning_rate": 4.556284954349179e-06, "loss": 0.0144, "step": 2228 }, { "epoch": 1.39, "learning_rate": 4.547838512614773e-06, "loss": 0.0149, "step": 2229 }, { "epoch": 1.39, "learning_rate": 4.53939760231354e-06, "loss": 0.0195, "step": 2230 }, { "epoch": 1.39, "learning_rate": 4.530962232009116e-06, "loss": 0.0145, "step": 2231 }, { "epoch": 1.39, "learning_rate": 4.5225324102595215e-06, "loss": 0.0148, "step": 2232 }, { "epoch": 1.39, "learning_rate": 4.514108145617135e-06, "loss": 0.0124, "step": 2233 }, { "epoch": 1.39, "learning_rate": 4.505689446628712e-06, "loss": 0.0125, "step": 2234 }, { "epoch": 1.39, "learning_rate": 4.497276321835357e-06, "loss": 0.0173, "step": 2235 }, { "epoch": 1.39, "learning_rate": 4.488868779772509e-06, "loss": 0.0149, "step": 2236 }, { "epoch": 1.39, "learning_rate": 4.480466828969958e-06, "loss": 0.0152, "step": 2237 }, { "epoch": 1.39, "learning_rate": 4.472070477951809e-06, "loss": 0.0146, "step": 2238 }, { "epoch": 1.39, "learning_rate": 4.4636797352365035e-06, "loss": 0.0194, "step": 2239 }, { "epoch": 1.39, "learning_rate": 4.455294609336771e-06, "loss": 0.0225, "step": 2240 }, { "epoch": 1.39, "learning_rate": 4.4469151087596555e-06, "loss": 0.01, "step": 2241 }, { "epoch": 1.39, "learning_rate": 4.438541242006495e-06, "loss": 0.0135, "step": 2242 }, { "epoch": 1.39, "learning_rate": 4.430173017572902e-06, "loss": 0.0164, "step": 2243 }, { "epoch": 1.4, "learning_rate": 4.421810443948774e-06, "loss": 0.0131, "step": 2244 }, { "epoch": 1.4, "learning_rate": 4.413453529618274e-06, "loss": 0.0128, "step": 2245 }, { "epoch": 1.4, "learning_rate": 4.4051022830598135e-06, "loss": 0.0128, "step": 2246 }, { "epoch": 1.4, "learning_rate": 4.3967567127460705e-06, "loss": 0.0109, "step": 2247 }, { "epoch": 1.4, "learning_rate": 4.388416827143958e-06, "loss": 0.0154, "step": 2248 }, { "epoch": 1.4, "learning_rate": 4.38008263471461e-06, "loss": 0.0118, "step": 2249 }, { "epoch": 1.4, "learning_rate": 4.371754143913398e-06, "loss": 0.0128, "step": 2250 }, { "epoch": 1.4, "learning_rate": 4.36343136318991e-06, "loss": 0.0117, "step": 2251 }, { "epoch": 1.4, "learning_rate": 4.355114300987926e-06, "loss": 0.0264, "step": 2252 }, { "epoch": 1.4, "learning_rate": 4.34680296574544e-06, "loss": 0.0166, "step": 2253 }, { "epoch": 1.4, "learning_rate": 4.338497365894628e-06, "loss": 0.016, "step": 2254 }, { "epoch": 1.4, "learning_rate": 4.330197509861851e-06, "loss": 0.0162, "step": 2255 }, { "epoch": 1.4, "learning_rate": 4.32190340606764e-06, "loss": 0.0159, "step": 2256 }, { "epoch": 1.4, "learning_rate": 4.313615062926691e-06, "loss": 0.0173, "step": 2257 }, { "epoch": 1.4, "learning_rate": 4.305332488847857e-06, "loss": 0.0159, "step": 2258 }, { "epoch": 1.4, "learning_rate": 4.297055692234133e-06, "loss": 0.0178, "step": 2259 }, { "epoch": 1.41, "learning_rate": 4.288784681482657e-06, "loss": 0.0168, "step": 2260 }, { "epoch": 1.41, "learning_rate": 4.2805194649847014e-06, "loss": 0.0138, "step": 2261 }, { "epoch": 1.41, "learning_rate": 4.272260051125646e-06, "loss": 0.011, "step": 2262 }, { "epoch": 1.41, "learning_rate": 4.264006448284996e-06, "loss": 0.0105, "step": 2263 }, { "epoch": 1.41, "learning_rate": 4.25575866483636e-06, "loss": 0.0167, "step": 2264 }, { "epoch": 1.41, "learning_rate": 4.247516709147437e-06, "loss": 0.0196, "step": 2265 }, { "epoch": 1.41, "learning_rate": 4.2392805895800185e-06, "loss": 0.0242, "step": 2266 }, { "epoch": 1.41, "learning_rate": 4.231050314489975e-06, "loss": 0.015, "step": 2267 }, { "epoch": 1.41, "learning_rate": 4.222825892227239e-06, "loss": 0.0159, "step": 2268 }, { "epoch": 1.41, "learning_rate": 4.214607331135817e-06, "loss": 0.0283, "step": 2269 }, { "epoch": 1.41, "learning_rate": 4.206394639553766e-06, "loss": 0.0184, "step": 2270 }, { "epoch": 1.41, "learning_rate": 4.198187825813179e-06, "loss": 0.0162, "step": 2271 }, { "epoch": 1.41, "learning_rate": 4.189986898240196e-06, "loss": 0.0157, "step": 2272 }, { "epoch": 1.41, "learning_rate": 4.181791865154982e-06, "loss": 0.0131, "step": 2273 }, { "epoch": 1.41, "learning_rate": 4.173602734871723e-06, "loss": 0.0189, "step": 2274 }, { "epoch": 1.41, "learning_rate": 4.165419515698613e-06, "loss": 0.0172, "step": 2275 }, { "epoch": 1.42, "learning_rate": 4.157242215937854e-06, "loss": 0.0136, "step": 2276 }, { "epoch": 1.42, "learning_rate": 4.149070843885639e-06, "loss": 0.0181, "step": 2277 }, { "epoch": 1.42, "learning_rate": 4.140905407832144e-06, "loss": 0.0122, "step": 2278 }, { "epoch": 1.42, "learning_rate": 4.132745916061528e-06, "loss": 0.0206, "step": 2279 }, { "epoch": 1.42, "learning_rate": 4.1245923768519235e-06, "loss": 0.0223, "step": 2280 }, { "epoch": 1.42, "learning_rate": 4.1164447984754084e-06, "loss": 0.0176, "step": 2281 }, { "epoch": 1.42, "learning_rate": 4.108303189198024e-06, "loss": 0.0144, "step": 2282 }, { "epoch": 1.42, "learning_rate": 4.100167557279765e-06, "loss": 0.0141, "step": 2283 }, { "epoch": 1.42, "learning_rate": 4.09203791097454e-06, "loss": 0.0178, "step": 2284 }, { "epoch": 1.42, "learning_rate": 4.083914258530202e-06, "loss": 0.021, "step": 2285 }, { "epoch": 1.42, "learning_rate": 4.075796608188518e-06, "loss": 0.0133, "step": 2286 }, { "epoch": 1.42, "learning_rate": 4.0676849681851594e-06, "loss": 0.0106, "step": 2287 }, { "epoch": 1.42, "learning_rate": 4.059579346749709e-06, "loss": 0.0121, "step": 2288 }, { "epoch": 1.42, "learning_rate": 4.051479752105642e-06, "loss": 0.0123, "step": 2289 }, { "epoch": 1.42, "learning_rate": 4.043386192470309e-06, "loss": 0.0178, "step": 2290 }, { "epoch": 1.42, "learning_rate": 4.035298676054955e-06, "loss": 0.0176, "step": 2291 }, { "epoch": 1.43, "learning_rate": 4.02721721106468e-06, "loss": 0.0227, "step": 2292 }, { "epoch": 1.43, "learning_rate": 4.019141805698456e-06, "loss": 0.0241, "step": 2293 }, { "epoch": 1.43, "learning_rate": 4.01107246814909e-06, "loss": 0.0227, "step": 2294 }, { "epoch": 1.43, "learning_rate": 4.00300920660325e-06, "loss": 0.0139, "step": 2295 }, { "epoch": 1.43, "learning_rate": 3.994952029241436e-06, "loss": 0.0214, "step": 2296 }, { "epoch": 1.43, "learning_rate": 3.986900944237966e-06, "loss": 0.0227, "step": 2297 }, { "epoch": 1.43, "learning_rate": 3.9788559597609856e-06, "loss": 0.0149, "step": 2298 }, { "epoch": 1.43, "learning_rate": 3.970817083972451e-06, "loss": 0.0108, "step": 2299 }, { "epoch": 1.43, "learning_rate": 3.962784325028119e-06, "loss": 0.0152, "step": 2300 }, { "epoch": 1.43, "learning_rate": 3.9547576910775405e-06, "loss": 0.0128, "step": 2301 }, { "epoch": 1.43, "learning_rate": 3.946737190264057e-06, "loss": 0.0132, "step": 2302 }, { "epoch": 1.43, "learning_rate": 3.938722830724775e-06, "loss": 0.0144, "step": 2303 }, { "epoch": 1.43, "learning_rate": 3.930714620590582e-06, "loss": 0.0171, "step": 2304 }, { "epoch": 1.43, "learning_rate": 3.9227125679861286e-06, "loss": 0.0179, "step": 2305 }, { "epoch": 1.43, "learning_rate": 3.914716681029805e-06, "loss": 0.0117, "step": 2306 }, { "epoch": 1.43, "learning_rate": 3.906726967833759e-06, "loss": 0.0132, "step": 2307 }, { "epoch": 1.44, "learning_rate": 3.898743436503869e-06, "loss": 0.0205, "step": 2308 }, { "epoch": 1.44, "learning_rate": 3.890766095139744e-06, "loss": 0.0124, "step": 2309 }, { "epoch": 1.44, "learning_rate": 3.882794951834711e-06, "loss": 0.0147, "step": 2310 }, { "epoch": 1.44, "learning_rate": 3.87483001467581e-06, "loss": 0.0182, "step": 2311 }, { "epoch": 1.44, "learning_rate": 3.866871291743787e-06, "loss": 0.0131, "step": 2312 }, { "epoch": 1.44, "learning_rate": 3.858918791113075e-06, "loss": 0.0205, "step": 2313 }, { "epoch": 1.44, "learning_rate": 3.850972520851804e-06, "loss": 0.0139, "step": 2314 }, { "epoch": 1.44, "learning_rate": 3.8430324890217805e-06, "loss": 0.0134, "step": 2315 }, { "epoch": 1.44, "learning_rate": 3.835098703678475e-06, "loss": 0.019, "step": 2316 }, { "epoch": 1.44, "learning_rate": 3.827171172871029e-06, "loss": 0.0221, "step": 2317 }, { "epoch": 1.44, "learning_rate": 3.819249904642235e-06, "loss": 0.0152, "step": 2318 }, { "epoch": 1.44, "learning_rate": 3.8113349070285344e-06, "loss": 0.0185, "step": 2319 }, { "epoch": 1.44, "learning_rate": 3.8034261880600034e-06, "loss": 0.0199, "step": 2320 }, { "epoch": 1.44, "learning_rate": 3.795523755760353e-06, "loss": 0.0161, "step": 2321 }, { "epoch": 1.44, "learning_rate": 3.7876276181469064e-06, "loss": 0.0122, "step": 2322 }, { "epoch": 1.44, "learning_rate": 3.77973778323061e-06, "loss": 0.0192, "step": 2323 }, { "epoch": 1.45, "learning_rate": 3.771854259016019e-06, "loss": 0.0157, "step": 2324 }, { "epoch": 1.45, "learning_rate": 3.76397705350127e-06, "loss": 0.0166, "step": 2325 }, { "epoch": 1.45, "learning_rate": 3.7561061746781004e-06, "loss": 0.019, "step": 2326 }, { "epoch": 1.45, "learning_rate": 3.748241630531838e-06, "loss": 0.0104, "step": 2327 }, { "epoch": 1.45, "learning_rate": 3.740383429041362e-06, "loss": 0.0132, "step": 2328 }, { "epoch": 1.45, "learning_rate": 3.7325315781791337e-06, "loss": 0.0155, "step": 2329 }, { "epoch": 1.45, "learning_rate": 3.7246860859111655e-06, "loss": 0.0138, "step": 2330 }, { "epoch": 1.45, "learning_rate": 3.7168469601970214e-06, "loss": 0.0273, "step": 2331 }, { "epoch": 1.45, "learning_rate": 3.7090142089897972e-06, "loss": 0.0186, "step": 2332 }, { "epoch": 1.45, "learning_rate": 3.7011878402361335e-06, "loss": 0.0211, "step": 2333 }, { "epoch": 1.45, "learning_rate": 3.693367861876188e-06, "loss": 0.0176, "step": 2334 }, { "epoch": 1.45, "learning_rate": 3.6855542818436397e-06, "loss": 0.0163, "step": 2335 }, { "epoch": 1.45, "learning_rate": 3.6777471080656744e-06, "loss": 0.0172, "step": 2336 }, { "epoch": 1.45, "learning_rate": 3.6699463484629795e-06, "loss": 0.0168, "step": 2337 }, { "epoch": 1.45, "learning_rate": 3.662152010949729e-06, "loss": 0.0158, "step": 2338 }, { "epoch": 1.45, "learning_rate": 3.6543641034335873e-06, "loss": 0.0204, "step": 2339 }, { "epoch": 1.46, "learning_rate": 3.6465826338157007e-06, "loss": 0.013, "step": 2340 }, { "epoch": 1.46, "learning_rate": 3.638807609990668e-06, "loss": 0.0156, "step": 2341 }, { "epoch": 1.46, "learning_rate": 3.6310390398465623e-06, "loss": 0.0123, "step": 2342 }, { "epoch": 1.46, "learning_rate": 3.6232769312649053e-06, "loss": 0.0128, "step": 2343 }, { "epoch": 1.46, "learning_rate": 3.615521292120663e-06, "loss": 0.0167, "step": 2344 }, { "epoch": 1.46, "learning_rate": 3.6077721302822355e-06, "loss": 0.0188, "step": 2345 }, { "epoch": 1.46, "learning_rate": 3.600029453611459e-06, "loss": 0.0185, "step": 2346 }, { "epoch": 1.46, "learning_rate": 3.592293269963577e-06, "loss": 0.0137, "step": 2347 }, { "epoch": 1.46, "learning_rate": 3.5845635871872575e-06, "loss": 0.0159, "step": 2348 }, { "epoch": 1.46, "learning_rate": 3.5768404131245695e-06, "loss": 0.0133, "step": 2349 }, { "epoch": 1.46, "learning_rate": 3.5691237556109794e-06, "loss": 0.0149, "step": 2350 }, { "epoch": 1.46, "learning_rate": 3.561413622475336e-06, "loss": 0.0139, "step": 2351 }, { "epoch": 1.46, "learning_rate": 3.5537100215398766e-06, "loss": 0.0187, "step": 2352 }, { "epoch": 1.46, "learning_rate": 3.5460129606202086e-06, "loss": 0.0168, "step": 2353 }, { "epoch": 1.46, "learning_rate": 3.5383224475253043e-06, "loss": 0.0234, "step": 2354 }, { "epoch": 1.46, "learning_rate": 3.5306384900574943e-06, "loss": 0.0182, "step": 2355 }, { "epoch": 1.47, "learning_rate": 3.5229610960124585e-06, "loss": 0.0209, "step": 2356 }, { "epoch": 1.47, "learning_rate": 3.5152902731792104e-06, "loss": 0.014, "step": 2357 }, { "epoch": 1.47, "learning_rate": 3.5076260293401053e-06, "loss": 0.0171, "step": 2358 }, { "epoch": 1.47, "learning_rate": 3.4999683722708265e-06, "loss": 0.0203, "step": 2359 }, { "epoch": 1.47, "learning_rate": 3.492317309740362e-06, "loss": 0.014, "step": 2360 }, { "epoch": 1.47, "learning_rate": 3.484672849511019e-06, "loss": 0.0227, "step": 2361 }, { "epoch": 1.47, "learning_rate": 3.4770349993384057e-06, "loss": 0.0167, "step": 2362 }, { "epoch": 1.47, "learning_rate": 3.469403766971423e-06, "loss": 0.015, "step": 2363 }, { "epoch": 1.47, "learning_rate": 3.4617791601522565e-06, "loss": 0.0183, "step": 2364 }, { "epoch": 1.47, "learning_rate": 3.4541611866163748e-06, "loss": 0.0116, "step": 2365 }, { "epoch": 1.47, "learning_rate": 3.446549854092506e-06, "loss": 0.0182, "step": 2366 }, { "epoch": 1.47, "learning_rate": 3.438945170302651e-06, "loss": 0.0219, "step": 2367 }, { "epoch": 1.47, "learning_rate": 3.4313471429620628e-06, "loss": 0.0121, "step": 2368 }, { "epoch": 1.47, "learning_rate": 3.423755779779243e-06, "loss": 0.0197, "step": 2369 }, { "epoch": 1.47, "learning_rate": 3.4161710884559186e-06, "loss": 0.0146, "step": 2370 }, { "epoch": 1.47, "learning_rate": 3.408593076687069e-06, "loss": 0.0131, "step": 2371 }, { "epoch": 1.48, "learning_rate": 3.401021752160887e-06, "loss": 0.0151, "step": 2372 }, { "epoch": 1.48, "learning_rate": 3.393457122558772e-06, "loss": 0.0233, "step": 2373 }, { "epoch": 1.48, "learning_rate": 3.3858991955553455e-06, "loss": 0.0104, "step": 2374 }, { "epoch": 1.48, "learning_rate": 3.378347978818425e-06, "loss": 0.0163, "step": 2375 }, { "epoch": 1.48, "learning_rate": 3.3708034800090106e-06, "loss": 0.0143, "step": 2376 }, { "epoch": 1.48, "learning_rate": 3.363265706781299e-06, "loss": 0.0188, "step": 2377 }, { "epoch": 1.48, "learning_rate": 3.3557346667826573e-06, "loss": 0.0183, "step": 2378 }, { "epoch": 1.48, "learning_rate": 3.348210367653625e-06, "loss": 0.0126, "step": 2379 }, { "epoch": 1.48, "learning_rate": 3.3406928170278997e-06, "loss": 0.0212, "step": 2380 }, { "epoch": 1.48, "learning_rate": 3.3331820225323364e-06, "loss": 0.0112, "step": 2381 }, { "epoch": 1.48, "learning_rate": 3.3256779917869263e-06, "loss": 0.0134, "step": 2382 }, { "epoch": 1.48, "learning_rate": 3.3181807324048087e-06, "loss": 0.0145, "step": 2383 }, { "epoch": 1.48, "learning_rate": 3.3106902519922523e-06, "loss": 0.0157, "step": 2384 }, { "epoch": 1.48, "learning_rate": 3.3032065581486393e-06, "loss": 0.0157, "step": 2385 }, { "epoch": 1.48, "learning_rate": 3.2957296584664757e-06, "loss": 0.0171, "step": 2386 }, { "epoch": 1.48, "learning_rate": 3.2882595605313715e-06, "loss": 0.0208, "step": 2387 }, { "epoch": 1.49, "learning_rate": 3.2807962719220366e-06, "loss": 0.0143, "step": 2388 }, { "epoch": 1.49, "learning_rate": 3.27333980021027e-06, "loss": 0.02, "step": 2389 }, { "epoch": 1.49, "learning_rate": 3.26589015296096e-06, "loss": 0.0176, "step": 2390 }, { "epoch": 1.49, "learning_rate": 3.258447337732068e-06, "loss": 0.0179, "step": 2391 }, { "epoch": 1.49, "learning_rate": 3.2510113620746196e-06, "loss": 0.011, "step": 2392 }, { "epoch": 1.49, "learning_rate": 3.2435822335327096e-06, "loss": 0.013, "step": 2393 }, { "epoch": 1.49, "learning_rate": 3.236159959643482e-06, "loss": 0.0164, "step": 2394 }, { "epoch": 1.49, "learning_rate": 3.228744547937125e-06, "loss": 0.0136, "step": 2395 }, { "epoch": 1.49, "learning_rate": 3.221336005936867e-06, "loss": 0.0102, "step": 2396 }, { "epoch": 1.49, "learning_rate": 3.2139343411589663e-06, "loss": 0.018, "step": 2397 }, { "epoch": 1.49, "learning_rate": 3.2065395611127057e-06, "loss": 0.012, "step": 2398 }, { "epoch": 1.49, "learning_rate": 3.1991516733003813e-06, "loss": 0.0127, "step": 2399 }, { "epoch": 1.49, "learning_rate": 3.1917706852173003e-06, "loss": 0.0192, "step": 2400 }, { "epoch": 1.49, "learning_rate": 3.184396604351759e-06, "loss": 0.0198, "step": 2401 }, { "epoch": 1.49, "learning_rate": 3.1770294381850573e-06, "loss": 0.0172, "step": 2402 }, { "epoch": 1.49, "learning_rate": 3.1696691941914813e-06, "loss": 0.0209, "step": 2403 }, { "epoch": 1.5, "learning_rate": 3.1623158798382813e-06, "loss": 0.0175, "step": 2404 }, { "epoch": 1.5, "learning_rate": 3.1549695025856873e-06, "loss": 0.0121, "step": 2405 }, { "epoch": 1.5, "learning_rate": 3.147630069886891e-06, "loss": 0.0158, "step": 2406 }, { "epoch": 1.5, "learning_rate": 3.140297589188036e-06, "loss": 0.0121, "step": 2407 }, { "epoch": 1.5, "learning_rate": 3.132972067928215e-06, "loss": 0.0195, "step": 2408 }, { "epoch": 1.5, "learning_rate": 3.125653513539456e-06, "loss": 0.0122, "step": 2409 }, { "epoch": 1.5, "learning_rate": 3.118341933446726e-06, "loss": 0.0133, "step": 2410 }, { "epoch": 1.5, "learning_rate": 3.1110373350679057e-06, "loss": 0.0177, "step": 2411 }, { "epoch": 1.5, "learning_rate": 3.1037397258138026e-06, "loss": 0.0121, "step": 2412 }, { "epoch": 1.5, "learning_rate": 3.0964491130881303e-06, "loss": 0.0151, "step": 2413 }, { "epoch": 1.5, "learning_rate": 3.089165504287499e-06, "loss": 0.0102, "step": 2414 }, { "epoch": 1.5, "learning_rate": 3.0818889068014167e-06, "loss": 0.0127, "step": 2415 }, { "epoch": 1.5, "learning_rate": 3.074619328012289e-06, "loss": 0.0205, "step": 2416 }, { "epoch": 1.5, "learning_rate": 3.0673567752953816e-06, "loss": 0.0159, "step": 2417 }, { "epoch": 1.5, "learning_rate": 3.0601012560188427e-06, "loss": 0.0133, "step": 2418 }, { "epoch": 1.5, "learning_rate": 3.052852777543687e-06, "loss": 0.0178, "step": 2419 }, { "epoch": 1.5, "learning_rate": 3.045611347223776e-06, "loss": 0.0156, "step": 2420 }, { "epoch": 1.51, "learning_rate": 3.038376972405831e-06, "loss": 0.0173, "step": 2421 }, { "epoch": 1.51, "learning_rate": 3.031149660429412e-06, "loss": 0.014, "step": 2422 }, { "epoch": 1.51, "learning_rate": 3.0239294186269053e-06, "loss": 0.016, "step": 2423 }, { "epoch": 1.51, "learning_rate": 3.0167162543235384e-06, "loss": 0.0102, "step": 2424 }, { "epoch": 1.51, "learning_rate": 3.00951017483735e-06, "loss": 0.0096, "step": 2425 }, { "epoch": 1.51, "learning_rate": 3.0023111874791954e-06, "loss": 0.0147, "step": 2426 }, { "epoch": 1.51, "learning_rate": 2.9951192995527243e-06, "loss": 0.0114, "step": 2427 }, { "epoch": 1.51, "learning_rate": 2.9879345183543963e-06, "loss": 0.0139, "step": 2428 }, { "epoch": 1.51, "learning_rate": 2.9807568511734564e-06, "loss": 0.0107, "step": 2429 }, { "epoch": 1.51, "learning_rate": 2.9735863052919256e-06, "loss": 0.0147, "step": 2430 }, { "epoch": 1.51, "learning_rate": 2.9664228879846104e-06, "loss": 0.0227, "step": 2431 }, { "epoch": 1.51, "learning_rate": 2.959266606519079e-06, "loss": 0.0182, "step": 2432 }, { "epoch": 1.51, "learning_rate": 2.952117468155663e-06, "loss": 0.0172, "step": 2433 }, { "epoch": 1.51, "learning_rate": 2.944975480147445e-06, "loss": 0.0124, "step": 2434 }, { "epoch": 1.51, "learning_rate": 2.9378406497402577e-06, "loss": 0.0169, "step": 2435 }, { "epoch": 1.51, "learning_rate": 2.930712984172661e-06, "loss": 0.0174, "step": 2436 }, { "epoch": 1.52, "learning_rate": 2.923592490675956e-06, "loss": 0.0142, "step": 2437 }, { "epoch": 1.52, "learning_rate": 2.9164791764741697e-06, "loss": 0.0162, "step": 2438 }, { "epoch": 1.52, "learning_rate": 2.909373048784032e-06, "loss": 0.0188, "step": 2439 }, { "epoch": 1.52, "learning_rate": 2.902274114814995e-06, "loss": 0.0139, "step": 2440 }, { "epoch": 1.52, "learning_rate": 2.8951823817692048e-06, "loss": 0.014, "step": 2441 }, { "epoch": 1.52, "learning_rate": 2.8880978568415084e-06, "loss": 0.0176, "step": 2442 }, { "epoch": 1.52, "learning_rate": 2.881020547219433e-06, "loss": 0.0117, "step": 2443 }, { "epoch": 1.52, "learning_rate": 2.873950460083191e-06, "loss": 0.0175, "step": 2444 }, { "epoch": 1.52, "learning_rate": 2.866887602605667e-06, "loss": 0.012, "step": 2445 }, { "epoch": 1.52, "learning_rate": 2.859831981952402e-06, "loss": 0.0111, "step": 2446 }, { "epoch": 1.52, "learning_rate": 2.8527836052816073e-06, "loss": 0.0165, "step": 2447 }, { "epoch": 1.52, "learning_rate": 2.8457424797441402e-06, "loss": 0.0153, "step": 2448 }, { "epoch": 1.52, "learning_rate": 2.8387086124834952e-06, "loss": 0.0138, "step": 2449 }, { "epoch": 1.52, "learning_rate": 2.831682010635811e-06, "loss": 0.0169, "step": 2450 }, { "epoch": 1.52, "learning_rate": 2.824662681329854e-06, "loss": 0.0175, "step": 2451 }, { "epoch": 1.52, "learning_rate": 2.81765063168701e-06, "loss": 0.0103, "step": 2452 }, { "epoch": 1.53, "learning_rate": 2.8106458688212812e-06, "loss": 0.0151, "step": 2453 }, { "epoch": 1.53, "learning_rate": 2.8036483998392784e-06, "loss": 0.0124, "step": 2454 }, { "epoch": 1.53, "learning_rate": 2.7966582318402046e-06, "loss": 0.0182, "step": 2455 }, { "epoch": 1.53, "learning_rate": 2.7896753719158644e-06, "loss": 0.0078, "step": 2456 }, { "epoch": 1.53, "learning_rate": 2.7826998271506477e-06, "loss": 0.0064, "step": 2457 }, { "epoch": 1.53, "learning_rate": 2.7757316046215145e-06, "loss": 0.0105, "step": 2458 }, { "epoch": 1.53, "learning_rate": 2.768770711398001e-06, "loss": 0.0101, "step": 2459 }, { "epoch": 1.53, "learning_rate": 2.7618171545422156e-06, "loss": 0.008, "step": 2460 }, { "epoch": 1.53, "learning_rate": 2.754870941108815e-06, "loss": 0.0125, "step": 2461 }, { "epoch": 1.53, "learning_rate": 2.7479320781450024e-06, "loss": 0.0112, "step": 2462 }, { "epoch": 1.53, "learning_rate": 2.7410005726905297e-06, "loss": 0.0194, "step": 2463 }, { "epoch": 1.53, "learning_rate": 2.734076431777688e-06, "loss": 0.0139, "step": 2464 }, { "epoch": 1.53, "learning_rate": 2.7271596624312845e-06, "loss": 0.0156, "step": 2465 }, { "epoch": 1.53, "learning_rate": 2.7202502716686573e-06, "loss": 0.0205, "step": 2466 }, { "epoch": 1.53, "learning_rate": 2.713348266499658e-06, "loss": 0.0168, "step": 2467 }, { "epoch": 1.53, "learning_rate": 2.706453653926642e-06, "loss": 0.0119, "step": 2468 }, { "epoch": 1.54, "learning_rate": 2.6995664409444665e-06, "loss": 0.0169, "step": 2469 }, { "epoch": 1.54, "learning_rate": 2.6926866345404846e-06, "loss": 0.0195, "step": 2470 }, { "epoch": 1.54, "learning_rate": 2.6858142416945243e-06, "loss": 0.0185, "step": 2471 }, { "epoch": 1.54, "learning_rate": 2.6789492693789043e-06, "loss": 0.0185, "step": 2472 }, { "epoch": 1.54, "learning_rate": 2.672091724558411e-06, "loss": 0.0132, "step": 2473 }, { "epoch": 1.54, "learning_rate": 2.6652416141902913e-06, "loss": 0.0141, "step": 2474 }, { "epoch": 1.54, "learning_rate": 2.658398945224253e-06, "loss": 0.0134, "step": 2475 }, { "epoch": 1.54, "learning_rate": 2.6515637246024563e-06, "loss": 0.01, "step": 2476 }, { "epoch": 1.54, "learning_rate": 2.644735959259501e-06, "loss": 0.0078, "step": 2477 }, { "epoch": 1.54, "learning_rate": 2.637915656122425e-06, "loss": 0.0138, "step": 2478 }, { "epoch": 1.54, "learning_rate": 2.631102822110695e-06, "loss": 0.0088, "step": 2479 }, { "epoch": 1.54, "learning_rate": 2.624297464136204e-06, "loss": 0.0128, "step": 2480 }, { "epoch": 1.54, "learning_rate": 2.6174995891032495e-06, "loss": 0.014, "step": 2481 }, { "epoch": 1.54, "learning_rate": 2.610709203908548e-06, "loss": 0.0185, "step": 2482 }, { "epoch": 1.54, "learning_rate": 2.6039263154412165e-06, "loss": 0.0156, "step": 2483 }, { "epoch": 1.54, "learning_rate": 2.597150930582757e-06, "loss": 0.0189, "step": 2484 }, { "epoch": 1.55, "learning_rate": 2.5903830562070675e-06, "loss": 0.0138, "step": 2485 }, { "epoch": 1.55, "learning_rate": 2.5836226991804248e-06, "loss": 0.0156, "step": 2486 }, { "epoch": 1.55, "learning_rate": 2.5768698663614777e-06, "loss": 0.013, "step": 2487 }, { "epoch": 1.55, "learning_rate": 2.570124564601241e-06, "loss": 0.0147, "step": 2488 }, { "epoch": 1.55, "learning_rate": 2.563386800743094e-06, "loss": 0.0144, "step": 2489 }, { "epoch": 1.55, "learning_rate": 2.5566565816227585e-06, "loss": 0.0224, "step": 2490 }, { "epoch": 1.55, "learning_rate": 2.549933914068309e-06, "loss": 0.0138, "step": 2491 }, { "epoch": 1.55, "learning_rate": 2.5432188049001615e-06, "loss": 0.0133, "step": 2492 }, { "epoch": 1.55, "learning_rate": 2.5365112609310526e-06, "loss": 0.0079, "step": 2493 }, { "epoch": 1.55, "learning_rate": 2.5298112889660544e-06, "loss": 0.0179, "step": 2494 }, { "epoch": 1.55, "learning_rate": 2.523118895802553e-06, "loss": 0.0167, "step": 2495 }, { "epoch": 1.55, "learning_rate": 2.5164340882302453e-06, "loss": 0.0143, "step": 2496 }, { "epoch": 1.55, "learning_rate": 2.509756873031133e-06, "loss": 0.0105, "step": 2497 }, { "epoch": 1.55, "learning_rate": 2.503087256979514e-06, "loss": 0.017, "step": 2498 }, { "epoch": 1.55, "learning_rate": 2.4964252468419802e-06, "loss": 0.0098, "step": 2499 }, { "epoch": 1.55, "learning_rate": 2.4897708493773997e-06, "loss": 0.0154, "step": 2500 }, { "epoch": 1.56, "learning_rate": 2.483124071336922e-06, "loss": 0.0117, "step": 2501 }, { "epoch": 1.56, "learning_rate": 2.4764849194639707e-06, "loss": 0.0153, "step": 2502 }, { "epoch": 1.56, "learning_rate": 2.4698534004942185e-06, "loss": 0.0124, "step": 2503 }, { "epoch": 1.56, "learning_rate": 2.463229521155611e-06, "loss": 0.0104, "step": 2504 }, { "epoch": 1.56, "learning_rate": 2.456613288168337e-06, "loss": 0.0119, "step": 2505 }, { "epoch": 1.56, "learning_rate": 2.4500047082448196e-06, "loss": 0.0105, "step": 2506 }, { "epoch": 1.56, "learning_rate": 2.4434037880897266e-06, "loss": 0.0179, "step": 2507 }, { "epoch": 1.56, "learning_rate": 2.436810534399956e-06, "loss": 0.0116, "step": 2508 }, { "epoch": 1.56, "learning_rate": 2.430224953864617e-06, "loss": 0.0092, "step": 2509 }, { "epoch": 1.56, "learning_rate": 2.423647053165046e-06, "loss": 0.0185, "step": 2510 }, { "epoch": 1.56, "learning_rate": 2.4170768389747814e-06, "loss": 0.0129, "step": 2511 }, { "epoch": 1.56, "learning_rate": 2.4105143179595657e-06, "loss": 0.0234, "step": 2512 }, { "epoch": 1.56, "learning_rate": 2.403959496777336e-06, "loss": 0.0104, "step": 2513 }, { "epoch": 1.56, "learning_rate": 2.397412382078219e-06, "loss": 0.0119, "step": 2514 }, { "epoch": 1.56, "learning_rate": 2.390872980504516e-06, "loss": 0.0093, "step": 2515 }, { "epoch": 1.56, "learning_rate": 2.384341298690712e-06, "loss": 0.0143, "step": 2516 }, { "epoch": 1.57, "learning_rate": 2.3778173432634555e-06, "loss": 0.0121, "step": 2517 }, { "epoch": 1.57, "learning_rate": 2.371301120841558e-06, "loss": 0.0167, "step": 2518 }, { "epoch": 1.57, "learning_rate": 2.364792638035982e-06, "loss": 0.0156, "step": 2519 }, { "epoch": 1.57, "learning_rate": 2.358291901449842e-06, "loss": 0.0127, "step": 2520 }, { "epoch": 1.57, "learning_rate": 2.3517989176783907e-06, "loss": 0.0132, "step": 2521 }, { "epoch": 1.57, "learning_rate": 2.345313693309018e-06, "loss": 0.0199, "step": 2522 }, { "epoch": 1.57, "learning_rate": 2.33883623492124e-06, "loss": 0.0139, "step": 2523 }, { "epoch": 1.57, "learning_rate": 2.3323665490866964e-06, "loss": 0.0108, "step": 2524 }, { "epoch": 1.57, "learning_rate": 2.3259046423691333e-06, "loss": 0.0112, "step": 2525 }, { "epoch": 1.57, "learning_rate": 2.3194505213244136e-06, "loss": 0.0148, "step": 2526 }, { "epoch": 1.57, "learning_rate": 2.313004192500501e-06, "loss": 0.0153, "step": 2527 }, { "epoch": 1.57, "learning_rate": 2.3065656624374453e-06, "loss": 0.014, "step": 2528 }, { "epoch": 1.57, "learning_rate": 2.300134937667391e-06, "loss": 0.013, "step": 2529 }, { "epoch": 1.57, "learning_rate": 2.293712024714564e-06, "loss": 0.0163, "step": 2530 }, { "epoch": 1.57, "learning_rate": 2.287296930095261e-06, "loss": 0.0132, "step": 2531 }, { "epoch": 1.57, "learning_rate": 2.280889660317851e-06, "loss": 0.022, "step": 2532 }, { "epoch": 1.58, "learning_rate": 2.2744902218827646e-06, "loss": 0.0092, "step": 2533 }, { "epoch": 1.58, "learning_rate": 2.2680986212824786e-06, "loss": 0.0143, "step": 2534 }, { "epoch": 1.58, "learning_rate": 2.2617148650015275e-06, "loss": 0.0123, "step": 2535 }, { "epoch": 1.58, "learning_rate": 2.2553389595164845e-06, "loss": 0.012, "step": 2536 }, { "epoch": 1.58, "learning_rate": 2.24897091129596e-06, "loss": 0.0095, "step": 2537 }, { "epoch": 1.58, "learning_rate": 2.242610726800584e-06, "loss": 0.0115, "step": 2538 }, { "epoch": 1.58, "learning_rate": 2.2362584124830167e-06, "loss": 0.0149, "step": 2539 }, { "epoch": 1.58, "learning_rate": 2.2299139747879385e-06, "loss": 0.0141, "step": 2540 }, { "epoch": 1.58, "learning_rate": 2.2235774201520242e-06, "loss": 0.0081, "step": 2541 }, { "epoch": 1.58, "learning_rate": 2.2172487550039625e-06, "loss": 0.0156, "step": 2542 }, { "epoch": 1.58, "learning_rate": 2.210927985764435e-06, "loss": 0.011, "step": 2543 }, { "epoch": 1.58, "learning_rate": 2.204615118846107e-06, "loss": 0.0131, "step": 2544 }, { "epoch": 1.58, "learning_rate": 2.198310160653636e-06, "loss": 0.0079, "step": 2545 }, { "epoch": 1.58, "learning_rate": 2.1920131175836523e-06, "loss": 0.0106, "step": 2546 }, { "epoch": 1.58, "learning_rate": 2.1857239960247455e-06, "loss": 0.0112, "step": 2547 }, { "epoch": 1.58, "learning_rate": 2.1794428023574876e-06, "loss": 0.0202, "step": 2548 }, { "epoch": 1.59, "learning_rate": 2.1731695429543974e-06, "loss": 0.0238, "step": 2549 }, { "epoch": 1.59, "learning_rate": 2.1669042241799378e-06, "loss": 0.011, "step": 2550 }, { "epoch": 1.59, "learning_rate": 2.1606468523905265e-06, "loss": 0.0075, "step": 2551 }, { "epoch": 1.59, "learning_rate": 2.1543974339345166e-06, "loss": 0.0154, "step": 2552 }, { "epoch": 1.59, "learning_rate": 2.148155975152183e-06, "loss": 0.0124, "step": 2553 }, { "epoch": 1.59, "learning_rate": 2.141922482375737e-06, "loss": 0.0099, "step": 2554 }, { "epoch": 1.59, "learning_rate": 2.1356969619293023e-06, "loss": 0.0224, "step": 2555 }, { "epoch": 1.59, "learning_rate": 2.129479420128915e-06, "loss": 0.0146, "step": 2556 }, { "epoch": 1.59, "learning_rate": 2.123269863282518e-06, "loss": 0.0129, "step": 2557 }, { "epoch": 1.59, "learning_rate": 2.117068297689949e-06, "loss": 0.0194, "step": 2558 }, { "epoch": 1.59, "learning_rate": 2.1108747296429477e-06, "loss": 0.0213, "step": 2559 }, { "epoch": 1.59, "learning_rate": 2.1046891654251266e-06, "loss": 0.0171, "step": 2560 }, { "epoch": 1.59, "learning_rate": 2.098511611311986e-06, "loss": 0.0143, "step": 2561 }, { "epoch": 1.59, "learning_rate": 2.0923420735709043e-06, "loss": 0.0126, "step": 2562 }, { "epoch": 1.59, "learning_rate": 2.0861805584611137e-06, "loss": 0.0143, "step": 2563 }, { "epoch": 1.59, "learning_rate": 2.080027072233718e-06, "loss": 0.0164, "step": 2564 }, { "epoch": 1.6, "learning_rate": 2.0738816211316716e-06, "loss": 0.0132, "step": 2565 }, { "epoch": 1.6, "learning_rate": 2.067744211389777e-06, "loss": 0.0101, "step": 2566 }, { "epoch": 1.6, "learning_rate": 2.06161484923468e-06, "loss": 0.0153, "step": 2567 }, { "epoch": 1.6, "learning_rate": 2.055493540884863e-06, "loss": 0.0078, "step": 2568 }, { "epoch": 1.6, "learning_rate": 2.049380292550629e-06, "loss": 0.0133, "step": 2569 }, { "epoch": 1.6, "learning_rate": 2.043275110434113e-06, "loss": 0.012, "step": 2570 }, { "epoch": 1.6, "learning_rate": 2.037178000729266e-06, "loss": 0.0105, "step": 2571 }, { "epoch": 1.6, "learning_rate": 2.0310889696218416e-06, "loss": 0.0235, "step": 2572 }, { "epoch": 1.6, "learning_rate": 2.025008023289404e-06, "loss": 0.017, "step": 2573 }, { "epoch": 1.6, "learning_rate": 2.018935167901316e-06, "loss": 0.0163, "step": 2574 }, { "epoch": 1.6, "learning_rate": 2.0128704096187258e-06, "loss": 0.0176, "step": 2575 }, { "epoch": 1.6, "learning_rate": 2.006813754594573e-06, "loss": 0.0105, "step": 2576 }, { "epoch": 1.6, "learning_rate": 2.0007652089735707e-06, "loss": 0.0147, "step": 2577 }, { "epoch": 1.6, "learning_rate": 1.994724778892213e-06, "loss": 0.0181, "step": 2578 }, { "epoch": 1.6, "learning_rate": 1.9886924704787482e-06, "loss": 0.0106, "step": 2579 }, { "epoch": 1.6, "learning_rate": 1.9826682898531923e-06, "loss": 0.0157, "step": 2580 }, { "epoch": 1.61, "learning_rate": 1.9766522431273194e-06, "loss": 0.0084, "step": 2581 }, { "epoch": 1.61, "learning_rate": 1.9706443364046413e-06, "loss": 0.0104, "step": 2582 }, { "epoch": 1.61, "learning_rate": 1.964644575780418e-06, "loss": 0.0107, "step": 2583 }, { "epoch": 1.61, "learning_rate": 1.9586529673416433e-06, "loss": 0.0139, "step": 2584 }, { "epoch": 1.61, "learning_rate": 1.9526695171670428e-06, "loss": 0.0161, "step": 2585 }, { "epoch": 1.61, "learning_rate": 1.946694231327061e-06, "loss": 0.0159, "step": 2586 }, { "epoch": 1.61, "learning_rate": 1.9407271158838646e-06, "loss": 0.0189, "step": 2587 }, { "epoch": 1.61, "learning_rate": 1.934768176891323e-06, "loss": 0.0134, "step": 2588 }, { "epoch": 1.61, "learning_rate": 1.928817420395018e-06, "loss": 0.0158, "step": 2589 }, { "epoch": 1.61, "learning_rate": 1.9228748524322283e-06, "loss": 0.0187, "step": 2590 }, { "epoch": 1.61, "learning_rate": 1.916940479031918e-06, "loss": 0.0105, "step": 2591 }, { "epoch": 1.61, "learning_rate": 1.91101430621475e-06, "loss": 0.0127, "step": 2592 }, { "epoch": 1.61, "learning_rate": 1.9050963399930578e-06, "loss": 0.012, "step": 2593 }, { "epoch": 1.61, "learning_rate": 1.8991865863708547e-06, "loss": 0.0108, "step": 2594 }, { "epoch": 1.61, "learning_rate": 1.8932850513438139e-06, "loss": 0.0076, "step": 2595 }, { "epoch": 1.61, "learning_rate": 1.8873917408992792e-06, "loss": 0.0185, "step": 2596 }, { "epoch": 1.62, "learning_rate": 1.881506661016249e-06, "loss": 0.0159, "step": 2597 }, { "epoch": 1.62, "learning_rate": 1.8756298176653632e-06, "loss": 0.012, "step": 2598 }, { "epoch": 1.62, "learning_rate": 1.8697612168089152e-06, "loss": 0.0158, "step": 2599 }, { "epoch": 1.62, "learning_rate": 1.8639008644008317e-06, "loss": 0.0098, "step": 2600 }, { "epoch": 1.62, "learning_rate": 1.858048766386672e-06, "loss": 0.018, "step": 2601 }, { "epoch": 1.62, "learning_rate": 1.8522049287036204e-06, "loss": 0.0169, "step": 2602 }, { "epoch": 1.62, "learning_rate": 1.846369357280484e-06, "loss": 0.0134, "step": 2603 }, { "epoch": 1.62, "learning_rate": 1.8405420580376755e-06, "loss": 0.0131, "step": 2604 }, { "epoch": 1.62, "learning_rate": 1.8347230368872227e-06, "loss": 0.0137, "step": 2605 }, { "epoch": 1.62, "learning_rate": 1.8289122997327536e-06, "loss": 0.0124, "step": 2606 }, { "epoch": 1.62, "learning_rate": 1.8231098524694868e-06, "loss": 0.0162, "step": 2607 }, { "epoch": 1.62, "learning_rate": 1.8173157009842368e-06, "loss": 0.0096, "step": 2608 }, { "epoch": 1.62, "learning_rate": 1.811529851155398e-06, "loss": 0.0144, "step": 2609 }, { "epoch": 1.62, "learning_rate": 1.805752308852945e-06, "loss": 0.0155, "step": 2610 }, { "epoch": 1.62, "learning_rate": 1.7999830799384221e-06, "loss": 0.0188, "step": 2611 }, { "epoch": 1.62, "learning_rate": 1.79422217026494e-06, "loss": 0.0139, "step": 2612 }, { "epoch": 1.62, "learning_rate": 1.7884695856771704e-06, "loss": 0.0136, "step": 2613 }, { "epoch": 1.63, "learning_rate": 1.7827253320113347e-06, "loss": 0.0191, "step": 2614 }, { "epoch": 1.63, "learning_rate": 1.776989415095206e-06, "loss": 0.0133, "step": 2615 }, { "epoch": 1.63, "learning_rate": 1.771261840748102e-06, "loss": 0.018, "step": 2616 }, { "epoch": 1.63, "learning_rate": 1.7655426147808675e-06, "loss": 0.0179, "step": 2617 }, { "epoch": 1.63, "learning_rate": 1.7598317429958855e-06, "loss": 0.016, "step": 2618 }, { "epoch": 1.63, "learning_rate": 1.7541292311870616e-06, "loss": 0.0108, "step": 2619 }, { "epoch": 1.63, "learning_rate": 1.748435085139818e-06, "loss": 0.0119, "step": 2620 }, { "epoch": 1.63, "learning_rate": 1.7427493106310933e-06, "loss": 0.0124, "step": 2621 }, { "epoch": 1.63, "learning_rate": 1.7370719134293302e-06, "loss": 0.0155, "step": 2622 }, { "epoch": 1.63, "learning_rate": 1.7314028992944688e-06, "loss": 0.0104, "step": 2623 }, { "epoch": 1.63, "learning_rate": 1.7257422739779495e-06, "loss": 0.0192, "step": 2624 }, { "epoch": 1.63, "learning_rate": 1.720090043222704e-06, "loss": 0.0107, "step": 2625 }, { "epoch": 1.63, "learning_rate": 1.714446212763138e-06, "loss": 0.0184, "step": 2626 }, { "epoch": 1.63, "learning_rate": 1.7088107883251414e-06, "loss": 0.0111, "step": 2627 }, { "epoch": 1.63, "learning_rate": 1.703183775626075e-06, "loss": 0.0102, "step": 2628 }, { "epoch": 1.63, "learning_rate": 1.6975651803747716e-06, "loss": 0.0163, "step": 2629 }, { "epoch": 1.64, "learning_rate": 1.6919550082715108e-06, "loss": 0.0135, "step": 2630 }, { "epoch": 1.64, "learning_rate": 1.686353265008035e-06, "loss": 0.0134, "step": 2631 }, { "epoch": 1.64, "learning_rate": 1.6807599562675392e-06, "loss": 0.0172, "step": 2632 }, { "epoch": 1.64, "learning_rate": 1.6751750877246475e-06, "loss": 0.0139, "step": 2633 }, { "epoch": 1.64, "learning_rate": 1.6695986650454355e-06, "loss": 0.0131, "step": 2634 }, { "epoch": 1.64, "learning_rate": 1.6640306938874052e-06, "loss": 0.0107, "step": 2635 }, { "epoch": 1.64, "learning_rate": 1.6584711798994758e-06, "loss": 0.0106, "step": 2636 }, { "epoch": 1.64, "learning_rate": 1.652920128722001e-06, "loss": 0.0152, "step": 2637 }, { "epoch": 1.64, "learning_rate": 1.6473775459867424e-06, "loss": 0.0132, "step": 2638 }, { "epoch": 1.64, "learning_rate": 1.6418434373168623e-06, "loss": 0.0167, "step": 2639 }, { "epoch": 1.64, "learning_rate": 1.6363178083269381e-06, "loss": 0.0161, "step": 2640 }, { "epoch": 1.64, "learning_rate": 1.6308006646229379e-06, "loss": 0.0125, "step": 2641 }, { "epoch": 1.64, "learning_rate": 1.6252920118022186e-06, "loss": 0.0103, "step": 2642 }, { "epoch": 1.64, "learning_rate": 1.6197918554535285e-06, "loss": 0.0134, "step": 2643 }, { "epoch": 1.64, "learning_rate": 1.614300201156994e-06, "loss": 0.0138, "step": 2644 }, { "epoch": 1.64, "learning_rate": 1.608817054484113e-06, "loss": 0.0116, "step": 2645 }, { "epoch": 1.65, "learning_rate": 1.6033424209977577e-06, "loss": 0.026, "step": 2646 }, { "epoch": 1.65, "learning_rate": 1.5978763062521575e-06, "loss": 0.0115, "step": 2647 }, { "epoch": 1.65, "learning_rate": 1.592418715792904e-06, "loss": 0.018, "step": 2648 }, { "epoch": 1.65, "learning_rate": 1.5869696551569346e-06, "loss": 0.0149, "step": 2649 }, { "epoch": 1.65, "learning_rate": 1.5815291298725387e-06, "loss": 0.0146, "step": 2650 }, { "epoch": 1.65, "learning_rate": 1.5760971454593476e-06, "loss": 0.0091, "step": 2651 }, { "epoch": 1.65, "learning_rate": 1.5706737074283163e-06, "loss": 0.0093, "step": 2652 }, { "epoch": 1.65, "learning_rate": 1.565258821281742e-06, "loss": 0.0237, "step": 2653 }, { "epoch": 1.65, "learning_rate": 1.5598524925132396e-06, "loss": 0.0129, "step": 2654 }, { "epoch": 1.65, "learning_rate": 1.5544547266077425e-06, "loss": 0.0216, "step": 2655 }, { "epoch": 1.65, "learning_rate": 1.549065529041499e-06, "loss": 0.0188, "step": 2656 }, { "epoch": 1.65, "learning_rate": 1.5436849052820646e-06, "loss": 0.0144, "step": 2657 }, { "epoch": 1.65, "learning_rate": 1.538312860788288e-06, "loss": 0.0116, "step": 2658 }, { "epoch": 1.65, "learning_rate": 1.5329494010103263e-06, "loss": 0.0115, "step": 2659 }, { "epoch": 1.65, "learning_rate": 1.52759453138962e-06, "loss": 0.017, "step": 2660 }, { "epoch": 1.65, "learning_rate": 1.5222482573588949e-06, "loss": 0.0151, "step": 2661 }, { "epoch": 1.66, "learning_rate": 1.516910584342156e-06, "loss": 0.0115, "step": 2662 }, { "epoch": 1.66, "learning_rate": 1.5115815177546854e-06, "loss": 0.0179, "step": 2663 }, { "epoch": 1.66, "learning_rate": 1.5062610630030317e-06, "loss": 0.016, "step": 2664 }, { "epoch": 1.66, "learning_rate": 1.5009492254850056e-06, "loss": 0.0115, "step": 2665 }, { "epoch": 1.66, "learning_rate": 1.4956460105896763e-06, "loss": 0.0172, "step": 2666 }, { "epoch": 1.66, "learning_rate": 1.4903514236973682e-06, "loss": 0.0132, "step": 2667 }, { "epoch": 1.66, "learning_rate": 1.4850654701796418e-06, "loss": 0.0093, "step": 2668 }, { "epoch": 1.66, "learning_rate": 1.4797881553993099e-06, "loss": 0.0154, "step": 2669 }, { "epoch": 1.66, "learning_rate": 1.4745194847104184e-06, "loss": 0.0236, "step": 2670 }, { "epoch": 1.66, "learning_rate": 1.469259463458238e-06, "loss": 0.012, "step": 2671 }, { "epoch": 1.66, "learning_rate": 1.4640080969792669e-06, "loss": 0.0122, "step": 2672 }, { "epoch": 1.66, "learning_rate": 1.4587653906012311e-06, "loss": 0.0169, "step": 2673 }, { "epoch": 1.66, "learning_rate": 1.4535313496430558e-06, "loss": 0.0093, "step": 2674 }, { "epoch": 1.66, "learning_rate": 1.4483059794148869e-06, "loss": 0.0135, "step": 2675 }, { "epoch": 1.66, "learning_rate": 1.4430892852180688e-06, "loss": 0.0112, "step": 2676 }, { "epoch": 1.66, "learning_rate": 1.4378812723451397e-06, "loss": 0.0143, "step": 2677 }, { "epoch": 1.67, "learning_rate": 1.4326819460798369e-06, "loss": 0.0077, "step": 2678 }, { "epoch": 1.67, "learning_rate": 1.4274913116970846e-06, "loss": 0.0099, "step": 2679 }, { "epoch": 1.67, "learning_rate": 1.4223093744629802e-06, "loss": 0.0126, "step": 2680 }, { "epoch": 1.67, "learning_rate": 1.4171361396348093e-06, "loss": 0.0117, "step": 2681 }, { "epoch": 1.67, "learning_rate": 1.4119716124610228e-06, "loss": 0.0144, "step": 2682 }, { "epoch": 1.67, "learning_rate": 1.4068157981812347e-06, "loss": 0.0108, "step": 2683 }, { "epoch": 1.67, "learning_rate": 1.4016687020262231e-06, "loss": 0.015, "step": 2684 }, { "epoch": 1.67, "learning_rate": 1.3965303292179211e-06, "loss": 0.0169, "step": 2685 }, { "epoch": 1.67, "learning_rate": 1.3914006849694117e-06, "loss": 0.0172, "step": 2686 }, { "epoch": 1.67, "learning_rate": 1.3862797744849188e-06, "loss": 0.0122, "step": 2687 }, { "epoch": 1.67, "learning_rate": 1.3811676029598099e-06, "loss": 0.0109, "step": 2688 }, { "epoch": 1.67, "learning_rate": 1.3760641755805848e-06, "loss": 0.0131, "step": 2689 }, { "epoch": 1.67, "learning_rate": 1.370969497524872e-06, "loss": 0.017, "step": 2690 }, { "epoch": 1.67, "learning_rate": 1.365883573961424e-06, "loss": 0.0104, "step": 2691 }, { "epoch": 1.67, "learning_rate": 1.360806410050114e-06, "loss": 0.0101, "step": 2692 }, { "epoch": 1.67, "learning_rate": 1.3557380109419193e-06, "loss": 0.0172, "step": 2693 }, { "epoch": 1.68, "learning_rate": 1.3506783817789337e-06, "loss": 0.0082, "step": 2694 }, { "epoch": 1.68, "learning_rate": 1.3456275276943543e-06, "loss": 0.0189, "step": 2695 }, { "epoch": 1.68, "learning_rate": 1.3405854538124664e-06, "loss": 0.0125, "step": 2696 }, { "epoch": 1.68, "learning_rate": 1.3355521652486558e-06, "loss": 0.0132, "step": 2697 }, { "epoch": 1.68, "learning_rate": 1.3305276671093914e-06, "loss": 0.0166, "step": 2698 }, { "epoch": 1.68, "learning_rate": 1.3255119644922266e-06, "loss": 0.0082, "step": 2699 }, { "epoch": 1.68, "learning_rate": 1.3205050624857895e-06, "loss": 0.0132, "step": 2700 }, { "epoch": 1.68, "learning_rate": 1.3155069661697816e-06, "loss": 0.0068, "step": 2701 }, { "epoch": 1.68, "learning_rate": 1.3105176806149645e-06, "loss": 0.0158, "step": 2702 }, { "epoch": 1.68, "learning_rate": 1.3055372108831688e-06, "loss": 0.0169, "step": 2703 }, { "epoch": 1.68, "learning_rate": 1.300565562027276e-06, "loss": 0.0158, "step": 2704 }, { "epoch": 1.68, "learning_rate": 1.295602739091224e-06, "loss": 0.0155, "step": 2705 }, { "epoch": 1.68, "learning_rate": 1.2906487471099871e-06, "loss": 0.0165, "step": 2706 }, { "epoch": 1.68, "learning_rate": 1.2857035911095894e-06, "loss": 0.0139, "step": 2707 }, { "epoch": 1.68, "learning_rate": 1.2807672761070855e-06, "loss": 0.0151, "step": 2708 }, { "epoch": 1.68, "learning_rate": 1.2758398071105626e-06, "loss": 0.0113, "step": 2709 }, { "epoch": 1.69, "learning_rate": 1.2709211891191331e-06, "loss": 0.014, "step": 2710 }, { "epoch": 1.69, "learning_rate": 1.2660114271229306e-06, "loss": 0.0146, "step": 2711 }, { "epoch": 1.69, "learning_rate": 1.2611105261030987e-06, "loss": 0.0109, "step": 2712 }, { "epoch": 1.69, "learning_rate": 1.2562184910317965e-06, "loss": 0.0079, "step": 2713 }, { "epoch": 1.69, "learning_rate": 1.2513353268721907e-06, "loss": 0.0115, "step": 2714 }, { "epoch": 1.69, "learning_rate": 1.2464610385784381e-06, "loss": 0.0108, "step": 2715 }, { "epoch": 1.69, "learning_rate": 1.2415956310956968e-06, "loss": 0.0152, "step": 2716 }, { "epoch": 1.69, "learning_rate": 1.2367391093601233e-06, "loss": 0.0181, "step": 2717 }, { "epoch": 1.69, "learning_rate": 1.231891478298841e-06, "loss": 0.0115, "step": 2718 }, { "epoch": 1.69, "learning_rate": 1.2270527428299684e-06, "loss": 0.0117, "step": 2719 }, { "epoch": 1.69, "learning_rate": 1.2222229078625935e-06, "loss": 0.0115, "step": 2720 }, { "epoch": 1.69, "learning_rate": 1.2174019782967728e-06, "loss": 0.0131, "step": 2721 }, { "epoch": 1.69, "learning_rate": 1.212589959023529e-06, "loss": 0.0145, "step": 2722 }, { "epoch": 1.69, "learning_rate": 1.2077868549248484e-06, "loss": 0.0105, "step": 2723 }, { "epoch": 1.69, "learning_rate": 1.2029926708736673e-06, "loss": 0.0113, "step": 2724 }, { "epoch": 1.69, "learning_rate": 1.198207411733877e-06, "loss": 0.0112, "step": 2725 }, { "epoch": 1.7, "learning_rate": 1.1934310823603113e-06, "loss": 0.0154, "step": 2726 }, { "epoch": 1.7, "learning_rate": 1.1886636875987456e-06, "loss": 0.0106, "step": 2727 }, { "epoch": 1.7, "learning_rate": 1.183905232285888e-06, "loss": 0.0101, "step": 2728 }, { "epoch": 1.7, "learning_rate": 1.179155721249381e-06, "loss": 0.0137, "step": 2729 }, { "epoch": 1.7, "learning_rate": 1.1744151593077925e-06, "loss": 0.0096, "step": 2730 }, { "epoch": 1.7, "learning_rate": 1.1696835512706052e-06, "loss": 0.0132, "step": 2731 }, { "epoch": 1.7, "learning_rate": 1.1649609019382258e-06, "loss": 0.0124, "step": 2732 }, { "epoch": 1.7, "learning_rate": 1.1602472161019685e-06, "loss": 0.0174, "step": 2733 }, { "epoch": 1.7, "learning_rate": 1.1555424985440522e-06, "loss": 0.0152, "step": 2734 }, { "epoch": 1.7, "learning_rate": 1.1508467540376e-06, "loss": 0.011, "step": 2735 }, { "epoch": 1.7, "learning_rate": 1.1461599873466334e-06, "loss": 0.0099, "step": 2736 }, { "epoch": 1.7, "learning_rate": 1.1414822032260552e-06, "loss": 0.0085, "step": 2737 }, { "epoch": 1.7, "learning_rate": 1.1368134064216663e-06, "loss": 0.0134, "step": 2738 }, { "epoch": 1.7, "learning_rate": 1.1321536016701473e-06, "loss": 0.0124, "step": 2739 }, { "epoch": 1.7, "learning_rate": 1.127502793699049e-06, "loss": 0.0116, "step": 2740 }, { "epoch": 1.7, "learning_rate": 1.1228609872268038e-06, "loss": 0.0152, "step": 2741 }, { "epoch": 1.71, "learning_rate": 1.1182281869627054e-06, "loss": 0.0123, "step": 2742 }, { "epoch": 1.71, "learning_rate": 1.1136043976069133e-06, "loss": 0.0156, "step": 2743 }, { "epoch": 1.71, "learning_rate": 1.1089896238504461e-06, "loss": 0.0163, "step": 2744 }, { "epoch": 1.71, "learning_rate": 1.104383870375172e-06, "loss": 0.0115, "step": 2745 }, { "epoch": 1.71, "learning_rate": 1.0997871418538131e-06, "loss": 0.0113, "step": 2746 }, { "epoch": 1.71, "learning_rate": 1.095199442949928e-06, "loss": 0.013, "step": 2747 }, { "epoch": 1.71, "learning_rate": 1.090620778317919e-06, "loss": 0.0123, "step": 2748 }, { "epoch": 1.71, "learning_rate": 1.086051152603026e-06, "loss": 0.0157, "step": 2749 }, { "epoch": 1.71, "learning_rate": 1.0814905704413093e-06, "loss": 0.0085, "step": 2750 }, { "epoch": 1.71, "learning_rate": 1.0769390364596633e-06, "loss": 0.013, "step": 2751 }, { "epoch": 1.71, "learning_rate": 1.0723965552757975e-06, "loss": 0.0138, "step": 2752 }, { "epoch": 1.71, "learning_rate": 1.0678631314982391e-06, "loss": 0.0117, "step": 2753 }, { "epoch": 1.71, "learning_rate": 1.0633387697263254e-06, "loss": 0.0167, "step": 2754 }, { "epoch": 1.71, "learning_rate": 1.0588234745502012e-06, "loss": 0.0123, "step": 2755 }, { "epoch": 1.71, "learning_rate": 1.0543172505508104e-06, "loss": 0.0109, "step": 2756 }, { "epoch": 1.71, "learning_rate": 1.0498201022998943e-06, "loss": 0.0105, "step": 2757 }, { "epoch": 1.72, "learning_rate": 1.0453320343599915e-06, "loss": 0.016, "step": 2758 }, { "epoch": 1.72, "learning_rate": 1.0408530512844196e-06, "loss": 0.0092, "step": 2759 }, { "epoch": 1.72, "learning_rate": 1.0363831576172834e-06, "loss": 0.0109, "step": 2760 }, { "epoch": 1.72, "learning_rate": 1.0319223578934712e-06, "loss": 0.017, "step": 2761 }, { "epoch": 1.72, "learning_rate": 1.0274706566386395e-06, "loss": 0.0197, "step": 2762 }, { "epoch": 1.72, "learning_rate": 1.023028058369212e-06, "loss": 0.0124, "step": 2763 }, { "epoch": 1.72, "learning_rate": 1.0185945675923813e-06, "loss": 0.0145, "step": 2764 }, { "epoch": 1.72, "learning_rate": 1.0141701888061018e-06, "loss": 0.0109, "step": 2765 }, { "epoch": 1.72, "learning_rate": 1.009754926499077e-06, "loss": 0.0093, "step": 2766 }, { "epoch": 1.72, "learning_rate": 1.0053487851507648e-06, "loss": 0.0086, "step": 2767 }, { "epoch": 1.72, "learning_rate": 1.000951769231372e-06, "loss": 0.0105, "step": 2768 }, { "epoch": 1.72, "learning_rate": 9.965638832018432e-07, "loss": 0.0103, "step": 2769 }, { "epoch": 1.72, "learning_rate": 9.92185131513862e-07, "loss": 0.0188, "step": 2770 }, { "epoch": 1.72, "learning_rate": 9.878155186098481e-07, "loss": 0.0126, "step": 2771 }, { "epoch": 1.72, "learning_rate": 9.834550489229421e-07, "loss": 0.0128, "step": 2772 }, { "epoch": 1.72, "learning_rate": 9.791037268770144e-07, "loss": 0.0082, "step": 2773 }, { "epoch": 1.73, "learning_rate": 9.747615568866553e-07, "loss": 0.0159, "step": 2774 }, { "epoch": 1.73, "learning_rate": 9.704285433571637e-07, "loss": 0.0143, "step": 2775 }, { "epoch": 1.73, "learning_rate": 9.661046906845551e-07, "loss": 0.0135, "step": 2776 }, { "epoch": 1.73, "learning_rate": 9.617900032555482e-07, "loss": 0.0113, "step": 2777 }, { "epoch": 1.73, "learning_rate": 9.57484485447564e-07, "loss": 0.0116, "step": 2778 }, { "epoch": 1.73, "learning_rate": 9.531881416287203e-07, "loss": 0.015, "step": 2779 }, { "epoch": 1.73, "learning_rate": 9.489009761578294e-07, "loss": 0.0192, "step": 2780 }, { "epoch": 1.73, "learning_rate": 9.446229933843886e-07, "loss": 0.0085, "step": 2781 }, { "epoch": 1.73, "learning_rate": 9.403541976485797e-07, "loss": 0.0141, "step": 2782 }, { "epoch": 1.73, "learning_rate": 9.360945932812648e-07, "loss": 0.0126, "step": 2783 }, { "epoch": 1.73, "learning_rate": 9.318441846039828e-07, "loss": 0.0088, "step": 2784 }, { "epoch": 1.73, "learning_rate": 9.276029759289384e-07, "loss": 0.0119, "step": 2785 }, { "epoch": 1.73, "learning_rate": 9.233709715590056e-07, "loss": 0.0135, "step": 2786 }, { "epoch": 1.73, "learning_rate": 9.191481757877219e-07, "loss": 0.0069, "step": 2787 }, { "epoch": 1.73, "learning_rate": 9.149345928992781e-07, "loss": 0.0151, "step": 2788 }, { "epoch": 1.73, "learning_rate": 9.107302271685226e-07, "loss": 0.0131, "step": 2789 }, { "epoch": 1.74, "learning_rate": 9.065350828609509e-07, "loss": 0.012, "step": 2790 }, { "epoch": 1.74, "learning_rate": 9.023491642326998e-07, "loss": 0.0173, "step": 2791 }, { "epoch": 1.74, "learning_rate": 8.981724755305499e-07, "loss": 0.0142, "step": 2792 }, { "epoch": 1.74, "learning_rate": 8.940050209919193e-07, "loss": 0.0115, "step": 2793 }, { "epoch": 1.74, "learning_rate": 8.898468048448528e-07, "loss": 0.0146, "step": 2794 }, { "epoch": 1.74, "learning_rate": 8.856978313080244e-07, "loss": 0.0227, "step": 2795 }, { "epoch": 1.74, "learning_rate": 8.815581045907329e-07, "loss": 0.0106, "step": 2796 }, { "epoch": 1.74, "learning_rate": 8.774276288928951e-07, "loss": 0.0179, "step": 2797 }, { "epoch": 1.74, "learning_rate": 8.733064084050402e-07, "loss": 0.0123, "step": 2798 }, { "epoch": 1.74, "learning_rate": 8.691944473083114e-07, "loss": 0.017, "step": 2799 }, { "epoch": 1.74, "learning_rate": 8.650917497744548e-07, "loss": 0.0129, "step": 2800 }, { "epoch": 1.74, "learning_rate": 8.609983199658168e-07, "loss": 0.0183, "step": 2801 }, { "epoch": 1.74, "learning_rate": 8.569141620353461e-07, "loss": 0.0179, "step": 2802 }, { "epoch": 1.74, "learning_rate": 8.528392801265828e-07, "loss": 0.0119, "step": 2803 }, { "epoch": 1.74, "learning_rate": 8.487736783736533e-07, "loss": 0.0121, "step": 2804 }, { "epoch": 1.74, "learning_rate": 8.447173609012693e-07, "loss": 0.0102, "step": 2805 }, { "epoch": 1.75, "learning_rate": 8.406703318247323e-07, "loss": 0.0108, "step": 2806 }, { "epoch": 1.75, "learning_rate": 8.366325952499066e-07, "loss": 0.0105, "step": 2807 }, { "epoch": 1.75, "learning_rate": 8.326041552732378e-07, "loss": 0.0062, "step": 2808 }, { "epoch": 1.75, "learning_rate": 8.285850159817388e-07, "loss": 0.0114, "step": 2809 }, { "epoch": 1.75, "learning_rate": 8.245751814529812e-07, "loss": 0.0153, "step": 2810 }, { "epoch": 1.75, "learning_rate": 8.205746557551031e-07, "loss": 0.0114, "step": 2811 }, { "epoch": 1.75, "learning_rate": 8.165834429467956e-07, "loss": 0.0089, "step": 2812 }, { "epoch": 1.75, "learning_rate": 8.126015470773019e-07, "loss": 0.012, "step": 2813 }, { "epoch": 1.75, "learning_rate": 8.086289721864127e-07, "loss": 0.0107, "step": 2814 }, { "epoch": 1.75, "learning_rate": 8.046657223044618e-07, "loss": 0.0155, "step": 2815 }, { "epoch": 1.75, "learning_rate": 8.007118014523241e-07, "loss": 0.0107, "step": 2816 }, { "epoch": 1.75, "learning_rate": 7.967672136414051e-07, "loss": 0.0102, "step": 2817 }, { "epoch": 1.75, "learning_rate": 7.928319628736481e-07, "loss": 0.0111, "step": 2818 }, { "epoch": 1.75, "learning_rate": 7.889060531415193e-07, "loss": 0.0068, "step": 2819 }, { "epoch": 1.75, "learning_rate": 7.849894884280085e-07, "loss": 0.0128, "step": 2820 }, { "epoch": 1.75, "learning_rate": 7.81082272706627e-07, "loss": 0.0097, "step": 2821 }, { "epoch": 1.75, "learning_rate": 7.771844099413972e-07, "loss": 0.0151, "step": 2822 }, { "epoch": 1.76, "learning_rate": 7.73295904086856e-07, "loss": 0.0178, "step": 2823 }, { "epoch": 1.76, "learning_rate": 7.694167590880475e-07, "loss": 0.0104, "step": 2824 }, { "epoch": 1.76, "learning_rate": 7.65546978880517e-07, "loss": 0.0111, "step": 2825 }, { "epoch": 1.76, "learning_rate": 7.616865673903073e-07, "loss": 0.0181, "step": 2826 }, { "epoch": 1.76, "learning_rate": 7.578355285339589e-07, "loss": 0.0144, "step": 2827 }, { "epoch": 1.76, "learning_rate": 7.539938662185064e-07, "loss": 0.0132, "step": 2828 }, { "epoch": 1.76, "learning_rate": 7.501615843414623e-07, "loss": 0.0082, "step": 2829 }, { "epoch": 1.76, "learning_rate": 7.463386867908318e-07, "loss": 0.0157, "step": 2830 }, { "epoch": 1.76, "learning_rate": 7.425251774450948e-07, "loss": 0.0112, "step": 2831 }, { "epoch": 1.76, "learning_rate": 7.387210601732053e-07, "loss": 0.0132, "step": 2832 }, { "epoch": 1.76, "learning_rate": 7.349263388345929e-07, "loss": 0.0161, "step": 2833 }, { "epoch": 1.76, "learning_rate": 7.311410172791522e-07, "loss": 0.0226, "step": 2834 }, { "epoch": 1.76, "learning_rate": 7.273650993472414e-07, "loss": 0.01, "step": 2835 }, { "epoch": 1.76, "learning_rate": 7.235985888696762e-07, "loss": 0.0221, "step": 2836 }, { "epoch": 1.76, "learning_rate": 7.198414896677308e-07, "loss": 0.0189, "step": 2837 }, { "epoch": 1.76, "learning_rate": 7.160938055531341e-07, "loss": 0.0062, "step": 2838 }, { "epoch": 1.77, "learning_rate": 7.123555403280558e-07, "loss": 0.0067, "step": 2839 }, { "epoch": 1.77, "learning_rate": 7.086266977851141e-07, "loss": 0.0114, "step": 2840 }, { "epoch": 1.77, "learning_rate": 7.049072817073677e-07, "loss": 0.0106, "step": 2841 }, { "epoch": 1.77, "learning_rate": 7.011972958683122e-07, "loss": 0.018, "step": 2842 }, { "epoch": 1.77, "learning_rate": 6.974967440318736e-07, "loss": 0.0128, "step": 2843 }, { "epoch": 1.77, "learning_rate": 6.938056299524099e-07, "loss": 0.0145, "step": 2844 }, { "epoch": 1.77, "learning_rate": 6.901239573746987e-07, "loss": 0.0173, "step": 2845 }, { "epoch": 1.77, "learning_rate": 6.864517300339435e-07, "loss": 0.0115, "step": 2846 }, { "epoch": 1.77, "learning_rate": 6.827889516557673e-07, "loss": 0.0192, "step": 2847 }, { "epoch": 1.77, "learning_rate": 6.791356259561976e-07, "loss": 0.013, "step": 2848 }, { "epoch": 1.77, "learning_rate": 6.754917566416796e-07, "loss": 0.0133, "step": 2849 }, { "epoch": 1.77, "learning_rate": 6.718573474090673e-07, "loss": 0.0222, "step": 2850 }, { "epoch": 1.77, "learning_rate": 6.682324019456065e-07, "loss": 0.0126, "step": 2851 }, { "epoch": 1.77, "learning_rate": 6.64616923928949e-07, "loss": 0.0124, "step": 2852 }, { "epoch": 1.77, "learning_rate": 6.610109170271417e-07, "loss": 0.0171, "step": 2853 }, { "epoch": 1.77, "learning_rate": 6.574143848986226e-07, "loss": 0.0149, "step": 2854 }, { "epoch": 1.78, "learning_rate": 6.538273311922105e-07, "loss": 0.0086, "step": 2855 }, { "epoch": 1.78, "learning_rate": 6.502497595471157e-07, "loss": 0.0127, "step": 2856 }, { "epoch": 1.78, "learning_rate": 6.46681673592926e-07, "loss": 0.0117, "step": 2857 }, { "epoch": 1.78, "learning_rate": 6.431230769496056e-07, "loss": 0.0139, "step": 2858 }, { "epoch": 1.78, "learning_rate": 6.395739732274919e-07, "loss": 0.0069, "step": 2859 }, { "epoch": 1.78, "learning_rate": 6.360343660272905e-07, "loss": 0.0139, "step": 2860 }, { "epoch": 1.78, "learning_rate": 6.325042589400709e-07, "loss": 0.0049, "step": 2861 }, { "epoch": 1.78, "learning_rate": 6.289836555472683e-07, "loss": 0.0172, "step": 2862 }, { "epoch": 1.78, "learning_rate": 6.254725594206745e-07, "loss": 0.0055, "step": 2863 }, { "epoch": 1.78, "learning_rate": 6.219709741224322e-07, "loss": 0.0153, "step": 2864 }, { "epoch": 1.78, "learning_rate": 6.184789032050408e-07, "loss": 0.0206, "step": 2865 }, { "epoch": 1.78, "learning_rate": 6.149963502113421e-07, "loss": 0.0131, "step": 2866 }, { "epoch": 1.78, "learning_rate": 6.11523318674524e-07, "loss": 0.0095, "step": 2867 }, { "epoch": 1.78, "learning_rate": 6.080598121181158e-07, "loss": 0.0118, "step": 2868 }, { "epoch": 1.78, "learning_rate": 6.046058340559824e-07, "loss": 0.0149, "step": 2869 }, { "epoch": 1.78, "learning_rate": 6.011613879923161e-07, "loss": 0.0139, "step": 2870 }, { "epoch": 1.79, "learning_rate": 5.977264774216485e-07, "loss": 0.0121, "step": 2871 }, { "epoch": 1.79, "learning_rate": 5.943011058288295e-07, "loss": 0.0127, "step": 2872 }, { "epoch": 1.79, "learning_rate": 5.908852766890361e-07, "loss": 0.0124, "step": 2873 }, { "epoch": 1.79, "learning_rate": 5.874789934677583e-07, "loss": 0.0137, "step": 2874 }, { "epoch": 1.79, "learning_rate": 5.840822596208073e-07, "loss": 0.0151, "step": 2875 }, { "epoch": 1.79, "learning_rate": 5.806950785943033e-07, "loss": 0.0117, "step": 2876 }, { "epoch": 1.79, "learning_rate": 5.773174538246762e-07, "loss": 0.0134, "step": 2877 }, { "epoch": 1.79, "learning_rate": 5.739493887386583e-07, "loss": 0.0119, "step": 2878 }, { "epoch": 1.79, "learning_rate": 5.705908867532862e-07, "loss": 0.0215, "step": 2879 }, { "epoch": 1.79, "learning_rate": 5.672419512758909e-07, "loss": 0.0115, "step": 2880 }, { "epoch": 1.79, "learning_rate": 5.639025857041014e-07, "loss": 0.0085, "step": 2881 }, { "epoch": 1.79, "learning_rate": 5.605727934258354e-07, "loss": 0.012, "step": 2882 }, { "epoch": 1.79, "learning_rate": 5.572525778192983e-07, "loss": 0.0102, "step": 2883 }, { "epoch": 1.79, "learning_rate": 5.53941942252979e-07, "loss": 0.0124, "step": 2884 }, { "epoch": 1.79, "learning_rate": 5.506408900856508e-07, "loss": 0.0096, "step": 2885 }, { "epoch": 1.79, "learning_rate": 5.473494246663591e-07, "loss": 0.0123, "step": 2886 }, { "epoch": 1.8, "learning_rate": 5.440675493344261e-07, "loss": 0.0107, "step": 2887 }, { "epoch": 1.8, "learning_rate": 5.407952674194461e-07, "loss": 0.0094, "step": 2888 }, { "epoch": 1.8, "learning_rate": 5.375325822412747e-07, "loss": 0.0165, "step": 2889 }, { "epoch": 1.8, "learning_rate": 5.342794971100374e-07, "loss": 0.0116, "step": 2890 }, { "epoch": 1.8, "learning_rate": 5.310360153261162e-07, "loss": 0.0124, "step": 2891 }, { "epoch": 1.8, "learning_rate": 5.278021401801536e-07, "loss": 0.0123, "step": 2892 }, { "epoch": 1.8, "learning_rate": 5.245778749530406e-07, "loss": 0.0142, "step": 2893 }, { "epoch": 1.8, "learning_rate": 5.213632229159227e-07, "loss": 0.0129, "step": 2894 }, { "epoch": 1.8, "learning_rate": 5.181581873301944e-07, "loss": 0.0153, "step": 2895 }, { "epoch": 1.8, "learning_rate": 5.14962771447487e-07, "loss": 0.0155, "step": 2896 }, { "epoch": 1.8, "learning_rate": 5.117769785096771e-07, "loss": 0.0178, "step": 2897 }, { "epoch": 1.8, "learning_rate": 5.086008117488795e-07, "loss": 0.016, "step": 2898 }, { "epoch": 1.8, "learning_rate": 5.054342743874386e-07, "loss": 0.0184, "step": 2899 }, { "epoch": 1.8, "learning_rate": 5.022773696379313e-07, "loss": 0.0112, "step": 2900 }, { "epoch": 1.8, "learning_rate": 4.991301007031623e-07, "loss": 0.0124, "step": 2901 }, { "epoch": 1.8, "learning_rate": 4.959924707761621e-07, "loss": 0.0103, "step": 2902 }, { "epoch": 1.81, "learning_rate": 4.928644830401796e-07, "loss": 0.0061, "step": 2903 }, { "epoch": 1.81, "learning_rate": 4.897461406686821e-07, "loss": 0.0157, "step": 2904 }, { "epoch": 1.81, "learning_rate": 4.866374468253487e-07, "loss": 0.013, "step": 2905 }, { "epoch": 1.81, "learning_rate": 4.835384046640734e-07, "loss": 0.0085, "step": 2906 }, { "epoch": 1.81, "learning_rate": 4.8044901732896e-07, "loss": 0.0094, "step": 2907 }, { "epoch": 1.81, "learning_rate": 4.773692879543091e-07, "loss": 0.015, "step": 2908 }, { "epoch": 1.81, "learning_rate": 4.742992196646301e-07, "loss": 0.0159, "step": 2909 }, { "epoch": 1.81, "learning_rate": 4.7123881557462834e-07, "loss": 0.0081, "step": 2910 }, { "epoch": 1.81, "learning_rate": 4.681880787892046e-07, "loss": 0.0126, "step": 2911 }, { "epoch": 1.81, "learning_rate": 4.651470124034518e-07, "loss": 0.0126, "step": 2912 }, { "epoch": 1.81, "learning_rate": 4.6211561950265217e-07, "loss": 0.0105, "step": 2913 }, { "epoch": 1.81, "learning_rate": 4.590939031622743e-07, "loss": 0.0113, "step": 2914 }, { "epoch": 1.81, "learning_rate": 4.560818664479671e-07, "loss": 0.0143, "step": 2915 }, { "epoch": 1.81, "learning_rate": 4.5307951241556067e-07, "loss": 0.0097, "step": 2916 }, { "epoch": 1.81, "learning_rate": 4.500868441110651e-07, "loss": 0.0151, "step": 2917 }, { "epoch": 1.81, "learning_rate": 4.47103864570656e-07, "loss": 0.013, "step": 2918 }, { "epoch": 1.82, "learning_rate": 4.4413057682068606e-07, "loss": 0.0102, "step": 2919 }, { "epoch": 1.82, "learning_rate": 4.411669838776733e-07, "loss": 0.0082, "step": 2920 }, { "epoch": 1.82, "learning_rate": 4.382130887483005e-07, "loss": 0.0113, "step": 2921 }, { "epoch": 1.82, "learning_rate": 4.3526889442940925e-07, "loss": 0.0073, "step": 2922 }, { "epoch": 1.82, "learning_rate": 4.323344039080024e-07, "loss": 0.0164, "step": 2923 }, { "epoch": 1.82, "learning_rate": 4.2940962016123524e-07, "loss": 0.0121, "step": 2924 }, { "epoch": 1.82, "learning_rate": 4.2649454615641737e-07, "loss": 0.0088, "step": 2925 }, { "epoch": 1.82, "learning_rate": 4.2358918485100765e-07, "loss": 0.0112, "step": 2926 }, { "epoch": 1.82, "learning_rate": 4.2069353919260594e-07, "loss": 0.0128, "step": 2927 }, { "epoch": 1.82, "learning_rate": 4.1780761211896226e-07, "loss": 0.0164, "step": 2928 }, { "epoch": 1.82, "learning_rate": 4.149314065579624e-07, "loss": 0.0108, "step": 2929 }, { "epoch": 1.82, "learning_rate": 4.120649254276321e-07, "loss": 0.0125, "step": 2930 }, { "epoch": 1.82, "learning_rate": 4.0920817163612845e-07, "loss": 0.0143, "step": 2931 }, { "epoch": 1.82, "learning_rate": 4.063611480817409e-07, "loss": 0.0126, "step": 2932 }, { "epoch": 1.82, "learning_rate": 4.0352385765288883e-07, "loss": 0.0097, "step": 2933 }, { "epoch": 1.82, "learning_rate": 4.0069630322811303e-07, "loss": 0.0105, "step": 2934 }, { "epoch": 1.83, "learning_rate": 3.9787848767607997e-07, "loss": 0.0121, "step": 2935 }, { "epoch": 1.83, "learning_rate": 3.950704138555772e-07, "loss": 0.0094, "step": 2936 }, { "epoch": 1.83, "learning_rate": 3.9227208461550146e-07, "loss": 0.0118, "step": 2937 }, { "epoch": 1.83, "learning_rate": 3.894835027948718e-07, "loss": 0.0143, "step": 2938 }, { "epoch": 1.83, "learning_rate": 3.867046712228162e-07, "loss": 0.0103, "step": 2939 }, { "epoch": 1.83, "learning_rate": 3.839355927185662e-07, "loss": 0.0185, "step": 2940 }, { "epoch": 1.83, "learning_rate": 3.811762700914612e-07, "loss": 0.0147, "step": 2941 }, { "epoch": 1.83, "learning_rate": 3.784267061409452e-07, "loss": 0.0082, "step": 2942 }, { "epoch": 1.83, "learning_rate": 3.7568690365655693e-07, "loss": 0.0114, "step": 2943 }, { "epoch": 1.83, "learning_rate": 3.729568654179361e-07, "loss": 0.0147, "step": 2944 }, { "epoch": 1.83, "learning_rate": 3.7023659419481275e-07, "loss": 0.0143, "step": 2945 }, { "epoch": 1.83, "learning_rate": 3.6752609274701145e-07, "loss": 0.0093, "step": 2946 }, { "epoch": 1.83, "learning_rate": 3.6482536382444033e-07, "loss": 0.0082, "step": 2947 }, { "epoch": 1.83, "learning_rate": 3.621344101670976e-07, "loss": 0.0156, "step": 2948 }, { "epoch": 1.83, "learning_rate": 3.5945323450506387e-07, "loss": 0.0061, "step": 2949 }, { "epoch": 1.83, "learning_rate": 3.5678183955849323e-07, "loss": 0.0107, "step": 2950 }, { "epoch": 1.84, "learning_rate": 3.541202280376232e-07, "loss": 0.0087, "step": 2951 }, { "epoch": 1.84, "learning_rate": 3.5146840264276617e-07, "loss": 0.0137, "step": 2952 }, { "epoch": 1.84, "learning_rate": 3.488263660642999e-07, "loss": 0.0165, "step": 2953 }, { "epoch": 1.84, "learning_rate": 3.4619412098267693e-07, "loss": 0.0075, "step": 2954 }, { "epoch": 1.84, "learning_rate": 3.435716700684133e-07, "loss": 0.0062, "step": 2955 }, { "epoch": 1.84, "learning_rate": 3.409590159820897e-07, "loss": 0.0119, "step": 2956 }, { "epoch": 1.84, "learning_rate": 3.383561613743469e-07, "loss": 0.0126, "step": 2957 }, { "epoch": 1.84, "learning_rate": 3.357631088858837e-07, "loss": 0.011, "step": 2958 }, { "epoch": 1.84, "learning_rate": 3.331798611474535e-07, "loss": 0.0091, "step": 2959 }, { "epoch": 1.84, "learning_rate": 3.3060642077986313e-07, "loss": 0.0117, "step": 2960 }, { "epoch": 1.84, "learning_rate": 3.2804279039397093e-07, "loss": 0.0112, "step": 2961 }, { "epoch": 1.84, "learning_rate": 3.2548897259067844e-07, "loss": 0.0101, "step": 2962 }, { "epoch": 1.84, "learning_rate": 3.2294496996093773e-07, "loss": 0.0155, "step": 2963 }, { "epoch": 1.84, "learning_rate": 3.204107850857374e-07, "loss": 0.0152, "step": 2964 }, { "epoch": 1.84, "learning_rate": 3.1788642053610984e-07, "loss": 0.0096, "step": 2965 }, { "epoch": 1.84, "learning_rate": 3.15371878873123e-07, "loss": 0.0074, "step": 2966 }, { "epoch": 1.85, "learning_rate": 3.1286716264787633e-07, "loss": 0.0075, "step": 2967 }, { "epoch": 1.85, "learning_rate": 3.10372274401507e-07, "loss": 0.0111, "step": 2968 }, { "epoch": 1.85, "learning_rate": 3.0788721666517365e-07, "loss": 0.0196, "step": 2969 }, { "epoch": 1.85, "learning_rate": 3.054119919600673e-07, "loss": 0.0165, "step": 2970 }, { "epoch": 1.85, "learning_rate": 3.0294660279740376e-07, "loss": 0.0145, "step": 2971 }, { "epoch": 1.85, "learning_rate": 3.004910516784121e-07, "loss": 0.0094, "step": 2972 }, { "epoch": 1.85, "learning_rate": 2.9804534109434956e-07, "loss": 0.0147, "step": 2973 }, { "epoch": 1.85, "learning_rate": 2.9560947352648697e-07, "loss": 0.0132, "step": 2974 }, { "epoch": 1.85, "learning_rate": 2.9318345144610627e-07, "loss": 0.0118, "step": 2975 }, { "epoch": 1.85, "learning_rate": 2.9076727731450203e-07, "loss": 0.0125, "step": 2976 }, { "epoch": 1.85, "learning_rate": 2.883609535829812e-07, "loss": 0.014, "step": 2977 }, { "epoch": 1.85, "learning_rate": 2.8596448269285205e-07, "loss": 0.0116, "step": 2978 }, { "epoch": 1.85, "learning_rate": 2.8357786707542854e-07, "loss": 0.0084, "step": 2979 }, { "epoch": 1.85, "learning_rate": 2.8120110915202945e-07, "loss": 0.0175, "step": 2980 }, { "epoch": 1.85, "learning_rate": 2.788342113339648e-07, "loss": 0.0128, "step": 2981 }, { "epoch": 1.85, "learning_rate": 2.764771760225493e-07, "loss": 0.0129, "step": 2982 }, { "epoch": 1.86, "learning_rate": 2.74130005609089e-07, "loss": 0.0185, "step": 2983 }, { "epoch": 1.86, "learning_rate": 2.71792702474879e-07, "loss": 0.0112, "step": 2984 }, { "epoch": 1.86, "learning_rate": 2.6946526899120693e-07, "loss": 0.0135, "step": 2985 }, { "epoch": 1.86, "learning_rate": 2.6714770751934406e-07, "loss": 0.0118, "step": 2986 }, { "epoch": 1.86, "learning_rate": 2.648400204105517e-07, "loss": 0.016, "step": 2987 }, { "epoch": 1.86, "learning_rate": 2.6254221000606485e-07, "loss": 0.0111, "step": 2988 }, { "epoch": 1.86, "learning_rate": 2.602542786371065e-07, "loss": 0.0112, "step": 2989 }, { "epoch": 1.86, "learning_rate": 2.5797622862487104e-07, "loss": 0.0126, "step": 2990 }, { "epoch": 1.86, "learning_rate": 2.5570806228053193e-07, "loss": 0.0095, "step": 2991 }, { "epoch": 1.86, "learning_rate": 2.5344978190523283e-07, "loss": 0.0104, "step": 2992 }, { "epoch": 1.86, "learning_rate": 2.5120138979009003e-07, "loss": 0.011, "step": 2993 }, { "epoch": 1.86, "learning_rate": 2.489628882161832e-07, "loss": 0.0113, "step": 2994 }, { "epoch": 1.86, "learning_rate": 2.467342794545613e-07, "loss": 0.018, "step": 2995 }, { "epoch": 1.86, "learning_rate": 2.4451556576623794e-07, "loss": 0.0109, "step": 2996 }, { "epoch": 1.86, "learning_rate": 2.423067494021847e-07, "loss": 0.0156, "step": 2997 }, { "epoch": 1.86, "learning_rate": 2.4010783260333127e-07, "loss": 0.0141, "step": 2998 }, { "epoch": 1.87, "learning_rate": 2.3791881760056756e-07, "loss": 0.0092, "step": 2999 }, { "epoch": 1.87, "learning_rate": 2.35739706614736e-07, "loss": 0.0168, "step": 3000 }, { "epoch": 1.87, "learning_rate": 2.335705018566292e-07, "loss": 0.0076, "step": 3001 }, { "epoch": 1.87, "learning_rate": 2.3141120552699236e-07, "loss": 0.0103, "step": 3002 }, { "epoch": 1.87, "learning_rate": 2.292618198165164e-07, "loss": 0.0123, "step": 3003 }, { "epoch": 1.87, "learning_rate": 2.2712234690583813e-07, "loss": 0.0126, "step": 3004 }, { "epoch": 1.87, "learning_rate": 2.2499278896553678e-07, "loss": 0.0151, "step": 3005 }, { "epoch": 1.87, "learning_rate": 2.2287314815613524e-07, "loss": 0.0145, "step": 3006 }, { "epoch": 1.87, "learning_rate": 2.2076342662809003e-07, "loss": 0.0098, "step": 3007 }, { "epoch": 1.87, "learning_rate": 2.1866362652179673e-07, "loss": 0.0136, "step": 3008 }, { "epoch": 1.87, "learning_rate": 2.1657374996758795e-07, "loss": 0.0132, "step": 3009 }, { "epoch": 1.87, "learning_rate": 2.1449379908572431e-07, "loss": 0.0076, "step": 3010 }, { "epoch": 1.87, "learning_rate": 2.1242377598639896e-07, "loss": 0.0119, "step": 3011 }, { "epoch": 1.87, "learning_rate": 2.1036368276973195e-07, "loss": 0.0192, "step": 3012 }, { "epoch": 1.87, "learning_rate": 2.0831352152576924e-07, "loss": 0.0072, "step": 3013 }, { "epoch": 1.87, "learning_rate": 2.0627329433447917e-07, "loss": 0.0121, "step": 3014 }, { "epoch": 1.88, "learning_rate": 2.0424300326575497e-07, "loss": 0.013, "step": 3015 }, { "epoch": 1.88, "learning_rate": 2.022226503794056e-07, "loss": 0.0157, "step": 3016 }, { "epoch": 1.88, "learning_rate": 2.0021223772515807e-07, "loss": 0.0126, "step": 3017 }, { "epoch": 1.88, "learning_rate": 1.9821176734265757e-07, "loss": 0.0096, "step": 3018 }, { "epoch": 1.88, "learning_rate": 1.9622124126145837e-07, "loss": 0.0112, "step": 3019 }, { "epoch": 1.88, "learning_rate": 1.942406615010295e-07, "loss": 0.0112, "step": 3020 }, { "epoch": 1.88, "learning_rate": 1.9227003007074696e-07, "loss": 0.0129, "step": 3021 }, { "epoch": 1.88, "learning_rate": 1.9030934896989373e-07, "loss": 0.0058, "step": 3022 }, { "epoch": 1.88, "learning_rate": 1.883586201876575e-07, "loss": 0.0103, "step": 3023 }, { "epoch": 1.88, "learning_rate": 1.864178457031318e-07, "loss": 0.0136, "step": 3024 }, { "epoch": 1.88, "learning_rate": 1.8448702748530723e-07, "loss": 0.0126, "step": 3025 }, { "epoch": 1.88, "learning_rate": 1.8256616749307455e-07, "loss": 0.0086, "step": 3026 }, { "epoch": 1.88, "learning_rate": 1.8065526767522268e-07, "loss": 0.0075, "step": 3027 }, { "epoch": 1.88, "learning_rate": 1.787543299704375e-07, "loss": 0.0085, "step": 3028 }, { "epoch": 1.88, "learning_rate": 1.768633563072919e-07, "loss": 0.0117, "step": 3029 }, { "epoch": 1.88, "learning_rate": 1.7498234860425567e-07, "loss": 0.0129, "step": 3030 }, { "epoch": 1.88, "learning_rate": 1.7311130876968452e-07, "loss": 0.0109, "step": 3031 }, { "epoch": 1.89, "learning_rate": 1.712502387018211e-07, "loss": 0.0109, "step": 3032 }, { "epoch": 1.89, "learning_rate": 1.6939914028879623e-07, "loss": 0.0101, "step": 3033 }, { "epoch": 1.89, "learning_rate": 1.6755801540862092e-07, "loss": 0.0142, "step": 3034 }, { "epoch": 1.89, "learning_rate": 1.6572686592919107e-07, "loss": 0.0145, "step": 3035 }, { "epoch": 1.89, "learning_rate": 1.6390569370827835e-07, "loss": 0.0095, "step": 3036 }, { "epoch": 1.89, "learning_rate": 1.6209450059353482e-07, "loss": 0.0104, "step": 3037 }, { "epoch": 1.89, "learning_rate": 1.6029328842248725e-07, "loss": 0.0101, "step": 3038 }, { "epoch": 1.89, "learning_rate": 1.5850205902253613e-07, "loss": 0.0149, "step": 3039 }, { "epoch": 1.89, "learning_rate": 1.5672081421095552e-07, "loss": 0.0135, "step": 3040 }, { "epoch": 1.89, "learning_rate": 1.549495557948877e-07, "loss": 0.0135, "step": 3041 }, { "epoch": 1.89, "learning_rate": 1.53188285571344e-07, "loss": 0.0137, "step": 3042 }, { "epoch": 1.89, "learning_rate": 1.5143700532720295e-07, "loss": 0.0058, "step": 3043 }, { "epoch": 1.89, "learning_rate": 1.4969571683920768e-07, "loss": 0.0123, "step": 3044 }, { "epoch": 1.89, "learning_rate": 1.4796442187396397e-07, "loss": 0.0103, "step": 3045 }, { "epoch": 1.89, "learning_rate": 1.4624312218793902e-07, "loss": 0.0098, "step": 3046 }, { "epoch": 1.89, "learning_rate": 1.4453181952746144e-07, "loss": 0.0088, "step": 3047 }, { "epoch": 1.9, "learning_rate": 1.4283051562871132e-07, "loss": 0.0123, "step": 3048 }, { "epoch": 1.9, "learning_rate": 1.411392122177302e-07, "loss": 0.0083, "step": 3049 }, { "epoch": 1.9, "learning_rate": 1.394579110104144e-07, "loss": 0.0124, "step": 3050 }, { "epoch": 1.9, "learning_rate": 1.3778661371250835e-07, "loss": 0.0132, "step": 3051 }, { "epoch": 1.9, "learning_rate": 1.3612532201960904e-07, "loss": 0.0113, "step": 3052 }, { "epoch": 1.9, "learning_rate": 1.344740376171627e-07, "loss": 0.0112, "step": 3053 }, { "epoch": 1.9, "learning_rate": 1.3283276218046259e-07, "loss": 0.0077, "step": 3054 }, { "epoch": 1.9, "learning_rate": 1.3120149737464893e-07, "loss": 0.0071, "step": 3055 }, { "epoch": 1.9, "learning_rate": 1.2958024485470234e-07, "loss": 0.0084, "step": 3056 }, { "epoch": 1.9, "learning_rate": 1.2796900626544818e-07, "loss": 0.0119, "step": 3057 }, { "epoch": 1.9, "learning_rate": 1.263677832415522e-07, "loss": 0.0107, "step": 3058 }, { "epoch": 1.9, "learning_rate": 1.2477657740751714e-07, "loss": 0.0106, "step": 3059 }, { "epoch": 1.9, "learning_rate": 1.2319539037768614e-07, "loss": 0.0123, "step": 3060 }, { "epoch": 1.9, "learning_rate": 1.2162422375623373e-07, "loss": 0.014, "step": 3061 }, { "epoch": 1.9, "learning_rate": 1.2006307913717154e-07, "loss": 0.0162, "step": 3062 }, { "epoch": 1.9, "learning_rate": 1.1851195810434369e-07, "loss": 0.0166, "step": 3063 }, { "epoch": 1.91, "learning_rate": 1.169708622314214e-07, "loss": 0.0127, "step": 3064 }, { "epoch": 1.91, "learning_rate": 1.1543979308190844e-07, "loss": 0.0105, "step": 3065 }, { "epoch": 1.91, "learning_rate": 1.1391875220913452e-07, "loss": 0.012, "step": 3066 }, { "epoch": 1.91, "learning_rate": 1.1240774115625631e-07, "loss": 0.0133, "step": 3067 }, { "epoch": 1.91, "learning_rate": 1.1090676145625423e-07, "loss": 0.0067, "step": 3068 }, { "epoch": 1.91, "learning_rate": 1.0941581463193129e-07, "loss": 0.0177, "step": 3069 }, { "epoch": 1.91, "learning_rate": 1.0793490219591085e-07, "loss": 0.0095, "step": 3070 }, { "epoch": 1.91, "learning_rate": 1.0646402565063774e-07, "loss": 0.0121, "step": 3071 }, { "epoch": 1.91, "learning_rate": 1.0500318648837604e-07, "loss": 0.0069, "step": 3072 }, { "epoch": 1.91, "learning_rate": 1.0355238619120245e-07, "loss": 0.0187, "step": 3073 }, { "epoch": 1.91, "learning_rate": 1.021116262310129e-07, "loss": 0.0159, "step": 3074 }, { "epoch": 1.91, "learning_rate": 1.0068090806951369e-07, "loss": 0.0105, "step": 3075 }, { "epoch": 1.91, "learning_rate": 9.926023315822597e-08, "loss": 0.0085, "step": 3076 }, { "epoch": 1.91, "learning_rate": 9.7849602938479e-08, "loss": 0.0159, "step": 3077 }, { "epoch": 1.91, "learning_rate": 9.644901884141466e-08, "loss": 0.0148, "step": 3078 }, { "epoch": 1.91, "learning_rate": 9.505848228798076e-08, "loss": 0.0168, "step": 3079 }, { "epoch": 1.92, "learning_rate": 9.367799468893102e-08, "loss": 0.0109, "step": 3080 }, { "epoch": 1.92, "learning_rate": 9.2307557444824e-08, "loss": 0.0116, "step": 3081 }, { "epoch": 1.92, "learning_rate": 9.09471719460242e-08, "loss": 0.0116, "step": 3082 }, { "epoch": 1.92, "learning_rate": 8.959683957269428e-08, "loss": 0.0152, "step": 3083 }, { "epoch": 1.92, "learning_rate": 8.825656169480056e-08, "loss": 0.0126, "step": 3084 }, { "epoch": 1.92, "learning_rate": 8.692633967210872e-08, "loss": 0.0224, "step": 3085 }, { "epoch": 1.92, "learning_rate": 8.560617485418143e-08, "loss": 0.0088, "step": 3086 }, { "epoch": 1.92, "learning_rate": 8.42960685803762e-08, "loss": 0.0095, "step": 3087 }, { "epoch": 1.92, "learning_rate": 8.299602217984982e-08, "loss": 0.0131, "step": 3088 }, { "epoch": 1.92, "learning_rate": 8.170603697154944e-08, "loss": 0.0143, "step": 3089 }, { "epoch": 1.92, "learning_rate": 8.042611426421598e-08, "loss": 0.0111, "step": 3090 }, { "epoch": 1.92, "learning_rate": 7.915625535638294e-08, "loss": 0.0159, "step": 3091 }, { "epoch": 1.92, "learning_rate": 7.789646153637087e-08, "loss": 0.0117, "step": 3092 }, { "epoch": 1.92, "learning_rate": 7.664673408229073e-08, "loss": 0.0133, "step": 3093 }, { "epoch": 1.92, "learning_rate": 7.540707426204163e-08, "loss": 0.0114, "step": 3094 }, { "epoch": 1.92, "learning_rate": 7.417748333330533e-08, "loss": 0.0065, "step": 3095 }, { "epoch": 1.93, "learning_rate": 7.29579625435517e-08, "loss": 0.0138, "step": 3096 }, { "epoch": 1.93, "learning_rate": 7.174851313003105e-08, "loss": 0.0083, "step": 3097 }, { "epoch": 1.93, "learning_rate": 7.054913631978078e-08, "loss": 0.0176, "step": 3098 }, { "epoch": 1.93, "learning_rate": 6.935983332961305e-08, "loss": 0.0162, "step": 3099 }, { "epoch": 1.93, "learning_rate": 6.818060536612381e-08, "loss": 0.0079, "step": 3100 }, { "epoch": 1.93, "learning_rate": 6.701145362568829e-08, "loss": 0.0207, "step": 3101 }, { "epoch": 1.93, "learning_rate": 6.58523792944532e-08, "loss": 0.0131, "step": 3102 }, { "epoch": 1.93, "learning_rate": 6.470338354834793e-08, "loss": 0.014, "step": 3103 }, { "epoch": 1.93, "learning_rate": 6.356446755307444e-08, "loss": 0.015, "step": 3104 }, { "epoch": 1.93, "learning_rate": 6.243563246410622e-08, "loss": 0.0139, "step": 3105 }, { "epoch": 1.93, "learning_rate": 6.131687942669051e-08, "loss": 0.0119, "step": 3106 }, { "epoch": 1.93, "learning_rate": 6.020820957584939e-08, "loss": 0.0118, "step": 3107 }, { "epoch": 1.93, "learning_rate": 5.910962403636977e-08, "loss": 0.0077, "step": 3108 }, { "epoch": 1.93, "learning_rate": 5.802112392281123e-08, "loss": 0.0106, "step": 3109 }, { "epoch": 1.93, "learning_rate": 5.694271033950038e-08, "loss": 0.0161, "step": 3110 }, { "epoch": 1.93, "learning_rate": 5.587438438052872e-08, "loss": 0.0134, "step": 3111 }, { "epoch": 1.94, "learning_rate": 5.4816147129757027e-08, "loss": 0.0128, "step": 3112 }, { "epoch": 1.94, "learning_rate": 5.376799966080759e-08, "loss": 0.0138, "step": 3113 }, { "epoch": 1.94, "learning_rate": 5.272994303706758e-08, "loss": 0.012, "step": 3114 }, { "epoch": 1.94, "learning_rate": 5.170197831168677e-08, "loss": 0.0121, "step": 3115 }, { "epoch": 1.94, "learning_rate": 5.0684106527575384e-08, "loss": 0.0123, "step": 3116 }, { "epoch": 1.94, "learning_rate": 4.967632871740624e-08, "loss": 0.0098, "step": 3117 }, { "epoch": 1.94, "learning_rate": 4.8678645903608155e-08, "loss": 0.0224, "step": 3118 }, { "epoch": 1.94, "learning_rate": 4.769105909836924e-08, "loss": 0.0107, "step": 3119 }, { "epoch": 1.94, "learning_rate": 4.671356930363691e-08, "loss": 0.0138, "step": 3120 }, { "epoch": 1.94, "learning_rate": 4.5746177511112325e-08, "loss": 0.0111, "step": 3121 }, { "epoch": 1.94, "learning_rate": 4.4788884702252624e-08, "loss": 0.0149, "step": 3122 }, { "epoch": 1.94, "learning_rate": 4.384169184826981e-08, "loss": 0.0162, "step": 3123 }, { "epoch": 1.94, "learning_rate": 4.2904599910127406e-08, "loss": 0.0121, "step": 3124 }, { "epoch": 1.94, "learning_rate": 4.197760983854271e-08, "loss": 0.0131, "step": 3125 }, { "epoch": 1.94, "learning_rate": 4.106072257398452e-08, "loss": 0.0116, "step": 3126 }, { "epoch": 1.94, "learning_rate": 4.015393904666987e-08, "loss": 0.0124, "step": 3127 }, { "epoch": 1.95, "learning_rate": 3.925726017656728e-08, "loss": 0.0087, "step": 3128 }, { "epoch": 1.95, "learning_rate": 3.837068687339351e-08, "loss": 0.0083, "step": 3129 }, { "epoch": 1.95, "learning_rate": 3.749422003661018e-08, "loss": 0.0103, "step": 3130 }, { "epoch": 1.95, "learning_rate": 3.662786055542822e-08, "loss": 0.0094, "step": 3131 }, { "epoch": 1.95, "learning_rate": 3.5771609308804524e-08, "loss": 0.0147, "step": 3132 }, { "epoch": 1.95, "learning_rate": 3.4925467165439765e-08, "loss": 0.0156, "step": 3133 }, { "epoch": 1.95, "learning_rate": 3.408943498377726e-08, "loss": 0.0122, "step": 3134 }, { "epoch": 1.95, "learning_rate": 3.3263513612006305e-08, "loss": 0.0125, "step": 3135 }, { "epoch": 1.95, "learning_rate": 3.2447703888055516e-08, "loss": 0.0104, "step": 3136 }, { "epoch": 1.95, "learning_rate": 3.164200663959727e-08, "loss": 0.0093, "step": 3137 }, { "epoch": 1.95, "learning_rate": 3.084642268404214e-08, "loss": 0.0083, "step": 3138 }, { "epoch": 1.95, "learning_rate": 3.006095282854116e-08, "loss": 0.0179, "step": 3139 }, { "epoch": 1.95, "learning_rate": 2.928559786998575e-08, "loss": 0.0088, "step": 3140 }, { "epoch": 1.95, "learning_rate": 2.852035859500335e-08, "loss": 0.0117, "step": 3141 }, { "epoch": 1.95, "learning_rate": 2.7765235779959597e-08, "loss": 0.018, "step": 3142 }, { "epoch": 1.95, "learning_rate": 2.702023019095612e-08, "loss": 0.0147, "step": 3143 }, { "epoch": 1.96, "learning_rate": 2.628534258383164e-08, "loss": 0.0152, "step": 3144 }, { "epoch": 1.96, "learning_rate": 2.5560573704157543e-08, "loss": 0.008, "step": 3145 }, { "epoch": 1.96, "learning_rate": 2.4845924287241196e-08, "loss": 0.0117, "step": 3146 }, { "epoch": 1.96, "learning_rate": 2.414139505812152e-08, "loss": 0.0103, "step": 3147 }, { "epoch": 1.96, "learning_rate": 2.3446986731572308e-08, "loss": 0.0145, "step": 3148 }, { "epoch": 1.96, "learning_rate": 2.2762700012097795e-08, "loss": 0.0158, "step": 3149 }, { "epoch": 1.96, "learning_rate": 2.2088535593933756e-08, "loss": 0.0123, "step": 3150 }, { "epoch": 1.96, "learning_rate": 2.1424494161047528e-08, "loss": 0.0117, "step": 3151 }, { "epoch": 1.96, "learning_rate": 2.0770576387134643e-08, "loss": 0.0085, "step": 3152 }, { "epoch": 1.96, "learning_rate": 2.012678293562109e-08, "loss": 0.0066, "step": 3153 }, { "epoch": 1.96, "learning_rate": 1.9493114459659956e-08, "loss": 0.0181, "step": 3154 }, { "epoch": 1.96, "learning_rate": 1.8869571602132543e-08, "loss": 0.0118, "step": 3155 }, { "epoch": 1.96, "learning_rate": 1.8256154995649478e-08, "loss": 0.0114, "step": 3156 }, { "epoch": 1.96, "learning_rate": 1.7652865262544062e-08, "loss": 0.0091, "step": 3157 }, { "epoch": 1.96, "learning_rate": 1.7059703014878914e-08, "loss": 0.0103, "step": 3158 }, { "epoch": 1.96, "learning_rate": 1.6476668854440435e-08, "loss": 0.0173, "step": 3159 }, { "epoch": 1.97, "learning_rate": 1.59037633727388e-08, "loss": 0.0092, "step": 3160 }, { "epoch": 1.97, "learning_rate": 1.5340987151011288e-08, "loss": 0.0135, "step": 3161 }, { "epoch": 1.97, "learning_rate": 1.4788340760214514e-08, "loss": 0.008, "step": 3162 }, { "epoch": 1.97, "learning_rate": 1.424582476103109e-08, "loss": 0.0112, "step": 3163 }, { "epoch": 1.97, "learning_rate": 1.3713439703865183e-08, "loss": 0.0126, "step": 3164 }, { "epoch": 1.97, "learning_rate": 1.3191186128841404e-08, "loss": 0.0146, "step": 3165 }, { "epoch": 1.97, "learning_rate": 1.2679064565807032e-08, "loss": 0.0119, "step": 3166 }, { "epoch": 1.97, "learning_rate": 1.2177075534330895e-08, "loss": 0.0096, "step": 3167 }, { "epoch": 1.97, "learning_rate": 1.1685219543698945e-08, "loss": 0.0119, "step": 3168 }, { "epoch": 1.97, "learning_rate": 1.120349709291868e-08, "loss": 0.0102, "step": 3169 }, { "epoch": 1.97, "learning_rate": 1.073190867071805e-08, "loss": 0.0084, "step": 3170 }, { "epoch": 1.97, "learning_rate": 1.0270454755541004e-08, "loss": 0.0161, "step": 3171 }, { "epoch": 1.97, "learning_rate": 9.819135815549719e-09, "loss": 0.0092, "step": 3172 }, { "epoch": 1.97, "learning_rate": 9.377952308627925e-09, "loss": 0.0191, "step": 3173 }, { "epoch": 1.97, "learning_rate": 8.946904682370917e-09, "loss": 0.0117, "step": 3174 }, { "epoch": 1.97, "learning_rate": 8.525993374095543e-09, "loss": 0.012, "step": 3175 }, { "epoch": 1.98, "learning_rate": 8.115218810831327e-09, "loss": 0.0085, "step": 3176 }, { "epoch": 1.98, "learning_rate": 7.714581409326017e-09, "loss": 0.0084, "step": 3177 }, { "epoch": 1.98, "learning_rate": 7.3240815760433625e-09, "loss": 0.0103, "step": 3178 }, { "epoch": 1.98, "learning_rate": 6.943719707158681e-09, "loss": 0.0157, "step": 3179 }, { "epoch": 1.98, "learning_rate": 6.573496188565509e-09, "loss": 0.0102, "step": 3180 }, { "epoch": 1.98, "learning_rate": 6.213411395868951e-09, "loss": 0.0092, "step": 3181 }, { "epoch": 1.98, "learning_rate": 5.863465694390114e-09, "loss": 0.0097, "step": 3182 }, { "epoch": 1.98, "learning_rate": 5.523659439163886e-09, "loss": 0.0079, "step": 3183 }, { "epoch": 1.98, "learning_rate": 5.193992974935613e-09, "loss": 0.0176, "step": 3184 }, { "epoch": 1.98, "learning_rate": 4.874466636164421e-09, "loss": 0.0129, "step": 3185 }, { "epoch": 1.98, "learning_rate": 4.565080747023221e-09, "loss": 0.0108, "step": 3186 }, { "epoch": 1.98, "learning_rate": 4.265835621397596e-09, "loss": 0.017, "step": 3187 }, { "epoch": 1.98, "learning_rate": 3.976731562881364e-09, "loss": 0.0138, "step": 3188 }, { "epoch": 1.98, "learning_rate": 3.697768864782125e-09, "loss": 0.0171, "step": 3189 }, { "epoch": 1.98, "learning_rate": 3.4289478101201536e-09, "loss": 0.0121, "step": 3190 }, { "epoch": 1.98, "learning_rate": 3.170268671623955e-09, "loss": 0.0139, "step": 3191 }, { "epoch": 1.99, "learning_rate": 2.92173171173471e-09, "loss": 0.0137, "step": 3192 }, { "epoch": 1.99, "learning_rate": 2.6833371826018307e-09, "loss": 0.0122, "step": 3193 }, { "epoch": 1.99, "learning_rate": 2.4550853260851826e-09, "loss": 0.0177, "step": 3194 }, { "epoch": 1.99, "learning_rate": 2.2369763737573046e-09, "loss": 0.0175, "step": 3195 }, { "epoch": 1.99, "learning_rate": 2.0290105468978584e-09, "loss": 0.0116, "step": 3196 }, { "epoch": 1.99, "learning_rate": 1.8311880564958472e-09, "loss": 0.012, "step": 3197 }, { "epoch": 1.99, "learning_rate": 1.6435091032496187e-09, "loss": 0.0083, "step": 3198 }, { "epoch": 1.99, "learning_rate": 1.4659738775679721e-09, "loss": 0.0124, "step": 3199 }, { "epoch": 1.99, "learning_rate": 1.29858255956683e-09, "loss": 0.0098, "step": 3200 }, { "epoch": 1.99, "learning_rate": 1.141335319071457e-09, "loss": 0.0125, "step": 3201 }, { "epoch": 1.99, "learning_rate": 9.94232315615351e-10, "loss": 0.0106, "step": 3202 }, { "epoch": 1.99, "learning_rate": 8.572736984402419e-10, "loss": 0.0079, "step": 3203 }, { "epoch": 1.99, "learning_rate": 7.30459606494982e-10, "loss": 0.0146, "step": 3204 }, { "epoch": 1.99, "learning_rate": 6.137901684399871e-10, "loss": 0.018, "step": 3205 }, { "epoch": 1.99, "learning_rate": 5.072655026394646e-10, "loss": 0.0103, "step": 3206 }, { "epoch": 1.99, "learning_rate": 4.1088571716585425e-10, "loss": 0.0168, "step": 3207 }, { "epoch": 2.0, "learning_rate": 3.2465090980204894e-10, "loss": 0.013, "step": 3208 }, { "epoch": 2.0, "learning_rate": 2.4856116803695375e-10, "loss": 0.0191, "step": 3209 }, { "epoch": 2.0, "learning_rate": 1.8261656906437553e-10, "loss": 0.0107, "step": 3210 }, { "epoch": 2.0, "learning_rate": 1.2681717978968445e-10, "loss": 0.0057, "step": 3211 }, { "epoch": 2.0, "learning_rate": 8.11630568231525e-11, "loss": 0.0174, "step": 3212 }, { "epoch": 2.0, "learning_rate": 4.565424648106387e-11, "loss": 0.0155, "step": 3213 }, { "epoch": 2.0, "learning_rate": 2.0290784791265893e-11, "loss": 0.0056, "step": 3214 }, { "epoch": 2.0, "learning_rate": 5.0726974842874035e-12, "loss": 0.0166, "step": 3215 }, { "epoch": 2.0, "learning_rate": 0.0, "loss": 0.0114, "step": 3216 }, { "epoch": 2.0, "step": 3216, "total_flos": 163648376930304.0, "train_loss": 0.031938219622944006, "train_runtime": 43898.5386, "train_samples_per_second": 8.201, "train_steps_per_second": 0.073 } ], "logging_steps": 1.0, "max_steps": 3216, "num_input_tokens_seen": 0, "num_train_epochs": 2, "save_steps": 10000, "total_flos": 163648376930304.0, "train_batch_size": 16, "trial_name": null, "trial_params": null }