{ "best_metric": null, "best_model_checkpoint": null, "epoch": 2.0, "global_step": 2028, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 3.278688524590164e-07, "loss": 2.3379, "step": 1 }, { "epoch": 0.0, "learning_rate": 6.557377049180328e-07, "loss": 2.2812, "step": 2 }, { "epoch": 0.0, "learning_rate": 9.836065573770493e-07, "loss": 2.3184, "step": 3 }, { "epoch": 0.0, "learning_rate": 1.3114754098360657e-06, "loss": 2.2578, "step": 4 }, { "epoch": 0.0, "learning_rate": 1.6393442622950819e-06, "loss": 2.2383, "step": 5 }, { "epoch": 0.01, "learning_rate": 1.9672131147540985e-06, "loss": 2.1426, "step": 6 }, { "epoch": 0.01, "learning_rate": 2.295081967213115e-06, "loss": 1.7578, "step": 7 }, { "epoch": 0.01, "learning_rate": 2.6229508196721314e-06, "loss": 1.583, "step": 8 }, { "epoch": 0.01, "learning_rate": 2.9508196721311478e-06, "loss": 0.7607, "step": 9 }, { "epoch": 0.01, "learning_rate": 3.2786885245901638e-06, "loss": 0.7065, "step": 10 }, { "epoch": 0.01, "learning_rate": 3.6065573770491806e-06, "loss": 0.5913, "step": 11 }, { "epoch": 0.01, "learning_rate": 3.934426229508197e-06, "loss": 0.3, "step": 12 }, { "epoch": 0.01, "learning_rate": 4.2622950819672135e-06, "loss": 0.2467, "step": 13 }, { "epoch": 0.01, "learning_rate": 4.59016393442623e-06, "loss": 0.1715, "step": 14 }, { "epoch": 0.01, "learning_rate": 4.918032786885246e-06, "loss": 0.1462, "step": 15 }, { "epoch": 0.02, "learning_rate": 5.245901639344263e-06, "loss": 0.2114, "step": 16 }, { "epoch": 0.02, "learning_rate": 5.573770491803278e-06, "loss": 0.1772, "step": 17 }, { "epoch": 0.02, "learning_rate": 5.9016393442622956e-06, "loss": 0.1154, "step": 18 }, { "epoch": 0.02, "learning_rate": 6.229508196721312e-06, "loss": 0.1335, "step": 19 }, { "epoch": 0.02, "learning_rate": 6.5573770491803276e-06, "loss": 0.1257, "step": 20 }, { "epoch": 0.02, "learning_rate": 6.885245901639345e-06, "loss": 0.1159, "step": 21 }, { "epoch": 0.02, "learning_rate": 7.213114754098361e-06, "loss": 0.1118, "step": 22 }, { "epoch": 0.02, "learning_rate": 7.540983606557377e-06, "loss": 0.1097, "step": 23 }, { "epoch": 0.02, "learning_rate": 7.868852459016394e-06, "loss": 0.1077, "step": 24 }, { "epoch": 0.02, "learning_rate": 8.19672131147541e-06, "loss": 0.1033, "step": 25 }, { "epoch": 0.03, "learning_rate": 8.524590163934427e-06, "loss": 0.1043, "step": 26 }, { "epoch": 0.03, "learning_rate": 8.852459016393443e-06, "loss": 0.1125, "step": 27 }, { "epoch": 0.03, "learning_rate": 9.18032786885246e-06, "loss": 0.1064, "step": 28 }, { "epoch": 0.03, "learning_rate": 9.508196721311476e-06, "loss": 0.1105, "step": 29 }, { "epoch": 0.03, "learning_rate": 9.836065573770493e-06, "loss": 0.1046, "step": 30 }, { "epoch": 0.03, "learning_rate": 1.0163934426229509e-05, "loss": 0.1006, "step": 31 }, { "epoch": 0.03, "learning_rate": 1.0491803278688525e-05, "loss": 0.1026, "step": 32 }, { "epoch": 0.03, "learning_rate": 1.0819672131147544e-05, "loss": 0.0991, "step": 33 }, { "epoch": 0.03, "learning_rate": 1.1147540983606557e-05, "loss": 0.1032, "step": 34 }, { "epoch": 0.03, "learning_rate": 1.1475409836065575e-05, "loss": 0.1244, "step": 35 }, { "epoch": 0.04, "learning_rate": 1.1803278688524591e-05, "loss": 0.1017, "step": 36 }, { "epoch": 0.04, "learning_rate": 1.2131147540983608e-05, "loss": 0.1038, "step": 37 }, { "epoch": 0.04, "learning_rate": 1.2459016393442624e-05, "loss": 0.1038, "step": 38 }, { "epoch": 0.04, "learning_rate": 1.2786885245901642e-05, "loss": 0.1022, "step": 39 }, { "epoch": 0.04, "learning_rate": 1.3114754098360655e-05, "loss": 0.0973, "step": 40 }, { "epoch": 0.04, "learning_rate": 1.3442622950819673e-05, "loss": 0.1075, "step": 41 }, { "epoch": 0.04, "learning_rate": 1.377049180327869e-05, "loss": 0.101, "step": 42 }, { "epoch": 0.04, "learning_rate": 1.4098360655737706e-05, "loss": 0.0956, "step": 43 }, { "epoch": 0.04, "learning_rate": 1.4426229508196722e-05, "loss": 0.0958, "step": 44 }, { "epoch": 0.04, "learning_rate": 1.4754098360655739e-05, "loss": 0.0968, "step": 45 }, { "epoch": 0.05, "learning_rate": 1.5081967213114754e-05, "loss": 0.0921, "step": 46 }, { "epoch": 0.05, "learning_rate": 1.5409836065573772e-05, "loss": 0.0948, "step": 47 }, { "epoch": 0.05, "learning_rate": 1.5737704918032788e-05, "loss": 0.087, "step": 48 }, { "epoch": 0.05, "learning_rate": 1.6065573770491805e-05, "loss": 0.0918, "step": 49 }, { "epoch": 0.05, "learning_rate": 1.639344262295082e-05, "loss": 0.0928, "step": 50 }, { "epoch": 0.05, "learning_rate": 1.6721311475409837e-05, "loss": 0.088, "step": 51 }, { "epoch": 0.05, "learning_rate": 1.7049180327868854e-05, "loss": 0.0864, "step": 52 }, { "epoch": 0.05, "learning_rate": 1.737704918032787e-05, "loss": 0.1, "step": 53 }, { "epoch": 0.05, "learning_rate": 1.7704918032786887e-05, "loss": 0.0837, "step": 54 }, { "epoch": 0.05, "learning_rate": 1.8032786885245903e-05, "loss": 0.1064, "step": 55 }, { "epoch": 0.06, "learning_rate": 1.836065573770492e-05, "loss": 0.0945, "step": 56 }, { "epoch": 0.06, "learning_rate": 1.8688524590163936e-05, "loss": 0.0964, "step": 57 }, { "epoch": 0.06, "learning_rate": 1.9016393442622952e-05, "loss": 0.1091, "step": 58 }, { "epoch": 0.06, "learning_rate": 1.934426229508197e-05, "loss": 0.0839, "step": 59 }, { "epoch": 0.06, "learning_rate": 1.9672131147540985e-05, "loss": 0.088, "step": 60 }, { "epoch": 0.06, "learning_rate": 2e-05, "loss": 0.0983, "step": 61 }, { "epoch": 0.06, "learning_rate": 1.9999987245573438e-05, "loss": 0.0942, "step": 62 }, { "epoch": 0.06, "learning_rate": 1.9999948982326284e-05, "loss": 0.0954, "step": 63 }, { "epoch": 0.06, "learning_rate": 1.9999885210356147e-05, "loss": 0.1002, "step": 64 }, { "epoch": 0.06, "learning_rate": 1.99997959298257e-05, "loss": 0.0923, "step": 65 }, { "epoch": 0.07, "learning_rate": 1.9999681140962683e-05, "loss": 0.0855, "step": 66 }, { "epoch": 0.07, "learning_rate": 1.9999540844059916e-05, "loss": 0.0947, "step": 67 }, { "epoch": 0.07, "learning_rate": 1.9999375039475278e-05, "loss": 0.0881, "step": 68 }, { "epoch": 0.07, "learning_rate": 1.9999183727631714e-05, "loss": 0.101, "step": 69 }, { "epoch": 0.07, "learning_rate": 1.9998966909017246e-05, "loss": 0.0815, "step": 70 }, { "epoch": 0.07, "learning_rate": 1.9998724584184947e-05, "loss": 0.0944, "step": 71 }, { "epoch": 0.07, "learning_rate": 1.999845675375296e-05, "loss": 0.083, "step": 72 }, { "epoch": 0.07, "learning_rate": 1.9998163418404494e-05, "loss": 0.0942, "step": 73 }, { "epoch": 0.07, "learning_rate": 1.999784457888781e-05, "loss": 0.0794, "step": 74 }, { "epoch": 0.07, "learning_rate": 1.9997500236016233e-05, "loss": 0.0807, "step": 75 }, { "epoch": 0.07, "learning_rate": 1.9997130390668144e-05, "loss": 0.1028, "step": 76 }, { "epoch": 0.08, "learning_rate": 1.999673504378697e-05, "loss": 0.0795, "step": 77 }, { "epoch": 0.08, "learning_rate": 1.9996314196381203e-05, "loss": 0.082, "step": 78 }, { "epoch": 0.08, "learning_rate": 1.999586784952437e-05, "loss": 0.0885, "step": 79 }, { "epoch": 0.08, "learning_rate": 1.9995396004355055e-05, "loss": 0.0771, "step": 80 }, { "epoch": 0.08, "learning_rate": 1.9994898662076874e-05, "loss": 0.0779, "step": 81 }, { "epoch": 0.08, "learning_rate": 1.9994375823958504e-05, "loss": 0.0979, "step": 82 }, { "epoch": 0.08, "learning_rate": 1.999382749133363e-05, "loss": 0.0928, "step": 83 }, { "epoch": 0.08, "learning_rate": 1.9993253665600996e-05, "loss": 0.091, "step": 84 }, { "epoch": 0.08, "learning_rate": 1.999265434822436e-05, "loss": 0.0776, "step": 85 }, { "epoch": 0.08, "learning_rate": 1.9992029540732515e-05, "loss": 0.085, "step": 86 }, { "epoch": 0.09, "learning_rate": 1.999137924471927e-05, "loss": 0.0827, "step": 87 }, { "epoch": 0.09, "learning_rate": 1.999070346184346e-05, "loss": 0.0745, "step": 88 }, { "epoch": 0.09, "learning_rate": 1.9990002193828923e-05, "loss": 0.0859, "step": 89 }, { "epoch": 0.09, "learning_rate": 1.9989275442464523e-05, "loss": 0.0928, "step": 90 }, { "epoch": 0.09, "learning_rate": 1.9988523209604106e-05, "loss": 0.0804, "step": 91 }, { "epoch": 0.09, "learning_rate": 1.9987745497166546e-05, "loss": 0.0775, "step": 92 }, { "epoch": 0.09, "learning_rate": 1.998694230713569e-05, "loss": 0.076, "step": 93 }, { "epoch": 0.09, "learning_rate": 1.9986113641560387e-05, "loss": 0.0826, "step": 94 }, { "epoch": 0.09, "learning_rate": 1.9985259502554464e-05, "loss": 0.0791, "step": 95 }, { "epoch": 0.09, "learning_rate": 1.998437989229673e-05, "loss": 0.0814, "step": 96 }, { "epoch": 0.1, "learning_rate": 1.998347481303098e-05, "loss": 0.0756, "step": 97 }, { "epoch": 0.1, "learning_rate": 1.9982544267065962e-05, "loss": 0.0818, "step": 98 }, { "epoch": 0.1, "learning_rate": 1.9981588256775388e-05, "loss": 0.082, "step": 99 }, { "epoch": 0.1, "learning_rate": 1.9980606784597935e-05, "loss": 0.0859, "step": 100 }, { "epoch": 0.1, "learning_rate": 1.9979599853037223e-05, "loss": 0.0757, "step": 101 }, { "epoch": 0.1, "learning_rate": 1.997856746466182e-05, "loss": 0.0793, "step": 102 }, { "epoch": 0.1, "learning_rate": 1.9977509622105233e-05, "loss": 0.0779, "step": 103 }, { "epoch": 0.1, "learning_rate": 1.9976426328065895e-05, "loss": 0.081, "step": 104 }, { "epoch": 0.1, "learning_rate": 1.9975317585307164e-05, "loss": 0.088, "step": 105 }, { "epoch": 0.1, "learning_rate": 1.997418339665732e-05, "loss": 0.0776, "step": 106 }, { "epoch": 0.11, "learning_rate": 1.997302376500954e-05, "loss": 0.0768, "step": 107 }, { "epoch": 0.11, "learning_rate": 1.997183869332192e-05, "loss": 0.0767, "step": 108 }, { "epoch": 0.11, "learning_rate": 1.9970628184617436e-05, "loss": 0.0788, "step": 109 }, { "epoch": 0.11, "learning_rate": 1.9969392241983957e-05, "loss": 0.0776, "step": 110 }, { "epoch": 0.11, "learning_rate": 1.9968130868574237e-05, "loss": 0.0743, "step": 111 }, { "epoch": 0.11, "learning_rate": 1.9966844067605888e-05, "loss": 0.0823, "step": 112 }, { "epoch": 0.11, "learning_rate": 1.9965531842361393e-05, "loss": 0.0748, "step": 113 }, { "epoch": 0.11, "learning_rate": 1.9964194196188095e-05, "loss": 0.0769, "step": 114 }, { "epoch": 0.11, "learning_rate": 1.9962831132498165e-05, "loss": 0.0706, "step": 115 }, { "epoch": 0.11, "learning_rate": 1.996144265476863e-05, "loss": 0.0742, "step": 116 }, { "epoch": 0.12, "learning_rate": 1.9960028766541336e-05, "loss": 0.0684, "step": 117 }, { "epoch": 0.12, "learning_rate": 1.995858947142295e-05, "loss": 0.0819, "step": 118 }, { "epoch": 0.12, "learning_rate": 1.9957124773084943e-05, "loss": 0.0737, "step": 119 }, { "epoch": 0.12, "learning_rate": 1.99556346752636e-05, "loss": 0.0783, "step": 120 }, { "epoch": 0.12, "learning_rate": 1.9954119181759987e-05, "loss": 0.0767, "step": 121 }, { "epoch": 0.12, "learning_rate": 1.995257829643995e-05, "loss": 0.0767, "step": 122 }, { "epoch": 0.12, "learning_rate": 1.995101202323412e-05, "loss": 0.0807, "step": 123 }, { "epoch": 0.12, "learning_rate": 1.994942036613787e-05, "loss": 0.0837, "step": 124 }, { "epoch": 0.12, "learning_rate": 1.9947803329211343e-05, "loss": 0.0702, "step": 125 }, { "epoch": 0.12, "learning_rate": 1.994616091657941e-05, "loss": 0.0739, "step": 126 }, { "epoch": 0.13, "learning_rate": 1.994449313243168e-05, "loss": 0.082, "step": 127 }, { "epoch": 0.13, "learning_rate": 1.9942799981022472e-05, "loss": 0.0743, "step": 128 }, { "epoch": 0.13, "learning_rate": 1.9941081466670833e-05, "loss": 0.0779, "step": 129 }, { "epoch": 0.13, "learning_rate": 1.993933759376049e-05, "loss": 0.0804, "step": 130 }, { "epoch": 0.13, "learning_rate": 1.9937568366739858e-05, "loss": 0.0765, "step": 131 }, { "epoch": 0.13, "learning_rate": 1.9935773790122043e-05, "loss": 0.0746, "step": 132 }, { "epoch": 0.13, "learning_rate": 1.9933953868484793e-05, "loss": 0.0843, "step": 133 }, { "epoch": 0.13, "learning_rate": 1.9932108606470525e-05, "loss": 0.0754, "step": 134 }, { "epoch": 0.13, "learning_rate": 1.9930238008786288e-05, "loss": 0.0782, "step": 135 }, { "epoch": 0.13, "learning_rate": 1.992834208020377e-05, "loss": 0.0753, "step": 136 }, { "epoch": 0.14, "learning_rate": 1.9926420825559257e-05, "loss": 0.0847, "step": 137 }, { "epoch": 0.14, "learning_rate": 1.9924474249753656e-05, "loss": 0.0776, "step": 138 }, { "epoch": 0.14, "learning_rate": 1.9922502357752453e-05, "loss": 0.0814, "step": 139 }, { "epoch": 0.14, "learning_rate": 1.9920505154585726e-05, "loss": 0.0745, "step": 140 }, { "epoch": 0.14, "learning_rate": 1.9918482645348104e-05, "loss": 0.0795, "step": 141 }, { "epoch": 0.14, "learning_rate": 1.991643483519878e-05, "loss": 0.0775, "step": 142 }, { "epoch": 0.14, "learning_rate": 1.9914361729361484e-05, "loss": 0.0753, "step": 143 }, { "epoch": 0.14, "learning_rate": 1.991226333312447e-05, "loss": 0.0789, "step": 144 }, { "epoch": 0.14, "learning_rate": 1.9910139651840497e-05, "loss": 0.0794, "step": 145 }, { "epoch": 0.14, "learning_rate": 1.990799069092685e-05, "loss": 0.079, "step": 146 }, { "epoch": 0.14, "learning_rate": 1.990581645586527e-05, "loss": 0.0801, "step": 147 }, { "epoch": 0.15, "learning_rate": 1.9903616952201982e-05, "loss": 0.0749, "step": 148 }, { "epoch": 0.15, "learning_rate": 1.990139218554767e-05, "loss": 0.0798, "step": 149 }, { "epoch": 0.15, "learning_rate": 1.989914216157746e-05, "loss": 0.0769, "step": 150 }, { "epoch": 0.15, "learning_rate": 1.9896866886030903e-05, "loss": 0.0697, "step": 151 }, { "epoch": 0.15, "learning_rate": 1.9894566364711965e-05, "loss": 0.0803, "step": 152 }, { "epoch": 0.15, "learning_rate": 1.989224060348901e-05, "loss": 0.0756, "step": 153 }, { "epoch": 0.15, "learning_rate": 1.9889889608294797e-05, "loss": 0.083, "step": 154 }, { "epoch": 0.15, "learning_rate": 1.9887513385126434e-05, "loss": 0.0769, "step": 155 }, { "epoch": 0.15, "learning_rate": 1.9885111940045405e-05, "loss": 0.0751, "step": 156 }, { "epoch": 0.15, "learning_rate": 1.988268527917751e-05, "loss": 0.079, "step": 157 }, { "epoch": 0.16, "learning_rate": 1.9880233408712892e-05, "loss": 0.073, "step": 158 }, { "epoch": 0.16, "learning_rate": 1.9877756334905983e-05, "loss": 0.0831, "step": 159 }, { "epoch": 0.16, "learning_rate": 1.9875254064075522e-05, "loss": 0.0907, "step": 160 }, { "epoch": 0.16, "learning_rate": 1.987272660260451e-05, "loss": 0.0825, "step": 161 }, { "epoch": 0.16, "learning_rate": 1.987017395694021e-05, "loss": 0.0889, "step": 162 }, { "epoch": 0.16, "learning_rate": 1.9867596133594133e-05, "loss": 0.0809, "step": 163 }, { "epoch": 0.16, "learning_rate": 1.986499313914201e-05, "loss": 0.072, "step": 164 }, { "epoch": 0.16, "learning_rate": 1.986236498022378e-05, "loss": 0.0784, "step": 165 }, { "epoch": 0.16, "learning_rate": 1.9859711663543573e-05, "loss": 0.0776, "step": 166 }, { "epoch": 0.16, "learning_rate": 1.9857033195869694e-05, "loss": 0.0764, "step": 167 }, { "epoch": 0.17, "learning_rate": 1.9854329584034614e-05, "loss": 0.0744, "step": 168 }, { "epoch": 0.17, "learning_rate": 1.9851600834934936e-05, "loss": 0.0823, "step": 169 }, { "epoch": 0.17, "learning_rate": 1.984884695553138e-05, "loss": 0.0762, "step": 170 }, { "epoch": 0.17, "learning_rate": 1.984606795284878e-05, "loss": 0.0806, "step": 171 }, { "epoch": 0.17, "learning_rate": 1.9843263833976052e-05, "loss": 0.0739, "step": 172 }, { "epoch": 0.17, "learning_rate": 1.9840434606066182e-05, "loss": 0.0843, "step": 173 }, { "epoch": 0.17, "learning_rate": 1.9837580276336205e-05, "loss": 0.0675, "step": 174 }, { "epoch": 0.17, "learning_rate": 1.983470085206719e-05, "loss": 0.0702, "step": 175 }, { "epoch": 0.17, "learning_rate": 1.9831796340604224e-05, "loss": 0.0783, "step": 176 }, { "epoch": 0.17, "learning_rate": 1.982886674935637e-05, "loss": 0.0683, "step": 177 }, { "epoch": 0.18, "learning_rate": 1.982591208579669e-05, "loss": 0.0696, "step": 178 }, { "epoch": 0.18, "learning_rate": 1.9822932357462185e-05, "loss": 0.0737, "step": 179 }, { "epoch": 0.18, "learning_rate": 1.9819927571953804e-05, "loss": 0.0799, "step": 180 }, { "epoch": 0.18, "learning_rate": 1.9816897736936413e-05, "loss": 0.0789, "step": 181 }, { "epoch": 0.18, "learning_rate": 1.9813842860138764e-05, "loss": 0.0691, "step": 182 }, { "epoch": 0.18, "learning_rate": 1.9810762949353507e-05, "loss": 0.0714, "step": 183 }, { "epoch": 0.18, "learning_rate": 1.9807658012437136e-05, "loss": 0.0748, "step": 184 }, { "epoch": 0.18, "learning_rate": 1.980452805730999e-05, "loss": 0.0758, "step": 185 }, { "epoch": 0.18, "learning_rate": 1.980137309195623e-05, "loss": 0.0677, "step": 186 }, { "epoch": 0.18, "learning_rate": 1.9798193124423804e-05, "loss": 0.0738, "step": 187 }, { "epoch": 0.19, "learning_rate": 1.979498816282445e-05, "loss": 0.0758, "step": 188 }, { "epoch": 0.19, "learning_rate": 1.979175821533365e-05, "loss": 0.0781, "step": 189 }, { "epoch": 0.19, "learning_rate": 1.978850329019064e-05, "loss": 0.0766, "step": 190 }, { "epoch": 0.19, "learning_rate": 1.978522339569835e-05, "loss": 0.0721, "step": 191 }, { "epoch": 0.19, "learning_rate": 1.978191854022342e-05, "loss": 0.0795, "step": 192 }, { "epoch": 0.19, "learning_rate": 1.977858873219616e-05, "loss": 0.0744, "step": 193 }, { "epoch": 0.19, "learning_rate": 1.9775233980110524e-05, "loss": 0.0742, "step": 194 }, { "epoch": 0.19, "learning_rate": 1.9771854292524103e-05, "loss": 0.08, "step": 195 }, { "epoch": 0.19, "learning_rate": 1.9768449678058086e-05, "loss": 0.0811, "step": 196 }, { "epoch": 0.19, "learning_rate": 1.9765020145397264e-05, "loss": 0.0667, "step": 197 }, { "epoch": 0.2, "learning_rate": 1.9761565703289977e-05, "loss": 0.0828, "step": 198 }, { "epoch": 0.2, "learning_rate": 1.9758086360548107e-05, "loss": 0.0687, "step": 199 }, { "epoch": 0.2, "learning_rate": 1.975458212604706e-05, "loss": 0.0795, "step": 200 }, { "epoch": 0.2, "learning_rate": 1.9751053008725736e-05, "loss": 0.0764, "step": 201 }, { "epoch": 0.2, "learning_rate": 1.9747499017586516e-05, "loss": 0.0848, "step": 202 }, { "epoch": 0.2, "learning_rate": 1.9743920161695212e-05, "loss": 0.0717, "step": 203 }, { "epoch": 0.2, "learning_rate": 1.9740316450181086e-05, "loss": 0.0758, "step": 204 }, { "epoch": 0.2, "learning_rate": 1.9736687892236786e-05, "loss": 0.0783, "step": 205 }, { "epoch": 0.2, "learning_rate": 1.973303449711835e-05, "loss": 0.0702, "step": 206 }, { "epoch": 0.2, "learning_rate": 1.9729356274145165e-05, "loss": 0.0848, "step": 207 }, { "epoch": 0.21, "learning_rate": 1.9725653232699962e-05, "loss": 0.0862, "step": 208 }, { "epoch": 0.21, "learning_rate": 1.9721925382228774e-05, "loss": 0.0736, "step": 209 }, { "epoch": 0.21, "learning_rate": 1.971817273224092e-05, "loss": 0.0718, "step": 210 }, { "epoch": 0.21, "learning_rate": 1.971439529230897e-05, "loss": 0.0816, "step": 211 }, { "epoch": 0.21, "learning_rate": 1.9710593072068756e-05, "loss": 0.0759, "step": 212 }, { "epoch": 0.21, "learning_rate": 1.97067660812193e-05, "loss": 0.0662, "step": 213 }, { "epoch": 0.21, "learning_rate": 1.970291432952281e-05, "loss": 0.0724, "step": 214 }, { "epoch": 0.21, "learning_rate": 1.969903782680467e-05, "loss": 0.069, "step": 215 }, { "epoch": 0.21, "learning_rate": 1.969513658295339e-05, "loss": 0.0725, "step": 216 }, { "epoch": 0.21, "learning_rate": 1.9691210607920597e-05, "loss": 0.067, "step": 217 }, { "epoch": 0.21, "learning_rate": 1.9687259911721004e-05, "loss": 0.0697, "step": 218 }, { "epoch": 0.22, "learning_rate": 1.9683284504432384e-05, "loss": 0.0782, "step": 219 }, { "epoch": 0.22, "learning_rate": 1.9679284396195544e-05, "loss": 0.0647, "step": 220 }, { "epoch": 0.22, "learning_rate": 1.9675259597214304e-05, "loss": 0.0681, "step": 221 }, { "epoch": 0.22, "learning_rate": 1.967121011775546e-05, "loss": 0.0723, "step": 222 }, { "epoch": 0.22, "learning_rate": 1.966713596814877e-05, "loss": 0.0734, "step": 223 }, { "epoch": 0.22, "learning_rate": 1.966303715878693e-05, "loss": 0.0729, "step": 224 }, { "epoch": 0.22, "learning_rate": 1.965891370012552e-05, "loss": 0.0652, "step": 225 }, { "epoch": 0.22, "learning_rate": 1.9654765602683022e-05, "loss": 0.0753, "step": 226 }, { "epoch": 0.22, "learning_rate": 1.965059287704075e-05, "loss": 0.0766, "step": 227 }, { "epoch": 0.22, "learning_rate": 1.9646395533842852e-05, "loss": 0.0751, "step": 228 }, { "epoch": 0.23, "learning_rate": 1.9642173583796265e-05, "loss": 0.0732, "step": 229 }, { "epoch": 0.23, "learning_rate": 1.9637927037670704e-05, "loss": 0.0673, "step": 230 }, { "epoch": 0.23, "learning_rate": 1.963365590629862e-05, "loss": 0.0752, "step": 231 }, { "epoch": 0.23, "learning_rate": 1.9629360200575175e-05, "loss": 0.0803, "step": 232 }, { "epoch": 0.23, "learning_rate": 1.9625039931458228e-05, "loss": 0.07, "step": 233 }, { "epoch": 0.23, "learning_rate": 1.9620695109968285e-05, "loss": 0.0734, "step": 234 }, { "epoch": 0.23, "learning_rate": 1.9616325747188494e-05, "loss": 0.0735, "step": 235 }, { "epoch": 0.23, "learning_rate": 1.961193185426459e-05, "loss": 0.0674, "step": 236 }, { "epoch": 0.23, "learning_rate": 1.9607513442404892e-05, "loss": 0.0724, "step": 237 }, { "epoch": 0.23, "learning_rate": 1.960307052288027e-05, "loss": 0.0696, "step": 238 }, { "epoch": 0.24, "learning_rate": 1.959860310702409e-05, "loss": 0.0706, "step": 239 }, { "epoch": 0.24, "learning_rate": 1.9594111206232224e-05, "loss": 0.0743, "step": 240 }, { "epoch": 0.24, "learning_rate": 1.9589594831962998e-05, "loss": 0.0672, "step": 241 }, { "epoch": 0.24, "learning_rate": 1.958505399573716e-05, "loss": 0.0727, "step": 242 }, { "epoch": 0.24, "learning_rate": 1.958048870913786e-05, "loss": 0.0666, "step": 243 }, { "epoch": 0.24, "learning_rate": 1.957589898381063e-05, "loss": 0.0729, "step": 244 }, { "epoch": 0.24, "learning_rate": 1.9571284831463326e-05, "loss": 0.0785, "step": 245 }, { "epoch": 0.24, "learning_rate": 1.9566646263866124e-05, "loss": 0.0703, "step": 246 }, { "epoch": 0.24, "learning_rate": 1.9561983292851477e-05, "loss": 0.0844, "step": 247 }, { "epoch": 0.24, "learning_rate": 1.955729593031409e-05, "loss": 0.073, "step": 248 }, { "epoch": 0.25, "learning_rate": 1.9552584188210883e-05, "loss": 0.0671, "step": 249 }, { "epoch": 0.25, "learning_rate": 1.9547848078560975e-05, "loss": 0.0714, "step": 250 }, { "epoch": 0.25, "learning_rate": 1.954308761344564e-05, "loss": 0.0727, "step": 251 }, { "epoch": 0.25, "learning_rate": 1.9538302805008273e-05, "loss": 0.0715, "step": 252 }, { "epoch": 0.25, "learning_rate": 1.9533493665454372e-05, "loss": 0.0746, "step": 253 }, { "epoch": 0.25, "learning_rate": 1.9528660207051507e-05, "loss": 0.0756, "step": 254 }, { "epoch": 0.25, "learning_rate": 1.9523802442129267e-05, "loss": 0.0738, "step": 255 }, { "epoch": 0.25, "learning_rate": 1.951892038307926e-05, "loss": 0.0764, "step": 256 }, { "epoch": 0.25, "learning_rate": 1.9514014042355057e-05, "loss": 0.0772, "step": 257 }, { "epoch": 0.25, "learning_rate": 1.950908343247217e-05, "loss": 0.0844, "step": 258 }, { "epoch": 0.26, "learning_rate": 1.9504128566008017e-05, "loss": 0.077, "step": 259 }, { "epoch": 0.26, "learning_rate": 1.9499149455601897e-05, "loss": 0.0727, "step": 260 }, { "epoch": 0.26, "learning_rate": 1.949414611395495e-05, "loss": 0.0694, "step": 261 }, { "epoch": 0.26, "learning_rate": 1.9489118553830123e-05, "loss": 0.0778, "step": 262 }, { "epoch": 0.26, "learning_rate": 1.948406678805215e-05, "loss": 0.0851, "step": 263 }, { "epoch": 0.26, "learning_rate": 1.9478990829507507e-05, "loss": 0.0729, "step": 264 }, { "epoch": 0.26, "learning_rate": 1.9473890691144376e-05, "loss": 0.0708, "step": 265 }, { "epoch": 0.26, "learning_rate": 1.946876638597263e-05, "loss": 0.0788, "step": 266 }, { "epoch": 0.26, "learning_rate": 1.946361792706378e-05, "loss": 0.0687, "step": 267 }, { "epoch": 0.26, "learning_rate": 1.9458445327550957e-05, "loss": 0.0695, "step": 268 }, { "epoch": 0.27, "learning_rate": 1.9453248600628867e-05, "loss": 0.0697, "step": 269 }, { "epoch": 0.27, "learning_rate": 1.9448027759553768e-05, "loss": 0.0739, "step": 270 }, { "epoch": 0.27, "learning_rate": 1.9442782817643425e-05, "loss": 0.0762, "step": 271 }, { "epoch": 0.27, "learning_rate": 1.9437513788277077e-05, "loss": 0.0719, "step": 272 }, { "epoch": 0.27, "learning_rate": 1.9432220684895427e-05, "loss": 0.0729, "step": 273 }, { "epoch": 0.27, "learning_rate": 1.942690352100056e-05, "loss": 0.066, "step": 274 }, { "epoch": 0.27, "learning_rate": 1.9421562310155964e-05, "loss": 0.0703, "step": 275 }, { "epoch": 0.27, "learning_rate": 1.941619706598645e-05, "loss": 0.0764, "step": 276 }, { "epoch": 0.27, "learning_rate": 1.9410807802178138e-05, "loss": 0.0668, "step": 277 }, { "epoch": 0.27, "learning_rate": 1.9405394532478422e-05, "loss": 0.0702, "step": 278 }, { "epoch": 0.28, "learning_rate": 1.9399957270695937e-05, "loss": 0.0758, "step": 279 }, { "epoch": 0.28, "learning_rate": 1.939449603070051e-05, "loss": 0.0725, "step": 280 }, { "epoch": 0.28, "learning_rate": 1.9389010826423142e-05, "loss": 0.073, "step": 281 }, { "epoch": 0.28, "learning_rate": 1.9383501671855957e-05, "loss": 0.0649, "step": 282 }, { "epoch": 0.28, "learning_rate": 1.9377968581052176e-05, "loss": 0.0755, "step": 283 }, { "epoch": 0.28, "learning_rate": 1.937241156812608e-05, "loss": 0.0746, "step": 284 }, { "epoch": 0.28, "learning_rate": 1.9366830647252974e-05, "loss": 0.074, "step": 285 }, { "epoch": 0.28, "learning_rate": 1.936122583266914e-05, "loss": 0.0673, "step": 286 }, { "epoch": 0.28, "learning_rate": 1.9355597138671826e-05, "loss": 0.0765, "step": 287 }, { "epoch": 0.28, "learning_rate": 1.9349944579619177e-05, "loss": 0.0692, "step": 288 }, { "epoch": 0.29, "learning_rate": 1.9344268169930228e-05, "loss": 0.0706, "step": 289 }, { "epoch": 0.29, "learning_rate": 1.933856792408485e-05, "loss": 0.0692, "step": 290 }, { "epoch": 0.29, "learning_rate": 1.933284385662371e-05, "loss": 0.0721, "step": 291 }, { "epoch": 0.29, "learning_rate": 1.9327095982148258e-05, "loss": 0.0733, "step": 292 }, { "epoch": 0.29, "learning_rate": 1.9321324315320652e-05, "loss": 0.0792, "step": 293 }, { "epoch": 0.29, "learning_rate": 1.9315528870863757e-05, "loss": 0.0767, "step": 294 }, { "epoch": 0.29, "learning_rate": 1.9309709663561085e-05, "loss": 0.076, "step": 295 }, { "epoch": 0.29, "learning_rate": 1.930386670825677e-05, "loss": 0.0703, "step": 296 }, { "epoch": 0.29, "learning_rate": 1.9298000019855517e-05, "loss": 0.0717, "step": 297 }, { "epoch": 0.29, "learning_rate": 1.929210961332258e-05, "loss": 0.0681, "step": 298 }, { "epoch": 0.29, "learning_rate": 1.928619550368371e-05, "loss": 0.0732, "step": 299 }, { "epoch": 0.3, "learning_rate": 1.9280257706025114e-05, "loss": 0.0676, "step": 300 }, { "epoch": 0.3, "learning_rate": 1.9274296235493444e-05, "loss": 0.077, "step": 301 }, { "epoch": 0.3, "learning_rate": 1.926831110729572e-05, "loss": 0.0778, "step": 302 }, { "epoch": 0.3, "learning_rate": 1.926230233669932e-05, "loss": 0.0786, "step": 303 }, { "epoch": 0.3, "learning_rate": 1.925626993903193e-05, "loss": 0.0676, "step": 304 }, { "epoch": 0.3, "learning_rate": 1.9250213929681502e-05, "loss": 0.0607, "step": 305 }, { "epoch": 0.3, "learning_rate": 1.9244134324096223e-05, "loss": 0.0769, "step": 306 }, { "epoch": 0.3, "learning_rate": 1.923803113778447e-05, "loss": 0.0747, "step": 307 }, { "epoch": 0.3, "learning_rate": 1.9231904386314772e-05, "loss": 0.0779, "step": 308 }, { "epoch": 0.3, "learning_rate": 1.9225754085315766e-05, "loss": 0.0712, "step": 309 }, { "epoch": 0.31, "learning_rate": 1.9219580250476167e-05, "loss": 0.0725, "step": 310 }, { "epoch": 0.31, "learning_rate": 1.9213382897544722e-05, "loss": 0.0709, "step": 311 }, { "epoch": 0.31, "learning_rate": 1.920716204233016e-05, "loss": 0.074, "step": 312 }, { "epoch": 0.31, "learning_rate": 1.9200917700701176e-05, "loss": 0.0811, "step": 313 }, { "epoch": 0.31, "learning_rate": 1.9194649888586366e-05, "loss": 0.069, "step": 314 }, { "epoch": 0.31, "learning_rate": 1.9188358621974202e-05, "loss": 0.0766, "step": 315 }, { "epoch": 0.31, "learning_rate": 1.9182043916912984e-05, "loss": 0.0788, "step": 316 }, { "epoch": 0.31, "learning_rate": 1.9175705789510798e-05, "loss": 0.0814, "step": 317 }, { "epoch": 0.31, "learning_rate": 1.9169344255935484e-05, "loss": 0.0847, "step": 318 }, { "epoch": 0.31, "learning_rate": 1.916295933241458e-05, "loss": 0.081, "step": 319 }, { "epoch": 0.32, "learning_rate": 1.915655103523529e-05, "loss": 0.0748, "step": 320 }, { "epoch": 0.32, "learning_rate": 1.9150119380744458e-05, "loss": 0.0681, "step": 321 }, { "epoch": 0.32, "learning_rate": 1.9143664385348487e-05, "loss": 0.0676, "step": 322 }, { "epoch": 0.32, "learning_rate": 1.9137186065513332e-05, "loss": 0.0776, "step": 323 }, { "epoch": 0.32, "learning_rate": 1.9130684437764445e-05, "loss": 0.0806, "step": 324 }, { "epoch": 0.32, "learning_rate": 1.912415951868673e-05, "loss": 0.0699, "step": 325 }, { "epoch": 0.32, "learning_rate": 1.9117611324924513e-05, "loss": 0.0834, "step": 326 }, { "epoch": 0.32, "learning_rate": 1.9111039873181478e-05, "loss": 0.0725, "step": 327 }, { "epoch": 0.32, "learning_rate": 1.9104445180220648e-05, "loss": 0.0754, "step": 328 }, { "epoch": 0.32, "learning_rate": 1.909782726286433e-05, "loss": 0.0739, "step": 329 }, { "epoch": 0.33, "learning_rate": 1.9091186137994074e-05, "loss": 0.0687, "step": 330 }, { "epoch": 0.33, "learning_rate": 1.9084521822550623e-05, "loss": 0.0772, "step": 331 }, { "epoch": 0.33, "learning_rate": 1.9077834333533885e-05, "loss": 0.0771, "step": 332 }, { "epoch": 0.33, "learning_rate": 1.9071123688002875e-05, "loss": 0.0716, "step": 333 }, { "epoch": 0.33, "learning_rate": 1.9064389903075676e-05, "loss": 0.0687, "step": 334 }, { "epoch": 0.33, "learning_rate": 1.9057632995929413e-05, "loss": 0.0764, "step": 335 }, { "epoch": 0.33, "learning_rate": 1.905085298380017e-05, "loss": 0.073, "step": 336 }, { "epoch": 0.33, "learning_rate": 1.9044049883982984e-05, "loss": 0.0754, "step": 337 }, { "epoch": 0.33, "learning_rate": 1.9037223713831786e-05, "loss": 0.0711, "step": 338 }, { "epoch": 0.33, "learning_rate": 1.903037449075935e-05, "loss": 0.0711, "step": 339 }, { "epoch": 0.34, "learning_rate": 1.9023502232237258e-05, "loss": 0.0717, "step": 340 }, { "epoch": 0.34, "learning_rate": 1.901660695579585e-05, "loss": 0.0721, "step": 341 }, { "epoch": 0.34, "learning_rate": 1.900968867902419e-05, "loss": 0.0713, "step": 342 }, { "epoch": 0.34, "learning_rate": 1.9002747419570014e-05, "loss": 0.0696, "step": 343 }, { "epoch": 0.34, "learning_rate": 1.899578319513967e-05, "loss": 0.0685, "step": 344 }, { "epoch": 0.34, "learning_rate": 1.8988796023498096e-05, "loss": 0.0695, "step": 345 }, { "epoch": 0.34, "learning_rate": 1.898178592246877e-05, "loss": 0.0676, "step": 346 }, { "epoch": 0.34, "learning_rate": 1.8974752909933653e-05, "loss": 0.0692, "step": 347 }, { "epoch": 0.34, "learning_rate": 1.8967697003833156e-05, "loss": 0.0717, "step": 348 }, { "epoch": 0.34, "learning_rate": 1.896061822216608e-05, "loss": 0.0656, "step": 349 }, { "epoch": 0.35, "learning_rate": 1.8953516582989597e-05, "loss": 0.0725, "step": 350 }, { "epoch": 0.35, "learning_rate": 1.8946392104419162e-05, "loss": 0.0709, "step": 351 }, { "epoch": 0.35, "learning_rate": 1.893924480462851e-05, "loss": 0.0678, "step": 352 }, { "epoch": 0.35, "learning_rate": 1.8932074701849576e-05, "loss": 0.0723, "step": 353 }, { "epoch": 0.35, "learning_rate": 1.892488181437248e-05, "loss": 0.0753, "step": 354 }, { "epoch": 0.35, "learning_rate": 1.8917666160545446e-05, "loss": 0.0687, "step": 355 }, { "epoch": 0.35, "learning_rate": 1.8910427758774775e-05, "loss": 0.0704, "step": 356 }, { "epoch": 0.35, "learning_rate": 1.8903166627524812e-05, "loss": 0.0782, "step": 357 }, { "epoch": 0.35, "learning_rate": 1.889588278531786e-05, "loss": 0.0718, "step": 358 }, { "epoch": 0.35, "learning_rate": 1.888857625073417e-05, "loss": 0.0725, "step": 359 }, { "epoch": 0.36, "learning_rate": 1.8881247042411878e-05, "loss": 0.0699, "step": 360 }, { "epoch": 0.36, "learning_rate": 1.887389517904694e-05, "loss": 0.0711, "step": 361 }, { "epoch": 0.36, "learning_rate": 1.8866520679393127e-05, "loss": 0.0656, "step": 362 }, { "epoch": 0.36, "learning_rate": 1.885912356226194e-05, "loss": 0.0685, "step": 363 }, { "epoch": 0.36, "learning_rate": 1.885170384652258e-05, "loss": 0.0738, "step": 364 }, { "epoch": 0.36, "learning_rate": 1.884426155110188e-05, "loss": 0.0664, "step": 365 }, { "epoch": 0.36, "learning_rate": 1.883679669498429e-05, "loss": 0.0693, "step": 366 }, { "epoch": 0.36, "learning_rate": 1.8829309297211804e-05, "loss": 0.0704, "step": 367 }, { "epoch": 0.36, "learning_rate": 1.8821799376883912e-05, "loss": 0.0685, "step": 368 }, { "epoch": 0.36, "learning_rate": 1.8814266953157557e-05, "loss": 0.0745, "step": 369 }, { "epoch": 0.36, "learning_rate": 1.880671204524709e-05, "loss": 0.0697, "step": 370 }, { "epoch": 0.37, "learning_rate": 1.8799134672424214e-05, "loss": 0.0754, "step": 371 }, { "epoch": 0.37, "learning_rate": 1.879153485401794e-05, "loss": 0.0677, "step": 372 }, { "epoch": 0.37, "learning_rate": 1.8783912609414535e-05, "loss": 0.0672, "step": 373 }, { "epoch": 0.37, "learning_rate": 1.8776267958057467e-05, "loss": 0.074, "step": 374 }, { "epoch": 0.37, "learning_rate": 1.876860091944736e-05, "loss": 0.0742, "step": 375 }, { "epoch": 0.37, "learning_rate": 1.876091151314196e-05, "loss": 0.0707, "step": 376 }, { "epoch": 0.37, "learning_rate": 1.875319975875606e-05, "loss": 0.0728, "step": 377 }, { "epoch": 0.37, "learning_rate": 1.8745465675961453e-05, "loss": 0.0719, "step": 378 }, { "epoch": 0.37, "learning_rate": 1.8737709284486905e-05, "loss": 0.0722, "step": 379 }, { "epoch": 0.37, "learning_rate": 1.8729930604118074e-05, "loss": 0.0677, "step": 380 }, { "epoch": 0.38, "learning_rate": 1.872212965469749e-05, "loss": 0.0784, "step": 381 }, { "epoch": 0.38, "learning_rate": 1.871430645612447e-05, "loss": 0.067, "step": 382 }, { "epoch": 0.38, "learning_rate": 1.8706461028355107e-05, "loss": 0.0703, "step": 383 }, { "epoch": 0.38, "learning_rate": 1.8698593391402177e-05, "loss": 0.0659, "step": 384 }, { "epoch": 0.38, "learning_rate": 1.8690703565335127e-05, "loss": 0.0673, "step": 385 }, { "epoch": 0.38, "learning_rate": 1.8682791570279998e-05, "loss": 0.0688, "step": 386 }, { "epoch": 0.38, "learning_rate": 1.8674857426419383e-05, "loss": 0.0747, "step": 387 }, { "epoch": 0.38, "learning_rate": 1.8666901153992367e-05, "loss": 0.0679, "step": 388 }, { "epoch": 0.38, "learning_rate": 1.865892277329449e-05, "loss": 0.075, "step": 389 }, { "epoch": 0.38, "learning_rate": 1.865092230467769e-05, "loss": 0.0654, "step": 390 }, { "epoch": 0.39, "learning_rate": 1.8642899768550243e-05, "loss": 0.0768, "step": 391 }, { "epoch": 0.39, "learning_rate": 1.8634855185376722e-05, "loss": 0.0724, "step": 392 }, { "epoch": 0.39, "learning_rate": 1.8626788575677926e-05, "loss": 0.0686, "step": 393 }, { "epoch": 0.39, "learning_rate": 1.8618699960030865e-05, "loss": 0.0714, "step": 394 }, { "epoch": 0.39, "learning_rate": 1.8610589359068658e-05, "loss": 0.075, "step": 395 }, { "epoch": 0.39, "learning_rate": 1.8602456793480522e-05, "loss": 0.07, "step": 396 }, { "epoch": 0.39, "learning_rate": 1.8594302284011704e-05, "loss": 0.0731, "step": 397 }, { "epoch": 0.39, "learning_rate": 1.8586125851463414e-05, "loss": 0.0714, "step": 398 }, { "epoch": 0.39, "learning_rate": 1.85779275166928e-05, "loss": 0.0731, "step": 399 }, { "epoch": 0.39, "learning_rate": 1.8569707300612866e-05, "loss": 0.0694, "step": 400 }, { "epoch": 0.4, "learning_rate": 1.8561465224192453e-05, "loss": 0.0742, "step": 401 }, { "epoch": 0.4, "learning_rate": 1.8553201308456138e-05, "loss": 0.0714, "step": 402 }, { "epoch": 0.4, "learning_rate": 1.854491557448424e-05, "loss": 0.0722, "step": 403 }, { "epoch": 0.4, "learning_rate": 1.85366080434127e-05, "loss": 0.0671, "step": 404 }, { "epoch": 0.4, "learning_rate": 1.852827873643308e-05, "loss": 0.0695, "step": 405 }, { "epoch": 0.4, "learning_rate": 1.8519927674792496e-05, "loss": 0.0698, "step": 406 }, { "epoch": 0.4, "learning_rate": 1.8511554879793537e-05, "loss": 0.0689, "step": 407 }, { "epoch": 0.4, "learning_rate": 1.8503160372794253e-05, "loss": 0.0776, "step": 408 }, { "epoch": 0.4, "learning_rate": 1.8494744175208065e-05, "loss": 0.0778, "step": 409 }, { "epoch": 0.4, "learning_rate": 1.848630630850372e-05, "loss": 0.0729, "step": 410 }, { "epoch": 0.41, "learning_rate": 1.8477846794205258e-05, "loss": 0.0723, "step": 411 }, { "epoch": 0.41, "learning_rate": 1.846936565389193e-05, "loss": 0.0716, "step": 412 }, { "epoch": 0.41, "learning_rate": 1.8460862909198143e-05, "loss": 0.0713, "step": 413 }, { "epoch": 0.41, "learning_rate": 1.8452338581813433e-05, "loss": 0.0683, "step": 414 }, { "epoch": 0.41, "learning_rate": 1.8443792693482377e-05, "loss": 0.0673, "step": 415 }, { "epoch": 0.41, "learning_rate": 1.8435225266004555e-05, "loss": 0.0713, "step": 416 }, { "epoch": 0.41, "learning_rate": 1.8426636321234497e-05, "loss": 0.0673, "step": 417 }, { "epoch": 0.41, "learning_rate": 1.8418025881081612e-05, "loss": 0.0667, "step": 418 }, { "epoch": 0.41, "learning_rate": 1.8409393967510144e-05, "loss": 0.0719, "step": 419 }, { "epoch": 0.41, "learning_rate": 1.8400740602539117e-05, "loss": 0.0662, "step": 420 }, { "epoch": 0.42, "learning_rate": 1.8392065808242278e-05, "loss": 0.0671, "step": 421 }, { "epoch": 0.42, "learning_rate": 1.838336960674802e-05, "loss": 0.067, "step": 422 }, { "epoch": 0.42, "learning_rate": 1.8374652020239365e-05, "loss": 0.0675, "step": 423 }, { "epoch": 0.42, "learning_rate": 1.8365913070953875e-05, "loss": 0.0711, "step": 424 }, { "epoch": 0.42, "learning_rate": 1.8357152781183606e-05, "loss": 0.0617, "step": 425 }, { "epoch": 0.42, "learning_rate": 1.834837117327505e-05, "loss": 0.0689, "step": 426 }, { "epoch": 0.42, "learning_rate": 1.8339568269629085e-05, "loss": 0.0638, "step": 427 }, { "epoch": 0.42, "learning_rate": 1.8330744092700913e-05, "loss": 0.0624, "step": 428 }, { "epoch": 0.42, "learning_rate": 1.8321898664999986e-05, "loss": 0.0729, "step": 429 }, { "epoch": 0.42, "learning_rate": 1.8313032009089985e-05, "loss": 0.0723, "step": 430 }, { "epoch": 0.43, "learning_rate": 1.830414414758873e-05, "loss": 0.0684, "step": 431 }, { "epoch": 0.43, "learning_rate": 1.829523510316813e-05, "loss": 0.0672, "step": 432 }, { "epoch": 0.43, "learning_rate": 1.8286304898554145e-05, "loss": 0.0705, "step": 433 }, { "epoch": 0.43, "learning_rate": 1.82773535565267e-05, "loss": 0.0678, "step": 434 }, { "epoch": 0.43, "learning_rate": 1.826838109991964e-05, "loss": 0.0685, "step": 435 }, { "epoch": 0.43, "learning_rate": 1.8259387551620674e-05, "loss": 0.0618, "step": 436 }, { "epoch": 0.43, "learning_rate": 1.8250372934571313e-05, "loss": 0.0676, "step": 437 }, { "epoch": 0.43, "learning_rate": 1.8241337271766807e-05, "loss": 0.0663, "step": 438 }, { "epoch": 0.43, "learning_rate": 1.82322805862561e-05, "loss": 0.0744, "step": 439 }, { "epoch": 0.43, "learning_rate": 1.8223202901141758e-05, "loss": 0.0621, "step": 440 }, { "epoch": 0.43, "learning_rate": 1.821410423957991e-05, "loss": 0.0688, "step": 441 }, { "epoch": 0.44, "learning_rate": 1.8204984624780205e-05, "loss": 0.0624, "step": 442 }, { "epoch": 0.44, "learning_rate": 1.8195844080005728e-05, "loss": 0.0704, "step": 443 }, { "epoch": 0.44, "learning_rate": 1.8186682628572966e-05, "loss": 0.0671, "step": 444 }, { "epoch": 0.44, "learning_rate": 1.8177500293851728e-05, "loss": 0.0645, "step": 445 }, { "epoch": 0.44, "learning_rate": 1.8168297099265094e-05, "loss": 0.071, "step": 446 }, { "epoch": 0.44, "learning_rate": 1.8159073068289365e-05, "loss": 0.0641, "step": 447 }, { "epoch": 0.44, "learning_rate": 1.8149828224453977e-05, "loss": 0.0701, "step": 448 }, { "epoch": 0.44, "learning_rate": 1.8140562591341476e-05, "loss": 0.0677, "step": 449 }, { "epoch": 0.44, "learning_rate": 1.8131276192587423e-05, "loss": 0.0618, "step": 450 }, { "epoch": 0.44, "learning_rate": 1.812196905188036e-05, "loss": 0.0806, "step": 451 }, { "epoch": 0.45, "learning_rate": 1.811264119296173e-05, "loss": 0.0676, "step": 452 }, { "epoch": 0.45, "learning_rate": 1.810329263962584e-05, "loss": 0.0609, "step": 453 }, { "epoch": 0.45, "learning_rate": 1.8093923415719765e-05, "loss": 0.0685, "step": 454 }, { "epoch": 0.45, "learning_rate": 1.8084533545143333e-05, "loss": 0.0724, "step": 455 }, { "epoch": 0.45, "learning_rate": 1.807512305184902e-05, "loss": 0.0707, "step": 456 }, { "epoch": 0.45, "learning_rate": 1.8065691959841918e-05, "loss": 0.0648, "step": 457 }, { "epoch": 0.45, "learning_rate": 1.805624029317966e-05, "loss": 0.0718, "step": 458 }, { "epoch": 0.45, "learning_rate": 1.804676807597236e-05, "loss": 0.071, "step": 459 }, { "epoch": 0.45, "learning_rate": 1.803727533238257e-05, "loss": 0.0704, "step": 460 }, { "epoch": 0.45, "learning_rate": 1.8027762086625175e-05, "loss": 0.0706, "step": 461 }, { "epoch": 0.46, "learning_rate": 1.8018228362967377e-05, "loss": 0.075, "step": 462 }, { "epoch": 0.46, "learning_rate": 1.800867418572862e-05, "loss": 0.0675, "step": 463 }, { "epoch": 0.46, "learning_rate": 1.7999099579280512e-05, "loss": 0.07, "step": 464 }, { "epoch": 0.46, "learning_rate": 1.798950456804677e-05, "loss": 0.0689, "step": 465 }, { "epoch": 0.46, "learning_rate": 1.7979889176503174e-05, "loss": 0.0675, "step": 466 }, { "epoch": 0.46, "learning_rate": 1.7970253429177477e-05, "loss": 0.0686, "step": 467 }, { "epoch": 0.46, "learning_rate": 1.7960597350649377e-05, "loss": 0.068, "step": 468 }, { "epoch": 0.46, "learning_rate": 1.795092096555041e-05, "loss": 0.0681, "step": 469 }, { "epoch": 0.46, "learning_rate": 1.7941224298563934e-05, "loss": 0.073, "step": 470 }, { "epoch": 0.46, "learning_rate": 1.7931507374425028e-05, "loss": 0.064, "step": 471 }, { "epoch": 0.47, "learning_rate": 1.792177021792046e-05, "loss": 0.067, "step": 472 }, { "epoch": 0.47, "learning_rate": 1.7912012853888585e-05, "loss": 0.076, "step": 473 }, { "epoch": 0.47, "learning_rate": 1.7902235307219333e-05, "loss": 0.0688, "step": 474 }, { "epoch": 0.47, "learning_rate": 1.78924376028541e-05, "loss": 0.0642, "step": 475 }, { "epoch": 0.47, "learning_rate": 1.7882619765785705e-05, "loss": 0.0712, "step": 476 }, { "epoch": 0.47, "learning_rate": 1.7872781821058324e-05, "loss": 0.0693, "step": 477 }, { "epoch": 0.47, "learning_rate": 1.786292379376743e-05, "loss": 0.0753, "step": 478 }, { "epoch": 0.47, "learning_rate": 1.7853045709059713e-05, "loss": 0.0659, "step": 479 }, { "epoch": 0.47, "learning_rate": 1.784314759213304e-05, "loss": 0.0676, "step": 480 }, { "epoch": 0.47, "learning_rate": 1.7833229468236367e-05, "loss": 0.0641, "step": 481 }, { "epoch": 0.48, "learning_rate": 1.7823291362669697e-05, "loss": 0.0661, "step": 482 }, { "epoch": 0.48, "learning_rate": 1.7813333300783996e-05, "loss": 0.0703, "step": 483 }, { "epoch": 0.48, "learning_rate": 1.7803355307981132e-05, "loss": 0.0709, "step": 484 }, { "epoch": 0.48, "learning_rate": 1.779335740971382e-05, "loss": 0.0787, "step": 485 }, { "epoch": 0.48, "learning_rate": 1.7783339631485566e-05, "loss": 0.0732, "step": 486 }, { "epoch": 0.48, "learning_rate": 1.777330199885056e-05, "loss": 0.0674, "step": 487 }, { "epoch": 0.48, "learning_rate": 1.776324453741365e-05, "loss": 0.0693, "step": 488 }, { "epoch": 0.48, "learning_rate": 1.7753167272830276e-05, "loss": 0.0739, "step": 489 }, { "epoch": 0.48, "learning_rate": 1.774307023080638e-05, "loss": 0.0708, "step": 490 }, { "epoch": 0.48, "learning_rate": 1.7732953437098358e-05, "loss": 0.0689, "step": 491 }, { "epoch": 0.49, "learning_rate": 1.772281691751299e-05, "loss": 0.0714, "step": 492 }, { "epoch": 0.49, "learning_rate": 1.7712660697907377e-05, "loss": 0.0627, "step": 493 }, { "epoch": 0.49, "learning_rate": 1.7702484804188867e-05, "loss": 0.0712, "step": 494 }, { "epoch": 0.49, "learning_rate": 1.7692289262315e-05, "loss": 0.0647, "step": 495 }, { "epoch": 0.49, "learning_rate": 1.7682074098293434e-05, "loss": 0.0647, "step": 496 }, { "epoch": 0.49, "learning_rate": 1.767183933818188e-05, "loss": 0.0689, "step": 497 }, { "epoch": 0.49, "learning_rate": 1.766158500808804e-05, "loss": 0.0676, "step": 498 }, { "epoch": 0.49, "learning_rate": 1.765131113416953e-05, "loss": 0.0675, "step": 499 }, { "epoch": 0.49, "learning_rate": 1.764101774263383e-05, "loss": 0.0717, "step": 500 }, { "epoch": 0.49, "learning_rate": 1.7630704859738192e-05, "loss": 0.068, "step": 501 }, { "epoch": 0.5, "learning_rate": 1.7620372511789607e-05, "loss": 0.0708, "step": 502 }, { "epoch": 0.5, "learning_rate": 1.76100207251447e-05, "loss": 0.0647, "step": 503 }, { "epoch": 0.5, "learning_rate": 1.7599649526209702e-05, "loss": 0.0674, "step": 504 }, { "epoch": 0.5, "learning_rate": 1.7589258941440346e-05, "loss": 0.0634, "step": 505 }, { "epoch": 0.5, "learning_rate": 1.7578848997341823e-05, "loss": 0.0699, "step": 506 }, { "epoch": 0.5, "learning_rate": 1.7568419720468706e-05, "loss": 0.0674, "step": 507 }, { "epoch": 0.5, "learning_rate": 1.7557971137424886e-05, "loss": 0.0706, "step": 508 }, { "epoch": 0.5, "learning_rate": 1.75475032748635e-05, "loss": 0.0718, "step": 509 }, { "epoch": 0.5, "learning_rate": 1.753701615948686e-05, "loss": 0.0751, "step": 510 }, { "epoch": 0.5, "learning_rate": 1.75265098180464e-05, "loss": 0.0721, "step": 511 }, { "epoch": 0.5, "learning_rate": 1.751598427734259e-05, "loss": 0.0663, "step": 512 }, { "epoch": 0.51, "learning_rate": 1.750543956422488e-05, "loss": 0.064, "step": 513 }, { "epoch": 0.51, "learning_rate": 1.7494875705591617e-05, "loss": 0.0805, "step": 514 }, { "epoch": 0.51, "learning_rate": 1.748429272839e-05, "loss": 0.0711, "step": 515 }, { "epoch": 0.51, "learning_rate": 1.747369065961599e-05, "loss": 0.0747, "step": 516 }, { "epoch": 0.51, "learning_rate": 1.746306952631424e-05, "loss": 0.0715, "step": 517 }, { "epoch": 0.51, "learning_rate": 1.7452429355578053e-05, "loss": 0.0775, "step": 518 }, { "epoch": 0.51, "learning_rate": 1.7441770174549276e-05, "loss": 0.0672, "step": 519 }, { "epoch": 0.51, "learning_rate": 1.7431092010418266e-05, "loss": 0.0759, "step": 520 }, { "epoch": 0.51, "learning_rate": 1.7420394890423786e-05, "loss": 0.0733, "step": 521 }, { "epoch": 0.51, "learning_rate": 1.7409678841852968e-05, "loss": 0.0719, "step": 522 }, { "epoch": 0.52, "learning_rate": 1.7398943892041223e-05, "loss": 0.075, "step": 523 }, { "epoch": 0.52, "learning_rate": 1.7388190068372174e-05, "loss": 0.0716, "step": 524 }, { "epoch": 0.52, "learning_rate": 1.7377417398277593e-05, "loss": 0.0675, "step": 525 }, { "epoch": 0.52, "learning_rate": 1.7366625909237328e-05, "loss": 0.0712, "step": 526 }, { "epoch": 0.52, "learning_rate": 1.7355815628779227e-05, "loss": 0.0654, "step": 527 }, { "epoch": 0.52, "learning_rate": 1.734498658447908e-05, "loss": 0.0688, "step": 528 }, { "epoch": 0.52, "learning_rate": 1.7334138803960527e-05, "loss": 0.0631, "step": 529 }, { "epoch": 0.52, "learning_rate": 1.7323272314895022e-05, "loss": 0.0641, "step": 530 }, { "epoch": 0.52, "learning_rate": 1.731238714500173e-05, "loss": 0.0753, "step": 531 }, { "epoch": 0.52, "learning_rate": 1.7301483322047467e-05, "loss": 0.0781, "step": 532 }, { "epoch": 0.53, "learning_rate": 1.729056087384664e-05, "loss": 0.0662, "step": 533 }, { "epoch": 0.53, "learning_rate": 1.7279619828261164e-05, "loss": 0.0784, "step": 534 }, { "epoch": 0.53, "learning_rate": 1.726866021320038e-05, "loss": 0.0637, "step": 535 }, { "epoch": 0.53, "learning_rate": 1.7257682056621022e-05, "loss": 0.0689, "step": 536 }, { "epoch": 0.53, "learning_rate": 1.7246685386527098e-05, "loss": 0.0776, "step": 537 }, { "epoch": 0.53, "learning_rate": 1.723567023096986e-05, "loss": 0.0737, "step": 538 }, { "epoch": 0.53, "learning_rate": 1.72246366180477e-05, "loss": 0.0712, "step": 539 }, { "epoch": 0.53, "learning_rate": 1.7213584575906104e-05, "loss": 0.0807, "step": 540 }, { "epoch": 0.53, "learning_rate": 1.7202514132737566e-05, "loss": 0.0761, "step": 541 }, { "epoch": 0.53, "learning_rate": 1.7191425316781508e-05, "loss": 0.0649, "step": 542 }, { "epoch": 0.54, "learning_rate": 1.7180318156324235e-05, "loss": 0.0703, "step": 543 }, { "epoch": 0.54, "learning_rate": 1.7169192679698837e-05, "loss": 0.0694, "step": 544 }, { "epoch": 0.54, "learning_rate": 1.715804891528513e-05, "loss": 0.074, "step": 545 }, { "epoch": 0.54, "learning_rate": 1.714688689150958e-05, "loss": 0.0706, "step": 546 }, { "epoch": 0.54, "learning_rate": 1.7135706636845224e-05, "loss": 0.0682, "step": 547 }, { "epoch": 0.54, "learning_rate": 1.7124508179811617e-05, "loss": 0.0687, "step": 548 }, { "epoch": 0.54, "learning_rate": 1.7113291548974735e-05, "loss": 0.0651, "step": 549 }, { "epoch": 0.54, "learning_rate": 1.7102056772946912e-05, "loss": 0.0739, "step": 550 }, { "epoch": 0.54, "learning_rate": 1.7090803880386784e-05, "loss": 0.0658, "step": 551 }, { "epoch": 0.54, "learning_rate": 1.707953289999918e-05, "loss": 0.0718, "step": 552 }, { "epoch": 0.55, "learning_rate": 1.7068243860535084e-05, "loss": 0.0778, "step": 553 }, { "epoch": 0.55, "learning_rate": 1.7056936790791536e-05, "loss": 0.0669, "step": 554 }, { "epoch": 0.55, "learning_rate": 1.704561171961158e-05, "loss": 0.0739, "step": 555 }, { "epoch": 0.55, "learning_rate": 1.7034268675884167e-05, "loss": 0.0693, "step": 556 }, { "epoch": 0.55, "learning_rate": 1.7022907688544106e-05, "loss": 0.0745, "step": 557 }, { "epoch": 0.55, "learning_rate": 1.701152878657197e-05, "loss": 0.0703, "step": 558 }, { "epoch": 0.55, "learning_rate": 1.700013199899403e-05, "loss": 0.065, "step": 559 }, { "epoch": 0.55, "learning_rate": 1.6988717354882192e-05, "loss": 0.0813, "step": 560 }, { "epoch": 0.55, "learning_rate": 1.69772848833539e-05, "loss": 0.0638, "step": 561 }, { "epoch": 0.55, "learning_rate": 1.6965834613572082e-05, "loss": 0.0654, "step": 562 }, { "epoch": 0.56, "learning_rate": 1.695436657474505e-05, "loss": 0.0692, "step": 563 }, { "epoch": 0.56, "learning_rate": 1.6942880796126472e-05, "loss": 0.0726, "step": 564 }, { "epoch": 0.56, "learning_rate": 1.693137730701524e-05, "loss": 0.0738, "step": 565 }, { "epoch": 0.56, "learning_rate": 1.6919856136755438e-05, "loss": 0.0705, "step": 566 }, { "epoch": 0.56, "learning_rate": 1.6908317314736256e-05, "loss": 0.0676, "step": 567 }, { "epoch": 0.56, "learning_rate": 1.6896760870391898e-05, "loss": 0.0684, "step": 568 }, { "epoch": 0.56, "learning_rate": 1.6885186833201533e-05, "loss": 0.0781, "step": 569 }, { "epoch": 0.56, "learning_rate": 1.68735952326892e-05, "loss": 0.0781, "step": 570 }, { "epoch": 0.56, "learning_rate": 1.6861986098423745e-05, "loss": 0.0724, "step": 571 }, { "epoch": 0.56, "learning_rate": 1.6850359460018737e-05, "loss": 0.0677, "step": 572 }, { "epoch": 0.57, "learning_rate": 1.6838715347132394e-05, "loss": 0.0668, "step": 573 }, { "epoch": 0.57, "learning_rate": 1.682705378946752e-05, "loss": 0.0721, "step": 574 }, { "epoch": 0.57, "learning_rate": 1.68153748167714e-05, "loss": 0.0637, "step": 575 }, { "epoch": 0.57, "learning_rate": 1.6803678458835765e-05, "loss": 0.0733, "step": 576 }, { "epoch": 0.57, "learning_rate": 1.6791964745496675e-05, "loss": 0.0798, "step": 577 }, { "epoch": 0.57, "learning_rate": 1.6780233706634478e-05, "loss": 0.0676, "step": 578 }, { "epoch": 0.57, "learning_rate": 1.6768485372173696e-05, "loss": 0.0789, "step": 579 }, { "epoch": 0.57, "learning_rate": 1.6756719772082993e-05, "loss": 0.0711, "step": 580 }, { "epoch": 0.57, "learning_rate": 1.6744936936375063e-05, "loss": 0.0698, "step": 581 }, { "epoch": 0.57, "learning_rate": 1.673313689510657e-05, "loss": 0.0728, "step": 582 }, { "epoch": 0.57, "learning_rate": 1.672131967837806e-05, "loss": 0.0746, "step": 583 }, { "epoch": 0.58, "learning_rate": 1.67094853163339e-05, "loss": 0.0736, "step": 584 }, { "epoch": 0.58, "learning_rate": 1.6697633839162196e-05, "loss": 0.0717, "step": 585 }, { "epoch": 0.58, "learning_rate": 1.6685765277094702e-05, "loss": 0.0732, "step": 586 }, { "epoch": 0.58, "learning_rate": 1.6673879660406756e-05, "loss": 0.0769, "step": 587 }, { "epoch": 0.58, "learning_rate": 1.666197701941721e-05, "loss": 0.069, "step": 588 }, { "epoch": 0.58, "learning_rate": 1.665005738448833e-05, "loss": 0.072, "step": 589 }, { "epoch": 0.58, "learning_rate": 1.6638120786025745e-05, "loss": 0.0728, "step": 590 }, { "epoch": 0.58, "learning_rate": 1.6626167254478337e-05, "loss": 0.0715, "step": 591 }, { "epoch": 0.58, "learning_rate": 1.6614196820338206e-05, "loss": 0.0639, "step": 592 }, { "epoch": 0.58, "learning_rate": 1.6602209514140552e-05, "loss": 0.075, "step": 593 }, { "epoch": 0.59, "learning_rate": 1.6590205366463617e-05, "loss": 0.0699, "step": 594 }, { "epoch": 0.59, "learning_rate": 1.6578184407928604e-05, "loss": 0.0601, "step": 595 }, { "epoch": 0.59, "learning_rate": 1.6566146669199603e-05, "loss": 0.0746, "step": 596 }, { "epoch": 0.59, "learning_rate": 1.6554092180983503e-05, "loss": 0.0696, "step": 597 }, { "epoch": 0.59, "learning_rate": 1.654202097402992e-05, "loss": 0.0649, "step": 598 }, { "epoch": 0.59, "learning_rate": 1.6529933079131125e-05, "loss": 0.0689, "step": 599 }, { "epoch": 0.59, "learning_rate": 1.6517828527121942e-05, "loss": 0.0648, "step": 600 }, { "epoch": 0.59, "learning_rate": 1.65057073488797e-05, "loss": 0.0688, "step": 601 }, { "epoch": 0.59, "learning_rate": 1.6493569575324137e-05, "loss": 0.0644, "step": 602 }, { "epoch": 0.59, "learning_rate": 1.6481415237417313e-05, "loss": 0.0693, "step": 603 }, { "epoch": 0.6, "learning_rate": 1.646924436616356e-05, "loss": 0.065, "step": 604 }, { "epoch": 0.6, "learning_rate": 1.6457056992609368e-05, "loss": 0.0696, "step": 605 }, { "epoch": 0.6, "learning_rate": 1.644485314784333e-05, "loss": 0.0764, "step": 606 }, { "epoch": 0.6, "learning_rate": 1.6432632862996056e-05, "loss": 0.0662, "step": 607 }, { "epoch": 0.6, "learning_rate": 1.6420396169240095e-05, "loss": 0.0654, "step": 608 }, { "epoch": 0.6, "learning_rate": 1.640814309778984e-05, "loss": 0.0607, "step": 609 }, { "epoch": 0.6, "learning_rate": 1.6395873679901474e-05, "loss": 0.072, "step": 610 }, { "epoch": 0.6, "learning_rate": 1.6383587946872885e-05, "loss": 0.0616, "step": 611 }, { "epoch": 0.6, "learning_rate": 1.6371285930043554e-05, "loss": 0.0705, "step": 612 }, { "epoch": 0.6, "learning_rate": 1.6358967660794526e-05, "loss": 0.0724, "step": 613 }, { "epoch": 0.61, "learning_rate": 1.634663317054829e-05, "loss": 0.0694, "step": 614 }, { "epoch": 0.61, "learning_rate": 1.6334282490768712e-05, "loss": 0.0713, "step": 615 }, { "epoch": 0.61, "learning_rate": 1.6321915652960963e-05, "loss": 0.07, "step": 616 }, { "epoch": 0.61, "learning_rate": 1.6309532688671428e-05, "loss": 0.0688, "step": 617 }, { "epoch": 0.61, "learning_rate": 1.629713362948763e-05, "loss": 0.0683, "step": 618 }, { "epoch": 0.61, "learning_rate": 1.6284718507038143e-05, "loss": 0.0671, "step": 619 }, { "epoch": 0.61, "learning_rate": 1.6272287352992522e-05, "loss": 0.0691, "step": 620 }, { "epoch": 0.61, "learning_rate": 1.6259840199061215e-05, "loss": 0.0718, "step": 621 }, { "epoch": 0.61, "learning_rate": 1.624737707699549e-05, "loss": 0.0688, "step": 622 }, { "epoch": 0.61, "learning_rate": 1.6234898018587336e-05, "loss": 0.0691, "step": 623 }, { "epoch": 0.62, "learning_rate": 1.62224030556694e-05, "loss": 0.0653, "step": 624 }, { "epoch": 0.62, "learning_rate": 1.62098922201149e-05, "loss": 0.0737, "step": 625 }, { "epoch": 0.62, "learning_rate": 1.6197365543837547e-05, "loss": 0.0668, "step": 626 }, { "epoch": 0.62, "learning_rate": 1.6184823058791446e-05, "loss": 0.0661, "step": 627 }, { "epoch": 0.62, "learning_rate": 1.617226479697105e-05, "loss": 0.071, "step": 628 }, { "epoch": 0.62, "learning_rate": 1.6159690790411034e-05, "loss": 0.0639, "step": 629 }, { "epoch": 0.62, "learning_rate": 1.6147101071186254e-05, "loss": 0.076, "step": 630 }, { "epoch": 0.62, "learning_rate": 1.6134495671411632e-05, "loss": 0.0616, "step": 631 }, { "epoch": 0.62, "learning_rate": 1.6121874623242104e-05, "loss": 0.0668, "step": 632 }, { "epoch": 0.62, "learning_rate": 1.6109237958872514e-05, "loss": 0.0698, "step": 633 }, { "epoch": 0.63, "learning_rate": 1.6096585710537542e-05, "loss": 0.0626, "step": 634 }, { "epoch": 0.63, "learning_rate": 1.608391791051163e-05, "loss": 0.0696, "step": 635 }, { "epoch": 0.63, "learning_rate": 1.6071234591108874e-05, "loss": 0.0723, "step": 636 }, { "epoch": 0.63, "learning_rate": 1.605853578468297e-05, "loss": 0.0678, "step": 637 }, { "epoch": 0.63, "learning_rate": 1.6045821523627117e-05, "loss": 0.0713, "step": 638 }, { "epoch": 0.63, "learning_rate": 1.6033091840373936e-05, "loss": 0.0662, "step": 639 }, { "epoch": 0.63, "learning_rate": 1.6020346767395393e-05, "loss": 0.0692, "step": 640 }, { "epoch": 0.63, "learning_rate": 1.6007586337202702e-05, "loss": 0.064, "step": 641 }, { "epoch": 0.63, "learning_rate": 1.599481058234626e-05, "loss": 0.0709, "step": 642 }, { "epoch": 0.63, "learning_rate": 1.5982019535415553e-05, "loss": 0.0634, "step": 643 }, { "epoch": 0.64, "learning_rate": 1.5969213229039073e-05, "loss": 0.0601, "step": 644 }, { "epoch": 0.64, "learning_rate": 1.5956391695884242e-05, "loss": 0.072, "step": 645 }, { "epoch": 0.64, "learning_rate": 1.5943554968657312e-05, "loss": 0.0672, "step": 646 }, { "epoch": 0.64, "learning_rate": 1.5930703080103314e-05, "loss": 0.0683, "step": 647 }, { "epoch": 0.64, "learning_rate": 1.5917836063005933e-05, "loss": 0.0724, "step": 648 }, { "epoch": 0.64, "learning_rate": 1.5904953950187458e-05, "loss": 0.0674, "step": 649 }, { "epoch": 0.64, "learning_rate": 1.589205677450868e-05, "loss": 0.0728, "step": 650 }, { "epoch": 0.64, "learning_rate": 1.5879144568868814e-05, "loss": 0.0762, "step": 651 }, { "epoch": 0.64, "learning_rate": 1.586621736620542e-05, "loss": 0.0618, "step": 652 }, { "epoch": 0.64, "learning_rate": 1.5853275199494302e-05, "loss": 0.063, "step": 653 }, { "epoch": 0.64, "learning_rate": 1.584031810174945e-05, "loss": 0.0768, "step": 654 }, { "epoch": 0.65, "learning_rate": 1.582734610602293e-05, "loss": 0.0678, "step": 655 }, { "epoch": 0.65, "learning_rate": 1.5814359245404818e-05, "loss": 0.0694, "step": 656 }, { "epoch": 0.65, "learning_rate": 1.5801357553023105e-05, "loss": 0.0697, "step": 657 }, { "epoch": 0.65, "learning_rate": 1.578834106204362e-05, "loss": 0.0707, "step": 658 }, { "epoch": 0.65, "learning_rate": 1.5775309805669932e-05, "loss": 0.0636, "step": 659 }, { "epoch": 0.65, "learning_rate": 1.5762263817143284e-05, "loss": 0.0643, "step": 660 }, { "epoch": 0.65, "learning_rate": 1.57492031297425e-05, "loss": 0.0673, "step": 661 }, { "epoch": 0.65, "learning_rate": 1.5736127776783894e-05, "loss": 0.075, "step": 662 }, { "epoch": 0.65, "learning_rate": 1.5723037791621193e-05, "loss": 0.0673, "step": 663 }, { "epoch": 0.65, "learning_rate": 1.570993320764544e-05, "loss": 0.063, "step": 664 }, { "epoch": 0.66, "learning_rate": 1.5696814058284937e-05, "loss": 0.0655, "step": 665 }, { "epoch": 0.66, "learning_rate": 1.568368037700512e-05, "loss": 0.0714, "step": 666 }, { "epoch": 0.66, "learning_rate": 1.5670532197308515e-05, "loss": 0.0634, "step": 667 }, { "epoch": 0.66, "learning_rate": 1.5657369552734608e-05, "loss": 0.0713, "step": 668 }, { "epoch": 0.66, "learning_rate": 1.5644192476859803e-05, "loss": 0.0702, "step": 669 }, { "epoch": 0.66, "learning_rate": 1.563100100329731e-05, "loss": 0.065, "step": 670 }, { "epoch": 0.66, "learning_rate": 1.5617795165697057e-05, "loss": 0.0684, "step": 671 }, { "epoch": 0.66, "learning_rate": 1.560457499774563e-05, "loss": 0.063, "step": 672 }, { "epoch": 0.66, "learning_rate": 1.5591340533166156e-05, "loss": 0.0655, "step": 673 }, { "epoch": 0.66, "learning_rate": 1.5578091805718242e-05, "loss": 0.0761, "step": 674 }, { "epoch": 0.67, "learning_rate": 1.556482884919787e-05, "loss": 0.0748, "step": 675 }, { "epoch": 0.67, "learning_rate": 1.5551551697437318e-05, "loss": 0.0646, "step": 676 }, { "epoch": 0.67, "learning_rate": 1.5538260384305076e-05, "loss": 0.068, "step": 677 }, { "epoch": 0.67, "learning_rate": 1.5524954943705765e-05, "loss": 0.0632, "step": 678 }, { "epoch": 0.67, "learning_rate": 1.5511635409580032e-05, "loss": 0.0769, "step": 679 }, { "epoch": 0.67, "learning_rate": 1.549830181590449e-05, "loss": 0.0665, "step": 680 }, { "epoch": 0.67, "learning_rate": 1.54849541966916e-05, "loss": 0.0689, "step": 681 }, { "epoch": 0.67, "learning_rate": 1.5471592585989608e-05, "loss": 0.0649, "step": 682 }, { "epoch": 0.67, "learning_rate": 1.545821701788245e-05, "loss": 0.0657, "step": 683 }, { "epoch": 0.67, "learning_rate": 1.5444827526489675e-05, "loss": 0.0652, "step": 684 }, { "epoch": 0.68, "learning_rate": 1.543142414596633e-05, "loss": 0.0808, "step": 685 }, { "epoch": 0.68, "learning_rate": 1.54180069105029e-05, "loss": 0.0693, "step": 686 }, { "epoch": 0.68, "learning_rate": 1.5404575854325226e-05, "loss": 0.0725, "step": 687 }, { "epoch": 0.68, "learning_rate": 1.539113101169438e-05, "loss": 0.0642, "step": 688 }, { "epoch": 0.68, "learning_rate": 1.5377672416906625e-05, "loss": 0.0728, "step": 689 }, { "epoch": 0.68, "learning_rate": 1.5364200104293277e-05, "loss": 0.0645, "step": 690 }, { "epoch": 0.68, "learning_rate": 1.5350714108220673e-05, "loss": 0.0698, "step": 691 }, { "epoch": 0.68, "learning_rate": 1.5337214463090044e-05, "loss": 0.0667, "step": 692 }, { "epoch": 0.68, "learning_rate": 1.5323701203337425e-05, "loss": 0.0707, "step": 693 }, { "epoch": 0.68, "learning_rate": 1.53101743634336e-05, "loss": 0.0662, "step": 694 }, { "epoch": 0.69, "learning_rate": 1.5296633977883985e-05, "loss": 0.0634, "step": 695 }, { "epoch": 0.69, "learning_rate": 1.5283080081228547e-05, "loss": 0.0711, "step": 696 }, { "epoch": 0.69, "learning_rate": 1.5269512708041725e-05, "loss": 0.0678, "step": 697 }, { "epoch": 0.69, "learning_rate": 1.5255931892932333e-05, "loss": 0.0734, "step": 698 }, { "epoch": 0.69, "learning_rate": 1.524233767054347e-05, "loss": 0.0642, "step": 699 }, { "epoch": 0.69, "learning_rate": 1.5228730075552443e-05, "loss": 0.0675, "step": 700 }, { "epoch": 0.69, "learning_rate": 1.521510914267066e-05, "loss": 0.079, "step": 701 }, { "epoch": 0.69, "learning_rate": 1.520147490664356e-05, "loss": 0.0713, "step": 702 }, { "epoch": 0.69, "learning_rate": 1.5187827402250522e-05, "loss": 0.0724, "step": 703 }, { "epoch": 0.69, "learning_rate": 1.5174166664304757e-05, "loss": 0.0655, "step": 704 }, { "epoch": 0.7, "learning_rate": 1.5160492727653241e-05, "loss": 0.0714, "step": 705 }, { "epoch": 0.7, "learning_rate": 1.5146805627176624e-05, "loss": 0.0772, "step": 706 }, { "epoch": 0.7, "learning_rate": 1.5133105397789125e-05, "loss": 0.0727, "step": 707 }, { "epoch": 0.7, "learning_rate": 1.511939207443846e-05, "loss": 0.0656, "step": 708 }, { "epoch": 0.7, "learning_rate": 1.5105665692105742e-05, "loss": 0.066, "step": 709 }, { "epoch": 0.7, "learning_rate": 1.5091926285805398e-05, "loss": 0.0719, "step": 710 }, { "epoch": 0.7, "learning_rate": 1.5078173890585081e-05, "loss": 0.0682, "step": 711 }, { "epoch": 0.7, "learning_rate": 1.5064408541525573e-05, "loss": 0.0681, "step": 712 }, { "epoch": 0.7, "learning_rate": 1.5050630273740696e-05, "loss": 0.0746, "step": 713 }, { "epoch": 0.7, "learning_rate": 1.5036839122377236e-05, "loss": 0.0748, "step": 714 }, { "epoch": 0.71, "learning_rate": 1.5023035122614837e-05, "loss": 0.0675, "step": 715 }, { "epoch": 0.71, "learning_rate": 1.500921830966592e-05, "loss": 0.0606, "step": 716 }, { "epoch": 0.71, "learning_rate": 1.499538871877559e-05, "loss": 0.075, "step": 717 }, { "epoch": 0.71, "learning_rate": 1.4981546385221545e-05, "loss": 0.0698, "step": 718 }, { "epoch": 0.71, "learning_rate": 1.4967691344313995e-05, "loss": 0.0697, "step": 719 }, { "epoch": 0.71, "learning_rate": 1.4953823631395554e-05, "loss": 0.0702, "step": 720 }, { "epoch": 0.71, "learning_rate": 1.4939943281841172e-05, "loss": 0.0643, "step": 721 }, { "epoch": 0.71, "learning_rate": 1.4926050331058029e-05, "loss": 0.0683, "step": 722 }, { "epoch": 0.71, "learning_rate": 1.4912144814485445e-05, "loss": 0.0619, "step": 723 }, { "epoch": 0.71, "learning_rate": 1.4898226767594801e-05, "loss": 0.0714, "step": 724 }, { "epoch": 0.71, "learning_rate": 1.4884296225889438e-05, "loss": 0.0722, "step": 725 }, { "epoch": 0.72, "learning_rate": 1.4870353224904572e-05, "loss": 0.0673, "step": 726 }, { "epoch": 0.72, "learning_rate": 1.4856397800207198e-05, "loss": 0.0661, "step": 727 }, { "epoch": 0.72, "learning_rate": 1.4842429987396001e-05, "loss": 0.0666, "step": 728 }, { "epoch": 0.72, "learning_rate": 1.4828449822101274e-05, "loss": 0.0643, "step": 729 }, { "epoch": 0.72, "learning_rate": 1.4814457339984811e-05, "loss": 0.0677, "step": 730 }, { "epoch": 0.72, "learning_rate": 1.4800452576739835e-05, "loss": 0.0671, "step": 731 }, { "epoch": 0.72, "learning_rate": 1.4786435568090885e-05, "loss": 0.0702, "step": 732 }, { "epoch": 0.72, "learning_rate": 1.4772406349793744e-05, "loss": 0.0647, "step": 733 }, { "epoch": 0.72, "learning_rate": 1.475836495763534e-05, "loss": 0.0621, "step": 734 }, { "epoch": 0.72, "learning_rate": 1.4744311427433658e-05, "loss": 0.0668, "step": 735 }, { "epoch": 0.73, "learning_rate": 1.4730245795037631e-05, "loss": 0.0677, "step": 736 }, { "epoch": 0.73, "learning_rate": 1.4716168096327086e-05, "loss": 0.068, "step": 737 }, { "epoch": 0.73, "learning_rate": 1.470207836721261e-05, "loss": 0.069, "step": 738 }, { "epoch": 0.73, "learning_rate": 1.4687976643635488e-05, "loss": 0.0762, "step": 739 }, { "epoch": 0.73, "learning_rate": 1.4673862961567602e-05, "loss": 0.0719, "step": 740 }, { "epoch": 0.73, "learning_rate": 1.4659737357011334e-05, "loss": 0.0616, "step": 741 }, { "epoch": 0.73, "learning_rate": 1.464559986599948e-05, "loss": 0.0647, "step": 742 }, { "epoch": 0.73, "learning_rate": 1.4631450524595164e-05, "loss": 0.0618, "step": 743 }, { "epoch": 0.73, "learning_rate": 1.4617289368891725e-05, "loss": 0.0691, "step": 744 }, { "epoch": 0.73, "learning_rate": 1.4603116435012653e-05, "loss": 0.0653, "step": 745 }, { "epoch": 0.74, "learning_rate": 1.4588931759111476e-05, "loss": 0.0686, "step": 746 }, { "epoch": 0.74, "learning_rate": 1.457473537737167e-05, "loss": 0.0633, "step": 747 }, { "epoch": 0.74, "learning_rate": 1.4560527326006584e-05, "loss": 0.0739, "step": 748 }, { "epoch": 0.74, "learning_rate": 1.4546307641259326e-05, "loss": 0.062, "step": 749 }, { "epoch": 0.74, "learning_rate": 1.4532076359402677e-05, "loss": 0.0661, "step": 750 }, { "epoch": 0.74, "learning_rate": 1.4517833516739006e-05, "loss": 0.0644, "step": 751 }, { "epoch": 0.74, "learning_rate": 1.4503579149600176e-05, "loss": 0.0657, "step": 752 }, { "epoch": 0.74, "learning_rate": 1.4489313294347438e-05, "loss": 0.0669, "step": 753 }, { "epoch": 0.74, "learning_rate": 1.4475035987371355e-05, "loss": 0.0679, "step": 754 }, { "epoch": 0.74, "learning_rate": 1.4460747265091699e-05, "loss": 0.0749, "step": 755 }, { "epoch": 0.75, "learning_rate": 1.444644716395736e-05, "loss": 0.0697, "step": 756 }, { "epoch": 0.75, "learning_rate": 1.4432135720446256e-05, "loss": 0.0631, "step": 757 }, { "epoch": 0.75, "learning_rate": 1.4417812971065241e-05, "loss": 0.0647, "step": 758 }, { "epoch": 0.75, "learning_rate": 1.4403478952350005e-05, "loss": 0.0645, "step": 759 }, { "epoch": 0.75, "learning_rate": 1.4389133700864986e-05, "loss": 0.068, "step": 760 }, { "epoch": 0.75, "learning_rate": 1.4374777253203273e-05, "loss": 0.0659, "step": 761 }, { "epoch": 0.75, "learning_rate": 1.436040964598652e-05, "loss": 0.0656, "step": 762 }, { "epoch": 0.75, "learning_rate": 1.4346030915864845e-05, "loss": 0.0704, "step": 763 }, { "epoch": 0.75, "learning_rate": 1.4331641099516738e-05, "loss": 0.0676, "step": 764 }, { "epoch": 0.75, "learning_rate": 1.431724023364897e-05, "loss": 0.0713, "step": 765 }, { "epoch": 0.76, "learning_rate": 1.43028283549965e-05, "loss": 0.064, "step": 766 }, { "epoch": 0.76, "learning_rate": 1.4288405500322378e-05, "loss": 0.0667, "step": 767 }, { "epoch": 0.76, "learning_rate": 1.4273971706417648e-05, "loss": 0.0623, "step": 768 }, { "epoch": 0.76, "learning_rate": 1.4259527010101268e-05, "loss": 0.069, "step": 769 }, { "epoch": 0.76, "learning_rate": 1.4245071448219997e-05, "loss": 0.0634, "step": 770 }, { "epoch": 0.76, "learning_rate": 1.4230605057648321e-05, "loss": 0.0678, "step": 771 }, { "epoch": 0.76, "learning_rate": 1.4216127875288342e-05, "loss": 0.0765, "step": 772 }, { "epoch": 0.76, "learning_rate": 1.4201639938069687e-05, "loss": 0.0685, "step": 773 }, { "epoch": 0.76, "learning_rate": 1.4187141282949424e-05, "loss": 0.0648, "step": 774 }, { "epoch": 0.76, "learning_rate": 1.4172631946911964e-05, "loss": 0.0717, "step": 775 }, { "epoch": 0.77, "learning_rate": 1.4158111966968953e-05, "loss": 0.0646, "step": 776 }, { "epoch": 0.77, "learning_rate": 1.41435813801592e-05, "loss": 0.0679, "step": 777 }, { "epoch": 0.77, "learning_rate": 1.412904022354856e-05, "loss": 0.0658, "step": 778 }, { "epoch": 0.77, "learning_rate": 1.4114488534229862e-05, "loss": 0.0708, "step": 779 }, { "epoch": 0.77, "learning_rate": 1.409992634932279e-05, "loss": 0.072, "step": 780 }, { "epoch": 0.77, "learning_rate": 1.4085353705973815e-05, "loss": 0.0718, "step": 781 }, { "epoch": 0.77, "learning_rate": 1.407077064135607e-05, "loss": 0.0698, "step": 782 }, { "epoch": 0.77, "learning_rate": 1.4056177192669289e-05, "loss": 0.0703, "step": 783 }, { "epoch": 0.77, "learning_rate": 1.404157339713968e-05, "loss": 0.0675, "step": 784 }, { "epoch": 0.77, "learning_rate": 1.4026959292019853e-05, "loss": 0.0672, "step": 785 }, { "epoch": 0.78, "learning_rate": 1.4012334914588715e-05, "loss": 0.0761, "step": 786 }, { "epoch": 0.78, "learning_rate": 1.3997700302151374e-05, "loss": 0.0698, "step": 787 }, { "epoch": 0.78, "learning_rate": 1.3983055492039049e-05, "loss": 0.0703, "step": 788 }, { "epoch": 0.78, "learning_rate": 1.3968400521608969e-05, "loss": 0.0726, "step": 789 }, { "epoch": 0.78, "learning_rate": 1.3953735428244284e-05, "loss": 0.0657, "step": 790 }, { "epoch": 0.78, "learning_rate": 1.3939060249353966e-05, "loss": 0.0625, "step": 791 }, { "epoch": 0.78, "learning_rate": 1.3924375022372712e-05, "loss": 0.0756, "step": 792 }, { "epoch": 0.78, "learning_rate": 1.3909679784760853e-05, "loss": 0.0714, "step": 793 }, { "epoch": 0.78, "learning_rate": 1.3894974574004254e-05, "loss": 0.0656, "step": 794 }, { "epoch": 0.78, "learning_rate": 1.388025942761422e-05, "loss": 0.0725, "step": 795 }, { "epoch": 0.79, "learning_rate": 1.3865534383127406e-05, "loss": 0.069, "step": 796 }, { "epoch": 0.79, "learning_rate": 1.3850799478105708e-05, "loss": 0.0676, "step": 797 }, { "epoch": 0.79, "learning_rate": 1.3836054750136182e-05, "loss": 0.0652, "step": 798 }, { "epoch": 0.79, "learning_rate": 1.382130023683093e-05, "loss": 0.062, "step": 799 }, { "epoch": 0.79, "learning_rate": 1.380653597582703e-05, "loss": 0.0699, "step": 800 }, { "epoch": 0.79, "learning_rate": 1.3791762004786422e-05, "loss": 0.0636, "step": 801 }, { "epoch": 0.79, "learning_rate": 1.3776978361395803e-05, "loss": 0.0656, "step": 802 }, { "epoch": 0.79, "learning_rate": 1.3762185083366557e-05, "loss": 0.0667, "step": 803 }, { "epoch": 0.79, "learning_rate": 1.3747382208434641e-05, "loss": 0.0661, "step": 804 }, { "epoch": 0.79, "learning_rate": 1.3732569774360486e-05, "loss": 0.0685, "step": 805 }, { "epoch": 0.79, "learning_rate": 1.3717747818928917e-05, "loss": 0.0675, "step": 806 }, { "epoch": 0.8, "learning_rate": 1.3702916379949045e-05, "loss": 0.0661, "step": 807 }, { "epoch": 0.8, "learning_rate": 1.368807549525416e-05, "loss": 0.06, "step": 808 }, { "epoch": 0.8, "learning_rate": 1.3673225202701664e-05, "loss": 0.0648, "step": 809 }, { "epoch": 0.8, "learning_rate": 1.3658365540172948e-05, "loss": 0.0627, "step": 810 }, { "epoch": 0.8, "learning_rate": 1.364349654557331e-05, "loss": 0.0713, "step": 811 }, { "epoch": 0.8, "learning_rate": 1.3628618256831848e-05, "loss": 0.0649, "step": 812 }, { "epoch": 0.8, "learning_rate": 1.3613730711901365e-05, "loss": 0.0714, "step": 813 }, { "epoch": 0.8, "learning_rate": 1.359883394875829e-05, "loss": 0.0595, "step": 814 }, { "epoch": 0.8, "learning_rate": 1.358392800540255e-05, "loss": 0.0701, "step": 815 }, { "epoch": 0.8, "learning_rate": 1.35690129198575e-05, "loss": 0.0602, "step": 816 }, { "epoch": 0.81, "learning_rate": 1.3554088730169814e-05, "loss": 0.0665, "step": 817 }, { "epoch": 0.81, "learning_rate": 1.3539155474409387e-05, "loss": 0.067, "step": 818 }, { "epoch": 0.81, "learning_rate": 1.3524213190669238e-05, "loss": 0.0603, "step": 819 }, { "epoch": 0.81, "learning_rate": 1.3509261917065426e-05, "loss": 0.0692, "step": 820 }, { "epoch": 0.81, "learning_rate": 1.349430169173693e-05, "loss": 0.0666, "step": 821 }, { "epoch": 0.81, "learning_rate": 1.3479332552845573e-05, "loss": 0.0618, "step": 822 }, { "epoch": 0.81, "learning_rate": 1.3464354538575908e-05, "loss": 0.0799, "step": 823 }, { "epoch": 0.81, "learning_rate": 1.3449367687135134e-05, "loss": 0.0682, "step": 824 }, { "epoch": 0.81, "learning_rate": 1.343437203675299e-05, "loss": 0.0702, "step": 825 }, { "epoch": 0.81, "learning_rate": 1.3419367625681656e-05, "loss": 0.0726, "step": 826 }, { "epoch": 0.82, "learning_rate": 1.3404354492195671e-05, "loss": 0.0692, "step": 827 }, { "epoch": 0.82, "learning_rate": 1.3389332674591811e-05, "loss": 0.0664, "step": 828 }, { "epoch": 0.82, "learning_rate": 1.3374302211189014e-05, "loss": 0.066, "step": 829 }, { "epoch": 0.82, "learning_rate": 1.3359263140328265e-05, "loss": 0.0699, "step": 830 }, { "epoch": 0.82, "learning_rate": 1.334421550037251e-05, "loss": 0.0703, "step": 831 }, { "epoch": 0.82, "learning_rate": 1.3329159329706554e-05, "loss": 0.0699, "step": 832 }, { "epoch": 0.82, "learning_rate": 1.331409466673696e-05, "loss": 0.0717, "step": 833 }, { "epoch": 0.82, "learning_rate": 1.3299021549891958e-05, "loss": 0.0693, "step": 834 }, { "epoch": 0.82, "learning_rate": 1.3283940017621337e-05, "loss": 0.0723, "step": 835 }, { "epoch": 0.82, "learning_rate": 1.3268850108396359e-05, "loss": 0.0643, "step": 836 }, { "epoch": 0.83, "learning_rate": 1.3253751860709647e-05, "loss": 0.0671, "step": 837 }, { "epoch": 0.83, "learning_rate": 1.3238645313075104e-05, "loss": 0.0685, "step": 838 }, { "epoch": 0.83, "learning_rate": 1.3223530504027802e-05, "loss": 0.0641, "step": 839 }, { "epoch": 0.83, "learning_rate": 1.320840747212388e-05, "loss": 0.0699, "step": 840 }, { "epoch": 0.83, "learning_rate": 1.3193276255940464e-05, "loss": 0.0674, "step": 841 }, { "epoch": 0.83, "learning_rate": 1.3178136894075546e-05, "loss": 0.0674, "step": 842 }, { "epoch": 0.83, "learning_rate": 1.3162989425147902e-05, "loss": 0.0745, "step": 843 }, { "epoch": 0.83, "learning_rate": 1.3147833887796992e-05, "loss": 0.073, "step": 844 }, { "epoch": 0.83, "learning_rate": 1.313267032068285e-05, "loss": 0.0634, "step": 845 }, { "epoch": 0.83, "learning_rate": 1.3117498762485998e-05, "loss": 0.068, "step": 846 }, { "epoch": 0.84, "learning_rate": 1.3102319251907341e-05, "loss": 0.0722, "step": 847 }, { "epoch": 0.84, "learning_rate": 1.3087131827668068e-05, "loss": 0.0604, "step": 848 }, { "epoch": 0.84, "learning_rate": 1.307193652850956e-05, "loss": 0.0659, "step": 849 }, { "epoch": 0.84, "learning_rate": 1.3056733393193279e-05, "loss": 0.0659, "step": 850 }, { "epoch": 0.84, "learning_rate": 1.3041522460500684e-05, "loss": 0.0647, "step": 851 }, { "epoch": 0.84, "learning_rate": 1.3026303769233112e-05, "loss": 0.0705, "step": 852 }, { "epoch": 0.84, "learning_rate": 1.3011077358211707e-05, "loss": 0.068, "step": 853 }, { "epoch": 0.84, "learning_rate": 1.2995843266277293e-05, "loss": 0.063, "step": 854 }, { "epoch": 0.84, "learning_rate": 1.2980601532290292e-05, "loss": 0.0717, "step": 855 }, { "epoch": 0.84, "learning_rate": 1.2965352195130621e-05, "loss": 0.0691, "step": 856 }, { "epoch": 0.85, "learning_rate": 1.2950095293697586e-05, "loss": 0.0665, "step": 857 }, { "epoch": 0.85, "learning_rate": 1.2934830866909797e-05, "loss": 0.0666, "step": 858 }, { "epoch": 0.85, "learning_rate": 1.2919558953705055e-05, "loss": 0.0703, "step": 859 }, { "epoch": 0.85, "learning_rate": 1.2904279593040259e-05, "loss": 0.0585, "step": 860 }, { "epoch": 0.85, "learning_rate": 1.2888992823891307e-05, "loss": 0.0671, "step": 861 }, { "epoch": 0.85, "learning_rate": 1.287369868525299e-05, "loss": 0.0675, "step": 862 }, { "epoch": 0.85, "learning_rate": 1.2858397216138905e-05, "loss": 0.0716, "step": 863 }, { "epoch": 0.85, "learning_rate": 1.2843088455581342e-05, "loss": 0.0664, "step": 864 }, { "epoch": 0.85, "learning_rate": 1.28277724426312e-05, "loss": 0.0652, "step": 865 }, { "epoch": 0.85, "learning_rate": 1.2812449216357863e-05, "loss": 0.0629, "step": 866 }, { "epoch": 0.86, "learning_rate": 1.2797118815849129e-05, "loss": 0.0612, "step": 867 }, { "epoch": 0.86, "learning_rate": 1.2781781280211089e-05, "loss": 0.0641, "step": 868 }, { "epoch": 0.86, "learning_rate": 1.2766436648568041e-05, "loss": 0.0691, "step": 869 }, { "epoch": 0.86, "learning_rate": 1.275108496006238e-05, "loss": 0.0686, "step": 870 }, { "epoch": 0.86, "learning_rate": 1.2735726253854494e-05, "loss": 0.0714, "step": 871 }, { "epoch": 0.86, "learning_rate": 1.2720360569122693e-05, "loss": 0.0732, "step": 872 }, { "epoch": 0.86, "learning_rate": 1.270498794506307e-05, "loss": 0.0728, "step": 873 }, { "epoch": 0.86, "learning_rate": 1.2689608420889426e-05, "loss": 0.06, "step": 874 }, { "epoch": 0.86, "learning_rate": 1.2674222035833163e-05, "loss": 0.0665, "step": 875 }, { "epoch": 0.86, "learning_rate": 1.2658828829143191e-05, "loss": 0.0651, "step": 876 }, { "epoch": 0.86, "learning_rate": 1.264342884008581e-05, "loss": 0.0661, "step": 877 }, { "epoch": 0.87, "learning_rate": 1.2628022107944624e-05, "loss": 0.065, "step": 878 }, { "epoch": 0.87, "learning_rate": 1.2612608672020442e-05, "loss": 0.0678, "step": 879 }, { "epoch": 0.87, "learning_rate": 1.259718857163117e-05, "loss": 0.0619, "step": 880 }, { "epoch": 0.87, "learning_rate": 1.2581761846111718e-05, "loss": 0.067, "step": 881 }, { "epoch": 0.87, "learning_rate": 1.2566328534813893e-05, "loss": 0.0657, "step": 882 }, { "epoch": 0.87, "learning_rate": 1.2550888677106299e-05, "loss": 0.0698, "step": 883 }, { "epoch": 0.87, "learning_rate": 1.2535442312374247e-05, "loss": 0.067, "step": 884 }, { "epoch": 0.87, "learning_rate": 1.2519989480019636e-05, "loss": 0.0613, "step": 885 }, { "epoch": 0.87, "learning_rate": 1.2504530219460877e-05, "loss": 0.0716, "step": 886 }, { "epoch": 0.87, "learning_rate": 1.2489064570132764e-05, "loss": 0.0721, "step": 887 }, { "epoch": 0.88, "learning_rate": 1.2473592571486397e-05, "loss": 0.0641, "step": 888 }, { "epoch": 0.88, "learning_rate": 1.2458114262989072e-05, "loss": 0.071, "step": 889 }, { "epoch": 0.88, "learning_rate": 1.2442629684124174e-05, "loss": 0.0699, "step": 890 }, { "epoch": 0.88, "learning_rate": 1.2427138874391094e-05, "loss": 0.074, "step": 891 }, { "epoch": 0.88, "learning_rate": 1.2411641873305104e-05, "loss": 0.0649, "step": 892 }, { "epoch": 0.88, "learning_rate": 1.2396138720397283e-05, "loss": 0.0644, "step": 893 }, { "epoch": 0.88, "learning_rate": 1.2380629455214392e-05, "loss": 0.0722, "step": 894 }, { "epoch": 0.88, "learning_rate": 1.2365114117318793e-05, "loss": 0.0644, "step": 895 }, { "epoch": 0.88, "learning_rate": 1.2349592746288327e-05, "loss": 0.0694, "step": 896 }, { "epoch": 0.88, "learning_rate": 1.2334065381716231e-05, "loss": 0.0691, "step": 897 }, { "epoch": 0.89, "learning_rate": 1.2318532063211036e-05, "loss": 0.0697, "step": 898 }, { "epoch": 0.89, "learning_rate": 1.2302992830396457e-05, "loss": 0.0652, "step": 899 }, { "epoch": 0.89, "learning_rate": 1.2287447722911288e-05, "loss": 0.065, "step": 900 }, { "epoch": 0.89, "learning_rate": 1.2271896780409321e-05, "loss": 0.0641, "step": 901 }, { "epoch": 0.89, "learning_rate": 1.2256340042559226e-05, "loss": 0.0674, "step": 902 }, { "epoch": 0.89, "learning_rate": 1.2240777549044456e-05, "loss": 0.0661, "step": 903 }, { "epoch": 0.89, "learning_rate": 1.2225209339563144e-05, "loss": 0.0683, "step": 904 }, { "epoch": 0.89, "learning_rate": 1.2209635453828015e-05, "loss": 0.0608, "step": 905 }, { "epoch": 0.89, "learning_rate": 1.2194055931566257e-05, "loss": 0.0693, "step": 906 }, { "epoch": 0.89, "learning_rate": 1.2178470812519451e-05, "loss": 0.0604, "step": 907 }, { "epoch": 0.9, "learning_rate": 1.2162880136443447e-05, "loss": 0.0651, "step": 908 }, { "epoch": 0.9, "learning_rate": 1.2147283943108267e-05, "loss": 0.0677, "step": 909 }, { "epoch": 0.9, "learning_rate": 1.2131682272298016e-05, "loss": 0.0638, "step": 910 }, { "epoch": 0.9, "learning_rate": 1.2116075163810765e-05, "loss": 0.062, "step": 911 }, { "epoch": 0.9, "learning_rate": 1.2100462657458459e-05, "loss": 0.0736, "step": 912 }, { "epoch": 0.9, "learning_rate": 1.208484479306681e-05, "loss": 0.0687, "step": 913 }, { "epoch": 0.9, "learning_rate": 1.20692216104752e-05, "loss": 0.0692, "step": 914 }, { "epoch": 0.9, "learning_rate": 1.2053593149536576e-05, "loss": 0.0695, "step": 915 }, { "epoch": 0.9, "learning_rate": 1.2037959450117347e-05, "loss": 0.0714, "step": 916 }, { "epoch": 0.9, "learning_rate": 1.2022320552097291e-05, "loss": 0.0626, "step": 917 }, { "epoch": 0.91, "learning_rate": 1.200667649536944e-05, "loss": 0.0625, "step": 918 }, { "epoch": 0.91, "learning_rate": 1.199102731983999e-05, "loss": 0.0636, "step": 919 }, { "epoch": 0.91, "learning_rate": 1.1975373065428193e-05, "loss": 0.0806, "step": 920 }, { "epoch": 0.91, "learning_rate": 1.1959713772066258e-05, "loss": 0.0669, "step": 921 }, { "epoch": 0.91, "learning_rate": 1.1944049479699244e-05, "loss": 0.0618, "step": 922 }, { "epoch": 0.91, "learning_rate": 1.1928380228284965e-05, "loss": 0.0683, "step": 923 }, { "epoch": 0.91, "learning_rate": 1.1912706057793883e-05, "loss": 0.0686, "step": 924 }, { "epoch": 0.91, "learning_rate": 1.1897027008209015e-05, "loss": 0.0742, "step": 925 }, { "epoch": 0.91, "learning_rate": 1.188134311952581e-05, "loss": 0.0668, "step": 926 }, { "epoch": 0.91, "learning_rate": 1.1865654431752076e-05, "loss": 0.0598, "step": 927 }, { "epoch": 0.92, "learning_rate": 1.1849960984907853e-05, "loss": 0.0644, "step": 928 }, { "epoch": 0.92, "learning_rate": 1.1834262819025326e-05, "loss": 0.0639, "step": 929 }, { "epoch": 0.92, "learning_rate": 1.1818559974148715e-05, "loss": 0.069, "step": 930 }, { "epoch": 0.92, "learning_rate": 1.1802852490334173e-05, "loss": 0.0637, "step": 931 }, { "epoch": 0.92, "learning_rate": 1.1787140407649695e-05, "loss": 0.0663, "step": 932 }, { "epoch": 0.92, "learning_rate": 1.1771423766175e-05, "loss": 0.0707, "step": 933 }, { "epoch": 0.92, "learning_rate": 1.1755702606001435e-05, "loss": 0.0686, "step": 934 }, { "epoch": 0.92, "learning_rate": 1.1739976967231881e-05, "loss": 0.0689, "step": 935 }, { "epoch": 0.92, "learning_rate": 1.1724246889980638e-05, "loss": 0.0702, "step": 936 }, { "epoch": 0.92, "learning_rate": 1.1708512414373325e-05, "loss": 0.0723, "step": 937 }, { "epoch": 0.93, "learning_rate": 1.169277358054679e-05, "loss": 0.0736, "step": 938 }, { "epoch": 0.93, "learning_rate": 1.167703042864899e-05, "loss": 0.069, "step": 939 }, { "epoch": 0.93, "learning_rate": 1.16612829988389e-05, "loss": 0.0642, "step": 940 }, { "epoch": 0.93, "learning_rate": 1.164553133128641e-05, "loss": 0.0689, "step": 941 }, { "epoch": 0.93, "learning_rate": 1.1629775466172214e-05, "loss": 0.0656, "step": 942 }, { "epoch": 0.93, "learning_rate": 1.1614015443687723e-05, "loss": 0.0704, "step": 943 }, { "epoch": 0.93, "learning_rate": 1.1598251304034938e-05, "loss": 0.0765, "step": 944 }, { "epoch": 0.93, "learning_rate": 1.1582483087426377e-05, "loss": 0.0783, "step": 945 }, { "epoch": 0.93, "learning_rate": 1.156671083408495e-05, "loss": 0.0724, "step": 946 }, { "epoch": 0.93, "learning_rate": 1.1550934584243868e-05, "loss": 0.0642, "step": 947 }, { "epoch": 0.93, "learning_rate": 1.1535154378146538e-05, "loss": 0.071, "step": 948 }, { "epoch": 0.94, "learning_rate": 1.1519370256046449e-05, "loss": 0.07, "step": 949 }, { "epoch": 0.94, "learning_rate": 1.150358225820709e-05, "loss": 0.0714, "step": 950 }, { "epoch": 0.94, "learning_rate": 1.1487790424901833e-05, "loss": 0.0689, "step": 951 }, { "epoch": 0.94, "learning_rate": 1.147199479641383e-05, "loss": 0.0626, "step": 952 }, { "epoch": 0.94, "learning_rate": 1.1456195413035922e-05, "loss": 0.0651, "step": 953 }, { "epoch": 0.94, "learning_rate": 1.1440392315070523e-05, "loss": 0.0698, "step": 954 }, { "epoch": 0.94, "learning_rate": 1.1424585542829522e-05, "loss": 0.0709, "step": 955 }, { "epoch": 0.94, "learning_rate": 1.1408775136634185e-05, "loss": 0.066, "step": 956 }, { "epoch": 0.94, "learning_rate": 1.1392961136815046e-05, "loss": 0.0683, "step": 957 }, { "epoch": 0.94, "learning_rate": 1.1377143583711797e-05, "loss": 0.0681, "step": 958 }, { "epoch": 0.95, "learning_rate": 1.136132251767321e-05, "loss": 0.0694, "step": 959 }, { "epoch": 0.95, "learning_rate": 1.1345497979057005e-05, "loss": 0.0627, "step": 960 }, { "epoch": 0.95, "learning_rate": 1.132967000822977e-05, "loss": 0.0677, "step": 961 }, { "epoch": 0.95, "learning_rate": 1.1313838645566837e-05, "loss": 0.0676, "step": 962 }, { "epoch": 0.95, "learning_rate": 1.1298003931452202e-05, "loss": 0.0618, "step": 963 }, { "epoch": 0.95, "learning_rate": 1.1282165906278402e-05, "loss": 0.0695, "step": 964 }, { "epoch": 0.95, "learning_rate": 1.1266324610446425e-05, "loss": 0.062, "step": 965 }, { "epoch": 0.95, "learning_rate": 1.1250480084365598e-05, "loss": 0.0684, "step": 966 }, { "epoch": 0.95, "learning_rate": 1.123463236845349e-05, "loss": 0.0677, "step": 967 }, { "epoch": 0.95, "learning_rate": 1.1218781503135805e-05, "loss": 0.0696, "step": 968 }, { "epoch": 0.96, "learning_rate": 1.1202927528846286e-05, "loss": 0.0589, "step": 969 }, { "epoch": 0.96, "learning_rate": 1.1187070486026602e-05, "loss": 0.0779, "step": 970 }, { "epoch": 0.96, "learning_rate": 1.1171210415126248e-05, "loss": 0.0704, "step": 971 }, { "epoch": 0.96, "learning_rate": 1.1155347356602448e-05, "loss": 0.0701, "step": 972 }, { "epoch": 0.96, "learning_rate": 1.1139481350920046e-05, "loss": 0.0655, "step": 973 }, { "epoch": 0.96, "learning_rate": 1.1123612438551406e-05, "loss": 0.0631, "step": 974 }, { "epoch": 0.96, "learning_rate": 1.1107740659976297e-05, "loss": 0.0667, "step": 975 }, { "epoch": 0.96, "learning_rate": 1.1091866055681806e-05, "loss": 0.0695, "step": 976 }, { "epoch": 0.96, "learning_rate": 1.1075988666162232e-05, "loss": 0.0695, "step": 977 }, { "epoch": 0.96, "learning_rate": 1.1060108531918972e-05, "loss": 0.0661, "step": 978 }, { "epoch": 0.97, "learning_rate": 1.104422569346043e-05, "loss": 0.0686, "step": 979 }, { "epoch": 0.97, "learning_rate": 1.10283401913019e-05, "loss": 0.0706, "step": 980 }, { "epoch": 0.97, "learning_rate": 1.1012452065965481e-05, "loss": 0.0638, "step": 981 }, { "epoch": 0.97, "learning_rate": 1.099656135797996e-05, "loss": 0.0663, "step": 982 }, { "epoch": 0.97, "learning_rate": 1.0980668107880707e-05, "loss": 0.069, "step": 983 }, { "epoch": 0.97, "learning_rate": 1.0964772356209581e-05, "loss": 0.07, "step": 984 }, { "epoch": 0.97, "learning_rate": 1.094887414351482e-05, "loss": 0.0588, "step": 985 }, { "epoch": 0.97, "learning_rate": 1.0932973510350944e-05, "loss": 0.0722, "step": 986 }, { "epoch": 0.97, "learning_rate": 1.0917070497278644e-05, "loss": 0.0671, "step": 987 }, { "epoch": 0.97, "learning_rate": 1.090116514486468e-05, "loss": 0.0604, "step": 988 }, { "epoch": 0.98, "learning_rate": 1.0885257493681784e-05, "loss": 0.0724, "step": 989 }, { "epoch": 0.98, "learning_rate": 1.0869347584308551e-05, "loss": 0.0638, "step": 990 }, { "epoch": 0.98, "learning_rate": 1.0853435457329334e-05, "loss": 0.0612, "step": 991 }, { "epoch": 0.98, "learning_rate": 1.0837521153334143e-05, "loss": 0.0565, "step": 992 }, { "epoch": 0.98, "learning_rate": 1.0821604712918544e-05, "loss": 0.0611, "step": 993 }, { "epoch": 0.98, "learning_rate": 1.0805686176683547e-05, "loss": 0.0728, "step": 994 }, { "epoch": 0.98, "learning_rate": 1.0789765585235518e-05, "loss": 0.0718, "step": 995 }, { "epoch": 0.98, "learning_rate": 1.0773842979186057e-05, "loss": 0.0617, "step": 996 }, { "epoch": 0.98, "learning_rate": 1.0757918399151905e-05, "loss": 0.0665, "step": 997 }, { "epoch": 0.98, "learning_rate": 1.0741991885754842e-05, "loss": 0.0674, "step": 998 }, { "epoch": 0.99, "learning_rate": 1.0726063479621574e-05, "loss": 0.0685, "step": 999 }, { "epoch": 0.99, "learning_rate": 1.0710133221383641e-05, "loss": 0.0734, "step": 1000 }, { "epoch": 0.99, "learning_rate": 1.0694201151677307e-05, "loss": 0.0752, "step": 1001 }, { "epoch": 0.99, "learning_rate": 1.0678267311143447e-05, "loss": 0.0658, "step": 1002 }, { "epoch": 0.99, "learning_rate": 1.0662331740427465e-05, "loss": 0.0655, "step": 1003 }, { "epoch": 0.99, "learning_rate": 1.0646394480179173e-05, "loss": 0.0691, "step": 1004 }, { "epoch": 0.99, "learning_rate": 1.0630455571052697e-05, "loss": 0.0669, "step": 1005 }, { "epoch": 0.99, "learning_rate": 1.0614515053706367e-05, "loss": 0.0732, "step": 1006 }, { "epoch": 0.99, "learning_rate": 1.059857296880261e-05, "loss": 0.0632, "step": 1007 }, { "epoch": 0.99, "learning_rate": 1.0582629357007857e-05, "loss": 0.0578, "step": 1008 }, { "epoch": 1.0, "learning_rate": 1.0566684258992437e-05, "loss": 0.0669, "step": 1009 }, { "epoch": 1.0, "learning_rate": 1.0550737715430463e-05, "loss": 0.0727, "step": 1010 }, { "epoch": 1.0, "learning_rate": 1.053478976699974e-05, "loss": 0.0675, "step": 1011 }, { "epoch": 1.0, "learning_rate": 1.0518840454381654e-05, "loss": 0.0698, "step": 1012 }, { "epoch": 1.0, "learning_rate": 1.0502889818261075e-05, "loss": 0.0691, "step": 1013 }, { "epoch": 1.0, "learning_rate": 1.0486937899326241e-05, "loss": 0.0645, "step": 1014 }, { "epoch": 1.0, "learning_rate": 1.0470984738268675e-05, "loss": 0.0533, "step": 1015 }, { "epoch": 1.0, "learning_rate": 1.0455030375783058e-05, "loss": 0.0585, "step": 1016 }, { "epoch": 1.0, "learning_rate": 1.0439074852567137e-05, "loss": 0.0592, "step": 1017 }, { "epoch": 1.0, "learning_rate": 1.0423118209321624e-05, "loss": 0.0537, "step": 1018 }, { "epoch": 1.0, "learning_rate": 1.0407160486750087e-05, "loss": 0.0512, "step": 1019 }, { "epoch": 1.01, "learning_rate": 1.0391201725558842e-05, "loss": 0.059, "step": 1020 }, { "epoch": 1.01, "learning_rate": 1.0375241966456862e-05, "loss": 0.0578, "step": 1021 }, { "epoch": 1.01, "learning_rate": 1.0359281250155662e-05, "loss": 0.0586, "step": 1022 }, { "epoch": 1.01, "learning_rate": 1.0343319617369196e-05, "loss": 0.0603, "step": 1023 }, { "epoch": 1.01, "learning_rate": 1.032735710881376e-05, "loss": 0.0633, "step": 1024 }, { "epoch": 1.01, "learning_rate": 1.0311393765207884e-05, "loss": 0.0557, "step": 1025 }, { "epoch": 1.01, "learning_rate": 1.0295429627272228e-05, "loss": 0.0569, "step": 1026 }, { "epoch": 1.01, "learning_rate": 1.0279464735729472e-05, "loss": 0.0478, "step": 1027 }, { "epoch": 1.01, "learning_rate": 1.0263499131304228e-05, "loss": 0.0588, "step": 1028 }, { "epoch": 1.01, "learning_rate": 1.024753285472292e-05, "loss": 0.0572, "step": 1029 }, { "epoch": 1.02, "learning_rate": 1.0231565946713688e-05, "loss": 0.0544, "step": 1030 }, { "epoch": 1.02, "learning_rate": 1.0215598448006284e-05, "loss": 0.0469, "step": 1031 }, { "epoch": 1.02, "learning_rate": 1.0199630399331965e-05, "loss": 0.0642, "step": 1032 }, { "epoch": 1.02, "learning_rate": 1.0183661841423394e-05, "loss": 0.0526, "step": 1033 }, { "epoch": 1.02, "learning_rate": 1.0167692815014527e-05, "loss": 0.0541, "step": 1034 }, { "epoch": 1.02, "learning_rate": 1.0151723360840522e-05, "loss": 0.0535, "step": 1035 }, { "epoch": 1.02, "learning_rate": 1.0135753519637625e-05, "loss": 0.0494, "step": 1036 }, { "epoch": 1.02, "learning_rate": 1.011978333214307e-05, "loss": 0.0614, "step": 1037 }, { "epoch": 1.02, "learning_rate": 1.0103812839094969e-05, "loss": 0.047, "step": 1038 }, { "epoch": 1.02, "learning_rate": 1.0087842081232225e-05, "loss": 0.0515, "step": 1039 }, { "epoch": 1.03, "learning_rate": 1.0071871099294401e-05, "loss": 0.0568, "step": 1040 }, { "epoch": 1.03, "learning_rate": 1.0055899934021649e-05, "loss": 0.0598, "step": 1041 }, { "epoch": 1.03, "learning_rate": 1.0039928626154576e-05, "loss": 0.0522, "step": 1042 }, { "epoch": 1.03, "learning_rate": 1.0023957216434157e-05, "loss": 0.0584, "step": 1043 }, { "epoch": 1.03, "learning_rate": 1.0007985745601626e-05, "loss": 0.0615, "step": 1044 }, { "epoch": 1.03, "learning_rate": 9.992014254398377e-06, "loss": 0.0612, "step": 1045 }, { "epoch": 1.03, "learning_rate": 9.976042783565843e-06, "loss": 0.052, "step": 1046 }, { "epoch": 1.03, "learning_rate": 9.960071373845427e-06, "loss": 0.0525, "step": 1047 }, { "epoch": 1.03, "learning_rate": 9.944100065978351e-06, "loss": 0.061, "step": 1048 }, { "epoch": 1.03, "learning_rate": 9.9281289007056e-06, "loss": 0.0533, "step": 1049 }, { "epoch": 1.04, "learning_rate": 9.912157918767782e-06, "loss": 0.0495, "step": 1050 }, { "epoch": 1.04, "learning_rate": 9.896187160905033e-06, "loss": 0.0532, "step": 1051 }, { "epoch": 1.04, "learning_rate": 9.880216667856935e-06, "loss": 0.0558, "step": 1052 }, { "epoch": 1.04, "learning_rate": 9.864246480362376e-06, "loss": 0.0622, "step": 1053 }, { "epoch": 1.04, "learning_rate": 9.848276639159481e-06, "loss": 0.0542, "step": 1054 }, { "epoch": 1.04, "learning_rate": 9.832307184985475e-06, "loss": 0.0638, "step": 1055 }, { "epoch": 1.04, "learning_rate": 9.816338158576611e-06, "loss": 0.0541, "step": 1056 }, { "epoch": 1.04, "learning_rate": 9.800369600668036e-06, "loss": 0.053, "step": 1057 }, { "epoch": 1.04, "learning_rate": 9.78440155199372e-06, "loss": 0.0456, "step": 1058 }, { "epoch": 1.04, "learning_rate": 9.768434053286315e-06, "loss": 0.0546, "step": 1059 }, { "epoch": 1.05, "learning_rate": 9.752467145277082e-06, "loss": 0.0621, "step": 1060 }, { "epoch": 1.05, "learning_rate": 9.736500868695773e-06, "loss": 0.0517, "step": 1061 }, { "epoch": 1.05, "learning_rate": 9.720535264270529e-06, "loss": 0.0521, "step": 1062 }, { "epoch": 1.05, "learning_rate": 9.704570372727775e-06, "loss": 0.0584, "step": 1063 }, { "epoch": 1.05, "learning_rate": 9.688606234792118e-06, "loss": 0.0558, "step": 1064 }, { "epoch": 1.05, "learning_rate": 9.672642891186243e-06, "loss": 0.0533, "step": 1065 }, { "epoch": 1.05, "learning_rate": 9.656680382630807e-06, "loss": 0.0462, "step": 1066 }, { "epoch": 1.05, "learning_rate": 9.640718749844343e-06, "loss": 0.0548, "step": 1067 }, { "epoch": 1.05, "learning_rate": 9.624758033543143e-06, "loss": 0.0486, "step": 1068 }, { "epoch": 1.05, "learning_rate": 9.60879827444116e-06, "loss": 0.0557, "step": 1069 }, { "epoch": 1.06, "learning_rate": 9.592839513249918e-06, "loss": 0.0598, "step": 1070 }, { "epoch": 1.06, "learning_rate": 9.576881790678378e-06, "loss": 0.0573, "step": 1071 }, { "epoch": 1.06, "learning_rate": 9.560925147432866e-06, "loss": 0.0591, "step": 1072 }, { "epoch": 1.06, "learning_rate": 9.544969624216944e-06, "loss": 0.052, "step": 1073 }, { "epoch": 1.06, "learning_rate": 9.529015261731326e-06, "loss": 0.0517, "step": 1074 }, { "epoch": 1.06, "learning_rate": 9.513062100673759e-06, "loss": 0.0549, "step": 1075 }, { "epoch": 1.06, "learning_rate": 9.497110181738928e-06, "loss": 0.0495, "step": 1076 }, { "epoch": 1.06, "learning_rate": 9.481159545618351e-06, "loss": 0.0557, "step": 1077 }, { "epoch": 1.06, "learning_rate": 9.465210233000262e-06, "loss": 0.0582, "step": 1078 }, { "epoch": 1.06, "learning_rate": 9.449262284569542e-06, "loss": 0.0473, "step": 1079 }, { "epoch": 1.07, "learning_rate": 9.433315741007564e-06, "loss": 0.0513, "step": 1080 }, { "epoch": 1.07, "learning_rate": 9.417370642992146e-06, "loss": 0.0582, "step": 1081 }, { "epoch": 1.07, "learning_rate": 9.401427031197392e-06, "loss": 0.057, "step": 1082 }, { "epoch": 1.07, "learning_rate": 9.385484946293636e-06, "loss": 0.0464, "step": 1083 }, { "epoch": 1.07, "learning_rate": 9.369544428947303e-06, "loss": 0.0548, "step": 1084 }, { "epoch": 1.07, "learning_rate": 9.353605519820828e-06, "loss": 0.0555, "step": 1085 }, { "epoch": 1.07, "learning_rate": 9.337668259572541e-06, "loss": 0.05, "step": 1086 }, { "epoch": 1.07, "learning_rate": 9.321732688856558e-06, "loss": 0.0626, "step": 1087 }, { "epoch": 1.07, "learning_rate": 9.3057988483227e-06, "loss": 0.0594, "step": 1088 }, { "epoch": 1.07, "learning_rate": 9.28986677861636e-06, "loss": 0.0497, "step": 1089 }, { "epoch": 1.07, "learning_rate": 9.273936520378428e-06, "loss": 0.0534, "step": 1090 }, { "epoch": 1.08, "learning_rate": 9.258008114245158e-06, "loss": 0.0525, "step": 1091 }, { "epoch": 1.08, "learning_rate": 9.242081600848097e-06, "loss": 0.0527, "step": 1092 }, { "epoch": 1.08, "learning_rate": 9.226157020813945e-06, "loss": 0.0556, "step": 1093 }, { "epoch": 1.08, "learning_rate": 9.210234414764484e-06, "loss": 0.0497, "step": 1094 }, { "epoch": 1.08, "learning_rate": 9.194313823316457e-06, "loss": 0.0551, "step": 1095 }, { "epoch": 1.08, "learning_rate": 9.17839528708146e-06, "loss": 0.0485, "step": 1096 }, { "epoch": 1.08, "learning_rate": 9.16247884666586e-06, "loss": 0.0534, "step": 1097 }, { "epoch": 1.08, "learning_rate": 9.146564542670668e-06, "loss": 0.058, "step": 1098 }, { "epoch": 1.08, "learning_rate": 9.130652415691452e-06, "loss": 0.0517, "step": 1099 }, { "epoch": 1.08, "learning_rate": 9.114742506318216e-06, "loss": 0.0605, "step": 1100 }, { "epoch": 1.09, "learning_rate": 9.098834855135324e-06, "loss": 0.0535, "step": 1101 }, { "epoch": 1.09, "learning_rate": 9.08292950272136e-06, "loss": 0.0542, "step": 1102 }, { "epoch": 1.09, "learning_rate": 9.06702648964906e-06, "loss": 0.0571, "step": 1103 }, { "epoch": 1.09, "learning_rate": 9.051125856485183e-06, "loss": 0.0546, "step": 1104 }, { "epoch": 1.09, "learning_rate": 9.035227643790422e-06, "loss": 0.0554, "step": 1105 }, { "epoch": 1.09, "learning_rate": 9.019331892119296e-06, "loss": 0.055, "step": 1106 }, { "epoch": 1.09, "learning_rate": 9.003438642020042e-06, "loss": 0.053, "step": 1107 }, { "epoch": 1.09, "learning_rate": 8.98754793403452e-06, "loss": 0.053, "step": 1108 }, { "epoch": 1.09, "learning_rate": 8.971659808698102e-06, "loss": 0.0608, "step": 1109 }, { "epoch": 1.09, "learning_rate": 8.955774306539574e-06, "loss": 0.0447, "step": 1110 }, { "epoch": 1.1, "learning_rate": 8.939891468081033e-06, "loss": 0.05, "step": 1111 }, { "epoch": 1.1, "learning_rate": 8.924011333837771e-06, "loss": 0.0545, "step": 1112 }, { "epoch": 1.1, "learning_rate": 8.908133944318199e-06, "loss": 0.0546, "step": 1113 }, { "epoch": 1.1, "learning_rate": 8.892259340023708e-06, "loss": 0.0536, "step": 1114 }, { "epoch": 1.1, "learning_rate": 8.8763875614486e-06, "loss": 0.0505, "step": 1115 }, { "epoch": 1.1, "learning_rate": 8.860518649079953e-06, "loss": 0.0574, "step": 1116 }, { "epoch": 1.1, "learning_rate": 8.844652643397554e-06, "loss": 0.053, "step": 1117 }, { "epoch": 1.1, "learning_rate": 8.828789584873754e-06, "loss": 0.0585, "step": 1118 }, { "epoch": 1.1, "learning_rate": 8.812929513973403e-06, "loss": 0.0598, "step": 1119 }, { "epoch": 1.1, "learning_rate": 8.797072471153721e-06, "loss": 0.0564, "step": 1120 }, { "epoch": 1.11, "learning_rate": 8.781218496864198e-06, "loss": 0.0478, "step": 1121 }, { "epoch": 1.11, "learning_rate": 8.765367631546516e-06, "loss": 0.0469, "step": 1122 }, { "epoch": 1.11, "learning_rate": 8.749519915634405e-06, "loss": 0.0515, "step": 1123 }, { "epoch": 1.11, "learning_rate": 8.733675389553579e-06, "loss": 0.0493, "step": 1124 }, { "epoch": 1.11, "learning_rate": 8.717834093721598e-06, "loss": 0.0507, "step": 1125 }, { "epoch": 1.11, "learning_rate": 8.7019960685478e-06, "loss": 0.0574, "step": 1126 }, { "epoch": 1.11, "learning_rate": 8.686161354433163e-06, "loss": 0.0608, "step": 1127 }, { "epoch": 1.11, "learning_rate": 8.670329991770234e-06, "loss": 0.0562, "step": 1128 }, { "epoch": 1.11, "learning_rate": 8.654502020942998e-06, "loss": 0.0521, "step": 1129 }, { "epoch": 1.11, "learning_rate": 8.638677482326792e-06, "loss": 0.0525, "step": 1130 }, { "epoch": 1.12, "learning_rate": 8.622856416288206e-06, "loss": 0.0532, "step": 1131 }, { "epoch": 1.12, "learning_rate": 8.607038863184957e-06, "loss": 0.0523, "step": 1132 }, { "epoch": 1.12, "learning_rate": 8.591224863365817e-06, "loss": 0.06, "step": 1133 }, { "epoch": 1.12, "learning_rate": 8.575414457170476e-06, "loss": 0.0503, "step": 1134 }, { "epoch": 1.12, "learning_rate": 8.55960768492948e-06, "loss": 0.0526, "step": 1135 }, { "epoch": 1.12, "learning_rate": 8.543804586964078e-06, "loss": 0.0537, "step": 1136 }, { "epoch": 1.12, "learning_rate": 8.528005203586172e-06, "loss": 0.0508, "step": 1137 }, { "epoch": 1.12, "learning_rate": 8.512209575098174e-06, "loss": 0.0533, "step": 1138 }, { "epoch": 1.12, "learning_rate": 8.496417741792912e-06, "loss": 0.0543, "step": 1139 }, { "epoch": 1.12, "learning_rate": 8.480629743953556e-06, "loss": 0.0602, "step": 1140 }, { "epoch": 1.13, "learning_rate": 8.464845621853465e-06, "loss": 0.053, "step": 1141 }, { "epoch": 1.13, "learning_rate": 8.449065415756133e-06, "loss": 0.0547, "step": 1142 }, { "epoch": 1.13, "learning_rate": 8.43328916591505e-06, "loss": 0.0524, "step": 1143 }, { "epoch": 1.13, "learning_rate": 8.417516912573625e-06, "loss": 0.0588, "step": 1144 }, { "epoch": 1.13, "learning_rate": 8.401748695965062e-06, "loss": 0.0608, "step": 1145 }, { "epoch": 1.13, "learning_rate": 8.385984556312282e-06, "loss": 0.0493, "step": 1146 }, { "epoch": 1.13, "learning_rate": 8.370224533827787e-06, "loss": 0.0677, "step": 1147 }, { "epoch": 1.13, "learning_rate": 8.354468668713593e-06, "loss": 0.0618, "step": 1148 }, { "epoch": 1.13, "learning_rate": 8.338717001161101e-06, "loss": 0.0538, "step": 1149 }, { "epoch": 1.13, "learning_rate": 8.322969571351013e-06, "loss": 0.0652, "step": 1150 }, { "epoch": 1.14, "learning_rate": 8.307226419453214e-06, "loss": 0.0517, "step": 1151 }, { "epoch": 1.14, "learning_rate": 8.291487585626677e-06, "loss": 0.0598, "step": 1152 }, { "epoch": 1.14, "learning_rate": 8.275753110019367e-06, "loss": 0.0557, "step": 1153 }, { "epoch": 1.14, "learning_rate": 8.260023032768122e-06, "loss": 0.0576, "step": 1154 }, { "epoch": 1.14, "learning_rate": 8.244297393998566e-06, "loss": 0.0522, "step": 1155 }, { "epoch": 1.14, "learning_rate": 8.228576233825006e-06, "loss": 0.0587, "step": 1156 }, { "epoch": 1.14, "learning_rate": 8.212859592350308e-06, "loss": 0.056, "step": 1157 }, { "epoch": 1.14, "learning_rate": 8.19714750966583e-06, "loss": 0.0545, "step": 1158 }, { "epoch": 1.14, "learning_rate": 8.181440025851288e-06, "loss": 0.0543, "step": 1159 }, { "epoch": 1.14, "learning_rate": 8.165737180974678e-06, "loss": 0.0513, "step": 1160 }, { "epoch": 1.14, "learning_rate": 8.150039015092147e-06, "loss": 0.0522, "step": 1161 }, { "epoch": 1.15, "learning_rate": 8.134345568247927e-06, "loss": 0.0587, "step": 1162 }, { "epoch": 1.15, "learning_rate": 8.11865688047419e-06, "loss": 0.0605, "step": 1163 }, { "epoch": 1.15, "learning_rate": 8.10297299179099e-06, "loss": 0.0586, "step": 1164 }, { "epoch": 1.15, "learning_rate": 8.08729394220612e-06, "loss": 0.0593, "step": 1165 }, { "epoch": 1.15, "learning_rate": 8.071619771715038e-06, "loss": 0.0541, "step": 1166 }, { "epoch": 1.15, "learning_rate": 8.05595052030076e-06, "loss": 0.049, "step": 1167 }, { "epoch": 1.15, "learning_rate": 8.040286227933744e-06, "loss": 0.0537, "step": 1168 }, { "epoch": 1.15, "learning_rate": 8.02462693457181e-06, "loss": 0.0541, "step": 1169 }, { "epoch": 1.15, "learning_rate": 8.00897268016001e-06, "loss": 0.051, "step": 1170 }, { "epoch": 1.15, "learning_rate": 7.993323504630564e-06, "loss": 0.0491, "step": 1171 }, { "epoch": 1.16, "learning_rate": 7.97767944790271e-06, "loss": 0.0586, "step": 1172 }, { "epoch": 1.16, "learning_rate": 7.962040549882656e-06, "loss": 0.0521, "step": 1173 }, { "epoch": 1.16, "learning_rate": 7.94640685046343e-06, "loss": 0.0449, "step": 1174 }, { "epoch": 1.16, "learning_rate": 7.930778389524803e-06, "loss": 0.0526, "step": 1175 }, { "epoch": 1.16, "learning_rate": 7.915155206933194e-06, "loss": 0.0536, "step": 1176 }, { "epoch": 1.16, "learning_rate": 7.899537342541543e-06, "loss": 0.0518, "step": 1177 }, { "epoch": 1.16, "learning_rate": 7.883924836189238e-06, "loss": 0.0645, "step": 1178 }, { "epoch": 1.16, "learning_rate": 7.868317727701985e-06, "loss": 0.0488, "step": 1179 }, { "epoch": 1.16, "learning_rate": 7.852716056891736e-06, "loss": 0.0556, "step": 1180 }, { "epoch": 1.16, "learning_rate": 7.837119863556554e-06, "loss": 0.0567, "step": 1181 }, { "epoch": 1.17, "learning_rate": 7.82152918748055e-06, "loss": 0.0536, "step": 1182 }, { "epoch": 1.17, "learning_rate": 7.805944068433745e-06, "loss": 0.0478, "step": 1183 }, { "epoch": 1.17, "learning_rate": 7.790364546171988e-06, "loss": 0.0624, "step": 1184 }, { "epoch": 1.17, "learning_rate": 7.774790660436857e-06, "loss": 0.0564, "step": 1185 }, { "epoch": 1.17, "learning_rate": 7.759222450955547e-06, "loss": 0.0594, "step": 1186 }, { "epoch": 1.17, "learning_rate": 7.743659957440778e-06, "loss": 0.0575, "step": 1187 }, { "epoch": 1.17, "learning_rate": 7.72810321959068e-06, "loss": 0.0606, "step": 1188 }, { "epoch": 1.17, "learning_rate": 7.712552277088716e-06, "loss": 0.0571, "step": 1189 }, { "epoch": 1.17, "learning_rate": 7.697007169603547e-06, "loss": 0.0495, "step": 1190 }, { "epoch": 1.17, "learning_rate": 7.681467936788967e-06, "loss": 0.0605, "step": 1191 }, { "epoch": 1.18, "learning_rate": 7.665934618283774e-06, "loss": 0.0551, "step": 1192 }, { "epoch": 1.18, "learning_rate": 7.650407253711678e-06, "loss": 0.0536, "step": 1193 }, { "epoch": 1.18, "learning_rate": 7.634885882681214e-06, "loss": 0.0544, "step": 1194 }, { "epoch": 1.18, "learning_rate": 7.619370544785608e-06, "loss": 0.0536, "step": 1195 }, { "epoch": 1.18, "learning_rate": 7.60386127960272e-06, "loss": 0.0522, "step": 1196 }, { "epoch": 1.18, "learning_rate": 7.588358126694896e-06, "loss": 0.0588, "step": 1197 }, { "epoch": 1.18, "learning_rate": 7.57286112560891e-06, "loss": 0.0526, "step": 1198 }, { "epoch": 1.18, "learning_rate": 7.55737031587583e-06, "loss": 0.058, "step": 1199 }, { "epoch": 1.18, "learning_rate": 7.541885737010931e-06, "loss": 0.0552, "step": 1200 }, { "epoch": 1.18, "learning_rate": 7.526407428513607e-06, "loss": 0.0522, "step": 1201 }, { "epoch": 1.19, "learning_rate": 7.510935429867237e-06, "loss": 0.0511, "step": 1202 }, { "epoch": 1.19, "learning_rate": 7.495469780539128e-06, "loss": 0.0635, "step": 1203 }, { "epoch": 1.19, "learning_rate": 7.4800105199803645e-06, "loss": 0.0592, "step": 1204 }, { "epoch": 1.19, "learning_rate": 7.4645576876257574e-06, "loss": 0.0517, "step": 1205 }, { "epoch": 1.19, "learning_rate": 7.449111322893702e-06, "loss": 0.0535, "step": 1206 }, { "epoch": 1.19, "learning_rate": 7.4336714651861116e-06, "loss": 0.0554, "step": 1207 }, { "epoch": 1.19, "learning_rate": 7.418238153888286e-06, "loss": 0.0529, "step": 1208 }, { "epoch": 1.19, "learning_rate": 7.402811428368832e-06, "loss": 0.0534, "step": 1209 }, { "epoch": 1.19, "learning_rate": 7.3873913279795624e-06, "loss": 0.0548, "step": 1210 }, { "epoch": 1.19, "learning_rate": 7.37197789205538e-06, "loss": 0.0549, "step": 1211 }, { "epoch": 1.2, "learning_rate": 7.356571159914195e-06, "loss": 0.0511, "step": 1212 }, { "epoch": 1.2, "learning_rate": 7.34117117085681e-06, "loss": 0.0594, "step": 1213 }, { "epoch": 1.2, "learning_rate": 7.325777964166838e-06, "loss": 0.0509, "step": 1214 }, { "epoch": 1.2, "learning_rate": 7.3103915791105755e-06, "loss": 0.0571, "step": 1215 }, { "epoch": 1.2, "learning_rate": 7.295012054936934e-06, "loss": 0.0531, "step": 1216 }, { "epoch": 1.2, "learning_rate": 7.2796394308773125e-06, "loss": 0.0636, "step": 1217 }, { "epoch": 1.2, "learning_rate": 7.264273746145508e-06, "loss": 0.0534, "step": 1218 }, { "epoch": 1.2, "learning_rate": 7.2489150399376276e-06, "loss": 0.0598, "step": 1219 }, { "epoch": 1.2, "learning_rate": 7.23356335143196e-06, "loss": 0.0584, "step": 1220 }, { "epoch": 1.2, "learning_rate": 7.218218719788914e-06, "loss": 0.0568, "step": 1221 }, { "epoch": 1.21, "learning_rate": 7.2028811841508714e-06, "loss": 0.0538, "step": 1222 }, { "epoch": 1.21, "learning_rate": 7.187550783642141e-06, "loss": 0.0532, "step": 1223 }, { "epoch": 1.21, "learning_rate": 7.172227557368803e-06, "loss": 0.054, "step": 1224 }, { "epoch": 1.21, "learning_rate": 7.156911544418659e-06, "loss": 0.0586, "step": 1225 }, { "epoch": 1.21, "learning_rate": 7.141602783861099e-06, "loss": 0.055, "step": 1226 }, { "epoch": 1.21, "learning_rate": 7.126301314747013e-06, "loss": 0.0518, "step": 1227 }, { "epoch": 1.21, "learning_rate": 7.111007176108697e-06, "loss": 0.059, "step": 1228 }, { "epoch": 1.21, "learning_rate": 7.095720406959743e-06, "loss": 0.0596, "step": 1229 }, { "epoch": 1.21, "learning_rate": 7.080441046294948e-06, "loss": 0.0504, "step": 1230 }, { "epoch": 1.21, "learning_rate": 7.065169133090206e-06, "loss": 0.0526, "step": 1231 }, { "epoch": 1.21, "learning_rate": 7.049904706302417e-06, "loss": 0.049, "step": 1232 }, { "epoch": 1.22, "learning_rate": 7.034647804869383e-06, "loss": 0.0583, "step": 1233 }, { "epoch": 1.22, "learning_rate": 7.019398467709711e-06, "loss": 0.0556, "step": 1234 }, { "epoch": 1.22, "learning_rate": 7.0041567337227125e-06, "loss": 0.0514, "step": 1235 }, { "epoch": 1.22, "learning_rate": 6.988922641788296e-06, "loss": 0.0516, "step": 1236 }, { "epoch": 1.22, "learning_rate": 6.973696230766891e-06, "loss": 0.0575, "step": 1237 }, { "epoch": 1.22, "learning_rate": 6.9584775394993195e-06, "loss": 0.0498, "step": 1238 }, { "epoch": 1.22, "learning_rate": 6.943266606806725e-06, "loss": 0.0525, "step": 1239 }, { "epoch": 1.22, "learning_rate": 6.9280634714904405e-06, "loss": 0.059, "step": 1240 }, { "epoch": 1.22, "learning_rate": 6.912868172331935e-06, "loss": 0.0604, "step": 1241 }, { "epoch": 1.22, "learning_rate": 6.897680748092661e-06, "loss": 0.057, "step": 1242 }, { "epoch": 1.23, "learning_rate": 6.882501237514006e-06, "loss": 0.0538, "step": 1243 }, { "epoch": 1.23, "learning_rate": 6.8673296793171555e-06, "loss": 0.0554, "step": 1244 }, { "epoch": 1.23, "learning_rate": 6.852166112203011e-06, "loss": 0.0594, "step": 1245 }, { "epoch": 1.23, "learning_rate": 6.837010574852102e-06, "loss": 0.0594, "step": 1246 }, { "epoch": 1.23, "learning_rate": 6.821863105924457e-06, "loss": 0.0588, "step": 1247 }, { "epoch": 1.23, "learning_rate": 6.80672374405954e-06, "loss": 0.0536, "step": 1248 }, { "epoch": 1.23, "learning_rate": 6.79159252787612e-06, "loss": 0.0572, "step": 1249 }, { "epoch": 1.23, "learning_rate": 6.776469495972201e-06, "loss": 0.0546, "step": 1250 }, { "epoch": 1.23, "learning_rate": 6.761354686924895e-06, "loss": 0.0559, "step": 1251 }, { "epoch": 1.23, "learning_rate": 6.746248139290355e-06, "loss": 0.0527, "step": 1252 }, { "epoch": 1.24, "learning_rate": 6.731149891603648e-06, "loss": 0.0603, "step": 1253 }, { "epoch": 1.24, "learning_rate": 6.716059982378667e-06, "loss": 0.0506, "step": 1254 }, { "epoch": 1.24, "learning_rate": 6.700978450108047e-06, "loss": 0.059, "step": 1255 }, { "epoch": 1.24, "learning_rate": 6.6859053332630406e-06, "loss": 0.0557, "step": 1256 }, { "epoch": 1.24, "learning_rate": 6.67084067029345e-06, "loss": 0.0583, "step": 1257 }, { "epoch": 1.24, "learning_rate": 6.655784499627491e-06, "loss": 0.0504, "step": 1258 }, { "epoch": 1.24, "learning_rate": 6.640736859671739e-06, "loss": 0.0596, "step": 1259 }, { "epoch": 1.24, "learning_rate": 6.625697788810987e-06, "loss": 0.0535, "step": 1260 }, { "epoch": 1.24, "learning_rate": 6.610667325408191e-06, "loss": 0.045, "step": 1261 }, { "epoch": 1.24, "learning_rate": 6.595645507804333e-06, "loss": 0.0509, "step": 1262 }, { "epoch": 1.25, "learning_rate": 6.5806323743183455e-06, "loss": 0.055, "step": 1263 }, { "epoch": 1.25, "learning_rate": 6.565627963247015e-06, "loss": 0.0494, "step": 1264 }, { "epoch": 1.25, "learning_rate": 6.550632312864869e-06, "loss": 0.0547, "step": 1265 }, { "epoch": 1.25, "learning_rate": 6.535645461424096e-06, "loss": 0.055, "step": 1266 }, { "epoch": 1.25, "learning_rate": 6.520667447154429e-06, "loss": 0.0599, "step": 1267 }, { "epoch": 1.25, "learning_rate": 6.505698308263072e-06, "loss": 0.0536, "step": 1268 }, { "epoch": 1.25, "learning_rate": 6.490738082934576e-06, "loss": 0.0515, "step": 1269 }, { "epoch": 1.25, "learning_rate": 6.4757868093307645e-06, "loss": 0.0552, "step": 1270 }, { "epoch": 1.25, "learning_rate": 6.460844525590617e-06, "loss": 0.052, "step": 1271 }, { "epoch": 1.25, "learning_rate": 6.445911269830189e-06, "loss": 0.0533, "step": 1272 }, { "epoch": 1.26, "learning_rate": 6.430987080142502e-06, "loss": 0.0613, "step": 1273 }, { "epoch": 1.26, "learning_rate": 6.416071994597452e-06, "loss": 0.0534, "step": 1274 }, { "epoch": 1.26, "learning_rate": 6.401166051241714e-06, "loss": 0.0539, "step": 1275 }, { "epoch": 1.26, "learning_rate": 6.386269288098637e-06, "loss": 0.0598, "step": 1276 }, { "epoch": 1.26, "learning_rate": 6.371381743168156e-06, "loss": 0.0562, "step": 1277 }, { "epoch": 1.26, "learning_rate": 6.356503454426692e-06, "loss": 0.0517, "step": 1278 }, { "epoch": 1.26, "learning_rate": 6.341634459827053e-06, "loss": 0.0526, "step": 1279 }, { "epoch": 1.26, "learning_rate": 6.32677479729834e-06, "loss": 0.0465, "step": 1280 }, { "epoch": 1.26, "learning_rate": 6.311924504745842e-06, "loss": 0.063, "step": 1281 }, { "epoch": 1.26, "learning_rate": 6.2970836200509614e-06, "loss": 0.0579, "step": 1282 }, { "epoch": 1.27, "learning_rate": 6.282252181071082e-06, "loss": 0.0515, "step": 1283 }, { "epoch": 1.27, "learning_rate": 6.267430225639515e-06, "loss": 0.0527, "step": 1284 }, { "epoch": 1.27, "learning_rate": 6.25261779156536e-06, "loss": 0.0558, "step": 1285 }, { "epoch": 1.27, "learning_rate": 6.237814916633444e-06, "loss": 0.0533, "step": 1286 }, { "epoch": 1.27, "learning_rate": 6.2230216386041965e-06, "loss": 0.0547, "step": 1287 }, { "epoch": 1.27, "learning_rate": 6.208237995213582e-06, "loss": 0.0527, "step": 1288 }, { "epoch": 1.27, "learning_rate": 6.193464024172973e-06, "loss": 0.0529, "step": 1289 }, { "epoch": 1.27, "learning_rate": 6.178699763169073e-06, "loss": 0.0536, "step": 1290 }, { "epoch": 1.27, "learning_rate": 6.163945249863825e-06, "loss": 0.0523, "step": 1291 }, { "epoch": 1.27, "learning_rate": 6.149200521894294e-06, "loss": 0.0581, "step": 1292 }, { "epoch": 1.28, "learning_rate": 6.134465616872598e-06, "loss": 0.0488, "step": 1293 }, { "epoch": 1.28, "learning_rate": 6.11974057238578e-06, "loss": 0.0609, "step": 1294 }, { "epoch": 1.28, "learning_rate": 6.10502542599575e-06, "loss": 0.0468, "step": 1295 }, { "epoch": 1.28, "learning_rate": 6.090320215239151e-06, "loss": 0.0535, "step": 1296 }, { "epoch": 1.28, "learning_rate": 6.07562497762729e-06, "loss": 0.0642, "step": 1297 }, { "epoch": 1.28, "learning_rate": 6.060939750646039e-06, "loss": 0.0468, "step": 1298 }, { "epoch": 1.28, "learning_rate": 6.046264571755718e-06, "loss": 0.0469, "step": 1299 }, { "epoch": 1.28, "learning_rate": 6.0315994783910345e-06, "loss": 0.0566, "step": 1300 }, { "epoch": 1.28, "learning_rate": 6.0169445079609535e-06, "loss": 0.0595, "step": 1301 }, { "epoch": 1.28, "learning_rate": 6.002299697848629e-06, "loss": 0.0558, "step": 1302 }, { "epoch": 1.29, "learning_rate": 5.987665085411285e-06, "loss": 0.047, "step": 1303 }, { "epoch": 1.29, "learning_rate": 5.973040707980149e-06, "loss": 0.0656, "step": 1304 }, { "epoch": 1.29, "learning_rate": 5.958426602860322e-06, "loss": 0.0601, "step": 1305 }, { "epoch": 1.29, "learning_rate": 5.943822807330715e-06, "loss": 0.0527, "step": 1306 }, { "epoch": 1.29, "learning_rate": 5.929229358643932e-06, "loss": 0.053, "step": 1307 }, { "epoch": 1.29, "learning_rate": 5.914646294026189e-06, "loss": 0.0583, "step": 1308 }, { "epoch": 1.29, "learning_rate": 5.900073650677212e-06, "loss": 0.0588, "step": 1309 }, { "epoch": 1.29, "learning_rate": 5.885511465770142e-06, "loss": 0.0531, "step": 1310 }, { "epoch": 1.29, "learning_rate": 5.8709597764514414e-06, "loss": 0.0531, "step": 1311 }, { "epoch": 1.29, "learning_rate": 5.856418619840805e-06, "loss": 0.059, "step": 1312 }, { "epoch": 1.29, "learning_rate": 5.841888033031049e-06, "loss": 0.0522, "step": 1313 }, { "epoch": 1.3, "learning_rate": 5.827368053088043e-06, "loss": 0.062, "step": 1314 }, { "epoch": 1.3, "learning_rate": 5.8128587170505805e-06, "loss": 0.0582, "step": 1315 }, { "epoch": 1.3, "learning_rate": 5.79836006193032e-06, "loss": 0.0507, "step": 1316 }, { "epoch": 1.3, "learning_rate": 5.7838721247116634e-06, "loss": 0.0527, "step": 1317 }, { "epoch": 1.3, "learning_rate": 5.769394942351684e-06, "loss": 0.0537, "step": 1318 }, { "epoch": 1.3, "learning_rate": 5.7549285517800035e-06, "loss": 0.0567, "step": 1319 }, { "epoch": 1.3, "learning_rate": 5.740472989898738e-06, "loss": 0.0508, "step": 1320 }, { "epoch": 1.3, "learning_rate": 5.726028293582355e-06, "loss": 0.0519, "step": 1321 }, { "epoch": 1.3, "learning_rate": 5.711594499677628e-06, "loss": 0.0562, "step": 1322 }, { "epoch": 1.3, "learning_rate": 5.697171645003507e-06, "loss": 0.0603, "step": 1323 }, { "epoch": 1.31, "learning_rate": 5.682759766351033e-06, "loss": 0.0541, "step": 1324 }, { "epoch": 1.31, "learning_rate": 5.668358900483268e-06, "loss": 0.0521, "step": 1325 }, { "epoch": 1.31, "learning_rate": 5.65396908413516e-06, "loss": 0.0551, "step": 1326 }, { "epoch": 1.31, "learning_rate": 5.639590354013485e-06, "loss": 0.0571, "step": 1327 }, { "epoch": 1.31, "learning_rate": 5.62522274679673e-06, "loss": 0.0511, "step": 1328 }, { "epoch": 1.31, "learning_rate": 5.610866299135019e-06, "loss": 0.0479, "step": 1329 }, { "epoch": 1.31, "learning_rate": 5.5965210476499965e-06, "loss": 0.0566, "step": 1330 }, { "epoch": 1.31, "learning_rate": 5.582187028934762e-06, "loss": 0.0536, "step": 1331 }, { "epoch": 1.31, "learning_rate": 5.5678642795537455e-06, "loss": 0.0569, "step": 1332 }, { "epoch": 1.31, "learning_rate": 5.553552836042645e-06, "loss": 0.0504, "step": 1333 }, { "epoch": 1.32, "learning_rate": 5.539252734908305e-06, "loss": 0.0481, "step": 1334 }, { "epoch": 1.32, "learning_rate": 5.524964012628648e-06, "loss": 0.0487, "step": 1335 }, { "epoch": 1.32, "learning_rate": 5.510686705652563e-06, "loss": 0.0526, "step": 1336 }, { "epoch": 1.32, "learning_rate": 5.496420850399826e-06, "loss": 0.0518, "step": 1337 }, { "epoch": 1.32, "learning_rate": 5.482166483260993e-06, "loss": 0.0492, "step": 1338 }, { "epoch": 1.32, "learning_rate": 5.467923640597326e-06, "loss": 0.0534, "step": 1339 }, { "epoch": 1.32, "learning_rate": 5.4536923587406765e-06, "loss": 0.0544, "step": 1340 }, { "epoch": 1.32, "learning_rate": 5.439472673993418e-06, "loss": 0.0505, "step": 1341 }, { "epoch": 1.32, "learning_rate": 5.42526462262833e-06, "loss": 0.0602, "step": 1342 }, { "epoch": 1.32, "learning_rate": 5.411068240888529e-06, "loss": 0.0534, "step": 1343 }, { "epoch": 1.33, "learning_rate": 5.396883564987347e-06, "loss": 0.0511, "step": 1344 }, { "epoch": 1.33, "learning_rate": 5.3827106311082765e-06, "loss": 0.0442, "step": 1345 }, { "epoch": 1.33, "learning_rate": 5.368549475404836e-06, "loss": 0.0483, "step": 1346 }, { "epoch": 1.33, "learning_rate": 5.354400134000519e-06, "loss": 0.0521, "step": 1347 }, { "epoch": 1.33, "learning_rate": 5.340262642988665e-06, "loss": 0.0534, "step": 1348 }, { "epoch": 1.33, "learning_rate": 5.326137038432399e-06, "loss": 0.0534, "step": 1349 }, { "epoch": 1.33, "learning_rate": 5.312023356364513e-06, "loss": 0.0494, "step": 1350 }, { "epoch": 1.33, "learning_rate": 5.29792163278739e-06, "loss": 0.0559, "step": 1351 }, { "epoch": 1.33, "learning_rate": 5.283831903672918e-06, "loss": 0.0519, "step": 1352 }, { "epoch": 1.33, "learning_rate": 5.269754204962369e-06, "loss": 0.0583, "step": 1353 }, { "epoch": 1.34, "learning_rate": 5.2556885725663465e-06, "loss": 0.0421, "step": 1354 }, { "epoch": 1.34, "learning_rate": 5.241635042364659e-06, "loss": 0.0502, "step": 1355 }, { "epoch": 1.34, "learning_rate": 5.227593650206258e-06, "loss": 0.0482, "step": 1356 }, { "epoch": 1.34, "learning_rate": 5.213564431909116e-06, "loss": 0.0483, "step": 1357 }, { "epoch": 1.34, "learning_rate": 5.199547423260168e-06, "loss": 0.0556, "step": 1358 }, { "epoch": 1.34, "learning_rate": 5.185542660015192e-06, "loss": 0.0533, "step": 1359 }, { "epoch": 1.34, "learning_rate": 5.171550177898729e-06, "loss": 0.052, "step": 1360 }, { "epoch": 1.34, "learning_rate": 5.157570012604003e-06, "loss": 0.0514, "step": 1361 }, { "epoch": 1.34, "learning_rate": 5.143602199792805e-06, "loss": 0.0554, "step": 1362 }, { "epoch": 1.34, "learning_rate": 5.129646775095432e-06, "loss": 0.052, "step": 1363 }, { "epoch": 1.35, "learning_rate": 5.115703774110562e-06, "loss": 0.0452, "step": 1364 }, { "epoch": 1.35, "learning_rate": 5.101773232405201e-06, "loss": 0.0516, "step": 1365 }, { "epoch": 1.35, "learning_rate": 5.087855185514556e-06, "loss": 0.0586, "step": 1366 }, { "epoch": 1.35, "learning_rate": 5.073949668941975e-06, "loss": 0.0595, "step": 1367 }, { "epoch": 1.35, "learning_rate": 5.060056718158832e-06, "loss": 0.0533, "step": 1368 }, { "epoch": 1.35, "learning_rate": 5.0461763686044485e-06, "loss": 0.0531, "step": 1369 }, { "epoch": 1.35, "learning_rate": 5.032308655686011e-06, "loss": 0.0529, "step": 1370 }, { "epoch": 1.35, "learning_rate": 5.018453614778457e-06, "loss": 0.0534, "step": 1371 }, { "epoch": 1.35, "learning_rate": 5.004611281224414e-06, "loss": 0.0497, "step": 1372 }, { "epoch": 1.35, "learning_rate": 4.990781690334081e-06, "loss": 0.0575, "step": 1373 }, { "epoch": 1.36, "learning_rate": 4.976964877385166e-06, "loss": 0.0551, "step": 1374 }, { "epoch": 1.36, "learning_rate": 4.9631608776227645e-06, "loss": 0.0505, "step": 1375 }, { "epoch": 1.36, "learning_rate": 4.9493697262593065e-06, "loss": 0.0548, "step": 1376 }, { "epoch": 1.36, "learning_rate": 4.935591458474433e-06, "loss": 0.0593, "step": 1377 }, { "epoch": 1.36, "learning_rate": 4.921826109414921e-06, "loss": 0.0513, "step": 1378 }, { "epoch": 1.36, "learning_rate": 4.908073714194605e-06, "loss": 0.0537, "step": 1379 }, { "epoch": 1.36, "learning_rate": 4.894334307894259e-06, "loss": 0.0647, "step": 1380 }, { "epoch": 1.36, "learning_rate": 4.880607925561543e-06, "loss": 0.0549, "step": 1381 }, { "epoch": 1.36, "learning_rate": 4.866894602210875e-06, "loss": 0.0525, "step": 1382 }, { "epoch": 1.36, "learning_rate": 4.853194372823379e-06, "loss": 0.0545, "step": 1383 }, { "epoch": 1.36, "learning_rate": 4.8395072723467585e-06, "loss": 0.0578, "step": 1384 }, { "epoch": 1.37, "learning_rate": 4.825833335695247e-06, "loss": 0.0571, "step": 1385 }, { "epoch": 1.37, "learning_rate": 4.812172597749483e-06, "loss": 0.0519, "step": 1386 }, { "epoch": 1.37, "learning_rate": 4.798525093356441e-06, "loss": 0.0535, "step": 1387 }, { "epoch": 1.37, "learning_rate": 4.784890857329345e-06, "loss": 0.0557, "step": 1388 }, { "epoch": 1.37, "learning_rate": 4.771269924447561e-06, "loss": 0.0511, "step": 1389 }, { "epoch": 1.37, "learning_rate": 4.757662329456534e-06, "loss": 0.0583, "step": 1390 }, { "epoch": 1.37, "learning_rate": 4.74406810706767e-06, "loss": 0.0516, "step": 1391 }, { "epoch": 1.37, "learning_rate": 4.730487291958279e-06, "loss": 0.0518, "step": 1392 }, { "epoch": 1.37, "learning_rate": 4.71691991877146e-06, "loss": 0.0514, "step": 1393 }, { "epoch": 1.37, "learning_rate": 4.703366022116022e-06, "loss": 0.0493, "step": 1394 }, { "epoch": 1.38, "learning_rate": 4.6898256365664065e-06, "loss": 0.0546, "step": 1395 }, { "epoch": 1.38, "learning_rate": 4.676298796662579e-06, "loss": 0.0481, "step": 1396 }, { "epoch": 1.38, "learning_rate": 4.662785536909963e-06, "loss": 0.0543, "step": 1397 }, { "epoch": 1.38, "learning_rate": 4.649285891779327e-06, "loss": 0.0519, "step": 1398 }, { "epoch": 1.38, "learning_rate": 4.635799895706725e-06, "loss": 0.0555, "step": 1399 }, { "epoch": 1.38, "learning_rate": 4.622327583093379e-06, "loss": 0.0503, "step": 1400 }, { "epoch": 1.38, "learning_rate": 4.60886898830562e-06, "loss": 0.0509, "step": 1401 }, { "epoch": 1.38, "learning_rate": 4.5954241456747785e-06, "loss": 0.0549, "step": 1402 }, { "epoch": 1.38, "learning_rate": 4.5819930894971e-06, "loss": 0.0487, "step": 1403 }, { "epoch": 1.38, "learning_rate": 4.568575854033675e-06, "loss": 0.0482, "step": 1404 }, { "epoch": 1.39, "learning_rate": 4.5551724735103285e-06, "loss": 0.0532, "step": 1405 }, { "epoch": 1.39, "learning_rate": 4.541782982117552e-06, "loss": 0.05, "step": 1406 }, { "epoch": 1.39, "learning_rate": 4.528407414010395e-06, "loss": 0.0516, "step": 1407 }, { "epoch": 1.39, "learning_rate": 4.515045803308407e-06, "loss": 0.0471, "step": 1408 }, { "epoch": 1.39, "learning_rate": 4.501698184095513e-06, "loss": 0.0512, "step": 1409 }, { "epoch": 1.39, "learning_rate": 4.488364590419971e-06, "loss": 0.0513, "step": 1410 }, { "epoch": 1.39, "learning_rate": 4.475045056294239e-06, "loss": 0.0615, "step": 1411 }, { "epoch": 1.39, "learning_rate": 4.461739615694929e-06, "loss": 0.0616, "step": 1412 }, { "epoch": 1.39, "learning_rate": 4.448448302562687e-06, "loss": 0.0498, "step": 1413 }, { "epoch": 1.39, "learning_rate": 4.435171150802135e-06, "loss": 0.0583, "step": 1414 }, { "epoch": 1.4, "learning_rate": 4.4219081942817586e-06, "loss": 0.0577, "step": 1415 }, { "epoch": 1.4, "learning_rate": 4.4086594668338454e-06, "loss": 0.0557, "step": 1416 }, { "epoch": 1.4, "learning_rate": 4.395425002254371e-06, "loss": 0.0568, "step": 1417 }, { "epoch": 1.4, "learning_rate": 4.382204834302945e-06, "loss": 0.0549, "step": 1418 }, { "epoch": 1.4, "learning_rate": 4.368998996702694e-06, "loss": 0.048, "step": 1419 }, { "epoch": 1.4, "learning_rate": 4.355807523140199e-06, "loss": 0.0529, "step": 1420 }, { "epoch": 1.4, "learning_rate": 4.342630447265392e-06, "loss": 0.0499, "step": 1421 }, { "epoch": 1.4, "learning_rate": 4.329467802691489e-06, "loss": 0.0525, "step": 1422 }, { "epoch": 1.4, "learning_rate": 4.316319622994878e-06, "loss": 0.0545, "step": 1423 }, { "epoch": 1.4, "learning_rate": 4.3031859417150655e-06, "loss": 0.0548, "step": 1424 }, { "epoch": 1.41, "learning_rate": 4.290066792354559e-06, "loss": 0.0534, "step": 1425 }, { "epoch": 1.41, "learning_rate": 4.276962208378811e-06, "loss": 0.0492, "step": 1426 }, { "epoch": 1.41, "learning_rate": 4.263872223216106e-06, "loss": 0.0502, "step": 1427 }, { "epoch": 1.41, "learning_rate": 4.2507968702575005e-06, "loss": 0.0495, "step": 1428 }, { "epoch": 1.41, "learning_rate": 4.237736182856719e-06, "loss": 0.0521, "step": 1429 }, { "epoch": 1.41, "learning_rate": 4.22469019433007e-06, "loss": 0.0444, "step": 1430 }, { "epoch": 1.41, "learning_rate": 4.211658937956384e-06, "loss": 0.0543, "step": 1431 }, { "epoch": 1.41, "learning_rate": 4.198642446976895e-06, "loss": 0.0489, "step": 1432 }, { "epoch": 1.41, "learning_rate": 4.185640754595183e-06, "loss": 0.0508, "step": 1433 }, { "epoch": 1.41, "learning_rate": 4.172653893977069e-06, "loss": 0.0571, "step": 1434 }, { "epoch": 1.42, "learning_rate": 4.159681898250552e-06, "loss": 0.0561, "step": 1435 }, { "epoch": 1.42, "learning_rate": 4.146724800505697e-06, "loss": 0.0531, "step": 1436 }, { "epoch": 1.42, "learning_rate": 4.133782633794583e-06, "loss": 0.0533, "step": 1437 }, { "epoch": 1.42, "learning_rate": 4.120855431131189e-06, "loss": 0.0508, "step": 1438 }, { "epoch": 1.42, "learning_rate": 4.107943225491322e-06, "loss": 0.0548, "step": 1439 }, { "epoch": 1.42, "learning_rate": 4.095046049812545e-06, "loss": 0.0503, "step": 1440 }, { "epoch": 1.42, "learning_rate": 4.082163936994068e-06, "loss": 0.0483, "step": 1441 }, { "epoch": 1.42, "learning_rate": 4.069296919896689e-06, "loss": 0.0485, "step": 1442 }, { "epoch": 1.42, "learning_rate": 4.056445031342687e-06, "loss": 0.0492, "step": 1443 }, { "epoch": 1.42, "learning_rate": 4.0436083041157625e-06, "loss": 0.0552, "step": 1444 }, { "epoch": 1.43, "learning_rate": 4.030786770960927e-06, "loss": 0.0475, "step": 1445 }, { "epoch": 1.43, "learning_rate": 4.017980464584449e-06, "loss": 0.0546, "step": 1446 }, { "epoch": 1.43, "learning_rate": 4.005189417653743e-06, "loss": 0.0525, "step": 1447 }, { "epoch": 1.43, "learning_rate": 3.9924136627973e-06, "loss": 0.0501, "step": 1448 }, { "epoch": 1.43, "learning_rate": 3.979653232604612e-06, "loss": 0.0524, "step": 1449 }, { "epoch": 1.43, "learning_rate": 3.966908159626066e-06, "loss": 0.0478, "step": 1450 }, { "epoch": 1.43, "learning_rate": 3.954178476372887e-06, "loss": 0.0519, "step": 1451 }, { "epoch": 1.43, "learning_rate": 3.941464215317033e-06, "loss": 0.0509, "step": 1452 }, { "epoch": 1.43, "learning_rate": 3.928765408891131e-06, "loss": 0.0459, "step": 1453 }, { "epoch": 1.43, "learning_rate": 3.916082089488372e-06, "loss": 0.0474, "step": 1454 }, { "epoch": 1.43, "learning_rate": 3.903414289462458e-06, "loss": 0.0555, "step": 1455 }, { "epoch": 1.44, "learning_rate": 3.89076204112749e-06, "loss": 0.0536, "step": 1456 }, { "epoch": 1.44, "learning_rate": 3.878125376757899e-06, "loss": 0.0545, "step": 1457 }, { "epoch": 1.44, "learning_rate": 3.865504328588373e-06, "loss": 0.0568, "step": 1458 }, { "epoch": 1.44, "learning_rate": 3.852898928813751e-06, "loss": 0.0572, "step": 1459 }, { "epoch": 1.44, "learning_rate": 3.84030920958897e-06, "loss": 0.0476, "step": 1460 }, { "epoch": 1.44, "learning_rate": 3.827735203028953e-06, "loss": 0.052, "step": 1461 }, { "epoch": 1.44, "learning_rate": 3.815176941208555e-06, "loss": 0.0485, "step": 1462 }, { "epoch": 1.44, "learning_rate": 3.8026344561624553e-06, "loss": 0.0513, "step": 1463 }, { "epoch": 1.44, "learning_rate": 3.7901077798851006e-06, "loss": 0.0489, "step": 1464 }, { "epoch": 1.44, "learning_rate": 3.7775969443306024e-06, "loss": 0.0555, "step": 1465 }, { "epoch": 1.45, "learning_rate": 3.7651019814126656e-06, "loss": 0.051, "step": 1466 }, { "epoch": 1.45, "learning_rate": 3.752622923004513e-06, "loss": 0.0527, "step": 1467 }, { "epoch": 1.45, "learning_rate": 3.740159800938784e-06, "loss": 0.0523, "step": 1468 }, { "epoch": 1.45, "learning_rate": 3.7277126470074807e-06, "loss": 0.0568, "step": 1469 }, { "epoch": 1.45, "learning_rate": 3.7152814929618597e-06, "loss": 0.0473, "step": 1470 }, { "epoch": 1.45, "learning_rate": 3.7028663705123757e-06, "loss": 0.059, "step": 1471 }, { "epoch": 1.45, "learning_rate": 3.690467311328574e-06, "loss": 0.0498, "step": 1472 }, { "epoch": 1.45, "learning_rate": 3.6780843470390415e-06, "loss": 0.0507, "step": 1473 }, { "epoch": 1.45, "learning_rate": 3.665717509231295e-06, "loss": 0.0544, "step": 1474 }, { "epoch": 1.45, "learning_rate": 3.6533668294517154e-06, "loss": 0.0416, "step": 1475 }, { "epoch": 1.46, "learning_rate": 3.641032339205478e-06, "loss": 0.0534, "step": 1476 }, { "epoch": 1.46, "learning_rate": 3.6287140699564474e-06, "loss": 0.047, "step": 1477 }, { "epoch": 1.46, "learning_rate": 3.616412053127121e-06, "loss": 0.0524, "step": 1478 }, { "epoch": 1.46, "learning_rate": 3.6041263200985253e-06, "loss": 0.0565, "step": 1479 }, { "epoch": 1.46, "learning_rate": 3.5918569022101646e-06, "loss": 0.0464, "step": 1480 }, { "epoch": 1.46, "learning_rate": 3.579603830759909e-06, "loss": 0.058, "step": 1481 }, { "epoch": 1.46, "learning_rate": 3.5673671370039464e-06, "loss": 0.0499, "step": 1482 }, { "epoch": 1.46, "learning_rate": 3.5551468521566754e-06, "loss": 0.0533, "step": 1483 }, { "epoch": 1.46, "learning_rate": 3.5429430073906366e-06, "loss": 0.0591, "step": 1484 }, { "epoch": 1.46, "learning_rate": 3.530755633836446e-06, "loss": 0.0518, "step": 1485 }, { "epoch": 1.47, "learning_rate": 3.5185847625826897e-06, "loss": 0.0564, "step": 1486 }, { "epoch": 1.47, "learning_rate": 3.5064304246758694e-06, "loss": 0.0502, "step": 1487 }, { "epoch": 1.47, "learning_rate": 3.4942926511203023e-06, "loss": 0.0575, "step": 1488 }, { "epoch": 1.47, "learning_rate": 3.482171472878062e-06, "loss": 0.0497, "step": 1489 }, { "epoch": 1.47, "learning_rate": 3.4700669208688787e-06, "loss": 0.0517, "step": 1490 }, { "epoch": 1.47, "learning_rate": 3.4579790259700816e-06, "loss": 0.0508, "step": 1491 }, { "epoch": 1.47, "learning_rate": 3.4459078190164985e-06, "loss": 0.0532, "step": 1492 }, { "epoch": 1.47, "learning_rate": 3.4338533308004006e-06, "loss": 0.0546, "step": 1493 }, { "epoch": 1.47, "learning_rate": 3.421815592071398e-06, "loss": 0.0513, "step": 1494 }, { "epoch": 1.47, "learning_rate": 3.4097946335363873e-06, "loss": 0.0566, "step": 1495 }, { "epoch": 1.48, "learning_rate": 3.39779048585945e-06, "loss": 0.0566, "step": 1496 }, { "epoch": 1.48, "learning_rate": 3.385803179661796e-06, "loss": 0.0497, "step": 1497 }, { "epoch": 1.48, "learning_rate": 3.3738327455216625e-06, "loss": 0.0525, "step": 1498 }, { "epoch": 1.48, "learning_rate": 3.3618792139742597e-06, "loss": 0.0516, "step": 1499 }, { "epoch": 1.48, "learning_rate": 3.3499426155116697e-06, "loss": 0.0537, "step": 1500 }, { "epoch": 1.48, "learning_rate": 3.3380229805827925e-06, "loss": 0.0526, "step": 1501 }, { "epoch": 1.48, "learning_rate": 3.326120339593244e-06, "loss": 0.0575, "step": 1502 }, { "epoch": 1.48, "learning_rate": 3.314234722905302e-06, "loss": 0.0515, "step": 1503 }, { "epoch": 1.48, "learning_rate": 3.3023661608378054e-06, "loss": 0.0518, "step": 1504 }, { "epoch": 1.48, "learning_rate": 3.2905146836661005e-06, "loss": 0.0523, "step": 1505 }, { "epoch": 1.49, "learning_rate": 3.278680321621941e-06, "loss": 0.0521, "step": 1506 }, { "epoch": 1.49, "learning_rate": 3.2668631048934343e-06, "loss": 0.0514, "step": 1507 }, { "epoch": 1.49, "learning_rate": 3.255063063624939e-06, "loss": 0.0478, "step": 1508 }, { "epoch": 1.49, "learning_rate": 3.243280227917007e-06, "loss": 0.0491, "step": 1509 }, { "epoch": 1.49, "learning_rate": 3.2315146278263053e-06, "loss": 0.0511, "step": 1510 }, { "epoch": 1.49, "learning_rate": 3.219766293365525e-06, "loss": 0.0531, "step": 1511 }, { "epoch": 1.49, "learning_rate": 3.2080352545033254e-06, "loss": 0.0514, "step": 1512 }, { "epoch": 1.49, "learning_rate": 3.1963215411642358e-06, "loss": 0.0501, "step": 1513 }, { "epoch": 1.49, "learning_rate": 3.1846251832286e-06, "loss": 0.057, "step": 1514 }, { "epoch": 1.49, "learning_rate": 3.1729462105324816e-06, "loss": 0.0535, "step": 1515 }, { "epoch": 1.5, "learning_rate": 3.1612846528676065e-06, "loss": 0.049, "step": 1516 }, { "epoch": 1.5, "learning_rate": 3.149640539981267e-06, "loss": 0.0531, "step": 1517 }, { "epoch": 1.5, "learning_rate": 3.1380139015762558e-06, "loss": 0.0465, "step": 1518 }, { "epoch": 1.5, "learning_rate": 3.1264047673108012e-06, "loss": 0.0495, "step": 1519 }, { "epoch": 1.5, "learning_rate": 3.1148131667984682e-06, "loss": 0.0587, "step": 1520 }, { "epoch": 1.5, "learning_rate": 3.103239129608104e-06, "loss": 0.0536, "step": 1521 }, { "epoch": 1.5, "learning_rate": 3.091682685263746e-06, "loss": 0.0479, "step": 1522 }, { "epoch": 1.5, "learning_rate": 3.0801438632445633e-06, "loss": 0.0499, "step": 1523 }, { "epoch": 1.5, "learning_rate": 3.0686226929847617e-06, "loss": 0.0518, "step": 1524 }, { "epoch": 1.5, "learning_rate": 3.057119203873532e-06, "loss": 0.0527, "step": 1525 }, { "epoch": 1.5, "learning_rate": 3.045633425254951e-06, "loss": 0.0548, "step": 1526 }, { "epoch": 1.51, "learning_rate": 3.034165386427922e-06, "loss": 0.058, "step": 1527 }, { "epoch": 1.51, "learning_rate": 3.0227151166461e-06, "loss": 0.0546, "step": 1528 }, { "epoch": 1.51, "learning_rate": 3.011282645117807e-06, "loss": 0.0526, "step": 1529 }, { "epoch": 1.51, "learning_rate": 2.99986800100597e-06, "loss": 0.0464, "step": 1530 }, { "epoch": 1.51, "learning_rate": 2.9884712134280324e-06, "loss": 0.049, "step": 1531 }, { "epoch": 1.51, "learning_rate": 2.977092311455898e-06, "loss": 0.0542, "step": 1532 }, { "epoch": 1.51, "learning_rate": 2.9657313241158347e-06, "loss": 0.0471, "step": 1533 }, { "epoch": 1.51, "learning_rate": 2.954388280388425e-06, "loss": 0.055, "step": 1534 }, { "epoch": 1.51, "learning_rate": 2.9430632092084676e-06, "loss": 0.058, "step": 1535 }, { "epoch": 1.51, "learning_rate": 2.9317561394649197e-06, "loss": 0.0474, "step": 1536 }, { "epoch": 1.52, "learning_rate": 2.9204671000008232e-06, "loss": 0.0562, "step": 1537 }, { "epoch": 1.52, "learning_rate": 2.909196119613218e-06, "loss": 0.0483, "step": 1538 }, { "epoch": 1.52, "learning_rate": 2.8979432270530895e-06, "loss": 0.048, "step": 1539 }, { "epoch": 1.52, "learning_rate": 2.8867084510252684e-06, "loss": 0.0465, "step": 1540 }, { "epoch": 1.52, "learning_rate": 2.8754918201883863e-06, "loss": 0.0545, "step": 1541 }, { "epoch": 1.52, "learning_rate": 2.8642933631547766e-06, "loss": 0.0517, "step": 1542 }, { "epoch": 1.52, "learning_rate": 2.8531131084904253e-06, "loss": 0.0501, "step": 1543 }, { "epoch": 1.52, "learning_rate": 2.8419510847148747e-06, "loss": 0.0559, "step": 1544 }, { "epoch": 1.52, "learning_rate": 2.8308073203011667e-06, "loss": 0.0475, "step": 1545 }, { "epoch": 1.52, "learning_rate": 2.81968184367577e-06, "loss": 0.056, "step": 1546 }, { "epoch": 1.53, "learning_rate": 2.808574683218496e-06, "loss": 0.0517, "step": 1547 }, { "epoch": 1.53, "learning_rate": 2.7974858672624417e-06, "loss": 0.0488, "step": 1548 }, { "epoch": 1.53, "learning_rate": 2.786415424093898e-06, "loss": 0.0491, "step": 1549 }, { "epoch": 1.53, "learning_rate": 2.775363381952304e-06, "loss": 0.0487, "step": 1550 }, { "epoch": 1.53, "learning_rate": 2.764329769030144e-06, "loss": 0.0496, "step": 1551 }, { "epoch": 1.53, "learning_rate": 2.753314613472906e-06, "loss": 0.0574, "step": 1552 }, { "epoch": 1.53, "learning_rate": 2.742317943378985e-06, "loss": 0.0551, "step": 1553 }, { "epoch": 1.53, "learning_rate": 2.731339786799623e-06, "loss": 0.0514, "step": 1554 }, { "epoch": 1.53, "learning_rate": 2.7203801717388423e-06, "loss": 0.0465, "step": 1555 }, { "epoch": 1.53, "learning_rate": 2.7094391261533605e-06, "loss": 0.0477, "step": 1556 }, { "epoch": 1.54, "learning_rate": 2.698516677952535e-06, "loss": 0.0508, "step": 1557 }, { "epoch": 1.54, "learning_rate": 2.6876128549982725e-06, "loss": 0.056, "step": 1558 }, { "epoch": 1.54, "learning_rate": 2.6767276851049818e-06, "loss": 0.0518, "step": 1559 }, { "epoch": 1.54, "learning_rate": 2.665861196039475e-06, "loss": 0.056, "step": 1560 }, { "epoch": 1.54, "learning_rate": 2.6550134155209274e-06, "loss": 0.0508, "step": 1561 }, { "epoch": 1.54, "learning_rate": 2.644184371220778e-06, "loss": 0.0494, "step": 1562 }, { "epoch": 1.54, "learning_rate": 2.6333740907626747e-06, "loss": 0.0567, "step": 1563 }, { "epoch": 1.54, "learning_rate": 2.62258260172241e-06, "loss": 0.0528, "step": 1564 }, { "epoch": 1.54, "learning_rate": 2.6118099316278287e-06, "loss": 0.0459, "step": 1565 }, { "epoch": 1.54, "learning_rate": 2.6010561079587817e-06, "loss": 0.0509, "step": 1566 }, { "epoch": 1.55, "learning_rate": 2.5903211581470346e-06, "loss": 0.0507, "step": 1567 }, { "epoch": 1.55, "learning_rate": 2.5796051095762175e-06, "loss": 0.0591, "step": 1568 }, { "epoch": 1.55, "learning_rate": 2.568907989581738e-06, "loss": 0.045, "step": 1569 }, { "epoch": 1.55, "learning_rate": 2.5582298254507274e-06, "loss": 0.053, "step": 1570 }, { "epoch": 1.55, "learning_rate": 2.5475706444219505e-06, "loss": 0.0511, "step": 1571 }, { "epoch": 1.55, "learning_rate": 2.5369304736857635e-06, "loss": 0.051, "step": 1572 }, { "epoch": 1.55, "learning_rate": 2.5263093403840145e-06, "loss": 0.0533, "step": 1573 }, { "epoch": 1.55, "learning_rate": 2.5157072716100028e-06, "loss": 0.0493, "step": 1574 }, { "epoch": 1.55, "learning_rate": 2.505124294408383e-06, "loss": 0.0501, "step": 1575 }, { "epoch": 1.55, "learning_rate": 2.494560435775124e-06, "loss": 0.0513, "step": 1576 }, { "epoch": 1.56, "learning_rate": 2.48401572265741e-06, "loss": 0.0527, "step": 1577 }, { "epoch": 1.56, "learning_rate": 2.4734901819536015e-06, "loss": 0.0548, "step": 1578 }, { "epoch": 1.56, "learning_rate": 2.46298384051314e-06, "loss": 0.0524, "step": 1579 }, { "epoch": 1.56, "learning_rate": 2.452496725136503e-06, "loss": 0.0506, "step": 1580 }, { "epoch": 1.56, "learning_rate": 2.4420288625751133e-06, "loss": 0.0547, "step": 1581 }, { "epoch": 1.56, "learning_rate": 2.431580279531295e-06, "loss": 0.0533, "step": 1582 }, { "epoch": 1.56, "learning_rate": 2.4211510026581765e-06, "loss": 0.0535, "step": 1583 }, { "epoch": 1.56, "learning_rate": 2.4107410585596547e-06, "loss": 0.0508, "step": 1584 }, { "epoch": 1.56, "learning_rate": 2.400350473790296e-06, "loss": 0.0557, "step": 1585 }, { "epoch": 1.56, "learning_rate": 2.3899792748552987e-06, "loss": 0.053, "step": 1586 }, { "epoch": 1.57, "learning_rate": 2.3796274882103964e-06, "loss": 0.0486, "step": 1587 }, { "epoch": 1.57, "learning_rate": 2.3692951402618082e-06, "loss": 0.054, "step": 1588 }, { "epoch": 1.57, "learning_rate": 2.3589822573661737e-06, "loss": 0.056, "step": 1589 }, { "epoch": 1.57, "learning_rate": 2.3486888658304694e-06, "loss": 0.0501, "step": 1590 }, { "epoch": 1.57, "learning_rate": 2.3384149919119616e-06, "loss": 0.0512, "step": 1591 }, { "epoch": 1.57, "learning_rate": 2.32816066181812e-06, "loss": 0.0473, "step": 1592 }, { "epoch": 1.57, "learning_rate": 2.317925901706568e-06, "loss": 0.0523, "step": 1593 }, { "epoch": 1.57, "learning_rate": 2.3077107376850005e-06, "loss": 0.0479, "step": 1594 }, { "epoch": 1.57, "learning_rate": 2.297515195811135e-06, "loss": 0.0547, "step": 1595 }, { "epoch": 1.57, "learning_rate": 2.287339302092627e-06, "loss": 0.0506, "step": 1596 }, { "epoch": 1.57, "learning_rate": 2.277183082487011e-06, "loss": 0.05, "step": 1597 }, { "epoch": 1.58, "learning_rate": 2.2670465629016447e-06, "loss": 0.0538, "step": 1598 }, { "epoch": 1.58, "learning_rate": 2.256929769193621e-06, "loss": 0.0552, "step": 1599 }, { "epoch": 1.58, "learning_rate": 2.2468327271697256e-06, "loss": 0.051, "step": 1600 }, { "epoch": 1.58, "learning_rate": 2.2367554625863496e-06, "loss": 0.0479, "step": 1601 }, { "epoch": 1.58, "learning_rate": 2.226698001149444e-06, "loss": 0.0439, "step": 1602 }, { "epoch": 1.58, "learning_rate": 2.2166603685144363e-06, "loss": 0.0565, "step": 1603 }, { "epoch": 1.58, "learning_rate": 2.206642590286179e-06, "loss": 0.0429, "step": 1604 }, { "epoch": 1.58, "learning_rate": 2.1966446920188734e-06, "loss": 0.0512, "step": 1605 }, { "epoch": 1.58, "learning_rate": 2.186666699216008e-06, "loss": 0.0522, "step": 1606 }, { "epoch": 1.58, "learning_rate": 2.1767086373303057e-06, "loss": 0.0491, "step": 1607 }, { "epoch": 1.59, "learning_rate": 2.1667705317636333e-06, "loss": 0.0475, "step": 1608 }, { "epoch": 1.59, "learning_rate": 2.156852407866964e-06, "loss": 0.0469, "step": 1609 }, { "epoch": 1.59, "learning_rate": 2.1469542909402887e-06, "loss": 0.0573, "step": 1610 }, { "epoch": 1.59, "learning_rate": 2.1370762062325746e-06, "loss": 0.0496, "step": 1611 }, { "epoch": 1.59, "learning_rate": 2.1272181789416756e-06, "loss": 0.0533, "step": 1612 }, { "epoch": 1.59, "learning_rate": 2.1173802342142967e-06, "loss": 0.0515, "step": 1613 }, { "epoch": 1.59, "learning_rate": 2.1075623971459038e-06, "loss": 0.0478, "step": 1614 }, { "epoch": 1.59, "learning_rate": 2.0977646927806682e-06, "loss": 0.0513, "step": 1615 }, { "epoch": 1.59, "learning_rate": 2.0879871461114178e-06, "loss": 0.0487, "step": 1616 }, { "epoch": 1.59, "learning_rate": 2.0782297820795462e-06, "loss": 0.0459, "step": 1617 }, { "epoch": 1.6, "learning_rate": 2.068492625574975e-06, "loss": 0.0571, "step": 1618 }, { "epoch": 1.6, "learning_rate": 2.0587757014360686e-06, "loss": 0.0572, "step": 1619 }, { "epoch": 1.6, "learning_rate": 2.0490790344495923e-06, "loss": 0.0484, "step": 1620 }, { "epoch": 1.6, "learning_rate": 2.039402649350626e-06, "loss": 0.0522, "step": 1621 }, { "epoch": 1.6, "learning_rate": 2.029746570822524e-06, "loss": 0.0518, "step": 1622 }, { "epoch": 1.6, "learning_rate": 2.0201108234968305e-06, "loss": 0.0511, "step": 1623 }, { "epoch": 1.6, "learning_rate": 2.010495431953231e-06, "loss": 0.0475, "step": 1624 }, { "epoch": 1.6, "learning_rate": 2.0009004207194914e-06, "loss": 0.052, "step": 1625 }, { "epoch": 1.6, "learning_rate": 1.9913258142713798e-06, "loss": 0.0516, "step": 1626 }, { "epoch": 1.6, "learning_rate": 1.9817716370326236e-06, "loss": 0.05, "step": 1627 }, { "epoch": 1.61, "learning_rate": 1.972237913374829e-06, "loss": 0.0526, "step": 1628 }, { "epoch": 1.61, "learning_rate": 1.9627246676174363e-06, "loss": 0.0534, "step": 1629 }, { "epoch": 1.61, "learning_rate": 1.95323192402764e-06, "loss": 0.0589, "step": 1630 }, { "epoch": 1.61, "learning_rate": 1.943759706820344e-06, "loss": 0.0575, "step": 1631 }, { "epoch": 1.61, "learning_rate": 1.934308040158087e-06, "loss": 0.0443, "step": 1632 }, { "epoch": 1.61, "learning_rate": 1.9248769481509834e-06, "loss": 0.0536, "step": 1633 }, { "epoch": 1.61, "learning_rate": 1.9154664548566715e-06, "loss": 0.0595, "step": 1634 }, { "epoch": 1.61, "learning_rate": 1.9060765842802364e-06, "loss": 0.0565, "step": 1635 }, { "epoch": 1.61, "learning_rate": 1.896707360374167e-06, "loss": 0.0563, "step": 1636 }, { "epoch": 1.61, "learning_rate": 1.887358807038272e-06, "loss": 0.0462, "step": 1637 }, { "epoch": 1.62, "learning_rate": 1.8780309481196446e-06, "loss": 0.0573, "step": 1638 }, { "epoch": 1.62, "learning_rate": 1.868723807412578e-06, "loss": 0.0431, "step": 1639 }, { "epoch": 1.62, "learning_rate": 1.8594374086585265e-06, "loss": 0.0496, "step": 1640 }, { "epoch": 1.62, "learning_rate": 1.8501717755460246e-06, "loss": 0.0494, "step": 1641 }, { "epoch": 1.62, "learning_rate": 1.8409269317106382e-06, "loss": 0.0466, "step": 1642 }, { "epoch": 1.62, "learning_rate": 1.8317029007349086e-06, "loss": 0.055, "step": 1643 }, { "epoch": 1.62, "learning_rate": 1.8224997061482751e-06, "loss": 0.0521, "step": 1644 }, { "epoch": 1.62, "learning_rate": 1.8133173714270368e-06, "loss": 0.0504, "step": 1645 }, { "epoch": 1.62, "learning_rate": 1.8041559199942726e-06, "loss": 0.0585, "step": 1646 }, { "epoch": 1.62, "learning_rate": 1.7950153752197984e-06, "loss": 0.0446, "step": 1647 }, { "epoch": 1.63, "learning_rate": 1.78589576042009e-06, "loss": 0.0555, "step": 1648 }, { "epoch": 1.63, "learning_rate": 1.7767970988582461e-06, "loss": 0.0472, "step": 1649 }, { "epoch": 1.63, "learning_rate": 1.7677194137439036e-06, "loss": 0.0497, "step": 1650 }, { "epoch": 1.63, "learning_rate": 1.758662728233196e-06, "loss": 0.059, "step": 1651 }, { "epoch": 1.63, "learning_rate": 1.7496270654286918e-06, "loss": 0.0543, "step": 1652 }, { "epoch": 1.63, "learning_rate": 1.7406124483793286e-06, "loss": 0.0497, "step": 1653 }, { "epoch": 1.63, "learning_rate": 1.731618900080363e-06, "loss": 0.0527, "step": 1654 }, { "epoch": 1.63, "learning_rate": 1.7226464434733015e-06, "loss": 0.0507, "step": 1655 }, { "epoch": 1.63, "learning_rate": 1.7136951014458569e-06, "loss": 0.0614, "step": 1656 }, { "epoch": 1.63, "learning_rate": 1.7047648968318697e-06, "loss": 0.0469, "step": 1657 }, { "epoch": 1.64, "learning_rate": 1.6958558524112757e-06, "loss": 0.0587, "step": 1658 }, { "epoch": 1.64, "learning_rate": 1.686967990910018e-06, "loss": 0.0542, "step": 1659 }, { "epoch": 1.64, "learning_rate": 1.6781013350000165e-06, "loss": 0.0489, "step": 1660 }, { "epoch": 1.64, "learning_rate": 1.669255907299091e-06, "loss": 0.0572, "step": 1661 }, { "epoch": 1.64, "learning_rate": 1.660431730370916e-06, "loss": 0.056, "step": 1662 }, { "epoch": 1.64, "learning_rate": 1.6516288267249513e-06, "loss": 0.0475, "step": 1663 }, { "epoch": 1.64, "learning_rate": 1.642847218816398e-06, "loss": 0.0538, "step": 1664 }, { "epoch": 1.64, "learning_rate": 1.634086929046127e-06, "loss": 0.0524, "step": 1665 }, { "epoch": 1.64, "learning_rate": 1.6253479797606364e-06, "loss": 0.0508, "step": 1666 }, { "epoch": 1.64, "learning_rate": 1.61663039325198e-06, "loss": 0.0505, "step": 1667 }, { "epoch": 1.64, "learning_rate": 1.6079341917577262e-06, "loss": 0.0533, "step": 1668 }, { "epoch": 1.65, "learning_rate": 1.5992593974608817e-06, "loss": 0.0509, "step": 1669 }, { "epoch": 1.65, "learning_rate": 1.5906060324898575e-06, "loss": 0.0505, "step": 1670 }, { "epoch": 1.65, "learning_rate": 1.5819741189183902e-06, "loss": 0.049, "step": 1671 }, { "epoch": 1.65, "learning_rate": 1.573363678765506e-06, "loss": 0.0565, "step": 1672 }, { "epoch": 1.65, "learning_rate": 1.5647747339954445e-06, "loss": 0.0443, "step": 1673 }, { "epoch": 1.65, "learning_rate": 1.5562073065176253e-06, "loss": 0.0534, "step": 1674 }, { "epoch": 1.65, "learning_rate": 1.547661418186568e-06, "loss": 0.05, "step": 1675 }, { "epoch": 1.65, "learning_rate": 1.5391370908018589e-06, "loss": 0.05, "step": 1676 }, { "epoch": 1.65, "learning_rate": 1.5306343461080753e-06, "loss": 0.0545, "step": 1677 }, { "epoch": 1.65, "learning_rate": 1.522153205794742e-06, "loss": 0.0573, "step": 1678 }, { "epoch": 1.66, "learning_rate": 1.5136936914962818e-06, "loss": 0.0467, "step": 1679 }, { "epoch": 1.66, "learning_rate": 1.505255824791939e-06, "loss": 0.0509, "step": 1680 }, { "epoch": 1.66, "learning_rate": 1.4968396272057484e-06, "loss": 0.0526, "step": 1681 }, { "epoch": 1.66, "learning_rate": 1.4884451202064609e-06, "loss": 0.0424, "step": 1682 }, { "epoch": 1.66, "learning_rate": 1.4800723252075067e-06, "loss": 0.0518, "step": 1683 }, { "epoch": 1.66, "learning_rate": 1.4717212635669208e-06, "loss": 0.0488, "step": 1684 }, { "epoch": 1.66, "learning_rate": 1.4633919565873033e-06, "loss": 0.0529, "step": 1685 }, { "epoch": 1.66, "learning_rate": 1.4550844255157647e-06, "loss": 0.0544, "step": 1686 }, { "epoch": 1.66, "learning_rate": 1.4467986915438602e-06, "loss": 0.0488, "step": 1687 }, { "epoch": 1.66, "learning_rate": 1.4385347758075518e-06, "loss": 0.0496, "step": 1688 }, { "epoch": 1.67, "learning_rate": 1.4302926993871347e-06, "loss": 0.0517, "step": 1689 }, { "epoch": 1.67, "learning_rate": 1.4220724833072052e-06, "loss": 0.0485, "step": 1690 }, { "epoch": 1.67, "learning_rate": 1.4138741485365892e-06, "loss": 0.0445, "step": 1691 }, { "epoch": 1.67, "learning_rate": 1.4056977159883011e-06, "loss": 0.0566, "step": 1692 }, { "epoch": 1.67, "learning_rate": 1.3975432065194804e-06, "loss": 0.0537, "step": 1693 }, { "epoch": 1.67, "learning_rate": 1.3894106409313445e-06, "loss": 0.0487, "step": 1694 }, { "epoch": 1.67, "learning_rate": 1.38130003996914e-06, "loss": 0.0525, "step": 1695 }, { "epoch": 1.67, "learning_rate": 1.373211424322074e-06, "loss": 0.0568, "step": 1696 }, { "epoch": 1.67, "learning_rate": 1.3651448146232816e-06, "loss": 0.0511, "step": 1697 }, { "epoch": 1.67, "learning_rate": 1.357100231449756e-06, "loss": 0.0526, "step": 1698 }, { "epoch": 1.68, "learning_rate": 1.3490776953223107e-06, "loss": 0.0548, "step": 1699 }, { "epoch": 1.68, "learning_rate": 1.3410772267055093e-06, "loss": 0.0493, "step": 1700 }, { "epoch": 1.68, "learning_rate": 1.3330988460076367e-06, "loss": 0.0509, "step": 1701 }, { "epoch": 1.68, "learning_rate": 1.3251425735806212e-06, "loss": 0.0517, "step": 1702 }, { "epoch": 1.68, "learning_rate": 1.3172084297200028e-06, "loss": 0.05, "step": 1703 }, { "epoch": 1.68, "learning_rate": 1.309296434664874e-06, "loss": 0.0431, "step": 1704 }, { "epoch": 1.68, "learning_rate": 1.3014066085978238e-06, "loss": 0.0464, "step": 1705 }, { "epoch": 1.68, "learning_rate": 1.2935389716448976e-06, "loss": 0.0533, "step": 1706 }, { "epoch": 1.68, "learning_rate": 1.2856935438755314e-06, "loss": 0.0539, "step": 1707 }, { "epoch": 1.68, "learning_rate": 1.2778703453025155e-06, "loss": 0.0565, "step": 1708 }, { "epoch": 1.69, "learning_rate": 1.2700693958819266e-06, "loss": 0.0605, "step": 1709 }, { "epoch": 1.69, "learning_rate": 1.262290715513098e-06, "loss": 0.0478, "step": 1710 }, { "epoch": 1.69, "learning_rate": 1.25453432403855e-06, "loss": 0.0549, "step": 1711 }, { "epoch": 1.69, "learning_rate": 1.2468002412439427e-06, "loss": 0.0486, "step": 1712 }, { "epoch": 1.69, "learning_rate": 1.23908848685804e-06, "loss": 0.0487, "step": 1713 }, { "epoch": 1.69, "learning_rate": 1.2313990805526388e-06, "loss": 0.0503, "step": 1714 }, { "epoch": 1.69, "learning_rate": 1.2237320419425369e-06, "loss": 0.0529, "step": 1715 }, { "epoch": 1.69, "learning_rate": 1.2160873905854665e-06, "loss": 0.0591, "step": 1716 }, { "epoch": 1.69, "learning_rate": 1.2084651459820606e-06, "loss": 0.0546, "step": 1717 }, { "epoch": 1.69, "learning_rate": 1.2008653275757854e-06, "loss": 0.056, "step": 1718 }, { "epoch": 1.7, "learning_rate": 1.193287954752912e-06, "loss": 0.0547, "step": 1719 }, { "epoch": 1.7, "learning_rate": 1.1857330468424466e-06, "loss": 0.0487, "step": 1720 }, { "epoch": 1.7, "learning_rate": 1.1782006231160914e-06, "loss": 0.053, "step": 1721 }, { "epoch": 1.7, "learning_rate": 1.1706907027881987e-06, "loss": 0.0536, "step": 1722 }, { "epoch": 1.7, "learning_rate": 1.1632033050157099e-06, "loss": 0.0481, "step": 1723 }, { "epoch": 1.7, "learning_rate": 1.1557384488981227e-06, "loss": 0.0513, "step": 1724 }, { "epoch": 1.7, "learning_rate": 1.148296153477425e-06, "loss": 0.049, "step": 1725 }, { "epoch": 1.7, "learning_rate": 1.140876437738062e-06, "loss": 0.0493, "step": 1726 }, { "epoch": 1.7, "learning_rate": 1.1334793206068739e-06, "loss": 0.0521, "step": 1727 }, { "epoch": 1.7, "learning_rate": 1.1261048209530635e-06, "loss": 0.0513, "step": 1728 }, { "epoch": 1.71, "learning_rate": 1.1187529575881283e-06, "loss": 0.0446, "step": 1729 }, { "epoch": 1.71, "learning_rate": 1.1114237492658308e-06, "loss": 0.052, "step": 1730 }, { "epoch": 1.71, "learning_rate": 1.1041172146821422e-06, "loss": 0.0502, "step": 1731 }, { "epoch": 1.71, "learning_rate": 1.09683337247519e-06, "loss": 0.0547, "step": 1732 }, { "epoch": 1.71, "learning_rate": 1.0895722412252264e-06, "loss": 0.0549, "step": 1733 }, { "epoch": 1.71, "learning_rate": 1.082333839454559e-06, "loss": 0.0497, "step": 1734 }, { "epoch": 1.71, "learning_rate": 1.0751181856275238e-06, "loss": 0.0486, "step": 1735 }, { "epoch": 1.71, "learning_rate": 1.067925298150425e-06, "loss": 0.0552, "step": 1736 }, { "epoch": 1.71, "learning_rate": 1.0607551953714945e-06, "loss": 0.051, "step": 1737 }, { "epoch": 1.71, "learning_rate": 1.0536078955808394e-06, "loss": 0.0475, "step": 1738 }, { "epoch": 1.71, "learning_rate": 1.0464834170104066e-06, "loss": 0.0517, "step": 1739 }, { "epoch": 1.72, "learning_rate": 1.0393817778339188e-06, "loss": 0.0509, "step": 1740 }, { "epoch": 1.72, "learning_rate": 1.0323029961668463e-06, "loss": 0.0481, "step": 1741 }, { "epoch": 1.72, "learning_rate": 1.025247090066348e-06, "loss": 0.0484, "step": 1742 }, { "epoch": 1.72, "learning_rate": 1.0182140775312321e-06, "loss": 0.0587, "step": 1743 }, { "epoch": 1.72, "learning_rate": 1.0112039765019055e-06, "loss": 0.0436, "step": 1744 }, { "epoch": 1.72, "learning_rate": 1.004216804860334e-06, "loss": 0.0539, "step": 1745 }, { "epoch": 1.72, "learning_rate": 9.972525804299881e-07, "loss": 0.0556, "step": 1746 }, { "epoch": 1.72, "learning_rate": 9.903113209758098e-07, "loss": 0.0493, "step": 1747 }, { "epoch": 1.72, "learning_rate": 9.833930442041506e-07, "loss": 0.0464, "step": 1748 }, { "epoch": 1.72, "learning_rate": 9.764977677627463e-07, "loss": 0.0465, "step": 1749 }, { "epoch": 1.73, "learning_rate": 9.696255092406525e-07, "loss": 0.0528, "step": 1750 }, { "epoch": 1.73, "learning_rate": 9.627762861682155e-07, "loss": 0.0598, "step": 1751 }, { "epoch": 1.73, "learning_rate": 9.55950116017015e-07, "loss": 0.0542, "step": 1752 }, { "epoch": 1.73, "learning_rate": 9.491470161998317e-07, "loss": 0.0497, "step": 1753 }, { "epoch": 1.73, "learning_rate": 9.42367004070589e-07, "loss": 0.0446, "step": 1754 }, { "epoch": 1.73, "learning_rate": 9.356100969243231e-07, "loss": 0.0589, "step": 1755 }, { "epoch": 1.73, "learning_rate": 9.288763119971289e-07, "loss": 0.0502, "step": 1756 }, { "epoch": 1.73, "learning_rate": 9.221656664661161e-07, "loss": 0.0509, "step": 1757 }, { "epoch": 1.73, "learning_rate": 9.15478177449377e-07, "loss": 0.0505, "step": 1758 }, { "epoch": 1.73, "learning_rate": 9.088138620059251e-07, "loss": 0.0511, "step": 1759 }, { "epoch": 1.74, "learning_rate": 9.021727371356681e-07, "loss": 0.0455, "step": 1760 }, { "epoch": 1.74, "learning_rate": 8.955548197793496e-07, "loss": 0.0515, "step": 1761 }, { "epoch": 1.74, "learning_rate": 8.889601268185233e-07, "loss": 0.0572, "step": 1762 }, { "epoch": 1.74, "learning_rate": 8.823886750754884e-07, "loss": 0.0551, "step": 1763 }, { "epoch": 1.74, "learning_rate": 8.758404813132704e-07, "loss": 0.0548, "step": 1764 }, { "epoch": 1.74, "learning_rate": 8.693155622355587e-07, "loss": 0.0567, "step": 1765 }, { "epoch": 1.74, "learning_rate": 8.628139344866704e-07, "loss": 0.0466, "step": 1766 }, { "epoch": 1.74, "learning_rate": 8.563356146515156e-07, "loss": 0.0484, "step": 1767 }, { "epoch": 1.74, "learning_rate": 8.498806192555431e-07, "loss": 0.0531, "step": 1768 }, { "epoch": 1.74, "learning_rate": 8.434489647647093e-07, "loss": 0.0503, "step": 1769 }, { "epoch": 1.75, "learning_rate": 8.370406675854237e-07, "loss": 0.0612, "step": 1770 }, { "epoch": 1.75, "learning_rate": 8.30655744064519e-07, "loss": 0.0521, "step": 1771 }, { "epoch": 1.75, "learning_rate": 8.242942104892015e-07, "loss": 0.045, "step": 1772 }, { "epoch": 1.75, "learning_rate": 8.17956083087017e-07, "loss": 0.0515, "step": 1773 }, { "epoch": 1.75, "learning_rate": 8.116413780257992e-07, "loss": 0.0495, "step": 1774 }, { "epoch": 1.75, "learning_rate": 8.053501114136342e-07, "loss": 0.0429, "step": 1775 }, { "epoch": 1.75, "learning_rate": 7.990822992988267e-07, "loss": 0.0539, "step": 1776 }, { "epoch": 1.75, "learning_rate": 7.928379576698419e-07, "loss": 0.0526, "step": 1777 }, { "epoch": 1.75, "learning_rate": 7.866171024552827e-07, "loss": 0.0577, "step": 1778 }, { "epoch": 1.75, "learning_rate": 7.804197495238341e-07, "loss": 0.049, "step": 1779 }, { "epoch": 1.76, "learning_rate": 7.742459146842362e-07, "loss": 0.0474, "step": 1780 }, { "epoch": 1.76, "learning_rate": 7.680956136852302e-07, "loss": 0.0514, "step": 1781 }, { "epoch": 1.76, "learning_rate": 7.619688622155319e-07, "loss": 0.0562, "step": 1782 }, { "epoch": 1.76, "learning_rate": 7.558656759037796e-07, "loss": 0.0508, "step": 1783 }, { "epoch": 1.76, "learning_rate": 7.497860703185e-07, "loss": 0.0516, "step": 1784 }, { "epoch": 1.76, "learning_rate": 7.437300609680742e-07, "loss": 0.0448, "step": 1785 }, { "epoch": 1.76, "learning_rate": 7.376976633006828e-07, "loss": 0.0477, "step": 1786 }, { "epoch": 1.76, "learning_rate": 7.316888927042842e-07, "loss": 0.0555, "step": 1787 }, { "epoch": 1.76, "learning_rate": 7.25703764506559e-07, "loss": 0.053, "step": 1788 }, { "epoch": 1.76, "learning_rate": 7.197422939748877e-07, "loss": 0.0453, "step": 1789 }, { "epoch": 1.77, "learning_rate": 7.13804496316296e-07, "loss": 0.0526, "step": 1790 }, { "epoch": 1.77, "learning_rate": 7.078903866774212e-07, "loss": 0.0502, "step": 1791 }, { "epoch": 1.77, "learning_rate": 7.019999801444844e-07, "loss": 0.0533, "step": 1792 }, { "epoch": 1.77, "learning_rate": 6.961332917432318e-07, "loss": 0.0596, "step": 1793 }, { "epoch": 1.77, "learning_rate": 6.902903364389179e-07, "loss": 0.0547, "step": 1794 }, { "epoch": 1.77, "learning_rate": 6.84471129136246e-07, "loss": 0.0533, "step": 1795 }, { "epoch": 1.77, "learning_rate": 6.786756846793518e-07, "loss": 0.0602, "step": 1796 }, { "epoch": 1.77, "learning_rate": 6.729040178517454e-07, "loss": 0.049, "step": 1797 }, { "epoch": 1.77, "learning_rate": 6.671561433762908e-07, "loss": 0.0488, "step": 1798 }, { "epoch": 1.77, "learning_rate": 6.614320759151538e-07, "loss": 0.0482, "step": 1799 }, { "epoch": 1.78, "learning_rate": 6.557318300697735e-07, "loss": 0.0506, "step": 1800 }, { "epoch": 1.78, "learning_rate": 6.500554203808262e-07, "loss": 0.0468, "step": 1801 }, { "epoch": 1.78, "learning_rate": 6.444028613281783e-07, "loss": 0.0546, "step": 1802 }, { "epoch": 1.78, "learning_rate": 6.387741673308634e-07, "loss": 0.0486, "step": 1803 }, { "epoch": 1.78, "learning_rate": 6.331693527470306e-07, "loss": 0.0524, "step": 1804 }, { "epoch": 1.78, "learning_rate": 6.275884318739234e-07, "loss": 0.0449, "step": 1805 }, { "epoch": 1.78, "learning_rate": 6.22031418947826e-07, "loss": 0.0436, "step": 1806 }, { "epoch": 1.78, "learning_rate": 6.164983281440462e-07, "loss": 0.0566, "step": 1807 }, { "epoch": 1.78, "learning_rate": 6.109891735768603e-07, "loss": 0.0504, "step": 1808 }, { "epoch": 1.78, "learning_rate": 6.055039692994902e-07, "loss": 0.0482, "step": 1809 }, { "epoch": 1.79, "learning_rate": 6.000427293040645e-07, "loss": 0.049, "step": 1810 }, { "epoch": 1.79, "learning_rate": 5.946054675215785e-07, "loss": 0.045, "step": 1811 }, { "epoch": 1.79, "learning_rate": 5.891921978218662e-07, "loss": 0.0501, "step": 1812 }, { "epoch": 1.79, "learning_rate": 5.838029340135531e-07, "loss": 0.0438, "step": 1813 }, { "epoch": 1.79, "learning_rate": 5.78437689844038e-07, "loss": 0.054, "step": 1814 }, { "epoch": 1.79, "learning_rate": 5.730964789994386e-07, "loss": 0.0482, "step": 1815 }, { "epoch": 1.79, "learning_rate": 5.677793151045763e-07, "loss": 0.0515, "step": 1816 }, { "epoch": 1.79, "learning_rate": 5.624862117229224e-07, "loss": 0.0458, "step": 1817 }, { "epoch": 1.79, "learning_rate": 5.572171823565797e-07, "loss": 0.0458, "step": 1818 }, { "epoch": 1.79, "learning_rate": 5.519722404462335e-07, "loss": 0.0576, "step": 1819 }, { "epoch": 1.79, "learning_rate": 5.467513993711349e-07, "loss": 0.0582, "step": 1820 }, { "epoch": 1.8, "learning_rate": 5.41554672449045e-07, "loss": 0.0525, "step": 1821 }, { "epoch": 1.8, "learning_rate": 5.363820729362224e-07, "loss": 0.0539, "step": 1822 }, { "epoch": 1.8, "learning_rate": 5.312336140273721e-07, "loss": 0.0437, "step": 1823 }, { "epoch": 1.8, "learning_rate": 5.261093088556257e-07, "loss": 0.0512, "step": 1824 }, { "epoch": 1.8, "learning_rate": 5.210091704924947e-07, "loss": 0.0557, "step": 1825 }, { "epoch": 1.8, "learning_rate": 5.159332119478488e-07, "loss": 0.055, "step": 1826 }, { "epoch": 1.8, "learning_rate": 5.108814461698752e-07, "loss": 0.0538, "step": 1827 }, { "epoch": 1.8, "learning_rate": 5.058538860450513e-07, "loss": 0.0498, "step": 1828 }, { "epoch": 1.8, "learning_rate": 5.008505443981038e-07, "loss": 0.05, "step": 1829 }, { "epoch": 1.8, "learning_rate": 4.958714339919845e-07, "loss": 0.0454, "step": 1830 }, { "epoch": 1.81, "learning_rate": 4.909165675278327e-07, "loss": 0.0555, "step": 1831 }, { "epoch": 1.81, "learning_rate": 4.859859576449444e-07, "loss": 0.0476, "step": 1832 }, { "epoch": 1.81, "learning_rate": 4.810796169207398e-07, "loss": 0.044, "step": 1833 }, { "epoch": 1.81, "learning_rate": 4.7619755787073343e-07, "loss": 0.0485, "step": 1834 }, { "epoch": 1.81, "learning_rate": 4.713397929484964e-07, "loss": 0.0511, "step": 1835 }, { "epoch": 1.81, "learning_rate": 4.6650633454562776e-07, "loss": 0.0511, "step": 1836 }, { "epoch": 1.81, "learning_rate": 4.6169719499173037e-07, "loss": 0.0557, "step": 1837 }, { "epoch": 1.81, "learning_rate": 4.569123865543623e-07, "loss": 0.0539, "step": 1838 }, { "epoch": 1.81, "learning_rate": 4.5215192143902577e-07, "loss": 0.0506, "step": 1839 }, { "epoch": 1.81, "learning_rate": 4.474158117891181e-07, "loss": 0.0522, "step": 1840 }, { "epoch": 1.82, "learning_rate": 4.4270406968591416e-07, "loss": 0.0494, "step": 1841 }, { "epoch": 1.82, "learning_rate": 4.380167071485253e-07, "loss": 0.0534, "step": 1842 }, { "epoch": 1.82, "learning_rate": 4.33353736133878e-07, "loss": 0.052, "step": 1843 }, { "epoch": 1.82, "learning_rate": 4.287151685366764e-07, "loss": 0.0427, "step": 1844 }, { "epoch": 1.82, "learning_rate": 4.241010161893711e-07, "loss": 0.0534, "step": 1845 }, { "epoch": 1.82, "learning_rate": 4.1951129086214015e-07, "loss": 0.0515, "step": 1846 }, { "epoch": 1.82, "learning_rate": 4.149460042628439e-07, "loss": 0.0544, "step": 1847 }, { "epoch": 1.82, "learning_rate": 4.104051680370058e-07, "loss": 0.0551, "step": 1848 }, { "epoch": 1.82, "learning_rate": 4.058887937677769e-07, "loss": 0.0506, "step": 1849 }, { "epoch": 1.82, "learning_rate": 4.0139689297591265e-07, "loss": 0.0459, "step": 1850 }, { "epoch": 1.83, "learning_rate": 3.9692947711973293e-07, "loss": 0.0497, "step": 1851 }, { "epoch": 1.83, "learning_rate": 3.9248655759510755e-07, "loss": 0.0483, "step": 1852 }, { "epoch": 1.83, "learning_rate": 3.8806814573541185e-07, "loss": 0.0555, "step": 1853 }, { "epoch": 1.83, "learning_rate": 3.836742528115078e-07, "loss": 0.0547, "step": 1854 }, { "epoch": 1.83, "learning_rate": 3.793048900317142e-07, "loss": 0.0523, "step": 1855 }, { "epoch": 1.83, "learning_rate": 3.74960068541772e-07, "loss": 0.0449, "step": 1856 }, { "epoch": 1.83, "learning_rate": 3.706397994248256e-07, "loss": 0.0521, "step": 1857 }, { "epoch": 1.83, "learning_rate": 3.663440937013818e-07, "loss": 0.0515, "step": 1858 }, { "epoch": 1.83, "learning_rate": 3.6207296232929736e-07, "loss": 0.0512, "step": 1859 }, { "epoch": 1.83, "learning_rate": 3.578264162037348e-07, "loss": 0.0486, "step": 1860 }, { "epoch": 1.84, "learning_rate": 3.536044661571503e-07, "loss": 0.0457, "step": 1861 }, { "epoch": 1.84, "learning_rate": 3.4940712295925215e-07, "loss": 0.0539, "step": 1862 }, { "epoch": 1.84, "learning_rate": 3.4523439731698026e-07, "loss": 0.0541, "step": 1863 }, { "epoch": 1.84, "learning_rate": 3.410862998744813e-07, "loss": 0.0511, "step": 1864 }, { "epoch": 1.84, "learning_rate": 3.369628412130743e-07, "loss": 0.0513, "step": 1865 }, { "epoch": 1.84, "learning_rate": 3.3286403185123104e-07, "loss": 0.0582, "step": 1866 }, { "epoch": 1.84, "learning_rate": 3.2878988224454346e-07, "loss": 0.0493, "step": 1867 }, { "epoch": 1.84, "learning_rate": 3.247404027856993e-07, "loss": 0.0512, "step": 1868 }, { "epoch": 1.84, "learning_rate": 3.2071560380445674e-07, "loss": 0.0572, "step": 1869 }, { "epoch": 1.84, "learning_rate": 3.167154955676177e-07, "loss": 0.0457, "step": 1870 }, { "epoch": 1.85, "learning_rate": 3.127400882789988e-07, "loss": 0.0483, "step": 1871 }, { "epoch": 1.85, "learning_rate": 3.0878939207940496e-07, "loss": 0.0472, "step": 1872 }, { "epoch": 1.85, "learning_rate": 3.048634170466136e-07, "loss": 0.0436, "step": 1873 }, { "epoch": 1.85, "learning_rate": 3.0096217319533386e-07, "loss": 0.0521, "step": 1874 }, { "epoch": 1.85, "learning_rate": 2.9708567047719405e-07, "loss": 0.0546, "step": 1875 }, { "epoch": 1.85, "learning_rate": 2.932339187807043e-07, "loss": 0.0441, "step": 1876 }, { "epoch": 1.85, "learning_rate": 2.8940692793124615e-07, "loss": 0.0521, "step": 1877 }, { "epoch": 1.85, "learning_rate": 2.8560470769102953e-07, "loss": 0.0525, "step": 1878 }, { "epoch": 1.85, "learning_rate": 2.81827267759085e-07, "loss": 0.0483, "step": 1879 }, { "epoch": 1.85, "learning_rate": 2.78074617771229e-07, "loss": 0.046, "step": 1880 }, { "epoch": 1.86, "learning_rate": 2.7434676730003886e-07, "loss": 0.0552, "step": 1881 }, { "epoch": 1.86, "learning_rate": 2.7064372585483667e-07, "loss": 0.0514, "step": 1882 }, { "epoch": 1.86, "learning_rate": 2.6696550288165315e-07, "loss": 0.0471, "step": 1883 }, { "epoch": 1.86, "learning_rate": 2.6331210776321614e-07, "loss": 0.0493, "step": 1884 }, { "epoch": 1.86, "learning_rate": 2.5968354981891433e-07, "loss": 0.0476, "step": 1885 }, { "epoch": 1.86, "learning_rate": 2.560798383047869e-07, "loss": 0.0498, "step": 1886 }, { "epoch": 1.86, "learning_rate": 2.525009824134872e-07, "loss": 0.0459, "step": 1887 }, { "epoch": 1.86, "learning_rate": 2.489469912742637e-07, "loss": 0.047, "step": 1888 }, { "epoch": 1.86, "learning_rate": 2.4541787395294335e-07, "loss": 0.0515, "step": 1889 }, { "epoch": 1.86, "learning_rate": 2.4191363945189726e-07, "loss": 0.0546, "step": 1890 }, { "epoch": 1.86, "learning_rate": 2.384342967100284e-07, "loss": 0.0505, "step": 1891 }, { "epoch": 1.87, "learning_rate": 2.349798546027382e-07, "loss": 0.0526, "step": 1892 }, { "epoch": 1.87, "learning_rate": 2.3155032194191573e-07, "loss": 0.0476, "step": 1893 }, { "epoch": 1.87, "learning_rate": 2.281457074759008e-07, "loss": 0.0495, "step": 1894 }, { "epoch": 1.87, "learning_rate": 2.2476601988947965e-07, "loss": 0.0446, "step": 1895 }, { "epoch": 1.87, "learning_rate": 2.214112678038438e-07, "loss": 0.0497, "step": 1896 }, { "epoch": 1.87, "learning_rate": 2.1808145977658236e-07, "loss": 0.0525, "step": 1897 }, { "epoch": 1.87, "learning_rate": 2.1477660430165415e-07, "loss": 0.0537, "step": 1898 }, { "epoch": 1.87, "learning_rate": 2.1149670980936566e-07, "loss": 0.0495, "step": 1899 }, { "epoch": 1.87, "learning_rate": 2.0824178466635315e-07, "loss": 0.0525, "step": 1900 }, { "epoch": 1.87, "learning_rate": 2.0501183717555494e-07, "loss": 0.0502, "step": 1901 }, { "epoch": 1.88, "learning_rate": 2.0180687557619816e-07, "loss": 0.0531, "step": 1902 }, { "epoch": 1.88, "learning_rate": 1.9862690804377194e-07, "loss": 0.0524, "step": 1903 }, { "epoch": 1.88, "learning_rate": 1.954719426900098e-07, "loss": 0.0512, "step": 1904 }, { "epoch": 1.88, "learning_rate": 1.9234198756286514e-07, "loss": 0.055, "step": 1905 }, { "epoch": 1.88, "learning_rate": 1.8923705064649468e-07, "loss": 0.0461, "step": 1906 }, { "epoch": 1.88, "learning_rate": 1.8615713986123607e-07, "loss": 0.0481, "step": 1907 }, { "epoch": 1.88, "learning_rate": 1.8310226306359036e-07, "loss": 0.0527, "step": 1908 }, { "epoch": 1.88, "learning_rate": 1.800724280461963e-07, "loss": 0.0536, "step": 1909 }, { "epoch": 1.88, "learning_rate": 1.7706764253781595e-07, "loss": 0.0525, "step": 1910 }, { "epoch": 1.88, "learning_rate": 1.740879142033125e-07, "loss": 0.0445, "step": 1911 }, { "epoch": 1.89, "learning_rate": 1.711332506436314e-07, "loss": 0.0455, "step": 1912 }, { "epoch": 1.89, "learning_rate": 1.682036593957792e-07, "loss": 0.0515, "step": 1913 }, { "epoch": 1.89, "learning_rate": 1.652991479328092e-07, "loss": 0.0492, "step": 1914 }, { "epoch": 1.89, "learning_rate": 1.624197236637959e-07, "loss": 0.0435, "step": 1915 }, { "epoch": 1.89, "learning_rate": 1.5956539393382043e-07, "loss": 0.0522, "step": 1916 }, { "epoch": 1.89, "learning_rate": 1.567361660239508e-07, "loss": 0.0506, "step": 1917 }, { "epoch": 1.89, "learning_rate": 1.539320471512229e-07, "loss": 0.0479, "step": 1918 }, { "epoch": 1.89, "learning_rate": 1.5115304446862268e-07, "loss": 0.0516, "step": 1919 }, { "epoch": 1.89, "learning_rate": 1.4839916506506623e-07, "loss": 0.051, "step": 1920 }, { "epoch": 1.89, "learning_rate": 1.4567041596538545e-07, "loss": 0.0485, "step": 1921 }, { "epoch": 1.9, "learning_rate": 1.429668041303056e-07, "loss": 0.057, "step": 1922 }, { "epoch": 1.9, "learning_rate": 1.4028833645643113e-07, "loss": 0.0514, "step": 1923 }, { "epoch": 1.9, "learning_rate": 1.3763501977622329e-07, "loss": 0.0571, "step": 1924 }, { "epoch": 1.9, "learning_rate": 1.3500686085799242e-07, "loss": 0.0468, "step": 1925 }, { "epoch": 1.9, "learning_rate": 1.324038664058669e-07, "loss": 0.0504, "step": 1926 }, { "epoch": 1.9, "learning_rate": 1.2982604305979085e-07, "loss": 0.0562, "step": 1927 }, { "epoch": 1.9, "learning_rate": 1.2727339739549206e-07, "loss": 0.0444, "step": 1928 }, { "epoch": 1.9, "learning_rate": 1.2474593592447958e-07, "loss": 0.0465, "step": 1929 }, { "epoch": 1.9, "learning_rate": 1.2224366509401732e-07, "loss": 0.0549, "step": 1930 }, { "epoch": 1.9, "learning_rate": 1.1976659128711044e-07, "loss": 0.0557, "step": 1931 }, { "epoch": 1.91, "learning_rate": 1.1731472082249007e-07, "loss": 0.0547, "step": 1932 }, { "epoch": 1.91, "learning_rate": 1.1488805995459761e-07, "loss": 0.0455, "step": 1933 }, { "epoch": 1.91, "learning_rate": 1.1248661487356593e-07, "loss": 0.0548, "step": 1934 }, { "epoch": 1.91, "learning_rate": 1.1011039170520488e-07, "loss": 0.0555, "step": 1935 }, { "epoch": 1.91, "learning_rate": 1.0775939651099021e-07, "loss": 0.0505, "step": 1936 }, { "epoch": 1.91, "learning_rate": 1.0543363528803696e-07, "loss": 0.0493, "step": 1937 }, { "epoch": 1.91, "learning_rate": 1.0313311396909942e-07, "loss": 0.0505, "step": 1938 }, { "epoch": 1.91, "learning_rate": 1.0085783842254115e-07, "loss": 0.0473, "step": 1939 }, { "epoch": 1.91, "learning_rate": 9.860781445233058e-08, "loss": 0.0447, "step": 1940 }, { "epoch": 1.91, "learning_rate": 9.638304779801877e-08, "loss": 0.0478, "step": 1941 }, { "epoch": 1.92, "learning_rate": 9.418354413473163e-08, "loss": 0.0451, "step": 1942 }, { "epoch": 1.92, "learning_rate": 9.20093090731522e-08, "loss": 0.0514, "step": 1943 }, { "epoch": 1.92, "learning_rate": 8.986034815950173e-08, "loss": 0.0558, "step": 1944 }, { "epoch": 1.92, "learning_rate": 8.773666687553417e-08, "loss": 0.0529, "step": 1945 }, { "epoch": 1.92, "learning_rate": 8.563827063851837e-08, "loss": 0.0479, "step": 1946 }, { "epoch": 1.92, "learning_rate": 8.356516480122034e-08, "loss": 0.0477, "step": 1947 }, { "epoch": 1.92, "learning_rate": 8.151735465189658e-08, "loss": 0.047, "step": 1948 }, { "epoch": 1.92, "learning_rate": 7.949484541427633e-08, "loss": 0.0548, "step": 1949 }, { "epoch": 1.92, "learning_rate": 7.749764224754819e-08, "loss": 0.0484, "step": 1950 }, { "epoch": 1.92, "learning_rate": 7.55257502463469e-08, "loss": 0.0501, "step": 1951 }, { "epoch": 1.93, "learning_rate": 7.357917444074548e-08, "loss": 0.0518, "step": 1952 }, { "epoch": 1.93, "learning_rate": 7.165791979623305e-08, "loss": 0.0569, "step": 1953 }, { "epoch": 1.93, "learning_rate": 6.976199121371263e-08, "loss": 0.0497, "step": 1954 }, { "epoch": 1.93, "learning_rate": 6.789139352947783e-08, "loss": 0.0471, "step": 1955 }, { "epoch": 1.93, "learning_rate": 6.604613151521055e-08, "loss": 0.046, "step": 1956 }, { "epoch": 1.93, "learning_rate": 6.422620987796114e-08, "loss": 0.0553, "step": 1957 }, { "epoch": 1.93, "learning_rate": 6.243163326014268e-08, "loss": 0.0428, "step": 1958 }, { "epoch": 1.93, "learning_rate": 6.066240623951336e-08, "loss": 0.0533, "step": 1959 }, { "epoch": 1.93, "learning_rate": 5.8918533329167525e-08, "loss": 0.0511, "step": 1960 }, { "epoch": 1.93, "learning_rate": 5.7200018977526805e-08, "loss": 0.05, "step": 1961 }, { "epoch": 1.93, "learning_rate": 5.550686756832346e-08, "loss": 0.0498, "step": 1962 }, { "epoch": 1.94, "learning_rate": 5.383908342059263e-08, "loss": 0.044, "step": 1963 }, { "epoch": 1.94, "learning_rate": 5.219667078865898e-08, "loss": 0.0495, "step": 1964 }, { "epoch": 1.94, "learning_rate": 5.057963386213116e-08, "loss": 0.0526, "step": 1965 }, { "epoch": 1.94, "learning_rate": 4.898797676588296e-08, "loss": 0.0494, "step": 1966 }, { "epoch": 1.94, "learning_rate": 4.7421703560051045e-08, "loss": 0.0551, "step": 1967 }, { "epoch": 1.94, "learning_rate": 4.588081824001611e-08, "loss": 0.0518, "step": 1968 }, { "epoch": 1.94, "learning_rate": 4.436532473640287e-08, "loss": 0.0523, "step": 1969 }, { "epoch": 1.94, "learning_rate": 4.287522691505897e-08, "loss": 0.0485, "step": 1970 }, { "epoch": 1.94, "learning_rate": 4.1410528577053856e-08, "loss": 0.0458, "step": 1971 }, { "epoch": 1.94, "learning_rate": 3.9971233458665495e-08, "loss": 0.0501, "step": 1972 }, { "epoch": 1.95, "learning_rate": 3.855734523137034e-08, "loss": 0.051, "step": 1973 }, { "epoch": 1.95, "learning_rate": 3.716886750183557e-08, "loss": 0.0418, "step": 1974 }, { "epoch": 1.95, "learning_rate": 3.5805803811906904e-08, "loss": 0.0583, "step": 1975 }, { "epoch": 1.95, "learning_rate": 3.446815763860634e-08, "loss": 0.0568, "step": 1976 }, { "epoch": 1.95, "learning_rate": 3.31559323941133e-08, "loss": 0.045, "step": 1977 }, { "epoch": 1.95, "learning_rate": 3.186913142576464e-08, "loss": 0.0524, "step": 1978 }, { "epoch": 1.95, "learning_rate": 3.0607758016043546e-08, "loss": 0.0468, "step": 1979 }, { "epoch": 1.95, "learning_rate": 2.937181538256617e-08, "loss": 0.0522, "step": 1980 }, { "epoch": 1.95, "learning_rate": 2.816130667808281e-08, "loss": 0.0456, "step": 1981 }, { "epoch": 1.95, "learning_rate": 2.6976234990460095e-08, "loss": 0.0488, "step": 1982 }, { "epoch": 1.96, "learning_rate": 2.581660334268321e-08, "loss": 0.0499, "step": 1983 }, { "epoch": 1.96, "learning_rate": 2.4682414692835945e-08, "loss": 0.0447, "step": 1984 }, { "epoch": 1.96, "learning_rate": 2.35736719341062e-08, "loss": 0.0447, "step": 1985 }, { "epoch": 1.96, "learning_rate": 2.2490377894768266e-08, "loss": 0.0493, "step": 1986 }, { "epoch": 1.96, "learning_rate": 2.1432535338180572e-08, "loss": 0.0545, "step": 1987 }, { "epoch": 1.96, "learning_rate": 2.0400146962779034e-08, "loss": 0.0515, "step": 1988 }, { "epoch": 1.96, "learning_rate": 1.9393215402068176e-08, "loss": 0.0463, "step": 1989 }, { "epoch": 1.96, "learning_rate": 1.8411743224614477e-08, "loss": 0.0473, "step": 1990 }, { "epoch": 1.96, "learning_rate": 1.7455732934041904e-08, "loss": 0.0494, "step": 1991 }, { "epoch": 1.96, "learning_rate": 1.6525186969020833e-08, "loss": 0.0532, "step": 1992 }, { "epoch": 1.97, "learning_rate": 1.562010770326916e-08, "loss": 0.0522, "step": 1993 }, { "epoch": 1.97, "learning_rate": 1.474049744553896e-08, "loss": 0.0526, "step": 1994 }, { "epoch": 1.97, "learning_rate": 1.3886358439616498e-08, "loss": 0.0467, "step": 1995 }, { "epoch": 1.97, "learning_rate": 1.3057692864311134e-08, "loss": 0.0456, "step": 1996 }, { "epoch": 1.97, "learning_rate": 1.2254502833455307e-08, "loss": 0.0453, "step": 1997 }, { "epoch": 1.97, "learning_rate": 1.1476790395893444e-08, "loss": 0.0477, "step": 1998 }, { "epoch": 1.97, "learning_rate": 1.072455753548085e-08, "loss": 0.0531, "step": 1999 }, { "epoch": 1.97, "learning_rate": 9.99780617107815e-09, "loss": 0.0408, "step": 2000 }, { "epoch": 1.97, "learning_rate": 9.296538156543523e-09, "loss": 0.0435, "step": 2001 }, { "epoch": 1.97, "learning_rate": 8.620755280731585e-09, "loss": 0.0457, "step": 2002 }, { "epoch": 1.98, "learning_rate": 7.970459267487852e-09, "loss": 0.0565, "step": 2003 }, { "epoch": 1.98, "learning_rate": 7.3456517756420596e-09, "loss": 0.0435, "step": 2004 }, { "epoch": 1.98, "learning_rate": 6.74633439900596e-09, "loss": 0.0568, "step": 2005 }, { "epoch": 1.98, "learning_rate": 6.172508666371091e-09, "loss": 0.0488, "step": 2006 }, { "epoch": 1.98, "learning_rate": 5.6241760414987856e-09, "loss": 0.0513, "step": 2007 }, { "epoch": 1.98, "learning_rate": 5.101337923124616e-09, "loss": 0.05, "step": 2008 }, { "epoch": 1.98, "learning_rate": 4.6039956449484e-09, "loss": 0.0472, "step": 2009 }, { "epoch": 1.98, "learning_rate": 4.13215047563309e-09, "loss": 0.0516, "step": 2010 }, { "epoch": 1.98, "learning_rate": 3.685803618800332e-09, "loss": 0.0585, "step": 2011 }, { "epoch": 1.98, "learning_rate": 3.2649562130315782e-09, "loss": 0.0497, "step": 2012 }, { "epoch": 1.99, "learning_rate": 2.8696093318592023e-09, "loss": 0.0414, "step": 2013 }, { "epoch": 1.99, "learning_rate": 2.4997639837687217e-09, "loss": 0.0506, "step": 2014 }, { "epoch": 1.99, "learning_rate": 2.155421112191025e-09, "loss": 0.0473, "step": 2015 }, { "epoch": 1.99, "learning_rate": 1.8365815955079247e-09, "loss": 0.0438, "step": 2016 }, { "epoch": 1.99, "learning_rate": 1.5432462470421627e-09, "loss": 0.057, "step": 2017 }, { "epoch": 1.99, "learning_rate": 1.2754158150563023e-09, "loss": 0.0485, "step": 2018 }, { "epoch": 1.99, "learning_rate": 1.0330909827571677e-09, "loss": 0.0504, "step": 2019 }, { "epoch": 1.99, "learning_rate": 8.162723682869634e-10, "loss": 0.0508, "step": 2020 }, { "epoch": 1.99, "learning_rate": 6.24960524725493e-10, "loss": 0.0492, "step": 2021 }, { "epoch": 1.99, "learning_rate": 4.591559400857204e-10, "loss": 0.0514, "step": 2022 }, { "epoch": 2.0, "learning_rate": 3.188590373182088e-10, "loss": 0.0541, "step": 2023 }, { "epoch": 2.0, "learning_rate": 2.040701743033502e-10, "loss": 0.0499, "step": 2024 }, { "epoch": 2.0, "learning_rate": 1.1478964385469582e-10, "loss": 0.0524, "step": 2025 }, { "epoch": 2.0, "learning_rate": 5.1017673715625384e-11, "loss": 0.052, "step": 2026 }, { "epoch": 2.0, "learning_rate": 1.275442656267778e-11, "loss": 0.0483, "step": 2027 }, { "epoch": 2.0, "learning_rate": 0.0, "loss": 0.0503, "step": 2028 }, { "epoch": 2.0, "step": 2028, "total_flos": 251392139722752.0, "train_loss": 0.07178987883016673, "train_runtime": 7975.1458, "train_samples_per_second": 65.087, "train_steps_per_second": 0.254 } ], "max_steps": 2028, "num_train_epochs": 2, "total_flos": 251392139722752.0, "trial_name": null, "trial_params": null }