{ "best_metric": null, "best_model_checkpoint": null, "epoch": 5.998306519898391, "eval_steps": 500, "global_step": 3540, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 2.2727272727272729e-07, "loss": 4.6048, "step": 1 }, { "epoch": 0.0, "learning_rate": 4.5454545454545457e-07, "loss": 4.5919, "step": 2 }, { "epoch": 0.01, "learning_rate": 6.818181818181818e-07, "loss": 4.5966, "step": 3 }, { "epoch": 0.01, "learning_rate": 9.090909090909091e-07, "loss": 4.6159, "step": 4 }, { "epoch": 0.01, "learning_rate": 1.1363636363636364e-06, "loss": 4.5303, "step": 5 }, { "epoch": 0.01, "learning_rate": 1.3636363636363636e-06, "loss": 4.4178, "step": 6 }, { "epoch": 0.01, "learning_rate": 1.590909090909091e-06, "loss": 4.3337, "step": 7 }, { "epoch": 0.01, "learning_rate": 1.8181818181818183e-06, "loss": 3.8756, "step": 8 }, { "epoch": 0.02, "learning_rate": 2.0454545454545457e-06, "loss": 3.7636, "step": 9 }, { "epoch": 0.02, "learning_rate": 2.2727272727272728e-06, "loss": 3.0337, "step": 10 }, { "epoch": 0.02, "learning_rate": 2.5e-06, "loss": 2.9706, "step": 11 }, { "epoch": 0.02, "learning_rate": 2.7272727272727272e-06, "loss": 2.8538, "step": 12 }, { "epoch": 0.02, "learning_rate": 2.954545454545455e-06, "loss": 2.7651, "step": 13 }, { "epoch": 0.02, "learning_rate": 3.181818181818182e-06, "loss": 2.4846, "step": 14 }, { "epoch": 0.03, "learning_rate": 3.409090909090909e-06, "loss": 2.45, "step": 15 }, { "epoch": 0.03, "learning_rate": 3.6363636363636366e-06, "loss": 2.3373, "step": 16 }, { "epoch": 0.03, "learning_rate": 3.863636363636364e-06, "loss": 2.2242, "step": 17 }, { "epoch": 0.03, "learning_rate": 4.0909090909090915e-06, "loss": 2.0935, "step": 18 }, { "epoch": 0.03, "learning_rate": 4.3181818181818185e-06, "loss": 1.9616, "step": 19 }, { "epoch": 0.03, "learning_rate": 4.5454545454545455e-06, "loss": 1.8981, "step": 20 }, { "epoch": 0.04, "learning_rate": 4.772727272727273e-06, "loss": 1.8464, "step": 21 }, { "epoch": 0.04, "learning_rate": 5e-06, "loss": 1.8432, "step": 22 }, { "epoch": 0.04, "learning_rate": 5.2272727272727274e-06, "loss": 1.8214, "step": 23 }, { "epoch": 0.04, "learning_rate": 5.4545454545454545e-06, "loss": 1.8138, "step": 24 }, { "epoch": 0.04, "learning_rate": 5.681818181818183e-06, "loss": 1.7246, "step": 25 }, { "epoch": 0.04, "learning_rate": 5.90909090909091e-06, "loss": 1.6939, "step": 26 }, { "epoch": 0.05, "learning_rate": 6.136363636363637e-06, "loss": 1.6796, "step": 27 }, { "epoch": 0.05, "learning_rate": 6.363636363636364e-06, "loss": 1.6156, "step": 28 }, { "epoch": 0.05, "learning_rate": 6.590909090909091e-06, "loss": 1.6367, "step": 29 }, { "epoch": 0.05, "learning_rate": 6.818181818181818e-06, "loss": 1.5645, "step": 30 }, { "epoch": 0.05, "learning_rate": 7.045454545454546e-06, "loss": 1.5761, "step": 31 }, { "epoch": 0.05, "learning_rate": 7.272727272727273e-06, "loss": 1.6215, "step": 32 }, { "epoch": 0.06, "learning_rate": 7.500000000000001e-06, "loss": 1.5298, "step": 33 }, { "epoch": 0.06, "learning_rate": 7.727272727272727e-06, "loss": 1.5329, "step": 34 }, { "epoch": 0.06, "learning_rate": 7.954545454545455e-06, "loss": 1.5422, "step": 35 }, { "epoch": 0.06, "learning_rate": 8.181818181818183e-06, "loss": 1.5106, "step": 36 }, { "epoch": 0.06, "learning_rate": 8.40909090909091e-06, "loss": 1.4848, "step": 37 }, { "epoch": 0.07, "learning_rate": 8.636363636363637e-06, "loss": 1.4204, "step": 38 }, { "epoch": 0.07, "learning_rate": 8.863636363636365e-06, "loss": 1.4161, "step": 39 }, { "epoch": 0.07, "learning_rate": 9.090909090909091e-06, "loss": 1.4423, "step": 40 }, { "epoch": 0.07, "learning_rate": 9.318181818181819e-06, "loss": 1.4263, "step": 41 }, { "epoch": 0.07, "learning_rate": 9.545454545454547e-06, "loss": 1.4693, "step": 42 }, { "epoch": 0.07, "learning_rate": 9.772727272727273e-06, "loss": 1.4208, "step": 43 }, { "epoch": 0.08, "learning_rate": 1e-05, "loss": 1.3863, "step": 44 }, { "epoch": 0.08, "learning_rate": 1.0227272727272729e-05, "loss": 1.3988, "step": 45 }, { "epoch": 0.08, "learning_rate": 1.0454545454545455e-05, "loss": 1.3822, "step": 46 }, { "epoch": 0.08, "learning_rate": 1.0681818181818183e-05, "loss": 1.3656, "step": 47 }, { "epoch": 0.08, "learning_rate": 1.0909090909090909e-05, "loss": 1.3591, "step": 48 }, { "epoch": 0.08, "learning_rate": 1.1136363636363637e-05, "loss": 1.3512, "step": 49 }, { "epoch": 0.09, "learning_rate": 1.1363636363636366e-05, "loss": 1.3744, "step": 50 }, { "epoch": 0.09, "learning_rate": 1.1590909090909093e-05, "loss": 1.3765, "step": 51 }, { "epoch": 0.09, "learning_rate": 1.181818181818182e-05, "loss": 1.3294, "step": 52 }, { "epoch": 0.09, "learning_rate": 1.2045454545454547e-05, "loss": 1.3043, "step": 53 }, { "epoch": 0.09, "learning_rate": 1.2272727272727274e-05, "loss": 1.3017, "step": 54 }, { "epoch": 0.09, "learning_rate": 1.25e-05, "loss": 1.3303, "step": 55 }, { "epoch": 0.1, "learning_rate": 1.2727272727272728e-05, "loss": 1.3034, "step": 56 }, { "epoch": 0.1, "learning_rate": 1.2954545454545455e-05, "loss": 1.3379, "step": 57 }, { "epoch": 0.1, "learning_rate": 1.3181818181818183e-05, "loss": 1.2582, "step": 58 }, { "epoch": 0.1, "learning_rate": 1.3409090909090909e-05, "loss": 1.2925, "step": 59 }, { "epoch": 0.1, "learning_rate": 1.3636363636363637e-05, "loss": 1.2797, "step": 60 }, { "epoch": 0.1, "learning_rate": 1.3863636363636366e-05, "loss": 1.2774, "step": 61 }, { "epoch": 0.11, "learning_rate": 1.4090909090909092e-05, "loss": 1.2831, "step": 62 }, { "epoch": 0.11, "learning_rate": 1.431818181818182e-05, "loss": 1.2814, "step": 63 }, { "epoch": 0.11, "learning_rate": 1.4545454545454546e-05, "loss": 1.2209, "step": 64 }, { "epoch": 0.11, "learning_rate": 1.4772727272727274e-05, "loss": 1.2428, "step": 65 }, { "epoch": 0.11, "learning_rate": 1.5000000000000002e-05, "loss": 1.2764, "step": 66 }, { "epoch": 0.12, "learning_rate": 1.5227272727272728e-05, "loss": 1.2087, "step": 67 }, { "epoch": 0.12, "learning_rate": 1.5454545454545454e-05, "loss": 1.2932, "step": 68 }, { "epoch": 0.12, "learning_rate": 1.5681818181818182e-05, "loss": 1.275, "step": 69 }, { "epoch": 0.12, "learning_rate": 1.590909090909091e-05, "loss": 1.239, "step": 70 }, { "epoch": 0.12, "learning_rate": 1.6136363636363638e-05, "loss": 1.2077, "step": 71 }, { "epoch": 0.12, "learning_rate": 1.6363636363636366e-05, "loss": 1.2443, "step": 72 }, { "epoch": 0.13, "learning_rate": 1.6590909090909094e-05, "loss": 1.2476, "step": 73 }, { "epoch": 0.13, "learning_rate": 1.681818181818182e-05, "loss": 1.2344, "step": 74 }, { "epoch": 0.13, "learning_rate": 1.7045454545454546e-05, "loss": 1.2136, "step": 75 }, { "epoch": 0.13, "learning_rate": 1.7272727272727274e-05, "loss": 1.2069, "step": 76 }, { "epoch": 0.13, "learning_rate": 1.7500000000000002e-05, "loss": 1.1996, "step": 77 }, { "epoch": 0.13, "learning_rate": 1.772727272727273e-05, "loss": 1.1667, "step": 78 }, { "epoch": 0.14, "learning_rate": 1.7954545454545454e-05, "loss": 1.198, "step": 79 }, { "epoch": 0.14, "learning_rate": 1.8181818181818182e-05, "loss": 1.2032, "step": 80 }, { "epoch": 0.14, "learning_rate": 1.840909090909091e-05, "loss": 1.235, "step": 81 }, { "epoch": 0.14, "learning_rate": 1.8636363636363638e-05, "loss": 1.1846, "step": 82 }, { "epoch": 0.14, "learning_rate": 1.8863636363636366e-05, "loss": 1.2161, "step": 83 }, { "epoch": 0.14, "learning_rate": 1.9090909090909094e-05, "loss": 1.1832, "step": 84 }, { "epoch": 0.15, "learning_rate": 1.931818181818182e-05, "loss": 1.2051, "step": 85 }, { "epoch": 0.15, "learning_rate": 1.9545454545454546e-05, "loss": 1.2308, "step": 86 }, { "epoch": 0.15, "learning_rate": 1.9772727272727274e-05, "loss": 1.1997, "step": 87 }, { "epoch": 0.15, "learning_rate": 2e-05, "loss": 1.1871, "step": 88 }, { "epoch": 0.15, "learning_rate": 1.999999380336828e-05, "loss": 1.172, "step": 89 }, { "epoch": 0.15, "learning_rate": 1.9999975213480787e-05, "loss": 1.1942, "step": 90 }, { "epoch": 0.16, "learning_rate": 1.999994423036057e-05, "loss": 1.1457, "step": 91 }, { "epoch": 0.16, "learning_rate": 1.9999900854046022e-05, "loss": 1.1975, "step": 92 }, { "epoch": 0.16, "learning_rate": 1.99998450845909e-05, "loss": 1.1856, "step": 93 }, { "epoch": 0.16, "learning_rate": 1.9999776922064324e-05, "loss": 1.152, "step": 94 }, { "epoch": 0.16, "learning_rate": 1.9999696366550767e-05, "loss": 1.1885, "step": 95 }, { "epoch": 0.16, "learning_rate": 1.9999603418150065e-05, "loss": 1.1494, "step": 96 }, { "epoch": 0.17, "learning_rate": 1.9999498076977407e-05, "loss": 1.1624, "step": 97 }, { "epoch": 0.17, "learning_rate": 1.9999380343163356e-05, "loss": 1.1523, "step": 98 }, { "epoch": 0.17, "learning_rate": 1.999925021685381e-05, "loss": 1.1335, "step": 99 }, { "epoch": 0.17, "learning_rate": 1.9999107698210043e-05, "loss": 1.1572, "step": 100 }, { "epoch": 0.17, "learning_rate": 1.9998952787408687e-05, "loss": 1.1425, "step": 101 }, { "epoch": 0.18, "learning_rate": 1.999878548464172e-05, "loss": 1.1643, "step": 102 }, { "epoch": 0.18, "learning_rate": 1.9998605790116486e-05, "loss": 1.1443, "step": 103 }, { "epoch": 0.18, "learning_rate": 1.9998413704055687e-05, "loss": 1.1305, "step": 104 }, { "epoch": 0.18, "learning_rate": 1.999820922669738e-05, "loss": 1.1344, "step": 105 }, { "epoch": 0.18, "learning_rate": 1.9997992358294976e-05, "loss": 1.1462, "step": 106 }, { "epoch": 0.18, "learning_rate": 1.9997763099117253e-05, "loss": 1.1092, "step": 107 }, { "epoch": 0.19, "learning_rate": 1.999752144944833e-05, "loss": 1.1139, "step": 108 }, { "epoch": 0.19, "learning_rate": 1.9997267409587697e-05, "loss": 1.1312, "step": 109 }, { "epoch": 0.19, "learning_rate": 1.999700097985019e-05, "loss": 1.1119, "step": 110 }, { "epoch": 0.19, "learning_rate": 1.9996722160565996e-05, "loss": 1.1271, "step": 111 }, { "epoch": 0.19, "learning_rate": 1.999643095208067e-05, "loss": 1.1241, "step": 112 }, { "epoch": 0.19, "learning_rate": 1.9996127354755112e-05, "loss": 1.1291, "step": 113 }, { "epoch": 0.2, "learning_rate": 1.999581136896558e-05, "loss": 1.1357, "step": 114 }, { "epoch": 0.2, "learning_rate": 1.999548299510368e-05, "loss": 1.0925, "step": 115 }, { "epoch": 0.2, "learning_rate": 1.999514223357638e-05, "loss": 1.1301, "step": 116 }, { "epoch": 0.2, "learning_rate": 1.999478908480599e-05, "loss": 1.1286, "step": 117 }, { "epoch": 0.2, "learning_rate": 1.9994423549230175e-05, "loss": 1.1161, "step": 118 }, { "epoch": 0.2, "learning_rate": 1.9994045627301957e-05, "loss": 1.0664, "step": 119 }, { "epoch": 0.21, "learning_rate": 1.9993655319489703e-05, "loss": 1.0997, "step": 120 }, { "epoch": 0.21, "learning_rate": 1.9993252626277133e-05, "loss": 1.1149, "step": 121 }, { "epoch": 0.21, "learning_rate": 1.9992837548163315e-05, "loss": 1.1151, "step": 122 }, { "epoch": 0.21, "learning_rate": 1.9992410085662667e-05, "loss": 1.1308, "step": 123 }, { "epoch": 0.21, "learning_rate": 1.999197023930495e-05, "loss": 1.0972, "step": 124 }, { "epoch": 0.21, "learning_rate": 1.999151800963528e-05, "loss": 1.1152, "step": 125 }, { "epoch": 0.22, "learning_rate": 1.9991053397214122e-05, "loss": 1.1021, "step": 126 }, { "epoch": 0.22, "learning_rate": 1.9990576402617273e-05, "loss": 1.1191, "step": 127 }, { "epoch": 0.22, "learning_rate": 1.9990087026435893e-05, "loss": 1.108, "step": 128 }, { "epoch": 0.22, "learning_rate": 1.9989585269276475e-05, "loss": 1.0738, "step": 129 }, { "epoch": 0.22, "learning_rate": 1.998907113176086e-05, "loss": 1.0983, "step": 130 }, { "epoch": 0.22, "learning_rate": 1.9988544614526232e-05, "loss": 1.1008, "step": 131 }, { "epoch": 0.23, "learning_rate": 1.9988005718225117e-05, "loss": 1.0731, "step": 132 }, { "epoch": 0.23, "learning_rate": 1.9987454443525386e-05, "loss": 1.0558, "step": 133 }, { "epoch": 0.23, "learning_rate": 1.998689079111025e-05, "loss": 1.0913, "step": 134 }, { "epoch": 0.23, "learning_rate": 1.998631476167825e-05, "loss": 1.0972, "step": 135 }, { "epoch": 0.23, "learning_rate": 1.9985726355943284e-05, "loss": 1.1019, "step": 136 }, { "epoch": 0.24, "learning_rate": 1.9985125574634573e-05, "loss": 1.0786, "step": 137 }, { "epoch": 0.24, "learning_rate": 1.9984512418496683e-05, "loss": 1.0895, "step": 138 }, { "epoch": 0.24, "learning_rate": 1.9983886888289515e-05, "loss": 1.0853, "step": 139 }, { "epoch": 0.24, "learning_rate": 1.9983248984788304e-05, "loss": 1.1047, "step": 140 }, { "epoch": 0.24, "learning_rate": 1.998259870878362e-05, "loss": 1.0756, "step": 141 }, { "epoch": 0.24, "learning_rate": 1.9981936061081366e-05, "loss": 1.089, "step": 142 }, { "epoch": 0.25, "learning_rate": 1.9981261042502783e-05, "loss": 1.0643, "step": 143 }, { "epoch": 0.25, "learning_rate": 1.9980573653884437e-05, "loss": 1.091, "step": 144 }, { "epoch": 0.25, "learning_rate": 1.9979873896078226e-05, "loss": 1.1353, "step": 145 }, { "epoch": 0.25, "learning_rate": 1.9979161769951377e-05, "loss": 1.0779, "step": 146 }, { "epoch": 0.25, "learning_rate": 1.9978437276386454e-05, "loss": 1.067, "step": 147 }, { "epoch": 0.25, "learning_rate": 1.9977700416281332e-05, "loss": 1.0674, "step": 148 }, { "epoch": 0.26, "learning_rate": 1.9976951190549228e-05, "loss": 1.0541, "step": 149 }, { "epoch": 0.26, "learning_rate": 1.9976189600118673e-05, "loss": 1.0929, "step": 150 }, { "epoch": 0.26, "learning_rate": 1.9975415645933525e-05, "loss": 1.0849, "step": 151 }, { "epoch": 0.26, "learning_rate": 1.997462932895297e-05, "loss": 1.0758, "step": 152 }, { "epoch": 0.26, "learning_rate": 1.997383065015151e-05, "loss": 1.0684, "step": 153 }, { "epoch": 0.26, "learning_rate": 1.9973019610518965e-05, "loss": 1.0569, "step": 154 }, { "epoch": 0.27, "learning_rate": 1.9972196211060482e-05, "loss": 1.1028, "step": 155 }, { "epoch": 0.27, "learning_rate": 1.9971360452796523e-05, "loss": 1.0667, "step": 156 }, { "epoch": 0.27, "learning_rate": 1.997051233676286e-05, "loss": 1.089, "step": 157 }, { "epoch": 0.27, "learning_rate": 1.996965186401059e-05, "loss": 1.0783, "step": 158 }, { "epoch": 0.27, "learning_rate": 1.9968779035606118e-05, "loss": 1.0875, "step": 159 }, { "epoch": 0.27, "learning_rate": 1.996789385263116e-05, "loss": 1.0428, "step": 160 }, { "epoch": 0.28, "learning_rate": 1.9966996316182752e-05, "loss": 1.092, "step": 161 }, { "epoch": 0.28, "learning_rate": 1.996608642737323e-05, "loss": 1.0708, "step": 162 }, { "epoch": 0.28, "learning_rate": 1.996516418733025e-05, "loss": 1.0813, "step": 163 }, { "epoch": 0.28, "learning_rate": 1.9964229597196758e-05, "loss": 1.0473, "step": 164 }, { "epoch": 0.28, "learning_rate": 1.9963282658131024e-05, "loss": 1.07, "step": 165 }, { "epoch": 0.28, "learning_rate": 1.9962323371306616e-05, "loss": 1.0473, "step": 166 }, { "epoch": 0.29, "learning_rate": 1.9961351737912394e-05, "loss": 1.059, "step": 167 }, { "epoch": 0.29, "learning_rate": 1.9960367759152543e-05, "loss": 1.0558, "step": 168 }, { "epoch": 0.29, "learning_rate": 1.9959371436246516e-05, "loss": 1.0791, "step": 169 }, { "epoch": 0.29, "learning_rate": 1.9958362770429098e-05, "loss": 1.0501, "step": 170 }, { "epoch": 0.29, "learning_rate": 1.9957341762950346e-05, "loss": 1.0552, "step": 171 }, { "epoch": 0.3, "learning_rate": 1.9956308415075627e-05, "loss": 1.0285, "step": 172 }, { "epoch": 0.3, "learning_rate": 1.9955262728085592e-05, "loss": 1.0297, "step": 173 }, { "epoch": 0.3, "learning_rate": 1.995420470327619e-05, "loss": 1.0385, "step": 174 }, { "epoch": 0.3, "learning_rate": 1.9953134341958656e-05, "loss": 1.077, "step": 175 }, { "epoch": 0.3, "learning_rate": 1.9952051645459524e-05, "loss": 1.0436, "step": 176 }, { "epoch": 0.3, "learning_rate": 1.9950956615120604e-05, "loss": 1.0085, "step": 177 }, { "epoch": 0.31, "learning_rate": 1.9949849252298996e-05, "loss": 1.0689, "step": 178 }, { "epoch": 0.31, "learning_rate": 1.9948729558367084e-05, "loss": 1.0219, "step": 179 }, { "epoch": 0.31, "learning_rate": 1.9947597534712533e-05, "loss": 1.0664, "step": 180 }, { "epoch": 0.31, "learning_rate": 1.9946453182738296e-05, "loss": 1.0587, "step": 181 }, { "epoch": 0.31, "learning_rate": 1.994529650386259e-05, "loss": 1.0235, "step": 182 }, { "epoch": 0.31, "learning_rate": 1.994412749951892e-05, "loss": 1.0362, "step": 183 }, { "epoch": 0.32, "learning_rate": 1.9942946171156062e-05, "loss": 1.0384, "step": 184 }, { "epoch": 0.32, "learning_rate": 1.9941752520238076e-05, "loss": 1.0348, "step": 185 }, { "epoch": 0.32, "learning_rate": 1.994054654824428e-05, "loss": 1.0507, "step": 186 }, { "epoch": 0.32, "learning_rate": 1.9939328256669266e-05, "loss": 1.0269, "step": 187 }, { "epoch": 0.32, "learning_rate": 1.9938097647022895e-05, "loss": 1.0239, "step": 188 }, { "epoch": 0.32, "learning_rate": 1.993685472083029e-05, "loss": 0.9986, "step": 189 }, { "epoch": 0.33, "learning_rate": 1.993559947963185e-05, "loss": 1.0027, "step": 190 }, { "epoch": 0.33, "learning_rate": 1.9934331924983222e-05, "loss": 1.0489, "step": 191 }, { "epoch": 0.33, "learning_rate": 1.9933052058455325e-05, "loss": 1.0024, "step": 192 }, { "epoch": 0.33, "learning_rate": 1.9931759881634334e-05, "loss": 1.0453, "step": 193 }, { "epoch": 0.33, "learning_rate": 1.9930455396121665e-05, "loss": 1.0471, "step": 194 }, { "epoch": 0.33, "learning_rate": 1.9929138603534013e-05, "loss": 1.0357, "step": 195 }, { "epoch": 0.34, "learning_rate": 1.9927809505503307e-05, "loss": 1.049, "step": 196 }, { "epoch": 0.34, "learning_rate": 1.992646810367674e-05, "loss": 1.0105, "step": 197 }, { "epoch": 0.34, "learning_rate": 1.992511439971674e-05, "loss": 1.0599, "step": 198 }, { "epoch": 0.34, "learning_rate": 1.992374839530099e-05, "loss": 1.0403, "step": 199 }, { "epoch": 0.34, "learning_rate": 1.992237009212242e-05, "loss": 1.0761, "step": 200 }, { "epoch": 0.35, "learning_rate": 1.992097949188919e-05, "loss": 1.0113, "step": 201 }, { "epoch": 0.35, "learning_rate": 1.991957659632471e-05, "loss": 1.0153, "step": 202 }, { "epoch": 0.35, "learning_rate": 1.9918161407167628e-05, "loss": 1.0195, "step": 203 }, { "epoch": 0.35, "learning_rate": 1.9916733926171824e-05, "loss": 1.0144, "step": 204 }, { "epoch": 0.35, "learning_rate": 1.991529415510641e-05, "loss": 1.02, "step": 205 }, { "epoch": 0.35, "learning_rate": 1.9913842095755737e-05, "loss": 1.0045, "step": 206 }, { "epoch": 0.36, "learning_rate": 1.9912377749919373e-05, "loss": 0.9759, "step": 207 }, { "epoch": 0.36, "learning_rate": 1.9910901119412132e-05, "loss": 1.0206, "step": 208 }, { "epoch": 0.36, "learning_rate": 1.990941220606403e-05, "loss": 1.0314, "step": 209 }, { "epoch": 0.36, "learning_rate": 1.9907911011720323e-05, "loss": 1.0273, "step": 210 }, { "epoch": 0.36, "learning_rate": 1.9906397538241474e-05, "loss": 1.0294, "step": 211 }, { "epoch": 0.36, "learning_rate": 1.990487178750318e-05, "loss": 1.0122, "step": 212 }, { "epoch": 0.37, "learning_rate": 1.990333376139634e-05, "loss": 0.9883, "step": 213 }, { "epoch": 0.37, "learning_rate": 1.9901783461827066e-05, "loss": 1.0024, "step": 214 }, { "epoch": 0.37, "learning_rate": 1.990022089071669e-05, "loss": 1.0126, "step": 215 }, { "epoch": 0.37, "learning_rate": 1.989864605000175e-05, "loss": 1.0098, "step": 216 }, { "epoch": 0.37, "learning_rate": 1.989705894163398e-05, "loss": 1.0458, "step": 217 }, { "epoch": 0.37, "learning_rate": 1.9895459567580326e-05, "loss": 1.0036, "step": 218 }, { "epoch": 0.38, "learning_rate": 1.989384792982294e-05, "loss": 1.0005, "step": 219 }, { "epoch": 0.38, "learning_rate": 1.9892224030359167e-05, "loss": 1.0349, "step": 220 }, { "epoch": 0.38, "learning_rate": 1.989058787120154e-05, "loss": 0.9846, "step": 221 }, { "epoch": 0.38, "learning_rate": 1.9888939454377806e-05, "loss": 0.9921, "step": 222 }, { "epoch": 0.38, "learning_rate": 1.988727878193088e-05, "loss": 1.0009, "step": 223 }, { "epoch": 0.38, "learning_rate": 1.9885605855918887e-05, "loss": 1.03, "step": 224 }, { "epoch": 0.39, "learning_rate": 1.9883920678415124e-05, "loss": 1.0012, "step": 225 }, { "epoch": 0.39, "learning_rate": 1.9882223251508074e-05, "loss": 0.9785, "step": 226 }, { "epoch": 0.39, "learning_rate": 1.9880513577301403e-05, "loss": 0.9956, "step": 227 }, { "epoch": 0.39, "learning_rate": 1.987879165791396e-05, "loss": 0.9917, "step": 228 }, { "epoch": 0.39, "learning_rate": 1.9877057495479758e-05, "loss": 1.0047, "step": 229 }, { "epoch": 0.39, "learning_rate": 1.9875311092148e-05, "loss": 0.9834, "step": 230 }, { "epoch": 0.4, "learning_rate": 1.9873552450083036e-05, "loss": 0.9924, "step": 231 }, { "epoch": 0.4, "learning_rate": 1.987178157146441e-05, "loss": 1.0282, "step": 232 }, { "epoch": 0.4, "learning_rate": 1.986999845848681e-05, "loss": 1.0272, "step": 233 }, { "epoch": 0.4, "learning_rate": 1.9868203113360102e-05, "loss": 0.9747, "step": 234 }, { "epoch": 0.4, "learning_rate": 1.9866395538309302e-05, "loss": 0.9953, "step": 235 }, { "epoch": 0.41, "learning_rate": 1.9864575735574583e-05, "loss": 0.9715, "step": 236 }, { "epoch": 0.41, "learning_rate": 1.9862743707411275e-05, "loss": 1.0198, "step": 237 }, { "epoch": 0.41, "learning_rate": 1.9860899456089863e-05, "loss": 1.0128, "step": 238 }, { "epoch": 0.41, "learning_rate": 1.9859042983895967e-05, "loss": 0.9861, "step": 239 }, { "epoch": 0.41, "learning_rate": 1.9857174293130374e-05, "loss": 0.9902, "step": 240 }, { "epoch": 0.41, "learning_rate": 1.9855293386108995e-05, "loss": 0.9766, "step": 241 }, { "epoch": 0.42, "learning_rate": 1.9853400265162884e-05, "loss": 1.004, "step": 242 }, { "epoch": 0.42, "learning_rate": 1.9851494932638246e-05, "loss": 1.0048, "step": 243 }, { "epoch": 0.42, "learning_rate": 1.98495773908964e-05, "loss": 0.979, "step": 244 }, { "epoch": 0.42, "learning_rate": 1.9847647642313806e-05, "loss": 1.0031, "step": 245 }, { "epoch": 0.42, "learning_rate": 1.984570568928206e-05, "loss": 1.0074, "step": 246 }, { "epoch": 0.42, "learning_rate": 1.9843751534207868e-05, "loss": 0.9804, "step": 247 }, { "epoch": 0.43, "learning_rate": 1.9841785179513074e-05, "loss": 0.9842, "step": 248 }, { "epoch": 0.43, "learning_rate": 1.9839806627634628e-05, "loss": 1.0484, "step": 249 }, { "epoch": 0.43, "learning_rate": 1.98378158810246e-05, "loss": 0.9898, "step": 250 }, { "epoch": 0.43, "learning_rate": 1.983581294215018e-05, "loss": 1.0107, "step": 251 }, { "epoch": 0.43, "learning_rate": 1.9833797813493656e-05, "loss": 0.984, "step": 252 }, { "epoch": 0.43, "learning_rate": 1.9831770497552432e-05, "loss": 0.9865, "step": 253 }, { "epoch": 0.44, "learning_rate": 1.982973099683902e-05, "loss": 1.0329, "step": 254 }, { "epoch": 0.44, "learning_rate": 1.982767931388102e-05, "loss": 1.0019, "step": 255 }, { "epoch": 0.44, "learning_rate": 1.9825615451221142e-05, "loss": 1.0093, "step": 256 }, { "epoch": 0.44, "learning_rate": 1.9823539411417178e-05, "loss": 0.9765, "step": 257 }, { "epoch": 0.44, "learning_rate": 1.9821451197042028e-05, "loss": 1.0192, "step": 258 }, { "epoch": 0.44, "learning_rate": 1.9819350810683665e-05, "loss": 0.9923, "step": 259 }, { "epoch": 0.45, "learning_rate": 1.9817238254945158e-05, "loss": 0.9883, "step": 260 }, { "epoch": 0.45, "learning_rate": 1.9815113532444645e-05, "loss": 1.0008, "step": 261 }, { "epoch": 0.45, "learning_rate": 1.9812976645815357e-05, "loss": 0.9824, "step": 262 }, { "epoch": 0.45, "learning_rate": 1.9810827597705595e-05, "loss": 1.0164, "step": 263 }, { "epoch": 0.45, "learning_rate": 1.9808666390778724e-05, "loss": 0.9878, "step": 264 }, { "epoch": 0.45, "learning_rate": 1.980649302771319e-05, "loss": 0.9994, "step": 265 }, { "epoch": 0.46, "learning_rate": 1.98043075112025e-05, "loss": 0.9953, "step": 266 }, { "epoch": 0.46, "learning_rate": 1.980210984395522e-05, "loss": 1.0099, "step": 267 }, { "epoch": 0.46, "learning_rate": 1.9799900028694972e-05, "loss": 0.9745, "step": 268 }, { "epoch": 0.46, "learning_rate": 1.9797678068160448e-05, "loss": 0.9838, "step": 269 }, { "epoch": 0.46, "learning_rate": 1.9795443965105377e-05, "loss": 0.9794, "step": 270 }, { "epoch": 0.47, "learning_rate": 1.9793197722298544e-05, "loss": 0.9967, "step": 271 }, { "epoch": 0.47, "learning_rate": 1.9790939342523773e-05, "loss": 0.9661, "step": 272 }, { "epoch": 0.47, "learning_rate": 1.9788668828579936e-05, "loss": 1.0142, "step": 273 }, { "epoch": 0.47, "learning_rate": 1.9786386183280937e-05, "loss": 0.988, "step": 274 }, { "epoch": 0.47, "learning_rate": 1.9784091409455728e-05, "loss": 0.994, "step": 275 }, { "epoch": 0.47, "learning_rate": 1.9781784509948275e-05, "loss": 0.9771, "step": 276 }, { "epoch": 0.48, "learning_rate": 1.9779465487617583e-05, "loss": 0.9586, "step": 277 }, { "epoch": 0.48, "learning_rate": 1.977713434533767e-05, "loss": 0.9943, "step": 278 }, { "epoch": 0.48, "learning_rate": 1.9774791085997593e-05, "loss": 0.9671, "step": 279 }, { "epoch": 0.48, "learning_rate": 1.9772435712501404e-05, "loss": 0.9905, "step": 280 }, { "epoch": 0.48, "learning_rate": 1.9770068227768188e-05, "loss": 0.9555, "step": 281 }, { "epoch": 0.48, "learning_rate": 1.9767688634732028e-05, "loss": 0.9612, "step": 282 }, { "epoch": 0.49, "learning_rate": 1.976529693634202e-05, "loss": 0.9908, "step": 283 }, { "epoch": 0.49, "learning_rate": 1.976289313556225e-05, "loss": 1.0064, "step": 284 }, { "epoch": 0.49, "learning_rate": 1.9760477235371817e-05, "loss": 0.9765, "step": 285 }, { "epoch": 0.49, "learning_rate": 1.9758049238764815e-05, "loss": 1.0056, "step": 286 }, { "epoch": 0.49, "learning_rate": 1.975560914875031e-05, "loss": 0.9608, "step": 287 }, { "epoch": 0.49, "learning_rate": 1.9753156968352388e-05, "loss": 0.9891, "step": 288 }, { "epoch": 0.5, "learning_rate": 1.9750692700610087e-05, "loss": 0.9758, "step": 289 }, { "epoch": 0.5, "learning_rate": 1.974821634857744e-05, "loss": 0.9417, "step": 290 }, { "epoch": 0.5, "learning_rate": 1.9745727915323468e-05, "loss": 0.9905, "step": 291 }, { "epoch": 0.5, "learning_rate": 1.9743227403932135e-05, "loss": 0.9628, "step": 292 }, { "epoch": 0.5, "learning_rate": 1.97407148175024e-05, "loss": 0.9732, "step": 293 }, { "epoch": 0.5, "learning_rate": 1.973819015914818e-05, "loss": 0.9874, "step": 294 }, { "epoch": 0.51, "learning_rate": 1.973565343199834e-05, "loss": 1.009, "step": 295 }, { "epoch": 0.51, "learning_rate": 1.9733104639196724e-05, "loss": 0.9348, "step": 296 }, { "epoch": 0.51, "learning_rate": 1.973054378390211e-05, "loss": 0.9507, "step": 297 }, { "epoch": 0.51, "learning_rate": 1.972797086928824e-05, "loss": 0.9439, "step": 298 }, { "epoch": 0.51, "learning_rate": 1.9725385898543784e-05, "loss": 0.9715, "step": 299 }, { "epoch": 0.52, "learning_rate": 1.972278887487238e-05, "loss": 0.9656, "step": 300 }, { "epoch": 0.52, "learning_rate": 1.9720179801492577e-05, "loss": 0.9835, "step": 301 }, { "epoch": 0.52, "learning_rate": 1.9717558681637873e-05, "loss": 0.9308, "step": 302 }, { "epoch": 0.52, "learning_rate": 1.9714925518556686e-05, "loss": 0.9935, "step": 303 }, { "epoch": 0.52, "learning_rate": 1.971228031551237e-05, "loss": 0.9778, "step": 304 }, { "epoch": 0.52, "learning_rate": 1.970962307578319e-05, "loss": 0.9556, "step": 305 }, { "epoch": 0.53, "learning_rate": 1.9706953802662333e-05, "loss": 0.9522, "step": 306 }, { "epoch": 0.53, "learning_rate": 1.9704272499457908e-05, "loss": 0.945, "step": 307 }, { "epoch": 0.53, "learning_rate": 1.9701579169492916e-05, "loss": 0.9564, "step": 308 }, { "epoch": 0.53, "learning_rate": 1.9698873816105272e-05, "loss": 1.013, "step": 309 }, { "epoch": 0.53, "learning_rate": 1.96961564426478e-05, "loss": 0.9664, "step": 310 }, { "epoch": 0.53, "learning_rate": 1.96934270524882e-05, "loss": 0.9664, "step": 311 }, { "epoch": 0.54, "learning_rate": 1.969068564900909e-05, "loss": 0.9607, "step": 312 }, { "epoch": 0.54, "learning_rate": 1.9687932235607953e-05, "loss": 0.9337, "step": 313 }, { "epoch": 0.54, "learning_rate": 1.968516681569717e-05, "loss": 0.9565, "step": 314 }, { "epoch": 0.54, "learning_rate": 1.9682389392704003e-05, "loss": 0.9398, "step": 315 }, { "epoch": 0.54, "learning_rate": 1.967959997007058e-05, "loss": 0.9594, "step": 316 }, { "epoch": 0.54, "learning_rate": 1.967679855125391e-05, "loss": 0.9766, "step": 317 }, { "epoch": 0.55, "learning_rate": 1.9673985139725864e-05, "loss": 0.9708, "step": 318 }, { "epoch": 0.55, "learning_rate": 1.9671159738973177e-05, "loss": 0.9469, "step": 319 }, { "epoch": 0.55, "learning_rate": 1.966832235249744e-05, "loss": 0.9715, "step": 320 }, { "epoch": 0.55, "learning_rate": 1.96654729838151e-05, "loss": 0.974, "step": 321 }, { "epoch": 0.55, "learning_rate": 1.9662611636457464e-05, "loss": 0.9593, "step": 322 }, { "epoch": 0.55, "learning_rate": 1.9659738313970663e-05, "loss": 0.9798, "step": 323 }, { "epoch": 0.56, "learning_rate": 1.9656853019915692e-05, "loss": 0.9539, "step": 324 }, { "epoch": 0.56, "learning_rate": 1.9653955757868363e-05, "loss": 0.9854, "step": 325 }, { "epoch": 0.56, "learning_rate": 1.9651046531419335e-05, "loss": 0.9724, "step": 326 }, { "epoch": 0.56, "learning_rate": 1.9648125344174085e-05, "loss": 0.9528, "step": 327 }, { "epoch": 0.56, "learning_rate": 1.964519219975292e-05, "loss": 0.9636, "step": 328 }, { "epoch": 0.56, "learning_rate": 1.9642247101790966e-05, "loss": 0.9664, "step": 329 }, { "epoch": 0.57, "learning_rate": 1.9639290053938153e-05, "loss": 0.9392, "step": 330 }, { "epoch": 0.57, "learning_rate": 1.9636321059859233e-05, "loss": 0.9514, "step": 331 }, { "epoch": 0.57, "learning_rate": 1.963334012323376e-05, "loss": 0.9516, "step": 332 }, { "epoch": 0.57, "learning_rate": 1.9630347247756083e-05, "loss": 0.9286, "step": 333 }, { "epoch": 0.57, "learning_rate": 1.9627342437135356e-05, "loss": 0.9668, "step": 334 }, { "epoch": 0.58, "learning_rate": 1.9624325695095514e-05, "loss": 0.9675, "step": 335 }, { "epoch": 0.58, "learning_rate": 1.962129702537529e-05, "loss": 0.9361, "step": 336 }, { "epoch": 0.58, "learning_rate": 1.961825643172819e-05, "loss": 0.9571, "step": 337 }, { "epoch": 0.58, "learning_rate": 1.961520391792251e-05, "loss": 0.9581, "step": 338 }, { "epoch": 0.58, "learning_rate": 1.96121394877413e-05, "loss": 0.9301, "step": 339 }, { "epoch": 0.58, "learning_rate": 1.9609063144982398e-05, "loss": 0.9019, "step": 340 }, { "epoch": 0.59, "learning_rate": 1.960597489345839e-05, "loss": 0.9677, "step": 341 }, { "epoch": 0.59, "learning_rate": 1.9602874736996632e-05, "loss": 0.9331, "step": 342 }, { "epoch": 0.59, "learning_rate": 1.959976267943923e-05, "loss": 0.9571, "step": 343 }, { "epoch": 0.59, "learning_rate": 1.9596638724643034e-05, "loss": 0.9443, "step": 344 }, { "epoch": 0.59, "learning_rate": 1.9593502876479648e-05, "loss": 0.9775, "step": 345 }, { "epoch": 0.59, "learning_rate": 1.959035513883541e-05, "loss": 0.9507, "step": 346 }, { "epoch": 0.6, "learning_rate": 1.958719551561139e-05, "loss": 0.9468, "step": 347 }, { "epoch": 0.6, "learning_rate": 1.95840240107234e-05, "loss": 0.9263, "step": 348 }, { "epoch": 0.6, "learning_rate": 1.958084062810196e-05, "loss": 0.929, "step": 349 }, { "epoch": 0.6, "learning_rate": 1.957764537169233e-05, "loss": 0.9359, "step": 350 }, { "epoch": 0.6, "learning_rate": 1.9574438245454472e-05, "loss": 0.9269, "step": 351 }, { "epoch": 0.6, "learning_rate": 1.9571219253363057e-05, "loss": 0.9481, "step": 352 }, { "epoch": 0.61, "learning_rate": 1.9567988399407473e-05, "loss": 0.9591, "step": 353 }, { "epoch": 0.61, "learning_rate": 1.9564745687591804e-05, "loss": 0.9274, "step": 354 }, { "epoch": 0.61, "learning_rate": 1.956149112193482e-05, "loss": 0.9404, "step": 355 }, { "epoch": 0.61, "learning_rate": 1.9558224706469996e-05, "loss": 0.9704, "step": 356 }, { "epoch": 0.61, "learning_rate": 1.9554946445245485e-05, "loss": 0.9475, "step": 357 }, { "epoch": 0.61, "learning_rate": 1.9551656342324118e-05, "loss": 0.8951, "step": 358 }, { "epoch": 0.62, "learning_rate": 1.9548354401783416e-05, "loss": 0.9275, "step": 359 }, { "epoch": 0.62, "learning_rate": 1.9545040627715554e-05, "loss": 0.9465, "step": 360 }, { "epoch": 0.62, "learning_rate": 1.9541715024227382e-05, "loss": 0.9232, "step": 361 }, { "epoch": 0.62, "learning_rate": 1.9538377595440404e-05, "loss": 0.9269, "step": 362 }, { "epoch": 0.62, "learning_rate": 1.9535028345490792e-05, "loss": 0.9368, "step": 363 }, { "epoch": 0.62, "learning_rate": 1.953166727852935e-05, "loss": 0.9263, "step": 364 }, { "epoch": 0.63, "learning_rate": 1.9528294398721543e-05, "loss": 0.9272, "step": 365 }, { "epoch": 0.63, "learning_rate": 1.9524909710247465e-05, "loss": 0.9601, "step": 366 }, { "epoch": 0.63, "learning_rate": 1.952151321730186e-05, "loss": 0.9379, "step": 367 }, { "epoch": 0.63, "learning_rate": 1.9518104924094077e-05, "loss": 0.9169, "step": 368 }, { "epoch": 0.63, "learning_rate": 1.951468483484811e-05, "loss": 0.9491, "step": 369 }, { "epoch": 0.64, "learning_rate": 1.9511252953802568e-05, "loss": 0.927, "step": 370 }, { "epoch": 0.64, "learning_rate": 1.9507809285210672e-05, "loss": 0.8988, "step": 371 }, { "epoch": 0.64, "learning_rate": 1.9504353833340244e-05, "loss": 0.9542, "step": 372 }, { "epoch": 0.64, "learning_rate": 1.9500886602473725e-05, "loss": 0.9245, "step": 373 }, { "epoch": 0.64, "learning_rate": 1.949740759690814e-05, "loss": 0.9331, "step": 374 }, { "epoch": 0.64, "learning_rate": 1.9493916820955114e-05, "loss": 0.9075, "step": 375 }, { "epoch": 0.65, "learning_rate": 1.949041427894086e-05, "loss": 0.945, "step": 376 }, { "epoch": 0.65, "learning_rate": 1.9486899975206165e-05, "loss": 0.9509, "step": 377 }, { "epoch": 0.65, "learning_rate": 1.9483373914106407e-05, "loss": 0.9151, "step": 378 }, { "epoch": 0.65, "learning_rate": 1.9479836100011512e-05, "loss": 0.9532, "step": 379 }, { "epoch": 0.65, "learning_rate": 1.9476286537306002e-05, "loss": 0.9389, "step": 380 }, { "epoch": 0.65, "learning_rate": 1.9472725230388936e-05, "loss": 0.9205, "step": 381 }, { "epoch": 0.66, "learning_rate": 1.9469152183673936e-05, "loss": 0.9243, "step": 382 }, { "epoch": 0.66, "learning_rate": 1.946556740158917e-05, "loss": 0.9117, "step": 383 }, { "epoch": 0.66, "learning_rate": 1.946197088857736e-05, "loss": 0.924, "step": 384 }, { "epoch": 0.66, "learning_rate": 1.9458362649095753e-05, "loss": 0.9271, "step": 385 }, { "epoch": 0.66, "learning_rate": 1.945474268761614e-05, "loss": 0.9379, "step": 386 }, { "epoch": 0.66, "learning_rate": 1.945111100862483e-05, "loss": 0.9613, "step": 387 }, { "epoch": 0.67, "learning_rate": 1.944746761662266e-05, "loss": 0.9112, "step": 388 }, { "epoch": 0.67, "learning_rate": 1.9443812516124986e-05, "loss": 0.9269, "step": 389 }, { "epoch": 0.67, "learning_rate": 1.9440145711661663e-05, "loss": 0.9067, "step": 390 }, { "epoch": 0.67, "learning_rate": 1.9436467207777062e-05, "loss": 0.9117, "step": 391 }, { "epoch": 0.67, "learning_rate": 1.9432777009030054e-05, "loss": 0.9131, "step": 392 }, { "epoch": 0.67, "learning_rate": 1.9429075119993992e-05, "loss": 0.9031, "step": 393 }, { "epoch": 0.68, "learning_rate": 1.942536154525673e-05, "loss": 0.9055, "step": 394 }, { "epoch": 0.68, "learning_rate": 1.9421636289420593e-05, "loss": 0.9139, "step": 395 }, { "epoch": 0.68, "learning_rate": 1.9417899357102395e-05, "loss": 0.932, "step": 396 }, { "epoch": 0.68, "learning_rate": 1.9414150752933416e-05, "loss": 0.9122, "step": 397 }, { "epoch": 0.68, "learning_rate": 1.941039048155939e-05, "loss": 0.9271, "step": 398 }, { "epoch": 0.68, "learning_rate": 1.940661854764053e-05, "loss": 0.8919, "step": 399 }, { "epoch": 0.69, "learning_rate": 1.940283495585149e-05, "loss": 0.9208, "step": 400 }, { "epoch": 0.69, "learning_rate": 1.9399039710881375e-05, "loss": 0.9092, "step": 401 }, { "epoch": 0.69, "learning_rate": 1.9395232817433728e-05, "loss": 0.9464, "step": 402 }, { "epoch": 0.69, "learning_rate": 1.939141428022654e-05, "loss": 0.9274, "step": 403 }, { "epoch": 0.69, "learning_rate": 1.9387584103992217e-05, "loss": 0.9004, "step": 404 }, { "epoch": 0.7, "learning_rate": 1.9383742293477604e-05, "loss": 0.9227, "step": 405 }, { "epoch": 0.7, "learning_rate": 1.9379888853443954e-05, "loss": 0.9133, "step": 406 }, { "epoch": 0.7, "learning_rate": 1.937602378866694e-05, "loss": 0.9027, "step": 407 }, { "epoch": 0.7, "learning_rate": 1.9372147103936637e-05, "loss": 0.9422, "step": 408 }, { "epoch": 0.7, "learning_rate": 1.936825880405752e-05, "loss": 0.9251, "step": 409 }, { "epoch": 0.7, "learning_rate": 1.9364358893848464e-05, "loss": 0.9101, "step": 410 }, { "epoch": 0.71, "learning_rate": 1.936044737814273e-05, "loss": 0.9156, "step": 411 }, { "epoch": 0.71, "learning_rate": 1.9356524261787962e-05, "loss": 0.9086, "step": 412 }, { "epoch": 0.71, "learning_rate": 1.9352589549646183e-05, "loss": 0.9144, "step": 413 }, { "epoch": 0.71, "learning_rate": 1.9348643246593782e-05, "loss": 0.9389, "step": 414 }, { "epoch": 0.71, "learning_rate": 1.934468535752152e-05, "loss": 0.8898, "step": 415 }, { "epoch": 0.71, "learning_rate": 1.934071588733451e-05, "loss": 0.9069, "step": 416 }, { "epoch": 0.72, "learning_rate": 1.9336734840952223e-05, "loss": 0.8943, "step": 417 }, { "epoch": 0.72, "learning_rate": 1.9332742223308477e-05, "loss": 0.8981, "step": 418 }, { "epoch": 0.72, "learning_rate": 1.9328738039351427e-05, "loss": 0.908, "step": 419 }, { "epoch": 0.72, "learning_rate": 1.932472229404356e-05, "loss": 0.9237, "step": 420 }, { "epoch": 0.72, "learning_rate": 1.9320694992361697e-05, "loss": 0.9051, "step": 421 }, { "epoch": 0.72, "learning_rate": 1.931665613929698e-05, "loss": 0.8846, "step": 422 }, { "epoch": 0.73, "learning_rate": 1.9312605739854862e-05, "loss": 0.9173, "step": 423 }, { "epoch": 0.73, "learning_rate": 1.9308543799055118e-05, "loss": 0.8901, "step": 424 }, { "epoch": 0.73, "learning_rate": 1.930447032193181e-05, "loss": 0.9142, "step": 425 }, { "epoch": 0.73, "learning_rate": 1.9300385313533313e-05, "loss": 0.9013, "step": 426 }, { "epoch": 0.73, "learning_rate": 1.929628877892228e-05, "loss": 0.9155, "step": 427 }, { "epoch": 0.73, "learning_rate": 1.9292180723175656e-05, "loss": 0.9239, "step": 428 }, { "epoch": 0.74, "learning_rate": 1.9288061151384658e-05, "loss": 0.9041, "step": 429 }, { "epoch": 0.74, "learning_rate": 1.9283930068654787e-05, "loss": 0.9249, "step": 430 }, { "epoch": 0.74, "learning_rate": 1.9279787480105802e-05, "loss": 0.8781, "step": 431 }, { "epoch": 0.74, "learning_rate": 1.9275633390871718e-05, "loss": 0.8893, "step": 432 }, { "epoch": 0.74, "learning_rate": 1.927146780610081e-05, "loss": 0.9183, "step": 433 }, { "epoch": 0.75, "learning_rate": 1.9267290730955593e-05, "loss": 0.8787, "step": 434 }, { "epoch": 0.75, "learning_rate": 1.9263102170612833e-05, "loss": 0.9265, "step": 435 }, { "epoch": 0.75, "learning_rate": 1.9258902130263518e-05, "loss": 0.8827, "step": 436 }, { "epoch": 0.75, "learning_rate": 1.925469061511287e-05, "loss": 0.878, "step": 437 }, { "epoch": 0.75, "learning_rate": 1.9250467630380335e-05, "loss": 0.8852, "step": 438 }, { "epoch": 0.75, "learning_rate": 1.924623318129956e-05, "loss": 0.886, "step": 439 }, { "epoch": 0.76, "learning_rate": 1.9241987273118416e-05, "loss": 0.9026, "step": 440 }, { "epoch": 0.76, "learning_rate": 1.9237729911098972e-05, "loss": 0.8944, "step": 441 }, { "epoch": 0.76, "learning_rate": 1.923346110051748e-05, "loss": 0.8925, "step": 442 }, { "epoch": 0.76, "learning_rate": 1.922918084666439e-05, "loss": 0.894, "step": 443 }, { "epoch": 0.76, "learning_rate": 1.922488915484434e-05, "loss": 0.8986, "step": 444 }, { "epoch": 0.76, "learning_rate": 1.9220586030376135e-05, "loss": 0.8967, "step": 445 }, { "epoch": 0.77, "learning_rate": 1.9216271478592753e-05, "loss": 0.8735, "step": 446 }, { "epoch": 0.77, "learning_rate": 1.9211945504841324e-05, "loss": 0.9223, "step": 447 }, { "epoch": 0.77, "learning_rate": 1.9207608114483144e-05, "loss": 0.9239, "step": 448 }, { "epoch": 0.77, "learning_rate": 1.920325931289366e-05, "loss": 0.8759, "step": 449 }, { "epoch": 0.77, "learning_rate": 1.9198899105462447e-05, "loss": 0.8834, "step": 450 }, { "epoch": 0.77, "learning_rate": 1.9194527497593233e-05, "loss": 0.9048, "step": 451 }, { "epoch": 0.78, "learning_rate": 1.9190144494703865e-05, "loss": 0.8962, "step": 452 }, { "epoch": 0.78, "learning_rate": 1.9185750102226318e-05, "loss": 0.8893, "step": 453 }, { "epoch": 0.78, "learning_rate": 1.9181344325606667e-05, "loss": 0.8866, "step": 454 }, { "epoch": 0.78, "learning_rate": 1.917692717030512e-05, "loss": 0.8964, "step": 455 }, { "epoch": 0.78, "learning_rate": 1.9172498641795964e-05, "loss": 0.876, "step": 456 }, { "epoch": 0.78, "learning_rate": 1.916805874556759e-05, "loss": 0.9194, "step": 457 }, { "epoch": 0.79, "learning_rate": 1.916360748712249e-05, "loss": 0.896, "step": 458 }, { "epoch": 0.79, "learning_rate": 1.915914487197722e-05, "loss": 0.8943, "step": 459 }, { "epoch": 0.79, "learning_rate": 1.9154670905662412e-05, "loss": 0.8686, "step": 460 }, { "epoch": 0.79, "learning_rate": 1.9150185593722778e-05, "loss": 0.8699, "step": 461 }, { "epoch": 0.79, "learning_rate": 1.9145688941717074e-05, "loss": 0.8842, "step": 462 }, { "epoch": 0.79, "learning_rate": 1.9141180955218127e-05, "loss": 0.8861, "step": 463 }, { "epoch": 0.8, "learning_rate": 1.91366616398128e-05, "loss": 0.917, "step": 464 }, { "epoch": 0.8, "learning_rate": 1.9132131001102e-05, "loss": 0.8724, "step": 465 }, { "epoch": 0.8, "learning_rate": 1.912758904470067e-05, "loss": 0.9378, "step": 466 }, { "epoch": 0.8, "learning_rate": 1.912303577623777e-05, "loss": 0.8852, "step": 467 }, { "epoch": 0.8, "learning_rate": 1.9118471201356293e-05, "loss": 0.9224, "step": 468 }, { "epoch": 0.81, "learning_rate": 1.911389532571323e-05, "loss": 0.876, "step": 469 }, { "epoch": 0.81, "learning_rate": 1.9109308154979588e-05, "loss": 0.8958, "step": 470 }, { "epoch": 0.81, "learning_rate": 1.910470969484037e-05, "loss": 0.8742, "step": 471 }, { "epoch": 0.81, "learning_rate": 1.9100099950994564e-05, "loss": 0.8821, "step": 472 }, { "epoch": 0.81, "learning_rate": 1.909547892915515e-05, "loss": 0.884, "step": 473 }, { "epoch": 0.81, "learning_rate": 1.9090846635049084e-05, "loss": 0.8621, "step": 474 }, { "epoch": 0.82, "learning_rate": 1.9086203074417288e-05, "loss": 0.9075, "step": 475 }, { "epoch": 0.82, "learning_rate": 1.9081548253014645e-05, "loss": 0.9102, "step": 476 }, { "epoch": 0.82, "learning_rate": 1.9076882176610003e-05, "loss": 0.9048, "step": 477 }, { "epoch": 0.82, "learning_rate": 1.9072204850986153e-05, "loss": 0.884, "step": 478 }, { "epoch": 0.82, "learning_rate": 1.9067516281939826e-05, "loss": 0.8995, "step": 479 }, { "epoch": 0.82, "learning_rate": 1.9062816475281693e-05, "loss": 0.8425, "step": 480 }, { "epoch": 0.83, "learning_rate": 1.9058105436836345e-05, "loss": 0.8835, "step": 481 }, { "epoch": 0.83, "learning_rate": 1.905338317244229e-05, "loss": 0.895, "step": 482 }, { "epoch": 0.83, "learning_rate": 1.904864968795197e-05, "loss": 0.8659, "step": 483 }, { "epoch": 0.83, "learning_rate": 1.9043904989231703e-05, "loss": 0.8561, "step": 484 }, { "epoch": 0.83, "learning_rate": 1.9039149082161724e-05, "loss": 0.8953, "step": 485 }, { "epoch": 0.83, "learning_rate": 1.903438197263616e-05, "loss": 0.8909, "step": 486 }, { "epoch": 0.84, "learning_rate": 1.9029603666563003e-05, "loss": 0.8961, "step": 487 }, { "epoch": 0.84, "learning_rate": 1.902481416986414e-05, "loss": 0.8884, "step": 488 }, { "epoch": 0.84, "learning_rate": 1.902001348847532e-05, "loss": 0.8847, "step": 489 }, { "epoch": 0.84, "learning_rate": 1.901520162834616e-05, "loss": 0.8945, "step": 490 }, { "epoch": 0.84, "learning_rate": 1.9010378595440116e-05, "loss": 0.857, "step": 491 }, { "epoch": 0.84, "learning_rate": 1.9005544395734504e-05, "loss": 0.8914, "step": 492 }, { "epoch": 0.85, "learning_rate": 1.9000699035220473e-05, "loss": 0.8602, "step": 493 }, { "epoch": 0.85, "learning_rate": 1.8995842519903013e-05, "loss": 0.8713, "step": 494 }, { "epoch": 0.85, "learning_rate": 1.8990974855800918e-05, "loss": 0.8928, "step": 495 }, { "epoch": 0.85, "learning_rate": 1.8986096048946826e-05, "loss": 0.8546, "step": 496 }, { "epoch": 0.85, "learning_rate": 1.898120610538716e-05, "loss": 0.8535, "step": 497 }, { "epoch": 0.85, "learning_rate": 1.897630503118216e-05, "loss": 0.8535, "step": 498 }, { "epoch": 0.86, "learning_rate": 1.897139283240586e-05, "loss": 0.8585, "step": 499 }, { "epoch": 0.86, "learning_rate": 1.8966469515146076e-05, "loss": 0.8591, "step": 500 }, { "epoch": 0.86, "learning_rate": 1.8961535085504402e-05, "loss": 0.8621, "step": 501 }, { "epoch": 0.86, "learning_rate": 1.8956589549596207e-05, "loss": 0.8972, "step": 502 }, { "epoch": 0.86, "learning_rate": 1.8951632913550625e-05, "loss": 0.9082, "step": 503 }, { "epoch": 0.87, "learning_rate": 1.8946665183510546e-05, "loss": 0.8826, "step": 504 }, { "epoch": 0.87, "learning_rate": 1.8941686365632607e-05, "loss": 0.8701, "step": 505 }, { "epoch": 0.87, "learning_rate": 1.893669646608719e-05, "loss": 0.8873, "step": 506 }, { "epoch": 0.87, "learning_rate": 1.893169549105841e-05, "loss": 0.8976, "step": 507 }, { "epoch": 0.87, "learning_rate": 1.8926683446744102e-05, "loss": 0.8813, "step": 508 }, { "epoch": 0.87, "learning_rate": 1.8921660339355828e-05, "loss": 0.8769, "step": 509 }, { "epoch": 0.88, "learning_rate": 1.8916626175118863e-05, "loss": 0.871, "step": 510 }, { "epoch": 0.88, "learning_rate": 1.8911580960272168e-05, "loss": 0.8386, "step": 511 }, { "epoch": 0.88, "learning_rate": 1.890652470106842e-05, "loss": 0.9033, "step": 512 }, { "epoch": 0.88, "learning_rate": 1.890145740377397e-05, "loss": 0.852, "step": 513 }, { "epoch": 0.88, "learning_rate": 1.8896379074668848e-05, "loss": 0.8568, "step": 514 }, { "epoch": 0.88, "learning_rate": 1.8891289720046774e-05, "loss": 0.8493, "step": 515 }, { "epoch": 0.89, "learning_rate": 1.8886189346215107e-05, "loss": 0.9159, "step": 516 }, { "epoch": 0.89, "learning_rate": 1.8881077959494882e-05, "loss": 0.8917, "step": 517 }, { "epoch": 0.89, "learning_rate": 1.887595556622077e-05, "loss": 0.8711, "step": 518 }, { "epoch": 0.89, "learning_rate": 1.8870822172741093e-05, "loss": 0.8845, "step": 519 }, { "epoch": 0.89, "learning_rate": 1.88656777854178e-05, "loss": 0.8465, "step": 520 }, { "epoch": 0.89, "learning_rate": 1.8860522410626462e-05, "loss": 0.8295, "step": 521 }, { "epoch": 0.9, "learning_rate": 1.8855356054756274e-05, "loss": 0.835, "step": 522 }, { "epoch": 0.9, "learning_rate": 1.8850178724210037e-05, "loss": 0.8881, "step": 523 }, { "epoch": 0.9, "learning_rate": 1.8844990425404147e-05, "loss": 0.8545, "step": 524 }, { "epoch": 0.9, "learning_rate": 1.883979116476861e-05, "loss": 0.8724, "step": 525 }, { "epoch": 0.9, "learning_rate": 1.8834580948746997e-05, "loss": 0.8561, "step": 526 }, { "epoch": 0.9, "learning_rate": 1.8829359783796474e-05, "loss": 0.8595, "step": 527 }, { "epoch": 0.91, "learning_rate": 1.8824127676387764e-05, "loss": 0.8641, "step": 528 }, { "epoch": 0.91, "learning_rate": 1.8818884633005153e-05, "loss": 0.8217, "step": 529 }, { "epoch": 0.91, "learning_rate": 1.881363066014649e-05, "loss": 0.8534, "step": 530 }, { "epoch": 0.91, "learning_rate": 1.8808365764323157e-05, "loss": 0.9072, "step": 531 }, { "epoch": 0.91, "learning_rate": 1.8803089952060075e-05, "loss": 0.8896, "step": 532 }, { "epoch": 0.92, "learning_rate": 1.8797803229895702e-05, "loss": 0.8777, "step": 533 }, { "epoch": 0.92, "learning_rate": 1.8792505604382013e-05, "loss": 0.8399, "step": 534 }, { "epoch": 0.92, "learning_rate": 1.8787197082084493e-05, "loss": 0.8512, "step": 535 }, { "epoch": 0.92, "learning_rate": 1.878187766958213e-05, "loss": 0.8685, "step": 536 }, { "epoch": 0.92, "learning_rate": 1.877654737346742e-05, "loss": 0.8392, "step": 537 }, { "epoch": 0.92, "learning_rate": 1.8771206200346332e-05, "loss": 0.8407, "step": 538 }, { "epoch": 0.93, "learning_rate": 1.876585415683833e-05, "loss": 0.8447, "step": 539 }, { "epoch": 0.93, "learning_rate": 1.876049124957633e-05, "loss": 0.8844, "step": 540 }, { "epoch": 0.93, "learning_rate": 1.8755117485206736e-05, "loss": 0.8707, "step": 541 }, { "epoch": 0.93, "learning_rate": 1.874973287038939e-05, "loss": 0.8821, "step": 542 }, { "epoch": 0.93, "learning_rate": 1.8744337411797594e-05, "loss": 0.8531, "step": 543 }, { "epoch": 0.93, "learning_rate": 1.8738931116118076e-05, "loss": 0.854, "step": 544 }, { "epoch": 0.94, "learning_rate": 1.8733513990051e-05, "loss": 0.8432, "step": 545 }, { "epoch": 0.94, "learning_rate": 1.8728086040309952e-05, "loss": 0.8489, "step": 546 }, { "epoch": 0.94, "learning_rate": 1.872264727362194e-05, "loss": 0.8402, "step": 547 }, { "epoch": 0.94, "learning_rate": 1.8717197696727364e-05, "loss": 0.8214, "step": 548 }, { "epoch": 0.94, "learning_rate": 1.8711737316380033e-05, "loss": 0.8557, "step": 549 }, { "epoch": 0.94, "learning_rate": 1.8706266139347136e-05, "loss": 0.8626, "step": 550 }, { "epoch": 0.95, "learning_rate": 1.870078417240925e-05, "loss": 0.8506, "step": 551 }, { "epoch": 0.95, "learning_rate": 1.8695291422360317e-05, "loss": 0.8532, "step": 552 }, { "epoch": 0.95, "learning_rate": 1.8689787896007653e-05, "loss": 0.8458, "step": 553 }, { "epoch": 0.95, "learning_rate": 1.868427360017192e-05, "loss": 0.8206, "step": 554 }, { "epoch": 0.95, "learning_rate": 1.8678748541687127e-05, "loss": 0.8109, "step": 555 }, { "epoch": 0.95, "learning_rate": 1.867321272740063e-05, "loss": 0.8463, "step": 556 }, { "epoch": 0.96, "learning_rate": 1.8667666164173106e-05, "loss": 0.8823, "step": 557 }, { "epoch": 0.96, "learning_rate": 1.8662108858878558e-05, "loss": 0.8496, "step": 558 }, { "epoch": 0.96, "learning_rate": 1.8656540818404304e-05, "loss": 0.8571, "step": 559 }, { "epoch": 0.96, "learning_rate": 1.8650962049650955e-05, "loss": 0.8688, "step": 560 }, { "epoch": 0.96, "learning_rate": 1.8645372559532434e-05, "loss": 0.8496, "step": 561 }, { "epoch": 0.96, "learning_rate": 1.863977235497594e-05, "loss": 0.8506, "step": 562 }, { "epoch": 0.97, "learning_rate": 1.8634161442921956e-05, "loss": 0.861, "step": 563 }, { "epoch": 0.97, "learning_rate": 1.862853983032423e-05, "loss": 0.8369, "step": 564 }, { "epoch": 0.97, "learning_rate": 1.8622907524149778e-05, "loss": 0.8472, "step": 565 }, { "epoch": 0.97, "learning_rate": 1.861726453137886e-05, "loss": 0.8167, "step": 566 }, { "epoch": 0.97, "learning_rate": 1.8611610859004993e-05, "loss": 0.848, "step": 567 }, { "epoch": 0.98, "learning_rate": 1.8605946514034917e-05, "loss": 0.8405, "step": 568 }, { "epoch": 0.98, "learning_rate": 1.8600271503488608e-05, "loss": 0.8291, "step": 569 }, { "epoch": 0.98, "learning_rate": 1.859458583439925e-05, "loss": 0.8053, "step": 570 }, { "epoch": 0.98, "learning_rate": 1.8588889513813246e-05, "loss": 0.861, "step": 571 }, { "epoch": 0.98, "learning_rate": 1.8583182548790196e-05, "loss": 0.8256, "step": 572 }, { "epoch": 0.98, "learning_rate": 1.8577464946402893e-05, "loss": 0.8561, "step": 573 }, { "epoch": 0.99, "learning_rate": 1.8571736713737313e-05, "loss": 0.8495, "step": 574 }, { "epoch": 0.99, "learning_rate": 1.85659978578926e-05, "loss": 0.8569, "step": 575 }, { "epoch": 0.99, "learning_rate": 1.856024838598108e-05, "loss": 0.8421, "step": 576 }, { "epoch": 0.99, "learning_rate": 1.8554488305128214e-05, "loss": 0.8643, "step": 577 }, { "epoch": 0.99, "learning_rate": 1.8548717622472628e-05, "loss": 0.8353, "step": 578 }, { "epoch": 0.99, "learning_rate": 1.8542936345166075e-05, "loss": 0.8101, "step": 579 }, { "epoch": 1.0, "learning_rate": 1.8537144480373455e-05, "loss": 0.815, "step": 580 }, { "epoch": 1.0, "learning_rate": 1.8531342035272768e-05, "loss": 0.8404, "step": 581 }, { "epoch": 1.0, "learning_rate": 1.8525529017055144e-05, "loss": 0.8731, "step": 582 }, { "epoch": 1.0, "learning_rate": 1.8519705432924806e-05, "loss": 0.7601, "step": 583 }, { "epoch": 1.0, "learning_rate": 1.8513871290099074e-05, "loss": 0.759, "step": 584 }, { "epoch": 1.0, "learning_rate": 1.850802659580836e-05, "loss": 0.7238, "step": 585 }, { "epoch": 1.01, "learning_rate": 1.8502171357296144e-05, "loss": 0.7459, "step": 586 }, { "epoch": 1.01, "learning_rate": 1.8496305581818978e-05, "loss": 0.7406, "step": 587 }, { "epoch": 1.01, "learning_rate": 1.849042927664647e-05, "loss": 0.726, "step": 588 }, { "epoch": 1.01, "learning_rate": 1.8484542449061284e-05, "loss": 0.7317, "step": 589 }, { "epoch": 1.01, "learning_rate": 1.8478645106359118e-05, "loss": 0.7201, "step": 590 }, { "epoch": 1.01, "learning_rate": 1.8472737255848705e-05, "loss": 0.724, "step": 591 }, { "epoch": 1.02, "learning_rate": 1.84668189048518e-05, "loss": 0.7195, "step": 592 }, { "epoch": 1.02, "learning_rate": 1.846089006070317e-05, "loss": 0.748, "step": 593 }, { "epoch": 1.02, "learning_rate": 1.8454950730750586e-05, "loss": 0.7335, "step": 594 }, { "epoch": 1.02, "learning_rate": 1.8449000922354823e-05, "loss": 0.7454, "step": 595 }, { "epoch": 1.02, "learning_rate": 1.844304064288963e-05, "loss": 0.7306, "step": 596 }, { "epoch": 1.02, "learning_rate": 1.8437069899741736e-05, "loss": 0.7246, "step": 597 }, { "epoch": 1.03, "learning_rate": 1.8431088700310846e-05, "loss": 0.7355, "step": 598 }, { "epoch": 1.03, "learning_rate": 1.8425097052009615e-05, "loss": 0.713, "step": 599 }, { "epoch": 1.03, "learning_rate": 1.841909496226365e-05, "loss": 0.7363, "step": 600 }, { "epoch": 1.03, "learning_rate": 1.8413082438511504e-05, "loss": 0.7037, "step": 601 }, { "epoch": 1.03, "learning_rate": 1.8407059488204647e-05, "loss": 0.7218, "step": 602 }, { "epoch": 1.04, "learning_rate": 1.8401026118807487e-05, "loss": 0.7086, "step": 603 }, { "epoch": 1.04, "learning_rate": 1.839498233779734e-05, "loss": 0.7152, "step": 604 }, { "epoch": 1.04, "learning_rate": 1.838892815266441e-05, "loss": 0.7433, "step": 605 }, { "epoch": 1.04, "learning_rate": 1.8382863570911823e-05, "loss": 0.7195, "step": 606 }, { "epoch": 1.04, "learning_rate": 1.8376788600055565e-05, "loss": 0.73, "step": 607 }, { "epoch": 1.04, "learning_rate": 1.8370703247624514e-05, "loss": 0.7106, "step": 608 }, { "epoch": 1.05, "learning_rate": 1.836460752116041e-05, "loss": 0.7296, "step": 609 }, { "epoch": 1.05, "learning_rate": 1.8358501428217835e-05, "loss": 0.7299, "step": 610 }, { "epoch": 1.05, "learning_rate": 1.8352384976364236e-05, "loss": 0.7402, "step": 611 }, { "epoch": 1.05, "learning_rate": 1.8346258173179904e-05, "loss": 0.7292, "step": 612 }, { "epoch": 1.05, "learning_rate": 1.834012102625793e-05, "loss": 0.6933, "step": 613 }, { "epoch": 1.05, "learning_rate": 1.8333973543204256e-05, "loss": 0.7171, "step": 614 }, { "epoch": 1.06, "learning_rate": 1.8327815731637612e-05, "loss": 0.7055, "step": 615 }, { "epoch": 1.06, "learning_rate": 1.8321647599189538e-05, "loss": 0.7174, "step": 616 }, { "epoch": 1.06, "learning_rate": 1.8315469153504365e-05, "loss": 0.7028, "step": 617 }, { "epoch": 1.06, "learning_rate": 1.83092804022392e-05, "loss": 0.7282, "step": 618 }, { "epoch": 1.06, "learning_rate": 1.8303081353063928e-05, "loss": 0.7299, "step": 619 }, { "epoch": 1.06, "learning_rate": 1.8296872013661192e-05, "loss": 0.7254, "step": 620 }, { "epoch": 1.07, "learning_rate": 1.829065239172639e-05, "loss": 0.7294, "step": 621 }, { "epoch": 1.07, "learning_rate": 1.8284422494967668e-05, "loss": 0.7464, "step": 622 }, { "epoch": 1.07, "learning_rate": 1.82781823311059e-05, "loss": 0.726, "step": 623 }, { "epoch": 1.07, "learning_rate": 1.8271931907874675e-05, "loss": 0.7273, "step": 624 }, { "epoch": 1.07, "learning_rate": 1.826567123302032e-05, "loss": 0.7075, "step": 625 }, { "epoch": 1.07, "learning_rate": 1.825940031430185e-05, "loss": 0.7014, "step": 626 }, { "epoch": 1.08, "learning_rate": 1.8253119159490983e-05, "loss": 0.6809, "step": 627 }, { "epoch": 1.08, "learning_rate": 1.824682777637211e-05, "loss": 0.715, "step": 628 }, { "epoch": 1.08, "learning_rate": 1.8240526172742314e-05, "loss": 0.7328, "step": 629 }, { "epoch": 1.08, "learning_rate": 1.8234214356411343e-05, "loss": 0.7247, "step": 630 }, { "epoch": 1.08, "learning_rate": 1.8227892335201586e-05, "loss": 0.7256, "step": 631 }, { "epoch": 1.08, "learning_rate": 1.8221560116948103e-05, "loss": 0.7239, "step": 632 }, { "epoch": 1.09, "learning_rate": 1.821521770949857e-05, "loss": 0.7135, "step": 633 }, { "epoch": 1.09, "learning_rate": 1.82088651207133e-05, "loss": 0.7287, "step": 634 }, { "epoch": 1.09, "learning_rate": 1.820250235846523e-05, "loss": 0.7108, "step": 635 }, { "epoch": 1.09, "learning_rate": 1.8196129430639895e-05, "loss": 0.7372, "step": 636 }, { "epoch": 1.09, "learning_rate": 1.818974634513543e-05, "loss": 0.7085, "step": 637 }, { "epoch": 1.1, "learning_rate": 1.8183353109862563e-05, "loss": 0.7313, "step": 638 }, { "epoch": 1.1, "learning_rate": 1.81769497327446e-05, "loss": 0.7261, "step": 639 }, { "epoch": 1.1, "learning_rate": 1.8170536221717416e-05, "loss": 0.735, "step": 640 }, { "epoch": 1.1, "learning_rate": 1.816411258472944e-05, "loss": 0.7045, "step": 641 }, { "epoch": 1.1, "learning_rate": 1.815767882974166e-05, "loss": 0.7056, "step": 642 }, { "epoch": 1.1, "learning_rate": 1.8151234964727592e-05, "loss": 0.7021, "step": 643 }, { "epoch": 1.11, "learning_rate": 1.8144780997673293e-05, "loss": 0.6887, "step": 644 }, { "epoch": 1.11, "learning_rate": 1.8138316936577333e-05, "loss": 0.7123, "step": 645 }, { "epoch": 1.11, "learning_rate": 1.8131842789450787e-05, "loss": 0.7079, "step": 646 }, { "epoch": 1.11, "learning_rate": 1.8125358564317242e-05, "loss": 0.7352, "step": 647 }, { "epoch": 1.11, "learning_rate": 1.8118864269212775e-05, "loss": 0.7446, "step": 648 }, { "epoch": 1.11, "learning_rate": 1.8112359912185923e-05, "loss": 0.7095, "step": 649 }, { "epoch": 1.12, "learning_rate": 1.8105845501297722e-05, "loss": 0.7107, "step": 650 }, { "epoch": 1.12, "learning_rate": 1.8099321044621637e-05, "loss": 0.7157, "step": 651 }, { "epoch": 1.12, "learning_rate": 1.8092786550243613e-05, "loss": 0.7124, "step": 652 }, { "epoch": 1.12, "learning_rate": 1.8086242026262014e-05, "loss": 0.7334, "step": 653 }, { "epoch": 1.12, "learning_rate": 1.807968748078764e-05, "loss": 0.741, "step": 654 }, { "epoch": 1.12, "learning_rate": 1.807312292194372e-05, "loss": 0.7043, "step": 655 }, { "epoch": 1.13, "learning_rate": 1.8066548357865876e-05, "loss": 0.712, "step": 656 }, { "epoch": 1.13, "learning_rate": 1.8059963796702143e-05, "loss": 0.7097, "step": 657 }, { "epoch": 1.13, "learning_rate": 1.805336924661294e-05, "loss": 0.7058, "step": 658 }, { "epoch": 1.13, "learning_rate": 1.804676471577106e-05, "loss": 0.7045, "step": 659 }, { "epoch": 1.13, "learning_rate": 1.8040150212361687e-05, "loss": 0.6893, "step": 660 }, { "epoch": 1.13, "learning_rate": 1.8033525744582335e-05, "loss": 0.7189, "step": 661 }, { "epoch": 1.14, "learning_rate": 1.802689132064289e-05, "loss": 0.7039, "step": 662 }, { "epoch": 1.14, "learning_rate": 1.8020246948765565e-05, "loss": 0.7163, "step": 663 }, { "epoch": 1.14, "learning_rate": 1.8013592637184903e-05, "loss": 0.7076, "step": 664 }, { "epoch": 1.14, "learning_rate": 1.8006928394147777e-05, "loss": 0.7417, "step": 665 }, { "epoch": 1.14, "learning_rate": 1.8000254227913346e-05, "loss": 0.7422, "step": 666 }, { "epoch": 1.15, "learning_rate": 1.7993570146753088e-05, "loss": 0.6987, "step": 667 }, { "epoch": 1.15, "learning_rate": 1.798687615895076e-05, "loss": 0.7186, "step": 668 }, { "epoch": 1.15, "learning_rate": 1.7980172272802398e-05, "loss": 0.6964, "step": 669 }, { "epoch": 1.15, "learning_rate": 1.79734584966163e-05, "loss": 0.7275, "step": 670 }, { "epoch": 1.15, "learning_rate": 1.7966734838713032e-05, "loss": 0.7005, "step": 671 }, { "epoch": 1.15, "learning_rate": 1.7960001307425394e-05, "loss": 0.7047, "step": 672 }, { "epoch": 1.16, "learning_rate": 1.7953257911098435e-05, "loss": 0.7499, "step": 673 }, { "epoch": 1.16, "learning_rate": 1.794650465808942e-05, "loss": 0.7292, "step": 674 }, { "epoch": 1.16, "learning_rate": 1.7939741556767837e-05, "loss": 0.7262, "step": 675 }, { "epoch": 1.16, "learning_rate": 1.7932968615515368e-05, "loss": 0.7028, "step": 676 }, { "epoch": 1.16, "learning_rate": 1.7926185842725904e-05, "loss": 0.7059, "step": 677 }, { "epoch": 1.16, "learning_rate": 1.7919393246805513e-05, "loss": 0.7238, "step": 678 }, { "epoch": 1.17, "learning_rate": 1.7912590836172438e-05, "loss": 0.715, "step": 679 }, { "epoch": 1.17, "learning_rate": 1.7905778619257087e-05, "loss": 0.6913, "step": 680 }, { "epoch": 1.17, "learning_rate": 1.7898956604502012e-05, "loss": 0.6934, "step": 681 }, { "epoch": 1.17, "learning_rate": 1.7892124800361927e-05, "loss": 0.7053, "step": 682 }, { "epoch": 1.17, "learning_rate": 1.788528321530366e-05, "loss": 0.7006, "step": 683 }, { "epoch": 1.17, "learning_rate": 1.787843185780617e-05, "loss": 0.7412, "step": 684 }, { "epoch": 1.18, "learning_rate": 1.787157073636052e-05, "loss": 0.684, "step": 685 }, { "epoch": 1.18, "learning_rate": 1.7864699859469887e-05, "loss": 0.7071, "step": 686 }, { "epoch": 1.18, "learning_rate": 1.7857819235649525e-05, "loss": 0.7047, "step": 687 }, { "epoch": 1.18, "learning_rate": 1.785092887342677e-05, "loss": 0.7177, "step": 688 }, { "epoch": 1.18, "learning_rate": 1.7844028781341033e-05, "loss": 0.7065, "step": 689 }, { "epoch": 1.18, "learning_rate": 1.783711896794378e-05, "loss": 0.7147, "step": 690 }, { "epoch": 1.19, "learning_rate": 1.7830199441798526e-05, "loss": 0.7027, "step": 691 }, { "epoch": 1.19, "learning_rate": 1.7823270211480817e-05, "loss": 0.7098, "step": 692 }, { "epoch": 1.19, "learning_rate": 1.781633128557823e-05, "loss": 0.714, "step": 693 }, { "epoch": 1.19, "learning_rate": 1.7809382672690368e-05, "loss": 0.6818, "step": 694 }, { "epoch": 1.19, "learning_rate": 1.7802424381428823e-05, "loss": 0.7487, "step": 695 }, { "epoch": 1.19, "learning_rate": 1.779545642041719e-05, "loss": 0.7013, "step": 696 }, { "epoch": 1.2, "learning_rate": 1.7788478798291043e-05, "loss": 0.703, "step": 697 }, { "epoch": 1.2, "learning_rate": 1.7781491523697938e-05, "loss": 0.684, "step": 698 }, { "epoch": 1.2, "learning_rate": 1.777449460529739e-05, "loss": 0.6817, "step": 699 }, { "epoch": 1.2, "learning_rate": 1.7767488051760858e-05, "loss": 0.7026, "step": 700 }, { "epoch": 1.2, "learning_rate": 1.7760471871771752e-05, "loss": 0.6681, "step": 701 }, { "epoch": 1.21, "learning_rate": 1.775344607402541e-05, "loss": 0.7262, "step": 702 }, { "epoch": 1.21, "learning_rate": 1.7746410667229083e-05, "loss": 0.6904, "step": 703 }, { "epoch": 1.21, "learning_rate": 1.773936566010194e-05, "loss": 0.6818, "step": 704 }, { "epoch": 1.21, "learning_rate": 1.7732311061375048e-05, "loss": 0.6862, "step": 705 }, { "epoch": 1.21, "learning_rate": 1.772524687979135e-05, "loss": 0.7132, "step": 706 }, { "epoch": 1.21, "learning_rate": 1.7718173124105674e-05, "loss": 0.6989, "step": 707 }, { "epoch": 1.22, "learning_rate": 1.7711089803084714e-05, "loss": 0.6787, "step": 708 }, { "epoch": 1.22, "learning_rate": 1.7703996925507014e-05, "loss": 0.6837, "step": 709 }, { "epoch": 1.22, "learning_rate": 1.7696894500162964e-05, "loss": 0.6883, "step": 710 }, { "epoch": 1.22, "learning_rate": 1.7689782535854786e-05, "loss": 0.6828, "step": 711 }, { "epoch": 1.22, "learning_rate": 1.7682661041396532e-05, "loss": 0.7217, "step": 712 }, { "epoch": 1.22, "learning_rate": 1.767553002561405e-05, "loss": 0.6975, "step": 713 }, { "epoch": 1.23, "learning_rate": 1.7668389497345e-05, "loss": 0.7147, "step": 714 }, { "epoch": 1.23, "learning_rate": 1.766123946543882e-05, "loss": 0.673, "step": 715 }, { "epoch": 1.23, "learning_rate": 1.765407993875674e-05, "loss": 0.7232, "step": 716 }, { "epoch": 1.23, "learning_rate": 1.7646910926171747e-05, "loss": 0.7093, "step": 717 }, { "epoch": 1.23, "learning_rate": 1.7639732436568588e-05, "loss": 0.6856, "step": 718 }, { "epoch": 1.23, "learning_rate": 1.7632544478843756e-05, "loss": 0.6835, "step": 719 }, { "epoch": 1.24, "learning_rate": 1.7625347061905475e-05, "loss": 0.6903, "step": 720 }, { "epoch": 1.24, "learning_rate": 1.761814019467369e-05, "loss": 0.7369, "step": 721 }, { "epoch": 1.24, "learning_rate": 1.7610923886080066e-05, "loss": 0.6942, "step": 722 }, { "epoch": 1.24, "learning_rate": 1.7603698145067962e-05, "loss": 0.6994, "step": 723 }, { "epoch": 1.24, "learning_rate": 1.759646298059243e-05, "loss": 0.7048, "step": 724 }, { "epoch": 1.24, "learning_rate": 1.75892184016202e-05, "loss": 0.7021, "step": 725 }, { "epoch": 1.25, "learning_rate": 1.7581964417129673e-05, "loss": 0.6894, "step": 726 }, { "epoch": 1.25, "learning_rate": 1.7574701036110894e-05, "loss": 0.6855, "step": 727 }, { "epoch": 1.25, "learning_rate": 1.7567428267565572e-05, "loss": 0.6828, "step": 728 }, { "epoch": 1.25, "learning_rate": 1.7560146120507036e-05, "loss": 0.7038, "step": 729 }, { "epoch": 1.25, "learning_rate": 1.755285460396024e-05, "loss": 0.6811, "step": 730 }, { "epoch": 1.25, "learning_rate": 1.7545553726961763e-05, "loss": 0.7058, "step": 731 }, { "epoch": 1.26, "learning_rate": 1.753824349855976e-05, "loss": 0.6988, "step": 732 }, { "epoch": 1.26, "learning_rate": 1.7530923927814e-05, "loss": 0.6974, "step": 733 }, { "epoch": 1.26, "learning_rate": 1.7523595023795814e-05, "loss": 0.6888, "step": 734 }, { "epoch": 1.26, "learning_rate": 1.751625679558811e-05, "loss": 0.7215, "step": 735 }, { "epoch": 1.26, "learning_rate": 1.750890925228535e-05, "loss": 0.6579, "step": 736 }, { "epoch": 1.27, "learning_rate": 1.7501552402993523e-05, "loss": 0.6991, "step": 737 }, { "epoch": 1.27, "learning_rate": 1.749418625683019e-05, "loss": 0.6669, "step": 738 }, { "epoch": 1.27, "learning_rate": 1.7486810822924388e-05, "loss": 0.6772, "step": 739 }, { "epoch": 1.27, "learning_rate": 1.74794261104167e-05, "loss": 0.7143, "step": 740 }, { "epoch": 1.27, "learning_rate": 1.747203212845919e-05, "loss": 0.6921, "step": 741 }, { "epoch": 1.27, "learning_rate": 1.7464628886215416e-05, "loss": 0.7107, "step": 742 }, { "epoch": 1.28, "learning_rate": 1.7457216392860406e-05, "loss": 0.6854, "step": 743 }, { "epoch": 1.28, "learning_rate": 1.7449794657580666e-05, "loss": 0.7054, "step": 744 }, { "epoch": 1.28, "learning_rate": 1.7442363689574143e-05, "loss": 0.6821, "step": 745 }, { "epoch": 1.28, "learning_rate": 1.7434923498050234e-05, "loss": 0.7164, "step": 746 }, { "epoch": 1.28, "learning_rate": 1.7427474092229763e-05, "loss": 0.7189, "step": 747 }, { "epoch": 1.28, "learning_rate": 1.7420015481344972e-05, "loss": 0.6801, "step": 748 }, { "epoch": 1.29, "learning_rate": 1.741254767463952e-05, "loss": 0.6759, "step": 749 }, { "epoch": 1.29, "learning_rate": 1.7405070681368456e-05, "loss": 0.6841, "step": 750 }, { "epoch": 1.29, "learning_rate": 1.7397584510798208e-05, "loss": 0.6862, "step": 751 }, { "epoch": 1.29, "learning_rate": 1.7390089172206594e-05, "loss": 0.6884, "step": 752 }, { "epoch": 1.29, "learning_rate": 1.7382584674882776e-05, "loss": 0.6549, "step": 753 }, { "epoch": 1.29, "learning_rate": 1.7375071028127277e-05, "loss": 0.6771, "step": 754 }, { "epoch": 1.3, "learning_rate": 1.7367548241251962e-05, "loss": 0.6846, "step": 755 }, { "epoch": 1.3, "learning_rate": 1.7360016323580017e-05, "loss": 0.7005, "step": 756 }, { "epoch": 1.3, "learning_rate": 1.735247528444594e-05, "loss": 0.6693, "step": 757 }, { "epoch": 1.3, "learning_rate": 1.734492513319555e-05, "loss": 0.6913, "step": 758 }, { "epoch": 1.3, "learning_rate": 1.733736587918594e-05, "loss": 0.6533, "step": 759 }, { "epoch": 1.3, "learning_rate": 1.73297975317855e-05, "loss": 0.6721, "step": 760 }, { "epoch": 1.31, "learning_rate": 1.732222010037387e-05, "loss": 0.6647, "step": 761 }, { "epoch": 1.31, "learning_rate": 1.7314633594341974e-05, "loss": 0.6671, "step": 762 }, { "epoch": 1.31, "learning_rate": 1.730703802309196e-05, "loss": 0.6782, "step": 763 }, { "epoch": 1.31, "learning_rate": 1.7299433396037223e-05, "loss": 0.6799, "step": 764 }, { "epoch": 1.31, "learning_rate": 1.729181972260238e-05, "loss": 0.6704, "step": 765 }, { "epoch": 1.32, "learning_rate": 1.7284197012223253e-05, "loss": 0.6823, "step": 766 }, { "epoch": 1.32, "learning_rate": 1.7276565274346868e-05, "loss": 0.6672, "step": 767 }, { "epoch": 1.32, "learning_rate": 1.7268924518431437e-05, "loss": 0.6811, "step": 768 }, { "epoch": 1.32, "learning_rate": 1.7261274753946356e-05, "loss": 0.6772, "step": 769 }, { "epoch": 1.32, "learning_rate": 1.7253615990372177e-05, "loss": 0.6662, "step": 770 }, { "epoch": 1.32, "learning_rate": 1.7245948237200608e-05, "loss": 0.6679, "step": 771 }, { "epoch": 1.33, "learning_rate": 1.7238271503934494e-05, "loss": 0.6739, "step": 772 }, { "epoch": 1.33, "learning_rate": 1.7230585800087816e-05, "loss": 0.6695, "step": 773 }, { "epoch": 1.33, "learning_rate": 1.7222891135185672e-05, "loss": 0.6783, "step": 774 }, { "epoch": 1.33, "learning_rate": 1.7215187518764253e-05, "loss": 0.6488, "step": 775 }, { "epoch": 1.33, "learning_rate": 1.7207474960370865e-05, "loss": 0.6894, "step": 776 }, { "epoch": 1.33, "learning_rate": 1.719975346956388e-05, "loss": 0.6542, "step": 777 }, { "epoch": 1.34, "learning_rate": 1.7192023055912743e-05, "loss": 0.6645, "step": 778 }, { "epoch": 1.34, "learning_rate": 1.7184283728997965e-05, "loss": 0.6633, "step": 779 }, { "epoch": 1.34, "learning_rate": 1.7176535498411093e-05, "loss": 0.6805, "step": 780 }, { "epoch": 1.34, "learning_rate": 1.7168778373754708e-05, "loss": 0.6798, "step": 781 }, { "epoch": 1.34, "learning_rate": 1.7161012364642432e-05, "loss": 0.6558, "step": 782 }, { "epoch": 1.34, "learning_rate": 1.7153237480698877e-05, "loss": 0.6779, "step": 783 }, { "epoch": 1.35, "learning_rate": 1.714545373155966e-05, "loss": 0.6635, "step": 784 }, { "epoch": 1.35, "learning_rate": 1.713766112687139e-05, "loss": 0.6742, "step": 785 }, { "epoch": 1.35, "learning_rate": 1.7129859676291645e-05, "loss": 0.6788, "step": 786 }, { "epoch": 1.35, "learning_rate": 1.7122049389488972e-05, "loss": 0.6818, "step": 787 }, { "epoch": 1.35, "learning_rate": 1.7114230276142866e-05, "loss": 0.6552, "step": 788 }, { "epoch": 1.35, "learning_rate": 1.7106402345943748e-05, "loss": 0.6781, "step": 789 }, { "epoch": 1.36, "learning_rate": 1.7098565608592993e-05, "loss": 0.6618, "step": 790 }, { "epoch": 1.36, "learning_rate": 1.709072007380287e-05, "loss": 0.6475, "step": 791 }, { "epoch": 1.36, "learning_rate": 1.7082865751296553e-05, "loss": 0.6611, "step": 792 }, { "epoch": 1.36, "learning_rate": 1.707500265080812e-05, "loss": 0.6653, "step": 793 }, { "epoch": 1.36, "learning_rate": 1.706713078208251e-05, "loss": 0.6744, "step": 794 }, { "epoch": 1.36, "learning_rate": 1.705925015487554e-05, "loss": 0.6952, "step": 795 }, { "epoch": 1.37, "learning_rate": 1.705136077895388e-05, "loss": 0.6548, "step": 796 }, { "epoch": 1.37, "learning_rate": 1.704346266409504e-05, "loss": 0.6493, "step": 797 }, { "epoch": 1.37, "learning_rate": 1.7035555820087366e-05, "loss": 0.6428, "step": 798 }, { "epoch": 1.37, "learning_rate": 1.7027640256730013e-05, "loss": 0.6774, "step": 799 }, { "epoch": 1.37, "learning_rate": 1.701971598383295e-05, "loss": 0.6493, "step": 800 }, { "epoch": 1.38, "learning_rate": 1.701178301121693e-05, "loss": 0.6533, "step": 801 }, { "epoch": 1.38, "learning_rate": 1.700384134871351e-05, "loss": 0.6551, "step": 802 }, { "epoch": 1.38, "learning_rate": 1.699589100616499e-05, "loss": 0.6661, "step": 803 }, { "epoch": 1.38, "learning_rate": 1.6987931993424437e-05, "loss": 0.6654, "step": 804 }, { "epoch": 1.38, "learning_rate": 1.697996432035568e-05, "loss": 0.6718, "step": 805 }, { "epoch": 1.38, "learning_rate": 1.6971987996833242e-05, "loss": 0.6605, "step": 806 }, { "epoch": 1.39, "learning_rate": 1.6964003032742415e-05, "loss": 0.6431, "step": 807 }, { "epoch": 1.39, "learning_rate": 1.695600943797916e-05, "loss": 0.6619, "step": 808 }, { "epoch": 1.39, "learning_rate": 1.6948007222450158e-05, "loss": 0.6551, "step": 809 }, { "epoch": 1.39, "learning_rate": 1.693999639607276e-05, "loss": 0.6717, "step": 810 }, { "epoch": 1.39, "learning_rate": 1.6931976968774998e-05, "loss": 0.6605, "step": 811 }, { "epoch": 1.39, "learning_rate": 1.6923948950495562e-05, "loss": 0.6441, "step": 812 }, { "epoch": 1.4, "learning_rate": 1.6915912351183778e-05, "loss": 0.6878, "step": 813 }, { "epoch": 1.4, "learning_rate": 1.690786718079962e-05, "loss": 0.6706, "step": 814 }, { "epoch": 1.4, "learning_rate": 1.6899813449313678e-05, "loss": 0.6496, "step": 815 }, { "epoch": 1.4, "learning_rate": 1.6891751166707156e-05, "loss": 0.6636, "step": 816 }, { "epoch": 1.4, "learning_rate": 1.6883680342971854e-05, "loss": 0.6333, "step": 817 }, { "epoch": 1.4, "learning_rate": 1.6875600988110155e-05, "loss": 0.6637, "step": 818 }, { "epoch": 1.41, "learning_rate": 1.6867513112135012e-05, "loss": 0.6585, "step": 819 }, { "epoch": 1.41, "learning_rate": 1.6859416725069947e-05, "loss": 0.6371, "step": 820 }, { "epoch": 1.41, "learning_rate": 1.6851311836949033e-05, "loss": 0.6874, "step": 821 }, { "epoch": 1.41, "learning_rate": 1.6843198457816855e-05, "loss": 0.679, "step": 822 }, { "epoch": 1.41, "learning_rate": 1.6835076597728554e-05, "loss": 0.6453, "step": 823 }, { "epoch": 1.41, "learning_rate": 1.6826946266749753e-05, "loss": 0.6501, "step": 824 }, { "epoch": 1.42, "learning_rate": 1.681880747495659e-05, "loss": 0.6665, "step": 825 }, { "epoch": 1.42, "learning_rate": 1.6810660232435685e-05, "loss": 0.6615, "step": 826 }, { "epoch": 1.42, "learning_rate": 1.680250454928413e-05, "loss": 0.6815, "step": 827 }, { "epoch": 1.42, "learning_rate": 1.6794340435609474e-05, "loss": 0.6713, "step": 828 }, { "epoch": 1.42, "learning_rate": 1.6786167901529724e-05, "loss": 0.6619, "step": 829 }, { "epoch": 1.42, "learning_rate": 1.6777986957173315e-05, "loss": 0.6278, "step": 830 }, { "epoch": 1.43, "learning_rate": 1.676979761267911e-05, "loss": 0.652, "step": 831 }, { "epoch": 1.43, "learning_rate": 1.676159987819637e-05, "loss": 0.651, "step": 832 }, { "epoch": 1.43, "learning_rate": 1.675339376388477e-05, "loss": 0.6614, "step": 833 }, { "epoch": 1.43, "learning_rate": 1.674517927991436e-05, "loss": 0.6544, "step": 834 }, { "epoch": 1.43, "learning_rate": 1.6736956436465573e-05, "loss": 0.6342, "step": 835 }, { "epoch": 1.44, "learning_rate": 1.672872524372919e-05, "loss": 0.6493, "step": 836 }, { "epoch": 1.44, "learning_rate": 1.6720485711906346e-05, "loss": 0.6415, "step": 837 }, { "epoch": 1.44, "learning_rate": 1.671223785120851e-05, "loss": 0.6539, "step": 838 }, { "epoch": 1.44, "learning_rate": 1.6703981671857477e-05, "loss": 0.6489, "step": 839 }, { "epoch": 1.44, "learning_rate": 1.6695717184085342e-05, "loss": 0.6378, "step": 840 }, { "epoch": 1.44, "learning_rate": 1.6687444398134504e-05, "loss": 0.6631, "step": 841 }, { "epoch": 1.45, "learning_rate": 1.6679163324257643e-05, "loss": 0.6608, "step": 842 }, { "epoch": 1.45, "learning_rate": 1.6670873972717715e-05, "loss": 0.6408, "step": 843 }, { "epoch": 1.45, "learning_rate": 1.666257635378793e-05, "loss": 0.6534, "step": 844 }, { "epoch": 1.45, "learning_rate": 1.6654270477751748e-05, "loss": 0.6783, "step": 845 }, { "epoch": 1.45, "learning_rate": 1.664595635490286e-05, "loss": 0.6358, "step": 846 }, { "epoch": 1.45, "learning_rate": 1.663763399554517e-05, "loss": 0.645, "step": 847 }, { "epoch": 1.46, "learning_rate": 1.662930340999281e-05, "loss": 0.6571, "step": 848 }, { "epoch": 1.46, "learning_rate": 1.6620964608570083e-05, "loss": 0.659, "step": 849 }, { "epoch": 1.46, "learning_rate": 1.661261760161149e-05, "loss": 0.6495, "step": 850 }, { "epoch": 1.46, "learning_rate": 1.6604262399461696e-05, "loss": 0.6568, "step": 851 }, { "epoch": 1.46, "learning_rate": 1.6595899012475523e-05, "loss": 0.6437, "step": 852 }, { "epoch": 1.46, "learning_rate": 1.658752745101794e-05, "loss": 0.6223, "step": 853 }, { "epoch": 1.47, "learning_rate": 1.6579147725464038e-05, "loss": 0.6621, "step": 854 }, { "epoch": 1.47, "learning_rate": 1.6570759846199035e-05, "loss": 0.6601, "step": 855 }, { "epoch": 1.47, "learning_rate": 1.656236382361825e-05, "loss": 0.65, "step": 856 }, { "epoch": 1.47, "learning_rate": 1.6553959668127096e-05, "loss": 0.6703, "step": 857 }, { "epoch": 1.47, "learning_rate": 1.6545547390141063e-05, "loss": 0.622, "step": 858 }, { "epoch": 1.47, "learning_rate": 1.6537127000085708e-05, "loss": 0.6526, "step": 859 }, { "epoch": 1.48, "learning_rate": 1.6528698508396647e-05, "loss": 0.5989, "step": 860 }, { "epoch": 1.48, "learning_rate": 1.6520261925519523e-05, "loss": 0.6242, "step": 861 }, { "epoch": 1.48, "learning_rate": 1.6511817261910024e-05, "loss": 0.6351, "step": 862 }, { "epoch": 1.48, "learning_rate": 1.6503364528033838e-05, "loss": 0.6343, "step": 863 }, { "epoch": 1.48, "learning_rate": 1.6494903734366664e-05, "loss": 0.6479, "step": 864 }, { "epoch": 1.48, "learning_rate": 1.648643489139419e-05, "loss": 0.6268, "step": 865 }, { "epoch": 1.49, "learning_rate": 1.6477958009612068e-05, "loss": 0.6295, "step": 866 }, { "epoch": 1.49, "learning_rate": 1.6469473099525926e-05, "loss": 0.638, "step": 867 }, { "epoch": 1.49, "learning_rate": 1.6460980171651338e-05, "loss": 0.6153, "step": 868 }, { "epoch": 1.49, "learning_rate": 1.6452479236513806e-05, "loss": 0.651, "step": 869 }, { "epoch": 1.49, "learning_rate": 1.644397030464877e-05, "loss": 0.6382, "step": 870 }, { "epoch": 1.5, "learning_rate": 1.6435453386601575e-05, "loss": 0.638, "step": 871 }, { "epoch": 1.5, "learning_rate": 1.6426928492927453e-05, "loss": 0.6324, "step": 872 }, { "epoch": 1.5, "learning_rate": 1.6418395634191536e-05, "loss": 0.6292, "step": 873 }, { "epoch": 1.5, "learning_rate": 1.640985482096882e-05, "loss": 0.6145, "step": 874 }, { "epoch": 1.5, "learning_rate": 1.640130606384416e-05, "loss": 0.6078, "step": 875 }, { "epoch": 1.5, "learning_rate": 1.6392749373412253e-05, "loss": 0.6283, "step": 876 }, { "epoch": 1.51, "learning_rate": 1.638418476027763e-05, "loss": 0.6219, "step": 877 }, { "epoch": 1.51, "learning_rate": 1.637561223505465e-05, "loss": 0.6113, "step": 878 }, { "epoch": 1.51, "learning_rate": 1.636703180836746e-05, "loss": 0.6416, "step": 879 }, { "epoch": 1.51, "learning_rate": 1.6358443490850016e-05, "loss": 0.6372, "step": 880 }, { "epoch": 1.51, "learning_rate": 1.634984729314604e-05, "loss": 0.6338, "step": 881 }, { "epoch": 1.51, "learning_rate": 1.634124322590903e-05, "loss": 0.6007, "step": 882 }, { "epoch": 1.52, "learning_rate": 1.633263129980223e-05, "loss": 0.6038, "step": 883 }, { "epoch": 1.52, "learning_rate": 1.632401152549863e-05, "loss": 0.641, "step": 884 }, { "epoch": 1.52, "learning_rate": 1.6315383913680945e-05, "loss": 0.6159, "step": 885 }, { "epoch": 1.52, "learning_rate": 1.6306748475041595e-05, "loss": 0.6421, "step": 886 }, { "epoch": 1.52, "learning_rate": 1.6298105220282715e-05, "loss": 0.6064, "step": 887 }, { "epoch": 1.52, "learning_rate": 1.628945416011611e-05, "loss": 0.6338, "step": 888 }, { "epoch": 1.53, "learning_rate": 1.6280795305263276e-05, "loss": 0.6402, "step": 889 }, { "epoch": 1.53, "learning_rate": 1.6272128666455348e-05, "loss": 0.6186, "step": 890 }, { "epoch": 1.53, "learning_rate": 1.626345425443313e-05, "loss": 0.6681, "step": 891 }, { "epoch": 1.53, "learning_rate": 1.6254772079947048e-05, "loss": 0.6305, "step": 892 }, { "epoch": 1.53, "learning_rate": 1.6246082153757145e-05, "loss": 0.6444, "step": 893 }, { "epoch": 1.53, "learning_rate": 1.6237384486633077e-05, "loss": 0.6342, "step": 894 }, { "epoch": 1.54, "learning_rate": 1.6228679089354097e-05, "loss": 0.6323, "step": 895 }, { "epoch": 1.54, "learning_rate": 1.6219965972709023e-05, "loss": 0.6167, "step": 896 }, { "epoch": 1.54, "learning_rate": 1.621124514749626e-05, "loss": 0.6479, "step": 897 }, { "epoch": 1.54, "learning_rate": 1.6202516624523755e-05, "loss": 0.6231, "step": 898 }, { "epoch": 1.54, "learning_rate": 1.619378041460899e-05, "loss": 0.6403, "step": 899 }, { "epoch": 1.55, "learning_rate": 1.6185036528578985e-05, "loss": 0.6116, "step": 900 }, { "epoch": 1.55, "learning_rate": 1.6176284977270266e-05, "loss": 0.6442, "step": 901 }, { "epoch": 1.55, "learning_rate": 1.6167525771528866e-05, "loss": 0.5888, "step": 902 }, { "epoch": 1.55, "learning_rate": 1.6158758922210293e-05, "loss": 0.6195, "step": 903 }, { "epoch": 1.55, "learning_rate": 1.614998444017954e-05, "loss": 0.6485, "step": 904 }, { "epoch": 1.55, "learning_rate": 1.6141202336311046e-05, "loss": 0.6006, "step": 905 }, { "epoch": 1.56, "learning_rate": 1.613241262148871e-05, "loss": 0.6122, "step": 906 }, { "epoch": 1.56, "learning_rate": 1.6123615306605863e-05, "loss": 0.6045, "step": 907 }, { "epoch": 1.56, "learning_rate": 1.6114810402565237e-05, "loss": 0.594, "step": 908 }, { "epoch": 1.56, "learning_rate": 1.6105997920278983e-05, "loss": 0.598, "step": 909 }, { "epoch": 1.56, "learning_rate": 1.609717787066865e-05, "loss": 0.6178, "step": 910 }, { "epoch": 1.56, "learning_rate": 1.6088350264665155e-05, "loss": 0.629, "step": 911 }, { "epoch": 1.57, "learning_rate": 1.6079515113208778e-05, "loss": 0.5831, "step": 912 }, { "epoch": 1.57, "learning_rate": 1.607067242724916e-05, "loss": 0.6295, "step": 913 }, { "epoch": 1.57, "learning_rate": 1.6061822217745272e-05, "loss": 0.6223, "step": 914 }, { "epoch": 1.57, "learning_rate": 1.6052964495665414e-05, "loss": 0.6238, "step": 915 }, { "epoch": 1.57, "learning_rate": 1.6044099271987194e-05, "loss": 0.5992, "step": 916 }, { "epoch": 1.57, "learning_rate": 1.6035226557697514e-05, "loss": 0.5916, "step": 917 }, { "epoch": 1.58, "learning_rate": 1.6026346363792565e-05, "loss": 0.6168, "step": 918 }, { "epoch": 1.58, "learning_rate": 1.6017458701277807e-05, "loss": 0.5952, "step": 919 }, { "epoch": 1.58, "learning_rate": 1.6008563581167953e-05, "loss": 0.6127, "step": 920 }, { "epoch": 1.58, "learning_rate": 1.5999661014486956e-05, "loss": 0.6161, "step": 921 }, { "epoch": 1.58, "learning_rate": 1.5990751012268007e-05, "loss": 0.6289, "step": 922 }, { "epoch": 1.58, "learning_rate": 1.5981833585553504e-05, "loss": 0.5873, "step": 923 }, { "epoch": 1.59, "learning_rate": 1.5972908745395054e-05, "loss": 0.6016, "step": 924 }, { "epoch": 1.59, "learning_rate": 1.5963976502853434e-05, "loss": 0.6133, "step": 925 }, { "epoch": 1.59, "learning_rate": 1.595503686899862e-05, "loss": 0.63, "step": 926 }, { "epoch": 1.59, "learning_rate": 1.5946089854909726e-05, "loss": 0.5878, "step": 927 }, { "epoch": 1.59, "learning_rate": 1.5937135471675033e-05, "loss": 0.609, "step": 928 }, { "epoch": 1.59, "learning_rate": 1.5928173730391934e-05, "loss": 0.6274, "step": 929 }, { "epoch": 1.6, "learning_rate": 1.5919204642166956e-05, "loss": 0.6445, "step": 930 }, { "epoch": 1.6, "learning_rate": 1.5910228218115723e-05, "loss": 0.6311, "step": 931 }, { "epoch": 1.6, "learning_rate": 1.5901244469362962e-05, "loss": 0.6048, "step": 932 }, { "epoch": 1.6, "learning_rate": 1.589225340704246e-05, "loss": 0.5966, "step": 933 }, { "epoch": 1.6, "learning_rate": 1.5883255042297083e-05, "loss": 0.5853, "step": 934 }, { "epoch": 1.61, "learning_rate": 1.587424938627874e-05, "loss": 0.608, "step": 935 }, { "epoch": 1.61, "learning_rate": 1.5865236450148372e-05, "loss": 0.6076, "step": 936 }, { "epoch": 1.61, "learning_rate": 1.585621624507596e-05, "loss": 0.5844, "step": 937 }, { "epoch": 1.61, "learning_rate": 1.5847188782240473e-05, "loss": 0.5946, "step": 938 }, { "epoch": 1.61, "learning_rate": 1.5838154072829886e-05, "loss": 0.6243, "step": 939 }, { "epoch": 1.61, "learning_rate": 1.582911212804115e-05, "loss": 0.6113, "step": 940 }, { "epoch": 1.62, "learning_rate": 1.582006295908019e-05, "loss": 0.5988, "step": 941 }, { "epoch": 1.62, "learning_rate": 1.5811006577161882e-05, "loss": 0.5996, "step": 942 }, { "epoch": 1.62, "learning_rate": 1.5801942993510026e-05, "loss": 0.5947, "step": 943 }, { "epoch": 1.62, "learning_rate": 1.5792872219357372e-05, "loss": 0.6022, "step": 944 }, { "epoch": 1.62, "learning_rate": 1.5783794265945567e-05, "loss": 0.6039, "step": 945 }, { "epoch": 1.62, "learning_rate": 1.577470914452515e-05, "loss": 0.6045, "step": 946 }, { "epoch": 1.63, "learning_rate": 1.5765616866355562e-05, "loss": 0.5761, "step": 947 }, { "epoch": 1.63, "learning_rate": 1.57565174427051e-05, "loss": 0.6037, "step": 948 }, { "epoch": 1.63, "learning_rate": 1.5747410884850914e-05, "loss": 0.6093, "step": 949 }, { "epoch": 1.63, "learning_rate": 1.5738297204079005e-05, "loss": 0.6032, "step": 950 }, { "epoch": 1.63, "learning_rate": 1.57291764116842e-05, "loss": 0.5634, "step": 951 }, { "epoch": 1.63, "learning_rate": 1.572004851897013e-05, "loss": 0.6194, "step": 952 }, { "epoch": 1.64, "learning_rate": 1.5710913537249242e-05, "loss": 0.6292, "step": 953 }, { "epoch": 1.64, "learning_rate": 1.5701771477842755e-05, "loss": 0.5399, "step": 954 }, { "epoch": 1.64, "learning_rate": 1.5692622352080662e-05, "loss": 0.6033, "step": 955 }, { "epoch": 1.64, "learning_rate": 1.568346617130172e-05, "loss": 0.593, "step": 956 }, { "epoch": 1.64, "learning_rate": 1.567430294685342e-05, "loss": 0.5956, "step": 957 }, { "epoch": 1.64, "learning_rate": 1.5665132690091994e-05, "loss": 0.5767, "step": 958 }, { "epoch": 1.65, "learning_rate": 1.5655955412382375e-05, "loss": 0.567, "step": 959 }, { "epoch": 1.65, "learning_rate": 1.5646771125098213e-05, "loss": 0.5979, "step": 960 }, { "epoch": 1.65, "learning_rate": 1.5637579839621833e-05, "loss": 0.5972, "step": 961 }, { "epoch": 1.65, "learning_rate": 1.5628381567344236e-05, "loss": 0.581, "step": 962 }, { "epoch": 1.65, "learning_rate": 1.5619176319665083e-05, "loss": 0.5887, "step": 963 }, { "epoch": 1.65, "learning_rate": 1.5609964107992685e-05, "loss": 0.575, "step": 964 }, { "epoch": 1.66, "learning_rate": 1.5600744943743976e-05, "loss": 0.5775, "step": 965 }, { "epoch": 1.66, "learning_rate": 1.5591518838344507e-05, "loss": 0.5678, "step": 966 }, { "epoch": 1.66, "learning_rate": 1.5582285803228436e-05, "loss": 0.5701, "step": 967 }, { "epoch": 1.66, "learning_rate": 1.55730458498385e-05, "loss": 0.6333, "step": 968 }, { "epoch": 1.66, "learning_rate": 1.5563798989626027e-05, "loss": 0.6259, "step": 969 }, { "epoch": 1.67, "learning_rate": 1.5554545234050892e-05, "loss": 0.6078, "step": 970 }, { "epoch": 1.67, "learning_rate": 1.554528459458151e-05, "loss": 0.6015, "step": 971 }, { "epoch": 1.67, "learning_rate": 1.5536017082694846e-05, "loss": 0.5707, "step": 972 }, { "epoch": 1.67, "learning_rate": 1.5526742709876365e-05, "loss": 0.5969, "step": 973 }, { "epoch": 1.67, "learning_rate": 1.5517461487620047e-05, "loss": 0.5866, "step": 974 }, { "epoch": 1.67, "learning_rate": 1.5508173427428347e-05, "loss": 0.566, "step": 975 }, { "epoch": 1.68, "learning_rate": 1.549887854081221e-05, "loss": 0.5967, "step": 976 }, { "epoch": 1.68, "learning_rate": 1.5489576839291028e-05, "loss": 0.5859, "step": 977 }, { "epoch": 1.68, "learning_rate": 1.548026833439265e-05, "loss": 0.5772, "step": 978 }, { "epoch": 1.68, "learning_rate": 1.5470953037653348e-05, "loss": 0.5828, "step": 979 }, { "epoch": 1.68, "learning_rate": 1.5461630960617822e-05, "loss": 0.5877, "step": 980 }, { "epoch": 1.68, "learning_rate": 1.5452302114839155e-05, "loss": 0.5919, "step": 981 }, { "epoch": 1.69, "learning_rate": 1.544296651187884e-05, "loss": 0.6007, "step": 982 }, { "epoch": 1.69, "learning_rate": 1.5433624163306736e-05, "loss": 0.5793, "step": 983 }, { "epoch": 1.69, "learning_rate": 1.5424275080701056e-05, "loss": 0.5821, "step": 984 }, { "epoch": 1.69, "learning_rate": 1.541491927564837e-05, "loss": 0.5737, "step": 985 }, { "epoch": 1.69, "learning_rate": 1.540555675974357e-05, "loss": 0.5727, "step": 986 }, { "epoch": 1.69, "learning_rate": 1.5396187544589874e-05, "loss": 0.572, "step": 987 }, { "epoch": 1.7, "learning_rate": 1.5386811641798785e-05, "loss": 0.5855, "step": 988 }, { "epoch": 1.7, "learning_rate": 1.5377429062990122e-05, "loss": 0.5916, "step": 989 }, { "epoch": 1.7, "learning_rate": 1.536803981979195e-05, "loss": 0.5606, "step": 990 }, { "epoch": 1.7, "learning_rate": 1.5358643923840614e-05, "loss": 0.5973, "step": 991 }, { "epoch": 1.7, "learning_rate": 1.534924138678069e-05, "loss": 0.5697, "step": 992 }, { "epoch": 1.7, "learning_rate": 1.533983222026499e-05, "loss": 0.5936, "step": 993 }, { "epoch": 1.71, "learning_rate": 1.5330416435954548e-05, "loss": 0.5758, "step": 994 }, { "epoch": 1.71, "learning_rate": 1.5320994045518583e-05, "loss": 0.5635, "step": 995 }, { "epoch": 1.71, "learning_rate": 1.531156506063452e-05, "loss": 0.5613, "step": 996 }, { "epoch": 1.71, "learning_rate": 1.5302129492987944e-05, "loss": 0.5663, "step": 997 }, { "epoch": 1.71, "learning_rate": 1.529268735427261e-05, "loss": 0.5697, "step": 998 }, { "epoch": 1.72, "learning_rate": 1.52832386561904e-05, "loss": 0.5682, "step": 999 }, { "epoch": 1.72, "learning_rate": 1.5273783410451336e-05, "loss": 0.5549, "step": 1000 }, { "epoch": 1.72, "learning_rate": 1.526432162877356e-05, "loss": 0.5699, "step": 1001 }, { "epoch": 1.72, "learning_rate": 1.5254853322883301e-05, "loss": 0.5705, "step": 1002 }, { "epoch": 1.72, "learning_rate": 1.5245378504514881e-05, "loss": 0.5593, "step": 1003 }, { "epoch": 1.72, "learning_rate": 1.523589718541069e-05, "loss": 0.5719, "step": 1004 }, { "epoch": 1.73, "learning_rate": 1.5226409377321182e-05, "loss": 0.5847, "step": 1005 }, { "epoch": 1.73, "learning_rate": 1.5216915092004847e-05, "loss": 0.5408, "step": 1006 }, { "epoch": 1.73, "learning_rate": 1.5207414341228198e-05, "loss": 0.5918, "step": 1007 }, { "epoch": 1.73, "learning_rate": 1.5197907136765768e-05, "loss": 0.569, "step": 1008 }, { "epoch": 1.73, "learning_rate": 1.5188393490400088e-05, "loss": 0.5722, "step": 1009 }, { "epoch": 1.73, "learning_rate": 1.5178873413921666e-05, "loss": 0.576, "step": 1010 }, { "epoch": 1.74, "learning_rate": 1.5169346919128986e-05, "loss": 0.5653, "step": 1011 }, { "epoch": 1.74, "learning_rate": 1.5159814017828488e-05, "loss": 0.6109, "step": 1012 }, { "epoch": 1.74, "learning_rate": 1.5150274721834546e-05, "loss": 0.5692, "step": 1013 }, { "epoch": 1.74, "learning_rate": 1.5140729042969453e-05, "loss": 0.56, "step": 1014 }, { "epoch": 1.74, "learning_rate": 1.5131176993063431e-05, "loss": 0.5678, "step": 1015 }, { "epoch": 1.74, "learning_rate": 1.512161858395458e-05, "loss": 0.5689, "step": 1016 }, { "epoch": 1.75, "learning_rate": 1.511205382748889e-05, "loss": 0.5514, "step": 1017 }, { "epoch": 1.75, "learning_rate": 1.5102482735520218e-05, "loss": 0.5886, "step": 1018 }, { "epoch": 1.75, "learning_rate": 1.5092905319910268e-05, "loss": 0.5335, "step": 1019 }, { "epoch": 1.75, "learning_rate": 1.5083321592528584e-05, "loss": 0.5619, "step": 1020 }, { "epoch": 1.75, "learning_rate": 1.5073731565252532e-05, "loss": 0.57, "step": 1021 }, { "epoch": 1.75, "learning_rate": 1.5064135249967286e-05, "loss": 0.5425, "step": 1022 }, { "epoch": 1.76, "learning_rate": 1.505453265856581e-05, "loss": 0.5735, "step": 1023 }, { "epoch": 1.76, "learning_rate": 1.5044923802948853e-05, "loss": 0.5821, "step": 1024 }, { "epoch": 1.76, "learning_rate": 1.5035308695024914e-05, "loss": 0.562, "step": 1025 }, { "epoch": 1.76, "learning_rate": 1.5025687346710262e-05, "loss": 0.5613, "step": 1026 }, { "epoch": 1.76, "learning_rate": 1.501605976992888e-05, "loss": 0.58, "step": 1027 }, { "epoch": 1.76, "learning_rate": 1.5006425976612477e-05, "loss": 0.5703, "step": 1028 }, { "epoch": 1.77, "learning_rate": 1.4996785978700472e-05, "loss": 0.5628, "step": 1029 }, { "epoch": 1.77, "learning_rate": 1.4987139788139962e-05, "loss": 0.5746, "step": 1030 }, { "epoch": 1.77, "learning_rate": 1.497748741688573e-05, "loss": 0.5294, "step": 1031 }, { "epoch": 1.77, "learning_rate": 1.496782887690021e-05, "loss": 0.5547, "step": 1032 }, { "epoch": 1.77, "learning_rate": 1.4958164180153488e-05, "loss": 0.5495, "step": 1033 }, { "epoch": 1.78, "learning_rate": 1.4948493338623276e-05, "loss": 0.5392, "step": 1034 }, { "epoch": 1.78, "learning_rate": 1.4938816364294903e-05, "loss": 0.5499, "step": 1035 }, { "epoch": 1.78, "learning_rate": 1.4929133269161298e-05, "loss": 0.5709, "step": 1036 }, { "epoch": 1.78, "learning_rate": 1.4919444065222979e-05, "loss": 0.552, "step": 1037 }, { "epoch": 1.78, "learning_rate": 1.4909748764488025e-05, "loss": 0.5468, "step": 1038 }, { "epoch": 1.78, "learning_rate": 1.4900047378972083e-05, "loss": 0.5491, "step": 1039 }, { "epoch": 1.79, "learning_rate": 1.4890339920698334e-05, "loss": 0.5228, "step": 1040 }, { "epoch": 1.79, "learning_rate": 1.4880626401697487e-05, "loss": 0.5298, "step": 1041 }, { "epoch": 1.79, "learning_rate": 1.487090683400776e-05, "loss": 0.55, "step": 1042 }, { "epoch": 1.79, "learning_rate": 1.4861181229674877e-05, "loss": 0.5487, "step": 1043 }, { "epoch": 1.79, "learning_rate": 1.4851449600752027e-05, "loss": 0.5415, "step": 1044 }, { "epoch": 1.79, "learning_rate": 1.4841711959299874e-05, "loss": 0.5502, "step": 1045 }, { "epoch": 1.8, "learning_rate": 1.4831968317386537e-05, "loss": 0.5527, "step": 1046 }, { "epoch": 1.8, "learning_rate": 1.4822218687087571e-05, "loss": 0.5499, "step": 1047 }, { "epoch": 1.8, "learning_rate": 1.4812463080485944e-05, "loss": 0.5675, "step": 1048 }, { "epoch": 1.8, "learning_rate": 1.4802701509672038e-05, "loss": 0.5263, "step": 1049 }, { "epoch": 1.8, "learning_rate": 1.479293398674363e-05, "loss": 0.5574, "step": 1050 }, { "epoch": 1.8, "learning_rate": 1.4783160523805859e-05, "loss": 0.5672, "step": 1051 }, { "epoch": 1.81, "learning_rate": 1.4773381132971243e-05, "loss": 0.5266, "step": 1052 }, { "epoch": 1.81, "learning_rate": 1.4763595826359636e-05, "loss": 0.5381, "step": 1053 }, { "epoch": 1.81, "learning_rate": 1.4753804616098224e-05, "loss": 0.5322, "step": 1054 }, { "epoch": 1.81, "learning_rate": 1.4744007514321511e-05, "loss": 0.5605, "step": 1055 }, { "epoch": 1.81, "learning_rate": 1.4734204533171312e-05, "loss": 0.5386, "step": 1056 }, { "epoch": 1.81, "learning_rate": 1.472439568479671e-05, "loss": 0.5363, "step": 1057 }, { "epoch": 1.82, "learning_rate": 1.4714580981354077e-05, "loss": 0.5453, "step": 1058 }, { "epoch": 1.82, "learning_rate": 1.4704760435007029e-05, "loss": 0.5374, "step": 1059 }, { "epoch": 1.82, "learning_rate": 1.4694934057926427e-05, "loss": 0.5364, "step": 1060 }, { "epoch": 1.82, "learning_rate": 1.468510186229036e-05, "loss": 0.5332, "step": 1061 }, { "epoch": 1.82, "learning_rate": 1.4675263860284128e-05, "loss": 0.5272, "step": 1062 }, { "epoch": 1.82, "learning_rate": 1.4665420064100226e-05, "loss": 0.5501, "step": 1063 }, { "epoch": 1.83, "learning_rate": 1.4655570485938332e-05, "loss": 0.5296, "step": 1064 }, { "epoch": 1.83, "learning_rate": 1.4645715138005281e-05, "loss": 0.5402, "step": 1065 }, { "epoch": 1.83, "learning_rate": 1.4635854032515073e-05, "loss": 0.533, "step": 1066 }, { "epoch": 1.83, "learning_rate": 1.462598718168883e-05, "loss": 0.5325, "step": 1067 }, { "epoch": 1.83, "learning_rate": 1.4616114597754805e-05, "loss": 0.5224, "step": 1068 }, { "epoch": 1.84, "learning_rate": 1.4606236292948347e-05, "loss": 0.5088, "step": 1069 }, { "epoch": 1.84, "learning_rate": 1.4596352279511903e-05, "loss": 0.5375, "step": 1070 }, { "epoch": 1.84, "learning_rate": 1.4586462569694988e-05, "loss": 0.5488, "step": 1071 }, { "epoch": 1.84, "learning_rate": 1.4576567175754183e-05, "loss": 0.4931, "step": 1072 }, { "epoch": 1.84, "learning_rate": 1.4566666109953109e-05, "loss": 0.5345, "step": 1073 }, { "epoch": 1.84, "learning_rate": 1.4556759384562418e-05, "loss": 0.5331, "step": 1074 }, { "epoch": 1.85, "learning_rate": 1.4546847011859775e-05, "loss": 0.5349, "step": 1075 }, { "epoch": 1.85, "learning_rate": 1.4536929004129845e-05, "loss": 0.5249, "step": 1076 }, { "epoch": 1.85, "learning_rate": 1.4527005373664274e-05, "loss": 0.565, "step": 1077 }, { "epoch": 1.85, "learning_rate": 1.4517076132761686e-05, "loss": 0.5179, "step": 1078 }, { "epoch": 1.85, "learning_rate": 1.450714129372764e-05, "loss": 0.5193, "step": 1079 }, { "epoch": 1.85, "learning_rate": 1.4497200868874653e-05, "loss": 0.5478, "step": 1080 }, { "epoch": 1.86, "learning_rate": 1.448725487052215e-05, "loss": 0.5166, "step": 1081 }, { "epoch": 1.86, "learning_rate": 1.4477303310996474e-05, "loss": 0.5303, "step": 1082 }, { "epoch": 1.86, "learning_rate": 1.4467346202630847e-05, "loss": 0.5136, "step": 1083 }, { "epoch": 1.86, "learning_rate": 1.4457383557765385e-05, "loss": 0.5438, "step": 1084 }, { "epoch": 1.86, "learning_rate": 1.4447415388747048e-05, "loss": 0.5231, "step": 1085 }, { "epoch": 1.86, "learning_rate": 1.4437441707929657e-05, "loss": 0.5005, "step": 1086 }, { "epoch": 1.87, "learning_rate": 1.4427462527673852e-05, "loss": 0.5049, "step": 1087 }, { "epoch": 1.87, "learning_rate": 1.4417477860347098e-05, "loss": 0.5286, "step": 1088 }, { "epoch": 1.87, "learning_rate": 1.4407487718323653e-05, "loss": 0.5332, "step": 1089 }, { "epoch": 1.87, "learning_rate": 1.4397492113984569e-05, "loss": 0.5249, "step": 1090 }, { "epoch": 1.87, "learning_rate": 1.4387491059717653e-05, "loss": 0.5574, "step": 1091 }, { "epoch": 1.87, "learning_rate": 1.4377484567917482e-05, "loss": 0.5086, "step": 1092 }, { "epoch": 1.88, "learning_rate": 1.4367472650985362e-05, "loss": 0.5394, "step": 1093 }, { "epoch": 1.88, "learning_rate": 1.4357455321329327e-05, "loss": 0.5267, "step": 1094 }, { "epoch": 1.88, "learning_rate": 1.4347432591364114e-05, "loss": 0.5353, "step": 1095 }, { "epoch": 1.88, "learning_rate": 1.433740447351116e-05, "loss": 0.5494, "step": 1096 }, { "epoch": 1.88, "learning_rate": 1.4327370980198579e-05, "loss": 0.5253, "step": 1097 }, { "epoch": 1.88, "learning_rate": 1.4317332123861133e-05, "loss": 0.5264, "step": 1098 }, { "epoch": 1.89, "learning_rate": 1.4307287916940249e-05, "loss": 0.5165, "step": 1099 }, { "epoch": 1.89, "learning_rate": 1.4297238371883974e-05, "loss": 0.5093, "step": 1100 }, { "epoch": 1.89, "learning_rate": 1.428718350114698e-05, "loss": 0.5376, "step": 1101 }, { "epoch": 1.89, "learning_rate": 1.4277123317190524e-05, "loss": 0.5436, "step": 1102 }, { "epoch": 1.89, "learning_rate": 1.4267057832482465e-05, "loss": 0.5185, "step": 1103 }, { "epoch": 1.9, "learning_rate": 1.425698705949722e-05, "loss": 0.5181, "step": 1104 }, { "epoch": 1.9, "learning_rate": 1.424691101071576e-05, "loss": 0.545, "step": 1105 }, { "epoch": 1.9, "learning_rate": 1.4236829698625602e-05, "loss": 0.5196, "step": 1106 }, { "epoch": 1.9, "learning_rate": 1.4226743135720782e-05, "loss": 0.5095, "step": 1107 }, { "epoch": 1.9, "learning_rate": 1.421665133450184e-05, "loss": 0.5295, "step": 1108 }, { "epoch": 1.9, "learning_rate": 1.4206554307475816e-05, "loss": 0.5035, "step": 1109 }, { "epoch": 1.91, "learning_rate": 1.4196452067156217e-05, "loss": 0.4914, "step": 1110 }, { "epoch": 1.91, "learning_rate": 1.4186344626063017e-05, "loss": 0.4748, "step": 1111 }, { "epoch": 1.91, "learning_rate": 1.4176231996722634e-05, "loss": 0.5127, "step": 1112 }, { "epoch": 1.91, "learning_rate": 1.4166114191667914e-05, "loss": 0.495, "step": 1113 }, { "epoch": 1.91, "learning_rate": 1.4155991223438123e-05, "loss": 0.505, "step": 1114 }, { "epoch": 1.91, "learning_rate": 1.4145863104578921e-05, "loss": 0.528, "step": 1115 }, { "epoch": 1.92, "learning_rate": 1.413572984764235e-05, "loss": 0.5056, "step": 1116 }, { "epoch": 1.92, "learning_rate": 1.412559146518683e-05, "loss": 0.5188, "step": 1117 }, { "epoch": 1.92, "learning_rate": 1.4115447969777114e-05, "loss": 0.5074, "step": 1118 }, { "epoch": 1.92, "learning_rate": 1.4105299373984312e-05, "loss": 0.5147, "step": 1119 }, { "epoch": 1.92, "learning_rate": 1.4095145690385843e-05, "loss": 0.5046, "step": 1120 }, { "epoch": 1.92, "learning_rate": 1.4084986931565433e-05, "loss": 0.5121, "step": 1121 }, { "epoch": 1.93, "learning_rate": 1.40748231101131e-05, "loss": 0.4931, "step": 1122 }, { "epoch": 1.93, "learning_rate": 1.406465423862514e-05, "loss": 0.5174, "step": 1123 }, { "epoch": 1.93, "learning_rate": 1.40544803297041e-05, "loss": 0.5003, "step": 1124 }, { "epoch": 1.93, "learning_rate": 1.404430139595877e-05, "loss": 0.5209, "step": 1125 }, { "epoch": 1.93, "learning_rate": 1.4034117450004178e-05, "loss": 0.5103, "step": 1126 }, { "epoch": 1.93, "learning_rate": 1.4023928504461552e-05, "loss": 0.4798, "step": 1127 }, { "epoch": 1.94, "learning_rate": 1.4013734571958319e-05, "loss": 0.4833, "step": 1128 }, { "epoch": 1.94, "learning_rate": 1.4003535665128092e-05, "loss": 0.4955, "step": 1129 }, { "epoch": 1.94, "learning_rate": 1.3993331796610643e-05, "loss": 0.4895, "step": 1130 }, { "epoch": 1.94, "learning_rate": 1.3983122979051897e-05, "loss": 0.4997, "step": 1131 }, { "epoch": 1.94, "learning_rate": 1.397290922510391e-05, "loss": 0.5259, "step": 1132 }, { "epoch": 1.95, "learning_rate": 1.3962690547424853e-05, "loss": 0.5084, "step": 1133 }, { "epoch": 1.95, "learning_rate": 1.3952466958679006e-05, "loss": 0.5115, "step": 1134 }, { "epoch": 1.95, "learning_rate": 1.3942238471536729e-05, "loss": 0.5128, "step": 1135 }, { "epoch": 1.95, "learning_rate": 1.3932005098674458e-05, "loss": 0.489, "step": 1136 }, { "epoch": 1.95, "learning_rate": 1.3921766852774678e-05, "loss": 0.491, "step": 1137 }, { "epoch": 1.95, "learning_rate": 1.3911523746525924e-05, "loss": 0.4955, "step": 1138 }, { "epoch": 1.96, "learning_rate": 1.3901275792622744e-05, "loss": 0.4971, "step": 1139 }, { "epoch": 1.96, "learning_rate": 1.3891023003765694e-05, "loss": 0.4976, "step": 1140 }, { "epoch": 1.96, "learning_rate": 1.3880765392661327e-05, "loss": 0.4723, "step": 1141 }, { "epoch": 1.96, "learning_rate": 1.3870502972022175e-05, "loss": 0.5067, "step": 1142 }, { "epoch": 1.96, "learning_rate": 1.3860235754566718e-05, "loss": 0.4962, "step": 1143 }, { "epoch": 1.96, "learning_rate": 1.3849963753019395e-05, "loss": 0.4745, "step": 1144 }, { "epoch": 1.97, "learning_rate": 1.3839686980110569e-05, "loss": 0.4997, "step": 1145 }, { "epoch": 1.97, "learning_rate": 1.3829405448576511e-05, "loss": 0.5017, "step": 1146 }, { "epoch": 1.97, "learning_rate": 1.3819119171159399e-05, "loss": 0.4894, "step": 1147 }, { "epoch": 1.97, "learning_rate": 1.3808828160607283e-05, "loss": 0.4846, "step": 1148 }, { "epoch": 1.97, "learning_rate": 1.3798532429674086e-05, "loss": 0.4855, "step": 1149 }, { "epoch": 1.97, "learning_rate": 1.3788231991119577e-05, "loss": 0.4756, "step": 1150 }, { "epoch": 1.98, "learning_rate": 1.3777926857709359e-05, "loss": 0.4745, "step": 1151 }, { "epoch": 1.98, "learning_rate": 1.3767617042214862e-05, "loss": 0.5069, "step": 1152 }, { "epoch": 1.98, "learning_rate": 1.3757302557413306e-05, "loss": 0.4897, "step": 1153 }, { "epoch": 1.98, "learning_rate": 1.3746983416087708e-05, "loss": 0.4813, "step": 1154 }, { "epoch": 1.98, "learning_rate": 1.3736659631026847e-05, "loss": 0.4629, "step": 1155 }, { "epoch": 1.98, "learning_rate": 1.3726331215025266e-05, "loss": 0.4763, "step": 1156 }, { "epoch": 1.99, "learning_rate": 1.371599818088324e-05, "loss": 0.4695, "step": 1157 }, { "epoch": 1.99, "learning_rate": 1.370566054140677e-05, "loss": 0.4894, "step": 1158 }, { "epoch": 1.99, "learning_rate": 1.369531830940757e-05, "loss": 0.494, "step": 1159 }, { "epoch": 1.99, "learning_rate": 1.3684971497703035e-05, "loss": 0.4713, "step": 1160 }, { "epoch": 1.99, "learning_rate": 1.3674620119116247e-05, "loss": 0.4794, "step": 1161 }, { "epoch": 1.99, "learning_rate": 1.3664264186475935e-05, "loss": 0.4648, "step": 1162 }, { "epoch": 2.0, "learning_rate": 1.3653903712616486e-05, "loss": 0.4807, "step": 1163 }, { "epoch": 2.0, "learning_rate": 1.3643538710377901e-05, "loss": 0.4998, "step": 1164 }, { "epoch": 2.0, "learning_rate": 1.3633169192605806e-05, "loss": 0.4694, "step": 1165 }, { "epoch": 2.0, "learning_rate": 1.3622795172151417e-05, "loss": 0.3211, "step": 1166 }, { "epoch": 2.0, "learning_rate": 1.3612416661871532e-05, "loss": 0.3106, "step": 1167 }, { "epoch": 2.01, "learning_rate": 1.3602033674628508e-05, "loss": 0.3171, "step": 1168 }, { "epoch": 2.01, "learning_rate": 1.3591646223290256e-05, "loss": 0.317, "step": 1169 }, { "epoch": 2.01, "learning_rate": 1.3581254320730217e-05, "loss": 0.3259, "step": 1170 }, { "epoch": 2.01, "learning_rate": 1.3570857979827357e-05, "loss": 0.313, "step": 1171 }, { "epoch": 2.01, "learning_rate": 1.3560457213466122e-05, "loss": 0.3114, "step": 1172 }, { "epoch": 2.01, "learning_rate": 1.3550052034536467e-05, "loss": 0.3101, "step": 1173 }, { "epoch": 2.02, "learning_rate": 1.3539642455933803e-05, "loss": 0.3181, "step": 1174 }, { "epoch": 2.02, "learning_rate": 1.352922849055899e-05, "loss": 0.3107, "step": 1175 }, { "epoch": 2.02, "learning_rate": 1.351881015131833e-05, "loss": 0.3223, "step": 1176 }, { "epoch": 2.02, "learning_rate": 1.3508387451123555e-05, "loss": 0.3056, "step": 1177 }, { "epoch": 2.02, "learning_rate": 1.349796040289178e-05, "loss": 0.3054, "step": 1178 }, { "epoch": 2.02, "learning_rate": 1.3487529019545523e-05, "loss": 0.3092, "step": 1179 }, { "epoch": 2.03, "learning_rate": 1.3477093314012677e-05, "loss": 0.3171, "step": 1180 }, { "epoch": 2.03, "learning_rate": 1.3466653299226482e-05, "loss": 0.3126, "step": 1181 }, { "epoch": 2.03, "learning_rate": 1.3456208988125528e-05, "loss": 0.3153, "step": 1182 }, { "epoch": 2.03, "learning_rate": 1.3445760393653722e-05, "loss": 0.3103, "step": 1183 }, { "epoch": 2.03, "learning_rate": 1.3435307528760283e-05, "loss": 0.3145, "step": 1184 }, { "epoch": 2.03, "learning_rate": 1.3424850406399719e-05, "loss": 0.3083, "step": 1185 }, { "epoch": 2.04, "learning_rate": 1.341438903953182e-05, "loss": 0.3299, "step": 1186 }, { "epoch": 2.04, "learning_rate": 1.3403923441121635e-05, "loss": 0.3141, "step": 1187 }, { "epoch": 2.04, "learning_rate": 1.3393453624139454e-05, "loss": 0.3147, "step": 1188 }, { "epoch": 2.04, "learning_rate": 1.3382979601560796e-05, "loss": 0.3186, "step": 1189 }, { "epoch": 2.04, "learning_rate": 1.3372501386366399e-05, "loss": 0.3004, "step": 1190 }, { "epoch": 2.04, "learning_rate": 1.3362018991542183e-05, "loss": 0.3086, "step": 1191 }, { "epoch": 2.05, "learning_rate": 1.3351532430079258e-05, "loss": 0.2992, "step": 1192 }, { "epoch": 2.05, "learning_rate": 1.3341041714973901e-05, "loss": 0.305, "step": 1193 }, { "epoch": 2.05, "learning_rate": 1.3330546859227524e-05, "loss": 0.3204, "step": 1194 }, { "epoch": 2.05, "learning_rate": 1.3320047875846682e-05, "loss": 0.3218, "step": 1195 }, { "epoch": 2.05, "learning_rate": 1.3309544777843046e-05, "loss": 0.3053, "step": 1196 }, { "epoch": 2.05, "learning_rate": 1.3299037578233376e-05, "loss": 0.3132, "step": 1197 }, { "epoch": 2.06, "learning_rate": 1.3288526290039524e-05, "loss": 0.314, "step": 1198 }, { "epoch": 2.06, "learning_rate": 1.3278010926288405e-05, "loss": 0.3097, "step": 1199 }, { "epoch": 2.06, "learning_rate": 1.3267491500011987e-05, "loss": 0.3137, "step": 1200 }, { "epoch": 2.06, "learning_rate": 1.3256968024247271e-05, "loss": 0.3067, "step": 1201 }, { "epoch": 2.06, "learning_rate": 1.324644051203628e-05, "loss": 0.3221, "step": 1202 }, { "epoch": 2.07, "learning_rate": 1.3235908976426036e-05, "loss": 0.2982, "step": 1203 }, { "epoch": 2.07, "learning_rate": 1.3225373430468547e-05, "loss": 0.315, "step": 1204 }, { "epoch": 2.07, "learning_rate": 1.3214833887220794e-05, "loss": 0.2978, "step": 1205 }, { "epoch": 2.07, "learning_rate": 1.3204290359744716e-05, "loss": 0.3127, "step": 1206 }, { "epoch": 2.07, "learning_rate": 1.3193742861107174e-05, "loss": 0.3022, "step": 1207 }, { "epoch": 2.07, "learning_rate": 1.318319140437997e-05, "loss": 0.3115, "step": 1208 }, { "epoch": 2.08, "learning_rate": 1.3172636002639793e-05, "loss": 0.3018, "step": 1209 }, { "epoch": 2.08, "learning_rate": 1.316207666896824e-05, "loss": 0.3109, "step": 1210 }, { "epoch": 2.08, "learning_rate": 1.3151513416451766e-05, "loss": 0.2866, "step": 1211 }, { "epoch": 2.08, "learning_rate": 1.3140946258181693e-05, "loss": 0.3125, "step": 1212 }, { "epoch": 2.08, "learning_rate": 1.3130375207254172e-05, "loss": 0.3252, "step": 1213 }, { "epoch": 2.08, "learning_rate": 1.3119800276770187e-05, "loss": 0.3147, "step": 1214 }, { "epoch": 2.09, "learning_rate": 1.310922147983553e-05, "loss": 0.3019, "step": 1215 }, { "epoch": 2.09, "learning_rate": 1.309863882956078e-05, "loss": 0.3087, "step": 1216 }, { "epoch": 2.09, "learning_rate": 1.3088052339061296e-05, "loss": 0.2997, "step": 1217 }, { "epoch": 2.09, "learning_rate": 1.3077462021457197e-05, "loss": 0.3053, "step": 1218 }, { "epoch": 2.09, "learning_rate": 1.3066867889873341e-05, "loss": 0.3149, "step": 1219 }, { "epoch": 2.09, "learning_rate": 1.3056269957439309e-05, "loss": 0.3059, "step": 1220 }, { "epoch": 2.1, "learning_rate": 1.3045668237289406e-05, "loss": 0.2991, "step": 1221 }, { "epoch": 2.1, "learning_rate": 1.3035062742562618e-05, "loss": 0.3116, "step": 1222 }, { "epoch": 2.1, "learning_rate": 1.3024453486402617e-05, "loss": 0.3057, "step": 1223 }, { "epoch": 2.1, "learning_rate": 1.3013840481957729e-05, "loss": 0.3072, "step": 1224 }, { "epoch": 2.1, "learning_rate": 1.3003223742380935e-05, "loss": 0.298, "step": 1225 }, { "epoch": 2.1, "learning_rate": 1.2992603280829838e-05, "loss": 0.2978, "step": 1226 }, { "epoch": 2.11, "learning_rate": 1.2981979110466655e-05, "loss": 0.2959, "step": 1227 }, { "epoch": 2.11, "learning_rate": 1.2971351244458203e-05, "loss": 0.2938, "step": 1228 }, { "epoch": 2.11, "learning_rate": 1.2960719695975875e-05, "loss": 0.2972, "step": 1229 }, { "epoch": 2.11, "learning_rate": 1.2950084478195626e-05, "loss": 0.3014, "step": 1230 }, { "epoch": 2.11, "learning_rate": 1.2939445604297965e-05, "loss": 0.3114, "step": 1231 }, { "epoch": 2.12, "learning_rate": 1.2928803087467928e-05, "loss": 0.2881, "step": 1232 }, { "epoch": 2.12, "learning_rate": 1.2918156940895066e-05, "loss": 0.3018, "step": 1233 }, { "epoch": 2.12, "learning_rate": 1.290750717777343e-05, "loss": 0.3024, "step": 1234 }, { "epoch": 2.12, "learning_rate": 1.2896853811301553e-05, "loss": 0.2966, "step": 1235 }, { "epoch": 2.12, "learning_rate": 1.2886196854682428e-05, "loss": 0.3145, "step": 1236 }, { "epoch": 2.12, "learning_rate": 1.2875536321123508e-05, "loss": 0.3076, "step": 1237 }, { "epoch": 2.13, "learning_rate": 1.2864872223836667e-05, "loss": 0.3023, "step": 1238 }, { "epoch": 2.13, "learning_rate": 1.2854204576038206e-05, "loss": 0.2997, "step": 1239 }, { "epoch": 2.13, "learning_rate": 1.2843533390948821e-05, "loss": 0.3074, "step": 1240 }, { "epoch": 2.13, "learning_rate": 1.2832858681793594e-05, "loss": 0.2958, "step": 1241 }, { "epoch": 2.13, "learning_rate": 1.2822180461801969e-05, "loss": 0.3018, "step": 1242 }, { "epoch": 2.13, "learning_rate": 1.281149874420775e-05, "loss": 0.2937, "step": 1243 }, { "epoch": 2.14, "learning_rate": 1.2800813542249073e-05, "loss": 0.2992, "step": 1244 }, { "epoch": 2.14, "learning_rate": 1.2790124869168382e-05, "loss": 0.3166, "step": 1245 }, { "epoch": 2.14, "learning_rate": 1.2779432738212437e-05, "loss": 0.3054, "step": 1246 }, { "epoch": 2.14, "learning_rate": 1.2768737162632274e-05, "loss": 0.3127, "step": 1247 }, { "epoch": 2.14, "learning_rate": 1.2758038155683205e-05, "loss": 0.3046, "step": 1248 }, { "epoch": 2.14, "learning_rate": 1.2747335730624792e-05, "loss": 0.3042, "step": 1249 }, { "epoch": 2.15, "learning_rate": 1.2736629900720832e-05, "loss": 0.2961, "step": 1250 }, { "epoch": 2.15, "learning_rate": 1.2725920679239337e-05, "loss": 0.2947, "step": 1251 }, { "epoch": 2.15, "learning_rate": 1.271520807945253e-05, "loss": 0.3006, "step": 1252 }, { "epoch": 2.15, "learning_rate": 1.2704492114636822e-05, "loss": 0.3104, "step": 1253 }, { "epoch": 2.15, "learning_rate": 1.2693772798072785e-05, "loss": 0.3046, "step": 1254 }, { "epoch": 2.15, "learning_rate": 1.2683050143045153e-05, "loss": 0.3078, "step": 1255 }, { "epoch": 2.16, "learning_rate": 1.2672324162842797e-05, "loss": 0.309, "step": 1256 }, { "epoch": 2.16, "learning_rate": 1.2661594870758702e-05, "loss": 0.2923, "step": 1257 }, { "epoch": 2.16, "learning_rate": 1.2650862280089966e-05, "loss": 0.3105, "step": 1258 }, { "epoch": 2.16, "learning_rate": 1.264012640413777e-05, "loss": 0.2998, "step": 1259 }, { "epoch": 2.16, "learning_rate": 1.2629387256207366e-05, "loss": 0.2978, "step": 1260 }, { "epoch": 2.16, "learning_rate": 1.2618644849608068e-05, "loss": 0.312, "step": 1261 }, { "epoch": 2.17, "learning_rate": 1.260789919765322e-05, "loss": 0.3231, "step": 1262 }, { "epoch": 2.17, "learning_rate": 1.2597150313660194e-05, "loss": 0.3009, "step": 1263 }, { "epoch": 2.17, "learning_rate": 1.2586398210950359e-05, "loss": 0.3046, "step": 1264 }, { "epoch": 2.17, "learning_rate": 1.2575642902849085e-05, "loss": 0.3016, "step": 1265 }, { "epoch": 2.17, "learning_rate": 1.2564884402685711e-05, "loss": 0.3131, "step": 1266 }, { "epoch": 2.18, "learning_rate": 1.255412272379352e-05, "loss": 0.3003, "step": 1267 }, { "epoch": 2.18, "learning_rate": 1.254335787950975e-05, "loss": 0.2821, "step": 1268 }, { "epoch": 2.18, "learning_rate": 1.2532589883175561e-05, "loss": 0.2855, "step": 1269 }, { "epoch": 2.18, "learning_rate": 1.2521818748136005e-05, "loss": 0.3136, "step": 1270 }, { "epoch": 2.18, "learning_rate": 1.2511044487740041e-05, "loss": 0.3026, "step": 1271 }, { "epoch": 2.18, "learning_rate": 1.250026711534049e-05, "loss": 0.3039, "step": 1272 }, { "epoch": 2.19, "learning_rate": 1.2489486644294034e-05, "loss": 0.2977, "step": 1273 }, { "epoch": 2.19, "learning_rate": 1.2478703087961193e-05, "loss": 0.2904, "step": 1274 }, { "epoch": 2.19, "learning_rate": 1.2467916459706316e-05, "loss": 0.3179, "step": 1275 }, { "epoch": 2.19, "learning_rate": 1.2457126772897555e-05, "loss": 0.3001, "step": 1276 }, { "epoch": 2.19, "learning_rate": 1.2446334040906849e-05, "loss": 0.3004, "step": 1277 }, { "epoch": 2.19, "learning_rate": 1.2435538277109919e-05, "loss": 0.3111, "step": 1278 }, { "epoch": 2.2, "learning_rate": 1.2424739494886245e-05, "loss": 0.294, "step": 1279 }, { "epoch": 2.2, "learning_rate": 1.241393770761903e-05, "loss": 0.303, "step": 1280 }, { "epoch": 2.2, "learning_rate": 1.240313292869522e-05, "loss": 0.2979, "step": 1281 }, { "epoch": 2.2, "learning_rate": 1.239232517150546e-05, "loss": 0.28, "step": 1282 }, { "epoch": 2.2, "learning_rate": 1.2381514449444094e-05, "loss": 0.2892, "step": 1283 }, { "epoch": 2.2, "learning_rate": 1.237070077590913e-05, "loss": 0.2976, "step": 1284 }, { "epoch": 2.21, "learning_rate": 1.235988416430224e-05, "loss": 0.2921, "step": 1285 }, { "epoch": 2.21, "learning_rate": 1.2349064628028733e-05, "loss": 0.3062, "step": 1286 }, { "epoch": 2.21, "learning_rate": 1.2338242180497545e-05, "loss": 0.3073, "step": 1287 }, { "epoch": 2.21, "learning_rate": 1.2327416835121228e-05, "loss": 0.3039, "step": 1288 }, { "epoch": 2.21, "learning_rate": 1.231658860531591e-05, "loss": 0.2986, "step": 1289 }, { "epoch": 2.21, "learning_rate": 1.23057575045013e-05, "loss": 0.3117, "step": 1290 }, { "epoch": 2.22, "learning_rate": 1.2294923546100672e-05, "loss": 0.3188, "step": 1291 }, { "epoch": 2.22, "learning_rate": 1.2284086743540837e-05, "loss": 0.2993, "step": 1292 }, { "epoch": 2.22, "learning_rate": 1.2273247110252124e-05, "loss": 0.2914, "step": 1293 }, { "epoch": 2.22, "learning_rate": 1.226240465966838e-05, "loss": 0.2934, "step": 1294 }, { "epoch": 2.22, "learning_rate": 1.2251559405226943e-05, "loss": 0.3003, "step": 1295 }, { "epoch": 2.22, "learning_rate": 1.2240711360368614e-05, "loss": 0.3033, "step": 1296 }, { "epoch": 2.23, "learning_rate": 1.2229860538537666e-05, "loss": 0.2899, "step": 1297 }, { "epoch": 2.23, "learning_rate": 1.221900695318181e-05, "loss": 0.2983, "step": 1298 }, { "epoch": 2.23, "learning_rate": 1.2208150617752175e-05, "loss": 0.2955, "step": 1299 }, { "epoch": 2.23, "learning_rate": 1.2197291545703307e-05, "loss": 0.3081, "step": 1300 }, { "epoch": 2.23, "learning_rate": 1.2186429750493143e-05, "loss": 0.3037, "step": 1301 }, { "epoch": 2.24, "learning_rate": 1.2175565245582983e-05, "loss": 0.2972, "step": 1302 }, { "epoch": 2.24, "learning_rate": 1.2164698044437503e-05, "loss": 0.2891, "step": 1303 }, { "epoch": 2.24, "learning_rate": 1.2153828160524707e-05, "loss": 0.303, "step": 1304 }, { "epoch": 2.24, "learning_rate": 1.2142955607315929e-05, "loss": 0.2855, "step": 1305 }, { "epoch": 2.24, "learning_rate": 1.2132080398285812e-05, "loss": 0.297, "step": 1306 }, { "epoch": 2.24, "learning_rate": 1.2121202546912292e-05, "loss": 0.3012, "step": 1307 }, { "epoch": 2.25, "learning_rate": 1.211032206667657e-05, "loss": 0.3097, "step": 1308 }, { "epoch": 2.25, "learning_rate": 1.2099438971063115e-05, "loss": 0.2921, "step": 1309 }, { "epoch": 2.25, "learning_rate": 1.2088553273559637e-05, "loss": 0.2983, "step": 1310 }, { "epoch": 2.25, "learning_rate": 1.2077664987657064e-05, "loss": 0.3031, "step": 1311 }, { "epoch": 2.25, "learning_rate": 1.206677412684953e-05, "loss": 0.3, "step": 1312 }, { "epoch": 2.25, "learning_rate": 1.2055880704634377e-05, "loss": 0.3017, "step": 1313 }, { "epoch": 2.26, "learning_rate": 1.2044984734512107e-05, "loss": 0.2902, "step": 1314 }, { "epoch": 2.26, "learning_rate": 1.203408622998638e-05, "loss": 0.2935, "step": 1315 }, { "epoch": 2.26, "learning_rate": 1.2023185204563999e-05, "loss": 0.3005, "step": 1316 }, { "epoch": 2.26, "learning_rate": 1.2012281671754898e-05, "loss": 0.2933, "step": 1317 }, { "epoch": 2.26, "learning_rate": 1.2001375645072105e-05, "loss": 0.3006, "step": 1318 }, { "epoch": 2.26, "learning_rate": 1.1990467138031747e-05, "loss": 0.2949, "step": 1319 }, { "epoch": 2.27, "learning_rate": 1.197955616415303e-05, "loss": 0.2844, "step": 1320 }, { "epoch": 2.27, "learning_rate": 1.1968642736958208e-05, "loss": 0.2873, "step": 1321 }, { "epoch": 2.27, "learning_rate": 1.1957726869972577e-05, "loss": 0.3084, "step": 1322 }, { "epoch": 2.27, "learning_rate": 1.1946808576724465e-05, "loss": 0.2948, "step": 1323 }, { "epoch": 2.27, "learning_rate": 1.193588787074519e-05, "loss": 0.2819, "step": 1324 }, { "epoch": 2.27, "learning_rate": 1.1924964765569081e-05, "loss": 0.2826, "step": 1325 }, { "epoch": 2.28, "learning_rate": 1.1914039274733422e-05, "loss": 0.2993, "step": 1326 }, { "epoch": 2.28, "learning_rate": 1.1903111411778465e-05, "loss": 0.2965, "step": 1327 }, { "epoch": 2.28, "learning_rate": 1.18921811902474e-05, "loss": 0.3017, "step": 1328 }, { "epoch": 2.28, "learning_rate": 1.188124862368634e-05, "loss": 0.3025, "step": 1329 }, { "epoch": 2.28, "learning_rate": 1.1870313725644296e-05, "loss": 0.2986, "step": 1330 }, { "epoch": 2.28, "learning_rate": 1.185937650967318e-05, "loss": 0.3006, "step": 1331 }, { "epoch": 2.29, "learning_rate": 1.1848436989327769e-05, "loss": 0.2941, "step": 1332 }, { "epoch": 2.29, "learning_rate": 1.1837495178165706e-05, "loss": 0.2906, "step": 1333 }, { "epoch": 2.29, "learning_rate": 1.1826551089747456e-05, "loss": 0.2932, "step": 1334 }, { "epoch": 2.29, "learning_rate": 1.1815604737636321e-05, "loss": 0.2877, "step": 1335 }, { "epoch": 2.29, "learning_rate": 1.1804656135398405e-05, "loss": 0.302, "step": 1336 }, { "epoch": 2.3, "learning_rate": 1.1793705296602595e-05, "loss": 0.2916, "step": 1337 }, { "epoch": 2.3, "learning_rate": 1.1782752234820557e-05, "loss": 0.2876, "step": 1338 }, { "epoch": 2.3, "learning_rate": 1.1771796963626711e-05, "loss": 0.2857, "step": 1339 }, { "epoch": 2.3, "learning_rate": 1.1760839496598208e-05, "loss": 0.2902, "step": 1340 }, { "epoch": 2.3, "learning_rate": 1.174987984731493e-05, "loss": 0.3027, "step": 1341 }, { "epoch": 2.3, "learning_rate": 1.1738918029359454e-05, "loss": 0.2852, "step": 1342 }, { "epoch": 2.31, "learning_rate": 1.1727954056317055e-05, "loss": 0.2859, "step": 1343 }, { "epoch": 2.31, "learning_rate": 1.171698794177567e-05, "loss": 0.2876, "step": 1344 }, { "epoch": 2.31, "learning_rate": 1.1706019699325898e-05, "loss": 0.2823, "step": 1345 }, { "epoch": 2.31, "learning_rate": 1.1695049342560969e-05, "loss": 0.2921, "step": 1346 }, { "epoch": 2.31, "learning_rate": 1.1684076885076732e-05, "loss": 0.2903, "step": 1347 }, { "epoch": 2.31, "learning_rate": 1.1673102340471645e-05, "loss": 0.298, "step": 1348 }, { "epoch": 2.32, "learning_rate": 1.166212572234675e-05, "loss": 0.2994, "step": 1349 }, { "epoch": 2.32, "learning_rate": 1.1651147044305657e-05, "loss": 0.2973, "step": 1350 }, { "epoch": 2.32, "learning_rate": 1.1640166319954535e-05, "loss": 0.2967, "step": 1351 }, { "epoch": 2.32, "learning_rate": 1.1629183562902087e-05, "loss": 0.2852, "step": 1352 }, { "epoch": 2.32, "learning_rate": 1.1618198786759524e-05, "loss": 0.3014, "step": 1353 }, { "epoch": 2.32, "learning_rate": 1.1607212005140576e-05, "loss": 0.2905, "step": 1354 }, { "epoch": 2.33, "learning_rate": 1.159622323166145e-05, "loss": 0.2911, "step": 1355 }, { "epoch": 2.33, "learning_rate": 1.1585232479940816e-05, "loss": 0.2933, "step": 1356 }, { "epoch": 2.33, "learning_rate": 1.157423976359981e-05, "loss": 0.2999, "step": 1357 }, { "epoch": 2.33, "learning_rate": 1.1563245096261994e-05, "loss": 0.2825, "step": 1358 }, { "epoch": 2.33, "learning_rate": 1.1552248491553344e-05, "loss": 0.2916, "step": 1359 }, { "epoch": 2.33, "learning_rate": 1.1541249963102246e-05, "loss": 0.2885, "step": 1360 }, { "epoch": 2.34, "learning_rate": 1.1530249524539467e-05, "loss": 0.2928, "step": 1361 }, { "epoch": 2.34, "learning_rate": 1.1519247189498136e-05, "loss": 0.2921, "step": 1362 }, { "epoch": 2.34, "learning_rate": 1.1508242971613741e-05, "loss": 0.2909, "step": 1363 }, { "epoch": 2.34, "learning_rate": 1.1497236884524095e-05, "loss": 0.2924, "step": 1364 }, { "epoch": 2.34, "learning_rate": 1.1486228941869335e-05, "loss": 0.2824, "step": 1365 }, { "epoch": 2.35, "learning_rate": 1.1475219157291892e-05, "loss": 0.2799, "step": 1366 }, { "epoch": 2.35, "learning_rate": 1.1464207544436483e-05, "loss": 0.2881, "step": 1367 }, { "epoch": 2.35, "learning_rate": 1.1453194116950092e-05, "loss": 0.2808, "step": 1368 }, { "epoch": 2.35, "learning_rate": 1.1442178888481947e-05, "loss": 0.2936, "step": 1369 }, { "epoch": 2.35, "learning_rate": 1.1431161872683512e-05, "loss": 0.2785, "step": 1370 }, { "epoch": 2.35, "learning_rate": 1.1420143083208463e-05, "loss": 0.2857, "step": 1371 }, { "epoch": 2.36, "learning_rate": 1.140912253371268e-05, "loss": 0.2992, "step": 1372 }, { "epoch": 2.36, "learning_rate": 1.1398100237854218e-05, "loss": 0.2914, "step": 1373 }, { "epoch": 2.36, "learning_rate": 1.1387076209293299e-05, "loss": 0.2842, "step": 1374 }, { "epoch": 2.36, "learning_rate": 1.137605046169229e-05, "loss": 0.293, "step": 1375 }, { "epoch": 2.36, "learning_rate": 1.1365023008715693e-05, "loss": 0.2997, "step": 1376 }, { "epoch": 2.36, "learning_rate": 1.1353993864030121e-05, "loss": 0.2961, "step": 1377 }, { "epoch": 2.37, "learning_rate": 1.134296304130428e-05, "loss": 0.2974, "step": 1378 }, { "epoch": 2.37, "learning_rate": 1.1331930554208966e-05, "loss": 0.2807, "step": 1379 }, { "epoch": 2.37, "learning_rate": 1.1320896416417026e-05, "loss": 0.291, "step": 1380 }, { "epoch": 2.37, "learning_rate": 1.1309860641603357e-05, "loss": 0.2898, "step": 1381 }, { "epoch": 2.37, "learning_rate": 1.1298823243444888e-05, "loss": 0.2831, "step": 1382 }, { "epoch": 2.37, "learning_rate": 1.1287784235620554e-05, "loss": 0.2868, "step": 1383 }, { "epoch": 2.38, "learning_rate": 1.1276743631811295e-05, "loss": 0.2845, "step": 1384 }, { "epoch": 2.38, "learning_rate": 1.1265701445700015e-05, "loss": 0.2778, "step": 1385 }, { "epoch": 2.38, "learning_rate": 1.1254657690971587e-05, "loss": 0.292, "step": 1386 }, { "epoch": 2.38, "learning_rate": 1.1243612381312831e-05, "loss": 0.2877, "step": 1387 }, { "epoch": 2.38, "learning_rate": 1.123256553041249e-05, "loss": 0.2841, "step": 1388 }, { "epoch": 2.38, "learning_rate": 1.1221517151961211e-05, "loss": 0.2903, "step": 1389 }, { "epoch": 2.39, "learning_rate": 1.1210467259651552e-05, "loss": 0.2836, "step": 1390 }, { "epoch": 2.39, "learning_rate": 1.1199415867177926e-05, "loss": 0.2855, "step": 1391 }, { "epoch": 2.39, "learning_rate": 1.1188362988236615e-05, "loss": 0.2933, "step": 1392 }, { "epoch": 2.39, "learning_rate": 1.1177308636525745e-05, "loss": 0.2843, "step": 1393 }, { "epoch": 2.39, "learning_rate": 1.116625282574527e-05, "loss": 0.2884, "step": 1394 }, { "epoch": 2.39, "learning_rate": 1.1155195569596939e-05, "loss": 0.2858, "step": 1395 }, { "epoch": 2.4, "learning_rate": 1.1144136881784311e-05, "loss": 0.2947, "step": 1396 }, { "epoch": 2.4, "learning_rate": 1.1133076776012702e-05, "loss": 0.2748, "step": 1397 }, { "epoch": 2.4, "learning_rate": 1.112201526598919e-05, "loss": 0.2847, "step": 1398 }, { "epoch": 2.4, "learning_rate": 1.1110952365422605e-05, "loss": 0.2906, "step": 1399 }, { "epoch": 2.4, "learning_rate": 1.109988808802348e-05, "loss": 0.2898, "step": 1400 }, { "epoch": 2.41, "learning_rate": 1.1088822447504075e-05, "loss": 0.2785, "step": 1401 }, { "epoch": 2.41, "learning_rate": 1.1077755457578326e-05, "loss": 0.2909, "step": 1402 }, { "epoch": 2.41, "learning_rate": 1.1066687131961844e-05, "loss": 0.2915, "step": 1403 }, { "epoch": 2.41, "learning_rate": 1.1055617484371899e-05, "loss": 0.2894, "step": 1404 }, { "epoch": 2.41, "learning_rate": 1.1044546528527395e-05, "loss": 0.2984, "step": 1405 }, { "epoch": 2.41, "learning_rate": 1.1033474278148863e-05, "loss": 0.295, "step": 1406 }, { "epoch": 2.42, "learning_rate": 1.1022400746958428e-05, "loss": 0.2875, "step": 1407 }, { "epoch": 2.42, "learning_rate": 1.1011325948679812e-05, "loss": 0.2945, "step": 1408 }, { "epoch": 2.42, "learning_rate": 1.1000249897038307e-05, "loss": 0.2834, "step": 1409 }, { "epoch": 2.42, "learning_rate": 1.0989172605760752e-05, "loss": 0.275, "step": 1410 }, { "epoch": 2.42, "learning_rate": 1.0978094088575529e-05, "loss": 0.2877, "step": 1411 }, { "epoch": 2.42, "learning_rate": 1.0967014359212533e-05, "loss": 0.2731, "step": 1412 }, { "epoch": 2.43, "learning_rate": 1.0955933431403167e-05, "loss": 0.2883, "step": 1413 }, { "epoch": 2.43, "learning_rate": 1.0944851318880314e-05, "loss": 0.2853, "step": 1414 }, { "epoch": 2.43, "learning_rate": 1.0933768035378331e-05, "loss": 0.2883, "step": 1415 }, { "epoch": 2.43, "learning_rate": 1.092268359463302e-05, "loss": 0.2875, "step": 1416 }, { "epoch": 2.43, "learning_rate": 1.0911598010381626e-05, "loss": 0.2849, "step": 1417 }, { "epoch": 2.43, "learning_rate": 1.0900511296362802e-05, "loss": 0.2923, "step": 1418 }, { "epoch": 2.44, "learning_rate": 1.0889423466316602e-05, "loss": 0.2796, "step": 1419 }, { "epoch": 2.44, "learning_rate": 1.0878334533984468e-05, "loss": 0.2663, "step": 1420 }, { "epoch": 2.44, "learning_rate": 1.0867244513109207e-05, "loss": 0.2835, "step": 1421 }, { "epoch": 2.44, "learning_rate": 1.0856153417434976e-05, "loss": 0.2787, "step": 1422 }, { "epoch": 2.44, "learning_rate": 1.0845061260707257e-05, "loss": 0.2773, "step": 1423 }, { "epoch": 2.44, "learning_rate": 1.0833968056672854e-05, "loss": 0.2747, "step": 1424 }, { "epoch": 2.45, "learning_rate": 1.0822873819079873e-05, "loss": 0.2809, "step": 1425 }, { "epoch": 2.45, "learning_rate": 1.0811778561677687e-05, "loss": 0.2897, "step": 1426 }, { "epoch": 2.45, "learning_rate": 1.0800682298216942e-05, "loss": 0.2858, "step": 1427 }, { "epoch": 2.45, "learning_rate": 1.0789585042449534e-05, "loss": 0.2643, "step": 1428 }, { "epoch": 2.45, "learning_rate": 1.0778486808128579e-05, "loss": 0.2854, "step": 1429 }, { "epoch": 2.45, "learning_rate": 1.0767387609008413e-05, "loss": 0.2886, "step": 1430 }, { "epoch": 2.46, "learning_rate": 1.075628745884457e-05, "loss": 0.281, "step": 1431 }, { "epoch": 2.46, "learning_rate": 1.0745186371393752e-05, "loss": 0.2855, "step": 1432 }, { "epoch": 2.46, "learning_rate": 1.0734084360413833e-05, "loss": 0.3023, "step": 1433 }, { "epoch": 2.46, "learning_rate": 1.0722981439663829e-05, "loss": 0.2903, "step": 1434 }, { "epoch": 2.46, "learning_rate": 1.0711877622903878e-05, "loss": 0.2861, "step": 1435 }, { "epoch": 2.47, "learning_rate": 1.0700772923895235e-05, "loss": 0.2876, "step": 1436 }, { "epoch": 2.47, "learning_rate": 1.0689667356400244e-05, "loss": 0.2663, "step": 1437 }, { "epoch": 2.47, "learning_rate": 1.0678560934182331e-05, "loss": 0.2765, "step": 1438 }, { "epoch": 2.47, "learning_rate": 1.0667453671005976e-05, "loss": 0.287, "step": 1439 }, { "epoch": 2.47, "learning_rate": 1.0656345580636702e-05, "loss": 0.2769, "step": 1440 }, { "epoch": 2.47, "learning_rate": 1.0645236676841061e-05, "loss": 0.2695, "step": 1441 }, { "epoch": 2.48, "learning_rate": 1.0634126973386607e-05, "loss": 0.2725, "step": 1442 }, { "epoch": 2.48, "learning_rate": 1.0623016484041886e-05, "loss": 0.2868, "step": 1443 }, { "epoch": 2.48, "learning_rate": 1.0611905222576426e-05, "loss": 0.2863, "step": 1444 }, { "epoch": 2.48, "learning_rate": 1.0600793202760702e-05, "loss": 0.2889, "step": 1445 }, { "epoch": 2.48, "learning_rate": 1.0589680438366135e-05, "loss": 0.2883, "step": 1446 }, { "epoch": 2.48, "learning_rate": 1.0578566943165067e-05, "loss": 0.2791, "step": 1447 }, { "epoch": 2.49, "learning_rate": 1.0567452730930743e-05, "loss": 0.2782, "step": 1448 }, { "epoch": 2.49, "learning_rate": 1.0556337815437299e-05, "loss": 0.2939, "step": 1449 }, { "epoch": 2.49, "learning_rate": 1.0545222210459744e-05, "loss": 0.283, "step": 1450 }, { "epoch": 2.49, "learning_rate": 1.0534105929773941e-05, "loss": 0.2673, "step": 1451 }, { "epoch": 2.49, "learning_rate": 1.0522988987156586e-05, "loss": 0.2806, "step": 1452 }, { "epoch": 2.49, "learning_rate": 1.0511871396385204e-05, "loss": 0.287, "step": 1453 }, { "epoch": 2.5, "learning_rate": 1.0500753171238116e-05, "loss": 0.2791, "step": 1454 }, { "epoch": 2.5, "learning_rate": 1.048963432549443e-05, "loss": 0.2873, "step": 1455 }, { "epoch": 2.5, "learning_rate": 1.0478514872934022e-05, "loss": 0.2799, "step": 1456 }, { "epoch": 2.5, "learning_rate": 1.046739482733753e-05, "loss": 0.2829, "step": 1457 }, { "epoch": 2.5, "learning_rate": 1.0456274202486313e-05, "loss": 0.264, "step": 1458 }, { "epoch": 2.5, "learning_rate": 1.0445153012162457e-05, "loss": 0.2791, "step": 1459 }, { "epoch": 2.51, "learning_rate": 1.0434031270148743e-05, "loss": 0.2634, "step": 1460 }, { "epoch": 2.51, "learning_rate": 1.0422908990228644e-05, "loss": 0.2832, "step": 1461 }, { "epoch": 2.51, "learning_rate": 1.0411786186186293e-05, "loss": 0.2783, "step": 1462 }, { "epoch": 2.51, "learning_rate": 1.0400662871806474e-05, "loss": 0.2743, "step": 1463 }, { "epoch": 2.51, "learning_rate": 1.0389539060874599e-05, "loss": 0.2759, "step": 1464 }, { "epoch": 2.52, "learning_rate": 1.0378414767176706e-05, "loss": 0.2917, "step": 1465 }, { "epoch": 2.52, "learning_rate": 1.036729000449942e-05, "loss": 0.2748, "step": 1466 }, { "epoch": 2.52, "learning_rate": 1.0356164786629956e-05, "loss": 0.2756, "step": 1467 }, { "epoch": 2.52, "learning_rate": 1.0345039127356091e-05, "loss": 0.2753, "step": 1468 }, { "epoch": 2.52, "learning_rate": 1.0333913040466145e-05, "loss": 0.2795, "step": 1469 }, { "epoch": 2.52, "learning_rate": 1.0322786539748973e-05, "loss": 0.2885, "step": 1470 }, { "epoch": 2.53, "learning_rate": 1.0311659638993933e-05, "loss": 0.2669, "step": 1471 }, { "epoch": 2.53, "learning_rate": 1.0300532351990899e-05, "loss": 0.275, "step": 1472 }, { "epoch": 2.53, "learning_rate": 1.0289404692530197e-05, "loss": 0.2753, "step": 1473 }, { "epoch": 2.53, "learning_rate": 1.0278276674402639e-05, "loss": 0.2812, "step": 1474 }, { "epoch": 2.53, "learning_rate": 1.0267148311399464e-05, "loss": 0.2726, "step": 1475 }, { "epoch": 2.53, "learning_rate": 1.0256019617312354e-05, "loss": 0.2645, "step": 1476 }, { "epoch": 2.54, "learning_rate": 1.0244890605933384e-05, "loss": 0.2767, "step": 1477 }, { "epoch": 2.54, "learning_rate": 1.0233761291055036e-05, "loss": 0.264, "step": 1478 }, { "epoch": 2.54, "learning_rate": 1.0222631686470162e-05, "loss": 0.2918, "step": 1479 }, { "epoch": 2.54, "learning_rate": 1.0211501805971973e-05, "loss": 0.276, "step": 1480 }, { "epoch": 2.54, "learning_rate": 1.0200371663354023e-05, "loss": 0.2636, "step": 1481 }, { "epoch": 2.54, "learning_rate": 1.0189241272410191e-05, "loss": 0.2742, "step": 1482 }, { "epoch": 2.55, "learning_rate": 1.0178110646934667e-05, "loss": 0.2671, "step": 1483 }, { "epoch": 2.55, "learning_rate": 1.0166979800721923e-05, "loss": 0.2694, "step": 1484 }, { "epoch": 2.55, "learning_rate": 1.0155848747566719e-05, "loss": 0.2713, "step": 1485 }, { "epoch": 2.55, "learning_rate": 1.0144717501264052e-05, "loss": 0.2711, "step": 1486 }, { "epoch": 2.55, "learning_rate": 1.0133586075609172e-05, "loss": 0.2751, "step": 1487 }, { "epoch": 2.55, "learning_rate": 1.012245448439755e-05, "loss": 0.26, "step": 1488 }, { "epoch": 2.56, "learning_rate": 1.011132274142486e-05, "loss": 0.2657, "step": 1489 }, { "epoch": 2.56, "learning_rate": 1.0100190860486964e-05, "loss": 0.2711, "step": 1490 }, { "epoch": 2.56, "learning_rate": 1.0089058855379894e-05, "loss": 0.2779, "step": 1491 }, { "epoch": 2.56, "learning_rate": 1.0077926739899842e-05, "loss": 0.2666, "step": 1492 }, { "epoch": 2.56, "learning_rate": 1.0066794527843125e-05, "loss": 0.2658, "step": 1493 }, { "epoch": 2.56, "learning_rate": 1.005566223300619e-05, "loss": 0.2779, "step": 1494 }, { "epoch": 2.57, "learning_rate": 1.0044529869185584e-05, "loss": 0.2796, "step": 1495 }, { "epoch": 2.57, "learning_rate": 1.0033397450177938e-05, "loss": 0.2649, "step": 1496 }, { "epoch": 2.57, "learning_rate": 1.0022264989779951e-05, "loss": 0.2676, "step": 1497 }, { "epoch": 2.57, "learning_rate": 1.0011132501788378e-05, "loss": 0.2627, "step": 1498 }, { "epoch": 2.57, "learning_rate": 1e-05, "loss": 0.2784, "step": 1499 }, { "epoch": 2.58, "learning_rate": 9.988867498211623e-06, "loss": 0.2733, "step": 1500 }, { "epoch": 2.58, "learning_rate": 9.977735010220049e-06, "loss": 0.2772, "step": 1501 }, { "epoch": 2.58, "learning_rate": 9.966602549822064e-06, "loss": 0.2808, "step": 1502 }, { "epoch": 2.58, "learning_rate": 9.95547013081442e-06, "loss": 0.2661, "step": 1503 }, { "epoch": 2.58, "learning_rate": 9.944337766993813e-06, "loss": 0.2748, "step": 1504 }, { "epoch": 2.58, "learning_rate": 9.933205472156877e-06, "loss": 0.277, "step": 1505 }, { "epoch": 2.59, "learning_rate": 9.922073260100161e-06, "loss": 0.256, "step": 1506 }, { "epoch": 2.59, "learning_rate": 9.910941144620106e-06, "loss": 0.2526, "step": 1507 }, { "epoch": 2.59, "learning_rate": 9.899809139513038e-06, "loss": 0.2581, "step": 1508 }, { "epoch": 2.59, "learning_rate": 9.888677258575141e-06, "loss": 0.2765, "step": 1509 }, { "epoch": 2.59, "learning_rate": 9.877545515602453e-06, "loss": 0.2643, "step": 1510 }, { "epoch": 2.59, "learning_rate": 9.866413924390831e-06, "loss": 0.2682, "step": 1511 }, { "epoch": 2.6, "learning_rate": 9.855282498735952e-06, "loss": 0.2652, "step": 1512 }, { "epoch": 2.6, "learning_rate": 9.844151252433285e-06, "loss": 0.2587, "step": 1513 }, { "epoch": 2.6, "learning_rate": 9.833020199278076e-06, "loss": 0.2624, "step": 1514 }, { "epoch": 2.6, "learning_rate": 9.82188935306534e-06, "loss": 0.2636, "step": 1515 }, { "epoch": 2.6, "learning_rate": 9.810758727589814e-06, "loss": 0.2717, "step": 1516 }, { "epoch": 2.6, "learning_rate": 9.799628336645979e-06, "loss": 0.2601, "step": 1517 }, { "epoch": 2.61, "learning_rate": 9.78849819402803e-06, "loss": 0.2722, "step": 1518 }, { "epoch": 2.61, "learning_rate": 9.777368313529841e-06, "loss": 0.2708, "step": 1519 }, { "epoch": 2.61, "learning_rate": 9.766238708944964e-06, "loss": 0.2707, "step": 1520 }, { "epoch": 2.61, "learning_rate": 9.755109394066616e-06, "loss": 0.2561, "step": 1521 }, { "epoch": 2.61, "learning_rate": 9.743980382687651e-06, "loss": 0.2629, "step": 1522 }, { "epoch": 2.61, "learning_rate": 9.73285168860054e-06, "loss": 0.269, "step": 1523 }, { "epoch": 2.62, "learning_rate": 9.721723325597366e-06, "loss": 0.2587, "step": 1524 }, { "epoch": 2.62, "learning_rate": 9.710595307469805e-06, "loss": 0.2744, "step": 1525 }, { "epoch": 2.62, "learning_rate": 9.699467648009104e-06, "loss": 0.266, "step": 1526 }, { "epoch": 2.62, "learning_rate": 9.688340361006067e-06, "loss": 0.2558, "step": 1527 }, { "epoch": 2.62, "learning_rate": 9.677213460251034e-06, "loss": 0.2676, "step": 1528 }, { "epoch": 2.62, "learning_rate": 9.66608695953386e-06, "loss": 0.2552, "step": 1529 }, { "epoch": 2.63, "learning_rate": 9.654960872643912e-06, "loss": 0.2557, "step": 1530 }, { "epoch": 2.63, "learning_rate": 9.643835213370047e-06, "loss": 0.265, "step": 1531 }, { "epoch": 2.63, "learning_rate": 9.632709995500584e-06, "loss": 0.2642, "step": 1532 }, { "epoch": 2.63, "learning_rate": 9.6215852328233e-06, "loss": 0.2586, "step": 1533 }, { "epoch": 2.63, "learning_rate": 9.610460939125406e-06, "loss": 0.2754, "step": 1534 }, { "epoch": 2.64, "learning_rate": 9.599337128193532e-06, "loss": 0.2658, "step": 1535 }, { "epoch": 2.64, "learning_rate": 9.588213813813712e-06, "loss": 0.2582, "step": 1536 }, { "epoch": 2.64, "learning_rate": 9.577091009771359e-06, "loss": 0.2751, "step": 1537 }, { "epoch": 2.64, "learning_rate": 9.565968729851259e-06, "loss": 0.2581, "step": 1538 }, { "epoch": 2.64, "learning_rate": 9.554846987837548e-06, "loss": 0.2615, "step": 1539 }, { "epoch": 2.64, "learning_rate": 9.54372579751369e-06, "loss": 0.2669, "step": 1540 }, { "epoch": 2.65, "learning_rate": 9.532605172662475e-06, "loss": 0.2571, "step": 1541 }, { "epoch": 2.65, "learning_rate": 9.52148512706598e-06, "loss": 0.2611, "step": 1542 }, { "epoch": 2.65, "learning_rate": 9.510365674505576e-06, "loss": 0.2635, "step": 1543 }, { "epoch": 2.65, "learning_rate": 9.499246828761887e-06, "loss": 0.2671, "step": 1544 }, { "epoch": 2.65, "learning_rate": 9.488128603614797e-06, "loss": 0.2512, "step": 1545 }, { "epoch": 2.65, "learning_rate": 9.477011012843414e-06, "loss": 0.2716, "step": 1546 }, { "epoch": 2.66, "learning_rate": 9.465894070226064e-06, "loss": 0.2658, "step": 1547 }, { "epoch": 2.66, "learning_rate": 9.454777789540259e-06, "loss": 0.2592, "step": 1548 }, { "epoch": 2.66, "learning_rate": 9.443662184562706e-06, "loss": 0.2558, "step": 1549 }, { "epoch": 2.66, "learning_rate": 9.43254726906926e-06, "loss": 0.2668, "step": 1550 }, { "epoch": 2.66, "learning_rate": 9.421433056834936e-06, "loss": 0.2703, "step": 1551 }, { "epoch": 2.66, "learning_rate": 9.410319561633867e-06, "loss": 0.2622, "step": 1552 }, { "epoch": 2.67, "learning_rate": 9.399206797239298e-06, "loss": 0.2462, "step": 1553 }, { "epoch": 2.67, "learning_rate": 9.388094777423577e-06, "loss": 0.2718, "step": 1554 }, { "epoch": 2.67, "learning_rate": 9.376983515958116e-06, "loss": 0.2624, "step": 1555 }, { "epoch": 2.67, "learning_rate": 9.365873026613396e-06, "loss": 0.2567, "step": 1556 }, { "epoch": 2.67, "learning_rate": 9.354763323158942e-06, "loss": 0.2541, "step": 1557 }, { "epoch": 2.67, "learning_rate": 9.343654419363299e-06, "loss": 0.2556, "step": 1558 }, { "epoch": 2.68, "learning_rate": 9.332546328994024e-06, "loss": 0.264, "step": 1559 }, { "epoch": 2.68, "learning_rate": 9.321439065817674e-06, "loss": 0.2537, "step": 1560 }, { "epoch": 2.68, "learning_rate": 9.310332643599757e-06, "loss": 0.2554, "step": 1561 }, { "epoch": 2.68, "learning_rate": 9.299227076104768e-06, "loss": 0.2577, "step": 1562 }, { "epoch": 2.68, "learning_rate": 9.288122377096124e-06, "loss": 0.2554, "step": 1563 }, { "epoch": 2.68, "learning_rate": 9.277018560336175e-06, "loss": 0.2511, "step": 1564 }, { "epoch": 2.69, "learning_rate": 9.265915639586169e-06, "loss": 0.2628, "step": 1565 }, { "epoch": 2.69, "learning_rate": 9.254813628606254e-06, "loss": 0.2695, "step": 1566 }, { "epoch": 2.69, "learning_rate": 9.243712541155437e-06, "loss": 0.2536, "step": 1567 }, { "epoch": 2.69, "learning_rate": 9.23261239099159e-06, "loss": 0.2659, "step": 1568 }, { "epoch": 2.69, "learning_rate": 9.221513191871425e-06, "loss": 0.2573, "step": 1569 }, { "epoch": 2.7, "learning_rate": 9.210414957550468e-06, "loss": 0.2636, "step": 1570 }, { "epoch": 2.7, "learning_rate": 9.19931770178306e-06, "loss": 0.2673, "step": 1571 }, { "epoch": 2.7, "learning_rate": 9.188221438322315e-06, "loss": 0.2433, "step": 1572 }, { "epoch": 2.7, "learning_rate": 9.177126180920132e-06, "loss": 0.2488, "step": 1573 }, { "epoch": 2.7, "learning_rate": 9.166031943327147e-06, "loss": 0.2658, "step": 1574 }, { "epoch": 2.7, "learning_rate": 9.154938739292745e-06, "loss": 0.2598, "step": 1575 }, { "epoch": 2.71, "learning_rate": 9.143846582565029e-06, "loss": 0.2613, "step": 1576 }, { "epoch": 2.71, "learning_rate": 9.132755486890795e-06, "loss": 0.2658, "step": 1577 }, { "epoch": 2.71, "learning_rate": 9.121665466015532e-06, "loss": 0.2572, "step": 1578 }, { "epoch": 2.71, "learning_rate": 9.110576533683405e-06, "loss": 0.2567, "step": 1579 }, { "epoch": 2.71, "learning_rate": 9.099488703637205e-06, "loss": 0.2647, "step": 1580 }, { "epoch": 2.71, "learning_rate": 9.088401989618378e-06, "loss": 0.252, "step": 1581 }, { "epoch": 2.72, "learning_rate": 9.07731640536698e-06, "loss": 0.2483, "step": 1582 }, { "epoch": 2.72, "learning_rate": 9.066231964621672e-06, "loss": 0.2596, "step": 1583 }, { "epoch": 2.72, "learning_rate": 9.055148681119688e-06, "loss": 0.2562, "step": 1584 }, { "epoch": 2.72, "learning_rate": 9.044066568596838e-06, "loss": 0.2601, "step": 1585 }, { "epoch": 2.72, "learning_rate": 9.03298564078747e-06, "loss": 0.2538, "step": 1586 }, { "epoch": 2.72, "learning_rate": 9.021905911424476e-06, "loss": 0.2643, "step": 1587 }, { "epoch": 2.73, "learning_rate": 9.01082739423925e-06, "loss": 0.2666, "step": 1588 }, { "epoch": 2.73, "learning_rate": 8.999750102961694e-06, "loss": 0.2488, "step": 1589 }, { "epoch": 2.73, "learning_rate": 8.98867405132019e-06, "loss": 0.251, "step": 1590 }, { "epoch": 2.73, "learning_rate": 8.977599253041575e-06, "loss": 0.2549, "step": 1591 }, { "epoch": 2.73, "learning_rate": 8.966525721851142e-06, "loss": 0.2473, "step": 1592 }, { "epoch": 2.73, "learning_rate": 8.955453471472607e-06, "loss": 0.2561, "step": 1593 }, { "epoch": 2.74, "learning_rate": 8.944382515628104e-06, "loss": 0.2589, "step": 1594 }, { "epoch": 2.74, "learning_rate": 8.933312868038157e-06, "loss": 0.2579, "step": 1595 }, { "epoch": 2.74, "learning_rate": 8.922244542421676e-06, "loss": 0.2615, "step": 1596 }, { "epoch": 2.74, "learning_rate": 8.911177552495926e-06, "loss": 0.2468, "step": 1597 }, { "epoch": 2.74, "learning_rate": 8.900111911976523e-06, "loss": 0.2563, "step": 1598 }, { "epoch": 2.75, "learning_rate": 8.8890476345774e-06, "loss": 0.2493, "step": 1599 }, { "epoch": 2.75, "learning_rate": 8.877984734010813e-06, "loss": 0.2529, "step": 1600 }, { "epoch": 2.75, "learning_rate": 8.866923223987303e-06, "loss": 0.2497, "step": 1601 }, { "epoch": 2.75, "learning_rate": 8.855863118215692e-06, "loss": 0.2658, "step": 1602 }, { "epoch": 2.75, "learning_rate": 8.84480443040306e-06, "loss": 0.2446, "step": 1603 }, { "epoch": 2.75, "learning_rate": 8.833747174254737e-06, "loss": 0.2528, "step": 1604 }, { "epoch": 2.76, "learning_rate": 8.822691363474257e-06, "loss": 0.2427, "step": 1605 }, { "epoch": 2.76, "learning_rate": 8.811637011763389e-06, "loss": 0.2498, "step": 1606 }, { "epoch": 2.76, "learning_rate": 8.800584132822079e-06, "loss": 0.2527, "step": 1607 }, { "epoch": 2.76, "learning_rate": 8.789532740348452e-06, "loss": 0.2458, "step": 1608 }, { "epoch": 2.76, "learning_rate": 8.778482848038787e-06, "loss": 0.2622, "step": 1609 }, { "epoch": 2.76, "learning_rate": 8.767434469587512e-06, "loss": 0.2469, "step": 1610 }, { "epoch": 2.77, "learning_rate": 8.756387618687174e-06, "loss": 0.2486, "step": 1611 }, { "epoch": 2.77, "learning_rate": 8.745342309028418e-06, "loss": 0.2489, "step": 1612 }, { "epoch": 2.77, "learning_rate": 8.734298554299989e-06, "loss": 0.2595, "step": 1613 }, { "epoch": 2.77, "learning_rate": 8.723256368188708e-06, "loss": 0.2393, "step": 1614 }, { "epoch": 2.77, "learning_rate": 8.712215764379446e-06, "loss": 0.2559, "step": 1615 }, { "epoch": 2.77, "learning_rate": 8.701176756555114e-06, "loss": 0.2571, "step": 1616 }, { "epoch": 2.78, "learning_rate": 8.690139358396648e-06, "loss": 0.2393, "step": 1617 }, { "epoch": 2.78, "learning_rate": 8.67910358358298e-06, "loss": 0.2502, "step": 1618 }, { "epoch": 2.78, "learning_rate": 8.668069445791039e-06, "loss": 0.2491, "step": 1619 }, { "epoch": 2.78, "learning_rate": 8.657036958695722e-06, "loss": 0.2409, "step": 1620 }, { "epoch": 2.78, "learning_rate": 8.64600613596988e-06, "loss": 0.2504, "step": 1621 }, { "epoch": 2.78, "learning_rate": 8.634976991284307e-06, "loss": 0.2423, "step": 1622 }, { "epoch": 2.79, "learning_rate": 8.623949538307715e-06, "loss": 0.2395, "step": 1623 }, { "epoch": 2.79, "learning_rate": 8.612923790706706e-06, "loss": 0.2464, "step": 1624 }, { "epoch": 2.79, "learning_rate": 8.601899762145787e-06, "loss": 0.2481, "step": 1625 }, { "epoch": 2.79, "learning_rate": 8.590877466287322e-06, "loss": 0.2493, "step": 1626 }, { "epoch": 2.79, "learning_rate": 8.579856916791539e-06, "loss": 0.242, "step": 1627 }, { "epoch": 2.79, "learning_rate": 8.568838127316491e-06, "loss": 0.2371, "step": 1628 }, { "epoch": 2.8, "learning_rate": 8.557821111518056e-06, "loss": 0.2463, "step": 1629 }, { "epoch": 2.8, "learning_rate": 8.546805883049913e-06, "loss": 0.2377, "step": 1630 }, { "epoch": 2.8, "learning_rate": 8.53579245556352e-06, "loss": 0.2461, "step": 1631 }, { "epoch": 2.8, "learning_rate": 8.524780842708111e-06, "loss": 0.2454, "step": 1632 }, { "epoch": 2.8, "learning_rate": 8.513771058130669e-06, "loss": 0.2476, "step": 1633 }, { "epoch": 2.81, "learning_rate": 8.502763115475908e-06, "loss": 0.2397, "step": 1634 }, { "epoch": 2.81, "learning_rate": 8.491757028386262e-06, "loss": 0.2441, "step": 1635 }, { "epoch": 2.81, "learning_rate": 8.480752810501867e-06, "loss": 0.2424, "step": 1636 }, { "epoch": 2.81, "learning_rate": 8.469750475460537e-06, "loss": 0.2516, "step": 1637 }, { "epoch": 2.81, "learning_rate": 8.458750036897756e-06, "loss": 0.2608, "step": 1638 }, { "epoch": 2.81, "learning_rate": 8.447751508446658e-06, "loss": 0.2419, "step": 1639 }, { "epoch": 2.82, "learning_rate": 8.43675490373801e-06, "loss": 0.2408, "step": 1640 }, { "epoch": 2.82, "learning_rate": 8.425760236400191e-06, "loss": 0.2362, "step": 1641 }, { "epoch": 2.82, "learning_rate": 8.414767520059184e-06, "loss": 0.2499, "step": 1642 }, { "epoch": 2.82, "learning_rate": 8.403776768338555e-06, "loss": 0.2463, "step": 1643 }, { "epoch": 2.82, "learning_rate": 8.392787994859427e-06, "loss": 0.247, "step": 1644 }, { "epoch": 2.82, "learning_rate": 8.381801213240478e-06, "loss": 0.2484, "step": 1645 }, { "epoch": 2.83, "learning_rate": 8.370816437097917e-06, "loss": 0.2692, "step": 1646 }, { "epoch": 2.83, "learning_rate": 8.359833680045463e-06, "loss": 0.2406, "step": 1647 }, { "epoch": 2.83, "learning_rate": 8.348852955694343e-06, "loss": 0.2382, "step": 1648 }, { "epoch": 2.83, "learning_rate": 8.337874277653254e-06, "loss": 0.2441, "step": 1649 }, { "epoch": 2.83, "learning_rate": 8.326897659528359e-06, "loss": 0.2607, "step": 1650 }, { "epoch": 2.83, "learning_rate": 8.315923114923271e-06, "loss": 0.2458, "step": 1651 }, { "epoch": 2.84, "learning_rate": 8.304950657439034e-06, "loss": 0.2416, "step": 1652 }, { "epoch": 2.84, "learning_rate": 8.293980300674103e-06, "loss": 0.2441, "step": 1653 }, { "epoch": 2.84, "learning_rate": 8.28301205822433e-06, "loss": 0.2461, "step": 1654 }, { "epoch": 2.84, "learning_rate": 8.27204594368295e-06, "loss": 0.2452, "step": 1655 }, { "epoch": 2.84, "learning_rate": 8.26108197064055e-06, "loss": 0.2477, "step": 1656 }, { "epoch": 2.84, "learning_rate": 8.250120152685074e-06, "loss": 0.25, "step": 1657 }, { "epoch": 2.85, "learning_rate": 8.239160503401794e-06, "loss": 0.2268, "step": 1658 }, { "epoch": 2.85, "learning_rate": 8.22820303637329e-06, "loss": 0.2356, "step": 1659 }, { "epoch": 2.85, "learning_rate": 8.217247765179442e-06, "loss": 0.2526, "step": 1660 }, { "epoch": 2.85, "learning_rate": 8.206294703397405e-06, "loss": 0.2328, "step": 1661 }, { "epoch": 2.85, "learning_rate": 8.1953438646016e-06, "loss": 0.229, "step": 1662 }, { "epoch": 2.85, "learning_rate": 8.184395262363682e-06, "loss": 0.2263, "step": 1663 }, { "epoch": 2.86, "learning_rate": 8.173448910252549e-06, "loss": 0.2439, "step": 1664 }, { "epoch": 2.86, "learning_rate": 8.162504821834296e-06, "loss": 0.2443, "step": 1665 }, { "epoch": 2.86, "learning_rate": 8.15156301067223e-06, "loss": 0.2352, "step": 1666 }, { "epoch": 2.86, "learning_rate": 8.14062349032682e-06, "loss": 0.241, "step": 1667 }, { "epoch": 2.86, "learning_rate": 8.129686274355707e-06, "loss": 0.2424, "step": 1668 }, { "epoch": 2.87, "learning_rate": 8.118751376313666e-06, "loss": 0.2382, "step": 1669 }, { "epoch": 2.87, "learning_rate": 8.107818809752603e-06, "loss": 0.2361, "step": 1670 }, { "epoch": 2.87, "learning_rate": 8.096888588221536e-06, "loss": 0.2391, "step": 1671 }, { "epoch": 2.87, "learning_rate": 8.085960725266581e-06, "loss": 0.2437, "step": 1672 }, { "epoch": 2.87, "learning_rate": 8.07503523443092e-06, "loss": 0.2406, "step": 1673 }, { "epoch": 2.87, "learning_rate": 8.064112129254814e-06, "loss": 0.2492, "step": 1674 }, { "epoch": 2.88, "learning_rate": 8.053191423275542e-06, "loss": 0.2474, "step": 1675 }, { "epoch": 2.88, "learning_rate": 8.042273130027427e-06, "loss": 0.2405, "step": 1676 }, { "epoch": 2.88, "learning_rate": 8.031357263041796e-06, "loss": 0.2328, "step": 1677 }, { "epoch": 2.88, "learning_rate": 8.020443835846972e-06, "loss": 0.2354, "step": 1678 }, { "epoch": 2.88, "learning_rate": 8.009532861968255e-06, "loss": 0.2413, "step": 1679 }, { "epoch": 2.88, "learning_rate": 7.9986243549279e-06, "loss": 0.2278, "step": 1680 }, { "epoch": 2.89, "learning_rate": 7.987718328245109e-06, "loss": 0.2277, "step": 1681 }, { "epoch": 2.89, "learning_rate": 7.976814795436004e-06, "loss": 0.232, "step": 1682 }, { "epoch": 2.89, "learning_rate": 7.965913770013625e-06, "loss": 0.2392, "step": 1683 }, { "epoch": 2.89, "learning_rate": 7.955015265487895e-06, "loss": 0.2356, "step": 1684 }, { "epoch": 2.89, "learning_rate": 7.944119295365624e-06, "loss": 0.236, "step": 1685 }, { "epoch": 2.89, "learning_rate": 7.93322587315047e-06, "loss": 0.2349, "step": 1686 }, { "epoch": 2.9, "learning_rate": 7.922335012342943e-06, "loss": 0.2559, "step": 1687 }, { "epoch": 2.9, "learning_rate": 7.911446726440368e-06, "loss": 0.2277, "step": 1688 }, { "epoch": 2.9, "learning_rate": 7.900561028936887e-06, "loss": 0.2506, "step": 1689 }, { "epoch": 2.9, "learning_rate": 7.889677933323431e-06, "loss": 0.2497, "step": 1690 }, { "epoch": 2.9, "learning_rate": 7.878797453087712e-06, "loss": 0.236, "step": 1691 }, { "epoch": 2.9, "learning_rate": 7.867919601714188e-06, "loss": 0.2306, "step": 1692 }, { "epoch": 2.91, "learning_rate": 7.857044392684073e-06, "loss": 0.2424, "step": 1693 }, { "epoch": 2.91, "learning_rate": 7.846171839475296e-06, "loss": 0.2397, "step": 1694 }, { "epoch": 2.91, "learning_rate": 7.8353019555625e-06, "loss": 0.2355, "step": 1695 }, { "epoch": 2.91, "learning_rate": 7.824434754417019e-06, "loss": 0.2362, "step": 1696 }, { "epoch": 2.91, "learning_rate": 7.81357024950686e-06, "loss": 0.2344, "step": 1697 }, { "epoch": 2.92, "learning_rate": 7.802708454296693e-06, "loss": 0.2306, "step": 1698 }, { "epoch": 2.92, "learning_rate": 7.791849382247825e-06, "loss": 0.2259, "step": 1699 }, { "epoch": 2.92, "learning_rate": 7.780993046818194e-06, "loss": 0.2309, "step": 1700 }, { "epoch": 2.92, "learning_rate": 7.770139461462335e-06, "loss": 0.2366, "step": 1701 }, { "epoch": 2.92, "learning_rate": 7.759288639631388e-06, "loss": 0.2328, "step": 1702 }, { "epoch": 2.92, "learning_rate": 7.74844059477306e-06, "loss": 0.2287, "step": 1703 }, { "epoch": 2.93, "learning_rate": 7.73759534033162e-06, "loss": 0.2269, "step": 1704 }, { "epoch": 2.93, "learning_rate": 7.726752889747876e-06, "loss": 0.2227, "step": 1705 }, { "epoch": 2.93, "learning_rate": 7.715913256459168e-06, "loss": 0.2296, "step": 1706 }, { "epoch": 2.93, "learning_rate": 7.705076453899331e-06, "loss": 0.2437, "step": 1707 }, { "epoch": 2.93, "learning_rate": 7.694242495498706e-06, "loss": 0.2422, "step": 1708 }, { "epoch": 2.93, "learning_rate": 7.683411394684096e-06, "loss": 0.2268, "step": 1709 }, { "epoch": 2.94, "learning_rate": 7.672583164878775e-06, "loss": 0.2335, "step": 1710 }, { "epoch": 2.94, "learning_rate": 7.661757819502454e-06, "loss": 0.2275, "step": 1711 }, { "epoch": 2.94, "learning_rate": 7.650935371971274e-06, "loss": 0.2356, "step": 1712 }, { "epoch": 2.94, "learning_rate": 7.640115835697765e-06, "loss": 0.2323, "step": 1713 }, { "epoch": 2.94, "learning_rate": 7.629299224090874e-06, "loss": 0.2176, "step": 1714 }, { "epoch": 2.94, "learning_rate": 7.618485550555909e-06, "loss": 0.2352, "step": 1715 }, { "epoch": 2.95, "learning_rate": 7.60767482849454e-06, "loss": 0.2272, "step": 1716 }, { "epoch": 2.95, "learning_rate": 7.596867071304781e-06, "loss": 0.2287, "step": 1717 }, { "epoch": 2.95, "learning_rate": 7.586062292380971e-06, "loss": 0.2389, "step": 1718 }, { "epoch": 2.95, "learning_rate": 7.575260505113762e-06, "loss": 0.2317, "step": 1719 }, { "epoch": 2.95, "learning_rate": 7.564461722890082e-06, "loss": 0.2248, "step": 1720 }, { "epoch": 2.95, "learning_rate": 7.553665959093154e-06, "loss": 0.236, "step": 1721 }, { "epoch": 2.96, "learning_rate": 7.542873227102449e-06, "loss": 0.2265, "step": 1722 }, { "epoch": 2.96, "learning_rate": 7.532083540293688e-06, "loss": 0.2365, "step": 1723 }, { "epoch": 2.96, "learning_rate": 7.52129691203881e-06, "loss": 0.228, "step": 1724 }, { "epoch": 2.96, "learning_rate": 7.510513355705973e-06, "loss": 0.2234, "step": 1725 }, { "epoch": 2.96, "learning_rate": 7.499732884659517e-06, "loss": 0.2244, "step": 1726 }, { "epoch": 2.96, "learning_rate": 7.488955512259963e-06, "loss": 0.2259, "step": 1727 }, { "epoch": 2.97, "learning_rate": 7.478181251863999e-06, "loss": 0.2155, "step": 1728 }, { "epoch": 2.97, "learning_rate": 7.467410116824443e-06, "loss": 0.2347, "step": 1729 }, { "epoch": 2.97, "learning_rate": 7.45664212049025e-06, "loss": 0.2319, "step": 1730 }, { "epoch": 2.97, "learning_rate": 7.445877276206485e-06, "loss": 0.2243, "step": 1731 }, { "epoch": 2.97, "learning_rate": 7.435115597314295e-06, "loss": 0.235, "step": 1732 }, { "epoch": 2.98, "learning_rate": 7.424357097150917e-06, "loss": 0.2207, "step": 1733 }, { "epoch": 2.98, "learning_rate": 7.413601789049644e-06, "loss": 0.2199, "step": 1734 }, { "epoch": 2.98, "learning_rate": 7.402849686339811e-06, "loss": 0.2487, "step": 1735 }, { "epoch": 2.98, "learning_rate": 7.392100802346782e-06, "loss": 0.2213, "step": 1736 }, { "epoch": 2.98, "learning_rate": 7.3813551503919335e-06, "loss": 0.2257, "step": 1737 }, { "epoch": 2.98, "learning_rate": 7.370612743792635e-06, "loss": 0.2339, "step": 1738 }, { "epoch": 2.99, "learning_rate": 7.359873595862233e-06, "loss": 0.2243, "step": 1739 }, { "epoch": 2.99, "learning_rate": 7.349137719910037e-06, "loss": 0.2254, "step": 1740 }, { "epoch": 2.99, "learning_rate": 7.3384051292412996e-06, "loss": 0.218, "step": 1741 }, { "epoch": 2.99, "learning_rate": 7.3276758371572064e-06, "loss": 0.2284, "step": 1742 }, { "epoch": 2.99, "learning_rate": 7.316949856954847e-06, "loss": 0.2296, "step": 1743 }, { "epoch": 2.99, "learning_rate": 7.3062272019272185e-06, "loss": 0.2192, "step": 1744 }, { "epoch": 3.0, "learning_rate": 7.295507885363182e-06, "loss": 0.2259, "step": 1745 }, { "epoch": 3.0, "learning_rate": 7.284791920547472e-06, "loss": 0.2208, "step": 1746 }, { "epoch": 3.0, "learning_rate": 7.274079320760666e-06, "loss": 0.2211, "step": 1747 }, { "epoch": 3.0, "learning_rate": 7.263370099279173e-06, "loss": 0.1941, "step": 1748 }, { "epoch": 3.0, "learning_rate": 7.252664269375209e-06, "loss": 0.1808, "step": 1749 }, { "epoch": 3.0, "learning_rate": 7.241961844316794e-06, "loss": 0.1807, "step": 1750 }, { "epoch": 3.01, "learning_rate": 7.2312628373677295e-06, "loss": 0.1769, "step": 1751 }, { "epoch": 3.01, "learning_rate": 7.220567261787567e-06, "loss": 0.1782, "step": 1752 }, { "epoch": 3.01, "learning_rate": 7.209875130831621e-06, "loss": 0.1764, "step": 1753 }, { "epoch": 3.01, "learning_rate": 7.199186457750931e-06, "loss": 0.1753, "step": 1754 }, { "epoch": 3.01, "learning_rate": 7.188501255792251e-06, "loss": 0.1755, "step": 1755 }, { "epoch": 3.01, "learning_rate": 7.177819538198031e-06, "loss": 0.18, "step": 1756 }, { "epoch": 3.02, "learning_rate": 7.167141318206412e-06, "loss": 0.1771, "step": 1757 }, { "epoch": 3.02, "learning_rate": 7.156466609051182e-06, "loss": 0.1731, "step": 1758 }, { "epoch": 3.02, "learning_rate": 7.145795423961798e-06, "loss": 0.1768, "step": 1759 }, { "epoch": 3.02, "learning_rate": 7.135127776163336e-06, "loss": 0.181, "step": 1760 }, { "epoch": 3.02, "learning_rate": 7.124463678876494e-06, "loss": 0.18, "step": 1761 }, { "epoch": 3.02, "learning_rate": 7.113803145317572e-06, "loss": 0.1809, "step": 1762 }, { "epoch": 3.03, "learning_rate": 7.103146188698453e-06, "loss": 0.1782, "step": 1763 }, { "epoch": 3.03, "learning_rate": 7.0924928222265736e-06, "loss": 0.1766, "step": 1764 }, { "epoch": 3.03, "learning_rate": 7.081843059104937e-06, "loss": 0.1727, "step": 1765 }, { "epoch": 3.03, "learning_rate": 7.071196912532075e-06, "loss": 0.1736, "step": 1766 }, { "epoch": 3.03, "learning_rate": 7.0605543957020375e-06, "loss": 0.179, "step": 1767 }, { "epoch": 3.04, "learning_rate": 7.049915521804377e-06, "loss": 0.1791, "step": 1768 }, { "epoch": 3.04, "learning_rate": 7.039280304024126e-06, "loss": 0.1772, "step": 1769 }, { "epoch": 3.04, "learning_rate": 7.0286487555418e-06, "loss": 0.1755, "step": 1770 }, { "epoch": 3.04, "learning_rate": 7.018020889533348e-06, "loss": 0.1793, "step": 1771 }, { "epoch": 3.04, "learning_rate": 7.007396719170165e-06, "loss": 0.1759, "step": 1772 }, { "epoch": 3.04, "learning_rate": 6.996776257619068e-06, "loss": 0.1804, "step": 1773 }, { "epoch": 3.05, "learning_rate": 6.986159518042273e-06, "loss": 0.172, "step": 1774 }, { "epoch": 3.05, "learning_rate": 6.975546513597387e-06, "loss": 0.1726, "step": 1775 }, { "epoch": 3.05, "learning_rate": 6.964937257437385e-06, "loss": 0.1767, "step": 1776 }, { "epoch": 3.05, "learning_rate": 6.954331762710597e-06, "loss": 0.1803, "step": 1777 }, { "epoch": 3.05, "learning_rate": 6.943730042560692e-06, "loss": 0.1789, "step": 1778 }, { "epoch": 3.05, "learning_rate": 6.933132110126663e-06, "loss": 0.1767, "step": 1779 }, { "epoch": 3.06, "learning_rate": 6.922537978542805e-06, "loss": 0.1766, "step": 1780 }, { "epoch": 3.06, "learning_rate": 6.911947660938705e-06, "loss": 0.1761, "step": 1781 }, { "epoch": 3.06, "learning_rate": 6.901361170439224e-06, "loss": 0.1728, "step": 1782 }, { "epoch": 3.06, "learning_rate": 6.890778520164474e-06, "loss": 0.1779, "step": 1783 }, { "epoch": 3.06, "learning_rate": 6.880199723229817e-06, "loss": 0.1727, "step": 1784 }, { "epoch": 3.06, "learning_rate": 6.869624792745832e-06, "loss": 0.1758, "step": 1785 }, { "epoch": 3.07, "learning_rate": 6.859053741818311e-06, "loss": 0.1764, "step": 1786 }, { "epoch": 3.07, "learning_rate": 6.848486583548235e-06, "loss": 0.1713, "step": 1787 }, { "epoch": 3.07, "learning_rate": 6.837923331031761e-06, "loss": 0.1789, "step": 1788 }, { "epoch": 3.07, "learning_rate": 6.827363997360208e-06, "loss": 0.1823, "step": 1789 }, { "epoch": 3.07, "learning_rate": 6.8168085956200345e-06, "loss": 0.173, "step": 1790 }, { "epoch": 3.07, "learning_rate": 6.8062571388928275e-06, "loss": 0.1749, "step": 1791 }, { "epoch": 3.08, "learning_rate": 6.795709640255287e-06, "loss": 0.1734, "step": 1792 }, { "epoch": 3.08, "learning_rate": 6.785166112779205e-06, "loss": 0.1754, "step": 1793 }, { "epoch": 3.08, "learning_rate": 6.774626569531453e-06, "loss": 0.1792, "step": 1794 }, { "epoch": 3.08, "learning_rate": 6.7640910235739685e-06, "loss": 0.1779, "step": 1795 }, { "epoch": 3.08, "learning_rate": 6.753559487963724e-06, "loss": 0.1784, "step": 1796 }, { "epoch": 3.08, "learning_rate": 6.743031975752732e-06, "loss": 0.1744, "step": 1797 }, { "epoch": 3.09, "learning_rate": 6.732508499988016e-06, "loss": 0.17, "step": 1798 }, { "epoch": 3.09, "learning_rate": 6.7219890737115964e-06, "loss": 0.1719, "step": 1799 }, { "epoch": 3.09, "learning_rate": 6.711473709960478e-06, "loss": 0.1768, "step": 1800 }, { "epoch": 3.09, "learning_rate": 6.700962421766628e-06, "loss": 0.1757, "step": 1801 }, { "epoch": 3.09, "learning_rate": 6.690455222156959e-06, "loss": 0.1733, "step": 1802 }, { "epoch": 3.1, "learning_rate": 6.67995212415332e-06, "loss": 0.1745, "step": 1803 }, { "epoch": 3.1, "learning_rate": 6.669453140772477e-06, "loss": 0.1784, "step": 1804 }, { "epoch": 3.1, "learning_rate": 6.6589582850261025e-06, "loss": 0.178, "step": 1805 }, { "epoch": 3.1, "learning_rate": 6.648467569920742e-06, "loss": 0.1759, "step": 1806 }, { "epoch": 3.1, "learning_rate": 6.637981008457819e-06, "loss": 0.1763, "step": 1807 }, { "epoch": 3.1, "learning_rate": 6.627498613633607e-06, "loss": 0.1721, "step": 1808 }, { "epoch": 3.11, "learning_rate": 6.617020398439205e-06, "loss": 0.172, "step": 1809 }, { "epoch": 3.11, "learning_rate": 6.606546375860549e-06, "loss": 0.1748, "step": 1810 }, { "epoch": 3.11, "learning_rate": 6.596076558878367e-06, "loss": 0.1763, "step": 1811 }, { "epoch": 3.11, "learning_rate": 6.585610960468181e-06, "loss": 0.1733, "step": 1812 }, { "epoch": 3.11, "learning_rate": 6.575149593600282e-06, "loss": 0.1763, "step": 1813 }, { "epoch": 3.11, "learning_rate": 6.564692471239723e-06, "loss": 0.1749, "step": 1814 }, { "epoch": 3.12, "learning_rate": 6.554239606346282e-06, "loss": 0.1752, "step": 1815 }, { "epoch": 3.12, "learning_rate": 6.5437910118744764e-06, "loss": 0.1767, "step": 1816 }, { "epoch": 3.12, "learning_rate": 6.53334670077352e-06, "loss": 0.1717, "step": 1817 }, { "epoch": 3.12, "learning_rate": 6.522906685987327e-06, "loss": 0.1717, "step": 1818 }, { "epoch": 3.12, "learning_rate": 6.51247098045448e-06, "loss": 0.1749, "step": 1819 }, { "epoch": 3.12, "learning_rate": 6.502039597108226e-06, "loss": 0.176, "step": 1820 }, { "epoch": 3.13, "learning_rate": 6.491612548876451e-06, "loss": 0.1742, "step": 1821 }, { "epoch": 3.13, "learning_rate": 6.48118984868167e-06, "loss": 0.1744, "step": 1822 }, { "epoch": 3.13, "learning_rate": 6.470771509441014e-06, "loss": 0.1757, "step": 1823 }, { "epoch": 3.13, "learning_rate": 6.4603575440662e-06, "loss": 0.1751, "step": 1824 }, { "epoch": 3.13, "learning_rate": 6.449947965463535e-06, "loss": 0.1761, "step": 1825 }, { "epoch": 3.13, "learning_rate": 6.439542786533879e-06, "loss": 0.1771, "step": 1826 }, { "epoch": 3.14, "learning_rate": 6.429142020172649e-06, "loss": 0.1765, "step": 1827 }, { "epoch": 3.14, "learning_rate": 6.418745679269785e-06, "loss": 0.1766, "step": 1828 }, { "epoch": 3.14, "learning_rate": 6.408353776709748e-06, "loss": 0.1722, "step": 1829 }, { "epoch": 3.14, "learning_rate": 6.397966325371496e-06, "loss": 0.172, "step": 1830 }, { "epoch": 3.14, "learning_rate": 6.387583338128471e-06, "loss": 0.1793, "step": 1831 }, { "epoch": 3.15, "learning_rate": 6.377204827848583e-06, "loss": 0.177, "step": 1832 }, { "epoch": 3.15, "learning_rate": 6.366830807394195e-06, "loss": 0.1744, "step": 1833 }, { "epoch": 3.15, "learning_rate": 6.356461289622101e-06, "loss": 0.1776, "step": 1834 }, { "epoch": 3.15, "learning_rate": 6.346096287383519e-06, "loss": 0.1738, "step": 1835 }, { "epoch": 3.15, "learning_rate": 6.3357358135240664e-06, "loss": 0.172, "step": 1836 }, { "epoch": 3.15, "learning_rate": 6.3253798808837555e-06, "loss": 0.1745, "step": 1837 }, { "epoch": 3.16, "learning_rate": 6.315028502296965e-06, "loss": 0.1758, "step": 1838 }, { "epoch": 3.16, "learning_rate": 6.304681690592431e-06, "loss": 0.1744, "step": 1839 }, { "epoch": 3.16, "learning_rate": 6.29433945859323e-06, "loss": 0.1773, "step": 1840 }, { "epoch": 3.16, "learning_rate": 6.284001819116763e-06, "loss": 0.1763, "step": 1841 }, { "epoch": 3.16, "learning_rate": 6.273668784974737e-06, "loss": 0.1771, "step": 1842 }, { "epoch": 3.16, "learning_rate": 6.263340368973154e-06, "loss": 0.1742, "step": 1843 }, { "epoch": 3.17, "learning_rate": 6.253016583912295e-06, "loss": 0.1787, "step": 1844 }, { "epoch": 3.17, "learning_rate": 6.242697442586693e-06, "loss": 0.1799, "step": 1845 }, { "epoch": 3.17, "learning_rate": 6.232382957785143e-06, "loss": 0.1732, "step": 1846 }, { "epoch": 3.17, "learning_rate": 6.222073142290644e-06, "loss": 0.1774, "step": 1847 }, { "epoch": 3.17, "learning_rate": 6.211768008880427e-06, "loss": 0.1779, "step": 1848 }, { "epoch": 3.17, "learning_rate": 6.201467570325918e-06, "loss": 0.175, "step": 1849 }, { "epoch": 3.18, "learning_rate": 6.19117183939272e-06, "loss": 0.1764, "step": 1850 }, { "epoch": 3.18, "learning_rate": 6.180880828840603e-06, "loss": 0.172, "step": 1851 }, { "epoch": 3.18, "learning_rate": 6.170594551423494e-06, "loss": 0.1759, "step": 1852 }, { "epoch": 3.18, "learning_rate": 6.1603130198894354e-06, "loss": 0.1767, "step": 1853 }, { "epoch": 3.18, "learning_rate": 6.150036246980608e-06, "loss": 0.1789, "step": 1854 }, { "epoch": 3.18, "learning_rate": 6.139764245433286e-06, "loss": 0.1747, "step": 1855 }, { "epoch": 3.19, "learning_rate": 6.129497027977829e-06, "loss": 0.1737, "step": 1856 }, { "epoch": 3.19, "learning_rate": 6.119234607338673e-06, "loss": 0.1763, "step": 1857 }, { "epoch": 3.19, "learning_rate": 6.108976996234307e-06, "loss": 0.1778, "step": 1858 }, { "epoch": 3.19, "learning_rate": 6.098724207377261e-06, "loss": 0.1747, "step": 1859 }, { "epoch": 3.19, "learning_rate": 6.088476253474078e-06, "loss": 0.1773, "step": 1860 }, { "epoch": 3.19, "learning_rate": 6.078233147225323e-06, "loss": 0.172, "step": 1861 }, { "epoch": 3.2, "learning_rate": 6.0679949013255455e-06, "loss": 0.1749, "step": 1862 }, { "epoch": 3.2, "learning_rate": 6.057761528463275e-06, "loss": 0.1758, "step": 1863 }, { "epoch": 3.2, "learning_rate": 6.047533041320998e-06, "loss": 0.1751, "step": 1864 }, { "epoch": 3.2, "learning_rate": 6.037309452575152e-06, "loss": 0.1725, "step": 1865 }, { "epoch": 3.2, "learning_rate": 6.027090774896095e-06, "loss": 0.1759, "step": 1866 }, { "epoch": 3.21, "learning_rate": 6.016877020948107e-06, "loss": 0.1785, "step": 1867 }, { "epoch": 3.21, "learning_rate": 6.006668203389359e-06, "loss": 0.1767, "step": 1868 }, { "epoch": 3.21, "learning_rate": 5.99646433487191e-06, "loss": 0.1743, "step": 1869 }, { "epoch": 3.21, "learning_rate": 5.986265428041682e-06, "loss": 0.1736, "step": 1870 }, { "epoch": 3.21, "learning_rate": 5.976071495538454e-06, "loss": 0.1748, "step": 1871 }, { "epoch": 3.21, "learning_rate": 5.965882549995825e-06, "loss": 0.1714, "step": 1872 }, { "epoch": 3.22, "learning_rate": 5.955698604041231e-06, "loss": 0.172, "step": 1873 }, { "epoch": 3.22, "learning_rate": 5.945519670295904e-06, "loss": 0.1713, "step": 1874 }, { "epoch": 3.22, "learning_rate": 5.935345761374862e-06, "loss": 0.1758, "step": 1875 }, { "epoch": 3.22, "learning_rate": 5.925176889886901e-06, "loss": 0.1748, "step": 1876 }, { "epoch": 3.22, "learning_rate": 5.91501306843457e-06, "loss": 0.1775, "step": 1877 }, { "epoch": 3.22, "learning_rate": 5.904854309614162e-06, "loss": 0.1773, "step": 1878 }, { "epoch": 3.23, "learning_rate": 5.894700626015691e-06, "loss": 0.1758, "step": 1879 }, { "epoch": 3.23, "learning_rate": 5.884552030222888e-06, "loss": 0.1761, "step": 1880 }, { "epoch": 3.23, "learning_rate": 5.874408534813174e-06, "loss": 0.171, "step": 1881 }, { "epoch": 3.23, "learning_rate": 5.8642701523576496e-06, "loss": 0.1755, "step": 1882 }, { "epoch": 3.23, "learning_rate": 5.8541368954210805e-06, "loss": 0.1736, "step": 1883 }, { "epoch": 3.23, "learning_rate": 5.84400877656188e-06, "loss": 0.1762, "step": 1884 }, { "epoch": 3.24, "learning_rate": 5.833885808332091e-06, "loss": 0.1739, "step": 1885 }, { "epoch": 3.24, "learning_rate": 5.823768003277372e-06, "loss": 0.1705, "step": 1886 }, { "epoch": 3.24, "learning_rate": 5.813655373936988e-06, "loss": 0.1713, "step": 1887 }, { "epoch": 3.24, "learning_rate": 5.803547932843787e-06, "loss": 0.1757, "step": 1888 }, { "epoch": 3.24, "learning_rate": 5.793445692524188e-06, "loss": 0.1801, "step": 1889 }, { "epoch": 3.24, "learning_rate": 5.78334866549816e-06, "loss": 0.1758, "step": 1890 }, { "epoch": 3.25, "learning_rate": 5.77325686427922e-06, "loss": 0.1761, "step": 1891 }, { "epoch": 3.25, "learning_rate": 5.763170301374399e-06, "loss": 0.177, "step": 1892 }, { "epoch": 3.25, "learning_rate": 5.753088989284241e-06, "loss": 0.1742, "step": 1893 }, { "epoch": 3.25, "learning_rate": 5.743012940502783e-06, "loss": 0.1709, "step": 1894 }, { "epoch": 3.25, "learning_rate": 5.732942167517537e-06, "loss": 0.1763, "step": 1895 }, { "epoch": 3.25, "learning_rate": 5.722876682809476e-06, "loss": 0.1708, "step": 1896 }, { "epoch": 3.26, "learning_rate": 5.712816498853026e-06, "loss": 0.1718, "step": 1897 }, { "epoch": 3.26, "learning_rate": 5.702761628116029e-06, "loss": 0.1739, "step": 1898 }, { "epoch": 3.26, "learning_rate": 5.692712083059755e-06, "loss": 0.1744, "step": 1899 }, { "epoch": 3.26, "learning_rate": 5.682667876138871e-06, "loss": 0.1776, "step": 1900 }, { "epoch": 3.26, "learning_rate": 5.672629019801428e-06, "loss": 0.1786, "step": 1901 }, { "epoch": 3.27, "learning_rate": 5.662595526488841e-06, "loss": 0.1761, "step": 1902 }, { "epoch": 3.27, "learning_rate": 5.652567408635887e-06, "loss": 0.178, "step": 1903 }, { "epoch": 3.27, "learning_rate": 5.642544678670676e-06, "loss": 0.1752, "step": 1904 }, { "epoch": 3.27, "learning_rate": 5.63252734901464e-06, "loss": 0.178, "step": 1905 }, { "epoch": 3.27, "learning_rate": 5.6225154320825205e-06, "loss": 0.174, "step": 1906 }, { "epoch": 3.27, "learning_rate": 5.612508940282348e-06, "loss": 0.1777, "step": 1907 }, { "epoch": 3.28, "learning_rate": 5.602507886015434e-06, "loss": 0.1728, "step": 1908 }, { "epoch": 3.28, "learning_rate": 5.59251228167635e-06, "loss": 0.1714, "step": 1909 }, { "epoch": 3.28, "learning_rate": 5.582522139652906e-06, "loss": 0.175, "step": 1910 }, { "epoch": 3.28, "learning_rate": 5.572537472326152e-06, "loss": 0.1722, "step": 1911 }, { "epoch": 3.28, "learning_rate": 5.562558292070347e-06, "loss": 0.1729, "step": 1912 }, { "epoch": 3.28, "learning_rate": 5.552584611252954e-06, "loss": 0.1756, "step": 1913 }, { "epoch": 3.29, "learning_rate": 5.542616442234618e-06, "loss": 0.1724, "step": 1914 }, { "epoch": 3.29, "learning_rate": 5.532653797369154e-06, "loss": 0.1728, "step": 1915 }, { "epoch": 3.29, "learning_rate": 5.522696689003533e-06, "loss": 0.1729, "step": 1916 }, { "epoch": 3.29, "learning_rate": 5.512745129477855e-06, "loss": 0.1723, "step": 1917 }, { "epoch": 3.29, "learning_rate": 5.5027991311253494e-06, "loss": 0.1737, "step": 1918 }, { "epoch": 3.29, "learning_rate": 5.49285870627236e-06, "loss": 0.1742, "step": 1919 }, { "epoch": 3.3, "learning_rate": 5.482923867238317e-06, "loss": 0.1726, "step": 1920 }, { "epoch": 3.3, "learning_rate": 5.4729946263357255e-06, "loss": 0.1717, "step": 1921 }, { "epoch": 3.3, "learning_rate": 5.46307099587016e-06, "loss": 0.1714, "step": 1922 }, { "epoch": 3.3, "learning_rate": 5.453152988140229e-06, "loss": 0.1733, "step": 1923 }, { "epoch": 3.3, "learning_rate": 5.443240615437586e-06, "loss": 0.1742, "step": 1924 }, { "epoch": 3.3, "learning_rate": 5.4333338900468935e-06, "loss": 0.1665, "step": 1925 }, { "epoch": 3.31, "learning_rate": 5.423432824245819e-06, "loss": 0.172, "step": 1926 }, { "epoch": 3.31, "learning_rate": 5.4135374303050136e-06, "loss": 0.1721, "step": 1927 }, { "epoch": 3.31, "learning_rate": 5.403647720488099e-06, "loss": 0.1754, "step": 1928 }, { "epoch": 3.31, "learning_rate": 5.393763707051658e-06, "loss": 0.1741, "step": 1929 }, { "epoch": 3.31, "learning_rate": 5.3838854022452e-06, "loss": 0.1712, "step": 1930 }, { "epoch": 3.32, "learning_rate": 5.374012818311174e-06, "loss": 0.1737, "step": 1931 }, { "epoch": 3.32, "learning_rate": 5.364145967484932e-06, "loss": 0.1728, "step": 1932 }, { "epoch": 3.32, "learning_rate": 5.354284861994723e-06, "loss": 0.1784, "step": 1933 }, { "epoch": 3.32, "learning_rate": 5.344429514061668e-06, "loss": 0.1727, "step": 1934 }, { "epoch": 3.32, "learning_rate": 5.334579935899777e-06, "loss": 0.1697, "step": 1935 }, { "epoch": 3.32, "learning_rate": 5.324736139715875e-06, "loss": 0.1734, "step": 1936 }, { "epoch": 3.33, "learning_rate": 5.314898137709643e-06, "loss": 0.176, "step": 1937 }, { "epoch": 3.33, "learning_rate": 5.305065942073576e-06, "loss": 0.1736, "step": 1938 }, { "epoch": 3.33, "learning_rate": 5.295239564992975e-06, "loss": 0.1723, "step": 1939 }, { "epoch": 3.33, "learning_rate": 5.285419018645926e-06, "loss": 0.1692, "step": 1940 }, { "epoch": 3.33, "learning_rate": 5.2756043152032934e-06, "loss": 0.1714, "step": 1941 }, { "epoch": 3.33, "learning_rate": 5.265795466828693e-06, "loss": 0.1737, "step": 1942 }, { "epoch": 3.34, "learning_rate": 5.255992485678492e-06, "loss": 0.1773, "step": 1943 }, { "epoch": 3.34, "learning_rate": 5.246195383901782e-06, "loss": 0.1734, "step": 1944 }, { "epoch": 3.34, "learning_rate": 5.236404173640371e-06, "loss": 0.1744, "step": 1945 }, { "epoch": 3.34, "learning_rate": 5.226618867028761e-06, "loss": 0.1756, "step": 1946 }, { "epoch": 3.34, "learning_rate": 5.216839476194144e-06, "loss": 0.1748, "step": 1947 }, { "epoch": 3.34, "learning_rate": 5.207066013256375e-06, "loss": 0.1754, "step": 1948 }, { "epoch": 3.35, "learning_rate": 5.197298490327962e-06, "loss": 0.1777, "step": 1949 }, { "epoch": 3.35, "learning_rate": 5.187536919514058e-06, "loss": 0.1756, "step": 1950 }, { "epoch": 3.35, "learning_rate": 5.177781312912431e-06, "loss": 0.1752, "step": 1951 }, { "epoch": 3.35, "learning_rate": 5.168031682613463e-06, "loss": 0.177, "step": 1952 }, { "epoch": 3.35, "learning_rate": 5.158288040700127e-06, "loss": 0.1778, "step": 1953 }, { "epoch": 3.35, "learning_rate": 5.14855039924798e-06, "loss": 0.1764, "step": 1954 }, { "epoch": 3.36, "learning_rate": 5.138818770325129e-06, "loss": 0.1751, "step": 1955 }, { "epoch": 3.36, "learning_rate": 5.129093165992241e-06, "loss": 0.1723, "step": 1956 }, { "epoch": 3.36, "learning_rate": 5.119373598302515e-06, "loss": 0.1753, "step": 1957 }, { "epoch": 3.36, "learning_rate": 5.109660079301668e-06, "loss": 0.1707, "step": 1958 }, { "epoch": 3.36, "learning_rate": 5.099952621027919e-06, "loss": 0.1701, "step": 1959 }, { "epoch": 3.36, "learning_rate": 5.09025123551198e-06, "loss": 0.1718, "step": 1960 }, { "epoch": 3.37, "learning_rate": 5.080555934777028e-06, "loss": 0.1786, "step": 1961 }, { "epoch": 3.37, "learning_rate": 5.070866730838703e-06, "loss": 0.1771, "step": 1962 }, { "epoch": 3.37, "learning_rate": 5.061183635705098e-06, "loss": 0.1767, "step": 1963 }, { "epoch": 3.37, "learning_rate": 5.051506661376725e-06, "loss": 0.1732, "step": 1964 }, { "epoch": 3.37, "learning_rate": 5.041835819846512e-06, "loss": 0.1745, "step": 1965 }, { "epoch": 3.38, "learning_rate": 5.032171123099789e-06, "loss": 0.1698, "step": 1966 }, { "epoch": 3.38, "learning_rate": 5.022512583114274e-06, "loss": 0.1713, "step": 1967 }, { "epoch": 3.38, "learning_rate": 5.01286021186004e-06, "loss": 0.1737, "step": 1968 }, { "epoch": 3.38, "learning_rate": 5.003214021299532e-06, "loss": 0.1689, "step": 1969 }, { "epoch": 3.38, "learning_rate": 4.993574023387524e-06, "loss": 0.1739, "step": 1970 }, { "epoch": 3.38, "learning_rate": 4.983940230071121e-06, "loss": 0.1732, "step": 1971 }, { "epoch": 3.39, "learning_rate": 4.974312653289739e-06, "loss": 0.1769, "step": 1972 }, { "epoch": 3.39, "learning_rate": 4.964691304975089e-06, "loss": 0.1707, "step": 1973 }, { "epoch": 3.39, "learning_rate": 4.955076197051154e-06, "loss": 0.1726, "step": 1974 }, { "epoch": 3.39, "learning_rate": 4.9454673414341945e-06, "loss": 0.1747, "step": 1975 }, { "epoch": 3.39, "learning_rate": 4.93586475003272e-06, "loss": 0.1737, "step": 1976 }, { "epoch": 3.39, "learning_rate": 4.926268434747473e-06, "loss": 0.1757, "step": 1977 }, { "epoch": 3.4, "learning_rate": 4.916678407471417e-06, "loss": 0.176, "step": 1978 }, { "epoch": 3.4, "learning_rate": 4.907094680089736e-06, "loss": 0.1709, "step": 1979 }, { "epoch": 3.4, "learning_rate": 4.897517264479785e-06, "loss": 0.1704, "step": 1980 }, { "epoch": 3.4, "learning_rate": 4.887946172511112e-06, "loss": 0.1749, "step": 1981 }, { "epoch": 3.4, "learning_rate": 4.878381416045422e-06, "loss": 0.1687, "step": 1982 }, { "epoch": 3.4, "learning_rate": 4.868823006936572e-06, "loss": 0.1735, "step": 1983 }, { "epoch": 3.41, "learning_rate": 4.859270957030547e-06, "loss": 0.1782, "step": 1984 }, { "epoch": 3.41, "learning_rate": 4.849725278165457e-06, "loss": 0.1749, "step": 1985 }, { "epoch": 3.41, "learning_rate": 4.840185982171515e-06, "loss": 0.178, "step": 1986 }, { "epoch": 3.41, "learning_rate": 4.830653080871015e-06, "loss": 0.1691, "step": 1987 }, { "epoch": 3.41, "learning_rate": 4.821126586078336e-06, "loss": 0.1698, "step": 1988 }, { "epoch": 3.41, "learning_rate": 4.811606509599917e-06, "loss": 0.1706, "step": 1989 }, { "epoch": 3.42, "learning_rate": 4.8020928632342345e-06, "loss": 0.1802, "step": 1990 }, { "epoch": 3.42, "learning_rate": 4.7925856587718045e-06, "loss": 0.1729, "step": 1991 }, { "epoch": 3.42, "learning_rate": 4.783084907995156e-06, "loss": 0.1722, "step": 1992 }, { "epoch": 3.42, "learning_rate": 4.7735906226788185e-06, "loss": 0.1754, "step": 1993 }, { "epoch": 3.42, "learning_rate": 4.76410281458931e-06, "loss": 0.1764, "step": 1994 }, { "epoch": 3.42, "learning_rate": 4.754621495485121e-06, "loss": 0.1755, "step": 1995 }, { "epoch": 3.43, "learning_rate": 4.745146677116701e-06, "loss": 0.1723, "step": 1996 }, { "epoch": 3.43, "learning_rate": 4.7356783712264405e-06, "loss": 0.1691, "step": 1997 }, { "epoch": 3.43, "learning_rate": 4.726216589548667e-06, "loss": 0.1747, "step": 1998 }, { "epoch": 3.43, "learning_rate": 4.716761343809605e-06, "loss": 0.1725, "step": 1999 }, { "epoch": 3.43, "learning_rate": 4.7073126457273945e-06, "loss": 0.1709, "step": 2000 }, { "epoch": 3.44, "learning_rate": 4.697870507012057e-06, "loss": 0.1712, "step": 2001 }, { "epoch": 3.44, "learning_rate": 4.6884349393654825e-06, "loss": 0.1739, "step": 2002 }, { "epoch": 3.44, "learning_rate": 4.6790059544814194e-06, "loss": 0.1732, "step": 2003 }, { "epoch": 3.44, "learning_rate": 4.669583564045457e-06, "loss": 0.176, "step": 2004 }, { "epoch": 3.44, "learning_rate": 4.660167779735011e-06, "loss": 0.1725, "step": 2005 }, { "epoch": 3.44, "learning_rate": 4.650758613219311e-06, "loss": 0.1696, "step": 2006 }, { "epoch": 3.45, "learning_rate": 4.641356076159387e-06, "loss": 0.1732, "step": 2007 }, { "epoch": 3.45, "learning_rate": 4.63196018020805e-06, "loss": 0.1729, "step": 2008 }, { "epoch": 3.45, "learning_rate": 4.622570937009879e-06, "loss": 0.1742, "step": 2009 }, { "epoch": 3.45, "learning_rate": 4.613188358201212e-06, "loss": 0.1701, "step": 2010 }, { "epoch": 3.45, "learning_rate": 4.603812455410132e-06, "loss": 0.1713, "step": 2011 }, { "epoch": 3.45, "learning_rate": 4.594443240256433e-06, "loss": 0.1716, "step": 2012 }, { "epoch": 3.46, "learning_rate": 4.585080724351633e-06, "loss": 0.1729, "step": 2013 }, { "epoch": 3.46, "learning_rate": 4.575724919298946e-06, "loss": 0.1686, "step": 2014 }, { "epoch": 3.46, "learning_rate": 4.566375836693268e-06, "loss": 0.1722, "step": 2015 }, { "epoch": 3.46, "learning_rate": 4.5570334881211605e-06, "loss": 0.1742, "step": 2016 }, { "epoch": 3.46, "learning_rate": 4.547697885160845e-06, "loss": 0.1716, "step": 2017 }, { "epoch": 3.46, "learning_rate": 4.538369039382184e-06, "loss": 0.1717, "step": 2018 }, { "epoch": 3.47, "learning_rate": 4.529046962346653e-06, "loss": 0.1678, "step": 2019 }, { "epoch": 3.47, "learning_rate": 4.519731665607353e-06, "loss": 0.1692, "step": 2020 }, { "epoch": 3.47, "learning_rate": 4.510423160708975e-06, "loss": 0.1773, "step": 2021 }, { "epoch": 3.47, "learning_rate": 4.5011214591877914e-06, "loss": 0.1684, "step": 2022 }, { "epoch": 3.47, "learning_rate": 4.4918265725716525e-06, "loss": 0.1725, "step": 2023 }, { "epoch": 3.47, "learning_rate": 4.4825385123799585e-06, "loss": 0.1706, "step": 2024 }, { "epoch": 3.48, "learning_rate": 4.473257290123637e-06, "loss": 0.1728, "step": 2025 }, { "epoch": 3.48, "learning_rate": 4.463982917305155e-06, "loss": 0.1737, "step": 2026 }, { "epoch": 3.48, "learning_rate": 4.454715405418489e-06, "loss": 0.1715, "step": 2027 }, { "epoch": 3.48, "learning_rate": 4.445454765949111e-06, "loss": 0.1697, "step": 2028 }, { "epoch": 3.48, "learning_rate": 4.436201010373973e-06, "loss": 0.1649, "step": 2029 }, { "epoch": 3.48, "learning_rate": 4.426954150161503e-06, "loss": 0.1705, "step": 2030 }, { "epoch": 3.49, "learning_rate": 4.417714196771571e-06, "loss": 0.1681, "step": 2031 }, { "epoch": 3.49, "learning_rate": 4.408481161655499e-06, "loss": 0.1722, "step": 2032 }, { "epoch": 3.49, "learning_rate": 4.399255056256028e-06, "loss": 0.1723, "step": 2033 }, { "epoch": 3.49, "learning_rate": 4.390035892007318e-06, "loss": 0.1705, "step": 2034 }, { "epoch": 3.49, "learning_rate": 4.380823680334919e-06, "loss": 0.1686, "step": 2035 }, { "epoch": 3.5, "learning_rate": 4.3716184326557676e-06, "loss": 0.1685, "step": 2036 }, { "epoch": 3.5, "learning_rate": 4.362420160378172e-06, "loss": 0.1741, "step": 2037 }, { "epoch": 3.5, "learning_rate": 4.353228874901789e-06, "loss": 0.1728, "step": 2038 }, { "epoch": 3.5, "learning_rate": 4.344044587617626e-06, "loss": 0.1717, "step": 2039 }, { "epoch": 3.5, "learning_rate": 4.334867309908009e-06, "loss": 0.1739, "step": 2040 }, { "epoch": 3.5, "learning_rate": 4.32569705314658e-06, "loss": 0.1705, "step": 2041 }, { "epoch": 3.51, "learning_rate": 4.3165338286982825e-06, "loss": 0.1708, "step": 2042 }, { "epoch": 3.51, "learning_rate": 4.307377647919343e-06, "loss": 0.1719, "step": 2043 }, { "epoch": 3.51, "learning_rate": 4.2982285221572505e-06, "loss": 0.1747, "step": 2044 }, { "epoch": 3.51, "learning_rate": 4.289086462750762e-06, "loss": 0.1709, "step": 2045 }, { "epoch": 3.51, "learning_rate": 4.2799514810298725e-06, "loss": 0.1725, "step": 2046 }, { "epoch": 3.51, "learning_rate": 4.2708235883158055e-06, "loss": 0.1708, "step": 2047 }, { "epoch": 3.52, "learning_rate": 4.261702795920998e-06, "loss": 0.1728, "step": 2048 }, { "epoch": 3.52, "learning_rate": 4.25258911514909e-06, "loss": 0.174, "step": 2049 }, { "epoch": 3.52, "learning_rate": 4.243482557294905e-06, "loss": 0.1744, "step": 2050 }, { "epoch": 3.52, "learning_rate": 4.23438313364444e-06, "loss": 0.172, "step": 2051 }, { "epoch": 3.52, "learning_rate": 4.22529085547485e-06, "loss": 0.1701, "step": 2052 }, { "epoch": 3.52, "learning_rate": 4.216205734054436e-06, "loss": 0.1698, "step": 2053 }, { "epoch": 3.53, "learning_rate": 4.207127780642628e-06, "loss": 0.169, "step": 2054 }, { "epoch": 3.53, "learning_rate": 4.198057006489973e-06, "loss": 0.1738, "step": 2055 }, { "epoch": 3.53, "learning_rate": 4.188993422838123e-06, "loss": 0.179, "step": 2056 }, { "epoch": 3.53, "learning_rate": 4.17993704091981e-06, "loss": 0.1693, "step": 2057 }, { "epoch": 3.53, "learning_rate": 4.1708878719588505e-06, "loss": 0.1683, "step": 2058 }, { "epoch": 3.53, "learning_rate": 4.161845927170117e-06, "loss": 0.1701, "step": 2059 }, { "epoch": 3.54, "learning_rate": 4.152811217759529e-06, "loss": 0.1739, "step": 2060 }, { "epoch": 3.54, "learning_rate": 4.143783754924042e-06, "loss": 0.1723, "step": 2061 }, { "epoch": 3.54, "learning_rate": 4.134763549851631e-06, "loss": 0.1723, "step": 2062 }, { "epoch": 3.54, "learning_rate": 4.125750613721268e-06, "loss": 0.1674, "step": 2063 }, { "epoch": 3.54, "learning_rate": 4.116744957702922e-06, "loss": 0.1689, "step": 2064 }, { "epoch": 3.55, "learning_rate": 4.107746592957545e-06, "loss": 0.1679, "step": 2065 }, { "epoch": 3.55, "learning_rate": 4.09875553063704e-06, "loss": 0.169, "step": 2066 }, { "epoch": 3.55, "learning_rate": 4.0897717818842755e-06, "loss": 0.1694, "step": 2067 }, { "epoch": 3.55, "learning_rate": 4.0807953578330475e-06, "loss": 0.1668, "step": 2068 }, { "epoch": 3.55, "learning_rate": 4.07182626960807e-06, "loss": 0.1695, "step": 2069 }, { "epoch": 3.55, "learning_rate": 4.0628645283249716e-06, "loss": 0.1716, "step": 2070 }, { "epoch": 3.56, "learning_rate": 4.053910145090275e-06, "loss": 0.1687, "step": 2071 }, { "epoch": 3.56, "learning_rate": 4.044963131001384e-06, "loss": 0.1734, "step": 2072 }, { "epoch": 3.56, "learning_rate": 4.036023497146567e-06, "loss": 0.1706, "step": 2073 }, { "epoch": 3.56, "learning_rate": 4.02709125460495e-06, "loss": 0.1725, "step": 2074 }, { "epoch": 3.56, "learning_rate": 4.0181664144464984e-06, "loss": 0.1704, "step": 2075 }, { "epoch": 3.56, "learning_rate": 4.009248987731995e-06, "loss": 0.1651, "step": 2076 }, { "epoch": 3.57, "learning_rate": 4.000338985513046e-06, "loss": 0.1736, "step": 2077 }, { "epoch": 3.57, "learning_rate": 3.991436418832051e-06, "loss": 0.1737, "step": 2078 }, { "epoch": 3.57, "learning_rate": 3.9825412987221965e-06, "loss": 0.1704, "step": 2079 }, { "epoch": 3.57, "learning_rate": 3.973653636207437e-06, "loss": 0.1671, "step": 2080 }, { "epoch": 3.57, "learning_rate": 3.96477344230249e-06, "loss": 0.1709, "step": 2081 }, { "epoch": 3.57, "learning_rate": 3.95590072801281e-06, "loss": 0.1701, "step": 2082 }, { "epoch": 3.58, "learning_rate": 3.947035504334589e-06, "loss": 0.1701, "step": 2083 }, { "epoch": 3.58, "learning_rate": 3.938177782254731e-06, "loss": 0.1751, "step": 2084 }, { "epoch": 3.58, "learning_rate": 3.9293275727508425e-06, "loss": 0.1708, "step": 2085 }, { "epoch": 3.58, "learning_rate": 3.920484886791225e-06, "loss": 0.1729, "step": 2086 }, { "epoch": 3.58, "learning_rate": 3.911649735334853e-06, "loss": 0.1705, "step": 2087 }, { "epoch": 3.58, "learning_rate": 3.902822129331355e-06, "loss": 0.1759, "step": 2088 }, { "epoch": 3.59, "learning_rate": 3.894002079721021e-06, "loss": 0.1691, "step": 2089 }, { "epoch": 3.59, "learning_rate": 3.88518959743477e-06, "loss": 0.1687, "step": 2090 }, { "epoch": 3.59, "learning_rate": 3.876384693394143e-06, "loss": 0.164, "step": 2091 }, { "epoch": 3.59, "learning_rate": 3.867587378511291e-06, "loss": 0.1674, "step": 2092 }, { "epoch": 3.59, "learning_rate": 3.858797663688955e-06, "loss": 0.1678, "step": 2093 }, { "epoch": 3.59, "learning_rate": 3.850015559820465e-06, "loss": 0.167, "step": 2094 }, { "epoch": 3.6, "learning_rate": 3.841241077789709e-06, "loss": 0.1704, "step": 2095 }, { "epoch": 3.6, "learning_rate": 3.832474228471137e-06, "loss": 0.1683, "step": 2096 }, { "epoch": 3.6, "learning_rate": 3.823715022729735e-06, "loss": 0.1698, "step": 2097 }, { "epoch": 3.6, "learning_rate": 3.814963471421017e-06, "loss": 0.1721, "step": 2098 }, { "epoch": 3.6, "learning_rate": 3.8062195853910123e-06, "loss": 0.1697, "step": 2099 }, { "epoch": 3.61, "learning_rate": 3.7974833754762506e-06, "loss": 0.1693, "step": 2100 }, { "epoch": 3.61, "learning_rate": 3.7887548525037434e-06, "loss": 0.17, "step": 2101 }, { "epoch": 3.61, "learning_rate": 3.7800340272909784e-06, "loss": 0.1655, "step": 2102 }, { "epoch": 3.61, "learning_rate": 3.771320910645908e-06, "loss": 0.1671, "step": 2103 }, { "epoch": 3.61, "learning_rate": 3.762615513366925e-06, "loss": 0.1645, "step": 2104 }, { "epoch": 3.61, "learning_rate": 3.7539178462428582e-06, "loss": 0.1652, "step": 2105 }, { "epoch": 3.62, "learning_rate": 3.7452279200529584e-06, "loss": 0.1668, "step": 2106 }, { "epoch": 3.62, "learning_rate": 3.7365457455668742e-06, "loss": 0.1677, "step": 2107 }, { "epoch": 3.62, "learning_rate": 3.7278713335446557e-06, "loss": 0.1705, "step": 2108 }, { "epoch": 3.62, "learning_rate": 3.719204694736728e-06, "loss": 0.1655, "step": 2109 }, { "epoch": 3.62, "learning_rate": 3.71054583988389e-06, "loss": 0.1674, "step": 2110 }, { "epoch": 3.62, "learning_rate": 3.7018947797172864e-06, "loss": 0.1694, "step": 2111 }, { "epoch": 3.63, "learning_rate": 3.6932515249584044e-06, "loss": 0.1704, "step": 2112 }, { "epoch": 3.63, "learning_rate": 3.6846160863190595e-06, "loss": 0.1699, "step": 2113 }, { "epoch": 3.63, "learning_rate": 3.675988474501373e-06, "loss": 0.1743, "step": 2114 }, { "epoch": 3.63, "learning_rate": 3.667368700197773e-06, "loss": 0.1685, "step": 2115 }, { "epoch": 3.63, "learning_rate": 3.6587567740909744e-06, "loss": 0.1703, "step": 2116 }, { "epoch": 3.63, "learning_rate": 3.6501527068539633e-06, "loss": 0.172, "step": 2117 }, { "epoch": 3.64, "learning_rate": 3.641556509149987e-06, "loss": 0.1699, "step": 2118 }, { "epoch": 3.64, "learning_rate": 3.6329681916325444e-06, "loss": 0.1645, "step": 2119 }, { "epoch": 3.64, "learning_rate": 3.6243877649453553e-06, "loss": 0.1699, "step": 2120 }, { "epoch": 3.64, "learning_rate": 3.615815239722371e-06, "loss": 0.1721, "step": 2121 }, { "epoch": 3.64, "learning_rate": 3.6072506265877515e-06, "loss": 0.1725, "step": 2122 }, { "epoch": 3.64, "learning_rate": 3.598693936155844e-06, "loss": 0.1688, "step": 2123 }, { "epoch": 3.65, "learning_rate": 3.5901451790311825e-06, "loss": 0.1685, "step": 2124 }, { "epoch": 3.65, "learning_rate": 3.581604365808465e-06, "loss": 0.1685, "step": 2125 }, { "epoch": 3.65, "learning_rate": 3.5730715070725486e-06, "loss": 0.1708, "step": 2126 }, { "epoch": 3.65, "learning_rate": 3.5645466133984286e-06, "loss": 0.1695, "step": 2127 }, { "epoch": 3.65, "learning_rate": 3.5560296953512296e-06, "loss": 0.1743, "step": 2128 }, { "epoch": 3.65, "learning_rate": 3.5475207634861943e-06, "loss": 0.1701, "step": 2129 }, { "epoch": 3.66, "learning_rate": 3.5390198283486653e-06, "loss": 0.1652, "step": 2130 }, { "epoch": 3.66, "learning_rate": 3.530526900474075e-06, "loss": 0.1666, "step": 2131 }, { "epoch": 3.66, "learning_rate": 3.5220419903879356e-06, "loss": 0.1693, "step": 2132 }, { "epoch": 3.66, "learning_rate": 3.5135651086058155e-06, "loss": 0.1675, "step": 2133 }, { "epoch": 3.66, "learning_rate": 3.5050962656333376e-06, "loss": 0.1682, "step": 2134 }, { "epoch": 3.67, "learning_rate": 3.4966354719661645e-06, "loss": 0.1654, "step": 2135 }, { "epoch": 3.67, "learning_rate": 3.48818273808998e-06, "loss": 0.1658, "step": 2136 }, { "epoch": 3.67, "learning_rate": 3.47973807448048e-06, "loss": 0.1676, "step": 2137 }, { "epoch": 3.67, "learning_rate": 3.4713014916033582e-06, "loss": 0.1686, "step": 2138 }, { "epoch": 3.67, "learning_rate": 3.462872999914294e-06, "loss": 0.1719, "step": 2139 }, { "epoch": 3.67, "learning_rate": 3.4544526098589393e-06, "loss": 0.1735, "step": 2140 }, { "epoch": 3.68, "learning_rate": 3.4460403318729064e-06, "loss": 0.1673, "step": 2141 }, { "epoch": 3.68, "learning_rate": 3.4376361763817513e-06, "loss": 0.1703, "step": 2142 }, { "epoch": 3.68, "learning_rate": 3.4292401538009658e-06, "loss": 0.172, "step": 2143 }, { "epoch": 3.68, "learning_rate": 3.420852274535963e-06, "loss": 0.1676, "step": 2144 }, { "epoch": 3.68, "learning_rate": 3.4124725489820643e-06, "loss": 0.168, "step": 2145 }, { "epoch": 3.68, "learning_rate": 3.404100987524479e-06, "loss": 0.1694, "step": 2146 }, { "epoch": 3.69, "learning_rate": 3.3957376005383068e-06, "loss": 0.1705, "step": 2147 }, { "epoch": 3.69, "learning_rate": 3.3873823983885135e-06, "loss": 0.1673, "step": 2148 }, { "epoch": 3.69, "learning_rate": 3.37903539142992e-06, "loss": 0.1689, "step": 2149 }, { "epoch": 3.69, "learning_rate": 3.370696590007194e-06, "loss": 0.1686, "step": 2150 }, { "epoch": 3.69, "learning_rate": 3.3623660044548335e-06, "loss": 0.1701, "step": 2151 }, { "epoch": 3.69, "learning_rate": 3.354043645097147e-06, "loss": 0.1646, "step": 2152 }, { "epoch": 3.7, "learning_rate": 3.3457295222482532e-06, "loss": 0.1677, "step": 2153 }, { "epoch": 3.7, "learning_rate": 3.33742364621207e-06, "loss": 0.1651, "step": 2154 }, { "epoch": 3.7, "learning_rate": 3.3291260272822857e-06, "loss": 0.1716, "step": 2155 }, { "epoch": 3.7, "learning_rate": 3.320836675742358e-06, "loss": 0.1701, "step": 2156 }, { "epoch": 3.7, "learning_rate": 3.3125556018655013e-06, "loss": 0.1687, "step": 2157 }, { "epoch": 3.7, "learning_rate": 3.304282815914662e-06, "loss": 0.166, "step": 2158 }, { "epoch": 3.71, "learning_rate": 3.2960183281425263e-06, "loss": 0.1686, "step": 2159 }, { "epoch": 3.71, "learning_rate": 3.28776214879149e-06, "loss": 0.1691, "step": 2160 }, { "epoch": 3.71, "learning_rate": 3.2795142880936546e-06, "loss": 0.1678, "step": 2161 }, { "epoch": 3.71, "learning_rate": 3.2712747562708115e-06, "loss": 0.1677, "step": 2162 }, { "epoch": 3.71, "learning_rate": 3.2630435635344283e-06, "loss": 0.1698, "step": 2163 }, { "epoch": 3.72, "learning_rate": 3.254820720085643e-06, "loss": 0.1687, "step": 2164 }, { "epoch": 3.72, "learning_rate": 3.246606236115235e-06, "loss": 0.1682, "step": 2165 }, { "epoch": 3.72, "learning_rate": 3.2384001218036353e-06, "loss": 0.1691, "step": 2166 }, { "epoch": 3.72, "learning_rate": 3.2302023873208966e-06, "loss": 0.1698, "step": 2167 }, { "epoch": 3.72, "learning_rate": 3.222013042826687e-06, "loss": 0.167, "step": 2168 }, { "epoch": 3.72, "learning_rate": 3.2138320984702763e-06, "loss": 0.1659, "step": 2169 }, { "epoch": 3.73, "learning_rate": 3.2056595643905274e-06, "loss": 0.1678, "step": 2170 }, { "epoch": 3.73, "learning_rate": 3.197495450715874e-06, "loss": 0.1697, "step": 2171 }, { "epoch": 3.73, "learning_rate": 3.189339767564318e-06, "loss": 0.1652, "step": 2172 }, { "epoch": 3.73, "learning_rate": 3.181192525043413e-06, "loss": 0.1643, "step": 2173 }, { "epoch": 3.73, "learning_rate": 3.17305373325025e-06, "loss": 0.1683, "step": 2174 }, { "epoch": 3.73, "learning_rate": 3.164923402271449e-06, "loss": 0.1665, "step": 2175 }, { "epoch": 3.74, "learning_rate": 3.156801542183148e-06, "loss": 0.1665, "step": 2176 }, { "epoch": 3.74, "learning_rate": 3.1486881630509736e-06, "loss": 0.1697, "step": 2177 }, { "epoch": 3.74, "learning_rate": 3.1405832749300546e-06, "loss": 0.1714, "step": 2178 }, { "epoch": 3.74, "learning_rate": 3.132486887864992e-06, "loss": 0.1679, "step": 2179 }, { "epoch": 3.74, "learning_rate": 3.1243990118898505e-06, "loss": 0.1695, "step": 2180 }, { "epoch": 3.74, "learning_rate": 3.1163196570281496e-06, "loss": 0.163, "step": 2181 }, { "epoch": 3.75, "learning_rate": 3.108248833292846e-06, "loss": 0.1613, "step": 2182 }, { "epoch": 3.75, "learning_rate": 3.100186550686324e-06, "loss": 0.1684, "step": 2183 }, { "epoch": 3.75, "learning_rate": 3.092132819200383e-06, "loss": 0.1709, "step": 2184 }, { "epoch": 3.75, "learning_rate": 3.0840876488162254e-06, "loss": 0.1722, "step": 2185 }, { "epoch": 3.75, "learning_rate": 3.0760510495044414e-06, "loss": 0.1658, "step": 2186 }, { "epoch": 3.75, "learning_rate": 3.0680230312250015e-06, "loss": 0.1693, "step": 2187 }, { "epoch": 3.76, "learning_rate": 3.0600036039272385e-06, "loss": 0.1681, "step": 2188 }, { "epoch": 3.76, "learning_rate": 3.0519927775498448e-06, "loss": 0.1654, "step": 2189 }, { "epoch": 3.76, "learning_rate": 3.043990562020842e-06, "loss": 0.1661, "step": 2190 }, { "epoch": 3.76, "learning_rate": 3.0359969672575886e-06, "loss": 0.1709, "step": 2191 }, { "epoch": 3.76, "learning_rate": 3.028012003166758e-06, "loss": 0.1683, "step": 2192 }, { "epoch": 3.76, "learning_rate": 3.0200356796443276e-06, "loss": 0.1698, "step": 2193 }, { "epoch": 3.77, "learning_rate": 3.0120680065755637e-06, "loss": 0.1726, "step": 2194 }, { "epoch": 3.77, "learning_rate": 3.0041089938350166e-06, "loss": 0.1695, "step": 2195 }, { "epoch": 3.77, "learning_rate": 2.9961586512864947e-06, "loss": 0.1653, "step": 2196 }, { "epoch": 3.77, "learning_rate": 2.9882169887830682e-06, "loss": 0.1639, "step": 2197 }, { "epoch": 3.77, "learning_rate": 2.9802840161670532e-06, "loss": 0.1694, "step": 2198 }, { "epoch": 3.78, "learning_rate": 2.972359743269988e-06, "loss": 0.1691, "step": 2199 }, { "epoch": 3.78, "learning_rate": 2.9644441799126346e-06, "loss": 0.1646, "step": 2200 }, { "epoch": 3.78, "learning_rate": 2.9565373359049586e-06, "loss": 0.1693, "step": 2201 }, { "epoch": 3.78, "learning_rate": 2.9486392210461225e-06, "loss": 0.1675, "step": 2202 }, { "epoch": 3.78, "learning_rate": 2.940749845124463e-06, "loss": 0.1668, "step": 2203 }, { "epoch": 3.78, "learning_rate": 2.9328692179174933e-06, "loss": 0.1685, "step": 2204 }, { "epoch": 3.79, "learning_rate": 2.924997349191884e-06, "loss": 0.1664, "step": 2205 }, { "epoch": 3.79, "learning_rate": 2.917134248703447e-06, "loss": 0.1682, "step": 2206 }, { "epoch": 3.79, "learning_rate": 2.9092799261971326e-06, "loss": 0.1712, "step": 2207 }, { "epoch": 3.79, "learning_rate": 2.901434391407011e-06, "loss": 0.1699, "step": 2208 }, { "epoch": 3.79, "learning_rate": 2.893597654056254e-06, "loss": 0.1675, "step": 2209 }, { "epoch": 3.79, "learning_rate": 2.88576972385714e-06, "loss": 0.1647, "step": 2210 }, { "epoch": 3.8, "learning_rate": 2.87795061051103e-06, "loss": 0.1661, "step": 2211 }, { "epoch": 3.8, "learning_rate": 2.870140323708356e-06, "loss": 0.1631, "step": 2212 }, { "epoch": 3.8, "learning_rate": 2.8623388731286097e-06, "loss": 0.1678, "step": 2213 }, { "epoch": 3.8, "learning_rate": 2.854546268440339e-06, "loss": 0.167, "step": 2214 }, { "epoch": 3.8, "learning_rate": 2.846762519301127e-06, "loss": 0.1708, "step": 2215 }, { "epoch": 3.8, "learning_rate": 2.8389876353575707e-06, "loss": 0.1689, "step": 2216 }, { "epoch": 3.81, "learning_rate": 2.831221626245293e-06, "loss": 0.1679, "step": 2217 }, { "epoch": 3.81, "learning_rate": 2.823464501588913e-06, "loss": 0.1651, "step": 2218 }, { "epoch": 3.81, "learning_rate": 2.815716271002039e-06, "loss": 0.1659, "step": 2219 }, { "epoch": 3.81, "learning_rate": 2.8079769440872582e-06, "loss": 0.1652, "step": 2220 }, { "epoch": 3.81, "learning_rate": 2.800246530436125e-06, "loss": 0.1667, "step": 2221 }, { "epoch": 3.81, "learning_rate": 2.792525039629138e-06, "loss": 0.1679, "step": 2222 }, { "epoch": 3.82, "learning_rate": 2.784812481235749e-06, "loss": 0.169, "step": 2223 }, { "epoch": 3.82, "learning_rate": 2.777108864814333e-06, "loss": 0.1682, "step": 2224 }, { "epoch": 3.82, "learning_rate": 2.769414199912186e-06, "loss": 0.167, "step": 2225 }, { "epoch": 3.82, "learning_rate": 2.7617284960655076e-06, "loss": 0.1652, "step": 2226 }, { "epoch": 3.82, "learning_rate": 2.7540517627993947e-06, "loss": 0.164, "step": 2227 }, { "epoch": 3.82, "learning_rate": 2.7463840096278238e-06, "loss": 0.1629, "step": 2228 }, { "epoch": 3.83, "learning_rate": 2.738725246053644e-06, "loss": 0.1664, "step": 2229 }, { "epoch": 3.83, "learning_rate": 2.7310754815685627e-06, "loss": 0.1684, "step": 2230 }, { "epoch": 3.83, "learning_rate": 2.7234347256531345e-06, "loss": 0.1666, "step": 2231 }, { "epoch": 3.83, "learning_rate": 2.715802987776749e-06, "loss": 0.1663, "step": 2232 }, { "epoch": 3.83, "learning_rate": 2.7081802773976207e-06, "loss": 0.1619, "step": 2233 }, { "epoch": 3.84, "learning_rate": 2.700566603962779e-06, "loss": 0.1685, "step": 2234 }, { "epoch": 3.84, "learning_rate": 2.6929619769080418e-06, "loss": 0.166, "step": 2235 }, { "epoch": 3.84, "learning_rate": 2.6853664056580287e-06, "loss": 0.1636, "step": 2236 }, { "epoch": 3.84, "learning_rate": 2.6777798996261306e-06, "loss": 0.1656, "step": 2237 }, { "epoch": 3.84, "learning_rate": 2.6702024682145047e-06, "loss": 0.1633, "step": 2238 }, { "epoch": 3.84, "learning_rate": 2.6626341208140603e-06, "loss": 0.1643, "step": 2239 }, { "epoch": 3.85, "learning_rate": 2.655074866804451e-06, "loss": 0.1601, "step": 2240 }, { "epoch": 3.85, "learning_rate": 2.6475247155540583e-06, "loss": 0.1629, "step": 2241 }, { "epoch": 3.85, "learning_rate": 2.639983676419985e-06, "loss": 0.1705, "step": 2242 }, { "epoch": 3.85, "learning_rate": 2.632451758748038e-06, "loss": 0.1694, "step": 2243 }, { "epoch": 3.85, "learning_rate": 2.6249289718727224e-06, "loss": 0.1654, "step": 2244 }, { "epoch": 3.85, "learning_rate": 2.6174153251172253e-06, "loss": 0.1644, "step": 2245 }, { "epoch": 3.86, "learning_rate": 2.6099108277934105e-06, "loss": 0.1652, "step": 2246 }, { "epoch": 3.86, "learning_rate": 2.6024154892017938e-06, "loss": 0.1705, "step": 2247 }, { "epoch": 3.86, "learning_rate": 2.594929318631547e-06, "loss": 0.1687, "step": 2248 }, { "epoch": 3.86, "learning_rate": 2.587452325360481e-06, "loss": 0.1684, "step": 2249 }, { "epoch": 3.86, "learning_rate": 2.5799845186550286e-06, "loss": 0.1676, "step": 2250 }, { "epoch": 3.86, "learning_rate": 2.57252590777024e-06, "loss": 0.1677, "step": 2251 }, { "epoch": 3.87, "learning_rate": 2.565076501949769e-06, "loss": 0.1651, "step": 2252 }, { "epoch": 3.87, "learning_rate": 2.5576363104258616e-06, "loss": 0.1672, "step": 2253 }, { "epoch": 3.87, "learning_rate": 2.5502053424193383e-06, "loss": 0.165, "step": 2254 }, { "epoch": 3.87, "learning_rate": 2.5427836071395973e-06, "loss": 0.1699, "step": 2255 }, { "epoch": 3.87, "learning_rate": 2.5353711137845893e-06, "loss": 0.173, "step": 2256 }, { "epoch": 3.87, "learning_rate": 2.527967871540811e-06, "loss": 0.1704, "step": 2257 }, { "epoch": 3.88, "learning_rate": 2.5205738895833e-06, "loss": 0.1631, "step": 2258 }, { "epoch": 3.88, "learning_rate": 2.5131891770756133e-06, "loss": 0.1672, "step": 2259 }, { "epoch": 3.88, "learning_rate": 2.505813743169815e-06, "loss": 0.1648, "step": 2260 }, { "epoch": 3.88, "learning_rate": 2.498447597006476e-06, "loss": 0.1692, "step": 2261 }, { "epoch": 3.88, "learning_rate": 2.491090747714655e-06, "loss": 0.1659, "step": 2262 }, { "epoch": 3.88, "learning_rate": 2.4837432044118914e-06, "loss": 0.1642, "step": 2263 }, { "epoch": 3.89, "learning_rate": 2.4764049762041874e-06, "loss": 0.1708, "step": 2264 }, { "epoch": 3.89, "learning_rate": 2.4690760721860053e-06, "loss": 0.1687, "step": 2265 }, { "epoch": 3.89, "learning_rate": 2.4617565014402445e-06, "loss": 0.1691, "step": 2266 }, { "epoch": 3.89, "learning_rate": 2.4544462730382433e-06, "loss": 0.1664, "step": 2267 }, { "epoch": 3.89, "learning_rate": 2.4471453960397616e-06, "loss": 0.1631, "step": 2268 }, { "epoch": 3.9, "learning_rate": 2.4398538794929683e-06, "loss": 0.1655, "step": 2269 }, { "epoch": 3.9, "learning_rate": 2.432571732434431e-06, "loss": 0.1651, "step": 2270 }, { "epoch": 3.9, "learning_rate": 2.425298963889107e-06, "loss": 0.1683, "step": 2271 }, { "epoch": 3.9, "learning_rate": 2.4180355828703307e-06, "loss": 0.167, "step": 2272 }, { "epoch": 3.9, "learning_rate": 2.4107815983798e-06, "loss": 0.1682, "step": 2273 }, { "epoch": 3.9, "learning_rate": 2.4035370194075703e-06, "loss": 0.1638, "step": 2274 }, { "epoch": 3.91, "learning_rate": 2.396301854932038e-06, "loss": 0.1697, "step": 2275 }, { "epoch": 3.91, "learning_rate": 2.3890761139199347e-06, "loss": 0.1685, "step": 2276 }, { "epoch": 3.91, "learning_rate": 2.3818598053263108e-06, "loss": 0.1657, "step": 2277 }, { "epoch": 3.91, "learning_rate": 2.3746529380945294e-06, "loss": 0.1652, "step": 2278 }, { "epoch": 3.91, "learning_rate": 2.3674555211562468e-06, "loss": 0.1695, "step": 2279 }, { "epoch": 3.91, "learning_rate": 2.360267563431413e-06, "loss": 0.1683, "step": 2280 }, { "epoch": 3.92, "learning_rate": 2.353089073828255e-06, "loss": 0.1673, "step": 2281 }, { "epoch": 3.92, "learning_rate": 2.3459200612432633e-06, "loss": 0.1709, "step": 2282 }, { "epoch": 3.92, "learning_rate": 2.338760534561183e-06, "loss": 0.1673, "step": 2283 }, { "epoch": 3.92, "learning_rate": 2.331610502655005e-06, "loss": 0.1644, "step": 2284 }, { "epoch": 3.92, "learning_rate": 2.3244699743859523e-06, "loss": 0.1607, "step": 2285 }, { "epoch": 3.92, "learning_rate": 2.3173389586034702e-06, "loss": 0.165, "step": 2286 }, { "epoch": 3.93, "learning_rate": 2.3102174641452134e-06, "loss": 0.1624, "step": 2287 }, { "epoch": 3.93, "learning_rate": 2.3031054998370373e-06, "loss": 0.1623, "step": 2288 }, { "epoch": 3.93, "learning_rate": 2.2960030744929885e-06, "loss": 0.1665, "step": 2289 }, { "epoch": 3.93, "learning_rate": 2.288910196915288e-06, "loss": 0.1619, "step": 2290 }, { "epoch": 3.93, "learning_rate": 2.2818268758943297e-06, "loss": 0.1631, "step": 2291 }, { "epoch": 3.93, "learning_rate": 2.274753120208654e-06, "loss": 0.1655, "step": 2292 }, { "epoch": 3.94, "learning_rate": 2.2676889386249555e-06, "loss": 0.164, "step": 2293 }, { "epoch": 3.94, "learning_rate": 2.26063433989806e-06, "loss": 0.163, "step": 2294 }, { "epoch": 3.94, "learning_rate": 2.253589332770919e-06, "loss": 0.1628, "step": 2295 }, { "epoch": 3.94, "learning_rate": 2.246553925974594e-06, "loss": 0.1621, "step": 2296 }, { "epoch": 3.94, "learning_rate": 2.2395281282282532e-06, "loss": 0.165, "step": 2297 }, { "epoch": 3.95, "learning_rate": 2.2325119482391466e-06, "loss": 0.1691, "step": 2298 }, { "epoch": 3.95, "learning_rate": 2.2255053947026164e-06, "loss": 0.1694, "step": 2299 }, { "epoch": 3.95, "learning_rate": 2.218508476302065e-06, "loss": 0.1645, "step": 2300 }, { "epoch": 3.95, "learning_rate": 2.2115212017089573e-06, "loss": 0.1681, "step": 2301 }, { "epoch": 3.95, "learning_rate": 2.204543579582813e-06, "loss": 0.1645, "step": 2302 }, { "epoch": 3.95, "learning_rate": 2.1975756185711782e-06, "loss": 0.1685, "step": 2303 }, { "epoch": 3.96, "learning_rate": 2.1906173273096343e-06, "loss": 0.1665, "step": 2304 }, { "epoch": 3.96, "learning_rate": 2.1836687144217705e-06, "loss": 0.1641, "step": 2305 }, { "epoch": 3.96, "learning_rate": 2.176729788519186e-06, "loss": 0.1618, "step": 2306 }, { "epoch": 3.96, "learning_rate": 2.169800558201477e-06, "loss": 0.1614, "step": 2307 }, { "epoch": 3.96, "learning_rate": 2.162881032056221e-06, "loss": 0.1629, "step": 2308 }, { "epoch": 3.96, "learning_rate": 2.1559712186589664e-06, "loss": 0.1656, "step": 2309 }, { "epoch": 3.97, "learning_rate": 2.1490711265732333e-06, "loss": 0.1675, "step": 2310 }, { "epoch": 3.97, "learning_rate": 2.142180764350479e-06, "loss": 0.1683, "step": 2311 }, { "epoch": 3.97, "learning_rate": 2.1353001405301154e-06, "loss": 0.1642, "step": 2312 }, { "epoch": 3.97, "learning_rate": 2.128429263639481e-06, "loss": 0.1652, "step": 2313 }, { "epoch": 3.97, "learning_rate": 2.1215681421938337e-06, "loss": 0.1596, "step": 2314 }, { "epoch": 3.97, "learning_rate": 2.114716784696342e-06, "loss": 0.159, "step": 2315 }, { "epoch": 3.98, "learning_rate": 2.107875199638075e-06, "loss": 0.1611, "step": 2316 }, { "epoch": 3.98, "learning_rate": 2.101043395497989e-06, "loss": 0.1649, "step": 2317 }, { "epoch": 3.98, "learning_rate": 2.0942213807429168e-06, "loss": 0.159, "step": 2318 }, { "epoch": 3.98, "learning_rate": 2.087409163827564e-06, "loss": 0.1677, "step": 2319 }, { "epoch": 3.98, "learning_rate": 2.0806067531944875e-06, "loss": 0.1669, "step": 2320 }, { "epoch": 3.98, "learning_rate": 2.0738141572740964e-06, "loss": 0.1635, "step": 2321 }, { "epoch": 3.99, "learning_rate": 2.0670313844846335e-06, "loss": 0.1644, "step": 2322 }, { "epoch": 3.99, "learning_rate": 2.060258443232168e-06, "loss": 0.1644, "step": 2323 }, { "epoch": 3.99, "learning_rate": 2.053495341910583e-06, "loss": 0.1643, "step": 2324 }, { "epoch": 3.99, "learning_rate": 2.0467420889015675e-06, "loss": 0.1631, "step": 2325 }, { "epoch": 3.99, "learning_rate": 2.0399986925746075e-06, "loss": 0.1668, "step": 2326 }, { "epoch": 3.99, "learning_rate": 2.033265161286971e-06, "loss": 0.1688, "step": 2327 }, { "epoch": 4.0, "learning_rate": 2.026541503383702e-06, "loss": 0.1686, "step": 2328 }, { "epoch": 4.0, "learning_rate": 2.019827727197605e-06, "loss": 0.1656, "step": 2329 }, { "epoch": 4.0, "learning_rate": 2.0131238410492417e-06, "loss": 0.1683, "step": 2330 }, { "epoch": 4.0, "learning_rate": 2.006429853246913e-06, "loss": 0.15, "step": 2331 }, { "epoch": 4.0, "learning_rate": 1.9997457720866554e-06, "loss": 0.1552, "step": 2332 }, { "epoch": 4.01, "learning_rate": 1.993071605852226e-06, "loss": 0.1537, "step": 2333 }, { "epoch": 4.01, "learning_rate": 1.986407362815096e-06, "loss": 0.1526, "step": 2334 }, { "epoch": 4.01, "learning_rate": 1.9797530512344385e-06, "loss": 0.1544, "step": 2335 }, { "epoch": 4.01, "learning_rate": 1.973108679357113e-06, "loss": 0.1543, "step": 2336 }, { "epoch": 4.01, "learning_rate": 1.9664742554176673e-06, "loss": 0.1559, "step": 2337 }, { "epoch": 4.01, "learning_rate": 1.959849787638317e-06, "loss": 0.1538, "step": 2338 }, { "epoch": 4.02, "learning_rate": 1.95323528422894e-06, "loss": 0.1481, "step": 2339 }, { "epoch": 4.02, "learning_rate": 1.946630753387064e-06, "loss": 0.1545, "step": 2340 }, { "epoch": 4.02, "learning_rate": 1.9400362032978593e-06, "loss": 0.1498, "step": 2341 }, { "epoch": 4.02, "learning_rate": 1.9334516421341276e-06, "loss": 0.1499, "step": 2342 }, { "epoch": 4.02, "learning_rate": 1.926877078056284e-06, "loss": 0.1486, "step": 2343 }, { "epoch": 4.02, "learning_rate": 1.920312519212358e-06, "loss": 0.15, "step": 2344 }, { "epoch": 4.03, "learning_rate": 1.9137579737379865e-06, "loss": 0.1505, "step": 2345 }, { "epoch": 4.03, "learning_rate": 1.9072134497563876e-06, "loss": 0.1492, "step": 2346 }, { "epoch": 4.03, "learning_rate": 1.9006789553783623e-06, "loss": 0.1551, "step": 2347 }, { "epoch": 4.03, "learning_rate": 1.894154498702283e-06, "loss": 0.1518, "step": 2348 }, { "epoch": 4.03, "learning_rate": 1.8876400878140776e-06, "loss": 0.1548, "step": 2349 }, { "epoch": 4.03, "learning_rate": 1.8811357307872291e-06, "loss": 0.1557, "step": 2350 }, { "epoch": 4.04, "learning_rate": 1.874641435682758e-06, "loss": 0.1574, "step": 2351 }, { "epoch": 4.04, "learning_rate": 1.8681572105492153e-06, "loss": 0.1501, "step": 2352 }, { "epoch": 4.04, "learning_rate": 1.8616830634226713e-06, "loss": 0.1502, "step": 2353 }, { "epoch": 4.04, "learning_rate": 1.8552190023267113e-06, "loss": 0.1508, "step": 2354 }, { "epoch": 4.04, "learning_rate": 1.8487650352724118e-06, "loss": 0.1567, "step": 2355 }, { "epoch": 4.04, "learning_rate": 1.8423211702583443e-06, "loss": 0.1558, "step": 2356 }, { "epoch": 4.05, "learning_rate": 1.8358874152705619e-06, "loss": 0.1562, "step": 2357 }, { "epoch": 4.05, "learning_rate": 1.8294637782825875e-06, "loss": 0.1566, "step": 2358 }, { "epoch": 4.05, "learning_rate": 1.823050267255402e-06, "loss": 0.1553, "step": 2359 }, { "epoch": 4.05, "learning_rate": 1.816646890137439e-06, "loss": 0.1522, "step": 2360 }, { "epoch": 4.05, "learning_rate": 1.8102536548645734e-06, "loss": 0.1514, "step": 2361 }, { "epoch": 4.05, "learning_rate": 1.803870569360109e-06, "loss": 0.1597, "step": 2362 }, { "epoch": 4.06, "learning_rate": 1.7974976415347723e-06, "loss": 0.1515, "step": 2363 }, { "epoch": 4.06, "learning_rate": 1.7911348792867e-06, "loss": 0.1538, "step": 2364 }, { "epoch": 4.06, "learning_rate": 1.7847822905014322e-06, "loss": 0.1571, "step": 2365 }, { "epoch": 4.06, "learning_rate": 1.7784398830519002e-06, "loss": 0.1568, "step": 2366 }, { "epoch": 4.06, "learning_rate": 1.7721076647984158e-06, "loss": 0.1531, "step": 2367 }, { "epoch": 4.07, "learning_rate": 1.7657856435886623e-06, "loss": 0.1488, "step": 2368 }, { "epoch": 4.07, "learning_rate": 1.7594738272576883e-06, "loss": 0.1524, "step": 2369 }, { "epoch": 4.07, "learning_rate": 1.7531722236278936e-06, "loss": 0.1556, "step": 2370 }, { "epoch": 4.07, "learning_rate": 1.746880840509022e-06, "loss": 0.1525, "step": 2371 }, { "epoch": 4.07, "learning_rate": 1.7405996856981512e-06, "loss": 0.1536, "step": 2372 }, { "epoch": 4.07, "learning_rate": 1.7343287669796794e-06, "loss": 0.1547, "step": 2373 }, { "epoch": 4.08, "learning_rate": 1.7280680921253246e-06, "loss": 0.1485, "step": 2374 }, { "epoch": 4.08, "learning_rate": 1.7218176688941035e-06, "loss": 0.1525, "step": 2375 }, { "epoch": 4.08, "learning_rate": 1.715577505032332e-06, "loss": 0.1569, "step": 2376 }, { "epoch": 4.08, "learning_rate": 1.7093476082736082e-06, "loss": 0.1506, "step": 2377 }, { "epoch": 4.08, "learning_rate": 1.7031279863388084e-06, "loss": 0.1492, "step": 2378 }, { "epoch": 4.08, "learning_rate": 1.6969186469360731e-06, "loss": 0.1508, "step": 2379 }, { "epoch": 4.09, "learning_rate": 1.6907195977608037e-06, "loss": 0.1546, "step": 2380 }, { "epoch": 4.09, "learning_rate": 1.684530846495639e-06, "loss": 0.1593, "step": 2381 }, { "epoch": 4.09, "learning_rate": 1.6783524008104646e-06, "loss": 0.1533, "step": 2382 }, { "epoch": 4.09, "learning_rate": 1.6721842683623911e-06, "loss": 0.1543, "step": 2383 }, { "epoch": 4.09, "learning_rate": 1.6660264567957473e-06, "loss": 0.153, "step": 2384 }, { "epoch": 4.09, "learning_rate": 1.6598789737420706e-06, "loss": 0.1506, "step": 2385 }, { "epoch": 4.1, "learning_rate": 1.6537418268201033e-06, "loss": 0.1521, "step": 2386 }, { "epoch": 4.1, "learning_rate": 1.6476150236357656e-06, "loss": 0.1542, "step": 2387 }, { "epoch": 4.1, "learning_rate": 1.6414985717821675e-06, "loss": 0.1564, "step": 2388 }, { "epoch": 4.1, "learning_rate": 1.6353924788395948e-06, "loss": 0.1594, "step": 2389 }, { "epoch": 4.1, "learning_rate": 1.6292967523754856e-06, "loss": 0.1517, "step": 2390 }, { "epoch": 4.1, "learning_rate": 1.6232113999444332e-06, "loss": 0.1543, "step": 2391 }, { "epoch": 4.11, "learning_rate": 1.6171364290881808e-06, "loss": 0.1523, "step": 2392 }, { "epoch": 4.11, "learning_rate": 1.6110718473355912e-06, "loss": 0.151, "step": 2393 }, { "epoch": 4.11, "learning_rate": 1.6050176622026658e-06, "loss": 0.1515, "step": 2394 }, { "epoch": 4.11, "learning_rate": 1.5989738811925148e-06, "loss": 0.1563, "step": 2395 }, { "epoch": 4.11, "learning_rate": 1.5929405117953556e-06, "loss": 0.1486, "step": 2396 }, { "epoch": 4.12, "learning_rate": 1.5869175614885003e-06, "loss": 0.1513, "step": 2397 }, { "epoch": 4.12, "learning_rate": 1.580905037736351e-06, "loss": 0.1545, "step": 2398 }, { "epoch": 4.12, "learning_rate": 1.5749029479903888e-06, "loss": 0.1569, "step": 2399 }, { "epoch": 4.12, "learning_rate": 1.5689112996891576e-06, "loss": 0.1523, "step": 2400 }, { "epoch": 4.12, "learning_rate": 1.5629301002582676e-06, "loss": 0.1554, "step": 2401 }, { "epoch": 4.12, "learning_rate": 1.5569593571103747e-06, "loss": 0.1535, "step": 2402 }, { "epoch": 4.13, "learning_rate": 1.55099907764518e-06, "loss": 0.1582, "step": 2403 }, { "epoch": 4.13, "learning_rate": 1.5450492692494145e-06, "loss": 0.1515, "step": 2404 }, { "epoch": 4.13, "learning_rate": 1.5391099392968333e-06, "loss": 0.1523, "step": 2405 }, { "epoch": 4.13, "learning_rate": 1.533181095148203e-06, "loss": 0.1505, "step": 2406 }, { "epoch": 4.13, "learning_rate": 1.5272627441512965e-06, "loss": 0.1485, "step": 2407 }, { "epoch": 4.13, "learning_rate": 1.5213548936408828e-06, "loss": 0.1553, "step": 2408 }, { "epoch": 4.14, "learning_rate": 1.515457550938717e-06, "loss": 0.1479, "step": 2409 }, { "epoch": 4.14, "learning_rate": 1.5095707233535307e-06, "loss": 0.152, "step": 2410 }, { "epoch": 4.14, "learning_rate": 1.5036944181810243e-06, "loss": 0.154, "step": 2411 }, { "epoch": 4.14, "learning_rate": 1.4978286427038602e-06, "loss": 0.1563, "step": 2412 }, { "epoch": 4.14, "learning_rate": 1.4919734041916435e-06, "loss": 0.1519, "step": 2413 }, { "epoch": 4.14, "learning_rate": 1.486128709900928e-06, "loss": 0.1559, "step": 2414 }, { "epoch": 4.15, "learning_rate": 1.4802945670751978e-06, "loss": 0.1561, "step": 2415 }, { "epoch": 4.15, "learning_rate": 1.474470982944859e-06, "loss": 0.1522, "step": 2416 }, { "epoch": 4.15, "learning_rate": 1.4686579647272337e-06, "loss": 0.1563, "step": 2417 }, { "epoch": 4.15, "learning_rate": 1.462855519626548e-06, "loss": 0.1551, "step": 2418 }, { "epoch": 4.15, "learning_rate": 1.4570636548339245e-06, "loss": 0.158, "step": 2419 }, { "epoch": 4.15, "learning_rate": 1.4512823775273743e-06, "loss": 0.1527, "step": 2420 }, { "epoch": 4.16, "learning_rate": 1.4455116948717873e-06, "loss": 0.1573, "step": 2421 }, { "epoch": 4.16, "learning_rate": 1.4397516140189216e-06, "loss": 0.1513, "step": 2422 }, { "epoch": 4.16, "learning_rate": 1.434002142107398e-06, "loss": 0.1529, "step": 2423 }, { "epoch": 4.16, "learning_rate": 1.4282632862626899e-06, "loss": 0.1547, "step": 2424 }, { "epoch": 4.16, "learning_rate": 1.4225350535971095e-06, "loss": 0.1545, "step": 2425 }, { "epoch": 4.16, "learning_rate": 1.4168174512098065e-06, "loss": 0.1543, "step": 2426 }, { "epoch": 4.17, "learning_rate": 1.4111104861867564e-06, "loss": 0.1581, "step": 2427 }, { "epoch": 4.17, "learning_rate": 1.4054141656007536e-06, "loss": 0.1555, "step": 2428 }, { "epoch": 4.17, "learning_rate": 1.3997284965113967e-06, "loss": 0.1529, "step": 2429 }, { "epoch": 4.17, "learning_rate": 1.3940534859650844e-06, "loss": 0.1581, "step": 2430 }, { "epoch": 4.17, "learning_rate": 1.3883891409950112e-06, "loss": 0.1556, "step": 2431 }, { "epoch": 4.18, "learning_rate": 1.3827354686211403e-06, "loss": 0.1492, "step": 2432 }, { "epoch": 4.18, "learning_rate": 1.3770924758502246e-06, "loss": 0.1557, "step": 2433 }, { "epoch": 4.18, "learning_rate": 1.3714601696757713e-06, "loss": 0.1534, "step": 2434 }, { "epoch": 4.18, "learning_rate": 1.3658385570780453e-06, "loss": 0.158, "step": 2435 }, { "epoch": 4.18, "learning_rate": 1.3602276450240603e-06, "loss": 0.1584, "step": 2436 }, { "epoch": 4.18, "learning_rate": 1.3546274404675685e-06, "loss": 0.1539, "step": 2437 }, { "epoch": 4.19, "learning_rate": 1.349037950349047e-06, "loss": 0.1524, "step": 2438 }, { "epoch": 4.19, "learning_rate": 1.3434591815956998e-06, "loss": 0.1534, "step": 2439 }, { "epoch": 4.19, "learning_rate": 1.3378911411214434e-06, "loss": 0.1514, "step": 2440 }, { "epoch": 4.19, "learning_rate": 1.3323338358268955e-06, "loss": 0.153, "step": 2441 }, { "epoch": 4.19, "learning_rate": 1.3267872725993712e-06, "loss": 0.1538, "step": 2442 }, { "epoch": 4.19, "learning_rate": 1.3212514583128767e-06, "loss": 0.1538, "step": 2443 }, { "epoch": 4.2, "learning_rate": 1.3157263998280845e-06, "loss": 0.1508, "step": 2444 }, { "epoch": 4.2, "learning_rate": 1.310212103992351e-06, "loss": 0.1511, "step": 2445 }, { "epoch": 4.2, "learning_rate": 1.304708577639685e-06, "loss": 0.1551, "step": 2446 }, { "epoch": 4.2, "learning_rate": 1.2992158275907541e-06, "loss": 0.151, "step": 2447 }, { "epoch": 4.2, "learning_rate": 1.2937338606528648e-06, "loss": 0.1528, "step": 2448 }, { "epoch": 4.2, "learning_rate": 1.2882626836199685e-06, "loss": 0.1548, "step": 2449 }, { "epoch": 4.21, "learning_rate": 1.2828023032726377e-06, "loss": 0.1534, "step": 2450 }, { "epoch": 4.21, "learning_rate": 1.2773527263780626e-06, "loss": 0.1596, "step": 2451 }, { "epoch": 4.21, "learning_rate": 1.2719139596900487e-06, "loss": 0.1578, "step": 2452 }, { "epoch": 4.21, "learning_rate": 1.2664860099490028e-06, "loss": 0.1525, "step": 2453 }, { "epoch": 4.21, "learning_rate": 1.2610688838819262e-06, "loss": 0.1554, "step": 2454 }, { "epoch": 4.21, "learning_rate": 1.2556625882024065e-06, "loss": 0.1517, "step": 2455 }, { "epoch": 4.22, "learning_rate": 1.2502671296106095e-06, "loss": 0.1511, "step": 2456 }, { "epoch": 4.22, "learning_rate": 1.2448825147932652e-06, "loss": 0.1549, "step": 2457 }, { "epoch": 4.22, "learning_rate": 1.2395087504236714e-06, "loss": 0.1548, "step": 2458 }, { "epoch": 4.22, "learning_rate": 1.2341458431616759e-06, "loss": 0.154, "step": 2459 }, { "epoch": 4.22, "learning_rate": 1.2287937996536691e-06, "loss": 0.1488, "step": 2460 }, { "epoch": 4.22, "learning_rate": 1.2234526265325818e-06, "loss": 0.1454, "step": 2461 }, { "epoch": 4.23, "learning_rate": 1.2181223304178703e-06, "loss": 0.1529, "step": 2462 }, { "epoch": 4.23, "learning_rate": 1.2128029179155087e-06, "loss": 0.1492, "step": 2463 }, { "epoch": 4.23, "learning_rate": 1.2074943956179884e-06, "loss": 0.1529, "step": 2464 }, { "epoch": 4.23, "learning_rate": 1.202196770104298e-06, "loss": 0.1512, "step": 2465 }, { "epoch": 4.23, "learning_rate": 1.1969100479399254e-06, "loss": 0.1475, "step": 2466 }, { "epoch": 4.24, "learning_rate": 1.1916342356768462e-06, "loss": 0.1505, "step": 2467 }, { "epoch": 4.24, "learning_rate": 1.1863693398535115e-06, "loss": 0.1593, "step": 2468 }, { "epoch": 4.24, "learning_rate": 1.1811153669948472e-06, "loss": 0.1524, "step": 2469 }, { "epoch": 4.24, "learning_rate": 1.1758723236122383e-06, "loss": 0.1578, "step": 2470 }, { "epoch": 4.24, "learning_rate": 1.1706402162035268e-06, "loss": 0.156, "step": 2471 }, { "epoch": 4.24, "learning_rate": 1.1654190512530016e-06, "loss": 0.1531, "step": 2472 }, { "epoch": 4.25, "learning_rate": 1.160208835231391e-06, "loss": 0.1539, "step": 2473 }, { "epoch": 4.25, "learning_rate": 1.1550095745958523e-06, "loss": 0.1566, "step": 2474 }, { "epoch": 4.25, "learning_rate": 1.1498212757899675e-06, "loss": 0.155, "step": 2475 }, { "epoch": 4.25, "learning_rate": 1.1446439452437275e-06, "loss": 0.1532, "step": 2476 }, { "epoch": 4.25, "learning_rate": 1.1394775893735387e-06, "loss": 0.1561, "step": 2477 }, { "epoch": 4.25, "learning_rate": 1.1343222145822009e-06, "loss": 0.1554, "step": 2478 }, { "epoch": 4.26, "learning_rate": 1.129177827258906e-06, "loss": 0.1618, "step": 2479 }, { "epoch": 4.26, "learning_rate": 1.124044433779229e-06, "loss": 0.1473, "step": 2480 }, { "epoch": 4.26, "learning_rate": 1.1189220405051204e-06, "loss": 0.1536, "step": 2481 }, { "epoch": 4.26, "learning_rate": 1.1138106537848946e-06, "loss": 0.1533, "step": 2482 }, { "epoch": 4.26, "learning_rate": 1.1087102799532301e-06, "loss": 0.1489, "step": 2483 }, { "epoch": 4.26, "learning_rate": 1.1036209253311525e-06, "loss": 0.1537, "step": 2484 }, { "epoch": 4.27, "learning_rate": 1.0985425962260342e-06, "loss": 0.1525, "step": 2485 }, { "epoch": 4.27, "learning_rate": 1.0934752989315833e-06, "loss": 0.147, "step": 2486 }, { "epoch": 4.27, "learning_rate": 1.0884190397278327e-06, "loss": 0.1484, "step": 2487 }, { "epoch": 4.27, "learning_rate": 1.0833738248811421e-06, "loss": 0.1491, "step": 2488 }, { "epoch": 4.27, "learning_rate": 1.0783396606441721e-06, "loss": 0.1511, "step": 2489 }, { "epoch": 4.27, "learning_rate": 1.0733165532558998e-06, "loss": 0.147, "step": 2490 }, { "epoch": 4.28, "learning_rate": 1.0683045089415945e-06, "loss": 0.1516, "step": 2491 }, { "epoch": 4.28, "learning_rate": 1.0633035339128106e-06, "loss": 0.1538, "step": 2492 }, { "epoch": 4.28, "learning_rate": 1.0583136343673927e-06, "loss": 0.1515, "step": 2493 }, { "epoch": 4.28, "learning_rate": 1.0533348164894574e-06, "loss": 0.1551, "step": 2494 }, { "epoch": 4.28, "learning_rate": 1.0483670864493777e-06, "loss": 0.1562, "step": 2495 }, { "epoch": 4.28, "learning_rate": 1.0434104504037957e-06, "loss": 0.1504, "step": 2496 }, { "epoch": 4.29, "learning_rate": 1.0384649144956005e-06, "loss": 0.1596, "step": 2497 }, { "epoch": 4.29, "learning_rate": 1.0335304848539263e-06, "loss": 0.1621, "step": 2498 }, { "epoch": 4.29, "learning_rate": 1.02860716759414e-06, "loss": 0.1561, "step": 2499 }, { "epoch": 4.29, "learning_rate": 1.02369496881784e-06, "loss": 0.1498, "step": 2500 }, { "epoch": 4.24, "learning_rate": 4.18972183404194e-06, "loss": 0.1731, "step": 2501 }, { "epoch": 4.24, "learning_rate": 4.182276289358331e-06, "loss": 0.1687, "step": 2502 }, { "epoch": 4.24, "learning_rate": 4.1748356166514e-06, "loss": 0.1686, "step": 2503 }, { "epoch": 4.24, "learning_rate": 4.16739982215224e-06, "loss": 0.2343, "step": 2504 }, { "epoch": 4.25, "learning_rate": 4.159968912087863e-06, "loss": 0.1685, "step": 2505 }, { "epoch": 4.25, "learning_rate": 4.152542892681185e-06, "loss": 0.1538, "step": 2506 }, { "epoch": 4.25, "learning_rate": 4.145121770151024e-06, "loss": 0.1579, "step": 2507 }, { "epoch": 4.25, "learning_rate": 4.137705550712106e-06, "loss": 0.1706, "step": 2508 }, { "epoch": 4.25, "learning_rate": 4.130294240575047e-06, "loss": 0.1683, "step": 2509 }, { "epoch": 4.25, "learning_rate": 4.122887845946352e-06, "loss": 0.1705, "step": 2510 }, { "epoch": 4.26, "learning_rate": 4.11548637302841e-06, "loss": 0.1894, "step": 2511 }, { "epoch": 4.26, "learning_rate": 4.108089828019491e-06, "loss": 0.1666, "step": 2512 }, { "epoch": 4.26, "learning_rate": 4.100698217113725e-06, "loss": 0.1809, "step": 2513 }, { "epoch": 4.26, "learning_rate": 4.093311546501129e-06, "loss": 0.1776, "step": 2514 }, { "epoch": 4.26, "learning_rate": 4.085929822367569e-06, "loss": 0.1721, "step": 2515 }, { "epoch": 4.26, "learning_rate": 4.078553050894776e-06, "loss": 0.1662, "step": 2516 }, { "epoch": 4.27, "learning_rate": 4.07118123826033e-06, "loss": 0.1735, "step": 2517 }, { "epoch": 4.27, "learning_rate": 4.0638143906376615e-06, "loss": 0.1832, "step": 2518 }, { "epoch": 4.27, "learning_rate": 4.056452514196037e-06, "loss": 0.1696, "step": 2519 }, { "epoch": 4.27, "learning_rate": 4.049095615100564e-06, "loss": 0.1726, "step": 2520 }, { "epoch": 4.27, "learning_rate": 4.041743699512181e-06, "loss": 0.1697, "step": 2521 }, { "epoch": 4.27, "learning_rate": 4.034396773587657e-06, "loss": 0.1788, "step": 2522 }, { "epoch": 4.28, "learning_rate": 4.027054843479576e-06, "loss": 0.1822, "step": 2523 }, { "epoch": 4.28, "learning_rate": 4.019717915336344e-06, "loss": 0.1848, "step": 2524 }, { "epoch": 4.28, "learning_rate": 4.012385995302177e-06, "loss": 0.1647, "step": 2525 }, { "epoch": 4.28, "learning_rate": 4.005059089517089e-06, "loss": 0.1706, "step": 2526 }, { "epoch": 4.28, "learning_rate": 3.997737204116906e-06, "loss": 0.1734, "step": 2527 }, { "epoch": 4.28, "learning_rate": 3.9904203452332444e-06, "loss": 0.1792, "step": 2528 }, { "epoch": 4.29, "learning_rate": 3.983108518993512e-06, "loss": 0.2064, "step": 2529 }, { "epoch": 4.29, "learning_rate": 3.975801731520902e-06, "loss": 0.181, "step": 2530 }, { "epoch": 4.29, "learning_rate": 3.968499988934392e-06, "loss": 0.1764, "step": 2531 }, { "epoch": 4.29, "learning_rate": 3.961203297348723e-06, "loss": 0.1687, "step": 2532 }, { "epoch": 4.29, "learning_rate": 3.95391166287442e-06, "loss": 0.1757, "step": 2533 }, { "epoch": 4.29, "learning_rate": 3.946625091617764e-06, "loss": 0.1751, "step": 2534 }, { "epoch": 4.3, "learning_rate": 3.9393435896807995e-06, "loss": 0.1718, "step": 2535 }, { "epoch": 4.3, "learning_rate": 3.932067163161327e-06, "loss": 0.1799, "step": 2536 }, { "epoch": 4.3, "learning_rate": 3.924795818152895e-06, "loss": 0.164, "step": 2537 }, { "epoch": 4.3, "learning_rate": 3.917529560744794e-06, "loss": 0.175, "step": 2538 }, { "epoch": 4.3, "learning_rate": 3.910268397022054e-06, "loss": 0.173, "step": 2539 }, { "epoch": 4.3, "learning_rate": 3.903012333065447e-06, "loss": 0.172, "step": 2540 }, { "epoch": 4.31, "learning_rate": 3.895761374951465e-06, "loss": 0.1675, "step": 2541 }, { "epoch": 4.31, "learning_rate": 3.888515528752331e-06, "loss": 0.1682, "step": 2542 }, { "epoch": 4.31, "learning_rate": 3.881274800535986e-06, "loss": 0.1708, "step": 2543 }, { "epoch": 4.31, "learning_rate": 3.874039196366076e-06, "loss": 0.1642, "step": 2544 }, { "epoch": 4.31, "learning_rate": 3.866808722301968e-06, "loss": 0.1752, "step": 2545 }, { "epoch": 4.31, "learning_rate": 3.859583384398727e-06, "loss": 0.1621, "step": 2546 }, { "epoch": 4.32, "learning_rate": 3.852363188707118e-06, "loss": 0.1663, "step": 2547 }, { "epoch": 4.32, "learning_rate": 3.845148141273599e-06, "loss": 0.1804, "step": 2548 }, { "epoch": 4.32, "learning_rate": 3.837938248140322e-06, "loss": 0.1702, "step": 2549 }, { "epoch": 4.32, "learning_rate": 3.830733515345109e-06, "loss": 0.1785, "step": 2550 }, { "epoch": 4.32, "learning_rate": 3.823533948921475e-06, "loss": 0.1679, "step": 2551 }, { "epoch": 4.33, "learning_rate": 3.816339554898601e-06, "loss": 0.1714, "step": 2552 }, { "epoch": 4.33, "learning_rate": 3.8091503393013373e-06, "loss": 0.163, "step": 2553 }, { "epoch": 4.33, "learning_rate": 3.801966308150202e-06, "loss": 0.166, "step": 2554 }, { "epoch": 4.33, "learning_rate": 3.7947874674613683e-06, "loss": 0.1702, "step": 2555 }, { "epoch": 4.33, "learning_rate": 3.787613823246661e-06, "loss": 0.1626, "step": 2556 }, { "epoch": 4.33, "learning_rate": 3.78044538151355e-06, "loss": 0.1649, "step": 2557 }, { "epoch": 4.34, "learning_rate": 3.7732821482651573e-06, "loss": 0.1629, "step": 2558 }, { "epoch": 4.34, "learning_rate": 3.766124129500238e-06, "loss": 0.1807, "step": 2559 }, { "epoch": 4.34, "learning_rate": 3.7589713312131836e-06, "loss": 0.173, "step": 2560 }, { "epoch": 4.34, "learning_rate": 3.751823759394009e-06, "loss": 0.1755, "step": 2561 }, { "epoch": 4.34, "learning_rate": 3.7446814200283598e-06, "loss": 0.1668, "step": 2562 }, { "epoch": 4.34, "learning_rate": 3.737544319097487e-06, "loss": 0.1705, "step": 2563 }, { "epoch": 4.35, "learning_rate": 3.7304124625782656e-06, "loss": 0.1654, "step": 2564 }, { "epoch": 4.35, "learning_rate": 3.7232858564431773e-06, "loss": 0.1548, "step": 2565 }, { "epoch": 4.35, "learning_rate": 3.7161645066603036e-06, "loss": 0.1668, "step": 2566 }, { "epoch": 4.35, "learning_rate": 3.7090484191933263e-06, "loss": 0.1606, "step": 2567 }, { "epoch": 4.35, "learning_rate": 3.701937600001524e-06, "loss": 0.1618, "step": 2568 }, { "epoch": 4.35, "learning_rate": 3.694832055039751e-06, "loss": 0.1709, "step": 2569 }, { "epoch": 4.36, "learning_rate": 3.6877317902584575e-06, "loss": 0.1766, "step": 2570 }, { "epoch": 4.36, "learning_rate": 3.6806368116036683e-06, "loss": 0.1663, "step": 2571 }, { "epoch": 4.36, "learning_rate": 3.6735471250169786e-06, "loss": 0.1693, "step": 2572 }, { "epoch": 4.36, "learning_rate": 3.6664627364355544e-06, "loss": 0.1685, "step": 2573 }, { "epoch": 4.36, "learning_rate": 3.659383651792124e-06, "loss": 0.1623, "step": 2574 }, { "epoch": 4.36, "learning_rate": 3.652309877014979e-06, "loss": 0.1654, "step": 2575 }, { "epoch": 4.37, "learning_rate": 3.6452414180279516e-06, "loss": 0.165, "step": 2576 }, { "epoch": 4.37, "learning_rate": 3.638178280750435e-06, "loss": 0.1684, "step": 2577 }, { "epoch": 4.37, "learning_rate": 3.631120471097359e-06, "loss": 0.1757, "step": 2578 }, { "epoch": 4.37, "learning_rate": 3.6240679949791967e-06, "loss": 0.1739, "step": 2579 }, { "epoch": 4.37, "learning_rate": 3.6170208583019506e-06, "loss": 0.1708, "step": 2580 }, { "epoch": 4.37, "learning_rate": 3.6099790669671564e-06, "loss": 0.1642, "step": 2581 }, { "epoch": 4.38, "learning_rate": 3.6029426268718647e-06, "loss": 0.1711, "step": 2582 }, { "epoch": 4.38, "learning_rate": 3.5959115439086523e-06, "loss": 0.1644, "step": 2583 }, { "epoch": 4.38, "learning_rate": 3.588885823965608e-06, "loss": 0.1682, "step": 2584 }, { "epoch": 4.38, "learning_rate": 3.5818654729263293e-06, "loss": 0.1711, "step": 2585 }, { "epoch": 4.38, "learning_rate": 3.574850496669917e-06, "loss": 0.1662, "step": 2586 }, { "epoch": 4.38, "learning_rate": 3.567840901070976e-06, "loss": 0.1753, "step": 2587 }, { "epoch": 4.39, "learning_rate": 3.5608366919995917e-06, "loss": 0.1661, "step": 2588 }, { "epoch": 4.39, "learning_rate": 3.5538378753213533e-06, "loss": 0.1662, "step": 2589 }, { "epoch": 4.39, "learning_rate": 3.546844456897326e-06, "loss": 0.1662, "step": 2590 }, { "epoch": 4.39, "learning_rate": 3.539856442584059e-06, "loss": 0.1703, "step": 2591 }, { "epoch": 4.39, "learning_rate": 3.5328738382335704e-06, "loss": 0.1747, "step": 2592 }, { "epoch": 4.39, "learning_rate": 3.5258966496933566e-06, "loss": 0.1686, "step": 2593 }, { "epoch": 4.4, "learning_rate": 3.518924882806366e-06, "loss": 0.1656, "step": 2594 }, { "epoch": 4.4, "learning_rate": 3.511958543411017e-06, "loss": 0.1614, "step": 2595 }, { "epoch": 4.4, "learning_rate": 3.50499763734118e-06, "loss": 0.16, "step": 2596 }, { "epoch": 4.4, "learning_rate": 3.4980421704261724e-06, "loss": 0.1641, "step": 2597 }, { "epoch": 4.4, "learning_rate": 3.4910921484907614e-06, "loss": 0.171, "step": 2598 }, { "epoch": 4.4, "learning_rate": 3.484147577355155e-06, "loss": 0.167, "step": 2599 }, { "epoch": 4.41, "learning_rate": 3.4772084628349855e-06, "loss": 0.1751, "step": 2600 }, { "epoch": 4.41, "learning_rate": 3.4702748107413274e-06, "loss": 0.1735, "step": 2601 }, { "epoch": 4.41, "learning_rate": 3.463346626880679e-06, "loss": 0.174, "step": 2602 }, { "epoch": 4.41, "learning_rate": 3.456423917054955e-06, "loss": 0.169, "step": 2603 }, { "epoch": 4.41, "learning_rate": 3.4495066870614903e-06, "loss": 0.181, "step": 2604 }, { "epoch": 4.41, "learning_rate": 3.4425949426930304e-06, "loss": 0.1655, "step": 2605 }, { "epoch": 4.42, "learning_rate": 3.4356886897377186e-06, "loss": 0.1677, "step": 2606 }, { "epoch": 4.42, "learning_rate": 3.428787933979114e-06, "loss": 0.1661, "step": 2607 }, { "epoch": 4.42, "learning_rate": 3.421892681196156e-06, "loss": 0.1725, "step": 2608 }, { "epoch": 4.42, "learning_rate": 3.415002937163189e-06, "loss": 0.1617, "step": 2609 }, { "epoch": 4.42, "learning_rate": 3.4081187076499388e-06, "loss": 0.172, "step": 2610 }, { "epoch": 4.43, "learning_rate": 3.4012399984215117e-06, "loss": 0.1696, "step": 2611 }, { "epoch": 4.43, "learning_rate": 3.394366815238398e-06, "loss": 0.1577, "step": 2612 }, { "epoch": 4.43, "learning_rate": 3.387499163856448e-06, "loss": 0.1663, "step": 2613 }, { "epoch": 4.43, "learning_rate": 3.3806370500268906e-06, "loss": 0.1677, "step": 2614 }, { "epoch": 4.43, "learning_rate": 3.3737804794963116e-06, "loss": 0.1721, "step": 2615 }, { "epoch": 4.43, "learning_rate": 3.366929458006657e-06, "loss": 0.1789, "step": 2616 }, { "epoch": 4.44, "learning_rate": 3.3600839912952256e-06, "loss": 0.1596, "step": 2617 }, { "epoch": 4.44, "learning_rate": 3.3532440850946636e-06, "loss": 0.1705, "step": 2618 }, { "epoch": 4.44, "learning_rate": 3.346409745132966e-06, "loss": 0.1682, "step": 2619 }, { "epoch": 4.44, "learning_rate": 3.339580977133452e-06, "loss": 0.1677, "step": 2620 }, { "epoch": 4.44, "learning_rate": 3.33275778681479e-06, "loss": 0.1751, "step": 2621 }, { "epoch": 4.44, "learning_rate": 3.3259401798909698e-06, "loss": 0.1824, "step": 2622 }, { "epoch": 4.45, "learning_rate": 3.319128162071308e-06, "loss": 0.1661, "step": 2623 }, { "epoch": 4.45, "learning_rate": 3.312321739060439e-06, "loss": 0.1808, "step": 2624 }, { "epoch": 4.45, "learning_rate": 3.305520916558318e-06, "loss": 0.178, "step": 2625 }, { "epoch": 4.45, "learning_rate": 3.2987257002601957e-06, "loss": 0.1704, "step": 2626 }, { "epoch": 4.45, "learning_rate": 3.2919360958566438e-06, "loss": 0.1757, "step": 2627 }, { "epoch": 4.45, "learning_rate": 3.285152109033525e-06, "loss": 0.1796, "step": 2628 }, { "epoch": 4.46, "learning_rate": 3.278373745472e-06, "loss": 0.1774, "step": 2629 }, { "epoch": 4.46, "learning_rate": 3.271601010848523e-06, "loss": 0.1594, "step": 2630 }, { "epoch": 4.46, "learning_rate": 3.264833910834835e-06, "loss": 0.1833, "step": 2631 }, { "epoch": 4.46, "learning_rate": 3.258072451097948e-06, "loss": 0.1782, "step": 2632 }, { "epoch": 4.46, "learning_rate": 3.2513166373001603e-06, "loss": 0.1708, "step": 2633 }, { "epoch": 4.46, "learning_rate": 3.2445664750990426e-06, "loss": 0.1658, "step": 2634 }, { "epoch": 4.47, "learning_rate": 3.237821970147428e-06, "loss": 0.1635, "step": 2635 }, { "epoch": 4.47, "learning_rate": 3.2310831280934143e-06, "loss": 0.1628, "step": 2636 }, { "epoch": 4.47, "learning_rate": 3.2243499545803615e-06, "loss": 0.1665, "step": 2637 }, { "epoch": 4.47, "learning_rate": 3.2176224552468683e-06, "loss": 0.1675, "step": 2638 }, { "epoch": 4.47, "learning_rate": 3.210900635726797e-06, "loss": 0.1715, "step": 2639 }, { "epoch": 4.47, "learning_rate": 3.204184501649246e-06, "loss": 0.161, "step": 2640 }, { "epoch": 4.48, "learning_rate": 3.1974740586385544e-06, "loss": 0.1643, "step": 2641 }, { "epoch": 4.48, "learning_rate": 3.190769312314296e-06, "loss": 0.1754, "step": 2642 }, { "epoch": 4.48, "learning_rate": 3.184070268291275e-06, "loss": 0.164, "step": 2643 }, { "epoch": 4.48, "learning_rate": 3.1773769321795113e-06, "loss": 0.1804, "step": 2644 }, { "epoch": 4.48, "learning_rate": 3.1706893095842563e-06, "loss": 0.1617, "step": 2645 }, { "epoch": 4.48, "learning_rate": 3.164007406105971e-06, "loss": 0.1704, "step": 2646 }, { "epoch": 4.49, "learning_rate": 3.15733122734033e-06, "loss": 0.1726, "step": 2647 }, { "epoch": 4.49, "learning_rate": 3.1506607788782097e-06, "loss": 0.1615, "step": 2648 }, { "epoch": 4.49, "learning_rate": 3.1439960663056944e-06, "loss": 0.1621, "step": 2649 }, { "epoch": 4.49, "learning_rate": 3.1373370952040536e-06, "loss": 0.1701, "step": 2650 }, { "epoch": 4.49, "learning_rate": 3.130683871149759e-06, "loss": 0.1685, "step": 2651 }, { "epoch": 4.49, "learning_rate": 3.1240363997144664e-06, "loss": 0.1685, "step": 2652 }, { "epoch": 4.5, "learning_rate": 3.117394686465013e-06, "loss": 0.1693, "step": 2653 }, { "epoch": 4.5, "learning_rate": 3.1107587369634163e-06, "loss": 0.1816, "step": 2654 }, { "epoch": 4.5, "learning_rate": 3.1041285567668667e-06, "loss": 0.18, "step": 2655 }, { "epoch": 4.5, "learning_rate": 3.0975041514277172e-06, "loss": 0.1654, "step": 2656 }, { "epoch": 4.5, "learning_rate": 3.0908855264934955e-06, "loss": 0.1689, "step": 2657 }, { "epoch": 4.5, "learning_rate": 3.084272687506875e-06, "loss": 0.1635, "step": 2658 }, { "epoch": 4.51, "learning_rate": 3.077665640005696e-06, "loss": 0.1641, "step": 2659 }, { "epoch": 4.51, "learning_rate": 3.071064389522944e-06, "loss": 0.164, "step": 2660 }, { "epoch": 4.51, "learning_rate": 3.0644689415867477e-06, "loss": 0.1672, "step": 2661 }, { "epoch": 4.51, "learning_rate": 3.0578793017203856e-06, "loss": 0.173, "step": 2662 }, { "epoch": 4.51, "learning_rate": 3.0512954754422563e-06, "loss": 0.1742, "step": 2663 }, { "epoch": 4.51, "learning_rate": 3.044717468265902e-06, "loss": 0.1661, "step": 2664 }, { "epoch": 4.52, "learning_rate": 3.0381452856999914e-06, "loss": 0.1692, "step": 2665 }, { "epoch": 4.52, "learning_rate": 3.0315789332483103e-06, "loss": 0.1667, "step": 2666 }, { "epoch": 4.52, "learning_rate": 3.0250184164097663e-06, "loss": 0.1691, "step": 2667 }, { "epoch": 4.52, "learning_rate": 3.018463740678378e-06, "loss": 0.1676, "step": 2668 }, { "epoch": 4.52, "learning_rate": 3.0119149115432754e-06, "loss": 0.166, "step": 2669 }, { "epoch": 4.52, "learning_rate": 3.0053719344886843e-06, "loss": 0.1702, "step": 2670 }, { "epoch": 4.53, "learning_rate": 2.9988348149939397e-06, "loss": 0.1613, "step": 2671 }, { "epoch": 4.53, "learning_rate": 2.9923035585334636e-06, "loss": 0.1599, "step": 2672 }, { "epoch": 4.53, "learning_rate": 2.985778170576774e-06, "loss": 0.1592, "step": 2673 }, { "epoch": 4.53, "learning_rate": 2.979258656588472e-06, "loss": 0.1649, "step": 2674 }, { "epoch": 4.53, "learning_rate": 2.9727450220282405e-06, "loss": 0.1763, "step": 2675 }, { "epoch": 4.54, "learning_rate": 2.9662372723508328e-06, "loss": 0.1629, "step": 2676 }, { "epoch": 4.54, "learning_rate": 2.9597354130060807e-06, "loss": 0.1673, "step": 2677 }, { "epoch": 4.54, "learning_rate": 2.9532394494388827e-06, "loss": 0.1715, "step": 2678 }, { "epoch": 4.54, "learning_rate": 2.9467493870891974e-06, "loss": 0.1802, "step": 2679 }, { "epoch": 4.54, "learning_rate": 2.9402652313920443e-06, "loss": 0.162, "step": 2680 }, { "epoch": 4.54, "learning_rate": 2.933786987777496e-06, "loss": 0.1614, "step": 2681 }, { "epoch": 4.55, "learning_rate": 2.9273146616706685e-06, "loss": 0.1575, "step": 2682 }, { "epoch": 4.55, "learning_rate": 2.9208482584917284e-06, "loss": 0.1594, "step": 2683 }, { "epoch": 4.55, "learning_rate": 2.9143877836558807e-06, "loss": 0.1659, "step": 2684 }, { "epoch": 4.55, "learning_rate": 2.9079332425733674e-06, "loss": 0.1745, "step": 2685 }, { "epoch": 4.55, "learning_rate": 2.901484640649457e-06, "loss": 0.1738, "step": 2686 }, { "epoch": 4.55, "learning_rate": 2.895041983284451e-06, "loss": 0.1721, "step": 2687 }, { "epoch": 4.56, "learning_rate": 2.8886052758736636e-06, "loss": 0.1602, "step": 2688 }, { "epoch": 4.56, "learning_rate": 2.8821745238074326e-06, "loss": 0.1744, "step": 2689 }, { "epoch": 4.56, "learning_rate": 2.875749732471108e-06, "loss": 0.1614, "step": 2690 }, { "epoch": 4.56, "learning_rate": 2.869330907245047e-06, "loss": 0.1704, "step": 2691 }, { "epoch": 4.56, "learning_rate": 2.862918053504612e-06, "loss": 0.1698, "step": 2692 }, { "epoch": 4.56, "learning_rate": 2.856511176620165e-06, "loss": 0.1684, "step": 2693 }, { "epoch": 4.57, "learning_rate": 2.8501102819570557e-06, "loss": 0.1664, "step": 2694 }, { "epoch": 4.57, "learning_rate": 2.843715374875634e-06, "loss": 0.1732, "step": 2695 }, { "epoch": 4.57, "learning_rate": 2.83732646073123e-06, "loss": 0.1645, "step": 2696 }, { "epoch": 4.57, "learning_rate": 2.830943544874156e-06, "loss": 0.1725, "step": 2697 }, { "epoch": 4.57, "learning_rate": 2.8245666326497023e-06, "loss": 0.1585, "step": 2698 }, { "epoch": 4.57, "learning_rate": 2.818195729398133e-06, "loss": 0.1596, "step": 2699 }, { "epoch": 4.58, "learning_rate": 2.8118308404546733e-06, "loss": 0.1763, "step": 2700 }, { "epoch": 4.58, "learning_rate": 2.805471971149517e-06, "loss": 0.1631, "step": 2701 }, { "epoch": 4.58, "learning_rate": 2.7991191268078176e-06, "loss": 0.1629, "step": 2702 }, { "epoch": 4.58, "learning_rate": 2.792772312749682e-06, "loss": 0.1644, "step": 2703 }, { "epoch": 4.58, "learning_rate": 2.786431534290165e-06, "loss": 0.1657, "step": 2704 }, { "epoch": 4.58, "learning_rate": 2.7800967967392722e-06, "loss": 0.1697, "step": 2705 }, { "epoch": 4.59, "learning_rate": 2.7737681054019415e-06, "loss": 0.1589, "step": 2706 }, { "epoch": 4.59, "learning_rate": 2.767445465578058e-06, "loss": 0.1588, "step": 2707 }, { "epoch": 4.59, "learning_rate": 2.7611288825624283e-06, "loss": 0.1595, "step": 2708 }, { "epoch": 4.59, "learning_rate": 2.7548183616447952e-06, "loss": 0.1701, "step": 2709 }, { "epoch": 4.59, "learning_rate": 2.7485139081098213e-06, "loss": 0.1751, "step": 2710 }, { "epoch": 4.59, "learning_rate": 2.7422155272370886e-06, "loss": 0.1685, "step": 2711 }, { "epoch": 4.6, "learning_rate": 2.735923224301099e-06, "loss": 0.1651, "step": 2712 }, { "epoch": 4.6, "learning_rate": 2.729637004571252e-06, "loss": 0.1664, "step": 2713 }, { "epoch": 4.6, "learning_rate": 2.723356873311861e-06, "loss": 0.158, "step": 2714 }, { "epoch": 4.6, "learning_rate": 2.7170828357821433e-06, "loss": 0.1598, "step": 2715 }, { "epoch": 4.6, "learning_rate": 2.710814897236208e-06, "loss": 0.1702, "step": 2716 }, { "epoch": 4.6, "learning_rate": 2.704553062923059e-06, "loss": 0.1699, "step": 2717 }, { "epoch": 4.61, "learning_rate": 2.698297338086586e-06, "loss": 0.1696, "step": 2718 }, { "epoch": 4.61, "learning_rate": 2.692047727965568e-06, "loss": 0.1667, "step": 2719 }, { "epoch": 4.61, "learning_rate": 2.6858042377936544e-06, "loss": 0.17, "step": 2720 }, { "epoch": 4.61, "learning_rate": 2.6795668727993753e-06, "loss": 0.1672, "step": 2721 }, { "epoch": 4.61, "learning_rate": 2.6733356382061324e-06, "loss": 0.1662, "step": 2722 }, { "epoch": 4.61, "learning_rate": 2.6671105392321904e-06, "loss": 0.1582, "step": 2723 }, { "epoch": 4.62, "learning_rate": 2.6608915810906775e-06, "loss": 0.1726, "step": 2724 }, { "epoch": 4.62, "learning_rate": 2.6546787689895825e-06, "loss": 0.1626, "step": 2725 }, { "epoch": 4.62, "learning_rate": 2.6484721081317367e-06, "loss": 0.1597, "step": 2726 }, { "epoch": 4.62, "learning_rate": 2.64227160371483e-06, "loss": 0.1692, "step": 2727 }, { "epoch": 4.62, "learning_rate": 2.6360772609313946e-06, "loss": 0.1698, "step": 2728 }, { "epoch": 4.62, "learning_rate": 2.6298890849688018e-06, "loss": 0.1731, "step": 2729 }, { "epoch": 4.63, "learning_rate": 2.6237070810092578e-06, "loss": 0.1657, "step": 2730 }, { "epoch": 4.63, "learning_rate": 2.617531254229806e-06, "loss": 0.1597, "step": 2731 }, { "epoch": 4.63, "learning_rate": 2.6113616098023033e-06, "loss": 0.1695, "step": 2732 }, { "epoch": 4.63, "learning_rate": 2.6051981528934424e-06, "loss": 0.1605, "step": 2733 }, { "epoch": 4.63, "learning_rate": 2.599040888664729e-06, "loss": 0.1601, "step": 2734 }, { "epoch": 4.64, "learning_rate": 2.592889822272484e-06, "loss": 0.1698, "step": 2735 }, { "epoch": 4.64, "learning_rate": 2.5867449588678386e-06, "loss": 0.166, "step": 2736 }, { "epoch": 4.64, "learning_rate": 2.5806063035967275e-06, "loss": 0.1787, "step": 2737 }, { "epoch": 4.64, "learning_rate": 2.5744738615998853e-06, "loss": 0.1697, "step": 2738 }, { "epoch": 4.64, "learning_rate": 2.568347638012848e-06, "loss": 0.1703, "step": 2739 }, { "epoch": 4.64, "learning_rate": 2.5622276379659394e-06, "loss": 0.1744, "step": 2740 }, { "epoch": 4.65, "learning_rate": 2.5561138665842745e-06, "loss": 0.1651, "step": 2741 }, { "epoch": 4.65, "learning_rate": 2.550006328987752e-06, "loss": 0.1723, "step": 2742 }, { "epoch": 4.65, "learning_rate": 2.543905030291051e-06, "loss": 0.1724, "step": 2743 }, { "epoch": 4.65, "learning_rate": 2.53780997560362e-06, "loss": 0.1649, "step": 2744 }, { "epoch": 4.65, "learning_rate": 2.531721170029684e-06, "loss": 0.1636, "step": 2745 }, { "epoch": 4.65, "learning_rate": 2.5256386186682346e-06, "loss": 0.1655, "step": 2746 }, { "epoch": 4.66, "learning_rate": 2.5195623266130264e-06, "loss": 0.1585, "step": 2747 }, { "epoch": 4.66, "learning_rate": 2.5134922989525677e-06, "loss": 0.1606, "step": 2748 }, { "epoch": 4.66, "learning_rate": 2.5074285407701292e-06, "loss": 0.168, "step": 2749 }, { "epoch": 4.66, "learning_rate": 2.50137105714372e-06, "loss": 0.159, "step": 2750 }, { "epoch": 4.66, "learning_rate": 2.495319853146102e-06, "loss": 0.1676, "step": 2751 }, { "epoch": 4.66, "learning_rate": 2.4892749338447774e-06, "loss": 0.1629, "step": 2752 }, { "epoch": 4.67, "learning_rate": 2.483236304301985e-06, "loss": 0.1701, "step": 2753 }, { "epoch": 4.67, "learning_rate": 2.4772039695746963e-06, "loss": 0.1752, "step": 2754 }, { "epoch": 4.67, "learning_rate": 2.471177934714615e-06, "loss": 0.1664, "step": 2755 }, { "epoch": 4.67, "learning_rate": 2.465158204768158e-06, "loss": 0.1691, "step": 2756 }, { "epoch": 4.67, "learning_rate": 2.4591447847764727e-06, "loss": 0.1735, "step": 2757 }, { "epoch": 4.67, "learning_rate": 2.4531376797754212e-06, "loss": 0.1596, "step": 2758 }, { "epoch": 4.68, "learning_rate": 2.447136894795571e-06, "loss": 0.1742, "step": 2759 }, { "epoch": 4.68, "learning_rate": 2.4411424348622035e-06, "loss": 0.1704, "step": 2760 }, { "epoch": 4.68, "learning_rate": 2.4351543049952995e-06, "loss": 0.1751, "step": 2761 }, { "epoch": 4.68, "learning_rate": 2.429172510209542e-06, "loss": 0.1621, "step": 2762 }, { "epoch": 4.68, "learning_rate": 2.423197055514308e-06, "loss": 0.1638, "step": 2763 }, { "epoch": 4.68, "learning_rate": 2.4172279459136592e-06, "loss": 0.1648, "step": 2764 }, { "epoch": 4.69, "learning_rate": 2.411265186406352e-06, "loss": 0.1743, "step": 2765 }, { "epoch": 4.69, "learning_rate": 2.405308781985819e-06, "loss": 0.1639, "step": 2766 }, { "epoch": 4.69, "learning_rate": 2.399358737640175e-06, "loss": 0.1674, "step": 2767 }, { "epoch": 4.69, "learning_rate": 2.3934150583522078e-06, "loss": 0.1745, "step": 2768 }, { "epoch": 4.69, "learning_rate": 2.387477749099374e-06, "loss": 0.1622, "step": 2769 }, { "epoch": 4.69, "learning_rate": 2.381546814853791e-06, "loss": 0.1678, "step": 2770 }, { "epoch": 4.7, "learning_rate": 2.3756222605822467e-06, "loss": 0.1609, "step": 2771 }, { "epoch": 4.7, "learning_rate": 2.36970409124618e-06, "loss": 0.168, "step": 2772 }, { "epoch": 4.7, "learning_rate": 2.363792311801685e-06, "loss": 0.16, "step": 2773 }, { "epoch": 4.7, "learning_rate": 2.357886927199503e-06, "loss": 0.1648, "step": 2774 }, { "epoch": 4.7, "learning_rate": 2.351987942385026e-06, "loss": 0.1631, "step": 2775 }, { "epoch": 4.7, "learning_rate": 2.346095362298274e-06, "loss": 0.1696, "step": 2776 }, { "epoch": 4.71, "learning_rate": 2.3402091918739157e-06, "loss": 0.1693, "step": 2777 }, { "epoch": 4.71, "learning_rate": 2.334329436041246e-06, "loss": 0.1638, "step": 2778 }, { "epoch": 4.71, "learning_rate": 2.328456099724191e-06, "loss": 0.1588, "step": 2779 }, { "epoch": 4.71, "learning_rate": 2.322589187841299e-06, "loss": 0.1617, "step": 2780 }, { "epoch": 4.71, "learning_rate": 2.316728705305741e-06, "loss": 0.1699, "step": 2781 }, { "epoch": 4.71, "learning_rate": 2.310874657025296e-06, "loss": 0.1676, "step": 2782 }, { "epoch": 4.72, "learning_rate": 2.305027047902364e-06, "loss": 0.1685, "step": 2783 }, { "epoch": 4.72, "learning_rate": 2.2991858828339473e-06, "loss": 0.1657, "step": 2784 }, { "epoch": 4.72, "learning_rate": 2.2933511667116547e-06, "loss": 0.1721, "step": 2785 }, { "epoch": 4.72, "learning_rate": 2.2875229044216917e-06, "loss": 0.155, "step": 2786 }, { "epoch": 4.72, "learning_rate": 2.281701100844864e-06, "loss": 0.1633, "step": 2787 }, { "epoch": 4.72, "learning_rate": 2.275885760856559e-06, "loss": 0.163, "step": 2788 }, { "epoch": 4.73, "learning_rate": 2.270076889326759e-06, "loss": 0.1663, "step": 2789 }, { "epoch": 4.73, "learning_rate": 2.26427449112003e-06, "loss": 0.1605, "step": 2790 }, { "epoch": 4.73, "learning_rate": 2.2584785710955114e-06, "loss": 0.1784, "step": 2791 }, { "epoch": 4.73, "learning_rate": 2.2526891341069233e-06, "loss": 0.1639, "step": 2792 }, { "epoch": 4.73, "learning_rate": 2.2469061850025563e-06, "loss": 0.1643, "step": 2793 }, { "epoch": 4.73, "learning_rate": 2.241129728625259e-06, "loss": 0.1656, "step": 2794 }, { "epoch": 4.74, "learning_rate": 2.235359769812453e-06, "loss": 0.1733, "step": 2795 }, { "epoch": 4.74, "learning_rate": 2.229596313396113e-06, "loss": 0.1621, "step": 2796 }, { "epoch": 4.74, "learning_rate": 2.223839364202773e-06, "loss": 0.1675, "step": 2797 }, { "epoch": 4.74, "learning_rate": 2.218088927053513e-06, "loss": 0.1631, "step": 2798 }, { "epoch": 4.74, "learning_rate": 2.2123450067639652e-06, "loss": 0.1618, "step": 2799 }, { "epoch": 4.75, "learning_rate": 2.2066076081442944e-06, "loss": 0.1665, "step": 2800 }, { "epoch": 4.75, "learning_rate": 2.200876735999212e-06, "loss": 0.1583, "step": 2801 }, { "epoch": 4.75, "learning_rate": 2.195152395127965e-06, "loss": 0.168, "step": 2802 }, { "epoch": 4.75, "learning_rate": 2.1894345903243243e-06, "loss": 0.1724, "step": 2803 }, { "epoch": 4.75, "learning_rate": 2.183723326376592e-06, "loss": 0.169, "step": 2804 }, { "epoch": 4.75, "learning_rate": 2.1780186080675957e-06, "loss": 0.1581, "step": 2805 }, { "epoch": 4.76, "learning_rate": 2.17232044017467e-06, "loss": 0.1635, "step": 2806 }, { "epoch": 4.76, "learning_rate": 2.1666288274696745e-06, "loss": 0.1702, "step": 2807 }, { "epoch": 4.76, "learning_rate": 2.1609437747189786e-06, "loss": 0.1639, "step": 2808 }, { "epoch": 4.76, "learning_rate": 2.155265286683449e-06, "loss": 0.1673, "step": 2809 }, { "epoch": 4.76, "learning_rate": 2.1495933681184654e-06, "loss": 0.1599, "step": 2810 }, { "epoch": 4.76, "learning_rate": 2.1439280237739014e-06, "loss": 0.1741, "step": 2811 }, { "epoch": 4.77, "learning_rate": 2.1382692583941257e-06, "loss": 0.1583, "step": 2812 }, { "epoch": 4.77, "learning_rate": 2.1326170767179987e-06, "loss": 0.1637, "step": 2813 }, { "epoch": 4.77, "learning_rate": 2.1269714834788626e-06, "loss": 0.1609, "step": 2814 }, { "epoch": 4.77, "learning_rate": 2.1213324834045477e-06, "loss": 0.1644, "step": 2815 }, { "epoch": 4.77, "learning_rate": 2.11570008121736e-06, "loss": 0.1652, "step": 2816 }, { "epoch": 4.77, "learning_rate": 2.1100742816340826e-06, "loss": 0.1661, "step": 2817 }, { "epoch": 4.78, "learning_rate": 2.1044550893659665e-06, "loss": 0.1616, "step": 2818 }, { "epoch": 4.78, "learning_rate": 2.0988425091187346e-06, "loss": 0.1611, "step": 2819 }, { "epoch": 4.78, "learning_rate": 2.093236545592563e-06, "loss": 0.1664, "step": 2820 }, { "epoch": 4.78, "learning_rate": 2.087637203482095e-06, "loss": 0.1704, "step": 2821 }, { "epoch": 4.78, "learning_rate": 2.0820444874764266e-06, "loss": 0.169, "step": 2822 }, { "epoch": 4.78, "learning_rate": 2.076458402259105e-06, "loss": 0.1701, "step": 2823 }, { "epoch": 4.79, "learning_rate": 2.0708789525081243e-06, "loss": 0.1636, "step": 2824 }, { "epoch": 4.79, "learning_rate": 2.065306142895924e-06, "loss": 0.1622, "step": 2825 }, { "epoch": 4.79, "learning_rate": 2.0597399780893754e-06, "loss": 0.1544, "step": 2826 }, { "epoch": 4.79, "learning_rate": 2.054180462749793e-06, "loss": 0.1592, "step": 2827 }, { "epoch": 4.79, "learning_rate": 2.0486276015329208e-06, "loss": 0.1631, "step": 2828 }, { "epoch": 4.79, "learning_rate": 2.04308139908893e-06, "loss": 0.1676, "step": 2829 }, { "epoch": 4.8, "learning_rate": 2.0375418600624142e-06, "loss": 0.1656, "step": 2830 }, { "epoch": 4.8, "learning_rate": 2.0320089890923923e-06, "loss": 0.1579, "step": 2831 }, { "epoch": 4.8, "learning_rate": 2.0264827908122877e-06, "loss": 0.1631, "step": 2832 }, { "epoch": 4.8, "learning_rate": 2.0209632698499457e-06, "loss": 0.1647, "step": 2833 }, { "epoch": 4.8, "learning_rate": 2.015450430827619e-06, "loss": 0.1662, "step": 2834 }, { "epoch": 4.8, "learning_rate": 2.0099442783619604e-06, "loss": 0.1646, "step": 2835 }, { "epoch": 4.81, "learning_rate": 2.0044448170640265e-06, "loss": 0.1711, "step": 2836 }, { "epoch": 4.81, "learning_rate": 1.9989520515392712e-06, "loss": 0.1727, "step": 2837 }, { "epoch": 4.81, "learning_rate": 1.993465986387534e-06, "loss": 0.1604, "step": 2838 }, { "epoch": 4.81, "learning_rate": 1.9879866262030523e-06, "loss": 0.1622, "step": 2839 }, { "epoch": 4.81, "learning_rate": 1.9825139755744426e-06, "loss": 0.1585, "step": 2840 }, { "epoch": 4.81, "learning_rate": 1.9770480390847056e-06, "loss": 0.1636, "step": 2841 }, { "epoch": 4.82, "learning_rate": 1.97158882131122e-06, "loss": 0.1591, "step": 2842 }, { "epoch": 4.82, "learning_rate": 1.966136326825736e-06, "loss": 0.17, "step": 2843 }, { "epoch": 4.82, "learning_rate": 1.9606905601943705e-06, "loss": 0.1578, "step": 2844 }, { "epoch": 4.82, "learning_rate": 1.9552515259776115e-06, "loss": 0.161, "step": 2845 }, { "epoch": 4.82, "learning_rate": 1.9498192287303087e-06, "loss": 0.1657, "step": 2846 }, { "epoch": 4.82, "learning_rate": 1.9443936730016655e-06, "loss": 0.1711, "step": 2847 }, { "epoch": 4.83, "learning_rate": 1.9389748633352454e-06, "loss": 0.1629, "step": 2848 }, { "epoch": 4.83, "learning_rate": 1.9335628042689602e-06, "loss": 0.1691, "step": 2849 }, { "epoch": 4.83, "learning_rate": 1.928157500335065e-06, "loss": 0.1649, "step": 2850 }, { "epoch": 4.83, "learning_rate": 1.922758956060161e-06, "loss": 0.1651, "step": 2851 }, { "epoch": 4.83, "learning_rate": 1.9173671759651903e-06, "loss": 0.1672, "step": 2852 }, { "epoch": 4.83, "learning_rate": 1.911982164565429e-06, "loss": 0.1586, "step": 2853 }, { "epoch": 4.84, "learning_rate": 1.9066039263704828e-06, "loss": 0.1649, "step": 2854 }, { "epoch": 4.84, "learning_rate": 1.9012324658842906e-06, "loss": 0.1573, "step": 2855 }, { "epoch": 4.84, "learning_rate": 1.8958677876051069e-06, "loss": 0.1576, "step": 2856 }, { "epoch": 4.84, "learning_rate": 1.8905098960255131e-06, "loss": 0.16, "step": 2857 }, { "epoch": 4.84, "learning_rate": 1.8851587956324057e-06, "loss": 0.1575, "step": 2858 }, { "epoch": 4.85, "learning_rate": 1.879814490906997e-06, "loss": 0.1671, "step": 2859 }, { "epoch": 4.85, "learning_rate": 1.8744769863248002e-06, "loss": 0.1675, "step": 2860 }, { "epoch": 4.85, "learning_rate": 1.8691462863556397e-06, "loss": 0.1674, "step": 2861 }, { "epoch": 4.85, "learning_rate": 1.8638223954636426e-06, "loss": 0.1679, "step": 2862 }, { "epoch": 4.85, "learning_rate": 1.8585053181072321e-06, "loss": 0.166, "step": 2863 }, { "epoch": 4.85, "learning_rate": 1.8531950587391224e-06, "loss": 0.1629, "step": 2864 }, { "epoch": 4.86, "learning_rate": 1.8478916218063202e-06, "loss": 0.1631, "step": 2865 }, { "epoch": 4.86, "learning_rate": 1.8425950117501224e-06, "loss": 0.1647, "step": 2866 }, { "epoch": 4.86, "learning_rate": 1.8373052330061048e-06, "loss": 0.1595, "step": 2867 }, { "epoch": 4.86, "learning_rate": 1.832022290004124e-06, "loss": 0.1701, "step": 2868 }, { "epoch": 4.86, "learning_rate": 1.8267461871683124e-06, "loss": 0.1588, "step": 2869 }, { "epoch": 4.86, "learning_rate": 1.8214769289170697e-06, "loss": 0.1639, "step": 2870 }, { "epoch": 4.87, "learning_rate": 1.8162145196630698e-06, "loss": 0.1679, "step": 2871 }, { "epoch": 4.87, "learning_rate": 1.8109589638132486e-06, "loss": 0.1647, "step": 2872 }, { "epoch": 4.87, "learning_rate": 1.8057102657688018e-06, "loss": 0.1623, "step": 2873 }, { "epoch": 4.87, "learning_rate": 1.800468429925184e-06, "loss": 0.1609, "step": 2874 }, { "epoch": 4.87, "learning_rate": 1.7952334606721046e-06, "loss": 0.1666, "step": 2875 }, { "epoch": 4.87, "learning_rate": 1.7900053623935143e-06, "loss": 0.169, "step": 2876 }, { "epoch": 4.88, "learning_rate": 1.7847841394676192e-06, "loss": 0.1612, "step": 2877 }, { "epoch": 4.88, "learning_rate": 1.7795697962668624e-06, "loss": 0.1591, "step": 2878 }, { "epoch": 4.88, "learning_rate": 1.7743623371579288e-06, "loss": 0.1612, "step": 2879 }, { "epoch": 4.88, "learning_rate": 1.769161766501738e-06, "loss": 0.1674, "step": 2880 }, { "epoch": 4.88, "learning_rate": 1.7639680886534404e-06, "loss": 0.167, "step": 2881 }, { "epoch": 4.88, "learning_rate": 1.7587813079624095e-06, "loss": 0.1646, "step": 2882 }, { "epoch": 4.89, "learning_rate": 1.7536014287722492e-06, "loss": 0.1563, "step": 2883 }, { "epoch": 4.89, "learning_rate": 1.748428455420782e-06, "loss": 0.1658, "step": 2884 }, { "epoch": 4.89, "learning_rate": 1.743262392240047e-06, "loss": 0.1647, "step": 2885 }, { "epoch": 4.89, "learning_rate": 1.738103243556296e-06, "loss": 0.1667, "step": 2886 }, { "epoch": 4.89, "learning_rate": 1.7329510136899929e-06, "loss": 0.1571, "step": 2887 }, { "epoch": 4.89, "learning_rate": 1.7278057069558008e-06, "loss": 0.1657, "step": 2888 }, { "epoch": 4.9, "learning_rate": 1.7226673276625915e-06, "loss": 0.1601, "step": 2889 }, { "epoch": 4.9, "learning_rate": 1.7175358801134334e-06, "loss": 0.1622, "step": 2890 }, { "epoch": 4.9, "learning_rate": 1.712411368605591e-06, "loss": 0.174, "step": 2891 }, { "epoch": 4.9, "learning_rate": 1.7072937974305193e-06, "loss": 0.1569, "step": 2892 }, { "epoch": 4.9, "learning_rate": 1.7021831708738635e-06, "loss": 0.1625, "step": 2893 }, { "epoch": 4.9, "learning_rate": 1.6970794932154454e-06, "loss": 0.1637, "step": 2894 }, { "epoch": 4.91, "learning_rate": 1.6919827687292778e-06, "loss": 0.1626, "step": 2895 }, { "epoch": 4.91, "learning_rate": 1.686893001683544e-06, "loss": 0.1683, "step": 2896 }, { "epoch": 4.91, "learning_rate": 1.681810196340603e-06, "loss": 0.1642, "step": 2897 }, { "epoch": 4.91, "learning_rate": 1.676734356956985e-06, "loss": 0.167, "step": 2898 }, { "epoch": 4.91, "learning_rate": 1.671665487783386e-06, "loss": 0.1629, "step": 2899 }, { "epoch": 4.91, "learning_rate": 1.6666035930646607e-06, "loss": 0.1732, "step": 2900 }, { "epoch": 4.92, "learning_rate": 1.6615486770398291e-06, "loss": 0.1655, "step": 2901 }, { "epoch": 4.92, "learning_rate": 1.6565007439420643e-06, "loss": 0.1717, "step": 2902 }, { "epoch": 4.92, "learning_rate": 1.6514597979986912e-06, "loss": 0.1567, "step": 2903 }, { "epoch": 4.92, "learning_rate": 1.6464258434311854e-06, "loss": 0.1624, "step": 2904 }, { "epoch": 4.92, "learning_rate": 1.6413988844551665e-06, "loss": 0.1624, "step": 2905 }, { "epoch": 4.92, "learning_rate": 1.6363789252803942e-06, "loss": 0.1662, "step": 2906 }, { "epoch": 4.93, "learning_rate": 1.6313659701107665e-06, "loss": 0.1641, "step": 2907 }, { "epoch": 4.93, "learning_rate": 1.62636002314432e-06, "loss": 0.1643, "step": 2908 }, { "epoch": 4.93, "learning_rate": 1.62136108857322e-06, "loss": 0.1655, "step": 2909 }, { "epoch": 4.93, "learning_rate": 1.616369170583756e-06, "loss": 0.1646, "step": 2910 }, { "epoch": 4.93, "learning_rate": 1.6113842733563446e-06, "loss": 0.1684, "step": 2911 }, { "epoch": 4.93, "learning_rate": 1.6064064010655246e-06, "loss": 0.1639, "step": 2912 }, { "epoch": 4.94, "learning_rate": 1.6014355578799524e-06, "loss": 0.1631, "step": 2913 }, { "epoch": 4.94, "learning_rate": 1.5964717479623893e-06, "loss": 0.1672, "step": 2914 }, { "epoch": 4.94, "learning_rate": 1.5915149754697178e-06, "loss": 0.1676, "step": 2915 }, { "epoch": 4.94, "learning_rate": 1.5865652445529201e-06, "loss": 0.1704, "step": 2916 }, { "epoch": 4.94, "learning_rate": 1.5816225593570843e-06, "loss": 0.1631, "step": 2917 }, { "epoch": 4.94, "learning_rate": 1.5766869240213978e-06, "loss": 0.1649, "step": 2918 }, { "epoch": 4.95, "learning_rate": 1.5717583426791472e-06, "loss": 0.1591, "step": 2919 }, { "epoch": 4.95, "learning_rate": 1.5668368194577021e-06, "loss": 0.1729, "step": 2920 }, { "epoch": 4.95, "learning_rate": 1.5619223584785304e-06, "loss": 0.1613, "step": 2921 }, { "epoch": 4.95, "learning_rate": 1.557014963857184e-06, "loss": 0.1616, "step": 2922 }, { "epoch": 4.95, "learning_rate": 1.552114639703296e-06, "loss": 0.1749, "step": 2923 }, { "epoch": 4.96, "learning_rate": 1.547221390120578e-06, "loss": 0.1603, "step": 2924 }, { "epoch": 4.96, "learning_rate": 1.542335219206822e-06, "loss": 0.1674, "step": 2925 }, { "epoch": 4.96, "learning_rate": 1.5374561310538816e-06, "loss": 0.1649, "step": 2926 }, { "epoch": 4.96, "learning_rate": 1.5325841297476896e-06, "loss": 0.1636, "step": 2927 }, { "epoch": 4.96, "learning_rate": 1.5277192193682378e-06, "loss": 0.1657, "step": 2928 }, { "epoch": 4.96, "learning_rate": 1.5228614039895828e-06, "loss": 0.1636, "step": 2929 }, { "epoch": 4.97, "learning_rate": 1.5180106876798384e-06, "loss": 0.1641, "step": 2930 }, { "epoch": 4.97, "learning_rate": 1.5131670745011762e-06, "loss": 0.1647, "step": 2931 }, { "epoch": 4.97, "learning_rate": 1.5083305685098126e-06, "loss": 0.159, "step": 2932 }, { "epoch": 4.97, "learning_rate": 1.5035011737560167e-06, "loss": 0.1528, "step": 2933 }, { "epoch": 4.97, "learning_rate": 1.4986788942841036e-06, "loss": 0.1623, "step": 2934 }, { "epoch": 4.97, "learning_rate": 1.493863734132428e-06, "loss": 0.1652, "step": 2935 }, { "epoch": 4.98, "learning_rate": 1.4890556973333826e-06, "loss": 0.1629, "step": 2936 }, { "epoch": 4.98, "learning_rate": 1.4842547879133973e-06, "loss": 0.162, "step": 2937 }, { "epoch": 4.98, "learning_rate": 1.4794610098929264e-06, "loss": 0.1641, "step": 2938 }, { "epoch": 4.98, "learning_rate": 1.4746743672864595e-06, "loss": 0.1621, "step": 2939 }, { "epoch": 4.98, "learning_rate": 1.4698948641025068e-06, "loss": 0.1622, "step": 2940 }, { "epoch": 4.98, "learning_rate": 1.4651225043436013e-06, "loss": 0.1578, "step": 2941 }, { "epoch": 4.99, "learning_rate": 1.4603572920062925e-06, "loss": 0.163, "step": 2942 }, { "epoch": 4.99, "learning_rate": 1.4555992310811495e-06, "loss": 0.1611, "step": 2943 }, { "epoch": 4.99, "learning_rate": 1.4508483255527417e-06, "loss": 0.171, "step": 2944 }, { "epoch": 4.99, "learning_rate": 1.4461045793996553e-06, "loss": 0.1671, "step": 2945 }, { "epoch": 4.99, "learning_rate": 1.4413679965944804e-06, "loss": 0.162, "step": 2946 }, { "epoch": 4.99, "learning_rate": 1.4366385811038041e-06, "loss": 0.1619, "step": 2947 }, { "epoch": 5.0, "learning_rate": 1.431916336888215e-06, "loss": 0.169, "step": 2948 }, { "epoch": 5.0, "learning_rate": 1.4272012679022962e-06, "loss": 0.161, "step": 2949 }, { "epoch": 5.0, "learning_rate": 1.422493378094616e-06, "loss": 0.158, "step": 2950 }, { "epoch": 5.0, "learning_rate": 1.4177926714077384e-06, "loss": 0.1557, "step": 2951 }, { "epoch": 5.0, "learning_rate": 1.4130991517782078e-06, "loss": 0.1521, "step": 2952 }, { "epoch": 5.0, "learning_rate": 1.4084128231365512e-06, "loss": 0.1582, "step": 2953 }, { "epoch": 5.01, "learning_rate": 1.403733689407274e-06, "loss": 0.1569, "step": 2954 }, { "epoch": 5.01, "learning_rate": 1.399061754508857e-06, "loss": 0.1545, "step": 2955 }, { "epoch": 5.01, "learning_rate": 1.394397022353746e-06, "loss": 0.1518, "step": 2956 }, { "epoch": 5.01, "learning_rate": 1.3897394968483636e-06, "loss": 0.1534, "step": 2957 }, { "epoch": 5.01, "learning_rate": 1.3850891818930922e-06, "loss": 0.1497, "step": 2958 }, { "epoch": 5.01, "learning_rate": 1.3804460813822784e-06, "loss": 0.1551, "step": 2959 }, { "epoch": 5.02, "learning_rate": 1.3758101992042284e-06, "loss": 0.1505, "step": 2960 }, { "epoch": 5.02, "learning_rate": 1.3711815392411953e-06, "loss": 0.1558, "step": 2961 }, { "epoch": 5.02, "learning_rate": 1.3665601053693944e-06, "loss": 0.158, "step": 2962 }, { "epoch": 5.02, "learning_rate": 1.3619459014589843e-06, "loss": 0.1532, "step": 2963 }, { "epoch": 5.02, "learning_rate": 1.3573389313740682e-06, "loss": 0.1559, "step": 2964 }, { "epoch": 5.02, "learning_rate": 1.3527391989726923e-06, "loss": 0.1525, "step": 2965 }, { "epoch": 5.03, "learning_rate": 1.3481467081068455e-06, "loss": 0.1532, "step": 2966 }, { "epoch": 5.03, "learning_rate": 1.343561462622447e-06, "loss": 0.1526, "step": 2967 }, { "epoch": 5.03, "learning_rate": 1.3389834663593516e-06, "loss": 0.1575, "step": 2968 }, { "epoch": 5.03, "learning_rate": 1.3344127231513449e-06, "loss": 0.1539, "step": 2969 }, { "epoch": 5.03, "learning_rate": 1.3298492368261295e-06, "loss": 0.1532, "step": 2970 }, { "epoch": 5.03, "learning_rate": 1.325293011205343e-06, "loss": 0.1555, "step": 2971 }, { "epoch": 5.04, "learning_rate": 1.3207440501045343e-06, "loss": 0.1539, "step": 2972 }, { "epoch": 5.04, "learning_rate": 1.3162023573331728e-06, "loss": 0.1554, "step": 2973 }, { "epoch": 5.04, "learning_rate": 1.3116679366946395e-06, "loss": 0.1527, "step": 2974 }, { "epoch": 5.04, "learning_rate": 1.3071407919862278e-06, "loss": 0.1544, "step": 2975 }, { "epoch": 5.04, "learning_rate": 1.3026209269991308e-06, "loss": 0.1578, "step": 2976 }, { "epoch": 5.04, "learning_rate": 1.298108345518453e-06, "loss": 0.1557, "step": 2977 }, { "epoch": 5.05, "learning_rate": 1.2936030513231979e-06, "loss": 0.153, "step": 2978 }, { "epoch": 5.05, "learning_rate": 1.2891050481862634e-06, "loss": 0.1543, "step": 2979 }, { "epoch": 5.05, "learning_rate": 1.2846143398744448e-06, "loss": 0.1551, "step": 2980 }, { "epoch": 5.05, "learning_rate": 1.2801309301484277e-06, "loss": 0.1504, "step": 2981 }, { "epoch": 5.05, "learning_rate": 1.275654822762783e-06, "loss": 0.1537, "step": 2982 }, { "epoch": 5.06, "learning_rate": 1.2711860214659677e-06, "loss": 0.1545, "step": 2983 }, { "epoch": 5.06, "learning_rate": 1.266724530000324e-06, "loss": 0.1525, "step": 2984 }, { "epoch": 5.06, "learning_rate": 1.2622703521020675e-06, "loss": 0.1581, "step": 2985 }, { "epoch": 5.06, "learning_rate": 1.257823491501291e-06, "loss": 0.1575, "step": 2986 }, { "epoch": 5.06, "learning_rate": 1.2533839519219637e-06, "loss": 0.1569, "step": 2987 }, { "epoch": 5.06, "learning_rate": 1.2489517370819148e-06, "loss": 0.1556, "step": 2988 }, { "epoch": 5.07, "learning_rate": 1.2445268506928466e-06, "loss": 0.1542, "step": 2989 }, { "epoch": 5.07, "learning_rate": 1.2401092964603222e-06, "loss": 0.152, "step": 2990 }, { "epoch": 5.07, "learning_rate": 1.2356990780837651e-06, "loss": 0.1507, "step": 2991 }, { "epoch": 5.07, "learning_rate": 1.2312961992564554e-06, "loss": 0.1553, "step": 2992 }, { "epoch": 5.07, "learning_rate": 1.2269006636655278e-06, "loss": 0.1604, "step": 2993 }, { "epoch": 5.07, "learning_rate": 1.2225124749919625e-06, "loss": 0.1579, "step": 2994 }, { "epoch": 5.08, "learning_rate": 1.2181316369105922e-06, "loss": 0.1597, "step": 2995 }, { "epoch": 5.08, "learning_rate": 1.2137581530900932e-06, "loss": 0.1562, "step": 2996 }, { "epoch": 5.08, "learning_rate": 1.2093920271929816e-06, "loss": 0.1561, "step": 2997 }, { "epoch": 5.08, "learning_rate": 1.2050332628756123e-06, "loss": 0.1603, "step": 2998 }, { "epoch": 5.08, "learning_rate": 1.200681863788179e-06, "loss": 0.1565, "step": 2999 }, { "epoch": 5.08, "learning_rate": 1.196337833574699e-06, "loss": 0.1538, "step": 3000 }, { "epoch": 5.09, "learning_rate": 1.1920011758730255e-06, "loss": 0.1546, "step": 3001 }, { "epoch": 5.09, "learning_rate": 1.1876718943148358e-06, "loss": 0.1551, "step": 3002 }, { "epoch": 5.09, "learning_rate": 1.1833499925256308e-06, "loss": 0.1529, "step": 3003 }, { "epoch": 5.09, "learning_rate": 1.179035474124729e-06, "loss": 0.1525, "step": 3004 }, { "epoch": 5.09, "learning_rate": 1.174728342725272e-06, "loss": 0.1571, "step": 3005 }, { "epoch": 5.09, "learning_rate": 1.1704286019342049e-06, "loss": 0.1525, "step": 3006 }, { "epoch": 5.1, "learning_rate": 1.1661362553522914e-06, "loss": 0.1492, "step": 3007 }, { "epoch": 5.1, "learning_rate": 1.1618513065741022e-06, "loss": 0.1471, "step": 3008 }, { "epoch": 5.1, "learning_rate": 1.15757375918801e-06, "loss": 0.1512, "step": 3009 }, { "epoch": 5.1, "learning_rate": 1.1533036167761945e-06, "loss": 0.1534, "step": 3010 }, { "epoch": 5.1, "learning_rate": 1.149040882914626e-06, "loss": 0.156, "step": 3011 }, { "epoch": 5.1, "learning_rate": 1.1447855611730773e-06, "loss": 0.1587, "step": 3012 }, { "epoch": 5.11, "learning_rate": 1.1405376551151137e-06, "loss": 0.1516, "step": 3013 }, { "epoch": 5.11, "learning_rate": 1.136297168298085e-06, "loss": 0.153, "step": 3014 }, { "epoch": 5.11, "learning_rate": 1.132064104273133e-06, "loss": 0.1525, "step": 3015 }, { "epoch": 5.11, "learning_rate": 1.1278384665851816e-06, "loss": 0.1538, "step": 3016 }, { "epoch": 5.11, "learning_rate": 1.1236202587729372e-06, "loss": 0.1555, "step": 3017 }, { "epoch": 5.11, "learning_rate": 1.119409484368882e-06, "loss": 0.157, "step": 3018 }, { "epoch": 5.12, "learning_rate": 1.1152061468992758e-06, "loss": 0.158, "step": 3019 }, { "epoch": 5.12, "learning_rate": 1.1110102498841447e-06, "loss": 0.1608, "step": 3020 }, { "epoch": 5.12, "learning_rate": 1.1068217968372908e-06, "loss": 0.1584, "step": 3021 }, { "epoch": 5.12, "learning_rate": 1.1026407912662774e-06, "loss": 0.1529, "step": 3022 }, { "epoch": 5.12, "learning_rate": 1.0984672366724348e-06, "loss": 0.1572, "step": 3023 }, { "epoch": 5.12, "learning_rate": 1.0943011365508494e-06, "loss": 0.154, "step": 3024 }, { "epoch": 5.13, "learning_rate": 1.0901424943903726e-06, "loss": 0.1559, "step": 3025 }, { "epoch": 5.13, "learning_rate": 1.0859913136735979e-06, "loss": 0.156, "step": 3026 }, { "epoch": 5.13, "learning_rate": 1.0818475978768817e-06, "loss": 0.1525, "step": 3027 }, { "epoch": 5.13, "learning_rate": 1.0777113504703228e-06, "loss": 0.1505, "step": 3028 }, { "epoch": 5.13, "learning_rate": 1.0735825749177685e-06, "loss": 0.1546, "step": 3029 }, { "epoch": 5.13, "learning_rate": 1.0694612746768073e-06, "loss": 0.1575, "step": 3030 }, { "epoch": 5.14, "learning_rate": 1.0653474531987705e-06, "loss": 0.1564, "step": 3031 }, { "epoch": 5.14, "learning_rate": 1.0612411139287204e-06, "loss": 0.1569, "step": 3032 }, { "epoch": 5.14, "learning_rate": 1.0571422603054593e-06, "loss": 0.1543, "step": 3033 }, { "epoch": 5.14, "learning_rate": 1.0530508957615182e-06, "loss": 0.1531, "step": 3034 }, { "epoch": 5.14, "learning_rate": 1.0489670237231564e-06, "loss": 0.1585, "step": 3035 }, { "epoch": 5.14, "learning_rate": 1.0448906476103605e-06, "loss": 0.1571, "step": 3036 }, { "epoch": 5.15, "learning_rate": 1.0408217708368385e-06, "loss": 0.1508, "step": 3037 }, { "epoch": 5.15, "learning_rate": 1.036760396810016e-06, "loss": 0.1516, "step": 3038 }, { "epoch": 5.15, "learning_rate": 1.032706528931038e-06, "loss": 0.1573, "step": 3039 }, { "epoch": 5.15, "learning_rate": 1.028660170594764e-06, "loss": 0.1552, "step": 3040 }, { "epoch": 5.15, "learning_rate": 1.024621325189764e-06, "loss": 0.1545, "step": 3041 }, { "epoch": 5.15, "learning_rate": 1.0205899960983167e-06, "loss": 0.155, "step": 3042 }, { "epoch": 5.16, "learning_rate": 1.016566186696406e-06, "loss": 0.1559, "step": 3043 }, { "epoch": 5.16, "learning_rate": 1.0125499003537176e-06, "loss": 0.1527, "step": 3044 }, { "epoch": 5.16, "learning_rate": 1.0085411404336365e-06, "loss": 0.1515, "step": 3045 }, { "epoch": 5.16, "learning_rate": 1.0045399102932485e-06, "loss": 0.1506, "step": 3046 }, { "epoch": 5.16, "learning_rate": 1.000546213283331e-06, "loss": 0.1532, "step": 3047 }, { "epoch": 5.17, "learning_rate": 9.965600527483543e-07, "loss": 0.1596, "step": 3048 }, { "epoch": 5.17, "learning_rate": 9.92581432026476e-07, "loss": 0.1542, "step": 3049 }, { "epoch": 5.17, "learning_rate": 9.886103544495384e-07, "loss": 0.1506, "step": 3050 }, { "epoch": 5.17, "learning_rate": 9.846468233430695e-07, "loss": 0.1493, "step": 3051 }, { "epoch": 5.17, "learning_rate": 9.806908420262762e-07, "loss": 0.1497, "step": 3052 }, { "epoch": 5.17, "learning_rate": 9.76742413812044e-07, "loss": 0.1499, "step": 3053 }, { "epoch": 5.18, "learning_rate": 9.728015420069327e-07, "loss": 0.1575, "step": 3054 }, { "epoch": 5.18, "learning_rate": 9.688682299111752e-07, "loss": 0.1542, "step": 3055 }, { "epoch": 5.18, "learning_rate": 9.649424808186692e-07, "loss": 0.1511, "step": 3056 }, { "epoch": 5.18, "learning_rate": 9.610242980169826e-07, "loss": 0.1544, "step": 3057 }, { "epoch": 5.18, "learning_rate": 9.571136847873485e-07, "loss": 0.1554, "step": 3058 }, { "epoch": 5.18, "learning_rate": 9.532106444046562e-07, "loss": 0.1536, "step": 3059 }, { "epoch": 5.19, "learning_rate": 9.493151801374578e-07, "loss": 0.1557, "step": 3060 }, { "epoch": 5.19, "learning_rate": 9.454272952479604e-07, "loss": 0.1541, "step": 3061 }, { "epoch": 5.19, "learning_rate": 9.415469929920174e-07, "loss": 0.1606, "step": 3062 }, { "epoch": 5.19, "learning_rate": 9.376742766191427e-07, "loss": 0.164, "step": 3063 }, { "epoch": 5.19, "learning_rate": 9.338091493724877e-07, "loss": 0.1547, "step": 3064 }, { "epoch": 5.19, "learning_rate": 9.299516144888543e-07, "loss": 0.1567, "step": 3065 }, { "epoch": 5.2, "learning_rate": 9.26101675198684e-07, "loss": 0.1617, "step": 3066 }, { "epoch": 5.2, "learning_rate": 9.2225933472606e-07, "loss": 0.1567, "step": 3067 }, { "epoch": 5.2, "learning_rate": 9.184245962886984e-07, "loss": 0.1564, "step": 3068 }, { "epoch": 5.2, "learning_rate": 9.145974630979526e-07, "loss": 0.1555, "step": 3069 }, { "epoch": 5.2, "learning_rate": 9.107779383588033e-07, "loss": 0.1554, "step": 3070 }, { "epoch": 5.2, "learning_rate": 9.069660252698608e-07, "loss": 0.1549, "step": 3071 }, { "epoch": 5.21, "learning_rate": 9.03161727023365e-07, "loss": 0.1571, "step": 3072 }, { "epoch": 5.21, "learning_rate": 8.99365046805174e-07, "loss": 0.1571, "step": 3073 }, { "epoch": 5.21, "learning_rate": 8.955759877947668e-07, "loss": 0.1558, "step": 3074 }, { "epoch": 5.21, "learning_rate": 8.917945531652461e-07, "loss": 0.1564, "step": 3075 }, { "epoch": 5.21, "learning_rate": 8.88020746083319e-07, "loss": 0.1567, "step": 3076 }, { "epoch": 5.21, "learning_rate": 8.84254569709313e-07, "loss": 0.153, "step": 3077 }, { "epoch": 5.22, "learning_rate": 8.804960271971641e-07, "loss": 0.1526, "step": 3078 }, { "epoch": 5.22, "learning_rate": 8.767451216944134e-07, "loss": 0.1575, "step": 3079 }, { "epoch": 5.22, "learning_rate": 8.730018563422094e-07, "loss": 0.1552, "step": 3080 }, { "epoch": 5.22, "learning_rate": 8.692662342753011e-07, "loss": 0.153, "step": 3081 }, { "epoch": 5.22, "learning_rate": 8.65538258622034e-07, "loss": 0.152, "step": 3082 }, { "epoch": 5.22, "learning_rate": 8.618179325043529e-07, "loss": 0.1561, "step": 3083 }, { "epoch": 5.23, "learning_rate": 8.581052590377981e-07, "loss": 0.1569, "step": 3084 }, { "epoch": 5.23, "learning_rate": 8.544002413314967e-07, "loss": 0.154, "step": 3085 }, { "epoch": 5.23, "learning_rate": 8.507028824881702e-07, "loss": 0.1538, "step": 3086 }, { "epoch": 5.23, "learning_rate": 8.470131856041242e-07, "loss": 0.1537, "step": 3087 }, { "epoch": 5.23, "learning_rate": 8.433311537692424e-07, "loss": 0.1568, "step": 3088 }, { "epoch": 5.23, "learning_rate": 8.396567900669983e-07, "loss": 0.1526, "step": 3089 }, { "epoch": 5.24, "learning_rate": 8.359900975744373e-07, "loss": 0.1584, "step": 3090 }, { "epoch": 5.24, "learning_rate": 8.323310793621853e-07, "loss": 0.1565, "step": 3091 }, { "epoch": 5.24, "learning_rate": 8.286797384944378e-07, "loss": 0.1561, "step": 3092 }, { "epoch": 5.24, "learning_rate": 8.250360780289646e-07, "loss": 0.1542, "step": 3093 }, { "epoch": 5.24, "learning_rate": 8.21400101017098e-07, "loss": 0.1528, "step": 3094 }, { "epoch": 5.24, "learning_rate": 8.177718105037424e-07, "loss": 0.1534, "step": 3095 }, { "epoch": 5.25, "learning_rate": 8.141512095273595e-07, "loss": 0.1528, "step": 3096 }, { "epoch": 5.25, "learning_rate": 8.105383011199775e-07, "loss": 0.1546, "step": 3097 }, { "epoch": 5.25, "learning_rate": 8.069330883071768e-07, "loss": 0.1554, "step": 3098 }, { "epoch": 5.25, "learning_rate": 8.033355741080983e-07, "loss": 0.1535, "step": 3099 }, { "epoch": 5.25, "learning_rate": 7.997457615354309e-07, "loss": 0.154, "step": 3100 }, { "epoch": 5.25, "learning_rate": 7.961636535954153e-07, "loss": 0.1591, "step": 3101 }, { "epoch": 5.26, "learning_rate": 7.925892532878443e-07, "loss": 0.1553, "step": 3102 }, { "epoch": 5.26, "learning_rate": 7.8902256360605e-07, "loss": 0.1576, "step": 3103 }, { "epoch": 5.26, "learning_rate": 7.854635875369121e-07, "loss": 0.1579, "step": 3104 }, { "epoch": 5.26, "learning_rate": 7.819123280608487e-07, "loss": 0.1557, "step": 3105 }, { "epoch": 5.26, "learning_rate": 7.78368788151812e-07, "loss": 0.1523, "step": 3106 }, { "epoch": 5.27, "learning_rate": 7.748329707772961e-07, "loss": 0.1507, "step": 3107 }, { "epoch": 5.27, "learning_rate": 7.713048788983257e-07, "loss": 0.1552, "step": 3108 }, { "epoch": 5.27, "learning_rate": 7.67784515469453e-07, "loss": 0.1557, "step": 3109 }, { "epoch": 5.27, "learning_rate": 7.642718834387619e-07, "loss": 0.1574, "step": 3110 }, { "epoch": 5.27, "learning_rate": 7.607669857478617e-07, "loss": 0.1525, "step": 3111 }, { "epoch": 5.27, "learning_rate": 7.572698253318789e-07, "loss": 0.1553, "step": 3112 }, { "epoch": 5.28, "learning_rate": 7.537804051194697e-07, "loss": 0.1597, "step": 3113 }, { "epoch": 5.28, "learning_rate": 7.502987280327989e-07, "loss": 0.1536, "step": 3114 }, { "epoch": 5.28, "learning_rate": 7.46824796987553e-07, "loss": 0.1582, "step": 3115 }, { "epoch": 5.28, "learning_rate": 7.433586148929305e-07, "loss": 0.1525, "step": 3116 }, { "epoch": 5.28, "learning_rate": 7.399001846516407e-07, "loss": 0.1515, "step": 3117 }, { "epoch": 5.28, "learning_rate": 7.36449509159901e-07, "loss": 0.1512, "step": 3118 }, { "epoch": 5.29, "learning_rate": 7.330065913074347e-07, "loss": 0.1499, "step": 3119 }, { "epoch": 5.29, "learning_rate": 7.295714339774662e-07, "loss": 0.153, "step": 3120 }, { "epoch": 5.29, "learning_rate": 7.261440400467246e-07, "loss": 0.1555, "step": 3121 }, { "epoch": 5.29, "learning_rate": 7.22724412385436e-07, "loss": 0.1555, "step": 3122 }, { "epoch": 5.29, "learning_rate": 7.193125538573232e-07, "loss": 0.1561, "step": 3123 }, { "epoch": 5.29, "learning_rate": 7.159084673196015e-07, "loss": 0.1543, "step": 3124 }, { "epoch": 5.3, "learning_rate": 7.125121556229819e-07, "loss": 0.1566, "step": 3125 }, { "epoch": 5.3, "learning_rate": 7.09123621611657e-07, "loss": 0.1564, "step": 3126 }, { "epoch": 5.3, "learning_rate": 7.057428681233114e-07, "loss": 0.1535, "step": 3127 }, { "epoch": 5.3, "learning_rate": 7.023698979891136e-07, "loss": 0.1539, "step": 3128 }, { "epoch": 5.3, "learning_rate": 6.990047140337142e-07, "loss": 0.1531, "step": 3129 }, { "epoch": 5.3, "learning_rate": 6.956473190752411e-07, "loss": 0.1515, "step": 3130 }, { "epoch": 5.31, "learning_rate": 6.922977159253042e-07, "loss": 0.1495, "step": 3131 }, { "epoch": 5.31, "learning_rate": 6.889559073889784e-07, "loss": 0.1533, "step": 3132 }, { "epoch": 5.31, "learning_rate": 6.856218962648231e-07, "loss": 0.1576, "step": 3133 }, { "epoch": 5.31, "learning_rate": 6.822956853448592e-07, "loss": 0.1531, "step": 3134 }, { "epoch": 5.31, "learning_rate": 6.789772774145808e-07, "loss": 0.1556, "step": 3135 }, { "epoch": 5.31, "learning_rate": 6.756666752529439e-07, "loss": 0.1556, "step": 3136 }, { "epoch": 5.32, "learning_rate": 6.723638816323707e-07, "loss": 0.1597, "step": 3137 }, { "epoch": 5.32, "learning_rate": 6.6906889931874e-07, "loss": 0.159, "step": 3138 }, { "epoch": 5.32, "learning_rate": 6.657817310713932e-07, "loss": 0.1505, "step": 3139 }, { "epoch": 5.32, "learning_rate": 6.625023796431252e-07, "loss": 0.1492, "step": 3140 }, { "epoch": 5.32, "learning_rate": 6.592308477801878e-07, "loss": 0.1482, "step": 3141 }, { "epoch": 5.32, "learning_rate": 6.55967138222281e-07, "loss": 0.1564, "step": 3142 }, { "epoch": 5.33, "learning_rate": 6.527112537025582e-07, "loss": 0.1602, "step": 3143 }, { "epoch": 5.33, "learning_rate": 6.49463196947614e-07, "loss": 0.1575, "step": 3144 }, { "epoch": 5.33, "learning_rate": 6.462229706774925e-07, "loss": 0.1548, "step": 3145 }, { "epoch": 5.33, "learning_rate": 6.429905776056789e-07, "loss": 0.1534, "step": 3146 }, { "epoch": 5.33, "learning_rate": 6.397660204390988e-07, "loss": 0.153, "step": 3147 }, { "epoch": 5.33, "learning_rate": 6.365493018781144e-07, "loss": 0.1537, "step": 3148 }, { "epoch": 5.34, "learning_rate": 6.333404246165276e-07, "loss": 0.1505, "step": 3149 }, { "epoch": 5.34, "learning_rate": 6.301393913415677e-07, "loss": 0.1518, "step": 3150 }, { "epoch": 5.34, "learning_rate": 6.269462047338981e-07, "loss": 0.1483, "step": 3151 }, { "epoch": 5.34, "learning_rate": 6.237608674676122e-07, "loss": 0.1542, "step": 3152 }, { "epoch": 5.34, "learning_rate": 6.205833822102281e-07, "loss": 0.1538, "step": 3153 }, { "epoch": 5.34, "learning_rate": 6.17413751622692e-07, "loss": 0.1546, "step": 3154 }, { "epoch": 5.35, "learning_rate": 6.142519783593681e-07, "loss": 0.159, "step": 3155 }, { "epoch": 5.35, "learning_rate": 6.110980650680399e-07, "loss": 0.1567, "step": 3156 }, { "epoch": 5.35, "learning_rate": 6.079520143899131e-07, "loss": 0.1477, "step": 3157 }, { "epoch": 5.35, "learning_rate": 6.048138289596073e-07, "loss": 0.1533, "step": 3158 }, { "epoch": 5.35, "learning_rate": 6.016835114051523e-07, "loss": 0.1552, "step": 3159 }, { "epoch": 5.35, "learning_rate": 5.985610643479945e-07, "loss": 0.156, "step": 3160 }, { "epoch": 5.36, "learning_rate": 5.954464904029844e-07, "loss": 0.1578, "step": 3161 }, { "epoch": 5.36, "learning_rate": 5.923397921783835e-07, "loss": 0.1545, "step": 3162 }, { "epoch": 5.36, "learning_rate": 5.89240972275854e-07, "loss": 0.1609, "step": 3163 }, { "epoch": 5.36, "learning_rate": 5.861500332904613e-07, "loss": 0.1579, "step": 3164 }, { "epoch": 5.36, "learning_rate": 5.830669778106723e-07, "loss": 0.1602, "step": 3165 }, { "epoch": 5.36, "learning_rate": 5.799918084183509e-07, "loss": 0.1558, "step": 3166 }, { "epoch": 5.37, "learning_rate": 5.769245276887603e-07, "loss": 0.1556, "step": 3167 }, { "epoch": 5.37, "learning_rate": 5.738651381905514e-07, "loss": 0.1612, "step": 3168 }, { "epoch": 5.37, "learning_rate": 5.708136424857735e-07, "loss": 0.1579, "step": 3169 }, { "epoch": 5.37, "learning_rate": 5.677700431298583e-07, "loss": 0.1537, "step": 3170 }, { "epoch": 5.37, "learning_rate": 5.6473434267163e-07, "loss": 0.153, "step": 3171 }, { "epoch": 5.38, "learning_rate": 5.617065436532953e-07, "loss": 0.1534, "step": 3172 }, { "epoch": 5.38, "learning_rate": 5.586866486104458e-07, "loss": 0.152, "step": 3173 }, { "epoch": 5.38, "learning_rate": 5.556746600720541e-07, "loss": 0.1563, "step": 3174 }, { "epoch": 5.38, "learning_rate": 5.526705805604693e-07, "loss": 0.1578, "step": 3175 }, { "epoch": 5.38, "learning_rate": 5.496744125914177e-07, "loss": 0.1524, "step": 3176 }, { "epoch": 5.38, "learning_rate": 5.466861586740024e-07, "loss": 0.1502, "step": 3177 }, { "epoch": 5.39, "learning_rate": 5.437058213106971e-07, "loss": 0.1494, "step": 3178 }, { "epoch": 5.39, "learning_rate": 5.407334029973466e-07, "loss": 0.1575, "step": 3179 }, { "epoch": 5.39, "learning_rate": 5.377689062231628e-07, "loss": 0.1573, "step": 3180 }, { "epoch": 5.39, "learning_rate": 5.348123334707278e-07, "loss": 0.154, "step": 3181 }, { "epoch": 5.39, "learning_rate": 5.31863687215981e-07, "loss": 0.1514, "step": 3182 }, { "epoch": 5.39, "learning_rate": 5.289229699282294e-07, "loss": 0.1559, "step": 3183 }, { "epoch": 5.4, "learning_rate": 5.259901840701376e-07, "loss": 0.1683, "step": 3184 }, { "epoch": 5.4, "learning_rate": 5.230653320977296e-07, "loss": 0.1574, "step": 3185 }, { "epoch": 5.4, "learning_rate": 5.201484164603831e-07, "loss": 0.1563, "step": 3186 }, { "epoch": 5.4, "learning_rate": 5.172394396008351e-07, "loss": 0.1543, "step": 3187 }, { "epoch": 5.4, "learning_rate": 5.143384039551646e-07, "loss": 0.1498, "step": 3188 }, { "epoch": 5.4, "learning_rate": 5.114453119528096e-07, "loss": 0.1548, "step": 3189 }, { "epoch": 5.41, "learning_rate": 5.085601660165496e-07, "loss": 0.1578, "step": 3190 }, { "epoch": 5.41, "learning_rate": 5.056829685625154e-07, "loss": 0.1525, "step": 3191 }, { "epoch": 5.41, "learning_rate": 5.028137220001761e-07, "loss": 0.1532, "step": 3192 }, { "epoch": 5.41, "learning_rate": 4.999524287323465e-07, "loss": 0.1525, "step": 3193 }, { "epoch": 5.41, "learning_rate": 4.970990911551777e-07, "loss": 0.1556, "step": 3194 }, { "epoch": 5.41, "learning_rate": 4.942537116581592e-07, "loss": 0.1581, "step": 3195 }, { "epoch": 5.42, "learning_rate": 4.914162926241195e-07, "loss": 0.1543, "step": 3196 }, { "epoch": 5.42, "learning_rate": 4.885868364292168e-07, "loss": 0.1511, "step": 3197 }, { "epoch": 5.42, "learning_rate": 4.857653454429412e-07, "loss": 0.1529, "step": 3198 }, { "epoch": 5.42, "learning_rate": 4.82951822028116e-07, "loss": 0.1571, "step": 3199 }, { "epoch": 5.42, "learning_rate": 4.801462685408875e-07, "loss": 0.1606, "step": 3200 }, { "epoch": 5.42, "learning_rate": 4.773486873307298e-07, "loss": 0.1552, "step": 3201 }, { "epoch": 5.43, "learning_rate": 4.74559080740441e-07, "loss": 0.157, "step": 3202 }, { "epoch": 5.43, "learning_rate": 4.7177745110614013e-07, "loss": 0.1568, "step": 3203 }, { "epoch": 5.43, "learning_rate": 4.6900380075726836e-07, "loss": 0.1567, "step": 3204 }, { "epoch": 5.43, "learning_rate": 4.662381320165832e-07, "loss": 0.1546, "step": 3205 }, { "epoch": 5.43, "learning_rate": 4.6348044720015526e-07, "loss": 0.1558, "step": 3206 }, { "epoch": 5.43, "learning_rate": 4.6073074861737286e-07, "loss": 0.1536, "step": 3207 }, { "epoch": 5.44, "learning_rate": 4.57989038570934e-07, "loss": 0.1562, "step": 3208 }, { "epoch": 5.44, "learning_rate": 4.55255319356851e-07, "loss": 0.1526, "step": 3209 }, { "epoch": 5.44, "learning_rate": 4.52529593264438e-07, "loss": 0.1515, "step": 3210 }, { "epoch": 5.44, "learning_rate": 4.498118625763204e-07, "loss": 0.1535, "step": 3211 }, { "epoch": 5.44, "learning_rate": 4.4710212956842636e-07, "loss": 0.1581, "step": 3212 }, { "epoch": 5.44, "learning_rate": 4.4440039650998746e-07, "loss": 0.158, "step": 3213 }, { "epoch": 5.45, "learning_rate": 4.417066656635305e-07, "loss": 0.1572, "step": 3214 }, { "epoch": 5.45, "learning_rate": 4.3902093928488877e-07, "loss": 0.1528, "step": 3215 }, { "epoch": 5.45, "learning_rate": 4.363432196231876e-07, "loss": 0.1538, "step": 3216 }, { "epoch": 5.45, "learning_rate": 4.336735089208477e-07, "loss": 0.1523, "step": 3217 }, { "epoch": 5.45, "learning_rate": 4.310118094135851e-07, "loss": 0.1555, "step": 3218 }, { "epoch": 5.45, "learning_rate": 4.283581233304046e-07, "loss": 0.1557, "step": 3219 }, { "epoch": 5.46, "learning_rate": 4.2571245289360077e-07, "loss": 0.1518, "step": 3220 }, { "epoch": 5.46, "learning_rate": 4.230748003187546e-07, "loss": 0.1508, "step": 3221 }, { "epoch": 5.46, "learning_rate": 4.2044516781473476e-07, "loss": 0.153, "step": 3222 }, { "epoch": 5.46, "learning_rate": 4.178235575836931e-07, "loss": 0.1571, "step": 3223 }, { "epoch": 5.46, "learning_rate": 4.1520997182106225e-07, "loss": 0.1572, "step": 3224 }, { "epoch": 5.46, "learning_rate": 4.1260441271555707e-07, "loss": 0.1518, "step": 3225 }, { "epoch": 5.47, "learning_rate": 4.100068824491665e-07, "loss": 0.1536, "step": 3226 }, { "epoch": 5.47, "learning_rate": 4.074173831971584e-07, "loss": 0.151, "step": 3227 }, { "epoch": 5.47, "learning_rate": 4.048359171280769e-07, "loss": 0.1528, "step": 3228 }, { "epoch": 5.47, "learning_rate": 4.02262486403735e-07, "loss": 0.1533, "step": 3229 }, { "epoch": 5.47, "learning_rate": 3.99697093179221e-07, "loss": 0.1511, "step": 3230 }, { "epoch": 5.48, "learning_rate": 3.9713973960288864e-07, "loss": 0.1547, "step": 3231 }, { "epoch": 5.48, "learning_rate": 3.9459042781636035e-07, "loss": 0.1519, "step": 3232 }, { "epoch": 5.48, "learning_rate": 3.9204915995452284e-07, "loss": 0.1511, "step": 3233 }, { "epoch": 5.48, "learning_rate": 3.8951593814552936e-07, "loss": 0.1598, "step": 3234 }, { "epoch": 5.48, "learning_rate": 3.8699076451079175e-07, "loss": 0.1533, "step": 3235 }, { "epoch": 5.48, "learning_rate": 3.844736411649852e-07, "loss": 0.1538, "step": 3236 }, { "epoch": 5.49, "learning_rate": 3.819645702160435e-07, "loss": 0.1578, "step": 3237 }, { "epoch": 5.49, "learning_rate": 3.7946355376515033e-07, "loss": 0.1561, "step": 3238 }, { "epoch": 5.49, "learning_rate": 3.769705939067525e-07, "loss": 0.157, "step": 3239 }, { "epoch": 5.49, "learning_rate": 3.7448569272854564e-07, "loss": 0.1523, "step": 3240 }, { "epoch": 5.49, "learning_rate": 3.7200885231147844e-07, "loss": 0.1559, "step": 3241 }, { "epoch": 5.49, "learning_rate": 3.695400747297484e-07, "loss": 0.1542, "step": 3242 }, { "epoch": 5.5, "learning_rate": 3.6707936205080174e-07, "loss": 0.1583, "step": 3243 }, { "epoch": 5.5, "learning_rate": 3.6462671633532896e-07, "loss": 0.1614, "step": 3244 }, { "epoch": 5.5, "learning_rate": 3.62182139637266e-07, "loss": 0.1508, "step": 3245 }, { "epoch": 5.5, "learning_rate": 3.597456340037919e-07, "loss": 0.1563, "step": 3246 }, { "epoch": 5.5, "learning_rate": 3.5731720147532676e-07, "loss": 0.1553, "step": 3247 }, { "epoch": 5.5, "learning_rate": 3.548968440855294e-07, "loss": 0.1604, "step": 3248 }, { "epoch": 5.51, "learning_rate": 3.524845638612984e-07, "loss": 0.1598, "step": 3249 }, { "epoch": 5.51, "learning_rate": 3.5008036282276335e-07, "loss": 0.1549, "step": 3250 }, { "epoch": 5.51, "learning_rate": 3.476842429832916e-07, "loss": 0.1503, "step": 3251 }, { "epoch": 5.51, "learning_rate": 3.452962063494836e-07, "loss": 0.1504, "step": 3252 }, { "epoch": 5.51, "learning_rate": 3.429162549211684e-07, "loss": 0.1507, "step": 3253 }, { "epoch": 5.51, "learning_rate": 3.405443906914063e-07, "loss": 0.1558, "step": 3254 }, { "epoch": 5.52, "learning_rate": 3.3818061564648284e-07, "loss": 0.1594, "step": 3255 }, { "epoch": 5.52, "learning_rate": 3.3582493176591233e-07, "loss": 0.1535, "step": 3256 }, { "epoch": 5.52, "learning_rate": 3.33477341022429e-07, "loss": 0.1532, "step": 3257 }, { "epoch": 5.52, "learning_rate": 3.3113784538199355e-07, "loss": 0.1533, "step": 3258 }, { "epoch": 5.52, "learning_rate": 3.2880644680378547e-07, "loss": 0.1546, "step": 3259 }, { "epoch": 5.52, "learning_rate": 3.264831472402041e-07, "loss": 0.1571, "step": 3260 }, { "epoch": 5.53, "learning_rate": 3.241679486368665e-07, "loss": 0.1557, "step": 3261 }, { "epoch": 5.53, "learning_rate": 3.2186085293260616e-07, "loss": 0.1529, "step": 3262 }, { "epoch": 5.53, "learning_rate": 3.1956186205946983e-07, "loss": 0.151, "step": 3263 }, { "epoch": 5.53, "learning_rate": 3.172709779427152e-07, "loss": 0.1511, "step": 3264 }, { "epoch": 5.53, "learning_rate": 3.149882025008155e-07, "loss": 0.1554, "step": 3265 }, { "epoch": 5.53, "learning_rate": 3.127135376454515e-07, "loss": 0.1539, "step": 3266 }, { "epoch": 5.54, "learning_rate": 3.104469852815095e-07, "loss": 0.1522, "step": 3267 }, { "epoch": 5.54, "learning_rate": 3.081885473070856e-07, "loss": 0.1556, "step": 3268 }, { "epoch": 5.54, "learning_rate": 3.059382256134802e-07, "loss": 0.1547, "step": 3269 }, { "epoch": 5.54, "learning_rate": 3.0369602208519364e-07, "loss": 0.1568, "step": 3270 }, { "epoch": 5.54, "learning_rate": 3.0146193859993066e-07, "loss": 0.1551, "step": 3271 }, { "epoch": 5.54, "learning_rate": 2.9923597702859577e-07, "loss": 0.154, "step": 3272 }, { "epoch": 5.55, "learning_rate": 2.970181392352911e-07, "loss": 0.1531, "step": 3273 }, { "epoch": 5.55, "learning_rate": 2.948084270773155e-07, "loss": 0.1551, "step": 3274 }, { "epoch": 5.55, "learning_rate": 2.9260684240516535e-07, "loss": 0.1552, "step": 3275 }, { "epoch": 5.55, "learning_rate": 2.904133870625259e-07, "loss": 0.1523, "step": 3276 }, { "epoch": 5.55, "learning_rate": 2.882280628862799e-07, "loss": 0.1564, "step": 3277 }, { "epoch": 5.55, "learning_rate": 2.8605087170649804e-07, "loss": 0.1541, "step": 3278 }, { "epoch": 5.56, "learning_rate": 2.8388181534644065e-07, "loss": 0.1537, "step": 3279 }, { "epoch": 5.56, "learning_rate": 2.817208956225548e-07, "loss": 0.152, "step": 3280 }, { "epoch": 5.56, "learning_rate": 2.7956811434447617e-07, "loss": 0.151, "step": 3281 }, { "epoch": 5.56, "learning_rate": 2.774234733150216e-07, "loss": 0.1572, "step": 3282 }, { "epoch": 5.56, "learning_rate": 2.7528697433019336e-07, "loss": 0.1574, "step": 3283 }, { "epoch": 5.56, "learning_rate": 2.7315861917917464e-07, "loss": 0.1562, "step": 3284 }, { "epoch": 5.57, "learning_rate": 2.710384096443286e-07, "loss": 0.1565, "step": 3285 }, { "epoch": 5.57, "learning_rate": 2.6892634750119606e-07, "loss": 0.1526, "step": 3286 }, { "epoch": 5.57, "learning_rate": 2.668224345184989e-07, "loss": 0.1506, "step": 3287 }, { "epoch": 5.57, "learning_rate": 2.647266724581288e-07, "loss": 0.1502, "step": 3288 }, { "epoch": 5.57, "learning_rate": 2.626390630751541e-07, "loss": 0.1522, "step": 3289 }, { "epoch": 5.57, "learning_rate": 2.605596081178163e-07, "loss": 0.1524, "step": 3290 }, { "epoch": 5.58, "learning_rate": 2.584883093275292e-07, "loss": 0.1586, "step": 3291 }, { "epoch": 5.58, "learning_rate": 2.564251684388741e-07, "loss": 0.1578, "step": 3292 }, { "epoch": 5.58, "learning_rate": 2.543701871796023e-07, "loss": 0.1551, "step": 3293 }, { "epoch": 5.58, "learning_rate": 2.523233672706282e-07, "loss": 0.1549, "step": 3294 }, { "epoch": 5.58, "learning_rate": 2.502847104260364e-07, "loss": 0.1528, "step": 3295 }, { "epoch": 5.59, "learning_rate": 2.4825421835307453e-07, "loss": 0.1553, "step": 3296 }, { "epoch": 5.59, "learning_rate": 2.46231892752149e-07, "loss": 0.1525, "step": 3297 }, { "epoch": 5.59, "learning_rate": 2.442177353168307e-07, "loss": 0.1556, "step": 3298 }, { "epoch": 5.59, "learning_rate": 2.422117477338515e-07, "loss": 0.1555, "step": 3299 }, { "epoch": 5.59, "learning_rate": 2.402139316830976e-07, "loss": 0.1538, "step": 3300 }, { "epoch": 5.59, "learning_rate": 2.3822428883761406e-07, "loss": 0.1521, "step": 3301 }, { "epoch": 5.6, "learning_rate": 2.362428208636014e-07, "loss": 0.1556, "step": 3302 }, { "epoch": 5.6, "learning_rate": 2.3426952942041337e-07, "loss": 0.1567, "step": 3303 }, { "epoch": 5.6, "learning_rate": 2.323044161605592e-07, "loss": 0.153, "step": 3304 }, { "epoch": 5.6, "learning_rate": 2.3034748272969476e-07, "loss": 0.1529, "step": 3305 }, { "epoch": 5.6, "learning_rate": 2.2839873076663132e-07, "loss": 0.1531, "step": 3306 }, { "epoch": 5.6, "learning_rate": 2.2645816190332348e-07, "loss": 0.155, "step": 3307 }, { "epoch": 5.61, "learning_rate": 2.2452577776487572e-07, "loss": 0.1523, "step": 3308 }, { "epoch": 5.61, "learning_rate": 2.2260157996953692e-07, "loss": 0.1533, "step": 3309 }, { "epoch": 5.61, "learning_rate": 2.206855701287036e-07, "loss": 0.1569, "step": 3310 }, { "epoch": 5.61, "learning_rate": 2.1877774984691126e-07, "loss": 0.1553, "step": 3311 }, { "epoch": 5.61, "learning_rate": 2.1687812072184177e-07, "loss": 0.1523, "step": 3312 }, { "epoch": 5.61, "learning_rate": 2.1498668434431157e-07, "loss": 0.156, "step": 3313 }, { "epoch": 5.62, "learning_rate": 2.1310344229828028e-07, "loss": 0.1564, "step": 3314 }, { "epoch": 5.62, "learning_rate": 2.1122839616084412e-07, "loss": 0.1535, "step": 3315 }, { "epoch": 5.62, "learning_rate": 2.0936154750223703e-07, "loss": 0.153, "step": 3316 }, { "epoch": 5.62, "learning_rate": 2.0750289788582512e-07, "loss": 0.1542, "step": 3317 }, { "epoch": 5.62, "learning_rate": 2.056524488681122e-07, "loss": 0.1495, "step": 3318 }, { "epoch": 5.62, "learning_rate": 2.0381020199873314e-07, "loss": 0.157, "step": 3319 }, { "epoch": 5.63, "learning_rate": 2.0197615882045162e-07, "loss": 0.1569, "step": 3320 }, { "epoch": 5.63, "learning_rate": 2.0015032086916353e-07, "loss": 0.1513, "step": 3321 }, { "epoch": 5.63, "learning_rate": 1.9833268967389241e-07, "loss": 0.1553, "step": 3322 }, { "epoch": 5.63, "learning_rate": 1.9652326675679067e-07, "loss": 0.1571, "step": 3323 }, { "epoch": 5.63, "learning_rate": 1.9472205363313735e-07, "loss": 0.157, "step": 3324 }, { "epoch": 5.63, "learning_rate": 1.929290518113325e-07, "loss": 0.1563, "step": 3325 }, { "epoch": 5.64, "learning_rate": 1.9114426279290276e-07, "loss": 0.1572, "step": 3326 }, { "epoch": 5.64, "learning_rate": 1.8936768807249595e-07, "loss": 0.1547, "step": 3327 }, { "epoch": 5.64, "learning_rate": 1.8759932913788192e-07, "loss": 0.1555, "step": 3328 }, { "epoch": 5.64, "learning_rate": 1.8583918746994943e-07, "loss": 0.1561, "step": 3329 }, { "epoch": 5.64, "learning_rate": 1.840872645427072e-07, "loss": 0.1528, "step": 3330 }, { "epoch": 5.64, "learning_rate": 1.823435618232794e-07, "loss": 0.1549, "step": 3331 }, { "epoch": 5.65, "learning_rate": 1.8060808077190684e-07, "loss": 0.1565, "step": 3332 }, { "epoch": 5.65, "learning_rate": 1.7888082284194476e-07, "loss": 0.1543, "step": 3333 }, { "epoch": 5.65, "learning_rate": 1.7716178947986384e-07, "loss": 0.152, "step": 3334 }, { "epoch": 5.65, "learning_rate": 1.754509821252459e-07, "loss": 0.1584, "step": 3335 }, { "epoch": 5.65, "learning_rate": 1.7374840221078492e-07, "loss": 0.1575, "step": 3336 }, { "epoch": 5.65, "learning_rate": 1.7205405116228258e-07, "loss": 0.1538, "step": 3337 }, { "epoch": 5.66, "learning_rate": 1.7036793039865274e-07, "loss": 0.1562, "step": 3338 }, { "epoch": 5.66, "learning_rate": 1.686900413319137e-07, "loss": 0.1546, "step": 3339 }, { "epoch": 5.66, "learning_rate": 1.6702038536719145e-07, "loss": 0.1527, "step": 3340 }, { "epoch": 5.66, "learning_rate": 1.6535896390271977e-07, "loss": 0.1547, "step": 3341 }, { "epoch": 5.66, "learning_rate": 1.6370577832983237e-07, "loss": 0.1523, "step": 3342 }, { "epoch": 5.66, "learning_rate": 1.6206083003296846e-07, "loss": 0.1547, "step": 3343 }, { "epoch": 5.67, "learning_rate": 1.6042412038966948e-07, "loss": 0.158, "step": 3344 }, { "epoch": 5.67, "learning_rate": 1.5879565077057458e-07, "loss": 0.1532, "step": 3345 }, { "epoch": 5.67, "learning_rate": 1.5717542253942507e-07, "loss": 0.151, "step": 3346 }, { "epoch": 5.67, "learning_rate": 1.555634370530612e-07, "loss": 0.1548, "step": 3347 }, { "epoch": 5.67, "learning_rate": 1.5395969566141755e-07, "loss": 0.1559, "step": 3348 }, { "epoch": 5.67, "learning_rate": 1.5236419970752758e-07, "loss": 0.1559, "step": 3349 }, { "epoch": 5.68, "learning_rate": 1.5077695052751807e-07, "loss": 0.1508, "step": 3350 }, { "epoch": 5.68, "learning_rate": 1.4919794945060907e-07, "loss": 0.1522, "step": 3351 }, { "epoch": 5.68, "learning_rate": 1.4762719779911616e-07, "loss": 0.1544, "step": 3352 }, { "epoch": 5.68, "learning_rate": 1.460646968884427e-07, "loss": 0.1561, "step": 3353 }, { "epoch": 5.68, "learning_rate": 1.4451044802708758e-07, "loss": 0.1565, "step": 3354 }, { "epoch": 5.69, "learning_rate": 1.4296445251663293e-07, "loss": 0.1536, "step": 3355 }, { "epoch": 5.69, "learning_rate": 1.4142671165175537e-07, "loss": 0.1543, "step": 3356 }, { "epoch": 5.69, "learning_rate": 1.3989722672021366e-07, "loss": 0.1596, "step": 3357 }, { "epoch": 5.69, "learning_rate": 1.3837599900285658e-07, "loss": 0.1541, "step": 3358 }, { "epoch": 5.69, "learning_rate": 1.368630297736151e-07, "loss": 0.1536, "step": 3359 }, { "epoch": 5.69, "learning_rate": 1.353583202995057e-07, "loss": 0.1547, "step": 3360 }, { "epoch": 5.7, "learning_rate": 1.3386187184062815e-07, "loss": 0.1509, "step": 3361 }, { "epoch": 5.7, "learning_rate": 1.323736856501645e-07, "loss": 0.1497, "step": 3362 }, { "epoch": 5.7, "learning_rate": 1.308937629743756e-07, "loss": 0.1537, "step": 3363 }, { "epoch": 5.7, "learning_rate": 1.2942210505260344e-07, "loss": 0.1552, "step": 3364 }, { "epoch": 5.7, "learning_rate": 1.279587131172688e-07, "loss": 0.1513, "step": 3365 }, { "epoch": 5.7, "learning_rate": 1.2650358839387034e-07, "loss": 0.1535, "step": 3366 }, { "epoch": 5.71, "learning_rate": 1.2505673210098323e-07, "loss": 0.1521, "step": 3367 }, { "epoch": 5.71, "learning_rate": 1.236181454502583e-07, "loss": 0.1522, "step": 3368 }, { "epoch": 5.71, "learning_rate": 1.2218782964642073e-07, "loss": 0.1541, "step": 3369 }, { "epoch": 5.71, "learning_rate": 1.2076578588727016e-07, "loss": 0.1559, "step": 3370 }, { "epoch": 5.71, "learning_rate": 1.1935201536367958e-07, "loss": 0.1529, "step": 3371 }, { "epoch": 5.71, "learning_rate": 1.1794651925959078e-07, "loss": 0.1532, "step": 3372 }, { "epoch": 5.72, "learning_rate": 1.1654929875201781e-07, "loss": 0.1502, "step": 3373 }, { "epoch": 5.72, "learning_rate": 1.1516035501104695e-07, "loss": 0.1539, "step": 3374 }, { "epoch": 5.72, "learning_rate": 1.1377968919983106e-07, "loss": 0.1546, "step": 3375 }, { "epoch": 5.72, "learning_rate": 1.1240730247458754e-07, "loss": 0.1533, "step": 3376 }, { "epoch": 5.72, "learning_rate": 1.11043195984607e-07, "loss": 0.151, "step": 3377 }, { "epoch": 5.72, "learning_rate": 1.0968737087224013e-07, "loss": 0.1529, "step": 3378 }, { "epoch": 5.73, "learning_rate": 1.0833982827290646e-07, "loss": 0.1538, "step": 3379 }, { "epoch": 5.73, "learning_rate": 1.0700056931508774e-07, "loss": 0.1557, "step": 3380 }, { "epoch": 5.73, "learning_rate": 1.0566959512032904e-07, "loss": 0.1525, "step": 3381 }, { "epoch": 5.73, "learning_rate": 1.0434690680323656e-07, "loss": 0.1533, "step": 3382 }, { "epoch": 5.73, "learning_rate": 1.0303250547147869e-07, "loss": 0.1554, "step": 3383 }, { "epoch": 5.73, "learning_rate": 1.0172639222578273e-07, "loss": 0.1562, "step": 3384 }, { "epoch": 5.74, "learning_rate": 1.0042856815993818e-07, "loss": 0.1572, "step": 3385 }, { "epoch": 5.74, "learning_rate": 9.913903436078898e-08, "loss": 0.1546, "step": 3386 }, { "epoch": 5.74, "learning_rate": 9.785779190824018e-08, "loss": 0.1589, "step": 3387 }, { "epoch": 5.74, "learning_rate": 9.658484187525019e-08, "loss": 0.1533, "step": 3388 }, { "epoch": 5.74, "learning_rate": 9.532018532783405e-08, "loss": 0.1564, "step": 3389 }, { "epoch": 5.74, "learning_rate": 9.406382332506347e-08, "loss": 0.154, "step": 3390 }, { "epoch": 5.75, "learning_rate": 9.28157569190602e-08, "loss": 0.1539, "step": 3391 }, { "epoch": 5.75, "learning_rate": 9.15759871550026e-08, "loss": 0.1513, "step": 3392 }, { "epoch": 5.75, "learning_rate": 9.034451507112019e-08, "loss": 0.1522, "step": 3393 }, { "epoch": 5.75, "learning_rate": 8.912134169869135e-08, "loss": 0.1513, "step": 3394 }, { "epoch": 5.75, "learning_rate": 8.790646806204783e-08, "loss": 0.1521, "step": 3395 }, { "epoch": 5.75, "learning_rate": 8.669989517857025e-08, "loss": 0.1542, "step": 3396 }, { "epoch": 5.76, "learning_rate": 8.550162405868589e-08, "loss": 0.1622, "step": 3397 }, { "epoch": 5.76, "learning_rate": 8.431165570587096e-08, "loss": 0.1548, "step": 3398 }, { "epoch": 5.76, "learning_rate": 8.312999111665054e-08, "loss": 0.1545, "step": 3399 }, { "epoch": 5.76, "learning_rate": 8.195663128059416e-08, "loss": 0.1538, "step": 3400 }, { "epoch": 5.76, "learning_rate": 8.079157718031583e-08, "loss": 0.1581, "step": 3401 }, { "epoch": 5.76, "learning_rate": 7.963482979147508e-08, "loss": 0.1527, "step": 3402 }, { "epoch": 5.77, "learning_rate": 7.84863900827737e-08, "loss": 0.1513, "step": 3403 }, { "epoch": 5.77, "learning_rate": 7.734625901595905e-08, "loss": 0.1529, "step": 3404 }, { "epoch": 5.77, "learning_rate": 7.621443754581958e-08, "loss": 0.1586, "step": 3405 }, { "epoch": 5.77, "learning_rate": 7.509092662018269e-08, "loss": 0.1515, "step": 3406 }, { "epoch": 5.77, "learning_rate": 7.397572717991908e-08, "loss": 0.1482, "step": 3407 }, { "epoch": 5.77, "learning_rate": 7.286884015893725e-08, "loss": 0.1544, "step": 3408 }, { "epoch": 5.78, "learning_rate": 7.177026648418461e-08, "loss": 0.1529, "step": 3409 }, { "epoch": 5.78, "learning_rate": 7.06800070756497e-08, "loss": 0.1511, "step": 3410 }, { "epoch": 5.78, "learning_rate": 6.959806284635328e-08, "loss": 0.154, "step": 3411 }, { "epoch": 5.78, "learning_rate": 6.852443470235726e-08, "loss": 0.1556, "step": 3412 }, { "epoch": 5.78, "learning_rate": 6.745912354275685e-08, "loss": 0.154, "step": 3413 }, { "epoch": 5.78, "learning_rate": 6.6402130259684e-08, "loss": 0.1547, "step": 3414 }, { "epoch": 5.79, "learning_rate": 6.535345573830176e-08, "loss": 0.1537, "step": 3415 }, { "epoch": 5.79, "learning_rate": 6.431310085680986e-08, "loss": 0.1545, "step": 3416 }, { "epoch": 5.79, "learning_rate": 6.328106648644029e-08, "loss": 0.1618, "step": 3417 }, { "epoch": 5.79, "learning_rate": 6.225735349145612e-08, "loss": 0.152, "step": 3418 }, { "epoch": 5.79, "learning_rate": 6.124196272915384e-08, "loss": 0.1508, "step": 3419 }, { "epoch": 5.8, "learning_rate": 6.02348950498577e-08, "loss": 0.1517, "step": 3420 }, { "epoch": 5.8, "learning_rate": 5.923615129692417e-08, "loss": 0.1535, "step": 3421 }, { "epoch": 5.8, "learning_rate": 5.824573230673758e-08, "loss": 0.155, "step": 3422 }, { "epoch": 5.8, "learning_rate": 5.726363890871112e-08, "loss": 0.1576, "step": 3423 }, { "epoch": 5.8, "learning_rate": 5.62898719252869e-08, "loss": 0.1576, "step": 3424 }, { "epoch": 5.8, "learning_rate": 5.532443217193373e-08, "loss": 0.1534, "step": 3425 }, { "epoch": 5.81, "learning_rate": 5.436732045714488e-08, "loss": 0.1506, "step": 3426 }, { "epoch": 5.81, "learning_rate": 5.341853758244253e-08, "loss": 0.1541, "step": 3427 }, { "epoch": 5.81, "learning_rate": 5.2478084342371116e-08, "loss": 0.1557, "step": 3428 }, { "epoch": 5.81, "learning_rate": 5.1545961524502866e-08, "loss": 0.1508, "step": 3429 }, { "epoch": 5.81, "learning_rate": 5.062216990943003e-08, "loss": 0.1529, "step": 3430 }, { "epoch": 5.81, "learning_rate": 4.970671027077045e-08, "loss": 0.1547, "step": 3431 }, { "epoch": 5.82, "learning_rate": 4.8799583375164216e-08, "loss": 0.1536, "step": 3432 }, { "epoch": 5.82, "learning_rate": 4.7900789982271434e-08, "loss": 0.1535, "step": 3433 }, { "epoch": 5.82, "learning_rate": 4.7010330844777796e-08, "loss": 0.1543, "step": 3434 }, { "epoch": 5.82, "learning_rate": 4.6128206708383475e-08, "loss": 0.1591, "step": 3435 }, { "epoch": 5.82, "learning_rate": 4.52544183118131e-08, "loss": 0.1547, "step": 3436 }, { "epoch": 5.82, "learning_rate": 4.4388966386810225e-08, "loss": 0.1539, "step": 3437 }, { "epoch": 5.83, "learning_rate": 4.353185165813623e-08, "loss": 0.1528, "step": 3438 }, { "epoch": 5.83, "learning_rate": 4.268307484356804e-08, "loss": 0.1534, "step": 3439 }, { "epoch": 5.83, "learning_rate": 4.184263665390598e-08, "loss": 0.1553, "step": 3440 }, { "epoch": 5.83, "learning_rate": 4.101053779296149e-08, "loss": 0.1544, "step": 3441 }, { "epoch": 5.83, "learning_rate": 4.0186778957566065e-08, "loss": 0.1531, "step": 3442 }, { "epoch": 5.83, "learning_rate": 3.937136083756676e-08, "loss": 0.1522, "step": 3443 }, { "epoch": 5.84, "learning_rate": 3.856428411582291e-08, "loss": 0.1556, "step": 3444 }, { "epoch": 5.84, "learning_rate": 3.776554946821165e-08, "loss": 0.1544, "step": 3445 }, { "epoch": 5.84, "learning_rate": 3.697515756362125e-08, "loss": 0.1564, "step": 3446 }, { "epoch": 5.84, "learning_rate": 3.619310906395779e-08, "loss": 0.1573, "step": 3447 }, { "epoch": 5.84, "learning_rate": 3.541940462413518e-08, "loss": 0.152, "step": 3448 }, { "epoch": 5.84, "learning_rate": 3.4654044892082905e-08, "loss": 0.1493, "step": 3449 }, { "epoch": 5.85, "learning_rate": 3.389703050874271e-08, "loss": 0.1516, "step": 3450 }, { "epoch": 5.85, "learning_rate": 3.3148362108064155e-08, "loss": 0.1542, "step": 3451 }, { "epoch": 5.85, "learning_rate": 3.240804031701017e-08, "loss": 0.1538, "step": 3452 }, { "epoch": 5.85, "learning_rate": 3.167606575555482e-08, "loss": 0.1524, "step": 3453 }, { "epoch": 5.85, "learning_rate": 3.095243903667999e-08, "loss": 0.1549, "step": 3454 }, { "epoch": 5.85, "learning_rate": 3.0237160766377616e-08, "loss": 0.1516, "step": 3455 }, { "epoch": 5.86, "learning_rate": 2.9530231543648534e-08, "loss": 0.1542, "step": 3456 }, { "epoch": 5.86, "learning_rate": 2.8831651960500307e-08, "loss": 0.1548, "step": 3457 }, { "epoch": 5.86, "learning_rate": 2.814142260194941e-08, "loss": 0.1526, "step": 3458 }, { "epoch": 5.86, "learning_rate": 2.7459544046020136e-08, "loss": 0.1554, "step": 3459 }, { "epoch": 5.86, "learning_rate": 2.6786016863743492e-08, "loss": 0.1524, "step": 3460 }, { "epoch": 5.86, "learning_rate": 2.612084161915385e-08, "loss": 0.1531, "step": 3461 }, { "epoch": 5.87, "learning_rate": 2.5464018869294503e-08, "loss": 0.152, "step": 3462 }, { "epoch": 5.87, "learning_rate": 2.481554916421325e-08, "loss": 0.1571, "step": 3463 }, { "epoch": 5.87, "learning_rate": 2.4175433046962348e-08, "loss": 0.1548, "step": 3464 }, { "epoch": 5.87, "learning_rate": 2.354367105359967e-08, "loss": 0.1524, "step": 3465 }, { "epoch": 5.87, "learning_rate": 2.292026371318534e-08, "loss": 0.1545, "step": 3466 }, { "epoch": 5.87, "learning_rate": 2.230521154778287e-08, "loss": 0.1507, "step": 3467 }, { "epoch": 5.88, "learning_rate": 2.1698515072460236e-08, "loss": 0.1507, "step": 3468 }, { "epoch": 5.88, "learning_rate": 2.1100174795287697e-08, "loss": 0.1513, "step": 3469 }, { "epoch": 5.88, "learning_rate": 2.0510191217337773e-08, "loss": 0.1517, "step": 3470 }, { "epoch": 5.88, "learning_rate": 1.9928564832683017e-08, "loss": 0.1551, "step": 3471 }, { "epoch": 5.88, "learning_rate": 1.935529612840048e-08, "loss": 0.1574, "step": 3472 }, { "epoch": 5.88, "learning_rate": 1.879038558456614e-08, "loss": 0.1514, "step": 3473 }, { "epoch": 5.89, "learning_rate": 1.8233833674257128e-08, "loss": 0.1562, "step": 3474 }, { "epoch": 5.89, "learning_rate": 1.7685640863550624e-08, "loss": 0.1544, "step": 3475 }, { "epoch": 5.89, "learning_rate": 1.7145807611521625e-08, "loss": 0.1527, "step": 3476 }, { "epoch": 5.89, "learning_rate": 1.661433437024962e-08, "loss": 0.1498, "step": 3477 }, { "epoch": 5.89, "learning_rate": 1.609122158480747e-08, "loss": 0.1526, "step": 3478 }, { "epoch": 5.9, "learning_rate": 1.5576469693269204e-08, "loss": 0.1535, "step": 3479 }, { "epoch": 5.9, "learning_rate": 1.507007912670888e-08, "loss": 0.155, "step": 3480 }, { "epoch": 5.9, "learning_rate": 1.4572050309195063e-08, "loss": 0.1532, "step": 3481 }, { "epoch": 5.9, "learning_rate": 1.408238365779635e-08, "loss": 0.1501, "step": 3482 }, { "epoch": 5.9, "learning_rate": 1.3601079582576949e-08, "loss": 0.1514, "step": 3483 }, { "epoch": 5.9, "learning_rate": 1.3128138486598885e-08, "loss": 0.1506, "step": 3484 }, { "epoch": 5.91, "learning_rate": 1.26635607659209e-08, "loss": 0.1513, "step": 3485 }, { "epoch": 5.91, "learning_rate": 1.2207346809597343e-08, "loss": 0.152, "step": 3486 }, { "epoch": 5.91, "learning_rate": 1.1759496999679265e-08, "loss": 0.1534, "step": 3487 }, { "epoch": 5.91, "learning_rate": 1.132001171121111e-08, "loss": 0.1571, "step": 3488 }, { "epoch": 5.91, "learning_rate": 1.0888891312234029e-08, "loss": 0.1586, "step": 3489 }, { "epoch": 5.91, "learning_rate": 1.046613616378589e-08, "loss": 0.1578, "step": 3490 }, { "epoch": 5.92, "learning_rate": 1.0051746619896829e-08, "loss": 0.1533, "step": 3491 }, { "epoch": 5.92, "learning_rate": 9.64572302759148e-09, "loss": 0.1536, "step": 3492 }, { "epoch": 5.92, "learning_rate": 9.248065726890076e-09, "loss": 0.1559, "step": 3493 }, { "epoch": 5.92, "learning_rate": 8.858775050804013e-09, "loss": 0.1562, "step": 3494 }, { "epoch": 5.92, "learning_rate": 8.477851325340291e-09, "loss": 0.1595, "step": 3495 }, { "epoch": 5.92, "learning_rate": 8.105294869498182e-09, "loss": 0.1581, "step": 3496 }, { "epoch": 5.93, "learning_rate": 7.741105995271447e-09, "loss": 0.1606, "step": 3497 }, { "epoch": 5.93, "learning_rate": 7.385285007643905e-09, "loss": 0.1536, "step": 3498 }, { "epoch": 5.93, "learning_rate": 7.037832204593864e-09, "loss": 0.1522, "step": 3499 }, { "epoch": 5.93, "learning_rate": 6.698747877091905e-09, "loss": 0.1548, "step": 3500 }, { "epoch": 5.93, "learning_rate": 6.36803230909866e-09, "loss": 0.154, "step": 3501 }, { "epoch": 5.93, "learning_rate": 6.045685777567034e-09, "loss": 0.1521, "step": 3502 }, { "epoch": 5.94, "learning_rate": 5.7317085524422055e-09, "loss": 0.1518, "step": 3503 }, { "epoch": 5.94, "learning_rate": 5.426100896661624e-09, "loss": 0.1554, "step": 3504 }, { "epoch": 5.94, "learning_rate": 5.128863066151679e-09, "loss": 0.1538, "step": 3505 }, { "epoch": 5.94, "learning_rate": 4.839995309829926e-09, "loss": 0.1569, "step": 3506 }, { "epoch": 5.94, "learning_rate": 4.559497869605078e-09, "loss": 0.1581, "step": 3507 }, { "epoch": 5.94, "learning_rate": 4.287370980377015e-09, "loss": 0.153, "step": 3508 }, { "epoch": 5.95, "learning_rate": 4.023614870032333e-09, "loss": 0.1536, "step": 3509 }, { "epoch": 5.95, "learning_rate": 3.768229759453235e-09, "loss": 0.1515, "step": 3510 }, { "epoch": 5.95, "learning_rate": 3.521215862506422e-09, "loss": 0.1524, "step": 3511 }, { "epoch": 5.95, "learning_rate": 3.282573386051979e-09, "loss": 0.1557, "step": 3512 }, { "epoch": 5.95, "learning_rate": 3.0523025299356025e-09, "loss": 0.1561, "step": 3513 }, { "epoch": 5.95, "learning_rate": 2.8304034869974797e-09, "loss": 0.1534, "step": 3514 }, { "epoch": 5.96, "learning_rate": 2.6168764430622996e-09, "loss": 0.1601, "step": 3515 }, { "epoch": 5.96, "learning_rate": 2.4117215769459136e-09, "loss": 0.1539, "step": 3516 }, { "epoch": 5.96, "learning_rate": 2.214939060453114e-09, "loss": 0.1539, "step": 3517 }, { "epoch": 5.96, "learning_rate": 2.0265290583754148e-09, "loss": 0.1545, "step": 3518 }, { "epoch": 5.96, "learning_rate": 1.8464917284966023e-09, "loss": 0.1525, "step": 3519 }, { "epoch": 5.96, "learning_rate": 1.6748272215838523e-09, "loss": 0.152, "step": 3520 }, { "epoch": 5.97, "learning_rate": 1.5115356813977244e-09, "loss": 0.1588, "step": 3521 }, { "epoch": 5.97, "learning_rate": 1.3566172446832782e-09, "loss": 0.1522, "step": 3522 }, { "epoch": 5.97, "learning_rate": 1.210072041174515e-09, "loss": 0.1499, "step": 3523 }, { "epoch": 5.97, "learning_rate": 1.0719001935954876e-09, "loss": 0.154, "step": 3524 }, { "epoch": 5.97, "learning_rate": 9.421018176547504e-10, "loss": 0.1533, "step": 3525 }, { "epoch": 5.97, "learning_rate": 8.206770220520189e-10, "loss": 0.1588, "step": 3526 }, { "epoch": 5.98, "learning_rate": 7.076259084703996e-10, "loss": 0.1562, "step": 3527 }, { "epoch": 5.98, "learning_rate": 6.029485715852712e-10, "loss": 0.1571, "step": 3528 }, { "epoch": 5.98, "learning_rate": 5.06645099056513e-10, "loss": 0.1589, "step": 3529 }, { "epoch": 5.98, "learning_rate": 4.1871557153294604e-10, "loss": 0.1557, "step": 3530 }, { "epoch": 5.98, "learning_rate": 3.391600626478919e-10, "loss": 0.1518, "step": 3531 }, { "epoch": 5.98, "learning_rate": 2.6797863902694457e-10, "loss": 0.1498, "step": 3532 }, { "epoch": 5.99, "learning_rate": 2.051713602768679e-10, "loss": 0.1491, "step": 3533 }, { "epoch": 5.99, "learning_rate": 1.5073827899780847e-10, "loss": 0.1558, "step": 3534 }, { "epoch": 5.99, "learning_rate": 1.0467944077219295e-10, "loss": 0.1584, "step": 3535 }, { "epoch": 5.99, "learning_rate": 6.699488417249989e-11, "loss": 0.1522, "step": 3536 }, { "epoch": 5.99, "learning_rate": 3.7684640755708546e-11, "loss": 0.1532, "step": 3537 }, { "epoch": 5.99, "learning_rate": 1.6748735068849997e-11, "loss": 0.1537, "step": 3538 }, { "epoch": 6.0, "learning_rate": 4.187184643456021e-12, "loss": 0.1578, "step": 3539 }, { "epoch": 6.0, "learning_rate": 0.0, "loss": 0.1548, "step": 3540 }, { "epoch": 6.0, "step": 3540, "total_flos": 2.0635876934025216e+16, "train_loss": 0.04700535153601803, "train_runtime": 12718.1036, "train_samples_per_second": 142.579, "train_steps_per_second": 0.278 } ], "logging_steps": 1.0, "max_steps": 3540, "num_input_tokens_seen": 0, "num_train_epochs": 6, "save_steps": 500, "total_flos": 2.0635876934025216e+16, "train_batch_size": 32, "trial_name": null, "trial_params": null }