{ "best_metric": null, "best_model_checkpoint": null, "epoch": 2.0, "eval_steps": 500, "global_step": 2718, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 2.439024390243903e-07, "loss": 4.418, "step": 1 }, { "epoch": 0.0, "learning_rate": 4.878048780487805e-07, "loss": 3.9336, "step": 2 }, { "epoch": 0.0, "learning_rate": 7.317073170731707e-07, "loss": 4.1699, "step": 3 }, { "epoch": 0.0, "learning_rate": 9.75609756097561e-07, "loss": 4.1855, "step": 4 }, { "epoch": 0.0, "learning_rate": 1.2195121951219514e-06, "loss": 4.084, "step": 5 }, { "epoch": 0.0, "learning_rate": 1.4634146341463414e-06, "loss": 3.9473, "step": 6 }, { "epoch": 0.01, "learning_rate": 1.707317073170732e-06, "loss": 4.0938, "step": 7 }, { "epoch": 0.01, "learning_rate": 1.951219512195122e-06, "loss": 3.4141, "step": 8 }, { "epoch": 0.01, "learning_rate": 2.1951219512195125e-06, "loss": 3.1094, "step": 9 }, { "epoch": 0.01, "learning_rate": 2.4390243902439027e-06, "loss": 2.0244, "step": 10 }, { "epoch": 0.01, "learning_rate": 2.682926829268293e-06, "loss": 1.9512, "step": 11 }, { "epoch": 0.01, "learning_rate": 2.926829268292683e-06, "loss": 1.6738, "step": 12 }, { "epoch": 0.01, "learning_rate": 3.1707317073170736e-06, "loss": 0.8945, "step": 13 }, { "epoch": 0.01, "learning_rate": 3.414634146341464e-06, "loss": 0.8062, "step": 14 }, { "epoch": 0.01, "learning_rate": 3.6585365853658537e-06, "loss": 0.7017, "step": 15 }, { "epoch": 0.01, "learning_rate": 3.902439024390244e-06, "loss": 0.4771, "step": 16 }, { "epoch": 0.01, "learning_rate": 4.146341463414634e-06, "loss": 0.3428, "step": 17 }, { "epoch": 0.01, "learning_rate": 4.390243902439025e-06, "loss": 0.2932, "step": 18 }, { "epoch": 0.01, "learning_rate": 4.634146341463416e-06, "loss": 0.375, "step": 19 }, { "epoch": 0.01, "learning_rate": 4.8780487804878055e-06, "loss": 0.2786, "step": 20 }, { "epoch": 0.02, "learning_rate": 5.121951219512195e-06, "loss": 0.2468, "step": 21 }, { "epoch": 0.02, "learning_rate": 5.365853658536586e-06, "loss": 0.225, "step": 22 }, { "epoch": 0.02, "learning_rate": 5.609756097560977e-06, "loss": 0.178, "step": 23 }, { "epoch": 0.02, "learning_rate": 5.853658536585366e-06, "loss": 0.1757, "step": 24 }, { "epoch": 0.02, "learning_rate": 6.0975609756097564e-06, "loss": 0.1447, "step": 25 }, { "epoch": 0.02, "learning_rate": 6.341463414634147e-06, "loss": 0.1465, "step": 26 }, { "epoch": 0.02, "learning_rate": 6.585365853658538e-06, "loss": 0.1489, "step": 27 }, { "epoch": 0.02, "learning_rate": 6.829268292682928e-06, "loss": 0.1659, "step": 28 }, { "epoch": 0.02, "learning_rate": 7.0731707317073175e-06, "loss": 0.1593, "step": 29 }, { "epoch": 0.02, "learning_rate": 7.317073170731707e-06, "loss": 0.1621, "step": 30 }, { "epoch": 0.02, "learning_rate": 7.560975609756098e-06, "loss": 0.1436, "step": 31 }, { "epoch": 0.02, "learning_rate": 7.804878048780489e-06, "loss": 0.1472, "step": 32 }, { "epoch": 0.02, "learning_rate": 8.048780487804879e-06, "loss": 0.1439, "step": 33 }, { "epoch": 0.03, "learning_rate": 8.292682926829268e-06, "loss": 0.1504, "step": 34 }, { "epoch": 0.03, "learning_rate": 8.536585365853658e-06, "loss": 0.1512, "step": 35 }, { "epoch": 0.03, "learning_rate": 8.78048780487805e-06, "loss": 0.1457, "step": 36 }, { "epoch": 0.03, "learning_rate": 9.02439024390244e-06, "loss": 0.1489, "step": 37 }, { "epoch": 0.03, "learning_rate": 9.268292682926831e-06, "loss": 0.1409, "step": 38 }, { "epoch": 0.03, "learning_rate": 9.51219512195122e-06, "loss": 0.1296, "step": 39 }, { "epoch": 0.03, "learning_rate": 9.756097560975611e-06, "loss": 0.1448, "step": 40 }, { "epoch": 0.03, "learning_rate": 1e-05, "loss": 0.143, "step": 41 }, { "epoch": 0.03, "learning_rate": 1.024390243902439e-05, "loss": 0.157, "step": 42 }, { "epoch": 0.03, "learning_rate": 1.0487804878048782e-05, "loss": 0.1382, "step": 43 }, { "epoch": 0.03, "learning_rate": 1.0731707317073172e-05, "loss": 0.1373, "step": 44 }, { "epoch": 0.03, "learning_rate": 1.0975609756097562e-05, "loss": 0.1407, "step": 45 }, { "epoch": 0.03, "learning_rate": 1.1219512195121953e-05, "loss": 0.1357, "step": 46 }, { "epoch": 0.03, "learning_rate": 1.1463414634146342e-05, "loss": 0.1332, "step": 47 }, { "epoch": 0.04, "learning_rate": 1.1707317073170731e-05, "loss": 0.1589, "step": 48 }, { "epoch": 0.04, "learning_rate": 1.1951219512195123e-05, "loss": 0.1425, "step": 49 }, { "epoch": 0.04, "learning_rate": 1.2195121951219513e-05, "loss": 0.1485, "step": 50 }, { "epoch": 0.04, "learning_rate": 1.2439024390243903e-05, "loss": 0.1262, "step": 51 }, { "epoch": 0.04, "learning_rate": 1.2682926829268294e-05, "loss": 0.1381, "step": 52 }, { "epoch": 0.04, "learning_rate": 1.2926829268292684e-05, "loss": 0.1332, "step": 53 }, { "epoch": 0.04, "learning_rate": 1.3170731707317076e-05, "loss": 0.1405, "step": 54 }, { "epoch": 0.04, "learning_rate": 1.3414634146341466e-05, "loss": 0.1469, "step": 55 }, { "epoch": 0.04, "learning_rate": 1.3658536585365855e-05, "loss": 0.1408, "step": 56 }, { "epoch": 0.04, "learning_rate": 1.3902439024390244e-05, "loss": 0.1354, "step": 57 }, { "epoch": 0.04, "learning_rate": 1.4146341463414635e-05, "loss": 0.1341, "step": 58 }, { "epoch": 0.04, "learning_rate": 1.4390243902439025e-05, "loss": 0.1323, "step": 59 }, { "epoch": 0.04, "learning_rate": 1.4634146341463415e-05, "loss": 0.1221, "step": 60 }, { "epoch": 0.04, "learning_rate": 1.4878048780487806e-05, "loss": 0.1286, "step": 61 }, { "epoch": 0.05, "learning_rate": 1.5121951219512196e-05, "loss": 0.1395, "step": 62 }, { "epoch": 0.05, "learning_rate": 1.5365853658536586e-05, "loss": 0.1334, "step": 63 }, { "epoch": 0.05, "learning_rate": 1.5609756097560978e-05, "loss": 0.1291, "step": 64 }, { "epoch": 0.05, "learning_rate": 1.585365853658537e-05, "loss": 0.1255, "step": 65 }, { "epoch": 0.05, "learning_rate": 1.6097560975609757e-05, "loss": 0.1282, "step": 66 }, { "epoch": 0.05, "learning_rate": 1.6341463414634145e-05, "loss": 0.1286, "step": 67 }, { "epoch": 0.05, "learning_rate": 1.6585365853658537e-05, "loss": 0.129, "step": 68 }, { "epoch": 0.05, "learning_rate": 1.682926829268293e-05, "loss": 0.1266, "step": 69 }, { "epoch": 0.05, "learning_rate": 1.7073170731707317e-05, "loss": 0.1207, "step": 70 }, { "epoch": 0.05, "learning_rate": 1.7317073170731708e-05, "loss": 0.1245, "step": 71 }, { "epoch": 0.05, "learning_rate": 1.75609756097561e-05, "loss": 0.1265, "step": 72 }, { "epoch": 0.05, "learning_rate": 1.7804878048780488e-05, "loss": 0.1301, "step": 73 }, { "epoch": 0.05, "learning_rate": 1.804878048780488e-05, "loss": 0.1174, "step": 74 }, { "epoch": 0.06, "learning_rate": 1.829268292682927e-05, "loss": 0.1273, "step": 75 }, { "epoch": 0.06, "learning_rate": 1.8536585365853663e-05, "loss": 0.1262, "step": 76 }, { "epoch": 0.06, "learning_rate": 1.878048780487805e-05, "loss": 0.1232, "step": 77 }, { "epoch": 0.06, "learning_rate": 1.902439024390244e-05, "loss": 0.1293, "step": 78 }, { "epoch": 0.06, "learning_rate": 1.926829268292683e-05, "loss": 0.1317, "step": 79 }, { "epoch": 0.06, "learning_rate": 1.9512195121951222e-05, "loss": 0.1285, "step": 80 }, { "epoch": 0.06, "learning_rate": 1.975609756097561e-05, "loss": 0.122, "step": 81 }, { "epoch": 0.06, "learning_rate": 2e-05, "loss": 0.1143, "step": 82 }, { "epoch": 0.06, "learning_rate": 1.9999992898029135e-05, "loss": 0.1152, "step": 83 }, { "epoch": 0.06, "learning_rate": 1.999997159212662e-05, "loss": 0.1202, "step": 84 }, { "epoch": 0.06, "learning_rate": 1.9999936082322722e-05, "loss": 0.1323, "step": 85 }, { "epoch": 0.06, "learning_rate": 1.9999886368667875e-05, "loss": 0.1147, "step": 86 }, { "epoch": 0.06, "learning_rate": 1.9999822451232694e-05, "loss": 0.1266, "step": 87 }, { "epoch": 0.06, "learning_rate": 1.9999744330107972e-05, "loss": 0.1214, "step": 88 }, { "epoch": 0.07, "learning_rate": 1.9999652005404663e-05, "loss": 0.116, "step": 89 }, { "epoch": 0.07, "learning_rate": 1.999954547725391e-05, "loss": 0.1227, "step": 90 }, { "epoch": 0.07, "learning_rate": 1.9999424745807024e-05, "loss": 0.1211, "step": 91 }, { "epoch": 0.07, "learning_rate": 1.9999289811235492e-05, "loss": 0.1216, "step": 92 }, { "epoch": 0.07, "learning_rate": 1.999914067373097e-05, "loss": 0.1044, "step": 93 }, { "epoch": 0.07, "learning_rate": 1.9998977333505298e-05, "loss": 0.1061, "step": 94 }, { "epoch": 0.07, "learning_rate": 1.999879979079048e-05, "loss": 0.1187, "step": 95 }, { "epoch": 0.07, "learning_rate": 1.9998608045838696e-05, "loss": 0.1119, "step": 96 }, { "epoch": 0.07, "learning_rate": 1.99984020989223e-05, "loss": 0.1187, "step": 97 }, { "epoch": 0.07, "learning_rate": 1.9998181950333825e-05, "loss": 0.1089, "step": 98 }, { "epoch": 0.07, "learning_rate": 1.9997947600385958e-05, "loss": 0.1197, "step": 99 }, { "epoch": 0.07, "learning_rate": 1.999769904941157e-05, "loss": 0.1098, "step": 100 }, { "epoch": 0.07, "learning_rate": 1.9997436297763707e-05, "loss": 0.0942, "step": 101 }, { "epoch": 0.08, "learning_rate": 1.9997159345815577e-05, "loss": 0.104, "step": 102 }, { "epoch": 0.08, "learning_rate": 1.9996868193960557e-05, "loss": 0.109, "step": 103 }, { "epoch": 0.08, "learning_rate": 1.9996562842612208e-05, "loss": 0.1096, "step": 104 }, { "epoch": 0.08, "learning_rate": 1.9996243292204237e-05, "loss": 0.1089, "step": 105 }, { "epoch": 0.08, "learning_rate": 1.999590954319054e-05, "loss": 0.0966, "step": 106 }, { "epoch": 0.08, "learning_rate": 1.9995561596045168e-05, "loss": 0.0996, "step": 107 }, { "epoch": 0.08, "learning_rate": 1.9995199451262348e-05, "loss": 0.0948, "step": 108 }, { "epoch": 0.08, "learning_rate": 1.999482310935646e-05, "loss": 0.0928, "step": 109 }, { "epoch": 0.08, "learning_rate": 1.999443257086206e-05, "loss": 0.1011, "step": 110 }, { "epoch": 0.08, "learning_rate": 1.9994027836333873e-05, "loss": 0.1107, "step": 111 }, { "epoch": 0.08, "learning_rate": 1.9993608906346778e-05, "loss": 0.1007, "step": 112 }, { "epoch": 0.08, "learning_rate": 1.9993175781495818e-05, "loss": 0.076, "step": 113 }, { "epoch": 0.08, "learning_rate": 1.9992728462396207e-05, "loss": 0.1108, "step": 114 }, { "epoch": 0.08, "learning_rate": 1.9992266949683307e-05, "loss": 0.1036, "step": 115 }, { "epoch": 0.09, "learning_rate": 1.9991791244012654e-05, "loss": 0.095, "step": 116 }, { "epoch": 0.09, "learning_rate": 1.9991301346059938e-05, "loss": 0.0898, "step": 117 }, { "epoch": 0.09, "learning_rate": 1.9990797256521e-05, "loss": 0.0995, "step": 118 }, { "epoch": 0.09, "learning_rate": 1.999027897611185e-05, "loss": 0.1006, "step": 119 }, { "epoch": 0.09, "learning_rate": 1.9989746505568655e-05, "loss": 0.0874, "step": 120 }, { "epoch": 0.09, "learning_rate": 1.9989199845647725e-05, "loss": 0.0883, "step": 121 }, { "epoch": 0.09, "learning_rate": 1.998863899712554e-05, "loss": 0.0931, "step": 122 }, { "epoch": 0.09, "learning_rate": 1.998806396079872e-05, "loss": 0.081, "step": 123 }, { "epoch": 0.09, "learning_rate": 1.998747473748405e-05, "loss": 0.0687, "step": 124 }, { "epoch": 0.09, "learning_rate": 1.998687132801845e-05, "loss": 0.0798, "step": 125 }, { "epoch": 0.09, "learning_rate": 1.9986253733259004e-05, "loss": 0.0882, "step": 126 }, { "epoch": 0.09, "learning_rate": 1.9985621954082944e-05, "loss": 0.0774, "step": 127 }, { "epoch": 0.09, "learning_rate": 1.998497599138764e-05, "loss": 0.0853, "step": 128 }, { "epoch": 0.09, "learning_rate": 1.9984315846090615e-05, "loss": 0.0824, "step": 129 }, { "epoch": 0.1, "learning_rate": 1.9983641519129534e-05, "loss": 0.0806, "step": 130 }, { "epoch": 0.1, "learning_rate": 1.9982953011462216e-05, "loss": 0.0702, "step": 131 }, { "epoch": 0.1, "learning_rate": 1.99822503240666e-05, "loss": 0.0708, "step": 132 }, { "epoch": 0.1, "learning_rate": 1.9981533457940786e-05, "loss": 0.0818, "step": 133 }, { "epoch": 0.1, "learning_rate": 1.998080241410301e-05, "loss": 0.0916, "step": 134 }, { "epoch": 0.1, "learning_rate": 1.9980057193591634e-05, "loss": 0.0735, "step": 135 }, { "epoch": 0.1, "learning_rate": 1.997929779746517e-05, "loss": 0.0854, "step": 136 }, { "epoch": 0.1, "learning_rate": 1.997852422680226e-05, "loss": 0.0822, "step": 137 }, { "epoch": 0.1, "learning_rate": 1.997773648270168e-05, "loss": 0.0835, "step": 138 }, { "epoch": 0.1, "learning_rate": 1.9976934566282337e-05, "loss": 0.0802, "step": 139 }, { "epoch": 0.1, "learning_rate": 1.997611847868326e-05, "loss": 0.079, "step": 140 }, { "epoch": 0.1, "learning_rate": 1.9975288221063628e-05, "loss": 0.0853, "step": 141 }, { "epoch": 0.1, "learning_rate": 1.9974443794602723e-05, "loss": 0.0786, "step": 142 }, { "epoch": 0.11, "learning_rate": 1.9973585200499973e-05, "loss": 0.0768, "step": 143 }, { "epoch": 0.11, "learning_rate": 1.9972712439974912e-05, "loss": 0.0746, "step": 144 }, { "epoch": 0.11, "learning_rate": 1.997182551426721e-05, "loss": 0.079, "step": 145 }, { "epoch": 0.11, "learning_rate": 1.9970924424636645e-05, "loss": 0.0755, "step": 146 }, { "epoch": 0.11, "learning_rate": 1.997000917236313e-05, "loss": 0.0731, "step": 147 }, { "epoch": 0.11, "learning_rate": 1.996907975874667e-05, "loss": 0.0796, "step": 148 }, { "epoch": 0.11, "learning_rate": 1.9968136185107407e-05, "loss": 0.0808, "step": 149 }, { "epoch": 0.11, "learning_rate": 1.9967178452785586e-05, "loss": 0.0644, "step": 150 }, { "epoch": 0.11, "learning_rate": 1.996620656314157e-05, "loss": 0.0766, "step": 151 }, { "epoch": 0.11, "learning_rate": 1.9965220517555814e-05, "loss": 0.0764, "step": 152 }, { "epoch": 0.11, "learning_rate": 1.99642203174289e-05, "loss": 0.0774, "step": 153 }, { "epoch": 0.11, "learning_rate": 1.9963205964181503e-05, "loss": 0.0683, "step": 154 }, { "epoch": 0.11, "learning_rate": 1.9962177459254404e-05, "loss": 0.0728, "step": 155 }, { "epoch": 0.11, "learning_rate": 1.996113480410849e-05, "loss": 0.0588, "step": 156 }, { "epoch": 0.12, "learning_rate": 1.9960078000224737e-05, "loss": 0.0809, "step": 157 }, { "epoch": 0.12, "learning_rate": 1.9959007049104223e-05, "loss": 0.0728, "step": 158 }, { "epoch": 0.12, "learning_rate": 1.9957921952268126e-05, "loss": 0.0862, "step": 159 }, { "epoch": 0.12, "learning_rate": 1.9956822711257708e-05, "loss": 0.0833, "step": 160 }, { "epoch": 0.12, "learning_rate": 1.9955709327634323e-05, "loss": 0.0803, "step": 161 }, { "epoch": 0.12, "learning_rate": 1.995458180297942e-05, "loss": 0.0824, "step": 162 }, { "epoch": 0.12, "learning_rate": 1.995344013889452e-05, "loss": 0.077, "step": 163 }, { "epoch": 0.12, "learning_rate": 1.9952284337001238e-05, "loss": 0.0683, "step": 164 }, { "epoch": 0.12, "learning_rate": 1.9951114398941277e-05, "loss": 0.0881, "step": 165 }, { "epoch": 0.12, "learning_rate": 1.9949930326376403e-05, "loss": 0.0814, "step": 166 }, { "epoch": 0.12, "learning_rate": 1.9948732120988467e-05, "loss": 0.0907, "step": 167 }, { "epoch": 0.12, "learning_rate": 1.994751978447939e-05, "loss": 0.0738, "step": 168 }, { "epoch": 0.12, "learning_rate": 1.9946293318571172e-05, "loss": 0.0726, "step": 169 }, { "epoch": 0.13, "learning_rate": 1.994505272500588e-05, "loss": 0.0652, "step": 170 }, { "epoch": 0.13, "learning_rate": 1.9943798005545635e-05, "loss": 0.0768, "step": 171 }, { "epoch": 0.13, "learning_rate": 1.9942529161972646e-05, "loss": 0.0832, "step": 172 }, { "epoch": 0.13, "learning_rate": 1.9941246196089165e-05, "loss": 0.0574, "step": 173 }, { "epoch": 0.13, "learning_rate": 1.993994910971751e-05, "loss": 0.0965, "step": 174 }, { "epoch": 0.13, "learning_rate": 1.9938637904700052e-05, "loss": 0.0815, "step": 175 }, { "epoch": 0.13, "learning_rate": 1.9937312582899224e-05, "loss": 0.0763, "step": 176 }, { "epoch": 0.13, "learning_rate": 1.9935973146197504e-05, "loss": 0.0836, "step": 177 }, { "epoch": 0.13, "learning_rate": 1.993461959649742e-05, "loss": 0.089, "step": 178 }, { "epoch": 0.13, "learning_rate": 1.993325193572154e-05, "loss": 0.0636, "step": 179 }, { "epoch": 0.13, "learning_rate": 1.9931870165812492e-05, "loss": 0.0725, "step": 180 }, { "epoch": 0.13, "learning_rate": 1.993047428873293e-05, "loss": 0.0638, "step": 181 }, { "epoch": 0.13, "learning_rate": 1.9929064306465543e-05, "loss": 0.083, "step": 182 }, { "epoch": 0.13, "learning_rate": 1.992764022101307e-05, "loss": 0.063, "step": 183 }, { "epoch": 0.14, "learning_rate": 1.992620203439827e-05, "loss": 0.0677, "step": 184 }, { "epoch": 0.14, "learning_rate": 1.9924749748663938e-05, "loss": 0.0675, "step": 185 }, { "epoch": 0.14, "learning_rate": 1.9923283365872886e-05, "loss": 0.0559, "step": 186 }, { "epoch": 0.14, "learning_rate": 1.9921802888107962e-05, "loss": 0.06, "step": 187 }, { "epoch": 0.14, "learning_rate": 1.9920308317472023e-05, "loss": 0.0948, "step": 188 }, { "epoch": 0.14, "learning_rate": 1.9918799656087954e-05, "loss": 0.0617, "step": 189 }, { "epoch": 0.14, "learning_rate": 1.9917276906098643e-05, "loss": 0.0643, "step": 190 }, { "epoch": 0.14, "learning_rate": 1.9915740069667e-05, "loss": 0.0793, "step": 191 }, { "epoch": 0.14, "learning_rate": 1.991418914897593e-05, "loss": 0.0519, "step": 192 }, { "epoch": 0.14, "learning_rate": 1.991262414622836e-05, "loss": 0.078, "step": 193 }, { "epoch": 0.14, "learning_rate": 1.9911045063647214e-05, "loss": 0.0861, "step": 194 }, { "epoch": 0.14, "learning_rate": 1.9909451903475407e-05, "loss": 0.0591, "step": 195 }, { "epoch": 0.14, "learning_rate": 1.9907844667975847e-05, "loss": 0.0691, "step": 196 }, { "epoch": 0.14, "learning_rate": 1.9906223359431454e-05, "loss": 0.0667, "step": 197 }, { "epoch": 0.15, "learning_rate": 1.9904587980145117e-05, "loss": 0.0536, "step": 198 }, { "epoch": 0.15, "learning_rate": 1.9902938532439723e-05, "loss": 0.0795, "step": 199 }, { "epoch": 0.15, "learning_rate": 1.990127501865814e-05, "loss": 0.0535, "step": 200 }, { "epoch": 0.15, "learning_rate": 1.989959744116321e-05, "loss": 0.0597, "step": 201 }, { "epoch": 0.15, "learning_rate": 1.989790580233775e-05, "loss": 0.0536, "step": 202 }, { "epoch": 0.15, "learning_rate": 1.9896200104584565e-05, "loss": 0.0681, "step": 203 }, { "epoch": 0.15, "learning_rate": 1.989448035032641e-05, "loss": 0.0599, "step": 204 }, { "epoch": 0.15, "learning_rate": 1.989274654200601e-05, "loss": 0.0619, "step": 205 }, { "epoch": 0.15, "learning_rate": 1.989099868208607e-05, "loss": 0.066, "step": 206 }, { "epoch": 0.15, "learning_rate": 1.9889236773049228e-05, "loss": 0.0476, "step": 207 }, { "epoch": 0.15, "learning_rate": 1.9887460817398093e-05, "loss": 0.0475, "step": 208 }, { "epoch": 0.15, "learning_rate": 1.9885670817655222e-05, "loss": 0.0684, "step": 209 }, { "epoch": 0.15, "learning_rate": 1.9883866776363123e-05, "loss": 0.066, "step": 210 }, { "epoch": 0.16, "learning_rate": 1.988204869608424e-05, "loss": 0.0885, "step": 211 }, { "epoch": 0.16, "learning_rate": 1.9880216579400972e-05, "loss": 0.0767, "step": 212 }, { "epoch": 0.16, "learning_rate": 1.9878370428915642e-05, "loss": 0.0665, "step": 213 }, { "epoch": 0.16, "learning_rate": 1.9876510247250506e-05, "loss": 0.0609, "step": 214 }, { "epoch": 0.16, "learning_rate": 1.9874636037047764e-05, "loss": 0.0639, "step": 215 }, { "epoch": 0.16, "learning_rate": 1.9872747800969526e-05, "loss": 0.0707, "step": 216 }, { "epoch": 0.16, "learning_rate": 1.987084554169784e-05, "loss": 0.0659, "step": 217 }, { "epoch": 0.16, "learning_rate": 1.9868929261934657e-05, "loss": 0.0587, "step": 218 }, { "epoch": 0.16, "learning_rate": 1.986699896440185e-05, "loss": 0.0529, "step": 219 }, { "epoch": 0.16, "learning_rate": 1.986505465184121e-05, "loss": 0.0569, "step": 220 }, { "epoch": 0.16, "learning_rate": 1.9863096327014415e-05, "loss": 0.0751, "step": 221 }, { "epoch": 0.16, "learning_rate": 1.986112399270307e-05, "loss": 0.0749, "step": 222 }, { "epoch": 0.16, "learning_rate": 1.985913765170866e-05, "loss": 0.0617, "step": 223 }, { "epoch": 0.16, "learning_rate": 1.985713730685257e-05, "loss": 0.0771, "step": 224 }, { "epoch": 0.17, "learning_rate": 1.9855122960976086e-05, "loss": 0.0483, "step": 225 }, { "epoch": 0.17, "learning_rate": 1.985309461694037e-05, "loss": 0.0685, "step": 226 }, { "epoch": 0.17, "learning_rate": 1.9851052277626466e-05, "loss": 0.063, "step": 227 }, { "epoch": 0.17, "learning_rate": 1.9848995945935305e-05, "loss": 0.0568, "step": 228 }, { "epoch": 0.17, "learning_rate": 1.984692562478769e-05, "loss": 0.0588, "step": 229 }, { "epoch": 0.17, "learning_rate": 1.984484131712429e-05, "loss": 0.06, "step": 230 }, { "epoch": 0.17, "learning_rate": 1.9842743025905645e-05, "loss": 0.0568, "step": 231 }, { "epoch": 0.17, "learning_rate": 1.9840630754112152e-05, "loss": 0.0669, "step": 232 }, { "epoch": 0.17, "learning_rate": 1.9838504504744075e-05, "loss": 0.0642, "step": 233 }, { "epoch": 0.17, "learning_rate": 1.9836364280821522e-05, "loss": 0.053, "step": 234 }, { "epoch": 0.17, "learning_rate": 1.983421008538446e-05, "loss": 0.0656, "step": 235 }, { "epoch": 0.17, "learning_rate": 1.9832041921492688e-05, "loss": 0.0702, "step": 236 }, { "epoch": 0.17, "learning_rate": 1.982985979222586e-05, "loss": 0.0563, "step": 237 }, { "epoch": 0.18, "learning_rate": 1.9827663700683454e-05, "loss": 0.0603, "step": 238 }, { "epoch": 0.18, "learning_rate": 1.9825453649984793e-05, "loss": 0.0729, "step": 239 }, { "epoch": 0.18, "learning_rate": 1.982322964326901e-05, "loss": 0.0701, "step": 240 }, { "epoch": 0.18, "learning_rate": 1.9820991683695082e-05, "loss": 0.0701, "step": 241 }, { "epoch": 0.18, "learning_rate": 1.9818739774441784e-05, "loss": 0.0569, "step": 242 }, { "epoch": 0.18, "learning_rate": 1.9816473918707722e-05, "loss": 0.0567, "step": 243 }, { "epoch": 0.18, "learning_rate": 1.98141941197113e-05, "loss": 0.0776, "step": 244 }, { "epoch": 0.18, "learning_rate": 1.9811900380690738e-05, "loss": 0.0646, "step": 245 }, { "epoch": 0.18, "learning_rate": 1.980959270490404e-05, "loss": 0.0678, "step": 246 }, { "epoch": 0.18, "learning_rate": 1.980727109562902e-05, "loss": 0.0689, "step": 247 }, { "epoch": 0.18, "learning_rate": 1.980493555616328e-05, "loss": 0.0582, "step": 248 }, { "epoch": 0.18, "learning_rate": 1.9802586089824203e-05, "loss": 0.0741, "step": 249 }, { "epoch": 0.18, "learning_rate": 1.980022269994896e-05, "loss": 0.0528, "step": 250 }, { "epoch": 0.18, "learning_rate": 1.9797845389894492e-05, "loss": 0.0608, "step": 251 }, { "epoch": 0.19, "learning_rate": 1.9795454163037523e-05, "loss": 0.0636, "step": 252 }, { "epoch": 0.19, "learning_rate": 1.9793049022774533e-05, "loss": 0.0656, "step": 253 }, { "epoch": 0.19, "learning_rate": 1.9790629972521772e-05, "loss": 0.0586, "step": 254 }, { "epoch": 0.19, "learning_rate": 1.9788197015715243e-05, "loss": 0.0607, "step": 255 }, { "epoch": 0.19, "learning_rate": 1.97857501558107e-05, "loss": 0.0703, "step": 256 }, { "epoch": 0.19, "learning_rate": 1.9783289396283657e-05, "loss": 0.0611, "step": 257 }, { "epoch": 0.19, "learning_rate": 1.9780814740629357e-05, "loss": 0.0473, "step": 258 }, { "epoch": 0.19, "learning_rate": 1.9778326192362786e-05, "loss": 0.0604, "step": 259 }, { "epoch": 0.19, "learning_rate": 1.9775823755018665e-05, "loss": 0.0633, "step": 260 }, { "epoch": 0.19, "learning_rate": 1.9773307432151444e-05, "loss": 0.0744, "step": 261 }, { "epoch": 0.19, "learning_rate": 1.9770777227335292e-05, "loss": 0.0652, "step": 262 }, { "epoch": 0.19, "learning_rate": 1.976823314416409e-05, "loss": 0.0561, "step": 263 }, { "epoch": 0.19, "learning_rate": 1.976567518625145e-05, "loss": 0.062, "step": 264 }, { "epoch": 0.19, "learning_rate": 1.9763103357230672e-05, "loss": 0.0522, "step": 265 }, { "epoch": 0.2, "learning_rate": 1.976051766075477e-05, "loss": 0.0641, "step": 266 }, { "epoch": 0.2, "learning_rate": 1.9757918100496456e-05, "loss": 0.042, "step": 267 }, { "epoch": 0.2, "learning_rate": 1.9755304680148125e-05, "loss": 0.0623, "step": 268 }, { "epoch": 0.2, "learning_rate": 1.9752677403421866e-05, "loss": 0.0696, "step": 269 }, { "epoch": 0.2, "learning_rate": 1.9750036274049447e-05, "loss": 0.0588, "step": 270 }, { "epoch": 0.2, "learning_rate": 1.9747381295782314e-05, "loss": 0.0652, "step": 271 }, { "epoch": 0.2, "learning_rate": 1.974471247239158e-05, "loss": 0.0643, "step": 272 }, { "epoch": 0.2, "learning_rate": 1.9742029807668027e-05, "loss": 0.0522, "step": 273 }, { "epoch": 0.2, "learning_rate": 1.97393333054221e-05, "loss": 0.054, "step": 274 }, { "epoch": 0.2, "learning_rate": 1.9736622969483893e-05, "loss": 0.0537, "step": 275 }, { "epoch": 0.2, "learning_rate": 1.9733898803703145e-05, "loss": 0.0648, "step": 276 }, { "epoch": 0.2, "learning_rate": 1.9731160811949258e-05, "loss": 0.0676, "step": 277 }, { "epoch": 0.2, "learning_rate": 1.972840899811125e-05, "loss": 0.0441, "step": 278 }, { "epoch": 0.21, "learning_rate": 1.972564336609778e-05, "loss": 0.0556, "step": 279 }, { "epoch": 0.21, "learning_rate": 1.9722863919837146e-05, "loss": 0.0549, "step": 280 }, { "epoch": 0.21, "learning_rate": 1.9720070663277246e-05, "loss": 0.0597, "step": 281 }, { "epoch": 0.21, "learning_rate": 1.9717263600385614e-05, "loss": 0.0655, "step": 282 }, { "epoch": 0.21, "learning_rate": 1.971444273514938e-05, "loss": 0.0677, "step": 283 }, { "epoch": 0.21, "learning_rate": 1.9711608071575285e-05, "loss": 0.057, "step": 284 }, { "epoch": 0.21, "learning_rate": 1.9708759613689666e-05, "loss": 0.0524, "step": 285 }, { "epoch": 0.21, "learning_rate": 1.970589736553847e-05, "loss": 0.0614, "step": 286 }, { "epoch": 0.21, "learning_rate": 1.97030213311872e-05, "loss": 0.0522, "step": 287 }, { "epoch": 0.21, "learning_rate": 1.970013151472097e-05, "loss": 0.0597, "step": 288 }, { "epoch": 0.21, "learning_rate": 1.9697227920244453e-05, "loss": 0.0533, "step": 289 }, { "epoch": 0.21, "learning_rate": 1.96943105518819e-05, "loss": 0.0535, "step": 290 }, { "epoch": 0.21, "learning_rate": 1.9691379413777118e-05, "loss": 0.0611, "step": 291 }, { "epoch": 0.21, "learning_rate": 1.968843451009349e-05, "loss": 0.0574, "step": 292 }, { "epoch": 0.22, "learning_rate": 1.9685475845013934e-05, "loss": 0.0634, "step": 293 }, { "epoch": 0.22, "learning_rate": 1.9682503422740915e-05, "loss": 0.0533, "step": 294 }, { "epoch": 0.22, "learning_rate": 1.967951724749645e-05, "loss": 0.0629, "step": 295 }, { "epoch": 0.22, "learning_rate": 1.967651732352209e-05, "loss": 0.0552, "step": 296 }, { "epoch": 0.22, "learning_rate": 1.9673503655078902e-05, "loss": 0.0693, "step": 297 }, { "epoch": 0.22, "learning_rate": 1.9670476246447484e-05, "loss": 0.0654, "step": 298 }, { "epoch": 0.22, "learning_rate": 1.966743510192795e-05, "loss": 0.051, "step": 299 }, { "epoch": 0.22, "learning_rate": 1.966438022583993e-05, "loss": 0.0552, "step": 300 }, { "epoch": 0.22, "learning_rate": 1.966131162252254e-05, "loss": 0.0546, "step": 301 }, { "epoch": 0.22, "learning_rate": 1.9658229296334416e-05, "loss": 0.0599, "step": 302 }, { "epoch": 0.22, "learning_rate": 1.9655133251653673e-05, "loss": 0.0586, "step": 303 }, { "epoch": 0.22, "learning_rate": 1.9652023492877915e-05, "loss": 0.0548, "step": 304 }, { "epoch": 0.22, "learning_rate": 1.9648900024424223e-05, "loss": 0.065, "step": 305 }, { "epoch": 0.23, "learning_rate": 1.964576285072916e-05, "loss": 0.0495, "step": 306 }, { "epoch": 0.23, "learning_rate": 1.9642611976248746e-05, "loss": 0.0553, "step": 307 }, { "epoch": 0.23, "learning_rate": 1.963944740545846e-05, "loss": 0.0561, "step": 308 }, { "epoch": 0.23, "learning_rate": 1.9636269142853247e-05, "loss": 0.0598, "step": 309 }, { "epoch": 0.23, "learning_rate": 1.9633077192947486e-05, "loss": 0.0625, "step": 310 }, { "epoch": 0.23, "learning_rate": 1.962987156027501e-05, "loss": 0.071, "step": 311 }, { "epoch": 0.23, "learning_rate": 1.9626652249389076e-05, "loss": 0.0474, "step": 312 }, { "epoch": 0.23, "learning_rate": 1.9623419264862376e-05, "loss": 0.0487, "step": 313 }, { "epoch": 0.23, "learning_rate": 1.9620172611287028e-05, "loss": 0.0531, "step": 314 }, { "epoch": 0.23, "learning_rate": 1.961691229327455e-05, "loss": 0.0573, "step": 315 }, { "epoch": 0.23, "learning_rate": 1.9613638315455888e-05, "loss": 0.0609, "step": 316 }, { "epoch": 0.23, "learning_rate": 1.9610350682481376e-05, "loss": 0.0577, "step": 317 }, { "epoch": 0.23, "learning_rate": 1.9607049399020746e-05, "loss": 0.0539, "step": 318 }, { "epoch": 0.23, "learning_rate": 1.9603734469763127e-05, "loss": 0.0627, "step": 319 }, { "epoch": 0.24, "learning_rate": 1.9600405899417026e-05, "loss": 0.0565, "step": 320 }, { "epoch": 0.24, "learning_rate": 1.959706369271032e-05, "loss": 0.0558, "step": 321 }, { "epoch": 0.24, "learning_rate": 1.9593707854390263e-05, "loss": 0.0796, "step": 322 }, { "epoch": 0.24, "learning_rate": 1.9590338389223466e-05, "loss": 0.0663, "step": 323 }, { "epoch": 0.24, "learning_rate": 1.9586955301995903e-05, "loss": 0.0604, "step": 324 }, { "epoch": 0.24, "learning_rate": 1.9583558597512882e-05, "loss": 0.0572, "step": 325 }, { "epoch": 0.24, "learning_rate": 1.9580148280599075e-05, "loss": 0.0493, "step": 326 }, { "epoch": 0.24, "learning_rate": 1.9576724356098466e-05, "loss": 0.0692, "step": 327 }, { "epoch": 0.24, "learning_rate": 1.957328682887438e-05, "loss": 0.0569, "step": 328 }, { "epoch": 0.24, "learning_rate": 1.956983570380946e-05, "loss": 0.0647, "step": 329 }, { "epoch": 0.24, "learning_rate": 1.9566370985805666e-05, "loss": 0.0627, "step": 330 }, { "epoch": 0.24, "learning_rate": 1.9562892679784262e-05, "loss": 0.0528, "step": 331 }, { "epoch": 0.24, "learning_rate": 1.9559400790685813e-05, "loss": 0.0545, "step": 332 }, { "epoch": 0.25, "learning_rate": 1.955589532347018e-05, "loss": 0.0637, "step": 333 }, { "epoch": 0.25, "learning_rate": 1.9552376283116508e-05, "loss": 0.0571, "step": 334 }, { "epoch": 0.25, "learning_rate": 1.9548843674623222e-05, "loss": 0.0534, "step": 335 }, { "epoch": 0.25, "learning_rate": 1.9545297503008014e-05, "loss": 0.0559, "step": 336 }, { "epoch": 0.25, "learning_rate": 1.954173777330785e-05, "loss": 0.0557, "step": 337 }, { "epoch": 0.25, "learning_rate": 1.953816449057895e-05, "loss": 0.0483, "step": 338 }, { "epoch": 0.25, "learning_rate": 1.9534577659896777e-05, "loss": 0.057, "step": 339 }, { "epoch": 0.25, "learning_rate": 1.9530977286356053e-05, "loss": 0.0482, "step": 340 }, { "epoch": 0.25, "learning_rate": 1.9527363375070723e-05, "loss": 0.0626, "step": 341 }, { "epoch": 0.25, "learning_rate": 1.9523735931173964e-05, "loss": 0.0574, "step": 342 }, { "epoch": 0.25, "learning_rate": 1.9520094959818183e-05, "loss": 0.0723, "step": 343 }, { "epoch": 0.25, "learning_rate": 1.951644046617499e-05, "loss": 0.0511, "step": 344 }, { "epoch": 0.25, "learning_rate": 1.95127724554352e-05, "loss": 0.0487, "step": 345 }, { "epoch": 0.25, "learning_rate": 1.950909093280885e-05, "loss": 0.0567, "step": 346 }, { "epoch": 0.26, "learning_rate": 1.9505395903525136e-05, "loss": 0.0594, "step": 347 }, { "epoch": 0.26, "learning_rate": 1.9501687372832466e-05, "loss": 0.0632, "step": 348 }, { "epoch": 0.26, "learning_rate": 1.9497965345998417e-05, "loss": 0.0511, "step": 349 }, { "epoch": 0.26, "learning_rate": 1.9494229828309724e-05, "loss": 0.0717, "step": 350 }, { "epoch": 0.26, "learning_rate": 1.9490480825072308e-05, "loss": 0.0623, "step": 351 }, { "epoch": 0.26, "learning_rate": 1.948671834161122e-05, "loss": 0.0525, "step": 352 }, { "epoch": 0.26, "learning_rate": 1.9482942383270677e-05, "loss": 0.0445, "step": 353 }, { "epoch": 0.26, "learning_rate": 1.947915295541402e-05, "loss": 0.0747, "step": 354 }, { "epoch": 0.26, "learning_rate": 1.9475350063423743e-05, "loss": 0.0565, "step": 355 }, { "epoch": 0.26, "learning_rate": 1.947153371270144e-05, "loss": 0.048, "step": 356 }, { "epoch": 0.26, "learning_rate": 1.946770390866784e-05, "loss": 0.0418, "step": 357 }, { "epoch": 0.26, "learning_rate": 1.946386065676277e-05, "loss": 0.0649, "step": 358 }, { "epoch": 0.26, "learning_rate": 1.9460003962445166e-05, "loss": 0.0723, "step": 359 }, { "epoch": 0.26, "learning_rate": 1.945613383119305e-05, "loss": 0.08, "step": 360 }, { "epoch": 0.27, "learning_rate": 1.945225026850354e-05, "loss": 0.0514, "step": 361 }, { "epoch": 0.27, "learning_rate": 1.944835327989282e-05, "loss": 0.0607, "step": 362 }, { "epoch": 0.27, "learning_rate": 1.944444287089615e-05, "loss": 0.0519, "step": 363 }, { "epoch": 0.27, "learning_rate": 1.944051904706786e-05, "loss": 0.0687, "step": 364 }, { "epoch": 0.27, "learning_rate": 1.9436581813981316e-05, "loss": 0.0755, "step": 365 }, { "epoch": 0.27, "learning_rate": 1.9432631177228948e-05, "loss": 0.0535, "step": 366 }, { "epoch": 0.27, "learning_rate": 1.9428667142422216e-05, "loss": 0.0583, "step": 367 }, { "epoch": 0.27, "learning_rate": 1.942468971519161e-05, "loss": 0.063, "step": 368 }, { "epoch": 0.27, "learning_rate": 1.9420698901186643e-05, "loss": 0.0614, "step": 369 }, { "epoch": 0.27, "learning_rate": 1.941669470607585e-05, "loss": 0.0596, "step": 370 }, { "epoch": 0.27, "learning_rate": 1.941267713554676e-05, "loss": 0.0676, "step": 371 }, { "epoch": 0.27, "learning_rate": 1.9408646195305914e-05, "loss": 0.0526, "step": 372 }, { "epoch": 0.27, "learning_rate": 1.940460189107883e-05, "loss": 0.0641, "step": 373 }, { "epoch": 0.28, "learning_rate": 1.940054422861002e-05, "loss": 0.0553, "step": 374 }, { "epoch": 0.28, "learning_rate": 1.9396473213662956e-05, "loss": 0.07, "step": 375 }, { "epoch": 0.28, "learning_rate": 1.939238885202009e-05, "loss": 0.0605, "step": 376 }, { "epoch": 0.28, "learning_rate": 1.938829114948282e-05, "loss": 0.0633, "step": 377 }, { "epoch": 0.28, "learning_rate": 1.9384180111871502e-05, "loss": 0.0565, "step": 378 }, { "epoch": 0.28, "learning_rate": 1.938005574502543e-05, "loss": 0.0618, "step": 379 }, { "epoch": 0.28, "learning_rate": 1.9375918054802836e-05, "loss": 0.0583, "step": 380 }, { "epoch": 0.28, "learning_rate": 1.9371767047080862e-05, "loss": 0.0473, "step": 381 }, { "epoch": 0.28, "learning_rate": 1.936760272775558e-05, "loss": 0.0529, "step": 382 }, { "epoch": 0.28, "learning_rate": 1.9363425102741966e-05, "loss": 0.0607, "step": 383 }, { "epoch": 0.28, "learning_rate": 1.935923417797389e-05, "loss": 0.0612, "step": 384 }, { "epoch": 0.28, "learning_rate": 1.935502995940412e-05, "loss": 0.0615, "step": 385 }, { "epoch": 0.28, "learning_rate": 1.9350812453004303e-05, "loss": 0.0739, "step": 386 }, { "epoch": 0.28, "learning_rate": 1.934658166476496e-05, "loss": 0.0546, "step": 387 }, { "epoch": 0.29, "learning_rate": 1.934233760069548e-05, "loss": 0.0735, "step": 388 }, { "epoch": 0.29, "learning_rate": 1.9338080266824102e-05, "loss": 0.0662, "step": 389 }, { "epoch": 0.29, "learning_rate": 1.933380966919792e-05, "loss": 0.0772, "step": 390 }, { "epoch": 0.29, "learning_rate": 1.9329525813882873e-05, "loss": 0.0495, "step": 391 }, { "epoch": 0.29, "learning_rate": 1.9325228706963716e-05, "loss": 0.0612, "step": 392 }, { "epoch": 0.29, "learning_rate": 1.9320918354544037e-05, "loss": 0.05, "step": 393 }, { "epoch": 0.29, "learning_rate": 1.9316594762746238e-05, "loss": 0.0576, "step": 394 }, { "epoch": 0.29, "learning_rate": 1.9312257937711518e-05, "loss": 0.0724, "step": 395 }, { "epoch": 0.29, "learning_rate": 1.9307907885599883e-05, "loss": 0.058, "step": 396 }, { "epoch": 0.29, "learning_rate": 1.930354461259012e-05, "loss": 0.049, "step": 397 }, { "epoch": 0.29, "learning_rate": 1.9299168124879798e-05, "loss": 0.0499, "step": 398 }, { "epoch": 0.29, "learning_rate": 1.9294778428685252e-05, "loss": 0.0489, "step": 399 }, { "epoch": 0.29, "learning_rate": 1.9290375530241577e-05, "loss": 0.0499, "step": 400 }, { "epoch": 0.3, "learning_rate": 1.9285959435802638e-05, "loss": 0.0646, "step": 401 }, { "epoch": 0.3, "learning_rate": 1.9281530151641016e-05, "loss": 0.0575, "step": 402 }, { "epoch": 0.3, "learning_rate": 1.9277087684048047e-05, "loss": 0.0568, "step": 403 }, { "epoch": 0.3, "learning_rate": 1.9272632039333784e-05, "loss": 0.0652, "step": 404 }, { "epoch": 0.3, "learning_rate": 1.9268163223827e-05, "loss": 0.0728, "step": 405 }, { "epoch": 0.3, "learning_rate": 1.9263681243875173e-05, "loss": 0.0509, "step": 406 }, { "epoch": 0.3, "learning_rate": 1.9259186105844483e-05, "loss": 0.0475, "step": 407 }, { "epoch": 0.3, "learning_rate": 1.92546778161198e-05, "loss": 0.0482, "step": 408 }, { "epoch": 0.3, "learning_rate": 1.9250156381104667e-05, "loss": 0.0504, "step": 409 }, { "epoch": 0.3, "learning_rate": 1.9245621807221306e-05, "loss": 0.058, "step": 410 }, { "epoch": 0.3, "learning_rate": 1.9241074100910603e-05, "loss": 0.0722, "step": 411 }, { "epoch": 0.3, "learning_rate": 1.9236513268632085e-05, "loss": 0.0652, "step": 412 }, { "epoch": 0.3, "learning_rate": 1.9231939316863938e-05, "loss": 0.0612, "step": 413 }, { "epoch": 0.3, "learning_rate": 1.922735225210298e-05, "loss": 0.0523, "step": 414 }, { "epoch": 0.31, "learning_rate": 1.9222752080864643e-05, "loss": 0.0624, "step": 415 }, { "epoch": 0.31, "learning_rate": 1.9218138809682988e-05, "loss": 0.0524, "step": 416 }, { "epoch": 0.31, "learning_rate": 1.9213512445110675e-05, "loss": 0.0489, "step": 417 }, { "epoch": 0.31, "learning_rate": 1.9208872993718967e-05, "loss": 0.0466, "step": 418 }, { "epoch": 0.31, "learning_rate": 1.920422046209772e-05, "loss": 0.0592, "step": 419 }, { "epoch": 0.31, "learning_rate": 1.919955485685535e-05, "loss": 0.0691, "step": 420 }, { "epoch": 0.31, "learning_rate": 1.919487618461887e-05, "loss": 0.0661, "step": 421 }, { "epoch": 0.31, "learning_rate": 1.9190184452033828e-05, "loss": 0.0515, "step": 422 }, { "epoch": 0.31, "learning_rate": 1.9185479665764336e-05, "loss": 0.039, "step": 423 }, { "epoch": 0.31, "learning_rate": 1.9180761832493045e-05, "loss": 0.0589, "step": 424 }, { "epoch": 0.31, "learning_rate": 1.917603095892114e-05, "loss": 0.0605, "step": 425 }, { "epoch": 0.31, "learning_rate": 1.917128705176833e-05, "loss": 0.0434, "step": 426 }, { "epoch": 0.31, "learning_rate": 1.9166530117772825e-05, "loss": 0.0526, "step": 427 }, { "epoch": 0.31, "learning_rate": 1.9161760163691347e-05, "loss": 0.0534, "step": 428 }, { "epoch": 0.32, "learning_rate": 1.9156977196299117e-05, "loss": 0.0533, "step": 429 }, { "epoch": 0.32, "learning_rate": 1.915218122238983e-05, "loss": 0.0606, "step": 430 }, { "epoch": 0.32, "learning_rate": 1.9147372248775663e-05, "loss": 0.065, "step": 431 }, { "epoch": 0.32, "learning_rate": 1.9142550282287247e-05, "loss": 0.0612, "step": 432 }, { "epoch": 0.32, "learning_rate": 1.913771532977368e-05, "loss": 0.0628, "step": 433 }, { "epoch": 0.32, "learning_rate": 1.9132867398102498e-05, "loss": 0.076, "step": 434 }, { "epoch": 0.32, "learning_rate": 1.912800649415968e-05, "loss": 0.0382, "step": 435 }, { "epoch": 0.32, "learning_rate": 1.912313262484962e-05, "loss": 0.0606, "step": 436 }, { "epoch": 0.32, "learning_rate": 1.9118245797095136e-05, "loss": 0.0536, "step": 437 }, { "epoch": 0.32, "learning_rate": 1.911334601783745e-05, "loss": 0.0668, "step": 438 }, { "epoch": 0.32, "learning_rate": 1.910843329403618e-05, "loss": 0.0525, "step": 439 }, { "epoch": 0.32, "learning_rate": 1.910350763266933e-05, "loss": 0.0578, "step": 440 }, { "epoch": 0.32, "learning_rate": 1.9098569040733278e-05, "loss": 0.0465, "step": 441 }, { "epoch": 0.33, "learning_rate": 1.9093617525242772e-05, "loss": 0.0536, "step": 442 }, { "epoch": 0.33, "learning_rate": 1.908865309323092e-05, "loss": 0.0519, "step": 443 }, { "epoch": 0.33, "learning_rate": 1.9083675751749174e-05, "loss": 0.039, "step": 444 }, { "epoch": 0.33, "learning_rate": 1.907868550786731e-05, "loss": 0.062, "step": 445 }, { "epoch": 0.33, "learning_rate": 1.907368236867345e-05, "loss": 0.045, "step": 446 }, { "epoch": 0.33, "learning_rate": 1.906866634127402e-05, "loss": 0.0506, "step": 447 }, { "epoch": 0.33, "learning_rate": 1.9063637432793757e-05, "loss": 0.0576, "step": 448 }, { "epoch": 0.33, "learning_rate": 1.90585956503757e-05, "loss": 0.0567, "step": 449 }, { "epoch": 0.33, "learning_rate": 1.9053541001181156e-05, "loss": 0.0429, "step": 450 }, { "epoch": 0.33, "learning_rate": 1.9048473492389724e-05, "loss": 0.0547, "step": 451 }, { "epoch": 0.33, "learning_rate": 1.9043393131199266e-05, "loss": 0.0419, "step": 452 }, { "epoch": 0.33, "learning_rate": 1.9038299924825897e-05, "loss": 0.0605, "step": 453 }, { "epoch": 0.33, "learning_rate": 1.9033193880503976e-05, "loss": 0.0599, "step": 454 }, { "epoch": 0.33, "learning_rate": 1.90280750054861e-05, "loss": 0.0569, "step": 455 }, { "epoch": 0.34, "learning_rate": 1.9022943307043085e-05, "loss": 0.0526, "step": 456 }, { "epoch": 0.34, "learning_rate": 1.9017798792463972e-05, "loss": 0.0445, "step": 457 }, { "epoch": 0.34, "learning_rate": 1.9012641469055998e-05, "loss": 0.0671, "step": 458 }, { "epoch": 0.34, "learning_rate": 1.900747134414459e-05, "loss": 0.0665, "step": 459 }, { "epoch": 0.34, "learning_rate": 1.9002288425073367e-05, "loss": 0.0491, "step": 460 }, { "epoch": 0.34, "learning_rate": 1.8997092719204118e-05, "loss": 0.0558, "step": 461 }, { "epoch": 0.34, "learning_rate": 1.8991884233916795e-05, "loss": 0.0369, "step": 462 }, { "epoch": 0.34, "learning_rate": 1.8986662976609495e-05, "loss": 0.0456, "step": 463 }, { "epoch": 0.34, "learning_rate": 1.8981428954698466e-05, "loss": 0.0572, "step": 464 }, { "epoch": 0.34, "learning_rate": 1.8976182175618076e-05, "loss": 0.0602, "step": 465 }, { "epoch": 0.34, "learning_rate": 1.8970922646820825e-05, "loss": 0.0709, "step": 466 }, { "epoch": 0.34, "learning_rate": 1.8965650375777318e-05, "loss": 0.0603, "step": 467 }, { "epoch": 0.34, "learning_rate": 1.8960365369976256e-05, "loss": 0.0488, "step": 468 }, { "epoch": 0.35, "learning_rate": 1.895506763692443e-05, "loss": 0.0566, "step": 469 }, { "epoch": 0.35, "learning_rate": 1.8949757184146706e-05, "loss": 0.0628, "step": 470 }, { "epoch": 0.35, "learning_rate": 1.8944434019186025e-05, "loss": 0.0426, "step": 471 }, { "epoch": 0.35, "learning_rate": 1.893909814960338e-05, "loss": 0.0607, "step": 472 }, { "epoch": 0.35, "learning_rate": 1.8933749582977804e-05, "loss": 0.0422, "step": 473 }, { "epoch": 0.35, "learning_rate": 1.8928388326906376e-05, "loss": 0.058, "step": 474 }, { "epoch": 0.35, "learning_rate": 1.892301438900419e-05, "loss": 0.0481, "step": 475 }, { "epoch": 0.35, "learning_rate": 1.8917627776904352e-05, "loss": 0.0628, "step": 476 }, { "epoch": 0.35, "learning_rate": 1.891222849825798e-05, "loss": 0.0546, "step": 477 }, { "epoch": 0.35, "learning_rate": 1.8906816560734182e-05, "loss": 0.056, "step": 478 }, { "epoch": 0.35, "learning_rate": 1.890139197202003e-05, "loss": 0.0444, "step": 479 }, { "epoch": 0.35, "learning_rate": 1.889595473982059e-05, "loss": 0.035, "step": 480 }, { "epoch": 0.35, "learning_rate": 1.8890504871858865e-05, "loss": 0.0578, "step": 481 }, { "epoch": 0.35, "learning_rate": 1.8885042375875825e-05, "loss": 0.0397, "step": 482 }, { "epoch": 0.36, "learning_rate": 1.887956725963036e-05, "loss": 0.0532, "step": 483 }, { "epoch": 0.36, "learning_rate": 1.8874079530899298e-05, "loss": 0.0554, "step": 484 }, { "epoch": 0.36, "learning_rate": 1.8868579197477377e-05, "loss": 0.0443, "step": 485 }, { "epoch": 0.36, "learning_rate": 1.8863066267177234e-05, "loss": 0.052, "step": 486 }, { "epoch": 0.36, "learning_rate": 1.8857540747829407e-05, "loss": 0.0589, "step": 487 }, { "epoch": 0.36, "learning_rate": 1.885200264728231e-05, "loss": 0.0586, "step": 488 }, { "epoch": 0.36, "learning_rate": 1.884645197340223e-05, "loss": 0.0622, "step": 489 }, { "epoch": 0.36, "learning_rate": 1.884088873407331e-05, "loss": 0.0459, "step": 490 }, { "epoch": 0.36, "learning_rate": 1.8835312937197546e-05, "loss": 0.0425, "step": 491 }, { "epoch": 0.36, "learning_rate": 1.882972459069476e-05, "loss": 0.0472, "step": 492 }, { "epoch": 0.36, "learning_rate": 1.8824123702502615e-05, "loss": 0.0583, "step": 493 }, { "epoch": 0.36, "learning_rate": 1.8818510280576577e-05, "loss": 0.0404, "step": 494 }, { "epoch": 0.36, "learning_rate": 1.881288433288992e-05, "loss": 0.0657, "step": 495 }, { "epoch": 0.36, "learning_rate": 1.88072458674337e-05, "loss": 0.0459, "step": 496 }, { "epoch": 0.37, "learning_rate": 1.880159489221677e-05, "loss": 0.0602, "step": 497 }, { "epoch": 0.37, "learning_rate": 1.8795931415265735e-05, "loss": 0.0567, "step": 498 }, { "epoch": 0.37, "learning_rate": 1.879025544462497e-05, "loss": 0.0568, "step": 499 }, { "epoch": 0.37, "learning_rate": 1.8784566988356586e-05, "loss": 0.0552, "step": 500 }, { "epoch": 0.37, "learning_rate": 1.8778866054540436e-05, "loss": 0.0494, "step": 501 }, { "epoch": 0.37, "learning_rate": 1.877315265127409e-05, "loss": 0.0533, "step": 502 }, { "epoch": 0.37, "learning_rate": 1.8767426786672834e-05, "loss": 0.0498, "step": 503 }, { "epoch": 0.37, "learning_rate": 1.8761688468869658e-05, "loss": 0.0535, "step": 504 }, { "epoch": 0.37, "learning_rate": 1.8755937706015232e-05, "loss": 0.0626, "step": 505 }, { "epoch": 0.37, "learning_rate": 1.8750174506277902e-05, "loss": 0.0546, "step": 506 }, { "epoch": 0.37, "learning_rate": 1.8744398877843687e-05, "loss": 0.0448, "step": 507 }, { "epoch": 0.37, "learning_rate": 1.8738610828916255e-05, "loss": 0.0496, "step": 508 }, { "epoch": 0.37, "learning_rate": 1.873281036771692e-05, "loss": 0.0547, "step": 509 }, { "epoch": 0.38, "learning_rate": 1.8726997502484617e-05, "loss": 0.0573, "step": 510 }, { "epoch": 0.38, "learning_rate": 1.872117224147591e-05, "loss": 0.0595, "step": 511 }, { "epoch": 0.38, "learning_rate": 1.8715334592964964e-05, "loss": 0.0466, "step": 512 }, { "epoch": 0.38, "learning_rate": 1.8709484565243545e-05, "loss": 0.0462, "step": 513 }, { "epoch": 0.38, "learning_rate": 1.8703622166620995e-05, "loss": 0.0543, "step": 514 }, { "epoch": 0.38, "learning_rate": 1.869774740542423e-05, "loss": 0.0451, "step": 515 }, { "epoch": 0.38, "learning_rate": 1.869186028999773e-05, "loss": 0.0521, "step": 516 }, { "epoch": 0.38, "learning_rate": 1.8685960828703518e-05, "loss": 0.0516, "step": 517 }, { "epoch": 0.38, "learning_rate": 1.868004902992115e-05, "loss": 0.0538, "step": 518 }, { "epoch": 0.38, "learning_rate": 1.8674124902047717e-05, "loss": 0.0551, "step": 519 }, { "epoch": 0.38, "learning_rate": 1.8668188453497814e-05, "loss": 0.0645, "step": 520 }, { "epoch": 0.38, "learning_rate": 1.8662239692703533e-05, "loss": 0.0589, "step": 521 }, { "epoch": 0.38, "learning_rate": 1.865627862811447e-05, "loss": 0.049, "step": 522 }, { "epoch": 0.38, "learning_rate": 1.8650305268197672e-05, "loss": 0.0472, "step": 523 }, { "epoch": 0.39, "learning_rate": 1.8644319621437677e-05, "loss": 0.0543, "step": 524 }, { "epoch": 0.39, "learning_rate": 1.8638321696336453e-05, "loss": 0.0533, "step": 525 }, { "epoch": 0.39, "learning_rate": 1.863231150141343e-05, "loss": 0.0576, "step": 526 }, { "epoch": 0.39, "learning_rate": 1.8626289045205442e-05, "loss": 0.0409, "step": 527 }, { "epoch": 0.39, "learning_rate": 1.8620254336266757e-05, "loss": 0.0537, "step": 528 }, { "epoch": 0.39, "learning_rate": 1.8614207383169036e-05, "loss": 0.0444, "step": 529 }, { "epoch": 0.39, "learning_rate": 1.8608148194501343e-05, "loss": 0.0609, "step": 530 }, { "epoch": 0.39, "learning_rate": 1.8602076778870106e-05, "loss": 0.0544, "step": 531 }, { "epoch": 0.39, "learning_rate": 1.8595993144899135e-05, "loss": 0.0475, "step": 532 }, { "epoch": 0.39, "learning_rate": 1.8589897301229587e-05, "loss": 0.0534, "step": 533 }, { "epoch": 0.39, "learning_rate": 1.858378925651996e-05, "loss": 0.0434, "step": 534 }, { "epoch": 0.39, "learning_rate": 1.8577669019446086e-05, "loss": 0.0424, "step": 535 }, { "epoch": 0.39, "learning_rate": 1.8571536598701114e-05, "loss": 0.0508, "step": 536 }, { "epoch": 0.4, "learning_rate": 1.8565392002995503e-05, "loss": 0.0509, "step": 537 }, { "epoch": 0.4, "learning_rate": 1.8559235241056994e-05, "loss": 0.0588, "step": 538 }, { "epoch": 0.4, "learning_rate": 1.855306632163062e-05, "loss": 0.0621, "step": 539 }, { "epoch": 0.4, "learning_rate": 1.8546885253478678e-05, "loss": 0.0528, "step": 540 }, { "epoch": 0.4, "learning_rate": 1.854069204538072e-05, "loss": 0.0473, "step": 541 }, { "epoch": 0.4, "learning_rate": 1.853448670613354e-05, "loss": 0.0371, "step": 542 }, { "epoch": 0.4, "learning_rate": 1.8528269244551177e-05, "loss": 0.0474, "step": 543 }, { "epoch": 0.4, "learning_rate": 1.8522039669464863e-05, "loss": 0.0574, "step": 544 }, { "epoch": 0.4, "learning_rate": 1.851579798972306e-05, "loss": 0.0401, "step": 545 }, { "epoch": 0.4, "learning_rate": 1.8509544214191403e-05, "loss": 0.0582, "step": 546 }, { "epoch": 0.4, "learning_rate": 1.850327835175273e-05, "loss": 0.053, "step": 547 }, { "epoch": 0.4, "learning_rate": 1.8497000411307035e-05, "loss": 0.0398, "step": 548 }, { "epoch": 0.4, "learning_rate": 1.8490710401771457e-05, "loss": 0.0432, "step": 549 }, { "epoch": 0.4, "learning_rate": 1.8484408332080298e-05, "loss": 0.0618, "step": 550 }, { "epoch": 0.41, "learning_rate": 1.8478094211184977e-05, "loss": 0.0546, "step": 551 }, { "epoch": 0.41, "learning_rate": 1.847176804805404e-05, "loss": 0.0544, "step": 552 }, { "epoch": 0.41, "learning_rate": 1.8465429851673123e-05, "loss": 0.0528, "step": 553 }, { "epoch": 0.41, "learning_rate": 1.845907963104497e-05, "loss": 0.0471, "step": 554 }, { "epoch": 0.41, "learning_rate": 1.84527173951894e-05, "loss": 0.0361, "step": 555 }, { "epoch": 0.41, "learning_rate": 1.844634315314329e-05, "loss": 0.0632, "step": 556 }, { "epoch": 0.41, "learning_rate": 1.8439956913960577e-05, "loss": 0.0429, "step": 557 }, { "epoch": 0.41, "learning_rate": 1.843355868671224e-05, "loss": 0.0495, "step": 558 }, { "epoch": 0.41, "learning_rate": 1.8427148480486283e-05, "loss": 0.0432, "step": 559 }, { "epoch": 0.41, "learning_rate": 1.8420726304387723e-05, "loss": 0.037, "step": 560 }, { "epoch": 0.41, "learning_rate": 1.841429216753859e-05, "loss": 0.0529, "step": 561 }, { "epoch": 0.41, "learning_rate": 1.840784607907788e-05, "loss": 0.0495, "step": 562 }, { "epoch": 0.41, "learning_rate": 1.8401388048161592e-05, "loss": 0.0562, "step": 563 }, { "epoch": 0.42, "learning_rate": 1.839491808396267e-05, "loss": 0.0524, "step": 564 }, { "epoch": 0.42, "learning_rate": 1.8388436195671015e-05, "loss": 0.0558, "step": 565 }, { "epoch": 0.42, "learning_rate": 1.8381942392493464e-05, "loss": 0.0471, "step": 566 }, { "epoch": 0.42, "learning_rate": 1.8375436683653775e-05, "loss": 0.0542, "step": 567 }, { "epoch": 0.42, "learning_rate": 1.836891907839262e-05, "loss": 0.0529, "step": 568 }, { "epoch": 0.42, "learning_rate": 1.836238958596757e-05, "loss": 0.0533, "step": 569 }, { "epoch": 0.42, "learning_rate": 1.8355848215653073e-05, "loss": 0.0518, "step": 570 }, { "epoch": 0.42, "learning_rate": 1.8349294976740455e-05, "loss": 0.056, "step": 571 }, { "epoch": 0.42, "learning_rate": 1.8342729878537903e-05, "loss": 0.0429, "step": 572 }, { "epoch": 0.42, "learning_rate": 1.8336152930370436e-05, "loss": 0.0527, "step": 573 }, { "epoch": 0.42, "learning_rate": 1.8329564141579924e-05, "loss": 0.0447, "step": 574 }, { "epoch": 0.42, "learning_rate": 1.8322963521525032e-05, "loss": 0.0493, "step": 575 }, { "epoch": 0.42, "learning_rate": 1.831635107958125e-05, "loss": 0.0548, "step": 576 }, { "epoch": 0.42, "learning_rate": 1.830972682514085e-05, "loss": 0.0531, "step": 577 }, { "epoch": 0.43, "learning_rate": 1.8303090767612882e-05, "loss": 0.0635, "step": 578 }, { "epoch": 0.43, "learning_rate": 1.8296442916423172e-05, "loss": 0.04, "step": 579 }, { "epoch": 0.43, "learning_rate": 1.828978328101428e-05, "loss": 0.0531, "step": 580 }, { "epoch": 0.43, "learning_rate": 1.8283111870845515e-05, "loss": 0.0478, "step": 581 }, { "epoch": 0.43, "learning_rate": 1.8276428695392908e-05, "loss": 0.0448, "step": 582 }, { "epoch": 0.43, "learning_rate": 1.826973376414921e-05, "loss": 0.0555, "step": 583 }, { "epoch": 0.43, "learning_rate": 1.8263027086623852e-05, "loss": 0.044, "step": 584 }, { "epoch": 0.43, "learning_rate": 1.8256308672342968e-05, "loss": 0.0491, "step": 585 }, { "epoch": 0.43, "learning_rate": 1.824957853084935e-05, "loss": 0.061, "step": 586 }, { "epoch": 0.43, "learning_rate": 1.8242836671702453e-05, "loss": 0.0575, "step": 587 }, { "epoch": 0.43, "learning_rate": 1.8236083104478373e-05, "loss": 0.0598, "step": 588 }, { "epoch": 0.43, "learning_rate": 1.8229317838769844e-05, "loss": 0.0477, "step": 589 }, { "epoch": 0.43, "learning_rate": 1.82225408841862e-05, "loss": 0.0475, "step": 590 }, { "epoch": 0.43, "learning_rate": 1.82157522503534e-05, "loss": 0.058, "step": 591 }, { "epoch": 0.44, "learning_rate": 1.8208951946913965e-05, "loss": 0.0482, "step": 592 }, { "epoch": 0.44, "learning_rate": 1.8202139983527018e-05, "loss": 0.0517, "step": 593 }, { "epoch": 0.44, "learning_rate": 1.819531636986823e-05, "loss": 0.0373, "step": 594 }, { "epoch": 0.44, "learning_rate": 1.8188481115629818e-05, "loss": 0.0435, "step": 595 }, { "epoch": 0.44, "learning_rate": 1.8181634230520537e-05, "loss": 0.0486, "step": 596 }, { "epoch": 0.44, "learning_rate": 1.8174775724265668e-05, "loss": 0.0507, "step": 597 }, { "epoch": 0.44, "learning_rate": 1.8167905606606995e-05, "loss": 0.0375, "step": 598 }, { "epoch": 0.44, "learning_rate": 1.816102388730278e-05, "loss": 0.045, "step": 599 }, { "epoch": 0.44, "learning_rate": 1.8154130576127794e-05, "loss": 0.0398, "step": 600 }, { "epoch": 0.44, "learning_rate": 1.8147225682873246e-05, "loss": 0.0516, "step": 601 }, { "epoch": 0.44, "learning_rate": 1.8140309217346805e-05, "loss": 0.0491, "step": 602 }, { "epoch": 0.44, "learning_rate": 1.8133381189372582e-05, "loss": 0.0644, "step": 603 }, { "epoch": 0.44, "learning_rate": 1.812644160879111e-05, "loss": 0.054, "step": 604 }, { "epoch": 0.45, "learning_rate": 1.811949048545932e-05, "loss": 0.0471, "step": 605 }, { "epoch": 0.45, "learning_rate": 1.8112527829250558e-05, "loss": 0.0526, "step": 606 }, { "epoch": 0.45, "learning_rate": 1.8105553650054534e-05, "loss": 0.0519, "step": 607 }, { "epoch": 0.45, "learning_rate": 1.809856795777733e-05, "loss": 0.0507, "step": 608 }, { "epoch": 0.45, "learning_rate": 1.809157076234139e-05, "loss": 0.0414, "step": 609 }, { "epoch": 0.45, "learning_rate": 1.8084562073685482e-05, "loss": 0.0423, "step": 610 }, { "epoch": 0.45, "learning_rate": 1.807754190176471e-05, "loss": 0.0563, "step": 611 }, { "epoch": 0.45, "learning_rate": 1.807051025655048e-05, "loss": 0.0649, "step": 612 }, { "epoch": 0.45, "learning_rate": 1.806346714803051e-05, "loss": 0.0568, "step": 613 }, { "epoch": 0.45, "learning_rate": 1.8056412586208784e-05, "loss": 0.0476, "step": 614 }, { "epoch": 0.45, "learning_rate": 1.8049346581105564e-05, "loss": 0.0509, "step": 615 }, { "epoch": 0.45, "learning_rate": 1.8042269142757354e-05, "loss": 0.0499, "step": 616 }, { "epoch": 0.45, "learning_rate": 1.8035180281216915e-05, "loss": 0.0503, "step": 617 }, { "epoch": 0.45, "learning_rate": 1.8028080006553223e-05, "loss": 0.0473, "step": 618 }, { "epoch": 0.46, "learning_rate": 1.8020968328851463e-05, "loss": 0.0556, "step": 619 }, { "epoch": 0.46, "learning_rate": 1.8013845258213024e-05, "loss": 0.0462, "step": 620 }, { "epoch": 0.46, "learning_rate": 1.8006710804755472e-05, "loss": 0.0587, "step": 621 }, { "epoch": 0.46, "learning_rate": 1.7999564978612544e-05, "loss": 0.0504, "step": 622 }, { "epoch": 0.46, "learning_rate": 1.7992407789934134e-05, "loss": 0.0439, "step": 623 }, { "epoch": 0.46, "learning_rate": 1.7985239248886264e-05, "loss": 0.0441, "step": 624 }, { "epoch": 0.46, "learning_rate": 1.797805936565109e-05, "loss": 0.0481, "step": 625 }, { "epoch": 0.46, "learning_rate": 1.797086815042688e-05, "loss": 0.0467, "step": 626 }, { "epoch": 0.46, "learning_rate": 1.796366561342799e-05, "loss": 0.053, "step": 627 }, { "epoch": 0.46, "learning_rate": 1.7956451764884862e-05, "loss": 0.0412, "step": 628 }, { "epoch": 0.46, "learning_rate": 1.7949226615044008e-05, "loss": 0.054, "step": 629 }, { "epoch": 0.46, "learning_rate": 1.7941990174167987e-05, "loss": 0.051, "step": 630 }, { "epoch": 0.46, "learning_rate": 1.7934742452535393e-05, "loss": 0.0528, "step": 631 }, { "epoch": 0.47, "learning_rate": 1.7927483460440857e-05, "loss": 0.0466, "step": 632 }, { "epoch": 0.47, "learning_rate": 1.7920213208194997e-05, "loss": 0.0474, "step": 633 }, { "epoch": 0.47, "learning_rate": 1.7912931706124447e-05, "loss": 0.0584, "step": 634 }, { "epoch": 0.47, "learning_rate": 1.7905638964571807e-05, "loss": 0.0498, "step": 635 }, { "epoch": 0.47, "learning_rate": 1.789833499389564e-05, "loss": 0.0429, "step": 636 }, { "epoch": 0.47, "learning_rate": 1.789101980447047e-05, "loss": 0.0623, "step": 637 }, { "epoch": 0.47, "learning_rate": 1.7883693406686746e-05, "loss": 0.0572, "step": 638 }, { "epoch": 0.47, "learning_rate": 1.7876355810950845e-05, "loss": 0.0507, "step": 639 }, { "epoch": 0.47, "learning_rate": 1.786900702768504e-05, "loss": 0.0558, "step": 640 }, { "epoch": 0.47, "learning_rate": 1.7861647067327502e-05, "loss": 0.0416, "step": 641 }, { "epoch": 0.47, "learning_rate": 1.7854275940332272e-05, "loss": 0.0436, "step": 642 }, { "epoch": 0.47, "learning_rate": 1.7846893657169263e-05, "loss": 0.0491, "step": 643 }, { "epoch": 0.47, "learning_rate": 1.7839500228324223e-05, "loss": 0.0531, "step": 644 }, { "epoch": 0.47, "learning_rate": 1.7832095664298733e-05, "loss": 0.042, "step": 645 }, { "epoch": 0.48, "learning_rate": 1.78246799756102e-05, "loss": 0.0533, "step": 646 }, { "epoch": 0.48, "learning_rate": 1.781725317279182e-05, "loss": 0.0381, "step": 647 }, { "epoch": 0.48, "learning_rate": 1.7809815266392575e-05, "loss": 0.0516, "step": 648 }, { "epoch": 0.48, "learning_rate": 1.7802366266977234e-05, "loss": 0.0405, "step": 649 }, { "epoch": 0.48, "learning_rate": 1.779490618512631e-05, "loss": 0.0549, "step": 650 }, { "epoch": 0.48, "learning_rate": 1.7787435031436053e-05, "loss": 0.034, "step": 651 }, { "epoch": 0.48, "learning_rate": 1.7779952816518454e-05, "loss": 0.0537, "step": 652 }, { "epoch": 0.48, "learning_rate": 1.77724595510012e-05, "loss": 0.0583, "step": 653 }, { "epoch": 0.48, "learning_rate": 1.7764955245527693e-05, "loss": 0.0587, "step": 654 }, { "epoch": 0.48, "learning_rate": 1.7757439910756995e-05, "loss": 0.0441, "step": 655 }, { "epoch": 0.48, "learning_rate": 1.7749913557363844e-05, "loss": 0.0524, "step": 656 }, { "epoch": 0.48, "learning_rate": 1.7742376196038636e-05, "loss": 0.0501, "step": 657 }, { "epoch": 0.48, "learning_rate": 1.7734827837487386e-05, "loss": 0.0448, "step": 658 }, { "epoch": 0.48, "learning_rate": 1.7727268492431746e-05, "loss": 0.0534, "step": 659 }, { "epoch": 0.49, "learning_rate": 1.771969817160896e-05, "loss": 0.0487, "step": 660 }, { "epoch": 0.49, "learning_rate": 1.771211688577187e-05, "loss": 0.0476, "step": 661 }, { "epoch": 0.49, "learning_rate": 1.770452464568889e-05, "loss": 0.05, "step": 662 }, { "epoch": 0.49, "learning_rate": 1.7696921462143996e-05, "loss": 0.061, "step": 663 }, { "epoch": 0.49, "learning_rate": 1.7689307345936705e-05, "loss": 0.0428, "step": 664 }, { "epoch": 0.49, "learning_rate": 1.7681682307882058e-05, "loss": 0.0484, "step": 665 }, { "epoch": 0.49, "learning_rate": 1.767404635881062e-05, "loss": 0.0676, "step": 666 }, { "epoch": 0.49, "learning_rate": 1.7666399509568446e-05, "loss": 0.045, "step": 667 }, { "epoch": 0.49, "learning_rate": 1.7658741771017076e-05, "loss": 0.0492, "step": 668 }, { "epoch": 0.49, "learning_rate": 1.765107315403352e-05, "loss": 0.0435, "step": 669 }, { "epoch": 0.49, "learning_rate": 1.7643393669510236e-05, "loss": 0.0428, "step": 670 }, { "epoch": 0.49, "learning_rate": 1.7635703328355113e-05, "loss": 0.0563, "step": 671 }, { "epoch": 0.49, "learning_rate": 1.7628002141491477e-05, "loss": 0.0527, "step": 672 }, { "epoch": 0.5, "learning_rate": 1.7620290119858046e-05, "loss": 0.0539, "step": 673 }, { "epoch": 0.5, "learning_rate": 1.7612567274408925e-05, "loss": 0.0461, "step": 674 }, { "epoch": 0.5, "learning_rate": 1.76048336161136e-05, "loss": 0.0563, "step": 675 }, { "epoch": 0.5, "learning_rate": 1.759708915595692e-05, "loss": 0.0506, "step": 676 }, { "epoch": 0.5, "learning_rate": 1.7589333904939065e-05, "loss": 0.0603, "step": 677 }, { "epoch": 0.5, "learning_rate": 1.7581567874075552e-05, "loss": 0.0474, "step": 678 }, { "epoch": 0.5, "learning_rate": 1.7573791074397207e-05, "loss": 0.0349, "step": 679 }, { "epoch": 0.5, "learning_rate": 1.7566003516950146e-05, "loss": 0.0598, "step": 680 }, { "epoch": 0.5, "learning_rate": 1.7558205212795772e-05, "loss": 0.0484, "step": 681 }, { "epoch": 0.5, "learning_rate": 1.755039617301075e-05, "loss": 0.0509, "step": 682 }, { "epoch": 0.5, "learning_rate": 1.7542576408687005e-05, "loss": 0.0544, "step": 683 }, { "epoch": 0.5, "learning_rate": 1.753474593093167e-05, "loss": 0.0583, "step": 684 }, { "epoch": 0.5, "learning_rate": 1.752690475086712e-05, "loss": 0.0556, "step": 685 }, { "epoch": 0.5, "learning_rate": 1.751905287963091e-05, "loss": 0.0618, "step": 686 }, { "epoch": 0.51, "learning_rate": 1.7511190328375804e-05, "loss": 0.0324, "step": 687 }, { "epoch": 0.51, "learning_rate": 1.7503317108269722e-05, "loss": 0.0555, "step": 688 }, { "epoch": 0.51, "learning_rate": 1.7495433230495736e-05, "loss": 0.0692, "step": 689 }, { "epoch": 0.51, "learning_rate": 1.7487538706252062e-05, "loss": 0.0379, "step": 690 }, { "epoch": 0.51, "learning_rate": 1.7479633546752034e-05, "loss": 0.0509, "step": 691 }, { "epoch": 0.51, "learning_rate": 1.7471717763224096e-05, "loss": 0.0478, "step": 692 }, { "epoch": 0.51, "learning_rate": 1.746379136691178e-05, "loss": 0.0541, "step": 693 }, { "epoch": 0.51, "learning_rate": 1.7455854369073703e-05, "loss": 0.048, "step": 694 }, { "epoch": 0.51, "learning_rate": 1.7447906780983516e-05, "loss": 0.0553, "step": 695 }, { "epoch": 0.51, "learning_rate": 1.7439948613929928e-05, "loss": 0.0546, "step": 696 }, { "epoch": 0.51, "learning_rate": 1.7431979879216683e-05, "loss": 0.061, "step": 697 }, { "epoch": 0.51, "learning_rate": 1.742400058816252e-05, "loss": 0.0531, "step": 698 }, { "epoch": 0.51, "learning_rate": 1.7416010752101175e-05, "loss": 0.052, "step": 699 }, { "epoch": 0.52, "learning_rate": 1.740801038238137e-05, "loss": 0.0512, "step": 700 }, { "epoch": 0.52, "learning_rate": 1.739999949036678e-05, "loss": 0.0532, "step": 701 }, { "epoch": 0.52, "learning_rate": 1.7391978087436032e-05, "loss": 0.0446, "step": 702 }, { "epoch": 0.52, "learning_rate": 1.7383946184982675e-05, "loss": 0.0559, "step": 703 }, { "epoch": 0.52, "learning_rate": 1.737590379441518e-05, "loss": 0.0535, "step": 704 }, { "epoch": 0.52, "learning_rate": 1.7367850927156914e-05, "loss": 0.0622, "step": 705 }, { "epoch": 0.52, "learning_rate": 1.735978759464612e-05, "loss": 0.0526, "step": 706 }, { "epoch": 0.52, "learning_rate": 1.7351713808335902e-05, "loss": 0.0441, "step": 707 }, { "epoch": 0.52, "learning_rate": 1.734362957969423e-05, "loss": 0.0526, "step": 708 }, { "epoch": 0.52, "learning_rate": 1.7335534920203892e-05, "loss": 0.0516, "step": 709 }, { "epoch": 0.52, "learning_rate": 1.7327429841362494e-05, "loss": 0.0567, "step": 710 }, { "epoch": 0.52, "learning_rate": 1.7319314354682443e-05, "loss": 0.0451, "step": 711 }, { "epoch": 0.52, "learning_rate": 1.7311188471690925e-05, "loss": 0.0377, "step": 712 }, { "epoch": 0.52, "learning_rate": 1.7303052203929903e-05, "loss": 0.0418, "step": 713 }, { "epoch": 0.53, "learning_rate": 1.729490556295608e-05, "loss": 0.0621, "step": 714 }, { "epoch": 0.53, "learning_rate": 1.7286748560340904e-05, "loss": 0.0596, "step": 715 }, { "epoch": 0.53, "learning_rate": 1.7278581207670522e-05, "loss": 0.0552, "step": 716 }, { "epoch": 0.53, "learning_rate": 1.727040351654581e-05, "loss": 0.0587, "step": 717 }, { "epoch": 0.53, "learning_rate": 1.72622154985823e-05, "loss": 0.0464, "step": 718 }, { "epoch": 0.53, "learning_rate": 1.7254017165410208e-05, "loss": 0.0506, "step": 719 }, { "epoch": 0.53, "learning_rate": 1.7245808528674403e-05, "loss": 0.0607, "step": 720 }, { "epoch": 0.53, "learning_rate": 1.7237589600034378e-05, "loss": 0.0584, "step": 721 }, { "epoch": 0.53, "learning_rate": 1.7229360391164256e-05, "loss": 0.0472, "step": 722 }, { "epoch": 0.53, "learning_rate": 1.7221120913752758e-05, "loss": 0.0552, "step": 723 }, { "epoch": 0.53, "learning_rate": 1.7212871179503188e-05, "loss": 0.0459, "step": 724 }, { "epoch": 0.53, "learning_rate": 1.720461120013342e-05, "loss": 0.049, "step": 725 }, { "epoch": 0.53, "learning_rate": 1.719634098737588e-05, "loss": 0.0399, "step": 726 }, { "epoch": 0.53, "learning_rate": 1.718806055297753e-05, "loss": 0.0608, "step": 727 }, { "epoch": 0.54, "learning_rate": 1.7179769908699856e-05, "loss": 0.0534, "step": 728 }, { "epoch": 0.54, "learning_rate": 1.7171469066318836e-05, "loss": 0.0441, "step": 729 }, { "epoch": 0.54, "learning_rate": 1.716315803762494e-05, "loss": 0.0505, "step": 730 }, { "epoch": 0.54, "learning_rate": 1.71548368344231e-05, "loss": 0.0464, "step": 731 }, { "epoch": 0.54, "learning_rate": 1.7146505468532707e-05, "loss": 0.0497, "step": 732 }, { "epoch": 0.54, "learning_rate": 1.713816395178759e-05, "loss": 0.0504, "step": 733 }, { "epoch": 0.54, "learning_rate": 1.7129812296035985e-05, "loss": 0.0521, "step": 734 }, { "epoch": 0.54, "learning_rate": 1.7121450513140537e-05, "loss": 0.0443, "step": 735 }, { "epoch": 0.54, "learning_rate": 1.711307861497827e-05, "loss": 0.0418, "step": 736 }, { "epoch": 0.54, "learning_rate": 1.7104696613440588e-05, "loss": 0.0372, "step": 737 }, { "epoch": 0.54, "learning_rate": 1.709630452043323e-05, "loss": 0.0485, "step": 738 }, { "epoch": 0.54, "learning_rate": 1.7087902347876277e-05, "loss": 0.0383, "step": 739 }, { "epoch": 0.54, "learning_rate": 1.707949010770413e-05, "loss": 0.0434, "step": 740 }, { "epoch": 0.55, "learning_rate": 1.7071067811865477e-05, "loss": 0.0495, "step": 741 }, { "epoch": 0.55, "learning_rate": 1.7062635472323306e-05, "loss": 0.0474, "step": 742 }, { "epoch": 0.55, "learning_rate": 1.7054193101054864e-05, "loss": 0.042, "step": 743 }, { "epoch": 0.55, "learning_rate": 1.7045740710051637e-05, "loss": 0.0339, "step": 744 }, { "epoch": 0.55, "learning_rate": 1.703727831131936e-05, "loss": 0.0454, "step": 745 }, { "epoch": 0.55, "learning_rate": 1.7028805916877975e-05, "loss": 0.0495, "step": 746 }, { "epoch": 0.55, "learning_rate": 1.7020323538761618e-05, "loss": 0.0407, "step": 747 }, { "epoch": 0.55, "learning_rate": 1.7011831189018607e-05, "loss": 0.048, "step": 748 }, { "epoch": 0.55, "learning_rate": 1.7003328879711435e-05, "loss": 0.0419, "step": 749 }, { "epoch": 0.55, "learning_rate": 1.6994816622916726e-05, "loss": 0.041, "step": 750 }, { "epoch": 0.55, "learning_rate": 1.698629443072524e-05, "loss": 0.0438, "step": 751 }, { "epoch": 0.55, "learning_rate": 1.697776231524185e-05, "loss": 0.0482, "step": 752 }, { "epoch": 0.55, "learning_rate": 1.6969220288585523e-05, "loss": 0.0556, "step": 753 }, { "epoch": 0.55, "learning_rate": 1.696066836288931e-05, "loss": 0.0411, "step": 754 }, { "epoch": 0.56, "learning_rate": 1.6952106550300304e-05, "loss": 0.0503, "step": 755 }, { "epoch": 0.56, "learning_rate": 1.694353486297966e-05, "loss": 0.0417, "step": 756 }, { "epoch": 0.56, "learning_rate": 1.693495331310256e-05, "loss": 0.0494, "step": 757 }, { "epoch": 0.56, "learning_rate": 1.6926361912858172e-05, "loss": 0.0445, "step": 758 }, { "epoch": 0.56, "learning_rate": 1.6917760674449686e-05, "loss": 0.0455, "step": 759 }, { "epoch": 0.56, "learning_rate": 1.6909149610094245e-05, "loss": 0.0538, "step": 760 }, { "epoch": 0.56, "learning_rate": 1.690052873202295e-05, "loss": 0.0472, "step": 761 }, { "epoch": 0.56, "learning_rate": 1.689189805248085e-05, "loss": 0.0328, "step": 762 }, { "epoch": 0.56, "learning_rate": 1.688325758372692e-05, "loss": 0.0479, "step": 763 }, { "epoch": 0.56, "learning_rate": 1.6874607338034015e-05, "loss": 0.062, "step": 764 }, { "epoch": 0.56, "learning_rate": 1.68659473276889e-05, "loss": 0.0437, "step": 765 }, { "epoch": 0.56, "learning_rate": 1.6857277564992212e-05, "loss": 0.0479, "step": 766 }, { "epoch": 0.56, "learning_rate": 1.6848598062258424e-05, "loss": 0.0427, "step": 767 }, { "epoch": 0.57, "learning_rate": 1.683990883181585e-05, "loss": 0.0421, "step": 768 }, { "epoch": 0.57, "learning_rate": 1.6831209886006624e-05, "loss": 0.0489, "step": 769 }, { "epoch": 0.57, "learning_rate": 1.6822501237186677e-05, "loss": 0.0425, "step": 770 }, { "epoch": 0.57, "learning_rate": 1.6813782897725727e-05, "loss": 0.0459, "step": 771 }, { "epoch": 0.57, "learning_rate": 1.680505488000725e-05, "loss": 0.0587, "step": 772 }, { "epoch": 0.57, "learning_rate": 1.679631719642847e-05, "loss": 0.0436, "step": 773 }, { "epoch": 0.57, "learning_rate": 1.678756985940034e-05, "loss": 0.0358, "step": 774 }, { "epoch": 0.57, "learning_rate": 1.677881288134753e-05, "loss": 0.0609, "step": 775 }, { "epoch": 0.57, "learning_rate": 1.6770046274708404e-05, "loss": 0.0577, "step": 776 }, { "epoch": 0.57, "learning_rate": 1.6761270051934998e-05, "loss": 0.0487, "step": 777 }, { "epoch": 0.57, "learning_rate": 1.6752484225493e-05, "loss": 0.0398, "step": 778 }, { "epoch": 0.57, "learning_rate": 1.674368880786175e-05, "loss": 0.0401, "step": 779 }, { "epoch": 0.57, "learning_rate": 1.673488381153421e-05, "loss": 0.0482, "step": 780 }, { "epoch": 0.57, "learning_rate": 1.672606924901695e-05, "loss": 0.0449, "step": 781 }, { "epoch": 0.58, "learning_rate": 1.6717245132830114e-05, "loss": 0.0526, "step": 782 }, { "epoch": 0.58, "learning_rate": 1.670841147550743e-05, "loss": 0.0499, "step": 783 }, { "epoch": 0.58, "learning_rate": 1.6699568289596175e-05, "loss": 0.0594, "step": 784 }, { "epoch": 0.58, "learning_rate": 1.6690715587657154e-05, "loss": 0.0567, "step": 785 }, { "epoch": 0.58, "learning_rate": 1.66818533822647e-05, "loss": 0.0527, "step": 786 }, { "epoch": 0.58, "learning_rate": 1.667298168600663e-05, "loss": 0.0416, "step": 787 }, { "epoch": 0.58, "learning_rate": 1.6664100511484252e-05, "loss": 0.0506, "step": 788 }, { "epoch": 0.58, "learning_rate": 1.6655209871312338e-05, "loss": 0.046, "step": 789 }, { "epoch": 0.58, "learning_rate": 1.66463097781191e-05, "loss": 0.0452, "step": 790 }, { "epoch": 0.58, "learning_rate": 1.663740024454618e-05, "loss": 0.0451, "step": 791 }, { "epoch": 0.58, "learning_rate": 1.662848128324862e-05, "loss": 0.0532, "step": 792 }, { "epoch": 0.58, "learning_rate": 1.661955290689487e-05, "loss": 0.0355, "step": 793 }, { "epoch": 0.58, "learning_rate": 1.6610615128166738e-05, "loss": 0.0428, "step": 794 }, { "epoch": 0.58, "learning_rate": 1.66016679597594e-05, "loss": 0.0609, "step": 795 }, { "epoch": 0.59, "learning_rate": 1.659271141438135e-05, "loss": 0.0699, "step": 796 }, { "epoch": 0.59, "learning_rate": 1.6583745504754423e-05, "loss": 0.0535, "step": 797 }, { "epoch": 0.59, "learning_rate": 1.657477024361374e-05, "loss": 0.0473, "step": 798 }, { "epoch": 0.59, "learning_rate": 1.656578564370771e-05, "loss": 0.0379, "step": 799 }, { "epoch": 0.59, "learning_rate": 1.6556791717798013e-05, "loss": 0.0364, "step": 800 }, { "epoch": 0.59, "learning_rate": 1.6547788478659557e-05, "loss": 0.0485, "step": 801 }, { "epoch": 0.59, "learning_rate": 1.65387759390805e-05, "loss": 0.0523, "step": 802 }, { "epoch": 0.59, "learning_rate": 1.6529754111862196e-05, "loss": 0.052, "step": 803 }, { "epoch": 0.59, "learning_rate": 1.65207230098192e-05, "loss": 0.046, "step": 804 }, { "epoch": 0.59, "learning_rate": 1.651168264577923e-05, "loss": 0.0614, "step": 805 }, { "epoch": 0.59, "learning_rate": 1.6502633032583173e-05, "loss": 0.0424, "step": 806 }, { "epoch": 0.59, "learning_rate": 1.6493574183085043e-05, "loss": 0.037, "step": 807 }, { "epoch": 0.59, "learning_rate": 1.6484506110151977e-05, "loss": 0.0402, "step": 808 }, { "epoch": 0.6, "learning_rate": 1.647542882666422e-05, "loss": 0.0509, "step": 809 }, { "epoch": 0.6, "learning_rate": 1.646634234551508e-05, "loss": 0.0358, "step": 810 }, { "epoch": 0.6, "learning_rate": 1.645724667961095e-05, "loss": 0.0372, "step": 811 }, { "epoch": 0.6, "learning_rate": 1.6448141841871262e-05, "loss": 0.0388, "step": 812 }, { "epoch": 0.6, "learning_rate": 1.643902784522847e-05, "loss": 0.053, "step": 813 }, { "epoch": 0.6, "learning_rate": 1.6429904702628044e-05, "loss": 0.0603, "step": 814 }, { "epoch": 0.6, "learning_rate": 1.6420772427028445e-05, "loss": 0.0411, "step": 815 }, { "epoch": 0.6, "learning_rate": 1.64116310314011e-05, "loss": 0.0606, "step": 816 }, { "epoch": 0.6, "learning_rate": 1.6402480528730396e-05, "loss": 0.032, "step": 817 }, { "epoch": 0.6, "learning_rate": 1.639332093201365e-05, "loss": 0.0458, "step": 818 }, { "epoch": 0.6, "learning_rate": 1.6384152254261107e-05, "loss": 0.0564, "step": 819 }, { "epoch": 0.6, "learning_rate": 1.6374974508495895e-05, "loss": 0.0336, "step": 820 }, { "epoch": 0.6, "learning_rate": 1.636578770775404e-05, "loss": 0.0452, "step": 821 }, { "epoch": 0.6, "learning_rate": 1.6356591865084413e-05, "loss": 0.06, "step": 822 }, { "epoch": 0.61, "learning_rate": 1.6347386993548736e-05, "loss": 0.0448, "step": 823 }, { "epoch": 0.61, "learning_rate": 1.633817310622156e-05, "loss": 0.0419, "step": 824 }, { "epoch": 0.61, "learning_rate": 1.632895021619023e-05, "loss": 0.0414, "step": 825 }, { "epoch": 0.61, "learning_rate": 1.631971833655489e-05, "loss": 0.045, "step": 826 }, { "epoch": 0.61, "learning_rate": 1.6310477480428447e-05, "loss": 0.0425, "step": 827 }, { "epoch": 0.61, "learning_rate": 1.630122766093656e-05, "loss": 0.0381, "step": 828 }, { "epoch": 0.61, "learning_rate": 1.6291968891217616e-05, "loss": 0.0377, "step": 829 }, { "epoch": 0.61, "learning_rate": 1.6282701184422717e-05, "loss": 0.0646, "step": 830 }, { "epoch": 0.61, "learning_rate": 1.6273424553715664e-05, "loss": 0.0473, "step": 831 }, { "epoch": 0.61, "learning_rate": 1.6264139012272927e-05, "loss": 0.0439, "step": 832 }, { "epoch": 0.61, "learning_rate": 1.6254844573283638e-05, "loss": 0.0406, "step": 833 }, { "epoch": 0.61, "learning_rate": 1.6245541249949558e-05, "loss": 0.0478, "step": 834 }, { "epoch": 0.61, "learning_rate": 1.6236229055485074e-05, "loss": 0.0349, "step": 835 }, { "epoch": 0.62, "learning_rate": 1.622690800311718e-05, "loss": 0.0531, "step": 836 }, { "epoch": 0.62, "learning_rate": 1.6217578106085437e-05, "loss": 0.0411, "step": 837 }, { "epoch": 0.62, "learning_rate": 1.620823937764198e-05, "loss": 0.0614, "step": 838 }, { "epoch": 0.62, "learning_rate": 1.6198891831051486e-05, "loss": 0.0528, "step": 839 }, { "epoch": 0.62, "learning_rate": 1.618953547959115e-05, "loss": 0.0491, "step": 840 }, { "epoch": 0.62, "learning_rate": 1.6180170336550682e-05, "loss": 0.0514, "step": 841 }, { "epoch": 0.62, "learning_rate": 1.6170796415232278e-05, "loss": 0.0392, "step": 842 }, { "epoch": 0.62, "learning_rate": 1.6161413728950598e-05, "loss": 0.0427, "step": 843 }, { "epoch": 0.62, "learning_rate": 1.615202229103276e-05, "loss": 0.0504, "step": 844 }, { "epoch": 0.62, "learning_rate": 1.61426221148183e-05, "loss": 0.0501, "step": 845 }, { "epoch": 0.62, "learning_rate": 1.613321321365918e-05, "loss": 0.0571, "step": 846 }, { "epoch": 0.62, "learning_rate": 1.6123795600919748e-05, "loss": 0.0498, "step": 847 }, { "epoch": 0.62, "learning_rate": 1.6114369289976727e-05, "loss": 0.0399, "step": 848 }, { "epoch": 0.62, "learning_rate": 1.610493429421919e-05, "loss": 0.0565, "step": 849 }, { "epoch": 0.63, "learning_rate": 1.609549062704855e-05, "loss": 0.0492, "step": 850 }, { "epoch": 0.63, "learning_rate": 1.6086038301878543e-05, "loss": 0.0424, "step": 851 }, { "epoch": 0.63, "learning_rate": 1.607657733213519e-05, "loss": 0.0466, "step": 852 }, { "epoch": 0.63, "learning_rate": 1.6067107731256798e-05, "loss": 0.0606, "step": 853 }, { "epoch": 0.63, "learning_rate": 1.6057629512693938e-05, "loss": 0.0376, "step": 854 }, { "epoch": 0.63, "learning_rate": 1.604814268990941e-05, "loss": 0.0525, "step": 855 }, { "epoch": 0.63, "learning_rate": 1.6038647276378246e-05, "loss": 0.046, "step": 856 }, { "epoch": 0.63, "learning_rate": 1.6029143285587675e-05, "loss": 0.0464, "step": 857 }, { "epoch": 0.63, "learning_rate": 1.601963073103711e-05, "loss": 0.0525, "step": 858 }, { "epoch": 0.63, "learning_rate": 1.6010109626238128e-05, "loss": 0.042, "step": 859 }, { "epoch": 0.63, "learning_rate": 1.6000579984714453e-05, "loss": 0.0366, "step": 860 }, { "epoch": 0.63, "learning_rate": 1.599104182000193e-05, "loss": 0.0533, "step": 861 }, { "epoch": 0.63, "learning_rate": 1.5981495145648507e-05, "loss": 0.0439, "step": 862 }, { "epoch": 0.64, "learning_rate": 1.5971939975214235e-05, "loss": 0.0459, "step": 863 }, { "epoch": 0.64, "learning_rate": 1.5962376322271218e-05, "loss": 0.0455, "step": 864 }, { "epoch": 0.64, "learning_rate": 1.595280420040361e-05, "loss": 0.0465, "step": 865 }, { "epoch": 0.64, "learning_rate": 1.5943223623207608e-05, "loss": 0.0477, "step": 866 }, { "epoch": 0.64, "learning_rate": 1.593363460429139e-05, "loss": 0.0424, "step": 867 }, { "epoch": 0.64, "learning_rate": 1.5924037157275156e-05, "loss": 0.0465, "step": 868 }, { "epoch": 0.64, "learning_rate": 1.591443129579106e-05, "loss": 0.0418, "step": 869 }, { "epoch": 0.64, "learning_rate": 1.5904817033483216e-05, "loss": 0.0492, "step": 870 }, { "epoch": 0.64, "learning_rate": 1.589519438400766e-05, "loss": 0.0426, "step": 871 }, { "epoch": 0.64, "learning_rate": 1.588556336103235e-05, "loss": 0.0374, "step": 872 }, { "epoch": 0.64, "learning_rate": 1.587592397823713e-05, "loss": 0.0318, "step": 873 }, { "epoch": 0.64, "learning_rate": 1.586627624931373e-05, "loss": 0.056, "step": 874 }, { "epoch": 0.64, "learning_rate": 1.5856620187965726e-05, "loss": 0.0583, "step": 875 }, { "epoch": 0.64, "learning_rate": 1.584695580790853e-05, "loss": 0.0579, "step": 876 }, { "epoch": 0.65, "learning_rate": 1.583728312286937e-05, "loss": 0.0629, "step": 877 }, { "epoch": 0.65, "learning_rate": 1.5827602146587277e-05, "loss": 0.0438, "step": 878 }, { "epoch": 0.65, "learning_rate": 1.5817912892813048e-05, "loss": 0.0399, "step": 879 }, { "epoch": 0.65, "learning_rate": 1.5808215375309243e-05, "loss": 0.0456, "step": 880 }, { "epoch": 0.65, "learning_rate": 1.5798509607850165e-05, "loss": 0.0558, "step": 881 }, { "epoch": 0.65, "learning_rate": 1.578879560422182e-05, "loss": 0.0397, "step": 882 }, { "epoch": 0.65, "learning_rate": 1.577907337822193e-05, "loss": 0.0432, "step": 883 }, { "epoch": 0.65, "learning_rate": 1.576934294365988e-05, "loss": 0.0396, "step": 884 }, { "epoch": 0.65, "learning_rate": 1.5759604314356737e-05, "loss": 0.0495, "step": 885 }, { "epoch": 0.65, "learning_rate": 1.574985750414518e-05, "loss": 0.0351, "step": 886 }, { "epoch": 0.65, "learning_rate": 1.5740102526869524e-05, "loss": 0.043, "step": 887 }, { "epoch": 0.65, "learning_rate": 1.5730339396385684e-05, "loss": 0.0466, "step": 888 }, { "epoch": 0.65, "learning_rate": 1.5720568126561158e-05, "loss": 0.0424, "step": 889 }, { "epoch": 0.65, "learning_rate": 1.571078873127499e-05, "loss": 0.0332, "step": 890 }, { "epoch": 0.66, "learning_rate": 1.5701001224417788e-05, "loss": 0.0295, "step": 891 }, { "epoch": 0.66, "learning_rate": 1.569120561989166e-05, "loss": 0.0466, "step": 892 }, { "epoch": 0.66, "learning_rate": 1.568140193161023e-05, "loss": 0.0494, "step": 893 }, { "epoch": 0.66, "learning_rate": 1.5671590173498602e-05, "loss": 0.0365, "step": 894 }, { "epoch": 0.66, "learning_rate": 1.566177035949333e-05, "loss": 0.0367, "step": 895 }, { "epoch": 0.66, "learning_rate": 1.5651942503542435e-05, "loss": 0.0441, "step": 896 }, { "epoch": 0.66, "learning_rate": 1.5642106619605336e-05, "loss": 0.0493, "step": 897 }, { "epoch": 0.66, "learning_rate": 1.563226272165287e-05, "loss": 0.0543, "step": 898 }, { "epoch": 0.66, "learning_rate": 1.5622410823667247e-05, "loss": 0.047, "step": 899 }, { "epoch": 0.66, "learning_rate": 1.561255093964205e-05, "loss": 0.0365, "step": 900 }, { "epoch": 0.66, "learning_rate": 1.5602683083582205e-05, "loss": 0.0286, "step": 901 }, { "epoch": 0.66, "learning_rate": 1.559280726950395e-05, "loss": 0.0449, "step": 902 }, { "epoch": 0.66, "learning_rate": 1.5582923511434833e-05, "loss": 0.0484, "step": 903 }, { "epoch": 0.67, "learning_rate": 1.557303182341369e-05, "loss": 0.0503, "step": 904 }, { "epoch": 0.67, "learning_rate": 1.556313221949062e-05, "loss": 0.0434, "step": 905 }, { "epoch": 0.67, "learning_rate": 1.5553224713726954e-05, "loss": 0.0397, "step": 906 }, { "epoch": 0.67, "learning_rate": 1.5543309320195265e-05, "loss": 0.0448, "step": 907 }, { "epoch": 0.67, "learning_rate": 1.553338605297931e-05, "loss": 0.0482, "step": 908 }, { "epoch": 0.67, "learning_rate": 1.5523454926174053e-05, "loss": 0.0489, "step": 909 }, { "epoch": 0.67, "learning_rate": 1.55135159538856e-05, "loss": 0.0365, "step": 910 }, { "epoch": 0.67, "learning_rate": 1.5503569150231207e-05, "loss": 0.0547, "step": 911 }, { "epoch": 0.67, "learning_rate": 1.549361452933926e-05, "loss": 0.0399, "step": 912 }, { "epoch": 0.67, "learning_rate": 1.5483652105349246e-05, "loss": 0.0448, "step": 913 }, { "epoch": 0.67, "learning_rate": 1.5473681892411733e-05, "loss": 0.049, "step": 914 }, { "epoch": 0.67, "learning_rate": 1.5463703904688353e-05, "loss": 0.039, "step": 915 }, { "epoch": 0.67, "learning_rate": 1.5453718156351775e-05, "loss": 0.0335, "step": 916 }, { "epoch": 0.67, "learning_rate": 1.544372466158571e-05, "loss": 0.0525, "step": 917 }, { "epoch": 0.68, "learning_rate": 1.543372343458485e-05, "loss": 0.0417, "step": 918 }, { "epoch": 0.68, "learning_rate": 1.5423714489554887e-05, "loss": 0.0624, "step": 919 }, { "epoch": 0.68, "learning_rate": 1.541369784071246e-05, "loss": 0.0429, "step": 920 }, { "epoch": 0.68, "learning_rate": 1.5403673502285167e-05, "loss": 0.042, "step": 921 }, { "epoch": 0.68, "learning_rate": 1.5393641488511514e-05, "loss": 0.0484, "step": 922 }, { "epoch": 0.68, "learning_rate": 1.538360181364092e-05, "loss": 0.0538, "step": 923 }, { "epoch": 0.68, "learning_rate": 1.537355449193367e-05, "loss": 0.0341, "step": 924 }, { "epoch": 0.68, "learning_rate": 1.5363499537660936e-05, "loss": 0.0436, "step": 925 }, { "epoch": 0.68, "learning_rate": 1.5353436965104708e-05, "loss": 0.059, "step": 926 }, { "epoch": 0.68, "learning_rate": 1.5343366788557807e-05, "loss": 0.0408, "step": 927 }, { "epoch": 0.68, "learning_rate": 1.533328902232385e-05, "loss": 0.0509, "step": 928 }, { "epoch": 0.68, "learning_rate": 1.5323203680717243e-05, "loss": 0.0487, "step": 929 }, { "epoch": 0.68, "learning_rate": 1.5313110778063142e-05, "loss": 0.0364, "step": 930 }, { "epoch": 0.69, "learning_rate": 1.530301032869745e-05, "loss": 0.0554, "step": 931 }, { "epoch": 0.69, "learning_rate": 1.5292902346966782e-05, "loss": 0.0432, "step": 932 }, { "epoch": 0.69, "learning_rate": 1.5282786847228463e-05, "loss": 0.053, "step": 933 }, { "epoch": 0.69, "learning_rate": 1.5272663843850484e-05, "loss": 0.0424, "step": 934 }, { "epoch": 0.69, "learning_rate": 1.5262533351211507e-05, "loss": 0.0467, "step": 935 }, { "epoch": 0.69, "learning_rate": 1.5252395383700815e-05, "loss": 0.039, "step": 936 }, { "epoch": 0.69, "learning_rate": 1.5242249955718325e-05, "loss": 0.0527, "step": 937 }, { "epoch": 0.69, "learning_rate": 1.5232097081674542e-05, "loss": 0.0467, "step": 938 }, { "epoch": 0.69, "learning_rate": 1.5221936775990548e-05, "loss": 0.0425, "step": 939 }, { "epoch": 0.69, "learning_rate": 1.5211769053097984e-05, "loss": 0.0605, "step": 940 }, { "epoch": 0.69, "learning_rate": 1.5201593927439025e-05, "loss": 0.0547, "step": 941 }, { "epoch": 0.69, "learning_rate": 1.5191411413466357e-05, "loss": 0.0349, "step": 942 }, { "epoch": 0.69, "learning_rate": 1.5181221525643165e-05, "loss": 0.0508, "step": 943 }, { "epoch": 0.69, "learning_rate": 1.5171024278443108e-05, "loss": 0.0488, "step": 944 }, { "epoch": 0.7, "learning_rate": 1.5160819686350295e-05, "loss": 0.0287, "step": 945 }, { "epoch": 0.7, "learning_rate": 1.5150607763859266e-05, "loss": 0.0523, "step": 946 }, { "epoch": 0.7, "learning_rate": 1.514038852547498e-05, "loss": 0.0439, "step": 947 }, { "epoch": 0.7, "learning_rate": 1.5130161985712786e-05, "loss": 0.0544, "step": 948 }, { "epoch": 0.7, "learning_rate": 1.5119928159098396e-05, "loss": 0.0341, "step": 949 }, { "epoch": 0.7, "learning_rate": 1.510968706016788e-05, "loss": 0.0354, "step": 950 }, { "epoch": 0.7, "learning_rate": 1.509943870346764e-05, "loss": 0.0365, "step": 951 }, { "epoch": 0.7, "learning_rate": 1.5089183103554372e-05, "loss": 0.039, "step": 952 }, { "epoch": 0.7, "learning_rate": 1.5078920274995078e-05, "loss": 0.0471, "step": 953 }, { "epoch": 0.7, "learning_rate": 1.506865023236702e-05, "loss": 0.0502, "step": 954 }, { "epoch": 0.7, "learning_rate": 1.50583729902577e-05, "loss": 0.046, "step": 955 }, { "epoch": 0.7, "learning_rate": 1.504808856326486e-05, "loss": 0.0394, "step": 956 }, { "epoch": 0.7, "learning_rate": 1.5037796965996437e-05, "loss": 0.053, "step": 957 }, { "epoch": 0.7, "learning_rate": 1.5027498213070558e-05, "loss": 0.0424, "step": 958 }, { "epoch": 0.71, "learning_rate": 1.5017192319115506e-05, "loss": 0.0551, "step": 959 }, { "epoch": 0.71, "learning_rate": 1.5006879298769721e-05, "loss": 0.0456, "step": 960 }, { "epoch": 0.71, "learning_rate": 1.499655916668175e-05, "loss": 0.0279, "step": 961 }, { "epoch": 0.71, "learning_rate": 1.498623193751025e-05, "loss": 0.0455, "step": 962 }, { "epoch": 0.71, "learning_rate": 1.4975897625923962e-05, "loss": 0.0338, "step": 963 }, { "epoch": 0.71, "learning_rate": 1.4965556246601677e-05, "loss": 0.0422, "step": 964 }, { "epoch": 0.71, "learning_rate": 1.495520781423223e-05, "loss": 0.0492, "step": 965 }, { "epoch": 0.71, "learning_rate": 1.4944852343514478e-05, "loss": 0.0366, "step": 966 }, { "epoch": 0.71, "learning_rate": 1.4934489849157266e-05, "loss": 0.0275, "step": 967 }, { "epoch": 0.71, "learning_rate": 1.4924120345879422e-05, "loss": 0.0568, "step": 968 }, { "epoch": 0.71, "learning_rate": 1.4913743848409732e-05, "loss": 0.0509, "step": 969 }, { "epoch": 0.71, "learning_rate": 1.490336037148691e-05, "loss": 0.0439, "step": 970 }, { "epoch": 0.71, "learning_rate": 1.4892969929859583e-05, "loss": 0.0549, "step": 971 }, { "epoch": 0.72, "learning_rate": 1.4882572538286279e-05, "loss": 0.0475, "step": 972 }, { "epoch": 0.72, "learning_rate": 1.487216821153539e-05, "loss": 0.0552, "step": 973 }, { "epoch": 0.72, "learning_rate": 1.486175696438516e-05, "loss": 0.0391, "step": 974 }, { "epoch": 0.72, "learning_rate": 1.4851338811623663e-05, "loss": 0.0388, "step": 975 }, { "epoch": 0.72, "learning_rate": 1.4840913768048788e-05, "loss": 0.0319, "step": 976 }, { "epoch": 0.72, "learning_rate": 1.4830481848468198e-05, "loss": 0.0477, "step": 977 }, { "epoch": 0.72, "learning_rate": 1.4820043067699342e-05, "loss": 0.033, "step": 978 }, { "epoch": 0.72, "learning_rate": 1.4809597440569391e-05, "loss": 0.0479, "step": 979 }, { "epoch": 0.72, "learning_rate": 1.479914498191526e-05, "loss": 0.0483, "step": 980 }, { "epoch": 0.72, "learning_rate": 1.4788685706583562e-05, "loss": 0.036, "step": 981 }, { "epoch": 0.72, "learning_rate": 1.4778219629430585e-05, "loss": 0.0529, "step": 982 }, { "epoch": 0.72, "learning_rate": 1.4767746765322288e-05, "loss": 0.0386, "step": 983 }, { "epoch": 0.72, "learning_rate": 1.4757267129134266e-05, "loss": 0.0359, "step": 984 }, { "epoch": 0.72, "learning_rate": 1.4746780735751728e-05, "loss": 0.0389, "step": 985 }, { "epoch": 0.73, "learning_rate": 1.4736287600069493e-05, "loss": 0.0478, "step": 986 }, { "epoch": 0.73, "learning_rate": 1.4725787736991946e-05, "loss": 0.052, "step": 987 }, { "epoch": 0.73, "learning_rate": 1.4715281161433032e-05, "loss": 0.0326, "step": 988 }, { "epoch": 0.73, "learning_rate": 1.4704767888316229e-05, "loss": 0.0647, "step": 989 }, { "epoch": 0.73, "learning_rate": 1.4694247932574533e-05, "loss": 0.0461, "step": 990 }, { "epoch": 0.73, "learning_rate": 1.4683721309150424e-05, "loss": 0.0504, "step": 991 }, { "epoch": 0.73, "learning_rate": 1.4673188032995858e-05, "loss": 0.0434, "step": 992 }, { "epoch": 0.73, "learning_rate": 1.4662648119072236e-05, "loss": 0.0293, "step": 993 }, { "epoch": 0.73, "learning_rate": 1.4652101582350394e-05, "loss": 0.0515, "step": 994 }, { "epoch": 0.73, "learning_rate": 1.464154843781057e-05, "loss": 0.0523, "step": 995 }, { "epoch": 0.73, "learning_rate": 1.4630988700442386e-05, "loss": 0.0343, "step": 996 }, { "epoch": 0.73, "learning_rate": 1.4620422385244837e-05, "loss": 0.0559, "step": 997 }, { "epoch": 0.73, "learning_rate": 1.4609849507226254e-05, "loss": 0.0556, "step": 998 }, { "epoch": 0.74, "learning_rate": 1.4599270081404287e-05, "loss": 0.0497, "step": 999 }, { "epoch": 0.74, "learning_rate": 1.4588684122805895e-05, "loss": 0.049, "step": 1000 }, { "epoch": 0.74, "learning_rate": 1.4578091646467316e-05, "loss": 0.0489, "step": 1001 }, { "epoch": 0.74, "learning_rate": 1.4567492667434031e-05, "loss": 0.044, "step": 1002 }, { "epoch": 0.74, "learning_rate": 1.4556887200760777e-05, "loss": 0.0397, "step": 1003 }, { "epoch": 0.74, "learning_rate": 1.4546275261511493e-05, "loss": 0.0552, "step": 1004 }, { "epoch": 0.74, "learning_rate": 1.4535656864759314e-05, "loss": 0.0496, "step": 1005 }, { "epoch": 0.74, "learning_rate": 1.4525032025586555e-05, "loss": 0.0434, "step": 1006 }, { "epoch": 0.74, "learning_rate": 1.451440075908467e-05, "loss": 0.0401, "step": 1007 }, { "epoch": 0.74, "learning_rate": 1.450376308035425e-05, "loss": 0.0396, "step": 1008 }, { "epoch": 0.74, "learning_rate": 1.4493119004504993e-05, "loss": 0.0408, "step": 1009 }, { "epoch": 0.74, "learning_rate": 1.4482468546655679e-05, "loss": 0.0496, "step": 1010 }, { "epoch": 0.74, "learning_rate": 1.447181172193416e-05, "loss": 0.033, "step": 1011 }, { "epoch": 0.74, "learning_rate": 1.4461148545477328e-05, "loss": 0.0546, "step": 1012 }, { "epoch": 0.75, "learning_rate": 1.4450479032431093e-05, "loss": 0.0413, "step": 1013 }, { "epoch": 0.75, "learning_rate": 1.443980319795037e-05, "loss": 0.0471, "step": 1014 }, { "epoch": 0.75, "learning_rate": 1.4429121057199055e-05, "loss": 0.0527, "step": 1015 }, { "epoch": 0.75, "learning_rate": 1.4418432625349997e-05, "loss": 0.0508, "step": 1016 }, { "epoch": 0.75, "learning_rate": 1.440773791758498e-05, "loss": 0.0324, "step": 1017 }, { "epoch": 0.75, "learning_rate": 1.439703694909471e-05, "loss": 0.0515, "step": 1018 }, { "epoch": 0.75, "learning_rate": 1.4386329735078774e-05, "loss": 0.0452, "step": 1019 }, { "epoch": 0.75, "learning_rate": 1.437561629074564e-05, "loss": 0.0343, "step": 1020 }, { "epoch": 0.75, "learning_rate": 1.4364896631312622e-05, "loss": 0.0354, "step": 1021 }, { "epoch": 0.75, "learning_rate": 1.4354170772005862e-05, "loss": 0.0491, "step": 1022 }, { "epoch": 0.75, "learning_rate": 1.4343438728060303e-05, "loss": 0.0429, "step": 1023 }, { "epoch": 0.75, "learning_rate": 1.4332700514719687e-05, "loss": 0.0502, "step": 1024 }, { "epoch": 0.75, "learning_rate": 1.4321956147236502e-05, "loss": 0.0363, "step": 1025 }, { "epoch": 0.75, "learning_rate": 1.4311205640871985e-05, "loss": 0.0377, "step": 1026 }, { "epoch": 0.76, "learning_rate": 1.4300449010896098e-05, "loss": 0.0512, "step": 1027 }, { "epoch": 0.76, "learning_rate": 1.4289686272587493e-05, "loss": 0.0517, "step": 1028 }, { "epoch": 0.76, "learning_rate": 1.4278917441233497e-05, "loss": 0.0453, "step": 1029 }, { "epoch": 0.76, "learning_rate": 1.4268142532130102e-05, "loss": 0.0504, "step": 1030 }, { "epoch": 0.76, "learning_rate": 1.4257361560581924e-05, "loss": 0.0493, "step": 1031 }, { "epoch": 0.76, "learning_rate": 1.4246574541902187e-05, "loss": 0.047, "step": 1032 }, { "epoch": 0.76, "learning_rate": 1.4235781491412715e-05, "loss": 0.0382, "step": 1033 }, { "epoch": 0.76, "learning_rate": 1.4224982424443894e-05, "loss": 0.0519, "step": 1034 }, { "epoch": 0.76, "learning_rate": 1.421417735633465e-05, "loss": 0.0515, "step": 1035 }, { "epoch": 0.76, "learning_rate": 1.4203366302432447e-05, "loss": 0.0381, "step": 1036 }, { "epoch": 0.76, "learning_rate": 1.4192549278093237e-05, "loss": 0.0404, "step": 1037 }, { "epoch": 0.76, "learning_rate": 1.4181726298681462e-05, "loss": 0.0428, "step": 1038 }, { "epoch": 0.76, "learning_rate": 1.4170897379570016e-05, "loss": 0.0425, "step": 1039 }, { "epoch": 0.77, "learning_rate": 1.4160062536140235e-05, "loss": 0.0418, "step": 1040 }, { "epoch": 0.77, "learning_rate": 1.4149221783781867e-05, "loss": 0.0172, "step": 1041 }, { "epoch": 0.77, "learning_rate": 1.413837513789305e-05, "loss": 0.0332, "step": 1042 }, { "epoch": 0.77, "learning_rate": 1.4127522613880303e-05, "loss": 0.0397, "step": 1043 }, { "epoch": 0.77, "learning_rate": 1.4116664227158481e-05, "loss": 0.0483, "step": 1044 }, { "epoch": 0.77, "learning_rate": 1.4105799993150778e-05, "loss": 0.0444, "step": 1045 }, { "epoch": 0.77, "learning_rate": 1.4094929927288688e-05, "loss": 0.0482, "step": 1046 }, { "epoch": 0.77, "learning_rate": 1.4084054045011989e-05, "loss": 0.0536, "step": 1047 }, { "epoch": 0.77, "learning_rate": 1.4073172361768715e-05, "loss": 0.0407, "step": 1048 }, { "epoch": 0.77, "learning_rate": 1.4062284893015156e-05, "loss": 0.0537, "step": 1049 }, { "epoch": 0.77, "learning_rate": 1.4051391654215803e-05, "loss": 0.0314, "step": 1050 }, { "epoch": 0.77, "learning_rate": 1.4040492660843347e-05, "loss": 0.0374, "step": 1051 }, { "epoch": 0.77, "learning_rate": 1.402958792837866e-05, "loss": 0.0319, "step": 1052 }, { "epoch": 0.77, "learning_rate": 1.401867747231076e-05, "loss": 0.0445, "step": 1053 }, { "epoch": 0.78, "learning_rate": 1.4007761308136791e-05, "loss": 0.0386, "step": 1054 }, { "epoch": 0.78, "learning_rate": 1.3996839451362012e-05, "loss": 0.0477, "step": 1055 }, { "epoch": 0.78, "learning_rate": 1.3985911917499764e-05, "loss": 0.0452, "step": 1056 }, { "epoch": 0.78, "learning_rate": 1.3974978722071448e-05, "loss": 0.035, "step": 1057 }, { "epoch": 0.78, "learning_rate": 1.3964039880606522e-05, "loss": 0.0443, "step": 1058 }, { "epoch": 0.78, "learning_rate": 1.3953095408642443e-05, "loss": 0.0464, "step": 1059 }, { "epoch": 0.78, "learning_rate": 1.3942145321724678e-05, "loss": 0.051, "step": 1060 }, { "epoch": 0.78, "learning_rate": 1.3931189635406668e-05, "loss": 0.0284, "step": 1061 }, { "epoch": 0.78, "learning_rate": 1.3920228365249807e-05, "loss": 0.0396, "step": 1062 }, { "epoch": 0.78, "learning_rate": 1.3909261526823413e-05, "loss": 0.039, "step": 1063 }, { "epoch": 0.78, "learning_rate": 1.3898289135704726e-05, "loss": 0.0337, "step": 1064 }, { "epoch": 0.78, "learning_rate": 1.3887311207478864e-05, "loss": 0.0455, "step": 1065 }, { "epoch": 0.78, "learning_rate": 1.387632775773881e-05, "loss": 0.0403, "step": 1066 }, { "epoch": 0.79, "learning_rate": 1.3865338802085398e-05, "loss": 0.0485, "step": 1067 }, { "epoch": 0.79, "learning_rate": 1.3854344356127272e-05, "loss": 0.0349, "step": 1068 }, { "epoch": 0.79, "learning_rate": 1.3843344435480877e-05, "loss": 0.0508, "step": 1069 }, { "epoch": 0.79, "learning_rate": 1.3832339055770443e-05, "loss": 0.0528, "step": 1070 }, { "epoch": 0.79, "learning_rate": 1.382132823262794e-05, "loss": 0.0363, "step": 1071 }, { "epoch": 0.79, "learning_rate": 1.3810311981693084e-05, "loss": 0.0432, "step": 1072 }, { "epoch": 0.79, "learning_rate": 1.3799290318613287e-05, "loss": 0.0318, "step": 1073 }, { "epoch": 0.79, "learning_rate": 1.378826325904366e-05, "loss": 0.0473, "step": 1074 }, { "epoch": 0.79, "learning_rate": 1.3777230818646974e-05, "loss": 0.0453, "step": 1075 }, { "epoch": 0.79, "learning_rate": 1.3766193013093637e-05, "loss": 0.0502, "step": 1076 }, { "epoch": 0.79, "learning_rate": 1.375514985806169e-05, "loss": 0.0465, "step": 1077 }, { "epoch": 0.79, "learning_rate": 1.3744101369236765e-05, "loss": 0.0395, "step": 1078 }, { "epoch": 0.79, "learning_rate": 1.3733047562312067e-05, "loss": 0.0527, "step": 1079 }, { "epoch": 0.79, "learning_rate": 1.3721988452988366e-05, "loss": 0.0339, "step": 1080 }, { "epoch": 0.8, "learning_rate": 1.3710924056973952e-05, "loss": 0.04, "step": 1081 }, { "epoch": 0.8, "learning_rate": 1.3699854389984626e-05, "loss": 0.0542, "step": 1082 }, { "epoch": 0.8, "learning_rate": 1.3688779467743685e-05, "loss": 0.0457, "step": 1083 }, { "epoch": 0.8, "learning_rate": 1.367769930598188e-05, "loss": 0.0503, "step": 1084 }, { "epoch": 0.8, "learning_rate": 1.3666613920437408e-05, "loss": 0.0474, "step": 1085 }, { "epoch": 0.8, "learning_rate": 1.3655523326855889e-05, "loss": 0.0456, "step": 1086 }, { "epoch": 0.8, "learning_rate": 1.3644427540990335e-05, "loss": 0.0555, "step": 1087 }, { "epoch": 0.8, "learning_rate": 1.3633326578601133e-05, "loss": 0.0487, "step": 1088 }, { "epoch": 0.8, "learning_rate": 1.3622220455456032e-05, "loss": 0.0326, "step": 1089 }, { "epoch": 0.8, "learning_rate": 1.36111091873301e-05, "loss": 0.0461, "step": 1090 }, { "epoch": 0.8, "learning_rate": 1.3599992790005719e-05, "loss": 0.0354, "step": 1091 }, { "epoch": 0.8, "learning_rate": 1.3588871279272553e-05, "loss": 0.0458, "step": 1092 }, { "epoch": 0.8, "learning_rate": 1.3577744670927536e-05, "loss": 0.0395, "step": 1093 }, { "epoch": 0.81, "learning_rate": 1.356661298077483e-05, "loss": 0.0441, "step": 1094 }, { "epoch": 0.81, "learning_rate": 1.3555476224625828e-05, "loss": 0.0461, "step": 1095 }, { "epoch": 0.81, "learning_rate": 1.3544334418299115e-05, "loss": 0.042, "step": 1096 }, { "epoch": 0.81, "learning_rate": 1.3533187577620441e-05, "loss": 0.05, "step": 1097 }, { "epoch": 0.81, "learning_rate": 1.3522035718422722e-05, "loss": 0.0407, "step": 1098 }, { "epoch": 0.81, "learning_rate": 1.3510878856545987e-05, "loss": 0.0432, "step": 1099 }, { "epoch": 0.81, "learning_rate": 1.3499717007837381e-05, "loss": 0.0392, "step": 1100 }, { "epoch": 0.81, "learning_rate": 1.348855018815113e-05, "loss": 0.0351, "step": 1101 }, { "epoch": 0.81, "learning_rate": 1.3477378413348516e-05, "loss": 0.0462, "step": 1102 }, { "epoch": 0.81, "learning_rate": 1.3466201699297868e-05, "loss": 0.0358, "step": 1103 }, { "epoch": 0.81, "learning_rate": 1.3455020061874517e-05, "loss": 0.0502, "step": 1104 }, { "epoch": 0.81, "learning_rate": 1.3443833516960805e-05, "loss": 0.0382, "step": 1105 }, { "epoch": 0.81, "learning_rate": 1.343264208044603e-05, "loss": 0.028, "step": 1106 }, { "epoch": 0.81, "learning_rate": 1.3421445768226442e-05, "loss": 0.0417, "step": 1107 }, { "epoch": 0.82, "learning_rate": 1.3410244596205222e-05, "loss": 0.0459, "step": 1108 }, { "epoch": 0.82, "learning_rate": 1.3399038580292449e-05, "loss": 0.0422, "step": 1109 }, { "epoch": 0.82, "learning_rate": 1.3387827736405079e-05, "loss": 0.0442, "step": 1110 }, { "epoch": 0.82, "learning_rate": 1.337661208046693e-05, "loss": 0.0345, "step": 1111 }, { "epoch": 0.82, "learning_rate": 1.336539162840866e-05, "loss": 0.0441, "step": 1112 }, { "epoch": 0.82, "learning_rate": 1.3354166396167725e-05, "loss": 0.0505, "step": 1113 }, { "epoch": 0.82, "learning_rate": 1.3342936399688387e-05, "loss": 0.0524, "step": 1114 }, { "epoch": 0.82, "learning_rate": 1.3331701654921663e-05, "loss": 0.0353, "step": 1115 }, { "epoch": 0.82, "learning_rate": 1.3320462177825321e-05, "loss": 0.0395, "step": 1116 }, { "epoch": 0.82, "learning_rate": 1.3309217984363849e-05, "loss": 0.0493, "step": 1117 }, { "epoch": 0.82, "learning_rate": 1.3297969090508434e-05, "loss": 0.0449, "step": 1118 }, { "epoch": 0.82, "learning_rate": 1.3286715512236934e-05, "loss": 0.0399, "step": 1119 }, { "epoch": 0.82, "learning_rate": 1.3275457265533876e-05, "loss": 0.0571, "step": 1120 }, { "epoch": 0.82, "learning_rate": 1.3264194366390401e-05, "loss": 0.0355, "step": 1121 }, { "epoch": 0.83, "learning_rate": 1.3252926830804264e-05, "loss": 0.0485, "step": 1122 }, { "epoch": 0.83, "learning_rate": 1.3241654674779814e-05, "loss": 0.0354, "step": 1123 }, { "epoch": 0.83, "learning_rate": 1.323037791432795e-05, "loss": 0.0487, "step": 1124 }, { "epoch": 0.83, "learning_rate": 1.3219096565466117e-05, "loss": 0.0507, "step": 1125 }, { "epoch": 0.83, "learning_rate": 1.3207810644218277e-05, "loss": 0.0274, "step": 1126 }, { "epoch": 0.83, "learning_rate": 1.3196520166614892e-05, "loss": 0.0375, "step": 1127 }, { "epoch": 0.83, "learning_rate": 1.3185225148692884e-05, "loss": 0.0422, "step": 1128 }, { "epoch": 0.83, "learning_rate": 1.3173925606495637e-05, "loss": 0.0541, "step": 1129 }, { "epoch": 0.83, "learning_rate": 1.316262155607295e-05, "loss": 0.0407, "step": 1130 }, { "epoch": 0.83, "learning_rate": 1.315131301348103e-05, "loss": 0.042, "step": 1131 }, { "epoch": 0.83, "learning_rate": 1.3139999994782468e-05, "loss": 0.0466, "step": 1132 }, { "epoch": 0.83, "learning_rate": 1.3128682516046211e-05, "loss": 0.0456, "step": 1133 }, { "epoch": 0.83, "learning_rate": 1.3117360593347535e-05, "loss": 0.032, "step": 1134 }, { "epoch": 0.84, "learning_rate": 1.3106034242768036e-05, "loss": 0.0552, "step": 1135 }, { "epoch": 0.84, "learning_rate": 1.3094703480395597e-05, "loss": 0.033, "step": 1136 }, { "epoch": 0.84, "learning_rate": 1.3083368322324363e-05, "loss": 0.0339, "step": 1137 }, { "epoch": 0.84, "learning_rate": 1.3072028784654732e-05, "loss": 0.043, "step": 1138 }, { "epoch": 0.84, "learning_rate": 1.3060684883493312e-05, "loss": 0.0351, "step": 1139 }, { "epoch": 0.84, "learning_rate": 1.3049336634952918e-05, "loss": 0.0424, "step": 1140 }, { "epoch": 0.84, "learning_rate": 1.3037984055152535e-05, "loss": 0.0417, "step": 1141 }, { "epoch": 0.84, "learning_rate": 1.3026627160217302e-05, "loss": 0.0459, "step": 1142 }, { "epoch": 0.84, "learning_rate": 1.3015265966278483e-05, "loss": 0.0424, "step": 1143 }, { "epoch": 0.84, "learning_rate": 1.3003900489473455e-05, "loss": 0.0534, "step": 1144 }, { "epoch": 0.84, "learning_rate": 1.2992530745945677e-05, "loss": 0.0374, "step": 1145 }, { "epoch": 0.84, "learning_rate": 1.2981156751844659e-05, "loss": 0.0435, "step": 1146 }, { "epoch": 0.84, "learning_rate": 1.2969778523325963e-05, "loss": 0.0366, "step": 1147 }, { "epoch": 0.84, "learning_rate": 1.2958396076551157e-05, "loss": 0.043, "step": 1148 }, { "epoch": 0.85, "learning_rate": 1.2947009427687795e-05, "loss": 0.0438, "step": 1149 }, { "epoch": 0.85, "learning_rate": 1.2935618592909419e-05, "loss": 0.0425, "step": 1150 }, { "epoch": 0.85, "learning_rate": 1.2924223588395494e-05, "loss": 0.0326, "step": 1151 }, { "epoch": 0.85, "learning_rate": 1.2912824430331425e-05, "loss": 0.0455, "step": 1152 }, { "epoch": 0.85, "learning_rate": 1.2901421134908505e-05, "loss": 0.0336, "step": 1153 }, { "epoch": 0.85, "learning_rate": 1.2890013718323913e-05, "loss": 0.0365, "step": 1154 }, { "epoch": 0.85, "learning_rate": 1.2878602196780671e-05, "loss": 0.039, "step": 1155 }, { "epoch": 0.85, "learning_rate": 1.2867186586487642e-05, "loss": 0.0412, "step": 1156 }, { "epoch": 0.85, "learning_rate": 1.2855766903659492e-05, "loss": 0.06, "step": 1157 }, { "epoch": 0.85, "learning_rate": 1.2844343164516671e-05, "loss": 0.0393, "step": 1158 }, { "epoch": 0.85, "learning_rate": 1.2832915385285389e-05, "loss": 0.0443, "step": 1159 }, { "epoch": 0.85, "learning_rate": 1.2821483582197604e-05, "loss": 0.0323, "step": 1160 }, { "epoch": 0.85, "learning_rate": 1.2810047771490976e-05, "loss": 0.0341, "step": 1161 }, { "epoch": 0.86, "learning_rate": 1.2798607969408865e-05, "loss": 0.0432, "step": 1162 }, { "epoch": 0.86, "learning_rate": 1.2787164192200304e-05, "loss": 0.0361, "step": 1163 }, { "epoch": 0.86, "learning_rate": 1.2775716456119962e-05, "loss": 0.0397, "step": 1164 }, { "epoch": 0.86, "learning_rate": 1.2764264777428137e-05, "loss": 0.0375, "step": 1165 }, { "epoch": 0.86, "learning_rate": 1.275280917239073e-05, "loss": 0.0343, "step": 1166 }, { "epoch": 0.86, "learning_rate": 1.2741349657279214e-05, "loss": 0.0335, "step": 1167 }, { "epoch": 0.86, "learning_rate": 1.2729886248370616e-05, "loss": 0.0503, "step": 1168 }, { "epoch": 0.86, "learning_rate": 1.2718418961947496e-05, "loss": 0.0316, "step": 1169 }, { "epoch": 0.86, "learning_rate": 1.2706947814297923e-05, "loss": 0.0428, "step": 1170 }, { "epoch": 0.86, "learning_rate": 1.2695472821715447e-05, "loss": 0.04, "step": 1171 }, { "epoch": 0.86, "learning_rate": 1.2683994000499078e-05, "loss": 0.0431, "step": 1172 }, { "epoch": 0.86, "learning_rate": 1.2672511366953271e-05, "loss": 0.0384, "step": 1173 }, { "epoch": 0.86, "learning_rate": 1.2661024937387888e-05, "loss": 0.0617, "step": 1174 }, { "epoch": 0.86, "learning_rate": 1.2649534728118191e-05, "loss": 0.0274, "step": 1175 }, { "epoch": 0.87, "learning_rate": 1.2638040755464802e-05, "loss": 0.0456, "step": 1176 }, { "epoch": 0.87, "learning_rate": 1.2626543035753694e-05, "loss": 0.0461, "step": 1177 }, { "epoch": 0.87, "learning_rate": 1.2615041585316163e-05, "loss": 0.0349, "step": 1178 }, { "epoch": 0.87, "learning_rate": 1.2603536420488802e-05, "loss": 0.0382, "step": 1179 }, { "epoch": 0.87, "learning_rate": 1.2592027557613476e-05, "loss": 0.027, "step": 1180 }, { "epoch": 0.87, "learning_rate": 1.2580515013037311e-05, "loss": 0.0482, "step": 1181 }, { "epoch": 0.87, "learning_rate": 1.2568998803112658e-05, "loss": 0.0451, "step": 1182 }, { "epoch": 0.87, "learning_rate": 1.255747894419707e-05, "loss": 0.0571, "step": 1183 }, { "epoch": 0.87, "learning_rate": 1.2545955452653294e-05, "loss": 0.0394, "step": 1184 }, { "epoch": 0.87, "learning_rate": 1.2534428344849227e-05, "loss": 0.0431, "step": 1185 }, { "epoch": 0.87, "learning_rate": 1.2522897637157905e-05, "loss": 0.04, "step": 1186 }, { "epoch": 0.87, "learning_rate": 1.2511363345957479e-05, "loss": 0.035, "step": 1187 }, { "epoch": 0.87, "learning_rate": 1.249982548763119e-05, "loss": 0.0615, "step": 1188 }, { "epoch": 0.87, "learning_rate": 1.2488284078567339e-05, "loss": 0.0396, "step": 1189 }, { "epoch": 0.88, "learning_rate": 1.2476739135159286e-05, "loss": 0.0358, "step": 1190 }, { "epoch": 0.88, "learning_rate": 1.2465190673805397e-05, "loss": 0.0593, "step": 1191 }, { "epoch": 0.88, "learning_rate": 1.2453638710909033e-05, "loss": 0.0376, "step": 1192 }, { "epoch": 0.88, "learning_rate": 1.2442083262878543e-05, "loss": 0.0416, "step": 1193 }, { "epoch": 0.88, "learning_rate": 1.2430524346127215e-05, "loss": 0.037, "step": 1194 }, { "epoch": 0.88, "learning_rate": 1.2418961977073268e-05, "loss": 0.0361, "step": 1195 }, { "epoch": 0.88, "learning_rate": 1.2407396172139822e-05, "loss": 0.0545, "step": 1196 }, { "epoch": 0.88, "learning_rate": 1.239582694775488e-05, "loss": 0.033, "step": 1197 }, { "epoch": 0.88, "learning_rate": 1.2384254320351301e-05, "loss": 0.0318, "step": 1198 }, { "epoch": 0.88, "learning_rate": 1.2372678306366778e-05, "loss": 0.0463, "step": 1199 }, { "epoch": 0.88, "learning_rate": 1.2361098922243812e-05, "loss": 0.0334, "step": 1200 }, { "epoch": 0.88, "learning_rate": 1.2349516184429698e-05, "loss": 0.0461, "step": 1201 }, { "epoch": 0.88, "learning_rate": 1.233793010937648e-05, "loss": 0.06, "step": 1202 }, { "epoch": 0.89, "learning_rate": 1.2326340713540955e-05, "loss": 0.052, "step": 1203 }, { "epoch": 0.89, "learning_rate": 1.2314748013384639e-05, "loss": 0.0421, "step": 1204 }, { "epoch": 0.89, "learning_rate": 1.2303152025373727e-05, "loss": 0.0381, "step": 1205 }, { "epoch": 0.89, "learning_rate": 1.2291552765979104e-05, "loss": 0.0345, "step": 1206 }, { "epoch": 0.89, "learning_rate": 1.2279950251676283e-05, "loss": 0.0484, "step": 1207 }, { "epoch": 0.89, "learning_rate": 1.2268344498945404e-05, "loss": 0.0374, "step": 1208 }, { "epoch": 0.89, "learning_rate": 1.2256735524271219e-05, "loss": 0.0317, "step": 1209 }, { "epoch": 0.89, "learning_rate": 1.2245123344143044e-05, "loss": 0.0415, "step": 1210 }, { "epoch": 0.89, "learning_rate": 1.223350797505475e-05, "loss": 0.0557, "step": 1211 }, { "epoch": 0.89, "learning_rate": 1.2221889433504743e-05, "loss": 0.0412, "step": 1212 }, { "epoch": 0.89, "learning_rate": 1.2210267735995934e-05, "loss": 0.0364, "step": 1213 }, { "epoch": 0.89, "learning_rate": 1.2198642899035704e-05, "loss": 0.042, "step": 1214 }, { "epoch": 0.89, "learning_rate": 1.2187014939135917e-05, "loss": 0.0355, "step": 1215 }, { "epoch": 0.89, "learning_rate": 1.2175383872812851e-05, "loss": 0.0353, "step": 1216 }, { "epoch": 0.9, "learning_rate": 1.2163749716587208e-05, "loss": 0.0582, "step": 1217 }, { "epoch": 0.9, "learning_rate": 1.2152112486984071e-05, "loss": 0.0395, "step": 1218 }, { "epoch": 0.9, "learning_rate": 1.21404722005329e-05, "loss": 0.0422, "step": 1219 }, { "epoch": 0.9, "learning_rate": 1.2128828873767487e-05, "loss": 0.0363, "step": 1220 }, { "epoch": 0.9, "learning_rate": 1.2117182523225944e-05, "loss": 0.0317, "step": 1221 }, { "epoch": 0.9, "learning_rate": 1.210553316545068e-05, "loss": 0.0323, "step": 1222 }, { "epoch": 0.9, "learning_rate": 1.2093880816988377e-05, "loss": 0.0493, "step": 1223 }, { "epoch": 0.9, "learning_rate": 1.2082225494389961e-05, "loss": 0.0285, "step": 1224 }, { "epoch": 0.9, "learning_rate": 1.2070567214210586e-05, "loss": 0.0563, "step": 1225 }, { "epoch": 0.9, "learning_rate": 1.2058905993009604e-05, "loss": 0.0439, "step": 1226 }, { "epoch": 0.9, "learning_rate": 1.2047241847350543e-05, "loss": 0.042, "step": 1227 }, { "epoch": 0.9, "learning_rate": 1.2035574793801095e-05, "loss": 0.052, "step": 1228 }, { "epoch": 0.9, "learning_rate": 1.2023904848933069e-05, "loss": 0.0404, "step": 1229 }, { "epoch": 0.91, "learning_rate": 1.2012232029322384e-05, "loss": 0.0312, "step": 1230 }, { "epoch": 0.91, "learning_rate": 1.200055635154905e-05, "loss": 0.0419, "step": 1231 }, { "epoch": 0.91, "learning_rate": 1.1988877832197135e-05, "loss": 0.0476, "step": 1232 }, { "epoch": 0.91, "learning_rate": 1.197719648785473e-05, "loss": 0.0464, "step": 1233 }, { "epoch": 0.91, "learning_rate": 1.1965512335113958e-05, "loss": 0.0376, "step": 1234 }, { "epoch": 0.91, "learning_rate": 1.1953825390570916e-05, "loss": 0.0488, "step": 1235 }, { "epoch": 0.91, "learning_rate": 1.1942135670825672e-05, "loss": 0.04, "step": 1236 }, { "epoch": 0.91, "learning_rate": 1.1930443192482238e-05, "loss": 0.0401, "step": 1237 }, { "epoch": 0.91, "learning_rate": 1.1918747972148541e-05, "loss": 0.0441, "step": 1238 }, { "epoch": 0.91, "learning_rate": 1.1907050026436405e-05, "loss": 0.0502, "step": 1239 }, { "epoch": 0.91, "learning_rate": 1.189534937196152e-05, "loss": 0.0394, "step": 1240 }, { "epoch": 0.91, "learning_rate": 1.1883646025343434e-05, "loss": 0.0411, "step": 1241 }, { "epoch": 0.91, "learning_rate": 1.1871940003205505e-05, "loss": 0.037, "step": 1242 }, { "epoch": 0.91, "learning_rate": 1.1860231322174904e-05, "loss": 0.0511, "step": 1243 }, { "epoch": 0.92, "learning_rate": 1.1848519998882572e-05, "loss": 0.0317, "step": 1244 }, { "epoch": 0.92, "learning_rate": 1.1836806049963205e-05, "loss": 0.0411, "step": 1245 }, { "epoch": 0.92, "learning_rate": 1.1825089492055227e-05, "loss": 0.0375, "step": 1246 }, { "epoch": 0.92, "learning_rate": 1.1813370341800766e-05, "loss": 0.0531, "step": 1247 }, { "epoch": 0.92, "learning_rate": 1.1801648615845638e-05, "loss": 0.044, "step": 1248 }, { "epoch": 0.92, "learning_rate": 1.1789924330839315e-05, "loss": 0.0329, "step": 1249 }, { "epoch": 0.92, "learning_rate": 1.17781975034349e-05, "loss": 0.0456, "step": 1250 }, { "epoch": 0.92, "learning_rate": 1.1766468150289109e-05, "loss": 0.04, "step": 1251 }, { "epoch": 0.92, "learning_rate": 1.1754736288062256e-05, "loss": 0.0522, "step": 1252 }, { "epoch": 0.92, "learning_rate": 1.17430019334182e-05, "loss": 0.0475, "step": 1253 }, { "epoch": 0.92, "learning_rate": 1.1731265103024351e-05, "loss": 0.0374, "step": 1254 }, { "epoch": 0.92, "learning_rate": 1.1719525813551643e-05, "loss": 0.035, "step": 1255 }, { "epoch": 0.92, "learning_rate": 1.1707784081674482e-05, "loss": 0.0443, "step": 1256 }, { "epoch": 0.92, "learning_rate": 1.1696039924070766e-05, "loss": 0.0445, "step": 1257 }, { "epoch": 0.93, "learning_rate": 1.1684293357421824e-05, "loss": 0.0468, "step": 1258 }, { "epoch": 0.93, "learning_rate": 1.1672544398412408e-05, "loss": 0.0337, "step": 1259 }, { "epoch": 0.93, "learning_rate": 1.1660793063730674e-05, "loss": 0.0475, "step": 1260 }, { "epoch": 0.93, "learning_rate": 1.1649039370068149e-05, "loss": 0.0507, "step": 1261 }, { "epoch": 0.93, "learning_rate": 1.1637283334119713e-05, "loss": 0.0457, "step": 1262 }, { "epoch": 0.93, "learning_rate": 1.1625524972583565e-05, "loss": 0.0362, "step": 1263 }, { "epoch": 0.93, "learning_rate": 1.1613764302161222e-05, "loss": 0.035, "step": 1264 }, { "epoch": 0.93, "learning_rate": 1.1602001339557464e-05, "loss": 0.0622, "step": 1265 }, { "epoch": 0.93, "learning_rate": 1.1590236101480339e-05, "loss": 0.0455, "step": 1266 }, { "epoch": 0.93, "learning_rate": 1.1578468604641117e-05, "loss": 0.0485, "step": 1267 }, { "epoch": 0.93, "learning_rate": 1.1566698865754291e-05, "loss": 0.0446, "step": 1268 }, { "epoch": 0.93, "learning_rate": 1.1554926901537521e-05, "loss": 0.0475, "step": 1269 }, { "epoch": 0.93, "learning_rate": 1.154315272871164e-05, "loss": 0.0466, "step": 1270 }, { "epoch": 0.94, "learning_rate": 1.1531376364000617e-05, "loss": 0.0366, "step": 1271 }, { "epoch": 0.94, "learning_rate": 1.1519597824131527e-05, "loss": 0.034, "step": 1272 }, { "epoch": 0.94, "learning_rate": 1.1507817125834542e-05, "loss": 0.0416, "step": 1273 }, { "epoch": 0.94, "learning_rate": 1.1496034285842897e-05, "loss": 0.0331, "step": 1274 }, { "epoch": 0.94, "learning_rate": 1.148424932089287e-05, "loss": 0.0502, "step": 1275 }, { "epoch": 0.94, "learning_rate": 1.1472462247723752e-05, "loss": 0.0357, "step": 1276 }, { "epoch": 0.94, "learning_rate": 1.146067308307784e-05, "loss": 0.0535, "step": 1277 }, { "epoch": 0.94, "learning_rate": 1.1448881843700392e-05, "loss": 0.05, "step": 1278 }, { "epoch": 0.94, "learning_rate": 1.1437088546339612e-05, "loss": 0.0423, "step": 1279 }, { "epoch": 0.94, "learning_rate": 1.1425293207746638e-05, "loss": 0.0315, "step": 1280 }, { "epoch": 0.94, "learning_rate": 1.1413495844675495e-05, "loss": 0.0356, "step": 1281 }, { "epoch": 0.94, "learning_rate": 1.1401696473883086e-05, "loss": 0.0421, "step": 1282 }, { "epoch": 0.94, "learning_rate": 1.1389895112129177e-05, "loss": 0.0346, "step": 1283 }, { "epoch": 0.94, "learning_rate": 1.1378091776176348e-05, "loss": 0.0397, "step": 1284 }, { "epoch": 0.95, "learning_rate": 1.1366286482789988e-05, "loss": 0.0408, "step": 1285 }, { "epoch": 0.95, "learning_rate": 1.1354479248738271e-05, "loss": 0.0498, "step": 1286 }, { "epoch": 0.95, "learning_rate": 1.134267009079212e-05, "loss": 0.0367, "step": 1287 }, { "epoch": 0.95, "learning_rate": 1.1330859025725193e-05, "loss": 0.0473, "step": 1288 }, { "epoch": 0.95, "learning_rate": 1.1319046070313862e-05, "loss": 0.0383, "step": 1289 }, { "epoch": 0.95, "learning_rate": 1.130723124133718e-05, "loss": 0.0455, "step": 1290 }, { "epoch": 0.95, "learning_rate": 1.1295414555576856e-05, "loss": 0.0526, "step": 1291 }, { "epoch": 0.95, "learning_rate": 1.1283596029817248e-05, "loss": 0.0448, "step": 1292 }, { "epoch": 0.95, "learning_rate": 1.1271775680845316e-05, "loss": 0.0473, "step": 1293 }, { "epoch": 0.95, "learning_rate": 1.1259953525450616e-05, "loss": 0.0438, "step": 1294 }, { "epoch": 0.95, "learning_rate": 1.124812958042527e-05, "loss": 0.04, "step": 1295 }, { "epoch": 0.95, "learning_rate": 1.1236303862563945e-05, "loss": 0.0567, "step": 1296 }, { "epoch": 0.95, "learning_rate": 1.122447638866381e-05, "loss": 0.0368, "step": 1297 }, { "epoch": 0.96, "learning_rate": 1.1212647175524551e-05, "loss": 0.0452, "step": 1298 }, { "epoch": 0.96, "learning_rate": 1.1200816239948312e-05, "loss": 0.0439, "step": 1299 }, { "epoch": 0.96, "learning_rate": 1.1188983598739675e-05, "loss": 0.0533, "step": 1300 }, { "epoch": 0.96, "learning_rate": 1.1177149268705667e-05, "loss": 0.0458, "step": 1301 }, { "epoch": 0.96, "learning_rate": 1.1165313266655698e-05, "loss": 0.0499, "step": 1302 }, { "epoch": 0.96, "learning_rate": 1.115347560940155e-05, "loss": 0.0361, "step": 1303 }, { "epoch": 0.96, "learning_rate": 1.1141636313757369e-05, "loss": 0.0323, "step": 1304 }, { "epoch": 0.96, "learning_rate": 1.1129795396539622e-05, "loss": 0.0398, "step": 1305 }, { "epoch": 0.96, "learning_rate": 1.1117952874567073e-05, "loss": 0.0475, "step": 1306 }, { "epoch": 0.96, "learning_rate": 1.1106108764660777e-05, "loss": 0.0451, "step": 1307 }, { "epoch": 0.96, "learning_rate": 1.1094263083644036e-05, "loss": 0.0433, "step": 1308 }, { "epoch": 0.96, "learning_rate": 1.1082415848342384e-05, "loss": 0.0324, "step": 1309 }, { "epoch": 0.96, "learning_rate": 1.1070567075583572e-05, "loss": 0.041, "step": 1310 }, { "epoch": 0.96, "learning_rate": 1.1058716782197522e-05, "loss": 0.0466, "step": 1311 }, { "epoch": 0.97, "learning_rate": 1.1046864985016326e-05, "loss": 0.0279, "step": 1312 }, { "epoch": 0.97, "learning_rate": 1.1035011700874199e-05, "loss": 0.0285, "step": 1313 }, { "epoch": 0.97, "learning_rate": 1.1023156946607485e-05, "loss": 0.0359, "step": 1314 }, { "epoch": 0.97, "learning_rate": 1.1011300739054608e-05, "loss": 0.0453, "step": 1315 }, { "epoch": 0.97, "learning_rate": 1.0999443095056051e-05, "loss": 0.0357, "step": 1316 }, { "epoch": 0.97, "learning_rate": 1.0987584031454346e-05, "loss": 0.0312, "step": 1317 }, { "epoch": 0.97, "learning_rate": 1.0975723565094036e-05, "loss": 0.054, "step": 1318 }, { "epoch": 0.97, "learning_rate": 1.0963861712821658e-05, "loss": 0.0418, "step": 1319 }, { "epoch": 0.97, "learning_rate": 1.0951998491485722e-05, "loss": 0.0325, "step": 1320 }, { "epoch": 0.97, "learning_rate": 1.0940133917936675e-05, "loss": 0.0423, "step": 1321 }, { "epoch": 0.97, "learning_rate": 1.0928268009026885e-05, "loss": 0.0386, "step": 1322 }, { "epoch": 0.97, "learning_rate": 1.0916400781610627e-05, "loss": 0.0311, "step": 1323 }, { "epoch": 0.97, "learning_rate": 1.090453225254404e-05, "loss": 0.0366, "step": 1324 }, { "epoch": 0.97, "learning_rate": 1.0892662438685108e-05, "loss": 0.0293, "step": 1325 }, { "epoch": 0.98, "learning_rate": 1.0880791356893652e-05, "loss": 0.041, "step": 1326 }, { "epoch": 0.98, "learning_rate": 1.0868919024031284e-05, "loss": 0.0449, "step": 1327 }, { "epoch": 0.98, "learning_rate": 1.0857045456961394e-05, "loss": 0.0441, "step": 1328 }, { "epoch": 0.98, "learning_rate": 1.0845170672549134e-05, "loss": 0.0353, "step": 1329 }, { "epoch": 0.98, "learning_rate": 1.0833294687661376e-05, "loss": 0.0405, "step": 1330 }, { "epoch": 0.98, "learning_rate": 1.0821417519166694e-05, "loss": 0.0466, "step": 1331 }, { "epoch": 0.98, "learning_rate": 1.0809539183935358e-05, "loss": 0.034, "step": 1332 }, { "epoch": 0.98, "learning_rate": 1.079765969883928e-05, "loss": 0.0458, "step": 1333 }, { "epoch": 0.98, "learning_rate": 1.0785779080752012e-05, "loss": 0.0465, "step": 1334 }, { "epoch": 0.98, "learning_rate": 1.0773897346548718e-05, "loss": 0.0557, "step": 1335 }, { "epoch": 0.98, "learning_rate": 1.0762014513106143e-05, "loss": 0.0297, "step": 1336 }, { "epoch": 0.98, "learning_rate": 1.0750130597302592e-05, "loss": 0.039, "step": 1337 }, { "epoch": 0.98, "learning_rate": 1.073824561601791e-05, "loss": 0.0386, "step": 1338 }, { "epoch": 0.99, "learning_rate": 1.072635958613346e-05, "loss": 0.0288, "step": 1339 }, { "epoch": 0.99, "learning_rate": 1.0714472524532085e-05, "loss": 0.0418, "step": 1340 }, { "epoch": 0.99, "learning_rate": 1.0702584448098096e-05, "loss": 0.0335, "step": 1341 }, { "epoch": 0.99, "learning_rate": 1.0690695373717254e-05, "loss": 0.0233, "step": 1342 }, { "epoch": 0.99, "learning_rate": 1.0678805318276727e-05, "loss": 0.0405, "step": 1343 }, { "epoch": 0.99, "learning_rate": 1.0666914298665079e-05, "loss": 0.0525, "step": 1344 }, { "epoch": 0.99, "learning_rate": 1.0655022331772249e-05, "loss": 0.0408, "step": 1345 }, { "epoch": 0.99, "learning_rate": 1.0643129434489514e-05, "loss": 0.0358, "step": 1346 }, { "epoch": 0.99, "learning_rate": 1.0631235623709477e-05, "loss": 0.0306, "step": 1347 }, { "epoch": 0.99, "learning_rate": 1.0619340916326039e-05, "loss": 0.0437, "step": 1348 }, { "epoch": 0.99, "learning_rate": 1.0607445329234374e-05, "loss": 0.0377, "step": 1349 }, { "epoch": 0.99, "learning_rate": 1.05955488793309e-05, "loss": 0.0515, "step": 1350 }, { "epoch": 0.99, "learning_rate": 1.058365158351327e-05, "loss": 0.0355, "step": 1351 }, { "epoch": 0.99, "learning_rate": 1.0571753458680329e-05, "loss": 0.0369, "step": 1352 }, { "epoch": 1.0, "learning_rate": 1.0559854521732107e-05, "loss": 0.0396, "step": 1353 }, { "epoch": 1.0, "learning_rate": 1.0547954789569785e-05, "loss": 0.0377, "step": 1354 }, { "epoch": 1.0, "learning_rate": 1.0536054279095674e-05, "loss": 0.0479, "step": 1355 }, { "epoch": 1.0, "learning_rate": 1.0524153007213185e-05, "loss": 0.0445, "step": 1356 }, { "epoch": 1.0, "learning_rate": 1.0512250990826821e-05, "loss": 0.0445, "step": 1357 }, { "epoch": 1.0, "learning_rate": 1.0500348246842136e-05, "loss": 0.0448, "step": 1358 }, { "epoch": 1.0, "learning_rate": 1.0488444792165714e-05, "loss": 0.0461, "step": 1359 }, { "epoch": 1.0, "learning_rate": 1.0476540643705153e-05, "loss": 0.0331, "step": 1360 }, { "epoch": 1.0, "learning_rate": 1.0464635818369043e-05, "loss": 0.0451, "step": 1361 }, { "epoch": 1.0, "learning_rate": 1.0452730333066919e-05, "loss": 0.0407, "step": 1362 }, { "epoch": 1.0, "learning_rate": 1.044082420470927e-05, "loss": 0.0408, "step": 1363 }, { "epoch": 1.0, "learning_rate": 1.0428917450207489e-05, "loss": 0.0316, "step": 1364 }, { "epoch": 1.0, "learning_rate": 1.0417010086473862e-05, "loss": 0.0426, "step": 1365 }, { "epoch": 1.01, "learning_rate": 1.0405102130421536e-05, "loss": 0.0308, "step": 1366 }, { "epoch": 1.01, "learning_rate": 1.0393193598964507e-05, "loss": 0.029, "step": 1367 }, { "epoch": 1.01, "learning_rate": 1.0381284509017578e-05, "loss": 0.0305, "step": 1368 }, { "epoch": 1.01, "learning_rate": 1.0369374877496355e-05, "loss": 0.038, "step": 1369 }, { "epoch": 1.01, "learning_rate": 1.035746472131721e-05, "loss": 0.0334, "step": 1370 }, { "epoch": 1.01, "learning_rate": 1.0345554057397254e-05, "loss": 0.0347, "step": 1371 }, { "epoch": 1.01, "learning_rate": 1.033364290265433e-05, "loss": 0.0401, "step": 1372 }, { "epoch": 1.01, "learning_rate": 1.0321731274006972e-05, "loss": 0.0364, "step": 1373 }, { "epoch": 1.01, "learning_rate": 1.0309819188374386e-05, "loss": 0.0324, "step": 1374 }, { "epoch": 1.01, "learning_rate": 1.029790666267643e-05, "loss": 0.0301, "step": 1375 }, { "epoch": 1.01, "learning_rate": 1.0285993713833586e-05, "loss": 0.0283, "step": 1376 }, { "epoch": 1.01, "learning_rate": 1.027408035876694e-05, "loss": 0.0346, "step": 1377 }, { "epoch": 1.01, "learning_rate": 1.0262166614398144e-05, "loss": 0.0328, "step": 1378 }, { "epoch": 1.01, "learning_rate": 1.025025249764942e-05, "loss": 0.036, "step": 1379 }, { "epoch": 1.02, "learning_rate": 1.0238338025443507e-05, "loss": 0.0327, "step": 1380 }, { "epoch": 1.02, "learning_rate": 1.0226423214703649e-05, "loss": 0.0273, "step": 1381 }, { "epoch": 1.02, "learning_rate": 1.0214508082353578e-05, "loss": 0.044, "step": 1382 }, { "epoch": 1.02, "learning_rate": 1.0202592645317475e-05, "loss": 0.035, "step": 1383 }, { "epoch": 1.02, "learning_rate": 1.019067692051996e-05, "loss": 0.0281, "step": 1384 }, { "epoch": 1.02, "learning_rate": 1.0178760924886058e-05, "loss": 0.0263, "step": 1385 }, { "epoch": 1.02, "learning_rate": 1.016684467534118e-05, "loss": 0.0361, "step": 1386 }, { "epoch": 1.02, "learning_rate": 1.0154928188811094e-05, "loss": 0.0323, "step": 1387 }, { "epoch": 1.02, "learning_rate": 1.0143011482221916e-05, "loss": 0.0368, "step": 1388 }, { "epoch": 1.02, "learning_rate": 1.0131094572500063e-05, "loss": 0.0396, "step": 1389 }, { "epoch": 1.02, "learning_rate": 1.0119177476572237e-05, "loss": 0.045, "step": 1390 }, { "epoch": 1.02, "learning_rate": 1.0107260211365421e-05, "loss": 0.0345, "step": 1391 }, { "epoch": 1.02, "learning_rate": 1.0095342793806828e-05, "loss": 0.0319, "step": 1392 }, { "epoch": 1.03, "learning_rate": 1.008342524082388e-05, "loss": 0.0248, "step": 1393 }, { "epoch": 1.03, "learning_rate": 1.0071507569344213e-05, "loss": 0.029, "step": 1394 }, { "epoch": 1.03, "learning_rate": 1.0059589796295609e-05, "loss": 0.0354, "step": 1395 }, { "epoch": 1.03, "learning_rate": 1.0047671938606002e-05, "loss": 0.0376, "step": 1396 }, { "epoch": 1.03, "learning_rate": 1.0035754013203454e-05, "loss": 0.0259, "step": 1397 }, { "epoch": 1.03, "learning_rate": 1.0023836037016115e-05, "loss": 0.0409, "step": 1398 }, { "epoch": 1.03, "learning_rate": 1.0011918026972205e-05, "loss": 0.0329, "step": 1399 }, { "epoch": 1.03, "learning_rate": 1e-05, "loss": 0.0402, "step": 1400 }, { "epoch": 1.03, "learning_rate": 9.988081973027795e-06, "loss": 0.0324, "step": 1401 }, { "epoch": 1.03, "learning_rate": 9.976163962983889e-06, "loss": 0.0406, "step": 1402 }, { "epoch": 1.03, "learning_rate": 9.964245986796546e-06, "loss": 0.0284, "step": 1403 }, { "epoch": 1.03, "learning_rate": 9.952328061394001e-06, "loss": 0.0382, "step": 1404 }, { "epoch": 1.03, "learning_rate": 9.940410203704395e-06, "loss": 0.0341, "step": 1405 }, { "epoch": 1.03, "learning_rate": 9.92849243065579e-06, "loss": 0.0404, "step": 1406 }, { "epoch": 1.04, "learning_rate": 9.916574759176118e-06, "loss": 0.0279, "step": 1407 }, { "epoch": 1.04, "learning_rate": 9.904657206193175e-06, "loss": 0.0305, "step": 1408 }, { "epoch": 1.04, "learning_rate": 9.892739788634579e-06, "loss": 0.0279, "step": 1409 }, { "epoch": 1.04, "learning_rate": 9.880822523427766e-06, "loss": 0.0275, "step": 1410 }, { "epoch": 1.04, "learning_rate": 9.86890542749994e-06, "loss": 0.0318, "step": 1411 }, { "epoch": 1.04, "learning_rate": 9.856988517778086e-06, "loss": 0.0234, "step": 1412 }, { "epoch": 1.04, "learning_rate": 9.845071811188904e-06, "loss": 0.023, "step": 1413 }, { "epoch": 1.04, "learning_rate": 9.833155324658823e-06, "loss": 0.032, "step": 1414 }, { "epoch": 1.04, "learning_rate": 9.821239075113943e-06, "loss": 0.0417, "step": 1415 }, { "epoch": 1.04, "learning_rate": 9.809323079480043e-06, "loss": 0.0393, "step": 1416 }, { "epoch": 1.04, "learning_rate": 9.797407354682527e-06, "loss": 0.0429, "step": 1417 }, { "epoch": 1.04, "learning_rate": 9.785491917646425e-06, "loss": 0.0281, "step": 1418 }, { "epoch": 1.04, "learning_rate": 9.773576785296353e-06, "loss": 0.0352, "step": 1419 }, { "epoch": 1.04, "learning_rate": 9.761661974556495e-06, "loss": 0.0318, "step": 1420 }, { "epoch": 1.05, "learning_rate": 9.74974750235058e-06, "loss": 0.0232, "step": 1421 }, { "epoch": 1.05, "learning_rate": 9.737833385601858e-06, "loss": 0.0294, "step": 1422 }, { "epoch": 1.05, "learning_rate": 9.725919641233064e-06, "loss": 0.0393, "step": 1423 }, { "epoch": 1.05, "learning_rate": 9.714006286166416e-06, "loss": 0.0348, "step": 1424 }, { "epoch": 1.05, "learning_rate": 9.70209333732357e-06, "loss": 0.0369, "step": 1425 }, { "epoch": 1.05, "learning_rate": 9.690180811625618e-06, "loss": 0.0423, "step": 1426 }, { "epoch": 1.05, "learning_rate": 9.67826872599303e-06, "loss": 0.0283, "step": 1427 }, { "epoch": 1.05, "learning_rate": 9.666357097345672e-06, "loss": 0.0407, "step": 1428 }, { "epoch": 1.05, "learning_rate": 9.654445942602746e-06, "loss": 0.0385, "step": 1429 }, { "epoch": 1.05, "learning_rate": 9.642535278682795e-06, "loss": 0.0368, "step": 1430 }, { "epoch": 1.05, "learning_rate": 9.630625122503645e-06, "loss": 0.0393, "step": 1431 }, { "epoch": 1.05, "learning_rate": 9.618715490982425e-06, "loss": 0.0383, "step": 1432 }, { "epoch": 1.05, "learning_rate": 9.606806401035495e-06, "loss": 0.0396, "step": 1433 }, { "epoch": 1.06, "learning_rate": 9.594897869578466e-06, "loss": 0.0488, "step": 1434 }, { "epoch": 1.06, "learning_rate": 9.58298991352614e-06, "loss": 0.0323, "step": 1435 }, { "epoch": 1.06, "learning_rate": 9.571082549792513e-06, "loss": 0.0247, "step": 1436 }, { "epoch": 1.06, "learning_rate": 9.55917579529073e-06, "loss": 0.0335, "step": 1437 }, { "epoch": 1.06, "learning_rate": 9.547269666933085e-06, "loss": 0.0417, "step": 1438 }, { "epoch": 1.06, "learning_rate": 9.535364181630959e-06, "loss": 0.0327, "step": 1439 }, { "epoch": 1.06, "learning_rate": 9.523459356294849e-06, "loss": 0.0398, "step": 1440 }, { "epoch": 1.06, "learning_rate": 9.511555207834288e-06, "loss": 0.0225, "step": 1441 }, { "epoch": 1.06, "learning_rate": 9.499651753157869e-06, "loss": 0.0293, "step": 1442 }, { "epoch": 1.06, "learning_rate": 9.48774900917318e-06, "loss": 0.0445, "step": 1443 }, { "epoch": 1.06, "learning_rate": 9.475846992786817e-06, "loss": 0.0397, "step": 1444 }, { "epoch": 1.06, "learning_rate": 9.463945720904328e-06, "loss": 0.0506, "step": 1445 }, { "epoch": 1.06, "learning_rate": 9.452045210430218e-06, "loss": 0.0329, "step": 1446 }, { "epoch": 1.06, "learning_rate": 9.440145478267893e-06, "loss": 0.0245, "step": 1447 }, { "epoch": 1.07, "learning_rate": 9.428246541319673e-06, "loss": 0.0374, "step": 1448 }, { "epoch": 1.07, "learning_rate": 9.416348416486732e-06, "loss": 0.0359, "step": 1449 }, { "epoch": 1.07, "learning_rate": 9.404451120669102e-06, "loss": 0.0478, "step": 1450 }, { "epoch": 1.07, "learning_rate": 9.392554670765628e-06, "loss": 0.0345, "step": 1451 }, { "epoch": 1.07, "learning_rate": 9.380659083673963e-06, "loss": 0.0331, "step": 1452 }, { "epoch": 1.07, "learning_rate": 9.368764376290522e-06, "loss": 0.0372, "step": 1453 }, { "epoch": 1.07, "learning_rate": 9.356870565510488e-06, "loss": 0.0384, "step": 1454 }, { "epoch": 1.07, "learning_rate": 9.344977668227751e-06, "loss": 0.0418, "step": 1455 }, { "epoch": 1.07, "learning_rate": 9.333085701334925e-06, "loss": 0.0388, "step": 1456 }, { "epoch": 1.07, "learning_rate": 9.321194681723275e-06, "loss": 0.0439, "step": 1457 }, { "epoch": 1.07, "learning_rate": 9.30930462628275e-06, "loss": 0.0425, "step": 1458 }, { "epoch": 1.07, "learning_rate": 9.297415551901904e-06, "loss": 0.0277, "step": 1459 }, { "epoch": 1.07, "learning_rate": 9.285527475467918e-06, "loss": 0.035, "step": 1460 }, { "epoch": 1.08, "learning_rate": 9.27364041386654e-06, "loss": 0.0546, "step": 1461 }, { "epoch": 1.08, "learning_rate": 9.261754383982093e-06, "loss": 0.045, "step": 1462 }, { "epoch": 1.08, "learning_rate": 9.24986940269741e-06, "loss": 0.0406, "step": 1463 }, { "epoch": 1.08, "learning_rate": 9.23798548689386e-06, "loss": 0.0405, "step": 1464 }, { "epoch": 1.08, "learning_rate": 9.226102653451284e-06, "loss": 0.0282, "step": 1465 }, { "epoch": 1.08, "learning_rate": 9.21422091924799e-06, "loss": 0.037, "step": 1466 }, { "epoch": 1.08, "learning_rate": 9.202340301160722e-06, "loss": 0.0269, "step": 1467 }, { "epoch": 1.08, "learning_rate": 9.190460816064649e-06, "loss": 0.0374, "step": 1468 }, { "epoch": 1.08, "learning_rate": 9.178582480833309e-06, "loss": 0.0452, "step": 1469 }, { "epoch": 1.08, "learning_rate": 9.16670531233863e-06, "loss": 0.0379, "step": 1470 }, { "epoch": 1.08, "learning_rate": 9.154829327450869e-06, "loss": 0.0377, "step": 1471 }, { "epoch": 1.08, "learning_rate": 9.14295454303861e-06, "loss": 0.0378, "step": 1472 }, { "epoch": 1.08, "learning_rate": 9.13108097596872e-06, "loss": 0.0378, "step": 1473 }, { "epoch": 1.08, "learning_rate": 9.119208643106353e-06, "loss": 0.0301, "step": 1474 }, { "epoch": 1.09, "learning_rate": 9.107337561314895e-06, "loss": 0.0332, "step": 1475 }, { "epoch": 1.09, "learning_rate": 9.095467747455965e-06, "loss": 0.0328, "step": 1476 }, { "epoch": 1.09, "learning_rate": 9.083599218389374e-06, "loss": 0.0378, "step": 1477 }, { "epoch": 1.09, "learning_rate": 9.071731990973118e-06, "loss": 0.025, "step": 1478 }, { "epoch": 1.09, "learning_rate": 9.059866082063328e-06, "loss": 0.0346, "step": 1479 }, { "epoch": 1.09, "learning_rate": 9.048001508514283e-06, "loss": 0.0274, "step": 1480 }, { "epoch": 1.09, "learning_rate": 9.036138287178344e-06, "loss": 0.0349, "step": 1481 }, { "epoch": 1.09, "learning_rate": 9.024276434905969e-06, "loss": 0.0305, "step": 1482 }, { "epoch": 1.09, "learning_rate": 9.012415968545658e-06, "loss": 0.0321, "step": 1483 }, { "epoch": 1.09, "learning_rate": 9.000556904943956e-06, "loss": 0.0238, "step": 1484 }, { "epoch": 1.09, "learning_rate": 8.988699260945394e-06, "loss": 0.0336, "step": 1485 }, { "epoch": 1.09, "learning_rate": 8.976843053392518e-06, "loss": 0.0307, "step": 1486 }, { "epoch": 1.09, "learning_rate": 8.964988299125804e-06, "loss": 0.0375, "step": 1487 }, { "epoch": 1.09, "learning_rate": 8.95313501498368e-06, "loss": 0.039, "step": 1488 }, { "epoch": 1.1, "learning_rate": 8.94128321780248e-06, "loss": 0.0645, "step": 1489 }, { "epoch": 1.1, "learning_rate": 8.929432924416433e-06, "loss": 0.0356, "step": 1490 }, { "epoch": 1.1, "learning_rate": 8.917584151657618e-06, "loss": 0.0406, "step": 1491 }, { "epoch": 1.1, "learning_rate": 8.905736916355969e-06, "loss": 0.0349, "step": 1492 }, { "epoch": 1.1, "learning_rate": 8.893891235339226e-06, "loss": 0.0328, "step": 1493 }, { "epoch": 1.1, "learning_rate": 8.882047125432929e-06, "loss": 0.0293, "step": 1494 }, { "epoch": 1.1, "learning_rate": 8.870204603460381e-06, "loss": 0.0226, "step": 1495 }, { "epoch": 1.1, "learning_rate": 8.858363686242635e-06, "loss": 0.0247, "step": 1496 }, { "epoch": 1.1, "learning_rate": 8.846524390598454e-06, "loss": 0.044, "step": 1497 }, { "epoch": 1.1, "learning_rate": 8.834686733344309e-06, "loss": 0.0307, "step": 1498 }, { "epoch": 1.1, "learning_rate": 8.822850731294336e-06, "loss": 0.0354, "step": 1499 }, { "epoch": 1.1, "learning_rate": 8.811016401260327e-06, "loss": 0.0303, "step": 1500 }, { "epoch": 1.1, "learning_rate": 8.799183760051693e-06, "loss": 0.0353, "step": 1501 }, { "epoch": 1.11, "learning_rate": 8.787352824475454e-06, "loss": 0.0274, "step": 1502 }, { "epoch": 1.11, "learning_rate": 8.775523611336191e-06, "loss": 0.0226, "step": 1503 }, { "epoch": 1.11, "learning_rate": 8.76369613743606e-06, "loss": 0.032, "step": 1504 }, { "epoch": 1.11, "learning_rate": 8.751870419574731e-06, "loss": 0.0372, "step": 1505 }, { "epoch": 1.11, "learning_rate": 8.740046474549387e-06, "loss": 0.0409, "step": 1506 }, { "epoch": 1.11, "learning_rate": 8.728224319154687e-06, "loss": 0.0424, "step": 1507 }, { "epoch": 1.11, "learning_rate": 8.716403970182759e-06, "loss": 0.0306, "step": 1508 }, { "epoch": 1.11, "learning_rate": 8.704585444423148e-06, "loss": 0.0357, "step": 1509 }, { "epoch": 1.11, "learning_rate": 8.692768758662827e-06, "loss": 0.0389, "step": 1510 }, { "epoch": 1.11, "learning_rate": 8.68095392968614e-06, "loss": 0.0337, "step": 1511 }, { "epoch": 1.11, "learning_rate": 8.66914097427481e-06, "loss": 0.0379, "step": 1512 }, { "epoch": 1.11, "learning_rate": 8.657329909207883e-06, "loss": 0.0451, "step": 1513 }, { "epoch": 1.11, "learning_rate": 8.645520751261736e-06, "loss": 0.0464, "step": 1514 }, { "epoch": 1.11, "learning_rate": 8.633713517210016e-06, "loss": 0.0413, "step": 1515 }, { "epoch": 1.12, "learning_rate": 8.621908223823657e-06, "loss": 0.0425, "step": 1516 }, { "epoch": 1.12, "learning_rate": 8.610104887870827e-06, "loss": 0.0219, "step": 1517 }, { "epoch": 1.12, "learning_rate": 8.598303526116916e-06, "loss": 0.0352, "step": 1518 }, { "epoch": 1.12, "learning_rate": 8.586504155324508e-06, "loss": 0.0364, "step": 1519 }, { "epoch": 1.12, "learning_rate": 8.574706792253367e-06, "loss": 0.0287, "step": 1520 }, { "epoch": 1.12, "learning_rate": 8.56291145366039e-06, "loss": 0.0533, "step": 1521 }, { "epoch": 1.12, "learning_rate": 8.551118156299613e-06, "loss": 0.0411, "step": 1522 }, { "epoch": 1.12, "learning_rate": 8.539326916922162e-06, "loss": 0.0298, "step": 1523 }, { "epoch": 1.12, "learning_rate": 8.527537752276251e-06, "loss": 0.0387, "step": 1524 }, { "epoch": 1.12, "learning_rate": 8.515750679107133e-06, "loss": 0.0379, "step": 1525 }, { "epoch": 1.12, "learning_rate": 8.503965714157108e-06, "loss": 0.0301, "step": 1526 }, { "epoch": 1.12, "learning_rate": 8.492182874165461e-06, "loss": 0.0342, "step": 1527 }, { "epoch": 1.12, "learning_rate": 8.480402175868477e-06, "loss": 0.0288, "step": 1528 }, { "epoch": 1.13, "learning_rate": 8.468623635999386e-06, "loss": 0.0337, "step": 1529 }, { "epoch": 1.13, "learning_rate": 8.456847271288365e-06, "loss": 0.0381, "step": 1530 }, { "epoch": 1.13, "learning_rate": 8.44507309846248e-06, "loss": 0.0258, "step": 1531 }, { "epoch": 1.13, "learning_rate": 8.43330113424571e-06, "loss": 0.0397, "step": 1532 }, { "epoch": 1.13, "learning_rate": 8.421531395358885e-06, "loss": 0.0256, "step": 1533 }, { "epoch": 1.13, "learning_rate": 8.409763898519665e-06, "loss": 0.0281, "step": 1534 }, { "epoch": 1.13, "learning_rate": 8.397998660442539e-06, "loss": 0.0305, "step": 1535 }, { "epoch": 1.13, "learning_rate": 8.38623569783878e-06, "loss": 0.0289, "step": 1536 }, { "epoch": 1.13, "learning_rate": 8.374475027416438e-06, "loss": 0.0356, "step": 1537 }, { "epoch": 1.13, "learning_rate": 8.362716665880289e-06, "loss": 0.0305, "step": 1538 }, { "epoch": 1.13, "learning_rate": 8.350960629931853e-06, "loss": 0.031, "step": 1539 }, { "epoch": 1.13, "learning_rate": 8.339206936269328e-06, "loss": 0.0292, "step": 1540 }, { "epoch": 1.13, "learning_rate": 8.327455601587595e-06, "loss": 0.0403, "step": 1541 }, { "epoch": 1.13, "learning_rate": 8.315706642578178e-06, "loss": 0.0343, "step": 1542 }, { "epoch": 1.14, "learning_rate": 8.303960075929239e-06, "loss": 0.0293, "step": 1543 }, { "epoch": 1.14, "learning_rate": 8.292215918325518e-06, "loss": 0.0262, "step": 1544 }, { "epoch": 1.14, "learning_rate": 8.280474186448362e-06, "loss": 0.0388, "step": 1545 }, { "epoch": 1.14, "learning_rate": 8.268734896975649e-06, "loss": 0.0418, "step": 1546 }, { "epoch": 1.14, "learning_rate": 8.256998066581803e-06, "loss": 0.0238, "step": 1547 }, { "epoch": 1.14, "learning_rate": 8.245263711937746e-06, "loss": 0.0418, "step": 1548 }, { "epoch": 1.14, "learning_rate": 8.233531849710893e-06, "loss": 0.0232, "step": 1549 }, { "epoch": 1.14, "learning_rate": 8.221802496565102e-06, "loss": 0.0389, "step": 1550 }, { "epoch": 1.14, "learning_rate": 8.210075669160688e-06, "loss": 0.0363, "step": 1551 }, { "epoch": 1.14, "learning_rate": 8.198351384154363e-06, "loss": 0.0218, "step": 1552 }, { "epoch": 1.14, "learning_rate": 8.186629658199237e-06, "loss": 0.0228, "step": 1553 }, { "epoch": 1.14, "learning_rate": 8.174910507944775e-06, "loss": 0.0219, "step": 1554 }, { "epoch": 1.14, "learning_rate": 8.163193950036798e-06, "loss": 0.0303, "step": 1555 }, { "epoch": 1.14, "learning_rate": 8.15148000111743e-06, "loss": 0.0262, "step": 1556 }, { "epoch": 1.15, "learning_rate": 8.139768677825098e-06, "loss": 0.0307, "step": 1557 }, { "epoch": 1.15, "learning_rate": 8.128059996794495e-06, "loss": 0.0363, "step": 1558 }, { "epoch": 1.15, "learning_rate": 8.116353974656569e-06, "loss": 0.0291, "step": 1559 }, { "epoch": 1.15, "learning_rate": 8.10465062803848e-06, "loss": 0.0271, "step": 1560 }, { "epoch": 1.15, "learning_rate": 8.0929499735636e-06, "loss": 0.0329, "step": 1561 }, { "epoch": 1.15, "learning_rate": 8.08125202785146e-06, "loss": 0.0354, "step": 1562 }, { "epoch": 1.15, "learning_rate": 8.069556807517764e-06, "loss": 0.0201, "step": 1563 }, { "epoch": 1.15, "learning_rate": 8.05786432917433e-06, "loss": 0.031, "step": 1564 }, { "epoch": 1.15, "learning_rate": 8.046174609429087e-06, "loss": 0.0444, "step": 1565 }, { "epoch": 1.15, "learning_rate": 8.034487664886042e-06, "loss": 0.0364, "step": 1566 }, { "epoch": 1.15, "learning_rate": 8.022803512145272e-06, "loss": 0.0335, "step": 1567 }, { "epoch": 1.15, "learning_rate": 8.011122167802869e-06, "loss": 0.0322, "step": 1568 }, { "epoch": 1.15, "learning_rate": 7.999443648450951e-06, "loss": 0.0466, "step": 1569 }, { "epoch": 1.16, "learning_rate": 7.987767970677618e-06, "loss": 0.0297, "step": 1570 }, { "epoch": 1.16, "learning_rate": 7.976095151066936e-06, "loss": 0.0407, "step": 1571 }, { "epoch": 1.16, "learning_rate": 7.964425206198907e-06, "loss": 0.0328, "step": 1572 }, { "epoch": 1.16, "learning_rate": 7.952758152649458e-06, "loss": 0.0397, "step": 1573 }, { "epoch": 1.16, "learning_rate": 7.941094006990398e-06, "loss": 0.0387, "step": 1574 }, { "epoch": 1.16, "learning_rate": 7.929432785789417e-06, "loss": 0.0344, "step": 1575 }, { "epoch": 1.16, "learning_rate": 7.917774505610039e-06, "loss": 0.0338, "step": 1576 }, { "epoch": 1.16, "learning_rate": 7.906119183011625e-06, "loss": 0.0276, "step": 1577 }, { "epoch": 1.16, "learning_rate": 7.89446683454932e-06, "loss": 0.0406, "step": 1578 }, { "epoch": 1.16, "learning_rate": 7.88281747677406e-06, "loss": 0.0363, "step": 1579 }, { "epoch": 1.16, "learning_rate": 7.871171126232516e-06, "loss": 0.0295, "step": 1580 }, { "epoch": 1.16, "learning_rate": 7.859527799467104e-06, "loss": 0.0259, "step": 1581 }, { "epoch": 1.16, "learning_rate": 7.84788751301593e-06, "loss": 0.0442, "step": 1582 }, { "epoch": 1.16, "learning_rate": 7.836250283412796e-06, "loss": 0.0279, "step": 1583 }, { "epoch": 1.17, "learning_rate": 7.82461612718715e-06, "loss": 0.0277, "step": 1584 }, { "epoch": 1.17, "learning_rate": 7.812985060864086e-06, "loss": 0.0282, "step": 1585 }, { "epoch": 1.17, "learning_rate": 7.801357100964295e-06, "loss": 0.0408, "step": 1586 }, { "epoch": 1.17, "learning_rate": 7.789732264004071e-06, "loss": 0.0381, "step": 1587 }, { "epoch": 1.17, "learning_rate": 7.778110566495256e-06, "loss": 0.0307, "step": 1588 }, { "epoch": 1.17, "learning_rate": 7.766492024945253e-06, "loss": 0.0234, "step": 1589 }, { "epoch": 1.17, "learning_rate": 7.754876655856957e-06, "loss": 0.0335, "step": 1590 }, { "epoch": 1.17, "learning_rate": 7.743264475728785e-06, "loss": 0.0292, "step": 1591 }, { "epoch": 1.17, "learning_rate": 7.731655501054597e-06, "loss": 0.0356, "step": 1592 }, { "epoch": 1.17, "learning_rate": 7.720049748323722e-06, "loss": 0.04, "step": 1593 }, { "epoch": 1.17, "learning_rate": 7.708447234020898e-06, "loss": 0.0313, "step": 1594 }, { "epoch": 1.17, "learning_rate": 7.696847974626274e-06, "loss": 0.0327, "step": 1595 }, { "epoch": 1.17, "learning_rate": 7.685251986615363e-06, "loss": 0.0313, "step": 1596 }, { "epoch": 1.18, "learning_rate": 7.673659286459047e-06, "loss": 0.0347, "step": 1597 }, { "epoch": 1.18, "learning_rate": 7.662069890623525e-06, "loss": 0.0262, "step": 1598 }, { "epoch": 1.18, "learning_rate": 7.650483815570309e-06, "loss": 0.038, "step": 1599 }, { "epoch": 1.18, "learning_rate": 7.63890107775619e-06, "loss": 0.0303, "step": 1600 }, { "epoch": 1.18, "learning_rate": 7.627321693633227e-06, "loss": 0.0303, "step": 1601 }, { "epoch": 1.18, "learning_rate": 7.615745679648702e-06, "loss": 0.0262, "step": 1602 }, { "epoch": 1.18, "learning_rate": 7.604173052245123e-06, "loss": 0.034, "step": 1603 }, { "epoch": 1.18, "learning_rate": 7.59260382786018e-06, "loss": 0.0467, "step": 1604 }, { "epoch": 1.18, "learning_rate": 7.581038022926735e-06, "loss": 0.033, "step": 1605 }, { "epoch": 1.18, "learning_rate": 7.569475653872787e-06, "loss": 0.0347, "step": 1606 }, { "epoch": 1.18, "learning_rate": 7.557916737121462e-06, "loss": 0.0279, "step": 1607 }, { "epoch": 1.18, "learning_rate": 7.546361289090971e-06, "loss": 0.0361, "step": 1608 }, { "epoch": 1.18, "learning_rate": 7.53480932619461e-06, "loss": 0.0336, "step": 1609 }, { "epoch": 1.18, "learning_rate": 7.5232608648407166e-06, "loss": 0.034, "step": 1610 }, { "epoch": 1.19, "learning_rate": 7.511715921432663e-06, "loss": 0.0392, "step": 1611 }, { "epoch": 1.19, "learning_rate": 7.500174512368814e-06, "loss": 0.0351, "step": 1612 }, { "epoch": 1.19, "learning_rate": 7.488636654042527e-06, "loss": 0.0319, "step": 1613 }, { "epoch": 1.19, "learning_rate": 7.477102362842099e-06, "loss": 0.0464, "step": 1614 }, { "epoch": 1.19, "learning_rate": 7.465571655150778e-06, "loss": 0.0301, "step": 1615 }, { "epoch": 1.19, "learning_rate": 7.454044547346708e-06, "loss": 0.0315, "step": 1616 }, { "epoch": 1.19, "learning_rate": 7.442521055802932e-06, "loss": 0.031, "step": 1617 }, { "epoch": 1.19, "learning_rate": 7.431001196887345e-06, "loss": 0.0406, "step": 1618 }, { "epoch": 1.19, "learning_rate": 7.419484986962694e-06, "loss": 0.0333, "step": 1619 }, { "epoch": 1.19, "learning_rate": 7.407972442386527e-06, "loss": 0.0291, "step": 1620 }, { "epoch": 1.19, "learning_rate": 7.396463579511204e-06, "loss": 0.0322, "step": 1621 }, { "epoch": 1.19, "learning_rate": 7.384958414683839e-06, "loss": 0.033, "step": 1622 }, { "epoch": 1.19, "learning_rate": 7.373456964246309e-06, "loss": 0.0245, "step": 1623 }, { "epoch": 1.19, "learning_rate": 7.361959244535199e-06, "loss": 0.0335, "step": 1624 }, { "epoch": 1.2, "learning_rate": 7.350465271881815e-06, "loss": 0.0322, "step": 1625 }, { "epoch": 1.2, "learning_rate": 7.338975062612115e-06, "loss": 0.0284, "step": 1626 }, { "epoch": 1.2, "learning_rate": 7.327488633046733e-06, "loss": 0.0282, "step": 1627 }, { "epoch": 1.2, "learning_rate": 7.316005999500924e-06, "loss": 0.0485, "step": 1628 }, { "epoch": 1.2, "learning_rate": 7.3045271782845576e-06, "loss": 0.0386, "step": 1629 }, { "epoch": 1.2, "learning_rate": 7.293052185702079e-06, "loss": 0.0365, "step": 1630 }, { "epoch": 1.2, "learning_rate": 7.2815810380525075e-06, "loss": 0.0186, "step": 1631 }, { "epoch": 1.2, "learning_rate": 7.270113751629388e-06, "loss": 0.0327, "step": 1632 }, { "epoch": 1.2, "learning_rate": 7.258650342720791e-06, "loss": 0.0463, "step": 1633 }, { "epoch": 1.2, "learning_rate": 7.247190827609273e-06, "loss": 0.0369, "step": 1634 }, { "epoch": 1.2, "learning_rate": 7.235735222571868e-06, "loss": 0.025, "step": 1635 }, { "epoch": 1.2, "learning_rate": 7.224283543880041e-06, "loss": 0.0327, "step": 1636 }, { "epoch": 1.2, "learning_rate": 7.2128358077997005e-06, "loss": 0.0363, "step": 1637 }, { "epoch": 1.21, "learning_rate": 7.201392030591137e-06, "loss": 0.0375, "step": 1638 }, { "epoch": 1.21, "learning_rate": 7.1899522285090275e-06, "loss": 0.0373, "step": 1639 }, { "epoch": 1.21, "learning_rate": 7.178516417802399e-06, "loss": 0.0355, "step": 1640 }, { "epoch": 1.21, "learning_rate": 7.1670846147146144e-06, "loss": 0.0226, "step": 1641 }, { "epoch": 1.21, "learning_rate": 7.155656835483331e-06, "loss": 0.0501, "step": 1642 }, { "epoch": 1.21, "learning_rate": 7.144233096340512e-06, "loss": 0.0278, "step": 1643 }, { "epoch": 1.21, "learning_rate": 7.132813413512361e-06, "loss": 0.0383, "step": 1644 }, { "epoch": 1.21, "learning_rate": 7.121397803219332e-06, "loss": 0.0365, "step": 1645 }, { "epoch": 1.21, "learning_rate": 7.10998628167609e-06, "loss": 0.0302, "step": 1646 }, { "epoch": 1.21, "learning_rate": 7.098578865091499e-06, "loss": 0.0347, "step": 1647 }, { "epoch": 1.21, "learning_rate": 7.087175569668576e-06, "loss": 0.0316, "step": 1648 }, { "epoch": 1.21, "learning_rate": 7.075776411604509e-06, "loss": 0.0273, "step": 1649 }, { "epoch": 1.21, "learning_rate": 7.064381407090584e-06, "loss": 0.0382, "step": 1650 }, { "epoch": 1.21, "learning_rate": 7.052990572312206e-06, "loss": 0.0199, "step": 1651 }, { "epoch": 1.22, "learning_rate": 7.041603923448847e-06, "loss": 0.0412, "step": 1652 }, { "epoch": 1.22, "learning_rate": 7.030221476674042e-06, "loss": 0.0194, "step": 1653 }, { "epoch": 1.22, "learning_rate": 7.018843248155345e-06, "loss": 0.0308, "step": 1654 }, { "epoch": 1.22, "learning_rate": 7.007469254054329e-06, "loss": 0.0345, "step": 1655 }, { "epoch": 1.22, "learning_rate": 6.996099510526546e-06, "loss": 0.0408, "step": 1656 }, { "epoch": 1.22, "learning_rate": 6.984734033721518e-06, "loss": 0.0283, "step": 1657 }, { "epoch": 1.22, "learning_rate": 6.973372839782699e-06, "loss": 0.0217, "step": 1658 }, { "epoch": 1.22, "learning_rate": 6.962015944847469e-06, "loss": 0.0251, "step": 1659 }, { "epoch": 1.22, "learning_rate": 6.950663365047083e-06, "loss": 0.031, "step": 1660 }, { "epoch": 1.22, "learning_rate": 6.939315116506691e-06, "loss": 0.0219, "step": 1661 }, { "epoch": 1.22, "learning_rate": 6.927971215345271e-06, "loss": 0.0356, "step": 1662 }, { "epoch": 1.22, "learning_rate": 6.9166316776756405e-06, "loss": 0.0439, "step": 1663 }, { "epoch": 1.22, "learning_rate": 6.905296519604407e-06, "loss": 0.0308, "step": 1664 }, { "epoch": 1.23, "learning_rate": 6.893965757231964e-06, "loss": 0.0274, "step": 1665 }, { "epoch": 1.23, "learning_rate": 6.8826394066524695e-06, "loss": 0.0284, "step": 1666 }, { "epoch": 1.23, "learning_rate": 6.871317483953792e-06, "loss": 0.0435, "step": 1667 }, { "epoch": 1.23, "learning_rate": 6.860000005217533e-06, "loss": 0.0341, "step": 1668 }, { "epoch": 1.23, "learning_rate": 6.8486869865189705e-06, "loss": 0.0325, "step": 1669 }, { "epoch": 1.23, "learning_rate": 6.837378443927052e-06, "loss": 0.0293, "step": 1670 }, { "epoch": 1.23, "learning_rate": 6.826074393504364e-06, "loss": 0.0306, "step": 1671 }, { "epoch": 1.23, "learning_rate": 6.814774851307118e-06, "loss": 0.0316, "step": 1672 }, { "epoch": 1.23, "learning_rate": 6.80347983338511e-06, "loss": 0.0244, "step": 1673 }, { "epoch": 1.23, "learning_rate": 6.7921893557817246e-06, "loss": 0.0222, "step": 1674 }, { "epoch": 1.23, "learning_rate": 6.780903434533885e-06, "loss": 0.0316, "step": 1675 }, { "epoch": 1.23, "learning_rate": 6.769622085672054e-06, "loss": 0.0385, "step": 1676 }, { "epoch": 1.23, "learning_rate": 6.758345325220188e-06, "loss": 0.0528, "step": 1677 }, { "epoch": 1.23, "learning_rate": 6.747073169195739e-06, "loss": 0.0304, "step": 1678 }, { "epoch": 1.24, "learning_rate": 6.735805633609602e-06, "loss": 0.0294, "step": 1679 }, { "epoch": 1.24, "learning_rate": 6.724542734466127e-06, "loss": 0.0356, "step": 1680 }, { "epoch": 1.24, "learning_rate": 6.713284487763066e-06, "loss": 0.0343, "step": 1681 }, { "epoch": 1.24, "learning_rate": 6.70203090949157e-06, "loss": 0.0531, "step": 1682 }, { "epoch": 1.24, "learning_rate": 6.690782015636152e-06, "loss": 0.0361, "step": 1683 }, { "epoch": 1.24, "learning_rate": 6.679537822174682e-06, "loss": 0.0299, "step": 1684 }, { "epoch": 1.24, "learning_rate": 6.668298345078339e-06, "loss": 0.0324, "step": 1685 }, { "epoch": 1.24, "learning_rate": 6.657063600311616e-06, "loss": 0.0352, "step": 1686 }, { "epoch": 1.24, "learning_rate": 6.645833603832276e-06, "loss": 0.0417, "step": 1687 }, { "epoch": 1.24, "learning_rate": 6.634608371591343e-06, "loss": 0.0309, "step": 1688 }, { "epoch": 1.24, "learning_rate": 6.62338791953307e-06, "loss": 0.0457, "step": 1689 }, { "epoch": 1.24, "learning_rate": 6.6121722635949244e-06, "loss": 0.0374, "step": 1690 }, { "epoch": 1.24, "learning_rate": 6.6009614197075545e-06, "loss": 0.0353, "step": 1691 }, { "epoch": 1.25, "learning_rate": 6.58975540379478e-06, "loss": 0.0477, "step": 1692 }, { "epoch": 1.25, "learning_rate": 6.578554231773558e-06, "loss": 0.0339, "step": 1693 }, { "epoch": 1.25, "learning_rate": 6.567357919553973e-06, "loss": 0.0432, "step": 1694 }, { "epoch": 1.25, "learning_rate": 6.556166483039196e-06, "loss": 0.0357, "step": 1695 }, { "epoch": 1.25, "learning_rate": 6.544979938125485e-06, "loss": 0.0403, "step": 1696 }, { "epoch": 1.25, "learning_rate": 6.5337983007021355e-06, "loss": 0.0258, "step": 1697 }, { "epoch": 1.25, "learning_rate": 6.522621586651485e-06, "loss": 0.039, "step": 1698 }, { "epoch": 1.25, "learning_rate": 6.511449811848871e-06, "loss": 0.0225, "step": 1699 }, { "epoch": 1.25, "learning_rate": 6.5002829921626206e-06, "loss": 0.025, "step": 1700 }, { "epoch": 1.25, "learning_rate": 6.489121143454013e-06, "loss": 0.0289, "step": 1701 }, { "epoch": 1.25, "learning_rate": 6.477964281577282e-06, "loss": 0.0476, "step": 1702 }, { "epoch": 1.25, "learning_rate": 6.4668124223795606e-06, "loss": 0.0325, "step": 1703 }, { "epoch": 1.25, "learning_rate": 6.4556655817008895e-06, "loss": 0.0336, "step": 1704 }, { "epoch": 1.25, "learning_rate": 6.444523775374173e-06, "loss": 0.0255, "step": 1705 }, { "epoch": 1.26, "learning_rate": 6.433387019225175e-06, "loss": 0.0311, "step": 1706 }, { "epoch": 1.26, "learning_rate": 6.422255329072467e-06, "loss": 0.0309, "step": 1707 }, { "epoch": 1.26, "learning_rate": 6.411128720727448e-06, "loss": 0.0422, "step": 1708 }, { "epoch": 1.26, "learning_rate": 6.400007209994284e-06, "loss": 0.0342, "step": 1709 }, { "epoch": 1.26, "learning_rate": 6.3888908126699015e-06, "loss": 0.0334, "step": 1710 }, { "epoch": 1.26, "learning_rate": 6.3777795445439684e-06, "loss": 0.0332, "step": 1711 }, { "epoch": 1.26, "learning_rate": 6.366673421398869e-06, "loss": 0.0423, "step": 1712 }, { "epoch": 1.26, "learning_rate": 6.355572459009665e-06, "loss": 0.0258, "step": 1713 }, { "epoch": 1.26, "learning_rate": 6.344476673144113e-06, "loss": 0.0267, "step": 1714 }, { "epoch": 1.26, "learning_rate": 6.333386079562593e-06, "loss": 0.0293, "step": 1715 }, { "epoch": 1.26, "learning_rate": 6.322300694018122e-06, "loss": 0.0254, "step": 1716 }, { "epoch": 1.26, "learning_rate": 6.311220532256315e-06, "loss": 0.0206, "step": 1717 }, { "epoch": 1.26, "learning_rate": 6.3001456100153754e-06, "loss": 0.0378, "step": 1718 }, { "epoch": 1.26, "learning_rate": 6.289075943026051e-06, "loss": 0.0336, "step": 1719 }, { "epoch": 1.27, "learning_rate": 6.278011547011638e-06, "loss": 0.034, "step": 1720 }, { "epoch": 1.27, "learning_rate": 6.2669524376879335e-06, "loss": 0.0155, "step": 1721 }, { "epoch": 1.27, "learning_rate": 6.255898630763238e-06, "loss": 0.0298, "step": 1722 }, { "epoch": 1.27, "learning_rate": 6.2448501419383105e-06, "loss": 0.0329, "step": 1723 }, { "epoch": 1.27, "learning_rate": 6.233806986906367e-06, "loss": 0.0269, "step": 1724 }, { "epoch": 1.27, "learning_rate": 6.22276918135303e-06, "loss": 0.0266, "step": 1725 }, { "epoch": 1.27, "learning_rate": 6.211736740956343e-06, "loss": 0.033, "step": 1726 }, { "epoch": 1.27, "learning_rate": 6.200709681386714e-06, "loss": 0.0308, "step": 1727 }, { "epoch": 1.27, "learning_rate": 6.189688018306919e-06, "loss": 0.0352, "step": 1728 }, { "epoch": 1.27, "learning_rate": 6.17867176737206e-06, "loss": 0.0289, "step": 1729 }, { "epoch": 1.27, "learning_rate": 6.167660944229561e-06, "loss": 0.0367, "step": 1730 }, { "epoch": 1.27, "learning_rate": 6.156655564519125e-06, "loss": 0.0368, "step": 1731 }, { "epoch": 1.27, "learning_rate": 6.145655643872733e-06, "loss": 0.0276, "step": 1732 }, { "epoch": 1.28, "learning_rate": 6.1346611979146055e-06, "loss": 0.025, "step": 1733 }, { "epoch": 1.28, "learning_rate": 6.123672242261191e-06, "loss": 0.036, "step": 1734 }, { "epoch": 1.28, "learning_rate": 6.1126887925211395e-06, "loss": 0.0227, "step": 1735 }, { "epoch": 1.28, "learning_rate": 6.101710864295279e-06, "loss": 0.0349, "step": 1736 }, { "epoch": 1.28, "learning_rate": 6.09073847317659e-06, "loss": 0.0358, "step": 1737 }, { "epoch": 1.28, "learning_rate": 6.0797716347502e-06, "loss": 0.0257, "step": 1738 }, { "epoch": 1.28, "learning_rate": 6.068810364593335e-06, "loss": 0.0356, "step": 1739 }, { "epoch": 1.28, "learning_rate": 6.057854678275326e-06, "loss": 0.0248, "step": 1740 }, { "epoch": 1.28, "learning_rate": 6.046904591357559e-06, "loss": 0.0379, "step": 1741 }, { "epoch": 1.28, "learning_rate": 6.035960119393483e-06, "loss": 0.0311, "step": 1742 }, { "epoch": 1.28, "learning_rate": 6.025021277928553e-06, "loss": 0.0352, "step": 1743 }, { "epoch": 1.28, "learning_rate": 6.014088082500241e-06, "loss": 0.0343, "step": 1744 }, { "epoch": 1.28, "learning_rate": 6.003160548637992e-06, "loss": 0.042, "step": 1745 }, { "epoch": 1.28, "learning_rate": 5.9922386918632145e-06, "loss": 0.0332, "step": 1746 }, { "epoch": 1.29, "learning_rate": 5.981322527689242e-06, "loss": 0.0304, "step": 1747 }, { "epoch": 1.29, "learning_rate": 5.9704120716213435e-06, "loss": 0.0319, "step": 1748 }, { "epoch": 1.29, "learning_rate": 5.9595073391566536e-06, "loss": 0.0276, "step": 1749 }, { "epoch": 1.29, "learning_rate": 5.948608345784201e-06, "loss": 0.0205, "step": 1750 }, { "epoch": 1.29, "learning_rate": 5.937715106984847e-06, "loss": 0.03, "step": 1751 }, { "epoch": 1.29, "learning_rate": 5.926827638231289e-06, "loss": 0.0312, "step": 1752 }, { "epoch": 1.29, "learning_rate": 5.915945954988016e-06, "loss": 0.0299, "step": 1753 }, { "epoch": 1.29, "learning_rate": 5.905070072711318e-06, "loss": 0.0318, "step": 1754 }, { "epoch": 1.29, "learning_rate": 5.8942000068492256e-06, "loss": 0.029, "step": 1755 }, { "epoch": 1.29, "learning_rate": 5.883335772841523e-06, "loss": 0.033, "step": 1756 }, { "epoch": 1.29, "learning_rate": 5.872477386119702e-06, "loss": 0.0253, "step": 1757 }, { "epoch": 1.29, "learning_rate": 5.8616248621069545e-06, "loss": 0.0255, "step": 1758 }, { "epoch": 1.29, "learning_rate": 5.850778216218136e-06, "loss": 0.0304, "step": 1759 }, { "epoch": 1.3, "learning_rate": 5.83993746385977e-06, "loss": 0.0219, "step": 1760 }, { "epoch": 1.3, "learning_rate": 5.829102620429985e-06, "loss": 0.0322, "step": 1761 }, { "epoch": 1.3, "learning_rate": 5.818273701318542e-06, "loss": 0.0423, "step": 1762 }, { "epoch": 1.3, "learning_rate": 5.807450721906764e-06, "loss": 0.0229, "step": 1763 }, { "epoch": 1.3, "learning_rate": 5.796633697567557e-06, "loss": 0.0341, "step": 1764 }, { "epoch": 1.3, "learning_rate": 5.785822643665352e-06, "loss": 0.0233, "step": 1765 }, { "epoch": 1.3, "learning_rate": 5.77501757555611e-06, "loss": 0.0313, "step": 1766 }, { "epoch": 1.3, "learning_rate": 5.764218508587286e-06, "loss": 0.032, "step": 1767 }, { "epoch": 1.3, "learning_rate": 5.753425458097817e-06, "loss": 0.0453, "step": 1768 }, { "epoch": 1.3, "learning_rate": 5.74263843941808e-06, "loss": 0.036, "step": 1769 }, { "epoch": 1.3, "learning_rate": 5.731857467869902e-06, "loss": 0.0299, "step": 1770 }, { "epoch": 1.3, "learning_rate": 5.7210825587665045e-06, "loss": 0.0207, "step": 1771 }, { "epoch": 1.3, "learning_rate": 5.710313727412513e-06, "loss": 0.0308, "step": 1772 }, { "epoch": 1.3, "learning_rate": 5.699550989103903e-06, "loss": 0.0258, "step": 1773 }, { "epoch": 1.31, "learning_rate": 5.688794359128018e-06, "loss": 0.0312, "step": 1774 }, { "epoch": 1.31, "learning_rate": 5.678043852763502e-06, "loss": 0.0323, "step": 1775 }, { "epoch": 1.31, "learning_rate": 5.6672994852803184e-06, "loss": 0.0311, "step": 1776 }, { "epoch": 1.31, "learning_rate": 5.656561271939699e-06, "loss": 0.032, "step": 1777 }, { "epoch": 1.31, "learning_rate": 5.645829227994146e-06, "loss": 0.0332, "step": 1778 }, { "epoch": 1.31, "learning_rate": 5.635103368687378e-06, "loss": 0.0269, "step": 1779 }, { "epoch": 1.31, "learning_rate": 5.624383709254363e-06, "loss": 0.0346, "step": 1780 }, { "epoch": 1.31, "learning_rate": 5.613670264921227e-06, "loss": 0.0354, "step": 1781 }, { "epoch": 1.31, "learning_rate": 5.602963050905296e-06, "loss": 0.0352, "step": 1782 }, { "epoch": 1.31, "learning_rate": 5.592262082415022e-06, "loss": 0.0278, "step": 1783 }, { "epoch": 1.31, "learning_rate": 5.58156737465001e-06, "loss": 0.0414, "step": 1784 }, { "epoch": 1.31, "learning_rate": 5.570878942800945e-06, "loss": 0.037, "step": 1785 }, { "epoch": 1.31, "learning_rate": 5.560196802049633e-06, "loss": 0.0314, "step": 1786 }, { "epoch": 1.31, "learning_rate": 5.54952096756891e-06, "loss": 0.0323, "step": 1787 }, { "epoch": 1.32, "learning_rate": 5.538851454522678e-06, "loss": 0.0377, "step": 1788 }, { "epoch": 1.32, "learning_rate": 5.528188278065842e-06, "loss": 0.0334, "step": 1789 }, { "epoch": 1.32, "learning_rate": 5.517531453344327e-06, "loss": 0.0237, "step": 1790 }, { "epoch": 1.32, "learning_rate": 5.506880995495013e-06, "loss": 0.0391, "step": 1791 }, { "epoch": 1.32, "learning_rate": 5.496236919645754e-06, "loss": 0.0412, "step": 1792 }, { "epoch": 1.32, "learning_rate": 5.485599240915334e-06, "loss": 0.0305, "step": 1793 }, { "epoch": 1.32, "learning_rate": 5.474967974413451e-06, "loss": 0.0345, "step": 1794 }, { "epoch": 1.32, "learning_rate": 5.4643431352406885e-06, "loss": 0.0297, "step": 1795 }, { "epoch": 1.32, "learning_rate": 5.453724738488511e-06, "loss": 0.0393, "step": 1796 }, { "epoch": 1.32, "learning_rate": 5.44311279923923e-06, "loss": 0.0288, "step": 1797 }, { "epoch": 1.32, "learning_rate": 5.432507332565968e-06, "loss": 0.0316, "step": 1798 }, { "epoch": 1.32, "learning_rate": 5.421908353532688e-06, "loss": 0.0276, "step": 1799 }, { "epoch": 1.32, "learning_rate": 5.411315877194104e-06, "loss": 0.0255, "step": 1800 }, { "epoch": 1.33, "learning_rate": 5.400729918595715e-06, "loss": 0.0305, "step": 1801 }, { "epoch": 1.33, "learning_rate": 5.390150492773749e-06, "loss": 0.0316, "step": 1802 }, { "epoch": 1.33, "learning_rate": 5.379577614755167e-06, "loss": 0.0406, "step": 1803 }, { "epoch": 1.33, "learning_rate": 5.369011299557617e-06, "loss": 0.0385, "step": 1804 }, { "epoch": 1.33, "learning_rate": 5.358451562189434e-06, "loss": 0.0336, "step": 1805 }, { "epoch": 1.33, "learning_rate": 5.347898417649609e-06, "loss": 0.0313, "step": 1806 }, { "epoch": 1.33, "learning_rate": 5.337351880927768e-06, "loss": 0.0402, "step": 1807 }, { "epoch": 1.33, "learning_rate": 5.3268119670041465e-06, "loss": 0.0392, "step": 1808 }, { "epoch": 1.33, "learning_rate": 5.3162786908495815e-06, "loss": 0.0261, "step": 1809 }, { "epoch": 1.33, "learning_rate": 5.305752067425469e-06, "loss": 0.0232, "step": 1810 }, { "epoch": 1.33, "learning_rate": 5.2952321116837706e-06, "loss": 0.0342, "step": 1811 }, { "epoch": 1.33, "learning_rate": 5.284718838566968e-06, "loss": 0.0397, "step": 1812 }, { "epoch": 1.33, "learning_rate": 5.2742122630080565e-06, "loss": 0.0331, "step": 1813 }, { "epoch": 1.33, "learning_rate": 5.26371239993051e-06, "loss": 0.0202, "step": 1814 }, { "epoch": 1.34, "learning_rate": 5.253219264248276e-06, "loss": 0.0349, "step": 1815 }, { "epoch": 1.34, "learning_rate": 5.242732870865739e-06, "loss": 0.0428, "step": 1816 }, { "epoch": 1.34, "learning_rate": 5.232253234677713e-06, "loss": 0.0347, "step": 1817 }, { "epoch": 1.34, "learning_rate": 5.221780370569415e-06, "loss": 0.0228, "step": 1818 }, { "epoch": 1.34, "learning_rate": 5.211314293416441e-06, "loss": 0.0309, "step": 1819 }, { "epoch": 1.34, "learning_rate": 5.2008550180847394e-06, "loss": 0.0321, "step": 1820 }, { "epoch": 1.34, "learning_rate": 5.190402559430612e-06, "loss": 0.0268, "step": 1821 }, { "epoch": 1.34, "learning_rate": 5.1799569323006615e-06, "loss": 0.0374, "step": 1822 }, { "epoch": 1.34, "learning_rate": 5.169518151531801e-06, "loss": 0.029, "step": 1823 }, { "epoch": 1.34, "learning_rate": 5.159086231951213e-06, "loss": 0.0369, "step": 1824 }, { "epoch": 1.34, "learning_rate": 5.148661188376338e-06, "loss": 0.0285, "step": 1825 }, { "epoch": 1.34, "learning_rate": 5.138243035614842e-06, "loss": 0.0288, "step": 1826 }, { "epoch": 1.34, "learning_rate": 5.127831788464615e-06, "loss": 0.0343, "step": 1827 }, { "epoch": 1.35, "learning_rate": 5.117427461713724e-06, "loss": 0.0331, "step": 1828 }, { "epoch": 1.35, "learning_rate": 5.107030070140418e-06, "loss": 0.0429, "step": 1829 }, { "epoch": 1.35, "learning_rate": 5.096639628513092e-06, "loss": 0.0299, "step": 1830 }, { "epoch": 1.35, "learning_rate": 5.086256151590271e-06, "loss": 0.0309, "step": 1831 }, { "epoch": 1.35, "learning_rate": 5.0758796541205794e-06, "loss": 0.036, "step": 1832 }, { "epoch": 1.35, "learning_rate": 5.065510150842738e-06, "loss": 0.0317, "step": 1833 }, { "epoch": 1.35, "learning_rate": 5.055147656485526e-06, "loss": 0.031, "step": 1834 }, { "epoch": 1.35, "learning_rate": 5.0447921857677705e-06, "loss": 0.0388, "step": 1835 }, { "epoch": 1.35, "learning_rate": 5.034443753398323e-06, "loss": 0.0335, "step": 1836 }, { "epoch": 1.35, "learning_rate": 5.024102374076041e-06, "loss": 0.035, "step": 1837 }, { "epoch": 1.35, "learning_rate": 5.01376806248975e-06, "loss": 0.0231, "step": 1838 }, { "epoch": 1.35, "learning_rate": 5.003440833318254e-06, "loss": 0.0296, "step": 1839 }, { "epoch": 1.35, "learning_rate": 4.993120701230283e-06, "loss": 0.0202, "step": 1840 }, { "epoch": 1.35, "learning_rate": 4.9828076808844945e-06, "loss": 0.0315, "step": 1841 }, { "epoch": 1.36, "learning_rate": 4.972501786929443e-06, "loss": 0.0189, "step": 1842 }, { "epoch": 1.36, "learning_rate": 4.9622030340035655e-06, "loss": 0.0285, "step": 1843 }, { "epoch": 1.36, "learning_rate": 4.951911436735142e-06, "loss": 0.0371, "step": 1844 }, { "epoch": 1.36, "learning_rate": 4.9416270097423035e-06, "loss": 0.0336, "step": 1845 }, { "epoch": 1.36, "learning_rate": 4.931349767632985e-06, "loss": 0.0329, "step": 1846 }, { "epoch": 1.36, "learning_rate": 4.921079725004923e-06, "loss": 0.0358, "step": 1847 }, { "epoch": 1.36, "learning_rate": 4.910816896445628e-06, "loss": 0.0162, "step": 1848 }, { "epoch": 1.36, "learning_rate": 4.900561296532365e-06, "loss": 0.0184, "step": 1849 }, { "epoch": 1.36, "learning_rate": 4.890312939832119e-06, "loss": 0.0248, "step": 1850 }, { "epoch": 1.36, "learning_rate": 4.880071840901607e-06, "loss": 0.0314, "step": 1851 }, { "epoch": 1.36, "learning_rate": 4.869838014287217e-06, "loss": 0.0279, "step": 1852 }, { "epoch": 1.36, "learning_rate": 4.85961147452502e-06, "loss": 0.0411, "step": 1853 }, { "epoch": 1.36, "learning_rate": 4.849392236140734e-06, "loss": 0.0343, "step": 1854 }, { "epoch": 1.36, "learning_rate": 4.839180313649709e-06, "loss": 0.0186, "step": 1855 }, { "epoch": 1.37, "learning_rate": 4.828975721556895e-06, "loss": 0.031, "step": 1856 }, { "epoch": 1.37, "learning_rate": 4.818778474356839e-06, "loss": 0.0289, "step": 1857 }, { "epoch": 1.37, "learning_rate": 4.808588586533646e-06, "loss": 0.0268, "step": 1858 }, { "epoch": 1.37, "learning_rate": 4.798406072560977e-06, "loss": 0.0296, "step": 1859 }, { "epoch": 1.37, "learning_rate": 4.788230946902015e-06, "loss": 0.0335, "step": 1860 }, { "epoch": 1.37, "learning_rate": 4.778063224009453e-06, "loss": 0.0246, "step": 1861 }, { "epoch": 1.37, "learning_rate": 4.76790291832546e-06, "loss": 0.0202, "step": 1862 }, { "epoch": 1.37, "learning_rate": 4.757750044281679e-06, "loss": 0.0231, "step": 1863 }, { "epoch": 1.37, "learning_rate": 4.747604616299189e-06, "loss": 0.0299, "step": 1864 }, { "epoch": 1.37, "learning_rate": 4.737466648788497e-06, "loss": 0.0322, "step": 1865 }, { "epoch": 1.37, "learning_rate": 4.727336156149516e-06, "loss": 0.0359, "step": 1866 }, { "epoch": 1.37, "learning_rate": 4.71721315277154e-06, "loss": 0.0278, "step": 1867 }, { "epoch": 1.37, "learning_rate": 4.707097653033219e-06, "loss": 0.0296, "step": 1868 }, { "epoch": 1.38, "learning_rate": 4.6969896713025555e-06, "loss": 0.0373, "step": 1869 }, { "epoch": 1.38, "learning_rate": 4.686889221936861e-06, "loss": 0.0287, "step": 1870 }, { "epoch": 1.38, "learning_rate": 4.67679631928276e-06, "loss": 0.0192, "step": 1871 }, { "epoch": 1.38, "learning_rate": 4.66671097767615e-06, "loss": 0.0209, "step": 1872 }, { "epoch": 1.38, "learning_rate": 4.656633211442196e-06, "loss": 0.0357, "step": 1873 }, { "epoch": 1.38, "learning_rate": 4.646563034895293e-06, "loss": 0.0282, "step": 1874 }, { "epoch": 1.38, "learning_rate": 4.636500462339066e-06, "loss": 0.0231, "step": 1875 }, { "epoch": 1.38, "learning_rate": 4.626445508066329e-06, "loss": 0.0254, "step": 1876 }, { "epoch": 1.38, "learning_rate": 4.616398186359083e-06, "loss": 0.0221, "step": 1877 }, { "epoch": 1.38, "learning_rate": 4.606358511488486e-06, "loss": 0.0309, "step": 1878 }, { "epoch": 1.38, "learning_rate": 4.596326497714835e-06, "loss": 0.0258, "step": 1879 }, { "epoch": 1.38, "learning_rate": 4.58630215928754e-06, "loss": 0.0341, "step": 1880 }, { "epoch": 1.38, "learning_rate": 4.576285510445117e-06, "loss": 0.0231, "step": 1881 }, { "epoch": 1.38, "learning_rate": 4.566276565415152e-06, "loss": 0.0385, "step": 1882 }, { "epoch": 1.39, "learning_rate": 4.556275338414296e-06, "loss": 0.0323, "step": 1883 }, { "epoch": 1.39, "learning_rate": 4.5462818436482245e-06, "loss": 0.0213, "step": 1884 }, { "epoch": 1.39, "learning_rate": 4.536296095311653e-06, "loss": 0.0326, "step": 1885 }, { "epoch": 1.39, "learning_rate": 4.52631810758827e-06, "loss": 0.034, "step": 1886 }, { "epoch": 1.39, "learning_rate": 4.516347894650759e-06, "loss": 0.0281, "step": 1887 }, { "epoch": 1.39, "learning_rate": 4.506385470660742e-06, "loss": 0.0311, "step": 1888 }, { "epoch": 1.39, "learning_rate": 4.4964308497687995e-06, "loss": 0.0282, "step": 1889 }, { "epoch": 1.39, "learning_rate": 4.486484046114403e-06, "loss": 0.0441, "step": 1890 }, { "epoch": 1.39, "learning_rate": 4.476545073825951e-06, "loss": 0.0357, "step": 1891 }, { "epoch": 1.39, "learning_rate": 4.466613947020689e-06, "loss": 0.0344, "step": 1892 }, { "epoch": 1.39, "learning_rate": 4.45669067980474e-06, "loss": 0.0236, "step": 1893 }, { "epoch": 1.39, "learning_rate": 4.4467752862730485e-06, "loss": 0.0255, "step": 1894 }, { "epoch": 1.39, "learning_rate": 4.436867780509387e-06, "loss": 0.0259, "step": 1895 }, { "epoch": 1.4, "learning_rate": 4.42696817658631e-06, "loss": 0.0437, "step": 1896 }, { "epoch": 1.4, "learning_rate": 4.417076488565171e-06, "loss": 0.0255, "step": 1897 }, { "epoch": 1.4, "learning_rate": 4.4071927304960534e-06, "loss": 0.0266, "step": 1898 }, { "epoch": 1.4, "learning_rate": 4.3973169164178e-06, "loss": 0.0284, "step": 1899 }, { "epoch": 1.4, "learning_rate": 4.38744906035795e-06, "loss": 0.035, "step": 1900 }, { "epoch": 1.4, "learning_rate": 4.377589176332757e-06, "loss": 0.0305, "step": 1901 }, { "epoch": 1.4, "learning_rate": 4.367737278347136e-06, "loss": 0.0248, "step": 1902 }, { "epoch": 1.4, "learning_rate": 4.357893380394668e-06, "loss": 0.033, "step": 1903 }, { "epoch": 1.4, "learning_rate": 4.348057496457567e-06, "loss": 0.041, "step": 1904 }, { "epoch": 1.4, "learning_rate": 4.338229640506672e-06, "loss": 0.0341, "step": 1905 }, { "epoch": 1.4, "learning_rate": 4.328409826501403e-06, "loss": 0.0333, "step": 1906 }, { "epoch": 1.4, "learning_rate": 4.3185980683897756e-06, "loss": 0.0255, "step": 1907 }, { "epoch": 1.4, "learning_rate": 4.3087943801083445e-06, "loss": 0.0315, "step": 1908 }, { "epoch": 1.4, "learning_rate": 4.298998775582216e-06, "loss": 0.0224, "step": 1909 }, { "epoch": 1.41, "learning_rate": 4.289211268725009e-06, "loss": 0.033, "step": 1910 }, { "epoch": 1.41, "learning_rate": 4.279431873438846e-06, "loss": 0.0187, "step": 1911 }, { "epoch": 1.41, "learning_rate": 4.269660603614316e-06, "loss": 0.0394, "step": 1912 }, { "epoch": 1.41, "learning_rate": 4.25989747313048e-06, "loss": 0.0305, "step": 1913 }, { "epoch": 1.41, "learning_rate": 4.250142495854825e-06, "loss": 0.0349, "step": 1914 }, { "epoch": 1.41, "learning_rate": 4.240395685643266e-06, "loss": 0.0278, "step": 1915 }, { "epoch": 1.41, "learning_rate": 4.2306570563401185e-06, "loss": 0.0372, "step": 1916 }, { "epoch": 1.41, "learning_rate": 4.220926621778076e-06, "loss": 0.0312, "step": 1917 }, { "epoch": 1.41, "learning_rate": 4.211204395778183e-06, "loss": 0.0358, "step": 1918 }, { "epoch": 1.41, "learning_rate": 4.201490392149843e-06, "loss": 0.0388, "step": 1919 }, { "epoch": 1.41, "learning_rate": 4.19178462469076e-06, "loss": 0.0275, "step": 1920 }, { "epoch": 1.41, "learning_rate": 4.1820871071869555e-06, "loss": 0.0301, "step": 1921 }, { "epoch": 1.41, "learning_rate": 4.172397853412725e-06, "loss": 0.0442, "step": 1922 }, { "epoch": 1.42, "learning_rate": 4.162716877130633e-06, "loss": 0.0389, "step": 1923 }, { "epoch": 1.42, "learning_rate": 4.1530441920914746e-06, "loss": 0.0319, "step": 1924 }, { "epoch": 1.42, "learning_rate": 4.14337981203428e-06, "loss": 0.0247, "step": 1925 }, { "epoch": 1.42, "learning_rate": 4.1337237506862744e-06, "loss": 0.0469, "step": 1926 }, { "epoch": 1.42, "learning_rate": 4.124076021762874e-06, "loss": 0.042, "step": 1927 }, { "epoch": 1.42, "learning_rate": 4.114436638967656e-06, "loss": 0.0297, "step": 1928 }, { "epoch": 1.42, "learning_rate": 4.104805615992342e-06, "loss": 0.0276, "step": 1929 }, { "epoch": 1.42, "learning_rate": 4.095182966516787e-06, "loss": 0.0382, "step": 1930 }, { "epoch": 1.42, "learning_rate": 4.08556870420894e-06, "loss": 0.0273, "step": 1931 }, { "epoch": 1.42, "learning_rate": 4.075962842724847e-06, "loss": 0.0374, "step": 1932 }, { "epoch": 1.42, "learning_rate": 4.066365395708611e-06, "loss": 0.0284, "step": 1933 }, { "epoch": 1.42, "learning_rate": 4.0567763767923965e-06, "loss": 0.0323, "step": 1934 }, { "epoch": 1.42, "learning_rate": 4.047195799596387e-06, "loss": 0.0263, "step": 1935 }, { "epoch": 1.42, "learning_rate": 4.037623677728783e-06, "loss": 0.031, "step": 1936 }, { "epoch": 1.43, "learning_rate": 4.028060024785765e-06, "loss": 0.0297, "step": 1937 }, { "epoch": 1.43, "learning_rate": 4.018504854351495e-06, "loss": 0.03, "step": 1938 }, { "epoch": 1.43, "learning_rate": 4.008958179998075e-06, "loss": 0.0339, "step": 1939 }, { "epoch": 1.43, "learning_rate": 3.999420015285549e-06, "loss": 0.0215, "step": 1940 }, { "epoch": 1.43, "learning_rate": 3.989890373761871e-06, "loss": 0.0234, "step": 1941 }, { "epoch": 1.43, "learning_rate": 3.980369268962893e-06, "loss": 0.0198, "step": 1942 }, { "epoch": 1.43, "learning_rate": 3.970856714412325e-06, "loss": 0.0386, "step": 1943 }, { "epoch": 1.43, "learning_rate": 3.961352723621757e-06, "loss": 0.0309, "step": 1944 }, { "epoch": 1.43, "learning_rate": 3.9518573100905925e-06, "loss": 0.0219, "step": 1945 }, { "epoch": 1.43, "learning_rate": 3.942370487306064e-06, "loss": 0.0282, "step": 1946 }, { "epoch": 1.43, "learning_rate": 3.9328922687432015e-06, "loss": 0.0288, "step": 1947 }, { "epoch": 1.43, "learning_rate": 3.923422667864814e-06, "loss": 0.0454, "step": 1948 }, { "epoch": 1.43, "learning_rate": 3.91396169812146e-06, "loss": 0.0327, "step": 1949 }, { "epoch": 1.43, "learning_rate": 3.904509372951453e-06, "loss": 0.0344, "step": 1950 }, { "epoch": 1.44, "learning_rate": 3.8950657057808125e-06, "loss": 0.0218, "step": 1951 }, { "epoch": 1.44, "learning_rate": 3.885630710023275e-06, "loss": 0.0285, "step": 1952 }, { "epoch": 1.44, "learning_rate": 3.87620439908025e-06, "loss": 0.0289, "step": 1953 }, { "epoch": 1.44, "learning_rate": 3.866786786340821e-06, "loss": 0.0306, "step": 1954 }, { "epoch": 1.44, "learning_rate": 3.8573778851817e-06, "loss": 0.0283, "step": 1955 }, { "epoch": 1.44, "learning_rate": 3.847977708967246e-06, "loss": 0.0246, "step": 1956 }, { "epoch": 1.44, "learning_rate": 3.838586271049405e-06, "loss": 0.0217, "step": 1957 }, { "epoch": 1.44, "learning_rate": 3.829203584767724e-06, "loss": 0.0318, "step": 1958 }, { "epoch": 1.44, "learning_rate": 3.81982966344932e-06, "loss": 0.0284, "step": 1959 }, { "epoch": 1.44, "learning_rate": 3.810464520408853e-06, "loss": 0.0212, "step": 1960 }, { "epoch": 1.44, "learning_rate": 3.8011081689485163e-06, "loss": 0.0319, "step": 1961 }, { "epoch": 1.44, "learning_rate": 3.7917606223580217e-06, "loss": 0.0342, "step": 1962 }, { "epoch": 1.44, "learning_rate": 3.7824218939145628e-06, "loss": 0.0309, "step": 1963 }, { "epoch": 1.45, "learning_rate": 3.7730919968828194e-06, "loss": 0.0309, "step": 1964 }, { "epoch": 1.45, "learning_rate": 3.7637709445149227e-06, "loss": 0.033, "step": 1965 }, { "epoch": 1.45, "learning_rate": 3.754458750050445e-06, "loss": 0.0274, "step": 1966 }, { "epoch": 1.45, "learning_rate": 3.7451554267163638e-06, "loss": 0.0294, "step": 1967 }, { "epoch": 1.45, "learning_rate": 3.7358609877270746e-06, "loss": 0.0285, "step": 1968 }, { "epoch": 1.45, "learning_rate": 3.7265754462843372e-06, "loss": 0.0242, "step": 1969 }, { "epoch": 1.45, "learning_rate": 3.717298815577284e-06, "loss": 0.0214, "step": 1970 }, { "epoch": 1.45, "learning_rate": 3.7080311087823862e-06, "loss": 0.0312, "step": 1971 }, { "epoch": 1.45, "learning_rate": 3.6987723390634447e-06, "loss": 0.0289, "step": 1972 }, { "epoch": 1.45, "learning_rate": 3.6895225195715546e-06, "loss": 0.0198, "step": 1973 }, { "epoch": 1.45, "learning_rate": 3.6802816634451144e-06, "loss": 0.0236, "step": 1974 }, { "epoch": 1.45, "learning_rate": 3.6710497838097723e-06, "loss": 0.0229, "step": 1975 }, { "epoch": 1.45, "learning_rate": 3.661826893778443e-06, "loss": 0.0288, "step": 1976 }, { "epoch": 1.45, "learning_rate": 3.6526130064512644e-06, "loss": 0.0242, "step": 1977 }, { "epoch": 1.46, "learning_rate": 3.6434081349155903e-06, "loss": 0.0348, "step": 1978 }, { "epoch": 1.46, "learning_rate": 3.6342122922459623e-06, "loss": 0.033, "step": 1979 }, { "epoch": 1.46, "learning_rate": 3.6250254915041073e-06, "loss": 0.0288, "step": 1980 }, { "epoch": 1.46, "learning_rate": 3.615847745738896e-06, "loss": 0.0245, "step": 1981 }, { "epoch": 1.46, "learning_rate": 3.6066790679863505e-06, "loss": 0.0273, "step": 1982 }, { "epoch": 1.46, "learning_rate": 3.597519471269606e-06, "loss": 0.0201, "step": 1983 }, { "epoch": 1.46, "learning_rate": 3.588368968598903e-06, "loss": 0.0268, "step": 1984 }, { "epoch": 1.46, "learning_rate": 3.5792275729715566e-06, "loss": 0.027, "step": 1985 }, { "epoch": 1.46, "learning_rate": 3.5700952973719573e-06, "loss": 0.0295, "step": 1986 }, { "epoch": 1.46, "learning_rate": 3.5609721547715313e-06, "loss": 0.0271, "step": 1987 }, { "epoch": 1.46, "learning_rate": 3.551858158128739e-06, "loss": 0.0327, "step": 1988 }, { "epoch": 1.46, "learning_rate": 3.5427533203890484e-06, "loss": 0.0265, "step": 1989 }, { "epoch": 1.46, "learning_rate": 3.533657654484922e-06, "loss": 0.0327, "step": 1990 }, { "epoch": 1.47, "learning_rate": 3.5245711733357846e-06, "loss": 0.0291, "step": 1991 }, { "epoch": 1.47, "learning_rate": 3.515493889848025e-06, "loss": 0.0445, "step": 1992 }, { "epoch": 1.47, "learning_rate": 3.50642581691496e-06, "loss": 0.0285, "step": 1993 }, { "epoch": 1.47, "learning_rate": 3.49736696741683e-06, "loss": 0.0324, "step": 1994 }, { "epoch": 1.47, "learning_rate": 3.4883173542207704e-06, "loss": 0.0261, "step": 1995 }, { "epoch": 1.47, "learning_rate": 3.4792769901808043e-06, "loss": 0.0257, "step": 1996 }, { "epoch": 1.47, "learning_rate": 3.4702458881378053e-06, "loss": 0.0336, "step": 1997 }, { "epoch": 1.47, "learning_rate": 3.4612240609195034e-06, "loss": 0.0437, "step": 1998 }, { "epoch": 1.47, "learning_rate": 3.452211521340445e-06, "loss": 0.0227, "step": 1999 }, { "epoch": 1.47, "learning_rate": 3.443208282201994e-06, "loss": 0.0221, "step": 2000 }, { "epoch": 1.47, "learning_rate": 3.434214356292288e-06, "loss": 0.0306, "step": 2001 }, { "epoch": 1.47, "learning_rate": 3.4252297563862625e-06, "loss": 0.0434, "step": 2002 }, { "epoch": 1.47, "learning_rate": 3.416254495245579e-06, "loss": 0.0287, "step": 2003 }, { "epoch": 1.47, "learning_rate": 3.407288585618654e-06, "loss": 0.0283, "step": 2004 }, { "epoch": 1.48, "learning_rate": 3.398332040240606e-06, "loss": 0.0405, "step": 2005 }, { "epoch": 1.48, "learning_rate": 3.3893848718332665e-06, "loss": 0.0268, "step": 2006 }, { "epoch": 1.48, "learning_rate": 3.3804470931051315e-06, "loss": 0.0251, "step": 2007 }, { "epoch": 1.48, "learning_rate": 3.371518716751383e-06, "loss": 0.032, "step": 2008 }, { "epoch": 1.48, "learning_rate": 3.362599755453825e-06, "loss": 0.0304, "step": 2009 }, { "epoch": 1.48, "learning_rate": 3.3536902218809043e-06, "loss": 0.0341, "step": 2010 }, { "epoch": 1.48, "learning_rate": 3.344790128687665e-06, "loss": 0.0246, "step": 2011 }, { "epoch": 1.48, "learning_rate": 3.3358994885157537e-06, "loss": 0.0335, "step": 2012 }, { "epoch": 1.48, "learning_rate": 3.3270183139933764e-06, "loss": 0.03, "step": 2013 }, { "epoch": 1.48, "learning_rate": 3.318146617735306e-06, "loss": 0.031, "step": 2014 }, { "epoch": 1.48, "learning_rate": 3.309284412342848e-06, "loss": 0.0202, "step": 2015 }, { "epoch": 1.48, "learning_rate": 3.3004317104038296e-06, "loss": 0.0264, "step": 2016 }, { "epoch": 1.48, "learning_rate": 3.2915885244925715e-06, "loss": 0.0419, "step": 2017 }, { "epoch": 1.48, "learning_rate": 3.2827548671698907e-06, "loss": 0.0295, "step": 2018 }, { "epoch": 1.49, "learning_rate": 3.2739307509830542e-06, "loss": 0.0319, "step": 2019 }, { "epoch": 1.49, "learning_rate": 3.26511618846579e-06, "loss": 0.0231, "step": 2020 }, { "epoch": 1.49, "learning_rate": 3.2563111921382506e-06, "loss": 0.019, "step": 2021 }, { "epoch": 1.49, "learning_rate": 3.247515774507005e-06, "loss": 0.0301, "step": 2022 }, { "epoch": 1.49, "learning_rate": 3.238729948065007e-06, "loss": 0.0274, "step": 2023 }, { "epoch": 1.49, "learning_rate": 3.2299537252915993e-06, "loss": 0.0246, "step": 2024 }, { "epoch": 1.49, "learning_rate": 3.22118711865247e-06, "loss": 0.0313, "step": 2025 }, { "epoch": 1.49, "learning_rate": 3.2124301405996616e-06, "loss": 0.0251, "step": 2026 }, { "epoch": 1.49, "learning_rate": 3.203682803571533e-06, "loss": 0.0251, "step": 2027 }, { "epoch": 1.49, "learning_rate": 3.194945119992755e-06, "loss": 0.0481, "step": 2028 }, { "epoch": 1.49, "learning_rate": 3.186217102274276e-06, "loss": 0.0262, "step": 2029 }, { "epoch": 1.49, "learning_rate": 3.177498762813327e-06, "loss": 0.0189, "step": 2030 }, { "epoch": 1.49, "learning_rate": 3.1687901139933796e-06, "loss": 0.029, "step": 2031 }, { "epoch": 1.5, "learning_rate": 3.160091168184154e-06, "loss": 0.0281, "step": 2032 }, { "epoch": 1.5, "learning_rate": 3.1514019377415784e-06, "loss": 0.0221, "step": 2033 }, { "epoch": 1.5, "learning_rate": 3.142722435007791e-06, "loss": 0.0249, "step": 2034 }, { "epoch": 1.5, "learning_rate": 3.1340526723110996e-06, "loss": 0.0369, "step": 2035 }, { "epoch": 1.5, "learning_rate": 3.1253926619659912e-06, "loss": 0.0335, "step": 2036 }, { "epoch": 1.5, "learning_rate": 3.116742416273089e-06, "loss": 0.0256, "step": 2037 }, { "epoch": 1.5, "learning_rate": 3.108101947519151e-06, "loss": 0.024, "step": 2038 }, { "epoch": 1.5, "learning_rate": 3.0994712679770523e-06, "loss": 0.0363, "step": 2039 }, { "epoch": 1.5, "learning_rate": 3.0908503899057605e-06, "loss": 0.0265, "step": 2040 }, { "epoch": 1.5, "learning_rate": 3.082239325550317e-06, "loss": 0.0246, "step": 2041 }, { "epoch": 1.5, "learning_rate": 3.0736380871418305e-06, "loss": 0.0444, "step": 2042 }, { "epoch": 1.5, "learning_rate": 3.065046686897446e-06, "loss": 0.0263, "step": 2043 }, { "epoch": 1.5, "learning_rate": 3.0564651370203414e-06, "loss": 0.0256, "step": 2044 }, { "epoch": 1.5, "learning_rate": 3.047893449699699e-06, "loss": 0.0252, "step": 2045 }, { "epoch": 1.51, "learning_rate": 3.039331637110697e-06, "loss": 0.0272, "step": 2046 }, { "epoch": 1.51, "learning_rate": 3.0307797114144787e-06, "loss": 0.0256, "step": 2047 }, { "epoch": 1.51, "learning_rate": 3.0222376847581546e-06, "loss": 0.0321, "step": 2048 }, { "epoch": 1.51, "learning_rate": 3.013705569274764e-06, "loss": 0.0276, "step": 2049 }, { "epoch": 1.51, "learning_rate": 3.005183377083277e-06, "loss": 0.0348, "step": 2050 }, { "epoch": 1.51, "learning_rate": 2.996671120288566e-06, "loss": 0.0296, "step": 2051 }, { "epoch": 1.51, "learning_rate": 2.9881688109813933e-06, "loss": 0.0254, "step": 2052 }, { "epoch": 1.51, "learning_rate": 2.9796764612383844e-06, "loss": 0.0157, "step": 2053 }, { "epoch": 1.51, "learning_rate": 2.971194083122029e-06, "loss": 0.0231, "step": 2054 }, { "epoch": 1.51, "learning_rate": 2.962721688680641e-06, "loss": 0.0195, "step": 2055 }, { "epoch": 1.51, "learning_rate": 2.9542592899483633e-06, "loss": 0.0257, "step": 2056 }, { "epoch": 1.51, "learning_rate": 2.9458068989451393e-06, "loss": 0.0219, "step": 2057 }, { "epoch": 1.51, "learning_rate": 2.937364527676697e-06, "loss": 0.025, "step": 2058 }, { "epoch": 1.52, "learning_rate": 2.9289321881345257e-06, "loss": 0.0219, "step": 2059 }, { "epoch": 1.52, "learning_rate": 2.920509892295875e-06, "loss": 0.0309, "step": 2060 }, { "epoch": 1.52, "learning_rate": 2.9120976521237278e-06, "loss": 0.0162, "step": 2061 }, { "epoch": 1.52, "learning_rate": 2.903695479566774e-06, "loss": 0.023, "step": 2062 }, { "epoch": 1.52, "learning_rate": 2.895303386559414e-06, "loss": 0.0294, "step": 2063 }, { "epoch": 1.52, "learning_rate": 2.886921385021729e-06, "loss": 0.0304, "step": 2064 }, { "epoch": 1.52, "learning_rate": 2.8785494868594667e-06, "loss": 0.0211, "step": 2065 }, { "epoch": 1.52, "learning_rate": 2.870187703964017e-06, "loss": 0.0201, "step": 2066 }, { "epoch": 1.52, "learning_rate": 2.861836048212414e-06, "loss": 0.0212, "step": 2067 }, { "epoch": 1.52, "learning_rate": 2.8534945314672946e-06, "loss": 0.0254, "step": 2068 }, { "epoch": 1.52, "learning_rate": 2.845163165576903e-06, "loss": 0.0227, "step": 2069 }, { "epoch": 1.52, "learning_rate": 2.8368419623750633e-06, "loss": 0.0245, "step": 2070 }, { "epoch": 1.52, "learning_rate": 2.8285309336811674e-06, "loss": 0.0369, "step": 2071 }, { "epoch": 1.52, "learning_rate": 2.8202300913001445e-06, "loss": 0.0252, "step": 2072 }, { "epoch": 1.53, "learning_rate": 2.8119394470224703e-06, "loss": 0.0311, "step": 2073 }, { "epoch": 1.53, "learning_rate": 2.8036590126241226e-06, "loss": 0.0225, "step": 2074 }, { "epoch": 1.53, "learning_rate": 2.7953887998665817e-06, "loss": 0.0292, "step": 2075 }, { "epoch": 1.53, "learning_rate": 2.7871288204968127e-06, "loss": 0.0266, "step": 2076 }, { "epoch": 1.53, "learning_rate": 2.7788790862472446e-06, "loss": 0.0332, "step": 2077 }, { "epoch": 1.53, "learning_rate": 2.7706396088357444e-06, "loss": 0.0321, "step": 2078 }, { "epoch": 1.53, "learning_rate": 2.762410399965625e-06, "loss": 0.0453, "step": 2079 }, { "epoch": 1.53, "learning_rate": 2.754191471325601e-06, "loss": 0.0299, "step": 2080 }, { "epoch": 1.53, "learning_rate": 2.745982834589793e-06, "loss": 0.0314, "step": 2081 }, { "epoch": 1.53, "learning_rate": 2.737784501417702e-06, "loss": 0.0334, "step": 2082 }, { "epoch": 1.53, "learning_rate": 2.729596483454193e-06, "loss": 0.0272, "step": 2083 }, { "epoch": 1.53, "learning_rate": 2.7214187923294766e-06, "loss": 0.0314, "step": 2084 }, { "epoch": 1.53, "learning_rate": 2.7132514396591e-06, "loss": 0.0257, "step": 2085 }, { "epoch": 1.53, "learning_rate": 2.70509443704392e-06, "loss": 0.0257, "step": 2086 }, { "epoch": 1.54, "learning_rate": 2.6969477960700985e-06, "loss": 0.0334, "step": 2087 }, { "epoch": 1.54, "learning_rate": 2.6888115283090754e-06, "loss": 0.0257, "step": 2088 }, { "epoch": 1.54, "learning_rate": 2.680685645317561e-06, "loss": 0.0271, "step": 2089 }, { "epoch": 1.54, "learning_rate": 2.6725701586375075e-06, "loss": 0.0231, "step": 2090 }, { "epoch": 1.54, "learning_rate": 2.6644650797961103e-06, "loss": 0.0235, "step": 2091 }, { "epoch": 1.54, "learning_rate": 2.6563704203057704e-06, "loss": 0.0335, "step": 2092 }, { "epoch": 1.54, "learning_rate": 2.6482861916640977e-06, "loss": 0.0301, "step": 2093 }, { "epoch": 1.54, "learning_rate": 2.6402124053538837e-06, "loss": 0.0378, "step": 2094 }, { "epoch": 1.54, "learning_rate": 2.6321490728430886e-06, "loss": 0.025, "step": 2095 }, { "epoch": 1.54, "learning_rate": 2.6240962055848196e-06, "loss": 0.0375, "step": 2096 }, { "epoch": 1.54, "learning_rate": 2.6160538150173274e-06, "loss": 0.0307, "step": 2097 }, { "epoch": 1.54, "learning_rate": 2.6080219125639703e-06, "loss": 0.0249, "step": 2098 }, { "epoch": 1.54, "learning_rate": 2.6000005096332204e-06, "loss": 0.0266, "step": 2099 }, { "epoch": 1.55, "learning_rate": 2.5919896176186287e-06, "loss": 0.0244, "step": 2100 }, { "epoch": 1.55, "learning_rate": 2.583989247898826e-06, "loss": 0.0257, "step": 2101 }, { "epoch": 1.55, "learning_rate": 2.575999411837481e-06, "loss": 0.0259, "step": 2102 }, { "epoch": 1.55, "learning_rate": 2.56802012078332e-06, "loss": 0.0326, "step": 2103 }, { "epoch": 1.55, "learning_rate": 2.560051386070073e-06, "loss": 0.038, "step": 2104 }, { "epoch": 1.55, "learning_rate": 2.552093219016488e-06, "loss": 0.035, "step": 2105 }, { "epoch": 1.55, "learning_rate": 2.5441456309263e-06, "loss": 0.033, "step": 2106 }, { "epoch": 1.55, "learning_rate": 2.5362086330882196e-06, "loss": 0.0178, "step": 2107 }, { "epoch": 1.55, "learning_rate": 2.5282822367759054e-06, "loss": 0.0186, "step": 2108 }, { "epoch": 1.55, "learning_rate": 2.52036645324797e-06, "loss": 0.0315, "step": 2109 }, { "epoch": 1.55, "learning_rate": 2.512461293747942e-06, "loss": 0.0233, "step": 2110 }, { "epoch": 1.55, "learning_rate": 2.50456676950427e-06, "loss": 0.0179, "step": 2111 }, { "epoch": 1.55, "learning_rate": 2.496682891730279e-06, "loss": 0.0338, "step": 2112 }, { "epoch": 1.55, "learning_rate": 2.488809671624196e-06, "loss": 0.0274, "step": 2113 }, { "epoch": 1.56, "learning_rate": 2.480947120369089e-06, "loss": 0.0148, "step": 2114 }, { "epoch": 1.56, "learning_rate": 2.4730952491328865e-06, "loss": 0.0228, "step": 2115 }, { "epoch": 1.56, "learning_rate": 2.4652540690683315e-06, "loss": 0.0507, "step": 2116 }, { "epoch": 1.56, "learning_rate": 2.4574235913129995e-06, "loss": 0.0277, "step": 2117 }, { "epoch": 1.56, "learning_rate": 2.4496038269892455e-06, "loss": 0.02, "step": 2118 }, { "epoch": 1.56, "learning_rate": 2.4417947872042293e-06, "loss": 0.0223, "step": 2119 }, { "epoch": 1.56, "learning_rate": 2.433996483049855e-06, "loss": 0.0253, "step": 2120 }, { "epoch": 1.56, "learning_rate": 2.4262089256027976e-06, "loss": 0.0329, "step": 2121 }, { "epoch": 1.56, "learning_rate": 2.418432125924449e-06, "loss": 0.03, "step": 2122 }, { "epoch": 1.56, "learning_rate": 2.4106660950609385e-06, "loss": 0.0257, "step": 2123 }, { "epoch": 1.56, "learning_rate": 2.4029108440430838e-06, "loss": 0.0313, "step": 2124 }, { "epoch": 1.56, "learning_rate": 2.3951663838864014e-06, "loss": 0.0256, "step": 2125 }, { "epoch": 1.56, "learning_rate": 2.387432725591078e-06, "loss": 0.026, "step": 2126 }, { "epoch": 1.57, "learning_rate": 2.3797098801419594e-06, "loss": 0.0307, "step": 2127 }, { "epoch": 1.57, "learning_rate": 2.3719978585085234e-06, "loss": 0.0265, "step": 2128 }, { "epoch": 1.57, "learning_rate": 2.364296671644889e-06, "loss": 0.0242, "step": 2129 }, { "epoch": 1.57, "learning_rate": 2.356606330489769e-06, "loss": 0.0238, "step": 2130 }, { "epoch": 1.57, "learning_rate": 2.3489268459664827e-06, "loss": 0.0224, "step": 2131 }, { "epoch": 1.57, "learning_rate": 2.3412582289829254e-06, "loss": 0.0307, "step": 2132 }, { "epoch": 1.57, "learning_rate": 2.333600490431558e-06, "loss": 0.0219, "step": 2133 }, { "epoch": 1.57, "learning_rate": 2.3259536411893836e-06, "loss": 0.0241, "step": 2134 }, { "epoch": 1.57, "learning_rate": 2.3183176921179473e-06, "loss": 0.033, "step": 2135 }, { "epoch": 1.57, "learning_rate": 2.3106926540633e-06, "loss": 0.0206, "step": 2136 }, { "epoch": 1.57, "learning_rate": 2.303078537856005e-06, "loss": 0.0252, "step": 2137 }, { "epoch": 1.57, "learning_rate": 2.2954753543111097e-06, "loss": 0.0181, "step": 2138 }, { "epoch": 1.57, "learning_rate": 2.2878831142281323e-06, "loss": 0.0249, "step": 2139 }, { "epoch": 1.57, "learning_rate": 2.2803018283910415e-06, "loss": 0.0259, "step": 2140 }, { "epoch": 1.58, "learning_rate": 2.2727315075682586e-06, "loss": 0.0231, "step": 2141 }, { "epoch": 1.58, "learning_rate": 2.2651721625126167e-06, "loss": 0.0314, "step": 2142 }, { "epoch": 1.58, "learning_rate": 2.2576238039613674e-06, "loss": 0.0236, "step": 2143 }, { "epoch": 1.58, "learning_rate": 2.2500864426361556e-06, "loss": 0.0249, "step": 2144 }, { "epoch": 1.58, "learning_rate": 2.242560089243009e-06, "loss": 0.0249, "step": 2145 }, { "epoch": 1.58, "learning_rate": 2.23504475447231e-06, "loss": 0.0187, "step": 2146 }, { "epoch": 1.58, "learning_rate": 2.2275404489988005e-06, "loss": 0.0286, "step": 2147 }, { "epoch": 1.58, "learning_rate": 2.2200471834815497e-06, "loss": 0.0351, "step": 2148 }, { "epoch": 1.58, "learning_rate": 2.2125649685639484e-06, "loss": 0.0277, "step": 2149 }, { "epoch": 1.58, "learning_rate": 2.2050938148736934e-06, "loss": 0.0269, "step": 2150 }, { "epoch": 1.58, "learning_rate": 2.197633733022768e-06, "loss": 0.015, "step": 2151 }, { "epoch": 1.58, "learning_rate": 2.1901847336074258e-06, "loss": 0.0457, "step": 2152 }, { "epoch": 1.58, "learning_rate": 2.1827468272081863e-06, "loss": 0.0401, "step": 2153 }, { "epoch": 1.58, "learning_rate": 2.1753200243898032e-06, "loss": 0.0335, "step": 2154 }, { "epoch": 1.59, "learning_rate": 2.1679043357012677e-06, "loss": 0.0143, "step": 2155 }, { "epoch": 1.59, "learning_rate": 2.160499771675778e-06, "loss": 0.0171, "step": 2156 }, { "epoch": 1.59, "learning_rate": 2.1531063428307395e-06, "loss": 0.0305, "step": 2157 }, { "epoch": 1.59, "learning_rate": 2.14572405966773e-06, "loss": 0.0208, "step": 2158 }, { "epoch": 1.59, "learning_rate": 2.138352932672504e-06, "loss": 0.0263, "step": 2159 }, { "epoch": 1.59, "learning_rate": 2.130992972314965e-06, "loss": 0.0405, "step": 2160 }, { "epoch": 1.59, "learning_rate": 2.1236441890491575e-06, "loss": 0.0292, "step": 2161 }, { "epoch": 1.59, "learning_rate": 2.1163065933132544e-06, "loss": 0.0304, "step": 2162 }, { "epoch": 1.59, "learning_rate": 2.108980195529533e-06, "loss": 0.0389, "step": 2163 }, { "epoch": 1.59, "learning_rate": 2.101665006104362e-06, "loss": 0.0334, "step": 2164 }, { "epoch": 1.59, "learning_rate": 2.0943610354281997e-06, "loss": 0.0312, "step": 2165 }, { "epoch": 1.59, "learning_rate": 2.087068293875557e-06, "loss": 0.0197, "step": 2166 }, { "epoch": 1.59, "learning_rate": 2.0797867918050053e-06, "loss": 0.0226, "step": 2167 }, { "epoch": 1.6, "learning_rate": 2.0725165395591472e-06, "loss": 0.049, "step": 2168 }, { "epoch": 1.6, "learning_rate": 2.0652575474646096e-06, "loss": 0.0188, "step": 2169 }, { "epoch": 1.6, "learning_rate": 2.0580098258320167e-06, "loss": 0.0246, "step": 2170 }, { "epoch": 1.6, "learning_rate": 2.050773384955995e-06, "loss": 0.0315, "step": 2171 }, { "epoch": 1.6, "learning_rate": 2.043548235115139e-06, "loss": 0.0255, "step": 2172 }, { "epoch": 1.6, "learning_rate": 2.036334386572012e-06, "loss": 0.0287, "step": 2173 }, { "epoch": 1.6, "learning_rate": 2.0291318495731215e-06, "loss": 0.0292, "step": 2174 }, { "epoch": 1.6, "learning_rate": 2.0219406343489122e-06, "loss": 0.0323, "step": 2175 }, { "epoch": 1.6, "learning_rate": 2.014760751113738e-06, "loss": 0.0262, "step": 2176 }, { "epoch": 1.6, "learning_rate": 2.0075922100658708e-06, "loss": 0.0302, "step": 2177 }, { "epoch": 1.6, "learning_rate": 2.000435021387457e-06, "loss": 0.0304, "step": 2178 }, { "epoch": 1.6, "learning_rate": 1.9932891952445286e-06, "loss": 0.0284, "step": 2179 }, { "epoch": 1.6, "learning_rate": 1.9861547417869776e-06, "loss": 0.0298, "step": 2180 }, { "epoch": 1.6, "learning_rate": 1.97903167114854e-06, "loss": 0.0293, "step": 2181 }, { "epoch": 1.61, "learning_rate": 1.9719199934467804e-06, "loss": 0.0313, "step": 2182 }, { "epoch": 1.61, "learning_rate": 1.9648197187830887e-06, "loss": 0.0179, "step": 2183 }, { "epoch": 1.61, "learning_rate": 1.957730857242649e-06, "loss": 0.0264, "step": 2184 }, { "epoch": 1.61, "learning_rate": 1.9506534188944406e-06, "loss": 0.0243, "step": 2185 }, { "epoch": 1.61, "learning_rate": 1.943587413791217e-06, "loss": 0.0257, "step": 2186 }, { "epoch": 1.61, "learning_rate": 1.9365328519694926e-06, "loss": 0.0196, "step": 2187 }, { "epoch": 1.61, "learning_rate": 1.9294897434495196e-06, "loss": 0.025, "step": 2188 }, { "epoch": 1.61, "learning_rate": 1.9224580982352956e-06, "loss": 0.0401, "step": 2189 }, { "epoch": 1.61, "learning_rate": 1.915437926314523e-06, "loss": 0.0316, "step": 2190 }, { "epoch": 1.61, "learning_rate": 1.9084292376586133e-06, "loss": 0.0353, "step": 2191 }, { "epoch": 1.61, "learning_rate": 1.9014320422226707e-06, "loss": 0.0247, "step": 2192 }, { "epoch": 1.61, "learning_rate": 1.8944463499454668e-06, "loss": 0.0205, "step": 2193 }, { "epoch": 1.61, "learning_rate": 1.8874721707494448e-06, "loss": 0.0376, "step": 2194 }, { "epoch": 1.62, "learning_rate": 1.8805095145406794e-06, "loss": 0.0448, "step": 2195 }, { "epoch": 1.62, "learning_rate": 1.8735583912088951e-06, "loss": 0.0275, "step": 2196 }, { "epoch": 1.62, "learning_rate": 1.8666188106274196e-06, "loss": 0.0193, "step": 2197 }, { "epoch": 1.62, "learning_rate": 1.8596907826531962e-06, "loss": 0.0291, "step": 2198 }, { "epoch": 1.62, "learning_rate": 1.8527743171267566e-06, "loss": 0.0181, "step": 2199 }, { "epoch": 1.62, "learning_rate": 1.8458694238722086e-06, "loss": 0.0325, "step": 2200 }, { "epoch": 1.62, "learning_rate": 1.8389761126972183e-06, "loss": 0.0278, "step": 2201 }, { "epoch": 1.62, "learning_rate": 1.8320943933930103e-06, "loss": 0.0351, "step": 2202 }, { "epoch": 1.62, "learning_rate": 1.825224275734332e-06, "loss": 0.0376, "step": 2203 }, { "epoch": 1.62, "learning_rate": 1.818365769479462e-06, "loss": 0.0396, "step": 2204 }, { "epoch": 1.62, "learning_rate": 1.8115188843701837e-06, "loss": 0.0325, "step": 2205 }, { "epoch": 1.62, "learning_rate": 1.8046836301317727e-06, "loss": 0.0259, "step": 2206 }, { "epoch": 1.62, "learning_rate": 1.7978600164729821e-06, "loss": 0.0271, "step": 2207 }, { "epoch": 1.62, "learning_rate": 1.7910480530860363e-06, "loss": 0.0318, "step": 2208 }, { "epoch": 1.63, "learning_rate": 1.7842477496466037e-06, "loss": 0.0173, "step": 2209 }, { "epoch": 1.63, "learning_rate": 1.7774591158137977e-06, "loss": 0.043, "step": 2210 }, { "epoch": 1.63, "learning_rate": 1.770682161230156e-06, "loss": 0.0267, "step": 2211 }, { "epoch": 1.63, "learning_rate": 1.7639168955216257e-06, "loss": 0.0243, "step": 2212 }, { "epoch": 1.63, "learning_rate": 1.7571633282975475e-06, "loss": 0.0197, "step": 2213 }, { "epoch": 1.63, "learning_rate": 1.7504214691506527e-06, "loss": 0.0272, "step": 2214 }, { "epoch": 1.63, "learning_rate": 1.7436913276570345e-06, "loss": 0.0243, "step": 2215 }, { "epoch": 1.63, "learning_rate": 1.7369729133761493e-06, "loss": 0.021, "step": 2216 }, { "epoch": 1.63, "learning_rate": 1.7302662358507927e-06, "loss": 0.0367, "step": 2217 }, { "epoch": 1.63, "learning_rate": 1.7235713046070935e-06, "loss": 0.0279, "step": 2218 }, { "epoch": 1.63, "learning_rate": 1.716888129154488e-06, "loss": 0.0249, "step": 2219 }, { "epoch": 1.63, "learning_rate": 1.7102167189857255e-06, "loss": 0.0411, "step": 2220 }, { "epoch": 1.63, "learning_rate": 1.703557083576831e-06, "loss": 0.0248, "step": 2221 }, { "epoch": 1.64, "learning_rate": 1.6969092323871195e-06, "loss": 0.0302, "step": 2222 }, { "epoch": 1.64, "learning_rate": 1.6902731748591506e-06, "loss": 0.0188, "step": 2223 }, { "epoch": 1.64, "learning_rate": 1.6836489204187511e-06, "loss": 0.0348, "step": 2224 }, { "epoch": 1.64, "learning_rate": 1.6770364784749682e-06, "loss": 0.0241, "step": 2225 }, { "epoch": 1.64, "learning_rate": 1.6704358584200809e-06, "loss": 0.0339, "step": 2226 }, { "epoch": 1.64, "learning_rate": 1.6638470696295638e-06, "loss": 0.0394, "step": 2227 }, { "epoch": 1.64, "learning_rate": 1.6572701214621013e-06, "loss": 0.0167, "step": 2228 }, { "epoch": 1.64, "learning_rate": 1.6507050232595445e-06, "loss": 0.0323, "step": 2229 }, { "epoch": 1.64, "learning_rate": 1.6441517843469302e-06, "loss": 0.0249, "step": 2230 }, { "epoch": 1.64, "learning_rate": 1.6376104140324334e-06, "loss": 0.0172, "step": 2231 }, { "epoch": 1.64, "learning_rate": 1.631080921607383e-06, "loss": 0.0233, "step": 2232 }, { "epoch": 1.64, "learning_rate": 1.6245633163462283e-06, "loss": 0.0297, "step": 2233 }, { "epoch": 1.64, "learning_rate": 1.6180576075065412e-06, "loss": 0.0345, "step": 2234 }, { "epoch": 1.64, "learning_rate": 1.611563804328986e-06, "loss": 0.026, "step": 2235 }, { "epoch": 1.65, "learning_rate": 1.6050819160373331e-06, "loss": 0.0257, "step": 2236 }, { "epoch": 1.65, "learning_rate": 1.5986119518384102e-06, "loss": 0.0282, "step": 2237 }, { "epoch": 1.65, "learning_rate": 1.5921539209221238e-06, "loss": 0.0216, "step": 2238 }, { "epoch": 1.65, "learning_rate": 1.585707832461416e-06, "loss": 0.0306, "step": 2239 }, { "epoch": 1.65, "learning_rate": 1.5792736956122801e-06, "loss": 0.028, "step": 2240 }, { "epoch": 1.65, "learning_rate": 1.5728515195137206e-06, "loss": 0.0393, "step": 2241 }, { "epoch": 1.65, "learning_rate": 1.566441313287762e-06, "loss": 0.0338, "step": 2242 }, { "epoch": 1.65, "learning_rate": 1.5600430860394234e-06, "loss": 0.0257, "step": 2243 }, { "epoch": 1.65, "learning_rate": 1.5536568468567126e-06, "loss": 0.0292, "step": 2244 }, { "epoch": 1.65, "learning_rate": 1.5472826048106016e-06, "loss": 0.0291, "step": 2245 }, { "epoch": 1.65, "learning_rate": 1.5409203689550313e-06, "loss": 0.0217, "step": 2246 }, { "epoch": 1.65, "learning_rate": 1.5345701483268793e-06, "loss": 0.0347, "step": 2247 }, { "epoch": 1.65, "learning_rate": 1.5282319519459643e-06, "loss": 0.0249, "step": 2248 }, { "epoch": 1.65, "learning_rate": 1.5219057888150236e-06, "loss": 0.0227, "step": 2249 }, { "epoch": 1.66, "learning_rate": 1.5155916679197057e-06, "loss": 0.0315, "step": 2250 }, { "epoch": 1.66, "learning_rate": 1.509289598228546e-06, "loss": 0.0266, "step": 2251 }, { "epoch": 1.66, "learning_rate": 1.5029995886929717e-06, "loss": 0.0238, "step": 2252 }, { "epoch": 1.66, "learning_rate": 1.4967216482472713e-06, "loss": 0.0296, "step": 2253 }, { "epoch": 1.66, "learning_rate": 1.4904557858085967e-06, "loss": 0.0277, "step": 2254 }, { "epoch": 1.66, "learning_rate": 1.484202010276944e-06, "loss": 0.0252, "step": 2255 }, { "epoch": 1.66, "learning_rate": 1.4779603305351397e-06, "loss": 0.021, "step": 2256 }, { "epoch": 1.66, "learning_rate": 1.471730755448827e-06, "loss": 0.0331, "step": 2257 }, { "epoch": 1.66, "learning_rate": 1.4655132938664607e-06, "loss": 0.037, "step": 2258 }, { "epoch": 1.66, "learning_rate": 1.4593079546192834e-06, "loss": 0.0264, "step": 2259 }, { "epoch": 1.66, "learning_rate": 1.4531147465213247e-06, "loss": 0.0311, "step": 2260 }, { "epoch": 1.66, "learning_rate": 1.446933678369381e-06, "loss": 0.0201, "step": 2261 }, { "epoch": 1.66, "learning_rate": 1.4407647589430084e-06, "loss": 0.0284, "step": 2262 }, { "epoch": 1.67, "learning_rate": 1.4346079970045002e-06, "loss": 0.0337, "step": 2263 }, { "epoch": 1.67, "learning_rate": 1.4284634012988886e-06, "loss": 0.027, "step": 2264 }, { "epoch": 1.67, "learning_rate": 1.4223309805539164e-06, "loss": 0.0427, "step": 2265 }, { "epoch": 1.67, "learning_rate": 1.4162107434800422e-06, "loss": 0.0357, "step": 2266 }, { "epoch": 1.67, "learning_rate": 1.4101026987704147e-06, "loss": 0.0368, "step": 2267 }, { "epoch": 1.67, "learning_rate": 1.4040068551008658e-06, "loss": 0.03, "step": 2268 }, { "epoch": 1.67, "learning_rate": 1.3979232211298942e-06, "loss": 0.0295, "step": 2269 }, { "epoch": 1.67, "learning_rate": 1.3918518054986607e-06, "loss": 0.0215, "step": 2270 }, { "epoch": 1.67, "learning_rate": 1.3857926168309654e-06, "loss": 0.0263, "step": 2271 }, { "epoch": 1.67, "learning_rate": 1.3797456637332451e-06, "loss": 0.0256, "step": 2272 }, { "epoch": 1.67, "learning_rate": 1.3737109547945593e-06, "loss": 0.0297, "step": 2273 }, { "epoch": 1.67, "learning_rate": 1.3676884985865735e-06, "loss": 0.026, "step": 2274 }, { "epoch": 1.67, "learning_rate": 1.3616783036635461e-06, "loss": 0.0279, "step": 2275 }, { "epoch": 1.67, "learning_rate": 1.3556803785623274e-06, "loss": 0.0182, "step": 2276 }, { "epoch": 1.68, "learning_rate": 1.3496947318023302e-06, "loss": 0.0267, "step": 2277 }, { "epoch": 1.68, "learning_rate": 1.3437213718855347e-06, "loss": 0.026, "step": 2278 }, { "epoch": 1.68, "learning_rate": 1.3377603072964673e-06, "loss": 0.0232, "step": 2279 }, { "epoch": 1.68, "learning_rate": 1.3318115465021896e-06, "loss": 0.0142, "step": 2280 }, { "epoch": 1.68, "learning_rate": 1.325875097952285e-06, "loss": 0.0277, "step": 2281 }, { "epoch": 1.68, "learning_rate": 1.3199509700788527e-06, "loss": 0.0223, "step": 2282 }, { "epoch": 1.68, "learning_rate": 1.3140391712964862e-06, "loss": 0.0185, "step": 2283 }, { "epoch": 1.68, "learning_rate": 1.3081397100022718e-06, "loss": 0.0243, "step": 2284 }, { "epoch": 1.68, "learning_rate": 1.3022525945757692e-06, "loss": 0.0321, "step": 2285 }, { "epoch": 1.68, "learning_rate": 1.2963778333790067e-06, "loss": 0.0282, "step": 2286 }, { "epoch": 1.68, "learning_rate": 1.2905154347564564e-06, "loss": 0.0281, "step": 2287 }, { "epoch": 1.68, "learning_rate": 1.2846654070350372e-06, "loss": 0.0394, "step": 2288 }, { "epoch": 1.68, "learning_rate": 1.2788277585240928e-06, "loss": 0.028, "step": 2289 }, { "epoch": 1.69, "learning_rate": 1.2730024975153854e-06, "loss": 0.0255, "step": 2290 }, { "epoch": 1.69, "learning_rate": 1.2671896322830835e-06, "loss": 0.0214, "step": 2291 }, { "epoch": 1.69, "learning_rate": 1.2613891710837467e-06, "loss": 0.0211, "step": 2292 }, { "epoch": 1.69, "learning_rate": 1.2556011221563148e-06, "loss": 0.023, "step": 2293 }, { "epoch": 1.69, "learning_rate": 1.249825493722101e-06, "loss": 0.0162, "step": 2294 }, { "epoch": 1.69, "learning_rate": 1.2440622939847713e-06, "loss": 0.0225, "step": 2295 }, { "epoch": 1.69, "learning_rate": 1.2383115311303417e-06, "loss": 0.0285, "step": 2296 }, { "epoch": 1.69, "learning_rate": 1.232573213327164e-06, "loss": 0.0388, "step": 2297 }, { "epoch": 1.69, "learning_rate": 1.2268473487259124e-06, "loss": 0.0285, "step": 2298 }, { "epoch": 1.69, "learning_rate": 1.2211339454595673e-06, "loss": 0.035, "step": 2299 }, { "epoch": 1.69, "learning_rate": 1.2154330116434188e-06, "loss": 0.0327, "step": 2300 }, { "epoch": 1.69, "learning_rate": 1.2097445553750342e-06, "loss": 0.026, "step": 2301 }, { "epoch": 1.69, "learning_rate": 1.204068584734267e-06, "loss": 0.0247, "step": 2302 }, { "epoch": 1.69, "learning_rate": 1.1984051077832315e-06, "loss": 0.0223, "step": 2303 }, { "epoch": 1.7, "learning_rate": 1.1927541325663018e-06, "loss": 0.0354, "step": 2304 }, { "epoch": 1.7, "learning_rate": 1.1871156671100836e-06, "loss": 0.0324, "step": 2305 }, { "epoch": 1.7, "learning_rate": 1.1814897194234253e-06, "loss": 0.0182, "step": 2306 }, { "epoch": 1.7, "learning_rate": 1.175876297497387e-06, "loss": 0.03, "step": 2307 }, { "epoch": 1.7, "learning_rate": 1.1702754093052415e-06, "loss": 0.0271, "step": 2308 }, { "epoch": 1.7, "learning_rate": 1.1646870628024575e-06, "loss": 0.0373, "step": 2309 }, { "epoch": 1.7, "learning_rate": 1.1591112659266934e-06, "loss": 0.0326, "step": 2310 }, { "epoch": 1.7, "learning_rate": 1.1535480265977727e-06, "loss": 0.0148, "step": 2311 }, { "epoch": 1.7, "learning_rate": 1.1479973527176935e-06, "loss": 0.0208, "step": 2312 }, { "epoch": 1.7, "learning_rate": 1.1424592521705957e-06, "loss": 0.0173, "step": 2313 }, { "epoch": 1.7, "learning_rate": 1.136933732822768e-06, "loss": 0.0385, "step": 2314 }, { "epoch": 1.7, "learning_rate": 1.1314208025226258e-06, "loss": 0.0301, "step": 2315 }, { "epoch": 1.7, "learning_rate": 1.125920469100704e-06, "loss": 0.0232, "step": 2316 }, { "epoch": 1.7, "learning_rate": 1.1204327403696414e-06, "loss": 0.0205, "step": 2317 }, { "epoch": 1.71, "learning_rate": 1.1149576241241788e-06, "loss": 0.0245, "step": 2318 }, { "epoch": 1.71, "learning_rate": 1.1094951281411358e-06, "loss": 0.023, "step": 2319 }, { "epoch": 1.71, "learning_rate": 1.104045260179415e-06, "loss": 0.0267, "step": 2320 }, { "epoch": 1.71, "learning_rate": 1.0986080279799694e-06, "loss": 0.0337, "step": 2321 }, { "epoch": 1.71, "learning_rate": 1.0931834392658213e-06, "loss": 0.0255, "step": 2322 }, { "epoch": 1.71, "learning_rate": 1.0877715017420187e-06, "loss": 0.0318, "step": 2323 }, { "epoch": 1.71, "learning_rate": 1.082372223095647e-06, "loss": 0.0211, "step": 2324 }, { "epoch": 1.71, "learning_rate": 1.0769856109958131e-06, "loss": 0.0256, "step": 2325 }, { "epoch": 1.71, "learning_rate": 1.0716116730936254e-06, "loss": 0.0331, "step": 2326 }, { "epoch": 1.71, "learning_rate": 1.0662504170221955e-06, "loss": 0.0294, "step": 2327 }, { "epoch": 1.71, "learning_rate": 1.0609018503966207e-06, "loss": 0.0307, "step": 2328 }, { "epoch": 1.71, "learning_rate": 1.0555659808139762e-06, "loss": 0.0303, "step": 2329 }, { "epoch": 1.71, "learning_rate": 1.0502428158532952e-06, "loss": 0.0243, "step": 2330 }, { "epoch": 1.72, "learning_rate": 1.0449323630755747e-06, "loss": 0.0244, "step": 2331 }, { "epoch": 1.72, "learning_rate": 1.039634630023747e-06, "loss": 0.0298, "step": 2332 }, { "epoch": 1.72, "learning_rate": 1.0343496242226848e-06, "loss": 0.0163, "step": 2333 }, { "epoch": 1.72, "learning_rate": 1.0290773531791743e-06, "loss": 0.0192, "step": 2334 }, { "epoch": 1.72, "learning_rate": 1.0238178243819264e-06, "loss": 0.0286, "step": 2335 }, { "epoch": 1.72, "learning_rate": 1.0185710453015374e-06, "loss": 0.0295, "step": 2336 }, { "epoch": 1.72, "learning_rate": 1.0133370233905083e-06, "loss": 0.0232, "step": 2337 }, { "epoch": 1.72, "learning_rate": 1.0081157660832086e-06, "loss": 0.038, "step": 2338 }, { "epoch": 1.72, "learning_rate": 1.002907280795884e-06, "loss": 0.0192, "step": 2339 }, { "epoch": 1.72, "learning_rate": 9.977115749266331e-07, "loss": 0.0281, "step": 2340 }, { "epoch": 1.72, "learning_rate": 9.925286558554125e-07, "loss": 0.0284, "step": 2341 }, { "epoch": 1.72, "learning_rate": 9.87358530944006e-07, "loss": 0.04, "step": 2342 }, { "epoch": 1.72, "learning_rate": 9.82201207536031e-07, "loss": 0.0279, "step": 2343 }, { "epoch": 1.72, "learning_rate": 9.77056692956916e-07, "loss": 0.0192, "step": 2344 }, { "epoch": 1.73, "learning_rate": 9.719249945139053e-07, "loss": 0.0226, "step": 2345 }, { "epoch": 1.73, "learning_rate": 9.668061194960255e-07, "loss": 0.0176, "step": 2346 }, { "epoch": 1.73, "learning_rate": 9.617000751741047e-07, "loss": 0.0287, "step": 2347 }, { "epoch": 1.73, "learning_rate": 9.566068688007346e-07, "loss": 0.0231, "step": 2348 }, { "epoch": 1.73, "learning_rate": 9.515265076102775e-07, "loss": 0.0286, "step": 2349 }, { "epoch": 1.73, "learning_rate": 9.464589988188466e-07, "loss": 0.0293, "step": 2350 }, { "epoch": 1.73, "learning_rate": 9.414043496243053e-07, "loss": 0.0177, "step": 2351 }, { "epoch": 1.73, "learning_rate": 9.363625672062427e-07, "loss": 0.0204, "step": 2352 }, { "epoch": 1.73, "learning_rate": 9.313336587259813e-07, "loss": 0.018, "step": 2353 }, { "epoch": 1.73, "learning_rate": 9.263176313265521e-07, "loss": 0.0239, "step": 2354 }, { "epoch": 1.73, "learning_rate": 9.213144921326922e-07, "loss": 0.0252, "step": 2355 }, { "epoch": 1.73, "learning_rate": 9.163242482508306e-07, "loss": 0.0192, "step": 2356 }, { "epoch": 1.73, "learning_rate": 9.113469067690806e-07, "loss": 0.0189, "step": 2357 }, { "epoch": 1.74, "learning_rate": 9.06382474757228e-07, "loss": 0.0271, "step": 2358 }, { "epoch": 1.74, "learning_rate": 9.014309592667236e-07, "loss": 0.0266, "step": 2359 }, { "epoch": 1.74, "learning_rate": 8.964923673306725e-07, "loss": 0.0244, "step": 2360 }, { "epoch": 1.74, "learning_rate": 8.915667059638233e-07, "loss": 0.0268, "step": 2361 }, { "epoch": 1.74, "learning_rate": 8.866539821625519e-07, "loss": 0.0311, "step": 2362 }, { "epoch": 1.74, "learning_rate": 8.817542029048664e-07, "loss": 0.034, "step": 2363 }, { "epoch": 1.74, "learning_rate": 8.76867375150382e-07, "loss": 0.0244, "step": 2364 }, { "epoch": 1.74, "learning_rate": 8.719935058403229e-07, "loss": 0.0283, "step": 2365 }, { "epoch": 1.74, "learning_rate": 8.671326018975024e-07, "loss": 0.0202, "step": 2366 }, { "epoch": 1.74, "learning_rate": 8.622846702263232e-07, "loss": 0.0225, "step": 2367 }, { "epoch": 1.74, "learning_rate": 8.574497177127561e-07, "loss": 0.0212, "step": 2368 }, { "epoch": 1.74, "learning_rate": 8.52627751224343e-07, "loss": 0.0219, "step": 2369 }, { "epoch": 1.74, "learning_rate": 8.47818777610172e-07, "loss": 0.0335, "step": 2370 }, { "epoch": 1.74, "learning_rate": 8.430228037008836e-07, "loss": 0.026, "step": 2371 }, { "epoch": 1.75, "learning_rate": 8.38239836308653e-07, "loss": 0.0167, "step": 2372 }, { "epoch": 1.75, "learning_rate": 8.334698822271781e-07, "loss": 0.0331, "step": 2373 }, { "epoch": 1.75, "learning_rate": 8.287129482316725e-07, "loss": 0.0345, "step": 2374 }, { "epoch": 1.75, "learning_rate": 8.239690410788604e-07, "loss": 0.0333, "step": 2375 }, { "epoch": 1.75, "learning_rate": 8.192381675069561e-07, "loss": 0.0272, "step": 2376 }, { "epoch": 1.75, "learning_rate": 8.145203342356667e-07, "loss": 0.0217, "step": 2377 }, { "epoch": 1.75, "learning_rate": 8.098155479661751e-07, "loss": 0.0292, "step": 2378 }, { "epoch": 1.75, "learning_rate": 8.051238153811336e-07, "loss": 0.0194, "step": 2379 }, { "epoch": 1.75, "learning_rate": 8.004451431446503e-07, "loss": 0.0341, "step": 2380 }, { "epoch": 1.75, "learning_rate": 7.957795379022848e-07, "loss": 0.0293, "step": 2381 }, { "epoch": 1.75, "learning_rate": 7.911270062810338e-07, "loss": 0.0293, "step": 2382 }, { "epoch": 1.75, "learning_rate": 7.86487554889328e-07, "loss": 0.0283, "step": 2383 }, { "epoch": 1.75, "learning_rate": 7.818611903170159e-07, "loss": 0.017, "step": 2384 }, { "epoch": 1.75, "learning_rate": 7.77247919135361e-07, "loss": 0.0332, "step": 2385 }, { "epoch": 1.76, "learning_rate": 7.72647747897024e-07, "loss": 0.0259, "step": 2386 }, { "epoch": 1.76, "learning_rate": 7.680606831360638e-07, "loss": 0.0249, "step": 2387 }, { "epoch": 1.76, "learning_rate": 7.634867313679172e-07, "loss": 0.0316, "step": 2388 }, { "epoch": 1.76, "learning_rate": 7.589258990894022e-07, "loss": 0.0282, "step": 2389 }, { "epoch": 1.76, "learning_rate": 7.543781927786953e-07, "loss": 0.03, "step": 2390 }, { "epoch": 1.76, "learning_rate": 7.498436188953362e-07, "loss": 0.0367, "step": 2391 }, { "epoch": 1.76, "learning_rate": 7.453221838802027e-07, "loss": 0.0183, "step": 2392 }, { "epoch": 1.76, "learning_rate": 7.408138941555176e-07, "loss": 0.0245, "step": 2393 }, { "epoch": 1.76, "learning_rate": 7.363187561248275e-07, "loss": 0.0335, "step": 2394 }, { "epoch": 1.76, "learning_rate": 7.318367761730006e-07, "loss": 0.0293, "step": 2395 }, { "epoch": 1.76, "learning_rate": 7.273679606662166e-07, "loss": 0.0224, "step": 2396 }, { "epoch": 1.76, "learning_rate": 7.229123159519547e-07, "loss": 0.027, "step": 2397 }, { "epoch": 1.76, "learning_rate": 7.184698483589858e-07, "loss": 0.0205, "step": 2398 }, { "epoch": 1.77, "learning_rate": 7.140405641973658e-07, "loss": 0.0214, "step": 2399 }, { "epoch": 1.77, "learning_rate": 7.096244697584221e-07, "loss": 0.0314, "step": 2400 }, { "epoch": 1.77, "learning_rate": 7.052215713147515e-07, "loss": 0.0282, "step": 2401 }, { "epoch": 1.77, "learning_rate": 7.008318751202048e-07, "loss": 0.0418, "step": 2402 }, { "epoch": 1.77, "learning_rate": 6.964553874098823e-07, "loss": 0.0251, "step": 2403 }, { "epoch": 1.77, "learning_rate": 6.92092114400118e-07, "loss": 0.0298, "step": 2404 }, { "epoch": 1.77, "learning_rate": 6.877420622884845e-07, "loss": 0.0269, "step": 2405 }, { "epoch": 1.77, "learning_rate": 6.834052372537658e-07, "loss": 0.0229, "step": 2406 }, { "epoch": 1.77, "learning_rate": 6.790816454559645e-07, "loss": 0.0215, "step": 2407 }, { "epoch": 1.77, "learning_rate": 6.747712930362848e-07, "loss": 0.0294, "step": 2408 }, { "epoch": 1.77, "learning_rate": 6.704741861171293e-07, "loss": 0.0241, "step": 2409 }, { "epoch": 1.77, "learning_rate": 6.661903308020801e-07, "loss": 0.0159, "step": 2410 }, { "epoch": 1.77, "learning_rate": 6.619197331759019e-07, "loss": 0.0225, "step": 2411 }, { "epoch": 1.77, "learning_rate": 6.57662399304525e-07, "loss": 0.0257, "step": 2412 }, { "epoch": 1.78, "learning_rate": 6.534183352350432e-07, "loss": 0.0187, "step": 2413 }, { "epoch": 1.78, "learning_rate": 6.491875469956998e-07, "loss": 0.0228, "step": 2414 }, { "epoch": 1.78, "learning_rate": 6.449700405958837e-07, "loss": 0.0281, "step": 2415 }, { "epoch": 1.78, "learning_rate": 6.407658220261126e-07, "loss": 0.0276, "step": 2416 }, { "epoch": 1.78, "learning_rate": 6.365748972580376e-07, "loss": 0.02, "step": 2417 }, { "epoch": 1.78, "learning_rate": 6.323972722444215e-07, "loss": 0.0267, "step": 2418 }, { "epoch": 1.78, "learning_rate": 6.282329529191389e-07, "loss": 0.0125, "step": 2419 }, { "epoch": 1.78, "learning_rate": 6.240819451971658e-07, "loss": 0.0277, "step": 2420 }, { "epoch": 1.78, "learning_rate": 6.199442549745693e-07, "loss": 0.0142, "step": 2421 }, { "epoch": 1.78, "learning_rate": 6.158198881284994e-07, "loss": 0.0168, "step": 2422 }, { "epoch": 1.78, "learning_rate": 6.117088505171842e-07, "loss": 0.0254, "step": 2423 }, { "epoch": 1.78, "learning_rate": 6.076111479799162e-07, "loss": 0.0235, "step": 2424 }, { "epoch": 1.78, "learning_rate": 6.035267863370475e-07, "loss": 0.024, "step": 2425 }, { "epoch": 1.79, "learning_rate": 5.994557713899829e-07, "loss": 0.0361, "step": 2426 }, { "epoch": 1.79, "learning_rate": 5.953981089211702e-07, "loss": 0.0336, "step": 2427 }, { "epoch": 1.79, "learning_rate": 5.913538046940859e-07, "loss": 0.0361, "step": 2428 }, { "epoch": 1.79, "learning_rate": 5.873228644532392e-07, "loss": 0.0308, "step": 2429 }, { "epoch": 1.79, "learning_rate": 5.833052939241513e-07, "loss": 0.0365, "step": 2430 }, { "epoch": 1.79, "learning_rate": 5.793010988133585e-07, "loss": 0.0236, "step": 2431 }, { "epoch": 1.79, "learning_rate": 5.753102848083924e-07, "loss": 0.0302, "step": 2432 }, { "epoch": 1.79, "learning_rate": 5.713328575777877e-07, "loss": 0.0336, "step": 2433 }, { "epoch": 1.79, "learning_rate": 5.673688227710539e-07, "loss": 0.0301, "step": 2434 }, { "epoch": 1.79, "learning_rate": 5.634181860186871e-07, "loss": 0.0264, "step": 2435 }, { "epoch": 1.79, "learning_rate": 5.594809529321443e-07, "loss": 0.0231, "step": 2436 }, { "epoch": 1.79, "learning_rate": 5.55557129103852e-07, "loss": 0.017, "step": 2437 }, { "epoch": 1.79, "learning_rate": 5.516467201071829e-07, "loss": 0.0163, "step": 2438 }, { "epoch": 1.79, "learning_rate": 5.477497314964642e-07, "loss": 0.0377, "step": 2439 }, { "epoch": 1.8, "learning_rate": 5.438661688069513e-07, "loss": 0.0194, "step": 2440 }, { "epoch": 1.8, "learning_rate": 5.399960375548374e-07, "loss": 0.0187, "step": 2441 }, { "epoch": 1.8, "learning_rate": 5.361393432372319e-07, "loss": 0.0228, "step": 2442 }, { "epoch": 1.8, "learning_rate": 5.322960913321629e-07, "loss": 0.0279, "step": 2443 }, { "epoch": 1.8, "learning_rate": 5.284662872985602e-07, "loss": 0.0242, "step": 2444 }, { "epoch": 1.8, "learning_rate": 5.246499365762592e-07, "loss": 0.0225, "step": 2445 }, { "epoch": 1.8, "learning_rate": 5.208470445859782e-07, "loss": 0.0317, "step": 2446 }, { "epoch": 1.8, "learning_rate": 5.170576167293262e-07, "loss": 0.016, "step": 2447 }, { "epoch": 1.8, "learning_rate": 5.132816583887812e-07, "loss": 0.023, "step": 2448 }, { "epoch": 1.8, "learning_rate": 5.095191749276962e-07, "loss": 0.0223, "step": 2449 }, { "epoch": 1.8, "learning_rate": 5.057701716902764e-07, "loss": 0.0283, "step": 2450 }, { "epoch": 1.8, "learning_rate": 5.020346540015875e-07, "loss": 0.026, "step": 2451 }, { "epoch": 1.8, "learning_rate": 4.983126271675354e-07, "loss": 0.0163, "step": 2452 }, { "epoch": 1.81, "learning_rate": 4.946040964748666e-07, "loss": 0.0185, "step": 2453 }, { "epoch": 1.81, "learning_rate": 4.909090671911554e-07, "loss": 0.042, "step": 2454 }, { "epoch": 1.81, "learning_rate": 4.872275445648e-07, "loss": 0.0367, "step": 2455 }, { "epoch": 1.81, "learning_rate": 4.835595338250155e-07, "loss": 0.0341, "step": 2456 }, { "epoch": 1.81, "learning_rate": 4.799050401818184e-07, "loss": 0.0185, "step": 2457 }, { "epoch": 1.81, "learning_rate": 4.762640688260356e-07, "loss": 0.022, "step": 2458 }, { "epoch": 1.81, "learning_rate": 4.7263662492927906e-07, "loss": 0.0175, "step": 2459 }, { "epoch": 1.81, "learning_rate": 4.690227136439496e-07, "loss": 0.0278, "step": 2460 }, { "epoch": 1.81, "learning_rate": 4.6542234010322427e-07, "loss": 0.0177, "step": 2461 }, { "epoch": 1.81, "learning_rate": 4.618355094210547e-07, "loss": 0.0244, "step": 2462 }, { "epoch": 1.81, "learning_rate": 4.5826222669215213e-07, "loss": 0.0336, "step": 2463 }, { "epoch": 1.81, "learning_rate": 4.5470249699198667e-07, "loss": 0.0213, "step": 2464 }, { "epoch": 1.81, "learning_rate": 4.5115632537678035e-07, "loss": 0.0296, "step": 2465 }, { "epoch": 1.81, "learning_rate": 4.476237168834929e-07, "loss": 0.0246, "step": 2466 }, { "epoch": 1.82, "learning_rate": 4.441046765298207e-07, "loss": 0.04, "step": 2467 }, { "epoch": 1.82, "learning_rate": 4.4059920931418866e-07, "loss": 0.025, "step": 2468 }, { "epoch": 1.82, "learning_rate": 4.3710732021574056e-07, "loss": 0.0239, "step": 2469 }, { "epoch": 1.82, "learning_rate": 4.336290141943367e-07, "loss": 0.036, "step": 2470 }, { "epoch": 1.82, "learning_rate": 4.3016429619054276e-07, "loss": 0.0283, "step": 2471 }, { "epoch": 1.82, "learning_rate": 4.267131711256245e-07, "loss": 0.0338, "step": 2472 }, { "epoch": 1.82, "learning_rate": 4.232756439015384e-07, "loss": 0.0219, "step": 2473 }, { "epoch": 1.82, "learning_rate": 4.1985171940092884e-07, "loss": 0.0233, "step": 2474 }, { "epoch": 1.82, "learning_rate": 4.1644140248711773e-07, "loss": 0.0299, "step": 2475 }, { "epoch": 1.82, "learning_rate": 4.130446980041003e-07, "loss": 0.014, "step": 2476 }, { "epoch": 1.82, "learning_rate": 4.0966161077653386e-07, "loss": 0.0252, "step": 2477 }, { "epoch": 1.82, "learning_rate": 4.0629214560973907e-07, "loss": 0.0386, "step": 2478 }, { "epoch": 1.82, "learning_rate": 4.029363072896819e-07, "loss": 0.0296, "step": 2479 }, { "epoch": 1.82, "learning_rate": 3.995941005829773e-07, "loss": 0.0239, "step": 2480 }, { "epoch": 1.83, "learning_rate": 3.962655302368734e-07, "loss": 0.0187, "step": 2481 }, { "epoch": 1.83, "learning_rate": 3.92950600979255e-07, "loss": 0.0251, "step": 2482 }, { "epoch": 1.83, "learning_rate": 3.896493175186278e-07, "loss": 0.0227, "step": 2483 }, { "epoch": 1.83, "learning_rate": 3.863616845441154e-07, "loss": 0.026, "step": 2484 }, { "epoch": 1.83, "learning_rate": 3.8308770672545017e-07, "loss": 0.0278, "step": 2485 }, { "epoch": 1.83, "learning_rate": 3.798273887129755e-07, "loss": 0.0325, "step": 2486 }, { "epoch": 1.83, "learning_rate": 3.7658073513762493e-07, "loss": 0.0248, "step": 2487 }, { "epoch": 1.83, "learning_rate": 3.733477506109262e-07, "loss": 0.0307, "step": 2488 }, { "epoch": 1.83, "learning_rate": 3.701284397249938e-07, "loss": 0.0277, "step": 2489 }, { "epoch": 1.83, "learning_rate": 3.669228070525177e-07, "loss": 0.0304, "step": 2490 }, { "epoch": 1.83, "learning_rate": 3.637308571467568e-07, "loss": 0.0257, "step": 2491 }, { "epoch": 1.83, "learning_rate": 3.6055259454154334e-07, "loss": 0.0247, "step": 2492 }, { "epoch": 1.83, "learning_rate": 3.573880237512584e-07, "loss": 0.0209, "step": 2493 }, { "epoch": 1.84, "learning_rate": 3.5423714927084186e-07, "loss": 0.0336, "step": 2494 }, { "epoch": 1.84, "learning_rate": 3.51099975575776e-07, "loss": 0.0335, "step": 2495 }, { "epoch": 1.84, "learning_rate": 3.4797650712208863e-07, "loss": 0.032, "step": 2496 }, { "epoch": 1.84, "learning_rate": 3.448667483463297e-07, "loss": 0.0157, "step": 2497 }, { "epoch": 1.84, "learning_rate": 3.417707036655882e-07, "loss": 0.0224, "step": 2498 }, { "epoch": 1.84, "learning_rate": 3.3868837747746296e-07, "loss": 0.0231, "step": 2499 }, { "epoch": 1.84, "learning_rate": 3.356197741600753e-07, "loss": 0.0166, "step": 2500 }, { "epoch": 1.84, "learning_rate": 3.3256489807205083e-07, "loss": 0.0209, "step": 2501 }, { "epoch": 1.84, "learning_rate": 3.2952375355251865e-07, "loss": 0.017, "step": 2502 }, { "epoch": 1.84, "learning_rate": 3.2649634492110115e-07, "loss": 0.0255, "step": 2503 }, { "epoch": 1.84, "learning_rate": 3.234826764779131e-07, "loss": 0.0308, "step": 2504 }, { "epoch": 1.84, "learning_rate": 3.2048275250354923e-07, "loss": 0.0341, "step": 2505 }, { "epoch": 1.84, "learning_rate": 3.174965772590866e-07, "loss": 0.0209, "step": 2506 }, { "epoch": 1.84, "learning_rate": 3.1452415498607004e-07, "loss": 0.0287, "step": 2507 }, { "epoch": 1.85, "learning_rate": 3.1156548990651237e-07, "loss": 0.0246, "step": 2508 }, { "epoch": 1.85, "learning_rate": 3.086205862228819e-07, "loss": 0.028, "step": 2509 }, { "epoch": 1.85, "learning_rate": 3.0568944811810497e-07, "loss": 0.0215, "step": 2510 }, { "epoch": 1.85, "learning_rate": 3.0277207975555114e-07, "loss": 0.0237, "step": 2511 }, { "epoch": 1.85, "learning_rate": 2.9986848527903347e-07, "loss": 0.0359, "step": 2512 }, { "epoch": 1.85, "learning_rate": 2.969786688128007e-07, "loss": 0.0322, "step": 2513 }, { "epoch": 1.85, "learning_rate": 2.9410263446153385e-07, "loss": 0.0165, "step": 2514 }, { "epoch": 1.85, "learning_rate": 2.9124038631033305e-07, "loss": 0.0258, "step": 2515 }, { "epoch": 1.85, "learning_rate": 2.8839192842471943e-07, "loss": 0.0369, "step": 2516 }, { "epoch": 1.85, "learning_rate": 2.855572648506244e-07, "loss": 0.0185, "step": 2517 }, { "epoch": 1.85, "learning_rate": 2.827363996143895e-07, "loss": 0.0221, "step": 2518 }, { "epoch": 1.85, "learning_rate": 2.7992933672275513e-07, "loss": 0.0378, "step": 2519 }, { "epoch": 1.85, "learning_rate": 2.771360801628575e-07, "loss": 0.0279, "step": 2520 }, { "epoch": 1.86, "learning_rate": 2.743566339022197e-07, "loss": 0.0227, "step": 2521 }, { "epoch": 1.86, "learning_rate": 2.7159100188875355e-07, "loss": 0.0429, "step": 2522 }, { "epoch": 1.86, "learning_rate": 2.688391880507446e-07, "loss": 0.0313, "step": 2523 }, { "epoch": 1.86, "learning_rate": 2.6610119629685517e-07, "loss": 0.0223, "step": 2524 }, { "epoch": 1.86, "learning_rate": 2.63377030516111e-07, "loss": 0.0221, "step": 2525 }, { "epoch": 1.86, "learning_rate": 2.606666945779024e-07, "loss": 0.0249, "step": 2526 }, { "epoch": 1.86, "learning_rate": 2.579701923319744e-07, "loss": 0.025, "step": 2527 }, { "epoch": 1.86, "learning_rate": 2.552875276084232e-07, "loss": 0.0209, "step": 2528 }, { "epoch": 1.86, "learning_rate": 2.5261870421768975e-07, "loss": 0.0228, "step": 2529 }, { "epoch": 1.86, "learning_rate": 2.4996372595055605e-07, "loss": 0.0208, "step": 2530 }, { "epoch": 1.86, "learning_rate": 2.4732259657813674e-07, "loss": 0.0234, "step": 2531 }, { "epoch": 1.86, "learning_rate": 2.446953198518776e-07, "loss": 0.0188, "step": 2532 }, { "epoch": 1.86, "learning_rate": 2.4208189950354564e-07, "loss": 0.0168, "step": 2533 }, { "epoch": 1.86, "learning_rate": 2.394823392452306e-07, "loss": 0.0357, "step": 2534 }, { "epoch": 1.87, "learning_rate": 2.3689664276932978e-07, "loss": 0.0343, "step": 2535 }, { "epoch": 1.87, "learning_rate": 2.343248137485532e-07, "loss": 0.0411, "step": 2536 }, { "epoch": 1.87, "learning_rate": 2.3176685583590986e-07, "loss": 0.0228, "step": 2537 }, { "epoch": 1.87, "learning_rate": 2.2922277266471226e-07, "loss": 0.0411, "step": 2538 }, { "epoch": 1.87, "learning_rate": 2.2669256784855654e-07, "loss": 0.0244, "step": 2539 }, { "epoch": 1.87, "learning_rate": 2.241762449813345e-07, "loss": 0.0286, "step": 2540 }, { "epoch": 1.87, "learning_rate": 2.216738076372149e-07, "loss": 0.0361, "step": 2541 }, { "epoch": 1.87, "learning_rate": 2.191852593706456e-07, "loss": 0.0312, "step": 2542 }, { "epoch": 1.87, "learning_rate": 2.1671060371634355e-07, "loss": 0.0259, "step": 2543 }, { "epoch": 1.87, "learning_rate": 2.142498441893004e-07, "loss": 0.0212, "step": 2544 }, { "epoch": 1.87, "learning_rate": 2.118029842847602e-07, "loss": 0.026, "step": 2545 }, { "epoch": 1.87, "learning_rate": 2.0937002747823067e-07, "loss": 0.0219, "step": 2546 }, { "epoch": 1.87, "learning_rate": 2.0695097722546742e-07, "loss": 0.035, "step": 2547 }, { "epoch": 1.87, "learning_rate": 2.0454583696247864e-07, "loss": 0.0342, "step": 2548 }, { "epoch": 1.88, "learning_rate": 2.0215461010550718e-07, "loss": 0.0233, "step": 2549 }, { "epoch": 1.88, "learning_rate": 1.997773000510428e-07, "loss": 0.0257, "step": 2550 }, { "epoch": 1.88, "learning_rate": 1.9741391017579882e-07, "loss": 0.0339, "step": 2551 }, { "epoch": 1.88, "learning_rate": 1.9506444383672328e-07, "loss": 0.0257, "step": 2552 }, { "epoch": 1.88, "learning_rate": 1.9272890437098124e-07, "loss": 0.0212, "step": 2553 }, { "epoch": 1.88, "learning_rate": 1.9040729509596235e-07, "loss": 0.026, "step": 2554 }, { "epoch": 1.88, "learning_rate": 1.8809961930926435e-07, "loss": 0.023, "step": 2555 }, { "epoch": 1.88, "learning_rate": 1.8580588028869972e-07, "loss": 0.0231, "step": 2556 }, { "epoch": 1.88, "learning_rate": 1.83526081292279e-07, "loss": 0.0233, "step": 2557 }, { "epoch": 1.88, "learning_rate": 1.8126022555821742e-07, "loss": 0.0276, "step": 2558 }, { "epoch": 1.88, "learning_rate": 1.7900831630492166e-07, "loss": 0.0293, "step": 2559 }, { "epoch": 1.88, "learning_rate": 1.7677035673099196e-07, "loss": 0.0264, "step": 2560 }, { "epoch": 1.88, "learning_rate": 1.7454635001521224e-07, "loss": 0.0231, "step": 2561 }, { "epoch": 1.89, "learning_rate": 1.7233629931654782e-07, "loss": 0.0292, "step": 2562 }, { "epoch": 1.89, "learning_rate": 1.701402077741432e-07, "loss": 0.0377, "step": 2563 }, { "epoch": 1.89, "learning_rate": 1.6795807850731428e-07, "loss": 0.0283, "step": 2564 }, { "epoch": 1.89, "learning_rate": 1.6578991461554282e-07, "loss": 0.0173, "step": 2565 }, { "epoch": 1.89, "learning_rate": 1.6363571917847875e-07, "loss": 0.0343, "step": 2566 }, { "epoch": 1.89, "learning_rate": 1.6149549525592666e-07, "loss": 0.016, "step": 2567 }, { "epoch": 1.89, "learning_rate": 1.593692458878482e-07, "loss": 0.0322, "step": 2568 }, { "epoch": 1.89, "learning_rate": 1.5725697409435746e-07, "loss": 0.0236, "step": 2569 }, { "epoch": 1.89, "learning_rate": 1.5515868287571124e-07, "loss": 0.0225, "step": 2570 }, { "epoch": 1.89, "learning_rate": 1.5307437521231095e-07, "loss": 0.0238, "step": 2571 }, { "epoch": 1.89, "learning_rate": 1.5100405406469508e-07, "loss": 0.0264, "step": 2572 }, { "epoch": 1.89, "learning_rate": 1.4894772237353472e-07, "loss": 0.0215, "step": 2573 }, { "epoch": 1.89, "learning_rate": 1.469053830596323e-07, "loss": 0.0212, "step": 2574 }, { "epoch": 1.89, "learning_rate": 1.4487703902391514e-07, "loss": 0.0251, "step": 2575 }, { "epoch": 1.9, "learning_rate": 1.4286269314743085e-07, "loss": 0.0311, "step": 2576 }, { "epoch": 1.9, "learning_rate": 1.4086234829134405e-07, "loss": 0.0329, "step": 2577 }, { "epoch": 1.9, "learning_rate": 1.3887600729693307e-07, "loss": 0.0182, "step": 2578 }, { "epoch": 1.9, "learning_rate": 1.3690367298558548e-07, "loss": 0.0311, "step": 2579 }, { "epoch": 1.9, "learning_rate": 1.3494534815879257e-07, "loss": 0.0298, "step": 2580 }, { "epoch": 1.9, "learning_rate": 1.3300103559814927e-07, "loss": 0.0253, "step": 2581 }, { "epoch": 1.9, "learning_rate": 1.310707380653442e-07, "loss": 0.0221, "step": 2582 }, { "epoch": 1.9, "learning_rate": 1.29154458302162e-07, "loss": 0.0234, "step": 2583 }, { "epoch": 1.9, "learning_rate": 1.2725219903047425e-07, "loss": 0.0267, "step": 2584 }, { "epoch": 1.9, "learning_rate": 1.2536396295223852e-07, "loss": 0.0158, "step": 2585 }, { "epoch": 1.9, "learning_rate": 1.2348975274949605e-07, "loss": 0.0282, "step": 2586 }, { "epoch": 1.9, "learning_rate": 1.216295710843618e-07, "loss": 0.0198, "step": 2587 }, { "epoch": 1.9, "learning_rate": 1.1978342059902892e-07, "loss": 0.0217, "step": 2588 }, { "epoch": 1.91, "learning_rate": 1.1795130391575871e-07, "loss": 0.0357, "step": 2589 }, { "epoch": 1.91, "learning_rate": 1.161332236368784e-07, "loss": 0.0284, "step": 2590 }, { "epoch": 1.91, "learning_rate": 1.1432918234477897e-07, "loss": 0.0222, "step": 2591 }, { "epoch": 1.91, "learning_rate": 1.1253918260190844e-07, "loss": 0.0283, "step": 2592 }, { "epoch": 1.91, "learning_rate": 1.1076322695077412e-07, "loss": 0.0202, "step": 2593 }, { "epoch": 1.91, "learning_rate": 1.0900131791393265e-07, "loss": 0.0328, "step": 2594 }, { "epoch": 1.91, "learning_rate": 1.072534579939899e-07, "loss": 0.0215, "step": 2595 }, { "epoch": 1.91, "learning_rate": 1.0551964967359441e-07, "loss": 0.0216, "step": 2596 }, { "epoch": 1.91, "learning_rate": 1.0379989541543845e-07, "loss": 0.0293, "step": 2597 }, { "epoch": 1.91, "learning_rate": 1.0209419766225026e-07, "loss": 0.0261, "step": 2598 }, { "epoch": 1.91, "learning_rate": 1.0040255883679295e-07, "loss": 0.0317, "step": 2599 }, { "epoch": 1.91, "learning_rate": 9.872498134186115e-08, "loss": 0.0226, "step": 2600 }, { "epoch": 1.91, "learning_rate": 9.706146756027657e-08, "loss": 0.0212, "step": 2601 }, { "epoch": 1.91, "learning_rate": 9.541201985488358e-08, "loss": 0.0292, "step": 2602 }, { "epoch": 1.92, "learning_rate": 9.377664056854807e-08, "loss": 0.0177, "step": 2603 }, { "epoch": 1.92, "learning_rate": 9.215533202415306e-08, "loss": 0.0254, "step": 2604 }, { "epoch": 1.92, "learning_rate": 9.054809652459751e-08, "loss": 0.0245, "step": 2605 }, { "epoch": 1.92, "learning_rate": 8.89549363527864e-08, "loss": 0.0312, "step": 2606 }, { "epoch": 1.92, "learning_rate": 8.737585377163849e-08, "loss": 0.0184, "step": 2607 }, { "epoch": 1.92, "learning_rate": 8.581085102407072e-08, "loss": 0.0263, "step": 2608 }, { "epoch": 1.92, "learning_rate": 8.425993033300494e-08, "loss": 0.0222, "step": 2609 }, { "epoch": 1.92, "learning_rate": 8.272309390136013e-08, "loss": 0.034, "step": 2610 }, { "epoch": 1.92, "learning_rate": 8.120034391205011e-08, "loss": 0.0203, "step": 2611 }, { "epoch": 1.92, "learning_rate": 7.96916825279781e-08, "loss": 0.0173, "step": 2612 }, { "epoch": 1.92, "learning_rate": 7.819711189204105e-08, "loss": 0.0263, "step": 2613 }, { "epoch": 1.92, "learning_rate": 7.671663412711527e-08, "loss": 0.0241, "step": 2614 }, { "epoch": 1.92, "learning_rate": 7.525025133606534e-08, "loss": 0.0313, "step": 2615 }, { "epoch": 1.92, "learning_rate": 7.37979656017318e-08, "loss": 0.0258, "step": 2616 }, { "epoch": 1.93, "learning_rate": 7.235977898693125e-08, "loss": 0.0366, "step": 2617 }, { "epoch": 1.93, "learning_rate": 7.09356935344585e-08, "loss": 0.0187, "step": 2618 }, { "epoch": 1.93, "learning_rate": 6.952571126707442e-08, "loss": 0.0264, "step": 2619 }, { "epoch": 1.93, "learning_rate": 6.812983418750917e-08, "loss": 0.0268, "step": 2620 }, { "epoch": 1.93, "learning_rate": 6.674806427846014e-08, "loss": 0.0205, "step": 2621 }, { "epoch": 1.93, "learning_rate": 6.538040350258401e-08, "loss": 0.0292, "step": 2622 }, { "epoch": 1.93, "learning_rate": 6.402685380249796e-08, "loss": 0.0251, "step": 2623 }, { "epoch": 1.93, "learning_rate": 6.268741710077741e-08, "loss": 0.0228, "step": 2624 }, { "epoch": 1.93, "learning_rate": 6.13620952999494e-08, "loss": 0.0263, "step": 2625 }, { "epoch": 1.93, "learning_rate": 6.005089028249366e-08, "loss": 0.0204, "step": 2626 }, { "epoch": 1.93, "learning_rate": 5.8753803910838184e-08, "loss": 0.0219, "step": 2627 }, { "epoch": 1.93, "learning_rate": 5.747083802735587e-08, "loss": 0.0282, "step": 2628 }, { "epoch": 1.93, "learning_rate": 5.620199445436569e-08, "loss": 0.0306, "step": 2629 }, { "epoch": 1.94, "learning_rate": 5.494727499412489e-08, "loss": 0.0161, "step": 2630 }, { "epoch": 1.94, "learning_rate": 5.3706681428830063e-08, "loss": 0.0308, "step": 2631 }, { "epoch": 1.94, "learning_rate": 5.2480215520611665e-08, "loss": 0.0387, "step": 2632 }, { "epoch": 1.94, "learning_rate": 5.1267879011537294e-08, "loss": 0.0376, "step": 2633 }, { "epoch": 1.94, "learning_rate": 5.006967362359949e-08, "loss": 0.0153, "step": 2634 }, { "epoch": 1.94, "learning_rate": 4.888560105872464e-08, "loss": 0.0266, "step": 2635 }, { "epoch": 1.94, "learning_rate": 4.7715662998760735e-08, "loss": 0.0175, "step": 2636 }, { "epoch": 1.94, "learning_rate": 4.655986110548294e-08, "loss": 0.0266, "step": 2637 }, { "epoch": 1.94, "learning_rate": 4.54181970205847e-08, "loss": 0.0234, "step": 2638 }, { "epoch": 1.94, "learning_rate": 4.429067236567886e-08, "loss": 0.0239, "step": 2639 }, { "epoch": 1.94, "learning_rate": 4.317728874229321e-08, "loss": 0.0238, "step": 2640 }, { "epoch": 1.94, "learning_rate": 4.207804773187496e-08, "loss": 0.0281, "step": 2641 }, { "epoch": 1.94, "learning_rate": 4.0992950895776265e-08, "loss": 0.0153, "step": 2642 }, { "epoch": 1.94, "learning_rate": 3.992199977526534e-08, "loss": 0.0349, "step": 2643 }, { "epoch": 1.95, "learning_rate": 3.8865195891512054e-08, "loss": 0.0195, "step": 2644 }, { "epoch": 1.95, "learning_rate": 3.782254074559677e-08, "loss": 0.0329, "step": 2645 }, { "epoch": 1.95, "learning_rate": 3.679403581849927e-08, "loss": 0.0252, "step": 2646 }, { "epoch": 1.95, "learning_rate": 3.577968257110209e-08, "loss": 0.019, "step": 2647 }, { "epoch": 1.95, "learning_rate": 3.477948244418716e-08, "loss": 0.0353, "step": 2648 }, { "epoch": 1.95, "learning_rate": 3.379343685843362e-08, "loss": 0.025, "step": 2649 }, { "epoch": 1.95, "learning_rate": 3.2821547214413327e-08, "loss": 0.0284, "step": 2650 }, { "epoch": 1.95, "learning_rate": 3.186381489259427e-08, "loss": 0.0245, "step": 2651 }, { "epoch": 1.95, "learning_rate": 3.0920241253331596e-08, "loss": 0.0292, "step": 2652 }, { "epoch": 1.95, "learning_rate": 2.9990827636874334e-08, "loss": 0.0234, "step": 2653 }, { "epoch": 1.95, "learning_rate": 2.9075575363355368e-08, "loss": 0.0207, "step": 2654 }, { "epoch": 1.95, "learning_rate": 2.817448573279258e-08, "loss": 0.0244, "step": 2655 }, { "epoch": 1.95, "learning_rate": 2.728756002508881e-08, "loss": 0.0248, "step": 2656 }, { "epoch": 1.96, "learning_rate": 2.6414799500029675e-08, "loss": 0.0268, "step": 2657 }, { "epoch": 1.96, "learning_rate": 2.555620539727799e-08, "loss": 0.0225, "step": 2658 }, { "epoch": 1.96, "learning_rate": 2.4711778936375996e-08, "loss": 0.0315, "step": 2659 }, { "epoch": 1.96, "learning_rate": 2.388152131674093e-08, "loss": 0.0206, "step": 2660 }, { "epoch": 1.96, "learning_rate": 2.306543371766723e-08, "loss": 0.0227, "step": 2661 }, { "epoch": 1.96, "learning_rate": 2.2263517298320992e-08, "loss": 0.0238, "step": 2662 }, { "epoch": 1.96, "learning_rate": 2.1475773197739968e-08, "loss": 0.0347, "step": 2663 }, { "epoch": 1.96, "learning_rate": 2.070220253483024e-08, "loss": 0.0229, "step": 2664 }, { "epoch": 1.96, "learning_rate": 1.9942806408367322e-08, "loss": 0.0311, "step": 2665 }, { "epoch": 1.96, "learning_rate": 1.919758589699283e-08, "loss": 0.0229, "step": 2666 }, { "epoch": 1.96, "learning_rate": 1.8466542059214497e-08, "loss": 0.0246, "step": 2667 }, { "epoch": 1.96, "learning_rate": 1.774967593340171e-08, "loss": 0.0246, "step": 2668 }, { "epoch": 1.96, "learning_rate": 1.7046988537787746e-08, "loss": 0.0274, "step": 2669 }, { "epoch": 1.96, "learning_rate": 1.635848087046532e-08, "loss": 0.017, "step": 2670 }, { "epoch": 1.97, "learning_rate": 1.568415390938771e-08, "loss": 0.0174, "step": 2671 }, { "epoch": 1.97, "learning_rate": 1.5024008612363196e-08, "loss": 0.0389, "step": 2672 }, { "epoch": 1.97, "learning_rate": 1.4378045917058381e-08, "loss": 0.0347, "step": 2673 }, { "epoch": 1.97, "learning_rate": 1.3746266740997104e-08, "loss": 0.0231, "step": 2674 }, { "epoch": 1.97, "learning_rate": 1.3128671981552654e-08, "loss": 0.0259, "step": 2675 }, { "epoch": 1.97, "learning_rate": 1.2525262515954429e-08, "loss": 0.0203, "step": 2676 }, { "epoch": 1.97, "learning_rate": 1.1936039201281279e-08, "loss": 0.0166, "step": 2677 }, { "epoch": 1.97, "learning_rate": 1.1361002874461512e-08, "loss": 0.0372, "step": 2678 }, { "epoch": 1.97, "learning_rate": 1.0800154352275105e-08, "loss": 0.0287, "step": 2679 }, { "epoch": 1.97, "learning_rate": 1.0253494431347045e-08, "loss": 0.0291, "step": 2680 }, { "epoch": 1.97, "learning_rate": 9.72102388814955e-09, "loss": 0.0152, "step": 2681 }, { "epoch": 1.97, "learning_rate": 9.202743479002074e-09, "loss": 0.0219, "step": 2682 }, { "epoch": 1.97, "learning_rate": 8.698653940065748e-09, "loss": 0.021, "step": 2683 }, { "epoch": 1.97, "learning_rate": 8.208755987346717e-09, "loss": 0.0232, "step": 2684 }, { "epoch": 1.98, "learning_rate": 7.733050316692803e-09, "loss": 0.0284, "step": 2685 }, { "epoch": 1.98, "learning_rate": 7.27153760379462e-09, "loss": 0.0286, "step": 2686 }, { "epoch": 1.98, "learning_rate": 6.824218504182245e-09, "loss": 0.0238, "step": 2687 }, { "epoch": 1.98, "learning_rate": 6.391093653224101e-09, "loss": 0.0378, "step": 2688 }, { "epoch": 1.98, "learning_rate": 5.972163666128073e-09, "loss": 0.019, "step": 2689 }, { "epoch": 1.98, "learning_rate": 5.567429137940395e-09, "loss": 0.0247, "step": 2690 }, { "epoch": 1.98, "learning_rate": 5.176890643544541e-09, "loss": 0.0267, "step": 2691 }, { "epoch": 1.98, "learning_rate": 4.800548737656785e-09, "loss": 0.0243, "step": 2692 }, { "epoch": 1.98, "learning_rate": 4.438403954833969e-09, "loss": 0.0206, "step": 2693 }, { "epoch": 1.98, "learning_rate": 4.090456809462407e-09, "loss": 0.0208, "step": 2694 }, { "epoch": 1.98, "learning_rate": 3.756707795763425e-09, "loss": 0.0207, "step": 2695 }, { "epoch": 1.98, "learning_rate": 3.4371573877944874e-09, "loss": 0.0265, "step": 2696 }, { "epoch": 1.98, "learning_rate": 3.13180603944252e-09, "loss": 0.0247, "step": 2697 }, { "epoch": 1.99, "learning_rate": 2.840654184425029e-09, "loss": 0.0193, "step": 2698 }, { "epoch": 1.99, "learning_rate": 2.563702236294541e-09, "loss": 0.0278, "step": 2699 }, { "epoch": 1.99, "learning_rate": 2.300950588430828e-09, "loss": 0.0221, "step": 2700 }, { "epoch": 1.99, "learning_rate": 2.0523996140453526e-09, "loss": 0.0219, "step": 2701 }, { "epoch": 1.99, "learning_rate": 1.8180496661779346e-09, "loss": 0.0261, "step": 2702 }, { "epoch": 1.99, "learning_rate": 1.5979010776989712e-09, "loss": 0.0405, "step": 2703 }, { "epoch": 1.99, "learning_rate": 1.391954161304998e-09, "loss": 0.0266, "step": 2704 }, { "epoch": 1.99, "learning_rate": 1.2002092095220186e-09, "loss": 0.0231, "step": 2705 }, { "epoch": 1.99, "learning_rate": 1.0226664947032838e-09, "loss": 0.0298, "step": 2706 }, { "epoch": 1.99, "learning_rate": 8.593262690304028e-10, "loss": 0.016, "step": 2707 }, { "epoch": 1.99, "learning_rate": 7.101887645100114e-10, "loss": 0.0181, "step": 2708 }, { "epoch": 1.99, "learning_rate": 5.752541929771038e-10, "loss": 0.0293, "step": 2709 }, { "epoch": 1.99, "learning_rate": 4.54522746090591e-10, "loss": 0.0292, "step": 2710 }, { "epoch": 1.99, "learning_rate": 3.479945953377417e-10, "loss": 0.0254, "step": 2711 }, { "epoch": 2.0, "learning_rate": 2.55669892030852e-10, "loss": 0.0214, "step": 2712 }, { "epoch": 2.0, "learning_rate": 1.775487673061349e-10, "loss": 0.03, "step": 2713 }, { "epoch": 2.0, "learning_rate": 1.1363133212705102e-10, "loss": 0.0335, "step": 2714 }, { "epoch": 2.0, "learning_rate": 6.391767728097797e-11, "loss": 0.0257, "step": 2715 }, { "epoch": 2.0, "learning_rate": 2.840787338254103e-11, "loss": 0.0156, "step": 2716 }, { "epoch": 2.0, "learning_rate": 7.101970866951746e-12, "loss": 0.0271, "step": 2717 }, { "epoch": 2.0, "learning_rate": 0.0, "loss": 0.0299, "step": 2718 }, { "epoch": 2.0, "step": 2718, "total_flos": 149987960586240.0, "train_loss": 0.0592351465035748, "train_runtime": 8874.3343, "train_samples_per_second": 39.201, "train_steps_per_second": 0.306 } ], "logging_steps": 1.0, "max_steps": 2718, "num_input_tokens_seen": 0, "num_train_epochs": 2, "save_steps": 11100, "total_flos": 149987960586240.0, "train_batch_size": 16, "trial_name": null, "trial_params": null }