{ "best_metric": null, "best_model_checkpoint": null, "epoch": 3.9984550019312475, "eval_steps": 500, "global_step": 2588, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 0.0, "loss": 1.4373, "step": 2 }, { "epoch": 0.01, "learning_rate": 0.0, "loss": 1.4385, "step": 4 }, { "epoch": 0.01, "learning_rate": 5.128205128205128e-08, "loss": 1.3737, "step": 6 }, { "epoch": 0.01, "learning_rate": 7.692307692307692e-08, "loss": 1.4679, "step": 8 }, { "epoch": 0.02, "learning_rate": 1.2820512820512818e-07, "loss": 1.4198, "step": 10 }, { "epoch": 0.02, "learning_rate": 1.7948717948717948e-07, "loss": 1.4308, "step": 12 }, { "epoch": 0.02, "learning_rate": 2.3076923076923078e-07, "loss": 1.4539, "step": 14 }, { "epoch": 0.02, "learning_rate": 2.8205128205128203e-07, "loss": 1.4101, "step": 16 }, { "epoch": 0.03, "learning_rate": 3.333333333333333e-07, "loss": 1.5199, "step": 18 }, { "epoch": 0.03, "learning_rate": 3.8461538461538463e-07, "loss": 1.5741, "step": 20 }, { "epoch": 0.03, "learning_rate": 4.358974358974359e-07, "loss": 1.4547, "step": 22 }, { "epoch": 0.04, "learning_rate": 4.871794871794871e-07, "loss": 1.5166, "step": 24 }, { "epoch": 0.04, "learning_rate": 5.384615384615384e-07, "loss": 1.379, "step": 26 }, { "epoch": 0.04, "learning_rate": 5.897435897435898e-07, "loss": 1.3922, "step": 28 }, { "epoch": 0.05, "learning_rate": 6.410256410256411e-07, "loss": 1.358, "step": 30 }, { "epoch": 0.05, "learning_rate": 6.923076923076922e-07, "loss": 1.5129, "step": 32 }, { "epoch": 0.05, "learning_rate": 7.435897435897435e-07, "loss": 1.5167, "step": 34 }, { "epoch": 0.06, "learning_rate": 7.948717948717948e-07, "loss": 1.5071, "step": 36 }, { "epoch": 0.06, "learning_rate": 8.461538461538461e-07, "loss": 1.4828, "step": 38 }, { "epoch": 0.06, "learning_rate": 8.974358974358974e-07, "loss": 1.4743, "step": 40 }, { "epoch": 0.06, "learning_rate": 9.487179487179486e-07, "loss": 1.4679, "step": 42 }, { "epoch": 0.07, "learning_rate": 1e-06, "loss": 1.4403, "step": 44 }, { "epoch": 0.07, "learning_rate": 1.0512820512820512e-06, "loss": 1.439, "step": 46 }, { "epoch": 0.07, "learning_rate": 1.1025641025641026e-06, "loss": 1.4671, "step": 48 }, { "epoch": 0.08, "learning_rate": 1.1538461538461536e-06, "loss": 1.4136, "step": 50 }, { "epoch": 0.08, "learning_rate": 1.205128205128205e-06, "loss": 1.5101, "step": 52 }, { "epoch": 0.08, "learning_rate": 1.2564102564102563e-06, "loss": 1.379, "step": 54 }, { "epoch": 0.09, "learning_rate": 1.3076923076923077e-06, "loss": 1.43, "step": 56 }, { "epoch": 0.09, "learning_rate": 1.358974358974359e-06, "loss": 1.5077, "step": 58 }, { "epoch": 0.09, "learning_rate": 1.4102564102564104e-06, "loss": 1.4259, "step": 60 }, { "epoch": 0.1, "learning_rate": 1.4615384615384614e-06, "loss": 1.5, "step": 62 }, { "epoch": 0.1, "learning_rate": 1.5128205128205128e-06, "loss": 1.4796, "step": 64 }, { "epoch": 0.1, "learning_rate": 1.564102564102564e-06, "loss": 1.4358, "step": 66 }, { "epoch": 0.11, "learning_rate": 1.6153846153846154e-06, "loss": 1.4724, "step": 68 }, { "epoch": 0.11, "learning_rate": 1.6666666666666667e-06, "loss": 1.4255, "step": 70 }, { "epoch": 0.11, "learning_rate": 1.7179487179487177e-06, "loss": 1.4795, "step": 72 }, { "epoch": 0.11, "learning_rate": 1.769230769230769e-06, "loss": 1.4737, "step": 74 }, { "epoch": 0.12, "learning_rate": 1.8205128205128203e-06, "loss": 1.4159, "step": 76 }, { "epoch": 0.12, "learning_rate": 1.8717948717948718e-06, "loss": 1.4782, "step": 78 }, { "epoch": 0.12, "learning_rate": 1.923076923076923e-06, "loss": 1.4622, "step": 80 }, { "epoch": 0.13, "learning_rate": 1.9743589743589744e-06, "loss": 1.384, "step": 82 }, { "epoch": 0.13, "learning_rate": 1.999999216710599e-06, "loss": 1.4845, "step": 84 }, { "epoch": 0.13, "learning_rate": 1.999992950402752e-06, "loss": 1.5691, "step": 86 }, { "epoch": 0.14, "learning_rate": 1.999980417826325e-06, "loss": 1.4263, "step": 88 }, { "epoch": 0.14, "learning_rate": 1.9999616190598506e-06, "loss": 1.4873, "step": 90 }, { "epoch": 0.14, "learning_rate": 1.999936554221128e-06, "loss": 1.5156, "step": 92 }, { "epoch": 0.15, "learning_rate": 1.9999052234672214e-06, "loss": 1.4825, "step": 94 }, { "epoch": 0.15, "learning_rate": 1.9998676269944592e-06, "loss": 1.3912, "step": 96 }, { "epoch": 0.15, "learning_rate": 1.999823765038432e-06, "loss": 1.4633, "step": 98 }, { "epoch": 0.15, "learning_rate": 1.9997736378739935e-06, "loss": 1.5286, "step": 100 }, { "epoch": 0.16, "learning_rate": 1.9997172458152555e-06, "loss": 1.5165, "step": 102 }, { "epoch": 0.16, "learning_rate": 1.999654589215588e-06, "loss": 1.4677, "step": 104 }, { "epoch": 0.16, "learning_rate": 1.999585668467618e-06, "loss": 1.4347, "step": 106 }, { "epoch": 0.17, "learning_rate": 1.9995104840032234e-06, "loss": 1.4824, "step": 108 }, { "epoch": 0.17, "learning_rate": 1.9994290362935343e-06, "loss": 1.5422, "step": 110 }, { "epoch": 0.17, "learning_rate": 1.999341325848927e-06, "loss": 1.486, "step": 112 }, { "epoch": 0.18, "learning_rate": 1.9992473532190224e-06, "loss": 1.4434, "step": 114 }, { "epoch": 0.18, "learning_rate": 1.999147118992683e-06, "loss": 1.4272, "step": 116 }, { "epoch": 0.18, "learning_rate": 1.999040623798008e-06, "loss": 1.4451, "step": 118 }, { "epoch": 0.19, "learning_rate": 1.9989278683023284e-06, "loss": 1.4454, "step": 120 }, { "epoch": 0.19, "learning_rate": 1.9988088532122064e-06, "loss": 1.5033, "step": 122 }, { "epoch": 0.19, "learning_rate": 1.998683579273427e-06, "loss": 1.4574, "step": 124 }, { "epoch": 0.19, "learning_rate": 1.998552047270997e-06, "loss": 1.5126, "step": 126 }, { "epoch": 0.2, "learning_rate": 1.9984142580291367e-06, "loss": 1.5314, "step": 128 }, { "epoch": 0.2, "learning_rate": 1.998270212411276e-06, "loss": 1.3909, "step": 130 }, { "epoch": 0.2, "learning_rate": 1.99811991132005e-06, "loss": 1.4447, "step": 132 }, { "epoch": 0.21, "learning_rate": 1.9979633556972927e-06, "loss": 1.4373, "step": 134 }, { "epoch": 0.21, "learning_rate": 1.9978005465240303e-06, "loss": 1.5037, "step": 136 }, { "epoch": 0.21, "learning_rate": 1.997631484820476e-06, "loss": 1.478, "step": 138 }, { "epoch": 0.22, "learning_rate": 1.997456171646023e-06, "loss": 1.3734, "step": 140 }, { "epoch": 0.22, "learning_rate": 1.9972746080992398e-06, "loss": 1.4159, "step": 142 }, { "epoch": 0.22, "learning_rate": 1.997086795317859e-06, "loss": 1.3781, "step": 144 }, { "epoch": 0.23, "learning_rate": 1.9968927344787744e-06, "loss": 1.4486, "step": 146 }, { "epoch": 0.23, "learning_rate": 1.9966924267980322e-06, "loss": 1.4606, "step": 148 }, { "epoch": 0.23, "learning_rate": 1.9964858735308234e-06, "loss": 1.3898, "step": 150 }, { "epoch": 0.23, "learning_rate": 1.9962730759714745e-06, "loss": 1.4843, "step": 152 }, { "epoch": 0.24, "learning_rate": 1.996054035453442e-06, "loss": 1.4658, "step": 154 }, { "epoch": 0.24, "learning_rate": 1.9958287533493024e-06, "loss": 1.4858, "step": 156 }, { "epoch": 0.24, "learning_rate": 1.995597231070744e-06, "loss": 1.4944, "step": 158 }, { "epoch": 0.25, "learning_rate": 1.995359470068557e-06, "loss": 1.3798, "step": 160 }, { "epoch": 0.25, "learning_rate": 1.995115471832627e-06, "loss": 1.4059, "step": 162 }, { "epoch": 0.25, "learning_rate": 1.994865237891923e-06, "loss": 1.524, "step": 164 }, { "epoch": 0.26, "learning_rate": 1.994608769814489e-06, "loss": 1.4069, "step": 166 }, { "epoch": 0.26, "learning_rate": 1.994346069207434e-06, "loss": 1.4051, "step": 168 }, { "epoch": 0.26, "learning_rate": 1.994077137716923e-06, "loss": 1.4122, "step": 170 }, { "epoch": 0.27, "learning_rate": 1.9938019770281636e-06, "loss": 1.4409, "step": 172 }, { "epoch": 0.27, "learning_rate": 1.9935205888654e-06, "loss": 1.4113, "step": 174 }, { "epoch": 0.27, "learning_rate": 1.993232974991898e-06, "loss": 1.4351, "step": 176 }, { "epoch": 0.28, "learning_rate": 1.992939137209935e-06, "loss": 1.5097, "step": 178 }, { "epoch": 0.28, "learning_rate": 1.992639077360792e-06, "loss": 1.499, "step": 180 }, { "epoch": 0.28, "learning_rate": 1.9923327973247374e-06, "loss": 1.4899, "step": 182 }, { "epoch": 0.28, "learning_rate": 1.992020299021017e-06, "loss": 1.4555, "step": 184 }, { "epoch": 0.29, "learning_rate": 1.9917015844078436e-06, "loss": 1.5307, "step": 186 }, { "epoch": 0.29, "learning_rate": 1.991376655482383e-06, "loss": 1.4961, "step": 188 }, { "epoch": 0.29, "learning_rate": 1.9910455142807404e-06, "loss": 1.426, "step": 190 }, { "epoch": 0.3, "learning_rate": 1.990708162877951e-06, "loss": 1.4407, "step": 192 }, { "epoch": 0.3, "learning_rate": 1.9903646033879636e-06, "loss": 1.5055, "step": 194 }, { "epoch": 0.3, "learning_rate": 1.9900148379636295e-06, "loss": 1.4929, "step": 196 }, { "epoch": 0.31, "learning_rate": 1.9896588687966888e-06, "loss": 1.3952, "step": 198 }, { "epoch": 0.31, "learning_rate": 1.989296698117755e-06, "loss": 1.4858, "step": 200 }, { "epoch": 0.31, "learning_rate": 1.9889283281963033e-06, "loss": 1.4649, "step": 202 }, { "epoch": 0.32, "learning_rate": 1.988553761340655e-06, "loss": 1.53, "step": 204 }, { "epoch": 0.32, "learning_rate": 1.9881729998979624e-06, "loss": 1.4344, "step": 206 }, { "epoch": 0.32, "learning_rate": 1.987786046254196e-06, "loss": 1.4173, "step": 208 }, { "epoch": 0.32, "learning_rate": 1.987392902834129e-06, "loss": 1.4983, "step": 210 }, { "epoch": 0.33, "learning_rate": 1.9869935721013206e-06, "loss": 1.4794, "step": 212 }, { "epoch": 0.33, "learning_rate": 1.9865880565581016e-06, "loss": 1.3269, "step": 214 }, { "epoch": 0.33, "learning_rate": 1.98617635874556e-06, "loss": 1.429, "step": 216 }, { "epoch": 0.34, "learning_rate": 1.9857584812435226e-06, "loss": 1.4152, "step": 218 }, { "epoch": 0.34, "learning_rate": 1.985334426670541e-06, "loss": 1.5482, "step": 220 }, { "epoch": 0.34, "learning_rate": 1.984904197683873e-06, "loss": 1.4803, "step": 222 }, { "epoch": 0.35, "learning_rate": 1.984467796979468e-06, "loss": 1.608, "step": 224 }, { "epoch": 0.35, "learning_rate": 1.9840252272919503e-06, "loss": 1.4862, "step": 226 }, { "epoch": 0.35, "learning_rate": 1.9835764913945995e-06, "loss": 1.4762, "step": 228 }, { "epoch": 0.36, "learning_rate": 1.983121592099335e-06, "loss": 1.4883, "step": 230 }, { "epoch": 0.36, "learning_rate": 1.9826605322566976e-06, "loss": 1.5459, "step": 232 }, { "epoch": 0.36, "learning_rate": 1.982193314755833e-06, "loss": 1.5177, "step": 234 }, { "epoch": 0.36, "learning_rate": 1.9817199425244724e-06, "loss": 1.5896, "step": 236 }, { "epoch": 0.37, "learning_rate": 1.9812404185289137e-06, "loss": 1.4123, "step": 238 }, { "epoch": 0.37, "learning_rate": 1.9807547457740045e-06, "loss": 1.4553, "step": 240 }, { "epoch": 0.37, "learning_rate": 1.980262927303122e-06, "loss": 1.5094, "step": 242 }, { "epoch": 0.38, "learning_rate": 1.9797649661981547e-06, "loss": 1.5127, "step": 244 }, { "epoch": 0.38, "learning_rate": 1.979260865579483e-06, "loss": 1.4149, "step": 246 }, { "epoch": 0.38, "learning_rate": 1.978750628605958e-06, "loss": 1.4845, "step": 248 }, { "epoch": 0.39, "learning_rate": 1.978234258474885e-06, "loss": 1.4417, "step": 250 }, { "epoch": 0.39, "learning_rate": 1.977711758422001e-06, "loss": 1.5293, "step": 252 }, { "epoch": 0.39, "learning_rate": 1.9771831317214538e-06, "loss": 1.4598, "step": 254 }, { "epoch": 0.4, "learning_rate": 1.9766483816857845e-06, "loss": 1.5163, "step": 256 }, { "epoch": 0.4, "learning_rate": 1.9761075116659033e-06, "loss": 1.4675, "step": 258 }, { "epoch": 0.4, "learning_rate": 1.9755605250510715e-06, "loss": 1.4856, "step": 260 }, { "epoch": 0.4, "learning_rate": 1.9750074252688786e-06, "loss": 1.5187, "step": 262 }, { "epoch": 0.41, "learning_rate": 1.9744482157852196e-06, "loss": 1.4705, "step": 264 }, { "epoch": 0.41, "learning_rate": 1.973882900104277e-06, "loss": 1.4752, "step": 266 }, { "epoch": 0.41, "learning_rate": 1.9733114817684955e-06, "loss": 1.4844, "step": 268 }, { "epoch": 0.42, "learning_rate": 1.9727339643585606e-06, "loss": 1.4824, "step": 270 }, { "epoch": 0.42, "learning_rate": 1.972150351493378e-06, "loss": 1.4551, "step": 272 }, { "epoch": 0.42, "learning_rate": 1.971560646830048e-06, "loss": 1.4157, "step": 274 }, { "epoch": 0.43, "learning_rate": 1.970964854063844e-06, "loss": 1.4279, "step": 276 }, { "epoch": 0.43, "learning_rate": 1.97036297692819e-06, "loss": 1.5939, "step": 278 }, { "epoch": 0.43, "learning_rate": 1.9697550191946365e-06, "loss": 1.5021, "step": 280 }, { "epoch": 0.44, "learning_rate": 1.9691409846728367e-06, "loss": 1.4876, "step": 282 }, { "epoch": 0.44, "learning_rate": 1.9685208772105234e-06, "loss": 1.5089, "step": 284 }, { "epoch": 0.44, "learning_rate": 1.9678947006934833e-06, "loss": 1.4859, "step": 286 }, { "epoch": 0.44, "learning_rate": 1.967262459045535e-06, "loss": 1.4721, "step": 288 }, { "epoch": 0.45, "learning_rate": 1.9666241562285014e-06, "loss": 1.5576, "step": 290 }, { "epoch": 0.45, "learning_rate": 1.9659797962421886e-06, "loss": 1.4888, "step": 292 }, { "epoch": 0.45, "learning_rate": 1.965329383124357e-06, "loss": 1.5399, "step": 294 }, { "epoch": 0.46, "learning_rate": 1.9646729209506992e-06, "loss": 1.4362, "step": 296 }, { "epoch": 0.46, "learning_rate": 1.964010413834812e-06, "loss": 1.5217, "step": 298 }, { "epoch": 0.46, "learning_rate": 1.963341865928173e-06, "loss": 1.4794, "step": 300 }, { "epoch": 0.47, "learning_rate": 1.9626672814201115e-06, "loss": 1.4082, "step": 302 }, { "epoch": 0.47, "learning_rate": 1.961986664537785e-06, "loss": 1.4707, "step": 304 }, { "epoch": 0.47, "learning_rate": 1.961300019546153e-06, "loss": 1.4886, "step": 306 }, { "epoch": 0.48, "learning_rate": 1.9606073507479463e-06, "loss": 1.5323, "step": 308 }, { "epoch": 0.48, "learning_rate": 1.959908662483645e-06, "loss": 1.4301, "step": 310 }, { "epoch": 0.48, "learning_rate": 1.9592039591314484e-06, "loss": 1.4645, "step": 312 }, { "epoch": 0.49, "learning_rate": 1.958493245107248e-06, "loss": 1.5434, "step": 314 }, { "epoch": 0.49, "learning_rate": 1.9577765248645997e-06, "loss": 1.5357, "step": 316 }, { "epoch": 0.49, "learning_rate": 1.957053802894697e-06, "loss": 1.4056, "step": 318 }, { "epoch": 0.49, "learning_rate": 1.956325083726342e-06, "loss": 1.4653, "step": 320 }, { "epoch": 0.5, "learning_rate": 1.9555903719259166e-06, "loss": 1.5146, "step": 322 }, { "epoch": 0.5, "learning_rate": 1.9548496720973547e-06, "loss": 1.4518, "step": 324 }, { "epoch": 0.5, "learning_rate": 1.9541029888821136e-06, "loss": 1.3948, "step": 326 }, { "epoch": 0.51, "learning_rate": 1.9533503269591435e-06, "loss": 1.3753, "step": 328 }, { "epoch": 0.51, "learning_rate": 1.952591691044859e-06, "loss": 1.4757, "step": 330 }, { "epoch": 0.51, "learning_rate": 1.951827085893111e-06, "loss": 1.507, "step": 332 }, { "epoch": 0.52, "learning_rate": 1.9510565162951534e-06, "loss": 1.4706, "step": 334 }, { "epoch": 0.52, "learning_rate": 1.950279987079617e-06, "loss": 1.4482, "step": 336 }, { "epoch": 0.52, "learning_rate": 1.9494975031124763e-06, "loss": 1.4257, "step": 338 }, { "epoch": 0.53, "learning_rate": 1.9487090692970213e-06, "loss": 1.4531, "step": 340 }, { "epoch": 0.53, "learning_rate": 1.9479146905738243e-06, "loss": 1.3992, "step": 342 }, { "epoch": 0.53, "learning_rate": 1.947114371920711e-06, "loss": 1.4718, "step": 344 }, { "epoch": 0.53, "learning_rate": 1.946308118352728e-06, "loss": 1.5628, "step": 346 }, { "epoch": 0.54, "learning_rate": 1.945495934922113e-06, "loss": 1.3921, "step": 348 }, { "epoch": 0.54, "learning_rate": 1.9446778267182602e-06, "loss": 1.4406, "step": 350 }, { "epoch": 0.54, "learning_rate": 1.943853798867693e-06, "loss": 1.5219, "step": 352 }, { "epoch": 0.55, "learning_rate": 1.9430238565340267e-06, "loss": 1.4163, "step": 354 }, { "epoch": 0.55, "learning_rate": 1.94218800491794e-06, "loss": 1.4737, "step": 356 }, { "epoch": 0.55, "learning_rate": 1.94134624925714e-06, "loss": 1.3702, "step": 358 }, { "epoch": 0.56, "learning_rate": 1.9404985948263315e-06, "loss": 1.4473, "step": 360 }, { "epoch": 0.56, "learning_rate": 1.9396450469371816e-06, "loss": 1.4089, "step": 362 }, { "epoch": 0.56, "learning_rate": 1.938785610938289e-06, "loss": 1.4384, "step": 364 }, { "epoch": 0.57, "learning_rate": 1.937920292215148e-06, "loss": 1.3912, "step": 366 }, { "epoch": 0.57, "learning_rate": 1.9370490961901166e-06, "loss": 1.5239, "step": 368 }, { "epoch": 0.57, "learning_rate": 1.9361720283223815e-06, "loss": 1.4259, "step": 370 }, { "epoch": 0.57, "learning_rate": 1.9352890941079247e-06, "loss": 1.5154, "step": 372 }, { "epoch": 0.58, "learning_rate": 1.934400299079487e-06, "loss": 1.4081, "step": 374 }, { "epoch": 0.58, "learning_rate": 1.933505648806537e-06, "loss": 1.4375, "step": 376 }, { "epoch": 0.58, "learning_rate": 1.932605148895233e-06, "loss": 1.4245, "step": 378 }, { "epoch": 0.59, "learning_rate": 1.9316988049883887e-06, "loss": 1.5242, "step": 380 }, { "epoch": 0.59, "learning_rate": 1.930786622765439e-06, "loss": 1.4617, "step": 382 }, { "epoch": 0.59, "learning_rate": 1.929868607942402e-06, "loss": 1.4312, "step": 384 }, { "epoch": 0.6, "learning_rate": 1.928944766271847e-06, "loss": 1.4371, "step": 386 }, { "epoch": 0.6, "learning_rate": 1.9280151035428544e-06, "loss": 1.3988, "step": 388 }, { "epoch": 0.6, "learning_rate": 1.927079625580981e-06, "loss": 1.412, "step": 390 }, { "epoch": 0.61, "learning_rate": 1.9261383382482247e-06, "loss": 1.3855, "step": 392 }, { "epoch": 0.61, "learning_rate": 1.9251912474429867e-06, "loss": 1.3964, "step": 394 }, { "epoch": 0.61, "learning_rate": 1.924238359100034e-06, "loss": 1.4483, "step": 396 }, { "epoch": 0.61, "learning_rate": 1.9232796791904625e-06, "loss": 1.4555, "step": 398 }, { "epoch": 0.62, "learning_rate": 1.9223152137216604e-06, "loss": 1.4338, "step": 400 }, { "epoch": 0.62, "learning_rate": 1.921344968737271e-06, "loss": 1.467, "step": 402 }, { "epoch": 0.62, "learning_rate": 1.9203689503171513e-06, "loss": 1.3961, "step": 404 }, { "epoch": 0.63, "learning_rate": 1.919387164577339e-06, "loss": 1.472, "step": 406 }, { "epoch": 0.63, "learning_rate": 1.9183996176700106e-06, "loss": 1.4186, "step": 408 }, { "epoch": 0.63, "learning_rate": 1.9174063157834434e-06, "loss": 1.4305, "step": 410 }, { "epoch": 0.64, "learning_rate": 1.916407265141978e-06, "loss": 1.4349, "step": 412 }, { "epoch": 0.64, "learning_rate": 1.915402472005978e-06, "loss": 1.4141, "step": 414 }, { "epoch": 0.64, "learning_rate": 1.9143919426717916e-06, "loss": 1.4687, "step": 416 }, { "epoch": 0.65, "learning_rate": 1.9133756834717115e-06, "loss": 1.3353, "step": 418 }, { "epoch": 0.65, "learning_rate": 1.912353700773936e-06, "loss": 1.4682, "step": 420 }, { "epoch": 0.65, "learning_rate": 1.911326000982528e-06, "loss": 1.524, "step": 422 }, { "epoch": 0.66, "learning_rate": 1.9102925905373757e-06, "loss": 1.4579, "step": 424 }, { "epoch": 0.66, "learning_rate": 1.9092534759141527e-06, "loss": 1.3845, "step": 426 }, { "epoch": 0.66, "learning_rate": 1.9082086636242755e-06, "loss": 1.4535, "step": 428 }, { "epoch": 0.66, "learning_rate": 1.9071581602148651e-06, "loss": 1.463, "step": 430 }, { "epoch": 0.67, "learning_rate": 1.9061019722687047e-06, "loss": 1.455, "step": 432 }, { "epoch": 0.67, "learning_rate": 1.9050401064041975e-06, "loss": 1.4252, "step": 434 }, { "epoch": 0.67, "learning_rate": 1.9039725692753279e-06, "loss": 1.599, "step": 436 }, { "epoch": 0.68, "learning_rate": 1.902899367571617e-06, "loss": 1.5299, "step": 438 }, { "epoch": 0.68, "learning_rate": 1.9018205080180825e-06, "loss": 1.4192, "step": 440 }, { "epoch": 0.68, "learning_rate": 1.9007359973751957e-06, "loss": 1.4457, "step": 442 }, { "epoch": 0.69, "learning_rate": 1.8996458424388396e-06, "loss": 1.507, "step": 444 }, { "epoch": 0.69, "learning_rate": 1.8985500500402654e-06, "loss": 1.5039, "step": 446 }, { "epoch": 0.69, "learning_rate": 1.8974486270460516e-06, "loss": 1.5077, "step": 448 }, { "epoch": 0.7, "learning_rate": 1.8963415803580593e-06, "loss": 1.4559, "step": 450 }, { "epoch": 0.7, "learning_rate": 1.895228916913389e-06, "loss": 1.4555, "step": 452 }, { "epoch": 0.7, "learning_rate": 1.8941106436843378e-06, "loss": 1.4938, "step": 454 }, { "epoch": 0.7, "learning_rate": 1.8929867676783553e-06, "loss": 1.4056, "step": 456 }, { "epoch": 0.71, "learning_rate": 1.8918572959380003e-06, "loss": 1.4415, "step": 458 }, { "epoch": 0.71, "learning_rate": 1.890722235540896e-06, "loss": 1.4591, "step": 460 }, { "epoch": 0.71, "learning_rate": 1.889581593599686e-06, "loss": 1.3919, "step": 462 }, { "epoch": 0.72, "learning_rate": 1.8884353772619889e-06, "loss": 1.5499, "step": 464 }, { "epoch": 0.72, "learning_rate": 1.8872835937103553e-06, "loss": 1.4263, "step": 466 }, { "epoch": 0.72, "learning_rate": 1.886126250162221e-06, "loss": 1.4777, "step": 468 }, { "epoch": 0.73, "learning_rate": 1.8849633538698627e-06, "loss": 1.4812, "step": 470 }, { "epoch": 0.73, "learning_rate": 1.8837949121203518e-06, "loss": 1.4986, "step": 472 }, { "epoch": 0.73, "learning_rate": 1.8826209322355102e-06, "loss": 1.5011, "step": 474 }, { "epoch": 0.74, "learning_rate": 1.881441421571863e-06, "loss": 1.3221, "step": 476 }, { "epoch": 0.74, "learning_rate": 1.880256387520593e-06, "loss": 1.4712, "step": 478 }, { "epoch": 0.74, "learning_rate": 1.8790658375074936e-06, "loss": 1.4071, "step": 480 }, { "epoch": 0.74, "learning_rate": 1.8784684963452718e-06, "loss": 1.4792, "step": 482 }, { "epoch": 0.75, "learning_rate": 1.8772696863883904e-06, "loss": 1.4771, "step": 484 }, { "epoch": 0.75, "learning_rate": 1.8760653791852836e-06, "loss": 1.438, "step": 486 }, { "epoch": 0.75, "learning_rate": 1.8748555822825169e-06, "loss": 1.4643, "step": 488 }, { "epoch": 0.76, "learning_rate": 1.873640303261056e-06, "loss": 1.3929, "step": 490 }, { "epoch": 0.76, "learning_rate": 1.8724195497362192e-06, "loss": 1.5291, "step": 492 }, { "epoch": 0.76, "learning_rate": 1.87119332935763e-06, "loss": 1.4385, "step": 494 }, { "epoch": 0.77, "learning_rate": 1.869961649809169e-06, "loss": 1.4657, "step": 496 }, { "epoch": 0.77, "learning_rate": 1.868724518808925e-06, "loss": 1.3716, "step": 498 }, { "epoch": 0.77, "learning_rate": 1.8674819441091483e-06, "loss": 1.526, "step": 500 }, { "epoch": 0.78, "learning_rate": 1.8662339334962001e-06, "loss": 1.516, "step": 502 }, { "epoch": 0.78, "learning_rate": 1.8649804947905055e-06, "loss": 1.6113, "step": 504 }, { "epoch": 0.78, "learning_rate": 1.8637216358465034e-06, "loss": 1.4179, "step": 506 }, { "epoch": 0.78, "learning_rate": 1.8624573645525978e-06, "loss": 1.3795, "step": 508 }, { "epoch": 0.79, "learning_rate": 1.8611876888311078e-06, "loss": 1.5263, "step": 510 }, { "epoch": 0.79, "learning_rate": 1.8599126166382186e-06, "loss": 1.53, "step": 512 }, { "epoch": 0.79, "learning_rate": 1.8586321559639315e-06, "loss": 1.4668, "step": 514 }, { "epoch": 0.8, "learning_rate": 1.8573463148320134e-06, "loss": 1.3984, "step": 516 }, { "epoch": 0.8, "learning_rate": 1.856055101299947e-06, "loss": 1.4846, "step": 518 }, { "epoch": 0.8, "learning_rate": 1.8547585234588803e-06, "loss": 1.5401, "step": 520 }, { "epoch": 0.81, "learning_rate": 1.8534565894335755e-06, "loss": 1.4178, "step": 522 }, { "epoch": 0.81, "learning_rate": 1.852149307382358e-06, "loss": 1.4143, "step": 524 }, { "epoch": 0.81, "learning_rate": 1.8508366854970668e-06, "loss": 1.4739, "step": 526 }, { "epoch": 0.82, "learning_rate": 1.8495187320030002e-06, "loss": 1.4018, "step": 528 }, { "epoch": 0.82, "learning_rate": 1.8481954551588679e-06, "loss": 1.5131, "step": 530 }, { "epoch": 0.82, "learning_rate": 1.8468668632567357e-06, "loss": 1.411, "step": 532 }, { "epoch": 0.83, "learning_rate": 1.8455329646219764e-06, "loss": 1.4803, "step": 534 }, { "epoch": 0.83, "learning_rate": 1.844193767613216e-06, "loss": 1.4606, "step": 536 }, { "epoch": 0.83, "learning_rate": 1.8428492806222819e-06, "loss": 1.4787, "step": 538 }, { "epoch": 0.83, "learning_rate": 1.8414995120741497e-06, "loss": 1.497, "step": 540 }, { "epoch": 0.84, "learning_rate": 1.8401444704268916e-06, "loss": 1.5337, "step": 542 }, { "epoch": 0.84, "learning_rate": 1.8387841641716222e-06, "loss": 1.4635, "step": 544 }, { "epoch": 0.84, "learning_rate": 1.837418601832446e-06, "loss": 1.3847, "step": 546 }, { "epoch": 0.85, "learning_rate": 1.8360477919664037e-06, "loss": 1.4126, "step": 548 }, { "epoch": 0.85, "learning_rate": 1.8346717431634186e-06, "loss": 1.4833, "step": 550 }, { "epoch": 0.85, "learning_rate": 1.833290464046243e-06, "loss": 1.4478, "step": 552 }, { "epoch": 0.86, "learning_rate": 1.831903963270404e-06, "loss": 1.5498, "step": 554 }, { "epoch": 0.86, "learning_rate": 1.8305122495241486e-06, "loss": 1.4709, "step": 556 }, { "epoch": 0.86, "learning_rate": 1.8291153315283904e-06, "loss": 1.4051, "step": 558 }, { "epoch": 0.87, "learning_rate": 1.8277132180366551e-06, "loss": 1.4307, "step": 560 }, { "epoch": 0.87, "learning_rate": 1.8263059178350237e-06, "loss": 1.5244, "step": 562 }, { "epoch": 0.87, "learning_rate": 1.82489343974208e-06, "loss": 1.436, "step": 564 }, { "epoch": 0.87, "learning_rate": 1.8234757926088528e-06, "loss": 1.4165, "step": 566 }, { "epoch": 0.88, "learning_rate": 1.8220529853187632e-06, "loss": 1.5133, "step": 568 }, { "epoch": 0.88, "learning_rate": 1.820625026787566e-06, "loss": 1.6104, "step": 570 }, { "epoch": 0.88, "learning_rate": 1.8191919259632962e-06, "loss": 1.365, "step": 572 }, { "epoch": 0.89, "learning_rate": 1.817753691826212e-06, "loss": 1.4513, "step": 574 }, { "epoch": 0.89, "learning_rate": 1.8163103333887377e-06, "loss": 1.4228, "step": 576 }, { "epoch": 0.89, "learning_rate": 1.8148618596954092e-06, "loss": 1.4639, "step": 578 }, { "epoch": 0.9, "learning_rate": 1.8134082798228158e-06, "loss": 1.4448, "step": 580 }, { "epoch": 0.9, "learning_rate": 1.8119496028795432e-06, "loss": 1.3983, "step": 582 }, { "epoch": 0.9, "learning_rate": 1.8104858380061176e-06, "loss": 1.4449, "step": 584 }, { "epoch": 0.91, "learning_rate": 1.8090169943749474e-06, "loss": 1.4484, "step": 586 }, { "epoch": 0.91, "learning_rate": 1.8075430811902662e-06, "loss": 1.4015, "step": 588 }, { "epoch": 0.91, "learning_rate": 1.806064107688075e-06, "loss": 1.5276, "step": 590 }, { "epoch": 0.91, "learning_rate": 1.804580083136084e-06, "loss": 1.4602, "step": 592 }, { "epoch": 0.92, "learning_rate": 1.8030910168336555e-06, "loss": 1.4111, "step": 594 }, { "epoch": 0.92, "learning_rate": 1.8015969181117447e-06, "loss": 1.3566, "step": 596 }, { "epoch": 0.92, "learning_rate": 1.8000977963328416e-06, "loss": 1.3934, "step": 598 }, { "epoch": 0.93, "learning_rate": 1.7985936608909121e-06, "loss": 1.4942, "step": 600 }, { "epoch": 0.93, "learning_rate": 1.7970845212113387e-06, "loss": 1.421, "step": 602 }, { "epoch": 0.93, "learning_rate": 1.795570386750863e-06, "loss": 1.445, "step": 604 }, { "epoch": 0.94, "learning_rate": 1.7940512669975254e-06, "loss": 1.4394, "step": 606 }, { "epoch": 0.94, "learning_rate": 1.7925271714706053e-06, "loss": 1.5447, "step": 608 }, { "epoch": 0.94, "learning_rate": 1.7909981097205615e-06, "loss": 1.3753, "step": 610 }, { "epoch": 0.95, "learning_rate": 1.7894640913289735e-06, "loss": 1.4693, "step": 612 }, { "epoch": 0.95, "learning_rate": 1.7879251259084802e-06, "loss": 1.4715, "step": 614 }, { "epoch": 0.95, "learning_rate": 1.7863812231027202e-06, "loss": 1.5163, "step": 616 }, { "epoch": 0.95, "learning_rate": 1.7848323925862715e-06, "loss": 1.507, "step": 618 }, { "epoch": 0.96, "learning_rate": 1.7832786440645907e-06, "loss": 1.409, "step": 620 }, { "epoch": 0.96, "learning_rate": 1.7817199872739514e-06, "loss": 1.5028, "step": 622 }, { "epoch": 0.96, "learning_rate": 1.780156431981385e-06, "loss": 1.5621, "step": 624 }, { "epoch": 0.97, "learning_rate": 1.778587987984618e-06, "loss": 1.4952, "step": 626 }, { "epoch": 0.97, "learning_rate": 1.7770146651120106e-06, "loss": 1.4556, "step": 628 }, { "epoch": 0.97, "learning_rate": 1.7754364732224963e-06, "loss": 1.4562, "step": 630 }, { "epoch": 0.98, "learning_rate": 1.7738534222055189e-06, "loss": 1.4505, "step": 632 }, { "epoch": 0.98, "learning_rate": 1.7722655219809714e-06, "loss": 1.3392, "step": 634 }, { "epoch": 0.98, "learning_rate": 1.7706727824991332e-06, "loss": 1.3622, "step": 636 }, { "epoch": 0.99, "learning_rate": 1.7690752137406078e-06, "loss": 1.4267, "step": 638 }, { "epoch": 0.99, "learning_rate": 1.7674728257162608e-06, "loss": 1.3587, "step": 640 }, { "epoch": 0.99, "learning_rate": 1.7658656284671566e-06, "loss": 1.4474, "step": 642 }, { "epoch": 0.99, "learning_rate": 1.7642536320644962e-06, "loss": 1.3939, "step": 644 }, { "epoch": 1.0, "learning_rate": 1.7626368466095532e-06, "loss": 1.4862, "step": 646 }, { "epoch": 1.0, "learning_rate": 1.7610152822336103e-06, "loss": 1.4565, "step": 648 }, { "epoch": 1.0, "learning_rate": 1.7593889490978976e-06, "loss": 1.3179, "step": 650 }, { "epoch": 1.01, "learning_rate": 1.757757857393527e-06, "loss": 1.3548, "step": 652 }, { "epoch": 1.01, "learning_rate": 1.7561220173414296e-06, "loss": 1.3662, "step": 654 }, { "epoch": 1.01, "learning_rate": 1.75448143919229e-06, "loss": 1.3691, "step": 656 }, { "epoch": 1.02, "learning_rate": 1.7528361332264844e-06, "loss": 1.3861, "step": 658 }, { "epoch": 1.02, "learning_rate": 1.7511861097540147e-06, "loss": 1.4284, "step": 660 }, { "epoch": 1.02, "learning_rate": 1.749531379114444e-06, "loss": 1.3508, "step": 662 }, { "epoch": 1.03, "learning_rate": 1.747871951676832e-06, "loss": 1.3871, "step": 664 }, { "epoch": 1.03, "learning_rate": 1.7462078378396702e-06, "loss": 1.3648, "step": 666 }, { "epoch": 1.03, "learning_rate": 1.7445390480308154e-06, "loss": 1.4301, "step": 668 }, { "epoch": 1.04, "learning_rate": 1.7428655927074277e-06, "loss": 1.4273, "step": 670 }, { "epoch": 1.04, "learning_rate": 1.7411874823559009e-06, "loss": 1.4083, "step": 672 }, { "epoch": 1.04, "learning_rate": 1.7395047274917993e-06, "loss": 1.3737, "step": 674 }, { "epoch": 1.04, "learning_rate": 1.7378173386597916e-06, "loss": 1.2966, "step": 676 }, { "epoch": 1.05, "learning_rate": 1.7361253264335836e-06, "loss": 1.3119, "step": 678 }, { "epoch": 1.05, "learning_rate": 1.734428701415853e-06, "loss": 1.2907, "step": 680 }, { "epoch": 1.05, "learning_rate": 1.732727474238183e-06, "loss": 1.3785, "step": 682 }, { "epoch": 1.06, "learning_rate": 1.731021655560995e-06, "loss": 1.3756, "step": 684 }, { "epoch": 1.06, "learning_rate": 1.7293112560734823e-06, "loss": 1.4106, "step": 686 }, { "epoch": 1.06, "learning_rate": 1.7275962864935433e-06, "loss": 1.3724, "step": 688 }, { "epoch": 1.07, "learning_rate": 1.7258767575677138e-06, "loss": 1.3926, "step": 690 }, { "epoch": 1.07, "learning_rate": 1.7241526800710995e-06, "loss": 1.4547, "step": 692 }, { "epoch": 1.07, "learning_rate": 1.7224240648073095e-06, "loss": 1.4637, "step": 694 }, { "epoch": 1.08, "learning_rate": 1.7206909226083871e-06, "loss": 1.3559, "step": 696 }, { "epoch": 1.08, "learning_rate": 1.718953264334744e-06, "loss": 1.3049, "step": 698 }, { "epoch": 1.08, "learning_rate": 1.7172111008750908e-06, "loss": 1.3698, "step": 700 }, { "epoch": 1.08, "learning_rate": 1.715464443146367e-06, "loss": 1.4218, "step": 702 }, { "epoch": 1.09, "learning_rate": 1.713713302093678e-06, "loss": 1.311, "step": 704 }, { "epoch": 1.09, "learning_rate": 1.7119576886902204e-06, "loss": 1.3742, "step": 706 }, { "epoch": 1.09, "learning_rate": 1.710197613937217e-06, "loss": 1.3536, "step": 708 }, { "epoch": 1.1, "learning_rate": 1.7084330888638463e-06, "loss": 1.3678, "step": 710 }, { "epoch": 1.1, "learning_rate": 1.706664124527175e-06, "loss": 1.381, "step": 712 }, { "epoch": 1.1, "learning_rate": 1.7048907320120863e-06, "loss": 1.3459, "step": 714 }, { "epoch": 1.11, "learning_rate": 1.703112922431213e-06, "loss": 1.3723, "step": 716 }, { "epoch": 1.11, "learning_rate": 1.7013307069248652e-06, "loss": 1.3054, "step": 718 }, { "epoch": 1.11, "learning_rate": 1.699544096660963e-06, "loss": 1.2933, "step": 720 }, { "epoch": 1.12, "learning_rate": 1.697753102834965e-06, "loss": 1.4572, "step": 722 }, { "epoch": 1.12, "learning_rate": 1.6959577366697989e-06, "loss": 1.3961, "step": 724 }, { "epoch": 1.12, "learning_rate": 1.6941580094157905e-06, "loss": 1.4541, "step": 726 }, { "epoch": 1.12, "learning_rate": 1.6923539323505933e-06, "loss": 1.3431, "step": 728 }, { "epoch": 1.13, "learning_rate": 1.690545516779119e-06, "loss": 1.3904, "step": 730 }, { "epoch": 1.13, "learning_rate": 1.6887327740334647e-06, "loss": 1.3672, "step": 732 }, { "epoch": 1.13, "learning_rate": 1.6869157154728435e-06, "loss": 1.3494, "step": 734 }, { "epoch": 1.14, "learning_rate": 1.685094352483513e-06, "loss": 1.2945, "step": 736 }, { "epoch": 1.14, "learning_rate": 1.6832686964787026e-06, "loss": 1.3793, "step": 738 }, { "epoch": 1.14, "learning_rate": 1.6814387588985443e-06, "loss": 1.3581, "step": 740 }, { "epoch": 1.15, "learning_rate": 1.6796045512099992e-06, "loss": 1.3424, "step": 742 }, { "epoch": 1.15, "learning_rate": 1.6777660849067867e-06, "loss": 1.4094, "step": 744 }, { "epoch": 1.15, "learning_rate": 1.675923371509311e-06, "loss": 1.3212, "step": 746 }, { "epoch": 1.16, "learning_rate": 1.674076422564591e-06, "loss": 1.4396, "step": 748 }, { "epoch": 1.16, "learning_rate": 1.6722252496461865e-06, "loss": 1.2924, "step": 750 }, { "epoch": 1.16, "learning_rate": 1.6703698643541255e-06, "loss": 1.3452, "step": 752 }, { "epoch": 1.16, "learning_rate": 1.6685102783148328e-06, "loss": 1.3087, "step": 754 }, { "epoch": 1.17, "learning_rate": 1.6666465031810561e-06, "loss": 1.3665, "step": 756 }, { "epoch": 1.17, "learning_rate": 1.6647785506317934e-06, "loss": 1.3527, "step": 758 }, { "epoch": 1.17, "learning_rate": 1.6629064323722196e-06, "loss": 1.2507, "step": 760 }, { "epoch": 1.18, "learning_rate": 1.6610301601336127e-06, "loss": 1.4094, "step": 762 }, { "epoch": 1.18, "learning_rate": 1.6591497456732824e-06, "loss": 1.3644, "step": 764 }, { "epoch": 1.18, "learning_rate": 1.6572652007744928e-06, "loss": 1.4727, "step": 766 }, { "epoch": 1.19, "learning_rate": 1.6553765372463921e-06, "loss": 1.426, "step": 768 }, { "epoch": 1.19, "learning_rate": 1.6534837669239367e-06, "loss": 1.5269, "step": 770 }, { "epoch": 1.19, "learning_rate": 1.6515869016678173e-06, "loss": 1.3328, "step": 772 }, { "epoch": 1.2, "learning_rate": 1.649685953364385e-06, "loss": 1.3446, "step": 774 }, { "epoch": 1.2, "learning_rate": 1.6477809339255759e-06, "loss": 1.3507, "step": 776 }, { "epoch": 1.2, "learning_rate": 1.645871855288838e-06, "loss": 1.4081, "step": 778 }, { "epoch": 1.21, "learning_rate": 1.6439587294170557e-06, "loss": 1.3495, "step": 780 }, { "epoch": 1.21, "learning_rate": 1.6420415682984725e-06, "loss": 1.4486, "step": 782 }, { "epoch": 1.21, "learning_rate": 1.640120383946621e-06, "loss": 1.376, "step": 784 }, { "epoch": 1.21, "learning_rate": 1.638195188400243e-06, "loss": 1.373, "step": 786 }, { "epoch": 1.22, "learning_rate": 1.636265993723215e-06, "loss": 1.3801, "step": 788 }, { "epoch": 1.22, "learning_rate": 1.6352999004875239e-06, "loss": 1.3762, "step": 790 }, { "epoch": 1.22, "learning_rate": 1.6333647297890897e-06, "loss": 1.4202, "step": 792 }, { "epoch": 1.23, "learning_rate": 1.631425590229171e-06, "loss": 1.4095, "step": 794 }, { "epoch": 1.23, "learning_rate": 1.6294824939590219e-06, "loss": 1.4065, "step": 796 }, { "epoch": 1.23, "learning_rate": 1.6275354531546918e-06, "loss": 1.4245, "step": 798 }, { "epoch": 1.24, "learning_rate": 1.6255844800169472e-06, "loss": 1.3452, "step": 800 }, { "epoch": 1.24, "learning_rate": 1.6236295867711961e-06, "loss": 1.2696, "step": 802 }, { "epoch": 1.24, "learning_rate": 1.621670785667411e-06, "loss": 1.4431, "step": 804 }, { "epoch": 1.25, "learning_rate": 1.619708088980052e-06, "loss": 1.3574, "step": 806 }, { "epoch": 1.25, "learning_rate": 1.6177415090079906e-06, "loss": 1.3617, "step": 808 }, { "epoch": 1.25, "learning_rate": 1.6157710580744319e-06, "loss": 1.3708, "step": 810 }, { "epoch": 1.25, "learning_rate": 1.6137967485268375e-06, "loss": 1.371, "step": 812 }, { "epoch": 1.26, "learning_rate": 1.611818592736849e-06, "loss": 1.2964, "step": 814 }, { "epoch": 1.26, "learning_rate": 1.6098366031002088e-06, "loss": 1.3985, "step": 816 }, { "epoch": 1.26, "learning_rate": 1.6078507920366843e-06, "loss": 1.2999, "step": 818 }, { "epoch": 1.27, "learning_rate": 1.605861171989988e-06, "loss": 1.3478, "step": 820 }, { "epoch": 1.27, "learning_rate": 1.6038677554277021e-06, "loss": 1.397, "step": 822 }, { "epoch": 1.27, "learning_rate": 1.6018705548411977e-06, "loss": 1.3495, "step": 824 }, { "epoch": 1.28, "learning_rate": 1.5998695827455583e-06, "loss": 1.4514, "step": 826 }, { "epoch": 1.28, "learning_rate": 1.5978648516795015e-06, "loss": 1.3361, "step": 828 }, { "epoch": 1.28, "learning_rate": 1.5958563742052985e-06, "loss": 1.2843, "step": 830 }, { "epoch": 1.29, "learning_rate": 1.5938441629086978e-06, "loss": 1.4017, "step": 832 }, { "epoch": 1.29, "learning_rate": 1.5918282303988444e-06, "loss": 1.4397, "step": 834 }, { "epoch": 1.29, "learning_rate": 1.5898085893082018e-06, "loss": 1.3588, "step": 836 }, { "epoch": 1.29, "learning_rate": 1.587785252292473e-06, "loss": 1.4295, "step": 838 }, { "epoch": 1.3, "learning_rate": 1.5857582320305205e-06, "loss": 1.3792, "step": 840 }, { "epoch": 1.3, "learning_rate": 1.5837275412242872e-06, "loss": 1.4565, "step": 842 }, { "epoch": 1.3, "learning_rate": 1.5816931925987165e-06, "loss": 1.3947, "step": 844 }, { "epoch": 1.31, "learning_rate": 1.5796551989016736e-06, "loss": 1.3763, "step": 846 }, { "epoch": 1.31, "learning_rate": 1.5776135729038642e-06, "loss": 1.3208, "step": 848 }, { "epoch": 1.31, "learning_rate": 1.5755683273987551e-06, "loss": 1.3798, "step": 850 }, { "epoch": 1.32, "learning_rate": 1.5735194752024952e-06, "loss": 1.3541, "step": 852 }, { "epoch": 1.32, "learning_rate": 1.5714670291538318e-06, "loss": 1.3397, "step": 854 }, { "epoch": 1.32, "learning_rate": 1.5694110021140348e-06, "loss": 1.3964, "step": 856 }, { "epoch": 1.33, "learning_rate": 1.567351406966812e-06, "loss": 1.368, "step": 858 }, { "epoch": 1.33, "learning_rate": 1.5652882566182312e-06, "loss": 1.4729, "step": 860 }, { "epoch": 1.33, "learning_rate": 1.5632215639966377e-06, "loss": 1.4207, "step": 862 }, { "epoch": 1.33, "learning_rate": 1.5611513420525737e-06, "loss": 1.4328, "step": 864 }, { "epoch": 1.34, "learning_rate": 1.5590776037586974e-06, "loss": 1.366, "step": 866 }, { "epoch": 1.34, "learning_rate": 1.5570003621097017e-06, "loss": 1.38, "step": 868 }, { "epoch": 1.34, "learning_rate": 1.554919630122232e-06, "loss": 1.3656, "step": 870 }, { "epoch": 1.35, "learning_rate": 1.5528354208348055e-06, "loss": 1.3077, "step": 872 }, { "epoch": 1.35, "learning_rate": 1.5507477473077306e-06, "loss": 1.3806, "step": 874 }, { "epoch": 1.35, "learning_rate": 1.5486566226230218e-06, "loss": 1.4179, "step": 876 }, { "epoch": 1.36, "learning_rate": 1.5465620598843202e-06, "loss": 1.3462, "step": 878 }, { "epoch": 1.36, "learning_rate": 1.5444640722168111e-06, "loss": 1.3775, "step": 880 }, { "epoch": 1.36, "learning_rate": 1.5423626727671418e-06, "loss": 1.347, "step": 882 }, { "epoch": 1.37, "learning_rate": 1.5402578747033382e-06, "loss": 1.3932, "step": 884 }, { "epoch": 1.37, "learning_rate": 1.5381496912147234e-06, "loss": 1.3797, "step": 886 }, { "epoch": 1.37, "learning_rate": 1.5360381355118338e-06, "loss": 1.3875, "step": 888 }, { "epoch": 1.38, "learning_rate": 1.533923220826339e-06, "loss": 1.4489, "step": 890 }, { "epoch": 1.38, "learning_rate": 1.5318049604109546e-06, "loss": 1.3263, "step": 892 }, { "epoch": 1.38, "learning_rate": 1.5296833675393642e-06, "loss": 1.3832, "step": 894 }, { "epoch": 1.38, "learning_rate": 1.527558455506131e-06, "loss": 1.3735, "step": 896 }, { "epoch": 1.39, "learning_rate": 1.5254302376266192e-06, "loss": 1.2865, "step": 898 }, { "epoch": 1.39, "learning_rate": 1.5232987272369073e-06, "loss": 1.379, "step": 900 }, { "epoch": 1.39, "learning_rate": 1.5211639376937063e-06, "loss": 1.355, "step": 902 }, { "epoch": 1.4, "learning_rate": 1.519025882374275e-06, "loss": 1.2924, "step": 904 }, { "epoch": 1.4, "learning_rate": 1.5168845746763363e-06, "loss": 1.4276, "step": 906 }, { "epoch": 1.4, "learning_rate": 1.5147400280179948e-06, "loss": 1.3739, "step": 908 }, { "epoch": 1.41, "learning_rate": 1.5125922558376498e-06, "loss": 1.3961, "step": 910 }, { "epoch": 1.41, "learning_rate": 1.5104412715939142e-06, "loss": 1.3555, "step": 912 }, { "epoch": 1.41, "learning_rate": 1.5082870887655274e-06, "loss": 1.3367, "step": 914 }, { "epoch": 1.42, "learning_rate": 1.5061297208512729e-06, "loss": 1.3621, "step": 916 }, { "epoch": 1.42, "learning_rate": 1.5039691813698928e-06, "loss": 1.2356, "step": 918 }, { "epoch": 1.42, "learning_rate": 1.5018054838600032e-06, "loss": 1.4455, "step": 920 }, { "epoch": 1.42, "learning_rate": 1.4996386418800097e-06, "loss": 1.3938, "step": 922 }, { "epoch": 1.43, "learning_rate": 1.4974686690080216e-06, "loss": 1.339, "step": 924 }, { "epoch": 1.43, "learning_rate": 1.4952955788417675e-06, "loss": 1.3551, "step": 926 }, { "epoch": 1.43, "learning_rate": 1.4931193849985101e-06, "loss": 1.4069, "step": 928 }, { "epoch": 1.44, "learning_rate": 1.4909401011149607e-06, "loss": 1.3863, "step": 930 }, { "epoch": 1.44, "learning_rate": 1.4887577408471937e-06, "loss": 1.422, "step": 932 }, { "epoch": 1.44, "learning_rate": 1.4865723178705607e-06, "loss": 1.3686, "step": 934 }, { "epoch": 1.45, "learning_rate": 1.4843838458796065e-06, "loss": 1.3777, "step": 936 }, { "epoch": 1.45, "learning_rate": 1.4821923385879797e-06, "loss": 1.2875, "step": 938 }, { "epoch": 1.45, "learning_rate": 1.4799978097283518e-06, "loss": 1.3339, "step": 940 }, { "epoch": 1.46, "learning_rate": 1.477800273052326e-06, "loss": 1.367, "step": 942 }, { "epoch": 1.46, "learning_rate": 1.4755997423303547e-06, "loss": 1.3696, "step": 944 }, { "epoch": 1.46, "learning_rate": 1.4733962313516522e-06, "loss": 1.3362, "step": 946 }, { "epoch": 1.46, "learning_rate": 1.4711897539241064e-06, "loss": 1.4226, "step": 948 }, { "epoch": 1.47, "learning_rate": 1.4689803238741953e-06, "loss": 1.3654, "step": 950 }, { "epoch": 1.47, "learning_rate": 1.4667679550468991e-06, "loss": 1.392, "step": 952 }, { "epoch": 1.47, "learning_rate": 1.4645526613056125e-06, "loss": 1.4032, "step": 954 }, { "epoch": 1.48, "learning_rate": 1.462334456532059e-06, "loss": 1.3927, "step": 956 }, { "epoch": 1.48, "learning_rate": 1.460113354626203e-06, "loss": 1.401, "step": 958 }, { "epoch": 1.48, "learning_rate": 1.4578893695061642e-06, "loss": 1.418, "step": 960 }, { "epoch": 1.49, "learning_rate": 1.455662515108129e-06, "loss": 1.4281, "step": 962 }, { "epoch": 1.49, "learning_rate": 1.4534328053862631e-06, "loss": 1.4172, "step": 964 }, { "epoch": 1.49, "learning_rate": 1.4512002543126253e-06, "loss": 1.3257, "step": 966 }, { "epoch": 1.5, "learning_rate": 1.4489648758770787e-06, "loss": 1.3802, "step": 968 }, { "epoch": 1.5, "learning_rate": 1.4467266840872038e-06, "loss": 1.4165, "step": 970 }, { "epoch": 1.5, "learning_rate": 1.4444856929682104e-06, "loss": 1.4026, "step": 972 }, { "epoch": 1.5, "learning_rate": 1.4422419165628495e-06, "loss": 1.4226, "step": 974 }, { "epoch": 1.51, "learning_rate": 1.439995368931326e-06, "loss": 1.4275, "step": 976 }, { "epoch": 1.51, "learning_rate": 1.4377460641512098e-06, "loss": 1.449, "step": 978 }, { "epoch": 1.51, "learning_rate": 1.4354940163173483e-06, "loss": 1.4227, "step": 980 }, { "epoch": 1.52, "learning_rate": 1.4332392395417778e-06, "loss": 1.2831, "step": 982 }, { "epoch": 1.52, "learning_rate": 1.4309817479536343e-06, "loss": 1.3581, "step": 984 }, { "epoch": 1.52, "learning_rate": 1.4287215556990664e-06, "loss": 1.4217, "step": 986 }, { "epoch": 1.53, "learning_rate": 1.4264586769411456e-06, "loss": 1.4527, "step": 988 }, { "epoch": 1.53, "learning_rate": 1.4241931258597778e-06, "loss": 1.3916, "step": 990 }, { "epoch": 1.53, "learning_rate": 1.4219249166516149e-06, "loss": 1.3416, "step": 992 }, { "epoch": 1.54, "learning_rate": 1.4196540635299653e-06, "loss": 1.4022, "step": 994 }, { "epoch": 1.54, "learning_rate": 1.4173805807247039e-06, "loss": 1.3323, "step": 996 }, { "epoch": 1.54, "learning_rate": 1.4151044824821866e-06, "loss": 1.362, "step": 998 }, { "epoch": 1.54, "learning_rate": 1.4128257830651553e-06, "loss": 1.4149, "step": 1000 }, { "epoch": 1.55, "learning_rate": 1.4105444967526543e-06, "loss": 1.4061, "step": 1002 }, { "epoch": 1.55, "learning_rate": 1.4082606378399367e-06, "loss": 1.4255, "step": 1004 }, { "epoch": 1.55, "learning_rate": 1.4059742206383764e-06, "loss": 1.3034, "step": 1006 }, { "epoch": 1.56, "learning_rate": 1.4036852594753795e-06, "loss": 1.3547, "step": 1008 }, { "epoch": 1.56, "learning_rate": 1.4013937686942918e-06, "loss": 1.4458, "step": 1010 }, { "epoch": 1.56, "learning_rate": 1.3990997626543114e-06, "loss": 1.4288, "step": 1012 }, { "epoch": 1.57, "learning_rate": 1.396803255730398e-06, "loss": 1.423, "step": 1014 }, { "epoch": 1.57, "learning_rate": 1.3945042623131817e-06, "loss": 1.3553, "step": 1016 }, { "epoch": 1.57, "learning_rate": 1.3922027968088745e-06, "loss": 1.3599, "step": 1018 }, { "epoch": 1.58, "learning_rate": 1.389898873639179e-06, "loss": 1.3863, "step": 1020 }, { "epoch": 1.58, "learning_rate": 1.3875925072411982e-06, "loss": 1.3867, "step": 1022 }, { "epoch": 1.58, "learning_rate": 1.385283712067346e-06, "loss": 1.3955, "step": 1024 }, { "epoch": 1.59, "learning_rate": 1.382972502585254e-06, "loss": 1.32, "step": 1026 }, { "epoch": 1.59, "learning_rate": 1.3806588932776845e-06, "loss": 1.3567, "step": 1028 }, { "epoch": 1.59, "learning_rate": 1.3783428986424363e-06, "loss": 1.4056, "step": 1030 }, { "epoch": 1.59, "learning_rate": 1.3760245331922568e-06, "loss": 1.4656, "step": 1032 }, { "epoch": 1.6, "learning_rate": 1.3737038114547485e-06, "loss": 1.3549, "step": 1034 }, { "epoch": 1.6, "learning_rate": 1.3713807479722802e-06, "loss": 1.4253, "step": 1036 }, { "epoch": 1.6, "learning_rate": 1.3690553573018934e-06, "loss": 1.362, "step": 1038 }, { "epoch": 1.61, "learning_rate": 1.3667276540152141e-06, "loss": 1.4008, "step": 1040 }, { "epoch": 1.61, "learning_rate": 1.3643976526983583e-06, "loss": 1.4261, "step": 1042 }, { "epoch": 1.61, "learning_rate": 1.362065367951844e-06, "loss": 1.3193, "step": 1044 }, { "epoch": 1.62, "learning_rate": 1.359730814390496e-06, "loss": 1.5017, "step": 1046 }, { "epoch": 1.62, "learning_rate": 1.3573940066433573e-06, "loss": 1.3437, "step": 1048 }, { "epoch": 1.62, "learning_rate": 1.3550549593535962e-06, "loss": 1.3105, "step": 1050 }, { "epoch": 1.63, "learning_rate": 1.3527136871784145e-06, "loss": 1.3723, "step": 1052 }, { "epoch": 1.63, "learning_rate": 1.350370204788956e-06, "loss": 1.3718, "step": 1054 }, { "epoch": 1.63, "learning_rate": 1.3480245268702142e-06, "loss": 1.3559, "step": 1056 }, { "epoch": 1.63, "learning_rate": 1.3456766681209404e-06, "loss": 1.3925, "step": 1058 }, { "epoch": 1.64, "learning_rate": 1.343326643253552e-06, "loss": 1.3163, "step": 1060 }, { "epoch": 1.64, "learning_rate": 1.3409744669940398e-06, "loss": 1.3007, "step": 1062 }, { "epoch": 1.64, "learning_rate": 1.338620154081876e-06, "loss": 1.3533, "step": 1064 }, { "epoch": 1.65, "learning_rate": 1.3362637192699215e-06, "loss": 1.3055, "step": 1066 }, { "epoch": 1.65, "learning_rate": 1.3339051773243335e-06, "loss": 1.3122, "step": 1068 }, { "epoch": 1.65, "learning_rate": 1.3315445430244744e-06, "loss": 1.3876, "step": 1070 }, { "epoch": 1.66, "learning_rate": 1.3291818311628158e-06, "loss": 1.3817, "step": 1072 }, { "epoch": 1.66, "learning_rate": 1.3268170565448501e-06, "loss": 1.4047, "step": 1074 }, { "epoch": 1.66, "learning_rate": 1.3244502339889944e-06, "loss": 1.2793, "step": 1076 }, { "epoch": 1.67, "learning_rate": 1.3220813783264988e-06, "loss": 1.3293, "step": 1078 }, { "epoch": 1.67, "learning_rate": 1.3197105044013543e-06, "loss": 1.4351, "step": 1080 }, { "epoch": 1.67, "learning_rate": 1.317337627070198e-06, "loss": 1.3743, "step": 1082 }, { "epoch": 1.67, "learning_rate": 1.314962761202221e-06, "loss": 1.3268, "step": 1084 }, { "epoch": 1.68, "learning_rate": 1.312585921679077e-06, "loss": 1.356, "step": 1086 }, { "epoch": 1.68, "learning_rate": 1.3102071233947841e-06, "loss": 1.3401, "step": 1088 }, { "epoch": 1.68, "learning_rate": 1.3078263812556376e-06, "loss": 1.2994, "step": 1090 }, { "epoch": 1.69, "learning_rate": 1.3054437101801114e-06, "loss": 1.3529, "step": 1092 }, { "epoch": 1.69, "learning_rate": 1.3030591250987686e-06, "loss": 1.3394, "step": 1094 }, { "epoch": 1.69, "learning_rate": 1.3006726409541644e-06, "loss": 1.4006, "step": 1096 }, { "epoch": 1.7, "learning_rate": 1.298284272700755e-06, "loss": 1.3451, "step": 1098 }, { "epoch": 1.7, "learning_rate": 1.295894035304803e-06, "loss": 1.3693, "step": 1100 }, { "epoch": 1.7, "learning_rate": 1.2935019437442833e-06, "loss": 1.4804, "step": 1102 }, { "epoch": 1.71, "learning_rate": 1.2911080130087898e-06, "loss": 1.4007, "step": 1104 }, { "epoch": 1.71, "learning_rate": 1.2887122580994415e-06, "loss": 1.3485, "step": 1106 }, { "epoch": 1.71, "learning_rate": 1.2863146940287875e-06, "loss": 1.3399, "step": 1108 }, { "epoch": 1.71, "learning_rate": 1.283915335820714e-06, "loss": 1.3057, "step": 1110 }, { "epoch": 1.72, "learning_rate": 1.2815141985103509e-06, "loss": 1.3, "step": 1112 }, { "epoch": 1.72, "learning_rate": 1.2791112971439744e-06, "loss": 1.3362, "step": 1114 }, { "epoch": 1.72, "learning_rate": 1.2767066467789166e-06, "loss": 1.3818, "step": 1116 }, { "epoch": 1.73, "learning_rate": 1.2743002624834687e-06, "loss": 1.254, "step": 1118 }, { "epoch": 1.73, "learning_rate": 1.2718921593367874e-06, "loss": 1.3373, "step": 1120 }, { "epoch": 1.73, "learning_rate": 1.2694823524287996e-06, "loss": 1.4446, "step": 1122 }, { "epoch": 1.74, "learning_rate": 1.26707085686011e-06, "loss": 1.3048, "step": 1124 }, { "epoch": 1.74, "learning_rate": 1.2646576877419031e-06, "loss": 1.394, "step": 1126 }, { "epoch": 1.74, "learning_rate": 1.2622428601958518e-06, "loss": 1.3731, "step": 1128 }, { "epoch": 1.75, "learning_rate": 1.2598263893540207e-06, "loss": 1.3905, "step": 1130 }, { "epoch": 1.75, "learning_rate": 1.2574082903587723e-06, "loss": 1.4701, "step": 1132 }, { "epoch": 1.75, "learning_rate": 1.25498857836267e-06, "loss": 1.3322, "step": 1134 }, { "epoch": 1.76, "learning_rate": 1.2525672685283871e-06, "loss": 1.3103, "step": 1136 }, { "epoch": 1.76, "learning_rate": 1.2501443760286076e-06, "loss": 1.4229, "step": 1138 }, { "epoch": 1.76, "learning_rate": 1.2477199160459344e-06, "loss": 1.3197, "step": 1140 }, { "epoch": 1.76, "learning_rate": 1.2452939037727911e-06, "loss": 1.4741, "step": 1142 }, { "epoch": 1.77, "learning_rate": 1.2428663544113303e-06, "loss": 1.3786, "step": 1144 }, { "epoch": 1.77, "learning_rate": 1.2404372831733348e-06, "loss": 1.3054, "step": 1146 }, { "epoch": 1.77, "learning_rate": 1.238006705280125e-06, "loss": 1.3435, "step": 1148 }, { "epoch": 1.78, "learning_rate": 1.235574635962462e-06, "loss": 1.3953, "step": 1150 }, { "epoch": 1.78, "learning_rate": 1.2331410904604527e-06, "loss": 1.2835, "step": 1152 }, { "epoch": 1.78, "learning_rate": 1.230706084023455e-06, "loss": 1.4189, "step": 1154 }, { "epoch": 1.79, "learning_rate": 1.2282696319099794e-06, "loss": 1.3468, "step": 1156 }, { "epoch": 1.79, "learning_rate": 1.2258317493875987e-06, "loss": 1.4159, "step": 1158 }, { "epoch": 1.79, "learning_rate": 1.2233924517328455e-06, "loss": 1.4191, "step": 1160 }, { "epoch": 1.8, "learning_rate": 1.2209517542311233e-06, "loss": 1.3973, "step": 1162 }, { "epoch": 1.8, "learning_rate": 1.2185096721766047e-06, "loss": 1.3472, "step": 1164 }, { "epoch": 1.8, "learning_rate": 1.2160662208721404e-06, "loss": 1.3707, "step": 1166 }, { "epoch": 1.8, "learning_rate": 1.2136214156291604e-06, "loss": 1.34, "step": 1168 }, { "epoch": 1.81, "learning_rate": 1.2111752717675786e-06, "loss": 1.3472, "step": 1170 }, { "epoch": 1.81, "learning_rate": 1.208727804615698e-06, "loss": 1.3973, "step": 1172 }, { "epoch": 1.81, "learning_rate": 1.2062790295101131e-06, "loss": 1.3337, "step": 1174 }, { "epoch": 1.82, "learning_rate": 1.203828961795614e-06, "loss": 1.35, "step": 1176 }, { "epoch": 1.82, "learning_rate": 1.2013776168250922e-06, "loss": 1.4096, "step": 1178 }, { "epoch": 1.82, "learning_rate": 1.198925009959441e-06, "loss": 1.3244, "step": 1180 }, { "epoch": 1.83, "learning_rate": 1.1964711565674628e-06, "loss": 1.332, "step": 1182 }, { "epoch": 1.83, "learning_rate": 1.1940160720257701e-06, "loss": 1.3098, "step": 1184 }, { "epoch": 1.83, "learning_rate": 1.1915597717186908e-06, "loss": 1.3608, "step": 1186 }, { "epoch": 1.84, "learning_rate": 1.1891022710381703e-06, "loss": 1.4674, "step": 1188 }, { "epoch": 1.84, "learning_rate": 1.1866435853836772e-06, "loss": 1.3297, "step": 1190 }, { "epoch": 1.84, "learning_rate": 1.1841837301621038e-06, "loss": 1.3161, "step": 1192 }, { "epoch": 1.84, "learning_rate": 1.1817227207876732e-06, "loss": 1.3483, "step": 1194 }, { "epoch": 1.85, "learning_rate": 1.1792605726818393e-06, "loss": 1.4145, "step": 1196 }, { "epoch": 1.85, "learning_rate": 1.176797301273192e-06, "loss": 1.331, "step": 1198 }, { "epoch": 1.85, "learning_rate": 1.1743329219973608e-06, "loss": 1.3765, "step": 1200 }, { "epoch": 1.86, "learning_rate": 1.1718674502969169e-06, "loss": 1.3729, "step": 1202 }, { "epoch": 1.86, "learning_rate": 1.1694009016212772e-06, "loss": 1.3774, "step": 1204 }, { "epoch": 1.86, "learning_rate": 1.1669332914266066e-06, "loss": 1.3093, "step": 1206 }, { "epoch": 1.87, "learning_rate": 1.1644646351757229e-06, "loss": 1.3111, "step": 1208 }, { "epoch": 1.87, "learning_rate": 1.1619949483379979e-06, "loss": 1.3799, "step": 1210 }, { "epoch": 1.87, "learning_rate": 1.1595242463892618e-06, "loss": 1.4424, "step": 1212 }, { "epoch": 1.88, "learning_rate": 1.1570525448117059e-06, "loss": 1.4249, "step": 1214 }, { "epoch": 1.88, "learning_rate": 1.1545798590937852e-06, "loss": 1.3885, "step": 1216 }, { "epoch": 1.88, "learning_rate": 1.1521062047301214e-06, "loss": 1.4057, "step": 1218 }, { "epoch": 1.88, "learning_rate": 1.1496315972214076e-06, "loss": 1.4125, "step": 1220 }, { "epoch": 1.89, "learning_rate": 1.1471560520743066e-06, "loss": 1.4385, "step": 1222 }, { "epoch": 1.89, "learning_rate": 1.1446795848013604e-06, "loss": 1.4181, "step": 1224 }, { "epoch": 1.89, "learning_rate": 1.1422022109208858e-06, "loss": 1.3709, "step": 1226 }, { "epoch": 1.9, "learning_rate": 1.1397239459568834e-06, "loss": 1.4187, "step": 1228 }, { "epoch": 1.9, "learning_rate": 1.1372448054389364e-06, "loss": 1.2999, "step": 1230 }, { "epoch": 1.9, "learning_rate": 1.1347648049021142e-06, "loss": 1.4299, "step": 1232 }, { "epoch": 1.91, "learning_rate": 1.1322839598868758e-06, "loss": 1.4281, "step": 1234 }, { "epoch": 1.91, "learning_rate": 1.1298022859389732e-06, "loss": 1.3667, "step": 1236 }, { "epoch": 1.91, "learning_rate": 1.1273197986093496e-06, "loss": 1.3922, "step": 1238 }, { "epoch": 1.92, "learning_rate": 1.1248365134540486e-06, "loss": 1.3518, "step": 1240 }, { "epoch": 1.92, "learning_rate": 1.1223524460341108e-06, "loss": 1.3279, "step": 1242 }, { "epoch": 1.92, "learning_rate": 1.1198676119154798e-06, "loss": 1.3524, "step": 1244 }, { "epoch": 1.93, "learning_rate": 1.1173820266689033e-06, "loss": 1.4115, "step": 1246 }, { "epoch": 1.93, "learning_rate": 1.1148957058698358e-06, "loss": 1.3661, "step": 1248 }, { "epoch": 1.93, "learning_rate": 1.1124086650983413e-06, "loss": 1.3845, "step": 1250 }, { "epoch": 1.93, "learning_rate": 1.1099209199389948e-06, "loss": 1.3446, "step": 1252 }, { "epoch": 1.94, "learning_rate": 1.1074324859807856e-06, "loss": 1.3697, "step": 1254 }, { "epoch": 1.94, "learning_rate": 1.104943378817019e-06, "loss": 1.3043, "step": 1256 }, { "epoch": 1.94, "learning_rate": 1.1024536140452194e-06, "loss": 1.3462, "step": 1258 }, { "epoch": 1.95, "learning_rate": 1.0999632072670313e-06, "loss": 1.4342, "step": 1260 }, { "epoch": 1.95, "learning_rate": 1.0974721740881223e-06, "loss": 1.3783, "step": 1262 }, { "epoch": 1.95, "learning_rate": 1.0949805301180855e-06, "loss": 1.2289, "step": 1264 }, { "epoch": 1.96, "learning_rate": 1.0924882909703417e-06, "loss": 1.4249, "step": 1266 }, { "epoch": 1.96, "learning_rate": 1.0899954722620398e-06, "loss": 1.4454, "step": 1268 }, { "epoch": 1.96, "learning_rate": 1.0875020896139628e-06, "loss": 1.3535, "step": 1270 }, { "epoch": 1.97, "learning_rate": 1.0850081586504253e-06, "loss": 1.3541, "step": 1272 }, { "epoch": 1.97, "learning_rate": 1.082513694999179e-06, "loss": 1.3616, "step": 1274 }, { "epoch": 1.97, "learning_rate": 1.0800187142913132e-06, "loss": 1.3535, "step": 1276 }, { "epoch": 1.97, "learning_rate": 1.0775232321611574e-06, "loss": 1.3361, "step": 1278 }, { "epoch": 1.98, "learning_rate": 1.0750272642461828e-06, "loss": 1.3421, "step": 1280 }, { "epoch": 1.98, "learning_rate": 1.072530826186906e-06, "loss": 1.3151, "step": 1282 }, { "epoch": 1.98, "learning_rate": 1.070033933626787e-06, "loss": 1.3755, "step": 1284 }, { "epoch": 1.99, "learning_rate": 1.0675366022121372e-06, "loss": 1.4795, "step": 1286 }, { "epoch": 1.99, "learning_rate": 1.0650388475920152e-06, "loss": 1.43, "step": 1288 }, { "epoch": 1.99, "learning_rate": 1.0625406854181328e-06, "loss": 1.2887, "step": 1290 }, { "epoch": 2.0, "learning_rate": 1.0600421313447558e-06, "loss": 1.3876, "step": 1292 }, { "epoch": 2.0, "learning_rate": 1.0575432010286053e-06, "loss": 1.4285, "step": 1294 }, { "epoch": 2.0, "learning_rate": 1.0550439101287603e-06, "loss": 1.3944, "step": 1296 }, { "epoch": 2.01, "learning_rate": 1.0525442743065592e-06, "loss": 1.2691, "step": 1298 }, { "epoch": 2.01, "learning_rate": 1.0500443092255016e-06, "loss": 1.3078, "step": 1300 }, { "epoch": 2.01, "learning_rate": 1.047544030551151e-06, "loss": 1.2873, "step": 1302 }, { "epoch": 2.01, "learning_rate": 1.0450434539510352e-06, "loss": 1.287, "step": 1304 }, { "epoch": 2.02, "learning_rate": 1.0425425950945492e-06, "loss": 1.3309, "step": 1306 }, { "epoch": 2.02, "learning_rate": 1.0400414696528572e-06, "loss": 1.2904, "step": 1308 }, { "epoch": 2.02, "learning_rate": 1.037540093298793e-06, "loss": 1.3287, "step": 1310 }, { "epoch": 2.03, "learning_rate": 1.0350384817067634e-06, "loss": 1.2095, "step": 1312 }, { "epoch": 2.03, "learning_rate": 1.032536650552649e-06, "loss": 1.2528, "step": 1314 }, { "epoch": 2.03, "learning_rate": 1.030034615513706e-06, "loss": 1.3013, "step": 1316 }, { "epoch": 2.04, "learning_rate": 1.0275323922684689e-06, "loss": 1.3909, "step": 1318 }, { "epoch": 2.04, "learning_rate": 1.0250299964966509e-06, "loss": 1.2865, "step": 1320 }, { "epoch": 2.04, "learning_rate": 1.0225274438790468e-06, "loss": 1.3705, "step": 1322 }, { "epoch": 2.05, "learning_rate": 1.0200247500974335e-06, "loss": 1.1833, "step": 1324 }, { "epoch": 2.05, "learning_rate": 1.017521930834473e-06, "loss": 1.243, "step": 1326 }, { "epoch": 2.05, "learning_rate": 1.0150190017736147e-06, "loss": 1.2941, "step": 1328 }, { "epoch": 2.05, "learning_rate": 1.012515978598993e-06, "loss": 1.2335, "step": 1330 }, { "epoch": 2.06, "learning_rate": 1.0100128769953355e-06, "loss": 1.2777, "step": 1332 }, { "epoch": 2.06, "learning_rate": 1.0075097126478593e-06, "loss": 1.2818, "step": 1334 }, { "epoch": 2.06, "learning_rate": 1.0050065012421746e-06, "loss": 1.3015, "step": 1336 }, { "epoch": 2.07, "learning_rate": 1.0025032584641875e-06, "loss": 1.36, "step": 1338 }, { "epoch": 2.07, "learning_rate": 1e-06, "loss": 1.3296, "step": 1340 }, { "epoch": 2.07, "learning_rate": 9.974967415358124e-07, "loss": 1.2917, "step": 1342 }, { "epoch": 2.08, "learning_rate": 9.94993498757825e-07, "loss": 1.3798, "step": 1344 }, { "epoch": 2.08, "learning_rate": 9.924902873521409e-07, "loss": 1.298, "step": 1346 }, { "epoch": 2.08, "learning_rate": 9.899871230046644e-07, "loss": 1.2795, "step": 1348 }, { "epoch": 2.09, "learning_rate": 9.874840214010068e-07, "loss": 1.3845, "step": 1350 }, { "epoch": 2.09, "learning_rate": 9.849809982263854e-07, "loss": 1.1923, "step": 1352 }, { "epoch": 2.09, "learning_rate": 9.824780691655266e-07, "loss": 1.2915, "step": 1354 }, { "epoch": 2.1, "learning_rate": 9.799752499025668e-07, "loss": 1.3265, "step": 1356 }, { "epoch": 2.1, "learning_rate": 9.774725561209535e-07, "loss": 1.24, "step": 1358 }, { "epoch": 2.1, "learning_rate": 9.74970003503349e-07, "loss": 1.2896, "step": 1360 }, { "epoch": 2.1, "learning_rate": 9.72467607731531e-07, "loss": 1.328, "step": 1362 }, { "epoch": 2.11, "learning_rate": 9.699653844862937e-07, "loss": 1.2692, "step": 1364 }, { "epoch": 2.11, "learning_rate": 9.674633494473513e-07, "loss": 1.3959, "step": 1366 }, { "epoch": 2.11, "learning_rate": 9.649615182932367e-07, "loss": 1.3372, "step": 1368 }, { "epoch": 2.12, "learning_rate": 9.62459906701207e-07, "loss": 1.3165, "step": 1370 }, { "epoch": 2.12, "learning_rate": 9.599585303471427e-07, "loss": 1.2525, "step": 1372 }, { "epoch": 2.12, "learning_rate": 9.574574049054507e-07, "loss": 1.272, "step": 1374 }, { "epoch": 2.13, "learning_rate": 9.549565460489647e-07, "loss": 1.2884, "step": 1376 }, { "epoch": 2.13, "learning_rate": 9.524559694488494e-07, "loss": 1.2702, "step": 1378 }, { "epoch": 2.13, "learning_rate": 9.499556907744983e-07, "loss": 1.3222, "step": 1380 }, { "epoch": 2.14, "learning_rate": 9.474557256934409e-07, "loss": 1.1503, "step": 1382 }, { "epoch": 2.14, "learning_rate": 9.449560898712397e-07, "loss": 1.2681, "step": 1384 }, { "epoch": 2.14, "learning_rate": 9.424567989713944e-07, "loss": 1.3875, "step": 1386 }, { "epoch": 2.14, "learning_rate": 9.399578686552442e-07, "loss": 1.2324, "step": 1388 }, { "epoch": 2.15, "learning_rate": 9.374593145818671e-07, "loss": 1.2543, "step": 1390 }, { "epoch": 2.15, "learning_rate": 9.349611524079849e-07, "loss": 1.2623, "step": 1392 }, { "epoch": 2.15, "learning_rate": 9.324633977878628e-07, "loss": 1.3132, "step": 1394 }, { "epoch": 2.16, "learning_rate": 9.299660663732126e-07, "loss": 1.2872, "step": 1396 }, { "epoch": 2.16, "learning_rate": 9.274691738130943e-07, "loss": 1.4077, "step": 1398 }, { "epoch": 2.16, "learning_rate": 9.249727357538171e-07, "loss": 1.3012, "step": 1400 }, { "epoch": 2.17, "learning_rate": 9.224767678388425e-07, "loss": 1.2441, "step": 1402 }, { "epoch": 2.17, "learning_rate": 9.199812857086868e-07, "loss": 1.3194, "step": 1404 }, { "epoch": 2.17, "learning_rate": 9.174863050008208e-07, "loss": 1.2616, "step": 1406 }, { "epoch": 2.18, "learning_rate": 9.149918413495749e-07, "loss": 1.2777, "step": 1408 }, { "epoch": 2.18, "learning_rate": 9.124979103860373e-07, "loss": 1.3831, "step": 1410 }, { "epoch": 2.18, "learning_rate": 9.1000452773796e-07, "loss": 1.2384, "step": 1412 }, { "epoch": 2.18, "learning_rate": 9.075117090296583e-07, "loss": 1.3303, "step": 1414 }, { "epoch": 2.19, "learning_rate": 9.050194698819142e-07, "loss": 1.2988, "step": 1416 }, { "epoch": 2.19, "learning_rate": 9.025278259118778e-07, "loss": 1.268, "step": 1418 }, { "epoch": 2.19, "learning_rate": 9.00036792732969e-07, "loss": 1.3187, "step": 1420 }, { "epoch": 2.2, "learning_rate": 8.975463859547805e-07, "loss": 1.4051, "step": 1422 }, { "epoch": 2.2, "learning_rate": 8.950566211829808e-07, "loss": 1.2737, "step": 1424 }, { "epoch": 2.2, "learning_rate": 8.925675140192145e-07, "loss": 1.3198, "step": 1426 }, { "epoch": 2.21, "learning_rate": 8.900790800610055e-07, "loss": 1.321, "step": 1428 }, { "epoch": 2.21, "learning_rate": 8.875913349016588e-07, "loss": 1.2694, "step": 1430 }, { "epoch": 2.21, "learning_rate": 8.851042941301642e-07, "loss": 1.2251, "step": 1432 }, { "epoch": 2.22, "learning_rate": 8.826179733310968e-07, "loss": 1.2013, "step": 1434 }, { "epoch": 2.22, "learning_rate": 8.801323880845201e-07, "loss": 1.2555, "step": 1436 }, { "epoch": 2.22, "learning_rate": 8.776475539658892e-07, "loss": 1.2456, "step": 1438 }, { "epoch": 2.22, "learning_rate": 8.751634865459516e-07, "loss": 1.3333, "step": 1440 }, { "epoch": 2.23, "learning_rate": 8.726802013906504e-07, "loss": 1.3212, "step": 1442 }, { "epoch": 2.23, "learning_rate": 8.70197714061027e-07, "loss": 1.2905, "step": 1444 }, { "epoch": 2.23, "learning_rate": 8.677160401131239e-07, "loss": 1.2636, "step": 1446 }, { "epoch": 2.24, "learning_rate": 8.652351950978856e-07, "loss": 1.3024, "step": 1448 }, { "epoch": 2.24, "learning_rate": 8.62755194561064e-07, "loss": 1.3297, "step": 1450 }, { "epoch": 2.24, "learning_rate": 8.602760540431166e-07, "loss": 1.2982, "step": 1452 }, { "epoch": 2.25, "learning_rate": 8.577977890791142e-07, "loss": 1.3317, "step": 1454 }, { "epoch": 2.25, "learning_rate": 8.553204151986397e-07, "loss": 1.2503, "step": 1456 }, { "epoch": 2.25, "learning_rate": 8.528439479256931e-07, "loss": 1.2602, "step": 1458 }, { "epoch": 2.26, "learning_rate": 8.503684027785928e-07, "loss": 1.3226, "step": 1460 }, { "epoch": 2.26, "learning_rate": 8.478937952698785e-07, "loss": 1.2821, "step": 1462 }, { "epoch": 2.26, "learning_rate": 8.454201409062148e-07, "loss": 1.2993, "step": 1464 }, { "epoch": 2.26, "learning_rate": 8.429474551882941e-07, "loss": 1.2411, "step": 1466 }, { "epoch": 2.27, "learning_rate": 8.404757536107381e-07, "loss": 1.2353, "step": 1468 }, { "epoch": 2.27, "learning_rate": 8.380050516620024e-07, "loss": 1.3444, "step": 1470 }, { "epoch": 2.27, "learning_rate": 8.355353648242772e-07, "loss": 1.3238, "step": 1472 }, { "epoch": 2.28, "learning_rate": 8.330667085733934e-07, "loss": 1.2901, "step": 1474 }, { "epoch": 2.28, "learning_rate": 8.30599098378723e-07, "loss": 1.3145, "step": 1476 }, { "epoch": 2.28, "learning_rate": 8.281325497030828e-07, "loss": 1.3424, "step": 1478 }, { "epoch": 2.29, "learning_rate": 8.256670780026392e-07, "loss": 1.2561, "step": 1480 }, { "epoch": 2.29, "learning_rate": 8.232026987268081e-07, "loss": 1.2541, "step": 1482 }, { "epoch": 2.29, "learning_rate": 8.207394273181609e-07, "loss": 1.3157, "step": 1484 }, { "epoch": 2.3, "learning_rate": 8.182772792123269e-07, "loss": 1.2702, "step": 1486 }, { "epoch": 2.3, "learning_rate": 8.158162698378962e-07, "loss": 1.2977, "step": 1488 }, { "epoch": 2.3, "learning_rate": 8.133564146163232e-07, "loss": 1.371, "step": 1490 }, { "epoch": 2.31, "learning_rate": 8.108977289618296e-07, "loss": 1.2255, "step": 1492 }, { "epoch": 2.31, "learning_rate": 8.084402282813092e-07, "loss": 1.2896, "step": 1494 }, { "epoch": 2.31, "learning_rate": 8.059839279742298e-07, "loss": 1.2613, "step": 1496 }, { "epoch": 2.31, "learning_rate": 8.03528843432537e-07, "loss": 1.2602, "step": 1498 }, { "epoch": 2.32, "learning_rate": 8.010749900405588e-07, "loss": 1.336, "step": 1500 }, { "epoch": 2.32, "learning_rate": 7.98622383174908e-07, "loss": 1.2883, "step": 1502 }, { "epoch": 2.32, "learning_rate": 7.961710382043861e-07, "loss": 1.2725, "step": 1504 }, { "epoch": 2.33, "learning_rate": 7.937209704898871e-07, "loss": 1.2452, "step": 1506 }, { "epoch": 2.33, "learning_rate": 7.912721953843019e-07, "loss": 1.319, "step": 1508 }, { "epoch": 2.33, "learning_rate": 7.888247282324211e-07, "loss": 1.2352, "step": 1510 }, { "epoch": 2.34, "learning_rate": 7.8637858437084e-07, "loss": 1.2971, "step": 1512 }, { "epoch": 2.34, "learning_rate": 7.839337791278597e-07, "loss": 1.2538, "step": 1514 }, { "epoch": 2.34, "learning_rate": 7.814903278233953e-07, "loss": 1.3021, "step": 1516 }, { "epoch": 2.35, "learning_rate": 7.790482457688768e-07, "loss": 1.3054, "step": 1518 }, { "epoch": 2.35, "learning_rate": 7.766075482671543e-07, "loss": 1.3242, "step": 1520 }, { "epoch": 2.35, "learning_rate": 7.741682506124016e-07, "loss": 1.3379, "step": 1522 }, { "epoch": 2.35, "learning_rate": 7.717303680900205e-07, "loss": 1.2872, "step": 1524 }, { "epoch": 2.36, "learning_rate": 7.692939159765453e-07, "loss": 1.2444, "step": 1526 }, { "epoch": 2.36, "learning_rate": 7.668589095395472e-07, "loss": 1.3124, "step": 1528 }, { "epoch": 2.36, "learning_rate": 7.644253640375381e-07, "loss": 1.338, "step": 1530 }, { "epoch": 2.37, "learning_rate": 7.619932947198754e-07, "loss": 1.289, "step": 1532 }, { "epoch": 2.37, "learning_rate": 7.595627168266653e-07, "loss": 1.2379, "step": 1534 }, { "epoch": 2.37, "learning_rate": 7.571336455886698e-07, "loss": 1.2872, "step": 1536 }, { "epoch": 2.38, "learning_rate": 7.547060962272087e-07, "loss": 1.2432, "step": 1538 }, { "epoch": 2.38, "learning_rate": 7.522800839540655e-07, "loss": 1.2653, "step": 1540 }, { "epoch": 2.38, "learning_rate": 7.498556239713923e-07, "loss": 1.2991, "step": 1542 }, { "epoch": 2.39, "learning_rate": 7.47432731471613e-07, "loss": 1.3241, "step": 1544 }, { "epoch": 2.39, "learning_rate": 7.4501142163733e-07, "loss": 1.3954, "step": 1546 }, { "epoch": 2.39, "learning_rate": 7.42591709641228e-07, "loss": 1.2713, "step": 1548 }, { "epoch": 2.39, "learning_rate": 7.401736106459792e-07, "loss": 1.2817, "step": 1550 }, { "epoch": 2.4, "learning_rate": 7.377571398041484e-07, "loss": 1.2259, "step": 1552 }, { "epoch": 2.4, "learning_rate": 7.353423122580972e-07, "loss": 1.2676, "step": 1554 }, { "epoch": 2.4, "learning_rate": 7.329291431398903e-07, "loss": 1.2814, "step": 1556 }, { "epoch": 2.41, "learning_rate": 7.305176475712004e-07, "loss": 1.3103, "step": 1558 }, { "epoch": 2.41, "learning_rate": 7.281078406632126e-07, "loss": 1.3342, "step": 1560 }, { "epoch": 2.41, "learning_rate": 7.256997375165313e-07, "loss": 1.3209, "step": 1562 }, { "epoch": 2.42, "learning_rate": 7.232933532210834e-07, "loss": 1.2584, "step": 1564 }, { "epoch": 2.42, "learning_rate": 7.208887028560257e-07, "loss": 1.2422, "step": 1566 }, { "epoch": 2.42, "learning_rate": 7.184858014896492e-07, "loss": 1.3057, "step": 1568 }, { "epoch": 2.43, "learning_rate": 7.160846641792857e-07, "loss": 1.3268, "step": 1570 }, { "epoch": 2.43, "learning_rate": 7.136853059712125e-07, "loss": 1.3263, "step": 1572 }, { "epoch": 2.43, "learning_rate": 7.112877419005588e-07, "loss": 1.3594, "step": 1574 }, { "epoch": 2.43, "learning_rate": 7.088919869912101e-07, "loss": 1.2028, "step": 1576 }, { "epoch": 2.44, "learning_rate": 7.064980562557168e-07, "loss": 1.276, "step": 1578 }, { "epoch": 2.44, "learning_rate": 7.04105964695197e-07, "loss": 1.2502, "step": 1580 }, { "epoch": 2.44, "learning_rate": 7.017157272992447e-07, "loss": 1.2704, "step": 1582 }, { "epoch": 2.45, "learning_rate": 6.993273590458358e-07, "loss": 1.3181, "step": 1584 }, { "epoch": 2.45, "learning_rate": 6.969408749012314e-07, "loss": 1.2728, "step": 1586 }, { "epoch": 2.45, "learning_rate": 6.945562898198884e-07, "loss": 1.3771, "step": 1588 }, { "epoch": 2.46, "learning_rate": 6.921736187443623e-07, "loss": 1.288, "step": 1590 }, { "epoch": 2.46, "learning_rate": 6.897928766052157e-07, "loss": 1.2084, "step": 1592 }, { "epoch": 2.46, "learning_rate": 6.874140783209232e-07, "loss": 1.2635, "step": 1594 }, { "epoch": 2.47, "learning_rate": 6.850372387977787e-07, "loss": 1.3313, "step": 1596 }, { "epoch": 2.47, "learning_rate": 6.82662372929802e-07, "loss": 1.2369, "step": 1598 }, { "epoch": 2.47, "learning_rate": 6.802894955986458e-07, "loss": 1.3295, "step": 1600 }, { "epoch": 2.48, "learning_rate": 6.779186216735009e-07, "loss": 1.2502, "step": 1602 }, { "epoch": 2.48, "learning_rate": 6.755497660110059e-07, "loss": 1.2821, "step": 1604 }, { "epoch": 2.48, "learning_rate": 6.731829434551499e-07, "loss": 1.2689, "step": 1606 }, { "epoch": 2.48, "learning_rate": 6.708181688371842e-07, "loss": 1.3866, "step": 1608 }, { "epoch": 2.49, "learning_rate": 6.684554569755258e-07, "loss": 1.3208, "step": 1610 }, { "epoch": 2.49, "learning_rate": 6.660948226756663e-07, "loss": 1.2689, "step": 1612 }, { "epoch": 2.49, "learning_rate": 6.637362807300789e-07, "loss": 1.2565, "step": 1614 }, { "epoch": 2.5, "learning_rate": 6.613798459181242e-07, "loss": 1.2888, "step": 1616 }, { "epoch": 2.5, "learning_rate": 6.590255330059602e-07, "loss": 1.2752, "step": 1618 }, { "epoch": 2.5, "learning_rate": 6.56673356746448e-07, "loss": 1.2917, "step": 1620 }, { "epoch": 2.51, "learning_rate": 6.543233318790596e-07, "loss": 1.3281, "step": 1622 }, { "epoch": 2.51, "learning_rate": 6.519754731297857e-07, "loss": 1.2629, "step": 1624 }, { "epoch": 2.51, "learning_rate": 6.496297952110441e-07, "loss": 1.3338, "step": 1626 }, { "epoch": 2.52, "learning_rate": 6.472863128215856e-07, "loss": 1.3199, "step": 1628 }, { "epoch": 2.52, "learning_rate": 6.449450406464038e-07, "loss": 1.3147, "step": 1630 }, { "epoch": 2.52, "learning_rate": 6.426059933566425e-07, "loss": 1.3007, "step": 1632 }, { "epoch": 2.52, "learning_rate": 6.402691856095039e-07, "loss": 1.2929, "step": 1634 }, { "epoch": 2.53, "learning_rate": 6.379346320481563e-07, "loss": 1.3528, "step": 1636 }, { "epoch": 2.53, "learning_rate": 6.356023473016416e-07, "loss": 1.2652, "step": 1638 }, { "epoch": 2.53, "learning_rate": 6.33272345984786e-07, "loss": 1.2366, "step": 1640 }, { "epoch": 2.54, "learning_rate": 6.309446426981065e-07, "loss": 1.3326, "step": 1642 }, { "epoch": 2.54, "learning_rate": 6.286192520277197e-07, "loss": 1.3241, "step": 1644 }, { "epoch": 2.54, "learning_rate": 6.262961885452516e-07, "loss": 1.169, "step": 1646 }, { "epoch": 2.55, "learning_rate": 6.239754668077431e-07, "loss": 1.2842, "step": 1648 }, { "epoch": 2.55, "learning_rate": 6.216571013575636e-07, "loss": 1.2439, "step": 1650 }, { "epoch": 2.55, "learning_rate": 6.193411067223155e-07, "loss": 1.3393, "step": 1652 }, { "epoch": 2.56, "learning_rate": 6.170274974147458e-07, "loss": 1.2259, "step": 1654 }, { "epoch": 2.56, "learning_rate": 6.147162879326541e-07, "loss": 1.2786, "step": 1656 }, { "epoch": 2.56, "learning_rate": 6.124074927588017e-07, "loss": 1.3283, "step": 1658 }, { "epoch": 2.56, "learning_rate": 6.101011263608209e-07, "loss": 1.3039, "step": 1660 }, { "epoch": 2.57, "learning_rate": 6.077972031911255e-07, "loss": 1.2748, "step": 1662 }, { "epoch": 2.57, "learning_rate": 6.054957376868184e-07, "loss": 1.2994, "step": 1664 }, { "epoch": 2.57, "learning_rate": 6.031967442696022e-07, "loss": 1.3036, "step": 1666 }, { "epoch": 2.58, "learning_rate": 6.009002373456885e-07, "loss": 1.2622, "step": 1668 }, { "epoch": 2.58, "learning_rate": 5.986062313057083e-07, "loss": 1.3724, "step": 1670 }, { "epoch": 2.58, "learning_rate": 5.963147405246206e-07, "loss": 1.2803, "step": 1672 }, { "epoch": 2.59, "learning_rate": 5.940257793616233e-07, "loss": 1.244, "step": 1674 }, { "epoch": 2.59, "learning_rate": 5.917393621600634e-07, "loss": 1.3376, "step": 1676 }, { "epoch": 2.59, "learning_rate": 5.894555032473459e-07, "loss": 1.2783, "step": 1678 }, { "epoch": 2.6, "learning_rate": 5.871742169348447e-07, "loss": 1.3045, "step": 1680 }, { "epoch": 2.6, "learning_rate": 5.848955175178135e-07, "loss": 1.3962, "step": 1682 }, { "epoch": 2.6, "learning_rate": 5.826194192752957e-07, "loss": 1.3172, "step": 1684 }, { "epoch": 2.6, "learning_rate": 5.803459364700347e-07, "loss": 1.2864, "step": 1686 }, { "epoch": 2.61, "learning_rate": 5.780750833483849e-07, "loss": 1.3316, "step": 1688 }, { "epoch": 2.61, "learning_rate": 5.758068741402222e-07, "loss": 1.2866, "step": 1690 }, { "epoch": 2.61, "learning_rate": 5.735413230588544e-07, "loss": 1.2924, "step": 1692 }, { "epoch": 2.62, "learning_rate": 5.712784443009334e-07, "loss": 1.2567, "step": 1694 }, { "epoch": 2.62, "learning_rate": 5.690182520463653e-07, "loss": 1.2812, "step": 1696 }, { "epoch": 2.62, "learning_rate": 5.667607604582226e-07, "loss": 1.2735, "step": 1698 }, { "epoch": 2.63, "learning_rate": 5.645059836826516e-07, "loss": 1.19, "step": 1700 }, { "epoch": 2.63, "learning_rate": 5.622539358487901e-07, "loss": 1.2715, "step": 1702 }, { "epoch": 2.63, "learning_rate": 5.600046310686739e-07, "loss": 1.3521, "step": 1704 }, { "epoch": 2.64, "learning_rate": 5.577580834371505e-07, "loss": 1.3243, "step": 1706 }, { "epoch": 2.64, "learning_rate": 5.555143070317899e-07, "loss": 1.2846, "step": 1708 }, { "epoch": 2.64, "learning_rate": 5.532733159127962e-07, "loss": 1.311, "step": 1710 }, { "epoch": 2.65, "learning_rate": 5.51035124122921e-07, "loss": 1.3173, "step": 1712 }, { "epoch": 2.65, "learning_rate": 5.487997456873747e-07, "loss": 1.2577, "step": 1714 }, { "epoch": 2.65, "learning_rate": 5.465671946137367e-07, "loss": 1.2883, "step": 1716 }, { "epoch": 2.65, "learning_rate": 5.44337484891871e-07, "loss": 1.2709, "step": 1718 }, { "epoch": 2.66, "learning_rate": 5.421106304938355e-07, "loss": 1.2471, "step": 1720 }, { "epoch": 2.66, "learning_rate": 5.398866453737971e-07, "loss": 1.2438, "step": 1722 }, { "epoch": 2.66, "learning_rate": 5.37665543467941e-07, "loss": 1.2812, "step": 1724 }, { "epoch": 2.67, "learning_rate": 5.354473386943872e-07, "loss": 1.2995, "step": 1726 }, { "epoch": 2.67, "learning_rate": 5.332320449531007e-07, "loss": 1.3223, "step": 1728 }, { "epoch": 2.67, "learning_rate": 5.310196761258047e-07, "loss": 1.2864, "step": 1730 }, { "epoch": 2.68, "learning_rate": 5.288102460758936e-07, "loss": 1.3242, "step": 1732 }, { "epoch": 2.68, "learning_rate": 5.266037686483478e-07, "loss": 1.2468, "step": 1734 }, { "epoch": 2.68, "learning_rate": 5.244002576696452e-07, "loss": 1.3184, "step": 1736 }, { "epoch": 2.69, "learning_rate": 5.221997269476741e-07, "loss": 1.2214, "step": 1738 }, { "epoch": 2.69, "learning_rate": 5.200021902716482e-07, "loss": 1.2707, "step": 1740 }, { "epoch": 2.69, "learning_rate": 5.178076614120199e-07, "loss": 1.3467, "step": 1742 }, { "epoch": 2.69, "learning_rate": 5.156161541203938e-07, "loss": 1.3531, "step": 1744 }, { "epoch": 2.7, "learning_rate": 5.13427682129439e-07, "loss": 1.3149, "step": 1746 }, { "epoch": 2.7, "learning_rate": 5.112422591528061e-07, "loss": 1.2814, "step": 1748 }, { "epoch": 2.7, "learning_rate": 5.090598988850392e-07, "loss": 1.2964, "step": 1750 }, { "epoch": 2.71, "learning_rate": 5.068806150014901e-07, "loss": 1.2173, "step": 1752 }, { "epoch": 2.71, "learning_rate": 5.047044211582325e-07, "loss": 1.3234, "step": 1754 }, { "epoch": 2.71, "learning_rate": 5.025313309919782e-07, "loss": 1.299, "step": 1756 }, { "epoch": 2.72, "learning_rate": 5.003613581199903e-07, "loss": 1.3087, "step": 1758 }, { "epoch": 2.72, "learning_rate": 4.981945161399968e-07, "loss": 1.3802, "step": 1760 }, { "epoch": 2.72, "learning_rate": 4.960308186301074e-07, "loss": 1.3113, "step": 1762 }, { "epoch": 2.73, "learning_rate": 4.938702791487271e-07, "loss": 1.3047, "step": 1764 }, { "epoch": 2.73, "learning_rate": 4.917129112344729e-07, "loss": 1.3667, "step": 1766 }, { "epoch": 2.73, "learning_rate": 4.89558728406086e-07, "loss": 1.2877, "step": 1768 }, { "epoch": 2.73, "learning_rate": 4.874077441623503e-07, "loss": 1.2303, "step": 1770 }, { "epoch": 2.74, "learning_rate": 4.852599719820053e-07, "loss": 1.2977, "step": 1772 }, { "epoch": 2.74, "learning_rate": 4.831154253236638e-07, "loss": 1.3207, "step": 1774 }, { "epoch": 2.74, "learning_rate": 4.809741176257251e-07, "loss": 1.2645, "step": 1776 }, { "epoch": 2.75, "learning_rate": 4.788360623062936e-07, "loss": 1.2638, "step": 1778 }, { "epoch": 2.75, "learning_rate": 4.767012727630927e-07, "loss": 1.2733, "step": 1780 }, { "epoch": 2.75, "learning_rate": 4.745697623733811e-07, "loss": 1.2801, "step": 1782 }, { "epoch": 2.76, "learning_rate": 4.724415444938691e-07, "loss": 1.3685, "step": 1784 }, { "epoch": 2.76, "learning_rate": 4.703166324606359e-07, "loss": 1.3424, "step": 1786 }, { "epoch": 2.76, "learning_rate": 4.6819503958904495e-07, "loss": 1.3777, "step": 1788 }, { "epoch": 2.77, "learning_rate": 4.660767791736615e-07, "loss": 1.2755, "step": 1790 }, { "epoch": 2.77, "learning_rate": 4.6396186448816623e-07, "loss": 1.3359, "step": 1792 }, { "epoch": 2.77, "learning_rate": 4.6185030878527676e-07, "loss": 1.2887, "step": 1794 }, { "epoch": 2.77, "learning_rate": 4.597421252966616e-07, "loss": 1.2658, "step": 1796 }, { "epoch": 2.78, "learning_rate": 4.576373272328581e-07, "loss": 1.3478, "step": 1798 }, { "epoch": 2.78, "learning_rate": 4.5553592778318896e-07, "loss": 1.2223, "step": 1800 }, { "epoch": 2.78, "learning_rate": 4.5343794011567993e-07, "loss": 1.3446, "step": 1802 }, { "epoch": 2.79, "learning_rate": 4.5134337737697814e-07, "loss": 1.2158, "step": 1804 }, { "epoch": 2.79, "learning_rate": 4.4925225269226943e-07, "loss": 1.3378, "step": 1806 }, { "epoch": 2.79, "learning_rate": 4.471645791651941e-07, "loss": 1.3065, "step": 1808 }, { "epoch": 2.8, "learning_rate": 4.450803698777683e-07, "loss": 1.3326, "step": 1810 }, { "epoch": 2.8, "learning_rate": 4.429996378902985e-07, "loss": 1.2375, "step": 1812 }, { "epoch": 2.8, "learning_rate": 4.4092239624130276e-07, "loss": 1.2856, "step": 1814 }, { "epoch": 2.81, "learning_rate": 4.388486579474262e-07, "loss": 1.407, "step": 1816 }, { "epoch": 2.81, "learning_rate": 4.3677843600336205e-07, "loss": 1.2768, "step": 1818 }, { "epoch": 2.81, "learning_rate": 4.3471174338176863e-07, "loss": 1.3072, "step": 1820 }, { "epoch": 2.81, "learning_rate": 4.326485930331881e-07, "loss": 1.3862, "step": 1822 }, { "epoch": 2.82, "learning_rate": 4.3058899788596524e-07, "loss": 1.3172, "step": 1824 }, { "epoch": 2.82, "learning_rate": 4.2853297084616803e-07, "loss": 1.263, "step": 1826 }, { "epoch": 2.82, "learning_rate": 4.2648052479750506e-07, "loss": 1.3237, "step": 1828 }, { "epoch": 2.83, "learning_rate": 4.2443167260124455e-07, "loss": 1.3506, "step": 1830 }, { "epoch": 2.83, "learning_rate": 4.223864270961358e-07, "loss": 1.243, "step": 1832 }, { "epoch": 2.83, "learning_rate": 4.2034480109832625e-07, "loss": 1.294, "step": 1834 }, { "epoch": 2.84, "learning_rate": 4.1830680740128356e-07, "loss": 1.2106, "step": 1836 }, { "epoch": 2.84, "learning_rate": 4.162724587757128e-07, "loss": 1.3059, "step": 1838 }, { "epoch": 2.84, "learning_rate": 4.142417679694793e-07, "loss": 1.2669, "step": 1840 }, { "epoch": 2.85, "learning_rate": 4.1221474770752696e-07, "loss": 1.281, "step": 1842 }, { "epoch": 2.85, "learning_rate": 4.101914106917983e-07, "loss": 1.292, "step": 1844 }, { "epoch": 2.85, "learning_rate": 4.0817176960115563e-07, "loss": 1.2152, "step": 1846 }, { "epoch": 2.86, "learning_rate": 4.0615583709130196e-07, "loss": 1.3004, "step": 1848 }, { "epoch": 2.86, "learning_rate": 4.041436257947014e-07, "loss": 1.317, "step": 1850 }, { "epoch": 2.86, "learning_rate": 4.021351483204987e-07, "loss": 1.3438, "step": 1852 }, { "epoch": 2.86, "learning_rate": 4.001304172544415e-07, "loss": 1.3237, "step": 1854 }, { "epoch": 2.87, "learning_rate": 3.981294451588023e-07, "loss": 1.3096, "step": 1856 }, { "epoch": 2.87, "learning_rate": 3.961322445722981e-07, "loss": 1.2552, "step": 1858 }, { "epoch": 2.87, "learning_rate": 3.94138828010012e-07, "loss": 1.2581, "step": 1860 }, { "epoch": 2.88, "learning_rate": 3.9214920796331597e-07, "loss": 1.1844, "step": 1862 }, { "epoch": 2.88, "learning_rate": 3.9016339689979117e-07, "loss": 1.3058, "step": 1864 }, { "epoch": 2.88, "learning_rate": 3.8818140726315116e-07, "loss": 1.2557, "step": 1866 }, { "epoch": 2.89, "learning_rate": 3.8620325147316234e-07, "loss": 1.2973, "step": 1868 }, { "epoch": 2.89, "learning_rate": 3.84228941925568e-07, "loss": 1.2521, "step": 1870 }, { "epoch": 2.89, "learning_rate": 3.822584909920095e-07, "loss": 1.2604, "step": 1872 }, { "epoch": 2.9, "learning_rate": 3.802919110199483e-07, "loss": 1.2632, "step": 1874 }, { "epoch": 2.9, "learning_rate": 3.783292143325891e-07, "loss": 1.3246, "step": 1876 }, { "epoch": 2.9, "learning_rate": 3.763704132288038e-07, "loss": 1.2799, "step": 1878 }, { "epoch": 2.9, "learning_rate": 3.7441551998305255e-07, "loss": 1.328, "step": 1880 }, { "epoch": 2.91, "learning_rate": 3.7246454684530824e-07, "loss": 1.3078, "step": 1882 }, { "epoch": 2.91, "learning_rate": 3.705175060409784e-07, "loss": 1.2281, "step": 1884 }, { "epoch": 2.91, "learning_rate": 3.6857440977082923e-07, "loss": 1.2588, "step": 1886 }, { "epoch": 2.92, "learning_rate": 3.676043446420555e-07, "loss": 1.3075, "step": 1888 }, { "epoch": 2.92, "learning_rate": 3.656671879955249e-07, "loss": 1.3335, "step": 1890 }, { "epoch": 2.92, "learning_rate": 3.6373400627678487e-07, "loss": 1.2353, "step": 1892 }, { "epoch": 2.93, "learning_rate": 3.618048115997574e-07, "loss": 1.2783, "step": 1894 }, { "epoch": 2.93, "learning_rate": 3.598796160533789e-07, "loss": 1.2207, "step": 1896 }, { "epoch": 2.93, "learning_rate": 3.579584317015272e-07, "loss": 1.2557, "step": 1898 }, { "epoch": 2.94, "learning_rate": 3.5604127058294465e-07, "loss": 1.3584, "step": 1900 }, { "epoch": 2.94, "learning_rate": 3.5412814471116203e-07, "loss": 1.2519, "step": 1902 }, { "epoch": 2.94, "learning_rate": 3.522190660744242e-07, "loss": 1.2663, "step": 1904 }, { "epoch": 2.94, "learning_rate": 3.5031404663561504e-07, "loss": 1.288, "step": 1906 }, { "epoch": 2.95, "learning_rate": 3.4841309833218244e-07, "loss": 1.3052, "step": 1908 }, { "epoch": 2.95, "learning_rate": 3.4651623307606346e-07, "loss": 1.2914, "step": 1910 }, { "epoch": 2.95, "learning_rate": 3.446234627536079e-07, "loss": 1.1929, "step": 1912 }, { "epoch": 2.96, "learning_rate": 3.4273479922550705e-07, "loss": 1.1871, "step": 1914 }, { "epoch": 2.96, "learning_rate": 3.4085025432671745e-07, "loss": 1.2929, "step": 1916 }, { "epoch": 2.96, "learning_rate": 3.3896983986638704e-07, "loss": 1.2538, "step": 1918 }, { "epoch": 2.97, "learning_rate": 3.370935676277806e-07, "loss": 1.2194, "step": 1920 }, { "epoch": 2.97, "learning_rate": 3.352214493682065e-07, "loss": 1.2177, "step": 1922 }, { "epoch": 2.97, "learning_rate": 3.333534968189436e-07, "loss": 1.2658, "step": 1924 }, { "epoch": 2.98, "learning_rate": 3.314897216851673e-07, "loss": 1.2524, "step": 1926 }, { "epoch": 2.98, "learning_rate": 3.296301356458745e-07, "loss": 1.2408, "step": 1928 }, { "epoch": 2.98, "learning_rate": 3.2777475035381376e-07, "loss": 1.3456, "step": 1930 }, { "epoch": 2.98, "learning_rate": 3.2592357743540887e-07, "loss": 1.3285, "step": 1932 }, { "epoch": 2.99, "learning_rate": 3.2407662849068907e-07, "loss": 1.3319, "step": 1934 }, { "epoch": 2.99, "learning_rate": 3.222339150932133e-07, "loss": 1.4025, "step": 1936 }, { "epoch": 2.99, "learning_rate": 3.2039544879000046e-07, "loss": 1.2531, "step": 1938 }, { "epoch": 3.0, "learning_rate": 3.1856124110145566e-07, "loss": 1.2561, "step": 1940 }, { "epoch": 3.0, "learning_rate": 3.167313035212976e-07, "loss": 1.3596, "step": 1942 }, { "epoch": 3.0, "learning_rate": 3.149056475164873e-07, "loss": 1.206, "step": 1944 }, { "epoch": 3.01, "learning_rate": 3.1308428452715643e-07, "loss": 1.2705, "step": 1946 }, { "epoch": 3.01, "learning_rate": 3.112672259665354e-07, "loss": 1.2057, "step": 1948 }, { "epoch": 3.01, "learning_rate": 3.0945448322088096e-07, "loss": 1.2028, "step": 1950 }, { "epoch": 3.02, "learning_rate": 3.076460676494067e-07, "loss": 1.1189, "step": 1952 }, { "epoch": 3.02, "learning_rate": 3.0584199058420954e-07, "loss": 1.2521, "step": 1954 }, { "epoch": 3.02, "learning_rate": 3.040422633302011e-07, "loss": 1.2433, "step": 1956 }, { "epoch": 3.03, "learning_rate": 3.022468971650348e-07, "loss": 1.2778, "step": 1958 }, { "epoch": 3.03, "learning_rate": 3.004559033390368e-07, "loss": 1.2586, "step": 1960 }, { "epoch": 3.03, "learning_rate": 2.986692930751349e-07, "loss": 1.1934, "step": 1962 }, { "epoch": 3.03, "learning_rate": 2.968870775687873e-07, "loss": 1.2758, "step": 1964 }, { "epoch": 3.04, "learning_rate": 2.9510926798791357e-07, "loss": 1.2388, "step": 1966 }, { "epoch": 3.04, "learning_rate": 2.933358754728249e-07, "loss": 1.2051, "step": 1968 }, { "epoch": 3.04, "learning_rate": 2.9156691113615374e-07, "loss": 1.2258, "step": 1970 }, { "epoch": 3.05, "learning_rate": 2.898023860627834e-07, "loss": 1.337, "step": 1972 }, { "epoch": 3.05, "learning_rate": 2.8804231130977974e-07, "loss": 1.2559, "step": 1974 }, { "epoch": 3.05, "learning_rate": 2.8628669790632185e-07, "loss": 1.2276, "step": 1976 }, { "epoch": 3.06, "learning_rate": 2.845355568536328e-07, "loss": 1.2712, "step": 1978 }, { "epoch": 3.06, "learning_rate": 2.8278889912490944e-07, "loss": 1.2461, "step": 1980 }, { "epoch": 3.06, "learning_rate": 2.810467356652559e-07, "loss": 1.2039, "step": 1982 }, { "epoch": 3.07, "learning_rate": 2.793090773916128e-07, "loss": 1.2636, "step": 1984 }, { "epoch": 3.07, "learning_rate": 2.7757593519269084e-07, "loss": 1.2969, "step": 1986 }, { "epoch": 3.07, "learning_rate": 2.7584731992890063e-07, "loss": 1.2965, "step": 1988 }, { "epoch": 3.07, "learning_rate": 2.7412324243228615e-07, "loss": 1.2887, "step": 1990 }, { "epoch": 3.08, "learning_rate": 2.724037135064566e-07, "loss": 1.2288, "step": 1992 }, { "epoch": 3.08, "learning_rate": 2.706887439265179e-07, "loss": 1.2549, "step": 1994 }, { "epoch": 3.08, "learning_rate": 2.6897834443900526e-07, "loss": 1.2483, "step": 1996 }, { "epoch": 3.09, "learning_rate": 2.6727252576181713e-07, "loss": 1.2822, "step": 1998 }, { "epoch": 3.09, "learning_rate": 2.6557129858414697e-07, "loss": 1.2106, "step": 2000 }, { "epoch": 3.09, "learning_rate": 2.6387467356641644e-07, "loss": 1.2121, "step": 2002 }, { "epoch": 3.1, "learning_rate": 2.621826613402085e-07, "loss": 1.1564, "step": 2004 }, { "epoch": 3.1, "learning_rate": 2.604952725082005e-07, "loss": 1.2837, "step": 2006 }, { "epoch": 3.1, "learning_rate": 2.5881251764409883e-07, "loss": 1.2022, "step": 2008 }, { "epoch": 3.11, "learning_rate": 2.571344072925723e-07, "loss": 1.2317, "step": 2010 }, { "epoch": 3.11, "learning_rate": 2.5546095196918425e-07, "loss": 1.2276, "step": 2012 }, { "epoch": 3.11, "learning_rate": 2.5379216216033005e-07, "loss": 1.2733, "step": 2014 }, { "epoch": 3.11, "learning_rate": 2.5212804832316783e-07, "loss": 1.3676, "step": 2016 }, { "epoch": 3.12, "learning_rate": 2.5046862088555586e-07, "loss": 1.2543, "step": 2018 }, { "epoch": 3.12, "learning_rate": 2.48813890245985e-07, "loss": 1.2282, "step": 2020 }, { "epoch": 3.12, "learning_rate": 2.471638667735153e-07, "loss": 1.2103, "step": 2022 }, { "epoch": 3.13, "learning_rate": 2.4551856080770993e-07, "loss": 1.2187, "step": 2024 }, { "epoch": 3.13, "learning_rate": 2.4387798265857073e-07, "loss": 1.2364, "step": 2026 }, { "epoch": 3.13, "learning_rate": 2.422421426064728e-07, "loss": 1.3343, "step": 2028 }, { "epoch": 3.14, "learning_rate": 2.406110509021021e-07, "loss": 1.3088, "step": 2030 }, { "epoch": 3.14, "learning_rate": 2.3898471776638964e-07, "loss": 1.1721, "step": 2032 }, { "epoch": 3.14, "learning_rate": 2.37363153390447e-07, "loss": 1.2153, "step": 2034 }, { "epoch": 3.15, "learning_rate": 2.3574636793550372e-07, "loss": 1.2624, "step": 2036 }, { "epoch": 3.15, "learning_rate": 2.341343715328431e-07, "loss": 1.1111, "step": 2038 }, { "epoch": 3.15, "learning_rate": 2.3252717428373925e-07, "loss": 1.1653, "step": 2040 }, { "epoch": 3.15, "learning_rate": 2.3092478625939217e-07, "loss": 1.3015, "step": 2042 }, { "epoch": 3.16, "learning_rate": 2.2932721750086704e-07, "loss": 1.2304, "step": 2044 }, { "epoch": 3.16, "learning_rate": 2.2773447801902858e-07, "loss": 1.1991, "step": 2046 }, { "epoch": 3.16, "learning_rate": 2.2614657779448122e-07, "loss": 1.2033, "step": 2048 }, { "epoch": 3.17, "learning_rate": 2.2456352677750366e-07, "loss": 1.1552, "step": 2050 }, { "epoch": 3.17, "learning_rate": 2.2298533488798932e-07, "loss": 1.2217, "step": 2052 }, { "epoch": 3.17, "learning_rate": 2.2141201201538205e-07, "loss": 1.2124, "step": 2054 }, { "epoch": 3.18, "learning_rate": 2.1984356801861505e-07, "loss": 1.2279, "step": 2056 }, { "epoch": 3.18, "learning_rate": 2.1828001272604846e-07, "loss": 1.203, "step": 2058 }, { "epoch": 3.18, "learning_rate": 2.1672135593540918e-07, "loss": 1.2635, "step": 2060 }, { "epoch": 3.19, "learning_rate": 2.1516760741372807e-07, "loss": 1.2939, "step": 2062 }, { "epoch": 3.19, "learning_rate": 2.136187768972796e-07, "loss": 1.2374, "step": 2064 }, { "epoch": 3.19, "learning_rate": 2.120748740915198e-07, "loss": 1.2229, "step": 2066 }, { "epoch": 3.2, "learning_rate": 2.1053590867102645e-07, "loss": 1.2336, "step": 2068 }, { "epoch": 3.2, "learning_rate": 2.0900189027943827e-07, "loss": 1.2652, "step": 2070 }, { "epoch": 3.2, "learning_rate": 2.0747282852939475e-07, "loss": 1.3194, "step": 2072 }, { "epoch": 3.2, "learning_rate": 2.059487330024743e-07, "loss": 1.1662, "step": 2074 }, { "epoch": 3.21, "learning_rate": 2.0442961324913688e-07, "loss": 1.3132, "step": 2076 }, { "epoch": 3.21, "learning_rate": 2.029154787886612e-07, "loss": 1.2005, "step": 2078 }, { "epoch": 3.21, "learning_rate": 2.014063391090881e-07, "loss": 1.1702, "step": 2080 }, { "epoch": 3.22, "learning_rate": 1.9990220366715826e-07, "loss": 1.1917, "step": 2082 }, { "epoch": 3.22, "learning_rate": 1.9840308188825495e-07, "loss": 1.2282, "step": 2084 }, { "epoch": 3.22, "learning_rate": 1.969089831663443e-07, "loss": 1.2071, "step": 2086 }, { "epoch": 3.23, "learning_rate": 1.954199168639161e-07, "loss": 1.2024, "step": 2088 }, { "epoch": 3.23, "learning_rate": 1.939358923119252e-07, "loss": 1.225, "step": 2090 }, { "epoch": 3.23, "learning_rate": 1.924569188097338e-07, "loss": 1.2441, "step": 2092 }, { "epoch": 3.24, "learning_rate": 1.9098300562505264e-07, "loss": 1.2748, "step": 2094 }, { "epoch": 3.24, "learning_rate": 1.8951416199388248e-07, "loss": 1.156, "step": 2096 }, { "epoch": 3.24, "learning_rate": 1.8805039712045677e-07, "loss": 1.2445, "step": 2098 }, { "epoch": 3.24, "learning_rate": 1.8659172017718404e-07, "loss": 1.2821, "step": 2100 }, { "epoch": 3.25, "learning_rate": 1.8513814030459062e-07, "loss": 1.2656, "step": 2102 }, { "epoch": 3.25, "learning_rate": 1.8368966661126218e-07, "loss": 1.2696, "step": 2104 }, { "epoch": 3.25, "learning_rate": 1.8224630817378827e-07, "loss": 1.2518, "step": 2106 }, { "epoch": 3.26, "learning_rate": 1.8080807403670373e-07, "loss": 1.2429, "step": 2108 }, { "epoch": 3.26, "learning_rate": 1.793749732124341e-07, "loss": 1.2765, "step": 2110 }, { "epoch": 3.26, "learning_rate": 1.7794701468123686e-07, "loss": 1.2524, "step": 2112 }, { "epoch": 3.27, "learning_rate": 1.7652420739114693e-07, "loss": 1.2642, "step": 2114 }, { "epoch": 3.27, "learning_rate": 1.7510656025792002e-07, "loss": 1.2246, "step": 2116 }, { "epoch": 3.27, "learning_rate": 1.7369408216497626e-07, "loss": 1.1556, "step": 2118 }, { "epoch": 3.28, "learning_rate": 1.72286781963345e-07, "loss": 1.2524, "step": 2120 }, { "epoch": 3.28, "learning_rate": 1.7088466847160944e-07, "loss": 1.2412, "step": 2122 }, { "epoch": 3.28, "learning_rate": 1.6948775047585163e-07, "loss": 1.2411, "step": 2124 }, { "epoch": 3.28, "learning_rate": 1.6809603672959617e-07, "loss": 1.2694, "step": 2126 }, { "epoch": 3.29, "learning_rate": 1.6670953595375703e-07, "loss": 1.1853, "step": 2128 }, { "epoch": 3.29, "learning_rate": 1.6532825683658126e-07, "loss": 1.1995, "step": 2130 }, { "epoch": 3.29, "learning_rate": 1.6395220803359645e-07, "loss": 1.2435, "step": 2132 }, { "epoch": 3.3, "learning_rate": 1.62581398167554e-07, "loss": 1.1665, "step": 2134 }, { "epoch": 3.3, "learning_rate": 1.6121583582837772e-07, "loss": 1.2665, "step": 2136 }, { "epoch": 3.3, "learning_rate": 1.5985552957310843e-07, "loss": 1.2008, "step": 2138 }, { "epoch": 3.31, "learning_rate": 1.5850048792585046e-07, "loss": 1.203, "step": 2140 }, { "epoch": 3.31, "learning_rate": 1.571507193777183e-07, "loss": 1.3006, "step": 2142 }, { "epoch": 3.31, "learning_rate": 1.5580623238678392e-07, "loss": 1.2363, "step": 2144 }, { "epoch": 3.32, "learning_rate": 1.5446703537802342e-07, "loss": 1.1915, "step": 2146 }, { "epoch": 3.32, "learning_rate": 1.5313313674326444e-07, "loss": 1.1936, "step": 2148 }, { "epoch": 3.32, "learning_rate": 1.5180454484113227e-07, "loss": 1.2466, "step": 2150 }, { "epoch": 3.32, "learning_rate": 1.5048126799699968e-07, "loss": 1.2654, "step": 2152 }, { "epoch": 3.33, "learning_rate": 1.4916331450293317e-07, "loss": 1.275, "step": 2154 }, { "epoch": 3.33, "learning_rate": 1.4785069261764183e-07, "loss": 1.2491, "step": 2156 }, { "epoch": 3.33, "learning_rate": 1.465434105664246e-07, "loss": 1.2955, "step": 2158 }, { "epoch": 3.34, "learning_rate": 1.4524147654111962e-07, "loss": 1.2143, "step": 2160 }, { "epoch": 3.34, "learning_rate": 1.4394489870005278e-07, "loss": 1.2623, "step": 2162 }, { "epoch": 3.34, "learning_rate": 1.426536851679866e-07, "loss": 1.2493, "step": 2164 }, { "epoch": 3.35, "learning_rate": 1.4136784403606837e-07, "loss": 1.1757, "step": 2166 }, { "epoch": 3.35, "learning_rate": 1.400873833617814e-07, "loss": 1.3191, "step": 2168 }, { "epoch": 3.35, "learning_rate": 1.3881231116889213e-07, "loss": 1.2315, "step": 2170 }, { "epoch": 3.36, "learning_rate": 1.375426354474023e-07, "loss": 1.1969, "step": 2172 }, { "epoch": 3.36, "learning_rate": 1.3627836415349647e-07, "loss": 1.2299, "step": 2174 }, { "epoch": 3.36, "learning_rate": 1.3501950520949433e-07, "loss": 1.1544, "step": 2176 }, { "epoch": 3.37, "learning_rate": 1.3376606650379985e-07, "loss": 1.2282, "step": 2178 }, { "epoch": 3.37, "learning_rate": 1.3251805589085185e-07, "loss": 1.1187, "step": 2180 }, { "epoch": 3.37, "learning_rate": 1.31275481191075e-07, "loss": 1.2209, "step": 2182 }, { "epoch": 3.37, "learning_rate": 1.3003835019083098e-07, "loss": 1.221, "step": 2184 }, { "epoch": 3.38, "learning_rate": 1.2880667064237006e-07, "loss": 1.2704, "step": 2186 }, { "epoch": 3.38, "learning_rate": 1.2758045026378094e-07, "loss": 1.2907, "step": 2188 }, { "epoch": 3.38, "learning_rate": 1.263596967389441e-07, "loss": 1.2621, "step": 2190 }, { "epoch": 3.39, "learning_rate": 1.2514441771748307e-07, "loss": 1.2384, "step": 2192 }, { "epoch": 3.39, "learning_rate": 1.239346208147164e-07, "loss": 1.2919, "step": 2194 }, { "epoch": 3.39, "learning_rate": 1.2273031361160957e-07, "loss": 1.2667, "step": 2196 }, { "epoch": 3.4, "learning_rate": 1.215315036547282e-07, "loss": 1.2127, "step": 2198 }, { "epoch": 3.4, "learning_rate": 1.2033819845619085e-07, "loss": 1.2614, "step": 2200 }, { "epoch": 3.4, "learning_rate": 1.1915040549362066e-07, "loss": 1.313, "step": 2202 }, { "epoch": 3.41, "learning_rate": 1.1796813221009983e-07, "loss": 1.1931, "step": 2204 }, { "epoch": 3.41, "learning_rate": 1.1679138601412253e-07, "loss": 1.1068, "step": 2206 }, { "epoch": 3.41, "learning_rate": 1.1562017427954873e-07, "loss": 1.2006, "step": 2208 }, { "epoch": 3.41, "learning_rate": 1.1445450434555736e-07, "loss": 1.2608, "step": 2210 }, { "epoch": 3.42, "learning_rate": 1.1329438351660048e-07, "loss": 1.213, "step": 2212 }, { "epoch": 3.42, "learning_rate": 1.1213981906235825e-07, "loss": 1.1575, "step": 2214 }, { "epoch": 3.42, "learning_rate": 1.1099081821769297e-07, "loss": 1.2441, "step": 2216 }, { "epoch": 3.43, "learning_rate": 1.0984738818260264e-07, "loss": 1.1996, "step": 2218 }, { "epoch": 3.43, "learning_rate": 1.0870953612217815e-07, "loss": 1.2542, "step": 2220 }, { "epoch": 3.43, "learning_rate": 1.0757726916655574e-07, "loss": 1.2398, "step": 2222 }, { "epoch": 3.44, "learning_rate": 1.0645059441087479e-07, "loss": 1.1478, "step": 2224 }, { "epoch": 3.44, "learning_rate": 1.0532951891523123e-07, "loss": 1.2675, "step": 2226 }, { "epoch": 3.44, "learning_rate": 1.042140497046351e-07, "loss": 1.2885, "step": 2228 }, { "epoch": 3.45, "learning_rate": 1.031041937689654e-07, "loss": 1.2146, "step": 2230 }, { "epoch": 3.45, "learning_rate": 1.0199995806292639e-07, "loss": 1.2069, "step": 2232 }, { "epoch": 3.45, "learning_rate": 1.0090134950600404e-07, "loss": 1.2761, "step": 2234 }, { "epoch": 3.45, "learning_rate": 9.980837498242356e-08, "loss": 1.2093, "step": 2236 }, { "epoch": 3.46, "learning_rate": 9.872104134110492e-08, "loss": 1.2282, "step": 2238 }, { "epoch": 3.46, "learning_rate": 9.76393553956214e-08, "loss": 1.2213, "step": 2240 }, { "epoch": 3.46, "learning_rate": 9.656332392415457e-08, "loss": 1.2209, "step": 2242 }, { "epoch": 3.47, "learning_rate": 9.549295366945454e-08, "loss": 1.2807, "step": 2244 }, { "epoch": 3.47, "learning_rate": 9.442825133879607e-08, "loss": 1.2876, "step": 2246 }, { "epoch": 3.47, "learning_rate": 9.336922360393717e-08, "loss": 1.2574, "step": 2248 }, { "epoch": 3.48, "learning_rate": 9.231587710107669e-08, "loss": 1.2931, "step": 2250 }, { "epoch": 3.48, "learning_rate": 9.1268218430813e-08, "loss": 1.2222, "step": 2252 }, { "epoch": 3.48, "learning_rate": 9.022625415810325e-08, "loss": 1.1666, "step": 2254 }, { "epoch": 3.49, "learning_rate": 8.918999081222156e-08, "loss": 1.2419, "step": 2256 }, { "epoch": 3.49, "learning_rate": 8.815943488671785e-08, "loss": 1.2463, "step": 2258 }, { "epoch": 3.49, "learning_rate": 8.713459283937796e-08, "loss": 1.2103, "step": 2260 }, { "epoch": 3.49, "learning_rate": 8.611547109218298e-08, "loss": 1.2491, "step": 2262 }, { "epoch": 3.5, "learning_rate": 8.510207603126818e-08, "loss": 1.375, "step": 2264 }, { "epoch": 3.5, "learning_rate": 8.4094414006884e-08, "loss": 1.2675, "step": 2266 }, { "epoch": 3.5, "learning_rate": 8.309249133335571e-08, "loss": 1.2348, "step": 2268 }, { "epoch": 3.51, "learning_rate": 8.209631428904462e-08, "loss": 1.1853, "step": 2270 }, { "epoch": 3.51, "learning_rate": 8.110588911630733e-08, "loss": 1.2002, "step": 2272 }, { "epoch": 3.51, "learning_rate": 8.012122202145755e-08, "loss": 1.2445, "step": 2274 }, { "epoch": 3.52, "learning_rate": 7.914231917472747e-08, "loss": 1.294, "step": 2276 }, { "epoch": 3.52, "learning_rate": 7.816918671022865e-08, "loss": 1.2664, "step": 2278 }, { "epoch": 3.52, "learning_rate": 7.720183072591302e-08, "loss": 1.3054, "step": 2280 }, { "epoch": 3.53, "learning_rate": 7.624025728353622e-08, "loss": 1.2253, "step": 2282 }, { "epoch": 3.53, "learning_rate": 7.528447240861779e-08, "loss": 1.2355, "step": 2284 }, { "epoch": 3.53, "learning_rate": 7.433448209040494e-08, "loss": 1.2828, "step": 2286 }, { "epoch": 3.53, "learning_rate": 7.33902922818339e-08, "loss": 1.2577, "step": 2288 }, { "epoch": 3.54, "learning_rate": 7.245190889949338e-08, "loss": 1.2012, "step": 2290 }, { "epoch": 3.54, "learning_rate": 7.151933782358743e-08, "loss": 1.273, "step": 2292 }, { "epoch": 3.54, "learning_rate": 7.059258489789777e-08, "loss": 1.1239, "step": 2294 }, { "epoch": 3.55, "learning_rate": 6.967165592974788e-08, "loss": 1.2897, "step": 2296 }, { "epoch": 3.55, "learning_rate": 6.875655668996705e-08, "loss": 1.1857, "step": 2298 }, { "epoch": 3.55, "learning_rate": 6.78472929128534e-08, "loss": 1.2372, "step": 2300 }, { "epoch": 3.56, "learning_rate": 6.694387029613802e-08, "loss": 1.2625, "step": 2302 }, { "epoch": 3.56, "learning_rate": 6.604629450094922e-08, "loss": 1.2234, "step": 2304 }, { "epoch": 3.56, "learning_rate": 6.515457115177803e-08, "loss": 1.1805, "step": 2306 }, { "epoch": 3.57, "learning_rate": 6.426870583644194e-08, "loss": 1.2279, "step": 2308 }, { "epoch": 3.57, "learning_rate": 6.338870410604969e-08, "loss": 1.2897, "step": 2310 }, { "epoch": 3.57, "learning_rate": 6.251457147496775e-08, "loss": 1.3317, "step": 2312 }, { "epoch": 3.58, "learning_rate": 6.164631342078419e-08, "loss": 1.2426, "step": 2314 }, { "epoch": 3.58, "learning_rate": 6.078393538427573e-08, "loss": 1.2329, "step": 2316 }, { "epoch": 3.58, "learning_rate": 5.992744276937312e-08, "loss": 1.2288, "step": 2318 }, { "epoch": 3.58, "learning_rate": 5.907684094312637e-08, "loss": 1.2772, "step": 2320 }, { "epoch": 3.59, "learning_rate": 5.823213523567305e-08, "loss": 1.2723, "step": 2322 }, { "epoch": 3.59, "learning_rate": 5.739333094020315e-08, "loss": 1.2579, "step": 2324 }, { "epoch": 3.59, "learning_rate": 5.6560433312926814e-08, "loss": 1.1759, "step": 2326 }, { "epoch": 3.6, "learning_rate": 5.573344757304077e-08, "loss": 1.1667, "step": 2328 }, { "epoch": 3.6, "learning_rate": 5.491237890269651e-08, "loss": 1.1735, "step": 2330 }, { "epoch": 3.6, "learning_rate": 5.4097232446967175e-08, "loss": 1.2583, "step": 2332 }, { "epoch": 3.61, "learning_rate": 5.328801331381549e-08, "loss": 1.3096, "step": 2334 }, { "epoch": 3.61, "learning_rate": 5.248472657406122e-08, "loss": 1.2832, "step": 2336 }, { "epoch": 3.61, "learning_rate": 5.168737726135053e-08, "loss": 1.2734, "step": 2338 }, { "epoch": 3.62, "learning_rate": 5.08959703721239e-08, "loss": 1.3077, "step": 2340 }, { "epoch": 3.62, "learning_rate": 5.0110510865583935e-08, "loss": 1.1907, "step": 2342 }, { "epoch": 3.62, "learning_rate": 4.933100366366583e-08, "loss": 1.2428, "step": 2344 }, { "epoch": 3.62, "learning_rate": 4.855745365100539e-08, "loss": 1.2412, "step": 2346 }, { "epoch": 3.63, "learning_rate": 4.7789865674909056e-08, "loss": 1.3106, "step": 2348 }, { "epoch": 3.63, "learning_rate": 4.702824454532295e-08, "loss": 1.1842, "step": 2350 }, { "epoch": 3.63, "learning_rate": 4.6272595034803294e-08, "loss": 1.2733, "step": 2352 }, { "epoch": 3.64, "learning_rate": 4.552292187848661e-08, "loss": 1.2256, "step": 2354 }, { "epoch": 3.64, "learning_rate": 4.4779229774059124e-08, "loss": 1.2803, "step": 2356 }, { "epoch": 3.64, "learning_rate": 4.404152338172795e-08, "loss": 1.1866, "step": 2358 }, { "epoch": 3.65, "learning_rate": 4.33098073241922e-08, "loss": 1.2153, "step": 2360 }, { "epoch": 3.65, "learning_rate": 4.25840861866138e-08, "loss": 1.2367, "step": 2362 }, { "epoch": 3.65, "learning_rate": 4.186436451658804e-08, "loss": 1.2001, "step": 2364 }, { "epoch": 3.66, "learning_rate": 4.115064682411606e-08, "loss": 1.2986, "step": 2366 }, { "epoch": 3.66, "learning_rate": 4.044293758157624e-08, "loss": 1.2348, "step": 2368 }, { "epoch": 3.66, "learning_rate": 3.974124122369604e-08, "loss": 1.2283, "step": 2370 }, { "epoch": 3.66, "learning_rate": 3.904556214752419e-08, "loss": 1.2122, "step": 2372 }, { "epoch": 3.67, "learning_rate": 3.8355904712403465e-08, "loss": 1.1923, "step": 2374 }, { "epoch": 3.67, "learning_rate": 3.767227323994293e-08, "loss": 1.2409, "step": 2376 }, { "epoch": 3.67, "learning_rate": 3.6994672013991425e-08, "loss": 1.3155, "step": 2378 }, { "epoch": 3.68, "learning_rate": 3.632310528060978e-08, "loss": 1.3095, "step": 2380 }, { "epoch": 3.68, "learning_rate": 3.565757724804552e-08, "loss": 1.2636, "step": 2382 }, { "epoch": 3.68, "learning_rate": 3.4998092086705234e-08, "loss": 1.2976, "step": 2384 }, { "epoch": 3.69, "learning_rate": 3.434465392912955e-08, "loss": 1.2326, "step": 2386 }, { "epoch": 3.69, "learning_rate": 3.369726686996588e-08, "loss": 1.2291, "step": 2388 }, { "epoch": 3.69, "learning_rate": 3.3055934965944164e-08, "loss": 1.2691, "step": 2390 }, { "epoch": 3.7, "learning_rate": 3.2420662235850714e-08, "loss": 1.2873, "step": 2392 }, { "epoch": 3.7, "learning_rate": 3.179145266050332e-08, "loss": 1.2119, "step": 2394 }, { "epoch": 3.7, "learning_rate": 3.116831018272581e-08, "loss": 1.2434, "step": 2396 }, { "epoch": 3.7, "learning_rate": 3.055123870732368e-08, "loss": 1.2346, "step": 2398 }, { "epoch": 3.71, "learning_rate": 2.9940242101059944e-08, "loss": 1.3219, "step": 2400 }, { "epoch": 3.71, "learning_rate": 2.933532419263063e-08, "loss": 1.2282, "step": 2402 }, { "epoch": 3.71, "learning_rate": 2.873648877264012e-08, "loss": 1.2417, "step": 2404 }, { "epoch": 3.72, "learning_rate": 2.8143739593578852e-08, "loss": 1.2608, "step": 2406 }, { "epoch": 3.72, "learning_rate": 2.7557080369798202e-08, "loss": 1.2054, "step": 2408 }, { "epoch": 3.72, "learning_rate": 2.6976514777488744e-08, "loss": 1.244, "step": 2410 }, { "epoch": 3.73, "learning_rate": 2.6402046454655713e-08, "loss": 1.2891, "step": 2412 }, { "epoch": 3.73, "learning_rate": 2.5833679001097343e-08, "loss": 1.2564, "step": 2414 }, { "epoch": 3.73, "learning_rate": 2.5271415978382116e-08, "loss": 1.2764, "step": 2416 }, { "epoch": 3.74, "learning_rate": 2.4715260909825786e-08, "loss": 1.2952, "step": 2418 }, { "epoch": 3.74, "learning_rate": 2.416521728046983e-08, "loss": 1.1775, "step": 2420 }, { "epoch": 3.74, "learning_rate": 2.3621288537059804e-08, "loss": 1.2392, "step": 2422 }, { "epoch": 3.75, "learning_rate": 2.3083478088023355e-08, "loss": 1.231, "step": 2424 }, { "epoch": 3.75, "learning_rate": 2.2551789303449032e-08, "loss": 1.2146, "step": 2426 }, { "epoch": 3.75, "learning_rate": 2.2026225515064834e-08, "loss": 1.2824, "step": 2428 }, { "epoch": 3.75, "learning_rate": 2.1506790016217913e-08, "loss": 1.2423, "step": 2430 }, { "epoch": 3.76, "learning_rate": 2.0993486061853583e-08, "loss": 1.2643, "step": 2432 }, { "epoch": 3.76, "learning_rate": 2.0486316868495002e-08, "loss": 1.2348, "step": 2434 }, { "epoch": 3.76, "learning_rate": 1.998528561422297e-08, "loss": 1.3133, "step": 2436 }, { "epoch": 3.77, "learning_rate": 1.9490395438655826e-08, "loss": 1.2554, "step": 2438 }, { "epoch": 3.77, "learning_rate": 1.900164944293048e-08, "loss": 1.2948, "step": 2440 }, { "epoch": 3.77, "learning_rate": 1.8519050689681957e-08, "loss": 1.2102, "step": 2442 }, { "epoch": 3.78, "learning_rate": 1.8042602203025116e-08, "loss": 1.2252, "step": 2444 }, { "epoch": 3.78, "learning_rate": 1.7572306968535176e-08, "loss": 1.3008, "step": 2446 }, { "epoch": 3.78, "learning_rate": 1.7108167933229224e-08, "loss": 1.2677, "step": 2448 }, { "epoch": 3.79, "learning_rate": 1.6650188005547405e-08, "loss": 1.2357, "step": 2450 }, { "epoch": 3.79, "learning_rate": 1.6198370055335196e-08, "loss": 1.0909, "step": 2452 }, { "epoch": 3.79, "learning_rate": 1.5752716913825513e-08, "loss": 1.2383, "step": 2454 }, { "epoch": 3.79, "learning_rate": 1.531323137361995e-08, "loss": 1.2518, "step": 2456 }, { "epoch": 3.8, "learning_rate": 1.487991618867257e-08, "loss": 1.2157, "step": 2458 }, { "epoch": 3.8, "learning_rate": 1.4452774074271812e-08, "loss": 1.2126, "step": 2460 }, { "epoch": 3.8, "learning_rate": 1.4031807707023724e-08, "loss": 1.2401, "step": 2462 }, { "epoch": 3.81, "learning_rate": 1.3617019724835089e-08, "loss": 1.1938, "step": 2464 }, { "epoch": 3.81, "learning_rate": 1.3208412726897322e-08, "loss": 1.2957, "step": 2466 }, { "epoch": 3.81, "learning_rate": 1.2805989273669604e-08, "loss": 1.2479, "step": 2468 }, { "epoch": 3.82, "learning_rate": 1.2409751886863217e-08, "loss": 1.2762, "step": 2470 }, { "epoch": 3.82, "learning_rate": 1.2019703049425455e-08, "loss": 1.2535, "step": 2472 }, { "epoch": 3.82, "learning_rate": 1.1635845205524297e-08, "loss": 1.2131, "step": 2474 }, { "epoch": 3.83, "learning_rate": 1.1258180760533087e-08, "loss": 1.2486, "step": 2476 }, { "epoch": 3.83, "learning_rate": 1.0886712081015548e-08, "loss": 1.2606, "step": 2478 }, { "epoch": 3.83, "learning_rate": 1.0521441494710348e-08, "loss": 1.1539, "step": 2480 }, { "epoch": 3.83, "learning_rate": 1.0162371290517224e-08, "loss": 1.1864, "step": 2482 }, { "epoch": 3.84, "learning_rate": 9.809503718482436e-09, "loss": 1.2244, "step": 2484 }, { "epoch": 3.84, "learning_rate": 9.462840989784671e-09, "loss": 1.1421, "step": 2486 }, { "epoch": 3.84, "learning_rate": 9.122385276720935e-09, "loss": 1.2504, "step": 2488 }, { "epoch": 3.85, "learning_rate": 8.788138712693126e-09, "loss": 1.2854, "step": 2490 }, { "epoch": 3.85, "learning_rate": 8.460103392194718e-09, "loss": 1.2353, "step": 2492 }, { "epoch": 3.85, "learning_rate": 8.138281370797861e-09, "loss": 1.2847, "step": 2494 }, { "epoch": 3.86, "learning_rate": 7.82267466513975e-09, "loss": 1.2959, "step": 2496 }, { "epoch": 3.86, "learning_rate": 7.513285252910839e-09, "loss": 1.2548, "step": 2498 }, { "epoch": 3.86, "learning_rate": 7.2101150728418606e-09, "loss": 1.2525, "step": 2500 }, { "epoch": 3.87, "learning_rate": 6.913166024692052e-09, "loss": 1.216, "step": 2502 }, { "epoch": 3.87, "learning_rate": 6.622439969236948e-09, "loss": 1.2235, "step": 2504 }, { "epoch": 3.87, "learning_rate": 6.337938728257053e-09, "loss": 1.2088, "step": 2506 }, { "epoch": 3.87, "learning_rate": 6.059664084526073e-09, "loss": 1.2675, "step": 2508 }, { "epoch": 3.88, "learning_rate": 5.787617781800036e-09, "loss": 1.2696, "step": 2510 }, { "epoch": 3.88, "learning_rate": 5.521801524805858e-09, "loss": 1.2372, "step": 2512 }, { "epoch": 3.88, "learning_rate": 5.262216979231682e-09, "loss": 1.1922, "step": 2514 }, { "epoch": 3.89, "learning_rate": 5.0088657717152205e-09, "loss": 1.2559, "step": 2516 }, { "epoch": 3.89, "learning_rate": 4.761749489834544e-09, "loss": 1.2807, "step": 2518 }, { "epoch": 3.89, "learning_rate": 4.520869682097528e-09, "loss": 1.2705, "step": 2520 }, { "epoch": 3.9, "learning_rate": 4.2862278579323115e-09, "loss": 1.2546, "step": 2522 }, { "epoch": 3.9, "learning_rate": 4.057825487678079e-09, "loss": 1.2622, "step": 2524 }, { "epoch": 3.9, "learning_rate": 3.8356640025752895e-09, "loss": 1.2713, "step": 2526 }, { "epoch": 3.91, "learning_rate": 3.619744794757462e-09, "loss": 1.2215, "step": 2528 }, { "epoch": 3.91, "learning_rate": 3.410069217241962e-09, "loss": 1.3269, "step": 2530 }, { "epoch": 3.91, "learning_rate": 3.2066385839213393e-09, "loss": 1.2081, "step": 2532 }, { "epoch": 3.92, "learning_rate": 3.0094541695556695e-09, "loss": 1.3215, "step": 2534 }, { "epoch": 3.92, "learning_rate": 2.818517209764115e-09, "loss": 1.2093, "step": 2536 }, { "epoch": 3.92, "learning_rate": 2.6338289010174874e-09, "loss": 1.2465, "step": 2538 }, { "epoch": 3.92, "learning_rate": 2.455390400630364e-09, "loss": 1.2276, "step": 2540 }, { "epoch": 3.93, "learning_rate": 2.283202826754316e-09, "loss": 1.3187, "step": 2542 }, { "epoch": 3.93, "learning_rate": 2.117267258370359e-09, "loss": 1.2377, "step": 2544 }, { "epoch": 3.93, "learning_rate": 1.957584735282847e-09, "loss": 1.3236, "step": 2546 }, { "epoch": 3.94, "learning_rate": 1.804156258112255e-09, "loss": 1.3, "step": 2548 }, { "epoch": 3.94, "learning_rate": 1.6569827882896292e-09, "loss": 1.2685, "step": 2550 }, { "epoch": 3.94, "learning_rate": 1.5160652480497028e-09, "loss": 1.2171, "step": 2552 }, { "epoch": 3.95, "learning_rate": 1.3814045204260105e-09, "loss": 1.24, "step": 2554 }, { "epoch": 3.95, "learning_rate": 1.2530014492446728e-09, "loss": 1.248, "step": 2556 }, { "epoch": 3.95, "learning_rate": 1.1308568391196205e-09, "loss": 1.2048, "step": 2558 }, { "epoch": 3.96, "learning_rate": 1.0149714554471555e-09, "loss": 1.3384, "step": 2560 }, { "epoch": 3.96, "learning_rate": 9.053460244013988e-10, "loss": 1.1843, "step": 2562 }, { "epoch": 3.96, "learning_rate": 8.019812329292941e-10, "loss": 1.2247, "step": 2564 }, { "epoch": 3.96, "learning_rate": 7.048777287472773e-10, "loss": 1.1788, "step": 2566 }, { "epoch": 3.97, "learning_rate": 6.140361203361699e-10, "loss": 1.1927, "step": 2568 }, { "epoch": 3.97, "learning_rate": 5.294569769378476e-10, "loss": 1.1989, "step": 2570 }, { "epoch": 3.97, "learning_rate": 4.511408285516882e-10, "loss": 1.249, "step": 2572 }, { "epoch": 3.98, "learning_rate": 3.790881659312406e-10, "loss": 1.2052, "step": 2574 }, { "epoch": 3.98, "learning_rate": 3.132994405808942e-10, "loss": 1.1948, "step": 2576 }, { "epoch": 3.98, "learning_rate": 2.537750647535475e-10, "loss": 1.249, "step": 2578 }, { "epoch": 3.99, "learning_rate": 2.005154114474994e-10, "loss": 1.2526, "step": 2580 }, { "epoch": 3.99, "learning_rate": 1.5352081440422881e-10, "loss": 1.2689, "step": 2582 }, { "epoch": 3.99, "learning_rate": 1.1279156810684032e-10, "loss": 1.2414, "step": 2584 }, { "epoch": 4.0, "learning_rate": 7.83279277773996e-11, "loss": 1.2747, "step": 2586 }, { "epoch": 4.0, "learning_rate": 5.013010937582329e-11, "loss": 1.2184, "step": 2588 }, { "epoch": 4.0, "step": 2588, "total_flos": 7.469990048641843e+16, "train_loss": 1.341316524852741, "train_runtime": 24321.1859, "train_samples_per_second": 6.813, "train_steps_per_second": 0.106 } ], "logging_steps": 2, "max_steps": 2588, "num_input_tokens_seen": 0, "num_train_epochs": 4, "save_steps": 9000, "total_flos": 7.469990048641843e+16, "train_batch_size": 2, "trial_name": null, "trial_params": null }