{ "best_metric": null, "best_model_checkpoint": null, "epoch": 2.9988146977479255, "global_step": 3795, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 1.7543859649122808e-07, "loss": 0.9175, "step": 1 }, { "epoch": 0.0, "learning_rate": 3.5087719298245616e-07, "loss": 0.9571, "step": 2 }, { "epoch": 0.0, "learning_rate": 5.263157894736843e-07, "loss": 0.9694, "step": 3 }, { "epoch": 0.0, "learning_rate": 7.017543859649123e-07, "loss": 0.9406, "step": 4 }, { "epoch": 0.0, "learning_rate": 8.771929824561404e-07, "loss": 0.9173, "step": 5 }, { "epoch": 0.0, "learning_rate": 1.0526315789473685e-06, "loss": 0.8673, "step": 6 }, { "epoch": 0.01, "learning_rate": 1.2280701754385965e-06, "loss": 0.9695, "step": 7 }, { "epoch": 0.01, "learning_rate": 1.4035087719298246e-06, "loss": 0.8544, "step": 8 }, { "epoch": 0.01, "learning_rate": 1.5789473684210526e-06, "loss": 0.8929, "step": 9 }, { "epoch": 0.01, "learning_rate": 1.7543859649122807e-06, "loss": 0.8506, "step": 10 }, { "epoch": 0.01, "learning_rate": 1.929824561403509e-06, "loss": 0.819, "step": 11 }, { "epoch": 0.01, "learning_rate": 2.105263157894737e-06, "loss": 0.794, "step": 12 }, { "epoch": 0.01, "learning_rate": 2.280701754385965e-06, "loss": 0.7997, "step": 13 }, { "epoch": 0.01, "learning_rate": 2.456140350877193e-06, "loss": 0.7402, "step": 14 }, { "epoch": 0.01, "learning_rate": 2.631578947368421e-06, "loss": 0.6854, "step": 15 }, { "epoch": 0.01, "learning_rate": 2.8070175438596493e-06, "loss": 0.6654, "step": 16 }, { "epoch": 0.01, "learning_rate": 2.9824561403508774e-06, "loss": 0.5882, "step": 17 }, { "epoch": 0.01, "learning_rate": 3.157894736842105e-06, "loss": 0.7334, "step": 18 }, { "epoch": 0.02, "learning_rate": 3.3333333333333333e-06, "loss": 0.6077, "step": 19 }, { "epoch": 0.02, "learning_rate": 3.5087719298245615e-06, "loss": 0.6363, "step": 20 }, { "epoch": 0.02, "learning_rate": 3.6842105263157896e-06, "loss": 0.5459, "step": 21 }, { "epoch": 0.02, "learning_rate": 3.859649122807018e-06, "loss": 0.6126, "step": 22 }, { "epoch": 0.02, "learning_rate": 4.035087719298246e-06, "loss": 0.595, "step": 23 }, { "epoch": 0.02, "learning_rate": 4.210526315789474e-06, "loss": 0.5921, "step": 24 }, { "epoch": 0.02, "learning_rate": 4.385964912280702e-06, "loss": 0.537, "step": 25 }, { "epoch": 0.02, "learning_rate": 4.56140350877193e-06, "loss": 0.4832, "step": 26 }, { "epoch": 0.02, "learning_rate": 4.736842105263158e-06, "loss": 0.5233, "step": 27 }, { "epoch": 0.02, "learning_rate": 4.912280701754386e-06, "loss": 0.5377, "step": 28 }, { "epoch": 0.02, "learning_rate": 5.087719298245615e-06, "loss": 0.5105, "step": 29 }, { "epoch": 0.02, "learning_rate": 5.263157894736842e-06, "loss": 0.5156, "step": 30 }, { "epoch": 0.02, "learning_rate": 5.438596491228071e-06, "loss": 0.4773, "step": 31 }, { "epoch": 0.03, "learning_rate": 5.6140350877192985e-06, "loss": 0.4834, "step": 32 }, { "epoch": 0.03, "learning_rate": 5.789473684210527e-06, "loss": 0.4922, "step": 33 }, { "epoch": 0.03, "learning_rate": 5.964912280701755e-06, "loss": 0.4956, "step": 34 }, { "epoch": 0.03, "learning_rate": 6.140350877192983e-06, "loss": 0.4263, "step": 35 }, { "epoch": 0.03, "learning_rate": 6.31578947368421e-06, "loss": 0.4614, "step": 36 }, { "epoch": 0.03, "learning_rate": 6.491228070175439e-06, "loss": 0.4971, "step": 37 }, { "epoch": 0.03, "learning_rate": 6.666666666666667e-06, "loss": 0.4768, "step": 38 }, { "epoch": 0.03, "learning_rate": 6.842105263157896e-06, "loss": 0.5321, "step": 39 }, { "epoch": 0.03, "learning_rate": 7.017543859649123e-06, "loss": 0.5106, "step": 40 }, { "epoch": 0.03, "learning_rate": 7.192982456140352e-06, "loss": 0.5049, "step": 41 }, { "epoch": 0.03, "learning_rate": 7.368421052631579e-06, "loss": 0.4402, "step": 42 }, { "epoch": 0.03, "learning_rate": 7.5438596491228074e-06, "loss": 0.4478, "step": 43 }, { "epoch": 0.03, "learning_rate": 7.719298245614036e-06, "loss": 0.4868, "step": 44 }, { "epoch": 0.04, "learning_rate": 7.894736842105265e-06, "loss": 0.4454, "step": 45 }, { "epoch": 0.04, "learning_rate": 8.070175438596492e-06, "loss": 0.4592, "step": 46 }, { "epoch": 0.04, "learning_rate": 8.24561403508772e-06, "loss": 0.4555, "step": 47 }, { "epoch": 0.04, "learning_rate": 8.421052631578948e-06, "loss": 0.444, "step": 48 }, { "epoch": 0.04, "learning_rate": 8.596491228070176e-06, "loss": 0.4071, "step": 49 }, { "epoch": 0.04, "learning_rate": 8.771929824561405e-06, "loss": 0.4274, "step": 50 }, { "epoch": 0.04, "learning_rate": 8.947368421052632e-06, "loss": 0.4605, "step": 51 }, { "epoch": 0.04, "learning_rate": 9.12280701754386e-06, "loss": 0.5379, "step": 52 }, { "epoch": 0.04, "learning_rate": 9.298245614035088e-06, "loss": 0.449, "step": 53 }, { "epoch": 0.04, "learning_rate": 9.473684210526315e-06, "loss": 0.4423, "step": 54 }, { "epoch": 0.04, "learning_rate": 9.649122807017545e-06, "loss": 0.4577, "step": 55 }, { "epoch": 0.04, "learning_rate": 9.824561403508772e-06, "loss": 0.5124, "step": 56 }, { "epoch": 0.05, "learning_rate": 1e-05, "loss": 0.4397, "step": 57 }, { "epoch": 0.05, "learning_rate": 1.017543859649123e-05, "loss": 0.4175, "step": 58 }, { "epoch": 0.05, "learning_rate": 1.0350877192982459e-05, "loss": 0.416, "step": 59 }, { "epoch": 0.05, "learning_rate": 1.0526315789473684e-05, "loss": 0.5025, "step": 60 }, { "epoch": 0.05, "learning_rate": 1.0701754385964913e-05, "loss": 0.4217, "step": 61 }, { "epoch": 0.05, "learning_rate": 1.0877192982456142e-05, "loss": 0.4337, "step": 62 }, { "epoch": 0.05, "learning_rate": 1.105263157894737e-05, "loss": 0.4581, "step": 63 }, { "epoch": 0.05, "learning_rate": 1.1228070175438597e-05, "loss": 0.4722, "step": 64 }, { "epoch": 0.05, "learning_rate": 1.1403508771929826e-05, "loss": 0.4918, "step": 65 }, { "epoch": 0.05, "learning_rate": 1.1578947368421053e-05, "loss": 0.4846, "step": 66 }, { "epoch": 0.05, "learning_rate": 1.1754385964912282e-05, "loss": 0.4425, "step": 67 }, { "epoch": 0.05, "learning_rate": 1.192982456140351e-05, "loss": 0.4186, "step": 68 }, { "epoch": 0.05, "learning_rate": 1.2105263157894737e-05, "loss": 0.4766, "step": 69 }, { "epoch": 0.06, "learning_rate": 1.2280701754385966e-05, "loss": 0.4219, "step": 70 }, { "epoch": 0.06, "learning_rate": 1.2456140350877195e-05, "loss": 0.4733, "step": 71 }, { "epoch": 0.06, "learning_rate": 1.263157894736842e-05, "loss": 0.4114, "step": 72 }, { "epoch": 0.06, "learning_rate": 1.280701754385965e-05, "loss": 0.4731, "step": 73 }, { "epoch": 0.06, "learning_rate": 1.2982456140350879e-05, "loss": 0.3987, "step": 74 }, { "epoch": 0.06, "learning_rate": 1.3157894736842108e-05, "loss": 0.4115, "step": 75 }, { "epoch": 0.06, "learning_rate": 1.3333333333333333e-05, "loss": 0.453, "step": 76 }, { "epoch": 0.06, "learning_rate": 1.3508771929824562e-05, "loss": 0.4482, "step": 77 }, { "epoch": 0.06, "learning_rate": 1.3684210526315791e-05, "loss": 0.4397, "step": 78 }, { "epoch": 0.06, "learning_rate": 1.385964912280702e-05, "loss": 0.4385, "step": 79 }, { "epoch": 0.06, "learning_rate": 1.4035087719298246e-05, "loss": 0.3958, "step": 80 }, { "epoch": 0.06, "learning_rate": 1.4210526315789475e-05, "loss": 0.4315, "step": 81 }, { "epoch": 0.06, "learning_rate": 1.4385964912280704e-05, "loss": 0.3988, "step": 82 }, { "epoch": 0.07, "learning_rate": 1.4561403508771931e-05, "loss": 0.3997, "step": 83 }, { "epoch": 0.07, "learning_rate": 1.4736842105263159e-05, "loss": 0.3686, "step": 84 }, { "epoch": 0.07, "learning_rate": 1.4912280701754388e-05, "loss": 0.4712, "step": 85 }, { "epoch": 0.07, "learning_rate": 1.5087719298245615e-05, "loss": 0.4239, "step": 86 }, { "epoch": 0.07, "learning_rate": 1.5263157894736846e-05, "loss": 0.427, "step": 87 }, { "epoch": 0.07, "learning_rate": 1.543859649122807e-05, "loss": 0.425, "step": 88 }, { "epoch": 0.07, "learning_rate": 1.56140350877193e-05, "loss": 0.4243, "step": 89 }, { "epoch": 0.07, "learning_rate": 1.578947368421053e-05, "loss": 0.3949, "step": 90 }, { "epoch": 0.07, "learning_rate": 1.5964912280701755e-05, "loss": 0.4023, "step": 91 }, { "epoch": 0.07, "learning_rate": 1.6140350877192984e-05, "loss": 0.4674, "step": 92 }, { "epoch": 0.07, "learning_rate": 1.6315789473684213e-05, "loss": 0.4055, "step": 93 }, { "epoch": 0.07, "learning_rate": 1.649122807017544e-05, "loss": 0.4447, "step": 94 }, { "epoch": 0.08, "learning_rate": 1.6666666666666667e-05, "loss": 0.38, "step": 95 }, { "epoch": 0.08, "learning_rate": 1.6842105263157896e-05, "loss": 0.4741, "step": 96 }, { "epoch": 0.08, "learning_rate": 1.7017543859649125e-05, "loss": 0.4231, "step": 97 }, { "epoch": 0.08, "learning_rate": 1.719298245614035e-05, "loss": 0.4495, "step": 98 }, { "epoch": 0.08, "learning_rate": 1.736842105263158e-05, "loss": 0.4012, "step": 99 }, { "epoch": 0.08, "learning_rate": 1.754385964912281e-05, "loss": 0.409, "step": 100 }, { "epoch": 0.08, "learning_rate": 1.7719298245614035e-05, "loss": 0.4863, "step": 101 }, { "epoch": 0.08, "learning_rate": 1.7894736842105264e-05, "loss": 0.406, "step": 102 }, { "epoch": 0.08, "learning_rate": 1.8070175438596493e-05, "loss": 0.4492, "step": 103 }, { "epoch": 0.08, "learning_rate": 1.824561403508772e-05, "loss": 0.4491, "step": 104 }, { "epoch": 0.08, "learning_rate": 1.8421052631578947e-05, "loss": 0.5007, "step": 105 }, { "epoch": 0.08, "learning_rate": 1.8596491228070176e-05, "loss": 0.4385, "step": 106 }, { "epoch": 0.08, "learning_rate": 1.8771929824561405e-05, "loss": 0.43, "step": 107 }, { "epoch": 0.09, "learning_rate": 1.894736842105263e-05, "loss": 0.4007, "step": 108 }, { "epoch": 0.09, "learning_rate": 1.912280701754386e-05, "loss": 0.4388, "step": 109 }, { "epoch": 0.09, "learning_rate": 1.929824561403509e-05, "loss": 0.4397, "step": 110 }, { "epoch": 0.09, "learning_rate": 1.9473684210526318e-05, "loss": 0.4102, "step": 111 }, { "epoch": 0.09, "learning_rate": 1.9649122807017544e-05, "loss": 0.4137, "step": 112 }, { "epoch": 0.09, "learning_rate": 1.9824561403508773e-05, "loss": 0.4273, "step": 113 }, { "epoch": 0.09, "learning_rate": 2e-05, "loss": 0.4198, "step": 114 }, { "epoch": 0.09, "learning_rate": 1.9999996358015542e-05, "loss": 0.3666, "step": 115 }, { "epoch": 0.09, "learning_rate": 1.999998543206481e-05, "loss": 0.4241, "step": 116 }, { "epoch": 0.09, "learning_rate": 1.999996722215577e-05, "loss": 0.4218, "step": 117 }, { "epoch": 0.09, "learning_rate": 1.9999941728301686e-05, "loss": 0.4542, "step": 118 }, { "epoch": 0.09, "learning_rate": 1.9999908950521122e-05, "loss": 0.45, "step": 119 }, { "epoch": 0.09, "learning_rate": 1.9999868888837957e-05, "loss": 0.4778, "step": 120 }, { "epoch": 0.1, "learning_rate": 1.999982154328137e-05, "loss": 0.4134, "step": 121 }, { "epoch": 0.1, "learning_rate": 1.999976691388585e-05, "loss": 0.4835, "step": 122 }, { "epoch": 0.1, "learning_rate": 1.999970500069119e-05, "loss": 0.431, "step": 123 }, { "epoch": 0.1, "learning_rate": 1.9999635803742482e-05, "loss": 0.4603, "step": 124 }, { "epoch": 0.1, "learning_rate": 1.9999559323090132e-05, "loss": 0.4487, "step": 125 }, { "epoch": 0.1, "learning_rate": 1.999947555878985e-05, "loss": 0.4043, "step": 126 }, { "epoch": 0.1, "learning_rate": 1.9999384510902643e-05, "loss": 0.3752, "step": 127 }, { "epoch": 0.1, "learning_rate": 1.999928617949484e-05, "loss": 0.407, "step": 128 }, { "epoch": 0.1, "learning_rate": 1.9999180564638056e-05, "loss": 0.4491, "step": 129 }, { "epoch": 0.1, "learning_rate": 1.9999067666409225e-05, "loss": 0.4022, "step": 130 }, { "epoch": 0.1, "learning_rate": 1.999894748489058e-05, "loss": 0.4192, "step": 131 }, { "epoch": 0.1, "learning_rate": 1.9998820020169668e-05, "loss": 0.3941, "step": 132 }, { "epoch": 0.11, "learning_rate": 1.9998685272339325e-05, "loss": 0.4694, "step": 133 }, { "epoch": 0.11, "learning_rate": 1.9998543241497702e-05, "loss": 0.3873, "step": 134 }, { "epoch": 0.11, "learning_rate": 1.9998393927748257e-05, "loss": 0.4235, "step": 135 }, { "epoch": 0.11, "learning_rate": 1.999823733119975e-05, "loss": 0.4316, "step": 136 }, { "epoch": 0.11, "learning_rate": 1.9998073451966244e-05, "loss": 0.3604, "step": 137 }, { "epoch": 0.11, "learning_rate": 1.9997902290167104e-05, "loss": 0.4102, "step": 138 }, { "epoch": 0.11, "learning_rate": 1.999772384592701e-05, "loss": 0.4146, "step": 139 }, { "epoch": 0.11, "learning_rate": 1.9997538119375938e-05, "loss": 0.4845, "step": 140 }, { "epoch": 0.11, "learning_rate": 1.999734511064917e-05, "loss": 0.4373, "step": 141 }, { "epoch": 0.11, "learning_rate": 1.9997144819887292e-05, "loss": 0.4559, "step": 142 }, { "epoch": 0.11, "learning_rate": 1.99969372472362e-05, "loss": 0.4722, "step": 143 }, { "epoch": 0.11, "learning_rate": 1.9996722392847082e-05, "loss": 0.4635, "step": 144 }, { "epoch": 0.11, "learning_rate": 1.9996500256876447e-05, "loss": 0.3857, "step": 145 }, { "epoch": 0.12, "learning_rate": 1.999627083948609e-05, "loss": 0.3903, "step": 146 }, { "epoch": 0.12, "learning_rate": 1.9996034140843113e-05, "loss": 0.4037, "step": 147 }, { "epoch": 0.12, "learning_rate": 1.9995790161119943e-05, "loss": 0.4756, "step": 148 }, { "epoch": 0.12, "learning_rate": 1.999553890049428e-05, "loss": 0.4159, "step": 149 }, { "epoch": 0.12, "learning_rate": 1.999528035914915e-05, "loss": 0.377, "step": 150 }, { "epoch": 0.12, "learning_rate": 1.999501453727287e-05, "loss": 0.4045, "step": 151 }, { "epoch": 0.12, "learning_rate": 1.999474143505906e-05, "loss": 0.4199, "step": 152 }, { "epoch": 0.12, "learning_rate": 1.9994461052706652e-05, "loss": 0.3716, "step": 153 }, { "epoch": 0.12, "learning_rate": 1.9994173390419876e-05, "loss": 0.4289, "step": 154 }, { "epoch": 0.12, "learning_rate": 1.9993878448408263e-05, "loss": 0.4451, "step": 155 }, { "epoch": 0.12, "learning_rate": 1.9993576226886644e-05, "loss": 0.4121, "step": 156 }, { "epoch": 0.12, "learning_rate": 1.9993266726075163e-05, "loss": 0.4132, "step": 157 }, { "epoch": 0.12, "learning_rate": 1.9992949946199255e-05, "loss": 0.422, "step": 158 }, { "epoch": 0.13, "learning_rate": 1.999262588748966e-05, "loss": 0.406, "step": 159 }, { "epoch": 0.13, "learning_rate": 1.999229455018243e-05, "loss": 0.3656, "step": 160 }, { "epoch": 0.13, "learning_rate": 1.99919559345189e-05, "loss": 0.3831, "step": 161 }, { "epoch": 0.13, "learning_rate": 1.9991610040745718e-05, "loss": 0.435, "step": 162 }, { "epoch": 0.13, "learning_rate": 1.999125686911484e-05, "loss": 0.445, "step": 163 }, { "epoch": 0.13, "learning_rate": 1.9990896419883506e-05, "loss": 0.4183, "step": 164 }, { "epoch": 0.13, "learning_rate": 1.9990528693314273e-05, "loss": 0.4415, "step": 165 }, { "epoch": 0.13, "learning_rate": 1.9990153689674983e-05, "loss": 0.4237, "step": 166 }, { "epoch": 0.13, "learning_rate": 1.9989771409238802e-05, "loss": 0.417, "step": 167 }, { "epoch": 0.13, "learning_rate": 1.9989381852284165e-05, "loss": 0.4081, "step": 168 }, { "epoch": 0.13, "learning_rate": 1.9988985019094837e-05, "loss": 0.402, "step": 169 }, { "epoch": 0.13, "learning_rate": 1.9988580909959864e-05, "loss": 0.42, "step": 170 }, { "epoch": 0.14, "learning_rate": 1.99881695251736e-05, "loss": 0.391, "step": 171 }, { "epoch": 0.14, "learning_rate": 1.9987750865035693e-05, "loss": 0.4206, "step": 172 }, { "epoch": 0.14, "learning_rate": 1.99873249298511e-05, "loss": 0.4424, "step": 173 }, { "epoch": 0.14, "learning_rate": 1.998689171993006e-05, "loss": 0.3529, "step": 174 }, { "epoch": 0.14, "learning_rate": 1.9986451235588135e-05, "loss": 0.4055, "step": 175 }, { "epoch": 0.14, "learning_rate": 1.998600347714616e-05, "loss": 0.4039, "step": 176 }, { "epoch": 0.14, "learning_rate": 1.9985548444930295e-05, "loss": 0.4236, "step": 177 }, { "epoch": 0.14, "learning_rate": 1.998508613927197e-05, "loss": 0.4209, "step": 178 }, { "epoch": 0.14, "learning_rate": 1.9984616560507938e-05, "loss": 0.3969, "step": 179 }, { "epoch": 0.14, "learning_rate": 1.9984139708980228e-05, "loss": 0.3919, "step": 180 }, { "epoch": 0.14, "learning_rate": 1.9983655585036187e-05, "loss": 0.3731, "step": 181 }, { "epoch": 0.14, "learning_rate": 1.9983164189028447e-05, "loss": 0.3965, "step": 182 }, { "epoch": 0.14, "learning_rate": 1.9982665521314934e-05, "loss": 0.3735, "step": 183 }, { "epoch": 0.15, "learning_rate": 1.9982159582258882e-05, "loss": 0.3908, "step": 184 }, { "epoch": 0.15, "learning_rate": 1.9981646372228813e-05, "loss": 0.3849, "step": 185 }, { "epoch": 0.15, "learning_rate": 1.9981125891598545e-05, "loss": 0.4001, "step": 186 }, { "epoch": 0.15, "learning_rate": 1.9980598140747197e-05, "loss": 0.3842, "step": 187 }, { "epoch": 0.15, "learning_rate": 1.998006312005919e-05, "loss": 0.3661, "step": 188 }, { "epoch": 0.15, "learning_rate": 1.9979520829924212e-05, "loss": 0.462, "step": 189 }, { "epoch": 0.15, "learning_rate": 1.997897127073728e-05, "loss": 0.4031, "step": 190 }, { "epoch": 0.15, "learning_rate": 1.9978414442898688e-05, "loss": 0.3761, "step": 191 }, { "epoch": 0.15, "learning_rate": 1.9977850346814026e-05, "loss": 0.4175, "step": 192 }, { "epoch": 0.15, "learning_rate": 1.9977278982894185e-05, "loss": 0.3788, "step": 193 }, { "epoch": 0.15, "learning_rate": 1.9976700351555337e-05, "loss": 0.3921, "step": 194 }, { "epoch": 0.15, "learning_rate": 1.997611445321896e-05, "loss": 0.4278, "step": 195 }, { "epoch": 0.15, "learning_rate": 1.9975521288311817e-05, "loss": 0.3816, "step": 196 }, { "epoch": 0.16, "learning_rate": 1.997492085726597e-05, "loss": 0.4293, "step": 197 }, { "epoch": 0.16, "learning_rate": 1.9974313160518776e-05, "loss": 0.3721, "step": 198 }, { "epoch": 0.16, "learning_rate": 1.997369819851287e-05, "loss": 0.4243, "step": 199 }, { "epoch": 0.16, "learning_rate": 1.9973075971696195e-05, "loss": 0.4047, "step": 200 }, { "epoch": 0.16, "learning_rate": 1.9972446480521972e-05, "loss": 0.4096, "step": 201 }, { "epoch": 0.16, "learning_rate": 1.997180972544873e-05, "loss": 0.4134, "step": 202 }, { "epoch": 0.16, "learning_rate": 1.9971165706940275e-05, "loss": 0.4021, "step": 203 }, { "epoch": 0.16, "learning_rate": 1.9970514425465706e-05, "loss": 0.3813, "step": 204 }, { "epoch": 0.16, "learning_rate": 1.9969855881499413e-05, "loss": 0.3991, "step": 205 }, { "epoch": 0.16, "learning_rate": 1.9969190075521086e-05, "loss": 0.3963, "step": 206 }, { "epoch": 0.16, "learning_rate": 1.996851700801569e-05, "loss": 0.404, "step": 207 }, { "epoch": 0.16, "learning_rate": 1.9967836679473483e-05, "loss": 0.4168, "step": 208 }, { "epoch": 0.17, "learning_rate": 1.9967149090390018e-05, "loss": 0.3938, "step": 209 }, { "epoch": 0.17, "learning_rate": 1.996645424126613e-05, "loss": 0.3682, "step": 210 }, { "epoch": 0.17, "learning_rate": 1.996575213260795e-05, "loss": 0.379, "step": 211 }, { "epoch": 0.17, "learning_rate": 1.9965042764926887e-05, "loss": 0.3883, "step": 212 }, { "epoch": 0.17, "learning_rate": 1.9964326138739645e-05, "loss": 0.4345, "step": 213 }, { "epoch": 0.17, "learning_rate": 1.9963602254568212e-05, "loss": 0.3644, "step": 214 }, { "epoch": 0.17, "learning_rate": 1.996287111293986e-05, "loss": 0.3846, "step": 215 }, { "epoch": 0.17, "learning_rate": 1.996213271438715e-05, "loss": 0.3946, "step": 216 }, { "epoch": 0.17, "learning_rate": 1.9961387059447936e-05, "loss": 0.378, "step": 217 }, { "epoch": 0.17, "learning_rate": 1.9960634148665344e-05, "loss": 0.3835, "step": 218 }, { "epoch": 0.17, "learning_rate": 1.9959873982587795e-05, "loss": 0.394, "step": 219 }, { "epoch": 0.17, "learning_rate": 1.9959106561768988e-05, "loss": 0.3696, "step": 220 }, { "epoch": 0.17, "learning_rate": 1.9958331886767913e-05, "loss": 0.3517, "step": 221 }, { "epoch": 0.18, "learning_rate": 1.9957549958148844e-05, "loss": 0.3816, "step": 222 }, { "epoch": 0.18, "learning_rate": 1.995676077648133e-05, "loss": 0.3796, "step": 223 }, { "epoch": 0.18, "learning_rate": 1.9955964342340212e-05, "loss": 0.3473, "step": 224 }, { "epoch": 0.18, "learning_rate": 1.9955160656305606e-05, "loss": 0.3864, "step": 225 }, { "epoch": 0.18, "learning_rate": 1.9954349718962918e-05, "loss": 0.3848, "step": 226 }, { "epoch": 0.18, "learning_rate": 1.995353153090283e-05, "loss": 0.3625, "step": 227 }, { "epoch": 0.18, "learning_rate": 1.995270609272131e-05, "loss": 0.3429, "step": 228 }, { "epoch": 0.18, "learning_rate": 1.9951873405019602e-05, "loss": 0.399, "step": 229 }, { "epoch": 0.18, "learning_rate": 1.995103346840424e-05, "loss": 0.4233, "step": 230 }, { "epoch": 0.18, "learning_rate": 1.995018628348702e-05, "loss": 0.4422, "step": 231 }, { "epoch": 0.18, "learning_rate": 1.9949331850885033e-05, "loss": 0.3764, "step": 232 }, { "epoch": 0.18, "learning_rate": 1.994847017122065e-05, "loss": 0.3702, "step": 233 }, { "epoch": 0.18, "learning_rate": 1.9947601245121514e-05, "loss": 0.3778, "step": 234 }, { "epoch": 0.19, "learning_rate": 1.9946725073220542e-05, "loss": 0.355, "step": 235 }, { "epoch": 0.19, "learning_rate": 1.9945841656155943e-05, "loss": 0.3906, "step": 236 }, { "epoch": 0.19, "learning_rate": 1.9944950994571192e-05, "loss": 0.3764, "step": 237 }, { "epoch": 0.19, "learning_rate": 1.9944053089115045e-05, "loss": 0.383, "step": 238 }, { "epoch": 0.19, "learning_rate": 1.994314794044153e-05, "loss": 0.404, "step": 239 }, { "epoch": 0.19, "learning_rate": 1.9942235549209955e-05, "loss": 0.418, "step": 240 }, { "epoch": 0.19, "learning_rate": 1.9941315916084905e-05, "loss": 0.3527, "step": 241 }, { "epoch": 0.19, "learning_rate": 1.994038904173624e-05, "loss": 0.3658, "step": 242 }, { "epoch": 0.19, "learning_rate": 1.993945492683909e-05, "loss": 0.3601, "step": 243 }, { "epoch": 0.19, "learning_rate": 1.993851357207386e-05, "loss": 0.387, "step": 244 }, { "epoch": 0.19, "learning_rate": 1.9937564978126233e-05, "loss": 0.3631, "step": 245 }, { "epoch": 0.19, "learning_rate": 1.993660914568716e-05, "loss": 0.3683, "step": 246 }, { "epoch": 0.2, "learning_rate": 1.9935646075452863e-05, "loss": 0.3786, "step": 247 }, { "epoch": 0.2, "learning_rate": 1.9934675768124848e-05, "loss": 0.3879, "step": 248 }, { "epoch": 0.2, "learning_rate": 1.9933698224409876e-05, "loss": 0.3979, "step": 249 }, { "epoch": 0.2, "learning_rate": 1.9932713445019993e-05, "loss": 0.3654, "step": 250 }, { "epoch": 0.2, "learning_rate": 1.99317214306725e-05, "loss": 0.3706, "step": 251 }, { "epoch": 0.2, "learning_rate": 1.993072218208999e-05, "loss": 0.3513, "step": 252 }, { "epoch": 0.2, "learning_rate": 1.9929715700000297e-05, "loss": 0.4279, "step": 253 }, { "epoch": 0.2, "learning_rate": 1.9928701985136554e-05, "loss": 0.3735, "step": 254 }, { "epoch": 0.2, "learning_rate": 1.992768103823714e-05, "loss": 0.3516, "step": 255 }, { "epoch": 0.2, "learning_rate": 1.9926652860045713e-05, "loss": 0.3632, "step": 256 }, { "epoch": 0.2, "learning_rate": 1.9925617451311193e-05, "loss": 0.382, "step": 257 }, { "epoch": 0.2, "learning_rate": 1.9924574812787766e-05, "loss": 0.3461, "step": 258 }, { "epoch": 0.2, "learning_rate": 1.9923524945234893e-05, "loss": 0.4273, "step": 259 }, { "epoch": 0.21, "learning_rate": 1.9922467849417288e-05, "loss": 0.3891, "step": 260 }, { "epoch": 0.21, "learning_rate": 1.992140352610494e-05, "loss": 0.3768, "step": 261 }, { "epoch": 0.21, "learning_rate": 1.9920331976073095e-05, "loss": 0.3941, "step": 262 }, { "epoch": 0.21, "learning_rate": 1.9919253200102273e-05, "loss": 0.4546, "step": 263 }, { "epoch": 0.21, "learning_rate": 1.9918167198978246e-05, "loss": 0.363, "step": 264 }, { "epoch": 0.21, "learning_rate": 1.9917073973492055e-05, "loss": 0.3854, "step": 265 }, { "epoch": 0.21, "learning_rate": 1.9915973524440002e-05, "loss": 0.4374, "step": 266 }, { "epoch": 0.21, "learning_rate": 1.991486585262365e-05, "loss": 0.3862, "step": 267 }, { "epoch": 0.21, "learning_rate": 1.991375095884983e-05, "loss": 0.3816, "step": 268 }, { "epoch": 0.21, "learning_rate": 1.9912628843930614e-05, "loss": 0.4171, "step": 269 }, { "epoch": 0.21, "learning_rate": 1.991149950868336e-05, "loss": 0.3653, "step": 270 }, { "epoch": 0.21, "learning_rate": 1.9910362953930668e-05, "loss": 0.3887, "step": 271 }, { "epoch": 0.21, "learning_rate": 1.9909219180500398e-05, "loss": 0.3803, "step": 272 }, { "epoch": 0.22, "learning_rate": 1.9908068189225672e-05, "loss": 0.4454, "step": 273 }, { "epoch": 0.22, "learning_rate": 1.990690998094487e-05, "loss": 0.3737, "step": 274 }, { "epoch": 0.22, "learning_rate": 1.9905744556501627e-05, "loss": 0.375, "step": 275 }, { "epoch": 0.22, "learning_rate": 1.9904571916744836e-05, "loss": 0.3679, "step": 276 }, { "epoch": 0.22, "learning_rate": 1.990339206252864e-05, "loss": 0.382, "step": 277 }, { "epoch": 0.22, "learning_rate": 1.990220499471245e-05, "loss": 0.3621, "step": 278 }, { "epoch": 0.22, "learning_rate": 1.990101071416091e-05, "loss": 0.4189, "step": 279 }, { "epoch": 0.22, "learning_rate": 1.989980922174394e-05, "loss": 0.3656, "step": 280 }, { "epoch": 0.22, "learning_rate": 1.9898600518336696e-05, "loss": 0.3642, "step": 281 }, { "epoch": 0.22, "learning_rate": 1.98973846048196e-05, "loss": 0.3975, "step": 282 }, { "epoch": 0.22, "learning_rate": 1.9896161482078317e-05, "loss": 0.4083, "step": 283 }, { "epoch": 0.22, "learning_rate": 1.9894931151003765e-05, "loss": 0.3875, "step": 284 }, { "epoch": 0.23, "learning_rate": 1.9893693612492116e-05, "loss": 0.3829, "step": 285 }, { "epoch": 0.23, "learning_rate": 1.9892448867444785e-05, "loss": 0.3418, "step": 286 }, { "epoch": 0.23, "learning_rate": 1.9891196916768444e-05, "loss": 0.3641, "step": 287 }, { "epoch": 0.23, "learning_rate": 1.9889937761375015e-05, "loss": 0.3729, "step": 288 }, { "epoch": 0.23, "learning_rate": 1.988867140218165e-05, "loss": 0.3944, "step": 289 }, { "epoch": 0.23, "learning_rate": 1.988739784011077e-05, "loss": 0.3513, "step": 290 }, { "epoch": 0.23, "learning_rate": 1.9886117076090033e-05, "loss": 0.3581, "step": 291 }, { "epoch": 0.23, "learning_rate": 1.988482911105234e-05, "loss": 0.38, "step": 292 }, { "epoch": 0.23, "learning_rate": 1.9883533945935842e-05, "loss": 0.416, "step": 293 }, { "epoch": 0.23, "learning_rate": 1.9882231581683938e-05, "loss": 0.3598, "step": 294 }, { "epoch": 0.23, "learning_rate": 1.9880922019245258e-05, "loss": 0.3285, "step": 295 }, { "epoch": 0.23, "learning_rate": 1.987960525957369e-05, "loss": 0.3412, "step": 296 }, { "epoch": 0.23, "learning_rate": 1.9878281303628352e-05, "loss": 0.3722, "step": 297 }, { "epoch": 0.24, "learning_rate": 1.9876950152373617e-05, "loss": 0.376, "step": 298 }, { "epoch": 0.24, "learning_rate": 1.9875611806779085e-05, "loss": 0.3882, "step": 299 }, { "epoch": 0.24, "learning_rate": 1.9874266267819604e-05, "loss": 0.3607, "step": 300 }, { "epoch": 0.24, "learning_rate": 1.9872913536475266e-05, "loss": 0.3357, "step": 301 }, { "epoch": 0.24, "learning_rate": 1.9871553613731386e-05, "loss": 0.389, "step": 302 }, { "epoch": 0.24, "learning_rate": 1.987018650057853e-05, "loss": 0.3971, "step": 303 }, { "epoch": 0.24, "learning_rate": 1.986881219801251e-05, "loss": 0.3766, "step": 304 }, { "epoch": 0.24, "learning_rate": 1.986743070703435e-05, "loss": 0.3762, "step": 305 }, { "epoch": 0.24, "learning_rate": 1.986604202865033e-05, "loss": 0.3952, "step": 306 }, { "epoch": 0.24, "learning_rate": 1.986464616387196e-05, "loss": 0.3809, "step": 307 }, { "epoch": 0.24, "learning_rate": 1.9863243113715983e-05, "loss": 0.3743, "step": 308 }, { "epoch": 0.24, "learning_rate": 1.986183287920437e-05, "loss": 0.4083, "step": 309 }, { "epoch": 0.24, "learning_rate": 1.9860415461364343e-05, "loss": 0.398, "step": 310 }, { "epoch": 0.25, "learning_rate": 1.9858990861228332e-05, "loss": 0.3768, "step": 311 }, { "epoch": 0.25, "learning_rate": 1.9857559079834023e-05, "loss": 0.3761, "step": 312 }, { "epoch": 0.25, "learning_rate": 1.9856120118224315e-05, "loss": 0.3785, "step": 313 }, { "epoch": 0.25, "learning_rate": 1.985467397744734e-05, "loss": 0.393, "step": 314 }, { "epoch": 0.25, "learning_rate": 1.9853220658556474e-05, "loss": 0.3554, "step": 315 }, { "epoch": 0.25, "learning_rate": 1.9851760162610298e-05, "loss": 0.3774, "step": 316 }, { "epoch": 0.25, "learning_rate": 1.9850292490672637e-05, "loss": 0.3618, "step": 317 }, { "epoch": 0.25, "learning_rate": 1.984881764381254e-05, "loss": 0.3972, "step": 318 }, { "epoch": 0.25, "learning_rate": 1.9847335623104278e-05, "loss": 0.4131, "step": 319 }, { "epoch": 0.25, "learning_rate": 1.984584642962735e-05, "loss": 0.3703, "step": 320 }, { "epoch": 0.25, "learning_rate": 1.9844350064466488e-05, "loss": 0.3557, "step": 321 }, { "epoch": 0.25, "learning_rate": 1.984284652871163e-05, "loss": 0.36, "step": 322 }, { "epoch": 0.26, "learning_rate": 1.9841335823457952e-05, "loss": 0.3691, "step": 323 }, { "epoch": 0.26, "learning_rate": 1.9839817949805843e-05, "loss": 0.3884, "step": 324 }, { "epoch": 0.26, "learning_rate": 1.9838292908860922e-05, "loss": 0.3938, "step": 325 }, { "epoch": 0.26, "learning_rate": 1.983676070173402e-05, "loss": 0.3837, "step": 326 }, { "epoch": 0.26, "learning_rate": 1.9835221329541197e-05, "loss": 0.3855, "step": 327 }, { "epoch": 0.26, "learning_rate": 1.9833674793403722e-05, "loss": 0.3741, "step": 328 }, { "epoch": 0.26, "learning_rate": 1.983212109444809e-05, "loss": 0.3794, "step": 329 }, { "epoch": 0.26, "learning_rate": 1.9830560233806006e-05, "loss": 0.3683, "step": 330 }, { "epoch": 0.26, "learning_rate": 1.98289922126144e-05, "loss": 0.3704, "step": 331 }, { "epoch": 0.26, "learning_rate": 1.9827417032015415e-05, "loss": 0.3554, "step": 332 }, { "epoch": 0.26, "learning_rate": 1.9825834693156408e-05, "loss": 0.4102, "step": 333 }, { "epoch": 0.26, "learning_rate": 1.9824245197189943e-05, "loss": 0.3636, "step": 334 }, { "epoch": 0.26, "learning_rate": 1.982264854527381e-05, "loss": 0.3612, "step": 335 }, { "epoch": 0.27, "learning_rate": 1.9821044738571008e-05, "loss": 0.3312, "step": 336 }, { "epoch": 0.27, "learning_rate": 1.9819433778249734e-05, "loss": 0.3609, "step": 337 }, { "epoch": 0.27, "learning_rate": 1.981781566548342e-05, "loss": 0.3742, "step": 338 }, { "epoch": 0.27, "learning_rate": 1.981619040145068e-05, "loss": 0.3491, "step": 339 }, { "epoch": 0.27, "learning_rate": 1.9814557987335363e-05, "loss": 0.3117, "step": 340 }, { "epoch": 0.27, "learning_rate": 1.9812918424326512e-05, "loss": 0.4405, "step": 341 }, { "epoch": 0.27, "learning_rate": 1.9811271713618372e-05, "loss": 0.3627, "step": 342 }, { "epoch": 0.27, "learning_rate": 1.980961785641041e-05, "loss": 0.3612, "step": 343 }, { "epoch": 0.27, "learning_rate": 1.9807956853907287e-05, "loss": 0.3673, "step": 344 }, { "epoch": 0.27, "learning_rate": 1.980628870731888e-05, "loss": 0.3531, "step": 345 }, { "epoch": 0.27, "learning_rate": 1.9804613417860244e-05, "loss": 0.3328, "step": 346 }, { "epoch": 0.27, "learning_rate": 1.9802930986751674e-05, "loss": 0.3863, "step": 347 }, { "epoch": 0.27, "learning_rate": 1.9801241415218636e-05, "loss": 0.3675, "step": 348 }, { "epoch": 0.28, "learning_rate": 1.9799544704491813e-05, "loss": 0.3677, "step": 349 }, { "epoch": 0.28, "learning_rate": 1.979784085580708e-05, "loss": 0.351, "step": 350 }, { "epoch": 0.28, "learning_rate": 1.979612987040552e-05, "loss": 0.4164, "step": 351 }, { "epoch": 0.28, "learning_rate": 1.9794411749533405e-05, "loss": 0.4201, "step": 352 }, { "epoch": 0.28, "learning_rate": 1.9792686494442214e-05, "loss": 0.3347, "step": 353 }, { "epoch": 0.28, "learning_rate": 1.9790954106388614e-05, "loss": 0.3665, "step": 354 }, { "epoch": 0.28, "learning_rate": 1.978921458663447e-05, "loss": 0.3694, "step": 355 }, { "epoch": 0.28, "learning_rate": 1.9787467936446846e-05, "loss": 0.395, "step": 356 }, { "epoch": 0.28, "learning_rate": 1.9785714157097992e-05, "loss": 0.3365, "step": 357 }, { "epoch": 0.28, "learning_rate": 1.978395324986536e-05, "loss": 0.3826, "step": 358 }, { "epoch": 0.28, "learning_rate": 1.9782185216031587e-05, "loss": 0.3436, "step": 359 }, { "epoch": 0.28, "learning_rate": 1.9780410056884505e-05, "loss": 0.3492, "step": 360 }, { "epoch": 0.29, "learning_rate": 1.9778627773717134e-05, "loss": 0.4213, "step": 361 }, { "epoch": 0.29, "learning_rate": 1.977683836782768e-05, "loss": 0.3264, "step": 362 }, { "epoch": 0.29, "learning_rate": 1.9775041840519547e-05, "loss": 0.3533, "step": 363 }, { "epoch": 0.29, "learning_rate": 1.9773238193101316e-05, "loss": 0.3557, "step": 364 }, { "epoch": 0.29, "learning_rate": 1.977142742688676e-05, "loss": 0.3567, "step": 365 }, { "epoch": 0.29, "learning_rate": 1.976960954319483e-05, "loss": 0.3666, "step": 366 }, { "epoch": 0.29, "learning_rate": 1.9767784543349674e-05, "loss": 0.3612, "step": 367 }, { "epoch": 0.29, "learning_rate": 1.976595242868061e-05, "loss": 0.3682, "step": 368 }, { "epoch": 0.29, "learning_rate": 1.9764113200522153e-05, "loss": 0.317, "step": 369 }, { "epoch": 0.29, "learning_rate": 1.9762266860213982e-05, "loss": 0.3803, "step": 370 }, { "epoch": 0.29, "learning_rate": 1.9760413409100973e-05, "loss": 0.3722, "step": 371 }, { "epoch": 0.29, "learning_rate": 1.9758552848533168e-05, "loss": 0.3585, "step": 372 }, { "epoch": 0.29, "learning_rate": 1.9756685179865795e-05, "loss": 0.3316, "step": 373 }, { "epoch": 0.3, "learning_rate": 1.9754810404459262e-05, "loss": 0.3601, "step": 374 }, { "epoch": 0.3, "learning_rate": 1.9752928523679145e-05, "loss": 0.3208, "step": 375 }, { "epoch": 0.3, "learning_rate": 1.9751039538896203e-05, "loss": 0.3304, "step": 376 }, { "epoch": 0.3, "learning_rate": 1.9749143451486363e-05, "loss": 0.3991, "step": 377 }, { "epoch": 0.3, "learning_rate": 1.9747240262830734e-05, "loss": 0.3754, "step": 378 }, { "epoch": 0.3, "learning_rate": 1.974532997431559e-05, "loss": 0.3935, "step": 379 }, { "epoch": 0.3, "learning_rate": 1.974341258733238e-05, "loss": 0.3035, "step": 380 }, { "epoch": 0.3, "learning_rate": 1.9741488103277722e-05, "loss": 0.3789, "step": 381 }, { "epoch": 0.3, "learning_rate": 1.9739556523553402e-05, "loss": 0.4083, "step": 382 }, { "epoch": 0.3, "learning_rate": 1.973761784956638e-05, "loss": 0.4145, "step": 383 }, { "epoch": 0.3, "learning_rate": 1.9735672082728785e-05, "loss": 0.3588, "step": 384 }, { "epoch": 0.3, "learning_rate": 1.9733719224457896e-05, "loss": 0.353, "step": 385 }, { "epoch": 0.31, "learning_rate": 1.9731759276176174e-05, "loss": 0.3775, "step": 386 }, { "epoch": 0.31, "learning_rate": 1.9729792239311243e-05, "loss": 0.349, "step": 387 }, { "epoch": 0.31, "learning_rate": 1.9727818115295882e-05, "loss": 0.3566, "step": 388 }, { "epoch": 0.31, "learning_rate": 1.972583690556804e-05, "loss": 0.3714, "step": 389 }, { "epoch": 0.31, "learning_rate": 1.972384861157082e-05, "loss": 0.3568, "step": 390 }, { "epoch": 0.31, "learning_rate": 1.972185323475249e-05, "loss": 0.3695, "step": 391 }, { "epoch": 0.31, "learning_rate": 1.971985077656648e-05, "loss": 0.3412, "step": 392 }, { "epoch": 0.31, "learning_rate": 1.9717841238471377e-05, "loss": 0.3204, "step": 393 }, { "epoch": 0.31, "learning_rate": 1.9715824621930907e-05, "loss": 0.3542, "step": 394 }, { "epoch": 0.31, "learning_rate": 1.9713800928413987e-05, "loss": 0.358, "step": 395 }, { "epoch": 0.31, "learning_rate": 1.9711770159394654e-05, "loss": 0.3406, "step": 396 }, { "epoch": 0.31, "learning_rate": 1.9709732316352118e-05, "loss": 0.3579, "step": 397 }, { "epoch": 0.31, "learning_rate": 1.9707687400770744e-05, "loss": 0.373, "step": 398 }, { "epoch": 0.32, "learning_rate": 1.9705635414140035e-05, "loss": 0.3316, "step": 399 }, { "epoch": 0.32, "learning_rate": 1.9703576357954653e-05, "loss": 0.3821, "step": 400 }, { "epoch": 0.32, "learning_rate": 1.970151023371441e-05, "loss": 0.3702, "step": 401 }, { "epoch": 0.32, "learning_rate": 1.9699437042924266e-05, "loss": 0.3485, "step": 402 }, { "epoch": 0.32, "learning_rate": 1.969735678709432e-05, "loss": 0.3448, "step": 403 }, { "epoch": 0.32, "learning_rate": 1.969526946773983e-05, "loss": 0.3574, "step": 404 }, { "epoch": 0.32, "learning_rate": 1.969317508638119e-05, "loss": 0.3613, "step": 405 }, { "epoch": 0.32, "learning_rate": 1.9691073644543942e-05, "loss": 0.3472, "step": 406 }, { "epoch": 0.32, "learning_rate": 1.9688965143758772e-05, "loss": 0.3837, "step": 407 }, { "epoch": 0.32, "learning_rate": 1.96868495855615e-05, "loss": 0.3382, "step": 408 }, { "epoch": 0.32, "learning_rate": 1.9684726971493095e-05, "loss": 0.342, "step": 409 }, { "epoch": 0.32, "learning_rate": 1.9682597303099663e-05, "loss": 0.3784, "step": 410 }, { "epoch": 0.32, "learning_rate": 1.9680460581932448e-05, "loss": 0.3443, "step": 411 }, { "epoch": 0.33, "learning_rate": 1.967831680954783e-05, "loss": 0.3521, "step": 412 }, { "epoch": 0.33, "learning_rate": 1.9676165987507328e-05, "loss": 0.3196, "step": 413 }, { "epoch": 0.33, "learning_rate": 1.967400811737759e-05, "loss": 0.3552, "step": 414 }, { "epoch": 0.33, "learning_rate": 1.9671843200730408e-05, "loss": 0.3262, "step": 415 }, { "epoch": 0.33, "learning_rate": 1.966967123914269e-05, "loss": 0.3516, "step": 416 }, { "epoch": 0.33, "learning_rate": 1.96674922341965e-05, "loss": 0.3436, "step": 417 }, { "epoch": 0.33, "learning_rate": 1.966530618747901e-05, "loss": 0.3292, "step": 418 }, { "epoch": 0.33, "learning_rate": 1.966311310058253e-05, "loss": 0.377, "step": 419 }, { "epoch": 0.33, "learning_rate": 1.96609129751045e-05, "loss": 0.3535, "step": 420 }, { "epoch": 0.33, "learning_rate": 1.965870581264748e-05, "loss": 0.3553, "step": 421 }, { "epoch": 0.33, "learning_rate": 1.965649161481917e-05, "loss": 0.3435, "step": 422 }, { "epoch": 0.33, "learning_rate": 1.9654270383232377e-05, "loss": 0.3297, "step": 423 }, { "epoch": 0.34, "learning_rate": 1.965204211950504e-05, "loss": 0.3391, "step": 424 }, { "epoch": 0.34, "learning_rate": 1.9649806825260215e-05, "loss": 0.3667, "step": 425 }, { "epoch": 0.34, "learning_rate": 1.9647564502126094e-05, "loss": 0.3948, "step": 426 }, { "epoch": 0.34, "learning_rate": 1.9645315151735972e-05, "loss": 0.3837, "step": 427 }, { "epoch": 0.34, "learning_rate": 1.9643058775728267e-05, "loss": 0.3236, "step": 428 }, { "epoch": 0.34, "learning_rate": 1.964079537574652e-05, "loss": 0.3496, "step": 429 }, { "epoch": 0.34, "learning_rate": 1.9638524953439385e-05, "loss": 0.3568, "step": 430 }, { "epoch": 0.34, "learning_rate": 1.9636247510460627e-05, "loss": 0.3631, "step": 431 }, { "epoch": 0.34, "learning_rate": 1.963396304846913e-05, "loss": 0.3505, "step": 432 }, { "epoch": 0.34, "learning_rate": 1.963167156912889e-05, "loss": 0.3399, "step": 433 }, { "epoch": 0.34, "learning_rate": 1.962937307410901e-05, "loss": 0.3386, "step": 434 }, { "epoch": 0.34, "learning_rate": 1.9627067565083716e-05, "loss": 0.3413, "step": 435 }, { "epoch": 0.34, "learning_rate": 1.962475504373232e-05, "loss": 0.3451, "step": 436 }, { "epoch": 0.35, "learning_rate": 1.9622435511739268e-05, "loss": 0.3305, "step": 437 }, { "epoch": 0.35, "learning_rate": 1.962010897079409e-05, "loss": 0.3397, "step": 438 }, { "epoch": 0.35, "learning_rate": 1.9617775422591438e-05, "loss": 0.3682, "step": 439 }, { "epoch": 0.35, "learning_rate": 1.9615434868831057e-05, "loss": 0.3394, "step": 440 }, { "epoch": 0.35, "learning_rate": 1.96130873112178e-05, "loss": 0.343, "step": 441 }, { "epoch": 0.35, "learning_rate": 1.9610732751461623e-05, "loss": 0.3322, "step": 442 }, { "epoch": 0.35, "learning_rate": 1.9608371191277577e-05, "loss": 0.381, "step": 443 }, { "epoch": 0.35, "learning_rate": 1.9606002632385817e-05, "loss": 0.3299, "step": 444 }, { "epoch": 0.35, "learning_rate": 1.9603627076511595e-05, "loss": 0.3501, "step": 445 }, { "epoch": 0.35, "learning_rate": 1.9601244525385255e-05, "loss": 0.2823, "step": 446 }, { "epoch": 0.35, "learning_rate": 1.959885498074224e-05, "loss": 0.3443, "step": 447 }, { "epoch": 0.35, "learning_rate": 1.9596458444323092e-05, "loss": 0.3401, "step": 448 }, { "epoch": 0.35, "learning_rate": 1.9594054917873437e-05, "loss": 0.3309, "step": 449 }, { "epoch": 0.36, "learning_rate": 1.9591644403143997e-05, "loss": 0.3329, "step": 450 }, { "epoch": 0.36, "learning_rate": 1.9589226901890582e-05, "loss": 0.3226, "step": 451 }, { "epoch": 0.36, "learning_rate": 1.9586802415874094e-05, "loss": 0.3656, "step": 452 }, { "epoch": 0.36, "learning_rate": 1.958437094686052e-05, "loss": 0.3769, "step": 453 }, { "epoch": 0.36, "learning_rate": 1.9581932496620933e-05, "loss": 0.3483, "step": 454 }, { "epoch": 0.36, "learning_rate": 1.9579487066931495e-05, "loss": 0.3062, "step": 455 }, { "epoch": 0.36, "learning_rate": 1.9577034659573452e-05, "loss": 0.3362, "step": 456 }, { "epoch": 0.36, "learning_rate": 1.9574575276333124e-05, "loss": 0.3537, "step": 457 }, { "epoch": 0.36, "learning_rate": 1.9572108919001922e-05, "loss": 0.343, "step": 458 }, { "epoch": 0.36, "learning_rate": 1.956963558937633e-05, "loss": 0.3989, "step": 459 }, { "epoch": 0.36, "learning_rate": 1.956715528925792e-05, "loss": 0.377, "step": 460 }, { "epoch": 0.36, "learning_rate": 1.9564668020453327e-05, "loss": 0.3948, "step": 461 }, { "epoch": 0.37, "learning_rate": 1.9562173784774274e-05, "loss": 0.3419, "step": 462 }, { "epoch": 0.37, "learning_rate": 1.9559672584037554e-05, "loss": 0.3932, "step": 463 }, { "epoch": 0.37, "learning_rate": 1.9557164420065032e-05, "loss": 0.3618, "step": 464 }, { "epoch": 0.37, "learning_rate": 1.955464929468365e-05, "loss": 0.3618, "step": 465 }, { "epoch": 0.37, "learning_rate": 1.9552127209725418e-05, "loss": 0.3642, "step": 466 }, { "epoch": 0.37, "learning_rate": 1.9549598167027407e-05, "loss": 0.3384, "step": 467 }, { "epoch": 0.37, "learning_rate": 1.9547062168431777e-05, "loss": 0.3185, "step": 468 }, { "epoch": 0.37, "learning_rate": 1.954451921578573e-05, "loss": 0.3542, "step": 469 }, { "epoch": 0.37, "learning_rate": 1.954196931094155e-05, "loss": 0.3606, "step": 470 }, { "epoch": 0.37, "learning_rate": 1.9539412455756578e-05, "loss": 0.3216, "step": 471 }, { "epoch": 0.37, "learning_rate": 1.953684865209322e-05, "loss": 0.3342, "step": 472 }, { "epoch": 0.37, "learning_rate": 1.9534277901818944e-05, "loss": 0.3256, "step": 473 }, { "epoch": 0.37, "learning_rate": 1.9531700206806274e-05, "loss": 0.3203, "step": 474 }, { "epoch": 0.38, "learning_rate": 1.9529115568932796e-05, "loss": 0.3271, "step": 475 }, { "epoch": 0.38, "learning_rate": 1.9526523990081155e-05, "loss": 0.3798, "step": 476 }, { "epoch": 0.38, "learning_rate": 1.952392547213904e-05, "loss": 0.3342, "step": 477 }, { "epoch": 0.38, "learning_rate": 1.952132001699922e-05, "loss": 0.3955, "step": 478 }, { "epoch": 0.38, "learning_rate": 1.9518707626559483e-05, "loss": 0.358, "step": 479 }, { "epoch": 0.38, "learning_rate": 1.9516088302722696e-05, "loss": 0.3522, "step": 480 }, { "epoch": 0.38, "learning_rate": 1.9513462047396763e-05, "loss": 0.3873, "step": 481 }, { "epoch": 0.38, "learning_rate": 1.9510828862494642e-05, "loss": 0.3352, "step": 482 }, { "epoch": 0.38, "learning_rate": 1.9508188749934333e-05, "loss": 0.3611, "step": 483 }, { "epoch": 0.38, "learning_rate": 1.9505541711638888e-05, "loss": 0.3469, "step": 484 }, { "epoch": 0.38, "learning_rate": 1.9502887749536406e-05, "loss": 0.3113, "step": 485 }, { "epoch": 0.38, "learning_rate": 1.9500226865560015e-05, "loss": 0.3847, "step": 486 }, { "epoch": 0.38, "learning_rate": 1.9497559061647907e-05, "loss": 0.3733, "step": 487 }, { "epoch": 0.39, "learning_rate": 1.949488433974329e-05, "loss": 0.3132, "step": 488 }, { "epoch": 0.39, "learning_rate": 1.9492202701794432e-05, "loss": 0.3463, "step": 489 }, { "epoch": 0.39, "learning_rate": 1.9489514149754624e-05, "loss": 0.3469, "step": 490 }, { "epoch": 0.39, "learning_rate": 1.9486818685582203e-05, "loss": 0.3394, "step": 491 }, { "epoch": 0.39, "learning_rate": 1.9484116311240534e-05, "loss": 0.2942, "step": 492 }, { "epoch": 0.39, "learning_rate": 1.9481407028698017e-05, "loss": 0.382, "step": 493 }, { "epoch": 0.39, "learning_rate": 1.947869083992809e-05, "loss": 0.3453, "step": 494 }, { "epoch": 0.39, "learning_rate": 1.9475967746909212e-05, "loss": 0.2924, "step": 495 }, { "epoch": 0.39, "learning_rate": 1.9473237751624877e-05, "loss": 0.3345, "step": 496 }, { "epoch": 0.39, "learning_rate": 1.9470500856063602e-05, "loss": 0.3102, "step": 497 }, { "epoch": 0.39, "learning_rate": 1.946775706221894e-05, "loss": 0.3255, "step": 498 }, { "epoch": 0.39, "learning_rate": 1.946500637208946e-05, "loss": 0.3503, "step": 499 }, { "epoch": 0.4, "learning_rate": 1.946224878767875e-05, "loss": 0.3114, "step": 500 }, { "epoch": 0.4, "learning_rate": 1.945948431099543e-05, "loss": 0.3436, "step": 501 }, { "epoch": 0.4, "learning_rate": 1.945671294405314e-05, "loss": 0.345, "step": 502 }, { "epoch": 0.4, "learning_rate": 1.9453934688870527e-05, "loss": 0.2926, "step": 503 }, { "epoch": 0.4, "learning_rate": 1.945114954747127e-05, "loss": 0.317, "step": 504 }, { "epoch": 0.4, "learning_rate": 1.9448357521884057e-05, "loss": 0.3688, "step": 505 }, { "epoch": 0.4, "learning_rate": 1.9445558614142584e-05, "loss": 0.3397, "step": 506 }, { "epoch": 0.4, "learning_rate": 1.9442752826285578e-05, "loss": 0.3032, "step": 507 }, { "epoch": 0.4, "learning_rate": 1.9439940160356757e-05, "loss": 0.3216, "step": 508 }, { "epoch": 0.4, "learning_rate": 1.943712061840486e-05, "loss": 0.3581, "step": 509 }, { "epoch": 0.4, "learning_rate": 1.9434294202483634e-05, "loss": 0.3459, "step": 510 }, { "epoch": 0.4, "learning_rate": 1.943146091465183e-05, "loss": 0.3187, "step": 511 }, { "epoch": 0.4, "learning_rate": 1.942862075697321e-05, "loss": 0.3245, "step": 512 }, { "epoch": 0.41, "learning_rate": 1.9425773731516534e-05, "loss": 0.346, "step": 513 }, { "epoch": 0.41, "learning_rate": 1.942291984035556e-05, "loss": 0.3262, "step": 514 }, { "epoch": 0.41, "learning_rate": 1.9420059085569062e-05, "loss": 0.3485, "step": 515 }, { "epoch": 0.41, "learning_rate": 1.9417191469240806e-05, "loss": 0.3546, "step": 516 }, { "epoch": 0.41, "learning_rate": 1.9414316993459545e-05, "loss": 0.3498, "step": 517 }, { "epoch": 0.41, "learning_rate": 1.941143566031905e-05, "loss": 0.334, "step": 518 }, { "epoch": 0.41, "learning_rate": 1.940854747191806e-05, "loss": 0.3385, "step": 519 }, { "epoch": 0.41, "learning_rate": 1.940565243036034e-05, "loss": 0.3941, "step": 520 }, { "epoch": 0.41, "learning_rate": 1.9402750537754618e-05, "loss": 0.3423, "step": 521 }, { "epoch": 0.41, "learning_rate": 1.9399841796214625e-05, "loss": 0.3752, "step": 522 }, { "epoch": 0.41, "learning_rate": 1.9396926207859085e-05, "loss": 0.3462, "step": 523 }, { "epoch": 0.41, "learning_rate": 1.9394003774811695e-05, "loss": 0.3597, "step": 524 }, { "epoch": 0.41, "learning_rate": 1.9391074499201155e-05, "loss": 0.3824, "step": 525 }, { "epoch": 0.42, "learning_rate": 1.9388138383161133e-05, "loss": 0.408, "step": 526 }, { "epoch": 0.42, "learning_rate": 1.938519542883029e-05, "loss": 0.3611, "step": 527 }, { "epoch": 0.42, "learning_rate": 1.938224563835226e-05, "loss": 0.3026, "step": 528 }, { "epoch": 0.42, "learning_rate": 1.9379289013875673e-05, "loss": 0.3176, "step": 529 }, { "epoch": 0.42, "learning_rate": 1.9376325557554113e-05, "loss": 0.332, "step": 530 }, { "epoch": 0.42, "learning_rate": 1.9373355271546156e-05, "loss": 0.3292, "step": 531 }, { "epoch": 0.42, "learning_rate": 1.937037815801535e-05, "loss": 0.3415, "step": 532 }, { "epoch": 0.42, "learning_rate": 1.936739421913022e-05, "loss": 0.3038, "step": 533 }, { "epoch": 0.42, "learning_rate": 1.9364403457064252e-05, "loss": 0.319, "step": 534 }, { "epoch": 0.42, "learning_rate": 1.9361405873995904e-05, "loss": 0.3131, "step": 535 }, { "epoch": 0.42, "learning_rate": 1.9358401472108613e-05, "loss": 0.3101, "step": 536 }, { "epoch": 0.42, "learning_rate": 1.9355390253590775e-05, "loss": 0.3367, "step": 537 }, { "epoch": 0.43, "learning_rate": 1.9352372220635753e-05, "loss": 0.3399, "step": 538 }, { "epoch": 0.43, "learning_rate": 1.9349347375441868e-05, "loss": 0.3122, "step": 539 }, { "epoch": 0.43, "learning_rate": 1.9346315720212416e-05, "loss": 0.3869, "step": 540 }, { "epoch": 0.43, "learning_rate": 1.9343277257155634e-05, "loss": 0.3258, "step": 541 }, { "epoch": 0.43, "learning_rate": 1.934023198848474e-05, "loss": 0.3768, "step": 542 }, { "epoch": 0.43, "learning_rate": 1.933717991641789e-05, "loss": 0.338, "step": 543 }, { "epoch": 0.43, "learning_rate": 1.933412104317821e-05, "loss": 0.3333, "step": 544 }, { "epoch": 0.43, "learning_rate": 1.933105537099377e-05, "loss": 0.3347, "step": 545 }, { "epoch": 0.43, "learning_rate": 1.9327982902097596e-05, "loss": 0.3554, "step": 546 }, { "epoch": 0.43, "learning_rate": 1.9324903638727664e-05, "loss": 0.3364, "step": 547 }, { "epoch": 0.43, "learning_rate": 1.9321817583126905e-05, "loss": 0.3231, "step": 548 }, { "epoch": 0.43, "learning_rate": 1.931872473754319e-05, "loss": 0.3376, "step": 549 }, { "epoch": 0.43, "learning_rate": 1.9315625104229336e-05, "loss": 0.3368, "step": 550 }, { "epoch": 0.44, "learning_rate": 1.9312518685443104e-05, "loss": 0.32, "step": 551 }, { "epoch": 0.44, "learning_rate": 1.9309405483447208e-05, "loss": 0.3163, "step": 552 }, { "epoch": 0.44, "learning_rate": 1.930628550050929e-05, "loss": 0.3496, "step": 553 }, { "epoch": 0.44, "learning_rate": 1.930315873890193e-05, "loss": 0.3234, "step": 554 }, { "epoch": 0.44, "learning_rate": 1.9300025200902666e-05, "loss": 0.3314, "step": 555 }, { "epoch": 0.44, "learning_rate": 1.929688488879394e-05, "loss": 0.3641, "step": 556 }, { "epoch": 0.44, "learning_rate": 1.929373780486316e-05, "loss": 0.3333, "step": 557 }, { "epoch": 0.44, "learning_rate": 1.9290583951402648e-05, "loss": 0.3224, "step": 558 }, { "epoch": 0.44, "learning_rate": 1.9287423330709653e-05, "loss": 0.3771, "step": 559 }, { "epoch": 0.44, "learning_rate": 1.928425594508637e-05, "loss": 0.3137, "step": 560 }, { "epoch": 0.44, "learning_rate": 1.9281081796839915e-05, "loss": 0.3371, "step": 561 }, { "epoch": 0.44, "learning_rate": 1.9277900888282314e-05, "loss": 0.3797, "step": 562 }, { "epoch": 0.44, "learning_rate": 1.9274713221730548e-05, "loss": 0.2982, "step": 563 }, { "epoch": 0.45, "learning_rate": 1.9271518799506494e-05, "loss": 0.3441, "step": 564 }, { "epoch": 0.45, "learning_rate": 1.9268317623936957e-05, "loss": 0.3754, "step": 565 }, { "epoch": 0.45, "learning_rate": 1.926510969735367e-05, "loss": 0.3079, "step": 566 }, { "epoch": 0.45, "learning_rate": 1.9261895022093275e-05, "loss": 0.3698, "step": 567 }, { "epoch": 0.45, "learning_rate": 1.9258673600497332e-05, "loss": 0.3456, "step": 568 }, { "epoch": 0.45, "learning_rate": 1.9255445434912308e-05, "loss": 0.3386, "step": 569 }, { "epoch": 0.45, "learning_rate": 1.9252210527689596e-05, "loss": 0.3397, "step": 570 }, { "epoch": 0.45, "learning_rate": 1.924896888118549e-05, "loss": 0.3378, "step": 571 }, { "epoch": 0.45, "learning_rate": 1.9245720497761194e-05, "loss": 0.3274, "step": 572 }, { "epoch": 0.45, "learning_rate": 1.9242465379782823e-05, "loss": 0.3453, "step": 573 }, { "epoch": 0.45, "learning_rate": 1.923920352962139e-05, "loss": 0.3609, "step": 574 }, { "epoch": 0.45, "learning_rate": 1.9235934949652825e-05, "loss": 0.2733, "step": 575 }, { "epoch": 0.46, "learning_rate": 1.9232659642257942e-05, "loss": 0.3219, "step": 576 }, { "epoch": 0.46, "learning_rate": 1.922937760982247e-05, "loss": 0.3374, "step": 577 }, { "epoch": 0.46, "learning_rate": 1.9226088854737033e-05, "loss": 0.3217, "step": 578 }, { "epoch": 0.46, "learning_rate": 1.9222793379397146e-05, "loss": 0.346, "step": 579 }, { "epoch": 0.46, "learning_rate": 1.9219491186203222e-05, "loss": 0.3423, "step": 580 }, { "epoch": 0.46, "learning_rate": 1.9216182277560572e-05, "loss": 0.3519, "step": 581 }, { "epoch": 0.46, "learning_rate": 1.9212866655879397e-05, "loss": 0.3613, "step": 582 }, { "epoch": 0.46, "learning_rate": 1.9209544323574782e-05, "loss": 0.2958, "step": 583 }, { "epoch": 0.46, "learning_rate": 1.92062152830667e-05, "loss": 0.3277, "step": 584 }, { "epoch": 0.46, "learning_rate": 1.9202879536780013e-05, "loss": 0.3658, "step": 585 }, { "epoch": 0.46, "learning_rate": 1.919953708714448e-05, "loss": 0.3364, "step": 586 }, { "epoch": 0.46, "learning_rate": 1.919618793659472e-05, "loss": 0.3185, "step": 587 }, { "epoch": 0.46, "learning_rate": 1.919283208757025e-05, "loss": 0.3644, "step": 588 }, { "epoch": 0.47, "learning_rate": 1.918946954251545e-05, "loss": 0.3682, "step": 589 }, { "epoch": 0.47, "learning_rate": 1.91861003038796e-05, "loss": 0.3253, "step": 590 }, { "epoch": 0.47, "learning_rate": 1.918272437411684e-05, "loss": 0.316, "step": 591 }, { "epoch": 0.47, "learning_rate": 1.917934175568618e-05, "loss": 0.3644, "step": 592 }, { "epoch": 0.47, "learning_rate": 1.9175952451051515e-05, "loss": 0.3273, "step": 593 }, { "epoch": 0.47, "learning_rate": 1.91725564626816e-05, "loss": 0.3137, "step": 594 }, { "epoch": 0.47, "learning_rate": 1.9169153793050065e-05, "loss": 0.3577, "step": 595 }, { "epoch": 0.47, "learning_rate": 1.9165744444635406e-05, "loss": 0.366, "step": 596 }, { "epoch": 0.47, "learning_rate": 1.9162328419920976e-05, "loss": 0.3743, "step": 597 }, { "epoch": 0.47, "learning_rate": 1.9158905721395e-05, "loss": 0.3219, "step": 598 }, { "epoch": 0.47, "learning_rate": 1.9155476351550563e-05, "loss": 0.3177, "step": 599 }, { "epoch": 0.47, "learning_rate": 1.9152040312885604e-05, "loss": 0.3568, "step": 600 }, { "epoch": 0.47, "learning_rate": 1.9148597607902922e-05, "loss": 0.3332, "step": 601 }, { "epoch": 0.48, "learning_rate": 1.914514823911018e-05, "loss": 0.3392, "step": 602 }, { "epoch": 0.48, "learning_rate": 1.914169220901988e-05, "loss": 0.3508, "step": 603 }, { "epoch": 0.48, "learning_rate": 1.9138229520149382e-05, "loss": 0.3453, "step": 604 }, { "epoch": 0.48, "learning_rate": 1.9134760175020906e-05, "loss": 0.3483, "step": 605 }, { "epoch": 0.48, "learning_rate": 1.9131284176161505e-05, "loss": 0.3508, "step": 606 }, { "epoch": 0.48, "learning_rate": 1.912780152610309e-05, "loss": 0.3424, "step": 607 }, { "epoch": 0.48, "learning_rate": 1.9124312227382412e-05, "loss": 0.358, "step": 608 }, { "epoch": 0.48, "learning_rate": 1.9120816282541062e-05, "loss": 0.3175, "step": 609 }, { "epoch": 0.48, "learning_rate": 1.9117313694125482e-05, "loss": 0.3114, "step": 610 }, { "epoch": 0.48, "learning_rate": 1.9113804464686938e-05, "loss": 0.3425, "step": 611 }, { "epoch": 0.48, "learning_rate": 1.911028859678155e-05, "loss": 0.329, "step": 612 }, { "epoch": 0.48, "learning_rate": 1.9106766092970257e-05, "loss": 0.3155, "step": 613 }, { "epoch": 0.49, "learning_rate": 1.9103236955818848e-05, "loss": 0.3342, "step": 614 }, { "epoch": 0.49, "learning_rate": 1.9099701187897927e-05, "loss": 0.309, "step": 615 }, { "epoch": 0.49, "learning_rate": 1.9096158791782943e-05, "loss": 0.2826, "step": 616 }, { "epoch": 0.49, "learning_rate": 1.9092609770054168e-05, "loss": 0.3254, "step": 617 }, { "epoch": 0.49, "learning_rate": 1.9089054125296692e-05, "loss": 0.3317, "step": 618 }, { "epoch": 0.49, "learning_rate": 1.9085491860100437e-05, "loss": 0.3372, "step": 619 }, { "epoch": 0.49, "learning_rate": 1.9081922977060146e-05, "loss": 0.3075, "step": 620 }, { "epoch": 0.49, "learning_rate": 1.907834747877539e-05, "loss": 0.3207, "step": 621 }, { "epoch": 0.49, "learning_rate": 1.9074765367850542e-05, "loss": 0.3201, "step": 622 }, { "epoch": 0.49, "learning_rate": 1.90711766468948e-05, "loss": 0.3644, "step": 623 }, { "epoch": 0.49, "learning_rate": 1.906758131852218e-05, "loss": 0.3034, "step": 624 }, { "epoch": 0.49, "learning_rate": 1.9063979385351512e-05, "loss": 0.2701, "step": 625 }, { "epoch": 0.49, "learning_rate": 1.9060370850006425e-05, "loss": 0.3137, "step": 626 }, { "epoch": 0.5, "learning_rate": 1.9056755715115372e-05, "loss": 0.3017, "step": 627 }, { "epoch": 0.5, "learning_rate": 1.9053133983311603e-05, "loss": 0.3647, "step": 628 }, { "epoch": 0.5, "learning_rate": 1.9049505657233177e-05, "loss": 0.329, "step": 629 }, { "epoch": 0.5, "learning_rate": 1.9045870739522953e-05, "loss": 0.3332, "step": 630 }, { "epoch": 0.5, "learning_rate": 1.9042229232828596e-05, "loss": 0.3217, "step": 631 }, { "epoch": 0.5, "learning_rate": 1.9038581139802567e-05, "loss": 0.3103, "step": 632 }, { "epoch": 0.5, "learning_rate": 1.9034926463102122e-05, "loss": 0.3289, "step": 633 }, { "epoch": 0.5, "learning_rate": 1.9031265205389328e-05, "loss": 0.3159, "step": 634 }, { "epoch": 0.5, "learning_rate": 1.902759736933102e-05, "loss": 0.2968, "step": 635 }, { "epoch": 0.5, "learning_rate": 1.9023922957598847e-05, "loss": 0.3669, "step": 636 }, { "epoch": 0.5, "learning_rate": 1.902024197286924e-05, "loss": 0.3091, "step": 637 }, { "epoch": 0.5, "learning_rate": 1.9016554417823404e-05, "loss": 0.3102, "step": 638 }, { "epoch": 0.5, "learning_rate": 1.901286029514736e-05, "loss": 0.3275, "step": 639 }, { "epoch": 0.51, "learning_rate": 1.9009159607531886e-05, "loss": 0.312, "step": 640 }, { "epoch": 0.51, "learning_rate": 1.900545235767255e-05, "loss": 0.3132, "step": 641 }, { "epoch": 0.51, "learning_rate": 1.9001738548269707e-05, "loss": 0.2904, "step": 642 }, { "epoch": 0.51, "learning_rate": 1.899801818202848e-05, "loss": 0.3036, "step": 643 }, { "epoch": 0.51, "learning_rate": 1.8994291261658775e-05, "loss": 0.2837, "step": 644 }, { "epoch": 0.51, "learning_rate": 1.8990557789875265e-05, "loss": 0.3256, "step": 645 }, { "epoch": 0.51, "learning_rate": 1.8986817769397406e-05, "loss": 0.3232, "step": 646 }, { "epoch": 0.51, "learning_rate": 1.898307120294941e-05, "loss": 0.3591, "step": 647 }, { "epoch": 0.51, "learning_rate": 1.8979318093260268e-05, "loss": 0.3142, "step": 648 }, { "epoch": 0.51, "learning_rate": 1.8975558443063734e-05, "loss": 0.3996, "step": 649 }, { "epoch": 0.51, "learning_rate": 1.8971792255098326e-05, "loss": 0.3462, "step": 650 }, { "epoch": 0.51, "learning_rate": 1.8968019532107318e-05, "loss": 0.3284, "step": 651 }, { "epoch": 0.52, "learning_rate": 1.896424027683876e-05, "loss": 0.3375, "step": 652 }, { "epoch": 0.52, "learning_rate": 1.896045449204544e-05, "loss": 0.3211, "step": 653 }, { "epoch": 0.52, "learning_rate": 1.8956662180484913e-05, "loss": 0.3334, "step": 654 }, { "epoch": 0.52, "learning_rate": 1.8952863344919495e-05, "loss": 0.3289, "step": 655 }, { "epoch": 0.52, "learning_rate": 1.8949057988116236e-05, "loss": 0.3092, "step": 656 }, { "epoch": 0.52, "learning_rate": 1.8945246112846952e-05, "loss": 0.3102, "step": 657 }, { "epoch": 0.52, "learning_rate": 1.8941427721888202e-05, "loss": 0.316, "step": 658 }, { "epoch": 0.52, "learning_rate": 1.8937602818021285e-05, "loss": 0.307, "step": 659 }, { "epoch": 0.52, "learning_rate": 1.893377140403225e-05, "loss": 0.2999, "step": 660 }, { "epoch": 0.52, "learning_rate": 1.8929933482711896e-05, "loss": 0.3128, "step": 661 }, { "epoch": 0.52, "learning_rate": 1.8926089056855743e-05, "loss": 0.3178, "step": 662 }, { "epoch": 0.52, "learning_rate": 1.892223812926406e-05, "loss": 0.3081, "step": 663 }, { "epoch": 0.52, "learning_rate": 1.891838070274185e-05, "loss": 0.3129, "step": 664 }, { "epoch": 0.53, "learning_rate": 1.891451678009886e-05, "loss": 0.3192, "step": 665 }, { "epoch": 0.53, "learning_rate": 1.8910646364149548e-05, "loss": 0.3125, "step": 666 }, { "epoch": 0.53, "learning_rate": 1.8906769457713117e-05, "loss": 0.2799, "step": 667 }, { "epoch": 0.53, "learning_rate": 1.8902886063613492e-05, "loss": 0.3601, "step": 668 }, { "epoch": 0.53, "learning_rate": 1.889899618467933e-05, "loss": 0.3058, "step": 669 }, { "epoch": 0.53, "learning_rate": 1.8895099823744005e-05, "loss": 0.3436, "step": 670 }, { "epoch": 0.53, "learning_rate": 1.889119698364561e-05, "loss": 0.3347, "step": 671 }, { "epoch": 0.53, "learning_rate": 1.8887287667226964e-05, "loss": 0.3202, "step": 672 }, { "epoch": 0.53, "learning_rate": 1.8883371877335602e-05, "loss": 0.3085, "step": 673 }, { "epoch": 0.53, "learning_rate": 1.8879449616823776e-05, "loss": 0.2993, "step": 674 }, { "epoch": 0.53, "learning_rate": 1.887552088854844e-05, "loss": 0.3608, "step": 675 }, { "epoch": 0.53, "learning_rate": 1.8871585695371278e-05, "loss": 0.2905, "step": 676 }, { "epoch": 0.53, "learning_rate": 1.886764404015866e-05, "loss": 0.3118, "step": 677 }, { "epoch": 0.54, "learning_rate": 1.8863695925781685e-05, "loss": 0.3212, "step": 678 }, { "epoch": 0.54, "learning_rate": 1.8859741355116148e-05, "loss": 0.3317, "step": 679 }, { "epoch": 0.54, "learning_rate": 1.8855780331042538e-05, "loss": 0.2864, "step": 680 }, { "epoch": 0.54, "learning_rate": 1.8851812856446062e-05, "loss": 0.3257, "step": 681 }, { "epoch": 0.54, "learning_rate": 1.8847838934216605e-05, "loss": 0.3512, "step": 682 }, { "epoch": 0.54, "learning_rate": 1.8843858567248772e-05, "loss": 0.3386, "step": 683 }, { "epoch": 0.54, "learning_rate": 1.8839871758441842e-05, "loss": 0.3251, "step": 684 }, { "epoch": 0.54, "learning_rate": 1.8835878510699793e-05, "loss": 0.3259, "step": 685 }, { "epoch": 0.54, "learning_rate": 1.8831878826931302e-05, "loss": 0.3282, "step": 686 }, { "epoch": 0.54, "learning_rate": 1.882787271004972e-05, "loss": 0.2953, "step": 687 }, { "epoch": 0.54, "learning_rate": 1.8823860162973088e-05, "loss": 0.3627, "step": 688 }, { "epoch": 0.54, "learning_rate": 1.881984118862414e-05, "loss": 0.2929, "step": 689 }, { "epoch": 0.55, "learning_rate": 1.8815815789930277e-05, "loss": 0.3333, "step": 690 }, { "epoch": 0.55, "learning_rate": 1.8811783969823595e-05, "loss": 0.3457, "step": 691 }, { "epoch": 0.55, "learning_rate": 1.8807745731240852e-05, "loss": 0.303, "step": 692 }, { "epoch": 0.55, "learning_rate": 1.8803701077123492e-05, "loss": 0.2742, "step": 693 }, { "epoch": 0.55, "learning_rate": 1.8799650010417627e-05, "loss": 0.36, "step": 694 }, { "epoch": 0.55, "learning_rate": 1.8795592534074045e-05, "loss": 0.3731, "step": 695 }, { "epoch": 0.55, "learning_rate": 1.8791528651048193e-05, "loss": 0.2656, "step": 696 }, { "epoch": 0.55, "learning_rate": 1.8787458364300198e-05, "loss": 0.3354, "step": 697 }, { "epoch": 0.55, "learning_rate": 1.878338167679484e-05, "loss": 0.3495, "step": 698 }, { "epoch": 0.55, "learning_rate": 1.8779298591501565e-05, "loss": 0.3647, "step": 699 }, { "epoch": 0.55, "learning_rate": 1.877520911139448e-05, "loss": 0.3411, "step": 700 }, { "epoch": 0.55, "learning_rate": 1.877111323945235e-05, "loss": 0.3065, "step": 701 }, { "epoch": 0.55, "learning_rate": 1.8767010978658597e-05, "loss": 0.3122, "step": 702 }, { "epoch": 0.56, "learning_rate": 1.876290233200129e-05, "loss": 0.2983, "step": 703 }, { "epoch": 0.56, "learning_rate": 1.875878730247316e-05, "loss": 0.3078, "step": 704 }, { "epoch": 0.56, "learning_rate": 1.8754665893071583e-05, "loss": 0.319, "step": 705 }, { "epoch": 0.56, "learning_rate": 1.8750538106798573e-05, "loss": 0.312, "step": 706 }, { "epoch": 0.56, "learning_rate": 1.8746403946660802e-05, "loss": 0.334, "step": 707 }, { "epoch": 0.56, "learning_rate": 1.874226341566958e-05, "loss": 0.2806, "step": 708 }, { "epoch": 0.56, "learning_rate": 1.8738116516840853e-05, "loss": 0.3315, "step": 709 }, { "epoch": 0.56, "learning_rate": 1.8733963253195217e-05, "loss": 0.2944, "step": 710 }, { "epoch": 0.56, "learning_rate": 1.872980362775789e-05, "loss": 0.3396, "step": 711 }, { "epoch": 0.56, "learning_rate": 1.8725637643558723e-05, "loss": 0.3185, "step": 712 }, { "epoch": 0.56, "learning_rate": 1.8721465303632222e-05, "loss": 0.2909, "step": 713 }, { "epoch": 0.56, "learning_rate": 1.87172866110175e-05, "loss": 0.3067, "step": 714 }, { "epoch": 0.56, "learning_rate": 1.8713101568758295e-05, "loss": 0.3155, "step": 715 }, { "epoch": 0.57, "learning_rate": 1.870891017990299e-05, "loss": 0.3094, "step": 716 }, { "epoch": 0.57, "learning_rate": 1.870471244750458e-05, "loss": 0.3108, "step": 717 }, { "epoch": 0.57, "learning_rate": 1.870050837462067e-05, "loss": 0.3495, "step": 718 }, { "epoch": 0.57, "learning_rate": 1.8696297964313503e-05, "loss": 0.3025, "step": 719 }, { "epoch": 0.57, "learning_rate": 1.8692081219649926e-05, "loss": 0.3109, "step": 720 }, { "epoch": 0.57, "learning_rate": 1.86878581437014e-05, "loss": 0.3424, "step": 721 }, { "epoch": 0.57, "learning_rate": 1.8683628739544004e-05, "loss": 0.3255, "step": 722 }, { "epoch": 0.57, "learning_rate": 1.867939301025842e-05, "loss": 0.3434, "step": 723 }, { "epoch": 0.57, "learning_rate": 1.867515095892994e-05, "loss": 0.285, "step": 724 }, { "epoch": 0.57, "learning_rate": 1.8670902588648467e-05, "loss": 0.3447, "step": 725 }, { "epoch": 0.57, "learning_rate": 1.8666647902508493e-05, "loss": 0.3648, "step": 726 }, { "epoch": 0.57, "learning_rate": 1.866238690360912e-05, "loss": 0.2935, "step": 727 }, { "epoch": 0.58, "learning_rate": 1.865811959505405e-05, "loss": 0.3519, "step": 728 }, { "epoch": 0.58, "learning_rate": 1.8653845979951577e-05, "loss": 0.3166, "step": 729 }, { "epoch": 0.58, "learning_rate": 1.8649566061414583e-05, "loss": 0.3048, "step": 730 }, { "epoch": 0.58, "learning_rate": 1.8645279842560554e-05, "loss": 0.3196, "step": 731 }, { "epoch": 0.58, "learning_rate": 1.864098732651155e-05, "loss": 0.2729, "step": 732 }, { "epoch": 0.58, "learning_rate": 1.863668851639424e-05, "loss": 0.3164, "step": 733 }, { "epoch": 0.58, "learning_rate": 1.8632383415339853e-05, "loss": 0.3069, "step": 734 }, { "epoch": 0.58, "learning_rate": 1.8628072026484215e-05, "loss": 0.3319, "step": 735 }, { "epoch": 0.58, "learning_rate": 1.8623754352967723e-05, "loss": 0.3142, "step": 736 }, { "epoch": 0.58, "learning_rate": 1.8619430397935366e-05, "loss": 0.3385, "step": 737 }, { "epoch": 0.58, "learning_rate": 1.8615100164536696e-05, "loss": 0.3163, "step": 738 }, { "epoch": 0.58, "learning_rate": 1.8610763655925838e-05, "loss": 0.3483, "step": 739 }, { "epoch": 0.58, "learning_rate": 1.8606420875261492e-05, "loss": 0.305, "step": 740 }, { "epoch": 0.59, "learning_rate": 1.8602071825706928e-05, "loss": 0.3235, "step": 741 }, { "epoch": 0.59, "learning_rate": 1.8597716510429986e-05, "loss": 0.3186, "step": 742 }, { "epoch": 0.59, "learning_rate": 1.8593354932603053e-05, "loss": 0.3211, "step": 743 }, { "epoch": 0.59, "learning_rate": 1.858898709540309e-05, "loss": 0.3147, "step": 744 }, { "epoch": 0.59, "learning_rate": 1.858461300201163e-05, "loss": 0.3385, "step": 745 }, { "epoch": 0.59, "learning_rate": 1.8580232655614732e-05, "loss": 0.3165, "step": 746 }, { "epoch": 0.59, "learning_rate": 1.8575846059403036e-05, "loss": 0.3224, "step": 747 }, { "epoch": 0.59, "learning_rate": 1.8571453216571723e-05, "loss": 0.3215, "step": 748 }, { "epoch": 0.59, "learning_rate": 1.856705413032052e-05, "loss": 0.281, "step": 749 }, { "epoch": 0.59, "learning_rate": 1.856264880385372e-05, "loss": 0.31, "step": 750 }, { "epoch": 0.59, "learning_rate": 1.855823724038014e-05, "loss": 0.2854, "step": 751 }, { "epoch": 0.59, "learning_rate": 1.855381944311315e-05, "loss": 0.3037, "step": 752 }, { "epoch": 0.6, "learning_rate": 1.8549395415270664e-05, "loss": 0.2828, "step": 753 }, { "epoch": 0.6, "learning_rate": 1.854496516007513e-05, "loss": 0.3373, "step": 754 }, { "epoch": 0.6, "learning_rate": 1.8540528680753525e-05, "loss": 0.334, "step": 755 }, { "epoch": 0.6, "learning_rate": 1.8536085980537374e-05, "loss": 0.3225, "step": 756 }, { "epoch": 0.6, "learning_rate": 1.8531637062662723e-05, "loss": 0.2998, "step": 757 }, { "epoch": 0.6, "learning_rate": 1.852718193037015e-05, "loss": 0.2926, "step": 758 }, { "epoch": 0.6, "learning_rate": 1.8522720586904758e-05, "loss": 0.3239, "step": 759 }, { "epoch": 0.6, "learning_rate": 1.851825303551618e-05, "loss": 0.337, "step": 760 }, { "epoch": 0.6, "learning_rate": 1.8513779279458566e-05, "loss": 0.2977, "step": 761 }, { "epoch": 0.6, "learning_rate": 1.8509299321990583e-05, "loss": 0.3537, "step": 762 }, { "epoch": 0.6, "learning_rate": 1.8504813166375416e-05, "loss": 0.3142, "step": 763 }, { "epoch": 0.6, "learning_rate": 1.8500320815880774e-05, "loss": 0.3409, "step": 764 }, { "epoch": 0.6, "learning_rate": 1.8495822273778867e-05, "loss": 0.2989, "step": 765 }, { "epoch": 0.61, "learning_rate": 1.8491317543346418e-05, "loss": 0.3644, "step": 766 }, { "epoch": 0.61, "learning_rate": 1.8486806627864663e-05, "loss": 0.3267, "step": 767 }, { "epoch": 0.61, "learning_rate": 1.8482289530619332e-05, "loss": 0.2944, "step": 768 }, { "epoch": 0.61, "learning_rate": 1.847776625490067e-05, "loss": 0.2709, "step": 769 }, { "epoch": 0.61, "learning_rate": 1.8473236804003412e-05, "loss": 0.3306, "step": 770 }, { "epoch": 0.61, "learning_rate": 1.8468701181226803e-05, "loss": 0.2924, "step": 771 }, { "epoch": 0.61, "learning_rate": 1.8464159389874574e-05, "loss": 0.2863, "step": 772 }, { "epoch": 0.61, "learning_rate": 1.845961143325495e-05, "loss": 0.3268, "step": 773 }, { "epoch": 0.61, "learning_rate": 1.8455057314680646e-05, "loss": 0.3482, "step": 774 }, { "epoch": 0.61, "learning_rate": 1.8450497037468876e-05, "loss": 0.2883, "step": 775 }, { "epoch": 0.61, "learning_rate": 1.8445930604941323e-05, "loss": 0.3125, "step": 776 }, { "epoch": 0.61, "learning_rate": 1.8441358020424168e-05, "loss": 0.3165, "step": 777 }, { "epoch": 0.61, "learning_rate": 1.843677928724806e-05, "loss": 0.3231, "step": 778 }, { "epoch": 0.62, "learning_rate": 1.8432194408748146e-05, "loss": 0.3524, "step": 779 }, { "epoch": 0.62, "learning_rate": 1.8427603388264027e-05, "loss": 0.2961, "step": 780 }, { "epoch": 0.62, "learning_rate": 1.842300622913979e-05, "loss": 0.2868, "step": 781 }, { "epoch": 0.62, "learning_rate": 1.8418402934723997e-05, "loss": 0.3128, "step": 782 }, { "epoch": 0.62, "learning_rate": 1.8413793508369667e-05, "loss": 0.2905, "step": 783 }, { "epoch": 0.62, "learning_rate": 1.8409177953434296e-05, "loss": 0.3218, "step": 784 }, { "epoch": 0.62, "learning_rate": 1.8404556273279835e-05, "loss": 0.2802, "step": 785 }, { "epoch": 0.62, "learning_rate": 1.839992847127271e-05, "loss": 0.2891, "step": 786 }, { "epoch": 0.62, "learning_rate": 1.8395294550783784e-05, "loss": 0.3274, "step": 787 }, { "epoch": 0.62, "learning_rate": 1.8390654515188404e-05, "loss": 0.3025, "step": 788 }, { "epoch": 0.62, "learning_rate": 1.838600836786635e-05, "loss": 0.2929, "step": 789 }, { "epoch": 0.62, "learning_rate": 1.8381356112201863e-05, "loss": 0.2758, "step": 790 }, { "epoch": 0.63, "learning_rate": 1.8376697751583635e-05, "loss": 0.2891, "step": 791 }, { "epoch": 0.63, "learning_rate": 1.8372033289404795e-05, "loss": 0.3274, "step": 792 }, { "epoch": 0.63, "learning_rate": 1.8367362729062924e-05, "loss": 0.3139, "step": 793 }, { "epoch": 0.63, "learning_rate": 1.8362686073960052e-05, "loss": 0.3289, "step": 794 }, { "epoch": 0.63, "learning_rate": 1.835800332750263e-05, "loss": 0.3041, "step": 795 }, { "epoch": 0.63, "learning_rate": 1.8353314493101556e-05, "loss": 0.32, "step": 796 }, { "epoch": 0.63, "learning_rate": 1.834861957417217e-05, "loss": 0.2886, "step": 797 }, { "epoch": 0.63, "learning_rate": 1.834391857413423e-05, "loss": 0.2684, "step": 798 }, { "epoch": 0.63, "learning_rate": 1.8339211496411935e-05, "loss": 0.2789, "step": 799 }, { "epoch": 0.63, "learning_rate": 1.8334498344433903e-05, "loss": 0.2735, "step": 800 }, { "epoch": 0.63, "learning_rate": 1.8329779121633177e-05, "loss": 0.2884, "step": 801 }, { "epoch": 0.63, "learning_rate": 1.8325053831447226e-05, "loss": 0.2897, "step": 802 }, { "epoch": 0.63, "learning_rate": 1.8320322477317938e-05, "loss": 0.3166, "step": 803 }, { "epoch": 0.64, "learning_rate": 1.8315585062691616e-05, "loss": 0.3173, "step": 804 }, { "epoch": 0.64, "learning_rate": 1.8310841591018977e-05, "loss": 0.3297, "step": 805 }, { "epoch": 0.64, "learning_rate": 1.8306092065755153e-05, "loss": 0.3107, "step": 806 }, { "epoch": 0.64, "learning_rate": 1.830133649035968e-05, "loss": 0.297, "step": 807 }, { "epoch": 0.64, "learning_rate": 1.829657486829651e-05, "loss": 0.3053, "step": 808 }, { "epoch": 0.64, "learning_rate": 1.8291807203033986e-05, "loss": 0.2639, "step": 809 }, { "epoch": 0.64, "learning_rate": 1.828703349804487e-05, "loss": 0.2803, "step": 810 }, { "epoch": 0.64, "learning_rate": 1.8282253756806304e-05, "loss": 0.2872, "step": 811 }, { "epoch": 0.64, "learning_rate": 1.8277467982799842e-05, "loss": 0.2822, "step": 812 }, { "epoch": 0.64, "learning_rate": 1.8272676179511428e-05, "loss": 0.2887, "step": 813 }, { "epoch": 0.64, "learning_rate": 1.8267878350431396e-05, "loss": 0.3044, "step": 814 }, { "epoch": 0.64, "learning_rate": 1.826307449905447e-05, "loss": 0.3195, "step": 815 }, { "epoch": 0.64, "learning_rate": 1.8258264628879753e-05, "loss": 0.3131, "step": 816 }, { "epoch": 0.65, "learning_rate": 1.825344874341075e-05, "loss": 0.2721, "step": 817 }, { "epoch": 0.65, "learning_rate": 1.824862684615533e-05, "loss": 0.3249, "step": 818 }, { "epoch": 0.65, "learning_rate": 1.8243798940625752e-05, "loss": 0.3079, "step": 819 }, { "epoch": 0.65, "learning_rate": 1.823896503033865e-05, "loss": 0.3191, "step": 820 }, { "epoch": 0.65, "learning_rate": 1.823412511881502e-05, "loss": 0.3235, "step": 821 }, { "epoch": 0.65, "learning_rate": 1.8229279209580245e-05, "loss": 0.3396, "step": 822 }, { "epoch": 0.65, "learning_rate": 1.822442730616407e-05, "loss": 0.2932, "step": 823 }, { "epoch": 0.65, "learning_rate": 1.8219569412100603e-05, "loss": 0.3117, "step": 824 }, { "epoch": 0.65, "learning_rate": 1.8214705530928322e-05, "loss": 0.2879, "step": 825 }, { "epoch": 0.65, "learning_rate": 1.8209835666190067e-05, "loss": 0.2545, "step": 826 }, { "epoch": 0.65, "learning_rate": 1.820495982143302e-05, "loss": 0.3451, "step": 827 }, { "epoch": 0.65, "learning_rate": 1.8200078000208745e-05, "loss": 0.2983, "step": 828 }, { "epoch": 0.66, "learning_rate": 1.8195190206073135e-05, "loss": 0.2944, "step": 829 }, { "epoch": 0.66, "learning_rate": 1.819029644258645e-05, "loss": 0.2977, "step": 830 }, { "epoch": 0.66, "learning_rate": 1.818539671331329e-05, "loss": 0.3208, "step": 831 }, { "epoch": 0.66, "learning_rate": 1.8180491021822603e-05, "loss": 0.2938, "step": 832 }, { "epoch": 0.66, "learning_rate": 1.8175579371687678e-05, "loss": 0.2912, "step": 833 }, { "epoch": 0.66, "learning_rate": 1.8170661766486147e-05, "loss": 0.3309, "step": 834 }, { "epoch": 0.66, "learning_rate": 1.816573820979998e-05, "loss": 0.3114, "step": 835 }, { "epoch": 0.66, "learning_rate": 1.8160808705215477e-05, "loss": 0.2755, "step": 836 }, { "epoch": 0.66, "learning_rate": 1.815587325632328e-05, "loss": 0.334, "step": 837 }, { "epoch": 0.66, "learning_rate": 1.8150931866718346e-05, "loss": 0.3034, "step": 838 }, { "epoch": 0.66, "learning_rate": 1.8145984539999972e-05, "loss": 0.2882, "step": 839 }, { "epoch": 0.66, "learning_rate": 1.8141031279771777e-05, "loss": 0.2931, "step": 840 }, { "epoch": 0.66, "learning_rate": 1.8136072089641698e-05, "loss": 0.284, "step": 841 }, { "epoch": 0.67, "learning_rate": 1.8131106973221994e-05, "loss": 0.3014, "step": 842 }, { "epoch": 0.67, "learning_rate": 1.812613593412924e-05, "loss": 0.2943, "step": 843 }, { "epoch": 0.67, "learning_rate": 1.812115897598433e-05, "loss": 0.2693, "step": 844 }, { "epoch": 0.67, "learning_rate": 1.811617610241246e-05, "loss": 0.3163, "step": 845 }, { "epoch": 0.67, "learning_rate": 1.8111187317043136e-05, "loss": 0.2924, "step": 846 }, { "epoch": 0.67, "learning_rate": 1.8106192623510183e-05, "loss": 0.3189, "step": 847 }, { "epoch": 0.67, "learning_rate": 1.8101192025451717e-05, "loss": 0.2701, "step": 848 }, { "epoch": 0.67, "learning_rate": 1.8096185526510154e-05, "loss": 0.3026, "step": 849 }, { "epoch": 0.67, "learning_rate": 1.8091173130332214e-05, "loss": 0.3246, "step": 850 }, { "epoch": 0.67, "learning_rate": 1.8086154840568914e-05, "loss": 0.3024, "step": 851 }, { "epoch": 0.67, "learning_rate": 1.8081130660875557e-05, "loss": 0.3033, "step": 852 }, { "epoch": 0.67, "learning_rate": 1.8076100594911742e-05, "loss": 0.2994, "step": 853 }, { "epoch": 0.67, "learning_rate": 1.8071064646341354e-05, "loss": 0.2576, "step": 854 }, { "epoch": 0.68, "learning_rate": 1.8066022818832564e-05, "loss": 0.3291, "step": 855 }, { "epoch": 0.68, "learning_rate": 1.8060975116057815e-05, "loss": 0.2738, "step": 856 }, { "epoch": 0.68, "learning_rate": 1.8055921541693845e-05, "loss": 0.269, "step": 857 }, { "epoch": 0.68, "learning_rate": 1.805086209942166e-05, "loss": 0.2887, "step": 858 }, { "epoch": 0.68, "learning_rate": 1.8045796792926548e-05, "loss": 0.307, "step": 859 }, { "epoch": 0.68, "learning_rate": 1.804072562589805e-05, "loss": 0.3316, "step": 860 }, { "epoch": 0.68, "learning_rate": 1.8035648602029997e-05, "loss": 0.2798, "step": 861 }, { "epoch": 0.68, "learning_rate": 1.8030565725020478e-05, "loss": 0.3074, "step": 862 }, { "epoch": 0.68, "learning_rate": 1.802547699857184e-05, "loss": 0.2992, "step": 863 }, { "epoch": 0.68, "learning_rate": 1.80203824263907e-05, "loss": 0.3191, "step": 864 }, { "epoch": 0.68, "learning_rate": 1.8015282012187927e-05, "loss": 0.3058, "step": 865 }, { "epoch": 0.68, "learning_rate": 1.8010175759678647e-05, "loss": 0.2799, "step": 866 }, { "epoch": 0.69, "learning_rate": 1.8005063672582236e-05, "loss": 0.3417, "step": 867 }, { "epoch": 0.69, "learning_rate": 1.799994575462232e-05, "loss": 0.3056, "step": 868 }, { "epoch": 0.69, "learning_rate": 1.7994822009526786e-05, "loss": 0.2844, "step": 869 }, { "epoch": 0.69, "learning_rate": 1.7989692441027744e-05, "loss": 0.3078, "step": 870 }, { "epoch": 0.69, "learning_rate": 1.798455705286156e-05, "loss": 0.2939, "step": 871 }, { "epoch": 0.69, "learning_rate": 1.7979415848768833e-05, "loss": 0.2426, "step": 872 }, { "epoch": 0.69, "learning_rate": 1.7974268832494397e-05, "loss": 0.3007, "step": 873 }, { "epoch": 0.69, "learning_rate": 1.796911600778733e-05, "loss": 0.3284, "step": 874 }, { "epoch": 0.69, "learning_rate": 1.796395737840093e-05, "loss": 0.2695, "step": 875 }, { "epoch": 0.69, "learning_rate": 1.7958792948092726e-05, "loss": 0.2628, "step": 876 }, { "epoch": 0.69, "learning_rate": 1.7953622720624468e-05, "loss": 0.2638, "step": 877 }, { "epoch": 0.69, "learning_rate": 1.7948446699762146e-05, "loss": 0.2914, "step": 878 }, { "epoch": 0.69, "learning_rate": 1.7943264889275944e-05, "loss": 0.3276, "step": 879 }, { "epoch": 0.7, "learning_rate": 1.7938077292940288e-05, "loss": 0.3311, "step": 880 }, { "epoch": 0.7, "learning_rate": 1.79328839145338e-05, "loss": 0.2806, "step": 881 }, { "epoch": 0.7, "learning_rate": 1.792768475783932e-05, "loss": 0.3044, "step": 882 }, { "epoch": 0.7, "learning_rate": 1.7922479826643905e-05, "loss": 0.2913, "step": 883 }, { "epoch": 0.7, "learning_rate": 1.7917269124738802e-05, "loss": 0.3121, "step": 884 }, { "epoch": 0.7, "learning_rate": 1.7912052655919478e-05, "loss": 0.2749, "step": 885 }, { "epoch": 0.7, "learning_rate": 1.790683042398559e-05, "loss": 0.3111, "step": 886 }, { "epoch": 0.7, "learning_rate": 1.790160243274099e-05, "loss": 0.3021, "step": 887 }, { "epoch": 0.7, "learning_rate": 1.7896368685993738e-05, "loss": 0.2728, "step": 888 }, { "epoch": 0.7, "learning_rate": 1.789112918755607e-05, "loss": 0.2894, "step": 889 }, { "epoch": 0.7, "learning_rate": 1.7885883941244432e-05, "loss": 0.3205, "step": 890 }, { "epoch": 0.7, "learning_rate": 1.7880632950879438e-05, "loss": 0.2786, "step": 891 }, { "epoch": 0.7, "learning_rate": 1.7875376220285893e-05, "loss": 0.3162, "step": 892 }, { "epoch": 0.71, "learning_rate": 1.7870113753292782e-05, "loss": 0.337, "step": 893 }, { "epoch": 0.71, "learning_rate": 1.7864845553733276e-05, "loss": 0.3079, "step": 894 }, { "epoch": 0.71, "learning_rate": 1.7859571625444712e-05, "loss": 0.3135, "step": 895 }, { "epoch": 0.71, "learning_rate": 1.78542919722686e-05, "loss": 0.2483, "step": 896 }, { "epoch": 0.71, "learning_rate": 1.7849006598050626e-05, "loss": 0.2862, "step": 897 }, { "epoch": 0.71, "learning_rate": 1.784371550664064e-05, "loss": 0.2834, "step": 898 }, { "epoch": 0.71, "learning_rate": 1.7838418701892653e-05, "loss": 0.2903, "step": 899 }, { "epoch": 0.71, "learning_rate": 1.7833116187664846e-05, "loss": 0.3196, "step": 900 }, { "epoch": 0.71, "learning_rate": 1.7827807967819556e-05, "loss": 0.2645, "step": 901 }, { "epoch": 0.71, "learning_rate": 1.7822494046223263e-05, "loss": 0.3361, "step": 902 }, { "epoch": 0.71, "learning_rate": 1.781717442674662e-05, "loss": 0.3428, "step": 903 }, { "epoch": 0.71, "learning_rate": 1.781184911326442e-05, "loss": 0.2917, "step": 904 }, { "epoch": 0.72, "learning_rate": 1.7806518109655604e-05, "loss": 0.3138, "step": 905 }, { "epoch": 0.72, "learning_rate": 1.7801181419803257e-05, "loss": 0.2921, "step": 906 }, { "epoch": 0.72, "learning_rate": 1.779583904759461e-05, "loss": 0.2629, "step": 907 }, { "epoch": 0.72, "learning_rate": 1.779049099692103e-05, "loss": 0.2903, "step": 908 }, { "epoch": 0.72, "learning_rate": 1.7785137271678013e-05, "loss": 0.2912, "step": 909 }, { "epoch": 0.72, "learning_rate": 1.777977787576521e-05, "loss": 0.2958, "step": 910 }, { "epoch": 0.72, "learning_rate": 1.7774412813086377e-05, "loss": 0.3172, "step": 911 }, { "epoch": 0.72, "learning_rate": 1.776904208754941e-05, "loss": 0.3054, "step": 912 }, { "epoch": 0.72, "learning_rate": 1.7763665703066337e-05, "loss": 0.2918, "step": 913 }, { "epoch": 0.72, "learning_rate": 1.775828366355329e-05, "loss": 0.3138, "step": 914 }, { "epoch": 0.72, "learning_rate": 1.7752895972930538e-05, "loss": 0.2726, "step": 915 }, { "epoch": 0.72, "learning_rate": 1.774750263512245e-05, "loss": 0.3006, "step": 916 }, { "epoch": 0.72, "learning_rate": 1.774210365405752e-05, "loss": 0.2968, "step": 917 }, { "epoch": 0.73, "learning_rate": 1.7736699033668353e-05, "loss": 0.2714, "step": 918 }, { "epoch": 0.73, "learning_rate": 1.773128877789165e-05, "loss": 0.315, "step": 919 }, { "epoch": 0.73, "learning_rate": 1.772587289066823e-05, "loss": 0.3122, "step": 920 }, { "epoch": 0.73, "learning_rate": 1.772045137594301e-05, "loss": 0.3255, "step": 921 }, { "epoch": 0.73, "learning_rate": 1.7715024237665002e-05, "loss": 0.2885, "step": 922 }, { "epoch": 0.73, "learning_rate": 1.770959147978731e-05, "loss": 0.2728, "step": 923 }, { "epoch": 0.73, "learning_rate": 1.770415310626715e-05, "loss": 0.273, "step": 924 }, { "epoch": 0.73, "learning_rate": 1.769870912106581e-05, "loss": 0.2852, "step": 925 }, { "epoch": 0.73, "learning_rate": 1.769325952814867e-05, "loss": 0.2898, "step": 926 }, { "epoch": 0.73, "learning_rate": 1.7687804331485203e-05, "loss": 0.3019, "step": 927 }, { "epoch": 0.73, "learning_rate": 1.768234353504895e-05, "loss": 0.3073, "step": 928 }, { "epoch": 0.73, "learning_rate": 1.7676877142817545e-05, "loss": 0.3082, "step": 929 }, { "epoch": 0.73, "learning_rate": 1.7671405158772686e-05, "loss": 0.2733, "step": 930 }, { "epoch": 0.74, "learning_rate": 1.7665927586900147e-05, "loss": 0.324, "step": 931 }, { "epoch": 0.74, "learning_rate": 1.766044443118978e-05, "loss": 0.3088, "step": 932 }, { "epoch": 0.74, "learning_rate": 1.7654955695635498e-05, "loss": 0.2817, "step": 933 }, { "epoch": 0.74, "learning_rate": 1.764946138423528e-05, "loss": 0.2948, "step": 934 }, { "epoch": 0.74, "learning_rate": 1.764396150099116e-05, "loss": 0.311, "step": 935 }, { "epoch": 0.74, "learning_rate": 1.7638456049909238e-05, "loss": 0.2958, "step": 936 }, { "epoch": 0.74, "learning_rate": 1.7632945034999667e-05, "loss": 0.3162, "step": 937 }, { "epoch": 0.74, "learning_rate": 1.762742846027666e-05, "loss": 0.3187, "step": 938 }, { "epoch": 0.74, "learning_rate": 1.7621906329758466e-05, "loss": 0.2938, "step": 939 }, { "epoch": 0.74, "learning_rate": 1.7616378647467387e-05, "loss": 0.2823, "step": 940 }, { "epoch": 0.74, "learning_rate": 1.761084541742977e-05, "loss": 0.3533, "step": 941 }, { "epoch": 0.74, "learning_rate": 1.760530664367601e-05, "loss": 0.2958, "step": 942 }, { "epoch": 0.75, "learning_rate": 1.759976233024052e-05, "loss": 0.3302, "step": 943 }, { "epoch": 0.75, "learning_rate": 1.759421248116177e-05, "loss": 0.2889, "step": 944 }, { "epoch": 0.75, "learning_rate": 1.758865710048225e-05, "loss": 0.293, "step": 945 }, { "epoch": 0.75, "learning_rate": 1.7583096192248483e-05, "loss": 0.3002, "step": 946 }, { "epoch": 0.75, "learning_rate": 1.7577529760511017e-05, "loss": 0.2997, "step": 947 }, { "epoch": 0.75, "learning_rate": 1.7571957809324422e-05, "loss": 0.2838, "step": 948 }, { "epoch": 0.75, "learning_rate": 1.756638034274729e-05, "loss": 0.2958, "step": 949 }, { "epoch": 0.75, "learning_rate": 1.7560797364842235e-05, "loss": 0.2914, "step": 950 }, { "epoch": 0.75, "learning_rate": 1.7555208879675875e-05, "loss": 0.2884, "step": 951 }, { "epoch": 0.75, "learning_rate": 1.7549614891318846e-05, "loss": 0.2786, "step": 952 }, { "epoch": 0.75, "learning_rate": 1.7544015403845795e-05, "loss": 0.2895, "step": 953 }, { "epoch": 0.75, "learning_rate": 1.7538410421335373e-05, "loss": 0.287, "step": 954 }, { "epoch": 0.75, "learning_rate": 1.7532799947870224e-05, "loss": 0.2825, "step": 955 }, { "epoch": 0.76, "learning_rate": 1.7527183987537e-05, "loss": 0.2735, "step": 956 }, { "epoch": 0.76, "learning_rate": 1.752156254442636e-05, "loss": 0.3212, "step": 957 }, { "epoch": 0.76, "learning_rate": 1.7515935622632932e-05, "loss": 0.3028, "step": 958 }, { "epoch": 0.76, "learning_rate": 1.7510303226255357e-05, "loss": 0.2924, "step": 959 }, { "epoch": 0.76, "learning_rate": 1.7504665359396255e-05, "loss": 0.2816, "step": 960 }, { "epoch": 0.76, "learning_rate": 1.7499022026162225e-05, "loss": 0.3046, "step": 961 }, { "epoch": 0.76, "learning_rate": 1.749337323066386e-05, "loss": 0.2764, "step": 962 }, { "epoch": 0.76, "learning_rate": 1.748771897701572e-05, "loss": 0.2904, "step": 963 }, { "epoch": 0.76, "learning_rate": 1.7482059269336353e-05, "loss": 0.2826, "step": 964 }, { "epoch": 0.76, "learning_rate": 1.7476394111748262e-05, "loss": 0.3295, "step": 965 }, { "epoch": 0.76, "learning_rate": 1.7470723508377935e-05, "loss": 0.2856, "step": 966 }, { "epoch": 0.76, "learning_rate": 1.7465047463355823e-05, "loss": 0.286, "step": 967 }, { "epoch": 0.76, "learning_rate": 1.745936598081634e-05, "loss": 0.2839, "step": 968 }, { "epoch": 0.77, "learning_rate": 1.745367906489786e-05, "loss": 0.2711, "step": 969 }, { "epoch": 0.77, "learning_rate": 1.7447986719742708e-05, "loss": 0.2767, "step": 970 }, { "epoch": 0.77, "learning_rate": 1.7442288949497182e-05, "loss": 0.2345, "step": 971 }, { "epoch": 0.77, "learning_rate": 1.7436585758311512e-05, "loss": 0.3018, "step": 972 }, { "epoch": 0.77, "learning_rate": 1.743087715033989e-05, "loss": 0.3196, "step": 973 }, { "epoch": 0.77, "learning_rate": 1.7425163129740442e-05, "loss": 0.2649, "step": 974 }, { "epoch": 0.77, "learning_rate": 1.7419443700675248e-05, "loss": 0.3227, "step": 975 }, { "epoch": 0.77, "learning_rate": 1.7413718867310325e-05, "loss": 0.2748, "step": 976 }, { "epoch": 0.77, "learning_rate": 1.7407988633815612e-05, "loss": 0.3009, "step": 977 }, { "epoch": 0.77, "learning_rate": 1.7402253004365007e-05, "loss": 0.2766, "step": 978 }, { "epoch": 0.77, "learning_rate": 1.7396511983136317e-05, "loss": 0.2898, "step": 979 }, { "epoch": 0.77, "learning_rate": 1.7390765574311287e-05, "loss": 0.2832, "step": 980 }, { "epoch": 0.78, "learning_rate": 1.7385013782075575e-05, "loss": 0.2741, "step": 981 }, { "epoch": 0.78, "learning_rate": 1.7379256610618785e-05, "loss": 0.2774, "step": 982 }, { "epoch": 0.78, "learning_rate": 1.737349406413441e-05, "loss": 0.2392, "step": 983 }, { "epoch": 0.78, "learning_rate": 1.736772614681987e-05, "loss": 0.2824, "step": 984 }, { "epoch": 0.78, "learning_rate": 1.7361952862876505e-05, "loss": 0.3016, "step": 985 }, { "epoch": 0.78, "learning_rate": 1.7356174216509557e-05, "loss": 0.2915, "step": 986 }, { "epoch": 0.78, "learning_rate": 1.7350390211928167e-05, "loss": 0.2981, "step": 987 }, { "epoch": 0.78, "learning_rate": 1.7344600853345396e-05, "loss": 0.313, "step": 988 }, { "epoch": 0.78, "learning_rate": 1.7338806144978183e-05, "loss": 0.3087, "step": 989 }, { "epoch": 0.78, "learning_rate": 1.7333006091047386e-05, "loss": 0.2487, "step": 990 }, { "epoch": 0.78, "learning_rate": 1.732720069577774e-05, "loss": 0.3059, "step": 991 }, { "epoch": 0.78, "learning_rate": 1.7321389963397883e-05, "loss": 0.2871, "step": 992 }, { "epoch": 0.78, "learning_rate": 1.7315573898140324e-05, "loss": 0.2831, "step": 993 }, { "epoch": 0.79, "learning_rate": 1.730975250424148e-05, "loss": 0.2765, "step": 994 }, { "epoch": 0.79, "learning_rate": 1.730392578594162e-05, "loss": 0.2793, "step": 995 }, { "epoch": 0.79, "learning_rate": 1.7298093747484923e-05, "loss": 0.2769, "step": 996 }, { "epoch": 0.79, "learning_rate": 1.7292256393119422e-05, "loss": 0.3171, "step": 997 }, { "epoch": 0.79, "learning_rate": 1.7286413727097026e-05, "loss": 0.2765, "step": 998 }, { "epoch": 0.79, "learning_rate": 1.7280565753673517e-05, "loss": 0.2765, "step": 999 }, { "epoch": 0.79, "learning_rate": 1.7274712477108538e-05, "loss": 0.2994, "step": 1000 }, { "epoch": 0.79, "learning_rate": 1.7268853901665605e-05, "loss": 0.3068, "step": 1001 }, { "epoch": 0.79, "learning_rate": 1.7262990031612072e-05, "loss": 0.2397, "step": 1002 }, { "epoch": 0.79, "learning_rate": 1.7257120871219176e-05, "loss": 0.2978, "step": 1003 }, { "epoch": 0.79, "learning_rate": 1.7251246424761992e-05, "loss": 0.3042, "step": 1004 }, { "epoch": 0.79, "learning_rate": 1.7245366696519448e-05, "loss": 0.3214, "step": 1005 }, { "epoch": 0.79, "learning_rate": 1.723948169077432e-05, "loss": 0.3069, "step": 1006 }, { "epoch": 0.8, "learning_rate": 1.7233591411813225e-05, "loss": 0.2733, "step": 1007 }, { "epoch": 0.8, "learning_rate": 1.7227695863926627e-05, "loss": 0.2914, "step": 1008 }, { "epoch": 0.8, "learning_rate": 1.7221795051408826e-05, "loss": 0.2797, "step": 1009 }, { "epoch": 0.8, "learning_rate": 1.7215888978557953e-05, "loss": 0.2844, "step": 1010 }, { "epoch": 0.8, "learning_rate": 1.7209977649675975e-05, "loss": 0.3068, "step": 1011 }, { "epoch": 0.8, "learning_rate": 1.7204061069068683e-05, "loss": 0.2938, "step": 1012 }, { "epoch": 0.8, "learning_rate": 1.7198139241045696e-05, "loss": 0.3032, "step": 1013 }, { "epoch": 0.8, "learning_rate": 1.7192212169920458e-05, "loss": 0.2854, "step": 1014 }, { "epoch": 0.8, "learning_rate": 1.7186279860010228e-05, "loss": 0.2744, "step": 1015 }, { "epoch": 0.8, "learning_rate": 1.718034231563608e-05, "loss": 0.292, "step": 1016 }, { "epoch": 0.8, "learning_rate": 1.717439954112291e-05, "loss": 0.2947, "step": 1017 }, { "epoch": 0.8, "learning_rate": 1.7168451540799403e-05, "loss": 0.2529, "step": 1018 }, { "epoch": 0.81, "learning_rate": 1.7162498318998078e-05, "loss": 0.253, "step": 1019 }, { "epoch": 0.81, "learning_rate": 1.7156539880055236e-05, "loss": 0.2767, "step": 1020 }, { "epoch": 0.81, "learning_rate": 1.715057622831098e-05, "loss": 0.2704, "step": 1021 }, { "epoch": 0.81, "learning_rate": 1.714460736810923e-05, "loss": 0.3043, "step": 1022 }, { "epoch": 0.81, "learning_rate": 1.7138633303797676e-05, "loss": 0.2586, "step": 1023 }, { "epoch": 0.81, "learning_rate": 1.713265403972781e-05, "loss": 0.2819, "step": 1024 }, { "epoch": 0.81, "learning_rate": 1.7126669580254908e-05, "loss": 0.3022, "step": 1025 }, { "epoch": 0.81, "learning_rate": 1.712067992973803e-05, "loss": 0.276, "step": 1026 }, { "epoch": 0.81, "learning_rate": 1.7114685092540023e-05, "loss": 0.309, "step": 1027 }, { "epoch": 0.81, "learning_rate": 1.71086850730275e-05, "loss": 0.2687, "step": 1028 }, { "epoch": 0.81, "learning_rate": 1.710267987557087e-05, "loss": 0.3023, "step": 1029 }, { "epoch": 0.81, "learning_rate": 1.7096669504544293e-05, "loss": 0.3084, "step": 1030 }, { "epoch": 0.81, "learning_rate": 1.70906539643257e-05, "loss": 0.2926, "step": 1031 }, { "epoch": 0.82, "learning_rate": 1.7084633259296798e-05, "loss": 0.2766, "step": 1032 }, { "epoch": 0.82, "learning_rate": 1.7078607393843048e-05, "loss": 0.2687, "step": 1033 }, { "epoch": 0.82, "learning_rate": 1.707257637235367e-05, "loss": 0.259, "step": 1034 }, { "epoch": 0.82, "learning_rate": 1.706654019922164e-05, "loss": 0.2829, "step": 1035 }, { "epoch": 0.82, "learning_rate": 1.7060498878843695e-05, "loss": 0.2915, "step": 1036 }, { "epoch": 0.82, "learning_rate": 1.705445241562031e-05, "loss": 0.3061, "step": 1037 }, { "epoch": 0.82, "learning_rate": 1.704840081395571e-05, "loss": 0.2641, "step": 1038 }, { "epoch": 0.82, "learning_rate": 1.7042344078257857e-05, "loss": 0.2558, "step": 1039 }, { "epoch": 0.82, "learning_rate": 1.7036282212938468e-05, "loss": 0.286, "step": 1040 }, { "epoch": 0.82, "learning_rate": 1.703021522241298e-05, "loss": 0.2813, "step": 1041 }, { "epoch": 0.82, "learning_rate": 1.7024143111100575e-05, "loss": 0.3093, "step": 1042 }, { "epoch": 0.82, "learning_rate": 1.701806588342416e-05, "loss": 0.3023, "step": 1043 }, { "epoch": 0.82, "learning_rate": 1.701198354381036e-05, "loss": 0.2735, "step": 1044 }, { "epoch": 0.83, "learning_rate": 1.7005896096689544e-05, "loss": 0.3101, "step": 1045 }, { "epoch": 0.83, "learning_rate": 1.6999803546495782e-05, "loss": 0.278, "step": 1046 }, { "epoch": 0.83, "learning_rate": 1.6993705897666873e-05, "loss": 0.2844, "step": 1047 }, { "epoch": 0.83, "learning_rate": 1.698760315464432e-05, "loss": 0.2786, "step": 1048 }, { "epoch": 0.83, "learning_rate": 1.6981495321873346e-05, "loss": 0.2435, "step": 1049 }, { "epoch": 0.83, "learning_rate": 1.697538240380288e-05, "loss": 0.2416, "step": 1050 }, { "epoch": 0.83, "learning_rate": 1.696926440488555e-05, "loss": 0.2796, "step": 1051 }, { "epoch": 0.83, "learning_rate": 1.6963141329577683e-05, "loss": 0.2913, "step": 1052 }, { "epoch": 0.83, "learning_rate": 1.695701318233931e-05, "loss": 0.2647, "step": 1053 }, { "epoch": 0.83, "learning_rate": 1.6950879967634166e-05, "loss": 0.3075, "step": 1054 }, { "epoch": 0.83, "learning_rate": 1.6944741689929646e-05, "loss": 0.2907, "step": 1055 }, { "epoch": 0.83, "learning_rate": 1.6938598353696864e-05, "loss": 0.2704, "step": 1056 }, { "epoch": 0.84, "learning_rate": 1.693244996341061e-05, "loss": 0.2854, "step": 1057 }, { "epoch": 0.84, "learning_rate": 1.6926296523549343e-05, "loss": 0.2859, "step": 1058 }, { "epoch": 0.84, "learning_rate": 1.6920138038595214e-05, "loss": 0.2577, "step": 1059 }, { "epoch": 0.84, "learning_rate": 1.6913974513034046e-05, "loss": 0.2859, "step": 1060 }, { "epoch": 0.84, "learning_rate": 1.6907805951355332e-05, "loss": 0.2729, "step": 1061 }, { "epoch": 0.84, "learning_rate": 1.6901632358052226e-05, "loss": 0.2677, "step": 1062 }, { "epoch": 0.84, "learning_rate": 1.6895453737621562e-05, "loss": 0.3135, "step": 1063 }, { "epoch": 0.84, "learning_rate": 1.6889270094563828e-05, "loss": 0.2581, "step": 1064 }, { "epoch": 0.84, "learning_rate": 1.6883081433383163e-05, "loss": 0.2897, "step": 1065 }, { "epoch": 0.84, "learning_rate": 1.687688775858738e-05, "loss": 0.2817, "step": 1066 }, { "epoch": 0.84, "learning_rate": 1.687068907468792e-05, "loss": 0.2588, "step": 1067 }, { "epoch": 0.84, "learning_rate": 1.6864485386199895e-05, "loss": 0.3079, "step": 1068 }, { "epoch": 0.84, "learning_rate": 1.6858276697642043e-05, "loss": 0.2767, "step": 1069 }, { "epoch": 0.85, "learning_rate": 1.6852063013536765e-05, "loss": 0.2536, "step": 1070 }, { "epoch": 0.85, "learning_rate": 1.6845844338410077e-05, "loss": 0.2542, "step": 1071 }, { "epoch": 0.85, "learning_rate": 1.6839620676791653e-05, "loss": 0.2618, "step": 1072 }, { "epoch": 0.85, "learning_rate": 1.6833392033214783e-05, "loss": 0.2911, "step": 1073 }, { "epoch": 0.85, "learning_rate": 1.6827158412216396e-05, "loss": 0.3203, "step": 1074 }, { "epoch": 0.85, "learning_rate": 1.6820919818337035e-05, "loss": 0.287, "step": 1075 }, { "epoch": 0.85, "learning_rate": 1.6814676256120877e-05, "loss": 0.295, "step": 1076 }, { "epoch": 0.85, "learning_rate": 1.6808427730115716e-05, "loss": 0.2788, "step": 1077 }, { "epoch": 0.85, "learning_rate": 1.6802174244872955e-05, "loss": 0.2857, "step": 1078 }, { "epoch": 0.85, "learning_rate": 1.679591580494761e-05, "loss": 0.3072, "step": 1079 }, { "epoch": 0.85, "learning_rate": 1.6789652414898315e-05, "loss": 0.3003, "step": 1080 }, { "epoch": 0.85, "learning_rate": 1.6783384079287307e-05, "loss": 0.3053, "step": 1081 }, { "epoch": 0.85, "learning_rate": 1.677711080268041e-05, "loss": 0.27, "step": 1082 }, { "epoch": 0.86, "learning_rate": 1.677083258964707e-05, "loss": 0.2726, "step": 1083 }, { "epoch": 0.86, "learning_rate": 1.676454944476032e-05, "loss": 0.3065, "step": 1084 }, { "epoch": 0.86, "learning_rate": 1.6758261372596768e-05, "loss": 0.2476, "step": 1085 }, { "epoch": 0.86, "learning_rate": 1.675196837773664e-05, "loss": 0.3151, "step": 1086 }, { "epoch": 0.86, "learning_rate": 1.674567046476373e-05, "loss": 0.2572, "step": 1087 }, { "epoch": 0.86, "learning_rate": 1.6739367638265415e-05, "loss": 0.2416, "step": 1088 }, { "epoch": 0.86, "learning_rate": 1.673305990283266e-05, "loss": 0.2624, "step": 1089 }, { "epoch": 0.86, "learning_rate": 1.6726747263059996e-05, "loss": 0.2781, "step": 1090 }, { "epoch": 0.86, "learning_rate": 1.672042972354553e-05, "loss": 0.2478, "step": 1091 }, { "epoch": 0.86, "learning_rate": 1.6714107288890943e-05, "loss": 0.2797, "step": 1092 }, { "epoch": 0.86, "learning_rate": 1.6707779963701472e-05, "loss": 0.2903, "step": 1093 }, { "epoch": 0.86, "learning_rate": 1.670144775258592e-05, "loss": 0.2647, "step": 1094 }, { "epoch": 0.87, "learning_rate": 1.6695110660156652e-05, "loss": 0.2689, "step": 1095 }, { "epoch": 0.87, "learning_rate": 1.6688768691029588e-05, "loss": 0.2943, "step": 1096 }, { "epoch": 0.87, "learning_rate": 1.6682421849824194e-05, "loss": 0.2644, "step": 1097 }, { "epoch": 0.87, "learning_rate": 1.6676070141163498e-05, "loss": 0.2882, "step": 1098 }, { "epoch": 0.87, "learning_rate": 1.6669713569674058e-05, "loss": 0.2634, "step": 1099 }, { "epoch": 0.87, "learning_rate": 1.6663352139985977e-05, "loss": 0.3407, "step": 1100 }, { "epoch": 0.87, "learning_rate": 1.665698585673291e-05, "loss": 0.2601, "step": 1101 }, { "epoch": 0.87, "learning_rate": 1.6650614724552035e-05, "loss": 0.2716, "step": 1102 }, { "epoch": 0.87, "learning_rate": 1.6644238748084062e-05, "loss": 0.2883, "step": 1103 }, { "epoch": 0.87, "learning_rate": 1.6637857931973233e-05, "loss": 0.2772, "step": 1104 }, { "epoch": 0.87, "learning_rate": 1.6631472280867314e-05, "loss": 0.2988, "step": 1105 }, { "epoch": 0.87, "learning_rate": 1.66250817994176e-05, "loss": 0.2535, "step": 1106 }, { "epoch": 0.87, "learning_rate": 1.6618686492278892e-05, "loss": 0.2952, "step": 1107 }, { "epoch": 0.88, "learning_rate": 1.6612286364109508e-05, "loss": 0.2852, "step": 1108 }, { "epoch": 0.88, "learning_rate": 1.660588141957129e-05, "loss": 0.269, "step": 1109 }, { "epoch": 0.88, "learning_rate": 1.6599471663329577e-05, "loss": 0.2781, "step": 1110 }, { "epoch": 0.88, "learning_rate": 1.659305710005321e-05, "loss": 0.2884, "step": 1111 }, { "epoch": 0.88, "learning_rate": 1.658663773441455e-05, "loss": 0.2928, "step": 1112 }, { "epoch": 0.88, "learning_rate": 1.6580213571089427e-05, "loss": 0.3222, "step": 1113 }, { "epoch": 0.88, "learning_rate": 1.657378461475719e-05, "loss": 0.2718, "step": 1114 }, { "epoch": 0.88, "learning_rate": 1.656735087010067e-05, "loss": 0.2592, "step": 1115 }, { "epoch": 0.88, "learning_rate": 1.656091234180619e-05, "loss": 0.285, "step": 1116 }, { "epoch": 0.88, "learning_rate": 1.6554469034563543e-05, "loss": 0.2628, "step": 1117 }, { "epoch": 0.88, "learning_rate": 1.654802095306603e-05, "loss": 0.3018, "step": 1118 }, { "epoch": 0.88, "learning_rate": 1.65415681020104e-05, "loss": 0.2701, "step": 1119 }, { "epoch": 0.89, "learning_rate": 1.653511048609689e-05, "loss": 0.2383, "step": 1120 }, { "epoch": 0.89, "learning_rate": 1.6528648110029218e-05, "loss": 0.2564, "step": 1121 }, { "epoch": 0.89, "learning_rate": 1.6522180978514556e-05, "loss": 0.2753, "step": 1122 }, { "epoch": 0.89, "learning_rate": 1.6515709096263536e-05, "loss": 0.2915, "step": 1123 }, { "epoch": 0.89, "learning_rate": 1.6509232467990256e-05, "loss": 0.2701, "step": 1124 }, { "epoch": 0.89, "learning_rate": 1.6502751098412282e-05, "loss": 0.2569, "step": 1125 }, { "epoch": 0.89, "learning_rate": 1.649626499225061e-05, "loss": 0.2672, "step": 1126 }, { "epoch": 0.89, "learning_rate": 1.6489774154229715e-05, "loss": 0.2724, "step": 1127 }, { "epoch": 0.89, "learning_rate": 1.648327858907749e-05, "loss": 0.2708, "step": 1128 }, { "epoch": 0.89, "learning_rate": 1.6476778301525292e-05, "loss": 0.264, "step": 1129 }, { "epoch": 0.89, "learning_rate": 1.6470273296307907e-05, "loss": 0.2896, "step": 1130 }, { "epoch": 0.89, "learning_rate": 1.6463763578163563e-05, "loss": 0.2841, "step": 1131 }, { "epoch": 0.89, "learning_rate": 1.6457249151833915e-05, "loss": 0.255, "step": 1132 }, { "epoch": 0.9, "learning_rate": 1.6450730022064053e-05, "loss": 0.3009, "step": 1133 }, { "epoch": 0.9, "learning_rate": 1.6444206193602493e-05, "loss": 0.2616, "step": 1134 }, { "epoch": 0.9, "learning_rate": 1.643767767120117e-05, "loss": 0.2693, "step": 1135 }, { "epoch": 0.9, "learning_rate": 1.6431144459615432e-05, "loss": 0.2821, "step": 1136 }, { "epoch": 0.9, "learning_rate": 1.642460656360406e-05, "loss": 0.2717, "step": 1137 }, { "epoch": 0.9, "learning_rate": 1.6418063987929234e-05, "loss": 0.2752, "step": 1138 }, { "epoch": 0.9, "learning_rate": 1.6411516737356547e-05, "loss": 0.3043, "step": 1139 }, { "epoch": 0.9, "learning_rate": 1.6404964816654993e-05, "loss": 0.2487, "step": 1140 }, { "epoch": 0.9, "learning_rate": 1.6398408230596975e-05, "loss": 0.2681, "step": 1141 }, { "epoch": 0.9, "learning_rate": 1.6391846983958283e-05, "loss": 0.2804, "step": 1142 }, { "epoch": 0.9, "learning_rate": 1.638528108151811e-05, "loss": 0.2593, "step": 1143 }, { "epoch": 0.9, "learning_rate": 1.637871052805905e-05, "loss": 0.2649, "step": 1144 }, { "epoch": 0.9, "learning_rate": 1.6372135328367058e-05, "loss": 0.2851, "step": 1145 }, { "epoch": 0.91, "learning_rate": 1.63655554872315e-05, "loss": 0.2823, "step": 1146 }, { "epoch": 0.91, "learning_rate": 1.6358971009445106e-05, "loss": 0.2574, "step": 1147 }, { "epoch": 0.91, "learning_rate": 1.6352381899803992e-05, "loss": 0.2675, "step": 1148 }, { "epoch": 0.91, "learning_rate": 1.6345788163107645e-05, "loss": 0.2739, "step": 1149 }, { "epoch": 0.91, "learning_rate": 1.6339189804158922e-05, "loss": 0.3103, "step": 1150 }, { "epoch": 0.91, "learning_rate": 1.6332586827764044e-05, "loss": 0.2715, "step": 1151 }, { "epoch": 0.91, "learning_rate": 1.6325979238732606e-05, "loss": 0.2771, "step": 1152 }, { "epoch": 0.91, "learning_rate": 1.6319367041877547e-05, "loss": 0.317, "step": 1153 }, { "epoch": 0.91, "learning_rate": 1.631275024201518e-05, "loss": 0.2574, "step": 1154 }, { "epoch": 0.91, "learning_rate": 1.630612884396515e-05, "loss": 0.3195, "step": 1155 }, { "epoch": 0.91, "learning_rate": 1.629950285255047e-05, "loss": 0.2824, "step": 1156 }, { "epoch": 0.91, "learning_rate": 1.6292872272597487e-05, "loss": 0.2971, "step": 1157 }, { "epoch": 0.92, "learning_rate": 1.62862371089359e-05, "loss": 0.2751, "step": 1158 }, { "epoch": 0.92, "learning_rate": 1.627959736639874e-05, "loss": 0.2852, "step": 1159 }, { "epoch": 0.92, "learning_rate": 1.6272953049822376e-05, "loss": 0.2556, "step": 1160 }, { "epoch": 0.92, "learning_rate": 1.6266304164046505e-05, "loss": 0.2651, "step": 1161 }, { "epoch": 0.92, "learning_rate": 1.625965071391415e-05, "loss": 0.297, "step": 1162 }, { "epoch": 0.92, "learning_rate": 1.6252992704271676e-05, "loss": 0.255, "step": 1163 }, { "epoch": 0.92, "learning_rate": 1.6246330139968748e-05, "loss": 0.284, "step": 1164 }, { "epoch": 0.92, "learning_rate": 1.6239663025858356e-05, "loss": 0.2518, "step": 1165 }, { "epoch": 0.92, "learning_rate": 1.623299136679681e-05, "loss": 0.2538, "step": 1166 }, { "epoch": 0.92, "learning_rate": 1.6226315167643723e-05, "loss": 0.2986, "step": 1167 }, { "epoch": 0.92, "learning_rate": 1.6219634433262017e-05, "loss": 0.273, "step": 1168 }, { "epoch": 0.92, "learning_rate": 1.621294916851792e-05, "loss": 0.2595, "step": 1169 }, { "epoch": 0.92, "learning_rate": 1.6206259378280956e-05, "loss": 0.2885, "step": 1170 }, { "epoch": 0.93, "learning_rate": 1.6199565067423956e-05, "loss": 0.2478, "step": 1171 }, { "epoch": 0.93, "learning_rate": 1.619286624082302e-05, "loss": 0.2769, "step": 1172 }, { "epoch": 0.93, "learning_rate": 1.6186162903357562e-05, "loss": 0.2491, "step": 1173 }, { "epoch": 0.93, "learning_rate": 1.6179455059910275e-05, "loss": 0.2576, "step": 1174 }, { "epoch": 0.93, "learning_rate": 1.6172742715367124e-05, "loss": 0.2605, "step": 1175 }, { "epoch": 0.93, "learning_rate": 1.616602587461736e-05, "loss": 0.2544, "step": 1176 }, { "epoch": 0.93, "learning_rate": 1.6159304542553518e-05, "loss": 0.285, "step": 1177 }, { "epoch": 0.93, "learning_rate": 1.6152578724071385e-05, "loss": 0.2568, "step": 1178 }, { "epoch": 0.93, "learning_rate": 1.6145848424070032e-05, "loss": 0.2744, "step": 1179 }, { "epoch": 0.93, "learning_rate": 1.613911364745179e-05, "loss": 0.2859, "step": 1180 }, { "epoch": 0.93, "learning_rate": 1.613237439912224e-05, "loss": 0.2444, "step": 1181 }, { "epoch": 0.93, "learning_rate": 1.612563068399024e-05, "loss": 0.3071, "step": 1182 }, { "epoch": 0.93, "learning_rate": 1.6118882506967887e-05, "loss": 0.2736, "step": 1183 }, { "epoch": 0.94, "learning_rate": 1.6112129872970533e-05, "loss": 0.2469, "step": 1184 }, { "epoch": 0.94, "learning_rate": 1.6105372786916776e-05, "loss": 0.2469, "step": 1185 }, { "epoch": 0.94, "learning_rate": 1.6098611253728454e-05, "loss": 0.2833, "step": 1186 }, { "epoch": 0.94, "learning_rate": 1.609184527833065e-05, "loss": 0.238, "step": 1187 }, { "epoch": 0.94, "learning_rate": 1.6085074865651672e-05, "loss": 0.2696, "step": 1188 }, { "epoch": 0.94, "learning_rate": 1.607830002062308e-05, "loss": 0.2741, "step": 1189 }, { "epoch": 0.94, "learning_rate": 1.607152074817964e-05, "loss": 0.273, "step": 1190 }, { "epoch": 0.94, "learning_rate": 1.6064737053259355e-05, "loss": 0.2915, "step": 1191 }, { "epoch": 0.94, "learning_rate": 1.605794894080345e-05, "loss": 0.261, "step": 1192 }, { "epoch": 0.94, "learning_rate": 1.6051156415756364e-05, "loss": 0.2676, "step": 1193 }, { "epoch": 0.94, "learning_rate": 1.604435948306575e-05, "loss": 0.2764, "step": 1194 }, { "epoch": 0.94, "learning_rate": 1.6037558147682473e-05, "loss": 0.2942, "step": 1195 }, { "epoch": 0.95, "learning_rate": 1.603075241456061e-05, "loss": 0.2933, "step": 1196 }, { "epoch": 0.95, "learning_rate": 1.6023942288657423e-05, "loss": 0.2664, "step": 1197 }, { "epoch": 0.95, "learning_rate": 1.60171277749334e-05, "loss": 0.2919, "step": 1198 }, { "epoch": 0.95, "learning_rate": 1.6010308878352204e-05, "loss": 0.2646, "step": 1199 }, { "epoch": 0.95, "learning_rate": 1.60034856038807e-05, "loss": 0.2551, "step": 1200 }, { "epoch": 0.95, "learning_rate": 1.5996657956488938e-05, "loss": 0.2654, "step": 1201 }, { "epoch": 0.95, "learning_rate": 1.5989825941150162e-05, "loss": 0.2561, "step": 1202 }, { "epoch": 0.95, "learning_rate": 1.5982989562840785e-05, "loss": 0.2667, "step": 1203 }, { "epoch": 0.95, "learning_rate": 1.59761488265404e-05, "loss": 0.2745, "step": 1204 }, { "epoch": 0.95, "learning_rate": 1.5969303737231786e-05, "loss": 0.2731, "step": 1205 }, { "epoch": 0.95, "learning_rate": 1.596245429990088e-05, "loss": 0.2602, "step": 1206 }, { "epoch": 0.95, "learning_rate": 1.5955600519536793e-05, "loss": 0.2874, "step": 1207 }, { "epoch": 0.95, "learning_rate": 1.5948742401131795e-05, "loss": 0.2816, "step": 1208 }, { "epoch": 0.96, "learning_rate": 1.5941879949681323e-05, "loss": 0.2949, "step": 1209 }, { "epoch": 0.96, "learning_rate": 1.593501317018396e-05, "loss": 0.2902, "step": 1210 }, { "epoch": 0.96, "learning_rate": 1.592814206764145e-05, "loss": 0.2867, "step": 1211 }, { "epoch": 0.96, "learning_rate": 1.5921266647058683e-05, "loss": 0.2678, "step": 1212 }, { "epoch": 0.96, "learning_rate": 1.5914386913443688e-05, "loss": 0.2695, "step": 1213 }, { "epoch": 0.96, "learning_rate": 1.590750287180765e-05, "loss": 0.2902, "step": 1214 }, { "epoch": 0.96, "learning_rate": 1.5900614527164876e-05, "loss": 0.2903, "step": 1215 }, { "epoch": 0.96, "learning_rate": 1.589372188453282e-05, "loss": 0.2661, "step": 1216 }, { "epoch": 0.96, "learning_rate": 1.5886824948932062e-05, "loss": 0.2529, "step": 1217 }, { "epoch": 0.96, "learning_rate": 1.5879923725386307e-05, "loss": 0.2612, "step": 1218 }, { "epoch": 0.96, "learning_rate": 1.5873018218922382e-05, "loss": 0.2597, "step": 1219 }, { "epoch": 0.96, "learning_rate": 1.586610843457024e-05, "loss": 0.2216, "step": 1220 }, { "epoch": 0.96, "learning_rate": 1.5859194377362942e-05, "loss": 0.2525, "step": 1221 }, { "epoch": 0.97, "learning_rate": 1.585227605233667e-05, "loss": 0.2539, "step": 1222 }, { "epoch": 0.97, "learning_rate": 1.584535346453071e-05, "loss": 0.2537, "step": 1223 }, { "epoch": 0.97, "learning_rate": 1.5838426618987455e-05, "loss": 0.2728, "step": 1224 }, { "epoch": 0.97, "learning_rate": 1.5831495520752395e-05, "loss": 0.2761, "step": 1225 }, { "epoch": 0.97, "learning_rate": 1.5824560174874117e-05, "loss": 0.2529, "step": 1226 }, { "epoch": 0.97, "learning_rate": 1.5817620586404315e-05, "loss": 0.2685, "step": 1227 }, { "epoch": 0.97, "learning_rate": 1.5810676760397755e-05, "loss": 0.2545, "step": 1228 }, { "epoch": 0.97, "learning_rate": 1.58037287019123e-05, "loss": 0.2648, "step": 1229 }, { "epoch": 0.97, "learning_rate": 1.5796776416008897e-05, "loss": 0.267, "step": 1230 }, { "epoch": 0.97, "learning_rate": 1.578981990775157e-05, "loss": 0.2528, "step": 1231 }, { "epoch": 0.97, "learning_rate": 1.5782859182207414e-05, "loss": 0.2452, "step": 1232 }, { "epoch": 0.97, "learning_rate": 1.5775894244446603e-05, "loss": 0.2803, "step": 1233 }, { "epoch": 0.98, "learning_rate": 1.5768925099542373e-05, "loss": 0.2845, "step": 1234 }, { "epoch": 0.98, "learning_rate": 1.5761951752571032e-05, "loss": 0.2567, "step": 1235 }, { "epoch": 0.98, "learning_rate": 1.575497420861194e-05, "loss": 0.2594, "step": 1236 }, { "epoch": 0.98, "learning_rate": 1.574799247274752e-05, "loss": 0.2745, "step": 1237 }, { "epoch": 0.98, "learning_rate": 1.5741006550063244e-05, "loss": 0.2702, "step": 1238 }, { "epoch": 0.98, "learning_rate": 1.573401644564764e-05, "loss": 0.2735, "step": 1239 }, { "epoch": 0.98, "learning_rate": 1.5727022164592282e-05, "loss": 0.2668, "step": 1240 }, { "epoch": 0.98, "learning_rate": 1.5720023711991777e-05, "loss": 0.2633, "step": 1241 }, { "epoch": 0.98, "learning_rate": 1.571302109294377e-05, "loss": 0.2935, "step": 1242 }, { "epoch": 0.98, "learning_rate": 1.5706014312548964e-05, "loss": 0.2533, "step": 1243 }, { "epoch": 0.98, "learning_rate": 1.569900337591106e-05, "loss": 0.2498, "step": 1244 }, { "epoch": 0.98, "learning_rate": 1.569198828813681e-05, "loss": 0.2516, "step": 1245 }, { "epoch": 0.98, "learning_rate": 1.5684969054335976e-05, "loss": 0.2753, "step": 1246 }, { "epoch": 0.99, "learning_rate": 1.567794567962136e-05, "loss": 0.27, "step": 1247 }, { "epoch": 0.99, "learning_rate": 1.567091816910875e-05, "loss": 0.2652, "step": 1248 }, { "epoch": 0.99, "learning_rate": 1.566388652791697e-05, "loss": 0.2805, "step": 1249 }, { "epoch": 0.99, "learning_rate": 1.5656850761167848e-05, "loss": 0.2497, "step": 1250 }, { "epoch": 0.99, "learning_rate": 1.5649810873986214e-05, "loss": 0.2774, "step": 1251 }, { "epoch": 0.99, "learning_rate": 1.5642766871499896e-05, "loss": 0.2688, "step": 1252 }, { "epoch": 0.99, "learning_rate": 1.563571875883972e-05, "loss": 0.2927, "step": 1253 }, { "epoch": 0.99, "learning_rate": 1.5628666541139523e-05, "loss": 0.2803, "step": 1254 }, { "epoch": 0.99, "learning_rate": 1.562161022353611e-05, "loss": 0.2743, "step": 1255 }, { "epoch": 0.99, "learning_rate": 1.5614549811169276e-05, "loss": 0.2361, "step": 1256 }, { "epoch": 0.99, "learning_rate": 1.5607485309181813e-05, "loss": 0.2808, "step": 1257 }, { "epoch": 0.99, "learning_rate": 1.5600416722719472e-05, "loss": 0.2293, "step": 1258 }, { "epoch": 0.99, "learning_rate": 1.5593344056931004e-05, "loss": 0.252, "step": 1259 }, { "epoch": 1.0, "learning_rate": 1.55862673169681e-05, "loss": 0.2732, "step": 1260 }, { "epoch": 1.0, "learning_rate": 1.557918650798545e-05, "loss": 0.2614, "step": 1261 }, { "epoch": 1.0, "learning_rate": 1.557210163514068e-05, "loss": 0.2704, "step": 1262 }, { "epoch": 1.0, "learning_rate": 1.5565012703594403e-05, "loss": 0.2369, "step": 1263 }, { "epoch": 1.0, "learning_rate": 1.5557919718510166e-05, "loss": 0.2444, "step": 1264 }, { "epoch": 1.0, "learning_rate": 1.5550822685054475e-05, "loss": 0.2814, "step": 1265 }, { "epoch": 1.0, "learning_rate": 1.55437216083968e-05, "loss": 0.1914, "step": 1266 }, { "epoch": 1.0, "learning_rate": 1.5536616493709528e-05, "loss": 0.1613, "step": 1267 }, { "epoch": 1.0, "learning_rate": 1.5529507346168003e-05, "loss": 0.1722, "step": 1268 }, { "epoch": 1.0, "learning_rate": 1.552239417095052e-05, "loss": 0.167, "step": 1269 }, { "epoch": 1.0, "learning_rate": 1.5515276973238286e-05, "loss": 0.174, "step": 1270 }, { "epoch": 1.0, "learning_rate": 1.5508155758215443e-05, "loss": 0.1723, "step": 1271 }, { "epoch": 1.01, "learning_rate": 1.5501030531069066e-05, "loss": 0.1464, "step": 1272 }, { "epoch": 1.01, "learning_rate": 1.5493901296989142e-05, "loss": 0.169, "step": 1273 }, { "epoch": 1.01, "learning_rate": 1.548676806116859e-05, "loss": 0.1668, "step": 1274 }, { "epoch": 1.01, "learning_rate": 1.5479630828803235e-05, "loss": 0.1722, "step": 1275 }, { "epoch": 1.01, "learning_rate": 1.5472489605091814e-05, "loss": 0.184, "step": 1276 }, { "epoch": 1.01, "learning_rate": 1.5465344395235974e-05, "loss": 0.1597, "step": 1277 }, { "epoch": 1.01, "learning_rate": 1.5458195204440255e-05, "loss": 0.1764, "step": 1278 }, { "epoch": 1.01, "learning_rate": 1.5451042037912117e-05, "loss": 0.1656, "step": 1279 }, { "epoch": 1.01, "learning_rate": 1.5443884900861904e-05, "loss": 0.1968, "step": 1280 }, { "epoch": 1.01, "learning_rate": 1.5436723798502842e-05, "loss": 0.1661, "step": 1281 }, { "epoch": 1.01, "learning_rate": 1.5429558736051062e-05, "loss": 0.1671, "step": 1282 }, { "epoch": 1.01, "learning_rate": 1.5422389718725575e-05, "loss": 0.1781, "step": 1283 }, { "epoch": 1.01, "learning_rate": 1.5415216751748264e-05, "loss": 0.1624, "step": 1284 }, { "epoch": 1.02, "learning_rate": 1.5408039840343903e-05, "loss": 0.152, "step": 1285 }, { "epoch": 1.02, "learning_rate": 1.5400858989740132e-05, "loss": 0.1444, "step": 1286 }, { "epoch": 1.02, "learning_rate": 1.5393674205167453e-05, "loss": 0.1919, "step": 1287 }, { "epoch": 1.02, "learning_rate": 1.5386485491859254e-05, "loss": 0.1376, "step": 1288 }, { "epoch": 1.02, "learning_rate": 1.5379292855051756e-05, "loss": 0.1478, "step": 1289 }, { "epoch": 1.02, "learning_rate": 1.5372096299984064e-05, "loss": 0.1732, "step": 1290 }, { "epoch": 1.02, "learning_rate": 1.536489583189812e-05, "loss": 0.1575, "step": 1291 }, { "epoch": 1.02, "learning_rate": 1.535769145603873e-05, "loss": 0.1735, "step": 1292 }, { "epoch": 1.02, "learning_rate": 1.5350483177653528e-05, "loss": 0.1605, "step": 1293 }, { "epoch": 1.02, "learning_rate": 1.534327100199301e-05, "loss": 0.1684, "step": 1294 }, { "epoch": 1.02, "learning_rate": 1.5336054934310502e-05, "loss": 0.1769, "step": 1295 }, { "epoch": 1.02, "learning_rate": 1.5328834979862158e-05, "loss": 0.1809, "step": 1296 }, { "epoch": 1.02, "learning_rate": 1.5321611143906984e-05, "loss": 0.1549, "step": 1297 }, { "epoch": 1.03, "learning_rate": 1.5314383431706785e-05, "loss": 0.1715, "step": 1298 }, { "epoch": 1.03, "learning_rate": 1.5307151848526213e-05, "loss": 0.185, "step": 1299 }, { "epoch": 1.03, "learning_rate": 1.5299916399632726e-05, "loss": 0.131, "step": 1300 }, { "epoch": 1.03, "learning_rate": 1.529267709029661e-05, "loss": 0.1731, "step": 1301 }, { "epoch": 1.03, "learning_rate": 1.5285433925790946e-05, "loss": 0.1741, "step": 1302 }, { "epoch": 1.03, "learning_rate": 1.5278186911391636e-05, "loss": 0.1777, "step": 1303 }, { "epoch": 1.03, "learning_rate": 1.5270936052377385e-05, "loss": 0.1469, "step": 1304 }, { "epoch": 1.03, "learning_rate": 1.5263681354029694e-05, "loss": 0.1651, "step": 1305 }, { "epoch": 1.03, "learning_rate": 1.5256422821632866e-05, "loss": 0.1817, "step": 1306 }, { "epoch": 1.03, "learning_rate": 1.5249160460473988e-05, "loss": 0.1731, "step": 1307 }, { "epoch": 1.03, "learning_rate": 1.5241894275842946e-05, "loss": 0.183, "step": 1308 }, { "epoch": 1.03, "learning_rate": 1.5234624273032404e-05, "loss": 0.1655, "step": 1309 }, { "epoch": 1.04, "learning_rate": 1.5227350457337809e-05, "loss": 0.165, "step": 1310 }, { "epoch": 1.04, "learning_rate": 1.5220072834057387e-05, "loss": 0.1758, "step": 1311 }, { "epoch": 1.04, "learning_rate": 1.5212791408492135e-05, "loss": 0.1535, "step": 1312 }, { "epoch": 1.04, "learning_rate": 1.5205506185945823e-05, "loss": 0.155, "step": 1313 }, { "epoch": 1.04, "learning_rate": 1.5198217171724982e-05, "loss": 0.1523, "step": 1314 }, { "epoch": 1.04, "learning_rate": 1.5190924371138908e-05, "loss": 0.1483, "step": 1315 }, { "epoch": 1.04, "learning_rate": 1.5183627789499653e-05, "loss": 0.1423, "step": 1316 }, { "epoch": 1.04, "learning_rate": 1.5176327432122028e-05, "loss": 0.1469, "step": 1317 }, { "epoch": 1.04, "learning_rate": 1.5169023304323591e-05, "loss": 0.1489, "step": 1318 }, { "epoch": 1.04, "learning_rate": 1.5161715411424641e-05, "loss": 0.1702, "step": 1319 }, { "epoch": 1.04, "learning_rate": 1.5154403758748228e-05, "loss": 0.1462, "step": 1320 }, { "epoch": 1.04, "learning_rate": 1.5147088351620136e-05, "loss": 0.1693, "step": 1321 }, { "epoch": 1.04, "learning_rate": 1.5139769195368885e-05, "loss": 0.1762, "step": 1322 }, { "epoch": 1.05, "learning_rate": 1.5132446295325722e-05, "loss": 0.1585, "step": 1323 }, { "epoch": 1.05, "learning_rate": 1.5125119656824634e-05, "loss": 0.1509, "step": 1324 }, { "epoch": 1.05, "learning_rate": 1.5117789285202313e-05, "loss": 0.1877, "step": 1325 }, { "epoch": 1.05, "learning_rate": 1.511045518579818e-05, "loss": 0.1669, "step": 1326 }, { "epoch": 1.05, "learning_rate": 1.5103117363954376e-05, "loss": 0.1511, "step": 1327 }, { "epoch": 1.05, "learning_rate": 1.5095775825015744e-05, "loss": 0.1646, "step": 1328 }, { "epoch": 1.05, "learning_rate": 1.5088430574329836e-05, "loss": 0.1473, "step": 1329 }, { "epoch": 1.05, "learning_rate": 1.5081081617246912e-05, "loss": 0.1628, "step": 1330 }, { "epoch": 1.05, "learning_rate": 1.5073728959119929e-05, "loss": 0.1512, "step": 1331 }, { "epoch": 1.05, "learning_rate": 1.5066372605304537e-05, "loss": 0.1614, "step": 1332 }, { "epoch": 1.05, "learning_rate": 1.5059012561159087e-05, "loss": 0.1562, "step": 1333 }, { "epoch": 1.05, "learning_rate": 1.5051648832044614e-05, "loss": 0.1408, "step": 1334 }, { "epoch": 1.05, "learning_rate": 1.5044281423324826e-05, "loss": 0.1659, "step": 1335 }, { "epoch": 1.06, "learning_rate": 1.5036910340366131e-05, "loss": 0.1688, "step": 1336 }, { "epoch": 1.06, "learning_rate": 1.5029535588537595e-05, "loss": 0.1493, "step": 1337 }, { "epoch": 1.06, "learning_rate": 1.5022157173210969e-05, "loss": 0.1557, "step": 1338 }, { "epoch": 1.06, "learning_rate": 1.5014775099760663e-05, "loss": 0.1501, "step": 1339 }, { "epoch": 1.06, "learning_rate": 1.500738937356376e-05, "loss": 0.143, "step": 1340 }, { "epoch": 1.06, "learning_rate": 1.5000000000000002e-05, "loss": 0.1668, "step": 1341 }, { "epoch": 1.06, "learning_rate": 1.499260698445178e-05, "loss": 0.1416, "step": 1342 }, { "epoch": 1.06, "learning_rate": 1.498521033230415e-05, "loss": 0.1476, "step": 1343 }, { "epoch": 1.06, "learning_rate": 1.4977810048944806e-05, "loss": 0.1455, "step": 1344 }, { "epoch": 1.06, "learning_rate": 1.4970406139764092e-05, "loss": 0.1502, "step": 1345 }, { "epoch": 1.06, "learning_rate": 1.4962998610154993e-05, "loss": 0.1457, "step": 1346 }, { "epoch": 1.06, "learning_rate": 1.495558746551313e-05, "loss": 0.1526, "step": 1347 }, { "epoch": 1.07, "learning_rate": 1.494817271123676e-05, "loss": 0.1517, "step": 1348 }, { "epoch": 1.07, "learning_rate": 1.4940754352726762e-05, "loss": 0.1367, "step": 1349 }, { "epoch": 1.07, "learning_rate": 1.4933332395386652e-05, "loss": 0.145, "step": 1350 }, { "epoch": 1.07, "learning_rate": 1.4925906844622558e-05, "loss": 0.1629, "step": 1351 }, { "epoch": 1.07, "learning_rate": 1.4918477705843222e-05, "loss": 0.1547, "step": 1352 }, { "epoch": 1.07, "learning_rate": 1.4911044984460015e-05, "loss": 0.1402, "step": 1353 }, { "epoch": 1.07, "learning_rate": 1.4903608685886904e-05, "loss": 0.1581, "step": 1354 }, { "epoch": 1.07, "learning_rate": 1.4896168815540464e-05, "loss": 0.175, "step": 1355 }, { "epoch": 1.07, "learning_rate": 1.4888725378839877e-05, "loss": 0.1728, "step": 1356 }, { "epoch": 1.07, "learning_rate": 1.4881278381206916e-05, "loss": 0.1568, "step": 1357 }, { "epoch": 1.07, "learning_rate": 1.4873827828065952e-05, "loss": 0.1719, "step": 1358 }, { "epoch": 1.07, "learning_rate": 1.4866373724843945e-05, "loss": 0.1507, "step": 1359 }, { "epoch": 1.07, "learning_rate": 1.4858916076970444e-05, "loss": 0.1465, "step": 1360 }, { "epoch": 1.08, "learning_rate": 1.4851454889877569e-05, "loss": 0.1617, "step": 1361 }, { "epoch": 1.08, "learning_rate": 1.484399016900003e-05, "loss": 0.1612, "step": 1362 }, { "epoch": 1.08, "learning_rate": 1.483652191977511e-05, "loss": 0.1826, "step": 1363 }, { "epoch": 1.08, "learning_rate": 1.4829050147642652e-05, "loss": 0.148, "step": 1364 }, { "epoch": 1.08, "learning_rate": 1.4821574858045073e-05, "loss": 0.1563, "step": 1365 }, { "epoch": 1.08, "learning_rate": 1.4814096056427356e-05, "loss": 0.1584, "step": 1366 }, { "epoch": 1.08, "learning_rate": 1.480661374823703e-05, "loss": 0.1822, "step": 1367 }, { "epoch": 1.08, "learning_rate": 1.479912793892419e-05, "loss": 0.159, "step": 1368 }, { "epoch": 1.08, "learning_rate": 1.4791638633941472e-05, "loss": 0.1524, "step": 1369 }, { "epoch": 1.08, "learning_rate": 1.4784145838744067e-05, "loss": 0.1498, "step": 1370 }, { "epoch": 1.08, "learning_rate": 1.4776649558789698e-05, "loss": 0.1638, "step": 1371 }, { "epoch": 1.08, "learning_rate": 1.4769149799538637e-05, "loss": 0.1462, "step": 1372 }, { "epoch": 1.08, "learning_rate": 1.4761646566453687e-05, "loss": 0.1526, "step": 1373 }, { "epoch": 1.09, "learning_rate": 1.475413986500017e-05, "loss": 0.1734, "step": 1374 }, { "epoch": 1.09, "learning_rate": 1.4746629700645955e-05, "loss": 0.1582, "step": 1375 }, { "epoch": 1.09, "learning_rate": 1.4739116078861417e-05, "loss": 0.1568, "step": 1376 }, { "epoch": 1.09, "learning_rate": 1.4731599005119454e-05, "loss": 0.1436, "step": 1377 }, { "epoch": 1.09, "learning_rate": 1.4724078484895483e-05, "loss": 0.18, "step": 1378 }, { "epoch": 1.09, "learning_rate": 1.4716554523667423e-05, "loss": 0.1643, "step": 1379 }, { "epoch": 1.09, "learning_rate": 1.470902712691571e-05, "loss": 0.1716, "step": 1380 }, { "epoch": 1.09, "learning_rate": 1.4701496300123267e-05, "loss": 0.185, "step": 1381 }, { "epoch": 1.09, "learning_rate": 1.4693962048775535e-05, "loss": 0.1628, "step": 1382 }, { "epoch": 1.09, "learning_rate": 1.4686424378360434e-05, "loss": 0.1712, "step": 1383 }, { "epoch": 1.09, "learning_rate": 1.4678883294368377e-05, "loss": 0.1738, "step": 1384 }, { "epoch": 1.09, "learning_rate": 1.4671338802292274e-05, "loss": 0.1649, "step": 1385 }, { "epoch": 1.1, "learning_rate": 1.4663790907627502e-05, "loss": 0.1619, "step": 1386 }, { "epoch": 1.1, "learning_rate": 1.4656239615871927e-05, "loss": 0.1678, "step": 1387 }, { "epoch": 1.1, "learning_rate": 1.4648684932525889e-05, "loss": 0.1397, "step": 1388 }, { "epoch": 1.1, "learning_rate": 1.4641126863092194e-05, "loss": 0.1378, "step": 1389 }, { "epoch": 1.1, "learning_rate": 1.4633565413076114e-05, "loss": 0.1409, "step": 1390 }, { "epoch": 1.1, "learning_rate": 1.4626000587985388e-05, "loss": 0.1622, "step": 1391 }, { "epoch": 1.1, "learning_rate": 1.4618432393330211e-05, "loss": 0.1655, "step": 1392 }, { "epoch": 1.1, "learning_rate": 1.4610860834623233e-05, "loss": 0.1598, "step": 1393 }, { "epoch": 1.1, "learning_rate": 1.4603285917379549e-05, "loss": 0.1459, "step": 1394 }, { "epoch": 1.1, "learning_rate": 1.4595707647116713e-05, "loss": 0.1452, "step": 1395 }, { "epoch": 1.1, "learning_rate": 1.4588126029354706e-05, "loss": 0.1456, "step": 1396 }, { "epoch": 1.1, "learning_rate": 1.4580541069615958e-05, "loss": 0.1719, "step": 1397 }, { "epoch": 1.1, "learning_rate": 1.4572952773425335e-05, "loss": 0.1495, "step": 1398 }, { "epoch": 1.11, "learning_rate": 1.4565361146310121e-05, "loss": 0.1485, "step": 1399 }, { "epoch": 1.11, "learning_rate": 1.4557766193800036e-05, "loss": 0.1528, "step": 1400 }, { "epoch": 1.11, "learning_rate": 1.455016792142722e-05, "loss": 0.1479, "step": 1401 }, { "epoch": 1.11, "learning_rate": 1.4542566334726234e-05, "loss": 0.1505, "step": 1402 }, { "epoch": 1.11, "learning_rate": 1.4534961439234043e-05, "loss": 0.1633, "step": 1403 }, { "epoch": 1.11, "learning_rate": 1.4527353240490039e-05, "loss": 0.1569, "step": 1404 }, { "epoch": 1.11, "learning_rate": 1.4519741744036e-05, "loss": 0.1687, "step": 1405 }, { "epoch": 1.11, "learning_rate": 1.4512126955416124e-05, "loss": 0.1677, "step": 1406 }, { "epoch": 1.11, "learning_rate": 1.4504508880176996e-05, "loss": 0.1461, "step": 1407 }, { "epoch": 1.11, "learning_rate": 1.4496887523867597e-05, "loss": 0.1716, "step": 1408 }, { "epoch": 1.11, "learning_rate": 1.4489262892039304e-05, "loss": 0.1468, "step": 1409 }, { "epoch": 1.11, "learning_rate": 1.4481634990245871e-05, "loss": 0.1331, "step": 1410 }, { "epoch": 1.11, "learning_rate": 1.447400382404344e-05, "loss": 0.1614, "step": 1411 }, { "epoch": 1.12, "learning_rate": 1.4466369398990527e-05, "loss": 0.182, "step": 1412 }, { "epoch": 1.12, "learning_rate": 1.4458731720648024e-05, "loss": 0.1639, "step": 1413 }, { "epoch": 1.12, "learning_rate": 1.4451090794579194e-05, "loss": 0.1736, "step": 1414 }, { "epoch": 1.12, "learning_rate": 1.4443446626349662e-05, "loss": 0.1433, "step": 1415 }, { "epoch": 1.12, "learning_rate": 1.4435799221527417e-05, "loss": 0.1809, "step": 1416 }, { "epoch": 1.12, "learning_rate": 1.4428148585682806e-05, "loss": 0.1618, "step": 1417 }, { "epoch": 1.12, "learning_rate": 1.4420494724388525e-05, "loss": 0.1399, "step": 1418 }, { "epoch": 1.12, "learning_rate": 1.4412837643219625e-05, "loss": 0.1556, "step": 1419 }, { "epoch": 1.12, "learning_rate": 1.4405177347753503e-05, "loss": 0.1832, "step": 1420 }, { "epoch": 1.12, "learning_rate": 1.4397513843569888e-05, "loss": 0.1383, "step": 1421 }, { "epoch": 1.12, "learning_rate": 1.4389847136250858e-05, "loss": 0.1803, "step": 1422 }, { "epoch": 1.12, "learning_rate": 1.4382177231380819e-05, "loss": 0.1414, "step": 1423 }, { "epoch": 1.13, "learning_rate": 1.4374504134546498e-05, "loss": 0.143, "step": 1424 }, { "epoch": 1.13, "learning_rate": 1.4366827851336964e-05, "loss": 0.1653, "step": 1425 }, { "epoch": 1.13, "learning_rate": 1.4359148387343594e-05, "loss": 0.1693, "step": 1426 }, { "epoch": 1.13, "learning_rate": 1.4351465748160085e-05, "loss": 0.1485, "step": 1427 }, { "epoch": 1.13, "learning_rate": 1.4343779939382451e-05, "loss": 0.1585, "step": 1428 }, { "epoch": 1.13, "learning_rate": 1.4336090966609006e-05, "loss": 0.1652, "step": 1429 }, { "epoch": 1.13, "learning_rate": 1.4328398835440381e-05, "loss": 0.1508, "step": 1430 }, { "epoch": 1.13, "learning_rate": 1.4320703551479494e-05, "loss": 0.1589, "step": 1431 }, { "epoch": 1.13, "learning_rate": 1.4313005120331566e-05, "loss": 0.1598, "step": 1432 }, { "epoch": 1.13, "learning_rate": 1.4305303547604115e-05, "loss": 0.185, "step": 1433 }, { "epoch": 1.13, "learning_rate": 1.4297598838906938e-05, "loss": 0.1422, "step": 1434 }, { "epoch": 1.13, "learning_rate": 1.4289890999852126e-05, "loss": 0.1687, "step": 1435 }, { "epoch": 1.13, "learning_rate": 1.4282180036054041e-05, "loss": 0.149, "step": 1436 }, { "epoch": 1.14, "learning_rate": 1.4274465953129326e-05, "loss": 0.1697, "step": 1437 }, { "epoch": 1.14, "learning_rate": 1.4266748756696895e-05, "loss": 0.1835, "step": 1438 }, { "epoch": 1.14, "learning_rate": 1.4259028452377927e-05, "loss": 0.1669, "step": 1439 }, { "epoch": 1.14, "learning_rate": 1.4251305045795874e-05, "loss": 0.1643, "step": 1440 }, { "epoch": 1.14, "learning_rate": 1.4243578542576438e-05, "loss": 0.1545, "step": 1441 }, { "epoch": 1.14, "learning_rate": 1.4235848948347578e-05, "loss": 0.1722, "step": 1442 }, { "epoch": 1.14, "learning_rate": 1.422811626873951e-05, "loss": 0.1561, "step": 1443 }, { "epoch": 1.14, "learning_rate": 1.4220380509384693e-05, "loss": 0.1533, "step": 1444 }, { "epoch": 1.14, "learning_rate": 1.4212641675917823e-05, "loss": 0.1577, "step": 1445 }, { "epoch": 1.14, "learning_rate": 1.4204899773975855e-05, "loss": 0.159, "step": 1446 }, { "epoch": 1.14, "learning_rate": 1.419715480919796e-05, "loss": 0.156, "step": 1447 }, { "epoch": 1.14, "learning_rate": 1.4189406787225541e-05, "loss": 0.1439, "step": 1448 }, { "epoch": 1.15, "learning_rate": 1.4181655713702242e-05, "loss": 0.157, "step": 1449 }, { "epoch": 1.15, "learning_rate": 1.4173901594273917e-05, "loss": 0.1467, "step": 1450 }, { "epoch": 1.15, "learning_rate": 1.416614443458864e-05, "loss": 0.1475, "step": 1451 }, { "epoch": 1.15, "learning_rate": 1.4158384240296707e-05, "loss": 0.1674, "step": 1452 }, { "epoch": 1.15, "learning_rate": 1.4150621017050616e-05, "loss": 0.1544, "step": 1453 }, { "epoch": 1.15, "learning_rate": 1.4142854770505076e-05, "loss": 0.1549, "step": 1454 }, { "epoch": 1.15, "learning_rate": 1.4135085506316997e-05, "loss": 0.1558, "step": 1455 }, { "epoch": 1.15, "learning_rate": 1.4127313230145486e-05, "loss": 0.1427, "step": 1456 }, { "epoch": 1.15, "learning_rate": 1.4119537947651844e-05, "loss": 0.1591, "step": 1457 }, { "epoch": 1.15, "learning_rate": 1.4111759664499562e-05, "loss": 0.1842, "step": 1458 }, { "epoch": 1.15, "learning_rate": 1.4103978386354325e-05, "loss": 0.1542, "step": 1459 }, { "epoch": 1.15, "learning_rate": 1.4096194118883982e-05, "loss": 0.1347, "step": 1460 }, { "epoch": 1.15, "learning_rate": 1.4088406867758573e-05, "loss": 0.1484, "step": 1461 }, { "epoch": 1.16, "learning_rate": 1.4080616638650307e-05, "loss": 0.1822, "step": 1462 }, { "epoch": 1.16, "learning_rate": 1.4072823437233565e-05, "loss": 0.1521, "step": 1463 }, { "epoch": 1.16, "learning_rate": 1.4065027269184888e-05, "loss": 0.1266, "step": 1464 }, { "epoch": 1.16, "learning_rate": 1.4057228140182982e-05, "loss": 0.1592, "step": 1465 }, { "epoch": 1.16, "learning_rate": 1.4049426055908709e-05, "loss": 0.1735, "step": 1466 }, { "epoch": 1.16, "learning_rate": 1.404162102204508e-05, "loss": 0.1557, "step": 1467 }, { "epoch": 1.16, "learning_rate": 1.4033813044277258e-05, "loss": 0.1502, "step": 1468 }, { "epoch": 1.16, "learning_rate": 1.4026002128292553e-05, "loss": 0.1548, "step": 1469 }, { "epoch": 1.16, "learning_rate": 1.4018188279780412e-05, "loss": 0.1693, "step": 1470 }, { "epoch": 1.16, "learning_rate": 1.4010371504432413e-05, "loss": 0.1584, "step": 1471 }, { "epoch": 1.16, "learning_rate": 1.4002551807942273e-05, "loss": 0.1637, "step": 1472 }, { "epoch": 1.16, "learning_rate": 1.3994729196005839e-05, "loss": 0.1566, "step": 1473 }, { "epoch": 1.16, "learning_rate": 1.3986903674321069e-05, "loss": 0.1539, "step": 1474 }, { "epoch": 1.17, "learning_rate": 1.3979075248588054e-05, "loss": 0.139, "step": 1475 }, { "epoch": 1.17, "learning_rate": 1.3971243924508996e-05, "loss": 0.1671, "step": 1476 }, { "epoch": 1.17, "learning_rate": 1.3963409707788205e-05, "loss": 0.1507, "step": 1477 }, { "epoch": 1.17, "learning_rate": 1.3955572604132102e-05, "loss": 0.1687, "step": 1478 }, { "epoch": 1.17, "learning_rate": 1.3947732619249206e-05, "loss": 0.1546, "step": 1479 }, { "epoch": 1.17, "learning_rate": 1.3939889758850138e-05, "loss": 0.1469, "step": 1480 }, { "epoch": 1.17, "learning_rate": 1.3932044028647615e-05, "loss": 0.1256, "step": 1481 }, { "epoch": 1.17, "learning_rate": 1.3924195434356443e-05, "loss": 0.1331, "step": 1482 }, { "epoch": 1.17, "learning_rate": 1.3916343981693512e-05, "loss": 0.1623, "step": 1483 }, { "epoch": 1.17, "learning_rate": 1.3908489676377795e-05, "loss": 0.166, "step": 1484 }, { "epoch": 1.17, "learning_rate": 1.3900632524130343e-05, "loss": 0.1686, "step": 1485 }, { "epoch": 1.17, "learning_rate": 1.3892772530674288e-05, "loss": 0.1823, "step": 1486 }, { "epoch": 1.18, "learning_rate": 1.3884909701734812e-05, "loss": 0.1581, "step": 1487 }, { "epoch": 1.18, "learning_rate": 1.3877044043039189e-05, "loss": 0.1582, "step": 1488 }, { "epoch": 1.18, "learning_rate": 1.3869175560316734e-05, "loss": 0.1449, "step": 1489 }, { "epoch": 1.18, "learning_rate": 1.3861304259298823e-05, "loss": 0.1832, "step": 1490 }, { "epoch": 1.18, "learning_rate": 1.3853430145718892e-05, "loss": 0.1705, "step": 1491 }, { "epoch": 1.18, "learning_rate": 1.3845553225312419e-05, "loss": 0.1606, "step": 1492 }, { "epoch": 1.18, "learning_rate": 1.3837673503816924e-05, "loss": 0.1639, "step": 1493 }, { "epoch": 1.18, "learning_rate": 1.382979098697198e-05, "loss": 0.152, "step": 1494 }, { "epoch": 1.18, "learning_rate": 1.3821905680519181e-05, "loss": 0.1477, "step": 1495 }, { "epoch": 1.18, "learning_rate": 1.3814017590202163e-05, "loss": 0.1596, "step": 1496 }, { "epoch": 1.18, "learning_rate": 1.3806126721766586e-05, "loss": 0.151, "step": 1497 }, { "epoch": 1.18, "learning_rate": 1.3798233080960136e-05, "loss": 0.1532, "step": 1498 }, { "epoch": 1.18, "learning_rate": 1.3790336673532511e-05, "loss": 0.165, "step": 1499 }, { "epoch": 1.19, "learning_rate": 1.378243750523543e-05, "loss": 0.16, "step": 1500 }, { "epoch": 1.19, "learning_rate": 1.377453558182263e-05, "loss": 0.1256, "step": 1501 }, { "epoch": 1.19, "learning_rate": 1.3766630909049843e-05, "loss": 0.1584, "step": 1502 }, { "epoch": 1.19, "learning_rate": 1.3758723492674803e-05, "loss": 0.1366, "step": 1503 }, { "epoch": 1.19, "learning_rate": 1.3750813338457258e-05, "loss": 0.1847, "step": 1504 }, { "epoch": 1.19, "learning_rate": 1.3742900452158932e-05, "loss": 0.1427, "step": 1505 }, { "epoch": 1.19, "learning_rate": 1.3734984839543547e-05, "loss": 0.1713, "step": 1506 }, { "epoch": 1.19, "learning_rate": 1.3727066506376816e-05, "loss": 0.1419, "step": 1507 }, { "epoch": 1.19, "learning_rate": 1.3719145458426424e-05, "loss": 0.1565, "step": 1508 }, { "epoch": 1.19, "learning_rate": 1.3711221701462037e-05, "loss": 0.1662, "step": 1509 }, { "epoch": 1.19, "learning_rate": 1.3703295241255296e-05, "loss": 0.1599, "step": 1510 }, { "epoch": 1.19, "learning_rate": 1.3695366083579815e-05, "loss": 0.1646, "step": 1511 }, { "epoch": 1.19, "learning_rate": 1.368743423421116e-05, "loss": 0.1492, "step": 1512 }, { "epoch": 1.2, "learning_rate": 1.3679499698926869e-05, "loss": 0.1761, "step": 1513 }, { "epoch": 1.2, "learning_rate": 1.3671562483506431e-05, "loss": 0.1468, "step": 1514 }, { "epoch": 1.2, "learning_rate": 1.3663622593731294e-05, "loss": 0.1604, "step": 1515 }, { "epoch": 1.2, "learning_rate": 1.3655680035384841e-05, "loss": 0.1455, "step": 1516 }, { "epoch": 1.2, "learning_rate": 1.3647734814252416e-05, "loss": 0.1827, "step": 1517 }, { "epoch": 1.2, "learning_rate": 1.3639786936121287e-05, "loss": 0.151, "step": 1518 }, { "epoch": 1.2, "learning_rate": 1.3631836406780665e-05, "loss": 0.1385, "step": 1519 }, { "epoch": 1.2, "learning_rate": 1.3623883232021693e-05, "loss": 0.1661, "step": 1520 }, { "epoch": 1.2, "learning_rate": 1.3615927417637435e-05, "loss": 0.1514, "step": 1521 }, { "epoch": 1.2, "learning_rate": 1.3607968969422884e-05, "loss": 0.1497, "step": 1522 }, { "epoch": 1.2, "learning_rate": 1.3600007893174951e-05, "loss": 0.143, "step": 1523 }, { "epoch": 1.2, "learning_rate": 1.3592044194692456e-05, "loss": 0.1533, "step": 1524 }, { "epoch": 1.21, "learning_rate": 1.3584077879776132e-05, "loss": 0.1267, "step": 1525 }, { "epoch": 1.21, "learning_rate": 1.357610895422862e-05, "loss": 0.1372, "step": 1526 }, { "epoch": 1.21, "learning_rate": 1.356813742385446e-05, "loss": 0.1931, "step": 1527 }, { "epoch": 1.21, "learning_rate": 1.3560163294460089e-05, "loss": 0.1374, "step": 1528 }, { "epoch": 1.21, "learning_rate": 1.3552186571853837e-05, "loss": 0.1507, "step": 1529 }, { "epoch": 1.21, "learning_rate": 1.3544207261845928e-05, "loss": 0.1584, "step": 1530 }, { "epoch": 1.21, "learning_rate": 1.3536225370248463e-05, "loss": 0.1287, "step": 1531 }, { "epoch": 1.21, "learning_rate": 1.3528240902875429e-05, "loss": 0.1709, "step": 1532 }, { "epoch": 1.21, "learning_rate": 1.3520253865542687e-05, "loss": 0.1511, "step": 1533 }, { "epoch": 1.21, "learning_rate": 1.3512264264067969e-05, "loss": 0.1416, "step": 1534 }, { "epoch": 1.21, "learning_rate": 1.3504272104270876e-05, "loss": 0.1668, "step": 1535 }, { "epoch": 1.21, "learning_rate": 1.3496277391972874e-05, "loss": 0.1551, "step": 1536 }, { "epoch": 1.21, "learning_rate": 1.3488280132997286e-05, "loss": 0.1436, "step": 1537 }, { "epoch": 1.22, "learning_rate": 1.3480280333169289e-05, "loss": 0.1683, "step": 1538 }, { "epoch": 1.22, "learning_rate": 1.3472277998315915e-05, "loss": 0.1639, "step": 1539 }, { "epoch": 1.22, "learning_rate": 1.3464273134266037e-05, "loss": 0.1545, "step": 1540 }, { "epoch": 1.22, "learning_rate": 1.3456265746850374e-05, "loss": 0.1413, "step": 1541 }, { "epoch": 1.22, "learning_rate": 1.3448255841901481e-05, "loss": 0.1519, "step": 1542 }, { "epoch": 1.22, "learning_rate": 1.3440243425253753e-05, "loss": 0.1677, "step": 1543 }, { "epoch": 1.22, "learning_rate": 1.3432228502743404e-05, "loss": 0.1663, "step": 1544 }, { "epoch": 1.22, "learning_rate": 1.3424211080208478e-05, "loss": 0.1441, "step": 1545 }, { "epoch": 1.22, "learning_rate": 1.3416191163488847e-05, "loss": 0.1892, "step": 1546 }, { "epoch": 1.22, "learning_rate": 1.3408168758426188e-05, "loss": 0.1495, "step": 1547 }, { "epoch": 1.22, "learning_rate": 1.3400143870863997e-05, "loss": 0.1578, "step": 1548 }, { "epoch": 1.22, "learning_rate": 1.339211650664758e-05, "loss": 0.15, "step": 1549 }, { "epoch": 1.22, "learning_rate": 1.338408667162404e-05, "loss": 0.159, "step": 1550 }, { "epoch": 1.23, "learning_rate": 1.3376054371642282e-05, "loss": 0.1259, "step": 1551 }, { "epoch": 1.23, "learning_rate": 1.3368019612553016e-05, "loss": 0.1382, "step": 1552 }, { "epoch": 1.23, "learning_rate": 1.335998240020873e-05, "loss": 0.1607, "step": 1553 }, { "epoch": 1.23, "learning_rate": 1.3351942740463705e-05, "loss": 0.1458, "step": 1554 }, { "epoch": 1.23, "learning_rate": 1.3343900639174007e-05, "loss": 0.164, "step": 1555 }, { "epoch": 1.23, "learning_rate": 1.3335856102197478e-05, "loss": 0.1536, "step": 1556 }, { "epoch": 1.23, "learning_rate": 1.3327809135393728e-05, "loss": 0.1776, "step": 1557 }, { "epoch": 1.23, "learning_rate": 1.3319759744624145e-05, "loss": 0.1444, "step": 1558 }, { "epoch": 1.23, "learning_rate": 1.3311707935751885e-05, "loss": 0.1425, "step": 1559 }, { "epoch": 1.23, "learning_rate": 1.3303653714641853e-05, "loss": 0.1633, "step": 1560 }, { "epoch": 1.23, "learning_rate": 1.3295597087160726e-05, "loss": 0.1448, "step": 1561 }, { "epoch": 1.23, "learning_rate": 1.328753805917692e-05, "loss": 0.1406, "step": 1562 }, { "epoch": 1.24, "learning_rate": 1.3279476636560608e-05, "loss": 0.1528, "step": 1563 }, { "epoch": 1.24, "learning_rate": 1.327141282518371e-05, "loss": 0.1588, "step": 1564 }, { "epoch": 1.24, "learning_rate": 1.3263346630919875e-05, "loss": 0.1628, "step": 1565 }, { "epoch": 1.24, "learning_rate": 1.3255278059644496e-05, "loss": 0.1512, "step": 1566 }, { "epoch": 1.24, "learning_rate": 1.3247207117234694e-05, "loss": 0.166, "step": 1567 }, { "epoch": 1.24, "learning_rate": 1.3239133809569324e-05, "loss": 0.1407, "step": 1568 }, { "epoch": 1.24, "learning_rate": 1.323105814252895e-05, "loss": 0.1524, "step": 1569 }, { "epoch": 1.24, "learning_rate": 1.3222980121995867e-05, "loss": 0.1522, "step": 1570 }, { "epoch": 1.24, "learning_rate": 1.321489975385408e-05, "loss": 0.1518, "step": 1571 }, { "epoch": 1.24, "learning_rate": 1.3206817043989301e-05, "loss": 0.1532, "step": 1572 }, { "epoch": 1.24, "learning_rate": 1.3198731998288958e-05, "loss": 0.1792, "step": 1573 }, { "epoch": 1.24, "learning_rate": 1.3190644622642169e-05, "loss": 0.1417, "step": 1574 }, { "epoch": 1.24, "learning_rate": 1.3182554922939748e-05, "loss": 0.1609, "step": 1575 }, { "epoch": 1.25, "learning_rate": 1.3174462905074215e-05, "loss": 0.1593, "step": 1576 }, { "epoch": 1.25, "learning_rate": 1.3166368574939768e-05, "loss": 0.1711, "step": 1577 }, { "epoch": 1.25, "learning_rate": 1.3158271938432288e-05, "loss": 0.1687, "step": 1578 }, { "epoch": 1.25, "learning_rate": 1.3150173001449347e-05, "loss": 0.1503, "step": 1579 }, { "epoch": 1.25, "learning_rate": 1.3142071769890182e-05, "loss": 0.1598, "step": 1580 }, { "epoch": 1.25, "learning_rate": 1.3133968249655701e-05, "loss": 0.1676, "step": 1581 }, { "epoch": 1.25, "learning_rate": 1.312586244664849e-05, "loss": 0.1643, "step": 1582 }, { "epoch": 1.25, "learning_rate": 1.3117754366772789e-05, "loss": 0.1629, "step": 1583 }, { "epoch": 1.25, "learning_rate": 1.3109644015934493e-05, "loss": 0.1635, "step": 1584 }, { "epoch": 1.25, "learning_rate": 1.3101531400041163e-05, "loss": 0.1596, "step": 1585 }, { "epoch": 1.25, "learning_rate": 1.3093416525002001e-05, "loss": 0.1427, "step": 1586 }, { "epoch": 1.25, "learning_rate": 1.3085299396727851e-05, "loss": 0.156, "step": 1587 }, { "epoch": 1.25, "learning_rate": 1.3077180021131213e-05, "loss": 0.1396, "step": 1588 }, { "epoch": 1.26, "learning_rate": 1.306905840412621e-05, "loss": 0.1459, "step": 1589 }, { "epoch": 1.26, "learning_rate": 1.3060934551628603e-05, "loss": 0.1623, "step": 1590 }, { "epoch": 1.26, "learning_rate": 1.3052808469555782e-05, "loss": 0.1496, "step": 1591 }, { "epoch": 1.26, "learning_rate": 1.3044680163826763e-05, "loss": 0.16, "step": 1592 }, { "epoch": 1.26, "learning_rate": 1.3036549640362169e-05, "loss": 0.1708, "step": 1593 }, { "epoch": 1.26, "learning_rate": 1.3028416905084256e-05, "loss": 0.1538, "step": 1594 }, { "epoch": 1.26, "learning_rate": 1.3020281963916883e-05, "loss": 0.1388, "step": 1595 }, { "epoch": 1.26, "learning_rate": 1.301214482278551e-05, "loss": 0.151, "step": 1596 }, { "epoch": 1.26, "learning_rate": 1.3004005487617213e-05, "loss": 0.1675, "step": 1597 }, { "epoch": 1.26, "learning_rate": 1.2995863964340655e-05, "loss": 0.1478, "step": 1598 }, { "epoch": 1.26, "learning_rate": 1.2987720258886094e-05, "loss": 0.1735, "step": 1599 }, { "epoch": 1.26, "learning_rate": 1.2979574377185385e-05, "loss": 0.1354, "step": 1600 }, { "epoch": 1.27, "learning_rate": 1.2971426325171956e-05, "loss": 0.147, "step": 1601 }, { "epoch": 1.27, "learning_rate": 1.2963276108780829e-05, "loss": 0.1555, "step": 1602 }, { "epoch": 1.27, "learning_rate": 1.295512373394859e-05, "loss": 0.1434, "step": 1603 }, { "epoch": 1.27, "learning_rate": 1.2946969206613411e-05, "loss": 0.1451, "step": 1604 }, { "epoch": 1.27, "learning_rate": 1.293881253271502e-05, "loss": 0.161, "step": 1605 }, { "epoch": 1.27, "learning_rate": 1.2930653718194712e-05, "loss": 0.1446, "step": 1606 }, { "epoch": 1.27, "learning_rate": 1.2922492768995345e-05, "loss": 0.143, "step": 1607 }, { "epoch": 1.27, "learning_rate": 1.2914329691061327e-05, "loss": 0.1383, "step": 1608 }, { "epoch": 1.27, "learning_rate": 1.290616449033862e-05, "loss": 0.1589, "step": 1609 }, { "epoch": 1.27, "learning_rate": 1.289799717277473e-05, "loss": 0.1321, "step": 1610 }, { "epoch": 1.27, "learning_rate": 1.2889827744318705e-05, "loss": 0.1452, "step": 1611 }, { "epoch": 1.27, "learning_rate": 1.2881656210921131e-05, "loss": 0.1361, "step": 1612 }, { "epoch": 1.27, "learning_rate": 1.287348257853413e-05, "loss": 0.1761, "step": 1613 }, { "epoch": 1.28, "learning_rate": 1.286530685311135e-05, "loss": 0.1399, "step": 1614 }, { "epoch": 1.28, "learning_rate": 1.2857129040607963e-05, "loss": 0.1546, "step": 1615 }, { "epoch": 1.28, "learning_rate": 1.2848949146980663e-05, "loss": 0.1701, "step": 1616 }, { "epoch": 1.28, "learning_rate": 1.2840767178187657e-05, "loss": 0.1542, "step": 1617 }, { "epoch": 1.28, "learning_rate": 1.2832583140188667e-05, "loss": 0.139, "step": 1618 }, { "epoch": 1.28, "learning_rate": 1.282439703894492e-05, "loss": 0.1571, "step": 1619 }, { "epoch": 1.28, "learning_rate": 1.281620888041915e-05, "loss": 0.1587, "step": 1620 }, { "epoch": 1.28, "learning_rate": 1.2808018670575582e-05, "loss": 0.1601, "step": 1621 }, { "epoch": 1.28, "learning_rate": 1.279982641537994e-05, "loss": 0.1298, "step": 1622 }, { "epoch": 1.28, "learning_rate": 1.279163212079944e-05, "loss": 0.1595, "step": 1623 }, { "epoch": 1.28, "learning_rate": 1.278343579280278e-05, "loss": 0.1743, "step": 1624 }, { "epoch": 1.28, "learning_rate": 1.2775237437360137e-05, "loss": 0.1597, "step": 1625 }, { "epoch": 1.28, "learning_rate": 1.2767037060443173e-05, "loss": 0.1566, "step": 1626 }, { "epoch": 1.29, "learning_rate": 1.2758834668025011e-05, "loss": 0.1484, "step": 1627 }, { "epoch": 1.29, "learning_rate": 1.275063026608025e-05, "loss": 0.1418, "step": 1628 }, { "epoch": 1.29, "learning_rate": 1.2742423860584954e-05, "loss": 0.1455, "step": 1629 }, { "epoch": 1.29, "learning_rate": 1.2734215457516639e-05, "loss": 0.156, "step": 1630 }, { "epoch": 1.29, "learning_rate": 1.2726005062854285e-05, "loss": 0.1328, "step": 1631 }, { "epoch": 1.29, "learning_rate": 1.271779268257831e-05, "loss": 0.1587, "step": 1632 }, { "epoch": 1.29, "learning_rate": 1.2709578322670597e-05, "loss": 0.1432, "step": 1633 }, { "epoch": 1.29, "learning_rate": 1.2701361989114454e-05, "loss": 0.1358, "step": 1634 }, { "epoch": 1.29, "learning_rate": 1.269314368789463e-05, "loss": 0.1606, "step": 1635 }, { "epoch": 1.29, "learning_rate": 1.2684923424997315e-05, "loss": 0.1711, "step": 1636 }, { "epoch": 1.29, "learning_rate": 1.2676701206410123e-05, "loss": 0.1794, "step": 1637 }, { "epoch": 1.29, "learning_rate": 1.266847703812209e-05, "loss": 0.1253, "step": 1638 }, { "epoch": 1.3, "learning_rate": 1.2660250926123676e-05, "loss": 0.1449, "step": 1639 }, { "epoch": 1.3, "learning_rate": 1.2652022876406756e-05, "loss": 0.1646, "step": 1640 }, { "epoch": 1.3, "learning_rate": 1.2643792894964611e-05, "loss": 0.1414, "step": 1641 }, { "epoch": 1.3, "learning_rate": 1.2635560987791941e-05, "loss": 0.1463, "step": 1642 }, { "epoch": 1.3, "learning_rate": 1.2627327160884839e-05, "loss": 0.1436, "step": 1643 }, { "epoch": 1.3, "learning_rate": 1.2619091420240795e-05, "loss": 0.1671, "step": 1644 }, { "epoch": 1.3, "learning_rate": 1.2610853771858702e-05, "loss": 0.1532, "step": 1645 }, { "epoch": 1.3, "learning_rate": 1.2602614221738838e-05, "loss": 0.1558, "step": 1646 }, { "epoch": 1.3, "learning_rate": 1.2594372775882862e-05, "loss": 0.1526, "step": 1647 }, { "epoch": 1.3, "learning_rate": 1.2586129440293817e-05, "loss": 0.1557, "step": 1648 }, { "epoch": 1.3, "learning_rate": 1.2577884220976127e-05, "loss": 0.1417, "step": 1649 }, { "epoch": 1.3, "learning_rate": 1.2569637123935581e-05, "loss": 0.1466, "step": 1650 }, { "epoch": 1.3, "learning_rate": 1.2561388155179336e-05, "loss": 0.1485, "step": 1651 }, { "epoch": 1.31, "learning_rate": 1.2553137320715927e-05, "loss": 0.1715, "step": 1652 }, { "epoch": 1.31, "learning_rate": 1.2544884626555225e-05, "loss": 0.1709, "step": 1653 }, { "epoch": 1.31, "learning_rate": 1.2536630078708466e-05, "loss": 0.1591, "step": 1654 }, { "epoch": 1.31, "learning_rate": 1.2528373683188247e-05, "loss": 0.1565, "step": 1655 }, { "epoch": 1.31, "learning_rate": 1.2520115446008493e-05, "loss": 0.1662, "step": 1656 }, { "epoch": 1.31, "learning_rate": 1.2511855373184478e-05, "loss": 0.1651, "step": 1657 }, { "epoch": 1.31, "learning_rate": 1.2503593470732818e-05, "loss": 0.1465, "step": 1658 }, { "epoch": 1.31, "learning_rate": 1.2495329744671457e-05, "loss": 0.1392, "step": 1659 }, { "epoch": 1.31, "learning_rate": 1.248706420101966e-05, "loss": 0.1439, "step": 1660 }, { "epoch": 1.31, "learning_rate": 1.2478796845798031e-05, "loss": 0.1607, "step": 1661 }, { "epoch": 1.31, "learning_rate": 1.2470527685028485e-05, "loss": 0.1403, "step": 1662 }, { "epoch": 1.31, "learning_rate": 1.2462256724734248e-05, "loss": 0.1719, "step": 1663 }, { "epoch": 1.31, "learning_rate": 1.2453983970939864e-05, "loss": 0.1563, "step": 1664 }, { "epoch": 1.32, "learning_rate": 1.2445709429671184e-05, "loss": 0.1391, "step": 1665 }, { "epoch": 1.32, "learning_rate": 1.243743310695536e-05, "loss": 0.1388, "step": 1666 }, { "epoch": 1.32, "learning_rate": 1.242915500882083e-05, "loss": 0.1312, "step": 1667 }, { "epoch": 1.32, "learning_rate": 1.2420875141297344e-05, "loss": 0.1343, "step": 1668 }, { "epoch": 1.32, "learning_rate": 1.241259351041593e-05, "loss": 0.1497, "step": 1669 }, { "epoch": 1.32, "learning_rate": 1.2404310122208895e-05, "loss": 0.1456, "step": 1670 }, { "epoch": 1.32, "learning_rate": 1.2396024982709845e-05, "loss": 0.1827, "step": 1671 }, { "epoch": 1.32, "learning_rate": 1.2387738097953644e-05, "loss": 0.1601, "step": 1672 }, { "epoch": 1.32, "learning_rate": 1.2379449473976428e-05, "loss": 0.156, "step": 1673 }, { "epoch": 1.32, "learning_rate": 1.2371159116815614e-05, "loss": 0.146, "step": 1674 }, { "epoch": 1.32, "learning_rate": 1.2362867032509871e-05, "loss": 0.1393, "step": 1675 }, { "epoch": 1.32, "learning_rate": 1.235457322709912e-05, "loss": 0.1397, "step": 1676 }, { "epoch": 1.33, "learning_rate": 1.234627770662455e-05, "loss": 0.141, "step": 1677 }, { "epoch": 1.33, "learning_rate": 1.233798047712859e-05, "loss": 0.1433, "step": 1678 }, { "epoch": 1.33, "learning_rate": 1.2329681544654919e-05, "loss": 0.1647, "step": 1679 }, { "epoch": 1.33, "learning_rate": 1.2321380915248446e-05, "loss": 0.1607, "step": 1680 }, { "epoch": 1.33, "learning_rate": 1.2313078594955333e-05, "loss": 0.1473, "step": 1681 }, { "epoch": 1.33, "learning_rate": 1.2304774589822958e-05, "loss": 0.1568, "step": 1682 }, { "epoch": 1.33, "learning_rate": 1.2296468905899937e-05, "loss": 0.1642, "step": 1683 }, { "epoch": 1.33, "learning_rate": 1.2288161549236099e-05, "loss": 0.1716, "step": 1684 }, { "epoch": 1.33, "learning_rate": 1.2279852525882504e-05, "loss": 0.1853, "step": 1685 }, { "epoch": 1.33, "learning_rate": 1.227154184189141e-05, "loss": 0.1246, "step": 1686 }, { "epoch": 1.33, "learning_rate": 1.22632295033163e-05, "loss": 0.1544, "step": 1687 }, { "epoch": 1.33, "learning_rate": 1.2254915516211854e-05, "loss": 0.1509, "step": 1688 }, { "epoch": 1.33, "learning_rate": 1.2246599886633951e-05, "loss": 0.1646, "step": 1689 }, { "epoch": 1.34, "learning_rate": 1.2238282620639677e-05, "loss": 0.1463, "step": 1690 }, { "epoch": 1.34, "learning_rate": 1.2229963724287294e-05, "loss": 0.1499, "step": 1691 }, { "epoch": 1.34, "learning_rate": 1.222164320363627e-05, "loss": 0.1254, "step": 1692 }, { "epoch": 1.34, "learning_rate": 1.2213321064747234e-05, "loss": 0.1652, "step": 1693 }, { "epoch": 1.34, "learning_rate": 1.2204997313682019e-05, "loss": 0.161, "step": 1694 }, { "epoch": 1.34, "learning_rate": 1.2196671956503611e-05, "loss": 0.1429, "step": 1695 }, { "epoch": 1.34, "learning_rate": 1.2188344999276176e-05, "loss": 0.1383, "step": 1696 }, { "epoch": 1.34, "learning_rate": 1.2180016448065046e-05, "loss": 0.1444, "step": 1697 }, { "epoch": 1.34, "learning_rate": 1.217168630893671e-05, "loss": 0.1545, "step": 1698 }, { "epoch": 1.34, "learning_rate": 1.2163354587958813e-05, "loss": 0.1461, "step": 1699 }, { "epoch": 1.34, "learning_rate": 1.2155021291200161e-05, "loss": 0.1669, "step": 1700 }, { "epoch": 1.34, "learning_rate": 1.2146686424730699e-05, "loss": 0.1721, "step": 1701 }, { "epoch": 1.34, "learning_rate": 1.2138349994621512e-05, "loss": 0.127, "step": 1702 }, { "epoch": 1.35, "learning_rate": 1.213001200694484e-05, "loss": 0.159, "step": 1703 }, { "epoch": 1.35, "learning_rate": 1.212167246777404e-05, "loss": 0.1436, "step": 1704 }, { "epoch": 1.35, "learning_rate": 1.2113331383183607e-05, "loss": 0.137, "step": 1705 }, { "epoch": 1.35, "learning_rate": 1.210498875924916e-05, "loss": 0.1466, "step": 1706 }, { "epoch": 1.35, "learning_rate": 1.2096644602047447e-05, "loss": 0.1362, "step": 1707 }, { "epoch": 1.35, "learning_rate": 1.2088298917656322e-05, "loss": 0.1578, "step": 1708 }, { "epoch": 1.35, "learning_rate": 1.2079951712154752e-05, "loss": 0.1295, "step": 1709 }, { "epoch": 1.35, "learning_rate": 1.2071602991622822e-05, "loss": 0.1506, "step": 1710 }, { "epoch": 1.35, "learning_rate": 1.206325276214171e-05, "loss": 0.1726, "step": 1711 }, { "epoch": 1.35, "learning_rate": 1.2054901029793698e-05, "loss": 0.1418, "step": 1712 }, { "epoch": 1.35, "learning_rate": 1.2046547800662163e-05, "loss": 0.17, "step": 1713 }, { "epoch": 1.35, "learning_rate": 1.203819308083157e-05, "loss": 0.1469, "step": 1714 }, { "epoch": 1.36, "learning_rate": 1.202983687638747e-05, "loss": 0.1393, "step": 1715 }, { "epoch": 1.36, "learning_rate": 1.2021479193416502e-05, "loss": 0.1554, "step": 1716 }, { "epoch": 1.36, "learning_rate": 1.2013120038006367e-05, "loss": 0.1698, "step": 1717 }, { "epoch": 1.36, "learning_rate": 1.2004759416245853e-05, "loss": 0.1366, "step": 1718 }, { "epoch": 1.36, "learning_rate": 1.1996397334224814e-05, "loss": 0.1546, "step": 1719 }, { "epoch": 1.36, "learning_rate": 1.198803379803416e-05, "loss": 0.1534, "step": 1720 }, { "epoch": 1.36, "learning_rate": 1.1979668813765863e-05, "loss": 0.1699, "step": 1721 }, { "epoch": 1.36, "learning_rate": 1.1971302387512958e-05, "loss": 0.1579, "step": 1722 }, { "epoch": 1.36, "learning_rate": 1.1962934525369519e-05, "loss": 0.1691, "step": 1723 }, { "epoch": 1.36, "learning_rate": 1.1954565233430672e-05, "loss": 0.1493, "step": 1724 }, { "epoch": 1.36, "learning_rate": 1.1946194517792584e-05, "loss": 0.1556, "step": 1725 }, { "epoch": 1.36, "learning_rate": 1.1937822384552456e-05, "loss": 0.1401, "step": 1726 }, { "epoch": 1.36, "learning_rate": 1.1929448839808528e-05, "loss": 0.1608, "step": 1727 }, { "epoch": 1.37, "learning_rate": 1.1921073889660061e-05, "loss": 0.1622, "step": 1728 }, { "epoch": 1.37, "learning_rate": 1.1912697540207341e-05, "loss": 0.1497, "step": 1729 }, { "epoch": 1.37, "learning_rate": 1.190431979755168e-05, "loss": 0.1528, "step": 1730 }, { "epoch": 1.37, "learning_rate": 1.1895940667795395e-05, "loss": 0.1437, "step": 1731 }, { "epoch": 1.37, "learning_rate": 1.1887560157041823e-05, "loss": 0.1497, "step": 1732 }, { "epoch": 1.37, "learning_rate": 1.1879178271395295e-05, "loss": 0.1511, "step": 1733 }, { "epoch": 1.37, "learning_rate": 1.1870795016961157e-05, "loss": 0.1532, "step": 1734 }, { "epoch": 1.37, "learning_rate": 1.1862410399845739e-05, "loss": 0.1635, "step": 1735 }, { "epoch": 1.37, "learning_rate": 1.1854024426156376e-05, "loss": 0.1154, "step": 1736 }, { "epoch": 1.37, "learning_rate": 1.1845637102001383e-05, "loss": 0.182, "step": 1737 }, { "epoch": 1.37, "learning_rate": 1.183724843349006e-05, "loss": 0.1406, "step": 1738 }, { "epoch": 1.37, "learning_rate": 1.182885842673269e-05, "loss": 0.1397, "step": 1739 }, { "epoch": 1.37, "learning_rate": 1.1820467087840526e-05, "loss": 0.1698, "step": 1740 }, { "epoch": 1.38, "learning_rate": 1.181207442292579e-05, "loss": 0.1488, "step": 1741 }, { "epoch": 1.38, "learning_rate": 1.1803680438101678e-05, "loss": 0.1535, "step": 1742 }, { "epoch": 1.38, "learning_rate": 1.1795285139482341e-05, "loss": 0.1331, "step": 1743 }, { "epoch": 1.38, "learning_rate": 1.1786888533182886e-05, "loss": 0.1408, "step": 1744 }, { "epoch": 1.38, "learning_rate": 1.1778490625319376e-05, "loss": 0.1473, "step": 1745 }, { "epoch": 1.38, "learning_rate": 1.1770091422008824e-05, "loss": 0.1368, "step": 1746 }, { "epoch": 1.38, "learning_rate": 1.176169092936918e-05, "loss": 0.1496, "step": 1747 }, { "epoch": 1.38, "learning_rate": 1.1753289153519341e-05, "loss": 0.1615, "step": 1748 }, { "epoch": 1.38, "learning_rate": 1.174488610057913e-05, "loss": 0.1326, "step": 1749 }, { "epoch": 1.38, "learning_rate": 1.1736481776669307e-05, "loss": 0.1528, "step": 1750 }, { "epoch": 1.38, "learning_rate": 1.172807618791155e-05, "loss": 0.1579, "step": 1751 }, { "epoch": 1.38, "learning_rate": 1.1719669340428472e-05, "loss": 0.1555, "step": 1752 }, { "epoch": 1.39, "learning_rate": 1.171126124034359e-05, "loss": 0.1329, "step": 1753 }, { "epoch": 1.39, "learning_rate": 1.1702851893781338e-05, "loss": 0.1461, "step": 1754 }, { "epoch": 1.39, "learning_rate": 1.1694441306867062e-05, "loss": 0.1322, "step": 1755 }, { "epoch": 1.39, "learning_rate": 1.1686029485727002e-05, "loss": 0.1425, "step": 1756 }, { "epoch": 1.39, "learning_rate": 1.1677616436488302e-05, "loss": 0.1648, "step": 1757 }, { "epoch": 1.39, "learning_rate": 1.1669202165279009e-05, "loss": 0.1461, "step": 1758 }, { "epoch": 1.39, "learning_rate": 1.1660786678228043e-05, "loss": 0.1468, "step": 1759 }, { "epoch": 1.39, "learning_rate": 1.1652369981465218e-05, "loss": 0.1656, "step": 1760 }, { "epoch": 1.39, "learning_rate": 1.164395208112124e-05, "loss": 0.1519, "step": 1761 }, { "epoch": 1.39, "learning_rate": 1.1635532983327674e-05, "loss": 0.1448, "step": 1762 }, { "epoch": 1.39, "learning_rate": 1.1627112694216967e-05, "loss": 0.1476, "step": 1763 }, { "epoch": 1.39, "learning_rate": 1.1618691219922426e-05, "loss": 0.144, "step": 1764 }, { "epoch": 1.39, "learning_rate": 1.1610268566578233e-05, "loss": 0.1412, "step": 1765 }, { "epoch": 1.4, "learning_rate": 1.1601844740319423e-05, "loss": 0.1456, "step": 1766 }, { "epoch": 1.4, "learning_rate": 1.159341974728188e-05, "loss": 0.1468, "step": 1767 }, { "epoch": 1.4, "learning_rate": 1.1584993593602345e-05, "loss": 0.1786, "step": 1768 }, { "epoch": 1.4, "learning_rate": 1.1576566285418404e-05, "loss": 0.1481, "step": 1769 }, { "epoch": 1.4, "learning_rate": 1.1568137828868478e-05, "loss": 0.1506, "step": 1770 }, { "epoch": 1.4, "learning_rate": 1.1559708230091834e-05, "loss": 0.1558, "step": 1771 }, { "epoch": 1.4, "learning_rate": 1.1551277495228561e-05, "loss": 0.1631, "step": 1772 }, { "epoch": 1.4, "learning_rate": 1.1542845630419579e-05, "loss": 0.1639, "step": 1773 }, { "epoch": 1.4, "learning_rate": 1.1534412641806638e-05, "loss": 0.1268, "step": 1774 }, { "epoch": 1.4, "learning_rate": 1.15259785355323e-05, "loss": 0.1233, "step": 1775 }, { "epoch": 1.4, "learning_rate": 1.1517543317739931e-05, "loss": 0.1551, "step": 1776 }, { "epoch": 1.4, "learning_rate": 1.1509106994573731e-05, "loss": 0.148, "step": 1777 }, { "epoch": 1.4, "learning_rate": 1.1500669572178686e-05, "loss": 0.1508, "step": 1778 }, { "epoch": 1.41, "learning_rate": 1.1492231056700592e-05, "loss": 0.1319, "step": 1779 }, { "epoch": 1.41, "learning_rate": 1.1483791454286027e-05, "loss": 0.1662, "step": 1780 }, { "epoch": 1.41, "learning_rate": 1.1475350771082381e-05, "loss": 0.1555, "step": 1781 }, { "epoch": 1.41, "learning_rate": 1.1466909013237819e-05, "loss": 0.1715, "step": 1782 }, { "epoch": 1.41, "learning_rate": 1.1458466186901288e-05, "loss": 0.1332, "step": 1783 }, { "epoch": 1.41, "learning_rate": 1.1450022298222522e-05, "loss": 0.1568, "step": 1784 }, { "epoch": 1.41, "learning_rate": 1.1441577353352023e-05, "loss": 0.1437, "step": 1785 }, { "epoch": 1.41, "learning_rate": 1.1433131358441054e-05, "loss": 0.1544, "step": 1786 }, { "epoch": 1.41, "learning_rate": 1.1424684319641661e-05, "loss": 0.1271, "step": 1787 }, { "epoch": 1.41, "learning_rate": 1.1416236243106638e-05, "loss": 0.1434, "step": 1788 }, { "epoch": 1.41, "learning_rate": 1.1407787134989532e-05, "loss": 0.1391, "step": 1789 }, { "epoch": 1.41, "learning_rate": 1.1399337001444658e-05, "loss": 0.1379, "step": 1790 }, { "epoch": 1.42, "learning_rate": 1.1390885848627058e-05, "loss": 0.1517, "step": 1791 }, { "epoch": 1.42, "learning_rate": 1.1382433682692531e-05, "loss": 0.164, "step": 1792 }, { "epoch": 1.42, "learning_rate": 1.1373980509797601e-05, "loss": 0.1286, "step": 1793 }, { "epoch": 1.42, "learning_rate": 1.1365526336099542e-05, "loss": 0.1305, "step": 1794 }, { "epoch": 1.42, "learning_rate": 1.1357071167756341e-05, "loss": 0.1434, "step": 1795 }, { "epoch": 1.42, "learning_rate": 1.134861501092672e-05, "loss": 0.1469, "step": 1796 }, { "epoch": 1.42, "learning_rate": 1.1340157871770117e-05, "loss": 0.1448, "step": 1797 }, { "epoch": 1.42, "learning_rate": 1.1331699756446686e-05, "loss": 0.1539, "step": 1798 }, { "epoch": 1.42, "learning_rate": 1.1323240671117289e-05, "loss": 0.1462, "step": 1799 }, { "epoch": 1.42, "learning_rate": 1.13147806219435e-05, "loss": 0.1557, "step": 1800 }, { "epoch": 1.42, "learning_rate": 1.1306319615087594e-05, "loss": 0.1548, "step": 1801 }, { "epoch": 1.42, "learning_rate": 1.1297857656712536e-05, "loss": 0.1529, "step": 1802 }, { "epoch": 1.42, "learning_rate": 1.1289394752982e-05, "loss": 0.1522, "step": 1803 }, { "epoch": 1.43, "learning_rate": 1.1280930910060328e-05, "loss": 0.149, "step": 1804 }, { "epoch": 1.43, "learning_rate": 1.1272466134112562e-05, "loss": 0.1421, "step": 1805 }, { "epoch": 1.43, "learning_rate": 1.1264000431304422e-05, "loss": 0.1609, "step": 1806 }, { "epoch": 1.43, "learning_rate": 1.1255533807802297e-05, "loss": 0.1602, "step": 1807 }, { "epoch": 1.43, "learning_rate": 1.1247066269773246e-05, "loss": 0.1517, "step": 1808 }, { "epoch": 1.43, "learning_rate": 1.1238597823385e-05, "loss": 0.1381, "step": 1809 }, { "epoch": 1.43, "learning_rate": 1.1230128474805948e-05, "loss": 0.1464, "step": 1810 }, { "epoch": 1.43, "learning_rate": 1.1221658230205142e-05, "loss": 0.1298, "step": 1811 }, { "epoch": 1.43, "learning_rate": 1.1213187095752271e-05, "loss": 0.1527, "step": 1812 }, { "epoch": 1.43, "learning_rate": 1.1204715077617694e-05, "loss": 0.1545, "step": 1813 }, { "epoch": 1.43, "learning_rate": 1.1196242181972399e-05, "loss": 0.1477, "step": 1814 }, { "epoch": 1.43, "learning_rate": 1.1187768414988015e-05, "loss": 0.1421, "step": 1815 }, { "epoch": 1.44, "learning_rate": 1.1179293782836809e-05, "loss": 0.1583, "step": 1816 }, { "epoch": 1.44, "learning_rate": 1.1170818291691678e-05, "loss": 0.1399, "step": 1817 }, { "epoch": 1.44, "learning_rate": 1.1162341947726139e-05, "loss": 0.1431, "step": 1818 }, { "epoch": 1.44, "learning_rate": 1.1153864757114342e-05, "loss": 0.1526, "step": 1819 }, { "epoch": 1.44, "learning_rate": 1.114538672603104e-05, "loss": 0.1623, "step": 1820 }, { "epoch": 1.44, "learning_rate": 1.1136907860651603e-05, "loss": 0.154, "step": 1821 }, { "epoch": 1.44, "learning_rate": 1.1128428167152016e-05, "loss": 0.152, "step": 1822 }, { "epoch": 1.44, "learning_rate": 1.1119947651708863e-05, "loss": 0.1499, "step": 1823 }, { "epoch": 1.44, "learning_rate": 1.1111466320499318e-05, "loss": 0.1621, "step": 1824 }, { "epoch": 1.44, "learning_rate": 1.1102984179701157e-05, "loss": 0.1233, "step": 1825 }, { "epoch": 1.44, "learning_rate": 1.109450123549275e-05, "loss": 0.1276, "step": 1826 }, { "epoch": 1.44, "learning_rate": 1.1086017494053046e-05, "loss": 0.1536, "step": 1827 }, { "epoch": 1.44, "learning_rate": 1.107753296156157e-05, "loss": 0.1742, "step": 1828 }, { "epoch": 1.45, "learning_rate": 1.1069047644198437e-05, "loss": 0.1478, "step": 1829 }, { "epoch": 1.45, "learning_rate": 1.1060561548144321e-05, "loss": 0.1563, "step": 1830 }, { "epoch": 1.45, "learning_rate": 1.1052074679580469e-05, "loss": 0.137, "step": 1831 }, { "epoch": 1.45, "learning_rate": 1.1043587044688692e-05, "loss": 0.1452, "step": 1832 }, { "epoch": 1.45, "learning_rate": 1.1035098649651355e-05, "loss": 0.1296, "step": 1833 }, { "epoch": 1.45, "learning_rate": 1.1026609500651375e-05, "loss": 0.1658, "step": 1834 }, { "epoch": 1.45, "learning_rate": 1.1018119603872228e-05, "loss": 0.1563, "step": 1835 }, { "epoch": 1.45, "learning_rate": 1.1009628965497927e-05, "loss": 0.1531, "step": 1836 }, { "epoch": 1.45, "learning_rate": 1.1001137591713024e-05, "loss": 0.132, "step": 1837 }, { "epoch": 1.45, "learning_rate": 1.099264548870261e-05, "loss": 0.1405, "step": 1838 }, { "epoch": 1.45, "learning_rate": 1.0984152662652307e-05, "loss": 0.1188, "step": 1839 }, { "epoch": 1.45, "learning_rate": 1.0975659119748265e-05, "loss": 0.1514, "step": 1840 }, { "epoch": 1.45, "learning_rate": 1.096716486617715e-05, "loss": 0.1472, "step": 1841 }, { "epoch": 1.46, "learning_rate": 1.0958669908126151e-05, "loss": 0.1514, "step": 1842 }, { "epoch": 1.46, "learning_rate": 1.0950174251782974e-05, "loss": 0.1607, "step": 1843 }, { "epoch": 1.46, "learning_rate": 1.0941677903335823e-05, "loss": 0.1492, "step": 1844 }, { "epoch": 1.46, "learning_rate": 1.0933180868973414e-05, "loss": 0.1294, "step": 1845 }, { "epoch": 1.46, "learning_rate": 1.092468315488496e-05, "loss": 0.1518, "step": 1846 }, { "epoch": 1.46, "learning_rate": 1.0916184767260169e-05, "loss": 0.1414, "step": 1847 }, { "epoch": 1.46, "learning_rate": 1.0907685712289244e-05, "loss": 0.1251, "step": 1848 }, { "epoch": 1.46, "learning_rate": 1.0899185996162865e-05, "loss": 0.1553, "step": 1849 }, { "epoch": 1.46, "learning_rate": 1.08906856250722e-05, "loss": 0.1801, "step": 1850 }, { "epoch": 1.46, "learning_rate": 1.0882184605208895e-05, "loss": 0.1371, "step": 1851 }, { "epoch": 1.46, "learning_rate": 1.0873682942765065e-05, "loss": 0.1513, "step": 1852 }, { "epoch": 1.46, "learning_rate": 1.0865180643933296e-05, "loss": 0.152, "step": 1853 }, { "epoch": 1.47, "learning_rate": 1.0856677714906632e-05, "loss": 0.1349, "step": 1854 }, { "epoch": 1.47, "learning_rate": 1.0848174161878584e-05, "loss": 0.1581, "step": 1855 }, { "epoch": 1.47, "learning_rate": 1.0839669991043113e-05, "loss": 0.1363, "step": 1856 }, { "epoch": 1.47, "learning_rate": 1.083116520859463e-05, "loss": 0.1525, "step": 1857 }, { "epoch": 1.47, "learning_rate": 1.0822659820727992e-05, "loss": 0.1263, "step": 1858 }, { "epoch": 1.47, "learning_rate": 1.0814153833638499e-05, "loss": 0.1547, "step": 1859 }, { "epoch": 1.47, "learning_rate": 1.080564725352188e-05, "loss": 0.1496, "step": 1860 }, { "epoch": 1.47, "learning_rate": 1.0797140086574309e-05, "loss": 0.1491, "step": 1861 }, { "epoch": 1.47, "learning_rate": 1.0788632338992375e-05, "loss": 0.1451, "step": 1862 }, { "epoch": 1.47, "learning_rate": 1.0780124016973095e-05, "loss": 0.1233, "step": 1863 }, { "epoch": 1.47, "learning_rate": 1.0771615126713908e-05, "loss": 0.1346, "step": 1864 }, { "epoch": 1.47, "learning_rate": 1.076310567441266e-05, "loss": 0.1411, "step": 1865 }, { "epoch": 1.47, "learning_rate": 1.0754595666267609e-05, "loss": 0.1569, "step": 1866 }, { "epoch": 1.48, "learning_rate": 1.074608510847742e-05, "loss": 0.1562, "step": 1867 }, { "epoch": 1.48, "learning_rate": 1.0737574007241158e-05, "loss": 0.1433, "step": 1868 }, { "epoch": 1.48, "learning_rate": 1.0729062368758278e-05, "loss": 0.156, "step": 1869 }, { "epoch": 1.48, "learning_rate": 1.072055019922864e-05, "loss": 0.1263, "step": 1870 }, { "epoch": 1.48, "learning_rate": 1.0712037504852473e-05, "loss": 0.153, "step": 1871 }, { "epoch": 1.48, "learning_rate": 1.0703524291830398e-05, "loss": 0.1772, "step": 1872 }, { "epoch": 1.48, "learning_rate": 1.069501056636342e-05, "loss": 0.1717, "step": 1873 }, { "epoch": 1.48, "learning_rate": 1.0686496334652905e-05, "loss": 0.1384, "step": 1874 }, { "epoch": 1.48, "learning_rate": 1.067798160290059e-05, "loss": 0.1478, "step": 1875 }, { "epoch": 1.48, "learning_rate": 1.0669466377308584e-05, "loss": 0.144, "step": 1876 }, { "epoch": 1.48, "learning_rate": 1.066095066407935e-05, "loss": 0.1662, "step": 1877 }, { "epoch": 1.48, "learning_rate": 1.0652434469415705e-05, "loss": 0.1373, "step": 1878 }, { "epoch": 1.48, "learning_rate": 1.0643917799520822e-05, "loss": 0.1345, "step": 1879 }, { "epoch": 1.49, "learning_rate": 1.0635400660598214e-05, "loss": 0.1264, "step": 1880 }, { "epoch": 1.49, "learning_rate": 1.0626883058851737e-05, "loss": 0.1326, "step": 1881 }, { "epoch": 1.49, "learning_rate": 1.0618365000485593e-05, "loss": 0.1383, "step": 1882 }, { "epoch": 1.49, "learning_rate": 1.0609846491704303e-05, "loss": 0.1457, "step": 1883 }, { "epoch": 1.49, "learning_rate": 1.0601327538712723e-05, "loss": 0.1296, "step": 1884 }, { "epoch": 1.49, "learning_rate": 1.0592808147716032e-05, "loss": 0.1269, "step": 1885 }, { "epoch": 1.49, "learning_rate": 1.058428832491973e-05, "loss": 0.1422, "step": 1886 }, { "epoch": 1.49, "learning_rate": 1.0575768076529627e-05, "loss": 0.1355, "step": 1887 }, { "epoch": 1.49, "learning_rate": 1.0567247408751845e-05, "loss": 0.1597, "step": 1888 }, { "epoch": 1.49, "learning_rate": 1.0558726327792817e-05, "loss": 0.1817, "step": 1889 }, { "epoch": 1.49, "learning_rate": 1.0550204839859265e-05, "loss": 0.1295, "step": 1890 }, { "epoch": 1.49, "learning_rate": 1.0541682951158218e-05, "loss": 0.1306, "step": 1891 }, { "epoch": 1.5, "learning_rate": 1.0533160667896993e-05, "loss": 0.1559, "step": 1892 }, { "epoch": 1.5, "learning_rate": 1.0524637996283195e-05, "loss": 0.153, "step": 1893 }, { "epoch": 1.5, "learning_rate": 1.0516114942524713e-05, "loss": 0.1574, "step": 1894 }, { "epoch": 1.5, "learning_rate": 1.0507591512829707e-05, "loss": 0.1296, "step": 1895 }, { "epoch": 1.5, "learning_rate": 1.0499067713406622e-05, "loss": 0.1402, "step": 1896 }, { "epoch": 1.5, "learning_rate": 1.049054355046417e-05, "loss": 0.1554, "step": 1897 }, { "epoch": 1.5, "learning_rate": 1.0482019030211316e-05, "loss": 0.1402, "step": 1898 }, { "epoch": 1.5, "learning_rate": 1.0473494158857298e-05, "loss": 0.1496, "step": 1899 }, { "epoch": 1.5, "learning_rate": 1.0464968942611608e-05, "loss": 0.1432, "step": 1900 }, { "epoch": 1.5, "learning_rate": 1.0456443387683986e-05, "loss": 0.1453, "step": 1901 }, { "epoch": 1.5, "learning_rate": 1.0447917500284415e-05, "loss": 0.1493, "step": 1902 }, { "epoch": 1.5, "learning_rate": 1.0439391286623133e-05, "loss": 0.1442, "step": 1903 }, { "epoch": 1.5, "learning_rate": 1.04308647529106e-05, "loss": 0.1503, "step": 1904 }, { "epoch": 1.51, "learning_rate": 1.0422337905357523e-05, "loss": 0.1436, "step": 1905 }, { "epoch": 1.51, "learning_rate": 1.0413810750174827e-05, "loss": 0.134, "step": 1906 }, { "epoch": 1.51, "learning_rate": 1.0405283293573666e-05, "loss": 0.1417, "step": 1907 }, { "epoch": 1.51, "learning_rate": 1.0396755541765413e-05, "loss": 0.1324, "step": 1908 }, { "epoch": 1.51, "learning_rate": 1.0388227500961659e-05, "loss": 0.147, "step": 1909 }, { "epoch": 1.51, "learning_rate": 1.0379699177374199e-05, "loss": 0.149, "step": 1910 }, { "epoch": 1.51, "learning_rate": 1.0371170577215036e-05, "loss": 0.177, "step": 1911 }, { "epoch": 1.51, "learning_rate": 1.036264170669638e-05, "loss": 0.1503, "step": 1912 }, { "epoch": 1.51, "learning_rate": 1.0354112572030632e-05, "loss": 0.1564, "step": 1913 }, { "epoch": 1.51, "learning_rate": 1.0345583179430387e-05, "loss": 0.1597, "step": 1914 }, { "epoch": 1.51, "learning_rate": 1.0337053535108427e-05, "loss": 0.1811, "step": 1915 }, { "epoch": 1.51, "learning_rate": 1.0328523645277721e-05, "loss": 0.1264, "step": 1916 }, { "epoch": 1.51, "learning_rate": 1.0319993516151412e-05, "loss": 0.1321, "step": 1917 }, { "epoch": 1.52, "learning_rate": 1.031146315394282e-05, "loss": 0.156, "step": 1918 }, { "epoch": 1.52, "learning_rate": 1.0302932564865436e-05, "loss": 0.1438, "step": 1919 }, { "epoch": 1.52, "learning_rate": 1.0294401755132912e-05, "loss": 0.1716, "step": 1920 }, { "epoch": 1.52, "learning_rate": 1.0285870730959064e-05, "loss": 0.146, "step": 1921 }, { "epoch": 1.52, "learning_rate": 1.0277339498557866e-05, "loss": 0.1253, "step": 1922 }, { "epoch": 1.52, "learning_rate": 1.0268808064143438e-05, "loss": 0.1636, "step": 1923 }, { "epoch": 1.52, "learning_rate": 1.026027643393005e-05, "loss": 0.1617, "step": 1924 }, { "epoch": 1.52, "learning_rate": 1.0251744614132117e-05, "loss": 0.1399, "step": 1925 }, { "epoch": 1.52, "learning_rate": 1.0243212610964192e-05, "loss": 0.1638, "step": 1926 }, { "epoch": 1.52, "learning_rate": 1.0234680430640958e-05, "loss": 0.1371, "step": 1927 }, { "epoch": 1.52, "learning_rate": 1.0226148079377223e-05, "loss": 0.1549, "step": 1928 }, { "epoch": 1.52, "learning_rate": 1.0217615563387932e-05, "loss": 0.1503, "step": 1929 }, { "epoch": 1.53, "learning_rate": 1.0209082888888143e-05, "loss": 0.1354, "step": 1930 }, { "epoch": 1.53, "learning_rate": 1.0200550062093023e-05, "loss": 0.1329, "step": 1931 }, { "epoch": 1.53, "learning_rate": 1.0192017089217863e-05, "loss": 0.1703, "step": 1932 }, { "epoch": 1.53, "learning_rate": 1.0183483976478052e-05, "loss": 0.1452, "step": 1933 }, { "epoch": 1.53, "learning_rate": 1.0174950730089081e-05, "loss": 0.1311, "step": 1934 }, { "epoch": 1.53, "learning_rate": 1.0166417356266546e-05, "loss": 0.1423, "step": 1935 }, { "epoch": 1.53, "learning_rate": 1.0157883861226126e-05, "loss": 0.13, "step": 1936 }, { "epoch": 1.53, "learning_rate": 1.0149350251183587e-05, "loss": 0.162, "step": 1937 }, { "epoch": 1.53, "learning_rate": 1.0140816532354793e-05, "loss": 0.1368, "step": 1938 }, { "epoch": 1.53, "learning_rate": 1.0132282710955675e-05, "loss": 0.1538, "step": 1939 }, { "epoch": 1.53, "learning_rate": 1.0123748793202242e-05, "loss": 0.1337, "step": 1940 }, { "epoch": 1.53, "learning_rate": 1.0115214785310567e-05, "loss": 0.1226, "step": 1941 }, { "epoch": 1.53, "learning_rate": 1.0106680693496807e-05, "loss": 0.1394, "step": 1942 }, { "epoch": 1.54, "learning_rate": 1.009814652397716e-05, "loss": 0.157, "step": 1943 }, { "epoch": 1.54, "learning_rate": 1.0089612282967884e-05, "loss": 0.1461, "step": 1944 }, { "epoch": 1.54, "learning_rate": 1.0081077976685307e-05, "loss": 0.1244, "step": 1945 }, { "epoch": 1.54, "learning_rate": 1.0072543611345782e-05, "loss": 0.1507, "step": 1946 }, { "epoch": 1.54, "learning_rate": 1.0064009193165713e-05, "loss": 0.1247, "step": 1947 }, { "epoch": 1.54, "learning_rate": 1.0055474728361549e-05, "loss": 0.1342, "step": 1948 }, { "epoch": 1.54, "learning_rate": 1.0046940223149766e-05, "loss": 0.1263, "step": 1949 }, { "epoch": 1.54, "learning_rate": 1.0038405683746868e-05, "loss": 0.1321, "step": 1950 }, { "epoch": 1.54, "learning_rate": 1.002987111636939e-05, "loss": 0.1264, "step": 1951 }, { "epoch": 1.54, "learning_rate": 1.0021336527233885e-05, "loss": 0.1467, "step": 1952 }, { "epoch": 1.54, "learning_rate": 1.0012801922556918e-05, "loss": 0.1498, "step": 1953 }, { "epoch": 1.54, "learning_rate": 1.0004267308555071e-05, "loss": 0.1091, "step": 1954 }, { "epoch": 1.54, "learning_rate": 9.995732691444932e-06, "loss": 0.1451, "step": 1955 }, { "epoch": 1.55, "learning_rate": 9.987198077443085e-06, "loss": 0.1498, "step": 1956 }, { "epoch": 1.55, "learning_rate": 9.97866347276612e-06, "loss": 0.1478, "step": 1957 }, { "epoch": 1.55, "learning_rate": 9.970128883630611e-06, "loss": 0.1325, "step": 1958 }, { "epoch": 1.55, "learning_rate": 9.961594316253134e-06, "loss": 0.1425, "step": 1959 }, { "epoch": 1.55, "learning_rate": 9.953059776850238e-06, "loss": 0.1671, "step": 1960 }, { "epoch": 1.55, "learning_rate": 9.944525271638455e-06, "loss": 0.1284, "step": 1961 }, { "epoch": 1.55, "learning_rate": 9.93599080683429e-06, "loss": 0.1261, "step": 1962 }, { "epoch": 1.55, "learning_rate": 9.927456388654225e-06, "loss": 0.1345, "step": 1963 }, { "epoch": 1.55, "learning_rate": 9.918922023314698e-06, "loss": 0.1355, "step": 1964 }, { "epoch": 1.55, "learning_rate": 9.910387717032115e-06, "loss": 0.1426, "step": 1965 }, { "epoch": 1.55, "learning_rate": 9.901853476022844e-06, "loss": 0.157, "step": 1966 }, { "epoch": 1.55, "learning_rate": 9.893319306503198e-06, "loss": 0.1367, "step": 1967 }, { "epoch": 1.56, "learning_rate": 9.884785214689435e-06, "loss": 0.1415, "step": 1968 }, { "epoch": 1.56, "learning_rate": 9.876251206797764e-06, "loss": 0.1372, "step": 1969 }, { "epoch": 1.56, "learning_rate": 9.86771728904433e-06, "loss": 0.1359, "step": 1970 }, { "epoch": 1.56, "learning_rate": 9.859183467645207e-06, "loss": 0.1501, "step": 1971 }, { "epoch": 1.56, "learning_rate": 9.850649748816413e-06, "loss": 0.1659, "step": 1972 }, { "epoch": 1.56, "learning_rate": 9.842116138773879e-06, "loss": 0.1685, "step": 1973 }, { "epoch": 1.56, "learning_rate": 9.833582643733457e-06, "loss": 0.1552, "step": 1974 }, { "epoch": 1.56, "learning_rate": 9.82504926991092e-06, "loss": 0.1466, "step": 1975 }, { "epoch": 1.56, "learning_rate": 9.816516023521953e-06, "loss": 0.1481, "step": 1976 }, { "epoch": 1.56, "learning_rate": 9.807982910782142e-06, "loss": 0.1347, "step": 1977 }, { "epoch": 1.56, "learning_rate": 9.799449937906979e-06, "loss": 0.1673, "step": 1978 }, { "epoch": 1.56, "learning_rate": 9.790917111111862e-06, "loss": 0.1419, "step": 1979 }, { "epoch": 1.56, "learning_rate": 9.782384436612072e-06, "loss": 0.1389, "step": 1980 }, { "epoch": 1.57, "learning_rate": 9.77385192062278e-06, "loss": 0.1485, "step": 1981 }, { "epoch": 1.57, "learning_rate": 9.765319569359048e-06, "loss": 0.1359, "step": 1982 }, { "epoch": 1.57, "learning_rate": 9.756787389035813e-06, "loss": 0.1521, "step": 1983 }, { "epoch": 1.57, "learning_rate": 9.748255385867882e-06, "loss": 0.1457, "step": 1984 }, { "epoch": 1.57, "learning_rate": 9.73972356606995e-06, "loss": 0.1356, "step": 1985 }, { "epoch": 1.57, "learning_rate": 9.731191935856566e-06, "loss": 0.1586, "step": 1986 }, { "epoch": 1.57, "learning_rate": 9.722660501442138e-06, "loss": 0.1176, "step": 1987 }, { "epoch": 1.57, "learning_rate": 9.714129269040939e-06, "loss": 0.1284, "step": 1988 }, { "epoch": 1.57, "learning_rate": 9.705598244867093e-06, "loss": 0.1496, "step": 1989 }, { "epoch": 1.57, "learning_rate": 9.697067435134564e-06, "loss": 0.1722, "step": 1990 }, { "epoch": 1.57, "learning_rate": 9.688536846057181e-06, "loss": 0.1721, "step": 1991 }, { "epoch": 1.57, "learning_rate": 9.68000648384859e-06, "loss": 0.1628, "step": 1992 }, { "epoch": 1.57, "learning_rate": 9.671476354722282e-06, "loss": 0.1615, "step": 1993 }, { "epoch": 1.58, "learning_rate": 9.662946464891576e-06, "loss": 0.144, "step": 1994 }, { "epoch": 1.58, "learning_rate": 9.654416820569618e-06, "loss": 0.1517, "step": 1995 }, { "epoch": 1.58, "learning_rate": 9.645887427969373e-06, "loss": 0.154, "step": 1996 }, { "epoch": 1.58, "learning_rate": 9.63735829330362e-06, "loss": 0.1492, "step": 1997 }, { "epoch": 1.58, "learning_rate": 9.628829422784965e-06, "loss": 0.1359, "step": 1998 }, { "epoch": 1.58, "learning_rate": 9.620300822625805e-06, "loss": 0.1375, "step": 1999 }, { "epoch": 1.58, "learning_rate": 9.611772499038345e-06, "loss": 0.1487, "step": 2000 }, { "epoch": 1.58, "learning_rate": 9.603244458234589e-06, "loss": 0.1263, "step": 2001 }, { "epoch": 1.58, "learning_rate": 9.594716706426339e-06, "loss": 0.1342, "step": 2002 }, { "epoch": 1.58, "learning_rate": 9.586189249825174e-06, "loss": 0.1432, "step": 2003 }, { "epoch": 1.58, "learning_rate": 9.577662094642478e-06, "loss": 0.1535, "step": 2004 }, { "epoch": 1.58, "learning_rate": 9.569135247089401e-06, "loss": 0.1447, "step": 2005 }, { "epoch": 1.59, "learning_rate": 9.56060871337687e-06, "loss": 0.1658, "step": 2006 }, { "epoch": 1.59, "learning_rate": 9.552082499715588e-06, "loss": 0.1382, "step": 2007 }, { "epoch": 1.59, "learning_rate": 9.54355661231602e-06, "loss": 0.1575, "step": 2008 }, { "epoch": 1.59, "learning_rate": 9.535031057388396e-06, "loss": 0.1562, "step": 2009 }, { "epoch": 1.59, "learning_rate": 9.526505841142702e-06, "loss": 0.1477, "step": 2010 }, { "epoch": 1.59, "learning_rate": 9.517980969788687e-06, "loss": 0.1412, "step": 2011 }, { "epoch": 1.59, "learning_rate": 9.509456449535834e-06, "loss": 0.1404, "step": 2012 }, { "epoch": 1.59, "learning_rate": 9.50093228659338e-06, "loss": 0.1342, "step": 2013 }, { "epoch": 1.59, "learning_rate": 9.492408487170294e-06, "loss": 0.1489, "step": 2014 }, { "epoch": 1.59, "learning_rate": 9.48388505747529e-06, "loss": 0.1523, "step": 2015 }, { "epoch": 1.59, "learning_rate": 9.475362003716804e-06, "loss": 0.1504, "step": 2016 }, { "epoch": 1.59, "learning_rate": 9.466839332103007e-06, "loss": 0.1374, "step": 2017 }, { "epoch": 1.59, "learning_rate": 9.458317048841784e-06, "loss": 0.1231, "step": 2018 }, { "epoch": 1.6, "learning_rate": 9.449795160140737e-06, "loss": 0.1286, "step": 2019 }, { "epoch": 1.6, "learning_rate": 9.441273672207187e-06, "loss": 0.1272, "step": 2020 }, { "epoch": 1.6, "learning_rate": 9.432752591248156e-06, "loss": 0.1338, "step": 2021 }, { "epoch": 1.6, "learning_rate": 9.424231923470378e-06, "loss": 0.1269, "step": 2022 }, { "epoch": 1.6, "learning_rate": 9.415711675080272e-06, "loss": 0.1646, "step": 2023 }, { "epoch": 1.6, "learning_rate": 9.407191852283971e-06, "loss": 0.1429, "step": 2024 }, { "epoch": 1.6, "learning_rate": 9.39867246128728e-06, "loss": 0.1424, "step": 2025 }, { "epoch": 1.6, "learning_rate": 9.3901535082957e-06, "loss": 0.1317, "step": 2026 }, { "epoch": 1.6, "learning_rate": 9.38163499951441e-06, "loss": 0.1269, "step": 2027 }, { "epoch": 1.6, "learning_rate": 9.373116941148264e-06, "loss": 0.1205, "step": 2028 }, { "epoch": 1.6, "learning_rate": 9.364599339401788e-06, "loss": 0.1387, "step": 2029 }, { "epoch": 1.6, "learning_rate": 9.35608220047918e-06, "loss": 0.1391, "step": 2030 }, { "epoch": 1.6, "learning_rate": 9.347565530584299e-06, "loss": 0.1329, "step": 2031 }, { "epoch": 1.61, "learning_rate": 9.339049335920652e-06, "loss": 0.123, "step": 2032 }, { "epoch": 1.61, "learning_rate": 9.33053362269142e-06, "loss": 0.1294, "step": 2033 }, { "epoch": 1.61, "learning_rate": 9.322018397099414e-06, "loss": 0.1233, "step": 2034 }, { "epoch": 1.61, "learning_rate": 9.313503665347099e-06, "loss": 0.1382, "step": 2035 }, { "epoch": 1.61, "learning_rate": 9.304989433636582e-06, "loss": 0.1372, "step": 2036 }, { "epoch": 1.61, "learning_rate": 9.296475708169603e-06, "loss": 0.1222, "step": 2037 }, { "epoch": 1.61, "learning_rate": 9.28796249514753e-06, "loss": 0.1507, "step": 2038 }, { "epoch": 1.61, "learning_rate": 9.279449800771364e-06, "loss": 0.1501, "step": 2039 }, { "epoch": 1.61, "learning_rate": 9.270937631241723e-06, "loss": 0.1656, "step": 2040 }, { "epoch": 1.61, "learning_rate": 9.262425992758847e-06, "loss": 0.1381, "step": 2041 }, { "epoch": 1.61, "learning_rate": 9.253914891522583e-06, "loss": 0.1384, "step": 2042 }, { "epoch": 1.61, "learning_rate": 9.245404333732395e-06, "loss": 0.1486, "step": 2043 }, { "epoch": 1.62, "learning_rate": 9.236894325587344e-06, "loss": 0.1266, "step": 2044 }, { "epoch": 1.62, "learning_rate": 9.228384873286096e-06, "loss": 0.1347, "step": 2045 }, { "epoch": 1.62, "learning_rate": 9.219875983026909e-06, "loss": 0.1321, "step": 2046 }, { "epoch": 1.62, "learning_rate": 9.21136766100763e-06, "loss": 0.1418, "step": 2047 }, { "epoch": 1.62, "learning_rate": 9.202859913425693e-06, "loss": 0.1721, "step": 2048 }, { "epoch": 1.62, "learning_rate": 9.194352746478122e-06, "loss": 0.1342, "step": 2049 }, { "epoch": 1.62, "learning_rate": 9.185846166361505e-06, "loss": 0.16, "step": 2050 }, { "epoch": 1.62, "learning_rate": 9.17734017927201e-06, "loss": 0.1377, "step": 2051 }, { "epoch": 1.62, "learning_rate": 9.168834791405374e-06, "loss": 0.1465, "step": 2052 }, { "epoch": 1.62, "learning_rate": 9.16033000895689e-06, "loss": 0.1294, "step": 2053 }, { "epoch": 1.62, "learning_rate": 9.15182583812142e-06, "loss": 0.1376, "step": 2054 }, { "epoch": 1.62, "learning_rate": 9.143322285093371e-06, "loss": 0.1619, "step": 2055 }, { "epoch": 1.62, "learning_rate": 9.134819356066706e-06, "loss": 0.1389, "step": 2056 }, { "epoch": 1.63, "learning_rate": 9.126317057234937e-06, "loss": 0.1823, "step": 2057 }, { "epoch": 1.63, "learning_rate": 9.117815394791107e-06, "loss": 0.1422, "step": 2058 }, { "epoch": 1.63, "learning_rate": 9.109314374927803e-06, "loss": 0.1344, "step": 2059 }, { "epoch": 1.63, "learning_rate": 9.10081400383714e-06, "loss": 0.1791, "step": 2060 }, { "epoch": 1.63, "learning_rate": 9.092314287710757e-06, "loss": 0.1344, "step": 2061 }, { "epoch": 1.63, "learning_rate": 9.083815232739831e-06, "loss": 0.1287, "step": 2062 }, { "epoch": 1.63, "learning_rate": 9.075316845115043e-06, "loss": 0.1448, "step": 2063 }, { "epoch": 1.63, "learning_rate": 9.066819131026588e-06, "loss": 0.1464, "step": 2064 }, { "epoch": 1.63, "learning_rate": 9.05832209666418e-06, "loss": 0.1302, "step": 2065 }, { "epoch": 1.63, "learning_rate": 9.049825748217031e-06, "loss": 0.1478, "step": 2066 }, { "epoch": 1.63, "learning_rate": 9.041330091873852e-06, "loss": 0.1285, "step": 2067 }, { "epoch": 1.63, "learning_rate": 9.032835133822852e-06, "loss": 0.1381, "step": 2068 }, { "epoch": 1.63, "learning_rate": 9.024340880251738e-06, "loss": 0.1319, "step": 2069 }, { "epoch": 1.64, "learning_rate": 9.015847337347695e-06, "loss": 0.1539, "step": 2070 }, { "epoch": 1.64, "learning_rate": 9.007354511297393e-06, "loss": 0.1552, "step": 2071 }, { "epoch": 1.64, "learning_rate": 8.998862408286981e-06, "loss": 0.1311, "step": 2072 }, { "epoch": 1.64, "learning_rate": 8.990371034502078e-06, "loss": 0.1489, "step": 2073 }, { "epoch": 1.64, "learning_rate": 8.981880396127771e-06, "loss": 0.1283, "step": 2074 }, { "epoch": 1.64, "learning_rate": 8.973390499348625e-06, "loss": 0.1457, "step": 2075 }, { "epoch": 1.64, "learning_rate": 8.964901350348648e-06, "loss": 0.14, "step": 2076 }, { "epoch": 1.64, "learning_rate": 8.956412955311311e-06, "loss": 0.1317, "step": 2077 }, { "epoch": 1.64, "learning_rate": 8.947925320419533e-06, "loss": 0.1437, "step": 2078 }, { "epoch": 1.64, "learning_rate": 8.939438451855684e-06, "loss": 0.1471, "step": 2079 }, { "epoch": 1.64, "learning_rate": 8.930952355801563e-06, "loss": 0.1603, "step": 2080 }, { "epoch": 1.64, "learning_rate": 8.92246703843843e-06, "loss": 0.1438, "step": 2081 }, { "epoch": 1.65, "learning_rate": 8.913982505946958e-06, "loss": 0.1433, "step": 2082 }, { "epoch": 1.65, "learning_rate": 8.905498764507251e-06, "loss": 0.1321, "step": 2083 }, { "epoch": 1.65, "learning_rate": 8.897015820298845e-06, "loss": 0.1292, "step": 2084 }, { "epoch": 1.65, "learning_rate": 8.888533679500688e-06, "loss": 0.1457, "step": 2085 }, { "epoch": 1.65, "learning_rate": 8.880052348291142e-06, "loss": 0.1257, "step": 2086 }, { "epoch": 1.65, "learning_rate": 8.871571832847984e-06, "loss": 0.1592, "step": 2087 }, { "epoch": 1.65, "learning_rate": 8.863092139348397e-06, "loss": 0.1179, "step": 2088 }, { "epoch": 1.65, "learning_rate": 8.854613273968963e-06, "loss": 0.1193, "step": 2089 }, { "epoch": 1.65, "learning_rate": 8.846135242885663e-06, "loss": 0.1397, "step": 2090 }, { "epoch": 1.65, "learning_rate": 8.837658052273863e-06, "loss": 0.1385, "step": 2091 }, { "epoch": 1.65, "learning_rate": 8.829181708308327e-06, "loss": 0.1416, "step": 2092 }, { "epoch": 1.65, "learning_rate": 8.820706217163191e-06, "loss": 0.1742, "step": 2093 }, { "epoch": 1.65, "learning_rate": 8.812231585011987e-06, "loss": 0.1629, "step": 2094 }, { "epoch": 1.66, "learning_rate": 8.803757818027603e-06, "loss": 0.1335, "step": 2095 }, { "epoch": 1.66, "learning_rate": 8.795284922382308e-06, "loss": 0.129, "step": 2096 }, { "epoch": 1.66, "learning_rate": 8.78681290424773e-06, "loss": 0.1181, "step": 2097 }, { "epoch": 1.66, "learning_rate": 8.778341769794865e-06, "loss": 0.1351, "step": 2098 }, { "epoch": 1.66, "learning_rate": 8.769871525194056e-06, "loss": 0.1379, "step": 2099 }, { "epoch": 1.66, "learning_rate": 8.761402176615002e-06, "loss": 0.1255, "step": 2100 }, { "epoch": 1.66, "learning_rate": 8.752933730226756e-06, "loss": 0.1328, "step": 2101 }, { "epoch": 1.66, "learning_rate": 8.744466192197706e-06, "loss": 0.1333, "step": 2102 }, { "epoch": 1.66, "learning_rate": 8.735999568695581e-06, "loss": 0.1291, "step": 2103 }, { "epoch": 1.66, "learning_rate": 8.72753386588744e-06, "loss": 0.145, "step": 2104 }, { "epoch": 1.66, "learning_rate": 8.719069089939676e-06, "loss": 0.1563, "step": 2105 }, { "epoch": 1.66, "learning_rate": 8.710605247018002e-06, "loss": 0.1306, "step": 2106 }, { "epoch": 1.66, "learning_rate": 8.702142343287464e-06, "loss": 0.1397, "step": 2107 }, { "epoch": 1.67, "learning_rate": 8.69368038491241e-06, "loss": 0.1443, "step": 2108 }, { "epoch": 1.67, "learning_rate": 8.685219378056503e-06, "loss": 0.1478, "step": 2109 }, { "epoch": 1.67, "learning_rate": 8.676759328882715e-06, "loss": 0.1401, "step": 2110 }, { "epoch": 1.67, "learning_rate": 8.668300243553319e-06, "loss": 0.1477, "step": 2111 }, { "epoch": 1.67, "learning_rate": 8.659842128229886e-06, "loss": 0.1556, "step": 2112 }, { "epoch": 1.67, "learning_rate": 8.65138498907328e-06, "loss": 0.141, "step": 2113 }, { "epoch": 1.67, "learning_rate": 8.64292883224366e-06, "loss": 0.1489, "step": 2114 }, { "epoch": 1.67, "learning_rate": 8.634473663900461e-06, "loss": 0.143, "step": 2115 }, { "epoch": 1.67, "learning_rate": 8.626019490202402e-06, "loss": 0.137, "step": 2116 }, { "epoch": 1.67, "learning_rate": 8.617566317307475e-06, "loss": 0.115, "step": 2117 }, { "epoch": 1.67, "learning_rate": 8.609114151372947e-06, "loss": 0.1622, "step": 2118 }, { "epoch": 1.67, "learning_rate": 8.600662998555344e-06, "loss": 0.1305, "step": 2119 }, { "epoch": 1.68, "learning_rate": 8.592212865010468e-06, "loss": 0.1398, "step": 2120 }, { "epoch": 1.68, "learning_rate": 8.583763756893366e-06, "loss": 0.1331, "step": 2121 }, { "epoch": 1.68, "learning_rate": 8.575315680358342e-06, "loss": 0.1563, "step": 2122 }, { "epoch": 1.68, "learning_rate": 8.566868641558948e-06, "loss": 0.1206, "step": 2123 }, { "epoch": 1.68, "learning_rate": 8.558422646647984e-06, "loss": 0.1345, "step": 2124 }, { "epoch": 1.68, "learning_rate": 8.549977701777483e-06, "loss": 0.1428, "step": 2125 }, { "epoch": 1.68, "learning_rate": 8.541533813098712e-06, "loss": 0.1441, "step": 2126 }, { "epoch": 1.68, "learning_rate": 8.533090986762183e-06, "loss": 0.1337, "step": 2127 }, { "epoch": 1.68, "learning_rate": 8.524649228917622e-06, "loss": 0.1378, "step": 2128 }, { "epoch": 1.68, "learning_rate": 8.516208545713977e-06, "loss": 0.1151, "step": 2129 }, { "epoch": 1.68, "learning_rate": 8.507768943299415e-06, "loss": 0.1417, "step": 2130 }, { "epoch": 1.68, "learning_rate": 8.499330427821317e-06, "loss": 0.1257, "step": 2131 }, { "epoch": 1.68, "learning_rate": 8.490893005426269e-06, "loss": 0.1327, "step": 2132 }, { "epoch": 1.69, "learning_rate": 8.482456682260069e-06, "loss": 0.1319, "step": 2133 }, { "epoch": 1.69, "learning_rate": 8.474021464467706e-06, "loss": 0.1311, "step": 2134 }, { "epoch": 1.69, "learning_rate": 8.465587358193364e-06, "loss": 0.1357, "step": 2135 }, { "epoch": 1.69, "learning_rate": 8.457154369580424e-06, "loss": 0.1269, "step": 2136 }, { "epoch": 1.69, "learning_rate": 8.448722504771444e-06, "loss": 0.1335, "step": 2137 }, { "epoch": 1.69, "learning_rate": 8.440291769908168e-06, "loss": 0.1092, "step": 2138 }, { "epoch": 1.69, "learning_rate": 8.431862171131524e-06, "loss": 0.1187, "step": 2139 }, { "epoch": 1.69, "learning_rate": 8.423433714581599e-06, "loss": 0.1267, "step": 2140 }, { "epoch": 1.69, "learning_rate": 8.415006406397658e-06, "loss": 0.1413, "step": 2141 }, { "epoch": 1.69, "learning_rate": 8.406580252718125e-06, "loss": 0.1526, "step": 2142 }, { "epoch": 1.69, "learning_rate": 8.398155259680583e-06, "loss": 0.1395, "step": 2143 }, { "epoch": 1.69, "learning_rate": 8.38973143342177e-06, "loss": 0.143, "step": 2144 }, { "epoch": 1.69, "learning_rate": 8.381308780077575e-06, "loss": 0.1387, "step": 2145 }, { "epoch": 1.7, "learning_rate": 8.372887305783038e-06, "loss": 0.1474, "step": 2146 }, { "epoch": 1.7, "learning_rate": 8.364467016672328e-06, "loss": 0.1475, "step": 2147 }, { "epoch": 1.7, "learning_rate": 8.356047918878762e-06, "loss": 0.1287, "step": 2148 }, { "epoch": 1.7, "learning_rate": 8.347630018534783e-06, "loss": 0.1412, "step": 2149 }, { "epoch": 1.7, "learning_rate": 8.339213321771964e-06, "loss": 0.1369, "step": 2150 }, { "epoch": 1.7, "learning_rate": 8.330797834720993e-06, "loss": 0.1598, "step": 2151 }, { "epoch": 1.7, "learning_rate": 8.322383563511698e-06, "loss": 0.1615, "step": 2152 }, { "epoch": 1.7, "learning_rate": 8.313970514273003e-06, "loss": 0.1336, "step": 2153 }, { "epoch": 1.7, "learning_rate": 8.305558693132943e-06, "loss": 0.141, "step": 2154 }, { "epoch": 1.7, "learning_rate": 8.297148106218664e-06, "loss": 0.135, "step": 2155 }, { "epoch": 1.7, "learning_rate": 8.288738759656415e-06, "loss": 0.1288, "step": 2156 }, { "epoch": 1.7, "learning_rate": 8.280330659571532e-06, "loss": 0.1281, "step": 2157 }, { "epoch": 1.71, "learning_rate": 8.271923812088451e-06, "loss": 0.1315, "step": 2158 }, { "epoch": 1.71, "learning_rate": 8.263518223330698e-06, "loss": 0.1399, "step": 2159 }, { "epoch": 1.71, "learning_rate": 8.255113899420873e-06, "loss": 0.1198, "step": 2160 }, { "epoch": 1.71, "learning_rate": 8.246710846480664e-06, "loss": 0.1337, "step": 2161 }, { "epoch": 1.71, "learning_rate": 8.238309070630821e-06, "loss": 0.1514, "step": 2162 }, { "epoch": 1.71, "learning_rate": 8.229908577991177e-06, "loss": 0.1255, "step": 2163 }, { "epoch": 1.71, "learning_rate": 8.221509374680623e-06, "loss": 0.1231, "step": 2164 }, { "epoch": 1.71, "learning_rate": 8.213111466817116e-06, "loss": 0.1515, "step": 2165 }, { "epoch": 1.71, "learning_rate": 8.204714860517662e-06, "loss": 0.1245, "step": 2166 }, { "epoch": 1.71, "learning_rate": 8.196319561898325e-06, "loss": 0.1407, "step": 2167 }, { "epoch": 1.71, "learning_rate": 8.187925577074213e-06, "loss": 0.1443, "step": 2168 }, { "epoch": 1.71, "learning_rate": 8.179532912159477e-06, "loss": 0.1482, "step": 2169 }, { "epoch": 1.71, "learning_rate": 8.171141573267312e-06, "loss": 0.148, "step": 2170 }, { "epoch": 1.72, "learning_rate": 8.16275156650994e-06, "loss": 0.1407, "step": 2171 }, { "epoch": 1.72, "learning_rate": 8.154362897998619e-06, "loss": 0.1438, "step": 2172 }, { "epoch": 1.72, "learning_rate": 8.145975573843627e-06, "loss": 0.1283, "step": 2173 }, { "epoch": 1.72, "learning_rate": 8.137589600154264e-06, "loss": 0.1461, "step": 2174 }, { "epoch": 1.72, "learning_rate": 8.129204983038847e-06, "loss": 0.1295, "step": 2175 }, { "epoch": 1.72, "learning_rate": 8.120821728604708e-06, "loss": 0.1601, "step": 2176 }, { "epoch": 1.72, "learning_rate": 8.112439842958179e-06, "loss": 0.1632, "step": 2177 }, { "epoch": 1.72, "learning_rate": 8.104059332204606e-06, "loss": 0.1501, "step": 2178 }, { "epoch": 1.72, "learning_rate": 8.095680202448321e-06, "loss": 0.1503, "step": 2179 }, { "epoch": 1.72, "learning_rate": 8.08730245979266e-06, "loss": 0.1229, "step": 2180 }, { "epoch": 1.72, "learning_rate": 8.07892611033994e-06, "loss": 0.1412, "step": 2181 }, { "epoch": 1.72, "learning_rate": 8.070551160191474e-06, "loss": 0.1243, "step": 2182 }, { "epoch": 1.73, "learning_rate": 8.062177615447544e-06, "loss": 0.1475, "step": 2183 }, { "epoch": 1.73, "learning_rate": 8.053805482207418e-06, "loss": 0.1288, "step": 2184 }, { "epoch": 1.73, "learning_rate": 8.045434766569331e-06, "loss": 0.1219, "step": 2185 }, { "epoch": 1.73, "learning_rate": 8.037065474630486e-06, "loss": 0.1204, "step": 2186 }, { "epoch": 1.73, "learning_rate": 8.028697612487046e-06, "loss": 0.1399, "step": 2187 }, { "epoch": 1.73, "learning_rate": 8.020331186234139e-06, "loss": 0.1118, "step": 2188 }, { "epoch": 1.73, "learning_rate": 8.011966201965846e-06, "loss": 0.1506, "step": 2189 }, { "epoch": 1.73, "learning_rate": 8.003602665775189e-06, "loss": 0.1142, "step": 2190 }, { "epoch": 1.73, "learning_rate": 7.995240583754149e-06, "loss": 0.1516, "step": 2191 }, { "epoch": 1.73, "learning_rate": 7.986879961993635e-06, "loss": 0.1327, "step": 2192 }, { "epoch": 1.73, "learning_rate": 7.978520806583503e-06, "loss": 0.1358, "step": 2193 }, { "epoch": 1.73, "learning_rate": 7.970163123612532e-06, "loss": 0.1548, "step": 2194 }, { "epoch": 1.73, "learning_rate": 7.961806919168435e-06, "loss": 0.1475, "step": 2195 }, { "epoch": 1.74, "learning_rate": 7.95345219933784e-06, "loss": 0.1256, "step": 2196 }, { "epoch": 1.74, "learning_rate": 7.945098970206304e-06, "loss": 0.1326, "step": 2197 }, { "epoch": 1.74, "learning_rate": 7.936747237858292e-06, "loss": 0.1271, "step": 2198 }, { "epoch": 1.74, "learning_rate": 7.92839700837718e-06, "loss": 0.1396, "step": 2199 }, { "epoch": 1.74, "learning_rate": 7.92004828784525e-06, "loss": 0.151, "step": 2200 }, { "epoch": 1.74, "learning_rate": 7.911701082343683e-06, "loss": 0.1636, "step": 2201 }, { "epoch": 1.74, "learning_rate": 7.903355397952557e-06, "loss": 0.1393, "step": 2202 }, { "epoch": 1.74, "learning_rate": 7.895011240750841e-06, "loss": 0.1417, "step": 2203 }, { "epoch": 1.74, "learning_rate": 7.886668616816395e-06, "loss": 0.1227, "step": 2204 }, { "epoch": 1.74, "learning_rate": 7.878327532225964e-06, "loss": 0.1451, "step": 2205 }, { "epoch": 1.74, "learning_rate": 7.869987993055163e-06, "loss": 0.1468, "step": 2206 }, { "epoch": 1.74, "learning_rate": 7.86165000537849e-06, "loss": 0.1485, "step": 2207 }, { "epoch": 1.74, "learning_rate": 7.853313575269306e-06, "loss": 0.1415, "step": 2208 }, { "epoch": 1.75, "learning_rate": 7.84497870879984e-06, "loss": 0.1241, "step": 2209 }, { "epoch": 1.75, "learning_rate": 7.836645412041187e-06, "loss": 0.1226, "step": 2210 }, { "epoch": 1.75, "learning_rate": 7.828313691063294e-06, "loss": 0.1454, "step": 2211 }, { "epoch": 1.75, "learning_rate": 7.819983551934957e-06, "loss": 0.159, "step": 2212 }, { "epoch": 1.75, "learning_rate": 7.811655000723827e-06, "loss": 0.1384, "step": 2213 }, { "epoch": 1.75, "learning_rate": 7.803328043496394e-06, "loss": 0.1312, "step": 2214 }, { "epoch": 1.75, "learning_rate": 7.795002686317988e-06, "loss": 0.1505, "step": 2215 }, { "epoch": 1.75, "learning_rate": 7.786678935252766e-06, "loss": 0.136, "step": 2216 }, { "epoch": 1.75, "learning_rate": 7.778356796363734e-06, "loss": 0.1415, "step": 2217 }, { "epoch": 1.75, "learning_rate": 7.770036275712708e-06, "loss": 0.1447, "step": 2218 }, { "epoch": 1.75, "learning_rate": 7.761717379360328e-06, "loss": 0.1498, "step": 2219 }, { "epoch": 1.75, "learning_rate": 7.753400113366052e-06, "loss": 0.1303, "step": 2220 }, { "epoch": 1.76, "learning_rate": 7.745084483788153e-06, "loss": 0.1355, "step": 2221 }, { "epoch": 1.76, "learning_rate": 7.736770496683701e-06, "loss": 0.1419, "step": 2222 }, { "epoch": 1.76, "learning_rate": 7.728458158108592e-06, "loss": 0.1352, "step": 2223 }, { "epoch": 1.76, "learning_rate": 7.7201474741175e-06, "loss": 0.1532, "step": 2224 }, { "epoch": 1.76, "learning_rate": 7.711838450763903e-06, "loss": 0.1413, "step": 2225 }, { "epoch": 1.76, "learning_rate": 7.703531094100068e-06, "loss": 0.158, "step": 2226 }, { "epoch": 1.76, "learning_rate": 7.695225410177045e-06, "loss": 0.1343, "step": 2227 }, { "epoch": 1.76, "learning_rate": 7.686921405044672e-06, "loss": 0.1481, "step": 2228 }, { "epoch": 1.76, "learning_rate": 7.678619084751554e-06, "loss": 0.1374, "step": 2229 }, { "epoch": 1.76, "learning_rate": 7.670318455345084e-06, "loss": 0.1283, "step": 2230 }, { "epoch": 1.76, "learning_rate": 7.662019522871413e-06, "loss": 0.1458, "step": 2231 }, { "epoch": 1.76, "learning_rate": 7.653722293375453e-06, "loss": 0.1341, "step": 2232 }, { "epoch": 1.76, "learning_rate": 7.645426772900884e-06, "loss": 0.1297, "step": 2233 }, { "epoch": 1.77, "learning_rate": 7.637132967490136e-06, "loss": 0.1264, "step": 2234 }, { "epoch": 1.77, "learning_rate": 7.628840883184385e-06, "loss": 0.1519, "step": 2235 }, { "epoch": 1.77, "learning_rate": 7.620550526023571e-06, "loss": 0.1493, "step": 2236 }, { "epoch": 1.77, "learning_rate": 7.612261902046361e-06, "loss": 0.1711, "step": 2237 }, { "epoch": 1.77, "learning_rate": 7.603975017290159e-06, "loss": 0.1266, "step": 2238 }, { "epoch": 1.77, "learning_rate": 7.595689877791107e-06, "loss": 0.1319, "step": 2239 }, { "epoch": 1.77, "learning_rate": 7.587406489584077e-06, "loss": 0.1487, "step": 2240 }, { "epoch": 1.77, "learning_rate": 7.579124858702658e-06, "loss": 0.131, "step": 2241 }, { "epoch": 1.77, "learning_rate": 7.570844991179172e-06, "loss": 0.1367, "step": 2242 }, { "epoch": 1.77, "learning_rate": 7.562566893044645e-06, "loss": 0.1459, "step": 2243 }, { "epoch": 1.77, "learning_rate": 7.5542905703288175e-06, "loss": 0.148, "step": 2244 }, { "epoch": 1.77, "learning_rate": 7.546016029060138e-06, "loss": 0.1244, "step": 2245 }, { "epoch": 1.77, "learning_rate": 7.537743275265757e-06, "loss": 0.162, "step": 2246 }, { "epoch": 1.78, "learning_rate": 7.529472314971522e-06, "loss": 0.1472, "step": 2247 }, { "epoch": 1.78, "learning_rate": 7.52120315420197e-06, "loss": 0.1445, "step": 2248 }, { "epoch": 1.78, "learning_rate": 7.512935798980341e-06, "loss": 0.1361, "step": 2249 }, { "epoch": 1.78, "learning_rate": 7.504670255328548e-06, "loss": 0.1409, "step": 2250 }, { "epoch": 1.78, "learning_rate": 7.496406529267184e-06, "loss": 0.1263, "step": 2251 }, { "epoch": 1.78, "learning_rate": 7.4881446268155236e-06, "loss": 0.1346, "step": 2252 }, { "epoch": 1.78, "learning_rate": 7.4798845539915126e-06, "loss": 0.1381, "step": 2253 }, { "epoch": 1.78, "learning_rate": 7.471626316811753e-06, "loss": 0.142, "step": 2254 }, { "epoch": 1.78, "learning_rate": 7.4633699212915335e-06, "loss": 0.1256, "step": 2255 }, { "epoch": 1.78, "learning_rate": 7.455115373444779e-06, "loss": 0.1243, "step": 2256 }, { "epoch": 1.78, "learning_rate": 7.446862679284077e-06, "loss": 0.1574, "step": 2257 }, { "epoch": 1.78, "learning_rate": 7.438611844820664e-06, "loss": 0.1339, "step": 2258 }, { "epoch": 1.79, "learning_rate": 7.430362876064424e-06, "loss": 0.1396, "step": 2259 }, { "epoch": 1.79, "learning_rate": 7.42211577902388e-06, "loss": 0.1078, "step": 2260 }, { "epoch": 1.79, "learning_rate": 7.413870559706185e-06, "loss": 0.1237, "step": 2261 }, { "epoch": 1.79, "learning_rate": 7.4056272241171425e-06, "loss": 0.1513, "step": 2262 }, { "epoch": 1.79, "learning_rate": 7.397385778261166e-06, "loss": 0.1291, "step": 2263 }, { "epoch": 1.79, "learning_rate": 7.389146228141299e-06, "loss": 0.1127, "step": 2264 }, { "epoch": 1.79, "learning_rate": 7.380908579759207e-06, "loss": 0.149, "step": 2265 }, { "epoch": 1.79, "learning_rate": 7.372672839115167e-06, "loss": 0.12, "step": 2266 }, { "epoch": 1.79, "learning_rate": 7.36443901220806e-06, "loss": 0.1396, "step": 2267 }, { "epoch": 1.79, "learning_rate": 7.356207105035389e-06, "loss": 0.1297, "step": 2268 }, { "epoch": 1.79, "learning_rate": 7.347977123593248e-06, "loss": 0.1383, "step": 2269 }, { "epoch": 1.79, "learning_rate": 7.339749073876327e-06, "loss": 0.1394, "step": 2270 }, { "epoch": 1.79, "learning_rate": 7.331522961877914e-06, "loss": 0.1257, "step": 2271 }, { "epoch": 1.8, "learning_rate": 7.323298793589881e-06, "loss": 0.1425, "step": 2272 }, { "epoch": 1.8, "learning_rate": 7.315076575002688e-06, "loss": 0.1587, "step": 2273 }, { "epoch": 1.8, "learning_rate": 7.30685631210537e-06, "loss": 0.1423, "step": 2274 }, { "epoch": 1.8, "learning_rate": 7.29863801088555e-06, "loss": 0.1579, "step": 2275 }, { "epoch": 1.8, "learning_rate": 7.2904216773294055e-06, "loss": 0.121, "step": 2276 }, { "epoch": 1.8, "learning_rate": 7.282207317421691e-06, "loss": 0.1316, "step": 2277 }, { "epoch": 1.8, "learning_rate": 7.273994937145721e-06, "loss": 0.1214, "step": 2278 }, { "epoch": 1.8, "learning_rate": 7.265784542483365e-06, "loss": 0.1182, "step": 2279 }, { "epoch": 1.8, "learning_rate": 7.2575761394150476e-06, "loss": 0.1693, "step": 2280 }, { "epoch": 1.8, "learning_rate": 7.249369733919751e-06, "loss": 0.1409, "step": 2281 }, { "epoch": 1.8, "learning_rate": 7.241165331974991e-06, "loss": 0.1354, "step": 2282 }, { "epoch": 1.8, "learning_rate": 7.232962939556831e-06, "loss": 0.1391, "step": 2283 }, { "epoch": 1.8, "learning_rate": 7.224762562639866e-06, "loss": 0.1408, "step": 2284 }, { "epoch": 1.81, "learning_rate": 7.216564207197224e-06, "loss": 0.1484, "step": 2285 }, { "epoch": 1.81, "learning_rate": 7.20836787920056e-06, "loss": 0.1262, "step": 2286 }, { "epoch": 1.81, "learning_rate": 7.20017358462006e-06, "loss": 0.137, "step": 2287 }, { "epoch": 1.81, "learning_rate": 7.19198132942442e-06, "loss": 0.1441, "step": 2288 }, { "epoch": 1.81, "learning_rate": 7.183791119580854e-06, "loss": 0.1296, "step": 2289 }, { "epoch": 1.81, "learning_rate": 7.175602961055084e-06, "loss": 0.1548, "step": 2290 }, { "epoch": 1.81, "learning_rate": 7.167416859811338e-06, "loss": 0.1206, "step": 2291 }, { "epoch": 1.81, "learning_rate": 7.159232821812348e-06, "loss": 0.1273, "step": 2292 }, { "epoch": 1.81, "learning_rate": 7.151050853019339e-06, "loss": 0.1142, "step": 2293 }, { "epoch": 1.81, "learning_rate": 7.142870959392039e-06, "loss": 0.1491, "step": 2294 }, { "epoch": 1.81, "learning_rate": 7.134693146888652e-06, "loss": 0.1247, "step": 2295 }, { "epoch": 1.81, "learning_rate": 7.126517421465871e-06, "loss": 0.1295, "step": 2296 }, { "epoch": 1.82, "learning_rate": 7.118343789078872e-06, "loss": 0.1217, "step": 2297 }, { "epoch": 1.82, "learning_rate": 7.1101722556813e-06, "loss": 0.1298, "step": 2298 }, { "epoch": 1.82, "learning_rate": 7.102002827225272e-06, "loss": 0.1408, "step": 2299 }, { "epoch": 1.82, "learning_rate": 7.093835509661382e-06, "loss": 0.1257, "step": 2300 }, { "epoch": 1.82, "learning_rate": 7.085670308938674e-06, "loss": 0.1234, "step": 2301 }, { "epoch": 1.82, "learning_rate": 7.077507231004658e-06, "loss": 0.1326, "step": 2302 }, { "epoch": 1.82, "learning_rate": 7.069346281805292e-06, "loss": 0.1465, "step": 2303 }, { "epoch": 1.82, "learning_rate": 7.061187467284985e-06, "loss": 0.1517, "step": 2304 }, { "epoch": 1.82, "learning_rate": 7.053030793386593e-06, "loss": 0.1205, "step": 2305 }, { "epoch": 1.82, "learning_rate": 7.04487626605141e-06, "loss": 0.1386, "step": 2306 }, { "epoch": 1.82, "learning_rate": 7.0367238912191734e-06, "loss": 0.1454, "step": 2307 }, { "epoch": 1.82, "learning_rate": 7.028573674828047e-06, "loss": 0.1262, "step": 2308 }, { "epoch": 1.82, "learning_rate": 7.020425622814619e-06, "loss": 0.128, "step": 2309 }, { "epoch": 1.83, "learning_rate": 7.012279741113909e-06, "loss": 0.1213, "step": 2310 }, { "epoch": 1.83, "learning_rate": 7.004136035659349e-06, "loss": 0.1441, "step": 2311 }, { "epoch": 1.83, "learning_rate": 6.995994512382787e-06, "loss": 0.1233, "step": 2312 }, { "epoch": 1.83, "learning_rate": 6.987855177214489e-06, "loss": 0.1127, "step": 2313 }, { "epoch": 1.83, "learning_rate": 6.979718036083119e-06, "loss": 0.1185, "step": 2314 }, { "epoch": 1.83, "learning_rate": 6.971583094915746e-06, "loss": 0.1306, "step": 2315 }, { "epoch": 1.83, "learning_rate": 6.963450359637835e-06, "loss": 0.1244, "step": 2316 }, { "epoch": 1.83, "learning_rate": 6.9553198361732425e-06, "loss": 0.1518, "step": 2317 }, { "epoch": 1.83, "learning_rate": 6.947191530444219e-06, "loss": 0.1294, "step": 2318 }, { "epoch": 1.83, "learning_rate": 6.939065448371398e-06, "loss": 0.117, "step": 2319 }, { "epoch": 1.83, "learning_rate": 6.930941595873793e-06, "loss": 0.1399, "step": 2320 }, { "epoch": 1.83, "learning_rate": 6.92281997886879e-06, "loss": 0.1236, "step": 2321 }, { "epoch": 1.83, "learning_rate": 6.914700603272151e-06, "loss": 0.1293, "step": 2322 }, { "epoch": 1.84, "learning_rate": 6.906583474998003e-06, "loss": 0.1485, "step": 2323 }, { "epoch": 1.84, "learning_rate": 6.898468599958841e-06, "loss": 0.1225, "step": 2324 }, { "epoch": 1.84, "learning_rate": 6.8903559840655075e-06, "loss": 0.1209, "step": 2325 }, { "epoch": 1.84, "learning_rate": 6.882245633227214e-06, "loss": 0.1258, "step": 2326 }, { "epoch": 1.84, "learning_rate": 6.874137553351512e-06, "loss": 0.1438, "step": 2327 }, { "epoch": 1.84, "learning_rate": 6.866031750344302e-06, "loss": 0.1386, "step": 2328 }, { "epoch": 1.84, "learning_rate": 6.857928230109821e-06, "loss": 0.1421, "step": 2329 }, { "epoch": 1.84, "learning_rate": 6.849826998550656e-06, "loss": 0.1243, "step": 2330 }, { "epoch": 1.84, "learning_rate": 6.841728061567713e-06, "loss": 0.1278, "step": 2331 }, { "epoch": 1.84, "learning_rate": 6.833631425060237e-06, "loss": 0.1342, "step": 2332 }, { "epoch": 1.84, "learning_rate": 6.8255370949257894e-06, "loss": 0.1275, "step": 2333 }, { "epoch": 1.84, "learning_rate": 6.817445077060256e-06, "loss": 0.1166, "step": 2334 }, { "epoch": 1.85, "learning_rate": 6.809355377357836e-06, "loss": 0.127, "step": 2335 }, { "epoch": 1.85, "learning_rate": 6.801268001711046e-06, "loss": 0.1522, "step": 2336 }, { "epoch": 1.85, "learning_rate": 6.7931829560107e-06, "loss": 0.1331, "step": 2337 }, { "epoch": 1.85, "learning_rate": 6.7851002461459225e-06, "loss": 0.1151, "step": 2338 }, { "epoch": 1.85, "learning_rate": 6.777019878004136e-06, "loss": 0.1478, "step": 2339 }, { "epoch": 1.85, "learning_rate": 6.768941857471054e-06, "loss": 0.1261, "step": 2340 }, { "epoch": 1.85, "learning_rate": 6.760866190430679e-06, "loss": 0.1382, "step": 2341 }, { "epoch": 1.85, "learning_rate": 6.752792882765306e-06, "loss": 0.1259, "step": 2342 }, { "epoch": 1.85, "learning_rate": 6.744721940355508e-06, "loss": 0.1644, "step": 2343 }, { "epoch": 1.85, "learning_rate": 6.736653369080126e-06, "loss": 0.1193, "step": 2344 }, { "epoch": 1.85, "learning_rate": 6.728587174816292e-06, "loss": 0.1109, "step": 2345 }, { "epoch": 1.85, "learning_rate": 6.720523363439393e-06, "loss": 0.1551, "step": 2346 }, { "epoch": 1.85, "learning_rate": 6.7124619408230815e-06, "loss": 0.1418, "step": 2347 }, { "epoch": 1.86, "learning_rate": 6.7044029128392785e-06, "loss": 0.13, "step": 2348 }, { "epoch": 1.86, "learning_rate": 6.69634628535815e-06, "loss": 0.145, "step": 2349 }, { "epoch": 1.86, "learning_rate": 6.6882920642481205e-06, "loss": 0.1266, "step": 2350 }, { "epoch": 1.86, "learning_rate": 6.6802402553758565e-06, "loss": 0.1241, "step": 2351 }, { "epoch": 1.86, "learning_rate": 6.672190864606276e-06, "loss": 0.1331, "step": 2352 }, { "epoch": 1.86, "learning_rate": 6.664143897802526e-06, "loss": 0.1315, "step": 2353 }, { "epoch": 1.86, "learning_rate": 6.656099360825994e-06, "loss": 0.1461, "step": 2354 }, { "epoch": 1.86, "learning_rate": 6.648057259536297e-06, "loss": 0.1439, "step": 2355 }, { "epoch": 1.86, "learning_rate": 6.640017599791273e-06, "loss": 0.1108, "step": 2356 }, { "epoch": 1.86, "learning_rate": 6.631980387446987e-06, "loss": 0.1627, "step": 2357 }, { "epoch": 1.86, "learning_rate": 6.623945628357718e-06, "loss": 0.1441, "step": 2358 }, { "epoch": 1.86, "learning_rate": 6.615913328375965e-06, "loss": 0.122, "step": 2359 }, { "epoch": 1.86, "learning_rate": 6.607883493352425e-06, "loss": 0.1346, "step": 2360 }, { "epoch": 1.87, "learning_rate": 6.599856129136006e-06, "loss": 0.1405, "step": 2361 }, { "epoch": 1.87, "learning_rate": 6.591831241573816e-06, "loss": 0.1296, "step": 2362 }, { "epoch": 1.87, "learning_rate": 6.583808836511157e-06, "loss": 0.1686, "step": 2363 }, { "epoch": 1.87, "learning_rate": 6.575788919791522e-06, "loss": 0.1346, "step": 2364 }, { "epoch": 1.87, "learning_rate": 6.567771497256599e-06, "loss": 0.1298, "step": 2365 }, { "epoch": 1.87, "learning_rate": 6.55975657474625e-06, "loss": 0.1422, "step": 2366 }, { "epoch": 1.87, "learning_rate": 6.551744158098521e-06, "loss": 0.1598, "step": 2367 }, { "epoch": 1.87, "learning_rate": 6.54373425314963e-06, "loss": 0.1535, "step": 2368 }, { "epoch": 1.87, "learning_rate": 6.535726865733969e-06, "loss": 0.1315, "step": 2369 }, { "epoch": 1.87, "learning_rate": 6.527722001684087e-06, "loss": 0.1109, "step": 2370 }, { "epoch": 1.87, "learning_rate": 6.519719666830713e-06, "loss": 0.1352, "step": 2371 }, { "epoch": 1.87, "learning_rate": 6.511719867002717e-06, "loss": 0.1205, "step": 2372 }, { "epoch": 1.88, "learning_rate": 6.503722608027129e-06, "loss": 0.1456, "step": 2373 }, { "epoch": 1.88, "learning_rate": 6.495727895729127e-06, "loss": 0.141, "step": 2374 }, { "epoch": 1.88, "learning_rate": 6.487735735932035e-06, "loss": 0.1382, "step": 2375 }, { "epoch": 1.88, "learning_rate": 6.479746134457319e-06, "loss": 0.1129, "step": 2376 }, { "epoch": 1.88, "learning_rate": 6.471759097124573e-06, "loss": 0.1441, "step": 2377 }, { "epoch": 1.88, "learning_rate": 6.463774629751538e-06, "loss": 0.1329, "step": 2378 }, { "epoch": 1.88, "learning_rate": 6.455792738154074e-06, "loss": 0.1258, "step": 2379 }, { "epoch": 1.88, "learning_rate": 6.447813428146166e-06, "loss": 0.1352, "step": 2380 }, { "epoch": 1.88, "learning_rate": 6.439836705539916e-06, "loss": 0.1279, "step": 2381 }, { "epoch": 1.88, "learning_rate": 6.431862576145546e-06, "loss": 0.1404, "step": 2382 }, { "epoch": 1.88, "learning_rate": 6.423891045771382e-06, "loss": 0.1174, "step": 2383 }, { "epoch": 1.88, "learning_rate": 6.41592212022387e-06, "loss": 0.1394, "step": 2384 }, { "epoch": 1.88, "learning_rate": 6.407955805307547e-06, "loss": 0.1369, "step": 2385 }, { "epoch": 1.89, "learning_rate": 6.399992106825053e-06, "loss": 0.1267, "step": 2386 }, { "epoch": 1.89, "learning_rate": 6.392031030577118e-06, "loss": 0.1374, "step": 2387 }, { "epoch": 1.89, "learning_rate": 6.3840725823625685e-06, "loss": 0.1374, "step": 2388 }, { "epoch": 1.89, "learning_rate": 6.376116767978309e-06, "loss": 0.135, "step": 2389 }, { "epoch": 1.89, "learning_rate": 6.3681635932193365e-06, "loss": 0.1194, "step": 2390 }, { "epoch": 1.89, "learning_rate": 6.3602130638787155e-06, "loss": 0.1212, "step": 2391 }, { "epoch": 1.89, "learning_rate": 6.352265185747588e-06, "loss": 0.1316, "step": 2392 }, { "epoch": 1.89, "learning_rate": 6.344319964615162e-06, "loss": 0.1235, "step": 2393 }, { "epoch": 1.89, "learning_rate": 6.336377406268712e-06, "loss": 0.1221, "step": 2394 }, { "epoch": 1.89, "learning_rate": 6.328437516493573e-06, "loss": 0.1119, "step": 2395 }, { "epoch": 1.89, "learning_rate": 6.320500301073134e-06, "loss": 0.1466, "step": 2396 }, { "epoch": 1.89, "learning_rate": 6.312565765788843e-06, "loss": 0.1344, "step": 2397 }, { "epoch": 1.89, "learning_rate": 6.304633916420189e-06, "loss": 0.1411, "step": 2398 }, { "epoch": 1.9, "learning_rate": 6.296704758744705e-06, "loss": 0.1311, "step": 2399 }, { "epoch": 1.9, "learning_rate": 6.288778298537966e-06, "loss": 0.1309, "step": 2400 }, { "epoch": 1.9, "learning_rate": 6.2808545415735815e-06, "loss": 0.1483, "step": 2401 }, { "epoch": 1.9, "learning_rate": 6.272933493623184e-06, "loss": 0.137, "step": 2402 }, { "epoch": 1.9, "learning_rate": 6.2650151604564534e-06, "loss": 0.1231, "step": 2403 }, { "epoch": 1.9, "learning_rate": 6.25709954784107e-06, "loss": 0.1412, "step": 2404 }, { "epoch": 1.9, "learning_rate": 6.249186661542746e-06, "loss": 0.1502, "step": 2405 }, { "epoch": 1.9, "learning_rate": 6.241276507325198e-06, "loss": 0.1404, "step": 2406 }, { "epoch": 1.9, "learning_rate": 6.233369090950162e-06, "loss": 0.1437, "step": 2407 }, { "epoch": 1.9, "learning_rate": 6.225464418177374e-06, "loss": 0.1201, "step": 2408 }, { "epoch": 1.9, "learning_rate": 6.217562494764569e-06, "loss": 0.1553, "step": 2409 }, { "epoch": 1.9, "learning_rate": 6.209663326467492e-06, "loss": 0.1027, "step": 2410 }, { "epoch": 1.91, "learning_rate": 6.201766919039869e-06, "loss": 0.1307, "step": 2411 }, { "epoch": 1.91, "learning_rate": 6.193873278233417e-06, "loss": 0.1172, "step": 2412 }, { "epoch": 1.91, "learning_rate": 6.1859824097978396e-06, "loss": 0.1358, "step": 2413 }, { "epoch": 1.91, "learning_rate": 6.178094319480823e-06, "loss": 0.1095, "step": 2414 }, { "epoch": 1.91, "learning_rate": 6.170209013028021e-06, "loss": 0.122, "step": 2415 }, { "epoch": 1.91, "learning_rate": 6.162326496183077e-06, "loss": 0.134, "step": 2416 }, { "epoch": 1.91, "learning_rate": 6.1544467746875844e-06, "loss": 0.1268, "step": 2417 }, { "epoch": 1.91, "learning_rate": 6.146569854281111e-06, "loss": 0.1264, "step": 2418 }, { "epoch": 1.91, "learning_rate": 6.138695740701179e-06, "loss": 0.1134, "step": 2419 }, { "epoch": 1.91, "learning_rate": 6.13082443968327e-06, "loss": 0.1275, "step": 2420 }, { "epoch": 1.91, "learning_rate": 6.1229559569608144e-06, "loss": 0.1231, "step": 2421 }, { "epoch": 1.91, "learning_rate": 6.115090298265187e-06, "loss": 0.131, "step": 2422 }, { "epoch": 1.91, "learning_rate": 6.107227469325716e-06, "loss": 0.1329, "step": 2423 }, { "epoch": 1.92, "learning_rate": 6.099367475869658e-06, "loss": 0.1129, "step": 2424 }, { "epoch": 1.92, "learning_rate": 6.0915103236222095e-06, "loss": 0.1497, "step": 2425 }, { "epoch": 1.92, "learning_rate": 6.083656018306493e-06, "loss": 0.1153, "step": 2426 }, { "epoch": 1.92, "learning_rate": 6.075804565643562e-06, "loss": 0.1349, "step": 2427 }, { "epoch": 1.92, "learning_rate": 6.067955971352385e-06, "loss": 0.139, "step": 2428 }, { "epoch": 1.92, "learning_rate": 6.060110241149864e-06, "loss": 0.1208, "step": 2429 }, { "epoch": 1.92, "learning_rate": 6.052267380750796e-06, "loss": 0.1308, "step": 2430 }, { "epoch": 1.92, "learning_rate": 6.044427395867901e-06, "loss": 0.1173, "step": 2431 }, { "epoch": 1.92, "learning_rate": 6.036590292211797e-06, "loss": 0.1269, "step": 2432 }, { "epoch": 1.92, "learning_rate": 6.028756075491007e-06, "loss": 0.1457, "step": 2433 }, { "epoch": 1.92, "learning_rate": 6.020924751411945e-06, "loss": 0.1148, "step": 2434 }, { "epoch": 1.92, "learning_rate": 6.013096325678932e-06, "loss": 0.1261, "step": 2435 }, { "epoch": 1.92, "learning_rate": 6.005270803994165e-06, "loss": 0.1591, "step": 2436 }, { "epoch": 1.93, "learning_rate": 5.9974481920577285e-06, "loss": 0.1201, "step": 2437 }, { "epoch": 1.93, "learning_rate": 5.989628495567591e-06, "loss": 0.1189, "step": 2438 }, { "epoch": 1.93, "learning_rate": 5.981811720219593e-06, "loss": 0.1152, "step": 2439 }, { "epoch": 1.93, "learning_rate": 5.973997871707451e-06, "loss": 0.1336, "step": 2440 }, { "epoch": 1.93, "learning_rate": 5.966186955722742e-06, "loss": 0.1299, "step": 2441 }, { "epoch": 1.93, "learning_rate": 5.958378977954922e-06, "loss": 0.1129, "step": 2442 }, { "epoch": 1.93, "learning_rate": 5.950573944091295e-06, "loss": 0.148, "step": 2443 }, { "epoch": 1.93, "learning_rate": 5.942771859817021e-06, "loss": 0.1212, "step": 2444 }, { "epoch": 1.93, "learning_rate": 5.934972730815115e-06, "loss": 0.1444, "step": 2445 }, { "epoch": 1.93, "learning_rate": 5.927176562766439e-06, "loss": 0.1243, "step": 2446 }, { "epoch": 1.93, "learning_rate": 5.919383361349693e-06, "loss": 0.1168, "step": 2447 }, { "epoch": 1.93, "learning_rate": 5.9115931322414285e-06, "loss": 0.1198, "step": 2448 }, { "epoch": 1.94, "learning_rate": 5.90380588111602e-06, "loss": 0.1303, "step": 2449 }, { "epoch": 1.94, "learning_rate": 5.896021613645679e-06, "loss": 0.1407, "step": 2450 }, { "epoch": 1.94, "learning_rate": 5.888240335500439e-06, "loss": 0.119, "step": 2451 }, { "epoch": 1.94, "learning_rate": 5.88046205234816e-06, "loss": 0.1251, "step": 2452 }, { "epoch": 1.94, "learning_rate": 5.8726867698545185e-06, "loss": 0.1297, "step": 2453 }, { "epoch": 1.94, "learning_rate": 5.8649144936830045e-06, "loss": 0.1311, "step": 2454 }, { "epoch": 1.94, "learning_rate": 5.857145229494925e-06, "loss": 0.1405, "step": 2455 }, { "epoch": 1.94, "learning_rate": 5.8493789829493855e-06, "loss": 0.1143, "step": 2456 }, { "epoch": 1.94, "learning_rate": 5.841615759703296e-06, "loss": 0.1462, "step": 2457 }, { "epoch": 1.94, "learning_rate": 5.8338555654113605e-06, "loss": 0.1254, "step": 2458 }, { "epoch": 1.94, "learning_rate": 5.826098405726086e-06, "loss": 0.1248, "step": 2459 }, { "epoch": 1.94, "learning_rate": 5.818344286297756e-06, "loss": 0.1194, "step": 2460 }, { "epoch": 1.94, "learning_rate": 5.810593212774457e-06, "loss": 0.1292, "step": 2461 }, { "epoch": 1.95, "learning_rate": 5.802845190802042e-06, "loss": 0.1187, "step": 2462 }, { "epoch": 1.95, "learning_rate": 5.795100226024145e-06, "loss": 0.1627, "step": 2463 }, { "epoch": 1.95, "learning_rate": 5.787358324082175e-06, "loss": 0.1069, "step": 2464 }, { "epoch": 1.95, "learning_rate": 5.779619490615312e-06, "loss": 0.1182, "step": 2465 }, { "epoch": 1.95, "learning_rate": 5.771883731260492e-06, "loss": 0.1308, "step": 2466 }, { "epoch": 1.95, "learning_rate": 5.764151051652421e-06, "loss": 0.1344, "step": 2467 }, { "epoch": 1.95, "learning_rate": 5.756421457423562e-06, "loss": 0.1199, "step": 2468 }, { "epoch": 1.95, "learning_rate": 5.748694954204126e-06, "loss": 0.129, "step": 2469 }, { "epoch": 1.95, "learning_rate": 5.740971547622074e-06, "loss": 0.1522, "step": 2470 }, { "epoch": 1.95, "learning_rate": 5.733251243303108e-06, "loss": 0.1312, "step": 2471 }, { "epoch": 1.95, "learning_rate": 5.7255340468706776e-06, "loss": 0.1396, "step": 2472 }, { "epoch": 1.95, "learning_rate": 5.717819963945959e-06, "loss": 0.1604, "step": 2473 }, { "epoch": 1.95, "learning_rate": 5.710109000147874e-06, "loss": 0.1457, "step": 2474 }, { "epoch": 1.96, "learning_rate": 5.702401161093061e-06, "loss": 0.1236, "step": 2475 }, { "epoch": 1.96, "learning_rate": 5.6946964523958855e-06, "loss": 0.1379, "step": 2476 }, { "epoch": 1.96, "learning_rate": 5.686994879668435e-06, "loss": 0.1564, "step": 2477 }, { "epoch": 1.96, "learning_rate": 5.679296448520509e-06, "loss": 0.1241, "step": 2478 }, { "epoch": 1.96, "learning_rate": 5.671601164559619e-06, "loss": 0.1123, "step": 2479 }, { "epoch": 1.96, "learning_rate": 5.663909033390993e-06, "loss": 0.1393, "step": 2480 }, { "epoch": 1.96, "learning_rate": 5.65622006061755e-06, "loss": 0.136, "step": 2481 }, { "epoch": 1.96, "learning_rate": 5.6485342518399145e-06, "loss": 0.1365, "step": 2482 }, { "epoch": 1.96, "learning_rate": 5.640851612656408e-06, "loss": 0.127, "step": 2483 }, { "epoch": 1.96, "learning_rate": 5.633172148663039e-06, "loss": 0.1225, "step": 2484 }, { "epoch": 1.96, "learning_rate": 5.625495865453505e-06, "loss": 0.1349, "step": 2485 }, { "epoch": 1.96, "learning_rate": 5.617822768619183e-06, "loss": 0.1263, "step": 2486 }, { "epoch": 1.97, "learning_rate": 5.610152863749143e-06, "loss": 0.1278, "step": 2487 }, { "epoch": 1.97, "learning_rate": 5.602486156430113e-06, "loss": 0.1546, "step": 2488 }, { "epoch": 1.97, "learning_rate": 5.5948226522465e-06, "loss": 0.1205, "step": 2489 }, { "epoch": 1.97, "learning_rate": 5.5871623567803756e-06, "loss": 0.1259, "step": 2490 }, { "epoch": 1.97, "learning_rate": 5.579505275611478e-06, "loss": 0.1282, "step": 2491 }, { "epoch": 1.97, "learning_rate": 5.5718514143171945e-06, "loss": 0.1233, "step": 2492 }, { "epoch": 1.97, "learning_rate": 5.564200778472583e-06, "loss": 0.1396, "step": 2493 }, { "epoch": 1.97, "learning_rate": 5.556553373650339e-06, "loss": 0.1284, "step": 2494 }, { "epoch": 1.97, "learning_rate": 5.548909205420807e-06, "loss": 0.1053, "step": 2495 }, { "epoch": 1.97, "learning_rate": 5.5412682793519765e-06, "loss": 0.122, "step": 2496 }, { "epoch": 1.97, "learning_rate": 5.533630601009477e-06, "loss": 0.1218, "step": 2497 }, { "epoch": 1.97, "learning_rate": 5.525996175956565e-06, "loss": 0.1225, "step": 2498 }, { "epoch": 1.97, "learning_rate": 5.5183650097541295e-06, "loss": 0.1073, "step": 2499 }, { "epoch": 1.98, "learning_rate": 5.510737107960698e-06, "loss": 0.1286, "step": 2500 }, { "epoch": 1.98, "learning_rate": 5.503112476132404e-06, "loss": 0.122, "step": 2501 }, { "epoch": 1.98, "learning_rate": 5.495491119823007e-06, "loss": 0.1252, "step": 2502 }, { "epoch": 1.98, "learning_rate": 5.48787304458388e-06, "loss": 0.1399, "step": 2503 }, { "epoch": 1.98, "learning_rate": 5.480258255964004e-06, "loss": 0.1235, "step": 2504 }, { "epoch": 1.98, "learning_rate": 5.472646759509963e-06, "loss": 0.1389, "step": 2505 }, { "epoch": 1.98, "learning_rate": 5.465038560765956e-06, "loss": 0.143, "step": 2506 }, { "epoch": 1.98, "learning_rate": 5.457433665273768e-06, "loss": 0.1716, "step": 2507 }, { "epoch": 1.98, "learning_rate": 5.449832078572781e-06, "loss": 0.1241, "step": 2508 }, { "epoch": 1.98, "learning_rate": 5.442233806199967e-06, "loss": 0.1383, "step": 2509 }, { "epoch": 1.98, "learning_rate": 5.4346388536898844e-06, "loss": 0.1395, "step": 2510 }, { "epoch": 1.98, "learning_rate": 5.427047226574671e-06, "loss": 0.114, "step": 2511 }, { "epoch": 1.98, "learning_rate": 5.419458930384042e-06, "loss": 0.1315, "step": 2512 }, { "epoch": 1.99, "learning_rate": 5.411873970645296e-06, "loss": 0.138, "step": 2513 }, { "epoch": 1.99, "learning_rate": 5.404292352883291e-06, "loss": 0.1238, "step": 2514 }, { "epoch": 1.99, "learning_rate": 5.396714082620455e-06, "loss": 0.1416, "step": 2515 }, { "epoch": 1.99, "learning_rate": 5.389139165376773e-06, "loss": 0.1288, "step": 2516 }, { "epoch": 1.99, "learning_rate": 5.3815676066697946e-06, "loss": 0.1319, "step": 2517 }, { "epoch": 1.99, "learning_rate": 5.3739994120146136e-06, "loss": 0.1343, "step": 2518 }, { "epoch": 1.99, "learning_rate": 5.3664345869238875e-06, "loss": 0.1443, "step": 2519 }, { "epoch": 1.99, "learning_rate": 5.358873136907808e-06, "loss": 0.1149, "step": 2520 }, { "epoch": 1.99, "learning_rate": 5.351315067474112e-06, "loss": 0.1258, "step": 2521 }, { "epoch": 1.99, "learning_rate": 5.343760384128074e-06, "loss": 0.1454, "step": 2522 }, { "epoch": 1.99, "learning_rate": 5.336209092372502e-06, "loss": 0.1459, "step": 2523 }, { "epoch": 1.99, "learning_rate": 5.328661197707732e-06, "loss": 0.1413, "step": 2524 }, { "epoch": 2.0, "learning_rate": 5.321116705631624e-06, "loss": 0.1126, "step": 2525 }, { "epoch": 2.0, "learning_rate": 5.313575621639568e-06, "loss": 0.1367, "step": 2526 }, { "epoch": 2.0, "learning_rate": 5.306037951224468e-06, "loss": 0.1364, "step": 2527 }, { "epoch": 2.0, "learning_rate": 5.298503699876735e-06, "loss": 0.133, "step": 2528 }, { "epoch": 2.0, "learning_rate": 5.290972873084296e-06, "loss": 0.1303, "step": 2529 }, { "epoch": 2.0, "learning_rate": 5.28344547633258e-06, "loss": 0.1264, "step": 2530 }, { "epoch": 2.0, "learning_rate": 5.275921515104519e-06, "loss": 0.1386, "step": 2531 }, { "epoch": 2.0, "learning_rate": 5.268400994880547e-06, "loss": 0.045, "step": 2532 }, { "epoch": 2.0, "learning_rate": 5.260883921138585e-06, "loss": 0.0541, "step": 2533 }, { "epoch": 2.0, "learning_rate": 5.253370299354048e-06, "loss": 0.05, "step": 2534 }, { "epoch": 2.0, "learning_rate": 5.245860134999831e-06, "loss": 0.056, "step": 2535 }, { "epoch": 2.0, "learning_rate": 5.238353433546319e-06, "loss": 0.0558, "step": 2536 }, { "epoch": 2.0, "learning_rate": 5.2308502004613635e-06, "loss": 0.0495, "step": 2537 }, { "epoch": 2.01, "learning_rate": 5.223350441210303e-06, "loss": 0.0519, "step": 2538 }, { "epoch": 2.01, "learning_rate": 5.2158541612559365e-06, "loss": 0.048, "step": 2539 }, { "epoch": 2.01, "learning_rate": 5.20836136605853e-06, "loss": 0.0369, "step": 2540 }, { "epoch": 2.01, "learning_rate": 5.200872061075814e-06, "loss": 0.0463, "step": 2541 }, { "epoch": 2.01, "learning_rate": 5.193386251762973e-06, "loss": 0.0396, "step": 2542 }, { "epoch": 2.01, "learning_rate": 5.185903943572649e-06, "loss": 0.0493, "step": 2543 }, { "epoch": 2.01, "learning_rate": 5.178425141954926e-06, "loss": 0.0393, "step": 2544 }, { "epoch": 2.01, "learning_rate": 5.1709498523573496e-06, "loss": 0.0478, "step": 2545 }, { "epoch": 2.01, "learning_rate": 5.163478080224893e-06, "loss": 0.0439, "step": 2546 }, { "epoch": 2.01, "learning_rate": 5.15600983099997e-06, "loss": 0.0421, "step": 2547 }, { "epoch": 2.01, "learning_rate": 5.148545110122434e-06, "loss": 0.0429, "step": 2548 }, { "epoch": 2.01, "learning_rate": 5.1410839230295615e-06, "loss": 0.0411, "step": 2549 }, { "epoch": 2.02, "learning_rate": 5.133626275156055e-06, "loss": 0.0457, "step": 2550 }, { "epoch": 2.02, "learning_rate": 5.12617217193405e-06, "loss": 0.0327, "step": 2551 }, { "epoch": 2.02, "learning_rate": 5.118721618793086e-06, "loss": 0.0386, "step": 2552 }, { "epoch": 2.02, "learning_rate": 5.111274621160127e-06, "loss": 0.0379, "step": 2553 }, { "epoch": 2.02, "learning_rate": 5.10383118445954e-06, "loss": 0.0468, "step": 2554 }, { "epoch": 2.02, "learning_rate": 5.0963913141131006e-06, "loss": 0.0422, "step": 2555 }, { "epoch": 2.02, "learning_rate": 5.088955015539989e-06, "loss": 0.0419, "step": 2556 }, { "epoch": 2.02, "learning_rate": 5.0815222941567775e-06, "loss": 0.0525, "step": 2557 }, { "epoch": 2.02, "learning_rate": 5.0740931553774464e-06, "loss": 0.039, "step": 2558 }, { "epoch": 2.02, "learning_rate": 5.06666760461335e-06, "loss": 0.0453, "step": 2559 }, { "epoch": 2.02, "learning_rate": 5.059245647273239e-06, "loss": 0.049, "step": 2560 }, { "epoch": 2.02, "learning_rate": 5.051827288763245e-06, "loss": 0.0373, "step": 2561 }, { "epoch": 2.02, "learning_rate": 5.044412534486873e-06, "loss": 0.0408, "step": 2562 }, { "epoch": 2.03, "learning_rate": 5.037001389845009e-06, "loss": 0.042, "step": 2563 }, { "epoch": 2.03, "learning_rate": 5.029593860235911e-06, "loss": 0.0437, "step": 2564 }, { "epoch": 2.03, "learning_rate": 5.0221899510551965e-06, "loss": 0.0346, "step": 2565 }, { "epoch": 2.03, "learning_rate": 5.014789667695853e-06, "loss": 0.0352, "step": 2566 }, { "epoch": 2.03, "learning_rate": 5.007393015548222e-06, "loss": 0.0425, "step": 2567 }, { "epoch": 2.03, "learning_rate": 5.000000000000003e-06, "loss": 0.0367, "step": 2568 }, { "epoch": 2.03, "learning_rate": 4.992610626436244e-06, "loss": 0.0372, "step": 2569 }, { "epoch": 2.03, "learning_rate": 4.985224900239338e-06, "loss": 0.0285, "step": 2570 }, { "epoch": 2.03, "learning_rate": 4.977842826789034e-06, "loss": 0.0567, "step": 2571 }, { "epoch": 2.03, "learning_rate": 4.970464411462408e-06, "loss": 0.0418, "step": 2572 }, { "epoch": 2.03, "learning_rate": 4.963089659633873e-06, "loss": 0.0479, "step": 2573 }, { "epoch": 2.03, "learning_rate": 4.955718576675176e-06, "loss": 0.0396, "step": 2574 }, { "epoch": 2.03, "learning_rate": 4.948351167955391e-06, "loss": 0.036, "step": 2575 }, { "epoch": 2.04, "learning_rate": 4.940987438840912e-06, "loss": 0.038, "step": 2576 }, { "epoch": 2.04, "learning_rate": 4.933627394695464e-06, "loss": 0.0354, "step": 2577 }, { "epoch": 2.04, "learning_rate": 4.9262710408800754e-06, "loss": 0.0346, "step": 2578 }, { "epoch": 2.04, "learning_rate": 4.918918382753093e-06, "loss": 0.0378, "step": 2579 }, { "epoch": 2.04, "learning_rate": 4.911569425670168e-06, "loss": 0.0346, "step": 2580 }, { "epoch": 2.04, "learning_rate": 4.904224174984261e-06, "loss": 0.0373, "step": 2581 }, { "epoch": 2.04, "learning_rate": 4.896882636045624e-06, "loss": 0.043, "step": 2582 }, { "epoch": 2.04, "learning_rate": 4.88954481420182e-06, "loss": 0.0382, "step": 2583 }, { "epoch": 2.04, "learning_rate": 4.882210714797689e-06, "loss": 0.0442, "step": 2584 }, { "epoch": 2.04, "learning_rate": 4.874880343175369e-06, "loss": 0.0333, "step": 2585 }, { "epoch": 2.04, "learning_rate": 4.867553704674279e-06, "loss": 0.0449, "step": 2586 }, { "epoch": 2.04, "learning_rate": 4.86023080463112e-06, "loss": 0.0481, "step": 2587 }, { "epoch": 2.05, "learning_rate": 4.8529116483798685e-06, "loss": 0.0416, "step": 2588 }, { "epoch": 2.05, "learning_rate": 4.845596241251773e-06, "loss": 0.0313, "step": 2589 }, { "epoch": 2.05, "learning_rate": 4.83828458857536e-06, "loss": 0.0345, "step": 2590 }, { "epoch": 2.05, "learning_rate": 4.8309766956764116e-06, "loss": 0.0326, "step": 2591 }, { "epoch": 2.05, "learning_rate": 4.823672567877973e-06, "loss": 0.0391, "step": 2592 }, { "epoch": 2.05, "learning_rate": 4.816372210500348e-06, "loss": 0.0346, "step": 2593 }, { "epoch": 2.05, "learning_rate": 4.809075628861095e-06, "loss": 0.0432, "step": 2594 }, { "epoch": 2.05, "learning_rate": 4.801782828275019e-06, "loss": 0.038, "step": 2595 }, { "epoch": 2.05, "learning_rate": 4.794493814054179e-06, "loss": 0.0406, "step": 2596 }, { "epoch": 2.05, "learning_rate": 4.787208591507867e-06, "loss": 0.0455, "step": 2597 }, { "epoch": 2.05, "learning_rate": 4.779927165942616e-06, "loss": 0.0456, "step": 2598 }, { "epoch": 2.05, "learning_rate": 4.7726495426621945e-06, "loss": 0.0392, "step": 2599 }, { "epoch": 2.05, "learning_rate": 4.7653757269676e-06, "loss": 0.036, "step": 2600 }, { "epoch": 2.06, "learning_rate": 4.758105724157058e-06, "loss": 0.0347, "step": 2601 }, { "epoch": 2.06, "learning_rate": 4.750839539526012e-06, "loss": 0.0369, "step": 2602 }, { "epoch": 2.06, "learning_rate": 4.743577178367136e-06, "loss": 0.0345, "step": 2603 }, { "epoch": 2.06, "learning_rate": 4.7363186459703055e-06, "loss": 0.0413, "step": 2604 }, { "epoch": 2.06, "learning_rate": 4.729063947622617e-06, "loss": 0.0492, "step": 2605 }, { "epoch": 2.06, "learning_rate": 4.721813088608366e-06, "loss": 0.044, "step": 2606 }, { "epoch": 2.06, "learning_rate": 4.714566074209058e-06, "loss": 0.0388, "step": 2607 }, { "epoch": 2.06, "learning_rate": 4.707322909703392e-06, "loss": 0.0347, "step": 2608 }, { "epoch": 2.06, "learning_rate": 4.700083600367273e-06, "loss": 0.0398, "step": 2609 }, { "epoch": 2.06, "learning_rate": 4.692848151473789e-06, "loss": 0.0386, "step": 2610 }, { "epoch": 2.06, "learning_rate": 4.685616568293218e-06, "loss": 0.037, "step": 2611 }, { "epoch": 2.06, "learning_rate": 4.678388856093021e-06, "loss": 0.0422, "step": 2612 }, { "epoch": 2.06, "learning_rate": 4.671165020137844e-06, "loss": 0.0389, "step": 2613 }, { "epoch": 2.07, "learning_rate": 4.663945065689503e-06, "loss": 0.0403, "step": 2614 }, { "epoch": 2.07, "learning_rate": 4.656728998006991e-06, "loss": 0.0483, "step": 2615 }, { "epoch": 2.07, "learning_rate": 4.649516822346474e-06, "loss": 0.0341, "step": 2616 }, { "epoch": 2.07, "learning_rate": 4.6423085439612754e-06, "loss": 0.045, "step": 2617 }, { "epoch": 2.07, "learning_rate": 4.635104168101883e-06, "loss": 0.0379, "step": 2618 }, { "epoch": 2.07, "learning_rate": 4.62790370001594e-06, "loss": 0.0334, "step": 2619 }, { "epoch": 2.07, "learning_rate": 4.620707144948247e-06, "loss": 0.0377, "step": 2620 }, { "epoch": 2.07, "learning_rate": 4.613514508140748e-06, "loss": 0.0479, "step": 2621 }, { "epoch": 2.07, "learning_rate": 4.606325794832545e-06, "loss": 0.044, "step": 2622 }, { "epoch": 2.07, "learning_rate": 4.59914101025987e-06, "loss": 0.0455, "step": 2623 }, { "epoch": 2.07, "learning_rate": 4.591960159656098e-06, "loss": 0.0431, "step": 2624 }, { "epoch": 2.07, "learning_rate": 4.584783248251738e-06, "loss": 0.04, "step": 2625 }, { "epoch": 2.08, "learning_rate": 4.577610281274431e-06, "loss": 0.0443, "step": 2626 }, { "epoch": 2.08, "learning_rate": 4.570441263948939e-06, "loss": 0.0372, "step": 2627 }, { "epoch": 2.08, "learning_rate": 4.563276201497161e-06, "loss": 0.0399, "step": 2628 }, { "epoch": 2.08, "learning_rate": 4.556115099138101e-06, "loss": 0.0435, "step": 2629 }, { "epoch": 2.08, "learning_rate": 4.5489579620878845e-06, "loss": 0.0425, "step": 2630 }, { "epoch": 2.08, "learning_rate": 4.5418047955597465e-06, "loss": 0.0375, "step": 2631 }, { "epoch": 2.08, "learning_rate": 4.5346556047640325e-06, "loss": 0.0343, "step": 2632 }, { "epoch": 2.08, "learning_rate": 4.527510394908191e-06, "loss": 0.0357, "step": 2633 }, { "epoch": 2.08, "learning_rate": 4.520369171196766e-06, "loss": 0.0394, "step": 2634 }, { "epoch": 2.08, "learning_rate": 4.51323193883141e-06, "loss": 0.0406, "step": 2635 }, { "epoch": 2.08, "learning_rate": 4.50609870301086e-06, "loss": 0.0375, "step": 2636 }, { "epoch": 2.08, "learning_rate": 4.4989694689309394e-06, "loss": 0.036, "step": 2637 }, { "epoch": 2.08, "learning_rate": 4.491844241784561e-06, "loss": 0.0418, "step": 2638 }, { "epoch": 2.09, "learning_rate": 4.484723026761719e-06, "loss": 0.0411, "step": 2639 }, { "epoch": 2.09, "learning_rate": 4.477605829049479e-06, "loss": 0.0376, "step": 2640 }, { "epoch": 2.09, "learning_rate": 4.470492653831996e-06, "loss": 0.038, "step": 2641 }, { "epoch": 2.09, "learning_rate": 4.463383506290478e-06, "loss": 0.0353, "step": 2642 }, { "epoch": 2.09, "learning_rate": 4.456278391603207e-06, "loss": 0.0423, "step": 2643 }, { "epoch": 2.09, "learning_rate": 4.449177314945526e-06, "loss": 0.0342, "step": 2644 }, { "epoch": 2.09, "learning_rate": 4.442080281489839e-06, "loss": 0.0432, "step": 2645 }, { "epoch": 2.09, "learning_rate": 4.434987296405602e-06, "loss": 0.0399, "step": 2646 }, { "epoch": 2.09, "learning_rate": 4.427898364859319e-06, "loss": 0.0373, "step": 2647 }, { "epoch": 2.09, "learning_rate": 4.420813492014553e-06, "loss": 0.0468, "step": 2648 }, { "epoch": 2.09, "learning_rate": 4.413732683031901e-06, "loss": 0.0403, "step": 2649 }, { "epoch": 2.09, "learning_rate": 4.4066559430690015e-06, "loss": 0.0295, "step": 2650 }, { "epoch": 2.09, "learning_rate": 4.399583277280529e-06, "loss": 0.0408, "step": 2651 }, { "epoch": 2.1, "learning_rate": 4.392514690818194e-06, "loss": 0.0438, "step": 2652 }, { "epoch": 2.1, "learning_rate": 4.385450188830726e-06, "loss": 0.0351, "step": 2653 }, { "epoch": 2.1, "learning_rate": 4.378389776463895e-06, "loss": 0.0375, "step": 2654 }, { "epoch": 2.1, "learning_rate": 4.37133345886048e-06, "loss": 0.0382, "step": 2655 }, { "epoch": 2.1, "learning_rate": 4.3642812411602806e-06, "loss": 0.0401, "step": 2656 }, { "epoch": 2.1, "learning_rate": 4.3572331285001106e-06, "loss": 0.0387, "step": 2657 }, { "epoch": 2.1, "learning_rate": 4.350189126013793e-06, "loss": 0.0439, "step": 2658 }, { "epoch": 2.1, "learning_rate": 4.343149238832156e-06, "loss": 0.0397, "step": 2659 }, { "epoch": 2.1, "learning_rate": 4.336113472083031e-06, "loss": 0.0391, "step": 2660 }, { "epoch": 2.1, "learning_rate": 4.329081830891253e-06, "loss": 0.0339, "step": 2661 }, { "epoch": 2.1, "learning_rate": 4.322054320378646e-06, "loss": 0.0462, "step": 2662 }, { "epoch": 2.1, "learning_rate": 4.315030945664026e-06, "loss": 0.0395, "step": 2663 }, { "epoch": 2.11, "learning_rate": 4.308011711863196e-06, "loss": 0.0499, "step": 2664 }, { "epoch": 2.11, "learning_rate": 4.3009966240889465e-06, "loss": 0.0425, "step": 2665 }, { "epoch": 2.11, "learning_rate": 4.29398568745104e-06, "loss": 0.0288, "step": 2666 }, { "epoch": 2.11, "learning_rate": 4.28697890705623e-06, "loss": 0.0369, "step": 2667 }, { "epoch": 2.11, "learning_rate": 4.279976288008228e-06, "loss": 0.0352, "step": 2668 }, { "epoch": 2.11, "learning_rate": 4.2729778354077224e-06, "loss": 0.0383, "step": 2669 }, { "epoch": 2.11, "learning_rate": 4.265983554352361e-06, "loss": 0.0345, "step": 2670 }, { "epoch": 2.11, "learning_rate": 4.258993449936759e-06, "loss": 0.0311, "step": 2671 }, { "epoch": 2.11, "learning_rate": 4.252007527252486e-06, "loss": 0.0437, "step": 2672 }, { "epoch": 2.11, "learning_rate": 4.245025791388063e-06, "loss": 0.04, "step": 2673 }, { "epoch": 2.11, "learning_rate": 4.238048247428972e-06, "loss": 0.0385, "step": 2674 }, { "epoch": 2.11, "learning_rate": 4.23107490045763e-06, "loss": 0.0368, "step": 2675 }, { "epoch": 2.11, "learning_rate": 4.224105755553402e-06, "loss": 0.0512, "step": 2676 }, { "epoch": 2.12, "learning_rate": 4.21714081779259e-06, "loss": 0.0477, "step": 2677 }, { "epoch": 2.12, "learning_rate": 4.210180092248435e-06, "loss": 0.0428, "step": 2678 }, { "epoch": 2.12, "learning_rate": 4.203223583991103e-06, "loss": 0.0344, "step": 2679 }, { "epoch": 2.12, "learning_rate": 4.196271298087701e-06, "loss": 0.0396, "step": 2680 }, { "epoch": 2.12, "learning_rate": 4.189323239602249e-06, "loss": 0.0402, "step": 2681 }, { "epoch": 2.12, "learning_rate": 4.18237941359569e-06, "loss": 0.038, "step": 2682 }, { "epoch": 2.12, "learning_rate": 4.175439825125885e-06, "loss": 0.035, "step": 2683 }, { "epoch": 2.12, "learning_rate": 4.168504479247611e-06, "loss": 0.038, "step": 2684 }, { "epoch": 2.12, "learning_rate": 4.161573381012547e-06, "loss": 0.0424, "step": 2685 }, { "epoch": 2.12, "learning_rate": 4.154646535469291e-06, "loss": 0.0363, "step": 2686 }, { "epoch": 2.12, "learning_rate": 4.1477239476633325e-06, "loss": 0.0361, "step": 2687 }, { "epoch": 2.12, "learning_rate": 4.140805622637062e-06, "loss": 0.0337, "step": 2688 }, { "epoch": 2.12, "learning_rate": 4.133891565429768e-06, "loss": 0.0363, "step": 2689 }, { "epoch": 2.13, "learning_rate": 4.126981781077625e-06, "loss": 0.0365, "step": 2690 }, { "epoch": 2.13, "learning_rate": 4.1200762746137e-06, "loss": 0.0453, "step": 2691 }, { "epoch": 2.13, "learning_rate": 4.1131750510679414e-06, "loss": 0.0433, "step": 2692 }, { "epoch": 2.13, "learning_rate": 4.106278115467181e-06, "loss": 0.0367, "step": 2693 }, { "epoch": 2.13, "learning_rate": 4.099385472835128e-06, "loss": 0.0377, "step": 2694 }, { "epoch": 2.13, "learning_rate": 4.092497128192356e-06, "loss": 0.0383, "step": 2695 }, { "epoch": 2.13, "learning_rate": 4.085613086556317e-06, "loss": 0.0317, "step": 2696 }, { "epoch": 2.13, "learning_rate": 4.078733352941322e-06, "loss": 0.0387, "step": 2697 }, { "epoch": 2.13, "learning_rate": 4.071857932358553e-06, "loss": 0.0442, "step": 2698 }, { "epoch": 2.13, "learning_rate": 4.064986829816042e-06, "loss": 0.037, "step": 2699 }, { "epoch": 2.13, "learning_rate": 4.05812005031868e-06, "loss": 0.0386, "step": 2700 }, { "epoch": 2.13, "learning_rate": 4.0512575988682064e-06, "loss": 0.0477, "step": 2701 }, { "epoch": 2.14, "learning_rate": 4.044399480463211e-06, "loss": 0.0397, "step": 2702 }, { "epoch": 2.14, "learning_rate": 4.0375457000991216e-06, "loss": 0.0326, "step": 2703 }, { "epoch": 2.14, "learning_rate": 4.030696262768217e-06, "loss": 0.0387, "step": 2704 }, { "epoch": 2.14, "learning_rate": 4.023851173459602e-06, "loss": 0.0421, "step": 2705 }, { "epoch": 2.14, "learning_rate": 4.01701043715922e-06, "loss": 0.0399, "step": 2706 }, { "epoch": 2.14, "learning_rate": 4.0101740588498416e-06, "loss": 0.0438, "step": 2707 }, { "epoch": 2.14, "learning_rate": 4.003342043511064e-06, "loss": 0.0411, "step": 2708 }, { "epoch": 2.14, "learning_rate": 3.996514396119301e-06, "loss": 0.0438, "step": 2709 }, { "epoch": 2.14, "learning_rate": 3.989691121647798e-06, "loss": 0.0387, "step": 2710 }, { "epoch": 2.14, "learning_rate": 3.982872225066603e-06, "loss": 0.0487, "step": 2711 }, { "epoch": 2.14, "learning_rate": 3.976057711342578e-06, "loss": 0.0419, "step": 2712 }, { "epoch": 2.14, "learning_rate": 3.969247585439395e-06, "loss": 0.042, "step": 2713 }, { "epoch": 2.14, "learning_rate": 3.962441852317529e-06, "loss": 0.0377, "step": 2714 }, { "epoch": 2.15, "learning_rate": 3.95564051693425e-06, "loss": 0.041, "step": 2715 }, { "epoch": 2.15, "learning_rate": 3.948843584243638e-06, "loss": 0.0386, "step": 2716 }, { "epoch": 2.15, "learning_rate": 3.942051059196552e-06, "loss": 0.0433, "step": 2717 }, { "epoch": 2.15, "learning_rate": 3.935262946740648e-06, "loss": 0.0362, "step": 2718 }, { "epoch": 2.15, "learning_rate": 3.928479251820365e-06, "loss": 0.0455, "step": 2719 }, { "epoch": 2.15, "learning_rate": 3.921699979376926e-06, "loss": 0.0375, "step": 2720 }, { "epoch": 2.15, "learning_rate": 3.914925134348328e-06, "loss": 0.0357, "step": 2721 }, { "epoch": 2.15, "learning_rate": 3.908154721669355e-06, "loss": 0.041, "step": 2722 }, { "epoch": 2.15, "learning_rate": 3.9013887462715495e-06, "loss": 0.0322, "step": 2723 }, { "epoch": 2.15, "learning_rate": 3.894627213083227e-06, "loss": 0.0339, "step": 2724 }, { "epoch": 2.15, "learning_rate": 3.887870127029469e-06, "loss": 0.033, "step": 2725 }, { "epoch": 2.15, "learning_rate": 3.881117493032116e-06, "loss": 0.0295, "step": 2726 }, { "epoch": 2.15, "learning_rate": 3.874369316009759e-06, "loss": 0.0358, "step": 2727 }, { "epoch": 2.16, "learning_rate": 3.8676256008777605e-06, "loss": 0.0388, "step": 2728 }, { "epoch": 2.16, "learning_rate": 3.860886352548214e-06, "loss": 0.0316, "step": 2729 }, { "epoch": 2.16, "learning_rate": 3.85415157592997e-06, "loss": 0.0511, "step": 2730 }, { "epoch": 2.16, "learning_rate": 3.8474212759286175e-06, "loss": 0.0469, "step": 2731 }, { "epoch": 2.16, "learning_rate": 3.8406954574464874e-06, "loss": 0.042, "step": 2732 }, { "epoch": 2.16, "learning_rate": 3.833974125382639e-06, "loss": 0.033, "step": 2733 }, { "epoch": 2.16, "learning_rate": 3.8272572846328795e-06, "loss": 0.0336, "step": 2734 }, { "epoch": 2.16, "learning_rate": 3.820544940089729e-06, "loss": 0.0604, "step": 2735 }, { "epoch": 2.16, "learning_rate": 3.8138370966424386e-06, "loss": 0.044, "step": 2736 }, { "epoch": 2.16, "learning_rate": 3.807133759176983e-06, "loss": 0.0406, "step": 2737 }, { "epoch": 2.16, "learning_rate": 3.800434932576047e-06, "loss": 0.0419, "step": 2738 }, { "epoch": 2.16, "learning_rate": 3.793740621719042e-06, "loss": 0.0412, "step": 2739 }, { "epoch": 2.17, "learning_rate": 3.787050831482081e-06, "loss": 0.0428, "step": 2740 }, { "epoch": 2.17, "learning_rate": 3.7803655667379847e-06, "loss": 0.037, "step": 2741 }, { "epoch": 2.17, "learning_rate": 3.7736848323562803e-06, "loss": 0.0445, "step": 2742 }, { "epoch": 2.17, "learning_rate": 3.7670086332031943e-06, "loss": 0.0462, "step": 2743 }, { "epoch": 2.17, "learning_rate": 3.760336974141644e-06, "loss": 0.0378, "step": 2744 }, { "epoch": 2.17, "learning_rate": 3.753669860031254e-06, "loss": 0.0407, "step": 2745 }, { "epoch": 2.17, "learning_rate": 3.7470072957283253e-06, "loss": 0.0396, "step": 2746 }, { "epoch": 2.17, "learning_rate": 3.740349286085849e-06, "loss": 0.0404, "step": 2747 }, { "epoch": 2.17, "learning_rate": 3.7336958359534992e-06, "loss": 0.0374, "step": 2748 }, { "epoch": 2.17, "learning_rate": 3.7270469501776274e-06, "loss": 0.0387, "step": 2749 }, { "epoch": 2.17, "learning_rate": 3.720402633601259e-06, "loss": 0.0341, "step": 2750 }, { "epoch": 2.17, "learning_rate": 3.7137628910640997e-06, "loss": 0.0349, "step": 2751 }, { "epoch": 2.17, "learning_rate": 3.7071277274025142e-06, "loss": 0.0358, "step": 2752 }, { "epoch": 2.18, "learning_rate": 3.7004971474495334e-06, "loss": 0.0391, "step": 2753 }, { "epoch": 2.18, "learning_rate": 3.693871156034854e-06, "loss": 0.0344, "step": 2754 }, { "epoch": 2.18, "learning_rate": 3.6872497579848253e-06, "loss": 0.0343, "step": 2755 }, { "epoch": 2.18, "learning_rate": 3.6806329581224508e-06, "loss": 0.0354, "step": 2756 }, { "epoch": 2.18, "learning_rate": 3.674020761267394e-06, "loss": 0.0314, "step": 2757 }, { "epoch": 2.18, "learning_rate": 3.667413172235954e-06, "loss": 0.0347, "step": 2758 }, { "epoch": 2.18, "learning_rate": 3.660810195841079e-06, "loss": 0.0418, "step": 2759 }, { "epoch": 2.18, "learning_rate": 3.6542118368923562e-06, "loss": 0.0377, "step": 2760 }, { "epoch": 2.18, "learning_rate": 3.64761810019601e-06, "loss": 0.0352, "step": 2761 }, { "epoch": 2.18, "learning_rate": 3.6410289905548967e-06, "loss": 0.031, "step": 2762 }, { "epoch": 2.18, "learning_rate": 3.634444512768501e-06, "loss": 0.0449, "step": 2763 }, { "epoch": 2.18, "learning_rate": 3.6278646716329423e-06, "loss": 0.035, "step": 2764 }, { "epoch": 2.18, "learning_rate": 3.621289471940954e-06, "loss": 0.0412, "step": 2765 }, { "epoch": 2.19, "learning_rate": 3.61471891848189e-06, "loss": 0.0457, "step": 2766 }, { "epoch": 2.19, "learning_rate": 3.608153016041721e-06, "loss": 0.0397, "step": 2767 }, { "epoch": 2.19, "learning_rate": 3.601591769403031e-06, "loss": 0.0378, "step": 2768 }, { "epoch": 2.19, "learning_rate": 3.595035183345007e-06, "loss": 0.0412, "step": 2769 }, { "epoch": 2.19, "learning_rate": 3.588483262643454e-06, "loss": 0.0354, "step": 2770 }, { "epoch": 2.19, "learning_rate": 3.581936012070766e-06, "loss": 0.0406, "step": 2771 }, { "epoch": 2.19, "learning_rate": 3.575393436395941e-06, "loss": 0.0384, "step": 2772 }, { "epoch": 2.19, "learning_rate": 3.5688555403845694e-06, "loss": 0.0379, "step": 2773 }, { "epoch": 2.19, "learning_rate": 3.562322328798836e-06, "loss": 0.0298, "step": 2774 }, { "epoch": 2.19, "learning_rate": 3.5557938063975105e-06, "loss": 0.0404, "step": 2775 }, { "epoch": 2.19, "learning_rate": 3.5492699779359463e-06, "loss": 0.0497, "step": 2776 }, { "epoch": 2.19, "learning_rate": 3.5427508481660856e-06, "loss": 0.0314, "step": 2777 }, { "epoch": 2.2, "learning_rate": 3.5362364218364387e-06, "loss": 0.0334, "step": 2778 }, { "epoch": 2.2, "learning_rate": 3.5297267036920935e-06, "loss": 0.0296, "step": 2779 }, { "epoch": 2.2, "learning_rate": 3.523221698474709e-06, "loss": 0.038, "step": 2780 }, { "epoch": 2.2, "learning_rate": 3.5167214109225113e-06, "loss": 0.0446, "step": 2781 }, { "epoch": 2.2, "learning_rate": 3.5102258457702853e-06, "loss": 0.0347, "step": 2782 }, { "epoch": 2.2, "learning_rate": 3.503735007749387e-06, "loss": 0.0372, "step": 2783 }, { "epoch": 2.2, "learning_rate": 3.497248901587721e-06, "loss": 0.0331, "step": 2784 }, { "epoch": 2.2, "learning_rate": 3.4907675320097443e-06, "loss": 0.0426, "step": 2785 }, { "epoch": 2.2, "learning_rate": 3.484290903736468e-06, "loss": 0.039, "step": 2786 }, { "epoch": 2.2, "learning_rate": 3.477819021485448e-06, "loss": 0.0379, "step": 2787 }, { "epoch": 2.2, "learning_rate": 3.4713518899707787e-06, "loss": 0.035, "step": 2788 }, { "epoch": 2.2, "learning_rate": 3.464889513903107e-06, "loss": 0.0281, "step": 2789 }, { "epoch": 2.2, "learning_rate": 3.4584318979896028e-06, "loss": 0.0377, "step": 2790 }, { "epoch": 2.21, "learning_rate": 3.4519790469339743e-06, "loss": 0.0402, "step": 2791 }, { "epoch": 2.21, "learning_rate": 3.4455309654364577e-06, "loss": 0.0459, "step": 2792 }, { "epoch": 2.21, "learning_rate": 3.439087658193816e-06, "loss": 0.0308, "step": 2793 }, { "epoch": 2.21, "learning_rate": 3.4326491298993336e-06, "loss": 0.0297, "step": 2794 }, { "epoch": 2.21, "learning_rate": 3.4262153852428114e-06, "loss": 0.0387, "step": 2795 }, { "epoch": 2.21, "learning_rate": 3.4197864289105763e-06, "loss": 0.0395, "step": 2796 }, { "epoch": 2.21, "learning_rate": 3.413362265585455e-06, "loss": 0.0422, "step": 2797 }, { "epoch": 2.21, "learning_rate": 3.4069428999467904e-06, "loss": 0.0359, "step": 2798 }, { "epoch": 2.21, "learning_rate": 3.4005283366704268e-06, "loss": 0.044, "step": 2799 }, { "epoch": 2.21, "learning_rate": 3.3941185804287125e-06, "loss": 0.0361, "step": 2800 }, { "epoch": 2.21, "learning_rate": 3.3877136358904915e-06, "loss": 0.0414, "step": 2801 }, { "epoch": 2.21, "learning_rate": 3.381313507721111e-06, "loss": 0.0372, "step": 2802 }, { "epoch": 2.21, "learning_rate": 3.374918200582402e-06, "loss": 0.0466, "step": 2803 }, { "epoch": 2.22, "learning_rate": 3.368527719132686e-06, "loss": 0.0436, "step": 2804 }, { "epoch": 2.22, "learning_rate": 3.36214206802677e-06, "loss": 0.0335, "step": 2805 }, { "epoch": 2.22, "learning_rate": 3.3557612519159423e-06, "loss": 0.048, "step": 2806 }, { "epoch": 2.22, "learning_rate": 3.3493852754479696e-06, "loss": 0.037, "step": 2807 }, { "epoch": 2.22, "learning_rate": 3.343014143267089e-06, "loss": 0.0362, "step": 2808 }, { "epoch": 2.22, "learning_rate": 3.336647860014023e-06, "loss": 0.0412, "step": 2809 }, { "epoch": 2.22, "learning_rate": 3.3302864303259453e-06, "loss": 0.0409, "step": 2810 }, { "epoch": 2.22, "learning_rate": 3.3239298588365045e-06, "loss": 0.0302, "step": 2811 }, { "epoch": 2.22, "learning_rate": 3.3175781501758054e-06, "loss": 0.0285, "step": 2812 }, { "epoch": 2.22, "learning_rate": 3.311231308970414e-06, "loss": 0.0464, "step": 2813 }, { "epoch": 2.22, "learning_rate": 3.304889339843347e-06, "loss": 0.0379, "step": 2814 }, { "epoch": 2.22, "learning_rate": 3.2985522474140807e-06, "loss": 0.0359, "step": 2815 }, { "epoch": 2.23, "learning_rate": 3.2922200362985303e-06, "loss": 0.0432, "step": 2816 }, { "epoch": 2.23, "learning_rate": 3.285892711109059e-06, "loss": 0.0325, "step": 2817 }, { "epoch": 2.23, "learning_rate": 3.27957027645447e-06, "loss": 0.0375, "step": 2818 }, { "epoch": 2.23, "learning_rate": 3.273252736940006e-06, "loss": 0.0389, "step": 2819 }, { "epoch": 2.23, "learning_rate": 3.2669400971673425e-06, "loss": 0.0378, "step": 2820 }, { "epoch": 2.23, "learning_rate": 3.2606323617345835e-06, "loss": 0.0427, "step": 2821 }, { "epoch": 2.23, "learning_rate": 3.2543295352362714e-06, "loss": 0.0336, "step": 2822 }, { "epoch": 2.23, "learning_rate": 3.2480316222633614e-06, "loss": 0.0423, "step": 2823 }, { "epoch": 2.23, "learning_rate": 3.2417386274032337e-06, "loss": 0.0275, "step": 2824 }, { "epoch": 2.23, "learning_rate": 3.2354505552396863e-06, "loss": 0.0381, "step": 2825 }, { "epoch": 2.23, "learning_rate": 3.229167410352931e-06, "loss": 0.033, "step": 2826 }, { "epoch": 2.23, "learning_rate": 3.2228891973195887e-06, "loss": 0.0363, "step": 2827 }, { "epoch": 2.23, "learning_rate": 3.2166159207126956e-06, "loss": 0.0371, "step": 2828 }, { "epoch": 2.24, "learning_rate": 3.210347585101684e-06, "loss": 0.0377, "step": 2829 }, { "epoch": 2.24, "learning_rate": 3.2040841950523914e-06, "loss": 0.043, "step": 2830 }, { "epoch": 2.24, "learning_rate": 3.19782575512705e-06, "loss": 0.0305, "step": 2831 }, { "epoch": 2.24, "learning_rate": 3.1915722698842877e-06, "loss": 0.0369, "step": 2832 }, { "epoch": 2.24, "learning_rate": 3.1853237438791216e-06, "loss": 0.04, "step": 2833 }, { "epoch": 2.24, "learning_rate": 3.1790801816629657e-06, "loss": 0.0357, "step": 2834 }, { "epoch": 2.24, "learning_rate": 3.1728415877836072e-06, "loss": 0.0295, "step": 2835 }, { "epoch": 2.24, "learning_rate": 3.166607966785218e-06, "loss": 0.0346, "step": 2836 }, { "epoch": 2.24, "learning_rate": 3.1603793232083493e-06, "loss": 0.04, "step": 2837 }, { "epoch": 2.24, "learning_rate": 3.154155661589924e-06, "loss": 0.0361, "step": 2838 }, { "epoch": 2.24, "learning_rate": 3.1479369864632403e-06, "loss": 0.0419, "step": 2839 }, { "epoch": 2.24, "learning_rate": 3.1417233023579565e-06, "loss": 0.0405, "step": 2840 }, { "epoch": 2.24, "learning_rate": 3.135514613800108e-06, "loss": 0.0363, "step": 2841 }, { "epoch": 2.25, "learning_rate": 3.1293109253120825e-06, "loss": 0.0276, "step": 2842 }, { "epoch": 2.25, "learning_rate": 3.1231122414126245e-06, "loss": 0.0366, "step": 2843 }, { "epoch": 2.25, "learning_rate": 3.1169185666168377e-06, "loss": 0.0402, "step": 2844 }, { "epoch": 2.25, "learning_rate": 3.1107299054361763e-06, "loss": 0.0328, "step": 2845 }, { "epoch": 2.25, "learning_rate": 3.104546262378438e-06, "loss": 0.0317, "step": 2846 }, { "epoch": 2.25, "learning_rate": 3.098367641947775e-06, "loss": 0.0415, "step": 2847 }, { "epoch": 2.25, "learning_rate": 3.0921940486446724e-06, "loss": 0.037, "step": 2848 }, { "epoch": 2.25, "learning_rate": 3.086025486965957e-06, "loss": 0.0305, "step": 2849 }, { "epoch": 2.25, "learning_rate": 3.0798619614047885e-06, "loss": 0.0412, "step": 2850 }, { "epoch": 2.25, "learning_rate": 3.073703476450661e-06, "loss": 0.0438, "step": 2851 }, { "epoch": 2.25, "learning_rate": 3.067550036589396e-06, "loss": 0.034, "step": 2852 }, { "epoch": 2.25, "learning_rate": 3.061401646303136e-06, "loss": 0.043, "step": 2853 }, { "epoch": 2.26, "learning_rate": 3.0552583100703558e-06, "loss": 0.0389, "step": 2854 }, { "epoch": 2.26, "learning_rate": 3.04912003236584e-06, "loss": 0.0377, "step": 2855 }, { "epoch": 2.26, "learning_rate": 3.04298681766069e-06, "loss": 0.0336, "step": 2856 }, { "epoch": 2.26, "learning_rate": 3.036858670422321e-06, "loss": 0.039, "step": 2857 }, { "epoch": 2.26, "learning_rate": 3.030735595114456e-06, "loss": 0.0371, "step": 2858 }, { "epoch": 2.26, "learning_rate": 3.024617596197121e-06, "loss": 0.0375, "step": 2859 }, { "epoch": 2.26, "learning_rate": 3.0185046781266537e-06, "loss": 0.0558, "step": 2860 }, { "epoch": 2.26, "learning_rate": 3.012396845355682e-06, "loss": 0.0363, "step": 2861 }, { "epoch": 2.26, "learning_rate": 3.0062941023331316e-06, "loss": 0.0341, "step": 2862 }, { "epoch": 2.26, "learning_rate": 3.0001964535042216e-06, "loss": 0.0382, "step": 2863 }, { "epoch": 2.26, "learning_rate": 2.9941039033104614e-06, "loss": 0.0357, "step": 2864 }, { "epoch": 2.26, "learning_rate": 2.988016456189644e-06, "loss": 0.0383, "step": 2865 }, { "epoch": 2.26, "learning_rate": 2.9819341165758455e-06, "loss": 0.0441, "step": 2866 }, { "epoch": 2.27, "learning_rate": 2.9758568888994275e-06, "loss": 0.0386, "step": 2867 }, { "epoch": 2.27, "learning_rate": 2.9697847775870227e-06, "loss": 0.038, "step": 2868 }, { "epoch": 2.27, "learning_rate": 2.9637177870615365e-06, "loss": 0.0343, "step": 2869 }, { "epoch": 2.27, "learning_rate": 2.9576559217421463e-06, "loss": 0.0461, "step": 2870 }, { "epoch": 2.27, "learning_rate": 2.9515991860442973e-06, "loss": 0.0464, "step": 2871 }, { "epoch": 2.27, "learning_rate": 2.945547584379692e-06, "loss": 0.044, "step": 2872 }, { "epoch": 2.27, "learning_rate": 2.939501121156306e-06, "loss": 0.0415, "step": 2873 }, { "epoch": 2.27, "learning_rate": 2.933459800778361e-06, "loss": 0.0355, "step": 2874 }, { "epoch": 2.27, "learning_rate": 2.927423627646334e-06, "loss": 0.0311, "step": 2875 }, { "epoch": 2.27, "learning_rate": 2.9213926061569575e-06, "loss": 0.03, "step": 2876 }, { "epoch": 2.27, "learning_rate": 2.9153667407032073e-06, "loss": 0.0392, "step": 2877 }, { "epoch": 2.27, "learning_rate": 2.909346035674301e-06, "loss": 0.0327, "step": 2878 }, { "epoch": 2.27, "learning_rate": 2.9033304954557105e-06, "loss": 0.041, "step": 2879 }, { "epoch": 2.28, "learning_rate": 2.8973201244291305e-06, "loss": 0.0353, "step": 2880 }, { "epoch": 2.28, "learning_rate": 2.8913149269724983e-06, "loss": 0.0405, "step": 2881 }, { "epoch": 2.28, "learning_rate": 2.8853149074599818e-06, "loss": 0.0413, "step": 2882 }, { "epoch": 2.28, "learning_rate": 2.879320070261974e-06, "loss": 0.0324, "step": 2883 }, { "epoch": 2.28, "learning_rate": 2.8733304197450985e-06, "loss": 0.0488, "step": 2884 }, { "epoch": 2.28, "learning_rate": 2.867345960272192e-06, "loss": 0.0424, "step": 2885 }, { "epoch": 2.28, "learning_rate": 2.861366696202326e-06, "loss": 0.0328, "step": 2886 }, { "epoch": 2.28, "learning_rate": 2.855392631890771e-06, "loss": 0.0415, "step": 2887 }, { "epoch": 2.28, "learning_rate": 2.8494237716890195e-06, "loss": 0.0303, "step": 2888 }, { "epoch": 2.28, "learning_rate": 2.8434601199447698e-06, "loss": 0.0349, "step": 2889 }, { "epoch": 2.28, "learning_rate": 2.8375016810019274e-06, "loss": 0.0362, "step": 2890 }, { "epoch": 2.28, "learning_rate": 2.831548459200597e-06, "loss": 0.0389, "step": 2891 }, { "epoch": 2.29, "learning_rate": 2.825600458877095e-06, "loss": 0.0428, "step": 2892 }, { "epoch": 2.29, "learning_rate": 2.819657684363921e-06, "loss": 0.0327, "step": 2893 }, { "epoch": 2.29, "learning_rate": 2.813720139989775e-06, "loss": 0.0386, "step": 2894 }, { "epoch": 2.29, "learning_rate": 2.8077878300795446e-06, "loss": 0.0391, "step": 2895 }, { "epoch": 2.29, "learning_rate": 2.8018607589543067e-06, "loss": 0.027, "step": 2896 }, { "epoch": 2.29, "learning_rate": 2.7959389309313214e-06, "loss": 0.0438, "step": 2897 }, { "epoch": 2.29, "learning_rate": 2.7900223503240265e-06, "loss": 0.0407, "step": 2898 }, { "epoch": 2.29, "learning_rate": 2.7841110214420485e-06, "loss": 0.0406, "step": 2899 }, { "epoch": 2.29, "learning_rate": 2.7782049485911753e-06, "loss": 0.0402, "step": 2900 }, { "epoch": 2.29, "learning_rate": 2.7723041360733737e-06, "loss": 0.0425, "step": 2901 }, { "epoch": 2.29, "learning_rate": 2.7664085881867777e-06, "loss": 0.0358, "step": 2902 }, { "epoch": 2.29, "learning_rate": 2.760518309225685e-06, "loss": 0.0521, "step": 2903 }, { "epoch": 2.29, "learning_rate": 2.7546333034805528e-06, "loss": 0.0349, "step": 2904 }, { "epoch": 2.3, "learning_rate": 2.748753575238009e-06, "loss": 0.0365, "step": 2905 }, { "epoch": 2.3, "learning_rate": 2.7428791287808256e-06, "loss": 0.0283, "step": 2906 }, { "epoch": 2.3, "learning_rate": 2.737009968387929e-06, "loss": 0.0325, "step": 2907 }, { "epoch": 2.3, "learning_rate": 2.7311460983344007e-06, "loss": 0.0336, "step": 2908 }, { "epoch": 2.3, "learning_rate": 2.7252875228914643e-06, "loss": 0.0415, "step": 2909 }, { "epoch": 2.3, "learning_rate": 2.719434246326487e-06, "loss": 0.0422, "step": 2910 }, { "epoch": 2.3, "learning_rate": 2.713586272902975e-06, "loss": 0.0443, "step": 2911 }, { "epoch": 2.3, "learning_rate": 2.7077436068805796e-06, "loss": 0.0438, "step": 2912 }, { "epoch": 2.3, "learning_rate": 2.7019062525150783e-06, "loss": 0.0331, "step": 2913 }, { "epoch": 2.3, "learning_rate": 2.696074214058382e-06, "loss": 0.0381, "step": 2914 }, { "epoch": 2.3, "learning_rate": 2.690247495758528e-06, "loss": 0.0322, "step": 2915 }, { "epoch": 2.3, "learning_rate": 2.6844261018596806e-06, "loss": 0.0379, "step": 2916 }, { "epoch": 2.31, "learning_rate": 2.678610036602122e-06, "loss": 0.0413, "step": 2917 }, { "epoch": 2.31, "learning_rate": 2.6727993042222624e-06, "loss": 0.0337, "step": 2918 }, { "epoch": 2.31, "learning_rate": 2.6669939089526177e-06, "loss": 0.045, "step": 2919 }, { "epoch": 2.31, "learning_rate": 2.66119385502182e-06, "loss": 0.0409, "step": 2920 }, { "epoch": 2.31, "learning_rate": 2.655399146654609e-06, "loss": 0.0399, "step": 2921 }, { "epoch": 2.31, "learning_rate": 2.6496097880718364e-06, "loss": 0.0389, "step": 2922 }, { "epoch": 2.31, "learning_rate": 2.643825783490449e-06, "loss": 0.0336, "step": 2923 }, { "epoch": 2.31, "learning_rate": 2.6380471371234963e-06, "loss": 0.0356, "step": 2924 }, { "epoch": 2.31, "learning_rate": 2.632273853180132e-06, "loss": 0.0335, "step": 2925 }, { "epoch": 2.31, "learning_rate": 2.6265059358655955e-06, "loss": 0.0292, "step": 2926 }, { "epoch": 2.31, "learning_rate": 2.6207433893812194e-06, "loss": 0.0302, "step": 2927 }, { "epoch": 2.31, "learning_rate": 2.6149862179244257e-06, "loss": 0.038, "step": 2928 }, { "epoch": 2.31, "learning_rate": 2.6092344256887193e-06, "loss": 0.0435, "step": 2929 }, { "epoch": 2.32, "learning_rate": 2.6034880168636846e-06, "loss": 0.034, "step": 2930 }, { "epoch": 2.32, "learning_rate": 2.5977469956349956e-06, "loss": 0.0485, "step": 2931 }, { "epoch": 2.32, "learning_rate": 2.5920113661843883e-06, "loss": 0.0383, "step": 2932 }, { "epoch": 2.32, "learning_rate": 2.58628113268968e-06, "loss": 0.0469, "step": 2933 }, { "epoch": 2.32, "learning_rate": 2.5805562993247536e-06, "loss": 0.0408, "step": 2934 }, { "epoch": 2.32, "learning_rate": 2.5748368702595606e-06, "loss": 0.0336, "step": 2935 }, { "epoch": 2.32, "learning_rate": 2.5691228496601117e-06, "loss": 0.0411, "step": 2936 }, { "epoch": 2.32, "learning_rate": 2.563414241688489e-06, "loss": 0.0341, "step": 2937 }, { "epoch": 2.32, "learning_rate": 2.55771105050282e-06, "loss": 0.0357, "step": 2938 }, { "epoch": 2.32, "learning_rate": 2.552013280257294e-06, "loss": 0.0338, "step": 2939 }, { "epoch": 2.32, "learning_rate": 2.5463209351021457e-06, "loss": 0.0359, "step": 2940 }, { "epoch": 2.32, "learning_rate": 2.5406340191836653e-06, "loss": 0.0392, "step": 2941 }, { "epoch": 2.32, "learning_rate": 2.5349525366441818e-06, "loss": 0.0361, "step": 2942 }, { "epoch": 2.33, "learning_rate": 2.529276491622067e-06, "loss": 0.0325, "step": 2943 }, { "epoch": 2.33, "learning_rate": 2.523605888251742e-06, "loss": 0.0365, "step": 2944 }, { "epoch": 2.33, "learning_rate": 2.517940730663652e-06, "loss": 0.0365, "step": 2945 }, { "epoch": 2.33, "learning_rate": 2.5122810229842807e-06, "loss": 0.0316, "step": 2946 }, { "epoch": 2.33, "learning_rate": 2.5066267693361424e-06, "loss": 0.0317, "step": 2947 }, { "epoch": 2.33, "learning_rate": 2.500977973837776e-06, "loss": 0.0385, "step": 2948 }, { "epoch": 2.33, "learning_rate": 2.495334640603746e-06, "loss": 0.04, "step": 2949 }, { "epoch": 2.33, "learning_rate": 2.4896967737446433e-06, "loss": 0.0312, "step": 2950 }, { "epoch": 2.33, "learning_rate": 2.4840643773670693e-06, "loss": 0.0346, "step": 2951 }, { "epoch": 2.33, "learning_rate": 2.4784374555736445e-06, "loss": 0.0411, "step": 2952 }, { "epoch": 2.33, "learning_rate": 2.4728160124630017e-06, "loss": 0.0401, "step": 2953 }, { "epoch": 2.33, "learning_rate": 2.4672000521297824e-06, "loss": 0.0311, "step": 2954 }, { "epoch": 2.34, "learning_rate": 2.4615895786646337e-06, "loss": 0.0452, "step": 2955 }, { "epoch": 2.34, "learning_rate": 2.455984596154205e-06, "loss": 0.0422, "step": 2956 }, { "epoch": 2.34, "learning_rate": 2.4503851086811547e-06, "loss": 0.0345, "step": 2957 }, { "epoch": 2.34, "learning_rate": 2.444791120324127e-06, "loss": 0.046, "step": 2958 }, { "epoch": 2.34, "learning_rate": 2.439202635157768e-06, "loss": 0.046, "step": 2959 }, { "epoch": 2.34, "learning_rate": 2.4336196572527114e-06, "loss": 0.0414, "step": 2960 }, { "epoch": 2.34, "learning_rate": 2.4280421906755814e-06, "loss": 0.0407, "step": 2961 }, { "epoch": 2.34, "learning_rate": 2.4224702394889843e-06, "loss": 0.0423, "step": 2962 }, { "epoch": 2.34, "learning_rate": 2.416903807751518e-06, "loss": 0.0356, "step": 2963 }, { "epoch": 2.34, "learning_rate": 2.4113428995177522e-06, "loss": 0.0304, "step": 2964 }, { "epoch": 2.34, "learning_rate": 2.405787518838233e-06, "loss": 0.0369, "step": 2965 }, { "epoch": 2.34, "learning_rate": 2.400237669759484e-06, "loss": 0.0306, "step": 2966 }, { "epoch": 2.34, "learning_rate": 2.394693356323997e-06, "loss": 0.0335, "step": 2967 }, { "epoch": 2.35, "learning_rate": 2.3891545825702333e-06, "loss": 0.0401, "step": 2968 }, { "epoch": 2.35, "learning_rate": 2.383621352532616e-06, "loss": 0.0419, "step": 2969 }, { "epoch": 2.35, "learning_rate": 2.378093670241538e-06, "loss": 0.0326, "step": 2970 }, { "epoch": 2.35, "learning_rate": 2.372571539723343e-06, "loss": 0.0416, "step": 2971 }, { "epoch": 2.35, "learning_rate": 2.3670549650003326e-06, "loss": 0.0386, "step": 2972 }, { "epoch": 2.35, "learning_rate": 2.3615439500907657e-06, "loss": 0.0355, "step": 2973 }, { "epoch": 2.35, "learning_rate": 2.3560384990088447e-06, "loss": 0.0362, "step": 2974 }, { "epoch": 2.35, "learning_rate": 2.3505386157647227e-06, "loss": 0.0288, "step": 2975 }, { "epoch": 2.35, "learning_rate": 2.3450443043645035e-06, "loss": 0.0343, "step": 2976 }, { "epoch": 2.35, "learning_rate": 2.339555568810221e-06, "loss": 0.0351, "step": 2977 }, { "epoch": 2.35, "learning_rate": 2.334072413099856e-06, "loss": 0.0308, "step": 2978 }, { "epoch": 2.35, "learning_rate": 2.3285948412273198e-06, "loss": 0.0328, "step": 2979 }, { "epoch": 2.35, "learning_rate": 2.3231228571824603e-06, "loss": 0.0369, "step": 2980 }, { "epoch": 2.36, "learning_rate": 2.3176564649510515e-06, "loss": 0.0401, "step": 2981 }, { "epoch": 2.36, "learning_rate": 2.3121956685147995e-06, "loss": 0.0426, "step": 2982 }, { "epoch": 2.36, "learning_rate": 2.306740471851331e-06, "loss": 0.0355, "step": 2983 }, { "epoch": 2.36, "learning_rate": 2.3012908789341927e-06, "loss": 0.0384, "step": 2984 }, { "epoch": 2.36, "learning_rate": 2.2958468937328528e-06, "loss": 0.0371, "step": 2985 }, { "epoch": 2.36, "learning_rate": 2.2904085202126914e-06, "loss": 0.0375, "step": 2986 }, { "epoch": 2.36, "learning_rate": 2.2849757623350046e-06, "loss": 0.0391, "step": 2987 }, { "epoch": 2.36, "learning_rate": 2.279548624056992e-06, "loss": 0.0372, "step": 2988 }, { "epoch": 2.36, "learning_rate": 2.274127109331771e-06, "loss": 0.0383, "step": 2989 }, { "epoch": 2.36, "learning_rate": 2.268711222108352e-06, "loss": 0.0408, "step": 2990 }, { "epoch": 2.36, "learning_rate": 2.263300966331652e-06, "loss": 0.0372, "step": 2991 }, { "epoch": 2.36, "learning_rate": 2.257896345942484e-06, "loss": 0.0435, "step": 2992 }, { "epoch": 2.37, "learning_rate": 2.252497364877556e-06, "loss": 0.0393, "step": 2993 }, { "epoch": 2.37, "learning_rate": 2.247104027069467e-06, "loss": 0.0352, "step": 2994 }, { "epoch": 2.37, "learning_rate": 2.241716336446712e-06, "loss": 0.0328, "step": 2995 }, { "epoch": 2.37, "learning_rate": 2.2363342969336666e-06, "loss": 0.0336, "step": 2996 }, { "epoch": 2.37, "learning_rate": 2.230957912450592e-06, "loss": 0.0454, "step": 2997 }, { "epoch": 2.37, "learning_rate": 2.225587186913628e-06, "loss": 0.0302, "step": 2998 }, { "epoch": 2.37, "learning_rate": 2.2202221242347956e-06, "loss": 0.034, "step": 2999 }, { "epoch": 2.37, "learning_rate": 2.214862728321987e-06, "loss": 0.039, "step": 3000 }, { "epoch": 2.37, "learning_rate": 2.2095090030789747e-06, "loss": 0.0358, "step": 3001 }, { "epoch": 2.37, "learning_rate": 2.204160952405393e-06, "loss": 0.0355, "step": 3002 }, { "epoch": 2.37, "learning_rate": 2.1988185801967464e-06, "loss": 0.0347, "step": 3003 }, { "epoch": 2.37, "learning_rate": 2.1934818903443998e-06, "loss": 0.0357, "step": 3004 }, { "epoch": 2.37, "learning_rate": 2.1881508867355837e-06, "loss": 0.0292, "step": 3005 }, { "epoch": 2.38, "learning_rate": 2.182825573253382e-06, "loss": 0.0321, "step": 3006 }, { "epoch": 2.38, "learning_rate": 2.1775059537767396e-06, "loss": 0.0365, "step": 3007 }, { "epoch": 2.38, "learning_rate": 2.1721920321804502e-06, "loss": 0.0391, "step": 3008 }, { "epoch": 2.38, "learning_rate": 2.1668838123351566e-06, "loss": 0.0337, "step": 3009 }, { "epoch": 2.38, "learning_rate": 2.1615812981073503e-06, "loss": 0.0358, "step": 3010 }, { "epoch": 2.38, "learning_rate": 2.156284493359365e-06, "loss": 0.0355, "step": 3011 }, { "epoch": 2.38, "learning_rate": 2.150993401949376e-06, "loss": 0.0299, "step": 3012 }, { "epoch": 2.38, "learning_rate": 2.1457080277314024e-06, "loss": 0.0464, "step": 3013 }, { "epoch": 2.38, "learning_rate": 2.140428374555291e-06, "loss": 0.0364, "step": 3014 }, { "epoch": 2.38, "learning_rate": 2.135154446266726e-06, "loss": 0.0307, "step": 3015 }, { "epoch": 2.38, "learning_rate": 2.12988624670722e-06, "loss": 0.0362, "step": 3016 }, { "epoch": 2.38, "learning_rate": 2.1246237797141122e-06, "loss": 0.0402, "step": 3017 }, { "epoch": 2.38, "learning_rate": 2.119367049120565e-06, "loss": 0.0349, "step": 3018 }, { "epoch": 2.39, "learning_rate": 2.114116058755571e-06, "loss": 0.0397, "step": 3019 }, { "epoch": 2.39, "learning_rate": 2.1088708124439315e-06, "loss": 0.0419, "step": 3020 }, { "epoch": 2.39, "learning_rate": 2.103631314006267e-06, "loss": 0.042, "step": 3021 }, { "epoch": 2.39, "learning_rate": 2.098397567259014e-06, "loss": 0.0309, "step": 3022 }, { "epoch": 2.39, "learning_rate": 2.0931695760144155e-06, "loss": 0.0363, "step": 3023 }, { "epoch": 2.39, "learning_rate": 2.087947344080522e-06, "loss": 0.0492, "step": 3024 }, { "epoch": 2.39, "learning_rate": 2.0827308752611973e-06, "loss": 0.0382, "step": 3025 }, { "epoch": 2.39, "learning_rate": 2.0775201733560967e-06, "loss": 0.041, "step": 3026 }, { "epoch": 2.39, "learning_rate": 2.0723152421606805e-06, "loss": 0.0313, "step": 3027 }, { "epoch": 2.39, "learning_rate": 2.067116085466203e-06, "loss": 0.0327, "step": 3028 }, { "epoch": 2.39, "learning_rate": 2.061922707059716e-06, "loss": 0.0396, "step": 3029 }, { "epoch": 2.39, "learning_rate": 2.0567351107240563e-06, "loss": 0.033, "step": 3030 }, { "epoch": 2.4, "learning_rate": 2.0515533002378574e-06, "loss": 0.0299, "step": 3031 }, { "epoch": 2.4, "learning_rate": 2.046377279375532e-06, "loss": 0.0408, "step": 3032 }, { "epoch": 2.4, "learning_rate": 2.041207051907279e-06, "loss": 0.0369, "step": 3033 }, { "epoch": 2.4, "learning_rate": 2.0360426215990738e-06, "loss": 0.0351, "step": 3034 }, { "epoch": 2.4, "learning_rate": 2.03088399221267e-06, "loss": 0.0397, "step": 3035 }, { "epoch": 2.4, "learning_rate": 2.0257311675056025e-06, "loss": 0.0345, "step": 3036 }, { "epoch": 2.4, "learning_rate": 2.0205841512311697e-06, "loss": 0.0298, "step": 3037 }, { "epoch": 2.4, "learning_rate": 2.0154429471384427e-06, "loss": 0.0336, "step": 3038 }, { "epoch": 2.4, "learning_rate": 2.0103075589722576e-06, "loss": 0.0394, "step": 3039 }, { "epoch": 2.4, "learning_rate": 2.0051779904732173e-06, "loss": 0.0393, "step": 3040 }, { "epoch": 2.4, "learning_rate": 2.0000542453776785e-06, "loss": 0.0397, "step": 3041 }, { "epoch": 2.4, "learning_rate": 1.9949363274177667e-06, "loss": 0.039, "step": 3042 }, { "epoch": 2.4, "learning_rate": 1.9898242403213574e-06, "loss": 0.0352, "step": 3043 }, { "epoch": 2.41, "learning_rate": 1.984717987812076e-06, "loss": 0.0333, "step": 3044 }, { "epoch": 2.41, "learning_rate": 1.9796175736093027e-06, "loss": 0.0289, "step": 3045 }, { "epoch": 2.41, "learning_rate": 1.9745230014281622e-06, "loss": 0.0366, "step": 3046 }, { "epoch": 2.41, "learning_rate": 1.9694342749795226e-06, "loss": 0.037, "step": 3047 }, { "epoch": 2.41, "learning_rate": 1.9643513979700035e-06, "loss": 0.0436, "step": 3048 }, { "epoch": 2.41, "learning_rate": 1.9592743741019515e-06, "loss": 0.0382, "step": 3049 }, { "epoch": 2.41, "learning_rate": 1.954203207073456e-06, "loss": 0.0326, "step": 3050 }, { "epoch": 2.41, "learning_rate": 1.9491379005783405e-06, "loss": 0.0431, "step": 3051 }, { "epoch": 2.41, "learning_rate": 1.9440784583061577e-06, "loss": 0.0434, "step": 3052 }, { "epoch": 2.41, "learning_rate": 1.9390248839421854e-06, "loss": 0.0355, "step": 3053 }, { "epoch": 2.41, "learning_rate": 1.933977181167439e-06, "loss": 0.0348, "step": 3054 }, { "epoch": 2.41, "learning_rate": 1.9289353536586466e-06, "loss": 0.0334, "step": 3055 }, { "epoch": 2.41, "learning_rate": 1.9238994050882576e-06, "loss": 0.0373, "step": 3056 }, { "epoch": 2.42, "learning_rate": 1.9188693391244438e-06, "loss": 0.0275, "step": 3057 }, { "epoch": 2.42, "learning_rate": 1.913845159431088e-06, "loss": 0.0378, "step": 3058 }, { "epoch": 2.42, "learning_rate": 1.9088268696677857e-06, "loss": 0.0334, "step": 3059 }, { "epoch": 2.42, "learning_rate": 1.9038144734898478e-06, "loss": 0.036, "step": 3060 }, { "epoch": 2.42, "learning_rate": 1.8988079745482858e-06, "loss": 0.0384, "step": 3061 }, { "epoch": 2.42, "learning_rate": 1.893807376489818e-06, "loss": 0.039, "step": 3062 }, { "epoch": 2.42, "learning_rate": 1.8888126829568642e-06, "loss": 0.0329, "step": 3063 }, { "epoch": 2.42, "learning_rate": 1.8838238975875434e-06, "loss": 0.0397, "step": 3064 }, { "epoch": 2.42, "learning_rate": 1.8788410240156706e-06, "loss": 0.0426, "step": 3065 }, { "epoch": 2.42, "learning_rate": 1.8738640658707585e-06, "loss": 0.036, "step": 3066 }, { "epoch": 2.42, "learning_rate": 1.8688930267780058e-06, "loss": 0.0326, "step": 3067 }, { "epoch": 2.42, "learning_rate": 1.863927910358303e-06, "loss": 0.0343, "step": 3068 }, { "epoch": 2.43, "learning_rate": 1.8589687202282247e-06, "loss": 0.0393, "step": 3069 }, { "epoch": 2.43, "learning_rate": 1.8540154600000294e-06, "loss": 0.0333, "step": 3070 }, { "epoch": 2.43, "learning_rate": 1.8490681332816574e-06, "loss": 0.0395, "step": 3071 }, { "epoch": 2.43, "learning_rate": 1.844126743676722e-06, "loss": 0.0302, "step": 3072 }, { "epoch": 2.43, "learning_rate": 1.8391912947845214e-06, "loss": 0.032, "step": 3073 }, { "epoch": 2.43, "learning_rate": 1.8342617902000203e-06, "loss": 0.0465, "step": 3074 }, { "epoch": 2.43, "learning_rate": 1.8293382335138533e-06, "loss": 0.0386, "step": 3075 }, { "epoch": 2.43, "learning_rate": 1.8244206283123234e-06, "loss": 0.0309, "step": 3076 }, { "epoch": 2.43, "learning_rate": 1.8195089781774001e-06, "loss": 0.031, "step": 3077 }, { "epoch": 2.43, "learning_rate": 1.8146032866867114e-06, "loss": 0.0364, "step": 3078 }, { "epoch": 2.43, "learning_rate": 1.8097035574135512e-06, "loss": 0.0419, "step": 3079 }, { "epoch": 2.43, "learning_rate": 1.8048097939268661e-06, "loss": 0.0411, "step": 3080 }, { "epoch": 2.43, "learning_rate": 1.7999219997912575e-06, "loss": 0.0412, "step": 3081 }, { "epoch": 2.44, "learning_rate": 1.7950401785669812e-06, "loss": 0.0255, "step": 3082 }, { "epoch": 2.44, "learning_rate": 1.7901643338099383e-06, "loss": 0.0303, "step": 3083 }, { "epoch": 2.44, "learning_rate": 1.7852944690716766e-06, "loss": 0.0325, "step": 3084 }, { "epoch": 2.44, "learning_rate": 1.7804305878993966e-06, "loss": 0.0363, "step": 3085 }, { "epoch": 2.44, "learning_rate": 1.7755726938359319e-06, "loss": 0.0403, "step": 3086 }, { "epoch": 2.44, "learning_rate": 1.7707207904197566e-06, "loss": 0.0326, "step": 3087 }, { "epoch": 2.44, "learning_rate": 1.7658748811849824e-06, "loss": 0.0262, "step": 3088 }, { "epoch": 2.44, "learning_rate": 1.761034969661355e-06, "loss": 0.0341, "step": 3089 }, { "epoch": 2.44, "learning_rate": 1.7562010593742496e-06, "loss": 0.0416, "step": 3090 }, { "epoch": 2.44, "learning_rate": 1.75137315384467e-06, "loss": 0.043, "step": 3091 }, { "epoch": 2.44, "learning_rate": 1.7465512565892517e-06, "loss": 0.0294, "step": 3092 }, { "epoch": 2.44, "learning_rate": 1.7417353711202478e-06, "loss": 0.0327, "step": 3093 }, { "epoch": 2.44, "learning_rate": 1.7369255009455355e-06, "loss": 0.0343, "step": 3094 }, { "epoch": 2.45, "learning_rate": 1.7321216495686066e-06, "loss": 0.045, "step": 3095 }, { "epoch": 2.45, "learning_rate": 1.7273238204885734e-06, "loss": 0.0328, "step": 3096 }, { "epoch": 2.45, "learning_rate": 1.7225320172001569e-06, "loss": 0.0474, "step": 3097 }, { "epoch": 2.45, "learning_rate": 1.7177462431936963e-06, "loss": 0.0387, "step": 3098 }, { "epoch": 2.45, "learning_rate": 1.7129665019551333e-06, "loss": 0.042, "step": 3099 }, { "epoch": 2.45, "learning_rate": 1.708192796966015e-06, "loss": 0.028, "step": 3100 }, { "epoch": 2.45, "learning_rate": 1.7034251317034933e-06, "loss": 0.0359, "step": 3101 }, { "epoch": 2.45, "learning_rate": 1.6986635096403213e-06, "loss": 0.0313, "step": 3102 }, { "epoch": 2.45, "learning_rate": 1.6939079342448505e-06, "loss": 0.0364, "step": 3103 }, { "epoch": 2.45, "learning_rate": 1.6891584089810242e-06, "loss": 0.0351, "step": 3104 }, { "epoch": 2.45, "learning_rate": 1.6844149373083852e-06, "loss": 0.0391, "step": 3105 }, { "epoch": 2.45, "learning_rate": 1.679677522682064e-06, "loss": 0.0416, "step": 3106 }, { "epoch": 2.46, "learning_rate": 1.6749461685527756e-06, "loss": 0.0399, "step": 3107 }, { "epoch": 2.46, "learning_rate": 1.670220878366826e-06, "loss": 0.0384, "step": 3108 }, { "epoch": 2.46, "learning_rate": 1.6655016555661009e-06, "loss": 0.0362, "step": 3109 }, { "epoch": 2.46, "learning_rate": 1.6607885035880655e-06, "loss": 0.0381, "step": 3110 }, { "epoch": 2.46, "learning_rate": 1.6560814258657687e-06, "loss": 0.0351, "step": 3111 }, { "epoch": 2.46, "learning_rate": 1.651380425827831e-06, "loss": 0.0454, "step": 3112 }, { "epoch": 2.46, "learning_rate": 1.6466855068984444e-06, "loss": 0.0448, "step": 3113 }, { "epoch": 2.46, "learning_rate": 1.6419966724973734e-06, "loss": 0.0314, "step": 3114 }, { "epoch": 2.46, "learning_rate": 1.6373139260399517e-06, "loss": 0.0435, "step": 3115 }, { "epoch": 2.46, "learning_rate": 1.6326372709370763e-06, "loss": 0.0421, "step": 3116 }, { "epoch": 2.46, "learning_rate": 1.6279667105952057e-06, "loss": 0.0278, "step": 3117 }, { "epoch": 2.46, "learning_rate": 1.6233022484163663e-06, "loss": 0.0294, "step": 3118 }, { "epoch": 2.46, "learning_rate": 1.618643887798137e-06, "loss": 0.0334, "step": 3119 }, { "epoch": 2.47, "learning_rate": 1.6139916321336513e-06, "loss": 0.0312, "step": 3120 }, { "epoch": 2.47, "learning_rate": 1.6093454848115976e-06, "loss": 0.0371, "step": 3121 }, { "epoch": 2.47, "learning_rate": 1.6047054492162173e-06, "loss": 0.0426, "step": 3122 }, { "epoch": 2.47, "learning_rate": 1.6000715287272938e-06, "loss": 0.0351, "step": 3123 }, { "epoch": 2.47, "learning_rate": 1.595443726720165e-06, "loss": 0.0364, "step": 3124 }, { "epoch": 2.47, "learning_rate": 1.5908220465657064e-06, "loss": 0.0371, "step": 3125 }, { "epoch": 2.47, "learning_rate": 1.5862064916303343e-06, "loss": 0.0333, "step": 3126 }, { "epoch": 2.47, "learning_rate": 1.581597065276005e-06, "loss": 0.0404, "step": 3127 }, { "epoch": 2.47, "learning_rate": 1.5769937708602113e-06, "loss": 0.0365, "step": 3128 }, { "epoch": 2.47, "learning_rate": 1.5723966117359745e-06, "loss": 0.0319, "step": 3129 }, { "epoch": 2.47, "learning_rate": 1.5678055912518565e-06, "loss": 0.0337, "step": 3130 }, { "epoch": 2.47, "learning_rate": 1.5632207127519394e-06, "loss": 0.0342, "step": 3131 }, { "epoch": 2.47, "learning_rate": 1.5586419795758356e-06, "loss": 0.0316, "step": 3132 }, { "epoch": 2.48, "learning_rate": 1.5540693950586805e-06, "loss": 0.0358, "step": 3133 }, { "epoch": 2.48, "learning_rate": 1.5495029625311275e-06, "loss": 0.0392, "step": 3134 }, { "epoch": 2.48, "learning_rate": 1.5449426853193549e-06, "loss": 0.0341, "step": 3135 }, { "epoch": 2.48, "learning_rate": 1.5403885667450514e-06, "loss": 0.0353, "step": 3136 }, { "epoch": 2.48, "learning_rate": 1.5358406101254275e-06, "loss": 0.0366, "step": 3137 }, { "epoch": 2.48, "learning_rate": 1.5312988187731969e-06, "loss": 0.0366, "step": 3138 }, { "epoch": 2.48, "learning_rate": 1.5267631959965878e-06, "loss": 0.0345, "step": 3139 }, { "epoch": 2.48, "learning_rate": 1.522233745099334e-06, "loss": 0.0383, "step": 3140 }, { "epoch": 2.48, "learning_rate": 1.5177104693806721e-06, "loss": 0.0416, "step": 3141 }, { "epoch": 2.48, "learning_rate": 1.5131933721353398e-06, "loss": 0.0325, "step": 3142 }, { "epoch": 2.48, "learning_rate": 1.5086824566535828e-06, "loss": 0.0406, "step": 3143 }, { "epoch": 2.48, "learning_rate": 1.5041777262211355e-06, "loss": 0.0374, "step": 3144 }, { "epoch": 2.49, "learning_rate": 1.4996791841192282e-06, "loss": 0.0406, "step": 3145 }, { "epoch": 2.49, "learning_rate": 1.4951868336245844e-06, "loss": 0.0497, "step": 3146 }, { "epoch": 2.49, "learning_rate": 1.4907006780094212e-06, "loss": 0.0413, "step": 3147 }, { "epoch": 2.49, "learning_rate": 1.4862207205414381e-06, "loss": 0.0411, "step": 3148 }, { "epoch": 2.49, "learning_rate": 1.48174696448382e-06, "loss": 0.0425, "step": 3149 }, { "epoch": 2.49, "learning_rate": 1.4772794130952416e-06, "loss": 0.0329, "step": 3150 }, { "epoch": 2.49, "learning_rate": 1.4728180696298521e-06, "loss": 0.0343, "step": 3151 }, { "epoch": 2.49, "learning_rate": 1.46836293733728e-06, "loss": 0.0382, "step": 3152 }, { "epoch": 2.49, "learning_rate": 1.4639140194626289e-06, "loss": 0.0413, "step": 3153 }, { "epoch": 2.49, "learning_rate": 1.4594713192464783e-06, "loss": 0.0391, "step": 3154 }, { "epoch": 2.49, "learning_rate": 1.4550348399248726e-06, "loss": 0.0417, "step": 3155 }, { "epoch": 2.49, "learning_rate": 1.450604584729336e-06, "loss": 0.0348, "step": 3156 }, { "epoch": 2.49, "learning_rate": 1.4461805568868503e-06, "loss": 0.0275, "step": 3157 }, { "epoch": 2.5, "learning_rate": 1.4417627596198625e-06, "loss": 0.0333, "step": 3158 }, { "epoch": 2.5, "learning_rate": 1.4373511961462828e-06, "loss": 0.0339, "step": 3159 }, { "epoch": 2.5, "learning_rate": 1.4329458696794817e-06, "loss": 0.0317, "step": 3160 }, { "epoch": 2.5, "learning_rate": 1.4285467834282829e-06, "loss": 0.0406, "step": 3161 }, { "epoch": 2.5, "learning_rate": 1.4241539405969662e-06, "loss": 0.0333, "step": 3162 }, { "epoch": 2.5, "learning_rate": 1.41976734438527e-06, "loss": 0.0378, "step": 3163 }, { "epoch": 2.5, "learning_rate": 1.415386997988374e-06, "loss": 0.0379, "step": 3164 }, { "epoch": 2.5, "learning_rate": 1.411012904596909e-06, "loss": 0.0402, "step": 3165 }, { "epoch": 2.5, "learning_rate": 1.40664506739695e-06, "loss": 0.0383, "step": 3166 }, { "epoch": 2.5, "learning_rate": 1.402283489570019e-06, "loss": 0.0374, "step": 3167 }, { "epoch": 2.5, "learning_rate": 1.3979281742930706e-06, "loss": 0.031, "step": 3168 }, { "epoch": 2.5, "learning_rate": 1.393579124738509e-06, "loss": 0.0384, "step": 3169 }, { "epoch": 2.5, "learning_rate": 1.3892363440741652e-06, "loss": 0.0345, "step": 3170 }, { "epoch": 2.51, "learning_rate": 1.3848998354633082e-06, "loss": 0.0378, "step": 3171 }, { "epoch": 2.51, "learning_rate": 1.3805696020646365e-06, "loss": 0.0328, "step": 3172 }, { "epoch": 2.51, "learning_rate": 1.3762456470322782e-06, "loss": 0.0334, "step": 3173 }, { "epoch": 2.51, "learning_rate": 1.3719279735157875e-06, "loss": 0.0379, "step": 3174 }, { "epoch": 2.51, "learning_rate": 1.3676165846601485e-06, "loss": 0.0366, "step": 3175 }, { "epoch": 2.51, "learning_rate": 1.3633114836057615e-06, "loss": 0.0395, "step": 3176 }, { "epoch": 2.51, "learning_rate": 1.359012673488449e-06, "loss": 0.0302, "step": 3177 }, { "epoch": 2.51, "learning_rate": 1.3547201574394497e-06, "loss": 0.0417, "step": 3178 }, { "epoch": 2.51, "learning_rate": 1.3504339385854193e-06, "loss": 0.0381, "step": 3179 }, { "epoch": 2.51, "learning_rate": 1.346154020048428e-06, "loss": 0.0396, "step": 3180 }, { "epoch": 2.51, "learning_rate": 1.34188040494595e-06, "loss": 0.0415, "step": 3181 }, { "epoch": 2.51, "learning_rate": 1.3376130963908796e-06, "loss": 0.0323, "step": 3182 }, { "epoch": 2.52, "learning_rate": 1.3333520974915093e-06, "loss": 0.0366, "step": 3183 }, { "epoch": 2.52, "learning_rate": 1.3290974113515364e-06, "loss": 0.0292, "step": 3184 }, { "epoch": 2.52, "learning_rate": 1.324849041070061e-06, "loss": 0.0455, "step": 3185 }, { "epoch": 2.52, "learning_rate": 1.320606989741583e-06, "loss": 0.0432, "step": 3186 }, { "epoch": 2.52, "learning_rate": 1.3163712604559986e-06, "loss": 0.0338, "step": 3187 }, { "epoch": 2.52, "learning_rate": 1.312141856298602e-06, "loss": 0.0307, "step": 3188 }, { "epoch": 2.52, "learning_rate": 1.307918780350077e-06, "loss": 0.0343, "step": 3189 }, { "epoch": 2.52, "learning_rate": 1.3037020356864982e-06, "loss": 0.0336, "step": 3190 }, { "epoch": 2.52, "learning_rate": 1.299491625379331e-06, "loss": 0.0313, "step": 3191 }, { "epoch": 2.52, "learning_rate": 1.2952875524954233e-06, "loss": 0.0421, "step": 3192 }, { "epoch": 2.52, "learning_rate": 1.2910898200970102e-06, "loss": 0.0398, "step": 3193 }, { "epoch": 2.52, "learning_rate": 1.2868984312417043e-06, "loss": 0.0345, "step": 3194 }, { "epoch": 2.52, "learning_rate": 1.2827133889825039e-06, "loss": 0.0393, "step": 3195 }, { "epoch": 2.53, "learning_rate": 1.2785346963677792e-06, "loss": 0.0384, "step": 3196 }, { "epoch": 2.53, "learning_rate": 1.2743623564412767e-06, "loss": 0.0402, "step": 3197 }, { "epoch": 2.53, "learning_rate": 1.2701963722421162e-06, "loss": 0.0438, "step": 3198 }, { "epoch": 2.53, "learning_rate": 1.2660367468047873e-06, "loss": 0.0442, "step": 3199 }, { "epoch": 2.53, "learning_rate": 1.2618834831591465e-06, "loss": 0.0366, "step": 3200 }, { "epoch": 2.53, "learning_rate": 1.2577365843304212e-06, "loss": 0.0366, "step": 3201 }, { "epoch": 2.53, "learning_rate": 1.2535960533392e-06, "loss": 0.0305, "step": 3202 }, { "epoch": 2.53, "learning_rate": 1.2494618932014302e-06, "loss": 0.0426, "step": 3203 }, { "epoch": 2.53, "learning_rate": 1.245334106928422e-06, "loss": 0.0431, "step": 3204 }, { "epoch": 2.53, "learning_rate": 1.2412126975268423e-06, "loss": 0.047, "step": 3205 }, { "epoch": 2.53, "learning_rate": 1.2370976679987123e-06, "loss": 0.0351, "step": 3206 }, { "epoch": 2.53, "learning_rate": 1.2329890213414063e-06, "loss": 0.0451, "step": 3207 }, { "epoch": 2.53, "learning_rate": 1.2288867605476528e-06, "loss": 0.0327, "step": 3208 }, { "epoch": 2.54, "learning_rate": 1.2247908886055226e-06, "loss": 0.0319, "step": 3209 }, { "epoch": 2.54, "learning_rate": 1.220701408498438e-06, "loss": 0.0316, "step": 3210 }, { "epoch": 2.54, "learning_rate": 1.2166183232051632e-06, "loss": 0.0344, "step": 3211 }, { "epoch": 2.54, "learning_rate": 1.2125416356998044e-06, "loss": 0.0332, "step": 3212 }, { "epoch": 2.54, "learning_rate": 1.2084713489518063e-06, "loss": 0.031, "step": 3213 }, { "epoch": 2.54, "learning_rate": 1.2044074659259564e-06, "loss": 0.0344, "step": 3214 }, { "epoch": 2.54, "learning_rate": 1.2003499895823734e-06, "loss": 0.0311, "step": 3215 }, { "epoch": 2.54, "learning_rate": 1.19629892287651e-06, "loss": 0.0423, "step": 3216 }, { "epoch": 2.54, "learning_rate": 1.1922542687591509e-06, "loss": 0.0344, "step": 3217 }, { "epoch": 2.54, "learning_rate": 1.188216030176409e-06, "loss": 0.0506, "step": 3218 }, { "epoch": 2.54, "learning_rate": 1.1841842100697253e-06, "loss": 0.036, "step": 3219 }, { "epoch": 2.54, "learning_rate": 1.1801588113758634e-06, "loss": 0.0357, "step": 3220 }, { "epoch": 2.55, "learning_rate": 1.176139837026915e-06, "loss": 0.0418, "step": 3221 }, { "epoch": 2.55, "learning_rate": 1.1721272899502856e-06, "loss": 0.0368, "step": 3222 }, { "epoch": 2.55, "learning_rate": 1.1681211730687026e-06, "loss": 0.0363, "step": 3223 }, { "epoch": 2.55, "learning_rate": 1.1641214893002096e-06, "loss": 0.0341, "step": 3224 }, { "epoch": 2.55, "learning_rate": 1.1601282415581627e-06, "loss": 0.0343, "step": 3225 }, { "epoch": 2.55, "learning_rate": 1.15614143275123e-06, "loss": 0.0337, "step": 3226 }, { "epoch": 2.55, "learning_rate": 1.1521610657833948e-06, "loss": 0.0249, "step": 3227 }, { "epoch": 2.55, "learning_rate": 1.1481871435539415e-06, "loss": 0.0431, "step": 3228 }, { "epoch": 2.55, "learning_rate": 1.1442196689574625e-06, "loss": 0.0327, "step": 3229 }, { "epoch": 2.55, "learning_rate": 1.1402586448838548e-06, "loss": 0.031, "step": 3230 }, { "epoch": 2.55, "learning_rate": 1.1363040742183162e-06, "loss": 0.0385, "step": 3231 }, { "epoch": 2.55, "learning_rate": 1.1323559598413402e-06, "loss": 0.0387, "step": 3232 }, { "epoch": 2.55, "learning_rate": 1.1284143046287255e-06, "loss": 0.031, "step": 3233 }, { "epoch": 2.56, "learning_rate": 1.1244791114515608e-06, "loss": 0.0374, "step": 3234 }, { "epoch": 2.56, "learning_rate": 1.1205503831762276e-06, "loss": 0.0357, "step": 3235 }, { "epoch": 2.56, "learning_rate": 1.1166281226643982e-06, "loss": 0.0302, "step": 3236 }, { "epoch": 2.56, "learning_rate": 1.1127123327730383e-06, "loss": 0.0357, "step": 3237 }, { "epoch": 2.56, "learning_rate": 1.1088030163543929e-06, "loss": 0.0291, "step": 3238 }, { "epoch": 2.56, "learning_rate": 1.1049001762559975e-06, "loss": 0.029, "step": 3239 }, { "epoch": 2.56, "learning_rate": 1.1010038153206703e-06, "loss": 0.0314, "step": 3240 }, { "epoch": 2.56, "learning_rate": 1.097113936386508e-06, "loss": 0.0368, "step": 3241 }, { "epoch": 2.56, "learning_rate": 1.093230542286886e-06, "loss": 0.0449, "step": 3242 }, { "epoch": 2.56, "learning_rate": 1.0893536358504553e-06, "loss": 0.0351, "step": 3243 }, { "epoch": 2.56, "learning_rate": 1.0854832199011444e-06, "loss": 0.0359, "step": 3244 }, { "epoch": 2.56, "learning_rate": 1.0816192972581485e-06, "loss": 0.0395, "step": 3245 }, { "epoch": 2.56, "learning_rate": 1.0777618707359427e-06, "loss": 0.0378, "step": 3246 }, { "epoch": 2.57, "learning_rate": 1.0739109431442607e-06, "loss": 0.0412, "step": 3247 }, { "epoch": 2.57, "learning_rate": 1.0700665172881063e-06, "loss": 0.0349, "step": 3248 }, { "epoch": 2.57, "learning_rate": 1.0662285959677499e-06, "loss": 0.0421, "step": 3249 }, { "epoch": 2.57, "learning_rate": 1.0623971819787182e-06, "loss": 0.0272, "step": 3250 }, { "epoch": 2.57, "learning_rate": 1.0585722781118024e-06, "loss": 0.0319, "step": 3251 }, { "epoch": 2.57, "learning_rate": 1.0547538871530482e-06, "loss": 0.0354, "step": 3252 }, { "epoch": 2.57, "learning_rate": 1.0509420118837654e-06, "loss": 0.0329, "step": 3253 }, { "epoch": 2.57, "learning_rate": 1.0471366550805083e-06, "loss": 0.0338, "step": 3254 }, { "epoch": 2.57, "learning_rate": 1.0433378195150889e-06, "loss": 0.0287, "step": 3255 }, { "epoch": 2.57, "learning_rate": 1.0395455079545646e-06, "loss": 0.0424, "step": 3256 }, { "epoch": 2.57, "learning_rate": 1.0357597231612448e-06, "loss": 0.0296, "step": 3257 }, { "epoch": 2.57, "learning_rate": 1.0319804678926825e-06, "loss": 0.0397, "step": 3258 }, { "epoch": 2.58, "learning_rate": 1.028207744901677e-06, "loss": 0.0435, "step": 3259 }, { "epoch": 2.58, "learning_rate": 1.0244415569362675e-06, "loss": 0.0286, "step": 3260 }, { "epoch": 2.58, "learning_rate": 1.0206819067397345e-06, "loss": 0.045, "step": 3261 }, { "epoch": 2.58, "learning_rate": 1.0169287970505926e-06, "loss": 0.0328, "step": 3262 }, { "epoch": 2.58, "learning_rate": 1.0131822306025985e-06, "loss": 0.0323, "step": 3263 }, { "epoch": 2.58, "learning_rate": 1.009442210124737e-06, "loss": 0.0407, "step": 3264 }, { "epoch": 2.58, "learning_rate": 1.0057087383412278e-06, "loss": 0.0354, "step": 3265 }, { "epoch": 2.58, "learning_rate": 1.0019818179715223e-06, "loss": 0.0406, "step": 3266 }, { "epoch": 2.58, "learning_rate": 9.982614517302958e-07, "loss": 0.0295, "step": 3267 }, { "epoch": 2.58, "learning_rate": 9.945476423274526e-07, "loss": 0.037, "step": 3268 }, { "epoch": 2.58, "learning_rate": 9.908403924681188e-07, "loss": 0.0314, "step": 3269 }, { "epoch": 2.58, "learning_rate": 9.871397048526431e-07, "loss": 0.0332, "step": 3270 }, { "epoch": 2.58, "learning_rate": 9.834455821765953e-07, "loss": 0.0399, "step": 3271 }, { "epoch": 2.59, "learning_rate": 9.79758027130766e-07, "loss": 0.0304, "step": 3272 }, { "epoch": 2.59, "learning_rate": 9.760770424011557e-07, "loss": 0.0309, "step": 3273 }, { "epoch": 2.59, "learning_rate": 9.724026306689826e-07, "loss": 0.0358, "step": 3274 }, { "epoch": 2.59, "learning_rate": 9.687347946106772e-07, "loss": 0.0291, "step": 3275 }, { "epoch": 2.59, "learning_rate": 9.650735368978793e-07, "loss": 0.0326, "step": 3276 }, { "epoch": 2.59, "learning_rate": 9.614188601974373e-07, "loss": 0.0377, "step": 3277 }, { "epoch": 2.59, "learning_rate": 9.577707671714076e-07, "loss": 0.0345, "step": 3278 }, { "epoch": 2.59, "learning_rate": 9.541292604770502e-07, "loss": 0.0376, "step": 3279 }, { "epoch": 2.59, "learning_rate": 9.504943427668267e-07, "loss": 0.0368, "step": 3280 }, { "epoch": 2.59, "learning_rate": 9.468660166883992e-07, "loss": 0.0254, "step": 3281 }, { "epoch": 2.59, "learning_rate": 9.432442848846291e-07, "loss": 0.0388, "step": 3282 }, { "epoch": 2.59, "learning_rate": 9.396291499935761e-07, "loss": 0.0311, "step": 3283 }, { "epoch": 2.6, "learning_rate": 9.360206146484896e-07, "loss": 0.0337, "step": 3284 }, { "epoch": 2.6, "learning_rate": 9.324186814778202e-07, "loss": 0.0368, "step": 3285 }, { "epoch": 2.6, "learning_rate": 9.288233531052015e-07, "loss": 0.0314, "step": 3286 }, { "epoch": 2.6, "learning_rate": 9.25234632149462e-07, "loss": 0.0301, "step": 3287 }, { "epoch": 2.6, "learning_rate": 9.216525212246131e-07, "loss": 0.0304, "step": 3288 }, { "epoch": 2.6, "learning_rate": 9.180770229398539e-07, "loss": 0.0349, "step": 3289 }, { "epoch": 2.6, "learning_rate": 9.145081398995647e-07, "loss": 0.0254, "step": 3290 }, { "epoch": 2.6, "learning_rate": 9.109458747033106e-07, "loss": 0.0367, "step": 3291 }, { "epoch": 2.6, "learning_rate": 9.07390229945836e-07, "loss": 0.0364, "step": 3292 }, { "epoch": 2.6, "learning_rate": 9.038412082170578e-07, "loss": 0.03, "step": 3293 }, { "epoch": 2.6, "learning_rate": 9.00298812102075e-07, "loss": 0.0427, "step": 3294 }, { "epoch": 2.6, "learning_rate": 8.967630441811581e-07, "loss": 0.0346, "step": 3295 }, { "epoch": 2.6, "learning_rate": 8.932339070297469e-07, "loss": 0.031, "step": 3296 }, { "epoch": 2.61, "learning_rate": 8.897114032184539e-07, "loss": 0.0357, "step": 3297 }, { "epoch": 2.61, "learning_rate": 8.861955353130646e-07, "loss": 0.0375, "step": 3298 }, { "epoch": 2.61, "learning_rate": 8.826863058745216e-07, "loss": 0.0329, "step": 3299 }, { "epoch": 2.61, "learning_rate": 8.791837174589401e-07, "loss": 0.0389, "step": 3300 }, { "epoch": 2.61, "learning_rate": 8.756877726175917e-07, "loss": 0.0369, "step": 3301 }, { "epoch": 2.61, "learning_rate": 8.721984738969124e-07, "loss": 0.0352, "step": 3302 }, { "epoch": 2.61, "learning_rate": 8.687158238384963e-07, "loss": 0.036, "step": 3303 }, { "epoch": 2.61, "learning_rate": 8.652398249790972e-07, "loss": 0.0354, "step": 3304 }, { "epoch": 2.61, "learning_rate": 8.617704798506199e-07, "loss": 0.0335, "step": 3305 }, { "epoch": 2.61, "learning_rate": 8.583077909801252e-07, "loss": 0.0416, "step": 3306 }, { "epoch": 2.61, "learning_rate": 8.548517608898244e-07, "loss": 0.0365, "step": 3307 }, { "epoch": 2.61, "learning_rate": 8.514023920970793e-07, "loss": 0.0408, "step": 3308 }, { "epoch": 2.61, "learning_rate": 8.47959687114398e-07, "loss": 0.0482, "step": 3309 }, { "epoch": 2.62, "learning_rate": 8.44523648449439e-07, "loss": 0.0349, "step": 3310 }, { "epoch": 2.62, "learning_rate": 8.410942786050014e-07, "loss": 0.0269, "step": 3311 }, { "epoch": 2.62, "learning_rate": 8.37671580079027e-07, "loss": 0.0368, "step": 3312 }, { "epoch": 2.62, "learning_rate": 8.342555553645982e-07, "loss": 0.0332, "step": 3313 }, { "epoch": 2.62, "learning_rate": 8.308462069499368e-07, "loss": 0.0332, "step": 3314 }, { "epoch": 2.62, "learning_rate": 8.274435373184009e-07, "loss": 0.0298, "step": 3315 }, { "epoch": 2.62, "learning_rate": 8.240475489484878e-07, "loss": 0.0449, "step": 3316 }, { "epoch": 2.62, "learning_rate": 8.206582443138222e-07, "loss": 0.0329, "step": 3317 }, { "epoch": 2.62, "learning_rate": 8.172756258831638e-07, "loss": 0.0365, "step": 3318 }, { "epoch": 2.62, "learning_rate": 8.138996961204015e-07, "loss": 0.0376, "step": 3319 }, { "epoch": 2.62, "learning_rate": 8.105304574845508e-07, "loss": 0.0345, "step": 3320 }, { "epoch": 2.62, "learning_rate": 8.071679124297537e-07, "loss": 0.0301, "step": 3321 }, { "epoch": 2.63, "learning_rate": 8.038120634052826e-07, "loss": 0.035, "step": 3322 }, { "epoch": 2.63, "learning_rate": 8.00462912855523e-07, "loss": 0.0331, "step": 3323 }, { "epoch": 2.63, "learning_rate": 7.971204632199869e-07, "loss": 0.0382, "step": 3324 }, { "epoch": 2.63, "learning_rate": 7.937847169333058e-07, "loss": 0.0298, "step": 3325 }, { "epoch": 2.63, "learning_rate": 7.904556764252247e-07, "loss": 0.0243, "step": 3326 }, { "epoch": 2.63, "learning_rate": 7.871333441206053e-07, "loss": 0.0297, "step": 3327 }, { "epoch": 2.63, "learning_rate": 7.838177224394272e-07, "loss": 0.0376, "step": 3328 }, { "epoch": 2.63, "learning_rate": 7.805088137967787e-07, "loss": 0.0442, "step": 3329 }, { "epoch": 2.63, "learning_rate": 7.772066206028572e-07, "loss": 0.0315, "step": 3330 }, { "epoch": 2.63, "learning_rate": 7.7391114526297e-07, "loss": 0.0369, "step": 3331 }, { "epoch": 2.63, "learning_rate": 7.7062239017753e-07, "loss": 0.0341, "step": 3332 }, { "epoch": 2.63, "learning_rate": 7.673403577420591e-07, "loss": 0.0238, "step": 3333 }, { "epoch": 2.63, "learning_rate": 7.640650503471792e-07, "loss": 0.0324, "step": 3334 }, { "epoch": 2.64, "learning_rate": 7.60796470378612e-07, "loss": 0.0405, "step": 3335 }, { "epoch": 2.64, "learning_rate": 7.575346202171819e-07, "loss": 0.0291, "step": 3336 }, { "epoch": 2.64, "learning_rate": 7.542795022388094e-07, "loss": 0.0388, "step": 3337 }, { "epoch": 2.64, "learning_rate": 7.510311188145137e-07, "loss": 0.0396, "step": 3338 }, { "epoch": 2.64, "learning_rate": 7.477894723104073e-07, "loss": 0.031, "step": 3339 }, { "epoch": 2.64, "learning_rate": 7.445545650876951e-07, "loss": 0.0292, "step": 3340 }, { "epoch": 2.64, "learning_rate": 7.413263995026731e-07, "loss": 0.0327, "step": 3341 }, { "epoch": 2.64, "learning_rate": 7.381049779067273e-07, "loss": 0.0393, "step": 3342 }, { "epoch": 2.64, "learning_rate": 7.348903026463317e-07, "loss": 0.0392, "step": 3343 }, { "epoch": 2.64, "learning_rate": 7.316823760630431e-07, "loss": 0.041, "step": 3344 }, { "epoch": 2.64, "learning_rate": 7.284812004935083e-07, "loss": 0.0415, "step": 3345 }, { "epoch": 2.64, "learning_rate": 7.25286778269454e-07, "loss": 0.0307, "step": 3346 }, { "epoch": 2.64, "learning_rate": 7.220991117176856e-07, "loss": 0.0322, "step": 3347 }, { "epoch": 2.65, "learning_rate": 7.189182031600906e-07, "loss": 0.0368, "step": 3348 }, { "epoch": 2.65, "learning_rate": 7.157440549136319e-07, "loss": 0.0314, "step": 3349 }, { "epoch": 2.65, "learning_rate": 7.12576669290348e-07, "loss": 0.0439, "step": 3350 }, { "epoch": 2.65, "learning_rate": 7.094160485973567e-07, "loss": 0.0432, "step": 3351 }, { "epoch": 2.65, "learning_rate": 7.062621951368409e-07, "loss": 0.0388, "step": 3352 }, { "epoch": 2.65, "learning_rate": 7.031151112060597e-07, "loss": 0.0344, "step": 3353 }, { "epoch": 2.65, "learning_rate": 6.999747990973382e-07, "loss": 0.0279, "step": 3354 }, { "epoch": 2.65, "learning_rate": 6.968412610980691e-07, "loss": 0.0345, "step": 3355 }, { "epoch": 2.65, "learning_rate": 6.937144994907119e-07, "loss": 0.0394, "step": 3356 }, { "epoch": 2.65, "learning_rate": 6.905945165527928e-07, "loss": 0.0411, "step": 3357 }, { "epoch": 2.65, "learning_rate": 6.87481314556896e-07, "loss": 0.0358, "step": 3358 }, { "epoch": 2.65, "learning_rate": 6.843748957706675e-07, "loss": 0.0285, "step": 3359 }, { "epoch": 2.66, "learning_rate": 6.812752624568131e-07, "loss": 0.0349, "step": 3360 }, { "epoch": 2.66, "learning_rate": 6.781824168730966e-07, "loss": 0.0282, "step": 3361 }, { "epoch": 2.66, "learning_rate": 6.750963612723361e-07, "loss": 0.0368, "step": 3362 }, { "epoch": 2.66, "learning_rate": 6.720170979024065e-07, "loss": 0.0402, "step": 3363 }, { "epoch": 2.66, "learning_rate": 6.68944629006234e-07, "loss": 0.0361, "step": 3364 }, { "epoch": 2.66, "learning_rate": 6.658789568217939e-07, "loss": 0.0364, "step": 3365 }, { "epoch": 2.66, "learning_rate": 6.628200835821119e-07, "loss": 0.0341, "step": 3366 }, { "epoch": 2.66, "learning_rate": 6.597680115152638e-07, "loss": 0.0444, "step": 3367 }, { "epoch": 2.66, "learning_rate": 6.567227428443668e-07, "loss": 0.0368, "step": 3368 }, { "epoch": 2.66, "learning_rate": 6.536842797875876e-07, "loss": 0.0276, "step": 3369 }, { "epoch": 2.66, "learning_rate": 6.506526245581312e-07, "loss": 0.0243, "step": 3370 }, { "epoch": 2.66, "learning_rate": 6.476277793642482e-07, "loss": 0.0322, "step": 3371 }, { "epoch": 2.66, "learning_rate": 6.446097464092249e-07, "loss": 0.0321, "step": 3372 }, { "epoch": 2.67, "learning_rate": 6.415985278913872e-07, "loss": 0.0325, "step": 3373 }, { "epoch": 2.67, "learning_rate": 6.385941260040962e-07, "loss": 0.0294, "step": 3374 }, { "epoch": 2.67, "learning_rate": 6.355965429357513e-07, "loss": 0.0361, "step": 3375 }, { "epoch": 2.67, "learning_rate": 6.326057808697816e-07, "loss": 0.0333, "step": 3376 }, { "epoch": 2.67, "learning_rate": 6.296218419846489e-07, "loss": 0.0332, "step": 3377 }, { "epoch": 2.67, "learning_rate": 6.266447284538446e-07, "loss": 0.036, "step": 3378 }, { "epoch": 2.67, "learning_rate": 6.236744424458896e-07, "loss": 0.0328, "step": 3379 }, { "epoch": 2.67, "learning_rate": 6.207109861243287e-07, "loss": 0.0377, "step": 3380 }, { "epoch": 2.67, "learning_rate": 6.177543616477377e-07, "loss": 0.0354, "step": 3381 }, { "epoch": 2.67, "learning_rate": 6.148045711697126e-07, "loss": 0.0298, "step": 3382 }, { "epoch": 2.67, "learning_rate": 6.118616168388692e-07, "loss": 0.0484, "step": 3383 }, { "epoch": 2.67, "learning_rate": 6.08925500798847e-07, "loss": 0.0392, "step": 3384 }, { "epoch": 2.67, "learning_rate": 6.059962251883045e-07, "loss": 0.0294, "step": 3385 }, { "epoch": 2.68, "learning_rate": 6.030737921409169e-07, "loss": 0.0302, "step": 3386 }, { "epoch": 2.68, "learning_rate": 6.001582037853726e-07, "loss": 0.044, "step": 3387 }, { "epoch": 2.68, "learning_rate": 5.972494622453817e-07, "loss": 0.0312, "step": 3388 }, { "epoch": 2.68, "learning_rate": 5.943475696396606e-07, "loss": 0.0352, "step": 3389 }, { "epoch": 2.68, "learning_rate": 5.914525280819383e-07, "loss": 0.0347, "step": 3390 }, { "epoch": 2.68, "learning_rate": 5.885643396809549e-07, "loss": 0.0325, "step": 3391 }, { "epoch": 2.68, "learning_rate": 5.856830065404562e-07, "loss": 0.0372, "step": 3392 }, { "epoch": 2.68, "learning_rate": 5.828085307591969e-07, "loss": 0.0406, "step": 3393 }, { "epoch": 2.68, "learning_rate": 5.799409144309376e-07, "loss": 0.0318, "step": 3394 }, { "epoch": 2.68, "learning_rate": 5.770801596444398e-07, "loss": 0.0338, "step": 3395 }, { "epoch": 2.68, "learning_rate": 5.742262684834698e-07, "loss": 0.0439, "step": 3396 }, { "epoch": 2.68, "learning_rate": 5.713792430267917e-07, "loss": 0.0401, "step": 3397 }, { "epoch": 2.69, "learning_rate": 5.685390853481709e-07, "loss": 0.0434, "step": 3398 }, { "epoch": 2.69, "learning_rate": 5.657057975163682e-07, "loss": 0.0333, "step": 3399 }, { "epoch": 2.69, "learning_rate": 5.628793815951417e-07, "loss": 0.0359, "step": 3400 }, { "epoch": 2.69, "learning_rate": 5.600598396432456e-07, "loss": 0.0335, "step": 3401 }, { "epoch": 2.69, "learning_rate": 5.572471737144247e-07, "loss": 0.034, "step": 3402 }, { "epoch": 2.69, "learning_rate": 5.54441385857415e-07, "loss": 0.04, "step": 3403 }, { "epoch": 2.69, "learning_rate": 5.516424781159458e-07, "loss": 0.0346, "step": 3404 }, { "epoch": 2.69, "learning_rate": 5.488504525287319e-07, "loss": 0.0358, "step": 3405 }, { "epoch": 2.69, "learning_rate": 5.460653111294733e-07, "loss": 0.035, "step": 3406 }, { "epoch": 2.69, "learning_rate": 5.432870559468628e-07, "loss": 0.0374, "step": 3407 }, { "epoch": 2.69, "learning_rate": 5.405156890045704e-07, "loss": 0.034, "step": 3408 }, { "epoch": 2.69, "learning_rate": 5.377512123212525e-07, "loss": 0.0342, "step": 3409 }, { "epoch": 2.69, "learning_rate": 5.349936279105439e-07, "loss": 0.0298, "step": 3410 }, { "epoch": 2.7, "learning_rate": 5.322429377810612e-07, "loss": 0.0261, "step": 3411 }, { "epoch": 2.7, "learning_rate": 5.294991439363983e-07, "loss": 0.036, "step": 3412 }, { "epoch": 2.7, "learning_rate": 5.267622483751245e-07, "loss": 0.0343, "step": 3413 }, { "epoch": 2.7, "learning_rate": 5.240322530907893e-07, "loss": 0.0357, "step": 3414 }, { "epoch": 2.7, "learning_rate": 5.213091600719122e-07, "loss": 0.0314, "step": 3415 }, { "epoch": 2.7, "learning_rate": 5.185929713019833e-07, "loss": 0.0311, "step": 3416 }, { "epoch": 2.7, "learning_rate": 5.158836887594687e-07, "loss": 0.0332, "step": 3417 }, { "epoch": 2.7, "learning_rate": 5.131813144177988e-07, "loss": 0.0308, "step": 3418 }, { "epoch": 2.7, "learning_rate": 5.104858502453758e-07, "loss": 0.0378, "step": 3419 }, { "epoch": 2.7, "learning_rate": 5.07797298205569e-07, "loss": 0.0316, "step": 3420 }, { "epoch": 2.7, "learning_rate": 5.051156602567097e-07, "loss": 0.032, "step": 3421 }, { "epoch": 2.7, "learning_rate": 5.024409383520946e-07, "loss": 0.0404, "step": 3422 }, { "epoch": 2.7, "learning_rate": 4.997731344399837e-07, "loss": 0.0337, "step": 3423 }, { "epoch": 2.71, "learning_rate": 4.971122504635961e-07, "loss": 0.031, "step": 3424 }, { "epoch": 2.71, "learning_rate": 4.944582883611115e-07, "loss": 0.0407, "step": 3425 }, { "epoch": 2.71, "learning_rate": 4.91811250065668e-07, "loss": 0.0275, "step": 3426 }, { "epoch": 2.71, "learning_rate": 4.89171137505361e-07, "loss": 0.034, "step": 3427 }, { "epoch": 2.71, "learning_rate": 4.865379526032388e-07, "loss": 0.0385, "step": 3428 }, { "epoch": 2.71, "learning_rate": 4.839116972773061e-07, "loss": 0.0311, "step": 3429 }, { "epoch": 2.71, "learning_rate": 4.812923734405184e-07, "loss": 0.0439, "step": 3430 }, { "epoch": 2.71, "learning_rate": 4.786799830007838e-07, "loss": 0.0392, "step": 3431 }, { "epoch": 2.71, "learning_rate": 4.7607452786095686e-07, "loss": 0.0357, "step": 3432 }, { "epoch": 2.71, "learning_rate": 4.7347600991884737e-07, "loss": 0.0376, "step": 3433 }, { "epoch": 2.71, "learning_rate": 4.708844310672045e-07, "loss": 0.036, "step": 3434 }, { "epoch": 2.71, "learning_rate": 4.682997931937283e-07, "loss": 0.0335, "step": 3435 }, { "epoch": 2.72, "learning_rate": 4.657220981810584e-07, "loss": 0.0353, "step": 3436 }, { "epoch": 2.72, "learning_rate": 4.6315134790678173e-07, "loss": 0.0302, "step": 3437 }, { "epoch": 2.72, "learning_rate": 4.605875442434238e-07, "loss": 0.0352, "step": 3438 }, { "epoch": 2.72, "learning_rate": 4.5803068905845183e-07, "loss": 0.0297, "step": 3439 }, { "epoch": 2.72, "learning_rate": 4.554807842142717e-07, "loss": 0.0396, "step": 3440 }, { "epoch": 2.72, "learning_rate": 4.5293783156822533e-07, "loss": 0.0344, "step": 3441 }, { "epoch": 2.72, "learning_rate": 4.5040183297259233e-07, "loss": 0.03, "step": 3442 }, { "epoch": 2.72, "learning_rate": 4.4787279027458496e-07, "loss": 0.0385, "step": 3443 }, { "epoch": 2.72, "learning_rate": 4.4535070531635195e-07, "loss": 0.0314, "step": 3444 }, { "epoch": 2.72, "learning_rate": 4.4283557993496817e-07, "loss": 0.0365, "step": 3445 }, { "epoch": 2.72, "learning_rate": 4.403274159624482e-07, "loss": 0.039, "step": 3446 }, { "epoch": 2.72, "learning_rate": 4.378262152257273e-07, "loss": 0.0345, "step": 3447 }, { "epoch": 2.72, "learning_rate": 4.3533197954667485e-07, "loss": 0.0381, "step": 3448 }, { "epoch": 2.73, "learning_rate": 4.32844710742083e-07, "loss": 0.0407, "step": 3449 }, { "epoch": 2.73, "learning_rate": 4.303644106236704e-07, "loss": 0.0395, "step": 3450 }, { "epoch": 2.73, "learning_rate": 4.278910809980785e-07, "loss": 0.0343, "step": 3451 }, { "epoch": 2.73, "learning_rate": 4.254247236668763e-07, "loss": 0.036, "step": 3452 }, { "epoch": 2.73, "learning_rate": 4.2296534042654993e-07, "loss": 0.0352, "step": 3453 }, { "epoch": 2.73, "learning_rate": 4.205129330685054e-07, "loss": 0.0319, "step": 3454 }, { "epoch": 2.73, "learning_rate": 4.180675033790682e-07, "loss": 0.0404, "step": 3455 }, { "epoch": 2.73, "learning_rate": 4.1562905313948354e-07, "loss": 0.0398, "step": 3456 }, { "epoch": 2.73, "learning_rate": 4.131975841259095e-07, "loss": 0.0357, "step": 3457 }, { "epoch": 2.73, "learning_rate": 4.107730981094193e-07, "loss": 0.03, "step": 3458 }, { "epoch": 2.73, "learning_rate": 4.083555968560049e-07, "loss": 0.0375, "step": 3459 }, { "epoch": 2.73, "learning_rate": 4.0594508212656425e-07, "loss": 0.0394, "step": 3460 }, { "epoch": 2.73, "learning_rate": 4.0354155567690845e-07, "loss": 0.0286, "step": 3461 }, { "epoch": 2.74, "learning_rate": 4.0114501925775927e-07, "loss": 0.0267, "step": 3462 }, { "epoch": 2.74, "learning_rate": 3.987554746147471e-07, "loss": 0.0338, "step": 3463 }, { "epoch": 2.74, "learning_rate": 3.9637292348840725e-07, "loss": 0.034, "step": 3464 }, { "epoch": 2.74, "learning_rate": 3.9399736761418395e-07, "loss": 0.0324, "step": 3465 }, { "epoch": 2.74, "learning_rate": 3.9162880872242406e-07, "loss": 0.0309, "step": 3466 }, { "epoch": 2.74, "learning_rate": 3.892672485383797e-07, "loss": 0.0425, "step": 3467 }, { "epoch": 2.74, "learning_rate": 3.8691268878220165e-07, "loss": 0.0291, "step": 3468 }, { "epoch": 2.74, "learning_rate": 3.845651311689458e-07, "loss": 0.0425, "step": 3469 }, { "epoch": 2.74, "learning_rate": 3.8222457740856536e-07, "loss": 0.0405, "step": 3470 }, { "epoch": 2.74, "learning_rate": 3.7989102920591103e-07, "loss": 0.0324, "step": 3471 }, { "epoch": 2.74, "learning_rate": 3.7756448826073435e-07, "loss": 0.0465, "step": 3472 }, { "epoch": 2.74, "learning_rate": 3.7524495626767966e-07, "loss": 0.0337, "step": 3473 }, { "epoch": 2.75, "learning_rate": 3.729324349162866e-07, "loss": 0.041, "step": 3474 }, { "epoch": 2.75, "learning_rate": 3.706269258909889e-07, "loss": 0.0347, "step": 3475 }, { "epoch": 2.75, "learning_rate": 3.683284308711121e-07, "loss": 0.0358, "step": 3476 }, { "epoch": 2.75, "learning_rate": 3.660369515308715e-07, "loss": 0.0378, "step": 3477 }, { "epoch": 2.75, "learning_rate": 3.6375248953937516e-07, "loss": 0.0294, "step": 3478 }, { "epoch": 2.75, "learning_rate": 3.614750465606176e-07, "loss": 0.0268, "step": 3479 }, { "epoch": 2.75, "learning_rate": 3.592046242534819e-07, "loss": 0.0398, "step": 3480 }, { "epoch": 2.75, "learning_rate": 3.5694122427173517e-07, "loss": 0.0427, "step": 3481 }, { "epoch": 2.75, "learning_rate": 3.5468484826403193e-07, "loss": 0.0334, "step": 3482 }, { "epoch": 2.75, "learning_rate": 3.524354978739075e-07, "loss": 0.0313, "step": 3483 }, { "epoch": 2.75, "learning_rate": 3.501931747397858e-07, "loss": 0.0364, "step": 3484 }, { "epoch": 2.75, "learning_rate": 3.479578804949646e-07, "loss": 0.0301, "step": 3485 }, { "epoch": 2.75, "learning_rate": 3.4572961676762715e-07, "loss": 0.0306, "step": 3486 }, { "epoch": 2.76, "learning_rate": 3.4350838518083185e-07, "loss": 0.0378, "step": 3487 }, { "epoch": 2.76, "learning_rate": 3.4129418735251886e-07, "loss": 0.0279, "step": 3488 }, { "epoch": 2.76, "learning_rate": 3.390870248955025e-07, "loss": 0.0369, "step": 3489 }, { "epoch": 2.76, "learning_rate": 3.368868994174712e-07, "loss": 0.0381, "step": 3490 }, { "epoch": 2.76, "learning_rate": 3.3469381252099197e-07, "loss": 0.0311, "step": 3491 }, { "epoch": 2.76, "learning_rate": 3.3250776580350143e-07, "loss": 0.0377, "step": 3492 }, { "epoch": 2.76, "learning_rate": 3.303287608573091e-07, "loss": 0.0301, "step": 3493 }, { "epoch": 2.76, "learning_rate": 3.2815679926959553e-07, "loss": 0.035, "step": 3494 }, { "epoch": 2.76, "learning_rate": 3.259918826224118e-07, "loss": 0.0369, "step": 3495 }, { "epoch": 2.76, "learning_rate": 3.2383401249267423e-07, "loss": 0.033, "step": 3496 }, { "epoch": 2.76, "learning_rate": 3.216831904521711e-07, "loss": 0.0432, "step": 3497 }, { "epoch": 2.76, "learning_rate": 3.1953941806755265e-07, "loss": 0.0301, "step": 3498 }, { "epoch": 2.76, "learning_rate": 3.1740269690033743e-07, "loss": 0.0264, "step": 3499 }, { "epoch": 2.77, "learning_rate": 3.152730285069061e-07, "loss": 0.0374, "step": 3500 }, { "epoch": 2.77, "learning_rate": 3.131504144385023e-07, "loss": 0.0325, "step": 3501 }, { "epoch": 2.77, "learning_rate": 3.110348562412313e-07, "loss": 0.0353, "step": 3502 }, { "epoch": 2.77, "learning_rate": 3.0892635545605843e-07, "loss": 0.0297, "step": 3503 }, { "epoch": 2.77, "learning_rate": 3.0682491361881064e-07, "loss": 0.0339, "step": 3504 }, { "epoch": 2.77, "learning_rate": 3.0473053226017234e-07, "loss": 0.0299, "step": 3505 }, { "epoch": 2.77, "learning_rate": 3.026432129056811e-07, "loss": 0.0427, "step": 3506 }, { "epoch": 2.77, "learning_rate": 3.0056295707573736e-07, "loss": 0.0357, "step": 3507 }, { "epoch": 2.77, "learning_rate": 2.984897662855912e-07, "loss": 0.0324, "step": 3508 }, { "epoch": 2.77, "learning_rate": 2.96423642045347e-07, "loss": 0.0358, "step": 3509 }, { "epoch": 2.77, "learning_rate": 2.943645858599653e-07, "loss": 0.0359, "step": 3510 }, { "epoch": 2.77, "learning_rate": 2.9231259922925747e-07, "loss": 0.0373, "step": 3511 }, { "epoch": 2.78, "learning_rate": 2.902676836478813e-07, "loss": 0.0303, "step": 3512 }, { "epoch": 2.78, "learning_rate": 2.8822984060534854e-07, "loss": 0.0379, "step": 3513 }, { "epoch": 2.78, "learning_rate": 2.8619907158601744e-07, "loss": 0.0288, "step": 3514 }, { "epoch": 2.78, "learning_rate": 2.8417537806909254e-07, "loss": 0.0364, "step": 3515 }, { "epoch": 2.78, "learning_rate": 2.82158761528627e-07, "loss": 0.0368, "step": 3516 }, { "epoch": 2.78, "learning_rate": 2.801492234335201e-07, "loss": 0.0342, "step": 3517 }, { "epoch": 2.78, "learning_rate": 2.7814676524751007e-07, "loss": 0.0316, "step": 3518 }, { "epoch": 2.78, "learning_rate": 2.761513884291822e-07, "loss": 0.0376, "step": 3519 }, { "epoch": 2.78, "learning_rate": 2.741630944319629e-07, "loss": 0.0375, "step": 3520 }, { "epoch": 2.78, "learning_rate": 2.721818847041202e-07, "loss": 0.0372, "step": 3521 }, { "epoch": 2.78, "learning_rate": 2.7020776068875876e-07, "loss": 0.0329, "step": 3522 }, { "epoch": 2.78, "learning_rate": 2.6824072382382716e-07, "loss": 0.0399, "step": 3523 }, { "epoch": 2.78, "learning_rate": 2.6628077554210705e-07, "loss": 0.0369, "step": 3524 }, { "epoch": 2.79, "learning_rate": 2.6432791727121984e-07, "loss": 0.0337, "step": 3525 }, { "epoch": 2.79, "learning_rate": 2.623821504336199e-07, "loss": 0.0361, "step": 3526 }, { "epoch": 2.79, "learning_rate": 2.6044347644659905e-07, "loss": 0.0343, "step": 3527 }, { "epoch": 2.79, "learning_rate": 2.5851189672228103e-07, "loss": 0.0298, "step": 3528 }, { "epoch": 2.79, "learning_rate": 2.5658741266762267e-07, "loss": 0.036, "step": 3529 }, { "epoch": 2.79, "learning_rate": 2.5467002568441144e-07, "loss": 0.0317, "step": 3530 }, { "epoch": 2.79, "learning_rate": 2.5275973716926804e-07, "loss": 0.044, "step": 3531 }, { "epoch": 2.79, "learning_rate": 2.508565485136383e-07, "loss": 0.0413, "step": 3532 }, { "epoch": 2.79, "learning_rate": 2.489604611037999e-07, "loss": 0.0329, "step": 3533 }, { "epoch": 2.79, "learning_rate": 2.4707147632085815e-07, "loss": 0.0323, "step": 3534 }, { "epoch": 2.79, "learning_rate": 2.4518959554074017e-07, "loss": 0.0304, "step": 3535 }, { "epoch": 2.79, "learning_rate": 2.4331482013420616e-07, "loss": 0.0327, "step": 3536 }, { "epoch": 2.79, "learning_rate": 2.414471514668348e-07, "loss": 0.0263, "step": 3537 }, { "epoch": 2.8, "learning_rate": 2.3958659089903136e-07, "loss": 0.0315, "step": 3538 }, { "epoch": 2.8, "learning_rate": 2.3773313978602053e-07, "loss": 0.0425, "step": 3539 }, { "epoch": 2.8, "learning_rate": 2.358867994778502e-07, "loss": 0.0333, "step": 3540 }, { "epoch": 2.8, "learning_rate": 2.3404757131939014e-07, "loss": 0.0294, "step": 3541 }, { "epoch": 2.8, "learning_rate": 2.3221545665032874e-07, "loss": 0.0357, "step": 3542 }, { "epoch": 2.8, "learning_rate": 2.3039045680517292e-07, "loss": 0.0306, "step": 3543 }, { "epoch": 2.8, "learning_rate": 2.2857257311324487e-07, "loss": 0.0314, "step": 3544 }, { "epoch": 2.8, "learning_rate": 2.2676180689868765e-07, "loss": 0.0298, "step": 3545 }, { "epoch": 2.8, "learning_rate": 2.249581594804562e-07, "loss": 0.0442, "step": 3546 }, { "epoch": 2.8, "learning_rate": 2.231616321723218e-07, "loss": 0.0335, "step": 3547 }, { "epoch": 2.8, "learning_rate": 2.213722262828688e-07, "loss": 0.0348, "step": 3548 }, { "epoch": 2.8, "learning_rate": 2.1958994311549797e-07, "loss": 0.0282, "step": 3549 }, { "epoch": 2.81, "learning_rate": 2.1781478396841526e-07, "loss": 0.0312, "step": 3550 }, { "epoch": 2.81, "learning_rate": 2.1604675013464306e-07, "loss": 0.0465, "step": 3551 }, { "epoch": 2.81, "learning_rate": 2.1428584290201116e-07, "loss": 0.0377, "step": 3552 }, { "epoch": 2.81, "learning_rate": 2.1253206355315913e-07, "loss": 0.0341, "step": 3553 }, { "epoch": 2.81, "learning_rate": 2.1078541336553292e-07, "loss": 0.0268, "step": 3554 }, { "epoch": 2.81, "learning_rate": 2.0904589361138927e-07, "loss": 0.0466, "step": 3555 }, { "epoch": 2.81, "learning_rate": 2.0731350555778907e-07, "loss": 0.0307, "step": 3556 }, { "epoch": 2.81, "learning_rate": 2.0558825046659626e-07, "loss": 0.0324, "step": 3557 }, { "epoch": 2.81, "learning_rate": 2.0387012959448227e-07, "loss": 0.0328, "step": 3558 }, { "epoch": 2.81, "learning_rate": 2.021591441929216e-07, "loss": 0.0286, "step": 3559 }, { "epoch": 2.81, "learning_rate": 2.0045529550819064e-07, "loss": 0.0318, "step": 3560 }, { "epoch": 2.81, "learning_rate": 1.9875858478136557e-07, "loss": 0.0344, "step": 3561 }, { "epoch": 2.81, "learning_rate": 1.9706901324832772e-07, "loss": 0.0279, "step": 3562 }, { "epoch": 2.82, "learning_rate": 1.9538658213975492e-07, "loss": 0.0291, "step": 3563 }, { "epoch": 2.82, "learning_rate": 1.9371129268112466e-07, "loss": 0.0387, "step": 3564 }, { "epoch": 2.82, "learning_rate": 1.9204314609271302e-07, "loss": 0.0321, "step": 3565 }, { "epoch": 2.82, "learning_rate": 1.903821435895914e-07, "loss": 0.0362, "step": 3566 }, { "epoch": 2.82, "learning_rate": 1.8872828638162866e-07, "loss": 0.0378, "step": 3567 }, { "epoch": 2.82, "learning_rate": 1.8708157567349228e-07, "loss": 0.0362, "step": 3568 }, { "epoch": 2.82, "learning_rate": 1.8544201266463834e-07, "loss": 0.0381, "step": 3569 }, { "epoch": 2.82, "learning_rate": 1.8380959854932045e-07, "loss": 0.0358, "step": 3570 }, { "epoch": 2.82, "learning_rate": 1.8218433451658412e-07, "loss": 0.0328, "step": 3571 }, { "epoch": 2.82, "learning_rate": 1.8056622175026683e-07, "loss": 0.0261, "step": 3572 }, { "epoch": 2.82, "learning_rate": 1.7895526142899466e-07, "loss": 0.0357, "step": 3573 }, { "epoch": 2.82, "learning_rate": 1.7735145472619008e-07, "loss": 0.0243, "step": 3574 }, { "epoch": 2.82, "learning_rate": 1.7575480281005752e-07, "loss": 0.0294, "step": 3575 }, { "epoch": 2.83, "learning_rate": 1.7416530684359444e-07, "loss": 0.0288, "step": 3576 }, { "epoch": 2.83, "learning_rate": 1.7258296798458585e-07, "loss": 0.0304, "step": 3577 }, { "epoch": 2.83, "learning_rate": 1.7100778738560086e-07, "loss": 0.036, "step": 3578 }, { "epoch": 2.83, "learning_rate": 1.6943976619399615e-07, "loss": 0.0361, "step": 3579 }, { "epoch": 2.83, "learning_rate": 1.6787890555191366e-07, "loss": 0.0258, "step": 3580 }, { "epoch": 2.83, "learning_rate": 1.663252065962806e-07, "loss": 0.0311, "step": 3581 }, { "epoch": 2.83, "learning_rate": 1.6477867045880613e-07, "loss": 0.0421, "step": 3582 }, { "epoch": 2.83, "learning_rate": 1.632392982659814e-07, "loss": 0.0408, "step": 3583 }, { "epoch": 2.83, "learning_rate": 1.6170709113908055e-07, "loss": 0.0311, "step": 3584 }, { "epoch": 2.83, "learning_rate": 1.6018205019415866e-07, "loss": 0.0287, "step": 3585 }, { "epoch": 2.83, "learning_rate": 1.5866417654205045e-07, "loss": 0.0292, "step": 3586 }, { "epoch": 2.83, "learning_rate": 1.5715347128837154e-07, "loss": 0.0327, "step": 3587 }, { "epoch": 2.84, "learning_rate": 1.5564993553351394e-07, "loss": 0.0315, "step": 3588 }, { "epoch": 2.84, "learning_rate": 1.5415357037264933e-07, "loss": 0.034, "step": 3589 }, { "epoch": 2.84, "learning_rate": 1.5266437689572478e-07, "loss": 0.0364, "step": 3590 }, { "epoch": 2.84, "learning_rate": 1.511823561874637e-07, "loss": 0.0297, "step": 3591 }, { "epoch": 2.84, "learning_rate": 1.497075093273659e-07, "loss": 0.0299, "step": 3592 }, { "epoch": 2.84, "learning_rate": 1.4823983738970538e-07, "loss": 0.0365, "step": 3593 }, { "epoch": 2.84, "learning_rate": 1.4677934144352923e-07, "loss": 0.0343, "step": 3594 }, { "epoch": 2.84, "learning_rate": 1.4532602255265982e-07, "loss": 0.0346, "step": 3595 }, { "epoch": 2.84, "learning_rate": 1.4387988177568814e-07, "loss": 0.029, "step": 3596 }, { "epoch": 2.84, "learning_rate": 1.4244092016597933e-07, "loss": 0.039, "step": 3597 }, { "epoch": 2.84, "learning_rate": 1.4100913877166834e-07, "loss": 0.0337, "step": 3598 }, { "epoch": 2.84, "learning_rate": 1.395845386356609e-07, "loss": 0.0309, "step": 3599 }, { "epoch": 2.84, "learning_rate": 1.3816712079563034e-07, "loss": 0.0332, "step": 3600 }, { "epoch": 2.85, "learning_rate": 1.3675688628402073e-07, "loss": 0.0369, "step": 3601 }, { "epoch": 2.85, "learning_rate": 1.3535383612804266e-07, "loss": 0.0363, "step": 3602 }, { "epoch": 2.85, "learning_rate": 1.3395797134967192e-07, "loss": 0.0393, "step": 3603 }, { "epoch": 2.85, "learning_rate": 1.3256929296565303e-07, "loss": 0.0394, "step": 3604 }, { "epoch": 2.85, "learning_rate": 1.3118780198749458e-07, "loss": 0.0329, "step": 3605 }, { "epoch": 2.85, "learning_rate": 1.2981349942146947e-07, "loss": 0.038, "step": 3606 }, { "epoch": 2.85, "learning_rate": 1.2844638626861806e-07, "loss": 0.0321, "step": 3607 }, { "epoch": 2.85, "learning_rate": 1.270864635247393e-07, "loss": 0.0403, "step": 3608 }, { "epoch": 2.85, "learning_rate": 1.257337321803964e-07, "loss": 0.0393, "step": 3609 }, { "epoch": 2.85, "learning_rate": 1.2438819322091678e-07, "loss": 0.0498, "step": 3610 }, { "epoch": 2.85, "learning_rate": 1.2304984762638527e-07, "loss": 0.0264, "step": 3611 }, { "epoch": 2.85, "learning_rate": 1.2171869637164769e-07, "loss": 0.0313, "step": 3612 }, { "epoch": 2.85, "learning_rate": 1.203947404263117e-07, "loss": 0.0339, "step": 3613 }, { "epoch": 2.86, "learning_rate": 1.1907798075474375e-07, "loss": 0.0344, "step": 3614 }, { "epoch": 2.86, "learning_rate": 1.1776841831606544e-07, "loss": 0.0372, "step": 3615 }, { "epoch": 2.86, "learning_rate": 1.1646605406415933e-07, "loss": 0.0332, "step": 3616 }, { "epoch": 2.86, "learning_rate": 1.1517088894766326e-07, "loss": 0.035, "step": 3617 }, { "epoch": 2.86, "learning_rate": 1.1388292390997035e-07, "loss": 0.0304, "step": 3618 }, { "epoch": 2.86, "learning_rate": 1.1260215988923239e-07, "loss": 0.0359, "step": 3619 }, { "epoch": 2.86, "learning_rate": 1.1132859781835204e-07, "loss": 0.0305, "step": 3620 }, { "epoch": 2.86, "learning_rate": 1.1006223862498944e-07, "loss": 0.029, "step": 3621 }, { "epoch": 2.86, "learning_rate": 1.0880308323155564e-07, "loss": 0.0313, "step": 3622 }, { "epoch": 2.86, "learning_rate": 1.0755113255521588e-07, "loss": 0.0322, "step": 3623 }, { "epoch": 2.86, "learning_rate": 1.0630638750788625e-07, "loss": 0.039, "step": 3624 }, { "epoch": 2.86, "learning_rate": 1.0506884899623704e-07, "loss": 0.0306, "step": 3625 }, { "epoch": 2.87, "learning_rate": 1.0383851792168609e-07, "loss": 0.0421, "step": 3626 }, { "epoch": 2.87, "learning_rate": 1.026153951804032e-07, "loss": 0.0334, "step": 3627 }, { "epoch": 2.87, "learning_rate": 1.0139948166330682e-07, "loss": 0.0304, "step": 3628 }, { "epoch": 2.87, "learning_rate": 1.0019077825606404e-07, "loss": 0.0373, "step": 3629 }, { "epoch": 2.87, "learning_rate": 9.898928583909284e-08, "loss": 0.0299, "step": 3630 }, { "epoch": 2.87, "learning_rate": 9.779500528755425e-08, "loss": 0.0328, "step": 3631 }, { "epoch": 2.87, "learning_rate": 9.660793747136021e-08, "loss": 0.0355, "step": 3632 }, { "epoch": 2.87, "learning_rate": 9.542808325516573e-08, "loss": 0.0318, "step": 3633 }, { "epoch": 2.87, "learning_rate": 9.425544349837445e-08, "loss": 0.0403, "step": 3634 }, { "epoch": 2.87, "learning_rate": 9.309001905513094e-08, "loss": 0.0351, "step": 3635 }, { "epoch": 2.87, "learning_rate": 9.193181077433055e-08, "loss": 0.027, "step": 3636 }, { "epoch": 2.87, "learning_rate": 9.078081949960515e-08, "loss": 0.0416, "step": 3637 }, { "epoch": 2.87, "learning_rate": 8.963704606933521e-08, "loss": 0.0308, "step": 3638 }, { "epoch": 2.88, "learning_rate": 8.850049131664206e-08, "loss": 0.0414, "step": 3639 }, { "epoch": 2.88, "learning_rate": 8.737115606938684e-08, "loss": 0.0277, "step": 3640 }, { "epoch": 2.88, "learning_rate": 8.624904115017374e-08, "loss": 0.0422, "step": 3641 }, { "epoch": 2.88, "learning_rate": 8.513414737635006e-08, "loss": 0.0311, "step": 3642 }, { "epoch": 2.88, "learning_rate": 8.402647555999954e-08, "loss": 0.0378, "step": 3643 }, { "epoch": 2.88, "learning_rate": 8.292602650794678e-08, "loss": 0.0276, "step": 3644 }, { "epoch": 2.88, "learning_rate": 8.183280102175617e-08, "loss": 0.0391, "step": 3645 }, { "epoch": 2.88, "learning_rate": 8.07467998977296e-08, "loss": 0.0303, "step": 3646 }, { "epoch": 2.88, "learning_rate": 7.966802392690543e-08, "loss": 0.033, "step": 3647 }, { "epoch": 2.88, "learning_rate": 7.859647389506176e-08, "loss": 0.0305, "step": 3648 }, { "epoch": 2.88, "learning_rate": 7.753215058271424e-08, "loss": 0.0419, "step": 3649 }, { "epoch": 2.88, "learning_rate": 7.647505476510942e-08, "loss": 0.0343, "step": 3650 }, { "epoch": 2.89, "learning_rate": 7.542518721223469e-08, "loss": 0.0385, "step": 3651 }, { "epoch": 2.89, "learning_rate": 7.438254868880946e-08, "loss": 0.0253, "step": 3652 }, { "epoch": 2.89, "learning_rate": 7.334713995428733e-08, "loss": 0.0366, "step": 3653 }, { "epoch": 2.89, "learning_rate": 7.231896176285946e-08, "loss": 0.038, "step": 3654 }, { "epoch": 2.89, "learning_rate": 7.129801486344678e-08, "loss": 0.0291, "step": 3655 }, { "epoch": 2.89, "learning_rate": 7.028429999970221e-08, "loss": 0.0385, "step": 3656 }, { "epoch": 2.89, "learning_rate": 6.927781791001398e-08, "loss": 0.0336, "step": 3657 }, { "epoch": 2.89, "learning_rate": 6.82785693275001e-08, "loss": 0.0378, "step": 3658 }, { "epoch": 2.89, "learning_rate": 6.728655498001058e-08, "loss": 0.0336, "step": 3659 }, { "epoch": 2.89, "learning_rate": 6.630177559012518e-08, "loss": 0.034, "step": 3660 }, { "epoch": 2.89, "learning_rate": 6.532423187515346e-08, "loss": 0.03, "step": 3661 }, { "epoch": 2.89, "learning_rate": 6.435392454713696e-08, "loss": 0.0334, "step": 3662 }, { "epoch": 2.89, "learning_rate": 6.339085431284253e-08, "loss": 0.0308, "step": 3663 }, { "epoch": 2.9, "learning_rate": 6.243502187376904e-08, "loss": 0.0409, "step": 3664 }, { "epoch": 2.9, "learning_rate": 6.148642792614068e-08, "loss": 0.0349, "step": 3665 }, { "epoch": 2.9, "learning_rate": 6.054507316091141e-08, "loss": 0.0323, "step": 3666 }, { "epoch": 2.9, "learning_rate": 5.961095826376051e-08, "loss": 0.0287, "step": 3667 }, { "epoch": 2.9, "learning_rate": 5.868408391509484e-08, "loss": 0.0402, "step": 3668 }, { "epoch": 2.9, "learning_rate": 5.7764450790046554e-08, "loss": 0.0394, "step": 3669 }, { "epoch": 2.9, "learning_rate": 5.685205955847317e-08, "loss": 0.0371, "step": 3670 }, { "epoch": 2.9, "learning_rate": 5.5946910884958625e-08, "loss": 0.0287, "step": 3671 }, { "epoch": 2.9, "learning_rate": 5.5049005428808865e-08, "loss": 0.0319, "step": 3672 }, { "epoch": 2.9, "learning_rate": 5.415834384405738e-08, "loss": 0.0373, "step": 3673 }, { "epoch": 2.9, "learning_rate": 5.327492677945745e-08, "loss": 0.0298, "step": 3674 }, { "epoch": 2.9, "learning_rate": 5.239875487848878e-08, "loss": 0.0382, "step": 3675 }, { "epoch": 2.9, "learning_rate": 5.152982877934976e-08, "loss": 0.0316, "step": 3676 }, { "epoch": 2.91, "learning_rate": 5.066814911496742e-08, "loss": 0.0488, "step": 3677 }, { "epoch": 2.91, "learning_rate": 4.981371651298306e-08, "loss": 0.0348, "step": 3678 }, { "epoch": 2.91, "learning_rate": 4.8966531595764366e-08, "loss": 0.0292, "step": 3679 }, { "epoch": 2.91, "learning_rate": 4.812659498039884e-08, "loss": 0.0323, "step": 3680 }, { "epoch": 2.91, "learning_rate": 4.729390727869154e-08, "loss": 0.0295, "step": 3681 }, { "epoch": 2.91, "learning_rate": 4.646846909717173e-08, "loss": 0.035, "step": 3682 }, { "epoch": 2.91, "learning_rate": 4.5650281037084024e-08, "loss": 0.0316, "step": 3683 }, { "epoch": 2.91, "learning_rate": 4.483934369439613e-08, "loss": 0.0315, "step": 3684 }, { "epoch": 2.91, "learning_rate": 4.403565765979112e-08, "loss": 0.0335, "step": 3685 }, { "epoch": 2.91, "learning_rate": 4.3239223518670715e-08, "loss": 0.0446, "step": 3686 }, { "epoch": 2.91, "learning_rate": 4.245004185115753e-08, "loss": 0.0337, "step": 3687 }, { "epoch": 2.91, "learning_rate": 4.166811323208619e-08, "loss": 0.0332, "step": 3688 }, { "epoch": 2.92, "learning_rate": 4.089343823101222e-08, "loss": 0.0272, "step": 3689 }, { "epoch": 2.92, "learning_rate": 4.0126017412207565e-08, "loss": 0.0308, "step": 3690 }, { "epoch": 2.92, "learning_rate": 3.936585133465842e-08, "loss": 0.0454, "step": 3691 }, { "epoch": 2.92, "learning_rate": 3.861294055206633e-08, "loss": 0.0329, "step": 3692 }, { "epoch": 2.92, "learning_rate": 3.786728561285036e-08, "loss": 0.0424, "step": 3693 }, { "epoch": 2.92, "learning_rate": 3.712888706014273e-08, "loss": 0.0442, "step": 3694 }, { "epoch": 2.92, "learning_rate": 3.6397745431790974e-08, "loss": 0.0319, "step": 3695 }, { "epoch": 2.92, "learning_rate": 3.567386126035577e-08, "loss": 0.0342, "step": 3696 }, { "epoch": 2.92, "learning_rate": 3.4957235073113103e-08, "loss": 0.0324, "step": 3697 }, { "epoch": 2.92, "learning_rate": 3.4247867392049885e-08, "loss": 0.0348, "step": 3698 }, { "epoch": 2.92, "learning_rate": 3.354575873386945e-08, "loss": 0.0347, "step": 3699 }, { "epoch": 2.92, "learning_rate": 3.285090960998383e-08, "loss": 0.0384, "step": 3700 }, { "epoch": 2.92, "learning_rate": 3.216332052651927e-08, "loss": 0.0325, "step": 3701 }, { "epoch": 2.93, "learning_rate": 3.1482991984312926e-08, "loss": 0.0353, "step": 3702 }, { "epoch": 2.93, "learning_rate": 3.080992447891618e-08, "loss": 0.0366, "step": 3703 }, { "epoch": 2.93, "learning_rate": 3.014411850058685e-08, "loss": 0.0374, "step": 3704 }, { "epoch": 2.93, "learning_rate": 2.948557453429701e-08, "loss": 0.0364, "step": 3705 }, { "epoch": 2.93, "learning_rate": 2.8834293059728514e-08, "loss": 0.0285, "step": 3706 }, { "epoch": 2.93, "learning_rate": 2.819027455127188e-08, "loss": 0.0319, "step": 3707 }, { "epoch": 2.93, "learning_rate": 2.7553519478028535e-08, "loss": 0.0366, "step": 3708 }, { "epoch": 2.93, "learning_rate": 2.6924028303808582e-08, "loss": 0.0313, "step": 3709 }, { "epoch": 2.93, "learning_rate": 2.6301801487133018e-08, "loss": 0.031, "step": 3710 }, { "epoch": 2.93, "learning_rate": 2.5686839481227077e-08, "loss": 0.028, "step": 3711 }, { "epoch": 2.93, "learning_rate": 2.5079142734030227e-08, "loss": 0.0258, "step": 3712 }, { "epoch": 2.93, "learning_rate": 2.4478711688183943e-08, "loss": 0.0362, "step": 3713 }, { "epoch": 2.93, "learning_rate": 2.3885546781042824e-08, "loss": 0.0283, "step": 3714 }, { "epoch": 2.94, "learning_rate": 2.329964844466459e-08, "loss": 0.0316, "step": 3715 }, { "epoch": 2.94, "learning_rate": 2.2721017105817865e-08, "loss": 0.0387, "step": 3716 }, { "epoch": 2.94, "learning_rate": 2.2149653185973285e-08, "loss": 0.0463, "step": 3717 }, { "epoch": 2.94, "learning_rate": 2.1585557101312383e-08, "loss": 0.0333, "step": 3718 }, { "epoch": 2.94, "learning_rate": 2.1028729262719817e-08, "loss": 0.0319, "step": 3719 }, { "epoch": 2.94, "learning_rate": 2.0479170075788924e-08, "loss": 0.0371, "step": 3720 }, { "epoch": 2.94, "learning_rate": 1.9936879940815058e-08, "loss": 0.0342, "step": 3721 }, { "epoch": 2.94, "learning_rate": 1.9401859252801138e-08, "loss": 0.0339, "step": 3722 }, { "epoch": 2.94, "learning_rate": 1.8874108401456538e-08, "loss": 0.0355, "step": 3723 }, { "epoch": 2.94, "learning_rate": 1.835362777119043e-08, "loss": 0.0367, "step": 3724 }, { "epoch": 2.94, "learning_rate": 1.7840417741120662e-08, "loss": 0.0427, "step": 3725 }, { "epoch": 2.94, "learning_rate": 1.7334478685068212e-08, "loss": 0.0349, "step": 3726 }, { "epoch": 2.95, "learning_rate": 1.6835810971556066e-08, "loss": 0.0341, "step": 3727 }, { "epoch": 2.95, "learning_rate": 1.634441496381367e-08, "loss": 0.0349, "step": 3728 }, { "epoch": 2.95, "learning_rate": 1.586029101977249e-08, "loss": 0.0334, "step": 3729 }, { "epoch": 2.95, "learning_rate": 1.5383439492066e-08, "loss": 0.0413, "step": 3730 }, { "epoch": 2.95, "learning_rate": 1.4913860728030806e-08, "loss": 0.0397, "step": 3731 }, { "epoch": 2.95, "learning_rate": 1.4451555069708856e-08, "loss": 0.0379, "step": 3732 }, { "epoch": 2.95, "learning_rate": 1.3996522853839679e-08, "loss": 0.034, "step": 3733 }, { "epoch": 2.95, "learning_rate": 1.3548764411868143e-08, "loss": 0.0391, "step": 3734 }, { "epoch": 2.95, "learning_rate": 1.3108280069941137e-08, "loss": 0.0462, "step": 3735 }, { "epoch": 2.95, "learning_rate": 1.2675070148905343e-08, "loss": 0.0394, "step": 3736 }, { "epoch": 2.95, "learning_rate": 1.2249134964309462e-08, "loss": 0.0334, "step": 3737 }, { "epoch": 2.95, "learning_rate": 1.1830474826404204e-08, "loss": 0.0287, "step": 3738 }, { "epoch": 2.95, "learning_rate": 1.1419090040137859e-08, "loss": 0.033, "step": 3739 }, { "epoch": 2.96, "learning_rate": 1.1014980905165174e-08, "loss": 0.0441, "step": 3740 }, { "epoch": 2.96, "learning_rate": 1.0618147715835137e-08, "loss": 0.0367, "step": 3741 }, { "epoch": 2.96, "learning_rate": 1.0228590761200975e-08, "loss": 0.0333, "step": 3742 }, { "epoch": 2.96, "learning_rate": 9.846310325015706e-09, "loss": 0.037, "step": 3743 }, { "epoch": 2.96, "learning_rate": 9.471306685728821e-09, "loss": 0.0298, "step": 3744 }, { "epoch": 2.96, "learning_rate": 9.103580116494038e-09, "loss": 0.0303, "step": 3745 }, { "epoch": 2.96, "learning_rate": 8.743130885161544e-09, "loss": 0.0355, "step": 3746 }, { "epoch": 2.96, "learning_rate": 8.389959254281322e-09, "loss": 0.0404, "step": 3747 }, { "epoch": 2.96, "learning_rate": 8.044065481103147e-09, "loss": 0.0499, "step": 3748 }, { "epoch": 2.96, "learning_rate": 7.705449817573263e-09, "loss": 0.0353, "step": 3749 }, { "epoch": 2.96, "learning_rate": 7.3741125103399254e-09, "loss": 0.0376, "step": 3750 }, { "epoch": 2.96, "learning_rate": 7.050053800747858e-09, "loss": 0.0406, "step": 3751 }, { "epoch": 2.96, "learning_rate": 6.7332739248393564e-09, "loss": 0.037, "step": 3752 }, { "epoch": 2.97, "learning_rate": 6.423773113357623e-09, "loss": 0.0417, "step": 3753 }, { "epoch": 2.97, "learning_rate": 6.121551591740105e-09, "loss": 0.0302, "step": 3754 }, { "epoch": 2.97, "learning_rate": 5.826609580126263e-09, "loss": 0.0345, "step": 3755 }, { "epoch": 2.97, "learning_rate": 5.538947293349806e-09, "loss": 0.0368, "step": 3756 }, { "epoch": 2.97, "learning_rate": 5.258564940942013e-09, "loss": 0.0291, "step": 3757 }, { "epoch": 2.97, "learning_rate": 4.985462727133961e-09, "loss": 0.0364, "step": 3758 }, { "epoch": 2.97, "learning_rate": 4.719640850852081e-09, "loss": 0.0391, "step": 3759 }, { "epoch": 2.97, "learning_rate": 4.461099505720379e-09, "loss": 0.033, "step": 3760 }, { "epoch": 2.97, "learning_rate": 4.209838880059325e-09, "loss": 0.0484, "step": 3761 }, { "epoch": 2.97, "learning_rate": 3.965859156885854e-09, "loss": 0.0364, "step": 3762 }, { "epoch": 2.97, "learning_rate": 3.729160513914476e-09, "loss": 0.0377, "step": 3763 }, { "epoch": 2.97, "learning_rate": 3.4997431235561653e-09, "loss": 0.0357, "step": 3764 }, { "epoch": 2.98, "learning_rate": 3.2776071529183608e-09, "loss": 0.0316, "step": 3765 }, { "epoch": 2.98, "learning_rate": 3.062752763801635e-09, "loss": 0.0369, "step": 3766 }, { "epoch": 2.98, "learning_rate": 2.8551801127085776e-09, "loss": 0.0316, "step": 3767 }, { "epoch": 2.98, "learning_rate": 2.65488935083158e-09, "loss": 0.0338, "step": 3768 }, { "epoch": 2.98, "learning_rate": 2.4618806240639392e-09, "loss": 0.0386, "step": 3769 }, { "epoch": 2.98, "learning_rate": 2.276154072992087e-09, "loss": 0.0308, "step": 3770 }, { "epoch": 2.98, "learning_rate": 2.0977098328978098e-09, "loss": 0.0338, "step": 3771 }, { "epoch": 2.98, "learning_rate": 1.9265480337593566e-09, "loss": 0.0358, "step": 3772 }, { "epoch": 2.98, "learning_rate": 1.7626688002514436e-09, "loss": 0.0339, "step": 3773 }, { "epoch": 2.98, "learning_rate": 1.6060722517430293e-09, "loss": 0.0288, "step": 3774 }, { "epoch": 2.98, "learning_rate": 1.4567585022984276e-09, "loss": 0.0357, "step": 3775 }, { "epoch": 2.98, "learning_rate": 1.3147276606761961e-09, "loss": 0.0403, "step": 3776 }, { "epoch": 2.98, "learning_rate": 1.1799798303335775e-09, "loss": 0.0281, "step": 3777 }, { "epoch": 2.99, "learning_rate": 1.0525151094176178e-09, "loss": 0.0309, "step": 3778 }, { "epoch": 2.99, "learning_rate": 9.323335907751585e-10, "loss": 0.0356, "step": 3779 }, { "epoch": 2.99, "learning_rate": 8.194353619450646e-10, "loss": 0.0343, "step": 3780 }, { "epoch": 2.99, "learning_rate": 7.138205051626657e-10, "loss": 0.0406, "step": 3781 }, { "epoch": 2.99, "learning_rate": 6.154890973575356e-10, "loss": 0.0389, "step": 3782 }, { "epoch": 2.99, "learning_rate": 5.244412101534924e-10, "loss": 0.0396, "step": 3783 }, { "epoch": 2.99, "learning_rate": 4.4067690986970836e-10, "loss": 0.0335, "step": 3784 }, { "epoch": 2.99, "learning_rate": 3.641962575207103e-10, "loss": 0.0295, "step": 3785 }, { "epoch": 2.99, "learning_rate": 2.949993088130487e-10, "loss": 0.0293, "step": 3786 }, { "epoch": 2.99, "learning_rate": 2.3308611415084893e-10, "loss": 0.0353, "step": 3787 }, { "epoch": 2.99, "learning_rate": 1.7845671863025993e-10, "loss": 0.0267, "step": 3788 }, { "epoch": 2.99, "learning_rate": 1.3111116204500562e-10, "loss": 0.0311, "step": 3789 }, { "epoch": 2.99, "learning_rate": 9.10494788808336e-11, "loss": 0.0351, "step": 3790 }, { "epoch": 3.0, "learning_rate": 5.82716983177356e-11, "loss": 0.0371, "step": 3791 }, { "epoch": 3.0, "learning_rate": 3.277784423105779e-11, "loss": 0.0314, "step": 3792 }, { "epoch": 3.0, "learning_rate": 1.456793519039046e-11, "loss": 0.0362, "step": 3793 }, { "epoch": 3.0, "learning_rate": 3.641984460678316e-12, "loss": 0.0327, "step": 3794 }, { "epoch": 3.0, "learning_rate": 0.0, "loss": 0.0344, "step": 3795 }, { "epoch": 3.0, "step": 3795, "total_flos": 1.0929844817714217e+18, "train_loss": 0.17524353075080473, "train_runtime": 24461.6831, "train_samples_per_second": 9.931, "train_steps_per_second": 0.155 } ], "max_steps": 3795, "num_train_epochs": 3, "total_flos": 1.0929844817714217e+18, "trial_name": null, "trial_params": null }