{ "best_metric": null, "best_model_checkpoint": null, "epoch": 1.5686958037387249, "eval_steps": 500, "global_step": 9000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 4.355400696864112e-08, "loss": 0.6309, "step": 1 }, { "epoch": 0.0, "learning_rate": 8.710801393728224e-08, "loss": 0.6181, "step": 2 }, { "epoch": 0.0, "learning_rate": 1.3066202090592336e-07, "loss": 0.6113, "step": 3 }, { "epoch": 0.0, "learning_rate": 1.7421602787456448e-07, "loss": 0.619, "step": 4 }, { "epoch": 0.0, "learning_rate": 2.1777003484320556e-07, "loss": 0.6133, "step": 5 }, { "epoch": 0.0, "learning_rate": 2.6132404181184673e-07, "loss": 0.5963, "step": 6 }, { "epoch": 0.0, "learning_rate": 3.0487804878048784e-07, "loss": 0.6367, "step": 7 }, { "epoch": 0.0, "learning_rate": 3.4843205574912896e-07, "loss": 0.6349, "step": 8 }, { "epoch": 0.0, "learning_rate": 3.9198606271777007e-07, "loss": 0.6022, "step": 9 }, { "epoch": 0.0, "learning_rate": 4.3554006968641113e-07, "loss": 0.6241, "step": 10 }, { "epoch": 0.0, "learning_rate": 4.790940766550523e-07, "loss": 0.6141, "step": 11 }, { "epoch": 0.0, "learning_rate": 5.226480836236935e-07, "loss": 0.6209, "step": 12 }, { "epoch": 0.0, "learning_rate": 5.662020905923345e-07, "loss": 0.6265, "step": 13 }, { "epoch": 0.0, "learning_rate": 6.097560975609757e-07, "loss": 0.6186, "step": 14 }, { "epoch": 0.0, "learning_rate": 6.533101045296167e-07, "loss": 0.6143, "step": 15 }, { "epoch": 0.0, "learning_rate": 6.968641114982579e-07, "loss": 0.6082, "step": 16 }, { "epoch": 0.0, "learning_rate": 7.40418118466899e-07, "loss": 0.5972, "step": 17 }, { "epoch": 0.0, "learning_rate": 7.839721254355401e-07, "loss": 0.595, "step": 18 }, { "epoch": 0.0, "learning_rate": 8.275261324041812e-07, "loss": 0.5574, "step": 19 }, { "epoch": 0.0, "learning_rate": 8.710801393728223e-07, "loss": 0.6042, "step": 20 }, { "epoch": 0.0, "learning_rate": 9.146341463414634e-07, "loss": 0.5634, "step": 21 }, { "epoch": 0.0, "learning_rate": 9.581881533101046e-07, "loss": 0.5671, "step": 22 }, { "epoch": 0.0, "learning_rate": 1.0017421602787455e-06, "loss": 0.5882, "step": 23 }, { "epoch": 0.0, "learning_rate": 1.045296167247387e-06, "loss": 0.5797, "step": 24 }, { "epoch": 0.0, "learning_rate": 1.088850174216028e-06, "loss": 0.5548, "step": 25 }, { "epoch": 0.0, "learning_rate": 1.132404181184669e-06, "loss": 0.5518, "step": 26 }, { "epoch": 0.0, "learning_rate": 1.1759581881533102e-06, "loss": 0.5445, "step": 27 }, { "epoch": 0.0, "learning_rate": 1.2195121951219514e-06, "loss": 0.5509, "step": 28 }, { "epoch": 0.01, "learning_rate": 1.2630662020905923e-06, "loss": 0.5398, "step": 29 }, { "epoch": 0.01, "learning_rate": 1.3066202090592335e-06, "loss": 0.5368, "step": 30 }, { "epoch": 0.01, "learning_rate": 1.3501742160278747e-06, "loss": 0.5364, "step": 31 }, { "epoch": 0.01, "learning_rate": 1.3937282229965158e-06, "loss": 0.53, "step": 32 }, { "epoch": 0.01, "learning_rate": 1.4372822299651568e-06, "loss": 0.529, "step": 33 }, { "epoch": 0.01, "learning_rate": 1.480836236933798e-06, "loss": 0.519, "step": 34 }, { "epoch": 0.01, "learning_rate": 1.5243902439024391e-06, "loss": 0.5226, "step": 35 }, { "epoch": 0.01, "learning_rate": 1.5679442508710803e-06, "loss": 0.5164, "step": 36 }, { "epoch": 0.01, "learning_rate": 1.6114982578397212e-06, "loss": 0.5038, "step": 37 }, { "epoch": 0.01, "learning_rate": 1.6550522648083624e-06, "loss": 0.5007, "step": 38 }, { "epoch": 0.01, "learning_rate": 1.6986062717770036e-06, "loss": 0.4941, "step": 39 }, { "epoch": 0.01, "learning_rate": 1.7421602787456445e-06, "loss": 0.4985, "step": 40 }, { "epoch": 0.01, "learning_rate": 1.7857142857142857e-06, "loss": 0.5031, "step": 41 }, { "epoch": 0.01, "learning_rate": 1.8292682926829268e-06, "loss": 0.4914, "step": 42 }, { "epoch": 0.01, "learning_rate": 1.872822299651568e-06, "loss": 0.4951, "step": 43 }, { "epoch": 0.01, "learning_rate": 1.916376306620209e-06, "loss": 0.4971, "step": 44 }, { "epoch": 0.01, "learning_rate": 1.95993031358885e-06, "loss": 0.4885, "step": 45 }, { "epoch": 0.01, "learning_rate": 2.003484320557491e-06, "loss": 0.4975, "step": 46 }, { "epoch": 0.01, "learning_rate": 2.0470383275261322e-06, "loss": 0.4969, "step": 47 }, { "epoch": 0.01, "learning_rate": 2.090592334494774e-06, "loss": 0.493, "step": 48 }, { "epoch": 0.01, "learning_rate": 2.134146341463415e-06, "loss": 0.4746, "step": 49 }, { "epoch": 0.01, "learning_rate": 2.177700348432056e-06, "loss": 0.4842, "step": 50 }, { "epoch": 0.01, "learning_rate": 2.221254355400697e-06, "loss": 0.4935, "step": 51 }, { "epoch": 0.01, "learning_rate": 2.264808362369338e-06, "loss": 0.4885, "step": 52 }, { "epoch": 0.01, "learning_rate": 2.3083623693379792e-06, "loss": 0.4905, "step": 53 }, { "epoch": 0.01, "learning_rate": 2.3519163763066204e-06, "loss": 0.5062, "step": 54 }, { "epoch": 0.01, "learning_rate": 2.3954703832752616e-06, "loss": 0.4877, "step": 55 }, { "epoch": 0.01, "learning_rate": 2.4390243902439027e-06, "loss": 0.4746, "step": 56 }, { "epoch": 0.01, "learning_rate": 2.482578397212544e-06, "loss": 0.4737, "step": 57 }, { "epoch": 0.01, "learning_rate": 2.5261324041811846e-06, "loss": 0.4684, "step": 58 }, { "epoch": 0.01, "learning_rate": 2.569686411149826e-06, "loss": 0.4817, "step": 59 }, { "epoch": 0.01, "learning_rate": 2.613240418118467e-06, "loss": 0.4789, "step": 60 }, { "epoch": 0.01, "learning_rate": 2.656794425087108e-06, "loss": 0.4831, "step": 61 }, { "epoch": 0.01, "learning_rate": 2.7003484320557493e-06, "loss": 0.4677, "step": 62 }, { "epoch": 0.01, "learning_rate": 2.7439024390243905e-06, "loss": 0.4733, "step": 63 }, { "epoch": 0.01, "learning_rate": 2.7874564459930316e-06, "loss": 0.4702, "step": 64 }, { "epoch": 0.01, "learning_rate": 2.8310104529616724e-06, "loss": 0.4718, "step": 65 }, { "epoch": 0.01, "learning_rate": 2.8745644599303136e-06, "loss": 0.4532, "step": 66 }, { "epoch": 0.01, "learning_rate": 2.9181184668989547e-06, "loss": 0.4555, "step": 67 }, { "epoch": 0.01, "learning_rate": 2.961672473867596e-06, "loss": 0.475, "step": 68 }, { "epoch": 0.01, "learning_rate": 3.005226480836237e-06, "loss": 0.4925, "step": 69 }, { "epoch": 0.01, "learning_rate": 3.0487804878048782e-06, "loss": 0.4669, "step": 70 }, { "epoch": 0.01, "learning_rate": 3.0923344947735194e-06, "loss": 0.4744, "step": 71 }, { "epoch": 0.01, "learning_rate": 3.1358885017421605e-06, "loss": 0.4597, "step": 72 }, { "epoch": 0.01, "learning_rate": 3.1794425087108017e-06, "loss": 0.4591, "step": 73 }, { "epoch": 0.01, "learning_rate": 3.2229965156794425e-06, "loss": 0.4763, "step": 74 }, { "epoch": 0.01, "learning_rate": 3.266550522648084e-06, "loss": 0.4632, "step": 75 }, { "epoch": 0.01, "learning_rate": 3.3101045296167248e-06, "loss": 0.4684, "step": 76 }, { "epoch": 0.01, "learning_rate": 3.3536585365853664e-06, "loss": 0.4632, "step": 77 }, { "epoch": 0.01, "learning_rate": 3.397212543554007e-06, "loss": 0.4985, "step": 78 }, { "epoch": 0.01, "learning_rate": 3.4407665505226483e-06, "loss": 0.4618, "step": 79 }, { "epoch": 0.01, "learning_rate": 3.484320557491289e-06, "loss": 0.4556, "step": 80 }, { "epoch": 0.01, "learning_rate": 3.5278745644599306e-06, "loss": 0.489, "step": 81 }, { "epoch": 0.01, "learning_rate": 3.5714285714285714e-06, "loss": 0.4752, "step": 82 }, { "epoch": 0.01, "learning_rate": 3.614982578397213e-06, "loss": 0.4538, "step": 83 }, { "epoch": 0.01, "learning_rate": 3.6585365853658537e-06, "loss": 0.4833, "step": 84 }, { "epoch": 0.01, "learning_rate": 3.7020905923344953e-06, "loss": 0.4764, "step": 85 }, { "epoch": 0.01, "learning_rate": 3.745644599303136e-06, "loss": 0.476, "step": 86 }, { "epoch": 0.02, "learning_rate": 3.789198606271777e-06, "loss": 0.4763, "step": 87 }, { "epoch": 0.02, "learning_rate": 3.832752613240418e-06, "loss": 0.4633, "step": 88 }, { "epoch": 0.02, "learning_rate": 3.876306620209059e-06, "loss": 0.4433, "step": 89 }, { "epoch": 0.02, "learning_rate": 3.9198606271777e-06, "loss": 0.4548, "step": 90 }, { "epoch": 0.02, "learning_rate": 3.9634146341463414e-06, "loss": 0.4502, "step": 91 }, { "epoch": 0.02, "learning_rate": 4.006968641114982e-06, "loss": 0.4411, "step": 92 }, { "epoch": 0.02, "learning_rate": 4.050522648083624e-06, "loss": 0.4553, "step": 93 }, { "epoch": 0.02, "learning_rate": 4.0940766550522645e-06, "loss": 0.4643, "step": 94 }, { "epoch": 0.02, "learning_rate": 4.137630662020906e-06, "loss": 0.452, "step": 95 }, { "epoch": 0.02, "learning_rate": 4.181184668989548e-06, "loss": 0.4791, "step": 96 }, { "epoch": 0.02, "learning_rate": 4.224738675958188e-06, "loss": 0.45, "step": 97 }, { "epoch": 0.02, "learning_rate": 4.26829268292683e-06, "loss": 0.4543, "step": 98 }, { "epoch": 0.02, "learning_rate": 4.311846689895471e-06, "loss": 0.4707, "step": 99 }, { "epoch": 0.02, "learning_rate": 4.355400696864112e-06, "loss": 0.4644, "step": 100 }, { "epoch": 0.02, "learning_rate": 4.398954703832753e-06, "loss": 0.4475, "step": 101 }, { "epoch": 0.02, "learning_rate": 4.442508710801394e-06, "loss": 0.4502, "step": 102 }, { "epoch": 0.02, "learning_rate": 4.4860627177700346e-06, "loss": 0.4455, "step": 103 }, { "epoch": 0.02, "learning_rate": 4.529616724738676e-06, "loss": 0.4411, "step": 104 }, { "epoch": 0.02, "learning_rate": 4.573170731707317e-06, "loss": 0.4704, "step": 105 }, { "epoch": 0.02, "learning_rate": 4.6167247386759585e-06, "loss": 0.4671, "step": 106 }, { "epoch": 0.02, "learning_rate": 4.660278745644599e-06, "loss": 0.4591, "step": 107 }, { "epoch": 0.02, "learning_rate": 4.703832752613241e-06, "loss": 0.4728, "step": 108 }, { "epoch": 0.02, "learning_rate": 4.7473867595818816e-06, "loss": 0.479, "step": 109 }, { "epoch": 0.02, "learning_rate": 4.790940766550523e-06, "loss": 0.4571, "step": 110 }, { "epoch": 0.02, "learning_rate": 4.834494773519164e-06, "loss": 0.4594, "step": 111 }, { "epoch": 0.02, "learning_rate": 4.8780487804878055e-06, "loss": 0.4476, "step": 112 }, { "epoch": 0.02, "learning_rate": 4.921602787456446e-06, "loss": 0.4434, "step": 113 }, { "epoch": 0.02, "learning_rate": 4.965156794425088e-06, "loss": 0.4675, "step": 114 }, { "epoch": 0.02, "learning_rate": 5.0087108013937286e-06, "loss": 0.428, "step": 115 }, { "epoch": 0.02, "learning_rate": 5.052264808362369e-06, "loss": 0.4514, "step": 116 }, { "epoch": 0.02, "learning_rate": 5.09581881533101e-06, "loss": 0.4565, "step": 117 }, { "epoch": 0.02, "learning_rate": 5.139372822299652e-06, "loss": 0.4395, "step": 118 }, { "epoch": 0.02, "learning_rate": 5.182926829268292e-06, "loss": 0.4549, "step": 119 }, { "epoch": 0.02, "learning_rate": 5.226480836236934e-06, "loss": 0.4613, "step": 120 }, { "epoch": 0.02, "learning_rate": 5.270034843205575e-06, "loss": 0.433, "step": 121 }, { "epoch": 0.02, "learning_rate": 5.313588850174216e-06, "loss": 0.4411, "step": 122 }, { "epoch": 0.02, "learning_rate": 5.357142857142857e-06, "loss": 0.4299, "step": 123 }, { "epoch": 0.02, "learning_rate": 5.400696864111499e-06, "loss": 0.453, "step": 124 }, { "epoch": 0.02, "learning_rate": 5.444250871080139e-06, "loss": 0.4609, "step": 125 }, { "epoch": 0.02, "learning_rate": 5.487804878048781e-06, "loss": 0.4577, "step": 126 }, { "epoch": 0.02, "learning_rate": 5.531358885017422e-06, "loss": 0.4571, "step": 127 }, { "epoch": 0.02, "learning_rate": 5.574912891986063e-06, "loss": 0.4483, "step": 128 }, { "epoch": 0.02, "learning_rate": 5.618466898954704e-06, "loss": 0.4625, "step": 129 }, { "epoch": 0.02, "learning_rate": 5.662020905923345e-06, "loss": 0.4591, "step": 130 }, { "epoch": 0.02, "learning_rate": 5.705574912891986e-06, "loss": 0.4528, "step": 131 }, { "epoch": 0.02, "learning_rate": 5.749128919860627e-06, "loss": 0.4452, "step": 132 }, { "epoch": 0.02, "learning_rate": 5.792682926829269e-06, "loss": 0.4513, "step": 133 }, { "epoch": 0.02, "learning_rate": 5.8362369337979094e-06, "loss": 0.4426, "step": 134 }, { "epoch": 0.02, "learning_rate": 5.879790940766551e-06, "loss": 0.4408, "step": 135 }, { "epoch": 0.02, "learning_rate": 5.923344947735192e-06, "loss": 0.4465, "step": 136 }, { "epoch": 0.02, "learning_rate": 5.966898954703833e-06, "loss": 0.4511, "step": 137 }, { "epoch": 0.02, "learning_rate": 6.010452961672474e-06, "loss": 0.4381, "step": 138 }, { "epoch": 0.02, "learning_rate": 6.054006968641116e-06, "loss": 0.4375, "step": 139 }, { "epoch": 0.02, "learning_rate": 6.0975609756097564e-06, "loss": 0.4644, "step": 140 }, { "epoch": 0.02, "learning_rate": 6.141114982578398e-06, "loss": 0.4391, "step": 141 }, { "epoch": 0.02, "learning_rate": 6.184668989547039e-06, "loss": 0.4412, "step": 142 }, { "epoch": 0.02, "learning_rate": 6.2282229965156795e-06, "loss": 0.4366, "step": 143 }, { "epoch": 0.03, "learning_rate": 6.271777003484321e-06, "loss": 0.4318, "step": 144 }, { "epoch": 0.03, "learning_rate": 6.315331010452962e-06, "loss": 0.4302, "step": 145 }, { "epoch": 0.03, "learning_rate": 6.3588850174216034e-06, "loss": 0.469, "step": 146 }, { "epoch": 0.03, "learning_rate": 6.402439024390244e-06, "loss": 0.4515, "step": 147 }, { "epoch": 0.03, "learning_rate": 6.445993031358885e-06, "loss": 0.4397, "step": 148 }, { "epoch": 0.03, "learning_rate": 6.489547038327526e-06, "loss": 0.4328, "step": 149 }, { "epoch": 0.03, "learning_rate": 6.533101045296168e-06, "loss": 0.4432, "step": 150 }, { "epoch": 0.03, "learning_rate": 6.576655052264809e-06, "loss": 0.4404, "step": 151 }, { "epoch": 0.03, "learning_rate": 6.6202090592334496e-06, "loss": 0.4377, "step": 152 }, { "epoch": 0.03, "learning_rate": 6.66376306620209e-06, "loss": 0.4416, "step": 153 }, { "epoch": 0.03, "learning_rate": 6.707317073170733e-06, "loss": 0.4317, "step": 154 }, { "epoch": 0.03, "learning_rate": 6.7508710801393735e-06, "loss": 0.4389, "step": 155 }, { "epoch": 0.03, "learning_rate": 6.794425087108014e-06, "loss": 0.4192, "step": 156 }, { "epoch": 0.03, "learning_rate": 6.837979094076655e-06, "loss": 0.437, "step": 157 }, { "epoch": 0.03, "learning_rate": 6.8815331010452966e-06, "loss": 0.4539, "step": 158 }, { "epoch": 0.03, "learning_rate": 6.925087108013937e-06, "loss": 0.4357, "step": 159 }, { "epoch": 0.03, "learning_rate": 6.968641114982578e-06, "loss": 0.4503, "step": 160 }, { "epoch": 0.03, "learning_rate": 7.0121951219512205e-06, "loss": 0.4272, "step": 161 }, { "epoch": 0.03, "learning_rate": 7.055749128919861e-06, "loss": 0.4337, "step": 162 }, { "epoch": 0.03, "learning_rate": 7.099303135888502e-06, "loss": 0.4561, "step": 163 }, { "epoch": 0.03, "learning_rate": 7.142857142857143e-06, "loss": 0.4302, "step": 164 }, { "epoch": 0.03, "learning_rate": 7.186411149825785e-06, "loss": 0.4453, "step": 165 }, { "epoch": 0.03, "learning_rate": 7.229965156794426e-06, "loss": 0.4487, "step": 166 }, { "epoch": 0.03, "learning_rate": 7.273519163763067e-06, "loss": 0.4287, "step": 167 }, { "epoch": 0.03, "learning_rate": 7.317073170731707e-06, "loss": 0.4452, "step": 168 }, { "epoch": 0.03, "learning_rate": 7.360627177700349e-06, "loss": 0.4494, "step": 169 }, { "epoch": 0.03, "learning_rate": 7.4041811846689906e-06, "loss": 0.4437, "step": 170 }, { "epoch": 0.03, "learning_rate": 7.447735191637631e-06, "loss": 0.4404, "step": 171 }, { "epoch": 0.03, "learning_rate": 7.491289198606272e-06, "loss": 0.4397, "step": 172 }, { "epoch": 0.03, "learning_rate": 7.534843205574914e-06, "loss": 0.4357, "step": 173 }, { "epoch": 0.03, "learning_rate": 7.578397212543554e-06, "loss": 0.4389, "step": 174 }, { "epoch": 0.03, "learning_rate": 7.621951219512195e-06, "loss": 0.4399, "step": 175 }, { "epoch": 0.03, "learning_rate": 7.665505226480837e-06, "loss": 0.4359, "step": 176 }, { "epoch": 0.03, "learning_rate": 7.709059233449478e-06, "loss": 0.4385, "step": 177 }, { "epoch": 0.03, "learning_rate": 7.752613240418118e-06, "loss": 0.4346, "step": 178 }, { "epoch": 0.03, "learning_rate": 7.79616724738676e-06, "loss": 0.431, "step": 179 }, { "epoch": 0.03, "learning_rate": 7.8397212543554e-06, "loss": 0.4342, "step": 180 }, { "epoch": 0.03, "learning_rate": 7.883275261324043e-06, "loss": 0.4407, "step": 181 }, { "epoch": 0.03, "learning_rate": 7.926829268292683e-06, "loss": 0.4327, "step": 182 }, { "epoch": 0.03, "learning_rate": 7.970383275261324e-06, "loss": 0.4484, "step": 183 }, { "epoch": 0.03, "learning_rate": 8.013937282229964e-06, "loss": 0.4291, "step": 184 }, { "epoch": 0.03, "learning_rate": 8.057491289198608e-06, "loss": 0.4238, "step": 185 }, { "epoch": 0.03, "learning_rate": 8.101045296167248e-06, "loss": 0.4211, "step": 186 }, { "epoch": 0.03, "learning_rate": 8.144599303135889e-06, "loss": 0.4384, "step": 187 }, { "epoch": 0.03, "learning_rate": 8.188153310104529e-06, "loss": 0.4307, "step": 188 }, { "epoch": 0.03, "learning_rate": 8.231707317073172e-06, "loss": 0.439, "step": 189 }, { "epoch": 0.03, "learning_rate": 8.275261324041812e-06, "loss": 0.4359, "step": 190 }, { "epoch": 0.03, "learning_rate": 8.318815331010454e-06, "loss": 0.436, "step": 191 }, { "epoch": 0.03, "learning_rate": 8.362369337979095e-06, "loss": 0.44, "step": 192 }, { "epoch": 0.03, "learning_rate": 8.405923344947735e-06, "loss": 0.4382, "step": 193 }, { "epoch": 0.03, "learning_rate": 8.449477351916377e-06, "loss": 0.4426, "step": 194 }, { "epoch": 0.03, "learning_rate": 8.493031358885017e-06, "loss": 0.4399, "step": 195 }, { "epoch": 0.03, "learning_rate": 8.53658536585366e-06, "loss": 0.4616, "step": 196 }, { "epoch": 0.03, "learning_rate": 8.5801393728223e-06, "loss": 0.4334, "step": 197 }, { "epoch": 0.03, "learning_rate": 8.623693379790942e-06, "loss": 0.4603, "step": 198 }, { "epoch": 0.03, "learning_rate": 8.667247386759581e-06, "loss": 0.4476, "step": 199 }, { "epoch": 0.03, "learning_rate": 8.710801393728225e-06, "loss": 0.4351, "step": 200 }, { "epoch": 0.04, "learning_rate": 8.754355400696865e-06, "loss": 0.4352, "step": 201 }, { "epoch": 0.04, "learning_rate": 8.797909407665506e-06, "loss": 0.43, "step": 202 }, { "epoch": 0.04, "learning_rate": 8.841463414634146e-06, "loss": 0.4619, "step": 203 }, { "epoch": 0.04, "learning_rate": 8.885017421602788e-06, "loss": 0.4376, "step": 204 }, { "epoch": 0.04, "learning_rate": 8.92857142857143e-06, "loss": 0.4321, "step": 205 }, { "epoch": 0.04, "learning_rate": 8.972125435540069e-06, "loss": 0.4279, "step": 206 }, { "epoch": 0.04, "learning_rate": 9.01567944250871e-06, "loss": 0.4408, "step": 207 }, { "epoch": 0.04, "learning_rate": 9.059233449477352e-06, "loss": 0.4308, "step": 208 }, { "epoch": 0.04, "learning_rate": 9.102787456445994e-06, "loss": 0.4193, "step": 209 }, { "epoch": 0.04, "learning_rate": 9.146341463414634e-06, "loss": 0.4433, "step": 210 }, { "epoch": 0.04, "learning_rate": 9.189895470383275e-06, "loss": 0.4136, "step": 211 }, { "epoch": 0.04, "learning_rate": 9.233449477351917e-06, "loss": 0.4335, "step": 212 }, { "epoch": 0.04, "learning_rate": 9.277003484320559e-06, "loss": 0.4256, "step": 213 }, { "epoch": 0.04, "learning_rate": 9.320557491289198e-06, "loss": 0.4335, "step": 214 }, { "epoch": 0.04, "learning_rate": 9.36411149825784e-06, "loss": 0.431, "step": 215 }, { "epoch": 0.04, "learning_rate": 9.407665505226482e-06, "loss": 0.4198, "step": 216 }, { "epoch": 0.04, "learning_rate": 9.451219512195123e-06, "loss": 0.4427, "step": 217 }, { "epoch": 0.04, "learning_rate": 9.494773519163763e-06, "loss": 0.4353, "step": 218 }, { "epoch": 0.04, "learning_rate": 9.538327526132405e-06, "loss": 0.4221, "step": 219 }, { "epoch": 0.04, "learning_rate": 9.581881533101046e-06, "loss": 0.437, "step": 220 }, { "epoch": 0.04, "learning_rate": 9.625435540069686e-06, "loss": 0.4318, "step": 221 }, { "epoch": 0.04, "learning_rate": 9.668989547038328e-06, "loss": 0.4323, "step": 222 }, { "epoch": 0.04, "learning_rate": 9.712543554006968e-06, "loss": 0.441, "step": 223 }, { "epoch": 0.04, "learning_rate": 9.756097560975611e-06, "loss": 0.4332, "step": 224 }, { "epoch": 0.04, "learning_rate": 9.799651567944251e-06, "loss": 0.4211, "step": 225 }, { "epoch": 0.04, "learning_rate": 9.843205574912892e-06, "loss": 0.4356, "step": 226 }, { "epoch": 0.04, "learning_rate": 9.886759581881534e-06, "loss": 0.4221, "step": 227 }, { "epoch": 0.04, "learning_rate": 9.930313588850176e-06, "loss": 0.4127, "step": 228 }, { "epoch": 0.04, "learning_rate": 9.973867595818816e-06, "loss": 0.4493, "step": 229 }, { "epoch": 0.04, "learning_rate": 1.0017421602787457e-05, "loss": 0.4278, "step": 230 }, { "epoch": 0.04, "learning_rate": 1.0060975609756099e-05, "loss": 0.4393, "step": 231 }, { "epoch": 0.04, "learning_rate": 1.0104529616724739e-05, "loss": 0.4343, "step": 232 }, { "epoch": 0.04, "learning_rate": 1.014808362369338e-05, "loss": 0.4237, "step": 233 }, { "epoch": 0.04, "learning_rate": 1.019163763066202e-05, "loss": 0.4429, "step": 234 }, { "epoch": 0.04, "learning_rate": 1.0235191637630663e-05, "loss": 0.4247, "step": 235 }, { "epoch": 0.04, "learning_rate": 1.0278745644599303e-05, "loss": 0.4498, "step": 236 }, { "epoch": 0.04, "learning_rate": 1.0322299651567945e-05, "loss": 0.429, "step": 237 }, { "epoch": 0.04, "learning_rate": 1.0365853658536585e-05, "loss": 0.4231, "step": 238 }, { "epoch": 0.04, "learning_rate": 1.0409407665505228e-05, "loss": 0.4137, "step": 239 }, { "epoch": 0.04, "learning_rate": 1.0452961672473868e-05, "loss": 0.4395, "step": 240 }, { "epoch": 0.04, "learning_rate": 1.049651567944251e-05, "loss": 0.4268, "step": 241 }, { "epoch": 0.04, "learning_rate": 1.054006968641115e-05, "loss": 0.4474, "step": 242 }, { "epoch": 0.04, "learning_rate": 1.0583623693379793e-05, "loss": 0.4382, "step": 243 }, { "epoch": 0.04, "learning_rate": 1.0627177700348433e-05, "loss": 0.4342, "step": 244 }, { "epoch": 0.04, "learning_rate": 1.0670731707317074e-05, "loss": 0.4261, "step": 245 }, { "epoch": 0.04, "learning_rate": 1.0714285714285714e-05, "loss": 0.4199, "step": 246 }, { "epoch": 0.04, "learning_rate": 1.0757839721254356e-05, "loss": 0.4145, "step": 247 }, { "epoch": 0.04, "learning_rate": 1.0801393728222997e-05, "loss": 0.4328, "step": 248 }, { "epoch": 0.04, "learning_rate": 1.0844947735191637e-05, "loss": 0.436, "step": 249 }, { "epoch": 0.04, "learning_rate": 1.0888501742160279e-05, "loss": 0.4286, "step": 250 }, { "epoch": 0.04, "learning_rate": 1.093205574912892e-05, "loss": 0.4451, "step": 251 }, { "epoch": 0.04, "learning_rate": 1.0975609756097562e-05, "loss": 0.4293, "step": 252 }, { "epoch": 0.04, "learning_rate": 1.1019163763066202e-05, "loss": 0.4222, "step": 253 }, { "epoch": 0.04, "learning_rate": 1.1062717770034843e-05, "loss": 0.4214, "step": 254 }, { "epoch": 0.04, "learning_rate": 1.1106271777003485e-05, "loss": 0.448, "step": 255 }, { "epoch": 0.04, "learning_rate": 1.1149825783972127e-05, "loss": 0.4143, "step": 256 }, { "epoch": 0.04, "learning_rate": 1.1193379790940766e-05, "loss": 0.4249, "step": 257 }, { "epoch": 0.04, "learning_rate": 1.1236933797909408e-05, "loss": 0.4186, "step": 258 }, { "epoch": 0.05, "learning_rate": 1.128048780487805e-05, "loss": 0.4121, "step": 259 }, { "epoch": 0.05, "learning_rate": 1.132404181184669e-05, "loss": 0.4464, "step": 260 }, { "epoch": 0.05, "learning_rate": 1.1367595818815331e-05, "loss": 0.4578, "step": 261 }, { "epoch": 0.05, "learning_rate": 1.1411149825783973e-05, "loss": 0.4527, "step": 262 }, { "epoch": 0.05, "learning_rate": 1.1454703832752614e-05, "loss": 0.4182, "step": 263 }, { "epoch": 0.05, "learning_rate": 1.1498257839721254e-05, "loss": 0.4069, "step": 264 }, { "epoch": 0.05, "learning_rate": 1.1541811846689896e-05, "loss": 0.438, "step": 265 }, { "epoch": 0.05, "learning_rate": 1.1585365853658537e-05, "loss": 0.4317, "step": 266 }, { "epoch": 0.05, "learning_rate": 1.1628919860627179e-05, "loss": 0.4373, "step": 267 }, { "epoch": 0.05, "learning_rate": 1.1672473867595819e-05, "loss": 0.4301, "step": 268 }, { "epoch": 0.05, "learning_rate": 1.171602787456446e-05, "loss": 0.4388, "step": 269 }, { "epoch": 0.05, "learning_rate": 1.1759581881533102e-05, "loss": 0.4284, "step": 270 }, { "epoch": 0.05, "learning_rate": 1.1803135888501744e-05, "loss": 0.4156, "step": 271 }, { "epoch": 0.05, "learning_rate": 1.1846689895470384e-05, "loss": 0.4269, "step": 272 }, { "epoch": 0.05, "learning_rate": 1.1890243902439025e-05, "loss": 0.4209, "step": 273 }, { "epoch": 0.05, "learning_rate": 1.1933797909407667e-05, "loss": 0.4223, "step": 274 }, { "epoch": 0.05, "learning_rate": 1.1977351916376307e-05, "loss": 0.4238, "step": 275 }, { "epoch": 0.05, "learning_rate": 1.2020905923344948e-05, "loss": 0.4223, "step": 276 }, { "epoch": 0.05, "learning_rate": 1.2064459930313588e-05, "loss": 0.4135, "step": 277 }, { "epoch": 0.05, "learning_rate": 1.2108013937282231e-05, "loss": 0.4279, "step": 278 }, { "epoch": 0.05, "learning_rate": 1.2151567944250871e-05, "loss": 0.4214, "step": 279 }, { "epoch": 0.05, "learning_rate": 1.2195121951219513e-05, "loss": 0.4337, "step": 280 }, { "epoch": 0.05, "learning_rate": 1.2238675958188153e-05, "loss": 0.4331, "step": 281 }, { "epoch": 0.05, "learning_rate": 1.2282229965156796e-05, "loss": 0.436, "step": 282 }, { "epoch": 0.05, "learning_rate": 1.2325783972125436e-05, "loss": 0.4177, "step": 283 }, { "epoch": 0.05, "learning_rate": 1.2369337979094078e-05, "loss": 0.4265, "step": 284 }, { "epoch": 0.05, "learning_rate": 1.2412891986062717e-05, "loss": 0.425, "step": 285 }, { "epoch": 0.05, "learning_rate": 1.2456445993031359e-05, "loss": 0.4299, "step": 286 }, { "epoch": 0.05, "learning_rate": 1.25e-05, "loss": 0.4191, "step": 287 }, { "epoch": 0.05, "learning_rate": 1.2543554006968642e-05, "loss": 0.4255, "step": 288 }, { "epoch": 0.05, "learning_rate": 1.2587108013937282e-05, "loss": 0.4405, "step": 289 }, { "epoch": 0.05, "learning_rate": 1.2630662020905924e-05, "loss": 0.428, "step": 290 }, { "epoch": 0.05, "learning_rate": 1.2674216027874567e-05, "loss": 0.4147, "step": 291 }, { "epoch": 0.05, "learning_rate": 1.2717770034843207e-05, "loss": 0.4155, "step": 292 }, { "epoch": 0.05, "learning_rate": 1.2761324041811848e-05, "loss": 0.4308, "step": 293 }, { "epoch": 0.05, "learning_rate": 1.2804878048780488e-05, "loss": 0.4158, "step": 294 }, { "epoch": 0.05, "learning_rate": 1.284843205574913e-05, "loss": 0.4331, "step": 295 }, { "epoch": 0.05, "learning_rate": 1.289198606271777e-05, "loss": 0.4377, "step": 296 }, { "epoch": 0.05, "learning_rate": 1.2935540069686411e-05, "loss": 0.4541, "step": 297 }, { "epoch": 0.05, "learning_rate": 1.2979094076655051e-05, "loss": 0.4124, "step": 298 }, { "epoch": 0.05, "learning_rate": 1.3022648083623695e-05, "loss": 0.4228, "step": 299 }, { "epoch": 0.05, "learning_rate": 1.3066202090592336e-05, "loss": 0.4159, "step": 300 }, { "epoch": 0.05, "learning_rate": 1.3109756097560976e-05, "loss": 0.4227, "step": 301 }, { "epoch": 0.05, "learning_rate": 1.3153310104529618e-05, "loss": 0.4199, "step": 302 }, { "epoch": 0.05, "learning_rate": 1.3196864111498258e-05, "loss": 0.4164, "step": 303 }, { "epoch": 0.05, "learning_rate": 1.3240418118466899e-05, "loss": 0.4132, "step": 304 }, { "epoch": 0.05, "learning_rate": 1.3283972125435539e-05, "loss": 0.4362, "step": 305 }, { "epoch": 0.05, "learning_rate": 1.332752613240418e-05, "loss": 0.4284, "step": 306 }, { "epoch": 0.05, "learning_rate": 1.3371080139372824e-05, "loss": 0.4187, "step": 307 }, { "epoch": 0.05, "learning_rate": 1.3414634146341466e-05, "loss": 0.4279, "step": 308 }, { "epoch": 0.05, "learning_rate": 1.3458188153310105e-05, "loss": 0.41, "step": 309 }, { "epoch": 0.05, "learning_rate": 1.3501742160278747e-05, "loss": 0.4312, "step": 310 }, { "epoch": 0.05, "learning_rate": 1.3545296167247387e-05, "loss": 0.4259, "step": 311 }, { "epoch": 0.05, "learning_rate": 1.3588850174216028e-05, "loss": 0.4263, "step": 312 }, { "epoch": 0.05, "learning_rate": 1.3632404181184668e-05, "loss": 0.4101, "step": 313 }, { "epoch": 0.05, "learning_rate": 1.367595818815331e-05, "loss": 0.4207, "step": 314 }, { "epoch": 0.05, "learning_rate": 1.3719512195121953e-05, "loss": 0.4243, "step": 315 }, { "epoch": 0.06, "learning_rate": 1.3763066202090593e-05, "loss": 0.417, "step": 316 }, { "epoch": 0.06, "learning_rate": 1.3806620209059235e-05, "loss": 0.4507, "step": 317 }, { "epoch": 0.06, "learning_rate": 1.3850174216027875e-05, "loss": 0.4323, "step": 318 }, { "epoch": 0.06, "learning_rate": 1.3893728222996516e-05, "loss": 0.4072, "step": 319 }, { "epoch": 0.06, "learning_rate": 1.3937282229965156e-05, "loss": 0.4414, "step": 320 }, { "epoch": 0.06, "learning_rate": 1.3980836236933798e-05, "loss": 0.4075, "step": 321 }, { "epoch": 0.06, "learning_rate": 1.4024390243902441e-05, "loss": 0.4243, "step": 322 }, { "epoch": 0.06, "learning_rate": 1.4067944250871083e-05, "loss": 0.4224, "step": 323 }, { "epoch": 0.06, "learning_rate": 1.4111498257839722e-05, "loss": 0.427, "step": 324 }, { "epoch": 0.06, "learning_rate": 1.4155052264808364e-05, "loss": 0.4373, "step": 325 }, { "epoch": 0.06, "learning_rate": 1.4198606271777004e-05, "loss": 0.4219, "step": 326 }, { "epoch": 0.06, "learning_rate": 1.4242160278745646e-05, "loss": 0.4111, "step": 327 }, { "epoch": 0.06, "learning_rate": 1.4285714285714285e-05, "loss": 0.4202, "step": 328 }, { "epoch": 0.06, "learning_rate": 1.4329268292682927e-05, "loss": 0.4429, "step": 329 }, { "epoch": 0.06, "learning_rate": 1.437282229965157e-05, "loss": 0.4317, "step": 330 }, { "epoch": 0.06, "learning_rate": 1.441637630662021e-05, "loss": 0.4027, "step": 331 }, { "epoch": 0.06, "learning_rate": 1.4459930313588852e-05, "loss": 0.4328, "step": 332 }, { "epoch": 0.06, "learning_rate": 1.4503484320557492e-05, "loss": 0.4154, "step": 333 }, { "epoch": 0.06, "learning_rate": 1.4547038327526133e-05, "loss": 0.4258, "step": 334 }, { "epoch": 0.06, "learning_rate": 1.4590592334494773e-05, "loss": 0.4162, "step": 335 }, { "epoch": 0.06, "learning_rate": 1.4634146341463415e-05, "loss": 0.4219, "step": 336 }, { "epoch": 0.06, "learning_rate": 1.4677700348432055e-05, "loss": 0.4198, "step": 337 }, { "epoch": 0.06, "learning_rate": 1.4721254355400698e-05, "loss": 0.4223, "step": 338 }, { "epoch": 0.06, "learning_rate": 1.476480836236934e-05, "loss": 0.4213, "step": 339 }, { "epoch": 0.06, "learning_rate": 1.4808362369337981e-05, "loss": 0.4218, "step": 340 }, { "epoch": 0.06, "learning_rate": 1.4851916376306621e-05, "loss": 0.4242, "step": 341 }, { "epoch": 0.06, "learning_rate": 1.4895470383275263e-05, "loss": 0.4256, "step": 342 }, { "epoch": 0.06, "learning_rate": 1.4939024390243902e-05, "loss": 0.4416, "step": 343 }, { "epoch": 0.06, "learning_rate": 1.4982578397212544e-05, "loss": 0.4242, "step": 344 }, { "epoch": 0.06, "learning_rate": 1.5026132404181184e-05, "loss": 0.4567, "step": 345 }, { "epoch": 0.06, "learning_rate": 1.5069686411149827e-05, "loss": 0.4173, "step": 346 }, { "epoch": 0.06, "learning_rate": 1.5113240418118469e-05, "loss": 0.4474, "step": 347 }, { "epoch": 0.06, "learning_rate": 1.5156794425087109e-05, "loss": 0.4138, "step": 348 }, { "epoch": 0.06, "learning_rate": 1.520034843205575e-05, "loss": 0.4235, "step": 349 }, { "epoch": 0.06, "learning_rate": 1.524390243902439e-05, "loss": 0.4154, "step": 350 }, { "epoch": 0.06, "learning_rate": 1.528745644599303e-05, "loss": 0.4361, "step": 351 }, { "epoch": 0.06, "learning_rate": 1.5331010452961673e-05, "loss": 0.4211, "step": 352 }, { "epoch": 0.06, "learning_rate": 1.5374564459930317e-05, "loss": 0.4346, "step": 353 }, { "epoch": 0.06, "learning_rate": 1.5418118466898957e-05, "loss": 0.4266, "step": 354 }, { "epoch": 0.06, "learning_rate": 1.5461672473867596e-05, "loss": 0.4141, "step": 355 }, { "epoch": 0.06, "learning_rate": 1.5505226480836236e-05, "loss": 0.4292, "step": 356 }, { "epoch": 0.06, "learning_rate": 1.554878048780488e-05, "loss": 0.4573, "step": 357 }, { "epoch": 0.06, "learning_rate": 1.559233449477352e-05, "loss": 0.4135, "step": 358 }, { "epoch": 0.06, "learning_rate": 1.563588850174216e-05, "loss": 0.4253, "step": 359 }, { "epoch": 0.06, "learning_rate": 1.56794425087108e-05, "loss": 0.4303, "step": 360 }, { "epoch": 0.06, "learning_rate": 1.5722996515679446e-05, "loss": 0.4356, "step": 361 }, { "epoch": 0.06, "learning_rate": 1.5766550522648086e-05, "loss": 0.422, "step": 362 }, { "epoch": 0.06, "learning_rate": 1.5810104529616726e-05, "loss": 0.4284, "step": 363 }, { "epoch": 0.06, "learning_rate": 1.5853658536585366e-05, "loss": 0.4243, "step": 364 }, { "epoch": 0.06, "learning_rate": 1.589721254355401e-05, "loss": 0.4221, "step": 365 }, { "epoch": 0.06, "learning_rate": 1.594076655052265e-05, "loss": 0.4196, "step": 366 }, { "epoch": 0.06, "learning_rate": 1.598432055749129e-05, "loss": 0.4204, "step": 367 }, { "epoch": 0.06, "learning_rate": 1.602787456445993e-05, "loss": 0.4216, "step": 368 }, { "epoch": 0.06, "learning_rate": 1.6071428571428572e-05, "loss": 0.4145, "step": 369 }, { "epoch": 0.06, "learning_rate": 1.6114982578397215e-05, "loss": 0.4206, "step": 370 }, { "epoch": 0.06, "learning_rate": 1.6158536585365855e-05, "loss": 0.4309, "step": 371 }, { "epoch": 0.06, "learning_rate": 1.6202090592334495e-05, "loss": 0.3972, "step": 372 }, { "epoch": 0.07, "learning_rate": 1.6245644599303135e-05, "loss": 0.4309, "step": 373 }, { "epoch": 0.07, "learning_rate": 1.6289198606271778e-05, "loss": 0.4166, "step": 374 }, { "epoch": 0.07, "learning_rate": 1.6332752613240418e-05, "loss": 0.4183, "step": 375 }, { "epoch": 0.07, "learning_rate": 1.6376306620209058e-05, "loss": 0.4109, "step": 376 }, { "epoch": 0.07, "learning_rate": 1.64198606271777e-05, "loss": 0.4144, "step": 377 }, { "epoch": 0.07, "learning_rate": 1.6463414634146345e-05, "loss": 0.4046, "step": 378 }, { "epoch": 0.07, "learning_rate": 1.6506968641114984e-05, "loss": 0.4199, "step": 379 }, { "epoch": 0.07, "learning_rate": 1.6550522648083624e-05, "loss": 0.4072, "step": 380 }, { "epoch": 0.07, "learning_rate": 1.6594076655052264e-05, "loss": 0.4035, "step": 381 }, { "epoch": 0.07, "learning_rate": 1.6637630662020908e-05, "loss": 0.4251, "step": 382 }, { "epoch": 0.07, "learning_rate": 1.6681184668989547e-05, "loss": 0.4247, "step": 383 }, { "epoch": 0.07, "learning_rate": 1.672473867595819e-05, "loss": 0.4314, "step": 384 }, { "epoch": 0.07, "learning_rate": 1.676829268292683e-05, "loss": 0.4075, "step": 385 }, { "epoch": 0.07, "learning_rate": 1.681184668989547e-05, "loss": 0.4153, "step": 386 }, { "epoch": 0.07, "learning_rate": 1.6855400696864114e-05, "loss": 0.4333, "step": 387 }, { "epoch": 0.07, "learning_rate": 1.6898954703832754e-05, "loss": 0.4166, "step": 388 }, { "epoch": 0.07, "learning_rate": 1.6942508710801394e-05, "loss": 0.4193, "step": 389 }, { "epoch": 0.07, "learning_rate": 1.6986062717770033e-05, "loss": 0.3925, "step": 390 }, { "epoch": 0.07, "learning_rate": 1.7029616724738677e-05, "loss": 0.4353, "step": 391 }, { "epoch": 0.07, "learning_rate": 1.707317073170732e-05, "loss": 0.43, "step": 392 }, { "epoch": 0.07, "learning_rate": 1.711672473867596e-05, "loss": 0.4063, "step": 393 }, { "epoch": 0.07, "learning_rate": 1.71602787456446e-05, "loss": 0.4124, "step": 394 }, { "epoch": 0.07, "learning_rate": 1.7203832752613243e-05, "loss": 0.4117, "step": 395 }, { "epoch": 0.07, "learning_rate": 1.7247386759581883e-05, "loss": 0.4378, "step": 396 }, { "epoch": 0.07, "learning_rate": 1.7290940766550523e-05, "loss": 0.4211, "step": 397 }, { "epoch": 0.07, "learning_rate": 1.7334494773519163e-05, "loss": 0.4108, "step": 398 }, { "epoch": 0.07, "learning_rate": 1.7378048780487806e-05, "loss": 0.4259, "step": 399 }, { "epoch": 0.07, "learning_rate": 1.742160278745645e-05, "loss": 0.4126, "step": 400 }, { "epoch": 0.07, "learning_rate": 1.746515679442509e-05, "loss": 0.4062, "step": 401 }, { "epoch": 0.07, "learning_rate": 1.750871080139373e-05, "loss": 0.4099, "step": 402 }, { "epoch": 0.07, "learning_rate": 1.755226480836237e-05, "loss": 0.4214, "step": 403 }, { "epoch": 0.07, "learning_rate": 1.7595818815331012e-05, "loss": 0.4128, "step": 404 }, { "epoch": 0.07, "learning_rate": 1.7639372822299652e-05, "loss": 0.4044, "step": 405 }, { "epoch": 0.07, "learning_rate": 1.7682926829268292e-05, "loss": 0.4213, "step": 406 }, { "epoch": 0.07, "learning_rate": 1.7726480836236932e-05, "loss": 0.4149, "step": 407 }, { "epoch": 0.07, "learning_rate": 1.7770034843205575e-05, "loss": 0.4214, "step": 408 }, { "epoch": 0.07, "learning_rate": 1.781358885017422e-05, "loss": 0.4303, "step": 409 }, { "epoch": 0.07, "learning_rate": 1.785714285714286e-05, "loss": 0.4109, "step": 410 }, { "epoch": 0.07, "learning_rate": 1.79006968641115e-05, "loss": 0.4394, "step": 411 }, { "epoch": 0.07, "learning_rate": 1.7944250871080138e-05, "loss": 0.4274, "step": 412 }, { "epoch": 0.07, "learning_rate": 1.798780487804878e-05, "loss": 0.4141, "step": 413 }, { "epoch": 0.07, "learning_rate": 1.803135888501742e-05, "loss": 0.4235, "step": 414 }, { "epoch": 0.07, "learning_rate": 1.8074912891986065e-05, "loss": 0.4152, "step": 415 }, { "epoch": 0.07, "learning_rate": 1.8118466898954705e-05, "loss": 0.4149, "step": 416 }, { "epoch": 0.07, "learning_rate": 1.8162020905923348e-05, "loss": 0.4084, "step": 417 }, { "epoch": 0.07, "learning_rate": 1.8205574912891988e-05, "loss": 0.419, "step": 418 }, { "epoch": 0.07, "learning_rate": 1.8249128919860628e-05, "loss": 0.4105, "step": 419 }, { "epoch": 0.07, "learning_rate": 1.8292682926829268e-05, "loss": 0.4093, "step": 420 }, { "epoch": 0.07, "learning_rate": 1.833623693379791e-05, "loss": 0.4323, "step": 421 }, { "epoch": 0.07, "learning_rate": 1.837979094076655e-05, "loss": 0.4211, "step": 422 }, { "epoch": 0.07, "learning_rate": 1.8423344947735194e-05, "loss": 0.4263, "step": 423 }, { "epoch": 0.07, "learning_rate": 1.8466898954703834e-05, "loss": 0.4203, "step": 424 }, { "epoch": 0.07, "learning_rate": 1.8510452961672474e-05, "loss": 0.4132, "step": 425 }, { "epoch": 0.07, "learning_rate": 1.8554006968641117e-05, "loss": 0.4078, "step": 426 }, { "epoch": 0.07, "learning_rate": 1.8597560975609757e-05, "loss": 0.402, "step": 427 }, { "epoch": 0.07, "learning_rate": 1.8641114982578397e-05, "loss": 0.418, "step": 428 }, { "epoch": 0.07, "learning_rate": 1.8684668989547037e-05, "loss": 0.4091, "step": 429 }, { "epoch": 0.07, "learning_rate": 1.872822299651568e-05, "loss": 0.4115, "step": 430 }, { "epoch": 0.08, "learning_rate": 1.8771777003484323e-05, "loss": 0.418, "step": 431 }, { "epoch": 0.08, "learning_rate": 1.8815331010452963e-05, "loss": 0.424, "step": 432 }, { "epoch": 0.08, "learning_rate": 1.8858885017421603e-05, "loss": 0.4326, "step": 433 }, { "epoch": 0.08, "learning_rate": 1.8902439024390246e-05, "loss": 0.423, "step": 434 }, { "epoch": 0.08, "learning_rate": 1.8945993031358886e-05, "loss": 0.4227, "step": 435 }, { "epoch": 0.08, "learning_rate": 1.8989547038327526e-05, "loss": 0.4177, "step": 436 }, { "epoch": 0.08, "learning_rate": 1.9033101045296166e-05, "loss": 0.402, "step": 437 }, { "epoch": 0.08, "learning_rate": 1.907665505226481e-05, "loss": 0.4179, "step": 438 }, { "epoch": 0.08, "learning_rate": 1.9120209059233453e-05, "loss": 0.4253, "step": 439 }, { "epoch": 0.08, "learning_rate": 1.9163763066202093e-05, "loss": 0.4287, "step": 440 }, { "epoch": 0.08, "learning_rate": 1.9207317073170733e-05, "loss": 0.3982, "step": 441 }, { "epoch": 0.08, "learning_rate": 1.9250871080139372e-05, "loss": 0.4234, "step": 442 }, { "epoch": 0.08, "learning_rate": 1.9294425087108016e-05, "loss": 0.4259, "step": 443 }, { "epoch": 0.08, "learning_rate": 1.9337979094076656e-05, "loss": 0.4228, "step": 444 }, { "epoch": 0.08, "learning_rate": 1.9381533101045295e-05, "loss": 0.4262, "step": 445 }, { "epoch": 0.08, "learning_rate": 1.9425087108013935e-05, "loss": 0.4234, "step": 446 }, { "epoch": 0.08, "learning_rate": 1.9468641114982582e-05, "loss": 0.4048, "step": 447 }, { "epoch": 0.08, "learning_rate": 1.9512195121951222e-05, "loss": 0.417, "step": 448 }, { "epoch": 0.08, "learning_rate": 1.9555749128919862e-05, "loss": 0.3964, "step": 449 }, { "epoch": 0.08, "learning_rate": 1.9599303135888502e-05, "loss": 0.4418, "step": 450 }, { "epoch": 0.08, "learning_rate": 1.9642857142857145e-05, "loss": 0.4179, "step": 451 }, { "epoch": 0.08, "learning_rate": 1.9686411149825785e-05, "loss": 0.4249, "step": 452 }, { "epoch": 0.08, "learning_rate": 1.9729965156794425e-05, "loss": 0.4141, "step": 453 }, { "epoch": 0.08, "learning_rate": 1.9773519163763068e-05, "loss": 0.4066, "step": 454 }, { "epoch": 0.08, "learning_rate": 1.9817073170731708e-05, "loss": 0.3997, "step": 455 }, { "epoch": 0.08, "learning_rate": 1.986062717770035e-05, "loss": 0.4056, "step": 456 }, { "epoch": 0.08, "learning_rate": 1.990418118466899e-05, "loss": 0.4261, "step": 457 }, { "epoch": 0.08, "learning_rate": 1.994773519163763e-05, "loss": 0.4289, "step": 458 }, { "epoch": 0.08, "learning_rate": 1.999128919860627e-05, "loss": 0.4291, "step": 459 }, { "epoch": 0.08, "learning_rate": 2.0034843205574914e-05, "loss": 0.4133, "step": 460 }, { "epoch": 0.08, "learning_rate": 2.0078397212543554e-05, "loss": 0.418, "step": 461 }, { "epoch": 0.08, "learning_rate": 2.0121951219512197e-05, "loss": 0.4288, "step": 462 }, { "epoch": 0.08, "learning_rate": 2.0165505226480837e-05, "loss": 0.4407, "step": 463 }, { "epoch": 0.08, "learning_rate": 2.0209059233449477e-05, "loss": 0.4214, "step": 464 }, { "epoch": 0.08, "learning_rate": 2.025261324041812e-05, "loss": 0.4333, "step": 465 }, { "epoch": 0.08, "learning_rate": 2.029616724738676e-05, "loss": 0.4209, "step": 466 }, { "epoch": 0.08, "learning_rate": 2.03397212543554e-05, "loss": 0.4003, "step": 467 }, { "epoch": 0.08, "learning_rate": 2.038327526132404e-05, "loss": 0.4122, "step": 468 }, { "epoch": 0.08, "learning_rate": 2.0426829268292683e-05, "loss": 0.4145, "step": 469 }, { "epoch": 0.08, "learning_rate": 2.0470383275261327e-05, "loss": 0.4205, "step": 470 }, { "epoch": 0.08, "learning_rate": 2.0513937282229967e-05, "loss": 0.4152, "step": 471 }, { "epoch": 0.08, "learning_rate": 2.0557491289198607e-05, "loss": 0.4251, "step": 472 }, { "epoch": 0.08, "learning_rate": 2.060104529616725e-05, "loss": 0.42, "step": 473 }, { "epoch": 0.08, "learning_rate": 2.064459930313589e-05, "loss": 0.4111, "step": 474 }, { "epoch": 0.08, "learning_rate": 2.068815331010453e-05, "loss": 0.4098, "step": 475 }, { "epoch": 0.08, "learning_rate": 2.073170731707317e-05, "loss": 0.4049, "step": 476 }, { "epoch": 0.08, "learning_rate": 2.0775261324041813e-05, "loss": 0.4023, "step": 477 }, { "epoch": 0.08, "learning_rate": 2.0818815331010456e-05, "loss": 0.4003, "step": 478 }, { "epoch": 0.08, "learning_rate": 2.0862369337979096e-05, "loss": 0.4102, "step": 479 }, { "epoch": 0.08, "learning_rate": 2.0905923344947736e-05, "loss": 0.4244, "step": 480 }, { "epoch": 0.08, "learning_rate": 2.0949477351916376e-05, "loss": 0.4248, "step": 481 }, { "epoch": 0.08, "learning_rate": 2.099303135888502e-05, "loss": 0.4019, "step": 482 }, { "epoch": 0.08, "learning_rate": 2.103658536585366e-05, "loss": 0.4085, "step": 483 }, { "epoch": 0.08, "learning_rate": 2.10801393728223e-05, "loss": 0.4108, "step": 484 }, { "epoch": 0.08, "learning_rate": 2.1123693379790942e-05, "loss": 0.4097, "step": 485 }, { "epoch": 0.08, "learning_rate": 2.1167247386759585e-05, "loss": 0.4062, "step": 486 }, { "epoch": 0.08, "learning_rate": 2.1210801393728225e-05, "loss": 0.4207, "step": 487 }, { "epoch": 0.09, "learning_rate": 2.1254355400696865e-05, "loss": 0.4263, "step": 488 }, { "epoch": 0.09, "learning_rate": 2.1297909407665505e-05, "loss": 0.4132, "step": 489 }, { "epoch": 0.09, "learning_rate": 2.134146341463415e-05, "loss": 0.4092, "step": 490 }, { "epoch": 0.09, "learning_rate": 2.1385017421602788e-05, "loss": 0.3962, "step": 491 }, { "epoch": 0.09, "learning_rate": 2.1428571428571428e-05, "loss": 0.4132, "step": 492 }, { "epoch": 0.09, "learning_rate": 2.147212543554007e-05, "loss": 0.4203, "step": 493 }, { "epoch": 0.09, "learning_rate": 2.151567944250871e-05, "loss": 0.4107, "step": 494 }, { "epoch": 0.09, "learning_rate": 2.1559233449477355e-05, "loss": 0.4047, "step": 495 }, { "epoch": 0.09, "learning_rate": 2.1602787456445995e-05, "loss": 0.4101, "step": 496 }, { "epoch": 0.09, "learning_rate": 2.1646341463414634e-05, "loss": 0.4074, "step": 497 }, { "epoch": 0.09, "learning_rate": 2.1689895470383274e-05, "loss": 0.4197, "step": 498 }, { "epoch": 0.09, "learning_rate": 2.1733449477351918e-05, "loss": 0.3989, "step": 499 }, { "epoch": 0.09, "learning_rate": 2.1777003484320557e-05, "loss": 0.4108, "step": 500 }, { "epoch": 0.09, "learning_rate": 2.18205574912892e-05, "loss": 0.4119, "step": 501 }, { "epoch": 0.09, "learning_rate": 2.186411149825784e-05, "loss": 0.4164, "step": 502 }, { "epoch": 0.09, "learning_rate": 2.1907665505226484e-05, "loss": 0.4209, "step": 503 }, { "epoch": 0.09, "learning_rate": 2.1951219512195124e-05, "loss": 0.4065, "step": 504 }, { "epoch": 0.09, "learning_rate": 2.1994773519163764e-05, "loss": 0.4033, "step": 505 }, { "epoch": 0.09, "learning_rate": 2.2038327526132404e-05, "loss": 0.4114, "step": 506 }, { "epoch": 0.09, "learning_rate": 2.2081881533101047e-05, "loss": 0.4182, "step": 507 }, { "epoch": 0.09, "learning_rate": 2.2125435540069687e-05, "loss": 0.4054, "step": 508 }, { "epoch": 0.09, "learning_rate": 2.216898954703833e-05, "loss": 0.4149, "step": 509 }, { "epoch": 0.09, "learning_rate": 2.221254355400697e-05, "loss": 0.4112, "step": 510 }, { "epoch": 0.09, "learning_rate": 2.225609756097561e-05, "loss": 0.4025, "step": 511 }, { "epoch": 0.09, "learning_rate": 2.2299651567944253e-05, "loss": 0.4108, "step": 512 }, { "epoch": 0.09, "learning_rate": 2.2343205574912893e-05, "loss": 0.4081, "step": 513 }, { "epoch": 0.09, "learning_rate": 2.2386759581881533e-05, "loss": 0.4201, "step": 514 }, { "epoch": 0.09, "learning_rate": 2.2430313588850173e-05, "loss": 0.4094, "step": 515 }, { "epoch": 0.09, "learning_rate": 2.2473867595818816e-05, "loss": 0.4322, "step": 516 }, { "epoch": 0.09, "learning_rate": 2.251742160278746e-05, "loss": 0.4187, "step": 517 }, { "epoch": 0.09, "learning_rate": 2.25609756097561e-05, "loss": 0.4115, "step": 518 }, { "epoch": 0.09, "learning_rate": 2.260452961672474e-05, "loss": 0.4071, "step": 519 }, { "epoch": 0.09, "learning_rate": 2.264808362369338e-05, "loss": 0.4098, "step": 520 }, { "epoch": 0.09, "learning_rate": 2.2691637630662022e-05, "loss": 0.4165, "step": 521 }, { "epoch": 0.09, "learning_rate": 2.2735191637630662e-05, "loss": 0.4136, "step": 522 }, { "epoch": 0.09, "learning_rate": 2.2778745644599302e-05, "loss": 0.4059, "step": 523 }, { "epoch": 0.09, "learning_rate": 2.2822299651567945e-05, "loss": 0.4253, "step": 524 }, { "epoch": 0.09, "learning_rate": 2.286585365853659e-05, "loss": 0.4378, "step": 525 }, { "epoch": 0.09, "learning_rate": 2.290940766550523e-05, "loss": 0.4074, "step": 526 }, { "epoch": 0.09, "learning_rate": 2.295296167247387e-05, "loss": 0.4234, "step": 527 }, { "epoch": 0.09, "learning_rate": 2.299651567944251e-05, "loss": 0.4263, "step": 528 }, { "epoch": 0.09, "learning_rate": 2.3040069686411152e-05, "loss": 0.4094, "step": 529 }, { "epoch": 0.09, "learning_rate": 2.308362369337979e-05, "loss": 0.4186, "step": 530 }, { "epoch": 0.09, "learning_rate": 2.312717770034843e-05, "loss": 0.4096, "step": 531 }, { "epoch": 0.09, "learning_rate": 2.3170731707317075e-05, "loss": 0.4037, "step": 532 }, { "epoch": 0.09, "learning_rate": 2.3214285714285715e-05, "loss": 0.4024, "step": 533 }, { "epoch": 0.09, "learning_rate": 2.3257839721254358e-05, "loss": 0.4018, "step": 534 }, { "epoch": 0.09, "learning_rate": 2.3301393728222998e-05, "loss": 0.4305, "step": 535 }, { "epoch": 0.09, "learning_rate": 2.3344947735191638e-05, "loss": 0.4025, "step": 536 }, { "epoch": 0.09, "learning_rate": 2.3388501742160278e-05, "loss": 0.4162, "step": 537 }, { "epoch": 0.09, "learning_rate": 2.343205574912892e-05, "loss": 0.4051, "step": 538 }, { "epoch": 0.09, "learning_rate": 2.347560975609756e-05, "loss": 0.4106, "step": 539 }, { "epoch": 0.09, "learning_rate": 2.3519163763066204e-05, "loss": 0.4341, "step": 540 }, { "epoch": 0.09, "learning_rate": 2.3562717770034844e-05, "loss": 0.4142, "step": 541 }, { "epoch": 0.09, "learning_rate": 2.3606271777003487e-05, "loss": 0.4192, "step": 542 }, { "epoch": 0.09, "learning_rate": 2.3649825783972127e-05, "loss": 0.4219, "step": 543 }, { "epoch": 0.09, "learning_rate": 2.3693379790940767e-05, "loss": 0.4401, "step": 544 }, { "epoch": 0.09, "learning_rate": 2.3736933797909407e-05, "loss": 0.4208, "step": 545 }, { "epoch": 0.1, "learning_rate": 2.378048780487805e-05, "loss": 0.3899, "step": 546 }, { "epoch": 0.1, "learning_rate": 2.3824041811846694e-05, "loss": 0.4024, "step": 547 }, { "epoch": 0.1, "learning_rate": 2.3867595818815333e-05, "loss": 0.4191, "step": 548 }, { "epoch": 0.1, "learning_rate": 2.3911149825783973e-05, "loss": 0.4197, "step": 549 }, { "epoch": 0.1, "learning_rate": 2.3954703832752613e-05, "loss": 0.4085, "step": 550 }, { "epoch": 0.1, "learning_rate": 2.3998257839721257e-05, "loss": 0.4009, "step": 551 }, { "epoch": 0.1, "learning_rate": 2.4041811846689896e-05, "loss": 0.422, "step": 552 }, { "epoch": 0.1, "learning_rate": 2.4085365853658536e-05, "loss": 0.4151, "step": 553 }, { "epoch": 0.1, "learning_rate": 2.4128919860627176e-05, "loss": 0.4162, "step": 554 }, { "epoch": 0.1, "learning_rate": 2.4172473867595823e-05, "loss": 0.4009, "step": 555 }, { "epoch": 0.1, "learning_rate": 2.4216027874564463e-05, "loss": 0.3911, "step": 556 }, { "epoch": 0.1, "learning_rate": 2.4259581881533103e-05, "loss": 0.3966, "step": 557 }, { "epoch": 0.1, "learning_rate": 2.4303135888501743e-05, "loss": 0.3978, "step": 558 }, { "epoch": 0.1, "learning_rate": 2.4346689895470386e-05, "loss": 0.4351, "step": 559 }, { "epoch": 0.1, "learning_rate": 2.4390243902439026e-05, "loss": 0.4251, "step": 560 }, { "epoch": 0.1, "learning_rate": 2.4433797909407666e-05, "loss": 0.3944, "step": 561 }, { "epoch": 0.1, "learning_rate": 2.4477351916376306e-05, "loss": 0.3916, "step": 562 }, { "epoch": 0.1, "learning_rate": 2.452090592334495e-05, "loss": 0.3997, "step": 563 }, { "epoch": 0.1, "learning_rate": 2.4564459930313592e-05, "loss": 0.4412, "step": 564 }, { "epoch": 0.1, "learning_rate": 2.4608013937282232e-05, "loss": 0.4287, "step": 565 }, { "epoch": 0.1, "learning_rate": 2.4651567944250872e-05, "loss": 0.4144, "step": 566 }, { "epoch": 0.1, "learning_rate": 2.4695121951219512e-05, "loss": 0.3981, "step": 567 }, { "epoch": 0.1, "learning_rate": 2.4738675958188155e-05, "loss": 0.4191, "step": 568 }, { "epoch": 0.1, "learning_rate": 2.4782229965156795e-05, "loss": 0.4046, "step": 569 }, { "epoch": 0.1, "learning_rate": 2.4825783972125435e-05, "loss": 0.437, "step": 570 }, { "epoch": 0.1, "learning_rate": 2.4869337979094078e-05, "loss": 0.3989, "step": 571 }, { "epoch": 0.1, "learning_rate": 2.4912891986062718e-05, "loss": 0.424, "step": 572 }, { "epoch": 0.1, "learning_rate": 2.495644599303136e-05, "loss": 0.4118, "step": 573 }, { "epoch": 0.1, "learning_rate": 2.5e-05, "loss": 0.4122, "step": 574 }, { "epoch": 0.1, "learning_rate": 2.5043554006968644e-05, "loss": 0.4208, "step": 575 }, { "epoch": 0.1, "learning_rate": 2.5087108013937284e-05, "loss": 0.4098, "step": 576 }, { "epoch": 0.1, "learning_rate": 2.5130662020905928e-05, "loss": 0.4405, "step": 577 }, { "epoch": 0.1, "learning_rate": 2.5174216027874564e-05, "loss": 0.3936, "step": 578 }, { "epoch": 0.1, "learning_rate": 2.5217770034843207e-05, "loss": 0.3907, "step": 579 }, { "epoch": 0.1, "learning_rate": 2.5261324041811847e-05, "loss": 0.4024, "step": 580 }, { "epoch": 0.1, "learning_rate": 2.530487804878049e-05, "loss": 0.4073, "step": 581 }, { "epoch": 0.1, "learning_rate": 2.5348432055749134e-05, "loss": 0.4097, "step": 582 }, { "epoch": 0.1, "learning_rate": 2.539198606271777e-05, "loss": 0.4139, "step": 583 }, { "epoch": 0.1, "learning_rate": 2.5435540069686414e-05, "loss": 0.4172, "step": 584 }, { "epoch": 0.1, "learning_rate": 2.5479094076655054e-05, "loss": 0.4051, "step": 585 }, { "epoch": 0.1, "learning_rate": 2.5522648083623697e-05, "loss": 0.4074, "step": 586 }, { "epoch": 0.1, "learning_rate": 2.5566202090592333e-05, "loss": 0.4048, "step": 587 }, { "epoch": 0.1, "learning_rate": 2.5609756097560977e-05, "loss": 0.4154, "step": 588 }, { "epoch": 0.1, "learning_rate": 2.5653310104529617e-05, "loss": 0.4026, "step": 589 }, { "epoch": 0.1, "learning_rate": 2.569686411149826e-05, "loss": 0.436, "step": 590 }, { "epoch": 0.1, "learning_rate": 2.5740418118466903e-05, "loss": 0.4043, "step": 591 }, { "epoch": 0.1, "learning_rate": 2.578397212543554e-05, "loss": 0.4061, "step": 592 }, { "epoch": 0.1, "learning_rate": 2.5827526132404183e-05, "loss": 0.4114, "step": 593 }, { "epoch": 0.1, "learning_rate": 2.5871080139372823e-05, "loss": 0.4236, "step": 594 }, { "epoch": 0.1, "learning_rate": 2.5914634146341466e-05, "loss": 0.4052, "step": 595 }, { "epoch": 0.1, "learning_rate": 2.5958188153310103e-05, "loss": 0.4212, "step": 596 }, { "epoch": 0.1, "learning_rate": 2.6001742160278746e-05, "loss": 0.3953, "step": 597 }, { "epoch": 0.1, "learning_rate": 2.604529616724739e-05, "loss": 0.4117, "step": 598 }, { "epoch": 0.1, "learning_rate": 2.608885017421603e-05, "loss": 0.4172, "step": 599 }, { "epoch": 0.1, "learning_rate": 2.6132404181184672e-05, "loss": 0.4288, "step": 600 }, { "epoch": 0.1, "learning_rate": 2.617595818815331e-05, "loss": 0.4111, "step": 601 }, { "epoch": 0.1, "learning_rate": 2.6219512195121952e-05, "loss": 0.4083, "step": 602 }, { "epoch": 0.11, "learning_rate": 2.6263066202090592e-05, "loss": 0.3878, "step": 603 }, { "epoch": 0.11, "learning_rate": 2.6306620209059235e-05, "loss": 0.4168, "step": 604 }, { "epoch": 0.11, "learning_rate": 2.6350174216027872e-05, "loss": 0.4142, "step": 605 }, { "epoch": 0.11, "learning_rate": 2.6393728222996515e-05, "loss": 0.402, "step": 606 }, { "epoch": 0.11, "learning_rate": 2.6437282229965162e-05, "loss": 0.4027, "step": 607 }, { "epoch": 0.11, "learning_rate": 2.6480836236933798e-05, "loss": 0.4001, "step": 608 }, { "epoch": 0.11, "learning_rate": 2.652439024390244e-05, "loss": 0.4132, "step": 609 }, { "epoch": 0.11, "learning_rate": 2.6567944250871078e-05, "loss": 0.4122, "step": 610 }, { "epoch": 0.11, "learning_rate": 2.6611498257839725e-05, "loss": 0.4103, "step": 611 }, { "epoch": 0.11, "learning_rate": 2.665505226480836e-05, "loss": 0.4317, "step": 612 }, { "epoch": 0.11, "learning_rate": 2.6698606271777005e-05, "loss": 0.3924, "step": 613 }, { "epoch": 0.11, "learning_rate": 2.6742160278745648e-05, "loss": 0.3969, "step": 614 }, { "epoch": 0.11, "learning_rate": 2.6785714285714288e-05, "loss": 0.4343, "step": 615 }, { "epoch": 0.11, "learning_rate": 2.682926829268293e-05, "loss": 0.4005, "step": 616 }, { "epoch": 0.11, "learning_rate": 2.6872822299651568e-05, "loss": 0.4232, "step": 617 }, { "epoch": 0.11, "learning_rate": 2.691637630662021e-05, "loss": 0.4206, "step": 618 }, { "epoch": 0.11, "learning_rate": 2.695993031358885e-05, "loss": 0.41, "step": 619 }, { "epoch": 0.11, "learning_rate": 2.7003484320557494e-05, "loss": 0.4237, "step": 620 }, { "epoch": 0.11, "learning_rate": 2.7047038327526137e-05, "loss": 0.3957, "step": 621 }, { "epoch": 0.11, "learning_rate": 2.7090592334494774e-05, "loss": 0.4052, "step": 622 }, { "epoch": 0.11, "learning_rate": 2.7134146341463417e-05, "loss": 0.4119, "step": 623 }, { "epoch": 0.11, "learning_rate": 2.7177700348432057e-05, "loss": 0.4118, "step": 624 }, { "epoch": 0.11, "learning_rate": 2.72212543554007e-05, "loss": 0.3962, "step": 625 }, { "epoch": 0.11, "learning_rate": 2.7264808362369337e-05, "loss": 0.3998, "step": 626 }, { "epoch": 0.11, "learning_rate": 2.730836236933798e-05, "loss": 0.42, "step": 627 }, { "epoch": 0.11, "learning_rate": 2.735191637630662e-05, "loss": 0.4161, "step": 628 }, { "epoch": 0.11, "learning_rate": 2.7395470383275263e-05, "loss": 0.4043, "step": 629 }, { "epoch": 0.11, "learning_rate": 2.7439024390243906e-05, "loss": 0.4108, "step": 630 }, { "epoch": 0.11, "learning_rate": 2.7482578397212543e-05, "loss": 0.4103, "step": 631 }, { "epoch": 0.11, "learning_rate": 2.7526132404181186e-05, "loss": 0.4057, "step": 632 }, { "epoch": 0.11, "learning_rate": 2.7569686411149826e-05, "loss": 0.4097, "step": 633 }, { "epoch": 0.11, "learning_rate": 2.761324041811847e-05, "loss": 0.4048, "step": 634 }, { "epoch": 0.11, "learning_rate": 2.7656794425087106e-05, "loss": 0.3976, "step": 635 }, { "epoch": 0.11, "learning_rate": 2.770034843205575e-05, "loss": 0.399, "step": 636 }, { "epoch": 0.11, "learning_rate": 2.7743902439024393e-05, "loss": 0.419, "step": 637 }, { "epoch": 0.11, "learning_rate": 2.7787456445993032e-05, "loss": 0.4207, "step": 638 }, { "epoch": 0.11, "learning_rate": 2.7831010452961676e-05, "loss": 0.4163, "step": 639 }, { "epoch": 0.11, "learning_rate": 2.7874564459930312e-05, "loss": 0.3925, "step": 640 }, { "epoch": 0.11, "learning_rate": 2.7918118466898955e-05, "loss": 0.4024, "step": 641 }, { "epoch": 0.11, "learning_rate": 2.7961672473867595e-05, "loss": 0.4146, "step": 642 }, { "epoch": 0.11, "learning_rate": 2.800522648083624e-05, "loss": 0.3947, "step": 643 }, { "epoch": 0.11, "learning_rate": 2.8048780487804882e-05, "loss": 0.4023, "step": 644 }, { "epoch": 0.11, "learning_rate": 2.809233449477352e-05, "loss": 0.407, "step": 645 }, { "epoch": 0.11, "learning_rate": 2.8135888501742165e-05, "loss": 0.4031, "step": 646 }, { "epoch": 0.11, "learning_rate": 2.81794425087108e-05, "loss": 0.4232, "step": 647 }, { "epoch": 0.11, "learning_rate": 2.8222996515679445e-05, "loss": 0.4283, "step": 648 }, { "epoch": 0.11, "learning_rate": 2.826655052264808e-05, "loss": 0.4045, "step": 649 }, { "epoch": 0.11, "learning_rate": 2.8310104529616728e-05, "loss": 0.3918, "step": 650 }, { "epoch": 0.11, "learning_rate": 2.8353658536585365e-05, "loss": 0.436, "step": 651 }, { "epoch": 0.11, "learning_rate": 2.8397212543554008e-05, "loss": 0.4058, "step": 652 }, { "epoch": 0.11, "learning_rate": 2.844076655052265e-05, "loss": 0.4254, "step": 653 }, { "epoch": 0.11, "learning_rate": 2.848432055749129e-05, "loss": 0.403, "step": 654 }, { "epoch": 0.11, "learning_rate": 2.8527874564459934e-05, "loss": 0.427, "step": 655 }, { "epoch": 0.11, "learning_rate": 2.857142857142857e-05, "loss": 0.4153, "step": 656 }, { "epoch": 0.11, "learning_rate": 2.8614982578397214e-05, "loss": 0.4054, "step": 657 }, { "epoch": 0.11, "learning_rate": 2.8658536585365854e-05, "loss": 0.4089, "step": 658 }, { "epoch": 0.11, "learning_rate": 2.8702090592334497e-05, "loss": 0.3784, "step": 659 }, { "epoch": 0.12, "learning_rate": 2.874564459930314e-05, "loss": 0.3962, "step": 660 }, { "epoch": 0.12, "learning_rate": 2.8789198606271777e-05, "loss": 0.4139, "step": 661 }, { "epoch": 0.12, "learning_rate": 2.883275261324042e-05, "loss": 0.3946, "step": 662 }, { "epoch": 0.12, "learning_rate": 2.887630662020906e-05, "loss": 0.3863, "step": 663 }, { "epoch": 0.12, "learning_rate": 2.8919860627177704e-05, "loss": 0.4105, "step": 664 }, { "epoch": 0.12, "learning_rate": 2.896341463414634e-05, "loss": 0.427, "step": 665 }, { "epoch": 0.12, "learning_rate": 2.9006968641114983e-05, "loss": 0.4064, "step": 666 }, { "epoch": 0.12, "learning_rate": 2.9050522648083623e-05, "loss": 0.4049, "step": 667 }, { "epoch": 0.12, "learning_rate": 2.9094076655052267e-05, "loss": 0.4278, "step": 668 }, { "epoch": 0.12, "learning_rate": 2.913763066202091e-05, "loss": 0.4122, "step": 669 }, { "epoch": 0.12, "learning_rate": 2.9181184668989546e-05, "loss": 0.4094, "step": 670 }, { "epoch": 0.12, "learning_rate": 2.922473867595819e-05, "loss": 0.4152, "step": 671 }, { "epoch": 0.12, "learning_rate": 2.926829268292683e-05, "loss": 0.4036, "step": 672 }, { "epoch": 0.12, "learning_rate": 2.9311846689895473e-05, "loss": 0.3935, "step": 673 }, { "epoch": 0.12, "learning_rate": 2.935540069686411e-05, "loss": 0.4249, "step": 674 }, { "epoch": 0.12, "learning_rate": 2.9398954703832753e-05, "loss": 0.4193, "step": 675 }, { "epoch": 0.12, "learning_rate": 2.9442508710801396e-05, "loss": 0.3982, "step": 676 }, { "epoch": 0.12, "learning_rate": 2.9486062717770036e-05, "loss": 0.409, "step": 677 }, { "epoch": 0.12, "learning_rate": 2.952961672473868e-05, "loss": 0.3981, "step": 678 }, { "epoch": 0.12, "learning_rate": 2.9573170731707316e-05, "loss": 0.4124, "step": 679 }, { "epoch": 0.12, "learning_rate": 2.9616724738675962e-05, "loss": 0.4008, "step": 680 }, { "epoch": 0.12, "learning_rate": 2.96602787456446e-05, "loss": 0.4145, "step": 681 }, { "epoch": 0.12, "learning_rate": 2.9703832752613242e-05, "loss": 0.4161, "step": 682 }, { "epoch": 0.12, "learning_rate": 2.9747386759581885e-05, "loss": 0.41, "step": 683 }, { "epoch": 0.12, "learning_rate": 2.9790940766550525e-05, "loss": 0.4133, "step": 684 }, { "epoch": 0.12, "learning_rate": 2.983449477351917e-05, "loss": 0.4132, "step": 685 }, { "epoch": 0.12, "learning_rate": 2.9878048780487805e-05, "loss": 0.4094, "step": 686 }, { "epoch": 0.12, "learning_rate": 2.9921602787456448e-05, "loss": 0.4324, "step": 687 }, { "epoch": 0.12, "learning_rate": 2.9965156794425088e-05, "loss": 0.4248, "step": 688 }, { "epoch": 0.12, "learning_rate": 3.000871080139373e-05, "loss": 0.4217, "step": 689 }, { "epoch": 0.12, "learning_rate": 3.0052264808362368e-05, "loss": 0.412, "step": 690 }, { "epoch": 0.12, "learning_rate": 3.009581881533101e-05, "loss": 0.4106, "step": 691 }, { "epoch": 0.12, "learning_rate": 3.0139372822299655e-05, "loss": 0.4129, "step": 692 }, { "epoch": 0.12, "learning_rate": 3.0182926829268294e-05, "loss": 0.4092, "step": 693 }, { "epoch": 0.12, "learning_rate": 3.0226480836236938e-05, "loss": 0.4175, "step": 694 }, { "epoch": 0.12, "learning_rate": 3.0270034843205574e-05, "loss": 0.4185, "step": 695 }, { "epoch": 0.12, "learning_rate": 3.0313588850174217e-05, "loss": 0.3998, "step": 696 }, { "epoch": 0.12, "learning_rate": 3.0357142857142857e-05, "loss": 0.3802, "step": 697 }, { "epoch": 0.12, "learning_rate": 3.04006968641115e-05, "loss": 0.4185, "step": 698 }, { "epoch": 0.12, "learning_rate": 3.0444250871080144e-05, "loss": 0.3994, "step": 699 }, { "epoch": 0.12, "learning_rate": 3.048780487804878e-05, "loss": 0.4048, "step": 700 }, { "epoch": 0.12, "learning_rate": 3.0531358885017424e-05, "loss": 0.4115, "step": 701 }, { "epoch": 0.12, "learning_rate": 3.057491289198606e-05, "loss": 0.4051, "step": 702 }, { "epoch": 0.12, "learning_rate": 3.0618466898954704e-05, "loss": 0.408, "step": 703 }, { "epoch": 0.12, "learning_rate": 3.066202090592335e-05, "loss": 0.4048, "step": 704 }, { "epoch": 0.12, "learning_rate": 3.070557491289199e-05, "loss": 0.3999, "step": 705 }, { "epoch": 0.12, "learning_rate": 3.074912891986063e-05, "loss": 0.3942, "step": 706 }, { "epoch": 0.12, "learning_rate": 3.079268292682927e-05, "loss": 0.4136, "step": 707 }, { "epoch": 0.12, "learning_rate": 3.083623693379791e-05, "loss": 0.4125, "step": 708 }, { "epoch": 0.12, "learning_rate": 3.087979094076655e-05, "loss": 0.4158, "step": 709 }, { "epoch": 0.12, "learning_rate": 3.092334494773519e-05, "loss": 0.408, "step": 710 }, { "epoch": 0.12, "learning_rate": 3.096689895470383e-05, "loss": 0.4156, "step": 711 }, { "epoch": 0.12, "learning_rate": 3.101045296167247e-05, "loss": 0.4, "step": 712 }, { "epoch": 0.12, "learning_rate": 3.1054006968641116e-05, "loss": 0.4293, "step": 713 }, { "epoch": 0.12, "learning_rate": 3.109756097560976e-05, "loss": 0.4063, "step": 714 }, { "epoch": 0.12, "learning_rate": 3.11411149825784e-05, "loss": 0.4086, "step": 715 }, { "epoch": 0.12, "learning_rate": 3.118466898954704e-05, "loss": 0.3979, "step": 716 }, { "epoch": 0.12, "learning_rate": 3.122822299651568e-05, "loss": 0.394, "step": 717 }, { "epoch": 0.13, "learning_rate": 3.127177700348432e-05, "loss": 0.4031, "step": 718 }, { "epoch": 0.13, "learning_rate": 3.131533101045296e-05, "loss": 0.387, "step": 719 }, { "epoch": 0.13, "learning_rate": 3.13588850174216e-05, "loss": 0.4016, "step": 720 }, { "epoch": 0.13, "learning_rate": 3.140243902439025e-05, "loss": 0.4052, "step": 721 }, { "epoch": 0.13, "learning_rate": 3.144599303135889e-05, "loss": 0.4012, "step": 722 }, { "epoch": 0.13, "learning_rate": 3.148954703832753e-05, "loss": 0.3998, "step": 723 }, { "epoch": 0.13, "learning_rate": 3.153310104529617e-05, "loss": 0.3952, "step": 724 }, { "epoch": 0.13, "learning_rate": 3.157665505226481e-05, "loss": 0.3887, "step": 725 }, { "epoch": 0.13, "learning_rate": 3.162020905923345e-05, "loss": 0.4193, "step": 726 }, { "epoch": 0.13, "learning_rate": 3.166376306620209e-05, "loss": 0.3907, "step": 727 }, { "epoch": 0.13, "learning_rate": 3.170731707317073e-05, "loss": 0.4118, "step": 728 }, { "epoch": 0.13, "learning_rate": 3.1750871080139375e-05, "loss": 0.414, "step": 729 }, { "epoch": 0.13, "learning_rate": 3.179442508710802e-05, "loss": 0.4282, "step": 730 }, { "epoch": 0.13, "learning_rate": 3.183797909407666e-05, "loss": 0.4195, "step": 731 }, { "epoch": 0.13, "learning_rate": 3.18815331010453e-05, "loss": 0.4082, "step": 732 }, { "epoch": 0.13, "learning_rate": 3.192508710801394e-05, "loss": 0.3976, "step": 733 }, { "epoch": 0.13, "learning_rate": 3.196864111498258e-05, "loss": 0.4098, "step": 734 }, { "epoch": 0.13, "learning_rate": 3.201219512195122e-05, "loss": 0.4078, "step": 735 }, { "epoch": 0.13, "learning_rate": 3.205574912891986e-05, "loss": 0.3983, "step": 736 }, { "epoch": 0.13, "learning_rate": 3.20993031358885e-05, "loss": 0.4183, "step": 737 }, { "epoch": 0.13, "learning_rate": 3.2142857142857144e-05, "loss": 0.4001, "step": 738 }, { "epoch": 0.13, "learning_rate": 3.218641114982579e-05, "loss": 0.3942, "step": 739 }, { "epoch": 0.13, "learning_rate": 3.222996515679443e-05, "loss": 0.4091, "step": 740 }, { "epoch": 0.13, "learning_rate": 3.227351916376307e-05, "loss": 0.3978, "step": 741 }, { "epoch": 0.13, "learning_rate": 3.231707317073171e-05, "loss": 0.4142, "step": 742 }, { "epoch": 0.13, "learning_rate": 3.236062717770035e-05, "loss": 0.3966, "step": 743 }, { "epoch": 0.13, "learning_rate": 3.240418118466899e-05, "loss": 0.4026, "step": 744 }, { "epoch": 0.13, "learning_rate": 3.244773519163763e-05, "loss": 0.4176, "step": 745 }, { "epoch": 0.13, "learning_rate": 3.249128919860627e-05, "loss": 0.4097, "step": 746 }, { "epoch": 0.13, "learning_rate": 3.253484320557491e-05, "loss": 0.4151, "step": 747 }, { "epoch": 0.13, "learning_rate": 3.2578397212543556e-05, "loss": 0.4215, "step": 748 }, { "epoch": 0.13, "learning_rate": 3.26219512195122e-05, "loss": 0.4223, "step": 749 }, { "epoch": 0.13, "learning_rate": 3.2665505226480836e-05, "loss": 0.3948, "step": 750 }, { "epoch": 0.13, "learning_rate": 3.270905923344948e-05, "loss": 0.4014, "step": 751 }, { "epoch": 0.13, "learning_rate": 3.2752613240418116e-05, "loss": 0.4096, "step": 752 }, { "epoch": 0.13, "learning_rate": 3.279616724738676e-05, "loss": 0.4218, "step": 753 }, { "epoch": 0.13, "learning_rate": 3.28397212543554e-05, "loss": 0.4193, "step": 754 }, { "epoch": 0.13, "learning_rate": 3.288327526132404e-05, "loss": 0.4001, "step": 755 }, { "epoch": 0.13, "learning_rate": 3.292682926829269e-05, "loss": 0.4114, "step": 756 }, { "epoch": 0.13, "learning_rate": 3.2970383275261326e-05, "loss": 0.3973, "step": 757 }, { "epoch": 0.13, "learning_rate": 3.301393728222997e-05, "loss": 0.385, "step": 758 }, { "epoch": 0.13, "learning_rate": 3.3057491289198605e-05, "loss": 0.4027, "step": 759 }, { "epoch": 0.13, "learning_rate": 3.310104529616725e-05, "loss": 0.4144, "step": 760 }, { "epoch": 0.13, "learning_rate": 3.314459930313589e-05, "loss": 0.4082, "step": 761 }, { "epoch": 0.13, "learning_rate": 3.318815331010453e-05, "loss": 0.4017, "step": 762 }, { "epoch": 0.13, "learning_rate": 3.323170731707317e-05, "loss": 0.3958, "step": 763 }, { "epoch": 0.13, "learning_rate": 3.3275261324041815e-05, "loss": 0.4146, "step": 764 }, { "epoch": 0.13, "learning_rate": 3.331881533101046e-05, "loss": 0.402, "step": 765 }, { "epoch": 0.13, "learning_rate": 3.3362369337979095e-05, "loss": 0.4182, "step": 766 }, { "epoch": 0.13, "learning_rate": 3.340592334494774e-05, "loss": 0.4191, "step": 767 }, { "epoch": 0.13, "learning_rate": 3.344947735191638e-05, "loss": 0.4192, "step": 768 }, { "epoch": 0.13, "learning_rate": 3.349303135888502e-05, "loss": 0.3892, "step": 769 }, { "epoch": 0.13, "learning_rate": 3.353658536585366e-05, "loss": 0.3925, "step": 770 }, { "epoch": 0.13, "learning_rate": 3.35801393728223e-05, "loss": 0.4102, "step": 771 }, { "epoch": 0.13, "learning_rate": 3.362369337979094e-05, "loss": 0.4085, "step": 772 }, { "epoch": 0.13, "learning_rate": 3.3667247386759584e-05, "loss": 0.399, "step": 773 }, { "epoch": 0.13, "learning_rate": 3.371080139372823e-05, "loss": 0.3936, "step": 774 }, { "epoch": 0.14, "learning_rate": 3.3754355400696864e-05, "loss": 0.3938, "step": 775 }, { "epoch": 0.14, "learning_rate": 3.379790940766551e-05, "loss": 0.4128, "step": 776 }, { "epoch": 0.14, "learning_rate": 3.384146341463415e-05, "loss": 0.3926, "step": 777 }, { "epoch": 0.14, "learning_rate": 3.388501742160279e-05, "loss": 0.4001, "step": 778 }, { "epoch": 0.14, "learning_rate": 3.392857142857143e-05, "loss": 0.3968, "step": 779 }, { "epoch": 0.14, "learning_rate": 3.397212543554007e-05, "loss": 0.3884, "step": 780 }, { "epoch": 0.14, "learning_rate": 3.401567944250871e-05, "loss": 0.3962, "step": 781 }, { "epoch": 0.14, "learning_rate": 3.4059233449477354e-05, "loss": 0.4007, "step": 782 }, { "epoch": 0.14, "learning_rate": 3.4102787456446e-05, "loss": 0.4028, "step": 783 }, { "epoch": 0.14, "learning_rate": 3.414634146341464e-05, "loss": 0.3959, "step": 784 }, { "epoch": 0.14, "learning_rate": 3.4189895470383277e-05, "loss": 0.4051, "step": 785 }, { "epoch": 0.14, "learning_rate": 3.423344947735192e-05, "loss": 0.4255, "step": 786 }, { "epoch": 0.14, "learning_rate": 3.4277003484320556e-05, "loss": 0.3901, "step": 787 }, { "epoch": 0.14, "learning_rate": 3.43205574912892e-05, "loss": 0.3946, "step": 788 }, { "epoch": 0.14, "learning_rate": 3.4364111498257836e-05, "loss": 0.3876, "step": 789 }, { "epoch": 0.14, "learning_rate": 3.4407665505226486e-05, "loss": 0.3974, "step": 790 }, { "epoch": 0.14, "learning_rate": 3.445121951219512e-05, "loss": 0.3935, "step": 791 }, { "epoch": 0.14, "learning_rate": 3.4494773519163766e-05, "loss": 0.3969, "step": 792 }, { "epoch": 0.14, "learning_rate": 3.453832752613241e-05, "loss": 0.4016, "step": 793 }, { "epoch": 0.14, "learning_rate": 3.4581881533101046e-05, "loss": 0.3971, "step": 794 }, { "epoch": 0.14, "learning_rate": 3.462543554006969e-05, "loss": 0.3932, "step": 795 }, { "epoch": 0.14, "learning_rate": 3.4668989547038326e-05, "loss": 0.404, "step": 796 }, { "epoch": 0.14, "learning_rate": 3.471254355400697e-05, "loss": 0.4069, "step": 797 }, { "epoch": 0.14, "learning_rate": 3.475609756097561e-05, "loss": 0.4026, "step": 798 }, { "epoch": 0.14, "learning_rate": 3.4799651567944255e-05, "loss": 0.3888, "step": 799 }, { "epoch": 0.14, "learning_rate": 3.48432055749129e-05, "loss": 0.4065, "step": 800 }, { "epoch": 0.14, "learning_rate": 3.4886759581881535e-05, "loss": 0.4089, "step": 801 }, { "epoch": 0.14, "learning_rate": 3.493031358885018e-05, "loss": 0.4133, "step": 802 }, { "epoch": 0.14, "learning_rate": 3.4973867595818815e-05, "loss": 0.4044, "step": 803 }, { "epoch": 0.14, "learning_rate": 3.501742160278746e-05, "loss": 0.4155, "step": 804 }, { "epoch": 0.14, "learning_rate": 3.5060975609756095e-05, "loss": 0.4298, "step": 805 }, { "epoch": 0.14, "learning_rate": 3.510452961672474e-05, "loss": 0.4035, "step": 806 }, { "epoch": 0.14, "learning_rate": 3.514808362369338e-05, "loss": 0.4169, "step": 807 }, { "epoch": 0.14, "learning_rate": 3.5191637630662025e-05, "loss": 0.4124, "step": 808 }, { "epoch": 0.14, "learning_rate": 3.523519163763067e-05, "loss": 0.3943, "step": 809 }, { "epoch": 0.14, "learning_rate": 3.5278745644599304e-05, "loss": 0.4115, "step": 810 }, { "epoch": 0.14, "learning_rate": 3.532229965156795e-05, "loss": 0.3895, "step": 811 }, { "epoch": 0.14, "learning_rate": 3.5365853658536584e-05, "loss": 0.4136, "step": 812 }, { "epoch": 0.14, "learning_rate": 3.540940766550523e-05, "loss": 0.3891, "step": 813 }, { "epoch": 0.14, "learning_rate": 3.5452961672473864e-05, "loss": 0.4063, "step": 814 }, { "epoch": 0.14, "learning_rate": 3.549651567944251e-05, "loss": 0.3959, "step": 815 }, { "epoch": 0.14, "learning_rate": 3.554006968641115e-05, "loss": 0.422, "step": 816 }, { "epoch": 0.14, "learning_rate": 3.5583623693379794e-05, "loss": 0.4069, "step": 817 }, { "epoch": 0.14, "learning_rate": 3.562717770034844e-05, "loss": 0.3976, "step": 818 }, { "epoch": 0.14, "learning_rate": 3.5670731707317074e-05, "loss": 0.4136, "step": 819 }, { "epoch": 0.14, "learning_rate": 3.571428571428572e-05, "loss": 0.3902, "step": 820 }, { "epoch": 0.14, "learning_rate": 3.5757839721254353e-05, "loss": 0.3983, "step": 821 }, { "epoch": 0.14, "learning_rate": 3.5801393728223e-05, "loss": 0.3988, "step": 822 }, { "epoch": 0.14, "learning_rate": 3.584494773519164e-05, "loss": 0.3821, "step": 823 }, { "epoch": 0.14, "learning_rate": 3.5888501742160277e-05, "loss": 0.3971, "step": 824 }, { "epoch": 0.14, "learning_rate": 3.5932055749128927e-05, "loss": 0.3924, "step": 825 }, { "epoch": 0.14, "learning_rate": 3.597560975609756e-05, "loss": 0.4062, "step": 826 }, { "epoch": 0.14, "learning_rate": 3.6019163763066206e-05, "loss": 0.3871, "step": 827 }, { "epoch": 0.14, "learning_rate": 3.606271777003484e-05, "loss": 0.4016, "step": 828 }, { "epoch": 0.14, "learning_rate": 3.6106271777003486e-05, "loss": 0.4008, "step": 829 }, { "epoch": 0.14, "learning_rate": 3.614982578397213e-05, "loss": 0.4172, "step": 830 }, { "epoch": 0.14, "learning_rate": 3.6193379790940766e-05, "loss": 0.4231, "step": 831 }, { "epoch": 0.15, "learning_rate": 3.623693379790941e-05, "loss": 0.4176, "step": 832 }, { "epoch": 0.15, "learning_rate": 3.628048780487805e-05, "loss": 0.4132, "step": 833 }, { "epoch": 0.15, "learning_rate": 3.6324041811846696e-05, "loss": 0.4085, "step": 834 }, { "epoch": 0.15, "learning_rate": 3.636759581881533e-05, "loss": 0.3927, "step": 835 }, { "epoch": 0.15, "learning_rate": 3.6411149825783976e-05, "loss": 0.4183, "step": 836 }, { "epoch": 0.15, "learning_rate": 3.645470383275261e-05, "loss": 0.3992, "step": 837 }, { "epoch": 0.15, "learning_rate": 3.6498257839721255e-05, "loss": 0.394, "step": 838 }, { "epoch": 0.15, "learning_rate": 3.65418118466899e-05, "loss": 0.4182, "step": 839 }, { "epoch": 0.15, "learning_rate": 3.6585365853658535e-05, "loss": 0.4211, "step": 840 }, { "epoch": 0.15, "learning_rate": 3.662891986062718e-05, "loss": 0.3972, "step": 841 }, { "epoch": 0.15, "learning_rate": 3.667247386759582e-05, "loss": 0.4031, "step": 842 }, { "epoch": 0.15, "learning_rate": 3.6716027874564465e-05, "loss": 0.4051, "step": 843 }, { "epoch": 0.15, "learning_rate": 3.67595818815331e-05, "loss": 0.4024, "step": 844 }, { "epoch": 0.15, "learning_rate": 3.6803135888501745e-05, "loss": 0.3845, "step": 845 }, { "epoch": 0.15, "learning_rate": 3.684668989547039e-05, "loss": 0.401, "step": 846 }, { "epoch": 0.15, "learning_rate": 3.6890243902439025e-05, "loss": 0.399, "step": 847 }, { "epoch": 0.15, "learning_rate": 3.693379790940767e-05, "loss": 0.403, "step": 848 }, { "epoch": 0.15, "learning_rate": 3.6977351916376304e-05, "loss": 0.3857, "step": 849 }, { "epoch": 0.15, "learning_rate": 3.702090592334495e-05, "loss": 0.4104, "step": 850 }, { "epoch": 0.15, "learning_rate": 3.706445993031359e-05, "loss": 0.3992, "step": 851 }, { "epoch": 0.15, "learning_rate": 3.7108013937282234e-05, "loss": 0.4143, "step": 852 }, { "epoch": 0.15, "learning_rate": 3.715156794425087e-05, "loss": 0.3894, "step": 853 }, { "epoch": 0.15, "learning_rate": 3.7195121951219514e-05, "loss": 0.3974, "step": 854 }, { "epoch": 0.15, "learning_rate": 3.723867595818816e-05, "loss": 0.3982, "step": 855 }, { "epoch": 0.15, "learning_rate": 3.7282229965156794e-05, "loss": 0.3946, "step": 856 }, { "epoch": 0.15, "learning_rate": 3.732578397212544e-05, "loss": 0.3883, "step": 857 }, { "epoch": 0.15, "learning_rate": 3.7369337979094074e-05, "loss": 0.4104, "step": 858 }, { "epoch": 0.15, "learning_rate": 3.741289198606272e-05, "loss": 0.4061, "step": 859 }, { "epoch": 0.15, "learning_rate": 3.745644599303136e-05, "loss": 0.4216, "step": 860 }, { "epoch": 0.15, "learning_rate": 3.7500000000000003e-05, "loss": 0.4013, "step": 861 }, { "epoch": 0.15, "learning_rate": 3.754355400696865e-05, "loss": 0.3927, "step": 862 }, { "epoch": 0.15, "learning_rate": 3.758710801393728e-05, "loss": 0.4089, "step": 863 }, { "epoch": 0.15, "learning_rate": 3.7630662020905927e-05, "loss": 0.3957, "step": 864 }, { "epoch": 0.15, "learning_rate": 3.767421602787456e-05, "loss": 0.4035, "step": 865 }, { "epoch": 0.15, "learning_rate": 3.7717770034843206e-05, "loss": 0.412, "step": 866 }, { "epoch": 0.15, "learning_rate": 3.776132404181185e-05, "loss": 0.4149, "step": 867 }, { "epoch": 0.15, "learning_rate": 3.780487804878049e-05, "loss": 0.424, "step": 868 }, { "epoch": 0.15, "learning_rate": 3.7848432055749136e-05, "loss": 0.4042, "step": 869 }, { "epoch": 0.15, "learning_rate": 3.789198606271777e-05, "loss": 0.4171, "step": 870 }, { "epoch": 0.15, "learning_rate": 3.7935540069686416e-05, "loss": 0.3981, "step": 871 }, { "epoch": 0.15, "learning_rate": 3.797909407665505e-05, "loss": 0.3842, "step": 872 }, { "epoch": 0.15, "learning_rate": 3.8022648083623696e-05, "loss": 0.3935, "step": 873 }, { "epoch": 0.15, "learning_rate": 3.806620209059233e-05, "loss": 0.402, "step": 874 }, { "epoch": 0.15, "learning_rate": 3.8109756097560976e-05, "loss": 0.4021, "step": 875 }, { "epoch": 0.15, "learning_rate": 3.815331010452962e-05, "loss": 0.4161, "step": 876 }, { "epoch": 0.15, "learning_rate": 3.819686411149826e-05, "loss": 0.3879, "step": 877 }, { "epoch": 0.15, "learning_rate": 3.8240418118466905e-05, "loss": 0.3933, "step": 878 }, { "epoch": 0.15, "learning_rate": 3.828397212543554e-05, "loss": 0.404, "step": 879 }, { "epoch": 0.15, "learning_rate": 3.8327526132404185e-05, "loss": 0.4078, "step": 880 }, { "epoch": 0.15, "learning_rate": 3.837108013937282e-05, "loss": 0.3875, "step": 881 }, { "epoch": 0.15, "learning_rate": 3.8414634146341465e-05, "loss": 0.3932, "step": 882 }, { "epoch": 0.15, "learning_rate": 3.84581881533101e-05, "loss": 0.4165, "step": 883 }, { "epoch": 0.15, "learning_rate": 3.8501742160278745e-05, "loss": 0.4029, "step": 884 }, { "epoch": 0.15, "learning_rate": 3.854529616724739e-05, "loss": 0.4168, "step": 885 }, { "epoch": 0.15, "learning_rate": 3.858885017421603e-05, "loss": 0.3991, "step": 886 }, { "epoch": 0.15, "learning_rate": 3.8632404181184675e-05, "loss": 0.4164, "step": 887 }, { "epoch": 0.15, "learning_rate": 3.867595818815331e-05, "loss": 0.3934, "step": 888 }, { "epoch": 0.15, "learning_rate": 3.8719512195121954e-05, "loss": 0.3912, "step": 889 }, { "epoch": 0.16, "learning_rate": 3.876306620209059e-05, "loss": 0.4115, "step": 890 }, { "epoch": 0.16, "learning_rate": 3.8806620209059234e-05, "loss": 0.3851, "step": 891 }, { "epoch": 0.16, "learning_rate": 3.885017421602787e-05, "loss": 0.3999, "step": 892 }, { "epoch": 0.16, "learning_rate": 3.8893728222996514e-05, "loss": 0.4194, "step": 893 }, { "epoch": 0.16, "learning_rate": 3.8937282229965164e-05, "loss": 0.4016, "step": 894 }, { "epoch": 0.16, "learning_rate": 3.89808362369338e-05, "loss": 0.3885, "step": 895 }, { "epoch": 0.16, "learning_rate": 3.9024390243902444e-05, "loss": 0.386, "step": 896 }, { "epoch": 0.16, "learning_rate": 3.906794425087108e-05, "loss": 0.4206, "step": 897 }, { "epoch": 0.16, "learning_rate": 3.9111498257839724e-05, "loss": 0.4196, "step": 898 }, { "epoch": 0.16, "learning_rate": 3.915505226480836e-05, "loss": 0.4023, "step": 899 }, { "epoch": 0.16, "learning_rate": 3.9198606271777003e-05, "loss": 0.4156, "step": 900 }, { "epoch": 0.16, "learning_rate": 3.924216027874565e-05, "loss": 0.3951, "step": 901 }, { "epoch": 0.16, "learning_rate": 3.928571428571429e-05, "loss": 0.3996, "step": 902 }, { "epoch": 0.16, "learning_rate": 3.932926829268293e-05, "loss": 0.3944, "step": 903 }, { "epoch": 0.16, "learning_rate": 3.937282229965157e-05, "loss": 0.413, "step": 904 }, { "epoch": 0.16, "learning_rate": 3.941637630662021e-05, "loss": 0.392, "step": 905 }, { "epoch": 0.16, "learning_rate": 3.945993031358885e-05, "loss": 0.4078, "step": 906 }, { "epoch": 0.16, "learning_rate": 3.950348432055749e-05, "loss": 0.3981, "step": 907 }, { "epoch": 0.16, "learning_rate": 3.9547038327526136e-05, "loss": 0.3888, "step": 908 }, { "epoch": 0.16, "learning_rate": 3.959059233449477e-05, "loss": 0.4176, "step": 909 }, { "epoch": 0.16, "learning_rate": 3.9634146341463416e-05, "loss": 0.4048, "step": 910 }, { "epoch": 0.16, "learning_rate": 3.967770034843206e-05, "loss": 0.3962, "step": 911 }, { "epoch": 0.16, "learning_rate": 3.97212543554007e-05, "loss": 0.4055, "step": 912 }, { "epoch": 0.16, "learning_rate": 3.976480836236934e-05, "loss": 0.4142, "step": 913 }, { "epoch": 0.16, "learning_rate": 3.980836236933798e-05, "loss": 0.3983, "step": 914 }, { "epoch": 0.16, "learning_rate": 3.985191637630662e-05, "loss": 0.3996, "step": 915 }, { "epoch": 0.16, "learning_rate": 3.989547038327526e-05, "loss": 0.4001, "step": 916 }, { "epoch": 0.16, "learning_rate": 3.9939024390243905e-05, "loss": 0.3997, "step": 917 }, { "epoch": 0.16, "learning_rate": 3.998257839721254e-05, "loss": 0.3929, "step": 918 }, { "epoch": 0.16, "learning_rate": 4.0026132404181185e-05, "loss": 0.4049, "step": 919 }, { "epoch": 0.16, "learning_rate": 4.006968641114983e-05, "loss": 0.4161, "step": 920 }, { "epoch": 0.16, "learning_rate": 4.011324041811847e-05, "loss": 0.4, "step": 921 }, { "epoch": 0.16, "learning_rate": 4.015679442508711e-05, "loss": 0.3847, "step": 922 }, { "epoch": 0.16, "learning_rate": 4.020034843205575e-05, "loss": 0.4098, "step": 923 }, { "epoch": 0.16, "learning_rate": 4.0243902439024395e-05, "loss": 0.4092, "step": 924 }, { "epoch": 0.16, "learning_rate": 4.028745644599303e-05, "loss": 0.3927, "step": 925 }, { "epoch": 0.16, "learning_rate": 4.0331010452961675e-05, "loss": 0.4025, "step": 926 }, { "epoch": 0.16, "learning_rate": 4.037456445993031e-05, "loss": 0.3962, "step": 927 }, { "epoch": 0.16, "learning_rate": 4.0418118466898954e-05, "loss": 0.4, "step": 928 }, { "epoch": 0.16, "learning_rate": 4.04616724738676e-05, "loss": 0.4106, "step": 929 }, { "epoch": 0.16, "learning_rate": 4.050522648083624e-05, "loss": 0.4006, "step": 930 }, { "epoch": 0.16, "learning_rate": 4.0548780487804884e-05, "loss": 0.3915, "step": 931 }, { "epoch": 0.16, "learning_rate": 4.059233449477352e-05, "loss": 0.3937, "step": 932 }, { "epoch": 0.16, "learning_rate": 4.0635888501742164e-05, "loss": 0.408, "step": 933 }, { "epoch": 0.16, "learning_rate": 4.06794425087108e-05, "loss": 0.4141, "step": 934 }, { "epoch": 0.16, "learning_rate": 4.0722996515679444e-05, "loss": 0.4097, "step": 935 }, { "epoch": 0.16, "learning_rate": 4.076655052264808e-05, "loss": 0.3989, "step": 936 }, { "epoch": 0.16, "learning_rate": 4.081010452961673e-05, "loss": 0.3973, "step": 937 }, { "epoch": 0.16, "learning_rate": 4.085365853658537e-05, "loss": 0.3899, "step": 938 }, { "epoch": 0.16, "learning_rate": 4.089721254355401e-05, "loss": 0.4271, "step": 939 }, { "epoch": 0.16, "learning_rate": 4.0940766550522653e-05, "loss": 0.4203, "step": 940 }, { "epoch": 0.16, "learning_rate": 4.098432055749129e-05, "loss": 0.4008, "step": 941 }, { "epoch": 0.16, "learning_rate": 4.102787456445993e-05, "loss": 0.4111, "step": 942 }, { "epoch": 0.16, "learning_rate": 4.107142857142857e-05, "loss": 0.403, "step": 943 }, { "epoch": 0.16, "learning_rate": 4.111498257839721e-05, "loss": 0.4063, "step": 944 }, { "epoch": 0.16, "learning_rate": 4.1158536585365856e-05, "loss": 0.3999, "step": 945 }, { "epoch": 0.16, "learning_rate": 4.12020905923345e-05, "loss": 0.4074, "step": 946 }, { "epoch": 0.17, "learning_rate": 4.124564459930314e-05, "loss": 0.3861, "step": 947 }, { "epoch": 0.17, "learning_rate": 4.128919860627178e-05, "loss": 0.3886, "step": 948 }, { "epoch": 0.17, "learning_rate": 4.133275261324042e-05, "loss": 0.4216, "step": 949 }, { "epoch": 0.17, "learning_rate": 4.137630662020906e-05, "loss": 0.4202, "step": 950 }, { "epoch": 0.17, "learning_rate": 4.14198606271777e-05, "loss": 0.4035, "step": 951 }, { "epoch": 0.17, "learning_rate": 4.146341463414634e-05, "loss": 0.4014, "step": 952 }, { "epoch": 0.17, "learning_rate": 4.150696864111498e-05, "loss": 0.4009, "step": 953 }, { "epoch": 0.17, "learning_rate": 4.1550522648083626e-05, "loss": 0.4051, "step": 954 }, { "epoch": 0.17, "learning_rate": 4.159407665505227e-05, "loss": 0.4047, "step": 955 }, { "epoch": 0.17, "learning_rate": 4.163763066202091e-05, "loss": 0.4016, "step": 956 }, { "epoch": 0.17, "learning_rate": 4.168118466898955e-05, "loss": 0.4201, "step": 957 }, { "epoch": 0.17, "learning_rate": 4.172473867595819e-05, "loss": 0.3915, "step": 958 }, { "epoch": 0.17, "learning_rate": 4.176829268292683e-05, "loss": 0.4191, "step": 959 }, { "epoch": 0.17, "learning_rate": 4.181184668989547e-05, "loss": 0.4072, "step": 960 }, { "epoch": 0.17, "learning_rate": 4.185540069686411e-05, "loss": 0.3882, "step": 961 }, { "epoch": 0.17, "learning_rate": 4.189895470383275e-05, "loss": 0.4122, "step": 962 }, { "epoch": 0.17, "learning_rate": 4.1942508710801395e-05, "loss": 0.3895, "step": 963 }, { "epoch": 0.17, "learning_rate": 4.198606271777004e-05, "loss": 0.4031, "step": 964 }, { "epoch": 0.17, "learning_rate": 4.202961672473868e-05, "loss": 0.3941, "step": 965 }, { "epoch": 0.17, "learning_rate": 4.207317073170732e-05, "loss": 0.3891, "step": 966 }, { "epoch": 0.17, "learning_rate": 4.211672473867596e-05, "loss": 0.4031, "step": 967 }, { "epoch": 0.17, "learning_rate": 4.21602787456446e-05, "loss": 0.3908, "step": 968 }, { "epoch": 0.17, "learning_rate": 4.220383275261324e-05, "loss": 0.3948, "step": 969 }, { "epoch": 0.17, "learning_rate": 4.2247386759581884e-05, "loss": 0.3892, "step": 970 }, { "epoch": 0.17, "learning_rate": 4.229094076655053e-05, "loss": 0.3939, "step": 971 }, { "epoch": 0.17, "learning_rate": 4.233449477351917e-05, "loss": 0.4116, "step": 972 }, { "epoch": 0.17, "learning_rate": 4.237804878048781e-05, "loss": 0.3998, "step": 973 }, { "epoch": 0.17, "learning_rate": 4.242160278745645e-05, "loss": 0.3953, "step": 974 }, { "epoch": 0.17, "learning_rate": 4.246515679442509e-05, "loss": 0.4145, "step": 975 }, { "epoch": 0.17, "learning_rate": 4.250871080139373e-05, "loss": 0.4015, "step": 976 }, { "epoch": 0.17, "learning_rate": 4.255226480836237e-05, "loss": 0.3902, "step": 977 }, { "epoch": 0.17, "learning_rate": 4.259581881533101e-05, "loss": 0.4201, "step": 978 }, { "epoch": 0.17, "learning_rate": 4.2639372822299653e-05, "loss": 0.4108, "step": 979 }, { "epoch": 0.17, "learning_rate": 4.26829268292683e-05, "loss": 0.4064, "step": 980 }, { "epoch": 0.17, "learning_rate": 4.272648083623694e-05, "loss": 0.3795, "step": 981 }, { "epoch": 0.17, "learning_rate": 4.2770034843205577e-05, "loss": 0.4078, "step": 982 }, { "epoch": 0.17, "learning_rate": 4.281358885017422e-05, "loss": 0.3907, "step": 983 }, { "epoch": 0.17, "learning_rate": 4.2857142857142856e-05, "loss": 0.4157, "step": 984 }, { "epoch": 0.17, "learning_rate": 4.29006968641115e-05, "loss": 0.4075, "step": 985 }, { "epoch": 0.17, "learning_rate": 4.294425087108014e-05, "loss": 0.3936, "step": 986 }, { "epoch": 0.17, "learning_rate": 4.298780487804878e-05, "loss": 0.3815, "step": 987 }, { "epoch": 0.17, "learning_rate": 4.303135888501742e-05, "loss": 0.4152, "step": 988 }, { "epoch": 0.17, "learning_rate": 4.3074912891986066e-05, "loss": 0.4222, "step": 989 }, { "epoch": 0.17, "learning_rate": 4.311846689895471e-05, "loss": 0.4167, "step": 990 }, { "epoch": 0.17, "learning_rate": 4.3162020905923346e-05, "loss": 0.4195, "step": 991 }, { "epoch": 0.17, "learning_rate": 4.320557491289199e-05, "loss": 0.4125, "step": 992 }, { "epoch": 0.17, "learning_rate": 4.324912891986063e-05, "loss": 0.4109, "step": 993 }, { "epoch": 0.17, "learning_rate": 4.329268292682927e-05, "loss": 0.398, "step": 994 }, { "epoch": 0.17, "learning_rate": 4.333623693379791e-05, "loss": 0.4115, "step": 995 }, { "epoch": 0.17, "learning_rate": 4.337979094076655e-05, "loss": 0.4105, "step": 996 }, { "epoch": 0.17, "learning_rate": 4.342334494773519e-05, "loss": 0.3873, "step": 997 }, { "epoch": 0.17, "learning_rate": 4.3466898954703835e-05, "loss": 0.404, "step": 998 }, { "epoch": 0.17, "learning_rate": 4.351045296167248e-05, "loss": 0.4045, "step": 999 }, { "epoch": 0.17, "learning_rate": 4.3554006968641115e-05, "loss": 0.4043, "step": 1000 }, { "epoch": 0.17, "learning_rate": 4.359756097560976e-05, "loss": 0.3924, "step": 1001 }, { "epoch": 0.17, "learning_rate": 4.36411149825784e-05, "loss": 0.4117, "step": 1002 }, { "epoch": 0.17, "learning_rate": 4.368466898954704e-05, "loss": 0.4209, "step": 1003 }, { "epoch": 0.17, "learning_rate": 4.372822299651568e-05, "loss": 0.4067, "step": 1004 }, { "epoch": 0.18, "learning_rate": 4.377177700348432e-05, "loss": 0.4131, "step": 1005 }, { "epoch": 0.18, "learning_rate": 4.381533101045297e-05, "loss": 0.3954, "step": 1006 }, { "epoch": 0.18, "learning_rate": 4.3858885017421604e-05, "loss": 0.4093, "step": 1007 }, { "epoch": 0.18, "learning_rate": 4.390243902439025e-05, "loss": 0.3952, "step": 1008 }, { "epoch": 0.18, "learning_rate": 4.394599303135889e-05, "loss": 0.4127, "step": 1009 }, { "epoch": 0.18, "learning_rate": 4.398954703832753e-05, "loss": 0.396, "step": 1010 }, { "epoch": 0.18, "learning_rate": 4.403310104529617e-05, "loss": 0.4099, "step": 1011 }, { "epoch": 0.18, "learning_rate": 4.407665505226481e-05, "loss": 0.4099, "step": 1012 }, { "epoch": 0.18, "learning_rate": 4.412020905923345e-05, "loss": 0.4082, "step": 1013 }, { "epoch": 0.18, "learning_rate": 4.4163763066202094e-05, "loss": 0.4224, "step": 1014 }, { "epoch": 0.18, "learning_rate": 4.420731707317074e-05, "loss": 0.4039, "step": 1015 }, { "epoch": 0.18, "learning_rate": 4.4250871080139374e-05, "loss": 0.4012, "step": 1016 }, { "epoch": 0.18, "learning_rate": 4.429442508710802e-05, "loss": 0.4045, "step": 1017 }, { "epoch": 0.18, "learning_rate": 4.433797909407666e-05, "loss": 0.3959, "step": 1018 }, { "epoch": 0.18, "learning_rate": 4.43815331010453e-05, "loss": 0.3966, "step": 1019 }, { "epoch": 0.18, "learning_rate": 4.442508710801394e-05, "loss": 0.3961, "step": 1020 }, { "epoch": 0.18, "learning_rate": 4.4468641114982576e-05, "loss": 0.3987, "step": 1021 }, { "epoch": 0.18, "learning_rate": 4.451219512195122e-05, "loss": 0.3981, "step": 1022 }, { "epoch": 0.18, "learning_rate": 4.455574912891986e-05, "loss": 0.4034, "step": 1023 }, { "epoch": 0.18, "learning_rate": 4.4599303135888506e-05, "loss": 0.4173, "step": 1024 }, { "epoch": 0.18, "learning_rate": 4.464285714285715e-05, "loss": 0.421, "step": 1025 }, { "epoch": 0.18, "learning_rate": 4.4686411149825786e-05, "loss": 0.3905, "step": 1026 }, { "epoch": 0.18, "learning_rate": 4.472996515679443e-05, "loss": 0.3987, "step": 1027 }, { "epoch": 0.18, "learning_rate": 4.4773519163763066e-05, "loss": 0.4033, "step": 1028 }, { "epoch": 0.18, "learning_rate": 4.481707317073171e-05, "loss": 0.3955, "step": 1029 }, { "epoch": 0.18, "learning_rate": 4.4860627177700346e-05, "loss": 0.4022, "step": 1030 }, { "epoch": 0.18, "learning_rate": 4.490418118466899e-05, "loss": 0.407, "step": 1031 }, { "epoch": 0.18, "learning_rate": 4.494773519163763e-05, "loss": 0.4023, "step": 1032 }, { "epoch": 0.18, "learning_rate": 4.4991289198606276e-05, "loss": 0.4283, "step": 1033 }, { "epoch": 0.18, "learning_rate": 4.503484320557492e-05, "loss": 0.4037, "step": 1034 }, { "epoch": 0.18, "learning_rate": 4.5078397212543555e-05, "loss": 0.3981, "step": 1035 }, { "epoch": 0.18, "learning_rate": 4.51219512195122e-05, "loss": 0.4032, "step": 1036 }, { "epoch": 0.18, "learning_rate": 4.5165505226480835e-05, "loss": 0.4017, "step": 1037 }, { "epoch": 0.18, "learning_rate": 4.520905923344948e-05, "loss": 0.4021, "step": 1038 }, { "epoch": 0.18, "learning_rate": 4.5252613240418115e-05, "loss": 0.4168, "step": 1039 }, { "epoch": 0.18, "learning_rate": 4.529616724738676e-05, "loss": 0.4067, "step": 1040 }, { "epoch": 0.18, "learning_rate": 4.533972125435541e-05, "loss": 0.3838, "step": 1041 }, { "epoch": 0.18, "learning_rate": 4.5383275261324045e-05, "loss": 0.394, "step": 1042 }, { "epoch": 0.18, "learning_rate": 4.542682926829269e-05, "loss": 0.3912, "step": 1043 }, { "epoch": 0.18, "learning_rate": 4.5470383275261325e-05, "loss": 0.3934, "step": 1044 }, { "epoch": 0.18, "learning_rate": 4.551393728222997e-05, "loss": 0.4039, "step": 1045 }, { "epoch": 0.18, "learning_rate": 4.5557491289198604e-05, "loss": 0.4076, "step": 1046 }, { "epoch": 0.18, "learning_rate": 4.560104529616725e-05, "loss": 0.3937, "step": 1047 }, { "epoch": 0.18, "learning_rate": 4.564459930313589e-05, "loss": 0.4585, "step": 1048 }, { "epoch": 0.18, "learning_rate": 4.5688153310104534e-05, "loss": 0.39, "step": 1049 }, { "epoch": 0.18, "learning_rate": 4.573170731707318e-05, "loss": 0.4094, "step": 1050 }, { "epoch": 0.18, "learning_rate": 4.5775261324041814e-05, "loss": 0.4075, "step": 1051 }, { "epoch": 0.18, "learning_rate": 4.581881533101046e-05, "loss": 0.4032, "step": 1052 }, { "epoch": 0.18, "learning_rate": 4.5862369337979094e-05, "loss": 0.421, "step": 1053 }, { "epoch": 0.18, "learning_rate": 4.590592334494774e-05, "loss": 0.4028, "step": 1054 }, { "epoch": 0.18, "learning_rate": 4.594947735191638e-05, "loss": 0.3949, "step": 1055 }, { "epoch": 0.18, "learning_rate": 4.599303135888502e-05, "loss": 0.4153, "step": 1056 }, { "epoch": 0.18, "learning_rate": 4.603658536585366e-05, "loss": 0.4054, "step": 1057 }, { "epoch": 0.18, "learning_rate": 4.6080139372822303e-05, "loss": 0.3893, "step": 1058 }, { "epoch": 0.18, "learning_rate": 4.612369337979095e-05, "loss": 0.3975, "step": 1059 }, { "epoch": 0.18, "learning_rate": 4.616724738675958e-05, "loss": 0.4102, "step": 1060 }, { "epoch": 0.18, "learning_rate": 4.6210801393728226e-05, "loss": 0.4135, "step": 1061 }, { "epoch": 0.19, "learning_rate": 4.625435540069686e-05, "loss": 0.3958, "step": 1062 }, { "epoch": 0.19, "learning_rate": 4.6297909407665506e-05, "loss": 0.4129, "step": 1063 }, { "epoch": 0.19, "learning_rate": 4.634146341463415e-05, "loss": 0.3989, "step": 1064 }, { "epoch": 0.19, "learning_rate": 4.6385017421602786e-05, "loss": 0.4212, "step": 1065 }, { "epoch": 0.19, "learning_rate": 4.642857142857143e-05, "loss": 0.3769, "step": 1066 }, { "epoch": 0.19, "learning_rate": 4.647212543554007e-05, "loss": 0.3946, "step": 1067 }, { "epoch": 0.19, "learning_rate": 4.6515679442508716e-05, "loss": 0.3957, "step": 1068 }, { "epoch": 0.19, "learning_rate": 4.655923344947735e-05, "loss": 0.4103, "step": 1069 }, { "epoch": 0.19, "learning_rate": 4.6602787456445996e-05, "loss": 0.3987, "step": 1070 }, { "epoch": 0.19, "learning_rate": 4.664634146341464e-05, "loss": 0.386, "step": 1071 }, { "epoch": 0.19, "learning_rate": 4.6689895470383275e-05, "loss": 0.4015, "step": 1072 }, { "epoch": 0.19, "learning_rate": 4.673344947735192e-05, "loss": 0.3895, "step": 1073 }, { "epoch": 0.19, "learning_rate": 4.6777003484320555e-05, "loss": 0.393, "step": 1074 }, { "epoch": 0.19, "learning_rate": 4.6820557491289205e-05, "loss": 0.4162, "step": 1075 }, { "epoch": 0.19, "learning_rate": 4.686411149825784e-05, "loss": 0.3984, "step": 1076 }, { "epoch": 0.19, "learning_rate": 4.6907665505226485e-05, "loss": 0.375, "step": 1077 }, { "epoch": 0.19, "learning_rate": 4.695121951219512e-05, "loss": 0.4103, "step": 1078 }, { "epoch": 0.19, "learning_rate": 4.6994773519163765e-05, "loss": 0.4005, "step": 1079 }, { "epoch": 0.19, "learning_rate": 4.703832752613241e-05, "loss": 0.3984, "step": 1080 }, { "epoch": 0.19, "learning_rate": 4.7081881533101045e-05, "loss": 0.3961, "step": 1081 }, { "epoch": 0.19, "learning_rate": 4.712543554006969e-05, "loss": 0.3925, "step": 1082 }, { "epoch": 0.19, "learning_rate": 4.716898954703833e-05, "loss": 0.4001, "step": 1083 }, { "epoch": 0.19, "learning_rate": 4.7212543554006975e-05, "loss": 0.3833, "step": 1084 }, { "epoch": 0.19, "learning_rate": 4.725609756097561e-05, "loss": 0.4531, "step": 1085 }, { "epoch": 0.19, "learning_rate": 4.7299651567944254e-05, "loss": 0.4066, "step": 1086 }, { "epoch": 0.19, "learning_rate": 4.73432055749129e-05, "loss": 0.4109, "step": 1087 }, { "epoch": 0.19, "learning_rate": 4.7386759581881534e-05, "loss": 0.4035, "step": 1088 }, { "epoch": 0.19, "learning_rate": 4.743031358885018e-05, "loss": 0.3968, "step": 1089 }, { "epoch": 0.19, "learning_rate": 4.7473867595818814e-05, "loss": 0.4053, "step": 1090 }, { "epoch": 0.19, "learning_rate": 4.751742160278746e-05, "loss": 0.4017, "step": 1091 }, { "epoch": 0.19, "learning_rate": 4.75609756097561e-05, "loss": 0.3913, "step": 1092 }, { "epoch": 0.19, "learning_rate": 4.7604529616724744e-05, "loss": 0.4092, "step": 1093 }, { "epoch": 0.19, "learning_rate": 4.764808362369339e-05, "loss": 0.4222, "step": 1094 }, { "epoch": 0.19, "learning_rate": 4.7691637630662024e-05, "loss": 0.4085, "step": 1095 }, { "epoch": 0.19, "learning_rate": 4.773519163763067e-05, "loss": 0.4336, "step": 1096 }, { "epoch": 0.19, "learning_rate": 4.77787456445993e-05, "loss": 0.3978, "step": 1097 }, { "epoch": 0.19, "learning_rate": 4.782229965156795e-05, "loss": 0.4134, "step": 1098 }, { "epoch": 0.19, "learning_rate": 4.786585365853658e-05, "loss": 0.4037, "step": 1099 }, { "epoch": 0.19, "learning_rate": 4.7909407665505226e-05, "loss": 0.3972, "step": 1100 }, { "epoch": 0.19, "learning_rate": 4.795296167247387e-05, "loss": 0.3993, "step": 1101 }, { "epoch": 0.19, "learning_rate": 4.799651567944251e-05, "loss": 0.4007, "step": 1102 }, { "epoch": 0.19, "learning_rate": 4.8040069686411156e-05, "loss": 0.4027, "step": 1103 }, { "epoch": 0.19, "learning_rate": 4.808362369337979e-05, "loss": 0.4, "step": 1104 }, { "epoch": 0.19, "learning_rate": 4.8127177700348436e-05, "loss": 0.4066, "step": 1105 }, { "epoch": 0.19, "learning_rate": 4.817073170731707e-05, "loss": 0.3943, "step": 1106 }, { "epoch": 0.19, "learning_rate": 4.8214285714285716e-05, "loss": 0.4007, "step": 1107 }, { "epoch": 0.19, "learning_rate": 4.825783972125435e-05, "loss": 0.3861, "step": 1108 }, { "epoch": 0.19, "learning_rate": 4.8301393728222996e-05, "loss": 0.3986, "step": 1109 }, { "epoch": 0.19, "learning_rate": 4.8344947735191646e-05, "loss": 0.4084, "step": 1110 }, { "epoch": 0.19, "learning_rate": 4.838850174216028e-05, "loss": 0.4171, "step": 1111 }, { "epoch": 0.19, "learning_rate": 4.8432055749128926e-05, "loss": 0.4147, "step": 1112 }, { "epoch": 0.19, "learning_rate": 4.847560975609756e-05, "loss": 0.4033, "step": 1113 }, { "epoch": 0.19, "learning_rate": 4.8519163763066205e-05, "loss": 0.3888, "step": 1114 }, { "epoch": 0.19, "learning_rate": 4.856271777003484e-05, "loss": 0.4088, "step": 1115 }, { "epoch": 0.19, "learning_rate": 4.8606271777003485e-05, "loss": 0.4026, "step": 1116 }, { "epoch": 0.19, "learning_rate": 4.864982578397213e-05, "loss": 0.3806, "step": 1117 }, { "epoch": 0.19, "learning_rate": 4.869337979094077e-05, "loss": 0.4048, "step": 1118 }, { "epoch": 0.2, "learning_rate": 4.8736933797909415e-05, "loss": 0.424, "step": 1119 }, { "epoch": 0.2, "learning_rate": 4.878048780487805e-05, "loss": 0.4111, "step": 1120 }, { "epoch": 0.2, "learning_rate": 4.8824041811846695e-05, "loss": 0.4041, "step": 1121 }, { "epoch": 0.2, "learning_rate": 4.886759581881533e-05, "loss": 0.3948, "step": 1122 }, { "epoch": 0.2, "learning_rate": 4.8911149825783975e-05, "loss": 0.4196, "step": 1123 }, { "epoch": 0.2, "learning_rate": 4.895470383275261e-05, "loss": 0.4163, "step": 1124 }, { "epoch": 0.2, "learning_rate": 4.8998257839721254e-05, "loss": 0.402, "step": 1125 }, { "epoch": 0.2, "learning_rate": 4.90418118466899e-05, "loss": 0.391, "step": 1126 }, { "epoch": 0.2, "learning_rate": 4.908536585365854e-05, "loss": 0.4015, "step": 1127 }, { "epoch": 0.2, "learning_rate": 4.9128919860627184e-05, "loss": 0.4033, "step": 1128 }, { "epoch": 0.2, "learning_rate": 4.917247386759582e-05, "loss": 0.3983, "step": 1129 }, { "epoch": 0.2, "learning_rate": 4.9216027874564464e-05, "loss": 0.4084, "step": 1130 }, { "epoch": 0.2, "learning_rate": 4.92595818815331e-05, "loss": 0.4062, "step": 1131 }, { "epoch": 0.2, "learning_rate": 4.9303135888501744e-05, "loss": 0.4045, "step": 1132 }, { "epoch": 0.2, "learning_rate": 4.934668989547039e-05, "loss": 0.3965, "step": 1133 }, { "epoch": 0.2, "learning_rate": 4.9390243902439024e-05, "loss": 0.3975, "step": 1134 }, { "epoch": 0.2, "learning_rate": 4.943379790940767e-05, "loss": 0.3794, "step": 1135 }, { "epoch": 0.2, "learning_rate": 4.947735191637631e-05, "loss": 0.4071, "step": 1136 }, { "epoch": 0.2, "learning_rate": 4.952090592334495e-05, "loss": 0.4062, "step": 1137 }, { "epoch": 0.2, "learning_rate": 4.956445993031359e-05, "loss": 0.3876, "step": 1138 }, { "epoch": 0.2, "learning_rate": 4.960801393728223e-05, "loss": 0.3978, "step": 1139 }, { "epoch": 0.2, "learning_rate": 4.965156794425087e-05, "loss": 0.3963, "step": 1140 }, { "epoch": 0.2, "learning_rate": 4.969512195121951e-05, "loss": 0.3854, "step": 1141 }, { "epoch": 0.2, "learning_rate": 4.9738675958188156e-05, "loss": 0.4061, "step": 1142 }, { "epoch": 0.2, "learning_rate": 4.978222996515679e-05, "loss": 0.3921, "step": 1143 }, { "epoch": 0.2, "learning_rate": 4.9825783972125436e-05, "loss": 0.4031, "step": 1144 }, { "epoch": 0.2, "learning_rate": 4.986933797909408e-05, "loss": 0.3978, "step": 1145 }, { "epoch": 0.2, "learning_rate": 4.991289198606272e-05, "loss": 0.3896, "step": 1146 }, { "epoch": 0.2, "learning_rate": 4.995644599303136e-05, "loss": 0.3933, "step": 1147 }, { "epoch": 0.2, "learning_rate": 5e-05, "loss": 0.4059, "step": 1148 }, { "epoch": 0.2, "learning_rate": 4.9999998958670857e-05, "loss": 0.4093, "step": 1149 }, { "epoch": 0.2, "learning_rate": 4.999999583468353e-05, "loss": 0.4011, "step": 1150 }, { "epoch": 0.2, "learning_rate": 4.999999062803829e-05, "loss": 0.3998, "step": 1151 }, { "epoch": 0.2, "learning_rate": 4.9999983338735634e-05, "loss": 0.399, "step": 1152 }, { "epoch": 0.2, "learning_rate": 4.9999973966776235e-05, "loss": 0.3987, "step": 1153 }, { "epoch": 0.2, "learning_rate": 4.999996251216096e-05, "loss": 0.4022, "step": 1154 }, { "epoch": 0.2, "learning_rate": 4.999994897489086e-05, "loss": 0.4031, "step": 1155 }, { "epoch": 0.2, "learning_rate": 4.999993335496721e-05, "loss": 0.3992, "step": 1156 }, { "epoch": 0.2, "learning_rate": 4.9999915652391433e-05, "loss": 0.3998, "step": 1157 }, { "epoch": 0.2, "learning_rate": 4.9999895867165174e-05, "loss": 0.391, "step": 1158 }, { "epoch": 0.2, "learning_rate": 4.999987399929027e-05, "loss": 0.3961, "step": 1159 }, { "epoch": 0.2, "learning_rate": 4.999985004876875e-05, "loss": 0.401, "step": 1160 }, { "epoch": 0.2, "learning_rate": 4.999982401560282e-05, "loss": 0.399, "step": 1161 }, { "epoch": 0.2, "learning_rate": 4.999979589979489e-05, "loss": 0.4026, "step": 1162 }, { "epoch": 0.2, "learning_rate": 4.9999765701347556e-05, "loss": 0.4346, "step": 1163 }, { "epoch": 0.2, "learning_rate": 4.9999733420263635e-05, "loss": 0.4002, "step": 1164 }, { "epoch": 0.2, "learning_rate": 4.9999699056546096e-05, "loss": 0.4193, "step": 1165 }, { "epoch": 0.2, "learning_rate": 4.999966261019813e-05, "loss": 0.4032, "step": 1166 }, { "epoch": 0.2, "learning_rate": 4.99996240812231e-05, "loss": 0.4023, "step": 1167 }, { "epoch": 0.2, "learning_rate": 4.9999583469624585e-05, "loss": 0.3973, "step": 1168 }, { "epoch": 0.2, "learning_rate": 4.999954077540634e-05, "loss": 0.4008, "step": 1169 }, { "epoch": 0.2, "learning_rate": 4.999949599857231e-05, "loss": 0.3952, "step": 1170 }, { "epoch": 0.2, "learning_rate": 4.999944913912665e-05, "loss": 0.3982, "step": 1171 }, { "epoch": 0.2, "learning_rate": 4.999940019707369e-05, "loss": 0.4082, "step": 1172 }, { "epoch": 0.2, "learning_rate": 4.9999349172417956e-05, "loss": 0.3993, "step": 1173 }, { "epoch": 0.2, "learning_rate": 4.9999296065164186e-05, "loss": 0.4126, "step": 1174 }, { "epoch": 0.2, "learning_rate": 4.9999240875317285e-05, "loss": 0.405, "step": 1175 }, { "epoch": 0.2, "learning_rate": 4.999918360288237e-05, "loss": 0.4142, "step": 1176 }, { "epoch": 0.21, "learning_rate": 4.999912424786472e-05, "loss": 0.3837, "step": 1177 }, { "epoch": 0.21, "learning_rate": 4.999906281026985e-05, "loss": 0.3909, "step": 1178 }, { "epoch": 0.21, "learning_rate": 4.999899929010345e-05, "loss": 0.3892, "step": 1179 }, { "epoch": 0.21, "learning_rate": 4.999893368737139e-05, "loss": 0.3958, "step": 1180 }, { "epoch": 0.21, "learning_rate": 4.999886600207974e-05, "loss": 0.3921, "step": 1181 }, { "epoch": 0.21, "learning_rate": 4.999879623423476e-05, "loss": 0.3995, "step": 1182 }, { "epoch": 0.21, "learning_rate": 4.999872438384293e-05, "loss": 0.3838, "step": 1183 }, { "epoch": 0.21, "learning_rate": 4.999865045091089e-05, "loss": 0.4024, "step": 1184 }, { "epoch": 0.21, "learning_rate": 4.999857443544547e-05, "loss": 0.4072, "step": 1185 }, { "epoch": 0.21, "learning_rate": 4.9998496337453726e-05, "loss": 0.4028, "step": 1186 }, { "epoch": 0.21, "learning_rate": 4.999841615694287e-05, "loss": 0.4261, "step": 1187 }, { "epoch": 0.21, "learning_rate": 4.9998333893920336e-05, "loss": 0.4023, "step": 1188 }, { "epoch": 0.21, "learning_rate": 4.999824954839374e-05, "loss": 0.3891, "step": 1189 }, { "epoch": 0.21, "learning_rate": 4.9998163120370874e-05, "loss": 0.3877, "step": 1190 }, { "epoch": 0.21, "learning_rate": 4.999807460985975e-05, "loss": 0.3918, "step": 1191 }, { "epoch": 0.21, "learning_rate": 4.999798401686856e-05, "loss": 0.4019, "step": 1192 }, { "epoch": 0.21, "learning_rate": 4.999789134140569e-05, "loss": 0.4045, "step": 1193 }, { "epoch": 0.21, "learning_rate": 4.99977965834797e-05, "loss": 0.398, "step": 1194 }, { "epoch": 0.21, "learning_rate": 4.99976997430994e-05, "loss": 0.4115, "step": 1195 }, { "epoch": 0.21, "learning_rate": 4.999760082027372e-05, "loss": 0.3897, "step": 1196 }, { "epoch": 0.21, "learning_rate": 4.999749981501183e-05, "loss": 0.3927, "step": 1197 }, { "epoch": 0.21, "learning_rate": 4.999739672732307e-05, "loss": 0.3793, "step": 1198 }, { "epoch": 0.21, "learning_rate": 4.999729155721699e-05, "loss": 0.4018, "step": 1199 }, { "epoch": 0.21, "learning_rate": 4.999718430470333e-05, "loss": 0.3906, "step": 1200 }, { "epoch": 0.21, "learning_rate": 4.999707496979201e-05, "loss": 0.389, "step": 1201 }, { "epoch": 0.21, "learning_rate": 4.9996963552493145e-05, "loss": 0.4051, "step": 1202 }, { "epoch": 0.21, "learning_rate": 4.9996850052817064e-05, "loss": 0.4068, "step": 1203 }, { "epoch": 0.21, "learning_rate": 4.999673447077425e-05, "loss": 0.4079, "step": 1204 }, { "epoch": 0.21, "learning_rate": 4.999661680637543e-05, "loss": 0.3901, "step": 1205 }, { "epoch": 0.21, "learning_rate": 4.999649705963148e-05, "loss": 0.3976, "step": 1206 }, { "epoch": 0.21, "learning_rate": 4.999637523055348e-05, "loss": 0.3943, "step": 1207 }, { "epoch": 0.21, "learning_rate": 4.9996251319152704e-05, "loss": 0.393, "step": 1208 }, { "epoch": 0.21, "learning_rate": 4.999612532544063e-05, "loss": 0.404, "step": 1209 }, { "epoch": 0.21, "learning_rate": 4.999599724942893e-05, "loss": 0.4022, "step": 1210 }, { "epoch": 0.21, "learning_rate": 4.999586709112944e-05, "loss": 0.3976, "step": 1211 }, { "epoch": 0.21, "learning_rate": 4.999573485055422e-05, "loss": 0.3909, "step": 1212 }, { "epoch": 0.21, "learning_rate": 4.99956005277155e-05, "loss": 0.3959, "step": 1213 }, { "epoch": 0.21, "learning_rate": 4.9995464122625726e-05, "loss": 0.4047, "step": 1214 }, { "epoch": 0.21, "learning_rate": 4.9995325635297506e-05, "loss": 0.3997, "step": 1215 }, { "epoch": 0.21, "learning_rate": 4.999518506574368e-05, "loss": 0.4032, "step": 1216 }, { "epoch": 0.21, "learning_rate": 4.999504241397725e-05, "loss": 0.3992, "step": 1217 }, { "epoch": 0.21, "learning_rate": 4.9994897680011416e-05, "loss": 0.3966, "step": 1218 }, { "epoch": 0.21, "learning_rate": 4.999475086385958e-05, "loss": 0.4064, "step": 1219 }, { "epoch": 0.21, "learning_rate": 4.9994601965535344e-05, "loss": 0.4024, "step": 1220 }, { "epoch": 0.21, "learning_rate": 4.999445098505247e-05, "loss": 0.3986, "step": 1221 }, { "epoch": 0.21, "learning_rate": 4.9994297922424926e-05, "loss": 0.3958, "step": 1222 }, { "epoch": 0.21, "learning_rate": 4.999414277766691e-05, "loss": 0.4069, "step": 1223 }, { "epoch": 0.21, "learning_rate": 4.999398555079276e-05, "loss": 0.3928, "step": 1224 }, { "epoch": 0.21, "learning_rate": 4.999382624181704e-05, "loss": 0.3752, "step": 1225 }, { "epoch": 0.21, "learning_rate": 4.999366485075448e-05, "loss": 0.3975, "step": 1226 }, { "epoch": 0.21, "learning_rate": 4.999350137762005e-05, "loss": 0.4147, "step": 1227 }, { "epoch": 0.21, "learning_rate": 4.9993335822428846e-05, "loss": 0.4062, "step": 1228 }, { "epoch": 0.21, "learning_rate": 4.999316818519621e-05, "loss": 0.4102, "step": 1229 }, { "epoch": 0.21, "learning_rate": 4.9992998465937674e-05, "loss": 0.3994, "step": 1230 }, { "epoch": 0.21, "learning_rate": 4.999282666466892e-05, "loss": 0.3896, "step": 1231 }, { "epoch": 0.21, "learning_rate": 4.999265278140586e-05, "loss": 0.3931, "step": 1232 }, { "epoch": 0.21, "learning_rate": 4.9992476816164594e-05, "loss": 0.4149, "step": 1233 }, { "epoch": 0.22, "learning_rate": 4.9992298768961406e-05, "loss": 0.4051, "step": 1234 }, { "epoch": 0.22, "learning_rate": 4.9992118639812775e-05, "loss": 0.4162, "step": 1235 }, { "epoch": 0.22, "learning_rate": 4.999193642873537e-05, "loss": 0.388, "step": 1236 }, { "epoch": 0.22, "learning_rate": 4.9991752135746076e-05, "loss": 0.4041, "step": 1237 }, { "epoch": 0.22, "learning_rate": 4.9991565760861936e-05, "loss": 0.4215, "step": 1238 }, { "epoch": 0.22, "learning_rate": 4.9991377304100204e-05, "loss": 0.3961, "step": 1239 }, { "epoch": 0.22, "learning_rate": 4.999118676547832e-05, "loss": 0.3993, "step": 1240 }, { "epoch": 0.22, "learning_rate": 4.9990994145013924e-05, "loss": 0.3879, "step": 1241 }, { "epoch": 0.22, "learning_rate": 4.9990799442724856e-05, "loss": 0.3935, "step": 1242 }, { "epoch": 0.22, "learning_rate": 4.999060265862912e-05, "loss": 0.412, "step": 1243 }, { "epoch": 0.22, "learning_rate": 4.999040379274494e-05, "loss": 0.3996, "step": 1244 }, { "epoch": 0.22, "learning_rate": 4.999020284509072e-05, "loss": 0.4077, "step": 1245 }, { "epoch": 0.22, "learning_rate": 4.9989999815685076e-05, "loss": 0.3937, "step": 1246 }, { "epoch": 0.22, "learning_rate": 4.998979470454677e-05, "loss": 0.3831, "step": 1247 }, { "epoch": 0.22, "learning_rate": 4.998958751169482e-05, "loss": 0.3926, "step": 1248 }, { "epoch": 0.22, "learning_rate": 4.998937823714838e-05, "loss": 0.3865, "step": 1249 }, { "epoch": 0.22, "learning_rate": 4.9989166880926834e-05, "loss": 0.3735, "step": 1250 }, { "epoch": 0.22, "learning_rate": 4.9988953443049755e-05, "loss": 0.4125, "step": 1251 }, { "epoch": 0.22, "learning_rate": 4.9988737923536874e-05, "loss": 0.3936, "step": 1252 }, { "epoch": 0.22, "learning_rate": 4.9988520322408155e-05, "loss": 0.3869, "step": 1253 }, { "epoch": 0.22, "learning_rate": 4.998830063968374e-05, "loss": 0.4058, "step": 1254 }, { "epoch": 0.22, "learning_rate": 4.9988078875383956e-05, "loss": 0.3942, "step": 1255 }, { "epoch": 0.22, "learning_rate": 4.998785502952934e-05, "loss": 0.3853, "step": 1256 }, { "epoch": 0.22, "learning_rate": 4.998762910214061e-05, "loss": 0.4064, "step": 1257 }, { "epoch": 0.22, "learning_rate": 4.998740109323867e-05, "loss": 0.3925, "step": 1258 }, { "epoch": 0.22, "learning_rate": 4.998717100284463e-05, "loss": 0.4029, "step": 1259 }, { "epoch": 0.22, "learning_rate": 4.998693883097979e-05, "loss": 0.4054, "step": 1260 }, { "epoch": 0.22, "learning_rate": 4.9986704577665646e-05, "loss": 0.4083, "step": 1261 }, { "epoch": 0.22, "learning_rate": 4.9986468242923864e-05, "loss": 0.4019, "step": 1262 }, { "epoch": 0.22, "learning_rate": 4.998622982677633e-05, "loss": 0.4036, "step": 1263 }, { "epoch": 0.22, "learning_rate": 4.998598932924511e-05, "loss": 0.4176, "step": 1264 }, { "epoch": 0.22, "learning_rate": 4.998574675035248e-05, "loss": 0.4009, "step": 1265 }, { "epoch": 0.22, "learning_rate": 4.9985502090120867e-05, "loss": 0.387, "step": 1266 }, { "epoch": 0.22, "learning_rate": 4.998525534857294e-05, "loss": 0.4021, "step": 1267 }, { "epoch": 0.22, "learning_rate": 4.9985006525731524e-05, "loss": 0.4026, "step": 1268 }, { "epoch": 0.22, "learning_rate": 4.998475562161966e-05, "loss": 0.4076, "step": 1269 }, { "epoch": 0.22, "learning_rate": 4.998450263626056e-05, "loss": 0.4041, "step": 1270 }, { "epoch": 0.22, "learning_rate": 4.998424756967766e-05, "loss": 0.3899, "step": 1271 }, { "epoch": 0.22, "learning_rate": 4.9983990421894555e-05, "loss": 0.3899, "step": 1272 }, { "epoch": 0.22, "learning_rate": 4.998373119293505e-05, "loss": 0.4104, "step": 1273 }, { "epoch": 0.22, "learning_rate": 4.998346988282314e-05, "loss": 0.3961, "step": 1274 }, { "epoch": 0.22, "learning_rate": 4.9983206491583014e-05, "loss": 0.4043, "step": 1275 }, { "epoch": 0.22, "learning_rate": 4.998294101923906e-05, "loss": 0.3888, "step": 1276 }, { "epoch": 0.22, "learning_rate": 4.9982673465815826e-05, "loss": 0.3811, "step": 1277 }, { "epoch": 0.22, "learning_rate": 4.9982403831338104e-05, "loss": 0.3855, "step": 1278 }, { "epoch": 0.22, "learning_rate": 4.998213211583084e-05, "loss": 0.3987, "step": 1279 }, { "epoch": 0.22, "learning_rate": 4.998185831931919e-05, "loss": 0.4034, "step": 1280 }, { "epoch": 0.22, "learning_rate": 4.998158244182849e-05, "loss": 0.3874, "step": 1281 }, { "epoch": 0.22, "learning_rate": 4.998130448338428e-05, "loss": 0.395, "step": 1282 }, { "epoch": 0.22, "learning_rate": 4.99810244440123e-05, "loss": 0.3873, "step": 1283 }, { "epoch": 0.22, "learning_rate": 4.998074232373845e-05, "loss": 0.4005, "step": 1284 }, { "epoch": 0.22, "learning_rate": 4.9980458122588855e-05, "loss": 0.3908, "step": 1285 }, { "epoch": 0.22, "learning_rate": 4.998017184058982e-05, "loss": 0.3961, "step": 1286 }, { "epoch": 0.22, "learning_rate": 4.9979883477767835e-05, "loss": 0.3861, "step": 1287 }, { "epoch": 0.22, "learning_rate": 4.9979593034149616e-05, "loss": 0.3875, "step": 1288 }, { "epoch": 0.22, "learning_rate": 4.9979300509762017e-05, "loss": 0.3869, "step": 1289 }, { "epoch": 0.22, "learning_rate": 4.9979005904632134e-05, "loss": 0.4003, "step": 1290 }, { "epoch": 0.23, "learning_rate": 4.9978709218787236e-05, "loss": 0.3953, "step": 1291 }, { "epoch": 0.23, "learning_rate": 4.997841045225478e-05, "loss": 0.4063, "step": 1292 }, { "epoch": 0.23, "learning_rate": 4.997810960506242e-05, "loss": 0.4104, "step": 1293 }, { "epoch": 0.23, "learning_rate": 4.997780667723801e-05, "loss": 0.3789, "step": 1294 }, { "epoch": 0.23, "learning_rate": 4.997750166880958e-05, "loss": 0.404, "step": 1295 }, { "epoch": 0.23, "learning_rate": 4.997719457980537e-05, "loss": 0.3984, "step": 1296 }, { "epoch": 0.23, "learning_rate": 4.99768854102538e-05, "loss": 0.3981, "step": 1297 }, { "epoch": 0.23, "learning_rate": 4.9976574160183486e-05, "loss": 0.4146, "step": 1298 }, { "epoch": 0.23, "learning_rate": 4.9976260829623244e-05, "loss": 0.3994, "step": 1299 }, { "epoch": 0.23, "learning_rate": 4.997594541860207e-05, "loss": 0.3855, "step": 1300 }, { "epoch": 0.23, "learning_rate": 4.9975627927149176e-05, "loss": 0.4172, "step": 1301 }, { "epoch": 0.23, "learning_rate": 4.9975308355293924e-05, "loss": 0.4036, "step": 1302 }, { "epoch": 0.23, "learning_rate": 4.997498670306591e-05, "loss": 0.4076, "step": 1303 }, { "epoch": 0.23, "learning_rate": 4.9974662970494915e-05, "loss": 0.3956, "step": 1304 }, { "epoch": 0.23, "learning_rate": 4.9974337157610885e-05, "loss": 0.4083, "step": 1305 }, { "epoch": 0.23, "learning_rate": 4.997400926444399e-05, "loss": 0.4056, "step": 1306 }, { "epoch": 0.23, "learning_rate": 4.997367929102458e-05, "loss": 0.3912, "step": 1307 }, { "epoch": 0.23, "learning_rate": 4.997334723738319e-05, "loss": 0.3906, "step": 1308 }, { "epoch": 0.23, "learning_rate": 4.9973013103550564e-05, "loss": 0.3967, "step": 1309 }, { "epoch": 0.23, "learning_rate": 4.997267688955763e-05, "loss": 0.3919, "step": 1310 }, { "epoch": 0.23, "learning_rate": 4.9972338595435505e-05, "loss": 0.4082, "step": 1311 }, { "epoch": 0.23, "learning_rate": 4.997199822121551e-05, "loss": 0.4115, "step": 1312 }, { "epoch": 0.23, "learning_rate": 4.9971655766929134e-05, "loss": 0.4049, "step": 1313 }, { "epoch": 0.23, "learning_rate": 4.99713112326081e-05, "loss": 0.389, "step": 1314 }, { "epoch": 0.23, "learning_rate": 4.9970964618284274e-05, "loss": 0.387, "step": 1315 }, { "epoch": 0.23, "learning_rate": 4.997061592398975e-05, "loss": 0.387, "step": 1316 }, { "epoch": 0.23, "learning_rate": 4.997026514975682e-05, "loss": 0.3973, "step": 1317 }, { "epoch": 0.23, "learning_rate": 4.9969912295617924e-05, "loss": 0.3894, "step": 1318 }, { "epoch": 0.23, "learning_rate": 4.9969557361605744e-05, "loss": 0.4058, "step": 1319 }, { "epoch": 0.23, "learning_rate": 4.9969200347753126e-05, "loss": 0.3861, "step": 1320 }, { "epoch": 0.23, "learning_rate": 4.996884125409312e-05, "loss": 0.4025, "step": 1321 }, { "epoch": 0.23, "learning_rate": 4.9968480080658956e-05, "loss": 0.4126, "step": 1322 }, { "epoch": 0.23, "learning_rate": 4.996811682748407e-05, "loss": 0.3906, "step": 1323 }, { "epoch": 0.23, "learning_rate": 4.996775149460209e-05, "loss": 0.3891, "step": 1324 }, { "epoch": 0.23, "learning_rate": 4.9967384082046825e-05, "loss": 0.3862, "step": 1325 }, { "epoch": 0.23, "learning_rate": 4.996701458985229e-05, "loss": 0.398, "step": 1326 }, { "epoch": 0.23, "learning_rate": 4.9966643018052686e-05, "loss": 0.3949, "step": 1327 }, { "epoch": 0.23, "learning_rate": 4.99662693666824e-05, "loss": 0.4015, "step": 1328 }, { "epoch": 0.23, "learning_rate": 4.9965893635776015e-05, "loss": 0.4009, "step": 1329 }, { "epoch": 0.23, "learning_rate": 4.996551582536832e-05, "loss": 0.3931, "step": 1330 }, { "epoch": 0.23, "learning_rate": 4.9965135935494286e-05, "loss": 0.3993, "step": 1331 }, { "epoch": 0.23, "learning_rate": 4.9964753966189075e-05, "loss": 0.3973, "step": 1332 }, { "epoch": 0.23, "learning_rate": 4.996436991748804e-05, "loss": 0.4012, "step": 1333 }, { "epoch": 0.23, "learning_rate": 4.996398378942673e-05, "loss": 0.3845, "step": 1334 }, { "epoch": 0.23, "learning_rate": 4.996359558204088e-05, "loss": 0.3928, "step": 1335 }, { "epoch": 0.23, "learning_rate": 4.996320529536644e-05, "loss": 0.3748, "step": 1336 }, { "epoch": 0.23, "learning_rate": 4.996281292943953e-05, "loss": 0.4259, "step": 1337 }, { "epoch": 0.23, "learning_rate": 4.996241848429645e-05, "loss": 0.3796, "step": 1338 }, { "epoch": 0.23, "learning_rate": 4.9962021959973736e-05, "loss": 0.383, "step": 1339 }, { "epoch": 0.23, "learning_rate": 4.996162335650808e-05, "loss": 0.4052, "step": 1340 }, { "epoch": 0.23, "learning_rate": 4.9961222673936376e-05, "loss": 0.3984, "step": 1341 }, { "epoch": 0.23, "learning_rate": 4.996081991229572e-05, "loss": 0.4195, "step": 1342 }, { "epoch": 0.23, "learning_rate": 4.996041507162338e-05, "loss": 0.3893, "step": 1343 }, { "epoch": 0.23, "learning_rate": 4.996000815195684e-05, "loss": 0.3924, "step": 1344 }, { "epoch": 0.23, "learning_rate": 4.995959915333376e-05, "loss": 0.3956, "step": 1345 }, { "epoch": 0.23, "learning_rate": 4.9959188075792e-05, "loss": 0.3929, "step": 1346 }, { "epoch": 0.23, "learning_rate": 4.9958774919369624e-05, "loss": 0.4082, "step": 1347 }, { "epoch": 0.23, "learning_rate": 4.9958359684104846e-05, "loss": 0.3916, "step": 1348 }, { "epoch": 0.24, "learning_rate": 4.9957942370036114e-05, "loss": 0.3994, "step": 1349 }, { "epoch": 0.24, "learning_rate": 4.995752297720207e-05, "loss": 0.3892, "step": 1350 }, { "epoch": 0.24, "learning_rate": 4.995710150564152e-05, "loss": 0.4012, "step": 1351 }, { "epoch": 0.24, "learning_rate": 4.995667795539347e-05, "loss": 0.3789, "step": 1352 }, { "epoch": 0.24, "learning_rate": 4.995625232649714e-05, "loss": 0.4005, "step": 1353 }, { "epoch": 0.24, "learning_rate": 4.9955824618991924e-05, "loss": 0.4121, "step": 1354 }, { "epoch": 0.24, "learning_rate": 4.9955394832917404e-05, "loss": 0.3949, "step": 1355 }, { "epoch": 0.24, "learning_rate": 4.995496296831337e-05, "loss": 0.3763, "step": 1356 }, { "epoch": 0.24, "learning_rate": 4.995452902521979e-05, "loss": 0.4061, "step": 1357 }, { "epoch": 0.24, "learning_rate": 4.995409300367684e-05, "loss": 0.3842, "step": 1358 }, { "epoch": 0.24, "learning_rate": 4.9953654903724866e-05, "loss": 0.4018, "step": 1359 }, { "epoch": 0.24, "learning_rate": 4.995321472540444e-05, "loss": 0.4083, "step": 1360 }, { "epoch": 0.24, "learning_rate": 4.995277246875628e-05, "loss": 0.3915, "step": 1361 }, { "epoch": 0.24, "learning_rate": 4.995232813382134e-05, "loss": 0.4051, "step": 1362 }, { "epoch": 0.24, "learning_rate": 4.995188172064075e-05, "loss": 0.3792, "step": 1363 }, { "epoch": 0.24, "learning_rate": 4.9951433229255824e-05, "loss": 0.3892, "step": 1364 }, { "epoch": 0.24, "learning_rate": 4.995098265970807e-05, "loss": 0.398, "step": 1365 }, { "epoch": 0.24, "learning_rate": 4.99505300120392e-05, "loss": 0.3908, "step": 1366 }, { "epoch": 0.24, "learning_rate": 4.9950075286291125e-05, "loss": 0.3924, "step": 1367 }, { "epoch": 0.24, "learning_rate": 4.994961848250592e-05, "loss": 0.411, "step": 1368 }, { "epoch": 0.24, "learning_rate": 4.9949159600725863e-05, "loss": 0.4042, "step": 1369 }, { "epoch": 0.24, "learning_rate": 4.994869864099345e-05, "loss": 0.3808, "step": 1370 }, { "epoch": 0.24, "learning_rate": 4.9948235603351326e-05, "loss": 0.4003, "step": 1371 }, { "epoch": 0.24, "learning_rate": 4.9947770487842366e-05, "loss": 0.3971, "step": 1372 }, { "epoch": 0.24, "learning_rate": 4.9947303294509626e-05, "loss": 0.3896, "step": 1373 }, { "epoch": 0.24, "learning_rate": 4.994683402339633e-05, "loss": 0.4108, "step": 1374 }, { "epoch": 0.24, "learning_rate": 4.994636267454593e-05, "loss": 0.3797, "step": 1375 }, { "epoch": 0.24, "learning_rate": 4.994588924800206e-05, "loss": 0.3883, "step": 1376 }, { "epoch": 0.24, "learning_rate": 4.994541374380853e-05, "loss": 0.4034, "step": 1377 }, { "epoch": 0.24, "learning_rate": 4.9944936162009365e-05, "loss": 0.4178, "step": 1378 }, { "epoch": 0.24, "learning_rate": 4.9944456502648754e-05, "loss": 0.3932, "step": 1379 }, { "epoch": 0.24, "learning_rate": 4.994397476577111e-05, "loss": 0.3807, "step": 1380 }, { "epoch": 0.24, "learning_rate": 4.994349095142102e-05, "loss": 0.4015, "step": 1381 }, { "epoch": 0.24, "learning_rate": 4.994300505964327e-05, "loss": 0.3754, "step": 1382 }, { "epoch": 0.24, "learning_rate": 4.994251709048283e-05, "loss": 0.3982, "step": 1383 }, { "epoch": 0.24, "learning_rate": 4.9942027043984874e-05, "loss": 0.4028, "step": 1384 }, { "epoch": 0.24, "learning_rate": 4.9941534920194757e-05, "loss": 0.3839, "step": 1385 }, { "epoch": 0.24, "learning_rate": 4.994104071915803e-05, "loss": 0.3869, "step": 1386 }, { "epoch": 0.24, "learning_rate": 4.9940544440920434e-05, "loss": 0.3889, "step": 1387 }, { "epoch": 0.24, "learning_rate": 4.9940046085527924e-05, "loss": 0.4038, "step": 1388 }, { "epoch": 0.24, "learning_rate": 4.9939545653026605e-05, "loss": 0.3751, "step": 1389 }, { "epoch": 0.24, "learning_rate": 4.993904314346282e-05, "loss": 0.3955, "step": 1390 }, { "epoch": 0.24, "learning_rate": 4.9938538556883065e-05, "loss": 0.3987, "step": 1391 }, { "epoch": 0.24, "learning_rate": 4.993803189333406e-05, "loss": 0.384, "step": 1392 }, { "epoch": 0.24, "learning_rate": 4.993752315286269e-05, "loss": 0.3899, "step": 1393 }, { "epoch": 0.24, "learning_rate": 4.993701233551606e-05, "loss": 0.402, "step": 1394 }, { "epoch": 0.24, "learning_rate": 4.993649944134144e-05, "loss": 0.3979, "step": 1395 }, { "epoch": 0.24, "learning_rate": 4.9935984470386304e-05, "loss": 0.3949, "step": 1396 }, { "epoch": 0.24, "learning_rate": 4.993546742269833e-05, "loss": 0.3879, "step": 1397 }, { "epoch": 0.24, "learning_rate": 4.993494829832537e-05, "loss": 0.4032, "step": 1398 }, { "epoch": 0.24, "learning_rate": 4.993442709731548e-05, "loss": 0.3874, "step": 1399 }, { "epoch": 0.24, "learning_rate": 4.99339038197169e-05, "loss": 0.389, "step": 1400 }, { "epoch": 0.24, "learning_rate": 4.993337846557806e-05, "loss": 0.398, "step": 1401 }, { "epoch": 0.24, "learning_rate": 4.99328510349476e-05, "loss": 0.4024, "step": 1402 }, { "epoch": 0.24, "learning_rate": 4.993232152787434e-05, "loss": 0.4079, "step": 1403 }, { "epoch": 0.24, "learning_rate": 4.9931789944407284e-05, "loss": 0.3883, "step": 1404 }, { "epoch": 0.24, "learning_rate": 4.993125628459564e-05, "loss": 0.3926, "step": 1405 }, { "epoch": 0.25, "learning_rate": 4.99307205484888e-05, "loss": 0.3931, "step": 1406 }, { "epoch": 0.25, "learning_rate": 4.993018273613637e-05, "loss": 0.4013, "step": 1407 }, { "epoch": 0.25, "learning_rate": 4.992964284758812e-05, "loss": 0.4004, "step": 1408 }, { "epoch": 0.25, "learning_rate": 4.992910088289401e-05, "loss": 0.422, "step": 1409 }, { "epoch": 0.25, "learning_rate": 4.9928556842104226e-05, "loss": 0.4109, "step": 1410 }, { "epoch": 0.25, "learning_rate": 4.992801072526912e-05, "loss": 0.4136, "step": 1411 }, { "epoch": 0.25, "learning_rate": 4.992746253243925e-05, "loss": 0.3972, "step": 1412 }, { "epoch": 0.25, "learning_rate": 4.9926912263665335e-05, "loss": 0.3857, "step": 1413 }, { "epoch": 0.25, "learning_rate": 4.992635991899832e-05, "loss": 0.3982, "step": 1414 }, { "epoch": 0.25, "learning_rate": 4.992580549848935e-05, "loss": 0.4037, "step": 1415 }, { "epoch": 0.25, "learning_rate": 4.992524900218972e-05, "loss": 0.3844, "step": 1416 }, { "epoch": 0.25, "learning_rate": 4.992469043015094e-05, "loss": 0.3774, "step": 1417 }, { "epoch": 0.25, "learning_rate": 4.992412978242474e-05, "loss": 0.3747, "step": 1418 }, { "epoch": 0.25, "learning_rate": 4.9923567059062985e-05, "loss": 0.4, "step": 1419 }, { "epoch": 0.25, "learning_rate": 4.992300226011778e-05, "loss": 0.4041, "step": 1420 }, { "epoch": 0.25, "learning_rate": 4.9922435385641406e-05, "loss": 0.4102, "step": 1421 }, { "epoch": 0.25, "learning_rate": 4.992186643568631e-05, "loss": 0.3913, "step": 1422 }, { "epoch": 0.25, "learning_rate": 4.992129541030518e-05, "loss": 0.3713, "step": 1423 }, { "epoch": 0.25, "learning_rate": 4.992072230955086e-05, "loss": 0.3946, "step": 1424 }, { "epoch": 0.25, "learning_rate": 4.9920147133476406e-05, "loss": 0.3805, "step": 1425 }, { "epoch": 0.25, "learning_rate": 4.9919569882135054e-05, "loss": 0.396, "step": 1426 }, { "epoch": 0.25, "learning_rate": 4.9918990555580234e-05, "loss": 0.3801, "step": 1427 }, { "epoch": 0.25, "learning_rate": 4.991840915386557e-05, "loss": 0.3749, "step": 1428 }, { "epoch": 0.25, "learning_rate": 4.991782567704488e-05, "loss": 0.3904, "step": 1429 }, { "epoch": 0.25, "learning_rate": 4.991724012517217e-05, "loss": 0.3999, "step": 1430 }, { "epoch": 0.25, "learning_rate": 4.991665249830164e-05, "loss": 0.3859, "step": 1431 }, { "epoch": 0.25, "learning_rate": 4.991606279648769e-05, "loss": 0.4042, "step": 1432 }, { "epoch": 0.25, "learning_rate": 4.9915471019784895e-05, "loss": 0.4227, "step": 1433 }, { "epoch": 0.25, "learning_rate": 4.9914877168248034e-05, "loss": 0.3944, "step": 1434 }, { "epoch": 0.25, "learning_rate": 4.991428124193207e-05, "loss": 0.3823, "step": 1435 }, { "epoch": 0.25, "learning_rate": 4.9913683240892175e-05, "loss": 0.3956, "step": 1436 }, { "epoch": 0.25, "learning_rate": 4.99130831651837e-05, "loss": 0.3903, "step": 1437 }, { "epoch": 0.25, "learning_rate": 4.9912481014862184e-05, "loss": 0.3884, "step": 1438 }, { "epoch": 0.25, "learning_rate": 4.991187678998336e-05, "loss": 0.4244, "step": 1439 }, { "epoch": 0.25, "learning_rate": 4.9911270490603166e-05, "loss": 0.4033, "step": 1440 }, { "epoch": 0.25, "learning_rate": 4.991066211677772e-05, "loss": 0.3919, "step": 1441 }, { "epoch": 0.25, "learning_rate": 4.991005166856332e-05, "loss": 0.4079, "step": 1442 }, { "epoch": 0.25, "learning_rate": 4.990943914601649e-05, "loss": 0.3992, "step": 1443 }, { "epoch": 0.25, "learning_rate": 4.9908824549193935e-05, "loss": 0.3965, "step": 1444 }, { "epoch": 0.25, "learning_rate": 4.9908207878152514e-05, "loss": 0.3924, "step": 1445 }, { "epoch": 0.25, "learning_rate": 4.990758913294933e-05, "loss": 0.3763, "step": 1446 }, { "epoch": 0.25, "learning_rate": 4.9906968313641644e-05, "loss": 0.4069, "step": 1447 }, { "epoch": 0.25, "learning_rate": 4.990634542028693e-05, "loss": 0.3807, "step": 1448 }, { "epoch": 0.25, "learning_rate": 4.990572045294283e-05, "loss": 0.3861, "step": 1449 }, { "epoch": 0.25, "learning_rate": 4.9905093411667206e-05, "loss": 0.4074, "step": 1450 }, { "epoch": 0.25, "learning_rate": 4.99044642965181e-05, "loss": 0.4102, "step": 1451 }, { "epoch": 0.25, "learning_rate": 4.9903833107553735e-05, "loss": 0.3922, "step": 1452 }, { "epoch": 0.25, "learning_rate": 4.9903199844832536e-05, "loss": 0.3787, "step": 1453 }, { "epoch": 0.25, "learning_rate": 4.9902564508413136e-05, "loss": 0.3782, "step": 1454 }, { "epoch": 0.25, "learning_rate": 4.990192709835432e-05, "loss": 0.3912, "step": 1455 }, { "epoch": 0.25, "learning_rate": 4.99012876147151e-05, "loss": 0.3945, "step": 1456 }, { "epoch": 0.25, "learning_rate": 4.9900646057554676e-05, "loss": 0.396, "step": 1457 }, { "epoch": 0.25, "learning_rate": 4.9900002426932414e-05, "loss": 0.3929, "step": 1458 }, { "epoch": 0.25, "learning_rate": 4.98993567229079e-05, "loss": 0.3839, "step": 1459 }, { "epoch": 0.25, "learning_rate": 4.98987089455409e-05, "loss": 0.3988, "step": 1460 }, { "epoch": 0.25, "learning_rate": 4.9898059094891384e-05, "loss": 0.3836, "step": 1461 }, { "epoch": 0.25, "learning_rate": 4.989740717101949e-05, "loss": 0.3922, "step": 1462 }, { "epoch": 0.26, "learning_rate": 4.989675317398556e-05, "loss": 0.402, "step": 1463 }, { "epoch": 0.26, "learning_rate": 4.9896097103850146e-05, "loss": 0.3885, "step": 1464 }, { "epoch": 0.26, "learning_rate": 4.9895438960673965e-05, "loss": 0.3813, "step": 1465 }, { "epoch": 0.26, "learning_rate": 4.989477874451794e-05, "loss": 0.3827, "step": 1466 }, { "epoch": 0.26, "learning_rate": 4.9894116455443176e-05, "loss": 0.4117, "step": 1467 }, { "epoch": 0.26, "learning_rate": 4.9893452093510975e-05, "loss": 0.3814, "step": 1468 }, { "epoch": 0.26, "learning_rate": 4.9892785658782846e-05, "loss": 0.3908, "step": 1469 }, { "epoch": 0.26, "learning_rate": 4.989211715132047e-05, "loss": 0.3819, "step": 1470 }, { "epoch": 0.26, "learning_rate": 4.989144657118572e-05, "loss": 0.3941, "step": 1471 }, { "epoch": 0.26, "learning_rate": 4.9890773918440664e-05, "loss": 0.4194, "step": 1472 }, { "epoch": 0.26, "learning_rate": 4.989009919314758e-05, "loss": 0.387, "step": 1473 }, { "epoch": 0.26, "learning_rate": 4.98894223953689e-05, "loss": 0.3765, "step": 1474 }, { "epoch": 0.26, "learning_rate": 4.988874352516729e-05, "loss": 0.3893, "step": 1475 }, { "epoch": 0.26, "learning_rate": 4.988806258260558e-05, "loss": 0.3852, "step": 1476 }, { "epoch": 0.26, "learning_rate": 4.98873795677468e-05, "loss": 0.3883, "step": 1477 }, { "epoch": 0.26, "learning_rate": 4.9886694480654174e-05, "loss": 0.3835, "step": 1478 }, { "epoch": 0.26, "learning_rate": 4.988600732139111e-05, "loss": 0.389, "step": 1479 }, { "epoch": 0.26, "learning_rate": 4.988531809002122e-05, "loss": 0.3855, "step": 1480 }, { "epoch": 0.26, "learning_rate": 4.98846267866083e-05, "loss": 0.4009, "step": 1481 }, { "epoch": 0.26, "learning_rate": 4.9883933411216324e-05, "loss": 0.3755, "step": 1482 }, { "epoch": 0.26, "learning_rate": 4.98832379639095e-05, "loss": 0.4004, "step": 1483 }, { "epoch": 0.26, "learning_rate": 4.988254044475218e-05, "loss": 0.3782, "step": 1484 }, { "epoch": 0.26, "learning_rate": 4.988184085380893e-05, "loss": 0.3914, "step": 1485 }, { "epoch": 0.26, "learning_rate": 4.988113919114451e-05, "loss": 0.3985, "step": 1486 }, { "epoch": 0.26, "learning_rate": 4.9880435456823876e-05, "loss": 0.407, "step": 1487 }, { "epoch": 0.26, "learning_rate": 4.9879729650912154e-05, "loss": 0.3736, "step": 1488 }, { "epoch": 0.26, "learning_rate": 4.987902177347468e-05, "loss": 0.378, "step": 1489 }, { "epoch": 0.26, "learning_rate": 4.987831182457698e-05, "loss": 0.4001, "step": 1490 }, { "epoch": 0.26, "learning_rate": 4.9877599804284755e-05, "loss": 0.3957, "step": 1491 }, { "epoch": 0.26, "learning_rate": 4.987688571266393e-05, "loss": 0.3906, "step": 1492 }, { "epoch": 0.26, "learning_rate": 4.98761695497806e-05, "loss": 0.3828, "step": 1493 }, { "epoch": 0.26, "learning_rate": 4.987545131570105e-05, "loss": 0.3913, "step": 1494 }, { "epoch": 0.26, "learning_rate": 4.9874731010491753e-05, "loss": 0.3768, "step": 1495 }, { "epoch": 0.26, "learning_rate": 4.98740086342194e-05, "loss": 0.3951, "step": 1496 }, { "epoch": 0.26, "learning_rate": 4.987328418695084e-05, "loss": 0.3961, "step": 1497 }, { "epoch": 0.26, "learning_rate": 4.987255766875315e-05, "loss": 0.3905, "step": 1498 }, { "epoch": 0.26, "learning_rate": 4.9871829079693554e-05, "loss": 0.3783, "step": 1499 }, { "epoch": 0.26, "learning_rate": 4.987109841983951e-05, "loss": 0.3892, "step": 1500 }, { "epoch": 0.26, "learning_rate": 4.9870365689258635e-05, "loss": 0.3943, "step": 1501 }, { "epoch": 0.26, "learning_rate": 4.986963088801876e-05, "loss": 0.3783, "step": 1502 }, { "epoch": 0.26, "learning_rate": 4.986889401618791e-05, "loss": 0.4006, "step": 1503 }, { "epoch": 0.26, "learning_rate": 4.9868155073834285e-05, "loss": 0.3867, "step": 1504 }, { "epoch": 0.26, "learning_rate": 4.9867414061026265e-05, "loss": 0.3918, "step": 1505 }, { "epoch": 0.26, "learning_rate": 4.986667097783247e-05, "loss": 0.3711, "step": 1506 }, { "epoch": 0.26, "learning_rate": 4.986592582432166e-05, "loss": 0.39, "step": 1507 }, { "epoch": 0.26, "learning_rate": 4.986517860056282e-05, "loss": 0.3881, "step": 1508 }, { "epoch": 0.26, "learning_rate": 4.986442930662511e-05, "loss": 0.3886, "step": 1509 }, { "epoch": 0.26, "learning_rate": 4.986367794257789e-05, "loss": 0.3817, "step": 1510 }, { "epoch": 0.26, "learning_rate": 4.98629245084907e-05, "loss": 0.3929, "step": 1511 }, { "epoch": 0.26, "learning_rate": 4.9862169004433296e-05, "loss": 0.3852, "step": 1512 }, { "epoch": 0.26, "learning_rate": 4.986141143047559e-05, "loss": 0.402, "step": 1513 }, { "epoch": 0.26, "learning_rate": 4.986065178668772e-05, "loss": 0.3995, "step": 1514 }, { "epoch": 0.26, "learning_rate": 4.985989007313999e-05, "loss": 0.4007, "step": 1515 }, { "epoch": 0.26, "learning_rate": 4.985912628990291e-05, "loss": 0.3894, "step": 1516 }, { "epoch": 0.26, "learning_rate": 4.985836043704718e-05, "loss": 0.3967, "step": 1517 }, { "epoch": 0.26, "learning_rate": 4.985759251464369e-05, "loss": 0.3921, "step": 1518 }, { "epoch": 0.26, "learning_rate": 4.985682252276352e-05, "loss": 0.4021, "step": 1519 }, { "epoch": 0.26, "learning_rate": 4.985605046147794e-05, "loss": 0.3859, "step": 1520 }, { "epoch": 0.27, "learning_rate": 4.985527633085841e-05, "loss": 0.3839, "step": 1521 }, { "epoch": 0.27, "learning_rate": 4.98545001309766e-05, "loss": 0.4013, "step": 1522 }, { "epoch": 0.27, "learning_rate": 4.985372186190434e-05, "loss": 0.3776, "step": 1523 }, { "epoch": 0.27, "learning_rate": 4.9852941523713685e-05, "loss": 0.4077, "step": 1524 }, { "epoch": 0.27, "learning_rate": 4.985215911647685e-05, "loss": 0.3844, "step": 1525 }, { "epoch": 0.27, "learning_rate": 4.9851374640266266e-05, "loss": 0.3908, "step": 1526 }, { "epoch": 0.27, "learning_rate": 4.9850588095154543e-05, "loss": 0.3702, "step": 1527 }, { "epoch": 0.27, "learning_rate": 4.984979948121448e-05, "loss": 0.3834, "step": 1528 }, { "epoch": 0.27, "learning_rate": 4.9849008798519086e-05, "loss": 0.3945, "step": 1529 }, { "epoch": 0.27, "learning_rate": 4.984821604714154e-05, "loss": 0.3933, "step": 1530 }, { "epoch": 0.27, "learning_rate": 4.9847421227155226e-05, "loss": 0.382, "step": 1531 }, { "epoch": 0.27, "learning_rate": 4.9846624338633707e-05, "loss": 0.378, "step": 1532 }, { "epoch": 0.27, "learning_rate": 4.984582538165075e-05, "loss": 0.377, "step": 1533 }, { "epoch": 0.27, "learning_rate": 4.9845024356280306e-05, "loss": 0.3796, "step": 1534 }, { "epoch": 0.27, "learning_rate": 4.9844221262596535e-05, "loss": 0.3854, "step": 1535 }, { "epoch": 0.27, "learning_rate": 4.984341610067375e-05, "loss": 0.3721, "step": 1536 }, { "epoch": 0.27, "learning_rate": 4.9842608870586485e-05, "loss": 0.3815, "step": 1537 }, { "epoch": 0.27, "learning_rate": 4.984179957240947e-05, "loss": 0.3906, "step": 1538 }, { "epoch": 0.27, "learning_rate": 4.984098820621762e-05, "loss": 0.4001, "step": 1539 }, { "epoch": 0.27, "learning_rate": 4.9840174772086015e-05, "loss": 0.3976, "step": 1540 }, { "epoch": 0.27, "learning_rate": 4.983935927008995e-05, "loss": 0.3971, "step": 1541 }, { "epoch": 0.27, "learning_rate": 4.983854170030494e-05, "loss": 0.3837, "step": 1542 }, { "epoch": 0.27, "learning_rate": 4.983772206280663e-05, "loss": 0.3824, "step": 1543 }, { "epoch": 0.27, "learning_rate": 4.983690035767091e-05, "loss": 0.3991, "step": 1544 }, { "epoch": 0.27, "learning_rate": 4.983607658497381e-05, "loss": 0.3968, "step": 1545 }, { "epoch": 0.27, "learning_rate": 4.9835250744791615e-05, "loss": 0.3916, "step": 1546 }, { "epoch": 0.27, "learning_rate": 4.983442283720074e-05, "loss": 0.3829, "step": 1547 }, { "epoch": 0.27, "learning_rate": 4.983359286227784e-05, "loss": 0.3717, "step": 1548 }, { "epoch": 0.27, "learning_rate": 4.9832760820099724e-05, "loss": 0.3896, "step": 1549 }, { "epoch": 0.27, "learning_rate": 4.983192671074341e-05, "loss": 0.397, "step": 1550 }, { "epoch": 0.27, "learning_rate": 4.9831090534286114e-05, "loss": 0.3875, "step": 1551 }, { "epoch": 0.27, "learning_rate": 4.983025229080523e-05, "loss": 0.3981, "step": 1552 }, { "epoch": 0.27, "learning_rate": 4.982941198037834e-05, "loss": 0.387, "step": 1553 }, { "epoch": 0.27, "learning_rate": 4.982856960308324e-05, "loss": 0.3968, "step": 1554 }, { "epoch": 0.27, "learning_rate": 4.982772515899789e-05, "loss": 0.4073, "step": 1555 }, { "epoch": 0.27, "learning_rate": 4.982687864820045e-05, "loss": 0.401, "step": 1556 }, { "epoch": 0.27, "learning_rate": 4.98260300707693e-05, "loss": 0.3839, "step": 1557 }, { "epoch": 0.27, "learning_rate": 4.982517942678296e-05, "loss": 0.3865, "step": 1558 }, { "epoch": 0.27, "learning_rate": 4.982432671632017e-05, "loss": 0.3901, "step": 1559 }, { "epoch": 0.27, "learning_rate": 4.9823471939459884e-05, "loss": 0.3982, "step": 1560 }, { "epoch": 0.27, "learning_rate": 4.9822615096281206e-05, "loss": 0.3946, "step": 1561 }, { "epoch": 0.27, "learning_rate": 4.9821756186863435e-05, "loss": 0.404, "step": 1562 }, { "epoch": 0.27, "learning_rate": 4.9820895211286096e-05, "loss": 0.3857, "step": 1563 }, { "epoch": 0.27, "learning_rate": 4.9820032169628866e-05, "loss": 0.3851, "step": 1564 }, { "epoch": 0.27, "learning_rate": 4.981916706197164e-05, "loss": 0.374, "step": 1565 }, { "epoch": 0.27, "learning_rate": 4.9818299888394495e-05, "loss": 0.3759, "step": 1566 }, { "epoch": 0.27, "learning_rate": 4.98174306489777e-05, "loss": 0.3818, "step": 1567 }, { "epoch": 0.27, "learning_rate": 4.98165593438017e-05, "loss": 0.3908, "step": 1568 }, { "epoch": 0.27, "learning_rate": 4.981568597294716e-05, "loss": 0.3945, "step": 1569 }, { "epoch": 0.27, "learning_rate": 4.9814810536494924e-05, "loss": 0.3969, "step": 1570 }, { "epoch": 0.27, "learning_rate": 4.9813933034526e-05, "loss": 0.3883, "step": 1571 }, { "epoch": 0.27, "learning_rate": 4.9813053467121646e-05, "loss": 0.4055, "step": 1572 }, { "epoch": 0.27, "learning_rate": 4.981217183436326e-05, "loss": 0.3761, "step": 1573 }, { "epoch": 0.27, "learning_rate": 4.981128813633244e-05, "loss": 0.3742, "step": 1574 }, { "epoch": 0.27, "learning_rate": 4.9810402373111e-05, "loss": 0.3894, "step": 1575 }, { "epoch": 0.27, "learning_rate": 4.980951454478091e-05, "loss": 0.3769, "step": 1576 }, { "epoch": 0.27, "learning_rate": 4.980862465142437e-05, "loss": 0.3797, "step": 1577 }, { "epoch": 0.28, "learning_rate": 4.9807732693123735e-05, "loss": 0.3822, "step": 1578 }, { "epoch": 0.28, "learning_rate": 4.980683866996158e-05, "loss": 0.3991, "step": 1579 }, { "epoch": 0.28, "learning_rate": 4.980594258202064e-05, "loss": 0.3943, "step": 1580 }, { "epoch": 0.28, "learning_rate": 4.9805044429383886e-05, "loss": 0.3938, "step": 1581 }, { "epoch": 0.28, "learning_rate": 4.980414421213442e-05, "loss": 0.3673, "step": 1582 }, { "epoch": 0.28, "learning_rate": 4.98032419303556e-05, "loss": 0.3853, "step": 1583 }, { "epoch": 0.28, "learning_rate": 4.9802337584130926e-05, "loss": 0.3809, "step": 1584 }, { "epoch": 0.28, "learning_rate": 4.9801431173544114e-05, "loss": 0.3969, "step": 1585 }, { "epoch": 0.28, "learning_rate": 4.9800522698679056e-05, "loss": 0.3842, "step": 1586 }, { "epoch": 0.28, "learning_rate": 4.979961215961984e-05, "loss": 0.3982, "step": 1587 }, { "epoch": 0.28, "learning_rate": 4.9798699556450765e-05, "loss": 0.3915, "step": 1588 }, { "epoch": 0.28, "learning_rate": 4.979778488925629e-05, "loss": 0.3773, "step": 1589 }, { "epoch": 0.28, "learning_rate": 4.979686815812108e-05, "loss": 0.3869, "step": 1590 }, { "epoch": 0.28, "learning_rate": 4.979594936312999e-05, "loss": 0.3916, "step": 1591 }, { "epoch": 0.28, "learning_rate": 4.979502850436808e-05, "loss": 0.4067, "step": 1592 }, { "epoch": 0.28, "learning_rate": 4.979410558192057e-05, "loss": 0.3911, "step": 1593 }, { "epoch": 0.28, "learning_rate": 4.979318059587289e-05, "loss": 0.3875, "step": 1594 }, { "epoch": 0.28, "learning_rate": 4.979225354631067e-05, "loss": 0.3928, "step": 1595 }, { "epoch": 0.28, "learning_rate": 4.979132443331971e-05, "loss": 0.3647, "step": 1596 }, { "epoch": 0.28, "learning_rate": 4.979039325698601e-05, "loss": 0.386, "step": 1597 }, { "epoch": 0.28, "learning_rate": 4.978946001739577e-05, "loss": 0.3924, "step": 1598 }, { "epoch": 0.28, "learning_rate": 4.9788524714635374e-05, "loss": 0.4023, "step": 1599 }, { "epoch": 0.28, "learning_rate": 4.978758734879138e-05, "loss": 0.3796, "step": 1600 }, { "epoch": 0.28, "learning_rate": 4.9786647919950575e-05, "loss": 0.396, "step": 1601 }, { "epoch": 0.28, "learning_rate": 4.9785706428199905e-05, "loss": 0.39, "step": 1602 }, { "epoch": 0.28, "learning_rate": 4.97847628736265e-05, "loss": 0.4112, "step": 1603 }, { "epoch": 0.28, "learning_rate": 4.978381725631773e-05, "loss": 0.3927, "step": 1604 }, { "epoch": 0.28, "learning_rate": 4.9782869576361105e-05, "loss": 0.3678, "step": 1605 }, { "epoch": 0.28, "learning_rate": 4.978191983384435e-05, "loss": 0.3838, "step": 1606 }, { "epoch": 0.28, "learning_rate": 4.9780968028855366e-05, "loss": 0.3907, "step": 1607 }, { "epoch": 0.28, "learning_rate": 4.978001416148227e-05, "loss": 0.3927, "step": 1608 }, { "epoch": 0.28, "learning_rate": 4.977905823181334e-05, "loss": 0.4204, "step": 1609 }, { "epoch": 0.28, "learning_rate": 4.977810023993706e-05, "loss": 0.3802, "step": 1610 }, { "epoch": 0.28, "learning_rate": 4.977714018594212e-05, "loss": 0.3792, "step": 1611 }, { "epoch": 0.28, "learning_rate": 4.9776178069917365e-05, "loss": 0.3719, "step": 1612 }, { "epoch": 0.28, "learning_rate": 4.977521389195187e-05, "loss": 0.3893, "step": 1613 }, { "epoch": 0.28, "learning_rate": 4.977424765213487e-05, "loss": 0.3733, "step": 1614 }, { "epoch": 0.28, "learning_rate": 4.977327935055581e-05, "loss": 0.3899, "step": 1615 }, { "epoch": 0.28, "learning_rate": 4.977230898730431e-05, "loss": 0.3775, "step": 1616 }, { "epoch": 0.28, "learning_rate": 4.977133656247018e-05, "loss": 0.3753, "step": 1617 }, { "epoch": 0.28, "learning_rate": 4.977036207614346e-05, "loss": 0.3742, "step": 1618 }, { "epoch": 0.28, "learning_rate": 4.976938552841433e-05, "loss": 0.3854, "step": 1619 }, { "epoch": 0.28, "learning_rate": 4.976840691937319e-05, "loss": 0.3993, "step": 1620 }, { "epoch": 0.28, "learning_rate": 4.976742624911061e-05, "loss": 0.3807, "step": 1621 }, { "epoch": 0.28, "learning_rate": 4.9766443517717375e-05, "loss": 0.3764, "step": 1622 }, { "epoch": 0.28, "learning_rate": 4.9765458725284445e-05, "loss": 0.3775, "step": 1623 }, { "epoch": 0.28, "learning_rate": 4.976447187190298e-05, "loss": 0.3951, "step": 1624 }, { "epoch": 0.28, "learning_rate": 4.976348295766432e-05, "loss": 0.3879, "step": 1625 }, { "epoch": 0.28, "learning_rate": 4.9762491982660004e-05, "loss": 0.3888, "step": 1626 }, { "epoch": 0.28, "learning_rate": 4.976149894698175e-05, "loss": 0.3914, "step": 1627 }, { "epoch": 0.28, "learning_rate": 4.97605038507215e-05, "loss": 0.3843, "step": 1628 }, { "epoch": 0.28, "learning_rate": 4.9759506693971336e-05, "loss": 0.4025, "step": 1629 }, { "epoch": 0.28, "learning_rate": 4.975850747682357e-05, "loss": 0.3863, "step": 1630 }, { "epoch": 0.28, "learning_rate": 4.9757506199370694e-05, "loss": 0.3791, "step": 1631 }, { "epoch": 0.28, "learning_rate": 4.975650286170539e-05, "loss": 0.384, "step": 1632 }, { "epoch": 0.28, "learning_rate": 4.9755497463920514e-05, "loss": 0.4017, "step": 1633 }, { "epoch": 0.28, "learning_rate": 4.9754490006109146e-05, "loss": 0.3904, "step": 1634 }, { "epoch": 0.28, "learning_rate": 4.9753480488364526e-05, "loss": 0.3966, "step": 1635 }, { "epoch": 0.29, "learning_rate": 4.9752468910780106e-05, "loss": 0.3882, "step": 1636 }, { "epoch": 0.29, "learning_rate": 4.975145527344952e-05, "loss": 0.3965, "step": 1637 }, { "epoch": 0.29, "learning_rate": 4.975043957646659e-05, "loss": 0.3839, "step": 1638 }, { "epoch": 0.29, "learning_rate": 4.974942181992534e-05, "loss": 0.3903, "step": 1639 }, { "epoch": 0.29, "learning_rate": 4.974840200391996e-05, "loss": 0.3745, "step": 1640 }, { "epoch": 0.29, "learning_rate": 4.9747380128544855e-05, "loss": 0.3817, "step": 1641 }, { "epoch": 0.29, "learning_rate": 4.974635619389462e-05, "loss": 0.3956, "step": 1642 }, { "epoch": 0.29, "learning_rate": 4.9745330200064014e-05, "loss": 0.3989, "step": 1643 }, { "epoch": 0.29, "learning_rate": 4.9744302147148023e-05, "loss": 0.3676, "step": 1644 }, { "epoch": 0.29, "learning_rate": 4.974327203524181e-05, "loss": 0.3673, "step": 1645 }, { "epoch": 0.29, "learning_rate": 4.97422398644407e-05, "loss": 0.3771, "step": 1646 }, { "epoch": 0.29, "learning_rate": 4.974120563484026e-05, "loss": 0.3755, "step": 1647 }, { "epoch": 0.29, "learning_rate": 4.9740169346536206e-05, "loss": 0.3892, "step": 1648 }, { "epoch": 0.29, "learning_rate": 4.9739130999624464e-05, "loss": 0.3734, "step": 1649 }, { "epoch": 0.29, "learning_rate": 4.9738090594201144e-05, "loss": 0.3925, "step": 1650 }, { "epoch": 0.29, "learning_rate": 4.973704813036255e-05, "loss": 0.3798, "step": 1651 }, { "epoch": 0.29, "learning_rate": 4.973600360820518e-05, "loss": 0.386, "step": 1652 }, { "epoch": 0.29, "learning_rate": 4.973495702782571e-05, "loss": 0.4028, "step": 1653 }, { "epoch": 0.29, "learning_rate": 4.973390838932102e-05, "loss": 0.3738, "step": 1654 }, { "epoch": 0.29, "learning_rate": 4.973285769278817e-05, "loss": 0.3824, "step": 1655 }, { "epoch": 0.29, "learning_rate": 4.973180493832441e-05, "loss": 0.3854, "step": 1656 }, { "epoch": 0.29, "learning_rate": 4.9730750126027205e-05, "loss": 0.3803, "step": 1657 }, { "epoch": 0.29, "learning_rate": 4.972969325599417e-05, "loss": 0.3688, "step": 1658 }, { "epoch": 0.29, "learning_rate": 4.972863432832315e-05, "loss": 0.4174, "step": 1659 }, { "epoch": 0.29, "learning_rate": 4.9727573343112136e-05, "loss": 0.3905, "step": 1660 }, { "epoch": 0.29, "learning_rate": 4.9726510300459364e-05, "loss": 0.3795, "step": 1661 }, { "epoch": 0.29, "learning_rate": 4.972544520046322e-05, "loss": 0.3854, "step": 1662 }, { "epoch": 0.29, "learning_rate": 4.972437804322229e-05, "loss": 0.3957, "step": 1663 }, { "epoch": 0.29, "learning_rate": 4.972330882883536e-05, "loss": 0.3847, "step": 1664 }, { "epoch": 0.29, "learning_rate": 4.972223755740139e-05, "loss": 0.3881, "step": 1665 }, { "epoch": 0.29, "learning_rate": 4.9721164229019544e-05, "loss": 0.3775, "step": 1666 }, { "epoch": 0.29, "learning_rate": 4.972008884378918e-05, "loss": 0.3893, "step": 1667 }, { "epoch": 0.29, "learning_rate": 4.971901140180983e-05, "loss": 0.3967, "step": 1668 }, { "epoch": 0.29, "learning_rate": 4.9717931903181215e-05, "loss": 0.3853, "step": 1669 }, { "epoch": 0.29, "learning_rate": 4.971685034800327e-05, "loss": 0.3805, "step": 1670 }, { "epoch": 0.29, "learning_rate": 4.9715766736376105e-05, "loss": 0.3729, "step": 1671 }, { "epoch": 0.29, "learning_rate": 4.9714681068400026e-05, "loss": 0.4004, "step": 1672 }, { "epoch": 0.29, "learning_rate": 4.971359334417551e-05, "loss": 0.3693, "step": 1673 }, { "epoch": 0.29, "learning_rate": 4.971250356380326e-05, "loss": 0.3841, "step": 1674 }, { "epoch": 0.29, "learning_rate": 4.9711411727384126e-05, "loss": 0.3839, "step": 1675 }, { "epoch": 0.29, "learning_rate": 4.971031783501919e-05, "loss": 0.3873, "step": 1676 }, { "epoch": 0.29, "learning_rate": 4.9709221886809695e-05, "loss": 0.394, "step": 1677 }, { "epoch": 0.29, "learning_rate": 4.97081238828571e-05, "loss": 0.3783, "step": 1678 }, { "epoch": 0.29, "learning_rate": 4.9707023823263014e-05, "loss": 0.3828, "step": 1679 }, { "epoch": 0.29, "learning_rate": 4.970592170812928e-05, "loss": 0.3624, "step": 1680 }, { "epoch": 0.29, "learning_rate": 4.970481753755791e-05, "loss": 0.3914, "step": 1681 }, { "epoch": 0.29, "learning_rate": 4.970371131165111e-05, "loss": 0.3921, "step": 1682 }, { "epoch": 0.29, "learning_rate": 4.970260303051126e-05, "loss": 0.3772, "step": 1683 }, { "epoch": 0.29, "learning_rate": 4.9701492694240967e-05, "loss": 0.3696, "step": 1684 }, { "epoch": 0.29, "learning_rate": 4.970038030294299e-05, "loss": 0.3904, "step": 1685 }, { "epoch": 0.29, "learning_rate": 4.96992658567203e-05, "loss": 0.3813, "step": 1686 }, { "epoch": 0.29, "learning_rate": 4.969814935567606e-05, "loss": 0.4034, "step": 1687 }, { "epoch": 0.29, "learning_rate": 4.9697030799913606e-05, "loss": 0.3856, "step": 1688 }, { "epoch": 0.29, "learning_rate": 4.9695910189536484e-05, "loss": 0.3843, "step": 1689 }, { "epoch": 0.29, "learning_rate": 4.9694787524648415e-05, "loss": 0.3902, "step": 1690 }, { "epoch": 0.29, "learning_rate": 4.969366280535331e-05, "loss": 0.388, "step": 1691 }, { "epoch": 0.29, "learning_rate": 4.9692536031755297e-05, "loss": 0.3767, "step": 1692 }, { "epoch": 0.3, "learning_rate": 4.969140720395865e-05, "loss": 0.3953, "step": 1693 }, { "epoch": 0.3, "learning_rate": 4.969027632206786e-05, "loss": 0.4064, "step": 1694 }, { "epoch": 0.3, "learning_rate": 4.968914338618761e-05, "loss": 0.3718, "step": 1695 }, { "epoch": 0.3, "learning_rate": 4.9688008396422774e-05, "loss": 0.3827, "step": 1696 }, { "epoch": 0.3, "learning_rate": 4.96868713528784e-05, "loss": 0.3709, "step": 1697 }, { "epoch": 0.3, "learning_rate": 4.9685732255659736e-05, "loss": 0.3748, "step": 1698 }, { "epoch": 0.3, "learning_rate": 4.968459110487222e-05, "loss": 0.3998, "step": 1699 }, { "epoch": 0.3, "learning_rate": 4.968344790062149e-05, "loss": 0.3692, "step": 1700 }, { "epoch": 0.3, "learning_rate": 4.968230264301334e-05, "loss": 0.3802, "step": 1701 }, { "epoch": 0.3, "learning_rate": 4.96811553321538e-05, "loss": 0.3762, "step": 1702 }, { "epoch": 0.3, "learning_rate": 4.968000596814907e-05, "loss": 0.3864, "step": 1703 }, { "epoch": 0.3, "learning_rate": 4.967885455110552e-05, "loss": 0.3805, "step": 1704 }, { "epoch": 0.3, "learning_rate": 4.967770108112974e-05, "loss": 0.3789, "step": 1705 }, { "epoch": 0.3, "learning_rate": 4.9676545558328504e-05, "loss": 0.3866, "step": 1706 }, { "epoch": 0.3, "learning_rate": 4.967538798280875e-05, "loss": 0.3909, "step": 1707 }, { "epoch": 0.3, "learning_rate": 4.967422835467764e-05, "loss": 0.3871, "step": 1708 }, { "epoch": 0.3, "learning_rate": 4.967306667404252e-05, "loss": 0.3748, "step": 1709 }, { "epoch": 0.3, "learning_rate": 4.9671902941010894e-05, "loss": 0.3851, "step": 1710 }, { "epoch": 0.3, "learning_rate": 4.9670737155690503e-05, "loss": 0.3707, "step": 1711 }, { "epoch": 0.3, "learning_rate": 4.966956931818925e-05, "loss": 0.39, "step": 1712 }, { "epoch": 0.3, "learning_rate": 4.966839942861523e-05, "loss": 0.3924, "step": 1713 }, { "epoch": 0.3, "learning_rate": 4.966722748707672e-05, "loss": 0.3959, "step": 1714 }, { "epoch": 0.3, "learning_rate": 4.966605349368222e-05, "loss": 0.3886, "step": 1715 }, { "epoch": 0.3, "learning_rate": 4.966487744854039e-05, "loss": 0.3788, "step": 1716 }, { "epoch": 0.3, "learning_rate": 4.966369935176007e-05, "loss": 0.396, "step": 1717 }, { "epoch": 0.3, "learning_rate": 4.966251920345034e-05, "loss": 0.3839, "step": 1718 }, { "epoch": 0.3, "learning_rate": 4.966133700372041e-05, "loss": 0.3919, "step": 1719 }, { "epoch": 0.3, "learning_rate": 4.9660152752679716e-05, "loss": 0.3737, "step": 1720 }, { "epoch": 0.3, "learning_rate": 4.9658966450437885e-05, "loss": 0.3798, "step": 1721 }, { "epoch": 0.3, "learning_rate": 4.965777809710471e-05, "loss": 0.387, "step": 1722 }, { "epoch": 0.3, "learning_rate": 4.965658769279019e-05, "loss": 0.3754, "step": 1723 }, { "epoch": 0.3, "learning_rate": 4.965539523760452e-05, "loss": 0.3939, "step": 1724 }, { "epoch": 0.3, "learning_rate": 4.965420073165807e-05, "loss": 0.371, "step": 1725 }, { "epoch": 0.3, "learning_rate": 4.965300417506141e-05, "loss": 0.378, "step": 1726 }, { "epoch": 0.3, "learning_rate": 4.96518055679253e-05, "loss": 0.3887, "step": 1727 }, { "epoch": 0.3, "learning_rate": 4.965060491036068e-05, "loss": 0.3873, "step": 1728 }, { "epoch": 0.3, "learning_rate": 4.964940220247869e-05, "loss": 0.3738, "step": 1729 }, { "epoch": 0.3, "learning_rate": 4.9648197444390654e-05, "loss": 0.3948, "step": 1730 }, { "epoch": 0.3, "learning_rate": 4.964699063620807e-05, "loss": 0.3902, "step": 1731 }, { "epoch": 0.3, "learning_rate": 4.964578177804268e-05, "loss": 0.4028, "step": 1732 }, { "epoch": 0.3, "learning_rate": 4.964457087000636e-05, "loss": 0.3785, "step": 1733 }, { "epoch": 0.3, "learning_rate": 4.964335791221118e-05, "loss": 0.3945, "step": 1734 }, { "epoch": 0.3, "learning_rate": 4.9642142904769435e-05, "loss": 0.3978, "step": 1735 }, { "epoch": 0.3, "learning_rate": 4.9640925847793586e-05, "loss": 0.3827, "step": 1736 }, { "epoch": 0.3, "learning_rate": 4.963970674139628e-05, "loss": 0.3925, "step": 1737 }, { "epoch": 0.3, "learning_rate": 4.9638485585690365e-05, "loss": 0.3795, "step": 1738 }, { "epoch": 0.3, "learning_rate": 4.963726238078888e-05, "loss": 0.397, "step": 1739 }, { "epoch": 0.3, "learning_rate": 4.963603712680503e-05, "loss": 0.3934, "step": 1740 }, { "epoch": 0.3, "learning_rate": 4.963480982385225e-05, "loss": 0.3799, "step": 1741 }, { "epoch": 0.3, "learning_rate": 4.963358047204413e-05, "loss": 0.3802, "step": 1742 }, { "epoch": 0.3, "learning_rate": 4.963234907149446e-05, "loss": 0.4092, "step": 1743 }, { "epoch": 0.3, "learning_rate": 4.963111562231723e-05, "loss": 0.3826, "step": 1744 }, { "epoch": 0.3, "learning_rate": 4.9629880124626596e-05, "loss": 0.372, "step": 1745 }, { "epoch": 0.3, "learning_rate": 4.962864257853694e-05, "loss": 0.373, "step": 1746 }, { "epoch": 0.3, "learning_rate": 4.9627402984162795e-05, "loss": 0.3957, "step": 1747 }, { "epoch": 0.3, "learning_rate": 4.9626161341618914e-05, "loss": 0.3856, "step": 1748 }, { "epoch": 0.3, "learning_rate": 4.9624917651020226e-05, "loss": 0.3942, "step": 1749 }, { "epoch": 0.31, "learning_rate": 4.962367191248183e-05, "loss": 0.3805, "step": 1750 }, { "epoch": 0.31, "learning_rate": 4.9622424126119066e-05, "loss": 0.3745, "step": 1751 }, { "epoch": 0.31, "learning_rate": 4.96211742920474e-05, "loss": 0.3882, "step": 1752 }, { "epoch": 0.31, "learning_rate": 4.961992241038255e-05, "loss": 0.3864, "step": 1753 }, { "epoch": 0.31, "learning_rate": 4.961866848124037e-05, "loss": 0.3794, "step": 1754 }, { "epoch": 0.31, "learning_rate": 4.9617412504736935e-05, "loss": 0.3891, "step": 1755 }, { "epoch": 0.31, "learning_rate": 4.9616154480988515e-05, "loss": 0.3817, "step": 1756 }, { "epoch": 0.31, "learning_rate": 4.9614894410111535e-05, "loss": 0.3742, "step": 1757 }, { "epoch": 0.31, "learning_rate": 4.961363229222264e-05, "loss": 0.3761, "step": 1758 }, { "epoch": 0.31, "learning_rate": 4.961236812743866e-05, "loss": 0.3884, "step": 1759 }, { "epoch": 0.31, "learning_rate": 4.961110191587659e-05, "loss": 0.3783, "step": 1760 }, { "epoch": 0.31, "learning_rate": 4.960983365765366e-05, "loss": 0.375, "step": 1761 }, { "epoch": 0.31, "learning_rate": 4.9608563352887245e-05, "loss": 0.3969, "step": 1762 }, { "epoch": 0.31, "learning_rate": 4.960729100169494e-05, "loss": 0.3777, "step": 1763 }, { "epoch": 0.31, "learning_rate": 4.96060166041945e-05, "loss": 0.3596, "step": 1764 }, { "epoch": 0.31, "learning_rate": 4.9604740160503906e-05, "loss": 0.3826, "step": 1765 }, { "epoch": 0.31, "learning_rate": 4.960346167074129e-05, "loss": 0.3793, "step": 1766 }, { "epoch": 0.31, "learning_rate": 4.9602181135025006e-05, "loss": 0.3627, "step": 1767 }, { "epoch": 0.31, "learning_rate": 4.960089855347358e-05, "loss": 0.3745, "step": 1768 }, { "epoch": 0.31, "learning_rate": 4.9599613926205734e-05, "loss": 0.3757, "step": 1769 }, { "epoch": 0.31, "learning_rate": 4.959832725334037e-05, "loss": 0.3698, "step": 1770 }, { "epoch": 0.31, "learning_rate": 4.959703853499659e-05, "loss": 0.3707, "step": 1771 }, { "epoch": 0.31, "learning_rate": 4.9595747771293686e-05, "loss": 0.4074, "step": 1772 }, { "epoch": 0.31, "learning_rate": 4.959445496235112e-05, "loss": 0.3826, "step": 1773 }, { "epoch": 0.31, "learning_rate": 4.959316010828857e-05, "loss": 0.3941, "step": 1774 }, { "epoch": 0.31, "learning_rate": 4.959186320922589e-05, "loss": 0.3658, "step": 1775 }, { "epoch": 0.31, "learning_rate": 4.959056426528311e-05, "loss": 0.3921, "step": 1776 }, { "epoch": 0.31, "learning_rate": 4.9589263276580484e-05, "loss": 0.3688, "step": 1777 }, { "epoch": 0.31, "learning_rate": 4.958796024323843e-05, "loss": 0.3873, "step": 1778 }, { "epoch": 0.31, "learning_rate": 4.958665516537755e-05, "loss": 0.3759, "step": 1779 }, { "epoch": 0.31, "learning_rate": 4.958534804311866e-05, "loss": 0.372, "step": 1780 }, { "epoch": 0.31, "learning_rate": 4.958403887658273e-05, "loss": 0.3795, "step": 1781 }, { "epoch": 0.31, "learning_rate": 4.958272766589096e-05, "loss": 0.3831, "step": 1782 }, { "epoch": 0.31, "learning_rate": 4.958141441116471e-05, "loss": 0.3652, "step": 1783 }, { "epoch": 0.31, "learning_rate": 4.958009911252555e-05, "loss": 0.3842, "step": 1784 }, { "epoch": 0.31, "learning_rate": 4.95787817700952e-05, "loss": 0.3604, "step": 1785 }, { "epoch": 0.31, "learning_rate": 4.9577462383995624e-05, "loss": 0.3908, "step": 1786 }, { "epoch": 0.31, "learning_rate": 4.957614095434893e-05, "loss": 0.3689, "step": 1787 }, { "epoch": 0.31, "learning_rate": 4.9574817481277444e-05, "loss": 0.3878, "step": 1788 }, { "epoch": 0.31, "learning_rate": 4.9573491964903665e-05, "loss": 0.3701, "step": 1789 }, { "epoch": 0.31, "learning_rate": 4.957216440535029e-05, "loss": 0.4017, "step": 1790 }, { "epoch": 0.31, "learning_rate": 4.9570834802740195e-05, "loss": 0.3831, "step": 1791 }, { "epoch": 0.31, "learning_rate": 4.956950315719646e-05, "loss": 0.3734, "step": 1792 }, { "epoch": 0.31, "learning_rate": 4.9568169468842344e-05, "loss": 0.3818, "step": 1793 }, { "epoch": 0.31, "learning_rate": 4.956683373780129e-05, "loss": 0.4034, "step": 1794 }, { "epoch": 0.31, "learning_rate": 4.956549596419694e-05, "loss": 0.3991, "step": 1795 }, { "epoch": 0.31, "learning_rate": 4.9564156148153125e-05, "loss": 0.3924, "step": 1796 }, { "epoch": 0.31, "learning_rate": 4.9562814289793855e-05, "loss": 0.3892, "step": 1797 }, { "epoch": 0.31, "learning_rate": 4.956147038924334e-05, "loss": 0.3836, "step": 1798 }, { "epoch": 0.31, "learning_rate": 4.9560124446625984e-05, "loss": 0.3596, "step": 1799 }, { "epoch": 0.31, "learning_rate": 4.955877646206635e-05, "loss": 0.3752, "step": 1800 }, { "epoch": 0.31, "learning_rate": 4.955742643568923e-05, "loss": 0.3917, "step": 1801 }, { "epoch": 0.31, "learning_rate": 4.955607436761958e-05, "loss": 0.391, "step": 1802 }, { "epoch": 0.31, "learning_rate": 4.955472025798254e-05, "loss": 0.3704, "step": 1803 }, { "epoch": 0.31, "learning_rate": 4.9553364106903466e-05, "loss": 0.3873, "step": 1804 }, { "epoch": 0.31, "learning_rate": 4.955200591450788e-05, "loss": 0.3623, "step": 1805 }, { "epoch": 0.31, "learning_rate": 4.95506456809215e-05, "loss": 0.3854, "step": 1806 }, { "epoch": 0.31, "learning_rate": 4.954928340627022e-05, "loss": 0.3938, "step": 1807 }, { "epoch": 0.32, "learning_rate": 4.9547919090680175e-05, "loss": 0.362, "step": 1808 }, { "epoch": 0.32, "learning_rate": 4.95465527342776e-05, "loss": 0.3776, "step": 1809 }, { "epoch": 0.32, "learning_rate": 4.954518433718901e-05, "loss": 0.3999, "step": 1810 }, { "epoch": 0.32, "learning_rate": 4.954381389954104e-05, "loss": 0.3858, "step": 1811 }, { "epoch": 0.32, "learning_rate": 4.954244142146055e-05, "loss": 0.3659, "step": 1812 }, { "epoch": 0.32, "learning_rate": 4.9541066903074584e-05, "loss": 0.3956, "step": 1813 }, { "epoch": 0.32, "learning_rate": 4.953969034451036e-05, "loss": 0.3819, "step": 1814 }, { "epoch": 0.32, "learning_rate": 4.953831174589531e-05, "loss": 0.3996, "step": 1815 }, { "epoch": 0.32, "learning_rate": 4.953693110735703e-05, "loss": 0.379, "step": 1816 }, { "epoch": 0.32, "learning_rate": 4.953554842902332e-05, "loss": 0.3663, "step": 1817 }, { "epoch": 0.32, "learning_rate": 4.9534163711022166e-05, "loss": 0.3853, "step": 1818 }, { "epoch": 0.32, "learning_rate": 4.9532776953481735e-05, "loss": 0.3847, "step": 1819 }, { "epoch": 0.32, "learning_rate": 4.95313881565304e-05, "loss": 0.395, "step": 1820 }, { "epoch": 0.32, "learning_rate": 4.95299973202967e-05, "loss": 0.3927, "step": 1821 }, { "epoch": 0.32, "learning_rate": 4.952860444490938e-05, "loss": 0.356, "step": 1822 }, { "epoch": 0.32, "learning_rate": 4.9527209530497365e-05, "loss": 0.3728, "step": 1823 }, { "epoch": 0.32, "learning_rate": 4.952581257718977e-05, "loss": 0.3926, "step": 1824 }, { "epoch": 0.32, "learning_rate": 4.952441358511591e-05, "loss": 0.4, "step": 1825 }, { "epoch": 0.32, "learning_rate": 4.952301255440527e-05, "loss": 0.4146, "step": 1826 }, { "epoch": 0.32, "learning_rate": 4.952160948518754e-05, "loss": 0.3785, "step": 1827 }, { "epoch": 0.32, "learning_rate": 4.952020437759258e-05, "loss": 0.4065, "step": 1828 }, { "epoch": 0.32, "learning_rate": 4.9518797231750474e-05, "loss": 0.3747, "step": 1829 }, { "epoch": 0.32, "learning_rate": 4.9517388047791444e-05, "loss": 0.371, "step": 1830 }, { "epoch": 0.32, "learning_rate": 4.951597682584594e-05, "loss": 0.3557, "step": 1831 }, { "epoch": 0.32, "learning_rate": 4.9514563566044594e-05, "loss": 0.3943, "step": 1832 }, { "epoch": 0.32, "learning_rate": 4.9513148268518216e-05, "loss": 0.3764, "step": 1833 }, { "epoch": 0.32, "learning_rate": 4.9511730933397807e-05, "loss": 0.3912, "step": 1834 }, { "epoch": 0.32, "learning_rate": 4.951031156081456e-05, "loss": 0.3891, "step": 1835 }, { "epoch": 0.32, "learning_rate": 4.950889015089986e-05, "loss": 0.3724, "step": 1836 }, { "epoch": 0.32, "learning_rate": 4.950746670378526e-05, "loss": 0.3731, "step": 1837 }, { "epoch": 0.32, "learning_rate": 4.950604121960254e-05, "loss": 0.3811, "step": 1838 }, { "epoch": 0.32, "learning_rate": 4.9504613698483645e-05, "loss": 0.3849, "step": 1839 }, { "epoch": 0.32, "learning_rate": 4.9503184140560686e-05, "loss": 0.3916, "step": 1840 }, { "epoch": 0.32, "learning_rate": 4.950175254596601e-05, "loss": 0.3751, "step": 1841 }, { "epoch": 0.32, "learning_rate": 4.950031891483213e-05, "loss": 0.4002, "step": 1842 }, { "epoch": 0.32, "learning_rate": 4.9498883247291725e-05, "loss": 0.3691, "step": 1843 }, { "epoch": 0.32, "learning_rate": 4.9497445543477706e-05, "loss": 0.3817, "step": 1844 }, { "epoch": 0.32, "learning_rate": 4.9496005803523144e-05, "loss": 0.3929, "step": 1845 }, { "epoch": 0.32, "learning_rate": 4.94945640275613e-05, "loss": 0.3839, "step": 1846 }, { "epoch": 0.32, "learning_rate": 4.949312021572563e-05, "loss": 0.3931, "step": 1847 }, { "epoch": 0.32, "learning_rate": 4.949167436814979e-05, "loss": 0.4098, "step": 1848 }, { "epoch": 0.32, "learning_rate": 4.949022648496759e-05, "loss": 0.3753, "step": 1849 }, { "epoch": 0.32, "learning_rate": 4.948877656631307e-05, "loss": 0.3746, "step": 1850 }, { "epoch": 0.32, "learning_rate": 4.948732461232041e-05, "loss": 0.3934, "step": 1851 }, { "epoch": 0.32, "learning_rate": 4.948587062312404e-05, "loss": 0.3805, "step": 1852 }, { "epoch": 0.32, "learning_rate": 4.948441459885853e-05, "loss": 0.3638, "step": 1853 }, { "epoch": 0.32, "learning_rate": 4.948295653965865e-05, "loss": 0.3895, "step": 1854 }, { "epoch": 0.32, "learning_rate": 4.948149644565936e-05, "loss": 0.3923, "step": 1855 }, { "epoch": 0.32, "learning_rate": 4.9480034316995826e-05, "loss": 0.3731, "step": 1856 }, { "epoch": 0.32, "learning_rate": 4.947857015380337e-05, "loss": 0.3673, "step": 1857 }, { "epoch": 0.32, "learning_rate": 4.947710395621753e-05, "loss": 0.3942, "step": 1858 }, { "epoch": 0.32, "learning_rate": 4.9475635724374005e-05, "loss": 0.3713, "step": 1859 }, { "epoch": 0.32, "learning_rate": 4.9474165458408714e-05, "loss": 0.3841, "step": 1860 }, { "epoch": 0.32, "learning_rate": 4.9472693158457746e-05, "loss": 0.3788, "step": 1861 }, { "epoch": 0.32, "learning_rate": 4.947121882465737e-05, "loss": 0.3867, "step": 1862 }, { "epoch": 0.32, "learning_rate": 4.9469742457144075e-05, "loss": 0.3736, "step": 1863 }, { "epoch": 0.32, "learning_rate": 4.9468264056054496e-05, "loss": 0.3621, "step": 1864 }, { "epoch": 0.33, "learning_rate": 4.9466783621525494e-05, "loss": 0.3809, "step": 1865 }, { "epoch": 0.33, "learning_rate": 4.9465301153694086e-05, "loss": 0.3867, "step": 1866 }, { "epoch": 0.33, "learning_rate": 4.946381665269751e-05, "loss": 0.3846, "step": 1867 }, { "epoch": 0.33, "learning_rate": 4.946233011867315e-05, "loss": 0.386, "step": 1868 }, { "epoch": 0.33, "learning_rate": 4.946084155175864e-05, "loss": 0.3815, "step": 1869 }, { "epoch": 0.33, "learning_rate": 4.945935095209173e-05, "loss": 0.3664, "step": 1870 }, { "epoch": 0.33, "learning_rate": 4.9457858319810426e-05, "loss": 0.3887, "step": 1871 }, { "epoch": 0.33, "learning_rate": 4.9456363655052864e-05, "loss": 0.3841, "step": 1872 }, { "epoch": 0.33, "learning_rate": 4.94548669579574e-05, "loss": 0.388, "step": 1873 }, { "epoch": 0.33, "learning_rate": 4.945336822866258e-05, "loss": 0.3572, "step": 1874 }, { "epoch": 0.33, "learning_rate": 4.945186746730713e-05, "loss": 0.3746, "step": 1875 }, { "epoch": 0.33, "learning_rate": 4.945036467402995e-05, "loss": 0.3959, "step": 1876 }, { "epoch": 0.33, "learning_rate": 4.944885984897016e-05, "loss": 0.3821, "step": 1877 }, { "epoch": 0.33, "learning_rate": 4.9447352992267045e-05, "loss": 0.3959, "step": 1878 }, { "epoch": 0.33, "learning_rate": 4.944584410406008e-05, "loss": 0.3876, "step": 1879 }, { "epoch": 0.33, "learning_rate": 4.9444333184488926e-05, "loss": 0.3583, "step": 1880 }, { "epoch": 0.33, "learning_rate": 4.944282023369345e-05, "loss": 0.3688, "step": 1881 }, { "epoch": 0.33, "learning_rate": 4.944130525181369e-05, "loss": 0.4121, "step": 1882 }, { "epoch": 0.33, "learning_rate": 4.943978823898988e-05, "loss": 0.3862, "step": 1883 }, { "epoch": 0.33, "learning_rate": 4.943826919536244e-05, "loss": 0.3911, "step": 1884 }, { "epoch": 0.33, "learning_rate": 4.943674812107196e-05, "loss": 0.3895, "step": 1885 }, { "epoch": 0.33, "learning_rate": 4.9435225016259244e-05, "loss": 0.3779, "step": 1886 }, { "epoch": 0.33, "learning_rate": 4.9433699881065286e-05, "loss": 0.3752, "step": 1887 }, { "epoch": 0.33, "learning_rate": 4.943217271563125e-05, "loss": 0.3841, "step": 1888 }, { "epoch": 0.33, "learning_rate": 4.943064352009848e-05, "loss": 0.3868, "step": 1889 }, { "epoch": 0.33, "learning_rate": 4.9429112294608546e-05, "loss": 0.3791, "step": 1890 }, { "epoch": 0.33, "learning_rate": 4.942757903930316e-05, "loss": 0.3659, "step": 1891 }, { "epoch": 0.33, "learning_rate": 4.942604375432425e-05, "loss": 0.38, "step": 1892 }, { "epoch": 0.33, "learning_rate": 4.942450643981394e-05, "loss": 0.3873, "step": 1893 }, { "epoch": 0.33, "learning_rate": 4.9422967095914515e-05, "loss": 0.3696, "step": 1894 }, { "epoch": 0.33, "learning_rate": 4.9421425722768466e-05, "loss": 0.3909, "step": 1895 }, { "epoch": 0.33, "learning_rate": 4.941988232051846e-05, "loss": 0.3835, "step": 1896 }, { "epoch": 0.33, "learning_rate": 4.9418336889307365e-05, "loss": 0.386, "step": 1897 }, { "epoch": 0.33, "learning_rate": 4.941678942927822e-05, "loss": 0.3942, "step": 1898 }, { "epoch": 0.33, "learning_rate": 4.941523994057428e-05, "loss": 0.3723, "step": 1899 }, { "epoch": 0.33, "learning_rate": 4.941368842333896e-05, "loss": 0.3915, "step": 1900 }, { "epoch": 0.33, "learning_rate": 4.941213487771586e-05, "loss": 0.3837, "step": 1901 }, { "epoch": 0.33, "learning_rate": 4.94105793038488e-05, "loss": 0.3716, "step": 1902 }, { "epoch": 0.33, "learning_rate": 4.940902170188176e-05, "loss": 0.3746, "step": 1903 }, { "epoch": 0.33, "learning_rate": 4.940746207195891e-05, "loss": 0.3803, "step": 1904 }, { "epoch": 0.33, "learning_rate": 4.940590041422462e-05, "loss": 0.3723, "step": 1905 }, { "epoch": 0.33, "learning_rate": 4.9404336728823444e-05, "loss": 0.3708, "step": 1906 }, { "epoch": 0.33, "learning_rate": 4.940277101590012e-05, "loss": 0.3926, "step": 1907 }, { "epoch": 0.33, "learning_rate": 4.9401203275599556e-05, "loss": 0.3804, "step": 1908 }, { "epoch": 0.33, "learning_rate": 4.939963350806689e-05, "loss": 0.3822, "step": 1909 }, { "epoch": 0.33, "learning_rate": 4.939806171344742e-05, "loss": 0.3714, "step": 1910 }, { "epoch": 0.33, "learning_rate": 4.939648789188662e-05, "loss": 0.3889, "step": 1911 }, { "epoch": 0.33, "learning_rate": 4.939491204353019e-05, "loss": 0.3732, "step": 1912 }, { "epoch": 0.33, "learning_rate": 4.9393334168523974e-05, "loss": 0.3921, "step": 1913 }, { "epoch": 0.33, "learning_rate": 4.939175426701403e-05, "loss": 0.38, "step": 1914 }, { "epoch": 0.33, "learning_rate": 4.939017233914661e-05, "loss": 0.3872, "step": 1915 }, { "epoch": 0.33, "learning_rate": 4.938858838506812e-05, "loss": 0.3837, "step": 1916 }, { "epoch": 0.33, "learning_rate": 4.938700240492518e-05, "loss": 0.3546, "step": 1917 }, { "epoch": 0.33, "learning_rate": 4.938541439886462e-05, "loss": 0.3778, "step": 1918 }, { "epoch": 0.33, "learning_rate": 4.9383824367033395e-05, "loss": 0.3774, "step": 1919 }, { "epoch": 0.33, "learning_rate": 4.9382232309578694e-05, "loss": 0.3771, "step": 1920 }, { "epoch": 0.33, "learning_rate": 4.9380638226647894e-05, "loss": 0.4018, "step": 1921 }, { "epoch": 0.34, "learning_rate": 4.937904211838853e-05, "loss": 0.376, "step": 1922 }, { "epoch": 0.34, "learning_rate": 4.9377443984948346e-05, "loss": 0.391, "step": 1923 }, { "epoch": 0.34, "learning_rate": 4.937584382647527e-05, "loss": 0.4104, "step": 1924 }, { "epoch": 0.34, "learning_rate": 4.9374241643117434e-05, "loss": 0.3832, "step": 1925 }, { "epoch": 0.34, "learning_rate": 4.9372637435023115e-05, "loss": 0.3971, "step": 1926 }, { "epoch": 0.34, "learning_rate": 4.937103120234081e-05, "loss": 0.3632, "step": 1927 }, { "epoch": 0.34, "learning_rate": 4.936942294521921e-05, "loss": 0.3684, "step": 1928 }, { "epoch": 0.34, "learning_rate": 4.936781266380717e-05, "loss": 0.3955, "step": 1929 }, { "epoch": 0.34, "learning_rate": 4.936620035825373e-05, "loss": 0.3845, "step": 1930 }, { "epoch": 0.34, "learning_rate": 4.936458602870814e-05, "loss": 0.3763, "step": 1931 }, { "epoch": 0.34, "learning_rate": 4.9362969675319836e-05, "loss": 0.3876, "step": 1932 }, { "epoch": 0.34, "learning_rate": 4.9361351298238414e-05, "loss": 0.3952, "step": 1933 }, { "epoch": 0.34, "learning_rate": 4.935973089761369e-05, "loss": 0.3662, "step": 1934 }, { "epoch": 0.34, "learning_rate": 4.9358108473595636e-05, "loss": 0.3947, "step": 1935 }, { "epoch": 0.34, "learning_rate": 4.9356484026334455e-05, "loss": 0.3752, "step": 1936 }, { "epoch": 0.34, "learning_rate": 4.935485755598047e-05, "loss": 0.3697, "step": 1937 }, { "epoch": 0.34, "learning_rate": 4.9353229062684266e-05, "loss": 0.4019, "step": 1938 }, { "epoch": 0.34, "learning_rate": 4.935159854659657e-05, "loss": 0.383, "step": 1939 }, { "epoch": 0.34, "learning_rate": 4.9349966007868306e-05, "loss": 0.3776, "step": 1940 }, { "epoch": 0.34, "learning_rate": 4.934833144665058e-05, "loss": 0.3926, "step": 1941 }, { "epoch": 0.34, "learning_rate": 4.934669486309471e-05, "loss": 0.3999, "step": 1942 }, { "epoch": 0.34, "learning_rate": 4.934505625735216e-05, "loss": 0.3896, "step": 1943 }, { "epoch": 0.34, "learning_rate": 4.93434156295746e-05, "loss": 0.3663, "step": 1944 }, { "epoch": 0.34, "learning_rate": 4.934177297991392e-05, "loss": 0.3873, "step": 1945 }, { "epoch": 0.34, "learning_rate": 4.9340128308522144e-05, "loss": 0.3793, "step": 1946 }, { "epoch": 0.34, "learning_rate": 4.9338481615551525e-05, "loss": 0.3869, "step": 1947 }, { "epoch": 0.34, "learning_rate": 4.9336832901154455e-05, "loss": 0.3778, "step": 1948 }, { "epoch": 0.34, "learning_rate": 4.933518216548358e-05, "loss": 0.3815, "step": 1949 }, { "epoch": 0.34, "learning_rate": 4.933352940869167e-05, "loss": 0.3827, "step": 1950 }, { "epoch": 0.34, "learning_rate": 4.933187463093172e-05, "loss": 0.38, "step": 1951 }, { "epoch": 0.34, "learning_rate": 4.933021783235689e-05, "loss": 0.3788, "step": 1952 }, { "epoch": 0.34, "learning_rate": 4.932855901312056e-05, "loss": 0.3701, "step": 1953 }, { "epoch": 0.34, "learning_rate": 4.932689817337626e-05, "loss": 0.3671, "step": 1954 }, { "epoch": 0.34, "learning_rate": 4.932523531327771e-05, "loss": 0.3823, "step": 1955 }, { "epoch": 0.34, "learning_rate": 4.932357043297884e-05, "loss": 0.3728, "step": 1956 }, { "epoch": 0.34, "learning_rate": 4.932190353263377e-05, "loss": 0.376, "step": 1957 }, { "epoch": 0.34, "learning_rate": 4.932023461239677e-05, "loss": 0.3845, "step": 1958 }, { "epoch": 0.34, "learning_rate": 4.931856367242233e-05, "loss": 0.3776, "step": 1959 }, { "epoch": 0.34, "learning_rate": 4.931689071286511e-05, "loss": 0.387, "step": 1960 }, { "epoch": 0.34, "learning_rate": 4.931521573387998e-05, "loss": 0.3748, "step": 1961 }, { "epoch": 0.34, "learning_rate": 4.9313538735621964e-05, "loss": 0.3798, "step": 1962 }, { "epoch": 0.34, "learning_rate": 4.931185971824629e-05, "loss": 0.3869, "step": 1963 }, { "epoch": 0.34, "learning_rate": 4.931017868190837e-05, "loss": 0.3796, "step": 1964 }, { "epoch": 0.34, "learning_rate": 4.930849562676382e-05, "loss": 0.368, "step": 1965 }, { "epoch": 0.34, "learning_rate": 4.930681055296842e-05, "loss": 0.3669, "step": 1966 }, { "epoch": 0.34, "learning_rate": 4.9305123460678146e-05, "loss": 0.3765, "step": 1967 }, { "epoch": 0.34, "learning_rate": 4.930343435004915e-05, "loss": 0.3616, "step": 1968 }, { "epoch": 0.34, "learning_rate": 4.9301743221237796e-05, "loss": 0.4006, "step": 1969 }, { "epoch": 0.34, "learning_rate": 4.9300050074400605e-05, "loss": 0.3847, "step": 1970 }, { "epoch": 0.34, "learning_rate": 4.9298354909694306e-05, "loss": 0.3934, "step": 1971 }, { "epoch": 0.34, "learning_rate": 4.9296657727275805e-05, "loss": 0.3885, "step": 1972 }, { "epoch": 0.34, "learning_rate": 4.929495852730221e-05, "loss": 0.3769, "step": 1973 }, { "epoch": 0.34, "learning_rate": 4.929325730993079e-05, "loss": 0.3723, "step": 1974 }, { "epoch": 0.34, "learning_rate": 4.9291554075319005e-05, "loss": 0.4167, "step": 1975 }, { "epoch": 0.34, "learning_rate": 4.928984882362454e-05, "loss": 0.3719, "step": 1976 }, { "epoch": 0.34, "learning_rate": 4.928814155500521e-05, "loss": 0.3783, "step": 1977 }, { "epoch": 0.34, "learning_rate": 4.928643226961906e-05, "loss": 0.3932, "step": 1978 }, { "epoch": 0.34, "learning_rate": 4.92847209676243e-05, "loss": 0.3674, "step": 1979 }, { "epoch": 0.35, "learning_rate": 4.928300764917933e-05, "loss": 0.3682, "step": 1980 }, { "epoch": 0.35, "learning_rate": 4.928129231444275e-05, "loss": 0.3836, "step": 1981 }, { "epoch": 0.35, "learning_rate": 4.927957496357332e-05, "loss": 0.3578, "step": 1982 }, { "epoch": 0.35, "learning_rate": 4.927785559673002e-05, "loss": 0.3904, "step": 1983 }, { "epoch": 0.35, "learning_rate": 4.9276134214071986e-05, "loss": 0.375, "step": 1984 }, { "epoch": 0.35, "learning_rate": 4.927441081575855e-05, "loss": 0.378, "step": 1985 }, { "epoch": 0.35, "learning_rate": 4.9272685401949245e-05, "loss": 0.3966, "step": 1986 }, { "epoch": 0.35, "learning_rate": 4.927095797280378e-05, "loss": 0.39, "step": 1987 }, { "epoch": 0.35, "learning_rate": 4.9269228528482046e-05, "loss": 0.3777, "step": 1988 }, { "epoch": 0.35, "learning_rate": 4.926749706914412e-05, "loss": 0.3835, "step": 1989 }, { "epoch": 0.35, "learning_rate": 4.926576359495028e-05, "loss": 0.3718, "step": 1990 }, { "epoch": 0.35, "learning_rate": 4.9264028106060984e-05, "loss": 0.3713, "step": 1991 }, { "epoch": 0.35, "learning_rate": 4.9262290602636854e-05, "loss": 0.3807, "step": 1992 }, { "epoch": 0.35, "learning_rate": 4.926055108483874e-05, "loss": 0.3556, "step": 1993 }, { "epoch": 0.35, "learning_rate": 4.925880955282764e-05, "loss": 0.3996, "step": 1994 }, { "epoch": 0.35, "learning_rate": 4.9257066006764764e-05, "loss": 0.385, "step": 1995 }, { "epoch": 0.35, "learning_rate": 4.925532044681149e-05, "loss": 0.3604, "step": 1996 }, { "epoch": 0.35, "learning_rate": 4.925357287312941e-05, "loss": 0.3754, "step": 1997 }, { "epoch": 0.35, "learning_rate": 4.925182328588026e-05, "loss": 0.3812, "step": 1998 }, { "epoch": 0.35, "learning_rate": 4.9250071685226004e-05, "loss": 0.3804, "step": 1999 }, { "epoch": 0.35, "learning_rate": 4.924831807132877e-05, "loss": 0.3811, "step": 2000 }, { "epoch": 0.35, "learning_rate": 4.9246562444350875e-05, "loss": 0.3701, "step": 2001 }, { "epoch": 0.35, "learning_rate": 4.924480480445482e-05, "loss": 0.3743, "step": 2002 }, { "epoch": 0.35, "learning_rate": 4.9243045151803306e-05, "loss": 0.3767, "step": 2003 }, { "epoch": 0.35, "learning_rate": 4.9241283486559204e-05, "loss": 0.3705, "step": 2004 }, { "epoch": 0.35, "learning_rate": 4.923951980888558e-05, "loss": 0.3683, "step": 2005 }, { "epoch": 0.35, "learning_rate": 4.923775411894569e-05, "loss": 0.3893, "step": 2006 }, { "epoch": 0.35, "learning_rate": 4.9235986416902965e-05, "loss": 0.385, "step": 2007 }, { "epoch": 0.35, "learning_rate": 4.9234216702921026e-05, "loss": 0.3853, "step": 2008 }, { "epoch": 0.35, "learning_rate": 4.9232444977163697e-05, "loss": 0.3618, "step": 2009 }, { "epoch": 0.35, "learning_rate": 4.923067123979495e-05, "loss": 0.373, "step": 2010 }, { "epoch": 0.35, "learning_rate": 4.922889549097898e-05, "loss": 0.3827, "step": 2011 }, { "epoch": 0.35, "learning_rate": 4.9227117730880154e-05, "loss": 0.3752, "step": 2012 }, { "epoch": 0.35, "learning_rate": 4.922533795966304e-05, "loss": 0.3917, "step": 2013 }, { "epoch": 0.35, "learning_rate": 4.9223556177492346e-05, "loss": 0.3856, "step": 2014 }, { "epoch": 0.35, "learning_rate": 4.922177238453303e-05, "loss": 0.3841, "step": 2015 }, { "epoch": 0.35, "learning_rate": 4.9219986580950186e-05, "loss": 0.3743, "step": 2016 }, { "epoch": 0.35, "learning_rate": 4.9218198766909116e-05, "loss": 0.4035, "step": 2017 }, { "epoch": 0.35, "learning_rate": 4.92164089425753e-05, "loss": 0.3887, "step": 2018 }, { "epoch": 0.35, "learning_rate": 4.921461710811443e-05, "loss": 0.3774, "step": 2019 }, { "epoch": 0.35, "learning_rate": 4.921282326369234e-05, "loss": 0.3621, "step": 2020 }, { "epoch": 0.35, "learning_rate": 4.921102740947509e-05, "loss": 0.3792, "step": 2021 }, { "epoch": 0.35, "learning_rate": 4.9209229545628886e-05, "loss": 0.3872, "step": 2022 }, { "epoch": 0.35, "learning_rate": 4.920742967232016e-05, "loss": 0.3695, "step": 2023 }, { "epoch": 0.35, "learning_rate": 4.920562778971552e-05, "loss": 0.362, "step": 2024 }, { "epoch": 0.35, "learning_rate": 4.9203823897981736e-05, "loss": 0.3781, "step": 2025 }, { "epoch": 0.35, "learning_rate": 4.9202017997285785e-05, "loss": 0.3665, "step": 2026 }, { "epoch": 0.35, "learning_rate": 4.920021008779483e-05, "loss": 0.3628, "step": 2027 }, { "epoch": 0.35, "learning_rate": 4.919840016967622e-05, "loss": 0.3854, "step": 2028 }, { "epoch": 0.35, "learning_rate": 4.9196588243097476e-05, "loss": 0.3887, "step": 2029 }, { "epoch": 0.35, "learning_rate": 4.919477430822632e-05, "loss": 0.3816, "step": 2030 }, { "epoch": 0.35, "learning_rate": 4.9192958365230646e-05, "loss": 0.3755, "step": 2031 }, { "epoch": 0.35, "learning_rate": 4.919114041427855e-05, "loss": 0.4, "step": 2032 }, { "epoch": 0.35, "learning_rate": 4.918932045553831e-05, "loss": 0.402, "step": 2033 }, { "epoch": 0.35, "learning_rate": 4.9187498489178384e-05, "loss": 0.3782, "step": 2034 }, { "epoch": 0.35, "learning_rate": 4.918567451536742e-05, "loss": 0.3693, "step": 2035 }, { "epoch": 0.35, "learning_rate": 4.918384853427424e-05, "loss": 0.3847, "step": 2036 }, { "epoch": 0.36, "learning_rate": 4.9182020546067856e-05, "loss": 0.3619, "step": 2037 }, { "epoch": 0.36, "learning_rate": 4.91801905509175e-05, "loss": 0.3918, "step": 2038 }, { "epoch": 0.36, "learning_rate": 4.9178358548992535e-05, "loss": 0.3943, "step": 2039 }, { "epoch": 0.36, "learning_rate": 4.917652454046254e-05, "loss": 0.3845, "step": 2040 }, { "epoch": 0.36, "learning_rate": 4.917468852549729e-05, "loss": 0.3565, "step": 2041 }, { "epoch": 0.36, "learning_rate": 4.9172850504266714e-05, "loss": 0.3544, "step": 2042 }, { "epoch": 0.36, "learning_rate": 4.917101047694096e-05, "loss": 0.3723, "step": 2043 }, { "epoch": 0.36, "learning_rate": 4.916916844369033e-05, "loss": 0.3776, "step": 2044 }, { "epoch": 0.36, "learning_rate": 4.916732440468533e-05, "loss": 0.3777, "step": 2045 }, { "epoch": 0.36, "learning_rate": 4.916547836009666e-05, "loss": 0.3665, "step": 2046 }, { "epoch": 0.36, "learning_rate": 4.916363031009519e-05, "loss": 0.3771, "step": 2047 }, { "epoch": 0.36, "learning_rate": 4.9161780254851966e-05, "loss": 0.3815, "step": 2048 }, { "epoch": 0.36, "learning_rate": 4.915992819453826e-05, "loss": 0.3758, "step": 2049 }, { "epoch": 0.36, "learning_rate": 4.915807412932548e-05, "loss": 0.3624, "step": 2050 }, { "epoch": 0.36, "learning_rate": 4.915621805938526e-05, "loss": 0.3604, "step": 2051 }, { "epoch": 0.36, "learning_rate": 4.9154359984889395e-05, "loss": 0.3726, "step": 2052 }, { "epoch": 0.36, "learning_rate": 4.915249990600987e-05, "loss": 0.3762, "step": 2053 }, { "epoch": 0.36, "learning_rate": 4.915063782291886e-05, "loss": 0.3662, "step": 2054 }, { "epoch": 0.36, "learning_rate": 4.9148773735788735e-05, "loss": 0.3836, "step": 2055 }, { "epoch": 0.36, "learning_rate": 4.914690764479202e-05, "loss": 0.3658, "step": 2056 }, { "epoch": 0.36, "learning_rate": 4.914503955010146e-05, "loss": 0.3649, "step": 2057 }, { "epoch": 0.36, "learning_rate": 4.914316945188997e-05, "loss": 0.3721, "step": 2058 }, { "epoch": 0.36, "learning_rate": 4.9141297350330654e-05, "loss": 0.3851, "step": 2059 }, { "epoch": 0.36, "learning_rate": 4.913942324559679e-05, "loss": 0.3805, "step": 2060 }, { "epoch": 0.36, "learning_rate": 4.9137547137861846e-05, "loss": 0.3925, "step": 2061 }, { "epoch": 0.36, "learning_rate": 4.913566902729949e-05, "loss": 0.3903, "step": 2062 }, { "epoch": 0.36, "learning_rate": 4.9133788914083555e-05, "loss": 0.4147, "step": 2063 }, { "epoch": 0.36, "learning_rate": 4.913190679838808e-05, "loss": 0.3594, "step": 2064 }, { "epoch": 0.36, "learning_rate": 4.9130022680387284e-05, "loss": 0.379, "step": 2065 }, { "epoch": 0.36, "learning_rate": 4.912813656025554e-05, "loss": 0.3757, "step": 2066 }, { "epoch": 0.36, "learning_rate": 4.912624843816746e-05, "loss": 0.3777, "step": 2067 }, { "epoch": 0.36, "learning_rate": 4.9124358314297794e-05, "loss": 0.369, "step": 2068 }, { "epoch": 0.36, "learning_rate": 4.91224661888215e-05, "loss": 0.3772, "step": 2069 }, { "epoch": 0.36, "learning_rate": 4.912057206191373e-05, "loss": 0.3644, "step": 2070 }, { "epoch": 0.36, "learning_rate": 4.91186759337498e-05, "loss": 0.3933, "step": 2071 }, { "epoch": 0.36, "learning_rate": 4.911677780450522e-05, "loss": 0.3566, "step": 2072 }, { "epoch": 0.36, "learning_rate": 4.911487767435569e-05, "loss": 0.3742, "step": 2073 }, { "epoch": 0.36, "learning_rate": 4.911297554347708e-05, "loss": 0.3814, "step": 2074 }, { "epoch": 0.36, "learning_rate": 4.911107141204547e-05, "loss": 0.373, "step": 2075 }, { "epoch": 0.36, "learning_rate": 4.910916528023711e-05, "loss": 0.3775, "step": 2076 }, { "epoch": 0.36, "learning_rate": 4.9107257148228434e-05, "loss": 0.3818, "step": 2077 }, { "epoch": 0.36, "learning_rate": 4.910534701619605e-05, "loss": 0.3668, "step": 2078 }, { "epoch": 0.36, "learning_rate": 4.910343488431679e-05, "loss": 0.3815, "step": 2079 }, { "epoch": 0.36, "learning_rate": 4.9101520752767625e-05, "loss": 0.3779, "step": 2080 }, { "epoch": 0.36, "learning_rate": 4.9099604621725745e-05, "loss": 0.3708, "step": 2081 }, { "epoch": 0.36, "learning_rate": 4.90976864913685e-05, "loss": 0.3947, "step": 2082 }, { "epoch": 0.36, "learning_rate": 4.9095766361873446e-05, "loss": 0.3726, "step": 2083 }, { "epoch": 0.36, "learning_rate": 4.9093844233418314e-05, "loss": 0.3647, "step": 2084 }, { "epoch": 0.36, "learning_rate": 4.909192010618102e-05, "loss": 0.3838, "step": 2085 }, { "epoch": 0.36, "learning_rate": 4.9089993980339663e-05, "loss": 0.3915, "step": 2086 }, { "epoch": 0.36, "learning_rate": 4.908806585607254e-05, "loss": 0.3896, "step": 2087 }, { "epoch": 0.36, "learning_rate": 4.908613573355812e-05, "loss": 0.3799, "step": 2088 }, { "epoch": 0.36, "learning_rate": 4.908420361297505e-05, "loss": 0.385, "step": 2089 }, { "epoch": 0.36, "learning_rate": 4.9082269494502176e-05, "loss": 0.3946, "step": 2090 }, { "epoch": 0.36, "learning_rate": 4.908033337831855e-05, "loss": 0.3763, "step": 2091 }, { "epoch": 0.36, "learning_rate": 4.9078395264603335e-05, "loss": 0.3687, "step": 2092 }, { "epoch": 0.36, "learning_rate": 4.9076455153535964e-05, "loss": 0.3921, "step": 2093 }, { "epoch": 0.36, "learning_rate": 4.9074513045296e-05, "loss": 0.382, "step": 2094 }, { "epoch": 0.37, "learning_rate": 4.907256894006324e-05, "loss": 0.3735, "step": 2095 }, { "epoch": 0.37, "learning_rate": 4.907062283801761e-05, "loss": 0.3672, "step": 2096 }, { "epoch": 0.37, "learning_rate": 4.9068674739339236e-05, "loss": 0.3696, "step": 2097 }, { "epoch": 0.37, "learning_rate": 4.906672464420847e-05, "loss": 0.3699, "step": 2098 }, { "epoch": 0.37, "learning_rate": 4.9064772552805795e-05, "loss": 0.3742, "step": 2099 }, { "epoch": 0.37, "learning_rate": 4.906281846531191e-05, "loss": 0.3723, "step": 2100 }, { "epoch": 0.37, "learning_rate": 4.906086238190769e-05, "loss": 0.3609, "step": 2101 }, { "epoch": 0.37, "learning_rate": 4.905890430277418e-05, "loss": 0.3974, "step": 2102 }, { "epoch": 0.37, "learning_rate": 4.905694422809265e-05, "loss": 0.3797, "step": 2103 }, { "epoch": 0.37, "learning_rate": 4.905498215804452e-05, "loss": 0.3866, "step": 2104 }, { "epoch": 0.37, "learning_rate": 4.90530180928114e-05, "loss": 0.3644, "step": 2105 }, { "epoch": 0.37, "learning_rate": 4.905105203257509e-05, "loss": 0.363, "step": 2106 }, { "epoch": 0.37, "learning_rate": 4.904908397751758e-05, "loss": 0.3646, "step": 2107 }, { "epoch": 0.37, "learning_rate": 4.9047113927821034e-05, "loss": 0.3792, "step": 2108 }, { "epoch": 0.37, "learning_rate": 4.9045141883667794e-05, "loss": 0.3921, "step": 2109 }, { "epoch": 0.37, "learning_rate": 4.904316784524041e-05, "loss": 0.386, "step": 2110 }, { "epoch": 0.37, "learning_rate": 4.904119181272162e-05, "loss": 0.3806, "step": 2111 }, { "epoch": 0.37, "learning_rate": 4.9039213786294293e-05, "loss": 0.3876, "step": 2112 }, { "epoch": 0.37, "learning_rate": 4.9037233766141554e-05, "loss": 0.376, "step": 2113 }, { "epoch": 0.37, "learning_rate": 4.9035251752446656e-05, "loss": 0.4053, "step": 2114 }, { "epoch": 0.37, "learning_rate": 4.9033267745393066e-05, "loss": 0.3665, "step": 2115 }, { "epoch": 0.37, "learning_rate": 4.9031281745164436e-05, "loss": 0.3855, "step": 2116 }, { "epoch": 0.37, "learning_rate": 4.902929375194459e-05, "loss": 0.3703, "step": 2117 }, { "epoch": 0.37, "learning_rate": 4.902730376591754e-05, "loss": 0.391, "step": 2118 }, { "epoch": 0.37, "learning_rate": 4.9025311787267485e-05, "loss": 0.38, "step": 2119 }, { "epoch": 0.37, "learning_rate": 4.902331781617882e-05, "loss": 0.3866, "step": 2120 }, { "epoch": 0.37, "learning_rate": 4.902132185283609e-05, "loss": 0.3641, "step": 2121 }, { "epoch": 0.37, "learning_rate": 4.901932389742406e-05, "loss": 0.3647, "step": 2122 }, { "epoch": 0.37, "learning_rate": 4.901732395012766e-05, "loss": 0.3796, "step": 2123 }, { "epoch": 0.37, "learning_rate": 4.9015322011132024e-05, "loss": 0.3723, "step": 2124 }, { "epoch": 0.37, "learning_rate": 4.901331808062244e-05, "loss": 0.3799, "step": 2125 }, { "epoch": 0.37, "learning_rate": 4.90113121587844e-05, "loss": 0.3794, "step": 2126 }, { "epoch": 0.37, "learning_rate": 4.9009304245803584e-05, "loss": 0.3757, "step": 2127 }, { "epoch": 0.37, "learning_rate": 4.900729434186585e-05, "loss": 0.4109, "step": 2128 }, { "epoch": 0.37, "learning_rate": 4.900528244715723e-05, "loss": 0.3676, "step": 2129 }, { "epoch": 0.37, "learning_rate": 4.900326856186396e-05, "loss": 0.3802, "step": 2130 }, { "epoch": 0.37, "learning_rate": 4.9001252686172456e-05, "loss": 0.378, "step": 2131 }, { "epoch": 0.37, "learning_rate": 4.899923482026929e-05, "loss": 0.3755, "step": 2132 }, { "epoch": 0.37, "learning_rate": 4.899721496434126e-05, "loss": 0.3693, "step": 2133 }, { "epoch": 0.37, "learning_rate": 4.8995193118575325e-05, "loss": 0.3709, "step": 2134 }, { "epoch": 0.37, "learning_rate": 4.899316928315863e-05, "loss": 0.3686, "step": 2135 }, { "epoch": 0.37, "learning_rate": 4.899114345827851e-05, "loss": 0.3729, "step": 2136 }, { "epoch": 0.37, "learning_rate": 4.8989115644122474e-05, "loss": 0.3724, "step": 2137 }, { "epoch": 0.37, "learning_rate": 4.8987085840878226e-05, "loss": 0.377, "step": 2138 }, { "epoch": 0.37, "learning_rate": 4.898505404873366e-05, "loss": 0.3723, "step": 2139 }, { "epoch": 0.37, "learning_rate": 4.898302026787682e-05, "loss": 0.3641, "step": 2140 }, { "epoch": 0.37, "learning_rate": 4.898098449849597e-05, "loss": 0.3768, "step": 2141 }, { "epoch": 0.37, "learning_rate": 4.897894674077956e-05, "loss": 0.3948, "step": 2142 }, { "epoch": 0.37, "learning_rate": 4.897690699491619e-05, "loss": 0.3645, "step": 2143 }, { "epoch": 0.37, "learning_rate": 4.8974865261094676e-05, "loss": 0.3571, "step": 2144 }, { "epoch": 0.37, "learning_rate": 4.8972821539504e-05, "loss": 0.3764, "step": 2145 }, { "epoch": 0.37, "learning_rate": 4.897077583033334e-05, "loss": 0.3681, "step": 2146 }, { "epoch": 0.37, "learning_rate": 4.896872813377204e-05, "loss": 0.3831, "step": 2147 }, { "epoch": 0.37, "learning_rate": 4.896667845000966e-05, "loss": 0.3734, "step": 2148 }, { "epoch": 0.37, "learning_rate": 4.89646267792359e-05, "loss": 0.3781, "step": 2149 }, { "epoch": 0.37, "learning_rate": 4.8962573121640684e-05, "loss": 0.3752, "step": 2150 }, { "epoch": 0.37, "learning_rate": 4.896051747741411e-05, "loss": 0.3795, "step": 2151 }, { "epoch": 0.38, "learning_rate": 4.895845984674643e-05, "loss": 0.3721, "step": 2152 }, { "epoch": 0.38, "learning_rate": 4.8956400229828124e-05, "loss": 0.3747, "step": 2153 }, { "epoch": 0.38, "learning_rate": 4.895433862684983e-05, "loss": 0.3661, "step": 2154 }, { "epoch": 0.38, "learning_rate": 4.8952275038002384e-05, "loss": 0.3669, "step": 2155 }, { "epoch": 0.38, "learning_rate": 4.895020946347677e-05, "loss": 0.3822, "step": 2156 }, { "epoch": 0.38, "learning_rate": 4.894814190346421e-05, "loss": 0.3702, "step": 2157 }, { "epoch": 0.38, "learning_rate": 4.894607235815608e-05, "loss": 0.3761, "step": 2158 }, { "epoch": 0.38, "learning_rate": 4.8944000827743924e-05, "loss": 0.3639, "step": 2159 }, { "epoch": 0.38, "learning_rate": 4.894192731241951e-05, "loss": 0.3756, "step": 2160 }, { "epoch": 0.38, "learning_rate": 4.893985181237475e-05, "loss": 0.3676, "step": 2161 }, { "epoch": 0.38, "learning_rate": 4.893777432780177e-05, "loss": 0.3725, "step": 2162 }, { "epoch": 0.38, "learning_rate": 4.893569485889287e-05, "loss": 0.3728, "step": 2163 }, { "epoch": 0.38, "learning_rate": 4.8933613405840506e-05, "loss": 0.3864, "step": 2164 }, { "epoch": 0.38, "learning_rate": 4.893152996883737e-05, "loss": 0.3812, "step": 2165 }, { "epoch": 0.38, "learning_rate": 4.89294445480763e-05, "loss": 0.3651, "step": 2166 }, { "epoch": 0.38, "learning_rate": 4.892735714375034e-05, "loss": 0.3695, "step": 2167 }, { "epoch": 0.38, "learning_rate": 4.8925267756052685e-05, "loss": 0.374, "step": 2168 }, { "epoch": 0.38, "learning_rate": 4.892317638517674e-05, "loss": 0.3641, "step": 2169 }, { "epoch": 0.38, "learning_rate": 4.89210830313161e-05, "loss": 0.3654, "step": 2170 }, { "epoch": 0.38, "learning_rate": 4.8918987694664515e-05, "loss": 0.3911, "step": 2171 }, { "epoch": 0.38, "learning_rate": 4.8916890375415944e-05, "loss": 0.3685, "step": 2172 }, { "epoch": 0.38, "learning_rate": 4.891479107376452e-05, "loss": 0.374, "step": 2173 }, { "epoch": 0.38, "learning_rate": 4.8912689789904566e-05, "loss": 0.3647, "step": 2174 }, { "epoch": 0.38, "learning_rate": 4.891058652403056e-05, "loss": 0.3647, "step": 2175 }, { "epoch": 0.38, "learning_rate": 4.890848127633721e-05, "loss": 0.3666, "step": 2176 }, { "epoch": 0.38, "learning_rate": 4.8906374047019374e-05, "loss": 0.3915, "step": 2177 }, { "epoch": 0.38, "learning_rate": 4.89042648362721e-05, "loss": 0.3749, "step": 2178 }, { "epoch": 0.38, "learning_rate": 4.890215364429062e-05, "loss": 0.3667, "step": 2179 }, { "epoch": 0.38, "learning_rate": 4.890004047127037e-05, "loss": 0.3837, "step": 2180 }, { "epoch": 0.38, "learning_rate": 4.889792531740693e-05, "loss": 0.3848, "step": 2181 }, { "epoch": 0.38, "learning_rate": 4.8895808182896086e-05, "loss": 0.3875, "step": 2182 }, { "epoch": 0.38, "learning_rate": 4.889368906793381e-05, "loss": 0.3799, "step": 2183 }, { "epoch": 0.38, "learning_rate": 4.889156797271626e-05, "loss": 0.371, "step": 2184 }, { "epoch": 0.38, "learning_rate": 4.8889444897439756e-05, "loss": 0.3702, "step": 2185 }, { "epoch": 0.38, "learning_rate": 4.888731984230082e-05, "loss": 0.3775, "step": 2186 }, { "epoch": 0.38, "learning_rate": 4.888519280749616e-05, "loss": 0.3963, "step": 2187 }, { "epoch": 0.38, "learning_rate": 4.888306379322266e-05, "loss": 0.373, "step": 2188 }, { "epoch": 0.38, "learning_rate": 4.888093279967737e-05, "loss": 0.3764, "step": 2189 }, { "epoch": 0.38, "learning_rate": 4.887879982705756e-05, "loss": 0.3835, "step": 2190 }, { "epoch": 0.38, "learning_rate": 4.887666487556065e-05, "loss": 0.3667, "step": 2191 }, { "epoch": 0.38, "learning_rate": 4.887452794538427e-05, "loss": 0.378, "step": 2192 }, { "epoch": 0.38, "learning_rate": 4.887238903672621e-05, "loss": 0.378, "step": 2193 }, { "epoch": 0.38, "learning_rate": 4.887024814978445e-05, "loss": 0.3748, "step": 2194 }, { "epoch": 0.38, "learning_rate": 4.8868105284757163e-05, "loss": 0.3621, "step": 2195 }, { "epoch": 0.38, "learning_rate": 4.8865960441842695e-05, "loss": 0.3811, "step": 2196 }, { "epoch": 0.38, "learning_rate": 4.886381362123959e-05, "loss": 0.3698, "step": 2197 }, { "epoch": 0.38, "learning_rate": 4.886166482314654e-05, "loss": 0.3771, "step": 2198 }, { "epoch": 0.38, "learning_rate": 4.885951404776246e-05, "loss": 0.3572, "step": 2199 }, { "epoch": 0.38, "learning_rate": 4.885736129528642e-05, "loss": 0.3867, "step": 2200 }, { "epoch": 0.38, "learning_rate": 4.8855206565917704e-05, "loss": 0.3643, "step": 2201 }, { "epoch": 0.38, "learning_rate": 4.885304985985575e-05, "loss": 0.374, "step": 2202 }, { "epoch": 0.38, "learning_rate": 4.8850891177300176e-05, "loss": 0.3686, "step": 2203 }, { "epoch": 0.38, "learning_rate": 4.88487305184508e-05, "loss": 0.3808, "step": 2204 }, { "epoch": 0.38, "learning_rate": 4.884656788350763e-05, "loss": 0.3667, "step": 2205 }, { "epoch": 0.38, "learning_rate": 4.884440327267084e-05, "loss": 0.3865, "step": 2206 }, { "epoch": 0.38, "learning_rate": 4.8842236686140784e-05, "loss": 0.3689, "step": 2207 }, { "epoch": 0.38, "learning_rate": 4.8840068124118014e-05, "loss": 0.3874, "step": 2208 }, { "epoch": 0.39, "learning_rate": 4.883789758680326e-05, "loss": 0.3661, "step": 2209 }, { "epoch": 0.39, "learning_rate": 4.8835725074397424e-05, "loss": 0.3657, "step": 2210 }, { "epoch": 0.39, "learning_rate": 4.883355058710161e-05, "loss": 0.3822, "step": 2211 }, { "epoch": 0.39, "learning_rate": 4.8831374125117083e-05, "loss": 0.3967, "step": 2212 }, { "epoch": 0.39, "learning_rate": 4.882919568864531e-05, "loss": 0.375, "step": 2213 }, { "epoch": 0.39, "learning_rate": 4.882701527788794e-05, "loss": 0.3753, "step": 2214 }, { "epoch": 0.39, "learning_rate": 4.8824832893046773e-05, "loss": 0.3639, "step": 2215 }, { "epoch": 0.39, "learning_rate": 4.8822648534323833e-05, "loss": 0.3785, "step": 2216 }, { "epoch": 0.39, "learning_rate": 4.882046220192131e-05, "loss": 0.3622, "step": 2217 }, { "epoch": 0.39, "learning_rate": 4.8818273896041575e-05, "loss": 0.3779, "step": 2218 }, { "epoch": 0.39, "learning_rate": 4.881608361688718e-05, "loss": 0.3837, "step": 2219 }, { "epoch": 0.39, "learning_rate": 4.8813891364660876e-05, "loss": 0.3891, "step": 2220 }, { "epoch": 0.39, "learning_rate": 4.881169713956556e-05, "loss": 0.3747, "step": 2221 }, { "epoch": 0.39, "learning_rate": 4.8809500941804354e-05, "loss": 0.3649, "step": 2222 }, { "epoch": 0.39, "learning_rate": 4.8807302771580535e-05, "loss": 0.3616, "step": 2223 }, { "epoch": 0.39, "learning_rate": 4.8805102629097574e-05, "loss": 0.3685, "step": 2224 }, { "epoch": 0.39, "learning_rate": 4.880290051455913e-05, "loss": 0.3759, "step": 2225 }, { "epoch": 0.39, "learning_rate": 4.880069642816902e-05, "loss": 0.3658, "step": 2226 }, { "epoch": 0.39, "learning_rate": 4.879849037013127e-05, "loss": 0.37, "step": 2227 }, { "epoch": 0.39, "learning_rate": 4.879628234065008e-05, "loss": 0.3565, "step": 2228 }, { "epoch": 0.39, "learning_rate": 4.879407233992982e-05, "loss": 0.3915, "step": 2229 }, { "epoch": 0.39, "learning_rate": 4.879186036817507e-05, "loss": 0.3682, "step": 2230 }, { "epoch": 0.39, "learning_rate": 4.878964642559057e-05, "loss": 0.366, "step": 2231 }, { "epoch": 0.39, "learning_rate": 4.878743051238123e-05, "loss": 0.3723, "step": 2232 }, { "epoch": 0.39, "learning_rate": 4.878521262875219e-05, "loss": 0.3935, "step": 2233 }, { "epoch": 0.39, "learning_rate": 4.8782992774908724e-05, "loss": 0.3751, "step": 2234 }, { "epoch": 0.39, "learning_rate": 4.878077095105632e-05, "loss": 0.3789, "step": 2235 }, { "epoch": 0.39, "learning_rate": 4.877854715740062e-05, "loss": 0.3737, "step": 2236 }, { "epoch": 0.39, "learning_rate": 4.877632139414747e-05, "loss": 0.3878, "step": 2237 }, { "epoch": 0.39, "learning_rate": 4.877409366150291e-05, "loss": 0.3636, "step": 2238 }, { "epoch": 0.39, "learning_rate": 4.877186395967312e-05, "loss": 0.4243, "step": 2239 }, { "epoch": 0.39, "learning_rate": 4.876963228886449e-05, "loss": 0.3886, "step": 2240 }, { "epoch": 0.39, "learning_rate": 4.87673986492836e-05, "loss": 0.3803, "step": 2241 }, { "epoch": 0.39, "learning_rate": 4.8765163041137196e-05, "loss": 0.379, "step": 2242 }, { "epoch": 0.39, "learning_rate": 4.876292546463222e-05, "loss": 0.3877, "step": 2243 }, { "epoch": 0.39, "learning_rate": 4.8760685919975773e-05, "loss": 0.3838, "step": 2244 }, { "epoch": 0.39, "learning_rate": 4.875844440737517e-05, "loss": 0.3751, "step": 2245 }, { "epoch": 0.39, "learning_rate": 4.875620092703788e-05, "loss": 0.3702, "step": 2246 }, { "epoch": 0.39, "learning_rate": 4.875395547917156e-05, "loss": 0.3872, "step": 2247 }, { "epoch": 0.39, "learning_rate": 4.875170806398406e-05, "loss": 0.3939, "step": 2248 }, { "epoch": 0.39, "learning_rate": 4.874945868168341e-05, "loss": 0.3747, "step": 2249 }, { "epoch": 0.39, "learning_rate": 4.8747207332477825e-05, "loss": 0.3721, "step": 2250 }, { "epoch": 0.39, "learning_rate": 4.8744954016575685e-05, "loss": 0.3697, "step": 2251 }, { "epoch": 0.39, "learning_rate": 4.8742698734185564e-05, "loss": 0.387, "step": 2252 }, { "epoch": 0.39, "learning_rate": 4.874044148551621e-05, "loss": 0.3881, "step": 2253 }, { "epoch": 0.39, "learning_rate": 4.873818227077658e-05, "loss": 0.3786, "step": 2254 }, { "epoch": 0.39, "learning_rate": 4.8735921090175784e-05, "loss": 0.3823, "step": 2255 }, { "epoch": 0.39, "learning_rate": 4.873365794392311e-05, "loss": 0.3902, "step": 2256 }, { "epoch": 0.39, "learning_rate": 4.8731392832228055e-05, "loss": 0.3691, "step": 2257 }, { "epoch": 0.39, "learning_rate": 4.8729125755300284e-05, "loss": 0.3855, "step": 2258 }, { "epoch": 0.39, "learning_rate": 4.8726856713349636e-05, "loss": 0.3831, "step": 2259 }, { "epoch": 0.39, "learning_rate": 4.8724585706586126e-05, "loss": 0.3649, "step": 2260 }, { "epoch": 0.39, "learning_rate": 4.8722312735219996e-05, "loss": 0.3793, "step": 2261 }, { "epoch": 0.39, "learning_rate": 4.872003779946162e-05, "loss": 0.3709, "step": 2262 }, { "epoch": 0.39, "learning_rate": 4.871776089952157e-05, "loss": 0.3945, "step": 2263 }, { "epoch": 0.39, "learning_rate": 4.871548203561061e-05, "loss": 0.3609, "step": 2264 }, { "epoch": 0.39, "learning_rate": 4.871320120793967e-05, "loss": 0.3849, "step": 2265 }, { "epoch": 0.39, "learning_rate": 4.871091841671987e-05, "loss": 0.3633, "step": 2266 }, { "epoch": 0.4, "learning_rate": 4.8708633662162525e-05, "loss": 0.361, "step": 2267 }, { "epoch": 0.4, "learning_rate": 4.87063469444791e-05, "loss": 0.358, "step": 2268 }, { "epoch": 0.4, "learning_rate": 4.870405826388126e-05, "loss": 0.3735, "step": 2269 }, { "epoch": 0.4, "learning_rate": 4.870176762058086e-05, "loss": 0.395, "step": 2270 }, { "epoch": 0.4, "learning_rate": 4.8699475014789936e-05, "loss": 0.3514, "step": 2271 }, { "epoch": 0.4, "learning_rate": 4.869718044672068e-05, "loss": 0.3812, "step": 2272 }, { "epoch": 0.4, "learning_rate": 4.8694883916585486e-05, "loss": 0.3795, "step": 2273 }, { "epoch": 0.4, "learning_rate": 4.8692585424596936e-05, "loss": 0.3793, "step": 2274 }, { "epoch": 0.4, "learning_rate": 4.869028497096778e-05, "loss": 0.3668, "step": 2275 }, { "epoch": 0.4, "learning_rate": 4.868798255591095e-05, "loss": 0.3692, "step": 2276 }, { "epoch": 0.4, "learning_rate": 4.868567817963957e-05, "loss": 0.3921, "step": 2277 }, { "epoch": 0.4, "learning_rate": 4.868337184236693e-05, "loss": 0.3728, "step": 2278 }, { "epoch": 0.4, "learning_rate": 4.868106354430652e-05, "loss": 0.3638, "step": 2279 }, { "epoch": 0.4, "learning_rate": 4.8678753285671996e-05, "loss": 0.3663, "step": 2280 }, { "epoch": 0.4, "learning_rate": 4.8676441066677206e-05, "loss": 0.3517, "step": 2281 }, { "epoch": 0.4, "learning_rate": 4.867412688753617e-05, "loss": 0.3773, "step": 2282 }, { "epoch": 0.4, "learning_rate": 4.86718107484631e-05, "loss": 0.3763, "step": 2283 }, { "epoch": 0.4, "learning_rate": 4.8669492649672384e-05, "loss": 0.3806, "step": 2284 }, { "epoch": 0.4, "learning_rate": 4.866717259137858e-05, "loss": 0.3712, "step": 2285 }, { "epoch": 0.4, "learning_rate": 4.866485057379645e-05, "loss": 0.3694, "step": 2286 }, { "epoch": 0.4, "learning_rate": 4.866252659714092e-05, "loss": 0.3905, "step": 2287 }, { "epoch": 0.4, "learning_rate": 4.866020066162711e-05, "loss": 0.3603, "step": 2288 }, { "epoch": 0.4, "learning_rate": 4.865787276747031e-05, "loss": 0.3894, "step": 2289 }, { "epoch": 0.4, "learning_rate": 4.8655542914885995e-05, "loss": 0.3638, "step": 2290 }, { "epoch": 0.4, "learning_rate": 4.865321110408983e-05, "loss": 0.3737, "step": 2291 }, { "epoch": 0.4, "learning_rate": 4.865087733529764e-05, "loss": 0.387, "step": 2292 }, { "epoch": 0.4, "learning_rate": 4.864854160872546e-05, "loss": 0.3701, "step": 2293 }, { "epoch": 0.4, "learning_rate": 4.864620392458947e-05, "loss": 0.3739, "step": 2294 }, { "epoch": 0.4, "learning_rate": 4.864386428310607e-05, "loss": 0.3516, "step": 2295 }, { "epoch": 0.4, "learning_rate": 4.8641522684491816e-05, "loss": 0.3721, "step": 2296 }, { "epoch": 0.4, "learning_rate": 4.8639179128963466e-05, "loss": 0.3858, "step": 2297 }, { "epoch": 0.4, "learning_rate": 4.863683361673793e-05, "loss": 0.3699, "step": 2298 }, { "epoch": 0.4, "learning_rate": 4.8634486148032314e-05, "loss": 0.3856, "step": 2299 }, { "epoch": 0.4, "learning_rate": 4.863213672306391e-05, "loss": 0.3626, "step": 2300 }, { "epoch": 0.4, "learning_rate": 4.862978534205019e-05, "loss": 0.3708, "step": 2301 }, { "epoch": 0.4, "learning_rate": 4.862743200520881e-05, "loss": 0.3694, "step": 2302 }, { "epoch": 0.4, "learning_rate": 4.8625076712757575e-05, "loss": 0.3716, "step": 2303 }, { "epoch": 0.4, "learning_rate": 4.862271946491452e-05, "loss": 0.3766, "step": 2304 }, { "epoch": 0.4, "learning_rate": 4.862036026189784e-05, "loss": 0.378, "step": 2305 }, { "epoch": 0.4, "learning_rate": 4.86179991039259e-05, "loss": 0.3834, "step": 2306 }, { "epoch": 0.4, "learning_rate": 4.861563599121725e-05, "loss": 0.3804, "step": 2307 }, { "epoch": 0.4, "learning_rate": 4.8613270923990636e-05, "loss": 0.3687, "step": 2308 }, { "epoch": 0.4, "learning_rate": 4.8610903902464966e-05, "loss": 0.3862, "step": 2309 }, { "epoch": 0.4, "learning_rate": 4.860853492685935e-05, "loss": 0.3886, "step": 2310 }, { "epoch": 0.4, "learning_rate": 4.860616399739305e-05, "loss": 0.364, "step": 2311 }, { "epoch": 0.4, "learning_rate": 4.860379111428555e-05, "loss": 0.3824, "step": 2312 }, { "epoch": 0.4, "learning_rate": 4.860141627775646e-05, "loss": 0.3681, "step": 2313 }, { "epoch": 0.4, "learning_rate": 4.8599039488025615e-05, "loss": 0.3726, "step": 2314 }, { "epoch": 0.4, "learning_rate": 4.8596660745313024e-05, "loss": 0.3938, "step": 2315 }, { "epoch": 0.4, "learning_rate": 4.8594280049838865e-05, "loss": 0.3714, "step": 2316 }, { "epoch": 0.4, "learning_rate": 4.859189740182349e-05, "loss": 0.3629, "step": 2317 }, { "epoch": 0.4, "learning_rate": 4.858951280148745e-05, "loss": 0.3814, "step": 2318 }, { "epoch": 0.4, "learning_rate": 4.8587126249051484e-05, "loss": 0.3779, "step": 2319 }, { "epoch": 0.4, "learning_rate": 4.858473774473648e-05, "loss": 0.3719, "step": 2320 }, { "epoch": 0.4, "learning_rate": 4.858234728876353e-05, "loss": 0.362, "step": 2321 }, { "epoch": 0.4, "learning_rate": 4.85799548813539e-05, "loss": 0.3704, "step": 2322 }, { "epoch": 0.4, "learning_rate": 4.857756052272904e-05, "loss": 0.3734, "step": 2323 }, { "epoch": 0.41, "learning_rate": 4.857516421311058e-05, "loss": 0.3822, "step": 2324 }, { "epoch": 0.41, "learning_rate": 4.857276595272032e-05, "loss": 0.3636, "step": 2325 }, { "epoch": 0.41, "learning_rate": 4.857036574178025e-05, "loss": 0.3645, "step": 2326 }, { "epoch": 0.41, "learning_rate": 4.856796358051255e-05, "loss": 0.3763, "step": 2327 }, { "epoch": 0.41, "learning_rate": 4.856555946913956e-05, "loss": 0.4094, "step": 2328 }, { "epoch": 0.41, "learning_rate": 4.856315340788382e-05, "loss": 0.3679, "step": 2329 }, { "epoch": 0.41, "learning_rate": 4.856074539696803e-05, "loss": 0.3775, "step": 2330 }, { "epoch": 0.41, "learning_rate": 4.855833543661508e-05, "loss": 0.3725, "step": 2331 }, { "epoch": 0.41, "learning_rate": 4.855592352704806e-05, "loss": 0.3769, "step": 2332 }, { "epoch": 0.41, "learning_rate": 4.855350966849021e-05, "loss": 0.3649, "step": 2333 }, { "epoch": 0.41, "learning_rate": 4.855109386116497e-05, "loss": 0.38, "step": 2334 }, { "epoch": 0.41, "learning_rate": 4.8548676105295946e-05, "loss": 0.3771, "step": 2335 }, { "epoch": 0.41, "learning_rate": 4.8546256401106936e-05, "loss": 0.3584, "step": 2336 }, { "epoch": 0.41, "learning_rate": 4.8543834748821904e-05, "loss": 0.3614, "step": 2337 }, { "epoch": 0.41, "learning_rate": 4.854141114866502e-05, "loss": 0.3699, "step": 2338 }, { "epoch": 0.41, "learning_rate": 4.853898560086061e-05, "loss": 0.3677, "step": 2339 }, { "epoch": 0.41, "learning_rate": 4.85365581056332e-05, "loss": 0.3756, "step": 2340 }, { "epoch": 0.41, "learning_rate": 4.853412866320746e-05, "loss": 0.395, "step": 2341 }, { "epoch": 0.41, "learning_rate": 4.853169727380829e-05, "loss": 0.3903, "step": 2342 }, { "epoch": 0.41, "learning_rate": 4.852926393766074e-05, "loss": 0.3857, "step": 2343 }, { "epoch": 0.41, "learning_rate": 4.8526828654990033e-05, "loss": 0.3721, "step": 2344 }, { "epoch": 0.41, "learning_rate": 4.8524391426021605e-05, "loss": 0.3785, "step": 2345 }, { "epoch": 0.41, "learning_rate": 4.852195225098104e-05, "loss": 0.3687, "step": 2346 }, { "epoch": 0.41, "learning_rate": 4.8519511130094115e-05, "loss": 0.379, "step": 2347 }, { "epoch": 0.41, "learning_rate": 4.8517068063586786e-05, "loss": 0.3777, "step": 2348 }, { "epoch": 0.41, "learning_rate": 4.851462305168519e-05, "loss": 0.3842, "step": 2349 }, { "epoch": 0.41, "learning_rate": 4.851217609461566e-05, "loss": 0.3702, "step": 2350 }, { "epoch": 0.41, "learning_rate": 4.850972719260466e-05, "loss": 0.3888, "step": 2351 }, { "epoch": 0.41, "learning_rate": 4.8507276345878894e-05, "loss": 0.3771, "step": 2352 }, { "epoch": 0.41, "learning_rate": 4.850482355466521e-05, "loss": 0.3631, "step": 2353 }, { "epoch": 0.41, "learning_rate": 4.850236881919064e-05, "loss": 0.3641, "step": 2354 }, { "epoch": 0.41, "learning_rate": 4.849991213968241e-05, "loss": 0.3627, "step": 2355 }, { "epoch": 0.41, "learning_rate": 4.8497453516367906e-05, "loss": 0.3798, "step": 2356 }, { "epoch": 0.41, "learning_rate": 4.849499294947471e-05, "loss": 0.3783, "step": 2357 }, { "epoch": 0.41, "learning_rate": 4.849253043923058e-05, "loss": 0.3615, "step": 2358 }, { "epoch": 0.41, "learning_rate": 4.849006598586345e-05, "loss": 0.3909, "step": 2359 }, { "epoch": 0.41, "learning_rate": 4.848759958960144e-05, "loss": 0.3597, "step": 2360 }, { "epoch": 0.41, "learning_rate": 4.848513125067285e-05, "loss": 0.3996, "step": 2361 }, { "epoch": 0.41, "learning_rate": 4.848266096930614e-05, "loss": 0.3915, "step": 2362 }, { "epoch": 0.41, "learning_rate": 4.848018874572997e-05, "loss": 0.3846, "step": 2363 }, { "epoch": 0.41, "learning_rate": 4.847771458017318e-05, "loss": 0.3908, "step": 2364 }, { "epoch": 0.41, "learning_rate": 4.84752384728648e-05, "loss": 0.3818, "step": 2365 }, { "epoch": 0.41, "learning_rate": 4.8472760424033996e-05, "loss": 0.3806, "step": 2366 }, { "epoch": 0.41, "learning_rate": 4.847028043391016e-05, "loss": 0.3635, "step": 2367 }, { "epoch": 0.41, "learning_rate": 4.846779850272284e-05, "loss": 0.3719, "step": 2368 }, { "epoch": 0.41, "learning_rate": 4.846531463070178e-05, "loss": 0.3692, "step": 2369 }, { "epoch": 0.41, "learning_rate": 4.8462828818076885e-05, "loss": 0.4045, "step": 2370 }, { "epoch": 0.41, "learning_rate": 4.8460341065078246e-05, "loss": 0.374, "step": 2371 }, { "epoch": 0.41, "learning_rate": 4.845785137193614e-05, "loss": 0.3712, "step": 2372 }, { "epoch": 0.41, "learning_rate": 4.845535973888102e-05, "loss": 0.3965, "step": 2373 }, { "epoch": 0.41, "learning_rate": 4.8452866166143526e-05, "loss": 0.3857, "step": 2374 }, { "epoch": 0.41, "learning_rate": 4.845037065395445e-05, "loss": 0.3937, "step": 2375 }, { "epoch": 0.41, "learning_rate": 4.844787320254479e-05, "loss": 0.368, "step": 2376 }, { "epoch": 0.41, "learning_rate": 4.844537381214573e-05, "loss": 0.381, "step": 2377 }, { "epoch": 0.41, "learning_rate": 4.84428724829886e-05, "loss": 0.3752, "step": 2378 }, { "epoch": 0.41, "learning_rate": 4.844036921530495e-05, "loss": 0.3654, "step": 2379 }, { "epoch": 0.41, "learning_rate": 4.843786400932648e-05, "loss": 0.3705, "step": 2380 }, { "epoch": 0.42, "learning_rate": 4.843535686528507e-05, "loss": 0.3598, "step": 2381 }, { "epoch": 0.42, "learning_rate": 4.84328477834128e-05, "loss": 0.3867, "step": 2382 }, { "epoch": 0.42, "learning_rate": 4.843033676394191e-05, "loss": 0.3736, "step": 2383 }, { "epoch": 0.42, "learning_rate": 4.8427823807104824e-05, "loss": 0.3663, "step": 2384 }, { "epoch": 0.42, "learning_rate": 4.842530891313416e-05, "loss": 0.3669, "step": 2385 }, { "epoch": 0.42, "learning_rate": 4.8422792082262686e-05, "loss": 0.369, "step": 2386 }, { "epoch": 0.42, "learning_rate": 4.8420273314723386e-05, "loss": 0.3926, "step": 2387 }, { "epoch": 0.42, "learning_rate": 4.8417752610749384e-05, "loss": 0.3691, "step": 2388 }, { "epoch": 0.42, "learning_rate": 4.841522997057402e-05, "loss": 0.3758, "step": 2389 }, { "epoch": 0.42, "learning_rate": 4.8412705394430784e-05, "loss": 0.3699, "step": 2390 }, { "epoch": 0.42, "learning_rate": 4.841017888255337e-05, "loss": 0.3707, "step": 2391 }, { "epoch": 0.42, "learning_rate": 4.8407650435175615e-05, "loss": 0.3801, "step": 2392 }, { "epoch": 0.42, "learning_rate": 4.8405120052531584e-05, "loss": 0.3795, "step": 2393 }, { "epoch": 0.42, "learning_rate": 4.840258773485549e-05, "loss": 0.3874, "step": 2394 }, { "epoch": 0.42, "learning_rate": 4.8400053482381715e-05, "loss": 0.3673, "step": 2395 }, { "epoch": 0.42, "learning_rate": 4.839751729534485e-05, "loss": 0.3547, "step": 2396 }, { "epoch": 0.42, "learning_rate": 4.8394979173979655e-05, "loss": 0.3841, "step": 2397 }, { "epoch": 0.42, "learning_rate": 4.839243911852105e-05, "loss": 0.3825, "step": 2398 }, { "epoch": 0.42, "learning_rate": 4.838989712920417e-05, "loss": 0.3665, "step": 2399 }, { "epoch": 0.42, "learning_rate": 4.838735320626429e-05, "loss": 0.3786, "step": 2400 }, { "epoch": 0.42, "learning_rate": 4.838480734993689e-05, "loss": 0.3842, "step": 2401 }, { "epoch": 0.42, "learning_rate": 4.838225956045761e-05, "loss": 0.3538, "step": 2402 }, { "epoch": 0.42, "learning_rate": 4.837970983806229e-05, "loss": 0.3945, "step": 2403 }, { "epoch": 0.42, "learning_rate": 4.837715818298695e-05, "loss": 0.3712, "step": 2404 }, { "epoch": 0.42, "learning_rate": 4.8374604595467754e-05, "loss": 0.373, "step": 2405 }, { "epoch": 0.42, "learning_rate": 4.837204907574109e-05, "loss": 0.3617, "step": 2406 }, { "epoch": 0.42, "learning_rate": 4.836949162404348e-05, "loss": 0.3731, "step": 2407 }, { "epoch": 0.42, "learning_rate": 4.8366932240611676e-05, "loss": 0.37, "step": 2408 }, { "epoch": 0.42, "learning_rate": 4.836437092568257e-05, "loss": 0.3583, "step": 2409 }, { "epoch": 0.42, "learning_rate": 4.8361807679493226e-05, "loss": 0.3846, "step": 2410 }, { "epoch": 0.42, "learning_rate": 4.8359242502280927e-05, "loss": 0.3758, "step": 2411 }, { "epoch": 0.42, "learning_rate": 4.8356675394283106e-05, "loss": 0.3687, "step": 2412 }, { "epoch": 0.42, "learning_rate": 4.8354106355737387e-05, "loss": 0.3834, "step": 2413 }, { "epoch": 0.42, "learning_rate": 4.835153538688155e-05, "loss": 0.3533, "step": 2414 }, { "epoch": 0.42, "learning_rate": 4.834896248795359e-05, "loss": 0.3752, "step": 2415 }, { "epoch": 0.42, "learning_rate": 4.834638765919165e-05, "loss": 0.3824, "step": 2416 }, { "epoch": 0.42, "learning_rate": 4.834381090083406e-05, "loss": 0.3881, "step": 2417 }, { "epoch": 0.42, "learning_rate": 4.8341232213119336e-05, "loss": 0.3809, "step": 2418 }, { "epoch": 0.42, "learning_rate": 4.8338651596286175e-05, "loss": 0.3666, "step": 2419 }, { "epoch": 0.42, "learning_rate": 4.833606905057343e-05, "loss": 0.3832, "step": 2420 }, { "epoch": 0.42, "learning_rate": 4.833348457622017e-05, "loss": 0.3665, "step": 2421 }, { "epoch": 0.42, "learning_rate": 4.83308981734656e-05, "loss": 0.3739, "step": 2422 }, { "epoch": 0.42, "learning_rate": 4.8328309842549135e-05, "loss": 0.3861, "step": 2423 }, { "epoch": 0.42, "learning_rate": 4.832571958371035e-05, "loss": 0.3652, "step": 2424 }, { "epoch": 0.42, "learning_rate": 4.8323127397189015e-05, "loss": 0.3748, "step": 2425 }, { "epoch": 0.42, "learning_rate": 4.8320533283225066e-05, "loss": 0.3761, "step": 2426 }, { "epoch": 0.42, "learning_rate": 4.831793724205861e-05, "loss": 0.3882, "step": 2427 }, { "epoch": 0.42, "learning_rate": 4.831533927392997e-05, "loss": 0.3573, "step": 2428 }, { "epoch": 0.42, "learning_rate": 4.8312739379079596e-05, "loss": 0.3799, "step": 2429 }, { "epoch": 0.42, "learning_rate": 4.831013755774814e-05, "loss": 0.3751, "step": 2430 }, { "epoch": 0.42, "learning_rate": 4.830753381017646e-05, "loss": 0.3986, "step": 2431 }, { "epoch": 0.42, "learning_rate": 4.830492813660553e-05, "loss": 0.3532, "step": 2432 }, { "epoch": 0.42, "learning_rate": 4.830232053727657e-05, "loss": 0.3802, "step": 2433 }, { "epoch": 0.42, "learning_rate": 4.829971101243093e-05, "loss": 0.3727, "step": 2434 }, { "epoch": 0.42, "learning_rate": 4.829709956231015e-05, "loss": 0.3791, "step": 2435 }, { "epoch": 0.42, "learning_rate": 4.8294486187155967e-05, "loss": 0.3653, "step": 2436 }, { "epoch": 0.42, "learning_rate": 4.829187088721026e-05, "loss": 0.3737, "step": 2437 }, { "epoch": 0.42, "learning_rate": 4.828925366271514e-05, "loss": 0.3676, "step": 2438 }, { "epoch": 0.43, "learning_rate": 4.828663451391284e-05, "loss": 0.3838, "step": 2439 }, { "epoch": 0.43, "learning_rate": 4.82840134410458e-05, "loss": 0.3813, "step": 2440 }, { "epoch": 0.43, "learning_rate": 4.828139044435663e-05, "loss": 0.3751, "step": 2441 }, { "epoch": 0.43, "learning_rate": 4.827876552408812e-05, "loss": 0.3679, "step": 2442 }, { "epoch": 0.43, "learning_rate": 4.827613868048326e-05, "loss": 0.3689, "step": 2443 }, { "epoch": 0.43, "learning_rate": 4.827350991378518e-05, "loss": 0.3604, "step": 2444 }, { "epoch": 0.43, "learning_rate": 4.8270879224237195e-05, "loss": 0.3766, "step": 2445 }, { "epoch": 0.43, "learning_rate": 4.826824661208283e-05, "loss": 0.3711, "step": 2446 }, { "epoch": 0.43, "learning_rate": 4.8265612077565755e-05, "loss": 0.3722, "step": 2447 }, { "epoch": 0.43, "learning_rate": 4.826297562092984e-05, "loss": 0.3794, "step": 2448 }, { "epoch": 0.43, "learning_rate": 4.8260337242419106e-05, "loss": 0.3743, "step": 2449 }, { "epoch": 0.43, "learning_rate": 4.825769694227778e-05, "loss": 0.3676, "step": 2450 }, { "epoch": 0.43, "learning_rate": 4.8255054720750254e-05, "loss": 0.3672, "step": 2451 }, { "epoch": 0.43, "learning_rate": 4.82524105780811e-05, "loss": 0.3759, "step": 2452 }, { "epoch": 0.43, "learning_rate": 4.8249764514515054e-05, "loss": 0.3704, "step": 2453 }, { "epoch": 0.43, "learning_rate": 4.824711653029706e-05, "loss": 0.3721, "step": 2454 }, { "epoch": 0.43, "learning_rate": 4.8244466625672205e-05, "loss": 0.3813, "step": 2455 }, { "epoch": 0.43, "learning_rate": 4.824181480088578e-05, "loss": 0.3645, "step": 2456 }, { "epoch": 0.43, "learning_rate": 4.823916105618326e-05, "loss": 0.3602, "step": 2457 }, { "epoch": 0.43, "learning_rate": 4.823650539181025e-05, "loss": 0.3892, "step": 2458 }, { "epoch": 0.43, "learning_rate": 4.8233847808012586e-05, "loss": 0.3566, "step": 2459 }, { "epoch": 0.43, "learning_rate": 4.823118830503626e-05, "loss": 0.3649, "step": 2460 }, { "epoch": 0.43, "learning_rate": 4.822852688312744e-05, "loss": 0.3827, "step": 2461 }, { "epoch": 0.43, "learning_rate": 4.822586354253247e-05, "loss": 0.3572, "step": 2462 }, { "epoch": 0.43, "learning_rate": 4.8223198283497886e-05, "loss": 0.3577, "step": 2463 }, { "epoch": 0.43, "learning_rate": 4.822053110627038e-05, "loss": 0.377, "step": 2464 }, { "epoch": 0.43, "learning_rate": 4.8217862011096844e-05, "loss": 0.364, "step": 2465 }, { "epoch": 0.43, "learning_rate": 4.821519099822432e-05, "loss": 0.3608, "step": 2466 }, { "epoch": 0.43, "learning_rate": 4.821251806790007e-05, "loss": 0.3617, "step": 2467 }, { "epoch": 0.43, "learning_rate": 4.8209843220371476e-05, "loss": 0.378, "step": 2468 }, { "epoch": 0.43, "learning_rate": 4.820716645588615e-05, "loss": 0.383, "step": 2469 }, { "epoch": 0.43, "learning_rate": 4.8204487774691855e-05, "loss": 0.3763, "step": 2470 }, { "epoch": 0.43, "learning_rate": 4.820180717703653e-05, "loss": 0.3714, "step": 2471 }, { "epoch": 0.43, "learning_rate": 4.819912466316831e-05, "loss": 0.3815, "step": 2472 }, { "epoch": 0.43, "learning_rate": 4.819644023333549e-05, "loss": 0.3791, "step": 2473 }, { "epoch": 0.43, "learning_rate": 4.819375388778654e-05, "loss": 0.3744, "step": 2474 }, { "epoch": 0.43, "learning_rate": 4.819106562677013e-05, "loss": 0.3651, "step": 2475 }, { "epoch": 0.43, "learning_rate": 4.818837545053508e-05, "loss": 0.3914, "step": 2476 }, { "epoch": 0.43, "learning_rate": 4.8185683359330406e-05, "loss": 0.3709, "step": 2477 }, { "epoch": 0.43, "learning_rate": 4.818298935340529e-05, "loss": 0.351, "step": 2478 }, { "epoch": 0.43, "learning_rate": 4.818029343300911e-05, "loss": 0.3592, "step": 2479 }, { "epoch": 0.43, "learning_rate": 4.817759559839139e-05, "loss": 0.3861, "step": 2480 }, { "epoch": 0.43, "learning_rate": 4.8174895849801855e-05, "loss": 0.3689, "step": 2481 }, { "epoch": 0.43, "learning_rate": 4.817219418749041e-05, "loss": 0.3628, "step": 2482 }, { "epoch": 0.43, "learning_rate": 4.816949061170711e-05, "loss": 0.386, "step": 2483 }, { "epoch": 0.43, "learning_rate": 4.816678512270221e-05, "loss": 0.3802, "step": 2484 }, { "epoch": 0.43, "learning_rate": 4.8164077720726144e-05, "loss": 0.3831, "step": 2485 }, { "epoch": 0.43, "learning_rate": 4.8161368406029514e-05, "loss": 0.3741, "step": 2486 }, { "epoch": 0.43, "learning_rate": 4.8158657178863115e-05, "loss": 0.3752, "step": 2487 }, { "epoch": 0.43, "learning_rate": 4.815594403947787e-05, "loss": 0.3705, "step": 2488 }, { "epoch": 0.43, "learning_rate": 4.815322898812494e-05, "loss": 0.3726, "step": 2489 }, { "epoch": 0.43, "learning_rate": 4.8150512025055647e-05, "loss": 0.3998, "step": 2490 }, { "epoch": 0.43, "learning_rate": 4.814779315052145e-05, "loss": 0.3724, "step": 2491 }, { "epoch": 0.43, "learning_rate": 4.814507236477404e-05, "loss": 0.362, "step": 2492 }, { "epoch": 0.43, "learning_rate": 4.8142349668065235e-05, "loss": 0.3737, "step": 2493 }, { "epoch": 0.43, "learning_rate": 4.813962506064708e-05, "loss": 0.3803, "step": 2494 }, { "epoch": 0.43, "learning_rate": 4.813689854277176e-05, "loss": 0.385, "step": 2495 }, { "epoch": 0.44, "learning_rate": 4.813417011469165e-05, "loss": 0.3677, "step": 2496 }, { "epoch": 0.44, "learning_rate": 4.813143977665931e-05, "loss": 0.3829, "step": 2497 }, { "epoch": 0.44, "learning_rate": 4.8128707528927445e-05, "loss": 0.3692, "step": 2498 }, { "epoch": 0.44, "learning_rate": 4.812597337174898e-05, "loss": 0.3692, "step": 2499 }, { "epoch": 0.44, "learning_rate": 4.8123237305376975e-05, "loss": 0.3672, "step": 2500 }, { "epoch": 0.44, "learning_rate": 4.812049933006471e-05, "loss": 0.3817, "step": 2501 }, { "epoch": 0.44, "learning_rate": 4.811775944606561e-05, "loss": 0.3517, "step": 2502 }, { "epoch": 0.44, "learning_rate": 4.8115017653633285e-05, "loss": 0.391, "step": 2503 }, { "epoch": 0.44, "learning_rate": 4.811227395302152e-05, "loss": 0.3525, "step": 2504 }, { "epoch": 0.44, "learning_rate": 4.810952834448429e-05, "loss": 0.3727, "step": 2505 }, { "epoch": 0.44, "learning_rate": 4.810678082827571e-05, "loss": 0.379, "step": 2506 }, { "epoch": 0.44, "learning_rate": 4.8104031404650126e-05, "loss": 0.384, "step": 2507 }, { "epoch": 0.44, "learning_rate": 4.810128007386202e-05, "loss": 0.3683, "step": 2508 }, { "epoch": 0.44, "learning_rate": 4.8098526836166055e-05, "loss": 0.3644, "step": 2509 }, { "epoch": 0.44, "learning_rate": 4.809577169181709e-05, "loss": 0.3684, "step": 2510 }, { "epoch": 0.44, "learning_rate": 4.809301464107014e-05, "loss": 0.3838, "step": 2511 }, { "epoch": 0.44, "learning_rate": 4.809025568418041e-05, "loss": 0.3631, "step": 2512 }, { "epoch": 0.44, "learning_rate": 4.8087494821403275e-05, "loss": 0.3673, "step": 2513 }, { "epoch": 0.44, "learning_rate": 4.808473205299428e-05, "loss": 0.3461, "step": 2514 }, { "epoch": 0.44, "learning_rate": 4.808196737920917e-05, "loss": 0.3777, "step": 2515 }, { "epoch": 0.44, "learning_rate": 4.807920080030384e-05, "loss": 0.3789, "step": 2516 }, { "epoch": 0.44, "learning_rate": 4.807643231653437e-05, "loss": 0.3547, "step": 2517 }, { "epoch": 0.44, "learning_rate": 4.8073661928157027e-05, "loss": 0.3645, "step": 2518 }, { "epoch": 0.44, "learning_rate": 4.807088963542824e-05, "loss": 0.3894, "step": 2519 }, { "epoch": 0.44, "learning_rate": 4.80681154386046e-05, "loss": 0.3887, "step": 2520 }, { "epoch": 0.44, "learning_rate": 4.806533933794293e-05, "loss": 0.3643, "step": 2521 }, { "epoch": 0.44, "learning_rate": 4.8062561333700166e-05, "loss": 0.3779, "step": 2522 }, { "epoch": 0.44, "learning_rate": 4.8059781426133465e-05, "loss": 0.3854, "step": 2523 }, { "epoch": 0.44, "learning_rate": 4.805699961550013e-05, "loss": 0.3713, "step": 2524 }, { "epoch": 0.44, "learning_rate": 4.8054215902057664e-05, "loss": 0.382, "step": 2525 }, { "epoch": 0.44, "learning_rate": 4.805143028606373e-05, "loss": 0.3866, "step": 2526 }, { "epoch": 0.44, "learning_rate": 4.804864276777615e-05, "loss": 0.3801, "step": 2527 }, { "epoch": 0.44, "learning_rate": 4.804585334745298e-05, "loss": 0.3756, "step": 2528 }, { "epoch": 0.44, "learning_rate": 4.80430620253524e-05, "loss": 0.3536, "step": 2529 }, { "epoch": 0.44, "learning_rate": 4.804026880173277e-05, "loss": 0.3634, "step": 2530 }, { "epoch": 0.44, "learning_rate": 4.803747367685266e-05, "loss": 0.3778, "step": 2531 }, { "epoch": 0.44, "learning_rate": 4.8034676650970784e-05, "loss": 0.3633, "step": 2532 }, { "epoch": 0.44, "learning_rate": 4.803187772434604e-05, "loss": 0.3686, "step": 2533 }, { "epoch": 0.44, "learning_rate": 4.8029076897237496e-05, "loss": 0.3957, "step": 2534 }, { "epoch": 0.44, "learning_rate": 4.802627416990442e-05, "loss": 0.3765, "step": 2535 }, { "epoch": 0.44, "learning_rate": 4.802346954260624e-05, "loss": 0.3719, "step": 2536 }, { "epoch": 0.44, "learning_rate": 4.802066301560254e-05, "loss": 0.3624, "step": 2537 }, { "epoch": 0.44, "learning_rate": 4.801785458915312e-05, "loss": 0.3765, "step": 2538 }, { "epoch": 0.44, "learning_rate": 4.801504426351793e-05, "loss": 0.3751, "step": 2539 }, { "epoch": 0.44, "learning_rate": 4.801223203895709e-05, "loss": 0.3656, "step": 2540 }, { "epoch": 0.44, "learning_rate": 4.800941791573092e-05, "loss": 0.3786, "step": 2541 }, { "epoch": 0.44, "learning_rate": 4.8006601894099896e-05, "loss": 0.3785, "step": 2542 }, { "epoch": 0.44, "learning_rate": 4.800378397432468e-05, "loss": 0.3552, "step": 2543 }, { "epoch": 0.44, "learning_rate": 4.8000964156666094e-05, "loss": 0.3932, "step": 2544 }, { "epoch": 0.44, "learning_rate": 4.799814244138516e-05, "loss": 0.3578, "step": 2545 }, { "epoch": 0.44, "learning_rate": 4.7995318828743074e-05, "loss": 0.3688, "step": 2546 }, { "epoch": 0.44, "learning_rate": 4.799249331900117e-05, "loss": 0.3507, "step": 2547 }, { "epoch": 0.44, "learning_rate": 4.7989665912421e-05, "loss": 0.3679, "step": 2548 }, { "epoch": 0.44, "learning_rate": 4.7986836609264285e-05, "loss": 0.3825, "step": 2549 }, { "epoch": 0.44, "learning_rate": 4.798400540979289e-05, "loss": 0.3851, "step": 2550 }, { "epoch": 0.44, "learning_rate": 4.798117231426889e-05, "loss": 0.3635, "step": 2551 }, { "epoch": 0.44, "learning_rate": 4.7978337322954536e-05, "loss": 0.3638, "step": 2552 }, { "epoch": 0.44, "learning_rate": 4.7975500436112214e-05, "loss": 0.3893, "step": 2553 }, { "epoch": 0.45, "learning_rate": 4.7972661654004536e-05, "loss": 0.377, "step": 2554 }, { "epoch": 0.45, "learning_rate": 4.796982097689426e-05, "loss": 0.3741, "step": 2555 }, { "epoch": 0.45, "learning_rate": 4.796697840504433e-05, "loss": 0.3681, "step": 2556 }, { "epoch": 0.45, "learning_rate": 4.796413393871785e-05, "loss": 0.3551, "step": 2557 }, { "epoch": 0.45, "learning_rate": 4.7961287578178126e-05, "loss": 0.3738, "step": 2558 }, { "epoch": 0.45, "learning_rate": 4.795843932368862e-05, "loss": 0.3694, "step": 2559 }, { "epoch": 0.45, "learning_rate": 4.7955589175512974e-05, "loss": 0.3732, "step": 2560 }, { "epoch": 0.45, "learning_rate": 4.7952737133914995e-05, "loss": 0.3628, "step": 2561 }, { "epoch": 0.45, "learning_rate": 4.794988319915868e-05, "loss": 0.3493, "step": 2562 }, { "epoch": 0.45, "learning_rate": 4.794702737150821e-05, "loss": 0.365, "step": 2563 }, { "epoch": 0.45, "learning_rate": 4.7944169651227915e-05, "loss": 0.3837, "step": 2564 }, { "epoch": 0.45, "learning_rate": 4.794131003858232e-05, "loss": 0.3782, "step": 2565 }, { "epoch": 0.45, "learning_rate": 4.7938448533836104e-05, "loss": 0.375, "step": 2566 }, { "epoch": 0.45, "learning_rate": 4.7935585137254155e-05, "loss": 0.3604, "step": 2567 }, { "epoch": 0.45, "learning_rate": 4.79327198491015e-05, "loss": 0.3887, "step": 2568 }, { "epoch": 0.45, "learning_rate": 4.792985266964337e-05, "loss": 0.3556, "step": 2569 }, { "epoch": 0.45, "learning_rate": 4.792698359914515e-05, "loss": 0.3623, "step": 2570 }, { "epoch": 0.45, "learning_rate": 4.792411263787241e-05, "loss": 0.38, "step": 2571 }, { "epoch": 0.45, "learning_rate": 4.7921239786090896e-05, "loss": 0.3664, "step": 2572 }, { "epoch": 0.45, "learning_rate": 4.7918365044066526e-05, "loss": 0.3775, "step": 2573 }, { "epoch": 0.45, "learning_rate": 4.7915488412065394e-05, "loss": 0.3614, "step": 2574 }, { "epoch": 0.45, "learning_rate": 4.791260989035376e-05, "loss": 0.3605, "step": 2575 }, { "epoch": 0.45, "learning_rate": 4.790972947919809e-05, "loss": 0.3657, "step": 2576 }, { "epoch": 0.45, "learning_rate": 4.7906847178864975e-05, "loss": 0.3785, "step": 2577 }, { "epoch": 0.45, "learning_rate": 4.7903962989621224e-05, "loss": 0.3767, "step": 2578 }, { "epoch": 0.45, "learning_rate": 4.79010769117338e-05, "loss": 0.3818, "step": 2579 }, { "epoch": 0.45, "learning_rate": 4.789818894546986e-05, "loss": 0.3614, "step": 2580 }, { "epoch": 0.45, "learning_rate": 4.78952990910967e-05, "loss": 0.3588, "step": 2581 }, { "epoch": 0.45, "learning_rate": 4.789240734888182e-05, "loss": 0.3737, "step": 2582 }, { "epoch": 0.45, "learning_rate": 4.788951371909289e-05, "loss": 0.3701, "step": 2583 }, { "epoch": 0.45, "learning_rate": 4.7886618201997747e-05, "loss": 0.3655, "step": 2584 }, { "epoch": 0.45, "learning_rate": 4.7883720797864404e-05, "loss": 0.354, "step": 2585 }, { "epoch": 0.45, "learning_rate": 4.788082150696108e-05, "loss": 0.3718, "step": 2586 }, { "epoch": 0.45, "learning_rate": 4.787792032955611e-05, "loss": 0.3683, "step": 2587 }, { "epoch": 0.45, "learning_rate": 4.7875017265918036e-05, "loss": 0.376, "step": 2588 }, { "epoch": 0.45, "learning_rate": 4.7872112316315596e-05, "loss": 0.3725, "step": 2589 }, { "epoch": 0.45, "learning_rate": 4.786920548101766e-05, "loss": 0.3648, "step": 2590 }, { "epoch": 0.45, "learning_rate": 4.786629676029329e-05, "loss": 0.3661, "step": 2591 }, { "epoch": 0.45, "learning_rate": 4.786338615441174e-05, "loss": 0.3527, "step": 2592 }, { "epoch": 0.45, "learning_rate": 4.786047366364242e-05, "loss": 0.3689, "step": 2593 }, { "epoch": 0.45, "learning_rate": 4.785755928825491e-05, "loss": 0.355, "step": 2594 }, { "epoch": 0.45, "learning_rate": 4.7854643028518974e-05, "loss": 0.3642, "step": 2595 }, { "epoch": 0.45, "learning_rate": 4.785172488470455e-05, "loss": 0.3813, "step": 2596 }, { "epoch": 0.45, "learning_rate": 4.7848804857081746e-05, "loss": 0.3575, "step": 2597 }, { "epoch": 0.45, "learning_rate": 4.784588294592085e-05, "loss": 0.3464, "step": 2598 }, { "epoch": 0.45, "learning_rate": 4.784295915149233e-05, "loss": 0.3563, "step": 2599 }, { "epoch": 0.45, "learning_rate": 4.78400334740668e-05, "loss": 0.3889, "step": 2600 }, { "epoch": 0.45, "learning_rate": 4.7837105913915086e-05, "loss": 0.3664, "step": 2601 }, { "epoch": 0.45, "learning_rate": 4.783417647130817e-05, "loss": 0.3635, "step": 2602 }, { "epoch": 0.45, "learning_rate": 4.78312451465172e-05, "loss": 0.3493, "step": 2603 }, { "epoch": 0.45, "learning_rate": 4.7828311939813515e-05, "loss": 0.3836, "step": 2604 }, { "epoch": 0.45, "learning_rate": 4.782537685146861e-05, "loss": 0.3447, "step": 2605 }, { "epoch": 0.45, "learning_rate": 4.782243988175417e-05, "loss": 0.3671, "step": 2606 }, { "epoch": 0.45, "learning_rate": 4.781950103094205e-05, "loss": 0.3692, "step": 2607 }, { "epoch": 0.45, "learning_rate": 4.781656029930427e-05, "loss": 0.3602, "step": 2608 }, { "epoch": 0.45, "learning_rate": 4.7813617687113044e-05, "loss": 0.3535, "step": 2609 }, { "epoch": 0.45, "learning_rate": 4.781067319464075e-05, "loss": 0.3788, "step": 2610 }, { "epoch": 0.46, "learning_rate": 4.780772682215992e-05, "loss": 0.3645, "step": 2611 }, { "epoch": 0.46, "learning_rate": 4.780477856994329e-05, "loss": 0.3649, "step": 2612 }, { "epoch": 0.46, "learning_rate": 4.7801828438263755e-05, "loss": 0.3889, "step": 2613 }, { "epoch": 0.46, "learning_rate": 4.779887642739438e-05, "loss": 0.3708, "step": 2614 }, { "epoch": 0.46, "learning_rate": 4.779592253760843e-05, "loss": 0.3741, "step": 2615 }, { "epoch": 0.46, "learning_rate": 4.779296676917931e-05, "loss": 0.3952, "step": 2616 }, { "epoch": 0.46, "learning_rate": 4.7790009122380614e-05, "loss": 0.3914, "step": 2617 }, { "epoch": 0.46, "learning_rate": 4.778704959748611e-05, "loss": 0.3843, "step": 2618 }, { "epoch": 0.46, "learning_rate": 4.778408819476974e-05, "loss": 0.3611, "step": 2619 }, { "epoch": 0.46, "learning_rate": 4.778112491450562e-05, "loss": 0.3645, "step": 2620 }, { "epoch": 0.46, "learning_rate": 4.7778159756968046e-05, "loss": 0.3921, "step": 2621 }, { "epoch": 0.46, "learning_rate": 4.7775192722431474e-05, "loss": 0.3396, "step": 2622 }, { "epoch": 0.46, "learning_rate": 4.777222381117053e-05, "loss": 0.355, "step": 2623 }, { "epoch": 0.46, "learning_rate": 4.776925302346004e-05, "loss": 0.3624, "step": 2624 }, { "epoch": 0.46, "learning_rate": 4.776628035957498e-05, "loss": 0.3615, "step": 2625 }, { "epoch": 0.46, "learning_rate": 4.7763305819790515e-05, "loss": 0.3613, "step": 2626 }, { "epoch": 0.46, "learning_rate": 4.776032940438197e-05, "loss": 0.3763, "step": 2627 }, { "epoch": 0.46, "learning_rate": 4.7757351113624846e-05, "loss": 0.3697, "step": 2628 }, { "epoch": 0.46, "learning_rate": 4.775437094779483e-05, "loss": 0.3748, "step": 2629 }, { "epoch": 0.46, "learning_rate": 4.7751388907167765e-05, "loss": 0.3671, "step": 2630 }, { "epoch": 0.46, "learning_rate": 4.774840499201969e-05, "loss": 0.3821, "step": 2631 }, { "epoch": 0.46, "learning_rate": 4.774541920262679e-05, "loss": 0.3671, "step": 2632 }, { "epoch": 0.46, "learning_rate": 4.7742431539265434e-05, "loss": 0.3622, "step": 2633 }, { "epoch": 0.46, "learning_rate": 4.773944200221218e-05, "loss": 0.3916, "step": 2634 }, { "epoch": 0.46, "learning_rate": 4.7736450591743745e-05, "loss": 0.3651, "step": 2635 }, { "epoch": 0.46, "learning_rate": 4.773345730813702e-05, "loss": 0.3903, "step": 2636 }, { "epoch": 0.46, "learning_rate": 4.773046215166908e-05, "loss": 0.3612, "step": 2637 }, { "epoch": 0.46, "learning_rate": 4.7727465122617144e-05, "loss": 0.3615, "step": 2638 }, { "epoch": 0.46, "learning_rate": 4.772446622125864e-05, "loss": 0.3766, "step": 2639 }, { "epoch": 0.46, "learning_rate": 4.7721465447871156e-05, "loss": 0.3632, "step": 2640 }, { "epoch": 0.46, "learning_rate": 4.7718462802732436e-05, "loss": 0.3462, "step": 2641 }, { "epoch": 0.46, "learning_rate": 4.771545828612044e-05, "loss": 0.3745, "step": 2642 }, { "epoch": 0.46, "learning_rate": 4.771245189831323e-05, "loss": 0.3703, "step": 2643 }, { "epoch": 0.46, "learning_rate": 4.770944363958912e-05, "loss": 0.3695, "step": 2644 }, { "epoch": 0.46, "learning_rate": 4.7706433510226564e-05, "loss": 0.3711, "step": 2645 }, { "epoch": 0.46, "learning_rate": 4.770342151050417e-05, "loss": 0.3574, "step": 2646 }, { "epoch": 0.46, "learning_rate": 4.770040764070074e-05, "loss": 0.3692, "step": 2647 }, { "epoch": 0.46, "learning_rate": 4.769739190109526e-05, "loss": 0.3524, "step": 2648 }, { "epoch": 0.46, "learning_rate": 4.769437429196687e-05, "loss": 0.3688, "step": 2649 }, { "epoch": 0.46, "learning_rate": 4.7691354813594854e-05, "loss": 0.3932, "step": 2650 }, { "epoch": 0.46, "learning_rate": 4.768833346625875e-05, "loss": 0.3771, "step": 2651 }, { "epoch": 0.46, "learning_rate": 4.7685310250238194e-05, "loss": 0.3705, "step": 2652 }, { "epoch": 0.46, "learning_rate": 4.768228516581304e-05, "loss": 0.3566, "step": 2653 }, { "epoch": 0.46, "learning_rate": 4.767925821326328e-05, "loss": 0.3835, "step": 2654 }, { "epoch": 0.46, "learning_rate": 4.76762293928691e-05, "loss": 0.3632, "step": 2655 }, { "epoch": 0.46, "learning_rate": 4.767319870491087e-05, "loss": 0.3611, "step": 2656 }, { "epoch": 0.46, "learning_rate": 4.767016614966911e-05, "loss": 0.3629, "step": 2657 }, { "epoch": 0.46, "learning_rate": 4.766713172742451e-05, "loss": 0.3746, "step": 2658 }, { "epoch": 0.46, "learning_rate": 4.766409543845796e-05, "loss": 0.3489, "step": 2659 }, { "epoch": 0.46, "learning_rate": 4.7661057283050505e-05, "loss": 0.3846, "step": 2660 }, { "epoch": 0.46, "learning_rate": 4.765801726148336e-05, "loss": 0.362, "step": 2661 }, { "epoch": 0.46, "learning_rate": 4.7654975374037907e-05, "loss": 0.3519, "step": 2662 }, { "epoch": 0.46, "learning_rate": 4.765193162099573e-05, "loss": 0.3664, "step": 2663 }, { "epoch": 0.46, "learning_rate": 4.7648886002638554e-05, "loss": 0.3561, "step": 2664 }, { "epoch": 0.46, "learning_rate": 4.7645838519248294e-05, "loss": 0.3641, "step": 2665 }, { "epoch": 0.46, "learning_rate": 4.7642789171107026e-05, "loss": 0.3783, "step": 2666 }, { "epoch": 0.46, "learning_rate": 4.763973795849702e-05, "loss": 0.3693, "step": 2667 }, { "epoch": 0.47, "learning_rate": 4.763668488170069e-05, "loss": 0.377, "step": 2668 }, { "epoch": 0.47, "learning_rate": 4.763362994100066e-05, "loss": 0.3604, "step": 2669 }, { "epoch": 0.47, "learning_rate": 4.7630573136679666e-05, "loss": 0.3858, "step": 2670 }, { "epoch": 0.47, "learning_rate": 4.762751446902068e-05, "loss": 0.3631, "step": 2671 }, { "epoch": 0.47, "learning_rate": 4.7624453938306815e-05, "loss": 0.3627, "step": 2672 }, { "epoch": 0.47, "learning_rate": 4.7621391544821356e-05, "loss": 0.3687, "step": 2673 }, { "epoch": 0.47, "learning_rate": 4.761832728884778e-05, "loss": 0.3739, "step": 2674 }, { "epoch": 0.47, "learning_rate": 4.76152611706697e-05, "loss": 0.3708, "step": 2675 }, { "epoch": 0.47, "learning_rate": 4.761219319057095e-05, "loss": 0.3596, "step": 2676 }, { "epoch": 0.47, "learning_rate": 4.760912334883549e-05, "loss": 0.3692, "step": 2677 }, { "epoch": 0.47, "learning_rate": 4.7606051645747484e-05, "loss": 0.3685, "step": 2678 }, { "epoch": 0.47, "learning_rate": 4.7602978081591246e-05, "loss": 0.3675, "step": 2679 }, { "epoch": 0.47, "learning_rate": 4.7599902656651285e-05, "loss": 0.3815, "step": 2680 }, { "epoch": 0.47, "learning_rate": 4.759682537121227e-05, "loss": 0.3717, "step": 2681 }, { "epoch": 0.47, "learning_rate": 4.759374622555903e-05, "loss": 0.3819, "step": 2682 }, { "epoch": 0.47, "learning_rate": 4.759066521997658e-05, "loss": 0.3655, "step": 2683 }, { "epoch": 0.47, "learning_rate": 4.758758235475011e-05, "loss": 0.3814, "step": 2684 }, { "epoch": 0.47, "learning_rate": 4.7584497630165e-05, "loss": 0.3633, "step": 2685 }, { "epoch": 0.47, "learning_rate": 4.7581411046506745e-05, "loss": 0.3745, "step": 2686 }, { "epoch": 0.47, "learning_rate": 4.757832260406106e-05, "loss": 0.362, "step": 2687 }, { "epoch": 0.47, "learning_rate": 4.757523230311383e-05, "loss": 0.3672, "step": 2688 }, { "epoch": 0.47, "learning_rate": 4.757214014395109e-05, "loss": 0.3743, "step": 2689 }, { "epoch": 0.47, "learning_rate": 4.756904612685906e-05, "loss": 0.3674, "step": 2690 }, { "epoch": 0.47, "learning_rate": 4.756595025212412e-05, "loss": 0.3596, "step": 2691 }, { "epoch": 0.47, "learning_rate": 4.7562852520032856e-05, "loss": 0.3776, "step": 2692 }, { "epoch": 0.47, "learning_rate": 4.7559752930871984e-05, "loss": 0.3593, "step": 2693 }, { "epoch": 0.47, "learning_rate": 4.755665148492842e-05, "loss": 0.3691, "step": 2694 }, { "epoch": 0.47, "learning_rate": 4.755354818248924e-05, "loss": 0.388, "step": 2695 }, { "epoch": 0.47, "learning_rate": 4.7550443023841676e-05, "loss": 0.3663, "step": 2696 }, { "epoch": 0.47, "learning_rate": 4.7547336009273184e-05, "loss": 0.3564, "step": 2697 }, { "epoch": 0.47, "learning_rate": 4.754422713907132e-05, "loss": 0.356, "step": 2698 }, { "epoch": 0.47, "learning_rate": 4.754111641352388e-05, "loss": 0.3759, "step": 2699 }, { "epoch": 0.47, "learning_rate": 4.753800383291878e-05, "loss": 0.3656, "step": 2700 }, { "epoch": 0.47, "learning_rate": 4.753488939754415e-05, "loss": 0.3579, "step": 2701 }, { "epoch": 0.47, "learning_rate": 4.753177310768824e-05, "loss": 0.364, "step": 2702 }, { "epoch": 0.47, "learning_rate": 4.7528654963639527e-05, "loss": 0.3791, "step": 2703 }, { "epoch": 0.47, "learning_rate": 4.7525534965686614e-05, "loss": 0.3975, "step": 2704 }, { "epoch": 0.47, "learning_rate": 4.7522413114118316e-05, "loss": 0.3739, "step": 2705 }, { "epoch": 0.47, "learning_rate": 4.751928940922359e-05, "loss": 0.379, "step": 2706 }, { "epoch": 0.47, "learning_rate": 4.751616385129158e-05, "loss": 0.3474, "step": 2707 }, { "epoch": 0.47, "learning_rate": 4.75130364406116e-05, "loss": 0.3748, "step": 2708 }, { "epoch": 0.47, "learning_rate": 4.7509907177473104e-05, "loss": 0.3784, "step": 2709 }, { "epoch": 0.47, "learning_rate": 4.750677606216577e-05, "loss": 0.3829, "step": 2710 }, { "epoch": 0.47, "learning_rate": 4.7503643094979425e-05, "loss": 0.3592, "step": 2711 }, { "epoch": 0.47, "learning_rate": 4.7500508276204046e-05, "loss": 0.3847, "step": 2712 }, { "epoch": 0.47, "learning_rate": 4.749737160612981e-05, "loss": 0.3676, "step": 2713 }, { "epoch": 0.47, "learning_rate": 4.7494233085047055e-05, "loss": 0.357, "step": 2714 }, { "epoch": 0.47, "learning_rate": 4.7491092713246283e-05, "loss": 0.364, "step": 2715 }, { "epoch": 0.47, "learning_rate": 4.748795049101819e-05, "loss": 0.3611, "step": 2716 }, { "epoch": 0.47, "learning_rate": 4.748480641865362e-05, "loss": 0.362, "step": 2717 }, { "epoch": 0.47, "learning_rate": 4.74816604964436e-05, "loss": 0.3695, "step": 2718 }, { "epoch": 0.47, "learning_rate": 4.747851272467931e-05, "loss": 0.3714, "step": 2719 }, { "epoch": 0.47, "learning_rate": 4.7475363103652135e-05, "loss": 0.3615, "step": 2720 }, { "epoch": 0.47, "learning_rate": 4.7472211633653604e-05, "loss": 0.3822, "step": 2721 }, { "epoch": 0.47, "learning_rate": 4.746905831497543e-05, "loss": 0.368, "step": 2722 }, { "epoch": 0.47, "learning_rate": 4.746590314790948e-05, "loss": 0.3515, "step": 2723 }, { "epoch": 0.47, "learning_rate": 4.746274613274782e-05, "loss": 0.3538, "step": 2724 }, { "epoch": 0.47, "learning_rate": 4.745958726978266e-05, "loss": 0.373, "step": 2725 }, { "epoch": 0.48, "learning_rate": 4.7456426559306385e-05, "loss": 0.3565, "step": 2726 }, { "epoch": 0.48, "learning_rate": 4.745326400161159e-05, "loss": 0.367, "step": 2727 }, { "epoch": 0.48, "learning_rate": 4.7450099596990975e-05, "loss": 0.3581, "step": 2728 }, { "epoch": 0.48, "learning_rate": 4.744693334573747e-05, "loss": 0.3649, "step": 2729 }, { "epoch": 0.48, "learning_rate": 4.744376524814413e-05, "loss": 0.377, "step": 2730 }, { "epoch": 0.48, "learning_rate": 4.744059530450422e-05, "loss": 0.3609, "step": 2731 }, { "epoch": 0.48, "learning_rate": 4.743742351511115e-05, "loss": 0.3634, "step": 2732 }, { "epoch": 0.48, "learning_rate": 4.743424988025852e-05, "loss": 0.3671, "step": 2733 }, { "epoch": 0.48, "learning_rate": 4.743107440024007e-05, "loss": 0.3782, "step": 2734 }, { "epoch": 0.48, "learning_rate": 4.7427897075349746e-05, "loss": 0.3626, "step": 2735 }, { "epoch": 0.48, "learning_rate": 4.7424717905881645e-05, "loss": 0.3825, "step": 2736 }, { "epoch": 0.48, "learning_rate": 4.742153689213004e-05, "loss": 0.3509, "step": 2737 }, { "epoch": 0.48, "learning_rate": 4.7418354034389376e-05, "loss": 0.3685, "step": 2738 }, { "epoch": 0.48, "learning_rate": 4.7415169332954254e-05, "loss": 0.3622, "step": 2739 }, { "epoch": 0.48, "learning_rate": 4.741198278811947e-05, "loss": 0.3725, "step": 2740 }, { "epoch": 0.48, "learning_rate": 4.740879440017999e-05, "loss": 0.3664, "step": 2741 }, { "epoch": 0.48, "learning_rate": 4.740560416943091e-05, "loss": 0.3701, "step": 2742 }, { "epoch": 0.48, "learning_rate": 4.7402412096167545e-05, "loss": 0.38, "step": 2743 }, { "epoch": 0.48, "learning_rate": 4.739921818068536e-05, "loss": 0.3684, "step": 2744 }, { "epoch": 0.48, "learning_rate": 4.739602242327999e-05, "loss": 0.3556, "step": 2745 }, { "epoch": 0.48, "learning_rate": 4.739282482424724e-05, "loss": 0.37, "step": 2746 }, { "epoch": 0.48, "learning_rate": 4.7389625383883104e-05, "loss": 0.3925, "step": 2747 }, { "epoch": 0.48, "learning_rate": 4.738642410248372e-05, "loss": 0.3663, "step": 2748 }, { "epoch": 0.48, "learning_rate": 4.738322098034539e-05, "loss": 0.3747, "step": 2749 }, { "epoch": 0.48, "learning_rate": 4.738001601776462e-05, "loss": 0.3739, "step": 2750 }, { "epoch": 0.48, "learning_rate": 4.737680921503807e-05, "loss": 0.3755, "step": 2751 }, { "epoch": 0.48, "learning_rate": 4.7373600572462565e-05, "loss": 0.3675, "step": 2752 }, { "epoch": 0.48, "learning_rate": 4.737039009033512e-05, "loss": 0.3617, "step": 2753 }, { "epoch": 0.48, "learning_rate": 4.7367177768952874e-05, "loss": 0.3653, "step": 2754 }, { "epoch": 0.48, "learning_rate": 4.7363963608613205e-05, "loss": 0.3741, "step": 2755 }, { "epoch": 0.48, "learning_rate": 4.73607476096136e-05, "loss": 0.3675, "step": 2756 }, { "epoch": 0.48, "learning_rate": 4.7357529772251743e-05, "loss": 0.3765, "step": 2757 }, { "epoch": 0.48, "learning_rate": 4.735431009682549e-05, "loss": 0.3656, "step": 2758 }, { "epoch": 0.48, "learning_rate": 4.735108858363286e-05, "loss": 0.3648, "step": 2759 }, { "epoch": 0.48, "learning_rate": 4.734786523297205e-05, "loss": 0.3716, "step": 2760 }, { "epoch": 0.48, "learning_rate": 4.734464004514141e-05, "loss": 0.3676, "step": 2761 }, { "epoch": 0.48, "learning_rate": 4.734141302043949e-05, "loss": 0.3715, "step": 2762 }, { "epoch": 0.48, "learning_rate": 4.7338184159164975e-05, "loss": 0.3533, "step": 2763 }, { "epoch": 0.48, "learning_rate": 4.733495346161675e-05, "loss": 0.3633, "step": 2764 }, { "epoch": 0.48, "learning_rate": 4.7331720928093845e-05, "loss": 0.3631, "step": 2765 }, { "epoch": 0.48, "learning_rate": 4.7328486558895474e-05, "loss": 0.354, "step": 2766 }, { "epoch": 0.48, "learning_rate": 4.732525035432101e-05, "loss": 0.3886, "step": 2767 }, { "epoch": 0.48, "learning_rate": 4.732201231467004e-05, "loss": 0.3646, "step": 2768 }, { "epoch": 0.48, "learning_rate": 4.7318772440242245e-05, "loss": 0.3542, "step": 2769 }, { "epoch": 0.48, "learning_rate": 4.731553073133754e-05, "loss": 0.3544, "step": 2770 }, { "epoch": 0.48, "learning_rate": 4.731228718825597e-05, "loss": 0.3566, "step": 2771 }, { "epoch": 0.48, "learning_rate": 4.7309041811297775e-05, "loss": 0.3706, "step": 2772 }, { "epoch": 0.48, "learning_rate": 4.7305794600763356e-05, "loss": 0.3642, "step": 2773 }, { "epoch": 0.48, "learning_rate": 4.730254555695327e-05, "loss": 0.3567, "step": 2774 }, { "epoch": 0.48, "learning_rate": 4.729929468016829e-05, "loss": 0.3605, "step": 2775 }, { "epoch": 0.48, "learning_rate": 4.729604197070929e-05, "loss": 0.3803, "step": 2776 }, { "epoch": 0.48, "learning_rate": 4.7292787428877366e-05, "loss": 0.3483, "step": 2777 }, { "epoch": 0.48, "learning_rate": 4.728953105497376e-05, "loss": 0.3604, "step": 2778 }, { "epoch": 0.48, "learning_rate": 4.7286272849299904e-05, "loss": 0.3571, "step": 2779 }, { "epoch": 0.48, "learning_rate": 4.728301281215736e-05, "loss": 0.366, "step": 2780 }, { "epoch": 0.48, "learning_rate": 4.7279750943847916e-05, "loss": 0.365, "step": 2781 }, { "epoch": 0.48, "learning_rate": 4.727648724467348e-05, "loss": 0.395, "step": 2782 }, { "epoch": 0.49, "learning_rate": 4.727322171493616e-05, "loss": 0.3707, "step": 2783 }, { "epoch": 0.49, "learning_rate": 4.72699543549382e-05, "loss": 0.3636, "step": 2784 }, { "epoch": 0.49, "learning_rate": 4.7266685164982053e-05, "loss": 0.3692, "step": 2785 }, { "epoch": 0.49, "learning_rate": 4.726341414537033e-05, "loss": 0.3737, "step": 2786 }, { "epoch": 0.49, "learning_rate": 4.7260141296405794e-05, "loss": 0.3596, "step": 2787 }, { "epoch": 0.49, "learning_rate": 4.725686661839139e-05, "loss": 0.3799, "step": 2788 }, { "epoch": 0.49, "learning_rate": 4.725359011163023e-05, "loss": 0.3799, "step": 2789 }, { "epoch": 0.49, "learning_rate": 4.7250311776425596e-05, "loss": 0.3523, "step": 2790 }, { "epoch": 0.49, "learning_rate": 4.724703161308094e-05, "loss": 0.3733, "step": 2791 }, { "epoch": 0.49, "learning_rate": 4.724374962189989e-05, "loss": 0.3941, "step": 2792 }, { "epoch": 0.49, "learning_rate": 4.724046580318622e-05, "loss": 0.3664, "step": 2793 }, { "epoch": 0.49, "learning_rate": 4.7237180157243895e-05, "loss": 0.3653, "step": 2794 }, { "epoch": 0.49, "learning_rate": 4.7233892684377055e-05, "loss": 0.383, "step": 2795 }, { "epoch": 0.49, "learning_rate": 4.723060338488997e-05, "loss": 0.3781, "step": 2796 }, { "epoch": 0.49, "learning_rate": 4.722731225908714e-05, "loss": 0.3561, "step": 2797 }, { "epoch": 0.49, "learning_rate": 4.722401930727317e-05, "loss": 0.3558, "step": 2798 }, { "epoch": 0.49, "learning_rate": 4.722072452975289e-05, "loss": 0.3703, "step": 2799 }, { "epoch": 0.49, "learning_rate": 4.721742792683125e-05, "loss": 0.3754, "step": 2800 }, { "epoch": 0.49, "learning_rate": 4.721412949881341e-05, "loss": 0.3746, "step": 2801 }, { "epoch": 0.49, "learning_rate": 4.721082924600467e-05, "loss": 0.3488, "step": 2802 }, { "epoch": 0.49, "learning_rate": 4.720752716871052e-05, "loss": 0.3729, "step": 2803 }, { "epoch": 0.49, "learning_rate": 4.720422326723659e-05, "loss": 0.3804, "step": 2804 }, { "epoch": 0.49, "learning_rate": 4.720091754188872e-05, "loss": 0.3734, "step": 2805 }, { "epoch": 0.49, "learning_rate": 4.719760999297288e-05, "loss": 0.3549, "step": 2806 }, { "epoch": 0.49, "learning_rate": 4.7194300620795234e-05, "loss": 0.3617, "step": 2807 }, { "epoch": 0.49, "learning_rate": 4.719098942566211e-05, "loss": 0.3601, "step": 2808 }, { "epoch": 0.49, "learning_rate": 4.718767640787999e-05, "loss": 0.3692, "step": 2809 }, { "epoch": 0.49, "learning_rate": 4.7184361567755534e-05, "loss": 0.365, "step": 2810 }, { "epoch": 0.49, "learning_rate": 4.718104490559559e-05, "loss": 0.3666, "step": 2811 }, { "epoch": 0.49, "learning_rate": 4.7177726421707135e-05, "loss": 0.3831, "step": 2812 }, { "epoch": 0.49, "learning_rate": 4.717440611639736e-05, "loss": 0.3569, "step": 2813 }, { "epoch": 0.49, "learning_rate": 4.717108398997357e-05, "loss": 0.3674, "step": 2814 }, { "epoch": 0.49, "learning_rate": 4.71677600427433e-05, "loss": 0.3679, "step": 2815 }, { "epoch": 0.49, "learning_rate": 4.716443427501421e-05, "loss": 0.3852, "step": 2816 }, { "epoch": 0.49, "learning_rate": 4.7161106687094134e-05, "loss": 0.36, "step": 2817 }, { "epoch": 0.49, "learning_rate": 4.715777727929109e-05, "loss": 0.3539, "step": 2818 }, { "epoch": 0.49, "learning_rate": 4.715444605191327e-05, "loss": 0.357, "step": 2819 }, { "epoch": 0.49, "learning_rate": 4.715111300526899e-05, "loss": 0.3478, "step": 2820 }, { "epoch": 0.49, "learning_rate": 4.71477781396668e-05, "loss": 0.3475, "step": 2821 }, { "epoch": 0.49, "learning_rate": 4.714444145541536e-05, "loss": 0.3739, "step": 2822 }, { "epoch": 0.49, "learning_rate": 4.7141102952823534e-05, "loss": 0.377, "step": 2823 }, { "epoch": 0.49, "learning_rate": 4.713776263220033e-05, "loss": 0.3816, "step": 2824 }, { "epoch": 0.49, "learning_rate": 4.713442049385495e-05, "loss": 0.3689, "step": 2825 }, { "epoch": 0.49, "learning_rate": 4.713107653809675e-05, "loss": 0.3482, "step": 2826 }, { "epoch": 0.49, "learning_rate": 4.712773076523524e-05, "loss": 0.3781, "step": 2827 }, { "epoch": 0.49, "learning_rate": 4.712438317558013e-05, "loss": 0.3676, "step": 2828 }, { "epoch": 0.49, "learning_rate": 4.712103376944128e-05, "loss": 0.3891, "step": 2829 }, { "epoch": 0.49, "learning_rate": 4.71176825471287e-05, "loss": 0.381, "step": 2830 }, { "epoch": 0.49, "learning_rate": 4.7114329508952624e-05, "loss": 0.4042, "step": 2831 }, { "epoch": 0.49, "learning_rate": 4.7110974655223386e-05, "loss": 0.3661, "step": 2832 }, { "epoch": 0.49, "learning_rate": 4.710761798625153e-05, "loss": 0.365, "step": 2833 }, { "epoch": 0.49, "learning_rate": 4.710425950234777e-05, "loss": 0.3694, "step": 2834 }, { "epoch": 0.49, "learning_rate": 4.710089920382295e-05, "loss": 0.3776, "step": 2835 }, { "epoch": 0.49, "learning_rate": 4.7097537090988126e-05, "loss": 0.3657, "step": 2836 }, { "epoch": 0.49, "learning_rate": 4.7094173164154504e-05, "loss": 0.3542, "step": 2837 }, { "epoch": 0.49, "learning_rate": 4.709080742363345e-05, "loss": 0.395, "step": 2838 }, { "epoch": 0.49, "learning_rate": 4.708743986973652e-05, "loss": 0.3736, "step": 2839 }, { "epoch": 0.5, "learning_rate": 4.708407050277541e-05, "loss": 0.362, "step": 2840 }, { "epoch": 0.5, "learning_rate": 4.7080699323062e-05, "loss": 0.3605, "step": 2841 }, { "epoch": 0.5, "learning_rate": 4.707732633090834e-05, "loss": 0.3747, "step": 2842 }, { "epoch": 0.5, "learning_rate": 4.707395152662663e-05, "loss": 0.3731, "step": 2843 }, { "epoch": 0.5, "learning_rate": 4.7070574910529266e-05, "loss": 0.3572, "step": 2844 }, { "epoch": 0.5, "learning_rate": 4.706719648292879e-05, "loss": 0.3809, "step": 2845 }, { "epoch": 0.5, "learning_rate": 4.706381624413792e-05, "loss": 0.3575, "step": 2846 }, { "epoch": 0.5, "learning_rate": 4.706043419446953e-05, "loss": 0.3665, "step": 2847 }, { "epoch": 0.5, "learning_rate": 4.70570503342367e-05, "loss": 0.367, "step": 2848 }, { "epoch": 0.5, "learning_rate": 4.70536646637526e-05, "loss": 0.3522, "step": 2849 }, { "epoch": 0.5, "learning_rate": 4.705027718333066e-05, "loss": 0.3596, "step": 2850 }, { "epoch": 0.5, "learning_rate": 4.704688789328442e-05, "loss": 0.3723, "step": 2851 }, { "epoch": 0.5, "learning_rate": 4.704349679392759e-05, "loss": 0.3638, "step": 2852 }, { "epoch": 0.5, "learning_rate": 4.704010388557408e-05, "loss": 0.3746, "step": 2853 }, { "epoch": 0.5, "learning_rate": 4.703670916853792e-05, "loss": 0.3712, "step": 2854 }, { "epoch": 0.5, "learning_rate": 4.703331264313336e-05, "loss": 0.3713, "step": 2855 }, { "epoch": 0.5, "learning_rate": 4.7029914309674785e-05, "loss": 0.3604, "step": 2856 }, { "epoch": 0.5, "learning_rate": 4.7026514168476736e-05, "loss": 0.3719, "step": 2857 }, { "epoch": 0.5, "learning_rate": 4.702311221985397e-05, "loss": 0.356, "step": 2858 }, { "epoch": 0.5, "learning_rate": 4.7019708464121344e-05, "loss": 0.3486, "step": 2859 }, { "epoch": 0.5, "learning_rate": 4.7016302901593935e-05, "loss": 0.3672, "step": 2860 }, { "epoch": 0.5, "learning_rate": 4.701289553258698e-05, "loss": 0.3716, "step": 2861 }, { "epoch": 0.5, "learning_rate": 4.700948635741586e-05, "loss": 0.3655, "step": 2862 }, { "epoch": 0.5, "learning_rate": 4.7006075376396155e-05, "loss": 0.3733, "step": 2863 }, { "epoch": 0.5, "learning_rate": 4.700266258984357e-05, "loss": 0.3559, "step": 2864 }, { "epoch": 0.5, "learning_rate": 4.6999247998074023e-05, "loss": 0.3502, "step": 2865 }, { "epoch": 0.5, "learning_rate": 4.6995831601403563e-05, "loss": 0.3637, "step": 2866 }, { "epoch": 0.5, "learning_rate": 4.6992413400148423e-05, "loss": 0.3641, "step": 2867 }, { "epoch": 0.5, "learning_rate": 4.6988993394625e-05, "loss": 0.3649, "step": 2868 }, { "epoch": 0.5, "learning_rate": 4.6985571585149876e-05, "loss": 0.3677, "step": 2869 }, { "epoch": 0.5, "learning_rate": 4.698214797203976e-05, "loss": 0.3713, "step": 2870 }, { "epoch": 0.5, "learning_rate": 4.697872255561155e-05, "loss": 0.3852, "step": 2871 }, { "epoch": 0.5, "learning_rate": 4.697529533618233e-05, "loss": 0.3436, "step": 2872 }, { "epoch": 0.5, "learning_rate": 4.6971866314069324e-05, "loss": 0.3545, "step": 2873 }, { "epoch": 0.5, "learning_rate": 4.696843548958992e-05, "loss": 0.3637, "step": 2874 }, { "epoch": 0.5, "learning_rate": 4.6965002863061705e-05, "loss": 0.3746, "step": 2875 }, { "epoch": 0.5, "learning_rate": 4.696156843480239e-05, "loss": 0.3613, "step": 2876 }, { "epoch": 0.5, "learning_rate": 4.695813220512989e-05, "loss": 0.3649, "step": 2877 }, { "epoch": 0.5, "learning_rate": 4.6954694174362264e-05, "loss": 0.3544, "step": 2878 }, { "epoch": 0.5, "learning_rate": 4.695125434281775e-05, "loss": 0.3701, "step": 2879 }, { "epoch": 0.5, "learning_rate": 4.694781271081474e-05, "loss": 0.3558, "step": 2880 }, { "epoch": 0.5, "learning_rate": 4.694436927867182e-05, "loss": 0.3659, "step": 2881 }, { "epoch": 0.5, "learning_rate": 4.694092404670769e-05, "loss": 0.3537, "step": 2882 }, { "epoch": 0.5, "learning_rate": 4.693747701524128e-05, "loss": 0.3685, "step": 2883 }, { "epoch": 0.5, "learning_rate": 4.693402818459164e-05, "loss": 0.374, "step": 2884 }, { "epoch": 0.5, "learning_rate": 4.693057755507801e-05, "loss": 0.3707, "step": 2885 }, { "epoch": 0.5, "learning_rate": 4.6927125127019775e-05, "loss": 0.3606, "step": 2886 }, { "epoch": 0.5, "learning_rate": 4.692367090073652e-05, "loss": 0.3698, "step": 2887 }, { "epoch": 0.5, "learning_rate": 4.6920214876547976e-05, "loss": 0.379, "step": 2888 }, { "epoch": 0.5, "learning_rate": 4.691675705477402e-05, "loss": 0.3662, "step": 2889 }, { "epoch": 0.5, "learning_rate": 4.6913297435734736e-05, "loss": 0.3732, "step": 2890 }, { "epoch": 0.5, "learning_rate": 4.690983601975035e-05, "loss": 0.3627, "step": 2891 }, { "epoch": 0.5, "learning_rate": 4.6906372807141265e-05, "loss": 0.3732, "step": 2892 }, { "epoch": 0.5, "learning_rate": 4.690290779822804e-05, "loss": 0.3563, "step": 2893 }, { "epoch": 0.5, "learning_rate": 4.6899440993331404e-05, "loss": 0.3568, "step": 2894 }, { "epoch": 0.5, "learning_rate": 4.689597239277226e-05, "loss": 0.3742, "step": 2895 }, { "epoch": 0.5, "learning_rate": 4.6892501996871654e-05, "loss": 0.3676, "step": 2896 }, { "epoch": 0.5, "learning_rate": 4.688902980595084e-05, "loss": 0.38, "step": 2897 }, { "epoch": 0.51, "learning_rate": 4.688555582033118e-05, "loss": 0.3656, "step": 2898 }, { "epoch": 0.51, "learning_rate": 4.688208004033427e-05, "loss": 0.3548, "step": 2899 }, { "epoch": 0.51, "learning_rate": 4.687860246628182e-05, "loss": 0.3659, "step": 2900 }, { "epoch": 0.51, "learning_rate": 4.687512309849572e-05, "loss": 0.3584, "step": 2901 }, { "epoch": 0.51, "learning_rate": 4.6871641937298036e-05, "loss": 0.3618, "step": 2902 }, { "epoch": 0.51, "learning_rate": 4.686815898301099e-05, "loss": 0.3773, "step": 2903 }, { "epoch": 0.51, "learning_rate": 4.686467423595698e-05, "loss": 0.3699, "step": 2904 }, { "epoch": 0.51, "learning_rate": 4.686118769645856e-05, "loss": 0.358, "step": 2905 }, { "epoch": 0.51, "learning_rate": 4.685769936483845e-05, "loss": 0.3511, "step": 2906 }, { "epoch": 0.51, "learning_rate": 4.685420924141953e-05, "loss": 0.353, "step": 2907 }, { "epoch": 0.51, "learning_rate": 4.685071732652487e-05, "loss": 0.3711, "step": 2908 }, { "epoch": 0.51, "learning_rate": 4.6847223620477686e-05, "loss": 0.3452, "step": 2909 }, { "epoch": 0.51, "learning_rate": 4.684372812360137e-05, "loss": 0.3637, "step": 2910 }, { "epoch": 0.51, "learning_rate": 4.684023083621946e-05, "loss": 0.3639, "step": 2911 }, { "epoch": 0.51, "learning_rate": 4.683673175865569e-05, "loss": 0.3494, "step": 2912 }, { "epoch": 0.51, "learning_rate": 4.683323089123393e-05, "loss": 0.3764, "step": 2913 }, { "epoch": 0.51, "learning_rate": 4.6829728234278234e-05, "loss": 0.3693, "step": 2914 }, { "epoch": 0.51, "learning_rate": 4.682622378811282e-05, "loss": 0.3685, "step": 2915 }, { "epoch": 0.51, "learning_rate": 4.682271755306207e-05, "loss": 0.3521, "step": 2916 }, { "epoch": 0.51, "learning_rate": 4.6819209529450526e-05, "loss": 0.3767, "step": 2917 }, { "epoch": 0.51, "learning_rate": 4.68156997176029e-05, "loss": 0.3545, "step": 2918 }, { "epoch": 0.51, "learning_rate": 4.6812188117844075e-05, "loss": 0.3491, "step": 2919 }, { "epoch": 0.51, "learning_rate": 4.680867473049908e-05, "loss": 0.3675, "step": 2920 }, { "epoch": 0.51, "learning_rate": 4.680515955589314e-05, "loss": 0.3763, "step": 2921 }, { "epoch": 0.51, "learning_rate": 4.680164259435162e-05, "loss": 0.3698, "step": 2922 }, { "epoch": 0.51, "learning_rate": 4.679812384620006e-05, "loss": 0.3603, "step": 2923 }, { "epoch": 0.51, "learning_rate": 4.679460331176416e-05, "loss": 0.3732, "step": 2924 }, { "epoch": 0.51, "learning_rate": 4.679108099136979e-05, "loss": 0.3523, "step": 2925 }, { "epoch": 0.51, "learning_rate": 4.6787556885343006e-05, "loss": 0.3646, "step": 2926 }, { "epoch": 0.51, "learning_rate": 4.678403099400999e-05, "loss": 0.3838, "step": 2927 }, { "epoch": 0.51, "learning_rate": 4.678050331769709e-05, "loss": 0.3701, "step": 2928 }, { "epoch": 0.51, "learning_rate": 4.6776973856730874e-05, "loss": 0.3653, "step": 2929 }, { "epoch": 0.51, "learning_rate": 4.677344261143802e-05, "loss": 0.395, "step": 2930 }, { "epoch": 0.51, "learning_rate": 4.6769909582145374e-05, "loss": 0.3675, "step": 2931 }, { "epoch": 0.51, "learning_rate": 4.676637476917999e-05, "loss": 0.3714, "step": 2932 }, { "epoch": 0.51, "learning_rate": 4.6762838172869046e-05, "loss": 0.3571, "step": 2933 }, { "epoch": 0.51, "learning_rate": 4.67592997935399e-05, "loss": 0.3499, "step": 2934 }, { "epoch": 0.51, "learning_rate": 4.6755759631520076e-05, "loss": 0.3897, "step": 2935 }, { "epoch": 0.51, "learning_rate": 4.6752217687137246e-05, "loss": 0.3857, "step": 2936 }, { "epoch": 0.51, "learning_rate": 4.6748673960719285e-05, "loss": 0.3541, "step": 2937 }, { "epoch": 0.51, "learning_rate": 4.67451284525942e-05, "loss": 0.3626, "step": 2938 }, { "epoch": 0.51, "learning_rate": 4.674158116309016e-05, "loss": 0.3616, "step": 2939 }, { "epoch": 0.51, "learning_rate": 4.6738032092535523e-05, "loss": 0.3577, "step": 2940 }, { "epoch": 0.51, "learning_rate": 4.67344812412588e-05, "loss": 0.4103, "step": 2941 }, { "epoch": 0.51, "learning_rate": 4.673092860958867e-05, "loss": 0.3731, "step": 2942 }, { "epoch": 0.51, "learning_rate": 4.6727374197853965e-05, "loss": 0.3705, "step": 2943 }, { "epoch": 0.51, "learning_rate": 4.67238180063837e-05, "loss": 0.3685, "step": 2944 }, { "epoch": 0.51, "learning_rate": 4.672026003550704e-05, "loss": 0.3631, "step": 2945 }, { "epoch": 0.51, "learning_rate": 4.671670028555332e-05, "loss": 0.3662, "step": 2946 }, { "epoch": 0.51, "learning_rate": 4.6713138756852035e-05, "loss": 0.3474, "step": 2947 }, { "epoch": 0.51, "learning_rate": 4.6709575449732864e-05, "loss": 0.3642, "step": 2948 }, { "epoch": 0.51, "learning_rate": 4.670601036452561e-05, "loss": 0.3732, "step": 2949 }, { "epoch": 0.51, "learning_rate": 4.67024435015603e-05, "loss": 0.3671, "step": 2950 }, { "epoch": 0.51, "learning_rate": 4.669887486116708e-05, "loss": 0.3635, "step": 2951 }, { "epoch": 0.51, "learning_rate": 4.669530444367625e-05, "loss": 0.3854, "step": 2952 }, { "epoch": 0.51, "learning_rate": 4.669173224941832e-05, "loss": 0.3728, "step": 2953 }, { "epoch": 0.51, "learning_rate": 4.668815827872394e-05, "loss": 0.3679, "step": 2954 }, { "epoch": 0.52, "learning_rate": 4.668458253192391e-05, "loss": 0.37, "step": 2955 }, { "epoch": 0.52, "learning_rate": 4.668100500934924e-05, "loss": 0.3768, "step": 2956 }, { "epoch": 0.52, "learning_rate": 4.667742571133105e-05, "loss": 0.3558, "step": 2957 }, { "epoch": 0.52, "learning_rate": 4.667384463820065e-05, "loss": 0.348, "step": 2958 }, { "epoch": 0.52, "learning_rate": 4.667026179028953e-05, "loss": 0.3478, "step": 2959 }, { "epoch": 0.52, "learning_rate": 4.666667716792931e-05, "loss": 0.3709, "step": 2960 }, { "epoch": 0.52, "learning_rate": 4.666309077145181e-05, "loss": 0.3616, "step": 2961 }, { "epoch": 0.52, "learning_rate": 4.6659502601188975e-05, "loss": 0.3682, "step": 2962 }, { "epoch": 0.52, "learning_rate": 4.6655912657472945e-05, "loss": 0.3625, "step": 2963 }, { "epoch": 0.52, "learning_rate": 4.665232094063603e-05, "loss": 0.3646, "step": 2964 }, { "epoch": 0.52, "learning_rate": 4.664872745101067e-05, "loss": 0.3695, "step": 2965 }, { "epoch": 0.52, "learning_rate": 4.664513218892948e-05, "loss": 0.3566, "step": 2966 }, { "epoch": 0.52, "learning_rate": 4.6641535154725265e-05, "loss": 0.3735, "step": 2967 }, { "epoch": 0.52, "learning_rate": 4.663793634873097e-05, "loss": 0.3782, "step": 2968 }, { "epoch": 0.52, "learning_rate": 4.6634335771279705e-05, "loss": 0.353, "step": 2969 }, { "epoch": 0.52, "learning_rate": 4.6630733422704756e-05, "loss": 0.3678, "step": 2970 }, { "epoch": 0.52, "learning_rate": 4.662712930333956e-05, "loss": 0.3772, "step": 2971 }, { "epoch": 0.52, "learning_rate": 4.6623523413517725e-05, "loss": 0.3592, "step": 2972 }, { "epoch": 0.52, "learning_rate": 4.661991575357304e-05, "loss": 0.3753, "step": 2973 }, { "epoch": 0.52, "learning_rate": 4.66163063238394e-05, "loss": 0.3575, "step": 2974 }, { "epoch": 0.52, "learning_rate": 4.661269512465094e-05, "loss": 0.3708, "step": 2975 }, { "epoch": 0.52, "learning_rate": 4.66090821563419e-05, "loss": 0.3702, "step": 2976 }, { "epoch": 0.52, "learning_rate": 4.660546741924672e-05, "loss": 0.3688, "step": 2977 }, { "epoch": 0.52, "learning_rate": 4.660185091369997e-05, "loss": 0.3599, "step": 2978 }, { "epoch": 0.52, "learning_rate": 4.659823264003642e-05, "loss": 0.3683, "step": 2979 }, { "epoch": 0.52, "learning_rate": 4.659461259859099e-05, "loss": 0.3853, "step": 2980 }, { "epoch": 0.52, "learning_rate": 4.659099078969875e-05, "loss": 0.3686, "step": 2981 }, { "epoch": 0.52, "learning_rate": 4.658736721369495e-05, "loss": 0.3596, "step": 2982 }, { "epoch": 0.52, "learning_rate": 4.658374187091498e-05, "loss": 0.3643, "step": 2983 }, { "epoch": 0.52, "learning_rate": 4.6580114761694435e-05, "loss": 0.3724, "step": 2984 }, { "epoch": 0.52, "learning_rate": 4.6576485886369045e-05, "loss": 0.3802, "step": 2985 }, { "epoch": 0.52, "learning_rate": 4.65728552452747e-05, "loss": 0.3504, "step": 2986 }, { "epoch": 0.52, "learning_rate": 4.6569222838747455e-05, "loss": 0.3689, "step": 2987 }, { "epoch": 0.52, "learning_rate": 4.6565588667123557e-05, "loss": 0.383, "step": 2988 }, { "epoch": 0.52, "learning_rate": 4.656195273073938e-05, "loss": 0.3686, "step": 2989 }, { "epoch": 0.52, "learning_rate": 4.6558315029931467e-05, "loss": 0.3578, "step": 2990 }, { "epoch": 0.52, "learning_rate": 4.655467556503655e-05, "loss": 0.3541, "step": 2991 }, { "epoch": 0.52, "learning_rate": 4.655103433639151e-05, "loss": 0.3669, "step": 2992 }, { "epoch": 0.52, "learning_rate": 4.654739134433338e-05, "loss": 0.3677, "step": 2993 }, { "epoch": 0.52, "learning_rate": 4.654374658919935e-05, "loss": 0.3561, "step": 2994 }, { "epoch": 0.52, "learning_rate": 4.65401000713268e-05, "loss": 0.3873, "step": 2995 }, { "epoch": 0.52, "learning_rate": 4.653645179105328e-05, "loss": 0.3692, "step": 2996 }, { "epoch": 0.52, "learning_rate": 4.6532801748716456e-05, "loss": 0.3718, "step": 2997 }, { "epoch": 0.52, "learning_rate": 4.65291499446542e-05, "loss": 0.3723, "step": 2998 }, { "epoch": 0.52, "learning_rate": 4.652549637920453e-05, "loss": 0.3642, "step": 2999 }, { "epoch": 0.52, "learning_rate": 4.6521841052705634e-05, "loss": 0.3624, "step": 3000 }, { "epoch": 0.52, "learning_rate": 4.651818396549585e-05, "loss": 0.3774, "step": 3001 }, { "epoch": 0.52, "learning_rate": 4.6514525117913686e-05, "loss": 0.3509, "step": 3002 }, { "epoch": 0.52, "learning_rate": 4.651086451029783e-05, "loss": 0.3438, "step": 3003 }, { "epoch": 0.52, "learning_rate": 4.65072021429871e-05, "loss": 0.3484, "step": 3004 }, { "epoch": 0.52, "learning_rate": 4.65035380163205e-05, "loss": 0.3631, "step": 3005 }, { "epoch": 0.52, "learning_rate": 4.649987213063719e-05, "loss": 0.3812, "step": 3006 }, { "epoch": 0.52, "learning_rate": 4.64962044862765e-05, "loss": 0.3744, "step": 3007 }, { "epoch": 0.52, "learning_rate": 4.649253508357792e-05, "loss": 0.3769, "step": 3008 }, { "epoch": 0.52, "learning_rate": 4.648886392288108e-05, "loss": 0.3543, "step": 3009 }, { "epoch": 0.52, "learning_rate": 4.648519100452581e-05, "loss": 0.374, "step": 3010 }, { "epoch": 0.52, "learning_rate": 4.6481516328852074e-05, "loss": 0.356, "step": 3011 }, { "epoch": 0.52, "learning_rate": 4.6477839896200016e-05, "loss": 0.3464, "step": 3012 }, { "epoch": 0.53, "learning_rate": 4.6474161706909936e-05, "loss": 0.3719, "step": 3013 }, { "epoch": 0.53, "learning_rate": 4.6470481761322296e-05, "loss": 0.3634, "step": 3014 }, { "epoch": 0.53, "learning_rate": 4.646680005977772e-05, "loss": 0.3629, "step": 3015 }, { "epoch": 0.53, "learning_rate": 4.6463116602617e-05, "loss": 0.382, "step": 3016 }, { "epoch": 0.53, "learning_rate": 4.6459431390181074e-05, "loss": 0.3697, "step": 3017 }, { "epoch": 0.53, "learning_rate": 4.6455744422811075e-05, "loss": 0.3533, "step": 3018 }, { "epoch": 0.53, "learning_rate": 4.645205570084825e-05, "loss": 0.3796, "step": 3019 }, { "epoch": 0.53, "learning_rate": 4.644836522463407e-05, "loss": 0.3677, "step": 3020 }, { "epoch": 0.53, "learning_rate": 4.6444672994510114e-05, "loss": 0.365, "step": 3021 }, { "epoch": 0.53, "learning_rate": 4.644097901081815e-05, "loss": 0.3576, "step": 3022 }, { "epoch": 0.53, "learning_rate": 4.64372832739001e-05, "loss": 0.3708, "step": 3023 }, { "epoch": 0.53, "learning_rate": 4.643358578409804e-05, "loss": 0.3655, "step": 3024 }, { "epoch": 0.53, "learning_rate": 4.642988654175425e-05, "loss": 0.3811, "step": 3025 }, { "epoch": 0.53, "learning_rate": 4.642618554721111e-05, "loss": 0.3801, "step": 3026 }, { "epoch": 0.53, "learning_rate": 4.642248280081122e-05, "loss": 0.3676, "step": 3027 }, { "epoch": 0.53, "learning_rate": 4.64187783028973e-05, "loss": 0.3657, "step": 3028 }, { "epoch": 0.53, "learning_rate": 4.641507205381224e-05, "loss": 0.3674, "step": 3029 }, { "epoch": 0.53, "learning_rate": 4.641136405389912e-05, "loss": 0.3645, "step": 3030 }, { "epoch": 0.53, "learning_rate": 4.6407654303501155e-05, "loss": 0.3435, "step": 3031 }, { "epoch": 0.53, "learning_rate": 4.640394280296172e-05, "loss": 0.3574, "step": 3032 }, { "epoch": 0.53, "learning_rate": 4.640022955262437e-05, "loss": 0.3532, "step": 3033 }, { "epoch": 0.53, "learning_rate": 4.639651455283282e-05, "loss": 0.377, "step": 3034 }, { "epoch": 0.53, "learning_rate": 4.6392797803930916e-05, "loss": 0.3695, "step": 3035 }, { "epoch": 0.53, "learning_rate": 4.638907930626271e-05, "loss": 0.3461, "step": 3036 }, { "epoch": 0.53, "learning_rate": 4.638535906017239e-05, "loss": 0.3672, "step": 3037 }, { "epoch": 0.53, "learning_rate": 4.6381637066004315e-05, "loss": 0.3578, "step": 3038 }, { "epoch": 0.53, "learning_rate": 4.6377913324103e-05, "loss": 0.356, "step": 3039 }, { "epoch": 0.53, "learning_rate": 4.637418783481312e-05, "loss": 0.3738, "step": 3040 }, { "epoch": 0.53, "learning_rate": 4.6370460598479535e-05, "loss": 0.3664, "step": 3041 }, { "epoch": 0.53, "learning_rate": 4.636673161544721e-05, "loss": 0.3792, "step": 3042 }, { "epoch": 0.53, "learning_rate": 4.636300088606134e-05, "loss": 0.3768, "step": 3043 }, { "epoch": 0.53, "learning_rate": 4.635926841066724e-05, "loss": 0.3743, "step": 3044 }, { "epoch": 0.53, "learning_rate": 4.635553418961042e-05, "loss": 0.3758, "step": 3045 }, { "epoch": 0.53, "learning_rate": 4.635179822323649e-05, "loss": 0.3652, "step": 3046 }, { "epoch": 0.53, "learning_rate": 4.6348060511891285e-05, "loss": 0.3619, "step": 3047 }, { "epoch": 0.53, "learning_rate": 4.634432105592077e-05, "loss": 0.3725, "step": 3048 }, { "epoch": 0.53, "learning_rate": 4.634057985567108e-05, "loss": 0.3618, "step": 3049 }, { "epoch": 0.53, "learning_rate": 4.633683691148852e-05, "loss": 0.3664, "step": 3050 }, { "epoch": 0.53, "learning_rate": 4.6333092223719545e-05, "loss": 0.3573, "step": 3051 }, { "epoch": 0.53, "learning_rate": 4.6329345792710746e-05, "loss": 0.3757, "step": 3052 }, { "epoch": 0.53, "learning_rate": 4.632559761880893e-05, "loss": 0.372, "step": 3053 }, { "epoch": 0.53, "learning_rate": 4.632184770236104e-05, "loss": 0.3503, "step": 3054 }, { "epoch": 0.53, "learning_rate": 4.631809604371415e-05, "loss": 0.3489, "step": 3055 }, { "epoch": 0.53, "learning_rate": 4.6314342643215556e-05, "loss": 0.3577, "step": 3056 }, { "epoch": 0.53, "learning_rate": 4.6310587501212666e-05, "loss": 0.3593, "step": 3057 }, { "epoch": 0.53, "learning_rate": 4.6306830618053064e-05, "loss": 0.3676, "step": 3058 }, { "epoch": 0.53, "learning_rate": 4.63030719940845e-05, "loss": 0.3828, "step": 3059 }, { "epoch": 0.53, "learning_rate": 4.629931162965489e-05, "loss": 0.3548, "step": 3060 }, { "epoch": 0.53, "learning_rate": 4.629554952511229e-05, "loss": 0.3755, "step": 3061 }, { "epoch": 0.53, "learning_rate": 4.629178568080493e-05, "loss": 0.345, "step": 3062 }, { "epoch": 0.53, "learning_rate": 4.628802009708121e-05, "loss": 0.3589, "step": 3063 }, { "epoch": 0.53, "learning_rate": 4.628425277428968e-05, "loss": 0.3815, "step": 3064 }, { "epoch": 0.53, "learning_rate": 4.628048371277906e-05, "loss": 0.3733, "step": 3065 }, { "epoch": 0.53, "learning_rate": 4.62767129128982e-05, "loss": 0.3639, "step": 3066 }, { "epoch": 0.53, "learning_rate": 4.627294037499616e-05, "loss": 0.3609, "step": 3067 }, { "epoch": 0.53, "learning_rate": 4.6269166099422126e-05, "loss": 0.3586, "step": 3068 }, { "epoch": 0.53, "learning_rate": 4.626539008652546e-05, "loss": 0.3755, "step": 3069 }, { "epoch": 0.54, "learning_rate": 4.626161233665567e-05, "loss": 0.3758, "step": 3070 }, { "epoch": 0.54, "learning_rate": 4.625783285016243e-05, "loss": 0.355, "step": 3071 }, { "epoch": 0.54, "learning_rate": 4.625405162739561e-05, "loss": 0.3533, "step": 3072 }, { "epoch": 0.54, "learning_rate": 4.6250268668705176e-05, "loss": 0.348, "step": 3073 }, { "epoch": 0.54, "learning_rate": 4.6246483974441304e-05, "loss": 0.3723, "step": 3074 }, { "epoch": 0.54, "learning_rate": 4.624269754495431e-05, "loss": 0.3623, "step": 3075 }, { "epoch": 0.54, "learning_rate": 4.6238909380594685e-05, "loss": 0.3607, "step": 3076 }, { "epoch": 0.54, "learning_rate": 4.623511948171305e-05, "loss": 0.3611, "step": 3077 }, { "epoch": 0.54, "learning_rate": 4.623132784866023e-05, "loss": 0.3653, "step": 3078 }, { "epoch": 0.54, "learning_rate": 4.622753448178718e-05, "loss": 0.3756, "step": 3079 }, { "epoch": 0.54, "learning_rate": 4.622373938144503e-05, "loss": 0.3654, "step": 3080 }, { "epoch": 0.54, "learning_rate": 4.6219942547985054e-05, "loss": 0.3767, "step": 3081 }, { "epoch": 0.54, "learning_rate": 4.62161439817587e-05, "loss": 0.3617, "step": 3082 }, { "epoch": 0.54, "learning_rate": 4.621234368311757e-05, "loss": 0.3702, "step": 3083 }, { "epoch": 0.54, "learning_rate": 4.620854165241344e-05, "loss": 0.365, "step": 3084 }, { "epoch": 0.54, "learning_rate": 4.620473788999823e-05, "loss": 0.3672, "step": 3085 }, { "epoch": 0.54, "learning_rate": 4.620093239622403e-05, "loss": 0.3514, "step": 3086 }, { "epoch": 0.54, "learning_rate": 4.619712517144307e-05, "loss": 0.3356, "step": 3087 }, { "epoch": 0.54, "learning_rate": 4.619331621600777e-05, "loss": 0.3501, "step": 3088 }, { "epoch": 0.54, "learning_rate": 4.61895055302707e-05, "loss": 0.3522, "step": 3089 }, { "epoch": 0.54, "learning_rate": 4.6185693114584586e-05, "loss": 0.3629, "step": 3090 }, { "epoch": 0.54, "learning_rate": 4.618187896930231e-05, "loss": 0.3485, "step": 3091 }, { "epoch": 0.54, "learning_rate": 4.617806309477692e-05, "loss": 0.3614, "step": 3092 }, { "epoch": 0.54, "learning_rate": 4.6174245491361616e-05, "loss": 0.3748, "step": 3093 }, { "epoch": 0.54, "learning_rate": 4.6170426159409777e-05, "loss": 0.365, "step": 3094 }, { "epoch": 0.54, "learning_rate": 4.616660509927493e-05, "loss": 0.3544, "step": 3095 }, { "epoch": 0.54, "learning_rate": 4.616278231131076e-05, "loss": 0.345, "step": 3096 }, { "epoch": 0.54, "learning_rate": 4.615895779587111e-05, "loss": 0.3552, "step": 3097 }, { "epoch": 0.54, "learning_rate": 4.6155131553309995e-05, "loss": 0.3758, "step": 3098 }, { "epoch": 0.54, "learning_rate": 4.615130358398157e-05, "loss": 0.3625, "step": 3099 }, { "epoch": 0.54, "learning_rate": 4.614747388824017e-05, "loss": 0.3563, "step": 3100 }, { "epoch": 0.54, "learning_rate": 4.6143642466440286e-05, "loss": 0.3665, "step": 3101 }, { "epoch": 0.54, "learning_rate": 4.613980931893655e-05, "loss": 0.3622, "step": 3102 }, { "epoch": 0.54, "learning_rate": 4.613597444608377e-05, "loss": 0.3666, "step": 3103 }, { "epoch": 0.54, "learning_rate": 4.613213784823694e-05, "loss": 0.4014, "step": 3104 }, { "epoch": 0.54, "learning_rate": 4.612829952575114e-05, "loss": 0.3649, "step": 3105 }, { "epoch": 0.54, "learning_rate": 4.6124459478981705e-05, "loss": 0.3639, "step": 3106 }, { "epoch": 0.54, "learning_rate": 4.612061770828404e-05, "loss": 0.3534, "step": 3107 }, { "epoch": 0.54, "learning_rate": 4.611677421401376e-05, "loss": 0.3536, "step": 3108 }, { "epoch": 0.54, "learning_rate": 4.6112928996526626e-05, "loss": 0.3571, "step": 3109 }, { "epoch": 0.54, "learning_rate": 4.610908205617857e-05, "loss": 0.3634, "step": 3110 }, { "epoch": 0.54, "learning_rate": 4.610523339332568e-05, "loss": 0.3715, "step": 3111 }, { "epoch": 0.54, "learning_rate": 4.6101383008324185e-05, "loss": 0.369, "step": 3112 }, { "epoch": 0.54, "learning_rate": 4.609753090153049e-05, "loss": 0.3789, "step": 3113 }, { "epoch": 0.54, "learning_rate": 4.609367707330116e-05, "loss": 0.3695, "step": 3114 }, { "epoch": 0.54, "learning_rate": 4.6089821523992916e-05, "loss": 0.3687, "step": 3115 }, { "epoch": 0.54, "learning_rate": 4.608596425396263e-05, "loss": 0.3544, "step": 3116 }, { "epoch": 0.54, "learning_rate": 4.608210526356734e-05, "loss": 0.3817, "step": 3117 }, { "epoch": 0.54, "learning_rate": 4.6078244553164254e-05, "loss": 0.3704, "step": 3118 }, { "epoch": 0.54, "learning_rate": 4.607438212311073e-05, "loss": 0.3663, "step": 3119 }, { "epoch": 0.54, "learning_rate": 4.607051797376427e-05, "loss": 0.3491, "step": 3120 }, { "epoch": 0.54, "learning_rate": 4.6066652105482564e-05, "loss": 0.3606, "step": 3121 }, { "epoch": 0.54, "learning_rate": 4.606278451862344e-05, "loss": 0.3614, "step": 3122 }, { "epoch": 0.54, "learning_rate": 4.6058915213544895e-05, "loss": 0.3599, "step": 3123 }, { "epoch": 0.54, "learning_rate": 4.605504419060508e-05, "loss": 0.3616, "step": 3124 }, { "epoch": 0.54, "learning_rate": 4.60511714501623e-05, "loss": 0.3456, "step": 3125 }, { "epoch": 0.54, "learning_rate": 4.604729699257504e-05, "loss": 0.3685, "step": 3126 }, { "epoch": 0.55, "learning_rate": 4.604342081820192e-05, "loss": 0.3816, "step": 3127 }, { "epoch": 0.55, "learning_rate": 4.6039542927401744e-05, "loss": 0.3597, "step": 3128 }, { "epoch": 0.55, "learning_rate": 4.603566332053343e-05, "loss": 0.3863, "step": 3129 }, { "epoch": 0.55, "learning_rate": 4.603178199795612e-05, "loss": 0.3628, "step": 3130 }, { "epoch": 0.55, "learning_rate": 4.602789896002905e-05, "loss": 0.3645, "step": 3131 }, { "epoch": 0.55, "learning_rate": 4.602401420711165e-05, "loss": 0.3596, "step": 3132 }, { "epoch": 0.55, "learning_rate": 4.602012773956352e-05, "loss": 0.3695, "step": 3133 }, { "epoch": 0.55, "learning_rate": 4.6016239557744384e-05, "loss": 0.3682, "step": 3134 }, { "epoch": 0.55, "learning_rate": 4.6012349662014145e-05, "loss": 0.3462, "step": 3135 }, { "epoch": 0.55, "learning_rate": 4.600845805273287e-05, "loss": 0.3734, "step": 3136 }, { "epoch": 0.55, "learning_rate": 4.6004564730260766e-05, "loss": 0.3505, "step": 3137 }, { "epoch": 0.55, "learning_rate": 4.600066969495822e-05, "loss": 0.3787, "step": 3138 }, { "epoch": 0.55, "learning_rate": 4.5996772947185754e-05, "loss": 0.3761, "step": 3139 }, { "epoch": 0.55, "learning_rate": 4.599287448730407e-05, "loss": 0.3579, "step": 3140 }, { "epoch": 0.55, "learning_rate": 4.598897431567401e-05, "loss": 0.3643, "step": 3141 }, { "epoch": 0.55, "learning_rate": 4.5985072432656614e-05, "loss": 0.3578, "step": 3142 }, { "epoch": 0.55, "learning_rate": 4.598116883861301e-05, "loss": 0.3581, "step": 3143 }, { "epoch": 0.55, "learning_rate": 4.597726353390455e-05, "loss": 0.3514, "step": 3144 }, { "epoch": 0.55, "learning_rate": 4.59733565188927e-05, "loss": 0.3653, "step": 3145 }, { "epoch": 0.55, "learning_rate": 4.5969447793939115e-05, "loss": 0.3647, "step": 3146 }, { "epoch": 0.55, "learning_rate": 4.59655373594056e-05, "loss": 0.3523, "step": 3147 }, { "epoch": 0.55, "learning_rate": 4.596162521565412e-05, "loss": 0.3428, "step": 3148 }, { "epoch": 0.55, "learning_rate": 4.595771136304678e-05, "loss": 0.3517, "step": 3149 }, { "epoch": 0.55, "learning_rate": 4.595379580194586e-05, "loss": 0.3633, "step": 3150 }, { "epoch": 0.55, "learning_rate": 4.594987853271379e-05, "loss": 0.352, "step": 3151 }, { "epoch": 0.55, "learning_rate": 4.5945959555713175e-05, "loss": 0.3682, "step": 3152 }, { "epoch": 0.55, "learning_rate": 4.5942038871306756e-05, "loss": 0.3826, "step": 3153 }, { "epoch": 0.55, "learning_rate": 4.593811647985745e-05, "loss": 0.3554, "step": 3154 }, { "epoch": 0.55, "learning_rate": 4.5934192381728316e-05, "loss": 0.3849, "step": 3155 }, { "epoch": 0.55, "learning_rate": 4.5930266577282576e-05, "loss": 0.3384, "step": 3156 }, { "epoch": 0.55, "learning_rate": 4.592633906688362e-05, "loss": 0.358, "step": 3157 }, { "epoch": 0.55, "learning_rate": 4.592240985089498e-05, "loss": 0.3738, "step": 3158 }, { "epoch": 0.55, "learning_rate": 4.591847892968037e-05, "loss": 0.3719, "step": 3159 }, { "epoch": 0.55, "learning_rate": 4.5914546303603633e-05, "loss": 0.3591, "step": 3160 }, { "epoch": 0.55, "learning_rate": 4.591061197302879e-05, "loss": 0.3738, "step": 3161 }, { "epoch": 0.55, "learning_rate": 4.5906675938319996e-05, "loss": 0.3565, "step": 3162 }, { "epoch": 0.55, "learning_rate": 4.590273819984161e-05, "loss": 0.3548, "step": 3163 }, { "epoch": 0.55, "learning_rate": 4.589879875795809e-05, "loss": 0.3563, "step": 3164 }, { "epoch": 0.55, "learning_rate": 4.589485761303411e-05, "loss": 0.3734, "step": 3165 }, { "epoch": 0.55, "learning_rate": 4.5890914765434445e-05, "loss": 0.3419, "step": 3166 }, { "epoch": 0.55, "learning_rate": 4.588697021552407e-05, "loss": 0.3775, "step": 3167 }, { "epoch": 0.55, "learning_rate": 4.58830239636681e-05, "loss": 0.3562, "step": 3168 }, { "epoch": 0.55, "learning_rate": 4.587907601023181e-05, "loss": 0.3575, "step": 3169 }, { "epoch": 0.55, "learning_rate": 4.587512635558064e-05, "loss": 0.35, "step": 3170 }, { "epoch": 0.55, "learning_rate": 4.587117500008017e-05, "loss": 0.3703, "step": 3171 }, { "epoch": 0.55, "learning_rate": 4.5867221944096144e-05, "loss": 0.381, "step": 3172 }, { "epoch": 0.55, "learning_rate": 4.586326718799448e-05, "loss": 0.3542, "step": 3173 }, { "epoch": 0.55, "learning_rate": 4.585931073214124e-05, "loss": 0.363, "step": 3174 }, { "epoch": 0.55, "learning_rate": 4.5855352576902624e-05, "loss": 0.3569, "step": 3175 }, { "epoch": 0.55, "learning_rate": 4.585139272264503e-05, "loss": 0.3638, "step": 3176 }, { "epoch": 0.55, "learning_rate": 4.584743116973499e-05, "loss": 0.3634, "step": 3177 }, { "epoch": 0.55, "learning_rate": 4.584346791853918e-05, "loss": 0.3641, "step": 3178 }, { "epoch": 0.55, "learning_rate": 4.583950296942448e-05, "loss": 0.352, "step": 3179 }, { "epoch": 0.55, "learning_rate": 4.583553632275786e-05, "loss": 0.3414, "step": 3180 }, { "epoch": 0.55, "learning_rate": 4.5831567978906504e-05, "loss": 0.3708, "step": 3181 }, { "epoch": 0.55, "learning_rate": 4.582759793823773e-05, "loss": 0.3561, "step": 3182 }, { "epoch": 0.55, "learning_rate": 4.5823626201119e-05, "loss": 0.3511, "step": 3183 }, { "epoch": 0.55, "learning_rate": 4.581965276791798e-05, "loss": 0.3904, "step": 3184 }, { "epoch": 0.56, "learning_rate": 4.581567763900243e-05, "loss": 0.3623, "step": 3185 }, { "epoch": 0.56, "learning_rate": 4.5811700814740305e-05, "loss": 0.3587, "step": 3186 }, { "epoch": 0.56, "learning_rate": 4.5807722295499734e-05, "loss": 0.3506, "step": 3187 }, { "epoch": 0.56, "learning_rate": 4.580374208164894e-05, "loss": 0.3771, "step": 3188 }, { "epoch": 0.56, "learning_rate": 4.5799760173556386e-05, "loss": 0.372, "step": 3189 }, { "epoch": 0.56, "learning_rate": 4.579577657159061e-05, "loss": 0.3599, "step": 3190 }, { "epoch": 0.56, "learning_rate": 4.5791791276120353e-05, "loss": 0.3606, "step": 3191 }, { "epoch": 0.56, "learning_rate": 4.578780428751452e-05, "loss": 0.3813, "step": 3192 }, { "epoch": 0.56, "learning_rate": 4.5783815606142145e-05, "loss": 0.3801, "step": 3193 }, { "epoch": 0.56, "learning_rate": 4.577982523237243e-05, "loss": 0.3707, "step": 3194 }, { "epoch": 0.56, "learning_rate": 4.577583316657474e-05, "loss": 0.3625, "step": 3195 }, { "epoch": 0.56, "learning_rate": 4.577183940911859e-05, "loss": 0.3675, "step": 3196 }, { "epoch": 0.56, "learning_rate": 4.576784396037364e-05, "loss": 0.3545, "step": 3197 }, { "epoch": 0.56, "learning_rate": 4.576384682070974e-05, "loss": 0.3741, "step": 3198 }, { "epoch": 0.56, "learning_rate": 4.575984799049686e-05, "loss": 0.3556, "step": 3199 }, { "epoch": 0.56, "learning_rate": 4.575584747010515e-05, "loss": 0.361, "step": 3200 }, { "epoch": 0.56, "learning_rate": 4.5751845259904905e-05, "loss": 0.3709, "step": 3201 }, { "epoch": 0.56, "learning_rate": 4.574784136026659e-05, "loss": 0.3772, "step": 3202 }, { "epoch": 0.56, "learning_rate": 4.57438357715608e-05, "loss": 0.3599, "step": 3203 }, { "epoch": 0.56, "learning_rate": 4.5739828494158305e-05, "loss": 0.3728, "step": 3204 }, { "epoch": 0.56, "learning_rate": 4.573581952843004e-05, "loss": 0.3711, "step": 3205 }, { "epoch": 0.56, "learning_rate": 4.573180887474707e-05, "loss": 0.3714, "step": 3206 }, { "epoch": 0.56, "learning_rate": 4.572779653348065e-05, "loss": 0.3811, "step": 3207 }, { "epoch": 0.56, "learning_rate": 4.5723782505002156e-05, "loss": 0.3594, "step": 3208 }, { "epoch": 0.56, "learning_rate": 4.5719766789683154e-05, "loss": 0.356, "step": 3209 }, { "epoch": 0.56, "learning_rate": 4.571574938789533e-05, "loss": 0.3767, "step": 3210 }, { "epoch": 0.56, "learning_rate": 4.571173030001056e-05, "loss": 0.3388, "step": 3211 }, { "epoch": 0.56, "learning_rate": 4.570770952640085e-05, "loss": 0.3771, "step": 3212 }, { "epoch": 0.56, "learning_rate": 4.570368706743838e-05, "loss": 0.3756, "step": 3213 }, { "epoch": 0.56, "learning_rate": 4.569966292349548e-05, "loss": 0.3518, "step": 3214 }, { "epoch": 0.56, "learning_rate": 4.569563709494464e-05, "loss": 0.3703, "step": 3215 }, { "epoch": 0.56, "learning_rate": 4.569160958215848e-05, "loss": 0.3747, "step": 3216 }, { "epoch": 0.56, "learning_rate": 4.568758038550981e-05, "loss": 0.3634, "step": 3217 }, { "epoch": 0.56, "learning_rate": 4.56835495053716e-05, "loss": 0.3599, "step": 3218 }, { "epoch": 0.56, "learning_rate": 4.5679516942116935e-05, "loss": 0.3607, "step": 3219 }, { "epoch": 0.56, "learning_rate": 4.567548269611909e-05, "loss": 0.3638, "step": 3220 }, { "epoch": 0.56, "learning_rate": 4.567144676775148e-05, "loss": 0.3529, "step": 3221 }, { "epoch": 0.56, "learning_rate": 4.566740915738769e-05, "loss": 0.3562, "step": 3222 }, { "epoch": 0.56, "learning_rate": 4.5663369865401435e-05, "loss": 0.3555, "step": 3223 }, { "epoch": 0.56, "learning_rate": 4.565932889216661e-05, "loss": 0.3682, "step": 3224 }, { "epoch": 0.56, "learning_rate": 4.565528623805727e-05, "loss": 0.3696, "step": 3225 }, { "epoch": 0.56, "learning_rate": 4.56512419034476e-05, "loss": 0.3585, "step": 3226 }, { "epoch": 0.56, "learning_rate": 4.5647195888711954e-05, "loss": 0.3569, "step": 3227 }, { "epoch": 0.56, "learning_rate": 4.564314819422485e-05, "loss": 0.3681, "step": 3228 }, { "epoch": 0.56, "learning_rate": 4.563909882036095e-05, "loss": 0.3455, "step": 3229 }, { "epoch": 0.56, "learning_rate": 4.563504776749507e-05, "loss": 0.3589, "step": 3230 }, { "epoch": 0.56, "learning_rate": 4.56309950360022e-05, "loss": 0.3549, "step": 3231 }, { "epoch": 0.56, "learning_rate": 4.5626940626257446e-05, "loss": 0.3633, "step": 3232 }, { "epoch": 0.56, "learning_rate": 4.5622884538636114e-05, "loss": 0.3604, "step": 3233 }, { "epoch": 0.56, "learning_rate": 4.561882677351365e-05, "loss": 0.3816, "step": 3234 }, { "epoch": 0.56, "learning_rate": 4.561476733126563e-05, "loss": 0.3734, "step": 3235 }, { "epoch": 0.56, "learning_rate": 4.561070621226782e-05, "loss": 0.3547, "step": 3236 }, { "epoch": 0.56, "learning_rate": 4.5606643416896134e-05, "loss": 0.3608, "step": 3237 }, { "epoch": 0.56, "learning_rate": 4.5602578945526625e-05, "loss": 0.3642, "step": 3238 }, { "epoch": 0.56, "learning_rate": 4.5598512798535513e-05, "loss": 0.3593, "step": 3239 }, { "epoch": 0.56, "learning_rate": 4.559444497629917e-05, "loss": 0.3525, "step": 3240 }, { "epoch": 0.56, "learning_rate": 4.559037547919414e-05, "loss": 0.3613, "step": 3241 }, { "epoch": 0.57, "learning_rate": 4.558630430759707e-05, "loss": 0.3613, "step": 3242 }, { "epoch": 0.57, "learning_rate": 4.5582231461884834e-05, "loss": 0.3639, "step": 3243 }, { "epoch": 0.57, "learning_rate": 4.557815694243441e-05, "loss": 0.367, "step": 3244 }, { "epoch": 0.57, "learning_rate": 4.5574080749622945e-05, "loss": 0.354, "step": 3245 }, { "epoch": 0.57, "learning_rate": 4.557000288382774e-05, "loss": 0.3769, "step": 3246 }, { "epoch": 0.57, "learning_rate": 4.556592334542626e-05, "loss": 0.3456, "step": 3247 }, { "epoch": 0.57, "learning_rate": 4.5561842134796115e-05, "loss": 0.3579, "step": 3248 }, { "epoch": 0.57, "learning_rate": 4.555775925231507e-05, "loss": 0.3689, "step": 3249 }, { "epoch": 0.57, "learning_rate": 4.555367469836105e-05, "loss": 0.3582, "step": 3250 }, { "epoch": 0.57, "learning_rate": 4.554958847331214e-05, "loss": 0.3711, "step": 3251 }, { "epoch": 0.57, "learning_rate": 4.5545500577546555e-05, "loss": 0.3555, "step": 3252 }, { "epoch": 0.57, "learning_rate": 4.55414110114427e-05, "loss": 0.3616, "step": 3253 }, { "epoch": 0.57, "learning_rate": 4.553731977537909e-05, "loss": 0.3569, "step": 3254 }, { "epoch": 0.57, "learning_rate": 4.5533226869734444e-05, "loss": 0.3685, "step": 3255 }, { "epoch": 0.57, "learning_rate": 4.55291322948876e-05, "loss": 0.364, "step": 3256 }, { "epoch": 0.57, "learning_rate": 4.552503605121756e-05, "loss": 0.389, "step": 3257 }, { "epoch": 0.57, "learning_rate": 4.55209381391035e-05, "loss": 0.3677, "step": 3258 }, { "epoch": 0.57, "learning_rate": 4.551683855892472e-05, "loss": 0.3558, "step": 3259 }, { "epoch": 0.57, "learning_rate": 4.551273731106069e-05, "loss": 0.353, "step": 3260 }, { "epoch": 0.57, "learning_rate": 4.550863439589102e-05, "loss": 0.3737, "step": 3261 }, { "epoch": 0.57, "learning_rate": 4.550452981379551e-05, "loss": 0.3514, "step": 3262 }, { "epoch": 0.57, "learning_rate": 4.5500423565154076e-05, "loss": 0.3756, "step": 3263 }, { "epoch": 0.57, "learning_rate": 4.549631565034681e-05, "loss": 0.3601, "step": 3264 }, { "epoch": 0.57, "learning_rate": 4.5492206069753943e-05, "loss": 0.3571, "step": 3265 }, { "epoch": 0.57, "learning_rate": 4.548809482375588e-05, "loss": 0.369, "step": 3266 }, { "epoch": 0.57, "learning_rate": 4.5483981912733154e-05, "loss": 0.3551, "step": 3267 }, { "epoch": 0.57, "learning_rate": 4.5479867337066476e-05, "loss": 0.3622, "step": 3268 }, { "epoch": 0.57, "learning_rate": 4.54757510971367e-05, "loss": 0.3693, "step": 3269 }, { "epoch": 0.57, "learning_rate": 4.547163319332484e-05, "loss": 0.3521, "step": 3270 }, { "epoch": 0.57, "learning_rate": 4.5467513626012046e-05, "loss": 0.3629, "step": 3271 }, { "epoch": 0.57, "learning_rate": 4.5463392395579654e-05, "loss": 0.3636, "step": 3272 }, { "epoch": 0.57, "learning_rate": 4.5459269502409125e-05, "loss": 0.3608, "step": 3273 }, { "epoch": 0.57, "learning_rate": 4.5455144946882084e-05, "loss": 0.3558, "step": 3274 }, { "epoch": 0.57, "learning_rate": 4.545101872938031e-05, "loss": 0.3707, "step": 3275 }, { "epoch": 0.57, "learning_rate": 4.5446890850285757e-05, "loss": 0.3643, "step": 3276 }, { "epoch": 0.57, "learning_rate": 4.5442761309980473e-05, "loss": 0.3658, "step": 3277 }, { "epoch": 0.57, "learning_rate": 4.543863010884673e-05, "loss": 0.3654, "step": 3278 }, { "epoch": 0.57, "learning_rate": 4.5434497247266914e-05, "loss": 0.3698, "step": 3279 }, { "epoch": 0.57, "learning_rate": 4.5430362725623575e-05, "loss": 0.3726, "step": 3280 }, { "epoch": 0.57, "learning_rate": 4.542622654429941e-05, "loss": 0.3633, "step": 3281 }, { "epoch": 0.57, "learning_rate": 4.5422088703677284e-05, "loss": 0.3627, "step": 3282 }, { "epoch": 0.57, "learning_rate": 4.54179492041402e-05, "loss": 0.3493, "step": 3283 }, { "epoch": 0.57, "learning_rate": 4.541380804607131e-05, "loss": 0.3591, "step": 3284 }, { "epoch": 0.57, "learning_rate": 4.540966522985395e-05, "loss": 0.3628, "step": 3285 }, { "epoch": 0.57, "learning_rate": 4.540552075587158e-05, "loss": 0.3633, "step": 3286 }, { "epoch": 0.57, "learning_rate": 4.540137462450782e-05, "loss": 0.3609, "step": 3287 }, { "epoch": 0.57, "learning_rate": 4.539722683614645e-05, "loss": 0.3857, "step": 3288 }, { "epoch": 0.57, "learning_rate": 4.5393077391171404e-05, "loss": 0.3652, "step": 3289 }, { "epoch": 0.57, "learning_rate": 4.538892628996677e-05, "loss": 0.3738, "step": 3290 }, { "epoch": 0.57, "learning_rate": 4.5384773532916754e-05, "loss": 0.3602, "step": 3291 }, { "epoch": 0.57, "learning_rate": 4.538061912040579e-05, "loss": 0.3681, "step": 3292 }, { "epoch": 0.57, "learning_rate": 4.537646305281839e-05, "loss": 0.3544, "step": 3293 }, { "epoch": 0.57, "learning_rate": 4.5372305330539266e-05, "loss": 0.3653, "step": 3294 }, { "epoch": 0.57, "learning_rate": 4.536814595395326e-05, "loss": 0.3625, "step": 3295 }, { "epoch": 0.57, "learning_rate": 4.536398492344537e-05, "loss": 0.368, "step": 3296 }, { "epoch": 0.57, "learning_rate": 4.535982223940077e-05, "loss": 0.363, "step": 3297 }, { "epoch": 0.57, "learning_rate": 4.535565790220475e-05, "loss": 0.3554, "step": 3298 }, { "epoch": 0.58, "learning_rate": 4.5351491912242785e-05, "loss": 0.3775, "step": 3299 }, { "epoch": 0.58, "learning_rate": 4.534732426990047e-05, "loss": 0.375, "step": 3300 }, { "epoch": 0.58, "learning_rate": 4.53431549755636e-05, "loss": 0.3806, "step": 3301 }, { "epoch": 0.58, "learning_rate": 4.5338984029618076e-05, "loss": 0.3789, "step": 3302 }, { "epoch": 0.58, "learning_rate": 4.533481143244998e-05, "loss": 0.382, "step": 3303 }, { "epoch": 0.58, "learning_rate": 4.533063718444554e-05, "loss": 0.3649, "step": 3304 }, { "epoch": 0.58, "learning_rate": 4.532646128599112e-05, "loss": 0.3713, "step": 3305 }, { "epoch": 0.58, "learning_rate": 4.532228373747327e-05, "loss": 0.3717, "step": 3306 }, { "epoch": 0.58, "learning_rate": 4.531810453927868e-05, "loss": 0.3409, "step": 3307 }, { "epoch": 0.58, "learning_rate": 4.531392369179416e-05, "loss": 0.3464, "step": 3308 }, { "epoch": 0.58, "learning_rate": 4.5309741195406716e-05, "loss": 0.366, "step": 3309 }, { "epoch": 0.58, "learning_rate": 4.53055570505035e-05, "loss": 0.3682, "step": 3310 }, { "epoch": 0.58, "learning_rate": 4.5301371257471785e-05, "loss": 0.362, "step": 3311 }, { "epoch": 0.58, "learning_rate": 4.529718381669905e-05, "loss": 0.3537, "step": 3312 }, { "epoch": 0.58, "learning_rate": 4.529299472857286e-05, "loss": 0.3632, "step": 3313 }, { "epoch": 0.58, "learning_rate": 4.528880399348099e-05, "loss": 0.3575, "step": 3314 }, { "epoch": 0.58, "learning_rate": 4.528461161181134e-05, "loss": 0.3551, "step": 3315 }, { "epoch": 0.58, "learning_rate": 4.528041758395196e-05, "loss": 0.3716, "step": 3316 }, { "epoch": 0.58, "learning_rate": 4.5276221910291085e-05, "loss": 0.3725, "step": 3317 }, { "epoch": 0.58, "learning_rate": 4.5272024591217046e-05, "loss": 0.341, "step": 3318 }, { "epoch": 0.58, "learning_rate": 4.5267825627118375e-05, "loss": 0.3613, "step": 3319 }, { "epoch": 0.58, "learning_rate": 4.526362501838374e-05, "loss": 0.3498, "step": 3320 }, { "epoch": 0.58, "learning_rate": 4.525942276540195e-05, "loss": 0.3523, "step": 3321 }, { "epoch": 0.58, "learning_rate": 4.5255218868561986e-05, "loss": 0.3586, "step": 3322 }, { "epoch": 0.58, "learning_rate": 4.5251013328252975e-05, "loss": 0.368, "step": 3323 }, { "epoch": 0.58, "learning_rate": 4.5246806144864174e-05, "loss": 0.3812, "step": 3324 }, { "epoch": 0.58, "learning_rate": 4.524259731878503e-05, "loss": 0.36, "step": 3325 }, { "epoch": 0.58, "learning_rate": 4.523838685040511e-05, "loss": 0.3568, "step": 3326 }, { "epoch": 0.58, "learning_rate": 4.523417474011416e-05, "loss": 0.3638, "step": 3327 }, { "epoch": 0.58, "learning_rate": 4.522996098830205e-05, "loss": 0.3749, "step": 3328 }, { "epoch": 0.58, "learning_rate": 4.522574559535882e-05, "loss": 0.375, "step": 3329 }, { "epoch": 0.58, "learning_rate": 4.522152856167466e-05, "loss": 0.3705, "step": 3330 }, { "epoch": 0.58, "learning_rate": 4.52173098876399e-05, "loss": 0.3855, "step": 3331 }, { "epoch": 0.58, "learning_rate": 4.5213089573645046e-05, "loss": 0.3619, "step": 3332 }, { "epoch": 0.58, "learning_rate": 4.520886762008074e-05, "loss": 0.3664, "step": 3333 }, { "epoch": 0.58, "learning_rate": 4.520464402733777e-05, "loss": 0.3551, "step": 3334 }, { "epoch": 0.58, "learning_rate": 4.5200418795807076e-05, "loss": 0.3514, "step": 3335 }, { "epoch": 0.58, "learning_rate": 4.519619192587977e-05, "loss": 0.3515, "step": 3336 }, { "epoch": 0.58, "learning_rate": 4.5191963417947106e-05, "loss": 0.3679, "step": 3337 }, { "epoch": 0.58, "learning_rate": 4.518773327240047e-05, "loss": 0.3527, "step": 3338 }, { "epoch": 0.58, "learning_rate": 4.5183501489631415e-05, "loss": 0.3594, "step": 3339 }, { "epoch": 0.58, "learning_rate": 4.5179268070031656e-05, "loss": 0.3636, "step": 3340 }, { "epoch": 0.58, "learning_rate": 4.517503301399304e-05, "loss": 0.3411, "step": 3341 }, { "epoch": 0.58, "learning_rate": 4.5170796321907594e-05, "loss": 0.3662, "step": 3342 }, { "epoch": 0.58, "learning_rate": 4.516655799416746e-05, "loss": 0.3556, "step": 3343 }, { "epoch": 0.58, "learning_rate": 4.5162318031164946e-05, "loss": 0.3587, "step": 3344 }, { "epoch": 0.58, "learning_rate": 4.5158076433292526e-05, "loss": 0.3571, "step": 3345 }, { "epoch": 0.58, "learning_rate": 4.51538332009428e-05, "loss": 0.3618, "step": 3346 }, { "epoch": 0.58, "learning_rate": 4.514958833450855e-05, "loss": 0.3557, "step": 3347 }, { "epoch": 0.58, "learning_rate": 4.514534183438268e-05, "loss": 0.355, "step": 3348 }, { "epoch": 0.58, "learning_rate": 4.514109370095826e-05, "loss": 0.357, "step": 3349 }, { "epoch": 0.58, "learning_rate": 4.51368439346285e-05, "loss": 0.3414, "step": 3350 }, { "epoch": 0.58, "learning_rate": 4.513259253578678e-05, "loss": 0.3664, "step": 3351 }, { "epoch": 0.58, "learning_rate": 4.5128339504826615e-05, "loss": 0.3504, "step": 3352 }, { "epoch": 0.58, "learning_rate": 4.5124084842141684e-05, "loss": 0.3548, "step": 3353 }, { "epoch": 0.58, "learning_rate": 4.51198285481258e-05, "loss": 0.3592, "step": 3354 }, { "epoch": 0.58, "learning_rate": 4.511557062317295e-05, "loss": 0.3844, "step": 3355 }, { "epoch": 0.58, "learning_rate": 4.5111311067677234e-05, "loss": 0.3637, "step": 3356 }, { "epoch": 0.59, "learning_rate": 4.510704988203295e-05, "loss": 0.3537, "step": 3357 }, { "epoch": 0.59, "learning_rate": 4.5102787066634526e-05, "loss": 0.3568, "step": 3358 }, { "epoch": 0.59, "learning_rate": 4.5098522621876515e-05, "loss": 0.3824, "step": 3359 }, { "epoch": 0.59, "learning_rate": 4.509425654815368e-05, "loss": 0.3572, "step": 3360 }, { "epoch": 0.59, "learning_rate": 4.508998884586087e-05, "loss": 0.3516, "step": 3361 }, { "epoch": 0.59, "learning_rate": 4.5085719515393114e-05, "loss": 0.3669, "step": 3362 }, { "epoch": 0.59, "learning_rate": 4.5081448557145614e-05, "loss": 0.3592, "step": 3363 }, { "epoch": 0.59, "learning_rate": 4.5077175971513696e-05, "loss": 0.3376, "step": 3364 }, { "epoch": 0.59, "learning_rate": 4.507290175889283e-05, "loss": 0.3617, "step": 3365 }, { "epoch": 0.59, "learning_rate": 4.506862591967865e-05, "loss": 0.3654, "step": 3366 }, { "epoch": 0.59, "learning_rate": 4.5064348454266954e-05, "loss": 0.3731, "step": 3367 }, { "epoch": 0.59, "learning_rate": 4.506006936305365e-05, "loss": 0.3531, "step": 3368 }, { "epoch": 0.59, "learning_rate": 4.505578864643485e-05, "loss": 0.3646, "step": 3369 }, { "epoch": 0.59, "learning_rate": 4.505150630480676e-05, "loss": 0.3692, "step": 3370 }, { "epoch": 0.59, "learning_rate": 4.504722233856579e-05, "loss": 0.3598, "step": 3371 }, { "epoch": 0.59, "learning_rate": 4.504293674810846e-05, "loss": 0.3677, "step": 3372 }, { "epoch": 0.59, "learning_rate": 4.503864953383146e-05, "loss": 0.3622, "step": 3373 }, { "epoch": 0.59, "learning_rate": 4.503436069613162e-05, "loss": 0.374, "step": 3374 }, { "epoch": 0.59, "learning_rate": 4.503007023540594e-05, "loss": 0.3579, "step": 3375 }, { "epoch": 0.59, "learning_rate": 4.502577815205154e-05, "loss": 0.3552, "step": 3376 }, { "epoch": 0.59, "learning_rate": 4.502148444646571e-05, "loss": 0.3632, "step": 3377 }, { "epoch": 0.59, "learning_rate": 4.5017189119045896e-05, "loss": 0.368, "step": 3378 }, { "epoch": 0.59, "learning_rate": 4.501289217018968e-05, "loss": 0.3575, "step": 3379 }, { "epoch": 0.59, "learning_rate": 4.500859360029479e-05, "loss": 0.3628, "step": 3380 }, { "epoch": 0.59, "learning_rate": 4.500429340975913e-05, "loss": 0.3576, "step": 3381 }, { "epoch": 0.59, "learning_rate": 4.499999159898071e-05, "loss": 0.3655, "step": 3382 }, { "epoch": 0.59, "learning_rate": 4.4995688168357747e-05, "loss": 0.357, "step": 3383 }, { "epoch": 0.59, "learning_rate": 4.499138311828856e-05, "loss": 0.3551, "step": 3384 }, { "epoch": 0.59, "learning_rate": 4.498707644917164e-05, "loss": 0.3583, "step": 3385 }, { "epoch": 0.59, "learning_rate": 4.498276816140562e-05, "loss": 0.3982, "step": 3386 }, { "epoch": 0.59, "learning_rate": 4.497845825538929e-05, "loss": 0.3542, "step": 3387 }, { "epoch": 0.59, "learning_rate": 4.497414673152159e-05, "loss": 0.378, "step": 3388 }, { "epoch": 0.59, "learning_rate": 4.49698335902016e-05, "loss": 0.3449, "step": 3389 }, { "epoch": 0.59, "learning_rate": 4.496551883182856e-05, "loss": 0.3585, "step": 3390 }, { "epoch": 0.59, "learning_rate": 4.496120245680185e-05, "loss": 0.3512, "step": 3391 }, { "epoch": 0.59, "learning_rate": 4.4956884465521e-05, "loss": 0.3505, "step": 3392 }, { "epoch": 0.59, "learning_rate": 4.495256485838571e-05, "loss": 0.3676, "step": 3393 }, { "epoch": 0.59, "learning_rate": 4.4948243635795794e-05, "loss": 0.3654, "step": 3394 }, { "epoch": 0.59, "learning_rate": 4.4943920798151255e-05, "loss": 0.3609, "step": 3395 }, { "epoch": 0.59, "learning_rate": 4.493959634585222e-05, "loss": 0.365, "step": 3396 }, { "epoch": 0.59, "learning_rate": 4.4935270279298965e-05, "loss": 0.3662, "step": 3397 }, { "epoch": 0.59, "learning_rate": 4.493094259889193e-05, "loss": 0.3581, "step": 3398 }, { "epoch": 0.59, "learning_rate": 4.4926613305031696e-05, "loss": 0.3402, "step": 3399 }, { "epoch": 0.59, "learning_rate": 4.492228239811898e-05, "loss": 0.3484, "step": 3400 }, { "epoch": 0.59, "learning_rate": 4.4917949878554675e-05, "loss": 0.3794, "step": 3401 }, { "epoch": 0.59, "learning_rate": 4.491361574673982e-05, "loss": 0.3379, "step": 3402 }, { "epoch": 0.59, "learning_rate": 4.490928000307557e-05, "loss": 0.3461, "step": 3403 }, { "epoch": 0.59, "learning_rate": 4.4904942647963256e-05, "loss": 0.3654, "step": 3404 }, { "epoch": 0.59, "learning_rate": 4.490060368180438e-05, "loss": 0.3582, "step": 3405 }, { "epoch": 0.59, "learning_rate": 4.489626310500054e-05, "loss": 0.369, "step": 3406 }, { "epoch": 0.59, "learning_rate": 4.489192091795352e-05, "loss": 0.3601, "step": 3407 }, { "epoch": 0.59, "learning_rate": 4.4887577121065244e-05, "loss": 0.3493, "step": 3408 }, { "epoch": 0.59, "learning_rate": 4.48832317147378e-05, "loss": 0.3627, "step": 3409 }, { "epoch": 0.59, "learning_rate": 4.487888469937339e-05, "loss": 0.3454, "step": 3410 }, { "epoch": 0.59, "learning_rate": 4.487453607537438e-05, "loss": 0.3714, "step": 3411 }, { "epoch": 0.59, "learning_rate": 4.487018584314331e-05, "loss": 0.3663, "step": 3412 }, { "epoch": 0.59, "learning_rate": 4.4865834003082854e-05, "loss": 0.3443, "step": 3413 }, { "epoch": 0.6, "learning_rate": 4.4861480555595794e-05, "loss": 0.3459, "step": 3414 }, { "epoch": 0.6, "learning_rate": 4.485712550108513e-05, "loss": 0.3689, "step": 3415 }, { "epoch": 0.6, "learning_rate": 4.485276883995397e-05, "loss": 0.3575, "step": 3416 }, { "epoch": 0.6, "learning_rate": 4.4848410572605565e-05, "loss": 0.3418, "step": 3417 }, { "epoch": 0.6, "learning_rate": 4.484405069944334e-05, "loss": 0.351, "step": 3418 }, { "epoch": 0.6, "learning_rate": 4.483968922087085e-05, "loss": 0.3654, "step": 3419 }, { "epoch": 0.6, "learning_rate": 4.483532613729181e-05, "loss": 0.3545, "step": 3420 }, { "epoch": 0.6, "learning_rate": 4.483096144911007e-05, "loss": 0.3612, "step": 3421 }, { "epoch": 0.6, "learning_rate": 4.482659515672964e-05, "loss": 0.3594, "step": 3422 }, { "epoch": 0.6, "learning_rate": 4.482222726055468e-05, "loss": 0.3668, "step": 3423 }, { "epoch": 0.6, "learning_rate": 4.481785776098949e-05, "loss": 0.3628, "step": 3424 }, { "epoch": 0.6, "learning_rate": 4.481348665843852e-05, "loss": 0.3692, "step": 3425 }, { "epoch": 0.6, "learning_rate": 4.480911395330638e-05, "loss": 0.3569, "step": 3426 }, { "epoch": 0.6, "learning_rate": 4.480473964599782e-05, "loss": 0.343, "step": 3427 }, { "epoch": 0.6, "learning_rate": 4.480036373691772e-05, "loss": 0.3878, "step": 3428 }, { "epoch": 0.6, "learning_rate": 4.4795986226471136e-05, "loss": 0.3596, "step": 3429 }, { "epoch": 0.6, "learning_rate": 4.479160711506326e-05, "loss": 0.3579, "step": 3430 }, { "epoch": 0.6, "learning_rate": 4.4787226403099444e-05, "loss": 0.3598, "step": 3431 }, { "epoch": 0.6, "learning_rate": 4.478284409098516e-05, "loss": 0.383, "step": 3432 }, { "epoch": 0.6, "learning_rate": 4.477846017912606e-05, "loss": 0.365, "step": 3433 }, { "epoch": 0.6, "learning_rate": 4.4774074667927916e-05, "loss": 0.3601, "step": 3434 }, { "epoch": 0.6, "learning_rate": 4.476968755779668e-05, "loss": 0.3632, "step": 3435 }, { "epoch": 0.6, "learning_rate": 4.476529884913843e-05, "loss": 0.352, "step": 3436 }, { "epoch": 0.6, "learning_rate": 4.4760908542359386e-05, "loss": 0.3488, "step": 3437 }, { "epoch": 0.6, "learning_rate": 4.4756516637865935e-05, "loss": 0.3848, "step": 3438 }, { "epoch": 0.6, "learning_rate": 4.47521231360646e-05, "loss": 0.3661, "step": 3439 }, { "epoch": 0.6, "learning_rate": 4.474772803736205e-05, "loss": 0.3772, "step": 3440 }, { "epoch": 0.6, "learning_rate": 4.474333134216512e-05, "loss": 0.3543, "step": 3441 }, { "epoch": 0.6, "learning_rate": 4.473893305088077e-05, "loss": 0.3568, "step": 3442 }, { "epoch": 0.6, "learning_rate": 4.4734533163916115e-05, "loss": 0.3628, "step": 3443 }, { "epoch": 0.6, "learning_rate": 4.473013168167843e-05, "loss": 0.3421, "step": 3444 }, { "epoch": 0.6, "learning_rate": 4.4725728604575114e-05, "loss": 0.3654, "step": 3445 }, { "epoch": 0.6, "learning_rate": 4.4721323933013735e-05, "loss": 0.3754, "step": 3446 }, { "epoch": 0.6, "learning_rate": 4.4716917667402005e-05, "loss": 0.3591, "step": 3447 }, { "epoch": 0.6, "learning_rate": 4.471250980814778e-05, "loss": 0.3627, "step": 3448 }, { "epoch": 0.6, "learning_rate": 4.470810035565905e-05, "loss": 0.364, "step": 3449 }, { "epoch": 0.6, "learning_rate": 4.470368931034397e-05, "loss": 0.3533, "step": 3450 }, { "epoch": 0.6, "learning_rate": 4.469927667261085e-05, "loss": 0.3512, "step": 3451 }, { "epoch": 0.6, "learning_rate": 4.4694862442868114e-05, "loss": 0.3714, "step": 3452 }, { "epoch": 0.6, "learning_rate": 4.469044662152437e-05, "loss": 0.3452, "step": 3453 }, { "epoch": 0.6, "learning_rate": 4.468602920898836e-05, "loss": 0.359, "step": 3454 }, { "epoch": 0.6, "learning_rate": 4.468161020566896e-05, "loss": 0.3393, "step": 3455 }, { "epoch": 0.6, "learning_rate": 4.4677189611975216e-05, "loss": 0.3442, "step": 3456 }, { "epoch": 0.6, "learning_rate": 4.4672767428316304e-05, "loss": 0.3601, "step": 3457 }, { "epoch": 0.6, "learning_rate": 4.466834365510155e-05, "loss": 0.3752, "step": 3458 }, { "epoch": 0.6, "learning_rate": 4.466391829274043e-05, "loss": 0.3564, "step": 3459 }, { "epoch": 0.6, "learning_rate": 4.465949134164257e-05, "loss": 0.3587, "step": 3460 }, { "epoch": 0.6, "learning_rate": 4.4655062802217737e-05, "loss": 0.3567, "step": 3461 }, { "epoch": 0.6, "learning_rate": 4.4650632674875855e-05, "loss": 0.3655, "step": 3462 }, { "epoch": 0.6, "learning_rate": 4.464620096002698e-05, "loss": 0.3608, "step": 3463 }, { "epoch": 0.6, "learning_rate": 4.464176765808133e-05, "loss": 0.3666, "step": 3464 }, { "epoch": 0.6, "learning_rate": 4.463733276944926e-05, "loss": 0.3802, "step": 3465 }, { "epoch": 0.6, "learning_rate": 4.463289629454127e-05, "loss": 0.3422, "step": 3466 }, { "epoch": 0.6, "learning_rate": 4.462845823376801e-05, "loss": 0.3768, "step": 3467 }, { "epoch": 0.6, "learning_rate": 4.462401858754029e-05, "loss": 0.3558, "step": 3468 }, { "epoch": 0.6, "learning_rate": 4.461957735626906e-05, "loss": 0.3536, "step": 3469 }, { "epoch": 0.6, "learning_rate": 4.4615134540365386e-05, "loss": 0.3497, "step": 3470 }, { "epoch": 0.6, "learning_rate": 4.461069014024053e-05, "loss": 0.3636, "step": 3471 }, { "epoch": 0.61, "learning_rate": 4.460624415630586e-05, "loss": 0.364, "step": 3472 }, { "epoch": 0.61, "learning_rate": 4.460179658897293e-05, "loss": 0.3793, "step": 3473 }, { "epoch": 0.61, "learning_rate": 4.4597347438653396e-05, "loss": 0.3674, "step": 3474 }, { "epoch": 0.61, "learning_rate": 4.4592896705759094e-05, "loss": 0.3447, "step": 3475 }, { "epoch": 0.61, "learning_rate": 4.4588444390701985e-05, "loss": 0.3472, "step": 3476 }, { "epoch": 0.61, "learning_rate": 4.458399049389421e-05, "loss": 0.3531, "step": 3477 }, { "epoch": 0.61, "learning_rate": 4.457953501574801e-05, "loss": 0.3697, "step": 3478 }, { "epoch": 0.61, "learning_rate": 4.457507795667581e-05, "loss": 0.3688, "step": 3479 }, { "epoch": 0.61, "learning_rate": 4.4570619317090153e-05, "loss": 0.3589, "step": 3480 }, { "epoch": 0.61, "learning_rate": 4.4566159097403756e-05, "loss": 0.356, "step": 3481 }, { "epoch": 0.61, "learning_rate": 4.456169729802946e-05, "loss": 0.3514, "step": 3482 }, { "epoch": 0.61, "learning_rate": 4.455723391938026e-05, "loss": 0.3631, "step": 3483 }, { "epoch": 0.61, "learning_rate": 4.45527689618693e-05, "loss": 0.3698, "step": 3484 }, { "epoch": 0.61, "learning_rate": 4.454830242590988e-05, "loss": 0.3452, "step": 3485 }, { "epoch": 0.61, "learning_rate": 4.454383431191541e-05, "loss": 0.3584, "step": 3486 }, { "epoch": 0.61, "learning_rate": 4.453936462029949e-05, "loss": 0.3584, "step": 3487 }, { "epoch": 0.61, "learning_rate": 4.4534893351475834e-05, "loss": 0.3556, "step": 3488 }, { "epoch": 0.61, "learning_rate": 4.453042050585833e-05, "loss": 0.3503, "step": 3489 }, { "epoch": 0.61, "learning_rate": 4.4525946083860984e-05, "loss": 0.367, "step": 3490 }, { "epoch": 0.61, "learning_rate": 4.4521470085897965e-05, "loss": 0.355, "step": 3491 }, { "epoch": 0.61, "learning_rate": 4.451699251238358e-05, "loss": 0.39, "step": 3492 }, { "epoch": 0.61, "learning_rate": 4.451251336373227e-05, "loss": 0.3728, "step": 3493 }, { "epoch": 0.61, "learning_rate": 4.450803264035867e-05, "loss": 0.3591, "step": 3494 }, { "epoch": 0.61, "learning_rate": 4.45035503426775e-05, "loss": 0.3515, "step": 3495 }, { "epoch": 0.61, "learning_rate": 4.449906647110367e-05, "loss": 0.3497, "step": 3496 }, { "epoch": 0.61, "learning_rate": 4.44945810260522e-05, "loss": 0.3586, "step": 3497 }, { "epoch": 0.61, "learning_rate": 4.449009400793829e-05, "loss": 0.3482, "step": 3498 }, { "epoch": 0.61, "learning_rate": 4.448560541717726e-05, "loss": 0.3501, "step": 3499 }, { "epoch": 0.61, "learning_rate": 4.4481115254184604e-05, "loss": 0.3443, "step": 3500 }, { "epoch": 0.61, "learning_rate": 4.4476623519375914e-05, "loss": 0.3461, "step": 3501 }, { "epoch": 0.61, "learning_rate": 4.447213021316699e-05, "loss": 0.3537, "step": 3502 }, { "epoch": 0.61, "learning_rate": 4.446763533597371e-05, "loss": 0.3513, "step": 3503 }, { "epoch": 0.61, "learning_rate": 4.446313888821216e-05, "loss": 0.3569, "step": 3504 }, { "epoch": 0.61, "learning_rate": 4.445864087029852e-05, "loss": 0.3631, "step": 3505 }, { "epoch": 0.61, "learning_rate": 4.445414128264916e-05, "loss": 0.349, "step": 3506 }, { "epoch": 0.61, "learning_rate": 4.444964012568056e-05, "loss": 0.359, "step": 3507 }, { "epoch": 0.61, "learning_rate": 4.4445137399809363e-05, "loss": 0.354, "step": 3508 }, { "epoch": 0.61, "learning_rate": 4.444063310545234e-05, "loss": 0.3574, "step": 3509 }, { "epoch": 0.61, "learning_rate": 4.4436127243026445e-05, "loss": 0.3409, "step": 3510 }, { "epoch": 0.61, "learning_rate": 4.443161981294873e-05, "loss": 0.3644, "step": 3511 }, { "epoch": 0.61, "learning_rate": 4.442711081563643e-05, "loss": 0.3657, "step": 3512 }, { "epoch": 0.61, "learning_rate": 4.44226002515069e-05, "loss": 0.3716, "step": 3513 }, { "epoch": 0.61, "learning_rate": 4.4418088120977645e-05, "loss": 0.3647, "step": 3514 }, { "epoch": 0.61, "learning_rate": 4.4413574424466336e-05, "loss": 0.3501, "step": 3515 }, { "epoch": 0.61, "learning_rate": 4.440905916239076e-05, "loss": 0.3657, "step": 3516 }, { "epoch": 0.61, "learning_rate": 4.4404542335168863e-05, "loss": 0.3556, "step": 3517 }, { "epoch": 0.61, "learning_rate": 4.440002394321874e-05, "loss": 0.3669, "step": 3518 }, { "epoch": 0.61, "learning_rate": 4.439550398695862e-05, "loss": 0.3728, "step": 3519 }, { "epoch": 0.61, "learning_rate": 4.4390982466806884e-05, "loss": 0.3485, "step": 3520 }, { "epoch": 0.61, "learning_rate": 4.438645938318205e-05, "loss": 0.3602, "step": 3521 }, { "epoch": 0.61, "learning_rate": 4.438193473650279e-05, "loss": 0.368, "step": 3522 }, { "epoch": 0.61, "learning_rate": 4.437740852718793e-05, "loss": 0.3881, "step": 3523 }, { "epoch": 0.61, "learning_rate": 4.43728807556564e-05, "loss": 0.3759, "step": 3524 }, { "epoch": 0.61, "learning_rate": 4.436835142232733e-05, "loss": 0.3621, "step": 3525 }, { "epoch": 0.61, "learning_rate": 4.4363820527619945e-05, "loss": 0.3479, "step": 3526 }, { "epoch": 0.61, "learning_rate": 4.435928807195365e-05, "loss": 0.3657, "step": 3527 }, { "epoch": 0.61, "learning_rate": 4.4354754055747966e-05, "loss": 0.3491, "step": 3528 }, { "epoch": 0.62, "learning_rate": 4.4350218479422597e-05, "loss": 0.3393, "step": 3529 }, { "epoch": 0.62, "learning_rate": 4.434568134339735e-05, "loss": 0.3606, "step": 3530 }, { "epoch": 0.62, "learning_rate": 4.43411426480922e-05, "loss": 0.3759, "step": 3531 }, { "epoch": 0.62, "learning_rate": 4.433660239392726e-05, "loss": 0.3543, "step": 3532 }, { "epoch": 0.62, "learning_rate": 4.433206058132278e-05, "loss": 0.3582, "step": 3533 }, { "epoch": 0.62, "learning_rate": 4.432751721069918e-05, "loss": 0.3643, "step": 3534 }, { "epoch": 0.62, "learning_rate": 4.432297228247698e-05, "loss": 0.3664, "step": 3535 }, { "epoch": 0.62, "learning_rate": 4.4318425797076896e-05, "loss": 0.3664, "step": 3536 }, { "epoch": 0.62, "learning_rate": 4.4313877754919754e-05, "loss": 0.3648, "step": 3537 }, { "epoch": 0.62, "learning_rate": 4.4309328156426524e-05, "loss": 0.3773, "step": 3538 }, { "epoch": 0.62, "learning_rate": 4.430477700201833e-05, "loss": 0.3563, "step": 3539 }, { "epoch": 0.62, "learning_rate": 4.430022429211645e-05, "loss": 0.3617, "step": 3540 }, { "epoch": 0.62, "learning_rate": 4.429567002714229e-05, "loss": 0.3538, "step": 3541 }, { "epoch": 0.62, "learning_rate": 4.42911142075174e-05, "loss": 0.3658, "step": 3542 }, { "epoch": 0.62, "learning_rate": 4.428655683366348e-05, "loss": 0.3649, "step": 3543 }, { "epoch": 0.62, "learning_rate": 4.428199790600238e-05, "loss": 0.3602, "step": 3544 }, { "epoch": 0.62, "learning_rate": 4.427743742495607e-05, "loss": 0.3667, "step": 3545 }, { "epoch": 0.62, "learning_rate": 4.42728753909467e-05, "loss": 0.3448, "step": 3546 }, { "epoch": 0.62, "learning_rate": 4.426831180439652e-05, "loss": 0.3568, "step": 3547 }, { "epoch": 0.62, "learning_rate": 4.4263746665727964e-05, "loss": 0.3413, "step": 3548 }, { "epoch": 0.62, "learning_rate": 4.4259179975363596e-05, "loss": 0.3418, "step": 3549 }, { "epoch": 0.62, "learning_rate": 4.4254611733726114e-05, "loss": 0.3694, "step": 3550 }, { "epoch": 0.62, "learning_rate": 4.425004194123836e-05, "loss": 0.3566, "step": 3551 }, { "epoch": 0.62, "learning_rate": 4.424547059832334e-05, "loss": 0.3665, "step": 3552 }, { "epoch": 0.62, "learning_rate": 4.424089770540418e-05, "loss": 0.3506, "step": 3553 }, { "epoch": 0.62, "learning_rate": 4.423632326290416e-05, "loss": 0.3487, "step": 3554 }, { "epoch": 0.62, "learning_rate": 4.4231747271246696e-05, "loss": 0.3557, "step": 3555 }, { "epoch": 0.62, "learning_rate": 4.422716973085537e-05, "loss": 0.3606, "step": 3556 }, { "epoch": 0.62, "learning_rate": 4.422259064215388e-05, "loss": 0.3663, "step": 3557 }, { "epoch": 0.62, "learning_rate": 4.421801000556608e-05, "loss": 0.3527, "step": 3558 }, { "epoch": 0.62, "learning_rate": 4.421342782151596e-05, "loss": 0.3486, "step": 3559 }, { "epoch": 0.62, "learning_rate": 4.420884409042766e-05, "loss": 0.3611, "step": 3560 }, { "epoch": 0.62, "learning_rate": 4.420425881272548e-05, "loss": 0.3525, "step": 3561 }, { "epoch": 0.62, "learning_rate": 4.419967198883382e-05, "loss": 0.3453, "step": 3562 }, { "epoch": 0.62, "learning_rate": 4.419508361917727e-05, "loss": 0.3438, "step": 3563 }, { "epoch": 0.62, "learning_rate": 4.4190493704180526e-05, "loss": 0.3461, "step": 3564 }, { "epoch": 0.62, "learning_rate": 4.4185902244268444e-05, "loss": 0.3499, "step": 3565 }, { "epoch": 0.62, "learning_rate": 4.418130923986603e-05, "loss": 0.3619, "step": 3566 }, { "epoch": 0.62, "learning_rate": 4.417671469139842e-05, "loss": 0.3874, "step": 3567 }, { "epoch": 0.62, "learning_rate": 4.4172118599290894e-05, "loss": 0.3564, "step": 3568 }, { "epoch": 0.62, "learning_rate": 4.416752096396888e-05, "loss": 0.3644, "step": 3569 }, { "epoch": 0.62, "learning_rate": 4.4162921785857945e-05, "loss": 0.3689, "step": 3570 }, { "epoch": 0.62, "learning_rate": 4.4158321065383806e-05, "loss": 0.3652, "step": 3571 }, { "epoch": 0.62, "learning_rate": 4.415371880297232e-05, "loss": 0.3548, "step": 3572 }, { "epoch": 0.62, "learning_rate": 4.414911499904948e-05, "loss": 0.3525, "step": 3573 }, { "epoch": 0.62, "learning_rate": 4.414450965404142e-05, "loss": 0.3547, "step": 3574 }, { "epoch": 0.62, "learning_rate": 4.413990276837443e-05, "loss": 0.3491, "step": 3575 }, { "epoch": 0.62, "learning_rate": 4.413529434247494e-05, "loss": 0.3559, "step": 3576 }, { "epoch": 0.62, "learning_rate": 4.41306843767695e-05, "loss": 0.3485, "step": 3577 }, { "epoch": 0.62, "learning_rate": 4.412607287168484e-05, "loss": 0.3476, "step": 3578 }, { "epoch": 0.62, "learning_rate": 4.41214598276478e-05, "loss": 0.3364, "step": 3579 }, { "epoch": 0.62, "learning_rate": 4.4116845245085386e-05, "loss": 0.3515, "step": 3580 }, { "epoch": 0.62, "learning_rate": 4.411222912442472e-05, "loss": 0.3538, "step": 3581 }, { "epoch": 0.62, "learning_rate": 4.41076114660931e-05, "loss": 0.3584, "step": 3582 }, { "epoch": 0.62, "learning_rate": 4.410299227051795e-05, "loss": 0.3613, "step": 3583 }, { "epoch": 0.62, "learning_rate": 4.4098371538126806e-05, "loss": 0.363, "step": 3584 }, { "epoch": 0.62, "learning_rate": 4.40937492693474e-05, "loss": 0.3668, "step": 3585 }, { "epoch": 0.63, "learning_rate": 4.408912546460757e-05, "loss": 0.3699, "step": 3586 }, { "epoch": 0.63, "learning_rate": 4.4084500124335324e-05, "loss": 0.3419, "step": 3587 }, { "epoch": 0.63, "learning_rate": 4.407987324895878e-05, "loss": 0.3645, "step": 3588 }, { "epoch": 0.63, "learning_rate": 4.4075244838906216e-05, "loss": 0.3687, "step": 3589 }, { "epoch": 0.63, "learning_rate": 4.4070614894606054e-05, "loss": 0.3373, "step": 3590 }, { "epoch": 0.63, "learning_rate": 4.406598341648685e-05, "loss": 0.3527, "step": 3591 }, { "epoch": 0.63, "learning_rate": 4.4061350404977295e-05, "loss": 0.3626, "step": 3592 }, { "epoch": 0.63, "learning_rate": 4.4056715860506255e-05, "loss": 0.3523, "step": 3593 }, { "epoch": 0.63, "learning_rate": 4.40520797835027e-05, "loss": 0.3617, "step": 3594 }, { "epoch": 0.63, "learning_rate": 4.404744217439577e-05, "loss": 0.3736, "step": 3595 }, { "epoch": 0.63, "learning_rate": 4.404280303361472e-05, "loss": 0.3519, "step": 3596 }, { "epoch": 0.63, "learning_rate": 4.403816236158896e-05, "loss": 0.3717, "step": 3597 }, { "epoch": 0.63, "learning_rate": 4.4033520158748055e-05, "loss": 0.374, "step": 3598 }, { "epoch": 0.63, "learning_rate": 4.4028876425521694e-05, "loss": 0.3506, "step": 3599 }, { "epoch": 0.63, "learning_rate": 4.4024231162339715e-05, "loss": 0.3351, "step": 3600 }, { "epoch": 0.63, "learning_rate": 4.40195843696321e-05, "loss": 0.3643, "step": 3601 }, { "epoch": 0.63, "learning_rate": 4.4014936047828945e-05, "loss": 0.3463, "step": 3602 }, { "epoch": 0.63, "learning_rate": 4.401028619736053e-05, "loss": 0.3668, "step": 3603 }, { "epoch": 0.63, "learning_rate": 4.400563481865726e-05, "loss": 0.3522, "step": 3604 }, { "epoch": 0.63, "learning_rate": 4.400098191214967e-05, "loss": 0.3427, "step": 3605 }, { "epoch": 0.63, "learning_rate": 4.399632747826845e-05, "loss": 0.3658, "step": 3606 }, { "epoch": 0.63, "learning_rate": 4.399167151744442e-05, "loss": 0.3412, "step": 3607 }, { "epoch": 0.63, "learning_rate": 4.3987014030108554e-05, "loss": 0.3558, "step": 3608 }, { "epoch": 0.63, "learning_rate": 4.398235501669196e-05, "loss": 0.3518, "step": 3609 }, { "epoch": 0.63, "learning_rate": 4.3977694477625895e-05, "loss": 0.3511, "step": 3610 }, { "epoch": 0.63, "learning_rate": 4.397303241334173e-05, "loss": 0.3533, "step": 3611 }, { "epoch": 0.63, "learning_rate": 4.3968368824271014e-05, "loss": 0.405, "step": 3612 }, { "epoch": 0.63, "learning_rate": 4.396370371084543e-05, "loss": 0.3483, "step": 3613 }, { "epoch": 0.63, "learning_rate": 4.395903707349677e-05, "loss": 0.354, "step": 3614 }, { "epoch": 0.63, "learning_rate": 4.3954368912657e-05, "loss": 0.3639, "step": 3615 }, { "epoch": 0.63, "learning_rate": 4.394969922875822e-05, "loss": 0.3543, "step": 3616 }, { "epoch": 0.63, "learning_rate": 4.394502802223267e-05, "loss": 0.3714, "step": 3617 }, { "epoch": 0.63, "learning_rate": 4.394035529351272e-05, "loss": 0.3602, "step": 3618 }, { "epoch": 0.63, "learning_rate": 4.3935681043030895e-05, "loss": 0.3706, "step": 3619 }, { "epoch": 0.63, "learning_rate": 4.393100527121986e-05, "loss": 0.3583, "step": 3620 }, { "epoch": 0.63, "learning_rate": 4.3926327978512404e-05, "loss": 0.3415, "step": 3621 }, { "epoch": 0.63, "learning_rate": 4.3921649165341486e-05, "loss": 0.3531, "step": 3622 }, { "epoch": 0.63, "learning_rate": 4.3916968832140176e-05, "loss": 0.3615, "step": 3623 }, { "epoch": 0.63, "learning_rate": 4.391228697934171e-05, "loss": 0.3569, "step": 3624 }, { "epoch": 0.63, "learning_rate": 4.390760360737943e-05, "loss": 0.3483, "step": 3625 }, { "epoch": 0.63, "learning_rate": 4.390291871668687e-05, "loss": 0.364, "step": 3626 }, { "epoch": 0.63, "learning_rate": 4.3898232307697654e-05, "loss": 0.3479, "step": 3627 }, { "epoch": 0.63, "learning_rate": 4.389354438084558e-05, "loss": 0.3692, "step": 3628 }, { "epoch": 0.63, "learning_rate": 4.388885493656456e-05, "loss": 0.3607, "step": 3629 }, { "epoch": 0.63, "learning_rate": 4.388416397528868e-05, "loss": 0.362, "step": 3630 }, { "epoch": 0.63, "learning_rate": 4.387947149745214e-05, "loss": 0.3477, "step": 3631 }, { "epoch": 0.63, "learning_rate": 4.3874777503489285e-05, "loss": 0.383, "step": 3632 }, { "epoch": 0.63, "learning_rate": 4.3870081993834606e-05, "loss": 0.3718, "step": 3633 }, { "epoch": 0.63, "learning_rate": 4.386538496892272e-05, "loss": 0.3745, "step": 3634 }, { "epoch": 0.63, "learning_rate": 4.3860686429188416e-05, "loss": 0.3903, "step": 3635 }, { "epoch": 0.63, "learning_rate": 4.385598637506659e-05, "loss": 0.3577, "step": 3636 }, { "epoch": 0.63, "learning_rate": 4.38512848069923e-05, "loss": 0.3516, "step": 3637 }, { "epoch": 0.63, "learning_rate": 4.3846581725400726e-05, "loss": 0.37, "step": 3638 }, { "epoch": 0.63, "learning_rate": 4.38418771307272e-05, "loss": 0.3569, "step": 3639 }, { "epoch": 0.63, "learning_rate": 4.3837171023407195e-05, "loss": 0.3523, "step": 3640 }, { "epoch": 0.63, "learning_rate": 4.383246340387631e-05, "loss": 0.3519, "step": 3641 }, { "epoch": 0.63, "learning_rate": 4.382775427257032e-05, "loss": 0.3574, "step": 3642 }, { "epoch": 0.63, "learning_rate": 4.382304362992509e-05, "loss": 0.3622, "step": 3643 }, { "epoch": 0.64, "learning_rate": 4.381833147637665e-05, "loss": 0.3623, "step": 3644 }, { "epoch": 0.64, "learning_rate": 4.381361781236118e-05, "loss": 0.3564, "step": 3645 }, { "epoch": 0.64, "learning_rate": 4.3808902638314975e-05, "loss": 0.3558, "step": 3646 }, { "epoch": 0.64, "learning_rate": 4.3804185954674506e-05, "loss": 0.3425, "step": 3647 }, { "epoch": 0.64, "learning_rate": 4.3799467761876344e-05, "loss": 0.3542, "step": 3648 }, { "epoch": 0.64, "learning_rate": 4.3794748060357224e-05, "loss": 0.34, "step": 3649 }, { "epoch": 0.64, "learning_rate": 4.3790026850554006e-05, "loss": 0.3586, "step": 3650 }, { "epoch": 0.64, "learning_rate": 4.3785304132903705e-05, "loss": 0.3681, "step": 3651 }, { "epoch": 0.64, "learning_rate": 4.378057990784346e-05, "loss": 0.3409, "step": 3652 }, { "epoch": 0.64, "learning_rate": 4.3775854175810575e-05, "loss": 0.3643, "step": 3653 }, { "epoch": 0.64, "learning_rate": 4.3771126937242454e-05, "loss": 0.3473, "step": 3654 }, { "epoch": 0.64, "learning_rate": 4.3766398192576684e-05, "loss": 0.3623, "step": 3655 }, { "epoch": 0.64, "learning_rate": 4.376166794225096e-05, "loss": 0.3797, "step": 3656 }, { "epoch": 0.64, "learning_rate": 4.375693618670311e-05, "loss": 0.3536, "step": 3657 }, { "epoch": 0.64, "learning_rate": 4.3752202926371145e-05, "loss": 0.3434, "step": 3658 }, { "epoch": 0.64, "learning_rate": 4.374746816169317e-05, "loss": 0.3632, "step": 3659 }, { "epoch": 0.64, "learning_rate": 4.3742731893107455e-05, "loss": 0.3733, "step": 3660 }, { "epoch": 0.64, "learning_rate": 4.3737994121052396e-05, "loss": 0.3485, "step": 3661 }, { "epoch": 0.64, "learning_rate": 4.373325484596654e-05, "loss": 0.3547, "step": 3662 }, { "epoch": 0.64, "learning_rate": 4.372851406828855e-05, "loss": 0.3618, "step": 3663 }, { "epoch": 0.64, "learning_rate": 4.372377178845727e-05, "loss": 0.3564, "step": 3664 }, { "epoch": 0.64, "learning_rate": 4.371902800691165e-05, "loss": 0.3451, "step": 3665 }, { "epoch": 0.64, "learning_rate": 4.371428272409077e-05, "loss": 0.3636, "step": 3666 }, { "epoch": 0.64, "learning_rate": 4.370953594043388e-05, "loss": 0.3418, "step": 3667 }, { "epoch": 0.64, "learning_rate": 4.370478765638036e-05, "loss": 0.3486, "step": 3668 }, { "epoch": 0.64, "learning_rate": 4.370003787236971e-05, "loss": 0.3434, "step": 3669 }, { "epoch": 0.64, "learning_rate": 4.369528658884158e-05, "loss": 0.3633, "step": 3670 }, { "epoch": 0.64, "learning_rate": 4.3690533806235776e-05, "loss": 0.3506, "step": 3671 }, { "epoch": 0.64, "learning_rate": 4.3685779524992224e-05, "loss": 0.3395, "step": 3672 }, { "epoch": 0.64, "learning_rate": 4.368102374555099e-05, "loss": 0.3515, "step": 3673 }, { "epoch": 0.64, "learning_rate": 4.367626646835227e-05, "loss": 0.3477, "step": 3674 }, { "epoch": 0.64, "learning_rate": 4.367150769383644e-05, "loss": 0.3662, "step": 3675 }, { "epoch": 0.64, "learning_rate": 4.366674742244395e-05, "loss": 0.3762, "step": 3676 }, { "epoch": 0.64, "learning_rate": 4.366198565461544e-05, "loss": 0.3613, "step": 3677 }, { "epoch": 0.64, "learning_rate": 4.365722239079167e-05, "loss": 0.3672, "step": 3678 }, { "epoch": 0.64, "learning_rate": 4.365245763141353e-05, "loss": 0.3425, "step": 3679 }, { "epoch": 0.64, "learning_rate": 4.3647691376922087e-05, "loss": 0.363, "step": 3680 }, { "epoch": 0.64, "learning_rate": 4.364292362775849e-05, "loss": 0.3621, "step": 3681 }, { "epoch": 0.64, "learning_rate": 4.363815438436407e-05, "loss": 0.3635, "step": 3682 }, { "epoch": 0.64, "learning_rate": 4.363338364718026e-05, "loss": 0.3561, "step": 3683 }, { "epoch": 0.64, "learning_rate": 4.362861141664868e-05, "loss": 0.3744, "step": 3684 }, { "epoch": 0.64, "learning_rate": 4.362383769321104e-05, "loss": 0.3434, "step": 3685 }, { "epoch": 0.64, "learning_rate": 4.361906247730921e-05, "loss": 0.3678, "step": 3686 }, { "epoch": 0.64, "learning_rate": 4.361428576938521e-05, "loss": 0.3639, "step": 3687 }, { "epoch": 0.64, "learning_rate": 4.360950756988117e-05, "loss": 0.359, "step": 3688 }, { "epoch": 0.64, "learning_rate": 4.360472787923937e-05, "loss": 0.3671, "step": 3689 }, { "epoch": 0.64, "learning_rate": 4.359994669790224e-05, "loss": 0.3663, "step": 3690 }, { "epoch": 0.64, "learning_rate": 4.359516402631234e-05, "loss": 0.37, "step": 3691 }, { "epoch": 0.64, "learning_rate": 4.359037986491237e-05, "loss": 0.3559, "step": 3692 }, { "epoch": 0.64, "learning_rate": 4.358559421414515e-05, "loss": 0.3353, "step": 3693 }, { "epoch": 0.64, "learning_rate": 4.358080707445365e-05, "loss": 0.3456, "step": 3694 }, { "epoch": 0.64, "learning_rate": 4.3576018446281006e-05, "loss": 0.3672, "step": 3695 }, { "epoch": 0.64, "learning_rate": 4.3571228330070434e-05, "loss": 0.3887, "step": 3696 }, { "epoch": 0.64, "learning_rate": 4.356643672626535e-05, "loss": 0.3504, "step": 3697 }, { "epoch": 0.64, "learning_rate": 4.356164363530925e-05, "loss": 0.3677, "step": 3698 }, { "epoch": 0.64, "learning_rate": 4.355684905764581e-05, "loss": 0.3771, "step": 3699 }, { "epoch": 0.64, "learning_rate": 4.355205299371883e-05, "loss": 0.3633, "step": 3700 }, { "epoch": 0.65, "learning_rate": 4.354725544397224e-05, "loss": 0.364, "step": 3701 }, { "epoch": 0.65, "learning_rate": 4.35424564088501e-05, "loss": 0.354, "step": 3702 }, { "epoch": 0.65, "learning_rate": 4.3537655888796664e-05, "loss": 0.3729, "step": 3703 }, { "epoch": 0.65, "learning_rate": 4.353285388425623e-05, "loss": 0.3616, "step": 3704 }, { "epoch": 0.65, "learning_rate": 4.352805039567332e-05, "loss": 0.3509, "step": 3705 }, { "epoch": 0.65, "learning_rate": 4.352324542349254e-05, "loss": 0.3736, "step": 3706 }, { "epoch": 0.65, "learning_rate": 4.3518438968158656e-05, "loss": 0.3568, "step": 3707 }, { "epoch": 0.65, "learning_rate": 4.3513631030116556e-05, "loss": 0.3563, "step": 3708 }, { "epoch": 0.65, "learning_rate": 4.35088216098113e-05, "loss": 0.3631, "step": 3709 }, { "epoch": 0.65, "learning_rate": 4.350401070768803e-05, "loss": 0.3581, "step": 3710 }, { "epoch": 0.65, "learning_rate": 4.3499198324192075e-05, "loss": 0.3575, "step": 3711 }, { "epoch": 0.65, "learning_rate": 4.3494384459768885e-05, "loss": 0.356, "step": 3712 }, { "epoch": 0.65, "learning_rate": 4.3489569114864023e-05, "loss": 0.3457, "step": 3713 }, { "epoch": 0.65, "learning_rate": 4.348475228992323e-05, "loss": 0.3532, "step": 3714 }, { "epoch": 0.65, "learning_rate": 4.347993398539235e-05, "loss": 0.3592, "step": 3715 }, { "epoch": 0.65, "learning_rate": 4.347511420171739e-05, "loss": 0.3645, "step": 3716 }, { "epoch": 0.65, "learning_rate": 4.347029293934448e-05, "loss": 0.3765, "step": 3717 }, { "epoch": 0.65, "learning_rate": 4.3465470198719876e-05, "loss": 0.3459, "step": 3718 }, { "epoch": 0.65, "learning_rate": 4.346064598028999e-05, "loss": 0.3536, "step": 3719 }, { "epoch": 0.65, "learning_rate": 4.345582028450137e-05, "loss": 0.3602, "step": 3720 }, { "epoch": 0.65, "learning_rate": 4.345099311180069e-05, "loss": 0.3686, "step": 3721 }, { "epoch": 0.65, "learning_rate": 4.3446164462634765e-05, "loss": 0.3533, "step": 3722 }, { "epoch": 0.65, "learning_rate": 4.344133433745055e-05, "loss": 0.3825, "step": 3723 }, { "epoch": 0.65, "learning_rate": 4.343650273669513e-05, "loss": 0.37, "step": 3724 }, { "epoch": 0.65, "learning_rate": 4.343166966081573e-05, "loss": 0.3536, "step": 3725 }, { "epoch": 0.65, "learning_rate": 4.342683511025972e-05, "loss": 0.3439, "step": 3726 }, { "epoch": 0.65, "learning_rate": 4.342199908547459e-05, "loss": 0.3433, "step": 3727 }, { "epoch": 0.65, "learning_rate": 4.341716158690797e-05, "loss": 0.3719, "step": 3728 }, { "epoch": 0.65, "learning_rate": 4.3412322615007656e-05, "loss": 0.3586, "step": 3729 }, { "epoch": 0.65, "learning_rate": 4.3407482170221535e-05, "loss": 0.3548, "step": 3730 }, { "epoch": 0.65, "learning_rate": 4.340264025299765e-05, "loss": 0.35, "step": 3731 }, { "epoch": 0.65, "learning_rate": 4.339779686378419e-05, "loss": 0.3594, "step": 3732 }, { "epoch": 0.65, "learning_rate": 4.339295200302947e-05, "loss": 0.3562, "step": 3733 }, { "epoch": 0.65, "learning_rate": 4.338810567118194e-05, "loss": 0.3469, "step": 3734 }, { "epoch": 0.65, "learning_rate": 4.338325786869019e-05, "loss": 0.3587, "step": 3735 }, { "epoch": 0.65, "learning_rate": 4.337840859600294e-05, "loss": 0.3505, "step": 3736 }, { "epoch": 0.65, "learning_rate": 4.337355785356907e-05, "loss": 0.3859, "step": 3737 }, { "epoch": 0.65, "learning_rate": 4.336870564183756e-05, "loss": 0.3833, "step": 3738 }, { "epoch": 0.65, "learning_rate": 4.336385196125754e-05, "loss": 0.3849, "step": 3739 }, { "epoch": 0.65, "learning_rate": 4.3358996812278296e-05, "loss": 0.3564, "step": 3740 }, { "epoch": 0.65, "learning_rate": 4.335414019534922e-05, "loss": 0.3632, "step": 3741 }, { "epoch": 0.65, "learning_rate": 4.334928211091985e-05, "loss": 0.3502, "step": 3742 }, { "epoch": 0.65, "learning_rate": 4.3344422559439876e-05, "loss": 0.359, "step": 3743 }, { "epoch": 0.65, "learning_rate": 4.3339561541359107e-05, "loss": 0.3509, "step": 3744 }, { "epoch": 0.65, "learning_rate": 4.333469905712749e-05, "loss": 0.3721, "step": 3745 }, { "epoch": 0.65, "learning_rate": 4.33298351071951e-05, "loss": 0.3586, "step": 3746 }, { "epoch": 0.65, "learning_rate": 4.332496969201217e-05, "loss": 0.3617, "step": 3747 }, { "epoch": 0.65, "learning_rate": 4.332010281202905e-05, "loss": 0.3585, "step": 3748 }, { "epoch": 0.65, "learning_rate": 4.331523446769623e-05, "loss": 0.351, "step": 3749 }, { "epoch": 0.65, "learning_rate": 4.331036465946434e-05, "loss": 0.3548, "step": 3750 }, { "epoch": 0.65, "learning_rate": 4.330549338778414e-05, "loss": 0.3475, "step": 3751 }, { "epoch": 0.65, "learning_rate": 4.330062065310652e-05, "loss": 0.3561, "step": 3752 }, { "epoch": 0.65, "learning_rate": 4.329574645588253e-05, "loss": 0.3504, "step": 3753 }, { "epoch": 0.65, "learning_rate": 4.329087079656333e-05, "loss": 0.3497, "step": 3754 }, { "epoch": 0.65, "learning_rate": 4.3285993675600214e-05, "loss": 0.3525, "step": 3755 }, { "epoch": 0.65, "learning_rate": 4.3281115093444637e-05, "loss": 0.3505, "step": 3756 }, { "epoch": 0.65, "learning_rate": 4.327623505054816e-05, "loss": 0.3629, "step": 3757 }, { "epoch": 0.66, "learning_rate": 4.3271353547362494e-05, "loss": 0.3549, "step": 3758 }, { "epoch": 0.66, "learning_rate": 4.32664705843395e-05, "loss": 0.3574, "step": 3759 }, { "epoch": 0.66, "learning_rate": 4.326158616193113e-05, "loss": 0.3301, "step": 3760 }, { "epoch": 0.66, "learning_rate": 4.3256700280589525e-05, "loss": 0.3523, "step": 3761 }, { "epoch": 0.66, "learning_rate": 4.3251812940766925e-05, "loss": 0.3607, "step": 3762 }, { "epoch": 0.66, "learning_rate": 4.324692414291571e-05, "loss": 0.3597, "step": 3763 }, { "epoch": 0.66, "learning_rate": 4.32420338874884e-05, "loss": 0.3583, "step": 3764 }, { "epoch": 0.66, "learning_rate": 4.3237142174937655e-05, "loss": 0.343, "step": 3765 }, { "epoch": 0.66, "learning_rate": 4.323224900571626e-05, "loss": 0.3607, "step": 3766 }, { "epoch": 0.66, "learning_rate": 4.322735438027715e-05, "loss": 0.3527, "step": 3767 }, { "epoch": 0.66, "learning_rate": 4.3222458299073375e-05, "loss": 0.3728, "step": 3768 }, { "epoch": 0.66, "learning_rate": 4.321756076255813e-05, "loss": 0.3439, "step": 3769 }, { "epoch": 0.66, "learning_rate": 4.321266177118474e-05, "loss": 0.3656, "step": 3770 }, { "epoch": 0.66, "learning_rate": 4.320776132540667e-05, "loss": 0.3674, "step": 3771 }, { "epoch": 0.66, "learning_rate": 4.3202859425677526e-05, "loss": 0.3759, "step": 3772 }, { "epoch": 0.66, "learning_rate": 4.319795607245103e-05, "loss": 0.3512, "step": 3773 }, { "epoch": 0.66, "learning_rate": 4.319305126618106e-05, "loss": 0.3495, "step": 3774 }, { "epoch": 0.66, "learning_rate": 4.3188145007321614e-05, "loss": 0.3394, "step": 3775 }, { "epoch": 0.66, "learning_rate": 4.318323729632682e-05, "loss": 0.3416, "step": 3776 }, { "epoch": 0.66, "learning_rate": 4.317832813365095e-05, "loss": 0.3592, "step": 3777 }, { "epoch": 0.66, "learning_rate": 4.3173417519748424e-05, "loss": 0.3618, "step": 3778 }, { "epoch": 0.66, "learning_rate": 4.316850545507376e-05, "loss": 0.3581, "step": 3779 }, { "epoch": 0.66, "learning_rate": 4.316359194008165e-05, "loss": 0.3452, "step": 3780 }, { "epoch": 0.66, "learning_rate": 4.315867697522688e-05, "loss": 0.3695, "step": 3781 }, { "epoch": 0.66, "learning_rate": 4.315376056096442e-05, "loss": 0.3701, "step": 3782 }, { "epoch": 0.66, "learning_rate": 4.314884269774932e-05, "loss": 0.3841, "step": 3783 }, { "epoch": 0.66, "learning_rate": 4.31439233860368e-05, "loss": 0.3603, "step": 3784 }, { "epoch": 0.66, "learning_rate": 4.313900262628221e-05, "loss": 0.3662, "step": 3785 }, { "epoch": 0.66, "learning_rate": 4.313408041894103e-05, "loss": 0.3537, "step": 3786 }, { "epoch": 0.66, "learning_rate": 4.312915676446886e-05, "loss": 0.3611, "step": 3787 }, { "epoch": 0.66, "learning_rate": 4.312423166332144e-05, "loss": 0.3604, "step": 3788 }, { "epoch": 0.66, "learning_rate": 4.3119305115954676e-05, "loss": 0.3653, "step": 3789 }, { "epoch": 0.66, "learning_rate": 4.3114377122824574e-05, "loss": 0.3485, "step": 3790 }, { "epoch": 0.66, "learning_rate": 4.310944768438726e-05, "loss": 0.3611, "step": 3791 }, { "epoch": 0.66, "learning_rate": 4.3104516801099046e-05, "loss": 0.3381, "step": 3792 }, { "epoch": 0.66, "learning_rate": 4.309958447341633e-05, "loss": 0.3559, "step": 3793 }, { "epoch": 0.66, "learning_rate": 4.309465070179566e-05, "loss": 0.3558, "step": 3794 }, { "epoch": 0.66, "learning_rate": 4.308971548669372e-05, "loss": 0.3392, "step": 3795 }, { "epoch": 0.66, "learning_rate": 4.3084778828567335e-05, "loss": 0.3606, "step": 3796 }, { "epoch": 0.66, "learning_rate": 4.3079840727873446e-05, "loss": 0.3541, "step": 3797 }, { "epoch": 0.66, "learning_rate": 4.3074901185069136e-05, "loss": 0.3739, "step": 3798 }, { "epoch": 0.66, "learning_rate": 4.306996020061163e-05, "loss": 0.3489, "step": 3799 }, { "epoch": 0.66, "learning_rate": 4.3065017774958265e-05, "loss": 0.3721, "step": 3800 }, { "epoch": 0.66, "learning_rate": 4.306007390856655e-05, "loss": 0.3686, "step": 3801 }, { "epoch": 0.66, "learning_rate": 4.3055128601894074e-05, "loss": 0.3487, "step": 3802 }, { "epoch": 0.66, "learning_rate": 4.3050181855398595e-05, "loss": 0.3338, "step": 3803 }, { "epoch": 0.66, "learning_rate": 4.3045233669538e-05, "loss": 0.3606, "step": 3804 }, { "epoch": 0.66, "learning_rate": 4.304028404477031e-05, "loss": 0.3584, "step": 3805 }, { "epoch": 0.66, "learning_rate": 4.3035332981553684e-05, "loss": 0.3424, "step": 3806 }, { "epoch": 0.66, "learning_rate": 4.303038048034638e-05, "loss": 0.3642, "step": 3807 }, { "epoch": 0.66, "learning_rate": 4.3025426541606826e-05, "loss": 0.3565, "step": 3808 }, { "epoch": 0.66, "learning_rate": 4.302047116579357e-05, "loss": 0.3455, "step": 3809 }, { "epoch": 0.66, "learning_rate": 4.3015514353365307e-05, "loss": 0.3601, "step": 3810 }, { "epoch": 0.66, "learning_rate": 4.301055610478083e-05, "loss": 0.3607, "step": 3811 }, { "epoch": 0.66, "learning_rate": 4.300559642049911e-05, "loss": 0.3563, "step": 3812 }, { "epoch": 0.66, "learning_rate": 4.300063530097921e-05, "loss": 0.3584, "step": 3813 }, { "epoch": 0.66, "learning_rate": 4.299567274668036e-05, "loss": 0.3415, "step": 3814 }, { "epoch": 0.66, "learning_rate": 4.2990708758061884e-05, "loss": 0.3421, "step": 3815 }, { "epoch": 0.67, "learning_rate": 4.298574333558329e-05, "loss": 0.365, "step": 3816 }, { "epoch": 0.67, "learning_rate": 4.298077647970417e-05, "loss": 0.3671, "step": 3817 }, { "epoch": 0.67, "learning_rate": 4.297580819088428e-05, "loss": 0.3428, "step": 3818 }, { "epoch": 0.67, "learning_rate": 4.297083846958349e-05, "loss": 0.3783, "step": 3819 }, { "epoch": 0.67, "learning_rate": 4.296586731626181e-05, "loss": 0.3649, "step": 3820 }, { "epoch": 0.67, "learning_rate": 4.2960894731379395e-05, "loss": 0.3436, "step": 3821 }, { "epoch": 0.67, "learning_rate": 4.295592071539651e-05, "loss": 0.3651, "step": 3822 }, { "epoch": 0.67, "learning_rate": 4.2950945268773555e-05, "loss": 0.3594, "step": 3823 }, { "epoch": 0.67, "learning_rate": 4.29459683919711e-05, "loss": 0.3512, "step": 3824 }, { "epoch": 0.67, "learning_rate": 4.2940990085449775e-05, "loss": 0.355, "step": 3825 }, { "epoch": 0.67, "learning_rate": 4.293601034967042e-05, "loss": 0.3474, "step": 3826 }, { "epoch": 0.67, "learning_rate": 4.293102918509396e-05, "loss": 0.3645, "step": 3827 }, { "epoch": 0.67, "learning_rate": 4.2926046592181456e-05, "loss": 0.3539, "step": 3828 }, { "epoch": 0.67, "learning_rate": 4.292106257139413e-05, "loss": 0.3503, "step": 3829 }, { "epoch": 0.67, "learning_rate": 4.291607712319329e-05, "loss": 0.3512, "step": 3830 }, { "epoch": 0.67, "learning_rate": 4.291109024804043e-05, "loss": 0.348, "step": 3831 }, { "epoch": 0.67, "learning_rate": 4.290610194639712e-05, "loss": 0.3479, "step": 3832 }, { "epoch": 0.67, "learning_rate": 4.290111221872511e-05, "loss": 0.3628, "step": 3833 }, { "epoch": 0.67, "learning_rate": 4.289612106548625e-05, "loss": 0.3439, "step": 3834 }, { "epoch": 0.67, "learning_rate": 4.289112848714254e-05, "loss": 0.3765, "step": 3835 }, { "epoch": 0.67, "learning_rate": 4.288613448415612e-05, "loss": 0.3524, "step": 3836 }, { "epoch": 0.67, "learning_rate": 4.288113905698922e-05, "loss": 0.369, "step": 3837 }, { "epoch": 0.67, "learning_rate": 4.287614220610426e-05, "loss": 0.3668, "step": 3838 }, { "epoch": 0.67, "learning_rate": 4.287114393196373e-05, "loss": 0.3524, "step": 3839 }, { "epoch": 0.67, "learning_rate": 4.2866144235030304e-05, "loss": 0.3625, "step": 3840 }, { "epoch": 0.67, "learning_rate": 4.286114311576676e-05, "loss": 0.3532, "step": 3841 }, { "epoch": 0.67, "learning_rate": 4.2856140574636015e-05, "loss": 0.3557, "step": 3842 }, { "epoch": 0.67, "learning_rate": 4.2851136612101114e-05, "loss": 0.3353, "step": 3843 }, { "epoch": 0.67, "learning_rate": 4.2846131228625246e-05, "loss": 0.3548, "step": 3844 }, { "epoch": 0.67, "learning_rate": 4.2841124424671716e-05, "loss": 0.3532, "step": 3845 }, { "epoch": 0.67, "learning_rate": 4.283611620070397e-05, "loss": 0.3483, "step": 3846 }, { "epoch": 0.67, "learning_rate": 4.283110655718557e-05, "loss": 0.3485, "step": 3847 }, { "epoch": 0.67, "learning_rate": 4.282609549458024e-05, "loss": 0.3431, "step": 3848 }, { "epoch": 0.67, "learning_rate": 4.282108301335181e-05, "loss": 0.3553, "step": 3849 }, { "epoch": 0.67, "learning_rate": 4.2816069113964245e-05, "loss": 0.3719, "step": 3850 }, { "epoch": 0.67, "learning_rate": 4.281105379688164e-05, "loss": 0.3557, "step": 3851 }, { "epoch": 0.67, "learning_rate": 4.280603706256824e-05, "loss": 0.3587, "step": 3852 }, { "epoch": 0.67, "learning_rate": 4.28010189114884e-05, "loss": 0.3723, "step": 3853 }, { "epoch": 0.67, "learning_rate": 4.2795999344106605e-05, "loss": 0.3539, "step": 3854 }, { "epoch": 0.67, "learning_rate": 4.2790978360887485e-05, "loss": 0.3415, "step": 3855 }, { "epoch": 0.67, "learning_rate": 4.27859559622958e-05, "loss": 0.3774, "step": 3856 }, { "epoch": 0.67, "learning_rate": 4.278093214879644e-05, "loss": 0.3611, "step": 3857 }, { "epoch": 0.67, "learning_rate": 4.27759069208544e-05, "loss": 0.3376, "step": 3858 }, { "epoch": 0.67, "learning_rate": 4.277088027893485e-05, "loss": 0.3609, "step": 3859 }, { "epoch": 0.67, "learning_rate": 4.276585222350306e-05, "loss": 0.3525, "step": 3860 }, { "epoch": 0.67, "learning_rate": 4.2760822755024446e-05, "loss": 0.3519, "step": 3861 }, { "epoch": 0.67, "learning_rate": 4.275579187396453e-05, "loss": 0.3437, "step": 3862 }, { "epoch": 0.67, "learning_rate": 4.275075958078901e-05, "loss": 0.3379, "step": 3863 }, { "epoch": 0.67, "learning_rate": 4.2745725875963684e-05, "loss": 0.3509, "step": 3864 }, { "epoch": 0.67, "learning_rate": 4.274069075995446e-05, "loss": 0.352, "step": 3865 }, { "epoch": 0.67, "learning_rate": 4.2735654233227424e-05, "loss": 0.3486, "step": 3866 }, { "epoch": 0.67, "learning_rate": 4.273061629624877e-05, "loss": 0.364, "step": 3867 }, { "epoch": 0.67, "learning_rate": 4.2725576949484805e-05, "loss": 0.3673, "step": 3868 }, { "epoch": 0.67, "learning_rate": 4.2720536193402e-05, "loss": 0.3471, "step": 3869 }, { "epoch": 0.67, "learning_rate": 4.271549402846693e-05, "loss": 0.3546, "step": 3870 }, { "epoch": 0.67, "learning_rate": 4.271045045514633e-05, "loss": 0.3787, "step": 3871 }, { "epoch": 0.67, "learning_rate": 4.2705405473907026e-05, "loss": 0.3631, "step": 3872 }, { "epoch": 0.68, "learning_rate": 4.2700359085216004e-05, "loss": 0.3449, "step": 3873 }, { "epoch": 0.68, "learning_rate": 4.2695311289540365e-05, "loss": 0.3466, "step": 3874 }, { "epoch": 0.68, "learning_rate": 4.269026208734734e-05, "loss": 0.3475, "step": 3875 }, { "epoch": 0.68, "learning_rate": 4.268521147910432e-05, "loss": 0.3354, "step": 3876 }, { "epoch": 0.68, "learning_rate": 4.268015946527878e-05, "loss": 0.3713, "step": 3877 }, { "epoch": 0.68, "learning_rate": 4.267510604633836e-05, "loss": 0.3549, "step": 3878 }, { "epoch": 0.68, "learning_rate": 4.267005122275081e-05, "loss": 0.3642, "step": 3879 }, { "epoch": 0.68, "learning_rate": 4.266499499498402e-05, "loss": 0.3539, "step": 3880 }, { "epoch": 0.68, "learning_rate": 4.265993736350601e-05, "loss": 0.3362, "step": 3881 }, { "epoch": 0.68, "learning_rate": 4.265487832878492e-05, "loss": 0.3678, "step": 3882 }, { "epoch": 0.68, "learning_rate": 4.264981789128904e-05, "loss": 0.3432, "step": 3883 }, { "epoch": 0.68, "learning_rate": 4.2644756051486764e-05, "loss": 0.3434, "step": 3884 }, { "epoch": 0.68, "learning_rate": 4.263969280984664e-05, "loss": 0.3549, "step": 3885 }, { "epoch": 0.68, "learning_rate": 4.2634628166837325e-05, "loss": 0.342, "step": 3886 }, { "epoch": 0.68, "learning_rate": 4.262956212292763e-05, "loss": 0.3335, "step": 3887 }, { "epoch": 0.68, "learning_rate": 4.2624494678586455e-05, "loss": 0.3705, "step": 3888 }, { "epoch": 0.68, "learning_rate": 4.261942583428288e-05, "loss": 0.3515, "step": 3889 }, { "epoch": 0.68, "learning_rate": 4.2614355590486095e-05, "loss": 0.3434, "step": 3890 }, { "epoch": 0.68, "learning_rate": 4.260928394766539e-05, "loss": 0.3436, "step": 3891 }, { "epoch": 0.68, "learning_rate": 4.2604210906290215e-05, "loss": 0.3572, "step": 3892 }, { "epoch": 0.68, "learning_rate": 4.2599136466830156e-05, "loss": 0.3582, "step": 3893 }, { "epoch": 0.68, "learning_rate": 4.2594060629754925e-05, "loss": 0.3519, "step": 3894 }, { "epoch": 0.68, "learning_rate": 4.258898339553432e-05, "loss": 0.3539, "step": 3895 }, { "epoch": 0.68, "learning_rate": 4.2583904764638334e-05, "loss": 0.3758, "step": 3896 }, { "epoch": 0.68, "learning_rate": 4.2578824737537044e-05, "loss": 0.3512, "step": 3897 }, { "epoch": 0.68, "learning_rate": 4.257374331470068e-05, "loss": 0.3554, "step": 3898 }, { "epoch": 0.68, "learning_rate": 4.2568660496599576e-05, "loss": 0.3662, "step": 3899 }, { "epoch": 0.68, "learning_rate": 4.256357628370423e-05, "loss": 0.3632, "step": 3900 }, { "epoch": 0.68, "learning_rate": 4.255849067648524e-05, "loss": 0.3489, "step": 3901 }, { "epoch": 0.68, "learning_rate": 4.255340367541334e-05, "loss": 0.3569, "step": 3902 }, { "epoch": 0.68, "learning_rate": 4.2548315280959394e-05, "loss": 0.3598, "step": 3903 }, { "epoch": 0.68, "learning_rate": 4.2543225493594414e-05, "loss": 0.3743, "step": 3904 }, { "epoch": 0.68, "learning_rate": 4.2538134313789506e-05, "loss": 0.3744, "step": 3905 }, { "epoch": 0.68, "learning_rate": 4.2533041742015925e-05, "loss": 0.3637, "step": 3906 }, { "epoch": 0.68, "learning_rate": 4.252794777874506e-05, "loss": 0.3738, "step": 3907 }, { "epoch": 0.68, "learning_rate": 4.252285242444842e-05, "loss": 0.3478, "step": 3908 }, { "epoch": 0.68, "learning_rate": 4.2517755679597646e-05, "loss": 0.3678, "step": 3909 }, { "epoch": 0.68, "learning_rate": 4.2512657544664506e-05, "loss": 0.347, "step": 3910 }, { "epoch": 0.68, "learning_rate": 4.2507558020120886e-05, "loss": 0.3576, "step": 3911 }, { "epoch": 0.68, "learning_rate": 4.2502457106438815e-05, "loss": 0.3533, "step": 3912 }, { "epoch": 0.68, "learning_rate": 4.249735480409045e-05, "loss": 0.3555, "step": 3913 }, { "epoch": 0.68, "learning_rate": 4.249225111354808e-05, "loss": 0.3601, "step": 3914 }, { "epoch": 0.68, "learning_rate": 4.24871460352841e-05, "loss": 0.349, "step": 3915 }, { "epoch": 0.68, "learning_rate": 4.248203956977107e-05, "loss": 0.3668, "step": 3916 }, { "epoch": 0.68, "learning_rate": 4.247693171748164e-05, "loss": 0.3497, "step": 3917 }, { "epoch": 0.68, "learning_rate": 4.2471822478888605e-05, "loss": 0.3618, "step": 3918 }, { "epoch": 0.68, "learning_rate": 4.246671185446491e-05, "loss": 0.3592, "step": 3919 }, { "epoch": 0.68, "learning_rate": 4.246159984468359e-05, "loss": 0.3538, "step": 3920 }, { "epoch": 0.68, "learning_rate": 4.245648645001782e-05, "loss": 0.359, "step": 3921 }, { "epoch": 0.68, "learning_rate": 4.245137167094093e-05, "loss": 0.3538, "step": 3922 }, { "epoch": 0.68, "learning_rate": 4.244625550792634e-05, "loss": 0.3653, "step": 3923 }, { "epoch": 0.68, "learning_rate": 4.244113796144763e-05, "loss": 0.3899, "step": 3924 }, { "epoch": 0.68, "learning_rate": 4.243601903197847e-05, "loss": 0.3579, "step": 3925 }, { "epoch": 0.68, "learning_rate": 4.243089871999271e-05, "loss": 0.3548, "step": 3926 }, { "epoch": 0.68, "learning_rate": 4.2425777025964284e-05, "loss": 0.3564, "step": 3927 }, { "epoch": 0.68, "learning_rate": 4.2420653950367264e-05, "loss": 0.3365, "step": 3928 }, { "epoch": 0.68, "learning_rate": 4.241552949367587e-05, "loss": 0.3538, "step": 3929 }, { "epoch": 0.68, "learning_rate": 4.241040365636442e-05, "loss": 0.3529, "step": 3930 }, { "epoch": 0.69, "learning_rate": 4.240527643890738e-05, "loss": 0.3424, "step": 3931 }, { "epoch": 0.69, "learning_rate": 4.240014784177935e-05, "loss": 0.3637, "step": 3932 }, { "epoch": 0.69, "learning_rate": 4.239501786545503e-05, "loss": 0.3629, "step": 3933 }, { "epoch": 0.69, "learning_rate": 4.238988651040927e-05, "loss": 0.3269, "step": 3934 }, { "epoch": 0.69, "learning_rate": 4.2384753777117045e-05, "loss": 0.3625, "step": 3935 }, { "epoch": 0.69, "learning_rate": 4.2379619666053443e-05, "loss": 0.3586, "step": 3936 }, { "epoch": 0.69, "learning_rate": 4.237448417769371e-05, "loss": 0.3482, "step": 3937 }, { "epoch": 0.69, "learning_rate": 4.236934731251318e-05, "loss": 0.3411, "step": 3938 }, { "epoch": 0.69, "learning_rate": 4.2364209070987346e-05, "loss": 0.353, "step": 3939 }, { "epoch": 0.69, "learning_rate": 4.2359069453591816e-05, "loss": 0.3381, "step": 3940 }, { "epoch": 0.69, "learning_rate": 4.2353928460802326e-05, "loss": 0.337, "step": 3941 }, { "epoch": 0.69, "learning_rate": 4.2348786093094733e-05, "loss": 0.3619, "step": 3942 }, { "epoch": 0.69, "learning_rate": 4.2343642350945036e-05, "loss": 0.3498, "step": 3943 }, { "epoch": 0.69, "learning_rate": 4.233849723482935e-05, "loss": 0.3421, "step": 3944 }, { "epoch": 0.69, "learning_rate": 4.2333350745223924e-05, "loss": 0.3553, "step": 3945 }, { "epoch": 0.69, "learning_rate": 4.2328202882605124e-05, "loss": 0.3759, "step": 3946 }, { "epoch": 0.69, "learning_rate": 4.232305364744945e-05, "loss": 0.3558, "step": 3947 }, { "epoch": 0.69, "learning_rate": 4.231790304023353e-05, "loss": 0.3535, "step": 3948 }, { "epoch": 0.69, "learning_rate": 4.231275106143412e-05, "loss": 0.3471, "step": 3949 }, { "epoch": 0.69, "learning_rate": 4.2307597711528106e-05, "loss": 0.3393, "step": 3950 }, { "epoch": 0.69, "learning_rate": 4.230244299099248e-05, "loss": 0.3501, "step": 3951 }, { "epoch": 0.69, "learning_rate": 4.229728690030438e-05, "loss": 0.3411, "step": 3952 }, { "epoch": 0.69, "learning_rate": 4.229212943994108e-05, "loss": 0.3656, "step": 3953 }, { "epoch": 0.69, "learning_rate": 4.2286970610379966e-05, "loss": 0.3634, "step": 3954 }, { "epoch": 0.69, "learning_rate": 4.228181041209853e-05, "loss": 0.3543, "step": 3955 }, { "epoch": 0.69, "learning_rate": 4.2276648845574444e-05, "loss": 0.3592, "step": 3956 }, { "epoch": 0.69, "learning_rate": 4.227148591128546e-05, "loss": 0.36, "step": 3957 }, { "epoch": 0.69, "learning_rate": 4.226632160970946e-05, "loss": 0.3445, "step": 3958 }, { "epoch": 0.69, "learning_rate": 4.226115594132449e-05, "loss": 0.345, "step": 3959 }, { "epoch": 0.69, "learning_rate": 4.225598890660868e-05, "loss": 0.3469, "step": 3960 }, { "epoch": 0.69, "learning_rate": 4.2250820506040326e-05, "loss": 0.3413, "step": 3961 }, { "epoch": 0.69, "learning_rate": 4.22456507400978e-05, "loss": 0.3605, "step": 3962 }, { "epoch": 0.69, "learning_rate": 4.224047960925965e-05, "loss": 0.3592, "step": 3963 }, { "epoch": 0.69, "learning_rate": 4.223530711400452e-05, "loss": 0.3651, "step": 3964 }, { "epoch": 0.69, "learning_rate": 4.2230133254811184e-05, "loss": 0.3677, "step": 3965 }, { "epoch": 0.69, "learning_rate": 4.222495803215856e-05, "loss": 0.35, "step": 3966 }, { "epoch": 0.69, "learning_rate": 4.2219781446525676e-05, "loss": 0.3568, "step": 3967 }, { "epoch": 0.69, "learning_rate": 4.221460349839169e-05, "loss": 0.3552, "step": 3968 }, { "epoch": 0.69, "learning_rate": 4.220942418823588e-05, "loss": 0.3483, "step": 3969 }, { "epoch": 0.69, "learning_rate": 4.2204243516537675e-05, "loss": 0.3614, "step": 3970 }, { "epoch": 0.69, "learning_rate": 4.219906148377659e-05, "loss": 0.3433, "step": 3971 }, { "epoch": 0.69, "learning_rate": 4.2193878090432296e-05, "loss": 0.3414, "step": 3972 }, { "epoch": 0.69, "learning_rate": 4.2188693336984576e-05, "loss": 0.3529, "step": 3973 }, { "epoch": 0.69, "learning_rate": 4.218350722391335e-05, "loss": 0.3587, "step": 3974 }, { "epoch": 0.69, "learning_rate": 4.2178319751698665e-05, "loss": 0.3453, "step": 3975 }, { "epoch": 0.69, "learning_rate": 4.217313092082068e-05, "loss": 0.3629, "step": 3976 }, { "epoch": 0.69, "learning_rate": 4.2167940731759674e-05, "loss": 0.3567, "step": 3977 }, { "epoch": 0.69, "learning_rate": 4.216274918499608e-05, "loss": 0.3581, "step": 3978 }, { "epoch": 0.69, "learning_rate": 4.215755628101045e-05, "loss": 0.3491, "step": 3979 }, { "epoch": 0.69, "learning_rate": 4.2152362020283426e-05, "loss": 0.3562, "step": 3980 }, { "epoch": 0.69, "learning_rate": 4.214716640329582e-05, "loss": 0.352, "step": 3981 }, { "epoch": 0.69, "learning_rate": 4.214196943052855e-05, "loss": 0.3491, "step": 3982 }, { "epoch": 0.69, "learning_rate": 4.2136771102462655e-05, "loss": 0.3566, "step": 3983 }, { "epoch": 0.69, "learning_rate": 4.213157141957931e-05, "loss": 0.3574, "step": 3984 }, { "epoch": 0.69, "learning_rate": 4.2126370382359815e-05, "loss": 0.3513, "step": 3985 }, { "epoch": 0.69, "learning_rate": 4.212116799128559e-05, "loss": 0.3812, "step": 3986 }, { "epoch": 0.69, "learning_rate": 4.211596424683817e-05, "loss": 0.3497, "step": 3987 }, { "epoch": 0.7, "learning_rate": 4.211075914949924e-05, "loss": 0.3502, "step": 3988 }, { "epoch": 0.7, "learning_rate": 4.2105552699750595e-05, "loss": 0.3483, "step": 3989 }, { "epoch": 0.7, "learning_rate": 4.210034489807416e-05, "loss": 0.3571, "step": 3990 }, { "epoch": 0.7, "learning_rate": 4.209513574495197e-05, "loss": 0.3522, "step": 3991 }, { "epoch": 0.7, "learning_rate": 4.2089925240866206e-05, "loss": 0.3526, "step": 3992 }, { "epoch": 0.7, "learning_rate": 4.208471338629918e-05, "loss": 0.3532, "step": 3993 }, { "epoch": 0.7, "learning_rate": 4.207950018173328e-05, "loss": 0.3481, "step": 3994 }, { "epoch": 0.7, "learning_rate": 4.2074285627651086e-05, "loss": 0.3718, "step": 3995 }, { "epoch": 0.7, "learning_rate": 4.206906972453527e-05, "loss": 0.3572, "step": 3996 }, { "epoch": 0.7, "learning_rate": 4.2063852472868595e-05, "loss": 0.3602, "step": 3997 }, { "epoch": 0.7, "learning_rate": 4.2058633873134025e-05, "loss": 0.3476, "step": 3998 }, { "epoch": 0.7, "learning_rate": 4.205341392581458e-05, "loss": 0.3658, "step": 3999 }, { "epoch": 0.7, "learning_rate": 4.204819263139344e-05, "loss": 0.3559, "step": 4000 }, { "epoch": 0.7, "learning_rate": 4.2042969990353916e-05, "loss": 0.3372, "step": 4001 }, { "epoch": 0.7, "learning_rate": 4.20377460031794e-05, "loss": 0.3505, "step": 4002 }, { "epoch": 0.7, "learning_rate": 4.2032520670353464e-05, "loss": 0.3527, "step": 4003 }, { "epoch": 0.7, "learning_rate": 4.202729399235976e-05, "loss": 0.3662, "step": 4004 }, { "epoch": 0.7, "learning_rate": 4.20220659696821e-05, "loss": 0.381, "step": 4005 }, { "epoch": 0.7, "learning_rate": 4.2016836602804386e-05, "loss": 0.3546, "step": 4006 }, { "epoch": 0.7, "learning_rate": 4.201160589221068e-05, "loss": 0.3602, "step": 4007 }, { "epoch": 0.7, "learning_rate": 4.2006373838385125e-05, "loss": 0.3526, "step": 4008 }, { "epoch": 0.7, "learning_rate": 4.200114044181204e-05, "loss": 0.3508, "step": 4009 }, { "epoch": 0.7, "learning_rate": 4.199590570297583e-05, "loss": 0.3603, "step": 4010 }, { "epoch": 0.7, "learning_rate": 4.199066962236103e-05, "loss": 0.3573, "step": 4011 }, { "epoch": 0.7, "learning_rate": 4.198543220045231e-05, "loss": 0.3481, "step": 4012 }, { "epoch": 0.7, "learning_rate": 4.198019343773447e-05, "loss": 0.3435, "step": 4013 }, { "epoch": 0.7, "learning_rate": 4.1974953334692406e-05, "loss": 0.3551, "step": 4014 }, { "epoch": 0.7, "learning_rate": 4.196971189181117e-05, "loss": 0.3519, "step": 4015 }, { "epoch": 0.7, "learning_rate": 4.196446910957591e-05, "loss": 0.3412, "step": 4016 }, { "epoch": 0.7, "learning_rate": 4.195922498847192e-05, "loss": 0.3641, "step": 4017 }, { "epoch": 0.7, "learning_rate": 4.195397952898461e-05, "loss": 0.3552, "step": 4018 }, { "epoch": 0.7, "learning_rate": 4.1948732731599515e-05, "loss": 0.3358, "step": 4019 }, { "epoch": 0.7, "learning_rate": 4.194348459680228e-05, "loss": 0.3532, "step": 4020 }, { "epoch": 0.7, "learning_rate": 4.19382351250787e-05, "loss": 0.3545, "step": 4021 }, { "epoch": 0.7, "learning_rate": 4.193298431691467e-05, "loss": 0.3597, "step": 4022 }, { "epoch": 0.7, "learning_rate": 4.192773217279623e-05, "loss": 0.3476, "step": 4023 }, { "epoch": 0.7, "learning_rate": 4.1922478693209525e-05, "loss": 0.3639, "step": 4024 }, { "epoch": 0.7, "learning_rate": 4.1917223878640824e-05, "loss": 0.3642, "step": 4025 }, { "epoch": 0.7, "learning_rate": 4.1911967729576526e-05, "loss": 0.3505, "step": 4026 }, { "epoch": 0.7, "learning_rate": 4.190671024650317e-05, "loss": 0.3557, "step": 4027 }, { "epoch": 0.7, "learning_rate": 4.190145142990739e-05, "loss": 0.3582, "step": 4028 }, { "epoch": 0.7, "learning_rate": 4.189619128027596e-05, "loss": 0.3633, "step": 4029 }, { "epoch": 0.7, "learning_rate": 4.1890929798095776e-05, "loss": 0.3496, "step": 4030 }, { "epoch": 0.7, "learning_rate": 4.1885666983853845e-05, "loss": 0.3693, "step": 4031 }, { "epoch": 0.7, "learning_rate": 4.188040283803731e-05, "loss": 0.3584, "step": 4032 }, { "epoch": 0.7, "learning_rate": 4.1875137361133435e-05, "loss": 0.3483, "step": 4033 }, { "epoch": 0.7, "learning_rate": 4.1869870553629606e-05, "loss": 0.3382, "step": 4034 }, { "epoch": 0.7, "learning_rate": 4.186460241601333e-05, "loss": 0.3525, "step": 4035 }, { "epoch": 0.7, "learning_rate": 4.185933294877225e-05, "loss": 0.3666, "step": 4036 }, { "epoch": 0.7, "learning_rate": 4.1854062152394115e-05, "loss": 0.3503, "step": 4037 }, { "epoch": 0.7, "learning_rate": 4.1848790027366805e-05, "loss": 0.352, "step": 4038 }, { "epoch": 0.7, "learning_rate": 4.18435165741783e-05, "loss": 0.359, "step": 4039 }, { "epoch": 0.7, "learning_rate": 4.183824179331677e-05, "loss": 0.3614, "step": 4040 }, { "epoch": 0.7, "learning_rate": 4.183296568527042e-05, "loss": 0.3509, "step": 4041 }, { "epoch": 0.7, "learning_rate": 4.182768825052763e-05, "loss": 0.3593, "step": 4042 }, { "epoch": 0.7, "learning_rate": 4.182240948957691e-05, "loss": 0.3755, "step": 4043 }, { "epoch": 0.7, "learning_rate": 4.1817129402906864e-05, "loss": 0.3572, "step": 4044 }, { "epoch": 0.71, "learning_rate": 4.1811847991006234e-05, "loss": 0.3553, "step": 4045 }, { "epoch": 0.71, "learning_rate": 4.180656525436387e-05, "loss": 0.3607, "step": 4046 }, { "epoch": 0.71, "learning_rate": 4.180128119346878e-05, "loss": 0.359, "step": 4047 }, { "epoch": 0.71, "learning_rate": 4.179599580881004e-05, "loss": 0.3448, "step": 4048 }, { "epoch": 0.71, "learning_rate": 4.17907091008769e-05, "loss": 0.3668, "step": 4049 }, { "epoch": 0.71, "learning_rate": 4.17854210701587e-05, "loss": 0.377, "step": 4050 }, { "epoch": 0.71, "learning_rate": 4.178013171714492e-05, "loss": 0.3833, "step": 4051 }, { "epoch": 0.71, "learning_rate": 4.177484104232516e-05, "loss": 0.355, "step": 4052 }, { "epoch": 0.71, "learning_rate": 4.176954904618913e-05, "loss": 0.3773, "step": 4053 }, { "epoch": 0.71, "learning_rate": 4.1764255729226676e-05, "loss": 0.3437, "step": 4054 }, { "epoch": 0.71, "learning_rate": 4.1758961091927765e-05, "loss": 0.3724, "step": 4055 }, { "epoch": 0.71, "learning_rate": 4.175366513478247e-05, "loss": 0.3533, "step": 4056 }, { "epoch": 0.71, "learning_rate": 4.174836785828101e-05, "loss": 0.3626, "step": 4057 }, { "epoch": 0.71, "learning_rate": 4.1743069262913706e-05, "loss": 0.3756, "step": 4058 }, { "epoch": 0.71, "learning_rate": 4.173776934917102e-05, "loss": 0.3288, "step": 4059 }, { "epoch": 0.71, "learning_rate": 4.173246811754351e-05, "loss": 0.3498, "step": 4060 }, { "epoch": 0.71, "learning_rate": 4.172716556852189e-05, "loss": 0.3539, "step": 4061 }, { "epoch": 0.71, "learning_rate": 4.172186170259697e-05, "loss": 0.3579, "step": 4062 }, { "epoch": 0.71, "learning_rate": 4.1716556520259684e-05, "loss": 0.3781, "step": 4063 }, { "epoch": 0.71, "learning_rate": 4.171125002200109e-05, "loss": 0.3608, "step": 4064 }, { "epoch": 0.71, "learning_rate": 4.170594220831239e-05, "loss": 0.3558, "step": 4065 }, { "epoch": 0.71, "learning_rate": 4.1700633079684884e-05, "loss": 0.3515, "step": 4066 }, { "epoch": 0.71, "learning_rate": 4.169532263660998e-05, "loss": 0.3725, "step": 4067 }, { "epoch": 0.71, "learning_rate": 4.169001087957924e-05, "loss": 0.3563, "step": 4068 }, { "epoch": 0.71, "learning_rate": 4.168469780908434e-05, "loss": 0.3553, "step": 4069 }, { "epoch": 0.71, "learning_rate": 4.167938342561706e-05, "loss": 0.3507, "step": 4070 }, { "epoch": 0.71, "learning_rate": 4.167406772966933e-05, "loss": 0.373, "step": 4071 }, { "epoch": 0.71, "learning_rate": 4.1668750721733166e-05, "loss": 0.3562, "step": 4072 }, { "epoch": 0.71, "learning_rate": 4.166343240230072e-05, "loss": 0.353, "step": 4073 }, { "epoch": 0.71, "learning_rate": 4.165811277186429e-05, "loss": 0.3317, "step": 4074 }, { "epoch": 0.71, "learning_rate": 4.1652791830916266e-05, "loss": 0.3599, "step": 4075 }, { "epoch": 0.71, "learning_rate": 4.164746957994916e-05, "loss": 0.3504, "step": 4076 }, { "epoch": 0.71, "learning_rate": 4.1642146019455633e-05, "loss": 0.3591, "step": 4077 }, { "epoch": 0.71, "learning_rate": 4.163682114992842e-05, "loss": 0.3559, "step": 4078 }, { "epoch": 0.71, "learning_rate": 4.163149497186043e-05, "loss": 0.3592, "step": 4079 }, { "epoch": 0.71, "learning_rate": 4.162616748574465e-05, "loss": 0.3404, "step": 4080 }, { "epoch": 0.71, "learning_rate": 4.162083869207422e-05, "loss": 0.3522, "step": 4081 }, { "epoch": 0.71, "learning_rate": 4.161550859134238e-05, "loss": 0.3457, "step": 4082 }, { "epoch": 0.71, "learning_rate": 4.161017718404249e-05, "loss": 0.3546, "step": 4083 }, { "epoch": 0.71, "learning_rate": 4.160484447066805e-05, "loss": 0.3352, "step": 4084 }, { "epoch": 0.71, "learning_rate": 4.159951045171267e-05, "loss": 0.3552, "step": 4085 }, { "epoch": 0.71, "learning_rate": 4.159417512767008e-05, "loss": 0.3557, "step": 4086 }, { "epoch": 0.71, "learning_rate": 4.1588838499034124e-05, "loss": 0.3483, "step": 4087 }, { "epoch": 0.71, "learning_rate": 4.158350056629878e-05, "loss": 0.3479, "step": 4088 }, { "epoch": 0.71, "learning_rate": 4.1578161329958135e-05, "loss": 0.336, "step": 4089 }, { "epoch": 0.71, "learning_rate": 4.157282079050642e-05, "loss": 0.3512, "step": 4090 }, { "epoch": 0.71, "learning_rate": 4.156747894843794e-05, "loss": 0.3491, "step": 4091 }, { "epoch": 0.71, "learning_rate": 4.1562135804247174e-05, "loss": 0.3504, "step": 4092 }, { "epoch": 0.71, "learning_rate": 4.15567913584287e-05, "loss": 0.3572, "step": 4093 }, { "epoch": 0.71, "learning_rate": 4.155144561147719e-05, "loss": 0.3751, "step": 4094 }, { "epoch": 0.71, "learning_rate": 4.154609856388748e-05, "loss": 0.3647, "step": 4095 }, { "epoch": 0.71, "learning_rate": 4.15407502161545e-05, "loss": 0.3325, "step": 4096 }, { "epoch": 0.71, "learning_rate": 4.1535400568773295e-05, "loss": 0.3429, "step": 4097 }, { "epoch": 0.71, "learning_rate": 4.153004962223907e-05, "loss": 0.3608, "step": 4098 }, { "epoch": 0.71, "learning_rate": 4.15246973770471e-05, "loss": 0.3467, "step": 4099 }, { "epoch": 0.71, "learning_rate": 4.15193438336928e-05, "loss": 0.3653, "step": 4100 }, { "epoch": 0.71, "learning_rate": 4.1513988992671734e-05, "loss": 0.3472, "step": 4101 }, { "epoch": 0.71, "learning_rate": 4.150863285447954e-05, "loss": 0.3478, "step": 4102 }, { "epoch": 0.72, "learning_rate": 4.150327541961199e-05, "loss": 0.3433, "step": 4103 }, { "epoch": 0.72, "learning_rate": 4.1497916688565e-05, "loss": 0.3588, "step": 4104 }, { "epoch": 0.72, "learning_rate": 4.1492556661834577e-05, "loss": 0.3479, "step": 4105 }, { "epoch": 0.72, "learning_rate": 4.1487195339916854e-05, "loss": 0.3535, "step": 4106 }, { "epoch": 0.72, "learning_rate": 4.1481832723308106e-05, "loss": 0.3397, "step": 4107 }, { "epoch": 0.72, "learning_rate": 4.147646881250471e-05, "loss": 0.3405, "step": 4108 }, { "epoch": 0.72, "learning_rate": 4.1471103608003134e-05, "loss": 0.3524, "step": 4109 }, { "epoch": 0.72, "learning_rate": 4.146573711030002e-05, "loss": 0.3606, "step": 4110 }, { "epoch": 0.72, "learning_rate": 4.14603693198921e-05, "loss": 0.353, "step": 4111 }, { "epoch": 0.72, "learning_rate": 4.145500023727624e-05, "loss": 0.353, "step": 4112 }, { "epoch": 0.72, "learning_rate": 4.1449629862949404e-05, "loss": 0.3562, "step": 4113 }, { "epoch": 0.72, "learning_rate": 4.144425819740868e-05, "loss": 0.3594, "step": 4114 }, { "epoch": 0.72, "learning_rate": 4.143888524115131e-05, "loss": 0.3454, "step": 4115 }, { "epoch": 0.72, "learning_rate": 4.143351099467461e-05, "loss": 0.3466, "step": 4116 }, { "epoch": 0.72, "learning_rate": 4.1428135458476024e-05, "loss": 0.3562, "step": 4117 }, { "epoch": 0.72, "learning_rate": 4.142275863305315e-05, "loss": 0.3402, "step": 4118 }, { "epoch": 0.72, "learning_rate": 4.141738051890367e-05, "loss": 0.3587, "step": 4119 }, { "epoch": 0.72, "learning_rate": 4.141200111652539e-05, "loss": 0.3352, "step": 4120 }, { "epoch": 0.72, "learning_rate": 4.140662042641625e-05, "loss": 0.3564, "step": 4121 }, { "epoch": 0.72, "learning_rate": 4.14012384490743e-05, "loss": 0.3522, "step": 4122 }, { "epoch": 0.72, "learning_rate": 4.139585518499771e-05, "loss": 0.3431, "step": 4123 }, { "epoch": 0.72, "learning_rate": 4.1390470634684756e-05, "loss": 0.3415, "step": 4124 }, { "epoch": 0.72, "learning_rate": 4.1385084798633865e-05, "loss": 0.3517, "step": 4125 }, { "epoch": 0.72, "learning_rate": 4.1379697677343555e-05, "loss": 0.3593, "step": 4126 }, { "epoch": 0.72, "learning_rate": 4.137430927131248e-05, "loss": 0.3566, "step": 4127 }, { "epoch": 0.72, "learning_rate": 4.136891958103938e-05, "loss": 0.3586, "step": 4128 }, { "epoch": 0.72, "learning_rate": 4.136352860702316e-05, "loss": 0.3823, "step": 4129 }, { "epoch": 0.72, "learning_rate": 4.135813634976282e-05, "loss": 0.3566, "step": 4130 }, { "epoch": 0.72, "learning_rate": 4.135274280975748e-05, "loss": 0.3425, "step": 4131 }, { "epoch": 0.72, "learning_rate": 4.1347347987506394e-05, "loss": 0.3587, "step": 4132 }, { "epoch": 0.72, "learning_rate": 4.134195188350888e-05, "loss": 0.3666, "step": 4133 }, { "epoch": 0.72, "learning_rate": 4.133655449826446e-05, "loss": 0.3617, "step": 4134 }, { "epoch": 0.72, "learning_rate": 4.1331155832272705e-05, "loss": 0.3582, "step": 4135 }, { "epoch": 0.72, "learning_rate": 4.132575588603334e-05, "loss": 0.3647, "step": 4136 }, { "epoch": 0.72, "learning_rate": 4.1320354660046195e-05, "loss": 0.3688, "step": 4137 }, { "epoch": 0.72, "learning_rate": 4.131495215481122e-05, "loss": 0.3518, "step": 4138 }, { "epoch": 0.72, "learning_rate": 4.130954837082848e-05, "loss": 0.3232, "step": 4139 }, { "epoch": 0.72, "learning_rate": 4.130414330859817e-05, "loss": 0.3545, "step": 4140 }, { "epoch": 0.72, "learning_rate": 4.12987369686206e-05, "loss": 0.3507, "step": 4141 }, { "epoch": 0.72, "learning_rate": 4.1293329351396185e-05, "loss": 0.3422, "step": 4142 }, { "epoch": 0.72, "learning_rate": 4.1287920457425476e-05, "loss": 0.3537, "step": 4143 }, { "epoch": 0.72, "learning_rate": 4.128251028720913e-05, "loss": 0.3514, "step": 4144 }, { "epoch": 0.72, "learning_rate": 4.127709884124792e-05, "loss": 0.3519, "step": 4145 }, { "epoch": 0.72, "learning_rate": 4.127168612004277e-05, "loss": 0.3664, "step": 4146 }, { "epoch": 0.72, "learning_rate": 4.126627212409467e-05, "loss": 0.3366, "step": 4147 }, { "epoch": 0.72, "learning_rate": 4.126085685390475e-05, "loss": 0.3471, "step": 4148 }, { "epoch": 0.72, "learning_rate": 4.125544030997428e-05, "loss": 0.3608, "step": 4149 }, { "epoch": 0.72, "learning_rate": 4.125002249280462e-05, "loss": 0.3431, "step": 4150 }, { "epoch": 0.72, "learning_rate": 4.124460340289726e-05, "loss": 0.3625, "step": 4151 }, { "epoch": 0.72, "learning_rate": 4.12391830407538e-05, "loss": 0.3481, "step": 4152 }, { "epoch": 0.72, "learning_rate": 4.123376140687597e-05, "loss": 0.3475, "step": 4153 }, { "epoch": 0.72, "learning_rate": 4.1228338501765605e-05, "loss": 0.3469, "step": 4154 }, { "epoch": 0.72, "learning_rate": 4.1222914325924664e-05, "loss": 0.3527, "step": 4155 }, { "epoch": 0.72, "learning_rate": 4.121748887985522e-05, "loss": 0.3541, "step": 4156 }, { "epoch": 0.72, "learning_rate": 4.1212062164059474e-05, "loss": 0.358, "step": 4157 }, { "epoch": 0.72, "learning_rate": 4.1206634179039725e-05, "loss": 0.3544, "step": 4158 }, { "epoch": 0.72, "learning_rate": 4.120120492529842e-05, "loss": 0.3572, "step": 4159 }, { "epoch": 0.73, "learning_rate": 4.119577440333809e-05, "loss": 0.3546, "step": 4160 }, { "epoch": 0.73, "learning_rate": 4.1190342613661404e-05, "loss": 0.3415, "step": 4161 }, { "epoch": 0.73, "learning_rate": 4.118490955677113e-05, "loss": 0.3502, "step": 4162 }, { "epoch": 0.73, "learning_rate": 4.1179475233170185e-05, "loss": 0.3507, "step": 4163 }, { "epoch": 0.73, "learning_rate": 4.1174039643361574e-05, "loss": 0.3508, "step": 4164 }, { "epoch": 0.73, "learning_rate": 4.1168602787848434e-05, "loss": 0.3575, "step": 4165 }, { "epoch": 0.73, "learning_rate": 4.1163164667134006e-05, "loss": 0.3425, "step": 4166 }, { "epoch": 0.73, "learning_rate": 4.115772528172166e-05, "loss": 0.3381, "step": 4167 }, { "epoch": 0.73, "learning_rate": 4.115228463211489e-05, "loss": 0.3477, "step": 4168 }, { "epoch": 0.73, "learning_rate": 4.114684271881728e-05, "loss": 0.3623, "step": 4169 }, { "epoch": 0.73, "learning_rate": 4.114139954233256e-05, "loss": 0.3536, "step": 4170 }, { "epoch": 0.73, "learning_rate": 4.113595510316456e-05, "loss": 0.3589, "step": 4171 }, { "epoch": 0.73, "learning_rate": 4.1130509401817224e-05, "loss": 0.3613, "step": 4172 }, { "epoch": 0.73, "learning_rate": 4.1125062438794623e-05, "loss": 0.3527, "step": 4173 }, { "epoch": 0.73, "learning_rate": 4.1119614214600964e-05, "loss": 0.3649, "step": 4174 }, { "epoch": 0.73, "learning_rate": 4.111416472974052e-05, "loss": 0.3499, "step": 4175 }, { "epoch": 0.73, "learning_rate": 4.110871398471773e-05, "loss": 0.3631, "step": 4176 }, { "epoch": 0.73, "learning_rate": 4.110326198003711e-05, "loss": 0.3604, "step": 4177 }, { "epoch": 0.73, "learning_rate": 4.109780871620332e-05, "loss": 0.36, "step": 4178 }, { "epoch": 0.73, "learning_rate": 4.109235419372114e-05, "loss": 0.348, "step": 4179 }, { "epoch": 0.73, "learning_rate": 4.108689841309545e-05, "loss": 0.3608, "step": 4180 }, { "epoch": 0.73, "learning_rate": 4.108144137483123e-05, "loss": 0.3896, "step": 4181 }, { "epoch": 0.73, "learning_rate": 4.1075983079433616e-05, "loss": 0.3533, "step": 4182 }, { "epoch": 0.73, "learning_rate": 4.107052352740784e-05, "loss": 0.3501, "step": 4183 }, { "epoch": 0.73, "learning_rate": 4.106506271925926e-05, "loss": 0.362, "step": 4184 }, { "epoch": 0.73, "learning_rate": 4.105960065549332e-05, "loss": 0.3677, "step": 4185 }, { "epoch": 0.73, "learning_rate": 4.105413733661563e-05, "loss": 0.3453, "step": 4186 }, { "epoch": 0.73, "learning_rate": 4.104867276313186e-05, "loss": 0.3524, "step": 4187 }, { "epoch": 0.73, "learning_rate": 4.104320693554785e-05, "loss": 0.3726, "step": 4188 }, { "epoch": 0.73, "learning_rate": 4.103773985436953e-05, "loss": 0.3695, "step": 4189 }, { "epoch": 0.73, "learning_rate": 4.1032271520102927e-05, "loss": 0.3689, "step": 4190 }, { "epoch": 0.73, "learning_rate": 4.102680193325423e-05, "loss": 0.3596, "step": 4191 }, { "epoch": 0.73, "learning_rate": 4.1021331094329686e-05, "loss": 0.3598, "step": 4192 }, { "epoch": 0.73, "learning_rate": 4.101585900383572e-05, "loss": 0.3614, "step": 4193 }, { "epoch": 0.73, "learning_rate": 4.1010385662278834e-05, "loss": 0.3434, "step": 4194 }, { "epoch": 0.73, "learning_rate": 4.100491107016564e-05, "loss": 0.3509, "step": 4195 }, { "epoch": 0.73, "learning_rate": 4.09994352280029e-05, "loss": 0.3481, "step": 4196 }, { "epoch": 0.73, "learning_rate": 4.0993958136297474e-05, "loss": 0.3457, "step": 4197 }, { "epoch": 0.73, "learning_rate": 4.0988479795556314e-05, "loss": 0.3731, "step": 4198 }, { "epoch": 0.73, "learning_rate": 4.098300020628652e-05, "loss": 0.3681, "step": 4199 }, { "epoch": 0.73, "learning_rate": 4.0977519368995305e-05, "loss": 0.3454, "step": 4200 }, { "epoch": 0.73, "learning_rate": 4.097203728418998e-05, "loss": 0.3678, "step": 4201 }, { "epoch": 0.73, "learning_rate": 4.096655395237799e-05, "loss": 0.35, "step": 4202 }, { "epoch": 0.73, "learning_rate": 4.096106937406688e-05, "loss": 0.3483, "step": 4203 }, { "epoch": 0.73, "learning_rate": 4.095558354976431e-05, "loss": 0.346, "step": 4204 }, { "epoch": 0.73, "learning_rate": 4.095009647997807e-05, "loss": 0.3565, "step": 4205 }, { "epoch": 0.73, "learning_rate": 4.094460816521606e-05, "loss": 0.3751, "step": 4206 }, { "epoch": 0.73, "learning_rate": 4.09391186059863e-05, "loss": 0.3681, "step": 4207 }, { "epoch": 0.73, "learning_rate": 4.093362780279689e-05, "loss": 0.3333, "step": 4208 }, { "epoch": 0.73, "learning_rate": 4.0928135756156105e-05, "loss": 0.3327, "step": 4209 }, { "epoch": 0.73, "learning_rate": 4.092264246657228e-05, "loss": 0.3584, "step": 4210 }, { "epoch": 0.73, "learning_rate": 4.09171479345539e-05, "loss": 0.3568, "step": 4211 }, { "epoch": 0.73, "learning_rate": 4.091165216060955e-05, "loss": 0.3544, "step": 4212 }, { "epoch": 0.73, "learning_rate": 4.0906155145247924e-05, "loss": 0.3809, "step": 4213 }, { "epoch": 0.73, "learning_rate": 4.090065688897786e-05, "loss": 0.359, "step": 4214 }, { "epoch": 0.73, "learning_rate": 4.0895157392308266e-05, "loss": 0.3477, "step": 4215 }, { "epoch": 0.73, "learning_rate": 4.088965665574822e-05, "loss": 0.3494, "step": 4216 }, { "epoch": 0.74, "learning_rate": 4.0884154679806856e-05, "loss": 0.3527, "step": 4217 }, { "epoch": 0.74, "learning_rate": 4.0878651464993466e-05, "loss": 0.3501, "step": 4218 }, { "epoch": 0.74, "learning_rate": 4.087314701181744e-05, "loss": 0.3676, "step": 4219 }, { "epoch": 0.74, "learning_rate": 4.0867641320788284e-05, "loss": 0.3403, "step": 4220 }, { "epoch": 0.74, "learning_rate": 4.0862134392415615e-05, "loss": 0.3791, "step": 4221 }, { "epoch": 0.74, "learning_rate": 4.085662622720917e-05, "loss": 0.3402, "step": 4222 }, { "epoch": 0.74, "learning_rate": 4.085111682567881e-05, "loss": 0.3506, "step": 4223 }, { "epoch": 0.74, "learning_rate": 4.084560618833448e-05, "loss": 0.3599, "step": 4224 }, { "epoch": 0.74, "learning_rate": 4.084009431568628e-05, "loss": 0.3508, "step": 4225 }, { "epoch": 0.74, "learning_rate": 4.0834581208244377e-05, "loss": 0.3588, "step": 4226 }, { "epoch": 0.74, "learning_rate": 4.082906686651911e-05, "loss": 0.3483, "step": 4227 }, { "epoch": 0.74, "learning_rate": 4.082355129102087e-05, "loss": 0.3682, "step": 4228 }, { "epoch": 0.74, "learning_rate": 4.0818034482260215e-05, "loss": 0.3494, "step": 4229 }, { "epoch": 0.74, "learning_rate": 4.0812516440747787e-05, "loss": 0.3561, "step": 4230 }, { "epoch": 0.74, "learning_rate": 4.080699716699435e-05, "loss": 0.3432, "step": 4231 }, { "epoch": 0.74, "learning_rate": 4.080147666151078e-05, "loss": 0.3596, "step": 4232 }, { "epoch": 0.74, "learning_rate": 4.0795954924808076e-05, "loss": 0.3521, "step": 4233 }, { "epoch": 0.74, "learning_rate": 4.079043195739734e-05, "loss": 0.3621, "step": 4234 }, { "epoch": 0.74, "learning_rate": 4.078490775978979e-05, "loss": 0.3604, "step": 4235 }, { "epoch": 0.74, "learning_rate": 4.0779382332496775e-05, "loss": 0.3675, "step": 4236 }, { "epoch": 0.74, "learning_rate": 4.077385567602972e-05, "loss": 0.3523, "step": 4237 }, { "epoch": 0.74, "learning_rate": 4.07683277909002e-05, "loss": 0.3382, "step": 4238 }, { "epoch": 0.74, "learning_rate": 4.076279867761988e-05, "loss": 0.3367, "step": 4239 }, { "epoch": 0.74, "learning_rate": 4.075726833670057e-05, "loss": 0.3676, "step": 4240 }, { "epoch": 0.74, "learning_rate": 4.0751736768654146e-05, "loss": 0.3595, "step": 4241 }, { "epoch": 0.74, "learning_rate": 4.074620397399265e-05, "loss": 0.3462, "step": 4242 }, { "epoch": 0.74, "learning_rate": 4.0740669953228195e-05, "loss": 0.3555, "step": 4243 }, { "epoch": 0.74, "learning_rate": 4.0735134706873025e-05, "loss": 0.3449, "step": 4244 }, { "epoch": 0.74, "learning_rate": 4.07295982354395e-05, "loss": 0.336, "step": 4245 }, { "epoch": 0.74, "learning_rate": 4.07240605394401e-05, "loss": 0.3611, "step": 4246 }, { "epoch": 0.74, "learning_rate": 4.071852161938739e-05, "loss": 0.3612, "step": 4247 }, { "epoch": 0.74, "learning_rate": 4.071298147579408e-05, "loss": 0.3443, "step": 4248 }, { "epoch": 0.74, "learning_rate": 4.070744010917297e-05, "loss": 0.3519, "step": 4249 }, { "epoch": 0.74, "learning_rate": 4.0701897520037e-05, "loss": 0.3587, "step": 4250 }, { "epoch": 0.74, "learning_rate": 4.069635370889919e-05, "loss": 0.3529, "step": 4251 }, { "epoch": 0.74, "learning_rate": 4.069080867627269e-05, "loss": 0.3555, "step": 4252 }, { "epoch": 0.74, "learning_rate": 4.068526242267078e-05, "loss": 0.3501, "step": 4253 }, { "epoch": 0.74, "learning_rate": 4.0679714948606815e-05, "loss": 0.3571, "step": 4254 }, { "epoch": 0.74, "learning_rate": 4.067416625459429e-05, "loss": 0.3431, "step": 4255 }, { "epoch": 0.74, "learning_rate": 4.066861634114681e-05, "loss": 0.3656, "step": 4256 }, { "epoch": 0.74, "learning_rate": 4.066306520877809e-05, "loss": 0.3616, "step": 4257 }, { "epoch": 0.74, "learning_rate": 4.065751285800196e-05, "loss": 0.3523, "step": 4258 }, { "epoch": 0.74, "learning_rate": 4.065195928933235e-05, "loss": 0.3587, "step": 4259 }, { "epoch": 0.74, "learning_rate": 4.0646404503283317e-05, "loss": 0.3466, "step": 4260 }, { "epoch": 0.74, "learning_rate": 4.064084850036902e-05, "loss": 0.3465, "step": 4261 }, { "epoch": 0.74, "learning_rate": 4.063529128110376e-05, "loss": 0.3528, "step": 4262 }, { "epoch": 0.74, "learning_rate": 4.0629732846001894e-05, "loss": 0.342, "step": 4263 }, { "epoch": 0.74, "learning_rate": 4.062417319557796e-05, "loss": 0.34, "step": 4264 }, { "epoch": 0.74, "learning_rate": 4.0618612330346546e-05, "loss": 0.3504, "step": 4265 }, { "epoch": 0.74, "learning_rate": 4.061305025082239e-05, "loss": 0.3396, "step": 4266 }, { "epoch": 0.74, "learning_rate": 4.060748695752034e-05, "loss": 0.3534, "step": 4267 }, { "epoch": 0.74, "learning_rate": 4.0601922450955335e-05, "loss": 0.3378, "step": 4268 }, { "epoch": 0.74, "learning_rate": 4.059635673164245e-05, "loss": 0.3613, "step": 4269 }, { "epoch": 0.74, "learning_rate": 4.0590789800096854e-05, "loss": 0.3445, "step": 4270 }, { "epoch": 0.74, "learning_rate": 4.058522165683385e-05, "loss": 0.3506, "step": 4271 }, { "epoch": 0.74, "learning_rate": 4.057965230236883e-05, "loss": 0.3463, "step": 4272 }, { "epoch": 0.74, "learning_rate": 4.057408173721731e-05, "loss": 0.3632, "step": 4273 }, { "epoch": 0.74, "learning_rate": 4.056850996189491e-05, "loss": 0.3565, "step": 4274 }, { "epoch": 0.75, "learning_rate": 4.056293697691739e-05, "loss": 0.3367, "step": 4275 }, { "epoch": 0.75, "learning_rate": 4.0557362782800564e-05, "loss": 0.3659, "step": 4276 }, { "epoch": 0.75, "learning_rate": 4.055178738006042e-05, "loss": 0.3429, "step": 4277 }, { "epoch": 0.75, "learning_rate": 4.054621076921303e-05, "loss": 0.3634, "step": 4278 }, { "epoch": 0.75, "learning_rate": 4.054063295077457e-05, "loss": 0.3362, "step": 4279 }, { "epoch": 0.75, "learning_rate": 4.053505392526134e-05, "loss": 0.3403, "step": 4280 }, { "epoch": 0.75, "learning_rate": 4.052947369318975e-05, "loss": 0.3518, "step": 4281 }, { "epoch": 0.75, "learning_rate": 4.0523892255076327e-05, "loss": 0.3374, "step": 4282 }, { "epoch": 0.75, "learning_rate": 4.051830961143769e-05, "loss": 0.355, "step": 4283 }, { "epoch": 0.75, "learning_rate": 4.0512725762790605e-05, "loss": 0.3537, "step": 4284 }, { "epoch": 0.75, "learning_rate": 4.050714070965191e-05, "loss": 0.3592, "step": 4285 }, { "epoch": 0.75, "learning_rate": 4.0501554452538566e-05, "loss": 0.3415, "step": 4286 }, { "epoch": 0.75, "learning_rate": 4.0495966991967664e-05, "loss": 0.3503, "step": 4287 }, { "epoch": 0.75, "learning_rate": 4.049037832845639e-05, "loss": 0.3625, "step": 4288 }, { "epoch": 0.75, "learning_rate": 4.048478846252205e-05, "loss": 0.3739, "step": 4289 }, { "epoch": 0.75, "learning_rate": 4.047919739468206e-05, "loss": 0.3533, "step": 4290 }, { "epoch": 0.75, "learning_rate": 4.047360512545392e-05, "loss": 0.3426, "step": 4291 }, { "epoch": 0.75, "learning_rate": 4.046801165535529e-05, "loss": 0.3559, "step": 4292 }, { "epoch": 0.75, "learning_rate": 4.046241698490391e-05, "loss": 0.3621, "step": 4293 }, { "epoch": 0.75, "learning_rate": 4.0456821114617625e-05, "loss": 0.3548, "step": 4294 }, { "epoch": 0.75, "learning_rate": 4.045122404501442e-05, "loss": 0.3319, "step": 4295 }, { "epoch": 0.75, "learning_rate": 4.044562577661237e-05, "loss": 0.3447, "step": 4296 }, { "epoch": 0.75, "learning_rate": 4.044002630992966e-05, "loss": 0.3441, "step": 4297 }, { "epoch": 0.75, "learning_rate": 4.0434425645484583e-05, "loss": 0.3557, "step": 4298 }, { "epoch": 0.75, "learning_rate": 4.042882378379557e-05, "loss": 0.3571, "step": 4299 }, { "epoch": 0.75, "learning_rate": 4.042322072538113e-05, "loss": 0.3439, "step": 4300 }, { "epoch": 0.75, "learning_rate": 4.0417616470759914e-05, "loss": 0.3528, "step": 4301 }, { "epoch": 0.75, "learning_rate": 4.0412011020450644e-05, "loss": 0.3573, "step": 4302 }, { "epoch": 0.75, "learning_rate": 4.040640437497218e-05, "loss": 0.3489, "step": 4303 }, { "epoch": 0.75, "learning_rate": 4.04007965348435e-05, "loss": 0.3648, "step": 4304 }, { "epoch": 0.75, "learning_rate": 4.039518750058368e-05, "loss": 0.3306, "step": 4305 }, { "epoch": 0.75, "learning_rate": 4.038957727271189e-05, "loss": 0.355, "step": 4306 }, { "epoch": 0.75, "learning_rate": 4.038396585174743e-05, "loss": 0.356, "step": 4307 }, { "epoch": 0.75, "learning_rate": 4.037835323820972e-05, "loss": 0.3499, "step": 4308 }, { "epoch": 0.75, "learning_rate": 4.0372739432618284e-05, "loss": 0.3468, "step": 4309 }, { "epoch": 0.75, "learning_rate": 4.036712443549273e-05, "loss": 0.3655, "step": 4310 }, { "epoch": 0.75, "learning_rate": 4.03615082473528e-05, "loss": 0.3404, "step": 4311 }, { "epoch": 0.75, "learning_rate": 4.0355890868718356e-05, "loss": 0.3443, "step": 4312 }, { "epoch": 0.75, "learning_rate": 4.0350272300109346e-05, "loss": 0.3519, "step": 4313 }, { "epoch": 0.75, "learning_rate": 4.034465254204585e-05, "loss": 0.3396, "step": 4314 }, { "epoch": 0.75, "learning_rate": 4.0339031595048024e-05, "loss": 0.352, "step": 4315 }, { "epoch": 0.75, "learning_rate": 4.033340945963619e-05, "loss": 0.3421, "step": 4316 }, { "epoch": 0.75, "learning_rate": 4.032778613633073e-05, "loss": 0.3537, "step": 4317 }, { "epoch": 0.75, "learning_rate": 4.032216162565215e-05, "loss": 0.3556, "step": 4318 }, { "epoch": 0.75, "learning_rate": 4.0316535928121066e-05, "loss": 0.3452, "step": 4319 }, { "epoch": 0.75, "learning_rate": 4.031090904425823e-05, "loss": 0.3562, "step": 4320 }, { "epoch": 0.75, "learning_rate": 4.030528097458446e-05, "loss": 0.3593, "step": 4321 }, { "epoch": 0.75, "learning_rate": 4.029965171962071e-05, "loss": 0.3445, "step": 4322 }, { "epoch": 0.75, "learning_rate": 4.029402127988804e-05, "loss": 0.3718, "step": 4323 }, { "epoch": 0.75, "learning_rate": 4.028838965590761e-05, "loss": 0.3483, "step": 4324 }, { "epoch": 0.75, "learning_rate": 4.028275684820071e-05, "loss": 0.3548, "step": 4325 }, { "epoch": 0.75, "learning_rate": 4.027712285728872e-05, "loss": 0.3498, "step": 4326 }, { "epoch": 0.75, "learning_rate": 4.027148768369314e-05, "loss": 0.353, "step": 4327 }, { "epoch": 0.75, "learning_rate": 4.0265851327935585e-05, "loss": 0.3528, "step": 4328 }, { "epoch": 0.75, "learning_rate": 4.026021379053775e-05, "loss": 0.3497, "step": 4329 }, { "epoch": 0.75, "learning_rate": 4.025457507202148e-05, "loss": 0.3577, "step": 4330 }, { "epoch": 0.75, "learning_rate": 4.024893517290869e-05, "loss": 0.3435, "step": 4331 }, { "epoch": 0.76, "learning_rate": 4.0243294093721436e-05, "loss": 0.3356, "step": 4332 }, { "epoch": 0.76, "learning_rate": 4.0237651834981866e-05, "loss": 0.3424, "step": 4333 }, { "epoch": 0.76, "learning_rate": 4.023200839721226e-05, "loss": 0.3624, "step": 4334 }, { "epoch": 0.76, "learning_rate": 4.022636378093496e-05, "loss": 0.3471, "step": 4335 }, { "epoch": 0.76, "learning_rate": 4.022071798667246e-05, "loss": 0.3515, "step": 4336 }, { "epoch": 0.76, "learning_rate": 4.021507101494735e-05, "loss": 0.3357, "step": 4337 }, { "epoch": 0.76, "learning_rate": 4.020942286628233e-05, "loss": 0.3456, "step": 4338 }, { "epoch": 0.76, "learning_rate": 4.0203773541200205e-05, "loss": 0.3689, "step": 4339 }, { "epoch": 0.76, "learning_rate": 4.0198123040223894e-05, "loss": 0.3536, "step": 4340 }, { "epoch": 0.76, "learning_rate": 4.0192471363876415e-05, "loss": 0.3471, "step": 4341 }, { "epoch": 0.76, "learning_rate": 4.018681851268091e-05, "loss": 0.3502, "step": 4342 }, { "epoch": 0.76, "learning_rate": 4.018116448716062e-05, "loss": 0.355, "step": 4343 }, { "epoch": 0.76, "learning_rate": 4.01755092878389e-05, "loss": 0.3582, "step": 4344 }, { "epoch": 0.76, "learning_rate": 4.016985291523919e-05, "loss": 0.3579, "step": 4345 }, { "epoch": 0.76, "learning_rate": 4.0164195369885075e-05, "loss": 0.34, "step": 4346 }, { "epoch": 0.76, "learning_rate": 4.015853665230024e-05, "loss": 0.3699, "step": 4347 }, { "epoch": 0.76, "learning_rate": 4.015287676300845e-05, "loss": 0.3702, "step": 4348 }, { "epoch": 0.76, "learning_rate": 4.0147215702533614e-05, "loss": 0.3484, "step": 4349 }, { "epoch": 0.76, "learning_rate": 4.014155347139972e-05, "loss": 0.3322, "step": 4350 }, { "epoch": 0.76, "learning_rate": 4.01358900701309e-05, "loss": 0.3614, "step": 4351 }, { "epoch": 0.76, "learning_rate": 4.013022549925136e-05, "loss": 0.355, "step": 4352 }, { "epoch": 0.76, "learning_rate": 4.012455975928542e-05, "loss": 0.3592, "step": 4353 }, { "epoch": 0.76, "learning_rate": 4.011889285075754e-05, "loss": 0.3491, "step": 4354 }, { "epoch": 0.76, "learning_rate": 4.011322477419224e-05, "loss": 0.3398, "step": 4355 }, { "epoch": 0.76, "learning_rate": 4.010755553011418e-05, "loss": 0.3573, "step": 4356 }, { "epoch": 0.76, "learning_rate": 4.010188511904812e-05, "loss": 0.3592, "step": 4357 }, { "epoch": 0.76, "learning_rate": 4.0096213541518924e-05, "loss": 0.3392, "step": 4358 }, { "epoch": 0.76, "learning_rate": 4.0090540798051574e-05, "loss": 0.3635, "step": 4359 }, { "epoch": 0.76, "learning_rate": 4.0084866889171156e-05, "loss": 0.3582, "step": 4360 }, { "epoch": 0.76, "learning_rate": 4.007919181540284e-05, "loss": 0.367, "step": 4361 }, { "epoch": 0.76, "learning_rate": 4.0073515577271964e-05, "loss": 0.3507, "step": 4362 }, { "epoch": 0.76, "learning_rate": 4.006783817530391e-05, "loss": 0.3553, "step": 4363 }, { "epoch": 0.76, "learning_rate": 4.006215961002419e-05, "loss": 0.3322, "step": 4364 }, { "epoch": 0.76, "learning_rate": 4.005647988195844e-05, "loss": 0.3411, "step": 4365 }, { "epoch": 0.76, "learning_rate": 4.005079899163238e-05, "loss": 0.3673, "step": 4366 }, { "epoch": 0.76, "learning_rate": 4.004511693957186e-05, "loss": 0.3673, "step": 4367 }, { "epoch": 0.76, "learning_rate": 4.0039433726302805e-05, "loss": 0.3446, "step": 4368 }, { "epoch": 0.76, "learning_rate": 4.003374935235128e-05, "loss": 0.3479, "step": 4369 }, { "epoch": 0.76, "learning_rate": 4.002806381824346e-05, "loss": 0.3397, "step": 4370 }, { "epoch": 0.76, "learning_rate": 4.002237712450559e-05, "loss": 0.3713, "step": 4371 }, { "epoch": 0.76, "learning_rate": 4.001668927166406e-05, "loss": 0.3656, "step": 4372 }, { "epoch": 0.76, "learning_rate": 4.001100026024534e-05, "loss": 0.3618, "step": 4373 }, { "epoch": 0.76, "learning_rate": 4.000531009077603e-05, "loss": 0.3542, "step": 4374 }, { "epoch": 0.76, "learning_rate": 3.9999618763782814e-05, "loss": 0.3526, "step": 4375 }, { "epoch": 0.76, "learning_rate": 3.9993926279792514e-05, "loss": 0.362, "step": 4376 }, { "epoch": 0.76, "learning_rate": 3.998823263933203e-05, "loss": 0.3498, "step": 4377 }, { "epoch": 0.76, "learning_rate": 3.998253784292838e-05, "loss": 0.3531, "step": 4378 }, { "epoch": 0.76, "learning_rate": 3.9976841891108685e-05, "loss": 0.3596, "step": 4379 }, { "epoch": 0.76, "learning_rate": 3.997114478440019e-05, "loss": 0.3359, "step": 4380 }, { "epoch": 0.76, "learning_rate": 3.996544652333023e-05, "loss": 0.3721, "step": 4381 }, { "epoch": 0.76, "learning_rate": 3.9959747108426246e-05, "loss": 0.345, "step": 4382 }, { "epoch": 0.76, "learning_rate": 3.995404654021579e-05, "loss": 0.3641, "step": 4383 }, { "epoch": 0.76, "learning_rate": 3.994834481922652e-05, "loss": 0.361, "step": 4384 }, { "epoch": 0.76, "learning_rate": 3.99426419459862e-05, "loss": 0.3435, "step": 4385 }, { "epoch": 0.76, "learning_rate": 3.993693792102273e-05, "loss": 0.3591, "step": 4386 }, { "epoch": 0.76, "learning_rate": 3.993123274486405e-05, "loss": 0.3528, "step": 4387 }, { "epoch": 0.76, "learning_rate": 3.992552641803827e-05, "loss": 0.3396, "step": 4388 }, { "epoch": 0.77, "learning_rate": 3.991981894107358e-05, "loss": 0.355, "step": 4389 }, { "epoch": 0.77, "learning_rate": 3.991411031449827e-05, "loss": 0.3375, "step": 4390 }, { "epoch": 0.77, "learning_rate": 3.9908400538840754e-05, "loss": 0.3457, "step": 4391 }, { "epoch": 0.77, "learning_rate": 3.990268961462953e-05, "loss": 0.3475, "step": 4392 }, { "epoch": 0.77, "learning_rate": 3.9896977542393236e-05, "loss": 0.3462, "step": 4393 }, { "epoch": 0.77, "learning_rate": 3.989126432266058e-05, "loss": 0.3585, "step": 4394 }, { "epoch": 0.77, "learning_rate": 3.98855499559604e-05, "loss": 0.3765, "step": 4395 }, { "epoch": 0.77, "learning_rate": 3.987983444282163e-05, "loss": 0.3431, "step": 4396 }, { "epoch": 0.77, "learning_rate": 3.987411778377332e-05, "loss": 0.3531, "step": 4397 }, { "epoch": 0.77, "learning_rate": 3.98683999793446e-05, "loss": 0.3447, "step": 4398 }, { "epoch": 0.77, "learning_rate": 3.9862681030064745e-05, "loss": 0.3498, "step": 4399 }, { "epoch": 0.77, "learning_rate": 3.985696093646312e-05, "loss": 0.3388, "step": 4400 }, { "epoch": 0.77, "learning_rate": 3.985123969906916e-05, "loss": 0.3593, "step": 4401 }, { "epoch": 0.77, "learning_rate": 3.9845517318412454e-05, "loss": 0.342, "step": 4402 }, { "epoch": 0.77, "learning_rate": 3.9839793795022704e-05, "loss": 0.3386, "step": 4403 }, { "epoch": 0.77, "learning_rate": 3.983406912942966e-05, "loss": 0.3486, "step": 4404 }, { "epoch": 0.77, "learning_rate": 3.982834332216323e-05, "loss": 0.3577, "step": 4405 }, { "epoch": 0.77, "learning_rate": 3.982261637375341e-05, "loss": 0.3425, "step": 4406 }, { "epoch": 0.77, "learning_rate": 3.981688828473028e-05, "loss": 0.3468, "step": 4407 }, { "epoch": 0.77, "learning_rate": 3.981115905562408e-05, "loss": 0.351, "step": 4408 }, { "epoch": 0.77, "learning_rate": 3.9805428686965106e-05, "loss": 0.3538, "step": 4409 }, { "epoch": 0.77, "learning_rate": 3.979969717928376e-05, "loss": 0.365, "step": 4410 }, { "epoch": 0.77, "learning_rate": 3.979396453311059e-05, "loss": 0.3428, "step": 4411 }, { "epoch": 0.77, "learning_rate": 3.9788230748976214e-05, "loss": 0.3616, "step": 4412 }, { "epoch": 0.77, "learning_rate": 3.978249582741136e-05, "loss": 0.3522, "step": 4413 }, { "epoch": 0.77, "learning_rate": 3.9776759768946894e-05, "loss": 0.3369, "step": 4414 }, { "epoch": 0.77, "learning_rate": 3.9771022574113723e-05, "loss": 0.3333, "step": 4415 }, { "epoch": 0.77, "learning_rate": 3.976528424344292e-05, "loss": 0.3528, "step": 4416 }, { "epoch": 0.77, "learning_rate": 3.975954477746563e-05, "loss": 0.3365, "step": 4417 }, { "epoch": 0.77, "learning_rate": 3.975380417671312e-05, "loss": 0.3732, "step": 4418 }, { "epoch": 0.77, "learning_rate": 3.974806244171675e-05, "loss": 0.342, "step": 4419 }, { "epoch": 0.77, "learning_rate": 3.974231957300799e-05, "loss": 0.3367, "step": 4420 }, { "epoch": 0.77, "learning_rate": 3.973657557111842e-05, "loss": 0.3559, "step": 4421 }, { "epoch": 0.77, "learning_rate": 3.973083043657971e-05, "loss": 0.3485, "step": 4422 }, { "epoch": 0.77, "learning_rate": 3.972508416992365e-05, "loss": 0.3527, "step": 4423 }, { "epoch": 0.77, "learning_rate": 3.971933677168214e-05, "loss": 0.3688, "step": 4424 }, { "epoch": 0.77, "learning_rate": 3.9713588242387164e-05, "loss": 0.3524, "step": 4425 }, { "epoch": 0.77, "learning_rate": 3.970783858257081e-05, "loss": 0.3452, "step": 4426 }, { "epoch": 0.77, "learning_rate": 3.97020877927653e-05, "loss": 0.3382, "step": 4427 }, { "epoch": 0.77, "learning_rate": 3.969633587350293e-05, "loss": 0.3313, "step": 4428 }, { "epoch": 0.77, "learning_rate": 3.969058282531611e-05, "loss": 0.3548, "step": 4429 }, { "epoch": 0.77, "learning_rate": 3.9684828648737384e-05, "loss": 0.3537, "step": 4430 }, { "epoch": 0.77, "learning_rate": 3.967907334429934e-05, "loss": 0.3518, "step": 4431 }, { "epoch": 0.77, "learning_rate": 3.967331691253471e-05, "loss": 0.3442, "step": 4432 }, { "epoch": 0.77, "learning_rate": 3.966755935397634e-05, "loss": 0.3655, "step": 4433 }, { "epoch": 0.77, "learning_rate": 3.9661800669157154e-05, "loss": 0.3524, "step": 4434 }, { "epoch": 0.77, "learning_rate": 3.96560408586102e-05, "loss": 0.3447, "step": 4435 }, { "epoch": 0.77, "learning_rate": 3.96502799228686e-05, "loss": 0.3441, "step": 4436 }, { "epoch": 0.77, "learning_rate": 3.9644517862465626e-05, "loss": 0.3509, "step": 4437 }, { "epoch": 0.77, "learning_rate": 3.963875467793461e-05, "loss": 0.3468, "step": 4438 }, { "epoch": 0.77, "learning_rate": 3.963299036980901e-05, "loss": 0.3428, "step": 4439 }, { "epoch": 0.77, "learning_rate": 3.96272249386224e-05, "loss": 0.3556, "step": 4440 }, { "epoch": 0.77, "learning_rate": 3.9621458384908446e-05, "loss": 0.3525, "step": 4441 }, { "epoch": 0.77, "learning_rate": 3.961569070920088e-05, "loss": 0.349, "step": 4442 }, { "epoch": 0.77, "learning_rate": 3.960992191203361e-05, "loss": 0.3651, "step": 4443 }, { "epoch": 0.77, "learning_rate": 3.9604151993940595e-05, "loss": 0.3497, "step": 4444 }, { "epoch": 0.77, "learning_rate": 3.9598380955455904e-05, "loss": 0.3418, "step": 4445 }, { "epoch": 0.77, "learning_rate": 3.9592608797113744e-05, "loss": 0.3552, "step": 4446 }, { "epoch": 0.78, "learning_rate": 3.958683551944838e-05, "loss": 0.352, "step": 4447 }, { "epoch": 0.78, "learning_rate": 3.958106112299422e-05, "loss": 0.3528, "step": 4448 }, { "epoch": 0.78, "learning_rate": 3.957528560828573e-05, "loss": 0.3422, "step": 4449 }, { "epoch": 0.78, "learning_rate": 3.9569508975857534e-05, "loss": 0.3467, "step": 4450 }, { "epoch": 0.78, "learning_rate": 3.9563731226244316e-05, "loss": 0.3558, "step": 4451 }, { "epoch": 0.78, "learning_rate": 3.955795235998088e-05, "loss": 0.3563, "step": 4452 }, { "epoch": 0.78, "learning_rate": 3.9552172377602145e-05, "loss": 0.3414, "step": 4453 }, { "epoch": 0.78, "learning_rate": 3.9546391279643116e-05, "loss": 0.3356, "step": 4454 }, { "epoch": 0.78, "learning_rate": 3.95406090666389e-05, "loss": 0.3442, "step": 4455 }, { "epoch": 0.78, "learning_rate": 3.953482573912472e-05, "loss": 0.3431, "step": 4456 }, { "epoch": 0.78, "learning_rate": 3.9529041297635884e-05, "loss": 0.3757, "step": 4457 }, { "epoch": 0.78, "learning_rate": 3.9523255742707835e-05, "loss": 0.3389, "step": 4458 }, { "epoch": 0.78, "learning_rate": 3.951746907487607e-05, "loss": 0.3355, "step": 4459 }, { "epoch": 0.78, "learning_rate": 3.9511681294676255e-05, "loss": 0.3522, "step": 4460 }, { "epoch": 0.78, "learning_rate": 3.950589240264409e-05, "loss": 0.3351, "step": 4461 }, { "epoch": 0.78, "learning_rate": 3.950010239931543e-05, "loss": 0.3469, "step": 4462 }, { "epoch": 0.78, "learning_rate": 3.9494311285226204e-05, "loss": 0.3526, "step": 4463 }, { "epoch": 0.78, "learning_rate": 3.9488519060912455e-05, "loss": 0.3397, "step": 4464 }, { "epoch": 0.78, "learning_rate": 3.948272572691032e-05, "loss": 0.3529, "step": 4465 }, { "epoch": 0.78, "learning_rate": 3.947693128375605e-05, "loss": 0.3438, "step": 4466 }, { "epoch": 0.78, "learning_rate": 3.9471135731985996e-05, "loss": 0.3714, "step": 4467 }, { "epoch": 0.78, "learning_rate": 3.9465339072136615e-05, "loss": 0.3364, "step": 4468 }, { "epoch": 0.78, "learning_rate": 3.945954130474444e-05, "loss": 0.3607, "step": 4469 }, { "epoch": 0.78, "learning_rate": 3.9453742430346154e-05, "loss": 0.3616, "step": 4470 }, { "epoch": 0.78, "learning_rate": 3.9447942449478495e-05, "loss": 0.3543, "step": 4471 }, { "epoch": 0.78, "learning_rate": 3.9442141362678324e-05, "loss": 0.3422, "step": 4472 }, { "epoch": 0.78, "learning_rate": 3.943633917048262e-05, "loss": 0.3433, "step": 4473 }, { "epoch": 0.78, "learning_rate": 3.943053587342844e-05, "loss": 0.354, "step": 4474 }, { "epoch": 0.78, "learning_rate": 3.942473147205295e-05, "loss": 0.3477, "step": 4475 }, { "epoch": 0.78, "learning_rate": 3.941892596689343e-05, "loss": 0.3495, "step": 4476 }, { "epoch": 0.78, "learning_rate": 3.941311935848724e-05, "loss": 0.3428, "step": 4477 }, { "epoch": 0.78, "learning_rate": 3.940731164737187e-05, "loss": 0.342, "step": 4478 }, { "epoch": 0.78, "learning_rate": 3.940150283408488e-05, "loss": 0.3357, "step": 4479 }, { "epoch": 0.78, "learning_rate": 3.939569291916396e-05, "loss": 0.3621, "step": 4480 }, { "epoch": 0.78, "learning_rate": 3.938988190314688e-05, "loss": 0.3354, "step": 4481 }, { "epoch": 0.78, "learning_rate": 3.9384069786571535e-05, "loss": 0.3435, "step": 4482 }, { "epoch": 0.78, "learning_rate": 3.93782565699759e-05, "loss": 0.354, "step": 4483 }, { "epoch": 0.78, "learning_rate": 3.9372442253898076e-05, "loss": 0.3506, "step": 4484 }, { "epoch": 0.78, "learning_rate": 3.936662683887624e-05, "loss": 0.3588, "step": 4485 }, { "epoch": 0.78, "learning_rate": 3.936081032544867e-05, "loss": 0.3724, "step": 4486 }, { "epoch": 0.78, "learning_rate": 3.935499271415378e-05, "loss": 0.3464, "step": 4487 }, { "epoch": 0.78, "learning_rate": 3.9349174005530056e-05, "loss": 0.3303, "step": 4488 }, { "epoch": 0.78, "learning_rate": 3.934335420011609e-05, "loss": 0.3544, "step": 4489 }, { "epoch": 0.78, "learning_rate": 3.9337533298450566e-05, "loss": 0.3491, "step": 4490 }, { "epoch": 0.78, "learning_rate": 3.933171130107231e-05, "loss": 0.3667, "step": 4491 }, { "epoch": 0.78, "learning_rate": 3.93258882085202e-05, "loss": 0.3392, "step": 4492 }, { "epoch": 0.78, "learning_rate": 3.932006402133324e-05, "loss": 0.3328, "step": 4493 }, { "epoch": 0.78, "learning_rate": 3.931423874005054e-05, "loss": 0.3448, "step": 4494 }, { "epoch": 0.78, "learning_rate": 3.93084123652113e-05, "loss": 0.3546, "step": 4495 }, { "epoch": 0.78, "learning_rate": 3.9302584897354804e-05, "loss": 0.3377, "step": 4496 }, { "epoch": 0.78, "learning_rate": 3.9296756337020484e-05, "loss": 0.3421, "step": 4497 }, { "epoch": 0.78, "learning_rate": 3.929092668474784e-05, "loss": 0.3385, "step": 4498 }, { "epoch": 0.78, "learning_rate": 3.9285095941076475e-05, "loss": 0.352, "step": 4499 }, { "epoch": 0.78, "learning_rate": 3.92792641065461e-05, "loss": 0.339, "step": 4500 }, { "epoch": 0.78, "learning_rate": 4.355400696864112e-08, "loss": 0.3501, "step": 4501 }, { "epoch": 0.78, "learning_rate": 8.710801393728224e-08, "loss": 0.3363, "step": 4502 }, { "epoch": 0.78, "learning_rate": 1.3066202090592336e-07, "loss": 0.3476, "step": 4503 }, { "epoch": 0.79, "learning_rate": 1.7421602787456448e-07, "loss": 0.3427, "step": 4504 }, { "epoch": 0.79, "learning_rate": 2.1777003484320556e-07, "loss": 0.3505, "step": 4505 }, { "epoch": 0.79, "learning_rate": 2.6132404181184673e-07, "loss": 0.3429, "step": 4506 }, { "epoch": 0.79, "learning_rate": 3.0487804878048784e-07, "loss": 0.3663, "step": 4507 }, { "epoch": 0.79, "learning_rate": 3.4843205574912896e-07, "loss": 0.3373, "step": 4508 }, { "epoch": 0.79, "learning_rate": 3.9198606271777007e-07, "loss": 0.3748, "step": 4509 }, { "epoch": 0.79, "learning_rate": 4.3554006968641113e-07, "loss": 0.3501, "step": 4510 }, { "epoch": 0.79, "learning_rate": 4.790940766550523e-07, "loss": 0.3469, "step": 4511 }, { "epoch": 0.79, "learning_rate": 5.226480836236935e-07, "loss": 0.3388, "step": 4512 }, { "epoch": 0.79, "learning_rate": 5.662020905923345e-07, "loss": 0.3418, "step": 4513 }, { "epoch": 0.79, "learning_rate": 6.097560975609757e-07, "loss": 0.3402, "step": 4514 }, { "epoch": 0.79, "learning_rate": 6.533101045296167e-07, "loss": 0.3537, "step": 4515 }, { "epoch": 0.79, "learning_rate": 6.968641114982579e-07, "loss": 0.3461, "step": 4516 }, { "epoch": 0.79, "learning_rate": 7.40418118466899e-07, "loss": 0.3603, "step": 4517 }, { "epoch": 0.79, "learning_rate": 7.839721254355401e-07, "loss": 0.3317, "step": 4518 }, { "epoch": 0.79, "learning_rate": 8.275261324041812e-07, "loss": 0.3462, "step": 4519 }, { "epoch": 0.79, "learning_rate": 8.710801393728223e-07, "loss": 0.3641, "step": 4520 }, { "epoch": 0.79, "learning_rate": 9.146341463414634e-07, "loss": 0.3583, "step": 4521 }, { "epoch": 0.79, "learning_rate": 9.581881533101046e-07, "loss": 0.3617, "step": 4522 }, { "epoch": 0.79, "learning_rate": 1.0017421602787455e-06, "loss": 0.3469, "step": 4523 }, { "epoch": 0.79, "learning_rate": 1.045296167247387e-06, "loss": 0.3541, "step": 4524 }, { "epoch": 0.79, "learning_rate": 1.088850174216028e-06, "loss": 0.3353, "step": 4525 }, { "epoch": 0.79, "learning_rate": 1.132404181184669e-06, "loss": 0.3386, "step": 4526 }, { "epoch": 0.79, "learning_rate": 1.1759581881533102e-06, "loss": 0.3592, "step": 4527 }, { "epoch": 0.79, "learning_rate": 1.2195121951219514e-06, "loss": 0.3328, "step": 4528 }, { "epoch": 0.79, "learning_rate": 1.2630662020905923e-06, "loss": 0.3522, "step": 4529 }, { "epoch": 0.79, "learning_rate": 1.3066202090592335e-06, "loss": 0.3405, "step": 4530 }, { "epoch": 0.79, "learning_rate": 1.3501742160278747e-06, "loss": 0.3616, "step": 4531 }, { "epoch": 0.79, "learning_rate": 1.3937282229965158e-06, "loss": 0.335, "step": 4532 }, { "epoch": 0.79, "learning_rate": 1.4372822299651568e-06, "loss": 0.3497, "step": 4533 }, { "epoch": 0.79, "learning_rate": 1.480836236933798e-06, "loss": 0.3455, "step": 4534 }, { "epoch": 0.79, "learning_rate": 1.5243902439024391e-06, "loss": 0.3303, "step": 4535 }, { "epoch": 0.79, "learning_rate": 1.5679442508710803e-06, "loss": 0.3607, "step": 4536 }, { "epoch": 0.79, "learning_rate": 1.6114982578397212e-06, "loss": 0.3653, "step": 4537 }, { "epoch": 0.79, "learning_rate": 1.6550522648083624e-06, "loss": 0.3481, "step": 4538 }, { "epoch": 0.79, "learning_rate": 1.6986062717770036e-06, "loss": 0.3411, "step": 4539 }, { "epoch": 0.79, "learning_rate": 1.7421602787456445e-06, "loss": 0.3661, "step": 4540 }, { "epoch": 0.79, "learning_rate": 1.7857142857142857e-06, "loss": 0.3373, "step": 4541 }, { "epoch": 0.79, "learning_rate": 1.8292682926829268e-06, "loss": 0.3583, "step": 4542 }, { "epoch": 0.79, "learning_rate": 1.872822299651568e-06, "loss": 0.3627, "step": 4543 }, { "epoch": 0.79, "learning_rate": 1.916376306620209e-06, "loss": 0.3493, "step": 4544 }, { "epoch": 0.79, "learning_rate": 1.95993031358885e-06, "loss": 0.3346, "step": 4545 }, { "epoch": 0.79, "learning_rate": 2.003484320557491e-06, "loss": 0.3463, "step": 4546 }, { "epoch": 0.79, "learning_rate": 2.0470383275261322e-06, "loss": 0.3492, "step": 4547 }, { "epoch": 0.79, "learning_rate": 2.090592334494774e-06, "loss": 0.3589, "step": 4548 }, { "epoch": 0.79, "learning_rate": 2.134146341463415e-06, "loss": 0.3461, "step": 4549 }, { "epoch": 0.79, "learning_rate": 2.177700348432056e-06, "loss": 0.3478, "step": 4550 }, { "epoch": 0.79, "learning_rate": 2.221254355400697e-06, "loss": 0.3431, "step": 4551 }, { "epoch": 0.79, "learning_rate": 2.264808362369338e-06, "loss": 0.3577, "step": 4552 }, { "epoch": 0.79, "learning_rate": 2.3083623693379792e-06, "loss": 0.3494, "step": 4553 }, { "epoch": 0.79, "learning_rate": 2.3519163763066204e-06, "loss": 0.3487, "step": 4554 }, { "epoch": 0.79, "learning_rate": 2.3954703832752616e-06, "loss": 0.3614, "step": 4555 }, { "epoch": 0.79, "learning_rate": 2.4390243902439027e-06, "loss": 0.3575, "step": 4556 }, { "epoch": 0.79, "learning_rate": 2.482578397212544e-06, "loss": 0.3642, "step": 4557 }, { "epoch": 0.79, "learning_rate": 2.5261324041811846e-06, "loss": 0.3355, "step": 4558 }, { "epoch": 0.79, "learning_rate": 2.569686411149826e-06, "loss": 0.3451, "step": 4559 }, { "epoch": 0.79, "learning_rate": 2.613240418118467e-06, "loss": 0.3444, "step": 4560 }, { "epoch": 0.79, "learning_rate": 2.656794425087108e-06, "loss": 0.3492, "step": 4561 }, { "epoch": 0.8, "learning_rate": 2.7003484320557493e-06, "loss": 0.3328, "step": 4562 }, { "epoch": 0.8, "learning_rate": 2.7439024390243905e-06, "loss": 0.3488, "step": 4563 }, { "epoch": 0.8, "learning_rate": 2.7874564459930316e-06, "loss": 0.3491, "step": 4564 }, { "epoch": 0.8, "learning_rate": 2.8310104529616724e-06, "loss": 0.3597, "step": 4565 }, { "epoch": 0.8, "learning_rate": 2.8745644599303136e-06, "loss": 0.352, "step": 4566 }, { "epoch": 0.8, "learning_rate": 2.9181184668989547e-06, "loss": 0.3465, "step": 4567 }, { "epoch": 0.8, "learning_rate": 2.961672473867596e-06, "loss": 0.3702, "step": 4568 }, { "epoch": 0.8, "learning_rate": 3.005226480836237e-06, "loss": 0.3485, "step": 4569 }, { "epoch": 0.8, "learning_rate": 3.0487804878048782e-06, "loss": 0.3376, "step": 4570 }, { "epoch": 0.8, "learning_rate": 3.0923344947735194e-06, "loss": 0.3432, "step": 4571 }, { "epoch": 0.8, "learning_rate": 3.1358885017421605e-06, "loss": 0.333, "step": 4572 }, { "epoch": 0.8, "learning_rate": 3.1794425087108017e-06, "loss": 0.3398, "step": 4573 }, { "epoch": 0.8, "learning_rate": 3.2229965156794425e-06, "loss": 0.3457, "step": 4574 }, { "epoch": 0.8, "learning_rate": 3.266550522648084e-06, "loss": 0.3529, "step": 4575 }, { "epoch": 0.8, "learning_rate": 3.3101045296167248e-06, "loss": 0.364, "step": 4576 }, { "epoch": 0.8, "learning_rate": 3.3536585365853664e-06, "loss": 0.3532, "step": 4577 }, { "epoch": 0.8, "learning_rate": 3.397212543554007e-06, "loss": 0.3546, "step": 4578 }, { "epoch": 0.8, "learning_rate": 3.4407665505226483e-06, "loss": 0.3437, "step": 4579 }, { "epoch": 0.8, "learning_rate": 3.484320557491289e-06, "loss": 0.3632, "step": 4580 }, { "epoch": 0.8, "learning_rate": 3.5278745644599306e-06, "loss": 0.3453, "step": 4581 }, { "epoch": 0.8, "learning_rate": 3.5714285714285714e-06, "loss": 0.3556, "step": 4582 }, { "epoch": 0.8, "learning_rate": 3.614982578397213e-06, "loss": 0.3547, "step": 4583 }, { "epoch": 0.8, "learning_rate": 3.6585365853658537e-06, "loss": 0.3612, "step": 4584 }, { "epoch": 0.8, "learning_rate": 3.7020905923344953e-06, "loss": 0.3301, "step": 4585 }, { "epoch": 0.8, "learning_rate": 3.745644599303136e-06, "loss": 0.348, "step": 4586 }, { "epoch": 0.8, "learning_rate": 3.789198606271777e-06, "loss": 0.3466, "step": 4587 }, { "epoch": 0.8, "learning_rate": 3.832752613240418e-06, "loss": 0.3433, "step": 4588 }, { "epoch": 0.8, "learning_rate": 3.876306620209059e-06, "loss": 0.3422, "step": 4589 }, { "epoch": 0.8, "learning_rate": 3.9198606271777e-06, "loss": 0.3515, "step": 4590 }, { "epoch": 0.8, "learning_rate": 3.9634146341463414e-06, "loss": 0.3667, "step": 4591 }, { "epoch": 0.8, "learning_rate": 4.006968641114982e-06, "loss": 0.3561, "step": 4592 }, { "epoch": 0.8, "learning_rate": 4.050522648083624e-06, "loss": 0.3474, "step": 4593 }, { "epoch": 0.8, "learning_rate": 4.0940766550522645e-06, "loss": 0.3479, "step": 4594 }, { "epoch": 0.8, "learning_rate": 4.137630662020906e-06, "loss": 0.3433, "step": 4595 }, { "epoch": 0.8, "learning_rate": 4.181184668989548e-06, "loss": 0.355, "step": 4596 }, { "epoch": 0.8, "learning_rate": 4.224738675958188e-06, "loss": 0.3462, "step": 4597 }, { "epoch": 0.8, "learning_rate": 4.26829268292683e-06, "loss": 0.3635, "step": 4598 }, { "epoch": 0.8, "learning_rate": 4.311846689895471e-06, "loss": 0.3443, "step": 4599 }, { "epoch": 0.8, "learning_rate": 4.355400696864112e-06, "loss": 0.3331, "step": 4600 }, { "epoch": 0.8, "learning_rate": 4.398954703832753e-06, "loss": 0.3482, "step": 4601 }, { "epoch": 0.8, "learning_rate": 4.442508710801394e-06, "loss": 0.345, "step": 4602 }, { "epoch": 0.8, "learning_rate": 4.4860627177700346e-06, "loss": 0.3583, "step": 4603 }, { "epoch": 0.8, "learning_rate": 4.529616724738676e-06, "loss": 0.3476, "step": 4604 }, { "epoch": 0.8, "learning_rate": 4.573170731707317e-06, "loss": 0.3549, "step": 4605 }, { "epoch": 0.8, "learning_rate": 4.6167247386759585e-06, "loss": 0.3569, "step": 4606 }, { "epoch": 0.8, "learning_rate": 4.660278745644599e-06, "loss": 0.3457, "step": 4607 }, { "epoch": 0.8, "learning_rate": 4.703832752613241e-06, "loss": 0.3453, "step": 4608 }, { "epoch": 0.8, "learning_rate": 4.7473867595818816e-06, "loss": 0.3658, "step": 4609 }, { "epoch": 0.8, "learning_rate": 4.790940766550523e-06, "loss": 0.3491, "step": 4610 }, { "epoch": 0.8, "learning_rate": 4.834494773519164e-06, "loss": 0.3483, "step": 4611 }, { "epoch": 0.8, "learning_rate": 4.8780487804878055e-06, "loss": 0.343, "step": 4612 }, { "epoch": 0.8, "learning_rate": 4.921602787456446e-06, "loss": 0.3584, "step": 4613 }, { "epoch": 0.8, "learning_rate": 4.965156794425088e-06, "loss": 0.3472, "step": 4614 }, { "epoch": 0.8, "learning_rate": 5.0087108013937286e-06, "loss": 0.3318, "step": 4615 }, { "epoch": 0.8, "learning_rate": 5.052264808362369e-06, "loss": 0.3488, "step": 4616 }, { "epoch": 0.8, "learning_rate": 5.09581881533101e-06, "loss": 0.3466, "step": 4617 }, { "epoch": 0.8, "learning_rate": 5.139372822299652e-06, "loss": 0.3371, "step": 4618 }, { "epoch": 0.81, "learning_rate": 5.182926829268292e-06, "loss": 0.3531, "step": 4619 }, { "epoch": 0.81, "learning_rate": 5.226480836236934e-06, "loss": 0.3496, "step": 4620 }, { "epoch": 0.81, "learning_rate": 5.270034843205575e-06, "loss": 0.3465, "step": 4621 }, { "epoch": 0.81, "learning_rate": 5.313588850174216e-06, "loss": 0.3377, "step": 4622 }, { "epoch": 0.81, "learning_rate": 5.357142857142857e-06, "loss": 0.3423, "step": 4623 }, { "epoch": 0.81, "learning_rate": 5.400696864111499e-06, "loss": 0.3496, "step": 4624 }, { "epoch": 0.81, "learning_rate": 5.444250871080139e-06, "loss": 0.3496, "step": 4625 }, { "epoch": 0.81, "learning_rate": 5.487804878048781e-06, "loss": 0.3475, "step": 4626 }, { "epoch": 0.81, "learning_rate": 5.531358885017422e-06, "loss": 0.346, "step": 4627 }, { "epoch": 0.81, "learning_rate": 5.574912891986063e-06, "loss": 0.3466, "step": 4628 }, { "epoch": 0.81, "learning_rate": 5.618466898954704e-06, "loss": 0.3404, "step": 4629 }, { "epoch": 0.81, "learning_rate": 5.662020905923345e-06, "loss": 0.3571, "step": 4630 }, { "epoch": 0.81, "learning_rate": 5.705574912891986e-06, "loss": 0.3748, "step": 4631 }, { "epoch": 0.81, "learning_rate": 5.749128919860627e-06, "loss": 0.3321, "step": 4632 }, { "epoch": 0.81, "learning_rate": 5.792682926829269e-06, "loss": 0.3441, "step": 4633 }, { "epoch": 0.81, "learning_rate": 5.8362369337979094e-06, "loss": 0.3688, "step": 4634 }, { "epoch": 0.81, "learning_rate": 5.879790940766551e-06, "loss": 0.3495, "step": 4635 }, { "epoch": 0.81, "learning_rate": 5.923344947735192e-06, "loss": 0.3387, "step": 4636 }, { "epoch": 0.81, "learning_rate": 5.966898954703833e-06, "loss": 0.3591, "step": 4637 }, { "epoch": 0.81, "learning_rate": 6.010452961672474e-06, "loss": 0.3526, "step": 4638 }, { "epoch": 0.81, "learning_rate": 6.054006968641116e-06, "loss": 0.3522, "step": 4639 }, { "epoch": 0.81, "learning_rate": 6.0975609756097564e-06, "loss": 0.3538, "step": 4640 }, { "epoch": 0.81, "learning_rate": 6.141114982578398e-06, "loss": 0.3404, "step": 4641 }, { "epoch": 0.81, "learning_rate": 6.184668989547039e-06, "loss": 0.343, "step": 4642 }, { "epoch": 0.81, "learning_rate": 6.2282229965156795e-06, "loss": 0.349, "step": 4643 }, { "epoch": 0.81, "learning_rate": 6.271777003484321e-06, "loss": 0.3578, "step": 4644 }, { "epoch": 0.81, "learning_rate": 6.315331010452962e-06, "loss": 0.3534, "step": 4645 }, { "epoch": 0.81, "learning_rate": 6.3588850174216034e-06, "loss": 0.3563, "step": 4646 }, { "epoch": 0.81, "learning_rate": 6.402439024390244e-06, "loss": 0.3341, "step": 4647 }, { "epoch": 0.81, "learning_rate": 6.445993031358885e-06, "loss": 0.3699, "step": 4648 }, { "epoch": 0.81, "learning_rate": 6.489547038327526e-06, "loss": 0.3355, "step": 4649 }, { "epoch": 0.81, "learning_rate": 6.533101045296168e-06, "loss": 0.3601, "step": 4650 }, { "epoch": 0.81, "learning_rate": 6.576655052264809e-06, "loss": 0.3439, "step": 4651 }, { "epoch": 0.81, "learning_rate": 6.6202090592334496e-06, "loss": 0.3493, "step": 4652 }, { "epoch": 0.81, "learning_rate": 6.66376306620209e-06, "loss": 0.3512, "step": 4653 }, { "epoch": 0.81, "learning_rate": 6.707317073170733e-06, "loss": 0.3573, "step": 4654 }, { "epoch": 0.81, "learning_rate": 6.7508710801393735e-06, "loss": 0.3461, "step": 4655 }, { "epoch": 0.81, "learning_rate": 6.794425087108014e-06, "loss": 0.3481, "step": 4656 }, { "epoch": 0.81, "learning_rate": 6.837979094076655e-06, "loss": 0.3482, "step": 4657 }, { "epoch": 0.81, "learning_rate": 6.8815331010452966e-06, "loss": 0.3502, "step": 4658 }, { "epoch": 0.81, "learning_rate": 6.925087108013937e-06, "loss": 0.361, "step": 4659 }, { "epoch": 0.81, "learning_rate": 6.968641114982578e-06, "loss": 0.3435, "step": 4660 }, { "epoch": 0.81, "learning_rate": 7.0121951219512205e-06, "loss": 0.3182, "step": 4661 }, { "epoch": 0.81, "learning_rate": 7.055749128919861e-06, "loss": 0.344, "step": 4662 }, { "epoch": 0.81, "learning_rate": 7.099303135888502e-06, "loss": 0.3472, "step": 4663 }, { "epoch": 0.81, "learning_rate": 7.142857142857143e-06, "loss": 0.3446, "step": 4664 }, { "epoch": 0.81, "learning_rate": 7.186411149825785e-06, "loss": 0.3443, "step": 4665 }, { "epoch": 0.81, "learning_rate": 7.229965156794426e-06, "loss": 0.3559, "step": 4666 }, { "epoch": 0.81, "learning_rate": 7.273519163763067e-06, "loss": 0.3453, "step": 4667 }, { "epoch": 0.81, "learning_rate": 7.317073170731707e-06, "loss": 0.3401, "step": 4668 }, { "epoch": 0.81, "learning_rate": 7.360627177700349e-06, "loss": 0.3315, "step": 4669 }, { "epoch": 0.81, "learning_rate": 7.4041811846689906e-06, "loss": 0.354, "step": 4670 }, { "epoch": 0.81, "learning_rate": 7.447735191637631e-06, "loss": 0.3498, "step": 4671 }, { "epoch": 0.81, "learning_rate": 7.491289198606272e-06, "loss": 0.3615, "step": 4672 }, { "epoch": 0.81, "learning_rate": 7.534843205574914e-06, "loss": 0.3555, "step": 4673 }, { "epoch": 0.81, "learning_rate": 7.578397212543554e-06, "loss": 0.3492, "step": 4674 }, { "epoch": 0.81, "learning_rate": 7.621951219512195e-06, "loss": 0.3596, "step": 4675 }, { "epoch": 0.82, "learning_rate": 7.665505226480837e-06, "loss": 0.3541, "step": 4676 }, { "epoch": 0.82, "learning_rate": 7.709059233449478e-06, "loss": 0.3464, "step": 4677 }, { "epoch": 0.82, "learning_rate": 7.752613240418118e-06, "loss": 0.3543, "step": 4678 }, { "epoch": 0.82, "learning_rate": 7.79616724738676e-06, "loss": 0.3413, "step": 4679 }, { "epoch": 0.82, "learning_rate": 7.8397212543554e-06, "loss": 0.3469, "step": 4680 }, { "epoch": 0.82, "learning_rate": 7.883275261324043e-06, "loss": 0.3381, "step": 4681 }, { "epoch": 0.82, "learning_rate": 7.926829268292683e-06, "loss": 0.3601, "step": 4682 }, { "epoch": 0.82, "learning_rate": 7.970383275261324e-06, "loss": 0.3565, "step": 4683 }, { "epoch": 0.82, "learning_rate": 8.013937282229964e-06, "loss": 0.3442, "step": 4684 }, { "epoch": 0.82, "learning_rate": 8.057491289198608e-06, "loss": 0.3563, "step": 4685 }, { "epoch": 0.82, "learning_rate": 8.101045296167248e-06, "loss": 0.371, "step": 4686 }, { "epoch": 0.82, "learning_rate": 8.144599303135889e-06, "loss": 0.3643, "step": 4687 }, { "epoch": 0.82, "learning_rate": 8.188153310104529e-06, "loss": 0.3376, "step": 4688 }, { "epoch": 0.82, "learning_rate": 8.231707317073172e-06, "loss": 0.3381, "step": 4689 }, { "epoch": 0.82, "learning_rate": 8.275261324041812e-06, "loss": 0.3472, "step": 4690 }, { "epoch": 0.82, "learning_rate": 8.318815331010454e-06, "loss": 0.3443, "step": 4691 }, { "epoch": 0.82, "learning_rate": 8.362369337979095e-06, "loss": 0.3629, "step": 4692 }, { "epoch": 0.82, "learning_rate": 8.405923344947735e-06, "loss": 0.3352, "step": 4693 }, { "epoch": 0.82, "learning_rate": 8.449477351916377e-06, "loss": 0.346, "step": 4694 }, { "epoch": 0.82, "learning_rate": 8.493031358885017e-06, "loss": 0.3649, "step": 4695 }, { "epoch": 0.82, "learning_rate": 8.53658536585366e-06, "loss": 0.3614, "step": 4696 }, { "epoch": 0.82, "learning_rate": 8.5801393728223e-06, "loss": 0.3448, "step": 4697 }, { "epoch": 0.82, "learning_rate": 8.623693379790942e-06, "loss": 0.3539, "step": 4698 }, { "epoch": 0.82, "learning_rate": 8.667247386759581e-06, "loss": 0.3723, "step": 4699 }, { "epoch": 0.82, "learning_rate": 8.710801393728225e-06, "loss": 0.358, "step": 4700 }, { "epoch": 0.82, "learning_rate": 8.754355400696865e-06, "loss": 0.3248, "step": 4701 }, { "epoch": 0.82, "learning_rate": 8.797909407665506e-06, "loss": 0.3437, "step": 4702 }, { "epoch": 0.82, "learning_rate": 8.841463414634146e-06, "loss": 0.3453, "step": 4703 }, { "epoch": 0.82, "learning_rate": 8.885017421602788e-06, "loss": 0.3339, "step": 4704 }, { "epoch": 0.82, "learning_rate": 8.92857142857143e-06, "loss": 0.3466, "step": 4705 }, { "epoch": 0.82, "learning_rate": 8.972125435540069e-06, "loss": 0.3448, "step": 4706 }, { "epoch": 0.82, "learning_rate": 9.01567944250871e-06, "loss": 0.3608, "step": 4707 }, { "epoch": 0.82, "learning_rate": 9.059233449477352e-06, "loss": 0.3518, "step": 4708 }, { "epoch": 0.82, "learning_rate": 9.102787456445994e-06, "loss": 0.3535, "step": 4709 }, { "epoch": 0.82, "learning_rate": 9.146341463414634e-06, "loss": 0.3366, "step": 4710 }, { "epoch": 0.82, "learning_rate": 9.189895470383275e-06, "loss": 0.3296, "step": 4711 }, { "epoch": 0.82, "learning_rate": 9.233449477351917e-06, "loss": 0.3601, "step": 4712 }, { "epoch": 0.82, "learning_rate": 9.277003484320559e-06, "loss": 0.3406, "step": 4713 }, { "epoch": 0.82, "learning_rate": 9.320557491289198e-06, "loss": 0.3556, "step": 4714 }, { "epoch": 0.82, "learning_rate": 9.36411149825784e-06, "loss": 0.3549, "step": 4715 }, { "epoch": 0.82, "learning_rate": 9.407665505226482e-06, "loss": 0.3462, "step": 4716 }, { "epoch": 0.82, "learning_rate": 9.451219512195123e-06, "loss": 0.3531, "step": 4717 }, { "epoch": 0.82, "learning_rate": 9.494773519163763e-06, "loss": 0.3491, "step": 4718 }, { "epoch": 0.82, "learning_rate": 9.538327526132405e-06, "loss": 0.3528, "step": 4719 }, { "epoch": 0.82, "learning_rate": 9.581881533101046e-06, "loss": 0.3453, "step": 4720 }, { "epoch": 0.82, "learning_rate": 9.625435540069686e-06, "loss": 0.3451, "step": 4721 }, { "epoch": 0.82, "learning_rate": 9.668989547038328e-06, "loss": 0.3536, "step": 4722 }, { "epoch": 0.82, "learning_rate": 9.712543554006968e-06, "loss": 0.3356, "step": 4723 }, { "epoch": 0.82, "learning_rate": 9.756097560975611e-06, "loss": 0.3455, "step": 4724 }, { "epoch": 0.82, "learning_rate": 9.799651567944251e-06, "loss": 0.3205, "step": 4725 }, { "epoch": 0.82, "learning_rate": 9.843205574912892e-06, "loss": 0.334, "step": 4726 }, { "epoch": 0.82, "learning_rate": 9.886759581881534e-06, "loss": 0.3469, "step": 4727 }, { "epoch": 0.82, "learning_rate": 9.930313588850176e-06, "loss": 0.3464, "step": 4728 }, { "epoch": 0.82, "learning_rate": 9.973867595818816e-06, "loss": 0.3384, "step": 4729 }, { "epoch": 0.82, "learning_rate": 1.0017421602787457e-05, "loss": 0.3422, "step": 4730 }, { "epoch": 0.82, "learning_rate": 1.0060975609756099e-05, "loss": 0.3352, "step": 4731 }, { "epoch": 0.82, "learning_rate": 1.0104529616724739e-05, "loss": 0.3382, "step": 4732 }, { "epoch": 0.82, "learning_rate": 1.014808362369338e-05, "loss": 0.3368, "step": 4733 }, { "epoch": 0.83, "learning_rate": 1.019163763066202e-05, "loss": 0.3538, "step": 4734 }, { "epoch": 0.83, "learning_rate": 1.0235191637630663e-05, "loss": 0.3525, "step": 4735 }, { "epoch": 0.83, "learning_rate": 1.0278745644599303e-05, "loss": 0.3502, "step": 4736 }, { "epoch": 0.83, "learning_rate": 1.0322299651567945e-05, "loss": 0.3523, "step": 4737 }, { "epoch": 0.83, "learning_rate": 1.0365853658536585e-05, "loss": 0.3616, "step": 4738 }, { "epoch": 0.83, "learning_rate": 1.0409407665505228e-05, "loss": 0.3633, "step": 4739 }, { "epoch": 0.83, "learning_rate": 1.0452961672473868e-05, "loss": 0.3388, "step": 4740 }, { "epoch": 0.83, "learning_rate": 1.049651567944251e-05, "loss": 0.3458, "step": 4741 }, { "epoch": 0.83, "learning_rate": 1.054006968641115e-05, "loss": 0.3659, "step": 4742 }, { "epoch": 0.83, "learning_rate": 1.0583623693379793e-05, "loss": 0.3501, "step": 4743 }, { "epoch": 0.83, "learning_rate": 1.0627177700348433e-05, "loss": 0.3528, "step": 4744 }, { "epoch": 0.83, "learning_rate": 1.0670731707317074e-05, "loss": 0.3497, "step": 4745 }, { "epoch": 0.83, "learning_rate": 1.0714285714285714e-05, "loss": 0.3547, "step": 4746 }, { "epoch": 0.83, "learning_rate": 1.0757839721254356e-05, "loss": 0.3512, "step": 4747 }, { "epoch": 0.83, "learning_rate": 1.0801393728222997e-05, "loss": 0.3502, "step": 4748 }, { "epoch": 0.83, "learning_rate": 1.0844947735191637e-05, "loss": 0.3646, "step": 4749 }, { "epoch": 0.83, "learning_rate": 1.0888501742160279e-05, "loss": 0.3413, "step": 4750 }, { "epoch": 0.83, "learning_rate": 1.093205574912892e-05, "loss": 0.3506, "step": 4751 }, { "epoch": 0.83, "learning_rate": 1.0975609756097562e-05, "loss": 0.3329, "step": 4752 }, { "epoch": 0.83, "learning_rate": 1.1019163763066202e-05, "loss": 0.3326, "step": 4753 }, { "epoch": 0.83, "learning_rate": 1.1062717770034843e-05, "loss": 0.341, "step": 4754 }, { "epoch": 0.83, "learning_rate": 1.1106271777003485e-05, "loss": 0.3482, "step": 4755 }, { "epoch": 0.83, "learning_rate": 1.1149825783972127e-05, "loss": 0.3403, "step": 4756 }, { "epoch": 0.83, "learning_rate": 1.1193379790940766e-05, "loss": 0.3556, "step": 4757 }, { "epoch": 0.83, "learning_rate": 1.1236933797909408e-05, "loss": 0.3513, "step": 4758 }, { "epoch": 0.83, "learning_rate": 1.128048780487805e-05, "loss": 0.3269, "step": 4759 }, { "epoch": 0.83, "learning_rate": 1.132404181184669e-05, "loss": 0.3586, "step": 4760 }, { "epoch": 0.83, "learning_rate": 1.1367595818815331e-05, "loss": 0.3433, "step": 4761 }, { "epoch": 0.83, "learning_rate": 1.1411149825783973e-05, "loss": 0.3681, "step": 4762 }, { "epoch": 0.83, "learning_rate": 1.1454703832752614e-05, "loss": 0.3404, "step": 4763 }, { "epoch": 0.83, "learning_rate": 1.1498257839721254e-05, "loss": 0.3569, "step": 4764 }, { "epoch": 0.83, "learning_rate": 1.1541811846689896e-05, "loss": 0.3482, "step": 4765 }, { "epoch": 0.83, "learning_rate": 1.1585365853658537e-05, "loss": 0.3402, "step": 4766 }, { "epoch": 0.83, "learning_rate": 1.1628919860627179e-05, "loss": 0.3452, "step": 4767 }, { "epoch": 0.83, "learning_rate": 1.1672473867595819e-05, "loss": 0.3346, "step": 4768 }, { "epoch": 0.83, "learning_rate": 1.171602787456446e-05, "loss": 0.3386, "step": 4769 }, { "epoch": 0.83, "learning_rate": 1.1759581881533102e-05, "loss": 0.3354, "step": 4770 }, { "epoch": 0.83, "learning_rate": 1.1803135888501744e-05, "loss": 0.3388, "step": 4771 }, { "epoch": 0.83, "learning_rate": 1.1846689895470384e-05, "loss": 0.3405, "step": 4772 }, { "epoch": 0.83, "learning_rate": 1.1890243902439025e-05, "loss": 0.3455, "step": 4773 }, { "epoch": 0.83, "learning_rate": 1.1933797909407667e-05, "loss": 0.3466, "step": 4774 }, { "epoch": 0.83, "learning_rate": 1.1977351916376307e-05, "loss": 0.3341, "step": 4775 }, { "epoch": 0.83, "learning_rate": 1.2020905923344948e-05, "loss": 0.3284, "step": 4776 }, { "epoch": 0.83, "learning_rate": 1.2064459930313588e-05, "loss": 0.3382, "step": 4777 }, { "epoch": 0.83, "learning_rate": 1.2108013937282231e-05, "loss": 0.3561, "step": 4778 }, { "epoch": 0.83, "learning_rate": 1.2151567944250871e-05, "loss": 0.3412, "step": 4779 }, { "epoch": 0.83, "learning_rate": 1.2195121951219513e-05, "loss": 0.3527, "step": 4780 }, { "epoch": 0.83, "learning_rate": 1.2238675958188153e-05, "loss": 0.3536, "step": 4781 }, { "epoch": 0.83, "learning_rate": 1.2282229965156796e-05, "loss": 0.3416, "step": 4782 }, { "epoch": 0.83, "learning_rate": 1.2325783972125436e-05, "loss": 0.3428, "step": 4783 }, { "epoch": 0.83, "learning_rate": 1.2369337979094078e-05, "loss": 0.3547, "step": 4784 }, { "epoch": 0.83, "learning_rate": 1.2412891986062717e-05, "loss": 0.342, "step": 4785 }, { "epoch": 0.83, "learning_rate": 1.2456445993031359e-05, "loss": 0.3606, "step": 4786 }, { "epoch": 0.83, "learning_rate": 1.25e-05, "loss": 0.3474, "step": 4787 }, { "epoch": 0.83, "learning_rate": 1.2543554006968642e-05, "loss": 0.3331, "step": 4788 }, { "epoch": 0.83, "learning_rate": 1.2587108013937282e-05, "loss": 0.3562, "step": 4789 }, { "epoch": 0.83, "learning_rate": 1.2630662020905924e-05, "loss": 0.3446, "step": 4790 }, { "epoch": 0.84, "learning_rate": 1.2674216027874567e-05, "loss": 0.3481, "step": 4791 }, { "epoch": 0.84, "learning_rate": 1.2717770034843207e-05, "loss": 0.3541, "step": 4792 }, { "epoch": 0.84, "learning_rate": 1.2761324041811848e-05, "loss": 0.3446, "step": 4793 }, { "epoch": 0.84, "learning_rate": 1.2804878048780488e-05, "loss": 0.3522, "step": 4794 }, { "epoch": 0.84, "learning_rate": 1.284843205574913e-05, "loss": 0.3589, "step": 4795 }, { "epoch": 0.84, "learning_rate": 1.289198606271777e-05, "loss": 0.3478, "step": 4796 }, { "epoch": 0.84, "learning_rate": 1.2935540069686411e-05, "loss": 0.3527, "step": 4797 }, { "epoch": 0.84, "learning_rate": 1.2979094076655051e-05, "loss": 0.3375, "step": 4798 }, { "epoch": 0.84, "learning_rate": 1.3022648083623695e-05, "loss": 0.3297, "step": 4799 }, { "epoch": 0.84, "learning_rate": 1.3066202090592336e-05, "loss": 0.3492, "step": 4800 }, { "epoch": 0.84, "learning_rate": 1.3109756097560976e-05, "loss": 0.3503, "step": 4801 }, { "epoch": 0.84, "learning_rate": 1.3153310104529618e-05, "loss": 0.3473, "step": 4802 }, { "epoch": 0.84, "learning_rate": 1.3196864111498258e-05, "loss": 0.3571, "step": 4803 }, { "epoch": 0.84, "learning_rate": 1.3240418118466899e-05, "loss": 0.3408, "step": 4804 }, { "epoch": 0.84, "learning_rate": 1.3283972125435539e-05, "loss": 0.35, "step": 4805 }, { "epoch": 0.84, "learning_rate": 1.332752613240418e-05, "loss": 0.355, "step": 4806 }, { "epoch": 0.84, "learning_rate": 1.3371080139372824e-05, "loss": 0.366, "step": 4807 }, { "epoch": 0.84, "learning_rate": 1.3414634146341466e-05, "loss": 0.3531, "step": 4808 }, { "epoch": 0.84, "learning_rate": 1.3458188153310105e-05, "loss": 0.3564, "step": 4809 }, { "epoch": 0.84, "learning_rate": 1.3501742160278747e-05, "loss": 0.3389, "step": 4810 }, { "epoch": 0.84, "learning_rate": 1.3545296167247387e-05, "loss": 0.3526, "step": 4811 }, { "epoch": 0.84, "learning_rate": 1.3588850174216028e-05, "loss": 0.3734, "step": 4812 }, { "epoch": 0.84, "learning_rate": 1.3632404181184668e-05, "loss": 0.346, "step": 4813 }, { "epoch": 0.84, "learning_rate": 1.367595818815331e-05, "loss": 0.3507, "step": 4814 }, { "epoch": 0.84, "learning_rate": 1.3719512195121953e-05, "loss": 0.3413, "step": 4815 }, { "epoch": 0.84, "learning_rate": 1.3763066202090593e-05, "loss": 0.3601, "step": 4816 }, { "epoch": 0.84, "learning_rate": 1.3806620209059235e-05, "loss": 0.3548, "step": 4817 }, { "epoch": 0.84, "learning_rate": 1.3850174216027875e-05, "loss": 0.3659, "step": 4818 }, { "epoch": 0.84, "learning_rate": 1.3893728222996516e-05, "loss": 0.3346, "step": 4819 }, { "epoch": 0.84, "learning_rate": 1.3937282229965156e-05, "loss": 0.3562, "step": 4820 }, { "epoch": 0.84, "learning_rate": 1.3980836236933798e-05, "loss": 0.3471, "step": 4821 }, { "epoch": 0.84, "learning_rate": 1.4024390243902441e-05, "loss": 0.3578, "step": 4822 }, { "epoch": 0.84, "learning_rate": 1.4067944250871083e-05, "loss": 0.364, "step": 4823 }, { "epoch": 0.84, "learning_rate": 1.4111498257839722e-05, "loss": 0.3412, "step": 4824 }, { "epoch": 0.84, "learning_rate": 1.4155052264808364e-05, "loss": 0.3426, "step": 4825 }, { "epoch": 0.84, "learning_rate": 1.4198606271777004e-05, "loss": 0.3524, "step": 4826 }, { "epoch": 0.84, "learning_rate": 1.4242160278745646e-05, "loss": 0.3344, "step": 4827 }, { "epoch": 0.84, "learning_rate": 1.4285714285714285e-05, "loss": 0.3525, "step": 4828 }, { "epoch": 0.84, "learning_rate": 1.4329268292682927e-05, "loss": 0.3509, "step": 4829 }, { "epoch": 0.84, "learning_rate": 1.437282229965157e-05, "loss": 0.349, "step": 4830 }, { "epoch": 0.84, "learning_rate": 1.441637630662021e-05, "loss": 0.3387, "step": 4831 }, { "epoch": 0.84, "learning_rate": 1.4459930313588852e-05, "loss": 0.3558, "step": 4832 }, { "epoch": 0.84, "learning_rate": 1.4503484320557492e-05, "loss": 0.3489, "step": 4833 }, { "epoch": 0.84, "learning_rate": 1.4547038327526133e-05, "loss": 0.3451, "step": 4834 }, { "epoch": 0.84, "learning_rate": 1.4590592334494773e-05, "loss": 0.3397, "step": 4835 }, { "epoch": 0.84, "learning_rate": 1.4634146341463415e-05, "loss": 0.3549, "step": 4836 }, { "epoch": 0.84, "learning_rate": 1.4677700348432055e-05, "loss": 0.338, "step": 4837 }, { "epoch": 0.84, "learning_rate": 1.4721254355400698e-05, "loss": 0.3468, "step": 4838 }, { "epoch": 0.84, "learning_rate": 1.476480836236934e-05, "loss": 0.3609, "step": 4839 }, { "epoch": 0.84, "learning_rate": 1.4808362369337981e-05, "loss": 0.3503, "step": 4840 }, { "epoch": 0.84, "learning_rate": 1.4851916376306621e-05, "loss": 0.3495, "step": 4841 }, { "epoch": 0.84, "learning_rate": 1.4895470383275263e-05, "loss": 0.3328, "step": 4842 }, { "epoch": 0.84, "learning_rate": 1.4939024390243902e-05, "loss": 0.3378, "step": 4843 }, { "epoch": 0.84, "learning_rate": 1.4982578397212544e-05, "loss": 0.3417, "step": 4844 }, { "epoch": 0.84, "learning_rate": 1.5026132404181184e-05, "loss": 0.3603, "step": 4845 }, { "epoch": 0.84, "learning_rate": 1.5069686411149827e-05, "loss": 0.345, "step": 4846 }, { "epoch": 0.84, "learning_rate": 1.5113240418118469e-05, "loss": 0.3466, "step": 4847 }, { "epoch": 0.85, "learning_rate": 1.5156794425087109e-05, "loss": 0.3713, "step": 4848 }, { "epoch": 0.85, "learning_rate": 1.520034843205575e-05, "loss": 0.3543, "step": 4849 }, { "epoch": 0.85, "learning_rate": 1.524390243902439e-05, "loss": 0.3385, "step": 4850 }, { "epoch": 0.85, "learning_rate": 1.528745644599303e-05, "loss": 0.3421, "step": 4851 }, { "epoch": 0.85, "learning_rate": 1.5331010452961673e-05, "loss": 0.3381, "step": 4852 }, { "epoch": 0.85, "learning_rate": 1.5374564459930317e-05, "loss": 0.3543, "step": 4853 }, { "epoch": 0.85, "learning_rate": 1.5418118466898957e-05, "loss": 0.3521, "step": 4854 }, { "epoch": 0.85, "learning_rate": 1.5461672473867596e-05, "loss": 0.3509, "step": 4855 }, { "epoch": 0.85, "learning_rate": 1.5505226480836236e-05, "loss": 0.3503, "step": 4856 }, { "epoch": 0.85, "learning_rate": 1.554878048780488e-05, "loss": 0.3464, "step": 4857 }, { "epoch": 0.85, "learning_rate": 1.559233449477352e-05, "loss": 0.3467, "step": 4858 }, { "epoch": 0.85, "learning_rate": 1.563588850174216e-05, "loss": 0.3515, "step": 4859 }, { "epoch": 0.85, "learning_rate": 1.56794425087108e-05, "loss": 0.3457, "step": 4860 }, { "epoch": 0.85, "learning_rate": 1.5722996515679446e-05, "loss": 0.3491, "step": 4861 }, { "epoch": 0.85, "learning_rate": 1.5766550522648086e-05, "loss": 0.355, "step": 4862 }, { "epoch": 0.85, "learning_rate": 1.5810104529616726e-05, "loss": 0.3328, "step": 4863 }, { "epoch": 0.85, "learning_rate": 1.5853658536585366e-05, "loss": 0.3345, "step": 4864 }, { "epoch": 0.85, "learning_rate": 1.589721254355401e-05, "loss": 0.3694, "step": 4865 }, { "epoch": 0.85, "learning_rate": 1.594076655052265e-05, "loss": 0.3477, "step": 4866 }, { "epoch": 0.85, "learning_rate": 1.598432055749129e-05, "loss": 0.3456, "step": 4867 }, { "epoch": 0.85, "learning_rate": 1.602787456445993e-05, "loss": 0.3419, "step": 4868 }, { "epoch": 0.85, "learning_rate": 1.6071428571428572e-05, "loss": 0.3459, "step": 4869 }, { "epoch": 0.85, "learning_rate": 1.6114982578397215e-05, "loss": 0.3411, "step": 4870 }, { "epoch": 0.85, "learning_rate": 1.6158536585365855e-05, "loss": 0.3338, "step": 4871 }, { "epoch": 0.85, "learning_rate": 1.6202090592334495e-05, "loss": 0.3381, "step": 4872 }, { "epoch": 0.85, "learning_rate": 1.6245644599303135e-05, "loss": 0.3408, "step": 4873 }, { "epoch": 0.85, "learning_rate": 1.6289198606271778e-05, "loss": 0.3487, "step": 4874 }, { "epoch": 0.85, "learning_rate": 1.6332752613240418e-05, "loss": 0.3266, "step": 4875 }, { "epoch": 0.85, "learning_rate": 1.6376306620209058e-05, "loss": 0.3693, "step": 4876 }, { "epoch": 0.85, "learning_rate": 1.64198606271777e-05, "loss": 0.3567, "step": 4877 }, { "epoch": 0.85, "learning_rate": 1.6463414634146345e-05, "loss": 0.3492, "step": 4878 }, { "epoch": 0.85, "learning_rate": 1.6506968641114984e-05, "loss": 0.3498, "step": 4879 }, { "epoch": 0.85, "learning_rate": 1.6550522648083624e-05, "loss": 0.3587, "step": 4880 }, { "epoch": 0.85, "learning_rate": 1.6594076655052264e-05, "loss": 0.3534, "step": 4881 }, { "epoch": 0.85, "learning_rate": 1.6637630662020908e-05, "loss": 0.3489, "step": 4882 }, { "epoch": 0.85, "learning_rate": 1.6681184668989547e-05, "loss": 0.344, "step": 4883 }, { "epoch": 0.85, "learning_rate": 1.672473867595819e-05, "loss": 0.3292, "step": 4884 }, { "epoch": 0.85, "learning_rate": 1.676829268292683e-05, "loss": 0.345, "step": 4885 }, { "epoch": 0.85, "learning_rate": 1.681184668989547e-05, "loss": 0.3429, "step": 4886 }, { "epoch": 0.85, "learning_rate": 1.6855400696864114e-05, "loss": 0.3414, "step": 4887 }, { "epoch": 0.85, "learning_rate": 1.6898954703832754e-05, "loss": 0.3432, "step": 4888 }, { "epoch": 0.85, "learning_rate": 1.6942508710801394e-05, "loss": 0.359, "step": 4889 }, { "epoch": 0.85, "learning_rate": 1.6986062717770033e-05, "loss": 0.347, "step": 4890 }, { "epoch": 0.85, "learning_rate": 1.7029616724738677e-05, "loss": 0.3403, "step": 4891 }, { "epoch": 0.85, "learning_rate": 1.707317073170732e-05, "loss": 0.3238, "step": 4892 }, { "epoch": 0.85, "learning_rate": 1.711672473867596e-05, "loss": 0.3401, "step": 4893 }, { "epoch": 0.85, "learning_rate": 1.71602787456446e-05, "loss": 0.3411, "step": 4894 }, { "epoch": 0.85, "learning_rate": 1.7203832752613243e-05, "loss": 0.3616, "step": 4895 }, { "epoch": 0.85, "learning_rate": 1.7247386759581883e-05, "loss": 0.3661, "step": 4896 }, { "epoch": 0.85, "learning_rate": 1.7290940766550523e-05, "loss": 0.3443, "step": 4897 }, { "epoch": 0.85, "learning_rate": 1.7334494773519163e-05, "loss": 0.3537, "step": 4898 }, { "epoch": 0.85, "learning_rate": 1.7378048780487806e-05, "loss": 0.3645, "step": 4899 }, { "epoch": 0.85, "learning_rate": 1.742160278745645e-05, "loss": 0.3537, "step": 4900 }, { "epoch": 0.85, "learning_rate": 1.746515679442509e-05, "loss": 0.3432, "step": 4901 }, { "epoch": 0.85, "learning_rate": 1.750871080139373e-05, "loss": 0.3387, "step": 4902 }, { "epoch": 0.85, "learning_rate": 1.755226480836237e-05, "loss": 0.3755, "step": 4903 }, { "epoch": 0.85, "learning_rate": 1.7595818815331012e-05, "loss": 0.3505, "step": 4904 }, { "epoch": 0.85, "learning_rate": 1.7639372822299652e-05, "loss": 0.3586, "step": 4905 }, { "epoch": 0.86, "learning_rate": 1.7682926829268292e-05, "loss": 0.3492, "step": 4906 }, { "epoch": 0.86, "learning_rate": 1.7726480836236932e-05, "loss": 0.3612, "step": 4907 }, { "epoch": 0.86, "learning_rate": 1.7770034843205575e-05, "loss": 0.3503, "step": 4908 }, { "epoch": 0.86, "learning_rate": 1.781358885017422e-05, "loss": 0.3602, "step": 4909 }, { "epoch": 0.86, "learning_rate": 1.785714285714286e-05, "loss": 0.3419, "step": 4910 }, { "epoch": 0.86, "learning_rate": 1.79006968641115e-05, "loss": 0.3448, "step": 4911 }, { "epoch": 0.86, "learning_rate": 1.7944250871080138e-05, "loss": 0.3579, "step": 4912 }, { "epoch": 0.86, "learning_rate": 1.798780487804878e-05, "loss": 0.3411, "step": 4913 }, { "epoch": 0.86, "learning_rate": 1.803135888501742e-05, "loss": 0.3552, "step": 4914 }, { "epoch": 0.86, "learning_rate": 1.8074912891986065e-05, "loss": 0.3475, "step": 4915 }, { "epoch": 0.86, "learning_rate": 1.8118466898954705e-05, "loss": 0.3347, "step": 4916 }, { "epoch": 0.86, "learning_rate": 1.8162020905923348e-05, "loss": 0.3434, "step": 4917 }, { "epoch": 0.86, "learning_rate": 1.8205574912891988e-05, "loss": 0.3512, "step": 4918 }, { "epoch": 0.86, "learning_rate": 1.8249128919860628e-05, "loss": 0.3472, "step": 4919 }, { "epoch": 0.86, "learning_rate": 1.8292682926829268e-05, "loss": 0.3596, "step": 4920 }, { "epoch": 0.86, "learning_rate": 1.833623693379791e-05, "loss": 0.3415, "step": 4921 }, { "epoch": 0.86, "learning_rate": 1.837979094076655e-05, "loss": 0.3559, "step": 4922 }, { "epoch": 0.86, "learning_rate": 1.8423344947735194e-05, "loss": 0.3542, "step": 4923 }, { "epoch": 0.86, "learning_rate": 1.8466898954703834e-05, "loss": 0.3343, "step": 4924 }, { "epoch": 0.86, "learning_rate": 1.8510452961672474e-05, "loss": 0.3691, "step": 4925 }, { "epoch": 0.86, "learning_rate": 1.8554006968641117e-05, "loss": 0.352, "step": 4926 }, { "epoch": 0.86, "learning_rate": 1.8597560975609757e-05, "loss": 0.3547, "step": 4927 }, { "epoch": 0.86, "learning_rate": 1.8641114982578397e-05, "loss": 0.3427, "step": 4928 }, { "epoch": 0.86, "learning_rate": 1.8684668989547037e-05, "loss": 0.3666, "step": 4929 }, { "epoch": 0.86, "learning_rate": 1.872822299651568e-05, "loss": 0.3539, "step": 4930 }, { "epoch": 0.86, "learning_rate": 1.8771777003484323e-05, "loss": 0.3477, "step": 4931 }, { "epoch": 0.86, "learning_rate": 1.8815331010452963e-05, "loss": 0.3569, "step": 4932 }, { "epoch": 0.86, "learning_rate": 1.8858885017421603e-05, "loss": 0.3529, "step": 4933 }, { "epoch": 0.86, "learning_rate": 1.8902439024390246e-05, "loss": 0.3432, "step": 4934 }, { "epoch": 0.86, "learning_rate": 1.8945993031358886e-05, "loss": 0.3472, "step": 4935 }, { "epoch": 0.86, "learning_rate": 1.8989547038327526e-05, "loss": 0.3407, "step": 4936 }, { "epoch": 0.86, "learning_rate": 1.9033101045296166e-05, "loss": 0.3409, "step": 4937 }, { "epoch": 0.86, "learning_rate": 1.907665505226481e-05, "loss": 0.3639, "step": 4938 }, { "epoch": 0.86, "learning_rate": 1.9120209059233453e-05, "loss": 0.35, "step": 4939 }, { "epoch": 0.86, "learning_rate": 1.9163763066202093e-05, "loss": 0.3442, "step": 4940 }, { "epoch": 0.86, "learning_rate": 1.9207317073170733e-05, "loss": 0.3474, "step": 4941 }, { "epoch": 0.86, "learning_rate": 1.9250871080139372e-05, "loss": 0.3514, "step": 4942 }, { "epoch": 0.86, "learning_rate": 1.9294425087108016e-05, "loss": 0.3342, "step": 4943 }, { "epoch": 0.86, "learning_rate": 1.9337979094076656e-05, "loss": 0.3568, "step": 4944 }, { "epoch": 0.86, "learning_rate": 1.9381533101045295e-05, "loss": 0.332, "step": 4945 }, { "epoch": 0.86, "learning_rate": 1.9425087108013935e-05, "loss": 0.3439, "step": 4946 }, { "epoch": 0.86, "learning_rate": 1.9468641114982582e-05, "loss": 0.3638, "step": 4947 }, { "epoch": 0.86, "learning_rate": 1.9512195121951222e-05, "loss": 0.3584, "step": 4948 }, { "epoch": 0.86, "learning_rate": 1.9555749128919862e-05, "loss": 0.3341, "step": 4949 }, { "epoch": 0.86, "learning_rate": 1.9599303135888502e-05, "loss": 0.3595, "step": 4950 }, { "epoch": 0.86, "learning_rate": 1.9642857142857145e-05, "loss": 0.3513, "step": 4951 }, { "epoch": 0.86, "learning_rate": 1.9686411149825785e-05, "loss": 0.3503, "step": 4952 }, { "epoch": 0.86, "learning_rate": 1.9729965156794425e-05, "loss": 0.36, "step": 4953 }, { "epoch": 0.86, "learning_rate": 1.9773519163763068e-05, "loss": 0.3482, "step": 4954 }, { "epoch": 0.86, "learning_rate": 1.9817073170731708e-05, "loss": 0.3284, "step": 4955 }, { "epoch": 0.86, "learning_rate": 1.986062717770035e-05, "loss": 0.3539, "step": 4956 }, { "epoch": 0.86, "learning_rate": 1.990418118466899e-05, "loss": 0.3453, "step": 4957 }, { "epoch": 0.86, "learning_rate": 1.994773519163763e-05, "loss": 0.3303, "step": 4958 }, { "epoch": 0.86, "learning_rate": 1.999128919860627e-05, "loss": 0.344, "step": 4959 }, { "epoch": 0.86, "learning_rate": 2.0034843205574914e-05, "loss": 0.3459, "step": 4960 }, { "epoch": 0.86, "learning_rate": 2.0078397212543554e-05, "loss": 0.3649, "step": 4961 }, { "epoch": 0.86, "learning_rate": 2.0121951219512197e-05, "loss": 0.3635, "step": 4962 }, { "epoch": 0.87, "learning_rate": 2.0165505226480837e-05, "loss": 0.343, "step": 4963 }, { "epoch": 0.87, "learning_rate": 2.0209059233449477e-05, "loss": 0.3485, "step": 4964 }, { "epoch": 0.87, "learning_rate": 2.025261324041812e-05, "loss": 0.3439, "step": 4965 }, { "epoch": 0.87, "learning_rate": 2.029616724738676e-05, "loss": 0.3372, "step": 4966 }, { "epoch": 0.87, "learning_rate": 2.03397212543554e-05, "loss": 0.342, "step": 4967 }, { "epoch": 0.87, "learning_rate": 2.038327526132404e-05, "loss": 0.3378, "step": 4968 }, { "epoch": 0.87, "learning_rate": 2.0426829268292683e-05, "loss": 0.3287, "step": 4969 }, { "epoch": 0.87, "learning_rate": 2.0470383275261327e-05, "loss": 0.3503, "step": 4970 }, { "epoch": 0.87, "learning_rate": 2.0513937282229967e-05, "loss": 0.3488, "step": 4971 }, { "epoch": 0.87, "learning_rate": 2.0557491289198607e-05, "loss": 0.3437, "step": 4972 }, { "epoch": 0.87, "learning_rate": 2.060104529616725e-05, "loss": 0.3425, "step": 4973 }, { "epoch": 0.87, "learning_rate": 2.064459930313589e-05, "loss": 0.3566, "step": 4974 }, { "epoch": 0.87, "learning_rate": 2.068815331010453e-05, "loss": 0.3381, "step": 4975 }, { "epoch": 0.87, "learning_rate": 2.073170731707317e-05, "loss": 0.3522, "step": 4976 }, { "epoch": 0.87, "learning_rate": 2.0775261324041813e-05, "loss": 0.3468, "step": 4977 }, { "epoch": 0.87, "learning_rate": 2.0818815331010456e-05, "loss": 0.3365, "step": 4978 }, { "epoch": 0.87, "learning_rate": 2.0862369337979096e-05, "loss": 0.3463, "step": 4979 }, { "epoch": 0.87, "learning_rate": 2.0905923344947736e-05, "loss": 0.3624, "step": 4980 }, { "epoch": 0.87, "learning_rate": 2.0949477351916376e-05, "loss": 0.3609, "step": 4981 }, { "epoch": 0.87, "learning_rate": 2.099303135888502e-05, "loss": 0.3521, "step": 4982 }, { "epoch": 0.87, "learning_rate": 2.103658536585366e-05, "loss": 0.3543, "step": 4983 }, { "epoch": 0.87, "learning_rate": 2.10801393728223e-05, "loss": 0.3527, "step": 4984 }, { "epoch": 0.87, "learning_rate": 2.1123693379790942e-05, "loss": 0.3547, "step": 4985 }, { "epoch": 0.87, "learning_rate": 2.1167247386759585e-05, "loss": 0.3433, "step": 4986 }, { "epoch": 0.87, "learning_rate": 2.1210801393728225e-05, "loss": 0.347, "step": 4987 }, { "epoch": 0.87, "learning_rate": 2.1254355400696865e-05, "loss": 0.3419, "step": 4988 }, { "epoch": 0.87, "learning_rate": 2.1297909407665505e-05, "loss": 0.3511, "step": 4989 }, { "epoch": 0.87, "learning_rate": 2.134146341463415e-05, "loss": 0.3416, "step": 4990 }, { "epoch": 0.87, "learning_rate": 2.1385017421602788e-05, "loss": 0.3565, "step": 4991 }, { "epoch": 0.87, "learning_rate": 2.1428571428571428e-05, "loss": 0.3648, "step": 4992 }, { "epoch": 0.87, "learning_rate": 2.147212543554007e-05, "loss": 0.3326, "step": 4993 }, { "epoch": 0.87, "learning_rate": 2.151567944250871e-05, "loss": 0.3501, "step": 4994 }, { "epoch": 0.87, "learning_rate": 2.1559233449477355e-05, "loss": 0.3285, "step": 4995 }, { "epoch": 0.87, "learning_rate": 2.1602787456445995e-05, "loss": 0.3466, "step": 4996 }, { "epoch": 0.87, "learning_rate": 2.1646341463414634e-05, "loss": 0.3355, "step": 4997 }, { "epoch": 0.87, "learning_rate": 2.1689895470383274e-05, "loss": 0.3461, "step": 4998 }, { "epoch": 0.87, "learning_rate": 2.1733449477351918e-05, "loss": 0.341, "step": 4999 }, { "epoch": 0.87, "learning_rate": 2.1777003484320557e-05, "loss": 0.3439, "step": 5000 }, { "epoch": 0.87, "learning_rate": 2.18205574912892e-05, "loss": 0.3434, "step": 5001 }, { "epoch": 0.87, "learning_rate": 2.186411149825784e-05, "loss": 0.3339, "step": 5002 }, { "epoch": 0.87, "learning_rate": 2.1907665505226484e-05, "loss": 0.3634, "step": 5003 }, { "epoch": 0.87, "learning_rate": 2.1951219512195124e-05, "loss": 0.3443, "step": 5004 }, { "epoch": 0.87, "learning_rate": 2.1994773519163764e-05, "loss": 0.3514, "step": 5005 }, { "epoch": 0.87, "learning_rate": 2.2038327526132404e-05, "loss": 0.3441, "step": 5006 }, { "epoch": 0.87, "learning_rate": 2.2081881533101047e-05, "loss": 0.3431, "step": 5007 }, { "epoch": 0.87, "learning_rate": 2.2125435540069687e-05, "loss": 0.3345, "step": 5008 }, { "epoch": 0.87, "learning_rate": 2.216898954703833e-05, "loss": 0.3465, "step": 5009 }, { "epoch": 0.87, "learning_rate": 2.221254355400697e-05, "loss": 0.3379, "step": 5010 }, { "epoch": 0.87, "learning_rate": 2.225609756097561e-05, "loss": 0.3347, "step": 5011 }, { "epoch": 0.87, "learning_rate": 2.2299651567944253e-05, "loss": 0.3543, "step": 5012 }, { "epoch": 0.87, "learning_rate": 2.2343205574912893e-05, "loss": 0.3628, "step": 5013 }, { "epoch": 0.87, "learning_rate": 2.2386759581881533e-05, "loss": 0.3345, "step": 5014 }, { "epoch": 0.87, "learning_rate": 2.2430313588850173e-05, "loss": 0.3422, "step": 5015 }, { "epoch": 0.87, "learning_rate": 2.2473867595818816e-05, "loss": 0.36, "step": 5016 }, { "epoch": 0.87, "learning_rate": 2.251742160278746e-05, "loss": 0.3371, "step": 5017 }, { "epoch": 0.87, "learning_rate": 2.25609756097561e-05, "loss": 0.3439, "step": 5018 }, { "epoch": 0.87, "learning_rate": 2.260452961672474e-05, "loss": 0.3544, "step": 5019 }, { "epoch": 0.87, "learning_rate": 2.264808362369338e-05, "loss": 0.3424, "step": 5020 }, { "epoch": 0.88, "learning_rate": 2.2691637630662022e-05, "loss": 0.351, "step": 5021 }, { "epoch": 0.88, "learning_rate": 2.2735191637630662e-05, "loss": 0.341, "step": 5022 }, { "epoch": 0.88, "learning_rate": 2.2778745644599302e-05, "loss": 0.3545, "step": 5023 }, { "epoch": 0.88, "learning_rate": 2.2822299651567945e-05, "loss": 0.338, "step": 5024 }, { "epoch": 0.88, "learning_rate": 2.286585365853659e-05, "loss": 0.3312, "step": 5025 }, { "epoch": 0.88, "learning_rate": 2.290940766550523e-05, "loss": 0.3644, "step": 5026 }, { "epoch": 0.88, "learning_rate": 2.295296167247387e-05, "loss": 0.3473, "step": 5027 }, { "epoch": 0.88, "learning_rate": 2.299651567944251e-05, "loss": 0.3471, "step": 5028 }, { "epoch": 0.88, "learning_rate": 2.3040069686411152e-05, "loss": 0.3469, "step": 5029 }, { "epoch": 0.88, "learning_rate": 2.308362369337979e-05, "loss": 0.3283, "step": 5030 }, { "epoch": 0.88, "learning_rate": 2.312717770034843e-05, "loss": 0.3526, "step": 5031 }, { "epoch": 0.88, "learning_rate": 2.3170731707317075e-05, "loss": 0.3348, "step": 5032 }, { "epoch": 0.88, "learning_rate": 2.3214285714285715e-05, "loss": 0.3547, "step": 5033 }, { "epoch": 0.88, "learning_rate": 2.3257839721254358e-05, "loss": 0.358, "step": 5034 }, { "epoch": 0.88, "learning_rate": 2.3301393728222998e-05, "loss": 0.3493, "step": 5035 }, { "epoch": 0.88, "learning_rate": 2.3344947735191638e-05, "loss": 0.3319, "step": 5036 }, { "epoch": 0.88, "learning_rate": 2.3388501742160278e-05, "loss": 0.3508, "step": 5037 }, { "epoch": 0.88, "learning_rate": 2.343205574912892e-05, "loss": 0.3431, "step": 5038 }, { "epoch": 0.88, "learning_rate": 2.347560975609756e-05, "loss": 0.3307, "step": 5039 }, { "epoch": 0.88, "learning_rate": 2.3519163763066204e-05, "loss": 0.3515, "step": 5040 }, { "epoch": 0.88, "learning_rate": 2.3562717770034844e-05, "loss": 0.3575, "step": 5041 }, { "epoch": 0.88, "learning_rate": 2.3606271777003487e-05, "loss": 0.3646, "step": 5042 }, { "epoch": 0.88, "learning_rate": 2.3649825783972127e-05, "loss": 0.3568, "step": 5043 }, { "epoch": 0.88, "learning_rate": 2.3693379790940767e-05, "loss": 0.3424, "step": 5044 }, { "epoch": 0.88, "learning_rate": 2.3736933797909407e-05, "loss": 0.3518, "step": 5045 }, { "epoch": 0.88, "learning_rate": 2.378048780487805e-05, "loss": 0.3498, "step": 5046 }, { "epoch": 0.88, "learning_rate": 2.3824041811846694e-05, "loss": 0.3445, "step": 5047 }, { "epoch": 0.88, "learning_rate": 2.3867595818815333e-05, "loss": 0.3465, "step": 5048 }, { "epoch": 0.88, "learning_rate": 2.3911149825783973e-05, "loss": 0.3566, "step": 5049 }, { "epoch": 0.88, "learning_rate": 2.3954703832752613e-05, "loss": 0.3511, "step": 5050 }, { "epoch": 0.88, "learning_rate": 2.3998257839721257e-05, "loss": 0.3548, "step": 5051 }, { "epoch": 0.88, "learning_rate": 2.4041811846689896e-05, "loss": 0.3394, "step": 5052 }, { "epoch": 0.88, "learning_rate": 2.4085365853658536e-05, "loss": 0.3535, "step": 5053 }, { "epoch": 0.88, "learning_rate": 2.4128919860627176e-05, "loss": 0.3376, "step": 5054 }, { "epoch": 0.88, "learning_rate": 2.4172473867595823e-05, "loss": 0.3428, "step": 5055 }, { "epoch": 0.88, "learning_rate": 2.4216027874564463e-05, "loss": 0.3514, "step": 5056 }, { "epoch": 0.88, "learning_rate": 2.4259581881533103e-05, "loss": 0.3539, "step": 5057 }, { "epoch": 0.88, "learning_rate": 2.4303135888501743e-05, "loss": 0.3385, "step": 5058 }, { "epoch": 0.88, "learning_rate": 2.4346689895470386e-05, "loss": 0.351, "step": 5059 }, { "epoch": 0.88, "learning_rate": 2.4390243902439026e-05, "loss": 0.331, "step": 5060 }, { "epoch": 0.88, "learning_rate": 2.4433797909407666e-05, "loss": 0.3471, "step": 5061 }, { "epoch": 0.88, "learning_rate": 2.4477351916376306e-05, "loss": 0.3446, "step": 5062 }, { "epoch": 0.88, "learning_rate": 2.452090592334495e-05, "loss": 0.3588, "step": 5063 }, { "epoch": 0.88, "learning_rate": 2.4564459930313592e-05, "loss": 0.3585, "step": 5064 }, { "epoch": 0.88, "learning_rate": 2.4608013937282232e-05, "loss": 0.354, "step": 5065 }, { "epoch": 0.88, "learning_rate": 2.4651567944250872e-05, "loss": 0.3518, "step": 5066 }, { "epoch": 0.88, "learning_rate": 2.4695121951219512e-05, "loss": 0.3239, "step": 5067 }, { "epoch": 0.88, "learning_rate": 2.4738675958188155e-05, "loss": 0.3629, "step": 5068 }, { "epoch": 0.88, "learning_rate": 2.4782229965156795e-05, "loss": 0.3556, "step": 5069 }, { "epoch": 0.88, "learning_rate": 2.4825783972125435e-05, "loss": 0.3427, "step": 5070 }, { "epoch": 0.88, "learning_rate": 2.4869337979094078e-05, "loss": 0.353, "step": 5071 }, { "epoch": 0.88, "learning_rate": 2.4912891986062718e-05, "loss": 0.3551, "step": 5072 }, { "epoch": 0.88, "learning_rate": 2.495644599303136e-05, "loss": 0.3505, "step": 5073 }, { "epoch": 0.88, "learning_rate": 2.5e-05, "loss": 0.3533, "step": 5074 }, { "epoch": 0.88, "learning_rate": 2.5043554006968644e-05, "loss": 0.3355, "step": 5075 }, { "epoch": 0.88, "learning_rate": 2.5087108013937284e-05, "loss": 0.3525, "step": 5076 }, { "epoch": 0.88, "learning_rate": 2.5130662020905928e-05, "loss": 0.3364, "step": 5077 }, { "epoch": 0.89, "learning_rate": 2.5174216027874564e-05, "loss": 0.3464, "step": 5078 }, { "epoch": 0.89, "learning_rate": 2.5217770034843207e-05, "loss": 0.3443, "step": 5079 }, { "epoch": 0.89, "learning_rate": 2.5261324041811847e-05, "loss": 0.3484, "step": 5080 }, { "epoch": 0.89, "learning_rate": 2.530487804878049e-05, "loss": 0.3598, "step": 5081 }, { "epoch": 0.89, "learning_rate": 2.5348432055749134e-05, "loss": 0.3542, "step": 5082 }, { "epoch": 0.89, "learning_rate": 2.539198606271777e-05, "loss": 0.3417, "step": 5083 }, { "epoch": 0.89, "learning_rate": 2.5435540069686414e-05, "loss": 0.346, "step": 5084 }, { "epoch": 0.89, "learning_rate": 2.5479094076655054e-05, "loss": 0.3685, "step": 5085 }, { "epoch": 0.89, "learning_rate": 2.5522648083623697e-05, "loss": 0.3404, "step": 5086 }, { "epoch": 0.89, "learning_rate": 2.5566202090592333e-05, "loss": 0.3387, "step": 5087 }, { "epoch": 0.89, "learning_rate": 2.5609756097560977e-05, "loss": 0.3503, "step": 5088 }, { "epoch": 0.89, "learning_rate": 2.5653310104529617e-05, "loss": 0.3412, "step": 5089 }, { "epoch": 0.89, "learning_rate": 2.569686411149826e-05, "loss": 0.3468, "step": 5090 }, { "epoch": 0.89, "learning_rate": 2.5740418118466903e-05, "loss": 0.3394, "step": 5091 }, { "epoch": 0.89, "learning_rate": 2.578397212543554e-05, "loss": 0.3333, "step": 5092 }, { "epoch": 0.89, "learning_rate": 2.5827526132404183e-05, "loss": 0.3377, "step": 5093 }, { "epoch": 0.89, "learning_rate": 2.5871080139372823e-05, "loss": 0.3609, "step": 5094 }, { "epoch": 0.89, "learning_rate": 2.5914634146341466e-05, "loss": 0.3441, "step": 5095 }, { "epoch": 0.89, "learning_rate": 2.5958188153310103e-05, "loss": 0.3567, "step": 5096 }, { "epoch": 0.89, "learning_rate": 2.6001742160278746e-05, "loss": 0.3601, "step": 5097 }, { "epoch": 0.89, "learning_rate": 2.604529616724739e-05, "loss": 0.3381, "step": 5098 }, { "epoch": 0.89, "learning_rate": 2.608885017421603e-05, "loss": 0.3322, "step": 5099 }, { "epoch": 0.89, "learning_rate": 2.6132404181184672e-05, "loss": 0.3569, "step": 5100 }, { "epoch": 0.89, "learning_rate": 2.617595818815331e-05, "loss": 0.3417, "step": 5101 }, { "epoch": 0.89, "learning_rate": 2.6219512195121952e-05, "loss": 0.3583, "step": 5102 }, { "epoch": 0.89, "learning_rate": 2.6263066202090592e-05, "loss": 0.3544, "step": 5103 }, { "epoch": 0.89, "learning_rate": 2.6306620209059235e-05, "loss": 0.3396, "step": 5104 }, { "epoch": 0.89, "learning_rate": 2.6350174216027872e-05, "loss": 0.3535, "step": 5105 }, { "epoch": 0.89, "learning_rate": 2.6393728222996515e-05, "loss": 0.3522, "step": 5106 }, { "epoch": 0.89, "learning_rate": 2.6437282229965162e-05, "loss": 0.3527, "step": 5107 }, { "epoch": 0.89, "learning_rate": 2.6480836236933798e-05, "loss": 0.3329, "step": 5108 }, { "epoch": 0.89, "learning_rate": 2.652439024390244e-05, "loss": 0.3605, "step": 5109 }, { "epoch": 0.89, "learning_rate": 2.6567944250871078e-05, "loss": 0.348, "step": 5110 }, { "epoch": 0.89, "learning_rate": 2.6611498257839725e-05, "loss": 0.3311, "step": 5111 }, { "epoch": 0.89, "learning_rate": 2.665505226480836e-05, "loss": 0.3486, "step": 5112 }, { "epoch": 0.89, "learning_rate": 2.6698606271777005e-05, "loss": 0.3256, "step": 5113 }, { "epoch": 0.89, "learning_rate": 2.6742160278745648e-05, "loss": 0.3334, "step": 5114 }, { "epoch": 0.89, "learning_rate": 2.6785714285714288e-05, "loss": 0.3513, "step": 5115 }, { "epoch": 0.89, "learning_rate": 2.682926829268293e-05, "loss": 0.3434, "step": 5116 }, { "epoch": 0.89, "learning_rate": 2.6872822299651568e-05, "loss": 0.3534, "step": 5117 }, { "epoch": 0.89, "learning_rate": 2.691637630662021e-05, "loss": 0.3438, "step": 5118 }, { "epoch": 0.89, "learning_rate": 2.695993031358885e-05, "loss": 0.3475, "step": 5119 }, { "epoch": 0.89, "learning_rate": 2.7003484320557494e-05, "loss": 0.3379, "step": 5120 }, { "epoch": 0.89, "learning_rate": 2.7047038327526137e-05, "loss": 0.3503, "step": 5121 }, { "epoch": 0.89, "learning_rate": 2.7090592334494774e-05, "loss": 0.3384, "step": 5122 }, { "epoch": 0.89, "learning_rate": 2.7134146341463417e-05, "loss": 0.3368, "step": 5123 }, { "epoch": 0.89, "learning_rate": 2.7177700348432057e-05, "loss": 0.346, "step": 5124 }, { "epoch": 0.89, "learning_rate": 2.72212543554007e-05, "loss": 0.3476, "step": 5125 }, { "epoch": 0.89, "learning_rate": 2.7264808362369337e-05, "loss": 0.3504, "step": 5126 }, { "epoch": 0.89, "learning_rate": 2.730836236933798e-05, "loss": 0.3485, "step": 5127 }, { "epoch": 0.89, "learning_rate": 2.735191637630662e-05, "loss": 0.3506, "step": 5128 }, { "epoch": 0.89, "learning_rate": 2.7395470383275263e-05, "loss": 0.3386, "step": 5129 }, { "epoch": 0.89, "learning_rate": 2.7439024390243906e-05, "loss": 0.3313, "step": 5130 }, { "epoch": 0.89, "learning_rate": 2.7482578397212543e-05, "loss": 0.3347, "step": 5131 }, { "epoch": 0.89, "learning_rate": 2.7526132404181186e-05, "loss": 0.3422, "step": 5132 }, { "epoch": 0.89, "learning_rate": 2.7569686411149826e-05, "loss": 0.3602, "step": 5133 }, { "epoch": 0.89, "learning_rate": 2.761324041811847e-05, "loss": 0.3474, "step": 5134 }, { "epoch": 0.9, "learning_rate": 2.7656794425087106e-05, "loss": 0.3549, "step": 5135 }, { "epoch": 0.9, "learning_rate": 2.770034843205575e-05, "loss": 0.3373, "step": 5136 }, { "epoch": 0.9, "learning_rate": 2.7743902439024393e-05, "loss": 0.3397, "step": 5137 }, { "epoch": 0.9, "learning_rate": 2.7787456445993032e-05, "loss": 0.3409, "step": 5138 }, { "epoch": 0.9, "learning_rate": 2.7831010452961676e-05, "loss": 0.3378, "step": 5139 }, { "epoch": 0.9, "learning_rate": 2.7874564459930312e-05, "loss": 0.3476, "step": 5140 }, { "epoch": 0.9, "learning_rate": 2.7918118466898955e-05, "loss": 0.35, "step": 5141 }, { "epoch": 0.9, "learning_rate": 2.7961672473867595e-05, "loss": 0.3499, "step": 5142 }, { "epoch": 0.9, "learning_rate": 2.800522648083624e-05, "loss": 0.3457, "step": 5143 }, { "epoch": 0.9, "learning_rate": 2.8048780487804882e-05, "loss": 0.3342, "step": 5144 }, { "epoch": 0.9, "learning_rate": 2.809233449477352e-05, "loss": 0.3586, "step": 5145 }, { "epoch": 0.9, "learning_rate": 2.8135888501742165e-05, "loss": 0.3528, "step": 5146 }, { "epoch": 0.9, "learning_rate": 2.81794425087108e-05, "loss": 0.3376, "step": 5147 }, { "epoch": 0.9, "learning_rate": 2.8222996515679445e-05, "loss": 0.3601, "step": 5148 }, { "epoch": 0.9, "learning_rate": 2.826655052264808e-05, "loss": 0.3512, "step": 5149 }, { "epoch": 0.9, "learning_rate": 2.8310104529616728e-05, "loss": 0.3614, "step": 5150 }, { "epoch": 0.9, "learning_rate": 2.8353658536585365e-05, "loss": 0.3406, "step": 5151 }, { "epoch": 0.9, "learning_rate": 2.8397212543554008e-05, "loss": 0.3402, "step": 5152 }, { "epoch": 0.9, "learning_rate": 2.844076655052265e-05, "loss": 0.3534, "step": 5153 }, { "epoch": 0.9, "learning_rate": 2.848432055749129e-05, "loss": 0.3559, "step": 5154 }, { "epoch": 0.9, "learning_rate": 2.8527874564459934e-05, "loss": 0.3446, "step": 5155 }, { "epoch": 0.9, "learning_rate": 2.857142857142857e-05, "loss": 0.3235, "step": 5156 }, { "epoch": 0.9, "learning_rate": 2.8614982578397214e-05, "loss": 0.3446, "step": 5157 }, { "epoch": 0.9, "learning_rate": 2.8658536585365854e-05, "loss": 0.3514, "step": 5158 }, { "epoch": 0.9, "learning_rate": 2.8702090592334497e-05, "loss": 0.3367, "step": 5159 }, { "epoch": 0.9, "learning_rate": 2.874564459930314e-05, "loss": 0.3584, "step": 5160 }, { "epoch": 0.9, "learning_rate": 2.8789198606271777e-05, "loss": 0.3461, "step": 5161 }, { "epoch": 0.9, "learning_rate": 2.883275261324042e-05, "loss": 0.3575, "step": 5162 }, { "epoch": 0.9, "learning_rate": 2.887630662020906e-05, "loss": 0.3405, "step": 5163 }, { "epoch": 0.9, "learning_rate": 2.8919860627177704e-05, "loss": 0.3336, "step": 5164 }, { "epoch": 0.9, "learning_rate": 2.896341463414634e-05, "loss": 0.3401, "step": 5165 }, { "epoch": 0.9, "learning_rate": 2.9006968641114983e-05, "loss": 0.3508, "step": 5166 }, { "epoch": 0.9, "learning_rate": 2.9050522648083623e-05, "loss": 0.3469, "step": 5167 }, { "epoch": 0.9, "learning_rate": 2.9094076655052267e-05, "loss": 0.3665, "step": 5168 }, { "epoch": 0.9, "learning_rate": 2.913763066202091e-05, "loss": 0.3438, "step": 5169 }, { "epoch": 0.9, "learning_rate": 2.9181184668989546e-05, "loss": 0.3447, "step": 5170 }, { "epoch": 0.9, "learning_rate": 2.922473867595819e-05, "loss": 0.3366, "step": 5171 }, { "epoch": 0.9, "learning_rate": 2.926829268292683e-05, "loss": 0.3535, "step": 5172 }, { "epoch": 0.9, "learning_rate": 2.9311846689895473e-05, "loss": 0.3468, "step": 5173 }, { "epoch": 0.9, "learning_rate": 2.935540069686411e-05, "loss": 0.3731, "step": 5174 }, { "epoch": 0.9, "learning_rate": 2.9398954703832753e-05, "loss": 0.3587, "step": 5175 }, { "epoch": 0.9, "learning_rate": 2.9442508710801396e-05, "loss": 0.3496, "step": 5176 }, { "epoch": 0.9, "learning_rate": 2.9486062717770036e-05, "loss": 0.3544, "step": 5177 }, { "epoch": 0.9, "learning_rate": 2.952961672473868e-05, "loss": 0.3483, "step": 5178 }, { "epoch": 0.9, "learning_rate": 2.9573170731707316e-05, "loss": 0.3429, "step": 5179 }, { "epoch": 0.9, "learning_rate": 2.9616724738675962e-05, "loss": 0.3442, "step": 5180 }, { "epoch": 0.9, "learning_rate": 2.96602787456446e-05, "loss": 0.3503, "step": 5181 }, { "epoch": 0.9, "learning_rate": 2.9703832752613242e-05, "loss": 0.3524, "step": 5182 }, { "epoch": 0.9, "learning_rate": 2.9747386759581885e-05, "loss": 0.3428, "step": 5183 }, { "epoch": 0.9, "learning_rate": 2.9790940766550525e-05, "loss": 0.3243, "step": 5184 }, { "epoch": 0.9, "learning_rate": 2.983449477351917e-05, "loss": 0.3508, "step": 5185 }, { "epoch": 0.9, "learning_rate": 2.9878048780487805e-05, "loss": 0.3294, "step": 5186 }, { "epoch": 0.9, "learning_rate": 2.9921602787456448e-05, "loss": 0.3406, "step": 5187 }, { "epoch": 0.9, "learning_rate": 2.9965156794425088e-05, "loss": 0.3298, "step": 5188 }, { "epoch": 0.9, "learning_rate": 3.000871080139373e-05, "loss": 0.361, "step": 5189 }, { "epoch": 0.9, "learning_rate": 3.0052264808362368e-05, "loss": 0.3459, "step": 5190 }, { "epoch": 0.9, "learning_rate": 3.009581881533101e-05, "loss": 0.3482, "step": 5191 }, { "epoch": 0.9, "learning_rate": 3.0139372822299655e-05, "loss": 0.3434, "step": 5192 }, { "epoch": 0.91, "learning_rate": 3.0182926829268294e-05, "loss": 0.341, "step": 5193 }, { "epoch": 0.91, "learning_rate": 3.0226480836236938e-05, "loss": 0.333, "step": 5194 }, { "epoch": 0.91, "learning_rate": 3.0270034843205574e-05, "loss": 0.3617, "step": 5195 }, { "epoch": 0.91, "learning_rate": 3.0313588850174217e-05, "loss": 0.3424, "step": 5196 }, { "epoch": 0.91, "learning_rate": 3.0357142857142857e-05, "loss": 0.3437, "step": 5197 }, { "epoch": 0.91, "learning_rate": 3.04006968641115e-05, "loss": 0.3478, "step": 5198 }, { "epoch": 0.91, "learning_rate": 3.0444250871080144e-05, "loss": 0.3491, "step": 5199 }, { "epoch": 0.91, "learning_rate": 3.048780487804878e-05, "loss": 0.3438, "step": 5200 }, { "epoch": 0.91, "learning_rate": 3.0531358885017424e-05, "loss": 0.353, "step": 5201 }, { "epoch": 0.91, "learning_rate": 3.057491289198606e-05, "loss": 0.3533, "step": 5202 }, { "epoch": 0.91, "learning_rate": 3.0618466898954704e-05, "loss": 0.3297, "step": 5203 }, { "epoch": 0.91, "learning_rate": 3.066202090592335e-05, "loss": 0.3485, "step": 5204 }, { "epoch": 0.91, "learning_rate": 3.070557491289199e-05, "loss": 0.3575, "step": 5205 }, { "epoch": 0.91, "learning_rate": 3.074912891986063e-05, "loss": 0.34, "step": 5206 }, { "epoch": 0.91, "learning_rate": 3.079268292682927e-05, "loss": 0.3571, "step": 5207 }, { "epoch": 0.91, "learning_rate": 3.083623693379791e-05, "loss": 0.3469, "step": 5208 }, { "epoch": 0.91, "learning_rate": 3.087979094076655e-05, "loss": 0.3397, "step": 5209 }, { "epoch": 0.91, "learning_rate": 3.092334494773519e-05, "loss": 0.3498, "step": 5210 }, { "epoch": 0.91, "learning_rate": 3.096689895470383e-05, "loss": 0.3423, "step": 5211 }, { "epoch": 0.91, "learning_rate": 3.101045296167247e-05, "loss": 0.3378, "step": 5212 }, { "epoch": 0.91, "learning_rate": 3.1054006968641116e-05, "loss": 0.3471, "step": 5213 }, { "epoch": 0.91, "learning_rate": 3.109756097560976e-05, "loss": 0.3469, "step": 5214 }, { "epoch": 0.91, "learning_rate": 3.11411149825784e-05, "loss": 0.347, "step": 5215 }, { "epoch": 0.91, "learning_rate": 3.118466898954704e-05, "loss": 0.3533, "step": 5216 }, { "epoch": 0.91, "learning_rate": 3.122822299651568e-05, "loss": 0.344, "step": 5217 }, { "epoch": 0.91, "learning_rate": 3.127177700348432e-05, "loss": 0.3273, "step": 5218 }, { "epoch": 0.91, "learning_rate": 3.131533101045296e-05, "loss": 0.3458, "step": 5219 }, { "epoch": 0.91, "learning_rate": 3.13588850174216e-05, "loss": 0.339, "step": 5220 }, { "epoch": 0.91, "learning_rate": 3.140243902439025e-05, "loss": 0.3341, "step": 5221 }, { "epoch": 0.91, "learning_rate": 3.144599303135889e-05, "loss": 0.3446, "step": 5222 }, { "epoch": 0.91, "learning_rate": 3.148954703832753e-05, "loss": 0.353, "step": 5223 }, { "epoch": 0.91, "learning_rate": 3.153310104529617e-05, "loss": 0.3356, "step": 5224 }, { "epoch": 0.91, "learning_rate": 3.157665505226481e-05, "loss": 0.3384, "step": 5225 }, { "epoch": 0.91, "learning_rate": 3.162020905923345e-05, "loss": 0.3492, "step": 5226 }, { "epoch": 0.91, "learning_rate": 3.166376306620209e-05, "loss": 0.3515, "step": 5227 }, { "epoch": 0.91, "learning_rate": 3.170731707317073e-05, "loss": 0.343, "step": 5228 }, { "epoch": 0.91, "learning_rate": 3.1750871080139375e-05, "loss": 0.3633, "step": 5229 }, { "epoch": 0.91, "learning_rate": 3.179442508710802e-05, "loss": 0.362, "step": 5230 }, { "epoch": 0.91, "learning_rate": 3.183797909407666e-05, "loss": 0.3392, "step": 5231 }, { "epoch": 0.91, "learning_rate": 3.18815331010453e-05, "loss": 0.3586, "step": 5232 }, { "epoch": 0.91, "learning_rate": 3.192508710801394e-05, "loss": 0.3367, "step": 5233 }, { "epoch": 0.91, "learning_rate": 3.196864111498258e-05, "loss": 0.3484, "step": 5234 }, { "epoch": 0.91, "learning_rate": 3.201219512195122e-05, "loss": 0.3512, "step": 5235 }, { "epoch": 0.91, "learning_rate": 3.205574912891986e-05, "loss": 0.3496, "step": 5236 }, { "epoch": 0.91, "learning_rate": 3.20993031358885e-05, "loss": 0.3347, "step": 5237 }, { "epoch": 0.91, "learning_rate": 3.2142857142857144e-05, "loss": 0.3687, "step": 5238 }, { "epoch": 0.91, "learning_rate": 3.218641114982579e-05, "loss": 0.3499, "step": 5239 }, { "epoch": 0.91, "learning_rate": 3.222996515679443e-05, "loss": 0.3472, "step": 5240 }, { "epoch": 0.91, "learning_rate": 3.227351916376307e-05, "loss": 0.3308, "step": 5241 }, { "epoch": 0.91, "learning_rate": 3.231707317073171e-05, "loss": 0.3414, "step": 5242 }, { "epoch": 0.91, "learning_rate": 3.236062717770035e-05, "loss": 0.3504, "step": 5243 }, { "epoch": 0.91, "learning_rate": 3.240418118466899e-05, "loss": 0.3534, "step": 5244 }, { "epoch": 0.91, "learning_rate": 3.244773519163763e-05, "loss": 0.3387, "step": 5245 }, { "epoch": 0.91, "learning_rate": 3.249128919860627e-05, "loss": 0.3413, "step": 5246 }, { "epoch": 0.91, "learning_rate": 3.253484320557491e-05, "loss": 0.3263, "step": 5247 }, { "epoch": 0.91, "learning_rate": 3.2578397212543556e-05, "loss": 0.3434, "step": 5248 }, { "epoch": 0.91, "learning_rate": 3.26219512195122e-05, "loss": 0.3699, "step": 5249 }, { "epoch": 0.92, "learning_rate": 3.2665505226480836e-05, "loss": 0.3437, "step": 5250 }, { "epoch": 0.92, "learning_rate": 3.270905923344948e-05, "loss": 0.3384, "step": 5251 }, { "epoch": 0.92, "learning_rate": 3.2752613240418116e-05, "loss": 0.35, "step": 5252 }, { "epoch": 0.92, "learning_rate": 3.279616724738676e-05, "loss": 0.3251, "step": 5253 }, { "epoch": 0.92, "learning_rate": 3.28397212543554e-05, "loss": 0.3494, "step": 5254 }, { "epoch": 0.92, "learning_rate": 3.288327526132404e-05, "loss": 0.3431, "step": 5255 }, { "epoch": 0.92, "learning_rate": 3.292682926829269e-05, "loss": 0.3438, "step": 5256 }, { "epoch": 0.92, "learning_rate": 3.2970383275261326e-05, "loss": 0.3468, "step": 5257 }, { "epoch": 0.92, "learning_rate": 3.301393728222997e-05, "loss": 0.3538, "step": 5258 }, { "epoch": 0.92, "learning_rate": 3.3057491289198605e-05, "loss": 0.3538, "step": 5259 }, { "epoch": 0.92, "learning_rate": 3.310104529616725e-05, "loss": 0.3587, "step": 5260 }, { "epoch": 0.92, "learning_rate": 3.314459930313589e-05, "loss": 0.3537, "step": 5261 }, { "epoch": 0.92, "learning_rate": 3.318815331010453e-05, "loss": 0.3474, "step": 5262 }, { "epoch": 0.92, "learning_rate": 3.323170731707317e-05, "loss": 0.3402, "step": 5263 }, { "epoch": 0.92, "learning_rate": 3.3275261324041815e-05, "loss": 0.354, "step": 5264 }, { "epoch": 0.92, "learning_rate": 3.331881533101046e-05, "loss": 0.3448, "step": 5265 }, { "epoch": 0.92, "learning_rate": 3.3362369337979095e-05, "loss": 0.3434, "step": 5266 }, { "epoch": 0.92, "learning_rate": 3.340592334494774e-05, "loss": 0.3448, "step": 5267 }, { "epoch": 0.92, "learning_rate": 3.344947735191638e-05, "loss": 0.3484, "step": 5268 }, { "epoch": 0.92, "learning_rate": 3.349303135888502e-05, "loss": 0.3511, "step": 5269 }, { "epoch": 0.92, "learning_rate": 3.353658536585366e-05, "loss": 0.3478, "step": 5270 }, { "epoch": 0.92, "learning_rate": 3.35801393728223e-05, "loss": 0.3386, "step": 5271 }, { "epoch": 0.92, "learning_rate": 3.362369337979094e-05, "loss": 0.3441, "step": 5272 }, { "epoch": 0.92, "learning_rate": 3.3667247386759584e-05, "loss": 0.3585, "step": 5273 }, { "epoch": 0.92, "learning_rate": 3.371080139372823e-05, "loss": 0.3368, "step": 5274 }, { "epoch": 0.92, "learning_rate": 3.3754355400696864e-05, "loss": 0.3601, "step": 5275 }, { "epoch": 0.92, "learning_rate": 3.379790940766551e-05, "loss": 0.3557, "step": 5276 }, { "epoch": 0.92, "learning_rate": 3.384146341463415e-05, "loss": 0.3485, "step": 5277 }, { "epoch": 0.92, "learning_rate": 3.388501742160279e-05, "loss": 0.3601, "step": 5278 }, { "epoch": 0.92, "learning_rate": 3.392857142857143e-05, "loss": 0.3732, "step": 5279 }, { "epoch": 0.92, "learning_rate": 3.397212543554007e-05, "loss": 0.3563, "step": 5280 }, { "epoch": 0.92, "learning_rate": 3.401567944250871e-05, "loss": 0.3436, "step": 5281 }, { "epoch": 0.92, "learning_rate": 3.4059233449477354e-05, "loss": 0.351, "step": 5282 }, { "epoch": 0.92, "learning_rate": 3.4102787456446e-05, "loss": 0.3493, "step": 5283 }, { "epoch": 0.92, "learning_rate": 3.414634146341464e-05, "loss": 0.3404, "step": 5284 }, { "epoch": 0.92, "learning_rate": 3.4189895470383277e-05, "loss": 0.355, "step": 5285 }, { "epoch": 0.92, "learning_rate": 3.423344947735192e-05, "loss": 0.3769, "step": 5286 }, { "epoch": 0.92, "learning_rate": 3.4277003484320556e-05, "loss": 0.3428, "step": 5287 }, { "epoch": 0.92, "learning_rate": 3.43205574912892e-05, "loss": 0.3311, "step": 5288 }, { "epoch": 0.92, "learning_rate": 3.4364111498257836e-05, "loss": 0.3345, "step": 5289 }, { "epoch": 0.92, "learning_rate": 3.4407665505226486e-05, "loss": 0.3423, "step": 5290 }, { "epoch": 0.92, "learning_rate": 3.445121951219512e-05, "loss": 0.3489, "step": 5291 }, { "epoch": 0.92, "learning_rate": 3.4494773519163766e-05, "loss": 0.3683, "step": 5292 }, { "epoch": 0.92, "learning_rate": 3.453832752613241e-05, "loss": 0.3533, "step": 5293 }, { "epoch": 0.92, "learning_rate": 3.4581881533101046e-05, "loss": 0.3486, "step": 5294 }, { "epoch": 0.92, "learning_rate": 3.462543554006969e-05, "loss": 0.3457, "step": 5295 }, { "epoch": 0.92, "learning_rate": 3.4668989547038326e-05, "loss": 0.3347, "step": 5296 }, { "epoch": 0.92, "learning_rate": 3.471254355400697e-05, "loss": 0.3399, "step": 5297 }, { "epoch": 0.92, "learning_rate": 3.475609756097561e-05, "loss": 0.3534, "step": 5298 }, { "epoch": 0.92, "learning_rate": 3.4799651567944255e-05, "loss": 0.3477, "step": 5299 }, { "epoch": 0.92, "learning_rate": 3.48432055749129e-05, "loss": 0.3513, "step": 5300 }, { "epoch": 0.92, "learning_rate": 3.4886759581881535e-05, "loss": 0.3428, "step": 5301 }, { "epoch": 0.92, "learning_rate": 3.493031358885018e-05, "loss": 0.3535, "step": 5302 }, { "epoch": 0.92, "learning_rate": 3.4973867595818815e-05, "loss": 0.3494, "step": 5303 }, { "epoch": 0.92, "learning_rate": 3.501742160278746e-05, "loss": 0.3653, "step": 5304 }, { "epoch": 0.92, "learning_rate": 3.5060975609756095e-05, "loss": 0.3393, "step": 5305 }, { "epoch": 0.92, "learning_rate": 3.510452961672474e-05, "loss": 0.3478, "step": 5306 }, { "epoch": 0.93, "learning_rate": 3.514808362369338e-05, "loss": 0.3393, "step": 5307 }, { "epoch": 0.93, "learning_rate": 3.5191637630662025e-05, "loss": 0.3311, "step": 5308 }, { "epoch": 0.93, "learning_rate": 3.523519163763067e-05, "loss": 0.3477, "step": 5309 }, { "epoch": 0.93, "learning_rate": 3.5278745644599304e-05, "loss": 0.3457, "step": 5310 }, { "epoch": 0.93, "learning_rate": 3.532229965156795e-05, "loss": 0.3429, "step": 5311 }, { "epoch": 0.93, "learning_rate": 3.5365853658536584e-05, "loss": 0.3539, "step": 5312 }, { "epoch": 0.93, "learning_rate": 3.540940766550523e-05, "loss": 0.3383, "step": 5313 }, { "epoch": 0.93, "learning_rate": 3.5452961672473864e-05, "loss": 0.3514, "step": 5314 }, { "epoch": 0.93, "learning_rate": 3.549651567944251e-05, "loss": 0.3512, "step": 5315 }, { "epoch": 0.93, "learning_rate": 3.554006968641115e-05, "loss": 0.3411, "step": 5316 }, { "epoch": 0.93, "learning_rate": 3.5583623693379794e-05, "loss": 0.3473, "step": 5317 }, { "epoch": 0.93, "learning_rate": 3.562717770034844e-05, "loss": 0.3487, "step": 5318 }, { "epoch": 0.93, "learning_rate": 3.5670731707317074e-05, "loss": 0.3261, "step": 5319 }, { "epoch": 0.93, "learning_rate": 3.571428571428572e-05, "loss": 0.3421, "step": 5320 }, { "epoch": 0.93, "learning_rate": 3.5757839721254353e-05, "loss": 0.3617, "step": 5321 }, { "epoch": 0.93, "learning_rate": 3.5801393728223e-05, "loss": 0.3482, "step": 5322 }, { "epoch": 0.93, "learning_rate": 3.584494773519164e-05, "loss": 0.3353, "step": 5323 }, { "epoch": 0.93, "learning_rate": 3.5888501742160277e-05, "loss": 0.3368, "step": 5324 }, { "epoch": 0.93, "learning_rate": 3.5932055749128927e-05, "loss": 0.3412, "step": 5325 }, { "epoch": 0.93, "learning_rate": 3.597560975609756e-05, "loss": 0.3342, "step": 5326 }, { "epoch": 0.93, "learning_rate": 3.6019163763066206e-05, "loss": 0.3561, "step": 5327 }, { "epoch": 0.93, "learning_rate": 3.606271777003484e-05, "loss": 0.3497, "step": 5328 }, { "epoch": 0.93, "learning_rate": 3.6106271777003486e-05, "loss": 0.3465, "step": 5329 }, { "epoch": 0.93, "learning_rate": 3.614982578397213e-05, "loss": 0.349, "step": 5330 }, { "epoch": 0.93, "learning_rate": 3.6193379790940766e-05, "loss": 0.3494, "step": 5331 }, { "epoch": 0.93, "learning_rate": 3.623693379790941e-05, "loss": 0.3436, "step": 5332 }, { "epoch": 0.93, "learning_rate": 3.628048780487805e-05, "loss": 0.3441, "step": 5333 }, { "epoch": 0.93, "learning_rate": 3.6324041811846696e-05, "loss": 0.3352, "step": 5334 }, { "epoch": 0.93, "learning_rate": 3.636759581881533e-05, "loss": 0.3404, "step": 5335 }, { "epoch": 0.93, "learning_rate": 3.6411149825783976e-05, "loss": 0.3412, "step": 5336 }, { "epoch": 0.93, "learning_rate": 3.645470383275261e-05, "loss": 0.331, "step": 5337 }, { "epoch": 0.93, "learning_rate": 3.6498257839721255e-05, "loss": 0.3536, "step": 5338 }, { "epoch": 0.93, "learning_rate": 3.65418118466899e-05, "loss": 0.3582, "step": 5339 }, { "epoch": 0.93, "learning_rate": 3.6585365853658535e-05, "loss": 0.3375, "step": 5340 }, { "epoch": 0.93, "learning_rate": 3.662891986062718e-05, "loss": 0.3361, "step": 5341 }, { "epoch": 0.93, "learning_rate": 3.667247386759582e-05, "loss": 0.3395, "step": 5342 }, { "epoch": 0.93, "learning_rate": 3.6716027874564465e-05, "loss": 0.333, "step": 5343 }, { "epoch": 0.93, "learning_rate": 3.67595818815331e-05, "loss": 0.3585, "step": 5344 }, { "epoch": 0.93, "learning_rate": 3.6803135888501745e-05, "loss": 0.3348, "step": 5345 }, { "epoch": 0.93, "learning_rate": 3.684668989547039e-05, "loss": 0.338, "step": 5346 }, { "epoch": 0.93, "learning_rate": 3.6890243902439025e-05, "loss": 0.3576, "step": 5347 }, { "epoch": 0.93, "learning_rate": 3.693379790940767e-05, "loss": 0.3278, "step": 5348 }, { "epoch": 0.93, "learning_rate": 3.6977351916376304e-05, "loss": 0.3388, "step": 5349 }, { "epoch": 0.93, "learning_rate": 3.702090592334495e-05, "loss": 0.3492, "step": 5350 }, { "epoch": 0.93, "learning_rate": 3.706445993031359e-05, "loss": 0.3369, "step": 5351 }, { "epoch": 0.93, "learning_rate": 3.7108013937282234e-05, "loss": 0.3362, "step": 5352 }, { "epoch": 0.93, "learning_rate": 3.715156794425087e-05, "loss": 0.3537, "step": 5353 }, { "epoch": 0.93, "learning_rate": 3.7195121951219514e-05, "loss": 0.3408, "step": 5354 }, { "epoch": 0.93, "learning_rate": 3.723867595818816e-05, "loss": 0.3545, "step": 5355 }, { "epoch": 0.93, "learning_rate": 3.7282229965156794e-05, "loss": 0.3403, "step": 5356 }, { "epoch": 0.93, "learning_rate": 3.732578397212544e-05, "loss": 0.3352, "step": 5357 }, { "epoch": 0.93, "learning_rate": 3.7369337979094074e-05, "loss": 0.345, "step": 5358 }, { "epoch": 0.93, "learning_rate": 3.741289198606272e-05, "loss": 0.3408, "step": 5359 }, { "epoch": 0.93, "learning_rate": 3.745644599303136e-05, "loss": 0.3246, "step": 5360 }, { "epoch": 0.93, "learning_rate": 3.7500000000000003e-05, "loss": 0.3455, "step": 5361 }, { "epoch": 0.93, "learning_rate": 3.754355400696865e-05, "loss": 0.3262, "step": 5362 }, { "epoch": 0.93, "learning_rate": 3.758710801393728e-05, "loss": 0.3355, "step": 5363 }, { "epoch": 0.93, "learning_rate": 3.7630662020905927e-05, "loss": 0.3486, "step": 5364 }, { "epoch": 0.94, "learning_rate": 3.767421602787456e-05, "loss": 0.349, "step": 5365 }, { "epoch": 0.94, "learning_rate": 3.7717770034843206e-05, "loss": 0.3514, "step": 5366 }, { "epoch": 0.94, "learning_rate": 3.776132404181185e-05, "loss": 0.347, "step": 5367 }, { "epoch": 0.94, "learning_rate": 3.780487804878049e-05, "loss": 0.3656, "step": 5368 }, { "epoch": 0.94, "learning_rate": 3.7848432055749136e-05, "loss": 0.3356, "step": 5369 }, { "epoch": 0.94, "learning_rate": 3.789198606271777e-05, "loss": 0.333, "step": 5370 }, { "epoch": 0.94, "learning_rate": 3.7935540069686416e-05, "loss": 0.3348, "step": 5371 }, { "epoch": 0.94, "learning_rate": 3.797909407665505e-05, "loss": 0.3302, "step": 5372 }, { "epoch": 0.94, "learning_rate": 3.8022648083623696e-05, "loss": 0.3416, "step": 5373 }, { "epoch": 0.94, "learning_rate": 3.806620209059233e-05, "loss": 0.3326, "step": 5374 }, { "epoch": 0.94, "learning_rate": 3.8109756097560976e-05, "loss": 0.3372, "step": 5375 }, { "epoch": 0.94, "learning_rate": 3.815331010452962e-05, "loss": 0.3539, "step": 5376 }, { "epoch": 0.94, "learning_rate": 3.819686411149826e-05, "loss": 0.3233, "step": 5377 }, { "epoch": 0.94, "learning_rate": 3.8240418118466905e-05, "loss": 0.3498, "step": 5378 }, { "epoch": 0.94, "learning_rate": 3.828397212543554e-05, "loss": 0.3518, "step": 5379 }, { "epoch": 0.94, "learning_rate": 3.8327526132404185e-05, "loss": 0.3449, "step": 5380 }, { "epoch": 0.94, "learning_rate": 3.837108013937282e-05, "loss": 0.3484, "step": 5381 }, { "epoch": 0.94, "learning_rate": 3.8414634146341465e-05, "loss": 0.3475, "step": 5382 }, { "epoch": 0.94, "learning_rate": 3.84581881533101e-05, "loss": 0.3418, "step": 5383 }, { "epoch": 0.94, "learning_rate": 3.8501742160278745e-05, "loss": 0.3514, "step": 5384 }, { "epoch": 0.94, "learning_rate": 3.854529616724739e-05, "loss": 0.3404, "step": 5385 }, { "epoch": 0.94, "learning_rate": 3.858885017421603e-05, "loss": 0.3498, "step": 5386 }, { "epoch": 0.94, "learning_rate": 3.8632404181184675e-05, "loss": 0.3464, "step": 5387 }, { "epoch": 0.94, "learning_rate": 3.867595818815331e-05, "loss": 0.3465, "step": 5388 }, { "epoch": 0.94, "learning_rate": 3.8719512195121954e-05, "loss": 0.3447, "step": 5389 }, { "epoch": 0.94, "learning_rate": 3.876306620209059e-05, "loss": 0.3449, "step": 5390 }, { "epoch": 0.94, "learning_rate": 3.8806620209059234e-05, "loss": 0.3592, "step": 5391 }, { "epoch": 0.94, "learning_rate": 3.885017421602787e-05, "loss": 0.3424, "step": 5392 }, { "epoch": 0.94, "learning_rate": 3.8893728222996514e-05, "loss": 0.3505, "step": 5393 }, { "epoch": 0.94, "learning_rate": 3.8937282229965164e-05, "loss": 0.3338, "step": 5394 }, { "epoch": 0.94, "learning_rate": 3.89808362369338e-05, "loss": 0.3597, "step": 5395 }, { "epoch": 0.94, "learning_rate": 3.9024390243902444e-05, "loss": 0.3457, "step": 5396 }, { "epoch": 0.94, "learning_rate": 3.906794425087108e-05, "loss": 0.3429, "step": 5397 }, { "epoch": 0.94, "learning_rate": 3.9111498257839724e-05, "loss": 0.346, "step": 5398 }, { "epoch": 0.94, "learning_rate": 3.915505226480836e-05, "loss": 0.3379, "step": 5399 }, { "epoch": 0.94, "learning_rate": 3.9198606271777003e-05, "loss": 0.357, "step": 5400 }, { "epoch": 0.94, "learning_rate": 3.924216027874565e-05, "loss": 0.3571, "step": 5401 }, { "epoch": 0.94, "learning_rate": 3.928571428571429e-05, "loss": 0.3466, "step": 5402 }, { "epoch": 0.94, "learning_rate": 3.932926829268293e-05, "loss": 0.3452, "step": 5403 }, { "epoch": 0.94, "learning_rate": 3.937282229965157e-05, "loss": 0.3413, "step": 5404 }, { "epoch": 0.94, "learning_rate": 3.941637630662021e-05, "loss": 0.3614, "step": 5405 }, { "epoch": 0.94, "learning_rate": 3.945993031358885e-05, "loss": 0.3383, "step": 5406 }, { "epoch": 0.94, "learning_rate": 3.950348432055749e-05, "loss": 0.345, "step": 5407 }, { "epoch": 0.94, "learning_rate": 3.9547038327526136e-05, "loss": 0.3398, "step": 5408 }, { "epoch": 0.94, "learning_rate": 3.959059233449477e-05, "loss": 0.3408, "step": 5409 }, { "epoch": 0.94, "learning_rate": 3.9634146341463416e-05, "loss": 0.3508, "step": 5410 }, { "epoch": 0.94, "learning_rate": 3.967770034843206e-05, "loss": 0.3466, "step": 5411 }, { "epoch": 0.94, "learning_rate": 3.97212543554007e-05, "loss": 0.3528, "step": 5412 }, { "epoch": 0.94, "learning_rate": 3.976480836236934e-05, "loss": 0.3432, "step": 5413 }, { "epoch": 0.94, "learning_rate": 3.980836236933798e-05, "loss": 0.3367, "step": 5414 }, { "epoch": 0.94, "learning_rate": 3.985191637630662e-05, "loss": 0.3437, "step": 5415 }, { "epoch": 0.94, "learning_rate": 3.989547038327526e-05, "loss": 0.3423, "step": 5416 }, { "epoch": 0.94, "learning_rate": 3.9939024390243905e-05, "loss": 0.3438, "step": 5417 }, { "epoch": 0.94, "learning_rate": 3.998257839721254e-05, "loss": 0.341, "step": 5418 }, { "epoch": 0.94, "learning_rate": 4.0026132404181185e-05, "loss": 0.3305, "step": 5419 }, { "epoch": 0.94, "learning_rate": 4.006968641114983e-05, "loss": 0.3447, "step": 5420 }, { "epoch": 0.94, "learning_rate": 4.011324041811847e-05, "loss": 0.3425, "step": 5421 }, { "epoch": 0.95, "learning_rate": 4.015679442508711e-05, "loss": 0.3493, "step": 5422 }, { "epoch": 0.95, "learning_rate": 4.020034843205575e-05, "loss": 0.3483, "step": 5423 }, { "epoch": 0.95, "learning_rate": 4.0243902439024395e-05, "loss": 0.3617, "step": 5424 }, { "epoch": 0.95, "learning_rate": 4.028745644599303e-05, "loss": 0.3199, "step": 5425 }, { "epoch": 0.95, "learning_rate": 4.0331010452961675e-05, "loss": 0.3493, "step": 5426 }, { "epoch": 0.95, "learning_rate": 4.037456445993031e-05, "loss": 0.3389, "step": 5427 }, { "epoch": 0.95, "learning_rate": 4.0418118466898954e-05, "loss": 0.3476, "step": 5428 }, { "epoch": 0.95, "learning_rate": 4.04616724738676e-05, "loss": 0.351, "step": 5429 }, { "epoch": 0.95, "learning_rate": 4.050522648083624e-05, "loss": 0.3397, "step": 5430 }, { "epoch": 0.95, "learning_rate": 4.0548780487804884e-05, "loss": 0.3309, "step": 5431 }, { "epoch": 0.95, "learning_rate": 4.059233449477352e-05, "loss": 0.3227, "step": 5432 }, { "epoch": 0.95, "learning_rate": 4.0635888501742164e-05, "loss": 0.3434, "step": 5433 }, { "epoch": 0.95, "learning_rate": 4.06794425087108e-05, "loss": 0.3528, "step": 5434 }, { "epoch": 0.95, "learning_rate": 4.0722996515679444e-05, "loss": 0.3302, "step": 5435 }, { "epoch": 0.95, "learning_rate": 4.076655052264808e-05, "loss": 0.3273, "step": 5436 }, { "epoch": 0.95, "learning_rate": 4.081010452961673e-05, "loss": 0.3296, "step": 5437 }, { "epoch": 0.95, "learning_rate": 4.085365853658537e-05, "loss": 0.3258, "step": 5438 }, { "epoch": 0.95, "learning_rate": 4.089721254355401e-05, "loss": 0.356, "step": 5439 }, { "epoch": 0.95, "learning_rate": 4.0940766550522653e-05, "loss": 0.3414, "step": 5440 }, { "epoch": 0.95, "learning_rate": 4.098432055749129e-05, "loss": 0.3327, "step": 5441 }, { "epoch": 0.95, "learning_rate": 4.102787456445993e-05, "loss": 0.3352, "step": 5442 }, { "epoch": 0.95, "learning_rate": 4.107142857142857e-05, "loss": 0.3476, "step": 5443 }, { "epoch": 0.95, "learning_rate": 4.111498257839721e-05, "loss": 0.3515, "step": 5444 }, { "epoch": 0.95, "learning_rate": 4.1158536585365856e-05, "loss": 0.3652, "step": 5445 }, { "epoch": 0.95, "learning_rate": 4.12020905923345e-05, "loss": 0.3461, "step": 5446 }, { "epoch": 0.95, "learning_rate": 4.124564459930314e-05, "loss": 0.3531, "step": 5447 }, { "epoch": 0.95, "learning_rate": 4.128919860627178e-05, "loss": 0.3438, "step": 5448 }, { "epoch": 0.95, "learning_rate": 4.133275261324042e-05, "loss": 0.3571, "step": 5449 }, { "epoch": 0.95, "learning_rate": 4.137630662020906e-05, "loss": 0.3508, "step": 5450 }, { "epoch": 0.95, "learning_rate": 4.14198606271777e-05, "loss": 0.3528, "step": 5451 }, { "epoch": 0.95, "learning_rate": 4.146341463414634e-05, "loss": 0.3378, "step": 5452 }, { "epoch": 0.95, "learning_rate": 4.150696864111498e-05, "loss": 0.329, "step": 5453 }, { "epoch": 0.95, "learning_rate": 4.1550522648083626e-05, "loss": 0.3364, "step": 5454 }, { "epoch": 0.95, "learning_rate": 4.159407665505227e-05, "loss": 0.3392, "step": 5455 }, { "epoch": 0.95, "learning_rate": 4.163763066202091e-05, "loss": 0.356, "step": 5456 }, { "epoch": 0.95, "learning_rate": 4.168118466898955e-05, "loss": 0.3438, "step": 5457 }, { "epoch": 0.95, "learning_rate": 4.172473867595819e-05, "loss": 0.3351, "step": 5458 }, { "epoch": 0.95, "learning_rate": 4.176829268292683e-05, "loss": 0.3497, "step": 5459 }, { "epoch": 0.95, "learning_rate": 4.181184668989547e-05, "loss": 0.3515, "step": 5460 }, { "epoch": 0.95, "learning_rate": 4.185540069686411e-05, "loss": 0.3339, "step": 5461 }, { "epoch": 0.95, "learning_rate": 4.189895470383275e-05, "loss": 0.3436, "step": 5462 }, { "epoch": 0.95, "learning_rate": 4.1942508710801395e-05, "loss": 0.3458, "step": 5463 }, { "epoch": 0.95, "learning_rate": 4.198606271777004e-05, "loss": 0.3555, "step": 5464 }, { "epoch": 0.95, "learning_rate": 4.202961672473868e-05, "loss": 0.3418, "step": 5465 }, { "epoch": 0.95, "learning_rate": 4.207317073170732e-05, "loss": 0.3495, "step": 5466 }, { "epoch": 0.95, "learning_rate": 4.211672473867596e-05, "loss": 0.3566, "step": 5467 }, { "epoch": 0.95, "learning_rate": 4.21602787456446e-05, "loss": 0.33, "step": 5468 }, { "epoch": 0.95, "learning_rate": 4.220383275261324e-05, "loss": 0.339, "step": 5469 }, { "epoch": 0.95, "learning_rate": 4.2247386759581884e-05, "loss": 0.3234, "step": 5470 }, { "epoch": 0.95, "learning_rate": 4.229094076655053e-05, "loss": 0.3196, "step": 5471 }, { "epoch": 0.95, "learning_rate": 4.233449477351917e-05, "loss": 0.3505, "step": 5472 }, { "epoch": 0.95, "learning_rate": 4.237804878048781e-05, "loss": 0.341, "step": 5473 }, { "epoch": 0.95, "learning_rate": 4.242160278745645e-05, "loss": 0.3593, "step": 5474 }, { "epoch": 0.95, "learning_rate": 4.246515679442509e-05, "loss": 0.3497, "step": 5475 }, { "epoch": 0.95, "learning_rate": 4.250871080139373e-05, "loss": 0.3427, "step": 5476 }, { "epoch": 0.95, "learning_rate": 4.255226480836237e-05, "loss": 0.347, "step": 5477 }, { "epoch": 0.95, "learning_rate": 4.259581881533101e-05, "loss": 0.3339, "step": 5478 }, { "epoch": 0.95, "learning_rate": 4.2639372822299653e-05, "loss": 0.3641, "step": 5479 }, { "epoch": 0.96, "learning_rate": 4.26829268292683e-05, "loss": 0.3311, "step": 5480 }, { "epoch": 0.96, "learning_rate": 4.272648083623694e-05, "loss": 0.3609, "step": 5481 }, { "epoch": 0.96, "learning_rate": 4.2770034843205577e-05, "loss": 0.3494, "step": 5482 }, { "epoch": 0.96, "learning_rate": 4.281358885017422e-05, "loss": 0.3598, "step": 5483 }, { "epoch": 0.96, "learning_rate": 4.2857142857142856e-05, "loss": 0.3378, "step": 5484 }, { "epoch": 0.96, "learning_rate": 4.29006968641115e-05, "loss": 0.346, "step": 5485 }, { "epoch": 0.96, "learning_rate": 4.294425087108014e-05, "loss": 0.3402, "step": 5486 }, { "epoch": 0.96, "learning_rate": 4.298780487804878e-05, "loss": 0.3424, "step": 5487 }, { "epoch": 0.96, "learning_rate": 4.303135888501742e-05, "loss": 0.337, "step": 5488 }, { "epoch": 0.96, "learning_rate": 4.3074912891986066e-05, "loss": 0.3567, "step": 5489 }, { "epoch": 0.96, "learning_rate": 4.311846689895471e-05, "loss": 0.3456, "step": 5490 }, { "epoch": 0.96, "learning_rate": 4.3162020905923346e-05, "loss": 0.344, "step": 5491 }, { "epoch": 0.96, "learning_rate": 4.320557491289199e-05, "loss": 0.3636, "step": 5492 }, { "epoch": 0.96, "learning_rate": 4.324912891986063e-05, "loss": 0.3354, "step": 5493 }, { "epoch": 0.96, "learning_rate": 4.329268292682927e-05, "loss": 0.3472, "step": 5494 }, { "epoch": 0.96, "learning_rate": 4.333623693379791e-05, "loss": 0.345, "step": 5495 }, { "epoch": 0.96, "learning_rate": 4.337979094076655e-05, "loss": 0.3227, "step": 5496 }, { "epoch": 0.96, "learning_rate": 4.342334494773519e-05, "loss": 0.3506, "step": 5497 }, { "epoch": 0.96, "learning_rate": 4.3466898954703835e-05, "loss": 0.3535, "step": 5498 }, { "epoch": 0.96, "learning_rate": 4.351045296167248e-05, "loss": 0.3508, "step": 5499 }, { "epoch": 0.96, "learning_rate": 4.3554006968641115e-05, "loss": 0.3425, "step": 5500 }, { "epoch": 0.96, "learning_rate": 4.359756097560976e-05, "loss": 0.3368, "step": 5501 }, { "epoch": 0.96, "learning_rate": 4.36411149825784e-05, "loss": 0.3449, "step": 5502 }, { "epoch": 0.96, "learning_rate": 4.368466898954704e-05, "loss": 0.3391, "step": 5503 }, { "epoch": 0.96, "learning_rate": 4.372822299651568e-05, "loss": 0.3438, "step": 5504 }, { "epoch": 0.96, "learning_rate": 4.377177700348432e-05, "loss": 0.3376, "step": 5505 }, { "epoch": 0.96, "learning_rate": 4.381533101045297e-05, "loss": 0.3322, "step": 5506 }, { "epoch": 0.96, "learning_rate": 4.3858885017421604e-05, "loss": 0.3464, "step": 5507 }, { "epoch": 0.96, "learning_rate": 4.390243902439025e-05, "loss": 0.3513, "step": 5508 }, { "epoch": 0.96, "learning_rate": 4.394599303135889e-05, "loss": 0.3399, "step": 5509 }, { "epoch": 0.96, "learning_rate": 4.398954703832753e-05, "loss": 0.3334, "step": 5510 }, { "epoch": 0.96, "learning_rate": 4.403310104529617e-05, "loss": 0.328, "step": 5511 }, { "epoch": 0.96, "learning_rate": 4.407665505226481e-05, "loss": 0.34, "step": 5512 }, { "epoch": 0.96, "learning_rate": 4.412020905923345e-05, "loss": 0.3336, "step": 5513 }, { "epoch": 0.96, "learning_rate": 4.4163763066202094e-05, "loss": 0.3489, "step": 5514 }, { "epoch": 0.96, "learning_rate": 4.420731707317074e-05, "loss": 0.3339, "step": 5515 }, { "epoch": 0.96, "learning_rate": 4.4250871080139374e-05, "loss": 0.3427, "step": 5516 }, { "epoch": 0.96, "learning_rate": 4.429442508710802e-05, "loss": 0.3646, "step": 5517 }, { "epoch": 0.96, "learning_rate": 4.433797909407666e-05, "loss": 0.3478, "step": 5518 }, { "epoch": 0.96, "learning_rate": 4.43815331010453e-05, "loss": 0.3495, "step": 5519 }, { "epoch": 0.96, "learning_rate": 4.442508710801394e-05, "loss": 0.353, "step": 5520 }, { "epoch": 0.96, "learning_rate": 4.4468641114982576e-05, "loss": 0.3336, "step": 5521 }, { "epoch": 0.96, "learning_rate": 4.451219512195122e-05, "loss": 0.3384, "step": 5522 }, { "epoch": 0.96, "learning_rate": 4.455574912891986e-05, "loss": 0.3446, "step": 5523 }, { "epoch": 0.96, "learning_rate": 4.4599303135888506e-05, "loss": 0.3316, "step": 5524 }, { "epoch": 0.96, "learning_rate": 4.464285714285715e-05, "loss": 0.3291, "step": 5525 }, { "epoch": 0.96, "learning_rate": 4.4686411149825786e-05, "loss": 0.352, "step": 5526 }, { "epoch": 0.96, "learning_rate": 4.472996515679443e-05, "loss": 0.3458, "step": 5527 }, { "epoch": 0.96, "learning_rate": 4.4773519163763066e-05, "loss": 0.3628, "step": 5528 }, { "epoch": 0.96, "learning_rate": 4.481707317073171e-05, "loss": 0.355, "step": 5529 }, { "epoch": 0.96, "learning_rate": 4.4860627177700346e-05, "loss": 0.3378, "step": 5530 }, { "epoch": 0.96, "learning_rate": 4.490418118466899e-05, "loss": 0.3439, "step": 5531 }, { "epoch": 0.96, "learning_rate": 4.494773519163763e-05, "loss": 0.3378, "step": 5532 }, { "epoch": 0.96, "learning_rate": 4.4991289198606276e-05, "loss": 0.35, "step": 5533 }, { "epoch": 0.96, "learning_rate": 4.503484320557492e-05, "loss": 0.3608, "step": 5534 }, { "epoch": 0.96, "learning_rate": 4.5078397212543555e-05, "loss": 0.3466, "step": 5535 }, { "epoch": 0.96, "learning_rate": 4.51219512195122e-05, "loss": 0.3473, "step": 5536 }, { "epoch": 0.97, "learning_rate": 4.5165505226480835e-05, "loss": 0.3533, "step": 5537 }, { "epoch": 0.97, "learning_rate": 4.520905923344948e-05, "loss": 0.3487, "step": 5538 }, { "epoch": 0.97, "learning_rate": 4.5252613240418115e-05, "loss": 0.3382, "step": 5539 }, { "epoch": 0.97, "learning_rate": 4.529616724738676e-05, "loss": 0.3441, "step": 5540 }, { "epoch": 0.97, "learning_rate": 4.533972125435541e-05, "loss": 0.3514, "step": 5541 }, { "epoch": 0.97, "learning_rate": 4.5383275261324045e-05, "loss": 0.3435, "step": 5542 }, { "epoch": 0.97, "learning_rate": 4.542682926829269e-05, "loss": 0.3453, "step": 5543 }, { "epoch": 0.97, "learning_rate": 4.5470383275261325e-05, "loss": 0.3556, "step": 5544 }, { "epoch": 0.97, "learning_rate": 4.551393728222997e-05, "loss": 0.3393, "step": 5545 }, { "epoch": 0.97, "learning_rate": 4.5557491289198604e-05, "loss": 0.3427, "step": 5546 }, { "epoch": 0.97, "learning_rate": 4.560104529616725e-05, "loss": 0.3618, "step": 5547 }, { "epoch": 0.97, "learning_rate": 4.564459930313589e-05, "loss": 0.3445, "step": 5548 }, { "epoch": 0.97, "learning_rate": 4.5688153310104534e-05, "loss": 0.3343, "step": 5549 }, { "epoch": 0.97, "learning_rate": 4.573170731707318e-05, "loss": 0.3269, "step": 5550 }, { "epoch": 0.97, "learning_rate": 4.5775261324041814e-05, "loss": 0.3292, "step": 5551 }, { "epoch": 0.97, "learning_rate": 4.581881533101046e-05, "loss": 0.3379, "step": 5552 }, { "epoch": 0.97, "learning_rate": 4.5862369337979094e-05, "loss": 0.3501, "step": 5553 }, { "epoch": 0.97, "learning_rate": 4.590592334494774e-05, "loss": 0.3553, "step": 5554 }, { "epoch": 0.97, "learning_rate": 4.594947735191638e-05, "loss": 0.3403, "step": 5555 }, { "epoch": 0.97, "learning_rate": 4.599303135888502e-05, "loss": 0.3375, "step": 5556 }, { "epoch": 0.97, "learning_rate": 4.603658536585366e-05, "loss": 0.3531, "step": 5557 }, { "epoch": 0.97, "learning_rate": 4.6080139372822303e-05, "loss": 0.3439, "step": 5558 }, { "epoch": 0.97, "learning_rate": 4.612369337979095e-05, "loss": 0.3522, "step": 5559 }, { "epoch": 0.97, "learning_rate": 4.616724738675958e-05, "loss": 0.3396, "step": 5560 }, { "epoch": 0.97, "learning_rate": 4.6210801393728226e-05, "loss": 0.3405, "step": 5561 }, { "epoch": 0.97, "learning_rate": 4.625435540069686e-05, "loss": 0.3558, "step": 5562 }, { "epoch": 0.97, "learning_rate": 4.6297909407665506e-05, "loss": 0.3413, "step": 5563 }, { "epoch": 0.97, "learning_rate": 4.634146341463415e-05, "loss": 0.3496, "step": 5564 }, { "epoch": 0.97, "learning_rate": 4.6385017421602786e-05, "loss": 0.3597, "step": 5565 }, { "epoch": 0.97, "learning_rate": 4.642857142857143e-05, "loss": 0.3355, "step": 5566 }, { "epoch": 0.97, "learning_rate": 4.647212543554007e-05, "loss": 0.3441, "step": 5567 }, { "epoch": 0.97, "learning_rate": 4.6515679442508716e-05, "loss": 0.3514, "step": 5568 }, { "epoch": 0.97, "learning_rate": 4.655923344947735e-05, "loss": 0.3474, "step": 5569 }, { "epoch": 0.97, "learning_rate": 4.6602787456445996e-05, "loss": 0.3506, "step": 5570 }, { "epoch": 0.97, "learning_rate": 4.664634146341464e-05, "loss": 0.3532, "step": 5571 }, { "epoch": 0.97, "learning_rate": 4.6689895470383275e-05, "loss": 0.34, "step": 5572 }, { "epoch": 0.97, "learning_rate": 4.673344947735192e-05, "loss": 0.3401, "step": 5573 }, { "epoch": 0.97, "learning_rate": 4.6777003484320555e-05, "loss": 0.3472, "step": 5574 }, { "epoch": 0.97, "learning_rate": 4.6820557491289205e-05, "loss": 0.3362, "step": 5575 }, { "epoch": 0.97, "learning_rate": 4.686411149825784e-05, "loss": 0.3418, "step": 5576 }, { "epoch": 0.97, "learning_rate": 4.6907665505226485e-05, "loss": 0.3451, "step": 5577 }, { "epoch": 0.97, "learning_rate": 4.695121951219512e-05, "loss": 0.3514, "step": 5578 }, { "epoch": 0.97, "learning_rate": 4.6994773519163765e-05, "loss": 0.3331, "step": 5579 }, { "epoch": 0.97, "learning_rate": 4.703832752613241e-05, "loss": 0.3695, "step": 5580 }, { "epoch": 0.97, "learning_rate": 4.7081881533101045e-05, "loss": 0.3316, "step": 5581 }, { "epoch": 0.97, "learning_rate": 4.712543554006969e-05, "loss": 0.3389, "step": 5582 }, { "epoch": 0.97, "learning_rate": 4.716898954703833e-05, "loss": 0.3434, "step": 5583 }, { "epoch": 0.97, "learning_rate": 4.7212543554006975e-05, "loss": 0.3337, "step": 5584 }, { "epoch": 0.97, "learning_rate": 4.725609756097561e-05, "loss": 0.3518, "step": 5585 }, { "epoch": 0.97, "learning_rate": 4.7299651567944254e-05, "loss": 0.3461, "step": 5586 }, { "epoch": 0.97, "learning_rate": 4.73432055749129e-05, "loss": 0.3504, "step": 5587 }, { "epoch": 0.97, "learning_rate": 4.7386759581881534e-05, "loss": 0.3329, "step": 5588 }, { "epoch": 0.97, "learning_rate": 4.743031358885018e-05, "loss": 0.345, "step": 5589 }, { "epoch": 0.97, "learning_rate": 4.7473867595818814e-05, "loss": 0.3275, "step": 5590 }, { "epoch": 0.97, "learning_rate": 4.751742160278746e-05, "loss": 0.3354, "step": 5591 }, { "epoch": 0.97, "learning_rate": 4.75609756097561e-05, "loss": 0.3446, "step": 5592 }, { "epoch": 0.97, "learning_rate": 4.7604529616724744e-05, "loss": 0.3625, "step": 5593 }, { "epoch": 0.98, "learning_rate": 4.764808362369339e-05, "loss": 0.3317, "step": 5594 }, { "epoch": 0.98, "learning_rate": 4.7691637630662024e-05, "loss": 0.3489, "step": 5595 }, { "epoch": 0.98, "learning_rate": 4.773519163763067e-05, "loss": 0.3422, "step": 5596 }, { "epoch": 0.98, "learning_rate": 4.77787456445993e-05, "loss": 0.3295, "step": 5597 }, { "epoch": 0.98, "learning_rate": 4.782229965156795e-05, "loss": 0.3487, "step": 5598 }, { "epoch": 0.98, "learning_rate": 4.786585365853658e-05, "loss": 0.3464, "step": 5599 }, { "epoch": 0.98, "learning_rate": 4.7909407665505226e-05, "loss": 0.3207, "step": 5600 }, { "epoch": 0.98, "learning_rate": 4.795296167247387e-05, "loss": 0.3566, "step": 5601 }, { "epoch": 0.98, "learning_rate": 4.799651567944251e-05, "loss": 0.3592, "step": 5602 }, { "epoch": 0.98, "learning_rate": 4.8040069686411156e-05, "loss": 0.3393, "step": 5603 }, { "epoch": 0.98, "learning_rate": 4.808362369337979e-05, "loss": 0.3364, "step": 5604 }, { "epoch": 0.98, "learning_rate": 4.8127177700348436e-05, "loss": 0.3335, "step": 5605 }, { "epoch": 0.98, "learning_rate": 4.817073170731707e-05, "loss": 0.3376, "step": 5606 }, { "epoch": 0.98, "learning_rate": 4.8214285714285716e-05, "loss": 0.321, "step": 5607 }, { "epoch": 0.98, "learning_rate": 4.825783972125435e-05, "loss": 0.3455, "step": 5608 }, { "epoch": 0.98, "learning_rate": 4.8301393728222996e-05, "loss": 0.3371, "step": 5609 }, { "epoch": 0.98, "learning_rate": 4.8344947735191646e-05, "loss": 0.3454, "step": 5610 }, { "epoch": 0.98, "learning_rate": 4.838850174216028e-05, "loss": 0.3442, "step": 5611 }, { "epoch": 0.98, "learning_rate": 4.8432055749128926e-05, "loss": 0.3445, "step": 5612 }, { "epoch": 0.98, "learning_rate": 4.847560975609756e-05, "loss": 0.3614, "step": 5613 }, { "epoch": 0.98, "learning_rate": 4.8519163763066205e-05, "loss": 0.3483, "step": 5614 }, { "epoch": 0.98, "learning_rate": 4.856271777003484e-05, "loss": 0.3542, "step": 5615 }, { "epoch": 0.98, "learning_rate": 4.8606271777003485e-05, "loss": 0.3503, "step": 5616 }, { "epoch": 0.98, "learning_rate": 4.864982578397213e-05, "loss": 0.3607, "step": 5617 }, { "epoch": 0.98, "learning_rate": 4.869337979094077e-05, "loss": 0.3644, "step": 5618 }, { "epoch": 0.98, "learning_rate": 4.8736933797909415e-05, "loss": 0.3395, "step": 5619 }, { "epoch": 0.98, "learning_rate": 4.878048780487805e-05, "loss": 0.3392, "step": 5620 }, { "epoch": 0.98, "learning_rate": 4.8824041811846695e-05, "loss": 0.3584, "step": 5621 }, { "epoch": 0.98, "learning_rate": 4.886759581881533e-05, "loss": 0.3534, "step": 5622 }, { "epoch": 0.98, "learning_rate": 4.8911149825783975e-05, "loss": 0.3361, "step": 5623 }, { "epoch": 0.98, "learning_rate": 4.895470383275261e-05, "loss": 0.3473, "step": 5624 }, { "epoch": 0.98, "learning_rate": 4.8998257839721254e-05, "loss": 0.3282, "step": 5625 }, { "epoch": 0.98, "learning_rate": 4.90418118466899e-05, "loss": 0.3309, "step": 5626 }, { "epoch": 0.98, "learning_rate": 4.908536585365854e-05, "loss": 0.342, "step": 5627 }, { "epoch": 0.98, "learning_rate": 4.9128919860627184e-05, "loss": 0.342, "step": 5628 }, { "epoch": 0.98, "learning_rate": 4.917247386759582e-05, "loss": 0.3469, "step": 5629 }, { "epoch": 0.98, "learning_rate": 4.9216027874564464e-05, "loss": 0.3322, "step": 5630 }, { "epoch": 0.98, "learning_rate": 4.92595818815331e-05, "loss": 0.348, "step": 5631 }, { "epoch": 0.98, "learning_rate": 4.9303135888501744e-05, "loss": 0.3272, "step": 5632 }, { "epoch": 0.98, "learning_rate": 4.934668989547039e-05, "loss": 0.3451, "step": 5633 }, { "epoch": 0.98, "learning_rate": 4.9390243902439024e-05, "loss": 0.3414, "step": 5634 }, { "epoch": 0.98, "learning_rate": 4.943379790940767e-05, "loss": 0.3502, "step": 5635 }, { "epoch": 0.98, "learning_rate": 4.947735191637631e-05, "loss": 0.3512, "step": 5636 }, { "epoch": 0.98, "learning_rate": 4.952090592334495e-05, "loss": 0.3499, "step": 5637 }, { "epoch": 0.98, "learning_rate": 4.956445993031359e-05, "loss": 0.3527, "step": 5638 }, { "epoch": 0.98, "learning_rate": 4.960801393728223e-05, "loss": 0.3383, "step": 5639 }, { "epoch": 0.98, "learning_rate": 4.965156794425087e-05, "loss": 0.3407, "step": 5640 }, { "epoch": 0.98, "learning_rate": 4.969512195121951e-05, "loss": 0.3389, "step": 5641 }, { "epoch": 0.98, "learning_rate": 4.9738675958188156e-05, "loss": 0.3291, "step": 5642 }, { "epoch": 0.98, "learning_rate": 4.978222996515679e-05, "loss": 0.3504, "step": 5643 }, { "epoch": 0.98, "learning_rate": 4.9825783972125436e-05, "loss": 0.348, "step": 5644 }, { "epoch": 0.98, "learning_rate": 4.986933797909408e-05, "loss": 0.3424, "step": 5645 }, { "epoch": 0.98, "learning_rate": 4.991289198606272e-05, "loss": 0.3407, "step": 5646 }, { "epoch": 0.98, "learning_rate": 4.995644599303136e-05, "loss": 0.35, "step": 5647 }, { "epoch": 0.98, "learning_rate": 5e-05, "loss": 0.3531, "step": 5648 }, { "epoch": 0.98, "learning_rate": 4.9999998958670857e-05, "loss": 0.3277, "step": 5649 }, { "epoch": 0.98, "learning_rate": 4.999999583468353e-05, "loss": 0.3488, "step": 5650 }, { "epoch": 0.98, "learning_rate": 4.999999062803829e-05, "loss": 0.3679, "step": 5651 }, { "epoch": 0.99, "learning_rate": 4.9999983338735634e-05, "loss": 0.3315, "step": 5652 }, { "epoch": 0.99, "learning_rate": 4.9999973966776235e-05, "loss": 0.3344, "step": 5653 }, { "epoch": 0.99, "learning_rate": 4.999996251216096e-05, "loss": 0.3446, "step": 5654 }, { "epoch": 0.99, "learning_rate": 4.999994897489086e-05, "loss": 0.334, "step": 5655 }, { "epoch": 0.99, "learning_rate": 4.999993335496721e-05, "loss": 0.3511, "step": 5656 }, { "epoch": 0.99, "learning_rate": 4.9999915652391433e-05, "loss": 0.3438, "step": 5657 }, { "epoch": 0.99, "learning_rate": 4.9999895867165174e-05, "loss": 0.3407, "step": 5658 }, { "epoch": 0.99, "learning_rate": 4.999987399929027e-05, "loss": 0.3292, "step": 5659 }, { "epoch": 0.99, "learning_rate": 4.999985004876875e-05, "loss": 0.346, "step": 5660 }, { "epoch": 0.99, "learning_rate": 4.999982401560282e-05, "loss": 0.345, "step": 5661 }, { "epoch": 0.99, "learning_rate": 4.999979589979489e-05, "loss": 0.3444, "step": 5662 }, { "epoch": 0.99, "learning_rate": 4.9999765701347556e-05, "loss": 0.3405, "step": 5663 }, { "epoch": 0.99, "learning_rate": 4.9999733420263635e-05, "loss": 0.3539, "step": 5664 }, { "epoch": 0.99, "learning_rate": 4.9999699056546096e-05, "loss": 0.3467, "step": 5665 }, { "epoch": 0.99, "learning_rate": 4.999966261019813e-05, "loss": 0.3392, "step": 5666 }, { "epoch": 0.99, "learning_rate": 4.99996240812231e-05, "loss": 0.3354, "step": 5667 }, { "epoch": 0.99, "learning_rate": 4.9999583469624585e-05, "loss": 0.3443, "step": 5668 }, { "epoch": 0.99, "learning_rate": 4.999954077540634e-05, "loss": 0.3523, "step": 5669 }, { "epoch": 0.99, "learning_rate": 4.999949599857231e-05, "loss": 0.3473, "step": 5670 }, { "epoch": 0.99, "learning_rate": 4.999944913912665e-05, "loss": 0.3424, "step": 5671 }, { "epoch": 0.99, "learning_rate": 4.999940019707369e-05, "loss": 0.3251, "step": 5672 }, { "epoch": 0.99, "learning_rate": 4.9999349172417956e-05, "loss": 0.3464, "step": 5673 }, { "epoch": 0.99, "learning_rate": 4.9999296065164186e-05, "loss": 0.3379, "step": 5674 }, { "epoch": 0.99, "learning_rate": 4.9999240875317285e-05, "loss": 0.3503, "step": 5675 }, { "epoch": 0.99, "learning_rate": 4.999918360288237e-05, "loss": 0.3527, "step": 5676 }, { "epoch": 0.99, "learning_rate": 4.999912424786472e-05, "loss": 0.3246, "step": 5677 }, { "epoch": 0.99, "learning_rate": 4.999906281026985e-05, "loss": 0.3559, "step": 5678 }, { "epoch": 0.99, "learning_rate": 4.999899929010345e-05, "loss": 0.3472, "step": 5679 }, { "epoch": 0.99, "learning_rate": 4.999893368737139e-05, "loss": 0.3447, "step": 5680 }, { "epoch": 0.99, "learning_rate": 4.999886600207974e-05, "loss": 0.3783, "step": 5681 }, { "epoch": 0.99, "learning_rate": 4.999879623423476e-05, "loss": 0.3317, "step": 5682 }, { "epoch": 0.99, "learning_rate": 4.999872438384293e-05, "loss": 0.3793, "step": 5683 }, { "epoch": 0.99, "learning_rate": 4.999865045091089e-05, "loss": 0.3505, "step": 5684 }, { "epoch": 0.99, "learning_rate": 4.999857443544547e-05, "loss": 0.3416, "step": 5685 }, { "epoch": 0.99, "learning_rate": 4.9998496337453726e-05, "loss": 0.3547, "step": 5686 }, { "epoch": 0.99, "learning_rate": 4.999841615694287e-05, "loss": 0.3472, "step": 5687 }, { "epoch": 0.99, "learning_rate": 4.9998333893920336e-05, "loss": 0.3374, "step": 5688 }, { "epoch": 0.99, "learning_rate": 4.999824954839374e-05, "loss": 0.3284, "step": 5689 }, { "epoch": 0.99, "learning_rate": 4.9998163120370874e-05, "loss": 0.3466, "step": 5690 }, { "epoch": 0.99, "learning_rate": 4.999807460985975e-05, "loss": 0.3541, "step": 5691 }, { "epoch": 0.99, "learning_rate": 4.999798401686856e-05, "loss": 0.347, "step": 5692 }, { "epoch": 0.99, "learning_rate": 4.999789134140569e-05, "loss": 0.3595, "step": 5693 }, { "epoch": 0.99, "learning_rate": 4.99977965834797e-05, "loss": 0.328, "step": 5694 }, { "epoch": 0.99, "learning_rate": 4.99976997430994e-05, "loss": 0.3356, "step": 5695 }, { "epoch": 0.99, "learning_rate": 4.999760082027372e-05, "loss": 0.356, "step": 5696 }, { "epoch": 0.99, "learning_rate": 4.999749981501183e-05, "loss": 0.3444, "step": 5697 }, { "epoch": 0.99, "learning_rate": 4.999739672732307e-05, "loss": 0.3643, "step": 5698 }, { "epoch": 0.99, "learning_rate": 4.999729155721699e-05, "loss": 0.3473, "step": 5699 }, { "epoch": 0.99, "learning_rate": 4.999718430470333e-05, "loss": 0.34, "step": 5700 }, { "epoch": 0.99, "learning_rate": 4.999707496979201e-05, "loss": 0.3387, "step": 5701 }, { "epoch": 0.99, "learning_rate": 4.9996963552493145e-05, "loss": 0.3427, "step": 5702 }, { "epoch": 0.99, "learning_rate": 4.9996850052817064e-05, "loss": 0.333, "step": 5703 }, { "epoch": 0.99, "learning_rate": 4.999673447077425e-05, "loss": 0.3657, "step": 5704 }, { "epoch": 0.99, "learning_rate": 4.999661680637543e-05, "loss": 0.3351, "step": 5705 }, { "epoch": 0.99, "learning_rate": 4.999649705963148e-05, "loss": 0.3389, "step": 5706 }, { "epoch": 0.99, "learning_rate": 4.999637523055348e-05, "loss": 0.3548, "step": 5707 }, { "epoch": 0.99, "learning_rate": 4.9996251319152704e-05, "loss": 0.3395, "step": 5708 }, { "epoch": 1.0, "learning_rate": 4.999612532544063e-05, "loss": 0.3602, "step": 5709 }, { "epoch": 1.0, "learning_rate": 4.999599724942893e-05, "loss": 0.3595, "step": 5710 }, { "epoch": 1.0, "learning_rate": 4.999586709112944e-05, "loss": 0.3404, "step": 5711 }, { "epoch": 1.0, "learning_rate": 4.999573485055422e-05, "loss": 0.3269, "step": 5712 }, { "epoch": 1.0, "learning_rate": 4.99956005277155e-05, "loss": 0.3452, "step": 5713 }, { "epoch": 1.0, "learning_rate": 4.9995464122625726e-05, "loss": 0.3477, "step": 5714 }, { "epoch": 1.0, "learning_rate": 4.9995325635297506e-05, "loss": 0.3611, "step": 5715 }, { "epoch": 1.0, "learning_rate": 4.999518506574368e-05, "loss": 0.3374, "step": 5716 }, { "epoch": 1.0, "learning_rate": 4.999504241397725e-05, "loss": 0.3408, "step": 5717 }, { "epoch": 1.0, "learning_rate": 4.9994897680011416e-05, "loss": 0.3481, "step": 5718 }, { "epoch": 1.0, "learning_rate": 4.999475086385958e-05, "loss": 0.3283, "step": 5719 }, { "epoch": 1.0, "learning_rate": 4.9994601965535344e-05, "loss": 0.3349, "step": 5720 }, { "epoch": 1.0, "learning_rate": 4.999445098505247e-05, "loss": 0.3366, "step": 5721 }, { "epoch": 1.0, "learning_rate": 4.9994297922424926e-05, "loss": 0.3464, "step": 5722 }, { "epoch": 1.0, "learning_rate": 4.999414277766691e-05, "loss": 0.3383, "step": 5723 }, { "epoch": 1.0, "learning_rate": 4.999398555079276e-05, "loss": 0.3459, "step": 5724 }, { "epoch": 1.0, "learning_rate": 4.999382624181704e-05, "loss": 0.3161, "step": 5725 }, { "epoch": 1.0, "learning_rate": 4.999366485075448e-05, "loss": 0.3439, "step": 5726 }, { "epoch": 1.0, "learning_rate": 4.999350137762005e-05, "loss": 0.3367, "step": 5727 }, { "epoch": 1.0, "learning_rate": 4.9993335822428846e-05, "loss": 0.3294, "step": 5728 }, { "epoch": 1.0, "learning_rate": 4.999316818519621e-05, "loss": 0.3505, "step": 5729 }, { "epoch": 1.0, "learning_rate": 4.9992998465937674e-05, "loss": 0.3431, "step": 5730 }, { "epoch": 1.0, "learning_rate": 4.999282666466892e-05, "loss": 0.348, "step": 5731 }, { "epoch": 1.0, "learning_rate": 4.999265278140586e-05, "loss": 0.3413, "step": 5732 }, { "epoch": 1.0, "learning_rate": 4.9992476816164594e-05, "loss": 0.3504, "step": 5733 }, { "epoch": 1.0, "learning_rate": 4.9992298768961406e-05, "loss": 0.3323, "step": 5734 }, { "epoch": 1.0, "learning_rate": 4.9992118639812775e-05, "loss": 0.3554, "step": 5735 }, { "epoch": 1.0, "learning_rate": 4.999193642873537e-05, "loss": 0.3407, "step": 5736 }, { "epoch": 1.0, "learning_rate": 4.9991752135746076e-05, "loss": 0.3422, "step": 5737 }, { "epoch": 1.0, "learning_rate": 4.9991565760861936e-05, "loss": 0.3256, "step": 5738 }, { "epoch": 1.0, "learning_rate": 4.9991377304100204e-05, "loss": 0.3114, "step": 5739 }, { "epoch": 1.0, "learning_rate": 4.999118676547832e-05, "loss": 0.2955, "step": 5740 }, { "epoch": 1.0, "learning_rate": 4.9990994145013924e-05, "loss": 0.2987, "step": 5741 }, { "epoch": 1.0, "learning_rate": 4.9990799442724856e-05, "loss": 0.2975, "step": 5742 }, { "epoch": 1.0, "learning_rate": 4.999060265862912e-05, "loss": 0.3025, "step": 5743 }, { "epoch": 1.0, "learning_rate": 4.999040379274494e-05, "loss": 0.3161, "step": 5744 }, { "epoch": 1.0, "learning_rate": 4.999020284509072e-05, "loss": 0.3013, "step": 5745 }, { "epoch": 1.0, "learning_rate": 4.9989999815685076e-05, "loss": 0.2989, "step": 5746 }, { "epoch": 1.0, "learning_rate": 4.998979470454677e-05, "loss": 0.3009, "step": 5747 }, { "epoch": 1.0, "learning_rate": 4.998958751169482e-05, "loss": 0.3104, "step": 5748 }, { "epoch": 1.0, "learning_rate": 4.998937823714838e-05, "loss": 0.3053, "step": 5749 }, { "epoch": 1.0, "learning_rate": 4.9989166880926834e-05, "loss": 0.3369, "step": 5750 }, { "epoch": 1.0, "learning_rate": 4.9988953443049755e-05, "loss": 0.2974, "step": 5751 }, { "epoch": 1.0, "learning_rate": 4.9988737923536874e-05, "loss": 0.301, "step": 5752 }, { "epoch": 1.0, "learning_rate": 4.9988520322408155e-05, "loss": 0.2998, "step": 5753 }, { "epoch": 1.0, "learning_rate": 4.998830063968374e-05, "loss": 0.2951, "step": 5754 }, { "epoch": 1.0, "learning_rate": 4.9988078875383956e-05, "loss": 0.2938, "step": 5755 }, { "epoch": 1.0, "learning_rate": 4.998785502952934e-05, "loss": 0.3129, "step": 5756 }, { "epoch": 1.0, "learning_rate": 4.998762910214061e-05, "loss": 0.3035, "step": 5757 }, { "epoch": 1.0, "learning_rate": 4.998740109323867e-05, "loss": 0.3139, "step": 5758 }, { "epoch": 1.0, "learning_rate": 4.998717100284463e-05, "loss": 0.3094, "step": 5759 }, { "epoch": 1.0, "learning_rate": 4.998693883097979e-05, "loss": 0.2992, "step": 5760 }, { "epoch": 1.0, "learning_rate": 4.9986704577665646e-05, "loss": 0.3089, "step": 5761 }, { "epoch": 1.0, "learning_rate": 4.9986468242923864e-05, "loss": 0.2794, "step": 5762 }, { "epoch": 1.0, "learning_rate": 4.998622982677633e-05, "loss": 0.3018, "step": 5763 }, { "epoch": 1.0, "learning_rate": 4.998598932924511e-05, "loss": 0.316, "step": 5764 }, { "epoch": 1.0, "learning_rate": 4.998574675035248e-05, "loss": 0.3017, "step": 5765 }, { "epoch": 1.01, "learning_rate": 4.9985502090120867e-05, "loss": 0.2999, "step": 5766 }, { "epoch": 1.01, "learning_rate": 4.998525534857294e-05, "loss": 0.3179, "step": 5767 }, { "epoch": 1.01, "learning_rate": 4.9985006525731524e-05, "loss": 0.3115, "step": 5768 }, { "epoch": 1.01, "learning_rate": 4.998475562161966e-05, "loss": 0.3044, "step": 5769 }, { "epoch": 1.01, "learning_rate": 4.998450263626056e-05, "loss": 0.314, "step": 5770 }, { "epoch": 1.01, "learning_rate": 4.998424756967766e-05, "loss": 0.3041, "step": 5771 }, { "epoch": 1.01, "learning_rate": 4.9983990421894555e-05, "loss": 0.3056, "step": 5772 }, { "epoch": 1.01, "learning_rate": 4.998373119293505e-05, "loss": 0.3004, "step": 5773 }, { "epoch": 1.01, "learning_rate": 4.998346988282314e-05, "loss": 0.297, "step": 5774 }, { "epoch": 1.01, "learning_rate": 4.9983206491583014e-05, "loss": 0.3148, "step": 5775 }, { "epoch": 1.01, "learning_rate": 4.998294101923906e-05, "loss": 0.3056, "step": 5776 }, { "epoch": 1.01, "learning_rate": 4.9982673465815826e-05, "loss": 0.3063, "step": 5777 }, { "epoch": 1.01, "learning_rate": 4.9982403831338104e-05, "loss": 0.3175, "step": 5778 }, { "epoch": 1.01, "learning_rate": 4.998213211583084e-05, "loss": 0.3033, "step": 5779 }, { "epoch": 1.01, "learning_rate": 4.998185831931919e-05, "loss": 0.317, "step": 5780 }, { "epoch": 1.01, "learning_rate": 4.998158244182849e-05, "loss": 0.3088, "step": 5781 }, { "epoch": 1.01, "learning_rate": 4.998130448338428e-05, "loss": 0.2992, "step": 5782 }, { "epoch": 1.01, "learning_rate": 4.99810244440123e-05, "loss": 0.2851, "step": 5783 }, { "epoch": 1.01, "learning_rate": 4.998074232373845e-05, "loss": 0.3153, "step": 5784 }, { "epoch": 1.01, "learning_rate": 4.9980458122588855e-05, "loss": 0.295, "step": 5785 }, { "epoch": 1.01, "learning_rate": 4.998017184058982e-05, "loss": 0.3067, "step": 5786 }, { "epoch": 1.01, "learning_rate": 4.9979883477767835e-05, "loss": 0.304, "step": 5787 }, { "epoch": 1.01, "learning_rate": 4.9979593034149616e-05, "loss": 0.2962, "step": 5788 }, { "epoch": 1.01, "learning_rate": 4.9979300509762017e-05, "loss": 0.3147, "step": 5789 }, { "epoch": 1.01, "learning_rate": 4.9979005904632134e-05, "loss": 0.2968, "step": 5790 }, { "epoch": 1.01, "learning_rate": 4.9978709218787236e-05, "loss": 0.3034, "step": 5791 }, { "epoch": 1.01, "learning_rate": 4.997841045225478e-05, "loss": 0.2887, "step": 5792 }, { "epoch": 1.01, "learning_rate": 4.997810960506242e-05, "loss": 0.3011, "step": 5793 }, { "epoch": 1.01, "learning_rate": 4.997780667723801e-05, "loss": 0.303, "step": 5794 }, { "epoch": 1.01, "learning_rate": 4.997750166880958e-05, "loss": 0.3008, "step": 5795 }, { "epoch": 1.01, "learning_rate": 4.997719457980537e-05, "loss": 0.2944, "step": 5796 }, { "epoch": 1.01, "learning_rate": 4.99768854102538e-05, "loss": 0.3052, "step": 5797 }, { "epoch": 1.01, "learning_rate": 4.9976574160183486e-05, "loss": 0.2994, "step": 5798 }, { "epoch": 1.01, "learning_rate": 4.9976260829623244e-05, "loss": 0.2951, "step": 5799 }, { "epoch": 1.01, "learning_rate": 4.997594541860207e-05, "loss": 0.3019, "step": 5800 }, { "epoch": 1.01, "learning_rate": 4.9975627927149176e-05, "loss": 0.2972, "step": 5801 }, { "epoch": 1.01, "learning_rate": 4.9975308355293924e-05, "loss": 0.3011, "step": 5802 }, { "epoch": 1.01, "learning_rate": 4.997498670306591e-05, "loss": 0.3095, "step": 5803 }, { "epoch": 1.01, "learning_rate": 4.9974662970494915e-05, "loss": 0.3173, "step": 5804 }, { "epoch": 1.01, "learning_rate": 4.9974337157610885e-05, "loss": 0.3088, "step": 5805 }, { "epoch": 1.01, "learning_rate": 4.997400926444399e-05, "loss": 0.317, "step": 5806 }, { "epoch": 1.01, "learning_rate": 4.997367929102458e-05, "loss": 0.3142, "step": 5807 }, { "epoch": 1.01, "learning_rate": 4.997334723738319e-05, "loss": 0.3076, "step": 5808 }, { "epoch": 1.01, "learning_rate": 4.9973013103550564e-05, "loss": 0.3149, "step": 5809 }, { "epoch": 1.01, "learning_rate": 4.997267688955763e-05, "loss": 0.3032, "step": 5810 }, { "epoch": 1.01, "learning_rate": 4.9972338595435505e-05, "loss": 0.3164, "step": 5811 }, { "epoch": 1.01, "learning_rate": 4.997199822121551e-05, "loss": 0.2968, "step": 5812 }, { "epoch": 1.01, "learning_rate": 4.9971655766929134e-05, "loss": 0.3007, "step": 5813 }, { "epoch": 1.01, "learning_rate": 4.99713112326081e-05, "loss": 0.3151, "step": 5814 }, { "epoch": 1.01, "learning_rate": 4.9970964618284274e-05, "loss": 0.3049, "step": 5815 }, { "epoch": 1.01, "learning_rate": 4.997061592398975e-05, "loss": 0.2915, "step": 5816 }, { "epoch": 1.01, "learning_rate": 4.997026514975682e-05, "loss": 0.3077, "step": 5817 }, { "epoch": 1.01, "learning_rate": 4.9969912295617924e-05, "loss": 0.3143, "step": 5818 }, { "epoch": 1.01, "learning_rate": 4.9969557361605744e-05, "loss": 0.307, "step": 5819 }, { "epoch": 1.01, "learning_rate": 4.9969200347753126e-05, "loss": 0.3121, "step": 5820 }, { "epoch": 1.01, "learning_rate": 4.996884125409312e-05, "loss": 0.2969, "step": 5821 }, { "epoch": 1.01, "learning_rate": 4.9968480080658956e-05, "loss": 0.2813, "step": 5822 }, { "epoch": 1.01, "learning_rate": 4.996811682748407e-05, "loss": 0.3061, "step": 5823 }, { "epoch": 1.02, "learning_rate": 4.996775149460209e-05, "loss": 0.2871, "step": 5824 }, { "epoch": 1.02, "learning_rate": 4.9967384082046825e-05, "loss": 0.3149, "step": 5825 }, { "epoch": 1.02, "learning_rate": 4.996701458985229e-05, "loss": 0.2944, "step": 5826 }, { "epoch": 1.02, "learning_rate": 4.9966643018052686e-05, "loss": 0.2837, "step": 5827 }, { "epoch": 1.02, "learning_rate": 4.99662693666824e-05, "loss": 0.2982, "step": 5828 }, { "epoch": 1.02, "learning_rate": 4.9965893635776015e-05, "loss": 0.3003, "step": 5829 }, { "epoch": 1.02, "learning_rate": 4.996551582536832e-05, "loss": 0.3112, "step": 5830 }, { "epoch": 1.02, "learning_rate": 4.9965135935494286e-05, "loss": 0.3069, "step": 5831 }, { "epoch": 1.02, "learning_rate": 4.9964753966189075e-05, "loss": 0.3031, "step": 5832 }, { "epoch": 1.02, "learning_rate": 4.996436991748804e-05, "loss": 0.3123, "step": 5833 }, { "epoch": 1.02, "learning_rate": 4.996398378942673e-05, "loss": 0.3007, "step": 5834 }, { "epoch": 1.02, "learning_rate": 4.996359558204088e-05, "loss": 0.2971, "step": 5835 }, { "epoch": 1.02, "learning_rate": 4.996320529536644e-05, "loss": 0.3081, "step": 5836 }, { "epoch": 1.02, "learning_rate": 4.996281292943953e-05, "loss": 0.3141, "step": 5837 }, { "epoch": 1.02, "learning_rate": 4.996241848429645e-05, "loss": 0.3164, "step": 5838 }, { "epoch": 1.02, "learning_rate": 4.9962021959973736e-05, "loss": 0.3106, "step": 5839 }, { "epoch": 1.02, "learning_rate": 4.996162335650808e-05, "loss": 0.3075, "step": 5840 }, { "epoch": 1.02, "learning_rate": 4.9961222673936376e-05, "loss": 0.3052, "step": 5841 }, { "epoch": 1.02, "learning_rate": 4.996081991229572e-05, "loss": 0.3044, "step": 5842 }, { "epoch": 1.02, "learning_rate": 4.996041507162338e-05, "loss": 0.3016, "step": 5843 }, { "epoch": 1.02, "learning_rate": 4.996000815195684e-05, "loss": 0.3084, "step": 5844 }, { "epoch": 1.02, "learning_rate": 4.995959915333376e-05, "loss": 0.3019, "step": 5845 }, { "epoch": 1.02, "learning_rate": 4.9959188075792e-05, "loss": 0.3049, "step": 5846 }, { "epoch": 1.02, "learning_rate": 4.9958774919369624e-05, "loss": 0.2984, "step": 5847 }, { "epoch": 1.02, "learning_rate": 4.9958359684104846e-05, "loss": 0.3061, "step": 5848 }, { "epoch": 1.02, "learning_rate": 4.9957942370036114e-05, "loss": 0.3036, "step": 5849 }, { "epoch": 1.02, "learning_rate": 4.995752297720207e-05, "loss": 0.3064, "step": 5850 }, { "epoch": 1.02, "learning_rate": 4.995710150564152e-05, "loss": 0.3087, "step": 5851 }, { "epoch": 1.02, "learning_rate": 4.995667795539347e-05, "loss": 0.3141, "step": 5852 }, { "epoch": 1.02, "learning_rate": 4.995625232649714e-05, "loss": 0.3066, "step": 5853 }, { "epoch": 1.02, "learning_rate": 4.9955824618991924e-05, "loss": 0.3066, "step": 5854 }, { "epoch": 1.02, "learning_rate": 4.9955394832917404e-05, "loss": 0.3121, "step": 5855 }, { "epoch": 1.02, "learning_rate": 4.995496296831337e-05, "loss": 0.2943, "step": 5856 }, { "epoch": 1.02, "learning_rate": 4.995452902521979e-05, "loss": 0.3195, "step": 5857 }, { "epoch": 1.02, "learning_rate": 4.995409300367684e-05, "loss": 0.3088, "step": 5858 }, { "epoch": 1.02, "learning_rate": 4.9953654903724866e-05, "loss": 0.2946, "step": 5859 }, { "epoch": 1.02, "learning_rate": 4.995321472540444e-05, "loss": 0.2987, "step": 5860 }, { "epoch": 1.02, "learning_rate": 4.995277246875628e-05, "loss": 0.3133, "step": 5861 }, { "epoch": 1.02, "learning_rate": 4.995232813382134e-05, "loss": 0.2958, "step": 5862 }, { "epoch": 1.02, "learning_rate": 4.995188172064075e-05, "loss": 0.3002, "step": 5863 }, { "epoch": 1.02, "learning_rate": 4.9951433229255824e-05, "loss": 0.3014, "step": 5864 }, { "epoch": 1.02, "learning_rate": 4.995098265970807e-05, "loss": 0.299, "step": 5865 }, { "epoch": 1.02, "learning_rate": 4.99505300120392e-05, "loss": 0.3025, "step": 5866 }, { "epoch": 1.02, "learning_rate": 4.9950075286291125e-05, "loss": 0.3159, "step": 5867 }, { "epoch": 1.02, "learning_rate": 4.994961848250592e-05, "loss": 0.3089, "step": 5868 }, { "epoch": 1.02, "learning_rate": 4.9949159600725863e-05, "loss": 0.3096, "step": 5869 }, { "epoch": 1.02, "learning_rate": 4.994869864099345e-05, "loss": 0.3021, "step": 5870 }, { "epoch": 1.02, "learning_rate": 4.9948235603351326e-05, "loss": 0.3156, "step": 5871 }, { "epoch": 1.02, "learning_rate": 4.9947770487842366e-05, "loss": 0.3005, "step": 5872 }, { "epoch": 1.02, "learning_rate": 4.9947303294509626e-05, "loss": 0.3168, "step": 5873 }, { "epoch": 1.02, "learning_rate": 4.994683402339633e-05, "loss": 0.2993, "step": 5874 }, { "epoch": 1.02, "learning_rate": 4.994636267454593e-05, "loss": 0.2935, "step": 5875 }, { "epoch": 1.02, "learning_rate": 4.994588924800206e-05, "loss": 0.2929, "step": 5876 }, { "epoch": 1.02, "learning_rate": 4.994541374380853e-05, "loss": 0.2976, "step": 5877 }, { "epoch": 1.02, "learning_rate": 4.9944936162009365e-05, "loss": 0.3027, "step": 5878 }, { "epoch": 1.02, "learning_rate": 4.9944456502648754e-05, "loss": 0.307, "step": 5879 }, { "epoch": 1.02, "learning_rate": 4.994397476577111e-05, "loss": 0.3067, "step": 5880 }, { "epoch": 1.03, "learning_rate": 4.994349095142102e-05, "loss": 0.3017, "step": 5881 }, { "epoch": 1.03, "learning_rate": 4.994300505964327e-05, "loss": 0.3051, "step": 5882 }, { "epoch": 1.03, "learning_rate": 4.994251709048283e-05, "loss": 0.3042, "step": 5883 }, { "epoch": 1.03, "learning_rate": 4.9942027043984874e-05, "loss": 0.3058, "step": 5884 }, { "epoch": 1.03, "learning_rate": 4.9941534920194757e-05, "loss": 0.3079, "step": 5885 }, { "epoch": 1.03, "learning_rate": 4.994104071915803e-05, "loss": 0.3032, "step": 5886 }, { "epoch": 1.03, "learning_rate": 4.9940544440920434e-05, "loss": 0.3065, "step": 5887 }, { "epoch": 1.03, "learning_rate": 4.9940046085527924e-05, "loss": 0.3191, "step": 5888 }, { "epoch": 1.03, "learning_rate": 4.9939545653026605e-05, "loss": 0.2969, "step": 5889 }, { "epoch": 1.03, "learning_rate": 4.993904314346282e-05, "loss": 0.2876, "step": 5890 }, { "epoch": 1.03, "learning_rate": 4.9938538556883065e-05, "loss": 0.305, "step": 5891 }, { "epoch": 1.03, "learning_rate": 4.993803189333406e-05, "loss": 0.318, "step": 5892 }, { "epoch": 1.03, "learning_rate": 4.993752315286269e-05, "loss": 0.3076, "step": 5893 }, { "epoch": 1.03, "learning_rate": 4.993701233551606e-05, "loss": 0.3127, "step": 5894 }, { "epoch": 1.03, "learning_rate": 4.993649944134144e-05, "loss": 0.3041, "step": 5895 }, { "epoch": 1.03, "learning_rate": 4.9935984470386304e-05, "loss": 0.3052, "step": 5896 }, { "epoch": 1.03, "learning_rate": 4.993546742269833e-05, "loss": 0.3062, "step": 5897 }, { "epoch": 1.03, "learning_rate": 4.993494829832537e-05, "loss": 0.3057, "step": 5898 }, { "epoch": 1.03, "learning_rate": 4.993442709731548e-05, "loss": 0.3057, "step": 5899 }, { "epoch": 1.03, "learning_rate": 4.99339038197169e-05, "loss": 0.3105, "step": 5900 }, { "epoch": 1.03, "learning_rate": 4.993337846557806e-05, "loss": 0.2897, "step": 5901 }, { "epoch": 1.03, "learning_rate": 4.99328510349476e-05, "loss": 0.311, "step": 5902 }, { "epoch": 1.03, "learning_rate": 4.993232152787434e-05, "loss": 0.3116, "step": 5903 }, { "epoch": 1.03, "learning_rate": 4.9931789944407284e-05, "loss": 0.2986, "step": 5904 }, { "epoch": 1.03, "learning_rate": 4.993125628459564e-05, "loss": 0.3028, "step": 5905 }, { "epoch": 1.03, "learning_rate": 4.99307205484888e-05, "loss": 0.3049, "step": 5906 }, { "epoch": 1.03, "learning_rate": 4.993018273613637e-05, "loss": 0.286, "step": 5907 }, { "epoch": 1.03, "learning_rate": 4.992964284758812e-05, "loss": 0.2994, "step": 5908 }, { "epoch": 1.03, "learning_rate": 4.992910088289401e-05, "loss": 0.3083, "step": 5909 }, { "epoch": 1.03, "learning_rate": 4.9928556842104226e-05, "loss": 0.3006, "step": 5910 }, { "epoch": 1.03, "learning_rate": 4.992801072526912e-05, "loss": 0.3231, "step": 5911 }, { "epoch": 1.03, "learning_rate": 4.992746253243925e-05, "loss": 0.313, "step": 5912 }, { "epoch": 1.03, "learning_rate": 4.9926912263665335e-05, "loss": 0.2867, "step": 5913 }, { "epoch": 1.03, "learning_rate": 4.992635991899832e-05, "loss": 0.3127, "step": 5914 }, { "epoch": 1.03, "learning_rate": 4.992580549848935e-05, "loss": 0.295, "step": 5915 }, { "epoch": 1.03, "learning_rate": 4.992524900218972e-05, "loss": 0.3106, "step": 5916 }, { "epoch": 1.03, "learning_rate": 4.992469043015094e-05, "loss": 0.2923, "step": 5917 }, { "epoch": 1.03, "learning_rate": 4.992412978242474e-05, "loss": 0.3009, "step": 5918 }, { "epoch": 1.03, "learning_rate": 4.9923567059062985e-05, "loss": 0.2911, "step": 5919 }, { "epoch": 1.03, "learning_rate": 4.992300226011778e-05, "loss": 0.2959, "step": 5920 }, { "epoch": 1.03, "learning_rate": 4.9922435385641406e-05, "loss": 0.2954, "step": 5921 }, { "epoch": 1.03, "learning_rate": 4.992186643568631e-05, "loss": 0.3115, "step": 5922 }, { "epoch": 1.03, "learning_rate": 4.992129541030518e-05, "loss": 0.3145, "step": 5923 }, { "epoch": 1.03, "learning_rate": 4.992072230955086e-05, "loss": 0.3054, "step": 5924 }, { "epoch": 1.03, "learning_rate": 4.9920147133476406e-05, "loss": 0.3034, "step": 5925 }, { "epoch": 1.03, "learning_rate": 4.9919569882135054e-05, "loss": 0.3144, "step": 5926 }, { "epoch": 1.03, "learning_rate": 4.9918990555580234e-05, "loss": 0.2904, "step": 5927 }, { "epoch": 1.03, "learning_rate": 4.991840915386557e-05, "loss": 0.2966, "step": 5928 }, { "epoch": 1.03, "learning_rate": 4.991782567704488e-05, "loss": 0.3079, "step": 5929 }, { "epoch": 1.03, "learning_rate": 4.991724012517217e-05, "loss": 0.3018, "step": 5930 }, { "epoch": 1.03, "learning_rate": 4.991665249830164e-05, "loss": 0.3059, "step": 5931 }, { "epoch": 1.03, "learning_rate": 4.991606279648769e-05, "loss": 0.3103, "step": 5932 }, { "epoch": 1.03, "learning_rate": 4.9915471019784895e-05, "loss": 0.3126, "step": 5933 }, { "epoch": 1.03, "learning_rate": 4.9914877168248034e-05, "loss": 0.3042, "step": 5934 }, { "epoch": 1.03, "learning_rate": 4.991428124193207e-05, "loss": 0.2916, "step": 5935 }, { "epoch": 1.03, "learning_rate": 4.9913683240892175e-05, "loss": 0.3075, "step": 5936 }, { "epoch": 1.03, "learning_rate": 4.99130831651837e-05, "loss": 0.3066, "step": 5937 }, { "epoch": 1.03, "learning_rate": 4.9912481014862184e-05, "loss": 0.3155, "step": 5938 }, { "epoch": 1.04, "learning_rate": 4.991187678998336e-05, "loss": 0.3067, "step": 5939 }, { "epoch": 1.04, "learning_rate": 4.9911270490603166e-05, "loss": 0.3075, "step": 5940 }, { "epoch": 1.04, "learning_rate": 4.991066211677772e-05, "loss": 0.3032, "step": 5941 }, { "epoch": 1.04, "learning_rate": 4.991005166856332e-05, "loss": 0.3017, "step": 5942 }, { "epoch": 1.04, "learning_rate": 4.990943914601649e-05, "loss": 0.3026, "step": 5943 }, { "epoch": 1.04, "learning_rate": 4.9908824549193935e-05, "loss": 0.3091, "step": 5944 }, { "epoch": 1.04, "learning_rate": 4.9908207878152514e-05, "loss": 0.3216, "step": 5945 }, { "epoch": 1.04, "learning_rate": 4.990758913294933e-05, "loss": 0.2876, "step": 5946 }, { "epoch": 1.04, "learning_rate": 4.9906968313641644e-05, "loss": 0.3024, "step": 5947 }, { "epoch": 1.04, "learning_rate": 4.990634542028693e-05, "loss": 0.3042, "step": 5948 }, { "epoch": 1.04, "learning_rate": 4.990572045294283e-05, "loss": 0.3002, "step": 5949 }, { "epoch": 1.04, "learning_rate": 4.9905093411667206e-05, "loss": 0.3222, "step": 5950 }, { "epoch": 1.04, "learning_rate": 4.99044642965181e-05, "loss": 0.3065, "step": 5951 }, { "epoch": 1.04, "learning_rate": 4.9903833107553735e-05, "loss": 0.3045, "step": 5952 }, { "epoch": 1.04, "learning_rate": 4.9903199844832536e-05, "loss": 0.2946, "step": 5953 }, { "epoch": 1.04, "learning_rate": 4.9902564508413136e-05, "loss": 0.3153, "step": 5954 }, { "epoch": 1.04, "learning_rate": 4.990192709835432e-05, "loss": 0.3108, "step": 5955 }, { "epoch": 1.04, "learning_rate": 4.99012876147151e-05, "loss": 0.2946, "step": 5956 }, { "epoch": 1.04, "learning_rate": 4.9900646057554676e-05, "loss": 0.3213, "step": 5957 }, { "epoch": 1.04, "learning_rate": 4.9900002426932414e-05, "loss": 0.3065, "step": 5958 }, { "epoch": 1.04, "learning_rate": 4.98993567229079e-05, "loss": 0.2967, "step": 5959 }, { "epoch": 1.04, "learning_rate": 4.98987089455409e-05, "loss": 0.2982, "step": 5960 }, { "epoch": 1.04, "learning_rate": 4.9898059094891384e-05, "loss": 0.3012, "step": 5961 }, { "epoch": 1.04, "learning_rate": 4.989740717101949e-05, "loss": 0.3181, "step": 5962 }, { "epoch": 1.04, "learning_rate": 4.989675317398556e-05, "loss": 0.3037, "step": 5963 }, { "epoch": 1.04, "learning_rate": 4.9896097103850146e-05, "loss": 0.3016, "step": 5964 }, { "epoch": 1.04, "learning_rate": 4.9895438960673965e-05, "loss": 0.3027, "step": 5965 }, { "epoch": 1.04, "learning_rate": 4.989477874451794e-05, "loss": 0.3144, "step": 5966 }, { "epoch": 1.04, "learning_rate": 4.9894116455443176e-05, "loss": 0.3256, "step": 5967 }, { "epoch": 1.04, "learning_rate": 4.9893452093510975e-05, "loss": 0.3072, "step": 5968 }, { "epoch": 1.04, "learning_rate": 4.9892785658782846e-05, "loss": 0.2967, "step": 5969 }, { "epoch": 1.04, "learning_rate": 4.989211715132047e-05, "loss": 0.3037, "step": 5970 }, { "epoch": 1.04, "learning_rate": 4.989144657118572e-05, "loss": 0.3005, "step": 5971 }, { "epoch": 1.04, "learning_rate": 4.9890773918440664e-05, "loss": 0.3048, "step": 5972 }, { "epoch": 1.04, "learning_rate": 4.989009919314758e-05, "loss": 0.3001, "step": 5973 }, { "epoch": 1.04, "learning_rate": 4.98894223953689e-05, "loss": 0.3211, "step": 5974 }, { "epoch": 1.04, "learning_rate": 4.988874352516729e-05, "loss": 0.3067, "step": 5975 }, { "epoch": 1.04, "learning_rate": 4.988806258260558e-05, "loss": 0.3003, "step": 5976 }, { "epoch": 1.04, "learning_rate": 4.98873795677468e-05, "loss": 0.3122, "step": 5977 }, { "epoch": 1.04, "learning_rate": 4.9886694480654174e-05, "loss": 0.29, "step": 5978 }, { "epoch": 1.04, "learning_rate": 4.988600732139111e-05, "loss": 0.3043, "step": 5979 }, { "epoch": 1.04, "learning_rate": 4.988531809002122e-05, "loss": 0.302, "step": 5980 }, { "epoch": 1.04, "learning_rate": 4.98846267866083e-05, "loss": 0.3042, "step": 5981 }, { "epoch": 1.04, "learning_rate": 4.9883933411216324e-05, "loss": 0.3179, "step": 5982 }, { "epoch": 1.04, "learning_rate": 4.98832379639095e-05, "loss": 0.316, "step": 5983 }, { "epoch": 1.04, "learning_rate": 4.988254044475218e-05, "loss": 0.2957, "step": 5984 }, { "epoch": 1.04, "learning_rate": 4.988184085380893e-05, "loss": 0.3025, "step": 5985 }, { "epoch": 1.04, "learning_rate": 4.988113919114451e-05, "loss": 0.3034, "step": 5986 }, { "epoch": 1.04, "learning_rate": 4.9880435456823876e-05, "loss": 0.2802, "step": 5987 }, { "epoch": 1.04, "learning_rate": 4.9879729650912154e-05, "loss": 0.2921, "step": 5988 }, { "epoch": 1.04, "learning_rate": 4.987902177347468e-05, "loss": 0.3094, "step": 5989 }, { "epoch": 1.04, "learning_rate": 4.987831182457698e-05, "loss": 0.3074, "step": 5990 }, { "epoch": 1.04, "learning_rate": 4.9877599804284755e-05, "loss": 0.3006, "step": 5991 }, { "epoch": 1.04, "learning_rate": 4.987688571266393e-05, "loss": 0.2928, "step": 5992 }, { "epoch": 1.04, "learning_rate": 4.98761695497806e-05, "loss": 0.3049, "step": 5993 }, { "epoch": 1.04, "learning_rate": 4.987545131570105e-05, "loss": 0.2877, "step": 5994 }, { "epoch": 1.04, "learning_rate": 4.9874731010491753e-05, "loss": 0.3018, "step": 5995 }, { "epoch": 1.05, "learning_rate": 4.98740086342194e-05, "loss": 0.3034, "step": 5996 }, { "epoch": 1.05, "learning_rate": 4.987328418695084e-05, "loss": 0.3144, "step": 5997 }, { "epoch": 1.05, "learning_rate": 4.987255766875315e-05, "loss": 0.3227, "step": 5998 }, { "epoch": 1.05, "learning_rate": 4.9871829079693554e-05, "loss": 0.3084, "step": 5999 }, { "epoch": 1.05, "learning_rate": 4.987109841983951e-05, "loss": 0.2995, "step": 6000 }, { "epoch": 1.05, "learning_rate": 4.9870365689258635e-05, "loss": 0.2886, "step": 6001 }, { "epoch": 1.05, "learning_rate": 4.986963088801876e-05, "loss": 0.326, "step": 6002 }, { "epoch": 1.05, "learning_rate": 4.986889401618791e-05, "loss": 0.2994, "step": 6003 }, { "epoch": 1.05, "learning_rate": 4.9868155073834285e-05, "loss": 0.3108, "step": 6004 }, { "epoch": 1.05, "learning_rate": 4.9867414061026265e-05, "loss": 0.2979, "step": 6005 }, { "epoch": 1.05, "learning_rate": 4.986667097783247e-05, "loss": 0.3164, "step": 6006 }, { "epoch": 1.05, "learning_rate": 4.986592582432166e-05, "loss": 0.2931, "step": 6007 }, { "epoch": 1.05, "learning_rate": 4.986517860056282e-05, "loss": 0.3095, "step": 6008 }, { "epoch": 1.05, "learning_rate": 4.986442930662511e-05, "loss": 0.3223, "step": 6009 }, { "epoch": 1.05, "learning_rate": 4.986367794257789e-05, "loss": 0.3146, "step": 6010 }, { "epoch": 1.05, "learning_rate": 4.98629245084907e-05, "loss": 0.3178, "step": 6011 }, { "epoch": 1.05, "learning_rate": 4.9862169004433296e-05, "loss": 0.3024, "step": 6012 }, { "epoch": 1.05, "learning_rate": 4.986141143047559e-05, "loss": 0.2941, "step": 6013 }, { "epoch": 1.05, "learning_rate": 4.986065178668772e-05, "loss": 0.3152, "step": 6014 }, { "epoch": 1.05, "learning_rate": 4.985989007313999e-05, "loss": 0.292, "step": 6015 }, { "epoch": 1.05, "learning_rate": 4.985912628990291e-05, "loss": 0.3035, "step": 6016 }, { "epoch": 1.05, "learning_rate": 4.985836043704718e-05, "loss": 0.3039, "step": 6017 }, { "epoch": 1.05, "learning_rate": 4.985759251464369e-05, "loss": 0.314, "step": 6018 }, { "epoch": 1.05, "learning_rate": 4.985682252276352e-05, "loss": 0.3033, "step": 6019 }, { "epoch": 1.05, "learning_rate": 4.985605046147794e-05, "loss": 0.3106, "step": 6020 }, { "epoch": 1.05, "learning_rate": 4.985527633085841e-05, "loss": 0.3134, "step": 6021 }, { "epoch": 1.05, "learning_rate": 4.98545001309766e-05, "loss": 0.3114, "step": 6022 }, { "epoch": 1.05, "learning_rate": 4.985372186190434e-05, "loss": 0.3176, "step": 6023 }, { "epoch": 1.05, "learning_rate": 4.9852941523713685e-05, "loss": 0.3174, "step": 6024 }, { "epoch": 1.05, "learning_rate": 4.985215911647685e-05, "loss": 0.2946, "step": 6025 }, { "epoch": 1.05, "learning_rate": 4.9851374640266266e-05, "loss": 0.3074, "step": 6026 }, { "epoch": 1.05, "learning_rate": 4.9850588095154543e-05, "loss": 0.3194, "step": 6027 }, { "epoch": 1.05, "learning_rate": 4.984979948121448e-05, "loss": 0.32, "step": 6028 }, { "epoch": 1.05, "learning_rate": 4.9849008798519086e-05, "loss": 0.2855, "step": 6029 }, { "epoch": 1.05, "learning_rate": 4.984821604714154e-05, "loss": 0.3025, "step": 6030 }, { "epoch": 1.05, "learning_rate": 4.9847421227155226e-05, "loss": 0.3057, "step": 6031 }, { "epoch": 1.05, "learning_rate": 4.9846624338633707e-05, "loss": 0.3132, "step": 6032 }, { "epoch": 1.05, "learning_rate": 4.984582538165075e-05, "loss": 0.3006, "step": 6033 }, { "epoch": 1.05, "learning_rate": 4.9845024356280306e-05, "loss": 0.2967, "step": 6034 }, { "epoch": 1.05, "learning_rate": 4.9844221262596535e-05, "loss": 0.3039, "step": 6035 }, { "epoch": 1.05, "learning_rate": 4.984341610067375e-05, "loss": 0.289, "step": 6036 }, { "epoch": 1.05, "learning_rate": 4.9842608870586485e-05, "loss": 0.2969, "step": 6037 }, { "epoch": 1.05, "learning_rate": 4.984179957240947e-05, "loss": 0.304, "step": 6038 }, { "epoch": 1.05, "learning_rate": 4.984098820621762e-05, "loss": 0.297, "step": 6039 }, { "epoch": 1.05, "learning_rate": 4.9840174772086015e-05, "loss": 0.3118, "step": 6040 }, { "epoch": 1.05, "learning_rate": 4.983935927008995e-05, "loss": 0.3082, "step": 6041 }, { "epoch": 1.05, "learning_rate": 4.983854170030494e-05, "loss": 0.3093, "step": 6042 }, { "epoch": 1.05, "learning_rate": 4.983772206280663e-05, "loss": 0.3033, "step": 6043 }, { "epoch": 1.05, "learning_rate": 4.983690035767091e-05, "loss": 0.3094, "step": 6044 }, { "epoch": 1.05, "learning_rate": 4.983607658497381e-05, "loss": 0.3181, "step": 6045 }, { "epoch": 1.05, "learning_rate": 4.9835250744791615e-05, "loss": 0.2981, "step": 6046 }, { "epoch": 1.05, "learning_rate": 4.983442283720074e-05, "loss": 0.3112, "step": 6047 }, { "epoch": 1.05, "learning_rate": 4.983359286227784e-05, "loss": 0.3107, "step": 6048 }, { "epoch": 1.05, "learning_rate": 4.9832760820099724e-05, "loss": 0.3108, "step": 6049 }, { "epoch": 1.05, "learning_rate": 4.983192671074341e-05, "loss": 0.2995, "step": 6050 }, { "epoch": 1.05, "learning_rate": 4.9831090534286114e-05, "loss": 0.2924, "step": 6051 }, { "epoch": 1.05, "learning_rate": 4.983025229080523e-05, "loss": 0.2955, "step": 6052 }, { "epoch": 1.06, "learning_rate": 4.982941198037834e-05, "loss": 0.311, "step": 6053 }, { "epoch": 1.06, "learning_rate": 4.982856960308324e-05, "loss": 0.3011, "step": 6054 }, { "epoch": 1.06, "learning_rate": 4.982772515899789e-05, "loss": 0.3044, "step": 6055 }, { "epoch": 1.06, "learning_rate": 4.982687864820045e-05, "loss": 0.313, "step": 6056 }, { "epoch": 1.06, "learning_rate": 4.98260300707693e-05, "loss": 0.2979, "step": 6057 }, { "epoch": 1.06, "learning_rate": 4.982517942678296e-05, "loss": 0.3101, "step": 6058 }, { "epoch": 1.06, "learning_rate": 4.982432671632017e-05, "loss": 0.3034, "step": 6059 }, { "epoch": 1.06, "learning_rate": 4.9823471939459884e-05, "loss": 0.3073, "step": 6060 }, { "epoch": 1.06, "learning_rate": 4.9822615096281206e-05, "loss": 0.2826, "step": 6061 }, { "epoch": 1.06, "learning_rate": 4.9821756186863435e-05, "loss": 0.2927, "step": 6062 }, { "epoch": 1.06, "learning_rate": 4.9820895211286096e-05, "loss": 0.3021, "step": 6063 }, { "epoch": 1.06, "learning_rate": 4.9820032169628866e-05, "loss": 0.3129, "step": 6064 }, { "epoch": 1.06, "learning_rate": 4.981916706197164e-05, "loss": 0.328, "step": 6065 }, { "epoch": 1.06, "learning_rate": 4.9818299888394495e-05, "loss": 0.2847, "step": 6066 }, { "epoch": 1.06, "learning_rate": 4.98174306489777e-05, "loss": 0.3237, "step": 6067 }, { "epoch": 1.06, "learning_rate": 4.98165593438017e-05, "loss": 0.3029, "step": 6068 }, { "epoch": 1.06, "learning_rate": 4.981568597294716e-05, "loss": 0.2986, "step": 6069 }, { "epoch": 1.06, "learning_rate": 4.9814810536494924e-05, "loss": 0.3005, "step": 6070 }, { "epoch": 1.06, "learning_rate": 4.9813933034526e-05, "loss": 0.2981, "step": 6071 }, { "epoch": 1.06, "learning_rate": 4.9813053467121646e-05, "loss": 0.3107, "step": 6072 }, { "epoch": 1.06, "learning_rate": 4.981217183436326e-05, "loss": 0.2955, "step": 6073 }, { "epoch": 1.06, "learning_rate": 4.981128813633244e-05, "loss": 0.3077, "step": 6074 }, { "epoch": 1.06, "learning_rate": 4.9810402373111e-05, "loss": 0.2901, "step": 6075 }, { "epoch": 1.06, "learning_rate": 4.980951454478091e-05, "loss": 0.294, "step": 6076 }, { "epoch": 1.06, "learning_rate": 4.980862465142437e-05, "loss": 0.3107, "step": 6077 }, { "epoch": 1.06, "learning_rate": 4.9807732693123735e-05, "loss": 0.2957, "step": 6078 }, { "epoch": 1.06, "learning_rate": 4.980683866996158e-05, "loss": 0.3058, "step": 6079 }, { "epoch": 1.06, "learning_rate": 4.980594258202064e-05, "loss": 0.3085, "step": 6080 }, { "epoch": 1.06, "learning_rate": 4.9805044429383886e-05, "loss": 0.2918, "step": 6081 }, { "epoch": 1.06, "learning_rate": 4.980414421213442e-05, "loss": 0.3208, "step": 6082 }, { "epoch": 1.06, "learning_rate": 4.98032419303556e-05, "loss": 0.2907, "step": 6083 }, { "epoch": 1.06, "learning_rate": 4.9802337584130926e-05, "loss": 0.291, "step": 6084 }, { "epoch": 1.06, "learning_rate": 4.9801431173544114e-05, "loss": 0.298, "step": 6085 }, { "epoch": 1.06, "learning_rate": 4.9800522698679056e-05, "loss": 0.2965, "step": 6086 }, { "epoch": 1.06, "learning_rate": 4.979961215961984e-05, "loss": 0.3001, "step": 6087 }, { "epoch": 1.06, "learning_rate": 4.9798699556450765e-05, "loss": 0.3001, "step": 6088 }, { "epoch": 1.06, "learning_rate": 4.979778488925629e-05, "loss": 0.3046, "step": 6089 }, { "epoch": 1.06, "learning_rate": 4.979686815812108e-05, "loss": 0.3119, "step": 6090 }, { "epoch": 1.06, "learning_rate": 4.979594936312999e-05, "loss": 0.309, "step": 6091 }, { "epoch": 1.06, "learning_rate": 4.979502850436808e-05, "loss": 0.309, "step": 6092 }, { "epoch": 1.06, "learning_rate": 4.979410558192057e-05, "loss": 0.3153, "step": 6093 }, { "epoch": 1.06, "learning_rate": 4.979318059587289e-05, "loss": 0.3051, "step": 6094 }, { "epoch": 1.06, "learning_rate": 4.979225354631067e-05, "loss": 0.3052, "step": 6095 }, { "epoch": 1.06, "learning_rate": 4.979132443331971e-05, "loss": 0.3063, "step": 6096 }, { "epoch": 1.06, "learning_rate": 4.979039325698601e-05, "loss": 0.2951, "step": 6097 }, { "epoch": 1.06, "learning_rate": 4.978946001739577e-05, "loss": 0.3155, "step": 6098 }, { "epoch": 1.06, "learning_rate": 4.9788524714635374e-05, "loss": 0.2927, "step": 6099 }, { "epoch": 1.06, "learning_rate": 4.978758734879138e-05, "loss": 0.3007, "step": 6100 }, { "epoch": 1.06, "learning_rate": 4.9786647919950575e-05, "loss": 0.3168, "step": 6101 }, { "epoch": 1.06, "learning_rate": 4.9785706428199905e-05, "loss": 0.2884, "step": 6102 }, { "epoch": 1.06, "learning_rate": 4.97847628736265e-05, "loss": 0.3035, "step": 6103 }, { "epoch": 1.06, "learning_rate": 4.978381725631773e-05, "loss": 0.3024, "step": 6104 }, { "epoch": 1.06, "learning_rate": 4.9782869576361105e-05, "loss": 0.2941, "step": 6105 }, { "epoch": 1.06, "learning_rate": 4.978191983384435e-05, "loss": 0.3067, "step": 6106 }, { "epoch": 1.06, "learning_rate": 4.9780968028855366e-05, "loss": 0.3096, "step": 6107 }, { "epoch": 1.06, "learning_rate": 4.978001416148227e-05, "loss": 0.3008, "step": 6108 }, { "epoch": 1.06, "learning_rate": 4.977905823181334e-05, "loss": 0.2979, "step": 6109 }, { "epoch": 1.06, "learning_rate": 4.977810023993706e-05, "loss": 0.31, "step": 6110 }, { "epoch": 1.07, "learning_rate": 4.977714018594212e-05, "loss": 0.3112, "step": 6111 }, { "epoch": 1.07, "learning_rate": 4.9776178069917365e-05, "loss": 0.2968, "step": 6112 }, { "epoch": 1.07, "learning_rate": 4.977521389195187e-05, "loss": 0.3206, "step": 6113 }, { "epoch": 1.07, "learning_rate": 4.977424765213487e-05, "loss": 0.2952, "step": 6114 }, { "epoch": 1.07, "learning_rate": 4.977327935055581e-05, "loss": 0.3045, "step": 6115 }, { "epoch": 1.07, "learning_rate": 4.977230898730431e-05, "loss": 0.3098, "step": 6116 }, { "epoch": 1.07, "learning_rate": 4.977133656247018e-05, "loss": 0.3084, "step": 6117 }, { "epoch": 1.07, "learning_rate": 4.977036207614346e-05, "loss": 0.3215, "step": 6118 }, { "epoch": 1.07, "learning_rate": 4.976938552841433e-05, "loss": 0.3138, "step": 6119 }, { "epoch": 1.07, "learning_rate": 4.976840691937319e-05, "loss": 0.3002, "step": 6120 }, { "epoch": 1.07, "learning_rate": 4.976742624911061e-05, "loss": 0.3031, "step": 6121 }, { "epoch": 1.07, "learning_rate": 4.9766443517717375e-05, "loss": 0.2999, "step": 6122 }, { "epoch": 1.07, "learning_rate": 4.9765458725284445e-05, "loss": 0.3, "step": 6123 }, { "epoch": 1.07, "learning_rate": 4.976447187190298e-05, "loss": 0.3015, "step": 6124 }, { "epoch": 1.07, "learning_rate": 4.976348295766432e-05, "loss": 0.2991, "step": 6125 }, { "epoch": 1.07, "learning_rate": 4.9762491982660004e-05, "loss": 0.3189, "step": 6126 }, { "epoch": 1.07, "learning_rate": 4.976149894698175e-05, "loss": 0.3092, "step": 6127 }, { "epoch": 1.07, "learning_rate": 4.97605038507215e-05, "loss": 0.3042, "step": 6128 }, { "epoch": 1.07, "learning_rate": 4.9759506693971336e-05, "loss": 0.2947, "step": 6129 }, { "epoch": 1.07, "learning_rate": 4.975850747682357e-05, "loss": 0.2971, "step": 6130 }, { "epoch": 1.07, "learning_rate": 4.9757506199370694e-05, "loss": 0.309, "step": 6131 }, { "epoch": 1.07, "learning_rate": 4.975650286170539e-05, "loss": 0.3055, "step": 6132 }, { "epoch": 1.07, "learning_rate": 4.9755497463920514e-05, "loss": 0.3089, "step": 6133 }, { "epoch": 1.07, "learning_rate": 4.9754490006109146e-05, "loss": 0.3099, "step": 6134 }, { "epoch": 1.07, "learning_rate": 4.9753480488364526e-05, "loss": 0.304, "step": 6135 }, { "epoch": 1.07, "learning_rate": 4.9752468910780106e-05, "loss": 0.3055, "step": 6136 }, { "epoch": 1.07, "learning_rate": 4.975145527344952e-05, "loss": 0.2956, "step": 6137 }, { "epoch": 1.07, "learning_rate": 4.975043957646659e-05, "loss": 0.307, "step": 6138 }, { "epoch": 1.07, "learning_rate": 4.974942181992534e-05, "loss": 0.3235, "step": 6139 }, { "epoch": 1.07, "learning_rate": 4.974840200391996e-05, "loss": 0.2932, "step": 6140 }, { "epoch": 1.07, "learning_rate": 4.9747380128544855e-05, "loss": 0.3108, "step": 6141 }, { "epoch": 1.07, "learning_rate": 4.974635619389462e-05, "loss": 0.3105, "step": 6142 }, { "epoch": 1.07, "learning_rate": 4.9745330200064014e-05, "loss": 0.3035, "step": 6143 }, { "epoch": 1.07, "learning_rate": 4.9744302147148023e-05, "loss": 0.3, "step": 6144 }, { "epoch": 1.07, "learning_rate": 4.974327203524181e-05, "loss": 0.303, "step": 6145 }, { "epoch": 1.07, "learning_rate": 4.97422398644407e-05, "loss": 0.31, "step": 6146 }, { "epoch": 1.07, "learning_rate": 4.974120563484026e-05, "loss": 0.2946, "step": 6147 }, { "epoch": 1.07, "learning_rate": 4.9740169346536206e-05, "loss": 0.3165, "step": 6148 }, { "epoch": 1.07, "learning_rate": 4.9739130999624464e-05, "loss": 0.3024, "step": 6149 }, { "epoch": 1.07, "learning_rate": 4.9738090594201144e-05, "loss": 0.2995, "step": 6150 }, { "epoch": 1.07, "learning_rate": 4.973704813036255e-05, "loss": 0.3146, "step": 6151 }, { "epoch": 1.07, "learning_rate": 4.973600360820518e-05, "loss": 0.3023, "step": 6152 }, { "epoch": 1.07, "learning_rate": 4.973495702782571e-05, "loss": 0.3173, "step": 6153 }, { "epoch": 1.07, "learning_rate": 4.973390838932102e-05, "loss": 0.3005, "step": 6154 }, { "epoch": 1.07, "learning_rate": 4.973285769278817e-05, "loss": 0.2944, "step": 6155 }, { "epoch": 1.07, "learning_rate": 4.973180493832441e-05, "loss": 0.3284, "step": 6156 }, { "epoch": 1.07, "learning_rate": 4.9730750126027205e-05, "loss": 0.3042, "step": 6157 }, { "epoch": 1.07, "learning_rate": 4.972969325599417e-05, "loss": 0.3043, "step": 6158 }, { "epoch": 1.07, "learning_rate": 4.972863432832315e-05, "loss": 0.296, "step": 6159 }, { "epoch": 1.07, "learning_rate": 4.9727573343112136e-05, "loss": 0.3146, "step": 6160 }, { "epoch": 1.07, "learning_rate": 4.9726510300459364e-05, "loss": 0.3094, "step": 6161 }, { "epoch": 1.07, "learning_rate": 4.972544520046322e-05, "loss": 0.3038, "step": 6162 }, { "epoch": 1.07, "learning_rate": 4.972437804322229e-05, "loss": 0.2961, "step": 6163 }, { "epoch": 1.07, "learning_rate": 4.972330882883536e-05, "loss": 0.2908, "step": 6164 }, { "epoch": 1.07, "learning_rate": 4.972223755740139e-05, "loss": 0.2988, "step": 6165 }, { "epoch": 1.07, "learning_rate": 4.9721164229019544e-05, "loss": 0.2912, "step": 6166 }, { "epoch": 1.07, "learning_rate": 4.972008884378918e-05, "loss": 0.3043, "step": 6167 }, { "epoch": 1.08, "learning_rate": 4.971901140180983e-05, "loss": 0.3071, "step": 6168 }, { "epoch": 1.08, "learning_rate": 4.9717931903181215e-05, "loss": 0.3089, "step": 6169 }, { "epoch": 1.08, "learning_rate": 4.971685034800327e-05, "loss": 0.3169, "step": 6170 }, { "epoch": 1.08, "learning_rate": 4.9715766736376105e-05, "loss": 0.3005, "step": 6171 }, { "epoch": 1.08, "learning_rate": 4.9714681068400026e-05, "loss": 0.3038, "step": 6172 }, { "epoch": 1.08, "learning_rate": 4.971359334417551e-05, "loss": 0.3118, "step": 6173 }, { "epoch": 1.08, "learning_rate": 4.971250356380326e-05, "loss": 0.3134, "step": 6174 }, { "epoch": 1.08, "learning_rate": 4.9711411727384126e-05, "loss": 0.3285, "step": 6175 }, { "epoch": 1.08, "learning_rate": 4.971031783501919e-05, "loss": 0.2985, "step": 6176 }, { "epoch": 1.08, "learning_rate": 4.9709221886809695e-05, "loss": 0.3089, "step": 6177 }, { "epoch": 1.08, "learning_rate": 4.97081238828571e-05, "loss": 0.2906, "step": 6178 }, { "epoch": 1.08, "learning_rate": 4.9707023823263014e-05, "loss": 0.3023, "step": 6179 }, { "epoch": 1.08, "learning_rate": 4.970592170812928e-05, "loss": 0.3173, "step": 6180 }, { "epoch": 1.08, "learning_rate": 4.970481753755791e-05, "loss": 0.3019, "step": 6181 }, { "epoch": 1.08, "learning_rate": 4.970371131165111e-05, "loss": 0.3175, "step": 6182 }, { "epoch": 1.08, "learning_rate": 4.970260303051126e-05, "loss": 0.307, "step": 6183 }, { "epoch": 1.08, "learning_rate": 4.9701492694240967e-05, "loss": 0.2999, "step": 6184 }, { "epoch": 1.08, "learning_rate": 4.970038030294299e-05, "loss": 0.2985, "step": 6185 }, { "epoch": 1.08, "learning_rate": 4.96992658567203e-05, "loss": 0.3057, "step": 6186 }, { "epoch": 1.08, "learning_rate": 4.969814935567606e-05, "loss": 0.3024, "step": 6187 }, { "epoch": 1.08, "learning_rate": 4.9697030799913606e-05, "loss": 0.3041, "step": 6188 }, { "epoch": 1.08, "learning_rate": 4.9695910189536484e-05, "loss": 0.2984, "step": 6189 }, { "epoch": 1.08, "learning_rate": 4.9694787524648415e-05, "loss": 0.3464, "step": 6190 }, { "epoch": 1.08, "learning_rate": 4.969366280535331e-05, "loss": 0.2998, "step": 6191 }, { "epoch": 1.08, "learning_rate": 4.9692536031755297e-05, "loss": 0.2989, "step": 6192 }, { "epoch": 1.08, "learning_rate": 4.969140720395865e-05, "loss": 0.3157, "step": 6193 }, { "epoch": 1.08, "learning_rate": 4.969027632206786e-05, "loss": 0.3076, "step": 6194 }, { "epoch": 1.08, "learning_rate": 4.968914338618761e-05, "loss": 0.3116, "step": 6195 }, { "epoch": 1.08, "learning_rate": 4.9688008396422774e-05, "loss": 0.3196, "step": 6196 }, { "epoch": 1.08, "learning_rate": 4.96868713528784e-05, "loss": 0.2946, "step": 6197 }, { "epoch": 1.08, "learning_rate": 4.9685732255659736e-05, "loss": 0.3008, "step": 6198 }, { "epoch": 1.08, "learning_rate": 4.968459110487222e-05, "loss": 0.3064, "step": 6199 }, { "epoch": 1.08, "learning_rate": 4.968344790062149e-05, "loss": 0.2982, "step": 6200 }, { "epoch": 1.08, "learning_rate": 4.968230264301334e-05, "loss": 0.2964, "step": 6201 }, { "epoch": 1.08, "learning_rate": 4.96811553321538e-05, "loss": 0.3063, "step": 6202 }, { "epoch": 1.08, "learning_rate": 4.968000596814907e-05, "loss": 0.2946, "step": 6203 }, { "epoch": 1.08, "learning_rate": 4.967885455110552e-05, "loss": 0.3192, "step": 6204 }, { "epoch": 1.08, "learning_rate": 4.967770108112974e-05, "loss": 0.3041, "step": 6205 }, { "epoch": 1.08, "learning_rate": 4.9676545558328504e-05, "loss": 0.2905, "step": 6206 }, { "epoch": 1.08, "learning_rate": 4.967538798280875e-05, "loss": 0.3056, "step": 6207 }, { "epoch": 1.08, "learning_rate": 4.967422835467764e-05, "loss": 0.3047, "step": 6208 }, { "epoch": 1.08, "learning_rate": 4.967306667404252e-05, "loss": 0.3099, "step": 6209 }, { "epoch": 1.08, "learning_rate": 4.9671902941010894e-05, "loss": 0.302, "step": 6210 }, { "epoch": 1.08, "learning_rate": 4.9670737155690503e-05, "loss": 0.2987, "step": 6211 }, { "epoch": 1.08, "learning_rate": 4.966956931818925e-05, "loss": 0.2991, "step": 6212 }, { "epoch": 1.08, "learning_rate": 4.966839942861523e-05, "loss": 0.316, "step": 6213 }, { "epoch": 1.08, "learning_rate": 4.966722748707672e-05, "loss": 0.3192, "step": 6214 }, { "epoch": 1.08, "learning_rate": 4.966605349368222e-05, "loss": 0.3131, "step": 6215 }, { "epoch": 1.08, "learning_rate": 4.966487744854039e-05, "loss": 0.3182, "step": 6216 }, { "epoch": 1.08, "learning_rate": 4.966369935176007e-05, "loss": 0.2976, "step": 6217 }, { "epoch": 1.08, "learning_rate": 4.966251920345034e-05, "loss": 0.312, "step": 6218 }, { "epoch": 1.08, "learning_rate": 4.966133700372041e-05, "loss": 0.3081, "step": 6219 }, { "epoch": 1.08, "learning_rate": 4.9660152752679716e-05, "loss": 0.3265, "step": 6220 }, { "epoch": 1.08, "learning_rate": 4.9658966450437885e-05, "loss": 0.316, "step": 6221 }, { "epoch": 1.08, "learning_rate": 4.965777809710471e-05, "loss": 0.3035, "step": 6222 }, { "epoch": 1.08, "learning_rate": 4.965658769279019e-05, "loss": 0.3083, "step": 6223 }, { "epoch": 1.08, "learning_rate": 4.965539523760452e-05, "loss": 0.3268, "step": 6224 }, { "epoch": 1.09, "learning_rate": 4.965420073165807e-05, "loss": 0.3092, "step": 6225 }, { "epoch": 1.09, "learning_rate": 4.965300417506141e-05, "loss": 0.3044, "step": 6226 }, { "epoch": 1.09, "learning_rate": 4.96518055679253e-05, "loss": 0.316, "step": 6227 }, { "epoch": 1.09, "learning_rate": 4.965060491036068e-05, "loss": 0.3006, "step": 6228 }, { "epoch": 1.09, "learning_rate": 4.964940220247869e-05, "loss": 0.2985, "step": 6229 }, { "epoch": 1.09, "learning_rate": 4.9648197444390654e-05, "loss": 0.3023, "step": 6230 }, { "epoch": 1.09, "learning_rate": 4.964699063620807e-05, "loss": 0.3002, "step": 6231 }, { "epoch": 1.09, "learning_rate": 4.964578177804268e-05, "loss": 0.3051, "step": 6232 }, { "epoch": 1.09, "learning_rate": 4.964457087000636e-05, "loss": 0.2902, "step": 6233 }, { "epoch": 1.09, "learning_rate": 4.964335791221118e-05, "loss": 0.3066, "step": 6234 }, { "epoch": 1.09, "learning_rate": 4.9642142904769435e-05, "loss": 0.2971, "step": 6235 }, { "epoch": 1.09, "learning_rate": 4.9640925847793586e-05, "loss": 0.2926, "step": 6236 }, { "epoch": 1.09, "learning_rate": 4.963970674139628e-05, "loss": 0.2966, "step": 6237 }, { "epoch": 1.09, "learning_rate": 4.9638485585690365e-05, "loss": 0.3084, "step": 6238 }, { "epoch": 1.09, "learning_rate": 4.963726238078888e-05, "loss": 0.3094, "step": 6239 }, { "epoch": 1.09, "learning_rate": 4.963603712680503e-05, "loss": 0.3112, "step": 6240 }, { "epoch": 1.09, "learning_rate": 4.963480982385225e-05, "loss": 0.2977, "step": 6241 }, { "epoch": 1.09, "learning_rate": 4.963358047204413e-05, "loss": 0.2938, "step": 6242 }, { "epoch": 1.09, "learning_rate": 4.963234907149446e-05, "loss": 0.3161, "step": 6243 }, { "epoch": 1.09, "learning_rate": 4.963111562231723e-05, "loss": 0.3166, "step": 6244 }, { "epoch": 1.09, "learning_rate": 4.9629880124626596e-05, "loss": 0.3011, "step": 6245 }, { "epoch": 1.09, "learning_rate": 4.962864257853694e-05, "loss": 0.3031, "step": 6246 }, { "epoch": 1.09, "learning_rate": 4.9627402984162795e-05, "loss": 0.3088, "step": 6247 }, { "epoch": 1.09, "learning_rate": 4.9626161341618914e-05, "loss": 0.3221, "step": 6248 }, { "epoch": 1.09, "learning_rate": 4.9624917651020226e-05, "loss": 0.307, "step": 6249 }, { "epoch": 1.09, "learning_rate": 4.962367191248183e-05, "loss": 0.3193, "step": 6250 }, { "epoch": 1.09, "learning_rate": 4.9622424126119066e-05, "loss": 0.313, "step": 6251 }, { "epoch": 1.09, "learning_rate": 4.96211742920474e-05, "loss": 0.291, "step": 6252 }, { "epoch": 1.09, "learning_rate": 4.961992241038255e-05, "loss": 0.3158, "step": 6253 }, { "epoch": 1.09, "learning_rate": 4.961866848124037e-05, "loss": 0.2981, "step": 6254 }, { "epoch": 1.09, "learning_rate": 4.9617412504736935e-05, "loss": 0.3007, "step": 6255 }, { "epoch": 1.09, "learning_rate": 4.9616154480988515e-05, "loss": 0.3048, "step": 6256 }, { "epoch": 1.09, "learning_rate": 4.9614894410111535e-05, "loss": 0.3014, "step": 6257 }, { "epoch": 1.09, "learning_rate": 4.961363229222264e-05, "loss": 0.3037, "step": 6258 }, { "epoch": 1.09, "learning_rate": 4.961236812743866e-05, "loss": 0.3109, "step": 6259 }, { "epoch": 1.09, "learning_rate": 4.961110191587659e-05, "loss": 0.3051, "step": 6260 }, { "epoch": 1.09, "learning_rate": 4.960983365765366e-05, "loss": 0.306, "step": 6261 }, { "epoch": 1.09, "learning_rate": 4.9608563352887245e-05, "loss": 0.3034, "step": 6262 }, { "epoch": 1.09, "learning_rate": 4.960729100169494e-05, "loss": 0.3144, "step": 6263 }, { "epoch": 1.09, "learning_rate": 4.96060166041945e-05, "loss": 0.3226, "step": 6264 }, { "epoch": 1.09, "learning_rate": 4.9604740160503906e-05, "loss": 0.3017, "step": 6265 }, { "epoch": 1.09, "learning_rate": 4.960346167074129e-05, "loss": 0.3084, "step": 6266 }, { "epoch": 1.09, "learning_rate": 4.9602181135025006e-05, "loss": 0.3207, "step": 6267 }, { "epoch": 1.09, "learning_rate": 4.960089855347358e-05, "loss": 0.294, "step": 6268 }, { "epoch": 1.09, "learning_rate": 4.9599613926205734e-05, "loss": 0.303, "step": 6269 }, { "epoch": 1.09, "learning_rate": 4.959832725334037e-05, "loss": 0.308, "step": 6270 }, { "epoch": 1.09, "learning_rate": 4.959703853499659e-05, "loss": 0.318, "step": 6271 }, { "epoch": 1.09, "learning_rate": 4.9595747771293686e-05, "loss": 0.3031, "step": 6272 }, { "epoch": 1.09, "learning_rate": 4.959445496235112e-05, "loss": 0.2909, "step": 6273 }, { "epoch": 1.09, "learning_rate": 4.959316010828857e-05, "loss": 0.3145, "step": 6274 }, { "epoch": 1.09, "learning_rate": 4.959186320922589e-05, "loss": 0.3145, "step": 6275 }, { "epoch": 1.09, "learning_rate": 4.959056426528311e-05, "loss": 0.3111, "step": 6276 }, { "epoch": 1.09, "learning_rate": 4.9589263276580484e-05, "loss": 0.3035, "step": 6277 }, { "epoch": 1.09, "learning_rate": 4.958796024323843e-05, "loss": 0.3122, "step": 6278 }, { "epoch": 1.09, "learning_rate": 4.958665516537755e-05, "loss": 0.3041, "step": 6279 }, { "epoch": 1.09, "learning_rate": 4.958534804311866e-05, "loss": 0.2977, "step": 6280 }, { "epoch": 1.09, "learning_rate": 4.958403887658273e-05, "loss": 0.3107, "step": 6281 }, { "epoch": 1.09, "learning_rate": 4.958272766589096e-05, "loss": 0.3036, "step": 6282 }, { "epoch": 1.1, "learning_rate": 4.958141441116471e-05, "loss": 0.3071, "step": 6283 }, { "epoch": 1.1, "learning_rate": 4.958009911252555e-05, "loss": 0.3084, "step": 6284 }, { "epoch": 1.1, "learning_rate": 4.95787817700952e-05, "loss": 0.2995, "step": 6285 }, { "epoch": 1.1, "learning_rate": 4.9577462383995624e-05, "loss": 0.3008, "step": 6286 }, { "epoch": 1.1, "learning_rate": 4.957614095434893e-05, "loss": 0.3214, "step": 6287 }, { "epoch": 1.1, "learning_rate": 4.9574817481277444e-05, "loss": 0.3032, "step": 6288 }, { "epoch": 1.1, "learning_rate": 4.9573491964903665e-05, "loss": 0.2972, "step": 6289 }, { "epoch": 1.1, "learning_rate": 4.957216440535029e-05, "loss": 0.2991, "step": 6290 }, { "epoch": 1.1, "learning_rate": 4.9570834802740195e-05, "loss": 0.3043, "step": 6291 }, { "epoch": 1.1, "learning_rate": 4.956950315719646e-05, "loss": 0.3089, "step": 6292 }, { "epoch": 1.1, "learning_rate": 4.9568169468842344e-05, "loss": 0.3224, "step": 6293 }, { "epoch": 1.1, "learning_rate": 4.956683373780129e-05, "loss": 0.3007, "step": 6294 }, { "epoch": 1.1, "learning_rate": 4.956549596419694e-05, "loss": 0.3109, "step": 6295 }, { "epoch": 1.1, "learning_rate": 4.9564156148153125e-05, "loss": 0.3062, "step": 6296 }, { "epoch": 1.1, "learning_rate": 4.9562814289793855e-05, "loss": 0.3078, "step": 6297 }, { "epoch": 1.1, "learning_rate": 4.956147038924334e-05, "loss": 0.3034, "step": 6298 }, { "epoch": 1.1, "learning_rate": 4.9560124446625984e-05, "loss": 0.2962, "step": 6299 }, { "epoch": 1.1, "learning_rate": 4.955877646206635e-05, "loss": 0.3052, "step": 6300 }, { "epoch": 1.1, "learning_rate": 4.955742643568923e-05, "loss": 0.3009, "step": 6301 }, { "epoch": 1.1, "learning_rate": 4.955607436761958e-05, "loss": 0.3084, "step": 6302 }, { "epoch": 1.1, "learning_rate": 4.955472025798254e-05, "loss": 0.3022, "step": 6303 }, { "epoch": 1.1, "learning_rate": 4.9553364106903466e-05, "loss": 0.2986, "step": 6304 }, { "epoch": 1.1, "learning_rate": 4.955200591450788e-05, "loss": 0.2945, "step": 6305 }, { "epoch": 1.1, "learning_rate": 4.95506456809215e-05, "loss": 0.3143, "step": 6306 }, { "epoch": 1.1, "learning_rate": 4.954928340627022e-05, "loss": 0.3012, "step": 6307 }, { "epoch": 1.1, "learning_rate": 4.9547919090680175e-05, "loss": 0.3036, "step": 6308 }, { "epoch": 1.1, "learning_rate": 4.95465527342776e-05, "loss": 0.2991, "step": 6309 }, { "epoch": 1.1, "learning_rate": 4.954518433718901e-05, "loss": 0.3187, "step": 6310 }, { "epoch": 1.1, "learning_rate": 4.954381389954104e-05, "loss": 0.3384, "step": 6311 }, { "epoch": 1.1, "learning_rate": 4.954244142146055e-05, "loss": 0.2897, "step": 6312 }, { "epoch": 1.1, "learning_rate": 4.9541066903074584e-05, "loss": 0.3106, "step": 6313 }, { "epoch": 1.1, "learning_rate": 4.953969034451036e-05, "loss": 0.3033, "step": 6314 }, { "epoch": 1.1, "learning_rate": 4.953831174589531e-05, "loss": 0.3088, "step": 6315 }, { "epoch": 1.1, "learning_rate": 4.953693110735703e-05, "loss": 0.3079, "step": 6316 }, { "epoch": 1.1, "learning_rate": 4.953554842902332e-05, "loss": 0.3002, "step": 6317 }, { "epoch": 1.1, "learning_rate": 4.9534163711022166e-05, "loss": 0.3124, "step": 6318 }, { "epoch": 1.1, "learning_rate": 4.9532776953481735e-05, "loss": 0.3135, "step": 6319 }, { "epoch": 1.1, "learning_rate": 4.95313881565304e-05, "loss": 0.321, "step": 6320 }, { "epoch": 1.1, "learning_rate": 4.95299973202967e-05, "loss": 0.3137, "step": 6321 }, { "epoch": 1.1, "learning_rate": 4.952860444490938e-05, "loss": 0.3044, "step": 6322 }, { "epoch": 1.1, "learning_rate": 4.9527209530497365e-05, "loss": 0.3071, "step": 6323 }, { "epoch": 1.1, "learning_rate": 4.952581257718977e-05, "loss": 0.3132, "step": 6324 }, { "epoch": 1.1, "learning_rate": 4.952441358511591e-05, "loss": 0.3298, "step": 6325 }, { "epoch": 1.1, "learning_rate": 4.952301255440527e-05, "loss": 0.2957, "step": 6326 }, { "epoch": 1.1, "learning_rate": 4.952160948518754e-05, "loss": 0.2999, "step": 6327 }, { "epoch": 1.1, "learning_rate": 4.952020437759258e-05, "loss": 0.3028, "step": 6328 }, { "epoch": 1.1, "learning_rate": 4.9518797231750474e-05, "loss": 0.3071, "step": 6329 }, { "epoch": 1.1, "learning_rate": 4.9517388047791444e-05, "loss": 0.2941, "step": 6330 }, { "epoch": 1.1, "learning_rate": 4.951597682584594e-05, "loss": 0.2988, "step": 6331 }, { "epoch": 1.1, "learning_rate": 4.9514563566044594e-05, "loss": 0.3092, "step": 6332 }, { "epoch": 1.1, "learning_rate": 4.9513148268518216e-05, "loss": 0.3129, "step": 6333 }, { "epoch": 1.1, "learning_rate": 4.9511730933397807e-05, "loss": 0.3135, "step": 6334 }, { "epoch": 1.1, "learning_rate": 4.951031156081456e-05, "loss": 0.3147, "step": 6335 }, { "epoch": 1.1, "learning_rate": 4.950889015089986e-05, "loss": 0.3105, "step": 6336 }, { "epoch": 1.1, "learning_rate": 4.950746670378526e-05, "loss": 0.3135, "step": 6337 }, { "epoch": 1.1, "learning_rate": 4.950604121960254e-05, "loss": 0.3026, "step": 6338 }, { "epoch": 1.1, "learning_rate": 4.9504613698483645e-05, "loss": 0.3159, "step": 6339 }, { "epoch": 1.11, "learning_rate": 4.9503184140560686e-05, "loss": 0.3069, "step": 6340 }, { "epoch": 1.11, "learning_rate": 4.950175254596601e-05, "loss": 0.3144, "step": 6341 }, { "epoch": 1.11, "learning_rate": 4.950031891483213e-05, "loss": 0.2871, "step": 6342 }, { "epoch": 1.11, "learning_rate": 4.9498883247291725e-05, "loss": 0.3009, "step": 6343 }, { "epoch": 1.11, "learning_rate": 4.9497445543477706e-05, "loss": 0.2837, "step": 6344 }, { "epoch": 1.11, "learning_rate": 4.9496005803523144e-05, "loss": 0.2992, "step": 6345 }, { "epoch": 1.11, "learning_rate": 4.94945640275613e-05, "loss": 0.3258, "step": 6346 }, { "epoch": 1.11, "learning_rate": 4.949312021572563e-05, "loss": 0.3054, "step": 6347 }, { "epoch": 1.11, "learning_rate": 4.949167436814979e-05, "loss": 0.3097, "step": 6348 }, { "epoch": 1.11, "learning_rate": 4.949022648496759e-05, "loss": 0.3098, "step": 6349 }, { "epoch": 1.11, "learning_rate": 4.948877656631307e-05, "loss": 0.302, "step": 6350 }, { "epoch": 1.11, "learning_rate": 4.948732461232041e-05, "loss": 0.3246, "step": 6351 }, { "epoch": 1.11, "learning_rate": 4.948587062312404e-05, "loss": 0.3016, "step": 6352 }, { "epoch": 1.11, "learning_rate": 4.948441459885853e-05, "loss": 0.3086, "step": 6353 }, { "epoch": 1.11, "learning_rate": 4.948295653965865e-05, "loss": 0.3138, "step": 6354 }, { "epoch": 1.11, "learning_rate": 4.948149644565936e-05, "loss": 0.3073, "step": 6355 }, { "epoch": 1.11, "learning_rate": 4.9480034316995826e-05, "loss": 0.2985, "step": 6356 }, { "epoch": 1.11, "learning_rate": 4.947857015380337e-05, "loss": 0.3075, "step": 6357 }, { "epoch": 1.11, "learning_rate": 4.947710395621753e-05, "loss": 0.3081, "step": 6358 }, { "epoch": 1.11, "learning_rate": 4.9475635724374005e-05, "loss": 0.3125, "step": 6359 }, { "epoch": 1.11, "learning_rate": 4.9474165458408714e-05, "loss": 0.3205, "step": 6360 }, { "epoch": 1.11, "learning_rate": 4.9472693158457746e-05, "loss": 0.2966, "step": 6361 }, { "epoch": 1.11, "learning_rate": 4.947121882465737e-05, "loss": 0.293, "step": 6362 }, { "epoch": 1.11, "learning_rate": 4.9469742457144075e-05, "loss": 0.299, "step": 6363 }, { "epoch": 1.11, "learning_rate": 4.9468264056054496e-05, "loss": 0.308, "step": 6364 }, { "epoch": 1.11, "learning_rate": 4.9466783621525494e-05, "loss": 0.3001, "step": 6365 }, { "epoch": 1.11, "learning_rate": 4.9465301153694086e-05, "loss": 0.3059, "step": 6366 }, { "epoch": 1.11, "learning_rate": 4.946381665269751e-05, "loss": 0.299, "step": 6367 }, { "epoch": 1.11, "learning_rate": 4.946233011867315e-05, "loss": 0.3036, "step": 6368 }, { "epoch": 1.11, "learning_rate": 4.946084155175864e-05, "loss": 0.3119, "step": 6369 }, { "epoch": 1.11, "learning_rate": 4.945935095209173e-05, "loss": 0.3127, "step": 6370 }, { "epoch": 1.11, "learning_rate": 4.9457858319810426e-05, "loss": 0.3024, "step": 6371 }, { "epoch": 1.11, "learning_rate": 4.9456363655052864e-05, "loss": 0.3008, "step": 6372 }, { "epoch": 1.11, "learning_rate": 4.94548669579574e-05, "loss": 0.3084, "step": 6373 }, { "epoch": 1.11, "learning_rate": 4.945336822866258e-05, "loss": 0.3012, "step": 6374 }, { "epoch": 1.11, "learning_rate": 4.945186746730713e-05, "loss": 0.309, "step": 6375 }, { "epoch": 1.11, "learning_rate": 4.945036467402995e-05, "loss": 0.295, "step": 6376 }, { "epoch": 1.11, "learning_rate": 4.944885984897016e-05, "loss": 0.3108, "step": 6377 }, { "epoch": 1.11, "learning_rate": 4.9447352992267045e-05, "loss": 0.303, "step": 6378 }, { "epoch": 1.11, "learning_rate": 4.944584410406008e-05, "loss": 0.2958, "step": 6379 }, { "epoch": 1.11, "learning_rate": 4.9444333184488926e-05, "loss": 0.3057, "step": 6380 }, { "epoch": 1.11, "learning_rate": 4.944282023369345e-05, "loss": 0.314, "step": 6381 }, { "epoch": 1.11, "learning_rate": 4.944130525181369e-05, "loss": 0.3045, "step": 6382 }, { "epoch": 1.11, "learning_rate": 4.943978823898988e-05, "loss": 0.309, "step": 6383 }, { "epoch": 1.11, "learning_rate": 4.943826919536244e-05, "loss": 0.312, "step": 6384 }, { "epoch": 1.11, "learning_rate": 4.943674812107196e-05, "loss": 0.2933, "step": 6385 }, { "epoch": 1.11, "learning_rate": 4.9435225016259244e-05, "loss": 0.2943, "step": 6386 }, { "epoch": 1.11, "learning_rate": 4.9433699881065286e-05, "loss": 0.3156, "step": 6387 }, { "epoch": 1.11, "learning_rate": 4.943217271563125e-05, "loss": 0.2842, "step": 6388 }, { "epoch": 1.11, "learning_rate": 4.943064352009848e-05, "loss": 0.303, "step": 6389 }, { "epoch": 1.11, "learning_rate": 4.9429112294608546e-05, "loss": 0.315, "step": 6390 }, { "epoch": 1.11, "learning_rate": 4.942757903930316e-05, "loss": 0.3094, "step": 6391 }, { "epoch": 1.11, "learning_rate": 4.942604375432425e-05, "loss": 0.3063, "step": 6392 }, { "epoch": 1.11, "learning_rate": 4.942450643981394e-05, "loss": 0.2942, "step": 6393 }, { "epoch": 1.11, "learning_rate": 4.9422967095914515e-05, "loss": 0.3109, "step": 6394 }, { "epoch": 1.11, "learning_rate": 4.9421425722768466e-05, "loss": 0.3055, "step": 6395 }, { "epoch": 1.11, "learning_rate": 4.941988232051846e-05, "loss": 0.3054, "step": 6396 }, { "epoch": 1.11, "learning_rate": 4.9418336889307365e-05, "loss": 0.3104, "step": 6397 }, { "epoch": 1.12, "learning_rate": 4.941678942927822e-05, "loss": 0.3137, "step": 6398 }, { "epoch": 1.12, "learning_rate": 4.941523994057428e-05, "loss": 0.3102, "step": 6399 }, { "epoch": 1.12, "learning_rate": 4.941368842333896e-05, "loss": 0.3041, "step": 6400 }, { "epoch": 1.12, "learning_rate": 4.941213487771586e-05, "loss": 0.3222, "step": 6401 }, { "epoch": 1.12, "learning_rate": 4.94105793038488e-05, "loss": 0.2945, "step": 6402 }, { "epoch": 1.12, "learning_rate": 4.940902170188176e-05, "loss": 0.3067, "step": 6403 }, { "epoch": 1.12, "learning_rate": 4.940746207195891e-05, "loss": 0.3073, "step": 6404 }, { "epoch": 1.12, "learning_rate": 4.940590041422462e-05, "loss": 0.3088, "step": 6405 }, { "epoch": 1.12, "learning_rate": 4.9404336728823444e-05, "loss": 0.3089, "step": 6406 }, { "epoch": 1.12, "learning_rate": 4.940277101590012e-05, "loss": 0.3197, "step": 6407 }, { "epoch": 1.12, "learning_rate": 4.9401203275599556e-05, "loss": 0.3049, "step": 6408 }, { "epoch": 1.12, "learning_rate": 4.939963350806689e-05, "loss": 0.3219, "step": 6409 }, { "epoch": 1.12, "learning_rate": 4.939806171344742e-05, "loss": 0.3211, "step": 6410 }, { "epoch": 1.12, "learning_rate": 4.939648789188662e-05, "loss": 0.3128, "step": 6411 }, { "epoch": 1.12, "learning_rate": 4.939491204353019e-05, "loss": 0.3028, "step": 6412 }, { "epoch": 1.12, "learning_rate": 4.9393334168523974e-05, "loss": 0.3069, "step": 6413 }, { "epoch": 1.12, "learning_rate": 4.939175426701403e-05, "loss": 0.304, "step": 6414 }, { "epoch": 1.12, "learning_rate": 4.939017233914661e-05, "loss": 0.3132, "step": 6415 }, { "epoch": 1.12, "learning_rate": 4.938858838506812e-05, "loss": 0.2912, "step": 6416 }, { "epoch": 1.12, "learning_rate": 4.938700240492518e-05, "loss": 0.309, "step": 6417 }, { "epoch": 1.12, "learning_rate": 4.938541439886462e-05, "loss": 0.3028, "step": 6418 }, { "epoch": 1.12, "learning_rate": 4.9383824367033395e-05, "loss": 0.3139, "step": 6419 }, { "epoch": 1.12, "learning_rate": 4.9382232309578694e-05, "loss": 0.3007, "step": 6420 }, { "epoch": 1.12, "learning_rate": 4.9380638226647894e-05, "loss": 0.297, "step": 6421 }, { "epoch": 1.12, "learning_rate": 4.937904211838853e-05, "loss": 0.3121, "step": 6422 }, { "epoch": 1.12, "learning_rate": 4.9377443984948346e-05, "loss": 0.296, "step": 6423 }, { "epoch": 1.12, "learning_rate": 4.937584382647527e-05, "loss": 0.306, "step": 6424 }, { "epoch": 1.12, "learning_rate": 4.9374241643117434e-05, "loss": 0.3084, "step": 6425 }, { "epoch": 1.12, "learning_rate": 4.9372637435023115e-05, "loss": 0.3117, "step": 6426 }, { "epoch": 1.12, "learning_rate": 4.937103120234081e-05, "loss": 0.3127, "step": 6427 }, { "epoch": 1.12, "learning_rate": 4.936942294521921e-05, "loss": 0.3402, "step": 6428 }, { "epoch": 1.12, "learning_rate": 4.936781266380717e-05, "loss": 0.3032, "step": 6429 }, { "epoch": 1.12, "learning_rate": 4.936620035825373e-05, "loss": 0.301, "step": 6430 }, { "epoch": 1.12, "learning_rate": 4.936458602870814e-05, "loss": 0.3316, "step": 6431 }, { "epoch": 1.12, "learning_rate": 4.9362969675319836e-05, "loss": 0.2864, "step": 6432 }, { "epoch": 1.12, "learning_rate": 4.9361351298238414e-05, "loss": 0.3121, "step": 6433 }, { "epoch": 1.12, "learning_rate": 4.935973089761369e-05, "loss": 0.2938, "step": 6434 }, { "epoch": 1.12, "learning_rate": 4.9358108473595636e-05, "loss": 0.3109, "step": 6435 }, { "epoch": 1.12, "learning_rate": 4.9356484026334455e-05, "loss": 0.298, "step": 6436 }, { "epoch": 1.12, "learning_rate": 4.935485755598047e-05, "loss": 0.2986, "step": 6437 }, { "epoch": 1.12, "learning_rate": 4.9353229062684266e-05, "loss": 0.3207, "step": 6438 }, { "epoch": 1.12, "learning_rate": 4.935159854659657e-05, "loss": 0.3059, "step": 6439 }, { "epoch": 1.12, "learning_rate": 4.9349966007868306e-05, "loss": 0.3168, "step": 6440 }, { "epoch": 1.12, "learning_rate": 4.934833144665058e-05, "loss": 0.3078, "step": 6441 }, { "epoch": 1.12, "learning_rate": 4.934669486309471e-05, "loss": 0.3058, "step": 6442 }, { "epoch": 1.12, "learning_rate": 4.934505625735216e-05, "loss": 0.3044, "step": 6443 }, { "epoch": 1.12, "learning_rate": 4.93434156295746e-05, "loss": 0.3123, "step": 6444 }, { "epoch": 1.12, "learning_rate": 4.934177297991392e-05, "loss": 0.3108, "step": 6445 }, { "epoch": 1.12, "learning_rate": 4.9340128308522144e-05, "loss": 0.3026, "step": 6446 }, { "epoch": 1.12, "learning_rate": 4.9338481615551525e-05, "loss": 0.3201, "step": 6447 }, { "epoch": 1.12, "learning_rate": 4.9336832901154455e-05, "loss": 0.3271, "step": 6448 }, { "epoch": 1.12, "learning_rate": 4.933518216548358e-05, "loss": 0.308, "step": 6449 }, { "epoch": 1.12, "learning_rate": 4.933352940869167e-05, "loss": 0.2967, "step": 6450 }, { "epoch": 1.12, "learning_rate": 4.933187463093172e-05, "loss": 0.3084, "step": 6451 }, { "epoch": 1.12, "learning_rate": 4.933021783235689e-05, "loss": 0.3053, "step": 6452 }, { "epoch": 1.12, "learning_rate": 4.932855901312056e-05, "loss": 0.3202, "step": 6453 }, { "epoch": 1.12, "learning_rate": 4.932689817337626e-05, "loss": 0.3016, "step": 6454 }, { "epoch": 1.13, "learning_rate": 4.932523531327771e-05, "loss": 0.2914, "step": 6455 }, { "epoch": 1.13, "learning_rate": 4.932357043297884e-05, "loss": 0.2943, "step": 6456 }, { "epoch": 1.13, "learning_rate": 4.932190353263377e-05, "loss": 0.3037, "step": 6457 }, { "epoch": 1.13, "learning_rate": 4.932023461239677e-05, "loss": 0.3304, "step": 6458 }, { "epoch": 1.13, "learning_rate": 4.931856367242233e-05, "loss": 0.3105, "step": 6459 }, { "epoch": 1.13, "learning_rate": 4.931689071286511e-05, "loss": 0.3015, "step": 6460 }, { "epoch": 1.13, "learning_rate": 4.931521573387998e-05, "loss": 0.2983, "step": 6461 }, { "epoch": 1.13, "learning_rate": 4.9313538735621964e-05, "loss": 0.3089, "step": 6462 }, { "epoch": 1.13, "learning_rate": 4.931185971824629e-05, "loss": 0.3078, "step": 6463 }, { "epoch": 1.13, "learning_rate": 4.931017868190837e-05, "loss": 0.3054, "step": 6464 }, { "epoch": 1.13, "learning_rate": 4.930849562676382e-05, "loss": 0.3159, "step": 6465 }, { "epoch": 1.13, "learning_rate": 4.930681055296842e-05, "loss": 0.3267, "step": 6466 }, { "epoch": 1.13, "learning_rate": 4.9305123460678146e-05, "loss": 0.302, "step": 6467 }, { "epoch": 1.13, "learning_rate": 4.930343435004915e-05, "loss": 0.2974, "step": 6468 }, { "epoch": 1.13, "learning_rate": 4.9301743221237796e-05, "loss": 0.3164, "step": 6469 }, { "epoch": 1.13, "learning_rate": 4.9300050074400605e-05, "loss": 0.3, "step": 6470 }, { "epoch": 1.13, "learning_rate": 4.9298354909694306e-05, "loss": 0.3057, "step": 6471 }, { "epoch": 1.13, "learning_rate": 4.9296657727275805e-05, "loss": 0.3172, "step": 6472 }, { "epoch": 1.13, "learning_rate": 4.929495852730221e-05, "loss": 0.3091, "step": 6473 }, { "epoch": 1.13, "learning_rate": 4.929325730993079e-05, "loss": 0.3012, "step": 6474 }, { "epoch": 1.13, "learning_rate": 4.9291554075319005e-05, "loss": 0.3028, "step": 6475 }, { "epoch": 1.13, "learning_rate": 4.928984882362454e-05, "loss": 0.2973, "step": 6476 }, { "epoch": 1.13, "learning_rate": 4.928814155500521e-05, "loss": 0.3098, "step": 6477 }, { "epoch": 1.13, "learning_rate": 4.928643226961906e-05, "loss": 0.3061, "step": 6478 }, { "epoch": 1.13, "learning_rate": 4.92847209676243e-05, "loss": 0.3023, "step": 6479 }, { "epoch": 1.13, "learning_rate": 4.928300764917933e-05, "loss": 0.2856, "step": 6480 }, { "epoch": 1.13, "learning_rate": 4.928129231444275e-05, "loss": 0.3071, "step": 6481 }, { "epoch": 1.13, "learning_rate": 4.927957496357332e-05, "loss": 0.3152, "step": 6482 }, { "epoch": 1.13, "learning_rate": 4.927785559673002e-05, "loss": 0.3111, "step": 6483 }, { "epoch": 1.13, "learning_rate": 4.9276134214071986e-05, "loss": 0.3056, "step": 6484 }, { "epoch": 1.13, "learning_rate": 4.927441081575855e-05, "loss": 0.3069, "step": 6485 }, { "epoch": 1.13, "learning_rate": 4.9272685401949245e-05, "loss": 0.3002, "step": 6486 }, { "epoch": 1.13, "learning_rate": 4.927095797280378e-05, "loss": 0.2958, "step": 6487 }, { "epoch": 1.13, "learning_rate": 4.9269228528482046e-05, "loss": 0.3047, "step": 6488 }, { "epoch": 1.13, "learning_rate": 4.926749706914412e-05, "loss": 0.3096, "step": 6489 }, { "epoch": 1.13, "learning_rate": 4.926576359495028e-05, "loss": 0.2991, "step": 6490 }, { "epoch": 1.13, "learning_rate": 4.9264028106060984e-05, "loss": 0.3233, "step": 6491 }, { "epoch": 1.13, "learning_rate": 4.9262290602636854e-05, "loss": 0.303, "step": 6492 }, { "epoch": 1.13, "learning_rate": 4.926055108483874e-05, "loss": 0.3048, "step": 6493 }, { "epoch": 1.13, "learning_rate": 4.925880955282764e-05, "loss": 0.3054, "step": 6494 }, { "epoch": 1.13, "learning_rate": 4.9257066006764764e-05, "loss": 0.3173, "step": 6495 }, { "epoch": 1.13, "learning_rate": 4.925532044681149e-05, "loss": 0.3153, "step": 6496 }, { "epoch": 1.13, "learning_rate": 4.925357287312941e-05, "loss": 0.3136, "step": 6497 }, { "epoch": 1.13, "learning_rate": 4.925182328588026e-05, "loss": 0.3196, "step": 6498 }, { "epoch": 1.13, "learning_rate": 4.9250071685226004e-05, "loss": 0.2915, "step": 6499 }, { "epoch": 1.13, "learning_rate": 4.924831807132877e-05, "loss": 0.2952, "step": 6500 }, { "epoch": 1.13, "learning_rate": 4.9246562444350875e-05, "loss": 0.3055, "step": 6501 }, { "epoch": 1.13, "learning_rate": 4.924480480445482e-05, "loss": 0.31, "step": 6502 }, { "epoch": 1.13, "learning_rate": 4.9243045151803306e-05, "loss": 0.3054, "step": 6503 }, { "epoch": 1.13, "learning_rate": 4.9241283486559204e-05, "loss": 0.3089, "step": 6504 }, { "epoch": 1.13, "learning_rate": 4.923951980888558e-05, "loss": 0.295, "step": 6505 }, { "epoch": 1.13, "learning_rate": 4.923775411894569e-05, "loss": 0.3034, "step": 6506 }, { "epoch": 1.13, "learning_rate": 4.9235986416902965e-05, "loss": 0.3014, "step": 6507 }, { "epoch": 1.13, "learning_rate": 4.9234216702921026e-05, "loss": 0.317, "step": 6508 }, { "epoch": 1.13, "learning_rate": 4.9232444977163697e-05, "loss": 0.323, "step": 6509 }, { "epoch": 1.13, "learning_rate": 4.923067123979495e-05, "loss": 0.2831, "step": 6510 }, { "epoch": 1.13, "learning_rate": 4.922889549097898e-05, "loss": 0.2934, "step": 6511 }, { "epoch": 1.14, "learning_rate": 4.9227117730880154e-05, "loss": 0.2998, "step": 6512 }, { "epoch": 1.14, "learning_rate": 4.922533795966304e-05, "loss": 0.3097, "step": 6513 }, { "epoch": 1.14, "learning_rate": 4.9223556177492346e-05, "loss": 0.3136, "step": 6514 }, { "epoch": 1.14, "learning_rate": 4.922177238453303e-05, "loss": 0.3041, "step": 6515 }, { "epoch": 1.14, "learning_rate": 4.9219986580950186e-05, "loss": 0.2958, "step": 6516 }, { "epoch": 1.14, "learning_rate": 4.9218198766909116e-05, "loss": 0.2878, "step": 6517 }, { "epoch": 1.14, "learning_rate": 4.92164089425753e-05, "loss": 0.3017, "step": 6518 }, { "epoch": 1.14, "learning_rate": 4.921461710811443e-05, "loss": 0.3056, "step": 6519 }, { "epoch": 1.14, "learning_rate": 4.921282326369234e-05, "loss": 0.292, "step": 6520 }, { "epoch": 1.14, "learning_rate": 4.921102740947509e-05, "loss": 0.3213, "step": 6521 }, { "epoch": 1.14, "learning_rate": 4.9209229545628886e-05, "loss": 0.2984, "step": 6522 }, { "epoch": 1.14, "learning_rate": 4.920742967232016e-05, "loss": 0.313, "step": 6523 }, { "epoch": 1.14, "learning_rate": 4.920562778971552e-05, "loss": 0.2955, "step": 6524 }, { "epoch": 1.14, "learning_rate": 4.9203823897981736e-05, "loss": 0.3005, "step": 6525 }, { "epoch": 1.14, "learning_rate": 4.9202017997285785e-05, "loss": 0.3088, "step": 6526 }, { "epoch": 1.14, "learning_rate": 4.920021008779483e-05, "loss": 0.2908, "step": 6527 }, { "epoch": 1.14, "learning_rate": 4.919840016967622e-05, "loss": 0.2993, "step": 6528 }, { "epoch": 1.14, "learning_rate": 4.9196588243097476e-05, "loss": 0.3161, "step": 6529 }, { "epoch": 1.14, "learning_rate": 4.919477430822632e-05, "loss": 0.3017, "step": 6530 }, { "epoch": 1.14, "learning_rate": 4.9192958365230646e-05, "loss": 0.3109, "step": 6531 }, { "epoch": 1.14, "learning_rate": 4.919114041427855e-05, "loss": 0.3251, "step": 6532 }, { "epoch": 1.14, "learning_rate": 4.918932045553831e-05, "loss": 0.3269, "step": 6533 }, { "epoch": 1.14, "learning_rate": 4.9187498489178384e-05, "loss": 0.3166, "step": 6534 }, { "epoch": 1.14, "learning_rate": 4.918567451536742e-05, "loss": 0.3224, "step": 6535 }, { "epoch": 1.14, "learning_rate": 4.918384853427424e-05, "loss": 0.3069, "step": 6536 }, { "epoch": 1.14, "learning_rate": 4.9182020546067856e-05, "loss": 0.2996, "step": 6537 }, { "epoch": 1.14, "learning_rate": 4.91801905509175e-05, "loss": 0.3134, "step": 6538 }, { "epoch": 1.14, "learning_rate": 4.9178358548992535e-05, "loss": 0.3081, "step": 6539 }, { "epoch": 1.14, "learning_rate": 4.917652454046254e-05, "loss": 0.294, "step": 6540 }, { "epoch": 1.14, "learning_rate": 4.917468852549729e-05, "loss": 0.3096, "step": 6541 }, { "epoch": 1.14, "learning_rate": 4.9172850504266714e-05, "loss": 0.2954, "step": 6542 }, { "epoch": 1.14, "learning_rate": 4.917101047694096e-05, "loss": 0.298, "step": 6543 }, { "epoch": 1.14, "learning_rate": 4.916916844369033e-05, "loss": 0.303, "step": 6544 }, { "epoch": 1.14, "learning_rate": 4.916732440468533e-05, "loss": 0.3207, "step": 6545 }, { "epoch": 1.14, "learning_rate": 4.916547836009666e-05, "loss": 0.2957, "step": 6546 }, { "epoch": 1.14, "learning_rate": 4.916363031009519e-05, "loss": 0.3096, "step": 6547 }, { "epoch": 1.14, "learning_rate": 4.9161780254851966e-05, "loss": 0.3093, "step": 6548 }, { "epoch": 1.14, "learning_rate": 4.915992819453826e-05, "loss": 0.3041, "step": 6549 }, { "epoch": 1.14, "learning_rate": 4.915807412932548e-05, "loss": 0.2998, "step": 6550 }, { "epoch": 1.14, "learning_rate": 4.915621805938526e-05, "loss": 0.31, "step": 6551 }, { "epoch": 1.14, "learning_rate": 4.9154359984889395e-05, "loss": 0.3097, "step": 6552 }, { "epoch": 1.14, "learning_rate": 4.915249990600987e-05, "loss": 0.3029, "step": 6553 }, { "epoch": 1.14, "learning_rate": 4.915063782291886e-05, "loss": 0.3011, "step": 6554 }, { "epoch": 1.14, "learning_rate": 4.9148773735788735e-05, "loss": 0.3121, "step": 6555 }, { "epoch": 1.14, "learning_rate": 4.914690764479202e-05, "loss": 0.3106, "step": 6556 }, { "epoch": 1.14, "learning_rate": 4.914503955010146e-05, "loss": 0.3082, "step": 6557 }, { "epoch": 1.14, "learning_rate": 4.914316945188997e-05, "loss": 0.301, "step": 6558 }, { "epoch": 1.14, "learning_rate": 4.9141297350330654e-05, "loss": 0.3159, "step": 6559 }, { "epoch": 1.14, "learning_rate": 4.913942324559679e-05, "loss": 0.3004, "step": 6560 }, { "epoch": 1.14, "learning_rate": 4.9137547137861846e-05, "loss": 0.3014, "step": 6561 }, { "epoch": 1.14, "learning_rate": 4.913566902729949e-05, "loss": 0.3008, "step": 6562 }, { "epoch": 1.14, "learning_rate": 4.9133788914083555e-05, "loss": 0.2942, "step": 6563 }, { "epoch": 1.14, "learning_rate": 4.913190679838808e-05, "loss": 0.3303, "step": 6564 }, { "epoch": 1.14, "learning_rate": 4.9130022680387284e-05, "loss": 0.2914, "step": 6565 }, { "epoch": 1.14, "learning_rate": 4.912813656025554e-05, "loss": 0.3114, "step": 6566 }, { "epoch": 1.14, "learning_rate": 4.912624843816746e-05, "loss": 0.2993, "step": 6567 }, { "epoch": 1.14, "learning_rate": 4.9124358314297794e-05, "loss": 0.306, "step": 6568 }, { "epoch": 1.14, "learning_rate": 4.91224661888215e-05, "loss": 0.3183, "step": 6569 }, { "epoch": 1.15, "learning_rate": 4.912057206191373e-05, "loss": 0.3177, "step": 6570 }, { "epoch": 1.15, "learning_rate": 4.91186759337498e-05, "loss": 0.3068, "step": 6571 }, { "epoch": 1.15, "learning_rate": 4.911677780450522e-05, "loss": 0.3001, "step": 6572 }, { "epoch": 1.15, "learning_rate": 4.911487767435569e-05, "loss": 0.308, "step": 6573 }, { "epoch": 1.15, "learning_rate": 4.911297554347708e-05, "loss": 0.3175, "step": 6574 }, { "epoch": 1.15, "learning_rate": 4.911107141204547e-05, "loss": 0.299, "step": 6575 }, { "epoch": 1.15, "learning_rate": 4.910916528023711e-05, "loss": 0.3201, "step": 6576 }, { "epoch": 1.15, "learning_rate": 4.9107257148228434e-05, "loss": 0.3254, "step": 6577 }, { "epoch": 1.15, "learning_rate": 4.910534701619605e-05, "loss": 0.2978, "step": 6578 }, { "epoch": 1.15, "learning_rate": 4.910343488431679e-05, "loss": 0.2992, "step": 6579 }, { "epoch": 1.15, "learning_rate": 4.9101520752767625e-05, "loss": 0.3078, "step": 6580 }, { "epoch": 1.15, "learning_rate": 4.9099604621725745e-05, "loss": 0.3058, "step": 6581 }, { "epoch": 1.15, "learning_rate": 4.90976864913685e-05, "loss": 0.3172, "step": 6582 }, { "epoch": 1.15, "learning_rate": 4.9095766361873446e-05, "loss": 0.2983, "step": 6583 }, { "epoch": 1.15, "learning_rate": 4.9093844233418314e-05, "loss": 0.3046, "step": 6584 }, { "epoch": 1.15, "learning_rate": 4.909192010618102e-05, "loss": 0.2931, "step": 6585 }, { "epoch": 1.15, "learning_rate": 4.9089993980339663e-05, "loss": 0.3129, "step": 6586 }, { "epoch": 1.15, "learning_rate": 4.908806585607254e-05, "loss": 0.309, "step": 6587 }, { "epoch": 1.15, "learning_rate": 4.908613573355812e-05, "loss": 0.2887, "step": 6588 }, { "epoch": 1.15, "learning_rate": 4.908420361297505e-05, "loss": 0.3125, "step": 6589 }, { "epoch": 1.15, "learning_rate": 4.9082269494502176e-05, "loss": 0.3088, "step": 6590 }, { "epoch": 1.15, "learning_rate": 4.908033337831855e-05, "loss": 0.343, "step": 6591 }, { "epoch": 1.15, "learning_rate": 4.9078395264603335e-05, "loss": 0.3029, "step": 6592 }, { "epoch": 1.15, "learning_rate": 4.9076455153535964e-05, "loss": 0.3043, "step": 6593 }, { "epoch": 1.15, "learning_rate": 4.9074513045296e-05, "loss": 0.3016, "step": 6594 }, { "epoch": 1.15, "learning_rate": 4.907256894006324e-05, "loss": 0.3153, "step": 6595 }, { "epoch": 1.15, "learning_rate": 4.907062283801761e-05, "loss": 0.2952, "step": 6596 }, { "epoch": 1.15, "learning_rate": 4.9068674739339236e-05, "loss": 0.311, "step": 6597 }, { "epoch": 1.15, "learning_rate": 4.906672464420847e-05, "loss": 0.2924, "step": 6598 }, { "epoch": 1.15, "learning_rate": 4.9064772552805795e-05, "loss": 0.2987, "step": 6599 }, { "epoch": 1.15, "learning_rate": 4.906281846531191e-05, "loss": 0.2926, "step": 6600 }, { "epoch": 1.15, "learning_rate": 4.906086238190769e-05, "loss": 0.3103, "step": 6601 }, { "epoch": 1.15, "learning_rate": 4.905890430277418e-05, "loss": 0.319, "step": 6602 }, { "epoch": 1.15, "learning_rate": 4.905694422809265e-05, "loss": 0.3076, "step": 6603 }, { "epoch": 1.15, "learning_rate": 4.905498215804452e-05, "loss": 0.3069, "step": 6604 }, { "epoch": 1.15, "learning_rate": 4.90530180928114e-05, "loss": 0.2989, "step": 6605 }, { "epoch": 1.15, "learning_rate": 4.905105203257509e-05, "loss": 0.3109, "step": 6606 }, { "epoch": 1.15, "learning_rate": 4.904908397751758e-05, "loss": 0.3214, "step": 6607 }, { "epoch": 1.15, "learning_rate": 4.9047113927821034e-05, "loss": 0.2989, "step": 6608 }, { "epoch": 1.15, "learning_rate": 4.9045141883667794e-05, "loss": 0.3034, "step": 6609 }, { "epoch": 1.15, "learning_rate": 4.904316784524041e-05, "loss": 0.3141, "step": 6610 }, { "epoch": 1.15, "learning_rate": 4.904119181272162e-05, "loss": 0.3157, "step": 6611 }, { "epoch": 1.15, "learning_rate": 4.9039213786294293e-05, "loss": 0.3023, "step": 6612 }, { "epoch": 1.15, "learning_rate": 4.9037233766141554e-05, "loss": 0.3048, "step": 6613 }, { "epoch": 1.15, "learning_rate": 4.9035251752446656e-05, "loss": 0.3048, "step": 6614 }, { "epoch": 1.15, "learning_rate": 4.9033267745393066e-05, "loss": 0.2876, "step": 6615 }, { "epoch": 1.15, "learning_rate": 4.9031281745164436e-05, "loss": 0.3063, "step": 6616 }, { "epoch": 1.15, "learning_rate": 4.902929375194459e-05, "loss": 0.3167, "step": 6617 }, { "epoch": 1.15, "learning_rate": 4.902730376591754e-05, "loss": 0.3127, "step": 6618 }, { "epoch": 1.15, "learning_rate": 4.9025311787267485e-05, "loss": 0.3055, "step": 6619 }, { "epoch": 1.15, "learning_rate": 4.902331781617882e-05, "loss": 0.3045, "step": 6620 }, { "epoch": 1.15, "learning_rate": 4.902132185283609e-05, "loss": 0.2913, "step": 6621 }, { "epoch": 1.15, "learning_rate": 4.901932389742406e-05, "loss": 0.31, "step": 6622 }, { "epoch": 1.15, "learning_rate": 4.901732395012766e-05, "loss": 0.312, "step": 6623 }, { "epoch": 1.15, "learning_rate": 4.9015322011132024e-05, "loss": 0.2912, "step": 6624 }, { "epoch": 1.15, "learning_rate": 4.901331808062244e-05, "loss": 0.3141, "step": 6625 }, { "epoch": 1.15, "learning_rate": 4.90113121587844e-05, "loss": 0.3037, "step": 6626 }, { "epoch": 1.16, "learning_rate": 4.9009304245803584e-05, "loss": 0.3182, "step": 6627 }, { "epoch": 1.16, "learning_rate": 4.900729434186585e-05, "loss": 0.2975, "step": 6628 }, { "epoch": 1.16, "learning_rate": 4.900528244715723e-05, "loss": 0.3009, "step": 6629 }, { "epoch": 1.16, "learning_rate": 4.900326856186396e-05, "loss": 0.2978, "step": 6630 }, { "epoch": 1.16, "learning_rate": 4.9001252686172456e-05, "loss": 0.3119, "step": 6631 }, { "epoch": 1.16, "learning_rate": 4.899923482026929e-05, "loss": 0.3147, "step": 6632 }, { "epoch": 1.16, "learning_rate": 4.899721496434126e-05, "loss": 0.297, "step": 6633 }, { "epoch": 1.16, "learning_rate": 4.8995193118575325e-05, "loss": 0.3136, "step": 6634 }, { "epoch": 1.16, "learning_rate": 4.899316928315863e-05, "loss": 0.3111, "step": 6635 }, { "epoch": 1.16, "learning_rate": 4.899114345827851e-05, "loss": 0.3159, "step": 6636 }, { "epoch": 1.16, "learning_rate": 4.8989115644122474e-05, "loss": 0.3085, "step": 6637 }, { "epoch": 1.16, "learning_rate": 4.8987085840878226e-05, "loss": 0.3119, "step": 6638 }, { "epoch": 1.16, "learning_rate": 4.898505404873366e-05, "loss": 0.2967, "step": 6639 }, { "epoch": 1.16, "learning_rate": 4.898302026787682e-05, "loss": 0.3059, "step": 6640 }, { "epoch": 1.16, "learning_rate": 4.898098449849597e-05, "loss": 0.3058, "step": 6641 }, { "epoch": 1.16, "learning_rate": 4.897894674077956e-05, "loss": 0.2978, "step": 6642 }, { "epoch": 1.16, "learning_rate": 4.897690699491619e-05, "loss": 0.3144, "step": 6643 }, { "epoch": 1.16, "learning_rate": 4.8974865261094676e-05, "loss": 0.3008, "step": 6644 }, { "epoch": 1.16, "learning_rate": 4.8972821539504e-05, "loss": 0.3115, "step": 6645 }, { "epoch": 1.16, "learning_rate": 4.897077583033334e-05, "loss": 0.3167, "step": 6646 }, { "epoch": 1.16, "learning_rate": 4.896872813377204e-05, "loss": 0.3081, "step": 6647 }, { "epoch": 1.16, "learning_rate": 4.896667845000966e-05, "loss": 0.3065, "step": 6648 }, { "epoch": 1.16, "learning_rate": 4.89646267792359e-05, "loss": 0.3055, "step": 6649 }, { "epoch": 1.16, "learning_rate": 4.8962573121640684e-05, "loss": 0.3102, "step": 6650 }, { "epoch": 1.16, "learning_rate": 4.896051747741411e-05, "loss": 0.3019, "step": 6651 }, { "epoch": 1.16, "learning_rate": 4.895845984674643e-05, "loss": 0.3043, "step": 6652 }, { "epoch": 1.16, "learning_rate": 4.8956400229828124e-05, "loss": 0.311, "step": 6653 }, { "epoch": 1.16, "learning_rate": 4.895433862684983e-05, "loss": 0.3005, "step": 6654 }, { "epoch": 1.16, "learning_rate": 4.8952275038002384e-05, "loss": 0.3138, "step": 6655 }, { "epoch": 1.16, "learning_rate": 4.895020946347677e-05, "loss": 0.314, "step": 6656 }, { "epoch": 1.16, "learning_rate": 4.894814190346421e-05, "loss": 0.2989, "step": 6657 }, { "epoch": 1.16, "learning_rate": 4.894607235815608e-05, "loss": 0.3084, "step": 6658 }, { "epoch": 1.16, "learning_rate": 4.8944000827743924e-05, "loss": 0.2904, "step": 6659 }, { "epoch": 1.16, "learning_rate": 4.894192731241951e-05, "loss": 0.3095, "step": 6660 }, { "epoch": 1.16, "learning_rate": 4.893985181237475e-05, "loss": 0.3083, "step": 6661 }, { "epoch": 1.16, "learning_rate": 4.893777432780177e-05, "loss": 0.2949, "step": 6662 }, { "epoch": 1.16, "learning_rate": 4.893569485889287e-05, "loss": 0.3141, "step": 6663 }, { "epoch": 1.16, "learning_rate": 4.8933613405840506e-05, "loss": 0.3072, "step": 6664 }, { "epoch": 1.16, "learning_rate": 4.893152996883737e-05, "loss": 0.2946, "step": 6665 }, { "epoch": 1.16, "learning_rate": 4.89294445480763e-05, "loss": 0.2921, "step": 6666 }, { "epoch": 1.16, "learning_rate": 4.892735714375034e-05, "loss": 0.2998, "step": 6667 }, { "epoch": 1.16, "learning_rate": 4.8925267756052685e-05, "loss": 0.2902, "step": 6668 }, { "epoch": 1.16, "learning_rate": 4.892317638517674e-05, "loss": 0.3161, "step": 6669 }, { "epoch": 1.16, "learning_rate": 4.89210830313161e-05, "loss": 0.3071, "step": 6670 }, { "epoch": 1.16, "learning_rate": 4.8918987694664515e-05, "loss": 0.3135, "step": 6671 }, { "epoch": 1.16, "learning_rate": 4.8916890375415944e-05, "loss": 0.3028, "step": 6672 }, { "epoch": 1.16, "learning_rate": 4.891479107376452e-05, "loss": 0.2989, "step": 6673 }, { "epoch": 1.16, "learning_rate": 4.8912689789904566e-05, "loss": 0.313, "step": 6674 }, { "epoch": 1.16, "learning_rate": 4.891058652403056e-05, "loss": 0.2955, "step": 6675 }, { "epoch": 1.16, "learning_rate": 4.890848127633721e-05, "loss": 0.3037, "step": 6676 }, { "epoch": 1.16, "learning_rate": 4.8906374047019374e-05, "loss": 0.3119, "step": 6677 }, { "epoch": 1.16, "learning_rate": 4.89042648362721e-05, "loss": 0.2989, "step": 6678 }, { "epoch": 1.16, "learning_rate": 4.890215364429062e-05, "loss": 0.3097, "step": 6679 }, { "epoch": 1.16, "learning_rate": 4.890004047127037e-05, "loss": 0.3122, "step": 6680 }, { "epoch": 1.16, "learning_rate": 4.889792531740693e-05, "loss": 0.3041, "step": 6681 }, { "epoch": 1.16, "learning_rate": 4.8895808182896086e-05, "loss": 0.2964, "step": 6682 }, { "epoch": 1.16, "learning_rate": 4.889368906793381e-05, "loss": 0.302, "step": 6683 }, { "epoch": 1.17, "learning_rate": 4.889156797271626e-05, "loss": 0.3076, "step": 6684 }, { "epoch": 1.17, "learning_rate": 4.8889444897439756e-05, "loss": 0.3269, "step": 6685 }, { "epoch": 1.17, "learning_rate": 4.888731984230082e-05, "loss": 0.3083, "step": 6686 }, { "epoch": 1.17, "learning_rate": 4.888519280749616e-05, "loss": 0.3176, "step": 6687 }, { "epoch": 1.17, "learning_rate": 4.888306379322266e-05, "loss": 0.304, "step": 6688 }, { "epoch": 1.17, "learning_rate": 4.888093279967737e-05, "loss": 0.3182, "step": 6689 }, { "epoch": 1.17, "learning_rate": 4.887879982705756e-05, "loss": 0.3142, "step": 6690 }, { "epoch": 1.17, "learning_rate": 4.887666487556065e-05, "loss": 0.3083, "step": 6691 }, { "epoch": 1.17, "learning_rate": 4.887452794538427e-05, "loss": 0.3202, "step": 6692 }, { "epoch": 1.17, "learning_rate": 4.887238903672621e-05, "loss": 0.3336, "step": 6693 }, { "epoch": 1.17, "learning_rate": 4.887024814978445e-05, "loss": 0.3111, "step": 6694 }, { "epoch": 1.17, "learning_rate": 4.8868105284757163e-05, "loss": 0.2931, "step": 6695 }, { "epoch": 1.17, "learning_rate": 4.8865960441842695e-05, "loss": 0.304, "step": 6696 }, { "epoch": 1.17, "learning_rate": 4.886381362123959e-05, "loss": 0.3039, "step": 6697 }, { "epoch": 1.17, "learning_rate": 4.886166482314654e-05, "loss": 0.3078, "step": 6698 }, { "epoch": 1.17, "learning_rate": 4.885951404776246e-05, "loss": 0.3103, "step": 6699 }, { "epoch": 1.17, "learning_rate": 4.885736129528642e-05, "loss": 0.3004, "step": 6700 }, { "epoch": 1.17, "learning_rate": 4.8855206565917704e-05, "loss": 0.2997, "step": 6701 }, { "epoch": 1.17, "learning_rate": 4.885304985985575e-05, "loss": 0.2869, "step": 6702 }, { "epoch": 1.17, "learning_rate": 4.8850891177300176e-05, "loss": 0.3138, "step": 6703 }, { "epoch": 1.17, "learning_rate": 4.88487305184508e-05, "loss": 0.3102, "step": 6704 }, { "epoch": 1.17, "learning_rate": 4.884656788350763e-05, "loss": 0.3062, "step": 6705 }, { "epoch": 1.17, "learning_rate": 4.884440327267084e-05, "loss": 0.3098, "step": 6706 }, { "epoch": 1.17, "learning_rate": 4.8842236686140784e-05, "loss": 0.2915, "step": 6707 }, { "epoch": 1.17, "learning_rate": 4.8840068124118014e-05, "loss": 0.3065, "step": 6708 }, { "epoch": 1.17, "learning_rate": 4.883789758680326e-05, "loss": 0.3237, "step": 6709 }, { "epoch": 1.17, "learning_rate": 4.8835725074397424e-05, "loss": 0.3142, "step": 6710 }, { "epoch": 1.17, "learning_rate": 4.883355058710161e-05, "loss": 0.3148, "step": 6711 }, { "epoch": 1.17, "learning_rate": 4.8831374125117083e-05, "loss": 0.313, "step": 6712 }, { "epoch": 1.17, "learning_rate": 4.882919568864531e-05, "loss": 0.3096, "step": 6713 }, { "epoch": 1.17, "learning_rate": 4.882701527788794e-05, "loss": 0.3066, "step": 6714 }, { "epoch": 1.17, "learning_rate": 4.8824832893046773e-05, "loss": 0.3151, "step": 6715 }, { "epoch": 1.17, "learning_rate": 4.8822648534323833e-05, "loss": 0.2963, "step": 6716 }, { "epoch": 1.17, "learning_rate": 4.882046220192131e-05, "loss": 0.3239, "step": 6717 }, { "epoch": 1.17, "learning_rate": 4.8818273896041575e-05, "loss": 0.3113, "step": 6718 }, { "epoch": 1.17, "learning_rate": 4.881608361688718e-05, "loss": 0.2955, "step": 6719 }, { "epoch": 1.17, "learning_rate": 4.8813891364660876e-05, "loss": 0.293, "step": 6720 }, { "epoch": 1.17, "learning_rate": 4.881169713956556e-05, "loss": 0.315, "step": 6721 }, { "epoch": 1.17, "learning_rate": 4.8809500941804354e-05, "loss": 0.2906, "step": 6722 }, { "epoch": 1.17, "learning_rate": 4.8807302771580535e-05, "loss": 0.2938, "step": 6723 }, { "epoch": 1.17, "learning_rate": 4.8805102629097574e-05, "loss": 0.2861, "step": 6724 }, { "epoch": 1.17, "learning_rate": 4.880290051455913e-05, "loss": 0.2979, "step": 6725 }, { "epoch": 1.17, "learning_rate": 4.880069642816902e-05, "loss": 0.304, "step": 6726 }, { "epoch": 1.17, "learning_rate": 4.879849037013127e-05, "loss": 0.3167, "step": 6727 }, { "epoch": 1.17, "learning_rate": 4.879628234065008e-05, "loss": 0.3004, "step": 6728 }, { "epoch": 1.17, "learning_rate": 4.879407233992982e-05, "loss": 0.3059, "step": 6729 }, { "epoch": 1.17, "learning_rate": 4.879186036817507e-05, "loss": 0.322, "step": 6730 }, { "epoch": 1.17, "learning_rate": 4.878964642559057e-05, "loss": 0.2997, "step": 6731 }, { "epoch": 1.17, "learning_rate": 4.878743051238123e-05, "loss": 0.3058, "step": 6732 }, { "epoch": 1.17, "learning_rate": 4.878521262875219e-05, "loss": 0.311, "step": 6733 }, { "epoch": 1.17, "learning_rate": 4.8782992774908724e-05, "loss": 0.3036, "step": 6734 }, { "epoch": 1.17, "learning_rate": 4.878077095105632e-05, "loss": 0.3004, "step": 6735 }, { "epoch": 1.17, "learning_rate": 4.877854715740062e-05, "loss": 0.3128, "step": 6736 }, { "epoch": 1.17, "learning_rate": 4.877632139414747e-05, "loss": 0.3229, "step": 6737 }, { "epoch": 1.17, "learning_rate": 4.877409366150291e-05, "loss": 0.3004, "step": 6738 }, { "epoch": 1.17, "learning_rate": 4.877186395967312e-05, "loss": 0.3254, "step": 6739 }, { "epoch": 1.17, "learning_rate": 4.876963228886449e-05, "loss": 0.3094, "step": 6740 }, { "epoch": 1.17, "learning_rate": 4.87673986492836e-05, "loss": 0.318, "step": 6741 }, { "epoch": 1.18, "learning_rate": 4.8765163041137196e-05, "loss": 0.3115, "step": 6742 }, { "epoch": 1.18, "learning_rate": 4.876292546463222e-05, "loss": 0.308, "step": 6743 }, { "epoch": 1.18, "learning_rate": 4.8760685919975773e-05, "loss": 0.316, "step": 6744 }, { "epoch": 1.18, "learning_rate": 4.875844440737517e-05, "loss": 0.2985, "step": 6745 }, { "epoch": 1.18, "learning_rate": 4.875620092703788e-05, "loss": 0.3112, "step": 6746 }, { "epoch": 1.18, "learning_rate": 4.875395547917156e-05, "loss": 0.3191, "step": 6747 }, { "epoch": 1.18, "learning_rate": 4.875170806398406e-05, "loss": 0.2994, "step": 6748 }, { "epoch": 1.18, "learning_rate": 4.874945868168341e-05, "loss": 0.303, "step": 6749 }, { "epoch": 1.18, "learning_rate": 4.8747207332477825e-05, "loss": 0.3048, "step": 6750 }, { "epoch": 1.18, "learning_rate": 4.8744954016575685e-05, "loss": 0.3249, "step": 6751 }, { "epoch": 1.18, "learning_rate": 4.8742698734185564e-05, "loss": 0.3143, "step": 6752 }, { "epoch": 1.18, "learning_rate": 4.874044148551621e-05, "loss": 0.3015, "step": 6753 }, { "epoch": 1.18, "learning_rate": 4.873818227077658e-05, "loss": 0.3072, "step": 6754 }, { "epoch": 1.18, "learning_rate": 4.8735921090175784e-05, "loss": 0.3168, "step": 6755 }, { "epoch": 1.18, "learning_rate": 4.873365794392311e-05, "loss": 0.304, "step": 6756 }, { "epoch": 1.18, "learning_rate": 4.8731392832228055e-05, "loss": 0.2984, "step": 6757 }, { "epoch": 1.18, "learning_rate": 4.8729125755300284e-05, "loss": 0.3083, "step": 6758 }, { "epoch": 1.18, "learning_rate": 4.8726856713349636e-05, "loss": 0.3074, "step": 6759 }, { "epoch": 1.18, "learning_rate": 4.8724585706586126e-05, "loss": 0.3214, "step": 6760 }, { "epoch": 1.18, "learning_rate": 4.8722312735219996e-05, "loss": 0.2936, "step": 6761 }, { "epoch": 1.18, "learning_rate": 4.872003779946162e-05, "loss": 0.2995, "step": 6762 }, { "epoch": 1.18, "learning_rate": 4.871776089952157e-05, "loss": 0.3143, "step": 6763 }, { "epoch": 1.18, "learning_rate": 4.871548203561061e-05, "loss": 0.3118, "step": 6764 }, { "epoch": 1.18, "learning_rate": 4.871320120793967e-05, "loss": 0.3098, "step": 6765 }, { "epoch": 1.18, "learning_rate": 4.871091841671987e-05, "loss": 0.3119, "step": 6766 }, { "epoch": 1.18, "learning_rate": 4.8708633662162525e-05, "loss": 0.3117, "step": 6767 }, { "epoch": 1.18, "learning_rate": 4.87063469444791e-05, "loss": 0.3109, "step": 6768 }, { "epoch": 1.18, "learning_rate": 4.870405826388126e-05, "loss": 0.3288, "step": 6769 }, { "epoch": 1.18, "learning_rate": 4.870176762058086e-05, "loss": 0.2867, "step": 6770 }, { "epoch": 1.18, "learning_rate": 4.8699475014789936e-05, "loss": 0.3078, "step": 6771 }, { "epoch": 1.18, "learning_rate": 4.869718044672068e-05, "loss": 0.3109, "step": 6772 }, { "epoch": 1.18, "learning_rate": 4.8694883916585486e-05, "loss": 0.3066, "step": 6773 }, { "epoch": 1.18, "learning_rate": 4.8692585424596936e-05, "loss": 0.3109, "step": 6774 }, { "epoch": 1.18, "learning_rate": 4.869028497096778e-05, "loss": 0.2986, "step": 6775 }, { "epoch": 1.18, "learning_rate": 4.868798255591095e-05, "loss": 0.3136, "step": 6776 }, { "epoch": 1.18, "learning_rate": 4.868567817963957e-05, "loss": 0.3254, "step": 6777 }, { "epoch": 1.18, "learning_rate": 4.868337184236693e-05, "loss": 0.3008, "step": 6778 }, { "epoch": 1.18, "learning_rate": 4.868106354430652e-05, "loss": 0.3098, "step": 6779 }, { "epoch": 1.18, "learning_rate": 4.8678753285671996e-05, "loss": 0.3092, "step": 6780 }, { "epoch": 1.18, "learning_rate": 4.8676441066677206e-05, "loss": 0.303, "step": 6781 }, { "epoch": 1.18, "learning_rate": 4.867412688753617e-05, "loss": 0.2951, "step": 6782 }, { "epoch": 1.18, "learning_rate": 4.86718107484631e-05, "loss": 0.3031, "step": 6783 }, { "epoch": 1.18, "learning_rate": 4.8669492649672384e-05, "loss": 0.3097, "step": 6784 }, { "epoch": 1.18, "learning_rate": 4.866717259137858e-05, "loss": 0.3217, "step": 6785 }, { "epoch": 1.18, "learning_rate": 4.866485057379645e-05, "loss": 0.3077, "step": 6786 }, { "epoch": 1.18, "learning_rate": 4.866252659714092e-05, "loss": 0.2946, "step": 6787 }, { "epoch": 1.18, "learning_rate": 4.866020066162711e-05, "loss": 0.3036, "step": 6788 }, { "epoch": 1.18, "learning_rate": 4.865787276747031e-05, "loss": 0.307, "step": 6789 }, { "epoch": 1.18, "learning_rate": 4.8655542914885995e-05, "loss": 0.3185, "step": 6790 }, { "epoch": 1.18, "learning_rate": 4.865321110408983e-05, "loss": 0.301, "step": 6791 }, { "epoch": 1.18, "learning_rate": 4.865087733529764e-05, "loss": 0.3102, "step": 6792 }, { "epoch": 1.18, "learning_rate": 4.864854160872546e-05, "loss": 0.2872, "step": 6793 }, { "epoch": 1.18, "learning_rate": 4.864620392458947e-05, "loss": 0.3059, "step": 6794 }, { "epoch": 1.18, "learning_rate": 4.864386428310607e-05, "loss": 0.3132, "step": 6795 }, { "epoch": 1.18, "learning_rate": 4.8641522684491816e-05, "loss": 0.307, "step": 6796 }, { "epoch": 1.18, "learning_rate": 4.8639179128963466e-05, "loss": 0.3084, "step": 6797 }, { "epoch": 1.18, "learning_rate": 4.863683361673793e-05, "loss": 0.2991, "step": 6798 }, { "epoch": 1.19, "learning_rate": 4.8634486148032314e-05, "loss": 0.3054, "step": 6799 }, { "epoch": 1.19, "learning_rate": 4.863213672306391e-05, "loss": 0.3148, "step": 6800 }, { "epoch": 1.19, "learning_rate": 4.862978534205019e-05, "loss": 0.3057, "step": 6801 }, { "epoch": 1.19, "learning_rate": 4.862743200520881e-05, "loss": 0.2961, "step": 6802 }, { "epoch": 1.19, "learning_rate": 4.8625076712757575e-05, "loss": 0.3056, "step": 6803 }, { "epoch": 1.19, "learning_rate": 4.862271946491452e-05, "loss": 0.2991, "step": 6804 }, { "epoch": 1.19, "learning_rate": 4.862036026189784e-05, "loss": 0.3046, "step": 6805 }, { "epoch": 1.19, "learning_rate": 4.86179991039259e-05, "loss": 0.327, "step": 6806 }, { "epoch": 1.19, "learning_rate": 4.861563599121725e-05, "loss": 0.3058, "step": 6807 }, { "epoch": 1.19, "learning_rate": 4.8613270923990636e-05, "loss": 0.3, "step": 6808 }, { "epoch": 1.19, "learning_rate": 4.8610903902464966e-05, "loss": 0.3097, "step": 6809 }, { "epoch": 1.19, "learning_rate": 4.860853492685935e-05, "loss": 0.2977, "step": 6810 }, { "epoch": 1.19, "learning_rate": 4.860616399739305e-05, "loss": 0.3233, "step": 6811 }, { "epoch": 1.19, "learning_rate": 4.860379111428555e-05, "loss": 0.2997, "step": 6812 }, { "epoch": 1.19, "learning_rate": 4.860141627775646e-05, "loss": 0.3104, "step": 6813 }, { "epoch": 1.19, "learning_rate": 4.8599039488025615e-05, "loss": 0.3231, "step": 6814 }, { "epoch": 1.19, "learning_rate": 4.8596660745313024e-05, "loss": 0.3006, "step": 6815 }, { "epoch": 1.19, "learning_rate": 4.8594280049838865e-05, "loss": 0.3011, "step": 6816 }, { "epoch": 1.19, "learning_rate": 4.859189740182349e-05, "loss": 0.3052, "step": 6817 }, { "epoch": 1.19, "learning_rate": 4.858951280148745e-05, "loss": 0.3139, "step": 6818 }, { "epoch": 1.19, "learning_rate": 4.8587126249051484e-05, "loss": 0.3156, "step": 6819 }, { "epoch": 1.19, "learning_rate": 4.858473774473648e-05, "loss": 0.3027, "step": 6820 }, { "epoch": 1.19, "learning_rate": 4.858234728876353e-05, "loss": 0.311, "step": 6821 }, { "epoch": 1.19, "learning_rate": 4.85799548813539e-05, "loss": 0.3078, "step": 6822 }, { "epoch": 1.19, "learning_rate": 4.857756052272904e-05, "loss": 0.2962, "step": 6823 }, { "epoch": 1.19, "learning_rate": 4.857516421311058e-05, "loss": 0.2994, "step": 6824 }, { "epoch": 1.19, "learning_rate": 4.857276595272032e-05, "loss": 0.2974, "step": 6825 }, { "epoch": 1.19, "learning_rate": 4.857036574178025e-05, "loss": 0.3052, "step": 6826 }, { "epoch": 1.19, "learning_rate": 4.856796358051255e-05, "loss": 0.2956, "step": 6827 }, { "epoch": 1.19, "learning_rate": 4.856555946913956e-05, "loss": 0.3067, "step": 6828 }, { "epoch": 1.19, "learning_rate": 4.856315340788382e-05, "loss": 0.3136, "step": 6829 }, { "epoch": 1.19, "learning_rate": 4.856074539696803e-05, "loss": 0.3032, "step": 6830 }, { "epoch": 1.19, "learning_rate": 4.855833543661508e-05, "loss": 0.2996, "step": 6831 }, { "epoch": 1.19, "learning_rate": 4.855592352704806e-05, "loss": 0.3292, "step": 6832 }, { "epoch": 1.19, "learning_rate": 4.855350966849021e-05, "loss": 0.3069, "step": 6833 }, { "epoch": 1.19, "learning_rate": 4.855109386116497e-05, "loss": 0.3124, "step": 6834 }, { "epoch": 1.19, "learning_rate": 4.8548676105295946e-05, "loss": 0.3093, "step": 6835 }, { "epoch": 1.19, "learning_rate": 4.8546256401106936e-05, "loss": 0.3232, "step": 6836 }, { "epoch": 1.19, "learning_rate": 4.8543834748821904e-05, "loss": 0.2998, "step": 6837 }, { "epoch": 1.19, "learning_rate": 4.854141114866502e-05, "loss": 0.2903, "step": 6838 }, { "epoch": 1.19, "learning_rate": 4.853898560086061e-05, "loss": 0.3274, "step": 6839 }, { "epoch": 1.19, "learning_rate": 4.85365581056332e-05, "loss": 0.3133, "step": 6840 }, { "epoch": 1.19, "learning_rate": 4.853412866320746e-05, "loss": 0.3164, "step": 6841 }, { "epoch": 1.19, "learning_rate": 4.853169727380829e-05, "loss": 0.3103, "step": 6842 }, { "epoch": 1.19, "learning_rate": 4.852926393766074e-05, "loss": 0.3056, "step": 6843 }, { "epoch": 1.19, "learning_rate": 4.8526828654990033e-05, "loss": 0.3287, "step": 6844 }, { "epoch": 1.19, "learning_rate": 4.8524391426021605e-05, "loss": 0.3252, "step": 6845 }, { "epoch": 1.19, "learning_rate": 4.852195225098104e-05, "loss": 0.3161, "step": 6846 }, { "epoch": 1.19, "learning_rate": 4.8519511130094115e-05, "loss": 0.3138, "step": 6847 }, { "epoch": 1.19, "learning_rate": 4.8517068063586786e-05, "loss": 0.3072, "step": 6848 }, { "epoch": 1.19, "learning_rate": 4.851462305168519e-05, "loss": 0.32, "step": 6849 }, { "epoch": 1.19, "learning_rate": 4.851217609461566e-05, "loss": 0.3016, "step": 6850 }, { "epoch": 1.19, "learning_rate": 4.850972719260466e-05, "loss": 0.3076, "step": 6851 }, { "epoch": 1.19, "learning_rate": 4.8507276345878894e-05, "loss": 0.3132, "step": 6852 }, { "epoch": 1.19, "learning_rate": 4.850482355466521e-05, "loss": 0.2925, "step": 6853 }, { "epoch": 1.19, "learning_rate": 4.850236881919064e-05, "loss": 0.3122, "step": 6854 }, { "epoch": 1.19, "learning_rate": 4.849991213968241e-05, "loss": 0.297, "step": 6855 }, { "epoch": 1.19, "learning_rate": 4.8497453516367906e-05, "loss": 0.3185, "step": 6856 }, { "epoch": 1.2, "learning_rate": 4.849499294947471e-05, "loss": 0.3027, "step": 6857 }, { "epoch": 1.2, "learning_rate": 4.849253043923058e-05, "loss": 0.3175, "step": 6858 }, { "epoch": 1.2, "learning_rate": 4.849006598586345e-05, "loss": 0.299, "step": 6859 }, { "epoch": 1.2, "learning_rate": 4.848759958960144e-05, "loss": 0.3122, "step": 6860 }, { "epoch": 1.2, "learning_rate": 4.848513125067285e-05, "loss": 0.3078, "step": 6861 }, { "epoch": 1.2, "learning_rate": 4.848266096930614e-05, "loss": 0.3062, "step": 6862 }, { "epoch": 1.2, "learning_rate": 4.848018874572997e-05, "loss": 0.3078, "step": 6863 }, { "epoch": 1.2, "learning_rate": 4.847771458017318e-05, "loss": 0.3013, "step": 6864 }, { "epoch": 1.2, "learning_rate": 4.84752384728648e-05, "loss": 0.3216, "step": 6865 }, { "epoch": 1.2, "learning_rate": 4.8472760424033996e-05, "loss": 0.3091, "step": 6866 }, { "epoch": 1.2, "learning_rate": 4.847028043391016e-05, "loss": 0.3001, "step": 6867 }, { "epoch": 1.2, "learning_rate": 4.846779850272284e-05, "loss": 0.3059, "step": 6868 }, { "epoch": 1.2, "learning_rate": 4.846531463070178e-05, "loss": 0.2999, "step": 6869 }, { "epoch": 1.2, "learning_rate": 4.8462828818076885e-05, "loss": 0.3009, "step": 6870 }, { "epoch": 1.2, "learning_rate": 4.8460341065078246e-05, "loss": 0.3033, "step": 6871 }, { "epoch": 1.2, "learning_rate": 4.845785137193614e-05, "loss": 0.3063, "step": 6872 }, { "epoch": 1.2, "learning_rate": 4.845535973888102e-05, "loss": 0.32, "step": 6873 }, { "epoch": 1.2, "learning_rate": 4.8452866166143526e-05, "loss": 0.3026, "step": 6874 }, { "epoch": 1.2, "learning_rate": 4.845037065395445e-05, "loss": 0.3045, "step": 6875 }, { "epoch": 1.2, "learning_rate": 4.844787320254479e-05, "loss": 0.2925, "step": 6876 }, { "epoch": 1.2, "learning_rate": 4.844537381214573e-05, "loss": 0.3043, "step": 6877 }, { "epoch": 1.2, "learning_rate": 4.84428724829886e-05, "loss": 0.3116, "step": 6878 }, { "epoch": 1.2, "learning_rate": 4.844036921530495e-05, "loss": 0.3146, "step": 6879 }, { "epoch": 1.2, "learning_rate": 4.843786400932648e-05, "loss": 0.2998, "step": 6880 }, { "epoch": 1.2, "learning_rate": 4.843535686528507e-05, "loss": 0.3181, "step": 6881 }, { "epoch": 1.2, "learning_rate": 4.84328477834128e-05, "loss": 0.3112, "step": 6882 }, { "epoch": 1.2, "learning_rate": 4.843033676394191e-05, "loss": 0.3111, "step": 6883 }, { "epoch": 1.2, "learning_rate": 4.8427823807104824e-05, "loss": 0.2992, "step": 6884 }, { "epoch": 1.2, "learning_rate": 4.842530891313416e-05, "loss": 0.314, "step": 6885 }, { "epoch": 1.2, "learning_rate": 4.8422792082262686e-05, "loss": 0.3134, "step": 6886 }, { "epoch": 1.2, "learning_rate": 4.8420273314723386e-05, "loss": 0.3049, "step": 6887 }, { "epoch": 1.2, "learning_rate": 4.8417752610749384e-05, "loss": 0.3219, "step": 6888 }, { "epoch": 1.2, "learning_rate": 4.841522997057402e-05, "loss": 0.297, "step": 6889 }, { "epoch": 1.2, "learning_rate": 4.8412705394430784e-05, "loss": 0.3038, "step": 6890 }, { "epoch": 1.2, "learning_rate": 4.841017888255337e-05, "loss": 0.3227, "step": 6891 }, { "epoch": 1.2, "learning_rate": 4.8407650435175615e-05, "loss": 0.3117, "step": 6892 }, { "epoch": 1.2, "learning_rate": 4.8405120052531584e-05, "loss": 0.3207, "step": 6893 }, { "epoch": 1.2, "learning_rate": 4.840258773485549e-05, "loss": 0.3191, "step": 6894 }, { "epoch": 1.2, "learning_rate": 4.8400053482381715e-05, "loss": 0.3047, "step": 6895 }, { "epoch": 1.2, "learning_rate": 4.839751729534485e-05, "loss": 0.2969, "step": 6896 }, { "epoch": 1.2, "learning_rate": 4.8394979173979655e-05, "loss": 0.2884, "step": 6897 }, { "epoch": 1.2, "learning_rate": 4.839243911852105e-05, "loss": 0.2998, "step": 6898 }, { "epoch": 1.2, "learning_rate": 4.838989712920417e-05, "loss": 0.3156, "step": 6899 }, { "epoch": 1.2, "learning_rate": 4.838735320626429e-05, "loss": 0.2986, "step": 6900 }, { "epoch": 1.2, "learning_rate": 4.838480734993689e-05, "loss": 0.2938, "step": 6901 }, { "epoch": 1.2, "learning_rate": 4.838225956045761e-05, "loss": 0.3045, "step": 6902 }, { "epoch": 1.2, "learning_rate": 4.837970983806229e-05, "loss": 0.3174, "step": 6903 }, { "epoch": 1.2, "learning_rate": 4.837715818298695e-05, "loss": 0.3047, "step": 6904 }, { "epoch": 1.2, "learning_rate": 4.8374604595467754e-05, "loss": 0.3174, "step": 6905 }, { "epoch": 1.2, "learning_rate": 4.837204907574109e-05, "loss": 0.3232, "step": 6906 }, { "epoch": 1.2, "learning_rate": 4.836949162404348e-05, "loss": 0.2991, "step": 6907 }, { "epoch": 1.2, "learning_rate": 4.8366932240611676e-05, "loss": 0.3105, "step": 6908 }, { "epoch": 1.2, "learning_rate": 4.836437092568257e-05, "loss": 0.2982, "step": 6909 }, { "epoch": 1.2, "learning_rate": 4.8361807679493226e-05, "loss": 0.3053, "step": 6910 }, { "epoch": 1.2, "learning_rate": 4.8359242502280927e-05, "loss": 0.3021, "step": 6911 }, { "epoch": 1.2, "learning_rate": 4.8356675394283106e-05, "loss": 0.3026, "step": 6912 }, { "epoch": 1.2, "learning_rate": 4.8354106355737387e-05, "loss": 0.3104, "step": 6913 }, { "epoch": 1.21, "learning_rate": 4.835153538688155e-05, "loss": 0.3076, "step": 6914 }, { "epoch": 1.21, "learning_rate": 4.834896248795359e-05, "loss": 0.3042, "step": 6915 }, { "epoch": 1.21, "learning_rate": 4.834638765919165e-05, "loss": 0.296, "step": 6916 }, { "epoch": 1.21, "learning_rate": 4.834381090083406e-05, "loss": 0.3047, "step": 6917 }, { "epoch": 1.21, "learning_rate": 4.8341232213119336e-05, "loss": 0.3062, "step": 6918 }, { "epoch": 1.21, "learning_rate": 4.8338651596286175e-05, "loss": 0.2965, "step": 6919 }, { "epoch": 1.21, "learning_rate": 4.833606905057343e-05, "loss": 0.3101, "step": 6920 }, { "epoch": 1.21, "learning_rate": 4.833348457622017e-05, "loss": 0.3161, "step": 6921 }, { "epoch": 1.21, "learning_rate": 4.83308981734656e-05, "loss": 0.3017, "step": 6922 }, { "epoch": 1.21, "learning_rate": 4.8328309842549135e-05, "loss": 0.308, "step": 6923 }, { "epoch": 1.21, "learning_rate": 4.832571958371035e-05, "loss": 0.3037, "step": 6924 }, { "epoch": 1.21, "learning_rate": 4.8323127397189015e-05, "loss": 0.3134, "step": 6925 }, { "epoch": 1.21, "learning_rate": 4.8320533283225066e-05, "loss": 0.2949, "step": 6926 }, { "epoch": 1.21, "learning_rate": 4.831793724205861e-05, "loss": 0.3141, "step": 6927 }, { "epoch": 1.21, "learning_rate": 4.831533927392997e-05, "loss": 0.3002, "step": 6928 }, { "epoch": 1.21, "learning_rate": 4.8312739379079596e-05, "loss": 0.3033, "step": 6929 }, { "epoch": 1.21, "learning_rate": 4.831013755774814e-05, "loss": 0.3092, "step": 6930 }, { "epoch": 1.21, "learning_rate": 4.830753381017646e-05, "loss": 0.3155, "step": 6931 }, { "epoch": 1.21, "learning_rate": 4.830492813660553e-05, "loss": 0.2936, "step": 6932 }, { "epoch": 1.21, "learning_rate": 4.830232053727657e-05, "loss": 0.2897, "step": 6933 }, { "epoch": 1.21, "learning_rate": 4.829971101243093e-05, "loss": 0.3089, "step": 6934 }, { "epoch": 1.21, "learning_rate": 4.829709956231015e-05, "loss": 0.3125, "step": 6935 }, { "epoch": 1.21, "learning_rate": 4.8294486187155967e-05, "loss": 0.3126, "step": 6936 }, { "epoch": 1.21, "learning_rate": 4.829187088721026e-05, "loss": 0.3093, "step": 6937 }, { "epoch": 1.21, "learning_rate": 4.828925366271514e-05, "loss": 0.3134, "step": 6938 }, { "epoch": 1.21, "learning_rate": 4.828663451391284e-05, "loss": 0.3142, "step": 6939 }, { "epoch": 1.21, "learning_rate": 4.82840134410458e-05, "loss": 0.3038, "step": 6940 }, { "epoch": 1.21, "learning_rate": 4.828139044435663e-05, "loss": 0.3077, "step": 6941 }, { "epoch": 1.21, "learning_rate": 4.827876552408812e-05, "loss": 0.3077, "step": 6942 }, { "epoch": 1.21, "learning_rate": 4.827613868048326e-05, "loss": 0.337, "step": 6943 }, { "epoch": 1.21, "learning_rate": 4.827350991378518e-05, "loss": 0.3027, "step": 6944 }, { "epoch": 1.21, "learning_rate": 4.8270879224237195e-05, "loss": 0.2976, "step": 6945 }, { "epoch": 1.21, "learning_rate": 4.826824661208283e-05, "loss": 0.3102, "step": 6946 }, { "epoch": 1.21, "learning_rate": 4.8265612077565755e-05, "loss": 0.3034, "step": 6947 }, { "epoch": 1.21, "learning_rate": 4.826297562092984e-05, "loss": 0.3037, "step": 6948 }, { "epoch": 1.21, "learning_rate": 4.8260337242419106e-05, "loss": 0.3116, "step": 6949 }, { "epoch": 1.21, "learning_rate": 4.825769694227778e-05, "loss": 0.3041, "step": 6950 }, { "epoch": 1.21, "learning_rate": 4.8255054720750254e-05, "loss": 0.3021, "step": 6951 }, { "epoch": 1.21, "learning_rate": 4.82524105780811e-05, "loss": 0.2951, "step": 6952 }, { "epoch": 1.21, "learning_rate": 4.8249764514515054e-05, "loss": 0.3093, "step": 6953 }, { "epoch": 1.21, "learning_rate": 4.824711653029706e-05, "loss": 0.3093, "step": 6954 }, { "epoch": 1.21, "learning_rate": 4.8244466625672205e-05, "loss": 0.3136, "step": 6955 }, { "epoch": 1.21, "learning_rate": 4.824181480088578e-05, "loss": 0.3039, "step": 6956 }, { "epoch": 1.21, "learning_rate": 4.823916105618326e-05, "loss": 0.3072, "step": 6957 }, { "epoch": 1.21, "learning_rate": 4.823650539181025e-05, "loss": 0.301, "step": 6958 }, { "epoch": 1.21, "learning_rate": 4.8233847808012586e-05, "loss": 0.3206, "step": 6959 }, { "epoch": 1.21, "learning_rate": 4.823118830503626e-05, "loss": 0.3123, "step": 6960 }, { "epoch": 1.21, "learning_rate": 4.822852688312744e-05, "loss": 0.2972, "step": 6961 }, { "epoch": 1.21, "learning_rate": 4.822586354253247e-05, "loss": 0.311, "step": 6962 }, { "epoch": 1.21, "learning_rate": 4.8223198283497886e-05, "loss": 0.3144, "step": 6963 }, { "epoch": 1.21, "learning_rate": 4.822053110627038e-05, "loss": 0.3053, "step": 6964 }, { "epoch": 1.21, "learning_rate": 4.8217862011096844e-05, "loss": 0.3203, "step": 6965 }, { "epoch": 1.21, "learning_rate": 4.821519099822432e-05, "loss": 0.3026, "step": 6966 }, { "epoch": 1.21, "learning_rate": 4.821251806790007e-05, "loss": 0.3165, "step": 6967 }, { "epoch": 1.21, "learning_rate": 4.8209843220371476e-05, "loss": 0.3083, "step": 6968 }, { "epoch": 1.21, "learning_rate": 4.820716645588615e-05, "loss": 0.3251, "step": 6969 }, { "epoch": 1.21, "learning_rate": 4.8204487774691855e-05, "loss": 0.3226, "step": 6970 }, { "epoch": 1.22, "learning_rate": 4.820180717703653e-05, "loss": 0.2959, "step": 6971 }, { "epoch": 1.22, "learning_rate": 4.819912466316831e-05, "loss": 0.3094, "step": 6972 }, { "epoch": 1.22, "learning_rate": 4.819644023333549e-05, "loss": 0.3005, "step": 6973 }, { "epoch": 1.22, "learning_rate": 4.819375388778654e-05, "loss": 0.3138, "step": 6974 }, { "epoch": 1.22, "learning_rate": 4.819106562677013e-05, "loss": 0.3319, "step": 6975 }, { "epoch": 1.22, "learning_rate": 4.818837545053508e-05, "loss": 0.3201, "step": 6976 }, { "epoch": 1.22, "learning_rate": 4.8185683359330406e-05, "loss": 0.3099, "step": 6977 }, { "epoch": 1.22, "learning_rate": 4.818298935340529e-05, "loss": 0.3172, "step": 6978 }, { "epoch": 1.22, "learning_rate": 4.818029343300911e-05, "loss": 0.3065, "step": 6979 }, { "epoch": 1.22, "learning_rate": 4.817759559839139e-05, "loss": 0.3135, "step": 6980 }, { "epoch": 1.22, "learning_rate": 4.8174895849801855e-05, "loss": 0.312, "step": 6981 }, { "epoch": 1.22, "learning_rate": 4.817219418749041e-05, "loss": 0.3066, "step": 6982 }, { "epoch": 1.22, "learning_rate": 4.816949061170711e-05, "loss": 0.3108, "step": 6983 }, { "epoch": 1.22, "learning_rate": 4.816678512270221e-05, "loss": 0.2983, "step": 6984 }, { "epoch": 1.22, "learning_rate": 4.8164077720726144e-05, "loss": 0.3036, "step": 6985 }, { "epoch": 1.22, "learning_rate": 4.8161368406029514e-05, "loss": 0.3089, "step": 6986 }, { "epoch": 1.22, "learning_rate": 4.8158657178863115e-05, "loss": 0.3007, "step": 6987 }, { "epoch": 1.22, "learning_rate": 4.815594403947787e-05, "loss": 0.3086, "step": 6988 }, { "epoch": 1.22, "learning_rate": 4.815322898812494e-05, "loss": 0.3121, "step": 6989 }, { "epoch": 1.22, "learning_rate": 4.8150512025055647e-05, "loss": 0.3179, "step": 6990 }, { "epoch": 1.22, "learning_rate": 4.814779315052145e-05, "loss": 0.2992, "step": 6991 }, { "epoch": 1.22, "learning_rate": 4.814507236477404e-05, "loss": 0.3156, "step": 6992 }, { "epoch": 1.22, "learning_rate": 4.8142349668065235e-05, "loss": 0.3114, "step": 6993 }, { "epoch": 1.22, "learning_rate": 4.813962506064708e-05, "loss": 0.3146, "step": 6994 }, { "epoch": 1.22, "learning_rate": 4.813689854277176e-05, "loss": 0.2985, "step": 6995 }, { "epoch": 1.22, "learning_rate": 4.813417011469165e-05, "loss": 0.3061, "step": 6996 }, { "epoch": 1.22, "learning_rate": 4.813143977665931e-05, "loss": 0.2961, "step": 6997 }, { "epoch": 1.22, "learning_rate": 4.8128707528927445e-05, "loss": 0.2945, "step": 6998 }, { "epoch": 1.22, "learning_rate": 4.812597337174898e-05, "loss": 0.3171, "step": 6999 }, { "epoch": 1.22, "learning_rate": 4.8123237305376975e-05, "loss": 0.3014, "step": 7000 }, { "epoch": 1.22, "learning_rate": 4.812049933006471e-05, "loss": 0.3219, "step": 7001 }, { "epoch": 1.22, "learning_rate": 4.811775944606561e-05, "loss": 0.3067, "step": 7002 }, { "epoch": 1.22, "learning_rate": 4.8115017653633285e-05, "loss": 0.3084, "step": 7003 }, { "epoch": 1.22, "learning_rate": 4.811227395302152e-05, "loss": 0.3027, "step": 7004 }, { "epoch": 1.22, "learning_rate": 4.810952834448429e-05, "loss": 0.3142, "step": 7005 }, { "epoch": 1.22, "learning_rate": 4.810678082827571e-05, "loss": 0.3143, "step": 7006 }, { "epoch": 1.22, "learning_rate": 4.8104031404650126e-05, "loss": 0.2991, "step": 7007 }, { "epoch": 1.22, "learning_rate": 4.810128007386202e-05, "loss": 0.3134, "step": 7008 }, { "epoch": 1.22, "learning_rate": 4.8098526836166055e-05, "loss": 0.3326, "step": 7009 }, { "epoch": 1.22, "learning_rate": 4.809577169181709e-05, "loss": 0.3109, "step": 7010 }, { "epoch": 1.22, "learning_rate": 4.809301464107014e-05, "loss": 0.2966, "step": 7011 }, { "epoch": 1.22, "learning_rate": 4.809025568418041e-05, "loss": 0.289, "step": 7012 }, { "epoch": 1.22, "learning_rate": 4.8087494821403275e-05, "loss": 0.3099, "step": 7013 }, { "epoch": 1.22, "learning_rate": 4.808473205299428e-05, "loss": 0.3136, "step": 7014 }, { "epoch": 1.22, "learning_rate": 4.808196737920917e-05, "loss": 0.3043, "step": 7015 }, { "epoch": 1.22, "learning_rate": 4.807920080030384e-05, "loss": 0.3066, "step": 7016 }, { "epoch": 1.22, "learning_rate": 4.807643231653437e-05, "loss": 0.3181, "step": 7017 }, { "epoch": 1.22, "learning_rate": 4.8073661928157027e-05, "loss": 0.2896, "step": 7018 }, { "epoch": 1.22, "learning_rate": 4.807088963542824e-05, "loss": 0.3323, "step": 7019 }, { "epoch": 1.22, "learning_rate": 4.80681154386046e-05, "loss": 0.2975, "step": 7020 }, { "epoch": 1.22, "learning_rate": 4.806533933794293e-05, "loss": 0.3031, "step": 7021 }, { "epoch": 1.22, "learning_rate": 4.8062561333700166e-05, "loss": 0.3116, "step": 7022 }, { "epoch": 1.22, "learning_rate": 4.8059781426133465e-05, "loss": 0.3, "step": 7023 }, { "epoch": 1.22, "learning_rate": 4.805699961550013e-05, "loss": 0.3288, "step": 7024 }, { "epoch": 1.22, "learning_rate": 4.8054215902057664e-05, "loss": 0.2966, "step": 7025 }, { "epoch": 1.22, "learning_rate": 4.805143028606373e-05, "loss": 0.2973, "step": 7026 }, { "epoch": 1.22, "learning_rate": 4.804864276777615e-05, "loss": 0.3029, "step": 7027 }, { "epoch": 1.22, "learning_rate": 4.804585334745298e-05, "loss": 0.2959, "step": 7028 }, { "epoch": 1.23, "learning_rate": 4.80430620253524e-05, "loss": 0.2962, "step": 7029 }, { "epoch": 1.23, "learning_rate": 4.804026880173277e-05, "loss": 0.3202, "step": 7030 }, { "epoch": 1.23, "learning_rate": 4.803747367685266e-05, "loss": 0.3202, "step": 7031 }, { "epoch": 1.23, "learning_rate": 4.8034676650970784e-05, "loss": 0.3098, "step": 7032 }, { "epoch": 1.23, "learning_rate": 4.803187772434604e-05, "loss": 0.3043, "step": 7033 }, { "epoch": 1.23, "learning_rate": 4.8029076897237496e-05, "loss": 0.3087, "step": 7034 }, { "epoch": 1.23, "learning_rate": 4.802627416990442e-05, "loss": 0.3043, "step": 7035 }, { "epoch": 1.23, "learning_rate": 4.802346954260624e-05, "loss": 0.3105, "step": 7036 }, { "epoch": 1.23, "learning_rate": 4.802066301560254e-05, "loss": 0.3219, "step": 7037 }, { "epoch": 1.23, "learning_rate": 4.801785458915312e-05, "loss": 0.3173, "step": 7038 }, { "epoch": 1.23, "learning_rate": 4.801504426351793e-05, "loss": 0.313, "step": 7039 }, { "epoch": 1.23, "learning_rate": 4.801223203895709e-05, "loss": 0.308, "step": 7040 }, { "epoch": 1.23, "learning_rate": 4.800941791573092e-05, "loss": 0.3034, "step": 7041 }, { "epoch": 1.23, "learning_rate": 4.8006601894099896e-05, "loss": 0.3115, "step": 7042 }, { "epoch": 1.23, "learning_rate": 4.800378397432468e-05, "loss": 0.3045, "step": 7043 }, { "epoch": 1.23, "learning_rate": 4.8000964156666094e-05, "loss": 0.3167, "step": 7044 }, { "epoch": 1.23, "learning_rate": 4.799814244138516e-05, "loss": 0.3063, "step": 7045 }, { "epoch": 1.23, "learning_rate": 4.7995318828743074e-05, "loss": 0.3224, "step": 7046 }, { "epoch": 1.23, "learning_rate": 4.799249331900117e-05, "loss": 0.3231, "step": 7047 }, { "epoch": 1.23, "learning_rate": 4.7989665912421e-05, "loss": 0.3147, "step": 7048 }, { "epoch": 1.23, "learning_rate": 4.7986836609264285e-05, "loss": 0.3188, "step": 7049 }, { "epoch": 1.23, "learning_rate": 4.798400540979289e-05, "loss": 0.3146, "step": 7050 }, { "epoch": 1.23, "learning_rate": 4.798117231426889e-05, "loss": 0.3126, "step": 7051 }, { "epoch": 1.23, "learning_rate": 4.7978337322954536e-05, "loss": 0.3106, "step": 7052 }, { "epoch": 1.23, "learning_rate": 4.7975500436112214e-05, "loss": 0.3137, "step": 7053 }, { "epoch": 1.23, "learning_rate": 4.7972661654004536e-05, "loss": 0.3015, "step": 7054 }, { "epoch": 1.23, "learning_rate": 4.796982097689426e-05, "loss": 0.3173, "step": 7055 }, { "epoch": 1.23, "learning_rate": 4.796697840504433e-05, "loss": 0.3006, "step": 7056 }, { "epoch": 1.23, "learning_rate": 4.796413393871785e-05, "loss": 0.3095, "step": 7057 }, { "epoch": 1.23, "learning_rate": 4.7961287578178126e-05, "loss": 0.3357, "step": 7058 }, { "epoch": 1.23, "learning_rate": 4.795843932368862e-05, "loss": 0.2984, "step": 7059 }, { "epoch": 1.23, "learning_rate": 4.7955589175512974e-05, "loss": 0.3032, "step": 7060 }, { "epoch": 1.23, "learning_rate": 4.7952737133914995e-05, "loss": 0.3085, "step": 7061 }, { "epoch": 1.23, "learning_rate": 4.794988319915868e-05, "loss": 0.3112, "step": 7062 }, { "epoch": 1.23, "learning_rate": 4.794702737150821e-05, "loss": 0.3207, "step": 7063 }, { "epoch": 1.23, "learning_rate": 4.7944169651227915e-05, "loss": 0.296, "step": 7064 }, { "epoch": 1.23, "learning_rate": 4.794131003858232e-05, "loss": 0.3059, "step": 7065 }, { "epoch": 1.23, "learning_rate": 4.7938448533836104e-05, "loss": 0.3016, "step": 7066 }, { "epoch": 1.23, "learning_rate": 4.7935585137254155e-05, "loss": 0.3162, "step": 7067 }, { "epoch": 1.23, "learning_rate": 4.79327198491015e-05, "loss": 0.3079, "step": 7068 }, { "epoch": 1.23, "learning_rate": 4.792985266964337e-05, "loss": 0.3106, "step": 7069 }, { "epoch": 1.23, "learning_rate": 4.792698359914515e-05, "loss": 0.2924, "step": 7070 }, { "epoch": 1.23, "learning_rate": 4.792411263787241e-05, "loss": 0.3234, "step": 7071 }, { "epoch": 1.23, "learning_rate": 4.7921239786090896e-05, "loss": 0.2993, "step": 7072 }, { "epoch": 1.23, "learning_rate": 4.7918365044066526e-05, "loss": 0.3141, "step": 7073 }, { "epoch": 1.23, "learning_rate": 4.7915488412065394e-05, "loss": 0.3196, "step": 7074 }, { "epoch": 1.23, "learning_rate": 4.791260989035376e-05, "loss": 0.3006, "step": 7075 }, { "epoch": 1.23, "learning_rate": 4.790972947919809e-05, "loss": 0.3101, "step": 7076 }, { "epoch": 1.23, "learning_rate": 4.7906847178864975e-05, "loss": 0.3019, "step": 7077 }, { "epoch": 1.23, "learning_rate": 4.7903962989621224e-05, "loss": 0.3099, "step": 7078 }, { "epoch": 1.23, "learning_rate": 4.79010769117338e-05, "loss": 0.315, "step": 7079 }, { "epoch": 1.23, "learning_rate": 4.789818894546986e-05, "loss": 0.3159, "step": 7080 }, { "epoch": 1.23, "learning_rate": 4.78952990910967e-05, "loss": 0.3053, "step": 7081 }, { "epoch": 1.23, "learning_rate": 4.789240734888182e-05, "loss": 0.3168, "step": 7082 }, { "epoch": 1.23, "learning_rate": 4.788951371909289e-05, "loss": 0.2982, "step": 7083 }, { "epoch": 1.23, "learning_rate": 4.7886618201997747e-05, "loss": 0.307, "step": 7084 }, { "epoch": 1.23, "learning_rate": 4.7883720797864404e-05, "loss": 0.305, "step": 7085 }, { "epoch": 1.24, "learning_rate": 4.788082150696108e-05, "loss": 0.3031, "step": 7086 }, { "epoch": 1.24, "learning_rate": 4.787792032955611e-05, "loss": 0.3056, "step": 7087 }, { "epoch": 1.24, "learning_rate": 4.7875017265918036e-05, "loss": 0.3091, "step": 7088 }, { "epoch": 1.24, "learning_rate": 4.7872112316315596e-05, "loss": 0.3016, "step": 7089 }, { "epoch": 1.24, "learning_rate": 4.786920548101766e-05, "loss": 0.3095, "step": 7090 }, { "epoch": 1.24, "learning_rate": 4.786629676029329e-05, "loss": 0.3178, "step": 7091 }, { "epoch": 1.24, "learning_rate": 4.786338615441174e-05, "loss": 0.3213, "step": 7092 }, { "epoch": 1.24, "learning_rate": 4.786047366364242e-05, "loss": 0.3002, "step": 7093 }, { "epoch": 1.24, "learning_rate": 4.785755928825491e-05, "loss": 0.3031, "step": 7094 }, { "epoch": 1.24, "learning_rate": 4.7854643028518974e-05, "loss": 0.3002, "step": 7095 }, { "epoch": 1.24, "learning_rate": 4.785172488470455e-05, "loss": 0.3035, "step": 7096 }, { "epoch": 1.24, "learning_rate": 4.7848804857081746e-05, "loss": 0.312, "step": 7097 }, { "epoch": 1.24, "learning_rate": 4.784588294592085e-05, "loss": 0.3006, "step": 7098 }, { "epoch": 1.24, "learning_rate": 4.784295915149233e-05, "loss": 0.3157, "step": 7099 }, { "epoch": 1.24, "learning_rate": 4.78400334740668e-05, "loss": 0.3142, "step": 7100 }, { "epoch": 1.24, "learning_rate": 4.7837105913915086e-05, "loss": 0.3088, "step": 7101 }, { "epoch": 1.24, "learning_rate": 4.783417647130817e-05, "loss": 0.2976, "step": 7102 }, { "epoch": 1.24, "learning_rate": 4.78312451465172e-05, "loss": 0.3028, "step": 7103 }, { "epoch": 1.24, "learning_rate": 4.7828311939813515e-05, "loss": 0.3331, "step": 7104 }, { "epoch": 1.24, "learning_rate": 4.782537685146861e-05, "loss": 0.2982, "step": 7105 }, { "epoch": 1.24, "learning_rate": 4.782243988175417e-05, "loss": 0.3128, "step": 7106 }, { "epoch": 1.24, "learning_rate": 4.781950103094205e-05, "loss": 0.3013, "step": 7107 }, { "epoch": 1.24, "learning_rate": 4.781656029930427e-05, "loss": 0.2932, "step": 7108 }, { "epoch": 1.24, "learning_rate": 4.7813617687113044e-05, "loss": 0.3088, "step": 7109 }, { "epoch": 1.24, "learning_rate": 4.781067319464075e-05, "loss": 0.2869, "step": 7110 }, { "epoch": 1.24, "learning_rate": 4.780772682215992e-05, "loss": 0.3139, "step": 7111 }, { "epoch": 1.24, "learning_rate": 4.780477856994329e-05, "loss": 0.3175, "step": 7112 }, { "epoch": 1.24, "learning_rate": 4.7801828438263755e-05, "loss": 0.3019, "step": 7113 }, { "epoch": 1.24, "learning_rate": 4.779887642739438e-05, "loss": 0.3201, "step": 7114 }, { "epoch": 1.24, "learning_rate": 4.779592253760843e-05, "loss": 0.3439, "step": 7115 }, { "epoch": 1.24, "learning_rate": 4.779296676917931e-05, "loss": 0.3118, "step": 7116 }, { "epoch": 1.24, "learning_rate": 4.7790009122380614e-05, "loss": 0.3099, "step": 7117 }, { "epoch": 1.24, "learning_rate": 4.778704959748611e-05, "loss": 0.3082, "step": 7118 }, { "epoch": 1.24, "learning_rate": 4.778408819476974e-05, "loss": 0.3023, "step": 7119 }, { "epoch": 1.24, "learning_rate": 4.778112491450562e-05, "loss": 0.3106, "step": 7120 }, { "epoch": 1.24, "learning_rate": 4.7778159756968046e-05, "loss": 0.3111, "step": 7121 }, { "epoch": 1.24, "learning_rate": 4.7775192722431474e-05, "loss": 0.3184, "step": 7122 }, { "epoch": 1.24, "learning_rate": 4.777222381117053e-05, "loss": 0.3176, "step": 7123 }, { "epoch": 1.24, "learning_rate": 4.776925302346004e-05, "loss": 0.3052, "step": 7124 }, { "epoch": 1.24, "learning_rate": 4.776628035957498e-05, "loss": 0.3147, "step": 7125 }, { "epoch": 1.24, "learning_rate": 4.7763305819790515e-05, "loss": 0.304, "step": 7126 }, { "epoch": 1.24, "learning_rate": 4.776032940438197e-05, "loss": 0.3264, "step": 7127 }, { "epoch": 1.24, "learning_rate": 4.7757351113624846e-05, "loss": 0.3161, "step": 7128 }, { "epoch": 1.24, "learning_rate": 4.775437094779483e-05, "loss": 0.2922, "step": 7129 }, { "epoch": 1.24, "learning_rate": 4.7751388907167765e-05, "loss": 0.3259, "step": 7130 }, { "epoch": 1.24, "learning_rate": 4.774840499201969e-05, "loss": 0.3093, "step": 7131 }, { "epoch": 1.24, "learning_rate": 4.774541920262679e-05, "loss": 0.3143, "step": 7132 }, { "epoch": 1.24, "learning_rate": 4.7742431539265434e-05, "loss": 0.3195, "step": 7133 }, { "epoch": 1.24, "learning_rate": 4.773944200221218e-05, "loss": 0.3053, "step": 7134 }, { "epoch": 1.24, "learning_rate": 4.7736450591743745e-05, "loss": 0.3005, "step": 7135 }, { "epoch": 1.24, "learning_rate": 4.773345730813702e-05, "loss": 0.3015, "step": 7136 }, { "epoch": 1.24, "learning_rate": 4.773046215166908e-05, "loss": 0.2903, "step": 7137 }, { "epoch": 1.24, "learning_rate": 4.7727465122617144e-05, "loss": 0.3115, "step": 7138 }, { "epoch": 1.24, "learning_rate": 4.772446622125864e-05, "loss": 0.3287, "step": 7139 }, { "epoch": 1.24, "learning_rate": 4.7721465447871156e-05, "loss": 0.3129, "step": 7140 }, { "epoch": 1.24, "learning_rate": 4.7718462802732436e-05, "loss": 0.3252, "step": 7141 }, { "epoch": 1.24, "learning_rate": 4.771545828612044e-05, "loss": 0.3141, "step": 7142 }, { "epoch": 1.25, "learning_rate": 4.771245189831323e-05, "loss": 0.3041, "step": 7143 }, { "epoch": 1.25, "learning_rate": 4.770944363958912e-05, "loss": 0.3115, "step": 7144 }, { "epoch": 1.25, "learning_rate": 4.7706433510226564e-05, "loss": 0.3195, "step": 7145 }, { "epoch": 1.25, "learning_rate": 4.770342151050417e-05, "loss": 0.3044, "step": 7146 }, { "epoch": 1.25, "learning_rate": 4.770040764070074e-05, "loss": 0.3037, "step": 7147 }, { "epoch": 1.25, "learning_rate": 4.769739190109526e-05, "loss": 0.3007, "step": 7148 }, { "epoch": 1.25, "learning_rate": 4.769437429196687e-05, "loss": 0.3077, "step": 7149 }, { "epoch": 1.25, "learning_rate": 4.7691354813594854e-05, "loss": 0.3001, "step": 7150 }, { "epoch": 1.25, "learning_rate": 4.768833346625875e-05, "loss": 0.3259, "step": 7151 }, { "epoch": 1.25, "learning_rate": 4.7685310250238194e-05, "loss": 0.3477, "step": 7152 }, { "epoch": 1.25, "learning_rate": 4.768228516581304e-05, "loss": 0.318, "step": 7153 }, { "epoch": 1.25, "learning_rate": 4.767925821326328e-05, "loss": 0.3106, "step": 7154 }, { "epoch": 1.25, "learning_rate": 4.76762293928691e-05, "loss": 0.315, "step": 7155 }, { "epoch": 1.25, "learning_rate": 4.767319870491087e-05, "loss": 0.3097, "step": 7156 }, { "epoch": 1.25, "learning_rate": 4.767016614966911e-05, "loss": 0.3033, "step": 7157 }, { "epoch": 1.25, "learning_rate": 4.766713172742451e-05, "loss": 0.3131, "step": 7158 }, { "epoch": 1.25, "learning_rate": 4.766409543845796e-05, "loss": 0.3098, "step": 7159 }, { "epoch": 1.25, "learning_rate": 4.7661057283050505e-05, "loss": 0.3128, "step": 7160 }, { "epoch": 1.25, "learning_rate": 4.765801726148336e-05, "loss": 0.3035, "step": 7161 }, { "epoch": 1.25, "learning_rate": 4.7654975374037907e-05, "loss": 0.3262, "step": 7162 }, { "epoch": 1.25, "learning_rate": 4.765193162099573e-05, "loss": 0.3042, "step": 7163 }, { "epoch": 1.25, "learning_rate": 4.7648886002638554e-05, "loss": 0.3103, "step": 7164 }, { "epoch": 1.25, "learning_rate": 4.7645838519248294e-05, "loss": 0.2983, "step": 7165 }, { "epoch": 1.25, "learning_rate": 4.7642789171107026e-05, "loss": 0.3167, "step": 7166 }, { "epoch": 1.25, "learning_rate": 4.763973795849702e-05, "loss": 0.2985, "step": 7167 }, { "epoch": 1.25, "learning_rate": 4.763668488170069e-05, "loss": 0.3175, "step": 7168 }, { "epoch": 1.25, "learning_rate": 4.763362994100066e-05, "loss": 0.3049, "step": 7169 }, { "epoch": 1.25, "learning_rate": 4.7630573136679666e-05, "loss": 0.3024, "step": 7170 }, { "epoch": 1.25, "learning_rate": 4.762751446902068e-05, "loss": 0.3068, "step": 7171 }, { "epoch": 1.25, "learning_rate": 4.7624453938306815e-05, "loss": 0.317, "step": 7172 }, { "epoch": 1.25, "learning_rate": 4.7621391544821356e-05, "loss": 0.3145, "step": 7173 }, { "epoch": 1.25, "learning_rate": 4.761832728884778e-05, "loss": 0.3108, "step": 7174 }, { "epoch": 1.25, "learning_rate": 4.76152611706697e-05, "loss": 0.3254, "step": 7175 }, { "epoch": 1.25, "learning_rate": 4.761219319057095e-05, "loss": 0.2995, "step": 7176 }, { "epoch": 1.25, "learning_rate": 4.760912334883549e-05, "loss": 0.3066, "step": 7177 }, { "epoch": 1.25, "learning_rate": 4.7606051645747484e-05, "loss": 0.3126, "step": 7178 }, { "epoch": 1.25, "learning_rate": 4.7602978081591246e-05, "loss": 0.3051, "step": 7179 }, { "epoch": 1.25, "learning_rate": 4.7599902656651285e-05, "loss": 0.3098, "step": 7180 }, { "epoch": 1.25, "learning_rate": 4.759682537121227e-05, "loss": 0.2935, "step": 7181 }, { "epoch": 1.25, "learning_rate": 4.759374622555903e-05, "loss": 0.3076, "step": 7182 }, { "epoch": 1.25, "learning_rate": 4.759066521997658e-05, "loss": 0.3253, "step": 7183 }, { "epoch": 1.25, "learning_rate": 4.758758235475011e-05, "loss": 0.3034, "step": 7184 }, { "epoch": 1.25, "learning_rate": 4.7584497630165e-05, "loss": 0.3131, "step": 7185 }, { "epoch": 1.25, "learning_rate": 4.7581411046506745e-05, "loss": 0.3067, "step": 7186 }, { "epoch": 1.25, "learning_rate": 4.757832260406106e-05, "loss": 0.3081, "step": 7187 }, { "epoch": 1.25, "learning_rate": 4.757523230311383e-05, "loss": 0.3138, "step": 7188 }, { "epoch": 1.25, "learning_rate": 4.757214014395109e-05, "loss": 0.303, "step": 7189 }, { "epoch": 1.25, "learning_rate": 4.756904612685906e-05, "loss": 0.3019, "step": 7190 }, { "epoch": 1.25, "learning_rate": 4.756595025212412e-05, "loss": 0.303, "step": 7191 }, { "epoch": 1.25, "learning_rate": 4.7562852520032856e-05, "loss": 0.2949, "step": 7192 }, { "epoch": 1.25, "learning_rate": 4.7559752930871984e-05, "loss": 0.2981, "step": 7193 }, { "epoch": 1.25, "learning_rate": 4.755665148492842e-05, "loss": 0.3086, "step": 7194 }, { "epoch": 1.25, "learning_rate": 4.755354818248924e-05, "loss": 0.3083, "step": 7195 }, { "epoch": 1.25, "learning_rate": 4.7550443023841676e-05, "loss": 0.3109, "step": 7196 }, { "epoch": 1.25, "learning_rate": 4.7547336009273184e-05, "loss": 0.3073, "step": 7197 }, { "epoch": 1.25, "learning_rate": 4.754422713907132e-05, "loss": 0.3081, "step": 7198 }, { "epoch": 1.25, "learning_rate": 4.754111641352388e-05, "loss": 0.2977, "step": 7199 }, { "epoch": 1.25, "learning_rate": 4.753800383291878e-05, "loss": 0.3216, "step": 7200 }, { "epoch": 1.26, "learning_rate": 4.753488939754415e-05, "loss": 0.3085, "step": 7201 }, { "epoch": 1.26, "learning_rate": 4.753177310768824e-05, "loss": 0.3231, "step": 7202 }, { "epoch": 1.26, "learning_rate": 4.7528654963639527e-05, "loss": 0.3302, "step": 7203 }, { "epoch": 1.26, "learning_rate": 4.7525534965686614e-05, "loss": 0.3159, "step": 7204 }, { "epoch": 1.26, "learning_rate": 4.7522413114118316e-05, "loss": 0.3053, "step": 7205 }, { "epoch": 1.26, "learning_rate": 4.751928940922359e-05, "loss": 0.3068, "step": 7206 }, { "epoch": 1.26, "learning_rate": 4.751616385129158e-05, "loss": 0.3262, "step": 7207 }, { "epoch": 1.26, "learning_rate": 4.75130364406116e-05, "loss": 0.2969, "step": 7208 }, { "epoch": 1.26, "learning_rate": 4.7509907177473104e-05, "loss": 0.3047, "step": 7209 }, { "epoch": 1.26, "learning_rate": 4.750677606216577e-05, "loss": 0.2911, "step": 7210 }, { "epoch": 1.26, "learning_rate": 4.7503643094979425e-05, "loss": 0.3031, "step": 7211 }, { "epoch": 1.26, "learning_rate": 4.7500508276204046e-05, "loss": 0.3166, "step": 7212 }, { "epoch": 1.26, "learning_rate": 4.749737160612981e-05, "loss": 0.3185, "step": 7213 }, { "epoch": 1.26, "learning_rate": 4.7494233085047055e-05, "loss": 0.316, "step": 7214 }, { "epoch": 1.26, "learning_rate": 4.7491092713246283e-05, "loss": 0.306, "step": 7215 }, { "epoch": 1.26, "learning_rate": 4.748795049101819e-05, "loss": 0.32, "step": 7216 }, { "epoch": 1.26, "learning_rate": 4.748480641865362e-05, "loss": 0.3092, "step": 7217 }, { "epoch": 1.26, "learning_rate": 4.74816604964436e-05, "loss": 0.3079, "step": 7218 }, { "epoch": 1.26, "learning_rate": 4.747851272467931e-05, "loss": 0.3015, "step": 7219 }, { "epoch": 1.26, "learning_rate": 4.7475363103652135e-05, "loss": 0.2934, "step": 7220 }, { "epoch": 1.26, "learning_rate": 4.7472211633653604e-05, "loss": 0.3108, "step": 7221 }, { "epoch": 1.26, "learning_rate": 4.746905831497543e-05, "loss": 0.3105, "step": 7222 }, { "epoch": 1.26, "learning_rate": 4.746590314790948e-05, "loss": 0.3011, "step": 7223 }, { "epoch": 1.26, "learning_rate": 4.746274613274782e-05, "loss": 0.317, "step": 7224 }, { "epoch": 1.26, "learning_rate": 4.745958726978266e-05, "loss": 0.3038, "step": 7225 }, { "epoch": 1.26, "learning_rate": 4.7456426559306385e-05, "loss": 0.3067, "step": 7226 }, { "epoch": 1.26, "learning_rate": 4.745326400161159e-05, "loss": 0.3102, "step": 7227 }, { "epoch": 1.26, "learning_rate": 4.7450099596990975e-05, "loss": 0.3056, "step": 7228 }, { "epoch": 1.26, "learning_rate": 4.744693334573747e-05, "loss": 0.3207, "step": 7229 }, { "epoch": 1.26, "learning_rate": 4.744376524814413e-05, "loss": 0.3183, "step": 7230 }, { "epoch": 1.26, "learning_rate": 4.744059530450422e-05, "loss": 0.3092, "step": 7231 }, { "epoch": 1.26, "learning_rate": 4.743742351511115e-05, "loss": 0.3078, "step": 7232 }, { "epoch": 1.26, "learning_rate": 4.743424988025852e-05, "loss": 0.3151, "step": 7233 }, { "epoch": 1.26, "learning_rate": 4.743107440024007e-05, "loss": 0.3008, "step": 7234 }, { "epoch": 1.26, "learning_rate": 4.7427897075349746e-05, "loss": 0.2952, "step": 7235 }, { "epoch": 1.26, "learning_rate": 4.7424717905881645e-05, "loss": 0.3129, "step": 7236 }, { "epoch": 1.26, "learning_rate": 4.742153689213004e-05, "loss": 0.3019, "step": 7237 }, { "epoch": 1.26, "learning_rate": 4.7418354034389376e-05, "loss": 0.3055, "step": 7238 }, { "epoch": 1.26, "learning_rate": 4.7415169332954254e-05, "loss": 0.3153, "step": 7239 }, { "epoch": 1.26, "learning_rate": 4.741198278811947e-05, "loss": 0.3029, "step": 7240 }, { "epoch": 1.26, "learning_rate": 4.740879440017999e-05, "loss": 0.304, "step": 7241 }, { "epoch": 1.26, "learning_rate": 4.740560416943091e-05, "loss": 0.3112, "step": 7242 }, { "epoch": 1.26, "learning_rate": 4.7402412096167545e-05, "loss": 0.308, "step": 7243 }, { "epoch": 1.26, "learning_rate": 4.739921818068536e-05, "loss": 0.3118, "step": 7244 }, { "epoch": 1.26, "learning_rate": 4.739602242327999e-05, "loss": 0.2923, "step": 7245 }, { "epoch": 1.26, "learning_rate": 4.739282482424724e-05, "loss": 0.3083, "step": 7246 }, { "epoch": 1.26, "learning_rate": 4.7389625383883104e-05, "loss": 0.2954, "step": 7247 }, { "epoch": 1.26, "learning_rate": 4.738642410248372e-05, "loss": 0.301, "step": 7248 }, { "epoch": 1.26, "learning_rate": 4.738322098034539e-05, "loss": 0.3112, "step": 7249 }, { "epoch": 1.26, "learning_rate": 4.738001601776462e-05, "loss": 0.2921, "step": 7250 }, { "epoch": 1.26, "learning_rate": 4.737680921503807e-05, "loss": 0.3088, "step": 7251 }, { "epoch": 1.26, "learning_rate": 4.7373600572462565e-05, "loss": 0.3025, "step": 7252 }, { "epoch": 1.26, "learning_rate": 4.737039009033512e-05, "loss": 0.3173, "step": 7253 }, { "epoch": 1.26, "learning_rate": 4.7367177768952874e-05, "loss": 0.297, "step": 7254 }, { "epoch": 1.26, "learning_rate": 4.7363963608613205e-05, "loss": 0.305, "step": 7255 }, { "epoch": 1.26, "learning_rate": 4.73607476096136e-05, "loss": 0.3103, "step": 7256 }, { "epoch": 1.26, "learning_rate": 4.7357529772251743e-05, "loss": 0.3014, "step": 7257 }, { "epoch": 1.27, "learning_rate": 4.735431009682549e-05, "loss": 0.316, "step": 7258 }, { "epoch": 1.27, "learning_rate": 4.735108858363286e-05, "loss": 0.3048, "step": 7259 }, { "epoch": 1.27, "learning_rate": 4.734786523297205e-05, "loss": 0.31, "step": 7260 }, { "epoch": 1.27, "learning_rate": 4.734464004514141e-05, "loss": 0.3075, "step": 7261 }, { "epoch": 1.27, "learning_rate": 4.734141302043949e-05, "loss": 0.291, "step": 7262 }, { "epoch": 1.27, "learning_rate": 4.7338184159164975e-05, "loss": 0.3144, "step": 7263 }, { "epoch": 1.27, "learning_rate": 4.733495346161675e-05, "loss": 0.3223, "step": 7264 }, { "epoch": 1.27, "learning_rate": 4.7331720928093845e-05, "loss": 0.308, "step": 7265 }, { "epoch": 1.27, "learning_rate": 4.7328486558895474e-05, "loss": 0.298, "step": 7266 }, { "epoch": 1.27, "learning_rate": 4.732525035432101e-05, "loss": 0.3107, "step": 7267 }, { "epoch": 1.27, "learning_rate": 4.732201231467004e-05, "loss": 0.2974, "step": 7268 }, { "epoch": 1.27, "learning_rate": 4.7318772440242245e-05, "loss": 0.2993, "step": 7269 }, { "epoch": 1.27, "learning_rate": 4.731553073133754e-05, "loss": 0.3184, "step": 7270 }, { "epoch": 1.27, "learning_rate": 4.731228718825597e-05, "loss": 0.2951, "step": 7271 }, { "epoch": 1.27, "learning_rate": 4.7309041811297775e-05, "loss": 0.3055, "step": 7272 }, { "epoch": 1.27, "learning_rate": 4.7305794600763356e-05, "loss": 0.2961, "step": 7273 }, { "epoch": 1.27, "learning_rate": 4.730254555695327e-05, "loss": 0.307, "step": 7274 }, { "epoch": 1.27, "learning_rate": 4.729929468016829e-05, "loss": 0.2956, "step": 7275 }, { "epoch": 1.27, "learning_rate": 4.729604197070929e-05, "loss": 0.3187, "step": 7276 }, { "epoch": 1.27, "learning_rate": 4.7292787428877366e-05, "loss": 0.3097, "step": 7277 }, { "epoch": 1.27, "learning_rate": 4.728953105497376e-05, "loss": 0.3001, "step": 7278 }, { "epoch": 1.27, "learning_rate": 4.7286272849299904e-05, "loss": 0.3092, "step": 7279 }, { "epoch": 1.27, "learning_rate": 4.728301281215736e-05, "loss": 0.2941, "step": 7280 }, { "epoch": 1.27, "learning_rate": 4.7279750943847916e-05, "loss": 0.3191, "step": 7281 }, { "epoch": 1.27, "learning_rate": 4.727648724467348e-05, "loss": 0.3021, "step": 7282 }, { "epoch": 1.27, "learning_rate": 4.727322171493616e-05, "loss": 0.297, "step": 7283 }, { "epoch": 1.27, "learning_rate": 4.72699543549382e-05, "loss": 0.3157, "step": 7284 }, { "epoch": 1.27, "learning_rate": 4.7266685164982053e-05, "loss": 0.3181, "step": 7285 }, { "epoch": 1.27, "learning_rate": 4.726341414537033e-05, "loss": 0.3207, "step": 7286 }, { "epoch": 1.27, "learning_rate": 4.7260141296405794e-05, "loss": 0.3167, "step": 7287 }, { "epoch": 1.27, "learning_rate": 4.725686661839139e-05, "loss": 0.3129, "step": 7288 }, { "epoch": 1.27, "learning_rate": 4.725359011163023e-05, "loss": 0.3144, "step": 7289 }, { "epoch": 1.27, "learning_rate": 4.7250311776425596e-05, "loss": 0.3156, "step": 7290 }, { "epoch": 1.27, "learning_rate": 4.724703161308094e-05, "loss": 0.3116, "step": 7291 }, { "epoch": 1.27, "learning_rate": 4.724374962189989e-05, "loss": 0.3163, "step": 7292 }, { "epoch": 1.27, "learning_rate": 4.724046580318622e-05, "loss": 0.3043, "step": 7293 }, { "epoch": 1.27, "learning_rate": 4.7237180157243895e-05, "loss": 0.3041, "step": 7294 }, { "epoch": 1.27, "learning_rate": 4.7233892684377055e-05, "loss": 0.3165, "step": 7295 }, { "epoch": 1.27, "learning_rate": 4.723060338488997e-05, "loss": 0.3123, "step": 7296 }, { "epoch": 1.27, "learning_rate": 4.722731225908714e-05, "loss": 0.291, "step": 7297 }, { "epoch": 1.27, "learning_rate": 4.722401930727317e-05, "loss": 0.3299, "step": 7298 }, { "epoch": 1.27, "learning_rate": 4.722072452975289e-05, "loss": 0.3066, "step": 7299 }, { "epoch": 1.27, "learning_rate": 4.721742792683125e-05, "loss": 0.3068, "step": 7300 }, { "epoch": 1.27, "learning_rate": 4.721412949881341e-05, "loss": 0.3066, "step": 7301 }, { "epoch": 1.27, "learning_rate": 4.721082924600467e-05, "loss": 0.3113, "step": 7302 }, { "epoch": 1.27, "learning_rate": 4.720752716871052e-05, "loss": 0.3124, "step": 7303 }, { "epoch": 1.27, "learning_rate": 4.720422326723659e-05, "loss": 0.3012, "step": 7304 }, { "epoch": 1.27, "learning_rate": 4.720091754188872e-05, "loss": 0.3004, "step": 7305 }, { "epoch": 1.27, "learning_rate": 4.719760999297288e-05, "loss": 0.2927, "step": 7306 }, { "epoch": 1.27, "learning_rate": 4.7194300620795234e-05, "loss": 0.3102, "step": 7307 }, { "epoch": 1.27, "learning_rate": 4.719098942566211e-05, "loss": 0.3123, "step": 7308 }, { "epoch": 1.27, "learning_rate": 4.718767640787999e-05, "loss": 0.3172, "step": 7309 }, { "epoch": 1.27, "learning_rate": 4.7184361567755534e-05, "loss": 0.3029, "step": 7310 }, { "epoch": 1.27, "learning_rate": 4.718104490559559e-05, "loss": 0.3044, "step": 7311 }, { "epoch": 1.27, "learning_rate": 4.7177726421707135e-05, "loss": 0.3023, "step": 7312 }, { "epoch": 1.27, "learning_rate": 4.717440611639736e-05, "loss": 0.31, "step": 7313 }, { "epoch": 1.27, "learning_rate": 4.717108398997357e-05, "loss": 0.3012, "step": 7314 }, { "epoch": 1.28, "learning_rate": 4.71677600427433e-05, "loss": 0.3264, "step": 7315 }, { "epoch": 1.28, "learning_rate": 4.716443427501421e-05, "loss": 0.3129, "step": 7316 }, { "epoch": 1.28, "learning_rate": 4.7161106687094134e-05, "loss": 0.3097, "step": 7317 }, { "epoch": 1.28, "learning_rate": 4.715777727929109e-05, "loss": 0.3096, "step": 7318 }, { "epoch": 1.28, "learning_rate": 4.715444605191327e-05, "loss": 0.2983, "step": 7319 }, { "epoch": 1.28, "learning_rate": 4.715111300526899e-05, "loss": 0.3015, "step": 7320 }, { "epoch": 1.28, "learning_rate": 4.71477781396668e-05, "loss": 0.3166, "step": 7321 }, { "epoch": 1.28, "learning_rate": 4.714444145541536e-05, "loss": 0.3135, "step": 7322 }, { "epoch": 1.28, "learning_rate": 4.7141102952823534e-05, "loss": 0.3258, "step": 7323 }, { "epoch": 1.28, "learning_rate": 4.713776263220033e-05, "loss": 0.302, "step": 7324 }, { "epoch": 1.28, "learning_rate": 4.713442049385495e-05, "loss": 0.3049, "step": 7325 }, { "epoch": 1.28, "learning_rate": 4.713107653809675e-05, "loss": 0.3191, "step": 7326 }, { "epoch": 1.28, "learning_rate": 4.712773076523524e-05, "loss": 0.3031, "step": 7327 }, { "epoch": 1.28, "learning_rate": 4.712438317558013e-05, "loss": 0.3081, "step": 7328 }, { "epoch": 1.28, "learning_rate": 4.712103376944128e-05, "loss": 0.2985, "step": 7329 }, { "epoch": 1.28, "learning_rate": 4.71176825471287e-05, "loss": 0.3079, "step": 7330 }, { "epoch": 1.28, "learning_rate": 4.7114329508952624e-05, "loss": 0.2986, "step": 7331 }, { "epoch": 1.28, "learning_rate": 4.7110974655223386e-05, "loss": 0.3149, "step": 7332 }, { "epoch": 1.28, "learning_rate": 4.710761798625153e-05, "loss": 0.2992, "step": 7333 }, { "epoch": 1.28, "learning_rate": 4.710425950234777e-05, "loss": 0.3113, "step": 7334 }, { "epoch": 1.28, "learning_rate": 4.710089920382295e-05, "loss": 0.3009, "step": 7335 }, { "epoch": 1.28, "learning_rate": 4.7097537090988126e-05, "loss": 0.3089, "step": 7336 }, { "epoch": 1.28, "learning_rate": 4.7094173164154504e-05, "loss": 0.3073, "step": 7337 }, { "epoch": 1.28, "learning_rate": 4.709080742363345e-05, "loss": 0.3192, "step": 7338 }, { "epoch": 1.28, "learning_rate": 4.708743986973652e-05, "loss": 0.3045, "step": 7339 }, { "epoch": 1.28, "learning_rate": 4.708407050277541e-05, "loss": 0.3074, "step": 7340 }, { "epoch": 1.28, "learning_rate": 4.7080699323062e-05, "loss": 0.3148, "step": 7341 }, { "epoch": 1.28, "learning_rate": 4.707732633090834e-05, "loss": 0.2903, "step": 7342 }, { "epoch": 1.28, "learning_rate": 4.707395152662663e-05, "loss": 0.3072, "step": 7343 }, { "epoch": 1.28, "learning_rate": 4.7070574910529266e-05, "loss": 0.3181, "step": 7344 }, { "epoch": 1.28, "learning_rate": 4.706719648292879e-05, "loss": 0.3016, "step": 7345 }, { "epoch": 1.28, "learning_rate": 4.706381624413792e-05, "loss": 0.303, "step": 7346 }, { "epoch": 1.28, "learning_rate": 4.706043419446953e-05, "loss": 0.3015, "step": 7347 }, { "epoch": 1.28, "learning_rate": 4.70570503342367e-05, "loss": 0.3082, "step": 7348 }, { "epoch": 1.28, "learning_rate": 4.70536646637526e-05, "loss": 0.3159, "step": 7349 }, { "epoch": 1.28, "learning_rate": 4.705027718333066e-05, "loss": 0.3053, "step": 7350 }, { "epoch": 1.28, "learning_rate": 4.704688789328442e-05, "loss": 0.3044, "step": 7351 }, { "epoch": 1.28, "learning_rate": 4.704349679392759e-05, "loss": 0.3087, "step": 7352 }, { "epoch": 1.28, "learning_rate": 4.704010388557408e-05, "loss": 0.3108, "step": 7353 }, { "epoch": 1.28, "learning_rate": 4.703670916853792e-05, "loss": 0.3221, "step": 7354 }, { "epoch": 1.28, "learning_rate": 4.703331264313336e-05, "loss": 0.2959, "step": 7355 }, { "epoch": 1.28, "learning_rate": 4.7029914309674785e-05, "loss": 0.3107, "step": 7356 }, { "epoch": 1.28, "learning_rate": 4.7026514168476736e-05, "loss": 0.3105, "step": 7357 }, { "epoch": 1.28, "learning_rate": 4.702311221985397e-05, "loss": 0.3105, "step": 7358 }, { "epoch": 1.28, "learning_rate": 4.7019708464121344e-05, "loss": 0.3131, "step": 7359 }, { "epoch": 1.28, "learning_rate": 4.7016302901593935e-05, "loss": 0.2999, "step": 7360 }, { "epoch": 1.28, "learning_rate": 4.701289553258698e-05, "loss": 0.3168, "step": 7361 }, { "epoch": 1.28, "learning_rate": 4.700948635741586e-05, "loss": 0.3098, "step": 7362 }, { "epoch": 1.28, "learning_rate": 4.7006075376396155e-05, "loss": 0.3148, "step": 7363 }, { "epoch": 1.28, "learning_rate": 4.700266258984357e-05, "loss": 0.3089, "step": 7364 }, { "epoch": 1.28, "learning_rate": 4.6999247998074023e-05, "loss": 0.31, "step": 7365 }, { "epoch": 1.28, "learning_rate": 4.6995831601403563e-05, "loss": 0.302, "step": 7366 }, { "epoch": 1.28, "learning_rate": 4.6992413400148423e-05, "loss": 0.3127, "step": 7367 }, { "epoch": 1.28, "learning_rate": 4.6988993394625e-05, "loss": 0.3046, "step": 7368 }, { "epoch": 1.28, "learning_rate": 4.6985571585149876e-05, "loss": 0.307, "step": 7369 }, { "epoch": 1.28, "learning_rate": 4.698214797203976e-05, "loss": 0.3128, "step": 7370 }, { "epoch": 1.28, "learning_rate": 4.697872255561155e-05, "loss": 0.3123, "step": 7371 }, { "epoch": 1.28, "learning_rate": 4.697529533618233e-05, "loss": 0.3069, "step": 7372 }, { "epoch": 1.29, "learning_rate": 4.6971866314069324e-05, "loss": 0.308, "step": 7373 }, { "epoch": 1.29, "learning_rate": 4.696843548958992e-05, "loss": 0.307, "step": 7374 }, { "epoch": 1.29, "learning_rate": 4.6965002863061705e-05, "loss": 0.3077, "step": 7375 }, { "epoch": 1.29, "learning_rate": 4.696156843480239e-05, "loss": 0.3151, "step": 7376 }, { "epoch": 1.29, "learning_rate": 4.695813220512989e-05, "loss": 0.3105, "step": 7377 }, { "epoch": 1.29, "learning_rate": 4.6954694174362264e-05, "loss": 0.3021, "step": 7378 }, { "epoch": 1.29, "learning_rate": 4.695125434281775e-05, "loss": 0.3066, "step": 7379 }, { "epoch": 1.29, "learning_rate": 4.694781271081474e-05, "loss": 0.3073, "step": 7380 }, { "epoch": 1.29, "learning_rate": 4.694436927867182e-05, "loss": 0.3177, "step": 7381 }, { "epoch": 1.29, "learning_rate": 4.694092404670769e-05, "loss": 0.3128, "step": 7382 }, { "epoch": 1.29, "learning_rate": 4.693747701524128e-05, "loss": 0.3067, "step": 7383 }, { "epoch": 1.29, "learning_rate": 4.693402818459164e-05, "loss": 0.3049, "step": 7384 }, { "epoch": 1.29, "learning_rate": 4.693057755507801e-05, "loss": 0.302, "step": 7385 }, { "epoch": 1.29, "learning_rate": 4.6927125127019775e-05, "loss": 0.311, "step": 7386 }, { "epoch": 1.29, "learning_rate": 4.692367090073652e-05, "loss": 0.3068, "step": 7387 }, { "epoch": 1.29, "learning_rate": 4.6920214876547976e-05, "loss": 0.3006, "step": 7388 }, { "epoch": 1.29, "learning_rate": 4.691675705477402e-05, "loss": 0.2998, "step": 7389 }, { "epoch": 1.29, "learning_rate": 4.6913297435734736e-05, "loss": 0.3036, "step": 7390 }, { "epoch": 1.29, "learning_rate": 4.690983601975035e-05, "loss": 0.3089, "step": 7391 }, { "epoch": 1.29, "learning_rate": 4.6906372807141265e-05, "loss": 0.299, "step": 7392 }, { "epoch": 1.29, "learning_rate": 4.690290779822804e-05, "loss": 0.3069, "step": 7393 }, { "epoch": 1.29, "learning_rate": 4.6899440993331404e-05, "loss": 0.314, "step": 7394 }, { "epoch": 1.29, "learning_rate": 4.689597239277226e-05, "loss": 0.3106, "step": 7395 }, { "epoch": 1.29, "learning_rate": 4.6892501996871654e-05, "loss": 0.3037, "step": 7396 }, { "epoch": 1.29, "learning_rate": 4.688902980595084e-05, "loss": 0.3173, "step": 7397 }, { "epoch": 1.29, "learning_rate": 4.688555582033118e-05, "loss": 0.3099, "step": 7398 }, { "epoch": 1.29, "learning_rate": 4.688208004033427e-05, "loss": 0.3004, "step": 7399 }, { "epoch": 1.29, "learning_rate": 4.687860246628182e-05, "loss": 0.3071, "step": 7400 }, { "epoch": 1.29, "learning_rate": 4.687512309849572e-05, "loss": 0.3043, "step": 7401 }, { "epoch": 1.29, "learning_rate": 4.6871641937298036e-05, "loss": 0.2965, "step": 7402 }, { "epoch": 1.29, "learning_rate": 4.686815898301099e-05, "loss": 0.3079, "step": 7403 }, { "epoch": 1.29, "learning_rate": 4.686467423595698e-05, "loss": 0.306, "step": 7404 }, { "epoch": 1.29, "learning_rate": 4.686118769645856e-05, "loss": 0.321, "step": 7405 }, { "epoch": 1.29, "learning_rate": 4.685769936483845e-05, "loss": 0.3078, "step": 7406 }, { "epoch": 1.29, "learning_rate": 4.685420924141953e-05, "loss": 0.3178, "step": 7407 }, { "epoch": 1.29, "learning_rate": 4.685071732652487e-05, "loss": 0.3067, "step": 7408 }, { "epoch": 1.29, "learning_rate": 4.6847223620477686e-05, "loss": 0.3378, "step": 7409 }, { "epoch": 1.29, "learning_rate": 4.684372812360137e-05, "loss": 0.2993, "step": 7410 }, { "epoch": 1.29, "learning_rate": 4.684023083621946e-05, "loss": 0.2947, "step": 7411 }, { "epoch": 1.29, "learning_rate": 4.683673175865569e-05, "loss": 0.3015, "step": 7412 }, { "epoch": 1.29, "learning_rate": 4.683323089123393e-05, "loss": 0.3218, "step": 7413 }, { "epoch": 1.29, "learning_rate": 4.6829728234278234e-05, "loss": 0.3023, "step": 7414 }, { "epoch": 1.29, "learning_rate": 4.682622378811282e-05, "loss": 0.3091, "step": 7415 }, { "epoch": 1.29, "learning_rate": 4.682271755306207e-05, "loss": 0.3343, "step": 7416 }, { "epoch": 1.29, "learning_rate": 4.6819209529450526e-05, "loss": 0.3106, "step": 7417 }, { "epoch": 1.29, "learning_rate": 4.68156997176029e-05, "loss": 0.3231, "step": 7418 }, { "epoch": 1.29, "learning_rate": 4.6812188117844075e-05, "loss": 0.3118, "step": 7419 }, { "epoch": 1.29, "learning_rate": 4.680867473049908e-05, "loss": 0.3012, "step": 7420 }, { "epoch": 1.29, "learning_rate": 4.680515955589314e-05, "loss": 0.316, "step": 7421 }, { "epoch": 1.29, "learning_rate": 4.680164259435162e-05, "loss": 0.3066, "step": 7422 }, { "epoch": 1.29, "learning_rate": 4.679812384620006e-05, "loss": 0.3166, "step": 7423 }, { "epoch": 1.29, "learning_rate": 4.679460331176416e-05, "loss": 0.3013, "step": 7424 }, { "epoch": 1.29, "learning_rate": 4.679108099136979e-05, "loss": 0.3155, "step": 7425 }, { "epoch": 1.29, "learning_rate": 4.6787556885343006e-05, "loss": 0.3071, "step": 7426 }, { "epoch": 1.29, "learning_rate": 4.678403099400999e-05, "loss": 0.3127, "step": 7427 }, { "epoch": 1.29, "learning_rate": 4.678050331769709e-05, "loss": 0.3149, "step": 7428 }, { "epoch": 1.29, "learning_rate": 4.6776973856730874e-05, "loss": 0.3114, "step": 7429 }, { "epoch": 1.3, "learning_rate": 4.677344261143802e-05, "loss": 0.3192, "step": 7430 }, { "epoch": 1.3, "learning_rate": 4.6769909582145374e-05, "loss": 0.3049, "step": 7431 }, { "epoch": 1.3, "learning_rate": 4.676637476917999e-05, "loss": 0.3013, "step": 7432 }, { "epoch": 1.3, "learning_rate": 4.6762838172869046e-05, "loss": 0.3155, "step": 7433 }, { "epoch": 1.3, "learning_rate": 4.67592997935399e-05, "loss": 0.3039, "step": 7434 }, { "epoch": 1.3, "learning_rate": 4.6755759631520076e-05, "loss": 0.3002, "step": 7435 }, { "epoch": 1.3, "learning_rate": 4.6752217687137246e-05, "loss": 0.3104, "step": 7436 }, { "epoch": 1.3, "learning_rate": 4.6748673960719285e-05, "loss": 0.3217, "step": 7437 }, { "epoch": 1.3, "learning_rate": 4.67451284525942e-05, "loss": 0.3171, "step": 7438 }, { "epoch": 1.3, "learning_rate": 4.674158116309016e-05, "loss": 0.3074, "step": 7439 }, { "epoch": 1.3, "learning_rate": 4.6738032092535523e-05, "loss": 0.3092, "step": 7440 }, { "epoch": 1.3, "learning_rate": 4.67344812412588e-05, "loss": 0.3086, "step": 7441 }, { "epoch": 1.3, "learning_rate": 4.673092860958867e-05, "loss": 0.3035, "step": 7442 }, { "epoch": 1.3, "learning_rate": 4.6727374197853965e-05, "loss": 0.3093, "step": 7443 }, { "epoch": 1.3, "learning_rate": 4.67238180063837e-05, "loss": 0.316, "step": 7444 }, { "epoch": 1.3, "learning_rate": 4.672026003550704e-05, "loss": 0.3165, "step": 7445 }, { "epoch": 1.3, "learning_rate": 4.671670028555332e-05, "loss": 0.3051, "step": 7446 }, { "epoch": 1.3, "learning_rate": 4.6713138756852035e-05, "loss": 0.3018, "step": 7447 }, { "epoch": 1.3, "learning_rate": 4.6709575449732864e-05, "loss": 0.3113, "step": 7448 }, { "epoch": 1.3, "learning_rate": 4.670601036452561e-05, "loss": 0.2955, "step": 7449 }, { "epoch": 1.3, "learning_rate": 4.67024435015603e-05, "loss": 0.3079, "step": 7450 }, { "epoch": 1.3, "learning_rate": 4.669887486116708e-05, "loss": 0.3121, "step": 7451 }, { "epoch": 1.3, "learning_rate": 4.669530444367625e-05, "loss": 0.3125, "step": 7452 }, { "epoch": 1.3, "learning_rate": 4.669173224941832e-05, "loss": 0.3096, "step": 7453 }, { "epoch": 1.3, "learning_rate": 4.668815827872394e-05, "loss": 0.3117, "step": 7454 }, { "epoch": 1.3, "learning_rate": 4.668458253192391e-05, "loss": 0.3066, "step": 7455 }, { "epoch": 1.3, "learning_rate": 4.668100500934924e-05, "loss": 0.3053, "step": 7456 }, { "epoch": 1.3, "learning_rate": 4.667742571133105e-05, "loss": 0.3037, "step": 7457 }, { "epoch": 1.3, "learning_rate": 4.667384463820065e-05, "loss": 0.3103, "step": 7458 }, { "epoch": 1.3, "learning_rate": 4.667026179028953e-05, "loss": 0.3098, "step": 7459 }, { "epoch": 1.3, "learning_rate": 4.666667716792931e-05, "loss": 0.3112, "step": 7460 }, { "epoch": 1.3, "learning_rate": 4.666309077145181e-05, "loss": 0.3216, "step": 7461 }, { "epoch": 1.3, "learning_rate": 4.6659502601188975e-05, "loss": 0.3112, "step": 7462 }, { "epoch": 1.3, "learning_rate": 4.6655912657472945e-05, "loss": 0.3008, "step": 7463 }, { "epoch": 1.3, "learning_rate": 4.665232094063603e-05, "loss": 0.2988, "step": 7464 }, { "epoch": 1.3, "learning_rate": 4.664872745101067e-05, "loss": 0.3081, "step": 7465 }, { "epoch": 1.3, "learning_rate": 4.664513218892948e-05, "loss": 0.292, "step": 7466 }, { "epoch": 1.3, "learning_rate": 4.6641535154725265e-05, "loss": 0.3211, "step": 7467 }, { "epoch": 1.3, "learning_rate": 4.663793634873097e-05, "loss": 0.3062, "step": 7468 }, { "epoch": 1.3, "learning_rate": 4.6634335771279705e-05, "loss": 0.311, "step": 7469 }, { "epoch": 1.3, "learning_rate": 4.6630733422704756e-05, "loss": 0.2957, "step": 7470 }, { "epoch": 1.3, "learning_rate": 4.662712930333956e-05, "loss": 0.3116, "step": 7471 }, { "epoch": 1.3, "learning_rate": 4.6623523413517725e-05, "loss": 0.3173, "step": 7472 }, { "epoch": 1.3, "learning_rate": 4.661991575357304e-05, "loss": 0.3179, "step": 7473 }, { "epoch": 1.3, "learning_rate": 4.66163063238394e-05, "loss": 0.3081, "step": 7474 }, { "epoch": 1.3, "learning_rate": 4.661269512465094e-05, "loss": 0.301, "step": 7475 }, { "epoch": 1.3, "learning_rate": 4.66090821563419e-05, "loss": 0.2985, "step": 7476 }, { "epoch": 1.3, "learning_rate": 4.660546741924672e-05, "loss": 0.3413, "step": 7477 }, { "epoch": 1.3, "learning_rate": 4.660185091369997e-05, "loss": 0.3116, "step": 7478 }, { "epoch": 1.3, "learning_rate": 4.659823264003642e-05, "loss": 0.3074, "step": 7479 }, { "epoch": 1.3, "learning_rate": 4.659461259859099e-05, "loss": 0.3019, "step": 7480 }, { "epoch": 1.3, "learning_rate": 4.659099078969875e-05, "loss": 0.3116, "step": 7481 }, { "epoch": 1.3, "learning_rate": 4.658736721369495e-05, "loss": 0.3152, "step": 7482 }, { "epoch": 1.3, "learning_rate": 4.658374187091498e-05, "loss": 0.3089, "step": 7483 }, { "epoch": 1.3, "learning_rate": 4.6580114761694435e-05, "loss": 0.2896, "step": 7484 }, { "epoch": 1.3, "learning_rate": 4.6576485886369045e-05, "loss": 0.3093, "step": 7485 }, { "epoch": 1.3, "learning_rate": 4.65728552452747e-05, "loss": 0.3106, "step": 7486 }, { "epoch": 1.3, "learning_rate": 4.6569222838747455e-05, "loss": 0.3161, "step": 7487 }, { "epoch": 1.31, "learning_rate": 4.6565588667123557e-05, "loss": 0.3107, "step": 7488 }, { "epoch": 1.31, "learning_rate": 4.656195273073938e-05, "loss": 0.3147, "step": 7489 }, { "epoch": 1.31, "learning_rate": 4.6558315029931467e-05, "loss": 0.3103, "step": 7490 }, { "epoch": 1.31, "learning_rate": 4.655467556503655e-05, "loss": 0.3088, "step": 7491 }, { "epoch": 1.31, "learning_rate": 4.655103433639151e-05, "loss": 0.3104, "step": 7492 }, { "epoch": 1.31, "learning_rate": 4.654739134433338e-05, "loss": 0.3119, "step": 7493 }, { "epoch": 1.31, "learning_rate": 4.654374658919935e-05, "loss": 0.299, "step": 7494 }, { "epoch": 1.31, "learning_rate": 4.65401000713268e-05, "loss": 0.3024, "step": 7495 }, { "epoch": 1.31, "learning_rate": 4.653645179105328e-05, "loss": 0.2934, "step": 7496 }, { "epoch": 1.31, "learning_rate": 4.6532801748716456e-05, "loss": 0.3063, "step": 7497 }, { "epoch": 1.31, "learning_rate": 4.65291499446542e-05, "loss": 0.3056, "step": 7498 }, { "epoch": 1.31, "learning_rate": 4.652549637920453e-05, "loss": 0.2998, "step": 7499 }, { "epoch": 1.31, "learning_rate": 4.6521841052705634e-05, "loss": 0.2993, "step": 7500 }, { "epoch": 1.31, "learning_rate": 4.651818396549585e-05, "loss": 0.3222, "step": 7501 }, { "epoch": 1.31, "learning_rate": 4.6514525117913686e-05, "loss": 0.3075, "step": 7502 }, { "epoch": 1.31, "learning_rate": 4.651086451029783e-05, "loss": 0.3241, "step": 7503 }, { "epoch": 1.31, "learning_rate": 4.65072021429871e-05, "loss": 0.3015, "step": 7504 }, { "epoch": 1.31, "learning_rate": 4.65035380163205e-05, "loss": 0.3282, "step": 7505 }, { "epoch": 1.31, "learning_rate": 4.649987213063719e-05, "loss": 0.3075, "step": 7506 }, { "epoch": 1.31, "learning_rate": 4.64962044862765e-05, "loss": 0.3152, "step": 7507 }, { "epoch": 1.31, "learning_rate": 4.649253508357792e-05, "loss": 0.3025, "step": 7508 }, { "epoch": 1.31, "learning_rate": 4.648886392288108e-05, "loss": 0.3025, "step": 7509 }, { "epoch": 1.31, "learning_rate": 4.648519100452581e-05, "loss": 0.3119, "step": 7510 }, { "epoch": 1.31, "learning_rate": 4.6481516328852074e-05, "loss": 0.3007, "step": 7511 }, { "epoch": 1.31, "learning_rate": 4.6477839896200016e-05, "loss": 0.3034, "step": 7512 }, { "epoch": 1.31, "learning_rate": 4.6474161706909936e-05, "loss": 0.3046, "step": 7513 }, { "epoch": 1.31, "learning_rate": 4.6470481761322296e-05, "loss": 0.3181, "step": 7514 }, { "epoch": 1.31, "learning_rate": 4.646680005977772e-05, "loss": 0.312, "step": 7515 }, { "epoch": 1.31, "learning_rate": 4.6463116602617e-05, "loss": 0.304, "step": 7516 }, { "epoch": 1.31, "learning_rate": 4.6459431390181074e-05, "loss": 0.3068, "step": 7517 }, { "epoch": 1.31, "learning_rate": 4.6455744422811075e-05, "loss": 0.296, "step": 7518 }, { "epoch": 1.31, "learning_rate": 4.645205570084825e-05, "loss": 0.3105, "step": 7519 }, { "epoch": 1.31, "learning_rate": 4.644836522463407e-05, "loss": 0.3092, "step": 7520 }, { "epoch": 1.31, "learning_rate": 4.6444672994510114e-05, "loss": 0.3209, "step": 7521 }, { "epoch": 1.31, "learning_rate": 4.644097901081815e-05, "loss": 0.3162, "step": 7522 }, { "epoch": 1.31, "learning_rate": 4.64372832739001e-05, "loss": 0.3062, "step": 7523 }, { "epoch": 1.31, "learning_rate": 4.643358578409804e-05, "loss": 0.3165, "step": 7524 }, { "epoch": 1.31, "learning_rate": 4.642988654175425e-05, "loss": 0.2978, "step": 7525 }, { "epoch": 1.31, "learning_rate": 4.642618554721111e-05, "loss": 0.3171, "step": 7526 }, { "epoch": 1.31, "learning_rate": 4.642248280081122e-05, "loss": 0.3137, "step": 7527 }, { "epoch": 1.31, "learning_rate": 4.64187783028973e-05, "loss": 0.3139, "step": 7528 }, { "epoch": 1.31, "learning_rate": 4.641507205381224e-05, "loss": 0.3082, "step": 7529 }, { "epoch": 1.31, "learning_rate": 4.641136405389912e-05, "loss": 0.3096, "step": 7530 }, { "epoch": 1.31, "learning_rate": 4.6407654303501155e-05, "loss": 0.3012, "step": 7531 }, { "epoch": 1.31, "learning_rate": 4.640394280296172e-05, "loss": 0.2991, "step": 7532 }, { "epoch": 1.31, "learning_rate": 4.640022955262437e-05, "loss": 0.3227, "step": 7533 }, { "epoch": 1.31, "learning_rate": 4.639651455283282e-05, "loss": 0.309, "step": 7534 }, { "epoch": 1.31, "learning_rate": 4.6392797803930916e-05, "loss": 0.2988, "step": 7535 }, { "epoch": 1.31, "learning_rate": 4.638907930626271e-05, "loss": 0.3022, "step": 7536 }, { "epoch": 1.31, "learning_rate": 4.638535906017239e-05, "loss": 0.3138, "step": 7537 }, { "epoch": 1.31, "learning_rate": 4.6381637066004315e-05, "loss": 0.3158, "step": 7538 }, { "epoch": 1.31, "learning_rate": 4.6377913324103e-05, "loss": 0.3175, "step": 7539 }, { "epoch": 1.31, "learning_rate": 4.637418783481312e-05, "loss": 0.2999, "step": 7540 }, { "epoch": 1.31, "learning_rate": 4.6370460598479535e-05, "loss": 0.3137, "step": 7541 }, { "epoch": 1.31, "learning_rate": 4.636673161544721e-05, "loss": 0.314, "step": 7542 }, { "epoch": 1.31, "learning_rate": 4.636300088606134e-05, "loss": 0.3069, "step": 7543 }, { "epoch": 1.31, "learning_rate": 4.635926841066724e-05, "loss": 0.3027, "step": 7544 }, { "epoch": 1.32, "learning_rate": 4.635553418961042e-05, "loss": 0.3065, "step": 7545 }, { "epoch": 1.32, "learning_rate": 4.635179822323649e-05, "loss": 0.3075, "step": 7546 }, { "epoch": 1.32, "learning_rate": 4.6348060511891285e-05, "loss": 0.2985, "step": 7547 }, { "epoch": 1.32, "learning_rate": 4.634432105592077e-05, "loss": 0.3075, "step": 7548 }, { "epoch": 1.32, "learning_rate": 4.634057985567108e-05, "loss": 0.3194, "step": 7549 }, { "epoch": 1.32, "learning_rate": 4.633683691148852e-05, "loss": 0.3087, "step": 7550 }, { "epoch": 1.32, "learning_rate": 4.6333092223719545e-05, "loss": 0.3108, "step": 7551 }, { "epoch": 1.32, "learning_rate": 4.6329345792710746e-05, "loss": 0.3022, "step": 7552 }, { "epoch": 1.32, "learning_rate": 4.632559761880893e-05, "loss": 0.3192, "step": 7553 }, { "epoch": 1.32, "learning_rate": 4.632184770236104e-05, "loss": 0.3095, "step": 7554 }, { "epoch": 1.32, "learning_rate": 4.631809604371415e-05, "loss": 0.3084, "step": 7555 }, { "epoch": 1.32, "learning_rate": 4.6314342643215556e-05, "loss": 0.3057, "step": 7556 }, { "epoch": 1.32, "learning_rate": 4.6310587501212666e-05, "loss": 0.3153, "step": 7557 }, { "epoch": 1.32, "learning_rate": 4.6306830618053064e-05, "loss": 0.3027, "step": 7558 }, { "epoch": 1.32, "learning_rate": 4.63030719940845e-05, "loss": 0.3223, "step": 7559 }, { "epoch": 1.32, "learning_rate": 4.629931162965489e-05, "loss": 0.313, "step": 7560 }, { "epoch": 1.32, "learning_rate": 4.629554952511229e-05, "loss": 0.315, "step": 7561 }, { "epoch": 1.32, "learning_rate": 4.629178568080493e-05, "loss": 0.3139, "step": 7562 }, { "epoch": 1.32, "learning_rate": 4.628802009708121e-05, "loss": 0.3002, "step": 7563 }, { "epoch": 1.32, "learning_rate": 4.628425277428968e-05, "loss": 0.3135, "step": 7564 }, { "epoch": 1.32, "learning_rate": 4.628048371277906e-05, "loss": 0.3121, "step": 7565 }, { "epoch": 1.32, "learning_rate": 4.62767129128982e-05, "loss": 0.3209, "step": 7566 }, { "epoch": 1.32, "learning_rate": 4.627294037499616e-05, "loss": 0.3242, "step": 7567 }, { "epoch": 1.32, "learning_rate": 4.6269166099422126e-05, "loss": 0.3067, "step": 7568 }, { "epoch": 1.32, "learning_rate": 4.626539008652546e-05, "loss": 0.3051, "step": 7569 }, { "epoch": 1.32, "learning_rate": 4.626161233665567e-05, "loss": 0.3035, "step": 7570 }, { "epoch": 1.32, "learning_rate": 4.625783285016243e-05, "loss": 0.3143, "step": 7571 }, { "epoch": 1.32, "learning_rate": 4.625405162739561e-05, "loss": 0.3034, "step": 7572 }, { "epoch": 1.32, "learning_rate": 4.6250268668705176e-05, "loss": 0.3053, "step": 7573 }, { "epoch": 1.32, "learning_rate": 4.6246483974441304e-05, "loss": 0.3029, "step": 7574 }, { "epoch": 1.32, "learning_rate": 4.624269754495431e-05, "loss": 0.3014, "step": 7575 }, { "epoch": 1.32, "learning_rate": 4.6238909380594685e-05, "loss": 0.3213, "step": 7576 }, { "epoch": 1.32, "learning_rate": 4.623511948171305e-05, "loss": 0.3024, "step": 7577 }, { "epoch": 1.32, "learning_rate": 4.623132784866023e-05, "loss": 0.3065, "step": 7578 }, { "epoch": 1.32, "learning_rate": 4.622753448178718e-05, "loss": 0.2948, "step": 7579 }, { "epoch": 1.32, "learning_rate": 4.622373938144503e-05, "loss": 0.3104, "step": 7580 }, { "epoch": 1.32, "learning_rate": 4.6219942547985054e-05, "loss": 0.2981, "step": 7581 }, { "epoch": 1.32, "learning_rate": 4.62161439817587e-05, "loss": 0.2912, "step": 7582 }, { "epoch": 1.32, "learning_rate": 4.621234368311757e-05, "loss": 0.335, "step": 7583 }, { "epoch": 1.32, "learning_rate": 4.620854165241344e-05, "loss": 0.2925, "step": 7584 }, { "epoch": 1.32, "learning_rate": 4.620473788999823e-05, "loss": 0.3192, "step": 7585 }, { "epoch": 1.32, "learning_rate": 4.620093239622403e-05, "loss": 0.3273, "step": 7586 }, { "epoch": 1.32, "learning_rate": 4.619712517144307e-05, "loss": 0.3057, "step": 7587 }, { "epoch": 1.32, "learning_rate": 4.619331621600777e-05, "loss": 0.3068, "step": 7588 }, { "epoch": 1.32, "learning_rate": 4.61895055302707e-05, "loss": 0.3024, "step": 7589 }, { "epoch": 1.32, "learning_rate": 4.6185693114584586e-05, "loss": 0.3059, "step": 7590 }, { "epoch": 1.32, "learning_rate": 4.618187896930231e-05, "loss": 0.3036, "step": 7591 }, { "epoch": 1.32, "learning_rate": 4.617806309477692e-05, "loss": 0.3101, "step": 7592 }, { "epoch": 1.32, "learning_rate": 4.6174245491361616e-05, "loss": 0.3023, "step": 7593 }, { "epoch": 1.32, "learning_rate": 4.6170426159409777e-05, "loss": 0.3126, "step": 7594 }, { "epoch": 1.32, "learning_rate": 4.616660509927493e-05, "loss": 0.297, "step": 7595 }, { "epoch": 1.32, "learning_rate": 4.616278231131076e-05, "loss": 0.3226, "step": 7596 }, { "epoch": 1.32, "learning_rate": 4.615895779587111e-05, "loss": 0.2924, "step": 7597 }, { "epoch": 1.32, "learning_rate": 4.6155131553309995e-05, "loss": 0.3047, "step": 7598 }, { "epoch": 1.32, "learning_rate": 4.615130358398157e-05, "loss": 0.3247, "step": 7599 }, { "epoch": 1.32, "learning_rate": 4.614747388824017e-05, "loss": 0.3055, "step": 7600 }, { "epoch": 1.32, "learning_rate": 4.6143642466440286e-05, "loss": 0.3034, "step": 7601 }, { "epoch": 1.33, "learning_rate": 4.613980931893655e-05, "loss": 0.2956, "step": 7602 }, { "epoch": 1.33, "learning_rate": 4.613597444608377e-05, "loss": 0.306, "step": 7603 }, { "epoch": 1.33, "learning_rate": 4.613213784823694e-05, "loss": 0.3111, "step": 7604 }, { "epoch": 1.33, "learning_rate": 4.612829952575114e-05, "loss": 0.3029, "step": 7605 }, { "epoch": 1.33, "learning_rate": 4.6124459478981705e-05, "loss": 0.3062, "step": 7606 }, { "epoch": 1.33, "learning_rate": 4.612061770828404e-05, "loss": 0.3131, "step": 7607 }, { "epoch": 1.33, "learning_rate": 4.611677421401376e-05, "loss": 0.316, "step": 7608 }, { "epoch": 1.33, "learning_rate": 4.6112928996526626e-05, "loss": 0.3132, "step": 7609 }, { "epoch": 1.33, "learning_rate": 4.610908205617857e-05, "loss": 0.2929, "step": 7610 }, { "epoch": 1.33, "learning_rate": 4.610523339332568e-05, "loss": 0.3137, "step": 7611 }, { "epoch": 1.33, "learning_rate": 4.6101383008324185e-05, "loss": 0.2961, "step": 7612 }, { "epoch": 1.33, "learning_rate": 4.609753090153049e-05, "loss": 0.3116, "step": 7613 }, { "epoch": 1.33, "learning_rate": 4.609367707330116e-05, "loss": 0.306, "step": 7614 }, { "epoch": 1.33, "learning_rate": 4.6089821523992916e-05, "loss": 0.3021, "step": 7615 }, { "epoch": 1.33, "learning_rate": 4.608596425396263e-05, "loss": 0.3033, "step": 7616 }, { "epoch": 1.33, "learning_rate": 4.608210526356734e-05, "loss": 0.3131, "step": 7617 }, { "epoch": 1.33, "learning_rate": 4.6078244553164254e-05, "loss": 0.3017, "step": 7618 }, { "epoch": 1.33, "learning_rate": 4.607438212311073e-05, "loss": 0.2942, "step": 7619 }, { "epoch": 1.33, "learning_rate": 4.607051797376427e-05, "loss": 0.299, "step": 7620 }, { "epoch": 1.33, "learning_rate": 4.6066652105482564e-05, "loss": 0.3033, "step": 7621 }, { "epoch": 1.33, "learning_rate": 4.606278451862344e-05, "loss": 0.2989, "step": 7622 }, { "epoch": 1.33, "learning_rate": 4.6058915213544895e-05, "loss": 0.31, "step": 7623 }, { "epoch": 1.33, "learning_rate": 4.605504419060508e-05, "loss": 0.307, "step": 7624 }, { "epoch": 1.33, "learning_rate": 4.60511714501623e-05, "loss": 0.3155, "step": 7625 }, { "epoch": 1.33, "learning_rate": 4.604729699257504e-05, "loss": 0.3109, "step": 7626 }, { "epoch": 1.33, "learning_rate": 4.604342081820192e-05, "loss": 0.3096, "step": 7627 }, { "epoch": 1.33, "learning_rate": 4.6039542927401744e-05, "loss": 0.309, "step": 7628 }, { "epoch": 1.33, "learning_rate": 4.603566332053343e-05, "loss": 0.3334, "step": 7629 }, { "epoch": 1.33, "learning_rate": 4.603178199795612e-05, "loss": 0.302, "step": 7630 }, { "epoch": 1.33, "learning_rate": 4.602789896002905e-05, "loss": 0.2997, "step": 7631 }, { "epoch": 1.33, "learning_rate": 4.602401420711165e-05, "loss": 0.3047, "step": 7632 }, { "epoch": 1.33, "learning_rate": 4.602012773956352e-05, "loss": 0.3108, "step": 7633 }, { "epoch": 1.33, "learning_rate": 4.6016239557744384e-05, "loss": 0.3077, "step": 7634 }, { "epoch": 1.33, "learning_rate": 4.6012349662014145e-05, "loss": 0.32, "step": 7635 }, { "epoch": 1.33, "learning_rate": 4.600845805273287e-05, "loss": 0.3145, "step": 7636 }, { "epoch": 1.33, "learning_rate": 4.6004564730260766e-05, "loss": 0.3117, "step": 7637 }, { "epoch": 1.33, "learning_rate": 4.600066969495822e-05, "loss": 0.3034, "step": 7638 }, { "epoch": 1.33, "learning_rate": 4.5996772947185754e-05, "loss": 0.3051, "step": 7639 }, { "epoch": 1.33, "learning_rate": 4.599287448730407e-05, "loss": 0.3093, "step": 7640 }, { "epoch": 1.33, "learning_rate": 4.598897431567401e-05, "loss": 0.3252, "step": 7641 }, { "epoch": 1.33, "learning_rate": 4.5985072432656614e-05, "loss": 0.321, "step": 7642 }, { "epoch": 1.33, "learning_rate": 4.598116883861301e-05, "loss": 0.3089, "step": 7643 }, { "epoch": 1.33, "learning_rate": 4.597726353390455e-05, "loss": 0.3025, "step": 7644 }, { "epoch": 1.33, "learning_rate": 4.59733565188927e-05, "loss": 0.3098, "step": 7645 }, { "epoch": 1.33, "learning_rate": 4.5969447793939115e-05, "loss": 0.3022, "step": 7646 }, { "epoch": 1.33, "learning_rate": 4.59655373594056e-05, "loss": 0.3372, "step": 7647 }, { "epoch": 1.33, "learning_rate": 4.596162521565412e-05, "loss": 0.3214, "step": 7648 }, { "epoch": 1.33, "learning_rate": 4.595771136304678e-05, "loss": 0.302, "step": 7649 }, { "epoch": 1.33, "learning_rate": 4.595379580194586e-05, "loss": 0.3069, "step": 7650 }, { "epoch": 1.33, "learning_rate": 4.594987853271379e-05, "loss": 0.3232, "step": 7651 }, { "epoch": 1.33, "learning_rate": 4.5945959555713175e-05, "loss": 0.3024, "step": 7652 }, { "epoch": 1.33, "learning_rate": 4.5942038871306756e-05, "loss": 0.3187, "step": 7653 }, { "epoch": 1.33, "learning_rate": 4.593811647985745e-05, "loss": 0.3209, "step": 7654 }, { "epoch": 1.33, "learning_rate": 4.5934192381728316e-05, "loss": 0.3214, "step": 7655 }, { "epoch": 1.33, "learning_rate": 4.5930266577282576e-05, "loss": 0.2988, "step": 7656 }, { "epoch": 1.33, "learning_rate": 4.592633906688362e-05, "loss": 0.2996, "step": 7657 }, { "epoch": 1.33, "learning_rate": 4.592240985089498e-05, "loss": 0.3028, "step": 7658 }, { "epoch": 1.33, "learning_rate": 4.591847892968037e-05, "loss": 0.3077, "step": 7659 }, { "epoch": 1.34, "learning_rate": 4.5914546303603633e-05, "loss": 0.3107, "step": 7660 }, { "epoch": 1.34, "learning_rate": 4.591061197302879e-05, "loss": 0.315, "step": 7661 }, { "epoch": 1.34, "learning_rate": 4.5906675938319996e-05, "loss": 0.307, "step": 7662 }, { "epoch": 1.34, "learning_rate": 4.590273819984161e-05, "loss": 0.306, "step": 7663 }, { "epoch": 1.34, "learning_rate": 4.589879875795809e-05, "loss": 0.3157, "step": 7664 }, { "epoch": 1.34, "learning_rate": 4.589485761303411e-05, "loss": 0.2974, "step": 7665 }, { "epoch": 1.34, "learning_rate": 4.5890914765434445e-05, "loss": 0.3073, "step": 7666 }, { "epoch": 1.34, "learning_rate": 4.588697021552407e-05, "loss": 0.331, "step": 7667 }, { "epoch": 1.34, "learning_rate": 4.58830239636681e-05, "loss": 0.3046, "step": 7668 }, { "epoch": 1.34, "learning_rate": 4.587907601023181e-05, "loss": 0.3028, "step": 7669 }, { "epoch": 1.34, "learning_rate": 4.587512635558064e-05, "loss": 0.3147, "step": 7670 }, { "epoch": 1.34, "learning_rate": 4.587117500008017e-05, "loss": 0.3217, "step": 7671 }, { "epoch": 1.34, "learning_rate": 4.5867221944096144e-05, "loss": 0.2994, "step": 7672 }, { "epoch": 1.34, "learning_rate": 4.586326718799448e-05, "loss": 0.3206, "step": 7673 }, { "epoch": 1.34, "learning_rate": 4.585931073214124e-05, "loss": 0.3157, "step": 7674 }, { "epoch": 1.34, "learning_rate": 4.5855352576902624e-05, "loss": 0.3096, "step": 7675 }, { "epoch": 1.34, "learning_rate": 4.585139272264503e-05, "loss": 0.2984, "step": 7676 }, { "epoch": 1.34, "learning_rate": 4.584743116973499e-05, "loss": 0.304, "step": 7677 }, { "epoch": 1.34, "learning_rate": 4.584346791853918e-05, "loss": 0.3174, "step": 7678 }, { "epoch": 1.34, "learning_rate": 4.583950296942448e-05, "loss": 0.3101, "step": 7679 }, { "epoch": 1.34, "learning_rate": 4.583553632275786e-05, "loss": 0.2952, "step": 7680 }, { "epoch": 1.34, "learning_rate": 4.5831567978906504e-05, "loss": 0.3099, "step": 7681 }, { "epoch": 1.34, "learning_rate": 4.582759793823773e-05, "loss": 0.3142, "step": 7682 }, { "epoch": 1.34, "learning_rate": 4.5823626201119e-05, "loss": 0.293, "step": 7683 }, { "epoch": 1.34, "learning_rate": 4.581965276791798e-05, "loss": 0.315, "step": 7684 }, { "epoch": 1.34, "learning_rate": 4.581567763900243e-05, "loss": 0.3151, "step": 7685 }, { "epoch": 1.34, "learning_rate": 4.5811700814740305e-05, "loss": 0.3218, "step": 7686 }, { "epoch": 1.34, "learning_rate": 4.5807722295499734e-05, "loss": 0.3196, "step": 7687 }, { "epoch": 1.34, "learning_rate": 4.580374208164894e-05, "loss": 0.3144, "step": 7688 }, { "epoch": 1.34, "learning_rate": 4.5799760173556386e-05, "loss": 0.3083, "step": 7689 }, { "epoch": 1.34, "learning_rate": 4.579577657159061e-05, "loss": 0.2953, "step": 7690 }, { "epoch": 1.34, "learning_rate": 4.5791791276120353e-05, "loss": 0.3002, "step": 7691 }, { "epoch": 1.34, "learning_rate": 4.578780428751452e-05, "loss": 0.2981, "step": 7692 }, { "epoch": 1.34, "learning_rate": 4.5783815606142145e-05, "loss": 0.3061, "step": 7693 }, { "epoch": 1.34, "learning_rate": 4.577982523237243e-05, "loss": 0.3053, "step": 7694 }, { "epoch": 1.34, "learning_rate": 4.577583316657474e-05, "loss": 0.3258, "step": 7695 }, { "epoch": 1.34, "learning_rate": 4.577183940911859e-05, "loss": 0.3007, "step": 7696 }, { "epoch": 1.34, "learning_rate": 4.576784396037364e-05, "loss": 0.2953, "step": 7697 }, { "epoch": 1.34, "learning_rate": 4.576384682070974e-05, "loss": 0.3062, "step": 7698 }, { "epoch": 1.34, "learning_rate": 4.575984799049686e-05, "loss": 0.3133, "step": 7699 }, { "epoch": 1.34, "learning_rate": 4.575584747010515e-05, "loss": 0.3026, "step": 7700 }, { "epoch": 1.34, "learning_rate": 4.5751845259904905e-05, "loss": 0.3144, "step": 7701 }, { "epoch": 1.34, "learning_rate": 4.574784136026659e-05, "loss": 0.3258, "step": 7702 }, { "epoch": 1.34, "learning_rate": 4.57438357715608e-05, "loss": 0.312, "step": 7703 }, { "epoch": 1.34, "learning_rate": 4.5739828494158305e-05, "loss": 0.31, "step": 7704 }, { "epoch": 1.34, "learning_rate": 4.573581952843004e-05, "loss": 0.2887, "step": 7705 }, { "epoch": 1.34, "learning_rate": 4.573180887474707e-05, "loss": 0.3059, "step": 7706 }, { "epoch": 1.34, "learning_rate": 4.572779653348065e-05, "loss": 0.3265, "step": 7707 }, { "epoch": 1.34, "learning_rate": 4.5723782505002156e-05, "loss": 0.3059, "step": 7708 }, { "epoch": 1.34, "learning_rate": 4.5719766789683154e-05, "loss": 0.3345, "step": 7709 }, { "epoch": 1.34, "learning_rate": 4.571574938789533e-05, "loss": 0.3238, "step": 7710 }, { "epoch": 1.34, "learning_rate": 4.571173030001056e-05, "loss": 0.3097, "step": 7711 }, { "epoch": 1.34, "learning_rate": 4.570770952640085e-05, "loss": 0.311, "step": 7712 }, { "epoch": 1.34, "learning_rate": 4.570368706743838e-05, "loss": 0.3089, "step": 7713 }, { "epoch": 1.34, "learning_rate": 4.569966292349548e-05, "loss": 0.3186, "step": 7714 }, { "epoch": 1.34, "learning_rate": 4.569563709494464e-05, "loss": 0.3145, "step": 7715 }, { "epoch": 1.34, "learning_rate": 4.569160958215848e-05, "loss": 0.3043, "step": 7716 }, { "epoch": 1.35, "learning_rate": 4.568758038550981e-05, "loss": 0.2892, "step": 7717 }, { "epoch": 1.35, "learning_rate": 4.56835495053716e-05, "loss": 0.3257, "step": 7718 }, { "epoch": 1.35, "learning_rate": 4.5679516942116935e-05, "loss": 0.3105, "step": 7719 }, { "epoch": 1.35, "learning_rate": 4.567548269611909e-05, "loss": 0.2974, "step": 7720 }, { "epoch": 1.35, "learning_rate": 4.567144676775148e-05, "loss": 0.2936, "step": 7721 }, { "epoch": 1.35, "learning_rate": 4.566740915738769e-05, "loss": 0.3059, "step": 7722 }, { "epoch": 1.35, "learning_rate": 4.5663369865401435e-05, "loss": 0.3088, "step": 7723 }, { "epoch": 1.35, "learning_rate": 4.565932889216661e-05, "loss": 0.3213, "step": 7724 }, { "epoch": 1.35, "learning_rate": 4.565528623805727e-05, "loss": 0.3119, "step": 7725 }, { "epoch": 1.35, "learning_rate": 4.56512419034476e-05, "loss": 0.3033, "step": 7726 }, { "epoch": 1.35, "learning_rate": 4.5647195888711954e-05, "loss": 0.3129, "step": 7727 }, { "epoch": 1.35, "learning_rate": 4.564314819422485e-05, "loss": 0.3205, "step": 7728 }, { "epoch": 1.35, "learning_rate": 4.563909882036095e-05, "loss": 0.3162, "step": 7729 }, { "epoch": 1.35, "learning_rate": 4.563504776749507e-05, "loss": 0.3001, "step": 7730 }, { "epoch": 1.35, "learning_rate": 4.56309950360022e-05, "loss": 0.3138, "step": 7731 }, { "epoch": 1.35, "learning_rate": 4.5626940626257446e-05, "loss": 0.3087, "step": 7732 }, { "epoch": 1.35, "learning_rate": 4.5622884538636114e-05, "loss": 0.3138, "step": 7733 }, { "epoch": 1.35, "learning_rate": 4.561882677351365e-05, "loss": 0.2974, "step": 7734 }, { "epoch": 1.35, "learning_rate": 4.561476733126563e-05, "loss": 0.314, "step": 7735 }, { "epoch": 1.35, "learning_rate": 4.561070621226782e-05, "loss": 0.3089, "step": 7736 }, { "epoch": 1.35, "learning_rate": 4.5606643416896134e-05, "loss": 0.2948, "step": 7737 }, { "epoch": 1.35, "learning_rate": 4.5602578945526625e-05, "loss": 0.3205, "step": 7738 }, { "epoch": 1.35, "learning_rate": 4.5598512798535513e-05, "loss": 0.303, "step": 7739 }, { "epoch": 1.35, "learning_rate": 4.559444497629917e-05, "loss": 0.3166, "step": 7740 }, { "epoch": 1.35, "learning_rate": 4.559037547919414e-05, "loss": 0.2997, "step": 7741 }, { "epoch": 1.35, "learning_rate": 4.558630430759707e-05, "loss": 0.3213, "step": 7742 }, { "epoch": 1.35, "learning_rate": 4.5582231461884834e-05, "loss": 0.309, "step": 7743 }, { "epoch": 1.35, "learning_rate": 4.557815694243441e-05, "loss": 0.3054, "step": 7744 }, { "epoch": 1.35, "learning_rate": 4.5574080749622945e-05, "loss": 0.2939, "step": 7745 }, { "epoch": 1.35, "learning_rate": 4.557000288382774e-05, "loss": 0.2997, "step": 7746 }, { "epoch": 1.35, "learning_rate": 4.556592334542626e-05, "loss": 0.3043, "step": 7747 }, { "epoch": 1.35, "learning_rate": 4.5561842134796115e-05, "loss": 0.3124, "step": 7748 }, { "epoch": 1.35, "learning_rate": 4.555775925231507e-05, "loss": 0.3136, "step": 7749 }, { "epoch": 1.35, "learning_rate": 4.555367469836105e-05, "loss": 0.3067, "step": 7750 }, { "epoch": 1.35, "learning_rate": 4.554958847331214e-05, "loss": 0.3142, "step": 7751 }, { "epoch": 1.35, "learning_rate": 4.5545500577546555e-05, "loss": 0.3036, "step": 7752 }, { "epoch": 1.35, "learning_rate": 4.55414110114427e-05, "loss": 0.3141, "step": 7753 }, { "epoch": 1.35, "learning_rate": 4.553731977537909e-05, "loss": 0.296, "step": 7754 }, { "epoch": 1.35, "learning_rate": 4.5533226869734444e-05, "loss": 0.3096, "step": 7755 }, { "epoch": 1.35, "learning_rate": 4.55291322948876e-05, "loss": 0.3054, "step": 7756 }, { "epoch": 1.35, "learning_rate": 4.552503605121756e-05, "loss": 0.3235, "step": 7757 }, { "epoch": 1.35, "learning_rate": 4.55209381391035e-05, "loss": 0.3154, "step": 7758 }, { "epoch": 1.35, "learning_rate": 4.551683855892472e-05, "loss": 0.3089, "step": 7759 }, { "epoch": 1.35, "learning_rate": 4.551273731106069e-05, "loss": 0.2983, "step": 7760 }, { "epoch": 1.35, "learning_rate": 4.550863439589102e-05, "loss": 0.3012, "step": 7761 }, { "epoch": 1.35, "learning_rate": 4.550452981379551e-05, "loss": 0.3274, "step": 7762 }, { "epoch": 1.35, "learning_rate": 4.5500423565154076e-05, "loss": 0.3009, "step": 7763 }, { "epoch": 1.35, "learning_rate": 4.549631565034681e-05, "loss": 0.3069, "step": 7764 }, { "epoch": 1.35, "learning_rate": 4.5492206069753943e-05, "loss": 0.2926, "step": 7765 }, { "epoch": 1.35, "learning_rate": 4.548809482375588e-05, "loss": 0.3039, "step": 7766 }, { "epoch": 1.35, "learning_rate": 4.5483981912733154e-05, "loss": 0.3032, "step": 7767 }, { "epoch": 1.35, "learning_rate": 4.5479867337066476e-05, "loss": 0.3115, "step": 7768 }, { "epoch": 1.35, "learning_rate": 4.54757510971367e-05, "loss": 0.3033, "step": 7769 }, { "epoch": 1.35, "learning_rate": 4.547163319332484e-05, "loss": 0.3076, "step": 7770 }, { "epoch": 1.35, "learning_rate": 4.5467513626012046e-05, "loss": 0.3127, "step": 7771 }, { "epoch": 1.35, "learning_rate": 4.5463392395579654e-05, "loss": 0.2986, "step": 7772 }, { "epoch": 1.35, "learning_rate": 4.5459269502409125e-05, "loss": 0.3051, "step": 7773 }, { "epoch": 1.36, "learning_rate": 4.5455144946882084e-05, "loss": 0.312, "step": 7774 }, { "epoch": 1.36, "learning_rate": 4.545101872938031e-05, "loss": 0.3086, "step": 7775 }, { "epoch": 1.36, "learning_rate": 4.5446890850285757e-05, "loss": 0.3096, "step": 7776 }, { "epoch": 1.36, "learning_rate": 4.5442761309980473e-05, "loss": 0.3183, "step": 7777 }, { "epoch": 1.36, "learning_rate": 4.543863010884673e-05, "loss": 0.3152, "step": 7778 }, { "epoch": 1.36, "learning_rate": 4.5434497247266914e-05, "loss": 0.3015, "step": 7779 }, { "epoch": 1.36, "learning_rate": 4.5430362725623575e-05, "loss": 0.315, "step": 7780 }, { "epoch": 1.36, "learning_rate": 4.542622654429941e-05, "loss": 0.3128, "step": 7781 }, { "epoch": 1.36, "learning_rate": 4.5422088703677284e-05, "loss": 0.3154, "step": 7782 }, { "epoch": 1.36, "learning_rate": 4.54179492041402e-05, "loss": 0.2887, "step": 7783 }, { "epoch": 1.36, "learning_rate": 4.541380804607131e-05, "loss": 0.3088, "step": 7784 }, { "epoch": 1.36, "learning_rate": 4.540966522985395e-05, "loss": 0.305, "step": 7785 }, { "epoch": 1.36, "learning_rate": 4.540552075587158e-05, "loss": 0.3134, "step": 7786 }, { "epoch": 1.36, "learning_rate": 4.540137462450782e-05, "loss": 0.3076, "step": 7787 }, { "epoch": 1.36, "learning_rate": 4.539722683614645e-05, "loss": 0.2964, "step": 7788 }, { "epoch": 1.36, "learning_rate": 4.5393077391171404e-05, "loss": 0.314, "step": 7789 }, { "epoch": 1.36, "learning_rate": 4.538892628996677e-05, "loss": 0.3218, "step": 7790 }, { "epoch": 1.36, "learning_rate": 4.5384773532916754e-05, "loss": 0.3135, "step": 7791 }, { "epoch": 1.36, "learning_rate": 4.538061912040579e-05, "loss": 0.3146, "step": 7792 }, { "epoch": 1.36, "learning_rate": 4.537646305281839e-05, "loss": 0.3132, "step": 7793 }, { "epoch": 1.36, "learning_rate": 4.5372305330539266e-05, "loss": 0.3157, "step": 7794 }, { "epoch": 1.36, "learning_rate": 4.536814595395326e-05, "loss": 0.3088, "step": 7795 }, { "epoch": 1.36, "learning_rate": 4.536398492344537e-05, "loss": 0.3234, "step": 7796 }, { "epoch": 1.36, "learning_rate": 4.535982223940077e-05, "loss": 0.3157, "step": 7797 }, { "epoch": 1.36, "learning_rate": 4.535565790220475e-05, "loss": 0.3119, "step": 7798 }, { "epoch": 1.36, "learning_rate": 4.5351491912242785e-05, "loss": 0.3, "step": 7799 }, { "epoch": 1.36, "learning_rate": 4.534732426990047e-05, "loss": 0.3029, "step": 7800 }, { "epoch": 1.36, "learning_rate": 4.53431549755636e-05, "loss": 0.3262, "step": 7801 }, { "epoch": 1.36, "learning_rate": 4.5338984029618076e-05, "loss": 0.3105, "step": 7802 }, { "epoch": 1.36, "learning_rate": 4.533481143244998e-05, "loss": 0.3092, "step": 7803 }, { "epoch": 1.36, "learning_rate": 4.533063718444554e-05, "loss": 0.3174, "step": 7804 }, { "epoch": 1.36, "learning_rate": 4.532646128599112e-05, "loss": 0.3037, "step": 7805 }, { "epoch": 1.36, "learning_rate": 4.532228373747327e-05, "loss": 0.309, "step": 7806 }, { "epoch": 1.36, "learning_rate": 4.531810453927868e-05, "loss": 0.308, "step": 7807 }, { "epoch": 1.36, "learning_rate": 4.531392369179416e-05, "loss": 0.3039, "step": 7808 }, { "epoch": 1.36, "learning_rate": 4.5309741195406716e-05, "loss": 0.3007, "step": 7809 }, { "epoch": 1.36, "learning_rate": 4.53055570505035e-05, "loss": 0.311, "step": 7810 }, { "epoch": 1.36, "learning_rate": 4.5301371257471785e-05, "loss": 0.3177, "step": 7811 }, { "epoch": 1.36, "learning_rate": 4.529718381669905e-05, "loss": 0.3041, "step": 7812 }, { "epoch": 1.36, "learning_rate": 4.529299472857286e-05, "loss": 0.3097, "step": 7813 }, { "epoch": 1.36, "learning_rate": 4.528880399348099e-05, "loss": 0.305, "step": 7814 }, { "epoch": 1.36, "learning_rate": 4.528461161181134e-05, "loss": 0.2958, "step": 7815 }, { "epoch": 1.36, "learning_rate": 4.528041758395196e-05, "loss": 0.2981, "step": 7816 }, { "epoch": 1.36, "learning_rate": 4.5276221910291085e-05, "loss": 0.3099, "step": 7817 }, { "epoch": 1.36, "learning_rate": 4.5272024591217046e-05, "loss": 0.3058, "step": 7818 }, { "epoch": 1.36, "learning_rate": 4.5267825627118375e-05, "loss": 0.3224, "step": 7819 }, { "epoch": 1.36, "learning_rate": 4.526362501838374e-05, "loss": 0.3118, "step": 7820 }, { "epoch": 1.36, "learning_rate": 4.525942276540195e-05, "loss": 0.3081, "step": 7821 }, { "epoch": 1.36, "learning_rate": 4.5255218868561986e-05, "loss": 0.3081, "step": 7822 }, { "epoch": 1.36, "learning_rate": 4.5251013328252975e-05, "loss": 0.3054, "step": 7823 }, { "epoch": 1.36, "learning_rate": 4.5246806144864174e-05, "loss": 0.3122, "step": 7824 }, { "epoch": 1.36, "learning_rate": 4.524259731878503e-05, "loss": 0.3133, "step": 7825 }, { "epoch": 1.36, "learning_rate": 4.523838685040511e-05, "loss": 0.3075, "step": 7826 }, { "epoch": 1.36, "learning_rate": 4.523417474011416e-05, "loss": 0.3014, "step": 7827 }, { "epoch": 1.36, "learning_rate": 4.522996098830205e-05, "loss": 0.3131, "step": 7828 }, { "epoch": 1.36, "learning_rate": 4.522574559535882e-05, "loss": 0.3089, "step": 7829 }, { "epoch": 1.36, "learning_rate": 4.522152856167466e-05, "loss": 0.3028, "step": 7830 }, { "epoch": 1.36, "learning_rate": 4.52173098876399e-05, "loss": 0.3038, "step": 7831 }, { "epoch": 1.37, "learning_rate": 4.5213089573645046e-05, "loss": 0.3119, "step": 7832 }, { "epoch": 1.37, "learning_rate": 4.520886762008074e-05, "loss": 0.3166, "step": 7833 }, { "epoch": 1.37, "learning_rate": 4.520464402733777e-05, "loss": 0.3132, "step": 7834 }, { "epoch": 1.37, "learning_rate": 4.5200418795807076e-05, "loss": 0.3201, "step": 7835 }, { "epoch": 1.37, "learning_rate": 4.519619192587977e-05, "loss": 0.3024, "step": 7836 }, { "epoch": 1.37, "learning_rate": 4.5191963417947106e-05, "loss": 0.2982, "step": 7837 }, { "epoch": 1.37, "learning_rate": 4.518773327240047e-05, "loss": 0.3144, "step": 7838 }, { "epoch": 1.37, "learning_rate": 4.5183501489631415e-05, "loss": 0.3144, "step": 7839 }, { "epoch": 1.37, "learning_rate": 4.5179268070031656e-05, "loss": 0.3087, "step": 7840 }, { "epoch": 1.37, "learning_rate": 4.517503301399304e-05, "loss": 0.3131, "step": 7841 }, { "epoch": 1.37, "learning_rate": 4.5170796321907594e-05, "loss": 0.2942, "step": 7842 }, { "epoch": 1.37, "learning_rate": 4.516655799416746e-05, "loss": 0.306, "step": 7843 }, { "epoch": 1.37, "learning_rate": 4.5162318031164946e-05, "loss": 0.3182, "step": 7844 }, { "epoch": 1.37, "learning_rate": 4.5158076433292526e-05, "loss": 0.3014, "step": 7845 }, { "epoch": 1.37, "learning_rate": 4.51538332009428e-05, "loss": 0.2966, "step": 7846 }, { "epoch": 1.37, "learning_rate": 4.514958833450855e-05, "loss": 0.3139, "step": 7847 }, { "epoch": 1.37, "learning_rate": 4.514534183438268e-05, "loss": 0.33, "step": 7848 }, { "epoch": 1.37, "learning_rate": 4.514109370095826e-05, "loss": 0.3136, "step": 7849 }, { "epoch": 1.37, "learning_rate": 4.51368439346285e-05, "loss": 0.2921, "step": 7850 }, { "epoch": 1.37, "learning_rate": 4.513259253578678e-05, "loss": 0.3084, "step": 7851 }, { "epoch": 1.37, "learning_rate": 4.5128339504826615e-05, "loss": 0.3287, "step": 7852 }, { "epoch": 1.37, "learning_rate": 4.5124084842141684e-05, "loss": 0.3028, "step": 7853 }, { "epoch": 1.37, "learning_rate": 4.51198285481258e-05, "loss": 0.3028, "step": 7854 }, { "epoch": 1.37, "learning_rate": 4.511557062317295e-05, "loss": 0.3256, "step": 7855 }, { "epoch": 1.37, "learning_rate": 4.5111311067677234e-05, "loss": 0.3072, "step": 7856 }, { "epoch": 1.37, "learning_rate": 4.510704988203295e-05, "loss": 0.3238, "step": 7857 }, { "epoch": 1.37, "learning_rate": 4.5102787066634526e-05, "loss": 0.3, "step": 7858 }, { "epoch": 1.37, "learning_rate": 4.5098522621876515e-05, "loss": 0.3061, "step": 7859 }, { "epoch": 1.37, "learning_rate": 4.509425654815368e-05, "loss": 0.3034, "step": 7860 }, { "epoch": 1.37, "learning_rate": 4.508998884586087e-05, "loss": 0.3009, "step": 7861 }, { "epoch": 1.37, "learning_rate": 4.5085719515393114e-05, "loss": 0.3127, "step": 7862 }, { "epoch": 1.37, "learning_rate": 4.5081448557145614e-05, "loss": 0.3078, "step": 7863 }, { "epoch": 1.37, "learning_rate": 4.5077175971513696e-05, "loss": 0.3216, "step": 7864 }, { "epoch": 1.37, "learning_rate": 4.507290175889283e-05, "loss": 0.3089, "step": 7865 }, { "epoch": 1.37, "learning_rate": 4.506862591967865e-05, "loss": 0.3147, "step": 7866 }, { "epoch": 1.37, "learning_rate": 4.5064348454266954e-05, "loss": 0.3038, "step": 7867 }, { "epoch": 1.37, "learning_rate": 4.506006936305365e-05, "loss": 0.3145, "step": 7868 }, { "epoch": 1.37, "learning_rate": 4.505578864643485e-05, "loss": 0.2986, "step": 7869 }, { "epoch": 1.37, "learning_rate": 4.505150630480676e-05, "loss": 0.3028, "step": 7870 }, { "epoch": 1.37, "learning_rate": 4.504722233856579e-05, "loss": 0.3033, "step": 7871 }, { "epoch": 1.37, "learning_rate": 4.504293674810846e-05, "loss": 0.2973, "step": 7872 }, { "epoch": 1.37, "learning_rate": 4.503864953383146e-05, "loss": 0.3157, "step": 7873 }, { "epoch": 1.37, "learning_rate": 4.503436069613162e-05, "loss": 0.2945, "step": 7874 }, { "epoch": 1.37, "learning_rate": 4.503007023540594e-05, "loss": 0.3148, "step": 7875 }, { "epoch": 1.37, "learning_rate": 4.502577815205154e-05, "loss": 0.3073, "step": 7876 }, { "epoch": 1.37, "learning_rate": 4.502148444646571e-05, "loss": 0.3114, "step": 7877 }, { "epoch": 1.37, "learning_rate": 4.5017189119045896e-05, "loss": 0.3048, "step": 7878 }, { "epoch": 1.37, "learning_rate": 4.501289217018968e-05, "loss": 0.3232, "step": 7879 }, { "epoch": 1.37, "learning_rate": 4.500859360029479e-05, "loss": 0.3118, "step": 7880 }, { "epoch": 1.37, "learning_rate": 4.500429340975913e-05, "loss": 0.3071, "step": 7881 }, { "epoch": 1.37, "learning_rate": 4.499999159898071e-05, "loss": 0.322, "step": 7882 }, { "epoch": 1.37, "learning_rate": 4.4995688168357747e-05, "loss": 0.2982, "step": 7883 }, { "epoch": 1.37, "learning_rate": 4.499138311828856e-05, "loss": 0.296, "step": 7884 }, { "epoch": 1.37, "learning_rate": 4.498707644917164e-05, "loss": 0.3179, "step": 7885 }, { "epoch": 1.37, "learning_rate": 4.498276816140562e-05, "loss": 0.3139, "step": 7886 }, { "epoch": 1.37, "learning_rate": 4.497845825538929e-05, "loss": 0.3208, "step": 7887 }, { "epoch": 1.37, "learning_rate": 4.497414673152159e-05, "loss": 0.3261, "step": 7888 }, { "epoch": 1.38, "learning_rate": 4.49698335902016e-05, "loss": 0.3144, "step": 7889 }, { "epoch": 1.38, "learning_rate": 4.496551883182856e-05, "loss": 0.3082, "step": 7890 }, { "epoch": 1.38, "learning_rate": 4.496120245680185e-05, "loss": 0.3078, "step": 7891 }, { "epoch": 1.38, "learning_rate": 4.4956884465521e-05, "loss": 0.3183, "step": 7892 }, { "epoch": 1.38, "learning_rate": 4.495256485838571e-05, "loss": 0.3023, "step": 7893 }, { "epoch": 1.38, "learning_rate": 4.4948243635795794e-05, "loss": 0.3177, "step": 7894 }, { "epoch": 1.38, "learning_rate": 4.4943920798151255e-05, "loss": 0.307, "step": 7895 }, { "epoch": 1.38, "learning_rate": 4.493959634585222e-05, "loss": 0.3305, "step": 7896 }, { "epoch": 1.38, "learning_rate": 4.4935270279298965e-05, "loss": 0.306, "step": 7897 }, { "epoch": 1.38, "learning_rate": 4.493094259889193e-05, "loss": 0.3196, "step": 7898 }, { "epoch": 1.38, "learning_rate": 4.4926613305031696e-05, "loss": 0.3072, "step": 7899 }, { "epoch": 1.38, "learning_rate": 4.492228239811898e-05, "loss": 0.3215, "step": 7900 }, { "epoch": 1.38, "learning_rate": 4.4917949878554675e-05, "loss": 0.3194, "step": 7901 }, { "epoch": 1.38, "learning_rate": 4.491361574673982e-05, "loss": 0.3028, "step": 7902 }, { "epoch": 1.38, "learning_rate": 4.490928000307557e-05, "loss": 0.3089, "step": 7903 }, { "epoch": 1.38, "learning_rate": 4.4904942647963256e-05, "loss": 0.3232, "step": 7904 }, { "epoch": 1.38, "learning_rate": 4.490060368180438e-05, "loss": 0.3028, "step": 7905 }, { "epoch": 1.38, "learning_rate": 4.489626310500054e-05, "loss": 0.3001, "step": 7906 }, { "epoch": 1.38, "learning_rate": 4.489192091795352e-05, "loss": 0.3052, "step": 7907 }, { "epoch": 1.38, "learning_rate": 4.4887577121065244e-05, "loss": 0.3124, "step": 7908 }, { "epoch": 1.38, "learning_rate": 4.48832317147378e-05, "loss": 0.294, "step": 7909 }, { "epoch": 1.38, "learning_rate": 4.487888469937339e-05, "loss": 0.3042, "step": 7910 }, { "epoch": 1.38, "learning_rate": 4.487453607537438e-05, "loss": 0.3079, "step": 7911 }, { "epoch": 1.38, "learning_rate": 4.487018584314331e-05, "loss": 0.311, "step": 7912 }, { "epoch": 1.38, "learning_rate": 4.4865834003082854e-05, "loss": 0.3149, "step": 7913 }, { "epoch": 1.38, "learning_rate": 4.4861480555595794e-05, "loss": 0.3086, "step": 7914 }, { "epoch": 1.38, "learning_rate": 4.485712550108513e-05, "loss": 0.3205, "step": 7915 }, { "epoch": 1.38, "learning_rate": 4.485276883995397e-05, "loss": 0.3374, "step": 7916 }, { "epoch": 1.38, "learning_rate": 4.4848410572605565e-05, "loss": 0.3177, "step": 7917 }, { "epoch": 1.38, "learning_rate": 4.484405069944334e-05, "loss": 0.2945, "step": 7918 }, { "epoch": 1.38, "learning_rate": 4.483968922087085e-05, "loss": 0.303, "step": 7919 }, { "epoch": 1.38, "learning_rate": 4.483532613729181e-05, "loss": 0.3066, "step": 7920 }, { "epoch": 1.38, "learning_rate": 4.483096144911007e-05, "loss": 0.2981, "step": 7921 }, { "epoch": 1.38, "learning_rate": 4.482659515672964e-05, "loss": 0.3159, "step": 7922 }, { "epoch": 1.38, "learning_rate": 4.482222726055468e-05, "loss": 0.2947, "step": 7923 }, { "epoch": 1.38, "learning_rate": 4.481785776098949e-05, "loss": 0.3216, "step": 7924 }, { "epoch": 1.38, "learning_rate": 4.481348665843852e-05, "loss": 0.3194, "step": 7925 }, { "epoch": 1.38, "learning_rate": 4.480911395330638e-05, "loss": 0.3088, "step": 7926 }, { "epoch": 1.38, "learning_rate": 4.480473964599782e-05, "loss": 0.3081, "step": 7927 }, { "epoch": 1.38, "learning_rate": 4.480036373691772e-05, "loss": 0.2852, "step": 7928 }, { "epoch": 1.38, "learning_rate": 4.4795986226471136e-05, "loss": 0.3055, "step": 7929 }, { "epoch": 1.38, "learning_rate": 4.479160711506326e-05, "loss": 0.3085, "step": 7930 }, { "epoch": 1.38, "learning_rate": 4.4787226403099444e-05, "loss": 0.3419, "step": 7931 }, { "epoch": 1.38, "learning_rate": 4.478284409098516e-05, "loss": 0.3091, "step": 7932 }, { "epoch": 1.38, "learning_rate": 4.477846017912606e-05, "loss": 0.3202, "step": 7933 }, { "epoch": 1.38, "learning_rate": 4.4774074667927916e-05, "loss": 0.3024, "step": 7934 }, { "epoch": 1.38, "learning_rate": 4.476968755779668e-05, "loss": 0.3057, "step": 7935 }, { "epoch": 1.38, "learning_rate": 4.476529884913843e-05, "loss": 0.3186, "step": 7936 }, { "epoch": 1.38, "learning_rate": 4.4760908542359386e-05, "loss": 0.3027, "step": 7937 }, { "epoch": 1.38, "learning_rate": 4.4756516637865935e-05, "loss": 0.3207, "step": 7938 }, { "epoch": 1.38, "learning_rate": 4.47521231360646e-05, "loss": 0.3049, "step": 7939 }, { "epoch": 1.38, "learning_rate": 4.474772803736205e-05, "loss": 0.3118, "step": 7940 }, { "epoch": 1.38, "learning_rate": 4.474333134216512e-05, "loss": 0.3198, "step": 7941 }, { "epoch": 1.38, "learning_rate": 4.473893305088077e-05, "loss": 0.3133, "step": 7942 }, { "epoch": 1.38, "learning_rate": 4.4734533163916115e-05, "loss": 0.3235, "step": 7943 }, { "epoch": 1.38, "learning_rate": 4.473013168167843e-05, "loss": 0.3095, "step": 7944 }, { "epoch": 1.38, "learning_rate": 4.4725728604575114e-05, "loss": 0.3096, "step": 7945 }, { "epoch": 1.38, "learning_rate": 4.4721323933013735e-05, "loss": 0.3058, "step": 7946 }, { "epoch": 1.39, "learning_rate": 4.4716917667402005e-05, "loss": 0.314, "step": 7947 }, { "epoch": 1.39, "learning_rate": 4.471250980814778e-05, "loss": 0.3223, "step": 7948 }, { "epoch": 1.39, "learning_rate": 4.470810035565905e-05, "loss": 0.3136, "step": 7949 }, { "epoch": 1.39, "learning_rate": 4.470368931034397e-05, "loss": 0.3126, "step": 7950 }, { "epoch": 1.39, "learning_rate": 4.469927667261085e-05, "loss": 0.3044, "step": 7951 }, { "epoch": 1.39, "learning_rate": 4.4694862442868114e-05, "loss": 0.3043, "step": 7952 }, { "epoch": 1.39, "learning_rate": 4.469044662152437e-05, "loss": 0.3127, "step": 7953 }, { "epoch": 1.39, "learning_rate": 4.468602920898836e-05, "loss": 0.3045, "step": 7954 }, { "epoch": 1.39, "learning_rate": 4.468161020566896e-05, "loss": 0.3148, "step": 7955 }, { "epoch": 1.39, "learning_rate": 4.4677189611975216e-05, "loss": 0.3033, "step": 7956 }, { "epoch": 1.39, "learning_rate": 4.4672767428316304e-05, "loss": 0.308, "step": 7957 }, { "epoch": 1.39, "learning_rate": 4.466834365510155e-05, "loss": 0.3043, "step": 7958 }, { "epoch": 1.39, "learning_rate": 4.466391829274043e-05, "loss": 0.3071, "step": 7959 }, { "epoch": 1.39, "learning_rate": 4.465949134164257e-05, "loss": 0.3167, "step": 7960 }, { "epoch": 1.39, "learning_rate": 4.4655062802217737e-05, "loss": 0.2909, "step": 7961 }, { "epoch": 1.39, "learning_rate": 4.4650632674875855e-05, "loss": 0.3075, "step": 7962 }, { "epoch": 1.39, "learning_rate": 4.464620096002698e-05, "loss": 0.3124, "step": 7963 }, { "epoch": 1.39, "learning_rate": 4.464176765808133e-05, "loss": 0.3107, "step": 7964 }, { "epoch": 1.39, "learning_rate": 4.463733276944926e-05, "loss": 0.3124, "step": 7965 }, { "epoch": 1.39, "learning_rate": 4.463289629454127e-05, "loss": 0.3099, "step": 7966 }, { "epoch": 1.39, "learning_rate": 4.462845823376801e-05, "loss": 0.3044, "step": 7967 }, { "epoch": 1.39, "learning_rate": 4.462401858754029e-05, "loss": 0.3254, "step": 7968 }, { "epoch": 1.39, "learning_rate": 4.461957735626906e-05, "loss": 0.3221, "step": 7969 }, { "epoch": 1.39, "learning_rate": 4.4615134540365386e-05, "loss": 0.3112, "step": 7970 }, { "epoch": 1.39, "learning_rate": 4.461069014024053e-05, "loss": 0.3267, "step": 7971 }, { "epoch": 1.39, "learning_rate": 4.460624415630586e-05, "loss": 0.3043, "step": 7972 }, { "epoch": 1.39, "learning_rate": 4.460179658897293e-05, "loss": 0.2991, "step": 7973 }, { "epoch": 1.39, "learning_rate": 4.4597347438653396e-05, "loss": 0.3097, "step": 7974 }, { "epoch": 1.39, "learning_rate": 4.4592896705759094e-05, "loss": 0.3196, "step": 7975 }, { "epoch": 1.39, "learning_rate": 4.4588444390701985e-05, "loss": 0.3031, "step": 7976 }, { "epoch": 1.39, "learning_rate": 4.458399049389421e-05, "loss": 0.3181, "step": 7977 }, { "epoch": 1.39, "learning_rate": 4.457953501574801e-05, "loss": 0.3196, "step": 7978 }, { "epoch": 1.39, "learning_rate": 4.457507795667581e-05, "loss": 0.3056, "step": 7979 }, { "epoch": 1.39, "learning_rate": 4.4570619317090153e-05, "loss": 0.3053, "step": 7980 }, { "epoch": 1.39, "learning_rate": 4.4566159097403756e-05, "loss": 0.3284, "step": 7981 }, { "epoch": 1.39, "learning_rate": 4.456169729802946e-05, "loss": 0.3034, "step": 7982 }, { "epoch": 1.39, "learning_rate": 4.455723391938026e-05, "loss": 0.3009, "step": 7983 }, { "epoch": 1.39, "learning_rate": 4.45527689618693e-05, "loss": 0.3134, "step": 7984 }, { "epoch": 1.39, "learning_rate": 4.454830242590988e-05, "loss": 0.3202, "step": 7985 }, { "epoch": 1.39, "learning_rate": 4.454383431191541e-05, "loss": 0.309, "step": 7986 }, { "epoch": 1.39, "learning_rate": 4.453936462029949e-05, "loss": 0.3057, "step": 7987 }, { "epoch": 1.39, "learning_rate": 4.4534893351475834e-05, "loss": 0.3033, "step": 7988 }, { "epoch": 1.39, "learning_rate": 4.453042050585833e-05, "loss": 0.3046, "step": 7989 }, { "epoch": 1.39, "learning_rate": 4.4525946083860984e-05, "loss": 0.3175, "step": 7990 }, { "epoch": 1.39, "learning_rate": 4.4521470085897965e-05, "loss": 0.3073, "step": 7991 }, { "epoch": 1.39, "learning_rate": 4.451699251238358e-05, "loss": 0.303, "step": 7992 }, { "epoch": 1.39, "learning_rate": 4.451251336373227e-05, "loss": 0.3104, "step": 7993 }, { "epoch": 1.39, "learning_rate": 4.450803264035867e-05, "loss": 0.3154, "step": 7994 }, { "epoch": 1.39, "learning_rate": 4.45035503426775e-05, "loss": 0.3209, "step": 7995 }, { "epoch": 1.39, "learning_rate": 4.449906647110367e-05, "loss": 0.3196, "step": 7996 }, { "epoch": 1.39, "learning_rate": 4.44945810260522e-05, "loss": 0.3106, "step": 7997 }, { "epoch": 1.39, "learning_rate": 4.449009400793829e-05, "loss": 0.3242, "step": 7998 }, { "epoch": 1.39, "learning_rate": 4.448560541717726e-05, "loss": 0.3178, "step": 7999 }, { "epoch": 1.39, "learning_rate": 4.4481115254184604e-05, "loss": 0.3123, "step": 8000 }, { "epoch": 1.39, "learning_rate": 4.4476623519375914e-05, "loss": 0.3078, "step": 8001 }, { "epoch": 1.39, "learning_rate": 4.447213021316699e-05, "loss": 0.2995, "step": 8002 }, { "epoch": 1.39, "learning_rate": 4.446763533597371e-05, "loss": 0.3092, "step": 8003 }, { "epoch": 1.4, "learning_rate": 4.446313888821216e-05, "loss": 0.3115, "step": 8004 }, { "epoch": 1.4, "learning_rate": 4.445864087029852e-05, "loss": 0.3149, "step": 8005 }, { "epoch": 1.4, "learning_rate": 4.445414128264916e-05, "loss": 0.3241, "step": 8006 }, { "epoch": 1.4, "learning_rate": 4.444964012568056e-05, "loss": 0.319, "step": 8007 }, { "epoch": 1.4, "learning_rate": 4.4445137399809363e-05, "loss": 0.3135, "step": 8008 }, { "epoch": 1.4, "learning_rate": 4.444063310545234e-05, "loss": 0.309, "step": 8009 }, { "epoch": 1.4, "learning_rate": 4.4436127243026445e-05, "loss": 0.3131, "step": 8010 }, { "epoch": 1.4, "learning_rate": 4.443161981294873e-05, "loss": 0.3032, "step": 8011 }, { "epoch": 1.4, "learning_rate": 4.442711081563643e-05, "loss": 0.3149, "step": 8012 }, { "epoch": 1.4, "learning_rate": 4.44226002515069e-05, "loss": 0.3076, "step": 8013 }, { "epoch": 1.4, "learning_rate": 4.4418088120977645e-05, "loss": 0.3016, "step": 8014 }, { "epoch": 1.4, "learning_rate": 4.4413574424466336e-05, "loss": 0.3131, "step": 8015 }, { "epoch": 1.4, "learning_rate": 4.440905916239076e-05, "loss": 0.3149, "step": 8016 }, { "epoch": 1.4, "learning_rate": 4.4404542335168863e-05, "loss": 0.3091, "step": 8017 }, { "epoch": 1.4, "learning_rate": 4.440002394321874e-05, "loss": 0.3062, "step": 8018 }, { "epoch": 1.4, "learning_rate": 4.439550398695862e-05, "loss": 0.3078, "step": 8019 }, { "epoch": 1.4, "learning_rate": 4.4390982466806884e-05, "loss": 0.2837, "step": 8020 }, { "epoch": 1.4, "learning_rate": 4.438645938318205e-05, "loss": 0.3143, "step": 8021 }, { "epoch": 1.4, "learning_rate": 4.438193473650279e-05, "loss": 0.3109, "step": 8022 }, { "epoch": 1.4, "learning_rate": 4.437740852718793e-05, "loss": 0.2949, "step": 8023 }, { "epoch": 1.4, "learning_rate": 4.43728807556564e-05, "loss": 0.302, "step": 8024 }, { "epoch": 1.4, "learning_rate": 4.436835142232733e-05, "loss": 0.2964, "step": 8025 }, { "epoch": 1.4, "learning_rate": 4.4363820527619945e-05, "loss": 0.3079, "step": 8026 }, { "epoch": 1.4, "learning_rate": 4.435928807195365e-05, "loss": 0.301, "step": 8027 }, { "epoch": 1.4, "learning_rate": 4.4354754055747966e-05, "loss": 0.3117, "step": 8028 }, { "epoch": 1.4, "learning_rate": 4.4350218479422597e-05, "loss": 0.3052, "step": 8029 }, { "epoch": 1.4, "learning_rate": 4.434568134339735e-05, "loss": 0.3172, "step": 8030 }, { "epoch": 1.4, "learning_rate": 4.43411426480922e-05, "loss": 0.3102, "step": 8031 }, { "epoch": 1.4, "learning_rate": 4.433660239392726e-05, "loss": 0.3183, "step": 8032 }, { "epoch": 1.4, "learning_rate": 4.433206058132278e-05, "loss": 0.3064, "step": 8033 }, { "epoch": 1.4, "learning_rate": 4.432751721069918e-05, "loss": 0.3237, "step": 8034 }, { "epoch": 1.4, "learning_rate": 4.432297228247698e-05, "loss": 0.3151, "step": 8035 }, { "epoch": 1.4, "learning_rate": 4.4318425797076896e-05, "loss": 0.3088, "step": 8036 }, { "epoch": 1.4, "learning_rate": 4.4313877754919754e-05, "loss": 0.3099, "step": 8037 }, { "epoch": 1.4, "learning_rate": 4.4309328156426524e-05, "loss": 0.3092, "step": 8038 }, { "epoch": 1.4, "learning_rate": 4.430477700201833e-05, "loss": 0.3037, "step": 8039 }, { "epoch": 1.4, "learning_rate": 4.430022429211645e-05, "loss": 0.312, "step": 8040 }, { "epoch": 1.4, "learning_rate": 4.429567002714229e-05, "loss": 0.3132, "step": 8041 }, { "epoch": 1.4, "learning_rate": 4.42911142075174e-05, "loss": 0.3005, "step": 8042 }, { "epoch": 1.4, "learning_rate": 4.428655683366348e-05, "loss": 0.3142, "step": 8043 }, { "epoch": 1.4, "learning_rate": 4.428199790600238e-05, "loss": 0.3137, "step": 8044 }, { "epoch": 1.4, "learning_rate": 4.427743742495607e-05, "loss": 0.3017, "step": 8045 }, { "epoch": 1.4, "learning_rate": 4.42728753909467e-05, "loss": 0.3054, "step": 8046 }, { "epoch": 1.4, "learning_rate": 4.426831180439652e-05, "loss": 0.3118, "step": 8047 }, { "epoch": 1.4, "learning_rate": 4.4263746665727964e-05, "loss": 0.3101, "step": 8048 }, { "epoch": 1.4, "learning_rate": 4.4259179975363596e-05, "loss": 0.3049, "step": 8049 }, { "epoch": 1.4, "learning_rate": 4.4254611733726114e-05, "loss": 0.3022, "step": 8050 }, { "epoch": 1.4, "learning_rate": 4.425004194123836e-05, "loss": 0.3077, "step": 8051 }, { "epoch": 1.4, "learning_rate": 4.424547059832334e-05, "loss": 0.3095, "step": 8052 }, { "epoch": 1.4, "learning_rate": 4.424089770540418e-05, "loss": 0.2899, "step": 8053 }, { "epoch": 1.4, "learning_rate": 4.423632326290416e-05, "loss": 0.3087, "step": 8054 }, { "epoch": 1.4, "learning_rate": 4.4231747271246696e-05, "loss": 0.3066, "step": 8055 }, { "epoch": 1.4, "learning_rate": 4.422716973085537e-05, "loss": 0.3223, "step": 8056 }, { "epoch": 1.4, "learning_rate": 4.422259064215388e-05, "loss": 0.3099, "step": 8057 }, { "epoch": 1.4, "learning_rate": 4.421801000556608e-05, "loss": 0.31, "step": 8058 }, { "epoch": 1.4, "learning_rate": 4.421342782151596e-05, "loss": 0.3099, "step": 8059 }, { "epoch": 1.4, "learning_rate": 4.420884409042766e-05, "loss": 0.3033, "step": 8060 }, { "epoch": 1.41, "learning_rate": 4.420425881272548e-05, "loss": 0.3125, "step": 8061 }, { "epoch": 1.41, "learning_rate": 4.419967198883382e-05, "loss": 0.3066, "step": 8062 }, { "epoch": 1.41, "learning_rate": 4.419508361917727e-05, "loss": 0.3071, "step": 8063 }, { "epoch": 1.41, "learning_rate": 4.4190493704180526e-05, "loss": 0.3065, "step": 8064 }, { "epoch": 1.41, "learning_rate": 4.4185902244268444e-05, "loss": 0.3059, "step": 8065 }, { "epoch": 1.41, "learning_rate": 4.418130923986603e-05, "loss": 0.307, "step": 8066 }, { "epoch": 1.41, "learning_rate": 4.417671469139842e-05, "loss": 0.3064, "step": 8067 }, { "epoch": 1.41, "learning_rate": 4.4172118599290894e-05, "loss": 0.3133, "step": 8068 }, { "epoch": 1.41, "learning_rate": 4.416752096396888e-05, "loss": 0.3163, "step": 8069 }, { "epoch": 1.41, "learning_rate": 4.4162921785857945e-05, "loss": 0.3071, "step": 8070 }, { "epoch": 1.41, "learning_rate": 4.4158321065383806e-05, "loss": 0.3087, "step": 8071 }, { "epoch": 1.41, "learning_rate": 4.415371880297232e-05, "loss": 0.3067, "step": 8072 }, { "epoch": 1.41, "learning_rate": 4.414911499904948e-05, "loss": 0.3005, "step": 8073 }, { "epoch": 1.41, "learning_rate": 4.414450965404142e-05, "loss": 0.3074, "step": 8074 }, { "epoch": 1.41, "learning_rate": 4.413990276837443e-05, "loss": 0.312, "step": 8075 }, { "epoch": 1.41, "learning_rate": 4.413529434247494e-05, "loss": 0.3269, "step": 8076 }, { "epoch": 1.41, "learning_rate": 4.41306843767695e-05, "loss": 0.3215, "step": 8077 }, { "epoch": 1.41, "learning_rate": 4.412607287168484e-05, "loss": 0.3189, "step": 8078 }, { "epoch": 1.41, "learning_rate": 4.41214598276478e-05, "loss": 0.3064, "step": 8079 }, { "epoch": 1.41, "learning_rate": 4.4116845245085386e-05, "loss": 0.2962, "step": 8080 }, { "epoch": 1.41, "learning_rate": 4.411222912442472e-05, "loss": 0.2878, "step": 8081 }, { "epoch": 1.41, "learning_rate": 4.41076114660931e-05, "loss": 0.3041, "step": 8082 }, { "epoch": 1.41, "learning_rate": 4.410299227051795e-05, "loss": 0.3049, "step": 8083 }, { "epoch": 1.41, "learning_rate": 4.4098371538126806e-05, "loss": 0.3396, "step": 8084 }, { "epoch": 1.41, "learning_rate": 4.40937492693474e-05, "loss": 0.2989, "step": 8085 }, { "epoch": 1.41, "learning_rate": 4.408912546460757e-05, "loss": 0.3084, "step": 8086 }, { "epoch": 1.41, "learning_rate": 4.4084500124335324e-05, "loss": 0.3144, "step": 8087 }, { "epoch": 1.41, "learning_rate": 4.407987324895878e-05, "loss": 0.324, "step": 8088 }, { "epoch": 1.41, "learning_rate": 4.4075244838906216e-05, "loss": 0.316, "step": 8089 }, { "epoch": 1.41, "learning_rate": 4.4070614894606054e-05, "loss": 0.3151, "step": 8090 }, { "epoch": 1.41, "learning_rate": 4.406598341648685e-05, "loss": 0.3125, "step": 8091 }, { "epoch": 1.41, "learning_rate": 4.4061350404977295e-05, "loss": 0.3017, "step": 8092 }, { "epoch": 1.41, "learning_rate": 4.4056715860506255e-05, "loss": 0.3117, "step": 8093 }, { "epoch": 1.41, "learning_rate": 4.40520797835027e-05, "loss": 0.3084, "step": 8094 }, { "epoch": 1.41, "learning_rate": 4.404744217439577e-05, "loss": 0.3161, "step": 8095 }, { "epoch": 1.41, "learning_rate": 4.404280303361472e-05, "loss": 0.3049, "step": 8096 }, { "epoch": 1.41, "learning_rate": 4.403816236158896e-05, "loss": 0.3093, "step": 8097 }, { "epoch": 1.41, "learning_rate": 4.4033520158748055e-05, "loss": 0.3068, "step": 8098 }, { "epoch": 1.41, "learning_rate": 4.4028876425521694e-05, "loss": 0.3219, "step": 8099 }, { "epoch": 1.41, "learning_rate": 4.4024231162339715e-05, "loss": 0.3065, "step": 8100 }, { "epoch": 1.41, "learning_rate": 4.40195843696321e-05, "loss": 0.3046, "step": 8101 }, { "epoch": 1.41, "learning_rate": 4.4014936047828945e-05, "loss": 0.3146, "step": 8102 }, { "epoch": 1.41, "learning_rate": 4.401028619736053e-05, "loss": 0.3194, "step": 8103 }, { "epoch": 1.41, "learning_rate": 4.400563481865726e-05, "loss": 0.3316, "step": 8104 }, { "epoch": 1.41, "learning_rate": 4.400098191214967e-05, "loss": 0.2998, "step": 8105 }, { "epoch": 1.41, "learning_rate": 4.399632747826845e-05, "loss": 0.3056, "step": 8106 }, { "epoch": 1.41, "learning_rate": 4.399167151744442e-05, "loss": 0.3043, "step": 8107 }, { "epoch": 1.41, "learning_rate": 4.3987014030108554e-05, "loss": 0.3075, "step": 8108 }, { "epoch": 1.41, "learning_rate": 4.398235501669196e-05, "loss": 0.3099, "step": 8109 }, { "epoch": 1.41, "learning_rate": 4.3977694477625895e-05, "loss": 0.3117, "step": 8110 }, { "epoch": 1.41, "learning_rate": 4.397303241334173e-05, "loss": 0.3188, "step": 8111 }, { "epoch": 1.41, "learning_rate": 4.3968368824271014e-05, "loss": 0.3348, "step": 8112 }, { "epoch": 1.41, "learning_rate": 4.396370371084543e-05, "loss": 0.3083, "step": 8113 }, { "epoch": 1.41, "learning_rate": 4.395903707349677e-05, "loss": 0.3122, "step": 8114 }, { "epoch": 1.41, "learning_rate": 4.3954368912657e-05, "loss": 0.3169, "step": 8115 }, { "epoch": 1.41, "learning_rate": 4.394969922875822e-05, "loss": 0.3042, "step": 8116 }, { "epoch": 1.41, "learning_rate": 4.394502802223267e-05, "loss": 0.3099, "step": 8117 }, { "epoch": 1.41, "learning_rate": 4.394035529351272e-05, "loss": 0.3024, "step": 8118 }, { "epoch": 1.42, "learning_rate": 4.3935681043030895e-05, "loss": 0.3075, "step": 8119 }, { "epoch": 1.42, "learning_rate": 4.393100527121986e-05, "loss": 0.3168, "step": 8120 }, { "epoch": 1.42, "learning_rate": 4.3926327978512404e-05, "loss": 0.2999, "step": 8121 }, { "epoch": 1.42, "learning_rate": 4.3921649165341486e-05, "loss": 0.3065, "step": 8122 }, { "epoch": 1.42, "learning_rate": 4.3916968832140176e-05, "loss": 0.2982, "step": 8123 }, { "epoch": 1.42, "learning_rate": 4.391228697934171e-05, "loss": 0.3147, "step": 8124 }, { "epoch": 1.42, "learning_rate": 4.390760360737943e-05, "loss": 0.309, "step": 8125 }, { "epoch": 1.42, "learning_rate": 4.390291871668687e-05, "loss": 0.3113, "step": 8126 }, { "epoch": 1.42, "learning_rate": 4.3898232307697654e-05, "loss": 0.3284, "step": 8127 }, { "epoch": 1.42, "learning_rate": 4.389354438084558e-05, "loss": 0.3137, "step": 8128 }, { "epoch": 1.42, "learning_rate": 4.388885493656456e-05, "loss": 0.2989, "step": 8129 }, { "epoch": 1.42, "learning_rate": 4.388416397528868e-05, "loss": 0.3103, "step": 8130 }, { "epoch": 1.42, "learning_rate": 4.387947149745214e-05, "loss": 0.2967, "step": 8131 }, { "epoch": 1.42, "learning_rate": 4.3874777503489285e-05, "loss": 0.3114, "step": 8132 }, { "epoch": 1.42, "learning_rate": 4.3870081993834606e-05, "loss": 0.3065, "step": 8133 }, { "epoch": 1.42, "learning_rate": 4.386538496892272e-05, "loss": 0.3073, "step": 8134 }, { "epoch": 1.42, "learning_rate": 4.3860686429188416e-05, "loss": 0.3058, "step": 8135 }, { "epoch": 1.42, "learning_rate": 4.385598637506659e-05, "loss": 0.305, "step": 8136 }, { "epoch": 1.42, "learning_rate": 4.38512848069923e-05, "loss": 0.2816, "step": 8137 }, { "epoch": 1.42, "learning_rate": 4.3846581725400726e-05, "loss": 0.3096, "step": 8138 }, { "epoch": 1.42, "learning_rate": 4.38418771307272e-05, "loss": 0.2986, "step": 8139 }, { "epoch": 1.42, "learning_rate": 4.3837171023407195e-05, "loss": 0.309, "step": 8140 }, { "epoch": 1.42, "learning_rate": 4.383246340387631e-05, "loss": 0.2937, "step": 8141 }, { "epoch": 1.42, "learning_rate": 4.382775427257032e-05, "loss": 0.3083, "step": 8142 }, { "epoch": 1.42, "learning_rate": 4.382304362992509e-05, "loss": 0.3085, "step": 8143 }, { "epoch": 1.42, "learning_rate": 4.381833147637665e-05, "loss": 0.3105, "step": 8144 }, { "epoch": 1.42, "learning_rate": 4.381361781236118e-05, "loss": 0.3255, "step": 8145 }, { "epoch": 1.42, "learning_rate": 4.3808902638314975e-05, "loss": 0.3252, "step": 8146 }, { "epoch": 1.42, "learning_rate": 4.3804185954674506e-05, "loss": 0.3179, "step": 8147 }, { "epoch": 1.42, "learning_rate": 4.3799467761876344e-05, "loss": 0.3205, "step": 8148 }, { "epoch": 1.42, "learning_rate": 4.3794748060357224e-05, "loss": 0.3123, "step": 8149 }, { "epoch": 1.42, "learning_rate": 4.3790026850554006e-05, "loss": 0.3141, "step": 8150 }, { "epoch": 1.42, "learning_rate": 4.3785304132903705e-05, "loss": 0.3109, "step": 8151 }, { "epoch": 1.42, "learning_rate": 4.378057990784346e-05, "loss": 0.3299, "step": 8152 }, { "epoch": 1.42, "learning_rate": 4.3775854175810575e-05, "loss": 0.3023, "step": 8153 }, { "epoch": 1.42, "learning_rate": 4.3771126937242454e-05, "loss": 0.3178, "step": 8154 }, { "epoch": 1.42, "learning_rate": 4.3766398192576684e-05, "loss": 0.3165, "step": 8155 }, { "epoch": 1.42, "learning_rate": 4.376166794225096e-05, "loss": 0.3114, "step": 8156 }, { "epoch": 1.42, "learning_rate": 4.375693618670311e-05, "loss": 0.3004, "step": 8157 }, { "epoch": 1.42, "learning_rate": 4.3752202926371145e-05, "loss": 0.3052, "step": 8158 }, { "epoch": 1.42, "learning_rate": 4.374746816169317e-05, "loss": 0.313, "step": 8159 }, { "epoch": 1.42, "learning_rate": 4.3742731893107455e-05, "loss": 0.3141, "step": 8160 }, { "epoch": 1.42, "learning_rate": 4.3737994121052396e-05, "loss": 0.3227, "step": 8161 }, { "epoch": 1.42, "learning_rate": 4.373325484596654e-05, "loss": 0.3202, "step": 8162 }, { "epoch": 1.42, "learning_rate": 4.372851406828855e-05, "loss": 0.3093, "step": 8163 }, { "epoch": 1.42, "learning_rate": 4.372377178845727e-05, "loss": 0.2965, "step": 8164 }, { "epoch": 1.42, "learning_rate": 4.371902800691165e-05, "loss": 0.2993, "step": 8165 }, { "epoch": 1.42, "learning_rate": 4.371428272409077e-05, "loss": 0.3052, "step": 8166 }, { "epoch": 1.42, "learning_rate": 4.370953594043388e-05, "loss": 0.3037, "step": 8167 }, { "epoch": 1.42, "learning_rate": 4.370478765638036e-05, "loss": 0.3102, "step": 8168 }, { "epoch": 1.42, "learning_rate": 4.370003787236971e-05, "loss": 0.3031, "step": 8169 }, { "epoch": 1.42, "learning_rate": 4.369528658884158e-05, "loss": 0.3132, "step": 8170 }, { "epoch": 1.42, "learning_rate": 4.3690533806235776e-05, "loss": 0.2948, "step": 8171 }, { "epoch": 1.42, "learning_rate": 4.3685779524992224e-05, "loss": 0.3026, "step": 8172 }, { "epoch": 1.42, "learning_rate": 4.368102374555099e-05, "loss": 0.3183, "step": 8173 }, { "epoch": 1.42, "learning_rate": 4.367626646835227e-05, "loss": 0.3244, "step": 8174 }, { "epoch": 1.42, "learning_rate": 4.367150769383644e-05, "loss": 0.2994, "step": 8175 }, { "epoch": 1.43, "learning_rate": 4.366674742244395e-05, "loss": 0.3162, "step": 8176 }, { "epoch": 1.43, "learning_rate": 4.366198565461544e-05, "loss": 0.309, "step": 8177 }, { "epoch": 1.43, "learning_rate": 4.365722239079167e-05, "loss": 0.313, "step": 8178 }, { "epoch": 1.43, "learning_rate": 4.365245763141353e-05, "loss": 0.3025, "step": 8179 }, { "epoch": 1.43, "learning_rate": 4.3647691376922087e-05, "loss": 0.3167, "step": 8180 }, { "epoch": 1.43, "learning_rate": 4.364292362775849e-05, "loss": 0.3149, "step": 8181 }, { "epoch": 1.43, "learning_rate": 4.363815438436407e-05, "loss": 0.3154, "step": 8182 }, { "epoch": 1.43, "learning_rate": 4.363338364718026e-05, "loss": 0.3073, "step": 8183 }, { "epoch": 1.43, "learning_rate": 4.362861141664868e-05, "loss": 0.3077, "step": 8184 }, { "epoch": 1.43, "learning_rate": 4.362383769321104e-05, "loss": 0.3103, "step": 8185 }, { "epoch": 1.43, "learning_rate": 4.361906247730921e-05, "loss": 0.3115, "step": 8186 }, { "epoch": 1.43, "learning_rate": 4.361428576938521e-05, "loss": 0.3011, "step": 8187 }, { "epoch": 1.43, "learning_rate": 4.360950756988117e-05, "loss": 0.3092, "step": 8188 }, { "epoch": 1.43, "learning_rate": 4.360472787923937e-05, "loss": 0.3006, "step": 8189 }, { "epoch": 1.43, "learning_rate": 4.359994669790224e-05, "loss": 0.3064, "step": 8190 }, { "epoch": 1.43, "learning_rate": 4.359516402631234e-05, "loss": 0.3078, "step": 8191 }, { "epoch": 1.43, "learning_rate": 4.359037986491237e-05, "loss": 0.3109, "step": 8192 }, { "epoch": 1.43, "learning_rate": 4.358559421414515e-05, "loss": 0.3113, "step": 8193 }, { "epoch": 1.43, "learning_rate": 4.358080707445365e-05, "loss": 0.3158, "step": 8194 }, { "epoch": 1.43, "learning_rate": 4.3576018446281006e-05, "loss": 0.3075, "step": 8195 }, { "epoch": 1.43, "learning_rate": 4.3571228330070434e-05, "loss": 0.3003, "step": 8196 }, { "epoch": 1.43, "learning_rate": 4.356643672626535e-05, "loss": 0.3056, "step": 8197 }, { "epoch": 1.43, "learning_rate": 4.356164363530925e-05, "loss": 0.3093, "step": 8198 }, { "epoch": 1.43, "learning_rate": 4.355684905764581e-05, "loss": 0.3061, "step": 8199 }, { "epoch": 1.43, "learning_rate": 4.355205299371883e-05, "loss": 0.3143, "step": 8200 }, { "epoch": 1.43, "learning_rate": 4.354725544397224e-05, "loss": 0.3038, "step": 8201 }, { "epoch": 1.43, "learning_rate": 4.35424564088501e-05, "loss": 0.2986, "step": 8202 }, { "epoch": 1.43, "learning_rate": 4.3537655888796664e-05, "loss": 0.3061, "step": 8203 }, { "epoch": 1.43, "learning_rate": 4.353285388425623e-05, "loss": 0.3223, "step": 8204 }, { "epoch": 1.43, "learning_rate": 4.352805039567332e-05, "loss": 0.32, "step": 8205 }, { "epoch": 1.43, "learning_rate": 4.352324542349254e-05, "loss": 0.3025, "step": 8206 }, { "epoch": 1.43, "learning_rate": 4.3518438968158656e-05, "loss": 0.3108, "step": 8207 }, { "epoch": 1.43, "learning_rate": 4.3513631030116556e-05, "loss": 0.3031, "step": 8208 }, { "epoch": 1.43, "learning_rate": 4.35088216098113e-05, "loss": 0.3198, "step": 8209 }, { "epoch": 1.43, "learning_rate": 4.350401070768803e-05, "loss": 0.3073, "step": 8210 }, { "epoch": 1.43, "learning_rate": 4.3499198324192075e-05, "loss": 0.3124, "step": 8211 }, { "epoch": 1.43, "learning_rate": 4.3494384459768885e-05, "loss": 0.301, "step": 8212 }, { "epoch": 1.43, "learning_rate": 4.3489569114864023e-05, "loss": 0.3068, "step": 8213 }, { "epoch": 1.43, "learning_rate": 4.348475228992323e-05, "loss": 0.2971, "step": 8214 }, { "epoch": 1.43, "learning_rate": 4.347993398539235e-05, "loss": 0.3224, "step": 8215 }, { "epoch": 1.43, "learning_rate": 4.347511420171739e-05, "loss": 0.3009, "step": 8216 }, { "epoch": 1.43, "learning_rate": 4.347029293934448e-05, "loss": 0.3224, "step": 8217 }, { "epoch": 1.43, "learning_rate": 4.3465470198719876e-05, "loss": 0.3038, "step": 8218 }, { "epoch": 1.43, "learning_rate": 4.346064598028999e-05, "loss": 0.3104, "step": 8219 }, { "epoch": 1.43, "learning_rate": 4.345582028450137e-05, "loss": 0.3103, "step": 8220 }, { "epoch": 1.43, "learning_rate": 4.345099311180069e-05, "loss": 0.3117, "step": 8221 }, { "epoch": 1.43, "learning_rate": 4.3446164462634765e-05, "loss": 0.3086, "step": 8222 }, { "epoch": 1.43, "learning_rate": 4.344133433745055e-05, "loss": 0.2997, "step": 8223 }, { "epoch": 1.43, "learning_rate": 4.343650273669513e-05, "loss": 0.3137, "step": 8224 }, { "epoch": 1.43, "learning_rate": 4.343166966081573e-05, "loss": 0.2897, "step": 8225 }, { "epoch": 1.43, "learning_rate": 4.342683511025972e-05, "loss": 0.3255, "step": 8226 }, { "epoch": 1.43, "learning_rate": 4.342199908547459e-05, "loss": 0.3015, "step": 8227 }, { "epoch": 1.43, "learning_rate": 4.341716158690797e-05, "loss": 0.319, "step": 8228 }, { "epoch": 1.43, "learning_rate": 4.3412322615007656e-05, "loss": 0.3148, "step": 8229 }, { "epoch": 1.43, "learning_rate": 4.3407482170221535e-05, "loss": 0.3071, "step": 8230 }, { "epoch": 1.43, "learning_rate": 4.340264025299765e-05, "loss": 0.3042, "step": 8231 }, { "epoch": 1.43, "learning_rate": 4.339779686378419e-05, "loss": 0.3105, "step": 8232 }, { "epoch": 1.44, "learning_rate": 4.339295200302947e-05, "loss": 0.3227, "step": 8233 }, { "epoch": 1.44, "learning_rate": 4.338810567118194e-05, "loss": 0.3125, "step": 8234 }, { "epoch": 1.44, "learning_rate": 4.338325786869019e-05, "loss": 0.3092, "step": 8235 }, { "epoch": 1.44, "learning_rate": 4.337840859600294e-05, "loss": 0.326, "step": 8236 }, { "epoch": 1.44, "learning_rate": 4.337355785356907e-05, "loss": 0.299, "step": 8237 }, { "epoch": 1.44, "learning_rate": 4.336870564183756e-05, "loss": 0.3048, "step": 8238 }, { "epoch": 1.44, "learning_rate": 4.336385196125754e-05, "loss": 0.3287, "step": 8239 }, { "epoch": 1.44, "learning_rate": 4.3358996812278296e-05, "loss": 0.3125, "step": 8240 }, { "epoch": 1.44, "learning_rate": 4.335414019534922e-05, "loss": 0.3111, "step": 8241 }, { "epoch": 1.44, "learning_rate": 4.334928211091985e-05, "loss": 0.3051, "step": 8242 }, { "epoch": 1.44, "learning_rate": 4.3344422559439876e-05, "loss": 0.3115, "step": 8243 }, { "epoch": 1.44, "learning_rate": 4.3339561541359107e-05, "loss": 0.3088, "step": 8244 }, { "epoch": 1.44, "learning_rate": 4.333469905712749e-05, "loss": 0.3199, "step": 8245 }, { "epoch": 1.44, "learning_rate": 4.33298351071951e-05, "loss": 0.3246, "step": 8246 }, { "epoch": 1.44, "learning_rate": 4.332496969201217e-05, "loss": 0.327, "step": 8247 }, { "epoch": 1.44, "learning_rate": 4.332010281202905e-05, "loss": 0.3037, "step": 8248 }, { "epoch": 1.44, "learning_rate": 4.331523446769623e-05, "loss": 0.3265, "step": 8249 }, { "epoch": 1.44, "learning_rate": 4.331036465946434e-05, "loss": 0.332, "step": 8250 }, { "epoch": 1.44, "learning_rate": 4.330549338778414e-05, "loss": 0.3195, "step": 8251 }, { "epoch": 1.44, "learning_rate": 4.330062065310652e-05, "loss": 0.3092, "step": 8252 }, { "epoch": 1.44, "learning_rate": 4.329574645588253e-05, "loss": 0.3075, "step": 8253 }, { "epoch": 1.44, "learning_rate": 4.329087079656333e-05, "loss": 0.3139, "step": 8254 }, { "epoch": 1.44, "learning_rate": 4.3285993675600214e-05, "loss": 0.2971, "step": 8255 }, { "epoch": 1.44, "learning_rate": 4.3281115093444637e-05, "loss": 0.3108, "step": 8256 }, { "epoch": 1.44, "learning_rate": 4.327623505054816e-05, "loss": 0.3053, "step": 8257 }, { "epoch": 1.44, "learning_rate": 4.3271353547362494e-05, "loss": 0.3142, "step": 8258 }, { "epoch": 1.44, "learning_rate": 4.32664705843395e-05, "loss": 0.3048, "step": 8259 }, { "epoch": 1.44, "learning_rate": 4.326158616193113e-05, "loss": 0.3003, "step": 8260 }, { "epoch": 1.44, "learning_rate": 4.3256700280589525e-05, "loss": 0.3224, "step": 8261 }, { "epoch": 1.44, "learning_rate": 4.3251812940766925e-05, "loss": 0.3151, "step": 8262 }, { "epoch": 1.44, "learning_rate": 4.324692414291571e-05, "loss": 0.3126, "step": 8263 }, { "epoch": 1.44, "learning_rate": 4.32420338874884e-05, "loss": 0.3041, "step": 8264 }, { "epoch": 1.44, "learning_rate": 4.3237142174937655e-05, "loss": 0.3186, "step": 8265 }, { "epoch": 1.44, "learning_rate": 4.323224900571626e-05, "loss": 0.3161, "step": 8266 }, { "epoch": 1.44, "learning_rate": 4.322735438027715e-05, "loss": 0.3146, "step": 8267 }, { "epoch": 1.44, "learning_rate": 4.3222458299073375e-05, "loss": 0.3094, "step": 8268 }, { "epoch": 1.44, "learning_rate": 4.321756076255813e-05, "loss": 0.2997, "step": 8269 }, { "epoch": 1.44, "learning_rate": 4.321266177118474e-05, "loss": 0.3136, "step": 8270 }, { "epoch": 1.44, "learning_rate": 4.320776132540667e-05, "loss": 0.3091, "step": 8271 }, { "epoch": 1.44, "learning_rate": 4.3202859425677526e-05, "loss": 0.3211, "step": 8272 }, { "epoch": 1.44, "learning_rate": 4.319795607245103e-05, "loss": 0.3036, "step": 8273 }, { "epoch": 1.44, "learning_rate": 4.319305126618106e-05, "loss": 0.3069, "step": 8274 }, { "epoch": 1.44, "learning_rate": 4.3188145007321614e-05, "loss": 0.3283, "step": 8275 }, { "epoch": 1.44, "learning_rate": 4.318323729632682e-05, "loss": 0.3115, "step": 8276 }, { "epoch": 1.44, "learning_rate": 4.317832813365095e-05, "loss": 0.3131, "step": 8277 }, { "epoch": 1.44, "learning_rate": 4.3173417519748424e-05, "loss": 0.3014, "step": 8278 }, { "epoch": 1.44, "learning_rate": 4.316850545507376e-05, "loss": 0.2943, "step": 8279 }, { "epoch": 1.44, "learning_rate": 4.316359194008165e-05, "loss": 0.3079, "step": 8280 }, { "epoch": 1.44, "learning_rate": 4.315867697522688e-05, "loss": 0.3092, "step": 8281 }, { "epoch": 1.44, "learning_rate": 4.315376056096442e-05, "loss": 0.3135, "step": 8282 }, { "epoch": 1.44, "learning_rate": 4.314884269774932e-05, "loss": 0.3306, "step": 8283 }, { "epoch": 1.44, "learning_rate": 4.31439233860368e-05, "loss": 0.312, "step": 8284 }, { "epoch": 1.44, "learning_rate": 4.313900262628221e-05, "loss": 0.3133, "step": 8285 }, { "epoch": 1.44, "learning_rate": 4.313408041894103e-05, "loss": 0.3177, "step": 8286 }, { "epoch": 1.44, "learning_rate": 4.312915676446886e-05, "loss": 0.2921, "step": 8287 }, { "epoch": 1.44, "learning_rate": 4.312423166332144e-05, "loss": 0.3148, "step": 8288 }, { "epoch": 1.44, "learning_rate": 4.3119305115954676e-05, "loss": 0.2893, "step": 8289 }, { "epoch": 1.44, "learning_rate": 4.3114377122824574e-05, "loss": 0.2989, "step": 8290 }, { "epoch": 1.45, "learning_rate": 4.310944768438726e-05, "loss": 0.3037, "step": 8291 }, { "epoch": 1.45, "learning_rate": 4.3104516801099046e-05, "loss": 0.3084, "step": 8292 }, { "epoch": 1.45, "learning_rate": 4.309958447341633e-05, "loss": 0.3336, "step": 8293 }, { "epoch": 1.45, "learning_rate": 4.309465070179566e-05, "loss": 0.3138, "step": 8294 }, { "epoch": 1.45, "learning_rate": 4.308971548669372e-05, "loss": 0.3095, "step": 8295 }, { "epoch": 1.45, "learning_rate": 4.3084778828567335e-05, "loss": 0.327, "step": 8296 }, { "epoch": 1.45, "learning_rate": 4.3079840727873446e-05, "loss": 0.3077, "step": 8297 }, { "epoch": 1.45, "learning_rate": 4.3074901185069136e-05, "loss": 0.3226, "step": 8298 }, { "epoch": 1.45, "learning_rate": 4.306996020061163e-05, "loss": 0.3131, "step": 8299 }, { "epoch": 1.45, "learning_rate": 4.3065017774958265e-05, "loss": 0.3132, "step": 8300 }, { "epoch": 1.45, "learning_rate": 4.306007390856655e-05, "loss": 0.3001, "step": 8301 }, { "epoch": 1.45, "learning_rate": 4.3055128601894074e-05, "loss": 0.3019, "step": 8302 }, { "epoch": 1.45, "learning_rate": 4.3050181855398595e-05, "loss": 0.3237, "step": 8303 }, { "epoch": 1.45, "learning_rate": 4.3045233669538e-05, "loss": 0.3062, "step": 8304 }, { "epoch": 1.45, "learning_rate": 4.304028404477031e-05, "loss": 0.3034, "step": 8305 }, { "epoch": 1.45, "learning_rate": 4.3035332981553684e-05, "loss": 0.3161, "step": 8306 }, { "epoch": 1.45, "learning_rate": 4.303038048034638e-05, "loss": 0.3058, "step": 8307 }, { "epoch": 1.45, "learning_rate": 4.3025426541606826e-05, "loss": 0.2986, "step": 8308 }, { "epoch": 1.45, "learning_rate": 4.302047116579357e-05, "loss": 0.305, "step": 8309 }, { "epoch": 1.45, "learning_rate": 4.3015514353365307e-05, "loss": 0.3113, "step": 8310 }, { "epoch": 1.45, "learning_rate": 4.301055610478083e-05, "loss": 0.3003, "step": 8311 }, { "epoch": 1.45, "learning_rate": 4.300559642049911e-05, "loss": 0.3005, "step": 8312 }, { "epoch": 1.45, "learning_rate": 4.300063530097921e-05, "loss": 0.3041, "step": 8313 }, { "epoch": 1.45, "learning_rate": 4.299567274668036e-05, "loss": 0.3032, "step": 8314 }, { "epoch": 1.45, "learning_rate": 4.2990708758061884e-05, "loss": 0.3188, "step": 8315 }, { "epoch": 1.45, "learning_rate": 4.298574333558329e-05, "loss": 0.3144, "step": 8316 }, { "epoch": 1.45, "learning_rate": 4.298077647970417e-05, "loss": 0.3127, "step": 8317 }, { "epoch": 1.45, "learning_rate": 4.297580819088428e-05, "loss": 0.3106, "step": 8318 }, { "epoch": 1.45, "learning_rate": 4.297083846958349e-05, "loss": 0.2965, "step": 8319 }, { "epoch": 1.45, "learning_rate": 4.296586731626181e-05, "loss": 0.3063, "step": 8320 }, { "epoch": 1.45, "learning_rate": 4.2960894731379395e-05, "loss": 0.3171, "step": 8321 }, { "epoch": 1.45, "learning_rate": 4.295592071539651e-05, "loss": 0.3108, "step": 8322 }, { "epoch": 1.45, "learning_rate": 4.2950945268773555e-05, "loss": 0.3023, "step": 8323 }, { "epoch": 1.45, "learning_rate": 4.29459683919711e-05, "loss": 0.3044, "step": 8324 }, { "epoch": 1.45, "learning_rate": 4.2940990085449775e-05, "loss": 0.3106, "step": 8325 }, { "epoch": 1.45, "learning_rate": 4.293601034967042e-05, "loss": 0.3178, "step": 8326 }, { "epoch": 1.45, "learning_rate": 4.293102918509396e-05, "loss": 0.314, "step": 8327 }, { "epoch": 1.45, "learning_rate": 4.2926046592181456e-05, "loss": 0.3109, "step": 8328 }, { "epoch": 1.45, "learning_rate": 4.292106257139413e-05, "loss": 0.3109, "step": 8329 }, { "epoch": 1.45, "learning_rate": 4.291607712319329e-05, "loss": 0.3165, "step": 8330 }, { "epoch": 1.45, "learning_rate": 4.291109024804043e-05, "loss": 0.3206, "step": 8331 }, { "epoch": 1.45, "learning_rate": 4.290610194639712e-05, "loss": 0.3091, "step": 8332 }, { "epoch": 1.45, "learning_rate": 4.290111221872511e-05, "loss": 0.2912, "step": 8333 }, { "epoch": 1.45, "learning_rate": 4.289612106548625e-05, "loss": 0.3015, "step": 8334 }, { "epoch": 1.45, "learning_rate": 4.289112848714254e-05, "loss": 0.288, "step": 8335 }, { "epoch": 1.45, "learning_rate": 4.288613448415612e-05, "loss": 0.3102, "step": 8336 }, { "epoch": 1.45, "learning_rate": 4.288113905698922e-05, "loss": 0.3113, "step": 8337 }, { "epoch": 1.45, "learning_rate": 4.287614220610426e-05, "loss": 0.2818, "step": 8338 }, { "epoch": 1.45, "learning_rate": 4.287114393196373e-05, "loss": 0.3073, "step": 8339 }, { "epoch": 1.45, "learning_rate": 4.2866144235030304e-05, "loss": 0.3087, "step": 8340 }, { "epoch": 1.45, "learning_rate": 4.286114311576676e-05, "loss": 0.3054, "step": 8341 }, { "epoch": 1.45, "learning_rate": 4.2856140574636015e-05, "loss": 0.306, "step": 8342 }, { "epoch": 1.45, "learning_rate": 4.2851136612101114e-05, "loss": 0.3139, "step": 8343 }, { "epoch": 1.45, "learning_rate": 4.2846131228625246e-05, "loss": 0.3166, "step": 8344 }, { "epoch": 1.45, "learning_rate": 4.2841124424671716e-05, "loss": 0.3028, "step": 8345 }, { "epoch": 1.45, "learning_rate": 4.283611620070397e-05, "loss": 0.3104, "step": 8346 }, { "epoch": 1.45, "learning_rate": 4.283110655718557e-05, "loss": 0.3195, "step": 8347 }, { "epoch": 1.46, "learning_rate": 4.282609549458024e-05, "loss": 0.2955, "step": 8348 }, { "epoch": 1.46, "learning_rate": 4.282108301335181e-05, "loss": 0.3051, "step": 8349 }, { "epoch": 1.46, "learning_rate": 4.2816069113964245e-05, "loss": 0.3232, "step": 8350 }, { "epoch": 1.46, "learning_rate": 4.281105379688164e-05, "loss": 0.3052, "step": 8351 }, { "epoch": 1.46, "learning_rate": 4.280603706256824e-05, "loss": 0.2984, "step": 8352 }, { "epoch": 1.46, "learning_rate": 4.28010189114884e-05, "loss": 0.3129, "step": 8353 }, { "epoch": 1.46, "learning_rate": 4.2795999344106605e-05, "loss": 0.3237, "step": 8354 }, { "epoch": 1.46, "learning_rate": 4.2790978360887485e-05, "loss": 0.3013, "step": 8355 }, { "epoch": 1.46, "learning_rate": 4.27859559622958e-05, "loss": 0.3202, "step": 8356 }, { "epoch": 1.46, "learning_rate": 4.278093214879644e-05, "loss": 0.3041, "step": 8357 }, { "epoch": 1.46, "learning_rate": 4.27759069208544e-05, "loss": 0.323, "step": 8358 }, { "epoch": 1.46, "learning_rate": 4.277088027893485e-05, "loss": 0.3039, "step": 8359 }, { "epoch": 1.46, "learning_rate": 4.276585222350306e-05, "loss": 0.3079, "step": 8360 }, { "epoch": 1.46, "learning_rate": 4.2760822755024446e-05, "loss": 0.2949, "step": 8361 }, { "epoch": 1.46, "learning_rate": 4.275579187396453e-05, "loss": 0.3107, "step": 8362 }, { "epoch": 1.46, "learning_rate": 4.275075958078901e-05, "loss": 0.2981, "step": 8363 }, { "epoch": 1.46, "learning_rate": 4.2745725875963684e-05, "loss": 0.2998, "step": 8364 }, { "epoch": 1.46, "learning_rate": 4.274069075995446e-05, "loss": 0.3312, "step": 8365 }, { "epoch": 1.46, "learning_rate": 4.2735654233227424e-05, "loss": 0.3284, "step": 8366 }, { "epoch": 1.46, "learning_rate": 4.273061629624877e-05, "loss": 0.3067, "step": 8367 }, { "epoch": 1.46, "learning_rate": 4.2725576949484805e-05, "loss": 0.3112, "step": 8368 }, { "epoch": 1.46, "learning_rate": 4.2720536193402e-05, "loss": 0.3179, "step": 8369 }, { "epoch": 1.46, "learning_rate": 4.271549402846693e-05, "loss": 0.3078, "step": 8370 }, { "epoch": 1.46, "learning_rate": 4.271045045514633e-05, "loss": 0.3097, "step": 8371 }, { "epoch": 1.46, "learning_rate": 4.2705405473907026e-05, "loss": 0.3073, "step": 8372 }, { "epoch": 1.46, "learning_rate": 4.2700359085216004e-05, "loss": 0.3007, "step": 8373 }, { "epoch": 1.46, "learning_rate": 4.2695311289540365e-05, "loss": 0.3114, "step": 8374 }, { "epoch": 1.46, "learning_rate": 4.269026208734734e-05, "loss": 0.3157, "step": 8375 }, { "epoch": 1.46, "learning_rate": 4.268521147910432e-05, "loss": 0.3291, "step": 8376 }, { "epoch": 1.46, "learning_rate": 4.268015946527878e-05, "loss": 0.3119, "step": 8377 }, { "epoch": 1.46, "learning_rate": 4.267510604633836e-05, "loss": 0.295, "step": 8378 }, { "epoch": 1.46, "learning_rate": 4.267005122275081e-05, "loss": 0.3135, "step": 8379 }, { "epoch": 1.46, "learning_rate": 4.266499499498402e-05, "loss": 0.3111, "step": 8380 }, { "epoch": 1.46, "learning_rate": 4.265993736350601e-05, "loss": 0.31, "step": 8381 }, { "epoch": 1.46, "learning_rate": 4.265487832878492e-05, "loss": 0.3059, "step": 8382 }, { "epoch": 1.46, "learning_rate": 4.264981789128904e-05, "loss": 0.3169, "step": 8383 }, { "epoch": 1.46, "learning_rate": 4.2644756051486764e-05, "loss": 0.3169, "step": 8384 }, { "epoch": 1.46, "learning_rate": 4.263969280984664e-05, "loss": 0.3062, "step": 8385 }, { "epoch": 1.46, "learning_rate": 4.2634628166837325e-05, "loss": 0.3059, "step": 8386 }, { "epoch": 1.46, "learning_rate": 4.262956212292763e-05, "loss": 0.3127, "step": 8387 }, { "epoch": 1.46, "learning_rate": 4.2624494678586455e-05, "loss": 0.3058, "step": 8388 }, { "epoch": 1.46, "learning_rate": 4.261942583428288e-05, "loss": 0.3102, "step": 8389 }, { "epoch": 1.46, "learning_rate": 4.2614355590486095e-05, "loss": 0.3225, "step": 8390 }, { "epoch": 1.46, "learning_rate": 4.260928394766539e-05, "loss": 0.3114, "step": 8391 }, { "epoch": 1.46, "learning_rate": 4.2604210906290215e-05, "loss": 0.315, "step": 8392 }, { "epoch": 1.46, "learning_rate": 4.2599136466830156e-05, "loss": 0.3062, "step": 8393 }, { "epoch": 1.46, "learning_rate": 4.2594060629754925e-05, "loss": 0.3178, "step": 8394 }, { "epoch": 1.46, "learning_rate": 4.258898339553432e-05, "loss": 0.319, "step": 8395 }, { "epoch": 1.46, "learning_rate": 4.2583904764638334e-05, "loss": 0.3071, "step": 8396 }, { "epoch": 1.46, "learning_rate": 4.2578824737537044e-05, "loss": 0.3062, "step": 8397 }, { "epoch": 1.46, "learning_rate": 4.257374331470068e-05, "loss": 0.3007, "step": 8398 }, { "epoch": 1.46, "learning_rate": 4.2568660496599576e-05, "loss": 0.318, "step": 8399 }, { "epoch": 1.46, "learning_rate": 4.256357628370423e-05, "loss": 0.3122, "step": 8400 }, { "epoch": 1.46, "learning_rate": 4.255849067648524e-05, "loss": 0.3116, "step": 8401 }, { "epoch": 1.46, "learning_rate": 4.255340367541334e-05, "loss": 0.2991, "step": 8402 }, { "epoch": 1.46, "learning_rate": 4.2548315280959394e-05, "loss": 0.2973, "step": 8403 }, { "epoch": 1.46, "learning_rate": 4.2543225493594414e-05, "loss": 0.3191, "step": 8404 }, { "epoch": 1.46, "learning_rate": 4.2538134313789506e-05, "loss": 0.3205, "step": 8405 }, { "epoch": 1.47, "learning_rate": 4.2533041742015925e-05, "loss": 0.3118, "step": 8406 }, { "epoch": 1.47, "learning_rate": 4.252794777874506e-05, "loss": 0.3157, "step": 8407 }, { "epoch": 1.47, "learning_rate": 4.252285242444842e-05, "loss": 0.3048, "step": 8408 }, { "epoch": 1.47, "learning_rate": 4.2517755679597646e-05, "loss": 0.3262, "step": 8409 }, { "epoch": 1.47, "learning_rate": 4.2512657544664506e-05, "loss": 0.3008, "step": 8410 }, { "epoch": 1.47, "learning_rate": 4.2507558020120886e-05, "loss": 0.309, "step": 8411 }, { "epoch": 1.47, "learning_rate": 4.2502457106438815e-05, "loss": 0.3032, "step": 8412 }, { "epoch": 1.47, "learning_rate": 4.249735480409045e-05, "loss": 0.3181, "step": 8413 }, { "epoch": 1.47, "learning_rate": 4.249225111354808e-05, "loss": 0.3141, "step": 8414 }, { "epoch": 1.47, "learning_rate": 4.24871460352841e-05, "loss": 0.3186, "step": 8415 }, { "epoch": 1.47, "learning_rate": 4.248203956977107e-05, "loss": 0.3306, "step": 8416 }, { "epoch": 1.47, "learning_rate": 4.247693171748164e-05, "loss": 0.3098, "step": 8417 }, { "epoch": 1.47, "learning_rate": 4.2471822478888605e-05, "loss": 0.3028, "step": 8418 }, { "epoch": 1.47, "learning_rate": 4.246671185446491e-05, "loss": 0.3097, "step": 8419 }, { "epoch": 1.47, "learning_rate": 4.246159984468359e-05, "loss": 0.3113, "step": 8420 }, { "epoch": 1.47, "learning_rate": 4.245648645001782e-05, "loss": 0.3134, "step": 8421 }, { "epoch": 1.47, "learning_rate": 4.245137167094093e-05, "loss": 0.3055, "step": 8422 }, { "epoch": 1.47, "learning_rate": 4.244625550792634e-05, "loss": 0.3018, "step": 8423 }, { "epoch": 1.47, "learning_rate": 4.244113796144763e-05, "loss": 0.3077, "step": 8424 }, { "epoch": 1.47, "learning_rate": 4.243601903197847e-05, "loss": 0.2954, "step": 8425 }, { "epoch": 1.47, "learning_rate": 4.243089871999271e-05, "loss": 0.3138, "step": 8426 }, { "epoch": 1.47, "learning_rate": 4.2425777025964284e-05, "loss": 0.2947, "step": 8427 }, { "epoch": 1.47, "learning_rate": 4.2420653950367264e-05, "loss": 0.3232, "step": 8428 }, { "epoch": 1.47, "learning_rate": 4.241552949367587e-05, "loss": 0.3193, "step": 8429 }, { "epoch": 1.47, "learning_rate": 4.241040365636442e-05, "loss": 0.3033, "step": 8430 }, { "epoch": 1.47, "learning_rate": 4.240527643890738e-05, "loss": 0.3051, "step": 8431 }, { "epoch": 1.47, "learning_rate": 4.240014784177935e-05, "loss": 0.3166, "step": 8432 }, { "epoch": 1.47, "learning_rate": 4.239501786545503e-05, "loss": 0.2924, "step": 8433 }, { "epoch": 1.47, "learning_rate": 4.238988651040927e-05, "loss": 0.3162, "step": 8434 }, { "epoch": 1.47, "learning_rate": 4.2384753777117045e-05, "loss": 0.3129, "step": 8435 }, { "epoch": 1.47, "learning_rate": 4.2379619666053443e-05, "loss": 0.3143, "step": 8436 }, { "epoch": 1.47, "learning_rate": 4.237448417769371e-05, "loss": 0.2966, "step": 8437 }, { "epoch": 1.47, "learning_rate": 4.236934731251318e-05, "loss": 0.2963, "step": 8438 }, { "epoch": 1.47, "learning_rate": 4.2364209070987346e-05, "loss": 0.3351, "step": 8439 }, { "epoch": 1.47, "learning_rate": 4.2359069453591816e-05, "loss": 0.3257, "step": 8440 }, { "epoch": 1.47, "learning_rate": 4.2353928460802326e-05, "loss": 0.3005, "step": 8441 }, { "epoch": 1.47, "learning_rate": 4.2348786093094733e-05, "loss": 0.3177, "step": 8442 }, { "epoch": 1.47, "learning_rate": 4.2343642350945036e-05, "loss": 0.3129, "step": 8443 }, { "epoch": 1.47, "learning_rate": 4.233849723482935e-05, "loss": 0.3179, "step": 8444 }, { "epoch": 1.47, "learning_rate": 4.2333350745223924e-05, "loss": 0.3061, "step": 8445 }, { "epoch": 1.47, "learning_rate": 4.2328202882605124e-05, "loss": 0.3116, "step": 8446 }, { "epoch": 1.47, "learning_rate": 4.232305364744945e-05, "loss": 0.3204, "step": 8447 }, { "epoch": 1.47, "learning_rate": 4.231790304023353e-05, "loss": 0.317, "step": 8448 }, { "epoch": 1.47, "learning_rate": 4.231275106143412e-05, "loss": 0.3212, "step": 8449 }, { "epoch": 1.47, "learning_rate": 4.2307597711528106e-05, "loss": 0.314, "step": 8450 }, { "epoch": 1.47, "learning_rate": 4.230244299099248e-05, "loss": 0.3179, "step": 8451 }, { "epoch": 1.47, "learning_rate": 4.229728690030438e-05, "loss": 0.3159, "step": 8452 }, { "epoch": 1.47, "learning_rate": 4.229212943994108e-05, "loss": 0.3025, "step": 8453 }, { "epoch": 1.47, "learning_rate": 4.2286970610379966e-05, "loss": 0.3225, "step": 8454 }, { "epoch": 1.47, "learning_rate": 4.228181041209853e-05, "loss": 0.3145, "step": 8455 }, { "epoch": 1.47, "learning_rate": 4.2276648845574444e-05, "loss": 0.3111, "step": 8456 }, { "epoch": 1.47, "learning_rate": 4.227148591128546e-05, "loss": 0.3125, "step": 8457 }, { "epoch": 1.47, "learning_rate": 4.226632160970946e-05, "loss": 0.2998, "step": 8458 }, { "epoch": 1.47, "learning_rate": 4.226115594132449e-05, "loss": 0.3012, "step": 8459 }, { "epoch": 1.47, "learning_rate": 4.225598890660868e-05, "loss": 0.3338, "step": 8460 }, { "epoch": 1.47, "learning_rate": 4.2250820506040326e-05, "loss": 0.3182, "step": 8461 }, { "epoch": 1.47, "learning_rate": 4.22456507400978e-05, "loss": 0.3081, "step": 8462 }, { "epoch": 1.48, "learning_rate": 4.224047960925965e-05, "loss": 0.3081, "step": 8463 }, { "epoch": 1.48, "learning_rate": 4.223530711400452e-05, "loss": 0.31, "step": 8464 }, { "epoch": 1.48, "learning_rate": 4.2230133254811184e-05, "loss": 0.2999, "step": 8465 }, { "epoch": 1.48, "learning_rate": 4.222495803215856e-05, "loss": 0.3168, "step": 8466 }, { "epoch": 1.48, "learning_rate": 4.2219781446525676e-05, "loss": 0.3141, "step": 8467 }, { "epoch": 1.48, "learning_rate": 4.221460349839169e-05, "loss": 0.3102, "step": 8468 }, { "epoch": 1.48, "learning_rate": 4.220942418823588e-05, "loss": 0.3047, "step": 8469 }, { "epoch": 1.48, "learning_rate": 4.2204243516537675e-05, "loss": 0.2944, "step": 8470 }, { "epoch": 1.48, "learning_rate": 4.219906148377659e-05, "loss": 0.3049, "step": 8471 }, { "epoch": 1.48, "learning_rate": 4.2193878090432296e-05, "loss": 0.3075, "step": 8472 }, { "epoch": 1.48, "learning_rate": 4.2188693336984576e-05, "loss": 0.3012, "step": 8473 }, { "epoch": 1.48, "learning_rate": 4.218350722391335e-05, "loss": 0.2996, "step": 8474 }, { "epoch": 1.48, "learning_rate": 4.2178319751698665e-05, "loss": 0.3124, "step": 8475 }, { "epoch": 1.48, "learning_rate": 4.217313092082068e-05, "loss": 0.3167, "step": 8476 }, { "epoch": 1.48, "learning_rate": 4.2167940731759674e-05, "loss": 0.3083, "step": 8477 }, { "epoch": 1.48, "learning_rate": 4.216274918499608e-05, "loss": 0.3049, "step": 8478 }, { "epoch": 1.48, "learning_rate": 4.215755628101045e-05, "loss": 0.3062, "step": 8479 }, { "epoch": 1.48, "learning_rate": 4.2152362020283426e-05, "loss": 0.286, "step": 8480 }, { "epoch": 1.48, "learning_rate": 4.214716640329582e-05, "loss": 0.307, "step": 8481 }, { "epoch": 1.48, "learning_rate": 4.214196943052855e-05, "loss": 0.2957, "step": 8482 }, { "epoch": 1.48, "learning_rate": 4.2136771102462655e-05, "loss": 0.326, "step": 8483 }, { "epoch": 1.48, "learning_rate": 4.213157141957931e-05, "loss": 0.308, "step": 8484 }, { "epoch": 1.48, "learning_rate": 4.2126370382359815e-05, "loss": 0.3086, "step": 8485 }, { "epoch": 1.48, "learning_rate": 4.212116799128559e-05, "loss": 0.3128, "step": 8486 }, { "epoch": 1.48, "learning_rate": 4.211596424683817e-05, "loss": 0.3106, "step": 8487 }, { "epoch": 1.48, "learning_rate": 4.211075914949924e-05, "loss": 0.3104, "step": 8488 }, { "epoch": 1.48, "learning_rate": 4.2105552699750595e-05, "loss": 0.2995, "step": 8489 }, { "epoch": 1.48, "learning_rate": 4.210034489807416e-05, "loss": 0.3001, "step": 8490 }, { "epoch": 1.48, "learning_rate": 4.209513574495197e-05, "loss": 0.3027, "step": 8491 }, { "epoch": 1.48, "learning_rate": 4.2089925240866206e-05, "loss": 0.3136, "step": 8492 }, { "epoch": 1.48, "learning_rate": 4.208471338629918e-05, "loss": 0.3077, "step": 8493 }, { "epoch": 1.48, "learning_rate": 4.207950018173328e-05, "loss": 0.3131, "step": 8494 }, { "epoch": 1.48, "learning_rate": 4.2074285627651086e-05, "loss": 0.3103, "step": 8495 }, { "epoch": 1.48, "learning_rate": 4.206906972453527e-05, "loss": 0.312, "step": 8496 }, { "epoch": 1.48, "learning_rate": 4.2063852472868595e-05, "loss": 0.3255, "step": 8497 }, { "epoch": 1.48, "learning_rate": 4.2058633873134025e-05, "loss": 0.3025, "step": 8498 }, { "epoch": 1.48, "learning_rate": 4.205341392581458e-05, "loss": 0.3189, "step": 8499 }, { "epoch": 1.48, "learning_rate": 4.204819263139344e-05, "loss": 0.3168, "step": 8500 }, { "epoch": 1.48, "learning_rate": 4.2042969990353916e-05, "loss": 0.2972, "step": 8501 }, { "epoch": 1.48, "learning_rate": 4.20377460031794e-05, "loss": 0.3097, "step": 8502 }, { "epoch": 1.48, "learning_rate": 4.2032520670353464e-05, "loss": 0.3071, "step": 8503 }, { "epoch": 1.48, "learning_rate": 4.202729399235976e-05, "loss": 0.2896, "step": 8504 }, { "epoch": 1.48, "learning_rate": 4.20220659696821e-05, "loss": 0.3344, "step": 8505 }, { "epoch": 1.48, "learning_rate": 4.2016836602804386e-05, "loss": 0.323, "step": 8506 }, { "epoch": 1.48, "learning_rate": 4.201160589221068e-05, "loss": 0.2967, "step": 8507 }, { "epoch": 1.48, "learning_rate": 4.2006373838385125e-05, "loss": 0.3295, "step": 8508 }, { "epoch": 1.48, "learning_rate": 4.200114044181204e-05, "loss": 0.3172, "step": 8509 }, { "epoch": 1.48, "learning_rate": 4.199590570297583e-05, "loss": 0.313, "step": 8510 }, { "epoch": 1.48, "learning_rate": 4.199066962236103e-05, "loss": 0.3047, "step": 8511 }, { "epoch": 1.48, "learning_rate": 4.198543220045231e-05, "loss": 0.3239, "step": 8512 }, { "epoch": 1.48, "learning_rate": 4.198019343773447e-05, "loss": 0.3048, "step": 8513 }, { "epoch": 1.48, "learning_rate": 4.1974953334692406e-05, "loss": 0.3114, "step": 8514 }, { "epoch": 1.48, "learning_rate": 4.196971189181117e-05, "loss": 0.3053, "step": 8515 }, { "epoch": 1.48, "learning_rate": 4.196446910957591e-05, "loss": 0.3196, "step": 8516 }, { "epoch": 1.48, "learning_rate": 4.195922498847192e-05, "loss": 0.3175, "step": 8517 }, { "epoch": 1.48, "learning_rate": 4.195397952898461e-05, "loss": 0.3122, "step": 8518 }, { "epoch": 1.48, "learning_rate": 4.1948732731599515e-05, "loss": 0.3142, "step": 8519 }, { "epoch": 1.49, "learning_rate": 4.194348459680228e-05, "loss": 0.2965, "step": 8520 }, { "epoch": 1.49, "learning_rate": 4.19382351250787e-05, "loss": 0.3252, "step": 8521 }, { "epoch": 1.49, "learning_rate": 4.193298431691467e-05, "loss": 0.3041, "step": 8522 }, { "epoch": 1.49, "learning_rate": 4.192773217279623e-05, "loss": 0.3045, "step": 8523 }, { "epoch": 1.49, "learning_rate": 4.1922478693209525e-05, "loss": 0.3041, "step": 8524 }, { "epoch": 1.49, "learning_rate": 4.1917223878640824e-05, "loss": 0.3149, "step": 8525 }, { "epoch": 1.49, "learning_rate": 4.1911967729576526e-05, "loss": 0.3296, "step": 8526 }, { "epoch": 1.49, "learning_rate": 4.190671024650317e-05, "loss": 0.3225, "step": 8527 }, { "epoch": 1.49, "learning_rate": 4.190145142990739e-05, "loss": 0.311, "step": 8528 }, { "epoch": 1.49, "learning_rate": 4.189619128027596e-05, "loss": 0.3096, "step": 8529 }, { "epoch": 1.49, "learning_rate": 4.1890929798095776e-05, "loss": 0.3119, "step": 8530 }, { "epoch": 1.49, "learning_rate": 4.1885666983853845e-05, "loss": 0.3097, "step": 8531 }, { "epoch": 1.49, "learning_rate": 4.188040283803731e-05, "loss": 0.3103, "step": 8532 }, { "epoch": 1.49, "learning_rate": 4.1875137361133435e-05, "loss": 0.3179, "step": 8533 }, { "epoch": 1.49, "learning_rate": 4.1869870553629606e-05, "loss": 0.3129, "step": 8534 }, { "epoch": 1.49, "learning_rate": 4.186460241601333e-05, "loss": 0.322, "step": 8535 }, { "epoch": 1.49, "learning_rate": 4.185933294877225e-05, "loss": 0.3058, "step": 8536 }, { "epoch": 1.49, "learning_rate": 4.1854062152394115e-05, "loss": 0.3038, "step": 8537 }, { "epoch": 1.49, "learning_rate": 4.1848790027366805e-05, "loss": 0.314, "step": 8538 }, { "epoch": 1.49, "learning_rate": 4.18435165741783e-05, "loss": 0.3049, "step": 8539 }, { "epoch": 1.49, "learning_rate": 4.183824179331677e-05, "loss": 0.3056, "step": 8540 }, { "epoch": 1.49, "learning_rate": 4.183296568527042e-05, "loss": 0.2954, "step": 8541 }, { "epoch": 1.49, "learning_rate": 4.182768825052763e-05, "loss": 0.3171, "step": 8542 }, { "epoch": 1.49, "learning_rate": 4.182240948957691e-05, "loss": 0.297, "step": 8543 }, { "epoch": 1.49, "learning_rate": 4.1817129402906864e-05, "loss": 0.3136, "step": 8544 }, { "epoch": 1.49, "learning_rate": 4.1811847991006234e-05, "loss": 0.306, "step": 8545 }, { "epoch": 1.49, "learning_rate": 4.180656525436387e-05, "loss": 0.2983, "step": 8546 }, { "epoch": 1.49, "learning_rate": 4.180128119346878e-05, "loss": 0.3139, "step": 8547 }, { "epoch": 1.49, "learning_rate": 4.179599580881004e-05, "loss": 0.3028, "step": 8548 }, { "epoch": 1.49, "learning_rate": 4.17907091008769e-05, "loss": 0.2996, "step": 8549 }, { "epoch": 1.49, "learning_rate": 4.17854210701587e-05, "loss": 0.316, "step": 8550 }, { "epoch": 1.49, "learning_rate": 4.178013171714492e-05, "loss": 0.3028, "step": 8551 }, { "epoch": 1.49, "learning_rate": 4.177484104232516e-05, "loss": 0.3109, "step": 8552 }, { "epoch": 1.49, "learning_rate": 4.176954904618913e-05, "loss": 0.3148, "step": 8553 }, { "epoch": 1.49, "learning_rate": 4.1764255729226676e-05, "loss": 0.3136, "step": 8554 }, { "epoch": 1.49, "learning_rate": 4.1758961091927765e-05, "loss": 0.317, "step": 8555 }, { "epoch": 1.49, "learning_rate": 4.175366513478247e-05, "loss": 0.3156, "step": 8556 }, { "epoch": 1.49, "learning_rate": 4.174836785828101e-05, "loss": 0.2898, "step": 8557 }, { "epoch": 1.49, "learning_rate": 4.1743069262913706e-05, "loss": 0.322, "step": 8558 }, { "epoch": 1.49, "learning_rate": 4.173776934917102e-05, "loss": 0.3045, "step": 8559 }, { "epoch": 1.49, "learning_rate": 4.173246811754351e-05, "loss": 0.3131, "step": 8560 }, { "epoch": 1.49, "learning_rate": 4.172716556852189e-05, "loss": 0.3105, "step": 8561 }, { "epoch": 1.49, "learning_rate": 4.172186170259697e-05, "loss": 0.3104, "step": 8562 }, { "epoch": 1.49, "learning_rate": 4.1716556520259684e-05, "loss": 0.3019, "step": 8563 }, { "epoch": 1.49, "learning_rate": 4.171125002200109e-05, "loss": 0.3226, "step": 8564 }, { "epoch": 1.49, "learning_rate": 4.170594220831239e-05, "loss": 0.3024, "step": 8565 }, { "epoch": 1.49, "learning_rate": 4.1700633079684884e-05, "loss": 0.2961, "step": 8566 }, { "epoch": 1.49, "learning_rate": 4.169532263660998e-05, "loss": 0.3142, "step": 8567 }, { "epoch": 1.49, "learning_rate": 4.169001087957924e-05, "loss": 0.3137, "step": 8568 }, { "epoch": 1.49, "learning_rate": 4.168469780908434e-05, "loss": 0.3049, "step": 8569 }, { "epoch": 1.49, "learning_rate": 4.167938342561706e-05, "loss": 0.2942, "step": 8570 }, { "epoch": 1.49, "learning_rate": 4.167406772966933e-05, "loss": 0.318, "step": 8571 }, { "epoch": 1.49, "learning_rate": 4.1668750721733166e-05, "loss": 0.3008, "step": 8572 }, { "epoch": 1.49, "learning_rate": 4.166343240230072e-05, "loss": 0.3359, "step": 8573 }, { "epoch": 1.49, "learning_rate": 4.165811277186429e-05, "loss": 0.3258, "step": 8574 }, { "epoch": 1.49, "learning_rate": 4.1652791830916266e-05, "loss": 0.3088, "step": 8575 }, { "epoch": 1.49, "learning_rate": 4.164746957994916e-05, "loss": 0.3054, "step": 8576 }, { "epoch": 1.49, "learning_rate": 4.1642146019455633e-05, "loss": 0.3146, "step": 8577 }, { "epoch": 1.5, "learning_rate": 4.163682114992842e-05, "loss": 0.3175, "step": 8578 }, { "epoch": 1.5, "learning_rate": 4.163149497186043e-05, "loss": 0.2997, "step": 8579 }, { "epoch": 1.5, "learning_rate": 4.162616748574465e-05, "loss": 0.3256, "step": 8580 }, { "epoch": 1.5, "learning_rate": 4.162083869207422e-05, "loss": 0.3038, "step": 8581 }, { "epoch": 1.5, "learning_rate": 4.161550859134238e-05, "loss": 0.3141, "step": 8582 }, { "epoch": 1.5, "learning_rate": 4.161017718404249e-05, "loss": 0.2922, "step": 8583 }, { "epoch": 1.5, "learning_rate": 4.160484447066805e-05, "loss": 0.3049, "step": 8584 }, { "epoch": 1.5, "learning_rate": 4.159951045171267e-05, "loss": 0.3158, "step": 8585 }, { "epoch": 1.5, "learning_rate": 4.159417512767008e-05, "loss": 0.3187, "step": 8586 }, { "epoch": 1.5, "learning_rate": 4.1588838499034124e-05, "loss": 0.3091, "step": 8587 }, { "epoch": 1.5, "learning_rate": 4.158350056629878e-05, "loss": 0.3125, "step": 8588 }, { "epoch": 1.5, "learning_rate": 4.1578161329958135e-05, "loss": 0.3079, "step": 8589 }, { "epoch": 1.5, "learning_rate": 4.157282079050642e-05, "loss": 0.3259, "step": 8590 }, { "epoch": 1.5, "learning_rate": 4.156747894843794e-05, "loss": 0.3064, "step": 8591 }, { "epoch": 1.5, "learning_rate": 4.1562135804247174e-05, "loss": 0.3128, "step": 8592 }, { "epoch": 1.5, "learning_rate": 4.15567913584287e-05, "loss": 0.3091, "step": 8593 }, { "epoch": 1.5, "learning_rate": 4.155144561147719e-05, "loss": 0.3009, "step": 8594 }, { "epoch": 1.5, "learning_rate": 4.154609856388748e-05, "loss": 0.3092, "step": 8595 }, { "epoch": 1.5, "learning_rate": 4.15407502161545e-05, "loss": 0.3132, "step": 8596 }, { "epoch": 1.5, "learning_rate": 4.1535400568773295e-05, "loss": 0.3116, "step": 8597 }, { "epoch": 1.5, "learning_rate": 4.153004962223907e-05, "loss": 0.3015, "step": 8598 }, { "epoch": 1.5, "learning_rate": 4.15246973770471e-05, "loss": 0.3014, "step": 8599 }, { "epoch": 1.5, "learning_rate": 4.15193438336928e-05, "loss": 0.3116, "step": 8600 }, { "epoch": 1.5, "learning_rate": 4.1513988992671734e-05, "loss": 0.3172, "step": 8601 }, { "epoch": 1.5, "learning_rate": 4.150863285447954e-05, "loss": 0.3139, "step": 8602 }, { "epoch": 1.5, "learning_rate": 4.150327541961199e-05, "loss": 0.3098, "step": 8603 }, { "epoch": 1.5, "learning_rate": 4.1497916688565e-05, "loss": 0.3093, "step": 8604 }, { "epoch": 1.5, "learning_rate": 4.1492556661834577e-05, "loss": 0.2965, "step": 8605 }, { "epoch": 1.5, "learning_rate": 4.1487195339916854e-05, "loss": 0.3176, "step": 8606 }, { "epoch": 1.5, "learning_rate": 4.1481832723308106e-05, "loss": 0.299, "step": 8607 }, { "epoch": 1.5, "learning_rate": 4.147646881250471e-05, "loss": 0.3056, "step": 8608 }, { "epoch": 1.5, "learning_rate": 4.1471103608003134e-05, "loss": 0.3106, "step": 8609 }, { "epoch": 1.5, "learning_rate": 4.146573711030002e-05, "loss": 0.3249, "step": 8610 }, { "epoch": 1.5, "learning_rate": 4.14603693198921e-05, "loss": 0.3134, "step": 8611 }, { "epoch": 1.5, "learning_rate": 4.145500023727624e-05, "loss": 0.3092, "step": 8612 }, { "epoch": 1.5, "learning_rate": 4.1449629862949404e-05, "loss": 0.3079, "step": 8613 }, { "epoch": 1.5, "learning_rate": 4.144425819740868e-05, "loss": 0.3199, "step": 8614 }, { "epoch": 1.5, "learning_rate": 4.143888524115131e-05, "loss": 0.3047, "step": 8615 }, { "epoch": 1.5, "learning_rate": 4.143351099467461e-05, "loss": 0.3349, "step": 8616 }, { "epoch": 1.5, "learning_rate": 4.1428135458476024e-05, "loss": 0.3109, "step": 8617 }, { "epoch": 1.5, "learning_rate": 4.142275863305315e-05, "loss": 0.3095, "step": 8618 }, { "epoch": 1.5, "learning_rate": 4.141738051890367e-05, "loss": 0.319, "step": 8619 }, { "epoch": 1.5, "learning_rate": 4.141200111652539e-05, "loss": 0.3064, "step": 8620 }, { "epoch": 1.5, "learning_rate": 4.140662042641625e-05, "loss": 0.3025, "step": 8621 }, { "epoch": 1.5, "learning_rate": 4.14012384490743e-05, "loss": 0.323, "step": 8622 }, { "epoch": 1.5, "learning_rate": 4.139585518499771e-05, "loss": 0.3045, "step": 8623 }, { "epoch": 1.5, "learning_rate": 4.1390470634684756e-05, "loss": 0.3093, "step": 8624 }, { "epoch": 1.5, "learning_rate": 4.1385084798633865e-05, "loss": 0.3033, "step": 8625 }, { "epoch": 1.5, "learning_rate": 4.1379697677343555e-05, "loss": 0.3201, "step": 8626 }, { "epoch": 1.5, "learning_rate": 4.137430927131248e-05, "loss": 0.3009, "step": 8627 }, { "epoch": 1.5, "learning_rate": 4.136891958103938e-05, "loss": 0.3342, "step": 8628 }, { "epoch": 1.5, "learning_rate": 4.136352860702316e-05, "loss": 0.3203, "step": 8629 }, { "epoch": 1.5, "learning_rate": 4.135813634976282e-05, "loss": 0.3068, "step": 8630 }, { "epoch": 1.5, "learning_rate": 4.135274280975748e-05, "loss": 0.2961, "step": 8631 }, { "epoch": 1.5, "learning_rate": 4.1347347987506394e-05, "loss": 0.3229, "step": 8632 }, { "epoch": 1.5, "learning_rate": 4.134195188350888e-05, "loss": 0.2933, "step": 8633 }, { "epoch": 1.5, "learning_rate": 4.133655449826446e-05, "loss": 0.3077, "step": 8634 }, { "epoch": 1.51, "learning_rate": 4.1331155832272705e-05, "loss": 0.3014, "step": 8635 }, { "epoch": 1.51, "learning_rate": 4.132575588603334e-05, "loss": 0.3084, "step": 8636 }, { "epoch": 1.51, "learning_rate": 4.1320354660046195e-05, "loss": 0.3135, "step": 8637 }, { "epoch": 1.51, "learning_rate": 4.131495215481122e-05, "loss": 0.3013, "step": 8638 }, { "epoch": 1.51, "learning_rate": 4.130954837082848e-05, "loss": 0.3069, "step": 8639 }, { "epoch": 1.51, "learning_rate": 4.130414330859817e-05, "loss": 0.3242, "step": 8640 }, { "epoch": 1.51, "learning_rate": 4.12987369686206e-05, "loss": 0.3237, "step": 8641 }, { "epoch": 1.51, "learning_rate": 4.1293329351396185e-05, "loss": 0.3025, "step": 8642 }, { "epoch": 1.51, "learning_rate": 4.1287920457425476e-05, "loss": 0.3123, "step": 8643 }, { "epoch": 1.51, "learning_rate": 4.128251028720913e-05, "loss": 0.313, "step": 8644 }, { "epoch": 1.51, "learning_rate": 4.127709884124792e-05, "loss": 0.3344, "step": 8645 }, { "epoch": 1.51, "learning_rate": 4.127168612004277e-05, "loss": 0.2974, "step": 8646 }, { "epoch": 1.51, "learning_rate": 4.126627212409467e-05, "loss": 0.3057, "step": 8647 }, { "epoch": 1.51, "learning_rate": 4.126085685390475e-05, "loss": 0.3017, "step": 8648 }, { "epoch": 1.51, "learning_rate": 4.125544030997428e-05, "loss": 0.3029, "step": 8649 }, { "epoch": 1.51, "learning_rate": 4.125002249280462e-05, "loss": 0.3257, "step": 8650 }, { "epoch": 1.51, "learning_rate": 4.124460340289726e-05, "loss": 0.3072, "step": 8651 }, { "epoch": 1.51, "learning_rate": 4.12391830407538e-05, "loss": 0.3066, "step": 8652 }, { "epoch": 1.51, "learning_rate": 4.123376140687597e-05, "loss": 0.3196, "step": 8653 }, { "epoch": 1.51, "learning_rate": 4.1228338501765605e-05, "loss": 0.3027, "step": 8654 }, { "epoch": 1.51, "learning_rate": 4.1222914325924664e-05, "loss": 0.3086, "step": 8655 }, { "epoch": 1.51, "learning_rate": 4.121748887985522e-05, "loss": 0.2986, "step": 8656 }, { "epoch": 1.51, "learning_rate": 4.1212062164059474e-05, "loss": 0.2974, "step": 8657 }, { "epoch": 1.51, "learning_rate": 4.1206634179039725e-05, "loss": 0.3192, "step": 8658 }, { "epoch": 1.51, "learning_rate": 4.120120492529842e-05, "loss": 0.3065, "step": 8659 }, { "epoch": 1.51, "learning_rate": 4.119577440333809e-05, "loss": 0.3125, "step": 8660 }, { "epoch": 1.51, "learning_rate": 4.1190342613661404e-05, "loss": 0.3146, "step": 8661 }, { "epoch": 1.51, "learning_rate": 4.118490955677113e-05, "loss": 0.3123, "step": 8662 }, { "epoch": 1.51, "learning_rate": 4.1179475233170185e-05, "loss": 0.3142, "step": 8663 }, { "epoch": 1.51, "learning_rate": 4.1174039643361574e-05, "loss": 0.2974, "step": 8664 }, { "epoch": 1.51, "learning_rate": 4.1168602787848434e-05, "loss": 0.3049, "step": 8665 }, { "epoch": 1.51, "learning_rate": 4.1163164667134006e-05, "loss": 0.3026, "step": 8666 }, { "epoch": 1.51, "learning_rate": 4.115772528172166e-05, "loss": 0.3042, "step": 8667 }, { "epoch": 1.51, "learning_rate": 4.115228463211489e-05, "loss": 0.3122, "step": 8668 }, { "epoch": 1.51, "learning_rate": 4.114684271881728e-05, "loss": 0.3045, "step": 8669 }, { "epoch": 1.51, "learning_rate": 4.114139954233256e-05, "loss": 0.2934, "step": 8670 }, { "epoch": 1.51, "learning_rate": 4.113595510316456e-05, "loss": 0.3092, "step": 8671 }, { "epoch": 1.51, "learning_rate": 4.1130509401817224e-05, "loss": 0.3116, "step": 8672 }, { "epoch": 1.51, "learning_rate": 4.1125062438794623e-05, "loss": 0.3114, "step": 8673 }, { "epoch": 1.51, "learning_rate": 4.1119614214600964e-05, "loss": 0.3087, "step": 8674 }, { "epoch": 1.51, "learning_rate": 4.111416472974052e-05, "loss": 0.3101, "step": 8675 }, { "epoch": 1.51, "learning_rate": 4.110871398471773e-05, "loss": 0.3015, "step": 8676 }, { "epoch": 1.51, "learning_rate": 4.110326198003711e-05, "loss": 0.3031, "step": 8677 }, { "epoch": 1.51, "learning_rate": 4.109780871620332e-05, "loss": 0.3058, "step": 8678 }, { "epoch": 1.51, "learning_rate": 4.109235419372114e-05, "loss": 0.3162, "step": 8679 }, { "epoch": 1.51, "learning_rate": 4.108689841309545e-05, "loss": 0.325, "step": 8680 }, { "epoch": 1.51, "learning_rate": 4.108144137483123e-05, "loss": 0.3093, "step": 8681 }, { "epoch": 1.51, "learning_rate": 4.1075983079433616e-05, "loss": 0.3144, "step": 8682 }, { "epoch": 1.51, "learning_rate": 4.107052352740784e-05, "loss": 0.3113, "step": 8683 }, { "epoch": 1.51, "learning_rate": 4.106506271925926e-05, "loss": 0.3069, "step": 8684 }, { "epoch": 1.51, "learning_rate": 4.105960065549332e-05, "loss": 0.315, "step": 8685 }, { "epoch": 1.51, "learning_rate": 4.105413733661563e-05, "loss": 0.3112, "step": 8686 }, { "epoch": 1.51, "learning_rate": 4.104867276313186e-05, "loss": 0.3077, "step": 8687 }, { "epoch": 1.51, "learning_rate": 4.104320693554785e-05, "loss": 0.2937, "step": 8688 }, { "epoch": 1.51, "learning_rate": 4.103773985436953e-05, "loss": 0.3069, "step": 8689 }, { "epoch": 1.51, "learning_rate": 4.1032271520102927e-05, "loss": 0.3072, "step": 8690 }, { "epoch": 1.51, "learning_rate": 4.102680193325423e-05, "loss": 0.2974, "step": 8691 }, { "epoch": 1.52, "learning_rate": 4.1021331094329686e-05, "loss": 0.2994, "step": 8692 }, { "epoch": 1.52, "learning_rate": 4.101585900383572e-05, "loss": 0.3212, "step": 8693 }, { "epoch": 1.52, "learning_rate": 4.1010385662278834e-05, "loss": 0.3048, "step": 8694 }, { "epoch": 1.52, "learning_rate": 4.100491107016564e-05, "loss": 0.3073, "step": 8695 }, { "epoch": 1.52, "learning_rate": 4.09994352280029e-05, "loss": 0.3187, "step": 8696 }, { "epoch": 1.52, "learning_rate": 4.0993958136297474e-05, "loss": 0.2855, "step": 8697 }, { "epoch": 1.52, "learning_rate": 4.0988479795556314e-05, "loss": 0.3023, "step": 8698 }, { "epoch": 1.52, "learning_rate": 4.098300020628652e-05, "loss": 0.3183, "step": 8699 }, { "epoch": 1.52, "learning_rate": 4.0977519368995305e-05, "loss": 0.3059, "step": 8700 }, { "epoch": 1.52, "learning_rate": 4.097203728418998e-05, "loss": 0.315, "step": 8701 }, { "epoch": 1.52, "learning_rate": 4.096655395237799e-05, "loss": 0.3025, "step": 8702 }, { "epoch": 1.52, "learning_rate": 4.096106937406688e-05, "loss": 0.3104, "step": 8703 }, { "epoch": 1.52, "learning_rate": 4.095558354976431e-05, "loss": 0.3198, "step": 8704 }, { "epoch": 1.52, "learning_rate": 4.095009647997807e-05, "loss": 0.3065, "step": 8705 }, { "epoch": 1.52, "learning_rate": 4.094460816521606e-05, "loss": 0.3094, "step": 8706 }, { "epoch": 1.52, "learning_rate": 4.09391186059863e-05, "loss": 0.309, "step": 8707 }, { "epoch": 1.52, "learning_rate": 4.093362780279689e-05, "loss": 0.2975, "step": 8708 }, { "epoch": 1.52, "learning_rate": 4.0928135756156105e-05, "loss": 0.2999, "step": 8709 }, { "epoch": 1.52, "learning_rate": 4.092264246657228e-05, "loss": 0.3034, "step": 8710 }, { "epoch": 1.52, "learning_rate": 4.09171479345539e-05, "loss": 0.3036, "step": 8711 }, { "epoch": 1.52, "learning_rate": 4.091165216060955e-05, "loss": 0.3034, "step": 8712 }, { "epoch": 1.52, "learning_rate": 4.0906155145247924e-05, "loss": 0.2969, "step": 8713 }, { "epoch": 1.52, "learning_rate": 4.090065688897786e-05, "loss": 0.3059, "step": 8714 }, { "epoch": 1.52, "learning_rate": 4.0895157392308266e-05, "loss": 0.3224, "step": 8715 }, { "epoch": 1.52, "learning_rate": 4.088965665574822e-05, "loss": 0.3009, "step": 8716 }, { "epoch": 1.52, "learning_rate": 4.0884154679806856e-05, "loss": 0.316, "step": 8717 }, { "epoch": 1.52, "learning_rate": 4.0878651464993466e-05, "loss": 0.2983, "step": 8718 }, { "epoch": 1.52, "learning_rate": 4.087314701181744e-05, "loss": 0.3128, "step": 8719 }, { "epoch": 1.52, "learning_rate": 4.0867641320788284e-05, "loss": 0.3176, "step": 8720 }, { "epoch": 1.52, "learning_rate": 4.0862134392415615e-05, "loss": 0.3226, "step": 8721 }, { "epoch": 1.52, "learning_rate": 4.085662622720917e-05, "loss": 0.3189, "step": 8722 }, { "epoch": 1.52, "learning_rate": 4.085111682567881e-05, "loss": 0.2972, "step": 8723 }, { "epoch": 1.52, "learning_rate": 4.084560618833448e-05, "loss": 0.2915, "step": 8724 }, { "epoch": 1.52, "learning_rate": 4.084009431568628e-05, "loss": 0.3186, "step": 8725 }, { "epoch": 1.52, "learning_rate": 4.0834581208244377e-05, "loss": 0.2942, "step": 8726 }, { "epoch": 1.52, "learning_rate": 4.082906686651911e-05, "loss": 0.32, "step": 8727 }, { "epoch": 1.52, "learning_rate": 4.082355129102087e-05, "loss": 0.3102, "step": 8728 }, { "epoch": 1.52, "learning_rate": 4.0818034482260215e-05, "loss": 0.3022, "step": 8729 }, { "epoch": 1.52, "learning_rate": 4.0812516440747787e-05, "loss": 0.3154, "step": 8730 }, { "epoch": 1.52, "learning_rate": 4.080699716699435e-05, "loss": 0.3146, "step": 8731 }, { "epoch": 1.52, "learning_rate": 4.080147666151078e-05, "loss": 0.3063, "step": 8732 }, { "epoch": 1.52, "learning_rate": 4.0795954924808076e-05, "loss": 0.3136, "step": 8733 }, { "epoch": 1.52, "learning_rate": 4.079043195739734e-05, "loss": 0.3048, "step": 8734 }, { "epoch": 1.52, "learning_rate": 4.078490775978979e-05, "loss": 0.3259, "step": 8735 }, { "epoch": 1.52, "learning_rate": 4.0779382332496775e-05, "loss": 0.3055, "step": 8736 }, { "epoch": 1.52, "learning_rate": 4.077385567602972e-05, "loss": 0.307, "step": 8737 }, { "epoch": 1.52, "learning_rate": 4.07683277909002e-05, "loss": 0.3293, "step": 8738 }, { "epoch": 1.52, "learning_rate": 4.076279867761988e-05, "loss": 0.3117, "step": 8739 }, { "epoch": 1.52, "learning_rate": 4.075726833670057e-05, "loss": 0.3084, "step": 8740 }, { "epoch": 1.52, "learning_rate": 4.0751736768654146e-05, "loss": 0.3041, "step": 8741 }, { "epoch": 1.52, "learning_rate": 4.074620397399265e-05, "loss": 0.3041, "step": 8742 }, { "epoch": 1.52, "learning_rate": 4.0740669953228195e-05, "loss": 0.3089, "step": 8743 }, { "epoch": 1.52, "learning_rate": 4.0735134706873025e-05, "loss": 0.3173, "step": 8744 }, { "epoch": 1.52, "learning_rate": 4.07295982354395e-05, "loss": 0.2948, "step": 8745 }, { "epoch": 1.52, "learning_rate": 4.07240605394401e-05, "loss": 0.3128, "step": 8746 }, { "epoch": 1.52, "learning_rate": 4.071852161938739e-05, "loss": 0.3028, "step": 8747 }, { "epoch": 1.52, "learning_rate": 4.071298147579408e-05, "loss": 0.3029, "step": 8748 }, { "epoch": 1.52, "learning_rate": 4.070744010917297e-05, "loss": 0.3118, "step": 8749 }, { "epoch": 1.53, "learning_rate": 4.0701897520037e-05, "loss": 0.3306, "step": 8750 }, { "epoch": 1.53, "learning_rate": 4.069635370889919e-05, "loss": 0.3178, "step": 8751 }, { "epoch": 1.53, "learning_rate": 4.069080867627269e-05, "loss": 0.3215, "step": 8752 }, { "epoch": 1.53, "learning_rate": 4.068526242267078e-05, "loss": 0.3046, "step": 8753 }, { "epoch": 1.53, "learning_rate": 4.0679714948606815e-05, "loss": 0.3009, "step": 8754 }, { "epoch": 1.53, "learning_rate": 4.067416625459429e-05, "loss": 0.3213, "step": 8755 }, { "epoch": 1.53, "learning_rate": 4.066861634114681e-05, "loss": 0.3139, "step": 8756 }, { "epoch": 1.53, "learning_rate": 4.066306520877809e-05, "loss": 0.3165, "step": 8757 }, { "epoch": 1.53, "learning_rate": 4.065751285800196e-05, "loss": 0.3147, "step": 8758 }, { "epoch": 1.53, "learning_rate": 4.065195928933235e-05, "loss": 0.3059, "step": 8759 }, { "epoch": 1.53, "learning_rate": 4.0646404503283317e-05, "loss": 0.299, "step": 8760 }, { "epoch": 1.53, "learning_rate": 4.064084850036902e-05, "loss": 0.3133, "step": 8761 }, { "epoch": 1.53, "learning_rate": 4.063529128110376e-05, "loss": 0.3195, "step": 8762 }, { "epoch": 1.53, "learning_rate": 4.0629732846001894e-05, "loss": 0.3055, "step": 8763 }, { "epoch": 1.53, "learning_rate": 4.062417319557796e-05, "loss": 0.3188, "step": 8764 }, { "epoch": 1.53, "learning_rate": 4.0618612330346546e-05, "loss": 0.3033, "step": 8765 }, { "epoch": 1.53, "learning_rate": 4.061305025082239e-05, "loss": 0.2952, "step": 8766 }, { "epoch": 1.53, "learning_rate": 4.060748695752034e-05, "loss": 0.3154, "step": 8767 }, { "epoch": 1.53, "learning_rate": 4.0601922450955335e-05, "loss": 0.2994, "step": 8768 }, { "epoch": 1.53, "learning_rate": 4.059635673164245e-05, "loss": 0.311, "step": 8769 }, { "epoch": 1.53, "learning_rate": 4.0590789800096854e-05, "loss": 0.312, "step": 8770 }, { "epoch": 1.53, "learning_rate": 4.058522165683385e-05, "loss": 0.3143, "step": 8771 }, { "epoch": 1.53, "learning_rate": 4.057965230236883e-05, "loss": 0.3139, "step": 8772 }, { "epoch": 1.53, "learning_rate": 4.057408173721731e-05, "loss": 0.3094, "step": 8773 }, { "epoch": 1.53, "learning_rate": 4.056850996189491e-05, "loss": 0.2869, "step": 8774 }, { "epoch": 1.53, "learning_rate": 4.056293697691739e-05, "loss": 0.315, "step": 8775 }, { "epoch": 1.53, "learning_rate": 4.0557362782800564e-05, "loss": 0.3081, "step": 8776 }, { "epoch": 1.53, "learning_rate": 4.055178738006042e-05, "loss": 0.3196, "step": 8777 }, { "epoch": 1.53, "learning_rate": 4.054621076921303e-05, "loss": 0.3105, "step": 8778 }, { "epoch": 1.53, "learning_rate": 4.054063295077457e-05, "loss": 0.3176, "step": 8779 }, { "epoch": 1.53, "learning_rate": 4.053505392526134e-05, "loss": 0.3304, "step": 8780 }, { "epoch": 1.53, "learning_rate": 4.052947369318975e-05, "loss": 0.3005, "step": 8781 }, { "epoch": 1.53, "learning_rate": 4.0523892255076327e-05, "loss": 0.3151, "step": 8782 }, { "epoch": 1.53, "learning_rate": 4.051830961143769e-05, "loss": 0.3089, "step": 8783 }, { "epoch": 1.53, "learning_rate": 4.0512725762790605e-05, "loss": 0.3084, "step": 8784 }, { "epoch": 1.53, "learning_rate": 4.050714070965191e-05, "loss": 0.3003, "step": 8785 }, { "epoch": 1.53, "learning_rate": 4.0501554452538566e-05, "loss": 0.3131, "step": 8786 }, { "epoch": 1.53, "learning_rate": 4.0495966991967664e-05, "loss": 0.3013, "step": 8787 }, { "epoch": 1.53, "learning_rate": 4.049037832845639e-05, "loss": 0.3216, "step": 8788 }, { "epoch": 1.53, "learning_rate": 4.048478846252205e-05, "loss": 0.3014, "step": 8789 }, { "epoch": 1.53, "learning_rate": 4.047919739468206e-05, "loss": 0.303, "step": 8790 }, { "epoch": 1.53, "learning_rate": 4.047360512545392e-05, "loss": 0.3168, "step": 8791 }, { "epoch": 1.53, "learning_rate": 4.046801165535529e-05, "loss": 0.3176, "step": 8792 }, { "epoch": 1.53, "learning_rate": 4.046241698490391e-05, "loss": 0.3141, "step": 8793 }, { "epoch": 1.53, "learning_rate": 4.0456821114617625e-05, "loss": 0.3301, "step": 8794 }, { "epoch": 1.53, "learning_rate": 4.045122404501442e-05, "loss": 0.3087, "step": 8795 }, { "epoch": 1.53, "learning_rate": 4.044562577661237e-05, "loss": 0.3072, "step": 8796 }, { "epoch": 1.53, "learning_rate": 4.044002630992966e-05, "loss": 0.3032, "step": 8797 }, { "epoch": 1.53, "learning_rate": 4.0434425645484583e-05, "loss": 0.3102, "step": 8798 }, { "epoch": 1.53, "learning_rate": 4.042882378379557e-05, "loss": 0.3036, "step": 8799 }, { "epoch": 1.53, "learning_rate": 4.042322072538113e-05, "loss": 0.2876, "step": 8800 }, { "epoch": 1.53, "learning_rate": 4.0417616470759914e-05, "loss": 0.3174, "step": 8801 }, { "epoch": 1.53, "learning_rate": 4.0412011020450644e-05, "loss": 0.3209, "step": 8802 }, { "epoch": 1.53, "learning_rate": 4.040640437497218e-05, "loss": 0.3128, "step": 8803 }, { "epoch": 1.53, "learning_rate": 4.04007965348435e-05, "loss": 0.3133, "step": 8804 }, { "epoch": 1.53, "learning_rate": 4.039518750058368e-05, "loss": 0.3135, "step": 8805 }, { "epoch": 1.53, "learning_rate": 4.038957727271189e-05, "loss": 0.3092, "step": 8806 }, { "epoch": 1.54, "learning_rate": 4.038396585174743e-05, "loss": 0.3097, "step": 8807 }, { "epoch": 1.54, "learning_rate": 4.037835323820972e-05, "loss": 0.3131, "step": 8808 }, { "epoch": 1.54, "learning_rate": 4.0372739432618284e-05, "loss": 0.2937, "step": 8809 }, { "epoch": 1.54, "learning_rate": 4.036712443549273e-05, "loss": 0.3153, "step": 8810 }, { "epoch": 1.54, "learning_rate": 4.03615082473528e-05, "loss": 0.3006, "step": 8811 }, { "epoch": 1.54, "learning_rate": 4.0355890868718356e-05, "loss": 0.318, "step": 8812 }, { "epoch": 1.54, "learning_rate": 4.0350272300109346e-05, "loss": 0.3178, "step": 8813 }, { "epoch": 1.54, "learning_rate": 4.034465254204585e-05, "loss": 0.3095, "step": 8814 }, { "epoch": 1.54, "learning_rate": 4.0339031595048024e-05, "loss": 0.3059, "step": 8815 }, { "epoch": 1.54, "learning_rate": 4.033340945963619e-05, "loss": 0.3014, "step": 8816 }, { "epoch": 1.54, "learning_rate": 4.032778613633073e-05, "loss": 0.3143, "step": 8817 }, { "epoch": 1.54, "learning_rate": 4.032216162565215e-05, "loss": 0.298, "step": 8818 }, { "epoch": 1.54, "learning_rate": 4.0316535928121066e-05, "loss": 0.3132, "step": 8819 }, { "epoch": 1.54, "learning_rate": 4.031090904425823e-05, "loss": 0.3246, "step": 8820 }, { "epoch": 1.54, "learning_rate": 4.030528097458446e-05, "loss": 0.3151, "step": 8821 }, { "epoch": 1.54, "learning_rate": 4.029965171962071e-05, "loss": 0.3027, "step": 8822 }, { "epoch": 1.54, "learning_rate": 4.029402127988804e-05, "loss": 0.3008, "step": 8823 }, { "epoch": 1.54, "learning_rate": 4.028838965590761e-05, "loss": 0.2916, "step": 8824 }, { "epoch": 1.54, "learning_rate": 4.028275684820071e-05, "loss": 0.3348, "step": 8825 }, { "epoch": 1.54, "learning_rate": 4.027712285728872e-05, "loss": 0.3153, "step": 8826 }, { "epoch": 1.54, "learning_rate": 4.027148768369314e-05, "loss": 0.3105, "step": 8827 }, { "epoch": 1.54, "learning_rate": 4.0265851327935585e-05, "loss": 0.3148, "step": 8828 }, { "epoch": 1.54, "learning_rate": 4.026021379053775e-05, "loss": 0.3185, "step": 8829 }, { "epoch": 1.54, "learning_rate": 4.025457507202148e-05, "loss": 0.3077, "step": 8830 }, { "epoch": 1.54, "learning_rate": 4.024893517290869e-05, "loss": 0.3209, "step": 8831 }, { "epoch": 1.54, "learning_rate": 4.0243294093721436e-05, "loss": 0.3122, "step": 8832 }, { "epoch": 1.54, "learning_rate": 4.0237651834981866e-05, "loss": 0.3243, "step": 8833 }, { "epoch": 1.54, "learning_rate": 4.023200839721226e-05, "loss": 0.3232, "step": 8834 }, { "epoch": 1.54, "learning_rate": 4.022636378093496e-05, "loss": 0.3142, "step": 8835 }, { "epoch": 1.54, "learning_rate": 4.022071798667246e-05, "loss": 0.309, "step": 8836 }, { "epoch": 1.54, "learning_rate": 4.021507101494735e-05, "loss": 0.2979, "step": 8837 }, { "epoch": 1.54, "learning_rate": 4.020942286628233e-05, "loss": 0.3361, "step": 8838 }, { "epoch": 1.54, "learning_rate": 4.0203773541200205e-05, "loss": 0.3083, "step": 8839 }, { "epoch": 1.54, "learning_rate": 4.0198123040223894e-05, "loss": 0.3139, "step": 8840 }, { "epoch": 1.54, "learning_rate": 4.0192471363876415e-05, "loss": 0.3121, "step": 8841 }, { "epoch": 1.54, "learning_rate": 4.018681851268091e-05, "loss": 0.3139, "step": 8842 }, { "epoch": 1.54, "learning_rate": 4.018116448716062e-05, "loss": 0.3158, "step": 8843 }, { "epoch": 1.54, "learning_rate": 4.01755092878389e-05, "loss": 0.3181, "step": 8844 }, { "epoch": 1.54, "learning_rate": 4.016985291523919e-05, "loss": 0.2933, "step": 8845 }, { "epoch": 1.54, "learning_rate": 4.0164195369885075e-05, "loss": 0.3132, "step": 8846 }, { "epoch": 1.54, "learning_rate": 4.015853665230024e-05, "loss": 0.3235, "step": 8847 }, { "epoch": 1.54, "learning_rate": 4.015287676300845e-05, "loss": 0.3074, "step": 8848 }, { "epoch": 1.54, "learning_rate": 4.0147215702533614e-05, "loss": 0.3171, "step": 8849 }, { "epoch": 1.54, "learning_rate": 4.014155347139972e-05, "loss": 0.3086, "step": 8850 }, { "epoch": 1.54, "learning_rate": 4.01358900701309e-05, "loss": 0.3091, "step": 8851 }, { "epoch": 1.54, "learning_rate": 4.013022549925136e-05, "loss": 0.2991, "step": 8852 }, { "epoch": 1.54, "learning_rate": 4.012455975928542e-05, "loss": 0.3003, "step": 8853 }, { "epoch": 1.54, "learning_rate": 4.011889285075754e-05, "loss": 0.3059, "step": 8854 }, { "epoch": 1.54, "learning_rate": 4.011322477419224e-05, "loss": 0.3164, "step": 8855 }, { "epoch": 1.54, "learning_rate": 4.010755553011418e-05, "loss": 0.3077, "step": 8856 }, { "epoch": 1.54, "learning_rate": 4.010188511904812e-05, "loss": 0.3209, "step": 8857 }, { "epoch": 1.54, "learning_rate": 4.0096213541518924e-05, "loss": 0.3157, "step": 8858 }, { "epoch": 1.54, "learning_rate": 4.0090540798051574e-05, "loss": 0.3127, "step": 8859 }, { "epoch": 1.54, "learning_rate": 4.0084866889171156e-05, "loss": 0.3049, "step": 8860 }, { "epoch": 1.54, "learning_rate": 4.007919181540284e-05, "loss": 0.3088, "step": 8861 }, { "epoch": 1.54, "learning_rate": 4.0073515577271964e-05, "loss": 0.3015, "step": 8862 }, { "epoch": 1.54, "learning_rate": 4.006783817530391e-05, "loss": 0.3067, "step": 8863 }, { "epoch": 1.54, "learning_rate": 4.006215961002419e-05, "loss": 0.3092, "step": 8864 }, { "epoch": 1.55, "learning_rate": 4.005647988195844e-05, "loss": 0.295, "step": 8865 }, { "epoch": 1.55, "learning_rate": 4.005079899163238e-05, "loss": 0.2974, "step": 8866 }, { "epoch": 1.55, "learning_rate": 4.004511693957186e-05, "loss": 0.3079, "step": 8867 }, { "epoch": 1.55, "learning_rate": 4.0039433726302805e-05, "loss": 0.3146, "step": 8868 }, { "epoch": 1.55, "learning_rate": 4.003374935235128e-05, "loss": 0.3122, "step": 8869 }, { "epoch": 1.55, "learning_rate": 4.002806381824346e-05, "loss": 0.3265, "step": 8870 }, { "epoch": 1.55, "learning_rate": 4.002237712450559e-05, "loss": 0.318, "step": 8871 }, { "epoch": 1.55, "learning_rate": 4.001668927166406e-05, "loss": 0.288, "step": 8872 }, { "epoch": 1.55, "learning_rate": 4.001100026024534e-05, "loss": 0.3031, "step": 8873 }, { "epoch": 1.55, "learning_rate": 4.000531009077603e-05, "loss": 0.3076, "step": 8874 }, { "epoch": 1.55, "learning_rate": 3.9999618763782814e-05, "loss": 0.3254, "step": 8875 }, { "epoch": 1.55, "learning_rate": 3.9993926279792514e-05, "loss": 0.3144, "step": 8876 }, { "epoch": 1.55, "learning_rate": 3.998823263933203e-05, "loss": 0.3079, "step": 8877 }, { "epoch": 1.55, "learning_rate": 3.998253784292838e-05, "loss": 0.3142, "step": 8878 }, { "epoch": 1.55, "learning_rate": 3.9976841891108685e-05, "loss": 0.2964, "step": 8879 }, { "epoch": 1.55, "learning_rate": 3.997114478440019e-05, "loss": 0.2967, "step": 8880 }, { "epoch": 1.55, "learning_rate": 3.996544652333023e-05, "loss": 0.2957, "step": 8881 }, { "epoch": 1.55, "learning_rate": 3.9959747108426246e-05, "loss": 0.3021, "step": 8882 }, { "epoch": 1.55, "learning_rate": 3.995404654021579e-05, "loss": 0.3086, "step": 8883 }, { "epoch": 1.55, "learning_rate": 3.994834481922652e-05, "loss": 0.3227, "step": 8884 }, { "epoch": 1.55, "learning_rate": 3.99426419459862e-05, "loss": 0.3089, "step": 8885 }, { "epoch": 1.55, "learning_rate": 3.993693792102273e-05, "loss": 0.3351, "step": 8886 }, { "epoch": 1.55, "learning_rate": 3.993123274486405e-05, "loss": 0.3202, "step": 8887 }, { "epoch": 1.55, "learning_rate": 3.992552641803827e-05, "loss": 0.2997, "step": 8888 }, { "epoch": 1.55, "learning_rate": 3.991981894107358e-05, "loss": 0.316, "step": 8889 }, { "epoch": 1.55, "learning_rate": 3.991411031449827e-05, "loss": 0.317, "step": 8890 }, { "epoch": 1.55, "learning_rate": 3.9908400538840754e-05, "loss": 0.3088, "step": 8891 }, { "epoch": 1.55, "learning_rate": 3.990268961462953e-05, "loss": 0.3155, "step": 8892 }, { "epoch": 1.55, "learning_rate": 3.9896977542393236e-05, "loss": 0.3059, "step": 8893 }, { "epoch": 1.55, "learning_rate": 3.989126432266058e-05, "loss": 0.3186, "step": 8894 }, { "epoch": 1.55, "learning_rate": 3.98855499559604e-05, "loss": 0.3141, "step": 8895 }, { "epoch": 1.55, "learning_rate": 3.987983444282163e-05, "loss": 0.317, "step": 8896 }, { "epoch": 1.55, "learning_rate": 3.987411778377332e-05, "loss": 0.3117, "step": 8897 }, { "epoch": 1.55, "learning_rate": 3.98683999793446e-05, "loss": 0.3131, "step": 8898 }, { "epoch": 1.55, "learning_rate": 3.9862681030064745e-05, "loss": 0.3026, "step": 8899 }, { "epoch": 1.55, "learning_rate": 3.985696093646312e-05, "loss": 0.2974, "step": 8900 }, { "epoch": 1.55, "learning_rate": 3.985123969906916e-05, "loss": 0.3025, "step": 8901 }, { "epoch": 1.55, "learning_rate": 3.9845517318412454e-05, "loss": 0.3053, "step": 8902 }, { "epoch": 1.55, "learning_rate": 3.9839793795022704e-05, "loss": 0.301, "step": 8903 }, { "epoch": 1.55, "learning_rate": 3.983406912942966e-05, "loss": 0.3201, "step": 8904 }, { "epoch": 1.55, "learning_rate": 3.982834332216323e-05, "loss": 0.3067, "step": 8905 }, { "epoch": 1.55, "learning_rate": 3.982261637375341e-05, "loss": 0.3279, "step": 8906 }, { "epoch": 1.55, "learning_rate": 3.981688828473028e-05, "loss": 0.3003, "step": 8907 }, { "epoch": 1.55, "learning_rate": 3.981115905562408e-05, "loss": 0.304, "step": 8908 }, { "epoch": 1.55, "learning_rate": 3.9805428686965106e-05, "loss": 0.314, "step": 8909 }, { "epoch": 1.55, "learning_rate": 3.979969717928376e-05, "loss": 0.3138, "step": 8910 }, { "epoch": 1.55, "learning_rate": 3.979396453311059e-05, "loss": 0.3111, "step": 8911 }, { "epoch": 1.55, "learning_rate": 3.9788230748976214e-05, "loss": 0.3074, "step": 8912 }, { "epoch": 1.55, "learning_rate": 3.978249582741136e-05, "loss": 0.3047, "step": 8913 }, { "epoch": 1.55, "learning_rate": 3.9776759768946894e-05, "loss": 0.3109, "step": 8914 }, { "epoch": 1.55, "learning_rate": 3.9771022574113723e-05, "loss": 0.3131, "step": 8915 }, { "epoch": 1.55, "learning_rate": 3.976528424344292e-05, "loss": 0.3145, "step": 8916 }, { "epoch": 1.55, "learning_rate": 3.975954477746563e-05, "loss": 0.3139, "step": 8917 }, { "epoch": 1.55, "learning_rate": 3.975380417671312e-05, "loss": 0.3135, "step": 8918 }, { "epoch": 1.55, "learning_rate": 3.974806244171675e-05, "loss": 0.3082, "step": 8919 }, { "epoch": 1.55, "learning_rate": 3.974231957300799e-05, "loss": 0.295, "step": 8920 }, { "epoch": 1.55, "learning_rate": 3.973657557111842e-05, "loss": 0.3092, "step": 8921 }, { "epoch": 1.56, "learning_rate": 3.973083043657971e-05, "loss": 0.3146, "step": 8922 }, { "epoch": 1.56, "learning_rate": 3.972508416992365e-05, "loss": 0.3081, "step": 8923 }, { "epoch": 1.56, "learning_rate": 3.971933677168214e-05, "loss": 0.3094, "step": 8924 }, { "epoch": 1.56, "learning_rate": 3.9713588242387164e-05, "loss": 0.3168, "step": 8925 }, { "epoch": 1.56, "learning_rate": 3.970783858257081e-05, "loss": 0.3145, "step": 8926 }, { "epoch": 1.56, "learning_rate": 3.97020877927653e-05, "loss": 0.3029, "step": 8927 }, { "epoch": 1.56, "learning_rate": 3.969633587350293e-05, "loss": 0.3001, "step": 8928 }, { "epoch": 1.56, "learning_rate": 3.969058282531611e-05, "loss": 0.3187, "step": 8929 }, { "epoch": 1.56, "learning_rate": 3.9684828648737384e-05, "loss": 0.3107, "step": 8930 }, { "epoch": 1.56, "learning_rate": 3.967907334429934e-05, "loss": 0.3002, "step": 8931 }, { "epoch": 1.56, "learning_rate": 3.967331691253471e-05, "loss": 0.308, "step": 8932 }, { "epoch": 1.56, "learning_rate": 3.966755935397634e-05, "loss": 0.305, "step": 8933 }, { "epoch": 1.56, "learning_rate": 3.9661800669157154e-05, "loss": 0.2995, "step": 8934 }, { "epoch": 1.56, "learning_rate": 3.96560408586102e-05, "loss": 0.3075, "step": 8935 }, { "epoch": 1.56, "learning_rate": 3.96502799228686e-05, "loss": 0.3164, "step": 8936 }, { "epoch": 1.56, "learning_rate": 3.9644517862465626e-05, "loss": 0.2952, "step": 8937 }, { "epoch": 1.56, "learning_rate": 3.963875467793461e-05, "loss": 0.3053, "step": 8938 }, { "epoch": 1.56, "learning_rate": 3.963299036980901e-05, "loss": 0.3115, "step": 8939 }, { "epoch": 1.56, "learning_rate": 3.96272249386224e-05, "loss": 0.307, "step": 8940 }, { "epoch": 1.56, "learning_rate": 3.9621458384908446e-05, "loss": 0.3045, "step": 8941 }, { "epoch": 1.56, "learning_rate": 3.961569070920088e-05, "loss": 0.3022, "step": 8942 }, { "epoch": 1.56, "learning_rate": 3.960992191203361e-05, "loss": 0.2929, "step": 8943 }, { "epoch": 1.56, "learning_rate": 3.9604151993940595e-05, "loss": 0.3298, "step": 8944 }, { "epoch": 1.56, "learning_rate": 3.9598380955455904e-05, "loss": 0.2961, "step": 8945 }, { "epoch": 1.56, "learning_rate": 3.9592608797113744e-05, "loss": 0.3137, "step": 8946 }, { "epoch": 1.56, "learning_rate": 3.958683551944838e-05, "loss": 0.3251, "step": 8947 }, { "epoch": 1.56, "learning_rate": 3.958106112299422e-05, "loss": 0.315, "step": 8948 }, { "epoch": 1.56, "learning_rate": 3.957528560828573e-05, "loss": 0.3172, "step": 8949 }, { "epoch": 1.56, "learning_rate": 3.9569508975857534e-05, "loss": 0.3146, "step": 8950 }, { "epoch": 1.56, "learning_rate": 3.9563731226244316e-05, "loss": 0.3134, "step": 8951 }, { "epoch": 1.56, "learning_rate": 3.955795235998088e-05, "loss": 0.3295, "step": 8952 }, { "epoch": 1.56, "learning_rate": 3.9552172377602145e-05, "loss": 0.315, "step": 8953 }, { "epoch": 1.56, "learning_rate": 3.9546391279643116e-05, "loss": 0.2962, "step": 8954 }, { "epoch": 1.56, "learning_rate": 3.95406090666389e-05, "loss": 0.3035, "step": 8955 }, { "epoch": 1.56, "learning_rate": 3.953482573912472e-05, "loss": 0.3121, "step": 8956 }, { "epoch": 1.56, "learning_rate": 3.9529041297635884e-05, "loss": 0.3169, "step": 8957 }, { "epoch": 1.56, "learning_rate": 3.9523255742707835e-05, "loss": 0.3162, "step": 8958 }, { "epoch": 1.56, "learning_rate": 3.951746907487607e-05, "loss": 0.3019, "step": 8959 }, { "epoch": 1.56, "learning_rate": 3.9511681294676255e-05, "loss": 0.296, "step": 8960 }, { "epoch": 1.56, "learning_rate": 3.950589240264409e-05, "loss": 0.31, "step": 8961 }, { "epoch": 1.56, "learning_rate": 3.950010239931543e-05, "loss": 0.303, "step": 8962 }, { "epoch": 1.56, "learning_rate": 3.9494311285226204e-05, "loss": 0.3098, "step": 8963 }, { "epoch": 1.56, "learning_rate": 3.9488519060912455e-05, "loss": 0.3105, "step": 8964 }, { "epoch": 1.56, "learning_rate": 3.948272572691032e-05, "loss": 0.3052, "step": 8965 }, { "epoch": 1.56, "learning_rate": 3.947693128375605e-05, "loss": 0.2991, "step": 8966 }, { "epoch": 1.56, "learning_rate": 3.9471135731985996e-05, "loss": 0.3223, "step": 8967 }, { "epoch": 1.56, "learning_rate": 3.9465339072136615e-05, "loss": 0.2999, "step": 8968 }, { "epoch": 1.56, "learning_rate": 3.945954130474444e-05, "loss": 0.3106, "step": 8969 }, { "epoch": 1.56, "learning_rate": 3.9453742430346154e-05, "loss": 0.3002, "step": 8970 }, { "epoch": 1.56, "learning_rate": 3.9447942449478495e-05, "loss": 0.2996, "step": 8971 }, { "epoch": 1.56, "learning_rate": 3.9442141362678324e-05, "loss": 0.3206, "step": 8972 }, { "epoch": 1.56, "learning_rate": 3.943633917048262e-05, "loss": 0.2972, "step": 8973 }, { "epoch": 1.56, "learning_rate": 3.943053587342844e-05, "loss": 0.3119, "step": 8974 }, { "epoch": 1.56, "learning_rate": 3.942473147205295e-05, "loss": 0.3182, "step": 8975 }, { "epoch": 1.56, "learning_rate": 3.941892596689343e-05, "loss": 0.3009, "step": 8976 }, { "epoch": 1.56, "learning_rate": 3.941311935848724e-05, "loss": 0.3135, "step": 8977 }, { "epoch": 1.56, "learning_rate": 3.940731164737187e-05, "loss": 0.3033, "step": 8978 }, { "epoch": 1.57, "learning_rate": 3.940150283408488e-05, "loss": 0.3205, "step": 8979 }, { "epoch": 1.57, "learning_rate": 3.939569291916396e-05, "loss": 0.3023, "step": 8980 }, { "epoch": 1.57, "learning_rate": 3.938988190314688e-05, "loss": 0.3111, "step": 8981 }, { "epoch": 1.57, "learning_rate": 3.9384069786571535e-05, "loss": 0.3072, "step": 8982 }, { "epoch": 1.57, "learning_rate": 3.93782565699759e-05, "loss": 0.3029, "step": 8983 }, { "epoch": 1.57, "learning_rate": 3.9372442253898076e-05, "loss": 0.2996, "step": 8984 }, { "epoch": 1.57, "learning_rate": 3.936662683887624e-05, "loss": 0.3173, "step": 8985 }, { "epoch": 1.57, "learning_rate": 3.936081032544867e-05, "loss": 0.3123, "step": 8986 }, { "epoch": 1.57, "learning_rate": 3.935499271415378e-05, "loss": 0.3095, "step": 8987 }, { "epoch": 1.57, "learning_rate": 3.9349174005530056e-05, "loss": 0.299, "step": 8988 }, { "epoch": 1.57, "learning_rate": 3.934335420011609e-05, "loss": 0.31, "step": 8989 }, { "epoch": 1.57, "learning_rate": 3.9337533298450566e-05, "loss": 0.3077, "step": 8990 }, { "epoch": 1.57, "learning_rate": 3.933171130107231e-05, "loss": 0.3224, "step": 8991 }, { "epoch": 1.57, "learning_rate": 3.93258882085202e-05, "loss": 0.3057, "step": 8992 }, { "epoch": 1.57, "learning_rate": 3.932006402133324e-05, "loss": 0.299, "step": 8993 }, { "epoch": 1.57, "learning_rate": 3.931423874005054e-05, "loss": 0.2961, "step": 8994 }, { "epoch": 1.57, "learning_rate": 3.93084123652113e-05, "loss": 0.2901, "step": 8995 }, { "epoch": 1.57, "learning_rate": 3.9302584897354804e-05, "loss": 0.2887, "step": 8996 }, { "epoch": 1.57, "learning_rate": 3.9296756337020484e-05, "loss": 0.3187, "step": 8997 }, { "epoch": 1.57, "learning_rate": 3.929092668474784e-05, "loss": 0.302, "step": 8998 }, { "epoch": 1.57, "learning_rate": 3.9285095941076475e-05, "loss": 0.3036, "step": 8999 }, { "epoch": 1.57, "learning_rate": 3.92792641065461e-05, "loss": 0.293, "step": 9000 } ], "logging_steps": 1.0, "max_steps": 11474, "num_train_epochs": 2, "save_steps": 500, "total_flos": 1.4597812314184666e+20, "trial_name": null, "trial_params": null }