{ "best_metric": null, "best_model_checkpoint": null, "epoch": 3.0, "global_step": 1737, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 3.773584905660378e-07, "loss": 1.8047, "step": 1 }, { "epoch": 0.0, "learning_rate": 7.547169811320755e-07, "loss": 1.8027, "step": 2 }, { "epoch": 0.01, "learning_rate": 1.1320754716981133e-06, "loss": 1.6191, "step": 3 }, { "epoch": 0.01, "learning_rate": 1.509433962264151e-06, "loss": 1.3994, "step": 4 }, { "epoch": 0.01, "learning_rate": 1.8867924528301889e-06, "loss": 1.1846, "step": 5 }, { "epoch": 0.01, "learning_rate": 2.2641509433962266e-06, "loss": 0.8877, "step": 6 }, { "epoch": 0.01, "learning_rate": 2.6415094339622644e-06, "loss": 0.6777, "step": 7 }, { "epoch": 0.01, "learning_rate": 3.018867924528302e-06, "loss": 0.4395, "step": 8 }, { "epoch": 0.02, "learning_rate": 3.3962264150943395e-06, "loss": 0.2571, "step": 9 }, { "epoch": 0.02, "learning_rate": 3.7735849056603777e-06, "loss": 0.147, "step": 10 }, { "epoch": 0.02, "learning_rate": 4.150943396226416e-06, "loss": 0.0666, "step": 11 }, { "epoch": 0.02, "learning_rate": 4.528301886792453e-06, "loss": 0.085, "step": 12 }, { "epoch": 0.02, "learning_rate": 4.905660377358491e-06, "loss": 0.0947, "step": 13 }, { "epoch": 0.02, "learning_rate": 5.283018867924529e-06, "loss": 0.0614, "step": 14 }, { "epoch": 0.03, "learning_rate": 5.660377358490566e-06, "loss": 0.0506, "step": 15 }, { "epoch": 0.03, "learning_rate": 6.037735849056604e-06, "loss": 0.0551, "step": 16 }, { "epoch": 0.03, "learning_rate": 6.415094339622642e-06, "loss": 0.0812, "step": 17 }, { "epoch": 0.03, "learning_rate": 6.792452830188679e-06, "loss": 0.0601, "step": 18 }, { "epoch": 0.03, "learning_rate": 7.169811320754717e-06, "loss": 0.0652, "step": 19 }, { "epoch": 0.03, "learning_rate": 7.5471698113207555e-06, "loss": 0.0608, "step": 20 }, { "epoch": 0.04, "learning_rate": 7.924528301886793e-06, "loss": 0.0514, "step": 21 }, { "epoch": 0.04, "learning_rate": 8.301886792452832e-06, "loss": 0.0624, "step": 22 }, { "epoch": 0.04, "learning_rate": 8.67924528301887e-06, "loss": 0.0635, "step": 23 }, { "epoch": 0.04, "learning_rate": 9.056603773584907e-06, "loss": 0.0715, "step": 24 }, { "epoch": 0.04, "learning_rate": 9.433962264150944e-06, "loss": 0.0473, "step": 25 }, { "epoch": 0.04, "learning_rate": 9.811320754716981e-06, "loss": 0.0524, "step": 26 }, { "epoch": 0.05, "learning_rate": 1.018867924528302e-05, "loss": 0.0704, "step": 27 }, { "epoch": 0.05, "learning_rate": 1.0566037735849058e-05, "loss": 0.0639, "step": 28 }, { "epoch": 0.05, "learning_rate": 1.0943396226415095e-05, "loss": 0.0646, "step": 29 }, { "epoch": 0.05, "learning_rate": 1.1320754716981132e-05, "loss": 0.0582, "step": 30 }, { "epoch": 0.05, "learning_rate": 1.169811320754717e-05, "loss": 0.0527, "step": 31 }, { "epoch": 0.06, "learning_rate": 1.2075471698113209e-05, "loss": 0.0617, "step": 32 }, { "epoch": 0.06, "learning_rate": 1.2452830188679246e-05, "loss": 0.0419, "step": 33 }, { "epoch": 0.06, "learning_rate": 1.2830188679245283e-05, "loss": 0.0462, "step": 34 }, { "epoch": 0.06, "learning_rate": 1.320754716981132e-05, "loss": 0.0471, "step": 35 }, { "epoch": 0.06, "learning_rate": 1.3584905660377358e-05, "loss": 0.0386, "step": 36 }, { "epoch": 0.06, "learning_rate": 1.3962264150943397e-05, "loss": 0.0608, "step": 37 }, { "epoch": 0.07, "learning_rate": 1.4339622641509435e-05, "loss": 0.0345, "step": 38 }, { "epoch": 0.07, "learning_rate": 1.4716981132075472e-05, "loss": 0.0569, "step": 39 }, { "epoch": 0.07, "learning_rate": 1.5094339622641511e-05, "loss": 0.0497, "step": 40 }, { "epoch": 0.07, "learning_rate": 1.547169811320755e-05, "loss": 0.047, "step": 41 }, { "epoch": 0.07, "learning_rate": 1.5849056603773586e-05, "loss": 0.0479, "step": 42 }, { "epoch": 0.07, "learning_rate": 1.6226415094339625e-05, "loss": 0.0486, "step": 43 }, { "epoch": 0.08, "learning_rate": 1.6603773584905664e-05, "loss": 0.0484, "step": 44 }, { "epoch": 0.08, "learning_rate": 1.69811320754717e-05, "loss": 0.0376, "step": 45 }, { "epoch": 0.08, "learning_rate": 1.735849056603774e-05, "loss": 0.0412, "step": 46 }, { "epoch": 0.08, "learning_rate": 1.7735849056603774e-05, "loss": 0.0678, "step": 47 }, { "epoch": 0.08, "learning_rate": 1.8113207547169813e-05, "loss": 0.0486, "step": 48 }, { "epoch": 0.08, "learning_rate": 1.8490566037735852e-05, "loss": 0.054, "step": 49 }, { "epoch": 0.09, "learning_rate": 1.8867924528301888e-05, "loss": 0.052, "step": 50 }, { "epoch": 0.09, "learning_rate": 1.9245283018867927e-05, "loss": 0.0367, "step": 51 }, { "epoch": 0.09, "learning_rate": 1.9622641509433963e-05, "loss": 0.0601, "step": 52 }, { "epoch": 0.09, "learning_rate": 2e-05, "loss": 0.048, "step": 53 }, { "epoch": 0.09, "learning_rate": 1.9999982598549543e-05, "loss": 0.0429, "step": 54 }, { "epoch": 0.09, "learning_rate": 1.999993039425873e-05, "loss": 0.0519, "step": 55 }, { "epoch": 0.1, "learning_rate": 1.9999843387309243e-05, "loss": 0.0463, "step": 56 }, { "epoch": 0.1, "learning_rate": 1.9999721578003894e-05, "loss": 0.0666, "step": 57 }, { "epoch": 0.1, "learning_rate": 1.999956496676662e-05, "loss": 0.0352, "step": 58 }, { "epoch": 0.1, "learning_rate": 1.9999373554142466e-05, "loss": 0.064, "step": 59 }, { "epoch": 0.1, "learning_rate": 1.9999147340797613e-05, "loss": 0.0644, "step": 60 }, { "epoch": 0.11, "learning_rate": 1.9998886327519337e-05, "loss": 0.0517, "step": 61 }, { "epoch": 0.11, "learning_rate": 1.9998590515216048e-05, "loss": 0.0467, "step": 62 }, { "epoch": 0.11, "learning_rate": 1.9998259904917257e-05, "loss": 0.0574, "step": 63 }, { "epoch": 0.11, "learning_rate": 1.9997894497773586e-05, "loss": 0.049, "step": 64 }, { "epoch": 0.11, "learning_rate": 1.999749429505675e-05, "loss": 0.0482, "step": 65 }, { "epoch": 0.11, "learning_rate": 1.999705929815958e-05, "loss": 0.0441, "step": 66 }, { "epoch": 0.12, "learning_rate": 1.9996589508595984e-05, "loss": 0.0487, "step": 67 }, { "epoch": 0.12, "learning_rate": 1.999608492800097e-05, "loss": 0.0519, "step": 68 }, { "epoch": 0.12, "learning_rate": 1.9995545558130624e-05, "loss": 0.0455, "step": 69 }, { "epoch": 0.12, "learning_rate": 1.999497140086211e-05, "loss": 0.0516, "step": 70 }, { "epoch": 0.12, "learning_rate": 1.9994362458193663e-05, "loss": 0.0613, "step": 71 }, { "epoch": 0.12, "learning_rate": 1.9993718732244574e-05, "loss": 0.0456, "step": 72 }, { "epoch": 0.13, "learning_rate": 1.9993040225255205e-05, "loss": 0.0485, "step": 73 }, { "epoch": 0.13, "learning_rate": 1.9992326939586953e-05, "loss": 0.0452, "step": 74 }, { "epoch": 0.13, "learning_rate": 1.9991578877722255e-05, "loss": 0.0556, "step": 75 }, { "epoch": 0.13, "learning_rate": 1.999079604226459e-05, "loss": 0.0362, "step": 76 }, { "epoch": 0.13, "learning_rate": 1.998997843593845e-05, "loss": 0.0425, "step": 77 }, { "epoch": 0.13, "learning_rate": 1.998912606158934e-05, "loss": 0.0489, "step": 78 }, { "epoch": 0.14, "learning_rate": 1.998823892218377e-05, "loss": 0.053, "step": 79 }, { "epoch": 0.14, "learning_rate": 1.9987317020809245e-05, "loss": 0.0434, "step": 80 }, { "epoch": 0.14, "learning_rate": 1.9986360360674252e-05, "loss": 0.0382, "step": 81 }, { "epoch": 0.14, "learning_rate": 1.9985368945108242e-05, "loss": 0.0593, "step": 82 }, { "epoch": 0.14, "learning_rate": 1.998434277756163e-05, "loss": 0.0526, "step": 83 }, { "epoch": 0.15, "learning_rate": 1.998328186160577e-05, "loss": 0.0508, "step": 84 }, { "epoch": 0.15, "learning_rate": 1.9982186200932964e-05, "loss": 0.0506, "step": 85 }, { "epoch": 0.15, "learning_rate": 1.9981055799356432e-05, "loss": 0.0454, "step": 86 }, { "epoch": 0.15, "learning_rate": 1.997989066081029e-05, "loss": 0.0525, "step": 87 }, { "epoch": 0.15, "learning_rate": 1.997869078934957e-05, "loss": 0.0478, "step": 88 }, { "epoch": 0.15, "learning_rate": 1.9977456189150164e-05, "loss": 0.0426, "step": 89 }, { "epoch": 0.16, "learning_rate": 1.997618686450884e-05, "loss": 0.0504, "step": 90 }, { "epoch": 0.16, "learning_rate": 1.997488281984322e-05, "loss": 0.0591, "step": 91 }, { "epoch": 0.16, "learning_rate": 1.9973544059691755e-05, "loss": 0.0445, "step": 92 }, { "epoch": 0.16, "learning_rate": 1.9972170588713715e-05, "loss": 0.034, "step": 93 }, { "epoch": 0.16, "learning_rate": 1.997076241168918e-05, "loss": 0.0495, "step": 94 }, { "epoch": 0.16, "learning_rate": 1.996931953351902e-05, "loss": 0.0392, "step": 95 }, { "epoch": 0.17, "learning_rate": 1.9967841959224864e-05, "loss": 0.0477, "step": 96 }, { "epoch": 0.17, "learning_rate": 1.9966329693949098e-05, "loss": 0.0511, "step": 97 }, { "epoch": 0.17, "learning_rate": 1.9964782742954848e-05, "loss": 0.0535, "step": 98 }, { "epoch": 0.17, "learning_rate": 1.996320111162595e-05, "loss": 0.0329, "step": 99 }, { "epoch": 0.17, "learning_rate": 1.996158480546694e-05, "loss": 0.0311, "step": 100 }, { "epoch": 0.17, "learning_rate": 1.9959933830103034e-05, "loss": 0.0667, "step": 101 }, { "epoch": 0.18, "learning_rate": 1.9958248191280107e-05, "loss": 0.0507, "step": 102 }, { "epoch": 0.18, "learning_rate": 1.9956527894864662e-05, "loss": 0.0367, "step": 103 }, { "epoch": 0.18, "learning_rate": 1.9954772946843838e-05, "loss": 0.0437, "step": 104 }, { "epoch": 0.18, "learning_rate": 1.9952983353325358e-05, "loss": 0.0363, "step": 105 }, { "epoch": 0.18, "learning_rate": 1.9951159120537533e-05, "loss": 0.0392, "step": 106 }, { "epoch": 0.18, "learning_rate": 1.9949300254829217e-05, "loss": 0.0467, "step": 107 }, { "epoch": 0.19, "learning_rate": 1.9947406762669802e-05, "loss": 0.0364, "step": 108 }, { "epoch": 0.19, "learning_rate": 1.9945478650649192e-05, "loss": 0.047, "step": 109 }, { "epoch": 0.19, "learning_rate": 1.9943515925477777e-05, "loss": 0.0398, "step": 110 }, { "epoch": 0.19, "learning_rate": 1.994151859398641e-05, "loss": 0.0617, "step": 111 }, { "epoch": 0.19, "learning_rate": 1.9939486663126376e-05, "loss": 0.0437, "step": 112 }, { "epoch": 0.2, "learning_rate": 1.9937420139969397e-05, "loss": 0.0488, "step": 113 }, { "epoch": 0.2, "learning_rate": 1.9935319031707564e-05, "loss": 0.0396, "step": 114 }, { "epoch": 0.2, "learning_rate": 1.9933183345653347e-05, "loss": 0.0708, "step": 115 }, { "epoch": 0.2, "learning_rate": 1.993101308923955e-05, "loss": 0.0488, "step": 116 }, { "epoch": 0.2, "learning_rate": 1.9928808270019297e-05, "loss": 0.0366, "step": 117 }, { "epoch": 0.2, "learning_rate": 1.9926568895666e-05, "loss": 0.0399, "step": 118 }, { "epoch": 0.21, "learning_rate": 1.992429497397333e-05, "loss": 0.0329, "step": 119 }, { "epoch": 0.21, "learning_rate": 1.992198651285519e-05, "loss": 0.0405, "step": 120 }, { "epoch": 0.21, "learning_rate": 1.9919643520345698e-05, "loss": 0.0587, "step": 121 }, { "epoch": 0.21, "learning_rate": 1.9917266004599152e-05, "loss": 0.0514, "step": 122 }, { "epoch": 0.21, "learning_rate": 1.9914853973889988e-05, "loss": 0.0383, "step": 123 }, { "epoch": 0.21, "learning_rate": 1.991240743661278e-05, "loss": 0.0372, "step": 124 }, { "epoch": 0.22, "learning_rate": 1.990992640128218e-05, "loss": 0.0539, "step": 125 }, { "epoch": 0.22, "learning_rate": 1.990741087653292e-05, "loss": 0.0324, "step": 126 }, { "epoch": 0.22, "learning_rate": 1.9904860871119746e-05, "loss": 0.0457, "step": 127 }, { "epoch": 0.22, "learning_rate": 1.9902276393917424e-05, "loss": 0.0461, "step": 128 }, { "epoch": 0.22, "learning_rate": 1.989965745392068e-05, "loss": 0.0458, "step": 129 }, { "epoch": 0.22, "learning_rate": 1.989700406024419e-05, "loss": 0.0526, "step": 130 }, { "epoch": 0.23, "learning_rate": 1.9894316222122526e-05, "loss": 0.0403, "step": 131 }, { "epoch": 0.23, "learning_rate": 1.9891593948910152e-05, "loss": 0.0353, "step": 132 }, { "epoch": 0.23, "learning_rate": 1.988883725008136e-05, "loss": 0.0528, "step": 133 }, { "epoch": 0.23, "learning_rate": 1.9886046135230272e-05, "loss": 0.0422, "step": 134 }, { "epoch": 0.23, "learning_rate": 1.988322061407077e-05, "loss": 0.0451, "step": 135 }, { "epoch": 0.23, "learning_rate": 1.988036069643649e-05, "loss": 0.0454, "step": 136 }, { "epoch": 0.24, "learning_rate": 1.9877466392280773e-05, "loss": 0.0468, "step": 137 }, { "epoch": 0.24, "learning_rate": 1.987453771167664e-05, "loss": 0.0416, "step": 138 }, { "epoch": 0.24, "learning_rate": 1.987157466481675e-05, "loss": 0.0288, "step": 139 }, { "epoch": 0.24, "learning_rate": 1.986857726201336e-05, "loss": 0.0423, "step": 140 }, { "epoch": 0.24, "learning_rate": 1.9865545513698304e-05, "loss": 0.0498, "step": 141 }, { "epoch": 0.25, "learning_rate": 1.9862479430422947e-05, "loss": 0.0391, "step": 142 }, { "epoch": 0.25, "learning_rate": 1.985937902285815e-05, "loss": 0.0377, "step": 143 }, { "epoch": 0.25, "learning_rate": 1.9856244301794226e-05, "loss": 0.0434, "step": 144 }, { "epoch": 0.25, "learning_rate": 1.9853075278140913e-05, "loss": 0.035, "step": 145 }, { "epoch": 0.25, "learning_rate": 1.984987196292734e-05, "loss": 0.056, "step": 146 }, { "epoch": 0.25, "learning_rate": 1.984663436730197e-05, "loss": 0.0421, "step": 147 }, { "epoch": 0.26, "learning_rate": 1.984336250253257e-05, "loss": 0.042, "step": 148 }, { "epoch": 0.26, "learning_rate": 1.984005638000618e-05, "loss": 0.0484, "step": 149 }, { "epoch": 0.26, "learning_rate": 1.9836716011229068e-05, "loss": 0.0321, "step": 150 }, { "epoch": 0.26, "learning_rate": 1.983334140782669e-05, "loss": 0.0426, "step": 151 }, { "epoch": 0.26, "learning_rate": 1.982993258154364e-05, "loss": 0.0408, "step": 152 }, { "epoch": 0.26, "learning_rate": 1.9826489544243623e-05, "loss": 0.0443, "step": 153 }, { "epoch": 0.27, "learning_rate": 1.9823012307909405e-05, "loss": 0.0347, "step": 154 }, { "epoch": 0.27, "learning_rate": 1.981950088464278e-05, "loss": 0.0597, "step": 155 }, { "epoch": 0.27, "learning_rate": 1.981595528666452e-05, "loss": 0.0361, "step": 156 }, { "epoch": 0.27, "learning_rate": 1.981237552631434e-05, "loss": 0.0296, "step": 157 }, { "epoch": 0.27, "learning_rate": 1.9808761616050833e-05, "loss": 0.0459, "step": 158 }, { "epoch": 0.27, "learning_rate": 1.9805113568451458e-05, "loss": 0.0383, "step": 159 }, { "epoch": 0.28, "learning_rate": 1.9801431396212486e-05, "loss": 0.0372, "step": 160 }, { "epoch": 0.28, "learning_rate": 1.9797715112148937e-05, "loss": 0.0468, "step": 161 }, { "epoch": 0.28, "learning_rate": 1.9793964729194563e-05, "loss": 0.04, "step": 162 }, { "epoch": 0.28, "learning_rate": 1.9790180260401778e-05, "loss": 0.0465, "step": 163 }, { "epoch": 0.28, "learning_rate": 1.9786361718941637e-05, "loss": 0.0347, "step": 164 }, { "epoch": 0.28, "learning_rate": 1.9782509118103773e-05, "loss": 0.041, "step": 165 }, { "epoch": 0.29, "learning_rate": 1.9778622471296353e-05, "loss": 0.0288, "step": 166 }, { "epoch": 0.29, "learning_rate": 1.977470179204603e-05, "loss": 0.0387, "step": 167 }, { "epoch": 0.29, "learning_rate": 1.977074709399791e-05, "loss": 0.0381, "step": 168 }, { "epoch": 0.29, "learning_rate": 1.9766758390915494e-05, "loss": 0.0575, "step": 169 }, { "epoch": 0.29, "learning_rate": 1.9762735696680617e-05, "loss": 0.0367, "step": 170 }, { "epoch": 0.3, "learning_rate": 1.9758679025293427e-05, "loss": 0.0504, "step": 171 }, { "epoch": 0.3, "learning_rate": 1.975458839087231e-05, "loss": 0.0462, "step": 172 }, { "epoch": 0.3, "learning_rate": 1.9750463807653873e-05, "loss": 0.0325, "step": 173 }, { "epoch": 0.3, "learning_rate": 1.9746305289992854e-05, "loss": 0.0455, "step": 174 }, { "epoch": 0.3, "learning_rate": 1.9742112852362104e-05, "loss": 0.0419, "step": 175 }, { "epoch": 0.3, "learning_rate": 1.9737886509352517e-05, "loss": 0.0459, "step": 176 }, { "epoch": 0.31, "learning_rate": 1.9733626275673e-05, "loss": 0.0311, "step": 177 }, { "epoch": 0.31, "learning_rate": 1.9729332166150393e-05, "loss": 0.028, "step": 178 }, { "epoch": 0.31, "learning_rate": 1.9725004195729452e-05, "loss": 0.0331, "step": 179 }, { "epoch": 0.31, "learning_rate": 1.9720642379472763e-05, "loss": 0.0389, "step": 180 }, { "epoch": 0.31, "learning_rate": 1.9716246732560715e-05, "loss": 0.0362, "step": 181 }, { "epoch": 0.31, "learning_rate": 1.9711817270291435e-05, "loss": 0.0438, "step": 182 }, { "epoch": 0.32, "learning_rate": 1.9707354008080736e-05, "loss": 0.0335, "step": 183 }, { "epoch": 0.32, "learning_rate": 1.9702856961462064e-05, "loss": 0.035, "step": 184 }, { "epoch": 0.32, "learning_rate": 1.9698326146086446e-05, "loss": 0.041, "step": 185 }, { "epoch": 0.32, "learning_rate": 1.9693761577722434e-05, "loss": 0.0383, "step": 186 }, { "epoch": 0.32, "learning_rate": 1.968916327225605e-05, "loss": 0.0393, "step": 187 }, { "epoch": 0.32, "learning_rate": 1.9684531245690735e-05, "loss": 0.0435, "step": 188 }, { "epoch": 0.33, "learning_rate": 1.967986551414728e-05, "loss": 0.0402, "step": 189 }, { "epoch": 0.33, "learning_rate": 1.9675166093863785e-05, "loss": 0.0398, "step": 190 }, { "epoch": 0.33, "learning_rate": 1.9670433001195595e-05, "loss": 0.039, "step": 191 }, { "epoch": 0.33, "learning_rate": 1.966566625261525e-05, "loss": 0.0286, "step": 192 }, { "epoch": 0.33, "learning_rate": 1.9660865864712413e-05, "loss": 0.0377, "step": 193 }, { "epoch": 0.34, "learning_rate": 1.965603185419383e-05, "loss": 0.0284, "step": 194 }, { "epoch": 0.34, "learning_rate": 1.965116423788326e-05, "loss": 0.0272, "step": 195 }, { "epoch": 0.34, "learning_rate": 1.9646263032721412e-05, "loss": 0.037, "step": 196 }, { "epoch": 0.34, "learning_rate": 1.9641328255765916e-05, "loss": 0.0388, "step": 197 }, { "epoch": 0.34, "learning_rate": 1.9636359924191216e-05, "loss": 0.0328, "step": 198 }, { "epoch": 0.34, "learning_rate": 1.9631358055288546e-05, "loss": 0.0407, "step": 199 }, { "epoch": 0.35, "learning_rate": 1.9626322666465874e-05, "loss": 0.0269, "step": 200 }, { "epoch": 0.35, "learning_rate": 1.96212537752478e-05, "loss": 0.035, "step": 201 }, { "epoch": 0.35, "learning_rate": 1.961615139927554e-05, "loss": 0.0297, "step": 202 }, { "epoch": 0.35, "learning_rate": 1.9611015556306845e-05, "loss": 0.0307, "step": 203 }, { "epoch": 0.35, "learning_rate": 1.9605846264215936e-05, "loss": 0.0424, "step": 204 }, { "epoch": 0.35, "learning_rate": 1.9600643540993453e-05, "loss": 0.0281, "step": 205 }, { "epoch": 0.36, "learning_rate": 1.9595407404746377e-05, "loss": 0.0417, "step": 206 }, { "epoch": 0.36, "learning_rate": 1.959013787369798e-05, "loss": 0.0458, "step": 207 }, { "epoch": 0.36, "learning_rate": 1.9584834966187768e-05, "loss": 0.0388, "step": 208 }, { "epoch": 0.36, "learning_rate": 1.9579498700671386e-05, "loss": 0.0353, "step": 209 }, { "epoch": 0.36, "learning_rate": 1.9574129095720597e-05, "loss": 0.0305, "step": 210 }, { "epoch": 0.36, "learning_rate": 1.9568726170023174e-05, "loss": 0.032, "step": 211 }, { "epoch": 0.37, "learning_rate": 1.956328994238287e-05, "loss": 0.0321, "step": 212 }, { "epoch": 0.37, "learning_rate": 1.9557820431719333e-05, "loss": 0.038, "step": 213 }, { "epoch": 0.37, "learning_rate": 1.955231765706805e-05, "loss": 0.0404, "step": 214 }, { "epoch": 0.37, "learning_rate": 1.9546781637580272e-05, "loss": 0.0363, "step": 215 }, { "epoch": 0.37, "learning_rate": 1.9541212392522953e-05, "loss": 0.0452, "step": 216 }, { "epoch": 0.37, "learning_rate": 1.9535609941278676e-05, "loss": 0.0372, "step": 217 }, { "epoch": 0.38, "learning_rate": 1.9529974303345605e-05, "loss": 0.0308, "step": 218 }, { "epoch": 0.38, "learning_rate": 1.9524305498337387e-05, "loss": 0.0401, "step": 219 }, { "epoch": 0.38, "learning_rate": 1.9518603545983114e-05, "loss": 0.0389, "step": 220 }, { "epoch": 0.38, "learning_rate": 1.9512868466127232e-05, "loss": 0.0272, "step": 221 }, { "epoch": 0.38, "learning_rate": 1.950710027872948e-05, "loss": 0.037, "step": 222 }, { "epoch": 0.39, "learning_rate": 1.9501299003864828e-05, "loss": 0.0512, "step": 223 }, { "epoch": 0.39, "learning_rate": 1.9495464661723394e-05, "loss": 0.0388, "step": 224 }, { "epoch": 0.39, "learning_rate": 1.9489597272610377e-05, "loss": 0.0308, "step": 225 }, { "epoch": 0.39, "learning_rate": 1.9483696856946e-05, "loss": 0.0428, "step": 226 }, { "epoch": 0.39, "learning_rate": 1.9477763435265417e-05, "loss": 0.0405, "step": 227 }, { "epoch": 0.39, "learning_rate": 1.9471797028218656e-05, "loss": 0.046, "step": 228 }, { "epoch": 0.4, "learning_rate": 1.9465797656570546e-05, "loss": 0.0352, "step": 229 }, { "epoch": 0.4, "learning_rate": 1.9459765341200643e-05, "loss": 0.0316, "step": 230 }, { "epoch": 0.4, "learning_rate": 1.945370010310315e-05, "loss": 0.0365, "step": 231 }, { "epoch": 0.4, "learning_rate": 1.9447601963386857e-05, "loss": 0.0509, "step": 232 }, { "epoch": 0.4, "learning_rate": 1.944147094327506e-05, "loss": 0.028, "step": 233 }, { "epoch": 0.4, "learning_rate": 1.9435307064105486e-05, "loss": 0.0354, "step": 234 }, { "epoch": 0.41, "learning_rate": 1.9429110347330225e-05, "loss": 0.0281, "step": 235 }, { "epoch": 0.41, "learning_rate": 1.9422880814515646e-05, "loss": 0.0543, "step": 236 }, { "epoch": 0.41, "learning_rate": 1.9416618487342333e-05, "loss": 0.0391, "step": 237 }, { "epoch": 0.41, "learning_rate": 1.9410323387604996e-05, "loss": 0.0386, "step": 238 }, { "epoch": 0.41, "learning_rate": 1.9403995537212418e-05, "loss": 0.0355, "step": 239 }, { "epoch": 0.41, "learning_rate": 1.9397634958187345e-05, "loss": 0.0383, "step": 240 }, { "epoch": 0.42, "learning_rate": 1.9391241672666438e-05, "loss": 0.0267, "step": 241 }, { "epoch": 0.42, "learning_rate": 1.9384815702900193e-05, "loss": 0.0403, "step": 242 }, { "epoch": 0.42, "learning_rate": 1.937835707125284e-05, "loss": 0.0352, "step": 243 }, { "epoch": 0.42, "learning_rate": 1.9371865800202296e-05, "loss": 0.0585, "step": 244 }, { "epoch": 0.42, "learning_rate": 1.936534191234006e-05, "loss": 0.0309, "step": 245 }, { "epoch": 0.42, "learning_rate": 1.9358785430371166e-05, "loss": 0.0431, "step": 246 }, { "epoch": 0.43, "learning_rate": 1.9352196377114064e-05, "loss": 0.0335, "step": 247 }, { "epoch": 0.43, "learning_rate": 1.9345574775500577e-05, "loss": 0.0323, "step": 248 }, { "epoch": 0.43, "learning_rate": 1.9338920648575798e-05, "loss": 0.028, "step": 249 }, { "epoch": 0.43, "learning_rate": 1.933223401949801e-05, "loss": 0.0308, "step": 250 }, { "epoch": 0.43, "learning_rate": 1.9325514911538634e-05, "loss": 0.0393, "step": 251 }, { "epoch": 0.44, "learning_rate": 1.931876334808211e-05, "loss": 0.0347, "step": 252 }, { "epoch": 0.44, "learning_rate": 1.9311979352625837e-05, "loss": 0.023, "step": 253 }, { "epoch": 0.44, "learning_rate": 1.930516294878009e-05, "loss": 0.0341, "step": 254 }, { "epoch": 0.44, "learning_rate": 1.9298314160267923e-05, "loss": 0.0408, "step": 255 }, { "epoch": 0.44, "learning_rate": 1.929143301092512e-05, "loss": 0.0294, "step": 256 }, { "epoch": 0.44, "learning_rate": 1.928451952470007e-05, "loss": 0.0371, "step": 257 }, { "epoch": 0.45, "learning_rate": 1.9277573725653705e-05, "loss": 0.0392, "step": 258 }, { "epoch": 0.45, "learning_rate": 1.927059563795943e-05, "loss": 0.0339, "step": 259 }, { "epoch": 0.45, "learning_rate": 1.926358528590301e-05, "loss": 0.0366, "step": 260 }, { "epoch": 0.45, "learning_rate": 1.9256542693882505e-05, "loss": 0.0393, "step": 261 }, { "epoch": 0.45, "learning_rate": 1.924946788640818e-05, "loss": 0.0398, "step": 262 }, { "epoch": 0.45, "learning_rate": 1.924236088810241e-05, "loss": 0.0348, "step": 263 }, { "epoch": 0.46, "learning_rate": 1.9235221723699623e-05, "loss": 0.0435, "step": 264 }, { "epoch": 0.46, "learning_rate": 1.922805041804617e-05, "loss": 0.0283, "step": 265 }, { "epoch": 0.46, "learning_rate": 1.9220846996100283e-05, "loss": 0.0372, "step": 266 }, { "epoch": 0.46, "learning_rate": 1.9213611482931954e-05, "loss": 0.0381, "step": 267 }, { "epoch": 0.46, "learning_rate": 1.9206343903722874e-05, "loss": 0.0333, "step": 268 }, { "epoch": 0.46, "learning_rate": 1.919904428376632e-05, "loss": 0.0377, "step": 269 }, { "epoch": 0.47, "learning_rate": 1.9191712648467093e-05, "loss": 0.0366, "step": 270 }, { "epoch": 0.47, "learning_rate": 1.918434902334141e-05, "loss": 0.039, "step": 271 }, { "epoch": 0.47, "learning_rate": 1.917695343401682e-05, "loss": 0.0395, "step": 272 }, { "epoch": 0.47, "learning_rate": 1.916952590623212e-05, "loss": 0.0325, "step": 273 }, { "epoch": 0.47, "learning_rate": 1.9162066465837263e-05, "loss": 0.0292, "step": 274 }, { "epoch": 0.47, "learning_rate": 1.9154575138793263e-05, "loss": 0.0387, "step": 275 }, { "epoch": 0.48, "learning_rate": 1.9147051951172112e-05, "loss": 0.0402, "step": 276 }, { "epoch": 0.48, "learning_rate": 1.9139496929156685e-05, "loss": 0.0417, "step": 277 }, { "epoch": 0.48, "learning_rate": 1.9131910099040655e-05, "loss": 0.0501, "step": 278 }, { "epoch": 0.48, "learning_rate": 1.9124291487228386e-05, "loss": 0.0406, "step": 279 }, { "epoch": 0.48, "learning_rate": 1.9116641120234858e-05, "loss": 0.0263, "step": 280 }, { "epoch": 0.49, "learning_rate": 1.910895902468557e-05, "loss": 0.041, "step": 281 }, { "epoch": 0.49, "learning_rate": 1.9101245227316442e-05, "loss": 0.0319, "step": 282 }, { "epoch": 0.49, "learning_rate": 1.909349975497372e-05, "loss": 0.0402, "step": 283 }, { "epoch": 0.49, "learning_rate": 1.9085722634613906e-05, "loss": 0.0353, "step": 284 }, { "epoch": 0.49, "learning_rate": 1.907791389330363e-05, "loss": 0.0313, "step": 285 }, { "epoch": 0.49, "learning_rate": 1.9070073558219576e-05, "loss": 0.0385, "step": 286 }, { "epoch": 0.5, "learning_rate": 1.9062201656648382e-05, "loss": 0.0355, "step": 287 }, { "epoch": 0.5, "learning_rate": 1.9054298215986557e-05, "loss": 0.0279, "step": 288 }, { "epoch": 0.5, "learning_rate": 1.904636326374036e-05, "loss": 0.0235, "step": 289 }, { "epoch": 0.5, "learning_rate": 1.903839682752573e-05, "loss": 0.0374, "step": 290 }, { "epoch": 0.5, "learning_rate": 1.9030398935068173e-05, "loss": 0.0217, "step": 291 }, { "epoch": 0.5, "learning_rate": 1.9022369614202677e-05, "loss": 0.048, "step": 292 }, { "epoch": 0.51, "learning_rate": 1.9014308892873612e-05, "loss": 0.0455, "step": 293 }, { "epoch": 0.51, "learning_rate": 1.9006216799134618e-05, "loss": 0.0343, "step": 294 }, { "epoch": 0.51, "learning_rate": 1.8998093361148534e-05, "loss": 0.035, "step": 295 }, { "epoch": 0.51, "learning_rate": 1.898993860718728e-05, "loss": 0.0313, "step": 296 }, { "epoch": 0.51, "learning_rate": 1.8981752565631767e-05, "loss": 0.0276, "step": 297 }, { "epoch": 0.51, "learning_rate": 1.897353526497179e-05, "loss": 0.0331, "step": 298 }, { "epoch": 0.52, "learning_rate": 1.8965286733805945e-05, "loss": 0.0329, "step": 299 }, { "epoch": 0.52, "learning_rate": 1.8957007000841508e-05, "loss": 0.0242, "step": 300 }, { "epoch": 0.52, "learning_rate": 1.8948696094894354e-05, "loss": 0.025, "step": 301 }, { "epoch": 0.52, "learning_rate": 1.8940354044888843e-05, "loss": 0.0435, "step": 302 }, { "epoch": 0.52, "learning_rate": 1.8931980879857737e-05, "loss": 0.0475, "step": 303 }, { "epoch": 0.53, "learning_rate": 1.8923576628942072e-05, "loss": 0.0354, "step": 304 }, { "epoch": 0.53, "learning_rate": 1.8915141321391083e-05, "loss": 0.0357, "step": 305 }, { "epoch": 0.53, "learning_rate": 1.890667498656208e-05, "loss": 0.0403, "step": 306 }, { "epoch": 0.53, "learning_rate": 1.8898177653920374e-05, "loss": 0.0413, "step": 307 }, { "epoch": 0.53, "learning_rate": 1.8889649353039145e-05, "loss": 0.0283, "step": 308 }, { "epoch": 0.53, "learning_rate": 1.8881090113599353e-05, "loss": 0.0226, "step": 309 }, { "epoch": 0.54, "learning_rate": 1.8872499965389633e-05, "loss": 0.0236, "step": 310 }, { "epoch": 0.54, "learning_rate": 1.8863878938306193e-05, "loss": 0.0284, "step": 311 }, { "epoch": 0.54, "learning_rate": 1.8855227062352707e-05, "loss": 0.0391, "step": 312 }, { "epoch": 0.54, "learning_rate": 1.8846544367640218e-05, "loss": 0.027, "step": 313 }, { "epoch": 0.54, "learning_rate": 1.8837830884387022e-05, "loss": 0.0287, "step": 314 }, { "epoch": 0.54, "learning_rate": 1.882908664291856e-05, "loss": 0.0373, "step": 315 }, { "epoch": 0.55, "learning_rate": 1.8820311673667336e-05, "loss": 0.0353, "step": 316 }, { "epoch": 0.55, "learning_rate": 1.881150600717279e-05, "loss": 0.044, "step": 317 }, { "epoch": 0.55, "learning_rate": 1.880266967408119e-05, "loss": 0.0264, "step": 318 }, { "epoch": 0.55, "learning_rate": 1.8793802705145542e-05, "loss": 0.0277, "step": 319 }, { "epoch": 0.55, "learning_rate": 1.8784905131225472e-05, "loss": 0.0396, "step": 320 }, { "epoch": 0.55, "learning_rate": 1.8775976983287117e-05, "loss": 0.0292, "step": 321 }, { "epoch": 0.56, "learning_rate": 1.8767018292403015e-05, "loss": 0.0432, "step": 322 }, { "epoch": 0.56, "learning_rate": 1.8758029089752023e-05, "loss": 0.0416, "step": 323 }, { "epoch": 0.56, "learning_rate": 1.874900940661916e-05, "loss": 0.0358, "step": 324 }, { "epoch": 0.56, "learning_rate": 1.873995927439555e-05, "loss": 0.0336, "step": 325 }, { "epoch": 0.56, "learning_rate": 1.873087872457827e-05, "loss": 0.0254, "step": 326 }, { "epoch": 0.56, "learning_rate": 1.8721767788770274e-05, "loss": 0.0389, "step": 327 }, { "epoch": 0.57, "learning_rate": 1.871262649868026e-05, "loss": 0.0267, "step": 328 }, { "epoch": 0.57, "learning_rate": 1.8703454886122568e-05, "loss": 0.0373, "step": 329 }, { "epoch": 0.57, "learning_rate": 1.869425298301707e-05, "loss": 0.0305, "step": 330 }, { "epoch": 0.57, "learning_rate": 1.8685020821389058e-05, "loss": 0.0342, "step": 331 }, { "epoch": 0.57, "learning_rate": 1.8675758433369137e-05, "loss": 0.0442, "step": 332 }, { "epoch": 0.58, "learning_rate": 1.86664658511931e-05, "loss": 0.0259, "step": 333 }, { "epoch": 0.58, "learning_rate": 1.865714310720183e-05, "loss": 0.0197, "step": 334 }, { "epoch": 0.58, "learning_rate": 1.8647790233841183e-05, "loss": 0.0346, "step": 335 }, { "epoch": 0.58, "learning_rate": 1.8638407263661865e-05, "loss": 0.0296, "step": 336 }, { "epoch": 0.58, "learning_rate": 1.862899422931934e-05, "loss": 0.0475, "step": 337 }, { "epoch": 0.58, "learning_rate": 1.8619551163573697e-05, "loss": 0.0336, "step": 338 }, { "epoch": 0.59, "learning_rate": 1.8610078099289544e-05, "loss": 0.0284, "step": 339 }, { "epoch": 0.59, "learning_rate": 1.860057506943589e-05, "loss": 0.03, "step": 340 }, { "epoch": 0.59, "learning_rate": 1.859104210708604e-05, "loss": 0.028, "step": 341 }, { "epoch": 0.59, "learning_rate": 1.8581479245417463e-05, "loss": 0.0223, "step": 342 }, { "epoch": 0.59, "learning_rate": 1.85718865177117e-05, "loss": 0.0315, "step": 343 }, { "epoch": 0.59, "learning_rate": 1.8562263957354214e-05, "loss": 0.0309, "step": 344 }, { "epoch": 0.6, "learning_rate": 1.855261159783432e-05, "loss": 0.0342, "step": 345 }, { "epoch": 0.6, "learning_rate": 1.8542929472745016e-05, "loss": 0.023, "step": 346 }, { "epoch": 0.6, "learning_rate": 1.8533217615782918e-05, "loss": 0.035, "step": 347 }, { "epoch": 0.6, "learning_rate": 1.85234760607481e-05, "loss": 0.0321, "step": 348 }, { "epoch": 0.6, "learning_rate": 1.8513704841543997e-05, "loss": 0.0353, "step": 349 }, { "epoch": 0.6, "learning_rate": 1.8503903992177293e-05, "loss": 0.0322, "step": 350 }, { "epoch": 0.61, "learning_rate": 1.849407354675778e-05, "loss": 0.0345, "step": 351 }, { "epoch": 0.61, "learning_rate": 1.8484213539498266e-05, "loss": 0.0261, "step": 352 }, { "epoch": 0.61, "learning_rate": 1.847432400471443e-05, "loss": 0.0343, "step": 353 }, { "epoch": 0.61, "learning_rate": 1.8464404976824727e-05, "loss": 0.023, "step": 354 }, { "epoch": 0.61, "learning_rate": 1.8454456490350254e-05, "loss": 0.037, "step": 355 }, { "epoch": 0.61, "learning_rate": 1.8444478579914618e-05, "loss": 0.0327, "step": 356 }, { "epoch": 0.62, "learning_rate": 1.8434471280243854e-05, "loss": 0.0283, "step": 357 }, { "epoch": 0.62, "learning_rate": 1.8424434626166263e-05, "loss": 0.0323, "step": 358 }, { "epoch": 0.62, "learning_rate": 1.841436865261231e-05, "loss": 0.0307, "step": 359 }, { "epoch": 0.62, "learning_rate": 1.8404273394614512e-05, "loss": 0.0331, "step": 360 }, { "epoch": 0.62, "learning_rate": 1.8394148887307286e-05, "loss": 0.0366, "step": 361 }, { "epoch": 0.63, "learning_rate": 1.838399516592686e-05, "loss": 0.0348, "step": 362 }, { "epoch": 0.63, "learning_rate": 1.8373812265811126e-05, "loss": 0.0354, "step": 363 }, { "epoch": 0.63, "learning_rate": 1.8363600222399534e-05, "loss": 0.0362, "step": 364 }, { "epoch": 0.63, "learning_rate": 1.8353359071232954e-05, "loss": 0.0276, "step": 365 }, { "epoch": 0.63, "learning_rate": 1.8343088847953568e-05, "loss": 0.0324, "step": 366 }, { "epoch": 0.63, "learning_rate": 1.8332789588304726e-05, "loss": 0.0294, "step": 367 }, { "epoch": 0.64, "learning_rate": 1.8322461328130846e-05, "loss": 0.0258, "step": 368 }, { "epoch": 0.64, "learning_rate": 1.8312104103377266e-05, "loss": 0.0465, "step": 369 }, { "epoch": 0.64, "learning_rate": 1.830171795009013e-05, "loss": 0.0262, "step": 370 }, { "epoch": 0.64, "learning_rate": 1.8291302904416266e-05, "loss": 0.0328, "step": 371 }, { "epoch": 0.64, "learning_rate": 1.828085900260306e-05, "loss": 0.0297, "step": 372 }, { "epoch": 0.64, "learning_rate": 1.827038628099831e-05, "loss": 0.0362, "step": 373 }, { "epoch": 0.65, "learning_rate": 1.8259884776050133e-05, "loss": 0.0326, "step": 374 }, { "epoch": 0.65, "learning_rate": 1.824935452430681e-05, "loss": 0.0289, "step": 375 }, { "epoch": 0.65, "learning_rate": 1.8238795562416672e-05, "loss": 0.0332, "step": 376 }, { "epoch": 0.65, "learning_rate": 1.822820792712797e-05, "loss": 0.0273, "step": 377 }, { "epoch": 0.65, "learning_rate": 1.821759165528875e-05, "loss": 0.0271, "step": 378 }, { "epoch": 0.65, "learning_rate": 1.8206946783846708e-05, "loss": 0.026, "step": 379 }, { "epoch": 0.66, "learning_rate": 1.819627334984909e-05, "loss": 0.0313, "step": 380 }, { "epoch": 0.66, "learning_rate": 1.8185571390442542e-05, "loss": 0.0357, "step": 381 }, { "epoch": 0.66, "learning_rate": 1.817484094287299e-05, "loss": 0.0394, "step": 382 }, { "epoch": 0.66, "learning_rate": 1.81640820444855e-05, "loss": 0.0339, "step": 383 }, { "epoch": 0.66, "learning_rate": 1.815329473272417e-05, "loss": 0.0324, "step": 384 }, { "epoch": 0.66, "learning_rate": 1.8142479045131956e-05, "loss": 0.0298, "step": 385 }, { "epoch": 0.67, "learning_rate": 1.8131635019350603e-05, "loss": 0.0303, "step": 386 }, { "epoch": 0.67, "learning_rate": 1.812076269312046e-05, "loss": 0.0302, "step": 387 }, { "epoch": 0.67, "learning_rate": 1.8109862104280374e-05, "loss": 0.0293, "step": 388 }, { "epoch": 0.67, "learning_rate": 1.809893329076757e-05, "loss": 0.0331, "step": 389 }, { "epoch": 0.67, "learning_rate": 1.808797629061747e-05, "loss": 0.0255, "step": 390 }, { "epoch": 0.68, "learning_rate": 1.807699114196363e-05, "loss": 0.0335, "step": 391 }, { "epoch": 0.68, "learning_rate": 1.8065977883037543e-05, "loss": 0.03, "step": 392 }, { "epoch": 0.68, "learning_rate": 1.8054936552168548e-05, "loss": 0.0352, "step": 393 }, { "epoch": 0.68, "learning_rate": 1.8043867187783682e-05, "loss": 0.0287, "step": 394 }, { "epoch": 0.68, "learning_rate": 1.8032769828407543e-05, "loss": 0.0233, "step": 395 }, { "epoch": 0.68, "learning_rate": 1.802164451266216e-05, "loss": 0.0311, "step": 396 }, { "epoch": 0.69, "learning_rate": 1.801049127926686e-05, "loss": 0.0253, "step": 397 }, { "epoch": 0.69, "learning_rate": 1.7999310167038126e-05, "loss": 0.0233, "step": 398 }, { "epoch": 0.69, "learning_rate": 1.798810121488948e-05, "loss": 0.0272, "step": 399 }, { "epoch": 0.69, "learning_rate": 1.7976864461831323e-05, "loss": 0.025, "step": 400 }, { "epoch": 0.69, "learning_rate": 1.7965599946970814e-05, "loss": 0.0337, "step": 401 }, { "epoch": 0.69, "learning_rate": 1.7954307709511735e-05, "loss": 0.0329, "step": 402 }, { "epoch": 0.7, "learning_rate": 1.7942987788754348e-05, "loss": 0.0331, "step": 403 }, { "epoch": 0.7, "learning_rate": 1.793164022409526e-05, "loss": 0.0344, "step": 404 }, { "epoch": 0.7, "learning_rate": 1.7920265055027285e-05, "loss": 0.0272, "step": 405 }, { "epoch": 0.7, "learning_rate": 1.7908862321139318e-05, "loss": 0.0335, "step": 406 }, { "epoch": 0.7, "learning_rate": 1.789743206211618e-05, "loss": 0.0291, "step": 407 }, { "epoch": 0.7, "learning_rate": 1.7885974317738477e-05, "loss": 0.0346, "step": 408 }, { "epoch": 0.71, "learning_rate": 1.7874489127882496e-05, "loss": 0.0292, "step": 409 }, { "epoch": 0.71, "learning_rate": 1.7862976532520025e-05, "loss": 0.0257, "step": 410 }, { "epoch": 0.71, "learning_rate": 1.7851436571718235e-05, "loss": 0.0397, "step": 411 }, { "epoch": 0.71, "learning_rate": 1.783986928563954e-05, "loss": 0.0317, "step": 412 }, { "epoch": 0.71, "learning_rate": 1.7828274714541445e-05, "loss": 0.0331, "step": 413 }, { "epoch": 0.72, "learning_rate": 1.781665289877643e-05, "loss": 0.0335, "step": 414 }, { "epoch": 0.72, "learning_rate": 1.7805003878791775e-05, "loss": 0.0274, "step": 415 }, { "epoch": 0.72, "learning_rate": 1.7793327695129455e-05, "loss": 0.0359, "step": 416 }, { "epoch": 0.72, "learning_rate": 1.7781624388425974e-05, "loss": 0.0279, "step": 417 }, { "epoch": 0.72, "learning_rate": 1.7769893999412238e-05, "loss": 0.0281, "step": 418 }, { "epoch": 0.72, "learning_rate": 1.7758136568913403e-05, "loss": 0.0347, "step": 419 }, { "epoch": 0.73, "learning_rate": 1.7746352137848732e-05, "loss": 0.0256, "step": 420 }, { "epoch": 0.73, "learning_rate": 1.773454074723147e-05, "loss": 0.0293, "step": 421 }, { "epoch": 0.73, "learning_rate": 1.772270243816868e-05, "loss": 0.0317, "step": 422 }, { "epoch": 0.73, "learning_rate": 1.771083725186111e-05, "loss": 0.022, "step": 423 }, { "epoch": 0.73, "learning_rate": 1.7698945229603057e-05, "loss": 0.0241, "step": 424 }, { "epoch": 0.73, "learning_rate": 1.76870264127822e-05, "loss": 0.0258, "step": 425 }, { "epoch": 0.74, "learning_rate": 1.767508084287949e-05, "loss": 0.0323, "step": 426 }, { "epoch": 0.74, "learning_rate": 1.766310856146896e-05, "loss": 0.0251, "step": 427 }, { "epoch": 0.74, "learning_rate": 1.7651109610217635e-05, "loss": 0.0275, "step": 428 }, { "epoch": 0.74, "learning_rate": 1.763908403088534e-05, "loss": 0.0245, "step": 429 }, { "epoch": 0.74, "learning_rate": 1.7627031865324587e-05, "loss": 0.0328, "step": 430 }, { "epoch": 0.74, "learning_rate": 1.76149531554804e-05, "loss": 0.0302, "step": 431 }, { "epoch": 0.75, "learning_rate": 1.7602847943390194e-05, "loss": 0.0318, "step": 432 }, { "epoch": 0.75, "learning_rate": 1.759071627118362e-05, "loss": 0.0257, "step": 433 }, { "epoch": 0.75, "learning_rate": 1.757855818108242e-05, "loss": 0.0292, "step": 434 }, { "epoch": 0.75, "learning_rate": 1.7566373715400274e-05, "loss": 0.0384, "step": 435 }, { "epoch": 0.75, "learning_rate": 1.755416291654265e-05, "loss": 0.026, "step": 436 }, { "epoch": 0.75, "learning_rate": 1.754192582700668e-05, "loss": 0.0208, "step": 437 }, { "epoch": 0.76, "learning_rate": 1.7529662489380975e-05, "loss": 0.0237, "step": 438 }, { "epoch": 0.76, "learning_rate": 1.751737294634552e-05, "loss": 0.0237, "step": 439 }, { "epoch": 0.76, "learning_rate": 1.750505724067148e-05, "loss": 0.0368, "step": 440 }, { "epoch": 0.76, "learning_rate": 1.7492715415221087e-05, "loss": 0.0266, "step": 441 }, { "epoch": 0.76, "learning_rate": 1.7480347512947474e-05, "loss": 0.029, "step": 442 }, { "epoch": 0.77, "learning_rate": 1.746795357689453e-05, "loss": 0.0293, "step": 443 }, { "epoch": 0.77, "learning_rate": 1.7455533650196744e-05, "loss": 0.0199, "step": 444 }, { "epoch": 0.77, "learning_rate": 1.7443087776079068e-05, "loss": 0.0246, "step": 445 }, { "epoch": 0.77, "learning_rate": 1.743061599785675e-05, "loss": 0.0273, "step": 446 }, { "epoch": 0.77, "learning_rate": 1.7418118358935196e-05, "loss": 0.0322, "step": 447 }, { "epoch": 0.77, "learning_rate": 1.740559490280982e-05, "loss": 0.0279, "step": 448 }, { "epoch": 0.78, "learning_rate": 1.739304567306588e-05, "loss": 0.0224, "step": 449 }, { "epoch": 0.78, "learning_rate": 1.7380470713378336e-05, "loss": 0.0212, "step": 450 }, { "epoch": 0.78, "learning_rate": 1.7367870067511695e-05, "loss": 0.0342, "step": 451 }, { "epoch": 0.78, "learning_rate": 1.735524377931986e-05, "loss": 0.0381, "step": 452 }, { "epoch": 0.78, "learning_rate": 1.7342591892745978e-05, "loss": 0.0303, "step": 453 }, { "epoch": 0.78, "learning_rate": 1.7329914451822278e-05, "loss": 0.0219, "step": 454 }, { "epoch": 0.79, "learning_rate": 1.7317211500669944e-05, "loss": 0.0426, "step": 455 }, { "epoch": 0.79, "learning_rate": 1.730448308349892e-05, "loss": 0.0375, "step": 456 }, { "epoch": 0.79, "learning_rate": 1.72917292446078e-05, "loss": 0.0352, "step": 457 }, { "epoch": 0.79, "learning_rate": 1.7278950028383632e-05, "loss": 0.0276, "step": 458 }, { "epoch": 0.79, "learning_rate": 1.72661454793018e-05, "loss": 0.0288, "step": 459 }, { "epoch": 0.79, "learning_rate": 1.7253315641925854e-05, "loss": 0.0204, "step": 460 }, { "epoch": 0.8, "learning_rate": 1.7240460560907345e-05, "loss": 0.0196, "step": 461 }, { "epoch": 0.8, "learning_rate": 1.7227580280985687e-05, "loss": 0.0317, "step": 462 }, { "epoch": 0.8, "learning_rate": 1.7214674846987992e-05, "loss": 0.0209, "step": 463 }, { "epoch": 0.8, "learning_rate": 1.7201744303828904e-05, "loss": 0.0302, "step": 464 }, { "epoch": 0.8, "learning_rate": 1.7188788696510477e-05, "loss": 0.0421, "step": 465 }, { "epoch": 0.8, "learning_rate": 1.7175808070121976e-05, "loss": 0.0312, "step": 466 }, { "epoch": 0.81, "learning_rate": 1.7162802469839747e-05, "loss": 0.0393, "step": 467 }, { "epoch": 0.81, "learning_rate": 1.7149771940927055e-05, "loss": 0.0296, "step": 468 }, { "epoch": 0.81, "learning_rate": 1.7136716528733916e-05, "loss": 0.0333, "step": 469 }, { "epoch": 0.81, "learning_rate": 1.7123636278696952e-05, "loss": 0.0307, "step": 470 }, { "epoch": 0.81, "learning_rate": 1.711053123633923e-05, "loss": 0.0233, "step": 471 }, { "epoch": 0.82, "learning_rate": 1.7097401447270103e-05, "loss": 0.042, "step": 472 }, { "epoch": 0.82, "learning_rate": 1.7084246957185036e-05, "loss": 0.034, "step": 473 }, { "epoch": 0.82, "learning_rate": 1.7071067811865477e-05, "loss": 0.0281, "step": 474 }, { "epoch": 0.82, "learning_rate": 1.7057864057178674e-05, "loss": 0.036, "step": 475 }, { "epoch": 0.82, "learning_rate": 1.7044635739077523e-05, "loss": 0.0253, "step": 476 }, { "epoch": 0.82, "learning_rate": 1.703138290360041e-05, "loss": 0.0248, "step": 477 }, { "epoch": 0.83, "learning_rate": 1.7018105596871046e-05, "loss": 0.0302, "step": 478 }, { "epoch": 0.83, "learning_rate": 1.700480386509831e-05, "loss": 0.0261, "step": 479 }, { "epoch": 0.83, "learning_rate": 1.6991477754576084e-05, "loss": 0.0254, "step": 480 }, { "epoch": 0.83, "learning_rate": 1.6978127311683103e-05, "loss": 0.0373, "step": 481 }, { "epoch": 0.83, "learning_rate": 1.696475258288278e-05, "loss": 0.0272, "step": 482 }, { "epoch": 0.83, "learning_rate": 1.695135361472305e-05, "loss": 0.0325, "step": 483 }, { "epoch": 0.84, "learning_rate": 1.6937930453836212e-05, "loss": 0.0255, "step": 484 }, { "epoch": 0.84, "learning_rate": 1.6924483146938756e-05, "loss": 0.027, "step": 485 }, { "epoch": 0.84, "learning_rate": 1.6911011740831212e-05, "loss": 0.0248, "step": 486 }, { "epoch": 0.84, "learning_rate": 1.6897516282397982e-05, "loss": 0.035, "step": 487 }, { "epoch": 0.84, "learning_rate": 1.688399681860718e-05, "loss": 0.0204, "step": 488 }, { "epoch": 0.84, "learning_rate": 1.6870453396510456e-05, "loss": 0.0407, "step": 489 }, { "epoch": 0.85, "learning_rate": 1.6856886063242847e-05, "loss": 0.0308, "step": 490 }, { "epoch": 0.85, "learning_rate": 1.684329486602261e-05, "loss": 0.025, "step": 491 }, { "epoch": 0.85, "learning_rate": 1.682967985215106e-05, "loss": 0.0353, "step": 492 }, { "epoch": 0.85, "learning_rate": 1.681604106901239e-05, "loss": 0.0232, "step": 493 }, { "epoch": 0.85, "learning_rate": 1.6802378564073524e-05, "loss": 0.0245, "step": 494 }, { "epoch": 0.85, "learning_rate": 1.6788692384883933e-05, "loss": 0.0254, "step": 495 }, { "epoch": 0.86, "learning_rate": 1.6774982579075503e-05, "loss": 0.0358, "step": 496 }, { "epoch": 0.86, "learning_rate": 1.676124919436233e-05, "loss": 0.0276, "step": 497 }, { "epoch": 0.86, "learning_rate": 1.6747492278540578e-05, "loss": 0.0273, "step": 498 }, { "epoch": 0.86, "learning_rate": 1.6733711879488304e-05, "loss": 0.0226, "step": 499 }, { "epoch": 0.86, "learning_rate": 1.6719908045165295e-05, "loss": 0.0237, "step": 500 }, { "epoch": 0.87, "learning_rate": 1.6706080823612897e-05, "loss": 0.0215, "step": 501 }, { "epoch": 0.87, "learning_rate": 1.669223026295385e-05, "loss": 0.018, "step": 502 }, { "epoch": 0.87, "learning_rate": 1.6678356411392135e-05, "loss": 0.0256, "step": 503 }, { "epoch": 0.87, "learning_rate": 1.6664459317212765e-05, "loss": 0.0301, "step": 504 }, { "epoch": 0.87, "learning_rate": 1.665053902878167e-05, "loss": 0.0331, "step": 505 }, { "epoch": 0.87, "learning_rate": 1.6636595594545486e-05, "loss": 0.0279, "step": 506 }, { "epoch": 0.88, "learning_rate": 1.6622629063031413e-05, "loss": 0.0251, "step": 507 }, { "epoch": 0.88, "learning_rate": 1.6608639482847027e-05, "loss": 0.034, "step": 508 }, { "epoch": 0.88, "learning_rate": 1.659462690268013e-05, "loss": 0.0279, "step": 509 }, { "epoch": 0.88, "learning_rate": 1.6580591371298563e-05, "loss": 0.0333, "step": 510 }, { "epoch": 0.88, "learning_rate": 1.6566532937550048e-05, "loss": 0.025, "step": 511 }, { "epoch": 0.88, "learning_rate": 1.6552451650362014e-05, "loss": 0.0359, "step": 512 }, { "epoch": 0.89, "learning_rate": 1.6538347558741424e-05, "loss": 0.0195, "step": 513 }, { "epoch": 0.89, "learning_rate": 1.6524220711774604e-05, "loss": 0.0334, "step": 514 }, { "epoch": 0.89, "learning_rate": 1.651007115862709e-05, "loss": 0.0319, "step": 515 }, { "epoch": 0.89, "learning_rate": 1.6495898948543426e-05, "loss": 0.028, "step": 516 }, { "epoch": 0.89, "learning_rate": 1.6481704130847013e-05, "loss": 0.0358, "step": 517 }, { "epoch": 0.89, "learning_rate": 1.646748675493993e-05, "loss": 0.0307, "step": 518 }, { "epoch": 0.9, "learning_rate": 1.6453246870302778e-05, "loss": 0.0287, "step": 519 }, { "epoch": 0.9, "learning_rate": 1.6438984526494483e-05, "loss": 0.0262, "step": 520 }, { "epoch": 0.9, "learning_rate": 1.642469977315214e-05, "loss": 0.0326, "step": 521 }, { "epoch": 0.9, "learning_rate": 1.6410392659990833e-05, "loss": 0.0295, "step": 522 }, { "epoch": 0.9, "learning_rate": 1.6396063236803465e-05, "loss": 0.0234, "step": 523 }, { "epoch": 0.91, "learning_rate": 1.6381711553460587e-05, "loss": 0.0257, "step": 524 }, { "epoch": 0.91, "learning_rate": 1.6367337659910223e-05, "loss": 0.0342, "step": 525 }, { "epoch": 0.91, "learning_rate": 1.635294160617769e-05, "loss": 0.0237, "step": 526 }, { "epoch": 0.91, "learning_rate": 1.633852344236543e-05, "loss": 0.036, "step": 527 }, { "epoch": 0.91, "learning_rate": 1.6324083218652836e-05, "loss": 0.0278, "step": 528 }, { "epoch": 0.91, "learning_rate": 1.6309620985296075e-05, "loss": 0.0304, "step": 529 }, { "epoch": 0.92, "learning_rate": 1.629513679262792e-05, "loss": 0.0318, "step": 530 }, { "epoch": 0.92, "learning_rate": 1.6280630691057556e-05, "loss": 0.035, "step": 531 }, { "epoch": 0.92, "learning_rate": 1.626610273107043e-05, "loss": 0.0285, "step": 532 }, { "epoch": 0.92, "learning_rate": 1.625155296322805e-05, "loss": 0.0306, "step": 533 }, { "epoch": 0.92, "learning_rate": 1.6236981438167835e-05, "loss": 0.0326, "step": 534 }, { "epoch": 0.92, "learning_rate": 1.6222388206602923e-05, "loss": 0.0221, "step": 535 }, { "epoch": 0.93, "learning_rate": 1.620777331932198e-05, "loss": 0.0264, "step": 536 }, { "epoch": 0.93, "learning_rate": 1.6193136827189067e-05, "loss": 0.0327, "step": 537 }, { "epoch": 0.93, "learning_rate": 1.6178478781143414e-05, "loss": 0.0246, "step": 538 }, { "epoch": 0.93, "learning_rate": 1.6163799232199276e-05, "loss": 0.0241, "step": 539 }, { "epoch": 0.93, "learning_rate": 1.6149098231445744e-05, "loss": 0.0257, "step": 540 }, { "epoch": 0.93, "learning_rate": 1.6134375830046566e-05, "loss": 0.0252, "step": 541 }, { "epoch": 0.94, "learning_rate": 1.611963207923996e-05, "loss": 0.0292, "step": 542 }, { "epoch": 0.94, "learning_rate": 1.610486703033847e-05, "loss": 0.0253, "step": 543 }, { "epoch": 0.94, "learning_rate": 1.609008073472874e-05, "loss": 0.0242, "step": 544 }, { "epoch": 0.94, "learning_rate": 1.607527324387137e-05, "loss": 0.0248, "step": 545 }, { "epoch": 0.94, "learning_rate": 1.6060444609300726e-05, "loss": 0.0258, "step": 546 }, { "epoch": 0.94, "learning_rate": 1.604559488262475e-05, "loss": 0.0324, "step": 547 }, { "epoch": 0.95, "learning_rate": 1.6030724115524812e-05, "loss": 0.0341, "step": 548 }, { "epoch": 0.95, "learning_rate": 1.6015832359755483e-05, "loss": 0.0238, "step": 549 }, { "epoch": 0.95, "learning_rate": 1.60009196671444e-05, "loss": 0.0313, "step": 550 }, { "epoch": 0.95, "learning_rate": 1.598598608959206e-05, "loss": 0.0287, "step": 551 }, { "epoch": 0.95, "learning_rate": 1.597103167907164e-05, "loss": 0.0355, "step": 552 }, { "epoch": 0.96, "learning_rate": 1.5956056487628832e-05, "loss": 0.024, "step": 553 }, { "epoch": 0.96, "learning_rate": 1.594106056738164e-05, "loss": 0.0231, "step": 554 }, { "epoch": 0.96, "learning_rate": 1.592604397052022e-05, "loss": 0.0234, "step": 555 }, { "epoch": 0.96, "learning_rate": 1.591100674930669e-05, "loss": 0.0418, "step": 556 }, { "epoch": 0.96, "learning_rate": 1.5895948956074937e-05, "loss": 0.0253, "step": 557 }, { "epoch": 0.96, "learning_rate": 1.5880870643230447e-05, "loss": 0.0331, "step": 558 }, { "epoch": 0.97, "learning_rate": 1.5865771863250125e-05, "loss": 0.0266, "step": 559 }, { "epoch": 0.97, "learning_rate": 1.585065266868211e-05, "loss": 0.0387, "step": 560 }, { "epoch": 0.97, "learning_rate": 1.5835513112145583e-05, "loss": 0.0317, "step": 561 }, { "epoch": 0.97, "learning_rate": 1.5820353246330588e-05, "loss": 0.0269, "step": 562 }, { "epoch": 0.97, "learning_rate": 1.5805173123997856e-05, "loss": 0.0252, "step": 563 }, { "epoch": 0.97, "learning_rate": 1.5789972797978623e-05, "loss": 0.0209, "step": 564 }, { "epoch": 0.98, "learning_rate": 1.5774752321174428e-05, "loss": 0.0235, "step": 565 }, { "epoch": 0.98, "learning_rate": 1.5759511746556948e-05, "loss": 0.0229, "step": 566 }, { "epoch": 0.98, "learning_rate": 1.57442511271678e-05, "loss": 0.0384, "step": 567 }, { "epoch": 0.98, "learning_rate": 1.5728970516118374e-05, "loss": 0.0296, "step": 568 }, { "epoch": 0.98, "learning_rate": 1.571366996658962e-05, "loss": 0.0251, "step": 569 }, { "epoch": 0.98, "learning_rate": 1.5698349531831895e-05, "loss": 0.0244, "step": 570 }, { "epoch": 0.99, "learning_rate": 1.5683009265164755e-05, "loss": 0.0232, "step": 571 }, { "epoch": 0.99, "learning_rate": 1.5667649219976773e-05, "loss": 0.0314, "step": 572 }, { "epoch": 0.99, "learning_rate": 1.5652269449725375e-05, "loss": 0.0375, "step": 573 }, { "epoch": 0.99, "learning_rate": 1.563687000793661e-05, "loss": 0.0328, "step": 574 }, { "epoch": 0.99, "learning_rate": 1.5621450948205008e-05, "loss": 0.0285, "step": 575 }, { "epoch": 0.99, "learning_rate": 1.560601232419337e-05, "loss": 0.0219, "step": 576 }, { "epoch": 1.0, "learning_rate": 1.5590554189632585e-05, "loss": 0.0285, "step": 577 }, { "epoch": 1.0, "learning_rate": 1.5575076598321445e-05, "loss": 0.0236, "step": 578 }, { "epoch": 1.0, "learning_rate": 1.5559579604126458e-05, "loss": 0.0203, "step": 579 }, { "epoch": 1.0, "learning_rate": 1.554406326098166e-05, "loss": 0.0244, "step": 580 }, { "epoch": 1.0, "learning_rate": 1.552852762288843e-05, "loss": 0.0228, "step": 581 }, { "epoch": 1.01, "learning_rate": 1.5512972743915287e-05, "loss": 0.0279, "step": 582 }, { "epoch": 1.01, "learning_rate": 1.549739867819773e-05, "loss": 0.0214, "step": 583 }, { "epoch": 1.01, "learning_rate": 1.5481805479938018e-05, "loss": 0.0271, "step": 584 }, { "epoch": 1.01, "learning_rate": 1.5466193203405017e-05, "loss": 0.0222, "step": 585 }, { "epoch": 1.01, "learning_rate": 1.545056190293396e-05, "loss": 0.0233, "step": 586 }, { "epoch": 1.01, "learning_rate": 1.5434911632926322e-05, "loss": 0.0229, "step": 587 }, { "epoch": 1.02, "learning_rate": 1.541924244784958e-05, "loss": 0.0226, "step": 588 }, { "epoch": 1.02, "learning_rate": 1.540355440223704e-05, "loss": 0.024, "step": 589 }, { "epoch": 1.02, "learning_rate": 1.5387847550687656e-05, "loss": 0.0156, "step": 590 }, { "epoch": 1.02, "learning_rate": 1.537212194786582e-05, "loss": 0.0246, "step": 591 }, { "epoch": 1.02, "learning_rate": 1.53563776485012e-05, "loss": 0.0262, "step": 592 }, { "epoch": 1.02, "learning_rate": 1.534061470738852e-05, "loss": 0.0188, "step": 593 }, { "epoch": 1.03, "learning_rate": 1.5324833179387392e-05, "loss": 0.0227, "step": 594 }, { "epoch": 1.03, "learning_rate": 1.5309033119422107e-05, "loss": 0.0228, "step": 595 }, { "epoch": 1.03, "learning_rate": 1.529321458248146e-05, "loss": 0.0248, "step": 596 }, { "epoch": 1.03, "learning_rate": 1.527737762361855e-05, "loss": 0.0303, "step": 597 }, { "epoch": 1.03, "learning_rate": 1.5261522297950578e-05, "loss": 0.0206, "step": 598 }, { "epoch": 1.03, "learning_rate": 1.524564866065869e-05, "loss": 0.0258, "step": 599 }, { "epoch": 1.04, "learning_rate": 1.5229756766987745e-05, "loss": 0.0225, "step": 600 }, { "epoch": 1.04, "learning_rate": 1.5213846672246139e-05, "loss": 0.0226, "step": 601 }, { "epoch": 1.04, "learning_rate": 1.5197918431805618e-05, "loss": 0.0167, "step": 602 }, { "epoch": 1.04, "learning_rate": 1.5181972101101083e-05, "loss": 0.0292, "step": 603 }, { "epoch": 1.04, "learning_rate": 1.5166007735630386e-05, "loss": 0.0246, "step": 604 }, { "epoch": 1.04, "learning_rate": 1.5150025390954153e-05, "loss": 0.0216, "step": 605 }, { "epoch": 1.05, "learning_rate": 1.5134025122695581e-05, "loss": 0.0304, "step": 606 }, { "epoch": 1.05, "learning_rate": 1.511800698654024e-05, "loss": 0.0263, "step": 607 }, { "epoch": 1.05, "learning_rate": 1.5101971038235897e-05, "loss": 0.0285, "step": 608 }, { "epoch": 1.05, "learning_rate": 1.5085917333592299e-05, "loss": 0.0255, "step": 609 }, { "epoch": 1.05, "learning_rate": 1.5069845928480997e-05, "loss": 0.0246, "step": 610 }, { "epoch": 1.06, "learning_rate": 1.5053756878835144e-05, "loss": 0.0207, "step": 611 }, { "epoch": 1.06, "learning_rate": 1.5037650240649297e-05, "loss": 0.0242, "step": 612 }, { "epoch": 1.06, "learning_rate": 1.5021526069979232e-05, "loss": 0.0205, "step": 613 }, { "epoch": 1.06, "learning_rate": 1.500538442294174e-05, "loss": 0.0166, "step": 614 }, { "epoch": 1.06, "learning_rate": 1.4989225355714436e-05, "loss": 0.0256, "step": 615 }, { "epoch": 1.06, "learning_rate": 1.4973048924535562e-05, "loss": 0.0208, "step": 616 }, { "epoch": 1.07, "learning_rate": 1.4956855185703787e-05, "loss": 0.0184, "step": 617 }, { "epoch": 1.07, "learning_rate": 1.4940644195578025e-05, "loss": 0.0198, "step": 618 }, { "epoch": 1.07, "learning_rate": 1.4924416010577218e-05, "loss": 0.0277, "step": 619 }, { "epoch": 1.07, "learning_rate": 1.4908170687180163e-05, "loss": 0.0268, "step": 620 }, { "epoch": 1.07, "learning_rate": 1.48919082819253e-05, "loss": 0.0221, "step": 621 }, { "epoch": 1.07, "learning_rate": 1.4875628851410509e-05, "loss": 0.0199, "step": 622 }, { "epoch": 1.08, "learning_rate": 1.4859332452292937e-05, "loss": 0.0233, "step": 623 }, { "epoch": 1.08, "learning_rate": 1.4843019141288777e-05, "loss": 0.0237, "step": 624 }, { "epoch": 1.08, "learning_rate": 1.4826688975173085e-05, "loss": 0.0164, "step": 625 }, { "epoch": 1.08, "learning_rate": 1.4810342010779578e-05, "loss": 0.0215, "step": 626 }, { "epoch": 1.08, "learning_rate": 1.479397830500043e-05, "loss": 0.0205, "step": 627 }, { "epoch": 1.08, "learning_rate": 1.4777597914786089e-05, "loss": 0.0241, "step": 628 }, { "epoch": 1.09, "learning_rate": 1.4761200897145063e-05, "loss": 0.0313, "step": 629 }, { "epoch": 1.09, "learning_rate": 1.4744787309143727e-05, "loss": 0.0233, "step": 630 }, { "epoch": 1.09, "learning_rate": 1.4728357207906133e-05, "loss": 0.0238, "step": 631 }, { "epoch": 1.09, "learning_rate": 1.4711910650613799e-05, "loss": 0.0179, "step": 632 }, { "epoch": 1.09, "learning_rate": 1.4695447694505512e-05, "loss": 0.0262, "step": 633 }, { "epoch": 1.09, "learning_rate": 1.4678968396877135e-05, "loss": 0.0226, "step": 634 }, { "epoch": 1.1, "learning_rate": 1.4662472815081408e-05, "loss": 0.0238, "step": 635 }, { "epoch": 1.1, "learning_rate": 1.464596100652774e-05, "loss": 0.0204, "step": 636 }, { "epoch": 1.1, "learning_rate": 1.4629433028682014e-05, "loss": 0.0269, "step": 637 }, { "epoch": 1.1, "learning_rate": 1.4612888939066388e-05, "loss": 0.0247, "step": 638 }, { "epoch": 1.1, "learning_rate": 1.459632879525909e-05, "loss": 0.0226, "step": 639 }, { "epoch": 1.11, "learning_rate": 1.4579752654894229e-05, "loss": 0.0257, "step": 640 }, { "epoch": 1.11, "learning_rate": 1.456316057566158e-05, "loss": 0.024, "step": 641 }, { "epoch": 1.11, "learning_rate": 1.4546552615306393e-05, "loss": 0.0135, "step": 642 }, { "epoch": 1.11, "learning_rate": 1.4529928831629185e-05, "loss": 0.0265, "step": 643 }, { "epoch": 1.11, "learning_rate": 1.4513289282485548e-05, "loss": 0.0206, "step": 644 }, { "epoch": 1.11, "learning_rate": 1.4496634025785938e-05, "loss": 0.0189, "step": 645 }, { "epoch": 1.12, "learning_rate": 1.4479963119495485e-05, "loss": 0.0319, "step": 646 }, { "epoch": 1.12, "learning_rate": 1.4463276621633775e-05, "loss": 0.0207, "step": 647 }, { "epoch": 1.12, "learning_rate": 1.444657459027466e-05, "loss": 0.0236, "step": 648 }, { "epoch": 1.12, "learning_rate": 1.4429857083546054e-05, "loss": 0.0216, "step": 649 }, { "epoch": 1.12, "learning_rate": 1.4413124159629734e-05, "loss": 0.0236, "step": 650 }, { "epoch": 1.12, "learning_rate": 1.439637587676113e-05, "loss": 0.0331, "step": 651 }, { "epoch": 1.13, "learning_rate": 1.437961229322912e-05, "loss": 0.021, "step": 652 }, { "epoch": 1.13, "learning_rate": 1.4362833467375839e-05, "loss": 0.0227, "step": 653 }, { "epoch": 1.13, "learning_rate": 1.434603945759647e-05, "loss": 0.0276, "step": 654 }, { "epoch": 1.13, "learning_rate": 1.4329230322339036e-05, "loss": 0.0287, "step": 655 }, { "epoch": 1.13, "learning_rate": 1.431240612010421e-05, "loss": 0.0275, "step": 656 }, { "epoch": 1.13, "learning_rate": 1.429556690944509e-05, "loss": 0.0224, "step": 657 }, { "epoch": 1.14, "learning_rate": 1.4278712748967017e-05, "loss": 0.0213, "step": 658 }, { "epoch": 1.14, "learning_rate": 1.4261843697327355e-05, "loss": 0.0276, "step": 659 }, { "epoch": 1.14, "learning_rate": 1.4244959813235304e-05, "loss": 0.0253, "step": 660 }, { "epoch": 1.14, "learning_rate": 1.4228061155451671e-05, "loss": 0.0237, "step": 661 }, { "epoch": 1.14, "learning_rate": 1.4211147782788695e-05, "loss": 0.0218, "step": 662 }, { "epoch": 1.15, "learning_rate": 1.4194219754109812e-05, "loss": 0.0252, "step": 663 }, { "epoch": 1.15, "learning_rate": 1.4177277128329479e-05, "loss": 0.0195, "step": 664 }, { "epoch": 1.15, "learning_rate": 1.4160319964412943e-05, "loss": 0.0253, "step": 665 }, { "epoch": 1.15, "learning_rate": 1.4143348321376056e-05, "loss": 0.0207, "step": 666 }, { "epoch": 1.15, "learning_rate": 1.412636225828506e-05, "loss": 0.0324, "step": 667 }, { "epoch": 1.15, "learning_rate": 1.4109361834256381e-05, "loss": 0.0294, "step": 668 }, { "epoch": 1.16, "learning_rate": 1.4092347108456425e-05, "loss": 0.0264, "step": 669 }, { "epoch": 1.16, "learning_rate": 1.4075318140101377e-05, "loss": 0.0189, "step": 670 }, { "epoch": 1.16, "learning_rate": 1.4058274988456983e-05, "loss": 0.0244, "step": 671 }, { "epoch": 1.16, "learning_rate": 1.404121771283836e-05, "loss": 0.0221, "step": 672 }, { "epoch": 1.16, "learning_rate": 1.402414637260977e-05, "loss": 0.0276, "step": 673 }, { "epoch": 1.16, "learning_rate": 1.4007061027184432e-05, "loss": 0.025, "step": 674 }, { "epoch": 1.17, "learning_rate": 1.3989961736024302e-05, "loss": 0.0229, "step": 675 }, { "epoch": 1.17, "learning_rate": 1.3972848558639878e-05, "loss": 0.0202, "step": 676 }, { "epoch": 1.17, "learning_rate": 1.3955721554589979e-05, "loss": 0.0211, "step": 677 }, { "epoch": 1.17, "learning_rate": 1.3938580783481545e-05, "loss": 0.0211, "step": 678 }, { "epoch": 1.17, "learning_rate": 1.3921426304969435e-05, "loss": 0.0237, "step": 679 }, { "epoch": 1.17, "learning_rate": 1.3904258178756209e-05, "loss": 0.0183, "step": 680 }, { "epoch": 1.18, "learning_rate": 1.388707646459193e-05, "loss": 0.0263, "step": 681 }, { "epoch": 1.18, "learning_rate": 1.3869881222273945e-05, "loss": 0.0317, "step": 682 }, { "epoch": 1.18, "learning_rate": 1.3852672511646683e-05, "loss": 0.0182, "step": 683 }, { "epoch": 1.18, "learning_rate": 1.3835450392601453e-05, "loss": 0.0177, "step": 684 }, { "epoch": 1.18, "learning_rate": 1.3818214925076226e-05, "loss": 0.0157, "step": 685 }, { "epoch": 1.18, "learning_rate": 1.3800966169055426e-05, "loss": 0.0157, "step": 686 }, { "epoch": 1.19, "learning_rate": 1.3783704184569727e-05, "loss": 0.0228, "step": 687 }, { "epoch": 1.19, "learning_rate": 1.3766429031695844e-05, "loss": 0.0153, "step": 688 }, { "epoch": 1.19, "learning_rate": 1.3749140770556322e-05, "loss": 0.0334, "step": 689 }, { "epoch": 1.19, "learning_rate": 1.373183946131932e-05, "loss": 0.0283, "step": 690 }, { "epoch": 1.19, "learning_rate": 1.3714525164198418e-05, "loss": 0.0223, "step": 691 }, { "epoch": 1.2, "learning_rate": 1.3697197939452392e-05, "loss": 0.0285, "step": 692 }, { "epoch": 1.2, "learning_rate": 1.367985784738501e-05, "loss": 0.0233, "step": 693 }, { "epoch": 1.2, "learning_rate": 1.3662504948344824e-05, "loss": 0.0182, "step": 694 }, { "epoch": 1.2, "learning_rate": 1.364513930272495e-05, "loss": 0.0229, "step": 695 }, { "epoch": 1.2, "learning_rate": 1.362776097096288e-05, "loss": 0.0232, "step": 696 }, { "epoch": 1.2, "learning_rate": 1.361037001354025e-05, "loss": 0.0245, "step": 697 }, { "epoch": 1.21, "learning_rate": 1.3592966490982631e-05, "loss": 0.0182, "step": 698 }, { "epoch": 1.21, "learning_rate": 1.3575550463859332e-05, "loss": 0.0208, "step": 699 }, { "epoch": 1.21, "learning_rate": 1.3558121992783183e-05, "loss": 0.0238, "step": 700 }, { "epoch": 1.21, "learning_rate": 1.3540681138410317e-05, "loss": 0.0159, "step": 701 }, { "epoch": 1.21, "learning_rate": 1.3523227961439965e-05, "loss": 0.014, "step": 702 }, { "epoch": 1.21, "learning_rate": 1.350576252261425e-05, "loss": 0.0218, "step": 703 }, { "epoch": 1.22, "learning_rate": 1.3488284882717961e-05, "loss": 0.022, "step": 704 }, { "epoch": 1.22, "learning_rate": 1.3470795102578358e-05, "loss": 0.0207, "step": 705 }, { "epoch": 1.22, "learning_rate": 1.3453293243064947e-05, "loss": 0.0206, "step": 706 }, { "epoch": 1.22, "learning_rate": 1.3435779365089281e-05, "loss": 0.0203, "step": 707 }, { "epoch": 1.22, "learning_rate": 1.3418253529604733e-05, "loss": 0.0194, "step": 708 }, { "epoch": 1.22, "learning_rate": 1.3400715797606293e-05, "loss": 0.0262, "step": 709 }, { "epoch": 1.23, "learning_rate": 1.3383166230130357e-05, "loss": 0.0202, "step": 710 }, { "epoch": 1.23, "learning_rate": 1.3365604888254511e-05, "loss": 0.0277, "step": 711 }, { "epoch": 1.23, "learning_rate": 1.3348031833097319e-05, "loss": 0.0235, "step": 712 }, { "epoch": 1.23, "learning_rate": 1.3330447125818115e-05, "loss": 0.0217, "step": 713 }, { "epoch": 1.23, "learning_rate": 1.3312850827616776e-05, "loss": 0.0261, "step": 714 }, { "epoch": 1.23, "learning_rate": 1.3295242999733524e-05, "loss": 0.0177, "step": 715 }, { "epoch": 1.24, "learning_rate": 1.3277623703448712e-05, "loss": 0.0211, "step": 716 }, { "epoch": 1.24, "learning_rate": 1.3259993000082599e-05, "loss": 0.0175, "step": 717 }, { "epoch": 1.24, "learning_rate": 1.3242350950995146e-05, "loss": 0.027, "step": 718 }, { "epoch": 1.24, "learning_rate": 1.3224697617585809e-05, "loss": 0.0142, "step": 719 }, { "epoch": 1.24, "learning_rate": 1.3207033061293299e-05, "loss": 0.0206, "step": 720 }, { "epoch": 1.25, "learning_rate": 1.3189357343595405e-05, "loss": 0.0241, "step": 721 }, { "epoch": 1.25, "learning_rate": 1.3171670526008749e-05, "loss": 0.0218, "step": 722 }, { "epoch": 1.25, "learning_rate": 1.3153972670088584e-05, "loss": 0.02, "step": 723 }, { "epoch": 1.25, "learning_rate": 1.3136263837428588e-05, "loss": 0.0278, "step": 724 }, { "epoch": 1.25, "learning_rate": 1.3118544089660635e-05, "loss": 0.0214, "step": 725 }, { "epoch": 1.25, "learning_rate": 1.3100813488454582e-05, "loss": 0.0229, "step": 726 }, { "epoch": 1.26, "learning_rate": 1.3083072095518068e-05, "loss": 0.0169, "step": 727 }, { "epoch": 1.26, "learning_rate": 1.3065319972596288e-05, "loss": 0.0261, "step": 728 }, { "epoch": 1.26, "learning_rate": 1.3047557181471784e-05, "loss": 0.0196, "step": 729 }, { "epoch": 1.26, "learning_rate": 1.3029783783964214e-05, "loss": 0.0249, "step": 730 }, { "epoch": 1.26, "learning_rate": 1.3011999841930159e-05, "loss": 0.0257, "step": 731 }, { "epoch": 1.26, "learning_rate": 1.2994205417262898e-05, "loss": 0.0173, "step": 732 }, { "epoch": 1.27, "learning_rate": 1.2976400571892189e-05, "loss": 0.0145, "step": 733 }, { "epoch": 1.27, "learning_rate": 1.2958585367784063e-05, "loss": 0.0264, "step": 734 }, { "epoch": 1.27, "learning_rate": 1.2940759866940593e-05, "loss": 0.0298, "step": 735 }, { "epoch": 1.27, "learning_rate": 1.29229241313997e-05, "loss": 0.0255, "step": 736 }, { "epoch": 1.27, "learning_rate": 1.2905078223234907e-05, "loss": 0.0189, "step": 737 }, { "epoch": 1.27, "learning_rate": 1.2887222204555165e-05, "loss": 0.0266, "step": 738 }, { "epoch": 1.28, "learning_rate": 1.2869356137504586e-05, "loss": 0.0212, "step": 739 }, { "epoch": 1.28, "learning_rate": 1.2851480084262273e-05, "loss": 0.0256, "step": 740 }, { "epoch": 1.28, "learning_rate": 1.2833594107042078e-05, "loss": 0.0226, "step": 741 }, { "epoch": 1.28, "learning_rate": 1.2815698268092387e-05, "loss": 0.0267, "step": 742 }, { "epoch": 1.28, "learning_rate": 1.2797792629695909e-05, "loss": 0.0265, "step": 743 }, { "epoch": 1.28, "learning_rate": 1.2779877254169466e-05, "loss": 0.0285, "step": 744 }, { "epoch": 1.29, "learning_rate": 1.2761952203863759e-05, "loss": 0.0196, "step": 745 }, { "epoch": 1.29, "learning_rate": 1.274401754116316e-05, "loss": 0.0273, "step": 746 }, { "epoch": 1.29, "learning_rate": 1.2726073328485502e-05, "loss": 0.0173, "step": 747 }, { "epoch": 1.29, "learning_rate": 1.270811962828185e-05, "loss": 0.0339, "step": 748 }, { "epoch": 1.29, "learning_rate": 1.2690156503036288e-05, "loss": 0.03, "step": 749 }, { "epoch": 1.3, "learning_rate": 1.2672184015265702e-05, "loss": 0.0188, "step": 750 }, { "epoch": 1.3, "learning_rate": 1.2654202227519565e-05, "loss": 0.0225, "step": 751 }, { "epoch": 1.3, "learning_rate": 1.2636211202379713e-05, "loss": 0.0186, "step": 752 }, { "epoch": 1.3, "learning_rate": 1.2618211002460135e-05, "loss": 0.0225, "step": 753 }, { "epoch": 1.3, "learning_rate": 1.2600201690406746e-05, "loss": 0.019, "step": 754 }, { "epoch": 1.3, "learning_rate": 1.2582183328897177e-05, "loss": 0.0259, "step": 755 }, { "epoch": 1.31, "learning_rate": 1.2564155980640552e-05, "loss": 0.0211, "step": 756 }, { "epoch": 1.31, "learning_rate": 1.2546119708377273e-05, "loss": 0.02, "step": 757 }, { "epoch": 1.31, "learning_rate": 1.2528074574878803e-05, "loss": 0.0257, "step": 758 }, { "epoch": 1.31, "learning_rate": 1.251002064294744e-05, "loss": 0.0182, "step": 759 }, { "epoch": 1.31, "learning_rate": 1.24919579754161e-05, "loss": 0.0188, "step": 760 }, { "epoch": 1.31, "learning_rate": 1.2473886635148109e-05, "loss": 0.0278, "step": 761 }, { "epoch": 1.32, "learning_rate": 1.2455806685036973e-05, "loss": 0.0297, "step": 762 }, { "epoch": 1.32, "learning_rate": 1.2437718188006165e-05, "loss": 0.0224, "step": 763 }, { "epoch": 1.32, "learning_rate": 1.2419621207008902e-05, "loss": 0.0191, "step": 764 }, { "epoch": 1.32, "learning_rate": 1.2401515805027924e-05, "loss": 0.0176, "step": 765 }, { "epoch": 1.32, "learning_rate": 1.2383402045075283e-05, "loss": 0.0211, "step": 766 }, { "epoch": 1.32, "learning_rate": 1.2365279990192123e-05, "loss": 0.0201, "step": 767 }, { "epoch": 1.33, "learning_rate": 1.2347149703448447e-05, "loss": 0.0189, "step": 768 }, { "epoch": 1.33, "learning_rate": 1.2329011247942915e-05, "loss": 0.0216, "step": 769 }, { "epoch": 1.33, "learning_rate": 1.2310864686802616e-05, "loss": 0.0269, "step": 770 }, { "epoch": 1.33, "learning_rate": 1.2292710083182837e-05, "loss": 0.0257, "step": 771 }, { "epoch": 1.33, "learning_rate": 1.2274547500266875e-05, "loss": 0.0266, "step": 772 }, { "epoch": 1.34, "learning_rate": 1.2256377001265785e-05, "loss": 0.0182, "step": 773 }, { "epoch": 1.34, "learning_rate": 1.2238198649418174e-05, "loss": 0.0207, "step": 774 }, { "epoch": 1.34, "learning_rate": 1.222001250798998e-05, "loss": 0.0208, "step": 775 }, { "epoch": 1.34, "learning_rate": 1.220181864027425e-05, "loss": 0.0246, "step": 776 }, { "epoch": 1.34, "learning_rate": 1.2183617109590925e-05, "loss": 0.0216, "step": 777 }, { "epoch": 1.34, "learning_rate": 1.2165407979286603e-05, "loss": 0.0204, "step": 778 }, { "epoch": 1.35, "learning_rate": 1.214719131273435e-05, "loss": 0.0238, "step": 779 }, { "epoch": 1.35, "learning_rate": 1.2128967173333448e-05, "loss": 0.0261, "step": 780 }, { "epoch": 1.35, "learning_rate": 1.2110735624509184e-05, "loss": 0.0238, "step": 781 }, { "epoch": 1.35, "learning_rate": 1.2092496729712639e-05, "loss": 0.0423, "step": 782 }, { "epoch": 1.35, "learning_rate": 1.2074250552420459e-05, "loss": 0.0213, "step": 783 }, { "epoch": 1.35, "learning_rate": 1.2055997156134633e-05, "loss": 0.0146, "step": 784 }, { "epoch": 1.36, "learning_rate": 1.2037736604382279e-05, "loss": 0.0219, "step": 785 }, { "epoch": 1.36, "learning_rate": 1.2019468960715407e-05, "loss": 0.0185, "step": 786 }, { "epoch": 1.36, "learning_rate": 1.2001194288710722e-05, "loss": 0.0189, "step": 787 }, { "epoch": 1.36, "learning_rate": 1.198291265196938e-05, "loss": 0.0289, "step": 788 }, { "epoch": 1.36, "learning_rate": 1.1964624114116784e-05, "loss": 0.0202, "step": 789 }, { "epoch": 1.36, "learning_rate": 1.1946328738802353e-05, "loss": 0.0216, "step": 790 }, { "epoch": 1.37, "learning_rate": 1.1928026589699292e-05, "loss": 0.0222, "step": 791 }, { "epoch": 1.37, "learning_rate": 1.1909717730504396e-05, "loss": 0.0176, "step": 792 }, { "epoch": 1.37, "learning_rate": 1.1891402224937805e-05, "loss": 0.0205, "step": 793 }, { "epoch": 1.37, "learning_rate": 1.187308013674279e-05, "loss": 0.026, "step": 794 }, { "epoch": 1.37, "learning_rate": 1.1854751529685533e-05, "loss": 0.0316, "step": 795 }, { "epoch": 1.37, "learning_rate": 1.1836416467554905e-05, "loss": 0.0214, "step": 796 }, { "epoch": 1.38, "learning_rate": 1.1818075014162243e-05, "loss": 0.0179, "step": 797 }, { "epoch": 1.38, "learning_rate": 1.1799727233341121e-05, "loss": 0.0294, "step": 798 }, { "epoch": 1.38, "learning_rate": 1.1781373188947141e-05, "loss": 0.0263, "step": 799 }, { "epoch": 1.38, "learning_rate": 1.1763012944857699e-05, "loss": 0.0156, "step": 800 }, { "epoch": 1.38, "learning_rate": 1.1744646564971777e-05, "loss": 0.0196, "step": 801 }, { "epoch": 1.39, "learning_rate": 1.17262741132097e-05, "loss": 0.0168, "step": 802 }, { "epoch": 1.39, "learning_rate": 1.170789565351293e-05, "loss": 0.024, "step": 803 }, { "epoch": 1.39, "learning_rate": 1.1689511249843839e-05, "loss": 0.0269, "step": 804 }, { "epoch": 1.39, "learning_rate": 1.1671120966185486e-05, "loss": 0.0245, "step": 805 }, { "epoch": 1.39, "learning_rate": 1.1652724866541393e-05, "loss": 0.0206, "step": 806 }, { "epoch": 1.39, "learning_rate": 1.1634323014935323e-05, "loss": 0.0242, "step": 807 }, { "epoch": 1.4, "learning_rate": 1.1615915475411054e-05, "loss": 0.018, "step": 808 }, { "epoch": 1.4, "learning_rate": 1.159750231203217e-05, "loss": 0.0316, "step": 809 }, { "epoch": 1.4, "learning_rate": 1.1579083588881814e-05, "loss": 0.0182, "step": 810 }, { "epoch": 1.4, "learning_rate": 1.1560659370062493e-05, "loss": 0.0203, "step": 811 }, { "epoch": 1.4, "learning_rate": 1.1542229719695828e-05, "loss": 0.0206, "step": 812 }, { "epoch": 1.4, "learning_rate": 1.1523794701922351e-05, "loss": 0.0203, "step": 813 }, { "epoch": 1.41, "learning_rate": 1.150535438090127e-05, "loss": 0.0197, "step": 814 }, { "epoch": 1.41, "learning_rate": 1.1486908820810253e-05, "loss": 0.0221, "step": 815 }, { "epoch": 1.41, "learning_rate": 1.14684580858452e-05, "loss": 0.0227, "step": 816 }, { "epoch": 1.41, "learning_rate": 1.145000224022002e-05, "loss": 0.0193, "step": 817 }, { "epoch": 1.41, "learning_rate": 1.143154134816641e-05, "loss": 0.024, "step": 818 }, { "epoch": 1.41, "learning_rate": 1.1413075473933629e-05, "loss": 0.0204, "step": 819 }, { "epoch": 1.42, "learning_rate": 1.1394604681788276e-05, "loss": 0.0209, "step": 820 }, { "epoch": 1.42, "learning_rate": 1.1376129036014073e-05, "loss": 0.0273, "step": 821 }, { "epoch": 1.42, "learning_rate": 1.1357648600911615e-05, "loss": 0.0145, "step": 822 }, { "epoch": 1.42, "learning_rate": 1.1339163440798187e-05, "loss": 0.0181, "step": 823 }, { "epoch": 1.42, "learning_rate": 1.13206736200075e-05, "loss": 0.0194, "step": 824 }, { "epoch": 1.42, "learning_rate": 1.1302179202889505e-05, "loss": 0.0232, "step": 825 }, { "epoch": 1.43, "learning_rate": 1.1283680253810128e-05, "loss": 0.0315, "step": 826 }, { "epoch": 1.43, "learning_rate": 1.1265176837151084e-05, "loss": 0.0248, "step": 827 }, { "epoch": 1.43, "learning_rate": 1.1246669017309629e-05, "loss": 0.0164, "step": 828 }, { "epoch": 1.43, "learning_rate": 1.1228156858698344e-05, "loss": 0.0134, "step": 829 }, { "epoch": 1.43, "learning_rate": 1.1209640425744912e-05, "loss": 0.0298, "step": 830 }, { "epoch": 1.44, "learning_rate": 1.1191119782891893e-05, "loss": 0.0212, "step": 831 }, { "epoch": 1.44, "learning_rate": 1.1172594994596497e-05, "loss": 0.0246, "step": 832 }, { "epoch": 1.44, "learning_rate": 1.1154066125330358e-05, "loss": 0.0242, "step": 833 }, { "epoch": 1.44, "learning_rate": 1.1135533239579318e-05, "loss": 0.0255, "step": 834 }, { "epoch": 1.44, "learning_rate": 1.1116996401843194e-05, "loss": 0.0202, "step": 835 }, { "epoch": 1.44, "learning_rate": 1.109845567663556e-05, "loss": 0.0204, "step": 836 }, { "epoch": 1.45, "learning_rate": 1.107991112848352e-05, "loss": 0.0187, "step": 837 }, { "epoch": 1.45, "learning_rate": 1.106136282192748e-05, "loss": 0.0223, "step": 838 }, { "epoch": 1.45, "learning_rate": 1.1042810821520924e-05, "loss": 0.0217, "step": 839 }, { "epoch": 1.45, "learning_rate": 1.10242551918302e-05, "loss": 0.0212, "step": 840 }, { "epoch": 1.45, "learning_rate": 1.100569599743428e-05, "loss": 0.0199, "step": 841 }, { "epoch": 1.45, "learning_rate": 1.0987133302924546e-05, "loss": 0.0188, "step": 842 }, { "epoch": 1.46, "learning_rate": 1.0968567172904558e-05, "loss": 0.024, "step": 843 }, { "epoch": 1.46, "learning_rate": 1.0949997671989833e-05, "loss": 0.0176, "step": 844 }, { "epoch": 1.46, "learning_rate": 1.0931424864807624e-05, "loss": 0.0238, "step": 845 }, { "epoch": 1.46, "learning_rate": 1.091284881599669e-05, "loss": 0.0174, "step": 846 }, { "epoch": 1.46, "learning_rate": 1.0894269590207062e-05, "loss": 0.0182, "step": 847 }, { "epoch": 1.46, "learning_rate": 1.0875687252099842e-05, "loss": 0.0195, "step": 848 }, { "epoch": 1.47, "learning_rate": 1.0857101866346953e-05, "loss": 0.0208, "step": 849 }, { "epoch": 1.47, "learning_rate": 1.0838513497630932e-05, "loss": 0.0247, "step": 850 }, { "epoch": 1.47, "learning_rate": 1.0819922210644695e-05, "loss": 0.0268, "step": 851 }, { "epoch": 1.47, "learning_rate": 1.080132807009131e-05, "loss": 0.0172, "step": 852 }, { "epoch": 1.47, "learning_rate": 1.0782731140683786e-05, "loss": 0.0233, "step": 853 }, { "epoch": 1.47, "learning_rate": 1.0764131487144824e-05, "loss": 0.0193, "step": 854 }, { "epoch": 1.48, "learning_rate": 1.0745529174206618e-05, "loss": 0.0246, "step": 855 }, { "epoch": 1.48, "learning_rate": 1.0726924266610616e-05, "loss": 0.0321, "step": 856 }, { "epoch": 1.48, "learning_rate": 1.0708316829107295e-05, "loss": 0.0229, "step": 857 }, { "epoch": 1.48, "learning_rate": 1.0689706926455927e-05, "loss": 0.0238, "step": 858 }, { "epoch": 1.48, "learning_rate": 1.0671094623424376e-05, "loss": 0.0195, "step": 859 }, { "epoch": 1.49, "learning_rate": 1.0652479984788857e-05, "loss": 0.0198, "step": 860 }, { "epoch": 1.49, "learning_rate": 1.0633863075333713e-05, "loss": 0.0235, "step": 861 }, { "epoch": 1.49, "learning_rate": 1.0615243959851189e-05, "loss": 0.0262, "step": 862 }, { "epoch": 1.49, "learning_rate": 1.0596622703141209e-05, "loss": 0.0234, "step": 863 }, { "epoch": 1.49, "learning_rate": 1.0577999370011142e-05, "loss": 0.0182, "step": 864 }, { "epoch": 1.49, "learning_rate": 1.0559374025275597e-05, "loss": 0.0209, "step": 865 }, { "epoch": 1.5, "learning_rate": 1.0540746733756173e-05, "loss": 0.0209, "step": 866 }, { "epoch": 1.5, "learning_rate": 1.0522117560281251e-05, "loss": 0.0151, "step": 867 }, { "epoch": 1.5, "learning_rate": 1.0503486569685756e-05, "loss": 0.0176, "step": 868 }, { "epoch": 1.5, "learning_rate": 1.0484853826810942e-05, "loss": 0.0215, "step": 869 }, { "epoch": 1.5, "learning_rate": 1.0466219396504158e-05, "loss": 0.0188, "step": 870 }, { "epoch": 1.5, "learning_rate": 1.0447583343618629e-05, "loss": 0.0192, "step": 871 }, { "epoch": 1.51, "learning_rate": 1.0428945733013223e-05, "loss": 0.0163, "step": 872 }, { "epoch": 1.51, "learning_rate": 1.0410306629552231e-05, "loss": 0.0209, "step": 873 }, { "epoch": 1.51, "learning_rate": 1.0391666098105143e-05, "loss": 0.0178, "step": 874 }, { "epoch": 1.51, "learning_rate": 1.0373024203546413e-05, "loss": 0.0139, "step": 875 }, { "epoch": 1.51, "learning_rate": 1.0354381010755245e-05, "loss": 0.0157, "step": 876 }, { "epoch": 1.51, "learning_rate": 1.0335736584615357e-05, "loss": 0.0165, "step": 877 }, { "epoch": 1.52, "learning_rate": 1.0317090990014759e-05, "loss": 0.0215, "step": 878 }, { "epoch": 1.52, "learning_rate": 1.029844429184553e-05, "loss": 0.0182, "step": 879 }, { "epoch": 1.52, "learning_rate": 1.0279796555003589e-05, "loss": 0.0201, "step": 880 }, { "epoch": 1.52, "learning_rate": 1.0261147844388472e-05, "loss": 0.0159, "step": 881 }, { "epoch": 1.52, "learning_rate": 1.0242498224903099e-05, "loss": 0.0243, "step": 882 }, { "epoch": 1.53, "learning_rate": 1.0223847761453558e-05, "loss": 0.0285, "step": 883 }, { "epoch": 1.53, "learning_rate": 1.0205196518948868e-05, "loss": 0.0211, "step": 884 }, { "epoch": 1.53, "learning_rate": 1.0186544562300766e-05, "loss": 0.0212, "step": 885 }, { "epoch": 1.53, "learning_rate": 1.0167891956423473e-05, "loss": 0.0178, "step": 886 }, { "epoch": 1.53, "learning_rate": 1.0149238766233466e-05, "loss": 0.0225, "step": 887 }, { "epoch": 1.53, "learning_rate": 1.0130585056649262e-05, "loss": 0.0206, "step": 888 }, { "epoch": 1.54, "learning_rate": 1.011193089259118e-05, "loss": 0.0269, "step": 889 }, { "epoch": 1.54, "learning_rate": 1.0093276338981116e-05, "loss": 0.017, "step": 890 }, { "epoch": 1.54, "learning_rate": 1.007462146074234e-05, "loss": 0.0138, "step": 891 }, { "epoch": 1.54, "learning_rate": 1.0055966322799228e-05, "loss": 0.0201, "step": 892 }, { "epoch": 1.54, "learning_rate": 1.0037310990077083e-05, "loss": 0.0137, "step": 893 }, { "epoch": 1.54, "learning_rate": 1.0018655527501867e-05, "loss": 0.0207, "step": 894 }, { "epoch": 1.55, "learning_rate": 1e-05, "loss": 0.0197, "step": 895 }, { "epoch": 1.55, "learning_rate": 9.981344472498138e-06, "loss": 0.02, "step": 896 }, { "epoch": 1.55, "learning_rate": 9.962689009922918e-06, "loss": 0.02, "step": 897 }, { "epoch": 1.55, "learning_rate": 9.944033677200773e-06, "loss": 0.0286, "step": 898 }, { "epoch": 1.55, "learning_rate": 9.925378539257664e-06, "loss": 0.0185, "step": 899 }, { "epoch": 1.55, "learning_rate": 9.906723661018887e-06, "loss": 0.0189, "step": 900 }, { "epoch": 1.56, "learning_rate": 9.888069107408824e-06, "loss": 0.0246, "step": 901 }, { "epoch": 1.56, "learning_rate": 9.869414943350743e-06, "loss": 0.0267, "step": 902 }, { "epoch": 1.56, "learning_rate": 9.850761233766537e-06, "loss": 0.0184, "step": 903 }, { "epoch": 1.56, "learning_rate": 9.832108043576532e-06, "loss": 0.0263, "step": 904 }, { "epoch": 1.56, "learning_rate": 9.813455437699238e-06, "loss": 0.0278, "step": 905 }, { "epoch": 1.56, "learning_rate": 9.794803481051137e-06, "loss": 0.0273, "step": 906 }, { "epoch": 1.57, "learning_rate": 9.776152238546447e-06, "loss": 0.0223, "step": 907 }, { "epoch": 1.57, "learning_rate": 9.757501775096904e-06, "loss": 0.0198, "step": 908 }, { "epoch": 1.57, "learning_rate": 9.738852155611531e-06, "loss": 0.0222, "step": 909 }, { "epoch": 1.57, "learning_rate": 9.720203444996413e-06, "loss": 0.0223, "step": 910 }, { "epoch": 1.57, "learning_rate": 9.70155570815447e-06, "loss": 0.0239, "step": 911 }, { "epoch": 1.58, "learning_rate": 9.682909009985243e-06, "loss": 0.0154, "step": 912 }, { "epoch": 1.58, "learning_rate": 9.664263415384644e-06, "loss": 0.0183, "step": 913 }, { "epoch": 1.58, "learning_rate": 9.645618989244758e-06, "loss": 0.0176, "step": 914 }, { "epoch": 1.58, "learning_rate": 9.626975796453587e-06, "loss": 0.024, "step": 915 }, { "epoch": 1.58, "learning_rate": 9.608333901894859e-06, "loss": 0.0155, "step": 916 }, { "epoch": 1.58, "learning_rate": 9.589693370447769e-06, "loss": 0.0204, "step": 917 }, { "epoch": 1.59, "learning_rate": 9.57105426698678e-06, "loss": 0.0247, "step": 918 }, { "epoch": 1.59, "learning_rate": 9.552416656381373e-06, "loss": 0.0246, "step": 919 }, { "epoch": 1.59, "learning_rate": 9.533780603495846e-06, "loss": 0.0178, "step": 920 }, { "epoch": 1.59, "learning_rate": 9.515146173189058e-06, "loss": 0.0235, "step": 921 }, { "epoch": 1.59, "learning_rate": 9.496513430314245e-06, "loss": 0.0157, "step": 922 }, { "epoch": 1.59, "learning_rate": 9.47788243971875e-06, "loss": 0.0233, "step": 923 }, { "epoch": 1.6, "learning_rate": 9.45925326624383e-06, "loss": 0.0189, "step": 924 }, { "epoch": 1.6, "learning_rate": 9.440625974724408e-06, "loss": 0.0202, "step": 925 }, { "epoch": 1.6, "learning_rate": 9.422000629988861e-06, "loss": 0.0277, "step": 926 }, { "epoch": 1.6, "learning_rate": 9.403377296858798e-06, "loss": 0.0181, "step": 927 }, { "epoch": 1.6, "learning_rate": 9.384756040148813e-06, "loss": 0.0234, "step": 928 }, { "epoch": 1.6, "learning_rate": 9.366136924666289e-06, "loss": 0.0245, "step": 929 }, { "epoch": 1.61, "learning_rate": 9.347520015211144e-06, "loss": 0.0205, "step": 930 }, { "epoch": 1.61, "learning_rate": 9.328905376575629e-06, "loss": 0.0184, "step": 931 }, { "epoch": 1.61, "learning_rate": 9.310293073544077e-06, "loss": 0.0214, "step": 932 }, { "epoch": 1.61, "learning_rate": 9.291683170892712e-06, "loss": 0.0291, "step": 933 }, { "epoch": 1.61, "learning_rate": 9.273075733389385e-06, "loss": 0.0176, "step": 934 }, { "epoch": 1.61, "learning_rate": 9.254470825793385e-06, "loss": 0.0242, "step": 935 }, { "epoch": 1.62, "learning_rate": 9.235868512855178e-06, "loss": 0.0238, "step": 936 }, { "epoch": 1.62, "learning_rate": 9.217268859316219e-06, "loss": 0.0194, "step": 937 }, { "epoch": 1.62, "learning_rate": 9.19867192990869e-06, "loss": 0.0185, "step": 938 }, { "epoch": 1.62, "learning_rate": 9.180077789355307e-06, "loss": 0.0182, "step": 939 }, { "epoch": 1.62, "learning_rate": 9.161486502369068e-06, "loss": 0.0145, "step": 940 }, { "epoch": 1.63, "learning_rate": 9.142898133653049e-06, "loss": 0.0206, "step": 941 }, { "epoch": 1.63, "learning_rate": 9.12431274790016e-06, "loss": 0.0178, "step": 942 }, { "epoch": 1.63, "learning_rate": 9.10573040979294e-06, "loss": 0.0239, "step": 943 }, { "epoch": 1.63, "learning_rate": 9.087151184003313e-06, "loss": 0.0165, "step": 944 }, { "epoch": 1.63, "learning_rate": 9.068575135192377e-06, "loss": 0.0268, "step": 945 }, { "epoch": 1.63, "learning_rate": 9.050002328010165e-06, "loss": 0.0263, "step": 946 }, { "epoch": 1.64, "learning_rate": 9.031432827095444e-06, "loss": 0.0265, "step": 947 }, { "epoch": 1.64, "learning_rate": 9.012866697075456e-06, "loss": 0.0214, "step": 948 }, { "epoch": 1.64, "learning_rate": 8.994304002565723e-06, "loss": 0.0224, "step": 949 }, { "epoch": 1.64, "learning_rate": 8.9757448081698e-06, "loss": 0.0162, "step": 950 }, { "epoch": 1.64, "learning_rate": 8.957189178479077e-06, "loss": 0.0239, "step": 951 }, { "epoch": 1.64, "learning_rate": 8.938637178072525e-06, "loss": 0.0204, "step": 952 }, { "epoch": 1.65, "learning_rate": 8.920088871516482e-06, "loss": 0.0242, "step": 953 }, { "epoch": 1.65, "learning_rate": 8.901544323364443e-06, "loss": 0.0258, "step": 954 }, { "epoch": 1.65, "learning_rate": 8.88300359815681e-06, "loss": 0.0128, "step": 955 }, { "epoch": 1.65, "learning_rate": 8.864466760420688e-06, "loss": 0.0265, "step": 956 }, { "epoch": 1.65, "learning_rate": 8.845933874669645e-06, "loss": 0.0201, "step": 957 }, { "epoch": 1.65, "learning_rate": 8.827405005403508e-06, "loss": 0.0179, "step": 958 }, { "epoch": 1.66, "learning_rate": 8.80888021710811e-06, "loss": 0.0292, "step": 959 }, { "epoch": 1.66, "learning_rate": 8.790359574255091e-06, "loss": 0.0222, "step": 960 }, { "epoch": 1.66, "learning_rate": 8.771843141301659e-06, "loss": 0.0178, "step": 961 }, { "epoch": 1.66, "learning_rate": 8.753330982690376e-06, "loss": 0.0191, "step": 962 }, { "epoch": 1.66, "learning_rate": 8.734823162848919e-06, "loss": 0.0208, "step": 963 }, { "epoch": 1.66, "learning_rate": 8.716319746189877e-06, "loss": 0.0212, "step": 964 }, { "epoch": 1.67, "learning_rate": 8.697820797110499e-06, "loss": 0.0249, "step": 965 }, { "epoch": 1.67, "learning_rate": 8.679326379992502e-06, "loss": 0.0261, "step": 966 }, { "epoch": 1.67, "learning_rate": 8.660836559201814e-06, "loss": 0.0179, "step": 967 }, { "epoch": 1.67, "learning_rate": 8.642351399088387e-06, "loss": 0.014, "step": 968 }, { "epoch": 1.67, "learning_rate": 8.62387096398593e-06, "loss": 0.0186, "step": 969 }, { "epoch": 1.68, "learning_rate": 8.605395318211725e-06, "loss": 0.0199, "step": 970 }, { "epoch": 1.68, "learning_rate": 8.586924526066371e-06, "loss": 0.0219, "step": 971 }, { "epoch": 1.68, "learning_rate": 8.568458651833592e-06, "loss": 0.0177, "step": 972 }, { "epoch": 1.68, "learning_rate": 8.549997759779981e-06, "loss": 0.0208, "step": 973 }, { "epoch": 1.68, "learning_rate": 8.531541914154803e-06, "loss": 0.0169, "step": 974 }, { "epoch": 1.68, "learning_rate": 8.513091179189749e-06, "loss": 0.0228, "step": 975 }, { "epoch": 1.69, "learning_rate": 8.494645619098734e-06, "loss": 0.0213, "step": 976 }, { "epoch": 1.69, "learning_rate": 8.47620529807765e-06, "loss": 0.0188, "step": 977 }, { "epoch": 1.69, "learning_rate": 8.457770280304173e-06, "loss": 0.0221, "step": 978 }, { "epoch": 1.69, "learning_rate": 8.439340629937508e-06, "loss": 0.0158, "step": 979 }, { "epoch": 1.69, "learning_rate": 8.420916411118188e-06, "loss": 0.022, "step": 980 }, { "epoch": 1.69, "learning_rate": 8.402497687967837e-06, "loss": 0.0199, "step": 981 }, { "epoch": 1.7, "learning_rate": 8.384084524588949e-06, "loss": 0.0243, "step": 982 }, { "epoch": 1.7, "learning_rate": 8.365676985064684e-06, "loss": 0.0255, "step": 983 }, { "epoch": 1.7, "learning_rate": 8.347275133458608e-06, "loss": 0.0207, "step": 984 }, { "epoch": 1.7, "learning_rate": 8.328879033814516e-06, "loss": 0.0179, "step": 985 }, { "epoch": 1.7, "learning_rate": 8.310488750156164e-06, "loss": 0.0221, "step": 986 }, { "epoch": 1.7, "learning_rate": 8.292104346487076e-06, "loss": 0.0247, "step": 987 }, { "epoch": 1.71, "learning_rate": 8.273725886790302e-06, "loss": 0.0211, "step": 988 }, { "epoch": 1.71, "learning_rate": 8.255353435028228e-06, "loss": 0.0171, "step": 989 }, { "epoch": 1.71, "learning_rate": 8.236987055142303e-06, "loss": 0.0286, "step": 990 }, { "epoch": 1.71, "learning_rate": 8.218626811052865e-06, "loss": 0.0185, "step": 991 }, { "epoch": 1.71, "learning_rate": 8.200272766658882e-06, "loss": 0.0294, "step": 992 }, { "epoch": 1.72, "learning_rate": 8.181924985837762e-06, "loss": 0.0135, "step": 993 }, { "epoch": 1.72, "learning_rate": 8.163583532445095e-06, "loss": 0.0298, "step": 994 }, { "epoch": 1.72, "learning_rate": 8.145248470314469e-06, "loss": 0.0265, "step": 995 }, { "epoch": 1.72, "learning_rate": 8.126919863257212e-06, "loss": 0.0134, "step": 996 }, { "epoch": 1.72, "learning_rate": 8.108597775062198e-06, "loss": 0.0221, "step": 997 }, { "epoch": 1.72, "learning_rate": 8.090282269495604e-06, "loss": 0.0171, "step": 998 }, { "epoch": 1.73, "learning_rate": 8.07197341030071e-06, "loss": 0.018, "step": 999 }, { "epoch": 1.73, "learning_rate": 8.05367126119765e-06, "loss": 0.0237, "step": 1000 }, { "epoch": 1.73, "learning_rate": 8.035375885883217e-06, "loss": 0.0151, "step": 1001 }, { "epoch": 1.73, "learning_rate": 8.01708734803062e-06, "loss": 0.018, "step": 1002 }, { "epoch": 1.73, "learning_rate": 7.998805711289281e-06, "loss": 0.0238, "step": 1003 }, { "epoch": 1.73, "learning_rate": 7.980531039284593e-06, "loss": 0.0198, "step": 1004 }, { "epoch": 1.74, "learning_rate": 7.962263395617724e-06, "loss": 0.0231, "step": 1005 }, { "epoch": 1.74, "learning_rate": 7.944002843865367e-06, "loss": 0.0198, "step": 1006 }, { "epoch": 1.74, "learning_rate": 7.925749447579543e-06, "loss": 0.0294, "step": 1007 }, { "epoch": 1.74, "learning_rate": 7.907503270287365e-06, "loss": 0.0198, "step": 1008 }, { "epoch": 1.74, "learning_rate": 7.88926437549082e-06, "loss": 0.022, "step": 1009 }, { "epoch": 1.74, "learning_rate": 7.871032826666557e-06, "loss": 0.0158, "step": 1010 }, { "epoch": 1.75, "learning_rate": 7.852808687265653e-06, "loss": 0.0134, "step": 1011 }, { "epoch": 1.75, "learning_rate": 7.8345920207134e-06, "loss": 0.0167, "step": 1012 }, { "epoch": 1.75, "learning_rate": 7.81638289040908e-06, "loss": 0.0217, "step": 1013 }, { "epoch": 1.75, "learning_rate": 7.798181359725754e-06, "loss": 0.019, "step": 1014 }, { "epoch": 1.75, "learning_rate": 7.779987492010022e-06, "loss": 0.0286, "step": 1015 }, { "epoch": 1.75, "learning_rate": 7.76180135058183e-06, "loss": 0.0287, "step": 1016 }, { "epoch": 1.76, "learning_rate": 7.743622998734217e-06, "loss": 0.0228, "step": 1017 }, { "epoch": 1.76, "learning_rate": 7.72545249973313e-06, "loss": 0.0186, "step": 1018 }, { "epoch": 1.76, "learning_rate": 7.707289916817164e-06, "loss": 0.0192, "step": 1019 }, { "epoch": 1.76, "learning_rate": 7.689135313197391e-06, "loss": 0.0134, "step": 1020 }, { "epoch": 1.76, "learning_rate": 7.670988752057088e-06, "loss": 0.0217, "step": 1021 }, { "epoch": 1.77, "learning_rate": 7.652850296551557e-06, "loss": 0.0207, "step": 1022 }, { "epoch": 1.77, "learning_rate": 7.634720009807879e-06, "loss": 0.0203, "step": 1023 }, { "epoch": 1.77, "learning_rate": 7.616597954924719e-06, "loss": 0.0199, "step": 1024 }, { "epoch": 1.77, "learning_rate": 7.598484194972076e-06, "loss": 0.0197, "step": 1025 }, { "epoch": 1.77, "learning_rate": 7.580378792991101e-06, "loss": 0.0159, "step": 1026 }, { "epoch": 1.77, "learning_rate": 7.562281811993835e-06, "loss": 0.017, "step": 1027 }, { "epoch": 1.78, "learning_rate": 7.544193314963029e-06, "loss": 0.0151, "step": 1028 }, { "epoch": 1.78, "learning_rate": 7.526113364851891e-06, "loss": 0.019, "step": 1029 }, { "epoch": 1.78, "learning_rate": 7.508042024583903e-06, "loss": 0.0195, "step": 1030 }, { "epoch": 1.78, "learning_rate": 7.489979357052563e-06, "loss": 0.0223, "step": 1031 }, { "epoch": 1.78, "learning_rate": 7.471925425121199e-06, "loss": 0.0208, "step": 1032 }, { "epoch": 1.78, "learning_rate": 7.453880291622726e-06, "loss": 0.021, "step": 1033 }, { "epoch": 1.79, "learning_rate": 7.435844019359451e-06, "loss": 0.0277, "step": 1034 }, { "epoch": 1.79, "learning_rate": 7.417816671102824e-06, "loss": 0.0195, "step": 1035 }, { "epoch": 1.79, "learning_rate": 7.3997983095932555e-06, "loss": 0.0181, "step": 1036 }, { "epoch": 1.79, "learning_rate": 7.381788997539869e-06, "loss": 0.0208, "step": 1037 }, { "epoch": 1.79, "learning_rate": 7.3637887976202895e-06, "loss": 0.0183, "step": 1038 }, { "epoch": 1.79, "learning_rate": 7.34579777248044e-06, "loss": 0.0227, "step": 1039 }, { "epoch": 1.8, "learning_rate": 7.3278159847343e-06, "loss": 0.0201, "step": 1040 }, { "epoch": 1.8, "learning_rate": 7.309843496963715e-06, "loss": 0.0195, "step": 1041 }, { "epoch": 1.8, "learning_rate": 7.291880371718154e-06, "loss": 0.0222, "step": 1042 }, { "epoch": 1.8, "learning_rate": 7.273926671514503e-06, "loss": 0.0206, "step": 1043 }, { "epoch": 1.8, "learning_rate": 7.2559824588368425e-06, "loss": 0.0211, "step": 1044 }, { "epoch": 1.8, "learning_rate": 7.238047796136247e-06, "loss": 0.0142, "step": 1045 }, { "epoch": 1.81, "learning_rate": 7.220122745830536e-06, "loss": 0.0168, "step": 1046 }, { "epoch": 1.81, "learning_rate": 7.202207370304093e-06, "loss": 0.022, "step": 1047 }, { "epoch": 1.81, "learning_rate": 7.1843017319076166e-06, "loss": 0.018, "step": 1048 }, { "epoch": 1.81, "learning_rate": 7.166405892957926e-06, "loss": 0.0185, "step": 1049 }, { "epoch": 1.81, "learning_rate": 7.148519915737727e-06, "loss": 0.0164, "step": 1050 }, { "epoch": 1.82, "learning_rate": 7.130643862495417e-06, "loss": 0.0217, "step": 1051 }, { "epoch": 1.82, "learning_rate": 7.112777795444839e-06, "loss": 0.0259, "step": 1052 }, { "epoch": 1.82, "learning_rate": 7.094921776765095e-06, "loss": 0.0153, "step": 1053 }, { "epoch": 1.82, "learning_rate": 7.077075868600303e-06, "loss": 0.022, "step": 1054 }, { "epoch": 1.82, "learning_rate": 7.059240133059408e-06, "loss": 0.0217, "step": 1055 }, { "epoch": 1.82, "learning_rate": 7.041414632215939e-06, "loss": 0.0141, "step": 1056 }, { "epoch": 1.83, "learning_rate": 7.023599428107815e-06, "loss": 0.0174, "step": 1057 }, { "epoch": 1.83, "learning_rate": 7.005794582737104e-06, "loss": 0.0237, "step": 1058 }, { "epoch": 1.83, "learning_rate": 6.988000158069845e-06, "loss": 0.0207, "step": 1059 }, { "epoch": 1.83, "learning_rate": 6.970216216035788e-06, "loss": 0.0274, "step": 1060 }, { "epoch": 1.83, "learning_rate": 6.95244281852822e-06, "loss": 0.0202, "step": 1061 }, { "epoch": 1.83, "learning_rate": 6.934680027403711e-06, "loss": 0.0189, "step": 1062 }, { "epoch": 1.84, "learning_rate": 6.916927904481934e-06, "loss": 0.0257, "step": 1063 }, { "epoch": 1.84, "learning_rate": 6.899186511545424e-06, "loss": 0.0187, "step": 1064 }, { "epoch": 1.84, "learning_rate": 6.881455910339369e-06, "loss": 0.0219, "step": 1065 }, { "epoch": 1.84, "learning_rate": 6.863736162571415e-06, "loss": 0.0196, "step": 1066 }, { "epoch": 1.84, "learning_rate": 6.846027329911417e-06, "loss": 0.0163, "step": 1067 }, { "epoch": 1.84, "learning_rate": 6.828329473991256e-06, "loss": 0.0172, "step": 1068 }, { "epoch": 1.85, "learning_rate": 6.8106426564045965e-06, "loss": 0.0194, "step": 1069 }, { "epoch": 1.85, "learning_rate": 6.792966938706704e-06, "loss": 0.0213, "step": 1070 }, { "epoch": 1.85, "learning_rate": 6.775302382414194e-06, "loss": 0.0201, "step": 1071 }, { "epoch": 1.85, "learning_rate": 6.757649049004856e-06, "loss": 0.018, "step": 1072 }, { "epoch": 1.85, "learning_rate": 6.740006999917406e-06, "loss": 0.0213, "step": 1073 }, { "epoch": 1.85, "learning_rate": 6.722376296551294e-06, "loss": 0.0161, "step": 1074 }, { "epoch": 1.86, "learning_rate": 6.704757000266478e-06, "loss": 0.0187, "step": 1075 }, { "epoch": 1.86, "learning_rate": 6.687149172383228e-06, "loss": 0.0171, "step": 1076 }, { "epoch": 1.86, "learning_rate": 6.669552874181888e-06, "loss": 0.0199, "step": 1077 }, { "epoch": 1.86, "learning_rate": 6.651968166902683e-06, "loss": 0.0178, "step": 1078 }, { "epoch": 1.86, "learning_rate": 6.63439511174549e-06, "loss": 0.021, "step": 1079 }, { "epoch": 1.87, "learning_rate": 6.616833769869646e-06, "loss": 0.0191, "step": 1080 }, { "epoch": 1.87, "learning_rate": 6.599284202393709e-06, "loss": 0.0215, "step": 1081 }, { "epoch": 1.87, "learning_rate": 6.581746470395271e-06, "loss": 0.0259, "step": 1082 }, { "epoch": 1.87, "learning_rate": 6.56422063491072e-06, "loss": 0.0136, "step": 1083 }, { "epoch": 1.87, "learning_rate": 6.546706756935055e-06, "loss": 0.0182, "step": 1084 }, { "epoch": 1.87, "learning_rate": 6.529204897421644e-06, "loss": 0.0246, "step": 1085 }, { "epoch": 1.88, "learning_rate": 6.511715117282041e-06, "loss": 0.0167, "step": 1086 }, { "epoch": 1.88, "learning_rate": 6.494237477385753e-06, "loss": 0.0162, "step": 1087 }, { "epoch": 1.88, "learning_rate": 6.476772038560039e-06, "loss": 0.0274, "step": 1088 }, { "epoch": 1.88, "learning_rate": 6.4593188615896855e-06, "loss": 0.0212, "step": 1089 }, { "epoch": 1.88, "learning_rate": 6.44187800721682e-06, "loss": 0.024, "step": 1090 }, { "epoch": 1.88, "learning_rate": 6.424449536140668e-06, "loss": 0.0139, "step": 1091 }, { "epoch": 1.89, "learning_rate": 6.407033509017372e-06, "loss": 0.0205, "step": 1092 }, { "epoch": 1.89, "learning_rate": 6.389629986459756e-06, "loss": 0.0179, "step": 1093 }, { "epoch": 1.89, "learning_rate": 6.372239029037122e-06, "loss": 0.0241, "step": 1094 }, { "epoch": 1.89, "learning_rate": 6.354860697275054e-06, "loss": 0.0145, "step": 1095 }, { "epoch": 1.89, "learning_rate": 6.337495051655181e-06, "loss": 0.0179, "step": 1096 }, { "epoch": 1.89, "learning_rate": 6.3201421526149945e-06, "loss": 0.0231, "step": 1097 }, { "epoch": 1.9, "learning_rate": 6.3028020605476105e-06, "loss": 0.0262, "step": 1098 }, { "epoch": 1.9, "learning_rate": 6.285474835801586e-06, "loss": 0.0181, "step": 1099 }, { "epoch": 1.9, "learning_rate": 6.2681605386806835e-06, "loss": 0.0247, "step": 1100 }, { "epoch": 1.9, "learning_rate": 6.250859229443684e-06, "loss": 0.0216, "step": 1101 }, { "epoch": 1.9, "learning_rate": 6.233570968304159e-06, "loss": 0.0159, "step": 1102 }, { "epoch": 1.91, "learning_rate": 6.216295815430277e-06, "loss": 0.0215, "step": 1103 }, { "epoch": 1.91, "learning_rate": 6.1990338309445785e-06, "loss": 0.0227, "step": 1104 }, { "epoch": 1.91, "learning_rate": 6.181785074923778e-06, "loss": 0.0213, "step": 1105 }, { "epoch": 1.91, "learning_rate": 6.164549607398547e-06, "loss": 0.0167, "step": 1106 }, { "epoch": 1.91, "learning_rate": 6.147327488353319e-06, "loss": 0.0099, "step": 1107 }, { "epoch": 1.91, "learning_rate": 6.130118777726059e-06, "loss": 0.0165, "step": 1108 }, { "epoch": 1.92, "learning_rate": 6.112923535408074e-06, "loss": 0.0149, "step": 1109 }, { "epoch": 1.92, "learning_rate": 6.095741821243791e-06, "loss": 0.0193, "step": 1110 }, { "epoch": 1.92, "learning_rate": 6.078573695030568e-06, "loss": 0.0211, "step": 1111 }, { "epoch": 1.92, "learning_rate": 6.061419216518456e-06, "loss": 0.0242, "step": 1112 }, { "epoch": 1.92, "learning_rate": 6.044278445410025e-06, "loss": 0.0212, "step": 1113 }, { "epoch": 1.92, "learning_rate": 6.027151441360124e-06, "loss": 0.0253, "step": 1114 }, { "epoch": 1.93, "learning_rate": 6.0100382639757e-06, "loss": 0.015, "step": 1115 }, { "epoch": 1.93, "learning_rate": 5.992938972815569e-06, "loss": 0.0203, "step": 1116 }, { "epoch": 1.93, "learning_rate": 5.975853627390233e-06, "loss": 0.0215, "step": 1117 }, { "epoch": 1.93, "learning_rate": 5.958782287161642e-06, "loss": 0.0197, "step": 1118 }, { "epoch": 1.93, "learning_rate": 5.941725011543018e-06, "loss": 0.028, "step": 1119 }, { "epoch": 1.93, "learning_rate": 5.924681859898628e-06, "loss": 0.0145, "step": 1120 }, { "epoch": 1.94, "learning_rate": 5.907652891543576e-06, "loss": 0.025, "step": 1121 }, { "epoch": 1.94, "learning_rate": 5.890638165743624e-06, "loss": 0.0182, "step": 1122 }, { "epoch": 1.94, "learning_rate": 5.873637741714941e-06, "loss": 0.022, "step": 1123 }, { "epoch": 1.94, "learning_rate": 5.856651678623948e-06, "loss": 0.0231, "step": 1124 }, { "epoch": 1.94, "learning_rate": 5.839680035587061e-06, "loss": 0.017, "step": 1125 }, { "epoch": 1.94, "learning_rate": 5.8227228716705255e-06, "loss": 0.0124, "step": 1126 }, { "epoch": 1.95, "learning_rate": 5.805780245890192e-06, "loss": 0.0172, "step": 1127 }, { "epoch": 1.95, "learning_rate": 5.788852217211311e-06, "loss": 0.0251, "step": 1128 }, { "epoch": 1.95, "learning_rate": 5.771938844548331e-06, "loss": 0.0184, "step": 1129 }, { "epoch": 1.95, "learning_rate": 5.755040186764702e-06, "loss": 0.0167, "step": 1130 }, { "epoch": 1.95, "learning_rate": 5.738156302672646e-06, "loss": 0.0209, "step": 1131 }, { "epoch": 1.96, "learning_rate": 5.721287251032988e-06, "loss": 0.0229, "step": 1132 }, { "epoch": 1.96, "learning_rate": 5.704433090554912e-06, "loss": 0.0154, "step": 1133 }, { "epoch": 1.96, "learning_rate": 5.687593879895793e-06, "loss": 0.0211, "step": 1134 }, { "epoch": 1.96, "learning_rate": 5.670769677660962e-06, "loss": 0.0217, "step": 1135 }, { "epoch": 1.96, "learning_rate": 5.653960542403531e-06, "loss": 0.0088, "step": 1136 }, { "epoch": 1.96, "learning_rate": 5.6371665326241635e-06, "loss": 0.021, "step": 1137 }, { "epoch": 1.97, "learning_rate": 5.620387706770884e-06, "loss": 0.0224, "step": 1138 }, { "epoch": 1.97, "learning_rate": 5.603624123238871e-06, "loss": 0.0118, "step": 1139 }, { "epoch": 1.97, "learning_rate": 5.586875840370267e-06, "loss": 0.0179, "step": 1140 }, { "epoch": 1.97, "learning_rate": 5.570142916453944e-06, "loss": 0.019, "step": 1141 }, { "epoch": 1.97, "learning_rate": 5.5534254097253425e-06, "loss": 0.0186, "step": 1142 }, { "epoch": 1.97, "learning_rate": 5.536723378366226e-06, "loss": 0.023, "step": 1143 }, { "epoch": 1.98, "learning_rate": 5.520036880504515e-06, "loss": 0.0212, "step": 1144 }, { "epoch": 1.98, "learning_rate": 5.503365974214059e-06, "loss": 0.0133, "step": 1145 }, { "epoch": 1.98, "learning_rate": 5.486710717514453e-06, "loss": 0.013, "step": 1146 }, { "epoch": 1.98, "learning_rate": 5.470071168370818e-06, "loss": 0.0182, "step": 1147 }, { "epoch": 1.98, "learning_rate": 5.453447384693613e-06, "loss": 0.0168, "step": 1148 }, { "epoch": 1.98, "learning_rate": 5.436839424338426e-06, "loss": 0.0301, "step": 1149 }, { "epoch": 1.99, "learning_rate": 5.4202473451057736e-06, "loss": 0.0196, "step": 1150 }, { "epoch": 1.99, "learning_rate": 5.403671204740915e-06, "loss": 0.0154, "step": 1151 }, { "epoch": 1.99, "learning_rate": 5.387111060933616e-06, "loss": 0.017, "step": 1152 }, { "epoch": 1.99, "learning_rate": 5.37056697131799e-06, "loss": 0.0159, "step": 1153 }, { "epoch": 1.99, "learning_rate": 5.354038993472262e-06, "loss": 0.0155, "step": 1154 }, { "epoch": 1.99, "learning_rate": 5.337527184918594e-06, "loss": 0.0236, "step": 1155 }, { "epoch": 2.0, "learning_rate": 5.321031603122866e-06, "loss": 0.0134, "step": 1156 }, { "epoch": 2.0, "learning_rate": 5.304552305494493e-06, "loss": 0.0165, "step": 1157 }, { "epoch": 2.0, "learning_rate": 5.288089349386208e-06, "loss": 0.0183, "step": 1158 }, { "epoch": 2.0, "learning_rate": 5.2716427920938714e-06, "loss": 0.0154, "step": 1159 }, { "epoch": 2.0, "learning_rate": 5.255212690856275e-06, "loss": 0.0161, "step": 1160 }, { "epoch": 2.01, "learning_rate": 5.238799102854941e-06, "loss": 0.0251, "step": 1161 }, { "epoch": 2.01, "learning_rate": 5.222402085213911e-06, "loss": 0.0126, "step": 1162 }, { "epoch": 2.01, "learning_rate": 5.206021694999571e-06, "loss": 0.0137, "step": 1163 }, { "epoch": 2.01, "learning_rate": 5.189657989220422e-06, "loss": 0.0229, "step": 1164 }, { "epoch": 2.01, "learning_rate": 5.173311024826916e-06, "loss": 0.0108, "step": 1165 }, { "epoch": 2.01, "learning_rate": 5.156980858711223e-06, "loss": 0.0173, "step": 1166 }, { "epoch": 2.02, "learning_rate": 5.140667547707065e-06, "loss": 0.0146, "step": 1167 }, { "epoch": 2.02, "learning_rate": 5.124371148589493e-06, "loss": 0.02, "step": 1168 }, { "epoch": 2.02, "learning_rate": 5.108091718074706e-06, "loss": 0.0175, "step": 1169 }, { "epoch": 2.02, "learning_rate": 5.091829312819836e-06, "loss": 0.0126, "step": 1170 }, { "epoch": 2.02, "learning_rate": 5.075583989422783e-06, "loss": 0.0138, "step": 1171 }, { "epoch": 2.02, "learning_rate": 5.059355804421977e-06, "loss": 0.0143, "step": 1172 }, { "epoch": 2.03, "learning_rate": 5.043144814296214e-06, "loss": 0.013, "step": 1173 }, { "epoch": 2.03, "learning_rate": 5.026951075464439e-06, "loss": 0.0143, "step": 1174 }, { "epoch": 2.03, "learning_rate": 5.010774644285564e-06, "loss": 0.0139, "step": 1175 }, { "epoch": 2.03, "learning_rate": 4.994615577058262e-06, "loss": 0.0144, "step": 1176 }, { "epoch": 2.03, "learning_rate": 4.9784739300207675e-06, "loss": 0.0194, "step": 1177 }, { "epoch": 2.03, "learning_rate": 4.962349759350704e-06, "loss": 0.0218, "step": 1178 }, { "epoch": 2.04, "learning_rate": 4.946243121164861e-06, "loss": 0.016, "step": 1179 }, { "epoch": 2.04, "learning_rate": 4.930154071519009e-06, "loss": 0.0131, "step": 1180 }, { "epoch": 2.04, "learning_rate": 4.914082666407705e-06, "loss": 0.0117, "step": 1181 }, { "epoch": 2.04, "learning_rate": 4.898028961764108e-06, "loss": 0.0182, "step": 1182 }, { "epoch": 2.04, "learning_rate": 4.881993013459762e-06, "loss": 0.0171, "step": 1183 }, { "epoch": 2.04, "learning_rate": 4.8659748773044235e-06, "loss": 0.0127, "step": 1184 }, { "epoch": 2.05, "learning_rate": 4.849974609045849e-06, "loss": 0.0125, "step": 1185 }, { "epoch": 2.05, "learning_rate": 4.833992264369617e-06, "loss": 0.0115, "step": 1186 }, { "epoch": 2.05, "learning_rate": 4.818027898898919e-06, "loss": 0.0141, "step": 1187 }, { "epoch": 2.05, "learning_rate": 4.802081568194383e-06, "loss": 0.0169, "step": 1188 }, { "epoch": 2.05, "learning_rate": 4.786153327753865e-06, "loss": 0.0126, "step": 1189 }, { "epoch": 2.06, "learning_rate": 4.770243233012261e-06, "loss": 0.0157, "step": 1190 }, { "epoch": 2.06, "learning_rate": 4.754351339341311e-06, "loss": 0.0187, "step": 1191 }, { "epoch": 2.06, "learning_rate": 4.738477702049423e-06, "loss": 0.0142, "step": 1192 }, { "epoch": 2.06, "learning_rate": 4.722622376381455e-06, "loss": 0.023, "step": 1193 }, { "epoch": 2.06, "learning_rate": 4.706785417518544e-06, "loss": 0.0167, "step": 1194 }, { "epoch": 2.06, "learning_rate": 4.690966880577894e-06, "loss": 0.0154, "step": 1195 }, { "epoch": 2.07, "learning_rate": 4.675166820612611e-06, "loss": 0.0216, "step": 1196 }, { "epoch": 2.07, "learning_rate": 4.659385292611479e-06, "loss": 0.0162, "step": 1197 }, { "epoch": 2.07, "learning_rate": 4.643622351498801e-06, "loss": 0.0146, "step": 1198 }, { "epoch": 2.07, "learning_rate": 4.627878052134182e-06, "loss": 0.0177, "step": 1199 }, { "epoch": 2.07, "learning_rate": 4.61215244931235e-06, "loss": 0.0161, "step": 1200 }, { "epoch": 2.07, "learning_rate": 4.59644559776296e-06, "loss": 0.0183, "step": 1201 }, { "epoch": 2.08, "learning_rate": 4.580757552150421e-06, "loss": 0.0183, "step": 1202 }, { "epoch": 2.08, "learning_rate": 4.565088367073675e-06, "loss": 0.012, "step": 1203 }, { "epoch": 2.08, "learning_rate": 4.549438097066039e-06, "loss": 0.0229, "step": 1204 }, { "epoch": 2.08, "learning_rate": 4.533806796594989e-06, "loss": 0.0149, "step": 1205 }, { "epoch": 2.08, "learning_rate": 4.5181945200619814e-06, "loss": 0.0175, "step": 1206 }, { "epoch": 2.08, "learning_rate": 4.502601321802273e-06, "loss": 0.0176, "step": 1207 }, { "epoch": 2.09, "learning_rate": 4.487027256084713e-06, "loss": 0.0134, "step": 1208 }, { "epoch": 2.09, "learning_rate": 4.471472377111574e-06, "loss": 0.0174, "step": 1209 }, { "epoch": 2.09, "learning_rate": 4.455936739018343e-06, "loss": 0.0141, "step": 1210 }, { "epoch": 2.09, "learning_rate": 4.440420395873547e-06, "loss": 0.0119, "step": 1211 }, { "epoch": 2.09, "learning_rate": 4.424923401678559e-06, "loss": 0.015, "step": 1212 }, { "epoch": 2.09, "learning_rate": 4.409445810367421e-06, "loss": 0.0142, "step": 1213 }, { "epoch": 2.1, "learning_rate": 4.393987675806634e-06, "loss": 0.0139, "step": 1214 }, { "epoch": 2.1, "learning_rate": 4.378549051794996e-06, "loss": 0.021, "step": 1215 }, { "epoch": 2.1, "learning_rate": 4.363129992063392e-06, "loss": 0.0129, "step": 1216 }, { "epoch": 2.1, "learning_rate": 4.347730550274628e-06, "loss": 0.0169, "step": 1217 }, { "epoch": 2.1, "learning_rate": 4.3323507800232265e-06, "loss": 0.0124, "step": 1218 }, { "epoch": 2.11, "learning_rate": 4.31699073483525e-06, "loss": 0.0193, "step": 1219 }, { "epoch": 2.11, "learning_rate": 4.301650468168108e-06, "loss": 0.0177, "step": 1220 }, { "epoch": 2.11, "learning_rate": 4.286330033410384e-06, "loss": 0.0149, "step": 1221 }, { "epoch": 2.11, "learning_rate": 4.271029483881629e-06, "loss": 0.0178, "step": 1222 }, { "epoch": 2.11, "learning_rate": 4.255748872832201e-06, "loss": 0.0129, "step": 1223 }, { "epoch": 2.11, "learning_rate": 4.240488253443053e-06, "loss": 0.0211, "step": 1224 }, { "epoch": 2.12, "learning_rate": 4.2252476788255735e-06, "loss": 0.0204, "step": 1225 }, { "epoch": 2.12, "learning_rate": 4.210027202021377e-06, "loss": 0.0191, "step": 1226 }, { "epoch": 2.12, "learning_rate": 4.194826876002145e-06, "loss": 0.0161, "step": 1227 }, { "epoch": 2.12, "learning_rate": 4.179646753669417e-06, "loss": 0.013, "step": 1228 }, { "epoch": 2.12, "learning_rate": 4.164486887854424e-06, "loss": 0.0129, "step": 1229 }, { "epoch": 2.12, "learning_rate": 4.149347331317892e-06, "loss": 0.0126, "step": 1230 }, { "epoch": 2.13, "learning_rate": 4.134228136749877e-06, "loss": 0.0217, "step": 1231 }, { "epoch": 2.13, "learning_rate": 4.119129356769559e-06, "loss": 0.0195, "step": 1232 }, { "epoch": 2.13, "learning_rate": 4.104051043925068e-06, "loss": 0.0155, "step": 1233 }, { "epoch": 2.13, "learning_rate": 4.088993250693315e-06, "loss": 0.0148, "step": 1234 }, { "epoch": 2.13, "learning_rate": 4.07395602947978e-06, "loss": 0.0187, "step": 1235 }, { "epoch": 2.13, "learning_rate": 4.058939432618364e-06, "loss": 0.0143, "step": 1236 }, { "epoch": 2.14, "learning_rate": 4.043943512371171e-06, "loss": 0.0154, "step": 1237 }, { "epoch": 2.14, "learning_rate": 4.028968320928363e-06, "loss": 0.0124, "step": 1238 }, { "epoch": 2.14, "learning_rate": 4.0140139104079445e-06, "loss": 0.0128, "step": 1239 }, { "epoch": 2.14, "learning_rate": 3.999080332855604e-06, "loss": 0.0136, "step": 1240 }, { "epoch": 2.14, "learning_rate": 3.984167640244518e-06, "loss": 0.0189, "step": 1241 }, { "epoch": 2.15, "learning_rate": 3.9692758844751936e-06, "loss": 0.0136, "step": 1242 }, { "epoch": 2.15, "learning_rate": 3.9544051173752504e-06, "loss": 0.0187, "step": 1243 }, { "epoch": 2.15, "learning_rate": 3.9395553906992804e-06, "loss": 0.0082, "step": 1244 }, { "epoch": 2.15, "learning_rate": 3.924726756128632e-06, "loss": 0.0131, "step": 1245 }, { "epoch": 2.15, "learning_rate": 3.9099192652712635e-06, "loss": 0.0175, "step": 1246 }, { "epoch": 2.15, "learning_rate": 3.89513296966153e-06, "loss": 0.0197, "step": 1247 }, { "epoch": 2.16, "learning_rate": 3.880367920760038e-06, "loss": 0.0173, "step": 1248 }, { "epoch": 2.16, "learning_rate": 3.8656241699534396e-06, "loss": 0.0178, "step": 1249 }, { "epoch": 2.16, "learning_rate": 3.85090176855426e-06, "loss": 0.0175, "step": 1250 }, { "epoch": 2.16, "learning_rate": 3.836200767800725e-06, "loss": 0.01, "step": 1251 }, { "epoch": 2.16, "learning_rate": 3.8215212188565894e-06, "loss": 0.0174, "step": 1252 }, { "epoch": 2.16, "learning_rate": 3.8068631728109364e-06, "loss": 0.0091, "step": 1253 }, { "epoch": 2.17, "learning_rate": 3.792226680678023e-06, "loss": 0.0197, "step": 1254 }, { "epoch": 2.17, "learning_rate": 3.7776117933970814e-06, "loss": 0.0153, "step": 1255 }, { "epoch": 2.17, "learning_rate": 3.763018561832166e-06, "loss": 0.0255, "step": 1256 }, { "epoch": 2.17, "learning_rate": 3.7484470367719493e-06, "loss": 0.0146, "step": 1257 }, { "epoch": 2.17, "learning_rate": 3.733897268929574e-06, "loss": 0.0159, "step": 1258 }, { "epoch": 2.17, "learning_rate": 3.7193693089424465e-06, "loss": 0.0159, "step": 1259 }, { "epoch": 2.18, "learning_rate": 3.7048632073720846e-06, "loss": 0.0133, "step": 1260 }, { "epoch": 2.18, "learning_rate": 3.6903790147039286e-06, "loss": 0.0168, "step": 1261 }, { "epoch": 2.18, "learning_rate": 3.6759167813471675e-06, "loss": 0.0148, "step": 1262 }, { "epoch": 2.18, "learning_rate": 3.6614765576345755e-06, "loss": 0.0144, "step": 1263 }, { "epoch": 2.18, "learning_rate": 3.6470583938223127e-06, "loss": 0.0123, "step": 1264 }, { "epoch": 2.18, "learning_rate": 3.6326623400897797e-06, "loss": 0.014, "step": 1265 }, { "epoch": 2.19, "learning_rate": 3.618288446539413e-06, "loss": 0.0171, "step": 1266 }, { "epoch": 2.19, "learning_rate": 3.6039367631965383e-06, "loss": 0.0088, "step": 1267 }, { "epoch": 2.19, "learning_rate": 3.589607340009169e-06, "loss": 0.0168, "step": 1268 }, { "epoch": 2.19, "learning_rate": 3.575300226847863e-06, "loss": 0.0163, "step": 1269 }, { "epoch": 2.19, "learning_rate": 3.5610154735055203e-06, "loss": 0.0211, "step": 1270 }, { "epoch": 2.2, "learning_rate": 3.546753129697227e-06, "loss": 0.0148, "step": 1271 }, { "epoch": 2.2, "learning_rate": 3.532513245060073e-06, "loss": 0.0159, "step": 1272 }, { "epoch": 2.2, "learning_rate": 3.5182958691529945e-06, "loss": 0.0163, "step": 1273 }, { "epoch": 2.2, "learning_rate": 3.5041010514565776e-06, "loss": 0.0165, "step": 1274 }, { "epoch": 2.2, "learning_rate": 3.4899288413729128e-06, "loss": 0.0229, "step": 1275 }, { "epoch": 2.2, "learning_rate": 3.475779288225394e-06, "loss": 0.0135, "step": 1276 }, { "epoch": 2.21, "learning_rate": 3.4616524412585797e-06, "loss": 0.0142, "step": 1277 }, { "epoch": 2.21, "learning_rate": 3.447548349637987e-06, "loss": 0.015, "step": 1278 }, { "epoch": 2.21, "learning_rate": 3.433467062449953e-06, "loss": 0.0125, "step": 1279 }, { "epoch": 2.21, "learning_rate": 3.4194086287014405e-06, "loss": 0.0171, "step": 1280 }, { "epoch": 2.21, "learning_rate": 3.405373097319875e-06, "loss": 0.0218, "step": 1281 }, { "epoch": 2.21, "learning_rate": 3.3913605171529763e-06, "loss": 0.0114, "step": 1282 }, { "epoch": 2.22, "learning_rate": 3.3773709369685924e-06, "loss": 0.0215, "step": 1283 }, { "epoch": 2.22, "learning_rate": 3.3634044054545145e-06, "loss": 0.0127, "step": 1284 }, { "epoch": 2.22, "learning_rate": 3.3494609712183323e-06, "loss": 0.0104, "step": 1285 }, { "epoch": 2.22, "learning_rate": 3.3355406827872338e-06, "loss": 0.009, "step": 1286 }, { "epoch": 2.22, "learning_rate": 3.321643588607868e-06, "loss": 0.0148, "step": 1287 }, { "epoch": 2.22, "learning_rate": 3.3077697370461503e-06, "loss": 0.0141, "step": 1288 }, { "epoch": 2.23, "learning_rate": 3.2939191763871047e-06, "loss": 0.0139, "step": 1289 }, { "epoch": 2.23, "learning_rate": 3.2800919548347086e-06, "loss": 0.0157, "step": 1290 }, { "epoch": 2.23, "learning_rate": 3.2662881205117003e-06, "loss": 0.0123, "step": 1291 }, { "epoch": 2.23, "learning_rate": 3.252507721459428e-06, "loss": 0.0166, "step": 1292 }, { "epoch": 2.23, "learning_rate": 3.2387508056376726e-06, "loss": 0.0208, "step": 1293 }, { "epoch": 2.23, "learning_rate": 3.225017420924501e-06, "loss": 0.0219, "step": 1294 }, { "epoch": 2.24, "learning_rate": 3.2113076151160683e-06, "loss": 0.0139, "step": 1295 }, { "epoch": 2.24, "learning_rate": 3.1976214359264834e-06, "loss": 0.0154, "step": 1296 }, { "epoch": 2.24, "learning_rate": 3.183958930987612e-06, "loss": 0.0169, "step": 1297 }, { "epoch": 2.24, "learning_rate": 3.1703201478489408e-06, "loss": 0.0167, "step": 1298 }, { "epoch": 2.24, "learning_rate": 3.1567051339773867e-06, "loss": 0.0141, "step": 1299 }, { "epoch": 2.25, "learning_rate": 3.143113936757154e-06, "loss": 0.0167, "step": 1300 }, { "epoch": 2.25, "learning_rate": 3.1295466034895482e-06, "loss": 0.0108, "step": 1301 }, { "epoch": 2.25, "learning_rate": 3.116003181392825e-06, "loss": 0.015, "step": 1302 }, { "epoch": 2.25, "learning_rate": 3.1024837176020173e-06, "loss": 0.0127, "step": 1303 }, { "epoch": 2.25, "learning_rate": 3.0889882591687894e-06, "loss": 0.0146, "step": 1304 }, { "epoch": 2.25, "learning_rate": 3.0755168530612444e-06, "loss": 0.0127, "step": 1305 }, { "epoch": 2.26, "learning_rate": 3.0620695461637905e-06, "loss": 0.0199, "step": 1306 }, { "epoch": 2.26, "learning_rate": 3.048646385276949e-06, "loss": 0.016, "step": 1307 }, { "epoch": 2.26, "learning_rate": 3.035247417117221e-06, "loss": 0.0158, "step": 1308 }, { "epoch": 2.26, "learning_rate": 3.021872688316896e-06, "loss": 0.0129, "step": 1309 }, { "epoch": 2.26, "learning_rate": 3.0085222454239173e-06, "loss": 0.0133, "step": 1310 }, { "epoch": 2.26, "learning_rate": 2.9951961349016944e-06, "loss": 0.0174, "step": 1311 }, { "epoch": 2.27, "learning_rate": 2.9818944031289586e-06, "loss": 0.0194, "step": 1312 }, { "epoch": 2.27, "learning_rate": 2.968617096399592e-06, "loss": 0.0181, "step": 1313 }, { "epoch": 2.27, "learning_rate": 2.9553642609224787e-06, "loss": 0.013, "step": 1314 }, { "epoch": 2.27, "learning_rate": 2.942135942821327e-06, "loss": 0.013, "step": 1315 }, { "epoch": 2.27, "learning_rate": 2.9289321881345257e-06, "loss": 0.0195, "step": 1316 }, { "epoch": 2.27, "learning_rate": 2.9157530428149683e-06, "loss": 0.0139, "step": 1317 }, { "epoch": 2.28, "learning_rate": 2.9025985527299016e-06, "loss": 0.0179, "step": 1318 }, { "epoch": 2.28, "learning_rate": 2.889468763660772e-06, "loss": 0.0156, "step": 1319 }, { "epoch": 2.28, "learning_rate": 2.8763637213030494e-06, "loss": 0.0154, "step": 1320 }, { "epoch": 2.28, "learning_rate": 2.8632834712660882e-06, "loss": 0.0154, "step": 1321 }, { "epoch": 2.28, "learning_rate": 2.8502280590729504e-06, "loss": 0.0168, "step": 1322 }, { "epoch": 2.28, "learning_rate": 2.8371975301602572e-06, "loss": 0.0126, "step": 1323 }, { "epoch": 2.29, "learning_rate": 2.8241919298780274e-06, "loss": 0.0159, "step": 1324 }, { "epoch": 2.29, "learning_rate": 2.8112113034895273e-06, "loss": 0.0192, "step": 1325 }, { "epoch": 2.29, "learning_rate": 2.798255696171096e-06, "loss": 0.018, "step": 1326 }, { "epoch": 2.29, "learning_rate": 2.785325153012014e-06, "loss": 0.0176, "step": 1327 }, { "epoch": 2.29, "learning_rate": 2.772419719014313e-06, "loss": 0.0183, "step": 1328 }, { "epoch": 2.3, "learning_rate": 2.7595394390926557e-06, "loss": 0.0188, "step": 1329 }, { "epoch": 2.3, "learning_rate": 2.7466843580741464e-06, "loss": 0.0176, "step": 1330 }, { "epoch": 2.3, "learning_rate": 2.7338545206982003e-06, "loss": 0.0161, "step": 1331 }, { "epoch": 2.3, "learning_rate": 2.721049971616373e-06, "loss": 0.0166, "step": 1332 }, { "epoch": 2.3, "learning_rate": 2.708270755392207e-06, "loss": 0.0127, "step": 1333 }, { "epoch": 2.3, "learning_rate": 2.695516916501082e-06, "loss": 0.0164, "step": 1334 }, { "epoch": 2.31, "learning_rate": 2.6827884993300602e-06, "loss": 0.0136, "step": 1335 }, { "epoch": 2.31, "learning_rate": 2.670085548177722e-06, "loss": 0.0171, "step": 1336 }, { "epoch": 2.31, "learning_rate": 2.657408107254027e-06, "loss": 0.021, "step": 1337 }, { "epoch": 2.31, "learning_rate": 2.6447562206801413e-06, "loss": 0.0184, "step": 1338 }, { "epoch": 2.31, "learning_rate": 2.6321299324883064e-06, "loss": 0.0172, "step": 1339 }, { "epoch": 2.31, "learning_rate": 2.6195292866216635e-06, "loss": 0.0124, "step": 1340 }, { "epoch": 2.32, "learning_rate": 2.60695432693412e-06, "loss": 0.0168, "step": 1341 }, { "epoch": 2.32, "learning_rate": 2.594405097190181e-06, "loss": 0.0213, "step": 1342 }, { "epoch": 2.32, "learning_rate": 2.581881641064806e-06, "loss": 0.0207, "step": 1343 }, { "epoch": 2.32, "learning_rate": 2.5693840021432537e-06, "loss": 0.0202, "step": 1344 }, { "epoch": 2.32, "learning_rate": 2.5569122239209366e-06, "loss": 0.0238, "step": 1345 }, { "epoch": 2.32, "learning_rate": 2.5444663498032597e-06, "loss": 0.0126, "step": 1346 }, { "epoch": 2.33, "learning_rate": 2.532046423105472e-06, "loss": 0.014, "step": 1347 }, { "epoch": 2.33, "learning_rate": 2.5196524870525274e-06, "loss": 0.0195, "step": 1348 }, { "epoch": 2.33, "learning_rate": 2.507284584778913e-06, "loss": 0.0139, "step": 1349 }, { "epoch": 2.33, "learning_rate": 2.4949427593285215e-06, "loss": 0.0188, "step": 1350 }, { "epoch": 2.33, "learning_rate": 2.4826270536544807e-06, "loss": 0.0155, "step": 1351 }, { "epoch": 2.34, "learning_rate": 2.470337510619024e-06, "loss": 0.0143, "step": 1352 }, { "epoch": 2.34, "learning_rate": 2.4580741729933246e-06, "loss": 0.0149, "step": 1353 }, { "epoch": 2.34, "learning_rate": 2.4458370834573532e-06, "loss": 0.0157, "step": 1354 }, { "epoch": 2.34, "learning_rate": 2.4336262845997306e-06, "loss": 0.0194, "step": 1355 }, { "epoch": 2.34, "learning_rate": 2.4214418189175837e-06, "loss": 0.018, "step": 1356 }, { "epoch": 2.34, "learning_rate": 2.4092837288163807e-06, "loss": 0.0208, "step": 1357 }, { "epoch": 2.35, "learning_rate": 2.3971520566098106e-06, "loss": 0.0235, "step": 1358 }, { "epoch": 2.35, "learning_rate": 2.385046844519604e-06, "loss": 0.0122, "step": 1359 }, { "epoch": 2.35, "learning_rate": 2.3729681346754175e-06, "loss": 0.0176, "step": 1360 }, { "epoch": 2.35, "learning_rate": 2.3609159691146577e-06, "loss": 0.0245, "step": 1361 }, { "epoch": 2.35, "learning_rate": 2.3488903897823656e-06, "loss": 0.0133, "step": 1362 }, { "epoch": 2.35, "learning_rate": 2.3368914385310415e-06, "loss": 0.0152, "step": 1363 }, { "epoch": 2.36, "learning_rate": 2.3249191571205153e-06, "loss": 0.0187, "step": 1364 }, { "epoch": 2.36, "learning_rate": 2.312973587217798e-06, "loss": 0.0179, "step": 1365 }, { "epoch": 2.36, "learning_rate": 2.3010547703969445e-06, "loss": 0.0198, "step": 1366 }, { "epoch": 2.36, "learning_rate": 2.289162748138888e-06, "loss": 0.0166, "step": 1367 }, { "epoch": 2.36, "learning_rate": 2.277297561831322e-06, "loss": 0.0142, "step": 1368 }, { "epoch": 2.36, "learning_rate": 2.2654592527685305e-06, "loss": 0.0178, "step": 1369 }, { "epoch": 2.37, "learning_rate": 2.253647862151269e-06, "loss": 0.0113, "step": 1370 }, { "epoch": 2.37, "learning_rate": 2.241863431086598e-06, "loss": 0.0177, "step": 1371 }, { "epoch": 2.37, "learning_rate": 2.2301060005877627e-06, "loss": 0.0191, "step": 1372 }, { "epoch": 2.37, "learning_rate": 2.2183756115740274e-06, "loss": 0.0214, "step": 1373 }, { "epoch": 2.37, "learning_rate": 2.2066723048705486e-06, "loss": 0.0165, "step": 1374 }, { "epoch": 2.37, "learning_rate": 2.1949961212082303e-06, "loss": 0.0164, "step": 1375 }, { "epoch": 2.38, "learning_rate": 2.1833471012235765e-06, "loss": 0.0165, "step": 1376 }, { "epoch": 2.38, "learning_rate": 2.171725285458559e-06, "loss": 0.0164, "step": 1377 }, { "epoch": 2.38, "learning_rate": 2.1601307143604643e-06, "loss": 0.0227, "step": 1378 }, { "epoch": 2.38, "learning_rate": 2.148563428281768e-06, "loss": 0.0165, "step": 1379 }, { "epoch": 2.38, "learning_rate": 2.137023467479976e-06, "loss": 0.0143, "step": 1380 }, { "epoch": 2.39, "learning_rate": 2.1255108721175066e-06, "loss": 0.0195, "step": 1381 }, { "epoch": 2.39, "learning_rate": 2.1140256822615233e-06, "loss": 0.0144, "step": 1382 }, { "epoch": 2.39, "learning_rate": 2.1025679378838247e-06, "loss": 0.0134, "step": 1383 }, { "epoch": 2.39, "learning_rate": 2.091137678860684e-06, "loss": 0.0176, "step": 1384 }, { "epoch": 2.39, "learning_rate": 2.079734944972717e-06, "loss": 0.0144, "step": 1385 }, { "epoch": 2.39, "learning_rate": 2.0683597759047424e-06, "loss": 0.0107, "step": 1386 }, { "epoch": 2.4, "learning_rate": 2.0570122112456547e-06, "loss": 0.0175, "step": 1387 }, { "epoch": 2.4, "learning_rate": 2.0456922904882657e-06, "loss": 0.0153, "step": 1388 }, { "epoch": 2.4, "learning_rate": 2.0344000530291875e-06, "loss": 0.0167, "step": 1389 }, { "epoch": 2.4, "learning_rate": 2.023135538168678e-06, "loss": 0.0172, "step": 1390 }, { "epoch": 2.4, "learning_rate": 2.0118987851105207e-06, "loss": 0.0172, "step": 1391 }, { "epoch": 2.4, "learning_rate": 2.0006898329618717e-06, "loss": 0.0173, "step": 1392 }, { "epoch": 2.41, "learning_rate": 1.9895087207331422e-06, "loss": 0.0149, "step": 1393 }, { "epoch": 2.41, "learning_rate": 1.978355487337842e-06, "loss": 0.0159, "step": 1394 }, { "epoch": 2.41, "learning_rate": 1.967230171592459e-06, "loss": 0.0108, "step": 1395 }, { "epoch": 2.41, "learning_rate": 1.956132812216318e-06, "loss": 0.0168, "step": 1396 }, { "epoch": 2.41, "learning_rate": 1.945063447831452e-06, "loss": 0.0113, "step": 1397 }, { "epoch": 2.41, "learning_rate": 1.9340221169624575e-06, "loss": 0.0172, "step": 1398 }, { "epoch": 2.42, "learning_rate": 1.9230088580363727e-06, "loss": 0.0154, "step": 1399 }, { "epoch": 2.42, "learning_rate": 1.912023709382529e-06, "loss": 0.0172, "step": 1400 }, { "epoch": 2.42, "learning_rate": 1.9010667092324342e-06, "loss": 0.0124, "step": 1401 }, { "epoch": 2.42, "learning_rate": 1.890137895719626e-06, "loss": 0.0141, "step": 1402 }, { "epoch": 2.42, "learning_rate": 1.8792373068795422e-06, "loss": 0.0181, "step": 1403 }, { "epoch": 2.42, "learning_rate": 1.8683649806494009e-06, "loss": 0.0097, "step": 1404 }, { "epoch": 2.43, "learning_rate": 1.8575209548680472e-06, "loss": 0.0137, "step": 1405 }, { "epoch": 2.43, "learning_rate": 1.846705267275838e-06, "loss": 0.0124, "step": 1406 }, { "epoch": 2.43, "learning_rate": 1.8359179555144991e-06, "loss": 0.0135, "step": 1407 }, { "epoch": 2.43, "learning_rate": 1.8251590571270117e-06, "loss": 0.0181, "step": 1408 }, { "epoch": 2.43, "learning_rate": 1.814428609557458e-06, "loss": 0.0235, "step": 1409 }, { "epoch": 2.44, "learning_rate": 1.8037266501509121e-06, "loss": 0.0119, "step": 1410 }, { "epoch": 2.44, "learning_rate": 1.7930532161532943e-06, "loss": 0.0137, "step": 1411 }, { "epoch": 2.44, "learning_rate": 1.7824083447112538e-06, "loss": 0.0214, "step": 1412 }, { "epoch": 2.44, "learning_rate": 1.7717920728720284e-06, "loss": 0.0204, "step": 1413 }, { "epoch": 2.44, "learning_rate": 1.761204437583327e-06, "loss": 0.0159, "step": 1414 }, { "epoch": 2.44, "learning_rate": 1.7506454756931911e-06, "loss": 0.0172, "step": 1415 }, { "epoch": 2.45, "learning_rate": 1.7401152239498697e-06, "loss": 0.0123, "step": 1416 }, { "epoch": 2.45, "learning_rate": 1.7296137190016916e-06, "loss": 0.0197, "step": 1417 }, { "epoch": 2.45, "learning_rate": 1.7191409973969452e-06, "loss": 0.0203, "step": 1418 }, { "epoch": 2.45, "learning_rate": 1.7086970955837344e-06, "loss": 0.0129, "step": 1419 }, { "epoch": 2.45, "learning_rate": 1.698282049909874e-06, "loss": 0.0126, "step": 1420 }, { "epoch": 2.45, "learning_rate": 1.6878958966227366e-06, "loss": 0.0161, "step": 1421 }, { "epoch": 2.46, "learning_rate": 1.6775386718691566e-06, "loss": 0.0153, "step": 1422 }, { "epoch": 2.46, "learning_rate": 1.6672104116952748e-06, "loss": 0.0168, "step": 1423 }, { "epoch": 2.46, "learning_rate": 1.6569111520464354e-06, "loss": 0.0126, "step": 1424 }, { "epoch": 2.46, "learning_rate": 1.646640928767047e-06, "loss": 0.0185, "step": 1425 }, { "epoch": 2.46, "learning_rate": 1.6363997776004691e-06, "loss": 0.0165, "step": 1426 }, { "epoch": 2.46, "learning_rate": 1.6261877341888754e-06, "loss": 0.0138, "step": 1427 }, { "epoch": 2.47, "learning_rate": 1.6160048340731439e-06, "loss": 0.0159, "step": 1428 }, { "epoch": 2.47, "learning_rate": 1.6058511126927178e-06, "loss": 0.0153, "step": 1429 }, { "epoch": 2.47, "learning_rate": 1.5957266053854913e-06, "loss": 0.0176, "step": 1430 }, { "epoch": 2.47, "learning_rate": 1.5856313473876916e-06, "loss": 0.0151, "step": 1431 }, { "epoch": 2.47, "learning_rate": 1.5755653738337395e-06, "loss": 0.0135, "step": 1432 }, { "epoch": 2.47, "learning_rate": 1.5655287197561497e-06, "loss": 0.0145, "step": 1433 }, { "epoch": 2.48, "learning_rate": 1.5555214200853851e-06, "loss": 0.0122, "step": 1434 }, { "epoch": 2.48, "learning_rate": 1.545543509649754e-06, "loss": 0.0157, "step": 1435 }, { "epoch": 2.48, "learning_rate": 1.535595023175276e-06, "loss": 0.0132, "step": 1436 }, { "epoch": 2.48, "learning_rate": 1.5256759952855737e-06, "loss": 0.0159, "step": 1437 }, { "epoch": 2.48, "learning_rate": 1.515786460501738e-06, "loss": 0.0164, "step": 1438 }, { "epoch": 2.49, "learning_rate": 1.5059264532422224e-06, "loss": 0.0092, "step": 1439 }, { "epoch": 2.49, "learning_rate": 1.4960960078227093e-06, "loss": 0.0098, "step": 1440 }, { "epoch": 2.49, "learning_rate": 1.4862951584560037e-06, "loss": 0.0179, "step": 1441 }, { "epoch": 2.49, "learning_rate": 1.4765239392519015e-06, "loss": 0.0171, "step": 1442 }, { "epoch": 2.49, "learning_rate": 1.4667823842170837e-06, "loss": 0.0167, "step": 1443 }, { "epoch": 2.49, "learning_rate": 1.4570705272549846e-06, "loss": 0.0175, "step": 1444 }, { "epoch": 2.5, "learning_rate": 1.447388402165686e-06, "loss": 0.0132, "step": 1445 }, { "epoch": 2.5, "learning_rate": 1.4377360426457876e-06, "loss": 0.0133, "step": 1446 }, { "epoch": 2.5, "learning_rate": 1.4281134822883058e-06, "loss": 0.0158, "step": 1447 }, { "epoch": 2.5, "learning_rate": 1.4185207545825375e-06, "loss": 0.0119, "step": 1448 }, { "epoch": 2.5, "learning_rate": 1.4089578929139635e-06, "loss": 0.0093, "step": 1449 }, { "epoch": 2.5, "learning_rate": 1.3994249305641105e-06, "loss": 0.0131, "step": 1450 }, { "epoch": 2.51, "learning_rate": 1.389921900710458e-06, "loss": 0.0179, "step": 1451 }, { "epoch": 2.51, "learning_rate": 1.3804488364263024e-06, "loss": 0.0144, "step": 1452 }, { "epoch": 2.51, "learning_rate": 1.371005770680659e-06, "loss": 0.0168, "step": 1453 }, { "epoch": 2.51, "learning_rate": 1.3615927363381342e-06, "loss": 0.0186, "step": 1454 }, { "epoch": 2.51, "learning_rate": 1.3522097661588195e-06, "loss": 0.0188, "step": 1455 }, { "epoch": 2.51, "learning_rate": 1.3428568927981689e-06, "loss": 0.0132, "step": 1456 }, { "epoch": 2.52, "learning_rate": 1.3335341488068997e-06, "loss": 0.0132, "step": 1457 }, { "epoch": 2.52, "learning_rate": 1.324241566630865e-06, "loss": 0.0221, "step": 1458 }, { "epoch": 2.52, "learning_rate": 1.3149791786109422e-06, "loss": 0.0174, "step": 1459 }, { "epoch": 2.52, "learning_rate": 1.3057470169829334e-06, "loss": 0.0181, "step": 1460 }, { "epoch": 2.52, "learning_rate": 1.2965451138774343e-06, "loss": 0.0092, "step": 1461 }, { "epoch": 2.53, "learning_rate": 1.2873735013197442e-06, "loss": 0.0184, "step": 1462 }, { "epoch": 2.53, "learning_rate": 1.2782322112297274e-06, "loss": 0.0156, "step": 1463 }, { "epoch": 2.53, "learning_rate": 1.2691212754217318e-06, "loss": 0.0155, "step": 1464 }, { "epoch": 2.53, "learning_rate": 1.2600407256044544e-06, "loss": 0.0204, "step": 1465 }, { "epoch": 2.53, "learning_rate": 1.250990593380843e-06, "loss": 0.0187, "step": 1466 }, { "epoch": 2.53, "learning_rate": 1.2419709102479804e-06, "loss": 0.0185, "step": 1467 }, { "epoch": 2.54, "learning_rate": 1.2329817075969851e-06, "loss": 0.0166, "step": 1468 }, { "epoch": 2.54, "learning_rate": 1.2240230167128863e-06, "loss": 0.0136, "step": 1469 }, { "epoch": 2.54, "learning_rate": 1.2150948687745301e-06, "loss": 0.0177, "step": 1470 }, { "epoch": 2.54, "learning_rate": 1.2061972948544575e-06, "loss": 0.0151, "step": 1471 }, { "epoch": 2.54, "learning_rate": 1.1973303259188107e-06, "loss": 0.0118, "step": 1472 }, { "epoch": 2.54, "learning_rate": 1.188493992827211e-06, "loss": 0.0136, "step": 1473 }, { "epoch": 2.55, "learning_rate": 1.1796883263326643e-06, "loss": 0.0139, "step": 1474 }, { "epoch": 2.55, "learning_rate": 1.1709133570814424e-06, "loss": 0.0126, "step": 1475 }, { "epoch": 2.55, "learning_rate": 1.1621691156129833e-06, "loss": 0.0104, "step": 1476 }, { "epoch": 2.55, "learning_rate": 1.1534556323597824e-06, "loss": 0.0154, "step": 1477 }, { "epoch": 2.55, "learning_rate": 1.1447729376472937e-06, "loss": 0.016, "step": 1478 }, { "epoch": 2.55, "learning_rate": 1.136121061693809e-06, "loss": 0.0154, "step": 1479 }, { "epoch": 2.56, "learning_rate": 1.1275000346103704e-06, "loss": 0.0195, "step": 1480 }, { "epoch": 2.56, "learning_rate": 1.1189098864006488e-06, "loss": 0.0136, "step": 1481 }, { "epoch": 2.56, "learning_rate": 1.110350646960856e-06, "loss": 0.0246, "step": 1482 }, { "epoch": 2.56, "learning_rate": 1.101822346079625e-06, "loss": 0.0141, "step": 1483 }, { "epoch": 2.56, "learning_rate": 1.0933250134379202e-06, "loss": 0.01, "step": 1484 }, { "epoch": 2.56, "learning_rate": 1.084858678608922e-06, "loss": 0.0176, "step": 1485 }, { "epoch": 2.57, "learning_rate": 1.076423371057932e-06, "loss": 0.0222, "step": 1486 }, { "epoch": 2.57, "learning_rate": 1.0680191201422686e-06, "loss": 0.0152, "step": 1487 }, { "epoch": 2.57, "learning_rate": 1.0596459551111582e-06, "loss": 0.0205, "step": 1488 }, { "epoch": 2.57, "learning_rate": 1.0513039051056507e-06, "loss": 0.017, "step": 1489 }, { "epoch": 2.57, "learning_rate": 1.0429929991584942e-06, "loss": 0.0126, "step": 1490 }, { "epoch": 2.58, "learning_rate": 1.0347132661940572e-06, "loss": 0.02, "step": 1491 }, { "epoch": 2.58, "learning_rate": 1.0264647350282097e-06, "loss": 0.0154, "step": 1492 }, { "epoch": 2.58, "learning_rate": 1.0182474343682346e-06, "loss": 0.0164, "step": 1493 }, { "epoch": 2.58, "learning_rate": 1.010061392812719e-06, "loss": 0.012, "step": 1494 }, { "epoch": 2.58, "learning_rate": 1.0019066388514664e-06, "loss": 0.0148, "step": 1495 }, { "epoch": 2.58, "learning_rate": 9.937832008653837e-07, "loss": 0.0107, "step": 1496 }, { "epoch": 2.59, "learning_rate": 9.85691107126392e-07, "loss": 0.0148, "step": 1497 }, { "epoch": 2.59, "learning_rate": 9.776303857973235e-07, "loss": 0.0165, "step": 1498 }, { "epoch": 2.59, "learning_rate": 9.696010649318288e-07, "loss": 0.0125, "step": 1499 }, { "epoch": 2.59, "learning_rate": 9.616031724742725e-07, "loss": 0.012, "step": 1500 }, { "epoch": 2.59, "learning_rate": 9.536367362596422e-07, "loss": 0.012, "step": 1501 }, { "epoch": 2.59, "learning_rate": 9.457017840134441e-07, "loss": 0.0179, "step": 1502 }, { "epoch": 2.6, "learning_rate": 9.377983433516181e-07, "loss": 0.0138, "step": 1503 }, { "epoch": 2.6, "learning_rate": 9.299264417804254e-07, "loss": 0.0145, "step": 1504 }, { "epoch": 2.6, "learning_rate": 9.220861066963715e-07, "loss": 0.0206, "step": 1505 }, { "epoch": 2.6, "learning_rate": 9.142773653860948e-07, "loss": 0.0134, "step": 1506 }, { "epoch": 2.6, "learning_rate": 9.065002450262816e-07, "loss": 0.0173, "step": 1507 }, { "epoch": 2.6, "learning_rate": 8.987547726835622e-07, "loss": 0.0144, "step": 1508 }, { "epoch": 2.61, "learning_rate": 8.910409753144344e-07, "loss": 0.0132, "step": 1509 }, { "epoch": 2.61, "learning_rate": 8.833588797651438e-07, "loss": 0.0207, "step": 1510 }, { "epoch": 2.61, "learning_rate": 8.757085127716169e-07, "loss": 0.0172, "step": 1511 }, { "epoch": 2.61, "learning_rate": 8.680899009593469e-07, "loss": 0.015, "step": 1512 }, { "epoch": 2.61, "learning_rate": 8.605030708433149e-07, "loss": 0.0151, "step": 1513 }, { "epoch": 2.61, "learning_rate": 8.529480488278908e-07, "loss": 0.0197, "step": 1514 }, { "epoch": 2.62, "learning_rate": 8.454248612067395e-07, "loss": 0.0163, "step": 1515 }, { "epoch": 2.62, "learning_rate": 8.379335341627404e-07, "loss": 0.0123, "step": 1516 }, { "epoch": 2.62, "learning_rate": 8.304740937678835e-07, "loss": 0.0144, "step": 1517 }, { "epoch": 2.62, "learning_rate": 8.230465659831843e-07, "loss": 0.0166, "step": 1518 }, { "epoch": 2.62, "learning_rate": 8.156509766585929e-07, "loss": 0.0157, "step": 1519 }, { "epoch": 2.63, "learning_rate": 8.082873515329093e-07, "loss": 0.0118, "step": 1520 }, { "epoch": 2.63, "learning_rate": 8.009557162336823e-07, "loss": 0.0145, "step": 1521 }, { "epoch": 2.63, "learning_rate": 7.936560962771312e-07, "loss": 0.0161, "step": 1522 }, { "epoch": 2.63, "learning_rate": 7.863885170680486e-07, "loss": 0.0109, "step": 1523 }, { "epoch": 2.63, "learning_rate": 7.79153003899722e-07, "loss": 0.0144, "step": 1524 }, { "epoch": 2.63, "learning_rate": 7.719495819538325e-07, "loss": 0.0135, "step": 1525 }, { "epoch": 2.64, "learning_rate": 7.647782763003808e-07, "loss": 0.0226, "step": 1526 }, { "epoch": 2.64, "learning_rate": 7.576391118975912e-07, "loss": 0.0171, "step": 1527 }, { "epoch": 2.64, "learning_rate": 7.505321135918242e-07, "loss": 0.0114, "step": 1528 }, { "epoch": 2.64, "learning_rate": 7.434573061174966e-07, "loss": 0.0153, "step": 1529 }, { "epoch": 2.64, "learning_rate": 7.364147140969924e-07, "loss": 0.0176, "step": 1530 }, { "epoch": 2.64, "learning_rate": 7.294043620405721e-07, "loss": 0.0118, "step": 1531 }, { "epoch": 2.65, "learning_rate": 7.224262743462973e-07, "loss": 0.0148, "step": 1532 }, { "epoch": 2.65, "learning_rate": 7.154804752999344e-07, "loss": 0.0131, "step": 1533 }, { "epoch": 2.65, "learning_rate": 7.085669890748826e-07, "loss": 0.0154, "step": 1534 }, { "epoch": 2.65, "learning_rate": 7.016858397320758e-07, "loss": 0.0095, "step": 1535 }, { "epoch": 2.65, "learning_rate": 6.948370512199143e-07, "loss": 0.0122, "step": 1536 }, { "epoch": 2.65, "learning_rate": 6.880206473741646e-07, "loss": 0.0127, "step": 1537 }, { "epoch": 2.66, "learning_rate": 6.812366519178926e-07, "loss": 0.0149, "step": 1538 }, { "epoch": 2.66, "learning_rate": 6.744850884613674e-07, "loss": 0.0208, "step": 1539 }, { "epoch": 2.66, "learning_rate": 6.677659805019909e-07, "loss": 0.0185, "step": 1540 }, { "epoch": 2.66, "learning_rate": 6.610793514242075e-07, "loss": 0.0176, "step": 1541 }, { "epoch": 2.66, "learning_rate": 6.544252244994242e-07, "loss": 0.0163, "step": 1542 }, { "epoch": 2.66, "learning_rate": 6.478036228859363e-07, "loss": 0.0154, "step": 1543 }, { "epoch": 2.67, "learning_rate": 6.412145696288353e-07, "loss": 0.0143, "step": 1544 }, { "epoch": 2.67, "learning_rate": 6.346580876599395e-07, "loss": 0.015, "step": 1545 }, { "epoch": 2.67, "learning_rate": 6.281341997977075e-07, "loss": 0.0165, "step": 1546 }, { "epoch": 2.67, "learning_rate": 6.216429287471626e-07, "loss": 0.0095, "step": 1547 }, { "epoch": 2.67, "learning_rate": 6.151842970998112e-07, "loss": 0.0087, "step": 1548 }, { "epoch": 2.68, "learning_rate": 6.08758327333564e-07, "loss": 0.0111, "step": 1549 }, { "epoch": 2.68, "learning_rate": 6.023650418126592e-07, "loss": 0.0171, "step": 1550 }, { "epoch": 2.68, "learning_rate": 5.96004462787586e-07, "loss": 0.0148, "step": 1551 }, { "epoch": 2.68, "learning_rate": 5.896766123950048e-07, "loss": 0.0138, "step": 1552 }, { "epoch": 2.68, "learning_rate": 5.833815126576714e-07, "loss": 0.0102, "step": 1553 }, { "epoch": 2.68, "learning_rate": 5.771191854843561e-07, "loss": 0.0144, "step": 1554 }, { "epoch": 2.69, "learning_rate": 5.708896526697782e-07, "loss": 0.0131, "step": 1555 }, { "epoch": 2.69, "learning_rate": 5.646929358945152e-07, "loss": 0.0187, "step": 1556 }, { "epoch": 2.69, "learning_rate": 5.585290567249424e-07, "loss": 0.0158, "step": 1557 }, { "epoch": 2.69, "learning_rate": 5.523980366131454e-07, "loss": 0.0199, "step": 1558 }, { "epoch": 2.69, "learning_rate": 5.462998968968525e-07, "loss": 0.016, "step": 1559 }, { "epoch": 2.69, "learning_rate": 5.402346587993589e-07, "loss": 0.0149, "step": 1560 }, { "epoch": 2.7, "learning_rate": 5.342023434294552e-07, "loss": 0.0138, "step": 1561 }, { "epoch": 2.7, "learning_rate": 5.282029717813453e-07, "loss": 0.0159, "step": 1562 }, { "epoch": 2.7, "learning_rate": 5.222365647345862e-07, "loss": 0.0147, "step": 1563 }, { "epoch": 2.7, "learning_rate": 5.163031430540022e-07, "loss": 0.0176, "step": 1564 }, { "epoch": 2.7, "learning_rate": 5.104027273896239e-07, "loss": 0.0143, "step": 1565 }, { "epoch": 2.7, "learning_rate": 5.045353382766083e-07, "loss": 0.0175, "step": 1566 }, { "epoch": 2.71, "learning_rate": 4.98700996135173e-07, "loss": 0.0172, "step": 1567 }, { "epoch": 2.71, "learning_rate": 4.928997212705211e-07, "loss": 0.0153, "step": 1568 }, { "epoch": 2.71, "learning_rate": 4.871315338727711e-07, "loss": 0.0167, "step": 1569 }, { "epoch": 2.71, "learning_rate": 4.813964540168892e-07, "loss": 0.0187, "step": 1570 }, { "epoch": 2.71, "learning_rate": 4.7569450166261356e-07, "loss": 0.0168, "step": 1571 }, { "epoch": 2.72, "learning_rate": 4.700256966543981e-07, "loss": 0.0143, "step": 1572 }, { "epoch": 2.72, "learning_rate": 4.6439005872132457e-07, "loss": 0.0209, "step": 1573 }, { "epoch": 2.72, "learning_rate": 4.5878760747705033e-07, "loss": 0.0187, "step": 1574 }, { "epoch": 2.72, "learning_rate": 4.532183624197284e-07, "loss": 0.0185, "step": 1575 }, { "epoch": 2.72, "learning_rate": 4.476823429319499e-07, "loss": 0.0132, "step": 1576 }, { "epoch": 2.72, "learning_rate": 4.421795682806662e-07, "loss": 0.0122, "step": 1577 }, { "epoch": 2.73, "learning_rate": 4.3671005761713126e-07, "loss": 0.0172, "step": 1578 }, { "epoch": 2.73, "learning_rate": 4.312738299768293e-07, "loss": 0.0146, "step": 1579 }, { "epoch": 2.73, "learning_rate": 4.2587090427940736e-07, "loss": 0.0188, "step": 1580 }, { "epoch": 2.73, "learning_rate": 4.2050129932861394e-07, "loss": 0.0161, "step": 1581 }, { "epoch": 2.73, "learning_rate": 4.1516503381223483e-07, "loss": 0.0172, "step": 1582 }, { "epoch": 2.73, "learning_rate": 4.0986212630201974e-07, "loss": 0.0137, "step": 1583 }, { "epoch": 2.74, "learning_rate": 4.045925952536267e-07, "loss": 0.0175, "step": 1584 }, { "epoch": 2.74, "learning_rate": 3.9935645900654906e-07, "loss": 0.0143, "step": 1585 }, { "epoch": 2.74, "learning_rate": 3.941537357840652e-07, "loss": 0.0183, "step": 1586 }, { "epoch": 2.74, "learning_rate": 3.889844436931556e-07, "loss": 0.0152, "step": 1587 }, { "epoch": 2.74, "learning_rate": 3.838486007244613e-07, "loss": 0.0195, "step": 1588 }, { "epoch": 2.74, "learning_rate": 3.7874622475220336e-07, "loss": 0.0143, "step": 1589 }, { "epoch": 2.75, "learning_rate": 3.7367733353413036e-07, "loss": 0.0136, "step": 1590 }, { "epoch": 2.75, "learning_rate": 3.686419447114531e-07, "loss": 0.012, "step": 1591 }, { "epoch": 2.75, "learning_rate": 3.6364007580878766e-07, "loss": 0.0152, "step": 1592 }, { "epoch": 2.75, "learning_rate": 3.58671744234087e-07, "loss": 0.0141, "step": 1593 }, { "epoch": 2.75, "learning_rate": 3.537369672785873e-07, "loss": 0.0146, "step": 1594 }, { "epoch": 2.75, "learning_rate": 3.4883576211674266e-07, "loss": 0.0158, "step": 1595 }, { "epoch": 2.76, "learning_rate": 3.439681458061717e-07, "loss": 0.0173, "step": 1596 }, { "epoch": 2.76, "learning_rate": 3.3913413528758877e-07, "loss": 0.0187, "step": 1597 }, { "epoch": 2.76, "learning_rate": 3.343337473847519e-07, "loss": 0.0111, "step": 1598 }, { "epoch": 2.76, "learning_rate": 3.295669988044059e-07, "loss": 0.0149, "step": 1599 }, { "epoch": 2.76, "learning_rate": 3.248339061362182e-07, "loss": 0.0133, "step": 1600 }, { "epoch": 2.77, "learning_rate": 3.2013448585272333e-07, "loss": 0.0114, "step": 1601 }, { "epoch": 2.77, "learning_rate": 3.1546875430926717e-07, "loss": 0.0203, "step": 1602 }, { "epoch": 2.77, "learning_rate": 3.1083672774395055e-07, "loss": 0.0116, "step": 1603 }, { "epoch": 2.77, "learning_rate": 3.0623842227756715e-07, "loss": 0.0103, "step": 1604 }, { "epoch": 2.77, "learning_rate": 3.016738539135566e-07, "loss": 0.0176, "step": 1605 }, { "epoch": 2.77, "learning_rate": 2.971430385379381e-07, "loss": 0.0159, "step": 1606 }, { "epoch": 2.78, "learning_rate": 2.926459919192659e-07, "loss": 0.012, "step": 1607 }, { "epoch": 2.78, "learning_rate": 2.8818272970856596e-07, "loss": 0.0128, "step": 1608 }, { "epoch": 2.78, "learning_rate": 2.837532674392862e-07, "loss": 0.0168, "step": 1609 }, { "epoch": 2.78, "learning_rate": 2.793576205272386e-07, "loss": 0.0107, "step": 1610 }, { "epoch": 2.78, "learning_rate": 2.749958042705514e-07, "loss": 0.0174, "step": 1611 }, { "epoch": 2.78, "learning_rate": 2.7066783384960825e-07, "loss": 0.0146, "step": 1612 }, { "epoch": 2.79, "learning_rate": 2.6637372432700483e-07, "loss": 0.0222, "step": 1613 }, { "epoch": 2.79, "learning_rate": 2.621134906474843e-07, "loss": 0.0148, "step": 1614 }, { "epoch": 2.79, "learning_rate": 2.578871476378997e-07, "loss": 0.0147, "step": 1615 }, { "epoch": 2.79, "learning_rate": 2.536947100071463e-07, "loss": 0.017, "step": 1616 }, { "epoch": 2.79, "learning_rate": 2.4953619234612816e-07, "loss": 0.013, "step": 1617 }, { "epoch": 2.79, "learning_rate": 2.4541160912768814e-07, "loss": 0.0175, "step": 1618 }, { "epoch": 2.8, "learning_rate": 2.41320974706577e-07, "loss": 0.0154, "step": 1619 }, { "epoch": 2.8, "learning_rate": 2.3726430331938665e-07, "loss": 0.0159, "step": 1620 }, { "epoch": 2.8, "learning_rate": 2.332416090845102e-07, "loss": 0.018, "step": 1621 }, { "epoch": 2.8, "learning_rate": 2.2925290600208981e-07, "loss": 0.0172, "step": 1622 }, { "epoch": 2.8, "learning_rate": 2.2529820795397228e-07, "loss": 0.0117, "step": 1623 }, { "epoch": 2.8, "learning_rate": 2.2137752870365125e-07, "loss": 0.0161, "step": 1624 }, { "epoch": 2.81, "learning_rate": 2.1749088189622848e-07, "loss": 0.0165, "step": 1625 }, { "epoch": 2.81, "learning_rate": 2.136382810583637e-07, "loss": 0.0125, "step": 1626 }, { "epoch": 2.81, "learning_rate": 2.0981973959822378e-07, "loss": 0.0113, "step": 1627 }, { "epoch": 2.81, "learning_rate": 2.060352708054425e-07, "loss": 0.015, "step": 1628 }, { "epoch": 2.81, "learning_rate": 2.0228488785106636e-07, "loss": 0.0179, "step": 1629 }, { "epoch": 2.82, "learning_rate": 1.9856860378751896e-07, "loss": 0.0132, "step": 1630 }, { "epoch": 2.82, "learning_rate": 1.9488643154854435e-07, "loss": 0.0119, "step": 1631 }, { "epoch": 2.82, "learning_rate": 1.912383839491727e-07, "loss": 0.0176, "step": 1632 }, { "epoch": 2.82, "learning_rate": 1.8762447368566582e-07, "loss": 0.0162, "step": 1633 }, { "epoch": 2.82, "learning_rate": 1.8404471333548058e-07, "loss": 0.0124, "step": 1634 }, { "epoch": 2.82, "learning_rate": 1.8049911535722108e-07, "loss": 0.022, "step": 1635 }, { "epoch": 2.83, "learning_rate": 1.7698769209059773e-07, "loss": 0.0166, "step": 1636 }, { "epoch": 2.83, "learning_rate": 1.7351045575638047e-07, "loss": 0.0125, "step": 1637 }, { "epoch": 2.83, "learning_rate": 1.700674184563622e-07, "loss": 0.0123, "step": 1638 }, { "epoch": 2.83, "learning_rate": 1.6665859217331105e-07, "loss": 0.0154, "step": 1639 }, { "epoch": 2.83, "learning_rate": 1.6328398877093144e-07, "loss": 0.0138, "step": 1640 }, { "epoch": 2.83, "learning_rate": 1.5994361999382202e-07, "loss": 0.0143, "step": 1641 }, { "epoch": 2.84, "learning_rate": 1.566374974674345e-07, "loss": 0.0156, "step": 1642 }, { "epoch": 2.84, "learning_rate": 1.5336563269803372e-07, "loss": 0.0184, "step": 1643 }, { "epoch": 2.84, "learning_rate": 1.5012803707266098e-07, "loss": 0.0231, "step": 1644 }, { "epoch": 2.84, "learning_rate": 1.4692472185908635e-07, "loss": 0.0155, "step": 1645 }, { "epoch": 2.84, "learning_rate": 1.437556982057764e-07, "loss": 0.013, "step": 1646 }, { "epoch": 2.84, "learning_rate": 1.4062097714185207e-07, "loss": 0.0185, "step": 1647 }, { "epoch": 2.85, "learning_rate": 1.375205695770532e-07, "loss": 0.0203, "step": 1648 }, { "epoch": 2.85, "learning_rate": 1.344544863016961e-07, "loss": 0.0136, "step": 1649 }, { "epoch": 2.85, "learning_rate": 1.3142273798664174e-07, "loss": 0.0182, "step": 1650 }, { "epoch": 2.85, "learning_rate": 1.2842533518325318e-07, "loss": 0.0178, "step": 1651 }, { "epoch": 2.85, "learning_rate": 1.2546228832336028e-07, "loss": 0.011, "step": 1652 }, { "epoch": 2.85, "learning_rate": 1.225336077192274e-07, "loss": 0.0159, "step": 1653 }, { "epoch": 2.86, "learning_rate": 1.196393035635113e-07, "loss": 0.0128, "step": 1654 }, { "epoch": 2.86, "learning_rate": 1.167793859292321e-07, "loss": 0.015, "step": 1655 }, { "epoch": 2.86, "learning_rate": 1.1395386476973024e-07, "loss": 0.0159, "step": 1656 }, { "epoch": 2.86, "learning_rate": 1.1116274991864073e-07, "loss": 0.0183, "step": 1657 }, { "epoch": 2.86, "learning_rate": 1.0840605108985214e-07, "loss": 0.0145, "step": 1658 }, { "epoch": 2.87, "learning_rate": 1.0568377787747663e-07, "loss": 0.0218, "step": 1659 }, { "epoch": 2.87, "learning_rate": 1.0299593975581446e-07, "loss": 0.0118, "step": 1660 }, { "epoch": 2.87, "learning_rate": 1.003425460793217e-07, "loss": 0.0117, "step": 1661 }, { "epoch": 2.87, "learning_rate": 9.772360608257814e-08, "loss": 0.0145, "step": 1662 }, { "epoch": 2.87, "learning_rate": 9.513912888025611e-08, "loss": 0.0139, "step": 1663 }, { "epoch": 2.87, "learning_rate": 9.258912346708394e-08, "loss": 0.0141, "step": 1664 }, { "epoch": 2.88, "learning_rate": 9.00735987178214e-08, "loss": 0.0147, "step": 1665 }, { "epoch": 2.88, "learning_rate": 8.759256338722321e-08, "loss": 0.0169, "step": 1666 }, { "epoch": 2.88, "learning_rate": 8.514602611001344e-08, "loss": 0.0131, "step": 1667 }, { "epoch": 2.88, "learning_rate": 8.273399540085104e-08, "loss": 0.0215, "step": 1668 }, { "epoch": 2.88, "learning_rate": 8.035647965430215e-08, "loss": 0.0142, "step": 1669 }, { "epoch": 2.88, "learning_rate": 7.801348714481238e-08, "loss": 0.0171, "step": 1670 }, { "epoch": 2.89, "learning_rate": 7.570502602667451e-08, "loss": 0.0154, "step": 1671 }, { "epoch": 2.89, "learning_rate": 7.343110433400191e-08, "loss": 0.013, "step": 1672 }, { "epoch": 2.89, "learning_rate": 7.119172998070412e-08, "loss": 0.0157, "step": 1673 }, { "epoch": 2.89, "learning_rate": 6.898691076045128e-08, "loss": 0.0157, "step": 1674 }, { "epoch": 2.89, "learning_rate": 6.68166543466553e-08, "loss": 0.014, "step": 1675 }, { "epoch": 2.89, "learning_rate": 6.468096829243654e-08, "loss": 0.0146, "step": 1676 }, { "epoch": 2.9, "learning_rate": 6.25798600306049e-08, "loss": 0.0114, "step": 1677 }, { "epoch": 2.9, "learning_rate": 6.051333687362327e-08, "loss": 0.013, "step": 1678 }, { "epoch": 2.9, "learning_rate": 5.848140601359409e-08, "loss": 0.0189, "step": 1679 }, { "epoch": 2.9, "learning_rate": 5.648407452222504e-08, "loss": 0.0191, "step": 1680 }, { "epoch": 2.9, "learning_rate": 5.4521349350808996e-08, "loss": 0.0132, "step": 1681 }, { "epoch": 2.91, "learning_rate": 5.259323733019961e-08, "loss": 0.018, "step": 1682 }, { "epoch": 2.91, "learning_rate": 5.0699745170785796e-08, "loss": 0.0146, "step": 1683 }, { "epoch": 2.91, "learning_rate": 4.884087946246951e-08, "loss": 0.0131, "step": 1684 }, { "epoch": 2.91, "learning_rate": 4.701664667464245e-08, "loss": 0.0171, "step": 1685 }, { "epoch": 2.91, "learning_rate": 4.5227053156164936e-08, "loss": 0.0182, "step": 1686 }, { "epoch": 2.91, "learning_rate": 4.347210513534039e-08, "loss": 0.0166, "step": 1687 }, { "epoch": 2.92, "learning_rate": 4.175180871989759e-08, "loss": 0.0157, "step": 1688 }, { "epoch": 2.92, "learning_rate": 4.006616989696621e-08, "loss": 0.0145, "step": 1689 }, { "epoch": 2.92, "learning_rate": 3.841519453305909e-08, "loss": 0.0166, "step": 1690 }, { "epoch": 2.92, "learning_rate": 3.679888837404999e-08, "loss": 0.0166, "step": 1691 }, { "epoch": 2.92, "learning_rate": 3.521725704515255e-08, "loss": 0.0156, "step": 1692 }, { "epoch": 2.92, "learning_rate": 3.367030605090249e-08, "loss": 0.0159, "step": 1693 }, { "epoch": 2.93, "learning_rate": 3.215804077513762e-08, "loss": 0.0201, "step": 1694 }, { "epoch": 2.93, "learning_rate": 3.068046648098122e-08, "loss": 0.0148, "step": 1695 }, { "epoch": 2.93, "learning_rate": 2.9237588310819797e-08, "loss": 0.0192, "step": 1696 }, { "epoch": 2.93, "learning_rate": 2.7829411286287577e-08, "loss": 0.0122, "step": 1697 }, { "epoch": 2.93, "learning_rate": 2.645594030824872e-08, "loss": 0.0176, "step": 1698 }, { "epoch": 2.93, "learning_rate": 2.5117180156781772e-08, "loss": 0.0123, "step": 1699 }, { "epoch": 2.94, "learning_rate": 2.381313549115971e-08, "loss": 0.0126, "step": 1700 }, { "epoch": 2.94, "learning_rate": 2.254381084983659e-08, "loss": 0.016, "step": 1701 }, { "epoch": 2.94, "learning_rate": 2.13092106504309e-08, "loss": 0.0186, "step": 1702 }, { "epoch": 2.94, "learning_rate": 2.010933918970781e-08, "loss": 0.0105, "step": 1703 }, { "epoch": 2.94, "learning_rate": 1.8944200643570275e-08, "loss": 0.0216, "step": 1704 }, { "epoch": 2.94, "learning_rate": 1.781379906703573e-08, "loss": 0.0147, "step": 1705 }, { "epoch": 2.95, "learning_rate": 1.6718138394232753e-08, "loss": 0.0167, "step": 1706 }, { "epoch": 2.95, "learning_rate": 1.565722243837553e-08, "loss": 0.015, "step": 1707 }, { "epoch": 2.95, "learning_rate": 1.4631054891760531e-08, "loss": 0.0147, "step": 1708 }, { "epoch": 2.95, "learning_rate": 1.3639639325748744e-08, "loss": 0.0187, "step": 1709 }, { "epoch": 2.95, "learning_rate": 1.2682979190754563e-08, "loss": 0.0193, "step": 1710 }, { "epoch": 2.96, "learning_rate": 1.176107781623026e-08, "loss": 0.0141, "step": 1711 }, { "epoch": 2.96, "learning_rate": 1.0873938410662643e-08, "loss": 0.019, "step": 1712 }, { "epoch": 2.96, "learning_rate": 1.0021564061554189e-08, "loss": 0.0147, "step": 1713 }, { "epoch": 2.96, "learning_rate": 9.203957735411938e-09, "loss": 0.0097, "step": 1714 }, { "epoch": 2.96, "learning_rate": 8.421122277746385e-09, "loss": 0.0187, "step": 1715 }, { "epoch": 2.96, "learning_rate": 7.673060413050382e-09, "loss": 0.0119, "step": 1716 }, { "epoch": 2.97, "learning_rate": 6.959774744796921e-09, "loss": 0.0125, "step": 1717 }, { "epoch": 2.97, "learning_rate": 6.281267755425813e-09, "loss": 0.0161, "step": 1718 }, { "epoch": 2.97, "learning_rate": 5.637541806340352e-09, "loss": 0.0136, "step": 1719 }, { "epoch": 2.97, "learning_rate": 5.028599137891777e-09, "loss": 0.0156, "step": 1720 }, { "epoch": 2.97, "learning_rate": 4.454441869377047e-09, "loss": 0.0175, "step": 1721 }, { "epoch": 2.97, "learning_rate": 3.9150719990321826e-09, "loss": 0.013, "step": 1722 }, { "epoch": 2.98, "learning_rate": 3.410491404017835e-09, "loss": 0.0157, "step": 1723 }, { "epoch": 2.98, "learning_rate": 2.9407018404237207e-09, "loss": 0.0163, "step": 1724 }, { "epoch": 2.98, "learning_rate": 2.5057049432519744e-09, "loss": 0.0177, "step": 1725 }, { "epoch": 2.98, "learning_rate": 2.1055022264182546e-09, "loss": 0.0143, "step": 1726 }, { "epoch": 2.98, "learning_rate": 1.7400950827439755e-09, "loss": 0.0129, "step": 1727 }, { "epoch": 2.98, "learning_rate": 1.4094847839529745e-09, "loss": 0.0131, "step": 1728 }, { "epoch": 2.99, "learning_rate": 1.1136724806637411e-09, "loss": 0.0202, "step": 1729 }, { "epoch": 2.99, "learning_rate": 8.526592023905267e-10, "loss": 0.0186, "step": 1730 }, { "epoch": 2.99, "learning_rate": 6.264458575333532e-10, "loss": 0.0148, "step": 1731 }, { "epoch": 2.99, "learning_rate": 4.350332333813434e-10, "loss": 0.0157, "step": 1732 }, { "epoch": 2.99, "learning_rate": 2.784219961060597e-10, "loss": 0.0166, "step": 1733 }, { "epoch": 2.99, "learning_rate": 1.5661269076039377e-10, "loss": 0.0127, "step": 1734 }, { "epoch": 3.0, "learning_rate": 6.960574127412578e-11, "loss": 0.0105, "step": 1735 }, { "epoch": 3.0, "learning_rate": 1.74014504594755e-11, "loss": 0.0175, "step": 1736 }, { "epoch": 3.0, "learning_rate": 0.0, "loss": 0.0127, "step": 1737 }, { "epoch": 3.0, "step": 1737, "total_flos": 4.942996663435264e+16, "train_loss": 0.03028348059096965, "train_runtime": 1221.8767, "train_samples_per_second": 45.427, "train_steps_per_second": 1.422 } ], "max_steps": 1737, "num_train_epochs": 3, "total_flos": 4.942996663435264e+16, "trial_name": null, "trial_params": null }