{ "best_metric": null, "best_model_checkpoint": null, "epoch": 4.0, "eval_steps": 500, "global_step": 13312, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 4.999812199519231e-05, "loss": 2.1019, "step": 1 }, { "epoch": 0.0, "learning_rate": 4.999624399038462e-05, "loss": 2.1138, "step": 2 }, { "epoch": 0.0, "learning_rate": 4.9994365985576923e-05, "loss": 2.4241, "step": 3 }, { "epoch": 0.0, "learning_rate": 4.9994365985576923e-05, "loss": 2.4366, "step": 4 }, { "epoch": 0.0, "learning_rate": 4.999248798076923e-05, "loss": 2.4304, "step": 5 }, { "epoch": 0.0, "learning_rate": 4.999060997596154e-05, "loss": 2.0541, "step": 6 }, { "epoch": 0.0, "learning_rate": 4.9988731971153845e-05, "loss": 2.1979, "step": 7 }, { "epoch": 0.0, "learning_rate": 4.9986853966346156e-05, "loss": 2.0999, "step": 8 }, { "epoch": 0.0, "learning_rate": 4.998497596153847e-05, "loss": 2.157, "step": 9 }, { "epoch": 0.0, "learning_rate": 4.998309795673077e-05, "loss": 2.8398, "step": 10 }, { "epoch": 0.0, "learning_rate": 4.998121995192308e-05, "loss": 2.2864, "step": 11 }, { "epoch": 0.0, "learning_rate": 4.998121995192308e-05, "loss": 2.5684, "step": 12 }, { "epoch": 0.0, "learning_rate": 4.997934194711539e-05, "loss": 2.4182, "step": 13 }, { "epoch": 0.0, "learning_rate": 4.9977463942307694e-05, "loss": 2.1895, "step": 14 }, { "epoch": 0.0, "learning_rate": 4.9975585937500005e-05, "loss": 2.2835, "step": 15 }, { "epoch": 0.0, "learning_rate": 4.997370793269231e-05, "loss": 2.3528, "step": 16 }, { "epoch": 0.01, "learning_rate": 4.9971829927884615e-05, "loss": 2.0907, "step": 17 }, { "epoch": 0.01, "learning_rate": 4.9969951923076926e-05, "loss": 2.417, "step": 18 }, { "epoch": 0.01, "learning_rate": 4.996807391826923e-05, "loss": 2.8072, "step": 19 }, { "epoch": 0.01, "learning_rate": 4.9966195913461536e-05, "loss": 2.1683, "step": 20 }, { "epoch": 0.01, "learning_rate": 4.996431790865385e-05, "loss": 2.2132, "step": 21 }, { "epoch": 0.01, "learning_rate": 4.996243990384616e-05, "loss": 2.3581, "step": 22 }, { "epoch": 0.01, "learning_rate": 4.9960561899038464e-05, "loss": 2.3062, "step": 23 }, { "epoch": 0.01, "learning_rate": 4.9958683894230775e-05, "loss": 2.209, "step": 24 }, { "epoch": 0.01, "learning_rate": 4.995680588942308e-05, "loss": 1.9566, "step": 25 }, { "epoch": 0.01, "learning_rate": 4.9954927884615385e-05, "loss": 2.174, "step": 26 }, { "epoch": 0.01, "learning_rate": 4.9953049879807696e-05, "loss": 2.2733, "step": 27 }, { "epoch": 0.01, "learning_rate": 4.9951171875e-05, "loss": 2.4282, "step": 28 }, { "epoch": 0.01, "learning_rate": 4.994929387019231e-05, "loss": 2.1145, "step": 29 }, { "epoch": 0.01, "learning_rate": 4.994741586538462e-05, "loss": 2.1823, "step": 30 }, { "epoch": 0.01, "learning_rate": 4.994553786057692e-05, "loss": 2.4109, "step": 31 }, { "epoch": 0.01, "learning_rate": 4.9943659855769234e-05, "loss": 2.3519, "step": 32 }, { "epoch": 0.01, "learning_rate": 4.994178185096154e-05, "loss": 2.1474, "step": 33 }, { "epoch": 0.01, "learning_rate": 4.993990384615384e-05, "loss": 2.135, "step": 34 }, { "epoch": 0.01, "learning_rate": 4.993802584134616e-05, "loss": 1.9446, "step": 35 }, { "epoch": 0.01, "learning_rate": 4.9936147836538466e-05, "loss": 2.2027, "step": 36 }, { "epoch": 0.01, "learning_rate": 4.993426983173077e-05, "loss": 2.203, "step": 37 }, { "epoch": 0.01, "learning_rate": 4.993239182692308e-05, "loss": 2.1324, "step": 38 }, { "epoch": 0.01, "learning_rate": 4.993051382211539e-05, "loss": 2.0978, "step": 39 }, { "epoch": 0.01, "learning_rate": 4.992863581730769e-05, "loss": 2.4978, "step": 40 }, { "epoch": 0.01, "learning_rate": 4.9926757812500004e-05, "loss": 2.1308, "step": 41 }, { "epoch": 0.01, "learning_rate": 4.992487980769231e-05, "loss": 2.6108, "step": 42 }, { "epoch": 0.01, "learning_rate": 4.992300180288462e-05, "loss": 2.0706, "step": 43 }, { "epoch": 0.01, "learning_rate": 4.9921123798076925e-05, "loss": 2.2108, "step": 44 }, { "epoch": 0.01, "learning_rate": 4.991924579326923e-05, "loss": 2.2057, "step": 45 }, { "epoch": 0.01, "learning_rate": 4.991736778846154e-05, "loss": 2.1869, "step": 46 }, { "epoch": 0.01, "learning_rate": 4.9915489783653846e-05, "loss": 2.3741, "step": 47 }, { "epoch": 0.01, "learning_rate": 4.991361177884616e-05, "loss": 2.1261, "step": 48 }, { "epoch": 0.01, "learning_rate": 4.991173377403847e-05, "loss": 2.0647, "step": 49 }, { "epoch": 0.02, "learning_rate": 4.9909855769230774e-05, "loss": 2.3066, "step": 50 }, { "epoch": 0.02, "learning_rate": 4.990797776442308e-05, "loss": 2.3429, "step": 51 }, { "epoch": 0.02, "learning_rate": 4.990609975961539e-05, "loss": 2.0841, "step": 52 }, { "epoch": 0.02, "learning_rate": 4.9904221754807695e-05, "loss": 2.2522, "step": 53 }, { "epoch": 0.02, "learning_rate": 4.990234375e-05, "loss": 2.2368, "step": 54 }, { "epoch": 0.02, "learning_rate": 4.990046574519231e-05, "loss": 2.2818, "step": 55 }, { "epoch": 0.02, "learning_rate": 4.9898587740384616e-05, "loss": 2.3308, "step": 56 }, { "epoch": 0.02, "learning_rate": 4.989670973557693e-05, "loss": 2.3175, "step": 57 }, { "epoch": 0.02, "learning_rate": 4.989483173076923e-05, "loss": 2.2404, "step": 58 }, { "epoch": 0.02, "learning_rate": 4.989295372596154e-05, "loss": 2.1038, "step": 59 }, { "epoch": 0.02, "learning_rate": 4.989107572115385e-05, "loss": 2.0055, "step": 60 }, { "epoch": 0.02, "learning_rate": 4.988919771634616e-05, "loss": 2.201, "step": 61 }, { "epoch": 0.02, "learning_rate": 4.9887319711538465e-05, "loss": 2.2555, "step": 62 }, { "epoch": 0.02, "learning_rate": 4.9885441706730776e-05, "loss": 2.1974, "step": 63 }, { "epoch": 0.02, "learning_rate": 4.988356370192308e-05, "loss": 2.3135, "step": 64 }, { "epoch": 0.02, "learning_rate": 4.9881685697115386e-05, "loss": 2.0469, "step": 65 }, { "epoch": 0.02, "learning_rate": 4.98798076923077e-05, "loss": 2.114, "step": 66 }, { "epoch": 0.02, "learning_rate": 4.98779296875e-05, "loss": 2.0074, "step": 67 }, { "epoch": 0.02, "learning_rate": 4.987605168269231e-05, "loss": 2.7015, "step": 68 }, { "epoch": 0.02, "learning_rate": 4.987417367788462e-05, "loss": 2.0351, "step": 69 }, { "epoch": 0.02, "learning_rate": 4.987229567307692e-05, "loss": 2.0542, "step": 70 }, { "epoch": 0.02, "learning_rate": 4.987041766826923e-05, "loss": 2.4132, "step": 71 }, { "epoch": 0.02, "learning_rate": 4.986853966346154e-05, "loss": 1.8847, "step": 72 }, { "epoch": 0.02, "learning_rate": 4.9866661658653844e-05, "loss": 2.212, "step": 73 }, { "epoch": 0.02, "learning_rate": 4.9864783653846156e-05, "loss": 2.2756, "step": 74 }, { "epoch": 0.02, "learning_rate": 4.986290564903847e-05, "loss": 1.835, "step": 75 }, { "epoch": 0.02, "learning_rate": 4.986102764423077e-05, "loss": 2.4462, "step": 76 }, { "epoch": 0.02, "learning_rate": 4.9859149639423084e-05, "loss": 2.0056, "step": 77 }, { "epoch": 0.02, "learning_rate": 4.985727163461539e-05, "loss": 1.9735, "step": 78 }, { "epoch": 0.02, "learning_rate": 4.9855393629807693e-05, "loss": 2.3894, "step": 79 }, { "epoch": 0.02, "learning_rate": 4.9853515625000005e-05, "loss": 2.1579, "step": 80 }, { "epoch": 0.02, "learning_rate": 4.985163762019231e-05, "loss": 2.1393, "step": 81 }, { "epoch": 0.02, "learning_rate": 4.9849759615384615e-05, "loss": 2.2285, "step": 82 }, { "epoch": 0.02, "learning_rate": 4.9847881610576926e-05, "loss": 2.1096, "step": 83 }, { "epoch": 0.03, "learning_rate": 4.984600360576923e-05, "loss": 2.4464, "step": 84 }, { "epoch": 0.03, "learning_rate": 4.9844125600961536e-05, "loss": 2.4865, "step": 85 }, { "epoch": 0.03, "learning_rate": 4.984224759615385e-05, "loss": 2.1318, "step": 86 }, { "epoch": 0.03, "learning_rate": 4.984036959134616e-05, "loss": 2.253, "step": 87 }, { "epoch": 0.03, "learning_rate": 4.9838491586538463e-05, "loss": 2.1453, "step": 88 }, { "epoch": 0.03, "learning_rate": 4.9836613581730775e-05, "loss": 2.5372, "step": 89 }, { "epoch": 0.03, "learning_rate": 4.983473557692308e-05, "loss": 2.1795, "step": 90 }, { "epoch": 0.03, "learning_rate": 4.9832857572115385e-05, "loss": 2.3242, "step": 91 }, { "epoch": 0.03, "learning_rate": 4.9830979567307696e-05, "loss": 2.1148, "step": 92 }, { "epoch": 0.03, "learning_rate": 4.98291015625e-05, "loss": 2.2242, "step": 93 }, { "epoch": 0.03, "learning_rate": 4.982722355769231e-05, "loss": 2.3132, "step": 94 }, { "epoch": 0.03, "learning_rate": 4.982534555288462e-05, "loss": 2.4635, "step": 95 }, { "epoch": 0.03, "learning_rate": 4.982346754807692e-05, "loss": 2.4463, "step": 96 }, { "epoch": 0.03, "learning_rate": 4.9821589543269234e-05, "loss": 2.1572, "step": 97 }, { "epoch": 0.03, "learning_rate": 4.981971153846154e-05, "loss": 2.2079, "step": 98 }, { "epoch": 0.03, "learning_rate": 4.981783353365384e-05, "loss": 2.116, "step": 99 }, { "epoch": 0.03, "learning_rate": 4.981595552884616e-05, "loss": 1.9718, "step": 100 }, { "epoch": 0.03, "learning_rate": 4.9814077524038466e-05, "loss": 2.2291, "step": 101 }, { "epoch": 0.03, "learning_rate": 4.981219951923077e-05, "loss": 2.1841, "step": 102 }, { "epoch": 0.03, "learning_rate": 4.981032151442308e-05, "loss": 2.2002, "step": 103 }, { "epoch": 0.03, "learning_rate": 4.980844350961539e-05, "loss": 2.2517, "step": 104 }, { "epoch": 0.03, "learning_rate": 4.980656550480769e-05, "loss": 2.1723, "step": 105 }, { "epoch": 0.03, "learning_rate": 4.9804687500000004e-05, "loss": 2.2835, "step": 106 }, { "epoch": 0.03, "learning_rate": 4.980280949519231e-05, "loss": 2.4319, "step": 107 }, { "epoch": 0.03, "learning_rate": 4.980093149038462e-05, "loss": 2.1771, "step": 108 }, { "epoch": 0.03, "learning_rate": 4.9799053485576925e-05, "loss": 2.2765, "step": 109 }, { "epoch": 0.03, "learning_rate": 4.979717548076923e-05, "loss": 2.1565, "step": 110 }, { "epoch": 0.03, "learning_rate": 4.979529747596154e-05, "loss": 2.2268, "step": 111 }, { "epoch": 0.03, "learning_rate": 4.9793419471153846e-05, "loss": 2.1223, "step": 112 }, { "epoch": 0.03, "learning_rate": 4.979154146634616e-05, "loss": 2.0351, "step": 113 }, { "epoch": 0.03, "learning_rate": 4.978966346153847e-05, "loss": 2.0348, "step": 114 }, { "epoch": 0.03, "learning_rate": 4.9787785456730774e-05, "loss": 2.3198, "step": 115 }, { "epoch": 0.03, "learning_rate": 4.978590745192308e-05, "loss": 2.2666, "step": 116 }, { "epoch": 0.04, "learning_rate": 4.978402944711539e-05, "loss": 2.2514, "step": 117 }, { "epoch": 0.04, "learning_rate": 4.9782151442307695e-05, "loss": 2.294, "step": 118 }, { "epoch": 0.04, "learning_rate": 4.97802734375e-05, "loss": 2.2837, "step": 119 }, { "epoch": 0.04, "learning_rate": 4.977839543269231e-05, "loss": 2.1227, "step": 120 }, { "epoch": 0.04, "learning_rate": 4.9776517427884616e-05, "loss": 2.328, "step": 121 }, { "epoch": 0.04, "learning_rate": 4.977463942307693e-05, "loss": 2.2519, "step": 122 }, { "epoch": 0.04, "learning_rate": 4.977276141826923e-05, "loss": 2.1454, "step": 123 }, { "epoch": 0.04, "learning_rate": 4.977088341346154e-05, "loss": 2.1206, "step": 124 }, { "epoch": 0.04, "learning_rate": 4.976900540865385e-05, "loss": 2.2643, "step": 125 }, { "epoch": 0.04, "learning_rate": 4.976712740384616e-05, "loss": 2.2422, "step": 126 }, { "epoch": 0.04, "learning_rate": 4.9765249399038465e-05, "loss": 2.2416, "step": 127 }, { "epoch": 0.04, "learning_rate": 4.9763371394230776e-05, "loss": 2.161, "step": 128 }, { "epoch": 0.04, "learning_rate": 4.976149338942308e-05, "loss": 2.1795, "step": 129 }, { "epoch": 0.04, "learning_rate": 4.9759615384615386e-05, "loss": 2.1558, "step": 130 }, { "epoch": 0.04, "learning_rate": 4.97577373798077e-05, "loss": 2.0926, "step": 131 }, { "epoch": 0.04, "learning_rate": 4.9755859375e-05, "loss": 2.0255, "step": 132 }, { "epoch": 0.04, "learning_rate": 4.975398137019231e-05, "loss": 2.0158, "step": 133 }, { "epoch": 0.04, "learning_rate": 4.975210336538462e-05, "loss": 2.2757, "step": 134 }, { "epoch": 0.04, "learning_rate": 4.975022536057692e-05, "loss": 2.0416, "step": 135 }, { "epoch": 0.04, "learning_rate": 4.9748347355769235e-05, "loss": 1.9642, "step": 136 }, { "epoch": 0.04, "learning_rate": 4.974646935096154e-05, "loss": 2.0433, "step": 137 }, { "epoch": 0.04, "learning_rate": 4.9744591346153844e-05, "loss": 2.1365, "step": 138 }, { "epoch": 0.04, "learning_rate": 4.9742713341346156e-05, "loss": 2.1576, "step": 139 }, { "epoch": 0.04, "learning_rate": 4.974083533653847e-05, "loss": 2.0088, "step": 140 }, { "epoch": 0.04, "learning_rate": 4.973895733173077e-05, "loss": 2.4674, "step": 141 }, { "epoch": 0.04, "learning_rate": 4.9737079326923084e-05, "loss": 2.1261, "step": 142 }, { "epoch": 0.04, "learning_rate": 4.973520132211539e-05, "loss": 2.1921, "step": 143 }, { "epoch": 0.04, "learning_rate": 4.973332331730769e-05, "loss": 2.5906, "step": 144 }, { "epoch": 0.04, "learning_rate": 4.9731445312500005e-05, "loss": 2.1653, "step": 145 }, { "epoch": 0.04, "learning_rate": 4.972956730769231e-05, "loss": 2.1459, "step": 146 }, { "epoch": 0.04, "learning_rate": 4.9727689302884614e-05, "loss": 2.0669, "step": 147 }, { "epoch": 0.04, "learning_rate": 4.9725811298076926e-05, "loss": 2.3641, "step": 148 }, { "epoch": 0.04, "learning_rate": 4.972393329326923e-05, "loss": 2.2283, "step": 149 }, { "epoch": 0.05, "learning_rate": 4.9722055288461536e-05, "loss": 2.6502, "step": 150 }, { "epoch": 0.05, "learning_rate": 4.972017728365385e-05, "loss": 2.2645, "step": 151 }, { "epoch": 0.05, "learning_rate": 4.971829927884616e-05, "loss": 2.2556, "step": 152 }, { "epoch": 0.05, "learning_rate": 4.971642127403846e-05, "loss": 2.0985, "step": 153 }, { "epoch": 0.05, "learning_rate": 4.9714543269230775e-05, "loss": 2.0752, "step": 154 }, { "epoch": 0.05, "learning_rate": 4.971266526442308e-05, "loss": 2.1127, "step": 155 }, { "epoch": 0.05, "learning_rate": 4.971078725961539e-05, "loss": 2.2283, "step": 156 }, { "epoch": 0.05, "learning_rate": 4.9708909254807696e-05, "loss": 2.0135, "step": 157 }, { "epoch": 0.05, "learning_rate": 4.970703125e-05, "loss": 2.3838, "step": 158 }, { "epoch": 0.05, "learning_rate": 4.970515324519231e-05, "loss": 1.9526, "step": 159 }, { "epoch": 0.05, "learning_rate": 4.970327524038462e-05, "loss": 2.1718, "step": 160 }, { "epoch": 0.05, "learning_rate": 4.970139723557692e-05, "loss": 2.0943, "step": 161 }, { "epoch": 0.05, "learning_rate": 4.9699519230769233e-05, "loss": 2.14, "step": 162 }, { "epoch": 0.05, "learning_rate": 4.969764122596154e-05, "loss": 2.2675, "step": 163 }, { "epoch": 0.05, "learning_rate": 4.969576322115384e-05, "loss": 2.1059, "step": 164 }, { "epoch": 0.05, "learning_rate": 4.969388521634616e-05, "loss": 1.9292, "step": 165 }, { "epoch": 0.05, "learning_rate": 4.9692007211538466e-05, "loss": 2.1599, "step": 166 }, { "epoch": 0.05, "learning_rate": 4.969012920673077e-05, "loss": 2.3706, "step": 167 }, { "epoch": 0.05, "learning_rate": 4.968825120192308e-05, "loss": 2.2414, "step": 168 }, { "epoch": 0.05, "learning_rate": 4.968637319711539e-05, "loss": 2.2092, "step": 169 }, { "epoch": 0.05, "learning_rate": 4.968449519230769e-05, "loss": 2.0028, "step": 170 }, { "epoch": 0.05, "learning_rate": 4.9682617187500003e-05, "loss": 2.1545, "step": 171 }, { "epoch": 0.05, "learning_rate": 4.968073918269231e-05, "loss": 2.2565, "step": 172 }, { "epoch": 0.05, "learning_rate": 4.967886117788462e-05, "loss": 2.0682, "step": 173 }, { "epoch": 0.05, "learning_rate": 4.9676983173076925e-05, "loss": 2.2662, "step": 174 }, { "epoch": 0.05, "learning_rate": 4.967510516826923e-05, "loss": 2.1574, "step": 175 }, { "epoch": 0.05, "learning_rate": 4.967322716346154e-05, "loss": 2.1784, "step": 176 }, { "epoch": 0.05, "learning_rate": 4.9671349158653846e-05, "loss": 2.2519, "step": 177 }, { "epoch": 0.05, "learning_rate": 4.966947115384616e-05, "loss": 2.1765, "step": 178 }, { "epoch": 0.05, "learning_rate": 4.966759314903847e-05, "loss": 2.318, "step": 179 }, { "epoch": 0.05, "learning_rate": 4.9665715144230774e-05, "loss": 2.2731, "step": 180 }, { "epoch": 0.05, "learning_rate": 4.966383713942308e-05, "loss": 2.1449, "step": 181 }, { "epoch": 0.05, "learning_rate": 4.966195913461539e-05, "loss": 2.2731, "step": 182 }, { "epoch": 0.05, "learning_rate": 4.9660081129807695e-05, "loss": 2.1532, "step": 183 }, { "epoch": 0.06, "learning_rate": 4.9658203125e-05, "loss": 2.3756, "step": 184 }, { "epoch": 0.06, "learning_rate": 4.965632512019231e-05, "loss": 2.0879, "step": 185 }, { "epoch": 0.06, "learning_rate": 4.9654447115384616e-05, "loss": 2.6703, "step": 186 }, { "epoch": 0.06, "learning_rate": 4.965256911057693e-05, "loss": 2.2504, "step": 187 }, { "epoch": 0.06, "learning_rate": 4.965069110576923e-05, "loss": 2.1941, "step": 188 }, { "epoch": 0.06, "learning_rate": 4.964881310096154e-05, "loss": 2.2432, "step": 189 }, { "epoch": 0.06, "learning_rate": 4.964693509615385e-05, "loss": 2.1805, "step": 190 }, { "epoch": 0.06, "learning_rate": 4.964505709134616e-05, "loss": 2.1082, "step": 191 }, { "epoch": 0.06, "learning_rate": 4.9643179086538465e-05, "loss": 2.612, "step": 192 }, { "epoch": 0.06, "learning_rate": 4.9641301081730776e-05, "loss": 2.2491, "step": 193 }, { "epoch": 0.06, "learning_rate": 4.963942307692308e-05, "loss": 2.302, "step": 194 }, { "epoch": 0.06, "learning_rate": 4.9637545072115386e-05, "loss": 2.1823, "step": 195 }, { "epoch": 0.06, "learning_rate": 4.96356670673077e-05, "loss": 2.3539, "step": 196 }, { "epoch": 0.06, "learning_rate": 4.96337890625e-05, "loss": 2.3837, "step": 197 }, { "epoch": 0.06, "learning_rate": 4.963191105769231e-05, "loss": 1.9222, "step": 198 }, { "epoch": 0.06, "learning_rate": 4.963003305288462e-05, "loss": 2.1745, "step": 199 }, { "epoch": 0.06, "learning_rate": 4.962815504807692e-05, "loss": 2.2745, "step": 200 }, { "epoch": 0.06, "learning_rate": 4.9626277043269235e-05, "loss": 2.2063, "step": 201 }, { "epoch": 0.06, "learning_rate": 4.962439903846154e-05, "loss": 2.412, "step": 202 }, { "epoch": 0.06, "learning_rate": 4.9622521033653844e-05, "loss": 2.2599, "step": 203 }, { "epoch": 0.06, "learning_rate": 4.9620643028846156e-05, "loss": 2.2187, "step": 204 }, { "epoch": 0.06, "learning_rate": 4.961876502403847e-05, "loss": 2.1911, "step": 205 }, { "epoch": 0.06, "learning_rate": 4.961688701923077e-05, "loss": 2.1087, "step": 206 }, { "epoch": 0.06, "learning_rate": 4.9615009014423084e-05, "loss": 2.36, "step": 207 }, { "epoch": 0.06, "learning_rate": 4.961313100961539e-05, "loss": 2.5764, "step": 208 }, { "epoch": 0.06, "learning_rate": 4.961125300480769e-05, "loss": 2.3015, "step": 209 }, { "epoch": 0.06, "learning_rate": 4.9609375000000005e-05, "loss": 2.0664, "step": 210 }, { "epoch": 0.06, "learning_rate": 4.960749699519231e-05, "loss": 2.1784, "step": 211 }, { "epoch": 0.06, "learning_rate": 4.9605618990384614e-05, "loss": 2.1845, "step": 212 }, { "epoch": 0.06, "learning_rate": 4.9603740985576926e-05, "loss": 2.1165, "step": 213 }, { "epoch": 0.06, "learning_rate": 4.960186298076923e-05, "loss": 2.1628, "step": 214 }, { "epoch": 0.06, "learning_rate": 4.9599984975961535e-05, "loss": 2.2031, "step": 215 }, { "epoch": 0.06, "learning_rate": 4.959810697115385e-05, "loss": 2.373, "step": 216 }, { "epoch": 0.07, "learning_rate": 4.959622896634616e-05, "loss": 2.3169, "step": 217 }, { "epoch": 0.07, "learning_rate": 4.959435096153846e-05, "loss": 2.1001, "step": 218 }, { "epoch": 0.07, "learning_rate": 4.9592472956730775e-05, "loss": 2.0814, "step": 219 }, { "epoch": 0.07, "learning_rate": 4.959059495192308e-05, "loss": 2.086, "step": 220 }, { "epoch": 0.07, "learning_rate": 4.958871694711539e-05, "loss": 2.216, "step": 221 }, { "epoch": 0.07, "learning_rate": 4.9586838942307696e-05, "loss": 2.2587, "step": 222 }, { "epoch": 0.07, "learning_rate": 4.95849609375e-05, "loss": 2.0993, "step": 223 }, { "epoch": 0.07, "learning_rate": 4.958308293269231e-05, "loss": 2.1712, "step": 224 }, { "epoch": 0.07, "learning_rate": 4.958120492788462e-05, "loss": 2.2605, "step": 225 }, { "epoch": 0.07, "learning_rate": 4.957932692307692e-05, "loss": 2.2602, "step": 226 }, { "epoch": 0.07, "learning_rate": 4.957744891826923e-05, "loss": 1.9052, "step": 227 }, { "epoch": 0.07, "learning_rate": 4.957557091346154e-05, "loss": 2.0861, "step": 228 }, { "epoch": 0.07, "learning_rate": 4.957369290865384e-05, "loss": 2.2485, "step": 229 }, { "epoch": 0.07, "learning_rate": 4.957181490384616e-05, "loss": 2.2975, "step": 230 }, { "epoch": 0.07, "learning_rate": 4.9569936899038466e-05, "loss": 2.3803, "step": 231 }, { "epoch": 0.07, "learning_rate": 4.956805889423077e-05, "loss": 2.2527, "step": 232 }, { "epoch": 0.07, "learning_rate": 4.956618088942308e-05, "loss": 2.1752, "step": 233 }, { "epoch": 0.07, "learning_rate": 4.956430288461539e-05, "loss": 2.1719, "step": 234 }, { "epoch": 0.07, "learning_rate": 4.956242487980769e-05, "loss": 2.085, "step": 235 }, { "epoch": 0.07, "learning_rate": 4.9560546875e-05, "loss": 2.5529, "step": 236 }, { "epoch": 0.07, "learning_rate": 4.955866887019231e-05, "loss": 2.0931, "step": 237 }, { "epoch": 0.07, "learning_rate": 4.955679086538462e-05, "loss": 2.0686, "step": 238 }, { "epoch": 0.07, "learning_rate": 4.9554912860576924e-05, "loss": 2.1789, "step": 239 }, { "epoch": 0.07, "learning_rate": 4.955303485576923e-05, "loss": 2.3809, "step": 240 }, { "epoch": 0.07, "learning_rate": 4.955115685096154e-05, "loss": 2.355, "step": 241 }, { "epoch": 0.07, "learning_rate": 4.9549278846153846e-05, "loss": 2.4239, "step": 242 }, { "epoch": 0.07, "learning_rate": 4.954740084134616e-05, "loss": 2.2032, "step": 243 }, { "epoch": 0.07, "learning_rate": 4.954552283653847e-05, "loss": 2.2828, "step": 244 }, { "epoch": 0.07, "learning_rate": 4.9543644831730773e-05, "loss": 2.0058, "step": 245 }, { "epoch": 0.07, "learning_rate": 4.954176682692308e-05, "loss": 2.0671, "step": 246 }, { "epoch": 0.07, "learning_rate": 4.953988882211539e-05, "loss": 2.3115, "step": 247 }, { "epoch": 0.07, "learning_rate": 4.9538010817307695e-05, "loss": 2.2888, "step": 248 }, { "epoch": 0.07, "learning_rate": 4.95361328125e-05, "loss": 2.2137, "step": 249 }, { "epoch": 0.08, "learning_rate": 4.953425480769231e-05, "loss": 2.1828, "step": 250 }, { "epoch": 0.08, "learning_rate": 4.9532376802884616e-05, "loss": 2.1222, "step": 251 }, { "epoch": 0.08, "learning_rate": 4.953049879807693e-05, "loss": 2.5337, "step": 252 }, { "epoch": 0.08, "learning_rate": 4.952862079326923e-05, "loss": 2.3559, "step": 253 }, { "epoch": 0.08, "learning_rate": 4.952674278846154e-05, "loss": 2.1317, "step": 254 }, { "epoch": 0.08, "learning_rate": 4.952486478365385e-05, "loss": 2.3484, "step": 255 }, { "epoch": 0.08, "learning_rate": 4.952298677884616e-05, "loss": 2.1749, "step": 256 }, { "epoch": 0.08, "learning_rate": 4.9521108774038465e-05, "loss": 2.031, "step": 257 }, { "epoch": 0.08, "learning_rate": 4.9519230769230776e-05, "loss": 2.0888, "step": 258 }, { "epoch": 0.08, "learning_rate": 4.951735276442308e-05, "loss": 1.9985, "step": 259 }, { "epoch": 0.08, "learning_rate": 4.9515474759615386e-05, "loss": 2.3177, "step": 260 }, { "epoch": 0.08, "learning_rate": 4.95135967548077e-05, "loss": 2.1355, "step": 261 }, { "epoch": 0.08, "learning_rate": 4.951171875e-05, "loss": 2.5003, "step": 262 }, { "epoch": 0.08, "learning_rate": 4.950984074519231e-05, "loss": 2.3224, "step": 263 }, { "epoch": 0.08, "learning_rate": 4.950796274038462e-05, "loss": 2.4475, "step": 264 }, { "epoch": 0.08, "learning_rate": 4.950608473557692e-05, "loss": 2.4284, "step": 265 }, { "epoch": 0.08, "learning_rate": 4.9504206730769235e-05, "loss": 2.2344, "step": 266 }, { "epoch": 0.08, "learning_rate": 4.950232872596154e-05, "loss": 2.1701, "step": 267 }, { "epoch": 0.08, "learning_rate": 4.9500450721153844e-05, "loss": 2.5279, "step": 268 }, { "epoch": 0.08, "learning_rate": 4.9498572716346156e-05, "loss": 2.7088, "step": 269 }, { "epoch": 0.08, "learning_rate": 4.949669471153847e-05, "loss": 2.2276, "step": 270 }, { "epoch": 0.08, "learning_rate": 4.949481670673077e-05, "loss": 2.1371, "step": 271 }, { "epoch": 0.08, "learning_rate": 4.9492938701923084e-05, "loss": 2.311, "step": 272 }, { "epoch": 0.08, "learning_rate": 4.949106069711539e-05, "loss": 2.1465, "step": 273 }, { "epoch": 0.08, "learning_rate": 4.948918269230769e-05, "loss": 2.3675, "step": 274 }, { "epoch": 0.08, "learning_rate": 4.9487304687500005e-05, "loss": 2.0586, "step": 275 }, { "epoch": 0.08, "learning_rate": 4.948542668269231e-05, "loss": 2.5294, "step": 276 }, { "epoch": 0.08, "learning_rate": 4.9483548677884614e-05, "loss": 2.018, "step": 277 }, { "epoch": 0.08, "learning_rate": 4.9481670673076926e-05, "loss": 2.2484, "step": 278 }, { "epoch": 0.08, "learning_rate": 4.947979266826923e-05, "loss": 2.4638, "step": 279 }, { "epoch": 0.08, "learning_rate": 4.947791466346154e-05, "loss": 2.0683, "step": 280 }, { "epoch": 0.08, "learning_rate": 4.947603665865385e-05, "loss": 1.9664, "step": 281 }, { "epoch": 0.08, "learning_rate": 4.947415865384616e-05, "loss": 2.2017, "step": 282 }, { "epoch": 0.09, "learning_rate": 4.947228064903846e-05, "loss": 1.8967, "step": 283 }, { "epoch": 0.09, "learning_rate": 4.9470402644230775e-05, "loss": 2.1114, "step": 284 }, { "epoch": 0.09, "learning_rate": 4.946852463942308e-05, "loss": 2.0602, "step": 285 }, { "epoch": 0.09, "learning_rate": 4.946664663461539e-05, "loss": 2.0996, "step": 286 }, { "epoch": 0.09, "learning_rate": 4.9464768629807696e-05, "loss": 2.3581, "step": 287 }, { "epoch": 0.09, "learning_rate": 4.9462890625e-05, "loss": 2.1525, "step": 288 }, { "epoch": 0.09, "learning_rate": 4.946101262019231e-05, "loss": 2.1422, "step": 289 }, { "epoch": 0.09, "learning_rate": 4.945913461538462e-05, "loss": 2.4018, "step": 290 }, { "epoch": 0.09, "learning_rate": 4.945725661057692e-05, "loss": 2.0936, "step": 291 }, { "epoch": 0.09, "learning_rate": 4.945537860576923e-05, "loss": 2.4236, "step": 292 }, { "epoch": 0.09, "learning_rate": 4.945350060096154e-05, "loss": 2.0162, "step": 293 }, { "epoch": 0.09, "learning_rate": 4.945162259615384e-05, "loss": 2.5273, "step": 294 }, { "epoch": 0.09, "learning_rate": 4.944974459134616e-05, "loss": 2.3119, "step": 295 }, { "epoch": 0.09, "learning_rate": 4.9447866586538466e-05, "loss": 2.3902, "step": 296 }, { "epoch": 0.09, "learning_rate": 4.944598858173077e-05, "loss": 2.0428, "step": 297 }, { "epoch": 0.09, "learning_rate": 4.944411057692308e-05, "loss": 2.5204, "step": 298 }, { "epoch": 0.09, "learning_rate": 4.944223257211539e-05, "loss": 2.3696, "step": 299 }, { "epoch": 0.09, "learning_rate": 4.94403545673077e-05, "loss": 1.9444, "step": 300 }, { "epoch": 0.09, "learning_rate": 4.94384765625e-05, "loss": 2.0067, "step": 301 }, { "epoch": 0.09, "learning_rate": 4.943659855769231e-05, "loss": 2.2436, "step": 302 }, { "epoch": 0.09, "learning_rate": 4.943472055288462e-05, "loss": 2.3744, "step": 303 }, { "epoch": 0.09, "learning_rate": 4.9432842548076924e-05, "loss": 2.3169, "step": 304 }, { "epoch": 0.09, "learning_rate": 4.943096454326923e-05, "loss": 2.1842, "step": 305 }, { "epoch": 0.09, "learning_rate": 4.942908653846154e-05, "loss": 2.2461, "step": 306 }, { "epoch": 0.09, "learning_rate": 4.9427208533653845e-05, "loss": 2.0583, "step": 307 }, { "epoch": 0.09, "learning_rate": 4.942533052884616e-05, "loss": 2.4163, "step": 308 }, { "epoch": 0.09, "learning_rate": 4.942345252403847e-05, "loss": 2.1694, "step": 309 }, { "epoch": 0.09, "learning_rate": 4.942157451923077e-05, "loss": 1.9452, "step": 310 }, { "epoch": 0.09, "learning_rate": 4.941969651442308e-05, "loss": 2.318, "step": 311 }, { "epoch": 0.09, "learning_rate": 4.941781850961539e-05, "loss": 2.4157, "step": 312 }, { "epoch": 0.09, "learning_rate": 4.9415940504807694e-05, "loss": 2.297, "step": 313 }, { "epoch": 0.09, "learning_rate": 4.94140625e-05, "loss": 2.375, "step": 314 }, { "epoch": 0.09, "learning_rate": 4.941218449519231e-05, "loss": 2.0808, "step": 315 }, { "epoch": 0.09, "learning_rate": 4.9410306490384615e-05, "loss": 2.1919, "step": 316 }, { "epoch": 0.1, "learning_rate": 4.940842848557693e-05, "loss": 2.3102, "step": 317 }, { "epoch": 0.1, "learning_rate": 4.940655048076923e-05, "loss": 2.2152, "step": 318 }, { "epoch": 0.1, "learning_rate": 4.9404672475961537e-05, "loss": 2.2708, "step": 319 }, { "epoch": 0.1, "learning_rate": 4.940279447115385e-05, "loss": 2.4566, "step": 320 }, { "epoch": 0.1, "learning_rate": 4.940091646634616e-05, "loss": 1.9145, "step": 321 }, { "epoch": 0.1, "learning_rate": 4.9399038461538464e-05, "loss": 2.3846, "step": 322 }, { "epoch": 0.1, "learning_rate": 4.9397160456730776e-05, "loss": 2.4575, "step": 323 }, { "epoch": 0.1, "learning_rate": 4.939528245192308e-05, "loss": 2.1409, "step": 324 }, { "epoch": 0.1, "learning_rate": 4.9393404447115386e-05, "loss": 2.22, "step": 325 }, { "epoch": 0.1, "learning_rate": 4.93915264423077e-05, "loss": 2.1246, "step": 326 }, { "epoch": 0.1, "learning_rate": 4.93896484375e-05, "loss": 2.4339, "step": 327 }, { "epoch": 0.1, "learning_rate": 4.938777043269231e-05, "loss": 2.1414, "step": 328 }, { "epoch": 0.1, "learning_rate": 4.938589242788462e-05, "loss": 2.4009, "step": 329 }, { "epoch": 0.1, "learning_rate": 4.938401442307692e-05, "loss": 2.185, "step": 330 }, { "epoch": 0.1, "learning_rate": 4.9382136418269234e-05, "loss": 2.2154, "step": 331 }, { "epoch": 0.1, "learning_rate": 4.938025841346154e-05, "loss": 2.3187, "step": 332 }, { "epoch": 0.1, "learning_rate": 4.9378380408653844e-05, "loss": 2.2601, "step": 333 }, { "epoch": 0.1, "learning_rate": 4.9376502403846156e-05, "loss": 2.2765, "step": 334 }, { "epoch": 0.1, "learning_rate": 4.937462439903847e-05, "loss": 2.0674, "step": 335 }, { "epoch": 0.1, "learning_rate": 4.937274639423077e-05, "loss": 2.155, "step": 336 }, { "epoch": 0.1, "learning_rate": 4.9370868389423083e-05, "loss": 2.218, "step": 337 }, { "epoch": 0.1, "learning_rate": 4.936899038461539e-05, "loss": 2.4188, "step": 338 }, { "epoch": 0.1, "learning_rate": 4.936711237980769e-05, "loss": 2.1882, "step": 339 }, { "epoch": 0.1, "learning_rate": 4.9365234375000005e-05, "loss": 2.2995, "step": 340 }, { "epoch": 0.1, "learning_rate": 4.936335637019231e-05, "loss": 2.3034, "step": 341 }, { "epoch": 0.1, "learning_rate": 4.9361478365384614e-05, "loss": 2.0427, "step": 342 }, { "epoch": 0.1, "learning_rate": 4.9359600360576926e-05, "loss": 2.3629, "step": 343 }, { "epoch": 0.1, "learning_rate": 4.935772235576923e-05, "loss": 2.1069, "step": 344 }, { "epoch": 0.1, "learning_rate": 4.935584435096154e-05, "loss": 1.927, "step": 345 }, { "epoch": 0.1, "learning_rate": 4.935396634615385e-05, "loss": 2.2611, "step": 346 }, { "epoch": 0.1, "learning_rate": 4.935208834134616e-05, "loss": 2.1965, "step": 347 }, { "epoch": 0.1, "learning_rate": 4.935021033653846e-05, "loss": 2.219, "step": 348 }, { "epoch": 0.1, "learning_rate": 4.9348332331730775e-05, "loss": 2.3209, "step": 349 }, { "epoch": 0.11, "learning_rate": 4.934645432692308e-05, "loss": 2.0778, "step": 350 }, { "epoch": 0.11, "learning_rate": 4.934457632211539e-05, "loss": 2.2416, "step": 351 }, { "epoch": 0.11, "learning_rate": 4.9342698317307696e-05, "loss": 2.2622, "step": 352 }, { "epoch": 0.11, "learning_rate": 4.93408203125e-05, "loss": 2.2522, "step": 353 }, { "epoch": 0.11, "learning_rate": 4.933894230769231e-05, "loss": 2.391, "step": 354 }, { "epoch": 0.11, "learning_rate": 4.933706430288462e-05, "loss": 2.2084, "step": 355 }, { "epoch": 0.11, "learning_rate": 4.933518629807692e-05, "loss": 2.312, "step": 356 }, { "epoch": 0.11, "learning_rate": 4.933330829326923e-05, "loss": 2.221, "step": 357 }, { "epoch": 0.11, "learning_rate": 4.933143028846154e-05, "loss": 2.2796, "step": 358 }, { "epoch": 0.11, "learning_rate": 4.932955228365384e-05, "loss": 2.3369, "step": 359 }, { "epoch": 0.11, "learning_rate": 4.932767427884616e-05, "loss": 2.0286, "step": 360 }, { "epoch": 0.11, "learning_rate": 4.9325796274038466e-05, "loss": 2.3084, "step": 361 }, { "epoch": 0.11, "learning_rate": 4.932391826923077e-05, "loss": 2.1091, "step": 362 }, { "epoch": 0.11, "learning_rate": 4.932204026442308e-05, "loss": 2.1962, "step": 363 }, { "epoch": 0.11, "learning_rate": 4.932016225961539e-05, "loss": 2.2727, "step": 364 }, { "epoch": 0.11, "learning_rate": 4.93182842548077e-05, "loss": 2.0867, "step": 365 }, { "epoch": 0.11, "learning_rate": 4.931640625e-05, "loss": 2.3972, "step": 366 }, { "epoch": 0.11, "learning_rate": 4.931452824519231e-05, "loss": 2.1119, "step": 367 }, { "epoch": 0.11, "learning_rate": 4.931265024038462e-05, "loss": 2.2127, "step": 368 }, { "epoch": 0.11, "learning_rate": 4.9310772235576924e-05, "loss": 2.1493, "step": 369 }, { "epoch": 0.11, "learning_rate": 4.930889423076923e-05, "loss": 2.349, "step": 370 }, { "epoch": 0.11, "learning_rate": 4.930701622596154e-05, "loss": 2.2979, "step": 371 }, { "epoch": 0.11, "learning_rate": 4.9305138221153845e-05, "loss": 2.2401, "step": 372 }, { "epoch": 0.11, "learning_rate": 4.930326021634616e-05, "loss": 2.1352, "step": 373 }, { "epoch": 0.11, "learning_rate": 4.930138221153847e-05, "loss": 2.118, "step": 374 }, { "epoch": 0.11, "learning_rate": 4.929950420673077e-05, "loss": 2.2502, "step": 375 }, { "epoch": 0.11, "learning_rate": 4.929762620192308e-05, "loss": 2.3149, "step": 376 }, { "epoch": 0.11, "learning_rate": 4.929574819711539e-05, "loss": 2.1822, "step": 377 }, { "epoch": 0.11, "learning_rate": 4.9293870192307694e-05, "loss": 2.1974, "step": 378 }, { "epoch": 0.11, "learning_rate": 4.92919921875e-05, "loss": 2.159, "step": 379 }, { "epoch": 0.11, "learning_rate": 4.929011418269231e-05, "loss": 2.3266, "step": 380 }, { "epoch": 0.11, "learning_rate": 4.9288236177884615e-05, "loss": 2.3263, "step": 381 }, { "epoch": 0.11, "learning_rate": 4.928635817307693e-05, "loss": 1.8783, "step": 382 }, { "epoch": 0.12, "learning_rate": 4.928448016826923e-05, "loss": 2.0821, "step": 383 }, { "epoch": 0.12, "learning_rate": 4.9282602163461536e-05, "loss": 2.108, "step": 384 }, { "epoch": 0.12, "learning_rate": 4.928072415865385e-05, "loss": 2.0389, "step": 385 }, { "epoch": 0.12, "learning_rate": 4.927884615384616e-05, "loss": 2.2843, "step": 386 }, { "epoch": 0.12, "learning_rate": 4.9276968149038464e-05, "loss": 2.1173, "step": 387 }, { "epoch": 0.12, "learning_rate": 4.9275090144230776e-05, "loss": 2.2042, "step": 388 }, { "epoch": 0.12, "learning_rate": 4.927321213942308e-05, "loss": 2.2558, "step": 389 }, { "epoch": 0.12, "learning_rate": 4.9271334134615385e-05, "loss": 2.1335, "step": 390 }, { "epoch": 0.12, "learning_rate": 4.92694561298077e-05, "loss": 2.383, "step": 391 }, { "epoch": 0.12, "learning_rate": 4.9267578125e-05, "loss": 2.1654, "step": 392 }, { "epoch": 0.12, "learning_rate": 4.9265700120192307e-05, "loss": 2.3441, "step": 393 }, { "epoch": 0.12, "learning_rate": 4.926382211538462e-05, "loss": 2.0608, "step": 394 }, { "epoch": 0.12, "learning_rate": 4.926194411057692e-05, "loss": 2.3655, "step": 395 }, { "epoch": 0.12, "learning_rate": 4.9260066105769234e-05, "loss": 2.0975, "step": 396 }, { "epoch": 0.12, "learning_rate": 4.925818810096154e-05, "loss": 2.44, "step": 397 }, { "epoch": 0.12, "learning_rate": 4.9256310096153844e-05, "loss": 1.8803, "step": 398 }, { "epoch": 0.12, "learning_rate": 4.9254432091346155e-05, "loss": 2.4008, "step": 399 }, { "epoch": 0.12, "learning_rate": 4.925255408653847e-05, "loss": 2.1933, "step": 400 }, { "epoch": 0.12, "learning_rate": 4.925067608173077e-05, "loss": 2.4058, "step": 401 }, { "epoch": 0.12, "learning_rate": 4.924879807692308e-05, "loss": 2.0054, "step": 402 }, { "epoch": 0.12, "learning_rate": 4.924692007211539e-05, "loss": 2.0747, "step": 403 }, { "epoch": 0.12, "learning_rate": 4.924504206730769e-05, "loss": 2.7113, "step": 404 }, { "epoch": 0.12, "learning_rate": 4.9243164062500004e-05, "loss": 2.1538, "step": 405 }, { "epoch": 0.12, "learning_rate": 4.924128605769231e-05, "loss": 2.1787, "step": 406 }, { "epoch": 0.12, "learning_rate": 4.9239408052884614e-05, "loss": 2.0409, "step": 407 }, { "epoch": 0.12, "learning_rate": 4.9237530048076926e-05, "loss": 2.3107, "step": 408 }, { "epoch": 0.12, "learning_rate": 4.923565204326923e-05, "loss": 2.3297, "step": 409 }, { "epoch": 0.12, "learning_rate": 4.923377403846154e-05, "loss": 2.2586, "step": 410 }, { "epoch": 0.12, "learning_rate": 4.923189603365385e-05, "loss": 2.2167, "step": 411 }, { "epoch": 0.12, "learning_rate": 4.923001802884616e-05, "loss": 2.1905, "step": 412 }, { "epoch": 0.12, "learning_rate": 4.922814002403846e-05, "loss": 2.412, "step": 413 }, { "epoch": 0.12, "learning_rate": 4.9226262019230774e-05, "loss": 2.1251, "step": 414 }, { "epoch": 0.12, "learning_rate": 4.922438401442308e-05, "loss": 2.2257, "step": 415 }, { "epoch": 0.12, "learning_rate": 4.922250600961539e-05, "loss": 2.2697, "step": 416 }, { "epoch": 0.13, "learning_rate": 4.9220628004807696e-05, "loss": 2.1048, "step": 417 }, { "epoch": 0.13, "learning_rate": 4.921875e-05, "loss": 2.1278, "step": 418 }, { "epoch": 0.13, "learning_rate": 4.921687199519231e-05, "loss": 2.27, "step": 419 }, { "epoch": 0.13, "learning_rate": 4.921499399038462e-05, "loss": 2.4765, "step": 420 }, { "epoch": 0.13, "learning_rate": 4.921311598557692e-05, "loss": 2.151, "step": 421 }, { "epoch": 0.13, "learning_rate": 4.921123798076923e-05, "loss": 2.2376, "step": 422 }, { "epoch": 0.13, "learning_rate": 4.920935997596154e-05, "loss": 2.0766, "step": 423 }, { "epoch": 0.13, "learning_rate": 4.920748197115385e-05, "loss": 2.2767, "step": 424 }, { "epoch": 0.13, "learning_rate": 4.920560396634616e-05, "loss": 2.013, "step": 425 }, { "epoch": 0.13, "learning_rate": 4.9203725961538466e-05, "loss": 2.1744, "step": 426 }, { "epoch": 0.13, "learning_rate": 4.920184795673077e-05, "loss": 2.1812, "step": 427 }, { "epoch": 0.13, "learning_rate": 4.919996995192308e-05, "loss": 2.3184, "step": 428 }, { "epoch": 0.13, "learning_rate": 4.919809194711539e-05, "loss": 2.0054, "step": 429 }, { "epoch": 0.13, "learning_rate": 4.91962139423077e-05, "loss": 2.5157, "step": 430 }, { "epoch": 0.13, "learning_rate": 4.91943359375e-05, "loss": 2.393, "step": 431 }, { "epoch": 0.13, "learning_rate": 4.919245793269231e-05, "loss": 2.2864, "step": 432 }, { "epoch": 0.13, "learning_rate": 4.919057992788462e-05, "loss": 2.3464, "step": 433 }, { "epoch": 0.13, "learning_rate": 4.9188701923076924e-05, "loss": 2.1009, "step": 434 }, { "epoch": 0.13, "learning_rate": 4.918682391826923e-05, "loss": 2.0867, "step": 435 }, { "epoch": 0.13, "learning_rate": 4.918494591346154e-05, "loss": 2.2589, "step": 436 }, { "epoch": 0.13, "learning_rate": 4.9183067908653845e-05, "loss": 2.1612, "step": 437 }, { "epoch": 0.13, "learning_rate": 4.918118990384616e-05, "loss": 2.3611, "step": 438 }, { "epoch": 0.13, "learning_rate": 4.917931189903847e-05, "loss": 2.1831, "step": 439 }, { "epoch": 0.13, "learning_rate": 4.917743389423077e-05, "loss": 2.5146, "step": 440 }, { "epoch": 0.13, "learning_rate": 4.917555588942308e-05, "loss": 2.0228, "step": 441 }, { "epoch": 0.13, "learning_rate": 4.917367788461539e-05, "loss": 2.016, "step": 442 }, { "epoch": 0.13, "learning_rate": 4.9171799879807694e-05, "loss": 2.022, "step": 443 }, { "epoch": 0.13, "learning_rate": 4.9169921875000006e-05, "loss": 2.4035, "step": 444 }, { "epoch": 0.13, "learning_rate": 4.916804387019231e-05, "loss": 2.4921, "step": 445 }, { "epoch": 0.13, "learning_rate": 4.9166165865384615e-05, "loss": 2.2173, "step": 446 }, { "epoch": 0.13, "learning_rate": 4.916428786057693e-05, "loss": 2.136, "step": 447 }, { "epoch": 0.13, "learning_rate": 4.916240985576923e-05, "loss": 2.1714, "step": 448 }, { "epoch": 0.13, "learning_rate": 4.9160531850961536e-05, "loss": 2.3007, "step": 449 }, { "epoch": 0.14, "learning_rate": 4.915865384615385e-05, "loss": 2.3295, "step": 450 }, { "epoch": 0.14, "learning_rate": 4.915677584134616e-05, "loss": 2.216, "step": 451 }, { "epoch": 0.14, "learning_rate": 4.9154897836538464e-05, "loss": 2.1151, "step": 452 }, { "epoch": 0.14, "learning_rate": 4.9153019831730776e-05, "loss": 2.2716, "step": 453 }, { "epoch": 0.14, "learning_rate": 4.915114182692308e-05, "loss": 2.4418, "step": 454 }, { "epoch": 0.14, "learning_rate": 4.9149263822115385e-05, "loss": 2.0288, "step": 455 }, { "epoch": 0.14, "learning_rate": 4.91473858173077e-05, "loss": 2.2664, "step": 456 }, { "epoch": 0.14, "learning_rate": 4.91455078125e-05, "loss": 2.1106, "step": 457 }, { "epoch": 0.14, "learning_rate": 4.9143629807692306e-05, "loss": 2.239, "step": 458 }, { "epoch": 0.14, "learning_rate": 4.914175180288462e-05, "loss": 2.2629, "step": 459 }, { "epoch": 0.14, "learning_rate": 4.913987379807692e-05, "loss": 2.3927, "step": 460 }, { "epoch": 0.14, "learning_rate": 4.9137995793269234e-05, "loss": 1.8208, "step": 461 }, { "epoch": 0.14, "learning_rate": 4.913611778846154e-05, "loss": 2.3371, "step": 462 }, { "epoch": 0.14, "learning_rate": 4.9134239783653844e-05, "loss": 2.3034, "step": 463 }, { "epoch": 0.14, "learning_rate": 4.913236177884616e-05, "loss": 2.3437, "step": 464 }, { "epoch": 0.14, "learning_rate": 4.913048377403847e-05, "loss": 2.3961, "step": 465 }, { "epoch": 0.14, "learning_rate": 4.912860576923077e-05, "loss": 2.2188, "step": 466 }, { "epoch": 0.14, "learning_rate": 4.912672776442308e-05, "loss": 2.2684, "step": 467 }, { "epoch": 0.14, "learning_rate": 4.912484975961539e-05, "loss": 1.9881, "step": 468 }, { "epoch": 0.14, "learning_rate": 4.912297175480769e-05, "loss": 2.0653, "step": 469 }, { "epoch": 0.14, "learning_rate": 4.9121093750000004e-05, "loss": 2.2586, "step": 470 }, { "epoch": 0.14, "learning_rate": 4.911921574519231e-05, "loss": 2.1904, "step": 471 }, { "epoch": 0.14, "learning_rate": 4.9117337740384614e-05, "loss": 2.2007, "step": 472 }, { "epoch": 0.14, "learning_rate": 4.9115459735576925e-05, "loss": 2.1604, "step": 473 }, { "epoch": 0.14, "learning_rate": 4.911358173076923e-05, "loss": 2.3588, "step": 474 }, { "epoch": 0.14, "learning_rate": 4.911170372596154e-05, "loss": 2.168, "step": 475 }, { "epoch": 0.14, "learning_rate": 4.9109825721153847e-05, "loss": 2.3788, "step": 476 }, { "epoch": 0.14, "learning_rate": 4.910794771634616e-05, "loss": 2.2712, "step": 477 }, { "epoch": 0.14, "learning_rate": 4.910606971153846e-05, "loss": 2.2237, "step": 478 }, { "epoch": 0.14, "learning_rate": 4.9104191706730774e-05, "loss": 2.2118, "step": 479 }, { "epoch": 0.14, "learning_rate": 4.910231370192308e-05, "loss": 2.2815, "step": 480 }, { "epoch": 0.14, "learning_rate": 4.910043569711539e-05, "loss": 2.2865, "step": 481 }, { "epoch": 0.14, "learning_rate": 4.9098557692307695e-05, "loss": 2.2325, "step": 482 }, { "epoch": 0.15, "learning_rate": 4.90966796875e-05, "loss": 2.0364, "step": 483 }, { "epoch": 0.15, "learning_rate": 4.909480168269231e-05, "loss": 2.0067, "step": 484 }, { "epoch": 0.15, "learning_rate": 4.9092923677884617e-05, "loss": 2.0019, "step": 485 }, { "epoch": 0.15, "learning_rate": 4.909104567307692e-05, "loss": 2.2746, "step": 486 }, { "epoch": 0.15, "learning_rate": 4.908916766826923e-05, "loss": 1.9774, "step": 487 }, { "epoch": 0.15, "learning_rate": 4.908728966346154e-05, "loss": 2.1881, "step": 488 }, { "epoch": 0.15, "learning_rate": 4.908541165865385e-05, "loss": 2.0667, "step": 489 }, { "epoch": 0.15, "learning_rate": 4.908353365384616e-05, "loss": 2.148, "step": 490 }, { "epoch": 0.15, "learning_rate": 4.9081655649038466e-05, "loss": 2.4582, "step": 491 }, { "epoch": 0.15, "learning_rate": 4.907977764423077e-05, "loss": 2.1503, "step": 492 }, { "epoch": 0.15, "learning_rate": 4.907789963942308e-05, "loss": 2.3103, "step": 493 }, { "epoch": 0.15, "learning_rate": 4.9076021634615387e-05, "loss": 2.3725, "step": 494 }, { "epoch": 0.15, "learning_rate": 4.90741436298077e-05, "loss": 2.0529, "step": 495 }, { "epoch": 0.15, "learning_rate": 4.9072265625e-05, "loss": 2.1273, "step": 496 }, { "epoch": 0.15, "learning_rate": 4.907038762019231e-05, "loss": 2.3283, "step": 497 }, { "epoch": 0.15, "learning_rate": 4.906850961538462e-05, "loss": 2.2111, "step": 498 }, { "epoch": 0.15, "learning_rate": 4.9066631610576924e-05, "loss": 2.3618, "step": 499 }, { "epoch": 0.15, "learning_rate": 4.906475360576923e-05, "loss": 2.3367, "step": 500 }, { "epoch": 0.15, "learning_rate": 4.906287560096154e-05, "loss": 2.2149, "step": 501 }, { "epoch": 0.15, "learning_rate": 4.9060997596153845e-05, "loss": 2.1458, "step": 502 }, { "epoch": 0.15, "learning_rate": 4.905911959134616e-05, "loss": 2.3243, "step": 503 }, { "epoch": 0.15, "learning_rate": 4.905724158653847e-05, "loss": 2.4537, "step": 504 }, { "epoch": 0.15, "learning_rate": 4.905536358173077e-05, "loss": 2.3841, "step": 505 }, { "epoch": 0.15, "learning_rate": 4.905348557692308e-05, "loss": 2.1074, "step": 506 }, { "epoch": 0.15, "learning_rate": 4.905160757211539e-05, "loss": 2.395, "step": 507 }, { "epoch": 0.15, "learning_rate": 4.9049729567307694e-05, "loss": 2.1486, "step": 508 }, { "epoch": 0.15, "learning_rate": 4.9047851562500006e-05, "loss": 2.2762, "step": 509 }, { "epoch": 0.15, "learning_rate": 4.904597355769231e-05, "loss": 2.1501, "step": 510 }, { "epoch": 0.15, "learning_rate": 4.9044095552884615e-05, "loss": 2.1781, "step": 511 }, { "epoch": 0.15, "learning_rate": 4.904221754807693e-05, "loss": 2.0434, "step": 512 }, { "epoch": 0.15, "learning_rate": 4.904033954326923e-05, "loss": 1.9484, "step": 513 }, { "epoch": 0.15, "learning_rate": 4.9038461538461536e-05, "loss": 2.3498, "step": 514 }, { "epoch": 0.15, "learning_rate": 4.903658353365385e-05, "loss": 2.2419, "step": 515 }, { "epoch": 0.16, "learning_rate": 4.903470552884616e-05, "loss": 2.194, "step": 516 }, { "epoch": 0.16, "learning_rate": 4.9032827524038464e-05, "loss": 2.475, "step": 517 }, { "epoch": 0.16, "learning_rate": 4.9030949519230776e-05, "loss": 2.0988, "step": 518 }, { "epoch": 0.16, "learning_rate": 4.902907151442308e-05, "loss": 2.4475, "step": 519 }, { "epoch": 0.16, "learning_rate": 4.9027193509615385e-05, "loss": 2.2012, "step": 520 }, { "epoch": 0.16, "learning_rate": 4.90253155048077e-05, "loss": 1.9965, "step": 521 }, { "epoch": 0.16, "learning_rate": 4.90234375e-05, "loss": 2.2444, "step": 522 }, { "epoch": 0.16, "learning_rate": 4.9021559495192306e-05, "loss": 2.4263, "step": 523 }, { "epoch": 0.16, "learning_rate": 4.901968149038462e-05, "loss": 2.1857, "step": 524 }, { "epoch": 0.16, "learning_rate": 4.901780348557692e-05, "loss": 2.0637, "step": 525 }, { "epoch": 0.16, "learning_rate": 4.9015925480769234e-05, "loss": 2.348, "step": 526 }, { "epoch": 0.16, "learning_rate": 4.901404747596154e-05, "loss": 2.0392, "step": 527 }, { "epoch": 0.16, "learning_rate": 4.9012169471153844e-05, "loss": 2.1082, "step": 528 }, { "epoch": 0.16, "learning_rate": 4.901029146634616e-05, "loss": 2.0703, "step": 529 }, { "epoch": 0.16, "learning_rate": 4.900841346153847e-05, "loss": 2.2452, "step": 530 }, { "epoch": 0.16, "learning_rate": 4.900653545673077e-05, "loss": 2.0815, "step": 531 }, { "epoch": 0.16, "learning_rate": 4.900465745192308e-05, "loss": 2.329, "step": 532 }, { "epoch": 0.16, "learning_rate": 4.900277944711539e-05, "loss": 2.2021, "step": 533 }, { "epoch": 0.16, "learning_rate": 4.900090144230769e-05, "loss": 2.1499, "step": 534 }, { "epoch": 0.16, "learning_rate": 4.8999023437500004e-05, "loss": 2.1798, "step": 535 }, { "epoch": 0.16, "learning_rate": 4.899714543269231e-05, "loss": 2.1284, "step": 536 }, { "epoch": 0.16, "learning_rate": 4.8995267427884614e-05, "loss": 2.2286, "step": 537 }, { "epoch": 0.16, "learning_rate": 4.8993389423076925e-05, "loss": 2.2997, "step": 538 }, { "epoch": 0.16, "learning_rate": 4.899151141826923e-05, "loss": 2.2504, "step": 539 }, { "epoch": 0.16, "learning_rate": 4.898963341346154e-05, "loss": 2.1208, "step": 540 }, { "epoch": 0.16, "learning_rate": 4.8987755408653846e-05, "loss": 2.0554, "step": 541 }, { "epoch": 0.16, "learning_rate": 4.898587740384616e-05, "loss": 2.1357, "step": 542 }, { "epoch": 0.16, "learning_rate": 4.898399939903846e-05, "loss": 2.0897, "step": 543 }, { "epoch": 0.16, "learning_rate": 4.8982121394230774e-05, "loss": 2.181, "step": 544 }, { "epoch": 0.16, "learning_rate": 4.898024338942308e-05, "loss": 2.5436, "step": 545 }, { "epoch": 0.16, "learning_rate": 4.897836538461539e-05, "loss": 2.4465, "step": 546 }, { "epoch": 0.16, "learning_rate": 4.8976487379807695e-05, "loss": 2.0509, "step": 547 }, { "epoch": 0.16, "learning_rate": 4.8974609375e-05, "loss": 2.3722, "step": 548 }, { "epoch": 0.16, "learning_rate": 4.897273137019231e-05, "loss": 2.4685, "step": 549 }, { "epoch": 0.17, "learning_rate": 4.8970853365384616e-05, "loss": 2.2926, "step": 550 }, { "epoch": 0.17, "learning_rate": 4.896897536057692e-05, "loss": 2.3707, "step": 551 }, { "epoch": 0.17, "learning_rate": 4.896709735576923e-05, "loss": 2.4126, "step": 552 }, { "epoch": 0.17, "learning_rate": 4.896521935096154e-05, "loss": 2.3198, "step": 553 }, { "epoch": 0.17, "learning_rate": 4.896334134615385e-05, "loss": 2.0349, "step": 554 }, { "epoch": 0.17, "learning_rate": 4.896146334134616e-05, "loss": 2.2428, "step": 555 }, { "epoch": 0.17, "learning_rate": 4.8959585336538465e-05, "loss": 2.1891, "step": 556 }, { "epoch": 0.17, "learning_rate": 4.895770733173077e-05, "loss": 2.5124, "step": 557 }, { "epoch": 0.17, "learning_rate": 4.895582932692308e-05, "loss": 2.246, "step": 558 }, { "epoch": 0.17, "learning_rate": 4.8953951322115386e-05, "loss": 2.3552, "step": 559 }, { "epoch": 0.17, "learning_rate": 4.89520733173077e-05, "loss": 2.2604, "step": 560 }, { "epoch": 0.17, "learning_rate": 4.89501953125e-05, "loss": 2.5368, "step": 561 }, { "epoch": 0.17, "learning_rate": 4.894831730769231e-05, "loss": 2.0569, "step": 562 }, { "epoch": 0.17, "learning_rate": 4.894643930288462e-05, "loss": 2.1446, "step": 563 }, { "epoch": 0.17, "learning_rate": 4.8944561298076924e-05, "loss": 2.328, "step": 564 }, { "epoch": 0.17, "learning_rate": 4.894268329326923e-05, "loss": 2.3606, "step": 565 }, { "epoch": 0.17, "learning_rate": 4.894080528846154e-05, "loss": 2.3852, "step": 566 }, { "epoch": 0.17, "learning_rate": 4.8938927283653845e-05, "loss": 2.1948, "step": 567 }, { "epoch": 0.17, "learning_rate": 4.8937049278846157e-05, "loss": 2.0912, "step": 568 }, { "epoch": 0.17, "learning_rate": 4.893517127403847e-05, "loss": 2.1803, "step": 569 }, { "epoch": 0.17, "learning_rate": 4.893329326923077e-05, "loss": 2.2532, "step": 570 }, { "epoch": 0.17, "learning_rate": 4.893141526442308e-05, "loss": 2.2333, "step": 571 }, { "epoch": 0.17, "learning_rate": 4.892953725961539e-05, "loss": 2.0845, "step": 572 }, { "epoch": 0.17, "learning_rate": 4.8927659254807694e-05, "loss": 2.188, "step": 573 }, { "epoch": 0.17, "learning_rate": 4.8925781250000006e-05, "loss": 2.0602, "step": 574 }, { "epoch": 0.17, "learning_rate": 4.892390324519231e-05, "loss": 2.3814, "step": 575 }, { "epoch": 0.17, "learning_rate": 4.8922025240384615e-05, "loss": 2.3789, "step": 576 }, { "epoch": 0.17, "learning_rate": 4.8920147235576927e-05, "loss": 2.2507, "step": 577 }, { "epoch": 0.17, "learning_rate": 4.891826923076923e-05, "loss": 2.0484, "step": 578 }, { "epoch": 0.17, "learning_rate": 4.8916391225961536e-05, "loss": 2.2521, "step": 579 }, { "epoch": 0.17, "learning_rate": 4.891451322115385e-05, "loss": 2.2921, "step": 580 }, { "epoch": 0.17, "learning_rate": 4.891263521634616e-05, "loss": 2.1542, "step": 581 }, { "epoch": 0.17, "learning_rate": 4.8910757211538464e-05, "loss": 2.1211, "step": 582 }, { "epoch": 0.18, "learning_rate": 4.8908879206730776e-05, "loss": 2.2869, "step": 583 }, { "epoch": 0.18, "learning_rate": 4.890700120192308e-05, "loss": 2.2061, "step": 584 }, { "epoch": 0.18, "learning_rate": 4.8905123197115385e-05, "loss": 2.2359, "step": 585 }, { "epoch": 0.18, "learning_rate": 4.89032451923077e-05, "loss": 2.199, "step": 586 }, { "epoch": 0.18, "learning_rate": 4.89013671875e-05, "loss": 2.4223, "step": 587 }, { "epoch": 0.18, "learning_rate": 4.889948918269231e-05, "loss": 2.3644, "step": 588 }, { "epoch": 0.18, "learning_rate": 4.889761117788462e-05, "loss": 2.1135, "step": 589 }, { "epoch": 0.18, "learning_rate": 4.889573317307692e-05, "loss": 2.0273, "step": 590 }, { "epoch": 0.18, "learning_rate": 4.8893855168269234e-05, "loss": 2.284, "step": 591 }, { "epoch": 0.18, "learning_rate": 4.889197716346154e-05, "loss": 2.1696, "step": 592 }, { "epoch": 0.18, "learning_rate": 4.8890099158653844e-05, "loss": 2.3255, "step": 593 }, { "epoch": 0.18, "learning_rate": 4.888822115384616e-05, "loss": 2.4223, "step": 594 }, { "epoch": 0.18, "learning_rate": 4.888634314903847e-05, "loss": 2.1394, "step": 595 }, { "epoch": 0.18, "learning_rate": 4.888446514423077e-05, "loss": 2.1219, "step": 596 }, { "epoch": 0.18, "learning_rate": 4.888258713942308e-05, "loss": 2.1466, "step": 597 }, { "epoch": 0.18, "learning_rate": 4.888070913461539e-05, "loss": 2.1852, "step": 598 }, { "epoch": 0.18, "learning_rate": 4.887883112980769e-05, "loss": 2.2541, "step": 599 }, { "epoch": 0.18, "learning_rate": 4.8876953125000004e-05, "loss": 2.1437, "step": 600 }, { "epoch": 0.18, "learning_rate": 4.887507512019231e-05, "loss": 2.1189, "step": 601 }, { "epoch": 0.18, "learning_rate": 4.8873197115384614e-05, "loss": 2.2795, "step": 602 }, { "epoch": 0.18, "learning_rate": 4.8871319110576925e-05, "loss": 1.8608, "step": 603 }, { "epoch": 0.18, "learning_rate": 4.886944110576923e-05, "loss": 2.0331, "step": 604 }, { "epoch": 0.18, "learning_rate": 4.886756310096154e-05, "loss": 2.1981, "step": 605 }, { "epoch": 0.18, "learning_rate": 4.8865685096153846e-05, "loss": 2.374, "step": 606 }, { "epoch": 0.18, "learning_rate": 4.886380709134616e-05, "loss": 2.268, "step": 607 }, { "epoch": 0.18, "learning_rate": 4.886192908653847e-05, "loss": 2.1762, "step": 608 }, { "epoch": 0.18, "learning_rate": 4.8860051081730774e-05, "loss": 2.4049, "step": 609 }, { "epoch": 0.18, "learning_rate": 4.885817307692308e-05, "loss": 2.4158, "step": 610 }, { "epoch": 0.18, "learning_rate": 4.885629507211539e-05, "loss": 2.5167, "step": 611 }, { "epoch": 0.18, "learning_rate": 4.8854417067307695e-05, "loss": 2.1388, "step": 612 }, { "epoch": 0.18, "learning_rate": 4.88525390625e-05, "loss": 2.3401, "step": 613 }, { "epoch": 0.18, "learning_rate": 4.885066105769231e-05, "loss": 2.3862, "step": 614 }, { "epoch": 0.18, "learning_rate": 4.8848783052884616e-05, "loss": 2.1043, "step": 615 }, { "epoch": 0.19, "learning_rate": 4.884690504807692e-05, "loss": 2.5028, "step": 616 }, { "epoch": 0.19, "learning_rate": 4.884502704326923e-05, "loss": 2.1637, "step": 617 }, { "epoch": 0.19, "learning_rate": 4.884314903846154e-05, "loss": 2.3774, "step": 618 }, { "epoch": 0.19, "learning_rate": 4.884127103365385e-05, "loss": 2.0467, "step": 619 }, { "epoch": 0.19, "learning_rate": 4.883939302884616e-05, "loss": 2.0324, "step": 620 }, { "epoch": 0.19, "learning_rate": 4.8837515024038465e-05, "loss": 2.3375, "step": 621 }, { "epoch": 0.19, "learning_rate": 4.883563701923077e-05, "loss": 2.0364, "step": 622 }, { "epoch": 0.19, "learning_rate": 4.883375901442308e-05, "loss": 2.0147, "step": 623 }, { "epoch": 0.19, "learning_rate": 4.8831881009615386e-05, "loss": 1.9866, "step": 624 }, { "epoch": 0.19, "learning_rate": 4.88300030048077e-05, "loss": 2.5263, "step": 625 }, { "epoch": 0.19, "learning_rate": 4.8828125e-05, "loss": 2.2024, "step": 626 }, { "epoch": 0.19, "learning_rate": 4.882624699519231e-05, "loss": 2.1105, "step": 627 }, { "epoch": 0.19, "learning_rate": 4.882436899038462e-05, "loss": 2.1321, "step": 628 }, { "epoch": 0.19, "learning_rate": 4.8822490985576924e-05, "loss": 2.0077, "step": 629 }, { "epoch": 0.19, "learning_rate": 4.882061298076923e-05, "loss": 2.3369, "step": 630 }, { "epoch": 0.19, "learning_rate": 4.881873497596154e-05, "loss": 2.1173, "step": 631 }, { "epoch": 0.19, "learning_rate": 4.8816856971153845e-05, "loss": 2.2733, "step": 632 }, { "epoch": 0.19, "learning_rate": 4.8814978966346156e-05, "loss": 2.1628, "step": 633 }, { "epoch": 0.19, "learning_rate": 4.881310096153847e-05, "loss": 2.4741, "step": 634 }, { "epoch": 0.19, "learning_rate": 4.881122295673077e-05, "loss": 1.88, "step": 635 }, { "epoch": 0.19, "learning_rate": 4.880934495192308e-05, "loss": 1.9597, "step": 636 }, { "epoch": 0.19, "learning_rate": 4.880746694711539e-05, "loss": 2.1688, "step": 637 }, { "epoch": 0.19, "learning_rate": 4.8805588942307694e-05, "loss": 2.4679, "step": 638 }, { "epoch": 0.19, "learning_rate": 4.8803710937500005e-05, "loss": 2.5494, "step": 639 }, { "epoch": 0.19, "learning_rate": 4.880183293269231e-05, "loss": 2.2211, "step": 640 }, { "epoch": 0.19, "learning_rate": 4.8799954927884615e-05, "loss": 2.1594, "step": 641 }, { "epoch": 0.19, "learning_rate": 4.8798076923076926e-05, "loss": 2.2456, "step": 642 }, { "epoch": 0.19, "learning_rate": 4.879619891826923e-05, "loss": 2.438, "step": 643 }, { "epoch": 0.19, "learning_rate": 4.8794320913461536e-05, "loss": 2.3017, "step": 644 }, { "epoch": 0.19, "learning_rate": 4.879244290865385e-05, "loss": 2.29, "step": 645 }, { "epoch": 0.19, "learning_rate": 4.879056490384616e-05, "loss": 2.5007, "step": 646 }, { "epoch": 0.19, "learning_rate": 4.8788686899038464e-05, "loss": 2.4076, "step": 647 }, { "epoch": 0.19, "learning_rate": 4.8786808894230775e-05, "loss": 2.1906, "step": 648 }, { "epoch": 0.2, "learning_rate": 4.878493088942308e-05, "loss": 2.2042, "step": 649 }, { "epoch": 0.2, "learning_rate": 4.8783052884615385e-05, "loss": 2.1219, "step": 650 }, { "epoch": 0.2, "learning_rate": 4.8781174879807697e-05, "loss": 1.9967, "step": 651 }, { "epoch": 0.2, "learning_rate": 4.8779296875e-05, "loss": 2.1836, "step": 652 }, { "epoch": 0.2, "learning_rate": 4.877741887019231e-05, "loss": 2.046, "step": 653 }, { "epoch": 0.2, "learning_rate": 4.877554086538462e-05, "loss": 2.0517, "step": 654 }, { "epoch": 0.2, "learning_rate": 4.877366286057692e-05, "loss": 1.9586, "step": 655 }, { "epoch": 0.2, "learning_rate": 4.8771784855769234e-05, "loss": 2.3735, "step": 656 }, { "epoch": 0.2, "learning_rate": 4.876990685096154e-05, "loss": 2.1413, "step": 657 }, { "epoch": 0.2, "learning_rate": 4.8768028846153843e-05, "loss": 2.144, "step": 658 }, { "epoch": 0.2, "learning_rate": 4.876615084134616e-05, "loss": 1.7736, "step": 659 }, { "epoch": 0.2, "learning_rate": 4.8764272836538467e-05, "loss": 2.2877, "step": 660 }, { "epoch": 0.2, "learning_rate": 4.876239483173077e-05, "loss": 2.1878, "step": 661 }, { "epoch": 0.2, "learning_rate": 4.876051682692308e-05, "loss": 2.2972, "step": 662 }, { "epoch": 0.2, "learning_rate": 4.875863882211539e-05, "loss": 2.2476, "step": 663 }, { "epoch": 0.2, "learning_rate": 4.875676081730769e-05, "loss": 2.3822, "step": 664 }, { "epoch": 0.2, "learning_rate": 4.8754882812500004e-05, "loss": 2.2069, "step": 665 }, { "epoch": 0.2, "learning_rate": 4.875300480769231e-05, "loss": 2.1104, "step": 666 }, { "epoch": 0.2, "learning_rate": 4.8751126802884614e-05, "loss": 1.9784, "step": 667 }, { "epoch": 0.2, "learning_rate": 4.8749248798076925e-05, "loss": 2.3716, "step": 668 }, { "epoch": 0.2, "learning_rate": 4.874737079326923e-05, "loss": 2.1546, "step": 669 }, { "epoch": 0.2, "learning_rate": 4.874549278846154e-05, "loss": 2.5782, "step": 670 }, { "epoch": 0.2, "learning_rate": 4.8743614783653846e-05, "loss": 2.2338, "step": 671 }, { "epoch": 0.2, "learning_rate": 4.874173677884616e-05, "loss": 2.3409, "step": 672 }, { "epoch": 0.2, "learning_rate": 4.873985877403847e-05, "loss": 2.2925, "step": 673 }, { "epoch": 0.2, "learning_rate": 4.8737980769230774e-05, "loss": 1.8414, "step": 674 }, { "epoch": 0.2, "learning_rate": 4.873610276442308e-05, "loss": 2.2709, "step": 675 }, { "epoch": 0.2, "learning_rate": 4.873422475961539e-05, "loss": 2.2311, "step": 676 }, { "epoch": 0.2, "learning_rate": 4.8732346754807695e-05, "loss": 2.1807, "step": 677 }, { "epoch": 0.2, "learning_rate": 4.873046875e-05, "loss": 2.2988, "step": 678 }, { "epoch": 0.2, "learning_rate": 4.872859074519231e-05, "loss": 2.2334, "step": 679 }, { "epoch": 0.2, "learning_rate": 4.8726712740384616e-05, "loss": 2.1194, "step": 680 }, { "epoch": 0.2, "learning_rate": 4.872483473557692e-05, "loss": 2.3538, "step": 681 }, { "epoch": 0.2, "learning_rate": 4.872295673076923e-05, "loss": 2.1378, "step": 682 }, { "epoch": 0.21, "learning_rate": 4.872107872596154e-05, "loss": 2.2401, "step": 683 }, { "epoch": 0.21, "learning_rate": 4.871920072115385e-05, "loss": 2.0857, "step": 684 }, { "epoch": 0.21, "learning_rate": 4.871732271634616e-05, "loss": 2.1647, "step": 685 }, { "epoch": 0.21, "learning_rate": 4.8715444711538465e-05, "loss": 2.0373, "step": 686 }, { "epoch": 0.21, "learning_rate": 4.871356670673077e-05, "loss": 2.4302, "step": 687 }, { "epoch": 0.21, "learning_rate": 4.871168870192308e-05, "loss": 2.1866, "step": 688 }, { "epoch": 0.21, "learning_rate": 4.8709810697115386e-05, "loss": 2.3891, "step": 689 }, { "epoch": 0.21, "learning_rate": 4.87079326923077e-05, "loss": 2.5311, "step": 690 }, { "epoch": 0.21, "learning_rate": 4.87060546875e-05, "loss": 2.2056, "step": 691 }, { "epoch": 0.21, "learning_rate": 4.870417668269231e-05, "loss": 2.2804, "step": 692 }, { "epoch": 0.21, "learning_rate": 4.870229867788462e-05, "loss": 2.262, "step": 693 }, { "epoch": 0.21, "learning_rate": 4.8700420673076924e-05, "loss": 2.2846, "step": 694 }, { "epoch": 0.21, "learning_rate": 4.869854266826923e-05, "loss": 2.2494, "step": 695 }, { "epoch": 0.21, "learning_rate": 4.869666466346154e-05, "loss": 2.0545, "step": 696 }, { "epoch": 0.21, "learning_rate": 4.8694786658653845e-05, "loss": 2.5078, "step": 697 }, { "epoch": 0.21, "learning_rate": 4.8692908653846156e-05, "loss": 2.1523, "step": 698 }, { "epoch": 0.21, "learning_rate": 4.869103064903847e-05, "loss": 2.3142, "step": 699 }, { "epoch": 0.21, "learning_rate": 4.868915264423077e-05, "loss": 2.1342, "step": 700 }, { "epoch": 0.21, "learning_rate": 4.868727463942308e-05, "loss": 1.8883, "step": 701 }, { "epoch": 0.21, "learning_rate": 4.868539663461539e-05, "loss": 1.9859, "step": 702 }, { "epoch": 0.21, "learning_rate": 4.8683518629807694e-05, "loss": 2.3521, "step": 703 }, { "epoch": 0.21, "learning_rate": 4.8681640625000005e-05, "loss": 2.1362, "step": 704 }, { "epoch": 0.21, "learning_rate": 4.867976262019231e-05, "loss": 2.1738, "step": 705 }, { "epoch": 0.21, "learning_rate": 4.8677884615384615e-05, "loss": 2.4612, "step": 706 }, { "epoch": 0.21, "learning_rate": 4.8676006610576926e-05, "loss": 2.0916, "step": 707 }, { "epoch": 0.21, "learning_rate": 4.867412860576923e-05, "loss": 2.2267, "step": 708 }, { "epoch": 0.21, "learning_rate": 4.8672250600961536e-05, "loss": 2.0653, "step": 709 }, { "epoch": 0.21, "learning_rate": 4.867037259615385e-05, "loss": 2.4385, "step": 710 }, { "epoch": 0.21, "learning_rate": 4.866849459134616e-05, "loss": 2.1707, "step": 711 }, { "epoch": 0.21, "learning_rate": 4.8666616586538464e-05, "loss": 2.3214, "step": 712 }, { "epoch": 0.21, "learning_rate": 4.8664738581730775e-05, "loss": 2.0906, "step": 713 }, { "epoch": 0.21, "learning_rate": 4.866286057692308e-05, "loss": 1.9859, "step": 714 }, { "epoch": 0.21, "learning_rate": 4.8660982572115385e-05, "loss": 1.7457, "step": 715 }, { "epoch": 0.22, "learning_rate": 4.8659104567307696e-05, "loss": 2.1222, "step": 716 }, { "epoch": 0.22, "learning_rate": 4.86572265625e-05, "loss": 2.121, "step": 717 }, { "epoch": 0.22, "learning_rate": 4.865534855769231e-05, "loss": 2.3016, "step": 718 }, { "epoch": 0.22, "learning_rate": 4.865347055288462e-05, "loss": 2.283, "step": 719 }, { "epoch": 0.22, "learning_rate": 4.865159254807692e-05, "loss": 2.2507, "step": 720 }, { "epoch": 0.22, "learning_rate": 4.8649714543269234e-05, "loss": 2.0882, "step": 721 }, { "epoch": 0.22, "learning_rate": 4.864783653846154e-05, "loss": 2.191, "step": 722 }, { "epoch": 0.22, "learning_rate": 4.864595853365384e-05, "loss": 2.2288, "step": 723 }, { "epoch": 0.22, "learning_rate": 4.864408052884616e-05, "loss": 2.5039, "step": 724 }, { "epoch": 0.22, "learning_rate": 4.8642202524038466e-05, "loss": 2.1638, "step": 725 }, { "epoch": 0.22, "learning_rate": 4.864032451923077e-05, "loss": 2.0002, "step": 726 }, { "epoch": 0.22, "learning_rate": 4.863844651442308e-05, "loss": 2.2729, "step": 727 }, { "epoch": 0.22, "learning_rate": 4.863656850961539e-05, "loss": 2.3664, "step": 728 }, { "epoch": 0.22, "learning_rate": 4.863469050480769e-05, "loss": 2.0375, "step": 729 }, { "epoch": 0.22, "learning_rate": 4.8632812500000004e-05, "loss": 2.3807, "step": 730 }, { "epoch": 0.22, "learning_rate": 4.863093449519231e-05, "loss": 2.3716, "step": 731 }, { "epoch": 0.22, "learning_rate": 4.862905649038462e-05, "loss": 2.3271, "step": 732 }, { "epoch": 0.22, "learning_rate": 4.8627178485576925e-05, "loss": 2.115, "step": 733 }, { "epoch": 0.22, "learning_rate": 4.862530048076923e-05, "loss": 2.2212, "step": 734 }, { "epoch": 0.22, "learning_rate": 4.862342247596154e-05, "loss": 2.2466, "step": 735 }, { "epoch": 0.22, "learning_rate": 4.8621544471153846e-05, "loss": 2.1542, "step": 736 }, { "epoch": 0.22, "learning_rate": 4.861966646634616e-05, "loss": 2.3671, "step": 737 }, { "epoch": 0.22, "learning_rate": 4.861778846153847e-05, "loss": 2.4722, "step": 738 }, { "epoch": 0.22, "learning_rate": 4.8615910456730774e-05, "loss": 2.4874, "step": 739 }, { "epoch": 0.22, "learning_rate": 4.861403245192308e-05, "loss": 2.3156, "step": 740 }, { "epoch": 0.22, "learning_rate": 4.861215444711539e-05, "loss": 2.1666, "step": 741 }, { "epoch": 0.22, "learning_rate": 4.8610276442307695e-05, "loss": 2.2498, "step": 742 }, { "epoch": 0.22, "learning_rate": 4.86083984375e-05, "loss": 2.4053, "step": 743 }, { "epoch": 0.22, "learning_rate": 4.860652043269231e-05, "loss": 2.3006, "step": 744 }, { "epoch": 0.22, "learning_rate": 4.8604642427884616e-05, "loss": 2.2588, "step": 745 }, { "epoch": 0.22, "learning_rate": 4.860276442307692e-05, "loss": 2.0118, "step": 746 }, { "epoch": 0.22, "learning_rate": 4.860088641826923e-05, "loss": 2.0682, "step": 747 }, { "epoch": 0.22, "learning_rate": 4.859900841346154e-05, "loss": 2.3357, "step": 748 }, { "epoch": 0.23, "learning_rate": 4.859713040865385e-05, "loss": 2.0669, "step": 749 }, { "epoch": 0.23, "learning_rate": 4.859525240384616e-05, "loss": 2.1531, "step": 750 }, { "epoch": 0.23, "learning_rate": 4.8593374399038465e-05, "loss": 2.1467, "step": 751 }, { "epoch": 0.23, "learning_rate": 4.859149639423078e-05, "loss": 2.1311, "step": 752 }, { "epoch": 0.23, "learning_rate": 4.858961838942308e-05, "loss": 2.306, "step": 753 }, { "epoch": 0.23, "learning_rate": 4.8587740384615386e-05, "loss": 2.0391, "step": 754 }, { "epoch": 0.23, "learning_rate": 4.85858623798077e-05, "loss": 2.3293, "step": 755 }, { "epoch": 0.23, "learning_rate": 4.8583984375e-05, "loss": 2.129, "step": 756 }, { "epoch": 0.23, "learning_rate": 4.858210637019231e-05, "loss": 2.2328, "step": 757 }, { "epoch": 0.23, "learning_rate": 4.858022836538462e-05, "loss": 2.3473, "step": 758 }, { "epoch": 0.23, "learning_rate": 4.8578350360576924e-05, "loss": 2.1439, "step": 759 }, { "epoch": 0.23, "learning_rate": 4.857647235576923e-05, "loss": 2.5653, "step": 760 }, { "epoch": 0.23, "learning_rate": 4.857459435096154e-05, "loss": 2.1356, "step": 761 }, { "epoch": 0.23, "learning_rate": 4.8572716346153845e-05, "loss": 2.0605, "step": 762 }, { "epoch": 0.23, "learning_rate": 4.8570838341346156e-05, "loss": 2.2119, "step": 763 }, { "epoch": 0.23, "learning_rate": 4.856896033653847e-05, "loss": 2.0198, "step": 764 }, { "epoch": 0.23, "learning_rate": 4.856708233173077e-05, "loss": 2.2731, "step": 765 }, { "epoch": 0.23, "learning_rate": 4.856520432692308e-05, "loss": 2.0263, "step": 766 }, { "epoch": 0.23, "learning_rate": 4.856332632211539e-05, "loss": 2.0964, "step": 767 }, { "epoch": 0.23, "learning_rate": 4.8561448317307694e-05, "loss": 2.1917, "step": 768 }, { "epoch": 0.23, "learning_rate": 4.8559570312500005e-05, "loss": 2.2324, "step": 769 }, { "epoch": 0.23, "learning_rate": 4.855769230769231e-05, "loss": 2.2083, "step": 770 }, { "epoch": 0.23, "learning_rate": 4.8555814302884615e-05, "loss": 2.4696, "step": 771 }, { "epoch": 0.23, "learning_rate": 4.8553936298076926e-05, "loss": 2.331, "step": 772 }, { "epoch": 0.23, "learning_rate": 4.855205829326923e-05, "loss": 2.1674, "step": 773 }, { "epoch": 0.23, "learning_rate": 4.8550180288461536e-05, "loss": 2.1942, "step": 774 }, { "epoch": 0.23, "learning_rate": 4.854830228365385e-05, "loss": 2.5216, "step": 775 }, { "epoch": 0.23, "learning_rate": 4.854642427884616e-05, "loss": 2.183, "step": 776 }, { "epoch": 0.23, "learning_rate": 4.8544546274038464e-05, "loss": 2.1777, "step": 777 }, { "epoch": 0.23, "learning_rate": 4.8542668269230775e-05, "loss": 2.257, "step": 778 }, { "epoch": 0.23, "learning_rate": 4.854079026442308e-05, "loss": 2.1034, "step": 779 }, { "epoch": 0.23, "learning_rate": 4.8538912259615385e-05, "loss": 2.1665, "step": 780 }, { "epoch": 0.23, "learning_rate": 4.8537034254807696e-05, "loss": 2.1556, "step": 781 }, { "epoch": 0.23, "learning_rate": 4.853515625e-05, "loss": 2.0346, "step": 782 }, { "epoch": 0.24, "learning_rate": 4.853327824519231e-05, "loss": 2.4852, "step": 783 }, { "epoch": 0.24, "learning_rate": 4.853140024038462e-05, "loss": 1.9088, "step": 784 }, { "epoch": 0.24, "learning_rate": 4.852952223557692e-05, "loss": 2.0765, "step": 785 }, { "epoch": 0.24, "learning_rate": 4.8527644230769234e-05, "loss": 2.1866, "step": 786 }, { "epoch": 0.24, "learning_rate": 4.852576622596154e-05, "loss": 2.469, "step": 787 }, { "epoch": 0.24, "learning_rate": 4.852388822115384e-05, "loss": 2.2122, "step": 788 }, { "epoch": 0.24, "learning_rate": 4.852201021634616e-05, "loss": 2.0837, "step": 789 }, { "epoch": 0.24, "learning_rate": 4.8520132211538466e-05, "loss": 2.3494, "step": 790 }, { "epoch": 0.24, "learning_rate": 4.851825420673077e-05, "loss": 2.1502, "step": 791 }, { "epoch": 0.24, "learning_rate": 4.851637620192308e-05, "loss": 2.2487, "step": 792 }, { "epoch": 0.24, "learning_rate": 4.851449819711539e-05, "loss": 2.3021, "step": 793 }, { "epoch": 0.24, "learning_rate": 4.851262019230769e-05, "loss": 2.24, "step": 794 }, { "epoch": 0.24, "learning_rate": 4.8510742187500004e-05, "loss": 2.3837, "step": 795 }, { "epoch": 0.24, "learning_rate": 4.850886418269231e-05, "loss": 2.4781, "step": 796 }, { "epoch": 0.24, "learning_rate": 4.850698617788462e-05, "loss": 2.0703, "step": 797 }, { "epoch": 0.24, "learning_rate": 4.8505108173076925e-05, "loss": 2.0254, "step": 798 }, { "epoch": 0.24, "learning_rate": 4.850323016826923e-05, "loss": 2.1532, "step": 799 }, { "epoch": 0.24, "learning_rate": 4.850135216346154e-05, "loss": 2.1843, "step": 800 }, { "epoch": 0.24, "learning_rate": 4.8499474158653846e-05, "loss": 2.1833, "step": 801 }, { "epoch": 0.24, "learning_rate": 4.849759615384616e-05, "loss": 2.5647, "step": 802 }, { "epoch": 0.24, "learning_rate": 4.849571814903847e-05, "loss": 2.15, "step": 803 }, { "epoch": 0.24, "learning_rate": 4.8493840144230774e-05, "loss": 1.9801, "step": 804 }, { "epoch": 0.24, "learning_rate": 4.849196213942308e-05, "loss": 2.402, "step": 805 }, { "epoch": 0.24, "learning_rate": 4.849008413461539e-05, "loss": 2.1225, "step": 806 }, { "epoch": 0.24, "learning_rate": 4.8488206129807695e-05, "loss": 2.4559, "step": 807 }, { "epoch": 0.24, "learning_rate": 4.8486328125e-05, "loss": 2.2326, "step": 808 }, { "epoch": 0.24, "learning_rate": 4.848445012019231e-05, "loss": 2.3829, "step": 809 }, { "epoch": 0.24, "learning_rate": 4.8482572115384616e-05, "loss": 1.8953, "step": 810 }, { "epoch": 0.24, "learning_rate": 4.848069411057692e-05, "loss": 2.2636, "step": 811 }, { "epoch": 0.24, "learning_rate": 4.847881610576923e-05, "loss": 2.2078, "step": 812 }, { "epoch": 0.24, "learning_rate": 4.847693810096154e-05, "loss": 2.1683, "step": 813 }, { "epoch": 0.24, "learning_rate": 4.847506009615385e-05, "loss": 2.09, "step": 814 }, { "epoch": 0.24, "learning_rate": 4.847318209134616e-05, "loss": 2.1133, "step": 815 }, { "epoch": 0.25, "learning_rate": 4.8471304086538465e-05, "loss": 2.1816, "step": 816 }, { "epoch": 0.25, "learning_rate": 4.8469426081730777e-05, "loss": 2.0806, "step": 817 }, { "epoch": 0.25, "learning_rate": 4.846754807692308e-05, "loss": 2.2255, "step": 818 }, { "epoch": 0.25, "learning_rate": 4.8465670072115386e-05, "loss": 2.1392, "step": 819 }, { "epoch": 0.25, "learning_rate": 4.84637920673077e-05, "loss": 2.2005, "step": 820 }, { "epoch": 0.25, "learning_rate": 4.84619140625e-05, "loss": 2.2087, "step": 821 }, { "epoch": 0.25, "learning_rate": 4.846003605769231e-05, "loss": 2.105, "step": 822 }, { "epoch": 0.25, "learning_rate": 4.845815805288462e-05, "loss": 1.8741, "step": 823 }, { "epoch": 0.25, "learning_rate": 4.8456280048076923e-05, "loss": 2.1229, "step": 824 }, { "epoch": 0.25, "learning_rate": 4.845440204326923e-05, "loss": 2.2681, "step": 825 }, { "epoch": 0.25, "learning_rate": 4.845252403846154e-05, "loss": 2.5036, "step": 826 }, { "epoch": 0.25, "learning_rate": 4.8450646033653845e-05, "loss": 2.0497, "step": 827 }, { "epoch": 0.25, "learning_rate": 4.8448768028846156e-05, "loss": 2.0464, "step": 828 }, { "epoch": 0.25, "learning_rate": 4.844689002403847e-05, "loss": 2.4016, "step": 829 }, { "epoch": 0.25, "learning_rate": 4.844501201923077e-05, "loss": 2.3499, "step": 830 }, { "epoch": 0.25, "learning_rate": 4.844313401442308e-05, "loss": 2.1472, "step": 831 }, { "epoch": 0.25, "learning_rate": 4.844125600961539e-05, "loss": 2.4409, "step": 832 }, { "epoch": 0.25, "learning_rate": 4.8439378004807694e-05, "loss": 2.1716, "step": 833 }, { "epoch": 0.25, "learning_rate": 4.8437500000000005e-05, "loss": 2.239, "step": 834 }, { "epoch": 0.25, "learning_rate": 4.843562199519231e-05, "loss": 2.1349, "step": 835 }, { "epoch": 0.25, "learning_rate": 4.8433743990384615e-05, "loss": 2.3059, "step": 836 }, { "epoch": 0.25, "learning_rate": 4.8431865985576926e-05, "loss": 2.4988, "step": 837 }, { "epoch": 0.25, "learning_rate": 4.842998798076923e-05, "loss": 2.1175, "step": 838 }, { "epoch": 0.25, "learning_rate": 4.8428109975961536e-05, "loss": 2.0121, "step": 839 }, { "epoch": 0.25, "learning_rate": 4.842623197115385e-05, "loss": 2.1997, "step": 840 }, { "epoch": 0.25, "learning_rate": 4.842435396634616e-05, "loss": 2.369, "step": 841 }, { "epoch": 0.25, "learning_rate": 4.8422475961538464e-05, "loss": 2.2978, "step": 842 }, { "epoch": 0.25, "learning_rate": 4.8420597956730775e-05, "loss": 2.0787, "step": 843 }, { "epoch": 0.25, "learning_rate": 4.841871995192308e-05, "loss": 2.3632, "step": 844 }, { "epoch": 0.25, "learning_rate": 4.8416841947115385e-05, "loss": 2.3025, "step": 845 }, { "epoch": 0.25, "learning_rate": 4.8414963942307696e-05, "loss": 2.161, "step": 846 }, { "epoch": 0.25, "learning_rate": 4.84130859375e-05, "loss": 2.2261, "step": 847 }, { "epoch": 0.25, "learning_rate": 4.841120793269231e-05, "loss": 2.1289, "step": 848 }, { "epoch": 0.26, "learning_rate": 4.840932992788462e-05, "loss": 2.3225, "step": 849 }, { "epoch": 0.26, "learning_rate": 4.840745192307692e-05, "loss": 1.8558, "step": 850 }, { "epoch": 0.26, "learning_rate": 4.8405573918269234e-05, "loss": 2.0578, "step": 851 }, { "epoch": 0.26, "learning_rate": 4.840369591346154e-05, "loss": 2.0502, "step": 852 }, { "epoch": 0.26, "learning_rate": 4.840181790865384e-05, "loss": 2.0923, "step": 853 }, { "epoch": 0.26, "learning_rate": 4.839993990384616e-05, "loss": 2.0201, "step": 854 }, { "epoch": 0.26, "learning_rate": 4.8398061899038466e-05, "loss": 2.2392, "step": 855 }, { "epoch": 0.26, "learning_rate": 4.839618389423077e-05, "loss": 2.2081, "step": 856 }, { "epoch": 0.26, "learning_rate": 4.839430588942308e-05, "loss": 2.2082, "step": 857 }, { "epoch": 0.26, "learning_rate": 4.839242788461539e-05, "loss": 1.9974, "step": 858 }, { "epoch": 0.26, "learning_rate": 4.839054987980769e-05, "loss": 2.0959, "step": 859 }, { "epoch": 0.26, "learning_rate": 4.8388671875000004e-05, "loss": 2.264, "step": 860 }, { "epoch": 0.26, "learning_rate": 4.838679387019231e-05, "loss": 1.862, "step": 861 }, { "epoch": 0.26, "learning_rate": 4.838491586538462e-05, "loss": 2.2986, "step": 862 }, { "epoch": 0.26, "learning_rate": 4.8383037860576925e-05, "loss": 2.0079, "step": 863 }, { "epoch": 0.26, "learning_rate": 4.838115985576923e-05, "loss": 2.5154, "step": 864 }, { "epoch": 0.26, "learning_rate": 4.837928185096154e-05, "loss": 2.1266, "step": 865 }, { "epoch": 0.26, "learning_rate": 4.8377403846153846e-05, "loss": 2.2245, "step": 866 }, { "epoch": 0.26, "learning_rate": 4.837552584134616e-05, "loss": 2.4506, "step": 867 }, { "epoch": 0.26, "learning_rate": 4.837364783653847e-05, "loss": 2.2146, "step": 868 }, { "epoch": 0.26, "learning_rate": 4.8371769831730774e-05, "loss": 1.9704, "step": 869 }, { "epoch": 0.26, "learning_rate": 4.836989182692308e-05, "loss": 1.9271, "step": 870 }, { "epoch": 0.26, "learning_rate": 4.836801382211539e-05, "loss": 2.3933, "step": 871 }, { "epoch": 0.26, "learning_rate": 4.8366135817307695e-05, "loss": 2.1217, "step": 872 }, { "epoch": 0.26, "learning_rate": 4.83642578125e-05, "loss": 2.2709, "step": 873 }, { "epoch": 0.26, "learning_rate": 4.836237980769231e-05, "loss": 2.2198, "step": 874 }, { "epoch": 0.26, "learning_rate": 4.8360501802884616e-05, "loss": 2.259, "step": 875 }, { "epoch": 0.26, "learning_rate": 4.835862379807693e-05, "loss": 2.2606, "step": 876 }, { "epoch": 0.26, "learning_rate": 4.835674579326923e-05, "loss": 2.3385, "step": 877 }, { "epoch": 0.26, "learning_rate": 4.835486778846154e-05, "loss": 2.3349, "step": 878 }, { "epoch": 0.26, "learning_rate": 4.835298978365385e-05, "loss": 2.3843, "step": 879 }, { "epoch": 0.26, "learning_rate": 4.835111177884616e-05, "loss": 2.5739, "step": 880 }, { "epoch": 0.26, "learning_rate": 4.8349233774038465e-05, "loss": 2.3151, "step": 881 }, { "epoch": 0.27, "learning_rate": 4.8347355769230776e-05, "loss": 2.1423, "step": 882 }, { "epoch": 0.27, "learning_rate": 4.834547776442308e-05, "loss": 2.2343, "step": 883 }, { "epoch": 0.27, "learning_rate": 4.8343599759615386e-05, "loss": 2.2592, "step": 884 }, { "epoch": 0.27, "learning_rate": 4.83417217548077e-05, "loss": 2.0929, "step": 885 }, { "epoch": 0.27, "learning_rate": 4.833984375e-05, "loss": 2.2996, "step": 886 }, { "epoch": 0.27, "learning_rate": 4.833796574519231e-05, "loss": 2.1362, "step": 887 }, { "epoch": 0.27, "learning_rate": 4.833608774038462e-05, "loss": 2.26, "step": 888 }, { "epoch": 0.27, "learning_rate": 4.833420973557692e-05, "loss": 2.2764, "step": 889 }, { "epoch": 0.27, "learning_rate": 4.833233173076923e-05, "loss": 2.4584, "step": 890 }, { "epoch": 0.27, "learning_rate": 4.833045372596154e-05, "loss": 2.4351, "step": 891 }, { "epoch": 0.27, "learning_rate": 4.8328575721153844e-05, "loss": 2.0768, "step": 892 }, { "epoch": 0.27, "learning_rate": 4.8326697716346156e-05, "loss": 2.061, "step": 893 }, { "epoch": 0.27, "learning_rate": 4.832481971153847e-05, "loss": 2.2232, "step": 894 }, { "epoch": 0.27, "learning_rate": 4.832294170673077e-05, "loss": 2.1076, "step": 895 }, { "epoch": 0.27, "learning_rate": 4.8321063701923084e-05, "loss": 2.5303, "step": 896 }, { "epoch": 0.27, "learning_rate": 4.831918569711539e-05, "loss": 2.3081, "step": 897 }, { "epoch": 0.27, "learning_rate": 4.8317307692307693e-05, "loss": 2.0107, "step": 898 }, { "epoch": 0.27, "learning_rate": 4.8315429687500005e-05, "loss": 2.0944, "step": 899 }, { "epoch": 0.27, "learning_rate": 4.831355168269231e-05, "loss": 2.2704, "step": 900 }, { "epoch": 0.27, "learning_rate": 4.8311673677884615e-05, "loss": 2.0219, "step": 901 }, { "epoch": 0.27, "learning_rate": 4.8309795673076926e-05, "loss": 2.1815, "step": 902 }, { "epoch": 0.27, "learning_rate": 4.830791766826923e-05, "loss": 2.2105, "step": 903 }, { "epoch": 0.27, "learning_rate": 4.8306039663461536e-05, "loss": 2.017, "step": 904 }, { "epoch": 0.27, "learning_rate": 4.830416165865385e-05, "loss": 1.9788, "step": 905 }, { "epoch": 0.27, "learning_rate": 4.830228365384616e-05, "loss": 2.3345, "step": 906 }, { "epoch": 0.27, "learning_rate": 4.8300405649038463e-05, "loss": 2.0534, "step": 907 }, { "epoch": 0.27, "learning_rate": 4.8298527644230775e-05, "loss": 2.3359, "step": 908 }, { "epoch": 0.27, "learning_rate": 4.829664963942308e-05, "loss": 2.3222, "step": 909 }, { "epoch": 0.27, "learning_rate": 4.8294771634615385e-05, "loss": 2.4026, "step": 910 }, { "epoch": 0.27, "learning_rate": 4.8292893629807696e-05, "loss": 2.2895, "step": 911 }, { "epoch": 0.27, "learning_rate": 4.8291015625e-05, "loss": 2.5739, "step": 912 }, { "epoch": 0.27, "learning_rate": 4.828913762019231e-05, "loss": 2.3598, "step": 913 }, { "epoch": 0.27, "learning_rate": 4.828725961538462e-05, "loss": 2.0937, "step": 914 }, { "epoch": 0.27, "learning_rate": 4.828538161057692e-05, "loss": 2.3558, "step": 915 }, { "epoch": 0.28, "learning_rate": 4.8283503605769234e-05, "loss": 2.2147, "step": 916 }, { "epoch": 0.28, "learning_rate": 4.828162560096154e-05, "loss": 2.5396, "step": 917 }, { "epoch": 0.28, "learning_rate": 4.827974759615384e-05, "loss": 2.1833, "step": 918 }, { "epoch": 0.28, "learning_rate": 4.827786959134616e-05, "loss": 2.3564, "step": 919 }, { "epoch": 0.28, "learning_rate": 4.8275991586538466e-05, "loss": 2.0427, "step": 920 }, { "epoch": 0.28, "learning_rate": 4.827411358173077e-05, "loss": 1.9432, "step": 921 }, { "epoch": 0.28, "learning_rate": 4.827223557692308e-05, "loss": 2.2361, "step": 922 }, { "epoch": 0.28, "learning_rate": 4.827035757211539e-05, "loss": 2.2371, "step": 923 }, { "epoch": 0.28, "learning_rate": 4.826847956730769e-05, "loss": 2.2036, "step": 924 }, { "epoch": 0.28, "learning_rate": 4.8266601562500004e-05, "loss": 2.1094, "step": 925 }, { "epoch": 0.28, "learning_rate": 4.826472355769231e-05, "loss": 2.3621, "step": 926 }, { "epoch": 0.28, "learning_rate": 4.826284555288462e-05, "loss": 2.2701, "step": 927 }, { "epoch": 0.28, "learning_rate": 4.8260967548076925e-05, "loss": 2.0703, "step": 928 }, { "epoch": 0.28, "learning_rate": 4.825908954326923e-05, "loss": 2.21, "step": 929 }, { "epoch": 0.28, "learning_rate": 4.825721153846154e-05, "loss": 2.1133, "step": 930 }, { "epoch": 0.28, "learning_rate": 4.8255333533653846e-05, "loss": 2.0301, "step": 931 }, { "epoch": 0.28, "learning_rate": 4.825345552884616e-05, "loss": 2.3174, "step": 932 }, { "epoch": 0.28, "learning_rate": 4.825157752403847e-05, "loss": 2.1932, "step": 933 }, { "epoch": 0.28, "learning_rate": 4.8249699519230774e-05, "loss": 2.2115, "step": 934 }, { "epoch": 0.28, "learning_rate": 4.824782151442308e-05, "loss": 2.3685, "step": 935 }, { "epoch": 0.28, "learning_rate": 4.824594350961539e-05, "loss": 2.2328, "step": 936 }, { "epoch": 0.28, "learning_rate": 4.8244065504807695e-05, "loss": 2.0032, "step": 937 }, { "epoch": 0.28, "learning_rate": 4.82421875e-05, "loss": 2.246, "step": 938 }, { "epoch": 0.28, "learning_rate": 4.824030949519231e-05, "loss": 2.1036, "step": 939 }, { "epoch": 0.28, "learning_rate": 4.8238431490384616e-05, "loss": 2.2579, "step": 940 }, { "epoch": 0.28, "learning_rate": 4.823655348557693e-05, "loss": 2.2193, "step": 941 }, { "epoch": 0.28, "learning_rate": 4.823467548076923e-05, "loss": 2.2799, "step": 942 }, { "epoch": 0.28, "learning_rate": 4.823279747596154e-05, "loss": 1.9664, "step": 943 }, { "epoch": 0.28, "learning_rate": 4.823091947115385e-05, "loss": 2.1279, "step": 944 }, { "epoch": 0.28, "learning_rate": 4.822904146634616e-05, "loss": 2.1696, "step": 945 }, { "epoch": 0.28, "learning_rate": 4.8227163461538465e-05, "loss": 2.4795, "step": 946 }, { "epoch": 0.28, "learning_rate": 4.8225285456730776e-05, "loss": 2.1481, "step": 947 }, { "epoch": 0.28, "learning_rate": 4.822340745192308e-05, "loss": 2.1228, "step": 948 }, { "epoch": 0.29, "learning_rate": 4.8221529447115386e-05, "loss": 2.2581, "step": 949 }, { "epoch": 0.29, "learning_rate": 4.82196514423077e-05, "loss": 1.9523, "step": 950 }, { "epoch": 0.29, "learning_rate": 4.82177734375e-05, "loss": 2.2271, "step": 951 }, { "epoch": 0.29, "learning_rate": 4.821589543269231e-05, "loss": 2.0706, "step": 952 }, { "epoch": 0.29, "learning_rate": 4.821401742788462e-05, "loss": 2.3105, "step": 953 }, { "epoch": 0.29, "learning_rate": 4.821213942307692e-05, "loss": 2.24, "step": 954 }, { "epoch": 0.29, "learning_rate": 4.821026141826923e-05, "loss": 2.4936, "step": 955 }, { "epoch": 0.29, "learning_rate": 4.820838341346154e-05, "loss": 2.3643, "step": 956 }, { "epoch": 0.29, "learning_rate": 4.8206505408653844e-05, "loss": 2.0355, "step": 957 }, { "epoch": 0.29, "learning_rate": 4.8204627403846156e-05, "loss": 2.0423, "step": 958 }, { "epoch": 0.29, "learning_rate": 4.820274939903847e-05, "loss": 2.3425, "step": 959 }, { "epoch": 0.29, "learning_rate": 4.820087139423077e-05, "loss": 2.3454, "step": 960 }, { "epoch": 0.29, "learning_rate": 4.8198993389423084e-05, "loss": 2.0486, "step": 961 }, { "epoch": 0.29, "learning_rate": 4.819711538461539e-05, "loss": 2.1863, "step": 962 }, { "epoch": 0.29, "learning_rate": 4.819523737980769e-05, "loss": 2.0522, "step": 963 }, { "epoch": 0.29, "learning_rate": 4.8193359375000005e-05, "loss": 1.8479, "step": 964 }, { "epoch": 0.29, "learning_rate": 4.819148137019231e-05, "loss": 2.4404, "step": 965 }, { "epoch": 0.29, "learning_rate": 4.8189603365384614e-05, "loss": 2.2513, "step": 966 }, { "epoch": 0.29, "learning_rate": 4.8187725360576926e-05, "loss": 2.1699, "step": 967 }, { "epoch": 0.29, "learning_rate": 4.818584735576923e-05, "loss": 2.1902, "step": 968 }, { "epoch": 0.29, "learning_rate": 4.8183969350961535e-05, "loss": 2.3651, "step": 969 }, { "epoch": 0.29, "learning_rate": 4.818209134615385e-05, "loss": 2.3029, "step": 970 }, { "epoch": 0.29, "learning_rate": 4.818021334134616e-05, "loss": 1.8395, "step": 971 }, { "epoch": 0.29, "learning_rate": 4.817833533653846e-05, "loss": 2.195, "step": 972 }, { "epoch": 0.29, "learning_rate": 4.8176457331730775e-05, "loss": 2.04, "step": 973 }, { "epoch": 0.29, "learning_rate": 4.817457932692308e-05, "loss": 2.2222, "step": 974 }, { "epoch": 0.29, "learning_rate": 4.8172701322115384e-05, "loss": 2.1705, "step": 975 }, { "epoch": 0.29, "learning_rate": 4.8170823317307696e-05, "loss": 2.2493, "step": 976 }, { "epoch": 0.29, "learning_rate": 4.81689453125e-05, "loss": 1.9376, "step": 977 }, { "epoch": 0.29, "learning_rate": 4.816706730769231e-05, "loss": 2.3742, "step": 978 }, { "epoch": 0.29, "learning_rate": 4.816518930288462e-05, "loss": 2.195, "step": 979 }, { "epoch": 0.29, "learning_rate": 4.816331129807692e-05, "loss": 2.3483, "step": 980 }, { "epoch": 0.29, "learning_rate": 4.8161433293269233e-05, "loss": 2.2839, "step": 981 }, { "epoch": 0.3, "learning_rate": 4.815955528846154e-05, "loss": 2.1339, "step": 982 }, { "epoch": 0.3, "learning_rate": 4.815767728365384e-05, "loss": 2.1998, "step": 983 }, { "epoch": 0.3, "learning_rate": 4.815579927884616e-05, "loss": 2.0702, "step": 984 }, { "epoch": 0.3, "learning_rate": 4.8153921274038466e-05, "loss": 2.2248, "step": 985 }, { "epoch": 0.3, "learning_rate": 4.815204326923077e-05, "loss": 2.4533, "step": 986 }, { "epoch": 0.3, "learning_rate": 4.815016526442308e-05, "loss": 2.2312, "step": 987 }, { "epoch": 0.3, "learning_rate": 4.814828725961539e-05, "loss": 2.3298, "step": 988 }, { "epoch": 0.3, "learning_rate": 4.814640925480769e-05, "loss": 2.3733, "step": 989 }, { "epoch": 0.3, "learning_rate": 4.8144531250000003e-05, "loss": 2.1128, "step": 990 }, { "epoch": 0.3, "learning_rate": 4.814265324519231e-05, "loss": 2.2317, "step": 991 }, { "epoch": 0.3, "learning_rate": 4.814077524038462e-05, "loss": 2.0809, "step": 992 }, { "epoch": 0.3, "learning_rate": 4.8138897235576925e-05, "loss": 2.1002, "step": 993 }, { "epoch": 0.3, "learning_rate": 4.813701923076923e-05, "loss": 2.0892, "step": 994 }, { "epoch": 0.3, "learning_rate": 4.813514122596154e-05, "loss": 2.3258, "step": 995 }, { "epoch": 0.3, "learning_rate": 4.8133263221153846e-05, "loss": 2.1991, "step": 996 }, { "epoch": 0.3, "learning_rate": 4.813138521634616e-05, "loss": 2.2213, "step": 997 }, { "epoch": 0.3, "learning_rate": 4.812950721153847e-05, "loss": 2.3987, "step": 998 }, { "epoch": 0.3, "learning_rate": 4.8127629206730774e-05, "loss": 1.9397, "step": 999 }, { "epoch": 0.3, "learning_rate": 4.812575120192308e-05, "loss": 2.3347, "step": 1000 }, { "epoch": 0.3, "learning_rate": 4.812387319711539e-05, "loss": 2.1813, "step": 1001 }, { "epoch": 0.3, "learning_rate": 4.8121995192307695e-05, "loss": 2.3565, "step": 1002 }, { "epoch": 0.3, "learning_rate": 4.81201171875e-05, "loss": 2.3646, "step": 1003 }, { "epoch": 0.3, "learning_rate": 4.811823918269231e-05, "loss": 2.2292, "step": 1004 }, { "epoch": 0.3, "learning_rate": 4.8116361177884616e-05, "loss": 2.2404, "step": 1005 }, { "epoch": 0.3, "learning_rate": 4.811448317307693e-05, "loss": 2.052, "step": 1006 }, { "epoch": 0.3, "learning_rate": 4.811260516826923e-05, "loss": 2.2381, "step": 1007 }, { "epoch": 0.3, "learning_rate": 4.811072716346154e-05, "loss": 2.3411, "step": 1008 }, { "epoch": 0.3, "learning_rate": 4.810884915865385e-05, "loss": 2.0055, "step": 1009 }, { "epoch": 0.3, "learning_rate": 4.810697115384616e-05, "loss": 2.3527, "step": 1010 }, { "epoch": 0.3, "learning_rate": 4.8105093149038465e-05, "loss": 2.1889, "step": 1011 }, { "epoch": 0.3, "learning_rate": 4.8103215144230776e-05, "loss": 2.2308, "step": 1012 }, { "epoch": 0.3, "learning_rate": 4.810133713942308e-05, "loss": 2.3953, "step": 1013 }, { "epoch": 0.3, "learning_rate": 4.8099459134615386e-05, "loss": 2.0835, "step": 1014 }, { "epoch": 0.3, "learning_rate": 4.80975811298077e-05, "loss": 2.2606, "step": 1015 }, { "epoch": 0.31, "learning_rate": 4.8095703125e-05, "loss": 2.1063, "step": 1016 }, { "epoch": 0.31, "learning_rate": 4.809382512019231e-05, "loss": 2.0029, "step": 1017 }, { "epoch": 0.31, "learning_rate": 4.809194711538462e-05, "loss": 2.6476, "step": 1018 }, { "epoch": 0.31, "learning_rate": 4.809006911057692e-05, "loss": 2.2402, "step": 1019 }, { "epoch": 0.31, "learning_rate": 4.8088191105769235e-05, "loss": 2.1294, "step": 1020 }, { "epoch": 0.31, "learning_rate": 4.808631310096154e-05, "loss": 2.4804, "step": 1021 }, { "epoch": 0.31, "learning_rate": 4.8084435096153844e-05, "loss": 2.3341, "step": 1022 }, { "epoch": 0.31, "learning_rate": 4.8082557091346156e-05, "loss": 2.3725, "step": 1023 }, { "epoch": 0.31, "learning_rate": 4.808067908653847e-05, "loss": 2.1234, "step": 1024 }, { "epoch": 0.31, "learning_rate": 4.807880108173077e-05, "loss": 2.3283, "step": 1025 }, { "epoch": 0.31, "learning_rate": 4.8076923076923084e-05, "loss": 2.3691, "step": 1026 }, { "epoch": 0.31, "learning_rate": 4.807504507211539e-05, "loss": 2.105, "step": 1027 }, { "epoch": 0.31, "learning_rate": 4.807316706730769e-05, "loss": 2.1931, "step": 1028 }, { "epoch": 0.31, "learning_rate": 4.8071289062500005e-05, "loss": 2.2498, "step": 1029 }, { "epoch": 0.31, "learning_rate": 4.806941105769231e-05, "loss": 2.0992, "step": 1030 }, { "epoch": 0.31, "learning_rate": 4.8067533052884614e-05, "loss": 2.1542, "step": 1031 }, { "epoch": 0.31, "learning_rate": 4.8065655048076926e-05, "loss": 2.0425, "step": 1032 }, { "epoch": 0.31, "learning_rate": 4.806377704326923e-05, "loss": 2.3017, "step": 1033 }, { "epoch": 0.31, "learning_rate": 4.8061899038461535e-05, "loss": 2.3096, "step": 1034 }, { "epoch": 0.31, "learning_rate": 4.806002103365385e-05, "loss": 2.5378, "step": 1035 }, { "epoch": 0.31, "learning_rate": 4.805814302884616e-05, "loss": 2.3854, "step": 1036 }, { "epoch": 0.31, "learning_rate": 4.805626502403846e-05, "loss": 2.2404, "step": 1037 }, { "epoch": 0.31, "learning_rate": 4.8054387019230775e-05, "loss": 2.2538, "step": 1038 }, { "epoch": 0.31, "learning_rate": 4.805250901442308e-05, "loss": 2.0224, "step": 1039 }, { "epoch": 0.31, "learning_rate": 4.805063100961539e-05, "loss": 2.1908, "step": 1040 }, { "epoch": 0.31, "learning_rate": 4.8048753004807696e-05, "loss": 2.2328, "step": 1041 }, { "epoch": 0.31, "learning_rate": 4.8046875e-05, "loss": 2.3423, "step": 1042 }, { "epoch": 0.31, "learning_rate": 4.804499699519231e-05, "loss": 2.1507, "step": 1043 }, { "epoch": 0.31, "learning_rate": 4.804311899038462e-05, "loss": 2.0393, "step": 1044 }, { "epoch": 0.31, "learning_rate": 4.804124098557692e-05, "loss": 2.0947, "step": 1045 }, { "epoch": 0.31, "learning_rate": 4.803936298076923e-05, "loss": 2.1282, "step": 1046 }, { "epoch": 0.31, "learning_rate": 4.803748497596154e-05, "loss": 2.0144, "step": 1047 }, { "epoch": 0.31, "learning_rate": 4.803560697115384e-05, "loss": 2.0647, "step": 1048 }, { "epoch": 0.32, "learning_rate": 4.803372896634616e-05, "loss": 2.3546, "step": 1049 }, { "epoch": 0.32, "learning_rate": 4.8031850961538466e-05, "loss": 2.1832, "step": 1050 }, { "epoch": 0.32, "learning_rate": 4.802997295673077e-05, "loss": 2.2596, "step": 1051 }, { "epoch": 0.32, "learning_rate": 4.802809495192308e-05, "loss": 2.2384, "step": 1052 }, { "epoch": 0.32, "learning_rate": 4.802621694711539e-05, "loss": 2.4053, "step": 1053 }, { "epoch": 0.32, "learning_rate": 4.802433894230769e-05, "loss": 2.1446, "step": 1054 }, { "epoch": 0.32, "learning_rate": 4.80224609375e-05, "loss": 2.2378, "step": 1055 }, { "epoch": 0.32, "learning_rate": 4.802058293269231e-05, "loss": 2.0783, "step": 1056 }, { "epoch": 0.32, "learning_rate": 4.801870492788462e-05, "loss": 2.3327, "step": 1057 }, { "epoch": 0.32, "learning_rate": 4.8016826923076924e-05, "loss": 2.3983, "step": 1058 }, { "epoch": 0.32, "learning_rate": 4.801494891826923e-05, "loss": 2.665, "step": 1059 }, { "epoch": 0.32, "learning_rate": 4.801307091346154e-05, "loss": 2.1017, "step": 1060 }, { "epoch": 0.32, "learning_rate": 4.8011192908653846e-05, "loss": 2.3091, "step": 1061 }, { "epoch": 0.32, "learning_rate": 4.800931490384616e-05, "loss": 2.2274, "step": 1062 }, { "epoch": 0.32, "learning_rate": 4.800743689903847e-05, "loss": 2.2174, "step": 1063 }, { "epoch": 0.32, "learning_rate": 4.8005558894230773e-05, "loss": 2.214, "step": 1064 }, { "epoch": 0.32, "learning_rate": 4.800368088942308e-05, "loss": 2.3794, "step": 1065 }, { "epoch": 0.32, "learning_rate": 4.800180288461539e-05, "loss": 2.3011, "step": 1066 }, { "epoch": 0.32, "learning_rate": 4.7999924879807694e-05, "loss": 2.1445, "step": 1067 }, { "epoch": 0.32, "learning_rate": 4.7998046875e-05, "loss": 2.0689, "step": 1068 }, { "epoch": 0.32, "learning_rate": 4.799616887019231e-05, "loss": 2.343, "step": 1069 }, { "epoch": 0.32, "learning_rate": 4.7994290865384616e-05, "loss": 2.2732, "step": 1070 }, { "epoch": 0.32, "learning_rate": 4.799241286057693e-05, "loss": 2.1717, "step": 1071 }, { "epoch": 0.32, "learning_rate": 4.799053485576923e-05, "loss": 2.2696, "step": 1072 }, { "epoch": 0.32, "learning_rate": 4.798865685096154e-05, "loss": 2.2059, "step": 1073 }, { "epoch": 0.32, "learning_rate": 4.798677884615385e-05, "loss": 1.9544, "step": 1074 }, { "epoch": 0.32, "learning_rate": 4.798490084134616e-05, "loss": 2.5589, "step": 1075 }, { "epoch": 0.32, "learning_rate": 4.7983022836538465e-05, "loss": 2.0326, "step": 1076 }, { "epoch": 0.32, "learning_rate": 4.7981144831730776e-05, "loss": 2.3515, "step": 1077 }, { "epoch": 0.32, "learning_rate": 4.797926682692308e-05, "loss": 2.1722, "step": 1078 }, { "epoch": 0.32, "learning_rate": 4.7977388822115386e-05, "loss": 2.1601, "step": 1079 }, { "epoch": 0.32, "learning_rate": 4.79755108173077e-05, "loss": 2.2522, "step": 1080 }, { "epoch": 0.32, "learning_rate": 4.79736328125e-05, "loss": 2.2589, "step": 1081 }, { "epoch": 0.33, "learning_rate": 4.797175480769231e-05, "loss": 1.9774, "step": 1082 }, { "epoch": 0.33, "learning_rate": 4.796987680288462e-05, "loss": 2.4678, "step": 1083 }, { "epoch": 0.33, "learning_rate": 4.796799879807692e-05, "loss": 2.1868, "step": 1084 }, { "epoch": 0.33, "learning_rate": 4.7966120793269235e-05, "loss": 2.1982, "step": 1085 }, { "epoch": 0.33, "learning_rate": 4.796424278846154e-05, "loss": 2.5536, "step": 1086 }, { "epoch": 0.33, "learning_rate": 4.7962364783653844e-05, "loss": 2.1885, "step": 1087 }, { "epoch": 0.33, "learning_rate": 4.7960486778846156e-05, "loss": 2.2075, "step": 1088 }, { "epoch": 0.33, "learning_rate": 4.795860877403847e-05, "loss": 2.4599, "step": 1089 }, { "epoch": 0.33, "learning_rate": 4.795673076923077e-05, "loss": 2.0257, "step": 1090 }, { "epoch": 0.33, "learning_rate": 4.7954852764423084e-05, "loss": 1.8164, "step": 1091 }, { "epoch": 0.33, "learning_rate": 4.795297475961539e-05, "loss": 2.3234, "step": 1092 }, { "epoch": 0.33, "learning_rate": 4.795109675480769e-05, "loss": 2.285, "step": 1093 }, { "epoch": 0.33, "learning_rate": 4.7949218750000005e-05, "loss": 2.1792, "step": 1094 }, { "epoch": 0.33, "learning_rate": 4.794734074519231e-05, "loss": 2.0168, "step": 1095 }, { "epoch": 0.33, "learning_rate": 4.7945462740384614e-05, "loss": 2.2981, "step": 1096 }, { "epoch": 0.33, "learning_rate": 4.7943584735576926e-05, "loss": 2.2318, "step": 1097 }, { "epoch": 0.33, "learning_rate": 4.794170673076923e-05, "loss": 2.2021, "step": 1098 }, { "epoch": 0.33, "learning_rate": 4.7939828725961535e-05, "loss": 2.476, "step": 1099 }, { "epoch": 0.33, "learning_rate": 4.793795072115385e-05, "loss": 2.2669, "step": 1100 }, { "epoch": 0.33, "learning_rate": 4.793607271634616e-05, "loss": 2.2987, "step": 1101 }, { "epoch": 0.33, "learning_rate": 4.793419471153846e-05, "loss": 2.2394, "step": 1102 }, { "epoch": 0.33, "learning_rate": 4.7932316706730775e-05, "loss": 1.9501, "step": 1103 }, { "epoch": 0.33, "learning_rate": 4.793043870192308e-05, "loss": 2.3367, "step": 1104 }, { "epoch": 0.33, "learning_rate": 4.792856069711539e-05, "loss": 2.4978, "step": 1105 }, { "epoch": 0.33, "learning_rate": 4.7926682692307696e-05, "loss": 2.1286, "step": 1106 }, { "epoch": 0.33, "learning_rate": 4.79248046875e-05, "loss": 2.237, "step": 1107 }, { "epoch": 0.33, "learning_rate": 4.792292668269231e-05, "loss": 1.9653, "step": 1108 }, { "epoch": 0.33, "learning_rate": 4.792104867788462e-05, "loss": 2.1797, "step": 1109 }, { "epoch": 0.33, "learning_rate": 4.791917067307692e-05, "loss": 2.3738, "step": 1110 }, { "epoch": 0.33, "learning_rate": 4.791729266826923e-05, "loss": 2.2635, "step": 1111 }, { "epoch": 0.33, "learning_rate": 4.791541466346154e-05, "loss": 2.2768, "step": 1112 }, { "epoch": 0.33, "learning_rate": 4.791353665865384e-05, "loss": 2.7832, "step": 1113 }, { "epoch": 0.33, "learning_rate": 4.791165865384616e-05, "loss": 1.8705, "step": 1114 }, { "epoch": 0.34, "learning_rate": 4.7909780649038466e-05, "loss": 2.2907, "step": 1115 }, { "epoch": 0.34, "learning_rate": 4.790790264423077e-05, "loss": 2.0893, "step": 1116 }, { "epoch": 0.34, "learning_rate": 4.790602463942308e-05, "loss": 2.1875, "step": 1117 }, { "epoch": 0.34, "learning_rate": 4.790414663461539e-05, "loss": 2.4047, "step": 1118 }, { "epoch": 0.34, "learning_rate": 4.790226862980769e-05, "loss": 2.2324, "step": 1119 }, { "epoch": 0.34, "learning_rate": 4.7900390625e-05, "loss": 2.1421, "step": 1120 }, { "epoch": 0.34, "learning_rate": 4.789851262019231e-05, "loss": 2.1348, "step": 1121 }, { "epoch": 0.34, "learning_rate": 4.789663461538462e-05, "loss": 2.0813, "step": 1122 }, { "epoch": 0.34, "learning_rate": 4.7894756610576924e-05, "loss": 2.125, "step": 1123 }, { "epoch": 0.34, "learning_rate": 4.789287860576923e-05, "loss": 2.0804, "step": 1124 }, { "epoch": 0.34, "learning_rate": 4.789100060096154e-05, "loss": 2.2281, "step": 1125 }, { "epoch": 0.34, "learning_rate": 4.7889122596153845e-05, "loss": 2.2189, "step": 1126 }, { "epoch": 0.34, "learning_rate": 4.788724459134616e-05, "loss": 2.1502, "step": 1127 }, { "epoch": 0.34, "learning_rate": 4.788536658653847e-05, "loss": 2.0885, "step": 1128 }, { "epoch": 0.34, "learning_rate": 4.788348858173077e-05, "loss": 2.1818, "step": 1129 }, { "epoch": 0.34, "learning_rate": 4.788161057692308e-05, "loss": 2.0548, "step": 1130 }, { "epoch": 0.34, "learning_rate": 4.787973257211539e-05, "loss": 2.7118, "step": 1131 }, { "epoch": 0.34, "learning_rate": 4.7877854567307694e-05, "loss": 2.043, "step": 1132 }, { "epoch": 0.34, "learning_rate": 4.78759765625e-05, "loss": 1.9761, "step": 1133 }, { "epoch": 0.34, "learning_rate": 4.787409855769231e-05, "loss": 2.259, "step": 1134 }, { "epoch": 0.34, "learning_rate": 4.7872220552884615e-05, "loss": 2.3442, "step": 1135 }, { "epoch": 0.34, "learning_rate": 4.787034254807693e-05, "loss": 2.3275, "step": 1136 }, { "epoch": 0.34, "learning_rate": 4.786846454326923e-05, "loss": 2.3806, "step": 1137 }, { "epoch": 0.34, "learning_rate": 4.7866586538461537e-05, "loss": 2.4864, "step": 1138 }, { "epoch": 0.34, "learning_rate": 4.786470853365385e-05, "loss": 2.1166, "step": 1139 }, { "epoch": 0.34, "learning_rate": 4.786283052884616e-05, "loss": 2.5579, "step": 1140 }, { "epoch": 0.34, "learning_rate": 4.7860952524038464e-05, "loss": 2.3542, "step": 1141 }, { "epoch": 0.34, "learning_rate": 4.7859074519230776e-05, "loss": 2.37, "step": 1142 }, { "epoch": 0.34, "learning_rate": 4.785719651442308e-05, "loss": 2.1063, "step": 1143 }, { "epoch": 0.34, "learning_rate": 4.7855318509615386e-05, "loss": 2.2035, "step": 1144 }, { "epoch": 0.34, "learning_rate": 4.78534405048077e-05, "loss": 2.3706, "step": 1145 }, { "epoch": 0.34, "learning_rate": 4.78515625e-05, "loss": 2.315, "step": 1146 }, { "epoch": 0.34, "learning_rate": 4.784968449519231e-05, "loss": 1.9803, "step": 1147 }, { "epoch": 0.34, "learning_rate": 4.784780649038462e-05, "loss": 1.9803, "step": 1148 }, { "epoch": 0.35, "learning_rate": 4.784592848557692e-05, "loss": 2.2797, "step": 1149 }, { "epoch": 0.35, "learning_rate": 4.7844050480769234e-05, "loss": 2.4011, "step": 1150 }, { "epoch": 0.35, "learning_rate": 4.784217247596154e-05, "loss": 2.6106, "step": 1151 }, { "epoch": 0.35, "learning_rate": 4.7840294471153844e-05, "loss": 2.2048, "step": 1152 }, { "epoch": 0.35, "learning_rate": 4.7838416466346156e-05, "loss": 2.3294, "step": 1153 }, { "epoch": 0.35, "learning_rate": 4.783653846153847e-05, "loss": 2.1909, "step": 1154 }, { "epoch": 0.35, "learning_rate": 4.783466045673077e-05, "loss": 2.0989, "step": 1155 }, { "epoch": 0.35, "learning_rate": 4.7832782451923083e-05, "loss": 2.2302, "step": 1156 }, { "epoch": 0.35, "learning_rate": 4.783090444711539e-05, "loss": 2.2745, "step": 1157 }, { "epoch": 0.35, "learning_rate": 4.782902644230769e-05, "loss": 2.1122, "step": 1158 }, { "epoch": 0.35, "learning_rate": 4.7827148437500005e-05, "loss": 2.7454, "step": 1159 }, { "epoch": 0.35, "learning_rate": 4.782527043269231e-05, "loss": 2.4314, "step": 1160 }, { "epoch": 0.35, "learning_rate": 4.7823392427884614e-05, "loss": 2.1534, "step": 1161 }, { "epoch": 0.35, "learning_rate": 4.7821514423076926e-05, "loss": 2.0932, "step": 1162 }, { "epoch": 0.35, "learning_rate": 4.781963641826923e-05, "loss": 2.2523, "step": 1163 }, { "epoch": 0.35, "learning_rate": 4.781775841346154e-05, "loss": 2.125, "step": 1164 }, { "epoch": 0.35, "learning_rate": 4.781588040865385e-05, "loss": 1.9611, "step": 1165 }, { "epoch": 0.35, "learning_rate": 4.781400240384616e-05, "loss": 2.2876, "step": 1166 }, { "epoch": 0.35, "learning_rate": 4.781212439903846e-05, "loss": 2.1269, "step": 1167 }, { "epoch": 0.35, "learning_rate": 4.7810246394230775e-05, "loss": 2.2044, "step": 1168 }, { "epoch": 0.35, "learning_rate": 4.780836838942308e-05, "loss": 2.1621, "step": 1169 }, { "epoch": 0.35, "learning_rate": 4.780649038461539e-05, "loss": 2.3796, "step": 1170 }, { "epoch": 0.35, "learning_rate": 4.7804612379807696e-05, "loss": 2.3937, "step": 1171 }, { "epoch": 0.35, "learning_rate": 4.7802734375e-05, "loss": 2.1281, "step": 1172 }, { "epoch": 0.35, "learning_rate": 4.780085637019231e-05, "loss": 2.0415, "step": 1173 }, { "epoch": 0.35, "learning_rate": 4.779897836538462e-05, "loss": 2.1304, "step": 1174 }, { "epoch": 0.35, "learning_rate": 4.779710036057692e-05, "loss": 1.964, "step": 1175 }, { "epoch": 0.35, "learning_rate": 4.779522235576923e-05, "loss": 2.3379, "step": 1176 }, { "epoch": 0.35, "learning_rate": 4.779334435096154e-05, "loss": 2.2265, "step": 1177 }, { "epoch": 0.35, "learning_rate": 4.779146634615384e-05, "loss": 2.3921, "step": 1178 }, { "epoch": 0.35, "learning_rate": 4.778958834134616e-05, "loss": 2.4342, "step": 1179 }, { "epoch": 0.35, "learning_rate": 4.7787710336538466e-05, "loss": 2.371, "step": 1180 }, { "epoch": 0.35, "learning_rate": 4.778583233173077e-05, "loss": 2.2007, "step": 1181 }, { "epoch": 0.36, "learning_rate": 4.778395432692308e-05, "loss": 2.2442, "step": 1182 }, { "epoch": 0.36, "learning_rate": 4.778207632211539e-05, "loss": 2.4464, "step": 1183 }, { "epoch": 0.36, "learning_rate": 4.77801983173077e-05, "loss": 2.0528, "step": 1184 }, { "epoch": 0.36, "learning_rate": 4.77783203125e-05, "loss": 2.3086, "step": 1185 }, { "epoch": 0.36, "learning_rate": 4.777644230769231e-05, "loss": 2.1962, "step": 1186 }, { "epoch": 0.36, "learning_rate": 4.777456430288462e-05, "loss": 2.035, "step": 1187 }, { "epoch": 0.36, "learning_rate": 4.7772686298076924e-05, "loss": 2.4807, "step": 1188 }, { "epoch": 0.36, "learning_rate": 4.777080829326923e-05, "loss": 2.2109, "step": 1189 }, { "epoch": 0.36, "learning_rate": 4.776893028846154e-05, "loss": 2.1828, "step": 1190 }, { "epoch": 0.36, "learning_rate": 4.7767052283653845e-05, "loss": 2.2904, "step": 1191 }, { "epoch": 0.36, "learning_rate": 4.776517427884616e-05, "loss": 2.2391, "step": 1192 }, { "epoch": 0.36, "learning_rate": 4.776329627403847e-05, "loss": 2.1991, "step": 1193 }, { "epoch": 0.36, "learning_rate": 4.776141826923077e-05, "loss": 2.3543, "step": 1194 }, { "epoch": 0.36, "learning_rate": 4.775954026442308e-05, "loss": 2.2829, "step": 1195 }, { "epoch": 0.36, "learning_rate": 4.775766225961539e-05, "loss": 2.304, "step": 1196 }, { "epoch": 0.36, "learning_rate": 4.7755784254807694e-05, "loss": 2.0132, "step": 1197 }, { "epoch": 0.36, "learning_rate": 4.775390625e-05, "loss": 2.0837, "step": 1198 }, { "epoch": 0.36, "learning_rate": 4.775202824519231e-05, "loss": 2.3275, "step": 1199 }, { "epoch": 0.36, "learning_rate": 4.7750150240384615e-05, "loss": 2.362, "step": 1200 }, { "epoch": 0.36, "learning_rate": 4.774827223557693e-05, "loss": 2.3226, "step": 1201 }, { "epoch": 0.36, "learning_rate": 4.774639423076923e-05, "loss": 2.3154, "step": 1202 }, { "epoch": 0.36, "learning_rate": 4.7744516225961536e-05, "loss": 2.2258, "step": 1203 }, { "epoch": 0.36, "learning_rate": 4.774263822115385e-05, "loss": 1.9638, "step": 1204 }, { "epoch": 0.36, "learning_rate": 4.774076021634616e-05, "loss": 2.5832, "step": 1205 }, { "epoch": 0.36, "learning_rate": 4.7738882211538464e-05, "loss": 2.1374, "step": 1206 }, { "epoch": 0.36, "learning_rate": 4.7737004206730776e-05, "loss": 2.328, "step": 1207 }, { "epoch": 0.36, "learning_rate": 4.773512620192308e-05, "loss": 2.2996, "step": 1208 }, { "epoch": 0.36, "learning_rate": 4.7733248197115385e-05, "loss": 2.3065, "step": 1209 }, { "epoch": 0.36, "learning_rate": 4.77313701923077e-05, "loss": 2.2253, "step": 1210 }, { "epoch": 0.36, "learning_rate": 4.77294921875e-05, "loss": 2.3821, "step": 1211 }, { "epoch": 0.36, "learning_rate": 4.7727614182692307e-05, "loss": 2.3582, "step": 1212 }, { "epoch": 0.36, "learning_rate": 4.772573617788462e-05, "loss": 2.2405, "step": 1213 }, { "epoch": 0.36, "learning_rate": 4.772385817307692e-05, "loss": 2.1475, "step": 1214 }, { "epoch": 0.37, "learning_rate": 4.7721980168269234e-05, "loss": 2.1011, "step": 1215 }, { "epoch": 0.37, "learning_rate": 4.772010216346154e-05, "loss": 2.4403, "step": 1216 }, { "epoch": 0.37, "learning_rate": 4.7718224158653844e-05, "loss": 2.3259, "step": 1217 }, { "epoch": 0.37, "learning_rate": 4.7716346153846155e-05, "loss": 2.1003, "step": 1218 }, { "epoch": 0.37, "learning_rate": 4.771446814903847e-05, "loss": 2.0528, "step": 1219 }, { "epoch": 0.37, "learning_rate": 4.771259014423077e-05, "loss": 2.3069, "step": 1220 }, { "epoch": 0.37, "learning_rate": 4.771071213942308e-05, "loss": 2.0052, "step": 1221 }, { "epoch": 0.37, "learning_rate": 4.770883413461539e-05, "loss": 2.5799, "step": 1222 }, { "epoch": 0.37, "learning_rate": 4.770695612980769e-05, "loss": 2.062, "step": 1223 }, { "epoch": 0.37, "learning_rate": 4.7705078125000004e-05, "loss": 2.3959, "step": 1224 }, { "epoch": 0.37, "learning_rate": 4.770320012019231e-05, "loss": 2.0899, "step": 1225 }, { "epoch": 0.37, "learning_rate": 4.7701322115384614e-05, "loss": 2.0655, "step": 1226 }, { "epoch": 0.37, "learning_rate": 4.7699444110576926e-05, "loss": 2.1519, "step": 1227 }, { "epoch": 0.37, "learning_rate": 4.769756610576923e-05, "loss": 2.0103, "step": 1228 }, { "epoch": 0.37, "learning_rate": 4.769568810096154e-05, "loss": 2.0097, "step": 1229 }, { "epoch": 0.37, "learning_rate": 4.7693810096153847e-05, "loss": 2.213, "step": 1230 }, { "epoch": 0.37, "learning_rate": 4.769193209134616e-05, "loss": 2.5435, "step": 1231 }, { "epoch": 0.37, "learning_rate": 4.769005408653846e-05, "loss": 2.193, "step": 1232 }, { "epoch": 0.37, "learning_rate": 4.7688176081730774e-05, "loss": 2.3743, "step": 1233 }, { "epoch": 0.37, "learning_rate": 4.768629807692308e-05, "loss": 2.1981, "step": 1234 }, { "epoch": 0.37, "learning_rate": 4.768442007211539e-05, "loss": 2.0903, "step": 1235 }, { "epoch": 0.37, "learning_rate": 4.7682542067307696e-05, "loss": 2.3301, "step": 1236 }, { "epoch": 0.37, "learning_rate": 4.76806640625e-05, "loss": 2.0387, "step": 1237 }, { "epoch": 0.37, "learning_rate": 4.767878605769231e-05, "loss": 2.2621, "step": 1238 }, { "epoch": 0.37, "learning_rate": 4.767690805288462e-05, "loss": 2.0379, "step": 1239 }, { "epoch": 0.37, "learning_rate": 4.767503004807692e-05, "loss": 2.1746, "step": 1240 }, { "epoch": 0.37, "learning_rate": 4.767315204326923e-05, "loss": 1.9945, "step": 1241 }, { "epoch": 0.37, "learning_rate": 4.767127403846154e-05, "loss": 2.1474, "step": 1242 }, { "epoch": 0.37, "learning_rate": 4.766939603365384e-05, "loss": 2.0966, "step": 1243 }, { "epoch": 0.37, "learning_rate": 4.766751802884616e-05, "loss": 2.1823, "step": 1244 }, { "epoch": 0.37, "learning_rate": 4.7665640024038466e-05, "loss": 2.3945, "step": 1245 }, { "epoch": 0.37, "learning_rate": 4.766376201923077e-05, "loss": 2.2588, "step": 1246 }, { "epoch": 0.37, "learning_rate": 4.766188401442308e-05, "loss": 2.1406, "step": 1247 }, { "epoch": 0.38, "learning_rate": 4.766000600961539e-05, "loss": 2.122, "step": 1248 }, { "epoch": 0.38, "learning_rate": 4.76581280048077e-05, "loss": 2.297, "step": 1249 }, { "epoch": 0.38, "learning_rate": 4.765625e-05, "loss": 2.0379, "step": 1250 }, { "epoch": 0.38, "learning_rate": 4.765437199519231e-05, "loss": 2.2862, "step": 1251 }, { "epoch": 0.38, "learning_rate": 4.765249399038462e-05, "loss": 2.1826, "step": 1252 }, { "epoch": 0.38, "learning_rate": 4.7650615985576924e-05, "loss": 2.2026, "step": 1253 }, { "epoch": 0.38, "learning_rate": 4.764873798076923e-05, "loss": 2.2076, "step": 1254 }, { "epoch": 0.38, "learning_rate": 4.764685997596154e-05, "loss": 2.3909, "step": 1255 }, { "epoch": 0.38, "learning_rate": 4.7644981971153845e-05, "loss": 2.2778, "step": 1256 }, { "epoch": 0.38, "learning_rate": 4.764310396634616e-05, "loss": 2.2287, "step": 1257 }, { "epoch": 0.38, "learning_rate": 4.764122596153847e-05, "loss": 2.3326, "step": 1258 }, { "epoch": 0.38, "learning_rate": 4.763934795673077e-05, "loss": 2.5842, "step": 1259 }, { "epoch": 0.38, "learning_rate": 4.763746995192308e-05, "loss": 2.635, "step": 1260 }, { "epoch": 0.38, "learning_rate": 4.763559194711539e-05, "loss": 2.1607, "step": 1261 }, { "epoch": 0.38, "learning_rate": 4.7633713942307694e-05, "loss": 2.2596, "step": 1262 }, { "epoch": 0.38, "learning_rate": 4.76318359375e-05, "loss": 2.0878, "step": 1263 }, { "epoch": 0.38, "learning_rate": 4.762995793269231e-05, "loss": 2.3433, "step": 1264 }, { "epoch": 0.38, "learning_rate": 4.7628079927884615e-05, "loss": 2.2255, "step": 1265 }, { "epoch": 0.38, "learning_rate": 4.762620192307693e-05, "loss": 2.3695, "step": 1266 }, { "epoch": 0.38, "learning_rate": 4.762432391826923e-05, "loss": 2.3238, "step": 1267 }, { "epoch": 0.38, "learning_rate": 4.7622445913461536e-05, "loss": 2.5318, "step": 1268 }, { "epoch": 0.38, "learning_rate": 4.762056790865385e-05, "loss": 2.0878, "step": 1269 }, { "epoch": 0.38, "learning_rate": 4.761868990384616e-05, "loss": 2.209, "step": 1270 }, { "epoch": 0.38, "learning_rate": 4.7616811899038464e-05, "loss": 2.2829, "step": 1271 }, { "epoch": 0.38, "learning_rate": 4.7614933894230776e-05, "loss": 2.3965, "step": 1272 }, { "epoch": 0.38, "learning_rate": 4.761305588942308e-05, "loss": 2.363, "step": 1273 }, { "epoch": 0.38, "learning_rate": 4.7611177884615385e-05, "loss": 2.4438, "step": 1274 }, { "epoch": 0.38, "learning_rate": 4.76092998798077e-05, "loss": 2.2273, "step": 1275 }, { "epoch": 0.38, "learning_rate": 4.7607421875e-05, "loss": 2.1763, "step": 1276 }, { "epoch": 0.38, "learning_rate": 4.7605543870192306e-05, "loss": 2.2035, "step": 1277 }, { "epoch": 0.38, "learning_rate": 4.760366586538462e-05, "loss": 2.178, "step": 1278 }, { "epoch": 0.38, "learning_rate": 4.760178786057692e-05, "loss": 2.5182, "step": 1279 }, { "epoch": 0.38, "learning_rate": 4.7599909855769234e-05, "loss": 2.3253, "step": 1280 }, { "epoch": 0.38, "learning_rate": 4.759803185096154e-05, "loss": 2.2708, "step": 1281 }, { "epoch": 0.39, "learning_rate": 4.7596153846153844e-05, "loss": 2.3851, "step": 1282 }, { "epoch": 0.39, "learning_rate": 4.7594275841346155e-05, "loss": 2.0667, "step": 1283 }, { "epoch": 0.39, "learning_rate": 4.759239783653847e-05, "loss": 2.0912, "step": 1284 }, { "epoch": 0.39, "learning_rate": 4.759051983173077e-05, "loss": 2.3846, "step": 1285 }, { "epoch": 0.39, "learning_rate": 4.758864182692308e-05, "loss": 2.0537, "step": 1286 }, { "epoch": 0.39, "learning_rate": 4.758676382211539e-05, "loss": 2.3026, "step": 1287 }, { "epoch": 0.39, "learning_rate": 4.758488581730769e-05, "loss": 2.3348, "step": 1288 }, { "epoch": 0.39, "learning_rate": 4.7583007812500004e-05, "loss": 2.2077, "step": 1289 }, { "epoch": 0.39, "learning_rate": 4.758112980769231e-05, "loss": 2.2773, "step": 1290 }, { "epoch": 0.39, "learning_rate": 4.7579251802884614e-05, "loss": 2.2908, "step": 1291 }, { "epoch": 0.39, "learning_rate": 4.7577373798076925e-05, "loss": 2.3524, "step": 1292 }, { "epoch": 0.39, "learning_rate": 4.757549579326923e-05, "loss": 2.0815, "step": 1293 }, { "epoch": 0.39, "learning_rate": 4.757361778846154e-05, "loss": 1.9512, "step": 1294 }, { "epoch": 0.39, "learning_rate": 4.7571739783653846e-05, "loss": 2.2058, "step": 1295 }, { "epoch": 0.39, "learning_rate": 4.756986177884616e-05, "loss": 2.2029, "step": 1296 }, { "epoch": 0.39, "learning_rate": 4.756798377403846e-05, "loss": 2.3892, "step": 1297 }, { "epoch": 0.39, "learning_rate": 4.7566105769230774e-05, "loss": 2.3451, "step": 1298 }, { "epoch": 0.39, "learning_rate": 4.756422776442308e-05, "loss": 2.3181, "step": 1299 }, { "epoch": 0.39, "learning_rate": 4.756234975961539e-05, "loss": 2.4074, "step": 1300 }, { "epoch": 0.39, "learning_rate": 4.7560471754807695e-05, "loss": 2.205, "step": 1301 }, { "epoch": 0.39, "learning_rate": 4.755859375e-05, "loss": 2.119, "step": 1302 }, { "epoch": 0.39, "learning_rate": 4.755671574519231e-05, "loss": 2.202, "step": 1303 }, { "epoch": 0.39, "learning_rate": 4.7554837740384617e-05, "loss": 2.2755, "step": 1304 }, { "epoch": 0.39, "learning_rate": 4.755295973557692e-05, "loss": 2.2279, "step": 1305 }, { "epoch": 0.39, "learning_rate": 4.755108173076923e-05, "loss": 2.3613, "step": 1306 }, { "epoch": 0.39, "learning_rate": 4.754920372596154e-05, "loss": 1.9422, "step": 1307 }, { "epoch": 0.39, "learning_rate": 4.754732572115385e-05, "loss": 2.2546, "step": 1308 }, { "epoch": 0.39, "learning_rate": 4.754544771634616e-05, "loss": 1.8943, "step": 1309 }, { "epoch": 0.39, "learning_rate": 4.7543569711538466e-05, "loss": 2.1036, "step": 1310 }, { "epoch": 0.39, "learning_rate": 4.754169170673077e-05, "loss": 2.1716, "step": 1311 }, { "epoch": 0.39, "learning_rate": 4.753981370192308e-05, "loss": 2.553, "step": 1312 }, { "epoch": 0.39, "learning_rate": 4.7537935697115387e-05, "loss": 2.3246, "step": 1313 }, { "epoch": 0.39, "learning_rate": 4.75360576923077e-05, "loss": 2.2917, "step": 1314 }, { "epoch": 0.4, "learning_rate": 4.75341796875e-05, "loss": 2.2202, "step": 1315 }, { "epoch": 0.4, "learning_rate": 4.753230168269231e-05, "loss": 2.3913, "step": 1316 }, { "epoch": 0.4, "learning_rate": 4.753042367788462e-05, "loss": 2.3359, "step": 1317 }, { "epoch": 0.4, "learning_rate": 4.7528545673076924e-05, "loss": 2.1354, "step": 1318 }, { "epoch": 0.4, "learning_rate": 4.752666766826923e-05, "loss": 2.4237, "step": 1319 }, { "epoch": 0.4, "learning_rate": 4.752478966346154e-05, "loss": 2.3592, "step": 1320 }, { "epoch": 0.4, "learning_rate": 4.7522911658653845e-05, "loss": 2.0933, "step": 1321 }, { "epoch": 0.4, "learning_rate": 4.752103365384616e-05, "loss": 2.0947, "step": 1322 }, { "epoch": 0.4, "learning_rate": 4.751915564903847e-05, "loss": 2.5059, "step": 1323 }, { "epoch": 0.4, "learning_rate": 4.751727764423077e-05, "loss": 2.2641, "step": 1324 }, { "epoch": 0.4, "learning_rate": 4.751539963942308e-05, "loss": 2.2594, "step": 1325 }, { "epoch": 0.4, "learning_rate": 4.751352163461539e-05, "loss": 2.2215, "step": 1326 }, { "epoch": 0.4, "learning_rate": 4.7511643629807694e-05, "loss": 2.1986, "step": 1327 }, { "epoch": 0.4, "learning_rate": 4.7509765625000006e-05, "loss": 2.2064, "step": 1328 }, { "epoch": 0.4, "learning_rate": 4.750788762019231e-05, "loss": 2.477, "step": 1329 }, { "epoch": 0.4, "learning_rate": 4.7506009615384615e-05, "loss": 2.0059, "step": 1330 }, { "epoch": 0.4, "learning_rate": 4.750413161057693e-05, "loss": 2.2777, "step": 1331 }, { "epoch": 0.4, "learning_rate": 4.750225360576923e-05, "loss": 2.274, "step": 1332 }, { "epoch": 0.4, "learning_rate": 4.7500375600961536e-05, "loss": 2.4623, "step": 1333 }, { "epoch": 0.4, "learning_rate": 4.749849759615385e-05, "loss": 2.2818, "step": 1334 }, { "epoch": 0.4, "learning_rate": 4.749661959134616e-05, "loss": 2.3329, "step": 1335 }, { "epoch": 0.4, "learning_rate": 4.7494741586538464e-05, "loss": 2.672, "step": 1336 }, { "epoch": 0.4, "learning_rate": 4.7492863581730776e-05, "loss": 2.0559, "step": 1337 }, { "epoch": 0.4, "learning_rate": 4.749098557692308e-05, "loss": 2.2628, "step": 1338 }, { "epoch": 0.4, "learning_rate": 4.7489107572115385e-05, "loss": 2.4546, "step": 1339 }, { "epoch": 0.4, "learning_rate": 4.74872295673077e-05, "loss": 1.9512, "step": 1340 }, { "epoch": 0.4, "learning_rate": 4.74853515625e-05, "loss": 2.1504, "step": 1341 }, { "epoch": 0.4, "learning_rate": 4.7483473557692306e-05, "loss": 2.118, "step": 1342 }, { "epoch": 0.4, "learning_rate": 4.748159555288462e-05, "loss": 2.2563, "step": 1343 }, { "epoch": 0.4, "learning_rate": 4.747971754807692e-05, "loss": 2.3474, "step": 1344 }, { "epoch": 0.4, "learning_rate": 4.7477839543269234e-05, "loss": 1.8929, "step": 1345 }, { "epoch": 0.4, "learning_rate": 4.747596153846154e-05, "loss": 2.3362, "step": 1346 }, { "epoch": 0.4, "learning_rate": 4.7474083533653844e-05, "loss": 2.0588, "step": 1347 }, { "epoch": 0.41, "learning_rate": 4.747220552884616e-05, "loss": 2.0303, "step": 1348 }, { "epoch": 0.41, "learning_rate": 4.747032752403847e-05, "loss": 1.9901, "step": 1349 }, { "epoch": 0.41, "learning_rate": 4.746844951923077e-05, "loss": 2.2957, "step": 1350 }, { "epoch": 0.41, "learning_rate": 4.746657151442308e-05, "loss": 2.2939, "step": 1351 }, { "epoch": 0.41, "learning_rate": 4.746469350961539e-05, "loss": 2.2853, "step": 1352 }, { "epoch": 0.41, "learning_rate": 4.746281550480769e-05, "loss": 2.3081, "step": 1353 }, { "epoch": 0.41, "learning_rate": 4.7460937500000004e-05, "loss": 2.184, "step": 1354 }, { "epoch": 0.41, "learning_rate": 4.745905949519231e-05, "loss": 2.0813, "step": 1355 }, { "epoch": 0.41, "learning_rate": 4.7457181490384614e-05, "loss": 1.8604, "step": 1356 }, { "epoch": 0.41, "learning_rate": 4.7455303485576925e-05, "loss": 2.3581, "step": 1357 }, { "epoch": 0.41, "learning_rate": 4.745342548076923e-05, "loss": 2.0301, "step": 1358 }, { "epoch": 0.41, "learning_rate": 4.745154747596154e-05, "loss": 2.2771, "step": 1359 }, { "epoch": 0.41, "learning_rate": 4.7449669471153846e-05, "loss": 2.3417, "step": 1360 }, { "epoch": 0.41, "learning_rate": 4.744779146634616e-05, "loss": 2.2529, "step": 1361 }, { "epoch": 0.41, "learning_rate": 4.744591346153846e-05, "loss": 2.3072, "step": 1362 }, { "epoch": 0.41, "learning_rate": 4.7444035456730774e-05, "loss": 2.3119, "step": 1363 }, { "epoch": 0.41, "learning_rate": 4.744215745192308e-05, "loss": 1.908, "step": 1364 }, { "epoch": 0.41, "learning_rate": 4.744027944711539e-05, "loss": 2.0141, "step": 1365 }, { "epoch": 0.41, "learning_rate": 4.7438401442307695e-05, "loss": 2.6037, "step": 1366 }, { "epoch": 0.41, "learning_rate": 4.74365234375e-05, "loss": 2.0712, "step": 1367 }, { "epoch": 0.41, "learning_rate": 4.743464543269231e-05, "loss": 2.1877, "step": 1368 }, { "epoch": 0.41, "learning_rate": 4.7432767427884616e-05, "loss": 2.4031, "step": 1369 }, { "epoch": 0.41, "learning_rate": 4.743088942307692e-05, "loss": 2.094, "step": 1370 }, { "epoch": 0.41, "learning_rate": 4.742901141826923e-05, "loss": 2.2342, "step": 1371 }, { "epoch": 0.41, "learning_rate": 4.742713341346154e-05, "loss": 1.9231, "step": 1372 }, { "epoch": 0.41, "learning_rate": 4.742525540865385e-05, "loss": 2.4859, "step": 1373 }, { "epoch": 0.41, "learning_rate": 4.742337740384616e-05, "loss": 2.3505, "step": 1374 }, { "epoch": 0.41, "learning_rate": 4.7421499399038465e-05, "loss": 2.1862, "step": 1375 }, { "epoch": 0.41, "learning_rate": 4.741962139423077e-05, "loss": 2.4013, "step": 1376 }, { "epoch": 0.41, "learning_rate": 4.741774338942308e-05, "loss": 2.2234, "step": 1377 }, { "epoch": 0.41, "learning_rate": 4.7415865384615386e-05, "loss": 2.1006, "step": 1378 }, { "epoch": 0.41, "learning_rate": 4.74139873798077e-05, "loss": 2.5135, "step": 1379 }, { "epoch": 0.41, "learning_rate": 4.7412109375e-05, "loss": 2.5369, "step": 1380 }, { "epoch": 0.41, "learning_rate": 4.741023137019231e-05, "loss": 2.2061, "step": 1381 }, { "epoch": 0.42, "learning_rate": 4.740835336538462e-05, "loss": 2.2681, "step": 1382 }, { "epoch": 0.42, "learning_rate": 4.7406475360576924e-05, "loss": 2.4005, "step": 1383 }, { "epoch": 0.42, "learning_rate": 4.740459735576923e-05, "loss": 2.1278, "step": 1384 }, { "epoch": 0.42, "learning_rate": 4.740271935096154e-05, "loss": 2.5396, "step": 1385 }, { "epoch": 0.42, "learning_rate": 4.7400841346153845e-05, "loss": 2.3337, "step": 1386 }, { "epoch": 0.42, "learning_rate": 4.7398963341346157e-05, "loss": 2.2271, "step": 1387 }, { "epoch": 0.42, "learning_rate": 4.739708533653847e-05, "loss": 2.3615, "step": 1388 }, { "epoch": 0.42, "learning_rate": 4.739520733173077e-05, "loss": 2.1838, "step": 1389 }, { "epoch": 0.42, "learning_rate": 4.739332932692308e-05, "loss": 2.4424, "step": 1390 }, { "epoch": 0.42, "learning_rate": 4.739145132211539e-05, "loss": 2.0177, "step": 1391 }, { "epoch": 0.42, "learning_rate": 4.7389573317307694e-05, "loss": 2.2981, "step": 1392 }, { "epoch": 0.42, "learning_rate": 4.7387695312500005e-05, "loss": 2.2105, "step": 1393 }, { "epoch": 0.42, "learning_rate": 4.738581730769231e-05, "loss": 2.1251, "step": 1394 }, { "epoch": 0.42, "learning_rate": 4.7383939302884615e-05, "loss": 2.1917, "step": 1395 }, { "epoch": 0.42, "learning_rate": 4.7382061298076927e-05, "loss": 2.0717, "step": 1396 }, { "epoch": 0.42, "learning_rate": 4.738018329326923e-05, "loss": 2.263, "step": 1397 }, { "epoch": 0.42, "learning_rate": 4.7378305288461536e-05, "loss": 2.2302, "step": 1398 }, { "epoch": 0.42, "learning_rate": 4.737642728365385e-05, "loss": 2.3253, "step": 1399 }, { "epoch": 0.42, "learning_rate": 4.737454927884616e-05, "loss": 2.4362, "step": 1400 }, { "epoch": 0.42, "learning_rate": 4.7372671274038464e-05, "loss": 2.0546, "step": 1401 }, { "epoch": 0.42, "learning_rate": 4.7370793269230776e-05, "loss": 2.2398, "step": 1402 }, { "epoch": 0.42, "learning_rate": 4.736891526442308e-05, "loss": 2.4164, "step": 1403 }, { "epoch": 0.42, "learning_rate": 4.7367037259615385e-05, "loss": 2.4172, "step": 1404 }, { "epoch": 0.42, "learning_rate": 4.73651592548077e-05, "loss": 2.1726, "step": 1405 }, { "epoch": 0.42, "learning_rate": 4.736328125e-05, "loss": 2.2635, "step": 1406 }, { "epoch": 0.42, "learning_rate": 4.7361403245192306e-05, "loss": 1.6974, "step": 1407 }, { "epoch": 0.42, "learning_rate": 4.735952524038462e-05, "loss": 1.9924, "step": 1408 }, { "epoch": 0.42, "learning_rate": 4.735764723557692e-05, "loss": 2.4253, "step": 1409 }, { "epoch": 0.42, "learning_rate": 4.7355769230769234e-05, "loss": 2.1491, "step": 1410 }, { "epoch": 0.42, "learning_rate": 4.735389122596154e-05, "loss": 2.2908, "step": 1411 }, { "epoch": 0.42, "learning_rate": 4.7352013221153844e-05, "loss": 2.3622, "step": 1412 }, { "epoch": 0.42, "learning_rate": 4.735013521634616e-05, "loss": 2.3482, "step": 1413 }, { "epoch": 0.42, "learning_rate": 4.734825721153847e-05, "loss": 2.173, "step": 1414 }, { "epoch": 0.43, "learning_rate": 4.734637920673077e-05, "loss": 2.4279, "step": 1415 }, { "epoch": 0.43, "learning_rate": 4.734450120192308e-05, "loss": 2.0693, "step": 1416 }, { "epoch": 0.43, "learning_rate": 4.734262319711539e-05, "loss": 2.1857, "step": 1417 }, { "epoch": 0.43, "learning_rate": 4.734074519230769e-05, "loss": 2.0168, "step": 1418 }, { "epoch": 0.43, "learning_rate": 4.7338867187500004e-05, "loss": 2.2008, "step": 1419 }, { "epoch": 0.43, "learning_rate": 4.733698918269231e-05, "loss": 1.9012, "step": 1420 }, { "epoch": 0.43, "learning_rate": 4.7335111177884614e-05, "loss": 2.8745, "step": 1421 }, { "epoch": 0.43, "learning_rate": 4.7333233173076925e-05, "loss": 2.3922, "step": 1422 }, { "epoch": 0.43, "learning_rate": 4.733135516826923e-05, "loss": 2.0749, "step": 1423 }, { "epoch": 0.43, "learning_rate": 4.732947716346154e-05, "loss": 2.2726, "step": 1424 }, { "epoch": 0.43, "learning_rate": 4.7327599158653846e-05, "loss": 2.2912, "step": 1425 }, { "epoch": 0.43, "learning_rate": 4.732572115384616e-05, "loss": 2.1459, "step": 1426 }, { "epoch": 0.43, "learning_rate": 4.732384314903846e-05, "loss": 2.1081, "step": 1427 }, { "epoch": 0.43, "learning_rate": 4.7321965144230774e-05, "loss": 2.298, "step": 1428 }, { "epoch": 0.43, "learning_rate": 4.732008713942308e-05, "loss": 2.2182, "step": 1429 }, { "epoch": 0.43, "learning_rate": 4.731820913461539e-05, "loss": 2.2038, "step": 1430 }, { "epoch": 0.43, "learning_rate": 4.7316331129807695e-05, "loss": 1.9905, "step": 1431 }, { "epoch": 0.43, "learning_rate": 4.7314453125e-05, "loss": 2.2631, "step": 1432 }, { "epoch": 0.43, "learning_rate": 4.731257512019231e-05, "loss": 2.0835, "step": 1433 }, { "epoch": 0.43, "learning_rate": 4.7310697115384616e-05, "loss": 2.1334, "step": 1434 }, { "epoch": 0.43, "learning_rate": 4.730881911057692e-05, "loss": 2.1743, "step": 1435 }, { "epoch": 0.43, "learning_rate": 4.730694110576923e-05, "loss": 2.305, "step": 1436 }, { "epoch": 0.43, "learning_rate": 4.730506310096154e-05, "loss": 2.5442, "step": 1437 }, { "epoch": 0.43, "learning_rate": 4.730318509615385e-05, "loss": 2.4368, "step": 1438 }, { "epoch": 0.43, "learning_rate": 4.730130709134616e-05, "loss": 2.1842, "step": 1439 }, { "epoch": 0.43, "learning_rate": 4.7299429086538465e-05, "loss": 2.1982, "step": 1440 }, { "epoch": 0.43, "learning_rate": 4.729755108173077e-05, "loss": 2.2668, "step": 1441 }, { "epoch": 0.43, "learning_rate": 4.729567307692308e-05, "loss": 2.1173, "step": 1442 }, { "epoch": 0.43, "learning_rate": 4.7293795072115386e-05, "loss": 2.238, "step": 1443 }, { "epoch": 0.43, "learning_rate": 4.72919170673077e-05, "loss": 1.9581, "step": 1444 }, { "epoch": 0.43, "learning_rate": 4.72900390625e-05, "loss": 2.2548, "step": 1445 }, { "epoch": 0.43, "learning_rate": 4.728816105769231e-05, "loss": 1.8372, "step": 1446 }, { "epoch": 0.43, "learning_rate": 4.728628305288462e-05, "loss": 2.2517, "step": 1447 }, { "epoch": 0.44, "learning_rate": 4.7284405048076924e-05, "loss": 2.2552, "step": 1448 }, { "epoch": 0.44, "learning_rate": 4.728252704326923e-05, "loss": 2.189, "step": 1449 }, { "epoch": 0.44, "learning_rate": 4.728064903846154e-05, "loss": 2.192, "step": 1450 }, { "epoch": 0.44, "learning_rate": 4.7278771033653845e-05, "loss": 2.2829, "step": 1451 }, { "epoch": 0.44, "learning_rate": 4.7276893028846156e-05, "loss": 2.3611, "step": 1452 }, { "epoch": 0.44, "learning_rate": 4.727501502403847e-05, "loss": 2.2343, "step": 1453 }, { "epoch": 0.44, "learning_rate": 4.727313701923077e-05, "loss": 2.1716, "step": 1454 }, { "epoch": 0.44, "learning_rate": 4.727125901442308e-05, "loss": 2.3085, "step": 1455 }, { "epoch": 0.44, "learning_rate": 4.726938100961539e-05, "loss": 2.0137, "step": 1456 }, { "epoch": 0.44, "learning_rate": 4.7267503004807694e-05, "loss": 2.3229, "step": 1457 }, { "epoch": 0.44, "learning_rate": 4.7265625000000005e-05, "loss": 2.0048, "step": 1458 }, { "epoch": 0.44, "learning_rate": 4.726374699519231e-05, "loss": 2.2712, "step": 1459 }, { "epoch": 0.44, "learning_rate": 4.7261868990384615e-05, "loss": 2.227, "step": 1460 }, { "epoch": 0.44, "learning_rate": 4.7259990985576926e-05, "loss": 2.4048, "step": 1461 }, { "epoch": 0.44, "learning_rate": 4.725811298076923e-05, "loss": 2.2608, "step": 1462 }, { "epoch": 0.44, "learning_rate": 4.7256234975961536e-05, "loss": 2.3166, "step": 1463 }, { "epoch": 0.44, "learning_rate": 4.725435697115385e-05, "loss": 2.0174, "step": 1464 }, { "epoch": 0.44, "learning_rate": 4.725247896634616e-05, "loss": 2.2126, "step": 1465 }, { "epoch": 0.44, "learning_rate": 4.7250600961538464e-05, "loss": 2.2279, "step": 1466 }, { "epoch": 0.44, "learning_rate": 4.7248722956730775e-05, "loss": 2.1204, "step": 1467 }, { "epoch": 0.44, "learning_rate": 4.724684495192308e-05, "loss": 2.1989, "step": 1468 }, { "epoch": 0.44, "learning_rate": 4.7244966947115385e-05, "loss": 2.2231, "step": 1469 }, { "epoch": 0.44, "learning_rate": 4.7243088942307697e-05, "loss": 2.2022, "step": 1470 }, { "epoch": 0.44, "learning_rate": 4.72412109375e-05, "loss": 2.3559, "step": 1471 }, { "epoch": 0.44, "learning_rate": 4.723933293269231e-05, "loss": 1.9928, "step": 1472 }, { "epoch": 0.44, "learning_rate": 4.723745492788462e-05, "loss": 2.2168, "step": 1473 }, { "epoch": 0.44, "learning_rate": 4.723557692307692e-05, "loss": 2.1202, "step": 1474 }, { "epoch": 0.44, "learning_rate": 4.7233698918269234e-05, "loss": 2.2737, "step": 1475 }, { "epoch": 0.44, "learning_rate": 4.723182091346154e-05, "loss": 2.0577, "step": 1476 }, { "epoch": 0.44, "learning_rate": 4.7229942908653843e-05, "loss": 2.0923, "step": 1477 }, { "epoch": 0.44, "learning_rate": 4.722806490384616e-05, "loss": 2.1208, "step": 1478 }, { "epoch": 0.44, "learning_rate": 4.7226186899038467e-05, "loss": 2.2472, "step": 1479 }, { "epoch": 0.44, "learning_rate": 4.722430889423077e-05, "loss": 2.1243, "step": 1480 }, { "epoch": 0.45, "learning_rate": 4.722243088942308e-05, "loss": 2.2995, "step": 1481 }, { "epoch": 0.45, "learning_rate": 4.722055288461539e-05, "loss": 2.287, "step": 1482 }, { "epoch": 0.45, "learning_rate": 4.721867487980769e-05, "loss": 2.431, "step": 1483 }, { "epoch": 0.45, "learning_rate": 4.7216796875000004e-05, "loss": 2.0972, "step": 1484 }, { "epoch": 0.45, "learning_rate": 4.721491887019231e-05, "loss": 2.2117, "step": 1485 }, { "epoch": 0.45, "learning_rate": 4.7213040865384614e-05, "loss": 2.3699, "step": 1486 }, { "epoch": 0.45, "learning_rate": 4.7211162860576925e-05, "loss": 2.3937, "step": 1487 }, { "epoch": 0.45, "learning_rate": 4.720928485576923e-05, "loss": 2.3438, "step": 1488 }, { "epoch": 0.45, "learning_rate": 4.720740685096154e-05, "loss": 2.3635, "step": 1489 }, { "epoch": 0.45, "learning_rate": 4.7205528846153846e-05, "loss": 1.9444, "step": 1490 }, { "epoch": 0.45, "learning_rate": 4.720365084134616e-05, "loss": 2.1363, "step": 1491 }, { "epoch": 0.45, "learning_rate": 4.720177283653847e-05, "loss": 2.1984, "step": 1492 }, { "epoch": 0.45, "learning_rate": 4.7199894831730774e-05, "loss": 2.215, "step": 1493 }, { "epoch": 0.45, "learning_rate": 4.719801682692308e-05, "loss": 2.2229, "step": 1494 }, { "epoch": 0.45, "learning_rate": 4.719613882211539e-05, "loss": 2.1464, "step": 1495 }, { "epoch": 0.45, "learning_rate": 4.7194260817307695e-05, "loss": 2.1957, "step": 1496 }, { "epoch": 0.45, "learning_rate": 4.71923828125e-05, "loss": 2.0109, "step": 1497 }, { "epoch": 0.45, "learning_rate": 4.719050480769231e-05, "loss": 2.327, "step": 1498 }, { "epoch": 0.45, "learning_rate": 4.7188626802884616e-05, "loss": 2.2587, "step": 1499 }, { "epoch": 0.45, "learning_rate": 4.718674879807692e-05, "loss": 2.1496, "step": 1500 }, { "epoch": 0.45, "learning_rate": 4.718487079326923e-05, "loss": 1.8909, "step": 1501 }, { "epoch": 0.45, "learning_rate": 4.718299278846154e-05, "loss": 2.3052, "step": 1502 }, { "epoch": 0.45, "learning_rate": 4.718111478365385e-05, "loss": 2.3763, "step": 1503 }, { "epoch": 0.45, "learning_rate": 4.717923677884616e-05, "loss": 2.02, "step": 1504 }, { "epoch": 0.45, "learning_rate": 4.7177358774038465e-05, "loss": 1.9713, "step": 1505 }, { "epoch": 0.45, "learning_rate": 4.717548076923077e-05, "loss": 2.2781, "step": 1506 }, { "epoch": 0.45, "learning_rate": 4.717360276442308e-05, "loss": 2.2566, "step": 1507 }, { "epoch": 0.45, "learning_rate": 4.7171724759615386e-05, "loss": 2.3991, "step": 1508 }, { "epoch": 0.45, "learning_rate": 4.71698467548077e-05, "loss": 2.2244, "step": 1509 }, { "epoch": 0.45, "learning_rate": 4.716796875e-05, "loss": 2.1176, "step": 1510 }, { "epoch": 0.45, "learning_rate": 4.716609074519231e-05, "loss": 2.4661, "step": 1511 }, { "epoch": 0.45, "learning_rate": 4.716421274038462e-05, "loss": 2.1815, "step": 1512 }, { "epoch": 0.45, "learning_rate": 4.7162334735576924e-05, "loss": 2.2451, "step": 1513 }, { "epoch": 0.45, "learning_rate": 4.716045673076923e-05, "loss": 2.1706, "step": 1514 }, { "epoch": 0.46, "learning_rate": 4.715857872596154e-05, "loss": 2.3374, "step": 1515 }, { "epoch": 0.46, "learning_rate": 4.7156700721153845e-05, "loss": 2.2614, "step": 1516 }, { "epoch": 0.46, "learning_rate": 4.7154822716346156e-05, "loss": 2.2218, "step": 1517 }, { "epoch": 0.46, "learning_rate": 4.715294471153847e-05, "loss": 2.126, "step": 1518 }, { "epoch": 0.46, "learning_rate": 4.715106670673077e-05, "loss": 2.1609, "step": 1519 }, { "epoch": 0.46, "learning_rate": 4.714918870192308e-05, "loss": 2.1601, "step": 1520 }, { "epoch": 0.46, "learning_rate": 4.714731069711539e-05, "loss": 2.3467, "step": 1521 }, { "epoch": 0.46, "learning_rate": 4.7145432692307694e-05, "loss": 2.1122, "step": 1522 }, { "epoch": 0.46, "learning_rate": 4.7143554687500005e-05, "loss": 1.9462, "step": 1523 }, { "epoch": 0.46, "learning_rate": 4.714167668269231e-05, "loss": 2.2664, "step": 1524 }, { "epoch": 0.46, "learning_rate": 4.7139798677884615e-05, "loss": 2.3107, "step": 1525 }, { "epoch": 0.46, "learning_rate": 4.7137920673076926e-05, "loss": 2.0887, "step": 1526 }, { "epoch": 0.46, "learning_rate": 4.713604266826923e-05, "loss": 1.9186, "step": 1527 }, { "epoch": 0.46, "learning_rate": 4.7134164663461536e-05, "loss": 2.1776, "step": 1528 }, { "epoch": 0.46, "learning_rate": 4.713228665865385e-05, "loss": 2.174, "step": 1529 }, { "epoch": 0.46, "learning_rate": 4.713040865384616e-05, "loss": 1.9483, "step": 1530 }, { "epoch": 0.46, "learning_rate": 4.7128530649038464e-05, "loss": 2.1706, "step": 1531 }, { "epoch": 0.46, "learning_rate": 4.7126652644230775e-05, "loss": 2.2476, "step": 1532 }, { "epoch": 0.46, "learning_rate": 4.712477463942308e-05, "loss": 2.3511, "step": 1533 }, { "epoch": 0.46, "learning_rate": 4.7122896634615385e-05, "loss": 1.9815, "step": 1534 }, { "epoch": 0.46, "learning_rate": 4.7121018629807696e-05, "loss": 2.2083, "step": 1535 }, { "epoch": 0.46, "learning_rate": 4.7119140625e-05, "loss": 2.2656, "step": 1536 }, { "epoch": 0.46, "learning_rate": 4.711726262019231e-05, "loss": 2.5057, "step": 1537 }, { "epoch": 0.46, "learning_rate": 4.711538461538462e-05, "loss": 2.2675, "step": 1538 }, { "epoch": 0.46, "learning_rate": 4.711350661057692e-05, "loss": 2.3721, "step": 1539 }, { "epoch": 0.46, "learning_rate": 4.7111628605769234e-05, "loss": 2.1243, "step": 1540 }, { "epoch": 0.46, "learning_rate": 4.710975060096154e-05, "loss": 2.1953, "step": 1541 }, { "epoch": 0.46, "learning_rate": 4.710787259615384e-05, "loss": 2.4029, "step": 1542 }, { "epoch": 0.46, "learning_rate": 4.710599459134616e-05, "loss": 2.0864, "step": 1543 }, { "epoch": 0.46, "learning_rate": 4.7104116586538466e-05, "loss": 2.3101, "step": 1544 }, { "epoch": 0.46, "learning_rate": 4.710223858173077e-05, "loss": 2.2423, "step": 1545 }, { "epoch": 0.46, "learning_rate": 4.710036057692308e-05, "loss": 2.1477, "step": 1546 }, { "epoch": 0.46, "learning_rate": 4.709848257211539e-05, "loss": 2.0443, "step": 1547 }, { "epoch": 0.47, "learning_rate": 4.709660456730769e-05, "loss": 2.0876, "step": 1548 }, { "epoch": 0.47, "learning_rate": 4.7094726562500004e-05, "loss": 2.3159, "step": 1549 }, { "epoch": 0.47, "learning_rate": 4.709284855769231e-05, "loss": 2.2173, "step": 1550 }, { "epoch": 0.47, "learning_rate": 4.7090970552884613e-05, "loss": 2.0378, "step": 1551 }, { "epoch": 0.47, "learning_rate": 4.7089092548076925e-05, "loss": 2.2953, "step": 1552 }, { "epoch": 0.47, "learning_rate": 4.708721454326923e-05, "loss": 2.2296, "step": 1553 }, { "epoch": 0.47, "learning_rate": 4.708533653846154e-05, "loss": 1.9383, "step": 1554 }, { "epoch": 0.47, "learning_rate": 4.7083458533653846e-05, "loss": 2.2814, "step": 1555 }, { "epoch": 0.47, "learning_rate": 4.708158052884616e-05, "loss": 2.4377, "step": 1556 }, { "epoch": 0.47, "learning_rate": 4.707970252403847e-05, "loss": 2.0905, "step": 1557 }, { "epoch": 0.47, "learning_rate": 4.7077824519230774e-05, "loss": 2.3124, "step": 1558 }, { "epoch": 0.47, "learning_rate": 4.707594651442308e-05, "loss": 1.9978, "step": 1559 }, { "epoch": 0.47, "learning_rate": 4.707406850961539e-05, "loss": 2.2612, "step": 1560 }, { "epoch": 0.47, "learning_rate": 4.7072190504807695e-05, "loss": 2.1745, "step": 1561 }, { "epoch": 0.47, "learning_rate": 4.70703125e-05, "loss": 1.8687, "step": 1562 }, { "epoch": 0.47, "learning_rate": 4.706843449519231e-05, "loss": 2.1982, "step": 1563 }, { "epoch": 0.47, "learning_rate": 4.7066556490384616e-05, "loss": 2.3274, "step": 1564 }, { "epoch": 0.47, "learning_rate": 4.706467848557692e-05, "loss": 2.1139, "step": 1565 }, { "epoch": 0.47, "learning_rate": 4.706280048076923e-05, "loss": 2.1443, "step": 1566 }, { "epoch": 0.47, "learning_rate": 4.706092247596154e-05, "loss": 2.3461, "step": 1567 }, { "epoch": 0.47, "learning_rate": 4.705904447115385e-05, "loss": 1.9206, "step": 1568 }, { "epoch": 0.47, "learning_rate": 4.705716646634616e-05, "loss": 2.0738, "step": 1569 }, { "epoch": 0.47, "learning_rate": 4.7055288461538465e-05, "loss": 1.9544, "step": 1570 }, { "epoch": 0.47, "learning_rate": 4.705341045673077e-05, "loss": 2.1575, "step": 1571 }, { "epoch": 0.47, "learning_rate": 4.705153245192308e-05, "loss": 2.4846, "step": 1572 }, { "epoch": 0.47, "learning_rate": 4.7049654447115386e-05, "loss": 2.2634, "step": 1573 }, { "epoch": 0.47, "learning_rate": 4.70477764423077e-05, "loss": 2.2896, "step": 1574 }, { "epoch": 0.47, "learning_rate": 4.70458984375e-05, "loss": 2.5765, "step": 1575 }, { "epoch": 0.47, "learning_rate": 4.704402043269231e-05, "loss": 2.1189, "step": 1576 }, { "epoch": 0.47, "learning_rate": 4.704214242788462e-05, "loss": 2.0905, "step": 1577 }, { "epoch": 0.47, "learning_rate": 4.7040264423076924e-05, "loss": 2.2525, "step": 1578 }, { "epoch": 0.47, "learning_rate": 4.703838641826923e-05, "loss": 2.1163, "step": 1579 }, { "epoch": 0.47, "learning_rate": 4.703650841346154e-05, "loss": 2.3812, "step": 1580 }, { "epoch": 0.48, "learning_rate": 4.7034630408653845e-05, "loss": 2.3076, "step": 1581 }, { "epoch": 0.48, "learning_rate": 4.7032752403846156e-05, "loss": 2.2544, "step": 1582 }, { "epoch": 0.48, "learning_rate": 4.703087439903847e-05, "loss": 1.9618, "step": 1583 }, { "epoch": 0.48, "learning_rate": 4.702899639423077e-05, "loss": 2.3686, "step": 1584 }, { "epoch": 0.48, "learning_rate": 4.702711838942308e-05, "loss": 1.9493, "step": 1585 }, { "epoch": 0.48, "learning_rate": 4.702524038461539e-05, "loss": 2.3493, "step": 1586 }, { "epoch": 0.48, "learning_rate": 4.7023362379807694e-05, "loss": 2.026, "step": 1587 }, { "epoch": 0.48, "learning_rate": 4.7021484375000005e-05, "loss": 2.395, "step": 1588 }, { "epoch": 0.48, "learning_rate": 4.701960637019231e-05, "loss": 2.1827, "step": 1589 }, { "epoch": 0.48, "learning_rate": 4.7017728365384615e-05, "loss": 2.2944, "step": 1590 }, { "epoch": 0.48, "learning_rate": 4.7015850360576926e-05, "loss": 2.1377, "step": 1591 }, { "epoch": 0.48, "learning_rate": 4.701397235576923e-05, "loss": 2.0906, "step": 1592 }, { "epoch": 0.48, "learning_rate": 4.7012094350961536e-05, "loss": 2.3192, "step": 1593 }, { "epoch": 0.48, "learning_rate": 4.701021634615385e-05, "loss": 2.1865, "step": 1594 }, { "epoch": 0.48, "learning_rate": 4.700833834134616e-05, "loss": 2.1757, "step": 1595 }, { "epoch": 0.48, "learning_rate": 4.7006460336538464e-05, "loss": 2.3942, "step": 1596 }, { "epoch": 0.48, "learning_rate": 4.7004582331730775e-05, "loss": 2.314, "step": 1597 }, { "epoch": 0.48, "learning_rate": 4.700270432692308e-05, "loss": 2.3642, "step": 1598 }, { "epoch": 0.48, "learning_rate": 4.7000826322115385e-05, "loss": 2.1535, "step": 1599 }, { "epoch": 0.48, "learning_rate": 4.6998948317307696e-05, "loss": 2.3045, "step": 1600 }, { "epoch": 0.48, "learning_rate": 4.69970703125e-05, "loss": 2.1423, "step": 1601 }, { "epoch": 0.48, "learning_rate": 4.699519230769231e-05, "loss": 2.1891, "step": 1602 }, { "epoch": 0.48, "learning_rate": 4.699331430288462e-05, "loss": 2.1692, "step": 1603 }, { "epoch": 0.48, "learning_rate": 4.699143629807692e-05, "loss": 2.1998, "step": 1604 }, { "epoch": 0.48, "learning_rate": 4.6989558293269234e-05, "loss": 1.9279, "step": 1605 }, { "epoch": 0.48, "learning_rate": 4.698768028846154e-05, "loss": 2.3855, "step": 1606 }, { "epoch": 0.48, "learning_rate": 4.698580228365384e-05, "loss": 2.2029, "step": 1607 }, { "epoch": 0.48, "learning_rate": 4.698392427884616e-05, "loss": 2.2544, "step": 1608 }, { "epoch": 0.48, "learning_rate": 4.6982046274038466e-05, "loss": 2.2714, "step": 1609 }, { "epoch": 0.48, "learning_rate": 4.698016826923077e-05, "loss": 2.1954, "step": 1610 }, { "epoch": 0.48, "learning_rate": 4.697829026442308e-05, "loss": 2.1118, "step": 1611 }, { "epoch": 0.48, "learning_rate": 4.697641225961539e-05, "loss": 2.1001, "step": 1612 }, { "epoch": 0.48, "learning_rate": 4.697453425480769e-05, "loss": 2.3575, "step": 1613 }, { "epoch": 0.48, "learning_rate": 4.6972656250000004e-05, "loss": 2.3569, "step": 1614 }, { "epoch": 0.49, "learning_rate": 4.697077824519231e-05, "loss": 2.2046, "step": 1615 }, { "epoch": 0.49, "learning_rate": 4.696890024038462e-05, "loss": 2.3248, "step": 1616 }, { "epoch": 0.49, "learning_rate": 4.6967022235576925e-05, "loss": 2.1386, "step": 1617 }, { "epoch": 0.49, "learning_rate": 4.696514423076923e-05, "loss": 2.1548, "step": 1618 }, { "epoch": 0.49, "learning_rate": 4.696326622596154e-05, "loss": 2.0461, "step": 1619 }, { "epoch": 0.49, "learning_rate": 4.6961388221153846e-05, "loss": 2.0383, "step": 1620 }, { "epoch": 0.49, "learning_rate": 4.695951021634616e-05, "loss": 1.9406, "step": 1621 }, { "epoch": 0.49, "learning_rate": 4.695763221153847e-05, "loss": 2.4657, "step": 1622 }, { "epoch": 0.49, "learning_rate": 4.6955754206730774e-05, "loss": 2.1145, "step": 1623 }, { "epoch": 0.49, "learning_rate": 4.695387620192308e-05, "loss": 1.8965, "step": 1624 }, { "epoch": 0.49, "learning_rate": 4.695199819711539e-05, "loss": 2.2919, "step": 1625 }, { "epoch": 0.49, "learning_rate": 4.6950120192307695e-05, "loss": 2.1689, "step": 1626 }, { "epoch": 0.49, "learning_rate": 4.69482421875e-05, "loss": 2.3569, "step": 1627 }, { "epoch": 0.49, "learning_rate": 4.694636418269231e-05, "loss": 2.2617, "step": 1628 }, { "epoch": 0.49, "learning_rate": 4.6944486177884616e-05, "loss": 2.5532, "step": 1629 }, { "epoch": 0.49, "learning_rate": 4.694260817307692e-05, "loss": 2.4408, "step": 1630 }, { "epoch": 0.49, "learning_rate": 4.694073016826923e-05, "loss": 2.2113, "step": 1631 }, { "epoch": 0.49, "learning_rate": 4.693885216346154e-05, "loss": 2.4904, "step": 1632 }, { "epoch": 0.49, "learning_rate": 4.693697415865385e-05, "loss": 1.9189, "step": 1633 }, { "epoch": 0.49, "learning_rate": 4.693509615384616e-05, "loss": 1.9534, "step": 1634 }, { "epoch": 0.49, "learning_rate": 4.6933218149038465e-05, "loss": 2.2566, "step": 1635 }, { "epoch": 0.49, "learning_rate": 4.6931340144230777e-05, "loss": 2.0915, "step": 1636 }, { "epoch": 0.49, "learning_rate": 4.692946213942308e-05, "loss": 2.1207, "step": 1637 }, { "epoch": 0.49, "learning_rate": 4.6927584134615386e-05, "loss": 1.9073, "step": 1638 }, { "epoch": 0.49, "learning_rate": 4.69257061298077e-05, "loss": 2.3132, "step": 1639 }, { "epoch": 0.49, "learning_rate": 4.6923828125e-05, "loss": 2.1576, "step": 1640 }, { "epoch": 0.49, "learning_rate": 4.692195012019231e-05, "loss": 2.5179, "step": 1641 }, { "epoch": 0.49, "learning_rate": 4.692007211538462e-05, "loss": 2.3535, "step": 1642 }, { "epoch": 0.49, "learning_rate": 4.6918194110576923e-05, "loss": 1.94, "step": 1643 }, { "epoch": 0.49, "learning_rate": 4.691631610576923e-05, "loss": 2.3775, "step": 1644 }, { "epoch": 0.49, "learning_rate": 4.691443810096154e-05, "loss": 2.243, "step": 1645 }, { "epoch": 0.49, "learning_rate": 4.6912560096153845e-05, "loss": 1.9082, "step": 1646 }, { "epoch": 0.49, "learning_rate": 4.6910682091346156e-05, "loss": 2.3278, "step": 1647 }, { "epoch": 0.5, "learning_rate": 4.690880408653847e-05, "loss": 2.672, "step": 1648 }, { "epoch": 0.5, "learning_rate": 4.690692608173077e-05, "loss": 2.1996, "step": 1649 }, { "epoch": 0.5, "learning_rate": 4.690504807692308e-05, "loss": 2.4316, "step": 1650 }, { "epoch": 0.5, "learning_rate": 4.690317007211539e-05, "loss": 2.2895, "step": 1651 }, { "epoch": 0.5, "learning_rate": 4.6901292067307694e-05, "loss": 2.2049, "step": 1652 }, { "epoch": 0.5, "learning_rate": 4.6899414062500005e-05, "loss": 2.2084, "step": 1653 }, { "epoch": 0.5, "learning_rate": 4.689753605769231e-05, "loss": 2.4822, "step": 1654 }, { "epoch": 0.5, "learning_rate": 4.6895658052884615e-05, "loss": 2.4278, "step": 1655 }, { "epoch": 0.5, "learning_rate": 4.6893780048076926e-05, "loss": 2.0265, "step": 1656 }, { "epoch": 0.5, "learning_rate": 4.689190204326923e-05, "loss": 2.3761, "step": 1657 }, { "epoch": 0.5, "learning_rate": 4.6890024038461536e-05, "loss": 2.3251, "step": 1658 }, { "epoch": 0.5, "learning_rate": 4.688814603365385e-05, "loss": 2.1984, "step": 1659 }, { "epoch": 0.5, "learning_rate": 4.688626802884616e-05, "loss": 2.2916, "step": 1660 }, { "epoch": 0.5, "learning_rate": 4.6884390024038464e-05, "loss": 2.0275, "step": 1661 }, { "epoch": 0.5, "learning_rate": 4.6882512019230775e-05, "loss": 2.2123, "step": 1662 }, { "epoch": 0.5, "learning_rate": 4.688063401442308e-05, "loss": 2.3227, "step": 1663 }, { "epoch": 0.5, "learning_rate": 4.6878756009615385e-05, "loss": 2.0498, "step": 1664 }, { "epoch": 0.5, "learning_rate": 4.6876878004807696e-05, "loss": 2.1553, "step": 1665 }, { "epoch": 0.5, "learning_rate": 4.6875e-05, "loss": 2.198, "step": 1666 }, { "epoch": 0.5, "learning_rate": 4.687312199519231e-05, "loss": 1.9594, "step": 1667 }, { "epoch": 0.5, "learning_rate": 4.687124399038462e-05, "loss": 2.3068, "step": 1668 }, { "epoch": 0.5, "learning_rate": 4.686936598557692e-05, "loss": 2.3984, "step": 1669 }, { "epoch": 0.5, "learning_rate": 4.6867487980769234e-05, "loss": 1.9125, "step": 1670 }, { "epoch": 0.5, "learning_rate": 4.686560997596154e-05, "loss": 1.792, "step": 1671 }, { "epoch": 0.5, "learning_rate": 4.686373197115384e-05, "loss": 2.1942, "step": 1672 }, { "epoch": 0.5, "learning_rate": 4.686185396634616e-05, "loss": 2.1845, "step": 1673 }, { "epoch": 0.5, "learning_rate": 4.6859975961538466e-05, "loss": 2.0573, "step": 1674 }, { "epoch": 0.5, "learning_rate": 4.685809795673077e-05, "loss": 2.2931, "step": 1675 }, { "epoch": 0.5, "learning_rate": 4.685621995192308e-05, "loss": 2.3434, "step": 1676 }, { "epoch": 0.5, "learning_rate": 4.685434194711539e-05, "loss": 2.4178, "step": 1677 }, { "epoch": 0.5, "learning_rate": 4.685246394230769e-05, "loss": 2.243, "step": 1678 }, { "epoch": 0.5, "learning_rate": 4.6850585937500004e-05, "loss": 2.2062, "step": 1679 }, { "epoch": 0.5, "learning_rate": 4.684870793269231e-05, "loss": 2.1657, "step": 1680 }, { "epoch": 0.51, "learning_rate": 4.684682992788462e-05, "loss": 2.0212, "step": 1681 }, { "epoch": 0.51, "learning_rate": 4.6844951923076925e-05, "loss": 2.39, "step": 1682 }, { "epoch": 0.51, "learning_rate": 4.684307391826923e-05, "loss": 1.9906, "step": 1683 }, { "epoch": 0.51, "learning_rate": 4.684119591346154e-05, "loss": 2.1522, "step": 1684 }, { "epoch": 0.51, "learning_rate": 4.6839317908653846e-05, "loss": 1.8074, "step": 1685 }, { "epoch": 0.51, "learning_rate": 4.683743990384616e-05, "loss": 2.0081, "step": 1686 }, { "epoch": 0.51, "learning_rate": 4.683556189903847e-05, "loss": 2.1205, "step": 1687 }, { "epoch": 0.51, "learning_rate": 4.6833683894230774e-05, "loss": 2.2923, "step": 1688 }, { "epoch": 0.51, "learning_rate": 4.683180588942308e-05, "loss": 2.35, "step": 1689 }, { "epoch": 0.51, "learning_rate": 4.682992788461539e-05, "loss": 2.1154, "step": 1690 }, { "epoch": 0.51, "learning_rate": 4.6828049879807695e-05, "loss": 2.3854, "step": 1691 }, { "epoch": 0.51, "learning_rate": 4.6826171875e-05, "loss": 2.0479, "step": 1692 }, { "epoch": 0.51, "learning_rate": 4.682429387019231e-05, "loss": 2.0873, "step": 1693 }, { "epoch": 0.51, "learning_rate": 4.6822415865384616e-05, "loss": 2.3363, "step": 1694 }, { "epoch": 0.51, "learning_rate": 4.682053786057692e-05, "loss": 2.1317, "step": 1695 }, { "epoch": 0.51, "learning_rate": 4.681865985576923e-05, "loss": 2.4833, "step": 1696 }, { "epoch": 0.51, "learning_rate": 4.681678185096154e-05, "loss": 2.2711, "step": 1697 }, { "epoch": 0.51, "learning_rate": 4.681490384615385e-05, "loss": 1.9761, "step": 1698 }, { "epoch": 0.51, "learning_rate": 4.681302584134616e-05, "loss": 2.282, "step": 1699 }, { "epoch": 0.51, "learning_rate": 4.6811147836538465e-05, "loss": 1.9212, "step": 1700 }, { "epoch": 0.51, "learning_rate": 4.6809269831730776e-05, "loss": 2.4575, "step": 1701 }, { "epoch": 0.51, "learning_rate": 4.680739182692308e-05, "loss": 2.3969, "step": 1702 }, { "epoch": 0.51, "learning_rate": 4.6805513822115386e-05, "loss": 2.6305, "step": 1703 }, { "epoch": 0.51, "learning_rate": 4.68036358173077e-05, "loss": 2.1476, "step": 1704 }, { "epoch": 0.51, "learning_rate": 4.68017578125e-05, "loss": 2.325, "step": 1705 }, { "epoch": 0.51, "learning_rate": 4.679987980769231e-05, "loss": 2.1822, "step": 1706 }, { "epoch": 0.51, "learning_rate": 4.679800180288462e-05, "loss": 2.0383, "step": 1707 }, { "epoch": 0.51, "learning_rate": 4.679612379807692e-05, "loss": 2.2255, "step": 1708 }, { "epoch": 0.51, "learning_rate": 4.679424579326923e-05, "loss": 2.1242, "step": 1709 }, { "epoch": 0.51, "learning_rate": 4.679236778846154e-05, "loss": 2.0055, "step": 1710 }, { "epoch": 0.51, "learning_rate": 4.6790489783653844e-05, "loss": 2.3797, "step": 1711 }, { "epoch": 0.51, "learning_rate": 4.6788611778846156e-05, "loss": 2.3626, "step": 1712 }, { "epoch": 0.51, "learning_rate": 4.678673377403847e-05, "loss": 2.317, "step": 1713 }, { "epoch": 0.52, "learning_rate": 4.678485576923077e-05, "loss": 2.2248, "step": 1714 }, { "epoch": 0.52, "learning_rate": 4.678297776442308e-05, "loss": 2.3072, "step": 1715 }, { "epoch": 0.52, "learning_rate": 4.678109975961539e-05, "loss": 2.1403, "step": 1716 }, { "epoch": 0.52, "learning_rate": 4.6779221754807693e-05, "loss": 2.0918, "step": 1717 }, { "epoch": 0.52, "learning_rate": 4.6777343750000005e-05, "loss": 2.1625, "step": 1718 }, { "epoch": 0.52, "learning_rate": 4.677546574519231e-05, "loss": 1.9072, "step": 1719 }, { "epoch": 0.52, "learning_rate": 4.6773587740384614e-05, "loss": 2.2811, "step": 1720 }, { "epoch": 0.52, "learning_rate": 4.6771709735576926e-05, "loss": 2.163, "step": 1721 }, { "epoch": 0.52, "learning_rate": 4.676983173076923e-05, "loss": 2.1809, "step": 1722 }, { "epoch": 0.52, "learning_rate": 4.6767953725961536e-05, "loss": 2.0116, "step": 1723 }, { "epoch": 0.52, "learning_rate": 4.676607572115385e-05, "loss": 2.3161, "step": 1724 }, { "epoch": 0.52, "learning_rate": 4.676419771634616e-05, "loss": 2.0415, "step": 1725 }, { "epoch": 0.52, "learning_rate": 4.6762319711538463e-05, "loss": 2.0428, "step": 1726 }, { "epoch": 0.52, "learning_rate": 4.6760441706730775e-05, "loss": 1.9974, "step": 1727 }, { "epoch": 0.52, "learning_rate": 4.675856370192308e-05, "loss": 2.3233, "step": 1728 }, { "epoch": 0.52, "learning_rate": 4.6756685697115385e-05, "loss": 2.2548, "step": 1729 }, { "epoch": 0.52, "learning_rate": 4.6754807692307696e-05, "loss": 2.257, "step": 1730 }, { "epoch": 0.52, "learning_rate": 4.67529296875e-05, "loss": 2.2782, "step": 1731 }, { "epoch": 0.52, "learning_rate": 4.675105168269231e-05, "loss": 2.1109, "step": 1732 }, { "epoch": 0.52, "learning_rate": 4.674917367788462e-05, "loss": 1.9774, "step": 1733 }, { "epoch": 0.52, "learning_rate": 4.674729567307692e-05, "loss": 2.1656, "step": 1734 }, { "epoch": 0.52, "learning_rate": 4.6745417668269234e-05, "loss": 2.1791, "step": 1735 }, { "epoch": 0.52, "learning_rate": 4.674353966346154e-05, "loss": 2.4566, "step": 1736 }, { "epoch": 0.52, "learning_rate": 4.674166165865384e-05, "loss": 2.0973, "step": 1737 }, { "epoch": 0.52, "learning_rate": 4.673978365384616e-05, "loss": 2.3695, "step": 1738 }, { "epoch": 0.52, "learning_rate": 4.6737905649038466e-05, "loss": 2.3386, "step": 1739 }, { "epoch": 0.52, "learning_rate": 4.673602764423077e-05, "loss": 2.1032, "step": 1740 }, { "epoch": 0.52, "learning_rate": 4.673414963942308e-05, "loss": 2.098, "step": 1741 }, { "epoch": 0.52, "learning_rate": 4.673227163461539e-05, "loss": 2.3565, "step": 1742 }, { "epoch": 0.52, "learning_rate": 4.673039362980769e-05, "loss": 2.2177, "step": 1743 }, { "epoch": 0.52, "learning_rate": 4.6728515625000004e-05, "loss": 2.1468, "step": 1744 }, { "epoch": 0.52, "learning_rate": 4.672663762019231e-05, "loss": 2.1994, "step": 1745 }, { "epoch": 0.52, "learning_rate": 4.672475961538462e-05, "loss": 1.9419, "step": 1746 }, { "epoch": 0.52, "learning_rate": 4.6722881610576925e-05, "loss": 2.0939, "step": 1747 }, { "epoch": 0.53, "learning_rate": 4.672100360576923e-05, "loss": 2.5101, "step": 1748 }, { "epoch": 0.53, "learning_rate": 4.671912560096154e-05, "loss": 2.1105, "step": 1749 }, { "epoch": 0.53, "learning_rate": 4.6717247596153846e-05, "loss": 2.153, "step": 1750 }, { "epoch": 0.53, "learning_rate": 4.671536959134616e-05, "loss": 2.0835, "step": 1751 }, { "epoch": 0.53, "learning_rate": 4.671349158653847e-05, "loss": 2.1555, "step": 1752 }, { "epoch": 0.53, "learning_rate": 4.6711613581730774e-05, "loss": 2.4143, "step": 1753 }, { "epoch": 0.53, "learning_rate": 4.670973557692308e-05, "loss": 2.0299, "step": 1754 }, { "epoch": 0.53, "learning_rate": 4.670785757211539e-05, "loss": 2.2642, "step": 1755 }, { "epoch": 0.53, "learning_rate": 4.6705979567307695e-05, "loss": 2.2116, "step": 1756 }, { "epoch": 0.53, "learning_rate": 4.67041015625e-05, "loss": 2.1597, "step": 1757 }, { "epoch": 0.53, "learning_rate": 4.670222355769231e-05, "loss": 2.0589, "step": 1758 }, { "epoch": 0.53, "learning_rate": 4.6700345552884616e-05, "loss": 2.2233, "step": 1759 }, { "epoch": 0.53, "learning_rate": 4.669846754807693e-05, "loss": 2.1236, "step": 1760 }, { "epoch": 0.53, "learning_rate": 4.669658954326923e-05, "loss": 2.2575, "step": 1761 }, { "epoch": 0.53, "learning_rate": 4.669471153846154e-05, "loss": 2.2407, "step": 1762 }, { "epoch": 0.53, "learning_rate": 4.669283353365385e-05, "loss": 1.8892, "step": 1763 }, { "epoch": 0.53, "learning_rate": 4.669095552884616e-05, "loss": 2.4514, "step": 1764 }, { "epoch": 0.53, "learning_rate": 4.6689077524038465e-05, "loss": 2.2399, "step": 1765 }, { "epoch": 0.53, "learning_rate": 4.6687199519230776e-05, "loss": 1.8581, "step": 1766 }, { "epoch": 0.53, "learning_rate": 4.668532151442308e-05, "loss": 2.4119, "step": 1767 }, { "epoch": 0.53, "learning_rate": 4.6683443509615386e-05, "loss": 1.9714, "step": 1768 }, { "epoch": 0.53, "learning_rate": 4.66815655048077e-05, "loss": 2.0647, "step": 1769 }, { "epoch": 0.53, "learning_rate": 4.66796875e-05, "loss": 2.3726, "step": 1770 }, { "epoch": 0.53, "learning_rate": 4.667780949519231e-05, "loss": 1.9961, "step": 1771 }, { "epoch": 0.53, "learning_rate": 4.667593149038462e-05, "loss": 2.2712, "step": 1772 }, { "epoch": 0.53, "learning_rate": 4.667405348557692e-05, "loss": 1.9339, "step": 1773 }, { "epoch": 0.53, "learning_rate": 4.667217548076923e-05, "loss": 2.1919, "step": 1774 }, { "epoch": 0.53, "learning_rate": 4.667029747596154e-05, "loss": 2.1472, "step": 1775 }, { "epoch": 0.53, "learning_rate": 4.6668419471153844e-05, "loss": 2.1861, "step": 1776 }, { "epoch": 0.53, "learning_rate": 4.6666541466346156e-05, "loss": 2.1308, "step": 1777 }, { "epoch": 0.53, "learning_rate": 4.666466346153847e-05, "loss": 2.2511, "step": 1778 }, { "epoch": 0.53, "learning_rate": 4.666278545673077e-05, "loss": 2.2631, "step": 1779 }, { "epoch": 0.53, "learning_rate": 4.6660907451923084e-05, "loss": 2.1526, "step": 1780 }, { "epoch": 0.54, "learning_rate": 4.665902944711539e-05, "loss": 2.3998, "step": 1781 }, { "epoch": 0.54, "learning_rate": 4.665715144230769e-05, "loss": 2.0666, "step": 1782 }, { "epoch": 0.54, "learning_rate": 4.6655273437500005e-05, "loss": 2.2279, "step": 1783 }, { "epoch": 0.54, "learning_rate": 4.665339543269231e-05, "loss": 2.0778, "step": 1784 }, { "epoch": 0.54, "learning_rate": 4.6651517427884614e-05, "loss": 2.5064, "step": 1785 }, { "epoch": 0.54, "learning_rate": 4.6649639423076926e-05, "loss": 2.1436, "step": 1786 }, { "epoch": 0.54, "learning_rate": 4.664776141826923e-05, "loss": 2.0242, "step": 1787 }, { "epoch": 0.54, "learning_rate": 4.6645883413461535e-05, "loss": 2.0537, "step": 1788 }, { "epoch": 0.54, "learning_rate": 4.664400540865385e-05, "loss": 2.0194, "step": 1789 }, { "epoch": 0.54, "learning_rate": 4.664212740384616e-05, "loss": 1.9538, "step": 1790 }, { "epoch": 0.54, "learning_rate": 4.664024939903846e-05, "loss": 2.0961, "step": 1791 }, { "epoch": 0.54, "learning_rate": 4.6638371394230775e-05, "loss": 2.4958, "step": 1792 }, { "epoch": 0.54, "learning_rate": 4.663649338942308e-05, "loss": 2.0391, "step": 1793 }, { "epoch": 0.54, "learning_rate": 4.6634615384615384e-05, "loss": 2.1526, "step": 1794 }, { "epoch": 0.54, "learning_rate": 4.6632737379807696e-05, "loss": 2.3222, "step": 1795 }, { "epoch": 0.54, "learning_rate": 4.6630859375e-05, "loss": 2.1845, "step": 1796 }, { "epoch": 0.54, "learning_rate": 4.662898137019231e-05, "loss": 2.0415, "step": 1797 }, { "epoch": 0.54, "learning_rate": 4.662710336538462e-05, "loss": 2.2911, "step": 1798 }, { "epoch": 0.54, "learning_rate": 4.662522536057692e-05, "loss": 2.0471, "step": 1799 }, { "epoch": 0.54, "learning_rate": 4.6623347355769233e-05, "loss": 2.2098, "step": 1800 }, { "epoch": 0.54, "learning_rate": 4.662146935096154e-05, "loss": 2.1431, "step": 1801 }, { "epoch": 0.54, "learning_rate": 4.661959134615384e-05, "loss": 1.8479, "step": 1802 }, { "epoch": 0.54, "learning_rate": 4.661771334134616e-05, "loss": 2.0511, "step": 1803 }, { "epoch": 0.54, "learning_rate": 4.6615835336538466e-05, "loss": 2.3194, "step": 1804 }, { "epoch": 0.54, "learning_rate": 4.661395733173077e-05, "loss": 2.0005, "step": 1805 }, { "epoch": 0.54, "learning_rate": 4.661207932692308e-05, "loss": 2.3443, "step": 1806 }, { "epoch": 0.54, "learning_rate": 4.661020132211539e-05, "loss": 2.1095, "step": 1807 }, { "epoch": 0.54, "learning_rate": 4.660832331730769e-05, "loss": 2.2155, "step": 1808 }, { "epoch": 0.54, "learning_rate": 4.6606445312500003e-05, "loss": 2.1857, "step": 1809 }, { "epoch": 0.54, "learning_rate": 4.660456730769231e-05, "loss": 2.192, "step": 1810 }, { "epoch": 0.54, "learning_rate": 4.660268930288462e-05, "loss": 2.3224, "step": 1811 }, { "epoch": 0.54, "learning_rate": 4.6600811298076925e-05, "loss": 2.0941, "step": 1812 }, { "epoch": 0.54, "learning_rate": 4.659893329326923e-05, "loss": 2.2326, "step": 1813 }, { "epoch": 0.55, "learning_rate": 4.659705528846154e-05, "loss": 2.0003, "step": 1814 }, { "epoch": 0.55, "learning_rate": 4.6595177283653846e-05, "loss": 2.3944, "step": 1815 }, { "epoch": 0.55, "learning_rate": 4.659329927884616e-05, "loss": 2.4993, "step": 1816 }, { "epoch": 0.55, "learning_rate": 4.659142127403847e-05, "loss": 2.2434, "step": 1817 }, { "epoch": 0.55, "learning_rate": 4.6589543269230773e-05, "loss": 2.445, "step": 1818 }, { "epoch": 0.55, "learning_rate": 4.658766526442308e-05, "loss": 2.0498, "step": 1819 }, { "epoch": 0.55, "learning_rate": 4.658578725961539e-05, "loss": 2.1766, "step": 1820 }, { "epoch": 0.55, "learning_rate": 4.6583909254807695e-05, "loss": 2.108, "step": 1821 }, { "epoch": 0.55, "learning_rate": 4.658203125e-05, "loss": 2.2093, "step": 1822 }, { "epoch": 0.55, "learning_rate": 4.658015324519231e-05, "loss": 2.1454, "step": 1823 }, { "epoch": 0.55, "learning_rate": 4.6578275240384616e-05, "loss": 1.9551, "step": 1824 }, { "epoch": 0.55, "learning_rate": 4.657639723557693e-05, "loss": 2.1024, "step": 1825 }, { "epoch": 0.55, "learning_rate": 4.657451923076923e-05, "loss": 2.2024, "step": 1826 }, { "epoch": 0.55, "learning_rate": 4.657264122596154e-05, "loss": 1.9092, "step": 1827 }, { "epoch": 0.55, "learning_rate": 4.657076322115385e-05, "loss": 1.836, "step": 1828 }, { "epoch": 0.55, "learning_rate": 4.656888521634616e-05, "loss": 2.1665, "step": 1829 }, { "epoch": 0.55, "learning_rate": 4.6567007211538465e-05, "loss": 2.041, "step": 1830 }, { "epoch": 0.55, "learning_rate": 4.6565129206730776e-05, "loss": 2.0528, "step": 1831 }, { "epoch": 0.55, "learning_rate": 4.656325120192308e-05, "loss": 2.1631, "step": 1832 }, { "epoch": 0.55, "learning_rate": 4.6561373197115386e-05, "loss": 2.4616, "step": 1833 }, { "epoch": 0.55, "learning_rate": 4.65594951923077e-05, "loss": 2.327, "step": 1834 }, { "epoch": 0.55, "learning_rate": 4.65576171875e-05, "loss": 2.3241, "step": 1835 }, { "epoch": 0.55, "learning_rate": 4.655573918269231e-05, "loss": 2.0038, "step": 1836 }, { "epoch": 0.55, "learning_rate": 4.655386117788462e-05, "loss": 2.3423, "step": 1837 }, { "epoch": 0.55, "learning_rate": 4.655198317307692e-05, "loss": 2.1971, "step": 1838 }, { "epoch": 0.55, "learning_rate": 4.655010516826923e-05, "loss": 2.0887, "step": 1839 }, { "epoch": 0.55, "learning_rate": 4.654822716346154e-05, "loss": 2.093, "step": 1840 }, { "epoch": 0.55, "learning_rate": 4.6546349158653844e-05, "loss": 2.0103, "step": 1841 }, { "epoch": 0.55, "learning_rate": 4.6544471153846156e-05, "loss": 2.0289, "step": 1842 }, { "epoch": 0.55, "learning_rate": 4.654259314903847e-05, "loss": 2.179, "step": 1843 }, { "epoch": 0.55, "learning_rate": 4.654071514423077e-05, "loss": 2.274, "step": 1844 }, { "epoch": 0.55, "learning_rate": 4.6538837139423084e-05, "loss": 1.9868, "step": 1845 }, { "epoch": 0.55, "learning_rate": 4.653695913461539e-05, "loss": 2.5569, "step": 1846 }, { "epoch": 0.55, "learning_rate": 4.653508112980769e-05, "loss": 2.1772, "step": 1847 }, { "epoch": 0.56, "learning_rate": 4.6533203125000005e-05, "loss": 2.0936, "step": 1848 }, { "epoch": 0.56, "learning_rate": 4.653132512019231e-05, "loss": 2.0966, "step": 1849 }, { "epoch": 0.56, "learning_rate": 4.6529447115384614e-05, "loss": 2.241, "step": 1850 }, { "epoch": 0.56, "learning_rate": 4.6527569110576926e-05, "loss": 2.1744, "step": 1851 }, { "epoch": 0.56, "learning_rate": 4.652569110576923e-05, "loss": 2.1002, "step": 1852 }, { "epoch": 0.56, "learning_rate": 4.6523813100961535e-05, "loss": 2.2556, "step": 1853 }, { "epoch": 0.56, "learning_rate": 4.652193509615385e-05, "loss": 2.1229, "step": 1854 }, { "epoch": 0.56, "learning_rate": 4.652005709134616e-05, "loss": 1.8335, "step": 1855 }, { "epoch": 0.56, "learning_rate": 4.651817908653846e-05, "loss": 2.2277, "step": 1856 }, { "epoch": 0.56, "learning_rate": 4.6516301081730775e-05, "loss": 2.117, "step": 1857 }, { "epoch": 0.56, "learning_rate": 4.651442307692308e-05, "loss": 2.2216, "step": 1858 }, { "epoch": 0.56, "learning_rate": 4.6512545072115384e-05, "loss": 2.1182, "step": 1859 }, { "epoch": 0.56, "learning_rate": 4.6510667067307696e-05, "loss": 2.075, "step": 1860 }, { "epoch": 0.56, "learning_rate": 4.65087890625e-05, "loss": 2.0923, "step": 1861 }, { "epoch": 0.56, "learning_rate": 4.650691105769231e-05, "loss": 2.1201, "step": 1862 }, { "epoch": 0.56, "learning_rate": 4.650503305288462e-05, "loss": 2.4418, "step": 1863 }, { "epoch": 0.56, "learning_rate": 4.650315504807692e-05, "loss": 2.2874, "step": 1864 }, { "epoch": 0.56, "learning_rate": 4.650127704326923e-05, "loss": 2.1476, "step": 1865 }, { "epoch": 0.56, "learning_rate": 4.649939903846154e-05, "loss": 2.209, "step": 1866 }, { "epoch": 0.56, "learning_rate": 4.649752103365384e-05, "loss": 2.3308, "step": 1867 }, { "epoch": 0.56, "learning_rate": 4.649564302884616e-05, "loss": 2.2165, "step": 1868 }, { "epoch": 0.56, "learning_rate": 4.6493765024038466e-05, "loss": 2.0637, "step": 1869 }, { "epoch": 0.56, "learning_rate": 4.649188701923077e-05, "loss": 2.1647, "step": 1870 }, { "epoch": 0.56, "learning_rate": 4.649000901442308e-05, "loss": 1.9537, "step": 1871 }, { "epoch": 0.56, "learning_rate": 4.648813100961539e-05, "loss": 2.2863, "step": 1872 }, { "epoch": 0.56, "learning_rate": 4.648625300480769e-05, "loss": 2.1892, "step": 1873 }, { "epoch": 0.56, "learning_rate": 4.6484375e-05, "loss": 2.1609, "step": 1874 }, { "epoch": 0.56, "learning_rate": 4.648249699519231e-05, "loss": 2.1707, "step": 1875 }, { "epoch": 0.56, "learning_rate": 4.648061899038462e-05, "loss": 2.1423, "step": 1876 }, { "epoch": 0.56, "learning_rate": 4.6478740985576924e-05, "loss": 2.3107, "step": 1877 }, { "epoch": 0.56, "learning_rate": 4.647686298076923e-05, "loss": 2.2444, "step": 1878 }, { "epoch": 0.56, "learning_rate": 4.647498497596154e-05, "loss": 2.3155, "step": 1879 }, { "epoch": 0.56, "learning_rate": 4.6473106971153846e-05, "loss": 2.153, "step": 1880 }, { "epoch": 0.57, "learning_rate": 4.647122896634616e-05, "loss": 2.4449, "step": 1881 }, { "epoch": 0.57, "learning_rate": 4.646935096153847e-05, "loss": 1.8569, "step": 1882 }, { "epoch": 0.57, "learning_rate": 4.6467472956730773e-05, "loss": 1.9722, "step": 1883 }, { "epoch": 0.57, "learning_rate": 4.646559495192308e-05, "loss": 2.0328, "step": 1884 }, { "epoch": 0.57, "learning_rate": 4.646371694711539e-05, "loss": 1.9308, "step": 1885 }, { "epoch": 0.57, "learning_rate": 4.6461838942307694e-05, "loss": 2.2006, "step": 1886 }, { "epoch": 0.57, "learning_rate": 4.64599609375e-05, "loss": 2.1448, "step": 1887 }, { "epoch": 0.57, "learning_rate": 4.645808293269231e-05, "loss": 1.9086, "step": 1888 }, { "epoch": 0.57, "learning_rate": 4.6456204927884616e-05, "loss": 2.1368, "step": 1889 }, { "epoch": 0.57, "learning_rate": 4.645432692307693e-05, "loss": 2.1161, "step": 1890 }, { "epoch": 0.57, "learning_rate": 4.645244891826923e-05, "loss": 2.3298, "step": 1891 }, { "epoch": 0.57, "learning_rate": 4.645057091346154e-05, "loss": 2.3055, "step": 1892 }, { "epoch": 0.57, "learning_rate": 4.644869290865385e-05, "loss": 2.3129, "step": 1893 }, { "epoch": 0.57, "learning_rate": 4.644681490384616e-05, "loss": 2.0372, "step": 1894 }, { "epoch": 0.57, "learning_rate": 4.6444936899038465e-05, "loss": 2.2297, "step": 1895 }, { "epoch": 0.57, "learning_rate": 4.6443058894230776e-05, "loss": 2.332, "step": 1896 }, { "epoch": 0.57, "learning_rate": 4.644118088942308e-05, "loss": 2.1868, "step": 1897 }, { "epoch": 0.57, "learning_rate": 4.6439302884615386e-05, "loss": 2.2229, "step": 1898 }, { "epoch": 0.57, "learning_rate": 4.64374248798077e-05, "loss": 2.4402, "step": 1899 }, { "epoch": 0.57, "learning_rate": 4.6435546875e-05, "loss": 2.3413, "step": 1900 }, { "epoch": 0.57, "learning_rate": 4.643366887019231e-05, "loss": 2.1974, "step": 1901 }, { "epoch": 0.57, "learning_rate": 4.643179086538462e-05, "loss": 2.1882, "step": 1902 }, { "epoch": 0.57, "learning_rate": 4.642991286057692e-05, "loss": 2.224, "step": 1903 }, { "epoch": 0.57, "learning_rate": 4.6428034855769235e-05, "loss": 2.0951, "step": 1904 }, { "epoch": 0.57, "learning_rate": 4.642615685096154e-05, "loss": 2.2133, "step": 1905 }, { "epoch": 0.57, "learning_rate": 4.6424278846153844e-05, "loss": 2.3146, "step": 1906 }, { "epoch": 0.57, "learning_rate": 4.6422400841346156e-05, "loss": 2.136, "step": 1907 }, { "epoch": 0.57, "learning_rate": 4.642052283653847e-05, "loss": 2.1227, "step": 1908 }, { "epoch": 0.57, "learning_rate": 4.641864483173077e-05, "loss": 2.24, "step": 1909 }, { "epoch": 0.57, "learning_rate": 4.6416766826923084e-05, "loss": 2.3091, "step": 1910 }, { "epoch": 0.57, "learning_rate": 4.641488882211539e-05, "loss": 2.2995, "step": 1911 }, { "epoch": 0.57, "learning_rate": 4.641301081730769e-05, "loss": 2.4242, "step": 1912 }, { "epoch": 0.57, "learning_rate": 4.6411132812500005e-05, "loss": 2.0721, "step": 1913 }, { "epoch": 0.58, "learning_rate": 4.640925480769231e-05, "loss": 1.9942, "step": 1914 }, { "epoch": 0.58, "learning_rate": 4.6407376802884614e-05, "loss": 2.2269, "step": 1915 }, { "epoch": 0.58, "learning_rate": 4.6405498798076926e-05, "loss": 2.4068, "step": 1916 }, { "epoch": 0.58, "learning_rate": 4.640362079326923e-05, "loss": 2.2335, "step": 1917 }, { "epoch": 0.58, "learning_rate": 4.6401742788461535e-05, "loss": 2.4289, "step": 1918 }, { "epoch": 0.58, "learning_rate": 4.639986478365385e-05, "loss": 2.4663, "step": 1919 }, { "epoch": 0.58, "learning_rate": 4.639798677884616e-05, "loss": 2.3424, "step": 1920 }, { "epoch": 0.58, "learning_rate": 4.639610877403846e-05, "loss": 2.1179, "step": 1921 }, { "epoch": 0.58, "learning_rate": 4.6394230769230775e-05, "loss": 2.0111, "step": 1922 }, { "epoch": 0.58, "learning_rate": 4.639235276442308e-05, "loss": 2.2195, "step": 1923 }, { "epoch": 0.58, "learning_rate": 4.639047475961539e-05, "loss": 2.4005, "step": 1924 }, { "epoch": 0.58, "learning_rate": 4.6388596754807696e-05, "loss": 1.9543, "step": 1925 }, { "epoch": 0.58, "learning_rate": 4.638671875e-05, "loss": 2.2673, "step": 1926 }, { "epoch": 0.58, "learning_rate": 4.638484074519231e-05, "loss": 2.3154, "step": 1927 }, { "epoch": 0.58, "learning_rate": 4.638296274038462e-05, "loss": 2.2486, "step": 1928 }, { "epoch": 0.58, "learning_rate": 4.638108473557692e-05, "loss": 2.3896, "step": 1929 }, { "epoch": 0.58, "learning_rate": 4.637920673076923e-05, "loss": 2.203, "step": 1930 }, { "epoch": 0.58, "learning_rate": 4.637732872596154e-05, "loss": 2.3127, "step": 1931 }, { "epoch": 0.58, "learning_rate": 4.637545072115384e-05, "loss": 2.1525, "step": 1932 }, { "epoch": 0.58, "learning_rate": 4.637357271634616e-05, "loss": 2.1924, "step": 1933 }, { "epoch": 0.58, "learning_rate": 4.6371694711538466e-05, "loss": 2.1287, "step": 1934 }, { "epoch": 0.58, "learning_rate": 4.636981670673077e-05, "loss": 2.2175, "step": 1935 }, { "epoch": 0.58, "learning_rate": 4.636793870192308e-05, "loss": 2.4702, "step": 1936 }, { "epoch": 0.58, "learning_rate": 4.636606069711539e-05, "loss": 2.2149, "step": 1937 }, { "epoch": 0.58, "learning_rate": 4.636418269230769e-05, "loss": 2.2648, "step": 1938 }, { "epoch": 0.58, "learning_rate": 4.63623046875e-05, "loss": 2.4737, "step": 1939 }, { "epoch": 0.58, "learning_rate": 4.636042668269231e-05, "loss": 1.9771, "step": 1940 }, { "epoch": 0.58, "learning_rate": 4.635854867788462e-05, "loss": 2.4398, "step": 1941 }, { "epoch": 0.58, "learning_rate": 4.6356670673076924e-05, "loss": 2.2824, "step": 1942 }, { "epoch": 0.58, "learning_rate": 4.635479266826923e-05, "loss": 2.2791, "step": 1943 }, { "epoch": 0.58, "learning_rate": 4.635291466346154e-05, "loss": 2.1666, "step": 1944 }, { "epoch": 0.58, "learning_rate": 4.6351036658653845e-05, "loss": 1.9779, "step": 1945 }, { "epoch": 0.58, "learning_rate": 4.634915865384616e-05, "loss": 2.073, "step": 1946 }, { "epoch": 0.59, "learning_rate": 4.634728064903847e-05, "loss": 2.3079, "step": 1947 }, { "epoch": 0.59, "learning_rate": 4.634540264423077e-05, "loss": 2.139, "step": 1948 }, { "epoch": 0.59, "learning_rate": 4.634352463942308e-05, "loss": 2.4321, "step": 1949 }, { "epoch": 0.59, "learning_rate": 4.634164663461539e-05, "loss": 2.1256, "step": 1950 }, { "epoch": 0.59, "learning_rate": 4.6339768629807694e-05, "loss": 2.0319, "step": 1951 }, { "epoch": 0.59, "learning_rate": 4.6337890625e-05, "loss": 2.0736, "step": 1952 }, { "epoch": 0.59, "learning_rate": 4.633601262019231e-05, "loss": 2.089, "step": 1953 }, { "epoch": 0.59, "learning_rate": 4.6334134615384615e-05, "loss": 2.1087, "step": 1954 }, { "epoch": 0.59, "learning_rate": 4.633225661057693e-05, "loss": 2.4038, "step": 1955 }, { "epoch": 0.59, "learning_rate": 4.633037860576923e-05, "loss": 2.4887, "step": 1956 }, { "epoch": 0.59, "learning_rate": 4.6328500600961537e-05, "loss": 2.1058, "step": 1957 }, { "epoch": 0.59, "learning_rate": 4.632662259615385e-05, "loss": 2.1672, "step": 1958 }, { "epoch": 0.59, "learning_rate": 4.632474459134616e-05, "loss": 2.0989, "step": 1959 }, { "epoch": 0.59, "learning_rate": 4.6322866586538464e-05, "loss": 2.1138, "step": 1960 }, { "epoch": 0.59, "learning_rate": 4.6320988581730776e-05, "loss": 2.167, "step": 1961 }, { "epoch": 0.59, "learning_rate": 4.631911057692308e-05, "loss": 2.056, "step": 1962 }, { "epoch": 0.59, "learning_rate": 4.6317232572115386e-05, "loss": 2.3263, "step": 1963 }, { "epoch": 0.59, "learning_rate": 4.63153545673077e-05, "loss": 2.2523, "step": 1964 }, { "epoch": 0.59, "learning_rate": 4.63134765625e-05, "loss": 2.1006, "step": 1965 }, { "epoch": 0.59, "learning_rate": 4.6311598557692307e-05, "loss": 2.0967, "step": 1966 }, { "epoch": 0.59, "learning_rate": 4.630972055288462e-05, "loss": 2.382, "step": 1967 }, { "epoch": 0.59, "learning_rate": 4.630784254807692e-05, "loss": 2.2222, "step": 1968 }, { "epoch": 0.59, "learning_rate": 4.6305964543269234e-05, "loss": 2.3319, "step": 1969 }, { "epoch": 0.59, "learning_rate": 4.630408653846154e-05, "loss": 2.4923, "step": 1970 }, { "epoch": 0.59, "learning_rate": 4.6302208533653844e-05, "loss": 2.3516, "step": 1971 }, { "epoch": 0.59, "learning_rate": 4.6300330528846156e-05, "loss": 2.2525, "step": 1972 }, { "epoch": 0.59, "learning_rate": 4.629845252403847e-05, "loss": 2.353, "step": 1973 }, { "epoch": 0.59, "learning_rate": 4.629657451923077e-05, "loss": 2.1641, "step": 1974 }, { "epoch": 0.59, "learning_rate": 4.6294696514423083e-05, "loss": 2.2761, "step": 1975 }, { "epoch": 0.59, "learning_rate": 4.629281850961539e-05, "loss": 2.0865, "step": 1976 }, { "epoch": 0.59, "learning_rate": 4.629094050480769e-05, "loss": 2.2239, "step": 1977 }, { "epoch": 0.59, "learning_rate": 4.6289062500000005e-05, "loss": 2.4412, "step": 1978 }, { "epoch": 0.59, "learning_rate": 4.628718449519231e-05, "loss": 2.3833, "step": 1979 }, { "epoch": 0.59, "learning_rate": 4.6285306490384614e-05, "loss": 1.9886, "step": 1980 }, { "epoch": 0.6, "learning_rate": 4.6283428485576926e-05, "loss": 2.1424, "step": 1981 }, { "epoch": 0.6, "learning_rate": 4.628155048076923e-05, "loss": 2.483, "step": 1982 }, { "epoch": 0.6, "learning_rate": 4.627967247596154e-05, "loss": 2.1889, "step": 1983 }, { "epoch": 0.6, "learning_rate": 4.627779447115385e-05, "loss": 2.1599, "step": 1984 }, { "epoch": 0.6, "learning_rate": 4.627591646634616e-05, "loss": 2.219, "step": 1985 }, { "epoch": 0.6, "learning_rate": 4.627403846153846e-05, "loss": 2.3488, "step": 1986 }, { "epoch": 0.6, "learning_rate": 4.6272160456730775e-05, "loss": 2.4368, "step": 1987 }, { "epoch": 0.6, "learning_rate": 4.627028245192308e-05, "loss": 2.2701, "step": 1988 }, { "epoch": 0.6, "learning_rate": 4.626840444711539e-05, "loss": 2.0566, "step": 1989 }, { "epoch": 0.6, "learning_rate": 4.6266526442307696e-05, "loss": 2.1964, "step": 1990 }, { "epoch": 0.6, "learning_rate": 4.62646484375e-05, "loss": 2.2665, "step": 1991 }, { "epoch": 0.6, "learning_rate": 4.626277043269231e-05, "loss": 2.1104, "step": 1992 }, { "epoch": 0.6, "learning_rate": 4.626089242788462e-05, "loss": 2.21, "step": 1993 }, { "epoch": 0.6, "learning_rate": 4.625901442307692e-05, "loss": 2.1843, "step": 1994 }, { "epoch": 0.6, "learning_rate": 4.625713641826923e-05, "loss": 1.9875, "step": 1995 }, { "epoch": 0.6, "learning_rate": 4.625525841346154e-05, "loss": 2.0898, "step": 1996 }, { "epoch": 0.6, "learning_rate": 4.625338040865384e-05, "loss": 2.4379, "step": 1997 }, { "epoch": 0.6, "learning_rate": 4.625150240384616e-05, "loss": 2.4372, "step": 1998 }, { "epoch": 0.6, "learning_rate": 4.6249624399038466e-05, "loss": 1.9789, "step": 1999 }, { "epoch": 0.6, "learning_rate": 4.624774639423077e-05, "loss": 1.8832, "step": 2000 }, { "epoch": 0.6, "learning_rate": 4.624586838942308e-05, "loss": 2.2031, "step": 2001 }, { "epoch": 0.6, "learning_rate": 4.624399038461539e-05, "loss": 2.0986, "step": 2002 }, { "epoch": 0.6, "learning_rate": 4.624211237980769e-05, "loss": 2.2991, "step": 2003 }, { "epoch": 0.6, "learning_rate": 4.6240234375e-05, "loss": 2.1609, "step": 2004 }, { "epoch": 0.6, "learning_rate": 4.623835637019231e-05, "loss": 2.0858, "step": 2005 }, { "epoch": 0.6, "learning_rate": 4.623647836538462e-05, "loss": 2.0543, "step": 2006 }, { "epoch": 0.6, "learning_rate": 4.6234600360576924e-05, "loss": 2.2404, "step": 2007 }, { "epoch": 0.6, "learning_rate": 4.623272235576923e-05, "loss": 2.2519, "step": 2008 }, { "epoch": 0.6, "learning_rate": 4.623084435096154e-05, "loss": 2.1049, "step": 2009 }, { "epoch": 0.6, "learning_rate": 4.6228966346153845e-05, "loss": 2.1126, "step": 2010 }, { "epoch": 0.6, "learning_rate": 4.622708834134616e-05, "loss": 1.8826, "step": 2011 }, { "epoch": 0.6, "learning_rate": 4.622521033653847e-05, "loss": 2.2502, "step": 2012 }, { "epoch": 0.6, "learning_rate": 4.622333233173077e-05, "loss": 2.1901, "step": 2013 }, { "epoch": 0.61, "learning_rate": 4.622145432692308e-05, "loss": 2.0124, "step": 2014 }, { "epoch": 0.61, "learning_rate": 4.621957632211539e-05, "loss": 2.2213, "step": 2015 }, { "epoch": 0.61, "learning_rate": 4.6217698317307694e-05, "loss": 2.0446, "step": 2016 }, { "epoch": 0.61, "learning_rate": 4.62158203125e-05, "loss": 2.347, "step": 2017 }, { "epoch": 0.61, "learning_rate": 4.621394230769231e-05, "loss": 2.2271, "step": 2018 }, { "epoch": 0.61, "learning_rate": 4.6212064302884615e-05, "loss": 2.3321, "step": 2019 }, { "epoch": 0.61, "learning_rate": 4.621018629807693e-05, "loss": 2.206, "step": 2020 }, { "epoch": 0.61, "learning_rate": 4.620830829326923e-05, "loss": 2.4361, "step": 2021 }, { "epoch": 0.61, "learning_rate": 4.6206430288461536e-05, "loss": 2.2835, "step": 2022 }, { "epoch": 0.61, "learning_rate": 4.620455228365385e-05, "loss": 2.2119, "step": 2023 }, { "epoch": 0.61, "learning_rate": 4.620267427884616e-05, "loss": 1.9586, "step": 2024 }, { "epoch": 0.61, "learning_rate": 4.6200796274038464e-05, "loss": 2.3926, "step": 2025 }, { "epoch": 0.61, "learning_rate": 4.6198918269230776e-05, "loss": 2.3022, "step": 2026 }, { "epoch": 0.61, "learning_rate": 4.619704026442308e-05, "loss": 2.0861, "step": 2027 }, { "epoch": 0.61, "learning_rate": 4.6195162259615385e-05, "loss": 1.9465, "step": 2028 }, { "epoch": 0.61, "learning_rate": 4.61932842548077e-05, "loss": 2.2847, "step": 2029 }, { "epoch": 0.61, "learning_rate": 4.619140625e-05, "loss": 2.1555, "step": 2030 }, { "epoch": 0.61, "learning_rate": 4.6189528245192306e-05, "loss": 1.8872, "step": 2031 }, { "epoch": 0.61, "learning_rate": 4.618765024038462e-05, "loss": 1.9925, "step": 2032 }, { "epoch": 0.61, "learning_rate": 4.618577223557692e-05, "loss": 2.3355, "step": 2033 }, { "epoch": 0.61, "learning_rate": 4.6183894230769234e-05, "loss": 2.1781, "step": 2034 }, { "epoch": 0.61, "learning_rate": 4.618201622596154e-05, "loss": 2.057, "step": 2035 }, { "epoch": 0.61, "learning_rate": 4.6180138221153844e-05, "loss": 2.1581, "step": 2036 }, { "epoch": 0.61, "learning_rate": 4.6178260216346155e-05, "loss": 2.031, "step": 2037 }, { "epoch": 0.61, "learning_rate": 4.617638221153847e-05, "loss": 2.1074, "step": 2038 }, { "epoch": 0.61, "learning_rate": 4.617450420673077e-05, "loss": 2.1693, "step": 2039 }, { "epoch": 0.61, "learning_rate": 4.617262620192308e-05, "loss": 1.9207, "step": 2040 }, { "epoch": 0.61, "learning_rate": 4.617074819711539e-05, "loss": 2.3931, "step": 2041 }, { "epoch": 0.61, "learning_rate": 4.616887019230769e-05, "loss": 2.2171, "step": 2042 }, { "epoch": 0.61, "learning_rate": 4.6166992187500004e-05, "loss": 2.6998, "step": 2043 }, { "epoch": 0.61, "learning_rate": 4.616511418269231e-05, "loss": 2.1855, "step": 2044 }, { "epoch": 0.61, "learning_rate": 4.6163236177884614e-05, "loss": 2.0861, "step": 2045 }, { "epoch": 0.61, "learning_rate": 4.6161358173076926e-05, "loss": 2.2932, "step": 2046 }, { "epoch": 0.62, "learning_rate": 4.615948016826923e-05, "loss": 2.2261, "step": 2047 }, { "epoch": 0.62, "learning_rate": 4.615760216346154e-05, "loss": 2.1009, "step": 2048 }, { "epoch": 0.62, "learning_rate": 4.6155724158653847e-05, "loss": 2.2688, "step": 2049 }, { "epoch": 0.62, "learning_rate": 4.615384615384616e-05, "loss": 2.2703, "step": 2050 }, { "epoch": 0.62, "learning_rate": 4.615196814903846e-05, "loss": 2.2154, "step": 2051 }, { "epoch": 0.62, "learning_rate": 4.6150090144230774e-05, "loss": 2.1372, "step": 2052 }, { "epoch": 0.62, "learning_rate": 4.614821213942308e-05, "loss": 2.2615, "step": 2053 }, { "epoch": 0.62, "learning_rate": 4.614633413461539e-05, "loss": 2.1242, "step": 2054 }, { "epoch": 0.62, "learning_rate": 4.6144456129807696e-05, "loss": 2.2625, "step": 2055 }, { "epoch": 0.62, "learning_rate": 4.6142578125e-05, "loss": 2.2138, "step": 2056 }, { "epoch": 0.62, "learning_rate": 4.614070012019231e-05, "loss": 2.2241, "step": 2057 }, { "epoch": 0.62, "learning_rate": 4.613882211538462e-05, "loss": 2.0521, "step": 2058 }, { "epoch": 0.62, "learning_rate": 4.613694411057692e-05, "loss": 2.2978, "step": 2059 }, { "epoch": 0.62, "learning_rate": 4.613506610576923e-05, "loss": 2.0771, "step": 2060 }, { "epoch": 0.62, "learning_rate": 4.613318810096154e-05, "loss": 2.0018, "step": 2061 }, { "epoch": 0.62, "learning_rate": 4.613131009615384e-05, "loss": 2.261, "step": 2062 }, { "epoch": 0.62, "learning_rate": 4.612943209134616e-05, "loss": 2.3031, "step": 2063 }, { "epoch": 0.62, "learning_rate": 4.6127554086538466e-05, "loss": 2.1859, "step": 2064 }, { "epoch": 0.62, "learning_rate": 4.612567608173077e-05, "loss": 1.9541, "step": 2065 }, { "epoch": 0.62, "learning_rate": 4.612379807692308e-05, "loss": 2.2851, "step": 2066 }, { "epoch": 0.62, "learning_rate": 4.612192007211539e-05, "loss": 2.2643, "step": 2067 }, { "epoch": 0.62, "learning_rate": 4.61200420673077e-05, "loss": 2.2086, "step": 2068 }, { "epoch": 0.62, "learning_rate": 4.61181640625e-05, "loss": 2.0494, "step": 2069 }, { "epoch": 0.62, "learning_rate": 4.611628605769231e-05, "loss": 2.0862, "step": 2070 }, { "epoch": 0.62, "learning_rate": 4.611440805288462e-05, "loss": 2.0249, "step": 2071 }, { "epoch": 0.62, "learning_rate": 4.6112530048076924e-05, "loss": 2.1388, "step": 2072 }, { "epoch": 0.62, "learning_rate": 4.611065204326923e-05, "loss": 1.9094, "step": 2073 }, { "epoch": 0.62, "learning_rate": 4.610877403846154e-05, "loss": 2.198, "step": 2074 }, { "epoch": 0.62, "learning_rate": 4.6106896033653845e-05, "loss": 2.3114, "step": 2075 }, { "epoch": 0.62, "learning_rate": 4.610501802884616e-05, "loss": 2.1082, "step": 2076 }, { "epoch": 0.62, "learning_rate": 4.610314002403847e-05, "loss": 2.1917, "step": 2077 }, { "epoch": 0.62, "learning_rate": 4.610126201923077e-05, "loss": 2.0176, "step": 2078 }, { "epoch": 0.62, "learning_rate": 4.609938401442308e-05, "loss": 2.39, "step": 2079 }, { "epoch": 0.62, "learning_rate": 4.609750600961539e-05, "loss": 2.2185, "step": 2080 }, { "epoch": 0.63, "learning_rate": 4.6095628004807694e-05, "loss": 1.9478, "step": 2081 }, { "epoch": 0.63, "learning_rate": 4.609375e-05, "loss": 1.9783, "step": 2082 }, { "epoch": 0.63, "learning_rate": 4.609187199519231e-05, "loss": 2.2243, "step": 2083 }, { "epoch": 0.63, "learning_rate": 4.6089993990384615e-05, "loss": 2.2544, "step": 2084 }, { "epoch": 0.63, "learning_rate": 4.608811598557693e-05, "loss": 2.2609, "step": 2085 }, { "epoch": 0.63, "learning_rate": 4.608623798076923e-05, "loss": 2.1309, "step": 2086 }, { "epoch": 0.63, "learning_rate": 4.6084359975961536e-05, "loss": 1.8668, "step": 2087 }, { "epoch": 0.63, "learning_rate": 4.608248197115385e-05, "loss": 2.1871, "step": 2088 }, { "epoch": 0.63, "learning_rate": 4.608060396634616e-05, "loss": 2.206, "step": 2089 }, { "epoch": 0.63, "learning_rate": 4.6078725961538464e-05, "loss": 2.2039, "step": 2090 }, { "epoch": 0.63, "learning_rate": 4.6076847956730776e-05, "loss": 2.3424, "step": 2091 }, { "epoch": 0.63, "learning_rate": 4.607496995192308e-05, "loss": 2.2068, "step": 2092 }, { "epoch": 0.63, "learning_rate": 4.6073091947115385e-05, "loss": 2.2524, "step": 2093 }, { "epoch": 0.63, "learning_rate": 4.60712139423077e-05, "loss": 2.1306, "step": 2094 }, { "epoch": 0.63, "learning_rate": 4.60693359375e-05, "loss": 2.1398, "step": 2095 }, { "epoch": 0.63, "learning_rate": 4.6067457932692306e-05, "loss": 2.1991, "step": 2096 }, { "epoch": 0.63, "learning_rate": 4.606557992788462e-05, "loss": 2.3777, "step": 2097 }, { "epoch": 0.63, "learning_rate": 4.606370192307692e-05, "loss": 2.1998, "step": 2098 }, { "epoch": 0.63, "learning_rate": 4.6061823918269234e-05, "loss": 2.9372, "step": 2099 }, { "epoch": 0.63, "learning_rate": 4.605994591346154e-05, "loss": 2.3911, "step": 2100 }, { "epoch": 0.63, "learning_rate": 4.6058067908653844e-05, "loss": 2.1962, "step": 2101 }, { "epoch": 0.63, "learning_rate": 4.6056189903846155e-05, "loss": 2.3645, "step": 2102 }, { "epoch": 0.63, "learning_rate": 4.605431189903847e-05, "loss": 2.5316, "step": 2103 }, { "epoch": 0.63, "learning_rate": 4.605243389423077e-05, "loss": 2.3585, "step": 2104 }, { "epoch": 0.63, "learning_rate": 4.605055588942308e-05, "loss": 2.0114, "step": 2105 }, { "epoch": 0.63, "learning_rate": 4.604867788461539e-05, "loss": 2.3478, "step": 2106 }, { "epoch": 0.63, "learning_rate": 4.604679987980769e-05, "loss": 2.1891, "step": 2107 }, { "epoch": 0.63, "learning_rate": 4.6044921875000004e-05, "loss": 2.3808, "step": 2108 }, { "epoch": 0.63, "learning_rate": 4.604304387019231e-05, "loss": 1.9891, "step": 2109 }, { "epoch": 0.63, "learning_rate": 4.6041165865384614e-05, "loss": 2.3784, "step": 2110 }, { "epoch": 0.63, "learning_rate": 4.6039287860576925e-05, "loss": 1.9207, "step": 2111 }, { "epoch": 0.63, "learning_rate": 4.603740985576923e-05, "loss": 2.0892, "step": 2112 }, { "epoch": 0.63, "learning_rate": 4.603553185096154e-05, "loss": 2.3575, "step": 2113 }, { "epoch": 0.64, "learning_rate": 4.6033653846153846e-05, "loss": 2.0631, "step": 2114 }, { "epoch": 0.64, "learning_rate": 4.603177584134616e-05, "loss": 2.2041, "step": 2115 }, { "epoch": 0.64, "learning_rate": 4.602989783653846e-05, "loss": 2.3525, "step": 2116 }, { "epoch": 0.64, "learning_rate": 4.6028019831730774e-05, "loss": 2.243, "step": 2117 }, { "epoch": 0.64, "learning_rate": 4.602614182692308e-05, "loss": 2.3192, "step": 2118 }, { "epoch": 0.64, "learning_rate": 4.602426382211539e-05, "loss": 2.3048, "step": 2119 }, { "epoch": 0.64, "learning_rate": 4.6022385817307695e-05, "loss": 1.9988, "step": 2120 }, { "epoch": 0.64, "learning_rate": 4.60205078125e-05, "loss": 2.292, "step": 2121 }, { "epoch": 0.64, "learning_rate": 4.601862980769231e-05, "loss": 2.3679, "step": 2122 }, { "epoch": 0.64, "learning_rate": 4.6016751802884617e-05, "loss": 2.3745, "step": 2123 }, { "epoch": 0.64, "learning_rate": 4.601487379807692e-05, "loss": 2.125, "step": 2124 }, { "epoch": 0.64, "learning_rate": 4.601299579326923e-05, "loss": 2.2385, "step": 2125 }, { "epoch": 0.64, "learning_rate": 4.601111778846154e-05, "loss": 2.1458, "step": 2126 }, { "epoch": 0.64, "learning_rate": 4.600923978365385e-05, "loss": 2.2784, "step": 2127 }, { "epoch": 0.64, "learning_rate": 4.600736177884616e-05, "loss": 2.1453, "step": 2128 }, { "epoch": 0.64, "learning_rate": 4.6005483774038465e-05, "loss": 2.1141, "step": 2129 }, { "epoch": 0.64, "learning_rate": 4.600360576923077e-05, "loss": 2.072, "step": 2130 }, { "epoch": 0.64, "learning_rate": 4.600172776442308e-05, "loss": 2.1463, "step": 2131 }, { "epoch": 0.64, "learning_rate": 4.5999849759615387e-05, "loss": 2.2591, "step": 2132 }, { "epoch": 0.64, "learning_rate": 4.59979717548077e-05, "loss": 1.9883, "step": 2133 }, { "epoch": 0.64, "learning_rate": 4.599609375e-05, "loss": 2.1679, "step": 2134 }, { "epoch": 0.64, "learning_rate": 4.599421574519231e-05, "loss": 2.271, "step": 2135 }, { "epoch": 0.64, "learning_rate": 4.599233774038462e-05, "loss": 2.1089, "step": 2136 }, { "epoch": 0.64, "learning_rate": 4.5990459735576924e-05, "loss": 2.2647, "step": 2137 }, { "epoch": 0.64, "learning_rate": 4.598858173076923e-05, "loss": 2.3278, "step": 2138 }, { "epoch": 0.64, "learning_rate": 4.598670372596154e-05, "loss": 2.0308, "step": 2139 }, { "epoch": 0.64, "learning_rate": 4.5984825721153845e-05, "loss": 2.0261, "step": 2140 }, { "epoch": 0.64, "learning_rate": 4.598294771634616e-05, "loss": 2.1196, "step": 2141 }, { "epoch": 0.64, "learning_rate": 4.598106971153847e-05, "loss": 2.4086, "step": 2142 }, { "epoch": 0.64, "learning_rate": 4.597919170673077e-05, "loss": 2.0834, "step": 2143 }, { "epoch": 0.64, "learning_rate": 4.597731370192308e-05, "loss": 2.2271, "step": 2144 }, { "epoch": 0.64, "learning_rate": 4.597543569711539e-05, "loss": 2.2276, "step": 2145 }, { "epoch": 0.64, "learning_rate": 4.5973557692307694e-05, "loss": 2.1262, "step": 2146 }, { "epoch": 0.65, "learning_rate": 4.59716796875e-05, "loss": 2.2248, "step": 2147 }, { "epoch": 0.65, "learning_rate": 4.596980168269231e-05, "loss": 1.896, "step": 2148 }, { "epoch": 0.65, "learning_rate": 4.5967923677884615e-05, "loss": 2.2368, "step": 2149 }, { "epoch": 0.65, "learning_rate": 4.596604567307693e-05, "loss": 2.1691, "step": 2150 }, { "epoch": 0.65, "learning_rate": 4.596416766826923e-05, "loss": 1.8367, "step": 2151 }, { "epoch": 0.65, "learning_rate": 4.5962289663461536e-05, "loss": 2.694, "step": 2152 }, { "epoch": 0.65, "learning_rate": 4.596041165865385e-05, "loss": 1.8543, "step": 2153 }, { "epoch": 0.65, "learning_rate": 4.595853365384616e-05, "loss": 2.1504, "step": 2154 }, { "epoch": 0.65, "learning_rate": 4.5956655649038464e-05, "loss": 2.1079, "step": 2155 }, { "epoch": 0.65, "learning_rate": 4.5954777644230776e-05, "loss": 2.1596, "step": 2156 }, { "epoch": 0.65, "learning_rate": 4.595289963942308e-05, "loss": 1.9662, "step": 2157 }, { "epoch": 0.65, "learning_rate": 4.5951021634615385e-05, "loss": 2.1482, "step": 2158 }, { "epoch": 0.65, "learning_rate": 4.59491436298077e-05, "loss": 1.8711, "step": 2159 }, { "epoch": 0.65, "learning_rate": 4.5947265625e-05, "loss": 2.0749, "step": 2160 }, { "epoch": 0.65, "learning_rate": 4.5945387620192306e-05, "loss": 2.2393, "step": 2161 }, { "epoch": 0.65, "learning_rate": 4.594350961538462e-05, "loss": 1.8353, "step": 2162 }, { "epoch": 0.65, "learning_rate": 4.594163161057692e-05, "loss": 2.2153, "step": 2163 }, { "epoch": 0.65, "learning_rate": 4.5939753605769234e-05, "loss": 2.0557, "step": 2164 }, { "epoch": 0.65, "learning_rate": 4.593787560096154e-05, "loss": 2.4084, "step": 2165 }, { "epoch": 0.65, "learning_rate": 4.5935997596153844e-05, "loss": 2.5414, "step": 2166 }, { "epoch": 0.65, "learning_rate": 4.5934119591346155e-05, "loss": 1.9585, "step": 2167 }, { "epoch": 0.65, "learning_rate": 4.593224158653847e-05, "loss": 2.2859, "step": 2168 }, { "epoch": 0.65, "learning_rate": 4.593036358173077e-05, "loss": 2.2132, "step": 2169 }, { "epoch": 0.65, "learning_rate": 4.592848557692308e-05, "loss": 2.2724, "step": 2170 }, { "epoch": 0.65, "learning_rate": 4.592660757211539e-05, "loss": 2.0909, "step": 2171 }, { "epoch": 0.65, "learning_rate": 4.592472956730769e-05, "loss": 2.2496, "step": 2172 }, { "epoch": 0.65, "learning_rate": 4.5922851562500004e-05, "loss": 2.289, "step": 2173 }, { "epoch": 0.65, "learning_rate": 4.592097355769231e-05, "loss": 2.0829, "step": 2174 }, { "epoch": 0.65, "learning_rate": 4.5919095552884614e-05, "loss": 2.154, "step": 2175 }, { "epoch": 0.65, "learning_rate": 4.5917217548076925e-05, "loss": 2.3544, "step": 2176 }, { "epoch": 0.65, "learning_rate": 4.591533954326923e-05, "loss": 2.0438, "step": 2177 }, { "epoch": 0.65, "learning_rate": 4.591346153846154e-05, "loss": 2.306, "step": 2178 }, { "epoch": 0.65, "learning_rate": 4.5911583533653846e-05, "loss": 1.7095, "step": 2179 }, { "epoch": 0.66, "learning_rate": 4.590970552884616e-05, "loss": 2.2078, "step": 2180 }, { "epoch": 0.66, "learning_rate": 4.590782752403846e-05, "loss": 2.1187, "step": 2181 }, { "epoch": 0.66, "learning_rate": 4.5905949519230774e-05, "loss": 1.9365, "step": 2182 }, { "epoch": 0.66, "learning_rate": 4.590407151442308e-05, "loss": 2.2593, "step": 2183 }, { "epoch": 0.66, "learning_rate": 4.590219350961539e-05, "loss": 2.3596, "step": 2184 }, { "epoch": 0.66, "learning_rate": 4.5900315504807695e-05, "loss": 2.1614, "step": 2185 }, { "epoch": 0.66, "learning_rate": 4.58984375e-05, "loss": 2.1873, "step": 2186 }, { "epoch": 0.66, "learning_rate": 4.589655949519231e-05, "loss": 1.9889, "step": 2187 }, { "epoch": 0.66, "learning_rate": 4.5894681490384616e-05, "loss": 1.9978, "step": 2188 }, { "epoch": 0.66, "learning_rate": 4.589280348557692e-05, "loss": 2.216, "step": 2189 }, { "epoch": 0.66, "learning_rate": 4.589092548076923e-05, "loss": 2.1648, "step": 2190 }, { "epoch": 0.66, "learning_rate": 4.588904747596154e-05, "loss": 1.9449, "step": 2191 }, { "epoch": 0.66, "learning_rate": 4.588716947115385e-05, "loss": 2.3185, "step": 2192 }, { "epoch": 0.66, "learning_rate": 4.588529146634616e-05, "loss": 2.3361, "step": 2193 }, { "epoch": 0.66, "learning_rate": 4.5883413461538465e-05, "loss": 2.1291, "step": 2194 }, { "epoch": 0.66, "learning_rate": 4.588153545673077e-05, "loss": 2.1361, "step": 2195 }, { "epoch": 0.66, "learning_rate": 4.587965745192308e-05, "loss": 2.1719, "step": 2196 }, { "epoch": 0.66, "learning_rate": 4.5877779447115386e-05, "loss": 2.0401, "step": 2197 }, { "epoch": 0.66, "learning_rate": 4.58759014423077e-05, "loss": 2.299, "step": 2198 }, { "epoch": 0.66, "learning_rate": 4.58740234375e-05, "loss": 2.3414, "step": 2199 }, { "epoch": 0.66, "learning_rate": 4.587214543269231e-05, "loss": 2.451, "step": 2200 }, { "epoch": 0.66, "learning_rate": 4.587026742788462e-05, "loss": 2.0387, "step": 2201 }, { "epoch": 0.66, "learning_rate": 4.5868389423076924e-05, "loss": 2.1369, "step": 2202 }, { "epoch": 0.66, "learning_rate": 4.586651141826923e-05, "loss": 1.9769, "step": 2203 }, { "epoch": 0.66, "learning_rate": 4.586463341346154e-05, "loss": 2.3513, "step": 2204 }, { "epoch": 0.66, "learning_rate": 4.5862755408653845e-05, "loss": 2.1858, "step": 2205 }, { "epoch": 0.66, "learning_rate": 4.5860877403846157e-05, "loss": 2.0076, "step": 2206 }, { "epoch": 0.66, "learning_rate": 4.585899939903847e-05, "loss": 2.1544, "step": 2207 }, { "epoch": 0.66, "learning_rate": 4.585712139423077e-05, "loss": 2.0334, "step": 2208 }, { "epoch": 0.66, "learning_rate": 4.585524338942308e-05, "loss": 2.1338, "step": 2209 }, { "epoch": 0.66, "learning_rate": 4.585336538461539e-05, "loss": 2.3358, "step": 2210 }, { "epoch": 0.66, "learning_rate": 4.5851487379807694e-05, "loss": 2.3698, "step": 2211 }, { "epoch": 0.66, "learning_rate": 4.5849609375000005e-05, "loss": 2.1452, "step": 2212 }, { "epoch": 0.66, "learning_rate": 4.584773137019231e-05, "loss": 2.274, "step": 2213 }, { "epoch": 0.67, "learning_rate": 4.5845853365384615e-05, "loss": 2.2008, "step": 2214 }, { "epoch": 0.67, "learning_rate": 4.5843975360576927e-05, "loss": 2.2117, "step": 2215 }, { "epoch": 0.67, "learning_rate": 4.584209735576923e-05, "loss": 1.7634, "step": 2216 }, { "epoch": 0.67, "learning_rate": 4.5840219350961536e-05, "loss": 2.0918, "step": 2217 }, { "epoch": 0.67, "learning_rate": 4.583834134615385e-05, "loss": 2.4259, "step": 2218 }, { "epoch": 0.67, "learning_rate": 4.583646334134616e-05, "loss": 2.3356, "step": 2219 }, { "epoch": 0.67, "learning_rate": 4.5834585336538464e-05, "loss": 2.3238, "step": 2220 }, { "epoch": 0.67, "learning_rate": 4.5832707331730776e-05, "loss": 2.0058, "step": 2221 }, { "epoch": 0.67, "learning_rate": 4.583082932692308e-05, "loss": 2.2047, "step": 2222 }, { "epoch": 0.67, "learning_rate": 4.5828951322115385e-05, "loss": 2.1262, "step": 2223 }, { "epoch": 0.67, "learning_rate": 4.58270733173077e-05, "loss": 2.4232, "step": 2224 }, { "epoch": 0.67, "learning_rate": 4.58251953125e-05, "loss": 2.1089, "step": 2225 }, { "epoch": 0.67, "learning_rate": 4.5823317307692306e-05, "loss": 2.2408, "step": 2226 }, { "epoch": 0.67, "learning_rate": 4.582143930288462e-05, "loss": 2.3812, "step": 2227 }, { "epoch": 0.67, "learning_rate": 4.581956129807692e-05, "loss": 2.002, "step": 2228 }, { "epoch": 0.67, "learning_rate": 4.5817683293269234e-05, "loss": 2.2235, "step": 2229 }, { "epoch": 0.67, "learning_rate": 4.581580528846154e-05, "loss": 2.3663, "step": 2230 }, { "epoch": 0.67, "learning_rate": 4.5813927283653844e-05, "loss": 2.3935, "step": 2231 }, { "epoch": 0.67, "learning_rate": 4.581204927884616e-05, "loss": 2.2655, "step": 2232 }, { "epoch": 0.67, "learning_rate": 4.581017127403847e-05, "loss": 2.0391, "step": 2233 }, { "epoch": 0.67, "learning_rate": 4.580829326923077e-05, "loss": 1.8375, "step": 2234 }, { "epoch": 0.67, "learning_rate": 4.580641526442308e-05, "loss": 2.1466, "step": 2235 }, { "epoch": 0.67, "learning_rate": 4.580453725961539e-05, "loss": 2.103, "step": 2236 }, { "epoch": 0.67, "learning_rate": 4.580265925480769e-05, "loss": 2.2678, "step": 2237 }, { "epoch": 0.67, "learning_rate": 4.5800781250000004e-05, "loss": 2.3389, "step": 2238 }, { "epoch": 0.67, "learning_rate": 4.579890324519231e-05, "loss": 2.3036, "step": 2239 }, { "epoch": 0.67, "learning_rate": 4.5797025240384614e-05, "loss": 2.2326, "step": 2240 }, { "epoch": 0.67, "learning_rate": 4.5795147235576925e-05, "loss": 2.1369, "step": 2241 }, { "epoch": 0.67, "learning_rate": 4.579326923076923e-05, "loss": 2.3708, "step": 2242 }, { "epoch": 0.67, "learning_rate": 4.579139122596154e-05, "loss": 2.0853, "step": 2243 }, { "epoch": 0.67, "learning_rate": 4.5789513221153846e-05, "loss": 2.2073, "step": 2244 }, { "epoch": 0.67, "learning_rate": 4.578763521634616e-05, "loss": 1.9318, "step": 2245 }, { "epoch": 0.67, "learning_rate": 4.578575721153846e-05, "loss": 2.1164, "step": 2246 }, { "epoch": 0.68, "learning_rate": 4.5783879206730774e-05, "loss": 2.2843, "step": 2247 }, { "epoch": 0.68, "learning_rate": 4.578200120192308e-05, "loss": 2.0805, "step": 2248 }, { "epoch": 0.68, "learning_rate": 4.578012319711539e-05, "loss": 2.3644, "step": 2249 }, { "epoch": 0.68, "learning_rate": 4.5778245192307695e-05, "loss": 2.3068, "step": 2250 }, { "epoch": 0.68, "learning_rate": 4.57763671875e-05, "loss": 2.0701, "step": 2251 }, { "epoch": 0.68, "learning_rate": 4.577448918269231e-05, "loss": 1.9694, "step": 2252 }, { "epoch": 0.68, "learning_rate": 4.5772611177884616e-05, "loss": 2.2541, "step": 2253 }, { "epoch": 0.68, "learning_rate": 4.577073317307692e-05, "loss": 1.9863, "step": 2254 }, { "epoch": 0.68, "learning_rate": 4.576885516826923e-05, "loss": 2.4125, "step": 2255 }, { "epoch": 0.68, "learning_rate": 4.576697716346154e-05, "loss": 2.2162, "step": 2256 }, { "epoch": 0.68, "learning_rate": 4.576509915865385e-05, "loss": 1.9822, "step": 2257 }, { "epoch": 0.68, "learning_rate": 4.576322115384616e-05, "loss": 1.8193, "step": 2258 }, { "epoch": 0.68, "learning_rate": 4.5761343149038465e-05, "loss": 2.0493, "step": 2259 }, { "epoch": 0.68, "learning_rate": 4.575946514423077e-05, "loss": 1.7765, "step": 2260 }, { "epoch": 0.68, "learning_rate": 4.575758713942308e-05, "loss": 2.317, "step": 2261 }, { "epoch": 0.68, "learning_rate": 4.5755709134615386e-05, "loss": 2.3117, "step": 2262 }, { "epoch": 0.68, "learning_rate": 4.57538311298077e-05, "loss": 2.2648, "step": 2263 }, { "epoch": 0.68, "learning_rate": 4.5751953125e-05, "loss": 2.3441, "step": 2264 }, { "epoch": 0.68, "learning_rate": 4.575007512019231e-05, "loss": 2.2043, "step": 2265 }, { "epoch": 0.68, "learning_rate": 4.574819711538462e-05, "loss": 2.2917, "step": 2266 }, { "epoch": 0.68, "learning_rate": 4.5746319110576924e-05, "loss": 2.4381, "step": 2267 }, { "epoch": 0.68, "learning_rate": 4.574444110576923e-05, "loss": 2.197, "step": 2268 }, { "epoch": 0.68, "learning_rate": 4.574256310096154e-05, "loss": 2.1925, "step": 2269 }, { "epoch": 0.68, "learning_rate": 4.5740685096153845e-05, "loss": 1.792, "step": 2270 }, { "epoch": 0.68, "learning_rate": 4.5738807091346156e-05, "loss": 2.0087, "step": 2271 }, { "epoch": 0.68, "learning_rate": 4.573692908653847e-05, "loss": 1.9769, "step": 2272 }, { "epoch": 0.68, "learning_rate": 4.573505108173077e-05, "loss": 2.0891, "step": 2273 }, { "epoch": 0.68, "learning_rate": 4.573317307692308e-05, "loss": 2.0472, "step": 2274 }, { "epoch": 0.68, "learning_rate": 4.573129507211539e-05, "loss": 2.1512, "step": 2275 }, { "epoch": 0.68, "learning_rate": 4.5729417067307694e-05, "loss": 2.2402, "step": 2276 }, { "epoch": 0.68, "learning_rate": 4.5727539062500005e-05, "loss": 2.2688, "step": 2277 }, { "epoch": 0.68, "learning_rate": 4.572566105769231e-05, "loss": 2.1078, "step": 2278 }, { "epoch": 0.68, "learning_rate": 4.5723783052884615e-05, "loss": 2.0643, "step": 2279 }, { "epoch": 0.69, "learning_rate": 4.5721905048076926e-05, "loss": 2.521, "step": 2280 }, { "epoch": 0.69, "learning_rate": 4.572002704326923e-05, "loss": 1.9651, "step": 2281 }, { "epoch": 0.69, "learning_rate": 4.5718149038461536e-05, "loss": 2.0518, "step": 2282 }, { "epoch": 0.69, "learning_rate": 4.571627103365385e-05, "loss": 2.2274, "step": 2283 }, { "epoch": 0.69, "learning_rate": 4.571439302884616e-05, "loss": 2.0478, "step": 2284 }, { "epoch": 0.69, "learning_rate": 4.5712515024038464e-05, "loss": 1.976, "step": 2285 }, { "epoch": 0.69, "learning_rate": 4.5710637019230775e-05, "loss": 2.0533, "step": 2286 }, { "epoch": 0.69, "learning_rate": 4.570875901442308e-05, "loss": 2.1057, "step": 2287 }, { "epoch": 0.69, "learning_rate": 4.5706881009615385e-05, "loss": 1.9978, "step": 2288 }, { "epoch": 0.69, "learning_rate": 4.5705003004807697e-05, "loss": 2.0513, "step": 2289 }, { "epoch": 0.69, "learning_rate": 4.5703125e-05, "loss": 2.3686, "step": 2290 }, { "epoch": 0.69, "learning_rate": 4.570124699519231e-05, "loss": 2.5323, "step": 2291 }, { "epoch": 0.69, "learning_rate": 4.569936899038462e-05, "loss": 2.3991, "step": 2292 }, { "epoch": 0.69, "learning_rate": 4.569749098557692e-05, "loss": 2.2668, "step": 2293 }, { "epoch": 0.69, "learning_rate": 4.5695612980769234e-05, "loss": 2.1703, "step": 2294 }, { "epoch": 0.69, "learning_rate": 4.569373497596154e-05, "loss": 2.2626, "step": 2295 }, { "epoch": 0.69, "learning_rate": 4.5691856971153843e-05, "loss": 2.1819, "step": 2296 }, { "epoch": 0.69, "learning_rate": 4.568997896634616e-05, "loss": 2.344, "step": 2297 }, { "epoch": 0.69, "learning_rate": 4.5688100961538467e-05, "loss": 2.1198, "step": 2298 }, { "epoch": 0.69, "learning_rate": 4.568622295673077e-05, "loss": 2.2718, "step": 2299 }, { "epoch": 0.69, "learning_rate": 4.568434495192308e-05, "loss": 2.35, "step": 2300 }, { "epoch": 0.69, "learning_rate": 4.568246694711539e-05, "loss": 2.284, "step": 2301 }, { "epoch": 0.69, "learning_rate": 4.568058894230769e-05, "loss": 2.1904, "step": 2302 }, { "epoch": 0.69, "learning_rate": 4.5678710937500004e-05, "loss": 2.3601, "step": 2303 }, { "epoch": 0.69, "learning_rate": 4.567683293269231e-05, "loss": 2.2524, "step": 2304 }, { "epoch": 0.69, "learning_rate": 4.5674954927884614e-05, "loss": 2.4494, "step": 2305 }, { "epoch": 0.69, "learning_rate": 4.5673076923076925e-05, "loss": 2.1692, "step": 2306 }, { "epoch": 0.69, "learning_rate": 4.567119891826923e-05, "loss": 1.9982, "step": 2307 }, { "epoch": 0.69, "learning_rate": 4.566932091346154e-05, "loss": 2.1939, "step": 2308 }, { "epoch": 0.69, "learning_rate": 4.5667442908653846e-05, "loss": 2.3707, "step": 2309 }, { "epoch": 0.69, "learning_rate": 4.566556490384616e-05, "loss": 2.1475, "step": 2310 }, { "epoch": 0.69, "learning_rate": 4.566368689903846e-05, "loss": 2.2053, "step": 2311 }, { "epoch": 0.69, "learning_rate": 4.5661808894230774e-05, "loss": 2.1231, "step": 2312 }, { "epoch": 0.7, "learning_rate": 4.565993088942308e-05, "loss": 2.1935, "step": 2313 }, { "epoch": 0.7, "learning_rate": 4.565805288461539e-05, "loss": 2.1535, "step": 2314 }, { "epoch": 0.7, "learning_rate": 4.5656174879807695e-05, "loss": 2.3463, "step": 2315 }, { "epoch": 0.7, "learning_rate": 4.5654296875e-05, "loss": 2.1682, "step": 2316 }, { "epoch": 0.7, "learning_rate": 4.565241887019231e-05, "loss": 2.1389, "step": 2317 }, { "epoch": 0.7, "learning_rate": 4.5650540865384616e-05, "loss": 2.3076, "step": 2318 }, { "epoch": 0.7, "learning_rate": 4.564866286057692e-05, "loss": 2.1957, "step": 2319 }, { "epoch": 0.7, "learning_rate": 4.564678485576923e-05, "loss": 2.2197, "step": 2320 }, { "epoch": 0.7, "learning_rate": 4.564490685096154e-05, "loss": 2.1202, "step": 2321 }, { "epoch": 0.7, "learning_rate": 4.564302884615385e-05, "loss": 2.2354, "step": 2322 }, { "epoch": 0.7, "learning_rate": 4.564115084134616e-05, "loss": 2.1376, "step": 2323 }, { "epoch": 0.7, "learning_rate": 4.5639272836538465e-05, "loss": 2.2667, "step": 2324 }, { "epoch": 0.7, "learning_rate": 4.563739483173077e-05, "loss": 2.4477, "step": 2325 }, { "epoch": 0.7, "learning_rate": 4.563551682692308e-05, "loss": 2.1734, "step": 2326 }, { "epoch": 0.7, "learning_rate": 4.5633638822115386e-05, "loss": 2.2278, "step": 2327 }, { "epoch": 0.7, "learning_rate": 4.56317608173077e-05, "loss": 2.3819, "step": 2328 }, { "epoch": 0.7, "learning_rate": 4.56298828125e-05, "loss": 2.2602, "step": 2329 }, { "epoch": 0.7, "learning_rate": 4.562800480769231e-05, "loss": 2.0793, "step": 2330 }, { "epoch": 0.7, "learning_rate": 4.562612680288462e-05, "loss": 2.2663, "step": 2331 }, { "epoch": 0.7, "learning_rate": 4.5624248798076924e-05, "loss": 2.3206, "step": 2332 }, { "epoch": 0.7, "learning_rate": 4.562237079326923e-05, "loss": 2.4191, "step": 2333 }, { "epoch": 0.7, "learning_rate": 4.562049278846154e-05, "loss": 2.257, "step": 2334 }, { "epoch": 0.7, "learning_rate": 4.5618614783653845e-05, "loss": 2.2884, "step": 2335 }, { "epoch": 0.7, "learning_rate": 4.5616736778846156e-05, "loss": 2.3269, "step": 2336 }, { "epoch": 0.7, "learning_rate": 4.561485877403847e-05, "loss": 2.0069, "step": 2337 }, { "epoch": 0.7, "learning_rate": 4.561298076923077e-05, "loss": 2.2275, "step": 2338 }, { "epoch": 0.7, "learning_rate": 4.561110276442308e-05, "loss": 2.0524, "step": 2339 }, { "epoch": 0.7, "learning_rate": 4.560922475961539e-05, "loss": 2.2085, "step": 2340 }, { "epoch": 0.7, "learning_rate": 4.5607346754807694e-05, "loss": 1.8526, "step": 2341 }, { "epoch": 0.7, "learning_rate": 4.5605468750000005e-05, "loss": 2.2348, "step": 2342 }, { "epoch": 0.7, "learning_rate": 4.560359074519231e-05, "loss": 2.0815, "step": 2343 }, { "epoch": 0.7, "learning_rate": 4.5601712740384615e-05, "loss": 2.6239, "step": 2344 }, { "epoch": 0.7, "learning_rate": 4.5599834735576926e-05, "loss": 1.9979, "step": 2345 }, { "epoch": 0.7, "learning_rate": 4.559795673076923e-05, "loss": 2.3972, "step": 2346 }, { "epoch": 0.71, "learning_rate": 4.5596078725961536e-05, "loss": 2.098, "step": 2347 }, { "epoch": 0.71, "learning_rate": 4.559420072115385e-05, "loss": 2.1849, "step": 2348 }, { "epoch": 0.71, "learning_rate": 4.559232271634616e-05, "loss": 2.1899, "step": 2349 }, { "epoch": 0.71, "learning_rate": 4.5590444711538464e-05, "loss": 1.8575, "step": 2350 }, { "epoch": 0.71, "learning_rate": 4.5588566706730775e-05, "loss": 1.8158, "step": 2351 }, { "epoch": 0.71, "learning_rate": 4.558668870192308e-05, "loss": 2.1457, "step": 2352 }, { "epoch": 0.71, "learning_rate": 4.5584810697115385e-05, "loss": 2.203, "step": 2353 }, { "epoch": 0.71, "learning_rate": 4.5582932692307696e-05, "loss": 2.3329, "step": 2354 }, { "epoch": 0.71, "learning_rate": 4.55810546875e-05, "loss": 1.9074, "step": 2355 }, { "epoch": 0.71, "learning_rate": 4.557917668269231e-05, "loss": 1.9333, "step": 2356 }, { "epoch": 0.71, "learning_rate": 4.557729867788462e-05, "loss": 2.2083, "step": 2357 }, { "epoch": 0.71, "learning_rate": 4.557542067307692e-05, "loss": 2.2448, "step": 2358 }, { "epoch": 0.71, "learning_rate": 4.5573542668269234e-05, "loss": 2.4468, "step": 2359 }, { "epoch": 0.71, "learning_rate": 4.557166466346154e-05, "loss": 2.2696, "step": 2360 }, { "epoch": 0.71, "learning_rate": 4.556978665865384e-05, "loss": 2.2269, "step": 2361 }, { "epoch": 0.71, "learning_rate": 4.556790865384616e-05, "loss": 2.2463, "step": 2362 }, { "epoch": 0.71, "learning_rate": 4.5566030649038466e-05, "loss": 2.24, "step": 2363 }, { "epoch": 0.71, "learning_rate": 4.556415264423077e-05, "loss": 2.1636, "step": 2364 }, { "epoch": 0.71, "learning_rate": 4.556227463942308e-05, "loss": 2.2761, "step": 2365 }, { "epoch": 0.71, "learning_rate": 4.556039663461539e-05, "loss": 2.1806, "step": 2366 }, { "epoch": 0.71, "learning_rate": 4.555851862980769e-05, "loss": 2.0346, "step": 2367 }, { "epoch": 0.71, "learning_rate": 4.5556640625000004e-05, "loss": 2.37, "step": 2368 }, { "epoch": 0.71, "learning_rate": 4.555476262019231e-05, "loss": 2.2911, "step": 2369 }, { "epoch": 0.71, "learning_rate": 4.5552884615384613e-05, "loss": 2.2805, "step": 2370 }, { "epoch": 0.71, "learning_rate": 4.5551006610576925e-05, "loss": 2.3369, "step": 2371 }, { "epoch": 0.71, "learning_rate": 4.554912860576923e-05, "loss": 2.3902, "step": 2372 }, { "epoch": 0.71, "learning_rate": 4.554725060096154e-05, "loss": 2.0022, "step": 2373 }, { "epoch": 0.71, "learning_rate": 4.5545372596153846e-05, "loss": 2.0054, "step": 2374 }, { "epoch": 0.71, "learning_rate": 4.554349459134616e-05, "loss": 2.1929, "step": 2375 }, { "epoch": 0.71, "learning_rate": 4.554161658653847e-05, "loss": 2.1347, "step": 2376 }, { "epoch": 0.71, "learning_rate": 4.5539738581730774e-05, "loss": 2.2813, "step": 2377 }, { "epoch": 0.71, "learning_rate": 4.553786057692308e-05, "loss": 2.3782, "step": 2378 }, { "epoch": 0.71, "learning_rate": 4.553598257211539e-05, "loss": 2.0134, "step": 2379 }, { "epoch": 0.72, "learning_rate": 4.5534104567307695e-05, "loss": 2.5759, "step": 2380 }, { "epoch": 0.72, "learning_rate": 4.55322265625e-05, "loss": 2.1192, "step": 2381 }, { "epoch": 0.72, "learning_rate": 4.553034855769231e-05, "loss": 1.8576, "step": 2382 }, { "epoch": 0.72, "learning_rate": 4.5528470552884616e-05, "loss": 2.1141, "step": 2383 }, { "epoch": 0.72, "learning_rate": 4.552659254807692e-05, "loss": 2.1783, "step": 2384 }, { "epoch": 0.72, "learning_rate": 4.552471454326923e-05, "loss": 2.1035, "step": 2385 }, { "epoch": 0.72, "learning_rate": 4.552283653846154e-05, "loss": 2.2603, "step": 2386 }, { "epoch": 0.72, "learning_rate": 4.552095853365385e-05, "loss": 2.1873, "step": 2387 }, { "epoch": 0.72, "learning_rate": 4.551908052884616e-05, "loss": 2.0845, "step": 2388 }, { "epoch": 0.72, "learning_rate": 4.5517202524038465e-05, "loss": 2.2123, "step": 2389 }, { "epoch": 0.72, "learning_rate": 4.551532451923077e-05, "loss": 2.2874, "step": 2390 }, { "epoch": 0.72, "learning_rate": 4.551344651442308e-05, "loss": 2.2752, "step": 2391 }, { "epoch": 0.72, "learning_rate": 4.5511568509615386e-05, "loss": 2.2647, "step": 2392 }, { "epoch": 0.72, "learning_rate": 4.55096905048077e-05, "loss": 2.2741, "step": 2393 }, { "epoch": 0.72, "learning_rate": 4.55078125e-05, "loss": 2.0415, "step": 2394 }, { "epoch": 0.72, "learning_rate": 4.550593449519231e-05, "loss": 2.0634, "step": 2395 }, { "epoch": 0.72, "learning_rate": 4.550405649038462e-05, "loss": 2.1721, "step": 2396 }, { "epoch": 0.72, "learning_rate": 4.5502178485576924e-05, "loss": 2.2752, "step": 2397 }, { "epoch": 0.72, "learning_rate": 4.550030048076923e-05, "loss": 2.4725, "step": 2398 }, { "epoch": 0.72, "learning_rate": 4.549842247596154e-05, "loss": 2.5783, "step": 2399 }, { "epoch": 0.72, "learning_rate": 4.5496544471153845e-05, "loss": 2.3444, "step": 2400 }, { "epoch": 0.72, "learning_rate": 4.5494666466346156e-05, "loss": 2.2662, "step": 2401 }, { "epoch": 0.72, "learning_rate": 4.549278846153847e-05, "loss": 2.3977, "step": 2402 }, { "epoch": 0.72, "learning_rate": 4.549091045673077e-05, "loss": 1.9576, "step": 2403 }, { "epoch": 0.72, "learning_rate": 4.548903245192308e-05, "loss": 2.2509, "step": 2404 }, { "epoch": 0.72, "learning_rate": 4.548715444711539e-05, "loss": 2.2088, "step": 2405 }, { "epoch": 0.72, "learning_rate": 4.5485276442307694e-05, "loss": 2.2362, "step": 2406 }, { "epoch": 0.72, "learning_rate": 4.5483398437500005e-05, "loss": 2.4972, "step": 2407 }, { "epoch": 0.72, "learning_rate": 4.548152043269231e-05, "loss": 2.5016, "step": 2408 }, { "epoch": 0.72, "learning_rate": 4.5479642427884615e-05, "loss": 1.8463, "step": 2409 }, { "epoch": 0.72, "learning_rate": 4.5477764423076926e-05, "loss": 2.0416, "step": 2410 }, { "epoch": 0.72, "learning_rate": 4.547588641826923e-05, "loss": 2.2254, "step": 2411 }, { "epoch": 0.72, "learning_rate": 4.5474008413461536e-05, "loss": 2.2627, "step": 2412 }, { "epoch": 0.73, "learning_rate": 4.547213040865385e-05, "loss": 2.4592, "step": 2413 }, { "epoch": 0.73, "learning_rate": 4.547025240384616e-05, "loss": 2.1077, "step": 2414 }, { "epoch": 0.73, "learning_rate": 4.5468374399038464e-05, "loss": 2.3516, "step": 2415 }, { "epoch": 0.73, "learning_rate": 4.5466496394230775e-05, "loss": 1.9473, "step": 2416 }, { "epoch": 0.73, "learning_rate": 4.546461838942308e-05, "loss": 1.9808, "step": 2417 }, { "epoch": 0.73, "learning_rate": 4.5462740384615385e-05, "loss": 2.2842, "step": 2418 }, { "epoch": 0.73, "learning_rate": 4.5460862379807696e-05, "loss": 2.17, "step": 2419 }, { "epoch": 0.73, "learning_rate": 4.5458984375e-05, "loss": 2.1073, "step": 2420 }, { "epoch": 0.73, "learning_rate": 4.545710637019231e-05, "loss": 2.0985, "step": 2421 }, { "epoch": 0.73, "learning_rate": 4.545522836538462e-05, "loss": 2.0336, "step": 2422 }, { "epoch": 0.73, "learning_rate": 4.545335036057692e-05, "loss": 2.2661, "step": 2423 }, { "epoch": 0.73, "learning_rate": 4.5451472355769234e-05, "loss": 1.9783, "step": 2424 }, { "epoch": 0.73, "learning_rate": 4.544959435096154e-05, "loss": 2.113, "step": 2425 }, { "epoch": 0.73, "learning_rate": 4.544771634615384e-05, "loss": 2.1613, "step": 2426 }, { "epoch": 0.73, "learning_rate": 4.544583834134616e-05, "loss": 2.1599, "step": 2427 }, { "epoch": 0.73, "learning_rate": 4.5443960336538466e-05, "loss": 2.3945, "step": 2428 }, { "epoch": 0.73, "learning_rate": 4.544208233173077e-05, "loss": 2.29, "step": 2429 }, { "epoch": 0.73, "learning_rate": 4.544020432692308e-05, "loss": 2.0782, "step": 2430 }, { "epoch": 0.73, "learning_rate": 4.543832632211539e-05, "loss": 2.1337, "step": 2431 }, { "epoch": 0.73, "learning_rate": 4.543644831730769e-05, "loss": 2.3862, "step": 2432 }, { "epoch": 0.73, "learning_rate": 4.5434570312500004e-05, "loss": 1.8569, "step": 2433 }, { "epoch": 0.73, "learning_rate": 4.543269230769231e-05, "loss": 2.0838, "step": 2434 }, { "epoch": 0.73, "learning_rate": 4.543081430288462e-05, "loss": 2.1039, "step": 2435 }, { "epoch": 0.73, "learning_rate": 4.5428936298076925e-05, "loss": 2.0509, "step": 2436 }, { "epoch": 0.73, "learning_rate": 4.542705829326923e-05, "loss": 2.3878, "step": 2437 }, { "epoch": 0.73, "learning_rate": 4.542518028846154e-05, "loss": 2.209, "step": 2438 }, { "epoch": 0.73, "learning_rate": 4.5423302283653846e-05, "loss": 1.9938, "step": 2439 }, { "epoch": 0.73, "learning_rate": 4.542142427884616e-05, "loss": 2.406, "step": 2440 }, { "epoch": 0.73, "learning_rate": 4.541954627403847e-05, "loss": 2.2707, "step": 2441 }, { "epoch": 0.73, "learning_rate": 4.5417668269230774e-05, "loss": 2.0047, "step": 2442 }, { "epoch": 0.73, "learning_rate": 4.541579026442308e-05, "loss": 2.1121, "step": 2443 }, { "epoch": 0.73, "learning_rate": 4.541391225961539e-05, "loss": 2.1514, "step": 2444 }, { "epoch": 0.73, "learning_rate": 4.5412034254807695e-05, "loss": 2.2368, "step": 2445 }, { "epoch": 0.73, "learning_rate": 4.541015625e-05, "loss": 2.2633, "step": 2446 }, { "epoch": 0.74, "learning_rate": 4.540827824519231e-05, "loss": 2.1529, "step": 2447 }, { "epoch": 0.74, "learning_rate": 4.5406400240384616e-05, "loss": 2.0818, "step": 2448 }, { "epoch": 0.74, "learning_rate": 4.540452223557692e-05, "loss": 2.1661, "step": 2449 }, { "epoch": 0.74, "learning_rate": 4.540264423076923e-05, "loss": 2.2066, "step": 2450 }, { "epoch": 0.74, "learning_rate": 4.540076622596154e-05, "loss": 2.0559, "step": 2451 }, { "epoch": 0.74, "learning_rate": 4.539888822115385e-05, "loss": 2.0571, "step": 2452 }, { "epoch": 0.74, "learning_rate": 4.539701021634616e-05, "loss": 2.3685, "step": 2453 }, { "epoch": 0.74, "learning_rate": 4.5395132211538465e-05, "loss": 2.2183, "step": 2454 }, { "epoch": 0.74, "learning_rate": 4.5393254206730777e-05, "loss": 2.4742, "step": 2455 }, { "epoch": 0.74, "learning_rate": 4.539137620192308e-05, "loss": 2.1136, "step": 2456 }, { "epoch": 0.74, "learning_rate": 4.5389498197115386e-05, "loss": 2.3277, "step": 2457 }, { "epoch": 0.74, "learning_rate": 4.53876201923077e-05, "loss": 2.3635, "step": 2458 }, { "epoch": 0.74, "learning_rate": 4.53857421875e-05, "loss": 2.2101, "step": 2459 }, { "epoch": 0.74, "learning_rate": 4.538386418269231e-05, "loss": 2.1662, "step": 2460 }, { "epoch": 0.74, "learning_rate": 4.538198617788462e-05, "loss": 2.0502, "step": 2461 }, { "epoch": 0.74, "learning_rate": 4.5380108173076923e-05, "loss": 2.2546, "step": 2462 }, { "epoch": 0.74, "learning_rate": 4.537823016826923e-05, "loss": 1.8596, "step": 2463 }, { "epoch": 0.74, "learning_rate": 4.537635216346154e-05, "loss": 2.5386, "step": 2464 }, { "epoch": 0.74, "learning_rate": 4.5374474158653845e-05, "loss": 2.3443, "step": 2465 }, { "epoch": 0.74, "learning_rate": 4.5372596153846156e-05, "loss": 2.6195, "step": 2466 }, { "epoch": 0.74, "learning_rate": 4.537071814903847e-05, "loss": 2.1264, "step": 2467 }, { "epoch": 0.74, "learning_rate": 4.536884014423077e-05, "loss": 2.2191, "step": 2468 }, { "epoch": 0.74, "learning_rate": 4.536696213942308e-05, "loss": 2.1209, "step": 2469 }, { "epoch": 0.74, "learning_rate": 4.536508413461539e-05, "loss": 2.0022, "step": 2470 }, { "epoch": 0.74, "learning_rate": 4.5363206129807694e-05, "loss": 2.3292, "step": 2471 }, { "epoch": 0.74, "learning_rate": 4.5361328125000005e-05, "loss": 2.2003, "step": 2472 }, { "epoch": 0.74, "learning_rate": 4.535945012019231e-05, "loss": 2.261, "step": 2473 }, { "epoch": 0.74, "learning_rate": 4.5357572115384615e-05, "loss": 2.1477, "step": 2474 }, { "epoch": 0.74, "learning_rate": 4.5355694110576926e-05, "loss": 2.4285, "step": 2475 }, { "epoch": 0.74, "learning_rate": 4.535381610576923e-05, "loss": 2.4046, "step": 2476 }, { "epoch": 0.74, "learning_rate": 4.5351938100961536e-05, "loss": 2.3065, "step": 2477 }, { "epoch": 0.74, "learning_rate": 4.535006009615385e-05, "loss": 2.4482, "step": 2478 }, { "epoch": 0.74, "learning_rate": 4.534818209134616e-05, "loss": 2.2553, "step": 2479 }, { "epoch": 0.75, "learning_rate": 4.5346304086538464e-05, "loss": 2.2852, "step": 2480 }, { "epoch": 0.75, "learning_rate": 4.5344426081730775e-05, "loss": 2.0646, "step": 2481 }, { "epoch": 0.75, "learning_rate": 4.534254807692308e-05, "loss": 2.2451, "step": 2482 }, { "epoch": 0.75, "learning_rate": 4.5340670072115385e-05, "loss": 2.1109, "step": 2483 }, { "epoch": 0.75, "learning_rate": 4.5338792067307696e-05, "loss": 2.0853, "step": 2484 }, { "epoch": 0.75, "learning_rate": 4.53369140625e-05, "loss": 2.1707, "step": 2485 }, { "epoch": 0.75, "learning_rate": 4.533503605769231e-05, "loss": 2.0336, "step": 2486 }, { "epoch": 0.75, "learning_rate": 4.533315805288462e-05, "loss": 2.107, "step": 2487 }, { "epoch": 0.75, "learning_rate": 4.533128004807692e-05, "loss": 2.4174, "step": 2488 }, { "epoch": 0.75, "learning_rate": 4.5329402043269234e-05, "loss": 2.167, "step": 2489 }, { "epoch": 0.75, "learning_rate": 4.532752403846154e-05, "loss": 2.3423, "step": 2490 }, { "epoch": 0.75, "learning_rate": 4.532564603365384e-05, "loss": 2.4065, "step": 2491 }, { "epoch": 0.75, "learning_rate": 4.532376802884616e-05, "loss": 2.2983, "step": 2492 }, { "epoch": 0.75, "learning_rate": 4.5321890024038466e-05, "loss": 2.0512, "step": 2493 }, { "epoch": 0.75, "learning_rate": 4.532001201923077e-05, "loss": 1.9734, "step": 2494 }, { "epoch": 0.75, "learning_rate": 4.531813401442308e-05, "loss": 2.0429, "step": 2495 }, { "epoch": 0.75, "learning_rate": 4.531625600961539e-05, "loss": 2.1366, "step": 2496 }, { "epoch": 0.75, "learning_rate": 4.531437800480769e-05, "loss": 2.0796, "step": 2497 }, { "epoch": 0.75, "learning_rate": 4.5312500000000004e-05, "loss": 2.2331, "step": 2498 }, { "epoch": 0.75, "learning_rate": 4.531062199519231e-05, "loss": 2.0454, "step": 2499 }, { "epoch": 0.75, "learning_rate": 4.530874399038462e-05, "loss": 2.1057, "step": 2500 }, { "epoch": 0.75, "learning_rate": 4.5306865985576925e-05, "loss": 2.2716, "step": 2501 }, { "epoch": 0.75, "learning_rate": 4.530498798076923e-05, "loss": 2.0163, "step": 2502 }, { "epoch": 0.75, "learning_rate": 4.530310997596154e-05, "loss": 2.3603, "step": 2503 }, { "epoch": 0.75, "learning_rate": 4.5301231971153846e-05, "loss": 2.2417, "step": 2504 }, { "epoch": 0.75, "learning_rate": 4.529935396634616e-05, "loss": 2.3501, "step": 2505 }, { "epoch": 0.75, "learning_rate": 4.529747596153847e-05, "loss": 2.2246, "step": 2506 }, { "epoch": 0.75, "learning_rate": 4.5295597956730774e-05, "loss": 2.0561, "step": 2507 }, { "epoch": 0.75, "learning_rate": 4.529371995192308e-05, "loss": 2.2529, "step": 2508 }, { "epoch": 0.75, "learning_rate": 4.529184194711539e-05, "loss": 2.1542, "step": 2509 }, { "epoch": 0.75, "learning_rate": 4.5289963942307695e-05, "loss": 2.2892, "step": 2510 }, { "epoch": 0.75, "learning_rate": 4.52880859375e-05, "loss": 2.2537, "step": 2511 }, { "epoch": 0.75, "learning_rate": 4.528620793269231e-05, "loss": 2.1761, "step": 2512 }, { "epoch": 0.76, "learning_rate": 4.5284329927884616e-05, "loss": 2.1457, "step": 2513 }, { "epoch": 0.76, "learning_rate": 4.528245192307692e-05, "loss": 2.2633, "step": 2514 }, { "epoch": 0.76, "learning_rate": 4.528057391826923e-05, "loss": 2.2268, "step": 2515 }, { "epoch": 0.76, "learning_rate": 4.527869591346154e-05, "loss": 2.1319, "step": 2516 }, { "epoch": 0.76, "learning_rate": 4.527681790865385e-05, "loss": 2.1244, "step": 2517 }, { "epoch": 0.76, "learning_rate": 4.527493990384616e-05, "loss": 2.028, "step": 2518 }, { "epoch": 0.76, "learning_rate": 4.5273061899038465e-05, "loss": 2.1001, "step": 2519 }, { "epoch": 0.76, "learning_rate": 4.5271183894230776e-05, "loss": 2.2484, "step": 2520 }, { "epoch": 0.76, "learning_rate": 4.526930588942308e-05, "loss": 2.0673, "step": 2521 }, { "epoch": 0.76, "learning_rate": 4.5267427884615386e-05, "loss": 2.0433, "step": 2522 }, { "epoch": 0.76, "learning_rate": 4.52655498798077e-05, "loss": 1.9172, "step": 2523 }, { "epoch": 0.76, "learning_rate": 4.5263671875e-05, "loss": 1.8365, "step": 2524 }, { "epoch": 0.76, "learning_rate": 4.526179387019231e-05, "loss": 2.148, "step": 2525 }, { "epoch": 0.76, "learning_rate": 4.525991586538462e-05, "loss": 2.2193, "step": 2526 }, { "epoch": 0.76, "learning_rate": 4.525803786057692e-05, "loss": 2.3937, "step": 2527 }, { "epoch": 0.76, "learning_rate": 4.525615985576923e-05, "loss": 2.1147, "step": 2528 }, { "epoch": 0.76, "learning_rate": 4.525428185096154e-05, "loss": 2.2686, "step": 2529 }, { "epoch": 0.76, "learning_rate": 4.5252403846153844e-05, "loss": 1.9442, "step": 2530 }, { "epoch": 0.76, "learning_rate": 4.5250525841346156e-05, "loss": 2.3499, "step": 2531 }, { "epoch": 0.76, "learning_rate": 4.524864783653847e-05, "loss": 2.359, "step": 2532 }, { "epoch": 0.76, "learning_rate": 4.524676983173077e-05, "loss": 2.0989, "step": 2533 }, { "epoch": 0.76, "learning_rate": 4.524489182692308e-05, "loss": 2.145, "step": 2534 }, { "epoch": 0.76, "learning_rate": 4.524301382211539e-05, "loss": 2.2218, "step": 2535 }, { "epoch": 0.76, "learning_rate": 4.5241135817307693e-05, "loss": 2.5938, "step": 2536 }, { "epoch": 0.76, "learning_rate": 4.5239257812500005e-05, "loss": 2.3684, "step": 2537 }, { "epoch": 0.76, "learning_rate": 4.523737980769231e-05, "loss": 2.262, "step": 2538 }, { "epoch": 0.76, "learning_rate": 4.5235501802884614e-05, "loss": 2.3951, "step": 2539 }, { "epoch": 0.76, "learning_rate": 4.5233623798076926e-05, "loss": 1.9141, "step": 2540 }, { "epoch": 0.76, "learning_rate": 4.523174579326923e-05, "loss": 2.3969, "step": 2541 }, { "epoch": 0.76, "learning_rate": 4.5229867788461536e-05, "loss": 2.2852, "step": 2542 }, { "epoch": 0.76, "learning_rate": 4.522798978365385e-05, "loss": 2.1675, "step": 2543 }, { "epoch": 0.76, "learning_rate": 4.522611177884616e-05, "loss": 2.208, "step": 2544 }, { "epoch": 0.76, "learning_rate": 4.5224233774038463e-05, "loss": 2.2355, "step": 2545 }, { "epoch": 0.77, "learning_rate": 4.5222355769230775e-05, "loss": 2.1185, "step": 2546 }, { "epoch": 0.77, "learning_rate": 4.522047776442308e-05, "loss": 2.0059, "step": 2547 }, { "epoch": 0.77, "learning_rate": 4.5218599759615385e-05, "loss": 2.1604, "step": 2548 }, { "epoch": 0.77, "learning_rate": 4.5216721754807696e-05, "loss": 2.2528, "step": 2549 }, { "epoch": 0.77, "learning_rate": 4.521484375e-05, "loss": 2.5142, "step": 2550 }, { "epoch": 0.77, "learning_rate": 4.521296574519231e-05, "loss": 1.8655, "step": 2551 }, { "epoch": 0.77, "learning_rate": 4.521108774038462e-05, "loss": 2.2675, "step": 2552 }, { "epoch": 0.77, "learning_rate": 4.520920973557692e-05, "loss": 2.3642, "step": 2553 }, { "epoch": 0.77, "learning_rate": 4.5207331730769233e-05, "loss": 1.9128, "step": 2554 }, { "epoch": 0.77, "learning_rate": 4.520545372596154e-05, "loss": 2.2085, "step": 2555 }, { "epoch": 0.77, "learning_rate": 4.520357572115384e-05, "loss": 2.5049, "step": 2556 }, { "epoch": 0.77, "learning_rate": 4.520169771634616e-05, "loss": 2.4866, "step": 2557 }, { "epoch": 0.77, "learning_rate": 4.5199819711538466e-05, "loss": 2.0002, "step": 2558 }, { "epoch": 0.77, "learning_rate": 4.519794170673077e-05, "loss": 2.006, "step": 2559 }, { "epoch": 0.77, "learning_rate": 4.519606370192308e-05, "loss": 2.4833, "step": 2560 }, { "epoch": 0.77, "learning_rate": 4.519418569711539e-05, "loss": 2.5368, "step": 2561 }, { "epoch": 0.77, "learning_rate": 4.519230769230769e-05, "loss": 2.1308, "step": 2562 }, { "epoch": 0.77, "learning_rate": 4.5190429687500004e-05, "loss": 2.3883, "step": 2563 }, { "epoch": 0.77, "learning_rate": 4.518855168269231e-05, "loss": 2.2157, "step": 2564 }, { "epoch": 0.77, "learning_rate": 4.518667367788462e-05, "loss": 2.3339, "step": 2565 }, { "epoch": 0.77, "learning_rate": 4.5184795673076925e-05, "loss": 2.3057, "step": 2566 }, { "epoch": 0.77, "learning_rate": 4.518291766826923e-05, "loss": 2.3439, "step": 2567 }, { "epoch": 0.77, "learning_rate": 4.518103966346154e-05, "loss": 2.1934, "step": 2568 }, { "epoch": 0.77, "learning_rate": 4.5179161658653846e-05, "loss": 2.2915, "step": 2569 }, { "epoch": 0.77, "learning_rate": 4.517728365384616e-05, "loss": 2.1316, "step": 2570 }, { "epoch": 0.77, "learning_rate": 4.517540564903847e-05, "loss": 2.527, "step": 2571 }, { "epoch": 0.77, "learning_rate": 4.5173527644230774e-05, "loss": 2.0685, "step": 2572 }, { "epoch": 0.77, "learning_rate": 4.517164963942308e-05, "loss": 2.2218, "step": 2573 }, { "epoch": 0.77, "learning_rate": 4.516977163461539e-05, "loss": 2.0445, "step": 2574 }, { "epoch": 0.77, "learning_rate": 4.5167893629807695e-05, "loss": 2.147, "step": 2575 }, { "epoch": 0.77, "learning_rate": 4.5166015625e-05, "loss": 2.2522, "step": 2576 }, { "epoch": 0.77, "learning_rate": 4.516413762019231e-05, "loss": 2.034, "step": 2577 }, { "epoch": 0.77, "learning_rate": 4.5162259615384616e-05, "loss": 1.9141, "step": 2578 }, { "epoch": 0.77, "learning_rate": 4.516038161057693e-05, "loss": 1.988, "step": 2579 }, { "epoch": 0.78, "learning_rate": 4.515850360576923e-05, "loss": 2.8407, "step": 2580 }, { "epoch": 0.78, "learning_rate": 4.515662560096154e-05, "loss": 2.0615, "step": 2581 }, { "epoch": 0.78, "learning_rate": 4.515474759615385e-05, "loss": 2.1409, "step": 2582 }, { "epoch": 0.78, "learning_rate": 4.515286959134616e-05, "loss": 2.3723, "step": 2583 }, { "epoch": 0.78, "learning_rate": 4.5150991586538465e-05, "loss": 1.8059, "step": 2584 }, { "epoch": 0.78, "learning_rate": 4.5149113581730776e-05, "loss": 2.1024, "step": 2585 }, { "epoch": 0.78, "learning_rate": 4.514723557692308e-05, "loss": 2.1622, "step": 2586 }, { "epoch": 0.78, "learning_rate": 4.5145357572115386e-05, "loss": 2.4263, "step": 2587 }, { "epoch": 0.78, "learning_rate": 4.51434795673077e-05, "loss": 2.1546, "step": 2588 }, { "epoch": 0.78, "learning_rate": 4.51416015625e-05, "loss": 2.4309, "step": 2589 }, { "epoch": 0.78, "learning_rate": 4.513972355769231e-05, "loss": 2.105, "step": 2590 }, { "epoch": 0.78, "learning_rate": 4.513784555288462e-05, "loss": 2.1415, "step": 2591 }, { "epoch": 0.78, "learning_rate": 4.513596754807692e-05, "loss": 1.999, "step": 2592 }, { "epoch": 0.78, "learning_rate": 4.513408954326923e-05, "loss": 2.0006, "step": 2593 }, { "epoch": 0.78, "learning_rate": 4.513221153846154e-05, "loss": 1.962, "step": 2594 }, { "epoch": 0.78, "learning_rate": 4.5130333533653844e-05, "loss": 2.0935, "step": 2595 }, { "epoch": 0.78, "learning_rate": 4.5128455528846156e-05, "loss": 2.3454, "step": 2596 }, { "epoch": 0.78, "learning_rate": 4.512657752403847e-05, "loss": 2.4476, "step": 2597 }, { "epoch": 0.78, "learning_rate": 4.512469951923077e-05, "loss": 1.9725, "step": 2598 }, { "epoch": 0.78, "learning_rate": 4.5122821514423084e-05, "loss": 1.929, "step": 2599 }, { "epoch": 0.78, "learning_rate": 4.512094350961539e-05, "loss": 2.234, "step": 2600 }, { "epoch": 0.78, "learning_rate": 4.511906550480769e-05, "loss": 1.8907, "step": 2601 }, { "epoch": 0.78, "learning_rate": 4.5117187500000005e-05, "loss": 1.8842, "step": 2602 }, { "epoch": 0.78, "learning_rate": 4.511530949519231e-05, "loss": 2.3146, "step": 2603 }, { "epoch": 0.78, "learning_rate": 4.5113431490384614e-05, "loss": 1.9002, "step": 2604 }, { "epoch": 0.78, "learning_rate": 4.5111553485576926e-05, "loss": 2.1716, "step": 2605 }, { "epoch": 0.78, "learning_rate": 4.510967548076923e-05, "loss": 2.4505, "step": 2606 }, { "epoch": 0.78, "learning_rate": 4.5107797475961535e-05, "loss": 2.2164, "step": 2607 }, { "epoch": 0.78, "learning_rate": 4.510591947115385e-05, "loss": 2.2706, "step": 2608 }, { "epoch": 0.78, "learning_rate": 4.510404146634616e-05, "loss": 1.8366, "step": 2609 }, { "epoch": 0.78, "learning_rate": 4.510216346153846e-05, "loss": 2.0839, "step": 2610 }, { "epoch": 0.78, "learning_rate": 4.5100285456730775e-05, "loss": 2.2453, "step": 2611 }, { "epoch": 0.78, "learning_rate": 4.509840745192308e-05, "loss": 2.0801, "step": 2612 }, { "epoch": 0.79, "learning_rate": 4.5096529447115384e-05, "loss": 2.2014, "step": 2613 }, { "epoch": 0.79, "learning_rate": 4.5094651442307696e-05, "loss": 2.3174, "step": 2614 }, { "epoch": 0.79, "learning_rate": 4.50927734375e-05, "loss": 2.1784, "step": 2615 }, { "epoch": 0.79, "learning_rate": 4.509089543269231e-05, "loss": 2.2415, "step": 2616 }, { "epoch": 0.79, "learning_rate": 4.508901742788462e-05, "loss": 2.3234, "step": 2617 }, { "epoch": 0.79, "learning_rate": 4.508713942307692e-05, "loss": 2.2784, "step": 2618 }, { "epoch": 0.79, "learning_rate": 4.508526141826923e-05, "loss": 2.1251, "step": 2619 }, { "epoch": 0.79, "learning_rate": 4.508338341346154e-05, "loss": 2.1941, "step": 2620 }, { "epoch": 0.79, "learning_rate": 4.508150540865384e-05, "loss": 2.3214, "step": 2621 }, { "epoch": 0.79, "learning_rate": 4.507962740384616e-05, "loss": 2.1404, "step": 2622 }, { "epoch": 0.79, "learning_rate": 4.5077749399038466e-05, "loss": 2.4102, "step": 2623 }, { "epoch": 0.79, "learning_rate": 4.507587139423077e-05, "loss": 2.6855, "step": 2624 }, { "epoch": 0.79, "learning_rate": 4.507399338942308e-05, "loss": 1.9197, "step": 2625 }, { "epoch": 0.79, "learning_rate": 4.507211538461539e-05, "loss": 1.8552, "step": 2626 }, { "epoch": 0.79, "learning_rate": 4.507023737980769e-05, "loss": 2.2762, "step": 2627 }, { "epoch": 0.79, "learning_rate": 4.5068359375000003e-05, "loss": 2.4595, "step": 2628 }, { "epoch": 0.79, "learning_rate": 4.506648137019231e-05, "loss": 2.3155, "step": 2629 }, { "epoch": 0.79, "learning_rate": 4.506460336538462e-05, "loss": 2.1041, "step": 2630 }, { "epoch": 0.79, "learning_rate": 4.5062725360576925e-05, "loss": 2.469, "step": 2631 }, { "epoch": 0.79, "learning_rate": 4.506084735576923e-05, "loss": 2.1261, "step": 2632 }, { "epoch": 0.79, "learning_rate": 4.505896935096154e-05, "loss": 2.4695, "step": 2633 }, { "epoch": 0.79, "learning_rate": 4.5057091346153846e-05, "loss": 2.0814, "step": 2634 }, { "epoch": 0.79, "learning_rate": 4.505521334134616e-05, "loss": 1.9803, "step": 2635 }, { "epoch": 0.79, "learning_rate": 4.505333533653847e-05, "loss": 2.178, "step": 2636 }, { "epoch": 0.79, "learning_rate": 4.5051457331730773e-05, "loss": 2.2474, "step": 2637 }, { "epoch": 0.79, "learning_rate": 4.504957932692308e-05, "loss": 2.0867, "step": 2638 }, { "epoch": 0.79, "learning_rate": 4.504770132211539e-05, "loss": 2.2194, "step": 2639 }, { "epoch": 0.79, "learning_rate": 4.5045823317307695e-05, "loss": 2.3877, "step": 2640 }, { "epoch": 0.79, "learning_rate": 4.50439453125e-05, "loss": 2.2519, "step": 2641 }, { "epoch": 0.79, "learning_rate": 4.504206730769231e-05, "loss": 2.055, "step": 2642 }, { "epoch": 0.79, "learning_rate": 4.5040189302884616e-05, "loss": 2.1242, "step": 2643 }, { "epoch": 0.79, "learning_rate": 4.503831129807693e-05, "loss": 2.0171, "step": 2644 }, { "epoch": 0.79, "learning_rate": 4.503643329326923e-05, "loss": 2.4294, "step": 2645 }, { "epoch": 0.8, "learning_rate": 4.503455528846154e-05, "loss": 2.1284, "step": 2646 }, { "epoch": 0.8, "learning_rate": 4.503267728365385e-05, "loss": 2.1904, "step": 2647 }, { "epoch": 0.8, "learning_rate": 4.503079927884616e-05, "loss": 2.029, "step": 2648 }, { "epoch": 0.8, "learning_rate": 4.5028921274038465e-05, "loss": 2.3838, "step": 2649 }, { "epoch": 0.8, "learning_rate": 4.5027043269230776e-05, "loss": 1.9753, "step": 2650 }, { "epoch": 0.8, "learning_rate": 4.502516526442308e-05, "loss": 2.3403, "step": 2651 }, { "epoch": 0.8, "learning_rate": 4.5023287259615386e-05, "loss": 2.1683, "step": 2652 }, { "epoch": 0.8, "learning_rate": 4.50214092548077e-05, "loss": 2.2424, "step": 2653 }, { "epoch": 0.8, "learning_rate": 4.501953125e-05, "loss": 1.814, "step": 2654 }, { "epoch": 0.8, "learning_rate": 4.501765324519231e-05, "loss": 2.3262, "step": 2655 }, { "epoch": 0.8, "learning_rate": 4.501577524038462e-05, "loss": 2.1069, "step": 2656 }, { "epoch": 0.8, "learning_rate": 4.501389723557692e-05, "loss": 2.1108, "step": 2657 }, { "epoch": 0.8, "learning_rate": 4.501201923076923e-05, "loss": 2.1387, "step": 2658 }, { "epoch": 0.8, "learning_rate": 4.501014122596154e-05, "loss": 1.9934, "step": 2659 }, { "epoch": 0.8, "learning_rate": 4.5008263221153844e-05, "loss": 2.2403, "step": 2660 }, { "epoch": 0.8, "learning_rate": 4.5006385216346156e-05, "loss": 2.2151, "step": 2661 }, { "epoch": 0.8, "learning_rate": 4.500450721153847e-05, "loss": 2.1761, "step": 2662 }, { "epoch": 0.8, "learning_rate": 4.500262920673077e-05, "loss": 2.1247, "step": 2663 }, { "epoch": 0.8, "learning_rate": 4.5000751201923084e-05, "loss": 2.035, "step": 2664 }, { "epoch": 0.8, "learning_rate": 4.499887319711539e-05, "loss": 2.115, "step": 2665 }, { "epoch": 0.8, "learning_rate": 4.499699519230769e-05, "loss": 2.047, "step": 2666 }, { "epoch": 0.8, "learning_rate": 4.4995117187500005e-05, "loss": 1.985, "step": 2667 }, { "epoch": 0.8, "learning_rate": 4.499323918269231e-05, "loss": 2.461, "step": 2668 }, { "epoch": 0.8, "learning_rate": 4.4991361177884614e-05, "loss": 2.2295, "step": 2669 }, { "epoch": 0.8, "learning_rate": 4.4989483173076926e-05, "loss": 2.3746, "step": 2670 }, { "epoch": 0.8, "learning_rate": 4.498760516826923e-05, "loss": 2.2719, "step": 2671 }, { "epoch": 0.8, "learning_rate": 4.4985727163461535e-05, "loss": 2.4965, "step": 2672 }, { "epoch": 0.8, "learning_rate": 4.498384915865385e-05, "loss": 2.4492, "step": 2673 }, { "epoch": 0.8, "learning_rate": 4.498197115384616e-05, "loss": 2.1097, "step": 2674 }, { "epoch": 0.8, "learning_rate": 4.498009314903846e-05, "loss": 2.2771, "step": 2675 }, { "epoch": 0.8, "learning_rate": 4.4978215144230775e-05, "loss": 2.3054, "step": 2676 }, { "epoch": 0.8, "learning_rate": 4.497633713942308e-05, "loss": 2.2737, "step": 2677 }, { "epoch": 0.8, "learning_rate": 4.4974459134615384e-05, "loss": 1.9045, "step": 2678 }, { "epoch": 0.8, "learning_rate": 4.4972581129807696e-05, "loss": 2.1915, "step": 2679 }, { "epoch": 0.81, "learning_rate": 4.4970703125e-05, "loss": 2.3502, "step": 2680 }, { "epoch": 0.81, "learning_rate": 4.496882512019231e-05, "loss": 2.0367, "step": 2681 }, { "epoch": 0.81, "learning_rate": 4.496694711538462e-05, "loss": 2.2951, "step": 2682 }, { "epoch": 0.81, "learning_rate": 4.496506911057692e-05, "loss": 2.2381, "step": 2683 }, { "epoch": 0.81, "learning_rate": 4.496319110576923e-05, "loss": 1.8546, "step": 2684 }, { "epoch": 0.81, "learning_rate": 4.496131310096154e-05, "loss": 2.3537, "step": 2685 }, { "epoch": 0.81, "learning_rate": 4.495943509615384e-05, "loss": 2.3807, "step": 2686 }, { "epoch": 0.81, "learning_rate": 4.495755709134616e-05, "loss": 2.2357, "step": 2687 }, { "epoch": 0.81, "learning_rate": 4.4955679086538466e-05, "loss": 2.0124, "step": 2688 }, { "epoch": 0.81, "learning_rate": 4.495380108173077e-05, "loss": 1.8298, "step": 2689 }, { "epoch": 0.81, "learning_rate": 4.495192307692308e-05, "loss": 2.2166, "step": 2690 }, { "epoch": 0.81, "learning_rate": 4.495004507211539e-05, "loss": 2.1138, "step": 2691 }, { "epoch": 0.81, "learning_rate": 4.494816706730769e-05, "loss": 2.1768, "step": 2692 }, { "epoch": 0.81, "learning_rate": 4.49462890625e-05, "loss": 2.412, "step": 2693 }, { "epoch": 0.81, "learning_rate": 4.494441105769231e-05, "loss": 2.396, "step": 2694 }, { "epoch": 0.81, "learning_rate": 4.494253305288462e-05, "loss": 2.1732, "step": 2695 }, { "epoch": 0.81, "learning_rate": 4.4940655048076924e-05, "loss": 2.1231, "step": 2696 }, { "epoch": 0.81, "learning_rate": 4.493877704326923e-05, "loss": 2.0688, "step": 2697 }, { "epoch": 0.81, "learning_rate": 4.493689903846154e-05, "loss": 2.2525, "step": 2698 }, { "epoch": 0.81, "learning_rate": 4.4935021033653846e-05, "loss": 2.1116, "step": 2699 }, { "epoch": 0.81, "learning_rate": 4.493314302884616e-05, "loss": 2.135, "step": 2700 }, { "epoch": 0.81, "learning_rate": 4.493126502403847e-05, "loss": 2.0045, "step": 2701 }, { "epoch": 0.81, "learning_rate": 4.492938701923077e-05, "loss": 2.3609, "step": 2702 }, { "epoch": 0.81, "learning_rate": 4.492750901442308e-05, "loss": 1.9935, "step": 2703 }, { "epoch": 0.81, "learning_rate": 4.492563100961539e-05, "loss": 2.2216, "step": 2704 }, { "epoch": 0.81, "learning_rate": 4.4923753004807694e-05, "loss": 2.1368, "step": 2705 }, { "epoch": 0.81, "learning_rate": 4.4921875e-05, "loss": 2.0414, "step": 2706 }, { "epoch": 0.81, "learning_rate": 4.491999699519231e-05, "loss": 2.172, "step": 2707 }, { "epoch": 0.81, "learning_rate": 4.4918118990384616e-05, "loss": 2.2359, "step": 2708 }, { "epoch": 0.81, "learning_rate": 4.491624098557693e-05, "loss": 2.1332, "step": 2709 }, { "epoch": 0.81, "learning_rate": 4.491436298076923e-05, "loss": 2.1421, "step": 2710 }, { "epoch": 0.81, "learning_rate": 4.491248497596154e-05, "loss": 2.2213, "step": 2711 }, { "epoch": 0.81, "learning_rate": 4.491060697115385e-05, "loss": 2.2543, "step": 2712 }, { "epoch": 0.82, "learning_rate": 4.490872896634616e-05, "loss": 2.1194, "step": 2713 }, { "epoch": 0.82, "learning_rate": 4.4906850961538465e-05, "loss": 2.3533, "step": 2714 }, { "epoch": 0.82, "learning_rate": 4.4904972956730776e-05, "loss": 1.876, "step": 2715 }, { "epoch": 0.82, "learning_rate": 4.490309495192308e-05, "loss": 2.3216, "step": 2716 }, { "epoch": 0.82, "learning_rate": 4.4901216947115386e-05, "loss": 2.2826, "step": 2717 }, { "epoch": 0.82, "learning_rate": 4.48993389423077e-05, "loss": 2.1132, "step": 2718 }, { "epoch": 0.82, "learning_rate": 4.48974609375e-05, "loss": 2.1525, "step": 2719 }, { "epoch": 0.82, "learning_rate": 4.489558293269231e-05, "loss": 2.3471, "step": 2720 }, { "epoch": 0.82, "learning_rate": 4.489370492788462e-05, "loss": 2.1663, "step": 2721 }, { "epoch": 0.82, "learning_rate": 4.489182692307692e-05, "loss": 2.188, "step": 2722 }, { "epoch": 0.82, "learning_rate": 4.4889948918269235e-05, "loss": 2.1166, "step": 2723 }, { "epoch": 0.82, "learning_rate": 4.488807091346154e-05, "loss": 2.2007, "step": 2724 }, { "epoch": 0.82, "learning_rate": 4.4886192908653844e-05, "loss": 2.1298, "step": 2725 }, { "epoch": 0.82, "learning_rate": 4.4884314903846156e-05, "loss": 2.0431, "step": 2726 }, { "epoch": 0.82, "learning_rate": 4.488243689903847e-05, "loss": 2.1389, "step": 2727 }, { "epoch": 0.82, "learning_rate": 4.488055889423077e-05, "loss": 2.1571, "step": 2728 }, { "epoch": 0.82, "learning_rate": 4.4878680889423084e-05, "loss": 2.5302, "step": 2729 }, { "epoch": 0.82, "learning_rate": 4.487680288461539e-05, "loss": 2.3826, "step": 2730 }, { "epoch": 0.82, "learning_rate": 4.487492487980769e-05, "loss": 2.0467, "step": 2731 }, { "epoch": 0.82, "learning_rate": 4.4873046875000005e-05, "loss": 1.8609, "step": 2732 }, { "epoch": 0.82, "learning_rate": 4.487116887019231e-05, "loss": 2.1367, "step": 2733 }, { "epoch": 0.82, "learning_rate": 4.4869290865384614e-05, "loss": 2.2238, "step": 2734 }, { "epoch": 0.82, "learning_rate": 4.4867412860576926e-05, "loss": 2.1687, "step": 2735 }, { "epoch": 0.82, "learning_rate": 4.486553485576923e-05, "loss": 1.9358, "step": 2736 }, { "epoch": 0.82, "learning_rate": 4.4863656850961535e-05, "loss": 2.1438, "step": 2737 }, { "epoch": 0.82, "learning_rate": 4.486177884615385e-05, "loss": 1.9836, "step": 2738 }, { "epoch": 0.82, "learning_rate": 4.485990084134616e-05, "loss": 2.2251, "step": 2739 }, { "epoch": 0.82, "learning_rate": 4.485802283653846e-05, "loss": 2.0876, "step": 2740 }, { "epoch": 0.82, "learning_rate": 4.4856144831730775e-05, "loss": 2.2826, "step": 2741 }, { "epoch": 0.82, "learning_rate": 4.485426682692308e-05, "loss": 2.189, "step": 2742 }, { "epoch": 0.82, "learning_rate": 4.485238882211539e-05, "loss": 2.4358, "step": 2743 }, { "epoch": 0.82, "learning_rate": 4.4850510817307696e-05, "loss": 2.2462, "step": 2744 }, { "epoch": 0.82, "learning_rate": 4.48486328125e-05, "loss": 2.0654, "step": 2745 }, { "epoch": 0.83, "learning_rate": 4.484675480769231e-05, "loss": 2.038, "step": 2746 }, { "epoch": 0.83, "learning_rate": 4.484487680288462e-05, "loss": 2.2447, "step": 2747 }, { "epoch": 0.83, "learning_rate": 4.484299879807692e-05, "loss": 2.2969, "step": 2748 }, { "epoch": 0.83, "learning_rate": 4.484112079326923e-05, "loss": 2.0421, "step": 2749 }, { "epoch": 0.83, "learning_rate": 4.483924278846154e-05, "loss": 2.1568, "step": 2750 }, { "epoch": 0.83, "learning_rate": 4.483736478365384e-05, "loss": 2.1413, "step": 2751 }, { "epoch": 0.83, "learning_rate": 4.483548677884616e-05, "loss": 2.0027, "step": 2752 }, { "epoch": 0.83, "learning_rate": 4.4833608774038466e-05, "loss": 2.2672, "step": 2753 }, { "epoch": 0.83, "learning_rate": 4.483173076923077e-05, "loss": 1.7456, "step": 2754 }, { "epoch": 0.83, "learning_rate": 4.482985276442308e-05, "loss": 2.1862, "step": 2755 }, { "epoch": 0.83, "learning_rate": 4.482797475961539e-05, "loss": 2.3768, "step": 2756 }, { "epoch": 0.83, "learning_rate": 4.482609675480769e-05, "loss": 2.2598, "step": 2757 }, { "epoch": 0.83, "learning_rate": 4.482421875e-05, "loss": 2.1092, "step": 2758 }, { "epoch": 0.83, "learning_rate": 4.482234074519231e-05, "loss": 2.2764, "step": 2759 }, { "epoch": 0.83, "learning_rate": 4.482046274038462e-05, "loss": 2.2298, "step": 2760 }, { "epoch": 0.83, "learning_rate": 4.4818584735576924e-05, "loss": 2.319, "step": 2761 }, { "epoch": 0.83, "learning_rate": 4.481670673076923e-05, "loss": 2.1528, "step": 2762 }, { "epoch": 0.83, "learning_rate": 4.481482872596154e-05, "loss": 2.1185, "step": 2763 }, { "epoch": 0.83, "learning_rate": 4.4812950721153845e-05, "loss": 2.4123, "step": 2764 }, { "epoch": 0.83, "learning_rate": 4.481107271634616e-05, "loss": 1.8695, "step": 2765 }, { "epoch": 0.83, "learning_rate": 4.480919471153847e-05, "loss": 2.2018, "step": 2766 }, { "epoch": 0.83, "learning_rate": 4.480731670673077e-05, "loss": 2.3084, "step": 2767 }, { "epoch": 0.83, "learning_rate": 4.480543870192308e-05, "loss": 2.1552, "step": 2768 }, { "epoch": 0.83, "learning_rate": 4.480356069711539e-05, "loss": 2.1487, "step": 2769 }, { "epoch": 0.83, "learning_rate": 4.4801682692307694e-05, "loss": 2.1125, "step": 2770 }, { "epoch": 0.83, "learning_rate": 4.47998046875e-05, "loss": 2.1348, "step": 2771 }, { "epoch": 0.83, "learning_rate": 4.479792668269231e-05, "loss": 2.2949, "step": 2772 }, { "epoch": 0.83, "learning_rate": 4.4796048677884615e-05, "loss": 2.0637, "step": 2773 }, { "epoch": 0.83, "learning_rate": 4.479417067307693e-05, "loss": 2.1647, "step": 2774 }, { "epoch": 0.83, "learning_rate": 4.479229266826923e-05, "loss": 2.1773, "step": 2775 }, { "epoch": 0.83, "learning_rate": 4.4790414663461537e-05, "loss": 2.1567, "step": 2776 }, { "epoch": 0.83, "learning_rate": 4.478853665865385e-05, "loss": 2.2021, "step": 2777 }, { "epoch": 0.83, "learning_rate": 4.478665865384616e-05, "loss": 2.291, "step": 2778 }, { "epoch": 0.84, "learning_rate": 4.4784780649038464e-05, "loss": 2.0269, "step": 2779 }, { "epoch": 0.84, "learning_rate": 4.4782902644230776e-05, "loss": 2.2754, "step": 2780 }, { "epoch": 0.84, "learning_rate": 4.478102463942308e-05, "loss": 2.1558, "step": 2781 }, { "epoch": 0.84, "learning_rate": 4.4779146634615386e-05, "loss": 2.4579, "step": 2782 }, { "epoch": 0.84, "learning_rate": 4.47772686298077e-05, "loss": 2.3362, "step": 2783 }, { "epoch": 0.84, "learning_rate": 4.4775390625e-05, "loss": 2.3689, "step": 2784 }, { "epoch": 0.84, "learning_rate": 4.4773512620192307e-05, "loss": 2.2238, "step": 2785 }, { "epoch": 0.84, "learning_rate": 4.477163461538462e-05, "loss": 2.093, "step": 2786 }, { "epoch": 0.84, "learning_rate": 4.476975661057692e-05, "loss": 2.2577, "step": 2787 }, { "epoch": 0.84, "learning_rate": 4.4767878605769234e-05, "loss": 2.1265, "step": 2788 }, { "epoch": 0.84, "learning_rate": 4.476600060096154e-05, "loss": 2.2753, "step": 2789 }, { "epoch": 0.84, "learning_rate": 4.4764122596153844e-05, "loss": 2.224, "step": 2790 }, { "epoch": 0.84, "learning_rate": 4.4762244591346156e-05, "loss": 2.1567, "step": 2791 }, { "epoch": 0.84, "learning_rate": 4.476036658653847e-05, "loss": 1.8743, "step": 2792 }, { "epoch": 0.84, "learning_rate": 4.475848858173077e-05, "loss": 2.0899, "step": 2793 }, { "epoch": 0.84, "learning_rate": 4.4756610576923083e-05, "loss": 2.1169, "step": 2794 }, { "epoch": 0.84, "learning_rate": 4.475473257211539e-05, "loss": 2.3887, "step": 2795 }, { "epoch": 0.84, "learning_rate": 4.475285456730769e-05, "loss": 2.3984, "step": 2796 }, { "epoch": 0.84, "learning_rate": 4.4750976562500005e-05, "loss": 1.8586, "step": 2797 }, { "epoch": 0.84, "learning_rate": 4.474909855769231e-05, "loss": 2.3191, "step": 2798 }, { "epoch": 0.84, "learning_rate": 4.4747220552884614e-05, "loss": 2.3494, "step": 2799 }, { "epoch": 0.84, "learning_rate": 4.4745342548076926e-05, "loss": 2.1828, "step": 2800 }, { "epoch": 0.84, "learning_rate": 4.474346454326923e-05, "loss": 2.3556, "step": 2801 }, { "epoch": 0.84, "learning_rate": 4.4741586538461535e-05, "loss": 2.1863, "step": 2802 }, { "epoch": 0.84, "learning_rate": 4.473970853365385e-05, "loss": 2.2817, "step": 2803 }, { "epoch": 0.84, "learning_rate": 4.473783052884616e-05, "loss": 1.9186, "step": 2804 }, { "epoch": 0.84, "learning_rate": 4.473595252403846e-05, "loss": 1.9479, "step": 2805 }, { "epoch": 0.84, "learning_rate": 4.4734074519230775e-05, "loss": 2.2648, "step": 2806 }, { "epoch": 0.84, "learning_rate": 4.473219651442308e-05, "loss": 2.1415, "step": 2807 }, { "epoch": 0.84, "learning_rate": 4.473031850961539e-05, "loss": 2.1309, "step": 2808 }, { "epoch": 0.84, "learning_rate": 4.4728440504807696e-05, "loss": 2.1211, "step": 2809 }, { "epoch": 0.84, "learning_rate": 4.47265625e-05, "loss": 2.3358, "step": 2810 }, { "epoch": 0.84, "learning_rate": 4.472468449519231e-05, "loss": 2.3652, "step": 2811 }, { "epoch": 0.84, "learning_rate": 4.472280649038462e-05, "loss": 2.3054, "step": 2812 }, { "epoch": 0.85, "learning_rate": 4.472092848557692e-05, "loss": 2.1297, "step": 2813 }, { "epoch": 0.85, "learning_rate": 4.471905048076923e-05, "loss": 2.0469, "step": 2814 }, { "epoch": 0.85, "learning_rate": 4.471717247596154e-05, "loss": 2.2256, "step": 2815 }, { "epoch": 0.85, "learning_rate": 4.471529447115384e-05, "loss": 1.9596, "step": 2816 }, { "epoch": 0.85, "learning_rate": 4.471341646634616e-05, "loss": 2.1091, "step": 2817 }, { "epoch": 0.85, "learning_rate": 4.4711538461538466e-05, "loss": 2.1895, "step": 2818 }, { "epoch": 0.85, "learning_rate": 4.470966045673077e-05, "loss": 2.1309, "step": 2819 }, { "epoch": 0.85, "learning_rate": 4.470778245192308e-05, "loss": 2.3742, "step": 2820 }, { "epoch": 0.85, "learning_rate": 4.470590444711539e-05, "loss": 2.339, "step": 2821 }, { "epoch": 0.85, "learning_rate": 4.470402644230769e-05, "loss": 2.1223, "step": 2822 }, { "epoch": 0.85, "learning_rate": 4.47021484375e-05, "loss": 1.9958, "step": 2823 }, { "epoch": 0.85, "learning_rate": 4.470027043269231e-05, "loss": 2.1797, "step": 2824 }, { "epoch": 0.85, "learning_rate": 4.469839242788462e-05, "loss": 2.1094, "step": 2825 }, { "epoch": 0.85, "learning_rate": 4.4696514423076924e-05, "loss": 1.9418, "step": 2826 }, { "epoch": 0.85, "learning_rate": 4.469463641826923e-05, "loss": 2.1014, "step": 2827 }, { "epoch": 0.85, "learning_rate": 4.469275841346154e-05, "loss": 2.0314, "step": 2828 }, { "epoch": 0.85, "learning_rate": 4.4690880408653845e-05, "loss": 1.8835, "step": 2829 }, { "epoch": 0.85, "learning_rate": 4.468900240384616e-05, "loss": 2.014, "step": 2830 }, { "epoch": 0.85, "learning_rate": 4.468712439903847e-05, "loss": 2.1534, "step": 2831 }, { "epoch": 0.85, "learning_rate": 4.468524639423077e-05, "loss": 2.13, "step": 2832 }, { "epoch": 0.85, "learning_rate": 4.468336838942308e-05, "loss": 2.5988, "step": 2833 }, { "epoch": 0.85, "learning_rate": 4.468149038461539e-05, "loss": 2.2439, "step": 2834 }, { "epoch": 0.85, "learning_rate": 4.4679612379807694e-05, "loss": 2.0789, "step": 2835 }, { "epoch": 0.85, "learning_rate": 4.4677734375e-05, "loss": 2.5194, "step": 2836 }, { "epoch": 0.85, "learning_rate": 4.467585637019231e-05, "loss": 1.9625, "step": 2837 }, { "epoch": 0.85, "learning_rate": 4.4673978365384615e-05, "loss": 2.2059, "step": 2838 }, { "epoch": 0.85, "learning_rate": 4.467210036057693e-05, "loss": 2.2135, "step": 2839 }, { "epoch": 0.85, "learning_rate": 4.467022235576923e-05, "loss": 2.1722, "step": 2840 }, { "epoch": 0.85, "learning_rate": 4.4668344350961536e-05, "loss": 2.1534, "step": 2841 }, { "epoch": 0.85, "learning_rate": 4.466646634615385e-05, "loss": 2.0825, "step": 2842 }, { "epoch": 0.85, "learning_rate": 4.466458834134616e-05, "loss": 2.4599, "step": 2843 }, { "epoch": 0.85, "learning_rate": 4.4662710336538464e-05, "loss": 2.0741, "step": 2844 }, { "epoch": 0.85, "learning_rate": 4.4660832331730776e-05, "loss": 2.4904, "step": 2845 }, { "epoch": 0.86, "learning_rate": 4.465895432692308e-05, "loss": 2.3402, "step": 2846 }, { "epoch": 0.86, "learning_rate": 4.4657076322115385e-05, "loss": 2.1473, "step": 2847 }, { "epoch": 0.86, "learning_rate": 4.46551983173077e-05, "loss": 2.236, "step": 2848 }, { "epoch": 0.86, "learning_rate": 4.46533203125e-05, "loss": 2.3237, "step": 2849 }, { "epoch": 0.86, "learning_rate": 4.4651442307692306e-05, "loss": 2.2527, "step": 2850 }, { "epoch": 0.86, "learning_rate": 4.464956430288462e-05, "loss": 2.0124, "step": 2851 }, { "epoch": 0.86, "learning_rate": 4.464768629807692e-05, "loss": 2.6227, "step": 2852 }, { "epoch": 0.86, "learning_rate": 4.4645808293269234e-05, "loss": 2.4069, "step": 2853 }, { "epoch": 0.86, "learning_rate": 4.464393028846154e-05, "loss": 1.9879, "step": 2854 }, { "epoch": 0.86, "learning_rate": 4.4642052283653844e-05, "loss": 1.755, "step": 2855 }, { "epoch": 0.86, "learning_rate": 4.4640174278846155e-05, "loss": 2.2463, "step": 2856 }, { "epoch": 0.86, "learning_rate": 4.463829627403847e-05, "loss": 2.077, "step": 2857 }, { "epoch": 0.86, "learning_rate": 4.463641826923077e-05, "loss": 2.2799, "step": 2858 }, { "epoch": 0.86, "learning_rate": 4.463454026442308e-05, "loss": 2.2904, "step": 2859 }, { "epoch": 0.86, "learning_rate": 4.463266225961539e-05, "loss": 2.1656, "step": 2860 }, { "epoch": 0.86, "learning_rate": 4.463078425480769e-05, "loss": 2.4864, "step": 2861 }, { "epoch": 0.86, "learning_rate": 4.4628906250000004e-05, "loss": 2.3827, "step": 2862 }, { "epoch": 0.86, "learning_rate": 4.462702824519231e-05, "loss": 1.9823, "step": 2863 }, { "epoch": 0.86, "learning_rate": 4.4625150240384614e-05, "loss": 2.3797, "step": 2864 }, { "epoch": 0.86, "learning_rate": 4.4623272235576925e-05, "loss": 2.0689, "step": 2865 }, { "epoch": 0.86, "learning_rate": 4.462139423076923e-05, "loss": 2.4093, "step": 2866 }, { "epoch": 0.86, "learning_rate": 4.461951622596154e-05, "loss": 2.1444, "step": 2867 }, { "epoch": 0.86, "learning_rate": 4.4617638221153847e-05, "loss": 2.2213, "step": 2868 }, { "epoch": 0.86, "learning_rate": 4.461576021634616e-05, "loss": 2.1013, "step": 2869 }, { "epoch": 0.86, "learning_rate": 4.461388221153846e-05, "loss": 2.2995, "step": 2870 }, { "epoch": 0.86, "learning_rate": 4.4612004206730774e-05, "loss": 2.2002, "step": 2871 }, { "epoch": 0.86, "learning_rate": 4.461012620192308e-05, "loss": 2.2, "step": 2872 }, { "epoch": 0.86, "learning_rate": 4.460824819711539e-05, "loss": 2.1392, "step": 2873 }, { "epoch": 0.86, "learning_rate": 4.4606370192307696e-05, "loss": 1.9554, "step": 2874 }, { "epoch": 0.86, "learning_rate": 4.46044921875e-05, "loss": 2.4137, "step": 2875 }, { "epoch": 0.86, "learning_rate": 4.460261418269231e-05, "loss": 2.593, "step": 2876 }, { "epoch": 0.86, "learning_rate": 4.460073617788462e-05, "loss": 2.0774, "step": 2877 }, { "epoch": 0.86, "learning_rate": 4.459885817307692e-05, "loss": 2.0335, "step": 2878 }, { "epoch": 0.87, "learning_rate": 4.459698016826923e-05, "loss": 2.4428, "step": 2879 }, { "epoch": 0.87, "learning_rate": 4.459510216346154e-05, "loss": 2.0916, "step": 2880 }, { "epoch": 0.87, "learning_rate": 4.459322415865384e-05, "loss": 2.223, "step": 2881 }, { "epoch": 0.87, "learning_rate": 4.459134615384616e-05, "loss": 2.1787, "step": 2882 }, { "epoch": 0.87, "learning_rate": 4.4589468149038466e-05, "loss": 2.2055, "step": 2883 }, { "epoch": 0.87, "learning_rate": 4.458759014423077e-05, "loss": 2.0418, "step": 2884 }, { "epoch": 0.87, "learning_rate": 4.458571213942308e-05, "loss": 2.2231, "step": 2885 }, { "epoch": 0.87, "learning_rate": 4.458383413461539e-05, "loss": 2.4091, "step": 2886 }, { "epoch": 0.87, "learning_rate": 4.45819561298077e-05, "loss": 2.4957, "step": 2887 }, { "epoch": 0.87, "learning_rate": 4.4580078125e-05, "loss": 2.3219, "step": 2888 }, { "epoch": 0.87, "learning_rate": 4.457820012019231e-05, "loss": 1.8498, "step": 2889 }, { "epoch": 0.87, "learning_rate": 4.457632211538462e-05, "loss": 2.1098, "step": 2890 }, { "epoch": 0.87, "learning_rate": 4.4574444110576924e-05, "loss": 2.3937, "step": 2891 }, { "epoch": 0.87, "learning_rate": 4.457256610576923e-05, "loss": 1.9858, "step": 2892 }, { "epoch": 0.87, "learning_rate": 4.457068810096154e-05, "loss": 2.0751, "step": 2893 }, { "epoch": 0.87, "learning_rate": 4.4568810096153845e-05, "loss": 2.3115, "step": 2894 }, { "epoch": 0.87, "learning_rate": 4.456693209134616e-05, "loss": 2.0187, "step": 2895 }, { "epoch": 0.87, "learning_rate": 4.456505408653847e-05, "loss": 1.9962, "step": 2896 }, { "epoch": 0.87, "learning_rate": 4.456317608173077e-05, "loss": 2.3207, "step": 2897 }, { "epoch": 0.87, "learning_rate": 4.456129807692308e-05, "loss": 1.8351, "step": 2898 }, { "epoch": 0.87, "learning_rate": 4.455942007211539e-05, "loss": 2.0429, "step": 2899 }, { "epoch": 0.87, "learning_rate": 4.4557542067307694e-05, "loss": 2.2631, "step": 2900 }, { "epoch": 0.87, "learning_rate": 4.45556640625e-05, "loss": 2.1521, "step": 2901 }, { "epoch": 0.87, "learning_rate": 4.455378605769231e-05, "loss": 2.2655, "step": 2902 }, { "epoch": 0.87, "learning_rate": 4.4551908052884615e-05, "loss": 2.2409, "step": 2903 }, { "epoch": 0.87, "learning_rate": 4.455003004807693e-05, "loss": 1.8843, "step": 2904 }, { "epoch": 0.87, "learning_rate": 4.454815204326923e-05, "loss": 2.1257, "step": 2905 }, { "epoch": 0.87, "learning_rate": 4.4546274038461536e-05, "loss": 2.0087, "step": 2906 }, { "epoch": 0.87, "learning_rate": 4.454439603365385e-05, "loss": 2.1949, "step": 2907 }, { "epoch": 0.87, "learning_rate": 4.454251802884616e-05, "loss": 1.8933, "step": 2908 }, { "epoch": 0.87, "learning_rate": 4.4540640024038464e-05, "loss": 2.4847, "step": 2909 }, { "epoch": 0.87, "learning_rate": 4.4538762019230776e-05, "loss": 2.2177, "step": 2910 }, { "epoch": 0.87, "learning_rate": 4.453688401442308e-05, "loss": 1.9978, "step": 2911 }, { "epoch": 0.88, "learning_rate": 4.4535006009615385e-05, "loss": 1.8307, "step": 2912 }, { "epoch": 0.88, "learning_rate": 4.45331280048077e-05, "loss": 2.2279, "step": 2913 }, { "epoch": 0.88, "learning_rate": 4.453125e-05, "loss": 2.379, "step": 2914 }, { "epoch": 0.88, "learning_rate": 4.4529371995192306e-05, "loss": 2.1811, "step": 2915 }, { "epoch": 0.88, "learning_rate": 4.452749399038462e-05, "loss": 2.2092, "step": 2916 }, { "epoch": 0.88, "learning_rate": 4.452561598557692e-05, "loss": 2.2825, "step": 2917 }, { "epoch": 0.88, "learning_rate": 4.4523737980769234e-05, "loss": 2.332, "step": 2918 }, { "epoch": 0.88, "learning_rate": 4.452185997596154e-05, "loss": 2.3683, "step": 2919 }, { "epoch": 0.88, "learning_rate": 4.4519981971153844e-05, "loss": 2.0517, "step": 2920 }, { "epoch": 0.88, "learning_rate": 4.4518103966346155e-05, "loss": 2.2065, "step": 2921 }, { "epoch": 0.88, "learning_rate": 4.451622596153847e-05, "loss": 2.1842, "step": 2922 }, { "epoch": 0.88, "learning_rate": 4.451434795673077e-05, "loss": 2.3104, "step": 2923 }, { "epoch": 0.88, "learning_rate": 4.451246995192308e-05, "loss": 2.363, "step": 2924 }, { "epoch": 0.88, "learning_rate": 4.451059194711539e-05, "loss": 2.1839, "step": 2925 }, { "epoch": 0.88, "learning_rate": 4.450871394230769e-05, "loss": 1.8655, "step": 2926 }, { "epoch": 0.88, "learning_rate": 4.4506835937500004e-05, "loss": 2.2633, "step": 2927 }, { "epoch": 0.88, "learning_rate": 4.450495793269231e-05, "loss": 2.6878, "step": 2928 }, { "epoch": 0.88, "learning_rate": 4.4503079927884614e-05, "loss": 2.1489, "step": 2929 }, { "epoch": 0.88, "learning_rate": 4.4501201923076925e-05, "loss": 2.0934, "step": 2930 }, { "epoch": 0.88, "learning_rate": 4.449932391826923e-05, "loss": 2.5305, "step": 2931 }, { "epoch": 0.88, "learning_rate": 4.449744591346154e-05, "loss": 2.309, "step": 2932 }, { "epoch": 0.88, "learning_rate": 4.4495567908653846e-05, "loss": 2.171, "step": 2933 }, { "epoch": 0.88, "learning_rate": 4.449368990384616e-05, "loss": 2.4283, "step": 2934 }, { "epoch": 0.88, "learning_rate": 4.449181189903846e-05, "loss": 1.8707, "step": 2935 }, { "epoch": 0.88, "learning_rate": 4.4489933894230774e-05, "loss": 2.2054, "step": 2936 }, { "epoch": 0.88, "learning_rate": 4.448805588942308e-05, "loss": 2.042, "step": 2937 }, { "epoch": 0.88, "learning_rate": 4.448617788461539e-05, "loss": 2.3811, "step": 2938 }, { "epoch": 0.88, "learning_rate": 4.4484299879807695e-05, "loss": 2.4559, "step": 2939 }, { "epoch": 0.88, "learning_rate": 4.4482421875e-05, "loss": 1.9943, "step": 2940 }, { "epoch": 0.88, "learning_rate": 4.448054387019231e-05, "loss": 2.1778, "step": 2941 }, { "epoch": 0.88, "learning_rate": 4.4478665865384617e-05, "loss": 2.0738, "step": 2942 }, { "epoch": 0.88, "learning_rate": 4.447678786057692e-05, "loss": 2.5897, "step": 2943 }, { "epoch": 0.88, "learning_rate": 4.447490985576923e-05, "loss": 2.184, "step": 2944 }, { "epoch": 0.88, "learning_rate": 4.447303185096154e-05, "loss": 2.0948, "step": 2945 }, { "epoch": 0.89, "learning_rate": 4.447115384615384e-05, "loss": 2.3349, "step": 2946 }, { "epoch": 0.89, "learning_rate": 4.446927584134616e-05, "loss": 2.3062, "step": 2947 }, { "epoch": 0.89, "learning_rate": 4.4467397836538465e-05, "loss": 2.2903, "step": 2948 }, { "epoch": 0.89, "learning_rate": 4.446551983173077e-05, "loss": 2.2217, "step": 2949 }, { "epoch": 0.89, "learning_rate": 4.446364182692308e-05, "loss": 2.2034, "step": 2950 }, { "epoch": 0.89, "learning_rate": 4.4461763822115387e-05, "loss": 2.4493, "step": 2951 }, { "epoch": 0.89, "learning_rate": 4.44598858173077e-05, "loss": 2.4452, "step": 2952 }, { "epoch": 0.89, "learning_rate": 4.44580078125e-05, "loss": 2.2361, "step": 2953 }, { "epoch": 0.89, "learning_rate": 4.445612980769231e-05, "loss": 2.1635, "step": 2954 }, { "epoch": 0.89, "learning_rate": 4.445425180288462e-05, "loss": 2.2226, "step": 2955 }, { "epoch": 0.89, "learning_rate": 4.4452373798076924e-05, "loss": 1.946, "step": 2956 }, { "epoch": 0.89, "learning_rate": 4.445049579326923e-05, "loss": 1.9984, "step": 2957 }, { "epoch": 0.89, "learning_rate": 4.444861778846154e-05, "loss": 2.0704, "step": 2958 }, { "epoch": 0.89, "learning_rate": 4.4446739783653845e-05, "loss": 2.2319, "step": 2959 }, { "epoch": 0.89, "learning_rate": 4.444486177884616e-05, "loss": 2.1, "step": 2960 }, { "epoch": 0.89, "learning_rate": 4.444298377403847e-05, "loss": 2.3534, "step": 2961 }, { "epoch": 0.89, "learning_rate": 4.444110576923077e-05, "loss": 2.0545, "step": 2962 }, { "epoch": 0.89, "learning_rate": 4.443922776442308e-05, "loss": 2.0174, "step": 2963 }, { "epoch": 0.89, "learning_rate": 4.443734975961539e-05, "loss": 2.1188, "step": 2964 }, { "epoch": 0.89, "learning_rate": 4.4435471754807694e-05, "loss": 2.1533, "step": 2965 }, { "epoch": 0.89, "learning_rate": 4.443359375e-05, "loss": 2.4659, "step": 2966 }, { "epoch": 0.89, "learning_rate": 4.443171574519231e-05, "loss": 2.2722, "step": 2967 }, { "epoch": 0.89, "learning_rate": 4.4429837740384615e-05, "loss": 2.0805, "step": 2968 }, { "epoch": 0.89, "learning_rate": 4.442795973557693e-05, "loss": 2.3571, "step": 2969 }, { "epoch": 0.89, "learning_rate": 4.442608173076923e-05, "loss": 2.0742, "step": 2970 }, { "epoch": 0.89, "learning_rate": 4.4424203725961536e-05, "loss": 2.3103, "step": 2971 }, { "epoch": 0.89, "learning_rate": 4.442232572115385e-05, "loss": 2.1149, "step": 2972 }, { "epoch": 0.89, "learning_rate": 4.442044771634616e-05, "loss": 2.082, "step": 2973 }, { "epoch": 0.89, "learning_rate": 4.4418569711538464e-05, "loss": 2.416, "step": 2974 }, { "epoch": 0.89, "learning_rate": 4.4416691706730776e-05, "loss": 2.1664, "step": 2975 }, { "epoch": 0.89, "learning_rate": 4.441481370192308e-05, "loss": 2.1576, "step": 2976 }, { "epoch": 0.89, "learning_rate": 4.4412935697115385e-05, "loss": 2.253, "step": 2977 }, { "epoch": 0.89, "learning_rate": 4.44110576923077e-05, "loss": 2.3068, "step": 2978 }, { "epoch": 0.9, "learning_rate": 4.44091796875e-05, "loss": 2.2957, "step": 2979 }, { "epoch": 0.9, "learning_rate": 4.4407301682692306e-05, "loss": 2.1006, "step": 2980 }, { "epoch": 0.9, "learning_rate": 4.440542367788462e-05, "loss": 2.5327, "step": 2981 }, { "epoch": 0.9, "learning_rate": 4.440354567307692e-05, "loss": 2.1336, "step": 2982 }, { "epoch": 0.9, "learning_rate": 4.4401667668269234e-05, "loss": 2.1697, "step": 2983 }, { "epoch": 0.9, "learning_rate": 4.439978966346154e-05, "loss": 1.9377, "step": 2984 }, { "epoch": 0.9, "learning_rate": 4.4397911658653844e-05, "loss": 2.2452, "step": 2985 }, { "epoch": 0.9, "learning_rate": 4.4396033653846155e-05, "loss": 1.8199, "step": 2986 }, { "epoch": 0.9, "learning_rate": 4.439415564903847e-05, "loss": 2.2652, "step": 2987 }, { "epoch": 0.9, "learning_rate": 4.439227764423077e-05, "loss": 2.0441, "step": 2988 }, { "epoch": 0.9, "learning_rate": 4.439039963942308e-05, "loss": 2.2336, "step": 2989 }, { "epoch": 0.9, "learning_rate": 4.438852163461539e-05, "loss": 2.3831, "step": 2990 }, { "epoch": 0.9, "learning_rate": 4.438664362980769e-05, "loss": 2.0669, "step": 2991 }, { "epoch": 0.9, "learning_rate": 4.4384765625000004e-05, "loss": 2.4189, "step": 2992 }, { "epoch": 0.9, "learning_rate": 4.438288762019231e-05, "loss": 1.952, "step": 2993 }, { "epoch": 0.9, "learning_rate": 4.4381009615384614e-05, "loss": 2.6218, "step": 2994 }, { "epoch": 0.9, "learning_rate": 4.4379131610576925e-05, "loss": 2.145, "step": 2995 }, { "epoch": 0.9, "learning_rate": 4.437725360576923e-05, "loss": 2.428, "step": 2996 }, { "epoch": 0.9, "learning_rate": 4.437537560096154e-05, "loss": 2.3607, "step": 2997 }, { "epoch": 0.9, "learning_rate": 4.4373497596153846e-05, "loss": 2.2701, "step": 2998 }, { "epoch": 0.9, "learning_rate": 4.437161959134616e-05, "loss": 2.4662, "step": 2999 }, { "epoch": 0.9, "learning_rate": 4.436974158653846e-05, "loss": 2.2158, "step": 3000 }, { "epoch": 0.9, "learning_rate": 4.4367863581730774e-05, "loss": 1.8585, "step": 3001 }, { "epoch": 0.9, "learning_rate": 4.436598557692308e-05, "loss": 2.4974, "step": 3002 }, { "epoch": 0.9, "learning_rate": 4.436410757211539e-05, "loss": 2.3536, "step": 3003 }, { "epoch": 0.9, "learning_rate": 4.4362229567307695e-05, "loss": 2.3398, "step": 3004 }, { "epoch": 0.9, "learning_rate": 4.43603515625e-05, "loss": 2.2038, "step": 3005 }, { "epoch": 0.9, "learning_rate": 4.435847355769231e-05, "loss": 2.1953, "step": 3006 }, { "epoch": 0.9, "learning_rate": 4.4356595552884616e-05, "loss": 2.3128, "step": 3007 }, { "epoch": 0.9, "learning_rate": 4.435471754807692e-05, "loss": 2.6437, "step": 3008 }, { "epoch": 0.9, "learning_rate": 4.435283954326923e-05, "loss": 1.9401, "step": 3009 }, { "epoch": 0.9, "learning_rate": 4.435096153846154e-05, "loss": 2.0243, "step": 3010 }, { "epoch": 0.9, "learning_rate": 4.434908353365385e-05, "loss": 2.1903, "step": 3011 }, { "epoch": 0.91, "learning_rate": 4.434720552884616e-05, "loss": 2.0897, "step": 3012 }, { "epoch": 0.91, "learning_rate": 4.4345327524038465e-05, "loss": 2.3658, "step": 3013 }, { "epoch": 0.91, "learning_rate": 4.434344951923077e-05, "loss": 2.1632, "step": 3014 }, { "epoch": 0.91, "learning_rate": 4.434157151442308e-05, "loss": 2.5342, "step": 3015 }, { "epoch": 0.91, "learning_rate": 4.4339693509615386e-05, "loss": 1.8246, "step": 3016 }, { "epoch": 0.91, "learning_rate": 4.43378155048077e-05, "loss": 2.0246, "step": 3017 }, { "epoch": 0.91, "learning_rate": 4.43359375e-05, "loss": 2.187, "step": 3018 }, { "epoch": 0.91, "learning_rate": 4.433405949519231e-05, "loss": 2.1887, "step": 3019 }, { "epoch": 0.91, "learning_rate": 4.433218149038462e-05, "loss": 2.1921, "step": 3020 }, { "epoch": 0.91, "learning_rate": 4.4330303485576924e-05, "loss": 2.0475, "step": 3021 }, { "epoch": 0.91, "learning_rate": 4.432842548076923e-05, "loss": 2.0155, "step": 3022 }, { "epoch": 0.91, "learning_rate": 4.432654747596154e-05, "loss": 2.2481, "step": 3023 }, { "epoch": 0.91, "learning_rate": 4.4324669471153845e-05, "loss": 2.0404, "step": 3024 }, { "epoch": 0.91, "learning_rate": 4.4322791466346157e-05, "loss": 2.1353, "step": 3025 }, { "epoch": 0.91, "learning_rate": 4.432091346153847e-05, "loss": 2.2711, "step": 3026 }, { "epoch": 0.91, "learning_rate": 4.431903545673077e-05, "loss": 1.9388, "step": 3027 }, { "epoch": 0.91, "learning_rate": 4.431715745192308e-05, "loss": 2.2199, "step": 3028 }, { "epoch": 0.91, "learning_rate": 4.431527944711539e-05, "loss": 2.2376, "step": 3029 }, { "epoch": 0.91, "learning_rate": 4.4313401442307694e-05, "loss": 2.1964, "step": 3030 }, { "epoch": 0.91, "learning_rate": 4.4311523437500005e-05, "loss": 1.9866, "step": 3031 }, { "epoch": 0.91, "learning_rate": 4.430964543269231e-05, "loss": 2.1177, "step": 3032 }, { "epoch": 0.91, "learning_rate": 4.4307767427884615e-05, "loss": 2.0279, "step": 3033 }, { "epoch": 0.91, "learning_rate": 4.4305889423076927e-05, "loss": 1.9939, "step": 3034 }, { "epoch": 0.91, "learning_rate": 4.430401141826923e-05, "loss": 2.2244, "step": 3035 }, { "epoch": 0.91, "learning_rate": 4.4302133413461536e-05, "loss": 2.2122, "step": 3036 }, { "epoch": 0.91, "learning_rate": 4.430025540865385e-05, "loss": 2.2909, "step": 3037 }, { "epoch": 0.91, "learning_rate": 4.429837740384616e-05, "loss": 2.3342, "step": 3038 }, { "epoch": 0.91, "learning_rate": 4.4296499399038464e-05, "loss": 2.3496, "step": 3039 }, { "epoch": 0.91, "learning_rate": 4.4294621394230776e-05, "loss": 1.7174, "step": 3040 }, { "epoch": 0.91, "learning_rate": 4.429274338942308e-05, "loss": 2.2044, "step": 3041 }, { "epoch": 0.91, "learning_rate": 4.4290865384615385e-05, "loss": 2.1977, "step": 3042 }, { "epoch": 0.91, "learning_rate": 4.42889873798077e-05, "loss": 2.0551, "step": 3043 }, { "epoch": 0.91, "learning_rate": 4.4287109375e-05, "loss": 2.2193, "step": 3044 }, { "epoch": 0.91, "learning_rate": 4.4285231370192306e-05, "loss": 2.3684, "step": 3045 }, { "epoch": 0.92, "learning_rate": 4.428335336538462e-05, "loss": 1.8207, "step": 3046 }, { "epoch": 0.92, "learning_rate": 4.428147536057692e-05, "loss": 2.5703, "step": 3047 }, { "epoch": 0.92, "learning_rate": 4.4279597355769234e-05, "loss": 2.1744, "step": 3048 }, { "epoch": 0.92, "learning_rate": 4.427771935096154e-05, "loss": 2.1893, "step": 3049 }, { "epoch": 0.92, "learning_rate": 4.4275841346153844e-05, "loss": 2.2375, "step": 3050 }, { "epoch": 0.92, "learning_rate": 4.427396334134616e-05, "loss": 2.0682, "step": 3051 }, { "epoch": 0.92, "learning_rate": 4.427208533653847e-05, "loss": 2.4486, "step": 3052 }, { "epoch": 0.92, "learning_rate": 4.427020733173077e-05, "loss": 2.2117, "step": 3053 }, { "epoch": 0.92, "learning_rate": 4.426832932692308e-05, "loss": 2.3189, "step": 3054 }, { "epoch": 0.92, "learning_rate": 4.426645132211539e-05, "loss": 2.2055, "step": 3055 }, { "epoch": 0.92, "learning_rate": 4.426457331730769e-05, "loss": 1.7718, "step": 3056 }, { "epoch": 0.92, "learning_rate": 4.4262695312500004e-05, "loss": 2.1079, "step": 3057 }, { "epoch": 0.92, "learning_rate": 4.426081730769231e-05, "loss": 2.1303, "step": 3058 }, { "epoch": 0.92, "learning_rate": 4.4258939302884614e-05, "loss": 2.0566, "step": 3059 }, { "epoch": 0.92, "learning_rate": 4.4257061298076925e-05, "loss": 2.289, "step": 3060 }, { "epoch": 0.92, "learning_rate": 4.425518329326923e-05, "loss": 2.3742, "step": 3061 }, { "epoch": 0.92, "learning_rate": 4.425330528846154e-05, "loss": 2.1952, "step": 3062 }, { "epoch": 0.92, "learning_rate": 4.4251427283653846e-05, "loss": 2.1366, "step": 3063 }, { "epoch": 0.92, "learning_rate": 4.424954927884616e-05, "loss": 2.0216, "step": 3064 }, { "epoch": 0.92, "learning_rate": 4.424767127403846e-05, "loss": 2.3995, "step": 3065 }, { "epoch": 0.92, "learning_rate": 4.4245793269230774e-05, "loss": 2.3003, "step": 3066 }, { "epoch": 0.92, "learning_rate": 4.424391526442308e-05, "loss": 2.2168, "step": 3067 }, { "epoch": 0.92, "learning_rate": 4.424203725961539e-05, "loss": 2.0833, "step": 3068 }, { "epoch": 0.92, "learning_rate": 4.4240159254807695e-05, "loss": 2.2272, "step": 3069 }, { "epoch": 0.92, "learning_rate": 4.423828125e-05, "loss": 1.9492, "step": 3070 }, { "epoch": 0.92, "learning_rate": 4.423640324519231e-05, "loss": 2.2424, "step": 3071 }, { "epoch": 0.92, "learning_rate": 4.4234525240384616e-05, "loss": 2.1505, "step": 3072 }, { "epoch": 0.92, "learning_rate": 4.423264723557692e-05, "loss": 2.0518, "step": 3073 }, { "epoch": 0.92, "learning_rate": 4.423076923076923e-05, "loss": 2.1383, "step": 3074 }, { "epoch": 0.92, "learning_rate": 4.422889122596154e-05, "loss": 1.9861, "step": 3075 }, { "epoch": 0.92, "learning_rate": 4.422701322115385e-05, "loss": 2.2463, "step": 3076 }, { "epoch": 0.92, "learning_rate": 4.422513521634616e-05, "loss": 1.8703, "step": 3077 }, { "epoch": 0.92, "learning_rate": 4.4223257211538465e-05, "loss": 2.1463, "step": 3078 }, { "epoch": 0.93, "learning_rate": 4.422137920673077e-05, "loss": 2.1963, "step": 3079 }, { "epoch": 0.93, "learning_rate": 4.421950120192308e-05, "loss": 2.1783, "step": 3080 }, { "epoch": 0.93, "learning_rate": 4.4217623197115386e-05, "loss": 2.4264, "step": 3081 }, { "epoch": 0.93, "learning_rate": 4.42157451923077e-05, "loss": 2.2938, "step": 3082 }, { "epoch": 0.93, "learning_rate": 4.42138671875e-05, "loss": 2.0419, "step": 3083 }, { "epoch": 0.93, "learning_rate": 4.421198918269231e-05, "loss": 2.3681, "step": 3084 }, { "epoch": 0.93, "learning_rate": 4.421011117788462e-05, "loss": 2.1373, "step": 3085 }, { "epoch": 0.93, "learning_rate": 4.4208233173076924e-05, "loss": 2.1332, "step": 3086 }, { "epoch": 0.93, "learning_rate": 4.420635516826923e-05, "loss": 2.045, "step": 3087 }, { "epoch": 0.93, "learning_rate": 4.420447716346154e-05, "loss": 2.1798, "step": 3088 }, { "epoch": 0.93, "learning_rate": 4.4202599158653845e-05, "loss": 2.2559, "step": 3089 }, { "epoch": 0.93, "learning_rate": 4.4200721153846156e-05, "loss": 2.1119, "step": 3090 }, { "epoch": 0.93, "learning_rate": 4.419884314903847e-05, "loss": 2.4109, "step": 3091 }, { "epoch": 0.93, "learning_rate": 4.419696514423077e-05, "loss": 2.0184, "step": 3092 }, { "epoch": 0.93, "learning_rate": 4.419508713942308e-05, "loss": 2.4729, "step": 3093 }, { "epoch": 0.93, "learning_rate": 4.419320913461539e-05, "loss": 2.1213, "step": 3094 }, { "epoch": 0.93, "learning_rate": 4.4191331129807694e-05, "loss": 2.3174, "step": 3095 }, { "epoch": 0.93, "learning_rate": 4.4189453125000005e-05, "loss": 1.9633, "step": 3096 }, { "epoch": 0.93, "learning_rate": 4.418757512019231e-05, "loss": 2.0938, "step": 3097 }, { "epoch": 0.93, "learning_rate": 4.4185697115384615e-05, "loss": 1.935, "step": 3098 }, { "epoch": 0.93, "learning_rate": 4.4183819110576926e-05, "loss": 2.3568, "step": 3099 }, { "epoch": 0.93, "learning_rate": 4.418194110576923e-05, "loss": 2.1787, "step": 3100 }, { "epoch": 0.93, "learning_rate": 4.4180063100961536e-05, "loss": 1.8013, "step": 3101 }, { "epoch": 0.93, "learning_rate": 4.417818509615385e-05, "loss": 2.0888, "step": 3102 }, { "epoch": 0.93, "learning_rate": 4.417630709134616e-05, "loss": 2.2289, "step": 3103 }, { "epoch": 0.93, "learning_rate": 4.4174429086538464e-05, "loss": 2.3416, "step": 3104 }, { "epoch": 0.93, "learning_rate": 4.4172551081730775e-05, "loss": 2.3552, "step": 3105 }, { "epoch": 0.93, "learning_rate": 4.417067307692308e-05, "loss": 2.2025, "step": 3106 }, { "epoch": 0.93, "learning_rate": 4.4168795072115385e-05, "loss": 2.1891, "step": 3107 }, { "epoch": 0.93, "learning_rate": 4.4166917067307697e-05, "loss": 2.2831, "step": 3108 }, { "epoch": 0.93, "learning_rate": 4.41650390625e-05, "loss": 1.7727, "step": 3109 }, { "epoch": 0.93, "learning_rate": 4.4163161057692306e-05, "loss": 1.9563, "step": 3110 }, { "epoch": 0.93, "learning_rate": 4.416128305288462e-05, "loss": 2.4866, "step": 3111 }, { "epoch": 0.94, "learning_rate": 4.415940504807692e-05, "loss": 2.1655, "step": 3112 }, { "epoch": 0.94, "learning_rate": 4.4157527043269234e-05, "loss": 2.2927, "step": 3113 }, { "epoch": 0.94, "learning_rate": 4.415564903846154e-05, "loss": 2.0418, "step": 3114 }, { "epoch": 0.94, "learning_rate": 4.4153771033653843e-05, "loss": 2.0362, "step": 3115 }, { "epoch": 0.94, "learning_rate": 4.415189302884616e-05, "loss": 2.1983, "step": 3116 }, { "epoch": 0.94, "learning_rate": 4.4150015024038467e-05, "loss": 2.1637, "step": 3117 }, { "epoch": 0.94, "learning_rate": 4.414813701923077e-05, "loss": 2.2759, "step": 3118 }, { "epoch": 0.94, "learning_rate": 4.414625901442308e-05, "loss": 1.5739, "step": 3119 }, { "epoch": 0.94, "learning_rate": 4.414438100961539e-05, "loss": 2.1231, "step": 3120 }, { "epoch": 0.94, "learning_rate": 4.414250300480769e-05, "loss": 2.2597, "step": 3121 }, { "epoch": 0.94, "learning_rate": 4.4140625000000004e-05, "loss": 2.1771, "step": 3122 }, { "epoch": 0.94, "learning_rate": 4.413874699519231e-05, "loss": 2.1958, "step": 3123 }, { "epoch": 0.94, "learning_rate": 4.4136868990384614e-05, "loss": 2.208, "step": 3124 }, { "epoch": 0.94, "learning_rate": 4.4134990985576925e-05, "loss": 2.1121, "step": 3125 }, { "epoch": 0.94, "learning_rate": 4.413311298076923e-05, "loss": 2.1343, "step": 3126 }, { "epoch": 0.94, "learning_rate": 4.413123497596154e-05, "loss": 2.4344, "step": 3127 }, { "epoch": 0.94, "learning_rate": 4.4129356971153846e-05, "loss": 1.9817, "step": 3128 }, { "epoch": 0.94, "learning_rate": 4.412747896634616e-05, "loss": 1.9504, "step": 3129 }, { "epoch": 0.94, "learning_rate": 4.412560096153846e-05, "loss": 2.0339, "step": 3130 }, { "epoch": 0.94, "learning_rate": 4.4123722956730774e-05, "loss": 2.3939, "step": 3131 }, { "epoch": 0.94, "learning_rate": 4.412184495192308e-05, "loss": 2.1188, "step": 3132 }, { "epoch": 0.94, "learning_rate": 4.411996694711539e-05, "loss": 2.1792, "step": 3133 }, { "epoch": 0.94, "learning_rate": 4.4118088942307695e-05, "loss": 2.3217, "step": 3134 }, { "epoch": 0.94, "learning_rate": 4.41162109375e-05, "loss": 2.2519, "step": 3135 }, { "epoch": 0.94, "learning_rate": 4.411433293269231e-05, "loss": 2.1516, "step": 3136 }, { "epoch": 0.94, "learning_rate": 4.4112454927884616e-05, "loss": 2.1075, "step": 3137 }, { "epoch": 0.94, "learning_rate": 4.411057692307692e-05, "loss": 1.9302, "step": 3138 }, { "epoch": 0.94, "learning_rate": 4.410869891826923e-05, "loss": 2.3033, "step": 3139 }, { "epoch": 0.94, "learning_rate": 4.410682091346154e-05, "loss": 2.0861, "step": 3140 }, { "epoch": 0.94, "learning_rate": 4.410494290865385e-05, "loss": 2.0648, "step": 3141 }, { "epoch": 0.94, "learning_rate": 4.410306490384616e-05, "loss": 2.1038, "step": 3142 }, { "epoch": 0.94, "learning_rate": 4.4101186899038465e-05, "loss": 2.0061, "step": 3143 }, { "epoch": 0.94, "learning_rate": 4.409930889423077e-05, "loss": 2.064, "step": 3144 }, { "epoch": 0.95, "learning_rate": 4.409743088942308e-05, "loss": 2.1478, "step": 3145 }, { "epoch": 0.95, "learning_rate": 4.4095552884615386e-05, "loss": 2.0669, "step": 3146 }, { "epoch": 0.95, "learning_rate": 4.40936748798077e-05, "loss": 2.3321, "step": 3147 }, { "epoch": 0.95, "learning_rate": 4.4091796875e-05, "loss": 1.9598, "step": 3148 }, { "epoch": 0.95, "learning_rate": 4.408991887019231e-05, "loss": 2.2164, "step": 3149 }, { "epoch": 0.95, "learning_rate": 4.408804086538462e-05, "loss": 2.1409, "step": 3150 }, { "epoch": 0.95, "learning_rate": 4.4086162860576924e-05, "loss": 2.2873, "step": 3151 }, { "epoch": 0.95, "learning_rate": 4.408428485576923e-05, "loss": 1.8119, "step": 3152 }, { "epoch": 0.95, "learning_rate": 4.408240685096154e-05, "loss": 2.0519, "step": 3153 }, { "epoch": 0.95, "learning_rate": 4.4080528846153845e-05, "loss": 2.2768, "step": 3154 }, { "epoch": 0.95, "learning_rate": 4.4078650841346156e-05, "loss": 2.2893, "step": 3155 }, { "epoch": 0.95, "learning_rate": 4.407677283653847e-05, "loss": 2.1945, "step": 3156 }, { "epoch": 0.95, "learning_rate": 4.407489483173077e-05, "loss": 2.2655, "step": 3157 }, { "epoch": 0.95, "learning_rate": 4.407301682692308e-05, "loss": 2.5486, "step": 3158 }, { "epoch": 0.95, "learning_rate": 4.407113882211539e-05, "loss": 1.896, "step": 3159 }, { "epoch": 0.95, "learning_rate": 4.4069260817307694e-05, "loss": 2.6342, "step": 3160 }, { "epoch": 0.95, "learning_rate": 4.4067382812500005e-05, "loss": 1.9573, "step": 3161 }, { "epoch": 0.95, "learning_rate": 4.406550480769231e-05, "loss": 2.0315, "step": 3162 }, { "epoch": 0.95, "learning_rate": 4.4063626802884615e-05, "loss": 2.1922, "step": 3163 }, { "epoch": 0.95, "learning_rate": 4.4061748798076926e-05, "loss": 2.1286, "step": 3164 }, { "epoch": 0.95, "learning_rate": 4.405987079326923e-05, "loss": 1.8297, "step": 3165 }, { "epoch": 0.95, "learning_rate": 4.4057992788461536e-05, "loss": 2.1536, "step": 3166 }, { "epoch": 0.95, "learning_rate": 4.405611478365385e-05, "loss": 2.1431, "step": 3167 }, { "epoch": 0.95, "learning_rate": 4.405423677884616e-05, "loss": 2.3868, "step": 3168 }, { "epoch": 0.95, "learning_rate": 4.4052358774038464e-05, "loss": 2.0607, "step": 3169 }, { "epoch": 0.95, "learning_rate": 4.4050480769230775e-05, "loss": 2.1458, "step": 3170 }, { "epoch": 0.95, "learning_rate": 4.404860276442308e-05, "loss": 2.046, "step": 3171 }, { "epoch": 0.95, "learning_rate": 4.4046724759615385e-05, "loss": 2.0172, "step": 3172 }, { "epoch": 0.95, "learning_rate": 4.4044846754807696e-05, "loss": 2.3091, "step": 3173 }, { "epoch": 0.95, "learning_rate": 4.404296875e-05, "loss": 2.2159, "step": 3174 }, { "epoch": 0.95, "learning_rate": 4.404109074519231e-05, "loss": 2.3761, "step": 3175 }, { "epoch": 0.95, "learning_rate": 4.403921274038462e-05, "loss": 2.2123, "step": 3176 }, { "epoch": 0.95, "learning_rate": 4.403733473557692e-05, "loss": 2.1903, "step": 3177 }, { "epoch": 0.95, "learning_rate": 4.4035456730769234e-05, "loss": 2.1356, "step": 3178 }, { "epoch": 0.96, "learning_rate": 4.403357872596154e-05, "loss": 2.3833, "step": 3179 }, { "epoch": 0.96, "learning_rate": 4.403170072115384e-05, "loss": 2.4983, "step": 3180 }, { "epoch": 0.96, "learning_rate": 4.402982271634616e-05, "loss": 2.1421, "step": 3181 }, { "epoch": 0.96, "learning_rate": 4.4027944711538466e-05, "loss": 1.9992, "step": 3182 }, { "epoch": 0.96, "learning_rate": 4.402606670673077e-05, "loss": 1.9979, "step": 3183 }, { "epoch": 0.96, "learning_rate": 4.402418870192308e-05, "loss": 2.1346, "step": 3184 }, { "epoch": 0.96, "learning_rate": 4.402231069711539e-05, "loss": 1.916, "step": 3185 }, { "epoch": 0.96, "learning_rate": 4.402043269230769e-05, "loss": 2.2161, "step": 3186 }, { "epoch": 0.96, "learning_rate": 4.4018554687500004e-05, "loss": 2.0616, "step": 3187 }, { "epoch": 0.96, "learning_rate": 4.401667668269231e-05, "loss": 2.2428, "step": 3188 }, { "epoch": 0.96, "learning_rate": 4.4014798677884613e-05, "loss": 2.0814, "step": 3189 }, { "epoch": 0.96, "learning_rate": 4.4012920673076925e-05, "loss": 2.223, "step": 3190 }, { "epoch": 0.96, "learning_rate": 4.401104266826923e-05, "loss": 2.0646, "step": 3191 }, { "epoch": 0.96, "learning_rate": 4.400916466346154e-05, "loss": 2.0643, "step": 3192 }, { "epoch": 0.96, "learning_rate": 4.4007286658653846e-05, "loss": 2.5788, "step": 3193 }, { "epoch": 0.96, "learning_rate": 4.400540865384616e-05, "loss": 2.2282, "step": 3194 }, { "epoch": 0.96, "learning_rate": 4.400353064903847e-05, "loss": 2.2908, "step": 3195 }, { "epoch": 0.96, "learning_rate": 4.4001652644230774e-05, "loss": 2.187, "step": 3196 }, { "epoch": 0.96, "learning_rate": 4.399977463942308e-05, "loss": 2.146, "step": 3197 }, { "epoch": 0.96, "learning_rate": 4.399789663461539e-05, "loss": 2.2327, "step": 3198 }, { "epoch": 0.96, "learning_rate": 4.3996018629807695e-05, "loss": 2.3328, "step": 3199 }, { "epoch": 0.96, "learning_rate": 4.3994140625e-05, "loss": 2.1435, "step": 3200 }, { "epoch": 0.96, "learning_rate": 4.399226262019231e-05, "loss": 2.2323, "step": 3201 }, { "epoch": 0.96, "learning_rate": 4.3990384615384616e-05, "loss": 2.2551, "step": 3202 }, { "epoch": 0.96, "learning_rate": 4.398850661057692e-05, "loss": 2.0716, "step": 3203 }, { "epoch": 0.96, "learning_rate": 4.398662860576923e-05, "loss": 2.2923, "step": 3204 }, { "epoch": 0.96, "learning_rate": 4.398475060096154e-05, "loss": 2.051, "step": 3205 }, { "epoch": 0.96, "learning_rate": 4.398287259615385e-05, "loss": 2.3675, "step": 3206 }, { "epoch": 0.96, "learning_rate": 4.398099459134616e-05, "loss": 2.026, "step": 3207 }, { "epoch": 0.96, "learning_rate": 4.3979116586538465e-05, "loss": 2.0827, "step": 3208 }, { "epoch": 0.96, "learning_rate": 4.397723858173077e-05, "loss": 2.1992, "step": 3209 }, { "epoch": 0.96, "learning_rate": 4.397536057692308e-05, "loss": 2.0897, "step": 3210 }, { "epoch": 0.96, "learning_rate": 4.3973482572115386e-05, "loss": 2.2338, "step": 3211 }, { "epoch": 0.97, "learning_rate": 4.39716045673077e-05, "loss": 2.0173, "step": 3212 }, { "epoch": 0.97, "learning_rate": 4.39697265625e-05, "loss": 2.0022, "step": 3213 }, { "epoch": 0.97, "learning_rate": 4.396784855769231e-05, "loss": 2.0395, "step": 3214 }, { "epoch": 0.97, "learning_rate": 4.396597055288462e-05, "loss": 2.2559, "step": 3215 }, { "epoch": 0.97, "learning_rate": 4.3964092548076924e-05, "loss": 2.4496, "step": 3216 }, { "epoch": 0.97, "learning_rate": 4.396221454326923e-05, "loss": 1.9542, "step": 3217 }, { "epoch": 0.97, "learning_rate": 4.396033653846154e-05, "loss": 1.9506, "step": 3218 }, { "epoch": 0.97, "learning_rate": 4.3958458533653845e-05, "loss": 2.185, "step": 3219 }, { "epoch": 0.97, "learning_rate": 4.3956580528846156e-05, "loss": 2.3053, "step": 3220 }, { "epoch": 0.97, "learning_rate": 4.395470252403847e-05, "loss": 2.1722, "step": 3221 }, { "epoch": 0.97, "learning_rate": 4.395282451923077e-05, "loss": 2.2406, "step": 3222 }, { "epoch": 0.97, "learning_rate": 4.395094651442308e-05, "loss": 2.0418, "step": 3223 }, { "epoch": 0.97, "learning_rate": 4.394906850961539e-05, "loss": 2.4299, "step": 3224 }, { "epoch": 0.97, "learning_rate": 4.3947190504807694e-05, "loss": 2.1212, "step": 3225 }, { "epoch": 0.97, "learning_rate": 4.3945312500000005e-05, "loss": 2.3411, "step": 3226 }, { "epoch": 0.97, "learning_rate": 4.394343449519231e-05, "loss": 2.2394, "step": 3227 }, { "epoch": 0.97, "learning_rate": 4.3941556490384615e-05, "loss": 1.8657, "step": 3228 }, { "epoch": 0.97, "learning_rate": 4.3939678485576926e-05, "loss": 2.1776, "step": 3229 }, { "epoch": 0.97, "learning_rate": 4.393780048076923e-05, "loss": 2.2135, "step": 3230 }, { "epoch": 0.97, "learning_rate": 4.3935922475961536e-05, "loss": 1.7788, "step": 3231 }, { "epoch": 0.97, "learning_rate": 4.393404447115385e-05, "loss": 2.1595, "step": 3232 }, { "epoch": 0.97, "learning_rate": 4.393216646634616e-05, "loss": 2.4013, "step": 3233 }, { "epoch": 0.97, "learning_rate": 4.3930288461538464e-05, "loss": 2.0901, "step": 3234 }, { "epoch": 0.97, "learning_rate": 4.3928410456730775e-05, "loss": 2.2068, "step": 3235 }, { "epoch": 0.97, "learning_rate": 4.392653245192308e-05, "loss": 2.6395, "step": 3236 }, { "epoch": 0.97, "learning_rate": 4.3924654447115385e-05, "loss": 2.3185, "step": 3237 }, { "epoch": 0.97, "learning_rate": 4.3922776442307696e-05, "loss": 2.3459, "step": 3238 }, { "epoch": 0.97, "learning_rate": 4.39208984375e-05, "loss": 2.0772, "step": 3239 }, { "epoch": 0.97, "learning_rate": 4.391902043269231e-05, "loss": 1.8894, "step": 3240 }, { "epoch": 0.97, "learning_rate": 4.391714242788462e-05, "loss": 2.6508, "step": 3241 }, { "epoch": 0.97, "learning_rate": 4.391526442307692e-05, "loss": 2.1976, "step": 3242 }, { "epoch": 0.97, "learning_rate": 4.3913386418269234e-05, "loss": 2.0709, "step": 3243 }, { "epoch": 0.97, "learning_rate": 4.391150841346154e-05, "loss": 2.1088, "step": 3244 }, { "epoch": 0.98, "learning_rate": 4.390963040865384e-05, "loss": 2.0117, "step": 3245 }, { "epoch": 0.98, "learning_rate": 4.390775240384616e-05, "loss": 1.7219, "step": 3246 }, { "epoch": 0.98, "learning_rate": 4.3905874399038466e-05, "loss": 2.1529, "step": 3247 }, { "epoch": 0.98, "learning_rate": 4.390399639423077e-05, "loss": 2.17, "step": 3248 }, { "epoch": 0.98, "learning_rate": 4.390211838942308e-05, "loss": 2.0937, "step": 3249 }, { "epoch": 0.98, "learning_rate": 4.390024038461539e-05, "loss": 2.1703, "step": 3250 }, { "epoch": 0.98, "learning_rate": 4.389836237980769e-05, "loss": 2.2861, "step": 3251 }, { "epoch": 0.98, "learning_rate": 4.3896484375000004e-05, "loss": 2.4706, "step": 3252 }, { "epoch": 0.98, "learning_rate": 4.389460637019231e-05, "loss": 2.251, "step": 3253 }, { "epoch": 0.98, "learning_rate": 4.389272836538461e-05, "loss": 2.2529, "step": 3254 }, { "epoch": 0.98, "learning_rate": 4.3890850360576925e-05, "loss": 2.1116, "step": 3255 }, { "epoch": 0.98, "learning_rate": 4.388897235576923e-05, "loss": 2.1035, "step": 3256 }, { "epoch": 0.98, "learning_rate": 4.388709435096154e-05, "loss": 2.2284, "step": 3257 }, { "epoch": 0.98, "learning_rate": 4.3885216346153846e-05, "loss": 2.1878, "step": 3258 }, { "epoch": 0.98, "learning_rate": 4.388333834134616e-05, "loss": 1.8985, "step": 3259 }, { "epoch": 0.98, "learning_rate": 4.388146033653847e-05, "loss": 2.4946, "step": 3260 }, { "epoch": 0.98, "learning_rate": 4.3879582331730774e-05, "loss": 2.4037, "step": 3261 }, { "epoch": 0.98, "learning_rate": 4.387770432692308e-05, "loss": 2.0733, "step": 3262 }, { "epoch": 0.98, "learning_rate": 4.387582632211539e-05, "loss": 2.1176, "step": 3263 }, { "epoch": 0.98, "learning_rate": 4.3873948317307695e-05, "loss": 2.1864, "step": 3264 }, { "epoch": 0.98, "learning_rate": 4.38720703125e-05, "loss": 2.1223, "step": 3265 }, { "epoch": 0.98, "learning_rate": 4.387019230769231e-05, "loss": 2.363, "step": 3266 }, { "epoch": 0.98, "learning_rate": 4.3868314302884616e-05, "loss": 2.1981, "step": 3267 }, { "epoch": 0.98, "learning_rate": 4.386643629807692e-05, "loss": 2.1797, "step": 3268 }, { "epoch": 0.98, "learning_rate": 4.386455829326923e-05, "loss": 2.0734, "step": 3269 }, { "epoch": 0.98, "learning_rate": 4.386268028846154e-05, "loss": 2.2755, "step": 3270 }, { "epoch": 0.98, "learning_rate": 4.386080228365385e-05, "loss": 2.0478, "step": 3271 }, { "epoch": 0.98, "learning_rate": 4.385892427884616e-05, "loss": 2.2295, "step": 3272 }, { "epoch": 0.98, "learning_rate": 4.3857046274038465e-05, "loss": 1.9999, "step": 3273 }, { "epoch": 0.98, "learning_rate": 4.385516826923077e-05, "loss": 2.3241, "step": 3274 }, { "epoch": 0.98, "learning_rate": 4.385329026442308e-05, "loss": 2.0409, "step": 3275 }, { "epoch": 0.98, "learning_rate": 4.3851412259615386e-05, "loss": 2.5314, "step": 3276 }, { "epoch": 0.98, "learning_rate": 4.38495342548077e-05, "loss": 2.2499, "step": 3277 }, { "epoch": 0.98, "learning_rate": 4.384765625e-05, "loss": 2.1801, "step": 3278 }, { "epoch": 0.99, "learning_rate": 4.384577824519231e-05, "loss": 2.3113, "step": 3279 }, { "epoch": 0.99, "learning_rate": 4.384390024038462e-05, "loss": 2.2457, "step": 3280 }, { "epoch": 0.99, "learning_rate": 4.3842022235576923e-05, "loss": 1.9878, "step": 3281 }, { "epoch": 0.99, "learning_rate": 4.384014423076923e-05, "loss": 1.7701, "step": 3282 }, { "epoch": 0.99, "learning_rate": 4.383826622596154e-05, "loss": 2.0505, "step": 3283 }, { "epoch": 0.99, "learning_rate": 4.3836388221153845e-05, "loss": 2.4855, "step": 3284 }, { "epoch": 0.99, "learning_rate": 4.3834510216346156e-05, "loss": 2.0155, "step": 3285 }, { "epoch": 0.99, "learning_rate": 4.383263221153847e-05, "loss": 2.1812, "step": 3286 }, { "epoch": 0.99, "learning_rate": 4.383075420673077e-05, "loss": 2.2866, "step": 3287 }, { "epoch": 0.99, "learning_rate": 4.382887620192308e-05, "loss": 2.3877, "step": 3288 }, { "epoch": 0.99, "learning_rate": 4.382699819711539e-05, "loss": 2.3139, "step": 3289 }, { "epoch": 0.99, "learning_rate": 4.3825120192307693e-05, "loss": 2.1315, "step": 3290 }, { "epoch": 0.99, "learning_rate": 4.3823242187500005e-05, "loss": 2.1628, "step": 3291 }, { "epoch": 0.99, "learning_rate": 4.382136418269231e-05, "loss": 2.2403, "step": 3292 }, { "epoch": 0.99, "learning_rate": 4.3819486177884615e-05, "loss": 2.2412, "step": 3293 }, { "epoch": 0.99, "learning_rate": 4.3817608173076926e-05, "loss": 2.1179, "step": 3294 }, { "epoch": 0.99, "learning_rate": 4.381573016826923e-05, "loss": 2.0959, "step": 3295 }, { "epoch": 0.99, "learning_rate": 4.3813852163461536e-05, "loss": 2.0803, "step": 3296 }, { "epoch": 0.99, "learning_rate": 4.381197415865385e-05, "loss": 2.2657, "step": 3297 }, { "epoch": 0.99, "learning_rate": 4.381009615384616e-05, "loss": 2.4587, "step": 3298 }, { "epoch": 0.99, "learning_rate": 4.3808218149038464e-05, "loss": 2.1596, "step": 3299 }, { "epoch": 0.99, "learning_rate": 4.3806340144230775e-05, "loss": 2.1202, "step": 3300 }, { "epoch": 0.99, "learning_rate": 4.380446213942308e-05, "loss": 2.0769, "step": 3301 }, { "epoch": 0.99, "learning_rate": 4.3802584134615385e-05, "loss": 2.4573, "step": 3302 }, { "epoch": 0.99, "learning_rate": 4.3800706129807696e-05, "loss": 2.0912, "step": 3303 }, { "epoch": 0.99, "learning_rate": 4.3798828125e-05, "loss": 2.2326, "step": 3304 }, { "epoch": 0.99, "learning_rate": 4.379695012019231e-05, "loss": 2.0648, "step": 3305 }, { "epoch": 0.99, "learning_rate": 4.379507211538462e-05, "loss": 2.4245, "step": 3306 }, { "epoch": 0.99, "learning_rate": 4.379319411057692e-05, "loss": 2.3415, "step": 3307 }, { "epoch": 0.99, "learning_rate": 4.3791316105769234e-05, "loss": 2.3076, "step": 3308 }, { "epoch": 0.99, "learning_rate": 4.378943810096154e-05, "loss": 2.2604, "step": 3309 }, { "epoch": 0.99, "learning_rate": 4.378756009615384e-05, "loss": 2.0605, "step": 3310 }, { "epoch": 0.99, "learning_rate": 4.378568209134616e-05, "loss": 2.1556, "step": 3311 }, { "epoch": 1.0, "learning_rate": 4.3783804086538466e-05, "loss": 2.2388, "step": 3312 }, { "epoch": 1.0, "learning_rate": 4.378192608173077e-05, "loss": 2.3333, "step": 3313 }, { "epoch": 1.0, "learning_rate": 4.378004807692308e-05, "loss": 2.2228, "step": 3314 }, { "epoch": 1.0, "learning_rate": 4.377817007211539e-05, "loss": 2.1229, "step": 3315 }, { "epoch": 1.0, "learning_rate": 4.377629206730769e-05, "loss": 2.1999, "step": 3316 }, { "epoch": 1.0, "learning_rate": 4.3774414062500004e-05, "loss": 2.3487, "step": 3317 }, { "epoch": 1.0, "learning_rate": 4.377253605769231e-05, "loss": 2.1793, "step": 3318 }, { "epoch": 1.0, "learning_rate": 4.377065805288462e-05, "loss": 2.0642, "step": 3319 }, { "epoch": 1.0, "learning_rate": 4.3768780048076925e-05, "loss": 2.4309, "step": 3320 }, { "epoch": 1.0, "learning_rate": 4.376690204326923e-05, "loss": 2.0556, "step": 3321 }, { "epoch": 1.0, "learning_rate": 4.376502403846154e-05, "loss": 2.0641, "step": 3322 }, { "epoch": 1.0, "learning_rate": 4.3763146033653846e-05, "loss": 2.2333, "step": 3323 }, { "epoch": 1.0, "learning_rate": 4.376126802884616e-05, "loss": 2.1985, "step": 3324 }, { "epoch": 1.0, "learning_rate": 4.375939002403847e-05, "loss": 2.2785, "step": 3325 }, { "epoch": 1.0, "learning_rate": 4.3757512019230774e-05, "loss": 2.1316, "step": 3326 }, { "epoch": 1.0, "learning_rate": 4.375563401442308e-05, "loss": 2.382, "step": 3327 }, { "epoch": 1.0, "learning_rate": 4.375375600961539e-05, "loss": 2.2647, "step": 3328 }, { "epoch": 1.0, "learning_rate": 4.3751878004807695e-05, "loss": 1.2718, "step": 3329 }, { "epoch": 1.0, "learning_rate": 4.375e-05, "loss": 1.2263, "step": 3330 }, { "epoch": 1.0, "learning_rate": 4.374812199519231e-05, "loss": 1.282, "step": 3331 }, { "epoch": 1.0, "learning_rate": 4.3746243990384616e-05, "loss": 1.1321, "step": 3332 }, { "epoch": 1.0, "learning_rate": 4.374436598557692e-05, "loss": 1.0453, "step": 3333 }, { "epoch": 1.0, "learning_rate": 4.374248798076923e-05, "loss": 1.2951, "step": 3334 }, { "epoch": 1.0, "learning_rate": 4.374060997596154e-05, "loss": 0.9454, "step": 3335 }, { "epoch": 1.0, "learning_rate": 4.373873197115385e-05, "loss": 1.3181, "step": 3336 }, { "epoch": 1.0, "learning_rate": 4.373685396634616e-05, "loss": 1.2866, "step": 3337 }, { "epoch": 1.0, "learning_rate": 4.3734975961538465e-05, "loss": 1.0707, "step": 3338 }, { "epoch": 1.0, "learning_rate": 4.3733097956730776e-05, "loss": 0.9933, "step": 3339 }, { "epoch": 1.0, "learning_rate": 4.373121995192308e-05, "loss": 1.2735, "step": 3340 }, { "epoch": 1.0, "learning_rate": 4.3729341947115386e-05, "loss": 1.1001, "step": 3341 }, { "epoch": 1.0, "learning_rate": 4.37274639423077e-05, "loss": 1.1553, "step": 3342 }, { "epoch": 1.0, "learning_rate": 4.37255859375e-05, "loss": 1.1717, "step": 3343 }, { "epoch": 1.0, "learning_rate": 4.372370793269231e-05, "loss": 0.9363, "step": 3344 }, { "epoch": 1.01, "learning_rate": 4.372182992788462e-05, "loss": 1.2932, "step": 3345 }, { "epoch": 1.01, "learning_rate": 4.371995192307692e-05, "loss": 1.0855, "step": 3346 }, { "epoch": 1.01, "learning_rate": 4.371807391826923e-05, "loss": 1.2776, "step": 3347 }, { "epoch": 1.01, "learning_rate": 4.371619591346154e-05, "loss": 1.305, "step": 3348 }, { "epoch": 1.01, "learning_rate": 4.3714317908653844e-05, "loss": 1.0124, "step": 3349 }, { "epoch": 1.01, "learning_rate": 4.3712439903846156e-05, "loss": 1.1958, "step": 3350 }, { "epoch": 1.01, "learning_rate": 4.371056189903847e-05, "loss": 1.3408, "step": 3351 }, { "epoch": 1.01, "learning_rate": 4.370868389423077e-05, "loss": 1.095, "step": 3352 }, { "epoch": 1.01, "learning_rate": 4.370680588942308e-05, "loss": 1.4353, "step": 3353 }, { "epoch": 1.01, "learning_rate": 4.370492788461539e-05, "loss": 1.0872, "step": 3354 }, { "epoch": 1.01, "learning_rate": 4.370304987980769e-05, "loss": 1.361, "step": 3355 }, { "epoch": 1.01, "learning_rate": 4.3701171875000005e-05, "loss": 1.1081, "step": 3356 }, { "epoch": 1.01, "learning_rate": 4.369929387019231e-05, "loss": 1.1673, "step": 3357 }, { "epoch": 1.01, "learning_rate": 4.3697415865384614e-05, "loss": 1.1694, "step": 3358 }, { "epoch": 1.01, "learning_rate": 4.3695537860576926e-05, "loss": 1.4851, "step": 3359 }, { "epoch": 1.01, "learning_rate": 4.369365985576923e-05, "loss": 1.0944, "step": 3360 }, { "epoch": 1.01, "learning_rate": 4.3691781850961536e-05, "loss": 1.0757, "step": 3361 }, { "epoch": 1.01, "learning_rate": 4.368990384615385e-05, "loss": 1.1968, "step": 3362 }, { "epoch": 1.01, "learning_rate": 4.368802584134616e-05, "loss": 1.0577, "step": 3363 }, { "epoch": 1.01, "learning_rate": 4.3686147836538463e-05, "loss": 1.102, "step": 3364 }, { "epoch": 1.01, "learning_rate": 4.3684269831730775e-05, "loss": 0.9948, "step": 3365 }, { "epoch": 1.01, "learning_rate": 4.368239182692308e-05, "loss": 1.1065, "step": 3366 }, { "epoch": 1.01, "learning_rate": 4.3680513822115385e-05, "loss": 1.3541, "step": 3367 }, { "epoch": 1.01, "learning_rate": 4.3678635817307696e-05, "loss": 0.9685, "step": 3368 }, { "epoch": 1.01, "learning_rate": 4.36767578125e-05, "loss": 0.9212, "step": 3369 }, { "epoch": 1.01, "learning_rate": 4.367487980769231e-05, "loss": 1.0741, "step": 3370 }, { "epoch": 1.01, "learning_rate": 4.367300180288462e-05, "loss": 0.8673, "step": 3371 }, { "epoch": 1.01, "learning_rate": 4.367112379807692e-05, "loss": 0.9257, "step": 3372 }, { "epoch": 1.01, "learning_rate": 4.3669245793269233e-05, "loss": 1.2326, "step": 3373 }, { "epoch": 1.01, "learning_rate": 4.366736778846154e-05, "loss": 1.0221, "step": 3374 }, { "epoch": 1.01, "learning_rate": 4.366548978365384e-05, "loss": 1.1987, "step": 3375 }, { "epoch": 1.01, "learning_rate": 4.366361177884616e-05, "loss": 1.1126, "step": 3376 }, { "epoch": 1.01, "learning_rate": 4.3661733774038466e-05, "loss": 1.0425, "step": 3377 }, { "epoch": 1.02, "learning_rate": 4.365985576923077e-05, "loss": 1.0279, "step": 3378 }, { "epoch": 1.02, "learning_rate": 4.365797776442308e-05, "loss": 0.9302, "step": 3379 }, { "epoch": 1.02, "learning_rate": 4.365609975961539e-05, "loss": 0.9231, "step": 3380 }, { "epoch": 1.02, "learning_rate": 4.365422175480769e-05, "loss": 0.814, "step": 3381 }, { "epoch": 1.02, "learning_rate": 4.3652343750000004e-05, "loss": 1.0095, "step": 3382 }, { "epoch": 1.02, "learning_rate": 4.365046574519231e-05, "loss": 1.0721, "step": 3383 }, { "epoch": 1.02, "learning_rate": 4.364858774038462e-05, "loss": 0.96, "step": 3384 }, { "epoch": 1.02, "learning_rate": 4.3646709735576925e-05, "loss": 0.9839, "step": 3385 }, { "epoch": 1.02, "learning_rate": 4.364483173076923e-05, "loss": 0.9815, "step": 3386 }, { "epoch": 1.02, "learning_rate": 4.364295372596154e-05, "loss": 1.2943, "step": 3387 }, { "epoch": 1.02, "learning_rate": 4.3641075721153846e-05, "loss": 1.3749, "step": 3388 }, { "epoch": 1.02, "learning_rate": 4.363919771634616e-05, "loss": 1.2451, "step": 3389 }, { "epoch": 1.02, "learning_rate": 4.363731971153847e-05, "loss": 1.1519, "step": 3390 }, { "epoch": 1.02, "learning_rate": 4.3635441706730774e-05, "loss": 0.8783, "step": 3391 }, { "epoch": 1.02, "learning_rate": 4.363356370192308e-05, "loss": 1.3056, "step": 3392 }, { "epoch": 1.02, "learning_rate": 4.363168569711539e-05, "loss": 1.1386, "step": 3393 }, { "epoch": 1.02, "learning_rate": 4.3629807692307695e-05, "loss": 0.9332, "step": 3394 }, { "epoch": 1.02, "learning_rate": 4.36279296875e-05, "loss": 0.9852, "step": 3395 }, { "epoch": 1.02, "learning_rate": 4.362605168269231e-05, "loss": 0.9771, "step": 3396 }, { "epoch": 1.02, "learning_rate": 4.3624173677884616e-05, "loss": 0.954, "step": 3397 }, { "epoch": 1.02, "learning_rate": 4.362229567307692e-05, "loss": 1.1007, "step": 3398 }, { "epoch": 1.02, "learning_rate": 4.362041766826923e-05, "loss": 1.1793, "step": 3399 }, { "epoch": 1.02, "learning_rate": 4.361853966346154e-05, "loss": 1.0452, "step": 3400 }, { "epoch": 1.02, "learning_rate": 4.361666165865385e-05, "loss": 1.0643, "step": 3401 }, { "epoch": 1.02, "learning_rate": 4.361478365384616e-05, "loss": 0.9848, "step": 3402 }, { "epoch": 1.02, "learning_rate": 4.3612905649038465e-05, "loss": 0.9068, "step": 3403 }, { "epoch": 1.02, "learning_rate": 4.3611027644230776e-05, "loss": 0.9207, "step": 3404 }, { "epoch": 1.02, "learning_rate": 4.360914963942308e-05, "loss": 0.9903, "step": 3405 }, { "epoch": 1.02, "learning_rate": 4.3607271634615386e-05, "loss": 0.997, "step": 3406 }, { "epoch": 1.02, "learning_rate": 4.36053936298077e-05, "loss": 1.0341, "step": 3407 }, { "epoch": 1.02, "learning_rate": 4.3603515625e-05, "loss": 1.1111, "step": 3408 }, { "epoch": 1.02, "learning_rate": 4.360163762019231e-05, "loss": 1.2468, "step": 3409 }, { "epoch": 1.02, "learning_rate": 4.359975961538462e-05, "loss": 1.2606, "step": 3410 }, { "epoch": 1.02, "learning_rate": 4.359788161057692e-05, "loss": 1.128, "step": 3411 }, { "epoch": 1.03, "learning_rate": 4.359600360576923e-05, "loss": 1.1525, "step": 3412 }, { "epoch": 1.03, "learning_rate": 4.359412560096154e-05, "loss": 1.0258, "step": 3413 }, { "epoch": 1.03, "learning_rate": 4.3592247596153844e-05, "loss": 1.182, "step": 3414 }, { "epoch": 1.03, "learning_rate": 4.3590369591346156e-05, "loss": 1.1109, "step": 3415 }, { "epoch": 1.03, "learning_rate": 4.358849158653847e-05, "loss": 1.0546, "step": 3416 }, { "epoch": 1.03, "learning_rate": 4.358661358173077e-05, "loss": 1.2524, "step": 3417 }, { "epoch": 1.03, "learning_rate": 4.358473557692308e-05, "loss": 1.0737, "step": 3418 }, { "epoch": 1.03, "learning_rate": 4.358285757211539e-05, "loss": 0.8423, "step": 3419 }, { "epoch": 1.03, "learning_rate": 4.358097956730769e-05, "loss": 0.915, "step": 3420 }, { "epoch": 1.03, "learning_rate": 4.3579101562500005e-05, "loss": 0.9294, "step": 3421 }, { "epoch": 1.03, "learning_rate": 4.357722355769231e-05, "loss": 1.0679, "step": 3422 }, { "epoch": 1.03, "learning_rate": 4.3575345552884614e-05, "loss": 1.0659, "step": 3423 }, { "epoch": 1.03, "learning_rate": 4.3573467548076926e-05, "loss": 0.8908, "step": 3424 }, { "epoch": 1.03, "learning_rate": 4.357158954326923e-05, "loss": 1.0254, "step": 3425 }, { "epoch": 1.03, "learning_rate": 4.3569711538461535e-05, "loss": 1.0077, "step": 3426 }, { "epoch": 1.03, "learning_rate": 4.356783353365385e-05, "loss": 1.01, "step": 3427 }, { "epoch": 1.03, "learning_rate": 4.356595552884616e-05, "loss": 1.036, "step": 3428 }, { "epoch": 1.03, "learning_rate": 4.356407752403846e-05, "loss": 1.1724, "step": 3429 }, { "epoch": 1.03, "learning_rate": 4.3562199519230775e-05, "loss": 0.9943, "step": 3430 }, { "epoch": 1.03, "learning_rate": 4.356032151442308e-05, "loss": 0.9439, "step": 3431 }, { "epoch": 1.03, "learning_rate": 4.3558443509615384e-05, "loss": 0.9661, "step": 3432 }, { "epoch": 1.03, "learning_rate": 4.3556565504807696e-05, "loss": 1.0607, "step": 3433 }, { "epoch": 1.03, "learning_rate": 4.35546875e-05, "loss": 1.0264, "step": 3434 }, { "epoch": 1.03, "learning_rate": 4.355280949519231e-05, "loss": 1.1043, "step": 3435 }, { "epoch": 1.03, "learning_rate": 4.355093149038462e-05, "loss": 1.0801, "step": 3436 }, { "epoch": 1.03, "learning_rate": 4.354905348557692e-05, "loss": 1.3867, "step": 3437 }, { "epoch": 1.03, "learning_rate": 4.354717548076923e-05, "loss": 1.2007, "step": 3438 }, { "epoch": 1.03, "learning_rate": 4.354529747596154e-05, "loss": 0.9745, "step": 3439 }, { "epoch": 1.03, "learning_rate": 4.354341947115384e-05, "loss": 0.9326, "step": 3440 }, { "epoch": 1.03, "learning_rate": 4.354154146634616e-05, "loss": 0.8026, "step": 3441 }, { "epoch": 1.03, "learning_rate": 4.3539663461538466e-05, "loss": 0.8298, "step": 3442 }, { "epoch": 1.03, "learning_rate": 4.353778545673077e-05, "loss": 1.1131, "step": 3443 }, { "epoch": 1.03, "learning_rate": 4.353590745192308e-05, "loss": 1.1508, "step": 3444 }, { "epoch": 1.04, "learning_rate": 4.353402944711539e-05, "loss": 1.0457, "step": 3445 }, { "epoch": 1.04, "learning_rate": 4.353215144230769e-05, "loss": 1.1542, "step": 3446 }, { "epoch": 1.04, "learning_rate": 4.3530273437500003e-05, "loss": 1.2559, "step": 3447 }, { "epoch": 1.04, "learning_rate": 4.352839543269231e-05, "loss": 1.4211, "step": 3448 }, { "epoch": 1.04, "learning_rate": 4.352651742788462e-05, "loss": 1.3002, "step": 3449 }, { "epoch": 1.04, "learning_rate": 4.3524639423076925e-05, "loss": 0.8068, "step": 3450 }, { "epoch": 1.04, "learning_rate": 4.352276141826923e-05, "loss": 1.2644, "step": 3451 }, { "epoch": 1.04, "learning_rate": 4.352088341346154e-05, "loss": 1.3394, "step": 3452 }, { "epoch": 1.04, "learning_rate": 4.3519005408653846e-05, "loss": 1.2468, "step": 3453 }, { "epoch": 1.04, "learning_rate": 4.351712740384616e-05, "loss": 0.9363, "step": 3454 }, { "epoch": 1.04, "learning_rate": 4.351524939903847e-05, "loss": 1.3494, "step": 3455 }, { "epoch": 1.04, "learning_rate": 4.3513371394230773e-05, "loss": 0.971, "step": 3456 }, { "epoch": 1.04, "learning_rate": 4.351149338942308e-05, "loss": 0.9289, "step": 3457 }, { "epoch": 1.04, "learning_rate": 4.350961538461539e-05, "loss": 1.0166, "step": 3458 }, { "epoch": 1.04, "learning_rate": 4.3507737379807695e-05, "loss": 0.9479, "step": 3459 }, { "epoch": 1.04, "learning_rate": 4.3505859375e-05, "loss": 1.0269, "step": 3460 }, { "epoch": 1.04, "learning_rate": 4.350398137019231e-05, "loss": 1.0112, "step": 3461 }, { "epoch": 1.04, "learning_rate": 4.3502103365384616e-05, "loss": 1.1207, "step": 3462 }, { "epoch": 1.04, "learning_rate": 4.350022536057693e-05, "loss": 0.9346, "step": 3463 }, { "epoch": 1.04, "learning_rate": 4.349834735576923e-05, "loss": 1.1003, "step": 3464 }, { "epoch": 1.04, "learning_rate": 4.349646935096154e-05, "loss": 1.0594, "step": 3465 }, { "epoch": 1.04, "learning_rate": 4.349459134615385e-05, "loss": 0.929, "step": 3466 }, { "epoch": 1.04, "learning_rate": 4.349271334134616e-05, "loss": 0.7117, "step": 3467 }, { "epoch": 1.04, "learning_rate": 4.3490835336538465e-05, "loss": 0.9519, "step": 3468 }, { "epoch": 1.04, "learning_rate": 4.3488957331730776e-05, "loss": 1.1432, "step": 3469 }, { "epoch": 1.04, "learning_rate": 4.348707932692308e-05, "loss": 1.0681, "step": 3470 }, { "epoch": 1.04, "learning_rate": 4.3485201322115386e-05, "loss": 1.05, "step": 3471 }, { "epoch": 1.04, "learning_rate": 4.34833233173077e-05, "loss": 1.0562, "step": 3472 }, { "epoch": 1.04, "learning_rate": 4.34814453125e-05, "loss": 1.1789, "step": 3473 }, { "epoch": 1.04, "learning_rate": 4.347956730769231e-05, "loss": 1.1536, "step": 3474 }, { "epoch": 1.04, "learning_rate": 4.347768930288462e-05, "loss": 0.9569, "step": 3475 }, { "epoch": 1.04, "learning_rate": 4.347581129807692e-05, "loss": 0.8709, "step": 3476 }, { "epoch": 1.04, "learning_rate": 4.347393329326923e-05, "loss": 1.0798, "step": 3477 }, { "epoch": 1.05, "learning_rate": 4.347205528846154e-05, "loss": 1.0561, "step": 3478 }, { "epoch": 1.05, "learning_rate": 4.3470177283653844e-05, "loss": 0.7878, "step": 3479 }, { "epoch": 1.05, "learning_rate": 4.3468299278846156e-05, "loss": 1.2304, "step": 3480 }, { "epoch": 1.05, "learning_rate": 4.346642127403847e-05, "loss": 1.1753, "step": 3481 }, { "epoch": 1.05, "learning_rate": 4.346454326923077e-05, "loss": 1.0323, "step": 3482 }, { "epoch": 1.05, "learning_rate": 4.3462665264423084e-05, "loss": 1.0701, "step": 3483 }, { "epoch": 1.05, "learning_rate": 4.346078725961539e-05, "loss": 0.9781, "step": 3484 }, { "epoch": 1.05, "learning_rate": 4.345890925480769e-05, "loss": 1.0167, "step": 3485 }, { "epoch": 1.05, "learning_rate": 4.3457031250000005e-05, "loss": 1.0056, "step": 3486 }, { "epoch": 1.05, "learning_rate": 4.345515324519231e-05, "loss": 1.0339, "step": 3487 }, { "epoch": 1.05, "learning_rate": 4.3453275240384614e-05, "loss": 0.8982, "step": 3488 }, { "epoch": 1.05, "learning_rate": 4.3451397235576926e-05, "loss": 1.0151, "step": 3489 }, { "epoch": 1.05, "learning_rate": 4.344951923076923e-05, "loss": 0.9024, "step": 3490 }, { "epoch": 1.05, "learning_rate": 4.3447641225961535e-05, "loss": 0.9998, "step": 3491 }, { "epoch": 1.05, "learning_rate": 4.344576322115385e-05, "loss": 1.1568, "step": 3492 }, { "epoch": 1.05, "learning_rate": 4.344388521634616e-05, "loss": 1.0818, "step": 3493 }, { "epoch": 1.05, "learning_rate": 4.344200721153846e-05, "loss": 0.7284, "step": 3494 }, { "epoch": 1.05, "learning_rate": 4.3440129206730775e-05, "loss": 1.149, "step": 3495 }, { "epoch": 1.05, "learning_rate": 4.343825120192308e-05, "loss": 1.3216, "step": 3496 }, { "epoch": 1.05, "learning_rate": 4.3436373197115384e-05, "loss": 1.0932, "step": 3497 }, { "epoch": 1.05, "learning_rate": 4.3434495192307696e-05, "loss": 1.1565, "step": 3498 }, { "epoch": 1.05, "learning_rate": 4.34326171875e-05, "loss": 1.1685, "step": 3499 }, { "epoch": 1.05, "learning_rate": 4.343073918269231e-05, "loss": 0.9683, "step": 3500 }, { "epoch": 1.05, "learning_rate": 4.342886117788462e-05, "loss": 0.9972, "step": 3501 }, { "epoch": 1.05, "learning_rate": 4.342698317307692e-05, "loss": 1.1563, "step": 3502 }, { "epoch": 1.05, "learning_rate": 4.342510516826923e-05, "loss": 1.2618, "step": 3503 }, { "epoch": 1.05, "learning_rate": 4.342322716346154e-05, "loss": 0.9392, "step": 3504 }, { "epoch": 1.05, "learning_rate": 4.342134915865384e-05, "loss": 1.2517, "step": 3505 }, { "epoch": 1.05, "learning_rate": 4.341947115384616e-05, "loss": 0.952, "step": 3506 }, { "epoch": 1.05, "learning_rate": 4.3417593149038466e-05, "loss": 1.0911, "step": 3507 }, { "epoch": 1.05, "learning_rate": 4.341571514423077e-05, "loss": 1.0287, "step": 3508 }, { "epoch": 1.05, "learning_rate": 4.341383713942308e-05, "loss": 0.9813, "step": 3509 }, { "epoch": 1.05, "learning_rate": 4.341195913461539e-05, "loss": 1.2176, "step": 3510 }, { "epoch": 1.05, "learning_rate": 4.341008112980769e-05, "loss": 1.1614, "step": 3511 }, { "epoch": 1.06, "learning_rate": 4.3408203125e-05, "loss": 0.995, "step": 3512 }, { "epoch": 1.06, "learning_rate": 4.340632512019231e-05, "loss": 1.1503, "step": 3513 }, { "epoch": 1.06, "learning_rate": 4.340444711538462e-05, "loss": 1.0493, "step": 3514 }, { "epoch": 1.06, "learning_rate": 4.3402569110576924e-05, "loss": 0.9667, "step": 3515 }, { "epoch": 1.06, "learning_rate": 4.340069110576923e-05, "loss": 0.9826, "step": 3516 }, { "epoch": 1.06, "learning_rate": 4.339881310096154e-05, "loss": 1.0974, "step": 3517 }, { "epoch": 1.06, "learning_rate": 4.3396935096153845e-05, "loss": 1.1041, "step": 3518 }, { "epoch": 1.06, "learning_rate": 4.339505709134616e-05, "loss": 0.9741, "step": 3519 }, { "epoch": 1.06, "learning_rate": 4.339317908653847e-05, "loss": 0.9711, "step": 3520 }, { "epoch": 1.06, "learning_rate": 4.339130108173077e-05, "loss": 0.9158, "step": 3521 }, { "epoch": 1.06, "learning_rate": 4.338942307692308e-05, "loss": 0.8701, "step": 3522 }, { "epoch": 1.06, "learning_rate": 4.338754507211539e-05, "loss": 1.098, "step": 3523 }, { "epoch": 1.06, "learning_rate": 4.3385667067307694e-05, "loss": 1.2416, "step": 3524 }, { "epoch": 1.06, "learning_rate": 4.33837890625e-05, "loss": 1.2297, "step": 3525 }, { "epoch": 1.06, "learning_rate": 4.338191105769231e-05, "loss": 0.742, "step": 3526 }, { "epoch": 1.06, "learning_rate": 4.3380033052884616e-05, "loss": 0.9557, "step": 3527 }, { "epoch": 1.06, "learning_rate": 4.337815504807693e-05, "loss": 1.2508, "step": 3528 }, { "epoch": 1.06, "learning_rate": 4.337627704326923e-05, "loss": 1.1705, "step": 3529 }, { "epoch": 1.06, "learning_rate": 4.337439903846154e-05, "loss": 1.3269, "step": 3530 }, { "epoch": 1.06, "learning_rate": 4.337252103365385e-05, "loss": 0.8903, "step": 3531 }, { "epoch": 1.06, "learning_rate": 4.337064302884616e-05, "loss": 1.2202, "step": 3532 }, { "epoch": 1.06, "learning_rate": 4.3368765024038465e-05, "loss": 1.3003, "step": 3533 }, { "epoch": 1.06, "learning_rate": 4.3366887019230776e-05, "loss": 0.9911, "step": 3534 }, { "epoch": 1.06, "learning_rate": 4.336500901442308e-05, "loss": 1.0138, "step": 3535 }, { "epoch": 1.06, "learning_rate": 4.3363131009615386e-05, "loss": 0.8694, "step": 3536 }, { "epoch": 1.06, "learning_rate": 4.33612530048077e-05, "loss": 1.0398, "step": 3537 }, { "epoch": 1.06, "learning_rate": 4.3359375e-05, "loss": 1.1609, "step": 3538 }, { "epoch": 1.06, "learning_rate": 4.335749699519231e-05, "loss": 1.0385, "step": 3539 }, { "epoch": 1.06, "learning_rate": 4.335561899038462e-05, "loss": 1.107, "step": 3540 }, { "epoch": 1.06, "learning_rate": 4.335374098557692e-05, "loss": 1.0814, "step": 3541 }, { "epoch": 1.06, "learning_rate": 4.335186298076923e-05, "loss": 1.2183, "step": 3542 }, { "epoch": 1.06, "learning_rate": 4.334998497596154e-05, "loss": 0.8952, "step": 3543 }, { "epoch": 1.06, "learning_rate": 4.3348106971153844e-05, "loss": 0.9137, "step": 3544 }, { "epoch": 1.07, "learning_rate": 4.3346228966346156e-05, "loss": 1.3015, "step": 3545 }, { "epoch": 1.07, "learning_rate": 4.334435096153847e-05, "loss": 0.957, "step": 3546 }, { "epoch": 1.07, "learning_rate": 4.334247295673077e-05, "loss": 1.0867, "step": 3547 }, { "epoch": 1.07, "learning_rate": 4.3340594951923084e-05, "loss": 1.1826, "step": 3548 }, { "epoch": 1.07, "learning_rate": 4.333871694711539e-05, "loss": 0.9408, "step": 3549 }, { "epoch": 1.07, "learning_rate": 4.333683894230769e-05, "loss": 0.8228, "step": 3550 }, { "epoch": 1.07, "learning_rate": 4.3334960937500005e-05, "loss": 1.0194, "step": 3551 }, { "epoch": 1.07, "learning_rate": 4.333308293269231e-05, "loss": 0.9976, "step": 3552 }, { "epoch": 1.07, "learning_rate": 4.3331204927884614e-05, "loss": 1.2183, "step": 3553 }, { "epoch": 1.07, "learning_rate": 4.3329326923076926e-05, "loss": 0.9269, "step": 3554 }, { "epoch": 1.07, "learning_rate": 4.332744891826923e-05, "loss": 1.2523, "step": 3555 }, { "epoch": 1.07, "learning_rate": 4.3325570913461535e-05, "loss": 0.9031, "step": 3556 }, { "epoch": 1.07, "learning_rate": 4.332369290865385e-05, "loss": 1.0322, "step": 3557 }, { "epoch": 1.07, "learning_rate": 4.332181490384616e-05, "loss": 1.0304, "step": 3558 }, { "epoch": 1.07, "learning_rate": 4.331993689903846e-05, "loss": 1.027, "step": 3559 }, { "epoch": 1.07, "learning_rate": 4.3318058894230775e-05, "loss": 0.6782, "step": 3560 }, { "epoch": 1.07, "learning_rate": 4.331618088942308e-05, "loss": 1.1466, "step": 3561 }, { "epoch": 1.07, "learning_rate": 4.3314302884615384e-05, "loss": 1.05, "step": 3562 }, { "epoch": 1.07, "learning_rate": 4.3312424879807696e-05, "loss": 1.0969, "step": 3563 }, { "epoch": 1.07, "learning_rate": 4.3310546875e-05, "loss": 1.2221, "step": 3564 }, { "epoch": 1.07, "learning_rate": 4.330866887019231e-05, "loss": 1.304, "step": 3565 }, { "epoch": 1.07, "learning_rate": 4.330679086538462e-05, "loss": 0.8567, "step": 3566 }, { "epoch": 1.07, "learning_rate": 4.330491286057692e-05, "loss": 1.4173, "step": 3567 }, { "epoch": 1.07, "learning_rate": 4.330303485576923e-05, "loss": 0.9253, "step": 3568 }, { "epoch": 1.07, "learning_rate": 4.330115685096154e-05, "loss": 0.8783, "step": 3569 }, { "epoch": 1.07, "learning_rate": 4.329927884615384e-05, "loss": 1.2153, "step": 3570 }, { "epoch": 1.07, "learning_rate": 4.329740084134616e-05, "loss": 1.0203, "step": 3571 }, { "epoch": 1.07, "learning_rate": 4.3295522836538466e-05, "loss": 1.1747, "step": 3572 }, { "epoch": 1.07, "learning_rate": 4.329364483173077e-05, "loss": 1.1489, "step": 3573 }, { "epoch": 1.07, "learning_rate": 4.329176682692308e-05, "loss": 1.2567, "step": 3574 }, { "epoch": 1.07, "learning_rate": 4.328988882211539e-05, "loss": 1.0428, "step": 3575 }, { "epoch": 1.07, "learning_rate": 4.328801081730769e-05, "loss": 1.0577, "step": 3576 }, { "epoch": 1.07, "learning_rate": 4.32861328125e-05, "loss": 1.4098, "step": 3577 }, { "epoch": 1.08, "learning_rate": 4.328425480769231e-05, "loss": 1.1247, "step": 3578 }, { "epoch": 1.08, "learning_rate": 4.328237680288462e-05, "loss": 0.9404, "step": 3579 }, { "epoch": 1.08, "learning_rate": 4.3280498798076924e-05, "loss": 1.1072, "step": 3580 }, { "epoch": 1.08, "learning_rate": 4.327862079326923e-05, "loss": 1.3759, "step": 3581 }, { "epoch": 1.08, "learning_rate": 4.327674278846154e-05, "loss": 0.9926, "step": 3582 }, { "epoch": 1.08, "learning_rate": 4.3274864783653845e-05, "loss": 1.1024, "step": 3583 }, { "epoch": 1.08, "learning_rate": 4.327298677884616e-05, "loss": 0.8724, "step": 3584 }, { "epoch": 1.08, "learning_rate": 4.327110877403847e-05, "loss": 0.9082, "step": 3585 }, { "epoch": 1.08, "learning_rate": 4.326923076923077e-05, "loss": 0.9764, "step": 3586 }, { "epoch": 1.08, "learning_rate": 4.326735276442308e-05, "loss": 1.119, "step": 3587 }, { "epoch": 1.08, "learning_rate": 4.326547475961539e-05, "loss": 0.8449, "step": 3588 }, { "epoch": 1.08, "learning_rate": 4.3263596754807694e-05, "loss": 1.1821, "step": 3589 }, { "epoch": 1.08, "learning_rate": 4.326171875e-05, "loss": 0.9812, "step": 3590 }, { "epoch": 1.08, "learning_rate": 4.325984074519231e-05, "loss": 1.155, "step": 3591 }, { "epoch": 1.08, "learning_rate": 4.3257962740384615e-05, "loss": 1.0678, "step": 3592 }, { "epoch": 1.08, "learning_rate": 4.325608473557693e-05, "loss": 1.1975, "step": 3593 }, { "epoch": 1.08, "learning_rate": 4.325420673076923e-05, "loss": 1.0934, "step": 3594 }, { "epoch": 1.08, "learning_rate": 4.3252328725961537e-05, "loss": 0.9186, "step": 3595 }, { "epoch": 1.08, "learning_rate": 4.325045072115385e-05, "loss": 1.0796, "step": 3596 }, { "epoch": 1.08, "learning_rate": 4.324857271634616e-05, "loss": 0.9319, "step": 3597 }, { "epoch": 1.08, "learning_rate": 4.3246694711538464e-05, "loss": 1.2604, "step": 3598 }, { "epoch": 1.08, "learning_rate": 4.3244816706730776e-05, "loss": 1.0741, "step": 3599 }, { "epoch": 1.08, "learning_rate": 4.324293870192308e-05, "loss": 1.171, "step": 3600 }, { "epoch": 1.08, "learning_rate": 4.3241060697115385e-05, "loss": 1.0711, "step": 3601 }, { "epoch": 1.08, "learning_rate": 4.32391826923077e-05, "loss": 1.0007, "step": 3602 }, { "epoch": 1.08, "learning_rate": 4.32373046875e-05, "loss": 0.9781, "step": 3603 }, { "epoch": 1.08, "learning_rate": 4.3235426682692307e-05, "loss": 0.8033, "step": 3604 }, { "epoch": 1.08, "learning_rate": 4.323354867788462e-05, "loss": 1.2153, "step": 3605 }, { "epoch": 1.08, "learning_rate": 4.323167067307692e-05, "loss": 0.8343, "step": 3606 }, { "epoch": 1.08, "learning_rate": 4.3229792668269234e-05, "loss": 1.2727, "step": 3607 }, { "epoch": 1.08, "learning_rate": 4.322791466346154e-05, "loss": 1.2627, "step": 3608 }, { "epoch": 1.08, "learning_rate": 4.3226036658653844e-05, "loss": 1.1473, "step": 3609 }, { "epoch": 1.08, "learning_rate": 4.3224158653846156e-05, "loss": 1.0621, "step": 3610 }, { "epoch": 1.09, "learning_rate": 4.322228064903847e-05, "loss": 1.3701, "step": 3611 }, { "epoch": 1.09, "learning_rate": 4.322040264423077e-05, "loss": 1.2571, "step": 3612 }, { "epoch": 1.09, "learning_rate": 4.3218524639423083e-05, "loss": 1.0774, "step": 3613 }, { "epoch": 1.09, "learning_rate": 4.321664663461539e-05, "loss": 1.189, "step": 3614 }, { "epoch": 1.09, "learning_rate": 4.321476862980769e-05, "loss": 1.0357, "step": 3615 }, { "epoch": 1.09, "learning_rate": 4.3212890625000004e-05, "loss": 1.0273, "step": 3616 }, { "epoch": 1.09, "learning_rate": 4.321101262019231e-05, "loss": 1.2156, "step": 3617 }, { "epoch": 1.09, "learning_rate": 4.3209134615384614e-05, "loss": 1.1277, "step": 3618 }, { "epoch": 1.09, "learning_rate": 4.3207256610576926e-05, "loss": 1.1111, "step": 3619 }, { "epoch": 1.09, "learning_rate": 4.320537860576923e-05, "loss": 1.1231, "step": 3620 }, { "epoch": 1.09, "learning_rate": 4.3203500600961535e-05, "loss": 0.9827, "step": 3621 }, { "epoch": 1.09, "learning_rate": 4.320162259615385e-05, "loss": 0.8849, "step": 3622 }, { "epoch": 1.09, "learning_rate": 4.319974459134616e-05, "loss": 1.0503, "step": 3623 }, { "epoch": 1.09, "learning_rate": 4.319786658653846e-05, "loss": 0.9216, "step": 3624 }, { "epoch": 1.09, "learning_rate": 4.3195988581730775e-05, "loss": 0.841, "step": 3625 }, { "epoch": 1.09, "learning_rate": 4.319411057692308e-05, "loss": 1.0868, "step": 3626 }, { "epoch": 1.09, "learning_rate": 4.319223257211539e-05, "loss": 0.737, "step": 3627 }, { "epoch": 1.09, "learning_rate": 4.3190354567307696e-05, "loss": 1.1092, "step": 3628 }, { "epoch": 1.09, "learning_rate": 4.31884765625e-05, "loss": 1.2936, "step": 3629 }, { "epoch": 1.09, "learning_rate": 4.318659855769231e-05, "loss": 1.094, "step": 3630 }, { "epoch": 1.09, "learning_rate": 4.318472055288462e-05, "loss": 1.1273, "step": 3631 }, { "epoch": 1.09, "learning_rate": 4.318284254807692e-05, "loss": 1.0992, "step": 3632 }, { "epoch": 1.09, "learning_rate": 4.318096454326923e-05, "loss": 1.0869, "step": 3633 }, { "epoch": 1.09, "learning_rate": 4.317908653846154e-05, "loss": 1.0986, "step": 3634 }, { "epoch": 1.09, "learning_rate": 4.317720853365384e-05, "loss": 1.0921, "step": 3635 }, { "epoch": 1.09, "learning_rate": 4.317533052884616e-05, "loss": 0.7573, "step": 3636 }, { "epoch": 1.09, "learning_rate": 4.3173452524038466e-05, "loss": 1.1549, "step": 3637 }, { "epoch": 1.09, "learning_rate": 4.317157451923077e-05, "loss": 1.0298, "step": 3638 }, { "epoch": 1.09, "learning_rate": 4.316969651442308e-05, "loss": 0.9326, "step": 3639 }, { "epoch": 1.09, "learning_rate": 4.316781850961539e-05, "loss": 1.2116, "step": 3640 }, { "epoch": 1.09, "learning_rate": 4.316594050480769e-05, "loss": 0.8957, "step": 3641 }, { "epoch": 1.09, "learning_rate": 4.31640625e-05, "loss": 0.8306, "step": 3642 }, { "epoch": 1.09, "learning_rate": 4.316218449519231e-05, "loss": 1.1526, "step": 3643 }, { "epoch": 1.09, "learning_rate": 4.316030649038462e-05, "loss": 1.2011, "step": 3644 }, { "epoch": 1.1, "learning_rate": 4.3158428485576924e-05, "loss": 1.0552, "step": 3645 }, { "epoch": 1.1, "learning_rate": 4.315655048076923e-05, "loss": 1.3002, "step": 3646 }, { "epoch": 1.1, "learning_rate": 4.315467247596154e-05, "loss": 1.0459, "step": 3647 }, { "epoch": 1.1, "learning_rate": 4.3152794471153845e-05, "loss": 1.3147, "step": 3648 }, { "epoch": 1.1, "learning_rate": 4.315091646634616e-05, "loss": 1.3283, "step": 3649 }, { "epoch": 1.1, "learning_rate": 4.314903846153847e-05, "loss": 0.9447, "step": 3650 }, { "epoch": 1.1, "learning_rate": 4.314716045673077e-05, "loss": 1.238, "step": 3651 }, { "epoch": 1.1, "learning_rate": 4.314528245192308e-05, "loss": 1.2525, "step": 3652 }, { "epoch": 1.1, "learning_rate": 4.314340444711539e-05, "loss": 1.2381, "step": 3653 }, { "epoch": 1.1, "learning_rate": 4.3141526442307694e-05, "loss": 1.2159, "step": 3654 }, { "epoch": 1.1, "learning_rate": 4.31396484375e-05, "loss": 1.3847, "step": 3655 }, { "epoch": 1.1, "learning_rate": 4.313777043269231e-05, "loss": 1.2698, "step": 3656 }, { "epoch": 1.1, "learning_rate": 4.3135892427884615e-05, "loss": 1.0587, "step": 3657 }, { "epoch": 1.1, "learning_rate": 4.313401442307693e-05, "loss": 1.0933, "step": 3658 }, { "epoch": 1.1, "learning_rate": 4.313213641826923e-05, "loss": 0.9169, "step": 3659 }, { "epoch": 1.1, "learning_rate": 4.3130258413461536e-05, "loss": 1.1287, "step": 3660 }, { "epoch": 1.1, "learning_rate": 4.312838040865385e-05, "loss": 1.194, "step": 3661 }, { "epoch": 1.1, "learning_rate": 4.312650240384616e-05, "loss": 1.1571, "step": 3662 }, { "epoch": 1.1, "learning_rate": 4.3124624399038464e-05, "loss": 1.2439, "step": 3663 }, { "epoch": 1.1, "learning_rate": 4.3122746394230776e-05, "loss": 1.0302, "step": 3664 }, { "epoch": 1.1, "learning_rate": 4.312086838942308e-05, "loss": 1.0849, "step": 3665 }, { "epoch": 1.1, "learning_rate": 4.3118990384615385e-05, "loss": 1.1107, "step": 3666 }, { "epoch": 1.1, "learning_rate": 4.31171123798077e-05, "loss": 1.2683, "step": 3667 }, { "epoch": 1.1, "learning_rate": 4.3115234375e-05, "loss": 0.9618, "step": 3668 }, { "epoch": 1.1, "learning_rate": 4.3113356370192306e-05, "loss": 1.2146, "step": 3669 }, { "epoch": 1.1, "learning_rate": 4.311147836538462e-05, "loss": 1.0614, "step": 3670 }, { "epoch": 1.1, "learning_rate": 4.310960036057692e-05, "loss": 0.9175, "step": 3671 }, { "epoch": 1.1, "learning_rate": 4.3107722355769234e-05, "loss": 1.1661, "step": 3672 }, { "epoch": 1.1, "learning_rate": 4.310584435096154e-05, "loss": 1.013, "step": 3673 }, { "epoch": 1.1, "learning_rate": 4.3103966346153844e-05, "loss": 1.0912, "step": 3674 }, { "epoch": 1.1, "learning_rate": 4.3102088341346155e-05, "loss": 1.3085, "step": 3675 }, { "epoch": 1.1, "learning_rate": 4.310021033653847e-05, "loss": 1.3471, "step": 3676 }, { "epoch": 1.1, "learning_rate": 4.309833233173077e-05, "loss": 0.7057, "step": 3677 }, { "epoch": 1.11, "learning_rate": 4.309645432692308e-05, "loss": 1.2245, "step": 3678 }, { "epoch": 1.11, "learning_rate": 4.309457632211539e-05, "loss": 1.0119, "step": 3679 }, { "epoch": 1.11, "learning_rate": 4.309269831730769e-05, "loss": 1.1769, "step": 3680 }, { "epoch": 1.11, "learning_rate": 4.3090820312500004e-05, "loss": 0.962, "step": 3681 }, { "epoch": 1.11, "learning_rate": 4.308894230769231e-05, "loss": 1.1021, "step": 3682 }, { "epoch": 1.11, "learning_rate": 4.3087064302884614e-05, "loss": 0.9008, "step": 3683 }, { "epoch": 1.11, "learning_rate": 4.3085186298076925e-05, "loss": 0.8972, "step": 3684 }, { "epoch": 1.11, "learning_rate": 4.308330829326923e-05, "loss": 0.6336, "step": 3685 }, { "epoch": 1.11, "learning_rate": 4.3081430288461535e-05, "loss": 0.991, "step": 3686 }, { "epoch": 1.11, "learning_rate": 4.3079552283653847e-05, "loss": 1.2045, "step": 3687 }, { "epoch": 1.11, "learning_rate": 4.307767427884616e-05, "loss": 1.0322, "step": 3688 }, { "epoch": 1.11, "learning_rate": 4.307579627403846e-05, "loss": 1.2576, "step": 3689 }, { "epoch": 1.11, "learning_rate": 4.3073918269230774e-05, "loss": 0.775, "step": 3690 }, { "epoch": 1.11, "learning_rate": 4.307204026442308e-05, "loss": 1.2971, "step": 3691 }, { "epoch": 1.11, "learning_rate": 4.307016225961539e-05, "loss": 1.0303, "step": 3692 }, { "epoch": 1.11, "learning_rate": 4.3068284254807696e-05, "loss": 1.2967, "step": 3693 }, { "epoch": 1.11, "learning_rate": 4.306640625e-05, "loss": 0.9476, "step": 3694 }, { "epoch": 1.11, "learning_rate": 4.306452824519231e-05, "loss": 0.8495, "step": 3695 }, { "epoch": 1.11, "learning_rate": 4.306265024038462e-05, "loss": 1.1049, "step": 3696 }, { "epoch": 1.11, "learning_rate": 4.306077223557692e-05, "loss": 1.0414, "step": 3697 }, { "epoch": 1.11, "learning_rate": 4.305889423076923e-05, "loss": 0.9054, "step": 3698 }, { "epoch": 1.11, "learning_rate": 4.305701622596154e-05, "loss": 1.0195, "step": 3699 }, { "epoch": 1.11, "learning_rate": 4.305513822115384e-05, "loss": 1.0001, "step": 3700 }, { "epoch": 1.11, "learning_rate": 4.305326021634616e-05, "loss": 1.1447, "step": 3701 }, { "epoch": 1.11, "learning_rate": 4.3051382211538466e-05, "loss": 0.9914, "step": 3702 }, { "epoch": 1.11, "learning_rate": 4.304950420673077e-05, "loss": 0.9852, "step": 3703 }, { "epoch": 1.11, "learning_rate": 4.304762620192308e-05, "loss": 0.7807, "step": 3704 }, { "epoch": 1.11, "learning_rate": 4.304574819711539e-05, "loss": 1.2792, "step": 3705 }, { "epoch": 1.11, "learning_rate": 4.304387019230769e-05, "loss": 1.0358, "step": 3706 }, { "epoch": 1.11, "learning_rate": 4.30419921875e-05, "loss": 1.1351, "step": 3707 }, { "epoch": 1.11, "learning_rate": 4.304011418269231e-05, "loss": 0.8913, "step": 3708 }, { "epoch": 1.11, "learning_rate": 4.303823617788462e-05, "loss": 1.0038, "step": 3709 }, { "epoch": 1.11, "learning_rate": 4.3036358173076924e-05, "loss": 1.082, "step": 3710 }, { "epoch": 1.12, "learning_rate": 4.303448016826923e-05, "loss": 1.1862, "step": 3711 }, { "epoch": 1.12, "learning_rate": 4.303260216346154e-05, "loss": 0.8119, "step": 3712 }, { "epoch": 1.12, "learning_rate": 4.3030724158653845e-05, "loss": 1.1294, "step": 3713 }, { "epoch": 1.12, "learning_rate": 4.302884615384616e-05, "loss": 1.1331, "step": 3714 }, { "epoch": 1.12, "learning_rate": 4.302696814903847e-05, "loss": 1.1483, "step": 3715 }, { "epoch": 1.12, "learning_rate": 4.302509014423077e-05, "loss": 1.0933, "step": 3716 }, { "epoch": 1.12, "learning_rate": 4.302321213942308e-05, "loss": 1.1251, "step": 3717 }, { "epoch": 1.12, "learning_rate": 4.302133413461539e-05, "loss": 0.929, "step": 3718 }, { "epoch": 1.12, "learning_rate": 4.3019456129807694e-05, "loss": 0.8707, "step": 3719 }, { "epoch": 1.12, "learning_rate": 4.3017578125e-05, "loss": 1.086, "step": 3720 }, { "epoch": 1.12, "learning_rate": 4.301570012019231e-05, "loss": 1.0773, "step": 3721 }, { "epoch": 1.12, "learning_rate": 4.3013822115384615e-05, "loss": 1.1035, "step": 3722 }, { "epoch": 1.12, "learning_rate": 4.301194411057693e-05, "loss": 1.1595, "step": 3723 }, { "epoch": 1.12, "learning_rate": 4.301006610576923e-05, "loss": 0.8258, "step": 3724 }, { "epoch": 1.12, "learning_rate": 4.3008188100961536e-05, "loss": 1.1883, "step": 3725 }, { "epoch": 1.12, "learning_rate": 4.300631009615385e-05, "loss": 0.7657, "step": 3726 }, { "epoch": 1.12, "learning_rate": 4.300443209134616e-05, "loss": 0.9817, "step": 3727 }, { "epoch": 1.12, "learning_rate": 4.3002554086538464e-05, "loss": 0.959, "step": 3728 }, { "epoch": 1.12, "learning_rate": 4.3000676081730776e-05, "loss": 1.0123, "step": 3729 }, { "epoch": 1.12, "learning_rate": 4.299879807692308e-05, "loss": 1.0198, "step": 3730 }, { "epoch": 1.12, "learning_rate": 4.2996920072115385e-05, "loss": 0.9282, "step": 3731 }, { "epoch": 1.12, "learning_rate": 4.29950420673077e-05, "loss": 1.0187, "step": 3732 }, { "epoch": 1.12, "learning_rate": 4.29931640625e-05, "loss": 0.994, "step": 3733 }, { "epoch": 1.12, "learning_rate": 4.2991286057692306e-05, "loss": 1.1393, "step": 3734 }, { "epoch": 1.12, "learning_rate": 4.298940805288462e-05, "loss": 1.0577, "step": 3735 }, { "epoch": 1.12, "learning_rate": 4.298753004807692e-05, "loss": 1.0638, "step": 3736 }, { "epoch": 1.12, "learning_rate": 4.2985652043269234e-05, "loss": 1.172, "step": 3737 }, { "epoch": 1.12, "learning_rate": 4.298377403846154e-05, "loss": 1.3046, "step": 3738 }, { "epoch": 1.12, "learning_rate": 4.2981896033653844e-05, "loss": 1.375, "step": 3739 }, { "epoch": 1.12, "learning_rate": 4.2980018028846155e-05, "loss": 0.9871, "step": 3740 }, { "epoch": 1.12, "learning_rate": 4.297814002403847e-05, "loss": 0.8692, "step": 3741 }, { "epoch": 1.12, "learning_rate": 4.297626201923077e-05, "loss": 1.1604, "step": 3742 }, { "epoch": 1.12, "learning_rate": 4.297438401442308e-05, "loss": 1.1352, "step": 3743 }, { "epoch": 1.12, "learning_rate": 4.297250600961539e-05, "loss": 1.1425, "step": 3744 }, { "epoch": 1.13, "learning_rate": 4.297062800480769e-05, "loss": 1.0475, "step": 3745 }, { "epoch": 1.13, "learning_rate": 4.2968750000000004e-05, "loss": 1.1071, "step": 3746 }, { "epoch": 1.13, "learning_rate": 4.296687199519231e-05, "loss": 1.1001, "step": 3747 }, { "epoch": 1.13, "learning_rate": 4.2964993990384614e-05, "loss": 1.2902, "step": 3748 }, { "epoch": 1.13, "learning_rate": 4.2963115985576925e-05, "loss": 1.0728, "step": 3749 }, { "epoch": 1.13, "learning_rate": 4.296123798076923e-05, "loss": 1.0858, "step": 3750 }, { "epoch": 1.13, "learning_rate": 4.295935997596154e-05, "loss": 0.9606, "step": 3751 }, { "epoch": 1.13, "learning_rate": 4.2957481971153846e-05, "loss": 0.8863, "step": 3752 }, { "epoch": 1.13, "learning_rate": 4.295560396634616e-05, "loss": 1.0474, "step": 3753 }, { "epoch": 1.13, "learning_rate": 4.295372596153846e-05, "loss": 1.1039, "step": 3754 }, { "epoch": 1.13, "learning_rate": 4.2951847956730774e-05, "loss": 1.0234, "step": 3755 }, { "epoch": 1.13, "learning_rate": 4.294996995192308e-05, "loss": 1.1676, "step": 3756 }, { "epoch": 1.13, "learning_rate": 4.294809194711539e-05, "loss": 0.9964, "step": 3757 }, { "epoch": 1.13, "learning_rate": 4.2946213942307695e-05, "loss": 1.3894, "step": 3758 }, { "epoch": 1.13, "learning_rate": 4.29443359375e-05, "loss": 1.0249, "step": 3759 }, { "epoch": 1.13, "learning_rate": 4.294245793269231e-05, "loss": 0.9752, "step": 3760 }, { "epoch": 1.13, "learning_rate": 4.2940579927884617e-05, "loss": 0.8821, "step": 3761 }, { "epoch": 1.13, "learning_rate": 4.293870192307692e-05, "loss": 1.2142, "step": 3762 }, { "epoch": 1.13, "learning_rate": 4.293682391826923e-05, "loss": 1.154, "step": 3763 }, { "epoch": 1.13, "learning_rate": 4.293494591346154e-05, "loss": 1.3825, "step": 3764 }, { "epoch": 1.13, "learning_rate": 4.293306790865384e-05, "loss": 1.2522, "step": 3765 }, { "epoch": 1.13, "learning_rate": 4.293118990384616e-05, "loss": 1.1215, "step": 3766 }, { "epoch": 1.13, "learning_rate": 4.2929311899038465e-05, "loss": 1.1606, "step": 3767 }, { "epoch": 1.13, "learning_rate": 4.292743389423077e-05, "loss": 1.0554, "step": 3768 }, { "epoch": 1.13, "learning_rate": 4.292555588942308e-05, "loss": 0.9267, "step": 3769 }, { "epoch": 1.13, "learning_rate": 4.2923677884615387e-05, "loss": 1.1114, "step": 3770 }, { "epoch": 1.13, "learning_rate": 4.29217998798077e-05, "loss": 1.2161, "step": 3771 }, { "epoch": 1.13, "learning_rate": 4.2919921875e-05, "loss": 1.1811, "step": 3772 }, { "epoch": 1.13, "learning_rate": 4.291804387019231e-05, "loss": 1.1301, "step": 3773 }, { "epoch": 1.13, "learning_rate": 4.291616586538462e-05, "loss": 1.1893, "step": 3774 }, { "epoch": 1.13, "learning_rate": 4.2914287860576924e-05, "loss": 1.2081, "step": 3775 }, { "epoch": 1.13, "learning_rate": 4.291240985576923e-05, "loss": 1.0922, "step": 3776 }, { "epoch": 1.13, "learning_rate": 4.291053185096154e-05, "loss": 1.2437, "step": 3777 }, { "epoch": 1.14, "learning_rate": 4.2908653846153845e-05, "loss": 1.1589, "step": 3778 }, { "epoch": 1.14, "learning_rate": 4.290677584134616e-05, "loss": 1.1416, "step": 3779 }, { "epoch": 1.14, "learning_rate": 4.290489783653847e-05, "loss": 1.2234, "step": 3780 }, { "epoch": 1.14, "learning_rate": 4.290301983173077e-05, "loss": 1.0311, "step": 3781 }, { "epoch": 1.14, "learning_rate": 4.290114182692308e-05, "loss": 1.0754, "step": 3782 }, { "epoch": 1.14, "learning_rate": 4.289926382211539e-05, "loss": 1.0727, "step": 3783 }, { "epoch": 1.14, "learning_rate": 4.2897385817307694e-05, "loss": 1.1083, "step": 3784 }, { "epoch": 1.14, "learning_rate": 4.28955078125e-05, "loss": 0.7146, "step": 3785 }, { "epoch": 1.14, "learning_rate": 4.289362980769231e-05, "loss": 1.0939, "step": 3786 }, { "epoch": 1.14, "learning_rate": 4.2891751802884615e-05, "loss": 0.9751, "step": 3787 }, { "epoch": 1.14, "learning_rate": 4.288987379807693e-05, "loss": 1.0836, "step": 3788 }, { "epoch": 1.14, "learning_rate": 4.288799579326923e-05, "loss": 1.0415, "step": 3789 }, { "epoch": 1.14, "learning_rate": 4.2886117788461536e-05, "loss": 1.0601, "step": 3790 }, { "epoch": 1.14, "learning_rate": 4.288423978365385e-05, "loss": 1.1104, "step": 3791 }, { "epoch": 1.14, "learning_rate": 4.288236177884616e-05, "loss": 1.1984, "step": 3792 }, { "epoch": 1.14, "learning_rate": 4.2880483774038464e-05, "loss": 0.9683, "step": 3793 }, { "epoch": 1.14, "learning_rate": 4.2878605769230776e-05, "loss": 1.0019, "step": 3794 }, { "epoch": 1.14, "learning_rate": 4.287672776442308e-05, "loss": 0.8896, "step": 3795 }, { "epoch": 1.14, "learning_rate": 4.2874849759615385e-05, "loss": 1.1013, "step": 3796 }, { "epoch": 1.14, "learning_rate": 4.28729717548077e-05, "loss": 1.0078, "step": 3797 }, { "epoch": 1.14, "learning_rate": 4.287109375e-05, "loss": 1.0999, "step": 3798 }, { "epoch": 1.14, "learning_rate": 4.2869215745192306e-05, "loss": 1.0979, "step": 3799 }, { "epoch": 1.14, "learning_rate": 4.286733774038462e-05, "loss": 1.0968, "step": 3800 }, { "epoch": 1.14, "learning_rate": 4.286545973557692e-05, "loss": 1.089, "step": 3801 }, { "epoch": 1.14, "learning_rate": 4.2863581730769234e-05, "loss": 1.1191, "step": 3802 }, { "epoch": 1.14, "learning_rate": 4.286170372596154e-05, "loss": 1.1721, "step": 3803 }, { "epoch": 1.14, "learning_rate": 4.2859825721153844e-05, "loss": 1.2751, "step": 3804 }, { "epoch": 1.14, "learning_rate": 4.2857947716346155e-05, "loss": 1.0177, "step": 3805 }, { "epoch": 1.14, "learning_rate": 4.285606971153847e-05, "loss": 0.8564, "step": 3806 }, { "epoch": 1.14, "learning_rate": 4.285419170673077e-05, "loss": 1.1051, "step": 3807 }, { "epoch": 1.14, "learning_rate": 4.285231370192308e-05, "loss": 1.336, "step": 3808 }, { "epoch": 1.14, "learning_rate": 4.285043569711539e-05, "loss": 1.0902, "step": 3809 }, { "epoch": 1.14, "learning_rate": 4.284855769230769e-05, "loss": 0.7585, "step": 3810 }, { "epoch": 1.15, "learning_rate": 4.2846679687500004e-05, "loss": 0.9944, "step": 3811 }, { "epoch": 1.15, "learning_rate": 4.284480168269231e-05, "loss": 1.2413, "step": 3812 }, { "epoch": 1.15, "learning_rate": 4.2842923677884614e-05, "loss": 1.1328, "step": 3813 }, { "epoch": 1.15, "learning_rate": 4.2841045673076925e-05, "loss": 0.9437, "step": 3814 }, { "epoch": 1.15, "learning_rate": 4.283916766826923e-05, "loss": 1.0158, "step": 3815 }, { "epoch": 1.15, "learning_rate": 4.283728966346154e-05, "loss": 0.9832, "step": 3816 }, { "epoch": 1.15, "learning_rate": 4.2835411658653846e-05, "loss": 1.2795, "step": 3817 }, { "epoch": 1.15, "learning_rate": 4.283353365384616e-05, "loss": 1.1196, "step": 3818 }, { "epoch": 1.15, "learning_rate": 4.283165564903846e-05, "loss": 0.8008, "step": 3819 }, { "epoch": 1.15, "learning_rate": 4.2829777644230774e-05, "loss": 1.0711, "step": 3820 }, { "epoch": 1.15, "learning_rate": 4.282789963942308e-05, "loss": 1.2477, "step": 3821 }, { "epoch": 1.15, "learning_rate": 4.282602163461539e-05, "loss": 1.0517, "step": 3822 }, { "epoch": 1.15, "learning_rate": 4.2824143629807695e-05, "loss": 0.9456, "step": 3823 }, { "epoch": 1.15, "learning_rate": 4.2822265625e-05, "loss": 1.0119, "step": 3824 }, { "epoch": 1.15, "learning_rate": 4.282038762019231e-05, "loss": 1.0785, "step": 3825 }, { "epoch": 1.15, "learning_rate": 4.2818509615384616e-05, "loss": 1.1779, "step": 3826 }, { "epoch": 1.15, "learning_rate": 4.281663161057692e-05, "loss": 1.0963, "step": 3827 }, { "epoch": 1.15, "learning_rate": 4.281475360576923e-05, "loss": 1.2318, "step": 3828 }, { "epoch": 1.15, "learning_rate": 4.281287560096154e-05, "loss": 1.3019, "step": 3829 }, { "epoch": 1.15, "learning_rate": 4.281099759615384e-05, "loss": 1.114, "step": 3830 }, { "epoch": 1.15, "learning_rate": 4.280911959134616e-05, "loss": 1.1413, "step": 3831 }, { "epoch": 1.15, "learning_rate": 4.2807241586538465e-05, "loss": 1.0908, "step": 3832 }, { "epoch": 1.15, "learning_rate": 4.280536358173077e-05, "loss": 1.0254, "step": 3833 }, { "epoch": 1.15, "learning_rate": 4.280348557692308e-05, "loss": 1.1251, "step": 3834 }, { "epoch": 1.15, "learning_rate": 4.2801607572115386e-05, "loss": 0.9781, "step": 3835 }, { "epoch": 1.15, "learning_rate": 4.27997295673077e-05, "loss": 0.9825, "step": 3836 }, { "epoch": 1.15, "learning_rate": 4.27978515625e-05, "loss": 0.9733, "step": 3837 }, { "epoch": 1.15, "learning_rate": 4.279597355769231e-05, "loss": 1.2433, "step": 3838 }, { "epoch": 1.15, "learning_rate": 4.279409555288462e-05, "loss": 1.2453, "step": 3839 }, { "epoch": 1.15, "learning_rate": 4.2792217548076924e-05, "loss": 1.0444, "step": 3840 }, { "epoch": 1.15, "learning_rate": 4.279033954326923e-05, "loss": 1.2646, "step": 3841 }, { "epoch": 1.15, "learning_rate": 4.278846153846154e-05, "loss": 0.9191, "step": 3842 }, { "epoch": 1.15, "learning_rate": 4.2786583533653845e-05, "loss": 1.0362, "step": 3843 }, { "epoch": 1.16, "learning_rate": 4.2784705528846157e-05, "loss": 1.0504, "step": 3844 }, { "epoch": 1.16, "learning_rate": 4.278282752403847e-05, "loss": 1.0874, "step": 3845 }, { "epoch": 1.16, "learning_rate": 4.278094951923077e-05, "loss": 1.0562, "step": 3846 }, { "epoch": 1.16, "learning_rate": 4.277907151442308e-05, "loss": 1.172, "step": 3847 }, { "epoch": 1.16, "learning_rate": 4.277719350961539e-05, "loss": 0.9217, "step": 3848 }, { "epoch": 1.16, "learning_rate": 4.2775315504807694e-05, "loss": 1.2607, "step": 3849 }, { "epoch": 1.16, "learning_rate": 4.27734375e-05, "loss": 1.025, "step": 3850 }, { "epoch": 1.16, "learning_rate": 4.277155949519231e-05, "loss": 1.2778, "step": 3851 }, { "epoch": 1.16, "learning_rate": 4.2769681490384615e-05, "loss": 1.0784, "step": 3852 }, { "epoch": 1.16, "learning_rate": 4.2767803485576927e-05, "loss": 0.9651, "step": 3853 }, { "epoch": 1.16, "learning_rate": 4.276592548076923e-05, "loss": 0.9975, "step": 3854 }, { "epoch": 1.16, "learning_rate": 4.2764047475961536e-05, "loss": 1.0867, "step": 3855 }, { "epoch": 1.16, "learning_rate": 4.276216947115385e-05, "loss": 1.026, "step": 3856 }, { "epoch": 1.16, "learning_rate": 4.276029146634616e-05, "loss": 0.913, "step": 3857 }, { "epoch": 1.16, "learning_rate": 4.2758413461538464e-05, "loss": 1.4016, "step": 3858 }, { "epoch": 1.16, "learning_rate": 4.2756535456730776e-05, "loss": 1.2677, "step": 3859 }, { "epoch": 1.16, "learning_rate": 4.275465745192308e-05, "loss": 1.4396, "step": 3860 }, { "epoch": 1.16, "learning_rate": 4.2752779447115385e-05, "loss": 1.0981, "step": 3861 }, { "epoch": 1.16, "learning_rate": 4.2750901442307697e-05, "loss": 1.1139, "step": 3862 }, { "epoch": 1.16, "learning_rate": 4.27490234375e-05, "loss": 1.0256, "step": 3863 }, { "epoch": 1.16, "learning_rate": 4.2747145432692306e-05, "loss": 1.106, "step": 3864 }, { "epoch": 1.16, "learning_rate": 4.274526742788462e-05, "loss": 1.0485, "step": 3865 }, { "epoch": 1.16, "learning_rate": 4.274338942307692e-05, "loss": 0.8122, "step": 3866 }, { "epoch": 1.16, "learning_rate": 4.2741511418269234e-05, "loss": 1.1503, "step": 3867 }, { "epoch": 1.16, "learning_rate": 4.273963341346154e-05, "loss": 1.2649, "step": 3868 }, { "epoch": 1.16, "learning_rate": 4.2737755408653844e-05, "loss": 0.919, "step": 3869 }, { "epoch": 1.16, "learning_rate": 4.2735877403846155e-05, "loss": 1.2058, "step": 3870 }, { "epoch": 1.16, "learning_rate": 4.273399939903847e-05, "loss": 0.978, "step": 3871 }, { "epoch": 1.16, "learning_rate": 4.273212139423077e-05, "loss": 0.9835, "step": 3872 }, { "epoch": 1.16, "learning_rate": 4.273024338942308e-05, "loss": 0.9919, "step": 3873 }, { "epoch": 1.16, "learning_rate": 4.272836538461539e-05, "loss": 1.0648, "step": 3874 }, { "epoch": 1.16, "learning_rate": 4.272648737980769e-05, "loss": 1.112, "step": 3875 }, { "epoch": 1.16, "learning_rate": 4.2724609375000004e-05, "loss": 0.9705, "step": 3876 }, { "epoch": 1.16, "learning_rate": 4.272273137019231e-05, "loss": 1.2463, "step": 3877 }, { "epoch": 1.17, "learning_rate": 4.2720853365384614e-05, "loss": 0.8689, "step": 3878 }, { "epoch": 1.17, "learning_rate": 4.2718975360576925e-05, "loss": 1.0365, "step": 3879 }, { "epoch": 1.17, "learning_rate": 4.271709735576923e-05, "loss": 1.0357, "step": 3880 }, { "epoch": 1.17, "learning_rate": 4.271521935096154e-05, "loss": 0.975, "step": 3881 }, { "epoch": 1.17, "learning_rate": 4.2713341346153846e-05, "loss": 1.2684, "step": 3882 }, { "epoch": 1.17, "learning_rate": 4.271146334134616e-05, "loss": 1.1411, "step": 3883 }, { "epoch": 1.17, "learning_rate": 4.270958533653846e-05, "loss": 1.125, "step": 3884 }, { "epoch": 1.17, "learning_rate": 4.2707707331730774e-05, "loss": 0.9501, "step": 3885 }, { "epoch": 1.17, "learning_rate": 4.270582932692308e-05, "loss": 1.1478, "step": 3886 }, { "epoch": 1.17, "learning_rate": 4.270395132211539e-05, "loss": 1.0213, "step": 3887 }, { "epoch": 1.17, "learning_rate": 4.2702073317307695e-05, "loss": 1.0773, "step": 3888 }, { "epoch": 1.17, "learning_rate": 4.27001953125e-05, "loss": 1.2148, "step": 3889 }, { "epoch": 1.17, "learning_rate": 4.269831730769231e-05, "loss": 1.1203, "step": 3890 }, { "epoch": 1.17, "learning_rate": 4.2696439302884616e-05, "loss": 0.9996, "step": 3891 }, { "epoch": 1.17, "learning_rate": 4.269456129807692e-05, "loss": 1.3846, "step": 3892 }, { "epoch": 1.17, "learning_rate": 4.269268329326923e-05, "loss": 1.1351, "step": 3893 }, { "epoch": 1.17, "learning_rate": 4.269080528846154e-05, "loss": 0.9751, "step": 3894 }, { "epoch": 1.17, "learning_rate": 4.268892728365385e-05, "loss": 0.986, "step": 3895 }, { "epoch": 1.17, "learning_rate": 4.268704927884616e-05, "loss": 1.0091, "step": 3896 }, { "epoch": 1.17, "learning_rate": 4.2685171274038465e-05, "loss": 1.1528, "step": 3897 }, { "epoch": 1.17, "learning_rate": 4.268329326923077e-05, "loss": 1.0246, "step": 3898 }, { "epoch": 1.17, "learning_rate": 4.268141526442308e-05, "loss": 1.4261, "step": 3899 }, { "epoch": 1.17, "learning_rate": 4.2679537259615386e-05, "loss": 1.3569, "step": 3900 }, { "epoch": 1.17, "learning_rate": 4.26776592548077e-05, "loss": 1.0208, "step": 3901 }, { "epoch": 1.17, "learning_rate": 4.267578125e-05, "loss": 1.2715, "step": 3902 }, { "epoch": 1.17, "learning_rate": 4.267390324519231e-05, "loss": 1.1098, "step": 3903 }, { "epoch": 1.17, "learning_rate": 4.267202524038462e-05, "loss": 1.2822, "step": 3904 }, { "epoch": 1.17, "learning_rate": 4.2670147235576924e-05, "loss": 1.0194, "step": 3905 }, { "epoch": 1.17, "learning_rate": 4.266826923076923e-05, "loss": 0.9648, "step": 3906 }, { "epoch": 1.17, "learning_rate": 4.266639122596154e-05, "loss": 1.0788, "step": 3907 }, { "epoch": 1.17, "learning_rate": 4.2664513221153845e-05, "loss": 1.0455, "step": 3908 }, { "epoch": 1.17, "learning_rate": 4.2662635216346156e-05, "loss": 1.0349, "step": 3909 }, { "epoch": 1.17, "learning_rate": 4.266075721153847e-05, "loss": 1.3076, "step": 3910 }, { "epoch": 1.18, "learning_rate": 4.265887920673077e-05, "loss": 1.062, "step": 3911 }, { "epoch": 1.18, "learning_rate": 4.265700120192308e-05, "loss": 1.2486, "step": 3912 }, { "epoch": 1.18, "learning_rate": 4.265512319711539e-05, "loss": 1.0527, "step": 3913 }, { "epoch": 1.18, "learning_rate": 4.2653245192307694e-05, "loss": 1.1544, "step": 3914 }, { "epoch": 1.18, "learning_rate": 4.2651367187500005e-05, "loss": 0.8791, "step": 3915 }, { "epoch": 1.18, "learning_rate": 4.264948918269231e-05, "loss": 0.9122, "step": 3916 }, { "epoch": 1.18, "learning_rate": 4.2647611177884615e-05, "loss": 1.1593, "step": 3917 }, { "epoch": 1.18, "learning_rate": 4.2645733173076926e-05, "loss": 0.8898, "step": 3918 }, { "epoch": 1.18, "learning_rate": 4.264385516826923e-05, "loss": 0.9975, "step": 3919 }, { "epoch": 1.18, "learning_rate": 4.2641977163461536e-05, "loss": 0.8298, "step": 3920 }, { "epoch": 1.18, "learning_rate": 4.264009915865385e-05, "loss": 1.0677, "step": 3921 }, { "epoch": 1.18, "learning_rate": 4.263822115384616e-05, "loss": 1.0061, "step": 3922 }, { "epoch": 1.18, "learning_rate": 4.2636343149038464e-05, "loss": 1.0124, "step": 3923 }, { "epoch": 1.18, "learning_rate": 4.2634465144230775e-05, "loss": 0.9644, "step": 3924 }, { "epoch": 1.18, "learning_rate": 4.263258713942308e-05, "loss": 1.1503, "step": 3925 }, { "epoch": 1.18, "learning_rate": 4.2630709134615385e-05, "loss": 0.9995, "step": 3926 }, { "epoch": 1.18, "learning_rate": 4.2628831129807696e-05, "loss": 1.207, "step": 3927 }, { "epoch": 1.18, "learning_rate": 4.2626953125e-05, "loss": 1.2266, "step": 3928 }, { "epoch": 1.18, "learning_rate": 4.2625075120192306e-05, "loss": 1.1301, "step": 3929 }, { "epoch": 1.18, "learning_rate": 4.262319711538462e-05, "loss": 1.2054, "step": 3930 }, { "epoch": 1.18, "learning_rate": 4.262131911057692e-05, "loss": 1.0027, "step": 3931 }, { "epoch": 1.18, "learning_rate": 4.2619441105769234e-05, "loss": 1.2083, "step": 3932 }, { "epoch": 1.18, "learning_rate": 4.261756310096154e-05, "loss": 1.2446, "step": 3933 }, { "epoch": 1.18, "learning_rate": 4.2615685096153843e-05, "loss": 1.0537, "step": 3934 }, { "epoch": 1.18, "learning_rate": 4.261380709134616e-05, "loss": 0.8533, "step": 3935 }, { "epoch": 1.18, "learning_rate": 4.2611929086538467e-05, "loss": 0.9978, "step": 3936 }, { "epoch": 1.18, "learning_rate": 4.261005108173077e-05, "loss": 1.204, "step": 3937 }, { "epoch": 1.18, "learning_rate": 4.260817307692308e-05, "loss": 1.0581, "step": 3938 }, { "epoch": 1.18, "learning_rate": 4.260629507211539e-05, "loss": 1.1808, "step": 3939 }, { "epoch": 1.18, "learning_rate": 4.260441706730769e-05, "loss": 1.3328, "step": 3940 }, { "epoch": 1.18, "learning_rate": 4.2602539062500004e-05, "loss": 0.8849, "step": 3941 }, { "epoch": 1.18, "learning_rate": 4.260066105769231e-05, "loss": 0.8938, "step": 3942 }, { "epoch": 1.18, "learning_rate": 4.2598783052884614e-05, "loss": 1.2268, "step": 3943 }, { "epoch": 1.19, "learning_rate": 4.2596905048076925e-05, "loss": 1.2719, "step": 3944 }, { "epoch": 1.19, "learning_rate": 4.259502704326923e-05, "loss": 1.2335, "step": 3945 }, { "epoch": 1.19, "learning_rate": 4.259314903846154e-05, "loss": 1.1911, "step": 3946 }, { "epoch": 1.19, "learning_rate": 4.2591271033653846e-05, "loss": 0.9254, "step": 3947 }, { "epoch": 1.19, "learning_rate": 4.258939302884616e-05, "loss": 0.9502, "step": 3948 }, { "epoch": 1.19, "learning_rate": 4.258751502403846e-05, "loss": 1.1553, "step": 3949 }, { "epoch": 1.19, "learning_rate": 4.2585637019230774e-05, "loss": 0.9494, "step": 3950 }, { "epoch": 1.19, "learning_rate": 4.258375901442308e-05, "loss": 1.1867, "step": 3951 }, { "epoch": 1.19, "learning_rate": 4.258188100961539e-05, "loss": 0.9989, "step": 3952 }, { "epoch": 1.19, "learning_rate": 4.2580003004807695e-05, "loss": 0.9633, "step": 3953 }, { "epoch": 1.19, "learning_rate": 4.2578125e-05, "loss": 1.1268, "step": 3954 }, { "epoch": 1.19, "learning_rate": 4.257624699519231e-05, "loss": 0.985, "step": 3955 }, { "epoch": 1.19, "learning_rate": 4.2574368990384616e-05, "loss": 0.9445, "step": 3956 }, { "epoch": 1.19, "learning_rate": 4.257249098557692e-05, "loss": 0.885, "step": 3957 }, { "epoch": 1.19, "learning_rate": 4.257061298076923e-05, "loss": 0.9862, "step": 3958 }, { "epoch": 1.19, "learning_rate": 4.256873497596154e-05, "loss": 1.1716, "step": 3959 }, { "epoch": 1.19, "learning_rate": 4.256685697115385e-05, "loss": 0.9717, "step": 3960 }, { "epoch": 1.19, "learning_rate": 4.256497896634616e-05, "loss": 1.1563, "step": 3961 }, { "epoch": 1.19, "learning_rate": 4.2563100961538465e-05, "loss": 1.1438, "step": 3962 }, { "epoch": 1.19, "learning_rate": 4.256122295673077e-05, "loss": 0.9838, "step": 3963 }, { "epoch": 1.19, "learning_rate": 4.255934495192308e-05, "loss": 1.0556, "step": 3964 }, { "epoch": 1.19, "learning_rate": 4.2557466947115386e-05, "loss": 1.144, "step": 3965 }, { "epoch": 1.19, "learning_rate": 4.25555889423077e-05, "loss": 1.1098, "step": 3966 }, { "epoch": 1.19, "learning_rate": 4.25537109375e-05, "loss": 1.3506, "step": 3967 }, { "epoch": 1.19, "learning_rate": 4.255183293269231e-05, "loss": 0.9898, "step": 3968 }, { "epoch": 1.19, "learning_rate": 4.254995492788462e-05, "loss": 1.01, "step": 3969 }, { "epoch": 1.19, "learning_rate": 4.2548076923076924e-05, "loss": 1.0805, "step": 3970 }, { "epoch": 1.19, "learning_rate": 4.254619891826923e-05, "loss": 1.2151, "step": 3971 }, { "epoch": 1.19, "learning_rate": 4.254432091346154e-05, "loss": 1.0879, "step": 3972 }, { "epoch": 1.19, "learning_rate": 4.2542442908653845e-05, "loss": 0.8997, "step": 3973 }, { "epoch": 1.19, "learning_rate": 4.2540564903846156e-05, "loss": 1.044, "step": 3974 }, { "epoch": 1.19, "learning_rate": 4.253868689903847e-05, "loss": 1.2161, "step": 3975 }, { "epoch": 1.19, "learning_rate": 4.253680889423077e-05, "loss": 1.215, "step": 3976 }, { "epoch": 1.2, "learning_rate": 4.253493088942308e-05, "loss": 1.0893, "step": 3977 }, { "epoch": 1.2, "learning_rate": 4.253305288461539e-05, "loss": 1.0546, "step": 3978 }, { "epoch": 1.2, "learning_rate": 4.2531174879807694e-05, "loss": 1.3039, "step": 3979 }, { "epoch": 1.2, "learning_rate": 4.2529296875000005e-05, "loss": 1.0264, "step": 3980 }, { "epoch": 1.2, "learning_rate": 4.252741887019231e-05, "loss": 1.109, "step": 3981 }, { "epoch": 1.2, "learning_rate": 4.2525540865384615e-05, "loss": 1.151, "step": 3982 }, { "epoch": 1.2, "learning_rate": 4.2523662860576926e-05, "loss": 0.887, "step": 3983 }, { "epoch": 1.2, "learning_rate": 4.252178485576923e-05, "loss": 1.1287, "step": 3984 }, { "epoch": 1.2, "learning_rate": 4.2519906850961536e-05, "loss": 0.9588, "step": 3985 }, { "epoch": 1.2, "learning_rate": 4.251802884615385e-05, "loss": 1.0663, "step": 3986 }, { "epoch": 1.2, "learning_rate": 4.251615084134616e-05, "loss": 0.8992, "step": 3987 }, { "epoch": 1.2, "learning_rate": 4.2514272836538464e-05, "loss": 1.0776, "step": 3988 }, { "epoch": 1.2, "learning_rate": 4.2512394831730775e-05, "loss": 1.1172, "step": 3989 }, { "epoch": 1.2, "learning_rate": 4.251051682692308e-05, "loss": 1.2517, "step": 3990 }, { "epoch": 1.2, "learning_rate": 4.2508638822115385e-05, "loss": 1.1166, "step": 3991 }, { "epoch": 1.2, "learning_rate": 4.2506760817307696e-05, "loss": 1.0764, "step": 3992 }, { "epoch": 1.2, "learning_rate": 4.25048828125e-05, "loss": 0.9289, "step": 3993 }, { "epoch": 1.2, "learning_rate": 4.2503004807692306e-05, "loss": 1.2598, "step": 3994 }, { "epoch": 1.2, "learning_rate": 4.250112680288462e-05, "loss": 1.0012, "step": 3995 }, { "epoch": 1.2, "learning_rate": 4.249924879807692e-05, "loss": 0.9125, "step": 3996 }, { "epoch": 1.2, "learning_rate": 4.2497370793269234e-05, "loss": 1.0022, "step": 3997 }, { "epoch": 1.2, "learning_rate": 4.249549278846154e-05, "loss": 1.0836, "step": 3998 }, { "epoch": 1.2, "learning_rate": 4.249361478365384e-05, "loss": 0.859, "step": 3999 }, { "epoch": 1.2, "learning_rate": 4.249173677884616e-05, "loss": 0.7493, "step": 4000 }, { "epoch": 1.2, "learning_rate": 4.2489858774038466e-05, "loss": 1.151, "step": 4001 }, { "epoch": 1.2, "learning_rate": 4.248798076923077e-05, "loss": 1.1558, "step": 4002 }, { "epoch": 1.2, "learning_rate": 4.248610276442308e-05, "loss": 1.0062, "step": 4003 }, { "epoch": 1.2, "learning_rate": 4.248422475961539e-05, "loss": 1.305, "step": 4004 }, { "epoch": 1.2, "learning_rate": 4.248234675480769e-05, "loss": 1.0768, "step": 4005 }, { "epoch": 1.2, "learning_rate": 4.2480468750000004e-05, "loss": 0.8963, "step": 4006 }, { "epoch": 1.2, "learning_rate": 4.247859074519231e-05, "loss": 0.8951, "step": 4007 }, { "epoch": 1.2, "learning_rate": 4.2476712740384613e-05, "loss": 0.877, "step": 4008 }, { "epoch": 1.2, "learning_rate": 4.2474834735576925e-05, "loss": 0.9803, "step": 4009 }, { "epoch": 1.2, "learning_rate": 4.247295673076923e-05, "loss": 1.0935, "step": 4010 }, { "epoch": 1.21, "learning_rate": 4.247107872596154e-05, "loss": 0.9536, "step": 4011 }, { "epoch": 1.21, "learning_rate": 4.2469200721153846e-05, "loss": 1.1368, "step": 4012 }, { "epoch": 1.21, "learning_rate": 4.246732271634616e-05, "loss": 0.9584, "step": 4013 }, { "epoch": 1.21, "learning_rate": 4.246544471153846e-05, "loss": 1.0333, "step": 4014 }, { "epoch": 1.21, "learning_rate": 4.2463566706730774e-05, "loss": 1.0394, "step": 4015 }, { "epoch": 1.21, "learning_rate": 4.246168870192308e-05, "loss": 1.0248, "step": 4016 }, { "epoch": 1.21, "learning_rate": 4.245981069711539e-05, "loss": 1.185, "step": 4017 }, { "epoch": 1.21, "learning_rate": 4.2457932692307695e-05, "loss": 1.0843, "step": 4018 }, { "epoch": 1.21, "learning_rate": 4.24560546875e-05, "loss": 1.1474, "step": 4019 }, { "epoch": 1.21, "learning_rate": 4.245417668269231e-05, "loss": 1.0393, "step": 4020 }, { "epoch": 1.21, "learning_rate": 4.2452298677884616e-05, "loss": 1.2501, "step": 4021 }, { "epoch": 1.21, "learning_rate": 4.245042067307692e-05, "loss": 1.1267, "step": 4022 }, { "epoch": 1.21, "learning_rate": 4.244854266826923e-05, "loss": 1.3342, "step": 4023 }, { "epoch": 1.21, "learning_rate": 4.244666466346154e-05, "loss": 1.2644, "step": 4024 }, { "epoch": 1.21, "learning_rate": 4.244478665865385e-05, "loss": 0.9406, "step": 4025 }, { "epoch": 1.21, "learning_rate": 4.244290865384616e-05, "loss": 1.2306, "step": 4026 }, { "epoch": 1.21, "learning_rate": 4.2441030649038465e-05, "loss": 1.0712, "step": 4027 }, { "epoch": 1.21, "learning_rate": 4.243915264423077e-05, "loss": 0.949, "step": 4028 }, { "epoch": 1.21, "learning_rate": 4.243727463942308e-05, "loss": 1.0485, "step": 4029 }, { "epoch": 1.21, "learning_rate": 4.2435396634615386e-05, "loss": 0.9959, "step": 4030 }, { "epoch": 1.21, "learning_rate": 4.24335186298077e-05, "loss": 0.9424, "step": 4031 }, { "epoch": 1.21, "learning_rate": 4.2431640625e-05, "loss": 0.96, "step": 4032 }, { "epoch": 1.21, "learning_rate": 4.242976262019231e-05, "loss": 1.0309, "step": 4033 }, { "epoch": 1.21, "learning_rate": 4.242788461538462e-05, "loss": 1.0776, "step": 4034 }, { "epoch": 1.21, "learning_rate": 4.2426006610576924e-05, "loss": 1.5688, "step": 4035 }, { "epoch": 1.21, "learning_rate": 4.242412860576923e-05, "loss": 1.1717, "step": 4036 }, { "epoch": 1.21, "learning_rate": 4.242225060096154e-05, "loss": 1.5447, "step": 4037 }, { "epoch": 1.21, "learning_rate": 4.2420372596153845e-05, "loss": 1.2617, "step": 4038 }, { "epoch": 1.21, "learning_rate": 4.2418494591346156e-05, "loss": 1.1123, "step": 4039 }, { "epoch": 1.21, "learning_rate": 4.241661658653847e-05, "loss": 1.428, "step": 4040 }, { "epoch": 1.21, "learning_rate": 4.241473858173077e-05, "loss": 1.2246, "step": 4041 }, { "epoch": 1.21, "learning_rate": 4.241286057692308e-05, "loss": 1.094, "step": 4042 }, { "epoch": 1.21, "learning_rate": 4.241098257211539e-05, "loss": 0.9798, "step": 4043 }, { "epoch": 1.22, "learning_rate": 4.2409104567307694e-05, "loss": 1.0332, "step": 4044 }, { "epoch": 1.22, "learning_rate": 4.2407226562500005e-05, "loss": 1.064, "step": 4045 }, { "epoch": 1.22, "learning_rate": 4.240534855769231e-05, "loss": 0.894, "step": 4046 }, { "epoch": 1.22, "learning_rate": 4.2403470552884615e-05, "loss": 1.2165, "step": 4047 }, { "epoch": 1.22, "learning_rate": 4.2401592548076926e-05, "loss": 0.9729, "step": 4048 }, { "epoch": 1.22, "learning_rate": 4.239971454326923e-05, "loss": 0.7996, "step": 4049 }, { "epoch": 1.22, "learning_rate": 4.2397836538461536e-05, "loss": 1.0537, "step": 4050 }, { "epoch": 1.22, "learning_rate": 4.239595853365385e-05, "loss": 1.2245, "step": 4051 }, { "epoch": 1.22, "learning_rate": 4.239408052884616e-05, "loss": 0.9783, "step": 4052 }, { "epoch": 1.22, "learning_rate": 4.2392202524038464e-05, "loss": 1.0395, "step": 4053 }, { "epoch": 1.22, "learning_rate": 4.2390324519230775e-05, "loss": 0.9337, "step": 4054 }, { "epoch": 1.22, "learning_rate": 4.238844651442308e-05, "loss": 1.1294, "step": 4055 }, { "epoch": 1.22, "learning_rate": 4.2386568509615385e-05, "loss": 1.1642, "step": 4056 }, { "epoch": 1.22, "learning_rate": 4.2384690504807696e-05, "loss": 1.1169, "step": 4057 }, { "epoch": 1.22, "learning_rate": 4.23828125e-05, "loss": 1.2289, "step": 4058 }, { "epoch": 1.22, "learning_rate": 4.238093449519231e-05, "loss": 1.0423, "step": 4059 }, { "epoch": 1.22, "learning_rate": 4.237905649038462e-05, "loss": 1.2176, "step": 4060 }, { "epoch": 1.22, "learning_rate": 4.237717848557692e-05, "loss": 1.1467, "step": 4061 }, { "epoch": 1.22, "learning_rate": 4.2375300480769234e-05, "loss": 1.017, "step": 4062 }, { "epoch": 1.22, "learning_rate": 4.237342247596154e-05, "loss": 1.1674, "step": 4063 }, { "epoch": 1.22, "learning_rate": 4.237154447115384e-05, "loss": 1.1481, "step": 4064 }, { "epoch": 1.22, "learning_rate": 4.236966646634616e-05, "loss": 1.2451, "step": 4065 }, { "epoch": 1.22, "learning_rate": 4.2367788461538466e-05, "loss": 1.086, "step": 4066 }, { "epoch": 1.22, "learning_rate": 4.236591045673077e-05, "loss": 1.01, "step": 4067 }, { "epoch": 1.22, "learning_rate": 4.236403245192308e-05, "loss": 1.1589, "step": 4068 }, { "epoch": 1.22, "learning_rate": 4.236215444711539e-05, "loss": 0.9593, "step": 4069 }, { "epoch": 1.22, "learning_rate": 4.236027644230769e-05, "loss": 0.8793, "step": 4070 }, { "epoch": 1.22, "learning_rate": 4.2358398437500004e-05, "loss": 0.9525, "step": 4071 }, { "epoch": 1.22, "learning_rate": 4.235652043269231e-05, "loss": 1.0075, "step": 4072 }, { "epoch": 1.22, "learning_rate": 4.235464242788461e-05, "loss": 0.8192, "step": 4073 }, { "epoch": 1.22, "learning_rate": 4.2352764423076925e-05, "loss": 0.8944, "step": 4074 }, { "epoch": 1.22, "learning_rate": 4.235088641826923e-05, "loss": 1.3116, "step": 4075 }, { "epoch": 1.22, "learning_rate": 4.234900841346154e-05, "loss": 1.0513, "step": 4076 }, { "epoch": 1.23, "learning_rate": 4.2347130408653846e-05, "loss": 1.0483, "step": 4077 }, { "epoch": 1.23, "learning_rate": 4.234525240384616e-05, "loss": 0.8785, "step": 4078 }, { "epoch": 1.23, "learning_rate": 4.234337439903847e-05, "loss": 1.2822, "step": 4079 }, { "epoch": 1.23, "learning_rate": 4.2341496394230774e-05, "loss": 1.3243, "step": 4080 }, { "epoch": 1.23, "learning_rate": 4.233961838942308e-05, "loss": 0.8881, "step": 4081 }, { "epoch": 1.23, "learning_rate": 4.233774038461539e-05, "loss": 1.0704, "step": 4082 }, { "epoch": 1.23, "learning_rate": 4.2335862379807695e-05, "loss": 1.187, "step": 4083 }, { "epoch": 1.23, "learning_rate": 4.2333984375e-05, "loss": 1.2537, "step": 4084 }, { "epoch": 1.23, "learning_rate": 4.233210637019231e-05, "loss": 1.0644, "step": 4085 }, { "epoch": 1.23, "learning_rate": 4.2330228365384616e-05, "loss": 1.118, "step": 4086 }, { "epoch": 1.23, "learning_rate": 4.232835036057692e-05, "loss": 0.9385, "step": 4087 }, { "epoch": 1.23, "learning_rate": 4.232647235576923e-05, "loss": 1.1094, "step": 4088 }, { "epoch": 1.23, "learning_rate": 4.232459435096154e-05, "loss": 1.1958, "step": 4089 }, { "epoch": 1.23, "learning_rate": 4.232271634615385e-05, "loss": 1.122, "step": 4090 }, { "epoch": 1.23, "learning_rate": 4.232083834134616e-05, "loss": 1.3781, "step": 4091 }, { "epoch": 1.23, "learning_rate": 4.2318960336538465e-05, "loss": 1.0636, "step": 4092 }, { "epoch": 1.23, "learning_rate": 4.231708233173077e-05, "loss": 0.9964, "step": 4093 }, { "epoch": 1.23, "learning_rate": 4.231520432692308e-05, "loss": 1.3065, "step": 4094 }, { "epoch": 1.23, "learning_rate": 4.2313326322115386e-05, "loss": 0.9677, "step": 4095 }, { "epoch": 1.23, "learning_rate": 4.23114483173077e-05, "loss": 1.3617, "step": 4096 }, { "epoch": 1.23, "learning_rate": 4.23095703125e-05, "loss": 1.1725, "step": 4097 }, { "epoch": 1.23, "learning_rate": 4.230769230769231e-05, "loss": 1.2161, "step": 4098 }, { "epoch": 1.23, "learning_rate": 4.230581430288462e-05, "loss": 1.2522, "step": 4099 }, { "epoch": 1.23, "learning_rate": 4.2303936298076923e-05, "loss": 0.951, "step": 4100 }, { "epoch": 1.23, "learning_rate": 4.230205829326923e-05, "loss": 1.153, "step": 4101 }, { "epoch": 1.23, "learning_rate": 4.230018028846154e-05, "loss": 0.9808, "step": 4102 }, { "epoch": 1.23, "learning_rate": 4.2298302283653845e-05, "loss": 1.1114, "step": 4103 }, { "epoch": 1.23, "learning_rate": 4.2296424278846156e-05, "loss": 0.9036, "step": 4104 }, { "epoch": 1.23, "learning_rate": 4.229454627403847e-05, "loss": 0.9846, "step": 4105 }, { "epoch": 1.23, "learning_rate": 4.229266826923077e-05, "loss": 0.8333, "step": 4106 }, { "epoch": 1.23, "learning_rate": 4.229079026442308e-05, "loss": 1.1028, "step": 4107 }, { "epoch": 1.23, "learning_rate": 4.228891225961539e-05, "loss": 0.8904, "step": 4108 }, { "epoch": 1.23, "learning_rate": 4.2287034254807693e-05, "loss": 1.0559, "step": 4109 }, { "epoch": 1.23, "learning_rate": 4.2285156250000005e-05, "loss": 0.9186, "step": 4110 }, { "epoch": 1.24, "learning_rate": 4.228327824519231e-05, "loss": 1.0801, "step": 4111 }, { "epoch": 1.24, "learning_rate": 4.2281400240384615e-05, "loss": 1.096, "step": 4112 }, { "epoch": 1.24, "learning_rate": 4.2279522235576926e-05, "loss": 1.1529, "step": 4113 }, { "epoch": 1.24, "learning_rate": 4.227764423076923e-05, "loss": 1.0825, "step": 4114 }, { "epoch": 1.24, "learning_rate": 4.2275766225961536e-05, "loss": 0.8896, "step": 4115 }, { "epoch": 1.24, "learning_rate": 4.227388822115385e-05, "loss": 1.1571, "step": 4116 }, { "epoch": 1.24, "learning_rate": 4.227201021634616e-05, "loss": 1.0952, "step": 4117 }, { "epoch": 1.24, "learning_rate": 4.2270132211538464e-05, "loss": 0.9699, "step": 4118 }, { "epoch": 1.24, "learning_rate": 4.2268254206730775e-05, "loss": 0.8472, "step": 4119 }, { "epoch": 1.24, "learning_rate": 4.226637620192308e-05, "loss": 0.9523, "step": 4120 }, { "epoch": 1.24, "learning_rate": 4.2264498197115385e-05, "loss": 1.1065, "step": 4121 }, { "epoch": 1.24, "learning_rate": 4.2262620192307696e-05, "loss": 1.2697, "step": 4122 }, { "epoch": 1.24, "learning_rate": 4.22607421875e-05, "loss": 1.1481, "step": 4123 }, { "epoch": 1.24, "learning_rate": 4.225886418269231e-05, "loss": 1.0491, "step": 4124 }, { "epoch": 1.24, "learning_rate": 4.225698617788462e-05, "loss": 0.8795, "step": 4125 }, { "epoch": 1.24, "learning_rate": 4.225510817307692e-05, "loss": 1.0267, "step": 4126 }, { "epoch": 1.24, "learning_rate": 4.2253230168269234e-05, "loss": 1.1091, "step": 4127 }, { "epoch": 1.24, "learning_rate": 4.225135216346154e-05, "loss": 1.342, "step": 4128 }, { "epoch": 1.24, "learning_rate": 4.224947415865384e-05, "loss": 1.0246, "step": 4129 }, { "epoch": 1.24, "learning_rate": 4.224759615384616e-05, "loss": 0.8934, "step": 4130 }, { "epoch": 1.24, "learning_rate": 4.2245718149038466e-05, "loss": 1.0029, "step": 4131 }, { "epoch": 1.24, "learning_rate": 4.224384014423077e-05, "loss": 1.0987, "step": 4132 }, { "epoch": 1.24, "learning_rate": 4.224196213942308e-05, "loss": 1.1973, "step": 4133 }, { "epoch": 1.24, "learning_rate": 4.224008413461539e-05, "loss": 0.9265, "step": 4134 }, { "epoch": 1.24, "learning_rate": 4.223820612980769e-05, "loss": 1.0349, "step": 4135 }, { "epoch": 1.24, "learning_rate": 4.2236328125000004e-05, "loss": 0.9766, "step": 4136 }, { "epoch": 1.24, "learning_rate": 4.223445012019231e-05, "loss": 0.9998, "step": 4137 }, { "epoch": 1.24, "learning_rate": 4.223257211538461e-05, "loss": 1.1765, "step": 4138 }, { "epoch": 1.24, "learning_rate": 4.2230694110576925e-05, "loss": 0.9106, "step": 4139 }, { "epoch": 1.24, "learning_rate": 4.222881610576923e-05, "loss": 0.9426, "step": 4140 }, { "epoch": 1.24, "learning_rate": 4.222693810096154e-05, "loss": 1.2173, "step": 4141 }, { "epoch": 1.24, "learning_rate": 4.2225060096153846e-05, "loss": 1.1604, "step": 4142 }, { "epoch": 1.24, "learning_rate": 4.222318209134616e-05, "loss": 1.3578, "step": 4143 }, { "epoch": 1.25, "learning_rate": 4.222130408653847e-05, "loss": 0.9051, "step": 4144 }, { "epoch": 1.25, "learning_rate": 4.2219426081730774e-05, "loss": 1.0047, "step": 4145 }, { "epoch": 1.25, "learning_rate": 4.221754807692308e-05, "loss": 1.0241, "step": 4146 }, { "epoch": 1.25, "learning_rate": 4.221567007211539e-05, "loss": 1.1193, "step": 4147 }, { "epoch": 1.25, "learning_rate": 4.2213792067307695e-05, "loss": 1.1032, "step": 4148 }, { "epoch": 1.25, "learning_rate": 4.22119140625e-05, "loss": 1.1822, "step": 4149 }, { "epoch": 1.25, "learning_rate": 4.221003605769231e-05, "loss": 1.0978, "step": 4150 }, { "epoch": 1.25, "learning_rate": 4.2208158052884616e-05, "loss": 0.9867, "step": 4151 }, { "epoch": 1.25, "learning_rate": 4.220628004807692e-05, "loss": 1.1387, "step": 4152 }, { "epoch": 1.25, "learning_rate": 4.220440204326923e-05, "loss": 1.2465, "step": 4153 }, { "epoch": 1.25, "learning_rate": 4.220252403846154e-05, "loss": 1.0038, "step": 4154 }, { "epoch": 1.25, "learning_rate": 4.220064603365385e-05, "loss": 0.9688, "step": 4155 }, { "epoch": 1.25, "learning_rate": 4.219876802884616e-05, "loss": 1.2527, "step": 4156 }, { "epoch": 1.25, "learning_rate": 4.2196890024038465e-05, "loss": 1.1227, "step": 4157 }, { "epoch": 1.25, "learning_rate": 4.219501201923077e-05, "loss": 1.093, "step": 4158 }, { "epoch": 1.25, "learning_rate": 4.219313401442308e-05, "loss": 0.9176, "step": 4159 }, { "epoch": 1.25, "learning_rate": 4.2191256009615386e-05, "loss": 0.9975, "step": 4160 }, { "epoch": 1.25, "learning_rate": 4.21893780048077e-05, "loss": 1.1665, "step": 4161 }, { "epoch": 1.25, "learning_rate": 4.21875e-05, "loss": 1.4294, "step": 4162 }, { "epoch": 1.25, "learning_rate": 4.21875e-05, "loss": 0.8843, "step": 4163 }, { "epoch": 1.25, "learning_rate": 4.218562199519231e-05, "loss": 0.988, "step": 4164 }, { "epoch": 1.25, "learning_rate": 4.218374399038462e-05, "loss": 1.0634, "step": 4165 }, { "epoch": 1.25, "learning_rate": 4.218186598557692e-05, "loss": 1.1277, "step": 4166 }, { "epoch": 1.25, "learning_rate": 4.217998798076923e-05, "loss": 1.0437, "step": 4167 }, { "epoch": 1.25, "learning_rate": 4.217810997596154e-05, "loss": 1.1291, "step": 4168 }, { "epoch": 1.25, "learning_rate": 4.2176231971153844e-05, "loss": 1.2693, "step": 4169 }, { "epoch": 1.25, "learning_rate": 4.2174353966346156e-05, "loss": 1.1431, "step": 4170 }, { "epoch": 1.25, "learning_rate": 4.217247596153847e-05, "loss": 1.0696, "step": 4171 }, { "epoch": 1.25, "learning_rate": 4.217059795673077e-05, "loss": 0.9734, "step": 4172 }, { "epoch": 1.25, "learning_rate": 4.216871995192308e-05, "loss": 1.1472, "step": 4173 }, { "epoch": 1.25, "learning_rate": 4.216684194711539e-05, "loss": 1.2031, "step": 4174 }, { "epoch": 1.25, "learning_rate": 4.216496394230769e-05, "loss": 1.092, "step": 4175 }, { "epoch": 1.25, "learning_rate": 4.2163085937500005e-05, "loss": 1.0506, "step": 4176 }, { "epoch": 1.26, "learning_rate": 4.216120793269231e-05, "loss": 0.9841, "step": 4177 }, { "epoch": 1.26, "learning_rate": 4.2159329927884614e-05, "loss": 0.9321, "step": 4178 }, { "epoch": 1.26, "learning_rate": 4.2157451923076926e-05, "loss": 0.9755, "step": 4179 }, { "epoch": 1.26, "learning_rate": 4.215557391826923e-05, "loss": 1.2388, "step": 4180 }, { "epoch": 1.26, "learning_rate": 4.2153695913461536e-05, "loss": 1.1799, "step": 4181 }, { "epoch": 1.26, "learning_rate": 4.215181790865385e-05, "loss": 1.3544, "step": 4182 }, { "epoch": 1.26, "learning_rate": 4.214993990384616e-05, "loss": 1.0858, "step": 4183 }, { "epoch": 1.26, "learning_rate": 4.2148061899038463e-05, "loss": 1.0532, "step": 4184 }, { "epoch": 1.26, "learning_rate": 4.2146183894230775e-05, "loss": 1.0384, "step": 4185 }, { "epoch": 1.26, "learning_rate": 4.214430588942308e-05, "loss": 1.2543, "step": 4186 }, { "epoch": 1.26, "learning_rate": 4.2142427884615385e-05, "loss": 0.9507, "step": 4187 }, { "epoch": 1.26, "learning_rate": 4.2140549879807696e-05, "loss": 1.0803, "step": 4188 }, { "epoch": 1.26, "learning_rate": 4.2138671875e-05, "loss": 1.094, "step": 4189 }, { "epoch": 1.26, "learning_rate": 4.213679387019231e-05, "loss": 1.0184, "step": 4190 }, { "epoch": 1.26, "learning_rate": 4.213491586538462e-05, "loss": 1.2675, "step": 4191 }, { "epoch": 1.26, "learning_rate": 4.213303786057692e-05, "loss": 1.2927, "step": 4192 }, { "epoch": 1.26, "learning_rate": 4.2131159855769233e-05, "loss": 1.1999, "step": 4193 }, { "epoch": 1.26, "learning_rate": 4.212928185096154e-05, "loss": 1.104, "step": 4194 }, { "epoch": 1.26, "learning_rate": 4.212740384615384e-05, "loss": 1.1953, "step": 4195 }, { "epoch": 1.26, "learning_rate": 4.212552584134616e-05, "loss": 1.1148, "step": 4196 }, { "epoch": 1.26, "learning_rate": 4.2123647836538466e-05, "loss": 1.2254, "step": 4197 }, { "epoch": 1.26, "learning_rate": 4.212176983173077e-05, "loss": 1.3512, "step": 4198 }, { "epoch": 1.26, "learning_rate": 4.211989182692308e-05, "loss": 0.9671, "step": 4199 }, { "epoch": 1.26, "learning_rate": 4.211801382211539e-05, "loss": 0.963, "step": 4200 }, { "epoch": 1.26, "learning_rate": 4.211613581730769e-05, "loss": 1.0657, "step": 4201 }, { "epoch": 1.26, "learning_rate": 4.2114257812500004e-05, "loss": 0.9679, "step": 4202 }, { "epoch": 1.26, "learning_rate": 4.211237980769231e-05, "loss": 0.9656, "step": 4203 }, { "epoch": 1.26, "learning_rate": 4.211050180288462e-05, "loss": 0.9939, "step": 4204 }, { "epoch": 1.26, "learning_rate": 4.2108623798076925e-05, "loss": 0.8021, "step": 4205 }, { "epoch": 1.26, "learning_rate": 4.210674579326923e-05, "loss": 0.9155, "step": 4206 }, { "epoch": 1.26, "learning_rate": 4.210486778846154e-05, "loss": 1.1539, "step": 4207 }, { "epoch": 1.26, "learning_rate": 4.2102989783653846e-05, "loss": 1.2486, "step": 4208 }, { "epoch": 1.26, "learning_rate": 4.210111177884616e-05, "loss": 1.1793, "step": 4209 }, { "epoch": 1.27, "learning_rate": 4.209923377403847e-05, "loss": 1.0092, "step": 4210 }, { "epoch": 1.27, "learning_rate": 4.2097355769230774e-05, "loss": 1.1529, "step": 4211 }, { "epoch": 1.27, "learning_rate": 4.209547776442308e-05, "loss": 1.1316, "step": 4212 }, { "epoch": 1.27, "learning_rate": 4.209359975961539e-05, "loss": 1.1312, "step": 4213 }, { "epoch": 1.27, "learning_rate": 4.2091721754807695e-05, "loss": 1.1558, "step": 4214 }, { "epoch": 1.27, "learning_rate": 4.208984375e-05, "loss": 1.098, "step": 4215 }, { "epoch": 1.27, "learning_rate": 4.208796574519231e-05, "loss": 1.1289, "step": 4216 }, { "epoch": 1.27, "learning_rate": 4.2086087740384616e-05, "loss": 1.1433, "step": 4217 }, { "epoch": 1.27, "learning_rate": 4.208420973557692e-05, "loss": 1.0157, "step": 4218 }, { "epoch": 1.27, "learning_rate": 4.208233173076923e-05, "loss": 1.2141, "step": 4219 }, { "epoch": 1.27, "learning_rate": 4.208045372596154e-05, "loss": 1.2023, "step": 4220 }, { "epoch": 1.27, "learning_rate": 4.207857572115385e-05, "loss": 1.0749, "step": 4221 }, { "epoch": 1.27, "learning_rate": 4.207669771634616e-05, "loss": 1.1252, "step": 4222 }, { "epoch": 1.27, "learning_rate": 4.2074819711538465e-05, "loss": 1.1344, "step": 4223 }, { "epoch": 1.27, "learning_rate": 4.2072941706730776e-05, "loss": 0.9578, "step": 4224 }, { "epoch": 1.27, "learning_rate": 4.207106370192308e-05, "loss": 1.1106, "step": 4225 }, { "epoch": 1.27, "learning_rate": 4.2069185697115386e-05, "loss": 1.2556, "step": 4226 }, { "epoch": 1.27, "learning_rate": 4.20673076923077e-05, "loss": 1.0667, "step": 4227 }, { "epoch": 1.27, "learning_rate": 4.20654296875e-05, "loss": 1.0387, "step": 4228 }, { "epoch": 1.27, "learning_rate": 4.206355168269231e-05, "loss": 1.1939, "step": 4229 }, { "epoch": 1.27, "learning_rate": 4.206167367788462e-05, "loss": 1.1854, "step": 4230 }, { "epoch": 1.27, "learning_rate": 4.205979567307692e-05, "loss": 1.1684, "step": 4231 }, { "epoch": 1.27, "learning_rate": 4.205791766826923e-05, "loss": 1.1473, "step": 4232 }, { "epoch": 1.27, "learning_rate": 4.205603966346154e-05, "loss": 1.191, "step": 4233 }, { "epoch": 1.27, "learning_rate": 4.2054161658653844e-05, "loss": 1.3124, "step": 4234 }, { "epoch": 1.27, "learning_rate": 4.2052283653846156e-05, "loss": 1.0457, "step": 4235 }, { "epoch": 1.27, "learning_rate": 4.205040564903847e-05, "loss": 1.1281, "step": 4236 }, { "epoch": 1.27, "learning_rate": 4.204852764423077e-05, "loss": 1.2457, "step": 4237 }, { "epoch": 1.27, "learning_rate": 4.204664963942308e-05, "loss": 1.2747, "step": 4238 }, { "epoch": 1.27, "learning_rate": 4.204477163461539e-05, "loss": 1.1098, "step": 4239 }, { "epoch": 1.27, "learning_rate": 4.204289362980769e-05, "loss": 1.0618, "step": 4240 }, { "epoch": 1.27, "learning_rate": 4.2041015625000005e-05, "loss": 0.9433, "step": 4241 }, { "epoch": 1.27, "learning_rate": 4.203913762019231e-05, "loss": 0.9972, "step": 4242 }, { "epoch": 1.27, "learning_rate": 4.2037259615384614e-05, "loss": 0.9755, "step": 4243 }, { "epoch": 1.28, "learning_rate": 4.2035381610576926e-05, "loss": 1.1738, "step": 4244 }, { "epoch": 1.28, "learning_rate": 4.203350360576923e-05, "loss": 1.081, "step": 4245 }, { "epoch": 1.28, "learning_rate": 4.2031625600961535e-05, "loss": 1.1201, "step": 4246 }, { "epoch": 1.28, "learning_rate": 4.202974759615385e-05, "loss": 0.9728, "step": 4247 }, { "epoch": 1.28, "learning_rate": 4.202786959134616e-05, "loss": 1.023, "step": 4248 }, { "epoch": 1.28, "learning_rate": 4.202599158653846e-05, "loss": 1.3067, "step": 4249 }, { "epoch": 1.28, "learning_rate": 4.2024113581730775e-05, "loss": 1.0556, "step": 4250 }, { "epoch": 1.28, "learning_rate": 4.202223557692308e-05, "loss": 1.0957, "step": 4251 }, { "epoch": 1.28, "learning_rate": 4.2020357572115384e-05, "loss": 0.9713, "step": 4252 }, { "epoch": 1.28, "learning_rate": 4.2018479567307696e-05, "loss": 1.1371, "step": 4253 }, { "epoch": 1.28, "learning_rate": 4.20166015625e-05, "loss": 0.9394, "step": 4254 }, { "epoch": 1.28, "learning_rate": 4.201472355769231e-05, "loss": 0.9993, "step": 4255 }, { "epoch": 1.28, "learning_rate": 4.201284555288462e-05, "loss": 0.7623, "step": 4256 }, { "epoch": 1.28, "learning_rate": 4.201096754807692e-05, "loss": 1.14, "step": 4257 }, { "epoch": 1.28, "learning_rate": 4.200908954326923e-05, "loss": 0.9455, "step": 4258 }, { "epoch": 1.28, "learning_rate": 4.200721153846154e-05, "loss": 0.9876, "step": 4259 }, { "epoch": 1.28, "learning_rate": 4.200533353365384e-05, "loss": 1.247, "step": 4260 }, { "epoch": 1.28, "learning_rate": 4.200345552884616e-05, "loss": 1.2739, "step": 4261 }, { "epoch": 1.28, "learning_rate": 4.2001577524038466e-05, "loss": 1.1814, "step": 4262 }, { "epoch": 1.28, "learning_rate": 4.199969951923077e-05, "loss": 0.8429, "step": 4263 }, { "epoch": 1.28, "learning_rate": 4.199782151442308e-05, "loss": 0.9397, "step": 4264 }, { "epoch": 1.28, "learning_rate": 4.199594350961539e-05, "loss": 0.8419, "step": 4265 }, { "epoch": 1.28, "learning_rate": 4.199406550480769e-05, "loss": 1.3451, "step": 4266 }, { "epoch": 1.28, "learning_rate": 4.1992187500000003e-05, "loss": 1.0655, "step": 4267 }, { "epoch": 1.28, "learning_rate": 4.199030949519231e-05, "loss": 0.9611, "step": 4268 }, { "epoch": 1.28, "learning_rate": 4.198843149038462e-05, "loss": 1.2141, "step": 4269 }, { "epoch": 1.28, "learning_rate": 4.1986553485576925e-05, "loss": 1.2224, "step": 4270 }, { "epoch": 1.28, "learning_rate": 4.198467548076923e-05, "loss": 0.9726, "step": 4271 }, { "epoch": 1.28, "learning_rate": 4.198279747596154e-05, "loss": 1.0024, "step": 4272 }, { "epoch": 1.28, "learning_rate": 4.1980919471153846e-05, "loss": 1.0107, "step": 4273 }, { "epoch": 1.28, "learning_rate": 4.197904146634616e-05, "loss": 1.1794, "step": 4274 }, { "epoch": 1.28, "learning_rate": 4.197716346153847e-05, "loss": 1.0717, "step": 4275 }, { "epoch": 1.28, "learning_rate": 4.1975285456730773e-05, "loss": 1.107, "step": 4276 }, { "epoch": 1.29, "learning_rate": 4.197340745192308e-05, "loss": 1.2261, "step": 4277 }, { "epoch": 1.29, "learning_rate": 4.197152944711539e-05, "loss": 0.8788, "step": 4278 }, { "epoch": 1.29, "learning_rate": 4.1969651442307695e-05, "loss": 1.229, "step": 4279 }, { "epoch": 1.29, "learning_rate": 4.19677734375e-05, "loss": 1.2741, "step": 4280 }, { "epoch": 1.29, "learning_rate": 4.196589543269231e-05, "loss": 1.2012, "step": 4281 }, { "epoch": 1.29, "learning_rate": 4.1964017427884616e-05, "loss": 1.1347, "step": 4282 }, { "epoch": 1.29, "learning_rate": 4.196213942307692e-05, "loss": 1.2291, "step": 4283 }, { "epoch": 1.29, "learning_rate": 4.196026141826923e-05, "loss": 1.2049, "step": 4284 }, { "epoch": 1.29, "learning_rate": 4.195838341346154e-05, "loss": 1.1674, "step": 4285 }, { "epoch": 1.29, "learning_rate": 4.195650540865385e-05, "loss": 1.0886, "step": 4286 }, { "epoch": 1.29, "learning_rate": 4.195462740384616e-05, "loss": 1.2411, "step": 4287 }, { "epoch": 1.29, "learning_rate": 4.1952749399038465e-05, "loss": 1.2515, "step": 4288 }, { "epoch": 1.29, "learning_rate": 4.1950871394230776e-05, "loss": 1.2256, "step": 4289 }, { "epoch": 1.29, "learning_rate": 4.194899338942308e-05, "loss": 1.1566, "step": 4290 }, { "epoch": 1.29, "learning_rate": 4.1947115384615386e-05, "loss": 0.8944, "step": 4291 }, { "epoch": 1.29, "learning_rate": 4.19452373798077e-05, "loss": 1.0315, "step": 4292 }, { "epoch": 1.29, "learning_rate": 4.1943359375e-05, "loss": 1.209, "step": 4293 }, { "epoch": 1.29, "learning_rate": 4.194148137019231e-05, "loss": 1.0144, "step": 4294 }, { "epoch": 1.29, "learning_rate": 4.193960336538462e-05, "loss": 1.375, "step": 4295 }, { "epoch": 1.29, "learning_rate": 4.193772536057692e-05, "loss": 1.0536, "step": 4296 }, { "epoch": 1.29, "learning_rate": 4.193584735576923e-05, "loss": 1.1031, "step": 4297 }, { "epoch": 1.29, "learning_rate": 4.193396935096154e-05, "loss": 0.8854, "step": 4298 }, { "epoch": 1.29, "learning_rate": 4.1932091346153844e-05, "loss": 1.1951, "step": 4299 }, { "epoch": 1.29, "learning_rate": 4.1930213341346156e-05, "loss": 0.973, "step": 4300 }, { "epoch": 1.29, "learning_rate": 4.192833533653847e-05, "loss": 1.2267, "step": 4301 }, { "epoch": 1.29, "learning_rate": 4.192645733173077e-05, "loss": 1.0168, "step": 4302 }, { "epoch": 1.29, "learning_rate": 4.192457932692308e-05, "loss": 1.1278, "step": 4303 }, { "epoch": 1.29, "learning_rate": 4.192270132211539e-05, "loss": 0.8998, "step": 4304 }, { "epoch": 1.29, "learning_rate": 4.192082331730769e-05, "loss": 1.1061, "step": 4305 }, { "epoch": 1.29, "learning_rate": 4.1918945312500005e-05, "loss": 1.2013, "step": 4306 }, { "epoch": 1.29, "learning_rate": 4.191706730769231e-05, "loss": 1.127, "step": 4307 }, { "epoch": 1.29, "learning_rate": 4.1915189302884614e-05, "loss": 1.1231, "step": 4308 }, { "epoch": 1.29, "learning_rate": 4.1913311298076926e-05, "loss": 1.2069, "step": 4309 }, { "epoch": 1.3, "learning_rate": 4.191143329326923e-05, "loss": 1.1538, "step": 4310 }, { "epoch": 1.3, "learning_rate": 4.1909555288461535e-05, "loss": 1.2189, "step": 4311 }, { "epoch": 1.3, "learning_rate": 4.190767728365385e-05, "loss": 0.9314, "step": 4312 }, { "epoch": 1.3, "learning_rate": 4.190579927884616e-05, "loss": 0.8757, "step": 4313 }, { "epoch": 1.3, "learning_rate": 4.190392127403846e-05, "loss": 0.996, "step": 4314 }, { "epoch": 1.3, "learning_rate": 4.1902043269230775e-05, "loss": 1.0706, "step": 4315 }, { "epoch": 1.3, "learning_rate": 4.190016526442308e-05, "loss": 1.0067, "step": 4316 }, { "epoch": 1.3, "learning_rate": 4.1898287259615384e-05, "loss": 0.9464, "step": 4317 }, { "epoch": 1.3, "learning_rate": 4.1896409254807696e-05, "loss": 0.9358, "step": 4318 }, { "epoch": 1.3, "learning_rate": 4.189453125e-05, "loss": 1.1486, "step": 4319 }, { "epoch": 1.3, "learning_rate": 4.189265324519231e-05, "loss": 1.2405, "step": 4320 }, { "epoch": 1.3, "learning_rate": 4.189077524038462e-05, "loss": 0.9572, "step": 4321 }, { "epoch": 1.3, "learning_rate": 4.188889723557692e-05, "loss": 0.994, "step": 4322 }, { "epoch": 1.3, "learning_rate": 4.188701923076923e-05, "loss": 0.9312, "step": 4323 }, { "epoch": 1.3, "learning_rate": 4.188514122596154e-05, "loss": 1.0344, "step": 4324 }, { "epoch": 1.3, "learning_rate": 4.188326322115384e-05, "loss": 1.0816, "step": 4325 }, { "epoch": 1.3, "learning_rate": 4.188138521634616e-05, "loss": 1.0664, "step": 4326 }, { "epoch": 1.3, "learning_rate": 4.1879507211538466e-05, "loss": 1.2551, "step": 4327 }, { "epoch": 1.3, "learning_rate": 4.187762920673077e-05, "loss": 1.1098, "step": 4328 }, { "epoch": 1.3, "learning_rate": 4.187575120192308e-05, "loss": 1.0029, "step": 4329 }, { "epoch": 1.3, "learning_rate": 4.187387319711539e-05, "loss": 1.0438, "step": 4330 }, { "epoch": 1.3, "learning_rate": 4.187199519230769e-05, "loss": 0.942, "step": 4331 }, { "epoch": 1.3, "learning_rate": 4.18701171875e-05, "loss": 0.9926, "step": 4332 }, { "epoch": 1.3, "learning_rate": 4.186823918269231e-05, "loss": 0.8562, "step": 4333 }, { "epoch": 1.3, "learning_rate": 4.186636117788462e-05, "loss": 1.0923, "step": 4334 }, { "epoch": 1.3, "learning_rate": 4.1864483173076924e-05, "loss": 1.0698, "step": 4335 }, { "epoch": 1.3, "learning_rate": 4.186260516826923e-05, "loss": 1.0936, "step": 4336 }, { "epoch": 1.3, "learning_rate": 4.186072716346154e-05, "loss": 1.0414, "step": 4337 }, { "epoch": 1.3, "learning_rate": 4.1858849158653845e-05, "loss": 1.0183, "step": 4338 }, { "epoch": 1.3, "learning_rate": 4.185697115384616e-05, "loss": 1.2191, "step": 4339 }, { "epoch": 1.3, "learning_rate": 4.185509314903847e-05, "loss": 1.0318, "step": 4340 }, { "epoch": 1.3, "learning_rate": 4.185321514423077e-05, "loss": 1.2314, "step": 4341 }, { "epoch": 1.3, "learning_rate": 4.185133713942308e-05, "loss": 0.9742, "step": 4342 }, { "epoch": 1.3, "learning_rate": 4.184945913461539e-05, "loss": 0.8871, "step": 4343 }, { "epoch": 1.31, "learning_rate": 4.1847581129807694e-05, "loss": 1.041, "step": 4344 }, { "epoch": 1.31, "learning_rate": 4.1845703125e-05, "loss": 0.8986, "step": 4345 }, { "epoch": 1.31, "learning_rate": 4.184382512019231e-05, "loss": 1.1745, "step": 4346 }, { "epoch": 1.31, "learning_rate": 4.1841947115384616e-05, "loss": 1.0843, "step": 4347 }, { "epoch": 1.31, "learning_rate": 4.184006911057693e-05, "loss": 1.1279, "step": 4348 }, { "epoch": 1.31, "learning_rate": 4.183819110576923e-05, "loss": 1.0864, "step": 4349 }, { "epoch": 1.31, "learning_rate": 4.183631310096154e-05, "loss": 1.0978, "step": 4350 }, { "epoch": 1.31, "learning_rate": 4.183443509615385e-05, "loss": 0.9359, "step": 4351 }, { "epoch": 1.31, "learning_rate": 4.183255709134616e-05, "loss": 0.9231, "step": 4352 }, { "epoch": 1.31, "learning_rate": 4.1830679086538464e-05, "loss": 1.0439, "step": 4353 }, { "epoch": 1.31, "learning_rate": 4.1828801081730776e-05, "loss": 0.8783, "step": 4354 }, { "epoch": 1.31, "learning_rate": 4.182692307692308e-05, "loss": 1.0168, "step": 4355 }, { "epoch": 1.31, "learning_rate": 4.1825045072115386e-05, "loss": 1.1097, "step": 4356 }, { "epoch": 1.31, "learning_rate": 4.18231670673077e-05, "loss": 1.1381, "step": 4357 }, { "epoch": 1.31, "learning_rate": 4.18212890625e-05, "loss": 1.1075, "step": 4358 }, { "epoch": 1.31, "learning_rate": 4.181941105769231e-05, "loss": 0.915, "step": 4359 }, { "epoch": 1.31, "learning_rate": 4.181753305288462e-05, "loss": 1.201, "step": 4360 }, { "epoch": 1.31, "learning_rate": 4.181565504807692e-05, "loss": 1.2566, "step": 4361 }, { "epoch": 1.31, "learning_rate": 4.181377704326923e-05, "loss": 1.0198, "step": 4362 }, { "epoch": 1.31, "learning_rate": 4.181189903846154e-05, "loss": 1.2104, "step": 4363 }, { "epoch": 1.31, "learning_rate": 4.1810021033653844e-05, "loss": 1.0759, "step": 4364 }, { "epoch": 1.31, "learning_rate": 4.1808143028846156e-05, "loss": 0.8959, "step": 4365 }, { "epoch": 1.31, "learning_rate": 4.180626502403847e-05, "loss": 1.0542, "step": 4366 }, { "epoch": 1.31, "learning_rate": 4.180438701923077e-05, "loss": 0.9926, "step": 4367 }, { "epoch": 1.31, "learning_rate": 4.1802509014423084e-05, "loss": 1.1064, "step": 4368 }, { "epoch": 1.31, "learning_rate": 4.180063100961539e-05, "loss": 1.2447, "step": 4369 }, { "epoch": 1.31, "learning_rate": 4.179875300480769e-05, "loss": 1.0952, "step": 4370 }, { "epoch": 1.31, "learning_rate": 4.1796875000000005e-05, "loss": 1.1697, "step": 4371 }, { "epoch": 1.31, "learning_rate": 4.179499699519231e-05, "loss": 1.0439, "step": 4372 }, { "epoch": 1.31, "learning_rate": 4.1793118990384614e-05, "loss": 1.025, "step": 4373 }, { "epoch": 1.31, "learning_rate": 4.1791240985576926e-05, "loss": 1.1637, "step": 4374 }, { "epoch": 1.31, "learning_rate": 4.178936298076923e-05, "loss": 1.0615, "step": 4375 }, { "epoch": 1.31, "learning_rate": 4.1787484975961535e-05, "loss": 1.1032, "step": 4376 }, { "epoch": 1.32, "learning_rate": 4.178560697115385e-05, "loss": 0.8782, "step": 4377 }, { "epoch": 1.32, "learning_rate": 4.178372896634616e-05, "loss": 1.0384, "step": 4378 }, { "epoch": 1.32, "learning_rate": 4.178185096153846e-05, "loss": 0.9473, "step": 4379 }, { "epoch": 1.32, "learning_rate": 4.1779972956730775e-05, "loss": 1.2457, "step": 4380 }, { "epoch": 1.32, "learning_rate": 4.177809495192308e-05, "loss": 1.0837, "step": 4381 }, { "epoch": 1.32, "learning_rate": 4.1776216947115384e-05, "loss": 1.1689, "step": 4382 }, { "epoch": 1.32, "learning_rate": 4.1774338942307696e-05, "loss": 0.9685, "step": 4383 }, { "epoch": 1.32, "learning_rate": 4.17724609375e-05, "loss": 1.0552, "step": 4384 }, { "epoch": 1.32, "learning_rate": 4.177058293269231e-05, "loss": 1.1805, "step": 4385 }, { "epoch": 1.32, "learning_rate": 4.176870492788462e-05, "loss": 1.111, "step": 4386 }, { "epoch": 1.32, "learning_rate": 4.176682692307692e-05, "loss": 0.9858, "step": 4387 }, { "epoch": 1.32, "learning_rate": 4.176494891826923e-05, "loss": 1.0075, "step": 4388 }, { "epoch": 1.32, "learning_rate": 4.176307091346154e-05, "loss": 0.8896, "step": 4389 }, { "epoch": 1.32, "learning_rate": 4.176119290865384e-05, "loss": 0.9233, "step": 4390 }, { "epoch": 1.32, "learning_rate": 4.175931490384616e-05, "loss": 1.2245, "step": 4391 }, { "epoch": 1.32, "learning_rate": 4.1757436899038466e-05, "loss": 1.1261, "step": 4392 }, { "epoch": 1.32, "learning_rate": 4.175555889423077e-05, "loss": 1.1761, "step": 4393 }, { "epoch": 1.32, "learning_rate": 4.175368088942308e-05, "loss": 1.0395, "step": 4394 }, { "epoch": 1.32, "learning_rate": 4.175180288461539e-05, "loss": 1.1226, "step": 4395 }, { "epoch": 1.32, "learning_rate": 4.174992487980769e-05, "loss": 1.1639, "step": 4396 }, { "epoch": 1.32, "learning_rate": 4.1748046875e-05, "loss": 1.1411, "step": 4397 }, { "epoch": 1.32, "learning_rate": 4.174616887019231e-05, "loss": 1.1535, "step": 4398 }, { "epoch": 1.32, "learning_rate": 4.174429086538462e-05, "loss": 1.2803, "step": 4399 }, { "epoch": 1.32, "learning_rate": 4.1742412860576924e-05, "loss": 0.9259, "step": 4400 }, { "epoch": 1.32, "learning_rate": 4.174053485576923e-05, "loss": 1.084, "step": 4401 }, { "epoch": 1.32, "learning_rate": 4.173865685096154e-05, "loss": 0.9721, "step": 4402 }, { "epoch": 1.32, "learning_rate": 4.1736778846153845e-05, "loss": 1.0047, "step": 4403 }, { "epoch": 1.32, "learning_rate": 4.173490084134616e-05, "loss": 1.0389, "step": 4404 }, { "epoch": 1.32, "learning_rate": 4.173302283653847e-05, "loss": 1.0103, "step": 4405 }, { "epoch": 1.32, "learning_rate": 4.173114483173077e-05, "loss": 1.1968, "step": 4406 }, { "epoch": 1.32, "learning_rate": 4.172926682692308e-05, "loss": 1.088, "step": 4407 }, { "epoch": 1.32, "learning_rate": 4.172738882211539e-05, "loss": 1.0274, "step": 4408 }, { "epoch": 1.32, "learning_rate": 4.1725510817307694e-05, "loss": 1.0985, "step": 4409 }, { "epoch": 1.33, "learning_rate": 4.17236328125e-05, "loss": 0.9922, "step": 4410 }, { "epoch": 1.33, "learning_rate": 4.172175480769231e-05, "loss": 1.2927, "step": 4411 }, { "epoch": 1.33, "learning_rate": 4.1719876802884615e-05, "loss": 1.0755, "step": 4412 }, { "epoch": 1.33, "learning_rate": 4.171799879807693e-05, "loss": 1.0151, "step": 4413 }, { "epoch": 1.33, "learning_rate": 4.171612079326923e-05, "loss": 1.0583, "step": 4414 }, { "epoch": 1.33, "learning_rate": 4.1714242788461537e-05, "loss": 1.0394, "step": 4415 }, { "epoch": 1.33, "learning_rate": 4.171236478365385e-05, "loss": 1.0114, "step": 4416 }, { "epoch": 1.33, "learning_rate": 4.171048677884616e-05, "loss": 1.143, "step": 4417 }, { "epoch": 1.33, "learning_rate": 4.1708608774038464e-05, "loss": 0.914, "step": 4418 }, { "epoch": 1.33, "learning_rate": 4.1706730769230776e-05, "loss": 1.099, "step": 4419 }, { "epoch": 1.33, "learning_rate": 4.170485276442308e-05, "loss": 1.1074, "step": 4420 }, { "epoch": 1.33, "learning_rate": 4.1702974759615385e-05, "loss": 1.0764, "step": 4421 }, { "epoch": 1.33, "learning_rate": 4.17010967548077e-05, "loss": 1.0934, "step": 4422 }, { "epoch": 1.33, "learning_rate": 4.169921875e-05, "loss": 0.888, "step": 4423 }, { "epoch": 1.33, "learning_rate": 4.1697340745192307e-05, "loss": 1.109, "step": 4424 }, { "epoch": 1.33, "learning_rate": 4.169546274038462e-05, "loss": 0.9375, "step": 4425 }, { "epoch": 1.33, "learning_rate": 4.169358473557692e-05, "loss": 1.3628, "step": 4426 }, { "epoch": 1.33, "learning_rate": 4.1691706730769234e-05, "loss": 1.0901, "step": 4427 }, { "epoch": 1.33, "learning_rate": 4.168982872596154e-05, "loss": 1.192, "step": 4428 }, { "epoch": 1.33, "learning_rate": 4.1687950721153844e-05, "loss": 1.1601, "step": 4429 }, { "epoch": 1.33, "learning_rate": 4.1686072716346156e-05, "loss": 1.3017, "step": 4430 }, { "epoch": 1.33, "learning_rate": 4.168419471153847e-05, "loss": 0.9093, "step": 4431 }, { "epoch": 1.33, "learning_rate": 4.168231670673077e-05, "loss": 1.3392, "step": 4432 }, { "epoch": 1.33, "learning_rate": 4.1680438701923083e-05, "loss": 0.831, "step": 4433 }, { "epoch": 1.33, "learning_rate": 4.167856069711539e-05, "loss": 1.0442, "step": 4434 }, { "epoch": 1.33, "learning_rate": 4.167668269230769e-05, "loss": 1.1118, "step": 4435 }, { "epoch": 1.33, "learning_rate": 4.1674804687500004e-05, "loss": 0.9843, "step": 4436 }, { "epoch": 1.33, "learning_rate": 4.167292668269231e-05, "loss": 0.9408, "step": 4437 }, { "epoch": 1.33, "learning_rate": 4.1671048677884614e-05, "loss": 1.1596, "step": 4438 }, { "epoch": 1.33, "learning_rate": 4.1669170673076926e-05, "loss": 1.1041, "step": 4439 }, { "epoch": 1.33, "learning_rate": 4.166729266826923e-05, "loss": 1.0326, "step": 4440 }, { "epoch": 1.33, "learning_rate": 4.1665414663461535e-05, "loss": 1.1785, "step": 4441 }, { "epoch": 1.33, "learning_rate": 4.166353665865385e-05, "loss": 1.2129, "step": 4442 }, { "epoch": 1.34, "learning_rate": 4.166165865384616e-05, "loss": 1.1691, "step": 4443 }, { "epoch": 1.34, "learning_rate": 4.165978064903846e-05, "loss": 1.0702, "step": 4444 }, { "epoch": 1.34, "learning_rate": 4.1657902644230775e-05, "loss": 0.9999, "step": 4445 }, { "epoch": 1.34, "learning_rate": 4.165602463942308e-05, "loss": 0.9339, "step": 4446 }, { "epoch": 1.34, "learning_rate": 4.1654146634615384e-05, "loss": 1.1442, "step": 4447 }, { "epoch": 1.34, "learning_rate": 4.1652268629807696e-05, "loss": 1.2169, "step": 4448 }, { "epoch": 1.34, "learning_rate": 4.1650390625e-05, "loss": 1.151, "step": 4449 }, { "epoch": 1.34, "learning_rate": 4.164851262019231e-05, "loss": 1.213, "step": 4450 }, { "epoch": 1.34, "learning_rate": 4.164663461538462e-05, "loss": 1.0251, "step": 4451 }, { "epoch": 1.34, "learning_rate": 4.164475661057692e-05, "loss": 1.0698, "step": 4452 }, { "epoch": 1.34, "learning_rate": 4.164287860576923e-05, "loss": 1.2066, "step": 4453 }, { "epoch": 1.34, "learning_rate": 4.164100060096154e-05, "loss": 1.1504, "step": 4454 }, { "epoch": 1.34, "learning_rate": 4.163912259615384e-05, "loss": 0.7998, "step": 4455 }, { "epoch": 1.34, "learning_rate": 4.163724459134616e-05, "loss": 1.1254, "step": 4456 }, { "epoch": 1.34, "learning_rate": 4.1635366586538466e-05, "loss": 1.1761, "step": 4457 }, { "epoch": 1.34, "learning_rate": 4.163348858173077e-05, "loss": 1.1703, "step": 4458 }, { "epoch": 1.34, "learning_rate": 4.163161057692308e-05, "loss": 0.9124, "step": 4459 }, { "epoch": 1.34, "learning_rate": 4.162973257211539e-05, "loss": 1.3377, "step": 4460 }, { "epoch": 1.34, "learning_rate": 4.162785456730769e-05, "loss": 1.0938, "step": 4461 }, { "epoch": 1.34, "learning_rate": 4.16259765625e-05, "loss": 1.1424, "step": 4462 }, { "epoch": 1.34, "learning_rate": 4.162409855769231e-05, "loss": 1.0974, "step": 4463 }, { "epoch": 1.34, "learning_rate": 4.162222055288462e-05, "loss": 0.7809, "step": 4464 }, { "epoch": 1.34, "learning_rate": 4.1620342548076924e-05, "loss": 0.851, "step": 4465 }, { "epoch": 1.34, "learning_rate": 4.161846454326923e-05, "loss": 1.307, "step": 4466 }, { "epoch": 1.34, "learning_rate": 4.161658653846154e-05, "loss": 1.2402, "step": 4467 }, { "epoch": 1.34, "learning_rate": 4.1614708533653845e-05, "loss": 1.0717, "step": 4468 }, { "epoch": 1.34, "learning_rate": 4.161283052884616e-05, "loss": 0.9802, "step": 4469 }, { "epoch": 1.34, "learning_rate": 4.161095252403847e-05, "loss": 1.0948, "step": 4470 }, { "epoch": 1.34, "learning_rate": 4.160907451923077e-05, "loss": 1.0881, "step": 4471 }, { "epoch": 1.34, "learning_rate": 4.160719651442308e-05, "loss": 0.9566, "step": 4472 }, { "epoch": 1.34, "learning_rate": 4.160531850961539e-05, "loss": 1.2176, "step": 4473 }, { "epoch": 1.34, "learning_rate": 4.1603440504807694e-05, "loss": 1.1675, "step": 4474 }, { "epoch": 1.34, "learning_rate": 4.16015625e-05, "loss": 1.0962, "step": 4475 }, { "epoch": 1.34, "learning_rate": 4.159968449519231e-05, "loss": 1.0413, "step": 4476 }, { "epoch": 1.35, "learning_rate": 4.1597806490384615e-05, "loss": 0.8745, "step": 4477 }, { "epoch": 1.35, "learning_rate": 4.159592848557693e-05, "loss": 1.2623, "step": 4478 }, { "epoch": 1.35, "learning_rate": 4.159405048076923e-05, "loss": 0.8135, "step": 4479 }, { "epoch": 1.35, "learning_rate": 4.1592172475961536e-05, "loss": 0.9027, "step": 4480 }, { "epoch": 1.35, "learning_rate": 4.159029447115385e-05, "loss": 0.9136, "step": 4481 }, { "epoch": 1.35, "learning_rate": 4.158841646634616e-05, "loss": 1.2565, "step": 4482 }, { "epoch": 1.35, "learning_rate": 4.1586538461538464e-05, "loss": 1.1401, "step": 4483 }, { "epoch": 1.35, "learning_rate": 4.1584660456730776e-05, "loss": 1.2296, "step": 4484 }, { "epoch": 1.35, "learning_rate": 4.158278245192308e-05, "loss": 1.0216, "step": 4485 }, { "epoch": 1.35, "learning_rate": 4.1580904447115385e-05, "loss": 1.0148, "step": 4486 }, { "epoch": 1.35, "learning_rate": 4.15790264423077e-05, "loss": 1.3631, "step": 4487 }, { "epoch": 1.35, "learning_rate": 4.15771484375e-05, "loss": 1.0621, "step": 4488 }, { "epoch": 1.35, "learning_rate": 4.1575270432692306e-05, "loss": 1.1323, "step": 4489 }, { "epoch": 1.35, "learning_rate": 4.157339242788462e-05, "loss": 1.0251, "step": 4490 }, { "epoch": 1.35, "learning_rate": 4.157151442307692e-05, "loss": 1.0101, "step": 4491 }, { "epoch": 1.35, "learning_rate": 4.1569636418269234e-05, "loss": 1.3107, "step": 4492 }, { "epoch": 1.35, "learning_rate": 4.156775841346154e-05, "loss": 0.9249, "step": 4493 }, { "epoch": 1.35, "learning_rate": 4.1565880408653844e-05, "loss": 1.1522, "step": 4494 }, { "epoch": 1.35, "learning_rate": 4.1564002403846155e-05, "loss": 1.0981, "step": 4495 }, { "epoch": 1.35, "learning_rate": 4.156212439903847e-05, "loss": 0.9653, "step": 4496 }, { "epoch": 1.35, "learning_rate": 4.156024639423077e-05, "loss": 1.1324, "step": 4497 }, { "epoch": 1.35, "learning_rate": 4.155836838942308e-05, "loss": 1.0894, "step": 4498 }, { "epoch": 1.35, "learning_rate": 4.155649038461539e-05, "loss": 1.0223, "step": 4499 }, { "epoch": 1.35, "learning_rate": 4.155461237980769e-05, "loss": 0.9709, "step": 4500 }, { "epoch": 1.35, "learning_rate": 4.1552734375000004e-05, "loss": 1.1388, "step": 4501 }, { "epoch": 1.35, "learning_rate": 4.155085637019231e-05, "loss": 0.6975, "step": 4502 }, { "epoch": 1.35, "learning_rate": 4.1548978365384614e-05, "loss": 1.0194, "step": 4503 }, { "epoch": 1.35, "learning_rate": 4.1547100360576925e-05, "loss": 1.2159, "step": 4504 }, { "epoch": 1.35, "learning_rate": 4.154522235576923e-05, "loss": 0.9112, "step": 4505 }, { "epoch": 1.35, "learning_rate": 4.1543344350961535e-05, "loss": 1.1364, "step": 4506 }, { "epoch": 1.35, "learning_rate": 4.1541466346153847e-05, "loss": 0.9342, "step": 4507 }, { "epoch": 1.35, "learning_rate": 4.153958834134616e-05, "loss": 1.3247, "step": 4508 }, { "epoch": 1.35, "learning_rate": 4.153771033653846e-05, "loss": 1.0782, "step": 4509 }, { "epoch": 1.36, "learning_rate": 4.1535832331730774e-05, "loss": 1.0762, "step": 4510 }, { "epoch": 1.36, "learning_rate": 4.153395432692308e-05, "loss": 1.0525, "step": 4511 }, { "epoch": 1.36, "learning_rate": 4.153207632211539e-05, "loss": 1.1308, "step": 4512 }, { "epoch": 1.36, "learning_rate": 4.1530198317307696e-05, "loss": 1.1214, "step": 4513 }, { "epoch": 1.36, "learning_rate": 4.15283203125e-05, "loss": 1.0036, "step": 4514 }, { "epoch": 1.36, "learning_rate": 4.152644230769231e-05, "loss": 1.1219, "step": 4515 }, { "epoch": 1.36, "learning_rate": 4.152456430288462e-05, "loss": 1.1868, "step": 4516 }, { "epoch": 1.36, "learning_rate": 4.152268629807692e-05, "loss": 1.1742, "step": 4517 }, { "epoch": 1.36, "learning_rate": 4.152080829326923e-05, "loss": 0.9998, "step": 4518 }, { "epoch": 1.36, "learning_rate": 4.151893028846154e-05, "loss": 1.1783, "step": 4519 }, { "epoch": 1.36, "learning_rate": 4.151705228365384e-05, "loss": 1.0021, "step": 4520 }, { "epoch": 1.36, "learning_rate": 4.151517427884616e-05, "loss": 1.1977, "step": 4521 }, { "epoch": 1.36, "learning_rate": 4.1513296274038466e-05, "loss": 1.1688, "step": 4522 }, { "epoch": 1.36, "learning_rate": 4.151141826923077e-05, "loss": 1.1317, "step": 4523 }, { "epoch": 1.36, "learning_rate": 4.150954026442308e-05, "loss": 1.3771, "step": 4524 }, { "epoch": 1.36, "learning_rate": 4.150766225961539e-05, "loss": 1.0374, "step": 4525 }, { "epoch": 1.36, "learning_rate": 4.150578425480769e-05, "loss": 1.312, "step": 4526 }, { "epoch": 1.36, "learning_rate": 4.150390625e-05, "loss": 1.2541, "step": 4527 }, { "epoch": 1.36, "learning_rate": 4.150202824519231e-05, "loss": 1.02, "step": 4528 }, { "epoch": 1.36, "learning_rate": 4.150015024038462e-05, "loss": 0.867, "step": 4529 }, { "epoch": 1.36, "learning_rate": 4.1498272235576924e-05, "loss": 0.8758, "step": 4530 }, { "epoch": 1.36, "learning_rate": 4.149639423076923e-05, "loss": 1.1708, "step": 4531 }, { "epoch": 1.36, "learning_rate": 4.149451622596154e-05, "loss": 1.3394, "step": 4532 }, { "epoch": 1.36, "learning_rate": 4.1492638221153845e-05, "loss": 1.2353, "step": 4533 }, { "epoch": 1.36, "learning_rate": 4.149076021634616e-05, "loss": 1.1982, "step": 4534 }, { "epoch": 1.36, "learning_rate": 4.148888221153847e-05, "loss": 0.9947, "step": 4535 }, { "epoch": 1.36, "learning_rate": 4.148700420673077e-05, "loss": 1.3495, "step": 4536 }, { "epoch": 1.36, "learning_rate": 4.148512620192308e-05, "loss": 0.9296, "step": 4537 }, { "epoch": 1.36, "learning_rate": 4.148324819711539e-05, "loss": 1.1062, "step": 4538 }, { "epoch": 1.36, "learning_rate": 4.1481370192307694e-05, "loss": 1.1106, "step": 4539 }, { "epoch": 1.36, "learning_rate": 4.14794921875e-05, "loss": 1.0317, "step": 4540 }, { "epoch": 1.36, "learning_rate": 4.147761418269231e-05, "loss": 1.1077, "step": 4541 }, { "epoch": 1.36, "learning_rate": 4.1475736177884615e-05, "loss": 1.3691, "step": 4542 }, { "epoch": 1.37, "learning_rate": 4.147385817307693e-05, "loss": 1.2078, "step": 4543 }, { "epoch": 1.37, "learning_rate": 4.147198016826923e-05, "loss": 0.843, "step": 4544 }, { "epoch": 1.37, "learning_rate": 4.1470102163461536e-05, "loss": 1.1016, "step": 4545 }, { "epoch": 1.37, "learning_rate": 4.146822415865385e-05, "loss": 1.1137, "step": 4546 }, { "epoch": 1.37, "learning_rate": 4.146634615384616e-05, "loss": 1.1932, "step": 4547 }, { "epoch": 1.37, "learning_rate": 4.1464468149038464e-05, "loss": 1.2465, "step": 4548 }, { "epoch": 1.37, "learning_rate": 4.1462590144230776e-05, "loss": 1.2269, "step": 4549 }, { "epoch": 1.37, "learning_rate": 4.146071213942308e-05, "loss": 0.905, "step": 4550 }, { "epoch": 1.37, "learning_rate": 4.1458834134615385e-05, "loss": 1.332, "step": 4551 }, { "epoch": 1.37, "learning_rate": 4.14569561298077e-05, "loss": 1.1282, "step": 4552 }, { "epoch": 1.37, "learning_rate": 4.1455078125e-05, "loss": 0.9571, "step": 4553 }, { "epoch": 1.37, "learning_rate": 4.1453200120192306e-05, "loss": 1.0567, "step": 4554 }, { "epoch": 1.37, "learning_rate": 4.145132211538462e-05, "loss": 1.2019, "step": 4555 }, { "epoch": 1.37, "learning_rate": 4.144944411057692e-05, "loss": 0.9522, "step": 4556 }, { "epoch": 1.37, "learning_rate": 4.1447566105769234e-05, "loss": 0.9919, "step": 4557 }, { "epoch": 1.37, "learning_rate": 4.144568810096154e-05, "loss": 1.2148, "step": 4558 }, { "epoch": 1.37, "learning_rate": 4.1443810096153844e-05, "loss": 1.2323, "step": 4559 }, { "epoch": 1.37, "learning_rate": 4.1441932091346155e-05, "loss": 0.9652, "step": 4560 }, { "epoch": 1.37, "learning_rate": 4.144005408653847e-05, "loss": 1.0467, "step": 4561 }, { "epoch": 1.37, "learning_rate": 4.143817608173077e-05, "loss": 1.1797, "step": 4562 }, { "epoch": 1.37, "learning_rate": 4.143629807692308e-05, "loss": 1.1207, "step": 4563 }, { "epoch": 1.37, "learning_rate": 4.143442007211539e-05, "loss": 1.0321, "step": 4564 }, { "epoch": 1.37, "learning_rate": 4.143254206730769e-05, "loss": 1.2476, "step": 4565 }, { "epoch": 1.37, "learning_rate": 4.1430664062500004e-05, "loss": 1.0785, "step": 4566 }, { "epoch": 1.37, "learning_rate": 4.142878605769231e-05, "loss": 1.0521, "step": 4567 }, { "epoch": 1.37, "learning_rate": 4.1426908052884614e-05, "loss": 1.3621, "step": 4568 }, { "epoch": 1.37, "learning_rate": 4.1425030048076925e-05, "loss": 0.9968, "step": 4569 }, { "epoch": 1.37, "learning_rate": 4.142315204326923e-05, "loss": 1.326, "step": 4570 }, { "epoch": 1.37, "learning_rate": 4.142127403846154e-05, "loss": 0.9325, "step": 4571 }, { "epoch": 1.37, "learning_rate": 4.1419396033653846e-05, "loss": 0.9573, "step": 4572 }, { "epoch": 1.37, "learning_rate": 4.141751802884616e-05, "loss": 1.3253, "step": 4573 }, { "epoch": 1.37, "learning_rate": 4.141564002403846e-05, "loss": 1.1808, "step": 4574 }, { "epoch": 1.37, "learning_rate": 4.1413762019230774e-05, "loss": 0.96, "step": 4575 }, { "epoch": 1.38, "learning_rate": 4.141188401442308e-05, "loss": 1.0691, "step": 4576 }, { "epoch": 1.38, "learning_rate": 4.141000600961539e-05, "loss": 1.2322, "step": 4577 }, { "epoch": 1.38, "learning_rate": 4.1408128004807695e-05, "loss": 1.0673, "step": 4578 }, { "epoch": 1.38, "learning_rate": 4.140625e-05, "loss": 1.0529, "step": 4579 }, { "epoch": 1.38, "learning_rate": 4.140437199519231e-05, "loss": 1.2534, "step": 4580 }, { "epoch": 1.38, "learning_rate": 4.1402493990384617e-05, "loss": 0.9547, "step": 4581 }, { "epoch": 1.38, "learning_rate": 4.140061598557692e-05, "loss": 0.8657, "step": 4582 }, { "epoch": 1.38, "learning_rate": 4.139873798076923e-05, "loss": 1.1556, "step": 4583 }, { "epoch": 1.38, "learning_rate": 4.139685997596154e-05, "loss": 1.1049, "step": 4584 }, { "epoch": 1.38, "learning_rate": 4.139498197115384e-05, "loss": 1.1783, "step": 4585 }, { "epoch": 1.38, "learning_rate": 4.139310396634616e-05, "loss": 0.7551, "step": 4586 }, { "epoch": 1.38, "learning_rate": 4.1391225961538465e-05, "loss": 1.141, "step": 4587 }, { "epoch": 1.38, "learning_rate": 4.138934795673077e-05, "loss": 1.0293, "step": 4588 }, { "epoch": 1.38, "learning_rate": 4.138746995192308e-05, "loss": 0.9221, "step": 4589 }, { "epoch": 1.38, "learning_rate": 4.1385591947115387e-05, "loss": 1.1511, "step": 4590 }, { "epoch": 1.38, "learning_rate": 4.13837139423077e-05, "loss": 1.0807, "step": 4591 }, { "epoch": 1.38, "learning_rate": 4.13818359375e-05, "loss": 1.1255, "step": 4592 }, { "epoch": 1.38, "learning_rate": 4.137995793269231e-05, "loss": 0.9702, "step": 4593 }, { "epoch": 1.38, "learning_rate": 4.137807992788462e-05, "loss": 1.0554, "step": 4594 }, { "epoch": 1.38, "learning_rate": 4.1376201923076924e-05, "loss": 1.1535, "step": 4595 }, { "epoch": 1.38, "learning_rate": 4.137432391826923e-05, "loss": 1.1667, "step": 4596 }, { "epoch": 1.38, "learning_rate": 4.137244591346154e-05, "loss": 1.4673, "step": 4597 }, { "epoch": 1.38, "learning_rate": 4.1370567908653845e-05, "loss": 1.2594, "step": 4598 }, { "epoch": 1.38, "learning_rate": 4.1368689903846157e-05, "loss": 1.0332, "step": 4599 }, { "epoch": 1.38, "learning_rate": 4.136681189903847e-05, "loss": 1.0874, "step": 4600 }, { "epoch": 1.38, "learning_rate": 4.136493389423077e-05, "loss": 0.9514, "step": 4601 }, { "epoch": 1.38, "learning_rate": 4.136305588942308e-05, "loss": 0.9796, "step": 4602 }, { "epoch": 1.38, "learning_rate": 4.136117788461539e-05, "loss": 1.2337, "step": 4603 }, { "epoch": 1.38, "learning_rate": 4.1359299879807694e-05, "loss": 0.9301, "step": 4604 }, { "epoch": 1.38, "learning_rate": 4.1357421875e-05, "loss": 0.8773, "step": 4605 }, { "epoch": 1.38, "learning_rate": 4.135554387019231e-05, "loss": 1.1068, "step": 4606 }, { "epoch": 1.38, "learning_rate": 4.1353665865384615e-05, "loss": 1.1568, "step": 4607 }, { "epoch": 1.38, "learning_rate": 4.135178786057693e-05, "loss": 1.0641, "step": 4608 }, { "epoch": 1.38, "learning_rate": 4.134990985576923e-05, "loss": 1.1587, "step": 4609 }, { "epoch": 1.39, "learning_rate": 4.1348031850961536e-05, "loss": 1.3995, "step": 4610 }, { "epoch": 1.39, "learning_rate": 4.134615384615385e-05, "loss": 0.9914, "step": 4611 }, { "epoch": 1.39, "learning_rate": 4.134427584134616e-05, "loss": 1.1201, "step": 4612 }, { "epoch": 1.39, "learning_rate": 4.1342397836538464e-05, "loss": 1.0867, "step": 4613 }, { "epoch": 1.39, "learning_rate": 4.1340519831730776e-05, "loss": 1.39, "step": 4614 }, { "epoch": 1.39, "learning_rate": 4.133864182692308e-05, "loss": 1.2794, "step": 4615 }, { "epoch": 1.39, "learning_rate": 4.1336763822115385e-05, "loss": 1.2302, "step": 4616 }, { "epoch": 1.39, "learning_rate": 4.13348858173077e-05, "loss": 1.1234, "step": 4617 }, { "epoch": 1.39, "learning_rate": 4.13330078125e-05, "loss": 1.1188, "step": 4618 }, { "epoch": 1.39, "learning_rate": 4.1331129807692306e-05, "loss": 0.8064, "step": 4619 }, { "epoch": 1.39, "learning_rate": 4.132925180288462e-05, "loss": 1.1233, "step": 4620 }, { "epoch": 1.39, "learning_rate": 4.132737379807692e-05, "loss": 1.0627, "step": 4621 }, { "epoch": 1.39, "learning_rate": 4.1325495793269234e-05, "loss": 0.8004, "step": 4622 }, { "epoch": 1.39, "learning_rate": 4.132361778846154e-05, "loss": 1.4913, "step": 4623 }, { "epoch": 1.39, "learning_rate": 4.1321739783653844e-05, "loss": 1.4387, "step": 4624 }, { "epoch": 1.39, "learning_rate": 4.1319861778846155e-05, "loss": 1.3797, "step": 4625 }, { "epoch": 1.39, "learning_rate": 4.131798377403847e-05, "loss": 1.1141, "step": 4626 }, { "epoch": 1.39, "learning_rate": 4.131610576923077e-05, "loss": 1.1802, "step": 4627 }, { "epoch": 1.39, "learning_rate": 4.131422776442308e-05, "loss": 1.1515, "step": 4628 }, { "epoch": 1.39, "learning_rate": 4.131234975961539e-05, "loss": 1.2205, "step": 4629 }, { "epoch": 1.39, "learning_rate": 4.131047175480769e-05, "loss": 1.0692, "step": 4630 }, { "epoch": 1.39, "learning_rate": 4.1308593750000004e-05, "loss": 1.4256, "step": 4631 }, { "epoch": 1.39, "learning_rate": 4.130671574519231e-05, "loss": 1.4765, "step": 4632 }, { "epoch": 1.39, "learning_rate": 4.1304837740384614e-05, "loss": 0.9888, "step": 4633 }, { "epoch": 1.39, "learning_rate": 4.1302959735576925e-05, "loss": 1.2862, "step": 4634 }, { "epoch": 1.39, "learning_rate": 4.130108173076923e-05, "loss": 1.1446, "step": 4635 }, { "epoch": 1.39, "learning_rate": 4.129920372596154e-05, "loss": 1.091, "step": 4636 }, { "epoch": 1.39, "learning_rate": 4.1297325721153846e-05, "loss": 1.1269, "step": 4637 }, { "epoch": 1.39, "learning_rate": 4.129544771634616e-05, "loss": 1.1025, "step": 4638 }, { "epoch": 1.39, "learning_rate": 4.129356971153846e-05, "loss": 0.9996, "step": 4639 }, { "epoch": 1.39, "learning_rate": 4.1291691706730774e-05, "loss": 1.0848, "step": 4640 }, { "epoch": 1.39, "learning_rate": 4.128981370192308e-05, "loss": 1.1085, "step": 4641 }, { "epoch": 1.39, "learning_rate": 4.128793569711539e-05, "loss": 1.2676, "step": 4642 }, { "epoch": 1.4, "learning_rate": 4.1286057692307695e-05, "loss": 0.9558, "step": 4643 }, { "epoch": 1.4, "learning_rate": 4.12841796875e-05, "loss": 1.0426, "step": 4644 }, { "epoch": 1.4, "learning_rate": 4.128230168269231e-05, "loss": 1.1288, "step": 4645 }, { "epoch": 1.4, "learning_rate": 4.1280423677884616e-05, "loss": 1.2406, "step": 4646 }, { "epoch": 1.4, "learning_rate": 4.127854567307692e-05, "loss": 1.234, "step": 4647 }, { "epoch": 1.4, "learning_rate": 4.127666766826923e-05, "loss": 1.1895, "step": 4648 }, { "epoch": 1.4, "learning_rate": 4.127478966346154e-05, "loss": 0.9174, "step": 4649 }, { "epoch": 1.4, "learning_rate": 4.127291165865384e-05, "loss": 1.2864, "step": 4650 }, { "epoch": 1.4, "learning_rate": 4.127103365384616e-05, "loss": 1.3069, "step": 4651 }, { "epoch": 1.4, "learning_rate": 4.1269155649038465e-05, "loss": 1.1426, "step": 4652 }, { "epoch": 1.4, "learning_rate": 4.126727764423077e-05, "loss": 1.1929, "step": 4653 }, { "epoch": 1.4, "learning_rate": 4.126539963942308e-05, "loss": 1.0501, "step": 4654 }, { "epoch": 1.4, "learning_rate": 4.1263521634615386e-05, "loss": 1.2399, "step": 4655 }, { "epoch": 1.4, "learning_rate": 4.12616436298077e-05, "loss": 0.8736, "step": 4656 }, { "epoch": 1.4, "learning_rate": 4.1259765625e-05, "loss": 1.0282, "step": 4657 }, { "epoch": 1.4, "learning_rate": 4.125788762019231e-05, "loss": 1.308, "step": 4658 }, { "epoch": 1.4, "learning_rate": 4.125600961538462e-05, "loss": 0.769, "step": 4659 }, { "epoch": 1.4, "learning_rate": 4.1254131610576924e-05, "loss": 1.2007, "step": 4660 }, { "epoch": 1.4, "learning_rate": 4.125225360576923e-05, "loss": 1.0472, "step": 4661 }, { "epoch": 1.4, "learning_rate": 4.125037560096154e-05, "loss": 1.1621, "step": 4662 }, { "epoch": 1.4, "learning_rate": 4.1248497596153845e-05, "loss": 1.2016, "step": 4663 }, { "epoch": 1.4, "learning_rate": 4.1246619591346156e-05, "loss": 1.1089, "step": 4664 }, { "epoch": 1.4, "learning_rate": 4.124474158653847e-05, "loss": 1.0847, "step": 4665 }, { "epoch": 1.4, "learning_rate": 4.124286358173077e-05, "loss": 0.9838, "step": 4666 }, { "epoch": 1.4, "learning_rate": 4.124098557692308e-05, "loss": 1.1104, "step": 4667 }, { "epoch": 1.4, "learning_rate": 4.123910757211539e-05, "loss": 1.1106, "step": 4668 }, { "epoch": 1.4, "learning_rate": 4.1237229567307694e-05, "loss": 1.135, "step": 4669 }, { "epoch": 1.4, "learning_rate": 4.12353515625e-05, "loss": 1.228, "step": 4670 }, { "epoch": 1.4, "learning_rate": 4.123347355769231e-05, "loss": 1.0687, "step": 4671 }, { "epoch": 1.4, "learning_rate": 4.1231595552884615e-05, "loss": 0.989, "step": 4672 }, { "epoch": 1.4, "learning_rate": 4.1229717548076927e-05, "loss": 0.9565, "step": 4673 }, { "epoch": 1.4, "learning_rate": 4.122783954326923e-05, "loss": 1.4499, "step": 4674 }, { "epoch": 1.4, "learning_rate": 4.1225961538461536e-05, "loss": 1.1529, "step": 4675 }, { "epoch": 1.41, "learning_rate": 4.122408353365385e-05, "loss": 1.0159, "step": 4676 }, { "epoch": 1.41, "learning_rate": 4.122220552884616e-05, "loss": 1.2027, "step": 4677 }, { "epoch": 1.41, "learning_rate": 4.1220327524038464e-05, "loss": 1.4455, "step": 4678 }, { "epoch": 1.41, "learning_rate": 4.1218449519230776e-05, "loss": 1.1105, "step": 4679 }, { "epoch": 1.41, "learning_rate": 4.121657151442308e-05, "loss": 1.2246, "step": 4680 }, { "epoch": 1.41, "learning_rate": 4.1214693509615385e-05, "loss": 1.1455, "step": 4681 }, { "epoch": 1.41, "learning_rate": 4.1212815504807697e-05, "loss": 1.0375, "step": 4682 }, { "epoch": 1.41, "learning_rate": 4.12109375e-05, "loss": 0.9696, "step": 4683 }, { "epoch": 1.41, "learning_rate": 4.1209059495192306e-05, "loss": 0.8664, "step": 4684 }, { "epoch": 1.41, "learning_rate": 4.120718149038462e-05, "loss": 1.2098, "step": 4685 }, { "epoch": 1.41, "learning_rate": 4.120530348557692e-05, "loss": 1.0245, "step": 4686 }, { "epoch": 1.41, "learning_rate": 4.1203425480769234e-05, "loss": 1.304, "step": 4687 }, { "epoch": 1.41, "learning_rate": 4.120154747596154e-05, "loss": 1.1492, "step": 4688 }, { "epoch": 1.41, "learning_rate": 4.1199669471153844e-05, "loss": 1.1899, "step": 4689 }, { "epoch": 1.41, "learning_rate": 4.1197791466346155e-05, "loss": 0.8863, "step": 4690 }, { "epoch": 1.41, "learning_rate": 4.119591346153847e-05, "loss": 1.2864, "step": 4691 }, { "epoch": 1.41, "learning_rate": 4.119403545673077e-05, "loss": 1.1141, "step": 4692 }, { "epoch": 1.41, "learning_rate": 4.119215745192308e-05, "loss": 1.3013, "step": 4693 }, { "epoch": 1.41, "learning_rate": 4.119027944711539e-05, "loss": 1.2387, "step": 4694 }, { "epoch": 1.41, "learning_rate": 4.118840144230769e-05, "loss": 1.2737, "step": 4695 }, { "epoch": 1.41, "learning_rate": 4.1186523437500004e-05, "loss": 0.8883, "step": 4696 }, { "epoch": 1.41, "learning_rate": 4.118464543269231e-05, "loss": 1.2447, "step": 4697 }, { "epoch": 1.41, "learning_rate": 4.1182767427884614e-05, "loss": 0.9324, "step": 4698 }, { "epoch": 1.41, "learning_rate": 4.1180889423076925e-05, "loss": 0.897, "step": 4699 }, { "epoch": 1.41, "learning_rate": 4.117901141826923e-05, "loss": 1.3602, "step": 4700 }, { "epoch": 1.41, "learning_rate": 4.117713341346154e-05, "loss": 1.1971, "step": 4701 }, { "epoch": 1.41, "learning_rate": 4.1175255408653846e-05, "loss": 1.0626, "step": 4702 }, { "epoch": 1.41, "learning_rate": 4.117337740384616e-05, "loss": 1.1342, "step": 4703 }, { "epoch": 1.41, "learning_rate": 4.117149939903846e-05, "loss": 1.1858, "step": 4704 }, { "epoch": 1.41, "learning_rate": 4.1169621394230774e-05, "loss": 1.1747, "step": 4705 }, { "epoch": 1.41, "learning_rate": 4.116774338942308e-05, "loss": 1.0059, "step": 4706 }, { "epoch": 1.41, "learning_rate": 4.116586538461539e-05, "loss": 1.1031, "step": 4707 }, { "epoch": 1.41, "learning_rate": 4.1163987379807695e-05, "loss": 0.9041, "step": 4708 }, { "epoch": 1.41, "learning_rate": 4.1162109375e-05, "loss": 1.1384, "step": 4709 }, { "epoch": 1.42, "learning_rate": 4.116023137019231e-05, "loss": 1.0637, "step": 4710 }, { "epoch": 1.42, "learning_rate": 4.1158353365384616e-05, "loss": 1.2731, "step": 4711 }, { "epoch": 1.42, "learning_rate": 4.115647536057692e-05, "loss": 1.08, "step": 4712 }, { "epoch": 1.42, "learning_rate": 4.115459735576923e-05, "loss": 1.1794, "step": 4713 }, { "epoch": 1.42, "learning_rate": 4.115271935096154e-05, "loss": 1.029, "step": 4714 }, { "epoch": 1.42, "learning_rate": 4.115084134615385e-05, "loss": 1.0376, "step": 4715 }, { "epoch": 1.42, "learning_rate": 4.114896334134616e-05, "loss": 0.9106, "step": 4716 }, { "epoch": 1.42, "learning_rate": 4.1147085336538465e-05, "loss": 1.0804, "step": 4717 }, { "epoch": 1.42, "learning_rate": 4.114520733173077e-05, "loss": 1.1201, "step": 4718 }, { "epoch": 1.42, "learning_rate": 4.114332932692308e-05, "loss": 1.1431, "step": 4719 }, { "epoch": 1.42, "learning_rate": 4.1141451322115386e-05, "loss": 0.8972, "step": 4720 }, { "epoch": 1.42, "learning_rate": 4.11395733173077e-05, "loss": 1.2841, "step": 4721 }, { "epoch": 1.42, "learning_rate": 4.11376953125e-05, "loss": 1.1206, "step": 4722 }, { "epoch": 1.42, "learning_rate": 4.113581730769231e-05, "loss": 1.3332, "step": 4723 }, { "epoch": 1.42, "learning_rate": 4.113393930288462e-05, "loss": 1.1934, "step": 4724 }, { "epoch": 1.42, "learning_rate": 4.1132061298076924e-05, "loss": 1.2379, "step": 4725 }, { "epoch": 1.42, "learning_rate": 4.113018329326923e-05, "loss": 1.1265, "step": 4726 }, { "epoch": 1.42, "learning_rate": 4.112830528846154e-05, "loss": 1.0362, "step": 4727 }, { "epoch": 1.42, "learning_rate": 4.1126427283653845e-05, "loss": 1.2251, "step": 4728 }, { "epoch": 1.42, "learning_rate": 4.1124549278846156e-05, "loss": 0.8414, "step": 4729 }, { "epoch": 1.42, "learning_rate": 4.112267127403847e-05, "loss": 1.1567, "step": 4730 }, { "epoch": 1.42, "learning_rate": 4.112079326923077e-05, "loss": 1.1507, "step": 4731 }, { "epoch": 1.42, "learning_rate": 4.111891526442308e-05, "loss": 1.0388, "step": 4732 }, { "epoch": 1.42, "learning_rate": 4.111703725961539e-05, "loss": 1.2816, "step": 4733 }, { "epoch": 1.42, "learning_rate": 4.1115159254807694e-05, "loss": 1.0933, "step": 4734 }, { "epoch": 1.42, "learning_rate": 4.1113281250000005e-05, "loss": 1.13, "step": 4735 }, { "epoch": 1.42, "learning_rate": 4.111140324519231e-05, "loss": 1.2411, "step": 4736 }, { "epoch": 1.42, "learning_rate": 4.1109525240384615e-05, "loss": 1.4372, "step": 4737 }, { "epoch": 1.42, "learning_rate": 4.1107647235576926e-05, "loss": 1.0236, "step": 4738 }, { "epoch": 1.42, "learning_rate": 4.110576923076923e-05, "loss": 0.8636, "step": 4739 }, { "epoch": 1.42, "learning_rate": 4.1103891225961536e-05, "loss": 1.1466, "step": 4740 }, { "epoch": 1.42, "learning_rate": 4.110201322115385e-05, "loss": 1.3574, "step": 4741 }, { "epoch": 1.42, "learning_rate": 4.110013521634616e-05, "loss": 1.1074, "step": 4742 }, { "epoch": 1.43, "learning_rate": 4.1098257211538464e-05, "loss": 1.1378, "step": 4743 }, { "epoch": 1.43, "learning_rate": 4.1096379206730775e-05, "loss": 1.1532, "step": 4744 }, { "epoch": 1.43, "learning_rate": 4.109450120192308e-05, "loss": 0.8719, "step": 4745 }, { "epoch": 1.43, "learning_rate": 4.1092623197115385e-05, "loss": 1.2424, "step": 4746 }, { "epoch": 1.43, "learning_rate": 4.1090745192307696e-05, "loss": 1.0481, "step": 4747 }, { "epoch": 1.43, "learning_rate": 4.10888671875e-05, "loss": 1.0355, "step": 4748 }, { "epoch": 1.43, "learning_rate": 4.1086989182692306e-05, "loss": 1.1003, "step": 4749 }, { "epoch": 1.43, "learning_rate": 4.108511117788462e-05, "loss": 0.9936, "step": 4750 }, { "epoch": 1.43, "learning_rate": 4.108323317307692e-05, "loss": 0.9868, "step": 4751 }, { "epoch": 1.43, "learning_rate": 4.1081355168269234e-05, "loss": 1.1233, "step": 4752 }, { "epoch": 1.43, "learning_rate": 4.107947716346154e-05, "loss": 1.1282, "step": 4753 }, { "epoch": 1.43, "learning_rate": 4.1077599158653843e-05, "loss": 1.1868, "step": 4754 }, { "epoch": 1.43, "learning_rate": 4.107572115384616e-05, "loss": 1.259, "step": 4755 }, { "epoch": 1.43, "learning_rate": 4.1073843149038467e-05, "loss": 1.1846, "step": 4756 }, { "epoch": 1.43, "learning_rate": 4.107196514423077e-05, "loss": 1.1738, "step": 4757 }, { "epoch": 1.43, "learning_rate": 4.107008713942308e-05, "loss": 0.9515, "step": 4758 }, { "epoch": 1.43, "learning_rate": 4.106820913461539e-05, "loss": 1.0744, "step": 4759 }, { "epoch": 1.43, "learning_rate": 4.106633112980769e-05, "loss": 1.0704, "step": 4760 }, { "epoch": 1.43, "learning_rate": 4.1064453125000004e-05, "loss": 1.0017, "step": 4761 }, { "epoch": 1.43, "learning_rate": 4.106257512019231e-05, "loss": 1.3854, "step": 4762 }, { "epoch": 1.43, "learning_rate": 4.1060697115384613e-05, "loss": 1.0112, "step": 4763 }, { "epoch": 1.43, "learning_rate": 4.1058819110576925e-05, "loss": 0.9833, "step": 4764 }, { "epoch": 1.43, "learning_rate": 4.105694110576923e-05, "loss": 1.2484, "step": 4765 }, { "epoch": 1.43, "learning_rate": 4.105506310096154e-05, "loss": 1.0082, "step": 4766 }, { "epoch": 1.43, "learning_rate": 4.1053185096153846e-05, "loss": 1.116, "step": 4767 }, { "epoch": 1.43, "learning_rate": 4.105130709134616e-05, "loss": 1.2619, "step": 4768 }, { "epoch": 1.43, "learning_rate": 4.104942908653846e-05, "loss": 1.0614, "step": 4769 }, { "epoch": 1.43, "learning_rate": 4.1047551081730774e-05, "loss": 1.2431, "step": 4770 }, { "epoch": 1.43, "learning_rate": 4.104567307692308e-05, "loss": 1.1815, "step": 4771 }, { "epoch": 1.43, "learning_rate": 4.104379507211539e-05, "loss": 1.081, "step": 4772 }, { "epoch": 1.43, "learning_rate": 4.1041917067307695e-05, "loss": 1.0227, "step": 4773 }, { "epoch": 1.43, "learning_rate": 4.10400390625e-05, "loss": 1.0491, "step": 4774 }, { "epoch": 1.43, "learning_rate": 4.103816105769231e-05, "loss": 1.3183, "step": 4775 }, { "epoch": 1.44, "learning_rate": 4.1036283052884616e-05, "loss": 1.3482, "step": 4776 }, { "epoch": 1.44, "learning_rate": 4.103440504807692e-05, "loss": 0.945, "step": 4777 }, { "epoch": 1.44, "learning_rate": 4.103252704326923e-05, "loss": 1.327, "step": 4778 }, { "epoch": 1.44, "learning_rate": 4.103064903846154e-05, "loss": 0.7786, "step": 4779 }, { "epoch": 1.44, "learning_rate": 4.102877103365385e-05, "loss": 1.2689, "step": 4780 }, { "epoch": 1.44, "learning_rate": 4.102689302884616e-05, "loss": 0.9863, "step": 4781 }, { "epoch": 1.44, "learning_rate": 4.1025015024038465e-05, "loss": 1.3212, "step": 4782 }, { "epoch": 1.44, "learning_rate": 4.102313701923077e-05, "loss": 0.9707, "step": 4783 }, { "epoch": 1.44, "learning_rate": 4.102125901442308e-05, "loss": 1.2754, "step": 4784 }, { "epoch": 1.44, "learning_rate": 4.1019381009615386e-05, "loss": 1.1929, "step": 4785 }, { "epoch": 1.44, "learning_rate": 4.10175030048077e-05, "loss": 1.2579, "step": 4786 }, { "epoch": 1.44, "learning_rate": 4.1015625e-05, "loss": 1.0498, "step": 4787 }, { "epoch": 1.44, "learning_rate": 4.101374699519231e-05, "loss": 0.9357, "step": 4788 }, { "epoch": 1.44, "learning_rate": 4.101186899038462e-05, "loss": 0.9303, "step": 4789 }, { "epoch": 1.44, "learning_rate": 4.1009990985576924e-05, "loss": 1.0028, "step": 4790 }, { "epoch": 1.44, "learning_rate": 4.100811298076923e-05, "loss": 1.1162, "step": 4791 }, { "epoch": 1.44, "learning_rate": 4.100623497596154e-05, "loss": 1.2835, "step": 4792 }, { "epoch": 1.44, "learning_rate": 4.1004356971153845e-05, "loss": 1.1969, "step": 4793 }, { "epoch": 1.44, "learning_rate": 4.1002478966346156e-05, "loss": 1.2668, "step": 4794 }, { "epoch": 1.44, "learning_rate": 4.100060096153847e-05, "loss": 1.3024, "step": 4795 }, { "epoch": 1.44, "learning_rate": 4.099872295673077e-05, "loss": 1.2065, "step": 4796 }, { "epoch": 1.44, "learning_rate": 4.099684495192308e-05, "loss": 1.0402, "step": 4797 }, { "epoch": 1.44, "learning_rate": 4.099496694711539e-05, "loss": 1.205, "step": 4798 }, { "epoch": 1.44, "learning_rate": 4.0993088942307694e-05, "loss": 1.1525, "step": 4799 }, { "epoch": 1.44, "learning_rate": 4.0991210937500005e-05, "loss": 1.1768, "step": 4800 }, { "epoch": 1.44, "learning_rate": 4.098933293269231e-05, "loss": 1.0858, "step": 4801 }, { "epoch": 1.44, "learning_rate": 4.0987454927884615e-05, "loss": 1.0592, "step": 4802 }, { "epoch": 1.44, "learning_rate": 4.0985576923076926e-05, "loss": 1.0972, "step": 4803 }, { "epoch": 1.44, "learning_rate": 4.098369891826923e-05, "loss": 1.1743, "step": 4804 }, { "epoch": 1.44, "learning_rate": 4.0981820913461536e-05, "loss": 1.108, "step": 4805 }, { "epoch": 1.44, "learning_rate": 4.097994290865385e-05, "loss": 1.0916, "step": 4806 }, { "epoch": 1.44, "learning_rate": 4.097806490384616e-05, "loss": 1.2955, "step": 4807 }, { "epoch": 1.44, "learning_rate": 4.0976186899038464e-05, "loss": 1.1352, "step": 4808 }, { "epoch": 1.45, "learning_rate": 4.0974308894230775e-05, "loss": 1.2848, "step": 4809 }, { "epoch": 1.45, "learning_rate": 4.097243088942308e-05, "loss": 0.9631, "step": 4810 }, { "epoch": 1.45, "learning_rate": 4.0970552884615385e-05, "loss": 0.9911, "step": 4811 }, { "epoch": 1.45, "learning_rate": 4.0968674879807696e-05, "loss": 1.1851, "step": 4812 }, { "epoch": 1.45, "learning_rate": 4.0966796875e-05, "loss": 1.0603, "step": 4813 }, { "epoch": 1.45, "learning_rate": 4.0964918870192306e-05, "loss": 1.4199, "step": 4814 }, { "epoch": 1.45, "learning_rate": 4.096304086538462e-05, "loss": 1.0843, "step": 4815 }, { "epoch": 1.45, "learning_rate": 4.096116286057692e-05, "loss": 1.0424, "step": 4816 }, { "epoch": 1.45, "learning_rate": 4.0959284855769234e-05, "loss": 1.0448, "step": 4817 }, { "epoch": 1.45, "learning_rate": 4.095740685096154e-05, "loss": 0.9257, "step": 4818 }, { "epoch": 1.45, "learning_rate": 4.095552884615384e-05, "loss": 1.2029, "step": 4819 }, { "epoch": 1.45, "learning_rate": 4.095365084134616e-05, "loss": 0.9622, "step": 4820 }, { "epoch": 1.45, "learning_rate": 4.0951772836538466e-05, "loss": 1.0126, "step": 4821 }, { "epoch": 1.45, "learning_rate": 4.094989483173077e-05, "loss": 0.9842, "step": 4822 }, { "epoch": 1.45, "learning_rate": 4.094801682692308e-05, "loss": 0.9073, "step": 4823 }, { "epoch": 1.45, "learning_rate": 4.094613882211539e-05, "loss": 1.2978, "step": 4824 }, { "epoch": 1.45, "learning_rate": 4.094426081730769e-05, "loss": 0.8367, "step": 4825 }, { "epoch": 1.45, "learning_rate": 4.0942382812500004e-05, "loss": 1.066, "step": 4826 }, { "epoch": 1.45, "learning_rate": 4.094050480769231e-05, "loss": 0.9328, "step": 4827 }, { "epoch": 1.45, "learning_rate": 4.093862680288461e-05, "loss": 1.0417, "step": 4828 }, { "epoch": 1.45, "learning_rate": 4.0936748798076925e-05, "loss": 1.1844, "step": 4829 }, { "epoch": 1.45, "learning_rate": 4.093487079326923e-05, "loss": 1.0716, "step": 4830 }, { "epoch": 1.45, "learning_rate": 4.093299278846154e-05, "loss": 0.9591, "step": 4831 }, { "epoch": 1.45, "learning_rate": 4.0931114783653846e-05, "loss": 1.1238, "step": 4832 }, { "epoch": 1.45, "learning_rate": 4.092923677884616e-05, "loss": 0.8832, "step": 4833 }, { "epoch": 1.45, "learning_rate": 4.092735877403846e-05, "loss": 1.0872, "step": 4834 }, { "epoch": 1.45, "learning_rate": 4.0925480769230774e-05, "loss": 1.191, "step": 4835 }, { "epoch": 1.45, "learning_rate": 4.092360276442308e-05, "loss": 1.1243, "step": 4836 }, { "epoch": 1.45, "learning_rate": 4.092172475961539e-05, "loss": 1.0468, "step": 4837 }, { "epoch": 1.45, "learning_rate": 4.0919846754807695e-05, "loss": 1.2593, "step": 4838 }, { "epoch": 1.45, "learning_rate": 4.091796875e-05, "loss": 1.0549, "step": 4839 }, { "epoch": 1.45, "learning_rate": 4.091609074519231e-05, "loss": 1.1614, "step": 4840 }, { "epoch": 1.45, "learning_rate": 4.0914212740384616e-05, "loss": 0.9565, "step": 4841 }, { "epoch": 1.45, "learning_rate": 4.091233473557692e-05, "loss": 1.2427, "step": 4842 }, { "epoch": 1.46, "learning_rate": 4.091045673076923e-05, "loss": 1.1628, "step": 4843 }, { "epoch": 1.46, "learning_rate": 4.091045673076923e-05, "loss": 0.9892, "step": 4844 }, { "epoch": 1.46, "learning_rate": 4.090857872596154e-05, "loss": 1.2091, "step": 4845 }, { "epoch": 1.46, "learning_rate": 4.090670072115385e-05, "loss": 1.2555, "step": 4846 }, { "epoch": 1.46, "learning_rate": 4.090482271634616e-05, "loss": 1.2656, "step": 4847 }, { "epoch": 1.46, "learning_rate": 4.0902944711538465e-05, "loss": 0.9232, "step": 4848 }, { "epoch": 1.46, "learning_rate": 4.090106670673077e-05, "loss": 0.897, "step": 4849 }, { "epoch": 1.46, "learning_rate": 4.089918870192308e-05, "loss": 0.9404, "step": 4850 }, { "epoch": 1.46, "learning_rate": 4.0897310697115386e-05, "loss": 0.9918, "step": 4851 }, { "epoch": 1.46, "learning_rate": 4.08954326923077e-05, "loss": 1.0236, "step": 4852 }, { "epoch": 1.46, "learning_rate": 4.08935546875e-05, "loss": 1.1908, "step": 4853 }, { "epoch": 1.46, "learning_rate": 4.089167668269231e-05, "loss": 1.034, "step": 4854 }, { "epoch": 1.46, "learning_rate": 4.088979867788462e-05, "loss": 0.921, "step": 4855 }, { "epoch": 1.46, "learning_rate": 4.0887920673076924e-05, "loss": 1.0684, "step": 4856 }, { "epoch": 1.46, "learning_rate": 4.088604266826923e-05, "loss": 1.4236, "step": 4857 }, { "epoch": 1.46, "learning_rate": 4.088416466346154e-05, "loss": 0.9626, "step": 4858 }, { "epoch": 1.46, "learning_rate": 4.0882286658653845e-05, "loss": 0.9981, "step": 4859 }, { "epoch": 1.46, "learning_rate": 4.0880408653846156e-05, "loss": 1.1201, "step": 4860 }, { "epoch": 1.46, "learning_rate": 4.087853064903847e-05, "loss": 0.8599, "step": 4861 }, { "epoch": 1.46, "learning_rate": 4.087665264423077e-05, "loss": 1.0118, "step": 4862 }, { "epoch": 1.46, "learning_rate": 4.087477463942308e-05, "loss": 1.213, "step": 4863 }, { "epoch": 1.46, "learning_rate": 4.087289663461539e-05, "loss": 1.4246, "step": 4864 }, { "epoch": 1.46, "learning_rate": 4.0871018629807694e-05, "loss": 1.2042, "step": 4865 }, { "epoch": 1.46, "learning_rate": 4.0869140625000005e-05, "loss": 0.9611, "step": 4866 }, { "epoch": 1.46, "learning_rate": 4.086726262019231e-05, "loss": 0.8963, "step": 4867 }, { "epoch": 1.46, "learning_rate": 4.0865384615384615e-05, "loss": 0.9891, "step": 4868 }, { "epoch": 1.46, "learning_rate": 4.0863506610576926e-05, "loss": 1.1927, "step": 4869 }, { "epoch": 1.46, "learning_rate": 4.086162860576923e-05, "loss": 1.4114, "step": 4870 }, { "epoch": 1.46, "learning_rate": 4.0859750600961536e-05, "loss": 0.9636, "step": 4871 }, { "epoch": 1.46, "learning_rate": 4.085787259615385e-05, "loss": 1.0756, "step": 4872 }, { "epoch": 1.46, "learning_rate": 4.085599459134616e-05, "loss": 1.1426, "step": 4873 }, { "epoch": 1.46, "learning_rate": 4.0854116586538464e-05, "loss": 1.2511, "step": 4874 }, { "epoch": 1.46, "learning_rate": 4.0852238581730775e-05, "loss": 0.8897, "step": 4875 }, { "epoch": 1.47, "learning_rate": 4.085036057692308e-05, "loss": 0.8404, "step": 4876 }, { "epoch": 1.47, "learning_rate": 4.0848482572115385e-05, "loss": 1.249, "step": 4877 }, { "epoch": 1.47, "learning_rate": 4.0846604567307696e-05, "loss": 1.1332, "step": 4878 }, { "epoch": 1.47, "learning_rate": 4.08447265625e-05, "loss": 1.1514, "step": 4879 }, { "epoch": 1.47, "learning_rate": 4.084284855769231e-05, "loss": 1.2879, "step": 4880 }, { "epoch": 1.47, "learning_rate": 4.084097055288462e-05, "loss": 1.2385, "step": 4881 }, { "epoch": 1.47, "learning_rate": 4.083909254807692e-05, "loss": 0.893, "step": 4882 }, { "epoch": 1.47, "learning_rate": 4.0837214543269234e-05, "loss": 1.0829, "step": 4883 }, { "epoch": 1.47, "learning_rate": 4.083533653846154e-05, "loss": 1.1552, "step": 4884 }, { "epoch": 1.47, "learning_rate": 4.083345853365384e-05, "loss": 1.3176, "step": 4885 }, { "epoch": 1.47, "learning_rate": 4.083158052884616e-05, "loss": 1.19, "step": 4886 }, { "epoch": 1.47, "learning_rate": 4.0829702524038466e-05, "loss": 0.8816, "step": 4887 }, { "epoch": 1.47, "learning_rate": 4.082782451923077e-05, "loss": 1.0105, "step": 4888 }, { "epoch": 1.47, "learning_rate": 4.082594651442308e-05, "loss": 1.2854, "step": 4889 }, { "epoch": 1.47, "learning_rate": 4.082406850961539e-05, "loss": 0.8746, "step": 4890 }, { "epoch": 1.47, "learning_rate": 4.082219050480769e-05, "loss": 1.2027, "step": 4891 }, { "epoch": 1.47, "learning_rate": 4.0820312500000004e-05, "loss": 1.3479, "step": 4892 }, { "epoch": 1.47, "learning_rate": 4.081843449519231e-05, "loss": 1.1361, "step": 4893 }, { "epoch": 1.47, "learning_rate": 4.081655649038461e-05, "loss": 1.4124, "step": 4894 }, { "epoch": 1.47, "learning_rate": 4.0814678485576925e-05, "loss": 1.0417, "step": 4895 }, { "epoch": 1.47, "learning_rate": 4.081280048076923e-05, "loss": 1.1509, "step": 4896 }, { "epoch": 1.47, "learning_rate": 4.081092247596154e-05, "loss": 1.2149, "step": 4897 }, { "epoch": 1.47, "learning_rate": 4.0809044471153846e-05, "loss": 1.2478, "step": 4898 }, { "epoch": 1.47, "learning_rate": 4.080716646634616e-05, "loss": 1.1055, "step": 4899 }, { "epoch": 1.47, "learning_rate": 4.080528846153847e-05, "loss": 1.001, "step": 4900 }, { "epoch": 1.47, "learning_rate": 4.0803410456730774e-05, "loss": 1.1566, "step": 4901 }, { "epoch": 1.47, "learning_rate": 4.080153245192308e-05, "loss": 1.0755, "step": 4902 }, { "epoch": 1.47, "learning_rate": 4.079965444711539e-05, "loss": 1.1104, "step": 4903 }, { "epoch": 1.47, "learning_rate": 4.0797776442307695e-05, "loss": 0.9823, "step": 4904 }, { "epoch": 1.47, "learning_rate": 4.07958984375e-05, "loss": 1.0676, "step": 4905 }, { "epoch": 1.47, "learning_rate": 4.079402043269231e-05, "loss": 1.0828, "step": 4906 }, { "epoch": 1.47, "learning_rate": 4.0792142427884616e-05, "loss": 1.1928, "step": 4907 }, { "epoch": 1.47, "learning_rate": 4.079026442307692e-05, "loss": 1.0003, "step": 4908 }, { "epoch": 1.48, "learning_rate": 4.078838641826923e-05, "loss": 1.1041, "step": 4909 }, { "epoch": 1.48, "learning_rate": 4.078650841346154e-05, "loss": 0.9613, "step": 4910 }, { "epoch": 1.48, "learning_rate": 4.078463040865385e-05, "loss": 1.022, "step": 4911 }, { "epoch": 1.48, "learning_rate": 4.078275240384616e-05, "loss": 1.1076, "step": 4912 }, { "epoch": 1.48, "learning_rate": 4.0780874399038465e-05, "loss": 1.145, "step": 4913 }, { "epoch": 1.48, "learning_rate": 4.077899639423077e-05, "loss": 1.0787, "step": 4914 }, { "epoch": 1.48, "learning_rate": 4.077711838942308e-05, "loss": 1.1841, "step": 4915 }, { "epoch": 1.48, "learning_rate": 4.0775240384615386e-05, "loss": 1.3059, "step": 4916 }, { "epoch": 1.48, "learning_rate": 4.07733623798077e-05, "loss": 1.156, "step": 4917 }, { "epoch": 1.48, "learning_rate": 4.0771484375e-05, "loss": 1.1158, "step": 4918 }, { "epoch": 1.48, "learning_rate": 4.076960637019231e-05, "loss": 1.0152, "step": 4919 }, { "epoch": 1.48, "learning_rate": 4.076772836538462e-05, "loss": 0.9588, "step": 4920 }, { "epoch": 1.48, "learning_rate": 4.0765850360576923e-05, "loss": 1.0999, "step": 4921 }, { "epoch": 1.48, "learning_rate": 4.076397235576923e-05, "loss": 1.1615, "step": 4922 }, { "epoch": 1.48, "learning_rate": 4.076209435096154e-05, "loss": 0.9228, "step": 4923 }, { "epoch": 1.48, "learning_rate": 4.0760216346153845e-05, "loss": 1.1483, "step": 4924 }, { "epoch": 1.48, "learning_rate": 4.0758338341346156e-05, "loss": 0.9184, "step": 4925 }, { "epoch": 1.48, "learning_rate": 4.075646033653847e-05, "loss": 1.1837, "step": 4926 }, { "epoch": 1.48, "learning_rate": 4.075458233173077e-05, "loss": 1.0589, "step": 4927 }, { "epoch": 1.48, "learning_rate": 4.075270432692308e-05, "loss": 1.0038, "step": 4928 }, { "epoch": 1.48, "learning_rate": 4.075082632211539e-05, "loss": 0.9143, "step": 4929 }, { "epoch": 1.48, "learning_rate": 4.0748948317307693e-05, "loss": 1.1163, "step": 4930 }, { "epoch": 1.48, "learning_rate": 4.0747070312500005e-05, "loss": 0.8945, "step": 4931 }, { "epoch": 1.48, "learning_rate": 4.074519230769231e-05, "loss": 1.126, "step": 4932 }, { "epoch": 1.48, "learning_rate": 4.0743314302884615e-05, "loss": 1.3152, "step": 4933 }, { "epoch": 1.48, "learning_rate": 4.0741436298076926e-05, "loss": 1.0686, "step": 4934 }, { "epoch": 1.48, "learning_rate": 4.073955829326923e-05, "loss": 1.0413, "step": 4935 }, { "epoch": 1.48, "learning_rate": 4.0737680288461536e-05, "loss": 1.0594, "step": 4936 }, { "epoch": 1.48, "learning_rate": 4.073580228365385e-05, "loss": 1.2133, "step": 4937 }, { "epoch": 1.48, "learning_rate": 4.073392427884616e-05, "loss": 0.8426, "step": 4938 }, { "epoch": 1.48, "learning_rate": 4.0732046274038464e-05, "loss": 1.2328, "step": 4939 }, { "epoch": 1.48, "learning_rate": 4.0730168269230775e-05, "loss": 1.321, "step": 4940 }, { "epoch": 1.48, "learning_rate": 4.072829026442308e-05, "loss": 0.8657, "step": 4941 }, { "epoch": 1.48, "learning_rate": 4.0726412259615385e-05, "loss": 1.1444, "step": 4942 }, { "epoch": 1.49, "learning_rate": 4.0724534254807696e-05, "loss": 1.0812, "step": 4943 }, { "epoch": 1.49, "learning_rate": 4.072265625e-05, "loss": 0.9577, "step": 4944 }, { "epoch": 1.49, "learning_rate": 4.072077824519231e-05, "loss": 1.1618, "step": 4945 }, { "epoch": 1.49, "learning_rate": 4.071890024038462e-05, "loss": 1.233, "step": 4946 }, { "epoch": 1.49, "learning_rate": 4.071702223557692e-05, "loss": 1.2712, "step": 4947 }, { "epoch": 1.49, "learning_rate": 4.0715144230769234e-05, "loss": 1.1093, "step": 4948 }, { "epoch": 1.49, "learning_rate": 4.071326622596154e-05, "loss": 1.0964, "step": 4949 }, { "epoch": 1.49, "learning_rate": 4.071138822115384e-05, "loss": 1.2375, "step": 4950 }, { "epoch": 1.49, "learning_rate": 4.070951021634616e-05, "loss": 1.0589, "step": 4951 }, { "epoch": 1.49, "learning_rate": 4.0707632211538466e-05, "loss": 0.9767, "step": 4952 }, { "epoch": 1.49, "learning_rate": 4.070575420673077e-05, "loss": 1.1064, "step": 4953 }, { "epoch": 1.49, "learning_rate": 4.070387620192308e-05, "loss": 0.9223, "step": 4954 }, { "epoch": 1.49, "learning_rate": 4.070199819711539e-05, "loss": 1.1268, "step": 4955 }, { "epoch": 1.49, "learning_rate": 4.070012019230769e-05, "loss": 1.0324, "step": 4956 }, { "epoch": 1.49, "learning_rate": 4.0698242187500004e-05, "loss": 1.3783, "step": 4957 }, { "epoch": 1.49, "learning_rate": 4.069636418269231e-05, "loss": 1.2933, "step": 4958 }, { "epoch": 1.49, "learning_rate": 4.069448617788461e-05, "loss": 1.0249, "step": 4959 }, { "epoch": 1.49, "learning_rate": 4.0692608173076925e-05, "loss": 1.1978, "step": 4960 }, { "epoch": 1.49, "learning_rate": 4.069073016826923e-05, "loss": 1.2523, "step": 4961 }, { "epoch": 1.49, "learning_rate": 4.068885216346154e-05, "loss": 1.2154, "step": 4962 }, { "epoch": 1.49, "learning_rate": 4.0686974158653846e-05, "loss": 0.9876, "step": 4963 }, { "epoch": 1.49, "learning_rate": 4.068509615384616e-05, "loss": 1.1529, "step": 4964 }, { "epoch": 1.49, "learning_rate": 4.068321814903847e-05, "loss": 1.0702, "step": 4965 }, { "epoch": 1.49, "learning_rate": 4.0681340144230774e-05, "loss": 0.9557, "step": 4966 }, { "epoch": 1.49, "learning_rate": 4.067946213942308e-05, "loss": 1.1228, "step": 4967 }, { "epoch": 1.49, "learning_rate": 4.067758413461539e-05, "loss": 1.2462, "step": 4968 }, { "epoch": 1.49, "learning_rate": 4.0675706129807695e-05, "loss": 1.0954, "step": 4969 }, { "epoch": 1.49, "learning_rate": 4.0673828125e-05, "loss": 0.9626, "step": 4970 }, { "epoch": 1.49, "learning_rate": 4.067195012019231e-05, "loss": 1.0553, "step": 4971 }, { "epoch": 1.49, "learning_rate": 4.0670072115384616e-05, "loss": 1.3093, "step": 4972 }, { "epoch": 1.49, "learning_rate": 4.066819411057692e-05, "loss": 0.9662, "step": 4973 }, { "epoch": 1.49, "learning_rate": 4.066631610576923e-05, "loss": 1.1441, "step": 4974 }, { "epoch": 1.49, "learning_rate": 4.066443810096154e-05, "loss": 0.9993, "step": 4975 }, { "epoch": 1.5, "learning_rate": 4.066256009615385e-05, "loss": 0.9627, "step": 4976 }, { "epoch": 1.5, "learning_rate": 4.066068209134616e-05, "loss": 0.8306, "step": 4977 }, { "epoch": 1.5, "learning_rate": 4.0658804086538465e-05, "loss": 0.9946, "step": 4978 }, { "epoch": 1.5, "learning_rate": 4.065692608173077e-05, "loss": 1.1571, "step": 4979 }, { "epoch": 1.5, "learning_rate": 4.065504807692308e-05, "loss": 1.1827, "step": 4980 }, { "epoch": 1.5, "learning_rate": 4.0653170072115386e-05, "loss": 1.0074, "step": 4981 }, { "epoch": 1.5, "learning_rate": 4.06512920673077e-05, "loss": 1.2077, "step": 4982 }, { "epoch": 1.5, "learning_rate": 4.06494140625e-05, "loss": 1.1554, "step": 4983 }, { "epoch": 1.5, "learning_rate": 4.064753605769231e-05, "loss": 0.8889, "step": 4984 }, { "epoch": 1.5, "learning_rate": 4.064565805288462e-05, "loss": 1.1369, "step": 4985 }, { "epoch": 1.5, "learning_rate": 4.064378004807692e-05, "loss": 1.2126, "step": 4986 }, { "epoch": 1.5, "learning_rate": 4.064190204326923e-05, "loss": 1.4075, "step": 4987 }, { "epoch": 1.5, "learning_rate": 4.064002403846154e-05, "loss": 1.1885, "step": 4988 }, { "epoch": 1.5, "learning_rate": 4.0638146033653844e-05, "loss": 1.0701, "step": 4989 }, { "epoch": 1.5, "learning_rate": 4.0636268028846156e-05, "loss": 0.9638, "step": 4990 }, { "epoch": 1.5, "learning_rate": 4.063439002403847e-05, "loss": 0.8701, "step": 4991 }, { "epoch": 1.5, "learning_rate": 4.063251201923077e-05, "loss": 1.0529, "step": 4992 }, { "epoch": 1.5, "learning_rate": 4.063063401442308e-05, "loss": 0.9161, "step": 4993 }, { "epoch": 1.5, "learning_rate": 4.062875600961539e-05, "loss": 1.1657, "step": 4994 }, { "epoch": 1.5, "learning_rate": 4.062687800480769e-05, "loss": 1.1026, "step": 4995 }, { "epoch": 1.5, "learning_rate": 4.0625000000000005e-05, "loss": 1.0964, "step": 4996 }, { "epoch": 1.5, "learning_rate": 4.062312199519231e-05, "loss": 1.2254, "step": 4997 }, { "epoch": 1.5, "learning_rate": 4.0621243990384614e-05, "loss": 1.1787, "step": 4998 }, { "epoch": 1.5, "learning_rate": 4.0619365985576926e-05, "loss": 1.3357, "step": 4999 }, { "epoch": 1.5, "learning_rate": 4.061748798076923e-05, "loss": 1.239, "step": 5000 }, { "epoch": 1.5, "learning_rate": 4.0615609975961536e-05, "loss": 0.9986, "step": 5001 }, { "epoch": 1.5, "learning_rate": 4.061373197115385e-05, "loss": 1.2542, "step": 5002 }, { "epoch": 1.5, "learning_rate": 4.061185396634616e-05, "loss": 1.0455, "step": 5003 }, { "epoch": 1.5, "learning_rate": 4.0609975961538463e-05, "loss": 1.1933, "step": 5004 }, { "epoch": 1.5, "learning_rate": 4.0608097956730775e-05, "loss": 1.092, "step": 5005 }, { "epoch": 1.5, "learning_rate": 4.060621995192308e-05, "loss": 1.2374, "step": 5006 }, { "epoch": 1.5, "learning_rate": 4.0604341947115385e-05, "loss": 1.0241, "step": 5007 }, { "epoch": 1.5, "learning_rate": 4.0602463942307696e-05, "loss": 1.0002, "step": 5008 }, { "epoch": 1.51, "learning_rate": 4.06005859375e-05, "loss": 1.211, "step": 5009 }, { "epoch": 1.51, "learning_rate": 4.059870793269231e-05, "loss": 1.0501, "step": 5010 }, { "epoch": 1.51, "learning_rate": 4.059682992788462e-05, "loss": 1.0844, "step": 5011 }, { "epoch": 1.51, "learning_rate": 4.059495192307692e-05, "loss": 1.0439, "step": 5012 }, { "epoch": 1.51, "learning_rate": 4.0593073918269233e-05, "loss": 1.1007, "step": 5013 }, { "epoch": 1.51, "learning_rate": 4.059119591346154e-05, "loss": 0.9853, "step": 5014 }, { "epoch": 1.51, "learning_rate": 4.058931790865384e-05, "loss": 1.0219, "step": 5015 }, { "epoch": 1.51, "learning_rate": 4.058743990384616e-05, "loss": 0.9171, "step": 5016 }, { "epoch": 1.51, "learning_rate": 4.0585561899038466e-05, "loss": 1.0506, "step": 5017 }, { "epoch": 1.51, "learning_rate": 4.058368389423077e-05, "loss": 1.1577, "step": 5018 }, { "epoch": 1.51, "learning_rate": 4.058180588942308e-05, "loss": 1.0874, "step": 5019 }, { "epoch": 1.51, "learning_rate": 4.057992788461539e-05, "loss": 1.0945, "step": 5020 }, { "epoch": 1.51, "learning_rate": 4.057804987980769e-05, "loss": 1.1035, "step": 5021 }, { "epoch": 1.51, "learning_rate": 4.0576171875000004e-05, "loss": 1.0403, "step": 5022 }, { "epoch": 1.51, "learning_rate": 4.057429387019231e-05, "loss": 1.1275, "step": 5023 }, { "epoch": 1.51, "learning_rate": 4.057241586538462e-05, "loss": 1.0469, "step": 5024 }, { "epoch": 1.51, "learning_rate": 4.0570537860576925e-05, "loss": 0.8853, "step": 5025 }, { "epoch": 1.51, "learning_rate": 4.056865985576923e-05, "loss": 0.9902, "step": 5026 }, { "epoch": 1.51, "learning_rate": 4.056678185096154e-05, "loss": 1.3133, "step": 5027 }, { "epoch": 1.51, "learning_rate": 4.0564903846153846e-05, "loss": 1.1563, "step": 5028 }, { "epoch": 1.51, "learning_rate": 4.056302584134616e-05, "loss": 1.1546, "step": 5029 }, { "epoch": 1.51, "learning_rate": 4.056114783653847e-05, "loss": 1.0794, "step": 5030 }, { "epoch": 1.51, "learning_rate": 4.0559269831730774e-05, "loss": 1.0454, "step": 5031 }, { "epoch": 1.51, "learning_rate": 4.055739182692308e-05, "loss": 1.1239, "step": 5032 }, { "epoch": 1.51, "learning_rate": 4.055551382211539e-05, "loss": 1.1062, "step": 5033 }, { "epoch": 1.51, "learning_rate": 4.0553635817307695e-05, "loss": 1.1423, "step": 5034 }, { "epoch": 1.51, "learning_rate": 4.05517578125e-05, "loss": 1.2495, "step": 5035 }, { "epoch": 1.51, "learning_rate": 4.054987980769231e-05, "loss": 1.3191, "step": 5036 }, { "epoch": 1.51, "learning_rate": 4.0548001802884616e-05, "loss": 1.1708, "step": 5037 }, { "epoch": 1.51, "learning_rate": 4.054612379807692e-05, "loss": 1.3047, "step": 5038 }, { "epoch": 1.51, "learning_rate": 4.054424579326923e-05, "loss": 1.2206, "step": 5039 }, { "epoch": 1.51, "learning_rate": 4.054236778846154e-05, "loss": 1.1538, "step": 5040 }, { "epoch": 1.51, "learning_rate": 4.054048978365385e-05, "loss": 1.3445, "step": 5041 }, { "epoch": 1.52, "learning_rate": 4.053861177884616e-05, "loss": 1.1013, "step": 5042 }, { "epoch": 1.52, "learning_rate": 4.0536733774038465e-05, "loss": 0.9678, "step": 5043 }, { "epoch": 1.52, "learning_rate": 4.0534855769230776e-05, "loss": 1.0677, "step": 5044 }, { "epoch": 1.52, "learning_rate": 4.053297776442308e-05, "loss": 1.0792, "step": 5045 }, { "epoch": 1.52, "learning_rate": 4.0531099759615386e-05, "loss": 1.1721, "step": 5046 }, { "epoch": 1.52, "learning_rate": 4.05292217548077e-05, "loss": 1.1692, "step": 5047 }, { "epoch": 1.52, "learning_rate": 4.052734375e-05, "loss": 1.2945, "step": 5048 }, { "epoch": 1.52, "learning_rate": 4.052546574519231e-05, "loss": 1.2032, "step": 5049 }, { "epoch": 1.52, "learning_rate": 4.052358774038462e-05, "loss": 1.1327, "step": 5050 }, { "epoch": 1.52, "learning_rate": 4.052170973557692e-05, "loss": 1.3149, "step": 5051 }, { "epoch": 1.52, "learning_rate": 4.051983173076923e-05, "loss": 1.0559, "step": 5052 }, { "epoch": 1.52, "learning_rate": 4.051795372596154e-05, "loss": 1.1864, "step": 5053 }, { "epoch": 1.52, "learning_rate": 4.0516075721153844e-05, "loss": 1.2215, "step": 5054 }, { "epoch": 1.52, "learning_rate": 4.0514197716346156e-05, "loss": 1.0194, "step": 5055 }, { "epoch": 1.52, "learning_rate": 4.051231971153847e-05, "loss": 1.1226, "step": 5056 }, { "epoch": 1.52, "learning_rate": 4.051044170673077e-05, "loss": 0.9506, "step": 5057 }, { "epoch": 1.52, "learning_rate": 4.050856370192308e-05, "loss": 1.1879, "step": 5058 }, { "epoch": 1.52, "learning_rate": 4.050668569711539e-05, "loss": 1.278, "step": 5059 }, { "epoch": 1.52, "learning_rate": 4.050480769230769e-05, "loss": 1.0392, "step": 5060 }, { "epoch": 1.52, "learning_rate": 4.0502929687500005e-05, "loss": 1.0973, "step": 5061 }, { "epoch": 1.52, "learning_rate": 4.050105168269231e-05, "loss": 1.0578, "step": 5062 }, { "epoch": 1.52, "learning_rate": 4.0499173677884614e-05, "loss": 1.0632, "step": 5063 }, { "epoch": 1.52, "learning_rate": 4.0497295673076926e-05, "loss": 1.2317, "step": 5064 }, { "epoch": 1.52, "learning_rate": 4.049541766826923e-05, "loss": 1.1738, "step": 5065 }, { "epoch": 1.52, "learning_rate": 4.0493539663461535e-05, "loss": 1.1535, "step": 5066 }, { "epoch": 1.52, "learning_rate": 4.049166165865385e-05, "loss": 1.4373, "step": 5067 }, { "epoch": 1.52, "learning_rate": 4.048978365384616e-05, "loss": 1.0411, "step": 5068 }, { "epoch": 1.52, "learning_rate": 4.048790564903846e-05, "loss": 1.1404, "step": 5069 }, { "epoch": 1.52, "learning_rate": 4.0486027644230775e-05, "loss": 1.139, "step": 5070 }, { "epoch": 1.52, "learning_rate": 4.048414963942308e-05, "loss": 1.0802, "step": 5071 }, { "epoch": 1.52, "learning_rate": 4.0482271634615384e-05, "loss": 0.8938, "step": 5072 }, { "epoch": 1.52, "learning_rate": 4.0480393629807696e-05, "loss": 1.4431, "step": 5073 }, { "epoch": 1.52, "learning_rate": 4.0478515625e-05, "loss": 1.2679, "step": 5074 }, { "epoch": 1.52, "learning_rate": 4.047663762019231e-05, "loss": 1.0671, "step": 5075 }, { "epoch": 1.53, "learning_rate": 4.047475961538462e-05, "loss": 1.071, "step": 5076 }, { "epoch": 1.53, "learning_rate": 4.047288161057692e-05, "loss": 1.3221, "step": 5077 }, { "epoch": 1.53, "learning_rate": 4.047100360576923e-05, "loss": 1.1157, "step": 5078 }, { "epoch": 1.53, "learning_rate": 4.046912560096154e-05, "loss": 1.1152, "step": 5079 }, { "epoch": 1.53, "learning_rate": 4.046724759615384e-05, "loss": 1.0561, "step": 5080 }, { "epoch": 1.53, "learning_rate": 4.046536959134616e-05, "loss": 1.124, "step": 5081 }, { "epoch": 1.53, "learning_rate": 4.0463491586538466e-05, "loss": 1.2677, "step": 5082 }, { "epoch": 1.53, "learning_rate": 4.046161358173077e-05, "loss": 1.1897, "step": 5083 }, { "epoch": 1.53, "learning_rate": 4.045973557692308e-05, "loss": 1.0035, "step": 5084 }, { "epoch": 1.53, "learning_rate": 4.045785757211539e-05, "loss": 1.2548, "step": 5085 }, { "epoch": 1.53, "learning_rate": 4.045597956730769e-05, "loss": 1.358, "step": 5086 }, { "epoch": 1.53, "learning_rate": 4.0454101562500003e-05, "loss": 1.1513, "step": 5087 }, { "epoch": 1.53, "learning_rate": 4.045222355769231e-05, "loss": 1.05, "step": 5088 }, { "epoch": 1.53, "learning_rate": 4.045034555288462e-05, "loss": 0.9241, "step": 5089 }, { "epoch": 1.53, "learning_rate": 4.0448467548076924e-05, "loss": 1.0443, "step": 5090 }, { "epoch": 1.53, "learning_rate": 4.044658954326923e-05, "loss": 1.1588, "step": 5091 }, { "epoch": 1.53, "learning_rate": 4.044471153846154e-05, "loss": 1.3121, "step": 5092 }, { "epoch": 1.53, "learning_rate": 4.0442833533653846e-05, "loss": 1.088, "step": 5093 }, { "epoch": 1.53, "learning_rate": 4.044095552884616e-05, "loss": 1.2211, "step": 5094 }, { "epoch": 1.53, "learning_rate": 4.043907752403847e-05, "loss": 1.0586, "step": 5095 }, { "epoch": 1.53, "learning_rate": 4.0437199519230773e-05, "loss": 1.1787, "step": 5096 }, { "epoch": 1.53, "learning_rate": 4.043532151442308e-05, "loss": 1.0923, "step": 5097 }, { "epoch": 1.53, "learning_rate": 4.043344350961539e-05, "loss": 1.0013, "step": 5098 }, { "epoch": 1.53, "learning_rate": 4.0431565504807695e-05, "loss": 1.0948, "step": 5099 }, { "epoch": 1.53, "learning_rate": 4.04296875e-05, "loss": 0.9855, "step": 5100 }, { "epoch": 1.53, "learning_rate": 4.042780949519231e-05, "loss": 1.0528, "step": 5101 }, { "epoch": 1.53, "learning_rate": 4.0425931490384616e-05, "loss": 1.0086, "step": 5102 }, { "epoch": 1.53, "learning_rate": 4.042405348557692e-05, "loss": 1.0591, "step": 5103 }, { "epoch": 1.53, "learning_rate": 4.042217548076923e-05, "loss": 0.857, "step": 5104 }, { "epoch": 1.53, "learning_rate": 4.042029747596154e-05, "loss": 0.9303, "step": 5105 }, { "epoch": 1.53, "learning_rate": 4.041841947115385e-05, "loss": 1.0141, "step": 5106 }, { "epoch": 1.53, "learning_rate": 4.041654146634616e-05, "loss": 1.1444, "step": 5107 }, { "epoch": 1.53, "learning_rate": 4.0414663461538465e-05, "loss": 1.266, "step": 5108 }, { "epoch": 1.54, "learning_rate": 4.0412785456730776e-05, "loss": 1.0709, "step": 5109 }, { "epoch": 1.54, "learning_rate": 4.041090745192308e-05, "loss": 1.0591, "step": 5110 }, { "epoch": 1.54, "learning_rate": 4.0409029447115386e-05, "loss": 1.0799, "step": 5111 }, { "epoch": 1.54, "learning_rate": 4.04071514423077e-05, "loss": 1.2175, "step": 5112 }, { "epoch": 1.54, "learning_rate": 4.04052734375e-05, "loss": 1.0578, "step": 5113 }, { "epoch": 1.54, "learning_rate": 4.040339543269231e-05, "loss": 1.4954, "step": 5114 }, { "epoch": 1.54, "learning_rate": 4.040151742788462e-05, "loss": 1.1439, "step": 5115 }, { "epoch": 1.54, "learning_rate": 4.039963942307692e-05, "loss": 1.0862, "step": 5116 }, { "epoch": 1.54, "learning_rate": 4.039776141826923e-05, "loss": 1.3249, "step": 5117 }, { "epoch": 1.54, "learning_rate": 4.039588341346154e-05, "loss": 1.0191, "step": 5118 }, { "epoch": 1.54, "learning_rate": 4.0394005408653844e-05, "loss": 1.3938, "step": 5119 }, { "epoch": 1.54, "learning_rate": 4.0392127403846156e-05, "loss": 1.0335, "step": 5120 }, { "epoch": 1.54, "learning_rate": 4.039024939903847e-05, "loss": 1.168, "step": 5121 }, { "epoch": 1.54, "learning_rate": 4.038837139423077e-05, "loss": 1.1786, "step": 5122 }, { "epoch": 1.54, "learning_rate": 4.038649338942308e-05, "loss": 0.9497, "step": 5123 }, { "epoch": 1.54, "learning_rate": 4.038461538461539e-05, "loss": 1.066, "step": 5124 }, { "epoch": 1.54, "learning_rate": 4.038273737980769e-05, "loss": 1.3588, "step": 5125 }, { "epoch": 1.54, "learning_rate": 4.0380859375000005e-05, "loss": 1.0068, "step": 5126 }, { "epoch": 1.54, "learning_rate": 4.037898137019231e-05, "loss": 1.1194, "step": 5127 }, { "epoch": 1.54, "learning_rate": 4.0377103365384614e-05, "loss": 1.0557, "step": 5128 }, { "epoch": 1.54, "learning_rate": 4.0375225360576926e-05, "loss": 1.0495, "step": 5129 }, { "epoch": 1.54, "learning_rate": 4.037334735576923e-05, "loss": 1.2734, "step": 5130 }, { "epoch": 1.54, "learning_rate": 4.0371469350961535e-05, "loss": 0.8879, "step": 5131 }, { "epoch": 1.54, "learning_rate": 4.036959134615385e-05, "loss": 1.0978, "step": 5132 }, { "epoch": 1.54, "learning_rate": 4.036771334134616e-05, "loss": 1.3075, "step": 5133 }, { "epoch": 1.54, "learning_rate": 4.036583533653846e-05, "loss": 1.1512, "step": 5134 }, { "epoch": 1.54, "learning_rate": 4.0363957331730775e-05, "loss": 1.1247, "step": 5135 }, { "epoch": 1.54, "learning_rate": 4.036207932692308e-05, "loss": 1.0169, "step": 5136 }, { "epoch": 1.54, "learning_rate": 4.0360201322115384e-05, "loss": 1.2289, "step": 5137 }, { "epoch": 1.54, "learning_rate": 4.0358323317307696e-05, "loss": 1.3159, "step": 5138 }, { "epoch": 1.54, "learning_rate": 4.03564453125e-05, "loss": 0.9555, "step": 5139 }, { "epoch": 1.54, "learning_rate": 4.035456730769231e-05, "loss": 0.894, "step": 5140 }, { "epoch": 1.54, "learning_rate": 4.035268930288462e-05, "loss": 1.1163, "step": 5141 }, { "epoch": 1.55, "learning_rate": 4.035081129807692e-05, "loss": 1.3864, "step": 5142 }, { "epoch": 1.55, "learning_rate": 4.034893329326923e-05, "loss": 1.1177, "step": 5143 }, { "epoch": 1.55, "learning_rate": 4.034705528846154e-05, "loss": 1.2817, "step": 5144 }, { "epoch": 1.55, "learning_rate": 4.034517728365384e-05, "loss": 0.9276, "step": 5145 }, { "epoch": 1.55, "learning_rate": 4.034329927884616e-05, "loss": 1.188, "step": 5146 }, { "epoch": 1.55, "learning_rate": 4.0341421274038466e-05, "loss": 0.9691, "step": 5147 }, { "epoch": 1.55, "learning_rate": 4.033954326923077e-05, "loss": 1.0747, "step": 5148 }, { "epoch": 1.55, "learning_rate": 4.033766526442308e-05, "loss": 0.8487, "step": 5149 }, { "epoch": 1.55, "learning_rate": 4.033578725961539e-05, "loss": 1.0581, "step": 5150 }, { "epoch": 1.55, "learning_rate": 4.033390925480769e-05, "loss": 1.2297, "step": 5151 }, { "epoch": 1.55, "learning_rate": 4.033203125e-05, "loss": 0.9269, "step": 5152 }, { "epoch": 1.55, "learning_rate": 4.033015324519231e-05, "loss": 1.1248, "step": 5153 }, { "epoch": 1.55, "learning_rate": 4.032827524038462e-05, "loss": 1.0896, "step": 5154 }, { "epoch": 1.55, "learning_rate": 4.0326397235576924e-05, "loss": 1.0022, "step": 5155 }, { "epoch": 1.55, "learning_rate": 4.032451923076923e-05, "loss": 1.0377, "step": 5156 }, { "epoch": 1.55, "learning_rate": 4.032264122596154e-05, "loss": 0.9665, "step": 5157 }, { "epoch": 1.55, "learning_rate": 4.0320763221153845e-05, "loss": 1.1468, "step": 5158 }, { "epoch": 1.55, "learning_rate": 4.031888521634616e-05, "loss": 0.9279, "step": 5159 }, { "epoch": 1.55, "learning_rate": 4.031700721153847e-05, "loss": 1.2224, "step": 5160 }, { "epoch": 1.55, "learning_rate": 4.031512920673077e-05, "loss": 1.1989, "step": 5161 }, { "epoch": 1.55, "learning_rate": 4.031325120192308e-05, "loss": 1.1782, "step": 5162 }, { "epoch": 1.55, "learning_rate": 4.031137319711539e-05, "loss": 1.2702, "step": 5163 }, { "epoch": 1.55, "learning_rate": 4.0309495192307694e-05, "loss": 1.2849, "step": 5164 }, { "epoch": 1.55, "learning_rate": 4.03076171875e-05, "loss": 1.1316, "step": 5165 }, { "epoch": 1.55, "learning_rate": 4.030573918269231e-05, "loss": 1.1938, "step": 5166 }, { "epoch": 1.55, "learning_rate": 4.0303861177884616e-05, "loss": 1.0414, "step": 5167 }, { "epoch": 1.55, "learning_rate": 4.030198317307693e-05, "loss": 1.0825, "step": 5168 }, { "epoch": 1.55, "learning_rate": 4.030010516826923e-05, "loss": 1.1937, "step": 5169 }, { "epoch": 1.55, "learning_rate": 4.029822716346154e-05, "loss": 0.9207, "step": 5170 }, { "epoch": 1.55, "learning_rate": 4.029634915865385e-05, "loss": 1.2975, "step": 5171 }, { "epoch": 1.55, "learning_rate": 4.029447115384616e-05, "loss": 1.4062, "step": 5172 }, { "epoch": 1.55, "learning_rate": 4.0292593149038464e-05, "loss": 1.2504, "step": 5173 }, { "epoch": 1.55, "learning_rate": 4.0290715144230776e-05, "loss": 1.2156, "step": 5174 }, { "epoch": 1.55, "learning_rate": 4.028883713942308e-05, "loss": 1.2526, "step": 5175 }, { "epoch": 1.56, "learning_rate": 4.0286959134615386e-05, "loss": 1.0216, "step": 5176 }, { "epoch": 1.56, "learning_rate": 4.02850811298077e-05, "loss": 1.0359, "step": 5177 }, { "epoch": 1.56, "learning_rate": 4.0283203125e-05, "loss": 1.1672, "step": 5178 }, { "epoch": 1.56, "learning_rate": 4.028132512019231e-05, "loss": 1.0755, "step": 5179 }, { "epoch": 1.56, "learning_rate": 4.027944711538462e-05, "loss": 1.241, "step": 5180 }, { "epoch": 1.56, "learning_rate": 4.027756911057692e-05, "loss": 1.1562, "step": 5181 }, { "epoch": 1.56, "learning_rate": 4.027569110576923e-05, "loss": 1.2441, "step": 5182 }, { "epoch": 1.56, "learning_rate": 4.027381310096154e-05, "loss": 1.1547, "step": 5183 }, { "epoch": 1.56, "learning_rate": 4.0271935096153844e-05, "loss": 1.151, "step": 5184 }, { "epoch": 1.56, "learning_rate": 4.0270057091346156e-05, "loss": 1.223, "step": 5185 }, { "epoch": 1.56, "learning_rate": 4.026817908653847e-05, "loss": 1.274, "step": 5186 }, { "epoch": 1.56, "learning_rate": 4.026630108173077e-05, "loss": 1.0355, "step": 5187 }, { "epoch": 1.56, "learning_rate": 4.0264423076923083e-05, "loss": 1.0752, "step": 5188 }, { "epoch": 1.56, "learning_rate": 4.026254507211539e-05, "loss": 1.0271, "step": 5189 }, { "epoch": 1.56, "learning_rate": 4.026066706730769e-05, "loss": 1.1716, "step": 5190 }, { "epoch": 1.56, "learning_rate": 4.0258789062500005e-05, "loss": 1.0812, "step": 5191 }, { "epoch": 1.56, "learning_rate": 4.025691105769231e-05, "loss": 1.0889, "step": 5192 }, { "epoch": 1.56, "learning_rate": 4.0255033052884614e-05, "loss": 1.2644, "step": 5193 }, { "epoch": 1.56, "learning_rate": 4.0253155048076926e-05, "loss": 1.0235, "step": 5194 }, { "epoch": 1.56, "learning_rate": 4.025127704326923e-05, "loss": 0.9418, "step": 5195 }, { "epoch": 1.56, "learning_rate": 4.0249399038461535e-05, "loss": 1.0761, "step": 5196 }, { "epoch": 1.56, "learning_rate": 4.024752103365385e-05, "loss": 1.0656, "step": 5197 }, { "epoch": 1.56, "learning_rate": 4.024564302884616e-05, "loss": 1.2167, "step": 5198 }, { "epoch": 1.56, "learning_rate": 4.024376502403846e-05, "loss": 1.4084, "step": 5199 }, { "epoch": 1.56, "learning_rate": 4.0241887019230775e-05, "loss": 1.113, "step": 5200 }, { "epoch": 1.56, "learning_rate": 4.024000901442308e-05, "loss": 1.0905, "step": 5201 }, { "epoch": 1.56, "learning_rate": 4.0238131009615384e-05, "loss": 1.0797, "step": 5202 }, { "epoch": 1.56, "learning_rate": 4.0236253004807696e-05, "loss": 1.1632, "step": 5203 }, { "epoch": 1.56, "learning_rate": 4.0234375e-05, "loss": 0.9548, "step": 5204 }, { "epoch": 1.56, "learning_rate": 4.023249699519231e-05, "loss": 1.1092, "step": 5205 }, { "epoch": 1.56, "learning_rate": 4.023061899038462e-05, "loss": 1.0911, "step": 5206 }, { "epoch": 1.56, "learning_rate": 4.022874098557692e-05, "loss": 0.8986, "step": 5207 }, { "epoch": 1.56, "learning_rate": 4.022686298076923e-05, "loss": 1.2518, "step": 5208 }, { "epoch": 1.57, "learning_rate": 4.022498497596154e-05, "loss": 1.0838, "step": 5209 }, { "epoch": 1.57, "learning_rate": 4.022310697115384e-05, "loss": 1.188, "step": 5210 }, { "epoch": 1.57, "learning_rate": 4.022122896634616e-05, "loss": 0.9032, "step": 5211 }, { "epoch": 1.57, "learning_rate": 4.0219350961538466e-05, "loss": 0.9976, "step": 5212 }, { "epoch": 1.57, "learning_rate": 4.021747295673077e-05, "loss": 1.0377, "step": 5213 }, { "epoch": 1.57, "learning_rate": 4.021559495192308e-05, "loss": 1.0119, "step": 5214 }, { "epoch": 1.57, "learning_rate": 4.021371694711539e-05, "loss": 1.0854, "step": 5215 }, { "epoch": 1.57, "learning_rate": 4.021183894230769e-05, "loss": 1.3717, "step": 5216 }, { "epoch": 1.57, "learning_rate": 4.02099609375e-05, "loss": 1.3818, "step": 5217 }, { "epoch": 1.57, "learning_rate": 4.020808293269231e-05, "loss": 1.2063, "step": 5218 }, { "epoch": 1.57, "learning_rate": 4.020620492788462e-05, "loss": 1.1706, "step": 5219 }, { "epoch": 1.57, "learning_rate": 4.0204326923076924e-05, "loss": 1.0051, "step": 5220 }, { "epoch": 1.57, "learning_rate": 4.020244891826923e-05, "loss": 1.0426, "step": 5221 }, { "epoch": 1.57, "learning_rate": 4.020057091346154e-05, "loss": 1.2098, "step": 5222 }, { "epoch": 1.57, "learning_rate": 4.0198692908653845e-05, "loss": 1.2241, "step": 5223 }, { "epoch": 1.57, "learning_rate": 4.019681490384616e-05, "loss": 1.0484, "step": 5224 }, { "epoch": 1.57, "learning_rate": 4.019493689903847e-05, "loss": 1.1253, "step": 5225 }, { "epoch": 1.57, "learning_rate": 4.019305889423077e-05, "loss": 1.2704, "step": 5226 }, { "epoch": 1.57, "learning_rate": 4.019118088942308e-05, "loss": 1.0562, "step": 5227 }, { "epoch": 1.57, "learning_rate": 4.018930288461539e-05, "loss": 1.1689, "step": 5228 }, { "epoch": 1.57, "learning_rate": 4.0187424879807694e-05, "loss": 1.1169, "step": 5229 }, { "epoch": 1.57, "learning_rate": 4.0185546875e-05, "loss": 1.2757, "step": 5230 }, { "epoch": 1.57, "learning_rate": 4.018366887019231e-05, "loss": 1.011, "step": 5231 }, { "epoch": 1.57, "learning_rate": 4.0181790865384615e-05, "loss": 1.4877, "step": 5232 }, { "epoch": 1.57, "learning_rate": 4.017991286057693e-05, "loss": 1.1759, "step": 5233 }, { "epoch": 1.57, "learning_rate": 4.017803485576923e-05, "loss": 1.3312, "step": 5234 }, { "epoch": 1.57, "learning_rate": 4.0176156850961537e-05, "loss": 1.109, "step": 5235 }, { "epoch": 1.57, "learning_rate": 4.017427884615385e-05, "loss": 1.0669, "step": 5236 }, { "epoch": 1.57, "learning_rate": 4.017240084134616e-05, "loss": 1.3533, "step": 5237 }, { "epoch": 1.57, "learning_rate": 4.0170522836538464e-05, "loss": 1.1632, "step": 5238 }, { "epoch": 1.57, "learning_rate": 4.0168644831730776e-05, "loss": 1.1746, "step": 5239 }, { "epoch": 1.57, "learning_rate": 4.016676682692308e-05, "loss": 1.3068, "step": 5240 }, { "epoch": 1.57, "learning_rate": 4.0164888822115385e-05, "loss": 0.9808, "step": 5241 }, { "epoch": 1.58, "learning_rate": 4.01630108173077e-05, "loss": 1.2302, "step": 5242 }, { "epoch": 1.58, "learning_rate": 4.01611328125e-05, "loss": 0.8798, "step": 5243 }, { "epoch": 1.58, "learning_rate": 4.0159254807692307e-05, "loss": 1.377, "step": 5244 }, { "epoch": 1.58, "learning_rate": 4.015737680288462e-05, "loss": 1.144, "step": 5245 }, { "epoch": 1.58, "learning_rate": 4.015549879807692e-05, "loss": 1.2657, "step": 5246 }, { "epoch": 1.58, "learning_rate": 4.015362079326923e-05, "loss": 1.1633, "step": 5247 }, { "epoch": 1.58, "learning_rate": 4.015174278846154e-05, "loss": 0.9168, "step": 5248 }, { "epoch": 1.58, "learning_rate": 4.0149864783653844e-05, "loss": 1.023, "step": 5249 }, { "epoch": 1.58, "learning_rate": 4.0147986778846156e-05, "loss": 1.281, "step": 5250 }, { "epoch": 1.58, "learning_rate": 4.014610877403847e-05, "loss": 1.4123, "step": 5251 }, { "epoch": 1.58, "learning_rate": 4.014423076923077e-05, "loss": 1.191, "step": 5252 }, { "epoch": 1.58, "learning_rate": 4.014235276442308e-05, "loss": 1.2437, "step": 5253 }, { "epoch": 1.58, "learning_rate": 4.014047475961539e-05, "loss": 1.1949, "step": 5254 }, { "epoch": 1.58, "learning_rate": 4.013859675480769e-05, "loss": 1.166, "step": 5255 }, { "epoch": 1.58, "learning_rate": 4.0136718750000004e-05, "loss": 1.2383, "step": 5256 }, { "epoch": 1.58, "learning_rate": 4.013484074519231e-05, "loss": 1.2447, "step": 5257 }, { "epoch": 1.58, "learning_rate": 4.0132962740384614e-05, "loss": 1.0353, "step": 5258 }, { "epoch": 1.58, "learning_rate": 4.0131084735576926e-05, "loss": 1.0626, "step": 5259 }, { "epoch": 1.58, "learning_rate": 4.012920673076923e-05, "loss": 0.9552, "step": 5260 }, { "epoch": 1.58, "learning_rate": 4.0127328725961535e-05, "loss": 1.1433, "step": 5261 }, { "epoch": 1.58, "learning_rate": 4.012545072115385e-05, "loss": 1.2069, "step": 5262 }, { "epoch": 1.58, "learning_rate": 4.012357271634616e-05, "loss": 1.2104, "step": 5263 }, { "epoch": 1.58, "learning_rate": 4.012169471153846e-05, "loss": 1.2533, "step": 5264 }, { "epoch": 1.58, "learning_rate": 4.0119816706730775e-05, "loss": 1.1819, "step": 5265 }, { "epoch": 1.58, "learning_rate": 4.011793870192308e-05, "loss": 1.1308, "step": 5266 }, { "epoch": 1.58, "learning_rate": 4.0116060697115384e-05, "loss": 1.0412, "step": 5267 }, { "epoch": 1.58, "learning_rate": 4.0114182692307696e-05, "loss": 1.2186, "step": 5268 }, { "epoch": 1.58, "learning_rate": 4.01123046875e-05, "loss": 1.0544, "step": 5269 }, { "epoch": 1.58, "learning_rate": 4.011042668269231e-05, "loss": 1.2037, "step": 5270 }, { "epoch": 1.58, "learning_rate": 4.010854867788462e-05, "loss": 0.8616, "step": 5271 }, { "epoch": 1.58, "learning_rate": 4.010667067307692e-05, "loss": 1.0414, "step": 5272 }, { "epoch": 1.58, "learning_rate": 4.010479266826923e-05, "loss": 1.0502, "step": 5273 }, { "epoch": 1.58, "learning_rate": 4.010291466346154e-05, "loss": 1.2429, "step": 5274 }, { "epoch": 1.59, "learning_rate": 4.010103665865384e-05, "loss": 1.1686, "step": 5275 }, { "epoch": 1.59, "learning_rate": 4.009915865384616e-05, "loss": 1.032, "step": 5276 }, { "epoch": 1.59, "learning_rate": 4.0097280649038466e-05, "loss": 0.9544, "step": 5277 }, { "epoch": 1.59, "learning_rate": 4.009540264423077e-05, "loss": 1.0722, "step": 5278 }, { "epoch": 1.59, "learning_rate": 4.009352463942308e-05, "loss": 0.9916, "step": 5279 }, { "epoch": 1.59, "learning_rate": 4.009164663461539e-05, "loss": 0.9156, "step": 5280 }, { "epoch": 1.59, "learning_rate": 4.008976862980769e-05, "loss": 1.3006, "step": 5281 }, { "epoch": 1.59, "learning_rate": 4.0087890625e-05, "loss": 0.8119, "step": 5282 }, { "epoch": 1.59, "learning_rate": 4.008601262019231e-05, "loss": 0.8524, "step": 5283 }, { "epoch": 1.59, "learning_rate": 4.008413461538462e-05, "loss": 1.0749, "step": 5284 }, { "epoch": 1.59, "learning_rate": 4.0082256610576924e-05, "loss": 0.9198, "step": 5285 }, { "epoch": 1.59, "learning_rate": 4.008037860576923e-05, "loss": 1.3091, "step": 5286 }, { "epoch": 1.59, "learning_rate": 4.007850060096154e-05, "loss": 1.0997, "step": 5287 }, { "epoch": 1.59, "learning_rate": 4.0076622596153845e-05, "loss": 1.1305, "step": 5288 }, { "epoch": 1.59, "learning_rate": 4.007474459134616e-05, "loss": 1.3673, "step": 5289 }, { "epoch": 1.59, "learning_rate": 4.007286658653847e-05, "loss": 1.1184, "step": 5290 }, { "epoch": 1.59, "learning_rate": 4.007098858173077e-05, "loss": 1.1387, "step": 5291 }, { "epoch": 1.59, "learning_rate": 4.006911057692308e-05, "loss": 1.021, "step": 5292 }, { "epoch": 1.59, "learning_rate": 4.006723257211539e-05, "loss": 1.0664, "step": 5293 }, { "epoch": 1.59, "learning_rate": 4.0065354567307694e-05, "loss": 1.1627, "step": 5294 }, { "epoch": 1.59, "learning_rate": 4.00634765625e-05, "loss": 1.0276, "step": 5295 }, { "epoch": 1.59, "learning_rate": 4.006159855769231e-05, "loss": 1.0523, "step": 5296 }, { "epoch": 1.59, "learning_rate": 4.0059720552884615e-05, "loss": 1.1479, "step": 5297 }, { "epoch": 1.59, "learning_rate": 4.005784254807693e-05, "loss": 0.8293, "step": 5298 }, { "epoch": 1.59, "learning_rate": 4.005596454326923e-05, "loss": 1.0112, "step": 5299 }, { "epoch": 1.59, "learning_rate": 4.0054086538461536e-05, "loss": 0.9611, "step": 5300 }, { "epoch": 1.59, "learning_rate": 4.005220853365385e-05, "loss": 1.1744, "step": 5301 }, { "epoch": 1.59, "learning_rate": 4.005033052884616e-05, "loss": 1.0973, "step": 5302 }, { "epoch": 1.59, "learning_rate": 4.0048452524038464e-05, "loss": 1.4103, "step": 5303 }, { "epoch": 1.59, "learning_rate": 4.0046574519230776e-05, "loss": 1.2944, "step": 5304 }, { "epoch": 1.59, "learning_rate": 4.004469651442308e-05, "loss": 1.1674, "step": 5305 }, { "epoch": 1.59, "learning_rate": 4.0042818509615385e-05, "loss": 0.8849, "step": 5306 }, { "epoch": 1.59, "learning_rate": 4.00409405048077e-05, "loss": 1.1697, "step": 5307 }, { "epoch": 1.59, "learning_rate": 4.00390625e-05, "loss": 1.1798, "step": 5308 }, { "epoch": 1.6, "learning_rate": 4.0037184495192306e-05, "loss": 1.0973, "step": 5309 }, { "epoch": 1.6, "learning_rate": 4.003530649038462e-05, "loss": 1.4036, "step": 5310 }, { "epoch": 1.6, "learning_rate": 4.003342848557692e-05, "loss": 1.3078, "step": 5311 }, { "epoch": 1.6, "learning_rate": 4.0031550480769234e-05, "loss": 1.2521, "step": 5312 }, { "epoch": 1.6, "learning_rate": 4.002967247596154e-05, "loss": 1.6061, "step": 5313 }, { "epoch": 1.6, "learning_rate": 4.0027794471153844e-05, "loss": 1.021, "step": 5314 }, { "epoch": 1.6, "learning_rate": 4.0025916466346155e-05, "loss": 1.086, "step": 5315 }, { "epoch": 1.6, "learning_rate": 4.002403846153847e-05, "loss": 1.0615, "step": 5316 }, { "epoch": 1.6, "learning_rate": 4.002216045673077e-05, "loss": 1.1072, "step": 5317 }, { "epoch": 1.6, "learning_rate": 4.002028245192308e-05, "loss": 1.2776, "step": 5318 }, { "epoch": 1.6, "learning_rate": 4.001840444711539e-05, "loss": 1.1693, "step": 5319 }, { "epoch": 1.6, "learning_rate": 4.001652644230769e-05, "loss": 0.9873, "step": 5320 }, { "epoch": 1.6, "learning_rate": 4.0014648437500004e-05, "loss": 0.9781, "step": 5321 }, { "epoch": 1.6, "learning_rate": 4.001277043269231e-05, "loss": 1.0717, "step": 5322 }, { "epoch": 1.6, "learning_rate": 4.0010892427884614e-05, "loss": 1.0432, "step": 5323 }, { "epoch": 1.6, "learning_rate": 4.0009014423076925e-05, "loss": 1.0892, "step": 5324 }, { "epoch": 1.6, "learning_rate": 4.000713641826923e-05, "loss": 1.1761, "step": 5325 }, { "epoch": 1.6, "learning_rate": 4.0005258413461535e-05, "loss": 1.226, "step": 5326 }, { "epoch": 1.6, "learning_rate": 4.0003380408653847e-05, "loss": 1.0627, "step": 5327 }, { "epoch": 1.6, "learning_rate": 4.000150240384616e-05, "loss": 1.0637, "step": 5328 }, { "epoch": 1.6, "learning_rate": 3.999962439903846e-05, "loss": 1.3748, "step": 5329 }, { "epoch": 1.6, "learning_rate": 3.9997746394230774e-05, "loss": 1.1967, "step": 5330 }, { "epoch": 1.6, "learning_rate": 3.999586838942308e-05, "loss": 1.1685, "step": 5331 }, { "epoch": 1.6, "learning_rate": 3.999399038461539e-05, "loss": 1.1275, "step": 5332 }, { "epoch": 1.6, "learning_rate": 3.9992112379807696e-05, "loss": 1.1548, "step": 5333 }, { "epoch": 1.6, "learning_rate": 3.9990234375e-05, "loss": 1.1708, "step": 5334 }, { "epoch": 1.6, "learning_rate": 3.998835637019231e-05, "loss": 1.117, "step": 5335 }, { "epoch": 1.6, "learning_rate": 3.9986478365384617e-05, "loss": 0.9911, "step": 5336 }, { "epoch": 1.6, "learning_rate": 3.998460036057692e-05, "loss": 1.0814, "step": 5337 }, { "epoch": 1.6, "learning_rate": 3.998272235576923e-05, "loss": 1.1887, "step": 5338 }, { "epoch": 1.6, "learning_rate": 3.998084435096154e-05, "loss": 1.0224, "step": 5339 }, { "epoch": 1.6, "learning_rate": 3.997896634615384e-05, "loss": 1.2317, "step": 5340 }, { "epoch": 1.6, "learning_rate": 3.997708834134616e-05, "loss": 1.4591, "step": 5341 }, { "epoch": 1.61, "learning_rate": 3.9975210336538466e-05, "loss": 1.318, "step": 5342 }, { "epoch": 1.61, "learning_rate": 3.997333233173077e-05, "loss": 0.9074, "step": 5343 }, { "epoch": 1.61, "learning_rate": 3.997145432692308e-05, "loss": 1.2353, "step": 5344 }, { "epoch": 1.61, "learning_rate": 3.996957632211539e-05, "loss": 1.0269, "step": 5345 }, { "epoch": 1.61, "learning_rate": 3.996769831730769e-05, "loss": 1.0707, "step": 5346 }, { "epoch": 1.61, "learning_rate": 3.99658203125e-05, "loss": 1.1452, "step": 5347 }, { "epoch": 1.61, "learning_rate": 3.996394230769231e-05, "loss": 1.3909, "step": 5348 }, { "epoch": 1.61, "learning_rate": 3.996206430288462e-05, "loss": 1.075, "step": 5349 }, { "epoch": 1.61, "learning_rate": 3.9960186298076924e-05, "loss": 1.0711, "step": 5350 }, { "epoch": 1.61, "learning_rate": 3.995830829326923e-05, "loss": 1.0308, "step": 5351 }, { "epoch": 1.61, "learning_rate": 3.995643028846154e-05, "loss": 1.054, "step": 5352 }, { "epoch": 1.61, "learning_rate": 3.9954552283653845e-05, "loss": 1.1256, "step": 5353 }, { "epoch": 1.61, "learning_rate": 3.995267427884616e-05, "loss": 1.1669, "step": 5354 }, { "epoch": 1.61, "learning_rate": 3.995079627403847e-05, "loss": 1.0025, "step": 5355 }, { "epoch": 1.61, "learning_rate": 3.994891826923077e-05, "loss": 1.2115, "step": 5356 }, { "epoch": 1.61, "learning_rate": 3.994704026442308e-05, "loss": 1.2957, "step": 5357 }, { "epoch": 1.61, "learning_rate": 3.994516225961539e-05, "loss": 0.9871, "step": 5358 }, { "epoch": 1.61, "learning_rate": 3.9943284254807694e-05, "loss": 1.1387, "step": 5359 }, { "epoch": 1.61, "learning_rate": 3.994140625e-05, "loss": 1.0922, "step": 5360 }, { "epoch": 1.61, "learning_rate": 3.993952824519231e-05, "loss": 1.2852, "step": 5361 }, { "epoch": 1.61, "learning_rate": 3.9937650240384615e-05, "loss": 1.293, "step": 5362 }, { "epoch": 1.61, "learning_rate": 3.993577223557693e-05, "loss": 1.1903, "step": 5363 }, { "epoch": 1.61, "learning_rate": 3.993389423076923e-05, "loss": 1.1054, "step": 5364 }, { "epoch": 1.61, "learning_rate": 3.9932016225961536e-05, "loss": 1.1155, "step": 5365 }, { "epoch": 1.61, "learning_rate": 3.993013822115385e-05, "loss": 1.0904, "step": 5366 }, { "epoch": 1.61, "learning_rate": 3.992826021634616e-05, "loss": 1.1984, "step": 5367 }, { "epoch": 1.61, "learning_rate": 3.9926382211538464e-05, "loss": 0.8053, "step": 5368 }, { "epoch": 1.61, "learning_rate": 3.9924504206730776e-05, "loss": 0.9872, "step": 5369 }, { "epoch": 1.61, "learning_rate": 3.992262620192308e-05, "loss": 1.1149, "step": 5370 }, { "epoch": 1.61, "learning_rate": 3.9920748197115385e-05, "loss": 1.1502, "step": 5371 }, { "epoch": 1.61, "learning_rate": 3.99188701923077e-05, "loss": 1.2463, "step": 5372 }, { "epoch": 1.61, "learning_rate": 3.99169921875e-05, "loss": 1.389, "step": 5373 }, { "epoch": 1.61, "learning_rate": 3.9915114182692306e-05, "loss": 1.2943, "step": 5374 }, { "epoch": 1.62, "learning_rate": 3.991323617788462e-05, "loss": 1.3752, "step": 5375 }, { "epoch": 1.62, "learning_rate": 3.991135817307692e-05, "loss": 0.9439, "step": 5376 }, { "epoch": 1.62, "learning_rate": 3.9909480168269234e-05, "loss": 1.1774, "step": 5377 }, { "epoch": 1.62, "learning_rate": 3.990760216346154e-05, "loss": 1.0671, "step": 5378 }, { "epoch": 1.62, "learning_rate": 3.9905724158653844e-05, "loss": 0.8919, "step": 5379 }, { "epoch": 1.62, "learning_rate": 3.9903846153846155e-05, "loss": 1.1577, "step": 5380 }, { "epoch": 1.62, "learning_rate": 3.990196814903847e-05, "loss": 1.1701, "step": 5381 }, { "epoch": 1.62, "learning_rate": 3.990009014423077e-05, "loss": 1.2211, "step": 5382 }, { "epoch": 1.62, "learning_rate": 3.989821213942308e-05, "loss": 1.1978, "step": 5383 }, { "epoch": 1.62, "learning_rate": 3.989633413461539e-05, "loss": 1.1141, "step": 5384 }, { "epoch": 1.62, "learning_rate": 3.989445612980769e-05, "loss": 1.0702, "step": 5385 }, { "epoch": 1.62, "learning_rate": 3.9892578125000004e-05, "loss": 0.9117, "step": 5386 }, { "epoch": 1.62, "learning_rate": 3.989070012019231e-05, "loss": 1.1456, "step": 5387 }, { "epoch": 1.62, "learning_rate": 3.9888822115384614e-05, "loss": 1.2026, "step": 5388 }, { "epoch": 1.62, "learning_rate": 3.9886944110576925e-05, "loss": 1.1437, "step": 5389 }, { "epoch": 1.62, "learning_rate": 3.988506610576923e-05, "loss": 0.9396, "step": 5390 }, { "epoch": 1.62, "learning_rate": 3.9883188100961535e-05, "loss": 1.1137, "step": 5391 }, { "epoch": 1.62, "learning_rate": 3.9881310096153846e-05, "loss": 1.3604, "step": 5392 }, { "epoch": 1.62, "learning_rate": 3.987943209134616e-05, "loss": 1.4211, "step": 5393 }, { "epoch": 1.62, "learning_rate": 3.987755408653846e-05, "loss": 1.158, "step": 5394 }, { "epoch": 1.62, "learning_rate": 3.9875676081730774e-05, "loss": 0.9956, "step": 5395 }, { "epoch": 1.62, "learning_rate": 3.987379807692308e-05, "loss": 1.0673, "step": 5396 }, { "epoch": 1.62, "learning_rate": 3.987192007211539e-05, "loss": 1.3078, "step": 5397 }, { "epoch": 1.62, "learning_rate": 3.9870042067307695e-05, "loss": 1.1921, "step": 5398 }, { "epoch": 1.62, "learning_rate": 3.98681640625e-05, "loss": 1.0368, "step": 5399 }, { "epoch": 1.62, "learning_rate": 3.986628605769231e-05, "loss": 1.0642, "step": 5400 }, { "epoch": 1.62, "learning_rate": 3.9864408052884616e-05, "loss": 0.9676, "step": 5401 }, { "epoch": 1.62, "learning_rate": 3.986253004807692e-05, "loss": 1.1389, "step": 5402 }, { "epoch": 1.62, "learning_rate": 3.986065204326923e-05, "loss": 1.323, "step": 5403 }, { "epoch": 1.62, "learning_rate": 3.985877403846154e-05, "loss": 1.0136, "step": 5404 }, { "epoch": 1.62, "learning_rate": 3.985689603365384e-05, "loss": 1.0279, "step": 5405 }, { "epoch": 1.62, "learning_rate": 3.985501802884616e-05, "loss": 1.1088, "step": 5406 }, { "epoch": 1.62, "learning_rate": 3.9853140024038465e-05, "loss": 0.9378, "step": 5407 }, { "epoch": 1.62, "learning_rate": 3.985126201923077e-05, "loss": 1.1917, "step": 5408 }, { "epoch": 1.63, "learning_rate": 3.984938401442308e-05, "loss": 1.0909, "step": 5409 }, { "epoch": 1.63, "learning_rate": 3.9847506009615387e-05, "loss": 1.0717, "step": 5410 }, { "epoch": 1.63, "learning_rate": 3.984562800480769e-05, "loss": 0.8011, "step": 5411 }, { "epoch": 1.63, "learning_rate": 3.984375e-05, "loss": 1.0648, "step": 5412 }, { "epoch": 1.63, "learning_rate": 3.984187199519231e-05, "loss": 1.2327, "step": 5413 }, { "epoch": 1.63, "learning_rate": 3.983999399038462e-05, "loss": 1.1737, "step": 5414 }, { "epoch": 1.63, "learning_rate": 3.9838115985576924e-05, "loss": 1.2229, "step": 5415 }, { "epoch": 1.63, "learning_rate": 3.983623798076923e-05, "loss": 0.9232, "step": 5416 }, { "epoch": 1.63, "learning_rate": 3.983435997596154e-05, "loss": 1.0821, "step": 5417 }, { "epoch": 1.63, "learning_rate": 3.9832481971153845e-05, "loss": 1.2103, "step": 5418 }, { "epoch": 1.63, "learning_rate": 3.9830603966346157e-05, "loss": 1.0958, "step": 5419 }, { "epoch": 1.63, "learning_rate": 3.982872596153847e-05, "loss": 1.0483, "step": 5420 }, { "epoch": 1.63, "learning_rate": 3.982684795673077e-05, "loss": 1.1714, "step": 5421 }, { "epoch": 1.63, "learning_rate": 3.982496995192308e-05, "loss": 1.1038, "step": 5422 }, { "epoch": 1.63, "learning_rate": 3.982309194711539e-05, "loss": 1.1173, "step": 5423 }, { "epoch": 1.63, "learning_rate": 3.9821213942307694e-05, "loss": 1.214, "step": 5424 }, { "epoch": 1.63, "learning_rate": 3.98193359375e-05, "loss": 1.2362, "step": 5425 }, { "epoch": 1.63, "learning_rate": 3.981745793269231e-05, "loss": 1.1475, "step": 5426 }, { "epoch": 1.63, "learning_rate": 3.9815579927884615e-05, "loss": 1.0139, "step": 5427 }, { "epoch": 1.63, "learning_rate": 3.981370192307693e-05, "loss": 1.2034, "step": 5428 }, { "epoch": 1.63, "learning_rate": 3.981182391826923e-05, "loss": 1.177, "step": 5429 }, { "epoch": 1.63, "learning_rate": 3.9809945913461536e-05, "loss": 1.3839, "step": 5430 }, { "epoch": 1.63, "learning_rate": 3.980806790865385e-05, "loss": 1.2878, "step": 5431 }, { "epoch": 1.63, "learning_rate": 3.980618990384616e-05, "loss": 1.2958, "step": 5432 }, { "epoch": 1.63, "learning_rate": 3.9804311899038464e-05, "loss": 0.9343, "step": 5433 }, { "epoch": 1.63, "learning_rate": 3.9802433894230776e-05, "loss": 1.1402, "step": 5434 }, { "epoch": 1.63, "learning_rate": 3.980055588942308e-05, "loss": 1.2994, "step": 5435 }, { "epoch": 1.63, "learning_rate": 3.9798677884615385e-05, "loss": 1.1987, "step": 5436 }, { "epoch": 1.63, "learning_rate": 3.97967998798077e-05, "loss": 1.0121, "step": 5437 }, { "epoch": 1.63, "learning_rate": 3.9794921875e-05, "loss": 1.28, "step": 5438 }, { "epoch": 1.63, "learning_rate": 3.9793043870192306e-05, "loss": 1.093, "step": 5439 }, { "epoch": 1.63, "learning_rate": 3.979116586538462e-05, "loss": 1.0596, "step": 5440 }, { "epoch": 1.63, "learning_rate": 3.978928786057692e-05, "loss": 1.4633, "step": 5441 }, { "epoch": 1.64, "learning_rate": 3.9787409855769234e-05, "loss": 1.4199, "step": 5442 }, { "epoch": 1.64, "learning_rate": 3.978553185096154e-05, "loss": 0.9022, "step": 5443 }, { "epoch": 1.64, "learning_rate": 3.9783653846153844e-05, "loss": 1.3913, "step": 5444 }, { "epoch": 1.64, "learning_rate": 3.9781775841346155e-05, "loss": 1.126, "step": 5445 }, { "epoch": 1.64, "learning_rate": 3.977989783653847e-05, "loss": 1.1691, "step": 5446 }, { "epoch": 1.64, "learning_rate": 3.977801983173077e-05, "loss": 1.1652, "step": 5447 }, { "epoch": 1.64, "learning_rate": 3.977614182692308e-05, "loss": 1.0101, "step": 5448 }, { "epoch": 1.64, "learning_rate": 3.977426382211539e-05, "loss": 1.1201, "step": 5449 }, { "epoch": 1.64, "learning_rate": 3.977238581730769e-05, "loss": 1.3249, "step": 5450 }, { "epoch": 1.64, "learning_rate": 3.9770507812500004e-05, "loss": 1.1021, "step": 5451 }, { "epoch": 1.64, "learning_rate": 3.976862980769231e-05, "loss": 1.2332, "step": 5452 }, { "epoch": 1.64, "learning_rate": 3.9766751802884614e-05, "loss": 1.073, "step": 5453 }, { "epoch": 1.64, "learning_rate": 3.9764873798076925e-05, "loss": 1.0801, "step": 5454 }, { "epoch": 1.64, "learning_rate": 3.976299579326923e-05, "loss": 1.203, "step": 5455 }, { "epoch": 1.64, "learning_rate": 3.976111778846154e-05, "loss": 0.9813, "step": 5456 }, { "epoch": 1.64, "learning_rate": 3.9759239783653846e-05, "loss": 1.0036, "step": 5457 }, { "epoch": 1.64, "learning_rate": 3.975736177884616e-05, "loss": 1.2076, "step": 5458 }, { "epoch": 1.64, "learning_rate": 3.975548377403846e-05, "loss": 1.3542, "step": 5459 }, { "epoch": 1.64, "learning_rate": 3.9753605769230774e-05, "loss": 1.2321, "step": 5460 }, { "epoch": 1.64, "learning_rate": 3.975172776442308e-05, "loss": 1.0039, "step": 5461 }, { "epoch": 1.64, "learning_rate": 3.974984975961539e-05, "loss": 1.0785, "step": 5462 }, { "epoch": 1.64, "learning_rate": 3.9747971754807695e-05, "loss": 1.1403, "step": 5463 }, { "epoch": 1.64, "learning_rate": 3.974609375e-05, "loss": 1.0524, "step": 5464 }, { "epoch": 1.64, "learning_rate": 3.974421574519231e-05, "loss": 1.1251, "step": 5465 }, { "epoch": 1.64, "learning_rate": 3.9742337740384616e-05, "loss": 1.1481, "step": 5466 }, { "epoch": 1.64, "learning_rate": 3.974045973557692e-05, "loss": 1.3305, "step": 5467 }, { "epoch": 1.64, "learning_rate": 3.973858173076923e-05, "loss": 1.3011, "step": 5468 }, { "epoch": 1.64, "learning_rate": 3.973670372596154e-05, "loss": 1.067, "step": 5469 }, { "epoch": 1.64, "learning_rate": 3.973482572115384e-05, "loss": 1.2785, "step": 5470 }, { "epoch": 1.64, "learning_rate": 3.973294771634616e-05, "loss": 1.1478, "step": 5471 }, { "epoch": 1.64, "learning_rate": 3.9731069711538465e-05, "loss": 1.1749, "step": 5472 }, { "epoch": 1.64, "learning_rate": 3.972919170673077e-05, "loss": 0.883, "step": 5473 }, { "epoch": 1.64, "learning_rate": 3.972731370192308e-05, "loss": 1.3158, "step": 5474 }, { "epoch": 1.65, "learning_rate": 3.9725435697115386e-05, "loss": 1.2024, "step": 5475 }, { "epoch": 1.65, "learning_rate": 3.97235576923077e-05, "loss": 1.0245, "step": 5476 }, { "epoch": 1.65, "learning_rate": 3.97216796875e-05, "loss": 1.214, "step": 5477 }, { "epoch": 1.65, "learning_rate": 3.971980168269231e-05, "loss": 1.0616, "step": 5478 }, { "epoch": 1.65, "learning_rate": 3.971792367788462e-05, "loss": 1.163, "step": 5479 }, { "epoch": 1.65, "learning_rate": 3.9716045673076924e-05, "loss": 1.0968, "step": 5480 }, { "epoch": 1.65, "learning_rate": 3.971416766826923e-05, "loss": 1.0237, "step": 5481 }, { "epoch": 1.65, "learning_rate": 3.971228966346154e-05, "loss": 1.2694, "step": 5482 }, { "epoch": 1.65, "learning_rate": 3.9710411658653845e-05, "loss": 1.2796, "step": 5483 }, { "epoch": 1.65, "learning_rate": 3.9708533653846156e-05, "loss": 1.1364, "step": 5484 }, { "epoch": 1.65, "learning_rate": 3.970665564903847e-05, "loss": 1.0427, "step": 5485 }, { "epoch": 1.65, "learning_rate": 3.970477764423077e-05, "loss": 1.3153, "step": 5486 }, { "epoch": 1.65, "learning_rate": 3.970289963942308e-05, "loss": 1.0592, "step": 5487 }, { "epoch": 1.65, "learning_rate": 3.970102163461539e-05, "loss": 1.0398, "step": 5488 }, { "epoch": 1.65, "learning_rate": 3.9699143629807694e-05, "loss": 1.0704, "step": 5489 }, { "epoch": 1.65, "learning_rate": 3.9697265625e-05, "loss": 0.9769, "step": 5490 }, { "epoch": 1.65, "learning_rate": 3.969538762019231e-05, "loss": 1.0888, "step": 5491 }, { "epoch": 1.65, "learning_rate": 3.9693509615384615e-05, "loss": 1.0753, "step": 5492 }, { "epoch": 1.65, "learning_rate": 3.9691631610576927e-05, "loss": 1.1947, "step": 5493 }, { "epoch": 1.65, "learning_rate": 3.968975360576923e-05, "loss": 1.0966, "step": 5494 }, { "epoch": 1.65, "learning_rate": 3.9687875600961536e-05, "loss": 1.0336, "step": 5495 }, { "epoch": 1.65, "learning_rate": 3.968599759615385e-05, "loss": 1.0459, "step": 5496 }, { "epoch": 1.65, "learning_rate": 3.968411959134616e-05, "loss": 1.1185, "step": 5497 }, { "epoch": 1.65, "learning_rate": 3.9682241586538464e-05, "loss": 1.3126, "step": 5498 }, { "epoch": 1.65, "learning_rate": 3.9680363581730775e-05, "loss": 1.1465, "step": 5499 }, { "epoch": 1.65, "learning_rate": 3.967848557692308e-05, "loss": 0.9917, "step": 5500 }, { "epoch": 1.65, "learning_rate": 3.9676607572115385e-05, "loss": 1.2918, "step": 5501 }, { "epoch": 1.65, "learning_rate": 3.9674729567307697e-05, "loss": 0.9373, "step": 5502 }, { "epoch": 1.65, "learning_rate": 3.96728515625e-05, "loss": 0.9632, "step": 5503 }, { "epoch": 1.65, "learning_rate": 3.9670973557692306e-05, "loss": 1.198, "step": 5504 }, { "epoch": 1.65, "learning_rate": 3.966909555288462e-05, "loss": 1.1404, "step": 5505 }, { "epoch": 1.65, "learning_rate": 3.966721754807692e-05, "loss": 1.1378, "step": 5506 }, { "epoch": 1.65, "learning_rate": 3.9665339543269234e-05, "loss": 1.1969, "step": 5507 }, { "epoch": 1.66, "learning_rate": 3.966346153846154e-05, "loss": 1.1119, "step": 5508 }, { "epoch": 1.66, "learning_rate": 3.9661583533653844e-05, "loss": 1.2544, "step": 5509 }, { "epoch": 1.66, "learning_rate": 3.9659705528846155e-05, "loss": 1.2853, "step": 5510 }, { "epoch": 1.66, "learning_rate": 3.965782752403847e-05, "loss": 1.0578, "step": 5511 }, { "epoch": 1.66, "learning_rate": 3.965594951923077e-05, "loss": 0.9611, "step": 5512 }, { "epoch": 1.66, "learning_rate": 3.965407151442308e-05, "loss": 1.0941, "step": 5513 }, { "epoch": 1.66, "learning_rate": 3.965219350961539e-05, "loss": 1.2107, "step": 5514 }, { "epoch": 1.66, "learning_rate": 3.965031550480769e-05, "loss": 0.9527, "step": 5515 }, { "epoch": 1.66, "learning_rate": 3.9648437500000004e-05, "loss": 1.2879, "step": 5516 }, { "epoch": 1.66, "learning_rate": 3.964655949519231e-05, "loss": 1.415, "step": 5517 }, { "epoch": 1.66, "learning_rate": 3.9644681490384614e-05, "loss": 1.196, "step": 5518 }, { "epoch": 1.66, "learning_rate": 3.9642803485576925e-05, "loss": 0.8341, "step": 5519 }, { "epoch": 1.66, "learning_rate": 3.964092548076923e-05, "loss": 1.003, "step": 5520 }, { "epoch": 1.66, "learning_rate": 3.963904747596154e-05, "loss": 0.882, "step": 5521 }, { "epoch": 1.66, "learning_rate": 3.9637169471153846e-05, "loss": 1.203, "step": 5522 }, { "epoch": 1.66, "learning_rate": 3.963529146634616e-05, "loss": 1.0919, "step": 5523 }, { "epoch": 1.66, "learning_rate": 3.963341346153846e-05, "loss": 1.3631, "step": 5524 }, { "epoch": 1.66, "learning_rate": 3.9631535456730774e-05, "loss": 1.2158, "step": 5525 }, { "epoch": 1.66, "learning_rate": 3.962965745192308e-05, "loss": 0.8836, "step": 5526 }, { "epoch": 1.66, "learning_rate": 3.962777944711539e-05, "loss": 0.9771, "step": 5527 }, { "epoch": 1.66, "learning_rate": 3.9625901442307695e-05, "loss": 1.1474, "step": 5528 }, { "epoch": 1.66, "learning_rate": 3.96240234375e-05, "loss": 1.0184, "step": 5529 }, { "epoch": 1.66, "learning_rate": 3.962214543269231e-05, "loss": 1.2138, "step": 5530 }, { "epoch": 1.66, "learning_rate": 3.9620267427884616e-05, "loss": 1.1445, "step": 5531 }, { "epoch": 1.66, "learning_rate": 3.961838942307692e-05, "loss": 1.166, "step": 5532 }, { "epoch": 1.66, "learning_rate": 3.961651141826923e-05, "loss": 1.0992, "step": 5533 }, { "epoch": 1.66, "learning_rate": 3.961463341346154e-05, "loss": 1.243, "step": 5534 }, { "epoch": 1.66, "learning_rate": 3.961275540865384e-05, "loss": 1.1728, "step": 5535 }, { "epoch": 1.66, "learning_rate": 3.961087740384616e-05, "loss": 1.2038, "step": 5536 }, { "epoch": 1.66, "learning_rate": 3.9608999399038465e-05, "loss": 1.2682, "step": 5537 }, { "epoch": 1.66, "learning_rate": 3.960712139423077e-05, "loss": 0.8896, "step": 5538 }, { "epoch": 1.66, "learning_rate": 3.960524338942308e-05, "loss": 1.1493, "step": 5539 }, { "epoch": 1.66, "learning_rate": 3.9603365384615386e-05, "loss": 1.2285, "step": 5540 }, { "epoch": 1.66, "learning_rate": 3.96014873798077e-05, "loss": 1.2646, "step": 5541 }, { "epoch": 1.67, "learning_rate": 3.9599609375e-05, "loss": 1.1046, "step": 5542 }, { "epoch": 1.67, "learning_rate": 3.959773137019231e-05, "loss": 1.3014, "step": 5543 }, { "epoch": 1.67, "learning_rate": 3.959585336538462e-05, "loss": 1.0706, "step": 5544 }, { "epoch": 1.67, "learning_rate": 3.9593975360576924e-05, "loss": 1.3837, "step": 5545 }, { "epoch": 1.67, "learning_rate": 3.959209735576923e-05, "loss": 1.0402, "step": 5546 }, { "epoch": 1.67, "learning_rate": 3.959021935096154e-05, "loss": 1.077, "step": 5547 }, { "epoch": 1.67, "learning_rate": 3.9588341346153845e-05, "loss": 1.2629, "step": 5548 }, { "epoch": 1.67, "learning_rate": 3.9586463341346156e-05, "loss": 0.8732, "step": 5549 }, { "epoch": 1.67, "learning_rate": 3.958458533653847e-05, "loss": 0.9797, "step": 5550 }, { "epoch": 1.67, "learning_rate": 3.958270733173077e-05, "loss": 1.1304, "step": 5551 }, { "epoch": 1.67, "learning_rate": 3.958082932692308e-05, "loss": 1.261, "step": 5552 }, { "epoch": 1.67, "learning_rate": 3.957895132211539e-05, "loss": 1.1922, "step": 5553 }, { "epoch": 1.67, "learning_rate": 3.9577073317307694e-05, "loss": 1.0285, "step": 5554 }, { "epoch": 1.67, "learning_rate": 3.95751953125e-05, "loss": 1.355, "step": 5555 }, { "epoch": 1.67, "learning_rate": 3.957331730769231e-05, "loss": 0.9313, "step": 5556 }, { "epoch": 1.67, "learning_rate": 3.9571439302884615e-05, "loss": 1.1502, "step": 5557 }, { "epoch": 1.67, "learning_rate": 3.9569561298076926e-05, "loss": 0.8557, "step": 5558 }, { "epoch": 1.67, "learning_rate": 3.956768329326923e-05, "loss": 1.2112, "step": 5559 }, { "epoch": 1.67, "learning_rate": 3.9565805288461536e-05, "loss": 0.9837, "step": 5560 }, { "epoch": 1.67, "learning_rate": 3.956392728365385e-05, "loss": 0.9535, "step": 5561 }, { "epoch": 1.67, "learning_rate": 3.956204927884616e-05, "loss": 1.0173, "step": 5562 }, { "epoch": 1.67, "learning_rate": 3.9560171274038464e-05, "loss": 1.174, "step": 5563 }, { "epoch": 1.67, "learning_rate": 3.9558293269230775e-05, "loss": 1.0432, "step": 5564 }, { "epoch": 1.67, "learning_rate": 3.955641526442308e-05, "loss": 1.4129, "step": 5565 }, { "epoch": 1.67, "learning_rate": 3.9554537259615385e-05, "loss": 1.0637, "step": 5566 }, { "epoch": 1.67, "learning_rate": 3.9552659254807696e-05, "loss": 1.1268, "step": 5567 }, { "epoch": 1.67, "learning_rate": 3.955078125e-05, "loss": 1.3969, "step": 5568 }, { "epoch": 1.67, "learning_rate": 3.9548903245192306e-05, "loss": 1.1921, "step": 5569 }, { "epoch": 1.67, "learning_rate": 3.954702524038462e-05, "loss": 1.2254, "step": 5570 }, { "epoch": 1.67, "learning_rate": 3.954514723557692e-05, "loss": 1.0372, "step": 5571 }, { "epoch": 1.67, "learning_rate": 3.9543269230769234e-05, "loss": 1.2466, "step": 5572 }, { "epoch": 1.67, "learning_rate": 3.954139122596154e-05, "loss": 1.1334, "step": 5573 }, { "epoch": 1.67, "learning_rate": 3.9539513221153843e-05, "loss": 1.0422, "step": 5574 }, { "epoch": 1.68, "learning_rate": 3.9537635216346155e-05, "loss": 1.3278, "step": 5575 }, { "epoch": 1.68, "learning_rate": 3.9535757211538467e-05, "loss": 1.0401, "step": 5576 }, { "epoch": 1.68, "learning_rate": 3.953387920673077e-05, "loss": 1.2887, "step": 5577 }, { "epoch": 1.68, "learning_rate": 3.953200120192308e-05, "loss": 0.9767, "step": 5578 }, { "epoch": 1.68, "learning_rate": 3.953012319711539e-05, "loss": 1.0126, "step": 5579 }, { "epoch": 1.68, "learning_rate": 3.952824519230769e-05, "loss": 1.0381, "step": 5580 }, { "epoch": 1.68, "learning_rate": 3.9526367187500004e-05, "loss": 1.2916, "step": 5581 }, { "epoch": 1.68, "learning_rate": 3.952448918269231e-05, "loss": 1.1698, "step": 5582 }, { "epoch": 1.68, "learning_rate": 3.9522611177884613e-05, "loss": 1.2002, "step": 5583 }, { "epoch": 1.68, "learning_rate": 3.9520733173076925e-05, "loss": 0.9529, "step": 5584 }, { "epoch": 1.68, "learning_rate": 3.951885516826923e-05, "loss": 0.9219, "step": 5585 }, { "epoch": 1.68, "learning_rate": 3.951697716346154e-05, "loss": 1.2132, "step": 5586 }, { "epoch": 1.68, "learning_rate": 3.9515099158653846e-05, "loss": 1.2808, "step": 5587 }, { "epoch": 1.68, "learning_rate": 3.951322115384616e-05, "loss": 1.0859, "step": 5588 }, { "epoch": 1.68, "learning_rate": 3.951134314903846e-05, "loss": 1.2253, "step": 5589 }, { "epoch": 1.68, "learning_rate": 3.9509465144230774e-05, "loss": 1.1258, "step": 5590 }, { "epoch": 1.68, "learning_rate": 3.950758713942308e-05, "loss": 0.9044, "step": 5591 }, { "epoch": 1.68, "learning_rate": 3.950570913461539e-05, "loss": 1.2525, "step": 5592 }, { "epoch": 1.68, "learning_rate": 3.9503831129807695e-05, "loss": 1.4217, "step": 5593 }, { "epoch": 1.68, "learning_rate": 3.9501953125e-05, "loss": 1.2275, "step": 5594 }, { "epoch": 1.68, "learning_rate": 3.950007512019231e-05, "loss": 1.1346, "step": 5595 }, { "epoch": 1.68, "learning_rate": 3.9498197115384616e-05, "loss": 1.2308, "step": 5596 }, { "epoch": 1.68, "learning_rate": 3.949631911057692e-05, "loss": 1.2548, "step": 5597 }, { "epoch": 1.68, "learning_rate": 3.949444110576923e-05, "loss": 1.004, "step": 5598 }, { "epoch": 1.68, "learning_rate": 3.949256310096154e-05, "loss": 1.3601, "step": 5599 }, { "epoch": 1.68, "learning_rate": 3.949068509615385e-05, "loss": 0.8949, "step": 5600 }, { "epoch": 1.68, "learning_rate": 3.948880709134616e-05, "loss": 1.0422, "step": 5601 }, { "epoch": 1.68, "learning_rate": 3.9486929086538465e-05, "loss": 1.121, "step": 5602 }, { "epoch": 1.68, "learning_rate": 3.948505108173077e-05, "loss": 1.221, "step": 5603 }, { "epoch": 1.68, "learning_rate": 3.948317307692308e-05, "loss": 1.079, "step": 5604 }, { "epoch": 1.68, "learning_rate": 3.9481295072115386e-05, "loss": 1.3453, "step": 5605 }, { "epoch": 1.68, "learning_rate": 3.94794170673077e-05, "loss": 1.1758, "step": 5606 }, { "epoch": 1.68, "learning_rate": 3.94775390625e-05, "loss": 1.1023, "step": 5607 }, { "epoch": 1.69, "learning_rate": 3.947566105769231e-05, "loss": 1.0441, "step": 5608 }, { "epoch": 1.69, "learning_rate": 3.947378305288462e-05, "loss": 1.097, "step": 5609 }, { "epoch": 1.69, "learning_rate": 3.9471905048076924e-05, "loss": 1.0198, "step": 5610 }, { "epoch": 1.69, "learning_rate": 3.947002704326923e-05, "loss": 1.133, "step": 5611 }, { "epoch": 1.69, "learning_rate": 3.946814903846154e-05, "loss": 1.1629, "step": 5612 }, { "epoch": 1.69, "learning_rate": 3.9466271033653845e-05, "loss": 1.2474, "step": 5613 }, { "epoch": 1.69, "learning_rate": 3.9464393028846156e-05, "loss": 1.1841, "step": 5614 }, { "epoch": 1.69, "learning_rate": 3.946251502403847e-05, "loss": 1.2335, "step": 5615 }, { "epoch": 1.69, "learning_rate": 3.946063701923077e-05, "loss": 1.33, "step": 5616 }, { "epoch": 1.69, "learning_rate": 3.945875901442308e-05, "loss": 1.1634, "step": 5617 }, { "epoch": 1.69, "learning_rate": 3.945688100961539e-05, "loss": 1.1008, "step": 5618 }, { "epoch": 1.69, "learning_rate": 3.9455003004807694e-05, "loss": 0.9759, "step": 5619 }, { "epoch": 1.69, "learning_rate": 3.9453125000000005e-05, "loss": 1.0869, "step": 5620 }, { "epoch": 1.69, "learning_rate": 3.945124699519231e-05, "loss": 1.2701, "step": 5621 }, { "epoch": 1.69, "learning_rate": 3.9449368990384615e-05, "loss": 1.1943, "step": 5622 }, { "epoch": 1.69, "learning_rate": 3.9447490985576926e-05, "loss": 0.7673, "step": 5623 }, { "epoch": 1.69, "learning_rate": 3.944561298076923e-05, "loss": 1.1128, "step": 5624 }, { "epoch": 1.69, "learning_rate": 3.9443734975961536e-05, "loss": 1.2908, "step": 5625 }, { "epoch": 1.69, "learning_rate": 3.944185697115385e-05, "loss": 1.2062, "step": 5626 }, { "epoch": 1.69, "learning_rate": 3.943997896634616e-05, "loss": 1.2291, "step": 5627 }, { "epoch": 1.69, "learning_rate": 3.9438100961538464e-05, "loss": 1.1831, "step": 5628 }, { "epoch": 1.69, "learning_rate": 3.9436222956730775e-05, "loss": 1.0773, "step": 5629 }, { "epoch": 1.69, "learning_rate": 3.943434495192308e-05, "loss": 1.4634, "step": 5630 }, { "epoch": 1.69, "learning_rate": 3.9432466947115385e-05, "loss": 1.1127, "step": 5631 }, { "epoch": 1.69, "learning_rate": 3.9430588942307696e-05, "loss": 1.0253, "step": 5632 }, { "epoch": 1.69, "learning_rate": 3.94287109375e-05, "loss": 1.3522, "step": 5633 }, { "epoch": 1.69, "learning_rate": 3.9426832932692306e-05, "loss": 1.1143, "step": 5634 }, { "epoch": 1.69, "learning_rate": 3.942495492788462e-05, "loss": 1.2738, "step": 5635 }, { "epoch": 1.69, "learning_rate": 3.942307692307692e-05, "loss": 0.971, "step": 5636 }, { "epoch": 1.69, "learning_rate": 3.9421198918269234e-05, "loss": 1.2017, "step": 5637 }, { "epoch": 1.69, "learning_rate": 3.941932091346154e-05, "loss": 1.1185, "step": 5638 }, { "epoch": 1.69, "learning_rate": 3.941744290865384e-05, "loss": 1.1567, "step": 5639 }, { "epoch": 1.69, "learning_rate": 3.941556490384616e-05, "loss": 0.9308, "step": 5640 }, { "epoch": 1.7, "learning_rate": 3.9413686899038466e-05, "loss": 1.2019, "step": 5641 }, { "epoch": 1.7, "learning_rate": 3.941180889423077e-05, "loss": 1.154, "step": 5642 }, { "epoch": 1.7, "learning_rate": 3.940993088942308e-05, "loss": 1.1647, "step": 5643 }, { "epoch": 1.7, "learning_rate": 3.940805288461539e-05, "loss": 1.1789, "step": 5644 }, { "epoch": 1.7, "learning_rate": 3.940617487980769e-05, "loss": 0.9255, "step": 5645 }, { "epoch": 1.7, "learning_rate": 3.9404296875000004e-05, "loss": 1.3055, "step": 5646 }, { "epoch": 1.7, "learning_rate": 3.940241887019231e-05, "loss": 1.4056, "step": 5647 }, { "epoch": 1.7, "learning_rate": 3.940054086538461e-05, "loss": 0.9536, "step": 5648 }, { "epoch": 1.7, "learning_rate": 3.9398662860576925e-05, "loss": 1.3147, "step": 5649 }, { "epoch": 1.7, "learning_rate": 3.939678485576923e-05, "loss": 1.2498, "step": 5650 }, { "epoch": 1.7, "learning_rate": 3.939490685096154e-05, "loss": 1.1897, "step": 5651 }, { "epoch": 1.7, "learning_rate": 3.9393028846153846e-05, "loss": 1.0145, "step": 5652 }, { "epoch": 1.7, "learning_rate": 3.939115084134616e-05, "loss": 1.1404, "step": 5653 }, { "epoch": 1.7, "learning_rate": 3.938927283653846e-05, "loss": 1.2454, "step": 5654 }, { "epoch": 1.7, "learning_rate": 3.9387394831730774e-05, "loss": 1.0442, "step": 5655 }, { "epoch": 1.7, "learning_rate": 3.938551682692308e-05, "loss": 1.3154, "step": 5656 }, { "epoch": 1.7, "learning_rate": 3.938363882211539e-05, "loss": 1.1782, "step": 5657 }, { "epoch": 1.7, "learning_rate": 3.9381760817307695e-05, "loss": 1.1513, "step": 5658 }, { "epoch": 1.7, "learning_rate": 3.93798828125e-05, "loss": 1.245, "step": 5659 }, { "epoch": 1.7, "learning_rate": 3.937800480769231e-05, "loss": 1.2517, "step": 5660 }, { "epoch": 1.7, "learning_rate": 3.9376126802884616e-05, "loss": 1.0365, "step": 5661 }, { "epoch": 1.7, "learning_rate": 3.937424879807692e-05, "loss": 1.0491, "step": 5662 }, { "epoch": 1.7, "learning_rate": 3.937237079326923e-05, "loss": 1.0557, "step": 5663 }, { "epoch": 1.7, "learning_rate": 3.937049278846154e-05, "loss": 1.3096, "step": 5664 }, { "epoch": 1.7, "learning_rate": 3.936861478365385e-05, "loss": 1.6054, "step": 5665 }, { "epoch": 1.7, "learning_rate": 3.936673677884616e-05, "loss": 1.2403, "step": 5666 }, { "epoch": 1.7, "learning_rate": 3.9364858774038465e-05, "loss": 1.2122, "step": 5667 }, { "epoch": 1.7, "learning_rate": 3.936298076923077e-05, "loss": 1.3516, "step": 5668 }, { "epoch": 1.7, "learning_rate": 3.936110276442308e-05, "loss": 0.9978, "step": 5669 }, { "epoch": 1.7, "learning_rate": 3.9359224759615386e-05, "loss": 1.0737, "step": 5670 }, { "epoch": 1.7, "learning_rate": 3.93573467548077e-05, "loss": 1.1142, "step": 5671 }, { "epoch": 1.7, "learning_rate": 3.935546875e-05, "loss": 1.1503, "step": 5672 }, { "epoch": 1.7, "learning_rate": 3.935359074519231e-05, "loss": 1.1834, "step": 5673 }, { "epoch": 1.7, "learning_rate": 3.935171274038462e-05, "loss": 1.156, "step": 5674 }, { "epoch": 1.71, "learning_rate": 3.9349834735576924e-05, "loss": 1.0244, "step": 5675 }, { "epoch": 1.71, "learning_rate": 3.934795673076923e-05, "loss": 1.2884, "step": 5676 }, { "epoch": 1.71, "learning_rate": 3.934607872596154e-05, "loss": 1.1565, "step": 5677 }, { "epoch": 1.71, "learning_rate": 3.9344200721153845e-05, "loss": 1.249, "step": 5678 }, { "epoch": 1.71, "learning_rate": 3.9342322716346156e-05, "loss": 0.9499, "step": 5679 }, { "epoch": 1.71, "learning_rate": 3.934044471153847e-05, "loss": 1.0256, "step": 5680 }, { "epoch": 1.71, "learning_rate": 3.933856670673077e-05, "loss": 1.2376, "step": 5681 }, { "epoch": 1.71, "learning_rate": 3.933668870192308e-05, "loss": 1.1688, "step": 5682 }, { "epoch": 1.71, "learning_rate": 3.933481069711539e-05, "loss": 1.2103, "step": 5683 }, { "epoch": 1.71, "learning_rate": 3.9332932692307694e-05, "loss": 1.2314, "step": 5684 }, { "epoch": 1.71, "learning_rate": 3.9331054687500005e-05, "loss": 1.055, "step": 5685 }, { "epoch": 1.71, "learning_rate": 3.932917668269231e-05, "loss": 1.2449, "step": 5686 }, { "epoch": 1.71, "learning_rate": 3.9327298677884615e-05, "loss": 0.9915, "step": 5687 }, { "epoch": 1.71, "learning_rate": 3.9325420673076926e-05, "loss": 1.508, "step": 5688 }, { "epoch": 1.71, "learning_rate": 3.932354266826923e-05, "loss": 1.2214, "step": 5689 }, { "epoch": 1.71, "learning_rate": 3.9321664663461536e-05, "loss": 1.285, "step": 5690 }, { "epoch": 1.71, "learning_rate": 3.931978665865385e-05, "loss": 1.123, "step": 5691 }, { "epoch": 1.71, "learning_rate": 3.931790865384616e-05, "loss": 1.0413, "step": 5692 }, { "epoch": 1.71, "learning_rate": 3.9316030649038464e-05, "loss": 1.3406, "step": 5693 }, { "epoch": 1.71, "learning_rate": 3.9314152644230775e-05, "loss": 1.1517, "step": 5694 }, { "epoch": 1.71, "learning_rate": 3.931227463942308e-05, "loss": 0.903, "step": 5695 }, { "epoch": 1.71, "learning_rate": 3.9310396634615385e-05, "loss": 1.041, "step": 5696 }, { "epoch": 1.71, "learning_rate": 3.9308518629807696e-05, "loss": 1.236, "step": 5697 }, { "epoch": 1.71, "learning_rate": 3.9306640625e-05, "loss": 1.132, "step": 5698 }, { "epoch": 1.71, "learning_rate": 3.9304762620192306e-05, "loss": 1.0697, "step": 5699 }, { "epoch": 1.71, "learning_rate": 3.930288461538462e-05, "loss": 1.2982, "step": 5700 }, { "epoch": 1.71, "learning_rate": 3.930100661057692e-05, "loss": 1.0549, "step": 5701 }, { "epoch": 1.71, "learning_rate": 3.9299128605769234e-05, "loss": 1.139, "step": 5702 }, { "epoch": 1.71, "learning_rate": 3.929725060096154e-05, "loss": 0.988, "step": 5703 }, { "epoch": 1.71, "learning_rate": 3.929537259615384e-05, "loss": 1.3163, "step": 5704 }, { "epoch": 1.71, "learning_rate": 3.929349459134616e-05, "loss": 0.9311, "step": 5705 }, { "epoch": 1.71, "learning_rate": 3.9291616586538466e-05, "loss": 1.2823, "step": 5706 }, { "epoch": 1.71, "learning_rate": 3.928973858173077e-05, "loss": 1.0365, "step": 5707 }, { "epoch": 1.72, "learning_rate": 3.928786057692308e-05, "loss": 1.0631, "step": 5708 }, { "epoch": 1.72, "learning_rate": 3.928598257211539e-05, "loss": 1.0962, "step": 5709 }, { "epoch": 1.72, "learning_rate": 3.928410456730769e-05, "loss": 1.014, "step": 5710 }, { "epoch": 1.72, "learning_rate": 3.9282226562500004e-05, "loss": 1.1111, "step": 5711 }, { "epoch": 1.72, "learning_rate": 3.928034855769231e-05, "loss": 1.2591, "step": 5712 }, { "epoch": 1.72, "learning_rate": 3.927847055288461e-05, "loss": 1.1041, "step": 5713 }, { "epoch": 1.72, "learning_rate": 3.9276592548076925e-05, "loss": 0.9992, "step": 5714 }, { "epoch": 1.72, "learning_rate": 3.927471454326923e-05, "loss": 0.9459, "step": 5715 }, { "epoch": 1.72, "learning_rate": 3.927283653846154e-05, "loss": 1.0427, "step": 5716 }, { "epoch": 1.72, "learning_rate": 3.9270958533653846e-05, "loss": 1.2209, "step": 5717 }, { "epoch": 1.72, "learning_rate": 3.926908052884616e-05, "loss": 1.3748, "step": 5718 }, { "epoch": 1.72, "learning_rate": 3.926720252403846e-05, "loss": 1.2568, "step": 5719 }, { "epoch": 1.72, "learning_rate": 3.9265324519230774e-05, "loss": 1.2665, "step": 5720 }, { "epoch": 1.72, "learning_rate": 3.926344651442308e-05, "loss": 1.1944, "step": 5721 }, { "epoch": 1.72, "learning_rate": 3.926156850961539e-05, "loss": 1.322, "step": 5722 }, { "epoch": 1.72, "learning_rate": 3.9259690504807695e-05, "loss": 1.0529, "step": 5723 }, { "epoch": 1.72, "learning_rate": 3.92578125e-05, "loss": 1.2198, "step": 5724 }, { "epoch": 1.72, "learning_rate": 3.925593449519231e-05, "loss": 1.2337, "step": 5725 }, { "epoch": 1.72, "learning_rate": 3.9254056490384616e-05, "loss": 1.269, "step": 5726 }, { "epoch": 1.72, "learning_rate": 3.925217848557692e-05, "loss": 1.1338, "step": 5727 }, { "epoch": 1.72, "learning_rate": 3.925030048076923e-05, "loss": 1.1918, "step": 5728 }, { "epoch": 1.72, "learning_rate": 3.924842247596154e-05, "loss": 1.3559, "step": 5729 }, { "epoch": 1.72, "learning_rate": 3.924654447115385e-05, "loss": 1.2967, "step": 5730 }, { "epoch": 1.72, "learning_rate": 3.924466646634616e-05, "loss": 0.8283, "step": 5731 }, { "epoch": 1.72, "learning_rate": 3.9242788461538465e-05, "loss": 1.1831, "step": 5732 }, { "epoch": 1.72, "learning_rate": 3.924091045673077e-05, "loss": 1.0269, "step": 5733 }, { "epoch": 1.72, "learning_rate": 3.923903245192308e-05, "loss": 0.9952, "step": 5734 }, { "epoch": 1.72, "learning_rate": 3.9237154447115386e-05, "loss": 1.1532, "step": 5735 }, { "epoch": 1.72, "learning_rate": 3.92352764423077e-05, "loss": 1.2555, "step": 5736 }, { "epoch": 1.72, "learning_rate": 3.92333984375e-05, "loss": 1.0986, "step": 5737 }, { "epoch": 1.72, "learning_rate": 3.923152043269231e-05, "loss": 1.308, "step": 5738 }, { "epoch": 1.72, "learning_rate": 3.922964242788462e-05, "loss": 1.2861, "step": 5739 }, { "epoch": 1.72, "learning_rate": 3.9227764423076923e-05, "loss": 1.1311, "step": 5740 }, { "epoch": 1.73, "learning_rate": 3.922588641826923e-05, "loss": 1.328, "step": 5741 }, { "epoch": 1.73, "learning_rate": 3.922400841346154e-05, "loss": 1.2193, "step": 5742 }, { "epoch": 1.73, "learning_rate": 3.9222130408653845e-05, "loss": 0.9712, "step": 5743 }, { "epoch": 1.73, "learning_rate": 3.9220252403846156e-05, "loss": 0.9539, "step": 5744 }, { "epoch": 1.73, "learning_rate": 3.921837439903847e-05, "loss": 1.284, "step": 5745 }, { "epoch": 1.73, "learning_rate": 3.921649639423077e-05, "loss": 1.212, "step": 5746 }, { "epoch": 1.73, "learning_rate": 3.921461838942308e-05, "loss": 1.0879, "step": 5747 }, { "epoch": 1.73, "learning_rate": 3.921274038461539e-05, "loss": 1.1478, "step": 5748 }, { "epoch": 1.73, "learning_rate": 3.9210862379807693e-05, "loss": 0.9443, "step": 5749 }, { "epoch": 1.73, "learning_rate": 3.9208984375000005e-05, "loss": 1.1689, "step": 5750 }, { "epoch": 1.73, "learning_rate": 3.920710637019231e-05, "loss": 1.2847, "step": 5751 }, { "epoch": 1.73, "learning_rate": 3.9205228365384615e-05, "loss": 1.2859, "step": 5752 }, { "epoch": 1.73, "learning_rate": 3.9203350360576926e-05, "loss": 1.0448, "step": 5753 }, { "epoch": 1.73, "learning_rate": 3.920147235576923e-05, "loss": 1.2622, "step": 5754 }, { "epoch": 1.73, "learning_rate": 3.9199594350961536e-05, "loss": 1.3653, "step": 5755 }, { "epoch": 1.73, "learning_rate": 3.919771634615385e-05, "loss": 1.22, "step": 5756 }, { "epoch": 1.73, "learning_rate": 3.919583834134616e-05, "loss": 1.2366, "step": 5757 }, { "epoch": 1.73, "learning_rate": 3.9193960336538464e-05, "loss": 1.1901, "step": 5758 }, { "epoch": 1.73, "learning_rate": 3.9192082331730775e-05, "loss": 1.0193, "step": 5759 }, { "epoch": 1.73, "learning_rate": 3.919020432692308e-05, "loss": 1.04, "step": 5760 }, { "epoch": 1.73, "learning_rate": 3.9188326322115385e-05, "loss": 1.1574, "step": 5761 }, { "epoch": 1.73, "learning_rate": 3.9186448317307696e-05, "loss": 1.1464, "step": 5762 }, { "epoch": 1.73, "learning_rate": 3.91845703125e-05, "loss": 1.2003, "step": 5763 }, { "epoch": 1.73, "learning_rate": 3.918269230769231e-05, "loss": 1.1654, "step": 5764 }, { "epoch": 1.73, "learning_rate": 3.918081430288462e-05, "loss": 1.0455, "step": 5765 }, { "epoch": 1.73, "learning_rate": 3.917893629807692e-05, "loss": 1.0182, "step": 5766 }, { "epoch": 1.73, "learning_rate": 3.9177058293269234e-05, "loss": 1.0241, "step": 5767 }, { "epoch": 1.73, "learning_rate": 3.917518028846154e-05, "loss": 1.1526, "step": 5768 }, { "epoch": 1.73, "learning_rate": 3.917330228365384e-05, "loss": 1.1232, "step": 5769 }, { "epoch": 1.73, "learning_rate": 3.917142427884616e-05, "loss": 1.1655, "step": 5770 }, { "epoch": 1.73, "learning_rate": 3.9169546274038466e-05, "loss": 1.1955, "step": 5771 }, { "epoch": 1.73, "learning_rate": 3.916766826923077e-05, "loss": 1.0618, "step": 5772 }, { "epoch": 1.73, "learning_rate": 3.916579026442308e-05, "loss": 0.9749, "step": 5773 }, { "epoch": 1.73, "learning_rate": 3.916391225961539e-05, "loss": 1.1542, "step": 5774 }, { "epoch": 1.74, "learning_rate": 3.916203425480769e-05, "loss": 0.9571, "step": 5775 }, { "epoch": 1.74, "learning_rate": 3.9160156250000004e-05, "loss": 1.2349, "step": 5776 }, { "epoch": 1.74, "learning_rate": 3.915827824519231e-05, "loss": 1.0667, "step": 5777 }, { "epoch": 1.74, "learning_rate": 3.915640024038461e-05, "loss": 1.2026, "step": 5778 }, { "epoch": 1.74, "learning_rate": 3.9154522235576925e-05, "loss": 1.0866, "step": 5779 }, { "epoch": 1.74, "learning_rate": 3.915264423076923e-05, "loss": 1.0389, "step": 5780 }, { "epoch": 1.74, "learning_rate": 3.915076622596154e-05, "loss": 1.1424, "step": 5781 }, { "epoch": 1.74, "learning_rate": 3.9148888221153846e-05, "loss": 1.1515, "step": 5782 }, { "epoch": 1.74, "learning_rate": 3.914701021634616e-05, "loss": 1.1617, "step": 5783 }, { "epoch": 1.74, "learning_rate": 3.914513221153847e-05, "loss": 1.0554, "step": 5784 }, { "epoch": 1.74, "learning_rate": 3.9143254206730774e-05, "loss": 1.0374, "step": 5785 }, { "epoch": 1.74, "learning_rate": 3.914137620192308e-05, "loss": 1.1894, "step": 5786 }, { "epoch": 1.74, "learning_rate": 3.913949819711539e-05, "loss": 1.2377, "step": 5787 }, { "epoch": 1.74, "learning_rate": 3.9137620192307695e-05, "loss": 1.1316, "step": 5788 }, { "epoch": 1.74, "learning_rate": 3.91357421875e-05, "loss": 1.1898, "step": 5789 }, { "epoch": 1.74, "learning_rate": 3.913386418269231e-05, "loss": 1.1063, "step": 5790 }, { "epoch": 1.74, "learning_rate": 3.9131986177884616e-05, "loss": 0.9698, "step": 5791 }, { "epoch": 1.74, "learning_rate": 3.913010817307692e-05, "loss": 1.224, "step": 5792 }, { "epoch": 1.74, "learning_rate": 3.912823016826923e-05, "loss": 1.2677, "step": 5793 }, { "epoch": 1.74, "learning_rate": 3.912635216346154e-05, "loss": 1.1165, "step": 5794 }, { "epoch": 1.74, "learning_rate": 3.912447415865385e-05, "loss": 1.2137, "step": 5795 }, { "epoch": 1.74, "learning_rate": 3.912259615384616e-05, "loss": 0.9565, "step": 5796 }, { "epoch": 1.74, "learning_rate": 3.9120718149038465e-05, "loss": 1.212, "step": 5797 }, { "epoch": 1.74, "learning_rate": 3.911884014423077e-05, "loss": 1.0977, "step": 5798 }, { "epoch": 1.74, "learning_rate": 3.911696213942308e-05, "loss": 1.2379, "step": 5799 }, { "epoch": 1.74, "learning_rate": 3.9115084134615386e-05, "loss": 1.0489, "step": 5800 }, { "epoch": 1.74, "learning_rate": 3.91132061298077e-05, "loss": 1.1291, "step": 5801 }, { "epoch": 1.74, "learning_rate": 3.9111328125e-05, "loss": 1.206, "step": 5802 }, { "epoch": 1.74, "learning_rate": 3.910945012019231e-05, "loss": 1.1074, "step": 5803 }, { "epoch": 1.74, "learning_rate": 3.910757211538462e-05, "loss": 0.7824, "step": 5804 }, { "epoch": 1.74, "learning_rate": 3.910569411057692e-05, "loss": 1.1279, "step": 5805 }, { "epoch": 1.74, "learning_rate": 3.910381610576923e-05, "loss": 1.2705, "step": 5806 }, { "epoch": 1.74, "learning_rate": 3.910193810096154e-05, "loss": 1.2664, "step": 5807 }, { "epoch": 1.75, "learning_rate": 3.9100060096153844e-05, "loss": 1.0635, "step": 5808 }, { "epoch": 1.75, "learning_rate": 3.9098182091346156e-05, "loss": 1.3311, "step": 5809 }, { "epoch": 1.75, "learning_rate": 3.909630408653847e-05, "loss": 1.2139, "step": 5810 }, { "epoch": 1.75, "learning_rate": 3.909442608173077e-05, "loss": 1.2328, "step": 5811 }, { "epoch": 1.75, "learning_rate": 3.909254807692308e-05, "loss": 1.1946, "step": 5812 }, { "epoch": 1.75, "learning_rate": 3.909067007211539e-05, "loss": 1.2669, "step": 5813 }, { "epoch": 1.75, "learning_rate": 3.908879206730769e-05, "loss": 1.1674, "step": 5814 }, { "epoch": 1.75, "learning_rate": 3.9086914062500005e-05, "loss": 1.1496, "step": 5815 }, { "epoch": 1.75, "learning_rate": 3.908503605769231e-05, "loss": 1.3532, "step": 5816 }, { "epoch": 1.75, "learning_rate": 3.9083158052884614e-05, "loss": 1.3005, "step": 5817 }, { "epoch": 1.75, "learning_rate": 3.9081280048076926e-05, "loss": 1.1276, "step": 5818 }, { "epoch": 1.75, "learning_rate": 3.907940204326923e-05, "loss": 1.1838, "step": 5819 }, { "epoch": 1.75, "learning_rate": 3.9077524038461536e-05, "loss": 1.2577, "step": 5820 }, { "epoch": 1.75, "learning_rate": 3.907564603365385e-05, "loss": 1.1827, "step": 5821 }, { "epoch": 1.75, "learning_rate": 3.907376802884616e-05, "loss": 0.9877, "step": 5822 }, { "epoch": 1.75, "learning_rate": 3.9071890024038463e-05, "loss": 1.053, "step": 5823 }, { "epoch": 1.75, "learning_rate": 3.9070012019230775e-05, "loss": 1.1685, "step": 5824 }, { "epoch": 1.75, "learning_rate": 3.906813401442308e-05, "loss": 1.2814, "step": 5825 }, { "epoch": 1.75, "learning_rate": 3.9066256009615384e-05, "loss": 1.2475, "step": 5826 }, { "epoch": 1.75, "learning_rate": 3.9064378004807696e-05, "loss": 0.8947, "step": 5827 }, { "epoch": 1.75, "learning_rate": 3.90625e-05, "loss": 1.0844, "step": 5828 }, { "epoch": 1.75, "learning_rate": 3.906062199519231e-05, "loss": 1.2673, "step": 5829 }, { "epoch": 1.75, "learning_rate": 3.905874399038462e-05, "loss": 1.2918, "step": 5830 }, { "epoch": 1.75, "learning_rate": 3.905686598557692e-05, "loss": 0.9601, "step": 5831 }, { "epoch": 1.75, "learning_rate": 3.9054987980769233e-05, "loss": 1.0242, "step": 5832 }, { "epoch": 1.75, "learning_rate": 3.905310997596154e-05, "loss": 1.2231, "step": 5833 }, { "epoch": 1.75, "learning_rate": 3.905123197115384e-05, "loss": 1.0218, "step": 5834 }, { "epoch": 1.75, "learning_rate": 3.904935396634616e-05, "loss": 1.0761, "step": 5835 }, { "epoch": 1.75, "learning_rate": 3.9047475961538466e-05, "loss": 1.2149, "step": 5836 }, { "epoch": 1.75, "learning_rate": 3.904559795673077e-05, "loss": 0.9256, "step": 5837 }, { "epoch": 1.75, "learning_rate": 3.904371995192308e-05, "loss": 1.0572, "step": 5838 }, { "epoch": 1.75, "learning_rate": 3.904184194711539e-05, "loss": 1.17, "step": 5839 }, { "epoch": 1.75, "learning_rate": 3.903996394230769e-05, "loss": 1.0048, "step": 5840 }, { "epoch": 1.76, "learning_rate": 3.9038085937500004e-05, "loss": 0.8832, "step": 5841 }, { "epoch": 1.76, "learning_rate": 3.903620793269231e-05, "loss": 1.0587, "step": 5842 }, { "epoch": 1.76, "learning_rate": 3.903432992788461e-05, "loss": 0.9668, "step": 5843 }, { "epoch": 1.76, "learning_rate": 3.9032451923076925e-05, "loss": 1.2008, "step": 5844 }, { "epoch": 1.76, "learning_rate": 3.903057391826923e-05, "loss": 1.1457, "step": 5845 }, { "epoch": 1.76, "learning_rate": 3.902869591346154e-05, "loss": 1.3298, "step": 5846 }, { "epoch": 1.76, "learning_rate": 3.9026817908653846e-05, "loss": 1.0757, "step": 5847 }, { "epoch": 1.76, "learning_rate": 3.902493990384616e-05, "loss": 1.2304, "step": 5848 }, { "epoch": 1.76, "learning_rate": 3.902306189903847e-05, "loss": 1.1908, "step": 5849 }, { "epoch": 1.76, "learning_rate": 3.9021183894230774e-05, "loss": 1.0906, "step": 5850 }, { "epoch": 1.76, "learning_rate": 3.901930588942308e-05, "loss": 1.2844, "step": 5851 }, { "epoch": 1.76, "learning_rate": 3.901742788461539e-05, "loss": 0.9529, "step": 5852 }, { "epoch": 1.76, "learning_rate": 3.9015549879807695e-05, "loss": 1.1836, "step": 5853 }, { "epoch": 1.76, "learning_rate": 3.9013671875e-05, "loss": 1.173, "step": 5854 }, { "epoch": 1.76, "learning_rate": 3.901179387019231e-05, "loss": 1.3303, "step": 5855 }, { "epoch": 1.76, "learning_rate": 3.9009915865384616e-05, "loss": 0.9751, "step": 5856 }, { "epoch": 1.76, "learning_rate": 3.900803786057692e-05, "loss": 1.135, "step": 5857 }, { "epoch": 1.76, "learning_rate": 3.900615985576923e-05, "loss": 1.0666, "step": 5858 }, { "epoch": 1.76, "learning_rate": 3.900428185096154e-05, "loss": 1.2098, "step": 5859 }, { "epoch": 1.76, "learning_rate": 3.900240384615385e-05, "loss": 1.4277, "step": 5860 }, { "epoch": 1.76, "learning_rate": 3.900052584134616e-05, "loss": 1.0505, "step": 5861 }, { "epoch": 1.76, "learning_rate": 3.8998647836538465e-05, "loss": 1.0927, "step": 5862 }, { "epoch": 1.76, "learning_rate": 3.899676983173077e-05, "loss": 1.0255, "step": 5863 }, { "epoch": 1.76, "learning_rate": 3.899489182692308e-05, "loss": 1.3134, "step": 5864 }, { "epoch": 1.76, "learning_rate": 3.8993013822115386e-05, "loss": 1.1063, "step": 5865 }, { "epoch": 1.76, "learning_rate": 3.89911358173077e-05, "loss": 1.0741, "step": 5866 }, { "epoch": 1.76, "learning_rate": 3.89892578125e-05, "loss": 0.9852, "step": 5867 }, { "epoch": 1.76, "learning_rate": 3.898737980769231e-05, "loss": 1.3057, "step": 5868 }, { "epoch": 1.76, "learning_rate": 3.898550180288462e-05, "loss": 1.0556, "step": 5869 }, { "epoch": 1.76, "learning_rate": 3.898362379807692e-05, "loss": 1.1674, "step": 5870 }, { "epoch": 1.76, "learning_rate": 3.898174579326923e-05, "loss": 1.1093, "step": 5871 }, { "epoch": 1.76, "learning_rate": 3.897986778846154e-05, "loss": 1.0647, "step": 5872 }, { "epoch": 1.76, "learning_rate": 3.8977989783653844e-05, "loss": 1.0386, "step": 5873 }, { "epoch": 1.77, "learning_rate": 3.8976111778846156e-05, "loss": 1.0935, "step": 5874 }, { "epoch": 1.77, "learning_rate": 3.897423377403847e-05, "loss": 1.0625, "step": 5875 }, { "epoch": 1.77, "learning_rate": 3.897235576923077e-05, "loss": 1.002, "step": 5876 }, { "epoch": 1.77, "learning_rate": 3.897047776442308e-05, "loss": 1.5153, "step": 5877 }, { "epoch": 1.77, "learning_rate": 3.896859975961539e-05, "loss": 1.3014, "step": 5878 }, { "epoch": 1.77, "learning_rate": 3.896672175480769e-05, "loss": 1.0139, "step": 5879 }, { "epoch": 1.77, "learning_rate": 3.8964843750000005e-05, "loss": 1.3831, "step": 5880 }, { "epoch": 1.77, "learning_rate": 3.896296574519231e-05, "loss": 0.9413, "step": 5881 }, { "epoch": 1.77, "learning_rate": 3.8961087740384614e-05, "loss": 0.9925, "step": 5882 }, { "epoch": 1.77, "learning_rate": 3.8959209735576926e-05, "loss": 1.305, "step": 5883 }, { "epoch": 1.77, "learning_rate": 3.895733173076923e-05, "loss": 1.1034, "step": 5884 }, { "epoch": 1.77, "learning_rate": 3.8955453725961535e-05, "loss": 1.0156, "step": 5885 }, { "epoch": 1.77, "learning_rate": 3.895357572115385e-05, "loss": 1.3468, "step": 5886 }, { "epoch": 1.77, "learning_rate": 3.895169771634616e-05, "loss": 0.991, "step": 5887 }, { "epoch": 1.77, "learning_rate": 3.894981971153846e-05, "loss": 1.0215, "step": 5888 }, { "epoch": 1.77, "learning_rate": 3.8947941706730775e-05, "loss": 1.1668, "step": 5889 }, { "epoch": 1.77, "learning_rate": 3.894606370192308e-05, "loss": 1.2227, "step": 5890 }, { "epoch": 1.77, "learning_rate": 3.8944185697115384e-05, "loss": 1.3283, "step": 5891 }, { "epoch": 1.77, "learning_rate": 3.8942307692307696e-05, "loss": 1.1716, "step": 5892 }, { "epoch": 1.77, "learning_rate": 3.89404296875e-05, "loss": 1.1702, "step": 5893 }, { "epoch": 1.77, "learning_rate": 3.893855168269231e-05, "loss": 1.2442, "step": 5894 }, { "epoch": 1.77, "learning_rate": 3.893667367788462e-05, "loss": 1.1864, "step": 5895 }, { "epoch": 1.77, "learning_rate": 3.893479567307692e-05, "loss": 0.8559, "step": 5896 }, { "epoch": 1.77, "learning_rate": 3.893291766826923e-05, "loss": 1.2996, "step": 5897 }, { "epoch": 1.77, "learning_rate": 3.893103966346154e-05, "loss": 1.0824, "step": 5898 }, { "epoch": 1.77, "learning_rate": 3.892916165865384e-05, "loss": 1.1099, "step": 5899 }, { "epoch": 1.77, "learning_rate": 3.892728365384616e-05, "loss": 1.4029, "step": 5900 }, { "epoch": 1.77, "learning_rate": 3.8925405649038466e-05, "loss": 1.0139, "step": 5901 }, { "epoch": 1.77, "learning_rate": 3.892352764423077e-05, "loss": 1.0318, "step": 5902 }, { "epoch": 1.77, "learning_rate": 3.892164963942308e-05, "loss": 1.0704, "step": 5903 }, { "epoch": 1.77, "learning_rate": 3.891977163461539e-05, "loss": 1.1042, "step": 5904 }, { "epoch": 1.77, "learning_rate": 3.891789362980769e-05, "loss": 1.3143, "step": 5905 }, { "epoch": 1.77, "learning_rate": 3.8916015625000003e-05, "loss": 1.0479, "step": 5906 }, { "epoch": 1.77, "learning_rate": 3.891413762019231e-05, "loss": 1.1329, "step": 5907 }, { "epoch": 1.78, "learning_rate": 3.891225961538462e-05, "loss": 1.1331, "step": 5908 }, { "epoch": 1.78, "learning_rate": 3.8910381610576924e-05, "loss": 1.077, "step": 5909 }, { "epoch": 1.78, "learning_rate": 3.890850360576923e-05, "loss": 1.1492, "step": 5910 }, { "epoch": 1.78, "learning_rate": 3.890662560096154e-05, "loss": 1.2741, "step": 5911 }, { "epoch": 1.78, "learning_rate": 3.8904747596153846e-05, "loss": 1.0649, "step": 5912 }, { "epoch": 1.78, "learning_rate": 3.890286959134616e-05, "loss": 1.0692, "step": 5913 }, { "epoch": 1.78, "learning_rate": 3.890099158653847e-05, "loss": 1.0411, "step": 5914 }, { "epoch": 1.78, "learning_rate": 3.8899113581730773e-05, "loss": 0.9576, "step": 5915 }, { "epoch": 1.78, "learning_rate": 3.889723557692308e-05, "loss": 1.1677, "step": 5916 }, { "epoch": 1.78, "learning_rate": 3.889535757211539e-05, "loss": 1.3458, "step": 5917 }, { "epoch": 1.78, "learning_rate": 3.8893479567307695e-05, "loss": 1.1414, "step": 5918 }, { "epoch": 1.78, "learning_rate": 3.88916015625e-05, "loss": 1.1182, "step": 5919 }, { "epoch": 1.78, "learning_rate": 3.888972355769231e-05, "loss": 1.0595, "step": 5920 }, { "epoch": 1.78, "learning_rate": 3.8887845552884616e-05, "loss": 1.2468, "step": 5921 }, { "epoch": 1.78, "learning_rate": 3.888596754807692e-05, "loss": 1.2392, "step": 5922 }, { "epoch": 1.78, "learning_rate": 3.888408954326923e-05, "loss": 1.4278, "step": 5923 }, { "epoch": 1.78, "learning_rate": 3.888221153846154e-05, "loss": 1.0318, "step": 5924 }, { "epoch": 1.78, "learning_rate": 3.888033353365385e-05, "loss": 1.0922, "step": 5925 }, { "epoch": 1.78, "learning_rate": 3.887845552884616e-05, "loss": 0.9942, "step": 5926 }, { "epoch": 1.78, "learning_rate": 3.8876577524038465e-05, "loss": 1.0504, "step": 5927 }, { "epoch": 1.78, "learning_rate": 3.8874699519230776e-05, "loss": 1.0729, "step": 5928 }, { "epoch": 1.78, "learning_rate": 3.887282151442308e-05, "loss": 1.1322, "step": 5929 }, { "epoch": 1.78, "learning_rate": 3.8870943509615386e-05, "loss": 1.2664, "step": 5930 }, { "epoch": 1.78, "learning_rate": 3.88690655048077e-05, "loss": 1.0991, "step": 5931 }, { "epoch": 1.78, "learning_rate": 3.88671875e-05, "loss": 1.1365, "step": 5932 }, { "epoch": 1.78, "learning_rate": 3.886530949519231e-05, "loss": 0.9936, "step": 5933 }, { "epoch": 1.78, "learning_rate": 3.886343149038462e-05, "loss": 1.2187, "step": 5934 }, { "epoch": 1.78, "learning_rate": 3.886155348557692e-05, "loss": 1.0595, "step": 5935 }, { "epoch": 1.78, "learning_rate": 3.885967548076923e-05, "loss": 1.1073, "step": 5936 }, { "epoch": 1.78, "learning_rate": 3.885779747596154e-05, "loss": 1.284, "step": 5937 }, { "epoch": 1.78, "learning_rate": 3.8855919471153844e-05, "loss": 1.0466, "step": 5938 }, { "epoch": 1.78, "learning_rate": 3.8854041466346156e-05, "loss": 1.5997, "step": 5939 }, { "epoch": 1.78, "learning_rate": 3.885216346153847e-05, "loss": 1.0629, "step": 5940 }, { "epoch": 1.79, "learning_rate": 3.885028545673077e-05, "loss": 1.2216, "step": 5941 }, { "epoch": 1.79, "learning_rate": 3.884840745192308e-05, "loss": 1.1121, "step": 5942 }, { "epoch": 1.79, "learning_rate": 3.884652944711539e-05, "loss": 1.0652, "step": 5943 }, { "epoch": 1.79, "learning_rate": 3.884465144230769e-05, "loss": 0.9915, "step": 5944 }, { "epoch": 1.79, "learning_rate": 3.8842773437500005e-05, "loss": 1.1557, "step": 5945 }, { "epoch": 1.79, "learning_rate": 3.884089543269231e-05, "loss": 1.1516, "step": 5946 }, { "epoch": 1.79, "learning_rate": 3.8839017427884614e-05, "loss": 1.0948, "step": 5947 }, { "epoch": 1.79, "learning_rate": 3.8837139423076926e-05, "loss": 1.3912, "step": 5948 }, { "epoch": 1.79, "learning_rate": 3.883526141826923e-05, "loss": 1.1557, "step": 5949 }, { "epoch": 1.79, "learning_rate": 3.8833383413461535e-05, "loss": 1.0661, "step": 5950 }, { "epoch": 1.79, "learning_rate": 3.883150540865385e-05, "loss": 1.2818, "step": 5951 }, { "epoch": 1.79, "learning_rate": 3.882962740384616e-05, "loss": 1.1614, "step": 5952 }, { "epoch": 1.79, "learning_rate": 3.882774939903846e-05, "loss": 1.0327, "step": 5953 }, { "epoch": 1.79, "learning_rate": 3.8825871394230775e-05, "loss": 0.8182, "step": 5954 }, { "epoch": 1.79, "learning_rate": 3.882399338942308e-05, "loss": 1.291, "step": 5955 }, { "epoch": 1.79, "learning_rate": 3.8822115384615384e-05, "loss": 1.1229, "step": 5956 }, { "epoch": 1.79, "learning_rate": 3.8820237379807696e-05, "loss": 1.1098, "step": 5957 }, { "epoch": 1.79, "learning_rate": 3.8818359375e-05, "loss": 1.1413, "step": 5958 }, { "epoch": 1.79, "learning_rate": 3.881648137019231e-05, "loss": 1.1665, "step": 5959 }, { "epoch": 1.79, "learning_rate": 3.881460336538462e-05, "loss": 1.111, "step": 5960 }, { "epoch": 1.79, "learning_rate": 3.881272536057692e-05, "loss": 1.2344, "step": 5961 }, { "epoch": 1.79, "learning_rate": 3.881084735576923e-05, "loss": 1.3215, "step": 5962 }, { "epoch": 1.79, "learning_rate": 3.880896935096154e-05, "loss": 1.1402, "step": 5963 }, { "epoch": 1.79, "learning_rate": 3.880709134615384e-05, "loss": 0.8527, "step": 5964 }, { "epoch": 1.79, "learning_rate": 3.880521334134616e-05, "loss": 1.0853, "step": 5965 }, { "epoch": 1.79, "learning_rate": 3.8803335336538466e-05, "loss": 1.0196, "step": 5966 }, { "epoch": 1.79, "learning_rate": 3.880145733173077e-05, "loss": 1.1452, "step": 5967 }, { "epoch": 1.79, "learning_rate": 3.879957932692308e-05, "loss": 1.0918, "step": 5968 }, { "epoch": 1.79, "learning_rate": 3.879770132211539e-05, "loss": 1.1431, "step": 5969 }, { "epoch": 1.79, "learning_rate": 3.879582331730769e-05, "loss": 1.1631, "step": 5970 }, { "epoch": 1.79, "learning_rate": 3.87939453125e-05, "loss": 1.0881, "step": 5971 }, { "epoch": 1.79, "learning_rate": 3.879206730769231e-05, "loss": 1.1707, "step": 5972 }, { "epoch": 1.79, "learning_rate": 3.879018930288462e-05, "loss": 1.2143, "step": 5973 }, { "epoch": 1.8, "learning_rate": 3.8788311298076924e-05, "loss": 0.9148, "step": 5974 }, { "epoch": 1.8, "learning_rate": 3.878643329326923e-05, "loss": 1.4749, "step": 5975 }, { "epoch": 1.8, "learning_rate": 3.878455528846154e-05, "loss": 1.3374, "step": 5976 }, { "epoch": 1.8, "learning_rate": 3.8782677283653845e-05, "loss": 1.0696, "step": 5977 }, { "epoch": 1.8, "learning_rate": 3.878079927884616e-05, "loss": 1.1726, "step": 5978 }, { "epoch": 1.8, "learning_rate": 3.877892127403847e-05, "loss": 1.1867, "step": 5979 }, { "epoch": 1.8, "learning_rate": 3.877704326923077e-05, "loss": 1.1801, "step": 5980 }, { "epoch": 1.8, "learning_rate": 3.877516526442308e-05, "loss": 1.0956, "step": 5981 }, { "epoch": 1.8, "learning_rate": 3.877328725961539e-05, "loss": 1.3642, "step": 5982 }, { "epoch": 1.8, "learning_rate": 3.8771409254807694e-05, "loss": 1.0942, "step": 5983 }, { "epoch": 1.8, "learning_rate": 3.876953125e-05, "loss": 1.0555, "step": 5984 }, { "epoch": 1.8, "learning_rate": 3.876765324519231e-05, "loss": 1.1046, "step": 5985 }, { "epoch": 1.8, "learning_rate": 3.8765775240384616e-05, "loss": 1.3222, "step": 5986 }, { "epoch": 1.8, "learning_rate": 3.876389723557692e-05, "loss": 1.0418, "step": 5987 }, { "epoch": 1.8, "learning_rate": 3.876201923076923e-05, "loss": 1.1732, "step": 5988 }, { "epoch": 1.8, "learning_rate": 3.8760141225961537e-05, "loss": 1.1588, "step": 5989 }, { "epoch": 1.8, "learning_rate": 3.875826322115385e-05, "loss": 1.1476, "step": 5990 }, { "epoch": 1.8, "learning_rate": 3.875638521634616e-05, "loss": 1.2788, "step": 5991 }, { "epoch": 1.8, "learning_rate": 3.8754507211538464e-05, "loss": 1.2264, "step": 5992 }, { "epoch": 1.8, "learning_rate": 3.8752629206730776e-05, "loss": 1.1359, "step": 5993 }, { "epoch": 1.8, "learning_rate": 3.875075120192308e-05, "loss": 1.1506, "step": 5994 }, { "epoch": 1.8, "learning_rate": 3.8748873197115386e-05, "loss": 1.1288, "step": 5995 }, { "epoch": 1.8, "learning_rate": 3.87469951923077e-05, "loss": 1.1624, "step": 5996 }, { "epoch": 1.8, "learning_rate": 3.87451171875e-05, "loss": 1.188, "step": 5997 }, { "epoch": 1.8, "learning_rate": 3.874323918269231e-05, "loss": 1.3253, "step": 5998 }, { "epoch": 1.8, "learning_rate": 3.874136117788462e-05, "loss": 1.1081, "step": 5999 }, { "epoch": 1.8, "learning_rate": 3.873948317307692e-05, "loss": 1.2115, "step": 6000 }, { "epoch": 1.8, "learning_rate": 3.873760516826923e-05, "loss": 1.0973, "step": 6001 }, { "epoch": 1.8, "learning_rate": 3.873572716346154e-05, "loss": 1.419, "step": 6002 }, { "epoch": 1.8, "learning_rate": 3.8733849158653844e-05, "loss": 1.2378, "step": 6003 }, { "epoch": 1.8, "learning_rate": 3.8731971153846156e-05, "loss": 1.0654, "step": 6004 }, { "epoch": 1.8, "learning_rate": 3.873009314903847e-05, "loss": 1.3647, "step": 6005 }, { "epoch": 1.8, "learning_rate": 3.872821514423077e-05, "loss": 0.9325, "step": 6006 }, { "epoch": 1.8, "learning_rate": 3.872633713942308e-05, "loss": 1.1788, "step": 6007 }, { "epoch": 1.81, "learning_rate": 3.872445913461539e-05, "loss": 1.0584, "step": 6008 }, { "epoch": 1.81, "learning_rate": 3.872258112980769e-05, "loss": 1.1399, "step": 6009 }, { "epoch": 1.81, "learning_rate": 3.8720703125000005e-05, "loss": 1.0841, "step": 6010 }, { "epoch": 1.81, "learning_rate": 3.871882512019231e-05, "loss": 1.1267, "step": 6011 }, { "epoch": 1.81, "learning_rate": 3.8716947115384614e-05, "loss": 1.1491, "step": 6012 }, { "epoch": 1.81, "learning_rate": 3.8715069110576926e-05, "loss": 0.9426, "step": 6013 }, { "epoch": 1.81, "learning_rate": 3.871319110576923e-05, "loss": 1.1631, "step": 6014 }, { "epoch": 1.81, "learning_rate": 3.8711313100961535e-05, "loss": 1.3663, "step": 6015 }, { "epoch": 1.81, "learning_rate": 3.870943509615385e-05, "loss": 1.2436, "step": 6016 }, { "epoch": 1.81, "learning_rate": 3.870755709134616e-05, "loss": 1.1294, "step": 6017 }, { "epoch": 1.81, "learning_rate": 3.870567908653846e-05, "loss": 1.1011, "step": 6018 }, { "epoch": 1.81, "learning_rate": 3.8703801081730775e-05, "loss": 1.4093, "step": 6019 }, { "epoch": 1.81, "learning_rate": 3.870192307692308e-05, "loss": 1.0878, "step": 6020 }, { "epoch": 1.81, "learning_rate": 3.8700045072115384e-05, "loss": 1.2909, "step": 6021 }, { "epoch": 1.81, "learning_rate": 3.8698167067307696e-05, "loss": 0.8587, "step": 6022 }, { "epoch": 1.81, "learning_rate": 3.86962890625e-05, "loss": 1.3916, "step": 6023 }, { "epoch": 1.81, "learning_rate": 3.869441105769231e-05, "loss": 1.1886, "step": 6024 }, { "epoch": 1.81, "learning_rate": 3.869253305288462e-05, "loss": 1.1817, "step": 6025 }, { "epoch": 1.81, "learning_rate": 3.869065504807692e-05, "loss": 1.1584, "step": 6026 }, { "epoch": 1.81, "learning_rate": 3.868877704326923e-05, "loss": 1.0956, "step": 6027 }, { "epoch": 1.81, "learning_rate": 3.868689903846154e-05, "loss": 1.0373, "step": 6028 }, { "epoch": 1.81, "learning_rate": 3.868502103365384e-05, "loss": 1.2636, "step": 6029 }, { "epoch": 1.81, "learning_rate": 3.868314302884616e-05, "loss": 0.9188, "step": 6030 }, { "epoch": 1.81, "learning_rate": 3.8681265024038466e-05, "loss": 1.1728, "step": 6031 }, { "epoch": 1.81, "learning_rate": 3.867938701923077e-05, "loss": 1.0842, "step": 6032 }, { "epoch": 1.81, "learning_rate": 3.867750901442308e-05, "loss": 0.9956, "step": 6033 }, { "epoch": 1.81, "learning_rate": 3.867563100961539e-05, "loss": 1.2785, "step": 6034 }, { "epoch": 1.81, "learning_rate": 3.867375300480769e-05, "loss": 1.248, "step": 6035 }, { "epoch": 1.81, "learning_rate": 3.8671875e-05, "loss": 1.2063, "step": 6036 }, { "epoch": 1.81, "learning_rate": 3.866999699519231e-05, "loss": 1.2538, "step": 6037 }, { "epoch": 1.81, "learning_rate": 3.866811899038462e-05, "loss": 1.1093, "step": 6038 }, { "epoch": 1.81, "learning_rate": 3.8666240985576924e-05, "loss": 1.199, "step": 6039 }, { "epoch": 1.81, "learning_rate": 3.866436298076923e-05, "loss": 1.0962, "step": 6040 }, { "epoch": 1.82, "learning_rate": 3.866248497596154e-05, "loss": 1.1487, "step": 6041 }, { "epoch": 1.82, "learning_rate": 3.8660606971153845e-05, "loss": 1.4366, "step": 6042 }, { "epoch": 1.82, "learning_rate": 3.865872896634616e-05, "loss": 1.079, "step": 6043 }, { "epoch": 1.82, "learning_rate": 3.865685096153847e-05, "loss": 0.9844, "step": 6044 }, { "epoch": 1.82, "learning_rate": 3.865497295673077e-05, "loss": 1.1857, "step": 6045 }, { "epoch": 1.82, "learning_rate": 3.865309495192308e-05, "loss": 1.3849, "step": 6046 }, { "epoch": 1.82, "learning_rate": 3.865121694711539e-05, "loss": 1.1157, "step": 6047 }, { "epoch": 1.82, "learning_rate": 3.8649338942307694e-05, "loss": 0.8957, "step": 6048 }, { "epoch": 1.82, "learning_rate": 3.86474609375e-05, "loss": 1.1642, "step": 6049 }, { "epoch": 1.82, "learning_rate": 3.864558293269231e-05, "loss": 0.8987, "step": 6050 }, { "epoch": 1.82, "learning_rate": 3.8643704927884615e-05, "loss": 1.1796, "step": 6051 }, { "epoch": 1.82, "learning_rate": 3.864182692307693e-05, "loss": 1.048, "step": 6052 }, { "epoch": 1.82, "learning_rate": 3.863994891826923e-05, "loss": 0.9118, "step": 6053 }, { "epoch": 1.82, "learning_rate": 3.8638070913461536e-05, "loss": 1.3948, "step": 6054 }, { "epoch": 1.82, "learning_rate": 3.863619290865385e-05, "loss": 1.3202, "step": 6055 }, { "epoch": 1.82, "learning_rate": 3.863431490384616e-05, "loss": 1.0531, "step": 6056 }, { "epoch": 1.82, "learning_rate": 3.8632436899038464e-05, "loss": 1.1925, "step": 6057 }, { "epoch": 1.82, "learning_rate": 3.8630558894230776e-05, "loss": 1.2792, "step": 6058 }, { "epoch": 1.82, "learning_rate": 3.862868088942308e-05, "loss": 1.2214, "step": 6059 }, { "epoch": 1.82, "learning_rate": 3.8626802884615385e-05, "loss": 1.1351, "step": 6060 }, { "epoch": 1.82, "learning_rate": 3.86249248798077e-05, "loss": 1.5383, "step": 6061 }, { "epoch": 1.82, "learning_rate": 3.8623046875e-05, "loss": 1.2184, "step": 6062 }, { "epoch": 1.82, "learning_rate": 3.8621168870192307e-05, "loss": 1.2601, "step": 6063 }, { "epoch": 1.82, "learning_rate": 3.861929086538462e-05, "loss": 1.2672, "step": 6064 }, { "epoch": 1.82, "learning_rate": 3.861741286057692e-05, "loss": 1.1404, "step": 6065 }, { "epoch": 1.82, "learning_rate": 3.861553485576923e-05, "loss": 0.9745, "step": 6066 }, { "epoch": 1.82, "learning_rate": 3.861365685096154e-05, "loss": 1.2207, "step": 6067 }, { "epoch": 1.82, "learning_rate": 3.8611778846153844e-05, "loss": 1.3091, "step": 6068 }, { "epoch": 1.82, "learning_rate": 3.8609900841346156e-05, "loss": 1.0313, "step": 6069 }, { "epoch": 1.82, "learning_rate": 3.860802283653847e-05, "loss": 0.9045, "step": 6070 }, { "epoch": 1.82, "learning_rate": 3.860614483173077e-05, "loss": 1.1592, "step": 6071 }, { "epoch": 1.82, "learning_rate": 3.860426682692308e-05, "loss": 1.4041, "step": 6072 }, { "epoch": 1.82, "learning_rate": 3.860238882211539e-05, "loss": 1.0925, "step": 6073 }, { "epoch": 1.83, "learning_rate": 3.860051081730769e-05, "loss": 1.0572, "step": 6074 }, { "epoch": 1.83, "learning_rate": 3.8598632812500004e-05, "loss": 1.083, "step": 6075 }, { "epoch": 1.83, "learning_rate": 3.859675480769231e-05, "loss": 0.9028, "step": 6076 }, { "epoch": 1.83, "learning_rate": 3.8594876802884614e-05, "loss": 1.1671, "step": 6077 }, { "epoch": 1.83, "learning_rate": 3.8592998798076926e-05, "loss": 1.3304, "step": 6078 }, { "epoch": 1.83, "learning_rate": 3.859112079326923e-05, "loss": 1.207, "step": 6079 }, { "epoch": 1.83, "learning_rate": 3.8589242788461535e-05, "loss": 0.9318, "step": 6080 }, { "epoch": 1.83, "learning_rate": 3.858736478365385e-05, "loss": 1.2624, "step": 6081 }, { "epoch": 1.83, "learning_rate": 3.858548677884616e-05, "loss": 1.2102, "step": 6082 }, { "epoch": 1.83, "learning_rate": 3.858360877403846e-05, "loss": 1.1856, "step": 6083 }, { "epoch": 1.83, "learning_rate": 3.8581730769230775e-05, "loss": 1.1982, "step": 6084 }, { "epoch": 1.83, "learning_rate": 3.857985276442308e-05, "loss": 1.1606, "step": 6085 }, { "epoch": 1.83, "learning_rate": 3.8577974759615384e-05, "loss": 1.317, "step": 6086 }, { "epoch": 1.83, "learning_rate": 3.8576096754807696e-05, "loss": 1.0856, "step": 6087 }, { "epoch": 1.83, "learning_rate": 3.857421875e-05, "loss": 1.3073, "step": 6088 }, { "epoch": 1.83, "learning_rate": 3.857234074519231e-05, "loss": 1.0676, "step": 6089 }, { "epoch": 1.83, "learning_rate": 3.857046274038462e-05, "loss": 1.2298, "step": 6090 }, { "epoch": 1.83, "learning_rate": 3.856858473557692e-05, "loss": 1.191, "step": 6091 }, { "epoch": 1.83, "learning_rate": 3.856670673076923e-05, "loss": 1.1883, "step": 6092 }, { "epoch": 1.83, "learning_rate": 3.856482872596154e-05, "loss": 1.1622, "step": 6093 }, { "epoch": 1.83, "learning_rate": 3.856295072115384e-05, "loss": 1.0651, "step": 6094 }, { "epoch": 1.83, "learning_rate": 3.856107271634616e-05, "loss": 1.2625, "step": 6095 }, { "epoch": 1.83, "learning_rate": 3.8559194711538466e-05, "loss": 1.2032, "step": 6096 }, { "epoch": 1.83, "learning_rate": 3.855731670673077e-05, "loss": 1.4011, "step": 6097 }, { "epoch": 1.83, "learning_rate": 3.855543870192308e-05, "loss": 1.0557, "step": 6098 }, { "epoch": 1.83, "learning_rate": 3.855356069711539e-05, "loss": 1.4609, "step": 6099 }, { "epoch": 1.83, "learning_rate": 3.855168269230769e-05, "loss": 1.0513, "step": 6100 }, { "epoch": 1.83, "learning_rate": 3.85498046875e-05, "loss": 0.9059, "step": 6101 }, { "epoch": 1.83, "learning_rate": 3.854792668269231e-05, "loss": 1.0919, "step": 6102 }, { "epoch": 1.83, "learning_rate": 3.854604867788462e-05, "loss": 1.0499, "step": 6103 }, { "epoch": 1.83, "learning_rate": 3.8544170673076924e-05, "loss": 1.1674, "step": 6104 }, { "epoch": 1.83, "learning_rate": 3.854229266826923e-05, "loss": 1.2272, "step": 6105 }, { "epoch": 1.83, "learning_rate": 3.854041466346154e-05, "loss": 1.3688, "step": 6106 }, { "epoch": 1.84, "learning_rate": 3.8538536658653845e-05, "loss": 1.1912, "step": 6107 }, { "epoch": 1.84, "learning_rate": 3.853665865384616e-05, "loss": 1.3148, "step": 6108 }, { "epoch": 1.84, "learning_rate": 3.853478064903847e-05, "loss": 0.8799, "step": 6109 }, { "epoch": 1.84, "learning_rate": 3.853290264423077e-05, "loss": 1.0312, "step": 6110 }, { "epoch": 1.84, "learning_rate": 3.853102463942308e-05, "loss": 1.4056, "step": 6111 }, { "epoch": 1.84, "learning_rate": 3.852914663461539e-05, "loss": 1.383, "step": 6112 }, { "epoch": 1.84, "learning_rate": 3.8527268629807694e-05, "loss": 1.0087, "step": 6113 }, { "epoch": 1.84, "learning_rate": 3.8525390625e-05, "loss": 1.014, "step": 6114 }, { "epoch": 1.84, "learning_rate": 3.852351262019231e-05, "loss": 1.1831, "step": 6115 }, { "epoch": 1.84, "learning_rate": 3.8521634615384615e-05, "loss": 1.2176, "step": 6116 }, { "epoch": 1.84, "learning_rate": 3.851975661057693e-05, "loss": 1.0252, "step": 6117 }, { "epoch": 1.84, "learning_rate": 3.851787860576923e-05, "loss": 0.9591, "step": 6118 }, { "epoch": 1.84, "learning_rate": 3.8516000600961536e-05, "loss": 1.0758, "step": 6119 }, { "epoch": 1.84, "learning_rate": 3.851412259615385e-05, "loss": 1.2649, "step": 6120 }, { "epoch": 1.84, "learning_rate": 3.851224459134616e-05, "loss": 1.1257, "step": 6121 }, { "epoch": 1.84, "learning_rate": 3.8510366586538464e-05, "loss": 1.07, "step": 6122 }, { "epoch": 1.84, "learning_rate": 3.8508488581730776e-05, "loss": 1.1053, "step": 6123 }, { "epoch": 1.84, "learning_rate": 3.850661057692308e-05, "loss": 1.3114, "step": 6124 }, { "epoch": 1.84, "learning_rate": 3.8504732572115385e-05, "loss": 1.1216, "step": 6125 }, { "epoch": 1.84, "learning_rate": 3.85028545673077e-05, "loss": 1.1298, "step": 6126 }, { "epoch": 1.84, "learning_rate": 3.85009765625e-05, "loss": 1.3619, "step": 6127 }, { "epoch": 1.84, "learning_rate": 3.8499098557692306e-05, "loss": 1.1236, "step": 6128 }, { "epoch": 1.84, "learning_rate": 3.849722055288462e-05, "loss": 0.9759, "step": 6129 }, { "epoch": 1.84, "learning_rate": 3.849534254807692e-05, "loss": 1.0921, "step": 6130 }, { "epoch": 1.84, "learning_rate": 3.849346454326923e-05, "loss": 1.1184, "step": 6131 }, { "epoch": 1.84, "learning_rate": 3.849158653846154e-05, "loss": 1.2435, "step": 6132 }, { "epoch": 1.84, "learning_rate": 3.8489708533653844e-05, "loss": 1.1047, "step": 6133 }, { "epoch": 1.84, "learning_rate": 3.8487830528846155e-05, "loss": 1.01, "step": 6134 }, { "epoch": 1.84, "learning_rate": 3.848595252403847e-05, "loss": 1.1948, "step": 6135 }, { "epoch": 1.84, "learning_rate": 3.848407451923077e-05, "loss": 0.9732, "step": 6136 }, { "epoch": 1.84, "learning_rate": 3.848219651442308e-05, "loss": 1.2232, "step": 6137 }, { "epoch": 1.84, "learning_rate": 3.848031850961539e-05, "loss": 1.1844, "step": 6138 }, { "epoch": 1.84, "learning_rate": 3.847844050480769e-05, "loss": 1.2596, "step": 6139 }, { "epoch": 1.84, "learning_rate": 3.8476562500000004e-05, "loss": 0.9164, "step": 6140 }, { "epoch": 1.85, "learning_rate": 3.847468449519231e-05, "loss": 1.306, "step": 6141 }, { "epoch": 1.85, "learning_rate": 3.8472806490384614e-05, "loss": 1.0227, "step": 6142 }, { "epoch": 1.85, "learning_rate": 3.8470928485576925e-05, "loss": 1.0698, "step": 6143 }, { "epoch": 1.85, "learning_rate": 3.846905048076923e-05, "loss": 0.8319, "step": 6144 }, { "epoch": 1.85, "learning_rate": 3.8467172475961535e-05, "loss": 1.1642, "step": 6145 }, { "epoch": 1.85, "learning_rate": 3.8465294471153847e-05, "loss": 1.0388, "step": 6146 }, { "epoch": 1.85, "learning_rate": 3.846341646634616e-05, "loss": 1.2704, "step": 6147 }, { "epoch": 1.85, "learning_rate": 3.846153846153846e-05, "loss": 1.3259, "step": 6148 }, { "epoch": 1.85, "learning_rate": 3.8459660456730774e-05, "loss": 1.2743, "step": 6149 }, { "epoch": 1.85, "learning_rate": 3.845778245192308e-05, "loss": 1.1454, "step": 6150 }, { "epoch": 1.85, "learning_rate": 3.8455904447115384e-05, "loss": 0.9724, "step": 6151 }, { "epoch": 1.85, "learning_rate": 3.8454026442307695e-05, "loss": 1.0515, "step": 6152 }, { "epoch": 1.85, "learning_rate": 3.84521484375e-05, "loss": 0.8467, "step": 6153 }, { "epoch": 1.85, "learning_rate": 3.845027043269231e-05, "loss": 0.9508, "step": 6154 }, { "epoch": 1.85, "learning_rate": 3.8448392427884617e-05, "loss": 1.1794, "step": 6155 }, { "epoch": 1.85, "learning_rate": 3.844651442307692e-05, "loss": 1.0073, "step": 6156 }, { "epoch": 1.85, "learning_rate": 3.844463641826923e-05, "loss": 1.0369, "step": 6157 }, { "epoch": 1.85, "learning_rate": 3.844275841346154e-05, "loss": 1.1008, "step": 6158 }, { "epoch": 1.85, "learning_rate": 3.844088040865384e-05, "loss": 1.331, "step": 6159 }, { "epoch": 1.85, "learning_rate": 3.843900240384616e-05, "loss": 0.9933, "step": 6160 }, { "epoch": 1.85, "learning_rate": 3.8437124399038466e-05, "loss": 0.8109, "step": 6161 }, { "epoch": 1.85, "learning_rate": 3.843524639423077e-05, "loss": 0.9289, "step": 6162 }, { "epoch": 1.85, "learning_rate": 3.843336838942308e-05, "loss": 1.061, "step": 6163 }, { "epoch": 1.85, "learning_rate": 3.843149038461539e-05, "loss": 1.2269, "step": 6164 }, { "epoch": 1.85, "learning_rate": 3.842961237980769e-05, "loss": 1.1563, "step": 6165 }, { "epoch": 1.85, "learning_rate": 3.8427734375e-05, "loss": 1.1528, "step": 6166 }, { "epoch": 1.85, "learning_rate": 3.842585637019231e-05, "loss": 0.9225, "step": 6167 }, { "epoch": 1.85, "learning_rate": 3.842397836538462e-05, "loss": 1.1975, "step": 6168 }, { "epoch": 1.85, "learning_rate": 3.8422100360576924e-05, "loss": 1.2749, "step": 6169 }, { "epoch": 1.85, "learning_rate": 3.842022235576923e-05, "loss": 1.1155, "step": 6170 }, { "epoch": 1.85, "learning_rate": 3.841834435096154e-05, "loss": 1.2507, "step": 6171 }, { "epoch": 1.85, "learning_rate": 3.8416466346153845e-05, "loss": 0.8342, "step": 6172 }, { "epoch": 1.85, "learning_rate": 3.841458834134616e-05, "loss": 1.0375, "step": 6173 }, { "epoch": 1.86, "learning_rate": 3.841271033653847e-05, "loss": 0.9558, "step": 6174 }, { "epoch": 1.86, "learning_rate": 3.841083233173077e-05, "loss": 1.2862, "step": 6175 }, { "epoch": 1.86, "learning_rate": 3.840895432692308e-05, "loss": 1.1609, "step": 6176 }, { "epoch": 1.86, "learning_rate": 3.840707632211539e-05, "loss": 1.2586, "step": 6177 }, { "epoch": 1.86, "learning_rate": 3.8405198317307694e-05, "loss": 1.3907, "step": 6178 }, { "epoch": 1.86, "learning_rate": 3.84033203125e-05, "loss": 1.3191, "step": 6179 }, { "epoch": 1.86, "learning_rate": 3.840144230769231e-05, "loss": 0.9435, "step": 6180 }, { "epoch": 1.86, "learning_rate": 3.8399564302884615e-05, "loss": 1.1605, "step": 6181 }, { "epoch": 1.86, "learning_rate": 3.839768629807693e-05, "loss": 1.1729, "step": 6182 }, { "epoch": 1.86, "learning_rate": 3.839580829326923e-05, "loss": 1.2584, "step": 6183 }, { "epoch": 1.86, "learning_rate": 3.8393930288461536e-05, "loss": 1.1012, "step": 6184 }, { "epoch": 1.86, "learning_rate": 3.839205228365385e-05, "loss": 0.9784, "step": 6185 }, { "epoch": 1.86, "learning_rate": 3.839017427884616e-05, "loss": 1.2372, "step": 6186 }, { "epoch": 1.86, "learning_rate": 3.8388296274038464e-05, "loss": 1.3331, "step": 6187 }, { "epoch": 1.86, "learning_rate": 3.8386418269230776e-05, "loss": 1.1161, "step": 6188 }, { "epoch": 1.86, "learning_rate": 3.838454026442308e-05, "loss": 1.1817, "step": 6189 }, { "epoch": 1.86, "learning_rate": 3.8382662259615385e-05, "loss": 1.3864, "step": 6190 }, { "epoch": 1.86, "learning_rate": 3.83807842548077e-05, "loss": 1.0636, "step": 6191 }, { "epoch": 1.86, "learning_rate": 3.837890625e-05, "loss": 1.1458, "step": 6192 }, { "epoch": 1.86, "learning_rate": 3.8377028245192306e-05, "loss": 1.2164, "step": 6193 }, { "epoch": 1.86, "learning_rate": 3.837515024038462e-05, "loss": 1.1695, "step": 6194 }, { "epoch": 1.86, "learning_rate": 3.837327223557692e-05, "loss": 0.9715, "step": 6195 }, { "epoch": 1.86, "learning_rate": 3.8371394230769234e-05, "loss": 1.2002, "step": 6196 }, { "epoch": 1.86, "learning_rate": 3.836951622596154e-05, "loss": 1.1649, "step": 6197 }, { "epoch": 1.86, "learning_rate": 3.8367638221153844e-05, "loss": 1.0648, "step": 6198 }, { "epoch": 1.86, "learning_rate": 3.8365760216346155e-05, "loss": 1.0521, "step": 6199 }, { "epoch": 1.86, "learning_rate": 3.836388221153847e-05, "loss": 1.198, "step": 6200 }, { "epoch": 1.86, "learning_rate": 3.836200420673077e-05, "loss": 1.0415, "step": 6201 }, { "epoch": 1.86, "learning_rate": 3.836012620192308e-05, "loss": 1.0151, "step": 6202 }, { "epoch": 1.86, "learning_rate": 3.835824819711539e-05, "loss": 1.0616, "step": 6203 }, { "epoch": 1.86, "learning_rate": 3.835637019230769e-05, "loss": 1.1881, "step": 6204 }, { "epoch": 1.86, "learning_rate": 3.8354492187500004e-05, "loss": 1.2247, "step": 6205 }, { "epoch": 1.86, "learning_rate": 3.835261418269231e-05, "loss": 1.328, "step": 6206 }, { "epoch": 1.87, "learning_rate": 3.8350736177884614e-05, "loss": 1.1881, "step": 6207 }, { "epoch": 1.87, "learning_rate": 3.8348858173076925e-05, "loss": 1.1205, "step": 6208 }, { "epoch": 1.87, "learning_rate": 3.834698016826923e-05, "loss": 1.045, "step": 6209 }, { "epoch": 1.87, "learning_rate": 3.8345102163461535e-05, "loss": 1.0731, "step": 6210 }, { "epoch": 1.87, "learning_rate": 3.8343224158653846e-05, "loss": 1.0506, "step": 6211 }, { "epoch": 1.87, "learning_rate": 3.834134615384616e-05, "loss": 1.2212, "step": 6212 }, { "epoch": 1.87, "learning_rate": 3.833946814903846e-05, "loss": 1.2744, "step": 6213 }, { "epoch": 1.87, "learning_rate": 3.8337590144230774e-05, "loss": 1.0461, "step": 6214 }, { "epoch": 1.87, "learning_rate": 3.833571213942308e-05, "loss": 1.3173, "step": 6215 }, { "epoch": 1.87, "learning_rate": 3.833383413461539e-05, "loss": 1.1152, "step": 6216 }, { "epoch": 1.87, "learning_rate": 3.8331956129807695e-05, "loss": 1.2965, "step": 6217 }, { "epoch": 1.87, "learning_rate": 3.8330078125e-05, "loss": 0.9952, "step": 6218 }, { "epoch": 1.87, "learning_rate": 3.832820012019231e-05, "loss": 1.2519, "step": 6219 }, { "epoch": 1.87, "learning_rate": 3.8326322115384616e-05, "loss": 1.3242, "step": 6220 }, { "epoch": 1.87, "learning_rate": 3.832444411057692e-05, "loss": 1.0961, "step": 6221 }, { "epoch": 1.87, "learning_rate": 3.832256610576923e-05, "loss": 1.1521, "step": 6222 }, { "epoch": 1.87, "learning_rate": 3.832068810096154e-05, "loss": 1.2549, "step": 6223 }, { "epoch": 1.87, "learning_rate": 3.831881009615384e-05, "loss": 1.09, "step": 6224 }, { "epoch": 1.87, "learning_rate": 3.831693209134616e-05, "loss": 1.3542, "step": 6225 }, { "epoch": 1.87, "learning_rate": 3.8315054086538465e-05, "loss": 1.1888, "step": 6226 }, { "epoch": 1.87, "learning_rate": 3.831317608173077e-05, "loss": 1.1885, "step": 6227 }, { "epoch": 1.87, "learning_rate": 3.831129807692308e-05, "loss": 1.0882, "step": 6228 }, { "epoch": 1.87, "learning_rate": 3.8309420072115387e-05, "loss": 1.1176, "step": 6229 }, { "epoch": 1.87, "learning_rate": 3.830754206730769e-05, "loss": 1.1205, "step": 6230 }, { "epoch": 1.87, "learning_rate": 3.83056640625e-05, "loss": 1.0335, "step": 6231 }, { "epoch": 1.87, "learning_rate": 3.830378605769231e-05, "loss": 1.053, "step": 6232 }, { "epoch": 1.87, "learning_rate": 3.830190805288462e-05, "loss": 1.3072, "step": 6233 }, { "epoch": 1.87, "learning_rate": 3.8300030048076924e-05, "loss": 1.2851, "step": 6234 }, { "epoch": 1.87, "learning_rate": 3.829815204326923e-05, "loss": 1.0815, "step": 6235 }, { "epoch": 1.87, "learning_rate": 3.829627403846154e-05, "loss": 0.8032, "step": 6236 }, { "epoch": 1.87, "learning_rate": 3.8294396033653845e-05, "loss": 1.1602, "step": 6237 }, { "epoch": 1.87, "learning_rate": 3.8292518028846157e-05, "loss": 1.0262, "step": 6238 }, { "epoch": 1.87, "learning_rate": 3.829064002403847e-05, "loss": 1.1288, "step": 6239 }, { "epoch": 1.88, "learning_rate": 3.828876201923077e-05, "loss": 1.4006, "step": 6240 }, { "epoch": 1.88, "learning_rate": 3.828688401442308e-05, "loss": 1.1902, "step": 6241 }, { "epoch": 1.88, "learning_rate": 3.828500600961539e-05, "loss": 1.2621, "step": 6242 }, { "epoch": 1.88, "learning_rate": 3.8283128004807694e-05, "loss": 1.0921, "step": 6243 }, { "epoch": 1.88, "learning_rate": 3.828125e-05, "loss": 0.9416, "step": 6244 }, { "epoch": 1.88, "learning_rate": 3.827937199519231e-05, "loss": 1.1719, "step": 6245 }, { "epoch": 1.88, "learning_rate": 3.8277493990384615e-05, "loss": 1.261, "step": 6246 }, { "epoch": 1.88, "learning_rate": 3.827561598557693e-05, "loss": 1.2177, "step": 6247 }, { "epoch": 1.88, "learning_rate": 3.827373798076923e-05, "loss": 1.2364, "step": 6248 }, { "epoch": 1.88, "learning_rate": 3.8271859975961536e-05, "loss": 1.0696, "step": 6249 }, { "epoch": 1.88, "learning_rate": 3.826998197115385e-05, "loss": 0.8523, "step": 6250 }, { "epoch": 1.88, "learning_rate": 3.826810396634616e-05, "loss": 1.1589, "step": 6251 }, { "epoch": 1.88, "learning_rate": 3.8266225961538464e-05, "loss": 1.1242, "step": 6252 }, { "epoch": 1.88, "learning_rate": 3.8264347956730776e-05, "loss": 1.1402, "step": 6253 }, { "epoch": 1.88, "learning_rate": 3.826246995192308e-05, "loss": 1.3442, "step": 6254 }, { "epoch": 1.88, "learning_rate": 3.8260591947115385e-05, "loss": 1.2304, "step": 6255 }, { "epoch": 1.88, "learning_rate": 3.82587139423077e-05, "loss": 1.0038, "step": 6256 }, { "epoch": 1.88, "learning_rate": 3.82568359375e-05, "loss": 1.2757, "step": 6257 }, { "epoch": 1.88, "learning_rate": 3.8254957932692306e-05, "loss": 1.239, "step": 6258 }, { "epoch": 1.88, "learning_rate": 3.825307992788462e-05, "loss": 0.8521, "step": 6259 }, { "epoch": 1.88, "learning_rate": 3.825120192307692e-05, "loss": 0.9478, "step": 6260 }, { "epoch": 1.88, "learning_rate": 3.8249323918269234e-05, "loss": 1.0852, "step": 6261 }, { "epoch": 1.88, "learning_rate": 3.824744591346154e-05, "loss": 1.2214, "step": 6262 }, { "epoch": 1.88, "learning_rate": 3.8245567908653844e-05, "loss": 1.0801, "step": 6263 }, { "epoch": 1.88, "learning_rate": 3.8243689903846155e-05, "loss": 1.1732, "step": 6264 }, { "epoch": 1.88, "learning_rate": 3.824181189903847e-05, "loss": 1.3238, "step": 6265 }, { "epoch": 1.88, "learning_rate": 3.823993389423077e-05, "loss": 1.2298, "step": 6266 }, { "epoch": 1.88, "learning_rate": 3.823805588942308e-05, "loss": 1.0248, "step": 6267 }, { "epoch": 1.88, "learning_rate": 3.823617788461539e-05, "loss": 1.2275, "step": 6268 }, { "epoch": 1.88, "learning_rate": 3.823429987980769e-05, "loss": 1.0258, "step": 6269 }, { "epoch": 1.88, "learning_rate": 3.8232421875000004e-05, "loss": 1.2185, "step": 6270 }, { "epoch": 1.88, "learning_rate": 3.823054387019231e-05, "loss": 0.9346, "step": 6271 }, { "epoch": 1.88, "learning_rate": 3.8228665865384614e-05, "loss": 1.1706, "step": 6272 }, { "epoch": 1.88, "learning_rate": 3.8226787860576925e-05, "loss": 1.172, "step": 6273 }, { "epoch": 1.89, "learning_rate": 3.822490985576923e-05, "loss": 0.9752, "step": 6274 }, { "epoch": 1.89, "learning_rate": 3.8223031850961535e-05, "loss": 1.1188, "step": 6275 }, { "epoch": 1.89, "learning_rate": 3.8221153846153846e-05, "loss": 0.9387, "step": 6276 }, { "epoch": 1.89, "learning_rate": 3.821927584134616e-05, "loss": 0.8118, "step": 6277 }, { "epoch": 1.89, "learning_rate": 3.821739783653846e-05, "loss": 1.1526, "step": 6278 }, { "epoch": 1.89, "learning_rate": 3.8215519831730774e-05, "loss": 0.9556, "step": 6279 }, { "epoch": 1.89, "learning_rate": 3.821364182692308e-05, "loss": 0.9693, "step": 6280 }, { "epoch": 1.89, "learning_rate": 3.821176382211539e-05, "loss": 1.4135, "step": 6281 }, { "epoch": 1.89, "learning_rate": 3.8209885817307695e-05, "loss": 1.1329, "step": 6282 }, { "epoch": 1.89, "learning_rate": 3.82080078125e-05, "loss": 1.0318, "step": 6283 }, { "epoch": 1.89, "learning_rate": 3.820612980769231e-05, "loss": 1.0154, "step": 6284 }, { "epoch": 1.89, "learning_rate": 3.8204251802884616e-05, "loss": 1.2198, "step": 6285 }, { "epoch": 1.89, "learning_rate": 3.820237379807692e-05, "loss": 1.1036, "step": 6286 }, { "epoch": 1.89, "learning_rate": 3.820049579326923e-05, "loss": 1.1679, "step": 6287 }, { "epoch": 1.89, "learning_rate": 3.819861778846154e-05, "loss": 1.201, "step": 6288 }, { "epoch": 1.89, "learning_rate": 3.819673978365384e-05, "loss": 0.9755, "step": 6289 }, { "epoch": 1.89, "learning_rate": 3.819486177884616e-05, "loss": 0.9174, "step": 6290 }, { "epoch": 1.89, "learning_rate": 3.8192983774038465e-05, "loss": 1.1942, "step": 6291 }, { "epoch": 1.89, "learning_rate": 3.819110576923077e-05, "loss": 1.1919, "step": 6292 }, { "epoch": 1.89, "learning_rate": 3.818922776442308e-05, "loss": 1.0761, "step": 6293 }, { "epoch": 1.89, "learning_rate": 3.8187349759615386e-05, "loss": 1.035, "step": 6294 }, { "epoch": 1.89, "learning_rate": 3.818547175480769e-05, "loss": 1.2162, "step": 6295 }, { "epoch": 1.89, "learning_rate": 3.818359375e-05, "loss": 1.3472, "step": 6296 }, { "epoch": 1.89, "learning_rate": 3.818171574519231e-05, "loss": 1.2068, "step": 6297 }, { "epoch": 1.89, "learning_rate": 3.817983774038462e-05, "loss": 1.1871, "step": 6298 }, { "epoch": 1.89, "learning_rate": 3.8177959735576924e-05, "loss": 1.321, "step": 6299 }, { "epoch": 1.89, "learning_rate": 3.817608173076923e-05, "loss": 1.2454, "step": 6300 }, { "epoch": 1.89, "learning_rate": 3.817420372596154e-05, "loss": 1.0388, "step": 6301 }, { "epoch": 1.89, "learning_rate": 3.8172325721153845e-05, "loss": 1.0497, "step": 6302 }, { "epoch": 1.89, "learning_rate": 3.8170447716346156e-05, "loss": 1.362, "step": 6303 }, { "epoch": 1.89, "learning_rate": 3.816856971153847e-05, "loss": 1.2678, "step": 6304 }, { "epoch": 1.89, "learning_rate": 3.816669170673077e-05, "loss": 0.9268, "step": 6305 }, { "epoch": 1.89, "learning_rate": 3.816481370192308e-05, "loss": 1.3133, "step": 6306 }, { "epoch": 1.9, "learning_rate": 3.816293569711539e-05, "loss": 1.1956, "step": 6307 }, { "epoch": 1.9, "learning_rate": 3.8161057692307694e-05, "loss": 0.9311, "step": 6308 }, { "epoch": 1.9, "learning_rate": 3.81591796875e-05, "loss": 0.8644, "step": 6309 }, { "epoch": 1.9, "learning_rate": 3.815730168269231e-05, "loss": 1.1107, "step": 6310 }, { "epoch": 1.9, "learning_rate": 3.8155423677884615e-05, "loss": 1.4323, "step": 6311 }, { "epoch": 1.9, "learning_rate": 3.8153545673076927e-05, "loss": 1.1621, "step": 6312 }, { "epoch": 1.9, "learning_rate": 3.815166766826923e-05, "loss": 1.1703, "step": 6313 }, { "epoch": 1.9, "learning_rate": 3.8149789663461536e-05, "loss": 1.1584, "step": 6314 }, { "epoch": 1.9, "learning_rate": 3.814791165865385e-05, "loss": 1.104, "step": 6315 }, { "epoch": 1.9, "learning_rate": 3.814603365384616e-05, "loss": 1.0871, "step": 6316 }, { "epoch": 1.9, "learning_rate": 3.8144155649038464e-05, "loss": 1.21, "step": 6317 }, { "epoch": 1.9, "learning_rate": 3.8142277644230775e-05, "loss": 1.2004, "step": 6318 }, { "epoch": 1.9, "learning_rate": 3.814039963942308e-05, "loss": 1.1848, "step": 6319 }, { "epoch": 1.9, "learning_rate": 3.8138521634615385e-05, "loss": 1.3159, "step": 6320 }, { "epoch": 1.9, "learning_rate": 3.8136643629807697e-05, "loss": 1.1388, "step": 6321 }, { "epoch": 1.9, "learning_rate": 3.8134765625e-05, "loss": 1.1063, "step": 6322 }, { "epoch": 1.9, "learning_rate": 3.8132887620192306e-05, "loss": 1.013, "step": 6323 }, { "epoch": 1.9, "learning_rate": 3.813100961538462e-05, "loss": 1.1378, "step": 6324 }, { "epoch": 1.9, "learning_rate": 3.812913161057692e-05, "loss": 1.1609, "step": 6325 }, { "epoch": 1.9, "learning_rate": 3.8127253605769234e-05, "loss": 1.0937, "step": 6326 }, { "epoch": 1.9, "learning_rate": 3.812537560096154e-05, "loss": 0.9542, "step": 6327 }, { "epoch": 1.9, "learning_rate": 3.8123497596153844e-05, "loss": 1.2078, "step": 6328 }, { "epoch": 1.9, "learning_rate": 3.8121619591346155e-05, "loss": 1.2528, "step": 6329 }, { "epoch": 1.9, "learning_rate": 3.811974158653847e-05, "loss": 0.9204, "step": 6330 }, { "epoch": 1.9, "learning_rate": 3.811786358173077e-05, "loss": 0.9735, "step": 6331 }, { "epoch": 1.9, "learning_rate": 3.811598557692308e-05, "loss": 1.0445, "step": 6332 }, { "epoch": 1.9, "learning_rate": 3.811410757211539e-05, "loss": 0.9813, "step": 6333 }, { "epoch": 1.9, "learning_rate": 3.811222956730769e-05, "loss": 1.1954, "step": 6334 }, { "epoch": 1.9, "learning_rate": 3.8110351562500004e-05, "loss": 1.1337, "step": 6335 }, { "epoch": 1.9, "learning_rate": 3.810847355769231e-05, "loss": 1.0766, "step": 6336 }, { "epoch": 1.9, "learning_rate": 3.8106595552884614e-05, "loss": 1.1128, "step": 6337 }, { "epoch": 1.9, "learning_rate": 3.8104717548076925e-05, "loss": 1.1722, "step": 6338 }, { "epoch": 1.9, "learning_rate": 3.810283954326923e-05, "loss": 1.0968, "step": 6339 }, { "epoch": 1.91, "learning_rate": 3.810096153846154e-05, "loss": 1.1921, "step": 6340 }, { "epoch": 1.91, "learning_rate": 3.8099083533653846e-05, "loss": 1.1206, "step": 6341 }, { "epoch": 1.91, "learning_rate": 3.809720552884616e-05, "loss": 1.1618, "step": 6342 }, { "epoch": 1.91, "learning_rate": 3.809532752403846e-05, "loss": 1.2758, "step": 6343 }, { "epoch": 1.91, "learning_rate": 3.8093449519230774e-05, "loss": 1.1812, "step": 6344 }, { "epoch": 1.91, "learning_rate": 3.809157151442308e-05, "loss": 1.219, "step": 6345 }, { "epoch": 1.91, "learning_rate": 3.808969350961539e-05, "loss": 1.1738, "step": 6346 }, { "epoch": 1.91, "learning_rate": 3.8087815504807695e-05, "loss": 1.1419, "step": 6347 }, { "epoch": 1.91, "learning_rate": 3.80859375e-05, "loss": 1.1445, "step": 6348 }, { "epoch": 1.91, "learning_rate": 3.808405949519231e-05, "loss": 0.9718, "step": 6349 }, { "epoch": 1.91, "learning_rate": 3.8082181490384616e-05, "loss": 1.221, "step": 6350 }, { "epoch": 1.91, "learning_rate": 3.808030348557692e-05, "loss": 1.0056, "step": 6351 }, { "epoch": 1.91, "learning_rate": 3.807842548076923e-05, "loss": 1.3189, "step": 6352 }, { "epoch": 1.91, "learning_rate": 3.807654747596154e-05, "loss": 1.3146, "step": 6353 }, { "epoch": 1.91, "learning_rate": 3.807466947115384e-05, "loss": 0.9964, "step": 6354 }, { "epoch": 1.91, "learning_rate": 3.807279146634616e-05, "loss": 1.3016, "step": 6355 }, { "epoch": 1.91, "learning_rate": 3.8070913461538465e-05, "loss": 1.4202, "step": 6356 }, { "epoch": 1.91, "learning_rate": 3.806903545673077e-05, "loss": 1.1567, "step": 6357 }, { "epoch": 1.91, "learning_rate": 3.806715745192308e-05, "loss": 1.0644, "step": 6358 }, { "epoch": 1.91, "learning_rate": 3.8065279447115386e-05, "loss": 0.9931, "step": 6359 }, { "epoch": 1.91, "learning_rate": 3.80634014423077e-05, "loss": 1.4339, "step": 6360 }, { "epoch": 1.91, "learning_rate": 3.80615234375e-05, "loss": 0.9523, "step": 6361 }, { "epoch": 1.91, "learning_rate": 3.805964543269231e-05, "loss": 1.3309, "step": 6362 }, { "epoch": 1.91, "learning_rate": 3.805776742788462e-05, "loss": 1.2717, "step": 6363 }, { "epoch": 1.91, "learning_rate": 3.8055889423076924e-05, "loss": 1.112, "step": 6364 }, { "epoch": 1.91, "learning_rate": 3.805401141826923e-05, "loss": 1.315, "step": 6365 }, { "epoch": 1.91, "learning_rate": 3.805213341346154e-05, "loss": 1.2861, "step": 6366 }, { "epoch": 1.91, "learning_rate": 3.8050255408653845e-05, "loss": 0.8386, "step": 6367 }, { "epoch": 1.91, "learning_rate": 3.8048377403846156e-05, "loss": 1.2251, "step": 6368 }, { "epoch": 1.91, "learning_rate": 3.804649939903847e-05, "loss": 1.0757, "step": 6369 }, { "epoch": 1.91, "learning_rate": 3.804462139423077e-05, "loss": 1.3056, "step": 6370 }, { "epoch": 1.91, "learning_rate": 3.804274338942308e-05, "loss": 0.9552, "step": 6371 }, { "epoch": 1.91, "learning_rate": 3.804086538461539e-05, "loss": 1.1131, "step": 6372 }, { "epoch": 1.91, "learning_rate": 3.8038987379807694e-05, "loss": 1.1985, "step": 6373 }, { "epoch": 1.92, "learning_rate": 3.8037109375e-05, "loss": 0.9636, "step": 6374 }, { "epoch": 1.92, "learning_rate": 3.803523137019231e-05, "loss": 0.812, "step": 6375 }, { "epoch": 1.92, "learning_rate": 3.8033353365384615e-05, "loss": 0.9772, "step": 6376 }, { "epoch": 1.92, "learning_rate": 3.8031475360576926e-05, "loss": 1.1041, "step": 6377 }, { "epoch": 1.92, "learning_rate": 3.802959735576923e-05, "loss": 1.0622, "step": 6378 }, { "epoch": 1.92, "learning_rate": 3.8027719350961536e-05, "loss": 1.1272, "step": 6379 }, { "epoch": 1.92, "learning_rate": 3.802584134615385e-05, "loss": 1.2301, "step": 6380 }, { "epoch": 1.92, "learning_rate": 3.802396334134616e-05, "loss": 1.2484, "step": 6381 }, { "epoch": 1.92, "learning_rate": 3.8022085336538464e-05, "loss": 0.9967, "step": 6382 }, { "epoch": 1.92, "learning_rate": 3.8020207331730775e-05, "loss": 1.213, "step": 6383 }, { "epoch": 1.92, "learning_rate": 3.801832932692308e-05, "loss": 0.9687, "step": 6384 }, { "epoch": 1.92, "learning_rate": 3.8016451322115385e-05, "loss": 0.9113, "step": 6385 }, { "epoch": 1.92, "learning_rate": 3.8014573317307696e-05, "loss": 1.1663, "step": 6386 }, { "epoch": 1.92, "learning_rate": 3.80126953125e-05, "loss": 0.9087, "step": 6387 }, { "epoch": 1.92, "learning_rate": 3.8010817307692306e-05, "loss": 1.144, "step": 6388 }, { "epoch": 1.92, "learning_rate": 3.800893930288462e-05, "loss": 0.9462, "step": 6389 }, { "epoch": 1.92, "learning_rate": 3.800706129807692e-05, "loss": 1.1078, "step": 6390 }, { "epoch": 1.92, "learning_rate": 3.8005183293269234e-05, "loss": 1.0645, "step": 6391 }, { "epoch": 1.92, "learning_rate": 3.800330528846154e-05, "loss": 1.1596, "step": 6392 }, { "epoch": 1.92, "learning_rate": 3.8001427283653843e-05, "loss": 0.9585, "step": 6393 }, { "epoch": 1.92, "learning_rate": 3.7999549278846155e-05, "loss": 1.1622, "step": 6394 }, { "epoch": 1.92, "learning_rate": 3.7997671274038467e-05, "loss": 0.9707, "step": 6395 }, { "epoch": 1.92, "learning_rate": 3.799579326923077e-05, "loss": 1.0362, "step": 6396 }, { "epoch": 1.92, "learning_rate": 3.799391526442308e-05, "loss": 1.1933, "step": 6397 }, { "epoch": 1.92, "learning_rate": 3.799203725961539e-05, "loss": 1.0518, "step": 6398 }, { "epoch": 1.92, "learning_rate": 3.799015925480769e-05, "loss": 1.1592, "step": 6399 }, { "epoch": 1.92, "learning_rate": 3.7988281250000004e-05, "loss": 1.175, "step": 6400 }, { "epoch": 1.92, "learning_rate": 3.798640324519231e-05, "loss": 0.9314, "step": 6401 }, { "epoch": 1.92, "learning_rate": 3.7984525240384613e-05, "loss": 0.8756, "step": 6402 }, { "epoch": 1.92, "learning_rate": 3.7982647235576925e-05, "loss": 1.1073, "step": 6403 }, { "epoch": 1.92, "learning_rate": 3.798076923076923e-05, "loss": 1.1094, "step": 6404 }, { "epoch": 1.92, "learning_rate": 3.797889122596154e-05, "loss": 1.0788, "step": 6405 }, { "epoch": 1.92, "learning_rate": 3.7977013221153846e-05, "loss": 1.3072, "step": 6406 }, { "epoch": 1.93, "learning_rate": 3.797513521634616e-05, "loss": 1.0085, "step": 6407 }, { "epoch": 1.93, "learning_rate": 3.797325721153846e-05, "loss": 1.0762, "step": 6408 }, { "epoch": 1.93, "learning_rate": 3.7971379206730774e-05, "loss": 1.1998, "step": 6409 }, { "epoch": 1.93, "learning_rate": 3.796950120192308e-05, "loss": 1.0912, "step": 6410 }, { "epoch": 1.93, "learning_rate": 3.796762319711539e-05, "loss": 1.3526, "step": 6411 }, { "epoch": 1.93, "learning_rate": 3.7965745192307695e-05, "loss": 1.4423, "step": 6412 }, { "epoch": 1.93, "learning_rate": 3.79638671875e-05, "loss": 1.0673, "step": 6413 }, { "epoch": 1.93, "learning_rate": 3.796198918269231e-05, "loss": 0.9668, "step": 6414 }, { "epoch": 1.93, "learning_rate": 3.7960111177884616e-05, "loss": 1.1795, "step": 6415 }, { "epoch": 1.93, "learning_rate": 3.795823317307692e-05, "loss": 1.1942, "step": 6416 }, { "epoch": 1.93, "learning_rate": 3.795635516826923e-05, "loss": 0.9707, "step": 6417 }, { "epoch": 1.93, "learning_rate": 3.795447716346154e-05, "loss": 1.2656, "step": 6418 }, { "epoch": 1.93, "learning_rate": 3.795259915865385e-05, "loss": 1.2964, "step": 6419 }, { "epoch": 1.93, "learning_rate": 3.795072115384616e-05, "loss": 0.9127, "step": 6420 }, { "epoch": 1.93, "learning_rate": 3.7948843149038465e-05, "loss": 1.147, "step": 6421 }, { "epoch": 1.93, "learning_rate": 3.794696514423077e-05, "loss": 1.0894, "step": 6422 }, { "epoch": 1.93, "learning_rate": 3.794508713942308e-05, "loss": 1.1881, "step": 6423 }, { "epoch": 1.93, "learning_rate": 3.7943209134615386e-05, "loss": 1.0947, "step": 6424 }, { "epoch": 1.93, "learning_rate": 3.79413311298077e-05, "loss": 1.0644, "step": 6425 }, { "epoch": 1.93, "learning_rate": 3.7939453125e-05, "loss": 0.9744, "step": 6426 }, { "epoch": 1.93, "learning_rate": 3.793757512019231e-05, "loss": 1.0798, "step": 6427 }, { "epoch": 1.93, "learning_rate": 3.793569711538462e-05, "loss": 1.1392, "step": 6428 }, { "epoch": 1.93, "learning_rate": 3.7933819110576924e-05, "loss": 1.1043, "step": 6429 }, { "epoch": 1.93, "learning_rate": 3.793194110576923e-05, "loss": 1.251, "step": 6430 }, { "epoch": 1.93, "learning_rate": 3.793006310096154e-05, "loss": 1.4162, "step": 6431 }, { "epoch": 1.93, "learning_rate": 3.7928185096153845e-05, "loss": 1.1995, "step": 6432 }, { "epoch": 1.93, "learning_rate": 3.7926307091346156e-05, "loss": 1.046, "step": 6433 }, { "epoch": 1.93, "learning_rate": 3.792442908653847e-05, "loss": 1.0802, "step": 6434 }, { "epoch": 1.93, "learning_rate": 3.792255108173077e-05, "loss": 1.3502, "step": 6435 }, { "epoch": 1.93, "learning_rate": 3.792067307692308e-05, "loss": 1.2242, "step": 6436 }, { "epoch": 1.93, "learning_rate": 3.791879507211539e-05, "loss": 1.1148, "step": 6437 }, { "epoch": 1.93, "learning_rate": 3.7916917067307694e-05, "loss": 1.0602, "step": 6438 }, { "epoch": 1.93, "learning_rate": 3.79150390625e-05, "loss": 1.1093, "step": 6439 }, { "epoch": 1.94, "learning_rate": 3.791316105769231e-05, "loss": 1.192, "step": 6440 }, { "epoch": 1.94, "learning_rate": 3.7911283052884615e-05, "loss": 1.2148, "step": 6441 }, { "epoch": 1.94, "learning_rate": 3.7909405048076926e-05, "loss": 1.0979, "step": 6442 }, { "epoch": 1.94, "learning_rate": 3.790752704326923e-05, "loss": 0.9063, "step": 6443 }, { "epoch": 1.94, "learning_rate": 3.7905649038461536e-05, "loss": 1.2169, "step": 6444 }, { "epoch": 1.94, "learning_rate": 3.790377103365385e-05, "loss": 1.2264, "step": 6445 }, { "epoch": 1.94, "learning_rate": 3.790189302884616e-05, "loss": 1.1357, "step": 6446 }, { "epoch": 1.94, "learning_rate": 3.7900015024038464e-05, "loss": 1.1622, "step": 6447 }, { "epoch": 1.94, "learning_rate": 3.7898137019230775e-05, "loss": 1.1053, "step": 6448 }, { "epoch": 1.94, "learning_rate": 3.789625901442308e-05, "loss": 0.9631, "step": 6449 }, { "epoch": 1.94, "learning_rate": 3.7894381009615385e-05, "loss": 1.2263, "step": 6450 }, { "epoch": 1.94, "learning_rate": 3.7892503004807696e-05, "loss": 1.033, "step": 6451 }, { "epoch": 1.94, "learning_rate": 3.7890625e-05, "loss": 1.1069, "step": 6452 }, { "epoch": 1.94, "learning_rate": 3.7888746995192306e-05, "loss": 1.2064, "step": 6453 }, { "epoch": 1.94, "learning_rate": 3.788686899038462e-05, "loss": 1.1495, "step": 6454 }, { "epoch": 1.94, "learning_rate": 3.788499098557692e-05, "loss": 1.2116, "step": 6455 }, { "epoch": 1.94, "learning_rate": 3.7883112980769234e-05, "loss": 1.287, "step": 6456 }, { "epoch": 1.94, "learning_rate": 3.788123497596154e-05, "loss": 1.2463, "step": 6457 }, { "epoch": 1.94, "learning_rate": 3.787935697115384e-05, "loss": 1.0649, "step": 6458 }, { "epoch": 1.94, "learning_rate": 3.7877478966346155e-05, "loss": 1.2985, "step": 6459 }, { "epoch": 1.94, "learning_rate": 3.7875600961538466e-05, "loss": 1.1455, "step": 6460 }, { "epoch": 1.94, "learning_rate": 3.787372295673077e-05, "loss": 1.4289, "step": 6461 }, { "epoch": 1.94, "learning_rate": 3.787184495192308e-05, "loss": 1.2805, "step": 6462 }, { "epoch": 1.94, "learning_rate": 3.786996694711539e-05, "loss": 1.2171, "step": 6463 }, { "epoch": 1.94, "learning_rate": 3.786808894230769e-05, "loss": 1.1277, "step": 6464 }, { "epoch": 1.94, "learning_rate": 3.7866210937500004e-05, "loss": 1.2295, "step": 6465 }, { "epoch": 1.94, "learning_rate": 3.786433293269231e-05, "loss": 1.3295, "step": 6466 }, { "epoch": 1.94, "learning_rate": 3.786245492788461e-05, "loss": 1.3114, "step": 6467 }, { "epoch": 1.94, "learning_rate": 3.7860576923076925e-05, "loss": 1.0666, "step": 6468 }, { "epoch": 1.94, "learning_rate": 3.785869891826923e-05, "loss": 1.0079, "step": 6469 }, { "epoch": 1.94, "learning_rate": 3.785682091346154e-05, "loss": 1.3057, "step": 6470 }, { "epoch": 1.94, "learning_rate": 3.7854942908653846e-05, "loss": 1.1507, "step": 6471 }, { "epoch": 1.94, "learning_rate": 3.785306490384616e-05, "loss": 0.9712, "step": 6472 }, { "epoch": 1.95, "learning_rate": 3.785118689903846e-05, "loss": 1.2758, "step": 6473 }, { "epoch": 1.95, "learning_rate": 3.7849308894230774e-05, "loss": 0.9904, "step": 6474 }, { "epoch": 1.95, "learning_rate": 3.784743088942308e-05, "loss": 0.9911, "step": 6475 }, { "epoch": 1.95, "learning_rate": 3.784555288461539e-05, "loss": 1.2707, "step": 6476 }, { "epoch": 1.95, "learning_rate": 3.7843674879807695e-05, "loss": 1.001, "step": 6477 }, { "epoch": 1.95, "learning_rate": 3.7841796875e-05, "loss": 1.2157, "step": 6478 }, { "epoch": 1.95, "learning_rate": 3.783991887019231e-05, "loss": 0.8493, "step": 6479 }, { "epoch": 1.95, "learning_rate": 3.7838040865384616e-05, "loss": 1.0479, "step": 6480 }, { "epoch": 1.95, "learning_rate": 3.783616286057692e-05, "loss": 1.0134, "step": 6481 }, { "epoch": 1.95, "learning_rate": 3.783428485576923e-05, "loss": 1.2698, "step": 6482 }, { "epoch": 1.95, "learning_rate": 3.783240685096154e-05, "loss": 1.1661, "step": 6483 }, { "epoch": 1.95, "learning_rate": 3.783052884615385e-05, "loss": 1.0601, "step": 6484 }, { "epoch": 1.95, "learning_rate": 3.782865084134616e-05, "loss": 1.2215, "step": 6485 }, { "epoch": 1.95, "learning_rate": 3.7826772836538465e-05, "loss": 1.2397, "step": 6486 }, { "epoch": 1.95, "learning_rate": 3.782489483173077e-05, "loss": 1.0999, "step": 6487 }, { "epoch": 1.95, "learning_rate": 3.782301682692308e-05, "loss": 1.1339, "step": 6488 }, { "epoch": 1.95, "learning_rate": 3.7821138822115386e-05, "loss": 1.0974, "step": 6489 }, { "epoch": 1.95, "learning_rate": 3.78192608173077e-05, "loss": 1.2632, "step": 6490 }, { "epoch": 1.95, "learning_rate": 3.78173828125e-05, "loss": 0.9534, "step": 6491 }, { "epoch": 1.95, "learning_rate": 3.781550480769231e-05, "loss": 0.9938, "step": 6492 }, { "epoch": 1.95, "learning_rate": 3.781362680288462e-05, "loss": 1.1939, "step": 6493 }, { "epoch": 1.95, "learning_rate": 3.7811748798076924e-05, "loss": 1.0098, "step": 6494 }, { "epoch": 1.95, "learning_rate": 3.780987079326923e-05, "loss": 1.0921, "step": 6495 }, { "epoch": 1.95, "learning_rate": 3.780799278846154e-05, "loss": 1.1987, "step": 6496 }, { "epoch": 1.95, "learning_rate": 3.7806114783653845e-05, "loss": 1.424, "step": 6497 }, { "epoch": 1.95, "learning_rate": 3.7804236778846156e-05, "loss": 1.0034, "step": 6498 }, { "epoch": 1.95, "learning_rate": 3.780235877403847e-05, "loss": 1.2786, "step": 6499 }, { "epoch": 1.95, "learning_rate": 3.780048076923077e-05, "loss": 1.487, "step": 6500 }, { "epoch": 1.95, "learning_rate": 3.779860276442308e-05, "loss": 0.9058, "step": 6501 }, { "epoch": 1.95, "learning_rate": 3.779672475961539e-05, "loss": 0.9579, "step": 6502 }, { "epoch": 1.95, "learning_rate": 3.7794846754807694e-05, "loss": 1.3576, "step": 6503 }, { "epoch": 1.95, "learning_rate": 3.7792968750000005e-05, "loss": 1.2509, "step": 6504 }, { "epoch": 1.95, "learning_rate": 3.779109074519231e-05, "loss": 0.9622, "step": 6505 }, { "epoch": 1.95, "learning_rate": 3.7789212740384615e-05, "loss": 1.1198, "step": 6506 }, { "epoch": 1.96, "learning_rate": 3.7787334735576926e-05, "loss": 1.3245, "step": 6507 }, { "epoch": 1.96, "learning_rate": 3.778545673076923e-05, "loss": 1.0203, "step": 6508 }, { "epoch": 1.96, "learning_rate": 3.7783578725961536e-05, "loss": 1.1621, "step": 6509 }, { "epoch": 1.96, "learning_rate": 3.778170072115385e-05, "loss": 1.0714, "step": 6510 }, { "epoch": 1.96, "learning_rate": 3.777982271634616e-05, "loss": 1.1933, "step": 6511 }, { "epoch": 1.96, "learning_rate": 3.7777944711538464e-05, "loss": 1.101, "step": 6512 }, { "epoch": 1.96, "learning_rate": 3.7776066706730775e-05, "loss": 0.9031, "step": 6513 }, { "epoch": 1.96, "learning_rate": 3.777418870192308e-05, "loss": 1.184, "step": 6514 }, { "epoch": 1.96, "learning_rate": 3.7772310697115385e-05, "loss": 1.258, "step": 6515 }, { "epoch": 1.96, "learning_rate": 3.7770432692307696e-05, "loss": 1.1083, "step": 6516 }, { "epoch": 1.96, "learning_rate": 3.77685546875e-05, "loss": 0.9838, "step": 6517 }, { "epoch": 1.96, "learning_rate": 3.7766676682692306e-05, "loss": 1.177, "step": 6518 }, { "epoch": 1.96, "learning_rate": 3.776479867788462e-05, "loss": 1.1793, "step": 6519 }, { "epoch": 1.96, "learning_rate": 3.776292067307692e-05, "loss": 1.1879, "step": 6520 }, { "epoch": 1.96, "learning_rate": 3.7761042668269234e-05, "loss": 1.0262, "step": 6521 }, { "epoch": 1.96, "learning_rate": 3.775916466346154e-05, "loss": 1.1893, "step": 6522 }, { "epoch": 1.96, "learning_rate": 3.775728665865384e-05, "loss": 1.1558, "step": 6523 }, { "epoch": 1.96, "learning_rate": 3.775540865384616e-05, "loss": 1.0743, "step": 6524 }, { "epoch": 1.96, "learning_rate": 3.7753530649038466e-05, "loss": 0.9972, "step": 6525 }, { "epoch": 1.96, "learning_rate": 3.775165264423077e-05, "loss": 1.3148, "step": 6526 }, { "epoch": 1.96, "learning_rate": 3.774977463942308e-05, "loss": 1.2179, "step": 6527 }, { "epoch": 1.96, "learning_rate": 3.774789663461539e-05, "loss": 1.166, "step": 6528 }, { "epoch": 1.96, "learning_rate": 3.774601862980769e-05, "loss": 1.1913, "step": 6529 }, { "epoch": 1.96, "learning_rate": 3.7744140625000004e-05, "loss": 1.0704, "step": 6530 }, { "epoch": 1.96, "learning_rate": 3.774226262019231e-05, "loss": 1.0706, "step": 6531 }, { "epoch": 1.96, "learning_rate": 3.774038461538461e-05, "loss": 1.0614, "step": 6532 }, { "epoch": 1.96, "learning_rate": 3.7738506610576925e-05, "loss": 1.1498, "step": 6533 }, { "epoch": 1.96, "learning_rate": 3.773662860576923e-05, "loss": 1.1463, "step": 6534 }, { "epoch": 1.96, "learning_rate": 3.773475060096154e-05, "loss": 1.0993, "step": 6535 }, { "epoch": 1.96, "learning_rate": 3.7732872596153846e-05, "loss": 1.1238, "step": 6536 }, { "epoch": 1.96, "learning_rate": 3.773099459134616e-05, "loss": 1.3208, "step": 6537 }, { "epoch": 1.96, "learning_rate": 3.772911658653846e-05, "loss": 1.158, "step": 6538 }, { "epoch": 1.96, "learning_rate": 3.7727238581730774e-05, "loss": 1.0697, "step": 6539 }, { "epoch": 1.97, "learning_rate": 3.772536057692308e-05, "loss": 1.1325, "step": 6540 }, { "epoch": 1.97, "learning_rate": 3.772348257211539e-05, "loss": 0.9945, "step": 6541 }, { "epoch": 1.97, "learning_rate": 3.7721604567307695e-05, "loss": 1.1696, "step": 6542 }, { "epoch": 1.97, "learning_rate": 3.77197265625e-05, "loss": 1.2287, "step": 6543 }, { "epoch": 1.97, "learning_rate": 3.771784855769231e-05, "loss": 1.3994, "step": 6544 }, { "epoch": 1.97, "learning_rate": 3.7715970552884616e-05, "loss": 1.3409, "step": 6545 }, { "epoch": 1.97, "learning_rate": 3.771409254807692e-05, "loss": 1.2347, "step": 6546 }, { "epoch": 1.97, "learning_rate": 3.771221454326923e-05, "loss": 1.0331, "step": 6547 }, { "epoch": 1.97, "learning_rate": 3.771033653846154e-05, "loss": 1.2507, "step": 6548 }, { "epoch": 1.97, "learning_rate": 3.770845853365385e-05, "loss": 1.186, "step": 6549 }, { "epoch": 1.97, "learning_rate": 3.770658052884616e-05, "loss": 1.0544, "step": 6550 }, { "epoch": 1.97, "learning_rate": 3.7704702524038465e-05, "loss": 1.0941, "step": 6551 }, { "epoch": 1.97, "learning_rate": 3.770282451923077e-05, "loss": 1.3045, "step": 6552 }, { "epoch": 1.97, "learning_rate": 3.770094651442308e-05, "loss": 0.9326, "step": 6553 }, { "epoch": 1.97, "learning_rate": 3.7699068509615386e-05, "loss": 1.2111, "step": 6554 }, { "epoch": 1.97, "learning_rate": 3.76971905048077e-05, "loss": 1.1396, "step": 6555 }, { "epoch": 1.97, "learning_rate": 3.76953125e-05, "loss": 1.3694, "step": 6556 }, { "epoch": 1.97, "learning_rate": 3.769343449519231e-05, "loss": 1.2994, "step": 6557 }, { "epoch": 1.97, "learning_rate": 3.769155649038462e-05, "loss": 1.0711, "step": 6558 }, { "epoch": 1.97, "learning_rate": 3.7689678485576923e-05, "loss": 1.1347, "step": 6559 }, { "epoch": 1.97, "learning_rate": 3.768780048076923e-05, "loss": 1.0611, "step": 6560 }, { "epoch": 1.97, "learning_rate": 3.768592247596154e-05, "loss": 1.3152, "step": 6561 }, { "epoch": 1.97, "learning_rate": 3.7684044471153844e-05, "loss": 1.0479, "step": 6562 }, { "epoch": 1.97, "learning_rate": 3.7682166466346156e-05, "loss": 1.061, "step": 6563 }, { "epoch": 1.97, "learning_rate": 3.768028846153847e-05, "loss": 1.3789, "step": 6564 }, { "epoch": 1.97, "learning_rate": 3.767841045673077e-05, "loss": 1.1055, "step": 6565 }, { "epoch": 1.97, "learning_rate": 3.767653245192308e-05, "loss": 1.1731, "step": 6566 }, { "epoch": 1.97, "learning_rate": 3.767465444711539e-05, "loss": 1.0547, "step": 6567 }, { "epoch": 1.97, "learning_rate": 3.7672776442307693e-05, "loss": 1.0945, "step": 6568 }, { "epoch": 1.97, "learning_rate": 3.7670898437500005e-05, "loss": 1.1176, "step": 6569 }, { "epoch": 1.97, "learning_rate": 3.766902043269231e-05, "loss": 1.2282, "step": 6570 }, { "epoch": 1.97, "learning_rate": 3.7667142427884615e-05, "loss": 1.0722, "step": 6571 }, { "epoch": 1.97, "learning_rate": 3.7665264423076926e-05, "loss": 0.9568, "step": 6572 }, { "epoch": 1.98, "learning_rate": 3.766338641826923e-05, "loss": 1.2861, "step": 6573 }, { "epoch": 1.98, "learning_rate": 3.7661508413461536e-05, "loss": 1.0543, "step": 6574 }, { "epoch": 1.98, "learning_rate": 3.765963040865385e-05, "loss": 1.246, "step": 6575 }, { "epoch": 1.98, "learning_rate": 3.765775240384616e-05, "loss": 0.9855, "step": 6576 }, { "epoch": 1.98, "learning_rate": 3.7655874399038464e-05, "loss": 1.288, "step": 6577 }, { "epoch": 1.98, "learning_rate": 3.7653996394230775e-05, "loss": 1.1828, "step": 6578 }, { "epoch": 1.98, "learning_rate": 3.765211838942308e-05, "loss": 1.0573, "step": 6579 }, { "epoch": 1.98, "learning_rate": 3.7650240384615385e-05, "loss": 1.1936, "step": 6580 }, { "epoch": 1.98, "learning_rate": 3.7648362379807696e-05, "loss": 1.0883, "step": 6581 }, { "epoch": 1.98, "learning_rate": 3.7646484375e-05, "loss": 1.2632, "step": 6582 }, { "epoch": 1.98, "learning_rate": 3.7644606370192306e-05, "loss": 1.1084, "step": 6583 }, { "epoch": 1.98, "learning_rate": 3.764272836538462e-05, "loss": 1.0332, "step": 6584 }, { "epoch": 1.98, "learning_rate": 3.764085036057692e-05, "loss": 1.0262, "step": 6585 }, { "epoch": 1.98, "learning_rate": 3.7638972355769234e-05, "loss": 1.2315, "step": 6586 }, { "epoch": 1.98, "learning_rate": 3.763709435096154e-05, "loss": 1.1249, "step": 6587 }, { "epoch": 1.98, "learning_rate": 3.763521634615384e-05, "loss": 1.2072, "step": 6588 }, { "epoch": 1.98, "learning_rate": 3.763333834134616e-05, "loss": 1.0861, "step": 6589 }, { "epoch": 1.98, "learning_rate": 3.7631460336538466e-05, "loss": 1.0496, "step": 6590 }, { "epoch": 1.98, "learning_rate": 3.762958233173077e-05, "loss": 1.1121, "step": 6591 }, { "epoch": 1.98, "learning_rate": 3.762770432692308e-05, "loss": 1.161, "step": 6592 }, { "epoch": 1.98, "learning_rate": 3.762582632211539e-05, "loss": 1.2122, "step": 6593 }, { "epoch": 1.98, "learning_rate": 3.762394831730769e-05, "loss": 1.0355, "step": 6594 }, { "epoch": 1.98, "learning_rate": 3.7622070312500004e-05, "loss": 1.0506, "step": 6595 }, { "epoch": 1.98, "learning_rate": 3.762019230769231e-05, "loss": 1.1843, "step": 6596 }, { "epoch": 1.98, "learning_rate": 3.761831430288461e-05, "loss": 1.1444, "step": 6597 }, { "epoch": 1.98, "learning_rate": 3.7616436298076925e-05, "loss": 1.1537, "step": 6598 }, { "epoch": 1.98, "learning_rate": 3.761455829326923e-05, "loss": 1.3535, "step": 6599 }, { "epoch": 1.98, "learning_rate": 3.761268028846154e-05, "loss": 1.086, "step": 6600 }, { "epoch": 1.98, "learning_rate": 3.7610802283653846e-05, "loss": 1.1249, "step": 6601 }, { "epoch": 1.98, "learning_rate": 3.760892427884616e-05, "loss": 1.3151, "step": 6602 }, { "epoch": 1.98, "learning_rate": 3.760704627403846e-05, "loss": 1.1161, "step": 6603 }, { "epoch": 1.98, "learning_rate": 3.7605168269230774e-05, "loss": 1.1172, "step": 6604 }, { "epoch": 1.98, "learning_rate": 3.760329026442308e-05, "loss": 1.2885, "step": 6605 }, { "epoch": 1.98, "learning_rate": 3.760141225961539e-05, "loss": 1.3603, "step": 6606 }, { "epoch": 1.99, "learning_rate": 3.7599534254807695e-05, "loss": 1.2106, "step": 6607 }, { "epoch": 1.99, "learning_rate": 3.759765625e-05, "loss": 1.0889, "step": 6608 }, { "epoch": 1.99, "learning_rate": 3.759577824519231e-05, "loss": 1.2555, "step": 6609 }, { "epoch": 1.99, "learning_rate": 3.7593900240384616e-05, "loss": 1.1059, "step": 6610 }, { "epoch": 1.99, "learning_rate": 3.759202223557692e-05, "loss": 1.2557, "step": 6611 }, { "epoch": 1.99, "learning_rate": 3.759014423076923e-05, "loss": 1.0538, "step": 6612 }, { "epoch": 1.99, "learning_rate": 3.758826622596154e-05, "loss": 1.2584, "step": 6613 }, { "epoch": 1.99, "learning_rate": 3.758638822115385e-05, "loss": 1.0856, "step": 6614 }, { "epoch": 1.99, "learning_rate": 3.758451021634616e-05, "loss": 1.0744, "step": 6615 }, { "epoch": 1.99, "learning_rate": 3.7582632211538465e-05, "loss": 1.2601, "step": 6616 }, { "epoch": 1.99, "learning_rate": 3.758075420673077e-05, "loss": 1.2142, "step": 6617 }, { "epoch": 1.99, "learning_rate": 3.757887620192308e-05, "loss": 1.0019, "step": 6618 }, { "epoch": 1.99, "learning_rate": 3.7576998197115386e-05, "loss": 0.8026, "step": 6619 }, { "epoch": 1.99, "learning_rate": 3.75751201923077e-05, "loss": 1.2032, "step": 6620 }, { "epoch": 1.99, "learning_rate": 3.75732421875e-05, "loss": 1.3788, "step": 6621 }, { "epoch": 1.99, "learning_rate": 3.757136418269231e-05, "loss": 1.2782, "step": 6622 }, { "epoch": 1.99, "learning_rate": 3.756948617788462e-05, "loss": 0.9977, "step": 6623 }, { "epoch": 1.99, "learning_rate": 3.756760817307692e-05, "loss": 1.2795, "step": 6624 }, { "epoch": 1.99, "learning_rate": 3.756573016826923e-05, "loss": 1.2638, "step": 6625 }, { "epoch": 1.99, "learning_rate": 3.756385216346154e-05, "loss": 1.2896, "step": 6626 }, { "epoch": 1.99, "learning_rate": 3.7561974158653844e-05, "loss": 1.1684, "step": 6627 }, { "epoch": 1.99, "learning_rate": 3.7560096153846156e-05, "loss": 1.2719, "step": 6628 }, { "epoch": 1.99, "learning_rate": 3.755821814903847e-05, "loss": 1.3295, "step": 6629 }, { "epoch": 1.99, "learning_rate": 3.755634014423077e-05, "loss": 1.1665, "step": 6630 }, { "epoch": 1.99, "learning_rate": 3.755446213942308e-05, "loss": 1.197, "step": 6631 }, { "epoch": 1.99, "learning_rate": 3.755258413461539e-05, "loss": 1.1401, "step": 6632 }, { "epoch": 1.99, "learning_rate": 3.755070612980769e-05, "loss": 1.2834, "step": 6633 }, { "epoch": 1.99, "learning_rate": 3.7548828125000005e-05, "loss": 0.9947, "step": 6634 }, { "epoch": 1.99, "learning_rate": 3.754695012019231e-05, "loss": 1.1078, "step": 6635 }, { "epoch": 1.99, "learning_rate": 3.7545072115384614e-05, "loss": 1.0183, "step": 6636 }, { "epoch": 1.99, "learning_rate": 3.7543194110576926e-05, "loss": 1.2907, "step": 6637 }, { "epoch": 1.99, "learning_rate": 3.754131610576923e-05, "loss": 1.0321, "step": 6638 }, { "epoch": 1.99, "learning_rate": 3.7539438100961536e-05, "loss": 1.1926, "step": 6639 }, { "epoch": 2.0, "learning_rate": 3.753756009615385e-05, "loss": 1.1833, "step": 6640 }, { "epoch": 2.0, "learning_rate": 3.753568209134616e-05, "loss": 0.9967, "step": 6641 }, { "epoch": 2.0, "learning_rate": 3.7533804086538463e-05, "loss": 1.0799, "step": 6642 }, { "epoch": 2.0, "learning_rate": 3.7531926081730775e-05, "loss": 1.2624, "step": 6643 }, { "epoch": 2.0, "learning_rate": 3.753004807692308e-05, "loss": 1.2274, "step": 6644 }, { "epoch": 2.0, "learning_rate": 3.7528170072115384e-05, "loss": 1.1906, "step": 6645 }, { "epoch": 2.0, "learning_rate": 3.7526292067307696e-05, "loss": 1.1045, "step": 6646 }, { "epoch": 2.0, "learning_rate": 3.75244140625e-05, "loss": 1.0717, "step": 6647 }, { "epoch": 2.0, "learning_rate": 3.752253605769231e-05, "loss": 1.2172, "step": 6648 }, { "epoch": 2.0, "learning_rate": 3.752065805288462e-05, "loss": 1.3091, "step": 6649 }, { "epoch": 2.0, "learning_rate": 3.751878004807692e-05, "loss": 1.0315, "step": 6650 }, { "epoch": 2.0, "learning_rate": 3.7516902043269233e-05, "loss": 1.0016, "step": 6651 }, { "epoch": 2.0, "learning_rate": 3.751502403846154e-05, "loss": 1.2681, "step": 6652 }, { "epoch": 2.0, "learning_rate": 3.751314603365384e-05, "loss": 0.9956, "step": 6653 }, { "epoch": 2.0, "learning_rate": 3.751126802884616e-05, "loss": 1.2268, "step": 6654 }, { "epoch": 2.0, "learning_rate": 3.7509390024038466e-05, "loss": 1.3173, "step": 6655 }, { "epoch": 2.0, "learning_rate": 3.750751201923077e-05, "loss": 1.0528, "step": 6656 }, { "epoch": 2.0, "learning_rate": 3.750563401442308e-05, "loss": 0.3246, "step": 6657 }, { "epoch": 2.0, "learning_rate": 3.750375600961539e-05, "loss": 0.3054, "step": 6658 }, { "epoch": 2.0, "learning_rate": 3.750187800480769e-05, "loss": 0.3015, "step": 6659 }, { "epoch": 2.0, "learning_rate": 3.7500000000000003e-05, "loss": 0.2723, "step": 6660 }, { "epoch": 2.0, "learning_rate": 3.749812199519231e-05, "loss": 0.2537, "step": 6661 }, { "epoch": 2.0, "learning_rate": 3.749624399038461e-05, "loss": 0.266, "step": 6662 }, { "epoch": 2.0, "learning_rate": 3.7494365985576925e-05, "loss": 0.3668, "step": 6663 }, { "epoch": 2.0, "learning_rate": 3.749248798076923e-05, "loss": 0.21, "step": 6664 }, { "epoch": 2.0, "learning_rate": 3.749060997596154e-05, "loss": 0.2852, "step": 6665 }, { "epoch": 2.0, "learning_rate": 3.7488731971153846e-05, "loss": 0.2907, "step": 6666 }, { "epoch": 2.0, "learning_rate": 3.748685396634616e-05, "loss": 0.277, "step": 6667 }, { "epoch": 2.0, "learning_rate": 3.748497596153847e-05, "loss": 0.3282, "step": 6668 }, { "epoch": 2.0, "learning_rate": 3.7483097956730774e-05, "loss": 0.316, "step": 6669 }, { "epoch": 2.0, "learning_rate": 3.748121995192308e-05, "loss": 0.4646, "step": 6670 }, { "epoch": 2.0, "learning_rate": 3.747934194711539e-05, "loss": 0.3849, "step": 6671 }, { "epoch": 2.0, "learning_rate": 3.7477463942307695e-05, "loss": 0.3145, "step": 6672 }, { "epoch": 2.01, "learning_rate": 3.74755859375e-05, "loss": 0.3386, "step": 6673 }, { "epoch": 2.01, "learning_rate": 3.747370793269231e-05, "loss": 0.3344, "step": 6674 }, { "epoch": 2.01, "learning_rate": 3.7471829927884616e-05, "loss": 0.3147, "step": 6675 }, { "epoch": 2.01, "learning_rate": 3.746995192307692e-05, "loss": 0.2793, "step": 6676 }, { "epoch": 2.01, "learning_rate": 3.746807391826923e-05, "loss": 0.3667, "step": 6677 }, { "epoch": 2.01, "learning_rate": 3.746619591346154e-05, "loss": 0.3554, "step": 6678 }, { "epoch": 2.01, "learning_rate": 3.746431790865385e-05, "loss": 0.3529, "step": 6679 }, { "epoch": 2.01, "learning_rate": 3.746243990384616e-05, "loss": 0.2406, "step": 6680 }, { "epoch": 2.01, "learning_rate": 3.7460561899038465e-05, "loss": 0.5007, "step": 6681 }, { "epoch": 2.01, "learning_rate": 3.745868389423077e-05, "loss": 0.2764, "step": 6682 }, { "epoch": 2.01, "learning_rate": 3.745680588942308e-05, "loss": 0.2727, "step": 6683 }, { "epoch": 2.01, "learning_rate": 3.7454927884615386e-05, "loss": 0.3134, "step": 6684 }, { "epoch": 2.01, "learning_rate": 3.74530498798077e-05, "loss": 0.2591, "step": 6685 }, { "epoch": 2.01, "learning_rate": 3.7451171875e-05, "loss": 0.489, "step": 6686 }, { "epoch": 2.01, "learning_rate": 3.744929387019231e-05, "loss": 0.3251, "step": 6687 }, { "epoch": 2.01, "learning_rate": 3.744741586538462e-05, "loss": 0.3538, "step": 6688 }, { "epoch": 2.01, "learning_rate": 3.744553786057692e-05, "loss": 0.2845, "step": 6689 }, { "epoch": 2.01, "learning_rate": 3.744365985576923e-05, "loss": 0.3424, "step": 6690 }, { "epoch": 2.01, "learning_rate": 3.744178185096154e-05, "loss": 0.2186, "step": 6691 }, { "epoch": 2.01, "learning_rate": 3.7439903846153844e-05, "loss": 0.3029, "step": 6692 }, { "epoch": 2.01, "learning_rate": 3.7438025841346156e-05, "loss": 0.3058, "step": 6693 }, { "epoch": 2.01, "learning_rate": 3.743614783653847e-05, "loss": 0.3252, "step": 6694 }, { "epoch": 2.01, "learning_rate": 3.743426983173077e-05, "loss": 0.4524, "step": 6695 }, { "epoch": 2.01, "learning_rate": 3.743239182692308e-05, "loss": 0.321, "step": 6696 }, { "epoch": 2.01, "learning_rate": 3.743051382211539e-05, "loss": 0.5213, "step": 6697 }, { "epoch": 2.01, "learning_rate": 3.742863581730769e-05, "loss": 0.3632, "step": 6698 }, { "epoch": 2.01, "learning_rate": 3.7426757812500005e-05, "loss": 0.4031, "step": 6699 }, { "epoch": 2.01, "learning_rate": 3.742487980769231e-05, "loss": 0.3388, "step": 6700 }, { "epoch": 2.01, "learning_rate": 3.7423001802884614e-05, "loss": 0.3046, "step": 6701 }, { "epoch": 2.01, "learning_rate": 3.7421123798076926e-05, "loss": 0.5217, "step": 6702 }, { "epoch": 2.01, "learning_rate": 3.741924579326923e-05, "loss": 0.3102, "step": 6703 }, { "epoch": 2.01, "learning_rate": 3.7417367788461535e-05, "loss": 0.3742, "step": 6704 }, { "epoch": 2.01, "learning_rate": 3.741548978365385e-05, "loss": 0.3422, "step": 6705 }, { "epoch": 2.02, "learning_rate": 3.741361177884616e-05, "loss": 0.289, "step": 6706 }, { "epoch": 2.02, "learning_rate": 3.741173377403846e-05, "loss": 0.4, "step": 6707 }, { "epoch": 2.02, "learning_rate": 3.7409855769230775e-05, "loss": 0.2147, "step": 6708 }, { "epoch": 2.02, "learning_rate": 3.740797776442308e-05, "loss": 0.2983, "step": 6709 }, { "epoch": 2.02, "learning_rate": 3.7406099759615384e-05, "loss": 0.3953, "step": 6710 }, { "epoch": 2.02, "learning_rate": 3.7404221754807696e-05, "loss": 0.346, "step": 6711 }, { "epoch": 2.02, "learning_rate": 3.740234375e-05, "loss": 0.3126, "step": 6712 }, { "epoch": 2.02, "learning_rate": 3.740046574519231e-05, "loss": 0.259, "step": 6713 }, { "epoch": 2.02, "learning_rate": 3.739858774038462e-05, "loss": 0.2818, "step": 6714 }, { "epoch": 2.02, "learning_rate": 3.739670973557692e-05, "loss": 0.3389, "step": 6715 }, { "epoch": 2.02, "learning_rate": 3.739483173076923e-05, "loss": 0.265, "step": 6716 }, { "epoch": 2.02, "learning_rate": 3.739295372596154e-05, "loss": 0.5012, "step": 6717 }, { "epoch": 2.02, "learning_rate": 3.739107572115384e-05, "loss": 0.1953, "step": 6718 }, { "epoch": 2.02, "learning_rate": 3.738919771634616e-05, "loss": 0.3692, "step": 6719 }, { "epoch": 2.02, "learning_rate": 3.7387319711538466e-05, "loss": 0.3792, "step": 6720 }, { "epoch": 2.02, "learning_rate": 3.738544170673077e-05, "loss": 0.3328, "step": 6721 }, { "epoch": 2.02, "learning_rate": 3.738356370192308e-05, "loss": 0.4416, "step": 6722 }, { "epoch": 2.02, "learning_rate": 3.738168569711539e-05, "loss": 0.3517, "step": 6723 }, { "epoch": 2.02, "learning_rate": 3.737980769230769e-05, "loss": 0.39, "step": 6724 }, { "epoch": 2.02, "learning_rate": 3.73779296875e-05, "loss": 0.499, "step": 6725 }, { "epoch": 2.02, "learning_rate": 3.737605168269231e-05, "loss": 0.4115, "step": 6726 }, { "epoch": 2.02, "learning_rate": 3.737417367788462e-05, "loss": 0.2652, "step": 6727 }, { "epoch": 2.02, "learning_rate": 3.7372295673076924e-05, "loss": 0.2375, "step": 6728 }, { "epoch": 2.02, "learning_rate": 3.737041766826923e-05, "loss": 0.3245, "step": 6729 }, { "epoch": 2.02, "learning_rate": 3.736853966346154e-05, "loss": 0.3419, "step": 6730 }, { "epoch": 2.02, "learning_rate": 3.7366661658653846e-05, "loss": 0.3962, "step": 6731 }, { "epoch": 2.02, "learning_rate": 3.736478365384616e-05, "loss": 0.3598, "step": 6732 }, { "epoch": 2.02, "learning_rate": 3.736290564903847e-05, "loss": 0.3793, "step": 6733 }, { "epoch": 2.02, "learning_rate": 3.7361027644230773e-05, "loss": 0.3501, "step": 6734 }, { "epoch": 2.02, "learning_rate": 3.735914963942308e-05, "loss": 0.3314, "step": 6735 }, { "epoch": 2.02, "learning_rate": 3.735727163461539e-05, "loss": 0.368, "step": 6736 }, { "epoch": 2.02, "learning_rate": 3.7355393629807695e-05, "loss": 0.4045, "step": 6737 }, { "epoch": 2.02, "learning_rate": 3.7353515625e-05, "loss": 0.2493, "step": 6738 }, { "epoch": 2.02, "learning_rate": 3.735163762019231e-05, "loss": 0.2819, "step": 6739 }, { "epoch": 2.03, "learning_rate": 3.7349759615384616e-05, "loss": 0.5055, "step": 6740 }, { "epoch": 2.03, "learning_rate": 3.734788161057692e-05, "loss": 0.346, "step": 6741 }, { "epoch": 2.03, "learning_rate": 3.734600360576923e-05, "loss": 0.3486, "step": 6742 }, { "epoch": 2.03, "learning_rate": 3.734412560096154e-05, "loss": 0.3436, "step": 6743 }, { "epoch": 2.03, "learning_rate": 3.734224759615385e-05, "loss": 0.2592, "step": 6744 }, { "epoch": 2.03, "learning_rate": 3.734036959134616e-05, "loss": 0.2933, "step": 6745 }, { "epoch": 2.03, "learning_rate": 3.7338491586538465e-05, "loss": 0.4179, "step": 6746 }, { "epoch": 2.03, "learning_rate": 3.733661358173077e-05, "loss": 0.4275, "step": 6747 }, { "epoch": 2.03, "learning_rate": 3.733473557692308e-05, "loss": 0.3502, "step": 6748 }, { "epoch": 2.03, "learning_rate": 3.7332857572115386e-05, "loss": 0.4425, "step": 6749 }, { "epoch": 2.03, "learning_rate": 3.73309795673077e-05, "loss": 0.2745, "step": 6750 }, { "epoch": 2.03, "learning_rate": 3.73291015625e-05, "loss": 0.3995, "step": 6751 }, { "epoch": 2.03, "learning_rate": 3.732722355769231e-05, "loss": 0.2734, "step": 6752 }, { "epoch": 2.03, "learning_rate": 3.732534555288462e-05, "loss": 0.399, "step": 6753 }, { "epoch": 2.03, "learning_rate": 3.732346754807692e-05, "loss": 0.2824, "step": 6754 }, { "epoch": 2.03, "learning_rate": 3.732158954326923e-05, "loss": 0.4113, "step": 6755 }, { "epoch": 2.03, "learning_rate": 3.731971153846154e-05, "loss": 0.2949, "step": 6756 }, { "epoch": 2.03, "learning_rate": 3.7317833533653844e-05, "loss": 0.3061, "step": 6757 }, { "epoch": 2.03, "learning_rate": 3.7315955528846156e-05, "loss": 0.2367, "step": 6758 }, { "epoch": 2.03, "learning_rate": 3.731407752403847e-05, "loss": 0.2751, "step": 6759 }, { "epoch": 2.03, "learning_rate": 3.731219951923077e-05, "loss": 0.2241, "step": 6760 }, { "epoch": 2.03, "learning_rate": 3.731032151442308e-05, "loss": 0.2641, "step": 6761 }, { "epoch": 2.03, "learning_rate": 3.730844350961539e-05, "loss": 0.442, "step": 6762 }, { "epoch": 2.03, "learning_rate": 3.730656550480769e-05, "loss": 0.4372, "step": 6763 }, { "epoch": 2.03, "learning_rate": 3.7304687500000005e-05, "loss": 0.3531, "step": 6764 }, { "epoch": 2.03, "learning_rate": 3.730280949519231e-05, "loss": 0.3974, "step": 6765 }, { "epoch": 2.03, "learning_rate": 3.7300931490384614e-05, "loss": 0.3134, "step": 6766 }, { "epoch": 2.03, "learning_rate": 3.7299053485576926e-05, "loss": 0.3794, "step": 6767 }, { "epoch": 2.03, "learning_rate": 3.729717548076923e-05, "loss": 0.2814, "step": 6768 }, { "epoch": 2.03, "learning_rate": 3.7295297475961535e-05, "loss": 0.2886, "step": 6769 }, { "epoch": 2.03, "learning_rate": 3.729341947115385e-05, "loss": 0.3855, "step": 6770 }, { "epoch": 2.03, "learning_rate": 3.729154146634616e-05, "loss": 0.3868, "step": 6771 }, { "epoch": 2.03, "learning_rate": 3.728966346153846e-05, "loss": 0.472, "step": 6772 }, { "epoch": 2.04, "learning_rate": 3.7287785456730775e-05, "loss": 0.4842, "step": 6773 }, { "epoch": 2.04, "learning_rate": 3.728590745192308e-05, "loss": 0.3843, "step": 6774 }, { "epoch": 2.04, "learning_rate": 3.7284029447115384e-05, "loss": 0.323, "step": 6775 }, { "epoch": 2.04, "learning_rate": 3.7282151442307696e-05, "loss": 0.3164, "step": 6776 }, { "epoch": 2.04, "learning_rate": 3.72802734375e-05, "loss": 0.2996, "step": 6777 }, { "epoch": 2.04, "learning_rate": 3.727839543269231e-05, "loss": 0.3544, "step": 6778 }, { "epoch": 2.04, "learning_rate": 3.727651742788462e-05, "loss": 0.2735, "step": 6779 }, { "epoch": 2.04, "learning_rate": 3.727463942307692e-05, "loss": 0.277, "step": 6780 }, { "epoch": 2.04, "learning_rate": 3.727276141826923e-05, "loss": 0.2962, "step": 6781 }, { "epoch": 2.04, "learning_rate": 3.727088341346154e-05, "loss": 0.5299, "step": 6782 }, { "epoch": 2.04, "learning_rate": 3.726900540865384e-05, "loss": 0.3413, "step": 6783 }, { "epoch": 2.04, "learning_rate": 3.726712740384616e-05, "loss": 0.3028, "step": 6784 }, { "epoch": 2.04, "learning_rate": 3.7265249399038466e-05, "loss": 0.4159, "step": 6785 }, { "epoch": 2.04, "learning_rate": 3.726337139423077e-05, "loss": 0.3792, "step": 6786 }, { "epoch": 2.04, "learning_rate": 3.726149338942308e-05, "loss": 0.3884, "step": 6787 }, { "epoch": 2.04, "learning_rate": 3.725961538461539e-05, "loss": 0.295, "step": 6788 }, { "epoch": 2.04, "learning_rate": 3.725773737980769e-05, "loss": 0.4817, "step": 6789 }, { "epoch": 2.04, "learning_rate": 3.7255859375e-05, "loss": 0.3049, "step": 6790 }, { "epoch": 2.04, "learning_rate": 3.725398137019231e-05, "loss": 0.2972, "step": 6791 }, { "epoch": 2.04, "learning_rate": 3.725210336538462e-05, "loss": 0.3712, "step": 6792 }, { "epoch": 2.04, "learning_rate": 3.7250225360576924e-05, "loss": 0.2875, "step": 6793 }, { "epoch": 2.04, "learning_rate": 3.724834735576923e-05, "loss": 0.3829, "step": 6794 }, { "epoch": 2.04, "learning_rate": 3.724646935096154e-05, "loss": 0.328, "step": 6795 }, { "epoch": 2.04, "learning_rate": 3.7244591346153845e-05, "loss": 0.2882, "step": 6796 }, { "epoch": 2.04, "learning_rate": 3.724271334134616e-05, "loss": 0.3052, "step": 6797 }, { "epoch": 2.04, "learning_rate": 3.724083533653847e-05, "loss": 0.3124, "step": 6798 }, { "epoch": 2.04, "learning_rate": 3.723895733173077e-05, "loss": 0.2651, "step": 6799 }, { "epoch": 2.04, "learning_rate": 3.723707932692308e-05, "loss": 0.5166, "step": 6800 }, { "epoch": 2.04, "learning_rate": 3.723520132211539e-05, "loss": 0.32, "step": 6801 }, { "epoch": 2.04, "learning_rate": 3.7233323317307694e-05, "loss": 0.2696, "step": 6802 }, { "epoch": 2.04, "learning_rate": 3.72314453125e-05, "loss": 0.3354, "step": 6803 }, { "epoch": 2.04, "learning_rate": 3.722956730769231e-05, "loss": 0.2779, "step": 6804 }, { "epoch": 2.04, "learning_rate": 3.7227689302884616e-05, "loss": 0.2313, "step": 6805 }, { "epoch": 2.05, "learning_rate": 3.722581129807692e-05, "loss": 0.1945, "step": 6806 }, { "epoch": 2.05, "learning_rate": 3.722393329326923e-05, "loss": 0.6656, "step": 6807 }, { "epoch": 2.05, "learning_rate": 3.7222055288461537e-05, "loss": 0.3758, "step": 6808 }, { "epoch": 2.05, "learning_rate": 3.722017728365385e-05, "loss": 0.3322, "step": 6809 }, { "epoch": 2.05, "learning_rate": 3.721829927884616e-05, "loss": 0.3054, "step": 6810 }, { "epoch": 2.05, "learning_rate": 3.7216421274038464e-05, "loss": 0.3937, "step": 6811 }, { "epoch": 2.05, "learning_rate": 3.7214543269230776e-05, "loss": 0.4302, "step": 6812 }, { "epoch": 2.05, "learning_rate": 3.721266526442308e-05, "loss": 0.2489, "step": 6813 }, { "epoch": 2.05, "learning_rate": 3.7210787259615386e-05, "loss": 0.2242, "step": 6814 }, { "epoch": 2.05, "learning_rate": 3.72089092548077e-05, "loss": 0.546, "step": 6815 }, { "epoch": 2.05, "learning_rate": 3.720703125e-05, "loss": 0.4598, "step": 6816 }, { "epoch": 2.05, "learning_rate": 3.720515324519231e-05, "loss": 0.3128, "step": 6817 }, { "epoch": 2.05, "learning_rate": 3.720327524038462e-05, "loss": 0.3887, "step": 6818 }, { "epoch": 2.05, "learning_rate": 3.720139723557692e-05, "loss": 0.3072, "step": 6819 }, { "epoch": 2.05, "learning_rate": 3.719951923076923e-05, "loss": 0.3444, "step": 6820 }, { "epoch": 2.05, "learning_rate": 3.719764122596154e-05, "loss": 0.2491, "step": 6821 }, { "epoch": 2.05, "learning_rate": 3.7195763221153844e-05, "loss": 0.2481, "step": 6822 }, { "epoch": 2.05, "learning_rate": 3.7193885216346156e-05, "loss": 0.3052, "step": 6823 }, { "epoch": 2.05, "learning_rate": 3.719200721153847e-05, "loss": 0.3151, "step": 6824 }, { "epoch": 2.05, "learning_rate": 3.719012920673077e-05, "loss": 0.3498, "step": 6825 }, { "epoch": 2.05, "learning_rate": 3.718825120192308e-05, "loss": 0.177, "step": 6826 }, { "epoch": 2.05, "learning_rate": 3.718637319711539e-05, "loss": 0.3736, "step": 6827 }, { "epoch": 2.05, "learning_rate": 3.718449519230769e-05, "loss": 0.3454, "step": 6828 }, { "epoch": 2.05, "learning_rate": 3.7182617187500005e-05, "loss": 0.3352, "step": 6829 }, { "epoch": 2.05, "learning_rate": 3.718073918269231e-05, "loss": 0.3271, "step": 6830 }, { "epoch": 2.05, "learning_rate": 3.7178861177884614e-05, "loss": 0.3488, "step": 6831 }, { "epoch": 2.05, "learning_rate": 3.7176983173076926e-05, "loss": 0.3382, "step": 6832 }, { "epoch": 2.05, "learning_rate": 3.717510516826923e-05, "loss": 0.3699, "step": 6833 }, { "epoch": 2.05, "learning_rate": 3.7173227163461535e-05, "loss": 0.2936, "step": 6834 }, { "epoch": 2.05, "learning_rate": 3.717134915865385e-05, "loss": 0.379, "step": 6835 }, { "epoch": 2.05, "learning_rate": 3.716947115384616e-05, "loss": 0.3575, "step": 6836 }, { "epoch": 2.05, "learning_rate": 3.716759314903846e-05, "loss": 0.2284, "step": 6837 }, { "epoch": 2.05, "learning_rate": 3.7165715144230775e-05, "loss": 0.3012, "step": 6838 }, { "epoch": 2.05, "learning_rate": 3.716383713942308e-05, "loss": 0.3273, "step": 6839 }, { "epoch": 2.06, "learning_rate": 3.7161959134615384e-05, "loss": 0.4226, "step": 6840 }, { "epoch": 2.06, "learning_rate": 3.7160081129807696e-05, "loss": 0.3902, "step": 6841 }, { "epoch": 2.06, "learning_rate": 3.7158203125e-05, "loss": 0.3005, "step": 6842 }, { "epoch": 2.06, "learning_rate": 3.715632512019231e-05, "loss": 0.3027, "step": 6843 }, { "epoch": 2.06, "learning_rate": 3.715444711538462e-05, "loss": 0.3569, "step": 6844 }, { "epoch": 2.06, "learning_rate": 3.715256911057692e-05, "loss": 0.3998, "step": 6845 }, { "epoch": 2.06, "learning_rate": 3.715069110576923e-05, "loss": 0.3061, "step": 6846 }, { "epoch": 2.06, "learning_rate": 3.714881310096154e-05, "loss": 0.3331, "step": 6847 }, { "epoch": 2.06, "learning_rate": 3.714693509615384e-05, "loss": 0.2635, "step": 6848 }, { "epoch": 2.06, "learning_rate": 3.714505709134616e-05, "loss": 0.4073, "step": 6849 }, { "epoch": 2.06, "learning_rate": 3.7143179086538466e-05, "loss": 0.3826, "step": 6850 }, { "epoch": 2.06, "learning_rate": 3.714130108173077e-05, "loss": 0.3931, "step": 6851 }, { "epoch": 2.06, "learning_rate": 3.713942307692308e-05, "loss": 0.4346, "step": 6852 }, { "epoch": 2.06, "learning_rate": 3.713754507211539e-05, "loss": 0.2596, "step": 6853 }, { "epoch": 2.06, "learning_rate": 3.713566706730769e-05, "loss": 0.3253, "step": 6854 }, { "epoch": 2.06, "learning_rate": 3.71337890625e-05, "loss": 0.2968, "step": 6855 }, { "epoch": 2.06, "learning_rate": 3.713191105769231e-05, "loss": 0.2746, "step": 6856 }, { "epoch": 2.06, "learning_rate": 3.713003305288462e-05, "loss": 0.2127, "step": 6857 }, { "epoch": 2.06, "learning_rate": 3.7128155048076924e-05, "loss": 0.319, "step": 6858 }, { "epoch": 2.06, "learning_rate": 3.712627704326923e-05, "loss": 0.2403, "step": 6859 }, { "epoch": 2.06, "learning_rate": 3.712439903846154e-05, "loss": 0.3021, "step": 6860 }, { "epoch": 2.06, "learning_rate": 3.7122521033653845e-05, "loss": 0.289, "step": 6861 }, { "epoch": 2.06, "learning_rate": 3.712064302884616e-05, "loss": 0.3987, "step": 6862 }, { "epoch": 2.06, "learning_rate": 3.711876502403847e-05, "loss": 0.2903, "step": 6863 }, { "epoch": 2.06, "learning_rate": 3.711688701923077e-05, "loss": 0.3881, "step": 6864 }, { "epoch": 2.06, "learning_rate": 3.711500901442308e-05, "loss": 0.3622, "step": 6865 }, { "epoch": 2.06, "learning_rate": 3.711313100961539e-05, "loss": 0.3811, "step": 6866 }, { "epoch": 2.06, "learning_rate": 3.7111253004807694e-05, "loss": 0.2647, "step": 6867 }, { "epoch": 2.06, "learning_rate": 3.7109375e-05, "loss": 0.2947, "step": 6868 }, { "epoch": 2.06, "learning_rate": 3.710749699519231e-05, "loss": 0.3956, "step": 6869 }, { "epoch": 2.06, "learning_rate": 3.7105618990384615e-05, "loss": 0.2073, "step": 6870 }, { "epoch": 2.06, "learning_rate": 3.710374098557693e-05, "loss": 0.2483, "step": 6871 }, { "epoch": 2.06, "learning_rate": 3.710186298076923e-05, "loss": 0.343, "step": 6872 }, { "epoch": 2.07, "learning_rate": 3.7099984975961536e-05, "loss": 0.3809, "step": 6873 }, { "epoch": 2.07, "learning_rate": 3.709810697115385e-05, "loss": 0.3491, "step": 6874 }, { "epoch": 2.07, "learning_rate": 3.709622896634616e-05, "loss": 0.3659, "step": 6875 }, { "epoch": 2.07, "learning_rate": 3.7094350961538464e-05, "loss": 0.2535, "step": 6876 }, { "epoch": 2.07, "learning_rate": 3.7092472956730776e-05, "loss": 0.3274, "step": 6877 }, { "epoch": 2.07, "learning_rate": 3.709059495192308e-05, "loss": 0.2375, "step": 6878 }, { "epoch": 2.07, "learning_rate": 3.7088716947115385e-05, "loss": 0.4461, "step": 6879 }, { "epoch": 2.07, "learning_rate": 3.70868389423077e-05, "loss": 0.3155, "step": 6880 }, { "epoch": 2.07, "learning_rate": 3.70849609375e-05, "loss": 0.3449, "step": 6881 }, { "epoch": 2.07, "learning_rate": 3.7083082932692307e-05, "loss": 0.3046, "step": 6882 }, { "epoch": 2.07, "learning_rate": 3.708120492788462e-05, "loss": 0.5656, "step": 6883 }, { "epoch": 2.07, "learning_rate": 3.707932692307692e-05, "loss": 0.3084, "step": 6884 }, { "epoch": 2.07, "learning_rate": 3.707744891826923e-05, "loss": 0.3224, "step": 6885 }, { "epoch": 2.07, "learning_rate": 3.707557091346154e-05, "loss": 0.3558, "step": 6886 }, { "epoch": 2.07, "learning_rate": 3.7073692908653844e-05, "loss": 0.3285, "step": 6887 }, { "epoch": 2.07, "learning_rate": 3.7071814903846155e-05, "loss": 0.3106, "step": 6888 }, { "epoch": 2.07, "learning_rate": 3.706993689903847e-05, "loss": 0.2701, "step": 6889 }, { "epoch": 2.07, "learning_rate": 3.706805889423077e-05, "loss": 0.3509, "step": 6890 }, { "epoch": 2.07, "learning_rate": 3.706618088942308e-05, "loss": 0.4485, "step": 6891 }, { "epoch": 2.07, "learning_rate": 3.706430288461539e-05, "loss": 0.3882, "step": 6892 }, { "epoch": 2.07, "learning_rate": 3.706242487980769e-05, "loss": 0.3207, "step": 6893 }, { "epoch": 2.07, "learning_rate": 3.7060546875000004e-05, "loss": 0.2783, "step": 6894 }, { "epoch": 2.07, "learning_rate": 3.705866887019231e-05, "loss": 0.3186, "step": 6895 }, { "epoch": 2.07, "learning_rate": 3.7056790865384614e-05, "loss": 0.375, "step": 6896 }, { "epoch": 2.07, "learning_rate": 3.7054912860576926e-05, "loss": 0.2966, "step": 6897 }, { "epoch": 2.07, "learning_rate": 3.705303485576923e-05, "loss": 0.2853, "step": 6898 }, { "epoch": 2.07, "learning_rate": 3.7051156850961535e-05, "loss": 0.3241, "step": 6899 }, { "epoch": 2.07, "learning_rate": 3.704927884615385e-05, "loss": 0.2693, "step": 6900 }, { "epoch": 2.07, "learning_rate": 3.704740084134616e-05, "loss": 0.4643, "step": 6901 }, { "epoch": 2.07, "learning_rate": 3.704552283653846e-05, "loss": 0.3241, "step": 6902 }, { "epoch": 2.07, "learning_rate": 3.7043644831730775e-05, "loss": 0.271, "step": 6903 }, { "epoch": 2.07, "learning_rate": 3.704176682692308e-05, "loss": 0.2086, "step": 6904 }, { "epoch": 2.07, "learning_rate": 3.7039888822115384e-05, "loss": 0.2954, "step": 6905 }, { "epoch": 2.08, "learning_rate": 3.7038010817307696e-05, "loss": 0.2566, "step": 6906 }, { "epoch": 2.08, "learning_rate": 3.70361328125e-05, "loss": 0.4031, "step": 6907 }, { "epoch": 2.08, "learning_rate": 3.703425480769231e-05, "loss": 0.2499, "step": 6908 }, { "epoch": 2.08, "learning_rate": 3.703237680288462e-05, "loss": 0.3982, "step": 6909 }, { "epoch": 2.08, "learning_rate": 3.703049879807692e-05, "loss": 0.4085, "step": 6910 }, { "epoch": 2.08, "learning_rate": 3.702862079326923e-05, "loss": 0.384, "step": 6911 }, { "epoch": 2.08, "learning_rate": 3.702674278846154e-05, "loss": 0.3557, "step": 6912 }, { "epoch": 2.08, "learning_rate": 3.702486478365384e-05, "loss": 0.3123, "step": 6913 }, { "epoch": 2.08, "learning_rate": 3.702298677884616e-05, "loss": 0.3907, "step": 6914 }, { "epoch": 2.08, "learning_rate": 3.7021108774038466e-05, "loss": 0.2772, "step": 6915 }, { "epoch": 2.08, "learning_rate": 3.701923076923077e-05, "loss": 0.2677, "step": 6916 }, { "epoch": 2.08, "learning_rate": 3.701735276442308e-05, "loss": 0.3829, "step": 6917 }, { "epoch": 2.08, "learning_rate": 3.701547475961539e-05, "loss": 0.2516, "step": 6918 }, { "epoch": 2.08, "learning_rate": 3.701359675480769e-05, "loss": 0.3392, "step": 6919 }, { "epoch": 2.08, "learning_rate": 3.701171875e-05, "loss": 0.3059, "step": 6920 }, { "epoch": 2.08, "learning_rate": 3.700984074519231e-05, "loss": 0.3277, "step": 6921 }, { "epoch": 2.08, "learning_rate": 3.700796274038462e-05, "loss": 0.3732, "step": 6922 }, { "epoch": 2.08, "learning_rate": 3.7006084735576924e-05, "loss": 0.3708, "step": 6923 }, { "epoch": 2.08, "learning_rate": 3.700420673076923e-05, "loss": 0.3485, "step": 6924 }, { "epoch": 2.08, "learning_rate": 3.700232872596154e-05, "loss": 0.304, "step": 6925 }, { "epoch": 2.08, "learning_rate": 3.7000450721153845e-05, "loss": 0.3398, "step": 6926 }, { "epoch": 2.08, "learning_rate": 3.699857271634616e-05, "loss": 0.3264, "step": 6927 }, { "epoch": 2.08, "learning_rate": 3.699669471153847e-05, "loss": 0.4736, "step": 6928 }, { "epoch": 2.08, "learning_rate": 3.699481670673077e-05, "loss": 0.2586, "step": 6929 }, { "epoch": 2.08, "learning_rate": 3.699293870192308e-05, "loss": 0.4141, "step": 6930 }, { "epoch": 2.08, "learning_rate": 3.699106069711539e-05, "loss": 0.2828, "step": 6931 }, { "epoch": 2.08, "learning_rate": 3.6989182692307694e-05, "loss": 0.2759, "step": 6932 }, { "epoch": 2.08, "learning_rate": 3.69873046875e-05, "loss": 0.3459, "step": 6933 }, { "epoch": 2.08, "learning_rate": 3.698542668269231e-05, "loss": 0.3273, "step": 6934 }, { "epoch": 2.08, "learning_rate": 3.6983548677884615e-05, "loss": 0.4127, "step": 6935 }, { "epoch": 2.08, "learning_rate": 3.698167067307693e-05, "loss": 0.3749, "step": 6936 }, { "epoch": 2.08, "learning_rate": 3.697979266826923e-05, "loss": 0.4394, "step": 6937 }, { "epoch": 2.08, "learning_rate": 3.6977914663461536e-05, "loss": 0.197, "step": 6938 }, { "epoch": 2.09, "learning_rate": 3.697603665865385e-05, "loss": 0.2524, "step": 6939 }, { "epoch": 2.09, "learning_rate": 3.697415865384616e-05, "loss": 0.2461, "step": 6940 }, { "epoch": 2.09, "learning_rate": 3.6972280649038464e-05, "loss": 0.3537, "step": 6941 }, { "epoch": 2.09, "learning_rate": 3.6970402644230776e-05, "loss": 0.2993, "step": 6942 }, { "epoch": 2.09, "learning_rate": 3.696852463942308e-05, "loss": 0.3155, "step": 6943 }, { "epoch": 2.09, "learning_rate": 3.6966646634615385e-05, "loss": 0.3778, "step": 6944 }, { "epoch": 2.09, "learning_rate": 3.69647686298077e-05, "loss": 0.2285, "step": 6945 }, { "epoch": 2.09, "learning_rate": 3.6962890625e-05, "loss": 0.3537, "step": 6946 }, { "epoch": 2.09, "learning_rate": 3.6961012620192306e-05, "loss": 0.3458, "step": 6947 }, { "epoch": 2.09, "learning_rate": 3.695913461538462e-05, "loss": 0.2686, "step": 6948 }, { "epoch": 2.09, "learning_rate": 3.695725661057692e-05, "loss": 0.4384, "step": 6949 }, { "epoch": 2.09, "learning_rate": 3.695537860576923e-05, "loss": 0.4226, "step": 6950 }, { "epoch": 2.09, "learning_rate": 3.695350060096154e-05, "loss": 0.4054, "step": 6951 }, { "epoch": 2.09, "learning_rate": 3.6951622596153844e-05, "loss": 0.3589, "step": 6952 }, { "epoch": 2.09, "learning_rate": 3.6949744591346155e-05, "loss": 0.2878, "step": 6953 }, { "epoch": 2.09, "learning_rate": 3.694786658653847e-05, "loss": 0.278, "step": 6954 }, { "epoch": 2.09, "learning_rate": 3.694598858173077e-05, "loss": 0.2678, "step": 6955 }, { "epoch": 2.09, "learning_rate": 3.694411057692308e-05, "loss": 0.3101, "step": 6956 }, { "epoch": 2.09, "learning_rate": 3.694223257211539e-05, "loss": 0.3693, "step": 6957 }, { "epoch": 2.09, "learning_rate": 3.694035456730769e-05, "loss": 0.3292, "step": 6958 }, { "epoch": 2.09, "learning_rate": 3.6938476562500004e-05, "loss": 0.3352, "step": 6959 }, { "epoch": 2.09, "learning_rate": 3.693659855769231e-05, "loss": 0.4719, "step": 6960 }, { "epoch": 2.09, "learning_rate": 3.6934720552884614e-05, "loss": 0.2695, "step": 6961 }, { "epoch": 2.09, "learning_rate": 3.6932842548076925e-05, "loss": 0.2189, "step": 6962 }, { "epoch": 2.09, "learning_rate": 3.693096454326923e-05, "loss": 0.2486, "step": 6963 }, { "epoch": 2.09, "learning_rate": 3.6929086538461535e-05, "loss": 0.4215, "step": 6964 }, { "epoch": 2.09, "learning_rate": 3.6927208533653847e-05, "loss": 0.2863, "step": 6965 }, { "epoch": 2.09, "learning_rate": 3.692533052884616e-05, "loss": 0.2739, "step": 6966 }, { "epoch": 2.09, "learning_rate": 3.692345252403846e-05, "loss": 0.385, "step": 6967 }, { "epoch": 2.09, "learning_rate": 3.6921574519230774e-05, "loss": 0.2301, "step": 6968 }, { "epoch": 2.09, "learning_rate": 3.691969651442308e-05, "loss": 0.3426, "step": 6969 }, { "epoch": 2.09, "learning_rate": 3.6917818509615384e-05, "loss": 0.2617, "step": 6970 }, { "epoch": 2.09, "learning_rate": 3.6915940504807695e-05, "loss": 0.4071, "step": 6971 }, { "epoch": 2.09, "learning_rate": 3.69140625e-05, "loss": 0.311, "step": 6972 }, { "epoch": 2.1, "learning_rate": 3.691218449519231e-05, "loss": 0.3199, "step": 6973 }, { "epoch": 2.1, "learning_rate": 3.6910306490384617e-05, "loss": 0.3118, "step": 6974 }, { "epoch": 2.1, "learning_rate": 3.690842848557692e-05, "loss": 0.4662, "step": 6975 }, { "epoch": 2.1, "learning_rate": 3.690655048076923e-05, "loss": 0.2448, "step": 6976 }, { "epoch": 2.1, "learning_rate": 3.690467247596154e-05, "loss": 0.4698, "step": 6977 }, { "epoch": 2.1, "learning_rate": 3.690279447115384e-05, "loss": 0.3878, "step": 6978 }, { "epoch": 2.1, "learning_rate": 3.690091646634616e-05, "loss": 0.3247, "step": 6979 }, { "epoch": 2.1, "learning_rate": 3.6899038461538466e-05, "loss": 0.3702, "step": 6980 }, { "epoch": 2.1, "learning_rate": 3.689716045673077e-05, "loss": 0.3934, "step": 6981 }, { "epoch": 2.1, "learning_rate": 3.689528245192308e-05, "loss": 0.4568, "step": 6982 }, { "epoch": 2.1, "learning_rate": 3.689340444711539e-05, "loss": 0.3143, "step": 6983 }, { "epoch": 2.1, "learning_rate": 3.689152644230769e-05, "loss": 0.455, "step": 6984 }, { "epoch": 2.1, "learning_rate": 3.68896484375e-05, "loss": 0.4072, "step": 6985 }, { "epoch": 2.1, "learning_rate": 3.688777043269231e-05, "loss": 0.3477, "step": 6986 }, { "epoch": 2.1, "learning_rate": 3.688589242788462e-05, "loss": 0.3666, "step": 6987 }, { "epoch": 2.1, "learning_rate": 3.6884014423076924e-05, "loss": 0.2563, "step": 6988 }, { "epoch": 2.1, "learning_rate": 3.688213641826923e-05, "loss": 0.2805, "step": 6989 }, { "epoch": 2.1, "learning_rate": 3.688025841346154e-05, "loss": 0.3553, "step": 6990 }, { "epoch": 2.1, "learning_rate": 3.6878380408653845e-05, "loss": 0.2391, "step": 6991 }, { "epoch": 2.1, "learning_rate": 3.687650240384616e-05, "loss": 0.5663, "step": 6992 }, { "epoch": 2.1, "learning_rate": 3.687462439903847e-05, "loss": 0.309, "step": 6993 }, { "epoch": 2.1, "learning_rate": 3.687274639423077e-05, "loss": 0.2626, "step": 6994 }, { "epoch": 2.1, "learning_rate": 3.687086838942308e-05, "loss": 0.4249, "step": 6995 }, { "epoch": 2.1, "learning_rate": 3.686899038461539e-05, "loss": 0.2243, "step": 6996 }, { "epoch": 2.1, "learning_rate": 3.6867112379807694e-05, "loss": 0.3042, "step": 6997 }, { "epoch": 2.1, "learning_rate": 3.6865234375e-05, "loss": 0.3437, "step": 6998 }, { "epoch": 2.1, "learning_rate": 3.686335637019231e-05, "loss": 0.4334, "step": 6999 }, { "epoch": 2.1, "learning_rate": 3.6861478365384615e-05, "loss": 0.3275, "step": 7000 }, { "epoch": 2.1, "learning_rate": 3.685960036057693e-05, "loss": 0.4604, "step": 7001 }, { "epoch": 2.1, "learning_rate": 3.685772235576923e-05, "loss": 0.4435, "step": 7002 }, { "epoch": 2.1, "learning_rate": 3.6855844350961536e-05, "loss": 0.3609, "step": 7003 }, { "epoch": 2.1, "learning_rate": 3.685396634615385e-05, "loss": 0.2837, "step": 7004 }, { "epoch": 2.1, "learning_rate": 3.685208834134616e-05, "loss": 0.2574, "step": 7005 }, { "epoch": 2.11, "learning_rate": 3.6850210336538464e-05, "loss": 0.3571, "step": 7006 }, { "epoch": 2.11, "learning_rate": 3.6848332331730776e-05, "loss": 0.3831, "step": 7007 }, { "epoch": 2.11, "learning_rate": 3.684645432692308e-05, "loss": 0.5037, "step": 7008 }, { "epoch": 2.11, "learning_rate": 3.6844576322115385e-05, "loss": 0.4603, "step": 7009 }, { "epoch": 2.11, "learning_rate": 3.68426983173077e-05, "loss": 0.3073, "step": 7010 }, { "epoch": 2.11, "learning_rate": 3.68408203125e-05, "loss": 0.3075, "step": 7011 }, { "epoch": 2.11, "learning_rate": 3.6838942307692306e-05, "loss": 0.2892, "step": 7012 }, { "epoch": 2.11, "learning_rate": 3.683706430288462e-05, "loss": 0.2076, "step": 7013 }, { "epoch": 2.11, "learning_rate": 3.683518629807692e-05, "loss": 0.323, "step": 7014 }, { "epoch": 2.11, "learning_rate": 3.6833308293269234e-05, "loss": 0.2533, "step": 7015 }, { "epoch": 2.11, "learning_rate": 3.683143028846154e-05, "loss": 0.5237, "step": 7016 }, { "epoch": 2.11, "learning_rate": 3.6829552283653844e-05, "loss": 0.3581, "step": 7017 }, { "epoch": 2.11, "learning_rate": 3.6827674278846155e-05, "loss": 0.3721, "step": 7018 }, { "epoch": 2.11, "learning_rate": 3.682579627403847e-05, "loss": 0.425, "step": 7019 }, { "epoch": 2.11, "learning_rate": 3.682391826923077e-05, "loss": 0.4276, "step": 7020 }, { "epoch": 2.11, "learning_rate": 3.682204026442308e-05, "loss": 0.3258, "step": 7021 }, { "epoch": 2.11, "learning_rate": 3.682016225961539e-05, "loss": 0.2169, "step": 7022 }, { "epoch": 2.11, "learning_rate": 3.681828425480769e-05, "loss": 0.4485, "step": 7023 }, { "epoch": 2.11, "learning_rate": 3.6816406250000004e-05, "loss": 0.2752, "step": 7024 }, { "epoch": 2.11, "learning_rate": 3.681452824519231e-05, "loss": 0.2998, "step": 7025 }, { "epoch": 2.11, "learning_rate": 3.6812650240384614e-05, "loss": 0.2676, "step": 7026 }, { "epoch": 2.11, "learning_rate": 3.6810772235576925e-05, "loss": 0.3542, "step": 7027 }, { "epoch": 2.11, "learning_rate": 3.680889423076923e-05, "loss": 0.4201, "step": 7028 }, { "epoch": 2.11, "learning_rate": 3.6807016225961535e-05, "loss": 0.3355, "step": 7029 }, { "epoch": 2.11, "learning_rate": 3.6805138221153846e-05, "loss": 0.2738, "step": 7030 }, { "epoch": 2.11, "learning_rate": 3.680326021634616e-05, "loss": 0.2534, "step": 7031 }, { "epoch": 2.11, "learning_rate": 3.680138221153846e-05, "loss": 0.3002, "step": 7032 }, { "epoch": 2.11, "learning_rate": 3.6799504206730774e-05, "loss": 0.2741, "step": 7033 }, { "epoch": 2.11, "learning_rate": 3.679762620192308e-05, "loss": 0.2954, "step": 7034 }, { "epoch": 2.11, "learning_rate": 3.679574819711539e-05, "loss": 0.3673, "step": 7035 }, { "epoch": 2.11, "learning_rate": 3.6793870192307695e-05, "loss": 0.3484, "step": 7036 }, { "epoch": 2.11, "learning_rate": 3.67919921875e-05, "loss": 0.2688, "step": 7037 }, { "epoch": 2.11, "learning_rate": 3.679011418269231e-05, "loss": 0.3966, "step": 7038 }, { "epoch": 2.12, "learning_rate": 3.6788236177884616e-05, "loss": 0.3153, "step": 7039 }, { "epoch": 2.12, "learning_rate": 3.678635817307692e-05, "loss": 0.3944, "step": 7040 }, { "epoch": 2.12, "learning_rate": 3.678448016826923e-05, "loss": 0.3387, "step": 7041 }, { "epoch": 2.12, "learning_rate": 3.678260216346154e-05, "loss": 0.2793, "step": 7042 }, { "epoch": 2.12, "learning_rate": 3.678072415865384e-05, "loss": 0.3257, "step": 7043 }, { "epoch": 2.12, "learning_rate": 3.677884615384616e-05, "loss": 0.2709, "step": 7044 }, { "epoch": 2.12, "learning_rate": 3.6776968149038465e-05, "loss": 0.4706, "step": 7045 }, { "epoch": 2.12, "learning_rate": 3.677509014423077e-05, "loss": 0.3591, "step": 7046 }, { "epoch": 2.12, "learning_rate": 3.677321213942308e-05, "loss": 0.3338, "step": 7047 }, { "epoch": 2.12, "learning_rate": 3.6771334134615387e-05, "loss": 0.3232, "step": 7048 }, { "epoch": 2.12, "learning_rate": 3.676945612980769e-05, "loss": 0.3632, "step": 7049 }, { "epoch": 2.12, "learning_rate": 3.6767578125e-05, "loss": 0.2841, "step": 7050 }, { "epoch": 2.12, "learning_rate": 3.676570012019231e-05, "loss": 0.2997, "step": 7051 }, { "epoch": 2.12, "learning_rate": 3.676382211538462e-05, "loss": 0.3551, "step": 7052 }, { "epoch": 2.12, "learning_rate": 3.6761944110576924e-05, "loss": 0.3188, "step": 7053 }, { "epoch": 2.12, "learning_rate": 3.676006610576923e-05, "loss": 0.2946, "step": 7054 }, { "epoch": 2.12, "learning_rate": 3.675818810096154e-05, "loss": 0.28, "step": 7055 }, { "epoch": 2.12, "learning_rate": 3.6756310096153845e-05, "loss": 0.3542, "step": 7056 }, { "epoch": 2.12, "learning_rate": 3.6754432091346157e-05, "loss": 0.3835, "step": 7057 }, { "epoch": 2.12, "learning_rate": 3.675255408653847e-05, "loss": 0.2769, "step": 7058 }, { "epoch": 2.12, "learning_rate": 3.675067608173077e-05, "loss": 0.2508, "step": 7059 }, { "epoch": 2.12, "learning_rate": 3.674879807692308e-05, "loss": 0.2996, "step": 7060 }, { "epoch": 2.12, "learning_rate": 3.674692007211539e-05, "loss": 0.3548, "step": 7061 }, { "epoch": 2.12, "learning_rate": 3.6745042067307694e-05, "loss": 0.3226, "step": 7062 }, { "epoch": 2.12, "learning_rate": 3.67431640625e-05, "loss": 0.3366, "step": 7063 }, { "epoch": 2.12, "learning_rate": 3.674128605769231e-05, "loss": 0.2361, "step": 7064 }, { "epoch": 2.12, "learning_rate": 3.6739408052884615e-05, "loss": 0.254, "step": 7065 }, { "epoch": 2.12, "learning_rate": 3.673753004807693e-05, "loss": 0.2973, "step": 7066 }, { "epoch": 2.12, "learning_rate": 3.673565204326923e-05, "loss": 0.3912, "step": 7067 }, { "epoch": 2.12, "learning_rate": 3.6733774038461536e-05, "loss": 0.4172, "step": 7068 }, { "epoch": 2.12, "learning_rate": 3.673189603365385e-05, "loss": 0.4356, "step": 7069 }, { "epoch": 2.12, "learning_rate": 3.673001802884616e-05, "loss": 0.3927, "step": 7070 }, { "epoch": 2.12, "learning_rate": 3.6728140024038464e-05, "loss": 0.2976, "step": 7071 }, { "epoch": 2.12, "learning_rate": 3.6726262019230776e-05, "loss": 0.3208, "step": 7072 }, { "epoch": 2.13, "learning_rate": 3.672438401442308e-05, "loss": 0.5479, "step": 7073 }, { "epoch": 2.13, "learning_rate": 3.6722506009615385e-05, "loss": 0.196, "step": 7074 }, { "epoch": 2.13, "learning_rate": 3.67206280048077e-05, "loss": 0.3777, "step": 7075 }, { "epoch": 2.13, "learning_rate": 3.671875e-05, "loss": 0.3122, "step": 7076 }, { "epoch": 2.13, "learning_rate": 3.6716871995192306e-05, "loss": 0.2783, "step": 7077 }, { "epoch": 2.13, "learning_rate": 3.671499399038462e-05, "loss": 0.3149, "step": 7078 }, { "epoch": 2.13, "learning_rate": 3.671311598557692e-05, "loss": 0.4237, "step": 7079 }, { "epoch": 2.13, "learning_rate": 3.6711237980769234e-05, "loss": 0.225, "step": 7080 }, { "epoch": 2.13, "learning_rate": 3.670935997596154e-05, "loss": 0.3839, "step": 7081 }, { "epoch": 2.13, "learning_rate": 3.6707481971153844e-05, "loss": 0.3158, "step": 7082 }, { "epoch": 2.13, "learning_rate": 3.6705603966346155e-05, "loss": 0.2372, "step": 7083 }, { "epoch": 2.13, "learning_rate": 3.670372596153847e-05, "loss": 0.3837, "step": 7084 }, { "epoch": 2.13, "learning_rate": 3.670184795673077e-05, "loss": 0.3706, "step": 7085 }, { "epoch": 2.13, "learning_rate": 3.669996995192308e-05, "loss": 0.2882, "step": 7086 }, { "epoch": 2.13, "learning_rate": 3.669809194711539e-05, "loss": 0.2996, "step": 7087 }, { "epoch": 2.13, "learning_rate": 3.669621394230769e-05, "loss": 0.4637, "step": 7088 }, { "epoch": 2.13, "learning_rate": 3.6694335937500004e-05, "loss": 0.3667, "step": 7089 }, { "epoch": 2.13, "learning_rate": 3.669245793269231e-05, "loss": 0.2323, "step": 7090 }, { "epoch": 2.13, "learning_rate": 3.6690579927884614e-05, "loss": 0.3799, "step": 7091 }, { "epoch": 2.13, "learning_rate": 3.6688701923076925e-05, "loss": 0.2968, "step": 7092 }, { "epoch": 2.13, "learning_rate": 3.668682391826923e-05, "loss": 0.3594, "step": 7093 }, { "epoch": 2.13, "learning_rate": 3.6684945913461535e-05, "loss": 0.2949, "step": 7094 }, { "epoch": 2.13, "learning_rate": 3.6683067908653846e-05, "loss": 0.2917, "step": 7095 }, { "epoch": 2.13, "learning_rate": 3.668118990384616e-05, "loss": 0.3944, "step": 7096 }, { "epoch": 2.13, "learning_rate": 3.667931189903846e-05, "loss": 0.3658, "step": 7097 }, { "epoch": 2.13, "learning_rate": 3.6677433894230774e-05, "loss": 0.4786, "step": 7098 }, { "epoch": 2.13, "learning_rate": 3.667555588942308e-05, "loss": 0.2557, "step": 7099 }, { "epoch": 2.13, "learning_rate": 3.667367788461539e-05, "loss": 0.441, "step": 7100 }, { "epoch": 2.13, "learning_rate": 3.6671799879807695e-05, "loss": 0.3966, "step": 7101 }, { "epoch": 2.13, "learning_rate": 3.6669921875e-05, "loss": 0.2746, "step": 7102 }, { "epoch": 2.13, "learning_rate": 3.666804387019231e-05, "loss": 0.2921, "step": 7103 }, { "epoch": 2.13, "learning_rate": 3.6666165865384616e-05, "loss": 0.3033, "step": 7104 }, { "epoch": 2.13, "learning_rate": 3.666428786057692e-05, "loss": 0.3136, "step": 7105 }, { "epoch": 2.14, "learning_rate": 3.666240985576923e-05, "loss": 0.4796, "step": 7106 }, { "epoch": 2.14, "learning_rate": 3.666053185096154e-05, "loss": 0.3356, "step": 7107 }, { "epoch": 2.14, "learning_rate": 3.665865384615384e-05, "loss": 0.3647, "step": 7108 }, { "epoch": 2.14, "learning_rate": 3.665677584134616e-05, "loss": 0.2211, "step": 7109 }, { "epoch": 2.14, "learning_rate": 3.6654897836538465e-05, "loss": 0.298, "step": 7110 }, { "epoch": 2.14, "learning_rate": 3.665301983173077e-05, "loss": 0.3603, "step": 7111 }, { "epoch": 2.14, "learning_rate": 3.665114182692308e-05, "loss": 0.369, "step": 7112 }, { "epoch": 2.14, "learning_rate": 3.6649263822115386e-05, "loss": 0.4985, "step": 7113 }, { "epoch": 2.14, "learning_rate": 3.664738581730769e-05, "loss": 0.3204, "step": 7114 }, { "epoch": 2.14, "learning_rate": 3.66455078125e-05, "loss": 0.3157, "step": 7115 }, { "epoch": 2.14, "learning_rate": 3.664362980769231e-05, "loss": 0.3158, "step": 7116 }, { "epoch": 2.14, "learning_rate": 3.664175180288462e-05, "loss": 0.2994, "step": 7117 }, { "epoch": 2.14, "learning_rate": 3.6639873798076924e-05, "loss": 0.376, "step": 7118 }, { "epoch": 2.14, "learning_rate": 3.663799579326923e-05, "loss": 0.446, "step": 7119 }, { "epoch": 2.14, "learning_rate": 3.663611778846154e-05, "loss": 0.2768, "step": 7120 }, { "epoch": 2.14, "learning_rate": 3.6634239783653845e-05, "loss": 0.372, "step": 7121 }, { "epoch": 2.14, "learning_rate": 3.6632361778846156e-05, "loss": 0.3584, "step": 7122 }, { "epoch": 2.14, "learning_rate": 3.663048377403847e-05, "loss": 0.273, "step": 7123 }, { "epoch": 2.14, "learning_rate": 3.662860576923077e-05, "loss": 0.2209, "step": 7124 }, { "epoch": 2.14, "learning_rate": 3.662672776442308e-05, "loss": 0.3905, "step": 7125 }, { "epoch": 2.14, "learning_rate": 3.662484975961539e-05, "loss": 0.242, "step": 7126 }, { "epoch": 2.14, "learning_rate": 3.6622971754807694e-05, "loss": 0.2871, "step": 7127 }, { "epoch": 2.14, "learning_rate": 3.662109375e-05, "loss": 0.2397, "step": 7128 }, { "epoch": 2.14, "learning_rate": 3.661921574519231e-05, "loss": 0.2365, "step": 7129 }, { "epoch": 2.14, "learning_rate": 3.6617337740384615e-05, "loss": 0.31, "step": 7130 }, { "epoch": 2.14, "learning_rate": 3.6615459735576927e-05, "loss": 0.4425, "step": 7131 }, { "epoch": 2.14, "learning_rate": 3.661358173076923e-05, "loss": 0.282, "step": 7132 }, { "epoch": 2.14, "learning_rate": 3.6611703725961536e-05, "loss": 0.4889, "step": 7133 }, { "epoch": 2.14, "learning_rate": 3.660982572115385e-05, "loss": 0.2597, "step": 7134 }, { "epoch": 2.14, "learning_rate": 3.660794771634616e-05, "loss": 0.3741, "step": 7135 }, { "epoch": 2.14, "learning_rate": 3.6606069711538464e-05, "loss": 0.3605, "step": 7136 }, { "epoch": 2.14, "learning_rate": 3.6604191706730775e-05, "loss": 0.2706, "step": 7137 }, { "epoch": 2.14, "learning_rate": 3.660231370192308e-05, "loss": 0.2775, "step": 7138 }, { "epoch": 2.15, "learning_rate": 3.6600435697115385e-05, "loss": 0.3126, "step": 7139 }, { "epoch": 2.15, "learning_rate": 3.6598557692307697e-05, "loss": 0.4099, "step": 7140 }, { "epoch": 2.15, "learning_rate": 3.65966796875e-05, "loss": 0.2902, "step": 7141 }, { "epoch": 2.15, "learning_rate": 3.6594801682692306e-05, "loss": 0.3032, "step": 7142 }, { "epoch": 2.15, "learning_rate": 3.659292367788462e-05, "loss": 0.3821, "step": 7143 }, { "epoch": 2.15, "learning_rate": 3.659104567307692e-05, "loss": 0.3697, "step": 7144 }, { "epoch": 2.15, "learning_rate": 3.6589167668269234e-05, "loss": 0.286, "step": 7145 }, { "epoch": 2.15, "learning_rate": 3.658728966346154e-05, "loss": 0.4522, "step": 7146 }, { "epoch": 2.15, "learning_rate": 3.6585411658653844e-05, "loss": 0.3173, "step": 7147 }, { "epoch": 2.15, "learning_rate": 3.6583533653846155e-05, "loss": 0.2612, "step": 7148 }, { "epoch": 2.15, "learning_rate": 3.6581655649038467e-05, "loss": 0.3271, "step": 7149 }, { "epoch": 2.15, "learning_rate": 3.657977764423077e-05, "loss": 0.4925, "step": 7150 }, { "epoch": 2.15, "learning_rate": 3.657789963942308e-05, "loss": 0.3152, "step": 7151 }, { "epoch": 2.15, "learning_rate": 3.657602163461539e-05, "loss": 0.37, "step": 7152 }, { "epoch": 2.15, "learning_rate": 3.657414362980769e-05, "loss": 0.3047, "step": 7153 }, { "epoch": 2.15, "learning_rate": 3.6572265625000004e-05, "loss": 0.3467, "step": 7154 }, { "epoch": 2.15, "learning_rate": 3.657038762019231e-05, "loss": 0.2404, "step": 7155 }, { "epoch": 2.15, "learning_rate": 3.6568509615384614e-05, "loss": 0.2354, "step": 7156 }, { "epoch": 2.15, "learning_rate": 3.6566631610576925e-05, "loss": 0.2704, "step": 7157 }, { "epoch": 2.15, "learning_rate": 3.656475360576923e-05, "loss": 0.2556, "step": 7158 }, { "epoch": 2.15, "learning_rate": 3.656287560096154e-05, "loss": 0.2999, "step": 7159 }, { "epoch": 2.15, "learning_rate": 3.6560997596153846e-05, "loss": 0.3896, "step": 7160 }, { "epoch": 2.15, "learning_rate": 3.655911959134616e-05, "loss": 0.244, "step": 7161 }, { "epoch": 2.15, "learning_rate": 3.655724158653846e-05, "loss": 0.3933, "step": 7162 }, { "epoch": 2.15, "learning_rate": 3.6555363581730774e-05, "loss": 0.3593, "step": 7163 }, { "epoch": 2.15, "learning_rate": 3.655348557692308e-05, "loss": 0.3062, "step": 7164 }, { "epoch": 2.15, "learning_rate": 3.655160757211539e-05, "loss": 0.351, "step": 7165 }, { "epoch": 2.15, "learning_rate": 3.6549729567307695e-05, "loss": 0.2808, "step": 7166 }, { "epoch": 2.15, "learning_rate": 3.65478515625e-05, "loss": 0.283, "step": 7167 }, { "epoch": 2.15, "learning_rate": 3.654597355769231e-05, "loss": 0.3367, "step": 7168 }, { "epoch": 2.15, "learning_rate": 3.6544095552884616e-05, "loss": 0.3844, "step": 7169 }, { "epoch": 2.15, "learning_rate": 3.654221754807692e-05, "loss": 0.2922, "step": 7170 }, { "epoch": 2.15, "learning_rate": 3.654033954326923e-05, "loss": 0.3516, "step": 7171 }, { "epoch": 2.16, "learning_rate": 3.653846153846154e-05, "loss": 0.3535, "step": 7172 }, { "epoch": 2.16, "learning_rate": 3.653658353365384e-05, "loss": 0.3048, "step": 7173 }, { "epoch": 2.16, "learning_rate": 3.653470552884616e-05, "loss": 0.3352, "step": 7174 }, { "epoch": 2.16, "learning_rate": 3.6532827524038465e-05, "loss": 0.4205, "step": 7175 }, { "epoch": 2.16, "learning_rate": 3.653094951923077e-05, "loss": 0.3042, "step": 7176 }, { "epoch": 2.16, "learning_rate": 3.652907151442308e-05, "loss": 0.4428, "step": 7177 }, { "epoch": 2.16, "learning_rate": 3.6527193509615386e-05, "loss": 0.2407, "step": 7178 }, { "epoch": 2.16, "learning_rate": 3.65253155048077e-05, "loss": 0.4028, "step": 7179 }, { "epoch": 2.16, "learning_rate": 3.65234375e-05, "loss": 0.3937, "step": 7180 }, { "epoch": 2.16, "learning_rate": 3.652155949519231e-05, "loss": 0.3526, "step": 7181 }, { "epoch": 2.16, "learning_rate": 3.651968149038462e-05, "loss": 0.2616, "step": 7182 }, { "epoch": 2.16, "learning_rate": 3.6517803485576924e-05, "loss": 0.3003, "step": 7183 }, { "epoch": 2.16, "learning_rate": 3.651592548076923e-05, "loss": 0.4103, "step": 7184 }, { "epoch": 2.16, "learning_rate": 3.651404747596154e-05, "loss": 0.4625, "step": 7185 }, { "epoch": 2.16, "learning_rate": 3.6512169471153845e-05, "loss": 0.2952, "step": 7186 }, { "epoch": 2.16, "learning_rate": 3.6510291466346156e-05, "loss": 0.3487, "step": 7187 }, { "epoch": 2.16, "learning_rate": 3.650841346153847e-05, "loss": 0.3592, "step": 7188 }, { "epoch": 2.16, "learning_rate": 3.650653545673077e-05, "loss": 0.5368, "step": 7189 }, { "epoch": 2.16, "learning_rate": 3.650465745192308e-05, "loss": 0.3335, "step": 7190 }, { "epoch": 2.16, "learning_rate": 3.650277944711539e-05, "loss": 0.2978, "step": 7191 }, { "epoch": 2.16, "learning_rate": 3.6500901442307694e-05, "loss": 0.363, "step": 7192 }, { "epoch": 2.16, "learning_rate": 3.64990234375e-05, "loss": 0.4231, "step": 7193 }, { "epoch": 2.16, "learning_rate": 3.649714543269231e-05, "loss": 0.2821, "step": 7194 }, { "epoch": 2.16, "learning_rate": 3.6495267427884615e-05, "loss": 0.3236, "step": 7195 }, { "epoch": 2.16, "learning_rate": 3.6493389423076926e-05, "loss": 0.255, "step": 7196 }, { "epoch": 2.16, "learning_rate": 3.649151141826923e-05, "loss": 0.3427, "step": 7197 }, { "epoch": 2.16, "learning_rate": 3.6489633413461536e-05, "loss": 0.3092, "step": 7198 }, { "epoch": 2.16, "learning_rate": 3.648775540865385e-05, "loss": 0.3939, "step": 7199 }, { "epoch": 2.16, "learning_rate": 3.648587740384616e-05, "loss": 0.2907, "step": 7200 }, { "epoch": 2.16, "learning_rate": 3.6483999399038464e-05, "loss": 0.3535, "step": 7201 }, { "epoch": 2.16, "learning_rate": 3.6482121394230775e-05, "loss": 0.388, "step": 7202 }, { "epoch": 2.16, "learning_rate": 3.648024338942308e-05, "loss": 0.4675, "step": 7203 }, { "epoch": 2.16, "learning_rate": 3.6478365384615385e-05, "loss": 0.3215, "step": 7204 }, { "epoch": 2.16, "learning_rate": 3.6476487379807696e-05, "loss": 0.3626, "step": 7205 }, { "epoch": 2.17, "learning_rate": 3.6474609375e-05, "loss": 0.3501, "step": 7206 }, { "epoch": 2.17, "learning_rate": 3.6472731370192306e-05, "loss": 0.3131, "step": 7207 }, { "epoch": 2.17, "learning_rate": 3.647085336538462e-05, "loss": 0.3471, "step": 7208 }, { "epoch": 2.17, "learning_rate": 3.646897536057692e-05, "loss": 0.3913, "step": 7209 }, { "epoch": 2.17, "learning_rate": 3.6467097355769234e-05, "loss": 0.3446, "step": 7210 }, { "epoch": 2.17, "learning_rate": 3.646521935096154e-05, "loss": 0.3615, "step": 7211 }, { "epoch": 2.17, "learning_rate": 3.6463341346153843e-05, "loss": 0.3846, "step": 7212 }, { "epoch": 2.17, "learning_rate": 3.6461463341346155e-05, "loss": 0.3881, "step": 7213 }, { "epoch": 2.17, "learning_rate": 3.6459585336538467e-05, "loss": 0.379, "step": 7214 }, { "epoch": 2.17, "learning_rate": 3.645770733173077e-05, "loss": 0.3763, "step": 7215 }, { "epoch": 2.17, "learning_rate": 3.645582932692308e-05, "loss": 0.4587, "step": 7216 }, { "epoch": 2.17, "learning_rate": 3.645395132211539e-05, "loss": 0.3962, "step": 7217 }, { "epoch": 2.17, "learning_rate": 3.645207331730769e-05, "loss": 0.3666, "step": 7218 }, { "epoch": 2.17, "learning_rate": 3.6450195312500004e-05, "loss": 0.1984, "step": 7219 }, { "epoch": 2.17, "learning_rate": 3.644831730769231e-05, "loss": 0.3571, "step": 7220 }, { "epoch": 2.17, "learning_rate": 3.6446439302884613e-05, "loss": 0.3949, "step": 7221 }, { "epoch": 2.17, "learning_rate": 3.6444561298076925e-05, "loss": 0.4148, "step": 7222 }, { "epoch": 2.17, "learning_rate": 3.644268329326923e-05, "loss": 0.1854, "step": 7223 }, { "epoch": 2.17, "learning_rate": 3.644080528846154e-05, "loss": 0.3003, "step": 7224 }, { "epoch": 2.17, "learning_rate": 3.6438927283653846e-05, "loss": 0.2883, "step": 7225 }, { "epoch": 2.17, "learning_rate": 3.643704927884616e-05, "loss": 0.4884, "step": 7226 }, { "epoch": 2.17, "learning_rate": 3.643517127403846e-05, "loss": 0.4202, "step": 7227 }, { "epoch": 2.17, "learning_rate": 3.6433293269230774e-05, "loss": 0.3398, "step": 7228 }, { "epoch": 2.17, "learning_rate": 3.643141526442308e-05, "loss": 0.3099, "step": 7229 }, { "epoch": 2.17, "learning_rate": 3.642953725961539e-05, "loss": 0.4253, "step": 7230 }, { "epoch": 2.17, "learning_rate": 3.6427659254807695e-05, "loss": 0.3558, "step": 7231 }, { "epoch": 2.17, "learning_rate": 3.642578125e-05, "loss": 0.2239, "step": 7232 }, { "epoch": 2.17, "learning_rate": 3.642390324519231e-05, "loss": 0.3667, "step": 7233 }, { "epoch": 2.17, "learning_rate": 3.6422025240384616e-05, "loss": 0.4072, "step": 7234 }, { "epoch": 2.17, "learning_rate": 3.642014723557692e-05, "loss": 0.3474, "step": 7235 }, { "epoch": 2.17, "learning_rate": 3.641826923076923e-05, "loss": 0.6335, "step": 7236 }, { "epoch": 2.17, "learning_rate": 3.641639122596154e-05, "loss": 0.3578, "step": 7237 }, { "epoch": 2.17, "learning_rate": 3.641451322115384e-05, "loss": 0.3279, "step": 7238 }, { "epoch": 2.18, "learning_rate": 3.641263521634616e-05, "loss": 0.3574, "step": 7239 }, { "epoch": 2.18, "learning_rate": 3.6410757211538465e-05, "loss": 0.2776, "step": 7240 }, { "epoch": 2.18, "learning_rate": 3.640887920673077e-05, "loss": 0.266, "step": 7241 }, { "epoch": 2.18, "learning_rate": 3.640700120192308e-05, "loss": 0.3133, "step": 7242 }, { "epoch": 2.18, "learning_rate": 3.6405123197115386e-05, "loss": 0.3251, "step": 7243 }, { "epoch": 2.18, "learning_rate": 3.64032451923077e-05, "loss": 0.2368, "step": 7244 }, { "epoch": 2.18, "learning_rate": 3.64013671875e-05, "loss": 0.3169, "step": 7245 }, { "epoch": 2.18, "learning_rate": 3.639948918269231e-05, "loss": 0.3226, "step": 7246 }, { "epoch": 2.18, "learning_rate": 3.639761117788462e-05, "loss": 0.2771, "step": 7247 }, { "epoch": 2.18, "learning_rate": 3.6395733173076924e-05, "loss": 0.3874, "step": 7248 }, { "epoch": 2.18, "learning_rate": 3.639385516826923e-05, "loss": 0.3669, "step": 7249 }, { "epoch": 2.18, "learning_rate": 3.639197716346154e-05, "loss": 0.5199, "step": 7250 }, { "epoch": 2.18, "learning_rate": 3.6390099158653845e-05, "loss": 0.3184, "step": 7251 }, { "epoch": 2.18, "learning_rate": 3.6388221153846156e-05, "loss": 0.3338, "step": 7252 }, { "epoch": 2.18, "learning_rate": 3.638634314903847e-05, "loss": 0.4069, "step": 7253 }, { "epoch": 2.18, "learning_rate": 3.638446514423077e-05, "loss": 0.2985, "step": 7254 }, { "epoch": 2.18, "learning_rate": 3.638258713942308e-05, "loss": 0.344, "step": 7255 }, { "epoch": 2.18, "learning_rate": 3.638070913461539e-05, "loss": 0.2678, "step": 7256 }, { "epoch": 2.18, "learning_rate": 3.6378831129807694e-05, "loss": 0.3425, "step": 7257 }, { "epoch": 2.18, "learning_rate": 3.6376953125e-05, "loss": 0.3114, "step": 7258 }, { "epoch": 2.18, "learning_rate": 3.637507512019231e-05, "loss": 0.2995, "step": 7259 }, { "epoch": 2.18, "learning_rate": 3.6373197115384615e-05, "loss": 0.4456, "step": 7260 }, { "epoch": 2.18, "learning_rate": 3.6371319110576926e-05, "loss": 0.3129, "step": 7261 }, { "epoch": 2.18, "learning_rate": 3.636944110576923e-05, "loss": 0.362, "step": 7262 }, { "epoch": 2.18, "learning_rate": 3.6367563100961536e-05, "loss": 0.4172, "step": 7263 }, { "epoch": 2.18, "learning_rate": 3.636568509615385e-05, "loss": 0.3023, "step": 7264 }, { "epoch": 2.18, "learning_rate": 3.636380709134616e-05, "loss": 0.2487, "step": 7265 }, { "epoch": 2.18, "learning_rate": 3.6361929086538464e-05, "loss": 0.4017, "step": 7266 }, { "epoch": 2.18, "learning_rate": 3.6360051081730775e-05, "loss": 0.3196, "step": 7267 }, { "epoch": 2.18, "learning_rate": 3.635817307692308e-05, "loss": 0.3549, "step": 7268 }, { "epoch": 2.18, "learning_rate": 3.6356295072115385e-05, "loss": 0.2498, "step": 7269 }, { "epoch": 2.18, "learning_rate": 3.6354417067307696e-05, "loss": 0.2707, "step": 7270 }, { "epoch": 2.18, "learning_rate": 3.63525390625e-05, "loss": 0.2596, "step": 7271 }, { "epoch": 2.19, "learning_rate": 3.6350661057692306e-05, "loss": 0.3581, "step": 7272 }, { "epoch": 2.19, "learning_rate": 3.634878305288462e-05, "loss": 0.2864, "step": 7273 }, { "epoch": 2.19, "learning_rate": 3.634690504807692e-05, "loss": 0.3961, "step": 7274 }, { "epoch": 2.19, "learning_rate": 3.6345027043269234e-05, "loss": 0.4392, "step": 7275 }, { "epoch": 2.19, "learning_rate": 3.634314903846154e-05, "loss": 0.3608, "step": 7276 }, { "epoch": 2.19, "learning_rate": 3.634127103365384e-05, "loss": 0.3458, "step": 7277 }, { "epoch": 2.19, "learning_rate": 3.6339393028846155e-05, "loss": 0.3445, "step": 7278 }, { "epoch": 2.19, "learning_rate": 3.6337515024038466e-05, "loss": 0.3915, "step": 7279 }, { "epoch": 2.19, "learning_rate": 3.633563701923077e-05, "loss": 0.3792, "step": 7280 }, { "epoch": 2.19, "learning_rate": 3.633375901442308e-05, "loss": 0.3071, "step": 7281 }, { "epoch": 2.19, "learning_rate": 3.633188100961539e-05, "loss": 0.4407, "step": 7282 }, { "epoch": 2.19, "learning_rate": 3.633000300480769e-05, "loss": 0.4064, "step": 7283 }, { "epoch": 2.19, "learning_rate": 3.6328125000000004e-05, "loss": 0.3097, "step": 7284 }, { "epoch": 2.19, "learning_rate": 3.632624699519231e-05, "loss": 0.5022, "step": 7285 }, { "epoch": 2.19, "learning_rate": 3.632436899038461e-05, "loss": 0.2292, "step": 7286 }, { "epoch": 2.19, "learning_rate": 3.6322490985576925e-05, "loss": 0.4197, "step": 7287 }, { "epoch": 2.19, "learning_rate": 3.632061298076923e-05, "loss": 0.3244, "step": 7288 }, { "epoch": 2.19, "learning_rate": 3.631873497596154e-05, "loss": 0.4377, "step": 7289 }, { "epoch": 2.19, "learning_rate": 3.6316856971153846e-05, "loss": 0.4937, "step": 7290 }, { "epoch": 2.19, "learning_rate": 3.631497896634616e-05, "loss": 0.3441, "step": 7291 }, { "epoch": 2.19, "learning_rate": 3.631310096153846e-05, "loss": 0.34, "step": 7292 }, { "epoch": 2.19, "learning_rate": 3.6311222956730774e-05, "loss": 0.3689, "step": 7293 }, { "epoch": 2.19, "learning_rate": 3.630934495192308e-05, "loss": 0.34, "step": 7294 }, { "epoch": 2.19, "learning_rate": 3.630746694711539e-05, "loss": 0.3756, "step": 7295 }, { "epoch": 2.19, "learning_rate": 3.6305588942307695e-05, "loss": 0.3991, "step": 7296 }, { "epoch": 2.19, "learning_rate": 3.63037109375e-05, "loss": 0.3293, "step": 7297 }, { "epoch": 2.19, "learning_rate": 3.630183293269231e-05, "loss": 0.4006, "step": 7298 }, { "epoch": 2.19, "learning_rate": 3.6299954927884616e-05, "loss": 0.3621, "step": 7299 }, { "epoch": 2.19, "learning_rate": 3.629807692307692e-05, "loss": 0.3588, "step": 7300 }, { "epoch": 2.19, "learning_rate": 3.629619891826923e-05, "loss": 0.3271, "step": 7301 }, { "epoch": 2.19, "learning_rate": 3.629432091346154e-05, "loss": 0.3415, "step": 7302 }, { "epoch": 2.19, "learning_rate": 3.629244290865385e-05, "loss": 0.346, "step": 7303 }, { "epoch": 2.19, "learning_rate": 3.629056490384616e-05, "loss": 0.3053, "step": 7304 }, { "epoch": 2.2, "learning_rate": 3.6288686899038465e-05, "loss": 0.2704, "step": 7305 }, { "epoch": 2.2, "learning_rate": 3.628680889423077e-05, "loss": 0.3708, "step": 7306 }, { "epoch": 2.2, "learning_rate": 3.628493088942308e-05, "loss": 0.3015, "step": 7307 }, { "epoch": 2.2, "learning_rate": 3.6283052884615386e-05, "loss": 0.4601, "step": 7308 }, { "epoch": 2.2, "learning_rate": 3.62811748798077e-05, "loss": 0.2517, "step": 7309 }, { "epoch": 2.2, "learning_rate": 3.6279296875e-05, "loss": 0.7274, "step": 7310 }, { "epoch": 2.2, "learning_rate": 3.627741887019231e-05, "loss": 0.2902, "step": 7311 }, { "epoch": 2.2, "learning_rate": 3.627554086538462e-05, "loss": 0.3188, "step": 7312 }, { "epoch": 2.2, "learning_rate": 3.6273662860576923e-05, "loss": 0.3019, "step": 7313 }, { "epoch": 2.2, "learning_rate": 3.627178485576923e-05, "loss": 0.4311, "step": 7314 }, { "epoch": 2.2, "learning_rate": 3.626990685096154e-05, "loss": 0.492, "step": 7315 }, { "epoch": 2.2, "learning_rate": 3.6268028846153845e-05, "loss": 0.3833, "step": 7316 }, { "epoch": 2.2, "learning_rate": 3.6266150841346156e-05, "loss": 0.3806, "step": 7317 }, { "epoch": 2.2, "learning_rate": 3.626427283653847e-05, "loss": 0.4995, "step": 7318 }, { "epoch": 2.2, "learning_rate": 3.626239483173077e-05, "loss": 0.3542, "step": 7319 }, { "epoch": 2.2, "learning_rate": 3.626051682692308e-05, "loss": 0.3118, "step": 7320 }, { "epoch": 2.2, "learning_rate": 3.625863882211539e-05, "loss": 0.3273, "step": 7321 }, { "epoch": 2.2, "learning_rate": 3.6256760817307694e-05, "loss": 0.3286, "step": 7322 }, { "epoch": 2.2, "learning_rate": 3.6254882812500005e-05, "loss": 0.3133, "step": 7323 }, { "epoch": 2.2, "learning_rate": 3.625300480769231e-05, "loss": 0.3083, "step": 7324 }, { "epoch": 2.2, "learning_rate": 3.6251126802884615e-05, "loss": 0.3042, "step": 7325 }, { "epoch": 2.2, "learning_rate": 3.6249248798076926e-05, "loss": 0.3655, "step": 7326 }, { "epoch": 2.2, "learning_rate": 3.624737079326923e-05, "loss": 0.2999, "step": 7327 }, { "epoch": 2.2, "learning_rate": 3.6245492788461536e-05, "loss": 0.3556, "step": 7328 }, { "epoch": 2.2, "learning_rate": 3.624361478365385e-05, "loss": 0.3568, "step": 7329 }, { "epoch": 2.2, "learning_rate": 3.624173677884616e-05, "loss": 0.4889, "step": 7330 }, { "epoch": 2.2, "learning_rate": 3.6239858774038464e-05, "loss": 0.3419, "step": 7331 }, { "epoch": 2.2, "learning_rate": 3.6237980769230775e-05, "loss": 0.4625, "step": 7332 }, { "epoch": 2.2, "learning_rate": 3.623610276442308e-05, "loss": 0.352, "step": 7333 }, { "epoch": 2.2, "learning_rate": 3.6234224759615385e-05, "loss": 0.2431, "step": 7334 }, { "epoch": 2.2, "learning_rate": 3.6232346754807696e-05, "loss": 0.2445, "step": 7335 }, { "epoch": 2.2, "learning_rate": 3.623046875e-05, "loss": 0.2595, "step": 7336 }, { "epoch": 2.2, "learning_rate": 3.6228590745192306e-05, "loss": 0.4268, "step": 7337 }, { "epoch": 2.2, "learning_rate": 3.622671274038462e-05, "loss": 0.3423, "step": 7338 }, { "epoch": 2.21, "learning_rate": 3.622483473557692e-05, "loss": 0.4371, "step": 7339 }, { "epoch": 2.21, "learning_rate": 3.6222956730769234e-05, "loss": 0.3687, "step": 7340 }, { "epoch": 2.21, "learning_rate": 3.622107872596154e-05, "loss": 0.2736, "step": 7341 }, { "epoch": 2.21, "learning_rate": 3.621920072115384e-05, "loss": 0.2689, "step": 7342 }, { "epoch": 2.21, "learning_rate": 3.621732271634616e-05, "loss": 0.3427, "step": 7343 }, { "epoch": 2.21, "learning_rate": 3.6215444711538466e-05, "loss": 0.3956, "step": 7344 }, { "epoch": 2.21, "learning_rate": 3.621356670673077e-05, "loss": 0.3105, "step": 7345 }, { "epoch": 2.21, "learning_rate": 3.621168870192308e-05, "loss": 0.2619, "step": 7346 }, { "epoch": 2.21, "learning_rate": 3.620981069711539e-05, "loss": 0.2984, "step": 7347 }, { "epoch": 2.21, "learning_rate": 3.620793269230769e-05, "loss": 0.3876, "step": 7348 }, { "epoch": 2.21, "learning_rate": 3.6206054687500004e-05, "loss": 0.3059, "step": 7349 }, { "epoch": 2.21, "learning_rate": 3.620417668269231e-05, "loss": 0.3716, "step": 7350 }, { "epoch": 2.21, "learning_rate": 3.620229867788461e-05, "loss": 0.2106, "step": 7351 }, { "epoch": 2.21, "learning_rate": 3.6200420673076925e-05, "loss": 0.347, "step": 7352 }, { "epoch": 2.21, "learning_rate": 3.619854266826923e-05, "loss": 0.4347, "step": 7353 }, { "epoch": 2.21, "learning_rate": 3.619666466346154e-05, "loss": 0.293, "step": 7354 }, { "epoch": 2.21, "learning_rate": 3.6194786658653846e-05, "loss": 0.4389, "step": 7355 }, { "epoch": 2.21, "learning_rate": 3.619290865384616e-05, "loss": 0.3809, "step": 7356 }, { "epoch": 2.21, "learning_rate": 3.619103064903846e-05, "loss": 0.3391, "step": 7357 }, { "epoch": 2.21, "learning_rate": 3.6189152644230774e-05, "loss": 0.3906, "step": 7358 }, { "epoch": 2.21, "learning_rate": 3.618727463942308e-05, "loss": 0.3729, "step": 7359 }, { "epoch": 2.21, "learning_rate": 3.618539663461539e-05, "loss": 0.2687, "step": 7360 }, { "epoch": 2.21, "learning_rate": 3.6183518629807695e-05, "loss": 0.289, "step": 7361 }, { "epoch": 2.21, "learning_rate": 3.6181640625e-05, "loss": 0.3301, "step": 7362 }, { "epoch": 2.21, "learning_rate": 3.617976262019231e-05, "loss": 0.2415, "step": 7363 }, { "epoch": 2.21, "learning_rate": 3.6177884615384616e-05, "loss": 0.4756, "step": 7364 }, { "epoch": 2.21, "learning_rate": 3.617600661057692e-05, "loss": 0.2303, "step": 7365 }, { "epoch": 2.21, "learning_rate": 3.617412860576923e-05, "loss": 0.3487, "step": 7366 }, { "epoch": 2.21, "learning_rate": 3.617225060096154e-05, "loss": 0.3935, "step": 7367 }, { "epoch": 2.21, "learning_rate": 3.617037259615385e-05, "loss": 0.4414, "step": 7368 }, { "epoch": 2.21, "learning_rate": 3.616849459134616e-05, "loss": 0.3055, "step": 7369 }, { "epoch": 2.21, "learning_rate": 3.6166616586538465e-05, "loss": 0.4469, "step": 7370 }, { "epoch": 2.21, "learning_rate": 3.616473858173077e-05, "loss": 0.3765, "step": 7371 }, { "epoch": 2.22, "learning_rate": 3.616286057692308e-05, "loss": 0.4211, "step": 7372 }, { "epoch": 2.22, "learning_rate": 3.6160982572115386e-05, "loss": 0.4524, "step": 7373 }, { "epoch": 2.22, "learning_rate": 3.61591045673077e-05, "loss": 0.3115, "step": 7374 }, { "epoch": 2.22, "learning_rate": 3.61572265625e-05, "loss": 0.2234, "step": 7375 }, { "epoch": 2.22, "learning_rate": 3.615534855769231e-05, "loss": 0.4075, "step": 7376 }, { "epoch": 2.22, "learning_rate": 3.615347055288462e-05, "loss": 0.2709, "step": 7377 }, { "epoch": 2.22, "learning_rate": 3.6151592548076923e-05, "loss": 0.2957, "step": 7378 }, { "epoch": 2.22, "learning_rate": 3.614971454326923e-05, "loss": 0.3062, "step": 7379 }, { "epoch": 2.22, "learning_rate": 3.614783653846154e-05, "loss": 0.3075, "step": 7380 }, { "epoch": 2.22, "learning_rate": 3.6145958533653844e-05, "loss": 0.3354, "step": 7381 }, { "epoch": 2.22, "learning_rate": 3.6144080528846156e-05, "loss": 0.2982, "step": 7382 }, { "epoch": 2.22, "learning_rate": 3.614220252403847e-05, "loss": 0.3697, "step": 7383 }, { "epoch": 2.22, "learning_rate": 3.614032451923077e-05, "loss": 0.3404, "step": 7384 }, { "epoch": 2.22, "learning_rate": 3.613844651442308e-05, "loss": 0.2768, "step": 7385 }, { "epoch": 2.22, "learning_rate": 3.613656850961539e-05, "loss": 0.4859, "step": 7386 }, { "epoch": 2.22, "learning_rate": 3.6134690504807693e-05, "loss": 0.3759, "step": 7387 }, { "epoch": 2.22, "learning_rate": 3.6132812500000005e-05, "loss": 0.1769, "step": 7388 }, { "epoch": 2.22, "learning_rate": 3.613093449519231e-05, "loss": 0.3883, "step": 7389 }, { "epoch": 2.22, "learning_rate": 3.6129056490384615e-05, "loss": 0.3433, "step": 7390 }, { "epoch": 2.22, "learning_rate": 3.6127178485576926e-05, "loss": 0.3703, "step": 7391 }, { "epoch": 2.22, "learning_rate": 3.612530048076923e-05, "loss": 0.3524, "step": 7392 }, { "epoch": 2.22, "learning_rate": 3.6123422475961536e-05, "loss": 0.287, "step": 7393 }, { "epoch": 2.22, "learning_rate": 3.612154447115385e-05, "loss": 0.2572, "step": 7394 }, { "epoch": 2.22, "learning_rate": 3.611966646634616e-05, "loss": 0.3746, "step": 7395 }, { "epoch": 2.22, "learning_rate": 3.6117788461538463e-05, "loss": 0.3965, "step": 7396 }, { "epoch": 2.22, "learning_rate": 3.6115910456730775e-05, "loss": 0.332, "step": 7397 }, { "epoch": 2.22, "learning_rate": 3.611403245192308e-05, "loss": 0.437, "step": 7398 }, { "epoch": 2.22, "learning_rate": 3.6112154447115385e-05, "loss": 0.3567, "step": 7399 }, { "epoch": 2.22, "learning_rate": 3.6110276442307696e-05, "loss": 0.4037, "step": 7400 }, { "epoch": 2.22, "learning_rate": 3.61083984375e-05, "loss": 0.3428, "step": 7401 }, { "epoch": 2.22, "learning_rate": 3.6106520432692306e-05, "loss": 0.3552, "step": 7402 }, { "epoch": 2.22, "learning_rate": 3.610464242788462e-05, "loss": 0.3428, "step": 7403 }, { "epoch": 2.22, "learning_rate": 3.610276442307692e-05, "loss": 0.34, "step": 7404 }, { "epoch": 2.23, "learning_rate": 3.6100886418269234e-05, "loss": 0.1936, "step": 7405 }, { "epoch": 2.23, "learning_rate": 3.609900841346154e-05, "loss": 0.3086, "step": 7406 }, { "epoch": 2.23, "learning_rate": 3.609713040865384e-05, "loss": 0.3433, "step": 7407 }, { "epoch": 2.23, "learning_rate": 3.609525240384616e-05, "loss": 0.2791, "step": 7408 }, { "epoch": 2.23, "learning_rate": 3.6093374399038466e-05, "loss": 0.2658, "step": 7409 }, { "epoch": 2.23, "learning_rate": 3.609149639423077e-05, "loss": 0.2685, "step": 7410 }, { "epoch": 2.23, "learning_rate": 3.608961838942308e-05, "loss": 0.3676, "step": 7411 }, { "epoch": 2.23, "learning_rate": 3.608774038461539e-05, "loss": 0.3931, "step": 7412 }, { "epoch": 2.23, "learning_rate": 3.608586237980769e-05, "loss": 0.3812, "step": 7413 }, { "epoch": 2.23, "learning_rate": 3.6083984375000004e-05, "loss": 0.4381, "step": 7414 }, { "epoch": 2.23, "learning_rate": 3.608210637019231e-05, "loss": 0.2299, "step": 7415 }, { "epoch": 2.23, "learning_rate": 3.608022836538461e-05, "loss": 0.346, "step": 7416 }, { "epoch": 2.23, "learning_rate": 3.6078350360576925e-05, "loss": 0.3169, "step": 7417 }, { "epoch": 2.23, "learning_rate": 3.607647235576923e-05, "loss": 0.429, "step": 7418 }, { "epoch": 2.23, "learning_rate": 3.607459435096154e-05, "loss": 0.3867, "step": 7419 }, { "epoch": 2.23, "learning_rate": 3.6072716346153846e-05, "loss": 0.4056, "step": 7420 }, { "epoch": 2.23, "learning_rate": 3.607083834134616e-05, "loss": 0.3732, "step": 7421 }, { "epoch": 2.23, "learning_rate": 3.606896033653846e-05, "loss": 0.325, "step": 7422 }, { "epoch": 2.23, "learning_rate": 3.6067082331730774e-05, "loss": 0.3292, "step": 7423 }, { "epoch": 2.23, "learning_rate": 3.606520432692308e-05, "loss": 0.3509, "step": 7424 }, { "epoch": 2.23, "learning_rate": 3.606332632211539e-05, "loss": 0.3143, "step": 7425 }, { "epoch": 2.23, "learning_rate": 3.6061448317307695e-05, "loss": 0.1628, "step": 7426 }, { "epoch": 2.23, "learning_rate": 3.60595703125e-05, "loss": 0.2836, "step": 7427 }, { "epoch": 2.23, "learning_rate": 3.605769230769231e-05, "loss": 0.3903, "step": 7428 }, { "epoch": 2.23, "learning_rate": 3.6055814302884616e-05, "loss": 0.2859, "step": 7429 }, { "epoch": 2.23, "learning_rate": 3.605393629807692e-05, "loss": 0.4239, "step": 7430 }, { "epoch": 2.23, "learning_rate": 3.605205829326923e-05, "loss": 0.2923, "step": 7431 }, { "epoch": 2.23, "learning_rate": 3.605018028846154e-05, "loss": 0.438, "step": 7432 }, { "epoch": 2.23, "learning_rate": 3.604830228365385e-05, "loss": 0.3728, "step": 7433 }, { "epoch": 2.23, "learning_rate": 3.604642427884616e-05, "loss": 0.2447, "step": 7434 }, { "epoch": 2.23, "learning_rate": 3.6044546274038465e-05, "loss": 0.2849, "step": 7435 }, { "epoch": 2.23, "learning_rate": 3.604266826923077e-05, "loss": 0.3101, "step": 7436 }, { "epoch": 2.23, "learning_rate": 3.604079026442308e-05, "loss": 0.2862, "step": 7437 }, { "epoch": 2.23, "learning_rate": 3.6038912259615386e-05, "loss": 0.3364, "step": 7438 }, { "epoch": 2.24, "learning_rate": 3.60370342548077e-05, "loss": 0.2747, "step": 7439 }, { "epoch": 2.24, "learning_rate": 3.603515625e-05, "loss": 0.4402, "step": 7440 }, { "epoch": 2.24, "learning_rate": 3.603327824519231e-05, "loss": 0.2793, "step": 7441 }, { "epoch": 2.24, "learning_rate": 3.603140024038462e-05, "loss": 0.4364, "step": 7442 }, { "epoch": 2.24, "learning_rate": 3.602952223557692e-05, "loss": 0.4487, "step": 7443 }, { "epoch": 2.24, "learning_rate": 3.602764423076923e-05, "loss": 0.367, "step": 7444 }, { "epoch": 2.24, "learning_rate": 3.602576622596154e-05, "loss": 0.4658, "step": 7445 }, { "epoch": 2.24, "learning_rate": 3.6023888221153844e-05, "loss": 0.3976, "step": 7446 }, { "epoch": 2.24, "learning_rate": 3.6022010216346156e-05, "loss": 0.3232, "step": 7447 }, { "epoch": 2.24, "learning_rate": 3.602013221153847e-05, "loss": 0.4412, "step": 7448 }, { "epoch": 2.24, "learning_rate": 3.601825420673077e-05, "loss": 0.3491, "step": 7449 }, { "epoch": 2.24, "learning_rate": 3.601637620192308e-05, "loss": 0.3874, "step": 7450 }, { "epoch": 2.24, "learning_rate": 3.601449819711539e-05, "loss": 0.4385, "step": 7451 }, { "epoch": 2.24, "learning_rate": 3.601262019230769e-05, "loss": 0.3282, "step": 7452 }, { "epoch": 2.24, "learning_rate": 3.6010742187500005e-05, "loss": 0.2841, "step": 7453 }, { "epoch": 2.24, "learning_rate": 3.600886418269231e-05, "loss": 0.3588, "step": 7454 }, { "epoch": 2.24, "learning_rate": 3.6006986177884614e-05, "loss": 0.286, "step": 7455 }, { "epoch": 2.24, "learning_rate": 3.6005108173076926e-05, "loss": 0.3358, "step": 7456 }, { "epoch": 2.24, "learning_rate": 3.600323016826923e-05, "loss": 0.3401, "step": 7457 }, { "epoch": 2.24, "learning_rate": 3.6001352163461536e-05, "loss": 0.3355, "step": 7458 }, { "epoch": 2.24, "learning_rate": 3.599947415865385e-05, "loss": 0.4177, "step": 7459 }, { "epoch": 2.24, "learning_rate": 3.599759615384616e-05, "loss": 0.3825, "step": 7460 }, { "epoch": 2.24, "learning_rate": 3.599571814903846e-05, "loss": 0.343, "step": 7461 }, { "epoch": 2.24, "learning_rate": 3.5993840144230775e-05, "loss": 0.3268, "step": 7462 }, { "epoch": 2.24, "learning_rate": 3.599196213942308e-05, "loss": 0.3098, "step": 7463 }, { "epoch": 2.24, "learning_rate": 3.5990084134615384e-05, "loss": 0.3491, "step": 7464 }, { "epoch": 2.24, "learning_rate": 3.5988206129807696e-05, "loss": 0.3798, "step": 7465 }, { "epoch": 2.24, "learning_rate": 3.5986328125e-05, "loss": 0.2294, "step": 7466 }, { "epoch": 2.24, "learning_rate": 3.598445012019231e-05, "loss": 0.5051, "step": 7467 }, { "epoch": 2.24, "learning_rate": 3.598257211538462e-05, "loss": 0.2745, "step": 7468 }, { "epoch": 2.24, "learning_rate": 3.598069411057692e-05, "loss": 0.3277, "step": 7469 }, { "epoch": 2.24, "learning_rate": 3.5978816105769233e-05, "loss": 0.331, "step": 7470 }, { "epoch": 2.24, "learning_rate": 3.597693810096154e-05, "loss": 0.4264, "step": 7471 }, { "epoch": 2.25, "learning_rate": 3.597506009615384e-05, "loss": 0.3581, "step": 7472 }, { "epoch": 2.25, "learning_rate": 3.597318209134616e-05, "loss": 0.3578, "step": 7473 }, { "epoch": 2.25, "learning_rate": 3.5971304086538466e-05, "loss": 0.3893, "step": 7474 }, { "epoch": 2.25, "learning_rate": 3.596942608173077e-05, "loss": 0.3347, "step": 7475 }, { "epoch": 2.25, "learning_rate": 3.596754807692308e-05, "loss": 0.3435, "step": 7476 }, { "epoch": 2.25, "learning_rate": 3.596567007211539e-05, "loss": 0.2778, "step": 7477 }, { "epoch": 2.25, "learning_rate": 3.596379206730769e-05, "loss": 0.4489, "step": 7478 }, { "epoch": 2.25, "learning_rate": 3.5961914062500003e-05, "loss": 0.3644, "step": 7479 }, { "epoch": 2.25, "learning_rate": 3.596003605769231e-05, "loss": 0.3424, "step": 7480 }, { "epoch": 2.25, "learning_rate": 3.595815805288461e-05, "loss": 0.4442, "step": 7481 }, { "epoch": 2.25, "learning_rate": 3.5956280048076925e-05, "loss": 0.2945, "step": 7482 }, { "epoch": 2.25, "learning_rate": 3.595440204326923e-05, "loss": 0.2782, "step": 7483 }, { "epoch": 2.25, "learning_rate": 3.595252403846154e-05, "loss": 0.3152, "step": 7484 }, { "epoch": 2.25, "learning_rate": 3.5950646033653846e-05, "loss": 0.3243, "step": 7485 }, { "epoch": 2.25, "learning_rate": 3.594876802884616e-05, "loss": 0.2274, "step": 7486 }, { "epoch": 2.25, "learning_rate": 3.594689002403847e-05, "loss": 0.2735, "step": 7487 }, { "epoch": 2.25, "learning_rate": 3.5945012019230774e-05, "loss": 0.4841, "step": 7488 }, { "epoch": 2.25, "learning_rate": 3.594313401442308e-05, "loss": 0.3763, "step": 7489 }, { "epoch": 2.25, "learning_rate": 3.594125600961539e-05, "loss": 0.3528, "step": 7490 }, { "epoch": 2.25, "learning_rate": 3.5939378004807695e-05, "loss": 0.4003, "step": 7491 }, { "epoch": 2.25, "learning_rate": 3.59375e-05, "loss": 0.3544, "step": 7492 }, { "epoch": 2.25, "learning_rate": 3.593562199519231e-05, "loss": 0.3604, "step": 7493 }, { "epoch": 2.25, "learning_rate": 3.5933743990384616e-05, "loss": 0.3442, "step": 7494 }, { "epoch": 2.25, "learning_rate": 3.593186598557692e-05, "loss": 0.2645, "step": 7495 }, { "epoch": 2.25, "learning_rate": 3.592998798076923e-05, "loss": 0.4224, "step": 7496 }, { "epoch": 2.25, "learning_rate": 3.592810997596154e-05, "loss": 0.4225, "step": 7497 }, { "epoch": 2.25, "learning_rate": 3.592623197115385e-05, "loss": 0.2797, "step": 7498 }, { "epoch": 2.25, "learning_rate": 3.592435396634616e-05, "loss": 0.333, "step": 7499 }, { "epoch": 2.25, "learning_rate": 3.5922475961538465e-05, "loss": 0.3114, "step": 7500 }, { "epoch": 2.25, "learning_rate": 3.592059795673077e-05, "loss": 0.256, "step": 7501 }, { "epoch": 2.25, "learning_rate": 3.591871995192308e-05, "loss": 0.4108, "step": 7502 }, { "epoch": 2.25, "learning_rate": 3.5916841947115386e-05, "loss": 0.3542, "step": 7503 }, { "epoch": 2.25, "learning_rate": 3.59149639423077e-05, "loss": 0.3501, "step": 7504 }, { "epoch": 2.26, "learning_rate": 3.59130859375e-05, "loss": 0.3984, "step": 7505 }, { "epoch": 2.26, "learning_rate": 3.591120793269231e-05, "loss": 0.4863, "step": 7506 }, { "epoch": 2.26, "learning_rate": 3.590932992788462e-05, "loss": 0.4378, "step": 7507 }, { "epoch": 2.26, "learning_rate": 3.590745192307692e-05, "loss": 0.3277, "step": 7508 }, { "epoch": 2.26, "learning_rate": 3.590557391826923e-05, "loss": 0.436, "step": 7509 }, { "epoch": 2.26, "learning_rate": 3.590369591346154e-05, "loss": 0.2859, "step": 7510 }, { "epoch": 2.26, "learning_rate": 3.5901817908653844e-05, "loss": 0.6163, "step": 7511 }, { "epoch": 2.26, "learning_rate": 3.5899939903846156e-05, "loss": 0.3538, "step": 7512 }, { "epoch": 2.26, "learning_rate": 3.589806189903847e-05, "loss": 0.307, "step": 7513 }, { "epoch": 2.26, "learning_rate": 3.589618389423077e-05, "loss": 0.3863, "step": 7514 }, { "epoch": 2.26, "learning_rate": 3.589430588942308e-05, "loss": 0.4247, "step": 7515 }, { "epoch": 2.26, "learning_rate": 3.589242788461539e-05, "loss": 0.395, "step": 7516 }, { "epoch": 2.26, "learning_rate": 3.589054987980769e-05, "loss": 0.431, "step": 7517 }, { "epoch": 2.26, "learning_rate": 3.5888671875000005e-05, "loss": 0.3358, "step": 7518 }, { "epoch": 2.26, "learning_rate": 3.588679387019231e-05, "loss": 0.5511, "step": 7519 }, { "epoch": 2.26, "learning_rate": 3.5884915865384614e-05, "loss": 0.2935, "step": 7520 }, { "epoch": 2.26, "learning_rate": 3.5883037860576926e-05, "loss": 0.3445, "step": 7521 }, { "epoch": 2.26, "learning_rate": 3.588115985576923e-05, "loss": 0.3748, "step": 7522 }, { "epoch": 2.26, "learning_rate": 3.5879281850961535e-05, "loss": 0.3992, "step": 7523 }, { "epoch": 2.26, "learning_rate": 3.587740384615385e-05, "loss": 0.3941, "step": 7524 }, { "epoch": 2.26, "learning_rate": 3.587552584134616e-05, "loss": 0.3885, "step": 7525 }, { "epoch": 2.26, "learning_rate": 3.587364783653846e-05, "loss": 0.3414, "step": 7526 }, { "epoch": 2.26, "learning_rate": 3.5871769831730775e-05, "loss": 0.2956, "step": 7527 }, { "epoch": 2.26, "learning_rate": 3.586989182692308e-05, "loss": 0.3707, "step": 7528 }, { "epoch": 2.26, "learning_rate": 3.5868013822115384e-05, "loss": 0.3248, "step": 7529 }, { "epoch": 2.26, "learning_rate": 3.5866135817307696e-05, "loss": 0.5112, "step": 7530 }, { "epoch": 2.26, "learning_rate": 3.58642578125e-05, "loss": 0.4071, "step": 7531 }, { "epoch": 2.26, "learning_rate": 3.586237980769231e-05, "loss": 0.3815, "step": 7532 }, { "epoch": 2.26, "learning_rate": 3.586050180288462e-05, "loss": 0.3608, "step": 7533 }, { "epoch": 2.26, "learning_rate": 3.585862379807692e-05, "loss": 0.3939, "step": 7534 }, { "epoch": 2.26, "learning_rate": 3.585674579326923e-05, "loss": 0.3463, "step": 7535 }, { "epoch": 2.26, "learning_rate": 3.585486778846154e-05, "loss": 0.2445, "step": 7536 }, { "epoch": 2.26, "learning_rate": 3.585298978365384e-05, "loss": 0.3578, "step": 7537 }, { "epoch": 2.27, "learning_rate": 3.585111177884616e-05, "loss": 0.4967, "step": 7538 }, { "epoch": 2.27, "learning_rate": 3.5849233774038466e-05, "loss": 0.3433, "step": 7539 }, { "epoch": 2.27, "learning_rate": 3.584735576923077e-05, "loss": 0.3663, "step": 7540 }, { "epoch": 2.27, "learning_rate": 3.584547776442308e-05, "loss": 0.4124, "step": 7541 }, { "epoch": 2.27, "learning_rate": 3.584359975961539e-05, "loss": 0.4589, "step": 7542 }, { "epoch": 2.27, "learning_rate": 3.584172175480769e-05, "loss": 0.4128, "step": 7543 }, { "epoch": 2.27, "learning_rate": 3.583984375e-05, "loss": 0.4438, "step": 7544 }, { "epoch": 2.27, "learning_rate": 3.583796574519231e-05, "loss": 0.3512, "step": 7545 }, { "epoch": 2.27, "learning_rate": 3.583608774038461e-05, "loss": 0.4053, "step": 7546 }, { "epoch": 2.27, "learning_rate": 3.5834209735576924e-05, "loss": 0.3598, "step": 7547 }, { "epoch": 2.27, "learning_rate": 3.583233173076923e-05, "loss": 0.3751, "step": 7548 }, { "epoch": 2.27, "learning_rate": 3.583045372596154e-05, "loss": 0.3119, "step": 7549 }, { "epoch": 2.27, "learning_rate": 3.5828575721153846e-05, "loss": 0.4045, "step": 7550 }, { "epoch": 2.27, "learning_rate": 3.582669771634616e-05, "loss": 0.2926, "step": 7551 }, { "epoch": 2.27, "learning_rate": 3.582481971153847e-05, "loss": 0.2994, "step": 7552 }, { "epoch": 2.27, "learning_rate": 3.5822941706730773e-05, "loss": 0.2984, "step": 7553 }, { "epoch": 2.27, "learning_rate": 3.582106370192308e-05, "loss": 0.3456, "step": 7554 }, { "epoch": 2.27, "learning_rate": 3.581918569711539e-05, "loss": 0.4274, "step": 7555 }, { "epoch": 2.27, "learning_rate": 3.5817307692307695e-05, "loss": 0.3537, "step": 7556 }, { "epoch": 2.27, "learning_rate": 3.58154296875e-05, "loss": 0.3478, "step": 7557 }, { "epoch": 2.27, "learning_rate": 3.581355168269231e-05, "loss": 0.3402, "step": 7558 }, { "epoch": 2.27, "learning_rate": 3.5811673677884616e-05, "loss": 0.4272, "step": 7559 }, { "epoch": 2.27, "learning_rate": 3.580979567307692e-05, "loss": 0.3508, "step": 7560 }, { "epoch": 2.27, "learning_rate": 3.580791766826923e-05, "loss": 0.3451, "step": 7561 }, { "epoch": 2.27, "learning_rate": 3.580603966346154e-05, "loss": 0.2527, "step": 7562 }, { "epoch": 2.27, "learning_rate": 3.580416165865385e-05, "loss": 0.3673, "step": 7563 }, { "epoch": 2.27, "learning_rate": 3.580228365384616e-05, "loss": 0.2951, "step": 7564 }, { "epoch": 2.27, "learning_rate": 3.5800405649038465e-05, "loss": 0.343, "step": 7565 }, { "epoch": 2.27, "learning_rate": 3.579852764423077e-05, "loss": 0.2833, "step": 7566 }, { "epoch": 2.27, "learning_rate": 3.579664963942308e-05, "loss": 0.2818, "step": 7567 }, { "epoch": 2.27, "learning_rate": 3.5794771634615386e-05, "loss": 0.4881, "step": 7568 }, { "epoch": 2.27, "learning_rate": 3.57928936298077e-05, "loss": 0.2503, "step": 7569 }, { "epoch": 2.27, "learning_rate": 3.5791015625e-05, "loss": 0.2925, "step": 7570 }, { "epoch": 2.27, "learning_rate": 3.578913762019231e-05, "loss": 0.4805, "step": 7571 }, { "epoch": 2.28, "learning_rate": 3.578725961538462e-05, "loss": 0.4634, "step": 7572 }, { "epoch": 2.28, "learning_rate": 3.578538161057692e-05, "loss": 0.3442, "step": 7573 }, { "epoch": 2.28, "learning_rate": 3.578350360576923e-05, "loss": 0.4158, "step": 7574 }, { "epoch": 2.28, "learning_rate": 3.578162560096154e-05, "loss": 0.4479, "step": 7575 }, { "epoch": 2.28, "learning_rate": 3.5779747596153844e-05, "loss": 0.471, "step": 7576 }, { "epoch": 2.28, "learning_rate": 3.5777869591346156e-05, "loss": 0.483, "step": 7577 }, { "epoch": 2.28, "learning_rate": 3.577599158653847e-05, "loss": 0.3387, "step": 7578 }, { "epoch": 2.28, "learning_rate": 3.577411358173077e-05, "loss": 0.3258, "step": 7579 }, { "epoch": 2.28, "learning_rate": 3.577223557692308e-05, "loss": 0.3223, "step": 7580 }, { "epoch": 2.28, "learning_rate": 3.577035757211539e-05, "loss": 0.6215, "step": 7581 }, { "epoch": 2.28, "learning_rate": 3.576847956730769e-05, "loss": 0.3363, "step": 7582 }, { "epoch": 2.28, "learning_rate": 3.5766601562500005e-05, "loss": 0.3629, "step": 7583 }, { "epoch": 2.28, "learning_rate": 3.576472355769231e-05, "loss": 0.3522, "step": 7584 }, { "epoch": 2.28, "learning_rate": 3.5762845552884614e-05, "loss": 0.4168, "step": 7585 }, { "epoch": 2.28, "learning_rate": 3.5760967548076926e-05, "loss": 0.2883, "step": 7586 }, { "epoch": 2.28, "learning_rate": 3.575908954326923e-05, "loss": 0.263, "step": 7587 }, { "epoch": 2.28, "learning_rate": 3.5757211538461535e-05, "loss": 0.3187, "step": 7588 }, { "epoch": 2.28, "learning_rate": 3.575533353365385e-05, "loss": 0.3795, "step": 7589 }, { "epoch": 2.28, "learning_rate": 3.575345552884616e-05, "loss": 0.4859, "step": 7590 }, { "epoch": 2.28, "learning_rate": 3.575157752403846e-05, "loss": 0.3041, "step": 7591 }, { "epoch": 2.28, "learning_rate": 3.5749699519230775e-05, "loss": 0.4795, "step": 7592 }, { "epoch": 2.28, "learning_rate": 3.574782151442308e-05, "loss": 0.2984, "step": 7593 }, { "epoch": 2.28, "learning_rate": 3.5745943509615384e-05, "loss": 0.3078, "step": 7594 }, { "epoch": 2.28, "learning_rate": 3.5744065504807696e-05, "loss": 0.4724, "step": 7595 }, { "epoch": 2.28, "learning_rate": 3.57421875e-05, "loss": 0.3683, "step": 7596 }, { "epoch": 2.28, "learning_rate": 3.574030949519231e-05, "loss": 0.3061, "step": 7597 }, { "epoch": 2.28, "learning_rate": 3.573843149038462e-05, "loss": 0.316, "step": 7598 }, { "epoch": 2.28, "learning_rate": 3.573655348557692e-05, "loss": 0.2241, "step": 7599 }, { "epoch": 2.28, "learning_rate": 3.573467548076923e-05, "loss": 0.3242, "step": 7600 }, { "epoch": 2.28, "learning_rate": 3.573279747596154e-05, "loss": 0.2889, "step": 7601 }, { "epoch": 2.28, "learning_rate": 3.573091947115384e-05, "loss": 0.535, "step": 7602 }, { "epoch": 2.28, "learning_rate": 3.572904146634616e-05, "loss": 0.4253, "step": 7603 }, { "epoch": 2.28, "learning_rate": 3.5727163461538466e-05, "loss": 0.3222, "step": 7604 }, { "epoch": 2.29, "learning_rate": 3.572528545673077e-05, "loss": 0.4918, "step": 7605 }, { "epoch": 2.29, "learning_rate": 3.572340745192308e-05, "loss": 0.3464, "step": 7606 }, { "epoch": 2.29, "learning_rate": 3.572152944711539e-05, "loss": 0.2704, "step": 7607 }, { "epoch": 2.29, "learning_rate": 3.571965144230769e-05, "loss": 0.4161, "step": 7608 }, { "epoch": 2.29, "learning_rate": 3.57177734375e-05, "loss": 0.2455, "step": 7609 }, { "epoch": 2.29, "learning_rate": 3.571589543269231e-05, "loss": 0.2783, "step": 7610 }, { "epoch": 2.29, "learning_rate": 3.571401742788462e-05, "loss": 0.3964, "step": 7611 }, { "epoch": 2.29, "learning_rate": 3.5712139423076924e-05, "loss": 0.3449, "step": 7612 }, { "epoch": 2.29, "learning_rate": 3.571026141826923e-05, "loss": 0.3976, "step": 7613 }, { "epoch": 2.29, "learning_rate": 3.570838341346154e-05, "loss": 0.3919, "step": 7614 }, { "epoch": 2.29, "learning_rate": 3.5706505408653845e-05, "loss": 0.3829, "step": 7615 }, { "epoch": 2.29, "learning_rate": 3.570462740384616e-05, "loss": 0.2997, "step": 7616 }, { "epoch": 2.29, "learning_rate": 3.570274939903847e-05, "loss": 0.3045, "step": 7617 }, { "epoch": 2.29, "learning_rate": 3.570087139423077e-05, "loss": 0.4083, "step": 7618 }, { "epoch": 2.29, "learning_rate": 3.569899338942308e-05, "loss": 0.3232, "step": 7619 }, { "epoch": 2.29, "learning_rate": 3.569711538461539e-05, "loss": 0.2433, "step": 7620 }, { "epoch": 2.29, "learning_rate": 3.5695237379807694e-05, "loss": 0.4534, "step": 7621 }, { "epoch": 2.29, "learning_rate": 3.5693359375e-05, "loss": 0.4642, "step": 7622 }, { "epoch": 2.29, "learning_rate": 3.569148137019231e-05, "loss": 0.265, "step": 7623 }, { "epoch": 2.29, "learning_rate": 3.5689603365384615e-05, "loss": 0.362, "step": 7624 }, { "epoch": 2.29, "learning_rate": 3.568772536057692e-05, "loss": 0.4111, "step": 7625 }, { "epoch": 2.29, "learning_rate": 3.568584735576923e-05, "loss": 0.4499, "step": 7626 }, { "epoch": 2.29, "learning_rate": 3.5683969350961537e-05, "loss": 0.319, "step": 7627 }, { "epoch": 2.29, "learning_rate": 3.568209134615385e-05, "loss": 0.429, "step": 7628 }, { "epoch": 2.29, "learning_rate": 3.568021334134616e-05, "loss": 0.4176, "step": 7629 }, { "epoch": 2.29, "learning_rate": 3.5678335336538464e-05, "loss": 0.415, "step": 7630 }, { "epoch": 2.29, "learning_rate": 3.5676457331730776e-05, "loss": 0.4309, "step": 7631 }, { "epoch": 2.29, "learning_rate": 3.567457932692308e-05, "loss": 0.3016, "step": 7632 }, { "epoch": 2.29, "learning_rate": 3.5672701322115386e-05, "loss": 0.3185, "step": 7633 }, { "epoch": 2.29, "learning_rate": 3.56708233173077e-05, "loss": 0.3822, "step": 7634 }, { "epoch": 2.29, "learning_rate": 3.56689453125e-05, "loss": 0.4068, "step": 7635 }, { "epoch": 2.29, "learning_rate": 3.566706730769231e-05, "loss": 0.2822, "step": 7636 }, { "epoch": 2.29, "learning_rate": 3.566518930288462e-05, "loss": 0.3899, "step": 7637 }, { "epoch": 2.3, "learning_rate": 3.566331129807692e-05, "loss": 0.3701, "step": 7638 }, { "epoch": 2.3, "learning_rate": 3.566143329326923e-05, "loss": 0.5238, "step": 7639 }, { "epoch": 2.3, "learning_rate": 3.565955528846154e-05, "loss": 0.2967, "step": 7640 }, { "epoch": 2.3, "learning_rate": 3.5657677283653844e-05, "loss": 0.2978, "step": 7641 }, { "epoch": 2.3, "learning_rate": 3.5655799278846156e-05, "loss": 0.3099, "step": 7642 }, { "epoch": 2.3, "learning_rate": 3.565392127403847e-05, "loss": 0.4147, "step": 7643 }, { "epoch": 2.3, "learning_rate": 3.565204326923077e-05, "loss": 0.3117, "step": 7644 }, { "epoch": 2.3, "learning_rate": 3.565016526442308e-05, "loss": 0.3446, "step": 7645 }, { "epoch": 2.3, "learning_rate": 3.564828725961539e-05, "loss": 0.4263, "step": 7646 }, { "epoch": 2.3, "learning_rate": 3.564640925480769e-05, "loss": 0.3077, "step": 7647 }, { "epoch": 2.3, "learning_rate": 3.5644531250000005e-05, "loss": 0.3327, "step": 7648 }, { "epoch": 2.3, "learning_rate": 3.564265324519231e-05, "loss": 0.3733, "step": 7649 }, { "epoch": 2.3, "learning_rate": 3.5640775240384614e-05, "loss": 0.2877, "step": 7650 }, { "epoch": 2.3, "learning_rate": 3.5638897235576926e-05, "loss": 0.5935, "step": 7651 }, { "epoch": 2.3, "learning_rate": 3.563701923076923e-05, "loss": 0.4066, "step": 7652 }, { "epoch": 2.3, "learning_rate": 3.5635141225961535e-05, "loss": 0.4013, "step": 7653 }, { "epoch": 2.3, "learning_rate": 3.563326322115385e-05, "loss": 0.473, "step": 7654 }, { "epoch": 2.3, "learning_rate": 3.563138521634616e-05, "loss": 0.5788, "step": 7655 }, { "epoch": 2.3, "learning_rate": 3.562950721153846e-05, "loss": 0.2694, "step": 7656 }, { "epoch": 2.3, "learning_rate": 3.5627629206730775e-05, "loss": 0.4106, "step": 7657 }, { "epoch": 2.3, "learning_rate": 3.562575120192308e-05, "loss": 0.2926, "step": 7658 }, { "epoch": 2.3, "learning_rate": 3.5623873197115384e-05, "loss": 0.3872, "step": 7659 }, { "epoch": 2.3, "learning_rate": 3.5621995192307696e-05, "loss": 0.3523, "step": 7660 }, { "epoch": 2.3, "learning_rate": 3.56201171875e-05, "loss": 0.3559, "step": 7661 }, { "epoch": 2.3, "learning_rate": 3.561823918269231e-05, "loss": 0.4236, "step": 7662 }, { "epoch": 2.3, "learning_rate": 3.561636117788462e-05, "loss": 0.4107, "step": 7663 }, { "epoch": 2.3, "learning_rate": 3.561448317307692e-05, "loss": 0.3375, "step": 7664 }, { "epoch": 2.3, "learning_rate": 3.561260516826923e-05, "loss": 0.4014, "step": 7665 }, { "epoch": 2.3, "learning_rate": 3.561072716346154e-05, "loss": 0.3166, "step": 7666 }, { "epoch": 2.3, "learning_rate": 3.560884915865384e-05, "loss": 0.3051, "step": 7667 }, { "epoch": 2.3, "learning_rate": 3.560697115384616e-05, "loss": 0.3415, "step": 7668 }, { "epoch": 2.3, "learning_rate": 3.5605093149038466e-05, "loss": 0.344, "step": 7669 }, { "epoch": 2.3, "learning_rate": 3.560321514423077e-05, "loss": 0.451, "step": 7670 }, { "epoch": 2.3, "learning_rate": 3.560133713942308e-05, "loss": 0.2469, "step": 7671 }, { "epoch": 2.31, "learning_rate": 3.559945913461539e-05, "loss": 0.2727, "step": 7672 }, { "epoch": 2.31, "learning_rate": 3.559758112980769e-05, "loss": 0.1991, "step": 7673 }, { "epoch": 2.31, "learning_rate": 3.5595703125e-05, "loss": 0.3924, "step": 7674 }, { "epoch": 2.31, "learning_rate": 3.559382512019231e-05, "loss": 0.3437, "step": 7675 }, { "epoch": 2.31, "learning_rate": 3.559194711538462e-05, "loss": 0.3259, "step": 7676 }, { "epoch": 2.31, "learning_rate": 3.5590069110576924e-05, "loss": 0.406, "step": 7677 }, { "epoch": 2.31, "learning_rate": 3.558819110576923e-05, "loss": 0.3496, "step": 7678 }, { "epoch": 2.31, "learning_rate": 3.558631310096154e-05, "loss": 0.5486, "step": 7679 }, { "epoch": 2.31, "learning_rate": 3.5584435096153845e-05, "loss": 0.3874, "step": 7680 }, { "epoch": 2.31, "learning_rate": 3.558255709134616e-05, "loss": 0.2989, "step": 7681 }, { "epoch": 2.31, "learning_rate": 3.558067908653847e-05, "loss": 0.365, "step": 7682 }, { "epoch": 2.31, "learning_rate": 3.557880108173077e-05, "loss": 0.3318, "step": 7683 }, { "epoch": 2.31, "learning_rate": 3.557692307692308e-05, "loss": 0.2922, "step": 7684 }, { "epoch": 2.31, "learning_rate": 3.557504507211539e-05, "loss": 0.2465, "step": 7685 }, { "epoch": 2.31, "learning_rate": 3.5573167067307694e-05, "loss": 0.3595, "step": 7686 }, { "epoch": 2.31, "learning_rate": 3.55712890625e-05, "loss": 0.3615, "step": 7687 }, { "epoch": 2.31, "learning_rate": 3.556941105769231e-05, "loss": 0.3014, "step": 7688 }, { "epoch": 2.31, "learning_rate": 3.5567533052884615e-05, "loss": 0.2151, "step": 7689 }, { "epoch": 2.31, "learning_rate": 3.556565504807692e-05, "loss": 0.3585, "step": 7690 }, { "epoch": 2.31, "learning_rate": 3.556377704326923e-05, "loss": 0.2696, "step": 7691 }, { "epoch": 2.31, "learning_rate": 3.5561899038461536e-05, "loss": 0.2922, "step": 7692 }, { "epoch": 2.31, "learning_rate": 3.556002103365385e-05, "loss": 0.3721, "step": 7693 }, { "epoch": 2.31, "learning_rate": 3.555814302884616e-05, "loss": 0.302, "step": 7694 }, { "epoch": 2.31, "learning_rate": 3.5556265024038464e-05, "loss": 0.4442, "step": 7695 }, { "epoch": 2.31, "learning_rate": 3.5554387019230776e-05, "loss": 0.3612, "step": 7696 }, { "epoch": 2.31, "learning_rate": 3.555250901442308e-05, "loss": 0.4576, "step": 7697 }, { "epoch": 2.31, "learning_rate": 3.5550631009615385e-05, "loss": 0.2535, "step": 7698 }, { "epoch": 2.31, "learning_rate": 3.55487530048077e-05, "loss": 0.2375, "step": 7699 }, { "epoch": 2.31, "learning_rate": 3.5546875e-05, "loss": 0.2924, "step": 7700 }, { "epoch": 2.31, "learning_rate": 3.5544996995192307e-05, "loss": 0.4295, "step": 7701 }, { "epoch": 2.31, "learning_rate": 3.554311899038462e-05, "loss": 0.3654, "step": 7702 }, { "epoch": 2.31, "learning_rate": 3.554124098557692e-05, "loss": 0.2155, "step": 7703 }, { "epoch": 2.31, "learning_rate": 3.553936298076923e-05, "loss": 0.314, "step": 7704 }, { "epoch": 2.32, "learning_rate": 3.553748497596154e-05, "loss": 0.3031, "step": 7705 }, { "epoch": 2.32, "learning_rate": 3.5535606971153844e-05, "loss": 0.3035, "step": 7706 }, { "epoch": 2.32, "learning_rate": 3.5533728966346155e-05, "loss": 0.3776, "step": 7707 }, { "epoch": 2.32, "learning_rate": 3.553185096153847e-05, "loss": 0.2647, "step": 7708 }, { "epoch": 2.32, "learning_rate": 3.552997295673077e-05, "loss": 0.4343, "step": 7709 }, { "epoch": 2.32, "learning_rate": 3.5528094951923077e-05, "loss": 0.5201, "step": 7710 }, { "epoch": 2.32, "learning_rate": 3.552621694711539e-05, "loss": 0.2866, "step": 7711 }, { "epoch": 2.32, "learning_rate": 3.552433894230769e-05, "loss": 0.3266, "step": 7712 }, { "epoch": 2.32, "learning_rate": 3.5522460937500004e-05, "loss": 0.3008, "step": 7713 }, { "epoch": 2.32, "learning_rate": 3.552058293269231e-05, "loss": 0.3657, "step": 7714 }, { "epoch": 2.32, "learning_rate": 3.5518704927884614e-05, "loss": 0.3221, "step": 7715 }, { "epoch": 2.32, "learning_rate": 3.5516826923076926e-05, "loss": 0.375, "step": 7716 }, { "epoch": 2.32, "learning_rate": 3.551494891826923e-05, "loss": 0.4124, "step": 7717 }, { "epoch": 2.32, "learning_rate": 3.5513070913461535e-05, "loss": 0.5475, "step": 7718 }, { "epoch": 2.32, "learning_rate": 3.551119290865385e-05, "loss": 0.4608, "step": 7719 }, { "epoch": 2.32, "learning_rate": 3.550931490384616e-05, "loss": 0.3545, "step": 7720 }, { "epoch": 2.32, "learning_rate": 3.550743689903846e-05, "loss": 0.3079, "step": 7721 }, { "epoch": 2.32, "learning_rate": 3.5505558894230774e-05, "loss": 0.2616, "step": 7722 }, { "epoch": 2.32, "learning_rate": 3.550368088942308e-05, "loss": 0.267, "step": 7723 }, { "epoch": 2.32, "learning_rate": 3.5501802884615384e-05, "loss": 0.2728, "step": 7724 }, { "epoch": 2.32, "learning_rate": 3.5499924879807696e-05, "loss": 0.2654, "step": 7725 }, { "epoch": 2.32, "learning_rate": 3.5498046875e-05, "loss": 0.5055, "step": 7726 }, { "epoch": 2.32, "learning_rate": 3.549616887019231e-05, "loss": 0.3708, "step": 7727 }, { "epoch": 2.32, "learning_rate": 3.549429086538462e-05, "loss": 0.2763, "step": 7728 }, { "epoch": 2.32, "learning_rate": 3.549241286057692e-05, "loss": 0.5158, "step": 7729 }, { "epoch": 2.32, "learning_rate": 3.549053485576923e-05, "loss": 0.3487, "step": 7730 }, { "epoch": 2.32, "learning_rate": 3.548865685096154e-05, "loss": 0.3975, "step": 7731 }, { "epoch": 2.32, "learning_rate": 3.548677884615384e-05, "loss": 0.2938, "step": 7732 }, { "epoch": 2.32, "learning_rate": 3.548490084134616e-05, "loss": 0.3137, "step": 7733 }, { "epoch": 2.32, "learning_rate": 3.5483022836538466e-05, "loss": 0.3184, "step": 7734 }, { "epoch": 2.32, "learning_rate": 3.548114483173077e-05, "loss": 0.3748, "step": 7735 }, { "epoch": 2.32, "learning_rate": 3.547926682692308e-05, "loss": 0.5, "step": 7736 }, { "epoch": 2.32, "learning_rate": 3.547738882211539e-05, "loss": 0.3766, "step": 7737 }, { "epoch": 2.33, "learning_rate": 3.547551081730769e-05, "loss": 0.5644, "step": 7738 }, { "epoch": 2.33, "learning_rate": 3.54736328125e-05, "loss": 0.3529, "step": 7739 }, { "epoch": 2.33, "learning_rate": 3.547175480769231e-05, "loss": 0.5177, "step": 7740 }, { "epoch": 2.33, "learning_rate": 3.546987680288462e-05, "loss": 0.4374, "step": 7741 }, { "epoch": 2.33, "learning_rate": 3.5467998798076924e-05, "loss": 0.3614, "step": 7742 }, { "epoch": 2.33, "learning_rate": 3.546612079326923e-05, "loss": 0.3895, "step": 7743 }, { "epoch": 2.33, "learning_rate": 3.546424278846154e-05, "loss": 0.3779, "step": 7744 }, { "epoch": 2.33, "learning_rate": 3.5462364783653845e-05, "loss": 0.3237, "step": 7745 }, { "epoch": 2.33, "learning_rate": 3.546048677884616e-05, "loss": 0.2961, "step": 7746 }, { "epoch": 2.33, "learning_rate": 3.545860877403847e-05, "loss": 0.3138, "step": 7747 }, { "epoch": 2.33, "learning_rate": 3.545673076923077e-05, "loss": 0.2719, "step": 7748 }, { "epoch": 2.33, "learning_rate": 3.545485276442308e-05, "loss": 0.3874, "step": 7749 }, { "epoch": 2.33, "learning_rate": 3.545297475961539e-05, "loss": 0.3586, "step": 7750 }, { "epoch": 2.33, "learning_rate": 3.5451096754807694e-05, "loss": 0.238, "step": 7751 }, { "epoch": 2.33, "learning_rate": 3.544921875e-05, "loss": 0.3976, "step": 7752 }, { "epoch": 2.33, "learning_rate": 3.544734074519231e-05, "loss": 0.3629, "step": 7753 }, { "epoch": 2.33, "learning_rate": 3.5445462740384615e-05, "loss": 0.3219, "step": 7754 }, { "epoch": 2.33, "learning_rate": 3.544358473557693e-05, "loss": 0.324, "step": 7755 }, { "epoch": 2.33, "learning_rate": 3.544170673076923e-05, "loss": 0.2647, "step": 7756 }, { "epoch": 2.33, "learning_rate": 3.5439828725961536e-05, "loss": 0.3402, "step": 7757 }, { "epoch": 2.33, "learning_rate": 3.543795072115385e-05, "loss": 0.4627, "step": 7758 }, { "epoch": 2.33, "learning_rate": 3.543607271634616e-05, "loss": 0.3854, "step": 7759 }, { "epoch": 2.33, "learning_rate": 3.5434194711538464e-05, "loss": 0.4088, "step": 7760 }, { "epoch": 2.33, "learning_rate": 3.5432316706730776e-05, "loss": 0.4134, "step": 7761 }, { "epoch": 2.33, "learning_rate": 3.543043870192308e-05, "loss": 0.5391, "step": 7762 }, { "epoch": 2.33, "learning_rate": 3.5428560697115385e-05, "loss": 0.424, "step": 7763 }, { "epoch": 2.33, "learning_rate": 3.54266826923077e-05, "loss": 0.2521, "step": 7764 }, { "epoch": 2.33, "learning_rate": 3.54248046875e-05, "loss": 0.2826, "step": 7765 }, { "epoch": 2.33, "learning_rate": 3.5422926682692306e-05, "loss": 0.2967, "step": 7766 }, { "epoch": 2.33, "learning_rate": 3.542104867788462e-05, "loss": 0.3404, "step": 7767 }, { "epoch": 2.33, "learning_rate": 3.541917067307692e-05, "loss": 0.308, "step": 7768 }, { "epoch": 2.33, "learning_rate": 3.541729266826923e-05, "loss": 0.3301, "step": 7769 }, { "epoch": 2.33, "learning_rate": 3.541541466346154e-05, "loss": 0.3085, "step": 7770 }, { "epoch": 2.34, "learning_rate": 3.5413536658653844e-05, "loss": 0.3129, "step": 7771 }, { "epoch": 2.34, "learning_rate": 3.5411658653846155e-05, "loss": 0.3061, "step": 7772 }, { "epoch": 2.34, "learning_rate": 3.540978064903847e-05, "loss": 0.391, "step": 7773 }, { "epoch": 2.34, "learning_rate": 3.540790264423077e-05, "loss": 0.3842, "step": 7774 }, { "epoch": 2.34, "learning_rate": 3.540602463942308e-05, "loss": 0.2077, "step": 7775 }, { "epoch": 2.34, "learning_rate": 3.540414663461539e-05, "loss": 0.4108, "step": 7776 }, { "epoch": 2.34, "learning_rate": 3.540226862980769e-05, "loss": 0.3412, "step": 7777 }, { "epoch": 2.34, "learning_rate": 3.5400390625000004e-05, "loss": 0.3732, "step": 7778 }, { "epoch": 2.34, "learning_rate": 3.539851262019231e-05, "loss": 0.2871, "step": 7779 }, { "epoch": 2.34, "learning_rate": 3.5396634615384614e-05, "loss": 0.3588, "step": 7780 }, { "epoch": 2.34, "learning_rate": 3.5394756610576925e-05, "loss": 0.2901, "step": 7781 }, { "epoch": 2.34, "learning_rate": 3.539287860576923e-05, "loss": 0.2371, "step": 7782 }, { "epoch": 2.34, "learning_rate": 3.5391000600961535e-05, "loss": 0.4074, "step": 7783 }, { "epoch": 2.34, "learning_rate": 3.5389122596153847e-05, "loss": 0.3135, "step": 7784 }, { "epoch": 2.34, "learning_rate": 3.538724459134616e-05, "loss": 0.459, "step": 7785 }, { "epoch": 2.34, "learning_rate": 3.538536658653846e-05, "loss": 0.3383, "step": 7786 }, { "epoch": 2.34, "learning_rate": 3.5383488581730774e-05, "loss": 0.235, "step": 7787 }, { "epoch": 2.34, "learning_rate": 3.538161057692308e-05, "loss": 0.2922, "step": 7788 }, { "epoch": 2.34, "learning_rate": 3.5379732572115384e-05, "loss": 0.314, "step": 7789 }, { "epoch": 2.34, "learning_rate": 3.5377854567307695e-05, "loss": 0.3833, "step": 7790 }, { "epoch": 2.34, "learning_rate": 3.53759765625e-05, "loss": 0.4794, "step": 7791 }, { "epoch": 2.34, "learning_rate": 3.537409855769231e-05, "loss": 0.3745, "step": 7792 }, { "epoch": 2.34, "learning_rate": 3.5372220552884617e-05, "loss": 0.2373, "step": 7793 }, { "epoch": 2.34, "learning_rate": 3.537034254807692e-05, "loss": 0.282, "step": 7794 }, { "epoch": 2.34, "learning_rate": 3.536846454326923e-05, "loss": 0.3152, "step": 7795 }, { "epoch": 2.34, "learning_rate": 3.536658653846154e-05, "loss": 0.3824, "step": 7796 }, { "epoch": 2.34, "learning_rate": 3.536470853365384e-05, "loss": 0.2624, "step": 7797 }, { "epoch": 2.34, "learning_rate": 3.536283052884616e-05, "loss": 0.3681, "step": 7798 }, { "epoch": 2.34, "learning_rate": 3.5360952524038466e-05, "loss": 0.3457, "step": 7799 }, { "epoch": 2.34, "learning_rate": 3.535907451923077e-05, "loss": 0.4024, "step": 7800 }, { "epoch": 2.34, "learning_rate": 3.535719651442308e-05, "loss": 0.441, "step": 7801 }, { "epoch": 2.34, "learning_rate": 3.535531850961539e-05, "loss": 0.3615, "step": 7802 }, { "epoch": 2.34, "learning_rate": 3.535344050480769e-05, "loss": 0.5178, "step": 7803 }, { "epoch": 2.34, "learning_rate": 3.53515625e-05, "loss": 0.3305, "step": 7804 }, { "epoch": 2.35, "learning_rate": 3.534968449519231e-05, "loss": 0.3352, "step": 7805 }, { "epoch": 2.35, "learning_rate": 3.534780649038462e-05, "loss": 0.3917, "step": 7806 }, { "epoch": 2.35, "learning_rate": 3.5345928485576924e-05, "loss": 0.3287, "step": 7807 }, { "epoch": 2.35, "learning_rate": 3.534405048076923e-05, "loss": 0.3739, "step": 7808 }, { "epoch": 2.35, "learning_rate": 3.534217247596154e-05, "loss": 0.387, "step": 7809 }, { "epoch": 2.35, "learning_rate": 3.5340294471153845e-05, "loss": 0.4948, "step": 7810 }, { "epoch": 2.35, "learning_rate": 3.533841646634616e-05, "loss": 0.3918, "step": 7811 }, { "epoch": 2.35, "learning_rate": 3.533653846153847e-05, "loss": 0.3107, "step": 7812 }, { "epoch": 2.35, "learning_rate": 3.533466045673077e-05, "loss": 0.2748, "step": 7813 }, { "epoch": 2.35, "learning_rate": 3.533278245192308e-05, "loss": 0.3541, "step": 7814 }, { "epoch": 2.35, "learning_rate": 3.533090444711539e-05, "loss": 0.3994, "step": 7815 }, { "epoch": 2.35, "learning_rate": 3.5329026442307694e-05, "loss": 0.3501, "step": 7816 }, { "epoch": 2.35, "learning_rate": 3.53271484375e-05, "loss": 0.2608, "step": 7817 }, { "epoch": 2.35, "learning_rate": 3.532527043269231e-05, "loss": 0.314, "step": 7818 }, { "epoch": 2.35, "learning_rate": 3.5323392427884615e-05, "loss": 0.3957, "step": 7819 }, { "epoch": 2.35, "learning_rate": 3.532151442307693e-05, "loss": 0.3466, "step": 7820 }, { "epoch": 2.35, "learning_rate": 3.531963641826923e-05, "loss": 0.344, "step": 7821 }, { "epoch": 2.35, "learning_rate": 3.5317758413461536e-05, "loss": 0.3516, "step": 7822 }, { "epoch": 2.35, "learning_rate": 3.531588040865385e-05, "loss": 0.3798, "step": 7823 }, { "epoch": 2.35, "learning_rate": 3.531400240384616e-05, "loss": 0.3209, "step": 7824 }, { "epoch": 2.35, "learning_rate": 3.5312124399038464e-05, "loss": 0.3335, "step": 7825 }, { "epoch": 2.35, "learning_rate": 3.5310246394230776e-05, "loss": 0.4367, "step": 7826 }, { "epoch": 2.35, "learning_rate": 3.530836838942308e-05, "loss": 0.3663, "step": 7827 }, { "epoch": 2.35, "learning_rate": 3.5306490384615385e-05, "loss": 0.3932, "step": 7828 }, { "epoch": 2.35, "learning_rate": 3.53046123798077e-05, "loss": 0.3352, "step": 7829 }, { "epoch": 2.35, "learning_rate": 3.5302734375e-05, "loss": 0.2185, "step": 7830 }, { "epoch": 2.35, "learning_rate": 3.5300856370192306e-05, "loss": 0.2854, "step": 7831 }, { "epoch": 2.35, "learning_rate": 3.529897836538462e-05, "loss": 0.3905, "step": 7832 }, { "epoch": 2.35, "learning_rate": 3.529710036057692e-05, "loss": 0.3056, "step": 7833 }, { "epoch": 2.35, "learning_rate": 3.529522235576923e-05, "loss": 0.3271, "step": 7834 }, { "epoch": 2.35, "learning_rate": 3.529334435096154e-05, "loss": 0.3809, "step": 7835 }, { "epoch": 2.35, "learning_rate": 3.5291466346153844e-05, "loss": 0.4504, "step": 7836 }, { "epoch": 2.35, "learning_rate": 3.5289588341346155e-05, "loss": 0.3155, "step": 7837 }, { "epoch": 2.36, "learning_rate": 3.528771033653847e-05, "loss": 0.344, "step": 7838 }, { "epoch": 2.36, "learning_rate": 3.528583233173077e-05, "loss": 0.3411, "step": 7839 }, { "epoch": 2.36, "learning_rate": 3.528395432692308e-05, "loss": 0.2917, "step": 7840 }, { "epoch": 2.36, "learning_rate": 3.528207632211539e-05, "loss": 0.3333, "step": 7841 }, { "epoch": 2.36, "learning_rate": 3.528019831730769e-05, "loss": 0.403, "step": 7842 }, { "epoch": 2.36, "learning_rate": 3.5278320312500004e-05, "loss": 0.308, "step": 7843 }, { "epoch": 2.36, "learning_rate": 3.527644230769231e-05, "loss": 0.419, "step": 7844 }, { "epoch": 2.36, "learning_rate": 3.5274564302884614e-05, "loss": 0.2659, "step": 7845 }, { "epoch": 2.36, "learning_rate": 3.5272686298076925e-05, "loss": 0.3793, "step": 7846 }, { "epoch": 2.36, "learning_rate": 3.527080829326923e-05, "loss": 0.3428, "step": 7847 }, { "epoch": 2.36, "learning_rate": 3.5268930288461535e-05, "loss": 0.2954, "step": 7848 }, { "epoch": 2.36, "learning_rate": 3.5267052283653846e-05, "loss": 0.4623, "step": 7849 }, { "epoch": 2.36, "learning_rate": 3.526517427884616e-05, "loss": 0.3435, "step": 7850 }, { "epoch": 2.36, "learning_rate": 3.526329627403846e-05, "loss": 0.3421, "step": 7851 }, { "epoch": 2.36, "learning_rate": 3.5261418269230774e-05, "loss": 0.3094, "step": 7852 }, { "epoch": 2.36, "learning_rate": 3.525954026442308e-05, "loss": 0.3588, "step": 7853 }, { "epoch": 2.36, "learning_rate": 3.5257662259615384e-05, "loss": 0.3736, "step": 7854 }, { "epoch": 2.36, "learning_rate": 3.5255784254807695e-05, "loss": 0.5011, "step": 7855 }, { "epoch": 2.36, "learning_rate": 3.525390625e-05, "loss": 0.3076, "step": 7856 }, { "epoch": 2.36, "learning_rate": 3.525202824519231e-05, "loss": 0.5591, "step": 7857 }, { "epoch": 2.36, "learning_rate": 3.5250150240384616e-05, "loss": 0.3491, "step": 7858 }, { "epoch": 2.36, "learning_rate": 3.524827223557692e-05, "loss": 0.3064, "step": 7859 }, { "epoch": 2.36, "learning_rate": 3.524639423076923e-05, "loss": 0.3754, "step": 7860 }, { "epoch": 2.36, "learning_rate": 3.524451622596154e-05, "loss": 0.3329, "step": 7861 }, { "epoch": 2.36, "learning_rate": 3.524263822115384e-05, "loss": 0.5165, "step": 7862 }, { "epoch": 2.36, "learning_rate": 3.524076021634616e-05, "loss": 0.2633, "step": 7863 }, { "epoch": 2.36, "learning_rate": 3.5238882211538465e-05, "loss": 0.3569, "step": 7864 }, { "epoch": 2.36, "learning_rate": 3.523700420673077e-05, "loss": 0.2838, "step": 7865 }, { "epoch": 2.36, "learning_rate": 3.523512620192308e-05, "loss": 0.4145, "step": 7866 }, { "epoch": 2.36, "learning_rate": 3.5233248197115387e-05, "loss": 0.344, "step": 7867 }, { "epoch": 2.36, "learning_rate": 3.523137019230769e-05, "loss": 0.2919, "step": 7868 }, { "epoch": 2.36, "learning_rate": 3.52294921875e-05, "loss": 0.4632, "step": 7869 }, { "epoch": 2.36, "learning_rate": 3.522761418269231e-05, "loss": 0.3098, "step": 7870 }, { "epoch": 2.37, "learning_rate": 3.522573617788462e-05, "loss": 0.452, "step": 7871 }, { "epoch": 2.37, "learning_rate": 3.5223858173076924e-05, "loss": 0.3613, "step": 7872 }, { "epoch": 2.37, "learning_rate": 3.522198016826923e-05, "loss": 0.3727, "step": 7873 }, { "epoch": 2.37, "learning_rate": 3.522010216346154e-05, "loss": 0.4429, "step": 7874 }, { "epoch": 2.37, "learning_rate": 3.5218224158653845e-05, "loss": 0.3354, "step": 7875 }, { "epoch": 2.37, "learning_rate": 3.5216346153846157e-05, "loss": 0.3395, "step": 7876 }, { "epoch": 2.37, "learning_rate": 3.521446814903847e-05, "loss": 0.3477, "step": 7877 }, { "epoch": 2.37, "learning_rate": 3.521259014423077e-05, "loss": 0.4252, "step": 7878 }, { "epoch": 2.37, "learning_rate": 3.521071213942308e-05, "loss": 0.3108, "step": 7879 }, { "epoch": 2.37, "learning_rate": 3.520883413461539e-05, "loss": 0.2951, "step": 7880 }, { "epoch": 2.37, "learning_rate": 3.5206956129807694e-05, "loss": 0.3463, "step": 7881 }, { "epoch": 2.37, "learning_rate": 3.5205078125e-05, "loss": 0.3078, "step": 7882 }, { "epoch": 2.37, "learning_rate": 3.520320012019231e-05, "loss": 0.2464, "step": 7883 }, { "epoch": 2.37, "learning_rate": 3.5201322115384615e-05, "loss": 0.3478, "step": 7884 }, { "epoch": 2.37, "learning_rate": 3.5199444110576927e-05, "loss": 0.3994, "step": 7885 }, { "epoch": 2.37, "learning_rate": 3.519756610576923e-05, "loss": 0.3961, "step": 7886 }, { "epoch": 2.37, "learning_rate": 3.5195688100961536e-05, "loss": 0.33, "step": 7887 }, { "epoch": 2.37, "learning_rate": 3.519381009615385e-05, "loss": 0.4268, "step": 7888 }, { "epoch": 2.37, "learning_rate": 3.519193209134616e-05, "loss": 0.3148, "step": 7889 }, { "epoch": 2.37, "learning_rate": 3.5190054086538464e-05, "loss": 0.3931, "step": 7890 }, { "epoch": 2.37, "learning_rate": 3.5188176081730776e-05, "loss": 0.337, "step": 7891 }, { "epoch": 2.37, "learning_rate": 3.518629807692308e-05, "loss": 0.321, "step": 7892 }, { "epoch": 2.37, "learning_rate": 3.5184420072115385e-05, "loss": 0.3615, "step": 7893 }, { "epoch": 2.37, "learning_rate": 3.51825420673077e-05, "loss": 0.3162, "step": 7894 }, { "epoch": 2.37, "learning_rate": 3.51806640625e-05, "loss": 0.3226, "step": 7895 }, { "epoch": 2.37, "learning_rate": 3.5178786057692306e-05, "loss": 0.3593, "step": 7896 }, { "epoch": 2.37, "learning_rate": 3.517690805288462e-05, "loss": 0.3009, "step": 7897 }, { "epoch": 2.37, "learning_rate": 3.517503004807692e-05, "loss": 0.4324, "step": 7898 }, { "epoch": 2.37, "learning_rate": 3.5173152043269234e-05, "loss": 0.3501, "step": 7899 }, { "epoch": 2.37, "learning_rate": 3.517127403846154e-05, "loss": 0.3484, "step": 7900 }, { "epoch": 2.37, "learning_rate": 3.5169396033653844e-05, "loss": 0.2567, "step": 7901 }, { "epoch": 2.37, "learning_rate": 3.5167518028846155e-05, "loss": 0.3354, "step": 7902 }, { "epoch": 2.37, "learning_rate": 3.516564002403847e-05, "loss": 0.2599, "step": 7903 }, { "epoch": 2.38, "learning_rate": 3.516376201923077e-05, "loss": 0.3722, "step": 7904 }, { "epoch": 2.38, "learning_rate": 3.516188401442308e-05, "loss": 0.4339, "step": 7905 }, { "epoch": 2.38, "learning_rate": 3.516000600961539e-05, "loss": 0.3705, "step": 7906 }, { "epoch": 2.38, "learning_rate": 3.515812800480769e-05, "loss": 0.3412, "step": 7907 }, { "epoch": 2.38, "learning_rate": 3.5156250000000004e-05, "loss": 0.5129, "step": 7908 }, { "epoch": 2.38, "learning_rate": 3.515437199519231e-05, "loss": 0.5218, "step": 7909 }, { "epoch": 2.38, "learning_rate": 3.5152493990384614e-05, "loss": 0.3826, "step": 7910 }, { "epoch": 2.38, "learning_rate": 3.5150615985576925e-05, "loss": 0.4272, "step": 7911 }, { "epoch": 2.38, "learning_rate": 3.514873798076923e-05, "loss": 0.2843, "step": 7912 }, { "epoch": 2.38, "learning_rate": 3.5146859975961535e-05, "loss": 0.4035, "step": 7913 }, { "epoch": 2.38, "learning_rate": 3.5144981971153846e-05, "loss": 0.2851, "step": 7914 }, { "epoch": 2.38, "learning_rate": 3.514310396634616e-05, "loss": 0.3096, "step": 7915 }, { "epoch": 2.38, "learning_rate": 3.514122596153846e-05, "loss": 0.3153, "step": 7916 }, { "epoch": 2.38, "learning_rate": 3.5139347956730774e-05, "loss": 0.3215, "step": 7917 }, { "epoch": 2.38, "learning_rate": 3.513746995192308e-05, "loss": 0.3073, "step": 7918 }, { "epoch": 2.38, "learning_rate": 3.513559194711539e-05, "loss": 0.3486, "step": 7919 }, { "epoch": 2.38, "learning_rate": 3.5133713942307695e-05, "loss": 0.2809, "step": 7920 }, { "epoch": 2.38, "learning_rate": 3.51318359375e-05, "loss": 0.3402, "step": 7921 }, { "epoch": 2.38, "learning_rate": 3.512995793269231e-05, "loss": 0.444, "step": 7922 }, { "epoch": 2.38, "learning_rate": 3.5128079927884616e-05, "loss": 0.2805, "step": 7923 }, { "epoch": 2.38, "learning_rate": 3.512620192307692e-05, "loss": 0.3312, "step": 7924 }, { "epoch": 2.38, "learning_rate": 3.512432391826923e-05, "loss": 0.4177, "step": 7925 }, { "epoch": 2.38, "learning_rate": 3.512244591346154e-05, "loss": 0.3036, "step": 7926 }, { "epoch": 2.38, "learning_rate": 3.512056790865384e-05, "loss": 0.3481, "step": 7927 }, { "epoch": 2.38, "learning_rate": 3.511868990384616e-05, "loss": 0.238, "step": 7928 }, { "epoch": 2.38, "learning_rate": 3.5116811899038465e-05, "loss": 0.3203, "step": 7929 }, { "epoch": 2.38, "learning_rate": 3.511493389423077e-05, "loss": 0.3623, "step": 7930 }, { "epoch": 2.38, "learning_rate": 3.511305588942308e-05, "loss": 0.3233, "step": 7931 }, { "epoch": 2.38, "learning_rate": 3.5111177884615386e-05, "loss": 0.399, "step": 7932 }, { "epoch": 2.38, "learning_rate": 3.510929987980769e-05, "loss": 0.2795, "step": 7933 }, { "epoch": 2.38, "learning_rate": 3.5107421875e-05, "loss": 0.3878, "step": 7934 }, { "epoch": 2.38, "learning_rate": 3.510554387019231e-05, "loss": 0.5034, "step": 7935 }, { "epoch": 2.38, "learning_rate": 3.510366586538462e-05, "loss": 0.1988, "step": 7936 }, { "epoch": 2.38, "learning_rate": 3.5101787860576924e-05, "loss": 0.3354, "step": 7937 }, { "epoch": 2.39, "learning_rate": 3.509990985576923e-05, "loss": 0.4693, "step": 7938 }, { "epoch": 2.39, "learning_rate": 3.509803185096154e-05, "loss": 0.2952, "step": 7939 }, { "epoch": 2.39, "learning_rate": 3.5096153846153845e-05, "loss": 0.4102, "step": 7940 }, { "epoch": 2.39, "learning_rate": 3.5094275841346156e-05, "loss": 0.2671, "step": 7941 }, { "epoch": 2.39, "learning_rate": 3.509239783653847e-05, "loss": 0.3998, "step": 7942 }, { "epoch": 2.39, "learning_rate": 3.509051983173077e-05, "loss": 0.5959, "step": 7943 }, { "epoch": 2.39, "learning_rate": 3.508864182692308e-05, "loss": 0.3207, "step": 7944 }, { "epoch": 2.39, "learning_rate": 3.508676382211539e-05, "loss": 0.3682, "step": 7945 }, { "epoch": 2.39, "learning_rate": 3.5084885817307694e-05, "loss": 0.2725, "step": 7946 }, { "epoch": 2.39, "learning_rate": 3.50830078125e-05, "loss": 0.2745, "step": 7947 }, { "epoch": 2.39, "learning_rate": 3.508112980769231e-05, "loss": 0.391, "step": 7948 }, { "epoch": 2.39, "learning_rate": 3.5079251802884615e-05, "loss": 0.4912, "step": 7949 }, { "epoch": 2.39, "learning_rate": 3.5077373798076927e-05, "loss": 0.3049, "step": 7950 }, { "epoch": 2.39, "learning_rate": 3.507549579326923e-05, "loss": 0.4468, "step": 7951 }, { "epoch": 2.39, "learning_rate": 3.5073617788461536e-05, "loss": 0.3542, "step": 7952 }, { "epoch": 2.39, "learning_rate": 3.507173978365385e-05, "loss": 0.3796, "step": 7953 }, { "epoch": 2.39, "learning_rate": 3.506986177884616e-05, "loss": 0.3127, "step": 7954 }, { "epoch": 2.39, "learning_rate": 3.5067983774038464e-05, "loss": 0.3773, "step": 7955 }, { "epoch": 2.39, "learning_rate": 3.5066105769230775e-05, "loss": 0.3981, "step": 7956 }, { "epoch": 2.39, "learning_rate": 3.506422776442308e-05, "loss": 0.4954, "step": 7957 }, { "epoch": 2.39, "learning_rate": 3.5062349759615385e-05, "loss": 0.3986, "step": 7958 }, { "epoch": 2.39, "learning_rate": 3.5060471754807697e-05, "loss": 0.34, "step": 7959 }, { "epoch": 2.39, "learning_rate": 3.505859375e-05, "loss": 0.4563, "step": 7960 }, { "epoch": 2.39, "learning_rate": 3.5056715745192306e-05, "loss": 0.4796, "step": 7961 }, { "epoch": 2.39, "learning_rate": 3.505483774038462e-05, "loss": 0.527, "step": 7962 }, { "epoch": 2.39, "learning_rate": 3.505295973557692e-05, "loss": 0.34, "step": 7963 }, { "epoch": 2.39, "learning_rate": 3.5051081730769234e-05, "loss": 0.4609, "step": 7964 }, { "epoch": 2.39, "learning_rate": 3.504920372596154e-05, "loss": 0.3703, "step": 7965 }, { "epoch": 2.39, "learning_rate": 3.5047325721153844e-05, "loss": 0.3456, "step": 7966 }, { "epoch": 2.39, "learning_rate": 3.5045447716346155e-05, "loss": 0.3829, "step": 7967 }, { "epoch": 2.39, "learning_rate": 3.5043569711538467e-05, "loss": 0.2841, "step": 7968 }, { "epoch": 2.39, "learning_rate": 3.504169170673077e-05, "loss": 0.4074, "step": 7969 }, { "epoch": 2.39, "learning_rate": 3.503981370192308e-05, "loss": 0.4237, "step": 7970 }, { "epoch": 2.4, "learning_rate": 3.503793569711539e-05, "loss": 0.3144, "step": 7971 }, { "epoch": 2.4, "learning_rate": 3.503605769230769e-05, "loss": 0.4497, "step": 7972 }, { "epoch": 2.4, "learning_rate": 3.5034179687500004e-05, "loss": 0.4485, "step": 7973 }, { "epoch": 2.4, "learning_rate": 3.503230168269231e-05, "loss": 0.3709, "step": 7974 }, { "epoch": 2.4, "learning_rate": 3.5030423677884614e-05, "loss": 0.3787, "step": 7975 }, { "epoch": 2.4, "learning_rate": 3.5028545673076925e-05, "loss": 0.3435, "step": 7976 }, { "epoch": 2.4, "learning_rate": 3.502666766826923e-05, "loss": 0.2971, "step": 7977 }, { "epoch": 2.4, "learning_rate": 3.5024789663461535e-05, "loss": 0.3014, "step": 7978 }, { "epoch": 2.4, "learning_rate": 3.5022911658653846e-05, "loss": 0.4756, "step": 7979 }, { "epoch": 2.4, "learning_rate": 3.502103365384616e-05, "loss": 0.4149, "step": 7980 }, { "epoch": 2.4, "learning_rate": 3.501915564903846e-05, "loss": 0.5161, "step": 7981 }, { "epoch": 2.4, "learning_rate": 3.5017277644230774e-05, "loss": 0.3602, "step": 7982 }, { "epoch": 2.4, "learning_rate": 3.501539963942308e-05, "loss": 0.6219, "step": 7983 }, { "epoch": 2.4, "learning_rate": 3.501352163461539e-05, "loss": 0.3503, "step": 7984 }, { "epoch": 2.4, "learning_rate": 3.5011643629807695e-05, "loss": 0.3497, "step": 7985 }, { "epoch": 2.4, "learning_rate": 3.5009765625e-05, "loss": 0.3354, "step": 7986 }, { "epoch": 2.4, "learning_rate": 3.500788762019231e-05, "loss": 0.3197, "step": 7987 }, { "epoch": 2.4, "learning_rate": 3.5006009615384616e-05, "loss": 0.3658, "step": 7988 }, { "epoch": 2.4, "learning_rate": 3.500413161057692e-05, "loss": 0.3144, "step": 7989 }, { "epoch": 2.4, "learning_rate": 3.500225360576923e-05, "loss": 0.4797, "step": 7990 }, { "epoch": 2.4, "learning_rate": 3.500037560096154e-05, "loss": 0.3787, "step": 7991 }, { "epoch": 2.4, "learning_rate": 3.499849759615384e-05, "loss": 0.4583, "step": 7992 }, { "epoch": 2.4, "learning_rate": 3.499661959134616e-05, "loss": 0.4289, "step": 7993 }, { "epoch": 2.4, "learning_rate": 3.4994741586538465e-05, "loss": 0.3288, "step": 7994 }, { "epoch": 2.4, "learning_rate": 3.499286358173077e-05, "loss": 0.3706, "step": 7995 }, { "epoch": 2.4, "learning_rate": 3.499098557692308e-05, "loss": 0.4907, "step": 7996 }, { "epoch": 2.4, "learning_rate": 3.4989107572115386e-05, "loss": 0.3108, "step": 7997 }, { "epoch": 2.4, "learning_rate": 3.498722956730769e-05, "loss": 0.3257, "step": 7998 }, { "epoch": 2.4, "learning_rate": 3.49853515625e-05, "loss": 0.3207, "step": 7999 }, { "epoch": 2.4, "learning_rate": 3.498347355769231e-05, "loss": 0.5471, "step": 8000 }, { "epoch": 2.4, "learning_rate": 3.498159555288462e-05, "loss": 0.3594, "step": 8001 }, { "epoch": 2.4, "learning_rate": 3.4979717548076924e-05, "loss": 0.308, "step": 8002 }, { "epoch": 2.4, "learning_rate": 3.497783954326923e-05, "loss": 0.3598, "step": 8003 }, { "epoch": 2.41, "learning_rate": 3.497596153846154e-05, "loss": 0.3221, "step": 8004 }, { "epoch": 2.41, "learning_rate": 3.4974083533653845e-05, "loss": 0.5111, "step": 8005 }, { "epoch": 2.41, "learning_rate": 3.4972205528846156e-05, "loss": 0.2875, "step": 8006 }, { "epoch": 2.41, "learning_rate": 3.497032752403847e-05, "loss": 0.3187, "step": 8007 }, { "epoch": 2.41, "learning_rate": 3.496844951923077e-05, "loss": 0.3469, "step": 8008 }, { "epoch": 2.41, "learning_rate": 3.496657151442308e-05, "loss": 0.3226, "step": 8009 }, { "epoch": 2.41, "learning_rate": 3.496469350961539e-05, "loss": 0.2671, "step": 8010 }, { "epoch": 2.41, "learning_rate": 3.4962815504807694e-05, "loss": 0.545, "step": 8011 }, { "epoch": 2.41, "learning_rate": 3.49609375e-05, "loss": 0.4304, "step": 8012 }, { "epoch": 2.41, "learning_rate": 3.495905949519231e-05, "loss": 0.47, "step": 8013 }, { "epoch": 2.41, "learning_rate": 3.4957181490384615e-05, "loss": 0.406, "step": 8014 }, { "epoch": 2.41, "learning_rate": 3.4955303485576926e-05, "loss": 0.4662, "step": 8015 }, { "epoch": 2.41, "learning_rate": 3.495342548076923e-05, "loss": 0.3697, "step": 8016 }, { "epoch": 2.41, "learning_rate": 3.4951547475961536e-05, "loss": 0.3637, "step": 8017 }, { "epoch": 2.41, "learning_rate": 3.494966947115385e-05, "loss": 0.4213, "step": 8018 }, { "epoch": 2.41, "learning_rate": 3.494779146634616e-05, "loss": 0.3285, "step": 8019 }, { "epoch": 2.41, "learning_rate": 3.4945913461538464e-05, "loss": 0.2876, "step": 8020 }, { "epoch": 2.41, "learning_rate": 3.4944035456730775e-05, "loss": 0.392, "step": 8021 }, { "epoch": 2.41, "learning_rate": 3.494215745192308e-05, "loss": 0.2961, "step": 8022 }, { "epoch": 2.41, "learning_rate": 3.4940279447115385e-05, "loss": 0.3191, "step": 8023 }, { "epoch": 2.41, "learning_rate": 3.4938401442307696e-05, "loss": 0.3409, "step": 8024 }, { "epoch": 2.41, "learning_rate": 3.49365234375e-05, "loss": 0.3028, "step": 8025 }, { "epoch": 2.41, "learning_rate": 3.4934645432692306e-05, "loss": 0.3629, "step": 8026 }, { "epoch": 2.41, "learning_rate": 3.493276742788462e-05, "loss": 0.4478, "step": 8027 }, { "epoch": 2.41, "learning_rate": 3.493088942307692e-05, "loss": 0.2423, "step": 8028 }, { "epoch": 2.41, "learning_rate": 3.4929011418269234e-05, "loss": 0.3299, "step": 8029 }, { "epoch": 2.41, "learning_rate": 3.492713341346154e-05, "loss": 0.5015, "step": 8030 }, { "epoch": 2.41, "learning_rate": 3.4925255408653843e-05, "loss": 0.2614, "step": 8031 }, { "epoch": 2.41, "learning_rate": 3.4923377403846155e-05, "loss": 0.2788, "step": 8032 }, { "epoch": 2.41, "learning_rate": 3.4921499399038466e-05, "loss": 0.3039, "step": 8033 }, { "epoch": 2.41, "learning_rate": 3.491962139423077e-05, "loss": 0.3245, "step": 8034 }, { "epoch": 2.41, "learning_rate": 3.491774338942308e-05, "loss": 0.3019, "step": 8035 }, { "epoch": 2.41, "learning_rate": 3.491586538461539e-05, "loss": 0.3335, "step": 8036 }, { "epoch": 2.41, "learning_rate": 3.491398737980769e-05, "loss": 0.5282, "step": 8037 }, { "epoch": 2.42, "learning_rate": 3.4912109375000004e-05, "loss": 0.3073, "step": 8038 }, { "epoch": 2.42, "learning_rate": 3.491023137019231e-05, "loss": 0.4385, "step": 8039 }, { "epoch": 2.42, "learning_rate": 3.4908353365384613e-05, "loss": 0.2922, "step": 8040 }, { "epoch": 2.42, "learning_rate": 3.4906475360576925e-05, "loss": 0.3651, "step": 8041 }, { "epoch": 2.42, "learning_rate": 3.490459735576923e-05, "loss": 0.4761, "step": 8042 }, { "epoch": 2.42, "learning_rate": 3.490271935096154e-05, "loss": 0.2583, "step": 8043 }, { "epoch": 2.42, "learning_rate": 3.4900841346153846e-05, "loss": 0.3563, "step": 8044 }, { "epoch": 2.42, "learning_rate": 3.489896334134616e-05, "loss": 0.4959, "step": 8045 }, { "epoch": 2.42, "learning_rate": 3.489708533653846e-05, "loss": 0.2788, "step": 8046 }, { "epoch": 2.42, "learning_rate": 3.4895207331730774e-05, "loss": 0.4358, "step": 8047 }, { "epoch": 2.42, "learning_rate": 3.489332932692308e-05, "loss": 0.4346, "step": 8048 }, { "epoch": 2.42, "learning_rate": 3.489145132211539e-05, "loss": 0.4128, "step": 8049 }, { "epoch": 2.42, "learning_rate": 3.4889573317307695e-05, "loss": 0.3486, "step": 8050 }, { "epoch": 2.42, "learning_rate": 3.48876953125e-05, "loss": 0.4578, "step": 8051 }, { "epoch": 2.42, "learning_rate": 3.488581730769231e-05, "loss": 0.4307, "step": 8052 }, { "epoch": 2.42, "learning_rate": 3.4883939302884616e-05, "loss": 0.3487, "step": 8053 }, { "epoch": 2.42, "learning_rate": 3.488206129807692e-05, "loss": 0.3273, "step": 8054 }, { "epoch": 2.42, "learning_rate": 3.488018329326923e-05, "loss": 0.2956, "step": 8055 }, { "epoch": 2.42, "learning_rate": 3.487830528846154e-05, "loss": 0.2421, "step": 8056 }, { "epoch": 2.42, "learning_rate": 3.487642728365384e-05, "loss": 0.4438, "step": 8057 }, { "epoch": 2.42, "learning_rate": 3.487454927884616e-05, "loss": 0.2911, "step": 8058 }, { "epoch": 2.42, "learning_rate": 3.4872671274038465e-05, "loss": 0.3303, "step": 8059 }, { "epoch": 2.42, "learning_rate": 3.487079326923077e-05, "loss": 0.2783, "step": 8060 }, { "epoch": 2.42, "learning_rate": 3.486891526442308e-05, "loss": 0.4318, "step": 8061 }, { "epoch": 2.42, "learning_rate": 3.4867037259615386e-05, "loss": 0.251, "step": 8062 }, { "epoch": 2.42, "learning_rate": 3.48651592548077e-05, "loss": 0.548, "step": 8063 }, { "epoch": 2.42, "learning_rate": 3.486328125e-05, "loss": 0.3805, "step": 8064 }, { "epoch": 2.42, "learning_rate": 3.486140324519231e-05, "loss": 0.2841, "step": 8065 }, { "epoch": 2.42, "learning_rate": 3.485952524038462e-05, "loss": 0.3417, "step": 8066 }, { "epoch": 2.42, "learning_rate": 3.4857647235576924e-05, "loss": 0.2997, "step": 8067 }, { "epoch": 2.42, "learning_rate": 3.485576923076923e-05, "loss": 0.316, "step": 8068 }, { "epoch": 2.42, "learning_rate": 3.485389122596154e-05, "loss": 0.3624, "step": 8069 }, { "epoch": 2.42, "learning_rate": 3.4852013221153845e-05, "loss": 0.2631, "step": 8070 }, { "epoch": 2.43, "learning_rate": 3.4850135216346156e-05, "loss": 0.2343, "step": 8071 }, { "epoch": 2.43, "learning_rate": 3.484825721153847e-05, "loss": 0.5123, "step": 8072 }, { "epoch": 2.43, "learning_rate": 3.484637920673077e-05, "loss": 0.3595, "step": 8073 }, { "epoch": 2.43, "learning_rate": 3.484450120192308e-05, "loss": 0.2648, "step": 8074 }, { "epoch": 2.43, "learning_rate": 3.484262319711539e-05, "loss": 0.4091, "step": 8075 }, { "epoch": 2.43, "learning_rate": 3.4840745192307694e-05, "loss": 0.4568, "step": 8076 }, { "epoch": 2.43, "learning_rate": 3.48388671875e-05, "loss": 0.3079, "step": 8077 }, { "epoch": 2.43, "learning_rate": 3.483698918269231e-05, "loss": 0.3296, "step": 8078 }, { "epoch": 2.43, "learning_rate": 3.4835111177884615e-05, "loss": 0.3501, "step": 8079 }, { "epoch": 2.43, "learning_rate": 3.4833233173076926e-05, "loss": 0.2611, "step": 8080 }, { "epoch": 2.43, "learning_rate": 3.483135516826923e-05, "loss": 0.4442, "step": 8081 }, { "epoch": 2.43, "learning_rate": 3.4829477163461536e-05, "loss": 0.3753, "step": 8082 }, { "epoch": 2.43, "learning_rate": 3.482759915865385e-05, "loss": 0.3263, "step": 8083 }, { "epoch": 2.43, "learning_rate": 3.482572115384616e-05, "loss": 0.3725, "step": 8084 }, { "epoch": 2.43, "learning_rate": 3.4823843149038464e-05, "loss": 0.4005, "step": 8085 }, { "epoch": 2.43, "learning_rate": 3.4821965144230775e-05, "loss": 0.4533, "step": 8086 }, { "epoch": 2.43, "learning_rate": 3.482008713942308e-05, "loss": 0.3792, "step": 8087 }, { "epoch": 2.43, "learning_rate": 3.4818209134615385e-05, "loss": 0.2341, "step": 8088 }, { "epoch": 2.43, "learning_rate": 3.4816331129807696e-05, "loss": 0.4318, "step": 8089 }, { "epoch": 2.43, "learning_rate": 3.4814453125e-05, "loss": 0.3199, "step": 8090 }, { "epoch": 2.43, "learning_rate": 3.4812575120192306e-05, "loss": 0.2103, "step": 8091 }, { "epoch": 2.43, "learning_rate": 3.481069711538462e-05, "loss": 0.4385, "step": 8092 }, { "epoch": 2.43, "learning_rate": 3.480881911057692e-05, "loss": 0.4275, "step": 8093 }, { "epoch": 2.43, "learning_rate": 3.4806941105769234e-05, "loss": 0.3651, "step": 8094 }, { "epoch": 2.43, "learning_rate": 3.480506310096154e-05, "loss": 0.3868, "step": 8095 }, { "epoch": 2.43, "learning_rate": 3.480318509615384e-05, "loss": 0.3893, "step": 8096 }, { "epoch": 2.43, "learning_rate": 3.4801307091346155e-05, "loss": 0.3439, "step": 8097 }, { "epoch": 2.43, "learning_rate": 3.4799429086538466e-05, "loss": 0.2719, "step": 8098 }, { "epoch": 2.43, "learning_rate": 3.479755108173077e-05, "loss": 0.3423, "step": 8099 }, { "epoch": 2.43, "learning_rate": 3.479567307692308e-05, "loss": 0.2615, "step": 8100 }, { "epoch": 2.43, "learning_rate": 3.479379507211539e-05, "loss": 0.3468, "step": 8101 }, { "epoch": 2.43, "learning_rate": 3.479191706730769e-05, "loss": 0.413, "step": 8102 }, { "epoch": 2.43, "learning_rate": 3.4790039062500004e-05, "loss": 0.3402, "step": 8103 }, { "epoch": 2.44, "learning_rate": 3.478816105769231e-05, "loss": 0.5737, "step": 8104 }, { "epoch": 2.44, "learning_rate": 3.478628305288461e-05, "loss": 0.3427, "step": 8105 }, { "epoch": 2.44, "learning_rate": 3.4784405048076925e-05, "loss": 0.2698, "step": 8106 }, { "epoch": 2.44, "learning_rate": 3.478252704326923e-05, "loss": 0.2789, "step": 8107 }, { "epoch": 2.44, "learning_rate": 3.478064903846154e-05, "loss": 0.4116, "step": 8108 }, { "epoch": 2.44, "learning_rate": 3.4778771033653846e-05, "loss": 0.3425, "step": 8109 }, { "epoch": 2.44, "learning_rate": 3.477689302884616e-05, "loss": 0.4141, "step": 8110 }, { "epoch": 2.44, "learning_rate": 3.477501502403846e-05, "loss": 0.3743, "step": 8111 }, { "epoch": 2.44, "learning_rate": 3.4773137019230774e-05, "loss": 0.3899, "step": 8112 }, { "epoch": 2.44, "learning_rate": 3.477125901442308e-05, "loss": 0.4244, "step": 8113 }, { "epoch": 2.44, "learning_rate": 3.476938100961539e-05, "loss": 0.4734, "step": 8114 }, { "epoch": 2.44, "learning_rate": 3.4767503004807695e-05, "loss": 0.3754, "step": 8115 }, { "epoch": 2.44, "learning_rate": 3.4765625e-05, "loss": 0.2395, "step": 8116 }, { "epoch": 2.44, "learning_rate": 3.476374699519231e-05, "loss": 0.3311, "step": 8117 }, { "epoch": 2.44, "learning_rate": 3.4761868990384616e-05, "loss": 0.344, "step": 8118 }, { "epoch": 2.44, "learning_rate": 3.475999098557692e-05, "loss": 0.2807, "step": 8119 }, { "epoch": 2.44, "learning_rate": 3.475811298076923e-05, "loss": 0.2983, "step": 8120 }, { "epoch": 2.44, "learning_rate": 3.475623497596154e-05, "loss": 0.2043, "step": 8121 }, { "epoch": 2.44, "learning_rate": 3.475435697115384e-05, "loss": 0.2379, "step": 8122 }, { "epoch": 2.44, "learning_rate": 3.475247896634616e-05, "loss": 0.3384, "step": 8123 }, { "epoch": 2.44, "learning_rate": 3.4750600961538465e-05, "loss": 0.423, "step": 8124 }, { "epoch": 2.44, "learning_rate": 3.474872295673077e-05, "loss": 0.2338, "step": 8125 }, { "epoch": 2.44, "learning_rate": 3.474684495192308e-05, "loss": 0.5127, "step": 8126 }, { "epoch": 2.44, "learning_rate": 3.4744966947115386e-05, "loss": 0.3617, "step": 8127 }, { "epoch": 2.44, "learning_rate": 3.47430889423077e-05, "loss": 0.2817, "step": 8128 }, { "epoch": 2.44, "learning_rate": 3.47412109375e-05, "loss": 0.3234, "step": 8129 }, { "epoch": 2.44, "learning_rate": 3.473933293269231e-05, "loss": 0.4885, "step": 8130 }, { "epoch": 2.44, "learning_rate": 3.473745492788462e-05, "loss": 0.3297, "step": 8131 }, { "epoch": 2.44, "learning_rate": 3.4735576923076923e-05, "loss": 0.5395, "step": 8132 }, { "epoch": 2.44, "learning_rate": 3.473369891826923e-05, "loss": 0.2993, "step": 8133 }, { "epoch": 2.44, "learning_rate": 3.473182091346154e-05, "loss": 0.4562, "step": 8134 }, { "epoch": 2.44, "learning_rate": 3.4729942908653845e-05, "loss": 0.3217, "step": 8135 }, { "epoch": 2.44, "learning_rate": 3.4728064903846156e-05, "loss": 0.4821, "step": 8136 }, { "epoch": 2.45, "learning_rate": 3.472618689903847e-05, "loss": 0.4129, "step": 8137 }, { "epoch": 2.45, "learning_rate": 3.472430889423077e-05, "loss": 0.1693, "step": 8138 }, { "epoch": 2.45, "learning_rate": 3.472243088942308e-05, "loss": 0.2882, "step": 8139 }, { "epoch": 2.45, "learning_rate": 3.472055288461539e-05, "loss": 0.3057, "step": 8140 }, { "epoch": 2.45, "learning_rate": 3.4718674879807694e-05, "loss": 0.3766, "step": 8141 }, { "epoch": 2.45, "learning_rate": 3.4716796875e-05, "loss": 0.3039, "step": 8142 }, { "epoch": 2.45, "learning_rate": 3.471491887019231e-05, "loss": 0.2702, "step": 8143 }, { "epoch": 2.45, "learning_rate": 3.4713040865384615e-05, "loss": 0.3577, "step": 8144 }, { "epoch": 2.45, "learning_rate": 3.4711162860576926e-05, "loss": 0.3689, "step": 8145 }, { "epoch": 2.45, "learning_rate": 3.470928485576923e-05, "loss": 0.4144, "step": 8146 }, { "epoch": 2.45, "learning_rate": 3.4707406850961536e-05, "loss": 0.3819, "step": 8147 }, { "epoch": 2.45, "learning_rate": 3.470552884615385e-05, "loss": 0.474, "step": 8148 }, { "epoch": 2.45, "learning_rate": 3.470365084134616e-05, "loss": 0.2369, "step": 8149 }, { "epoch": 2.45, "learning_rate": 3.4701772836538464e-05, "loss": 0.5019, "step": 8150 }, { "epoch": 2.45, "learning_rate": 3.4699894831730775e-05, "loss": 0.3735, "step": 8151 }, { "epoch": 2.45, "learning_rate": 3.469801682692308e-05, "loss": 0.3756, "step": 8152 }, { "epoch": 2.45, "learning_rate": 3.4696138822115385e-05, "loss": 0.3586, "step": 8153 }, { "epoch": 2.45, "learning_rate": 3.4694260817307696e-05, "loss": 0.337, "step": 8154 }, { "epoch": 2.45, "learning_rate": 3.46923828125e-05, "loss": 0.4728, "step": 8155 }, { "epoch": 2.45, "learning_rate": 3.4690504807692306e-05, "loss": 0.3364, "step": 8156 }, { "epoch": 2.45, "learning_rate": 3.468862680288462e-05, "loss": 0.5009, "step": 8157 }, { "epoch": 2.45, "learning_rate": 3.468674879807692e-05, "loss": 0.2881, "step": 8158 }, { "epoch": 2.45, "learning_rate": 3.4684870793269234e-05, "loss": 0.5773, "step": 8159 }, { "epoch": 2.45, "learning_rate": 3.468299278846154e-05, "loss": 0.3748, "step": 8160 }, { "epoch": 2.45, "learning_rate": 3.468111478365384e-05, "loss": 0.5562, "step": 8161 }, { "epoch": 2.45, "learning_rate": 3.4679236778846155e-05, "loss": 0.5624, "step": 8162 }, { "epoch": 2.45, "learning_rate": 3.4677358774038466e-05, "loss": 0.4623, "step": 8163 }, { "epoch": 2.45, "learning_rate": 3.467548076923077e-05, "loss": 0.3495, "step": 8164 }, { "epoch": 2.45, "learning_rate": 3.467360276442308e-05, "loss": 0.5464, "step": 8165 }, { "epoch": 2.45, "learning_rate": 3.467172475961539e-05, "loss": 0.3407, "step": 8166 }, { "epoch": 2.45, "learning_rate": 3.466984675480769e-05, "loss": 0.3428, "step": 8167 }, { "epoch": 2.45, "learning_rate": 3.4667968750000004e-05, "loss": 0.2559, "step": 8168 }, { "epoch": 2.45, "learning_rate": 3.466609074519231e-05, "loss": 0.4607, "step": 8169 }, { "epoch": 2.45, "learning_rate": 3.466421274038461e-05, "loss": 0.2774, "step": 8170 }, { "epoch": 2.46, "learning_rate": 3.4662334735576925e-05, "loss": 0.271, "step": 8171 }, { "epoch": 2.46, "learning_rate": 3.466045673076923e-05, "loss": 0.3689, "step": 8172 }, { "epoch": 2.46, "learning_rate": 3.465857872596154e-05, "loss": 0.5659, "step": 8173 }, { "epoch": 2.46, "learning_rate": 3.4656700721153846e-05, "loss": 0.3978, "step": 8174 }, { "epoch": 2.46, "learning_rate": 3.465482271634616e-05, "loss": 0.2296, "step": 8175 }, { "epoch": 2.46, "learning_rate": 3.465294471153846e-05, "loss": 0.4172, "step": 8176 }, { "epoch": 2.46, "learning_rate": 3.4651066706730774e-05, "loss": 0.4301, "step": 8177 }, { "epoch": 2.46, "learning_rate": 3.464918870192308e-05, "loss": 0.5944, "step": 8178 }, { "epoch": 2.46, "learning_rate": 3.464731069711539e-05, "loss": 0.2984, "step": 8179 }, { "epoch": 2.46, "learning_rate": 3.4645432692307695e-05, "loss": 0.4307, "step": 8180 }, { "epoch": 2.46, "learning_rate": 3.46435546875e-05, "loss": 0.3764, "step": 8181 }, { "epoch": 2.46, "learning_rate": 3.464167668269231e-05, "loss": 0.3414, "step": 8182 }, { "epoch": 2.46, "learning_rate": 3.4639798677884616e-05, "loss": 0.3216, "step": 8183 }, { "epoch": 2.46, "learning_rate": 3.463792067307692e-05, "loss": 0.4558, "step": 8184 }, { "epoch": 2.46, "learning_rate": 3.463604266826923e-05, "loss": 0.432, "step": 8185 }, { "epoch": 2.46, "learning_rate": 3.463416466346154e-05, "loss": 0.3698, "step": 8186 }, { "epoch": 2.46, "learning_rate": 3.463228665865385e-05, "loss": 0.4283, "step": 8187 }, { "epoch": 2.46, "learning_rate": 3.463040865384616e-05, "loss": 0.4002, "step": 8188 }, { "epoch": 2.46, "learning_rate": 3.4628530649038465e-05, "loss": 0.2513, "step": 8189 }, { "epoch": 2.46, "learning_rate": 3.462665264423077e-05, "loss": 0.3574, "step": 8190 }, { "epoch": 2.46, "learning_rate": 3.462477463942308e-05, "loss": 0.3775, "step": 8191 }, { "epoch": 2.46, "learning_rate": 3.4622896634615386e-05, "loss": 0.4036, "step": 8192 }, { "epoch": 2.46, "learning_rate": 3.46210186298077e-05, "loss": 0.4121, "step": 8193 }, { "epoch": 2.46, "learning_rate": 3.4619140625e-05, "loss": 0.3248, "step": 8194 }, { "epoch": 2.46, "learning_rate": 3.461726262019231e-05, "loss": 0.4251, "step": 8195 }, { "epoch": 2.46, "learning_rate": 3.461538461538462e-05, "loss": 0.3314, "step": 8196 }, { "epoch": 2.46, "learning_rate": 3.461350661057692e-05, "loss": 0.4165, "step": 8197 }, { "epoch": 2.46, "learning_rate": 3.461162860576923e-05, "loss": 0.3825, "step": 8198 }, { "epoch": 2.46, "learning_rate": 3.460975060096154e-05, "loss": 0.3845, "step": 8199 }, { "epoch": 2.46, "learning_rate": 3.4607872596153844e-05, "loss": 0.2957, "step": 8200 }, { "epoch": 2.46, "learning_rate": 3.4605994591346156e-05, "loss": 0.3825, "step": 8201 }, { "epoch": 2.46, "learning_rate": 3.460411658653847e-05, "loss": 0.2977, "step": 8202 }, { "epoch": 2.46, "learning_rate": 3.460223858173077e-05, "loss": 0.3184, "step": 8203 }, { "epoch": 2.47, "learning_rate": 3.460036057692308e-05, "loss": 0.4289, "step": 8204 }, { "epoch": 2.47, "learning_rate": 3.459848257211539e-05, "loss": 0.3866, "step": 8205 }, { "epoch": 2.47, "learning_rate": 3.4596604567307693e-05, "loss": 0.4111, "step": 8206 }, { "epoch": 2.47, "learning_rate": 3.4594726562500005e-05, "loss": 0.2506, "step": 8207 }, { "epoch": 2.47, "learning_rate": 3.459284855769231e-05, "loss": 0.2677, "step": 8208 }, { "epoch": 2.47, "learning_rate": 3.4590970552884615e-05, "loss": 0.4107, "step": 8209 }, { "epoch": 2.47, "learning_rate": 3.4589092548076926e-05, "loss": 0.293, "step": 8210 }, { "epoch": 2.47, "learning_rate": 3.458721454326923e-05, "loss": 0.3973, "step": 8211 }, { "epoch": 2.47, "learning_rate": 3.4585336538461536e-05, "loss": 0.3955, "step": 8212 }, { "epoch": 2.47, "learning_rate": 3.458345853365385e-05, "loss": 0.3996, "step": 8213 }, { "epoch": 2.47, "learning_rate": 3.458158052884616e-05, "loss": 0.2095, "step": 8214 }, { "epoch": 2.47, "learning_rate": 3.4579702524038463e-05, "loss": 0.3035, "step": 8215 }, { "epoch": 2.47, "learning_rate": 3.4577824519230775e-05, "loss": 0.4303, "step": 8216 }, { "epoch": 2.47, "learning_rate": 3.457594651442308e-05, "loss": 0.3443, "step": 8217 }, { "epoch": 2.47, "learning_rate": 3.4574068509615385e-05, "loss": 0.2816, "step": 8218 }, { "epoch": 2.47, "learning_rate": 3.4572190504807696e-05, "loss": 0.4508, "step": 8219 }, { "epoch": 2.47, "learning_rate": 3.45703125e-05, "loss": 0.3294, "step": 8220 }, { "epoch": 2.47, "learning_rate": 3.4568434495192306e-05, "loss": 0.2938, "step": 8221 }, { "epoch": 2.47, "learning_rate": 3.456655649038462e-05, "loss": 0.4005, "step": 8222 }, { "epoch": 2.47, "learning_rate": 3.456467848557692e-05, "loss": 0.4355, "step": 8223 }, { "epoch": 2.47, "learning_rate": 3.4562800480769234e-05, "loss": 0.3201, "step": 8224 }, { "epoch": 2.47, "learning_rate": 3.456092247596154e-05, "loss": 0.3213, "step": 8225 }, { "epoch": 2.47, "learning_rate": 3.455904447115384e-05, "loss": 0.3732, "step": 8226 }, { "epoch": 2.47, "learning_rate": 3.455716646634616e-05, "loss": 0.2988, "step": 8227 }, { "epoch": 2.47, "learning_rate": 3.4555288461538466e-05, "loss": 0.4243, "step": 8228 }, { "epoch": 2.47, "learning_rate": 3.455341045673077e-05, "loss": 0.3393, "step": 8229 }, { "epoch": 2.47, "learning_rate": 3.455153245192308e-05, "loss": 0.4799, "step": 8230 }, { "epoch": 2.47, "learning_rate": 3.454965444711539e-05, "loss": 0.5793, "step": 8231 }, { "epoch": 2.47, "learning_rate": 3.454777644230769e-05, "loss": 0.3329, "step": 8232 }, { "epoch": 2.47, "learning_rate": 3.4545898437500004e-05, "loss": 0.4191, "step": 8233 }, { "epoch": 2.47, "learning_rate": 3.454402043269231e-05, "loss": 0.4754, "step": 8234 }, { "epoch": 2.47, "learning_rate": 3.454214242788461e-05, "loss": 0.2095, "step": 8235 }, { "epoch": 2.47, "learning_rate": 3.4540264423076925e-05, "loss": 0.295, "step": 8236 }, { "epoch": 2.48, "learning_rate": 3.453838641826923e-05, "loss": 0.439, "step": 8237 }, { "epoch": 2.48, "learning_rate": 3.453650841346154e-05, "loss": 0.3025, "step": 8238 }, { "epoch": 2.48, "learning_rate": 3.4534630408653846e-05, "loss": 0.424, "step": 8239 }, { "epoch": 2.48, "learning_rate": 3.453275240384616e-05, "loss": 0.3206, "step": 8240 }, { "epoch": 2.48, "learning_rate": 3.453087439903846e-05, "loss": 0.3685, "step": 8241 }, { "epoch": 2.48, "learning_rate": 3.4528996394230774e-05, "loss": 0.3787, "step": 8242 }, { "epoch": 2.48, "learning_rate": 3.452711838942308e-05, "loss": 0.4236, "step": 8243 }, { "epoch": 2.48, "learning_rate": 3.452524038461539e-05, "loss": 0.4006, "step": 8244 }, { "epoch": 2.48, "learning_rate": 3.4523362379807695e-05, "loss": 0.4541, "step": 8245 }, { "epoch": 2.48, "learning_rate": 3.4521484375e-05, "loss": 0.4987, "step": 8246 }, { "epoch": 2.48, "learning_rate": 3.451960637019231e-05, "loss": 0.3496, "step": 8247 }, { "epoch": 2.48, "learning_rate": 3.4517728365384616e-05, "loss": 0.4591, "step": 8248 }, { "epoch": 2.48, "learning_rate": 3.451585036057692e-05, "loss": 0.334, "step": 8249 }, { "epoch": 2.48, "learning_rate": 3.451397235576923e-05, "loss": 0.3104, "step": 8250 }, { "epoch": 2.48, "learning_rate": 3.451209435096154e-05, "loss": 0.4631, "step": 8251 }, { "epoch": 2.48, "learning_rate": 3.451021634615385e-05, "loss": 0.3797, "step": 8252 }, { "epoch": 2.48, "learning_rate": 3.450833834134616e-05, "loss": 0.2434, "step": 8253 }, { "epoch": 2.48, "learning_rate": 3.4506460336538465e-05, "loss": 0.5151, "step": 8254 }, { "epoch": 2.48, "learning_rate": 3.450458233173077e-05, "loss": 0.2891, "step": 8255 }, { "epoch": 2.48, "learning_rate": 3.450270432692308e-05, "loss": 0.3275, "step": 8256 }, { "epoch": 2.48, "learning_rate": 3.4500826322115386e-05, "loss": 0.3918, "step": 8257 }, { "epoch": 2.48, "learning_rate": 3.44989483173077e-05, "loss": 0.4868, "step": 8258 }, { "epoch": 2.48, "learning_rate": 3.44970703125e-05, "loss": 0.4126, "step": 8259 }, { "epoch": 2.48, "learning_rate": 3.449519230769231e-05, "loss": 0.3457, "step": 8260 }, { "epoch": 2.48, "learning_rate": 3.449331430288462e-05, "loss": 0.2791, "step": 8261 }, { "epoch": 2.48, "learning_rate": 3.449143629807692e-05, "loss": 0.3367, "step": 8262 }, { "epoch": 2.48, "learning_rate": 3.448955829326923e-05, "loss": 0.3251, "step": 8263 }, { "epoch": 2.48, "learning_rate": 3.448768028846154e-05, "loss": 0.3555, "step": 8264 }, { "epoch": 2.48, "learning_rate": 3.4485802283653844e-05, "loss": 0.3845, "step": 8265 }, { "epoch": 2.48, "learning_rate": 3.4483924278846156e-05, "loss": 0.5569, "step": 8266 }, { "epoch": 2.48, "learning_rate": 3.448204627403847e-05, "loss": 0.4518, "step": 8267 }, { "epoch": 2.48, "learning_rate": 3.448016826923077e-05, "loss": 0.3622, "step": 8268 }, { "epoch": 2.48, "learning_rate": 3.447829026442308e-05, "loss": 0.4488, "step": 8269 }, { "epoch": 2.48, "learning_rate": 3.447641225961539e-05, "loss": 0.3095, "step": 8270 }, { "epoch": 2.49, "learning_rate": 3.447453425480769e-05, "loss": 0.3927, "step": 8271 }, { "epoch": 2.49, "learning_rate": 3.4472656250000005e-05, "loss": 0.3754, "step": 8272 }, { "epoch": 2.49, "learning_rate": 3.447077824519231e-05, "loss": 0.4348, "step": 8273 }, { "epoch": 2.49, "learning_rate": 3.4468900240384614e-05, "loss": 0.334, "step": 8274 }, { "epoch": 2.49, "learning_rate": 3.4467022235576926e-05, "loss": 0.4766, "step": 8275 }, { "epoch": 2.49, "learning_rate": 3.446514423076923e-05, "loss": 0.3522, "step": 8276 }, { "epoch": 2.49, "learning_rate": 3.4463266225961536e-05, "loss": 0.4153, "step": 8277 }, { "epoch": 2.49, "learning_rate": 3.446138822115385e-05, "loss": 0.3738, "step": 8278 }, { "epoch": 2.49, "learning_rate": 3.445951021634616e-05, "loss": 0.384, "step": 8279 }, { "epoch": 2.49, "learning_rate": 3.445763221153846e-05, "loss": 0.2985, "step": 8280 }, { "epoch": 2.49, "learning_rate": 3.4455754206730775e-05, "loss": 0.2916, "step": 8281 }, { "epoch": 2.49, "learning_rate": 3.445387620192308e-05, "loss": 0.3546, "step": 8282 }, { "epoch": 2.49, "learning_rate": 3.4451998197115384e-05, "loss": 0.3301, "step": 8283 }, { "epoch": 2.49, "learning_rate": 3.4450120192307696e-05, "loss": 0.4217, "step": 8284 }, { "epoch": 2.49, "learning_rate": 3.44482421875e-05, "loss": 0.3287, "step": 8285 }, { "epoch": 2.49, "learning_rate": 3.4446364182692306e-05, "loss": 0.425, "step": 8286 }, { "epoch": 2.49, "learning_rate": 3.444448617788462e-05, "loss": 0.4345, "step": 8287 }, { "epoch": 2.49, "learning_rate": 3.444260817307692e-05, "loss": 0.4072, "step": 8288 }, { "epoch": 2.49, "learning_rate": 3.4440730168269233e-05, "loss": 0.3457, "step": 8289 }, { "epoch": 2.49, "learning_rate": 3.443885216346154e-05, "loss": 0.3755, "step": 8290 }, { "epoch": 2.49, "learning_rate": 3.443697415865384e-05, "loss": 0.302, "step": 8291 }, { "epoch": 2.49, "learning_rate": 3.443509615384616e-05, "loss": 0.3819, "step": 8292 }, { "epoch": 2.49, "learning_rate": 3.4433218149038466e-05, "loss": 0.3969, "step": 8293 }, { "epoch": 2.49, "learning_rate": 3.443134014423077e-05, "loss": 0.4682, "step": 8294 }, { "epoch": 2.49, "learning_rate": 3.442946213942308e-05, "loss": 0.2978, "step": 8295 }, { "epoch": 2.49, "learning_rate": 3.442758413461539e-05, "loss": 0.4518, "step": 8296 }, { "epoch": 2.49, "learning_rate": 3.442570612980769e-05, "loss": 0.3826, "step": 8297 }, { "epoch": 2.49, "learning_rate": 3.4423828125000003e-05, "loss": 0.3573, "step": 8298 }, { "epoch": 2.49, "learning_rate": 3.442195012019231e-05, "loss": 0.3543, "step": 8299 }, { "epoch": 2.49, "learning_rate": 3.442007211538461e-05, "loss": 0.3942, "step": 8300 }, { "epoch": 2.49, "learning_rate": 3.4418194110576925e-05, "loss": 0.3608, "step": 8301 }, { "epoch": 2.49, "learning_rate": 3.441631610576923e-05, "loss": 0.4822, "step": 8302 }, { "epoch": 2.49, "learning_rate": 3.441443810096154e-05, "loss": 0.3075, "step": 8303 }, { "epoch": 2.5, "learning_rate": 3.4412560096153846e-05, "loss": 0.2622, "step": 8304 }, { "epoch": 2.5, "learning_rate": 3.441068209134616e-05, "loss": 0.3692, "step": 8305 }, { "epoch": 2.5, "learning_rate": 3.440880408653846e-05, "loss": 0.4712, "step": 8306 }, { "epoch": 2.5, "learning_rate": 3.4406926081730774e-05, "loss": 0.601, "step": 8307 }, { "epoch": 2.5, "learning_rate": 3.440504807692308e-05, "loss": 0.5809, "step": 8308 }, { "epoch": 2.5, "learning_rate": 3.440317007211539e-05, "loss": 0.4554, "step": 8309 }, { "epoch": 2.5, "learning_rate": 3.4401292067307695e-05, "loss": 0.332, "step": 8310 }, { "epoch": 2.5, "learning_rate": 3.43994140625e-05, "loss": 0.6002, "step": 8311 }, { "epoch": 2.5, "learning_rate": 3.439753605769231e-05, "loss": 0.3837, "step": 8312 }, { "epoch": 2.5, "learning_rate": 3.4395658052884616e-05, "loss": 0.3121, "step": 8313 }, { "epoch": 2.5, "learning_rate": 3.439378004807692e-05, "loss": 0.2728, "step": 8314 }, { "epoch": 2.5, "learning_rate": 3.439190204326923e-05, "loss": 0.3075, "step": 8315 }, { "epoch": 2.5, "learning_rate": 3.439002403846154e-05, "loss": 0.3337, "step": 8316 }, { "epoch": 2.5, "learning_rate": 3.438814603365385e-05, "loss": 0.3212, "step": 8317 }, { "epoch": 2.5, "learning_rate": 3.438626802884616e-05, "loss": 0.4918, "step": 8318 }, { "epoch": 2.5, "learning_rate": 3.4384390024038465e-05, "loss": 0.4229, "step": 8319 }, { "epoch": 2.5, "learning_rate": 3.438251201923077e-05, "loss": 0.2573, "step": 8320 }, { "epoch": 2.5, "learning_rate": 3.438063401442308e-05, "loss": 0.337, "step": 8321 }, { "epoch": 2.5, "learning_rate": 3.4378756009615386e-05, "loss": 0.3664, "step": 8322 }, { "epoch": 2.5, "learning_rate": 3.43768780048077e-05, "loss": 0.2785, "step": 8323 }, { "epoch": 2.5, "learning_rate": 3.4375e-05, "loss": 0.2988, "step": 8324 }, { "epoch": 2.5, "learning_rate": 3.437312199519231e-05, "loss": 0.2268, "step": 8325 }, { "epoch": 2.5, "learning_rate": 3.437124399038462e-05, "loss": 0.4966, "step": 8326 }, { "epoch": 2.5, "learning_rate": 3.436936598557692e-05, "loss": 0.4668, "step": 8327 }, { "epoch": 2.5, "learning_rate": 3.436748798076923e-05, "loss": 0.3285, "step": 8328 }, { "epoch": 2.5, "learning_rate": 3.436560997596154e-05, "loss": 0.395, "step": 8329 }, { "epoch": 2.5, "learning_rate": 3.4363731971153844e-05, "loss": 0.3245, "step": 8330 }, { "epoch": 2.5, "learning_rate": 3.4361853966346156e-05, "loss": 0.3317, "step": 8331 }, { "epoch": 2.5, "learning_rate": 3.435997596153847e-05, "loss": 0.2436, "step": 8332 }, { "epoch": 2.5, "learning_rate": 3.435809795673077e-05, "loss": 0.3877, "step": 8333 }, { "epoch": 2.5, "learning_rate": 3.435621995192308e-05, "loss": 0.3745, "step": 8334 }, { "epoch": 2.5, "learning_rate": 3.435434194711539e-05, "loss": 0.3165, "step": 8335 }, { "epoch": 2.5, "learning_rate": 3.435246394230769e-05, "loss": 0.3884, "step": 8336 }, { "epoch": 2.51, "learning_rate": 3.4350585937500005e-05, "loss": 0.3414, "step": 8337 }, { "epoch": 2.51, "learning_rate": 3.434870793269231e-05, "loss": 0.3985, "step": 8338 }, { "epoch": 2.51, "learning_rate": 3.4346829927884614e-05, "loss": 0.3634, "step": 8339 }, { "epoch": 2.51, "learning_rate": 3.4344951923076926e-05, "loss": 0.6302, "step": 8340 }, { "epoch": 2.51, "learning_rate": 3.434307391826923e-05, "loss": 0.4315, "step": 8341 }, { "epoch": 2.51, "learning_rate": 3.4341195913461535e-05, "loss": 0.3404, "step": 8342 }, { "epoch": 2.51, "learning_rate": 3.433931790865385e-05, "loss": 0.321, "step": 8343 }, { "epoch": 2.51, "learning_rate": 3.433743990384616e-05, "loss": 0.551, "step": 8344 }, { "epoch": 2.51, "learning_rate": 3.433556189903846e-05, "loss": 0.4005, "step": 8345 }, { "epoch": 2.51, "learning_rate": 3.4333683894230775e-05, "loss": 0.3761, "step": 8346 }, { "epoch": 2.51, "learning_rate": 3.433180588942308e-05, "loss": 0.429, "step": 8347 }, { "epoch": 2.51, "learning_rate": 3.4329927884615384e-05, "loss": 0.4163, "step": 8348 }, { "epoch": 2.51, "learning_rate": 3.4328049879807696e-05, "loss": 0.3527, "step": 8349 }, { "epoch": 2.51, "learning_rate": 3.4326171875e-05, "loss": 0.3282, "step": 8350 }, { "epoch": 2.51, "learning_rate": 3.432429387019231e-05, "loss": 0.2713, "step": 8351 }, { "epoch": 2.51, "learning_rate": 3.432241586538462e-05, "loss": 0.4019, "step": 8352 }, { "epoch": 2.51, "learning_rate": 3.432053786057692e-05, "loss": 0.3664, "step": 8353 }, { "epoch": 2.51, "learning_rate": 3.431865985576923e-05, "loss": 0.2794, "step": 8354 }, { "epoch": 2.51, "learning_rate": 3.431678185096154e-05, "loss": 0.3486, "step": 8355 }, { "epoch": 2.51, "learning_rate": 3.431490384615384e-05, "loss": 0.308, "step": 8356 }, { "epoch": 2.51, "learning_rate": 3.431302584134616e-05, "loss": 0.3879, "step": 8357 }, { "epoch": 2.51, "learning_rate": 3.4311147836538466e-05, "loss": 0.3539, "step": 8358 }, { "epoch": 2.51, "learning_rate": 3.430926983173077e-05, "loss": 0.3212, "step": 8359 }, { "epoch": 2.51, "learning_rate": 3.430739182692308e-05, "loss": 0.4276, "step": 8360 }, { "epoch": 2.51, "learning_rate": 3.430551382211539e-05, "loss": 0.338, "step": 8361 }, { "epoch": 2.51, "learning_rate": 3.430363581730769e-05, "loss": 0.3635, "step": 8362 }, { "epoch": 2.51, "learning_rate": 3.43017578125e-05, "loss": 0.3996, "step": 8363 }, { "epoch": 2.51, "learning_rate": 3.429987980769231e-05, "loss": 0.2849, "step": 8364 }, { "epoch": 2.51, "learning_rate": 3.429800180288461e-05, "loss": 0.3186, "step": 8365 }, { "epoch": 2.51, "learning_rate": 3.4296123798076924e-05, "loss": 0.4629, "step": 8366 }, { "epoch": 2.51, "learning_rate": 3.429424579326923e-05, "loss": 0.24, "step": 8367 }, { "epoch": 2.51, "learning_rate": 3.429236778846154e-05, "loss": 0.3439, "step": 8368 }, { "epoch": 2.51, "learning_rate": 3.4290489783653846e-05, "loss": 0.5965, "step": 8369 }, { "epoch": 2.52, "learning_rate": 3.428861177884616e-05, "loss": 0.4843, "step": 8370 }, { "epoch": 2.52, "learning_rate": 3.428673377403847e-05, "loss": 0.5215, "step": 8371 }, { "epoch": 2.52, "learning_rate": 3.4284855769230773e-05, "loss": 0.3896, "step": 8372 }, { "epoch": 2.52, "learning_rate": 3.428297776442308e-05, "loss": 0.2067, "step": 8373 }, { "epoch": 2.52, "learning_rate": 3.428109975961539e-05, "loss": 0.6486, "step": 8374 }, { "epoch": 2.52, "learning_rate": 3.4279221754807695e-05, "loss": 0.3335, "step": 8375 }, { "epoch": 2.52, "learning_rate": 3.427734375e-05, "loss": 0.3628, "step": 8376 }, { "epoch": 2.52, "learning_rate": 3.427546574519231e-05, "loss": 0.3741, "step": 8377 }, { "epoch": 2.52, "learning_rate": 3.4273587740384616e-05, "loss": 0.4158, "step": 8378 }, { "epoch": 2.52, "learning_rate": 3.427170973557692e-05, "loss": 0.5539, "step": 8379 }, { "epoch": 2.52, "learning_rate": 3.426983173076923e-05, "loss": 0.3477, "step": 8380 }, { "epoch": 2.52, "learning_rate": 3.426795372596154e-05, "loss": 0.3052, "step": 8381 }, { "epoch": 2.52, "learning_rate": 3.426607572115385e-05, "loss": 0.4134, "step": 8382 }, { "epoch": 2.52, "learning_rate": 3.426419771634616e-05, "loss": 0.307, "step": 8383 }, { "epoch": 2.52, "learning_rate": 3.4262319711538465e-05, "loss": 0.4479, "step": 8384 }, { "epoch": 2.52, "learning_rate": 3.426044170673077e-05, "loss": 0.2806, "step": 8385 }, { "epoch": 2.52, "learning_rate": 3.425856370192308e-05, "loss": 0.3212, "step": 8386 }, { "epoch": 2.52, "learning_rate": 3.4256685697115386e-05, "loss": 0.368, "step": 8387 }, { "epoch": 2.52, "learning_rate": 3.42548076923077e-05, "loss": 0.3363, "step": 8388 }, { "epoch": 2.52, "learning_rate": 3.42529296875e-05, "loss": 0.5595, "step": 8389 }, { "epoch": 2.52, "learning_rate": 3.425105168269231e-05, "loss": 0.3441, "step": 8390 }, { "epoch": 2.52, "learning_rate": 3.424917367788462e-05, "loss": 0.3827, "step": 8391 }, { "epoch": 2.52, "learning_rate": 3.424729567307692e-05, "loss": 0.3105, "step": 8392 }, { "epoch": 2.52, "learning_rate": 3.424541766826923e-05, "loss": 0.4515, "step": 8393 }, { "epoch": 2.52, "learning_rate": 3.424353966346154e-05, "loss": 0.275, "step": 8394 }, { "epoch": 2.52, "learning_rate": 3.4241661658653844e-05, "loss": 0.3987, "step": 8395 }, { "epoch": 2.52, "learning_rate": 3.4239783653846156e-05, "loss": 0.4054, "step": 8396 }, { "epoch": 2.52, "learning_rate": 3.423790564903847e-05, "loss": 0.3904, "step": 8397 }, { "epoch": 2.52, "learning_rate": 3.423602764423077e-05, "loss": 0.3234, "step": 8398 }, { "epoch": 2.52, "learning_rate": 3.423414963942308e-05, "loss": 0.363, "step": 8399 }, { "epoch": 2.52, "learning_rate": 3.423227163461539e-05, "loss": 0.4568, "step": 8400 }, { "epoch": 2.52, "learning_rate": 3.423039362980769e-05, "loss": 0.3359, "step": 8401 }, { "epoch": 2.52, "learning_rate": 3.4228515625000005e-05, "loss": 0.2743, "step": 8402 }, { "epoch": 2.52, "learning_rate": 3.422663762019231e-05, "loss": 0.493, "step": 8403 }, { "epoch": 2.53, "learning_rate": 3.4224759615384614e-05, "loss": 0.31, "step": 8404 }, { "epoch": 2.53, "learning_rate": 3.4222881610576926e-05, "loss": 0.2966, "step": 8405 }, { "epoch": 2.53, "learning_rate": 3.422100360576923e-05, "loss": 0.5285, "step": 8406 }, { "epoch": 2.53, "learning_rate": 3.4219125600961535e-05, "loss": 0.3549, "step": 8407 }, { "epoch": 2.53, "learning_rate": 3.421724759615385e-05, "loss": 0.3133, "step": 8408 }, { "epoch": 2.53, "learning_rate": 3.421536959134616e-05, "loss": 0.3497, "step": 8409 }, { "epoch": 2.53, "learning_rate": 3.421349158653846e-05, "loss": 0.5337, "step": 8410 }, { "epoch": 2.53, "learning_rate": 3.4211613581730775e-05, "loss": 0.3675, "step": 8411 }, { "epoch": 2.53, "learning_rate": 3.420973557692308e-05, "loss": 0.492, "step": 8412 }, { "epoch": 2.53, "learning_rate": 3.4207857572115384e-05, "loss": 0.3467, "step": 8413 }, { "epoch": 2.53, "learning_rate": 3.4205979567307696e-05, "loss": 0.3718, "step": 8414 }, { "epoch": 2.53, "learning_rate": 3.42041015625e-05, "loss": 0.2647, "step": 8415 }, { "epoch": 2.53, "learning_rate": 3.420222355769231e-05, "loss": 0.3699, "step": 8416 }, { "epoch": 2.53, "learning_rate": 3.420034555288462e-05, "loss": 0.3791, "step": 8417 }, { "epoch": 2.53, "learning_rate": 3.419846754807692e-05, "loss": 0.6525, "step": 8418 }, { "epoch": 2.53, "learning_rate": 3.419658954326923e-05, "loss": 0.381, "step": 8419 }, { "epoch": 2.53, "learning_rate": 3.419471153846154e-05, "loss": 0.467, "step": 8420 }, { "epoch": 2.53, "learning_rate": 3.419283353365384e-05, "loss": 0.5223, "step": 8421 }, { "epoch": 2.53, "learning_rate": 3.419095552884616e-05, "loss": 0.2947, "step": 8422 }, { "epoch": 2.53, "learning_rate": 3.4189077524038466e-05, "loss": 0.3856, "step": 8423 }, { "epoch": 2.53, "learning_rate": 3.418719951923077e-05, "loss": 0.5495, "step": 8424 }, { "epoch": 2.53, "learning_rate": 3.418532151442308e-05, "loss": 0.3654, "step": 8425 }, { "epoch": 2.53, "learning_rate": 3.418344350961539e-05, "loss": 0.3174, "step": 8426 }, { "epoch": 2.53, "learning_rate": 3.418156550480769e-05, "loss": 0.4331, "step": 8427 }, { "epoch": 2.53, "learning_rate": 3.41796875e-05, "loss": 0.3324, "step": 8428 }, { "epoch": 2.53, "learning_rate": 3.417780949519231e-05, "loss": 0.3339, "step": 8429 }, { "epoch": 2.53, "learning_rate": 3.417593149038461e-05, "loss": 0.5338, "step": 8430 }, { "epoch": 2.53, "learning_rate": 3.4174053485576924e-05, "loss": 0.3407, "step": 8431 }, { "epoch": 2.53, "learning_rate": 3.417217548076923e-05, "loss": 0.3491, "step": 8432 }, { "epoch": 2.53, "learning_rate": 3.417029747596154e-05, "loss": 0.2627, "step": 8433 }, { "epoch": 2.53, "learning_rate": 3.4168419471153845e-05, "loss": 0.3074, "step": 8434 }, { "epoch": 2.53, "learning_rate": 3.416654146634616e-05, "loss": 0.3532, "step": 8435 }, { "epoch": 2.53, "learning_rate": 3.416466346153847e-05, "loss": 0.4052, "step": 8436 }, { "epoch": 2.54, "learning_rate": 3.416278545673077e-05, "loss": 0.4119, "step": 8437 }, { "epoch": 2.54, "learning_rate": 3.416090745192308e-05, "loss": 0.3979, "step": 8438 }, { "epoch": 2.54, "learning_rate": 3.415902944711539e-05, "loss": 0.2969, "step": 8439 }, { "epoch": 2.54, "learning_rate": 3.4157151442307694e-05, "loss": 0.3929, "step": 8440 }, { "epoch": 2.54, "learning_rate": 3.41552734375e-05, "loss": 0.3792, "step": 8441 }, { "epoch": 2.54, "learning_rate": 3.415339543269231e-05, "loss": 0.3665, "step": 8442 }, { "epoch": 2.54, "learning_rate": 3.4151517427884615e-05, "loss": 0.3146, "step": 8443 }, { "epoch": 2.54, "learning_rate": 3.414963942307692e-05, "loss": 0.3246, "step": 8444 }, { "epoch": 2.54, "learning_rate": 3.414776141826923e-05, "loss": 0.4602, "step": 8445 }, { "epoch": 2.54, "learning_rate": 3.4145883413461537e-05, "loss": 0.3348, "step": 8446 }, { "epoch": 2.54, "learning_rate": 3.414400540865385e-05, "loss": 0.2427, "step": 8447 }, { "epoch": 2.54, "learning_rate": 3.414212740384616e-05, "loss": 0.2544, "step": 8448 }, { "epoch": 2.54, "learning_rate": 3.4140249399038464e-05, "loss": 0.4047, "step": 8449 }, { "epoch": 2.54, "learning_rate": 3.413837139423077e-05, "loss": 0.3028, "step": 8450 }, { "epoch": 2.54, "learning_rate": 3.413649338942308e-05, "loss": 0.3867, "step": 8451 }, { "epoch": 2.54, "learning_rate": 3.4134615384615386e-05, "loss": 0.3796, "step": 8452 }, { "epoch": 2.54, "learning_rate": 3.41327373798077e-05, "loss": 0.3397, "step": 8453 }, { "epoch": 2.54, "learning_rate": 3.4130859375e-05, "loss": 0.2583, "step": 8454 }, { "epoch": 2.54, "learning_rate": 3.412898137019231e-05, "loss": 0.3311, "step": 8455 }, { "epoch": 2.54, "learning_rate": 3.412710336538462e-05, "loss": 0.3973, "step": 8456 }, { "epoch": 2.54, "learning_rate": 3.412522536057692e-05, "loss": 0.2656, "step": 8457 }, { "epoch": 2.54, "learning_rate": 3.412334735576923e-05, "loss": 0.4764, "step": 8458 }, { "epoch": 2.54, "learning_rate": 3.412146935096154e-05, "loss": 0.3491, "step": 8459 }, { "epoch": 2.54, "learning_rate": 3.4119591346153844e-05, "loss": 0.3908, "step": 8460 }, { "epoch": 2.54, "learning_rate": 3.4117713341346156e-05, "loss": 0.4221, "step": 8461 }, { "epoch": 2.54, "learning_rate": 3.411583533653847e-05, "loss": 0.5018, "step": 8462 }, { "epoch": 2.54, "learning_rate": 3.411395733173077e-05, "loss": 0.3209, "step": 8463 }, { "epoch": 2.54, "learning_rate": 3.411207932692308e-05, "loss": 0.3684, "step": 8464 }, { "epoch": 2.54, "learning_rate": 3.411020132211539e-05, "loss": 0.5662, "step": 8465 }, { "epoch": 2.54, "learning_rate": 3.410832331730769e-05, "loss": 0.291, "step": 8466 }, { "epoch": 2.54, "learning_rate": 3.4106445312500005e-05, "loss": 0.3455, "step": 8467 }, { "epoch": 2.54, "learning_rate": 3.410456730769231e-05, "loss": 0.2731, "step": 8468 }, { "epoch": 2.54, "learning_rate": 3.4102689302884614e-05, "loss": 0.3254, "step": 8469 }, { "epoch": 2.55, "learning_rate": 3.4100811298076926e-05, "loss": 0.3744, "step": 8470 }, { "epoch": 2.55, "learning_rate": 3.409893329326923e-05, "loss": 0.4072, "step": 8471 }, { "epoch": 2.55, "learning_rate": 3.4097055288461535e-05, "loss": 0.4371, "step": 8472 }, { "epoch": 2.55, "learning_rate": 3.409517728365385e-05, "loss": 0.3112, "step": 8473 }, { "epoch": 2.55, "learning_rate": 3.409329927884616e-05, "loss": 0.2928, "step": 8474 }, { "epoch": 2.55, "learning_rate": 3.409142127403846e-05, "loss": 0.326, "step": 8475 }, { "epoch": 2.55, "learning_rate": 3.4089543269230775e-05, "loss": 0.2989, "step": 8476 }, { "epoch": 2.55, "learning_rate": 3.408766526442308e-05, "loss": 0.2762, "step": 8477 }, { "epoch": 2.55, "learning_rate": 3.4085787259615384e-05, "loss": 0.4008, "step": 8478 }, { "epoch": 2.55, "learning_rate": 3.4083909254807696e-05, "loss": 0.4705, "step": 8479 }, { "epoch": 2.55, "learning_rate": 3.408203125e-05, "loss": 0.3849, "step": 8480 }, { "epoch": 2.55, "learning_rate": 3.408015324519231e-05, "loss": 0.449, "step": 8481 }, { "epoch": 2.55, "learning_rate": 3.407827524038462e-05, "loss": 0.3223, "step": 8482 }, { "epoch": 2.55, "learning_rate": 3.407639723557692e-05, "loss": 0.3428, "step": 8483 }, { "epoch": 2.55, "learning_rate": 3.407451923076923e-05, "loss": 0.4422, "step": 8484 }, { "epoch": 2.55, "learning_rate": 3.407264122596154e-05, "loss": 0.4268, "step": 8485 }, { "epoch": 2.55, "learning_rate": 3.407076322115384e-05, "loss": 0.5696, "step": 8486 }, { "epoch": 2.55, "learning_rate": 3.406888521634616e-05, "loss": 0.3581, "step": 8487 }, { "epoch": 2.55, "learning_rate": 3.4067007211538466e-05, "loss": 0.5152, "step": 8488 }, { "epoch": 2.55, "learning_rate": 3.406512920673077e-05, "loss": 0.3383, "step": 8489 }, { "epoch": 2.55, "learning_rate": 3.406325120192308e-05, "loss": 0.3894, "step": 8490 }, { "epoch": 2.55, "learning_rate": 3.406137319711539e-05, "loss": 0.3584, "step": 8491 }, { "epoch": 2.55, "learning_rate": 3.405949519230769e-05, "loss": 0.2511, "step": 8492 }, { "epoch": 2.55, "learning_rate": 3.40576171875e-05, "loss": 0.3349, "step": 8493 }, { "epoch": 2.55, "learning_rate": 3.405573918269231e-05, "loss": 0.3487, "step": 8494 }, { "epoch": 2.55, "learning_rate": 3.405386117788462e-05, "loss": 0.3453, "step": 8495 }, { "epoch": 2.55, "learning_rate": 3.4051983173076924e-05, "loss": 0.2906, "step": 8496 }, { "epoch": 2.55, "learning_rate": 3.405010516826923e-05, "loss": 0.2558, "step": 8497 }, { "epoch": 2.55, "learning_rate": 3.404822716346154e-05, "loss": 0.4277, "step": 8498 }, { "epoch": 2.55, "learning_rate": 3.4046349158653845e-05, "loss": 0.3104, "step": 8499 }, { "epoch": 2.55, "learning_rate": 3.404447115384616e-05, "loss": 0.3749, "step": 8500 }, { "epoch": 2.55, "learning_rate": 3.404259314903847e-05, "loss": 0.3737, "step": 8501 }, { "epoch": 2.55, "learning_rate": 3.404071514423077e-05, "loss": 0.329, "step": 8502 }, { "epoch": 2.55, "learning_rate": 3.403883713942308e-05, "loss": 0.3864, "step": 8503 }, { "epoch": 2.56, "learning_rate": 3.403695913461539e-05, "loss": 0.5859, "step": 8504 }, { "epoch": 2.56, "learning_rate": 3.4035081129807694e-05, "loss": 0.3256, "step": 8505 }, { "epoch": 2.56, "learning_rate": 3.4033203125e-05, "loss": 0.4122, "step": 8506 }, { "epoch": 2.56, "learning_rate": 3.403132512019231e-05, "loss": 0.379, "step": 8507 }, { "epoch": 2.56, "learning_rate": 3.4029447115384615e-05, "loss": 0.4057, "step": 8508 }, { "epoch": 2.56, "learning_rate": 3.402756911057692e-05, "loss": 0.3696, "step": 8509 }, { "epoch": 2.56, "learning_rate": 3.402569110576923e-05, "loss": 0.3943, "step": 8510 }, { "epoch": 2.56, "learning_rate": 3.4023813100961536e-05, "loss": 0.4059, "step": 8511 }, { "epoch": 2.56, "learning_rate": 3.402193509615385e-05, "loss": 0.4807, "step": 8512 }, { "epoch": 2.56, "learning_rate": 3.402005709134616e-05, "loss": 0.3238, "step": 8513 }, { "epoch": 2.56, "learning_rate": 3.4018179086538464e-05, "loss": 0.3088, "step": 8514 }, { "epoch": 2.56, "learning_rate": 3.4016301081730776e-05, "loss": 0.4051, "step": 8515 }, { "epoch": 2.56, "learning_rate": 3.401442307692308e-05, "loss": 0.4071, "step": 8516 }, { "epoch": 2.56, "learning_rate": 3.4012545072115385e-05, "loss": 0.2838, "step": 8517 }, { "epoch": 2.56, "learning_rate": 3.40106670673077e-05, "loss": 0.3659, "step": 8518 }, { "epoch": 2.56, "learning_rate": 3.40087890625e-05, "loss": 0.3748, "step": 8519 }, { "epoch": 2.56, "learning_rate": 3.4006911057692307e-05, "loss": 0.3658, "step": 8520 }, { "epoch": 2.56, "learning_rate": 3.400503305288462e-05, "loss": 0.3815, "step": 8521 }, { "epoch": 2.56, "learning_rate": 3.400315504807692e-05, "loss": 0.3726, "step": 8522 }, { "epoch": 2.56, "learning_rate": 3.400127704326923e-05, "loss": 0.5316, "step": 8523 }, { "epoch": 2.56, "learning_rate": 3.399939903846154e-05, "loss": 0.4337, "step": 8524 }, { "epoch": 2.56, "learning_rate": 3.3997521033653844e-05, "loss": 0.26, "step": 8525 }, { "epoch": 2.56, "learning_rate": 3.3995643028846155e-05, "loss": 0.3208, "step": 8526 }, { "epoch": 2.56, "learning_rate": 3.399376502403847e-05, "loss": 0.3899, "step": 8527 }, { "epoch": 2.56, "learning_rate": 3.399188701923077e-05, "loss": 0.3682, "step": 8528 }, { "epoch": 2.56, "learning_rate": 3.3990009014423077e-05, "loss": 0.409, "step": 8529 }, { "epoch": 2.56, "learning_rate": 3.398813100961539e-05, "loss": 0.4339, "step": 8530 }, { "epoch": 2.56, "learning_rate": 3.398625300480769e-05, "loss": 0.3384, "step": 8531 }, { "epoch": 2.56, "learning_rate": 3.3984375000000004e-05, "loss": 0.2903, "step": 8532 }, { "epoch": 2.56, "learning_rate": 3.398249699519231e-05, "loss": 0.344, "step": 8533 }, { "epoch": 2.56, "learning_rate": 3.3980618990384614e-05, "loss": 0.4024, "step": 8534 }, { "epoch": 2.56, "learning_rate": 3.3978740985576926e-05, "loss": 0.4193, "step": 8535 }, { "epoch": 2.56, "learning_rate": 3.397686298076923e-05, "loss": 0.2355, "step": 8536 }, { "epoch": 2.57, "learning_rate": 3.3974984975961535e-05, "loss": 0.3675, "step": 8537 }, { "epoch": 2.57, "learning_rate": 3.397310697115385e-05, "loss": 0.3363, "step": 8538 }, { "epoch": 2.57, "learning_rate": 3.397122896634616e-05, "loss": 0.2916, "step": 8539 }, { "epoch": 2.57, "learning_rate": 3.396935096153846e-05, "loss": 0.4488, "step": 8540 }, { "epoch": 2.57, "learning_rate": 3.3967472956730774e-05, "loss": 0.3057, "step": 8541 }, { "epoch": 2.57, "learning_rate": 3.396559495192308e-05, "loss": 0.5253, "step": 8542 }, { "epoch": 2.57, "learning_rate": 3.3963716947115384e-05, "loss": 0.4945, "step": 8543 }, { "epoch": 2.57, "learning_rate": 3.3961838942307696e-05, "loss": 0.4278, "step": 8544 }, { "epoch": 2.57, "learning_rate": 3.39599609375e-05, "loss": 0.4217, "step": 8545 }, { "epoch": 2.57, "learning_rate": 3.395808293269231e-05, "loss": 0.4525, "step": 8546 }, { "epoch": 2.57, "learning_rate": 3.395620492788462e-05, "loss": 0.2864, "step": 8547 }, { "epoch": 2.57, "learning_rate": 3.395432692307692e-05, "loss": 0.3496, "step": 8548 }, { "epoch": 2.57, "learning_rate": 3.395244891826923e-05, "loss": 0.2913, "step": 8549 }, { "epoch": 2.57, "learning_rate": 3.395057091346154e-05, "loss": 0.2764, "step": 8550 }, { "epoch": 2.57, "learning_rate": 3.394869290865384e-05, "loss": 0.4318, "step": 8551 }, { "epoch": 2.57, "learning_rate": 3.394681490384616e-05, "loss": 0.4513, "step": 8552 }, { "epoch": 2.57, "learning_rate": 3.3944936899038466e-05, "loss": 0.3523, "step": 8553 }, { "epoch": 2.57, "learning_rate": 3.394305889423077e-05, "loss": 0.3198, "step": 8554 }, { "epoch": 2.57, "learning_rate": 3.394118088942308e-05, "loss": 0.4009, "step": 8555 }, { "epoch": 2.57, "learning_rate": 3.393930288461539e-05, "loss": 0.3652, "step": 8556 }, { "epoch": 2.57, "learning_rate": 3.393742487980769e-05, "loss": 0.4836, "step": 8557 }, { "epoch": 2.57, "learning_rate": 3.3935546875e-05, "loss": 0.3638, "step": 8558 }, { "epoch": 2.57, "learning_rate": 3.393366887019231e-05, "loss": 0.5196, "step": 8559 }, { "epoch": 2.57, "learning_rate": 3.393179086538462e-05, "loss": 0.4471, "step": 8560 }, { "epoch": 2.57, "learning_rate": 3.3929912860576924e-05, "loss": 0.2474, "step": 8561 }, { "epoch": 2.57, "learning_rate": 3.392803485576923e-05, "loss": 0.4908, "step": 8562 }, { "epoch": 2.57, "learning_rate": 3.392615685096154e-05, "loss": 0.2299, "step": 8563 }, { "epoch": 2.57, "learning_rate": 3.3924278846153845e-05, "loss": 0.3788, "step": 8564 }, { "epoch": 2.57, "learning_rate": 3.392240084134616e-05, "loss": 0.461, "step": 8565 }, { "epoch": 2.57, "learning_rate": 3.392052283653847e-05, "loss": 0.3056, "step": 8566 }, { "epoch": 2.57, "learning_rate": 3.391864483173077e-05, "loss": 0.3585, "step": 8567 }, { "epoch": 2.57, "learning_rate": 3.391676682692308e-05, "loss": 0.3297, "step": 8568 }, { "epoch": 2.57, "learning_rate": 3.391488882211539e-05, "loss": 0.4805, "step": 8569 }, { "epoch": 2.58, "learning_rate": 3.3913010817307694e-05, "loss": 0.3798, "step": 8570 }, { "epoch": 2.58, "learning_rate": 3.39111328125e-05, "loss": 0.2966, "step": 8571 }, { "epoch": 2.58, "learning_rate": 3.390925480769231e-05, "loss": 0.2899, "step": 8572 }, { "epoch": 2.58, "learning_rate": 3.3907376802884615e-05, "loss": 0.3134, "step": 8573 }, { "epoch": 2.58, "learning_rate": 3.390549879807692e-05, "loss": 0.3425, "step": 8574 }, { "epoch": 2.58, "learning_rate": 3.390362079326923e-05, "loss": 0.6004, "step": 8575 }, { "epoch": 2.58, "learning_rate": 3.3901742788461536e-05, "loss": 0.3766, "step": 8576 }, { "epoch": 2.58, "learning_rate": 3.389986478365385e-05, "loss": 0.3045, "step": 8577 }, { "epoch": 2.58, "learning_rate": 3.389798677884616e-05, "loss": 0.4722, "step": 8578 }, { "epoch": 2.58, "learning_rate": 3.3896108774038464e-05, "loss": 0.3731, "step": 8579 }, { "epoch": 2.58, "learning_rate": 3.3894230769230776e-05, "loss": 0.2555, "step": 8580 }, { "epoch": 2.58, "learning_rate": 3.389235276442308e-05, "loss": 0.4485, "step": 8581 }, { "epoch": 2.58, "learning_rate": 3.3890474759615385e-05, "loss": 0.3301, "step": 8582 }, { "epoch": 2.58, "learning_rate": 3.38885967548077e-05, "loss": 0.3068, "step": 8583 }, { "epoch": 2.58, "learning_rate": 3.388671875e-05, "loss": 0.3297, "step": 8584 }, { "epoch": 2.58, "learning_rate": 3.3884840745192306e-05, "loss": 0.3642, "step": 8585 }, { "epoch": 2.58, "learning_rate": 3.388296274038462e-05, "loss": 0.3725, "step": 8586 }, { "epoch": 2.58, "learning_rate": 3.388108473557692e-05, "loss": 0.3368, "step": 8587 }, { "epoch": 2.58, "learning_rate": 3.387920673076923e-05, "loss": 0.3592, "step": 8588 }, { "epoch": 2.58, "learning_rate": 3.387732872596154e-05, "loss": 0.3047, "step": 8589 }, { "epoch": 2.58, "learning_rate": 3.3875450721153844e-05, "loss": 0.2534, "step": 8590 }, { "epoch": 2.58, "learning_rate": 3.3873572716346155e-05, "loss": 0.4, "step": 8591 }, { "epoch": 2.58, "learning_rate": 3.387169471153847e-05, "loss": 0.3604, "step": 8592 }, { "epoch": 2.58, "learning_rate": 3.386981670673077e-05, "loss": 0.3803, "step": 8593 }, { "epoch": 2.58, "learning_rate": 3.3867938701923076e-05, "loss": 0.4043, "step": 8594 }, { "epoch": 2.58, "learning_rate": 3.386606069711539e-05, "loss": 0.421, "step": 8595 }, { "epoch": 2.58, "learning_rate": 3.386418269230769e-05, "loss": 0.433, "step": 8596 }, { "epoch": 2.58, "learning_rate": 3.3862304687500004e-05, "loss": 0.2929, "step": 8597 }, { "epoch": 2.58, "learning_rate": 3.386042668269231e-05, "loss": 0.41, "step": 8598 }, { "epoch": 2.58, "learning_rate": 3.3858548677884614e-05, "loss": 0.4438, "step": 8599 }, { "epoch": 2.58, "learning_rate": 3.3856670673076925e-05, "loss": 0.3493, "step": 8600 }, { "epoch": 2.58, "learning_rate": 3.385479266826923e-05, "loss": 0.4385, "step": 8601 }, { "epoch": 2.58, "learning_rate": 3.3852914663461535e-05, "loss": 0.361, "step": 8602 }, { "epoch": 2.59, "learning_rate": 3.3851036658653847e-05, "loss": 0.4997, "step": 8603 }, { "epoch": 2.59, "learning_rate": 3.384915865384616e-05, "loss": 0.6491, "step": 8604 }, { "epoch": 2.59, "learning_rate": 3.384728064903846e-05, "loss": 0.3832, "step": 8605 }, { "epoch": 2.59, "learning_rate": 3.3845402644230774e-05, "loss": 0.4194, "step": 8606 }, { "epoch": 2.59, "learning_rate": 3.384352463942308e-05, "loss": 0.2685, "step": 8607 }, { "epoch": 2.59, "learning_rate": 3.3841646634615384e-05, "loss": 0.3907, "step": 8608 }, { "epoch": 2.59, "learning_rate": 3.3839768629807695e-05, "loss": 0.2716, "step": 8609 }, { "epoch": 2.59, "learning_rate": 3.3837890625e-05, "loss": 0.391, "step": 8610 }, { "epoch": 2.59, "learning_rate": 3.383601262019231e-05, "loss": 0.3554, "step": 8611 }, { "epoch": 2.59, "learning_rate": 3.3834134615384617e-05, "loss": 0.4058, "step": 8612 }, { "epoch": 2.59, "learning_rate": 3.383225661057692e-05, "loss": 0.5275, "step": 8613 }, { "epoch": 2.59, "learning_rate": 3.383037860576923e-05, "loss": 0.4526, "step": 8614 }, { "epoch": 2.59, "learning_rate": 3.382850060096154e-05, "loss": 0.4785, "step": 8615 }, { "epoch": 2.59, "learning_rate": 3.382662259615384e-05, "loss": 0.4655, "step": 8616 }, { "epoch": 2.59, "learning_rate": 3.382474459134616e-05, "loss": 0.435, "step": 8617 }, { "epoch": 2.59, "learning_rate": 3.3822866586538466e-05, "loss": 0.2705, "step": 8618 }, { "epoch": 2.59, "learning_rate": 3.382098858173077e-05, "loss": 0.3144, "step": 8619 }, { "epoch": 2.59, "learning_rate": 3.381911057692308e-05, "loss": 0.5047, "step": 8620 }, { "epoch": 2.59, "learning_rate": 3.3817232572115387e-05, "loss": 0.4315, "step": 8621 }, { "epoch": 2.59, "learning_rate": 3.381535456730769e-05, "loss": 0.3997, "step": 8622 }, { "epoch": 2.59, "learning_rate": 3.38134765625e-05, "loss": 0.3031, "step": 8623 }, { "epoch": 2.59, "learning_rate": 3.381159855769231e-05, "loss": 0.4199, "step": 8624 }, { "epoch": 2.59, "learning_rate": 3.380972055288462e-05, "loss": 0.5348, "step": 8625 }, { "epoch": 2.59, "learning_rate": 3.3807842548076924e-05, "loss": 0.3283, "step": 8626 }, { "epoch": 2.59, "learning_rate": 3.380596454326923e-05, "loss": 0.332, "step": 8627 }, { "epoch": 2.59, "learning_rate": 3.380408653846154e-05, "loss": 0.4583, "step": 8628 }, { "epoch": 2.59, "learning_rate": 3.3802208533653845e-05, "loss": 0.3696, "step": 8629 }, { "epoch": 2.59, "learning_rate": 3.380033052884616e-05, "loss": 0.4039, "step": 8630 }, { "epoch": 2.59, "learning_rate": 3.379845252403847e-05, "loss": 0.4359, "step": 8631 }, { "epoch": 2.59, "learning_rate": 3.379657451923077e-05, "loss": 0.4666, "step": 8632 }, { "epoch": 2.59, "learning_rate": 3.379469651442308e-05, "loss": 0.385, "step": 8633 }, { "epoch": 2.59, "learning_rate": 3.379281850961539e-05, "loss": 0.3491, "step": 8634 }, { "epoch": 2.59, "learning_rate": 3.3790940504807694e-05, "loss": 0.3377, "step": 8635 }, { "epoch": 2.59, "learning_rate": 3.37890625e-05, "loss": 0.3104, "step": 8636 }, { "epoch": 2.6, "learning_rate": 3.378718449519231e-05, "loss": 0.4059, "step": 8637 }, { "epoch": 2.6, "learning_rate": 3.3785306490384615e-05, "loss": 0.5662, "step": 8638 }, { "epoch": 2.6, "learning_rate": 3.378342848557693e-05, "loss": 0.3351, "step": 8639 }, { "epoch": 2.6, "learning_rate": 3.378155048076923e-05, "loss": 0.3406, "step": 8640 }, { "epoch": 2.6, "learning_rate": 3.3779672475961536e-05, "loss": 0.3878, "step": 8641 }, { "epoch": 2.6, "learning_rate": 3.377779447115385e-05, "loss": 0.2449, "step": 8642 }, { "epoch": 2.6, "learning_rate": 3.377591646634616e-05, "loss": 0.2647, "step": 8643 }, { "epoch": 2.6, "learning_rate": 3.3774038461538464e-05, "loss": 0.3331, "step": 8644 }, { "epoch": 2.6, "learning_rate": 3.3772160456730776e-05, "loss": 0.4098, "step": 8645 }, { "epoch": 2.6, "learning_rate": 3.377028245192308e-05, "loss": 0.3677, "step": 8646 }, { "epoch": 2.6, "learning_rate": 3.3768404447115385e-05, "loss": 0.3795, "step": 8647 }, { "epoch": 2.6, "learning_rate": 3.37665264423077e-05, "loss": 0.6173, "step": 8648 }, { "epoch": 2.6, "learning_rate": 3.37646484375e-05, "loss": 0.3711, "step": 8649 }, { "epoch": 2.6, "learning_rate": 3.3762770432692306e-05, "loss": 0.4641, "step": 8650 }, { "epoch": 2.6, "learning_rate": 3.376089242788462e-05, "loss": 0.349, "step": 8651 }, { "epoch": 2.6, "learning_rate": 3.375901442307692e-05, "loss": 0.314, "step": 8652 }, { "epoch": 2.6, "learning_rate": 3.375713641826923e-05, "loss": 0.3718, "step": 8653 }, { "epoch": 2.6, "learning_rate": 3.375525841346154e-05, "loss": 0.264, "step": 8654 }, { "epoch": 2.6, "learning_rate": 3.3753380408653844e-05, "loss": 0.4032, "step": 8655 }, { "epoch": 2.6, "learning_rate": 3.3751502403846155e-05, "loss": 0.5066, "step": 8656 }, { "epoch": 2.6, "learning_rate": 3.374962439903847e-05, "loss": 0.3238, "step": 8657 }, { "epoch": 2.6, "learning_rate": 3.374774639423077e-05, "loss": 0.302, "step": 8658 }, { "epoch": 2.6, "learning_rate": 3.374586838942308e-05, "loss": 0.4995, "step": 8659 }, { "epoch": 2.6, "learning_rate": 3.374399038461539e-05, "loss": 0.4242, "step": 8660 }, { "epoch": 2.6, "learning_rate": 3.374211237980769e-05, "loss": 0.3236, "step": 8661 }, { "epoch": 2.6, "learning_rate": 3.3740234375000004e-05, "loss": 0.3065, "step": 8662 }, { "epoch": 2.6, "learning_rate": 3.373835637019231e-05, "loss": 0.3846, "step": 8663 }, { "epoch": 2.6, "learning_rate": 3.3736478365384614e-05, "loss": 0.2845, "step": 8664 }, { "epoch": 2.6, "learning_rate": 3.3734600360576925e-05, "loss": 0.3332, "step": 8665 }, { "epoch": 2.6, "learning_rate": 3.373272235576923e-05, "loss": 0.3744, "step": 8666 }, { "epoch": 2.6, "learning_rate": 3.3730844350961535e-05, "loss": 0.3891, "step": 8667 }, { "epoch": 2.6, "learning_rate": 3.3728966346153846e-05, "loss": 0.3079, "step": 8668 }, { "epoch": 2.6, "learning_rate": 3.372708834134616e-05, "loss": 0.3651, "step": 8669 }, { "epoch": 2.61, "learning_rate": 3.372521033653846e-05, "loss": 0.2687, "step": 8670 }, { "epoch": 2.61, "learning_rate": 3.3723332331730774e-05, "loss": 0.3307, "step": 8671 }, { "epoch": 2.61, "learning_rate": 3.372145432692308e-05, "loss": 0.3672, "step": 8672 }, { "epoch": 2.61, "learning_rate": 3.3719576322115384e-05, "loss": 0.3841, "step": 8673 }, { "epoch": 2.61, "learning_rate": 3.3717698317307695e-05, "loss": 0.4734, "step": 8674 }, { "epoch": 2.61, "learning_rate": 3.37158203125e-05, "loss": 0.4571, "step": 8675 }, { "epoch": 2.61, "learning_rate": 3.371394230769231e-05, "loss": 0.4848, "step": 8676 }, { "epoch": 2.61, "learning_rate": 3.3712064302884616e-05, "loss": 0.3403, "step": 8677 }, { "epoch": 2.61, "learning_rate": 3.371018629807692e-05, "loss": 0.2822, "step": 8678 }, { "epoch": 2.61, "learning_rate": 3.370830829326923e-05, "loss": 0.2775, "step": 8679 }, { "epoch": 2.61, "learning_rate": 3.370643028846154e-05, "loss": 0.4104, "step": 8680 }, { "epoch": 2.61, "learning_rate": 3.370455228365384e-05, "loss": 0.3552, "step": 8681 }, { "epoch": 2.61, "learning_rate": 3.370267427884616e-05, "loss": 0.4213, "step": 8682 }, { "epoch": 2.61, "learning_rate": 3.3700796274038465e-05, "loss": 0.3997, "step": 8683 }, { "epoch": 2.61, "learning_rate": 3.369891826923077e-05, "loss": 0.3831, "step": 8684 }, { "epoch": 2.61, "learning_rate": 3.369704026442308e-05, "loss": 0.4406, "step": 8685 }, { "epoch": 2.61, "learning_rate": 3.3695162259615386e-05, "loss": 0.5672, "step": 8686 }, { "epoch": 2.61, "learning_rate": 3.369328425480769e-05, "loss": 0.436, "step": 8687 }, { "epoch": 2.61, "learning_rate": 3.369140625e-05, "loss": 0.3645, "step": 8688 }, { "epoch": 2.61, "learning_rate": 3.368952824519231e-05, "loss": 0.2812, "step": 8689 }, { "epoch": 2.61, "learning_rate": 3.368765024038462e-05, "loss": 0.3205, "step": 8690 }, { "epoch": 2.61, "learning_rate": 3.3685772235576924e-05, "loss": 0.3658, "step": 8691 }, { "epoch": 2.61, "learning_rate": 3.368389423076923e-05, "loss": 0.4659, "step": 8692 }, { "epoch": 2.61, "learning_rate": 3.368201622596154e-05, "loss": 0.4005, "step": 8693 }, { "epoch": 2.61, "learning_rate": 3.3680138221153845e-05, "loss": 0.4358, "step": 8694 }, { "epoch": 2.61, "learning_rate": 3.3678260216346157e-05, "loss": 0.2744, "step": 8695 }, { "epoch": 2.61, "learning_rate": 3.367638221153847e-05, "loss": 0.4359, "step": 8696 }, { "epoch": 2.61, "learning_rate": 3.367450420673077e-05, "loss": 0.3752, "step": 8697 }, { "epoch": 2.61, "learning_rate": 3.367262620192308e-05, "loss": 0.4632, "step": 8698 }, { "epoch": 2.61, "learning_rate": 3.367074819711539e-05, "loss": 0.4619, "step": 8699 }, { "epoch": 2.61, "learning_rate": 3.3668870192307694e-05, "loss": 0.4352, "step": 8700 }, { "epoch": 2.61, "learning_rate": 3.36669921875e-05, "loss": 0.3492, "step": 8701 }, { "epoch": 2.61, "learning_rate": 3.366511418269231e-05, "loss": 0.32, "step": 8702 }, { "epoch": 2.62, "learning_rate": 3.3663236177884615e-05, "loss": 0.346, "step": 8703 }, { "epoch": 2.62, "learning_rate": 3.3661358173076927e-05, "loss": 0.3754, "step": 8704 }, { "epoch": 2.62, "learning_rate": 3.365948016826923e-05, "loss": 0.4527, "step": 8705 }, { "epoch": 2.62, "learning_rate": 3.3657602163461536e-05, "loss": 0.3258, "step": 8706 }, { "epoch": 2.62, "learning_rate": 3.365572415865385e-05, "loss": 0.4205, "step": 8707 }, { "epoch": 2.62, "learning_rate": 3.365384615384616e-05, "loss": 0.2519, "step": 8708 }, { "epoch": 2.62, "learning_rate": 3.3651968149038464e-05, "loss": 0.3968, "step": 8709 }, { "epoch": 2.62, "learning_rate": 3.3650090144230776e-05, "loss": 0.3819, "step": 8710 }, { "epoch": 2.62, "learning_rate": 3.364821213942308e-05, "loss": 0.4123, "step": 8711 }, { "epoch": 2.62, "learning_rate": 3.3646334134615385e-05, "loss": 0.3689, "step": 8712 }, { "epoch": 2.62, "learning_rate": 3.36444561298077e-05, "loss": 0.4936, "step": 8713 }, { "epoch": 2.62, "learning_rate": 3.3642578125e-05, "loss": 0.3764, "step": 8714 }, { "epoch": 2.62, "learning_rate": 3.3640700120192306e-05, "loss": 0.31, "step": 8715 }, { "epoch": 2.62, "learning_rate": 3.363882211538462e-05, "loss": 0.2709, "step": 8716 }, { "epoch": 2.62, "learning_rate": 3.363694411057692e-05, "loss": 0.2409, "step": 8717 }, { "epoch": 2.62, "learning_rate": 3.3635066105769234e-05, "loss": 0.2126, "step": 8718 }, { "epoch": 2.62, "learning_rate": 3.363318810096154e-05, "loss": 0.3352, "step": 8719 }, { "epoch": 2.62, "learning_rate": 3.3631310096153844e-05, "loss": 0.5148, "step": 8720 }, { "epoch": 2.62, "learning_rate": 3.3629432091346155e-05, "loss": 0.5353, "step": 8721 }, { "epoch": 2.62, "learning_rate": 3.362755408653847e-05, "loss": 0.386, "step": 8722 }, { "epoch": 2.62, "learning_rate": 3.362567608173077e-05, "loss": 0.4098, "step": 8723 }, { "epoch": 2.62, "learning_rate": 3.362379807692308e-05, "loss": 0.4223, "step": 8724 }, { "epoch": 2.62, "learning_rate": 3.362192007211539e-05, "loss": 0.3783, "step": 8725 }, { "epoch": 2.62, "learning_rate": 3.362004206730769e-05, "loss": 0.3694, "step": 8726 }, { "epoch": 2.62, "learning_rate": 3.3618164062500004e-05, "loss": 0.4189, "step": 8727 }, { "epoch": 2.62, "learning_rate": 3.361628605769231e-05, "loss": 0.3439, "step": 8728 }, { "epoch": 2.62, "learning_rate": 3.3614408052884614e-05, "loss": 0.5085, "step": 8729 }, { "epoch": 2.62, "learning_rate": 3.3612530048076925e-05, "loss": 0.3394, "step": 8730 }, { "epoch": 2.62, "learning_rate": 3.361065204326923e-05, "loss": 0.265, "step": 8731 }, { "epoch": 2.62, "learning_rate": 3.3608774038461535e-05, "loss": 0.4657, "step": 8732 }, { "epoch": 2.62, "learning_rate": 3.3606896033653846e-05, "loss": 0.3597, "step": 8733 }, { "epoch": 2.62, "learning_rate": 3.360501802884616e-05, "loss": 0.3757, "step": 8734 }, { "epoch": 2.62, "learning_rate": 3.360314002403846e-05, "loss": 0.3692, "step": 8735 }, { "epoch": 2.62, "learning_rate": 3.3601262019230774e-05, "loss": 0.3357, "step": 8736 }, { "epoch": 2.63, "learning_rate": 3.359938401442308e-05, "loss": 0.282, "step": 8737 }, { "epoch": 2.63, "learning_rate": 3.3597506009615384e-05, "loss": 0.4401, "step": 8738 }, { "epoch": 2.63, "learning_rate": 3.3595628004807695e-05, "loss": 0.3149, "step": 8739 }, { "epoch": 2.63, "learning_rate": 3.359375e-05, "loss": 0.3406, "step": 8740 }, { "epoch": 2.63, "learning_rate": 3.359187199519231e-05, "loss": 0.3925, "step": 8741 }, { "epoch": 2.63, "learning_rate": 3.3589993990384616e-05, "loss": 0.4406, "step": 8742 }, { "epoch": 2.63, "learning_rate": 3.358811598557692e-05, "loss": 0.5738, "step": 8743 }, { "epoch": 2.63, "learning_rate": 3.358623798076923e-05, "loss": 0.3055, "step": 8744 }, { "epoch": 2.63, "learning_rate": 3.358435997596154e-05, "loss": 0.4258, "step": 8745 }, { "epoch": 2.63, "learning_rate": 3.358248197115384e-05, "loss": 0.3751, "step": 8746 }, { "epoch": 2.63, "learning_rate": 3.358060396634616e-05, "loss": 0.3045, "step": 8747 }, { "epoch": 2.63, "learning_rate": 3.3578725961538465e-05, "loss": 0.4759, "step": 8748 }, { "epoch": 2.63, "learning_rate": 3.357684795673077e-05, "loss": 0.4364, "step": 8749 }, { "epoch": 2.63, "learning_rate": 3.357496995192308e-05, "loss": 0.422, "step": 8750 }, { "epoch": 2.63, "learning_rate": 3.3573091947115386e-05, "loss": 0.299, "step": 8751 }, { "epoch": 2.63, "learning_rate": 3.357121394230769e-05, "loss": 0.4132, "step": 8752 }, { "epoch": 2.63, "learning_rate": 3.35693359375e-05, "loss": 0.287, "step": 8753 }, { "epoch": 2.63, "learning_rate": 3.356745793269231e-05, "loss": 0.2578, "step": 8754 }, { "epoch": 2.63, "learning_rate": 3.356557992788462e-05, "loss": 0.305, "step": 8755 }, { "epoch": 2.63, "learning_rate": 3.3563701923076924e-05, "loss": 0.309, "step": 8756 }, { "epoch": 2.63, "learning_rate": 3.356182391826923e-05, "loss": 0.2744, "step": 8757 }, { "epoch": 2.63, "learning_rate": 3.355994591346154e-05, "loss": 0.4923, "step": 8758 }, { "epoch": 2.63, "learning_rate": 3.3558067908653845e-05, "loss": 0.4219, "step": 8759 }, { "epoch": 2.63, "learning_rate": 3.3556189903846156e-05, "loss": 0.3459, "step": 8760 }, { "epoch": 2.63, "learning_rate": 3.355431189903847e-05, "loss": 0.4093, "step": 8761 }, { "epoch": 2.63, "learning_rate": 3.355243389423077e-05, "loss": 0.4619, "step": 8762 }, { "epoch": 2.63, "learning_rate": 3.355055588942308e-05, "loss": 0.3516, "step": 8763 }, { "epoch": 2.63, "learning_rate": 3.354867788461539e-05, "loss": 0.317, "step": 8764 }, { "epoch": 2.63, "learning_rate": 3.3546799879807694e-05, "loss": 0.3747, "step": 8765 }, { "epoch": 2.63, "learning_rate": 3.3544921875e-05, "loss": 0.4062, "step": 8766 }, { "epoch": 2.63, "learning_rate": 3.354304387019231e-05, "loss": 0.4946, "step": 8767 }, { "epoch": 2.63, "learning_rate": 3.3541165865384615e-05, "loss": 0.3314, "step": 8768 }, { "epoch": 2.63, "learning_rate": 3.3539287860576926e-05, "loss": 0.3522, "step": 8769 }, { "epoch": 2.64, "learning_rate": 3.353740985576923e-05, "loss": 0.4255, "step": 8770 }, { "epoch": 2.64, "learning_rate": 3.3535531850961536e-05, "loss": 0.3008, "step": 8771 }, { "epoch": 2.64, "learning_rate": 3.353365384615385e-05, "loss": 0.3855, "step": 8772 }, { "epoch": 2.64, "learning_rate": 3.353177584134616e-05, "loss": 0.4269, "step": 8773 }, { "epoch": 2.64, "learning_rate": 3.3529897836538464e-05, "loss": 0.2837, "step": 8774 }, { "epoch": 2.64, "learning_rate": 3.3528019831730775e-05, "loss": 0.3578, "step": 8775 }, { "epoch": 2.64, "learning_rate": 3.352614182692308e-05, "loss": 0.2777, "step": 8776 }, { "epoch": 2.64, "learning_rate": 3.3524263822115385e-05, "loss": 0.3034, "step": 8777 }, { "epoch": 2.64, "learning_rate": 3.3522385817307697e-05, "loss": 0.3593, "step": 8778 }, { "epoch": 2.64, "learning_rate": 3.35205078125e-05, "loss": 0.304, "step": 8779 }, { "epoch": 2.64, "learning_rate": 3.3518629807692306e-05, "loss": 0.4399, "step": 8780 }, { "epoch": 2.64, "learning_rate": 3.351675180288462e-05, "loss": 0.3561, "step": 8781 }, { "epoch": 2.64, "learning_rate": 3.351487379807692e-05, "loss": 0.4107, "step": 8782 }, { "epoch": 2.64, "learning_rate": 3.3512995793269234e-05, "loss": 0.4438, "step": 8783 }, { "epoch": 2.64, "learning_rate": 3.351111778846154e-05, "loss": 0.4809, "step": 8784 }, { "epoch": 2.64, "learning_rate": 3.3509239783653843e-05, "loss": 0.3098, "step": 8785 }, { "epoch": 2.64, "learning_rate": 3.3507361778846155e-05, "loss": 0.3726, "step": 8786 }, { "epoch": 2.64, "learning_rate": 3.3505483774038467e-05, "loss": 0.3383, "step": 8787 }, { "epoch": 2.64, "learning_rate": 3.350360576923077e-05, "loss": 0.4213, "step": 8788 }, { "epoch": 2.64, "learning_rate": 3.350172776442308e-05, "loss": 0.3205, "step": 8789 }, { "epoch": 2.64, "learning_rate": 3.349984975961539e-05, "loss": 0.3143, "step": 8790 }, { "epoch": 2.64, "learning_rate": 3.349797175480769e-05, "loss": 0.5188, "step": 8791 }, { "epoch": 2.64, "learning_rate": 3.3496093750000004e-05, "loss": 0.2352, "step": 8792 }, { "epoch": 2.64, "learning_rate": 3.349421574519231e-05, "loss": 0.4231, "step": 8793 }, { "epoch": 2.64, "learning_rate": 3.3492337740384614e-05, "loss": 0.3457, "step": 8794 }, { "epoch": 2.64, "learning_rate": 3.3490459735576925e-05, "loss": 0.3018, "step": 8795 }, { "epoch": 2.64, "learning_rate": 3.348858173076923e-05, "loss": 0.4874, "step": 8796 }, { "epoch": 2.64, "learning_rate": 3.3486703725961535e-05, "loss": 0.3742, "step": 8797 }, { "epoch": 2.64, "learning_rate": 3.3484825721153846e-05, "loss": 0.4841, "step": 8798 }, { "epoch": 2.64, "learning_rate": 3.348294771634616e-05, "loss": 0.3031, "step": 8799 }, { "epoch": 2.64, "learning_rate": 3.348106971153846e-05, "loss": 0.2637, "step": 8800 }, { "epoch": 2.64, "learning_rate": 3.3479191706730774e-05, "loss": 0.3555, "step": 8801 }, { "epoch": 2.64, "learning_rate": 3.347731370192308e-05, "loss": 0.3551, "step": 8802 }, { "epoch": 2.65, "learning_rate": 3.347543569711539e-05, "loss": 0.2926, "step": 8803 }, { "epoch": 2.65, "learning_rate": 3.3473557692307695e-05, "loss": 0.3572, "step": 8804 }, { "epoch": 2.65, "learning_rate": 3.34716796875e-05, "loss": 0.4308, "step": 8805 }, { "epoch": 2.65, "learning_rate": 3.346980168269231e-05, "loss": 0.3524, "step": 8806 }, { "epoch": 2.65, "learning_rate": 3.3467923677884616e-05, "loss": 0.3399, "step": 8807 }, { "epoch": 2.65, "learning_rate": 3.346604567307692e-05, "loss": 0.4343, "step": 8808 }, { "epoch": 2.65, "learning_rate": 3.346416766826923e-05, "loss": 0.3363, "step": 8809 }, { "epoch": 2.65, "learning_rate": 3.346228966346154e-05, "loss": 0.2907, "step": 8810 }, { "epoch": 2.65, "learning_rate": 3.346041165865384e-05, "loss": 0.3406, "step": 8811 }, { "epoch": 2.65, "learning_rate": 3.345853365384616e-05, "loss": 0.2485, "step": 8812 }, { "epoch": 2.65, "learning_rate": 3.3456655649038465e-05, "loss": 0.3439, "step": 8813 }, { "epoch": 2.65, "learning_rate": 3.345477764423077e-05, "loss": 0.2828, "step": 8814 }, { "epoch": 2.65, "learning_rate": 3.345289963942308e-05, "loss": 0.3554, "step": 8815 }, { "epoch": 2.65, "learning_rate": 3.3451021634615386e-05, "loss": 0.3616, "step": 8816 }, { "epoch": 2.65, "learning_rate": 3.344914362980769e-05, "loss": 0.3008, "step": 8817 }, { "epoch": 2.65, "learning_rate": 3.3447265625e-05, "loss": 0.4234, "step": 8818 }, { "epoch": 2.65, "learning_rate": 3.344538762019231e-05, "loss": 0.3714, "step": 8819 }, { "epoch": 2.65, "learning_rate": 3.344350961538462e-05, "loss": 0.2811, "step": 8820 }, { "epoch": 2.65, "learning_rate": 3.3441631610576924e-05, "loss": 0.2743, "step": 8821 }, { "epoch": 2.65, "learning_rate": 3.343975360576923e-05, "loss": 0.4402, "step": 8822 }, { "epoch": 2.65, "learning_rate": 3.343787560096154e-05, "loss": 0.3757, "step": 8823 }, { "epoch": 2.65, "learning_rate": 3.3435997596153845e-05, "loss": 0.5637, "step": 8824 }, { "epoch": 2.65, "learning_rate": 3.3434119591346156e-05, "loss": 0.4307, "step": 8825 }, { "epoch": 2.65, "learning_rate": 3.343224158653847e-05, "loss": 0.3454, "step": 8826 }, { "epoch": 2.65, "learning_rate": 3.343036358173077e-05, "loss": 0.2586, "step": 8827 }, { "epoch": 2.65, "learning_rate": 3.342848557692308e-05, "loss": 0.3769, "step": 8828 }, { "epoch": 2.65, "learning_rate": 3.342660757211539e-05, "loss": 0.2928, "step": 8829 }, { "epoch": 2.65, "learning_rate": 3.3424729567307694e-05, "loss": 0.44, "step": 8830 }, { "epoch": 2.65, "learning_rate": 3.34228515625e-05, "loss": 0.5269, "step": 8831 }, { "epoch": 2.65, "learning_rate": 3.342097355769231e-05, "loss": 0.2925, "step": 8832 }, { "epoch": 2.65, "learning_rate": 3.3419095552884615e-05, "loss": 0.3264, "step": 8833 }, { "epoch": 2.65, "learning_rate": 3.3417217548076926e-05, "loss": 0.4987, "step": 8834 }, { "epoch": 2.65, "learning_rate": 3.341533954326923e-05, "loss": 0.3636, "step": 8835 }, { "epoch": 2.66, "learning_rate": 3.3413461538461536e-05, "loss": 0.2937, "step": 8836 }, { "epoch": 2.66, "learning_rate": 3.341158353365385e-05, "loss": 0.2979, "step": 8837 }, { "epoch": 2.66, "learning_rate": 3.340970552884616e-05, "loss": 0.4039, "step": 8838 }, { "epoch": 2.66, "learning_rate": 3.3407827524038464e-05, "loss": 0.4758, "step": 8839 }, { "epoch": 2.66, "learning_rate": 3.3405949519230775e-05, "loss": 0.357, "step": 8840 }, { "epoch": 2.66, "learning_rate": 3.340407151442308e-05, "loss": 0.4216, "step": 8841 }, { "epoch": 2.66, "learning_rate": 3.3402193509615385e-05, "loss": 0.3877, "step": 8842 }, { "epoch": 2.66, "learning_rate": 3.3400315504807696e-05, "loss": 0.4251, "step": 8843 }, { "epoch": 2.66, "learning_rate": 3.33984375e-05, "loss": 0.412, "step": 8844 }, { "epoch": 2.66, "learning_rate": 3.3396559495192306e-05, "loss": 0.2546, "step": 8845 }, { "epoch": 2.66, "learning_rate": 3.339468149038462e-05, "loss": 0.4757, "step": 8846 }, { "epoch": 2.66, "learning_rate": 3.339280348557692e-05, "loss": 0.2614, "step": 8847 }, { "epoch": 2.66, "learning_rate": 3.3390925480769234e-05, "loss": 0.4335, "step": 8848 }, { "epoch": 2.66, "learning_rate": 3.338904747596154e-05, "loss": 0.2296, "step": 8849 }, { "epoch": 2.66, "learning_rate": 3.338716947115384e-05, "loss": 0.3048, "step": 8850 }, { "epoch": 2.66, "learning_rate": 3.3385291466346155e-05, "loss": 0.3699, "step": 8851 }, { "epoch": 2.66, "learning_rate": 3.3383413461538466e-05, "loss": 0.3551, "step": 8852 }, { "epoch": 2.66, "learning_rate": 3.338153545673077e-05, "loss": 0.3867, "step": 8853 }, { "epoch": 2.66, "learning_rate": 3.337965745192308e-05, "loss": 0.3297, "step": 8854 }, { "epoch": 2.66, "learning_rate": 3.337777944711539e-05, "loss": 0.3943, "step": 8855 }, { "epoch": 2.66, "learning_rate": 3.337590144230769e-05, "loss": 0.4887, "step": 8856 }, { "epoch": 2.66, "learning_rate": 3.3374023437500004e-05, "loss": 0.464, "step": 8857 }, { "epoch": 2.66, "learning_rate": 3.337214543269231e-05, "loss": 0.3204, "step": 8858 }, { "epoch": 2.66, "learning_rate": 3.3370267427884613e-05, "loss": 0.308, "step": 8859 }, { "epoch": 2.66, "learning_rate": 3.3368389423076925e-05, "loss": 0.3361, "step": 8860 }, { "epoch": 2.66, "learning_rate": 3.336651141826923e-05, "loss": 0.2774, "step": 8861 }, { "epoch": 2.66, "learning_rate": 3.336463341346154e-05, "loss": 0.3906, "step": 8862 }, { "epoch": 2.66, "learning_rate": 3.3362755408653846e-05, "loss": 0.3422, "step": 8863 }, { "epoch": 2.66, "learning_rate": 3.336087740384616e-05, "loss": 0.5127, "step": 8864 }, { "epoch": 2.66, "learning_rate": 3.335899939903846e-05, "loss": 0.3265, "step": 8865 }, { "epoch": 2.66, "learning_rate": 3.3357121394230774e-05, "loss": 0.4074, "step": 8866 }, { "epoch": 2.66, "learning_rate": 3.335524338942308e-05, "loss": 0.3634, "step": 8867 }, { "epoch": 2.66, "learning_rate": 3.335336538461539e-05, "loss": 0.3871, "step": 8868 }, { "epoch": 2.66, "learning_rate": 3.3351487379807695e-05, "loss": 0.4734, "step": 8869 }, { "epoch": 2.67, "learning_rate": 3.3349609375e-05, "loss": 0.3109, "step": 8870 }, { "epoch": 2.67, "learning_rate": 3.334773137019231e-05, "loss": 0.3837, "step": 8871 }, { "epoch": 2.67, "learning_rate": 3.3345853365384616e-05, "loss": 0.3341, "step": 8872 }, { "epoch": 2.67, "learning_rate": 3.334397536057692e-05, "loss": 0.4736, "step": 8873 }, { "epoch": 2.67, "learning_rate": 3.334209735576923e-05, "loss": 0.377, "step": 8874 }, { "epoch": 2.67, "learning_rate": 3.334021935096154e-05, "loss": 0.3004, "step": 8875 }, { "epoch": 2.67, "learning_rate": 3.333834134615384e-05, "loss": 0.4512, "step": 8876 }, { "epoch": 2.67, "learning_rate": 3.333646334134616e-05, "loss": 0.4521, "step": 8877 }, { "epoch": 2.67, "learning_rate": 3.3334585336538465e-05, "loss": 0.3978, "step": 8878 }, { "epoch": 2.67, "learning_rate": 3.333270733173077e-05, "loss": 0.4671, "step": 8879 }, { "epoch": 2.67, "learning_rate": 3.333082932692308e-05, "loss": 0.3753, "step": 8880 }, { "epoch": 2.67, "learning_rate": 3.3328951322115386e-05, "loss": 0.3117, "step": 8881 }, { "epoch": 2.67, "learning_rate": 3.332707331730769e-05, "loss": 0.1825, "step": 8882 }, { "epoch": 2.67, "learning_rate": 3.33251953125e-05, "loss": 0.3656, "step": 8883 }, { "epoch": 2.67, "learning_rate": 3.332331730769231e-05, "loss": 0.2682, "step": 8884 }, { "epoch": 2.67, "learning_rate": 3.332143930288462e-05, "loss": 0.3751, "step": 8885 }, { "epoch": 2.67, "learning_rate": 3.3319561298076924e-05, "loss": 0.4082, "step": 8886 }, { "epoch": 2.67, "learning_rate": 3.331768329326923e-05, "loss": 0.3651, "step": 8887 }, { "epoch": 2.67, "learning_rate": 3.331580528846154e-05, "loss": 0.347, "step": 8888 }, { "epoch": 2.67, "learning_rate": 3.3313927283653845e-05, "loss": 0.4146, "step": 8889 }, { "epoch": 2.67, "learning_rate": 3.3312049278846156e-05, "loss": 0.3769, "step": 8890 }, { "epoch": 2.67, "learning_rate": 3.331017127403847e-05, "loss": 0.4337, "step": 8891 }, { "epoch": 2.67, "learning_rate": 3.330829326923077e-05, "loss": 0.2492, "step": 8892 }, { "epoch": 2.67, "learning_rate": 3.330641526442308e-05, "loss": 0.6046, "step": 8893 }, { "epoch": 2.67, "learning_rate": 3.330453725961539e-05, "loss": 0.3799, "step": 8894 }, { "epoch": 2.67, "learning_rate": 3.3302659254807694e-05, "loss": 0.461, "step": 8895 }, { "epoch": 2.67, "learning_rate": 3.330078125e-05, "loss": 0.4905, "step": 8896 }, { "epoch": 2.67, "learning_rate": 3.329890324519231e-05, "loss": 0.6557, "step": 8897 }, { "epoch": 2.67, "learning_rate": 3.3297025240384615e-05, "loss": 0.4818, "step": 8898 }, { "epoch": 2.67, "learning_rate": 3.3295147235576926e-05, "loss": 0.2473, "step": 8899 }, { "epoch": 2.67, "learning_rate": 3.329326923076923e-05, "loss": 0.2796, "step": 8900 }, { "epoch": 2.67, "learning_rate": 3.3291391225961536e-05, "loss": 0.2845, "step": 8901 }, { "epoch": 2.67, "learning_rate": 3.328951322115385e-05, "loss": 0.2764, "step": 8902 }, { "epoch": 2.68, "learning_rate": 3.328763521634616e-05, "loss": 0.5341, "step": 8903 }, { "epoch": 2.68, "learning_rate": 3.3285757211538464e-05, "loss": 0.3551, "step": 8904 }, { "epoch": 2.68, "learning_rate": 3.3283879206730775e-05, "loss": 0.4368, "step": 8905 }, { "epoch": 2.68, "learning_rate": 3.328200120192308e-05, "loss": 0.3877, "step": 8906 }, { "epoch": 2.68, "learning_rate": 3.3280123197115385e-05, "loss": 0.3335, "step": 8907 }, { "epoch": 2.68, "learning_rate": 3.3278245192307696e-05, "loss": 0.4088, "step": 8908 }, { "epoch": 2.68, "learning_rate": 3.32763671875e-05, "loss": 0.4094, "step": 8909 }, { "epoch": 2.68, "learning_rate": 3.3274489182692306e-05, "loss": 0.4036, "step": 8910 }, { "epoch": 2.68, "learning_rate": 3.327261117788462e-05, "loss": 0.3124, "step": 8911 }, { "epoch": 2.68, "learning_rate": 3.327073317307692e-05, "loss": 0.2329, "step": 8912 }, { "epoch": 2.68, "learning_rate": 3.3268855168269234e-05, "loss": 0.4861, "step": 8913 }, { "epoch": 2.68, "learning_rate": 3.326697716346154e-05, "loss": 0.4768, "step": 8914 }, { "epoch": 2.68, "learning_rate": 3.326509915865384e-05, "loss": 0.3833, "step": 8915 }, { "epoch": 2.68, "learning_rate": 3.3263221153846155e-05, "loss": 0.2857, "step": 8916 }, { "epoch": 2.68, "learning_rate": 3.3261343149038466e-05, "loss": 0.2421, "step": 8917 }, { "epoch": 2.68, "learning_rate": 3.325946514423077e-05, "loss": 0.4454, "step": 8918 }, { "epoch": 2.68, "learning_rate": 3.325758713942308e-05, "loss": 0.3899, "step": 8919 }, { "epoch": 2.68, "learning_rate": 3.325570913461539e-05, "loss": 0.3937, "step": 8920 }, { "epoch": 2.68, "learning_rate": 3.325383112980769e-05, "loss": 0.3418, "step": 8921 }, { "epoch": 2.68, "learning_rate": 3.3251953125000004e-05, "loss": 0.33, "step": 8922 }, { "epoch": 2.68, "learning_rate": 3.325007512019231e-05, "loss": 0.5701, "step": 8923 }, { "epoch": 2.68, "learning_rate": 3.324819711538461e-05, "loss": 0.3702, "step": 8924 }, { "epoch": 2.68, "learning_rate": 3.3246319110576925e-05, "loss": 0.3756, "step": 8925 }, { "epoch": 2.68, "learning_rate": 3.324444110576923e-05, "loss": 0.3685, "step": 8926 }, { "epoch": 2.68, "learning_rate": 3.324256310096154e-05, "loss": 0.2791, "step": 8927 }, { "epoch": 2.68, "learning_rate": 3.3240685096153846e-05, "loss": 0.3491, "step": 8928 }, { "epoch": 2.68, "learning_rate": 3.323880709134616e-05, "loss": 0.4003, "step": 8929 }, { "epoch": 2.68, "learning_rate": 3.323692908653846e-05, "loss": 0.3979, "step": 8930 }, { "epoch": 2.68, "learning_rate": 3.3235051081730774e-05, "loss": 0.5166, "step": 8931 }, { "epoch": 2.68, "learning_rate": 3.323317307692308e-05, "loss": 0.286, "step": 8932 }, { "epoch": 2.68, "learning_rate": 3.323129507211539e-05, "loss": 0.4124, "step": 8933 }, { "epoch": 2.68, "learning_rate": 3.3229417067307695e-05, "loss": 0.392, "step": 8934 }, { "epoch": 2.68, "learning_rate": 3.32275390625e-05, "loss": 0.4159, "step": 8935 }, { "epoch": 2.69, "learning_rate": 3.322566105769231e-05, "loss": 0.3855, "step": 8936 }, { "epoch": 2.69, "learning_rate": 3.3223783052884616e-05, "loss": 0.384, "step": 8937 }, { "epoch": 2.69, "learning_rate": 3.322190504807692e-05, "loss": 0.3558, "step": 8938 }, { "epoch": 2.69, "learning_rate": 3.322002704326923e-05, "loss": 0.399, "step": 8939 }, { "epoch": 2.69, "learning_rate": 3.321814903846154e-05, "loss": 0.3516, "step": 8940 }, { "epoch": 2.69, "learning_rate": 3.321627103365384e-05, "loss": 0.35, "step": 8941 }, { "epoch": 2.69, "learning_rate": 3.321439302884616e-05, "loss": 0.2856, "step": 8942 }, { "epoch": 2.69, "learning_rate": 3.3212515024038465e-05, "loss": 0.3671, "step": 8943 }, { "epoch": 2.69, "learning_rate": 3.321063701923077e-05, "loss": 0.3427, "step": 8944 }, { "epoch": 2.69, "learning_rate": 3.320875901442308e-05, "loss": 0.3502, "step": 8945 }, { "epoch": 2.69, "learning_rate": 3.3206881009615386e-05, "loss": 0.4897, "step": 8946 }, { "epoch": 2.69, "learning_rate": 3.32050030048077e-05, "loss": 0.3025, "step": 8947 }, { "epoch": 2.69, "learning_rate": 3.3203125e-05, "loss": 0.3635, "step": 8948 }, { "epoch": 2.69, "learning_rate": 3.320124699519231e-05, "loss": 0.2824, "step": 8949 }, { "epoch": 2.69, "learning_rate": 3.319936899038462e-05, "loss": 0.2974, "step": 8950 }, { "epoch": 2.69, "learning_rate": 3.3197490985576923e-05, "loss": 0.3934, "step": 8951 }, { "epoch": 2.69, "learning_rate": 3.319561298076923e-05, "loss": 0.3052, "step": 8952 }, { "epoch": 2.69, "learning_rate": 3.319373497596154e-05, "loss": 0.3315, "step": 8953 }, { "epoch": 2.69, "learning_rate": 3.3191856971153845e-05, "loss": 0.3469, "step": 8954 }, { "epoch": 2.69, "learning_rate": 3.3189978966346156e-05, "loss": 0.4259, "step": 8955 }, { "epoch": 2.69, "learning_rate": 3.318810096153847e-05, "loss": 0.299, "step": 8956 }, { "epoch": 2.69, "learning_rate": 3.318622295673077e-05, "loss": 0.3413, "step": 8957 }, { "epoch": 2.69, "learning_rate": 3.318434495192308e-05, "loss": 0.4397, "step": 8958 }, { "epoch": 2.69, "learning_rate": 3.318246694711539e-05, "loss": 0.4398, "step": 8959 }, { "epoch": 2.69, "learning_rate": 3.3180588942307694e-05, "loss": 0.3895, "step": 8960 }, { "epoch": 2.69, "learning_rate": 3.31787109375e-05, "loss": 0.3614, "step": 8961 }, { "epoch": 2.69, "learning_rate": 3.317683293269231e-05, "loss": 0.3447, "step": 8962 }, { "epoch": 2.69, "learning_rate": 3.3174954927884615e-05, "loss": 0.3823, "step": 8963 }, { "epoch": 2.69, "learning_rate": 3.3173076923076926e-05, "loss": 0.394, "step": 8964 }, { "epoch": 2.69, "learning_rate": 3.317119891826923e-05, "loss": 0.4209, "step": 8965 }, { "epoch": 2.69, "learning_rate": 3.3169320913461536e-05, "loss": 0.4634, "step": 8966 }, { "epoch": 2.69, "learning_rate": 3.316744290865385e-05, "loss": 0.269, "step": 8967 }, { "epoch": 2.69, "learning_rate": 3.316556490384616e-05, "loss": 0.2914, "step": 8968 }, { "epoch": 2.7, "learning_rate": 3.3163686899038464e-05, "loss": 0.3102, "step": 8969 }, { "epoch": 2.7, "learning_rate": 3.3161808894230775e-05, "loss": 0.4152, "step": 8970 }, { "epoch": 2.7, "learning_rate": 3.315993088942308e-05, "loss": 0.3606, "step": 8971 }, { "epoch": 2.7, "learning_rate": 3.3158052884615385e-05, "loss": 0.4248, "step": 8972 }, { "epoch": 2.7, "learning_rate": 3.3156174879807696e-05, "loss": 0.4111, "step": 8973 }, { "epoch": 2.7, "learning_rate": 3.3154296875e-05, "loss": 0.4186, "step": 8974 }, { "epoch": 2.7, "learning_rate": 3.3152418870192306e-05, "loss": 0.3693, "step": 8975 }, { "epoch": 2.7, "learning_rate": 3.315054086538462e-05, "loss": 0.5087, "step": 8976 }, { "epoch": 2.7, "learning_rate": 3.314866286057692e-05, "loss": 0.4608, "step": 8977 }, { "epoch": 2.7, "learning_rate": 3.3146784855769234e-05, "loss": 0.3611, "step": 8978 }, { "epoch": 2.7, "learning_rate": 3.314490685096154e-05, "loss": 0.2664, "step": 8979 }, { "epoch": 2.7, "learning_rate": 3.314302884615384e-05, "loss": 0.2699, "step": 8980 }, { "epoch": 2.7, "learning_rate": 3.3141150841346155e-05, "loss": 0.3011, "step": 8981 }, { "epoch": 2.7, "learning_rate": 3.3139272836538466e-05, "loss": 0.3819, "step": 8982 }, { "epoch": 2.7, "learning_rate": 3.313739483173077e-05, "loss": 0.5159, "step": 8983 }, { "epoch": 2.7, "learning_rate": 3.313551682692308e-05, "loss": 0.3433, "step": 8984 }, { "epoch": 2.7, "learning_rate": 3.313363882211539e-05, "loss": 0.3146, "step": 8985 }, { "epoch": 2.7, "learning_rate": 3.313176081730769e-05, "loss": 0.3395, "step": 8986 }, { "epoch": 2.7, "learning_rate": 3.3129882812500004e-05, "loss": 0.3833, "step": 8987 }, { "epoch": 2.7, "learning_rate": 3.312800480769231e-05, "loss": 0.3167, "step": 8988 }, { "epoch": 2.7, "learning_rate": 3.312612680288461e-05, "loss": 0.2334, "step": 8989 }, { "epoch": 2.7, "learning_rate": 3.3124248798076925e-05, "loss": 0.3348, "step": 8990 }, { "epoch": 2.7, "learning_rate": 3.312237079326923e-05, "loss": 0.3609, "step": 8991 }, { "epoch": 2.7, "learning_rate": 3.312049278846154e-05, "loss": 0.4086, "step": 8992 }, { "epoch": 2.7, "learning_rate": 3.3118614783653846e-05, "loss": 0.3741, "step": 8993 }, { "epoch": 2.7, "learning_rate": 3.311673677884616e-05, "loss": 0.3665, "step": 8994 }, { "epoch": 2.7, "learning_rate": 3.311485877403846e-05, "loss": 0.3592, "step": 8995 }, { "epoch": 2.7, "learning_rate": 3.3112980769230774e-05, "loss": 0.5145, "step": 8996 }, { "epoch": 2.7, "learning_rate": 3.311110276442308e-05, "loss": 0.4318, "step": 8997 }, { "epoch": 2.7, "learning_rate": 3.310922475961539e-05, "loss": 0.3276, "step": 8998 }, { "epoch": 2.7, "learning_rate": 3.3107346754807695e-05, "loss": 0.4004, "step": 8999 }, { "epoch": 2.7, "learning_rate": 3.310546875e-05, "loss": 0.3858, "step": 9000 }, { "epoch": 2.7, "learning_rate": 3.310359074519231e-05, "loss": 0.4853, "step": 9001 }, { "epoch": 2.7, "learning_rate": 3.3101712740384616e-05, "loss": 0.3223, "step": 9002 }, { "epoch": 2.71, "learning_rate": 3.309983473557692e-05, "loss": 0.3821, "step": 9003 }, { "epoch": 2.71, "learning_rate": 3.309795673076923e-05, "loss": 0.4052, "step": 9004 }, { "epoch": 2.71, "learning_rate": 3.309607872596154e-05, "loss": 0.2958, "step": 9005 }, { "epoch": 2.71, "learning_rate": 3.309420072115385e-05, "loss": 0.5667, "step": 9006 }, { "epoch": 2.71, "learning_rate": 3.309232271634616e-05, "loss": 0.2991, "step": 9007 }, { "epoch": 2.71, "learning_rate": 3.3090444711538465e-05, "loss": 0.3947, "step": 9008 }, { "epoch": 2.71, "learning_rate": 3.308856670673077e-05, "loss": 0.4227, "step": 9009 }, { "epoch": 2.71, "learning_rate": 3.308668870192308e-05, "loss": 0.3048, "step": 9010 }, { "epoch": 2.71, "learning_rate": 3.3084810697115386e-05, "loss": 0.4748, "step": 9011 }, { "epoch": 2.71, "learning_rate": 3.30829326923077e-05, "loss": 0.3381, "step": 9012 }, { "epoch": 2.71, "learning_rate": 3.30810546875e-05, "loss": 0.2553, "step": 9013 }, { "epoch": 2.71, "learning_rate": 3.307917668269231e-05, "loss": 0.2847, "step": 9014 }, { "epoch": 2.71, "learning_rate": 3.307729867788462e-05, "loss": 0.4076, "step": 9015 }, { "epoch": 2.71, "learning_rate": 3.307542067307692e-05, "loss": 0.3644, "step": 9016 }, { "epoch": 2.71, "learning_rate": 3.307354266826923e-05, "loss": 0.2222, "step": 9017 }, { "epoch": 2.71, "learning_rate": 3.307166466346154e-05, "loss": 0.2874, "step": 9018 }, { "epoch": 2.71, "learning_rate": 3.3069786658653844e-05, "loss": 0.3687, "step": 9019 }, { "epoch": 2.71, "learning_rate": 3.3067908653846156e-05, "loss": 0.3601, "step": 9020 }, { "epoch": 2.71, "learning_rate": 3.306603064903847e-05, "loss": 0.4952, "step": 9021 }, { "epoch": 2.71, "learning_rate": 3.306415264423077e-05, "loss": 0.384, "step": 9022 }, { "epoch": 2.71, "learning_rate": 3.306227463942308e-05, "loss": 0.3365, "step": 9023 }, { "epoch": 2.71, "learning_rate": 3.306039663461539e-05, "loss": 0.3703, "step": 9024 }, { "epoch": 2.71, "learning_rate": 3.3058518629807693e-05, "loss": 0.3364, "step": 9025 }, { "epoch": 2.71, "learning_rate": 3.3056640625000005e-05, "loss": 0.4455, "step": 9026 }, { "epoch": 2.71, "learning_rate": 3.305476262019231e-05, "loss": 0.4274, "step": 9027 }, { "epoch": 2.71, "learning_rate": 3.3052884615384615e-05, "loss": 0.3772, "step": 9028 }, { "epoch": 2.71, "learning_rate": 3.3051006610576926e-05, "loss": 0.5063, "step": 9029 }, { "epoch": 2.71, "learning_rate": 3.304912860576923e-05, "loss": 0.4046, "step": 9030 }, { "epoch": 2.71, "learning_rate": 3.3047250600961536e-05, "loss": 0.4254, "step": 9031 }, { "epoch": 2.71, "learning_rate": 3.304537259615385e-05, "loss": 0.3967, "step": 9032 }, { "epoch": 2.71, "learning_rate": 3.304349459134616e-05, "loss": 0.3664, "step": 9033 }, { "epoch": 2.71, "learning_rate": 3.3041616586538463e-05, "loss": 0.3088, "step": 9034 }, { "epoch": 2.71, "learning_rate": 3.3039738581730775e-05, "loss": 0.2767, "step": 9035 }, { "epoch": 2.72, "learning_rate": 3.303786057692308e-05, "loss": 0.3294, "step": 9036 }, { "epoch": 2.72, "learning_rate": 3.3035982572115385e-05, "loss": 0.3035, "step": 9037 }, { "epoch": 2.72, "learning_rate": 3.3034104567307696e-05, "loss": 0.4745, "step": 9038 }, { "epoch": 2.72, "learning_rate": 3.30322265625e-05, "loss": 0.4015, "step": 9039 }, { "epoch": 2.72, "learning_rate": 3.3030348557692306e-05, "loss": 0.458, "step": 9040 }, { "epoch": 2.72, "learning_rate": 3.302847055288462e-05, "loss": 0.1905, "step": 9041 }, { "epoch": 2.72, "learning_rate": 3.302659254807692e-05, "loss": 0.3781, "step": 9042 }, { "epoch": 2.72, "learning_rate": 3.3024714543269234e-05, "loss": 0.4282, "step": 9043 }, { "epoch": 2.72, "learning_rate": 3.302283653846154e-05, "loss": 0.3682, "step": 9044 }, { "epoch": 2.72, "learning_rate": 3.302095853365384e-05, "loss": 0.4294, "step": 9045 }, { "epoch": 2.72, "learning_rate": 3.3019080528846155e-05, "loss": 0.5778, "step": 9046 }, { "epoch": 2.72, "learning_rate": 3.3017202524038466e-05, "loss": 0.4444, "step": 9047 }, { "epoch": 2.72, "learning_rate": 3.301532451923077e-05, "loss": 0.3925, "step": 9048 }, { "epoch": 2.72, "learning_rate": 3.301344651442308e-05, "loss": 0.3597, "step": 9049 }, { "epoch": 2.72, "learning_rate": 3.301156850961539e-05, "loss": 0.3435, "step": 9050 }, { "epoch": 2.72, "learning_rate": 3.300969050480769e-05, "loss": 0.396, "step": 9051 }, { "epoch": 2.72, "learning_rate": 3.3007812500000004e-05, "loss": 0.4588, "step": 9052 }, { "epoch": 2.72, "learning_rate": 3.300593449519231e-05, "loss": 0.4866, "step": 9053 }, { "epoch": 2.72, "learning_rate": 3.300405649038461e-05, "loss": 0.3919, "step": 9054 }, { "epoch": 2.72, "learning_rate": 3.3002178485576925e-05, "loss": 0.3345, "step": 9055 }, { "epoch": 2.72, "learning_rate": 3.300030048076923e-05, "loss": 0.3454, "step": 9056 }, { "epoch": 2.72, "learning_rate": 3.299842247596154e-05, "loss": 0.4831, "step": 9057 }, { "epoch": 2.72, "learning_rate": 3.2996544471153846e-05, "loss": 0.4207, "step": 9058 }, { "epoch": 2.72, "learning_rate": 3.299466646634616e-05, "loss": 0.3583, "step": 9059 }, { "epoch": 2.72, "learning_rate": 3.299278846153846e-05, "loss": 0.337, "step": 9060 }, { "epoch": 2.72, "learning_rate": 3.2990910456730774e-05, "loss": 0.284, "step": 9061 }, { "epoch": 2.72, "learning_rate": 3.298903245192308e-05, "loss": 0.3835, "step": 9062 }, { "epoch": 2.72, "learning_rate": 3.298715444711539e-05, "loss": 0.3684, "step": 9063 }, { "epoch": 2.72, "learning_rate": 3.2985276442307695e-05, "loss": 0.4215, "step": 9064 }, { "epoch": 2.72, "learning_rate": 3.29833984375e-05, "loss": 0.3956, "step": 9065 }, { "epoch": 2.72, "learning_rate": 3.298152043269231e-05, "loss": 0.4117, "step": 9066 }, { "epoch": 2.72, "learning_rate": 3.2979642427884616e-05, "loss": 0.4767, "step": 9067 }, { "epoch": 2.72, "learning_rate": 3.297776442307692e-05, "loss": 0.3513, "step": 9068 }, { "epoch": 2.73, "learning_rate": 3.297588641826923e-05, "loss": 0.3182, "step": 9069 }, { "epoch": 2.73, "learning_rate": 3.297400841346154e-05, "loss": 0.3151, "step": 9070 }, { "epoch": 2.73, "learning_rate": 3.297213040865385e-05, "loss": 0.4242, "step": 9071 }, { "epoch": 2.73, "learning_rate": 3.297025240384616e-05, "loss": 0.267, "step": 9072 }, { "epoch": 2.73, "learning_rate": 3.2968374399038465e-05, "loss": 0.34, "step": 9073 }, { "epoch": 2.73, "learning_rate": 3.296649639423077e-05, "loss": 0.319, "step": 9074 }, { "epoch": 2.73, "learning_rate": 3.296461838942308e-05, "loss": 0.3818, "step": 9075 }, { "epoch": 2.73, "learning_rate": 3.2962740384615386e-05, "loss": 0.2963, "step": 9076 }, { "epoch": 2.73, "learning_rate": 3.29608623798077e-05, "loss": 0.3299, "step": 9077 }, { "epoch": 2.73, "learning_rate": 3.2958984375e-05, "loss": 0.3063, "step": 9078 }, { "epoch": 2.73, "learning_rate": 3.295710637019231e-05, "loss": 0.3554, "step": 9079 }, { "epoch": 2.73, "learning_rate": 3.295522836538462e-05, "loss": 0.3406, "step": 9080 }, { "epoch": 2.73, "learning_rate": 3.295335036057692e-05, "loss": 0.4423, "step": 9081 }, { "epoch": 2.73, "learning_rate": 3.295147235576923e-05, "loss": 0.358, "step": 9082 }, { "epoch": 2.73, "learning_rate": 3.294959435096154e-05, "loss": 0.3001, "step": 9083 }, { "epoch": 2.73, "learning_rate": 3.2947716346153844e-05, "loss": 0.3331, "step": 9084 }, { "epoch": 2.73, "learning_rate": 3.2945838341346156e-05, "loss": 0.4483, "step": 9085 }, { "epoch": 2.73, "learning_rate": 3.294396033653847e-05, "loss": 0.292, "step": 9086 }, { "epoch": 2.73, "learning_rate": 3.294208233173077e-05, "loss": 0.3308, "step": 9087 }, { "epoch": 2.73, "learning_rate": 3.294020432692308e-05, "loss": 0.4409, "step": 9088 }, { "epoch": 2.73, "learning_rate": 3.293832632211539e-05, "loss": 0.3289, "step": 9089 }, { "epoch": 2.73, "learning_rate": 3.293644831730769e-05, "loss": 0.5336, "step": 9090 }, { "epoch": 2.73, "learning_rate": 3.2934570312500005e-05, "loss": 0.3299, "step": 9091 }, { "epoch": 2.73, "learning_rate": 3.293269230769231e-05, "loss": 0.2963, "step": 9092 }, { "epoch": 2.73, "learning_rate": 3.2930814302884614e-05, "loss": 0.3864, "step": 9093 }, { "epoch": 2.73, "learning_rate": 3.2928936298076926e-05, "loss": 0.4455, "step": 9094 }, { "epoch": 2.73, "learning_rate": 3.292705829326923e-05, "loss": 0.3745, "step": 9095 }, { "epoch": 2.73, "learning_rate": 3.2925180288461535e-05, "loss": 0.3971, "step": 9096 }, { "epoch": 2.73, "learning_rate": 3.292330228365385e-05, "loss": 0.2999, "step": 9097 }, { "epoch": 2.73, "learning_rate": 3.292142427884616e-05, "loss": 0.3391, "step": 9098 }, { "epoch": 2.73, "learning_rate": 3.291954627403846e-05, "loss": 0.4675, "step": 9099 }, { "epoch": 2.73, "learning_rate": 3.2917668269230775e-05, "loss": 0.397, "step": 9100 }, { "epoch": 2.73, "learning_rate": 3.291579026442308e-05, "loss": 0.4136, "step": 9101 }, { "epoch": 2.73, "learning_rate": 3.2913912259615384e-05, "loss": 0.3383, "step": 9102 }, { "epoch": 2.74, "learning_rate": 3.2912034254807696e-05, "loss": 0.5208, "step": 9103 }, { "epoch": 2.74, "learning_rate": 3.291015625e-05, "loss": 0.3506, "step": 9104 }, { "epoch": 2.74, "learning_rate": 3.2908278245192306e-05, "loss": 0.4488, "step": 9105 }, { "epoch": 2.74, "learning_rate": 3.290640024038462e-05, "loss": 0.2698, "step": 9106 }, { "epoch": 2.74, "learning_rate": 3.290452223557692e-05, "loss": 0.4603, "step": 9107 }, { "epoch": 2.74, "learning_rate": 3.2902644230769233e-05, "loss": 0.3439, "step": 9108 }, { "epoch": 2.74, "learning_rate": 3.290076622596154e-05, "loss": 0.3599, "step": 9109 }, { "epoch": 2.74, "learning_rate": 3.289888822115384e-05, "loss": 0.3272, "step": 9110 }, { "epoch": 2.74, "learning_rate": 3.289701021634616e-05, "loss": 0.4622, "step": 9111 }, { "epoch": 2.74, "learning_rate": 3.2895132211538466e-05, "loss": 0.4133, "step": 9112 }, { "epoch": 2.74, "learning_rate": 3.289325420673077e-05, "loss": 0.3848, "step": 9113 }, { "epoch": 2.74, "learning_rate": 3.289137620192308e-05, "loss": 0.457, "step": 9114 }, { "epoch": 2.74, "learning_rate": 3.288949819711539e-05, "loss": 0.2383, "step": 9115 }, { "epoch": 2.74, "learning_rate": 3.288762019230769e-05, "loss": 0.3206, "step": 9116 }, { "epoch": 2.74, "learning_rate": 3.2885742187500003e-05, "loss": 0.4566, "step": 9117 }, { "epoch": 2.74, "learning_rate": 3.288386418269231e-05, "loss": 0.4001, "step": 9118 }, { "epoch": 2.74, "learning_rate": 3.288198617788461e-05, "loss": 0.3378, "step": 9119 }, { "epoch": 2.74, "learning_rate": 3.2880108173076925e-05, "loss": 0.4976, "step": 9120 }, { "epoch": 2.74, "learning_rate": 3.287823016826923e-05, "loss": 0.3601, "step": 9121 }, { "epoch": 2.74, "learning_rate": 3.287635216346154e-05, "loss": 0.4001, "step": 9122 }, { "epoch": 2.74, "learning_rate": 3.2874474158653846e-05, "loss": 0.4033, "step": 9123 }, { "epoch": 2.74, "learning_rate": 3.287259615384616e-05, "loss": 0.4049, "step": 9124 }, { "epoch": 2.74, "learning_rate": 3.287071814903846e-05, "loss": 0.3097, "step": 9125 }, { "epoch": 2.74, "learning_rate": 3.2868840144230774e-05, "loss": 0.4695, "step": 9126 }, { "epoch": 2.74, "learning_rate": 3.286696213942308e-05, "loss": 0.3611, "step": 9127 }, { "epoch": 2.74, "learning_rate": 3.286508413461539e-05, "loss": 0.3648, "step": 9128 }, { "epoch": 2.74, "learning_rate": 3.2863206129807695e-05, "loss": 0.2971, "step": 9129 }, { "epoch": 2.74, "learning_rate": 3.2861328125e-05, "loss": 0.3337, "step": 9130 }, { "epoch": 2.74, "learning_rate": 3.285945012019231e-05, "loss": 0.6253, "step": 9131 }, { "epoch": 2.74, "learning_rate": 3.2857572115384616e-05, "loss": 0.2834, "step": 9132 }, { "epoch": 2.74, "learning_rate": 3.285569411057692e-05, "loss": 0.3362, "step": 9133 }, { "epoch": 2.74, "learning_rate": 3.285381610576923e-05, "loss": 0.538, "step": 9134 }, { "epoch": 2.74, "learning_rate": 3.285193810096154e-05, "loss": 0.5058, "step": 9135 }, { "epoch": 2.75, "learning_rate": 3.285006009615385e-05, "loss": 0.3384, "step": 9136 }, { "epoch": 2.75, "learning_rate": 3.284818209134616e-05, "loss": 0.3863, "step": 9137 }, { "epoch": 2.75, "learning_rate": 3.2846304086538465e-05, "loss": 0.3865, "step": 9138 }, { "epoch": 2.75, "learning_rate": 3.284442608173077e-05, "loss": 0.3907, "step": 9139 }, { "epoch": 2.75, "learning_rate": 3.284254807692308e-05, "loss": 0.3276, "step": 9140 }, { "epoch": 2.75, "learning_rate": 3.2840670072115386e-05, "loss": 0.3446, "step": 9141 }, { "epoch": 2.75, "learning_rate": 3.28387920673077e-05, "loss": 0.402, "step": 9142 }, { "epoch": 2.75, "learning_rate": 3.28369140625e-05, "loss": 0.3697, "step": 9143 }, { "epoch": 2.75, "learning_rate": 3.283503605769231e-05, "loss": 0.4717, "step": 9144 }, { "epoch": 2.75, "learning_rate": 3.283315805288462e-05, "loss": 0.6024, "step": 9145 }, { "epoch": 2.75, "learning_rate": 3.283128004807692e-05, "loss": 0.3688, "step": 9146 }, { "epoch": 2.75, "learning_rate": 3.282940204326923e-05, "loss": 0.3242, "step": 9147 }, { "epoch": 2.75, "learning_rate": 3.282752403846154e-05, "loss": 0.4063, "step": 9148 }, { "epoch": 2.75, "learning_rate": 3.2825646033653844e-05, "loss": 0.3548, "step": 9149 }, { "epoch": 2.75, "learning_rate": 3.2823768028846156e-05, "loss": 0.2852, "step": 9150 }, { "epoch": 2.75, "learning_rate": 3.282189002403847e-05, "loss": 0.4343, "step": 9151 }, { "epoch": 2.75, "learning_rate": 3.282001201923077e-05, "loss": 0.3092, "step": 9152 }, { "epoch": 2.75, "learning_rate": 3.281813401442308e-05, "loss": 0.2987, "step": 9153 }, { "epoch": 2.75, "learning_rate": 3.281625600961539e-05, "loss": 0.3888, "step": 9154 }, { "epoch": 2.75, "learning_rate": 3.281437800480769e-05, "loss": 0.3829, "step": 9155 }, { "epoch": 2.75, "learning_rate": 3.2812500000000005e-05, "loss": 0.3961, "step": 9156 }, { "epoch": 2.75, "learning_rate": 3.281062199519231e-05, "loss": 0.3418, "step": 9157 }, { "epoch": 2.75, "learning_rate": 3.2808743990384614e-05, "loss": 0.2497, "step": 9158 }, { "epoch": 2.75, "learning_rate": 3.2806865985576926e-05, "loss": 0.3912, "step": 9159 }, { "epoch": 2.75, "learning_rate": 3.280498798076923e-05, "loss": 0.4045, "step": 9160 }, { "epoch": 2.75, "learning_rate": 3.2803109975961535e-05, "loss": 0.4827, "step": 9161 }, { "epoch": 2.75, "learning_rate": 3.280123197115385e-05, "loss": 0.4595, "step": 9162 }, { "epoch": 2.75, "learning_rate": 3.279935396634616e-05, "loss": 0.3841, "step": 9163 }, { "epoch": 2.75, "learning_rate": 3.279747596153846e-05, "loss": 0.3214, "step": 9164 }, { "epoch": 2.75, "learning_rate": 3.2795597956730775e-05, "loss": 0.349, "step": 9165 }, { "epoch": 2.75, "learning_rate": 3.279371995192308e-05, "loss": 0.503, "step": 9166 }, { "epoch": 2.75, "learning_rate": 3.2791841947115384e-05, "loss": 0.4007, "step": 9167 }, { "epoch": 2.75, "learning_rate": 3.2789963942307696e-05, "loss": 0.4221, "step": 9168 }, { "epoch": 2.76, "learning_rate": 3.27880859375e-05, "loss": 0.4759, "step": 9169 }, { "epoch": 2.76, "learning_rate": 3.278620793269231e-05, "loss": 0.438, "step": 9170 }, { "epoch": 2.76, "learning_rate": 3.278432992788462e-05, "loss": 0.4211, "step": 9171 }, { "epoch": 2.76, "learning_rate": 3.278245192307692e-05, "loss": 0.5171, "step": 9172 }, { "epoch": 2.76, "learning_rate": 3.278057391826923e-05, "loss": 0.3871, "step": 9173 }, { "epoch": 2.76, "learning_rate": 3.277869591346154e-05, "loss": 0.3965, "step": 9174 }, { "epoch": 2.76, "learning_rate": 3.277681790865384e-05, "loss": 0.4326, "step": 9175 }, { "epoch": 2.76, "learning_rate": 3.277493990384616e-05, "loss": 0.4269, "step": 9176 }, { "epoch": 2.76, "learning_rate": 3.2773061899038466e-05, "loss": 0.3883, "step": 9177 }, { "epoch": 2.76, "learning_rate": 3.277118389423077e-05, "loss": 0.478, "step": 9178 }, { "epoch": 2.76, "learning_rate": 3.276930588942308e-05, "loss": 0.4586, "step": 9179 }, { "epoch": 2.76, "learning_rate": 3.276742788461539e-05, "loss": 0.3659, "step": 9180 }, { "epoch": 2.76, "learning_rate": 3.276554987980769e-05, "loss": 0.3808, "step": 9181 }, { "epoch": 2.76, "learning_rate": 3.2763671875e-05, "loss": 0.3384, "step": 9182 }, { "epoch": 2.76, "learning_rate": 3.276179387019231e-05, "loss": 0.3662, "step": 9183 }, { "epoch": 2.76, "learning_rate": 3.275991586538461e-05, "loss": 0.3899, "step": 9184 }, { "epoch": 2.76, "learning_rate": 3.2758037860576924e-05, "loss": 0.389, "step": 9185 }, { "epoch": 2.76, "learning_rate": 3.275615985576923e-05, "loss": 0.3664, "step": 9186 }, { "epoch": 2.76, "learning_rate": 3.275428185096154e-05, "loss": 0.4023, "step": 9187 }, { "epoch": 2.76, "learning_rate": 3.2752403846153846e-05, "loss": 0.2731, "step": 9188 }, { "epoch": 2.76, "learning_rate": 3.275052584134616e-05, "loss": 0.2595, "step": 9189 }, { "epoch": 2.76, "learning_rate": 3.274864783653847e-05, "loss": 0.4297, "step": 9190 }, { "epoch": 2.76, "learning_rate": 3.2746769831730773e-05, "loss": 0.4299, "step": 9191 }, { "epoch": 2.76, "learning_rate": 3.274489182692308e-05, "loss": 0.5672, "step": 9192 }, { "epoch": 2.76, "learning_rate": 3.274301382211539e-05, "loss": 0.458, "step": 9193 }, { "epoch": 2.76, "learning_rate": 3.2741135817307694e-05, "loss": 0.3908, "step": 9194 }, { "epoch": 2.76, "learning_rate": 3.27392578125e-05, "loss": 0.3538, "step": 9195 }, { "epoch": 2.76, "learning_rate": 3.273737980769231e-05, "loss": 0.3342, "step": 9196 }, { "epoch": 2.76, "learning_rate": 3.2735501802884616e-05, "loss": 0.2801, "step": 9197 }, { "epoch": 2.76, "learning_rate": 3.273362379807692e-05, "loss": 0.3686, "step": 9198 }, { "epoch": 2.76, "learning_rate": 3.273174579326923e-05, "loss": 0.2964, "step": 9199 }, { "epoch": 2.76, "learning_rate": 3.272986778846154e-05, "loss": 0.3293, "step": 9200 }, { "epoch": 2.76, "learning_rate": 3.272798978365385e-05, "loss": 0.2517, "step": 9201 }, { "epoch": 2.77, "learning_rate": 3.272611177884616e-05, "loss": 0.6546, "step": 9202 }, { "epoch": 2.77, "learning_rate": 3.2724233774038465e-05, "loss": 0.5103, "step": 9203 }, { "epoch": 2.77, "learning_rate": 3.272235576923077e-05, "loss": 0.3767, "step": 9204 }, { "epoch": 2.77, "learning_rate": 3.272047776442308e-05, "loss": 0.3953, "step": 9205 }, { "epoch": 2.77, "learning_rate": 3.2718599759615386e-05, "loss": 0.387, "step": 9206 }, { "epoch": 2.77, "learning_rate": 3.27167217548077e-05, "loss": 0.3453, "step": 9207 }, { "epoch": 2.77, "learning_rate": 3.271484375e-05, "loss": 0.3532, "step": 9208 }, { "epoch": 2.77, "learning_rate": 3.271296574519231e-05, "loss": 0.3697, "step": 9209 }, { "epoch": 2.77, "learning_rate": 3.271108774038462e-05, "loss": 0.4959, "step": 9210 }, { "epoch": 2.77, "learning_rate": 3.270920973557692e-05, "loss": 0.3383, "step": 9211 }, { "epoch": 2.77, "learning_rate": 3.270733173076923e-05, "loss": 0.3231, "step": 9212 }, { "epoch": 2.77, "learning_rate": 3.270545372596154e-05, "loss": 0.3996, "step": 9213 }, { "epoch": 2.77, "learning_rate": 3.2703575721153844e-05, "loss": 0.3426, "step": 9214 }, { "epoch": 2.77, "learning_rate": 3.2701697716346156e-05, "loss": 0.3149, "step": 9215 }, { "epoch": 2.77, "learning_rate": 3.269981971153847e-05, "loss": 0.3247, "step": 9216 }, { "epoch": 2.77, "learning_rate": 3.269794170673077e-05, "loss": 0.3584, "step": 9217 }, { "epoch": 2.77, "learning_rate": 3.269606370192308e-05, "loss": 0.3244, "step": 9218 }, { "epoch": 2.77, "learning_rate": 3.269418569711539e-05, "loss": 0.3725, "step": 9219 }, { "epoch": 2.77, "learning_rate": 3.269230769230769e-05, "loss": 0.3619, "step": 9220 }, { "epoch": 2.77, "learning_rate": 3.2690429687500005e-05, "loss": 0.3216, "step": 9221 }, { "epoch": 2.77, "learning_rate": 3.268855168269231e-05, "loss": 0.3468, "step": 9222 }, { "epoch": 2.77, "learning_rate": 3.2686673677884614e-05, "loss": 0.2932, "step": 9223 }, { "epoch": 2.77, "learning_rate": 3.2684795673076926e-05, "loss": 0.6058, "step": 9224 }, { "epoch": 2.77, "learning_rate": 3.268291766826923e-05, "loss": 0.4099, "step": 9225 }, { "epoch": 2.77, "learning_rate": 3.2681039663461535e-05, "loss": 0.3048, "step": 9226 }, { "epoch": 2.77, "learning_rate": 3.267916165865385e-05, "loss": 0.4139, "step": 9227 }, { "epoch": 2.77, "learning_rate": 3.267728365384616e-05, "loss": 0.3499, "step": 9228 }, { "epoch": 2.77, "learning_rate": 3.267540564903846e-05, "loss": 0.4016, "step": 9229 }, { "epoch": 2.77, "learning_rate": 3.2673527644230775e-05, "loss": 0.4234, "step": 9230 }, { "epoch": 2.77, "learning_rate": 3.267164963942308e-05, "loss": 0.4139, "step": 9231 }, { "epoch": 2.77, "learning_rate": 3.2669771634615384e-05, "loss": 0.3096, "step": 9232 }, { "epoch": 2.77, "learning_rate": 3.2667893629807696e-05, "loss": 0.3938, "step": 9233 }, { "epoch": 2.77, "learning_rate": 3.2666015625e-05, "loss": 0.4411, "step": 9234 }, { "epoch": 2.77, "learning_rate": 3.266413762019231e-05, "loss": 0.3199, "step": 9235 }, { "epoch": 2.78, "learning_rate": 3.266225961538462e-05, "loss": 0.4704, "step": 9236 }, { "epoch": 2.78, "learning_rate": 3.266038161057692e-05, "loss": 0.4112, "step": 9237 }, { "epoch": 2.78, "learning_rate": 3.265850360576923e-05, "loss": 0.3454, "step": 9238 }, { "epoch": 2.78, "learning_rate": 3.265662560096154e-05, "loss": 0.3669, "step": 9239 }, { "epoch": 2.78, "learning_rate": 3.265474759615384e-05, "loss": 0.4096, "step": 9240 }, { "epoch": 2.78, "learning_rate": 3.265286959134616e-05, "loss": 0.4392, "step": 9241 }, { "epoch": 2.78, "learning_rate": 3.2650991586538466e-05, "loss": 0.4555, "step": 9242 }, { "epoch": 2.78, "learning_rate": 3.264911358173077e-05, "loss": 0.3622, "step": 9243 }, { "epoch": 2.78, "learning_rate": 3.264723557692308e-05, "loss": 0.3003, "step": 9244 }, { "epoch": 2.78, "learning_rate": 3.264535757211539e-05, "loss": 0.3732, "step": 9245 }, { "epoch": 2.78, "learning_rate": 3.264347956730769e-05, "loss": 0.3279, "step": 9246 }, { "epoch": 2.78, "learning_rate": 3.26416015625e-05, "loss": 0.2974, "step": 9247 }, { "epoch": 2.78, "learning_rate": 3.263972355769231e-05, "loss": 0.3989, "step": 9248 }, { "epoch": 2.78, "learning_rate": 3.263784555288461e-05, "loss": 0.4863, "step": 9249 }, { "epoch": 2.78, "learning_rate": 3.2635967548076924e-05, "loss": 0.4962, "step": 9250 }, { "epoch": 2.78, "learning_rate": 3.263408954326923e-05, "loss": 0.2319, "step": 9251 }, { "epoch": 2.78, "learning_rate": 3.263221153846154e-05, "loss": 0.4422, "step": 9252 }, { "epoch": 2.78, "learning_rate": 3.2630333533653845e-05, "loss": 0.4095, "step": 9253 }, { "epoch": 2.78, "learning_rate": 3.262845552884616e-05, "loss": 0.3616, "step": 9254 }, { "epoch": 2.78, "learning_rate": 3.262657752403847e-05, "loss": 0.4198, "step": 9255 }, { "epoch": 2.78, "learning_rate": 3.262469951923077e-05, "loss": 0.3068, "step": 9256 }, { "epoch": 2.78, "learning_rate": 3.262282151442308e-05, "loss": 0.3348, "step": 9257 }, { "epoch": 2.78, "learning_rate": 3.262094350961539e-05, "loss": 0.5105, "step": 9258 }, { "epoch": 2.78, "learning_rate": 3.2619065504807694e-05, "loss": 0.3378, "step": 9259 }, { "epoch": 2.78, "learning_rate": 3.26171875e-05, "loss": 0.3589, "step": 9260 }, { "epoch": 2.78, "learning_rate": 3.261530949519231e-05, "loss": 0.376, "step": 9261 }, { "epoch": 2.78, "learning_rate": 3.2613431490384615e-05, "loss": 0.4849, "step": 9262 }, { "epoch": 2.78, "learning_rate": 3.261155348557692e-05, "loss": 0.4119, "step": 9263 }, { "epoch": 2.78, "learning_rate": 3.260967548076923e-05, "loss": 0.5202, "step": 9264 }, { "epoch": 2.78, "learning_rate": 3.2607797475961537e-05, "loss": 0.3146, "step": 9265 }, { "epoch": 2.78, "learning_rate": 3.260591947115385e-05, "loss": 0.4447, "step": 9266 }, { "epoch": 2.78, "learning_rate": 3.260404146634616e-05, "loss": 0.5272, "step": 9267 }, { "epoch": 2.78, "learning_rate": 3.2602163461538464e-05, "loss": 0.3552, "step": 9268 }, { "epoch": 2.79, "learning_rate": 3.260028545673077e-05, "loss": 0.2849, "step": 9269 }, { "epoch": 2.79, "learning_rate": 3.259840745192308e-05, "loss": 0.343, "step": 9270 }, { "epoch": 2.79, "learning_rate": 3.2596529447115386e-05, "loss": 0.4285, "step": 9271 }, { "epoch": 2.79, "learning_rate": 3.25946514423077e-05, "loss": 0.3486, "step": 9272 }, { "epoch": 2.79, "learning_rate": 3.25927734375e-05, "loss": 0.4479, "step": 9273 }, { "epoch": 2.79, "learning_rate": 3.259089543269231e-05, "loss": 0.4744, "step": 9274 }, { "epoch": 2.79, "learning_rate": 3.258901742788462e-05, "loss": 0.3152, "step": 9275 }, { "epoch": 2.79, "learning_rate": 3.258713942307692e-05, "loss": 0.3114, "step": 9276 }, { "epoch": 2.79, "learning_rate": 3.258526141826923e-05, "loss": 0.4316, "step": 9277 }, { "epoch": 2.79, "learning_rate": 3.258338341346154e-05, "loss": 0.3163, "step": 9278 }, { "epoch": 2.79, "learning_rate": 3.2581505408653844e-05, "loss": 0.4494, "step": 9279 }, { "epoch": 2.79, "learning_rate": 3.2579627403846156e-05, "loss": 0.3011, "step": 9280 }, { "epoch": 2.79, "learning_rate": 3.257774939903847e-05, "loss": 0.4885, "step": 9281 }, { "epoch": 2.79, "learning_rate": 3.257587139423077e-05, "loss": 0.3322, "step": 9282 }, { "epoch": 2.79, "learning_rate": 3.257399338942308e-05, "loss": 0.2906, "step": 9283 }, { "epoch": 2.79, "learning_rate": 3.257211538461539e-05, "loss": 0.6015, "step": 9284 }, { "epoch": 2.79, "learning_rate": 3.257023737980769e-05, "loss": 0.3719, "step": 9285 }, { "epoch": 2.79, "learning_rate": 3.2568359375000005e-05, "loss": 0.468, "step": 9286 }, { "epoch": 2.79, "learning_rate": 3.256648137019231e-05, "loss": 0.4174, "step": 9287 }, { "epoch": 2.79, "learning_rate": 3.2564603365384614e-05, "loss": 0.2552, "step": 9288 }, { "epoch": 2.79, "learning_rate": 3.2562725360576926e-05, "loss": 0.459, "step": 9289 }, { "epoch": 2.79, "learning_rate": 3.256084735576923e-05, "loss": 0.4776, "step": 9290 }, { "epoch": 2.79, "learning_rate": 3.2558969350961535e-05, "loss": 0.4205, "step": 9291 }, { "epoch": 2.79, "learning_rate": 3.255709134615385e-05, "loss": 0.2849, "step": 9292 }, { "epoch": 2.79, "learning_rate": 3.255521334134616e-05, "loss": 0.4518, "step": 9293 }, { "epoch": 2.79, "learning_rate": 3.255333533653846e-05, "loss": 0.4494, "step": 9294 }, { "epoch": 2.79, "learning_rate": 3.2551457331730775e-05, "loss": 0.3106, "step": 9295 }, { "epoch": 2.79, "learning_rate": 3.254957932692308e-05, "loss": 0.4449, "step": 9296 }, { "epoch": 2.79, "learning_rate": 3.2547701322115384e-05, "loss": 0.4023, "step": 9297 }, { "epoch": 2.79, "learning_rate": 3.2545823317307696e-05, "loss": 0.4187, "step": 9298 }, { "epoch": 2.79, "learning_rate": 3.25439453125e-05, "loss": 0.424, "step": 9299 }, { "epoch": 2.79, "learning_rate": 3.254206730769231e-05, "loss": 0.2888, "step": 9300 }, { "epoch": 2.79, "learning_rate": 3.254018930288462e-05, "loss": 0.411, "step": 9301 }, { "epoch": 2.8, "learning_rate": 3.253831129807692e-05, "loss": 0.406, "step": 9302 }, { "epoch": 2.8, "learning_rate": 3.253643329326923e-05, "loss": 0.3701, "step": 9303 }, { "epoch": 2.8, "learning_rate": 3.253455528846154e-05, "loss": 0.4632, "step": 9304 }, { "epoch": 2.8, "learning_rate": 3.253267728365384e-05, "loss": 0.3327, "step": 9305 }, { "epoch": 2.8, "learning_rate": 3.253079927884616e-05, "loss": 0.3772, "step": 9306 }, { "epoch": 2.8, "learning_rate": 3.2528921274038466e-05, "loss": 0.4744, "step": 9307 }, { "epoch": 2.8, "learning_rate": 3.252704326923077e-05, "loss": 0.4513, "step": 9308 }, { "epoch": 2.8, "learning_rate": 3.252516526442308e-05, "loss": 0.4225, "step": 9309 }, { "epoch": 2.8, "learning_rate": 3.252328725961539e-05, "loss": 0.447, "step": 9310 }, { "epoch": 2.8, "learning_rate": 3.252140925480769e-05, "loss": 0.4043, "step": 9311 }, { "epoch": 2.8, "learning_rate": 3.251953125e-05, "loss": 0.4686, "step": 9312 }, { "epoch": 2.8, "learning_rate": 3.251765324519231e-05, "loss": 0.3816, "step": 9313 }, { "epoch": 2.8, "learning_rate": 3.251577524038462e-05, "loss": 0.3625, "step": 9314 }, { "epoch": 2.8, "learning_rate": 3.2513897235576924e-05, "loss": 0.3585, "step": 9315 }, { "epoch": 2.8, "learning_rate": 3.251201923076923e-05, "loss": 0.3667, "step": 9316 }, { "epoch": 2.8, "learning_rate": 3.251014122596154e-05, "loss": 0.3116, "step": 9317 }, { "epoch": 2.8, "learning_rate": 3.2508263221153845e-05, "loss": 0.5938, "step": 9318 }, { "epoch": 2.8, "learning_rate": 3.250638521634616e-05, "loss": 0.4891, "step": 9319 }, { "epoch": 2.8, "learning_rate": 3.250450721153847e-05, "loss": 0.4461, "step": 9320 }, { "epoch": 2.8, "learning_rate": 3.250262920673077e-05, "loss": 0.3868, "step": 9321 }, { "epoch": 2.8, "learning_rate": 3.250075120192308e-05, "loss": 0.3906, "step": 9322 }, { "epoch": 2.8, "learning_rate": 3.249887319711539e-05, "loss": 0.3231, "step": 9323 }, { "epoch": 2.8, "learning_rate": 3.2496995192307694e-05, "loss": 0.3083, "step": 9324 }, { "epoch": 2.8, "learning_rate": 3.24951171875e-05, "loss": 0.2699, "step": 9325 }, { "epoch": 2.8, "learning_rate": 3.249323918269231e-05, "loss": 0.3326, "step": 9326 }, { "epoch": 2.8, "learning_rate": 3.2491361177884615e-05, "loss": 0.2874, "step": 9327 }, { "epoch": 2.8, "learning_rate": 3.248948317307692e-05, "loss": 0.326, "step": 9328 }, { "epoch": 2.8, "learning_rate": 3.248760516826923e-05, "loss": 0.3305, "step": 9329 }, { "epoch": 2.8, "learning_rate": 3.2485727163461536e-05, "loss": 0.429, "step": 9330 }, { "epoch": 2.8, "learning_rate": 3.248384915865385e-05, "loss": 0.3549, "step": 9331 }, { "epoch": 2.8, "learning_rate": 3.248197115384616e-05, "loss": 0.4399, "step": 9332 }, { "epoch": 2.8, "learning_rate": 3.2480093149038464e-05, "loss": 0.4083, "step": 9333 }, { "epoch": 2.8, "learning_rate": 3.2478215144230776e-05, "loss": 0.347, "step": 9334 }, { "epoch": 2.8, "learning_rate": 3.247633713942308e-05, "loss": 0.3782, "step": 9335 }, { "epoch": 2.81, "learning_rate": 3.2474459134615385e-05, "loss": 0.3673, "step": 9336 }, { "epoch": 2.81, "learning_rate": 3.24725811298077e-05, "loss": 0.3493, "step": 9337 }, { "epoch": 2.81, "learning_rate": 3.2470703125e-05, "loss": 0.3961, "step": 9338 }, { "epoch": 2.81, "learning_rate": 3.2468825120192307e-05, "loss": 0.2903, "step": 9339 }, { "epoch": 2.81, "learning_rate": 3.246694711538462e-05, "loss": 0.4452, "step": 9340 }, { "epoch": 2.81, "learning_rate": 3.246506911057692e-05, "loss": 0.4396, "step": 9341 }, { "epoch": 2.81, "learning_rate": 3.246319110576923e-05, "loss": 0.4639, "step": 9342 }, { "epoch": 2.81, "learning_rate": 3.246131310096154e-05, "loss": 0.2981, "step": 9343 }, { "epoch": 2.81, "learning_rate": 3.2459435096153844e-05, "loss": 0.3472, "step": 9344 }, { "epoch": 2.81, "learning_rate": 3.2457557091346155e-05, "loss": 0.4785, "step": 9345 }, { "epoch": 2.81, "learning_rate": 3.245567908653847e-05, "loss": 0.3716, "step": 9346 }, { "epoch": 2.81, "learning_rate": 3.245380108173077e-05, "loss": 0.4168, "step": 9347 }, { "epoch": 2.81, "learning_rate": 3.2451923076923077e-05, "loss": 0.3538, "step": 9348 }, { "epoch": 2.81, "learning_rate": 3.245004507211539e-05, "loss": 0.4083, "step": 9349 }, { "epoch": 2.81, "learning_rate": 3.244816706730769e-05, "loss": 0.3628, "step": 9350 }, { "epoch": 2.81, "learning_rate": 3.2446289062500004e-05, "loss": 0.3415, "step": 9351 }, { "epoch": 2.81, "learning_rate": 3.244441105769231e-05, "loss": 0.2377, "step": 9352 }, { "epoch": 2.81, "learning_rate": 3.2442533052884614e-05, "loss": 0.2617, "step": 9353 }, { "epoch": 2.81, "learning_rate": 3.2440655048076926e-05, "loss": 0.4895, "step": 9354 }, { "epoch": 2.81, "learning_rate": 3.243877704326923e-05, "loss": 0.3445, "step": 9355 }, { "epoch": 2.81, "learning_rate": 3.2436899038461535e-05, "loss": 0.3418, "step": 9356 }, { "epoch": 2.81, "learning_rate": 3.2435021033653847e-05, "loss": 0.2596, "step": 9357 }, { "epoch": 2.81, "learning_rate": 3.243314302884616e-05, "loss": 0.289, "step": 9358 }, { "epoch": 2.81, "learning_rate": 3.243126502403846e-05, "loss": 0.4071, "step": 9359 }, { "epoch": 2.81, "learning_rate": 3.2429387019230774e-05, "loss": 0.3782, "step": 9360 }, { "epoch": 2.81, "learning_rate": 3.242750901442308e-05, "loss": 0.3153, "step": 9361 }, { "epoch": 2.81, "learning_rate": 3.2425631009615384e-05, "loss": 0.4193, "step": 9362 }, { "epoch": 2.81, "learning_rate": 3.2423753004807696e-05, "loss": 0.501, "step": 9363 }, { "epoch": 2.81, "learning_rate": 3.2421875e-05, "loss": 0.3085, "step": 9364 }, { "epoch": 2.81, "learning_rate": 3.241999699519231e-05, "loss": 0.3826, "step": 9365 }, { "epoch": 2.81, "learning_rate": 3.241811899038462e-05, "loss": 0.3794, "step": 9366 }, { "epoch": 2.81, "learning_rate": 3.241624098557692e-05, "loss": 0.2269, "step": 9367 }, { "epoch": 2.81, "learning_rate": 3.241436298076923e-05, "loss": 0.4123, "step": 9368 }, { "epoch": 2.82, "learning_rate": 3.241248497596154e-05, "loss": 0.4001, "step": 9369 }, { "epoch": 2.82, "learning_rate": 3.241060697115384e-05, "loss": 0.3785, "step": 9370 }, { "epoch": 2.82, "learning_rate": 3.240872896634616e-05, "loss": 0.3802, "step": 9371 }, { "epoch": 2.82, "learning_rate": 3.2406850961538466e-05, "loss": 0.3088, "step": 9372 }, { "epoch": 2.82, "learning_rate": 3.240497295673077e-05, "loss": 0.4867, "step": 9373 }, { "epoch": 2.82, "learning_rate": 3.240309495192308e-05, "loss": 0.3364, "step": 9374 }, { "epoch": 2.82, "learning_rate": 3.240121694711539e-05, "loss": 0.4293, "step": 9375 }, { "epoch": 2.82, "learning_rate": 3.239933894230769e-05, "loss": 0.3389, "step": 9376 }, { "epoch": 2.82, "learning_rate": 3.23974609375e-05, "loss": 0.3086, "step": 9377 }, { "epoch": 2.82, "learning_rate": 3.239558293269231e-05, "loss": 0.55, "step": 9378 }, { "epoch": 2.82, "learning_rate": 3.239370492788462e-05, "loss": 0.3584, "step": 9379 }, { "epoch": 2.82, "learning_rate": 3.2391826923076924e-05, "loss": 0.354, "step": 9380 }, { "epoch": 2.82, "learning_rate": 3.238994891826923e-05, "loss": 0.289, "step": 9381 }, { "epoch": 2.82, "learning_rate": 3.238807091346154e-05, "loss": 0.4973, "step": 9382 }, { "epoch": 2.82, "learning_rate": 3.2386192908653845e-05, "loss": 0.3647, "step": 9383 }, { "epoch": 2.82, "learning_rate": 3.238431490384616e-05, "loss": 0.3525, "step": 9384 }, { "epoch": 2.82, "learning_rate": 3.238243689903847e-05, "loss": 0.3731, "step": 9385 }, { "epoch": 2.82, "learning_rate": 3.238055889423077e-05, "loss": 0.5475, "step": 9386 }, { "epoch": 2.82, "learning_rate": 3.237868088942308e-05, "loss": 0.4789, "step": 9387 }, { "epoch": 2.82, "learning_rate": 3.237680288461539e-05, "loss": 0.4655, "step": 9388 }, { "epoch": 2.82, "learning_rate": 3.2374924879807694e-05, "loss": 0.3584, "step": 9389 }, { "epoch": 2.82, "learning_rate": 3.2373046875e-05, "loss": 0.4672, "step": 9390 }, { "epoch": 2.82, "learning_rate": 3.237116887019231e-05, "loss": 0.2789, "step": 9391 }, { "epoch": 2.82, "learning_rate": 3.2369290865384615e-05, "loss": 0.3639, "step": 9392 }, { "epoch": 2.82, "learning_rate": 3.236741286057692e-05, "loss": 0.3905, "step": 9393 }, { "epoch": 2.82, "learning_rate": 3.236553485576923e-05, "loss": 0.4128, "step": 9394 }, { "epoch": 2.82, "learning_rate": 3.2363656850961536e-05, "loss": 0.4805, "step": 9395 }, { "epoch": 2.82, "learning_rate": 3.236177884615385e-05, "loss": 0.369, "step": 9396 }, { "epoch": 2.82, "learning_rate": 3.235990084134616e-05, "loss": 0.3802, "step": 9397 }, { "epoch": 2.82, "learning_rate": 3.2358022836538464e-05, "loss": 0.3796, "step": 9398 }, { "epoch": 2.82, "learning_rate": 3.2356144831730776e-05, "loss": 0.3909, "step": 9399 }, { "epoch": 2.82, "learning_rate": 3.235426682692308e-05, "loss": 0.4341, "step": 9400 }, { "epoch": 2.82, "learning_rate": 3.2352388822115385e-05, "loss": 0.3435, "step": 9401 }, { "epoch": 2.83, "learning_rate": 3.23505108173077e-05, "loss": 0.3571, "step": 9402 }, { "epoch": 2.83, "learning_rate": 3.23486328125e-05, "loss": 0.2981, "step": 9403 }, { "epoch": 2.83, "learning_rate": 3.2346754807692306e-05, "loss": 0.3956, "step": 9404 }, { "epoch": 2.83, "learning_rate": 3.234487680288462e-05, "loss": 0.3501, "step": 9405 }, { "epoch": 2.83, "learning_rate": 3.234299879807692e-05, "loss": 0.3539, "step": 9406 }, { "epoch": 2.83, "learning_rate": 3.234112079326923e-05, "loss": 0.4083, "step": 9407 }, { "epoch": 2.83, "learning_rate": 3.233924278846154e-05, "loss": 0.3027, "step": 9408 }, { "epoch": 2.83, "learning_rate": 3.2337364783653844e-05, "loss": 0.3672, "step": 9409 }, { "epoch": 2.83, "learning_rate": 3.2335486778846155e-05, "loss": 0.3099, "step": 9410 }, { "epoch": 2.83, "learning_rate": 3.233360877403847e-05, "loss": 0.3399, "step": 9411 }, { "epoch": 2.83, "learning_rate": 3.233173076923077e-05, "loss": 0.382, "step": 9412 }, { "epoch": 2.83, "learning_rate": 3.2329852764423076e-05, "loss": 0.368, "step": 9413 }, { "epoch": 2.83, "learning_rate": 3.232797475961539e-05, "loss": 0.2847, "step": 9414 }, { "epoch": 2.83, "learning_rate": 3.232609675480769e-05, "loss": 0.3747, "step": 9415 }, { "epoch": 2.83, "learning_rate": 3.2324218750000004e-05, "loss": 0.4362, "step": 9416 }, { "epoch": 2.83, "learning_rate": 3.232234074519231e-05, "loss": 0.4872, "step": 9417 }, { "epoch": 2.83, "learning_rate": 3.2320462740384614e-05, "loss": 0.3557, "step": 9418 }, { "epoch": 2.83, "learning_rate": 3.2318584735576925e-05, "loss": 0.3155, "step": 9419 }, { "epoch": 2.83, "learning_rate": 3.231670673076923e-05, "loss": 0.3055, "step": 9420 }, { "epoch": 2.83, "learning_rate": 3.2314828725961535e-05, "loss": 0.3199, "step": 9421 }, { "epoch": 2.83, "learning_rate": 3.2312950721153846e-05, "loss": 0.3767, "step": 9422 }, { "epoch": 2.83, "learning_rate": 3.231107271634616e-05, "loss": 0.363, "step": 9423 }, { "epoch": 2.83, "learning_rate": 3.230919471153846e-05, "loss": 0.4249, "step": 9424 }, { "epoch": 2.83, "learning_rate": 3.2307316706730774e-05, "loss": 0.2317, "step": 9425 }, { "epoch": 2.83, "learning_rate": 3.230543870192308e-05, "loss": 0.4155, "step": 9426 }, { "epoch": 2.83, "learning_rate": 3.2303560697115384e-05, "loss": 0.2367, "step": 9427 }, { "epoch": 2.83, "learning_rate": 3.2301682692307695e-05, "loss": 0.2925, "step": 9428 }, { "epoch": 2.83, "learning_rate": 3.22998046875e-05, "loss": 0.3289, "step": 9429 }, { "epoch": 2.83, "learning_rate": 3.229792668269231e-05, "loss": 0.3281, "step": 9430 }, { "epoch": 2.83, "learning_rate": 3.2296048677884617e-05, "loss": 0.4157, "step": 9431 }, { "epoch": 2.83, "learning_rate": 3.229417067307692e-05, "loss": 0.4383, "step": 9432 }, { "epoch": 2.83, "learning_rate": 3.229229266826923e-05, "loss": 0.3465, "step": 9433 }, { "epoch": 2.83, "learning_rate": 3.229041466346154e-05, "loss": 0.3247, "step": 9434 }, { "epoch": 2.84, "learning_rate": 3.228853665865384e-05, "loss": 0.3159, "step": 9435 }, { "epoch": 2.84, "learning_rate": 3.228665865384616e-05, "loss": 0.3666, "step": 9436 }, { "epoch": 2.84, "learning_rate": 3.2284780649038466e-05, "loss": 0.5974, "step": 9437 }, { "epoch": 2.84, "learning_rate": 3.228290264423077e-05, "loss": 0.3333, "step": 9438 }, { "epoch": 2.84, "learning_rate": 3.228102463942308e-05, "loss": 0.3404, "step": 9439 }, { "epoch": 2.84, "learning_rate": 3.2279146634615387e-05, "loss": 0.4009, "step": 9440 }, { "epoch": 2.84, "learning_rate": 3.227726862980769e-05, "loss": 0.3809, "step": 9441 }, { "epoch": 2.84, "learning_rate": 3.2275390625e-05, "loss": 0.3024, "step": 9442 }, { "epoch": 2.84, "learning_rate": 3.227351262019231e-05, "loss": 0.4175, "step": 9443 }, { "epoch": 2.84, "learning_rate": 3.227163461538462e-05, "loss": 0.4003, "step": 9444 }, { "epoch": 2.84, "learning_rate": 3.2269756610576924e-05, "loss": 0.5497, "step": 9445 }, { "epoch": 2.84, "learning_rate": 3.226787860576923e-05, "loss": 0.5174, "step": 9446 }, { "epoch": 2.84, "learning_rate": 3.226600060096154e-05, "loss": 0.4798, "step": 9447 }, { "epoch": 2.84, "learning_rate": 3.2264122596153845e-05, "loss": 0.3029, "step": 9448 }, { "epoch": 2.84, "learning_rate": 3.226224459134616e-05, "loss": 0.3418, "step": 9449 }, { "epoch": 2.84, "learning_rate": 3.226036658653847e-05, "loss": 0.3535, "step": 9450 }, { "epoch": 2.84, "learning_rate": 3.225848858173077e-05, "loss": 0.4519, "step": 9451 }, { "epoch": 2.84, "learning_rate": 3.225661057692308e-05, "loss": 0.3952, "step": 9452 }, { "epoch": 2.84, "learning_rate": 3.225473257211539e-05, "loss": 0.3994, "step": 9453 }, { "epoch": 2.84, "learning_rate": 3.2252854567307694e-05, "loss": 0.3221, "step": 9454 }, { "epoch": 2.84, "learning_rate": 3.22509765625e-05, "loss": 0.299, "step": 9455 }, { "epoch": 2.84, "learning_rate": 3.224909855769231e-05, "loss": 0.3078, "step": 9456 }, { "epoch": 2.84, "learning_rate": 3.2247220552884615e-05, "loss": 0.5307, "step": 9457 }, { "epoch": 2.84, "learning_rate": 3.224534254807693e-05, "loss": 0.3506, "step": 9458 }, { "epoch": 2.84, "learning_rate": 3.224346454326923e-05, "loss": 0.2433, "step": 9459 }, { "epoch": 2.84, "learning_rate": 3.2241586538461536e-05, "loss": 0.4368, "step": 9460 }, { "epoch": 2.84, "learning_rate": 3.223970853365385e-05, "loss": 0.2836, "step": 9461 }, { "epoch": 2.84, "learning_rate": 3.223783052884616e-05, "loss": 0.2932, "step": 9462 }, { "epoch": 2.84, "learning_rate": 3.2235952524038464e-05, "loss": 0.4798, "step": 9463 }, { "epoch": 2.84, "learning_rate": 3.2234074519230776e-05, "loss": 0.3892, "step": 9464 }, { "epoch": 2.84, "learning_rate": 3.223219651442308e-05, "loss": 0.382, "step": 9465 }, { "epoch": 2.84, "learning_rate": 3.2230318509615385e-05, "loss": 0.4506, "step": 9466 }, { "epoch": 2.84, "learning_rate": 3.22284405048077e-05, "loss": 0.3242, "step": 9467 }, { "epoch": 2.84, "learning_rate": 3.22265625e-05, "loss": 0.3572, "step": 9468 }, { "epoch": 2.85, "learning_rate": 3.2224684495192306e-05, "loss": 0.2416, "step": 9469 }, { "epoch": 2.85, "learning_rate": 3.222280649038462e-05, "loss": 0.4015, "step": 9470 }, { "epoch": 2.85, "learning_rate": 3.222092848557692e-05, "loss": 0.4453, "step": 9471 }, { "epoch": 2.85, "learning_rate": 3.221905048076923e-05, "loss": 0.3936, "step": 9472 }, { "epoch": 2.85, "learning_rate": 3.221717247596154e-05, "loss": 0.3226, "step": 9473 }, { "epoch": 2.85, "learning_rate": 3.2215294471153844e-05, "loss": 0.4824, "step": 9474 }, { "epoch": 2.85, "learning_rate": 3.2213416466346155e-05, "loss": 0.4526, "step": 9475 }, { "epoch": 2.85, "learning_rate": 3.221153846153847e-05, "loss": 0.3845, "step": 9476 }, { "epoch": 2.85, "learning_rate": 3.220966045673077e-05, "loss": 0.4637, "step": 9477 }, { "epoch": 2.85, "learning_rate": 3.220778245192308e-05, "loss": 0.3826, "step": 9478 }, { "epoch": 2.85, "learning_rate": 3.220590444711539e-05, "loss": 0.2812, "step": 9479 }, { "epoch": 2.85, "learning_rate": 3.220402644230769e-05, "loss": 0.3078, "step": 9480 }, { "epoch": 2.85, "learning_rate": 3.2202148437500004e-05, "loss": 0.3045, "step": 9481 }, { "epoch": 2.85, "learning_rate": 3.220027043269231e-05, "loss": 0.3052, "step": 9482 }, { "epoch": 2.85, "learning_rate": 3.2198392427884614e-05, "loss": 0.2827, "step": 9483 }, { "epoch": 2.85, "learning_rate": 3.2196514423076925e-05, "loss": 0.4333, "step": 9484 }, { "epoch": 2.85, "learning_rate": 3.219463641826923e-05, "loss": 0.4427, "step": 9485 }, { "epoch": 2.85, "learning_rate": 3.2192758413461535e-05, "loss": 0.3152, "step": 9486 }, { "epoch": 2.85, "learning_rate": 3.2190880408653846e-05, "loss": 0.3271, "step": 9487 }, { "epoch": 2.85, "learning_rate": 3.218900240384616e-05, "loss": 0.438, "step": 9488 }, { "epoch": 2.85, "learning_rate": 3.218712439903846e-05, "loss": 0.2753, "step": 9489 }, { "epoch": 2.85, "learning_rate": 3.2185246394230774e-05, "loss": 0.485, "step": 9490 }, { "epoch": 2.85, "learning_rate": 3.218336838942308e-05, "loss": 0.343, "step": 9491 }, { "epoch": 2.85, "learning_rate": 3.2181490384615384e-05, "loss": 0.3967, "step": 9492 }, { "epoch": 2.85, "learning_rate": 3.2179612379807695e-05, "loss": 0.4739, "step": 9493 }, { "epoch": 2.85, "learning_rate": 3.2177734375e-05, "loss": 0.3789, "step": 9494 }, { "epoch": 2.85, "learning_rate": 3.217585637019231e-05, "loss": 0.4058, "step": 9495 }, { "epoch": 2.85, "learning_rate": 3.2173978365384616e-05, "loss": 0.46, "step": 9496 }, { "epoch": 2.85, "learning_rate": 3.217210036057692e-05, "loss": 0.3316, "step": 9497 }, { "epoch": 2.85, "learning_rate": 3.217022235576923e-05, "loss": 0.4032, "step": 9498 }, { "epoch": 2.85, "learning_rate": 3.216834435096154e-05, "loss": 0.349, "step": 9499 }, { "epoch": 2.85, "learning_rate": 3.216646634615384e-05, "loss": 0.3429, "step": 9500 }, { "epoch": 2.85, "learning_rate": 3.216458834134616e-05, "loss": 0.3213, "step": 9501 }, { "epoch": 2.86, "learning_rate": 3.2162710336538465e-05, "loss": 0.5232, "step": 9502 }, { "epoch": 2.86, "learning_rate": 3.216083233173077e-05, "loss": 0.5339, "step": 9503 }, { "epoch": 2.86, "learning_rate": 3.215895432692308e-05, "loss": 0.3916, "step": 9504 }, { "epoch": 2.86, "learning_rate": 3.2157076322115386e-05, "loss": 0.4049, "step": 9505 }, { "epoch": 2.86, "learning_rate": 3.215519831730769e-05, "loss": 0.29, "step": 9506 }, { "epoch": 2.86, "learning_rate": 3.21533203125e-05, "loss": 0.2303, "step": 9507 }, { "epoch": 2.86, "learning_rate": 3.215144230769231e-05, "loss": 0.5587, "step": 9508 }, { "epoch": 2.86, "learning_rate": 3.214956430288462e-05, "loss": 0.3133, "step": 9509 }, { "epoch": 2.86, "learning_rate": 3.2147686298076924e-05, "loss": 0.4646, "step": 9510 }, { "epoch": 2.86, "learning_rate": 3.214580829326923e-05, "loss": 0.2874, "step": 9511 }, { "epoch": 2.86, "learning_rate": 3.214393028846154e-05, "loss": 0.3419, "step": 9512 }, { "epoch": 2.86, "learning_rate": 3.2142052283653845e-05, "loss": 0.3022, "step": 9513 }, { "epoch": 2.86, "learning_rate": 3.2140174278846157e-05, "loss": 0.4846, "step": 9514 }, { "epoch": 2.86, "learning_rate": 3.213829627403847e-05, "loss": 0.3402, "step": 9515 }, { "epoch": 2.86, "learning_rate": 3.213641826923077e-05, "loss": 0.474, "step": 9516 }, { "epoch": 2.86, "learning_rate": 3.213454026442308e-05, "loss": 0.3025, "step": 9517 }, { "epoch": 2.86, "learning_rate": 3.213266225961539e-05, "loss": 0.3758, "step": 9518 }, { "epoch": 2.86, "learning_rate": 3.2130784254807694e-05, "loss": 0.3315, "step": 9519 }, { "epoch": 2.86, "learning_rate": 3.212890625e-05, "loss": 0.4771, "step": 9520 }, { "epoch": 2.86, "learning_rate": 3.212702824519231e-05, "loss": 0.311, "step": 9521 }, { "epoch": 2.86, "learning_rate": 3.2125150240384615e-05, "loss": 0.5045, "step": 9522 }, { "epoch": 2.86, "learning_rate": 3.2123272235576927e-05, "loss": 0.3087, "step": 9523 }, { "epoch": 2.86, "learning_rate": 3.212139423076923e-05, "loss": 0.361, "step": 9524 }, { "epoch": 2.86, "learning_rate": 3.2119516225961536e-05, "loss": 0.3331, "step": 9525 }, { "epoch": 2.86, "learning_rate": 3.211763822115385e-05, "loss": 0.4879, "step": 9526 }, { "epoch": 2.86, "learning_rate": 3.211576021634616e-05, "loss": 0.4494, "step": 9527 }, { "epoch": 2.86, "learning_rate": 3.2113882211538464e-05, "loss": 0.3603, "step": 9528 }, { "epoch": 2.86, "learning_rate": 3.2112004206730776e-05, "loss": 0.5075, "step": 9529 }, { "epoch": 2.86, "learning_rate": 3.211012620192308e-05, "loss": 0.3489, "step": 9530 }, { "epoch": 2.86, "learning_rate": 3.2108248197115385e-05, "loss": 0.4017, "step": 9531 }, { "epoch": 2.86, "learning_rate": 3.21063701923077e-05, "loss": 0.3794, "step": 9532 }, { "epoch": 2.86, "learning_rate": 3.21044921875e-05, "loss": 0.4514, "step": 9533 }, { "epoch": 2.86, "learning_rate": 3.2102614182692306e-05, "loss": 0.3708, "step": 9534 }, { "epoch": 2.87, "learning_rate": 3.210073617788462e-05, "loss": 0.3812, "step": 9535 }, { "epoch": 2.87, "learning_rate": 3.209885817307692e-05, "loss": 0.4524, "step": 9536 }, { "epoch": 2.87, "learning_rate": 3.209698016826923e-05, "loss": 0.3342, "step": 9537 }, { "epoch": 2.87, "learning_rate": 3.209510216346154e-05, "loss": 0.2663, "step": 9538 }, { "epoch": 2.87, "learning_rate": 3.2093224158653844e-05, "loss": 0.444, "step": 9539 }, { "epoch": 2.87, "learning_rate": 3.2091346153846155e-05, "loss": 0.3607, "step": 9540 }, { "epoch": 2.87, "learning_rate": 3.208946814903847e-05, "loss": 0.3759, "step": 9541 }, { "epoch": 2.87, "learning_rate": 3.208759014423077e-05, "loss": 0.4489, "step": 9542 }, { "epoch": 2.87, "learning_rate": 3.208571213942308e-05, "loss": 0.3862, "step": 9543 }, { "epoch": 2.87, "learning_rate": 3.208383413461539e-05, "loss": 0.4557, "step": 9544 }, { "epoch": 2.87, "learning_rate": 3.208195612980769e-05, "loss": 0.3656, "step": 9545 }, { "epoch": 2.87, "learning_rate": 3.2080078125000004e-05, "loss": 0.3226, "step": 9546 }, { "epoch": 2.87, "learning_rate": 3.207820012019231e-05, "loss": 0.4459, "step": 9547 }, { "epoch": 2.87, "learning_rate": 3.2076322115384614e-05, "loss": 0.3151, "step": 9548 }, { "epoch": 2.87, "learning_rate": 3.2074444110576925e-05, "loss": 0.3626, "step": 9549 }, { "epoch": 2.87, "learning_rate": 3.207256610576923e-05, "loss": 0.6116, "step": 9550 }, { "epoch": 2.87, "learning_rate": 3.2070688100961535e-05, "loss": 0.4541, "step": 9551 }, { "epoch": 2.87, "learning_rate": 3.2068810096153846e-05, "loss": 0.3744, "step": 9552 }, { "epoch": 2.87, "learning_rate": 3.206693209134616e-05, "loss": 0.4162, "step": 9553 }, { "epoch": 2.87, "learning_rate": 3.206505408653846e-05, "loss": 0.3825, "step": 9554 }, { "epoch": 2.87, "learning_rate": 3.2063176081730774e-05, "loss": 0.3809, "step": 9555 }, { "epoch": 2.87, "learning_rate": 3.206129807692308e-05, "loss": 0.3565, "step": 9556 }, { "epoch": 2.87, "learning_rate": 3.2059420072115384e-05, "loss": 0.374, "step": 9557 }, { "epoch": 2.87, "learning_rate": 3.2057542067307695e-05, "loss": 0.3654, "step": 9558 }, { "epoch": 2.87, "learning_rate": 3.20556640625e-05, "loss": 0.3451, "step": 9559 }, { "epoch": 2.87, "learning_rate": 3.205378605769231e-05, "loss": 0.2401, "step": 9560 }, { "epoch": 2.87, "learning_rate": 3.2051908052884616e-05, "loss": 0.4378, "step": 9561 }, { "epoch": 2.87, "learning_rate": 3.205003004807692e-05, "loss": 0.4174, "step": 9562 }, { "epoch": 2.87, "learning_rate": 3.204815204326923e-05, "loss": 0.3451, "step": 9563 }, { "epoch": 2.87, "learning_rate": 3.204627403846154e-05, "loss": 0.3352, "step": 9564 }, { "epoch": 2.87, "learning_rate": 3.204439603365384e-05, "loss": 0.3121, "step": 9565 }, { "epoch": 2.87, "learning_rate": 3.204251802884616e-05, "loss": 0.3286, "step": 9566 }, { "epoch": 2.87, "learning_rate": 3.2040640024038465e-05, "loss": 0.2515, "step": 9567 }, { "epoch": 2.88, "learning_rate": 3.203876201923077e-05, "loss": 0.424, "step": 9568 }, { "epoch": 2.88, "learning_rate": 3.203688401442308e-05, "loss": 0.3421, "step": 9569 }, { "epoch": 2.88, "learning_rate": 3.2035006009615386e-05, "loss": 0.3636, "step": 9570 }, { "epoch": 2.88, "learning_rate": 3.203312800480769e-05, "loss": 0.3539, "step": 9571 }, { "epoch": 2.88, "learning_rate": 3.203125e-05, "loss": 0.2875, "step": 9572 }, { "epoch": 2.88, "learning_rate": 3.202937199519231e-05, "loss": 0.462, "step": 9573 }, { "epoch": 2.88, "learning_rate": 3.202749399038462e-05, "loss": 0.2689, "step": 9574 }, { "epoch": 2.88, "learning_rate": 3.2025615985576924e-05, "loss": 0.4315, "step": 9575 }, { "epoch": 2.88, "learning_rate": 3.202373798076923e-05, "loss": 0.3973, "step": 9576 }, { "epoch": 2.88, "learning_rate": 3.202185997596154e-05, "loss": 0.4699, "step": 9577 }, { "epoch": 2.88, "learning_rate": 3.2019981971153845e-05, "loss": 0.3216, "step": 9578 }, { "epoch": 2.88, "learning_rate": 3.2018103966346156e-05, "loss": 0.4307, "step": 9579 }, { "epoch": 2.88, "learning_rate": 3.201622596153847e-05, "loss": 0.359, "step": 9580 }, { "epoch": 2.88, "learning_rate": 3.201434795673077e-05, "loss": 0.4287, "step": 9581 }, { "epoch": 2.88, "learning_rate": 3.201246995192308e-05, "loss": 0.3937, "step": 9582 }, { "epoch": 2.88, "learning_rate": 3.201059194711539e-05, "loss": 0.4157, "step": 9583 }, { "epoch": 2.88, "learning_rate": 3.2008713942307694e-05, "loss": 0.3069, "step": 9584 }, { "epoch": 2.88, "learning_rate": 3.20068359375e-05, "loss": 0.6007, "step": 9585 }, { "epoch": 2.88, "learning_rate": 3.200495793269231e-05, "loss": 0.3842, "step": 9586 }, { "epoch": 2.88, "learning_rate": 3.2003079927884615e-05, "loss": 0.3536, "step": 9587 }, { "epoch": 2.88, "learning_rate": 3.2001201923076926e-05, "loss": 0.3666, "step": 9588 }, { "epoch": 2.88, "learning_rate": 3.199932391826923e-05, "loss": 0.3424, "step": 9589 }, { "epoch": 2.88, "learning_rate": 3.1997445913461536e-05, "loss": 0.5037, "step": 9590 }, { "epoch": 2.88, "learning_rate": 3.199556790865385e-05, "loss": 0.3411, "step": 9591 }, { "epoch": 2.88, "learning_rate": 3.199368990384616e-05, "loss": 0.3239, "step": 9592 }, { "epoch": 2.88, "learning_rate": 3.1991811899038464e-05, "loss": 0.3673, "step": 9593 }, { "epoch": 2.88, "learning_rate": 3.1989933894230775e-05, "loss": 0.3222, "step": 9594 }, { "epoch": 2.88, "learning_rate": 3.198805588942308e-05, "loss": 0.324, "step": 9595 }, { "epoch": 2.88, "learning_rate": 3.1986177884615385e-05, "loss": 0.4005, "step": 9596 }, { "epoch": 2.88, "learning_rate": 3.1984299879807697e-05, "loss": 0.391, "step": 9597 }, { "epoch": 2.88, "learning_rate": 3.1982421875e-05, "loss": 0.4393, "step": 9598 }, { "epoch": 2.88, "learning_rate": 3.1980543870192306e-05, "loss": 0.3029, "step": 9599 }, { "epoch": 2.88, "learning_rate": 3.197866586538462e-05, "loss": 0.4322, "step": 9600 }, { "epoch": 2.88, "learning_rate": 3.197678786057692e-05, "loss": 0.4356, "step": 9601 }, { "epoch": 2.89, "learning_rate": 3.1974909855769234e-05, "loss": 0.3673, "step": 9602 }, { "epoch": 2.89, "learning_rate": 3.197303185096154e-05, "loss": 0.5023, "step": 9603 }, { "epoch": 2.89, "learning_rate": 3.1971153846153843e-05, "loss": 0.3442, "step": 9604 }, { "epoch": 2.89, "learning_rate": 3.1969275841346155e-05, "loss": 0.4928, "step": 9605 }, { "epoch": 2.89, "learning_rate": 3.1967397836538467e-05, "loss": 0.4435, "step": 9606 }, { "epoch": 2.89, "learning_rate": 3.196551983173077e-05, "loss": 0.3974, "step": 9607 }, { "epoch": 2.89, "learning_rate": 3.196364182692308e-05, "loss": 0.3727, "step": 9608 }, { "epoch": 2.89, "learning_rate": 3.196176382211539e-05, "loss": 0.2998, "step": 9609 }, { "epoch": 2.89, "learning_rate": 3.195988581730769e-05, "loss": 0.2229, "step": 9610 }, { "epoch": 2.89, "learning_rate": 3.1958007812500004e-05, "loss": 0.3803, "step": 9611 }, { "epoch": 2.89, "learning_rate": 3.195612980769231e-05, "loss": 0.3239, "step": 9612 }, { "epoch": 2.89, "learning_rate": 3.1954251802884614e-05, "loss": 0.567, "step": 9613 }, { "epoch": 2.89, "learning_rate": 3.1952373798076925e-05, "loss": 0.3841, "step": 9614 }, { "epoch": 2.89, "learning_rate": 3.195049579326923e-05, "loss": 0.5055, "step": 9615 }, { "epoch": 2.89, "learning_rate": 3.1948617788461535e-05, "loss": 0.3487, "step": 9616 }, { "epoch": 2.89, "learning_rate": 3.1946739783653846e-05, "loss": 0.3744, "step": 9617 }, { "epoch": 2.89, "learning_rate": 3.194486177884616e-05, "loss": 0.3783, "step": 9618 }, { "epoch": 2.89, "learning_rate": 3.194298377403846e-05, "loss": 0.3221, "step": 9619 }, { "epoch": 2.89, "learning_rate": 3.1941105769230774e-05, "loss": 0.3769, "step": 9620 }, { "epoch": 2.89, "learning_rate": 3.193922776442308e-05, "loss": 0.3087, "step": 9621 }, { "epoch": 2.89, "learning_rate": 3.193734975961539e-05, "loss": 0.3854, "step": 9622 }, { "epoch": 2.89, "learning_rate": 3.1935471754807695e-05, "loss": 0.4351, "step": 9623 }, { "epoch": 2.89, "learning_rate": 3.193359375e-05, "loss": 0.4162, "step": 9624 }, { "epoch": 2.89, "learning_rate": 3.193171574519231e-05, "loss": 0.3818, "step": 9625 }, { "epoch": 2.89, "learning_rate": 3.1929837740384616e-05, "loss": 0.2978, "step": 9626 }, { "epoch": 2.89, "learning_rate": 3.192795973557692e-05, "loss": 0.3701, "step": 9627 }, { "epoch": 2.89, "learning_rate": 3.192608173076923e-05, "loss": 0.4657, "step": 9628 }, { "epoch": 2.89, "learning_rate": 3.192420372596154e-05, "loss": 0.5781, "step": 9629 }, { "epoch": 2.89, "learning_rate": 3.192232572115384e-05, "loss": 0.2742, "step": 9630 }, { "epoch": 2.89, "learning_rate": 3.192044771634616e-05, "loss": 0.2702, "step": 9631 }, { "epoch": 2.89, "learning_rate": 3.1918569711538465e-05, "loss": 0.3467, "step": 9632 }, { "epoch": 2.89, "learning_rate": 3.191669170673077e-05, "loss": 0.291, "step": 9633 }, { "epoch": 2.89, "learning_rate": 3.191481370192308e-05, "loss": 0.3658, "step": 9634 }, { "epoch": 2.9, "learning_rate": 3.1912935697115386e-05, "loss": 0.3642, "step": 9635 }, { "epoch": 2.9, "learning_rate": 3.191105769230769e-05, "loss": 0.3373, "step": 9636 }, { "epoch": 2.9, "learning_rate": 3.19091796875e-05, "loss": 0.3451, "step": 9637 }, { "epoch": 2.9, "learning_rate": 3.190730168269231e-05, "loss": 0.3438, "step": 9638 }, { "epoch": 2.9, "learning_rate": 3.190542367788462e-05, "loss": 0.6308, "step": 9639 }, { "epoch": 2.9, "learning_rate": 3.1903545673076924e-05, "loss": 0.3237, "step": 9640 }, { "epoch": 2.9, "learning_rate": 3.190166766826923e-05, "loss": 0.3582, "step": 9641 }, { "epoch": 2.9, "learning_rate": 3.189978966346154e-05, "loss": 0.3815, "step": 9642 }, { "epoch": 2.9, "learning_rate": 3.1897911658653845e-05, "loss": 0.5133, "step": 9643 }, { "epoch": 2.9, "learning_rate": 3.1896033653846156e-05, "loss": 0.5444, "step": 9644 }, { "epoch": 2.9, "learning_rate": 3.189415564903847e-05, "loss": 0.5252, "step": 9645 }, { "epoch": 2.9, "learning_rate": 3.189227764423077e-05, "loss": 0.3515, "step": 9646 }, { "epoch": 2.9, "learning_rate": 3.189039963942308e-05, "loss": 0.3652, "step": 9647 }, { "epoch": 2.9, "learning_rate": 3.188852163461539e-05, "loss": 0.4255, "step": 9648 }, { "epoch": 2.9, "learning_rate": 3.1886643629807694e-05, "loss": 0.2309, "step": 9649 }, { "epoch": 2.9, "learning_rate": 3.1884765625e-05, "loss": 0.37, "step": 9650 }, { "epoch": 2.9, "learning_rate": 3.188288762019231e-05, "loss": 0.3319, "step": 9651 }, { "epoch": 2.9, "learning_rate": 3.1881009615384615e-05, "loss": 0.339, "step": 9652 }, { "epoch": 2.9, "learning_rate": 3.1879131610576926e-05, "loss": 0.4299, "step": 9653 }, { "epoch": 2.9, "learning_rate": 3.187725360576923e-05, "loss": 0.3958, "step": 9654 }, { "epoch": 2.9, "learning_rate": 3.1875375600961536e-05, "loss": 0.3771, "step": 9655 }, { "epoch": 2.9, "learning_rate": 3.187349759615385e-05, "loss": 0.2491, "step": 9656 }, { "epoch": 2.9, "learning_rate": 3.187161959134616e-05, "loss": 0.2025, "step": 9657 }, { "epoch": 2.9, "learning_rate": 3.1869741586538464e-05, "loss": 0.3099, "step": 9658 }, { "epoch": 2.9, "learning_rate": 3.1867863581730775e-05, "loss": 0.3583, "step": 9659 }, { "epoch": 2.9, "learning_rate": 3.186598557692308e-05, "loss": 0.755, "step": 9660 }, { "epoch": 2.9, "learning_rate": 3.1864107572115385e-05, "loss": 0.3082, "step": 9661 }, { "epoch": 2.9, "learning_rate": 3.1862229567307696e-05, "loss": 0.3355, "step": 9662 }, { "epoch": 2.9, "learning_rate": 3.18603515625e-05, "loss": 0.371, "step": 9663 }, { "epoch": 2.9, "learning_rate": 3.1858473557692306e-05, "loss": 0.2739, "step": 9664 }, { "epoch": 2.9, "learning_rate": 3.185659555288462e-05, "loss": 0.2814, "step": 9665 }, { "epoch": 2.9, "learning_rate": 3.185471754807692e-05, "loss": 0.3294, "step": 9666 }, { "epoch": 2.9, "learning_rate": 3.1852839543269234e-05, "loss": 0.3449, "step": 9667 }, { "epoch": 2.91, "learning_rate": 3.185096153846154e-05, "loss": 0.526, "step": 9668 }, { "epoch": 2.91, "learning_rate": 3.184908353365384e-05, "loss": 0.2727, "step": 9669 }, { "epoch": 2.91, "learning_rate": 3.1847205528846155e-05, "loss": 0.2598, "step": 9670 }, { "epoch": 2.91, "learning_rate": 3.1845327524038466e-05, "loss": 0.4424, "step": 9671 }, { "epoch": 2.91, "learning_rate": 3.184344951923077e-05, "loss": 0.4072, "step": 9672 }, { "epoch": 2.91, "learning_rate": 3.184157151442308e-05, "loss": 0.4252, "step": 9673 }, { "epoch": 2.91, "learning_rate": 3.183969350961539e-05, "loss": 0.3849, "step": 9674 }, { "epoch": 2.91, "learning_rate": 3.183781550480769e-05, "loss": 0.3893, "step": 9675 }, { "epoch": 2.91, "learning_rate": 3.1835937500000004e-05, "loss": 0.3357, "step": 9676 }, { "epoch": 2.91, "learning_rate": 3.183405949519231e-05, "loss": 0.3507, "step": 9677 }, { "epoch": 2.91, "learning_rate": 3.1832181490384613e-05, "loss": 0.3669, "step": 9678 }, { "epoch": 2.91, "learning_rate": 3.1830303485576925e-05, "loss": 0.4972, "step": 9679 }, { "epoch": 2.91, "learning_rate": 3.182842548076923e-05, "loss": 0.3709, "step": 9680 }, { "epoch": 2.91, "learning_rate": 3.1826547475961535e-05, "loss": 0.3482, "step": 9681 }, { "epoch": 2.91, "learning_rate": 3.1824669471153846e-05, "loss": 0.2962, "step": 9682 }, { "epoch": 2.91, "learning_rate": 3.182279146634616e-05, "loss": 0.4049, "step": 9683 }, { "epoch": 2.91, "learning_rate": 3.182091346153846e-05, "loss": 0.4076, "step": 9684 }, { "epoch": 2.91, "learning_rate": 3.1819035456730774e-05, "loss": 0.4573, "step": 9685 }, { "epoch": 2.91, "learning_rate": 3.181715745192308e-05, "loss": 0.4379, "step": 9686 }, { "epoch": 2.91, "learning_rate": 3.181527944711539e-05, "loss": 0.3962, "step": 9687 }, { "epoch": 2.91, "learning_rate": 3.1813401442307695e-05, "loss": 0.5227, "step": 9688 }, { "epoch": 2.91, "learning_rate": 3.18115234375e-05, "loss": 0.3494, "step": 9689 }, { "epoch": 2.91, "learning_rate": 3.180964543269231e-05, "loss": 0.2791, "step": 9690 }, { "epoch": 2.91, "learning_rate": 3.1807767427884616e-05, "loss": 0.366, "step": 9691 }, { "epoch": 2.91, "learning_rate": 3.180588942307692e-05, "loss": 0.4253, "step": 9692 }, { "epoch": 2.91, "learning_rate": 3.180401141826923e-05, "loss": 0.5334, "step": 9693 }, { "epoch": 2.91, "learning_rate": 3.180213341346154e-05, "loss": 0.4692, "step": 9694 }, { "epoch": 2.91, "learning_rate": 3.180025540865384e-05, "loss": 0.3624, "step": 9695 }, { "epoch": 2.91, "learning_rate": 3.179837740384616e-05, "loss": 0.441, "step": 9696 }, { "epoch": 2.91, "learning_rate": 3.1796499399038465e-05, "loss": 0.291, "step": 9697 }, { "epoch": 2.91, "learning_rate": 3.179462139423077e-05, "loss": 0.3936, "step": 9698 }, { "epoch": 2.91, "learning_rate": 3.179274338942308e-05, "loss": 0.4001, "step": 9699 }, { "epoch": 2.91, "learning_rate": 3.1790865384615386e-05, "loss": 0.4601, "step": 9700 }, { "epoch": 2.91, "learning_rate": 3.178898737980769e-05, "loss": 0.4265, "step": 9701 }, { "epoch": 2.92, "learning_rate": 3.1787109375e-05, "loss": 0.4469, "step": 9702 }, { "epoch": 2.92, "learning_rate": 3.178523137019231e-05, "loss": 0.4592, "step": 9703 }, { "epoch": 2.92, "learning_rate": 3.178335336538462e-05, "loss": 0.3241, "step": 9704 }, { "epoch": 2.92, "learning_rate": 3.1781475360576924e-05, "loss": 0.442, "step": 9705 }, { "epoch": 2.92, "learning_rate": 3.177959735576923e-05, "loss": 0.2708, "step": 9706 }, { "epoch": 2.92, "learning_rate": 3.177771935096154e-05, "loss": 0.4346, "step": 9707 }, { "epoch": 2.92, "learning_rate": 3.1775841346153845e-05, "loss": 0.2786, "step": 9708 }, { "epoch": 2.92, "learning_rate": 3.1773963341346156e-05, "loss": 0.39, "step": 9709 }, { "epoch": 2.92, "learning_rate": 3.177208533653847e-05, "loss": 0.3861, "step": 9710 }, { "epoch": 2.92, "learning_rate": 3.177020733173077e-05, "loss": 0.3163, "step": 9711 }, { "epoch": 2.92, "learning_rate": 3.176832932692308e-05, "loss": 0.312, "step": 9712 }, { "epoch": 2.92, "learning_rate": 3.176645132211539e-05, "loss": 0.4544, "step": 9713 }, { "epoch": 2.92, "learning_rate": 3.1764573317307694e-05, "loss": 0.4533, "step": 9714 }, { "epoch": 2.92, "learning_rate": 3.17626953125e-05, "loss": 0.4212, "step": 9715 }, { "epoch": 2.92, "learning_rate": 3.176081730769231e-05, "loss": 0.3372, "step": 9716 }, { "epoch": 2.92, "learning_rate": 3.1758939302884615e-05, "loss": 0.3752, "step": 9717 }, { "epoch": 2.92, "learning_rate": 3.1757061298076926e-05, "loss": 0.3813, "step": 9718 }, { "epoch": 2.92, "learning_rate": 3.175518329326923e-05, "loss": 0.3362, "step": 9719 }, { "epoch": 2.92, "learning_rate": 3.1753305288461536e-05, "loss": 0.5303, "step": 9720 }, { "epoch": 2.92, "learning_rate": 3.175142728365385e-05, "loss": 0.4186, "step": 9721 }, { "epoch": 2.92, "learning_rate": 3.174954927884616e-05, "loss": 0.3453, "step": 9722 }, { "epoch": 2.92, "learning_rate": 3.1747671274038464e-05, "loss": 0.3489, "step": 9723 }, { "epoch": 2.92, "learning_rate": 3.1745793269230775e-05, "loss": 0.3515, "step": 9724 }, { "epoch": 2.92, "learning_rate": 3.174391526442308e-05, "loss": 0.4378, "step": 9725 }, { "epoch": 2.92, "learning_rate": 3.1742037259615385e-05, "loss": 0.3993, "step": 9726 }, { "epoch": 2.92, "learning_rate": 3.1740159254807696e-05, "loss": 0.3883, "step": 9727 }, { "epoch": 2.92, "learning_rate": 3.173828125e-05, "loss": 0.5088, "step": 9728 }, { "epoch": 2.92, "learning_rate": 3.1736403245192306e-05, "loss": 0.3048, "step": 9729 }, { "epoch": 2.92, "learning_rate": 3.173452524038462e-05, "loss": 0.3481, "step": 9730 }, { "epoch": 2.92, "learning_rate": 3.173264723557692e-05, "loss": 0.3638, "step": 9731 }, { "epoch": 2.92, "learning_rate": 3.1730769230769234e-05, "loss": 0.3734, "step": 9732 }, { "epoch": 2.92, "learning_rate": 3.172889122596154e-05, "loss": 0.317, "step": 9733 }, { "epoch": 2.92, "learning_rate": 3.172701322115384e-05, "loss": 0.391, "step": 9734 }, { "epoch": 2.93, "learning_rate": 3.1725135216346155e-05, "loss": 0.3517, "step": 9735 }, { "epoch": 2.93, "learning_rate": 3.1723257211538466e-05, "loss": 0.3611, "step": 9736 }, { "epoch": 2.93, "learning_rate": 3.172137920673077e-05, "loss": 0.3913, "step": 9737 }, { "epoch": 2.93, "learning_rate": 3.171950120192308e-05, "loss": 0.417, "step": 9738 }, { "epoch": 2.93, "learning_rate": 3.171762319711539e-05, "loss": 0.4455, "step": 9739 }, { "epoch": 2.93, "learning_rate": 3.171574519230769e-05, "loss": 0.4061, "step": 9740 }, { "epoch": 2.93, "learning_rate": 3.1713867187500004e-05, "loss": 0.319, "step": 9741 }, { "epoch": 2.93, "learning_rate": 3.171198918269231e-05, "loss": 0.452, "step": 9742 }, { "epoch": 2.93, "learning_rate": 3.171011117788461e-05, "loss": 0.481, "step": 9743 }, { "epoch": 2.93, "learning_rate": 3.1708233173076925e-05, "loss": 0.3259, "step": 9744 }, { "epoch": 2.93, "learning_rate": 3.170635516826923e-05, "loss": 0.3718, "step": 9745 }, { "epoch": 2.93, "learning_rate": 3.170447716346154e-05, "loss": 0.3314, "step": 9746 }, { "epoch": 2.93, "learning_rate": 3.1702599158653846e-05, "loss": 0.4287, "step": 9747 }, { "epoch": 2.93, "learning_rate": 3.170072115384616e-05, "loss": 0.3271, "step": 9748 }, { "epoch": 2.93, "learning_rate": 3.169884314903846e-05, "loss": 0.4221, "step": 9749 }, { "epoch": 2.93, "learning_rate": 3.1696965144230774e-05, "loss": 0.2735, "step": 9750 }, { "epoch": 2.93, "learning_rate": 3.169508713942308e-05, "loss": 0.2628, "step": 9751 }, { "epoch": 2.93, "learning_rate": 3.169320913461539e-05, "loss": 0.3871, "step": 9752 }, { "epoch": 2.93, "learning_rate": 3.1691331129807695e-05, "loss": 0.4322, "step": 9753 }, { "epoch": 2.93, "learning_rate": 3.1689453125e-05, "loss": 0.4009, "step": 9754 }, { "epoch": 2.93, "learning_rate": 3.168757512019231e-05, "loss": 0.3251, "step": 9755 }, { "epoch": 2.93, "learning_rate": 3.1685697115384616e-05, "loss": 0.4446, "step": 9756 }, { "epoch": 2.93, "learning_rate": 3.168381911057692e-05, "loss": 0.431, "step": 9757 }, { "epoch": 2.93, "learning_rate": 3.168194110576923e-05, "loss": 0.3012, "step": 9758 }, { "epoch": 2.93, "learning_rate": 3.168006310096154e-05, "loss": 0.3675, "step": 9759 }, { "epoch": 2.93, "learning_rate": 3.167818509615384e-05, "loss": 0.4141, "step": 9760 }, { "epoch": 2.93, "learning_rate": 3.167630709134616e-05, "loss": 0.3646, "step": 9761 }, { "epoch": 2.93, "learning_rate": 3.1674429086538465e-05, "loss": 0.2184, "step": 9762 }, { "epoch": 2.93, "learning_rate": 3.167255108173077e-05, "loss": 0.5466, "step": 9763 }, { "epoch": 2.93, "learning_rate": 3.167067307692308e-05, "loss": 0.3471, "step": 9764 }, { "epoch": 2.93, "learning_rate": 3.1668795072115386e-05, "loss": 0.4944, "step": 9765 }, { "epoch": 2.93, "learning_rate": 3.16669170673077e-05, "loss": 0.4607, "step": 9766 }, { "epoch": 2.93, "learning_rate": 3.16650390625e-05, "loss": 0.3306, "step": 9767 }, { "epoch": 2.94, "learning_rate": 3.166316105769231e-05, "loss": 0.3206, "step": 9768 }, { "epoch": 2.94, "learning_rate": 3.166128305288462e-05, "loss": 0.3932, "step": 9769 }, { "epoch": 2.94, "learning_rate": 3.1659405048076923e-05, "loss": 0.3438, "step": 9770 }, { "epoch": 2.94, "learning_rate": 3.165752704326923e-05, "loss": 0.4552, "step": 9771 }, { "epoch": 2.94, "learning_rate": 3.165564903846154e-05, "loss": 0.4104, "step": 9772 }, { "epoch": 2.94, "learning_rate": 3.1653771033653845e-05, "loss": 0.2856, "step": 9773 }, { "epoch": 2.94, "learning_rate": 3.1651893028846156e-05, "loss": 0.3576, "step": 9774 }, { "epoch": 2.94, "learning_rate": 3.165001502403847e-05, "loss": 0.3762, "step": 9775 }, { "epoch": 2.94, "learning_rate": 3.164813701923077e-05, "loss": 0.4189, "step": 9776 }, { "epoch": 2.94, "learning_rate": 3.164625901442308e-05, "loss": 0.367, "step": 9777 }, { "epoch": 2.94, "learning_rate": 3.164438100961539e-05, "loss": 0.3528, "step": 9778 }, { "epoch": 2.94, "learning_rate": 3.1642503004807694e-05, "loss": 0.3743, "step": 9779 }, { "epoch": 2.94, "learning_rate": 3.1640625e-05, "loss": 0.3008, "step": 9780 }, { "epoch": 2.94, "learning_rate": 3.163874699519231e-05, "loss": 0.5585, "step": 9781 }, { "epoch": 2.94, "learning_rate": 3.1636868990384615e-05, "loss": 0.2692, "step": 9782 }, { "epoch": 2.94, "learning_rate": 3.1634990985576926e-05, "loss": 0.3509, "step": 9783 }, { "epoch": 2.94, "learning_rate": 3.163311298076923e-05, "loss": 0.4778, "step": 9784 }, { "epoch": 2.94, "learning_rate": 3.1631234975961536e-05, "loss": 0.32, "step": 9785 }, { "epoch": 2.94, "learning_rate": 3.162935697115385e-05, "loss": 0.3546, "step": 9786 }, { "epoch": 2.94, "learning_rate": 3.162747896634616e-05, "loss": 0.4564, "step": 9787 }, { "epoch": 2.94, "learning_rate": 3.1625600961538464e-05, "loss": 0.3191, "step": 9788 }, { "epoch": 2.94, "learning_rate": 3.1623722956730775e-05, "loss": 0.4028, "step": 9789 }, { "epoch": 2.94, "learning_rate": 3.162184495192308e-05, "loss": 0.4184, "step": 9790 }, { "epoch": 2.94, "learning_rate": 3.1619966947115385e-05, "loss": 0.4537, "step": 9791 }, { "epoch": 2.94, "learning_rate": 3.1618088942307696e-05, "loss": 0.4136, "step": 9792 }, { "epoch": 2.94, "learning_rate": 3.16162109375e-05, "loss": 0.3795, "step": 9793 }, { "epoch": 2.94, "learning_rate": 3.1614332932692306e-05, "loss": 0.2617, "step": 9794 }, { "epoch": 2.94, "learning_rate": 3.161245492788462e-05, "loss": 0.3466, "step": 9795 }, { "epoch": 2.94, "learning_rate": 3.161057692307692e-05, "loss": 0.3237, "step": 9796 }, { "epoch": 2.94, "learning_rate": 3.1608698918269234e-05, "loss": 0.3579, "step": 9797 }, { "epoch": 2.94, "learning_rate": 3.160682091346154e-05, "loss": 0.3479, "step": 9798 }, { "epoch": 2.94, "learning_rate": 3.160494290865384e-05, "loss": 0.4737, "step": 9799 }, { "epoch": 2.94, "learning_rate": 3.1603064903846155e-05, "loss": 0.2999, "step": 9800 }, { "epoch": 2.95, "learning_rate": 3.1601186899038466e-05, "loss": 0.4174, "step": 9801 }, { "epoch": 2.95, "learning_rate": 3.159930889423077e-05, "loss": 0.4371, "step": 9802 }, { "epoch": 2.95, "learning_rate": 3.159743088942308e-05, "loss": 0.246, "step": 9803 }, { "epoch": 2.95, "learning_rate": 3.159555288461539e-05, "loss": 0.3941, "step": 9804 }, { "epoch": 2.95, "learning_rate": 3.159367487980769e-05, "loss": 0.2996, "step": 9805 }, { "epoch": 2.95, "learning_rate": 3.1591796875000004e-05, "loss": 0.3061, "step": 9806 }, { "epoch": 2.95, "learning_rate": 3.158991887019231e-05, "loss": 0.4626, "step": 9807 }, { "epoch": 2.95, "learning_rate": 3.158804086538461e-05, "loss": 0.3882, "step": 9808 }, { "epoch": 2.95, "learning_rate": 3.1586162860576925e-05, "loss": 0.3216, "step": 9809 }, { "epoch": 2.95, "learning_rate": 3.158428485576923e-05, "loss": 0.4878, "step": 9810 }, { "epoch": 2.95, "learning_rate": 3.158240685096154e-05, "loss": 0.2791, "step": 9811 }, { "epoch": 2.95, "learning_rate": 3.1580528846153846e-05, "loss": 0.3126, "step": 9812 }, { "epoch": 2.95, "learning_rate": 3.157865084134616e-05, "loss": 0.3536, "step": 9813 }, { "epoch": 2.95, "learning_rate": 3.157677283653846e-05, "loss": 0.3632, "step": 9814 }, { "epoch": 2.95, "learning_rate": 3.1574894831730774e-05, "loss": 0.3633, "step": 9815 }, { "epoch": 2.95, "learning_rate": 3.157301682692308e-05, "loss": 0.3828, "step": 9816 }, { "epoch": 2.95, "learning_rate": 3.157113882211539e-05, "loss": 0.34, "step": 9817 }, { "epoch": 2.95, "learning_rate": 3.1569260817307695e-05, "loss": 0.4707, "step": 9818 }, { "epoch": 2.95, "learning_rate": 3.15673828125e-05, "loss": 0.2891, "step": 9819 }, { "epoch": 2.95, "learning_rate": 3.156550480769231e-05, "loss": 0.4532, "step": 9820 }, { "epoch": 2.95, "learning_rate": 3.1563626802884616e-05, "loss": 0.3353, "step": 9821 }, { "epoch": 2.95, "learning_rate": 3.156174879807692e-05, "loss": 0.3464, "step": 9822 }, { "epoch": 2.95, "learning_rate": 3.155987079326923e-05, "loss": 0.364, "step": 9823 }, { "epoch": 2.95, "learning_rate": 3.155799278846154e-05, "loss": 0.2755, "step": 9824 }, { "epoch": 2.95, "learning_rate": 3.155611478365384e-05, "loss": 0.4295, "step": 9825 }, { "epoch": 2.95, "learning_rate": 3.155423677884616e-05, "loss": 0.2653, "step": 9826 }, { "epoch": 2.95, "learning_rate": 3.1552358774038465e-05, "loss": 0.2773, "step": 9827 }, { "epoch": 2.95, "learning_rate": 3.155048076923077e-05, "loss": 0.4315, "step": 9828 }, { "epoch": 2.95, "learning_rate": 3.154860276442308e-05, "loss": 0.4237, "step": 9829 }, { "epoch": 2.95, "learning_rate": 3.1546724759615386e-05, "loss": 0.5002, "step": 9830 }, { "epoch": 2.95, "learning_rate": 3.15448467548077e-05, "loss": 0.4273, "step": 9831 }, { "epoch": 2.95, "learning_rate": 3.154296875e-05, "loss": 0.4081, "step": 9832 }, { "epoch": 2.95, "learning_rate": 3.154109074519231e-05, "loss": 0.4355, "step": 9833 }, { "epoch": 2.95, "learning_rate": 3.153921274038462e-05, "loss": 0.4184, "step": 9834 }, { "epoch": 2.96, "learning_rate": 3.153733473557692e-05, "loss": 0.3152, "step": 9835 }, { "epoch": 2.96, "learning_rate": 3.153545673076923e-05, "loss": 0.489, "step": 9836 }, { "epoch": 2.96, "learning_rate": 3.153357872596154e-05, "loss": 0.3349, "step": 9837 }, { "epoch": 2.96, "learning_rate": 3.1531700721153844e-05, "loss": 0.3518, "step": 9838 }, { "epoch": 2.96, "learning_rate": 3.1529822716346156e-05, "loss": 0.394, "step": 9839 }, { "epoch": 2.96, "learning_rate": 3.152794471153847e-05, "loss": 0.4095, "step": 9840 }, { "epoch": 2.96, "learning_rate": 3.152606670673077e-05, "loss": 0.3701, "step": 9841 }, { "epoch": 2.96, "learning_rate": 3.152418870192308e-05, "loss": 0.4323, "step": 9842 }, { "epoch": 2.96, "learning_rate": 3.152231069711539e-05, "loss": 0.3016, "step": 9843 }, { "epoch": 2.96, "learning_rate": 3.1520432692307693e-05, "loss": 0.3414, "step": 9844 }, { "epoch": 2.96, "learning_rate": 3.15185546875e-05, "loss": 0.4047, "step": 9845 }, { "epoch": 2.96, "learning_rate": 3.151667668269231e-05, "loss": 0.3404, "step": 9846 }, { "epoch": 2.96, "learning_rate": 3.1514798677884614e-05, "loss": 0.305, "step": 9847 }, { "epoch": 2.96, "learning_rate": 3.1512920673076926e-05, "loss": 0.3634, "step": 9848 }, { "epoch": 2.96, "learning_rate": 3.151104266826923e-05, "loss": 0.2626, "step": 9849 }, { "epoch": 2.96, "learning_rate": 3.1509164663461536e-05, "loss": 0.4822, "step": 9850 }, { "epoch": 2.96, "learning_rate": 3.150728665865385e-05, "loss": 0.3627, "step": 9851 }, { "epoch": 2.96, "learning_rate": 3.150540865384616e-05, "loss": 0.4638, "step": 9852 }, { "epoch": 2.96, "learning_rate": 3.1503530649038463e-05, "loss": 0.5722, "step": 9853 }, { "epoch": 2.96, "learning_rate": 3.1501652644230775e-05, "loss": 0.3158, "step": 9854 }, { "epoch": 2.96, "learning_rate": 3.149977463942308e-05, "loss": 0.3356, "step": 9855 }, { "epoch": 2.96, "learning_rate": 3.1497896634615385e-05, "loss": 0.5076, "step": 9856 }, { "epoch": 2.96, "learning_rate": 3.1496018629807696e-05, "loss": 0.4549, "step": 9857 }, { "epoch": 2.96, "learning_rate": 3.1494140625e-05, "loss": 0.3645, "step": 9858 }, { "epoch": 2.96, "learning_rate": 3.1492262620192306e-05, "loss": 0.3917, "step": 9859 }, { "epoch": 2.96, "learning_rate": 3.149038461538462e-05, "loss": 0.3442, "step": 9860 }, { "epoch": 2.96, "learning_rate": 3.148850661057692e-05, "loss": 0.3393, "step": 9861 }, { "epoch": 2.96, "learning_rate": 3.1486628605769234e-05, "loss": 0.3306, "step": 9862 }, { "epoch": 2.96, "learning_rate": 3.148475060096154e-05, "loss": 0.3452, "step": 9863 }, { "epoch": 2.96, "learning_rate": 3.148287259615384e-05, "loss": 0.514, "step": 9864 }, { "epoch": 2.96, "learning_rate": 3.1480994591346155e-05, "loss": 0.2906, "step": 9865 }, { "epoch": 2.96, "learning_rate": 3.1479116586538466e-05, "loss": 0.2702, "step": 9866 }, { "epoch": 2.96, "learning_rate": 3.147723858173077e-05, "loss": 0.4904, "step": 9867 }, { "epoch": 2.97, "learning_rate": 3.147536057692308e-05, "loss": 0.3511, "step": 9868 }, { "epoch": 2.97, "learning_rate": 3.147348257211539e-05, "loss": 0.446, "step": 9869 }, { "epoch": 2.97, "learning_rate": 3.147160456730769e-05, "loss": 0.4899, "step": 9870 }, { "epoch": 2.97, "learning_rate": 3.1469726562500004e-05, "loss": 0.5012, "step": 9871 }, { "epoch": 2.97, "learning_rate": 3.146784855769231e-05, "loss": 0.3048, "step": 9872 }, { "epoch": 2.97, "learning_rate": 3.146597055288461e-05, "loss": 0.3575, "step": 9873 }, { "epoch": 2.97, "learning_rate": 3.1464092548076925e-05, "loss": 0.3331, "step": 9874 }, { "epoch": 2.97, "learning_rate": 3.146221454326923e-05, "loss": 0.3381, "step": 9875 }, { "epoch": 2.97, "learning_rate": 3.146033653846154e-05, "loss": 0.3459, "step": 9876 }, { "epoch": 2.97, "learning_rate": 3.1458458533653846e-05, "loss": 0.3857, "step": 9877 }, { "epoch": 2.97, "learning_rate": 3.145658052884616e-05, "loss": 0.3922, "step": 9878 }, { "epoch": 2.97, "learning_rate": 3.145470252403846e-05, "loss": 0.2933, "step": 9879 }, { "epoch": 2.97, "learning_rate": 3.1452824519230774e-05, "loss": 0.3787, "step": 9880 }, { "epoch": 2.97, "learning_rate": 3.145094651442308e-05, "loss": 0.2791, "step": 9881 }, { "epoch": 2.97, "learning_rate": 3.144906850961539e-05, "loss": 0.3976, "step": 9882 }, { "epoch": 2.97, "learning_rate": 3.1447190504807695e-05, "loss": 0.3609, "step": 9883 }, { "epoch": 2.97, "learning_rate": 3.14453125e-05, "loss": 0.4184, "step": 9884 }, { "epoch": 2.97, "learning_rate": 3.144343449519231e-05, "loss": 0.5357, "step": 9885 }, { "epoch": 2.97, "learning_rate": 3.1441556490384616e-05, "loss": 0.3756, "step": 9886 }, { "epoch": 2.97, "learning_rate": 3.143967848557692e-05, "loss": 0.3963, "step": 9887 }, { "epoch": 2.97, "learning_rate": 3.143780048076923e-05, "loss": 0.2996, "step": 9888 }, { "epoch": 2.97, "learning_rate": 3.143592247596154e-05, "loss": 0.2931, "step": 9889 }, { "epoch": 2.97, "learning_rate": 3.143404447115385e-05, "loss": 0.4229, "step": 9890 }, { "epoch": 2.97, "learning_rate": 3.143216646634616e-05, "loss": 0.4444, "step": 9891 }, { "epoch": 2.97, "learning_rate": 3.1430288461538465e-05, "loss": 0.2521, "step": 9892 }, { "epoch": 2.97, "learning_rate": 3.142841045673077e-05, "loss": 0.55, "step": 9893 }, { "epoch": 2.97, "learning_rate": 3.142653245192308e-05, "loss": 0.4624, "step": 9894 }, { "epoch": 2.97, "learning_rate": 3.1424654447115386e-05, "loss": 0.353, "step": 9895 }, { "epoch": 2.97, "learning_rate": 3.14227764423077e-05, "loss": 0.3633, "step": 9896 }, { "epoch": 2.97, "learning_rate": 3.14208984375e-05, "loss": 0.3023, "step": 9897 }, { "epoch": 2.97, "learning_rate": 3.141902043269231e-05, "loss": 0.5247, "step": 9898 }, { "epoch": 2.97, "learning_rate": 3.141714242788462e-05, "loss": 0.3032, "step": 9899 }, { "epoch": 2.97, "learning_rate": 3.141526442307692e-05, "loss": 0.558, "step": 9900 }, { "epoch": 2.98, "learning_rate": 3.141338641826923e-05, "loss": 0.3706, "step": 9901 }, { "epoch": 2.98, "learning_rate": 3.141150841346154e-05, "loss": 0.5233, "step": 9902 }, { "epoch": 2.98, "learning_rate": 3.1409630408653844e-05, "loss": 0.3477, "step": 9903 }, { "epoch": 2.98, "learning_rate": 3.1407752403846156e-05, "loss": 0.3669, "step": 9904 }, { "epoch": 2.98, "learning_rate": 3.140587439903847e-05, "loss": 0.3918, "step": 9905 }, { "epoch": 2.98, "learning_rate": 3.140399639423077e-05, "loss": 0.29, "step": 9906 }, { "epoch": 2.98, "learning_rate": 3.140211838942308e-05, "loss": 0.3999, "step": 9907 }, { "epoch": 2.98, "learning_rate": 3.140024038461539e-05, "loss": 0.4061, "step": 9908 }, { "epoch": 2.98, "learning_rate": 3.139836237980769e-05, "loss": 0.2934, "step": 9909 }, { "epoch": 2.98, "learning_rate": 3.1396484375000005e-05, "loss": 0.2726, "step": 9910 }, { "epoch": 2.98, "learning_rate": 3.139460637019231e-05, "loss": 0.431, "step": 9911 }, { "epoch": 2.98, "learning_rate": 3.1392728365384614e-05, "loss": 0.4541, "step": 9912 }, { "epoch": 2.98, "learning_rate": 3.1390850360576926e-05, "loss": 0.4332, "step": 9913 }, { "epoch": 2.98, "learning_rate": 3.138897235576923e-05, "loss": 0.4, "step": 9914 }, { "epoch": 2.98, "learning_rate": 3.1387094350961535e-05, "loss": 0.4126, "step": 9915 }, { "epoch": 2.98, "learning_rate": 3.138521634615385e-05, "loss": 0.3395, "step": 9916 }, { "epoch": 2.98, "learning_rate": 3.138333834134616e-05, "loss": 0.312, "step": 9917 }, { "epoch": 2.98, "learning_rate": 3.138146033653846e-05, "loss": 0.4139, "step": 9918 }, { "epoch": 2.98, "learning_rate": 3.1379582331730775e-05, "loss": 0.54, "step": 9919 }, { "epoch": 2.98, "learning_rate": 3.137770432692308e-05, "loss": 0.2488, "step": 9920 }, { "epoch": 2.98, "learning_rate": 3.1375826322115384e-05, "loss": 0.3972, "step": 9921 }, { "epoch": 2.98, "learning_rate": 3.1373948317307696e-05, "loss": 0.4171, "step": 9922 }, { "epoch": 2.98, "learning_rate": 3.13720703125e-05, "loss": 0.4409, "step": 9923 }, { "epoch": 2.98, "learning_rate": 3.1370192307692306e-05, "loss": 0.5949, "step": 9924 }, { "epoch": 2.98, "learning_rate": 3.136831430288462e-05, "loss": 0.4219, "step": 9925 }, { "epoch": 2.98, "learning_rate": 3.136643629807692e-05, "loss": 0.3007, "step": 9926 }, { "epoch": 2.98, "learning_rate": 3.1364558293269233e-05, "loss": 0.4891, "step": 9927 }, { "epoch": 2.98, "learning_rate": 3.136268028846154e-05, "loss": 0.3289, "step": 9928 }, { "epoch": 2.98, "learning_rate": 3.136080228365384e-05, "loss": 0.3876, "step": 9929 }, { "epoch": 2.98, "learning_rate": 3.135892427884616e-05, "loss": 0.5951, "step": 9930 }, { "epoch": 2.98, "learning_rate": 3.1357046274038466e-05, "loss": 0.3328, "step": 9931 }, { "epoch": 2.98, "learning_rate": 3.135516826923077e-05, "loss": 0.4268, "step": 9932 }, { "epoch": 2.98, "learning_rate": 3.135329026442308e-05, "loss": 0.3066, "step": 9933 }, { "epoch": 2.98, "learning_rate": 3.135141225961539e-05, "loss": 0.3147, "step": 9934 }, { "epoch": 2.99, "learning_rate": 3.134953425480769e-05, "loss": 0.4493, "step": 9935 }, { "epoch": 2.99, "learning_rate": 3.1347656250000003e-05, "loss": 0.3801, "step": 9936 }, { "epoch": 2.99, "learning_rate": 3.134577824519231e-05, "loss": 0.3333, "step": 9937 }, { "epoch": 2.99, "learning_rate": 3.134390024038461e-05, "loss": 0.299, "step": 9938 }, { "epoch": 2.99, "learning_rate": 3.1342022235576925e-05, "loss": 0.3151, "step": 9939 }, { "epoch": 2.99, "learning_rate": 3.134014423076923e-05, "loss": 0.3072, "step": 9940 }, { "epoch": 2.99, "learning_rate": 3.133826622596154e-05, "loss": 0.3429, "step": 9941 }, { "epoch": 2.99, "learning_rate": 3.1336388221153846e-05, "loss": 0.3547, "step": 9942 }, { "epoch": 2.99, "learning_rate": 3.133451021634616e-05, "loss": 0.5195, "step": 9943 }, { "epoch": 2.99, "learning_rate": 3.133263221153846e-05, "loss": 0.3981, "step": 9944 }, { "epoch": 2.99, "learning_rate": 3.1330754206730773e-05, "loss": 0.3837, "step": 9945 }, { "epoch": 2.99, "learning_rate": 3.132887620192308e-05, "loss": 0.3544, "step": 9946 }, { "epoch": 2.99, "learning_rate": 3.132699819711539e-05, "loss": 0.4304, "step": 9947 }, { "epoch": 2.99, "learning_rate": 3.1325120192307695e-05, "loss": 0.3922, "step": 9948 }, { "epoch": 2.99, "learning_rate": 3.13232421875e-05, "loss": 0.4887, "step": 9949 }, { "epoch": 2.99, "learning_rate": 3.132136418269231e-05, "loss": 0.4251, "step": 9950 }, { "epoch": 2.99, "learning_rate": 3.1319486177884616e-05, "loss": 0.2983, "step": 9951 }, { "epoch": 2.99, "learning_rate": 3.131760817307692e-05, "loss": 0.3826, "step": 9952 }, { "epoch": 2.99, "learning_rate": 3.131573016826923e-05, "loss": 0.3391, "step": 9953 }, { "epoch": 2.99, "learning_rate": 3.131385216346154e-05, "loss": 0.2785, "step": 9954 }, { "epoch": 2.99, "learning_rate": 3.131197415865385e-05, "loss": 0.354, "step": 9955 }, { "epoch": 2.99, "learning_rate": 3.131009615384616e-05, "loss": 0.338, "step": 9956 }, { "epoch": 2.99, "learning_rate": 3.1308218149038465e-05, "loss": 0.3861, "step": 9957 }, { "epoch": 2.99, "learning_rate": 3.130634014423077e-05, "loss": 0.4176, "step": 9958 }, { "epoch": 2.99, "learning_rate": 3.130446213942308e-05, "loss": 0.2981, "step": 9959 }, { "epoch": 2.99, "learning_rate": 3.1302584134615386e-05, "loss": 0.3936, "step": 9960 }, { "epoch": 2.99, "learning_rate": 3.13007061298077e-05, "loss": 0.3794, "step": 9961 }, { "epoch": 2.99, "learning_rate": 3.1298828125e-05, "loss": 0.3623, "step": 9962 }, { "epoch": 2.99, "learning_rate": 3.129695012019231e-05, "loss": 0.2859, "step": 9963 }, { "epoch": 2.99, "learning_rate": 3.129507211538462e-05, "loss": 0.5755, "step": 9964 }, { "epoch": 2.99, "learning_rate": 3.129319411057692e-05, "loss": 0.4716, "step": 9965 }, { "epoch": 2.99, "learning_rate": 3.129131610576923e-05, "loss": 0.3123, "step": 9966 }, { "epoch": 2.99, "learning_rate": 3.128943810096154e-05, "loss": 0.3864, "step": 9967 }, { "epoch": 3.0, "learning_rate": 3.1287560096153844e-05, "loss": 0.3404, "step": 9968 }, { "epoch": 3.0, "learning_rate": 3.1285682091346156e-05, "loss": 0.4028, "step": 9969 }, { "epoch": 3.0, "learning_rate": 3.128380408653847e-05, "loss": 0.3952, "step": 9970 }, { "epoch": 3.0, "learning_rate": 3.128192608173077e-05, "loss": 0.3602, "step": 9971 }, { "epoch": 3.0, "learning_rate": 3.128004807692308e-05, "loss": 0.3349, "step": 9972 }, { "epoch": 3.0, "learning_rate": 3.127817007211539e-05, "loss": 0.3447, "step": 9973 }, { "epoch": 3.0, "learning_rate": 3.127629206730769e-05, "loss": 0.4759, "step": 9974 }, { "epoch": 3.0, "learning_rate": 3.1274414062500005e-05, "loss": 0.4331, "step": 9975 }, { "epoch": 3.0, "learning_rate": 3.127253605769231e-05, "loss": 0.3634, "step": 9976 }, { "epoch": 3.0, "learning_rate": 3.1270658052884614e-05, "loss": 0.4518, "step": 9977 }, { "epoch": 3.0, "learning_rate": 3.1268780048076926e-05, "loss": 0.3491, "step": 9978 }, { "epoch": 3.0, "learning_rate": 3.126690204326923e-05, "loss": 0.3458, "step": 9979 }, { "epoch": 3.0, "learning_rate": 3.1265024038461535e-05, "loss": 0.3101, "step": 9980 }, { "epoch": 3.0, "learning_rate": 3.126314603365385e-05, "loss": 0.3655, "step": 9981 }, { "epoch": 3.0, "learning_rate": 3.126126802884616e-05, "loss": 0.3308, "step": 9982 }, { "epoch": 3.0, "learning_rate": 3.125939002403846e-05, "loss": 0.3258, "step": 9983 }, { "epoch": 3.0, "learning_rate": 3.1257512019230775e-05, "loss": 0.277, "step": 9984 }, { "epoch": 3.0, "learning_rate": 3.125563401442308e-05, "loss": 0.1272, "step": 9985 }, { "epoch": 3.0, "learning_rate": 3.1253756009615384e-05, "loss": 0.1016, "step": 9986 }, { "epoch": 3.0, "learning_rate": 3.1251878004807696e-05, "loss": 0.1, "step": 9987 }, { "epoch": 3.0, "learning_rate": 3.125e-05, "loss": 0.1597, "step": 9988 }, { "epoch": 3.0, "learning_rate": 3.1248121995192305e-05, "loss": 0.117, "step": 9989 }, { "epoch": 3.0, "learning_rate": 3.124624399038462e-05, "loss": 0.1837, "step": 9990 }, { "epoch": 3.0, "learning_rate": 3.124436598557692e-05, "loss": 0.2031, "step": 9991 }, { "epoch": 3.0, "learning_rate": 3.124248798076923e-05, "loss": 0.1522, "step": 9992 }, { "epoch": 3.0, "learning_rate": 3.124060997596154e-05, "loss": 0.1075, "step": 9993 }, { "epoch": 3.0, "learning_rate": 3.123873197115384e-05, "loss": 0.2092, "step": 9994 }, { "epoch": 3.0, "learning_rate": 3.123685396634616e-05, "loss": 0.1068, "step": 9995 }, { "epoch": 3.0, "learning_rate": 3.1234975961538466e-05, "loss": 0.1024, "step": 9996 }, { "epoch": 3.0, "learning_rate": 3.123309795673077e-05, "loss": 0.0874, "step": 9997 }, { "epoch": 3.0, "learning_rate": 3.123121995192308e-05, "loss": 0.1302, "step": 9998 }, { "epoch": 3.0, "learning_rate": 3.122934194711539e-05, "loss": 0.1057, "step": 9999 }, { "epoch": 3.0, "learning_rate": 3.122746394230769e-05, "loss": 0.1106, "step": 10000 }, { "epoch": 3.01, "learning_rate": 3.12255859375e-05, "loss": 0.0935, "step": 10001 }, { "epoch": 3.01, "learning_rate": 3.122370793269231e-05, "loss": 0.103, "step": 10002 }, { "epoch": 3.01, "learning_rate": 3.122182992788461e-05, "loss": 0.0942, "step": 10003 }, { "epoch": 3.01, "learning_rate": 3.1219951923076924e-05, "loss": 0.1653, "step": 10004 }, { "epoch": 3.01, "learning_rate": 3.121807391826923e-05, "loss": 0.0929, "step": 10005 }, { "epoch": 3.01, "learning_rate": 3.121619591346154e-05, "loss": 0.1134, "step": 10006 }, { "epoch": 3.01, "learning_rate": 3.1214317908653846e-05, "loss": 0.1168, "step": 10007 }, { "epoch": 3.01, "learning_rate": 3.121243990384616e-05, "loss": 0.1471, "step": 10008 }, { "epoch": 3.01, "learning_rate": 3.121056189903846e-05, "loss": 0.1376, "step": 10009 }, { "epoch": 3.01, "learning_rate": 3.1208683894230773e-05, "loss": 0.1637, "step": 10010 }, { "epoch": 3.01, "learning_rate": 3.120680588942308e-05, "loss": 0.1481, "step": 10011 }, { "epoch": 3.01, "learning_rate": 3.120492788461539e-05, "loss": 0.1167, "step": 10012 }, { "epoch": 3.01, "learning_rate": 3.1203049879807694e-05, "loss": 0.1232, "step": 10013 }, { "epoch": 3.01, "learning_rate": 3.1201171875e-05, "loss": 0.1087, "step": 10014 }, { "epoch": 3.01, "learning_rate": 3.119929387019231e-05, "loss": 0.1751, "step": 10015 }, { "epoch": 3.01, "learning_rate": 3.1197415865384616e-05, "loss": 0.1486, "step": 10016 }, { "epoch": 3.01, "learning_rate": 3.119553786057692e-05, "loss": 0.1792, "step": 10017 }, { "epoch": 3.01, "learning_rate": 3.119365985576923e-05, "loss": 0.168, "step": 10018 }, { "epoch": 3.01, "learning_rate": 3.119178185096154e-05, "loss": 0.1721, "step": 10019 }, { "epoch": 3.01, "learning_rate": 3.118990384615385e-05, "loss": 0.1849, "step": 10020 }, { "epoch": 3.01, "learning_rate": 3.118802584134616e-05, "loss": 0.1151, "step": 10021 }, { "epoch": 3.01, "learning_rate": 3.1186147836538465e-05, "loss": 0.1168, "step": 10022 }, { "epoch": 3.01, "learning_rate": 3.118426983173077e-05, "loss": 0.1678, "step": 10023 }, { "epoch": 3.01, "learning_rate": 3.118239182692308e-05, "loss": 0.2147, "step": 10024 }, { "epoch": 3.01, "learning_rate": 3.1180513822115386e-05, "loss": 0.1921, "step": 10025 }, { "epoch": 3.01, "learning_rate": 3.11786358173077e-05, "loss": 0.125, "step": 10026 }, { "epoch": 3.01, "learning_rate": 3.11767578125e-05, "loss": 0.1314, "step": 10027 }, { "epoch": 3.01, "learning_rate": 3.117487980769231e-05, "loss": 0.1625, "step": 10028 }, { "epoch": 3.01, "learning_rate": 3.117300180288462e-05, "loss": 0.1061, "step": 10029 }, { "epoch": 3.01, "learning_rate": 3.117112379807692e-05, "loss": 0.1098, "step": 10030 }, { "epoch": 3.01, "learning_rate": 3.116924579326923e-05, "loss": 0.0738, "step": 10031 }, { "epoch": 3.01, "learning_rate": 3.116736778846154e-05, "loss": 0.1144, "step": 10032 }, { "epoch": 3.01, "learning_rate": 3.1165489783653844e-05, "loss": 0.1009, "step": 10033 }, { "epoch": 3.02, "learning_rate": 3.1163611778846156e-05, "loss": 0.1346, "step": 10034 }, { "epoch": 3.02, "learning_rate": 3.116173377403847e-05, "loss": 0.1527, "step": 10035 }, { "epoch": 3.02, "learning_rate": 3.115985576923077e-05, "loss": 0.1076, "step": 10036 }, { "epoch": 3.02, "learning_rate": 3.115797776442308e-05, "loss": 0.1105, "step": 10037 }, { "epoch": 3.02, "learning_rate": 3.115609975961539e-05, "loss": 0.0898, "step": 10038 }, { "epoch": 3.02, "learning_rate": 3.115422175480769e-05, "loss": 0.1041, "step": 10039 }, { "epoch": 3.02, "learning_rate": 3.1152343750000005e-05, "loss": 0.0968, "step": 10040 }, { "epoch": 3.02, "learning_rate": 3.115046574519231e-05, "loss": 0.1006, "step": 10041 }, { "epoch": 3.02, "learning_rate": 3.1148587740384614e-05, "loss": 0.1604, "step": 10042 }, { "epoch": 3.02, "learning_rate": 3.1146709735576926e-05, "loss": 0.0935, "step": 10043 }, { "epoch": 3.02, "learning_rate": 3.114483173076923e-05, "loss": 0.1238, "step": 10044 }, { "epoch": 3.02, "learning_rate": 3.1142953725961535e-05, "loss": 0.2056, "step": 10045 }, { "epoch": 3.02, "learning_rate": 3.114107572115385e-05, "loss": 0.1237, "step": 10046 }, { "epoch": 3.02, "learning_rate": 3.113919771634616e-05, "loss": 0.1011, "step": 10047 }, { "epoch": 3.02, "learning_rate": 3.113731971153846e-05, "loss": 0.1313, "step": 10048 }, { "epoch": 3.02, "learning_rate": 3.1135441706730775e-05, "loss": 0.1426, "step": 10049 }, { "epoch": 3.02, "learning_rate": 3.113356370192308e-05, "loss": 0.0836, "step": 10050 }, { "epoch": 3.02, "learning_rate": 3.1131685697115384e-05, "loss": 0.2106, "step": 10051 }, { "epoch": 3.02, "learning_rate": 3.1129807692307696e-05, "loss": 0.1354, "step": 10052 }, { "epoch": 3.02, "learning_rate": 3.11279296875e-05, "loss": 0.2128, "step": 10053 }, { "epoch": 3.02, "learning_rate": 3.112605168269231e-05, "loss": 0.162, "step": 10054 }, { "epoch": 3.02, "learning_rate": 3.112417367788462e-05, "loss": 0.1648, "step": 10055 }, { "epoch": 3.02, "learning_rate": 3.112229567307692e-05, "loss": 0.1203, "step": 10056 }, { "epoch": 3.02, "learning_rate": 3.112041766826923e-05, "loss": 0.1214, "step": 10057 }, { "epoch": 3.02, "learning_rate": 3.111853966346154e-05, "loss": 0.1071, "step": 10058 }, { "epoch": 3.02, "learning_rate": 3.111666165865384e-05, "loss": 0.1697, "step": 10059 }, { "epoch": 3.02, "learning_rate": 3.111478365384616e-05, "loss": 0.1499, "step": 10060 }, { "epoch": 3.02, "learning_rate": 3.1112905649038466e-05, "loss": 0.1575, "step": 10061 }, { "epoch": 3.02, "learning_rate": 3.111102764423077e-05, "loss": 0.1582, "step": 10062 }, { "epoch": 3.02, "learning_rate": 3.110914963942308e-05, "loss": 0.17, "step": 10063 }, { "epoch": 3.02, "learning_rate": 3.110727163461539e-05, "loss": 0.0997, "step": 10064 }, { "epoch": 3.02, "learning_rate": 3.110539362980769e-05, "loss": 0.1503, "step": 10065 }, { "epoch": 3.02, "learning_rate": 3.1103515625e-05, "loss": 0.14, "step": 10066 }, { "epoch": 3.02, "learning_rate": 3.110163762019231e-05, "loss": 0.1103, "step": 10067 }, { "epoch": 3.03, "learning_rate": 3.109975961538461e-05, "loss": 0.0942, "step": 10068 }, { "epoch": 3.03, "learning_rate": 3.1097881610576924e-05, "loss": 0.143, "step": 10069 }, { "epoch": 3.03, "learning_rate": 3.109600360576923e-05, "loss": 0.1354, "step": 10070 }, { "epoch": 3.03, "learning_rate": 3.109412560096154e-05, "loss": 0.1667, "step": 10071 }, { "epoch": 3.03, "learning_rate": 3.1092247596153845e-05, "loss": 0.116, "step": 10072 }, { "epoch": 3.03, "learning_rate": 3.109036959134616e-05, "loss": 0.1887, "step": 10073 }, { "epoch": 3.03, "learning_rate": 3.108849158653847e-05, "loss": 0.1958, "step": 10074 }, { "epoch": 3.03, "learning_rate": 3.108661358173077e-05, "loss": 0.1443, "step": 10075 }, { "epoch": 3.03, "learning_rate": 3.108473557692308e-05, "loss": 0.1282, "step": 10076 }, { "epoch": 3.03, "learning_rate": 3.108285757211539e-05, "loss": 0.165, "step": 10077 }, { "epoch": 3.03, "learning_rate": 3.1080979567307694e-05, "loss": 0.1091, "step": 10078 }, { "epoch": 3.03, "learning_rate": 3.10791015625e-05, "loss": 0.1668, "step": 10079 }, { "epoch": 3.03, "learning_rate": 3.107722355769231e-05, "loss": 0.1805, "step": 10080 }, { "epoch": 3.03, "learning_rate": 3.1075345552884615e-05, "loss": 0.084, "step": 10081 }, { "epoch": 3.03, "learning_rate": 3.107346754807692e-05, "loss": 0.1267, "step": 10082 }, { "epoch": 3.03, "learning_rate": 3.107158954326923e-05, "loss": 0.1107, "step": 10083 }, { "epoch": 3.03, "learning_rate": 3.1069711538461537e-05, "loss": 0.1133, "step": 10084 }, { "epoch": 3.03, "learning_rate": 3.106783353365385e-05, "loss": 0.206, "step": 10085 }, { "epoch": 3.03, "learning_rate": 3.106595552884616e-05, "loss": 0.0985, "step": 10086 }, { "epoch": 3.03, "learning_rate": 3.1064077524038464e-05, "loss": 0.1396, "step": 10087 }, { "epoch": 3.03, "learning_rate": 3.106219951923077e-05, "loss": 0.1126, "step": 10088 }, { "epoch": 3.03, "learning_rate": 3.106032151442308e-05, "loss": 0.174, "step": 10089 }, { "epoch": 3.03, "learning_rate": 3.1058443509615386e-05, "loss": 0.17, "step": 10090 }, { "epoch": 3.03, "learning_rate": 3.10565655048077e-05, "loss": 0.1881, "step": 10091 }, { "epoch": 3.03, "learning_rate": 3.10546875e-05, "loss": 0.0689, "step": 10092 }, { "epoch": 3.03, "learning_rate": 3.1052809495192307e-05, "loss": 0.1812, "step": 10093 }, { "epoch": 3.03, "learning_rate": 3.105093149038462e-05, "loss": 0.1531, "step": 10094 }, { "epoch": 3.03, "learning_rate": 3.104905348557692e-05, "loss": 0.163, "step": 10095 }, { "epoch": 3.03, "learning_rate": 3.104717548076923e-05, "loss": 0.2634, "step": 10096 }, { "epoch": 3.03, "learning_rate": 3.104529747596154e-05, "loss": 0.1439, "step": 10097 }, { "epoch": 3.03, "learning_rate": 3.1043419471153844e-05, "loss": 0.1867, "step": 10098 }, { "epoch": 3.03, "learning_rate": 3.1041541466346156e-05, "loss": 0.1155, "step": 10099 }, { "epoch": 3.03, "learning_rate": 3.103966346153847e-05, "loss": 0.1403, "step": 10100 }, { "epoch": 3.04, "learning_rate": 3.103778545673077e-05, "loss": 0.1458, "step": 10101 }, { "epoch": 3.04, "learning_rate": 3.103590745192308e-05, "loss": 0.1265, "step": 10102 }, { "epoch": 3.04, "learning_rate": 3.103402944711539e-05, "loss": 0.1179, "step": 10103 }, { "epoch": 3.04, "learning_rate": 3.103215144230769e-05, "loss": 0.1137, "step": 10104 }, { "epoch": 3.04, "learning_rate": 3.1030273437500005e-05, "loss": 0.1905, "step": 10105 }, { "epoch": 3.04, "learning_rate": 3.102839543269231e-05, "loss": 0.1798, "step": 10106 }, { "epoch": 3.04, "learning_rate": 3.1026517427884614e-05, "loss": 0.1045, "step": 10107 }, { "epoch": 3.04, "learning_rate": 3.1024639423076926e-05, "loss": 0.1561, "step": 10108 }, { "epoch": 3.04, "learning_rate": 3.102276141826923e-05, "loss": 0.1694, "step": 10109 }, { "epoch": 3.04, "learning_rate": 3.1020883413461535e-05, "loss": 0.133, "step": 10110 }, { "epoch": 3.04, "learning_rate": 3.101900540865385e-05, "loss": 0.1101, "step": 10111 }, { "epoch": 3.04, "learning_rate": 3.101712740384616e-05, "loss": 0.089, "step": 10112 }, { "epoch": 3.04, "learning_rate": 3.101524939903846e-05, "loss": 0.0897, "step": 10113 }, { "epoch": 3.04, "learning_rate": 3.1013371394230775e-05, "loss": 0.1355, "step": 10114 }, { "epoch": 3.04, "learning_rate": 3.101149338942308e-05, "loss": 0.153, "step": 10115 }, { "epoch": 3.04, "learning_rate": 3.1009615384615384e-05, "loss": 0.143, "step": 10116 }, { "epoch": 3.04, "learning_rate": 3.1007737379807696e-05, "loss": 0.1856, "step": 10117 }, { "epoch": 3.04, "learning_rate": 3.1005859375e-05, "loss": 0.1148, "step": 10118 }, { "epoch": 3.04, "learning_rate": 3.100398137019231e-05, "loss": 0.2471, "step": 10119 }, { "epoch": 3.04, "learning_rate": 3.100210336538462e-05, "loss": 0.1708, "step": 10120 }, { "epoch": 3.04, "learning_rate": 3.100022536057692e-05, "loss": 0.1974, "step": 10121 }, { "epoch": 3.04, "learning_rate": 3.099834735576923e-05, "loss": 0.1541, "step": 10122 }, { "epoch": 3.04, "learning_rate": 3.099646935096154e-05, "loss": 0.1174, "step": 10123 }, { "epoch": 3.04, "learning_rate": 3.099459134615384e-05, "loss": 0.1475, "step": 10124 }, { "epoch": 3.04, "learning_rate": 3.099271334134616e-05, "loss": 0.1061, "step": 10125 }, { "epoch": 3.04, "learning_rate": 3.0990835336538466e-05, "loss": 0.1451, "step": 10126 }, { "epoch": 3.04, "learning_rate": 3.098895733173077e-05, "loss": 0.1077, "step": 10127 }, { "epoch": 3.04, "learning_rate": 3.098707932692308e-05, "loss": 0.1385, "step": 10128 }, { "epoch": 3.04, "learning_rate": 3.098520132211539e-05, "loss": 0.1201, "step": 10129 }, { "epoch": 3.04, "learning_rate": 3.098332331730769e-05, "loss": 0.1737, "step": 10130 }, { "epoch": 3.04, "learning_rate": 3.09814453125e-05, "loss": 0.1495, "step": 10131 }, { "epoch": 3.04, "learning_rate": 3.097956730769231e-05, "loss": 0.1289, "step": 10132 }, { "epoch": 3.04, "learning_rate": 3.097768930288461e-05, "loss": 0.1089, "step": 10133 }, { "epoch": 3.05, "learning_rate": 3.0975811298076924e-05, "loss": 0.1497, "step": 10134 }, { "epoch": 3.05, "learning_rate": 3.097393329326923e-05, "loss": 0.2122, "step": 10135 }, { "epoch": 3.05, "learning_rate": 3.097205528846154e-05, "loss": 0.1161, "step": 10136 }, { "epoch": 3.05, "learning_rate": 3.0970177283653845e-05, "loss": 0.1408, "step": 10137 }, { "epoch": 3.05, "learning_rate": 3.096829927884616e-05, "loss": 0.1469, "step": 10138 }, { "epoch": 3.05, "learning_rate": 3.096642127403847e-05, "loss": 0.1232, "step": 10139 }, { "epoch": 3.05, "learning_rate": 3.096454326923077e-05, "loss": 0.1325, "step": 10140 }, { "epoch": 3.05, "learning_rate": 3.096266526442308e-05, "loss": 0.1713, "step": 10141 }, { "epoch": 3.05, "learning_rate": 3.096078725961539e-05, "loss": 0.1104, "step": 10142 }, { "epoch": 3.05, "learning_rate": 3.0958909254807694e-05, "loss": 0.1006, "step": 10143 }, { "epoch": 3.05, "learning_rate": 3.095703125e-05, "loss": 0.1481, "step": 10144 }, { "epoch": 3.05, "learning_rate": 3.095515324519231e-05, "loss": 0.1006, "step": 10145 }, { "epoch": 3.05, "learning_rate": 3.0953275240384615e-05, "loss": 0.1406, "step": 10146 }, { "epoch": 3.05, "learning_rate": 3.095139723557692e-05, "loss": 0.1401, "step": 10147 }, { "epoch": 3.05, "learning_rate": 3.094951923076923e-05, "loss": 0.1328, "step": 10148 }, { "epoch": 3.05, "learning_rate": 3.0947641225961536e-05, "loss": 0.1619, "step": 10149 }, { "epoch": 3.05, "learning_rate": 3.094576322115385e-05, "loss": 0.148, "step": 10150 }, { "epoch": 3.05, "learning_rate": 3.094388521634616e-05, "loss": 0.122, "step": 10151 }, { "epoch": 3.05, "learning_rate": 3.0942007211538464e-05, "loss": 0.1228, "step": 10152 }, { "epoch": 3.05, "learning_rate": 3.094012920673077e-05, "loss": 0.1266, "step": 10153 }, { "epoch": 3.05, "learning_rate": 3.093825120192308e-05, "loss": 0.1327, "step": 10154 }, { "epoch": 3.05, "learning_rate": 3.0936373197115385e-05, "loss": 0.1084, "step": 10155 }, { "epoch": 3.05, "learning_rate": 3.09344951923077e-05, "loss": 0.1742, "step": 10156 }, { "epoch": 3.05, "learning_rate": 3.09326171875e-05, "loss": 0.1658, "step": 10157 }, { "epoch": 3.05, "learning_rate": 3.0930739182692306e-05, "loss": 0.1104, "step": 10158 }, { "epoch": 3.05, "learning_rate": 3.092886117788462e-05, "loss": 0.1371, "step": 10159 }, { "epoch": 3.05, "learning_rate": 3.092698317307692e-05, "loss": 0.1707, "step": 10160 }, { "epoch": 3.05, "learning_rate": 3.092510516826923e-05, "loss": 0.1229, "step": 10161 }, { "epoch": 3.05, "learning_rate": 3.092322716346154e-05, "loss": 0.1114, "step": 10162 }, { "epoch": 3.05, "learning_rate": 3.0921349158653844e-05, "loss": 0.1418, "step": 10163 }, { "epoch": 3.05, "learning_rate": 3.0919471153846155e-05, "loss": 0.0901, "step": 10164 }, { "epoch": 3.05, "learning_rate": 3.091759314903847e-05, "loss": 0.1106, "step": 10165 }, { "epoch": 3.05, "learning_rate": 3.091571514423077e-05, "loss": 0.1599, "step": 10166 }, { "epoch": 3.05, "learning_rate": 3.0913837139423077e-05, "loss": 0.159, "step": 10167 }, { "epoch": 3.06, "learning_rate": 3.091195913461539e-05, "loss": 0.1206, "step": 10168 }, { "epoch": 3.06, "learning_rate": 3.091008112980769e-05, "loss": 0.1299, "step": 10169 }, { "epoch": 3.06, "learning_rate": 3.0908203125000004e-05, "loss": 0.1467, "step": 10170 }, { "epoch": 3.06, "learning_rate": 3.090632512019231e-05, "loss": 0.1697, "step": 10171 }, { "epoch": 3.06, "learning_rate": 3.0904447115384614e-05, "loss": 0.1402, "step": 10172 }, { "epoch": 3.06, "learning_rate": 3.0902569110576926e-05, "loss": 0.1485, "step": 10173 }, { "epoch": 3.06, "learning_rate": 3.090069110576923e-05, "loss": 0.1808, "step": 10174 }, { "epoch": 3.06, "learning_rate": 3.0898813100961535e-05, "loss": 0.1331, "step": 10175 }, { "epoch": 3.06, "learning_rate": 3.0896935096153847e-05, "loss": 0.1181, "step": 10176 }, { "epoch": 3.06, "learning_rate": 3.089505709134616e-05, "loss": 0.126, "step": 10177 }, { "epoch": 3.06, "learning_rate": 3.089317908653846e-05, "loss": 0.1661, "step": 10178 }, { "epoch": 3.06, "learning_rate": 3.0891301081730774e-05, "loss": 0.0803, "step": 10179 }, { "epoch": 3.06, "learning_rate": 3.088942307692308e-05, "loss": 0.1297, "step": 10180 }, { "epoch": 3.06, "learning_rate": 3.0887545072115384e-05, "loss": 0.1259, "step": 10181 }, { "epoch": 3.06, "learning_rate": 3.0885667067307696e-05, "loss": 0.1051, "step": 10182 }, { "epoch": 3.06, "learning_rate": 3.08837890625e-05, "loss": 0.0873, "step": 10183 }, { "epoch": 3.06, "learning_rate": 3.088191105769231e-05, "loss": 0.147, "step": 10184 }, { "epoch": 3.06, "learning_rate": 3.088003305288462e-05, "loss": 0.1041, "step": 10185 }, { "epoch": 3.06, "learning_rate": 3.087815504807692e-05, "loss": 0.0766, "step": 10186 }, { "epoch": 3.06, "learning_rate": 3.087627704326923e-05, "loss": 0.1775, "step": 10187 }, { "epoch": 3.06, "learning_rate": 3.087439903846154e-05, "loss": 0.1643, "step": 10188 }, { "epoch": 3.06, "learning_rate": 3.087252103365384e-05, "loss": 0.1876, "step": 10189 }, { "epoch": 3.06, "learning_rate": 3.087064302884616e-05, "loss": 0.1015, "step": 10190 }, { "epoch": 3.06, "learning_rate": 3.0868765024038466e-05, "loss": 0.1708, "step": 10191 }, { "epoch": 3.06, "learning_rate": 3.086688701923077e-05, "loss": 0.0912, "step": 10192 }, { "epoch": 3.06, "learning_rate": 3.086500901442308e-05, "loss": 0.165, "step": 10193 }, { "epoch": 3.06, "learning_rate": 3.086313100961539e-05, "loss": 0.1546, "step": 10194 }, { "epoch": 3.06, "learning_rate": 3.086125300480769e-05, "loss": 0.1706, "step": 10195 }, { "epoch": 3.06, "learning_rate": 3.0859375e-05, "loss": 0.1358, "step": 10196 }, { "epoch": 3.06, "learning_rate": 3.085749699519231e-05, "loss": 0.1566, "step": 10197 }, { "epoch": 3.06, "learning_rate": 3.085561899038462e-05, "loss": 0.1638, "step": 10198 }, { "epoch": 3.06, "learning_rate": 3.0853740985576924e-05, "loss": 0.0926, "step": 10199 }, { "epoch": 3.06, "learning_rate": 3.085186298076923e-05, "loss": 0.1022, "step": 10200 }, { "epoch": 3.07, "learning_rate": 3.084998497596154e-05, "loss": 0.1352, "step": 10201 }, { "epoch": 3.07, "learning_rate": 3.0848106971153845e-05, "loss": 0.1339, "step": 10202 }, { "epoch": 3.07, "learning_rate": 3.084622896634616e-05, "loss": 0.0966, "step": 10203 }, { "epoch": 3.07, "learning_rate": 3.084435096153847e-05, "loss": 0.109, "step": 10204 }, { "epoch": 3.07, "learning_rate": 3.084247295673077e-05, "loss": 0.1364, "step": 10205 }, { "epoch": 3.07, "learning_rate": 3.084059495192308e-05, "loss": 0.1759, "step": 10206 }, { "epoch": 3.07, "learning_rate": 3.083871694711539e-05, "loss": 0.1244, "step": 10207 }, { "epoch": 3.07, "learning_rate": 3.0836838942307694e-05, "loss": 0.1078, "step": 10208 }, { "epoch": 3.07, "learning_rate": 3.08349609375e-05, "loss": 0.1492, "step": 10209 }, { "epoch": 3.07, "learning_rate": 3.083308293269231e-05, "loss": 0.1202, "step": 10210 }, { "epoch": 3.07, "learning_rate": 3.0831204927884615e-05, "loss": 0.1401, "step": 10211 }, { "epoch": 3.07, "learning_rate": 3.082932692307692e-05, "loss": 0.0932, "step": 10212 }, { "epoch": 3.07, "learning_rate": 3.082744891826923e-05, "loss": 0.17, "step": 10213 }, { "epoch": 3.07, "learning_rate": 3.0825570913461536e-05, "loss": 0.1147, "step": 10214 }, { "epoch": 3.07, "learning_rate": 3.082369290865385e-05, "loss": 0.1776, "step": 10215 }, { "epoch": 3.07, "learning_rate": 3.082181490384616e-05, "loss": 0.2109, "step": 10216 }, { "epoch": 3.07, "learning_rate": 3.0819936899038464e-05, "loss": 0.134, "step": 10217 }, { "epoch": 3.07, "learning_rate": 3.0818058894230776e-05, "loss": 0.1832, "step": 10218 }, { "epoch": 3.07, "learning_rate": 3.081618088942308e-05, "loss": 0.3162, "step": 10219 }, { "epoch": 3.07, "learning_rate": 3.0814302884615385e-05, "loss": 0.2082, "step": 10220 }, { "epoch": 3.07, "learning_rate": 3.08124248798077e-05, "loss": 0.1233, "step": 10221 }, { "epoch": 3.07, "learning_rate": 3.0810546875e-05, "loss": 0.302, "step": 10222 }, { "epoch": 3.07, "learning_rate": 3.0808668870192306e-05, "loss": 0.1954, "step": 10223 }, { "epoch": 3.07, "learning_rate": 3.080679086538462e-05, "loss": 0.1252, "step": 10224 }, { "epoch": 3.07, "learning_rate": 3.080491286057692e-05, "loss": 0.1607, "step": 10225 }, { "epoch": 3.07, "learning_rate": 3.080303485576923e-05, "loss": 0.1468, "step": 10226 }, { "epoch": 3.07, "learning_rate": 3.080115685096154e-05, "loss": 0.1039, "step": 10227 }, { "epoch": 3.07, "learning_rate": 3.0799278846153844e-05, "loss": 0.138, "step": 10228 }, { "epoch": 3.07, "learning_rate": 3.0797400841346155e-05, "loss": 0.2057, "step": 10229 }, { "epoch": 3.07, "learning_rate": 3.079552283653847e-05, "loss": 0.1409, "step": 10230 }, { "epoch": 3.07, "learning_rate": 3.079364483173077e-05, "loss": 0.1454, "step": 10231 }, { "epoch": 3.07, "learning_rate": 3.0791766826923076e-05, "loss": 0.1245, "step": 10232 }, { "epoch": 3.07, "learning_rate": 3.078988882211539e-05, "loss": 0.1583, "step": 10233 }, { "epoch": 3.08, "learning_rate": 3.078801081730769e-05, "loss": 0.1646, "step": 10234 }, { "epoch": 3.08, "learning_rate": 3.0786132812500004e-05, "loss": 0.1541, "step": 10235 }, { "epoch": 3.08, "learning_rate": 3.078425480769231e-05, "loss": 0.1109, "step": 10236 }, { "epoch": 3.08, "learning_rate": 3.0782376802884614e-05, "loss": 0.1025, "step": 10237 }, { "epoch": 3.08, "learning_rate": 3.0780498798076925e-05, "loss": 0.1007, "step": 10238 }, { "epoch": 3.08, "learning_rate": 3.077862079326923e-05, "loss": 0.1524, "step": 10239 }, { "epoch": 3.08, "learning_rate": 3.0776742788461535e-05, "loss": 0.1421, "step": 10240 }, { "epoch": 3.08, "learning_rate": 3.0774864783653846e-05, "loss": 0.1193, "step": 10241 }, { "epoch": 3.08, "learning_rate": 3.077298677884616e-05, "loss": 0.1577, "step": 10242 }, { "epoch": 3.08, "learning_rate": 3.077110877403846e-05, "loss": 0.1303, "step": 10243 }, { "epoch": 3.08, "learning_rate": 3.0769230769230774e-05, "loss": 0.1453, "step": 10244 }, { "epoch": 3.08, "learning_rate": 3.076735276442308e-05, "loss": 0.0907, "step": 10245 }, { "epoch": 3.08, "learning_rate": 3.0765474759615384e-05, "loss": 0.1422, "step": 10246 }, { "epoch": 3.08, "learning_rate": 3.0763596754807695e-05, "loss": 0.1288, "step": 10247 }, { "epoch": 3.08, "learning_rate": 3.076171875e-05, "loss": 0.2493, "step": 10248 }, { "epoch": 3.08, "learning_rate": 3.075984074519231e-05, "loss": 0.094, "step": 10249 }, { "epoch": 3.08, "learning_rate": 3.0757962740384617e-05, "loss": 0.1869, "step": 10250 }, { "epoch": 3.08, "learning_rate": 3.075608473557692e-05, "loss": 0.2065, "step": 10251 }, { "epoch": 3.08, "learning_rate": 3.075420673076923e-05, "loss": 0.1381, "step": 10252 }, { "epoch": 3.08, "learning_rate": 3.075232872596154e-05, "loss": 0.1845, "step": 10253 }, { "epoch": 3.08, "learning_rate": 3.075045072115384e-05, "loss": 0.1399, "step": 10254 }, { "epoch": 3.08, "learning_rate": 3.074857271634616e-05, "loss": 0.135, "step": 10255 }, { "epoch": 3.08, "learning_rate": 3.0746694711538465e-05, "loss": 0.1253, "step": 10256 }, { "epoch": 3.08, "learning_rate": 3.074481670673077e-05, "loss": 0.1295, "step": 10257 }, { "epoch": 3.08, "learning_rate": 3.074293870192308e-05, "loss": 0.1656, "step": 10258 }, { "epoch": 3.08, "learning_rate": 3.0741060697115387e-05, "loss": 0.195, "step": 10259 }, { "epoch": 3.08, "learning_rate": 3.073918269230769e-05, "loss": 0.1322, "step": 10260 }, { "epoch": 3.08, "learning_rate": 3.07373046875e-05, "loss": 0.1439, "step": 10261 }, { "epoch": 3.08, "learning_rate": 3.073542668269231e-05, "loss": 0.1319, "step": 10262 }, { "epoch": 3.08, "learning_rate": 3.073354867788462e-05, "loss": 0.1376, "step": 10263 }, { "epoch": 3.08, "learning_rate": 3.0731670673076924e-05, "loss": 0.1531, "step": 10264 }, { "epoch": 3.08, "learning_rate": 3.072979266826923e-05, "loss": 0.1219, "step": 10265 }, { "epoch": 3.08, "learning_rate": 3.072791466346154e-05, "loss": 0.1078, "step": 10266 }, { "epoch": 3.09, "learning_rate": 3.0726036658653845e-05, "loss": 0.1359, "step": 10267 }, { "epoch": 3.09, "learning_rate": 3.072415865384616e-05, "loss": 0.1501, "step": 10268 }, { "epoch": 3.09, "learning_rate": 3.072228064903847e-05, "loss": 0.1373, "step": 10269 }, { "epoch": 3.09, "learning_rate": 3.072040264423077e-05, "loss": 0.1702, "step": 10270 }, { "epoch": 3.09, "learning_rate": 3.071852463942308e-05, "loss": 0.2502, "step": 10271 }, { "epoch": 3.09, "learning_rate": 3.071664663461539e-05, "loss": 0.1262, "step": 10272 }, { "epoch": 3.09, "learning_rate": 3.0714768629807694e-05, "loss": 0.1013, "step": 10273 }, { "epoch": 3.09, "learning_rate": 3.0712890625e-05, "loss": 0.1089, "step": 10274 }, { "epoch": 3.09, "learning_rate": 3.071101262019231e-05, "loss": 0.1244, "step": 10275 }, { "epoch": 3.09, "learning_rate": 3.0709134615384615e-05, "loss": 0.1235, "step": 10276 }, { "epoch": 3.09, "learning_rate": 3.070725661057692e-05, "loss": 0.1639, "step": 10277 }, { "epoch": 3.09, "learning_rate": 3.070537860576923e-05, "loss": 0.1337, "step": 10278 }, { "epoch": 3.09, "learning_rate": 3.0703500600961536e-05, "loss": 0.1345, "step": 10279 }, { "epoch": 3.09, "learning_rate": 3.070162259615385e-05, "loss": 0.1103, "step": 10280 }, { "epoch": 3.09, "learning_rate": 3.069974459134616e-05, "loss": 0.1335, "step": 10281 }, { "epoch": 3.09, "learning_rate": 3.0697866586538464e-05, "loss": 0.1621, "step": 10282 }, { "epoch": 3.09, "learning_rate": 3.0695988581730776e-05, "loss": 0.0685, "step": 10283 }, { "epoch": 3.09, "learning_rate": 3.069411057692308e-05, "loss": 0.1293, "step": 10284 }, { "epoch": 3.09, "learning_rate": 3.0692232572115385e-05, "loss": 0.1285, "step": 10285 }, { "epoch": 3.09, "learning_rate": 3.06903545673077e-05, "loss": 0.1164, "step": 10286 }, { "epoch": 3.09, "learning_rate": 3.06884765625e-05, "loss": 0.1663, "step": 10287 }, { "epoch": 3.09, "learning_rate": 3.0686598557692306e-05, "loss": 0.136, "step": 10288 }, { "epoch": 3.09, "learning_rate": 3.068472055288462e-05, "loss": 0.1372, "step": 10289 }, { "epoch": 3.09, "learning_rate": 3.068284254807692e-05, "loss": 0.1474, "step": 10290 }, { "epoch": 3.09, "learning_rate": 3.068096454326923e-05, "loss": 0.1321, "step": 10291 }, { "epoch": 3.09, "learning_rate": 3.067908653846154e-05, "loss": 0.1775, "step": 10292 }, { "epoch": 3.09, "learning_rate": 3.0677208533653844e-05, "loss": 0.1364, "step": 10293 }, { "epoch": 3.09, "learning_rate": 3.0675330528846155e-05, "loss": 0.1255, "step": 10294 }, { "epoch": 3.09, "learning_rate": 3.067345252403847e-05, "loss": 0.1766, "step": 10295 }, { "epoch": 3.09, "learning_rate": 3.067157451923077e-05, "loss": 0.1953, "step": 10296 }, { "epoch": 3.09, "learning_rate": 3.0669696514423076e-05, "loss": 0.0832, "step": 10297 }, { "epoch": 3.09, "learning_rate": 3.066781850961539e-05, "loss": 0.1516, "step": 10298 }, { "epoch": 3.09, "learning_rate": 3.066594050480769e-05, "loss": 0.2038, "step": 10299 }, { "epoch": 3.09, "learning_rate": 3.0664062500000004e-05, "loss": 0.1505, "step": 10300 }, { "epoch": 3.1, "learning_rate": 3.066218449519231e-05, "loss": 0.1868, "step": 10301 }, { "epoch": 3.1, "learning_rate": 3.0660306490384614e-05, "loss": 0.1225, "step": 10302 }, { "epoch": 3.1, "learning_rate": 3.0658428485576925e-05, "loss": 0.1327, "step": 10303 }, { "epoch": 3.1, "learning_rate": 3.065655048076923e-05, "loss": 0.1323, "step": 10304 }, { "epoch": 3.1, "learning_rate": 3.0654672475961535e-05, "loss": 0.2125, "step": 10305 }, { "epoch": 3.1, "learning_rate": 3.0652794471153846e-05, "loss": 0.0928, "step": 10306 }, { "epoch": 3.1, "learning_rate": 3.065091646634616e-05, "loss": 0.1313, "step": 10307 }, { "epoch": 3.1, "learning_rate": 3.064903846153846e-05, "loss": 0.1916, "step": 10308 }, { "epoch": 3.1, "learning_rate": 3.0647160456730774e-05, "loss": 0.2295, "step": 10309 }, { "epoch": 3.1, "learning_rate": 3.064528245192308e-05, "loss": 0.1165, "step": 10310 }, { "epoch": 3.1, "learning_rate": 3.0643404447115384e-05, "loss": 0.1352, "step": 10311 }, { "epoch": 3.1, "learning_rate": 3.0641526442307695e-05, "loss": 0.1486, "step": 10312 }, { "epoch": 3.1, "learning_rate": 3.06396484375e-05, "loss": 0.1318, "step": 10313 }, { "epoch": 3.1, "learning_rate": 3.063777043269231e-05, "loss": 0.3559, "step": 10314 }, { "epoch": 3.1, "learning_rate": 3.0635892427884616e-05, "loss": 0.0688, "step": 10315 }, { "epoch": 3.1, "learning_rate": 3.063401442307692e-05, "loss": 0.1758, "step": 10316 }, { "epoch": 3.1, "learning_rate": 3.063213641826923e-05, "loss": 0.1282, "step": 10317 }, { "epoch": 3.1, "learning_rate": 3.063025841346154e-05, "loss": 0.1206, "step": 10318 }, { "epoch": 3.1, "learning_rate": 3.062838040865384e-05, "loss": 0.237, "step": 10319 }, { "epoch": 3.1, "learning_rate": 3.062650240384616e-05, "loss": 0.1751, "step": 10320 }, { "epoch": 3.1, "learning_rate": 3.0624624399038465e-05, "loss": 0.1678, "step": 10321 }, { "epoch": 3.1, "learning_rate": 3.062274639423077e-05, "loss": 0.1643, "step": 10322 }, { "epoch": 3.1, "learning_rate": 3.062086838942308e-05, "loss": 0.1137, "step": 10323 }, { "epoch": 3.1, "learning_rate": 3.0618990384615386e-05, "loss": 0.1112, "step": 10324 }, { "epoch": 3.1, "learning_rate": 3.061711237980769e-05, "loss": 0.1747, "step": 10325 }, { "epoch": 3.1, "learning_rate": 3.0615234375e-05, "loss": 0.1275, "step": 10326 }, { "epoch": 3.1, "learning_rate": 3.061335637019231e-05, "loss": 0.2238, "step": 10327 }, { "epoch": 3.1, "learning_rate": 3.061147836538462e-05, "loss": 0.1973, "step": 10328 }, { "epoch": 3.1, "learning_rate": 3.0609600360576924e-05, "loss": 0.1583, "step": 10329 }, { "epoch": 3.1, "learning_rate": 3.060772235576923e-05, "loss": 0.1664, "step": 10330 }, { "epoch": 3.1, "learning_rate": 3.060584435096154e-05, "loss": 0.1313, "step": 10331 }, { "epoch": 3.1, "learning_rate": 3.0603966346153845e-05, "loss": 0.1225, "step": 10332 }, { "epoch": 3.1, "learning_rate": 3.0602088341346157e-05, "loss": 0.1329, "step": 10333 }, { "epoch": 3.11, "learning_rate": 3.060021033653847e-05, "loss": 0.1856, "step": 10334 }, { "epoch": 3.11, "learning_rate": 3.059833233173077e-05, "loss": 0.1062, "step": 10335 }, { "epoch": 3.11, "learning_rate": 3.059645432692308e-05, "loss": 0.1204, "step": 10336 }, { "epoch": 3.11, "learning_rate": 3.059457632211539e-05, "loss": 0.2283, "step": 10337 }, { "epoch": 3.11, "learning_rate": 3.0592698317307694e-05, "loss": 0.1465, "step": 10338 }, { "epoch": 3.11, "learning_rate": 3.05908203125e-05, "loss": 0.154, "step": 10339 }, { "epoch": 3.11, "learning_rate": 3.058894230769231e-05, "loss": 0.1451, "step": 10340 }, { "epoch": 3.11, "learning_rate": 3.0587064302884615e-05, "loss": 0.1183, "step": 10341 }, { "epoch": 3.11, "learning_rate": 3.0585186298076927e-05, "loss": 0.1506, "step": 10342 }, { "epoch": 3.11, "learning_rate": 3.058330829326923e-05, "loss": 0.1178, "step": 10343 }, { "epoch": 3.11, "learning_rate": 3.0581430288461536e-05, "loss": 0.1233, "step": 10344 }, { "epoch": 3.11, "learning_rate": 3.057955228365385e-05, "loss": 0.1414, "step": 10345 }, { "epoch": 3.11, "learning_rate": 3.057767427884616e-05, "loss": 0.1396, "step": 10346 }, { "epoch": 3.11, "learning_rate": 3.0575796274038464e-05, "loss": 0.0961, "step": 10347 }, { "epoch": 3.11, "learning_rate": 3.0573918269230776e-05, "loss": 0.1259, "step": 10348 }, { "epoch": 3.11, "learning_rate": 3.057204026442308e-05, "loss": 0.1755, "step": 10349 }, { "epoch": 3.11, "learning_rate": 3.0570162259615385e-05, "loss": 0.1581, "step": 10350 }, { "epoch": 3.11, "learning_rate": 3.05682842548077e-05, "loss": 0.1196, "step": 10351 }, { "epoch": 3.11, "learning_rate": 3.056640625e-05, "loss": 0.0893, "step": 10352 }, { "epoch": 3.11, "learning_rate": 3.0564528245192306e-05, "loss": 0.1361, "step": 10353 }, { "epoch": 3.11, "learning_rate": 3.056265024038462e-05, "loss": 0.1659, "step": 10354 }, { "epoch": 3.11, "learning_rate": 3.056077223557692e-05, "loss": 0.1463, "step": 10355 }, { "epoch": 3.11, "learning_rate": 3.055889423076923e-05, "loss": 0.1581, "step": 10356 }, { "epoch": 3.11, "learning_rate": 3.055701622596154e-05, "loss": 0.1323, "step": 10357 }, { "epoch": 3.11, "learning_rate": 3.0555138221153844e-05, "loss": 0.1495, "step": 10358 }, { "epoch": 3.11, "learning_rate": 3.0553260216346155e-05, "loss": 0.1973, "step": 10359 }, { "epoch": 3.11, "learning_rate": 3.055138221153847e-05, "loss": 0.1163, "step": 10360 }, { "epoch": 3.11, "learning_rate": 3.054950420673077e-05, "loss": 0.1413, "step": 10361 }, { "epoch": 3.11, "learning_rate": 3.054762620192308e-05, "loss": 0.1967, "step": 10362 }, { "epoch": 3.11, "learning_rate": 3.054574819711539e-05, "loss": 0.106, "step": 10363 }, { "epoch": 3.11, "learning_rate": 3.054387019230769e-05, "loss": 0.0905, "step": 10364 }, { "epoch": 3.11, "learning_rate": 3.0541992187500004e-05, "loss": 0.1478, "step": 10365 }, { "epoch": 3.11, "learning_rate": 3.054011418269231e-05, "loss": 0.1577, "step": 10366 }, { "epoch": 3.12, "learning_rate": 3.0538236177884614e-05, "loss": 0.1491, "step": 10367 }, { "epoch": 3.12, "learning_rate": 3.0536358173076925e-05, "loss": 0.1182, "step": 10368 }, { "epoch": 3.12, "learning_rate": 3.053448016826923e-05, "loss": 0.1498, "step": 10369 }, { "epoch": 3.12, "learning_rate": 3.0532602163461535e-05, "loss": 0.119, "step": 10370 }, { "epoch": 3.12, "learning_rate": 3.0530724158653846e-05, "loss": 0.0938, "step": 10371 }, { "epoch": 3.12, "learning_rate": 3.052884615384616e-05, "loss": 0.1262, "step": 10372 }, { "epoch": 3.12, "learning_rate": 3.052696814903846e-05, "loss": 0.1582, "step": 10373 }, { "epoch": 3.12, "learning_rate": 3.0525090144230774e-05, "loss": 0.0928, "step": 10374 }, { "epoch": 3.12, "learning_rate": 3.052321213942308e-05, "loss": 0.1134, "step": 10375 }, { "epoch": 3.12, "learning_rate": 3.0521334134615384e-05, "loss": 0.151, "step": 10376 }, { "epoch": 3.12, "learning_rate": 3.0519456129807695e-05, "loss": 0.1631, "step": 10377 }, { "epoch": 3.12, "learning_rate": 3.0517578125e-05, "loss": 0.1644, "step": 10378 }, { "epoch": 3.12, "learning_rate": 3.0515700120192308e-05, "loss": 0.1496, "step": 10379 }, { "epoch": 3.12, "learning_rate": 3.0513822115384616e-05, "loss": 0.1107, "step": 10380 }, { "epoch": 3.12, "learning_rate": 3.0511944110576924e-05, "loss": 0.1326, "step": 10381 }, { "epoch": 3.12, "learning_rate": 3.051006610576923e-05, "loss": 0.1767, "step": 10382 }, { "epoch": 3.12, "learning_rate": 3.0508188100961537e-05, "loss": 0.1263, "step": 10383 }, { "epoch": 3.12, "learning_rate": 3.0506310096153846e-05, "loss": 0.1744, "step": 10384 }, { "epoch": 3.12, "learning_rate": 3.0504432091346157e-05, "loss": 0.119, "step": 10385 }, { "epoch": 3.12, "learning_rate": 3.0502554086538465e-05, "loss": 0.1016, "step": 10386 }, { "epoch": 3.12, "learning_rate": 3.0500676081730773e-05, "loss": 0.1322, "step": 10387 }, { "epoch": 3.12, "learning_rate": 3.0498798076923078e-05, "loss": 0.1247, "step": 10388 }, { "epoch": 3.12, "learning_rate": 3.0496920072115386e-05, "loss": 0.1521, "step": 10389 }, { "epoch": 3.12, "learning_rate": 3.0495042067307695e-05, "loss": 0.1201, "step": 10390 }, { "epoch": 3.12, "learning_rate": 3.0493164062500003e-05, "loss": 0.2252, "step": 10391 }, { "epoch": 3.12, "learning_rate": 3.0491286057692307e-05, "loss": 0.1753, "step": 10392 }, { "epoch": 3.12, "learning_rate": 3.0489408052884616e-05, "loss": 0.1421, "step": 10393 }, { "epoch": 3.12, "learning_rate": 3.0487530048076924e-05, "loss": 0.1431, "step": 10394 }, { "epoch": 3.12, "learning_rate": 3.0485652043269232e-05, "loss": 0.15, "step": 10395 }, { "epoch": 3.12, "learning_rate": 3.0483774038461537e-05, "loss": 0.1069, "step": 10396 }, { "epoch": 3.12, "learning_rate": 3.0481896033653845e-05, "loss": 0.1169, "step": 10397 }, { "epoch": 3.12, "learning_rate": 3.0480018028846156e-05, "loss": 0.1487, "step": 10398 }, { "epoch": 3.12, "learning_rate": 3.0478140024038465e-05, "loss": 0.1233, "step": 10399 }, { "epoch": 3.12, "learning_rate": 3.0476262019230773e-05, "loss": 0.2451, "step": 10400 }, { "epoch": 3.13, "learning_rate": 3.047438401442308e-05, "loss": 0.167, "step": 10401 }, { "epoch": 3.13, "learning_rate": 3.0472506009615386e-05, "loss": 0.1596, "step": 10402 }, { "epoch": 3.13, "learning_rate": 3.0470628004807694e-05, "loss": 0.1235, "step": 10403 }, { "epoch": 3.13, "learning_rate": 3.0468750000000002e-05, "loss": 0.1582, "step": 10404 }, { "epoch": 3.13, "learning_rate": 3.046687199519231e-05, "loss": 0.1185, "step": 10405 }, { "epoch": 3.13, "learning_rate": 3.0464993990384615e-05, "loss": 0.1694, "step": 10406 }, { "epoch": 3.13, "learning_rate": 3.0463115985576923e-05, "loss": 0.1282, "step": 10407 }, { "epoch": 3.13, "learning_rate": 3.046123798076923e-05, "loss": 0.1154, "step": 10408 }, { "epoch": 3.13, "learning_rate": 3.045935997596154e-05, "loss": 0.1027, "step": 10409 }, { "epoch": 3.13, "learning_rate": 3.0457481971153844e-05, "loss": 0.1103, "step": 10410 }, { "epoch": 3.13, "learning_rate": 3.045560396634616e-05, "loss": 0.1495, "step": 10411 }, { "epoch": 3.13, "learning_rate": 3.0453725961538464e-05, "loss": 0.1236, "step": 10412 }, { "epoch": 3.13, "learning_rate": 3.0451847956730772e-05, "loss": 0.1558, "step": 10413 }, { "epoch": 3.13, "learning_rate": 3.044996995192308e-05, "loss": 0.1684, "step": 10414 }, { "epoch": 3.13, "learning_rate": 3.044809194711539e-05, "loss": 0.114, "step": 10415 }, { "epoch": 3.13, "learning_rate": 3.0446213942307693e-05, "loss": 0.1442, "step": 10416 }, { "epoch": 3.13, "learning_rate": 3.04443359375e-05, "loss": 0.1156, "step": 10417 }, { "epoch": 3.13, "learning_rate": 3.044245793269231e-05, "loss": 0.1673, "step": 10418 }, { "epoch": 3.13, "learning_rate": 3.0440579927884618e-05, "loss": 0.1207, "step": 10419 }, { "epoch": 3.13, "learning_rate": 3.0438701923076922e-05, "loss": 0.1492, "step": 10420 }, { "epoch": 3.13, "learning_rate": 3.043682391826923e-05, "loss": 0.2777, "step": 10421 }, { "epoch": 3.13, "learning_rate": 3.043494591346154e-05, "loss": 0.1254, "step": 10422 }, { "epoch": 3.13, "learning_rate": 3.0433067908653843e-05, "loss": 0.157, "step": 10423 }, { "epoch": 3.13, "learning_rate": 3.043118990384616e-05, "loss": 0.1272, "step": 10424 }, { "epoch": 3.13, "learning_rate": 3.0429311899038467e-05, "loss": 0.1211, "step": 10425 }, { "epoch": 3.13, "learning_rate": 3.042743389423077e-05, "loss": 0.3342, "step": 10426 }, { "epoch": 3.13, "learning_rate": 3.042555588942308e-05, "loss": 0.1255, "step": 10427 }, { "epoch": 3.13, "learning_rate": 3.0423677884615388e-05, "loss": 0.1369, "step": 10428 }, { "epoch": 3.13, "learning_rate": 3.0421799879807696e-05, "loss": 0.2411, "step": 10429 }, { "epoch": 3.13, "learning_rate": 3.0419921875e-05, "loss": 0.1178, "step": 10430 }, { "epoch": 3.13, "learning_rate": 3.041804387019231e-05, "loss": 0.1659, "step": 10431 }, { "epoch": 3.13, "learning_rate": 3.0416165865384617e-05, "loss": 0.1242, "step": 10432 }, { "epoch": 3.13, "learning_rate": 3.041428786057692e-05, "loss": 0.2283, "step": 10433 }, { "epoch": 3.14, "learning_rate": 3.041240985576923e-05, "loss": 0.1504, "step": 10434 }, { "epoch": 3.14, "learning_rate": 3.0410531850961538e-05, "loss": 0.1604, "step": 10435 }, { "epoch": 3.14, "learning_rate": 3.0408653846153846e-05, "loss": 0.1241, "step": 10436 }, { "epoch": 3.14, "learning_rate": 3.0406775841346158e-05, "loss": 0.1663, "step": 10437 }, { "epoch": 3.14, "learning_rate": 3.0404897836538466e-05, "loss": 0.151, "step": 10438 }, { "epoch": 3.14, "learning_rate": 3.0403019831730774e-05, "loss": 0.1733, "step": 10439 }, { "epoch": 3.14, "learning_rate": 3.040114182692308e-05, "loss": 0.1597, "step": 10440 }, { "epoch": 3.14, "learning_rate": 3.0399263822115387e-05, "loss": 0.08, "step": 10441 }, { "epoch": 3.14, "learning_rate": 3.0397385817307695e-05, "loss": 0.1622, "step": 10442 }, { "epoch": 3.14, "learning_rate": 3.03955078125e-05, "loss": 0.1575, "step": 10443 }, { "epoch": 3.14, "learning_rate": 3.0393629807692308e-05, "loss": 0.1943, "step": 10444 }, { "epoch": 3.14, "learning_rate": 3.0391751802884616e-05, "loss": 0.1271, "step": 10445 }, { "epoch": 3.14, "learning_rate": 3.0389873798076924e-05, "loss": 0.1262, "step": 10446 }, { "epoch": 3.14, "learning_rate": 3.038799579326923e-05, "loss": 0.1568, "step": 10447 }, { "epoch": 3.14, "learning_rate": 3.0386117788461537e-05, "loss": 0.1131, "step": 10448 }, { "epoch": 3.14, "learning_rate": 3.0384239783653845e-05, "loss": 0.128, "step": 10449 }, { "epoch": 3.14, "learning_rate": 3.0382361778846157e-05, "loss": 0.1412, "step": 10450 }, { "epoch": 3.14, "learning_rate": 3.0380483774038465e-05, "loss": 0.0899, "step": 10451 }, { "epoch": 3.14, "learning_rate": 3.0378605769230773e-05, "loss": 0.0946, "step": 10452 }, { "epoch": 3.14, "learning_rate": 3.0376727764423078e-05, "loss": 0.0942, "step": 10453 }, { "epoch": 3.14, "learning_rate": 3.0374849759615386e-05, "loss": 0.1741, "step": 10454 }, { "epoch": 3.14, "learning_rate": 3.0372971754807694e-05, "loss": 0.1318, "step": 10455 }, { "epoch": 3.14, "learning_rate": 3.0371093750000003e-05, "loss": 0.1363, "step": 10456 }, { "epoch": 3.14, "learning_rate": 3.0369215745192307e-05, "loss": 0.1424, "step": 10457 }, { "epoch": 3.14, "learning_rate": 3.0367337740384616e-05, "loss": 0.0832, "step": 10458 }, { "epoch": 3.14, "learning_rate": 3.0365459735576924e-05, "loss": 0.1763, "step": 10459 }, { "epoch": 3.14, "learning_rate": 3.0363581730769232e-05, "loss": 0.1152, "step": 10460 }, { "epoch": 3.14, "learning_rate": 3.0361703725961537e-05, "loss": 0.1388, "step": 10461 }, { "epoch": 3.14, "learning_rate": 3.0359825721153845e-05, "loss": 0.1501, "step": 10462 }, { "epoch": 3.14, "learning_rate": 3.0357947716346156e-05, "loss": 0.0971, "step": 10463 }, { "epoch": 3.14, "learning_rate": 3.0356069711538464e-05, "loss": 0.1483, "step": 10464 }, { "epoch": 3.14, "learning_rate": 3.0354191706730773e-05, "loss": 0.1297, "step": 10465 }, { "epoch": 3.14, "learning_rate": 3.035231370192308e-05, "loss": 0.1362, "step": 10466 }, { "epoch": 3.15, "learning_rate": 3.0350435697115386e-05, "loss": 0.2097, "step": 10467 }, { "epoch": 3.15, "learning_rate": 3.0348557692307694e-05, "loss": 0.1747, "step": 10468 }, { "epoch": 3.15, "learning_rate": 3.0346679687500002e-05, "loss": 0.1403, "step": 10469 }, { "epoch": 3.15, "learning_rate": 3.034480168269231e-05, "loss": 0.0927, "step": 10470 }, { "epoch": 3.15, "learning_rate": 3.0342923677884615e-05, "loss": 0.1818, "step": 10471 }, { "epoch": 3.15, "learning_rate": 3.0341045673076923e-05, "loss": 0.1163, "step": 10472 }, { "epoch": 3.15, "learning_rate": 3.033916766826923e-05, "loss": 0.1413, "step": 10473 }, { "epoch": 3.15, "learning_rate": 3.033728966346154e-05, "loss": 0.1067, "step": 10474 }, { "epoch": 3.15, "learning_rate": 3.0335411658653844e-05, "loss": 0.104, "step": 10475 }, { "epoch": 3.15, "learning_rate": 3.033353365384616e-05, "loss": 0.1447, "step": 10476 }, { "epoch": 3.15, "learning_rate": 3.0331655649038464e-05, "loss": 0.1514, "step": 10477 }, { "epoch": 3.15, "learning_rate": 3.0329777644230772e-05, "loss": 0.1452, "step": 10478 }, { "epoch": 3.15, "learning_rate": 3.032789963942308e-05, "loss": 0.1151, "step": 10479 }, { "epoch": 3.15, "learning_rate": 3.0326021634615388e-05, "loss": 0.0867, "step": 10480 }, { "epoch": 3.15, "learning_rate": 3.0324143629807693e-05, "loss": 0.1347, "step": 10481 }, { "epoch": 3.15, "learning_rate": 3.0322265625e-05, "loss": 0.188, "step": 10482 }, { "epoch": 3.15, "learning_rate": 3.032038762019231e-05, "loss": 0.115, "step": 10483 }, { "epoch": 3.15, "learning_rate": 3.0318509615384618e-05, "loss": 0.1482, "step": 10484 }, { "epoch": 3.15, "learning_rate": 3.0316631610576922e-05, "loss": 0.1346, "step": 10485 }, { "epoch": 3.15, "learning_rate": 3.031475360576923e-05, "loss": 0.1154, "step": 10486 }, { "epoch": 3.15, "learning_rate": 3.031287560096154e-05, "loss": 0.1743, "step": 10487 }, { "epoch": 3.15, "learning_rate": 3.0310997596153843e-05, "loss": 0.173, "step": 10488 }, { "epoch": 3.15, "learning_rate": 3.0309119591346158e-05, "loss": 0.0914, "step": 10489 }, { "epoch": 3.15, "learning_rate": 3.0307241586538466e-05, "loss": 0.1361, "step": 10490 }, { "epoch": 3.15, "learning_rate": 3.030536358173077e-05, "loss": 0.1018, "step": 10491 }, { "epoch": 3.15, "learning_rate": 3.030348557692308e-05, "loss": 0.1195, "step": 10492 }, { "epoch": 3.15, "learning_rate": 3.0301607572115388e-05, "loss": 0.2395, "step": 10493 }, { "epoch": 3.15, "learning_rate": 3.0299729567307696e-05, "loss": 0.1784, "step": 10494 }, { "epoch": 3.15, "learning_rate": 3.02978515625e-05, "loss": 0.1608, "step": 10495 }, { "epoch": 3.15, "learning_rate": 3.029597355769231e-05, "loss": 0.1453, "step": 10496 }, { "epoch": 3.15, "learning_rate": 3.0294095552884617e-05, "loss": 0.2102, "step": 10497 }, { "epoch": 3.15, "learning_rate": 3.029221754807692e-05, "loss": 0.1134, "step": 10498 }, { "epoch": 3.15, "learning_rate": 3.029033954326923e-05, "loss": 0.1619, "step": 10499 }, { "epoch": 3.16, "learning_rate": 3.0288461538461538e-05, "loss": 0.1165, "step": 10500 }, { "epoch": 3.16, "learning_rate": 3.0286583533653846e-05, "loss": 0.1422, "step": 10501 }, { "epoch": 3.16, "learning_rate": 3.0284705528846158e-05, "loss": 0.2093, "step": 10502 }, { "epoch": 3.16, "learning_rate": 3.0282827524038466e-05, "loss": 0.1834, "step": 10503 }, { "epoch": 3.16, "learning_rate": 3.0280949519230774e-05, "loss": 0.1556, "step": 10504 }, { "epoch": 3.16, "learning_rate": 3.027907151442308e-05, "loss": 0.1144, "step": 10505 }, { "epoch": 3.16, "learning_rate": 3.0277193509615387e-05, "loss": 0.1718, "step": 10506 }, { "epoch": 3.16, "learning_rate": 3.0275315504807695e-05, "loss": 0.1989, "step": 10507 }, { "epoch": 3.16, "learning_rate": 3.02734375e-05, "loss": 0.1198, "step": 10508 }, { "epoch": 3.16, "learning_rate": 3.0271559495192308e-05, "loss": 0.1263, "step": 10509 }, { "epoch": 3.16, "learning_rate": 3.0269681490384616e-05, "loss": 0.1894, "step": 10510 }, { "epoch": 3.16, "learning_rate": 3.0267803485576924e-05, "loss": 0.1253, "step": 10511 }, { "epoch": 3.16, "learning_rate": 3.026592548076923e-05, "loss": 0.1111, "step": 10512 }, { "epoch": 3.16, "learning_rate": 3.0264047475961537e-05, "loss": 0.1395, "step": 10513 }, { "epoch": 3.16, "learning_rate": 3.0262169471153845e-05, "loss": 0.0954, "step": 10514 }, { "epoch": 3.16, "learning_rate": 3.0260291466346157e-05, "loss": 0.1731, "step": 10515 }, { "epoch": 3.16, "learning_rate": 3.0258413461538465e-05, "loss": 0.1352, "step": 10516 }, { "epoch": 3.16, "learning_rate": 3.0256535456730773e-05, "loss": 0.116, "step": 10517 }, { "epoch": 3.16, "learning_rate": 3.0254657451923078e-05, "loss": 0.1306, "step": 10518 }, { "epoch": 3.16, "learning_rate": 3.0252779447115386e-05, "loss": 0.2169, "step": 10519 }, { "epoch": 3.16, "learning_rate": 3.0250901442307694e-05, "loss": 0.1456, "step": 10520 }, { "epoch": 3.16, "learning_rate": 3.0249023437500002e-05, "loss": 0.1113, "step": 10521 }, { "epoch": 3.16, "learning_rate": 3.0247145432692307e-05, "loss": 0.1626, "step": 10522 }, { "epoch": 3.16, "learning_rate": 3.0245267427884615e-05, "loss": 0.1032, "step": 10523 }, { "epoch": 3.16, "learning_rate": 3.0243389423076924e-05, "loss": 0.1728, "step": 10524 }, { "epoch": 3.16, "learning_rate": 3.0241511418269232e-05, "loss": 0.1048, "step": 10525 }, { "epoch": 3.16, "learning_rate": 3.0239633413461536e-05, "loss": 0.1059, "step": 10526 }, { "epoch": 3.16, "learning_rate": 3.0237755408653845e-05, "loss": 0.0904, "step": 10527 }, { "epoch": 3.16, "learning_rate": 3.0235877403846156e-05, "loss": 0.1577, "step": 10528 }, { "epoch": 3.16, "learning_rate": 3.0233999399038464e-05, "loss": 0.1531, "step": 10529 }, { "epoch": 3.16, "learning_rate": 3.0232121394230773e-05, "loss": 0.1542, "step": 10530 }, { "epoch": 3.16, "learning_rate": 3.023024338942308e-05, "loss": 0.1432, "step": 10531 }, { "epoch": 3.16, "learning_rate": 3.0228365384615385e-05, "loss": 0.1953, "step": 10532 }, { "epoch": 3.16, "learning_rate": 3.0226487379807694e-05, "loss": 0.1357, "step": 10533 }, { "epoch": 3.17, "learning_rate": 3.0224609375000002e-05, "loss": 0.1446, "step": 10534 }, { "epoch": 3.17, "learning_rate": 3.022273137019231e-05, "loss": 0.1451, "step": 10535 }, { "epoch": 3.17, "learning_rate": 3.0220853365384615e-05, "loss": 0.1555, "step": 10536 }, { "epoch": 3.17, "learning_rate": 3.0218975360576923e-05, "loss": 0.1383, "step": 10537 }, { "epoch": 3.17, "learning_rate": 3.021709735576923e-05, "loss": 0.1301, "step": 10538 }, { "epoch": 3.17, "learning_rate": 3.021521935096154e-05, "loss": 0.0961, "step": 10539 }, { "epoch": 3.17, "learning_rate": 3.0213341346153844e-05, "loss": 0.1958, "step": 10540 }, { "epoch": 3.17, "learning_rate": 3.021146334134616e-05, "loss": 0.1866, "step": 10541 }, { "epoch": 3.17, "learning_rate": 3.0209585336538464e-05, "loss": 0.1548, "step": 10542 }, { "epoch": 3.17, "learning_rate": 3.0207707331730772e-05, "loss": 0.1338, "step": 10543 }, { "epoch": 3.17, "learning_rate": 3.020582932692308e-05, "loss": 0.1442, "step": 10544 }, { "epoch": 3.17, "learning_rate": 3.0203951322115388e-05, "loss": 0.1306, "step": 10545 }, { "epoch": 3.17, "learning_rate": 3.0202073317307693e-05, "loss": 0.113, "step": 10546 }, { "epoch": 3.17, "learning_rate": 3.02001953125e-05, "loss": 0.1858, "step": 10547 }, { "epoch": 3.17, "learning_rate": 3.019831730769231e-05, "loss": 0.1612, "step": 10548 }, { "epoch": 3.17, "learning_rate": 3.0196439302884617e-05, "loss": 0.1897, "step": 10549 }, { "epoch": 3.17, "learning_rate": 3.0194561298076922e-05, "loss": 0.1929, "step": 10550 }, { "epoch": 3.17, "learning_rate": 3.019268329326923e-05, "loss": 0.1859, "step": 10551 }, { "epoch": 3.17, "learning_rate": 3.019080528846154e-05, "loss": 0.1685, "step": 10552 }, { "epoch": 3.17, "learning_rate": 3.0188927283653847e-05, "loss": 0.1406, "step": 10553 }, { "epoch": 3.17, "learning_rate": 3.0187049278846158e-05, "loss": 0.1114, "step": 10554 }, { "epoch": 3.17, "learning_rate": 3.0185171274038466e-05, "loss": 0.0926, "step": 10555 }, { "epoch": 3.17, "learning_rate": 3.018329326923077e-05, "loss": 0.1596, "step": 10556 }, { "epoch": 3.17, "learning_rate": 3.018141526442308e-05, "loss": 0.1509, "step": 10557 }, { "epoch": 3.17, "learning_rate": 3.0179537259615387e-05, "loss": 0.1647, "step": 10558 }, { "epoch": 3.17, "learning_rate": 3.0177659254807696e-05, "loss": 0.1281, "step": 10559 }, { "epoch": 3.17, "learning_rate": 3.017578125e-05, "loss": 0.116, "step": 10560 }, { "epoch": 3.17, "learning_rate": 3.017390324519231e-05, "loss": 0.1648, "step": 10561 }, { "epoch": 3.17, "learning_rate": 3.0172025240384617e-05, "loss": 0.1981, "step": 10562 }, { "epoch": 3.17, "learning_rate": 3.0170147235576925e-05, "loss": 0.1044, "step": 10563 }, { "epoch": 3.17, "learning_rate": 3.016826923076923e-05, "loss": 0.1461, "step": 10564 }, { "epoch": 3.17, "learning_rate": 3.0166391225961538e-05, "loss": 0.2652, "step": 10565 }, { "epoch": 3.17, "learning_rate": 3.0164513221153846e-05, "loss": 0.2004, "step": 10566 }, { "epoch": 3.18, "learning_rate": 3.0162635216346157e-05, "loss": 0.1865, "step": 10567 }, { "epoch": 3.18, "learning_rate": 3.0160757211538466e-05, "loss": 0.2688, "step": 10568 }, { "epoch": 3.18, "learning_rate": 3.0158879206730774e-05, "loss": 0.1089, "step": 10569 }, { "epoch": 3.18, "learning_rate": 3.015700120192308e-05, "loss": 0.0968, "step": 10570 }, { "epoch": 3.18, "learning_rate": 3.0155123197115387e-05, "loss": 0.1339, "step": 10571 }, { "epoch": 3.18, "learning_rate": 3.0153245192307695e-05, "loss": 0.1898, "step": 10572 }, { "epoch": 3.18, "learning_rate": 3.0151367187500003e-05, "loss": 0.1015, "step": 10573 }, { "epoch": 3.18, "learning_rate": 3.0149489182692308e-05, "loss": 0.1582, "step": 10574 }, { "epoch": 3.18, "learning_rate": 3.0147611177884616e-05, "loss": 0.1021, "step": 10575 }, { "epoch": 3.18, "learning_rate": 3.0145733173076924e-05, "loss": 0.1372, "step": 10576 }, { "epoch": 3.18, "learning_rate": 3.014385516826923e-05, "loss": 0.1002, "step": 10577 }, { "epoch": 3.18, "learning_rate": 3.0141977163461537e-05, "loss": 0.1774, "step": 10578 }, { "epoch": 3.18, "learning_rate": 3.0140099158653845e-05, "loss": 0.2828, "step": 10579 }, { "epoch": 3.18, "learning_rate": 3.0138221153846157e-05, "loss": 0.1362, "step": 10580 }, { "epoch": 3.18, "learning_rate": 3.0136343149038465e-05, "loss": 0.1196, "step": 10581 }, { "epoch": 3.18, "learning_rate": 3.0134465144230773e-05, "loss": 0.132, "step": 10582 }, { "epoch": 3.18, "learning_rate": 3.013258713942308e-05, "loss": 0.2225, "step": 10583 }, { "epoch": 3.18, "learning_rate": 3.0130709134615386e-05, "loss": 0.2087, "step": 10584 }, { "epoch": 3.18, "learning_rate": 3.0128831129807694e-05, "loss": 0.1332, "step": 10585 }, { "epoch": 3.18, "learning_rate": 3.0126953125000002e-05, "loss": 0.1364, "step": 10586 }, { "epoch": 3.18, "learning_rate": 3.0125075120192307e-05, "loss": 0.1507, "step": 10587 }, { "epoch": 3.18, "learning_rate": 3.0123197115384615e-05, "loss": 0.1787, "step": 10588 }, { "epoch": 3.18, "learning_rate": 3.0121319110576923e-05, "loss": 0.1669, "step": 10589 }, { "epoch": 3.18, "learning_rate": 3.011944110576923e-05, "loss": 0.184, "step": 10590 }, { "epoch": 3.18, "learning_rate": 3.0117563100961536e-05, "loss": 0.1264, "step": 10591 }, { "epoch": 3.18, "learning_rate": 3.0115685096153845e-05, "loss": 0.1116, "step": 10592 }, { "epoch": 3.18, "learning_rate": 3.011380709134616e-05, "loss": 0.1651, "step": 10593 }, { "epoch": 3.18, "learning_rate": 3.0111929086538464e-05, "loss": 0.1886, "step": 10594 }, { "epoch": 3.18, "learning_rate": 3.0110051081730772e-05, "loss": 0.2444, "step": 10595 }, { "epoch": 3.18, "learning_rate": 3.010817307692308e-05, "loss": 0.144, "step": 10596 }, { "epoch": 3.18, "learning_rate": 3.0106295072115385e-05, "loss": 0.2231, "step": 10597 }, { "epoch": 3.18, "learning_rate": 3.0104417067307694e-05, "loss": 0.1562, "step": 10598 }, { "epoch": 3.18, "learning_rate": 3.01025390625e-05, "loss": 0.1407, "step": 10599 }, { "epoch": 3.19, "learning_rate": 3.010066105769231e-05, "loss": 0.1741, "step": 10600 }, { "epoch": 3.19, "learning_rate": 3.0098783052884615e-05, "loss": 0.1003, "step": 10601 }, { "epoch": 3.19, "learning_rate": 3.0096905048076923e-05, "loss": 0.0936, "step": 10602 }, { "epoch": 3.19, "learning_rate": 3.009502704326923e-05, "loss": 0.1806, "step": 10603 }, { "epoch": 3.19, "learning_rate": 3.009314903846154e-05, "loss": 0.1787, "step": 10604 }, { "epoch": 3.19, "learning_rate": 3.0091271033653844e-05, "loss": 0.0914, "step": 10605 }, { "epoch": 3.19, "learning_rate": 3.008939302884616e-05, "loss": 0.1386, "step": 10606 }, { "epoch": 3.19, "learning_rate": 3.0087515024038464e-05, "loss": 0.1078, "step": 10607 }, { "epoch": 3.19, "learning_rate": 3.0085637019230772e-05, "loss": 0.1331, "step": 10608 }, { "epoch": 3.19, "learning_rate": 3.008375901442308e-05, "loss": 0.1671, "step": 10609 }, { "epoch": 3.19, "learning_rate": 3.0081881009615388e-05, "loss": 0.2906, "step": 10610 }, { "epoch": 3.19, "learning_rate": 3.0080003004807693e-05, "loss": 0.1111, "step": 10611 }, { "epoch": 3.19, "learning_rate": 3.0078125e-05, "loss": 0.1712, "step": 10612 }, { "epoch": 3.19, "learning_rate": 3.007624699519231e-05, "loss": 0.1665, "step": 10613 }, { "epoch": 3.19, "learning_rate": 3.0074368990384617e-05, "loss": 0.1458, "step": 10614 }, { "epoch": 3.19, "learning_rate": 3.0072490985576922e-05, "loss": 0.1568, "step": 10615 }, { "epoch": 3.19, "learning_rate": 3.007061298076923e-05, "loss": 0.1675, "step": 10616 }, { "epoch": 3.19, "learning_rate": 3.006873497596154e-05, "loss": 0.1357, "step": 10617 }, { "epoch": 3.19, "learning_rate": 3.0066856971153847e-05, "loss": 0.1284, "step": 10618 }, { "epoch": 3.19, "learning_rate": 3.0064978966346158e-05, "loss": 0.0975, "step": 10619 }, { "epoch": 3.19, "learning_rate": 3.0063100961538466e-05, "loss": 0.1383, "step": 10620 }, { "epoch": 3.19, "learning_rate": 3.006122295673077e-05, "loss": 0.1275, "step": 10621 }, { "epoch": 3.19, "learning_rate": 3.005934495192308e-05, "loss": 0.1727, "step": 10622 }, { "epoch": 3.19, "learning_rate": 3.0057466947115387e-05, "loss": 0.1069, "step": 10623 }, { "epoch": 3.19, "learning_rate": 3.0055588942307695e-05, "loss": 0.2842, "step": 10624 }, { "epoch": 3.19, "learning_rate": 3.00537109375e-05, "loss": 0.1683, "step": 10625 }, { "epoch": 3.19, "learning_rate": 3.005183293269231e-05, "loss": 0.1533, "step": 10626 }, { "epoch": 3.19, "learning_rate": 3.0049954927884617e-05, "loss": 0.1604, "step": 10627 }, { "epoch": 3.19, "learning_rate": 3.0048076923076925e-05, "loss": 0.1129, "step": 10628 }, { "epoch": 3.19, "learning_rate": 3.004619891826923e-05, "loss": 0.1681, "step": 10629 }, { "epoch": 3.19, "learning_rate": 3.0044320913461538e-05, "loss": 0.1602, "step": 10630 }, { "epoch": 3.19, "learning_rate": 3.0042442908653846e-05, "loss": 0.1751, "step": 10631 }, { "epoch": 3.19, "learning_rate": 3.0040564903846157e-05, "loss": 0.1826, "step": 10632 }, { "epoch": 3.2, "learning_rate": 3.0038686899038466e-05, "loss": 0.2149, "step": 10633 }, { "epoch": 3.2, "learning_rate": 3.0036808894230774e-05, "loss": 0.153, "step": 10634 }, { "epoch": 3.2, "learning_rate": 3.003493088942308e-05, "loss": 0.0927, "step": 10635 }, { "epoch": 3.2, "learning_rate": 3.0033052884615387e-05, "loss": 0.1832, "step": 10636 }, { "epoch": 3.2, "learning_rate": 3.0031174879807695e-05, "loss": 0.1674, "step": 10637 }, { "epoch": 3.2, "learning_rate": 3.0029296875000003e-05, "loss": 0.1303, "step": 10638 }, { "epoch": 3.2, "learning_rate": 3.0027418870192308e-05, "loss": 0.1514, "step": 10639 }, { "epoch": 3.2, "learning_rate": 3.0025540865384616e-05, "loss": 0.141, "step": 10640 }, { "epoch": 3.2, "learning_rate": 3.0023662860576924e-05, "loss": 0.1489, "step": 10641 }, { "epoch": 3.2, "learning_rate": 3.002178485576923e-05, "loss": 0.1536, "step": 10642 }, { "epoch": 3.2, "learning_rate": 3.0019906850961537e-05, "loss": 0.1592, "step": 10643 }, { "epoch": 3.2, "learning_rate": 3.0018028846153845e-05, "loss": 0.2539, "step": 10644 }, { "epoch": 3.2, "learning_rate": 3.0016150841346157e-05, "loss": 0.1002, "step": 10645 }, { "epoch": 3.2, "learning_rate": 3.0014272836538465e-05, "loss": 0.1234, "step": 10646 }, { "epoch": 3.2, "learning_rate": 3.0012394831730773e-05, "loss": 0.1376, "step": 10647 }, { "epoch": 3.2, "learning_rate": 3.001051682692308e-05, "loss": 0.1574, "step": 10648 }, { "epoch": 3.2, "learning_rate": 3.0008638822115386e-05, "loss": 0.1309, "step": 10649 }, { "epoch": 3.2, "learning_rate": 3.0006760817307694e-05, "loss": 0.1384, "step": 10650 }, { "epoch": 3.2, "learning_rate": 3.0004882812500002e-05, "loss": 0.1745, "step": 10651 }, { "epoch": 3.2, "learning_rate": 3.0003004807692307e-05, "loss": 0.1541, "step": 10652 }, { "epoch": 3.2, "learning_rate": 3.0001126802884615e-05, "loss": 0.1109, "step": 10653 }, { "epoch": 3.2, "learning_rate": 2.9999248798076923e-05, "loss": 0.1547, "step": 10654 }, { "epoch": 3.2, "learning_rate": 2.999737079326923e-05, "loss": 0.1045, "step": 10655 }, { "epoch": 3.2, "learning_rate": 2.9995492788461536e-05, "loss": 0.0967, "step": 10656 }, { "epoch": 3.2, "learning_rate": 2.9993614783653844e-05, "loss": 0.1709, "step": 10657 }, { "epoch": 3.2, "learning_rate": 2.999173677884616e-05, "loss": 0.1662, "step": 10658 }, { "epoch": 3.2, "learning_rate": 2.9989858774038464e-05, "loss": 0.1556, "step": 10659 }, { "epoch": 3.2, "learning_rate": 2.9987980769230772e-05, "loss": 0.147, "step": 10660 }, { "epoch": 3.2, "learning_rate": 2.998610276442308e-05, "loss": 0.2172, "step": 10661 }, { "epoch": 3.2, "learning_rate": 2.9984224759615385e-05, "loss": 0.1547, "step": 10662 }, { "epoch": 3.2, "learning_rate": 2.9982346754807693e-05, "loss": 0.1305, "step": 10663 }, { "epoch": 3.2, "learning_rate": 2.998046875e-05, "loss": 0.1076, "step": 10664 }, { "epoch": 3.2, "learning_rate": 2.997859074519231e-05, "loss": 0.1198, "step": 10665 }, { "epoch": 3.2, "learning_rate": 2.9976712740384614e-05, "loss": 0.1072, "step": 10666 }, { "epoch": 3.21, "learning_rate": 2.9974834735576923e-05, "loss": 0.2068, "step": 10667 }, { "epoch": 3.21, "learning_rate": 2.997295673076923e-05, "loss": 0.1281, "step": 10668 }, { "epoch": 3.21, "learning_rate": 2.997107872596154e-05, "loss": 0.1624, "step": 10669 }, { "epoch": 3.21, "learning_rate": 2.9969200721153844e-05, "loss": 0.1695, "step": 10670 }, { "epoch": 3.21, "learning_rate": 2.996732271634616e-05, "loss": 0.1517, "step": 10671 }, { "epoch": 3.21, "learning_rate": 2.9965444711538463e-05, "loss": 0.1442, "step": 10672 }, { "epoch": 3.21, "learning_rate": 2.996356670673077e-05, "loss": 0.0973, "step": 10673 }, { "epoch": 3.21, "learning_rate": 2.996168870192308e-05, "loss": 0.1551, "step": 10674 }, { "epoch": 3.21, "learning_rate": 2.9959810697115388e-05, "loss": 0.1158, "step": 10675 }, { "epoch": 3.21, "learning_rate": 2.9957932692307693e-05, "loss": 0.1296, "step": 10676 }, { "epoch": 3.21, "learning_rate": 2.99560546875e-05, "loss": 0.164, "step": 10677 }, { "epoch": 3.21, "learning_rate": 2.995417668269231e-05, "loss": 0.171, "step": 10678 }, { "epoch": 3.21, "learning_rate": 2.9952298677884617e-05, "loss": 0.1088, "step": 10679 }, { "epoch": 3.21, "learning_rate": 2.9950420673076922e-05, "loss": 0.1911, "step": 10680 }, { "epoch": 3.21, "learning_rate": 2.994854266826923e-05, "loss": 0.1082, "step": 10681 }, { "epoch": 3.21, "learning_rate": 2.9946664663461538e-05, "loss": 0.1268, "step": 10682 }, { "epoch": 3.21, "learning_rate": 2.9944786658653846e-05, "loss": 0.1247, "step": 10683 }, { "epoch": 3.21, "learning_rate": 2.9942908653846158e-05, "loss": 0.1128, "step": 10684 }, { "epoch": 3.21, "learning_rate": 2.9941030649038466e-05, "loss": 0.1242, "step": 10685 }, { "epoch": 3.21, "learning_rate": 2.993915264423077e-05, "loss": 0.0888, "step": 10686 }, { "epoch": 3.21, "learning_rate": 2.993727463942308e-05, "loss": 0.1381, "step": 10687 }, { "epoch": 3.21, "learning_rate": 2.9935396634615387e-05, "loss": 0.1466, "step": 10688 }, { "epoch": 3.21, "learning_rate": 2.9933518629807695e-05, "loss": 0.1297, "step": 10689 }, { "epoch": 3.21, "learning_rate": 2.9931640625e-05, "loss": 0.0865, "step": 10690 }, { "epoch": 3.21, "learning_rate": 2.992976262019231e-05, "loss": 0.1229, "step": 10691 }, { "epoch": 3.21, "learning_rate": 2.9927884615384616e-05, "loss": 0.1673, "step": 10692 }, { "epoch": 3.21, "learning_rate": 2.9926006610576925e-05, "loss": 0.1566, "step": 10693 }, { "epoch": 3.21, "learning_rate": 2.992412860576923e-05, "loss": 0.1403, "step": 10694 }, { "epoch": 3.21, "learning_rate": 2.9922250600961538e-05, "loss": 0.2099, "step": 10695 }, { "epoch": 3.21, "learning_rate": 2.9920372596153846e-05, "loss": 0.1354, "step": 10696 }, { "epoch": 3.21, "learning_rate": 2.9918494591346157e-05, "loss": 0.0975, "step": 10697 }, { "epoch": 3.21, "learning_rate": 2.9916616586538465e-05, "loss": 0.1635, "step": 10698 }, { "epoch": 3.21, "learning_rate": 2.9914738581730774e-05, "loss": 0.1501, "step": 10699 }, { "epoch": 3.22, "learning_rate": 2.991286057692308e-05, "loss": 0.1965, "step": 10700 }, { "epoch": 3.22, "learning_rate": 2.9910982572115387e-05, "loss": 0.1549, "step": 10701 }, { "epoch": 3.22, "learning_rate": 2.9909104567307695e-05, "loss": 0.1803, "step": 10702 }, { "epoch": 3.22, "learning_rate": 2.9907226562500003e-05, "loss": 0.1624, "step": 10703 }, { "epoch": 3.22, "learning_rate": 2.9905348557692308e-05, "loss": 0.1443, "step": 10704 }, { "epoch": 3.22, "learning_rate": 2.9903470552884616e-05, "loss": 0.13, "step": 10705 }, { "epoch": 3.22, "learning_rate": 2.9901592548076924e-05, "loss": 0.1039, "step": 10706 }, { "epoch": 3.22, "learning_rate": 2.9899714543269232e-05, "loss": 0.2361, "step": 10707 }, { "epoch": 3.22, "learning_rate": 2.9897836538461537e-05, "loss": 0.1994, "step": 10708 }, { "epoch": 3.22, "learning_rate": 2.9895958533653845e-05, "loss": 0.1432, "step": 10709 }, { "epoch": 3.22, "learning_rate": 2.9894080528846157e-05, "loss": 0.1371, "step": 10710 }, { "epoch": 3.22, "learning_rate": 2.9892202524038465e-05, "loss": 0.1221, "step": 10711 }, { "epoch": 3.22, "learning_rate": 2.9890324519230773e-05, "loss": 0.1377, "step": 10712 }, { "epoch": 3.22, "learning_rate": 2.988844651442308e-05, "loss": 0.1973, "step": 10713 }, { "epoch": 3.22, "learning_rate": 2.9886568509615386e-05, "loss": 0.1652, "step": 10714 }, { "epoch": 3.22, "learning_rate": 2.9884690504807694e-05, "loss": 0.2507, "step": 10715 }, { "epoch": 3.22, "learning_rate": 2.9882812500000002e-05, "loss": 0.1514, "step": 10716 }, { "epoch": 3.22, "learning_rate": 2.988093449519231e-05, "loss": 0.1308, "step": 10717 }, { "epoch": 3.22, "learning_rate": 2.9879056490384615e-05, "loss": 0.1327, "step": 10718 }, { "epoch": 3.22, "learning_rate": 2.9877178485576923e-05, "loss": 0.1685, "step": 10719 }, { "epoch": 3.22, "learning_rate": 2.987530048076923e-05, "loss": 0.1479, "step": 10720 }, { "epoch": 3.22, "learning_rate": 2.9873422475961536e-05, "loss": 0.1365, "step": 10721 }, { "epoch": 3.22, "learning_rate": 2.9871544471153844e-05, "loss": 0.1016, "step": 10722 }, { "epoch": 3.22, "learning_rate": 2.986966646634616e-05, "loss": 0.1178, "step": 10723 }, { "epoch": 3.22, "learning_rate": 2.9867788461538464e-05, "loss": 0.0846, "step": 10724 }, { "epoch": 3.22, "learning_rate": 2.9865910456730772e-05, "loss": 0.101, "step": 10725 }, { "epoch": 3.22, "learning_rate": 2.986403245192308e-05, "loss": 0.1187, "step": 10726 }, { "epoch": 3.22, "learning_rate": 2.986215444711539e-05, "loss": 0.1981, "step": 10727 }, { "epoch": 3.22, "learning_rate": 2.9860276442307693e-05, "loss": 0.1047, "step": 10728 }, { "epoch": 3.22, "learning_rate": 2.98583984375e-05, "loss": 0.1127, "step": 10729 }, { "epoch": 3.22, "learning_rate": 2.985652043269231e-05, "loss": 0.2519, "step": 10730 }, { "epoch": 3.22, "learning_rate": 2.9854642427884614e-05, "loss": 0.1112, "step": 10731 }, { "epoch": 3.22, "learning_rate": 2.9852764423076923e-05, "loss": 0.1563, "step": 10732 }, { "epoch": 3.23, "learning_rate": 2.985088641826923e-05, "loss": 0.1393, "step": 10733 }, { "epoch": 3.23, "learning_rate": 2.984900841346154e-05, "loss": 0.1461, "step": 10734 }, { "epoch": 3.23, "learning_rate": 2.9847130408653844e-05, "loss": 0.1422, "step": 10735 }, { "epoch": 3.23, "learning_rate": 2.984525240384616e-05, "loss": 0.1721, "step": 10736 }, { "epoch": 3.23, "learning_rate": 2.9843374399038467e-05, "loss": 0.2844, "step": 10737 }, { "epoch": 3.23, "learning_rate": 2.984149639423077e-05, "loss": 0.2067, "step": 10738 }, { "epoch": 3.23, "learning_rate": 2.983961838942308e-05, "loss": 0.1267, "step": 10739 }, { "epoch": 3.23, "learning_rate": 2.9837740384615388e-05, "loss": 0.1295, "step": 10740 }, { "epoch": 3.23, "learning_rate": 2.9835862379807693e-05, "loss": 0.2031, "step": 10741 }, { "epoch": 3.23, "learning_rate": 2.9833984375e-05, "loss": 0.1843, "step": 10742 }, { "epoch": 3.23, "learning_rate": 2.983210637019231e-05, "loss": 0.1554, "step": 10743 }, { "epoch": 3.23, "learning_rate": 2.9830228365384617e-05, "loss": 0.1414, "step": 10744 }, { "epoch": 3.23, "learning_rate": 2.9828350360576922e-05, "loss": 0.1304, "step": 10745 }, { "epoch": 3.23, "learning_rate": 2.982647235576923e-05, "loss": 0.1264, "step": 10746 }, { "epoch": 3.23, "learning_rate": 2.9824594350961538e-05, "loss": 0.1038, "step": 10747 }, { "epoch": 3.23, "learning_rate": 2.9822716346153846e-05, "loss": 0.1805, "step": 10748 }, { "epoch": 3.23, "learning_rate": 2.9820838341346158e-05, "loss": 0.147, "step": 10749 }, { "epoch": 3.23, "learning_rate": 2.9818960336538466e-05, "loss": 0.1503, "step": 10750 }, { "epoch": 3.23, "learning_rate": 2.981708233173077e-05, "loss": 0.1256, "step": 10751 }, { "epoch": 3.23, "learning_rate": 2.981520432692308e-05, "loss": 0.1863, "step": 10752 }, { "epoch": 3.23, "learning_rate": 2.9813326322115387e-05, "loss": 0.2004, "step": 10753 }, { "epoch": 3.23, "learning_rate": 2.9811448317307695e-05, "loss": 0.2154, "step": 10754 }, { "epoch": 3.23, "learning_rate": 2.98095703125e-05, "loss": 0.0968, "step": 10755 }, { "epoch": 3.23, "learning_rate": 2.9807692307692308e-05, "loss": 0.1256, "step": 10756 }, { "epoch": 3.23, "learning_rate": 2.9805814302884616e-05, "loss": 0.111, "step": 10757 }, { "epoch": 3.23, "learning_rate": 2.9803936298076925e-05, "loss": 0.1078, "step": 10758 }, { "epoch": 3.23, "learning_rate": 2.980205829326923e-05, "loss": 0.1109, "step": 10759 }, { "epoch": 3.23, "learning_rate": 2.9800180288461537e-05, "loss": 0.1782, "step": 10760 }, { "epoch": 3.23, "learning_rate": 2.9798302283653846e-05, "loss": 0.171, "step": 10761 }, { "epoch": 3.23, "learning_rate": 2.9796424278846157e-05, "loss": 0.2136, "step": 10762 }, { "epoch": 3.23, "learning_rate": 2.9794546274038465e-05, "loss": 0.1473, "step": 10763 }, { "epoch": 3.23, "learning_rate": 2.9792668269230773e-05, "loss": 0.1031, "step": 10764 }, { "epoch": 3.23, "learning_rate": 2.9790790264423078e-05, "loss": 0.1382, "step": 10765 }, { "epoch": 3.23, "learning_rate": 2.9788912259615386e-05, "loss": 0.1601, "step": 10766 }, { "epoch": 3.24, "learning_rate": 2.9787034254807695e-05, "loss": 0.1869, "step": 10767 }, { "epoch": 3.24, "learning_rate": 2.9785156250000003e-05, "loss": 0.1559, "step": 10768 }, { "epoch": 3.24, "learning_rate": 2.9783278245192308e-05, "loss": 0.1469, "step": 10769 }, { "epoch": 3.24, "learning_rate": 2.9781400240384616e-05, "loss": 0.1608, "step": 10770 }, { "epoch": 3.24, "learning_rate": 2.9779522235576924e-05, "loss": 0.129, "step": 10771 }, { "epoch": 3.24, "learning_rate": 2.9777644230769232e-05, "loss": 0.1651, "step": 10772 }, { "epoch": 3.24, "learning_rate": 2.9775766225961537e-05, "loss": 0.1296, "step": 10773 }, { "epoch": 3.24, "learning_rate": 2.9773888221153845e-05, "loss": 0.1188, "step": 10774 }, { "epoch": 3.24, "learning_rate": 2.9772010216346156e-05, "loss": 0.1798, "step": 10775 }, { "epoch": 3.24, "learning_rate": 2.9770132211538465e-05, "loss": 0.1207, "step": 10776 }, { "epoch": 3.24, "learning_rate": 2.9768254206730773e-05, "loss": 0.1157, "step": 10777 }, { "epoch": 3.24, "learning_rate": 2.976637620192308e-05, "loss": 0.1409, "step": 10778 }, { "epoch": 3.24, "learning_rate": 2.9764498197115386e-05, "loss": 0.1484, "step": 10779 }, { "epoch": 3.24, "learning_rate": 2.9762620192307694e-05, "loss": 0.1768, "step": 10780 }, { "epoch": 3.24, "learning_rate": 2.9760742187500002e-05, "loss": 0.1636, "step": 10781 }, { "epoch": 3.24, "learning_rate": 2.975886418269231e-05, "loss": 0.1689, "step": 10782 }, { "epoch": 3.24, "learning_rate": 2.9756986177884615e-05, "loss": 0.1396, "step": 10783 }, { "epoch": 3.24, "learning_rate": 2.9755108173076923e-05, "loss": 0.1411, "step": 10784 }, { "epoch": 3.24, "learning_rate": 2.975323016826923e-05, "loss": 0.1219, "step": 10785 }, { "epoch": 3.24, "learning_rate": 2.9751352163461536e-05, "loss": 0.1647, "step": 10786 }, { "epoch": 3.24, "learning_rate": 2.9749474158653844e-05, "loss": 0.2179, "step": 10787 }, { "epoch": 3.24, "learning_rate": 2.974759615384616e-05, "loss": 0.1169, "step": 10788 }, { "epoch": 3.24, "learning_rate": 2.9745718149038464e-05, "loss": 0.115, "step": 10789 }, { "epoch": 3.24, "learning_rate": 2.9743840144230772e-05, "loss": 0.1805, "step": 10790 }, { "epoch": 3.24, "learning_rate": 2.974196213942308e-05, "loss": 0.1098, "step": 10791 }, { "epoch": 3.24, "learning_rate": 2.974008413461539e-05, "loss": 0.1527, "step": 10792 }, { "epoch": 3.24, "learning_rate": 2.9738206129807693e-05, "loss": 0.1199, "step": 10793 }, { "epoch": 3.24, "learning_rate": 2.9736328125e-05, "loss": 0.1478, "step": 10794 }, { "epoch": 3.24, "learning_rate": 2.973445012019231e-05, "loss": 0.1164, "step": 10795 }, { "epoch": 3.24, "learning_rate": 2.9732572115384614e-05, "loss": 0.2214, "step": 10796 }, { "epoch": 3.24, "learning_rate": 2.9730694110576922e-05, "loss": 0.1307, "step": 10797 }, { "epoch": 3.24, "learning_rate": 2.972881610576923e-05, "loss": 0.215, "step": 10798 }, { "epoch": 3.24, "learning_rate": 2.972693810096154e-05, "loss": 0.1324, "step": 10799 }, { "epoch": 3.25, "learning_rate": 2.9725060096153844e-05, "loss": 0.1493, "step": 10800 }, { "epoch": 3.25, "learning_rate": 2.972318209134616e-05, "loss": 0.1546, "step": 10801 }, { "epoch": 3.25, "learning_rate": 2.9721304086538467e-05, "loss": 0.086, "step": 10802 }, { "epoch": 3.25, "learning_rate": 2.971942608173077e-05, "loss": 0.1416, "step": 10803 }, { "epoch": 3.25, "learning_rate": 2.971754807692308e-05, "loss": 0.1476, "step": 10804 }, { "epoch": 3.25, "learning_rate": 2.9715670072115388e-05, "loss": 0.1401, "step": 10805 }, { "epoch": 3.25, "learning_rate": 2.9713792067307692e-05, "loss": 0.1637, "step": 10806 }, { "epoch": 3.25, "learning_rate": 2.97119140625e-05, "loss": 0.1953, "step": 10807 }, { "epoch": 3.25, "learning_rate": 2.971003605769231e-05, "loss": 0.1536, "step": 10808 }, { "epoch": 3.25, "learning_rate": 2.9708158052884617e-05, "loss": 0.1511, "step": 10809 }, { "epoch": 3.25, "learning_rate": 2.9706280048076922e-05, "loss": 0.1034, "step": 10810 }, { "epoch": 3.25, "learning_rate": 2.970440204326923e-05, "loss": 0.1546, "step": 10811 }, { "epoch": 3.25, "learning_rate": 2.9702524038461538e-05, "loss": 0.1265, "step": 10812 }, { "epoch": 3.25, "learning_rate": 2.9700646033653846e-05, "loss": 0.1294, "step": 10813 }, { "epoch": 3.25, "learning_rate": 2.9698768028846158e-05, "loss": 0.2016, "step": 10814 }, { "epoch": 3.25, "learning_rate": 2.9696890024038466e-05, "loss": 0.1191, "step": 10815 }, { "epoch": 3.25, "learning_rate": 2.969501201923077e-05, "loss": 0.1229, "step": 10816 }, { "epoch": 3.25, "learning_rate": 2.969313401442308e-05, "loss": 0.1156, "step": 10817 }, { "epoch": 3.25, "learning_rate": 2.9691256009615387e-05, "loss": 0.1661, "step": 10818 }, { "epoch": 3.25, "learning_rate": 2.9689378004807695e-05, "loss": 0.1554, "step": 10819 }, { "epoch": 3.25, "learning_rate": 2.96875e-05, "loss": 0.128, "step": 10820 }, { "epoch": 3.25, "learning_rate": 2.9685621995192308e-05, "loss": 0.2022, "step": 10821 }, { "epoch": 3.25, "learning_rate": 2.9683743990384616e-05, "loss": 0.1137, "step": 10822 }, { "epoch": 3.25, "learning_rate": 2.9681865985576924e-05, "loss": 0.1401, "step": 10823 }, { "epoch": 3.25, "learning_rate": 2.967998798076923e-05, "loss": 0.153, "step": 10824 }, { "epoch": 3.25, "learning_rate": 2.9678109975961537e-05, "loss": 0.15, "step": 10825 }, { "epoch": 3.25, "learning_rate": 2.9676231971153846e-05, "loss": 0.1536, "step": 10826 }, { "epoch": 3.25, "learning_rate": 2.9674353966346157e-05, "loss": 0.1188, "step": 10827 }, { "epoch": 3.25, "learning_rate": 2.9672475961538465e-05, "loss": 0.1631, "step": 10828 }, { "epoch": 3.25, "learning_rate": 2.9670597956730773e-05, "loss": 0.1397, "step": 10829 }, { "epoch": 3.25, "learning_rate": 2.9668719951923078e-05, "loss": 0.1435, "step": 10830 }, { "epoch": 3.25, "learning_rate": 2.9666841947115386e-05, "loss": 0.1456, "step": 10831 }, { "epoch": 3.25, "learning_rate": 2.9664963942307694e-05, "loss": 0.1729, "step": 10832 }, { "epoch": 3.26, "learning_rate": 2.9663085937500003e-05, "loss": 0.1045, "step": 10833 }, { "epoch": 3.26, "learning_rate": 2.9661207932692307e-05, "loss": 0.1169, "step": 10834 }, { "epoch": 3.26, "learning_rate": 2.9659329927884616e-05, "loss": 0.1989, "step": 10835 }, { "epoch": 3.26, "learning_rate": 2.9657451923076924e-05, "loss": 0.1507, "step": 10836 }, { "epoch": 3.26, "learning_rate": 2.9655573918269232e-05, "loss": 0.2002, "step": 10837 }, { "epoch": 3.26, "learning_rate": 2.9653695913461537e-05, "loss": 0.1158, "step": 10838 }, { "epoch": 3.26, "learning_rate": 2.9651817908653845e-05, "loss": 0.1774, "step": 10839 }, { "epoch": 3.26, "learning_rate": 2.9649939903846156e-05, "loss": 0.2044, "step": 10840 }, { "epoch": 3.26, "learning_rate": 2.9648061899038465e-05, "loss": 0.1665, "step": 10841 }, { "epoch": 3.26, "learning_rate": 2.9646183894230773e-05, "loss": 0.1417, "step": 10842 }, { "epoch": 3.26, "learning_rate": 2.964430588942308e-05, "loss": 0.1863, "step": 10843 }, { "epoch": 3.26, "learning_rate": 2.9642427884615386e-05, "loss": 0.188, "step": 10844 }, { "epoch": 3.26, "learning_rate": 2.9640549879807694e-05, "loss": 0.121, "step": 10845 }, { "epoch": 3.26, "learning_rate": 2.9638671875000002e-05, "loss": 0.1242, "step": 10846 }, { "epoch": 3.26, "learning_rate": 2.963679387019231e-05, "loss": 0.1404, "step": 10847 }, { "epoch": 3.26, "learning_rate": 2.9634915865384615e-05, "loss": 0.168, "step": 10848 }, { "epoch": 3.26, "learning_rate": 2.9633037860576923e-05, "loss": 0.1637, "step": 10849 }, { "epoch": 3.26, "learning_rate": 2.963115985576923e-05, "loss": 0.1542, "step": 10850 }, { "epoch": 3.26, "learning_rate": 2.962928185096154e-05, "loss": 0.1419, "step": 10851 }, { "epoch": 3.26, "learning_rate": 2.9627403846153844e-05, "loss": 0.1843, "step": 10852 }, { "epoch": 3.26, "learning_rate": 2.962552584134616e-05, "loss": 0.1869, "step": 10853 }, { "epoch": 3.26, "learning_rate": 2.9623647836538464e-05, "loss": 0.1433, "step": 10854 }, { "epoch": 3.26, "learning_rate": 2.9621769831730772e-05, "loss": 0.1766, "step": 10855 }, { "epoch": 3.26, "learning_rate": 2.9621769831730772e-05, "loss": 0.2068, "step": 10856 }, { "epoch": 3.26, "learning_rate": 2.961989182692308e-05, "loss": 0.1333, "step": 10857 }, { "epoch": 3.26, "learning_rate": 2.9618013822115388e-05, "loss": 0.1673, "step": 10858 }, { "epoch": 3.26, "learning_rate": 2.9616135817307693e-05, "loss": 0.1996, "step": 10859 }, { "epoch": 3.26, "learning_rate": 2.96142578125e-05, "loss": 0.1114, "step": 10860 }, { "epoch": 3.26, "learning_rate": 2.961237980769231e-05, "loss": 0.1378, "step": 10861 }, { "epoch": 3.26, "learning_rate": 2.9610501802884618e-05, "loss": 0.2224, "step": 10862 }, { "epoch": 3.26, "learning_rate": 2.9608623798076922e-05, "loss": 0.164, "step": 10863 }, { "epoch": 3.26, "learning_rate": 2.960674579326923e-05, "loss": 0.1619, "step": 10864 }, { "epoch": 3.26, "learning_rate": 2.960486778846154e-05, "loss": 0.1299, "step": 10865 }, { "epoch": 3.27, "learning_rate": 2.9602989783653843e-05, "loss": 0.1753, "step": 10866 }, { "epoch": 3.27, "learning_rate": 2.960111177884616e-05, "loss": 0.2125, "step": 10867 }, { "epoch": 3.27, "learning_rate": 2.9599233774038467e-05, "loss": 0.1462, "step": 10868 }, { "epoch": 3.27, "learning_rate": 2.959735576923077e-05, "loss": 0.1264, "step": 10869 }, { "epoch": 3.27, "learning_rate": 2.959547776442308e-05, "loss": 0.139, "step": 10870 }, { "epoch": 3.27, "learning_rate": 2.9593599759615388e-05, "loss": 0.1633, "step": 10871 }, { "epoch": 3.27, "learning_rate": 2.9591721754807696e-05, "loss": 0.1414, "step": 10872 }, { "epoch": 3.27, "learning_rate": 2.958984375e-05, "loss": 0.1359, "step": 10873 }, { "epoch": 3.27, "learning_rate": 2.958796574519231e-05, "loss": 0.1237, "step": 10874 }, { "epoch": 3.27, "learning_rate": 2.9586087740384617e-05, "loss": 0.2294, "step": 10875 }, { "epoch": 3.27, "learning_rate": 2.958420973557692e-05, "loss": 0.1467, "step": 10876 }, { "epoch": 3.27, "learning_rate": 2.958233173076923e-05, "loss": 0.1488, "step": 10877 }, { "epoch": 3.27, "learning_rate": 2.9580453725961538e-05, "loss": 0.1215, "step": 10878 }, { "epoch": 3.27, "learning_rate": 2.9578575721153846e-05, "loss": 0.138, "step": 10879 }, { "epoch": 3.27, "learning_rate": 2.9576697716346158e-05, "loss": 0.1265, "step": 10880 }, { "epoch": 3.27, "learning_rate": 2.9574819711538466e-05, "loss": 0.1602, "step": 10881 }, { "epoch": 3.27, "learning_rate": 2.9572941706730774e-05, "loss": 0.1614, "step": 10882 }, { "epoch": 3.27, "learning_rate": 2.957106370192308e-05, "loss": 0.148, "step": 10883 }, { "epoch": 3.27, "learning_rate": 2.9569185697115387e-05, "loss": 0.2035, "step": 10884 }, { "epoch": 3.27, "learning_rate": 2.9567307692307695e-05, "loss": 0.11, "step": 10885 }, { "epoch": 3.27, "learning_rate": 2.95654296875e-05, "loss": 0.166, "step": 10886 }, { "epoch": 3.27, "learning_rate": 2.9563551682692308e-05, "loss": 0.0953, "step": 10887 }, { "epoch": 3.27, "learning_rate": 2.9561673677884616e-05, "loss": 0.1085, "step": 10888 }, { "epoch": 3.27, "learning_rate": 2.9559795673076924e-05, "loss": 0.1065, "step": 10889 }, { "epoch": 3.27, "learning_rate": 2.955791766826923e-05, "loss": 0.1222, "step": 10890 }, { "epoch": 3.27, "learning_rate": 2.9556039663461537e-05, "loss": 0.1181, "step": 10891 }, { "epoch": 3.27, "learning_rate": 2.9554161658653845e-05, "loss": 0.3456, "step": 10892 }, { "epoch": 3.27, "learning_rate": 2.9552283653846157e-05, "loss": 0.1479, "step": 10893 }, { "epoch": 3.27, "learning_rate": 2.9550405649038465e-05, "loss": 0.0799, "step": 10894 }, { "epoch": 3.27, "learning_rate": 2.9548527644230773e-05, "loss": 0.1961, "step": 10895 }, { "epoch": 3.27, "learning_rate": 2.9546649639423078e-05, "loss": 0.151, "step": 10896 }, { "epoch": 3.27, "learning_rate": 2.9544771634615386e-05, "loss": 0.2083, "step": 10897 }, { "epoch": 3.27, "learning_rate": 2.9542893629807694e-05, "loss": 0.0909, "step": 10898 }, { "epoch": 3.27, "learning_rate": 2.9541015625000003e-05, "loss": 0.2072, "step": 10899 }, { "epoch": 3.28, "learning_rate": 2.9539137620192307e-05, "loss": 0.1955, "step": 10900 }, { "epoch": 3.28, "learning_rate": 2.9537259615384615e-05, "loss": 0.2245, "step": 10901 }, { "epoch": 3.28, "learning_rate": 2.9535381610576924e-05, "loss": 0.1226, "step": 10902 }, { "epoch": 3.28, "learning_rate": 2.9533503605769232e-05, "loss": 0.0877, "step": 10903 }, { "epoch": 3.28, "learning_rate": 2.9531625600961537e-05, "loss": 0.0989, "step": 10904 }, { "epoch": 3.28, "learning_rate": 2.9529747596153845e-05, "loss": 0.1212, "step": 10905 }, { "epoch": 3.28, "learning_rate": 2.9527869591346156e-05, "loss": 0.1217, "step": 10906 }, { "epoch": 3.28, "learning_rate": 2.9525991586538464e-05, "loss": 0.1924, "step": 10907 }, { "epoch": 3.28, "learning_rate": 2.9524113581730773e-05, "loss": 0.1665, "step": 10908 }, { "epoch": 3.28, "learning_rate": 2.952223557692308e-05, "loss": 0.1184, "step": 10909 }, { "epoch": 3.28, "learning_rate": 2.9520357572115386e-05, "loss": 0.1803, "step": 10910 }, { "epoch": 3.28, "learning_rate": 2.9518479567307694e-05, "loss": 0.1644, "step": 10911 }, { "epoch": 3.28, "learning_rate": 2.9516601562500002e-05, "loss": 0.1121, "step": 10912 }, { "epoch": 3.28, "learning_rate": 2.951472355769231e-05, "loss": 0.1211, "step": 10913 }, { "epoch": 3.28, "learning_rate": 2.9512845552884615e-05, "loss": 0.2022, "step": 10914 }, { "epoch": 3.28, "learning_rate": 2.9510967548076923e-05, "loss": 0.1279, "step": 10915 }, { "epoch": 3.28, "learning_rate": 2.950908954326923e-05, "loss": 0.1026, "step": 10916 }, { "epoch": 3.28, "learning_rate": 2.950721153846154e-05, "loss": 0.1366, "step": 10917 }, { "epoch": 3.28, "learning_rate": 2.9505333533653844e-05, "loss": 0.181, "step": 10918 }, { "epoch": 3.28, "learning_rate": 2.950345552884616e-05, "loss": 0.1291, "step": 10919 }, { "epoch": 3.28, "learning_rate": 2.9501577524038464e-05, "loss": 0.1156, "step": 10920 }, { "epoch": 3.28, "learning_rate": 2.9499699519230772e-05, "loss": 0.1265, "step": 10921 }, { "epoch": 3.28, "learning_rate": 2.949782151442308e-05, "loss": 0.154, "step": 10922 }, { "epoch": 3.28, "learning_rate": 2.9495943509615388e-05, "loss": 0.1357, "step": 10923 }, { "epoch": 3.28, "learning_rate": 2.9494065504807693e-05, "loss": 0.0832, "step": 10924 }, { "epoch": 3.28, "learning_rate": 2.94921875e-05, "loss": 0.1314, "step": 10925 }, { "epoch": 3.28, "learning_rate": 2.949030949519231e-05, "loss": 0.1224, "step": 10926 }, { "epoch": 3.28, "learning_rate": 2.9488431490384617e-05, "loss": 0.1178, "step": 10927 }, { "epoch": 3.28, "learning_rate": 2.9486553485576922e-05, "loss": 0.2023, "step": 10928 }, { "epoch": 3.28, "learning_rate": 2.948467548076923e-05, "loss": 0.1361, "step": 10929 }, { "epoch": 3.28, "learning_rate": 2.948279747596154e-05, "loss": 0.2192, "step": 10930 }, { "epoch": 3.28, "learning_rate": 2.9480919471153847e-05, "loss": 0.2128, "step": 10931 }, { "epoch": 3.28, "learning_rate": 2.9479041466346158e-05, "loss": 0.2116, "step": 10932 }, { "epoch": 3.29, "learning_rate": 2.9477163461538466e-05, "loss": 0.0961, "step": 10933 }, { "epoch": 3.29, "learning_rate": 2.947528545673077e-05, "loss": 0.1647, "step": 10934 }, { "epoch": 3.29, "learning_rate": 2.947340745192308e-05, "loss": 0.1366, "step": 10935 }, { "epoch": 3.29, "learning_rate": 2.9471529447115387e-05, "loss": 0.1015, "step": 10936 }, { "epoch": 3.29, "learning_rate": 2.9469651442307696e-05, "loss": 0.2416, "step": 10937 }, { "epoch": 3.29, "learning_rate": 2.94677734375e-05, "loss": 0.1857, "step": 10938 }, { "epoch": 3.29, "learning_rate": 2.946589543269231e-05, "loss": 0.1715, "step": 10939 }, { "epoch": 3.29, "learning_rate": 2.9464017427884617e-05, "loss": 0.1592, "step": 10940 }, { "epoch": 3.29, "learning_rate": 2.946213942307692e-05, "loss": 0.1807, "step": 10941 }, { "epoch": 3.29, "learning_rate": 2.946026141826923e-05, "loss": 0.1404, "step": 10942 }, { "epoch": 3.29, "learning_rate": 2.9458383413461538e-05, "loss": 0.0993, "step": 10943 }, { "epoch": 3.29, "learning_rate": 2.9456505408653846e-05, "loss": 0.21, "step": 10944 }, { "epoch": 3.29, "learning_rate": 2.9454627403846158e-05, "loss": 0.1518, "step": 10945 }, { "epoch": 3.29, "learning_rate": 2.9452749399038466e-05, "loss": 0.1129, "step": 10946 }, { "epoch": 3.29, "learning_rate": 2.9450871394230774e-05, "loss": 0.236, "step": 10947 }, { "epoch": 3.29, "learning_rate": 2.944899338942308e-05, "loss": 0.1411, "step": 10948 }, { "epoch": 3.29, "learning_rate": 2.9447115384615387e-05, "loss": 0.2096, "step": 10949 }, { "epoch": 3.29, "learning_rate": 2.9445237379807695e-05, "loss": 0.2083, "step": 10950 }, { "epoch": 3.29, "learning_rate": 2.9443359375e-05, "loss": 0.1331, "step": 10951 }, { "epoch": 3.29, "learning_rate": 2.9441481370192308e-05, "loss": 0.1636, "step": 10952 }, { "epoch": 3.29, "learning_rate": 2.9439603365384616e-05, "loss": 0.1192, "step": 10953 }, { "epoch": 3.29, "learning_rate": 2.9437725360576924e-05, "loss": 0.2145, "step": 10954 }, { "epoch": 3.29, "learning_rate": 2.943584735576923e-05, "loss": 0.132, "step": 10955 }, { "epoch": 3.29, "learning_rate": 2.9433969350961537e-05, "loss": 0.1841, "step": 10956 }, { "epoch": 3.29, "learning_rate": 2.9432091346153845e-05, "loss": 0.2335, "step": 10957 }, { "epoch": 3.29, "learning_rate": 2.9430213341346157e-05, "loss": 0.1812, "step": 10958 }, { "epoch": 3.29, "learning_rate": 2.9428335336538465e-05, "loss": 0.1381, "step": 10959 }, { "epoch": 3.29, "learning_rate": 2.9426457331730773e-05, "loss": 0.1694, "step": 10960 }, { "epoch": 3.29, "learning_rate": 2.9424579326923078e-05, "loss": 0.101, "step": 10961 }, { "epoch": 3.29, "learning_rate": 2.9422701322115386e-05, "loss": 0.113, "step": 10962 }, { "epoch": 3.29, "learning_rate": 2.9420823317307694e-05, "loss": 0.1464, "step": 10963 }, { "epoch": 3.29, "learning_rate": 2.9418945312500002e-05, "loss": 0.1316, "step": 10964 }, { "epoch": 3.29, "learning_rate": 2.9417067307692307e-05, "loss": 0.1152, "step": 10965 }, { "epoch": 3.3, "learning_rate": 2.9415189302884615e-05, "loss": 0.0935, "step": 10966 }, { "epoch": 3.3, "learning_rate": 2.9413311298076924e-05, "loss": 0.1173, "step": 10967 }, { "epoch": 3.3, "learning_rate": 2.941143329326923e-05, "loss": 0.1142, "step": 10968 }, { "epoch": 3.3, "learning_rate": 2.9409555288461536e-05, "loss": 0.1328, "step": 10969 }, { "epoch": 3.3, "learning_rate": 2.9407677283653845e-05, "loss": 0.0933, "step": 10970 }, { "epoch": 3.3, "learning_rate": 2.9405799278846156e-05, "loss": 0.1878, "step": 10971 }, { "epoch": 3.3, "learning_rate": 2.9403921274038464e-05, "loss": 0.0966, "step": 10972 }, { "epoch": 3.3, "learning_rate": 2.9402043269230772e-05, "loss": 0.1253, "step": 10973 }, { "epoch": 3.3, "learning_rate": 2.940016526442308e-05, "loss": 0.13, "step": 10974 }, { "epoch": 3.3, "learning_rate": 2.9398287259615385e-05, "loss": 0.1982, "step": 10975 }, { "epoch": 3.3, "learning_rate": 2.9396409254807694e-05, "loss": 0.1936, "step": 10976 }, { "epoch": 3.3, "learning_rate": 2.9394531250000002e-05, "loss": 0.149, "step": 10977 }, { "epoch": 3.3, "learning_rate": 2.939265324519231e-05, "loss": 0.1817, "step": 10978 }, { "epoch": 3.3, "learning_rate": 2.9390775240384615e-05, "loss": 0.1941, "step": 10979 }, { "epoch": 3.3, "learning_rate": 2.9388897235576923e-05, "loss": 0.1188, "step": 10980 }, { "epoch": 3.3, "learning_rate": 2.938701923076923e-05, "loss": 0.1331, "step": 10981 }, { "epoch": 3.3, "learning_rate": 2.938514122596154e-05, "loss": 0.1674, "step": 10982 }, { "epoch": 3.3, "learning_rate": 2.9383263221153844e-05, "loss": 0.0856, "step": 10983 }, { "epoch": 3.3, "learning_rate": 2.938138521634616e-05, "loss": 0.1817, "step": 10984 }, { "epoch": 3.3, "learning_rate": 2.9379507211538464e-05, "loss": 0.1264, "step": 10985 }, { "epoch": 3.3, "learning_rate": 2.9377629206730772e-05, "loss": 0.1588, "step": 10986 }, { "epoch": 3.3, "learning_rate": 2.937575120192308e-05, "loss": 0.1197, "step": 10987 }, { "epoch": 3.3, "learning_rate": 2.9373873197115388e-05, "loss": 0.1245, "step": 10988 }, { "epoch": 3.3, "learning_rate": 2.9371995192307693e-05, "loss": 0.1337, "step": 10989 }, { "epoch": 3.3, "learning_rate": 2.93701171875e-05, "loss": 0.1788, "step": 10990 }, { "epoch": 3.3, "learning_rate": 2.936823918269231e-05, "loss": 0.1132, "step": 10991 }, { "epoch": 3.3, "learning_rate": 2.9366361177884617e-05, "loss": 0.1285, "step": 10992 }, { "epoch": 3.3, "learning_rate": 2.9364483173076922e-05, "loss": 0.1214, "step": 10993 }, { "epoch": 3.3, "learning_rate": 2.936260516826923e-05, "loss": 0.1609, "step": 10994 }, { "epoch": 3.3, "learning_rate": 2.936072716346154e-05, "loss": 0.1631, "step": 10995 }, { "epoch": 3.3, "learning_rate": 2.9358849158653847e-05, "loss": 0.1472, "step": 10996 }, { "epoch": 3.3, "learning_rate": 2.9356971153846158e-05, "loss": 0.1065, "step": 10997 }, { "epoch": 3.3, "learning_rate": 2.9355093149038466e-05, "loss": 0.1525, "step": 10998 }, { "epoch": 3.3, "learning_rate": 2.935321514423077e-05, "loss": 0.1383, "step": 10999 }, { "epoch": 3.31, "learning_rate": 2.935133713942308e-05, "loss": 0.1835, "step": 11000 }, { "epoch": 3.31, "learning_rate": 2.9349459134615387e-05, "loss": 0.1439, "step": 11001 }, { "epoch": 3.31, "learning_rate": 2.9347581129807696e-05, "loss": 0.1133, "step": 11002 }, { "epoch": 3.31, "learning_rate": 2.9345703125e-05, "loss": 0.1707, "step": 11003 }, { "epoch": 3.31, "learning_rate": 2.934382512019231e-05, "loss": 0.268, "step": 11004 }, { "epoch": 3.31, "learning_rate": 2.9341947115384617e-05, "loss": 0.2171, "step": 11005 }, { "epoch": 3.31, "learning_rate": 2.9340069110576925e-05, "loss": 0.1318, "step": 11006 }, { "epoch": 3.31, "learning_rate": 2.933819110576923e-05, "loss": 0.1725, "step": 11007 }, { "epoch": 3.31, "learning_rate": 2.9336313100961538e-05, "loss": 0.1227, "step": 11008 }, { "epoch": 3.31, "learning_rate": 2.9334435096153846e-05, "loss": 0.1836, "step": 11009 }, { "epoch": 3.31, "learning_rate": 2.9332557091346157e-05, "loss": 0.1438, "step": 11010 }, { "epoch": 3.31, "learning_rate": 2.9330679086538466e-05, "loss": 0.1757, "step": 11011 }, { "epoch": 3.31, "learning_rate": 2.9328801081730774e-05, "loss": 0.1612, "step": 11012 }, { "epoch": 3.31, "learning_rate": 2.932692307692308e-05, "loss": 0.1559, "step": 11013 }, { "epoch": 3.31, "learning_rate": 2.9325045072115387e-05, "loss": 0.1694, "step": 11014 }, { "epoch": 3.31, "learning_rate": 2.9323167067307695e-05, "loss": 0.1584, "step": 11015 }, { "epoch": 3.31, "learning_rate": 2.9321289062500003e-05, "loss": 0.2053, "step": 11016 }, { "epoch": 3.31, "learning_rate": 2.9319411057692308e-05, "loss": 0.1564, "step": 11017 }, { "epoch": 3.31, "learning_rate": 2.9317533052884616e-05, "loss": 0.1438, "step": 11018 }, { "epoch": 3.31, "learning_rate": 2.9315655048076924e-05, "loss": 0.1274, "step": 11019 }, { "epoch": 3.31, "learning_rate": 2.931377704326923e-05, "loss": 0.1564, "step": 11020 }, { "epoch": 3.31, "learning_rate": 2.9311899038461537e-05, "loss": 0.1478, "step": 11021 }, { "epoch": 3.31, "learning_rate": 2.9310021033653845e-05, "loss": 0.1414, "step": 11022 }, { "epoch": 3.31, "learning_rate": 2.9308143028846157e-05, "loss": 0.1977, "step": 11023 }, { "epoch": 3.31, "learning_rate": 2.9306265024038465e-05, "loss": 0.1286, "step": 11024 }, { "epoch": 3.31, "learning_rate": 2.9304387019230773e-05, "loss": 0.132, "step": 11025 }, { "epoch": 3.31, "learning_rate": 2.930250901442308e-05, "loss": 0.1043, "step": 11026 }, { "epoch": 3.31, "learning_rate": 2.9300631009615386e-05, "loss": 0.1771, "step": 11027 }, { "epoch": 3.31, "learning_rate": 2.9298753004807694e-05, "loss": 0.2412, "step": 11028 }, { "epoch": 3.31, "learning_rate": 2.9296875000000002e-05, "loss": 0.2049, "step": 11029 }, { "epoch": 3.31, "learning_rate": 2.9294996995192307e-05, "loss": 0.1365, "step": 11030 }, { "epoch": 3.31, "learning_rate": 2.9293118990384615e-05, "loss": 0.1463, "step": 11031 }, { "epoch": 3.31, "learning_rate": 2.9291240985576923e-05, "loss": 0.1032, "step": 11032 }, { "epoch": 3.32, "learning_rate": 2.928936298076923e-05, "loss": 0.1273, "step": 11033 }, { "epoch": 3.32, "learning_rate": 2.9287484975961536e-05, "loss": 0.2009, "step": 11034 }, { "epoch": 3.32, "learning_rate": 2.9285606971153844e-05, "loss": 0.1649, "step": 11035 }, { "epoch": 3.32, "learning_rate": 2.928372896634616e-05, "loss": 0.1441, "step": 11036 }, { "epoch": 3.32, "learning_rate": 2.9281850961538464e-05, "loss": 0.1107, "step": 11037 }, { "epoch": 3.32, "learning_rate": 2.9279972956730772e-05, "loss": 0.1695, "step": 11038 }, { "epoch": 3.32, "learning_rate": 2.927809495192308e-05, "loss": 0.1445, "step": 11039 }, { "epoch": 3.32, "learning_rate": 2.9276216947115385e-05, "loss": 0.1062, "step": 11040 }, { "epoch": 3.32, "learning_rate": 2.9274338942307693e-05, "loss": 0.187, "step": 11041 }, { "epoch": 3.32, "learning_rate": 2.92724609375e-05, "loss": 0.1466, "step": 11042 }, { "epoch": 3.32, "learning_rate": 2.927058293269231e-05, "loss": 0.1682, "step": 11043 }, { "epoch": 3.32, "learning_rate": 2.9268704927884615e-05, "loss": 0.2069, "step": 11044 }, { "epoch": 3.32, "learning_rate": 2.9266826923076923e-05, "loss": 0.2202, "step": 11045 }, { "epoch": 3.32, "learning_rate": 2.926494891826923e-05, "loss": 0.095, "step": 11046 }, { "epoch": 3.32, "learning_rate": 2.926307091346154e-05, "loss": 0.1409, "step": 11047 }, { "epoch": 3.32, "learning_rate": 2.9261192908653844e-05, "loss": 0.1436, "step": 11048 }, { "epoch": 3.32, "learning_rate": 2.925931490384616e-05, "loss": 0.1849, "step": 11049 }, { "epoch": 3.32, "learning_rate": 2.9257436899038463e-05, "loss": 0.1436, "step": 11050 }, { "epoch": 3.32, "learning_rate": 2.925555889423077e-05, "loss": 0.1874, "step": 11051 }, { "epoch": 3.32, "learning_rate": 2.925368088942308e-05, "loss": 0.1784, "step": 11052 }, { "epoch": 3.32, "learning_rate": 2.9251802884615388e-05, "loss": 0.1531, "step": 11053 }, { "epoch": 3.32, "learning_rate": 2.9249924879807693e-05, "loss": 0.0975, "step": 11054 }, { "epoch": 3.32, "learning_rate": 2.9248046875e-05, "loss": 0.1667, "step": 11055 }, { "epoch": 3.32, "learning_rate": 2.924616887019231e-05, "loss": 0.2331, "step": 11056 }, { "epoch": 3.32, "learning_rate": 2.9244290865384617e-05, "loss": 0.1922, "step": 11057 }, { "epoch": 3.32, "learning_rate": 2.9242412860576922e-05, "loss": 0.2053, "step": 11058 }, { "epoch": 3.32, "learning_rate": 2.924053485576923e-05, "loss": 0.1321, "step": 11059 }, { "epoch": 3.32, "learning_rate": 2.923865685096154e-05, "loss": 0.1714, "step": 11060 }, { "epoch": 3.32, "learning_rate": 2.9236778846153846e-05, "loss": 0.2171, "step": 11061 }, { "epoch": 3.32, "learning_rate": 2.9234900841346158e-05, "loss": 0.1684, "step": 11062 }, { "epoch": 3.32, "learning_rate": 2.9233022836538466e-05, "loss": 0.2124, "step": 11063 }, { "epoch": 3.32, "learning_rate": 2.923114483173077e-05, "loss": 0.1798, "step": 11064 }, { "epoch": 3.32, "learning_rate": 2.922926682692308e-05, "loss": 0.1086, "step": 11065 }, { "epoch": 3.33, "learning_rate": 2.9227388822115387e-05, "loss": 0.1492, "step": 11066 }, { "epoch": 3.33, "learning_rate": 2.9225510817307695e-05, "loss": 0.1538, "step": 11067 }, { "epoch": 3.33, "learning_rate": 2.92236328125e-05, "loss": 0.1406, "step": 11068 }, { "epoch": 3.33, "learning_rate": 2.922175480769231e-05, "loss": 0.1103, "step": 11069 }, { "epoch": 3.33, "learning_rate": 2.9219876802884617e-05, "loss": 0.1342, "step": 11070 }, { "epoch": 3.33, "learning_rate": 2.9217998798076925e-05, "loss": 0.1541, "step": 11071 }, { "epoch": 3.33, "learning_rate": 2.921612079326923e-05, "loss": 0.0774, "step": 11072 }, { "epoch": 3.33, "learning_rate": 2.9214242788461538e-05, "loss": 0.1911, "step": 11073 }, { "epoch": 3.33, "learning_rate": 2.9212364783653846e-05, "loss": 0.1697, "step": 11074 }, { "epoch": 3.33, "learning_rate": 2.9210486778846157e-05, "loss": 0.1432, "step": 11075 }, { "epoch": 3.33, "learning_rate": 2.9208608774038465e-05, "loss": 0.1711, "step": 11076 }, { "epoch": 3.33, "learning_rate": 2.9206730769230774e-05, "loss": 0.1655, "step": 11077 }, { "epoch": 3.33, "learning_rate": 2.920485276442308e-05, "loss": 0.2165, "step": 11078 }, { "epoch": 3.33, "learning_rate": 2.9202974759615387e-05, "loss": 0.0866, "step": 11079 }, { "epoch": 3.33, "learning_rate": 2.9201096754807695e-05, "loss": 0.1468, "step": 11080 }, { "epoch": 3.33, "learning_rate": 2.9199218750000003e-05, "loss": 0.1748, "step": 11081 }, { "epoch": 3.33, "learning_rate": 2.9197340745192308e-05, "loss": 0.1284, "step": 11082 }, { "epoch": 3.33, "learning_rate": 2.9195462740384616e-05, "loss": 0.2165, "step": 11083 }, { "epoch": 3.33, "learning_rate": 2.9193584735576924e-05, "loss": 0.0859, "step": 11084 }, { "epoch": 3.33, "learning_rate": 2.9191706730769232e-05, "loss": 0.2143, "step": 11085 }, { "epoch": 3.33, "learning_rate": 2.9189828725961537e-05, "loss": 0.1734, "step": 11086 }, { "epoch": 3.33, "learning_rate": 2.9187950721153845e-05, "loss": 0.1486, "step": 11087 }, { "epoch": 3.33, "learning_rate": 2.9186072716346157e-05, "loss": 0.1635, "step": 11088 }, { "epoch": 3.33, "learning_rate": 2.9184194711538465e-05, "loss": 0.1415, "step": 11089 }, { "epoch": 3.33, "learning_rate": 2.9182316706730773e-05, "loss": 0.0994, "step": 11090 }, { "epoch": 3.33, "learning_rate": 2.918043870192308e-05, "loss": 0.1546, "step": 11091 }, { "epoch": 3.33, "learning_rate": 2.9178560697115386e-05, "loss": 0.1863, "step": 11092 }, { "epoch": 3.33, "learning_rate": 2.9176682692307694e-05, "loss": 0.1331, "step": 11093 }, { "epoch": 3.33, "learning_rate": 2.9174804687500002e-05, "loss": 0.1121, "step": 11094 }, { "epoch": 3.33, "learning_rate": 2.917292668269231e-05, "loss": 0.1316, "step": 11095 }, { "epoch": 3.33, "learning_rate": 2.9171048677884615e-05, "loss": 0.1354, "step": 11096 }, { "epoch": 3.33, "learning_rate": 2.9169170673076923e-05, "loss": 0.1123, "step": 11097 }, { "epoch": 3.33, "learning_rate": 2.916729266826923e-05, "loss": 0.214, "step": 11098 }, { "epoch": 3.34, "learning_rate": 2.9165414663461536e-05, "loss": 0.1922, "step": 11099 }, { "epoch": 3.34, "learning_rate": 2.9163536658653844e-05, "loss": 0.1518, "step": 11100 }, { "epoch": 3.34, "learning_rate": 2.916165865384616e-05, "loss": 0.0951, "step": 11101 }, { "epoch": 3.34, "learning_rate": 2.9159780649038464e-05, "loss": 0.143, "step": 11102 }, { "epoch": 3.34, "learning_rate": 2.9157902644230772e-05, "loss": 0.1713, "step": 11103 }, { "epoch": 3.34, "learning_rate": 2.915602463942308e-05, "loss": 0.1319, "step": 11104 }, { "epoch": 3.34, "learning_rate": 2.9154146634615385e-05, "loss": 0.1056, "step": 11105 }, { "epoch": 3.34, "learning_rate": 2.9152268629807693e-05, "loss": 0.0631, "step": 11106 }, { "epoch": 3.34, "learning_rate": 2.9150390625e-05, "loss": 0.0924, "step": 11107 }, { "epoch": 3.34, "learning_rate": 2.914851262019231e-05, "loss": 0.1608, "step": 11108 }, { "epoch": 3.34, "learning_rate": 2.9146634615384614e-05, "loss": 0.1879, "step": 11109 }, { "epoch": 3.34, "learning_rate": 2.9144756610576923e-05, "loss": 0.1172, "step": 11110 }, { "epoch": 3.34, "learning_rate": 2.914287860576923e-05, "loss": 0.1188, "step": 11111 }, { "epoch": 3.34, "learning_rate": 2.914100060096154e-05, "loss": 0.1285, "step": 11112 }, { "epoch": 3.34, "learning_rate": 2.9139122596153844e-05, "loss": 0.1445, "step": 11113 }, { "epoch": 3.34, "learning_rate": 2.913724459134616e-05, "loss": 0.163, "step": 11114 }, { "epoch": 3.34, "learning_rate": 2.9135366586538463e-05, "loss": 0.2255, "step": 11115 }, { "epoch": 3.34, "learning_rate": 2.913348858173077e-05, "loss": 0.1439, "step": 11116 }, { "epoch": 3.34, "learning_rate": 2.913161057692308e-05, "loss": 0.1714, "step": 11117 }, { "epoch": 3.34, "learning_rate": 2.9129732572115388e-05, "loss": 0.1436, "step": 11118 }, { "epoch": 3.34, "learning_rate": 2.9127854567307693e-05, "loss": 0.0825, "step": 11119 }, { "epoch": 3.34, "learning_rate": 2.91259765625e-05, "loss": 0.1228, "step": 11120 }, { "epoch": 3.34, "learning_rate": 2.912409855769231e-05, "loss": 0.1718, "step": 11121 }, { "epoch": 3.34, "learning_rate": 2.9122220552884617e-05, "loss": 0.1525, "step": 11122 }, { "epoch": 3.34, "learning_rate": 2.9120342548076922e-05, "loss": 0.1623, "step": 11123 }, { "epoch": 3.34, "learning_rate": 2.911846454326923e-05, "loss": 0.1547, "step": 11124 }, { "epoch": 3.34, "learning_rate": 2.9116586538461538e-05, "loss": 0.133, "step": 11125 }, { "epoch": 3.34, "learning_rate": 2.9114708533653846e-05, "loss": 0.1388, "step": 11126 }, { "epoch": 3.34, "learning_rate": 2.9112830528846158e-05, "loss": 0.1399, "step": 11127 }, { "epoch": 3.34, "learning_rate": 2.9110952524038466e-05, "loss": 0.1295, "step": 11128 }, { "epoch": 3.34, "learning_rate": 2.910907451923077e-05, "loss": 0.1653, "step": 11129 }, { "epoch": 3.34, "learning_rate": 2.910719651442308e-05, "loss": 0.1664, "step": 11130 }, { "epoch": 3.34, "learning_rate": 2.9105318509615387e-05, "loss": 0.1513, "step": 11131 }, { "epoch": 3.34, "learning_rate": 2.9103440504807695e-05, "loss": 0.156, "step": 11132 }, { "epoch": 3.35, "learning_rate": 2.91015625e-05, "loss": 0.1421, "step": 11133 }, { "epoch": 3.35, "learning_rate": 2.9099684495192308e-05, "loss": 0.166, "step": 11134 }, { "epoch": 3.35, "learning_rate": 2.9097806490384616e-05, "loss": 0.187, "step": 11135 }, { "epoch": 3.35, "learning_rate": 2.9095928485576925e-05, "loss": 0.178, "step": 11136 }, { "epoch": 3.35, "learning_rate": 2.909405048076923e-05, "loss": 0.1609, "step": 11137 }, { "epoch": 3.35, "learning_rate": 2.9092172475961538e-05, "loss": 0.1074, "step": 11138 }, { "epoch": 3.35, "learning_rate": 2.9090294471153846e-05, "loss": 0.1663, "step": 11139 }, { "epoch": 3.35, "learning_rate": 2.9088416466346157e-05, "loss": 0.1526, "step": 11140 }, { "epoch": 3.35, "learning_rate": 2.9086538461538465e-05, "loss": 0.1224, "step": 11141 }, { "epoch": 3.35, "learning_rate": 2.9084660456730774e-05, "loss": 0.1042, "step": 11142 }, { "epoch": 3.35, "learning_rate": 2.908278245192308e-05, "loss": 0.2332, "step": 11143 }, { "epoch": 3.35, "learning_rate": 2.9080904447115386e-05, "loss": 0.1341, "step": 11144 }, { "epoch": 3.35, "learning_rate": 2.9079026442307695e-05, "loss": 0.1079, "step": 11145 }, { "epoch": 3.35, "learning_rate": 2.9077148437500003e-05, "loss": 0.1239, "step": 11146 }, { "epoch": 3.35, "learning_rate": 2.9075270432692308e-05, "loss": 0.1613, "step": 11147 }, { "epoch": 3.35, "learning_rate": 2.9073392427884616e-05, "loss": 0.1419, "step": 11148 }, { "epoch": 3.35, "learning_rate": 2.9071514423076924e-05, "loss": 0.1574, "step": 11149 }, { "epoch": 3.35, "learning_rate": 2.9069636418269232e-05, "loss": 0.1925, "step": 11150 }, { "epoch": 3.35, "learning_rate": 2.9067758413461537e-05, "loss": 0.1205, "step": 11151 }, { "epoch": 3.35, "learning_rate": 2.9065880408653845e-05, "loss": 0.0711, "step": 11152 }, { "epoch": 3.35, "learning_rate": 2.9064002403846157e-05, "loss": 0.124, "step": 11153 }, { "epoch": 3.35, "learning_rate": 2.9062124399038465e-05, "loss": 0.1428, "step": 11154 }, { "epoch": 3.35, "learning_rate": 2.9060246394230773e-05, "loss": 0.1266, "step": 11155 }, { "epoch": 3.35, "learning_rate": 2.905836838942308e-05, "loss": 0.1587, "step": 11156 }, { "epoch": 3.35, "learning_rate": 2.9056490384615386e-05, "loss": 0.1059, "step": 11157 }, { "epoch": 3.35, "learning_rate": 2.9054612379807694e-05, "loss": 0.1467, "step": 11158 }, { "epoch": 3.35, "learning_rate": 2.9052734375000002e-05, "loss": 0.1692, "step": 11159 }, { "epoch": 3.35, "learning_rate": 2.905085637019231e-05, "loss": 0.1622, "step": 11160 }, { "epoch": 3.35, "learning_rate": 2.9048978365384615e-05, "loss": 0.0949, "step": 11161 }, { "epoch": 3.35, "learning_rate": 2.9047100360576923e-05, "loss": 0.1564, "step": 11162 }, { "epoch": 3.35, "learning_rate": 2.904522235576923e-05, "loss": 0.1004, "step": 11163 }, { "epoch": 3.35, "learning_rate": 2.9043344350961536e-05, "loss": 0.1349, "step": 11164 }, { "epoch": 3.35, "learning_rate": 2.9041466346153844e-05, "loss": 0.1014, "step": 11165 }, { "epoch": 3.36, "learning_rate": 2.903958834134616e-05, "loss": 0.2162, "step": 11166 }, { "epoch": 3.36, "learning_rate": 2.9037710336538464e-05, "loss": 0.1089, "step": 11167 }, { "epoch": 3.36, "learning_rate": 2.9035832331730772e-05, "loss": 0.1388, "step": 11168 }, { "epoch": 3.36, "learning_rate": 2.903395432692308e-05, "loss": 0.127, "step": 11169 }, { "epoch": 3.36, "learning_rate": 2.903207632211539e-05, "loss": 0.168, "step": 11170 }, { "epoch": 3.36, "learning_rate": 2.9030198317307693e-05, "loss": 0.1471, "step": 11171 }, { "epoch": 3.36, "learning_rate": 2.90283203125e-05, "loss": 0.2717, "step": 11172 }, { "epoch": 3.36, "learning_rate": 2.902644230769231e-05, "loss": 0.1823, "step": 11173 }, { "epoch": 3.36, "learning_rate": 2.9024564302884614e-05, "loss": 0.1561, "step": 11174 }, { "epoch": 3.36, "learning_rate": 2.9022686298076922e-05, "loss": 0.2209, "step": 11175 }, { "epoch": 3.36, "learning_rate": 2.902080829326923e-05, "loss": 0.1532, "step": 11176 }, { "epoch": 3.36, "learning_rate": 2.901893028846154e-05, "loss": 0.158, "step": 11177 }, { "epoch": 3.36, "learning_rate": 2.9017052283653844e-05, "loss": 0.1355, "step": 11178 }, { "epoch": 3.36, "learning_rate": 2.901517427884616e-05, "loss": 0.1207, "step": 11179 }, { "epoch": 3.36, "learning_rate": 2.9013296274038467e-05, "loss": 0.1276, "step": 11180 }, { "epoch": 3.36, "learning_rate": 2.901141826923077e-05, "loss": 0.1622, "step": 11181 }, { "epoch": 3.36, "learning_rate": 2.900954026442308e-05, "loss": 0.1645, "step": 11182 }, { "epoch": 3.36, "learning_rate": 2.9007662259615388e-05, "loss": 0.1281, "step": 11183 }, { "epoch": 3.36, "learning_rate": 2.9005784254807693e-05, "loss": 0.1431, "step": 11184 }, { "epoch": 3.36, "learning_rate": 2.900390625e-05, "loss": 0.127, "step": 11185 }, { "epoch": 3.36, "learning_rate": 2.900202824519231e-05, "loss": 0.1629, "step": 11186 }, { "epoch": 3.36, "learning_rate": 2.9000150240384617e-05, "loss": 0.17, "step": 11187 }, { "epoch": 3.36, "learning_rate": 2.8998272235576922e-05, "loss": 0.1218, "step": 11188 }, { "epoch": 3.36, "learning_rate": 2.899639423076923e-05, "loss": 0.1732, "step": 11189 }, { "epoch": 3.36, "learning_rate": 2.8994516225961538e-05, "loss": 0.142, "step": 11190 }, { "epoch": 3.36, "learning_rate": 2.8992638221153846e-05, "loss": 0.1486, "step": 11191 }, { "epoch": 3.36, "learning_rate": 2.8990760216346158e-05, "loss": 0.1051, "step": 11192 }, { "epoch": 3.36, "learning_rate": 2.8988882211538466e-05, "loss": 0.1618, "step": 11193 }, { "epoch": 3.36, "learning_rate": 2.898700420673077e-05, "loss": 0.172, "step": 11194 }, { "epoch": 3.36, "learning_rate": 2.898512620192308e-05, "loss": 0.1819, "step": 11195 }, { "epoch": 3.36, "learning_rate": 2.8983248197115387e-05, "loss": 0.2046, "step": 11196 }, { "epoch": 3.36, "learning_rate": 2.8981370192307695e-05, "loss": 0.1245, "step": 11197 }, { "epoch": 3.36, "learning_rate": 2.89794921875e-05, "loss": 0.1516, "step": 11198 }, { "epoch": 3.37, "learning_rate": 2.8977614182692308e-05, "loss": 0.1151, "step": 11199 }, { "epoch": 3.37, "learning_rate": 2.8975736177884616e-05, "loss": 0.1082, "step": 11200 }, { "epoch": 3.37, "learning_rate": 2.8973858173076924e-05, "loss": 0.1885, "step": 11201 }, { "epoch": 3.37, "learning_rate": 2.897198016826923e-05, "loss": 0.1965, "step": 11202 }, { "epoch": 3.37, "learning_rate": 2.8970102163461537e-05, "loss": 0.1848, "step": 11203 }, { "epoch": 3.37, "learning_rate": 2.8968224158653846e-05, "loss": 0.2411, "step": 11204 }, { "epoch": 3.37, "learning_rate": 2.8966346153846157e-05, "loss": 0.1536, "step": 11205 }, { "epoch": 3.37, "learning_rate": 2.8964468149038465e-05, "loss": 0.2092, "step": 11206 }, { "epoch": 3.37, "learning_rate": 2.8962590144230773e-05, "loss": 0.1326, "step": 11207 }, { "epoch": 3.37, "learning_rate": 2.8960712139423078e-05, "loss": 0.0898, "step": 11208 }, { "epoch": 3.37, "learning_rate": 2.8958834134615386e-05, "loss": 0.1667, "step": 11209 }, { "epoch": 3.37, "learning_rate": 2.8956956129807695e-05, "loss": 0.1227, "step": 11210 }, { "epoch": 3.37, "learning_rate": 2.8955078125000003e-05, "loss": 0.1719, "step": 11211 }, { "epoch": 3.37, "learning_rate": 2.8953200120192307e-05, "loss": 0.1301, "step": 11212 }, { "epoch": 3.37, "learning_rate": 2.8951322115384616e-05, "loss": 0.1341, "step": 11213 }, { "epoch": 3.37, "learning_rate": 2.8949444110576924e-05, "loss": 0.1532, "step": 11214 }, { "epoch": 3.37, "learning_rate": 2.8947566105769232e-05, "loss": 0.1905, "step": 11215 }, { "epoch": 3.37, "learning_rate": 2.8945688100961537e-05, "loss": 0.1773, "step": 11216 }, { "epoch": 3.37, "learning_rate": 2.8943810096153845e-05, "loss": 0.1199, "step": 11217 }, { "epoch": 3.37, "learning_rate": 2.8941932091346156e-05, "loss": 0.1429, "step": 11218 }, { "epoch": 3.37, "learning_rate": 2.8940054086538465e-05, "loss": 0.1195, "step": 11219 }, { "epoch": 3.37, "learning_rate": 2.8938176081730773e-05, "loss": 0.1089, "step": 11220 }, { "epoch": 3.37, "learning_rate": 2.893629807692308e-05, "loss": 0.1581, "step": 11221 }, { "epoch": 3.37, "learning_rate": 2.8934420072115386e-05, "loss": 0.1292, "step": 11222 }, { "epoch": 3.37, "learning_rate": 2.8932542067307694e-05, "loss": 0.114, "step": 11223 }, { "epoch": 3.37, "learning_rate": 2.8930664062500002e-05, "loss": 0.1689, "step": 11224 }, { "epoch": 3.37, "learning_rate": 2.892878605769231e-05, "loss": 0.1187, "step": 11225 }, { "epoch": 3.37, "learning_rate": 2.8926908052884615e-05, "loss": 0.1057, "step": 11226 }, { "epoch": 3.37, "learning_rate": 2.8925030048076923e-05, "loss": 0.1249, "step": 11227 }, { "epoch": 3.37, "learning_rate": 2.892315204326923e-05, "loss": 0.2485, "step": 11228 }, { "epoch": 3.37, "learning_rate": 2.892127403846154e-05, "loss": 0.2068, "step": 11229 }, { "epoch": 3.37, "learning_rate": 2.8919396033653844e-05, "loss": 0.1487, "step": 11230 }, { "epoch": 3.37, "learning_rate": 2.891751802884616e-05, "loss": 0.101, "step": 11231 }, { "epoch": 3.38, "learning_rate": 2.8915640024038464e-05, "loss": 0.1436, "step": 11232 }, { "epoch": 3.38, "learning_rate": 2.8913762019230772e-05, "loss": 0.145, "step": 11233 }, { "epoch": 3.38, "learning_rate": 2.891188401442308e-05, "loss": 0.1459, "step": 11234 }, { "epoch": 3.38, "learning_rate": 2.891000600961539e-05, "loss": 0.0961, "step": 11235 }, { "epoch": 3.38, "learning_rate": 2.8908128004807693e-05, "loss": 0.1243, "step": 11236 }, { "epoch": 3.38, "learning_rate": 2.890625e-05, "loss": 0.1393, "step": 11237 }, { "epoch": 3.38, "learning_rate": 2.890437199519231e-05, "loss": 0.137, "step": 11238 }, { "epoch": 3.38, "learning_rate": 2.8902493990384618e-05, "loss": 0.1895, "step": 11239 }, { "epoch": 3.38, "learning_rate": 2.8900615985576922e-05, "loss": 0.201, "step": 11240 }, { "epoch": 3.38, "learning_rate": 2.889873798076923e-05, "loss": 0.184, "step": 11241 }, { "epoch": 3.38, "learning_rate": 2.889685997596154e-05, "loss": 0.1538, "step": 11242 }, { "epoch": 3.38, "learning_rate": 2.8894981971153843e-05, "loss": 0.1003, "step": 11243 }, { "epoch": 3.38, "learning_rate": 2.889310396634616e-05, "loss": 0.1562, "step": 11244 }, { "epoch": 3.38, "learning_rate": 2.8891225961538467e-05, "loss": 0.1629, "step": 11245 }, { "epoch": 3.38, "learning_rate": 2.888934795673077e-05, "loss": 0.1364, "step": 11246 }, { "epoch": 3.38, "learning_rate": 2.888746995192308e-05, "loss": 0.1581, "step": 11247 }, { "epoch": 3.38, "learning_rate": 2.8885591947115388e-05, "loss": 0.1337, "step": 11248 }, { "epoch": 3.38, "learning_rate": 2.8883713942307696e-05, "loss": 0.1465, "step": 11249 }, { "epoch": 3.38, "learning_rate": 2.88818359375e-05, "loss": 0.1659, "step": 11250 }, { "epoch": 3.38, "learning_rate": 2.887995793269231e-05, "loss": 0.1991, "step": 11251 }, { "epoch": 3.38, "learning_rate": 2.8878079927884617e-05, "loss": 0.1822, "step": 11252 }, { "epoch": 3.38, "learning_rate": 2.887620192307692e-05, "loss": 0.0892, "step": 11253 }, { "epoch": 3.38, "learning_rate": 2.887432391826923e-05, "loss": 0.1328, "step": 11254 }, { "epoch": 3.38, "learning_rate": 2.8872445913461538e-05, "loss": 0.164, "step": 11255 }, { "epoch": 3.38, "learning_rate": 2.8870567908653846e-05, "loss": 0.198, "step": 11256 }, { "epoch": 3.38, "learning_rate": 2.8868689903846158e-05, "loss": 0.1333, "step": 11257 }, { "epoch": 3.38, "learning_rate": 2.8866811899038466e-05, "loss": 0.1614, "step": 11258 }, { "epoch": 3.38, "learning_rate": 2.886493389423077e-05, "loss": 0.1409, "step": 11259 }, { "epoch": 3.38, "learning_rate": 2.886305588942308e-05, "loss": 0.1953, "step": 11260 }, { "epoch": 3.38, "learning_rate": 2.8861177884615387e-05, "loss": 0.2632, "step": 11261 }, { "epoch": 3.38, "learning_rate": 2.8859299879807695e-05, "loss": 0.134, "step": 11262 }, { "epoch": 3.38, "learning_rate": 2.8857421875e-05, "loss": 0.1593, "step": 11263 }, { "epoch": 3.38, "learning_rate": 2.8855543870192308e-05, "loss": 0.0996, "step": 11264 }, { "epoch": 3.38, "learning_rate": 2.8853665865384616e-05, "loss": 0.168, "step": 11265 }, { "epoch": 3.39, "learning_rate": 2.8851787860576924e-05, "loss": 0.1817, "step": 11266 }, { "epoch": 3.39, "learning_rate": 2.884990985576923e-05, "loss": 0.0964, "step": 11267 }, { "epoch": 3.39, "learning_rate": 2.8848031850961537e-05, "loss": 0.2006, "step": 11268 }, { "epoch": 3.39, "learning_rate": 2.8846153846153845e-05, "loss": 0.1746, "step": 11269 }, { "epoch": 3.39, "learning_rate": 2.8844275841346157e-05, "loss": 0.1704, "step": 11270 }, { "epoch": 3.39, "learning_rate": 2.8842397836538465e-05, "loss": 0.2207, "step": 11271 }, { "epoch": 3.39, "learning_rate": 2.8840519831730773e-05, "loss": 0.1339, "step": 11272 }, { "epoch": 3.39, "learning_rate": 2.8838641826923078e-05, "loss": 0.1617, "step": 11273 }, { "epoch": 3.39, "learning_rate": 2.8836763822115386e-05, "loss": 0.127, "step": 11274 }, { "epoch": 3.39, "learning_rate": 2.8834885817307694e-05, "loss": 0.1171, "step": 11275 }, { "epoch": 3.39, "learning_rate": 2.8833007812500003e-05, "loss": 0.1458, "step": 11276 }, { "epoch": 3.39, "learning_rate": 2.8831129807692307e-05, "loss": 0.1345, "step": 11277 }, { "epoch": 3.39, "learning_rate": 2.8829251802884616e-05, "loss": 0.1149, "step": 11278 }, { "epoch": 3.39, "learning_rate": 2.8827373798076924e-05, "loss": 0.1046, "step": 11279 }, { "epoch": 3.39, "learning_rate": 2.8825495793269232e-05, "loss": 0.1548, "step": 11280 }, { "epoch": 3.39, "learning_rate": 2.8823617788461537e-05, "loss": 0.1924, "step": 11281 }, { "epoch": 3.39, "learning_rate": 2.8821739783653845e-05, "loss": 0.116, "step": 11282 }, { "epoch": 3.39, "learning_rate": 2.8819861778846156e-05, "loss": 0.1354, "step": 11283 }, { "epoch": 3.39, "learning_rate": 2.8817983774038464e-05, "loss": 0.133, "step": 11284 }, { "epoch": 3.39, "learning_rate": 2.8816105769230773e-05, "loss": 0.157, "step": 11285 }, { "epoch": 3.39, "learning_rate": 2.881422776442308e-05, "loss": 0.1083, "step": 11286 }, { "epoch": 3.39, "learning_rate": 2.8812349759615386e-05, "loss": 0.0963, "step": 11287 }, { "epoch": 3.39, "learning_rate": 2.8810471754807694e-05, "loss": 0.1294, "step": 11288 }, { "epoch": 3.39, "learning_rate": 2.8808593750000002e-05, "loss": 0.1417, "step": 11289 }, { "epoch": 3.39, "learning_rate": 2.880671574519231e-05, "loss": 0.1309, "step": 11290 }, { "epoch": 3.39, "learning_rate": 2.8804837740384615e-05, "loss": 0.1822, "step": 11291 }, { "epoch": 3.39, "learning_rate": 2.8802959735576923e-05, "loss": 0.1649, "step": 11292 }, { "epoch": 3.39, "learning_rate": 2.880108173076923e-05, "loss": 0.1212, "step": 11293 }, { "epoch": 3.39, "learning_rate": 2.879920372596154e-05, "loss": 0.2002, "step": 11294 }, { "epoch": 3.39, "learning_rate": 2.8797325721153844e-05, "loss": 0.1755, "step": 11295 }, { "epoch": 3.39, "learning_rate": 2.879544771634616e-05, "loss": 0.1479, "step": 11296 }, { "epoch": 3.39, "learning_rate": 2.8793569711538464e-05, "loss": 0.1545, "step": 11297 }, { "epoch": 3.39, "learning_rate": 2.8791691706730772e-05, "loss": 0.192, "step": 11298 }, { "epoch": 3.4, "learning_rate": 2.878981370192308e-05, "loss": 0.141, "step": 11299 }, { "epoch": 3.4, "learning_rate": 2.8787935697115388e-05, "loss": 0.1528, "step": 11300 }, { "epoch": 3.4, "learning_rate": 2.8786057692307693e-05, "loss": 0.3112, "step": 11301 }, { "epoch": 3.4, "learning_rate": 2.87841796875e-05, "loss": 0.1272, "step": 11302 }, { "epoch": 3.4, "learning_rate": 2.878230168269231e-05, "loss": 0.1818, "step": 11303 }, { "epoch": 3.4, "learning_rate": 2.8780423677884617e-05, "loss": 0.1419, "step": 11304 }, { "epoch": 3.4, "learning_rate": 2.8778545673076922e-05, "loss": 0.1075, "step": 11305 }, { "epoch": 3.4, "learning_rate": 2.877666766826923e-05, "loss": 0.1595, "step": 11306 }, { "epoch": 3.4, "learning_rate": 2.877478966346154e-05, "loss": 0.2412, "step": 11307 }, { "epoch": 3.4, "learning_rate": 2.8772911658653843e-05, "loss": 0.1415, "step": 11308 }, { "epoch": 3.4, "learning_rate": 2.8771033653846158e-05, "loss": 0.1472, "step": 11309 }, { "epoch": 3.4, "learning_rate": 2.8769155649038466e-05, "loss": 0.1174, "step": 11310 }, { "epoch": 3.4, "learning_rate": 2.876727764423077e-05, "loss": 0.1292, "step": 11311 }, { "epoch": 3.4, "learning_rate": 2.876539963942308e-05, "loss": 0.099, "step": 11312 }, { "epoch": 3.4, "learning_rate": 2.8763521634615388e-05, "loss": 0.1496, "step": 11313 }, { "epoch": 3.4, "learning_rate": 2.8761643629807696e-05, "loss": 0.1609, "step": 11314 }, { "epoch": 3.4, "learning_rate": 2.8759765625e-05, "loss": 0.1472, "step": 11315 }, { "epoch": 3.4, "learning_rate": 2.875788762019231e-05, "loss": 0.1398, "step": 11316 }, { "epoch": 3.4, "learning_rate": 2.8756009615384617e-05, "loss": 0.1281, "step": 11317 }, { "epoch": 3.4, "learning_rate": 2.875413161057692e-05, "loss": 0.1724, "step": 11318 }, { "epoch": 3.4, "learning_rate": 2.875225360576923e-05, "loss": 0.1786, "step": 11319 }, { "epoch": 3.4, "learning_rate": 2.8750375600961538e-05, "loss": 0.1253, "step": 11320 }, { "epoch": 3.4, "learning_rate": 2.8748497596153846e-05, "loss": 0.1978, "step": 11321 }, { "epoch": 3.4, "learning_rate": 2.8746619591346158e-05, "loss": 0.1646, "step": 11322 }, { "epoch": 3.4, "learning_rate": 2.8744741586538466e-05, "loss": 0.1821, "step": 11323 }, { "epoch": 3.4, "learning_rate": 2.8742863581730774e-05, "loss": 0.1262, "step": 11324 }, { "epoch": 3.4, "learning_rate": 2.874098557692308e-05, "loss": 0.1839, "step": 11325 }, { "epoch": 3.4, "learning_rate": 2.8739107572115387e-05, "loss": 0.1784, "step": 11326 }, { "epoch": 3.4, "learning_rate": 2.8737229567307695e-05, "loss": 0.1398, "step": 11327 }, { "epoch": 3.4, "learning_rate": 2.87353515625e-05, "loss": 0.1841, "step": 11328 }, { "epoch": 3.4, "learning_rate": 2.8733473557692308e-05, "loss": 0.2016, "step": 11329 }, { "epoch": 3.4, "learning_rate": 2.8731595552884616e-05, "loss": 0.1472, "step": 11330 }, { "epoch": 3.4, "learning_rate": 2.8729717548076924e-05, "loss": 0.1333, "step": 11331 }, { "epoch": 3.41, "learning_rate": 2.872783954326923e-05, "loss": 0.1464, "step": 11332 }, { "epoch": 3.41, "learning_rate": 2.8725961538461537e-05, "loss": 0.1286, "step": 11333 }, { "epoch": 3.41, "learning_rate": 2.8724083533653845e-05, "loss": 0.2241, "step": 11334 }, { "epoch": 3.41, "learning_rate": 2.8722205528846157e-05, "loss": 0.145, "step": 11335 }, { "epoch": 3.41, "learning_rate": 2.8720327524038465e-05, "loss": 0.1462, "step": 11336 }, { "epoch": 3.41, "learning_rate": 2.8718449519230773e-05, "loss": 0.0972, "step": 11337 }, { "epoch": 3.41, "learning_rate": 2.8716571514423078e-05, "loss": 0.1287, "step": 11338 }, { "epoch": 3.41, "learning_rate": 2.8714693509615386e-05, "loss": 0.2427, "step": 11339 }, { "epoch": 3.41, "learning_rate": 2.8712815504807694e-05, "loss": 0.1632, "step": 11340 }, { "epoch": 3.41, "learning_rate": 2.8710937500000002e-05, "loss": 0.1833, "step": 11341 }, { "epoch": 3.41, "learning_rate": 2.8709059495192307e-05, "loss": 0.1458, "step": 11342 }, { "epoch": 3.41, "learning_rate": 2.8707181490384615e-05, "loss": 0.1984, "step": 11343 }, { "epoch": 3.41, "learning_rate": 2.8705303485576924e-05, "loss": 0.1592, "step": 11344 }, { "epoch": 3.41, "learning_rate": 2.8703425480769232e-05, "loss": 0.1619, "step": 11345 }, { "epoch": 3.41, "learning_rate": 2.8701547475961536e-05, "loss": 0.1954, "step": 11346 }, { "epoch": 3.41, "learning_rate": 2.8699669471153845e-05, "loss": 0.1296, "step": 11347 }, { "epoch": 3.41, "learning_rate": 2.8697791466346156e-05, "loss": 0.161, "step": 11348 }, { "epoch": 3.41, "learning_rate": 2.8695913461538464e-05, "loss": 0.1258, "step": 11349 }, { "epoch": 3.41, "learning_rate": 2.8694035456730773e-05, "loss": 0.1321, "step": 11350 }, { "epoch": 3.41, "learning_rate": 2.869215745192308e-05, "loss": 0.2218, "step": 11351 }, { "epoch": 3.41, "learning_rate": 2.8690279447115385e-05, "loss": 0.1198, "step": 11352 }, { "epoch": 3.41, "learning_rate": 2.8688401442307694e-05, "loss": 0.1535, "step": 11353 }, { "epoch": 3.41, "learning_rate": 2.8686523437500002e-05, "loss": 0.105, "step": 11354 }, { "epoch": 3.41, "learning_rate": 2.868464543269231e-05, "loss": 0.2101, "step": 11355 }, { "epoch": 3.41, "learning_rate": 2.8682767427884615e-05, "loss": 0.1275, "step": 11356 }, { "epoch": 3.41, "learning_rate": 2.8680889423076923e-05, "loss": 0.1511, "step": 11357 }, { "epoch": 3.41, "learning_rate": 2.867901141826923e-05, "loss": 0.1652, "step": 11358 }, { "epoch": 3.41, "learning_rate": 2.867713341346154e-05, "loss": 0.221, "step": 11359 }, { "epoch": 3.41, "learning_rate": 2.8675255408653844e-05, "loss": 0.1482, "step": 11360 }, { "epoch": 3.41, "learning_rate": 2.867337740384616e-05, "loss": 0.2123, "step": 11361 }, { "epoch": 3.41, "learning_rate": 2.8671499399038464e-05, "loss": 0.1779, "step": 11362 }, { "epoch": 3.41, "learning_rate": 2.8669621394230772e-05, "loss": 0.1606, "step": 11363 }, { "epoch": 3.41, "learning_rate": 2.866774338942308e-05, "loss": 0.1431, "step": 11364 }, { "epoch": 3.41, "learning_rate": 2.8665865384615388e-05, "loss": 0.1221, "step": 11365 }, { "epoch": 3.42, "learning_rate": 2.8663987379807693e-05, "loss": 0.1149, "step": 11366 }, { "epoch": 3.42, "learning_rate": 2.8662109375e-05, "loss": 0.1448, "step": 11367 }, { "epoch": 3.42, "learning_rate": 2.866023137019231e-05, "loss": 0.166, "step": 11368 }, { "epoch": 3.42, "learning_rate": 2.8658353365384617e-05, "loss": 0.187, "step": 11369 }, { "epoch": 3.42, "learning_rate": 2.8656475360576922e-05, "loss": 0.1315, "step": 11370 }, { "epoch": 3.42, "learning_rate": 2.865459735576923e-05, "loss": 0.1074, "step": 11371 }, { "epoch": 3.42, "learning_rate": 2.865271935096154e-05, "loss": 0.1613, "step": 11372 }, { "epoch": 3.42, "learning_rate": 2.8650841346153847e-05, "loss": 0.1493, "step": 11373 }, { "epoch": 3.42, "learning_rate": 2.8648963341346158e-05, "loss": 0.1841, "step": 11374 }, { "epoch": 3.42, "learning_rate": 2.8647085336538466e-05, "loss": 0.1625, "step": 11375 }, { "epoch": 3.42, "learning_rate": 2.864520733173077e-05, "loss": 0.145, "step": 11376 }, { "epoch": 3.42, "learning_rate": 2.864332932692308e-05, "loss": 0.2022, "step": 11377 }, { "epoch": 3.42, "learning_rate": 2.8641451322115387e-05, "loss": 0.217, "step": 11378 }, { "epoch": 3.42, "learning_rate": 2.8639573317307696e-05, "loss": 0.1474, "step": 11379 }, { "epoch": 3.42, "learning_rate": 2.86376953125e-05, "loss": 0.1279, "step": 11380 }, { "epoch": 3.42, "learning_rate": 2.863581730769231e-05, "loss": 0.1356, "step": 11381 }, { "epoch": 3.42, "learning_rate": 2.8633939302884617e-05, "loss": 0.2039, "step": 11382 }, { "epoch": 3.42, "learning_rate": 2.8632061298076925e-05, "loss": 0.1286, "step": 11383 }, { "epoch": 3.42, "learning_rate": 2.863018329326923e-05, "loss": 0.1701, "step": 11384 }, { "epoch": 3.42, "learning_rate": 2.8628305288461538e-05, "loss": 0.1593, "step": 11385 }, { "epoch": 3.42, "learning_rate": 2.8626427283653846e-05, "loss": 0.0884, "step": 11386 }, { "epoch": 3.42, "learning_rate": 2.8624549278846157e-05, "loss": 0.1491, "step": 11387 }, { "epoch": 3.42, "learning_rate": 2.8622671274038466e-05, "loss": 0.118, "step": 11388 }, { "epoch": 3.42, "learning_rate": 2.8620793269230774e-05, "loss": 0.1619, "step": 11389 }, { "epoch": 3.42, "learning_rate": 2.861891526442308e-05, "loss": 0.1326, "step": 11390 }, { "epoch": 3.42, "learning_rate": 2.8617037259615387e-05, "loss": 0.1257, "step": 11391 }, { "epoch": 3.42, "learning_rate": 2.8615159254807695e-05, "loss": 0.0787, "step": 11392 }, { "epoch": 3.42, "learning_rate": 2.8613281250000003e-05, "loss": 0.146, "step": 11393 }, { "epoch": 3.42, "learning_rate": 2.8611403245192308e-05, "loss": 0.1908, "step": 11394 }, { "epoch": 3.42, "learning_rate": 2.8609525240384616e-05, "loss": 0.2272, "step": 11395 }, { "epoch": 3.42, "learning_rate": 2.8607647235576924e-05, "loss": 0.1451, "step": 11396 }, { "epoch": 3.42, "learning_rate": 2.860576923076923e-05, "loss": 0.1862, "step": 11397 }, { "epoch": 3.42, "learning_rate": 2.8603891225961537e-05, "loss": 0.1383, "step": 11398 }, { "epoch": 3.43, "learning_rate": 2.8602013221153845e-05, "loss": 0.196, "step": 11399 }, { "epoch": 3.43, "learning_rate": 2.8600135216346157e-05, "loss": 0.2841, "step": 11400 }, { "epoch": 3.43, "learning_rate": 2.8598257211538465e-05, "loss": 0.2416, "step": 11401 }, { "epoch": 3.43, "learning_rate": 2.8596379206730773e-05, "loss": 0.1182, "step": 11402 }, { "epoch": 3.43, "learning_rate": 2.859450120192308e-05, "loss": 0.1149, "step": 11403 }, { "epoch": 3.43, "learning_rate": 2.8592623197115386e-05, "loss": 0.1567, "step": 11404 }, { "epoch": 3.43, "learning_rate": 2.8590745192307694e-05, "loss": 0.1579, "step": 11405 }, { "epoch": 3.43, "learning_rate": 2.8588867187500002e-05, "loss": 0.1277, "step": 11406 }, { "epoch": 3.43, "learning_rate": 2.8586989182692307e-05, "loss": 0.2196, "step": 11407 }, { "epoch": 3.43, "learning_rate": 2.8585111177884615e-05, "loss": 0.1575, "step": 11408 }, { "epoch": 3.43, "learning_rate": 2.8583233173076923e-05, "loss": 0.2144, "step": 11409 }, { "epoch": 3.43, "learning_rate": 2.858135516826923e-05, "loss": 0.125, "step": 11410 }, { "epoch": 3.43, "learning_rate": 2.8579477163461536e-05, "loss": 0.1342, "step": 11411 }, { "epoch": 3.43, "learning_rate": 2.8577599158653845e-05, "loss": 0.126, "step": 11412 }, { "epoch": 3.43, "learning_rate": 2.857572115384616e-05, "loss": 0.1592, "step": 11413 }, { "epoch": 3.43, "learning_rate": 2.8573843149038464e-05, "loss": 0.1492, "step": 11414 }, { "epoch": 3.43, "learning_rate": 2.8571965144230772e-05, "loss": 0.1504, "step": 11415 }, { "epoch": 3.43, "learning_rate": 2.857008713942308e-05, "loss": 0.1799, "step": 11416 }, { "epoch": 3.43, "learning_rate": 2.8568209134615385e-05, "loss": 0.127, "step": 11417 }, { "epoch": 3.43, "learning_rate": 2.8566331129807693e-05, "loss": 0.2527, "step": 11418 }, { "epoch": 3.43, "learning_rate": 2.8564453125e-05, "loss": 0.1208, "step": 11419 }, { "epoch": 3.43, "learning_rate": 2.856257512019231e-05, "loss": 0.1149, "step": 11420 }, { "epoch": 3.43, "learning_rate": 2.8560697115384615e-05, "loss": 0.1746, "step": 11421 }, { "epoch": 3.43, "learning_rate": 2.8558819110576923e-05, "loss": 0.1846, "step": 11422 }, { "epoch": 3.43, "learning_rate": 2.855694110576923e-05, "loss": 0.1898, "step": 11423 }, { "epoch": 3.43, "learning_rate": 2.855506310096154e-05, "loss": 0.1912, "step": 11424 }, { "epoch": 3.43, "learning_rate": 2.8553185096153844e-05, "loss": 0.1282, "step": 11425 }, { "epoch": 3.43, "learning_rate": 2.855130709134616e-05, "loss": 0.1746, "step": 11426 }, { "epoch": 3.43, "learning_rate": 2.8549429086538464e-05, "loss": 0.2173, "step": 11427 }, { "epoch": 3.43, "learning_rate": 2.8547551081730772e-05, "loss": 0.1502, "step": 11428 }, { "epoch": 3.43, "learning_rate": 2.854567307692308e-05, "loss": 0.1755, "step": 11429 }, { "epoch": 3.43, "learning_rate": 2.8543795072115388e-05, "loss": 0.1308, "step": 11430 }, { "epoch": 3.43, "learning_rate": 2.8541917067307693e-05, "loss": 0.1377, "step": 11431 }, { "epoch": 3.44, "learning_rate": 2.85400390625e-05, "loss": 0.1566, "step": 11432 }, { "epoch": 3.44, "learning_rate": 2.853816105769231e-05, "loss": 0.1223, "step": 11433 }, { "epoch": 3.44, "learning_rate": 2.8536283052884617e-05, "loss": 0.1267, "step": 11434 }, { "epoch": 3.44, "learning_rate": 2.8534405048076922e-05, "loss": 0.2119, "step": 11435 }, { "epoch": 3.44, "learning_rate": 2.853252704326923e-05, "loss": 0.1396, "step": 11436 }, { "epoch": 3.44, "learning_rate": 2.853064903846154e-05, "loss": 0.2268, "step": 11437 }, { "epoch": 3.44, "learning_rate": 2.8528771033653847e-05, "loss": 0.0745, "step": 11438 }, { "epoch": 3.44, "learning_rate": 2.8526893028846158e-05, "loss": 0.1989, "step": 11439 }, { "epoch": 3.44, "learning_rate": 2.8525015024038466e-05, "loss": 0.1386, "step": 11440 }, { "epoch": 3.44, "learning_rate": 2.852313701923077e-05, "loss": 0.1671, "step": 11441 }, { "epoch": 3.44, "learning_rate": 2.852125901442308e-05, "loss": 0.1354, "step": 11442 }, { "epoch": 3.44, "learning_rate": 2.8519381009615387e-05, "loss": 0.1508, "step": 11443 }, { "epoch": 3.44, "learning_rate": 2.8517503004807695e-05, "loss": 0.103, "step": 11444 }, { "epoch": 3.44, "learning_rate": 2.8515625e-05, "loss": 0.1593, "step": 11445 }, { "epoch": 3.44, "learning_rate": 2.851374699519231e-05, "loss": 0.1303, "step": 11446 }, { "epoch": 3.44, "learning_rate": 2.8511868990384617e-05, "loss": 0.1531, "step": 11447 }, { "epoch": 3.44, "learning_rate": 2.8509990985576925e-05, "loss": 0.1552, "step": 11448 }, { "epoch": 3.44, "learning_rate": 2.850811298076923e-05, "loss": 0.1296, "step": 11449 }, { "epoch": 3.44, "learning_rate": 2.8506234975961538e-05, "loss": 0.1161, "step": 11450 }, { "epoch": 3.44, "learning_rate": 2.8504356971153846e-05, "loss": 0.1479, "step": 11451 }, { "epoch": 3.44, "learning_rate": 2.8502478966346157e-05, "loss": 0.1561, "step": 11452 }, { "epoch": 3.44, "learning_rate": 2.8500600961538466e-05, "loss": 0.1236, "step": 11453 }, { "epoch": 3.44, "learning_rate": 2.8498722956730774e-05, "loss": 0.1835, "step": 11454 }, { "epoch": 3.44, "learning_rate": 2.849684495192308e-05, "loss": 0.1495, "step": 11455 }, { "epoch": 3.44, "learning_rate": 2.8494966947115387e-05, "loss": 0.1511, "step": 11456 }, { "epoch": 3.44, "learning_rate": 2.8493088942307695e-05, "loss": 0.1301, "step": 11457 }, { "epoch": 3.44, "learning_rate": 2.8491210937500003e-05, "loss": 0.1392, "step": 11458 }, { "epoch": 3.44, "learning_rate": 2.8489332932692308e-05, "loss": 0.1387, "step": 11459 }, { "epoch": 3.44, "learning_rate": 2.8487454927884616e-05, "loss": 0.2655, "step": 11460 }, { "epoch": 3.44, "learning_rate": 2.8485576923076924e-05, "loss": 0.2196, "step": 11461 }, { "epoch": 3.44, "learning_rate": 2.848369891826923e-05, "loss": 0.1748, "step": 11462 }, { "epoch": 3.44, "learning_rate": 2.8481820913461537e-05, "loss": 0.1472, "step": 11463 }, { "epoch": 3.44, "learning_rate": 2.8479942908653845e-05, "loss": 0.2043, "step": 11464 }, { "epoch": 3.45, "learning_rate": 2.8478064903846157e-05, "loss": 0.1733, "step": 11465 }, { "epoch": 3.45, "learning_rate": 2.8476186899038465e-05, "loss": 0.1377, "step": 11466 }, { "epoch": 3.45, "learning_rate": 2.8474308894230773e-05, "loss": 0.1643, "step": 11467 }, { "epoch": 3.45, "learning_rate": 2.847243088942308e-05, "loss": 0.1749, "step": 11468 }, { "epoch": 3.45, "learning_rate": 2.8470552884615386e-05, "loss": 0.1505, "step": 11469 }, { "epoch": 3.45, "learning_rate": 2.8468674879807694e-05, "loss": 0.1908, "step": 11470 }, { "epoch": 3.45, "learning_rate": 2.8466796875000002e-05, "loss": 0.1895, "step": 11471 }, { "epoch": 3.45, "learning_rate": 2.8464918870192307e-05, "loss": 0.1227, "step": 11472 }, { "epoch": 3.45, "learning_rate": 2.8463040865384615e-05, "loss": 0.1307, "step": 11473 }, { "epoch": 3.45, "learning_rate": 2.8461162860576923e-05, "loss": 0.1235, "step": 11474 }, { "epoch": 3.45, "learning_rate": 2.845928485576923e-05, "loss": 0.1289, "step": 11475 }, { "epoch": 3.45, "learning_rate": 2.8457406850961536e-05, "loss": 0.095, "step": 11476 }, { "epoch": 3.45, "learning_rate": 2.8455528846153844e-05, "loss": 0.1537, "step": 11477 }, { "epoch": 3.45, "learning_rate": 2.845365084134616e-05, "loss": 0.1477, "step": 11478 }, { "epoch": 3.45, "learning_rate": 2.8451772836538464e-05, "loss": 0.1268, "step": 11479 }, { "epoch": 3.45, "learning_rate": 2.8449894831730772e-05, "loss": 0.172, "step": 11480 }, { "epoch": 3.45, "learning_rate": 2.844801682692308e-05, "loss": 0.1141, "step": 11481 }, { "epoch": 3.45, "learning_rate": 2.8446138822115385e-05, "loss": 0.1202, "step": 11482 }, { "epoch": 3.45, "learning_rate": 2.8444260817307693e-05, "loss": 0.1435, "step": 11483 }, { "epoch": 3.45, "learning_rate": 2.84423828125e-05, "loss": 0.1782, "step": 11484 }, { "epoch": 3.45, "learning_rate": 2.844050480769231e-05, "loss": 0.0947, "step": 11485 }, { "epoch": 3.45, "learning_rate": 2.8438626802884614e-05, "loss": 0.1191, "step": 11486 }, { "epoch": 3.45, "learning_rate": 2.8436748798076923e-05, "loss": 0.12, "step": 11487 }, { "epoch": 3.45, "learning_rate": 2.843487079326923e-05, "loss": 0.1347, "step": 11488 }, { "epoch": 3.45, "learning_rate": 2.843299278846154e-05, "loss": 0.2018, "step": 11489 }, { "epoch": 3.45, "learning_rate": 2.8431114783653844e-05, "loss": 0.1328, "step": 11490 }, { "epoch": 3.45, "learning_rate": 2.842923677884616e-05, "loss": 0.082, "step": 11491 }, { "epoch": 3.45, "learning_rate": 2.8427358774038463e-05, "loss": 0.1384, "step": 11492 }, { "epoch": 3.45, "learning_rate": 2.842548076923077e-05, "loss": 0.188, "step": 11493 }, { "epoch": 3.45, "learning_rate": 2.842360276442308e-05, "loss": 0.1288, "step": 11494 }, { "epoch": 3.45, "learning_rate": 2.8421724759615388e-05, "loss": 0.1402, "step": 11495 }, { "epoch": 3.45, "learning_rate": 2.8419846754807693e-05, "loss": 0.1374, "step": 11496 }, { "epoch": 3.45, "learning_rate": 2.841796875e-05, "loss": 0.1626, "step": 11497 }, { "epoch": 3.45, "learning_rate": 2.841609074519231e-05, "loss": 0.1895, "step": 11498 }, { "epoch": 3.46, "learning_rate": 2.8414212740384617e-05, "loss": 0.1692, "step": 11499 }, { "epoch": 3.46, "learning_rate": 2.8412334735576922e-05, "loss": 0.1457, "step": 11500 }, { "epoch": 3.46, "learning_rate": 2.841045673076923e-05, "loss": 0.1826, "step": 11501 }, { "epoch": 3.46, "learning_rate": 2.8408578725961538e-05, "loss": 0.1827, "step": 11502 }, { "epoch": 3.46, "learning_rate": 2.8406700721153846e-05, "loss": 0.1417, "step": 11503 }, { "epoch": 3.46, "learning_rate": 2.8404822716346158e-05, "loss": 0.1355, "step": 11504 }, { "epoch": 3.46, "learning_rate": 2.8402944711538466e-05, "loss": 0.122, "step": 11505 }, { "epoch": 3.46, "learning_rate": 2.840106670673077e-05, "loss": 0.1155, "step": 11506 }, { "epoch": 3.46, "learning_rate": 2.839918870192308e-05, "loss": 0.1147, "step": 11507 }, { "epoch": 3.46, "learning_rate": 2.8397310697115387e-05, "loss": 0.2143, "step": 11508 }, { "epoch": 3.46, "learning_rate": 2.8395432692307695e-05, "loss": 0.1696, "step": 11509 }, { "epoch": 3.46, "learning_rate": 2.83935546875e-05, "loss": 0.1446, "step": 11510 }, { "epoch": 3.46, "learning_rate": 2.839167668269231e-05, "loss": 0.1534, "step": 11511 }, { "epoch": 3.46, "learning_rate": 2.8389798677884616e-05, "loss": 0.1515, "step": 11512 }, { "epoch": 3.46, "learning_rate": 2.8387920673076925e-05, "loss": 0.1792, "step": 11513 }, { "epoch": 3.46, "learning_rate": 2.838604266826923e-05, "loss": 0.1441, "step": 11514 }, { "epoch": 3.46, "learning_rate": 2.8384164663461538e-05, "loss": 0.1819, "step": 11515 }, { "epoch": 3.46, "learning_rate": 2.8382286658653846e-05, "loss": 0.1088, "step": 11516 }, { "epoch": 3.46, "learning_rate": 2.8380408653846157e-05, "loss": 0.1656, "step": 11517 }, { "epoch": 3.46, "learning_rate": 2.8378530649038465e-05, "loss": 0.1481, "step": 11518 }, { "epoch": 3.46, "learning_rate": 2.8376652644230774e-05, "loss": 0.1051, "step": 11519 }, { "epoch": 3.46, "learning_rate": 2.837477463942308e-05, "loss": 0.1295, "step": 11520 }, { "epoch": 3.46, "learning_rate": 2.8372896634615387e-05, "loss": 0.1928, "step": 11521 }, { "epoch": 3.46, "learning_rate": 2.8371018629807695e-05, "loss": 0.1609, "step": 11522 }, { "epoch": 3.46, "learning_rate": 2.8369140625000003e-05, "loss": 0.1575, "step": 11523 }, { "epoch": 3.46, "learning_rate": 2.8367262620192308e-05, "loss": 0.2162, "step": 11524 }, { "epoch": 3.46, "learning_rate": 2.8365384615384616e-05, "loss": 0.1425, "step": 11525 }, { "epoch": 3.46, "learning_rate": 2.8363506610576924e-05, "loss": 0.1331, "step": 11526 }, { "epoch": 3.46, "learning_rate": 2.8361628605769232e-05, "loss": 0.1503, "step": 11527 }, { "epoch": 3.46, "learning_rate": 2.8359750600961537e-05, "loss": 0.1824, "step": 11528 }, { "epoch": 3.46, "learning_rate": 2.8357872596153845e-05, "loss": 0.1744, "step": 11529 }, { "epoch": 3.46, "learning_rate": 2.8355994591346157e-05, "loss": 0.1276, "step": 11530 }, { "epoch": 3.46, "learning_rate": 2.8354116586538465e-05, "loss": 0.1227, "step": 11531 }, { "epoch": 3.47, "learning_rate": 2.8352238581730773e-05, "loss": 0.1312, "step": 11532 }, { "epoch": 3.47, "learning_rate": 2.835036057692308e-05, "loss": 0.1443, "step": 11533 }, { "epoch": 3.47, "learning_rate": 2.8348482572115386e-05, "loss": 0.0948, "step": 11534 }, { "epoch": 3.47, "learning_rate": 2.8346604567307694e-05, "loss": 0.1314, "step": 11535 }, { "epoch": 3.47, "learning_rate": 2.8344726562500002e-05, "loss": 0.1792, "step": 11536 }, { "epoch": 3.47, "learning_rate": 2.834284855769231e-05, "loss": 0.1502, "step": 11537 }, { "epoch": 3.47, "learning_rate": 2.8340970552884615e-05, "loss": 0.2979, "step": 11538 }, { "epoch": 3.47, "learning_rate": 2.8339092548076923e-05, "loss": 0.1211, "step": 11539 }, { "epoch": 3.47, "learning_rate": 2.833721454326923e-05, "loss": 0.1248, "step": 11540 }, { "epoch": 3.47, "learning_rate": 2.8335336538461536e-05, "loss": 0.1371, "step": 11541 }, { "epoch": 3.47, "learning_rate": 2.8333458533653844e-05, "loss": 0.1341, "step": 11542 }, { "epoch": 3.47, "learning_rate": 2.833158052884616e-05, "loss": 0.1179, "step": 11543 }, { "epoch": 3.47, "learning_rate": 2.8329702524038464e-05, "loss": 0.1652, "step": 11544 }, { "epoch": 3.47, "learning_rate": 2.8327824519230772e-05, "loss": 0.1711, "step": 11545 }, { "epoch": 3.47, "learning_rate": 2.832594651442308e-05, "loss": 0.1758, "step": 11546 }, { "epoch": 3.47, "learning_rate": 2.832406850961539e-05, "loss": 0.1122, "step": 11547 }, { "epoch": 3.47, "learning_rate": 2.8322190504807693e-05, "loss": 0.144, "step": 11548 }, { "epoch": 3.47, "learning_rate": 2.83203125e-05, "loss": 0.1503, "step": 11549 }, { "epoch": 3.47, "learning_rate": 2.831843449519231e-05, "loss": 0.1421, "step": 11550 }, { "epoch": 3.47, "learning_rate": 2.8316556490384614e-05, "loss": 0.1469, "step": 11551 }, { "epoch": 3.47, "learning_rate": 2.8314678485576923e-05, "loss": 0.1175, "step": 11552 }, { "epoch": 3.47, "learning_rate": 2.831280048076923e-05, "loss": 0.135, "step": 11553 }, { "epoch": 3.47, "learning_rate": 2.831092247596154e-05, "loss": 0.1452, "step": 11554 }, { "epoch": 3.47, "learning_rate": 2.8309044471153844e-05, "loss": 0.1296, "step": 11555 }, { "epoch": 3.47, "learning_rate": 2.830716646634616e-05, "loss": 0.1447, "step": 11556 }, { "epoch": 3.47, "learning_rate": 2.8305288461538467e-05, "loss": 0.1598, "step": 11557 }, { "epoch": 3.47, "learning_rate": 2.830341045673077e-05, "loss": 0.1532, "step": 11558 }, { "epoch": 3.47, "learning_rate": 2.830153245192308e-05, "loss": 0.2091, "step": 11559 }, { "epoch": 3.47, "learning_rate": 2.8299654447115388e-05, "loss": 0.1683, "step": 11560 }, { "epoch": 3.47, "learning_rate": 2.8297776442307693e-05, "loss": 0.167, "step": 11561 }, { "epoch": 3.47, "learning_rate": 2.82958984375e-05, "loss": 0.1952, "step": 11562 }, { "epoch": 3.47, "learning_rate": 2.829402043269231e-05, "loss": 0.084, "step": 11563 }, { "epoch": 3.47, "learning_rate": 2.8292142427884617e-05, "loss": 0.0995, "step": 11564 }, { "epoch": 3.48, "learning_rate": 2.8290264423076922e-05, "loss": 0.1663, "step": 11565 }, { "epoch": 3.48, "learning_rate": 2.828838641826923e-05, "loss": 0.1984, "step": 11566 }, { "epoch": 3.48, "learning_rate": 2.8286508413461538e-05, "loss": 0.1444, "step": 11567 }, { "epoch": 3.48, "learning_rate": 2.8284630408653846e-05, "loss": 0.1823, "step": 11568 }, { "epoch": 3.48, "learning_rate": 2.8282752403846158e-05, "loss": 0.2007, "step": 11569 }, { "epoch": 3.48, "learning_rate": 2.8280874399038466e-05, "loss": 0.1345, "step": 11570 }, { "epoch": 3.48, "learning_rate": 2.827899639423077e-05, "loss": 0.1144, "step": 11571 }, { "epoch": 3.48, "learning_rate": 2.827711838942308e-05, "loss": 0.1723, "step": 11572 }, { "epoch": 3.48, "learning_rate": 2.8275240384615387e-05, "loss": 0.1657, "step": 11573 }, { "epoch": 3.48, "learning_rate": 2.8273362379807695e-05, "loss": 0.1269, "step": 11574 }, { "epoch": 3.48, "learning_rate": 2.8271484375e-05, "loss": 0.154, "step": 11575 }, { "epoch": 3.48, "learning_rate": 2.8269606370192308e-05, "loss": 0.0995, "step": 11576 }, { "epoch": 3.48, "learning_rate": 2.8267728365384616e-05, "loss": 0.1425, "step": 11577 }, { "epoch": 3.48, "learning_rate": 2.8265850360576925e-05, "loss": 0.2091, "step": 11578 }, { "epoch": 3.48, "learning_rate": 2.826397235576923e-05, "loss": 0.1776, "step": 11579 }, { "epoch": 3.48, "learning_rate": 2.8262094350961537e-05, "loss": 0.1479, "step": 11580 }, { "epoch": 3.48, "learning_rate": 2.8260216346153846e-05, "loss": 0.1196, "step": 11581 }, { "epoch": 3.48, "learning_rate": 2.8258338341346157e-05, "loss": 0.1584, "step": 11582 }, { "epoch": 3.48, "learning_rate": 2.8256460336538465e-05, "loss": 0.1061, "step": 11583 }, { "epoch": 3.48, "learning_rate": 2.8254582331730773e-05, "loss": 0.1458, "step": 11584 }, { "epoch": 3.48, "learning_rate": 2.8252704326923078e-05, "loss": 0.1149, "step": 11585 }, { "epoch": 3.48, "learning_rate": 2.8250826322115386e-05, "loss": 0.1245, "step": 11586 }, { "epoch": 3.48, "learning_rate": 2.8248948317307695e-05, "loss": 0.1873, "step": 11587 }, { "epoch": 3.48, "learning_rate": 2.8247070312500003e-05, "loss": 0.1127, "step": 11588 }, { "epoch": 3.48, "learning_rate": 2.8245192307692307e-05, "loss": 0.1297, "step": 11589 }, { "epoch": 3.48, "learning_rate": 2.8243314302884616e-05, "loss": 0.1421, "step": 11590 }, { "epoch": 3.48, "learning_rate": 2.8241436298076924e-05, "loss": 0.2133, "step": 11591 }, { "epoch": 3.48, "learning_rate": 2.8239558293269232e-05, "loss": 0.1161, "step": 11592 }, { "epoch": 3.48, "learning_rate": 2.8237680288461537e-05, "loss": 0.1148, "step": 11593 }, { "epoch": 3.48, "learning_rate": 2.8235802283653845e-05, "loss": 0.1519, "step": 11594 }, { "epoch": 3.48, "learning_rate": 2.8233924278846156e-05, "loss": 0.1423, "step": 11595 }, { "epoch": 3.48, "learning_rate": 2.8232046274038465e-05, "loss": 0.1115, "step": 11596 }, { "epoch": 3.48, "learning_rate": 2.8230168269230773e-05, "loss": 0.1663, "step": 11597 }, { "epoch": 3.48, "learning_rate": 2.822829026442308e-05, "loss": 0.1279, "step": 11598 }, { "epoch": 3.49, "learning_rate": 2.8226412259615386e-05, "loss": 0.1643, "step": 11599 }, { "epoch": 3.49, "learning_rate": 2.8224534254807694e-05, "loss": 0.1797, "step": 11600 }, { "epoch": 3.49, "learning_rate": 2.8222656250000002e-05, "loss": 0.124, "step": 11601 }, { "epoch": 3.49, "learning_rate": 2.822077824519231e-05, "loss": 0.2297, "step": 11602 }, { "epoch": 3.49, "learning_rate": 2.8218900240384615e-05, "loss": 0.1264, "step": 11603 }, { "epoch": 3.49, "learning_rate": 2.8217022235576923e-05, "loss": 0.1311, "step": 11604 }, { "epoch": 3.49, "learning_rate": 2.821514423076923e-05, "loss": 0.1204, "step": 11605 }, { "epoch": 3.49, "learning_rate": 2.8213266225961536e-05, "loss": 0.1512, "step": 11606 }, { "epoch": 3.49, "learning_rate": 2.8211388221153844e-05, "loss": 0.1627, "step": 11607 }, { "epoch": 3.49, "learning_rate": 2.820951021634616e-05, "loss": 0.1305, "step": 11608 }, { "epoch": 3.49, "learning_rate": 2.8207632211538464e-05, "loss": 0.2082, "step": 11609 }, { "epoch": 3.49, "learning_rate": 2.8205754206730772e-05, "loss": 0.167, "step": 11610 }, { "epoch": 3.49, "learning_rate": 2.820387620192308e-05, "loss": 0.1888, "step": 11611 }, { "epoch": 3.49, "learning_rate": 2.820199819711539e-05, "loss": 0.15, "step": 11612 }, { "epoch": 3.49, "learning_rate": 2.8200120192307693e-05, "loss": 0.1535, "step": 11613 }, { "epoch": 3.49, "learning_rate": 2.81982421875e-05, "loss": 0.1563, "step": 11614 }, { "epoch": 3.49, "learning_rate": 2.819636418269231e-05, "loss": 0.1541, "step": 11615 }, { "epoch": 3.49, "learning_rate": 2.8194486177884614e-05, "loss": 0.112, "step": 11616 }, { "epoch": 3.49, "learning_rate": 2.8192608173076922e-05, "loss": 0.1647, "step": 11617 }, { "epoch": 3.49, "learning_rate": 2.819073016826923e-05, "loss": 0.1484, "step": 11618 }, { "epoch": 3.49, "learning_rate": 2.818885216346154e-05, "loss": 0.1827, "step": 11619 }, { "epoch": 3.49, "learning_rate": 2.8186974158653844e-05, "loss": 0.1863, "step": 11620 }, { "epoch": 3.49, "learning_rate": 2.818509615384616e-05, "loss": 0.1406, "step": 11621 }, { "epoch": 3.49, "learning_rate": 2.8183218149038467e-05, "loss": 0.177, "step": 11622 }, { "epoch": 3.49, "learning_rate": 2.818134014423077e-05, "loss": 0.1394, "step": 11623 }, { "epoch": 3.49, "learning_rate": 2.817946213942308e-05, "loss": 0.1373, "step": 11624 }, { "epoch": 3.49, "learning_rate": 2.8177584134615388e-05, "loss": 0.3123, "step": 11625 }, { "epoch": 3.49, "learning_rate": 2.8175706129807692e-05, "loss": 0.1807, "step": 11626 }, { "epoch": 3.49, "learning_rate": 2.8173828125e-05, "loss": 0.2267, "step": 11627 }, { "epoch": 3.49, "learning_rate": 2.817195012019231e-05, "loss": 0.1247, "step": 11628 }, { "epoch": 3.49, "learning_rate": 2.8170072115384617e-05, "loss": 0.1224, "step": 11629 }, { "epoch": 3.49, "learning_rate": 2.8168194110576922e-05, "loss": 0.1209, "step": 11630 }, { "epoch": 3.49, "learning_rate": 2.816631610576923e-05, "loss": 0.1622, "step": 11631 }, { "epoch": 3.5, "learning_rate": 2.8164438100961538e-05, "loss": 0.2522, "step": 11632 }, { "epoch": 3.5, "learning_rate": 2.8162560096153846e-05, "loss": 0.1134, "step": 11633 }, { "epoch": 3.5, "learning_rate": 2.8160682091346158e-05, "loss": 0.1936, "step": 11634 }, { "epoch": 3.5, "learning_rate": 2.8158804086538466e-05, "loss": 0.1941, "step": 11635 }, { "epoch": 3.5, "learning_rate": 2.815692608173077e-05, "loss": 0.1796, "step": 11636 }, { "epoch": 3.5, "learning_rate": 2.815504807692308e-05, "loss": 0.1545, "step": 11637 }, { "epoch": 3.5, "learning_rate": 2.8153170072115387e-05, "loss": 0.1192, "step": 11638 }, { "epoch": 3.5, "learning_rate": 2.8151292067307695e-05, "loss": 0.165, "step": 11639 }, { "epoch": 3.5, "learning_rate": 2.81494140625e-05, "loss": 0.2116, "step": 11640 }, { "epoch": 3.5, "learning_rate": 2.8147536057692308e-05, "loss": 0.1109, "step": 11641 }, { "epoch": 3.5, "learning_rate": 2.8145658052884616e-05, "loss": 0.1285, "step": 11642 }, { "epoch": 3.5, "learning_rate": 2.8143780048076924e-05, "loss": 0.1209, "step": 11643 }, { "epoch": 3.5, "learning_rate": 2.814190204326923e-05, "loss": 0.1091, "step": 11644 }, { "epoch": 3.5, "learning_rate": 2.8140024038461537e-05, "loss": 0.1523, "step": 11645 }, { "epoch": 3.5, "learning_rate": 2.8138146033653845e-05, "loss": 0.1299, "step": 11646 }, { "epoch": 3.5, "learning_rate": 2.8136268028846157e-05, "loss": 0.1316, "step": 11647 }, { "epoch": 3.5, "learning_rate": 2.8134390024038465e-05, "loss": 0.1204, "step": 11648 }, { "epoch": 3.5, "learning_rate": 2.8132512019230773e-05, "loss": 0.1126, "step": 11649 }, { "epoch": 3.5, "learning_rate": 2.8130634014423078e-05, "loss": 0.1306, "step": 11650 }, { "epoch": 3.5, "learning_rate": 2.8128756009615386e-05, "loss": 0.1498, "step": 11651 }, { "epoch": 3.5, "learning_rate": 2.8126878004807694e-05, "loss": 0.127, "step": 11652 }, { "epoch": 3.5, "learning_rate": 2.8125000000000003e-05, "loss": 0.1298, "step": 11653 }, { "epoch": 3.5, "learning_rate": 2.8123121995192307e-05, "loss": 0.1158, "step": 11654 }, { "epoch": 3.5, "learning_rate": 2.8121243990384616e-05, "loss": 0.1533, "step": 11655 }, { "epoch": 3.5, "learning_rate": 2.8119365985576924e-05, "loss": 0.1023, "step": 11656 }, { "epoch": 3.5, "learning_rate": 2.8117487980769232e-05, "loss": 0.1103, "step": 11657 }, { "epoch": 3.5, "learning_rate": 2.8115609975961537e-05, "loss": 0.1289, "step": 11658 }, { "epoch": 3.5, "learning_rate": 2.8113731971153845e-05, "loss": 0.1132, "step": 11659 }, { "epoch": 3.5, "learning_rate": 2.8111853966346156e-05, "loss": 0.1871, "step": 11660 }, { "epoch": 3.5, "learning_rate": 2.8109975961538465e-05, "loss": 0.1765, "step": 11661 }, { "epoch": 3.5, "learning_rate": 2.8108097956730773e-05, "loss": 0.1361, "step": 11662 }, { "epoch": 3.5, "learning_rate": 2.810621995192308e-05, "loss": 0.1498, "step": 11663 }, { "epoch": 3.5, "learning_rate": 2.8104341947115386e-05, "loss": 0.1765, "step": 11664 }, { "epoch": 3.51, "learning_rate": 2.8102463942307694e-05, "loss": 0.2551, "step": 11665 }, { "epoch": 3.51, "learning_rate": 2.8100585937500002e-05, "loss": 0.1877, "step": 11666 }, { "epoch": 3.51, "learning_rate": 2.809870793269231e-05, "loss": 0.1079, "step": 11667 }, { "epoch": 3.51, "learning_rate": 2.8096829927884615e-05, "loss": 0.1485, "step": 11668 }, { "epoch": 3.51, "learning_rate": 2.8094951923076923e-05, "loss": 0.1051, "step": 11669 }, { "epoch": 3.51, "learning_rate": 2.809307391826923e-05, "loss": 0.1814, "step": 11670 }, { "epoch": 3.51, "learning_rate": 2.809119591346154e-05, "loss": 0.1383, "step": 11671 }, { "epoch": 3.51, "learning_rate": 2.8089317908653844e-05, "loss": 0.1535, "step": 11672 }, { "epoch": 3.51, "learning_rate": 2.808743990384616e-05, "loss": 0.1557, "step": 11673 }, { "epoch": 3.51, "learning_rate": 2.8085561899038464e-05, "loss": 0.1301, "step": 11674 }, { "epoch": 3.51, "learning_rate": 2.8083683894230772e-05, "loss": 0.1556, "step": 11675 }, { "epoch": 3.51, "learning_rate": 2.808180588942308e-05, "loss": 0.1382, "step": 11676 }, { "epoch": 3.51, "learning_rate": 2.8079927884615388e-05, "loss": 0.1345, "step": 11677 }, { "epoch": 3.51, "learning_rate": 2.8078049879807693e-05, "loss": 0.1099, "step": 11678 }, { "epoch": 3.51, "learning_rate": 2.8076171875e-05, "loss": 0.1416, "step": 11679 }, { "epoch": 3.51, "learning_rate": 2.807429387019231e-05, "loss": 0.192, "step": 11680 }, { "epoch": 3.51, "learning_rate": 2.8072415865384618e-05, "loss": 0.196, "step": 11681 }, { "epoch": 3.51, "learning_rate": 2.8070537860576922e-05, "loss": 0.2669, "step": 11682 }, { "epoch": 3.51, "learning_rate": 2.806865985576923e-05, "loss": 0.172, "step": 11683 }, { "epoch": 3.51, "learning_rate": 2.806678185096154e-05, "loss": 0.2121, "step": 11684 }, { "epoch": 3.51, "learning_rate": 2.8064903846153843e-05, "loss": 0.1181, "step": 11685 }, { "epoch": 3.51, "learning_rate": 2.806302584134616e-05, "loss": 0.245, "step": 11686 }, { "epoch": 3.51, "learning_rate": 2.8061147836538466e-05, "loss": 0.1371, "step": 11687 }, { "epoch": 3.51, "learning_rate": 2.805926983173077e-05, "loss": 0.166, "step": 11688 }, { "epoch": 3.51, "learning_rate": 2.805739182692308e-05, "loss": 0.1686, "step": 11689 }, { "epoch": 3.51, "learning_rate": 2.8055513822115388e-05, "loss": 0.1389, "step": 11690 }, { "epoch": 3.51, "learning_rate": 2.8053635817307696e-05, "loss": 0.1329, "step": 11691 }, { "epoch": 3.51, "learning_rate": 2.80517578125e-05, "loss": 0.1429, "step": 11692 }, { "epoch": 3.51, "learning_rate": 2.804987980769231e-05, "loss": 0.1989, "step": 11693 }, { "epoch": 3.51, "learning_rate": 2.8048001802884617e-05, "loss": 0.1693, "step": 11694 }, { "epoch": 3.51, "learning_rate": 2.804612379807692e-05, "loss": 0.1534, "step": 11695 }, { "epoch": 3.51, "learning_rate": 2.804424579326923e-05, "loss": 0.1631, "step": 11696 }, { "epoch": 3.51, "learning_rate": 2.8042367788461538e-05, "loss": 0.1608, "step": 11697 }, { "epoch": 3.52, "learning_rate": 2.8040489783653846e-05, "loss": 0.118, "step": 11698 }, { "epoch": 3.52, "learning_rate": 2.8038611778846158e-05, "loss": 0.1134, "step": 11699 }, { "epoch": 3.52, "learning_rate": 2.8036733774038466e-05, "loss": 0.1292, "step": 11700 }, { "epoch": 3.52, "learning_rate": 2.8034855769230774e-05, "loss": 0.1678, "step": 11701 }, { "epoch": 3.52, "learning_rate": 2.803297776442308e-05, "loss": 0.1012, "step": 11702 }, { "epoch": 3.52, "learning_rate": 2.8031099759615387e-05, "loss": 0.088, "step": 11703 }, { "epoch": 3.52, "learning_rate": 2.8029221754807695e-05, "loss": 0.1365, "step": 11704 }, { "epoch": 3.52, "learning_rate": 2.802734375e-05, "loss": 0.167, "step": 11705 }, { "epoch": 3.52, "learning_rate": 2.8025465745192308e-05, "loss": 0.1376, "step": 11706 }, { "epoch": 3.52, "learning_rate": 2.8023587740384616e-05, "loss": 0.1418, "step": 11707 }, { "epoch": 3.52, "learning_rate": 2.8021709735576924e-05, "loss": 0.169, "step": 11708 }, { "epoch": 3.52, "learning_rate": 2.801983173076923e-05, "loss": 0.1282, "step": 11709 }, { "epoch": 3.52, "learning_rate": 2.8017953725961537e-05, "loss": 0.1213, "step": 11710 }, { "epoch": 3.52, "learning_rate": 2.8016075721153845e-05, "loss": 0.1737, "step": 11711 }, { "epoch": 3.52, "learning_rate": 2.8014197716346157e-05, "loss": 0.182, "step": 11712 }, { "epoch": 3.52, "learning_rate": 2.8012319711538465e-05, "loss": 0.1636, "step": 11713 }, { "epoch": 3.52, "learning_rate": 2.8010441706730773e-05, "loss": 0.1447, "step": 11714 }, { "epoch": 3.52, "learning_rate": 2.8008563701923078e-05, "loss": 0.1579, "step": 11715 }, { "epoch": 3.52, "learning_rate": 2.8006685697115386e-05, "loss": 0.1303, "step": 11716 }, { "epoch": 3.52, "learning_rate": 2.8004807692307694e-05, "loss": 0.1403, "step": 11717 }, { "epoch": 3.52, "learning_rate": 2.8002929687500003e-05, "loss": 0.1063, "step": 11718 }, { "epoch": 3.52, "learning_rate": 2.8001051682692307e-05, "loss": 0.1828, "step": 11719 }, { "epoch": 3.52, "learning_rate": 2.7999173677884615e-05, "loss": 0.1329, "step": 11720 }, { "epoch": 3.52, "learning_rate": 2.7997295673076924e-05, "loss": 0.1169, "step": 11721 }, { "epoch": 3.52, "learning_rate": 2.7995417668269232e-05, "loss": 0.1401, "step": 11722 }, { "epoch": 3.52, "learning_rate": 2.7993539663461537e-05, "loss": 0.2055, "step": 11723 }, { "epoch": 3.52, "learning_rate": 2.7991661658653845e-05, "loss": 0.247, "step": 11724 }, { "epoch": 3.52, "learning_rate": 2.7989783653846156e-05, "loss": 0.1644, "step": 11725 }, { "epoch": 3.52, "learning_rate": 2.7987905649038464e-05, "loss": 0.1592, "step": 11726 }, { "epoch": 3.52, "learning_rate": 2.7986027644230773e-05, "loss": 0.1863, "step": 11727 }, { "epoch": 3.52, "learning_rate": 2.798414963942308e-05, "loss": 0.1292, "step": 11728 }, { "epoch": 3.52, "learning_rate": 2.7982271634615385e-05, "loss": 0.1798, "step": 11729 }, { "epoch": 3.52, "learning_rate": 2.7980393629807694e-05, "loss": 0.1227, "step": 11730 }, { "epoch": 3.52, "learning_rate": 2.7978515625000002e-05, "loss": 0.1744, "step": 11731 }, { "epoch": 3.53, "learning_rate": 2.797663762019231e-05, "loss": 0.1946, "step": 11732 }, { "epoch": 3.53, "learning_rate": 2.7974759615384615e-05, "loss": 0.1124, "step": 11733 }, { "epoch": 3.53, "learning_rate": 2.7972881610576923e-05, "loss": 0.1777, "step": 11734 }, { "epoch": 3.53, "learning_rate": 2.797100360576923e-05, "loss": 0.1478, "step": 11735 }, { "epoch": 3.53, "learning_rate": 2.796912560096154e-05, "loss": 0.2132, "step": 11736 }, { "epoch": 3.53, "learning_rate": 2.7967247596153844e-05, "loss": 0.1344, "step": 11737 }, { "epoch": 3.53, "learning_rate": 2.796536959134616e-05, "loss": 0.1601, "step": 11738 }, { "epoch": 3.53, "learning_rate": 2.7963491586538464e-05, "loss": 0.1957, "step": 11739 }, { "epoch": 3.53, "learning_rate": 2.7961613581730772e-05, "loss": 0.1119, "step": 11740 }, { "epoch": 3.53, "learning_rate": 2.795973557692308e-05, "loss": 0.1543, "step": 11741 }, { "epoch": 3.53, "learning_rate": 2.7957857572115388e-05, "loss": 0.1235, "step": 11742 }, { "epoch": 3.53, "learning_rate": 2.7955979567307693e-05, "loss": 0.1081, "step": 11743 }, { "epoch": 3.53, "learning_rate": 2.79541015625e-05, "loss": 0.116, "step": 11744 }, { "epoch": 3.53, "learning_rate": 2.795222355769231e-05, "loss": 0.1405, "step": 11745 }, { "epoch": 3.53, "learning_rate": 2.7950345552884617e-05, "loss": 0.0995, "step": 11746 }, { "epoch": 3.53, "learning_rate": 2.7948467548076922e-05, "loss": 0.1199, "step": 11747 }, { "epoch": 3.53, "learning_rate": 2.794658954326923e-05, "loss": 0.1447, "step": 11748 }, { "epoch": 3.53, "learning_rate": 2.794471153846154e-05, "loss": 0.1093, "step": 11749 }, { "epoch": 3.53, "learning_rate": 2.7942833533653843e-05, "loss": 0.1404, "step": 11750 }, { "epoch": 3.53, "learning_rate": 2.7940955528846158e-05, "loss": 0.2093, "step": 11751 }, { "epoch": 3.53, "learning_rate": 2.7939077524038466e-05, "loss": 0.1378, "step": 11752 }, { "epoch": 3.53, "learning_rate": 2.793719951923077e-05, "loss": 0.16, "step": 11753 }, { "epoch": 3.53, "learning_rate": 2.793532151442308e-05, "loss": 0.188, "step": 11754 }, { "epoch": 3.53, "learning_rate": 2.7933443509615387e-05, "loss": 0.1297, "step": 11755 }, { "epoch": 3.53, "learning_rate": 2.7931565504807696e-05, "loss": 0.2445, "step": 11756 }, { "epoch": 3.53, "learning_rate": 2.79296875e-05, "loss": 0.2098, "step": 11757 }, { "epoch": 3.53, "learning_rate": 2.792780949519231e-05, "loss": 0.1337, "step": 11758 }, { "epoch": 3.53, "learning_rate": 2.7925931490384617e-05, "loss": 0.1134, "step": 11759 }, { "epoch": 3.53, "learning_rate": 2.792405348557692e-05, "loss": 0.1378, "step": 11760 }, { "epoch": 3.53, "learning_rate": 2.792217548076923e-05, "loss": 0.1529, "step": 11761 }, { "epoch": 3.53, "learning_rate": 2.7920297475961538e-05, "loss": 0.1754, "step": 11762 }, { "epoch": 3.53, "learning_rate": 2.7918419471153846e-05, "loss": 0.1216, "step": 11763 }, { "epoch": 3.53, "learning_rate": 2.7916541466346158e-05, "loss": 0.221, "step": 11764 }, { "epoch": 3.54, "learning_rate": 2.7914663461538466e-05, "loss": 0.2584, "step": 11765 }, { "epoch": 3.54, "learning_rate": 2.7912785456730774e-05, "loss": 0.1688, "step": 11766 }, { "epoch": 3.54, "learning_rate": 2.791090745192308e-05, "loss": 0.1951, "step": 11767 }, { "epoch": 3.54, "learning_rate": 2.7909029447115387e-05, "loss": 0.1396, "step": 11768 }, { "epoch": 3.54, "learning_rate": 2.7907151442307695e-05, "loss": 0.1378, "step": 11769 }, { "epoch": 3.54, "learning_rate": 2.79052734375e-05, "loss": 0.113, "step": 11770 }, { "epoch": 3.54, "learning_rate": 2.7903395432692308e-05, "loss": 0.1004, "step": 11771 }, { "epoch": 3.54, "learning_rate": 2.7901517427884616e-05, "loss": 0.1737, "step": 11772 }, { "epoch": 3.54, "learning_rate": 2.7899639423076924e-05, "loss": 0.2157, "step": 11773 }, { "epoch": 3.54, "learning_rate": 2.789776141826923e-05, "loss": 0.1803, "step": 11774 }, { "epoch": 3.54, "learning_rate": 2.7895883413461537e-05, "loss": 0.2017, "step": 11775 }, { "epoch": 3.54, "learning_rate": 2.7894005408653845e-05, "loss": 0.1672, "step": 11776 }, { "epoch": 3.54, "learning_rate": 2.7892127403846157e-05, "loss": 0.1418, "step": 11777 }, { "epoch": 3.54, "learning_rate": 2.7890249399038465e-05, "loss": 0.1779, "step": 11778 }, { "epoch": 3.54, "learning_rate": 2.7888371394230773e-05, "loss": 0.1222, "step": 11779 }, { "epoch": 3.54, "learning_rate": 2.7886493389423078e-05, "loss": 0.1287, "step": 11780 }, { "epoch": 3.54, "learning_rate": 2.7884615384615386e-05, "loss": 0.145, "step": 11781 }, { "epoch": 3.54, "learning_rate": 2.7882737379807694e-05, "loss": 0.1446, "step": 11782 }, { "epoch": 3.54, "learning_rate": 2.7880859375000002e-05, "loss": 0.1587, "step": 11783 }, { "epoch": 3.54, "learning_rate": 2.7878981370192307e-05, "loss": 0.1617, "step": 11784 }, { "epoch": 3.54, "learning_rate": 2.7877103365384615e-05, "loss": 0.1651, "step": 11785 }, { "epoch": 3.54, "learning_rate": 2.7875225360576923e-05, "loss": 0.1157, "step": 11786 }, { "epoch": 3.54, "learning_rate": 2.787334735576923e-05, "loss": 0.132, "step": 11787 }, { "epoch": 3.54, "learning_rate": 2.7871469350961536e-05, "loss": 0.1106, "step": 11788 }, { "epoch": 3.54, "learning_rate": 2.7869591346153845e-05, "loss": 0.159, "step": 11789 }, { "epoch": 3.54, "learning_rate": 2.7867713341346156e-05, "loss": 0.2278, "step": 11790 }, { "epoch": 3.54, "learning_rate": 2.7865835336538464e-05, "loss": 0.1537, "step": 11791 }, { "epoch": 3.54, "learning_rate": 2.7863957331730772e-05, "loss": 0.1869, "step": 11792 }, { "epoch": 3.54, "learning_rate": 2.786207932692308e-05, "loss": 0.1393, "step": 11793 }, { "epoch": 3.54, "learning_rate": 2.7860201322115385e-05, "loss": 0.1821, "step": 11794 }, { "epoch": 3.54, "learning_rate": 2.7858323317307694e-05, "loss": 0.1816, "step": 11795 }, { "epoch": 3.54, "learning_rate": 2.7856445312500002e-05, "loss": 0.1901, "step": 11796 }, { "epoch": 3.54, "learning_rate": 2.785456730769231e-05, "loss": 0.2243, "step": 11797 }, { "epoch": 3.55, "learning_rate": 2.7852689302884615e-05, "loss": 0.0428, "step": 11798 }, { "epoch": 3.55, "learning_rate": 2.7850811298076923e-05, "loss": 0.1503, "step": 11799 }, { "epoch": 3.55, "learning_rate": 2.784893329326923e-05, "loss": 0.2671, "step": 11800 }, { "epoch": 3.55, "learning_rate": 2.784705528846154e-05, "loss": 0.2554, "step": 11801 }, { "epoch": 3.55, "learning_rate": 2.7845177283653844e-05, "loss": 0.1789, "step": 11802 }, { "epoch": 3.55, "learning_rate": 2.784329927884616e-05, "loss": 0.1622, "step": 11803 }, { "epoch": 3.55, "learning_rate": 2.7841421274038464e-05, "loss": 0.1385, "step": 11804 }, { "epoch": 3.55, "learning_rate": 2.7839543269230772e-05, "loss": 0.1855, "step": 11805 }, { "epoch": 3.55, "learning_rate": 2.783766526442308e-05, "loss": 0.151, "step": 11806 }, { "epoch": 3.55, "learning_rate": 2.7835787259615388e-05, "loss": 0.1193, "step": 11807 }, { "epoch": 3.55, "learning_rate": 2.7833909254807693e-05, "loss": 0.1506, "step": 11808 }, { "epoch": 3.55, "learning_rate": 2.783203125e-05, "loss": 0.1237, "step": 11809 }, { "epoch": 3.55, "learning_rate": 2.783015324519231e-05, "loss": 0.1207, "step": 11810 }, { "epoch": 3.55, "learning_rate": 2.7828275240384617e-05, "loss": 0.1722, "step": 11811 }, { "epoch": 3.55, "learning_rate": 2.7826397235576922e-05, "loss": 0.1561, "step": 11812 }, { "epoch": 3.55, "learning_rate": 2.782451923076923e-05, "loss": 0.1391, "step": 11813 }, { "epoch": 3.55, "learning_rate": 2.782264122596154e-05, "loss": 0.1442, "step": 11814 }, { "epoch": 3.55, "learning_rate": 2.7820763221153847e-05, "loss": 0.1457, "step": 11815 }, { "epoch": 3.55, "learning_rate": 2.7818885216346158e-05, "loss": 0.0934, "step": 11816 }, { "epoch": 3.55, "learning_rate": 2.7817007211538466e-05, "loss": 0.1402, "step": 11817 }, { "epoch": 3.55, "learning_rate": 2.781512920673077e-05, "loss": 0.1926, "step": 11818 }, { "epoch": 3.55, "learning_rate": 2.781325120192308e-05, "loss": 0.1815, "step": 11819 }, { "epoch": 3.55, "learning_rate": 2.7811373197115387e-05, "loss": 0.1373, "step": 11820 }, { "epoch": 3.55, "learning_rate": 2.7809495192307696e-05, "loss": 0.2052, "step": 11821 }, { "epoch": 3.55, "learning_rate": 2.78076171875e-05, "loss": 0.1331, "step": 11822 }, { "epoch": 3.55, "learning_rate": 2.780573918269231e-05, "loss": 0.1647, "step": 11823 }, { "epoch": 3.55, "learning_rate": 2.7803861177884617e-05, "loss": 0.1557, "step": 11824 }, { "epoch": 3.55, "learning_rate": 2.7801983173076925e-05, "loss": 0.2463, "step": 11825 }, { "epoch": 3.55, "learning_rate": 2.780010516826923e-05, "loss": 0.1654, "step": 11826 }, { "epoch": 3.55, "learning_rate": 2.7798227163461538e-05, "loss": 0.1512, "step": 11827 }, { "epoch": 3.55, "learning_rate": 2.7796349158653846e-05, "loss": 0.2588, "step": 11828 }, { "epoch": 3.55, "learning_rate": 2.7794471153846157e-05, "loss": 0.1868, "step": 11829 }, { "epoch": 3.55, "learning_rate": 2.7792593149038466e-05, "loss": 0.1886, "step": 11830 }, { "epoch": 3.55, "learning_rate": 2.7790715144230774e-05, "loss": 0.106, "step": 11831 }, { "epoch": 3.56, "learning_rate": 2.778883713942308e-05, "loss": 0.1417, "step": 11832 }, { "epoch": 3.56, "learning_rate": 2.7786959134615387e-05, "loss": 0.1381, "step": 11833 }, { "epoch": 3.56, "learning_rate": 2.7785081129807695e-05, "loss": 0.1473, "step": 11834 }, { "epoch": 3.56, "learning_rate": 2.7783203125000003e-05, "loss": 0.1513, "step": 11835 }, { "epoch": 3.56, "learning_rate": 2.7781325120192308e-05, "loss": 0.1729, "step": 11836 }, { "epoch": 3.56, "learning_rate": 2.7779447115384616e-05, "loss": 0.1254, "step": 11837 }, { "epoch": 3.56, "learning_rate": 2.7777569110576924e-05, "loss": 0.0615, "step": 11838 }, { "epoch": 3.56, "learning_rate": 2.777569110576923e-05, "loss": 0.2422, "step": 11839 }, { "epoch": 3.56, "learning_rate": 2.7773813100961537e-05, "loss": 0.143, "step": 11840 }, { "epoch": 3.56, "learning_rate": 2.7771935096153845e-05, "loss": 0.1565, "step": 11841 }, { "epoch": 3.56, "learning_rate": 2.7770057091346157e-05, "loss": 0.1601, "step": 11842 }, { "epoch": 3.56, "learning_rate": 2.7768179086538465e-05, "loss": 0.1685, "step": 11843 }, { "epoch": 3.56, "learning_rate": 2.7766301081730773e-05, "loss": 0.1268, "step": 11844 }, { "epoch": 3.56, "learning_rate": 2.776442307692308e-05, "loss": 0.0954, "step": 11845 }, { "epoch": 3.56, "learning_rate": 2.7762545072115386e-05, "loss": 0.1281, "step": 11846 }, { "epoch": 3.56, "learning_rate": 2.7760667067307694e-05, "loss": 0.1346, "step": 11847 }, { "epoch": 3.56, "learning_rate": 2.7758789062500002e-05, "loss": 0.0933, "step": 11848 }, { "epoch": 3.56, "learning_rate": 2.7756911057692307e-05, "loss": 0.1161, "step": 11849 }, { "epoch": 3.56, "learning_rate": 2.7755033052884615e-05, "loss": 0.1446, "step": 11850 }, { "epoch": 3.56, "learning_rate": 2.7753155048076923e-05, "loss": 0.116, "step": 11851 }, { "epoch": 3.56, "learning_rate": 2.775127704326923e-05, "loss": 0.1163, "step": 11852 }, { "epoch": 3.56, "learning_rate": 2.7749399038461536e-05, "loss": 0.1451, "step": 11853 }, { "epoch": 3.56, "learning_rate": 2.7747521033653844e-05, "loss": 0.1449, "step": 11854 }, { "epoch": 3.56, "learning_rate": 2.774564302884616e-05, "loss": 0.0988, "step": 11855 }, { "epoch": 3.56, "learning_rate": 2.7743765024038464e-05, "loss": 0.1263, "step": 11856 }, { "epoch": 3.56, "learning_rate": 2.7741887019230772e-05, "loss": 0.1735, "step": 11857 }, { "epoch": 3.56, "learning_rate": 2.774000901442308e-05, "loss": 0.1662, "step": 11858 }, { "epoch": 3.56, "learning_rate": 2.7738131009615385e-05, "loss": 0.0868, "step": 11859 }, { "epoch": 3.56, "learning_rate": 2.7736253004807693e-05, "loss": 0.1656, "step": 11860 }, { "epoch": 3.56, "learning_rate": 2.7734375e-05, "loss": 0.3272, "step": 11861 }, { "epoch": 3.56, "learning_rate": 2.773249699519231e-05, "loss": 0.1457, "step": 11862 }, { "epoch": 3.56, "learning_rate": 2.7730618990384615e-05, "loss": 0.1894, "step": 11863 }, { "epoch": 3.56, "learning_rate": 2.7728740985576923e-05, "loss": 0.1562, "step": 11864 }, { "epoch": 3.57, "learning_rate": 2.772686298076923e-05, "loss": 0.1203, "step": 11865 }, { "epoch": 3.57, "learning_rate": 2.772498497596154e-05, "loss": 0.2296, "step": 11866 }, { "epoch": 3.57, "learning_rate": 2.7723106971153844e-05, "loss": 0.1883, "step": 11867 }, { "epoch": 3.57, "learning_rate": 2.772122896634616e-05, "loss": 0.1714, "step": 11868 }, { "epoch": 3.57, "learning_rate": 2.7719350961538463e-05, "loss": 0.2292, "step": 11869 }, { "epoch": 3.57, "learning_rate": 2.771747295673077e-05, "loss": 0.1614, "step": 11870 }, { "epoch": 3.57, "learning_rate": 2.771559495192308e-05, "loss": 0.1444, "step": 11871 }, { "epoch": 3.57, "learning_rate": 2.7713716947115388e-05, "loss": 0.1385, "step": 11872 }, { "epoch": 3.57, "learning_rate": 2.7711838942307693e-05, "loss": 0.083, "step": 11873 }, { "epoch": 3.57, "learning_rate": 2.77099609375e-05, "loss": 0.1653, "step": 11874 }, { "epoch": 3.57, "learning_rate": 2.770808293269231e-05, "loss": 0.118, "step": 11875 }, { "epoch": 3.57, "learning_rate": 2.7706204927884617e-05, "loss": 0.1564, "step": 11876 }, { "epoch": 3.57, "learning_rate": 2.7704326923076922e-05, "loss": 0.132, "step": 11877 }, { "epoch": 3.57, "learning_rate": 2.770244891826923e-05, "loss": 0.2426, "step": 11878 }, { "epoch": 3.57, "learning_rate": 2.7700570913461538e-05, "loss": 0.2036, "step": 11879 }, { "epoch": 3.57, "learning_rate": 2.7698692908653846e-05, "loss": 0.1232, "step": 11880 }, { "epoch": 3.57, "learning_rate": 2.7696814903846158e-05, "loss": 0.1116, "step": 11881 }, { "epoch": 3.57, "learning_rate": 2.7694936899038466e-05, "loss": 0.0914, "step": 11882 }, { "epoch": 3.57, "learning_rate": 2.769305889423077e-05, "loss": 0.1936, "step": 11883 }, { "epoch": 3.57, "learning_rate": 2.769118088942308e-05, "loss": 0.2047, "step": 11884 }, { "epoch": 3.57, "learning_rate": 2.7689302884615387e-05, "loss": 0.1423, "step": 11885 }, { "epoch": 3.57, "learning_rate": 2.7687424879807695e-05, "loss": 0.1586, "step": 11886 }, { "epoch": 3.57, "learning_rate": 2.7685546875e-05, "loss": 0.1398, "step": 11887 }, { "epoch": 3.57, "learning_rate": 2.768366887019231e-05, "loss": 0.2128, "step": 11888 }, { "epoch": 3.57, "learning_rate": 2.7681790865384617e-05, "loss": 0.1101, "step": 11889 }, { "epoch": 3.57, "learning_rate": 2.7679912860576925e-05, "loss": 0.1025, "step": 11890 }, { "epoch": 3.57, "learning_rate": 2.767803485576923e-05, "loss": 0.1845, "step": 11891 }, { "epoch": 3.57, "learning_rate": 2.7676156850961538e-05, "loss": 0.1544, "step": 11892 }, { "epoch": 3.57, "learning_rate": 2.7674278846153846e-05, "loss": 0.1582, "step": 11893 }, { "epoch": 3.57, "learning_rate": 2.7672400841346157e-05, "loss": 0.0901, "step": 11894 }, { "epoch": 3.57, "learning_rate": 2.7670522836538465e-05, "loss": 0.2016, "step": 11895 }, { "epoch": 3.57, "learning_rate": 2.7668644831730774e-05, "loss": 0.1491, "step": 11896 }, { "epoch": 3.57, "learning_rate": 2.766676682692308e-05, "loss": 0.1297, "step": 11897 }, { "epoch": 3.58, "learning_rate": 2.7664888822115387e-05, "loss": 0.1392, "step": 11898 }, { "epoch": 3.58, "learning_rate": 2.7663010817307695e-05, "loss": 0.1915, "step": 11899 }, { "epoch": 3.58, "learning_rate": 2.7661132812500003e-05, "loss": 0.0959, "step": 11900 }, { "epoch": 3.58, "learning_rate": 2.7659254807692308e-05, "loss": 0.1744, "step": 11901 }, { "epoch": 3.58, "learning_rate": 2.7657376802884616e-05, "loss": 0.1617, "step": 11902 }, { "epoch": 3.58, "learning_rate": 2.7655498798076924e-05, "loss": 0.0914, "step": 11903 }, { "epoch": 3.58, "learning_rate": 2.765362079326923e-05, "loss": 0.1596, "step": 11904 }, { "epoch": 3.58, "learning_rate": 2.7651742788461537e-05, "loss": 0.1256, "step": 11905 }, { "epoch": 3.58, "learning_rate": 2.7649864783653845e-05, "loss": 0.1538, "step": 11906 }, { "epoch": 3.58, "learning_rate": 2.7647986778846157e-05, "loss": 0.1525, "step": 11907 }, { "epoch": 3.58, "learning_rate": 2.7646108774038465e-05, "loss": 0.1254, "step": 11908 }, { "epoch": 3.58, "learning_rate": 2.7644230769230773e-05, "loss": 0.1367, "step": 11909 }, { "epoch": 3.58, "learning_rate": 2.764235276442308e-05, "loss": 0.1119, "step": 11910 }, { "epoch": 3.58, "learning_rate": 2.7640474759615386e-05, "loss": 0.1181, "step": 11911 }, { "epoch": 3.58, "learning_rate": 2.7638596754807694e-05, "loss": 0.1193, "step": 11912 }, { "epoch": 3.58, "learning_rate": 2.7636718750000002e-05, "loss": 0.1549, "step": 11913 }, { "epoch": 3.58, "learning_rate": 2.7634840745192307e-05, "loss": 0.2915, "step": 11914 }, { "epoch": 3.58, "learning_rate": 2.7632962740384615e-05, "loss": 0.1324, "step": 11915 }, { "epoch": 3.58, "learning_rate": 2.7631084735576923e-05, "loss": 0.1781, "step": 11916 }, { "epoch": 3.58, "learning_rate": 2.762920673076923e-05, "loss": 0.2125, "step": 11917 }, { "epoch": 3.58, "learning_rate": 2.7627328725961536e-05, "loss": 0.1729, "step": 11918 }, { "epoch": 3.58, "learning_rate": 2.7625450721153844e-05, "loss": 0.1545, "step": 11919 }, { "epoch": 3.58, "learning_rate": 2.762357271634616e-05, "loss": 0.1434, "step": 11920 }, { "epoch": 3.58, "learning_rate": 2.7621694711538464e-05, "loss": 0.1414, "step": 11921 }, { "epoch": 3.58, "learning_rate": 2.7619816706730772e-05, "loss": 0.1583, "step": 11922 }, { "epoch": 3.58, "learning_rate": 2.761793870192308e-05, "loss": 0.1964, "step": 11923 }, { "epoch": 3.58, "learning_rate": 2.7616060697115385e-05, "loss": 0.1616, "step": 11924 }, { "epoch": 3.58, "learning_rate": 2.7614182692307693e-05, "loss": 0.137, "step": 11925 }, { "epoch": 3.58, "learning_rate": 2.76123046875e-05, "loss": 0.1502, "step": 11926 }, { "epoch": 3.58, "learning_rate": 2.761042668269231e-05, "loss": 0.1799, "step": 11927 }, { "epoch": 3.58, "learning_rate": 2.7608548677884614e-05, "loss": 0.1705, "step": 11928 }, { "epoch": 3.58, "learning_rate": 2.7606670673076923e-05, "loss": 0.1531, "step": 11929 }, { "epoch": 3.58, "learning_rate": 2.760479266826923e-05, "loss": 0.1576, "step": 11930 }, { "epoch": 3.59, "learning_rate": 2.760291466346154e-05, "loss": 0.1462, "step": 11931 }, { "epoch": 3.59, "learning_rate": 2.7601036658653844e-05, "loss": 0.1206, "step": 11932 }, { "epoch": 3.59, "learning_rate": 2.759915865384616e-05, "loss": 0.1005, "step": 11933 }, { "epoch": 3.59, "learning_rate": 2.7597280649038463e-05, "loss": 0.1157, "step": 11934 }, { "epoch": 3.59, "learning_rate": 2.759540264423077e-05, "loss": 0.1454, "step": 11935 }, { "epoch": 3.59, "learning_rate": 2.759352463942308e-05, "loss": 0.1891, "step": 11936 }, { "epoch": 3.59, "learning_rate": 2.7591646634615388e-05, "loss": 0.1116, "step": 11937 }, { "epoch": 3.59, "learning_rate": 2.7589768629807693e-05, "loss": 0.104, "step": 11938 }, { "epoch": 3.59, "learning_rate": 2.7587890625e-05, "loss": 0.1086, "step": 11939 }, { "epoch": 3.59, "learning_rate": 2.758601262019231e-05, "loss": 0.1238, "step": 11940 }, { "epoch": 3.59, "learning_rate": 2.7584134615384617e-05, "loss": 0.1037, "step": 11941 }, { "epoch": 3.59, "learning_rate": 2.7582256610576922e-05, "loss": 0.3159, "step": 11942 }, { "epoch": 3.59, "learning_rate": 2.758037860576923e-05, "loss": 0.1591, "step": 11943 }, { "epoch": 3.59, "learning_rate": 2.7578500600961538e-05, "loss": 0.148, "step": 11944 }, { "epoch": 3.59, "learning_rate": 2.7576622596153846e-05, "loss": 0.1003, "step": 11945 }, { "epoch": 3.59, "learning_rate": 2.7574744591346158e-05, "loss": 0.1367, "step": 11946 }, { "epoch": 3.59, "learning_rate": 2.7572866586538466e-05, "loss": 0.1739, "step": 11947 }, { "epoch": 3.59, "learning_rate": 2.757098858173077e-05, "loss": 0.1449, "step": 11948 }, { "epoch": 3.59, "learning_rate": 2.756911057692308e-05, "loss": 0.1358, "step": 11949 }, { "epoch": 3.59, "learning_rate": 2.7567232572115387e-05, "loss": 0.2162, "step": 11950 }, { "epoch": 3.59, "learning_rate": 2.7565354567307695e-05, "loss": 0.1508, "step": 11951 }, { "epoch": 3.59, "learning_rate": 2.75634765625e-05, "loss": 0.1017, "step": 11952 }, { "epoch": 3.59, "learning_rate": 2.7561598557692308e-05, "loss": 0.1378, "step": 11953 }, { "epoch": 3.59, "learning_rate": 2.7559720552884616e-05, "loss": 0.1693, "step": 11954 }, { "epoch": 3.59, "learning_rate": 2.7557842548076925e-05, "loss": 0.1179, "step": 11955 }, { "epoch": 3.59, "learning_rate": 2.755596454326923e-05, "loss": 0.1152, "step": 11956 }, { "epoch": 3.59, "learning_rate": 2.7554086538461537e-05, "loss": 0.2716, "step": 11957 }, { "epoch": 3.59, "learning_rate": 2.7552208533653846e-05, "loss": 0.142, "step": 11958 }, { "epoch": 3.59, "learning_rate": 2.7550330528846157e-05, "loss": 0.1388, "step": 11959 }, { "epoch": 3.59, "learning_rate": 2.7548452524038465e-05, "loss": 0.1688, "step": 11960 }, { "epoch": 3.59, "learning_rate": 2.7546574519230774e-05, "loss": 0.2558, "step": 11961 }, { "epoch": 3.59, "learning_rate": 2.7544696514423078e-05, "loss": 0.1177, "step": 11962 }, { "epoch": 3.59, "learning_rate": 2.7542818509615386e-05, "loss": 0.1714, "step": 11963 }, { "epoch": 3.59, "learning_rate": 2.7540940504807695e-05, "loss": 0.1237, "step": 11964 }, { "epoch": 3.6, "learning_rate": 2.7539062500000003e-05, "loss": 0.1272, "step": 11965 }, { "epoch": 3.6, "learning_rate": 2.7537184495192308e-05, "loss": 0.3221, "step": 11966 }, { "epoch": 3.6, "learning_rate": 2.7535306490384616e-05, "loss": 0.1168, "step": 11967 }, { "epoch": 3.6, "learning_rate": 2.7533428485576924e-05, "loss": 0.1501, "step": 11968 }, { "epoch": 3.6, "learning_rate": 2.7531550480769232e-05, "loss": 0.2252, "step": 11969 }, { "epoch": 3.6, "learning_rate": 2.7529672475961537e-05, "loss": 0.1424, "step": 11970 }, { "epoch": 3.6, "learning_rate": 2.7527794471153845e-05, "loss": 0.2318, "step": 11971 }, { "epoch": 3.6, "learning_rate": 2.7525916466346157e-05, "loss": 0.1275, "step": 11972 }, { "epoch": 3.6, "learning_rate": 2.7524038461538465e-05, "loss": 0.117, "step": 11973 }, { "epoch": 3.6, "learning_rate": 2.7522160456730773e-05, "loss": 0.1185, "step": 11974 }, { "epoch": 3.6, "learning_rate": 2.752028245192308e-05, "loss": 0.2153, "step": 11975 }, { "epoch": 3.6, "learning_rate": 2.7518404447115386e-05, "loss": 0.1783, "step": 11976 }, { "epoch": 3.6, "learning_rate": 2.7516526442307694e-05, "loss": 0.1445, "step": 11977 }, { "epoch": 3.6, "learning_rate": 2.7514648437500002e-05, "loss": 0.101, "step": 11978 }, { "epoch": 3.6, "learning_rate": 2.751277043269231e-05, "loss": 0.1333, "step": 11979 }, { "epoch": 3.6, "learning_rate": 2.7510892427884615e-05, "loss": 0.1273, "step": 11980 }, { "epoch": 3.6, "learning_rate": 2.7509014423076923e-05, "loss": 0.1519, "step": 11981 }, { "epoch": 3.6, "learning_rate": 2.750713641826923e-05, "loss": 0.1652, "step": 11982 }, { "epoch": 3.6, "learning_rate": 2.7505258413461536e-05, "loss": 0.1283, "step": 11983 }, { "epoch": 3.6, "learning_rate": 2.7503380408653844e-05, "loss": 0.154, "step": 11984 }, { "epoch": 3.6, "learning_rate": 2.750150240384616e-05, "loss": 0.1772, "step": 11985 }, { "epoch": 3.6, "learning_rate": 2.7499624399038464e-05, "loss": 0.1572, "step": 11986 }, { "epoch": 3.6, "learning_rate": 2.7497746394230772e-05, "loss": 0.1542, "step": 11987 }, { "epoch": 3.6, "learning_rate": 2.749586838942308e-05, "loss": 0.111, "step": 11988 }, { "epoch": 3.6, "learning_rate": 2.749399038461539e-05, "loss": 0.093, "step": 11989 }, { "epoch": 3.6, "learning_rate": 2.7492112379807693e-05, "loss": 0.1548, "step": 11990 }, { "epoch": 3.6, "learning_rate": 2.7490234375e-05, "loss": 0.1447, "step": 11991 }, { "epoch": 3.6, "learning_rate": 2.748835637019231e-05, "loss": 0.1261, "step": 11992 }, { "epoch": 3.6, "learning_rate": 2.7486478365384614e-05, "loss": 0.1826, "step": 11993 }, { "epoch": 3.6, "learning_rate": 2.7484600360576922e-05, "loss": 0.1042, "step": 11994 }, { "epoch": 3.6, "learning_rate": 2.748272235576923e-05, "loss": 0.1448, "step": 11995 }, { "epoch": 3.6, "learning_rate": 2.748084435096154e-05, "loss": 0.2199, "step": 11996 }, { "epoch": 3.6, "learning_rate": 2.7478966346153844e-05, "loss": 0.2701, "step": 11997 }, { "epoch": 3.61, "learning_rate": 2.747708834134616e-05, "loss": 0.1637, "step": 11998 }, { "epoch": 3.61, "learning_rate": 2.7475210336538467e-05, "loss": 0.1535, "step": 11999 }, { "epoch": 3.61, "learning_rate": 2.747333233173077e-05, "loss": 0.1689, "step": 12000 }, { "epoch": 3.61, "learning_rate": 2.747145432692308e-05, "loss": 0.1272, "step": 12001 }, { "epoch": 3.61, "learning_rate": 2.7469576322115388e-05, "loss": 0.1913, "step": 12002 }, { "epoch": 3.61, "learning_rate": 2.7467698317307693e-05, "loss": 0.1632, "step": 12003 }, { "epoch": 3.61, "learning_rate": 2.74658203125e-05, "loss": 0.1166, "step": 12004 }, { "epoch": 3.61, "learning_rate": 2.746394230769231e-05, "loss": 0.1323, "step": 12005 }, { "epoch": 3.61, "learning_rate": 2.7462064302884617e-05, "loss": 0.2109, "step": 12006 }, { "epoch": 3.61, "learning_rate": 2.7460186298076922e-05, "loss": 0.2212, "step": 12007 }, { "epoch": 3.61, "learning_rate": 2.745830829326923e-05, "loss": 0.1533, "step": 12008 }, { "epoch": 3.61, "learning_rate": 2.7456430288461538e-05, "loss": 0.1703, "step": 12009 }, { "epoch": 3.61, "learning_rate": 2.7454552283653846e-05, "loss": 0.2642, "step": 12010 }, { "epoch": 3.61, "learning_rate": 2.7452674278846158e-05, "loss": 0.1963, "step": 12011 }, { "epoch": 3.61, "learning_rate": 2.7450796274038466e-05, "loss": 0.1861, "step": 12012 }, { "epoch": 3.61, "learning_rate": 2.744891826923077e-05, "loss": 0.1999, "step": 12013 }, { "epoch": 3.61, "learning_rate": 2.744704026442308e-05, "loss": 0.1044, "step": 12014 }, { "epoch": 3.61, "learning_rate": 2.7445162259615387e-05, "loss": 0.144, "step": 12015 }, { "epoch": 3.61, "learning_rate": 2.7443284254807695e-05, "loss": 0.1594, "step": 12016 }, { "epoch": 3.61, "learning_rate": 2.744140625e-05, "loss": 0.0916, "step": 12017 }, { "epoch": 3.61, "learning_rate": 2.7439528245192308e-05, "loss": 0.0955, "step": 12018 }, { "epoch": 3.61, "learning_rate": 2.7437650240384616e-05, "loss": 0.1502, "step": 12019 }, { "epoch": 3.61, "learning_rate": 2.7435772235576924e-05, "loss": 0.1042, "step": 12020 }, { "epoch": 3.61, "learning_rate": 2.743389423076923e-05, "loss": 0.1615, "step": 12021 }, { "epoch": 3.61, "learning_rate": 2.7432016225961537e-05, "loss": 0.1461, "step": 12022 }, { "epoch": 3.61, "learning_rate": 2.7430138221153846e-05, "loss": 0.1981, "step": 12023 }, { "epoch": 3.61, "learning_rate": 2.7428260216346157e-05, "loss": 0.1505, "step": 12024 }, { "epoch": 3.61, "learning_rate": 2.7426382211538465e-05, "loss": 0.1313, "step": 12025 }, { "epoch": 3.61, "learning_rate": 2.7424504206730773e-05, "loss": 0.1663, "step": 12026 }, { "epoch": 3.61, "learning_rate": 2.7422626201923078e-05, "loss": 0.1488, "step": 12027 }, { "epoch": 3.61, "learning_rate": 2.7420748197115386e-05, "loss": 0.113, "step": 12028 }, { "epoch": 3.61, "learning_rate": 2.7418870192307695e-05, "loss": 0.2248, "step": 12029 }, { "epoch": 3.61, "learning_rate": 2.7416992187500003e-05, "loss": 0.1517, "step": 12030 }, { "epoch": 3.62, "learning_rate": 2.7415114182692307e-05, "loss": 0.1794, "step": 12031 }, { "epoch": 3.62, "learning_rate": 2.7413236177884616e-05, "loss": 0.1257, "step": 12032 }, { "epoch": 3.62, "learning_rate": 2.7411358173076924e-05, "loss": 0.254, "step": 12033 }, { "epoch": 3.62, "learning_rate": 2.7409480168269232e-05, "loss": 0.1146, "step": 12034 }, { "epoch": 3.62, "learning_rate": 2.7407602163461537e-05, "loss": 0.1943, "step": 12035 }, { "epoch": 3.62, "learning_rate": 2.7405724158653845e-05, "loss": 0.1532, "step": 12036 }, { "epoch": 3.62, "learning_rate": 2.7403846153846156e-05, "loss": 0.1534, "step": 12037 }, { "epoch": 3.62, "learning_rate": 2.7401968149038465e-05, "loss": 0.1152, "step": 12038 }, { "epoch": 3.62, "learning_rate": 2.7400090144230773e-05, "loss": 0.1222, "step": 12039 }, { "epoch": 3.62, "learning_rate": 2.739821213942308e-05, "loss": 0.0914, "step": 12040 }, { "epoch": 3.62, "learning_rate": 2.7396334134615386e-05, "loss": 0.1741, "step": 12041 }, { "epoch": 3.62, "learning_rate": 2.7394456129807694e-05, "loss": 0.1192, "step": 12042 }, { "epoch": 3.62, "learning_rate": 2.7392578125000002e-05, "loss": 0.109, "step": 12043 }, { "epoch": 3.62, "learning_rate": 2.739070012019231e-05, "loss": 0.1508, "step": 12044 }, { "epoch": 3.62, "learning_rate": 2.7388822115384615e-05, "loss": 0.2013, "step": 12045 }, { "epoch": 3.62, "learning_rate": 2.7386944110576923e-05, "loss": 0.1611, "step": 12046 }, { "epoch": 3.62, "learning_rate": 2.738506610576923e-05, "loss": 0.1518, "step": 12047 }, { "epoch": 3.62, "learning_rate": 2.7383188100961536e-05, "loss": 0.0902, "step": 12048 }, { "epoch": 3.62, "learning_rate": 2.7381310096153844e-05, "loss": 0.1672, "step": 12049 }, { "epoch": 3.62, "learning_rate": 2.737943209134616e-05, "loss": 0.1421, "step": 12050 }, { "epoch": 3.62, "learning_rate": 2.7377554086538464e-05, "loss": 0.1357, "step": 12051 }, { "epoch": 3.62, "learning_rate": 2.7375676081730772e-05, "loss": 0.1472, "step": 12052 }, { "epoch": 3.62, "learning_rate": 2.737379807692308e-05, "loss": 0.1497, "step": 12053 }, { "epoch": 3.62, "learning_rate": 2.737192007211539e-05, "loss": 0.1512, "step": 12054 }, { "epoch": 3.62, "learning_rate": 2.7370042067307693e-05, "loss": 0.1358, "step": 12055 }, { "epoch": 3.62, "learning_rate": 2.73681640625e-05, "loss": 0.1283, "step": 12056 }, { "epoch": 3.62, "learning_rate": 2.736628605769231e-05, "loss": 0.277, "step": 12057 }, { "epoch": 3.62, "learning_rate": 2.7364408052884614e-05, "loss": 0.1624, "step": 12058 }, { "epoch": 3.62, "learning_rate": 2.7362530048076922e-05, "loss": 0.1367, "step": 12059 }, { "epoch": 3.62, "learning_rate": 2.736065204326923e-05, "loss": 0.1824, "step": 12060 }, { "epoch": 3.62, "learning_rate": 2.735877403846154e-05, "loss": 0.1475, "step": 12061 }, { "epoch": 3.62, "learning_rate": 2.7356896033653843e-05, "loss": 0.174, "step": 12062 }, { "epoch": 3.62, "learning_rate": 2.735501802884616e-05, "loss": 0.1283, "step": 12063 }, { "epoch": 3.62, "learning_rate": 2.7353140024038467e-05, "loss": 0.1772, "step": 12064 }, { "epoch": 3.63, "learning_rate": 2.735126201923077e-05, "loss": 0.1332, "step": 12065 }, { "epoch": 3.63, "learning_rate": 2.734938401442308e-05, "loss": 0.1213, "step": 12066 }, { "epoch": 3.63, "learning_rate": 2.7347506009615388e-05, "loss": 0.1332, "step": 12067 }, { "epoch": 3.63, "learning_rate": 2.7345628004807692e-05, "loss": 0.1312, "step": 12068 }, { "epoch": 3.63, "learning_rate": 2.734375e-05, "loss": 0.0995, "step": 12069 }, { "epoch": 3.63, "learning_rate": 2.734187199519231e-05, "loss": 0.1639, "step": 12070 }, { "epoch": 3.63, "learning_rate": 2.7339993990384617e-05, "loss": 0.1398, "step": 12071 }, { "epoch": 3.63, "learning_rate": 2.733811598557692e-05, "loss": 0.1462, "step": 12072 }, { "epoch": 3.63, "learning_rate": 2.733623798076923e-05, "loss": 0.1085, "step": 12073 }, { "epoch": 3.63, "learning_rate": 2.7334359975961538e-05, "loss": 0.1603, "step": 12074 }, { "epoch": 3.63, "learning_rate": 2.7332481971153846e-05, "loss": 0.1678, "step": 12075 }, { "epoch": 3.63, "learning_rate": 2.7330603966346158e-05, "loss": 0.1425, "step": 12076 }, { "epoch": 3.63, "learning_rate": 2.7328725961538466e-05, "loss": 0.1373, "step": 12077 }, { "epoch": 3.63, "learning_rate": 2.732684795673077e-05, "loss": 0.1983, "step": 12078 }, { "epoch": 3.63, "learning_rate": 2.732496995192308e-05, "loss": 0.1427, "step": 12079 }, { "epoch": 3.63, "learning_rate": 2.7323091947115387e-05, "loss": 0.2139, "step": 12080 }, { "epoch": 3.63, "learning_rate": 2.7321213942307695e-05, "loss": 0.1043, "step": 12081 }, { "epoch": 3.63, "learning_rate": 2.73193359375e-05, "loss": 0.1207, "step": 12082 }, { "epoch": 3.63, "learning_rate": 2.7317457932692308e-05, "loss": 0.1478, "step": 12083 }, { "epoch": 3.63, "learning_rate": 2.7315579927884616e-05, "loss": 0.1161, "step": 12084 }, { "epoch": 3.63, "learning_rate": 2.7313701923076924e-05, "loss": 0.1518, "step": 12085 }, { "epoch": 3.63, "learning_rate": 2.731182391826923e-05, "loss": 0.1033, "step": 12086 }, { "epoch": 3.63, "learning_rate": 2.7309945913461537e-05, "loss": 0.0828, "step": 12087 }, { "epoch": 3.63, "learning_rate": 2.7308067908653845e-05, "loss": 0.1037, "step": 12088 }, { "epoch": 3.63, "learning_rate": 2.7306189903846157e-05, "loss": 0.1623, "step": 12089 }, { "epoch": 3.63, "learning_rate": 2.7304311899038465e-05, "loss": 0.1315, "step": 12090 }, { "epoch": 3.63, "learning_rate": 2.7302433894230773e-05, "loss": 0.1678, "step": 12091 }, { "epoch": 3.63, "learning_rate": 2.7300555889423078e-05, "loss": 0.1834, "step": 12092 }, { "epoch": 3.63, "learning_rate": 2.7298677884615386e-05, "loss": 0.1315, "step": 12093 }, { "epoch": 3.63, "learning_rate": 2.7296799879807694e-05, "loss": 0.1148, "step": 12094 }, { "epoch": 3.63, "learning_rate": 2.7294921875000003e-05, "loss": 0.2796, "step": 12095 }, { "epoch": 3.63, "learning_rate": 2.7293043870192307e-05, "loss": 0.1629, "step": 12096 }, { "epoch": 3.63, "learning_rate": 2.7291165865384615e-05, "loss": 0.1998, "step": 12097 }, { "epoch": 3.64, "learning_rate": 2.7289287860576924e-05, "loss": 0.117, "step": 12098 }, { "epoch": 3.64, "learning_rate": 2.7287409855769232e-05, "loss": 0.1192, "step": 12099 }, { "epoch": 3.64, "learning_rate": 2.7285531850961537e-05, "loss": 0.1598, "step": 12100 }, { "epoch": 3.64, "learning_rate": 2.7283653846153845e-05, "loss": 0.1099, "step": 12101 }, { "epoch": 3.64, "learning_rate": 2.7281775841346156e-05, "loss": 0.1303, "step": 12102 }, { "epoch": 3.64, "learning_rate": 2.7279897836538464e-05, "loss": 0.1264, "step": 12103 }, { "epoch": 3.64, "learning_rate": 2.7278019831730773e-05, "loss": 0.1809, "step": 12104 }, { "epoch": 3.64, "learning_rate": 2.727614182692308e-05, "loss": 0.1654, "step": 12105 }, { "epoch": 3.64, "learning_rate": 2.7274263822115386e-05, "loss": 0.1333, "step": 12106 }, { "epoch": 3.64, "learning_rate": 2.7272385817307694e-05, "loss": 0.1053, "step": 12107 }, { "epoch": 3.64, "learning_rate": 2.7270507812500002e-05, "loss": 0.1469, "step": 12108 }, { "epoch": 3.64, "learning_rate": 2.726862980769231e-05, "loss": 0.1159, "step": 12109 }, { "epoch": 3.64, "learning_rate": 2.7266751802884615e-05, "loss": 0.1293, "step": 12110 }, { "epoch": 3.64, "learning_rate": 2.7264873798076923e-05, "loss": 0.1147, "step": 12111 }, { "epoch": 3.64, "learning_rate": 2.726299579326923e-05, "loss": 0.2143, "step": 12112 }, { "epoch": 3.64, "learning_rate": 2.726111778846154e-05, "loss": 0.1385, "step": 12113 }, { "epoch": 3.64, "learning_rate": 2.7259239783653844e-05, "loss": 0.1475, "step": 12114 }, { "epoch": 3.64, "learning_rate": 2.725736177884616e-05, "loss": 0.1182, "step": 12115 }, { "epoch": 3.64, "learning_rate": 2.7255483774038464e-05, "loss": 0.1384, "step": 12116 }, { "epoch": 3.64, "learning_rate": 2.7253605769230772e-05, "loss": 0.1868, "step": 12117 }, { "epoch": 3.64, "learning_rate": 2.725172776442308e-05, "loss": 0.1219, "step": 12118 }, { "epoch": 3.64, "learning_rate": 2.7249849759615388e-05, "loss": 0.222, "step": 12119 }, { "epoch": 3.64, "learning_rate": 2.7247971754807693e-05, "loss": 0.1605, "step": 12120 }, { "epoch": 3.64, "learning_rate": 2.724609375e-05, "loss": 0.2565, "step": 12121 }, { "epoch": 3.64, "learning_rate": 2.724421574519231e-05, "loss": 0.1104, "step": 12122 }, { "epoch": 3.64, "learning_rate": 2.7242337740384617e-05, "loss": 0.1013, "step": 12123 }, { "epoch": 3.64, "learning_rate": 2.7240459735576922e-05, "loss": 0.1199, "step": 12124 }, { "epoch": 3.64, "learning_rate": 2.723858173076923e-05, "loss": 0.1123, "step": 12125 }, { "epoch": 3.64, "learning_rate": 2.723670372596154e-05, "loss": 0.1385, "step": 12126 }, { "epoch": 3.64, "learning_rate": 2.7234825721153843e-05, "loss": 0.1581, "step": 12127 }, { "epoch": 3.64, "learning_rate": 2.7232947716346158e-05, "loss": 0.1537, "step": 12128 }, { "epoch": 3.64, "learning_rate": 2.7231069711538466e-05, "loss": 0.1425, "step": 12129 }, { "epoch": 3.64, "learning_rate": 2.722919170673077e-05, "loss": 0.1838, "step": 12130 }, { "epoch": 3.65, "learning_rate": 2.722731370192308e-05, "loss": 0.1955, "step": 12131 }, { "epoch": 3.65, "learning_rate": 2.7225435697115388e-05, "loss": 0.1517, "step": 12132 }, { "epoch": 3.65, "learning_rate": 2.7223557692307696e-05, "loss": 0.1796, "step": 12133 }, { "epoch": 3.65, "learning_rate": 2.72216796875e-05, "loss": 0.1174, "step": 12134 }, { "epoch": 3.65, "learning_rate": 2.721980168269231e-05, "loss": 0.1487, "step": 12135 }, { "epoch": 3.65, "learning_rate": 2.7217923677884617e-05, "loss": 0.1532, "step": 12136 }, { "epoch": 3.65, "learning_rate": 2.721604567307692e-05, "loss": 0.1752, "step": 12137 }, { "epoch": 3.65, "learning_rate": 2.721416766826923e-05, "loss": 0.1767, "step": 12138 }, { "epoch": 3.65, "learning_rate": 2.7212289663461538e-05, "loss": 0.1216, "step": 12139 }, { "epoch": 3.65, "learning_rate": 2.7210411658653846e-05, "loss": 0.152, "step": 12140 }, { "epoch": 3.65, "learning_rate": 2.7208533653846158e-05, "loss": 0.1092, "step": 12141 }, { "epoch": 3.65, "learning_rate": 2.7206655649038466e-05, "loss": 0.1243, "step": 12142 }, { "epoch": 3.65, "learning_rate": 2.7204777644230774e-05, "loss": 0.1598, "step": 12143 }, { "epoch": 3.65, "learning_rate": 2.720289963942308e-05, "loss": 0.1368, "step": 12144 }, { "epoch": 3.65, "learning_rate": 2.7201021634615387e-05, "loss": 0.1982, "step": 12145 }, { "epoch": 3.65, "learning_rate": 2.7199143629807695e-05, "loss": 0.1073, "step": 12146 }, { "epoch": 3.65, "learning_rate": 2.7197265625e-05, "loss": 0.1979, "step": 12147 }, { "epoch": 3.65, "learning_rate": 2.7195387620192308e-05, "loss": 0.116, "step": 12148 }, { "epoch": 3.65, "learning_rate": 2.7193509615384616e-05, "loss": 0.1792, "step": 12149 }, { "epoch": 3.65, "learning_rate": 2.7191631610576924e-05, "loss": 0.1265, "step": 12150 }, { "epoch": 3.65, "learning_rate": 2.718975360576923e-05, "loss": 0.1641, "step": 12151 }, { "epoch": 3.65, "learning_rate": 2.7187875600961537e-05, "loss": 0.1744, "step": 12152 }, { "epoch": 3.65, "learning_rate": 2.7185997596153845e-05, "loss": 0.1466, "step": 12153 }, { "epoch": 3.65, "learning_rate": 2.7184119591346157e-05, "loss": 0.1967, "step": 12154 }, { "epoch": 3.65, "learning_rate": 2.7182241586538465e-05, "loss": 0.1987, "step": 12155 }, { "epoch": 3.65, "learning_rate": 2.7180363581730773e-05, "loss": 0.1804, "step": 12156 }, { "epoch": 3.65, "learning_rate": 2.7178485576923078e-05, "loss": 0.1117, "step": 12157 }, { "epoch": 3.65, "learning_rate": 2.7176607572115386e-05, "loss": 0.1322, "step": 12158 }, { "epoch": 3.65, "learning_rate": 2.7174729567307694e-05, "loss": 0.2041, "step": 12159 }, { "epoch": 3.65, "learning_rate": 2.7172851562500002e-05, "loss": 0.1383, "step": 12160 }, { "epoch": 3.65, "learning_rate": 2.7170973557692307e-05, "loss": 0.0881, "step": 12161 }, { "epoch": 3.65, "learning_rate": 2.7169095552884615e-05, "loss": 0.1402, "step": 12162 }, { "epoch": 3.65, "learning_rate": 2.7167217548076924e-05, "loss": 0.3182, "step": 12163 }, { "epoch": 3.66, "learning_rate": 2.7165339543269232e-05, "loss": 0.2374, "step": 12164 }, { "epoch": 3.66, "learning_rate": 2.7163461538461536e-05, "loss": 0.1794, "step": 12165 }, { "epoch": 3.66, "learning_rate": 2.7161583533653845e-05, "loss": 0.1563, "step": 12166 }, { "epoch": 3.66, "learning_rate": 2.7159705528846156e-05, "loss": 0.1454, "step": 12167 }, { "epoch": 3.66, "learning_rate": 2.7157827524038464e-05, "loss": 0.1097, "step": 12168 }, { "epoch": 3.66, "learning_rate": 2.7155949519230773e-05, "loss": 0.139, "step": 12169 }, { "epoch": 3.66, "learning_rate": 2.715407151442308e-05, "loss": 0.1388, "step": 12170 }, { "epoch": 3.66, "learning_rate": 2.7152193509615385e-05, "loss": 0.1751, "step": 12171 }, { "epoch": 3.66, "learning_rate": 2.7150315504807694e-05, "loss": 0.1359, "step": 12172 }, { "epoch": 3.66, "learning_rate": 2.7148437500000002e-05, "loss": 0.1167, "step": 12173 }, { "epoch": 3.66, "learning_rate": 2.714655949519231e-05, "loss": 0.1663, "step": 12174 }, { "epoch": 3.66, "learning_rate": 2.7144681490384615e-05, "loss": 0.1205, "step": 12175 }, { "epoch": 3.66, "learning_rate": 2.7142803485576923e-05, "loss": 0.157, "step": 12176 }, { "epoch": 3.66, "learning_rate": 2.714092548076923e-05, "loss": 0.181, "step": 12177 }, { "epoch": 3.66, "learning_rate": 2.713904747596154e-05, "loss": 0.1776, "step": 12178 }, { "epoch": 3.66, "learning_rate": 2.7137169471153844e-05, "loss": 0.2342, "step": 12179 }, { "epoch": 3.66, "learning_rate": 2.713529146634616e-05, "loss": 0.177, "step": 12180 }, { "epoch": 3.66, "learning_rate": 2.7133413461538464e-05, "loss": 0.1779, "step": 12181 }, { "epoch": 3.66, "learning_rate": 2.7131535456730772e-05, "loss": 0.1283, "step": 12182 }, { "epoch": 3.66, "learning_rate": 2.712965745192308e-05, "loss": 0.2134, "step": 12183 }, { "epoch": 3.66, "learning_rate": 2.7127779447115388e-05, "loss": 0.1654, "step": 12184 }, { "epoch": 3.66, "learning_rate": 2.7125901442307693e-05, "loss": 0.1348, "step": 12185 }, { "epoch": 3.66, "learning_rate": 2.71240234375e-05, "loss": 0.1558, "step": 12186 }, { "epoch": 3.66, "learning_rate": 2.712214543269231e-05, "loss": 0.1028, "step": 12187 }, { "epoch": 3.66, "learning_rate": 2.7120267427884617e-05, "loss": 0.16, "step": 12188 }, { "epoch": 3.66, "learning_rate": 2.7118389423076922e-05, "loss": 0.1558, "step": 12189 }, { "epoch": 3.66, "learning_rate": 2.711651141826923e-05, "loss": 0.158, "step": 12190 }, { "epoch": 3.66, "learning_rate": 2.711463341346154e-05, "loss": 0.1322, "step": 12191 }, { "epoch": 3.66, "learning_rate": 2.7112755408653843e-05, "loss": 0.1371, "step": 12192 }, { "epoch": 3.66, "learning_rate": 2.7110877403846158e-05, "loss": 0.1554, "step": 12193 }, { "epoch": 3.66, "learning_rate": 2.7108999399038466e-05, "loss": 0.1385, "step": 12194 }, { "epoch": 3.66, "learning_rate": 2.710712139423077e-05, "loss": 0.1714, "step": 12195 }, { "epoch": 3.66, "learning_rate": 2.710524338942308e-05, "loss": 0.1181, "step": 12196 }, { "epoch": 3.66, "learning_rate": 2.7103365384615387e-05, "loss": 0.1853, "step": 12197 }, { "epoch": 3.67, "learning_rate": 2.7101487379807696e-05, "loss": 0.2507, "step": 12198 }, { "epoch": 3.67, "learning_rate": 2.7099609375e-05, "loss": 0.1619, "step": 12199 }, { "epoch": 3.67, "learning_rate": 2.709773137019231e-05, "loss": 0.1643, "step": 12200 }, { "epoch": 3.67, "learning_rate": 2.7095853365384617e-05, "loss": 0.1215, "step": 12201 }, { "epoch": 3.67, "learning_rate": 2.709397536057692e-05, "loss": 0.1223, "step": 12202 }, { "epoch": 3.67, "learning_rate": 2.709209735576923e-05, "loss": 0.1617, "step": 12203 }, { "epoch": 3.67, "learning_rate": 2.7090219350961538e-05, "loss": 0.1133, "step": 12204 }, { "epoch": 3.67, "learning_rate": 2.7088341346153846e-05, "loss": 0.1423, "step": 12205 }, { "epoch": 3.67, "learning_rate": 2.7086463341346157e-05, "loss": 0.1298, "step": 12206 }, { "epoch": 3.67, "learning_rate": 2.7084585336538466e-05, "loss": 0.1989, "step": 12207 }, { "epoch": 3.67, "learning_rate": 2.7082707331730774e-05, "loss": 0.1145, "step": 12208 }, { "epoch": 3.67, "learning_rate": 2.708082932692308e-05, "loss": 0.1156, "step": 12209 }, { "epoch": 3.67, "learning_rate": 2.7078951322115387e-05, "loss": 0.1683, "step": 12210 }, { "epoch": 3.67, "learning_rate": 2.7077073317307695e-05, "loss": 0.1491, "step": 12211 }, { "epoch": 3.67, "learning_rate": 2.70751953125e-05, "loss": 0.0882, "step": 12212 }, { "epoch": 3.67, "learning_rate": 2.7073317307692308e-05, "loss": 0.1734, "step": 12213 }, { "epoch": 3.67, "learning_rate": 2.7071439302884616e-05, "loss": 0.1774, "step": 12214 }, { "epoch": 3.67, "learning_rate": 2.7069561298076924e-05, "loss": 0.1633, "step": 12215 }, { "epoch": 3.67, "learning_rate": 2.706768329326923e-05, "loss": 0.1224, "step": 12216 }, { "epoch": 3.67, "learning_rate": 2.7065805288461537e-05, "loss": 0.1135, "step": 12217 }, { "epoch": 3.67, "learning_rate": 2.7063927283653845e-05, "loss": 0.1472, "step": 12218 }, { "epoch": 3.67, "learning_rate": 2.7062049278846157e-05, "loss": 0.1321, "step": 12219 }, { "epoch": 3.67, "learning_rate": 2.7060171274038465e-05, "loss": 0.1134, "step": 12220 }, { "epoch": 3.67, "learning_rate": 2.7058293269230773e-05, "loss": 0.1484, "step": 12221 }, { "epoch": 3.67, "learning_rate": 2.7056415264423078e-05, "loss": 0.1112, "step": 12222 }, { "epoch": 3.67, "learning_rate": 2.7054537259615386e-05, "loss": 0.1647, "step": 12223 }, { "epoch": 3.67, "learning_rate": 2.7052659254807694e-05, "loss": 0.1498, "step": 12224 }, { "epoch": 3.67, "learning_rate": 2.7050781250000002e-05, "loss": 0.1694, "step": 12225 }, { "epoch": 3.67, "learning_rate": 2.7048903245192307e-05, "loss": 0.1196, "step": 12226 }, { "epoch": 3.67, "learning_rate": 2.7047025240384615e-05, "loss": 0.1717, "step": 12227 }, { "epoch": 3.67, "learning_rate": 2.7045147235576923e-05, "loss": 0.151, "step": 12228 }, { "epoch": 3.67, "learning_rate": 2.704326923076923e-05, "loss": 0.2142, "step": 12229 }, { "epoch": 3.67, "learning_rate": 2.7041391225961536e-05, "loss": 0.1057, "step": 12230 }, { "epoch": 3.68, "learning_rate": 2.7039513221153845e-05, "loss": 0.1716, "step": 12231 }, { "epoch": 3.68, "learning_rate": 2.7037635216346156e-05, "loss": 0.1343, "step": 12232 }, { "epoch": 3.68, "learning_rate": 2.7035757211538464e-05, "loss": 0.2051, "step": 12233 }, { "epoch": 3.68, "learning_rate": 2.7033879206730772e-05, "loss": 0.1317, "step": 12234 }, { "epoch": 3.68, "learning_rate": 2.703200120192308e-05, "loss": 0.1692, "step": 12235 }, { "epoch": 3.68, "learning_rate": 2.7030123197115385e-05, "loss": 0.1382, "step": 12236 }, { "epoch": 3.68, "learning_rate": 2.7028245192307693e-05, "loss": 0.1586, "step": 12237 }, { "epoch": 3.68, "learning_rate": 2.70263671875e-05, "loss": 0.2384, "step": 12238 }, { "epoch": 3.68, "learning_rate": 2.702448918269231e-05, "loss": 0.1488, "step": 12239 }, { "epoch": 3.68, "learning_rate": 2.7022611177884615e-05, "loss": 0.1689, "step": 12240 }, { "epoch": 3.68, "learning_rate": 2.7020733173076923e-05, "loss": 0.1619, "step": 12241 }, { "epoch": 3.68, "learning_rate": 2.701885516826923e-05, "loss": 0.1462, "step": 12242 }, { "epoch": 3.68, "learning_rate": 2.701697716346154e-05, "loss": 0.1442, "step": 12243 }, { "epoch": 3.68, "learning_rate": 2.7015099158653844e-05, "loss": 0.1972, "step": 12244 }, { "epoch": 3.68, "learning_rate": 2.701322115384616e-05, "loss": 0.1336, "step": 12245 }, { "epoch": 3.68, "learning_rate": 2.7011343149038464e-05, "loss": 0.1333, "step": 12246 }, { "epoch": 3.68, "learning_rate": 2.700946514423077e-05, "loss": 0.1381, "step": 12247 }, { "epoch": 3.68, "learning_rate": 2.700758713942308e-05, "loss": 0.1323, "step": 12248 }, { "epoch": 3.68, "learning_rate": 2.7005709134615388e-05, "loss": 0.1532, "step": 12249 }, { "epoch": 3.68, "learning_rate": 2.7003831129807693e-05, "loss": 0.1838, "step": 12250 }, { "epoch": 3.68, "learning_rate": 2.7001953125e-05, "loss": 0.1761, "step": 12251 }, { "epoch": 3.68, "learning_rate": 2.700007512019231e-05, "loss": 0.1067, "step": 12252 }, { "epoch": 3.68, "learning_rate": 2.6998197115384617e-05, "loss": 0.1875, "step": 12253 }, { "epoch": 3.68, "learning_rate": 2.6996319110576922e-05, "loss": 0.1673, "step": 12254 }, { "epoch": 3.68, "learning_rate": 2.699444110576923e-05, "loss": 0.2016, "step": 12255 }, { "epoch": 3.68, "learning_rate": 2.699256310096154e-05, "loss": 0.1576, "step": 12256 }, { "epoch": 3.68, "learning_rate": 2.6990685096153847e-05, "loss": 0.1646, "step": 12257 }, { "epoch": 3.68, "learning_rate": 2.6988807091346158e-05, "loss": 0.1677, "step": 12258 }, { "epoch": 3.68, "learning_rate": 2.6986929086538466e-05, "loss": 0.0919, "step": 12259 }, { "epoch": 3.68, "learning_rate": 2.698505108173077e-05, "loss": 0.1851, "step": 12260 }, { "epoch": 3.68, "learning_rate": 2.698317307692308e-05, "loss": 0.1638, "step": 12261 }, { "epoch": 3.68, "learning_rate": 2.6981295072115387e-05, "loss": 0.129, "step": 12262 }, { "epoch": 3.68, "learning_rate": 2.6979417067307695e-05, "loss": 0.2108, "step": 12263 }, { "epoch": 3.69, "learning_rate": 2.69775390625e-05, "loss": 0.1269, "step": 12264 }, { "epoch": 3.69, "learning_rate": 2.697566105769231e-05, "loss": 0.1075, "step": 12265 }, { "epoch": 3.69, "learning_rate": 2.6973783052884617e-05, "loss": 0.117, "step": 12266 }, { "epoch": 3.69, "learning_rate": 2.6971905048076925e-05, "loss": 0.2065, "step": 12267 }, { "epoch": 3.69, "learning_rate": 2.697002704326923e-05, "loss": 0.1888, "step": 12268 }, { "epoch": 3.69, "learning_rate": 2.6968149038461538e-05, "loss": 0.1814, "step": 12269 }, { "epoch": 3.69, "learning_rate": 2.6966271033653846e-05, "loss": 0.1339, "step": 12270 }, { "epoch": 3.69, "learning_rate": 2.6964393028846157e-05, "loss": 0.1711, "step": 12271 }, { "epoch": 3.69, "learning_rate": 2.6962515024038466e-05, "loss": 0.168, "step": 12272 }, { "epoch": 3.69, "learning_rate": 2.6960637019230774e-05, "loss": 0.1028, "step": 12273 }, { "epoch": 3.69, "learning_rate": 2.695875901442308e-05, "loss": 0.1958, "step": 12274 }, { "epoch": 3.69, "learning_rate": 2.6956881009615387e-05, "loss": 0.1568, "step": 12275 }, { "epoch": 3.69, "learning_rate": 2.6955003004807695e-05, "loss": 0.1062, "step": 12276 }, { "epoch": 3.69, "learning_rate": 2.6953125000000003e-05, "loss": 0.1085, "step": 12277 }, { "epoch": 3.69, "learning_rate": 2.6951246995192308e-05, "loss": 0.1399, "step": 12278 }, { "epoch": 3.69, "learning_rate": 2.6949368990384616e-05, "loss": 0.1641, "step": 12279 }, { "epoch": 3.69, "learning_rate": 2.6947490985576924e-05, "loss": 0.1448, "step": 12280 }, { "epoch": 3.69, "learning_rate": 2.694561298076923e-05, "loss": 0.1215, "step": 12281 }, { "epoch": 3.69, "learning_rate": 2.6943734975961537e-05, "loss": 0.1163, "step": 12282 }, { "epoch": 3.69, "learning_rate": 2.6941856971153845e-05, "loss": 0.1624, "step": 12283 }, { "epoch": 3.69, "learning_rate": 2.6939978966346157e-05, "loss": 0.0879, "step": 12284 }, { "epoch": 3.69, "learning_rate": 2.6938100961538465e-05, "loss": 0.1581, "step": 12285 }, { "epoch": 3.69, "learning_rate": 2.6936222956730773e-05, "loss": 0.1375, "step": 12286 }, { "epoch": 3.69, "learning_rate": 2.693434495192308e-05, "loss": 0.2086, "step": 12287 }, { "epoch": 3.69, "learning_rate": 2.6932466947115386e-05, "loss": 0.0914, "step": 12288 }, { "epoch": 3.69, "learning_rate": 2.6930588942307694e-05, "loss": 0.1135, "step": 12289 }, { "epoch": 3.69, "learning_rate": 2.6928710937500002e-05, "loss": 0.1745, "step": 12290 }, { "epoch": 3.69, "learning_rate": 2.6926832932692307e-05, "loss": 0.1787, "step": 12291 }, { "epoch": 3.69, "learning_rate": 2.6924954927884615e-05, "loss": 0.1424, "step": 12292 }, { "epoch": 3.69, "learning_rate": 2.6923076923076923e-05, "loss": 0.1386, "step": 12293 }, { "epoch": 3.69, "learning_rate": 2.692119891826923e-05, "loss": 0.1094, "step": 12294 }, { "epoch": 3.69, "learning_rate": 2.6919320913461536e-05, "loss": 0.1473, "step": 12295 }, { "epoch": 3.69, "learning_rate": 2.6917442908653844e-05, "loss": 0.1178, "step": 12296 }, { "epoch": 3.7, "learning_rate": 2.691556490384616e-05, "loss": 0.1935, "step": 12297 }, { "epoch": 3.7, "learning_rate": 2.6913686899038464e-05, "loss": 0.1246, "step": 12298 }, { "epoch": 3.7, "learning_rate": 2.6911808894230772e-05, "loss": 0.1043, "step": 12299 }, { "epoch": 3.7, "learning_rate": 2.690993088942308e-05, "loss": 0.1727, "step": 12300 }, { "epoch": 3.7, "learning_rate": 2.6908052884615385e-05, "loss": 0.1643, "step": 12301 }, { "epoch": 3.7, "learning_rate": 2.6906174879807693e-05, "loss": 0.1804, "step": 12302 }, { "epoch": 3.7, "learning_rate": 2.6904296875e-05, "loss": 0.1152, "step": 12303 }, { "epoch": 3.7, "learning_rate": 2.690241887019231e-05, "loss": 0.1348, "step": 12304 }, { "epoch": 3.7, "learning_rate": 2.6900540865384614e-05, "loss": 0.1534, "step": 12305 }, { "epoch": 3.7, "learning_rate": 2.6898662860576923e-05, "loss": 0.1508, "step": 12306 }, { "epoch": 3.7, "learning_rate": 2.689678485576923e-05, "loss": 0.1207, "step": 12307 }, { "epoch": 3.7, "learning_rate": 2.689490685096154e-05, "loss": 0.1166, "step": 12308 }, { "epoch": 3.7, "learning_rate": 2.6893028846153844e-05, "loss": 0.1532, "step": 12309 }, { "epoch": 3.7, "learning_rate": 2.689115084134616e-05, "loss": 0.1712, "step": 12310 }, { "epoch": 3.7, "learning_rate": 2.6889272836538463e-05, "loss": 0.1328, "step": 12311 }, { "epoch": 3.7, "learning_rate": 2.688739483173077e-05, "loss": 0.186, "step": 12312 }, { "epoch": 3.7, "learning_rate": 2.688551682692308e-05, "loss": 0.1625, "step": 12313 }, { "epoch": 3.7, "learning_rate": 2.6883638822115388e-05, "loss": 0.1675, "step": 12314 }, { "epoch": 3.7, "learning_rate": 2.6881760817307693e-05, "loss": 0.1267, "step": 12315 }, { "epoch": 3.7, "learning_rate": 2.68798828125e-05, "loss": 0.1451, "step": 12316 }, { "epoch": 3.7, "learning_rate": 2.687800480769231e-05, "loss": 0.1555, "step": 12317 }, { "epoch": 3.7, "learning_rate": 2.6876126802884617e-05, "loss": 0.2681, "step": 12318 }, { "epoch": 3.7, "learning_rate": 2.6874248798076922e-05, "loss": 0.1395, "step": 12319 }, { "epoch": 3.7, "learning_rate": 2.687237079326923e-05, "loss": 0.2034, "step": 12320 }, { "epoch": 3.7, "learning_rate": 2.6870492788461538e-05, "loss": 0.2281, "step": 12321 }, { "epoch": 3.7, "learning_rate": 2.6868614783653846e-05, "loss": 0.12, "step": 12322 }, { "epoch": 3.7, "learning_rate": 2.6866736778846158e-05, "loss": 0.0938, "step": 12323 }, { "epoch": 3.7, "learning_rate": 2.6864858774038466e-05, "loss": 0.1199, "step": 12324 }, { "epoch": 3.7, "learning_rate": 2.686298076923077e-05, "loss": 0.1597, "step": 12325 }, { "epoch": 3.7, "learning_rate": 2.686110276442308e-05, "loss": 0.105, "step": 12326 }, { "epoch": 3.7, "learning_rate": 2.6859224759615387e-05, "loss": 0.0974, "step": 12327 }, { "epoch": 3.7, "learning_rate": 2.6857346754807695e-05, "loss": 0.168, "step": 12328 }, { "epoch": 3.7, "learning_rate": 2.685546875e-05, "loss": 0.135, "step": 12329 }, { "epoch": 3.7, "learning_rate": 2.6853590745192308e-05, "loss": 0.2087, "step": 12330 }, { "epoch": 3.71, "learning_rate": 2.6851712740384616e-05, "loss": 0.1422, "step": 12331 }, { "epoch": 3.71, "learning_rate": 2.6849834735576925e-05, "loss": 0.1322, "step": 12332 }, { "epoch": 3.71, "learning_rate": 2.684795673076923e-05, "loss": 0.1171, "step": 12333 }, { "epoch": 3.71, "learning_rate": 2.6846078725961538e-05, "loss": 0.1475, "step": 12334 }, { "epoch": 3.71, "learning_rate": 2.6844200721153846e-05, "loss": 0.1182, "step": 12335 }, { "epoch": 3.71, "learning_rate": 2.6842322716346157e-05, "loss": 0.1554, "step": 12336 }, { "epoch": 3.71, "learning_rate": 2.6840444711538465e-05, "loss": 0.162, "step": 12337 }, { "epoch": 3.71, "learning_rate": 2.6838566706730774e-05, "loss": 0.2099, "step": 12338 }, { "epoch": 3.71, "learning_rate": 2.683668870192308e-05, "loss": 0.1791, "step": 12339 }, { "epoch": 3.71, "learning_rate": 2.6834810697115387e-05, "loss": 0.1317, "step": 12340 }, { "epoch": 3.71, "learning_rate": 2.6832932692307695e-05, "loss": 0.1835, "step": 12341 }, { "epoch": 3.71, "learning_rate": 2.6831054687500003e-05, "loss": 0.1585, "step": 12342 }, { "epoch": 3.71, "learning_rate": 2.6829176682692308e-05, "loss": 0.1952, "step": 12343 }, { "epoch": 3.71, "learning_rate": 2.6827298677884616e-05, "loss": 0.1023, "step": 12344 }, { "epoch": 3.71, "learning_rate": 2.6825420673076924e-05, "loss": 0.1056, "step": 12345 }, { "epoch": 3.71, "learning_rate": 2.682354266826923e-05, "loss": 0.1432, "step": 12346 }, { "epoch": 3.71, "learning_rate": 2.6821664663461537e-05, "loss": 0.2016, "step": 12347 }, { "epoch": 3.71, "learning_rate": 2.6819786658653845e-05, "loss": 0.1482, "step": 12348 }, { "epoch": 3.71, "learning_rate": 2.6817908653846157e-05, "loss": 0.1607, "step": 12349 }, { "epoch": 3.71, "learning_rate": 2.6816030649038465e-05, "loss": 0.1429, "step": 12350 }, { "epoch": 3.71, "learning_rate": 2.6814152644230773e-05, "loss": 0.2453, "step": 12351 }, { "epoch": 3.71, "learning_rate": 2.681227463942308e-05, "loss": 0.1424, "step": 12352 }, { "epoch": 3.71, "learning_rate": 2.6810396634615386e-05, "loss": 0.1262, "step": 12353 }, { "epoch": 3.71, "learning_rate": 2.6808518629807694e-05, "loss": 0.1804, "step": 12354 }, { "epoch": 3.71, "learning_rate": 2.6806640625000002e-05, "loss": 0.1273, "step": 12355 }, { "epoch": 3.71, "learning_rate": 2.6804762620192307e-05, "loss": 0.1366, "step": 12356 }, { "epoch": 3.71, "learning_rate": 2.6802884615384615e-05, "loss": 0.0992, "step": 12357 }, { "epoch": 3.71, "learning_rate": 2.6801006610576923e-05, "loss": 0.1301, "step": 12358 }, { "epoch": 3.71, "learning_rate": 2.679912860576923e-05, "loss": 0.2489, "step": 12359 }, { "epoch": 3.71, "learning_rate": 2.6797250600961536e-05, "loss": 0.1685, "step": 12360 }, { "epoch": 3.71, "learning_rate": 2.6795372596153844e-05, "loss": 0.1295, "step": 12361 }, { "epoch": 3.71, "learning_rate": 2.679349459134616e-05, "loss": 0.1494, "step": 12362 }, { "epoch": 3.71, "learning_rate": 2.6791616586538464e-05, "loss": 0.1742, "step": 12363 }, { "epoch": 3.72, "learning_rate": 2.6789738581730772e-05, "loss": 0.129, "step": 12364 }, { "epoch": 3.72, "learning_rate": 2.678786057692308e-05, "loss": 0.1456, "step": 12365 }, { "epoch": 3.72, "learning_rate": 2.6785982572115385e-05, "loss": 0.1179, "step": 12366 }, { "epoch": 3.72, "learning_rate": 2.6784104567307693e-05, "loss": 0.1614, "step": 12367 }, { "epoch": 3.72, "learning_rate": 2.67822265625e-05, "loss": 0.204, "step": 12368 }, { "epoch": 3.72, "learning_rate": 2.678034855769231e-05, "loss": 0.1438, "step": 12369 }, { "epoch": 3.72, "learning_rate": 2.6778470552884614e-05, "loss": 0.1576, "step": 12370 }, { "epoch": 3.72, "learning_rate": 2.6776592548076923e-05, "loss": 0.1789, "step": 12371 }, { "epoch": 3.72, "learning_rate": 2.677471454326923e-05, "loss": 0.1147, "step": 12372 }, { "epoch": 3.72, "learning_rate": 2.677283653846154e-05, "loss": 0.1276, "step": 12373 }, { "epoch": 3.72, "learning_rate": 2.6770958533653844e-05, "loss": 0.1202, "step": 12374 }, { "epoch": 3.72, "learning_rate": 2.676908052884616e-05, "loss": 0.2056, "step": 12375 }, { "epoch": 3.72, "learning_rate": 2.6767202524038463e-05, "loss": 0.1075, "step": 12376 }, { "epoch": 3.72, "learning_rate": 2.676532451923077e-05, "loss": 0.17, "step": 12377 }, { "epoch": 3.72, "learning_rate": 2.676344651442308e-05, "loss": 0.1398, "step": 12378 }, { "epoch": 3.72, "learning_rate": 2.6761568509615388e-05, "loss": 0.1222, "step": 12379 }, { "epoch": 3.72, "learning_rate": 2.6759690504807693e-05, "loss": 0.2148, "step": 12380 }, { "epoch": 3.72, "learning_rate": 2.67578125e-05, "loss": 0.115, "step": 12381 }, { "epoch": 3.72, "learning_rate": 2.675593449519231e-05, "loss": 0.1065, "step": 12382 }, { "epoch": 3.72, "learning_rate": 2.6754056490384617e-05, "loss": 0.1571, "step": 12383 }, { "epoch": 3.72, "learning_rate": 2.6752178485576922e-05, "loss": 0.1921, "step": 12384 }, { "epoch": 3.72, "learning_rate": 2.675030048076923e-05, "loss": 0.1704, "step": 12385 }, { "epoch": 3.72, "learning_rate": 2.6748422475961538e-05, "loss": 0.1065, "step": 12386 }, { "epoch": 3.72, "learning_rate": 2.6746544471153846e-05, "loss": 0.2127, "step": 12387 }, { "epoch": 3.72, "learning_rate": 2.6744666466346158e-05, "loss": 0.2048, "step": 12388 }, { "epoch": 3.72, "learning_rate": 2.6742788461538466e-05, "loss": 0.131, "step": 12389 }, { "epoch": 3.72, "learning_rate": 2.674091045673077e-05, "loss": 0.1357, "step": 12390 }, { "epoch": 3.72, "learning_rate": 2.673903245192308e-05, "loss": 0.2125, "step": 12391 }, { "epoch": 3.72, "learning_rate": 2.6737154447115387e-05, "loss": 0.1115, "step": 12392 }, { "epoch": 3.72, "learning_rate": 2.6735276442307695e-05, "loss": 0.1239, "step": 12393 }, { "epoch": 3.72, "learning_rate": 2.67333984375e-05, "loss": 0.1361, "step": 12394 }, { "epoch": 3.72, "learning_rate": 2.6731520432692308e-05, "loss": 0.1123, "step": 12395 }, { "epoch": 3.72, "learning_rate": 2.6729642427884616e-05, "loss": 0.1038, "step": 12396 }, { "epoch": 3.73, "learning_rate": 2.6727764423076925e-05, "loss": 0.1336, "step": 12397 }, { "epoch": 3.73, "learning_rate": 2.672588641826923e-05, "loss": 0.1608, "step": 12398 }, { "epoch": 3.73, "learning_rate": 2.6724008413461537e-05, "loss": 0.1897, "step": 12399 }, { "epoch": 3.73, "learning_rate": 2.6722130408653846e-05, "loss": 0.1359, "step": 12400 }, { "epoch": 3.73, "learning_rate": 2.6720252403846157e-05, "loss": 0.1744, "step": 12401 }, { "epoch": 3.73, "learning_rate": 2.6718374399038465e-05, "loss": 0.1226, "step": 12402 }, { "epoch": 3.73, "learning_rate": 2.6716496394230773e-05, "loss": 0.1143, "step": 12403 }, { "epoch": 3.73, "learning_rate": 2.6714618389423078e-05, "loss": 0.2076, "step": 12404 }, { "epoch": 3.73, "learning_rate": 2.6712740384615386e-05, "loss": 0.1237, "step": 12405 }, { "epoch": 3.73, "learning_rate": 2.6710862379807695e-05, "loss": 0.1275, "step": 12406 }, { "epoch": 3.73, "learning_rate": 2.6708984375000003e-05, "loss": 0.1769, "step": 12407 }, { "epoch": 3.73, "learning_rate": 2.6707106370192307e-05, "loss": 0.0978, "step": 12408 }, { "epoch": 3.73, "learning_rate": 2.6705228365384616e-05, "loss": 0.1425, "step": 12409 }, { "epoch": 3.73, "learning_rate": 2.6703350360576924e-05, "loss": 0.1717, "step": 12410 }, { "epoch": 3.73, "learning_rate": 2.6701472355769232e-05, "loss": 0.2147, "step": 12411 }, { "epoch": 3.73, "learning_rate": 2.6699594350961537e-05, "loss": 0.108, "step": 12412 }, { "epoch": 3.73, "learning_rate": 2.6697716346153845e-05, "loss": 0.1772, "step": 12413 }, { "epoch": 3.73, "learning_rate": 2.6695838341346156e-05, "loss": 0.1278, "step": 12414 }, { "epoch": 3.73, "learning_rate": 2.6693960336538465e-05, "loss": 0.1548, "step": 12415 }, { "epoch": 3.73, "learning_rate": 2.6692082331730773e-05, "loss": 0.13, "step": 12416 }, { "epoch": 3.73, "learning_rate": 2.669020432692308e-05, "loss": 0.1841, "step": 12417 }, { "epoch": 3.73, "learning_rate": 2.6688326322115386e-05, "loss": 0.1328, "step": 12418 }, { "epoch": 3.73, "learning_rate": 2.6686448317307694e-05, "loss": 0.1212, "step": 12419 }, { "epoch": 3.73, "learning_rate": 2.6684570312500002e-05, "loss": 0.1133, "step": 12420 }, { "epoch": 3.73, "learning_rate": 2.668269230769231e-05, "loss": 0.1623, "step": 12421 }, { "epoch": 3.73, "learning_rate": 2.6680814302884615e-05, "loss": 0.1564, "step": 12422 }, { "epoch": 3.73, "learning_rate": 2.6678936298076923e-05, "loss": 0.1386, "step": 12423 }, { "epoch": 3.73, "learning_rate": 2.667705829326923e-05, "loss": 0.1925, "step": 12424 }, { "epoch": 3.73, "learning_rate": 2.6675180288461536e-05, "loss": 0.1174, "step": 12425 }, { "epoch": 3.73, "learning_rate": 2.6673302283653844e-05, "loss": 0.1492, "step": 12426 }, { "epoch": 3.73, "learning_rate": 2.667142427884616e-05, "loss": 0.0962, "step": 12427 }, { "epoch": 3.73, "learning_rate": 2.6669546274038464e-05, "loss": 0.1607, "step": 12428 }, { "epoch": 3.73, "learning_rate": 2.6667668269230772e-05, "loss": 0.148, "step": 12429 }, { "epoch": 3.73, "learning_rate": 2.666579026442308e-05, "loss": 0.0811, "step": 12430 }, { "epoch": 3.74, "learning_rate": 2.666391225961539e-05, "loss": 0.2099, "step": 12431 }, { "epoch": 3.74, "learning_rate": 2.6662034254807693e-05, "loss": 0.1531, "step": 12432 }, { "epoch": 3.74, "learning_rate": 2.666015625e-05, "loss": 0.113, "step": 12433 }, { "epoch": 3.74, "learning_rate": 2.665827824519231e-05, "loss": 0.1718, "step": 12434 }, { "epoch": 3.74, "learning_rate": 2.6656400240384614e-05, "loss": 0.1631, "step": 12435 }, { "epoch": 3.74, "learning_rate": 2.6654522235576922e-05, "loss": 0.1117, "step": 12436 }, { "epoch": 3.74, "learning_rate": 2.665264423076923e-05, "loss": 0.2189, "step": 12437 }, { "epoch": 3.74, "learning_rate": 2.665076622596154e-05, "loss": 0.1522, "step": 12438 }, { "epoch": 3.74, "learning_rate": 2.6648888221153843e-05, "loss": 0.1416, "step": 12439 }, { "epoch": 3.74, "learning_rate": 2.664701021634616e-05, "loss": 0.1698, "step": 12440 }, { "epoch": 3.74, "learning_rate": 2.6645132211538467e-05, "loss": 0.1378, "step": 12441 }, { "epoch": 3.74, "learning_rate": 2.664325420673077e-05, "loss": 0.1847, "step": 12442 }, { "epoch": 3.74, "learning_rate": 2.664137620192308e-05, "loss": 0.2039, "step": 12443 }, { "epoch": 3.74, "learning_rate": 2.6639498197115388e-05, "loss": 0.1669, "step": 12444 }, { "epoch": 3.74, "learning_rate": 2.6637620192307692e-05, "loss": 0.1626, "step": 12445 }, { "epoch": 3.74, "learning_rate": 2.66357421875e-05, "loss": 0.1584, "step": 12446 }, { "epoch": 3.74, "learning_rate": 2.663386418269231e-05, "loss": 0.1055, "step": 12447 }, { "epoch": 3.74, "learning_rate": 2.6631986177884617e-05, "loss": 0.1738, "step": 12448 }, { "epoch": 3.74, "learning_rate": 2.6630108173076922e-05, "loss": 0.141, "step": 12449 }, { "epoch": 3.74, "learning_rate": 2.662823016826923e-05, "loss": 0.14, "step": 12450 }, { "epoch": 3.74, "learning_rate": 2.6626352163461538e-05, "loss": 0.148, "step": 12451 }, { "epoch": 3.74, "learning_rate": 2.6624474158653846e-05, "loss": 0.1383, "step": 12452 }, { "epoch": 3.74, "learning_rate": 2.6622596153846158e-05, "loss": 0.1726, "step": 12453 }, { "epoch": 3.74, "learning_rate": 2.6620718149038466e-05, "loss": 0.1528, "step": 12454 }, { "epoch": 3.74, "learning_rate": 2.661884014423077e-05, "loss": 0.1596, "step": 12455 }, { "epoch": 3.74, "learning_rate": 2.661696213942308e-05, "loss": 0.134, "step": 12456 }, { "epoch": 3.74, "learning_rate": 2.6615084134615387e-05, "loss": 0.2218, "step": 12457 }, { "epoch": 3.74, "learning_rate": 2.6613206129807695e-05, "loss": 0.1397, "step": 12458 }, { "epoch": 3.74, "learning_rate": 2.6611328125e-05, "loss": 0.1493, "step": 12459 }, { "epoch": 3.74, "learning_rate": 2.6609450120192308e-05, "loss": 0.1566, "step": 12460 }, { "epoch": 3.74, "learning_rate": 2.6607572115384616e-05, "loss": 0.1267, "step": 12461 }, { "epoch": 3.74, "learning_rate": 2.6605694110576924e-05, "loss": 0.1304, "step": 12462 }, { "epoch": 3.74, "learning_rate": 2.660381610576923e-05, "loss": 0.2063, "step": 12463 }, { "epoch": 3.75, "learning_rate": 2.6601938100961537e-05, "loss": 0.1864, "step": 12464 }, { "epoch": 3.75, "learning_rate": 2.6600060096153845e-05, "loss": 0.1722, "step": 12465 }, { "epoch": 3.75, "learning_rate": 2.6598182091346157e-05, "loss": 0.1269, "step": 12466 }, { "epoch": 3.75, "learning_rate": 2.6596304086538465e-05, "loss": 0.1186, "step": 12467 }, { "epoch": 3.75, "learning_rate": 2.6594426081730773e-05, "loss": 0.1279, "step": 12468 }, { "epoch": 3.75, "learning_rate": 2.6592548076923078e-05, "loss": 0.1499, "step": 12469 }, { "epoch": 3.75, "learning_rate": 2.6590670072115386e-05, "loss": 0.1718, "step": 12470 }, { "epoch": 3.75, "learning_rate": 2.6588792067307694e-05, "loss": 0.1137, "step": 12471 }, { "epoch": 3.75, "learning_rate": 2.6586914062500003e-05, "loss": 0.1868, "step": 12472 }, { "epoch": 3.75, "learning_rate": 2.6585036057692307e-05, "loss": 0.1572, "step": 12473 }, { "epoch": 3.75, "learning_rate": 2.6583158052884616e-05, "loss": 0.147, "step": 12474 }, { "epoch": 3.75, "learning_rate": 2.6581280048076924e-05, "loss": 0.1727, "step": 12475 }, { "epoch": 3.75, "learning_rate": 2.6579402043269232e-05, "loss": 0.2121, "step": 12476 }, { "epoch": 3.75, "learning_rate": 2.6577524038461537e-05, "loss": 0.1209, "step": 12477 }, { "epoch": 3.75, "learning_rate": 2.6575646033653845e-05, "loss": 0.1499, "step": 12478 }, { "epoch": 3.75, "learning_rate": 2.6573768028846156e-05, "loss": 0.1299, "step": 12479 }, { "epoch": 3.75, "learning_rate": 2.6571890024038464e-05, "loss": 0.1317, "step": 12480 }, { "epoch": 3.75, "learning_rate": 2.6570012019230773e-05, "loss": 0.1361, "step": 12481 }, { "epoch": 3.75, "learning_rate": 2.656813401442308e-05, "loss": 0.1468, "step": 12482 }, { "epoch": 3.75, "learning_rate": 2.6566256009615386e-05, "loss": 0.2359, "step": 12483 }, { "epoch": 3.75, "learning_rate": 2.6564378004807694e-05, "loss": 0.1445, "step": 12484 }, { "epoch": 3.75, "learning_rate": 2.6562500000000002e-05, "loss": 0.1418, "step": 12485 }, { "epoch": 3.75, "learning_rate": 2.656062199519231e-05, "loss": 0.1782, "step": 12486 }, { "epoch": 3.75, "learning_rate": 2.6558743990384615e-05, "loss": 0.1805, "step": 12487 }, { "epoch": 3.75, "learning_rate": 2.6556865985576923e-05, "loss": 0.1593, "step": 12488 }, { "epoch": 3.75, "learning_rate": 2.655498798076923e-05, "loss": 0.3308, "step": 12489 }, { "epoch": 3.75, "learning_rate": 2.6553109975961536e-05, "loss": 0.1366, "step": 12490 }, { "epoch": 3.75, "learning_rate": 2.6551231971153844e-05, "loss": 0.1627, "step": 12491 }, { "epoch": 3.75, "learning_rate": 2.654935396634616e-05, "loss": 0.2418, "step": 12492 }, { "epoch": 3.75, "learning_rate": 2.6547475961538464e-05, "loss": 0.1234, "step": 12493 }, { "epoch": 3.75, "learning_rate": 2.6545597956730772e-05, "loss": 0.1347, "step": 12494 }, { "epoch": 3.75, "learning_rate": 2.654371995192308e-05, "loss": 0.1337, "step": 12495 }, { "epoch": 3.75, "learning_rate": 2.6541841947115388e-05, "loss": 0.1153, "step": 12496 }, { "epoch": 3.76, "learning_rate": 2.6539963942307693e-05, "loss": 0.1119, "step": 12497 }, { "epoch": 3.76, "learning_rate": 2.65380859375e-05, "loss": 0.1179, "step": 12498 }, { "epoch": 3.76, "learning_rate": 2.653620793269231e-05, "loss": 0.1582, "step": 12499 }, { "epoch": 3.76, "learning_rate": 2.6534329927884614e-05, "loss": 0.1236, "step": 12500 }, { "epoch": 3.76, "learning_rate": 2.6532451923076922e-05, "loss": 0.1572, "step": 12501 }, { "epoch": 3.76, "learning_rate": 2.653057391826923e-05, "loss": 0.1138, "step": 12502 }, { "epoch": 3.76, "learning_rate": 2.652869591346154e-05, "loss": 0.1805, "step": 12503 }, { "epoch": 3.76, "learning_rate": 2.6526817908653843e-05, "loss": 0.1815, "step": 12504 }, { "epoch": 3.76, "learning_rate": 2.652493990384616e-05, "loss": 0.0881, "step": 12505 }, { "epoch": 3.76, "learning_rate": 2.6523061899038466e-05, "loss": 0.1455, "step": 12506 }, { "epoch": 3.76, "learning_rate": 2.652118389423077e-05, "loss": 0.1596, "step": 12507 }, { "epoch": 3.76, "learning_rate": 2.651930588942308e-05, "loss": 0.1478, "step": 12508 }, { "epoch": 3.76, "learning_rate": 2.6517427884615388e-05, "loss": 0.1197, "step": 12509 }, { "epoch": 3.76, "learning_rate": 2.6515549879807692e-05, "loss": 0.089, "step": 12510 }, { "epoch": 3.76, "learning_rate": 2.6513671875e-05, "loss": 0.1235, "step": 12511 }, { "epoch": 3.76, "learning_rate": 2.651179387019231e-05, "loss": 0.1389, "step": 12512 }, { "epoch": 3.76, "learning_rate": 2.6509915865384617e-05, "loss": 0.1539, "step": 12513 }, { "epoch": 3.76, "learning_rate": 2.650803786057692e-05, "loss": 0.1483, "step": 12514 }, { "epoch": 3.76, "learning_rate": 2.650615985576923e-05, "loss": 0.1941, "step": 12515 }, { "epoch": 3.76, "learning_rate": 2.6504281850961538e-05, "loss": 0.2058, "step": 12516 }, { "epoch": 3.76, "learning_rate": 2.6502403846153846e-05, "loss": 0.1609, "step": 12517 }, { "epoch": 3.76, "learning_rate": 2.6500525841346158e-05, "loss": 0.1507, "step": 12518 }, { "epoch": 3.76, "learning_rate": 2.6498647836538466e-05, "loss": 0.1466, "step": 12519 }, { "epoch": 3.76, "learning_rate": 2.649676983173077e-05, "loss": 0.2132, "step": 12520 }, { "epoch": 3.76, "learning_rate": 2.649489182692308e-05, "loss": 0.1801, "step": 12521 }, { "epoch": 3.76, "learning_rate": 2.6493013822115387e-05, "loss": 0.1071, "step": 12522 }, { "epoch": 3.76, "learning_rate": 2.6491135817307695e-05, "loss": 0.1881, "step": 12523 }, { "epoch": 3.76, "learning_rate": 2.64892578125e-05, "loss": 0.1699, "step": 12524 }, { "epoch": 3.76, "learning_rate": 2.6487379807692308e-05, "loss": 0.2069, "step": 12525 }, { "epoch": 3.76, "learning_rate": 2.6485501802884616e-05, "loss": 0.1566, "step": 12526 }, { "epoch": 3.76, "learning_rate": 2.6483623798076924e-05, "loss": 0.1146, "step": 12527 }, { "epoch": 3.76, "learning_rate": 2.648174579326923e-05, "loss": 0.1186, "step": 12528 }, { "epoch": 3.76, "learning_rate": 2.6479867788461537e-05, "loss": 0.1736, "step": 12529 }, { "epoch": 3.77, "learning_rate": 2.6477989783653845e-05, "loss": 0.1513, "step": 12530 }, { "epoch": 3.77, "learning_rate": 2.6476111778846157e-05, "loss": 0.1713, "step": 12531 }, { "epoch": 3.77, "learning_rate": 2.6474233774038465e-05, "loss": 0.2047, "step": 12532 }, { "epoch": 3.77, "learning_rate": 2.6472355769230773e-05, "loss": 0.1959, "step": 12533 }, { "epoch": 3.77, "learning_rate": 2.6470477764423078e-05, "loss": 0.1594, "step": 12534 }, { "epoch": 3.77, "learning_rate": 2.6468599759615386e-05, "loss": 0.1326, "step": 12535 }, { "epoch": 3.77, "learning_rate": 2.6466721754807694e-05, "loss": 0.1673, "step": 12536 }, { "epoch": 3.77, "learning_rate": 2.6464843750000002e-05, "loss": 0.1807, "step": 12537 }, { "epoch": 3.77, "learning_rate": 2.6462965745192307e-05, "loss": 0.1422, "step": 12538 }, { "epoch": 3.77, "learning_rate": 2.6461087740384615e-05, "loss": 0.1317, "step": 12539 }, { "epoch": 3.77, "learning_rate": 2.6459209735576924e-05, "loss": 0.1495, "step": 12540 }, { "epoch": 3.77, "learning_rate": 2.6457331730769232e-05, "loss": 0.1645, "step": 12541 }, { "epoch": 3.77, "learning_rate": 2.6455453725961537e-05, "loss": 0.1524, "step": 12542 }, { "epoch": 3.77, "learning_rate": 2.6453575721153845e-05, "loss": 0.0956, "step": 12543 }, { "epoch": 3.77, "learning_rate": 2.6451697716346156e-05, "loss": 0.1291, "step": 12544 }, { "epoch": 3.77, "learning_rate": 2.6449819711538464e-05, "loss": 0.1191, "step": 12545 }, { "epoch": 3.77, "learning_rate": 2.6447941706730773e-05, "loss": 0.126, "step": 12546 }, { "epoch": 3.77, "learning_rate": 2.644606370192308e-05, "loss": 0.1433, "step": 12547 }, { "epoch": 3.77, "learning_rate": 2.6444185697115385e-05, "loss": 0.1795, "step": 12548 }, { "epoch": 3.77, "learning_rate": 2.6442307692307694e-05, "loss": 0.1491, "step": 12549 }, { "epoch": 3.77, "learning_rate": 2.6440429687500002e-05, "loss": 0.1738, "step": 12550 }, { "epoch": 3.77, "learning_rate": 2.643855168269231e-05, "loss": 0.1367, "step": 12551 }, { "epoch": 3.77, "learning_rate": 2.6436673677884615e-05, "loss": 0.1639, "step": 12552 }, { "epoch": 3.77, "learning_rate": 2.6434795673076923e-05, "loss": 0.1249, "step": 12553 }, { "epoch": 3.77, "learning_rate": 2.643291766826923e-05, "loss": 0.1197, "step": 12554 }, { "epoch": 3.77, "learning_rate": 2.643103966346154e-05, "loss": 0.1902, "step": 12555 }, { "epoch": 3.77, "learning_rate": 2.6429161658653844e-05, "loss": 0.1451, "step": 12556 }, { "epoch": 3.77, "learning_rate": 2.642728365384616e-05, "loss": 0.1608, "step": 12557 }, { "epoch": 3.77, "learning_rate": 2.6425405649038464e-05, "loss": 0.1377, "step": 12558 }, { "epoch": 3.77, "learning_rate": 2.6423527644230772e-05, "loss": 0.1016, "step": 12559 }, { "epoch": 3.77, "learning_rate": 2.642164963942308e-05, "loss": 0.1179, "step": 12560 }, { "epoch": 3.77, "learning_rate": 2.6419771634615388e-05, "loss": 0.1137, "step": 12561 }, { "epoch": 3.77, "learning_rate": 2.6417893629807693e-05, "loss": 0.1478, "step": 12562 }, { "epoch": 3.77, "learning_rate": 2.6416015625e-05, "loss": 0.1151, "step": 12563 }, { "epoch": 3.78, "learning_rate": 2.641413762019231e-05, "loss": 0.1685, "step": 12564 }, { "epoch": 3.78, "learning_rate": 2.6412259615384617e-05, "loss": 0.1264, "step": 12565 }, { "epoch": 3.78, "learning_rate": 2.6410381610576922e-05, "loss": 0.1285, "step": 12566 }, { "epoch": 3.78, "learning_rate": 2.640850360576923e-05, "loss": 0.1705, "step": 12567 }, { "epoch": 3.78, "learning_rate": 2.640662560096154e-05, "loss": 0.1453, "step": 12568 }, { "epoch": 3.78, "learning_rate": 2.6404747596153843e-05, "loss": 0.1272, "step": 12569 }, { "epoch": 3.78, "learning_rate": 2.6402869591346158e-05, "loss": 0.1485, "step": 12570 }, { "epoch": 3.78, "learning_rate": 2.6400991586538466e-05, "loss": 0.1652, "step": 12571 }, { "epoch": 3.78, "learning_rate": 2.639911358173077e-05, "loss": 0.1391, "step": 12572 }, { "epoch": 3.78, "learning_rate": 2.639723557692308e-05, "loss": 0.2135, "step": 12573 }, { "epoch": 3.78, "learning_rate": 2.6395357572115387e-05, "loss": 0.0967, "step": 12574 }, { "epoch": 3.78, "learning_rate": 2.6393479567307696e-05, "loss": 0.1105, "step": 12575 }, { "epoch": 3.78, "learning_rate": 2.63916015625e-05, "loss": 0.1334, "step": 12576 }, { "epoch": 3.78, "learning_rate": 2.638972355769231e-05, "loss": 0.1649, "step": 12577 }, { "epoch": 3.78, "learning_rate": 2.6387845552884617e-05, "loss": 0.1363, "step": 12578 }, { "epoch": 3.78, "learning_rate": 2.638596754807692e-05, "loss": 0.2058, "step": 12579 }, { "epoch": 3.78, "learning_rate": 2.638408954326923e-05, "loss": 0.1745, "step": 12580 }, { "epoch": 3.78, "learning_rate": 2.6382211538461538e-05, "loss": 0.1222, "step": 12581 }, { "epoch": 3.78, "learning_rate": 2.6380333533653846e-05, "loss": 0.1619, "step": 12582 }, { "epoch": 3.78, "learning_rate": 2.6378455528846158e-05, "loss": 0.2141, "step": 12583 }, { "epoch": 3.78, "learning_rate": 2.6376577524038466e-05, "loss": 0.1691, "step": 12584 }, { "epoch": 3.78, "learning_rate": 2.6374699519230774e-05, "loss": 0.1654, "step": 12585 }, { "epoch": 3.78, "learning_rate": 2.637282151442308e-05, "loss": 0.231, "step": 12586 }, { "epoch": 3.78, "learning_rate": 2.6370943509615387e-05, "loss": 0.1441, "step": 12587 }, { "epoch": 3.78, "learning_rate": 2.6369065504807695e-05, "loss": 0.1533, "step": 12588 }, { "epoch": 3.78, "learning_rate": 2.63671875e-05, "loss": 0.0867, "step": 12589 }, { "epoch": 3.78, "learning_rate": 2.6365309495192308e-05, "loss": 0.2873, "step": 12590 }, { "epoch": 3.78, "learning_rate": 2.6363431490384616e-05, "loss": 0.129, "step": 12591 }, { "epoch": 3.78, "learning_rate": 2.6361553485576924e-05, "loss": 0.1303, "step": 12592 }, { "epoch": 3.78, "learning_rate": 2.635967548076923e-05, "loss": 0.1662, "step": 12593 }, { "epoch": 3.78, "learning_rate": 2.6357797475961537e-05, "loss": 0.2055, "step": 12594 }, { "epoch": 3.78, "learning_rate": 2.6355919471153845e-05, "loss": 0.1245, "step": 12595 }, { "epoch": 3.78, "learning_rate": 2.6354041466346157e-05, "loss": 0.1441, "step": 12596 }, { "epoch": 3.79, "learning_rate": 2.6352163461538465e-05, "loss": 0.2034, "step": 12597 }, { "epoch": 3.79, "learning_rate": 2.6350285456730773e-05, "loss": 0.2335, "step": 12598 }, { "epoch": 3.79, "learning_rate": 2.6348407451923078e-05, "loss": 0.177, "step": 12599 }, { "epoch": 3.79, "learning_rate": 2.6346529447115386e-05, "loss": 0.1352, "step": 12600 }, { "epoch": 3.79, "learning_rate": 2.6344651442307694e-05, "loss": 0.1372, "step": 12601 }, { "epoch": 3.79, "learning_rate": 2.6342773437500002e-05, "loss": 0.1807, "step": 12602 }, { "epoch": 3.79, "learning_rate": 2.6340895432692307e-05, "loss": 0.1968, "step": 12603 }, { "epoch": 3.79, "learning_rate": 2.6339017427884615e-05, "loss": 0.1551, "step": 12604 }, { "epoch": 3.79, "learning_rate": 2.6337139423076923e-05, "loss": 0.2571, "step": 12605 }, { "epoch": 3.79, "learning_rate": 2.633526141826923e-05, "loss": 0.1418, "step": 12606 }, { "epoch": 3.79, "learning_rate": 2.6333383413461536e-05, "loss": 0.1725, "step": 12607 }, { "epoch": 3.79, "learning_rate": 2.6331505408653845e-05, "loss": 0.1412, "step": 12608 }, { "epoch": 3.79, "learning_rate": 2.6329627403846156e-05, "loss": 0.1475, "step": 12609 }, { "epoch": 3.79, "learning_rate": 2.6327749399038464e-05, "loss": 0.1726, "step": 12610 }, { "epoch": 3.79, "learning_rate": 2.6325871394230772e-05, "loss": 0.1132, "step": 12611 }, { "epoch": 3.79, "learning_rate": 2.632399338942308e-05, "loss": 0.1502, "step": 12612 }, { "epoch": 3.79, "learning_rate": 2.6322115384615385e-05, "loss": 0.148, "step": 12613 }, { "epoch": 3.79, "learning_rate": 2.6320237379807694e-05, "loss": 0.1361, "step": 12614 }, { "epoch": 3.79, "learning_rate": 2.6318359375e-05, "loss": 0.1323, "step": 12615 }, { "epoch": 3.79, "learning_rate": 2.631648137019231e-05, "loss": 0.1228, "step": 12616 }, { "epoch": 3.79, "learning_rate": 2.6314603365384615e-05, "loss": 0.1408, "step": 12617 }, { "epoch": 3.79, "learning_rate": 2.6312725360576923e-05, "loss": 0.1356, "step": 12618 }, { "epoch": 3.79, "learning_rate": 2.631084735576923e-05, "loss": 0.1703, "step": 12619 }, { "epoch": 3.79, "learning_rate": 2.630896935096154e-05, "loss": 0.0737, "step": 12620 }, { "epoch": 3.79, "learning_rate": 2.6307091346153844e-05, "loss": 0.1563, "step": 12621 }, { "epoch": 3.79, "learning_rate": 2.630521334134616e-05, "loss": 0.1944, "step": 12622 }, { "epoch": 3.79, "learning_rate": 2.6303335336538464e-05, "loss": 0.1238, "step": 12623 }, { "epoch": 3.79, "learning_rate": 2.6301457331730772e-05, "loss": 0.1435, "step": 12624 }, { "epoch": 3.79, "learning_rate": 2.629957932692308e-05, "loss": 0.1906, "step": 12625 }, { "epoch": 3.79, "learning_rate": 2.6297701322115388e-05, "loss": 0.1718, "step": 12626 }, { "epoch": 3.79, "learning_rate": 2.6295823317307693e-05, "loss": 0.1199, "step": 12627 }, { "epoch": 3.79, "learning_rate": 2.62939453125e-05, "loss": 0.1568, "step": 12628 }, { "epoch": 3.79, "learning_rate": 2.629206730769231e-05, "loss": 0.1119, "step": 12629 }, { "epoch": 3.8, "learning_rate": 2.6290189302884617e-05, "loss": 0.1367, "step": 12630 }, { "epoch": 3.8, "learning_rate": 2.6288311298076922e-05, "loss": 0.1359, "step": 12631 }, { "epoch": 3.8, "learning_rate": 2.628643329326923e-05, "loss": 0.1328, "step": 12632 }, { "epoch": 3.8, "learning_rate": 2.628455528846154e-05, "loss": 0.0897, "step": 12633 }, { "epoch": 3.8, "learning_rate": 2.6282677283653843e-05, "loss": 0.122, "step": 12634 }, { "epoch": 3.8, "learning_rate": 2.6280799278846158e-05, "loss": 0.179, "step": 12635 }, { "epoch": 3.8, "learning_rate": 2.6278921274038466e-05, "loss": 0.115, "step": 12636 }, { "epoch": 3.8, "learning_rate": 2.627704326923077e-05, "loss": 0.2194, "step": 12637 }, { "epoch": 3.8, "learning_rate": 2.627516526442308e-05, "loss": 0.1445, "step": 12638 }, { "epoch": 3.8, "learning_rate": 2.6273287259615387e-05, "loss": 0.1131, "step": 12639 }, { "epoch": 3.8, "learning_rate": 2.6271409254807696e-05, "loss": 0.1492, "step": 12640 }, { "epoch": 3.8, "learning_rate": 2.626953125e-05, "loss": 0.1354, "step": 12641 }, { "epoch": 3.8, "learning_rate": 2.626765324519231e-05, "loss": 0.1684, "step": 12642 }, { "epoch": 3.8, "learning_rate": 2.6265775240384617e-05, "loss": 0.1429, "step": 12643 }, { "epoch": 3.8, "learning_rate": 2.626389723557692e-05, "loss": 0.1566, "step": 12644 }, { "epoch": 3.8, "learning_rate": 2.626201923076923e-05, "loss": 0.1239, "step": 12645 }, { "epoch": 3.8, "learning_rate": 2.6260141225961538e-05, "loss": 0.2103, "step": 12646 }, { "epoch": 3.8, "learning_rate": 2.6258263221153846e-05, "loss": 0.1746, "step": 12647 }, { "epoch": 3.8, "learning_rate": 2.6256385216346157e-05, "loss": 0.1474, "step": 12648 }, { "epoch": 3.8, "learning_rate": 2.6254507211538466e-05, "loss": 0.1609, "step": 12649 }, { "epoch": 3.8, "learning_rate": 2.6252629206730774e-05, "loss": 0.1803, "step": 12650 }, { "epoch": 3.8, "learning_rate": 2.625075120192308e-05, "loss": 0.11, "step": 12651 }, { "epoch": 3.8, "learning_rate": 2.6248873197115387e-05, "loss": 0.1367, "step": 12652 }, { "epoch": 3.8, "learning_rate": 2.6246995192307695e-05, "loss": 0.1405, "step": 12653 }, { "epoch": 3.8, "learning_rate": 2.62451171875e-05, "loss": 0.182, "step": 12654 }, { "epoch": 3.8, "learning_rate": 2.6243239182692308e-05, "loss": 0.1581, "step": 12655 }, { "epoch": 3.8, "learning_rate": 2.6241361177884616e-05, "loss": 0.1328, "step": 12656 }, { "epoch": 3.8, "learning_rate": 2.6239483173076924e-05, "loss": 0.1742, "step": 12657 }, { "epoch": 3.8, "learning_rate": 2.623760516826923e-05, "loss": 0.1836, "step": 12658 }, { "epoch": 3.8, "learning_rate": 2.6235727163461537e-05, "loss": 0.197, "step": 12659 }, { "epoch": 3.8, "learning_rate": 2.6233849158653845e-05, "loss": 0.2488, "step": 12660 }, { "epoch": 3.8, "learning_rate": 2.6231971153846157e-05, "loss": 0.1653, "step": 12661 }, { "epoch": 3.8, "learning_rate": 2.6230093149038465e-05, "loss": 0.1902, "step": 12662 }, { "epoch": 3.8, "learning_rate": 2.6228215144230773e-05, "loss": 0.153, "step": 12663 }, { "epoch": 3.81, "learning_rate": 2.6226337139423078e-05, "loss": 0.1327, "step": 12664 }, { "epoch": 3.81, "learning_rate": 2.6224459134615386e-05, "loss": 0.1526, "step": 12665 }, { "epoch": 3.81, "learning_rate": 2.6222581129807694e-05, "loss": 0.1529, "step": 12666 }, { "epoch": 3.81, "learning_rate": 2.6220703125000002e-05, "loss": 0.2073, "step": 12667 }, { "epoch": 3.81, "learning_rate": 2.6218825120192307e-05, "loss": 0.184, "step": 12668 }, { "epoch": 3.81, "learning_rate": 2.6216947115384615e-05, "loss": 0.1319, "step": 12669 }, { "epoch": 3.81, "learning_rate": 2.6215069110576923e-05, "loss": 0.1621, "step": 12670 }, { "epoch": 3.81, "learning_rate": 2.621319110576923e-05, "loss": 0.1779, "step": 12671 }, { "epoch": 3.81, "learning_rate": 2.6211313100961536e-05, "loss": 0.1434, "step": 12672 }, { "epoch": 3.81, "learning_rate": 2.6209435096153844e-05, "loss": 0.182, "step": 12673 }, { "epoch": 3.81, "learning_rate": 2.6207557091346156e-05, "loss": 0.105, "step": 12674 }, { "epoch": 3.81, "learning_rate": 2.6205679086538464e-05, "loss": 0.1465, "step": 12675 }, { "epoch": 3.81, "learning_rate": 2.6203801081730772e-05, "loss": 0.156, "step": 12676 }, { "epoch": 3.81, "learning_rate": 2.620192307692308e-05, "loss": 0.1655, "step": 12677 }, { "epoch": 3.81, "learning_rate": 2.6200045072115385e-05, "loss": 0.1633, "step": 12678 }, { "epoch": 3.81, "learning_rate": 2.6198167067307693e-05, "loss": 0.1402, "step": 12679 }, { "epoch": 3.81, "learning_rate": 2.61962890625e-05, "loss": 0.2241, "step": 12680 }, { "epoch": 3.81, "learning_rate": 2.619441105769231e-05, "loss": 0.1608, "step": 12681 }, { "epoch": 3.81, "learning_rate": 2.6192533052884615e-05, "loss": 0.1086, "step": 12682 }, { "epoch": 3.81, "learning_rate": 2.6190655048076923e-05, "loss": 0.1683, "step": 12683 }, { "epoch": 3.81, "learning_rate": 2.618877704326923e-05, "loss": 0.1777, "step": 12684 }, { "epoch": 3.81, "learning_rate": 2.618689903846154e-05, "loss": 0.1433, "step": 12685 }, { "epoch": 3.81, "learning_rate": 2.6185021033653844e-05, "loss": 0.1878, "step": 12686 }, { "epoch": 3.81, "learning_rate": 2.618314302884616e-05, "loss": 0.1149, "step": 12687 }, { "epoch": 3.81, "learning_rate": 2.6181265024038463e-05, "loss": 0.1619, "step": 12688 }, { "epoch": 3.81, "learning_rate": 2.617938701923077e-05, "loss": 0.157, "step": 12689 }, { "epoch": 3.81, "learning_rate": 2.617750901442308e-05, "loss": 0.1761, "step": 12690 }, { "epoch": 3.81, "learning_rate": 2.6175631009615388e-05, "loss": 0.1548, "step": 12691 }, { "epoch": 3.81, "learning_rate": 2.6173753004807693e-05, "loss": 0.1108, "step": 12692 }, { "epoch": 3.81, "learning_rate": 2.6171875e-05, "loss": 0.1661, "step": 12693 }, { "epoch": 3.81, "learning_rate": 2.616999699519231e-05, "loss": 0.1278, "step": 12694 }, { "epoch": 3.81, "learning_rate": 2.6168118990384617e-05, "loss": 0.1333, "step": 12695 }, { "epoch": 3.81, "learning_rate": 2.6166240985576922e-05, "loss": 0.272, "step": 12696 }, { "epoch": 3.82, "learning_rate": 2.616436298076923e-05, "loss": 0.1481, "step": 12697 }, { "epoch": 3.82, "learning_rate": 2.6162484975961538e-05, "loss": 0.1301, "step": 12698 }, { "epoch": 3.82, "learning_rate": 2.6160606971153846e-05, "loss": 0.1356, "step": 12699 }, { "epoch": 3.82, "learning_rate": 2.6158728966346158e-05, "loss": 0.2052, "step": 12700 }, { "epoch": 3.82, "learning_rate": 2.6156850961538466e-05, "loss": 0.1367, "step": 12701 }, { "epoch": 3.82, "learning_rate": 2.615497295673077e-05, "loss": 0.1482, "step": 12702 }, { "epoch": 3.82, "learning_rate": 2.615309495192308e-05, "loss": 0.195, "step": 12703 }, { "epoch": 3.82, "learning_rate": 2.6151216947115387e-05, "loss": 0.1242, "step": 12704 }, { "epoch": 3.82, "learning_rate": 2.6149338942307695e-05, "loss": 0.1863, "step": 12705 }, { "epoch": 3.82, "learning_rate": 2.61474609375e-05, "loss": 0.1891, "step": 12706 }, { "epoch": 3.82, "learning_rate": 2.614558293269231e-05, "loss": 0.1585, "step": 12707 }, { "epoch": 3.82, "learning_rate": 2.6143704927884617e-05, "loss": 0.1589, "step": 12708 }, { "epoch": 3.82, "learning_rate": 2.6141826923076925e-05, "loss": 0.1169, "step": 12709 }, { "epoch": 3.82, "learning_rate": 2.613994891826923e-05, "loss": 0.1853, "step": 12710 }, { "epoch": 3.82, "learning_rate": 2.6138070913461538e-05, "loss": 0.1157, "step": 12711 }, { "epoch": 3.82, "learning_rate": 2.6136192908653846e-05, "loss": 0.1493, "step": 12712 }, { "epoch": 3.82, "learning_rate": 2.6134314903846157e-05, "loss": 0.1024, "step": 12713 }, { "epoch": 3.82, "learning_rate": 2.6132436899038465e-05, "loss": 0.2128, "step": 12714 }, { "epoch": 3.82, "learning_rate": 2.6130558894230774e-05, "loss": 0.1946, "step": 12715 }, { "epoch": 3.82, "learning_rate": 2.612868088942308e-05, "loss": 0.1551, "step": 12716 }, { "epoch": 3.82, "learning_rate": 2.6126802884615387e-05, "loss": 0.1747, "step": 12717 }, { "epoch": 3.82, "learning_rate": 2.6124924879807695e-05, "loss": 0.1257, "step": 12718 }, { "epoch": 3.82, "learning_rate": 2.6123046875000003e-05, "loss": 0.1618, "step": 12719 }, { "epoch": 3.82, "learning_rate": 2.6121168870192308e-05, "loss": 0.1377, "step": 12720 }, { "epoch": 3.82, "learning_rate": 2.6119290865384616e-05, "loss": 0.15, "step": 12721 }, { "epoch": 3.82, "learning_rate": 2.6117412860576924e-05, "loss": 0.2453, "step": 12722 }, { "epoch": 3.82, "learning_rate": 2.611553485576923e-05, "loss": 0.13, "step": 12723 }, { "epoch": 3.82, "learning_rate": 2.6113656850961537e-05, "loss": 0.1202, "step": 12724 }, { "epoch": 3.82, "learning_rate": 2.6111778846153845e-05, "loss": 0.1206, "step": 12725 }, { "epoch": 3.82, "learning_rate": 2.6109900841346157e-05, "loss": 0.1004, "step": 12726 }, { "epoch": 3.82, "learning_rate": 2.6108022836538465e-05, "loss": 0.1781, "step": 12727 }, { "epoch": 3.82, "learning_rate": 2.6106144831730773e-05, "loss": 0.2014, "step": 12728 }, { "epoch": 3.82, "learning_rate": 2.610426682692308e-05, "loss": 0.1072, "step": 12729 }, { "epoch": 3.83, "learning_rate": 2.6102388822115386e-05, "loss": 0.1772, "step": 12730 }, { "epoch": 3.83, "learning_rate": 2.6100510817307694e-05, "loss": 0.1553, "step": 12731 }, { "epoch": 3.83, "learning_rate": 2.6098632812500002e-05, "loss": 0.1238, "step": 12732 }, { "epoch": 3.83, "learning_rate": 2.6096754807692307e-05, "loss": 0.2226, "step": 12733 }, { "epoch": 3.83, "learning_rate": 2.6094876802884615e-05, "loss": 0.136, "step": 12734 }, { "epoch": 3.83, "learning_rate": 2.6092998798076923e-05, "loss": 0.1449, "step": 12735 }, { "epoch": 3.83, "learning_rate": 2.609112079326923e-05, "loss": 0.1556, "step": 12736 }, { "epoch": 3.83, "learning_rate": 2.6089242788461536e-05, "loss": 0.1358, "step": 12737 }, { "epoch": 3.83, "learning_rate": 2.6087364783653844e-05, "loss": 0.1539, "step": 12738 }, { "epoch": 3.83, "learning_rate": 2.608548677884616e-05, "loss": 0.1713, "step": 12739 }, { "epoch": 3.83, "learning_rate": 2.6083608774038464e-05, "loss": 0.1471, "step": 12740 }, { "epoch": 3.83, "learning_rate": 2.6081730769230772e-05, "loss": 0.1808, "step": 12741 }, { "epoch": 3.83, "learning_rate": 2.607985276442308e-05, "loss": 0.1531, "step": 12742 }, { "epoch": 3.83, "learning_rate": 2.6077974759615385e-05, "loss": 0.139, "step": 12743 }, { "epoch": 3.83, "learning_rate": 2.6076096754807693e-05, "loss": 0.1024, "step": 12744 }, { "epoch": 3.83, "learning_rate": 2.607421875e-05, "loss": 0.1468, "step": 12745 }, { "epoch": 3.83, "learning_rate": 2.607234074519231e-05, "loss": 0.1865, "step": 12746 }, { "epoch": 3.83, "learning_rate": 2.6070462740384614e-05, "loss": 0.1812, "step": 12747 }, { "epoch": 3.83, "learning_rate": 2.6068584735576923e-05, "loss": 0.154, "step": 12748 }, { "epoch": 3.83, "learning_rate": 2.606670673076923e-05, "loss": 0.1433, "step": 12749 }, { "epoch": 3.83, "learning_rate": 2.606482872596154e-05, "loss": 0.1956, "step": 12750 }, { "epoch": 3.83, "learning_rate": 2.6062950721153844e-05, "loss": 0.2101, "step": 12751 }, { "epoch": 3.83, "learning_rate": 2.606107271634616e-05, "loss": 0.1049, "step": 12752 }, { "epoch": 3.83, "learning_rate": 2.6059194711538463e-05, "loss": 0.1715, "step": 12753 }, { "epoch": 3.83, "learning_rate": 2.605731670673077e-05, "loss": 0.1829, "step": 12754 }, { "epoch": 3.83, "learning_rate": 2.605543870192308e-05, "loss": 0.1179, "step": 12755 }, { "epoch": 3.83, "learning_rate": 2.6053560697115388e-05, "loss": 0.138, "step": 12756 }, { "epoch": 3.83, "learning_rate": 2.6051682692307693e-05, "loss": 0.2068, "step": 12757 }, { "epoch": 3.83, "learning_rate": 2.60498046875e-05, "loss": 0.1581, "step": 12758 }, { "epoch": 3.83, "learning_rate": 2.604792668269231e-05, "loss": 0.1105, "step": 12759 }, { "epoch": 3.83, "learning_rate": 2.6046048677884617e-05, "loss": 0.1769, "step": 12760 }, { "epoch": 3.83, "learning_rate": 2.6044170673076922e-05, "loss": 0.0977, "step": 12761 }, { "epoch": 3.83, "learning_rate": 2.604229266826923e-05, "loss": 0.1324, "step": 12762 }, { "epoch": 3.84, "learning_rate": 2.6040414663461538e-05, "loss": 0.1261, "step": 12763 }, { "epoch": 3.84, "learning_rate": 2.6038536658653846e-05, "loss": 0.1588, "step": 12764 }, { "epoch": 3.84, "learning_rate": 2.6036658653846158e-05, "loss": 0.1158, "step": 12765 }, { "epoch": 3.84, "learning_rate": 2.6034780649038466e-05, "loss": 0.1451, "step": 12766 }, { "epoch": 3.84, "learning_rate": 2.603290264423077e-05, "loss": 0.1173, "step": 12767 }, { "epoch": 3.84, "learning_rate": 2.603102463942308e-05, "loss": 0.1986, "step": 12768 }, { "epoch": 3.84, "learning_rate": 2.6029146634615387e-05, "loss": 0.1541, "step": 12769 }, { "epoch": 3.84, "learning_rate": 2.6027268629807695e-05, "loss": 0.1207, "step": 12770 }, { "epoch": 3.84, "learning_rate": 2.6025390625e-05, "loss": 0.1361, "step": 12771 }, { "epoch": 3.84, "learning_rate": 2.6023512620192308e-05, "loss": 0.2637, "step": 12772 }, { "epoch": 3.84, "learning_rate": 2.6021634615384616e-05, "loss": 0.1559, "step": 12773 }, { "epoch": 3.84, "learning_rate": 2.6019756610576925e-05, "loss": 0.1586, "step": 12774 }, { "epoch": 3.84, "learning_rate": 2.601787860576923e-05, "loss": 0.1661, "step": 12775 }, { "epoch": 3.84, "learning_rate": 2.6016000600961537e-05, "loss": 0.1278, "step": 12776 }, { "epoch": 3.84, "learning_rate": 2.6014122596153846e-05, "loss": 0.127, "step": 12777 }, { "epoch": 3.84, "learning_rate": 2.6012244591346157e-05, "loss": 0.2323, "step": 12778 }, { "epoch": 3.84, "learning_rate": 2.6010366586538465e-05, "loss": 0.1123, "step": 12779 }, { "epoch": 3.84, "learning_rate": 2.6008488581730774e-05, "loss": 0.1815, "step": 12780 }, { "epoch": 3.84, "learning_rate": 2.6006610576923078e-05, "loss": 0.1381, "step": 12781 }, { "epoch": 3.84, "learning_rate": 2.6004732572115386e-05, "loss": 0.1478, "step": 12782 }, { "epoch": 3.84, "learning_rate": 2.6002854567307695e-05, "loss": 0.1299, "step": 12783 }, { "epoch": 3.84, "learning_rate": 2.6000976562500003e-05, "loss": 0.1277, "step": 12784 }, { "epoch": 3.84, "learning_rate": 2.5999098557692308e-05, "loss": 0.1253, "step": 12785 }, { "epoch": 3.84, "learning_rate": 2.5997220552884616e-05, "loss": 0.1022, "step": 12786 }, { "epoch": 3.84, "learning_rate": 2.5995342548076924e-05, "loss": 0.1631, "step": 12787 }, { "epoch": 3.84, "learning_rate": 2.599346454326923e-05, "loss": 0.1083, "step": 12788 }, { "epoch": 3.84, "learning_rate": 2.5991586538461537e-05, "loss": 0.1108, "step": 12789 }, { "epoch": 3.84, "learning_rate": 2.5989708533653845e-05, "loss": 0.1446, "step": 12790 }, { "epoch": 3.84, "learning_rate": 2.5987830528846156e-05, "loss": 0.1297, "step": 12791 }, { "epoch": 3.84, "learning_rate": 2.5985952524038465e-05, "loss": 0.2238, "step": 12792 }, { "epoch": 3.84, "learning_rate": 2.5984074519230773e-05, "loss": 0.1275, "step": 12793 }, { "epoch": 3.84, "learning_rate": 2.598219651442308e-05, "loss": 0.2339, "step": 12794 }, { "epoch": 3.84, "learning_rate": 2.5980318509615386e-05, "loss": 0.1678, "step": 12795 }, { "epoch": 3.84, "learning_rate": 2.5978440504807694e-05, "loss": 0.1746, "step": 12796 }, { "epoch": 3.85, "learning_rate": 2.5976562500000002e-05, "loss": 0.1547, "step": 12797 }, { "epoch": 3.85, "learning_rate": 2.5974684495192307e-05, "loss": 0.1799, "step": 12798 }, { "epoch": 3.85, "learning_rate": 2.5972806490384615e-05, "loss": 0.2913, "step": 12799 }, { "epoch": 3.85, "learning_rate": 2.5970928485576923e-05, "loss": 0.1718, "step": 12800 }, { "epoch": 3.85, "learning_rate": 2.596905048076923e-05, "loss": 0.128, "step": 12801 }, { "epoch": 3.85, "learning_rate": 2.5967172475961536e-05, "loss": 0.1765, "step": 12802 }, { "epoch": 3.85, "learning_rate": 2.5965294471153844e-05, "loss": 0.1683, "step": 12803 }, { "epoch": 3.85, "learning_rate": 2.596341646634616e-05, "loss": 0.1363, "step": 12804 }, { "epoch": 3.85, "learning_rate": 2.5961538461538464e-05, "loss": 0.09, "step": 12805 }, { "epoch": 3.85, "learning_rate": 2.5959660456730772e-05, "loss": 0.1499, "step": 12806 }, { "epoch": 3.85, "learning_rate": 2.595778245192308e-05, "loss": 0.188, "step": 12807 }, { "epoch": 3.85, "learning_rate": 2.5955904447115385e-05, "loss": 0.1123, "step": 12808 }, { "epoch": 3.85, "learning_rate": 2.5954026442307693e-05, "loss": 0.1543, "step": 12809 }, { "epoch": 3.85, "learning_rate": 2.59521484375e-05, "loss": 0.1178, "step": 12810 }, { "epoch": 3.85, "learning_rate": 2.595027043269231e-05, "loss": 0.1715, "step": 12811 }, { "epoch": 3.85, "learning_rate": 2.5948392427884614e-05, "loss": 0.1444, "step": 12812 }, { "epoch": 3.85, "learning_rate": 2.5946514423076922e-05, "loss": 0.1593, "step": 12813 }, { "epoch": 3.85, "learning_rate": 2.594463641826923e-05, "loss": 0.1636, "step": 12814 }, { "epoch": 3.85, "learning_rate": 2.594275841346154e-05, "loss": 0.1239, "step": 12815 }, { "epoch": 3.85, "learning_rate": 2.5940880408653844e-05, "loss": 0.1999, "step": 12816 }, { "epoch": 3.85, "learning_rate": 2.593900240384616e-05, "loss": 0.1427, "step": 12817 }, { "epoch": 3.85, "learning_rate": 2.5937124399038463e-05, "loss": 0.1589, "step": 12818 }, { "epoch": 3.85, "learning_rate": 2.593524639423077e-05, "loss": 0.1113, "step": 12819 }, { "epoch": 3.85, "learning_rate": 2.593336838942308e-05, "loss": 0.1325, "step": 12820 }, { "epoch": 3.85, "learning_rate": 2.5931490384615388e-05, "loss": 0.1185, "step": 12821 }, { "epoch": 3.85, "learning_rate": 2.5929612379807693e-05, "loss": 0.1532, "step": 12822 }, { "epoch": 3.85, "learning_rate": 2.5927734375e-05, "loss": 0.1072, "step": 12823 }, { "epoch": 3.85, "learning_rate": 2.592585637019231e-05, "loss": 0.1521, "step": 12824 }, { "epoch": 3.85, "learning_rate": 2.5923978365384617e-05, "loss": 0.1303, "step": 12825 }, { "epoch": 3.85, "learning_rate": 2.5922100360576922e-05, "loss": 0.1306, "step": 12826 }, { "epoch": 3.85, "learning_rate": 2.592022235576923e-05, "loss": 0.099, "step": 12827 }, { "epoch": 3.85, "learning_rate": 2.5918344350961538e-05, "loss": 0.1522, "step": 12828 }, { "epoch": 3.85, "learning_rate": 2.5916466346153846e-05, "loss": 0.141, "step": 12829 }, { "epoch": 3.86, "learning_rate": 2.5914588341346158e-05, "loss": 0.2255, "step": 12830 }, { "epoch": 3.86, "learning_rate": 2.5912710336538466e-05, "loss": 0.1594, "step": 12831 }, { "epoch": 3.86, "learning_rate": 2.591083233173077e-05, "loss": 0.1231, "step": 12832 }, { "epoch": 3.86, "learning_rate": 2.590895432692308e-05, "loss": 0.154, "step": 12833 }, { "epoch": 3.86, "learning_rate": 2.5907076322115387e-05, "loss": 0.158, "step": 12834 }, { "epoch": 3.86, "learning_rate": 2.5905198317307695e-05, "loss": 0.1554, "step": 12835 }, { "epoch": 3.86, "learning_rate": 2.59033203125e-05, "loss": 0.1627, "step": 12836 }, { "epoch": 3.86, "learning_rate": 2.5901442307692308e-05, "loss": 0.1464, "step": 12837 }, { "epoch": 3.86, "learning_rate": 2.5899564302884616e-05, "loss": 0.2097, "step": 12838 }, { "epoch": 3.86, "learning_rate": 2.5897686298076924e-05, "loss": 0.1824, "step": 12839 }, { "epoch": 3.86, "learning_rate": 2.589580829326923e-05, "loss": 0.1112, "step": 12840 }, { "epoch": 3.86, "learning_rate": 2.5893930288461537e-05, "loss": 0.1859, "step": 12841 }, { "epoch": 3.86, "learning_rate": 2.5892052283653846e-05, "loss": 0.1386, "step": 12842 }, { "epoch": 3.86, "learning_rate": 2.5890174278846157e-05, "loss": 0.2081, "step": 12843 }, { "epoch": 3.86, "learning_rate": 2.5888296274038465e-05, "loss": 0.1403, "step": 12844 }, { "epoch": 3.86, "learning_rate": 2.5886418269230773e-05, "loss": 0.1402, "step": 12845 }, { "epoch": 3.86, "learning_rate": 2.5884540264423078e-05, "loss": 0.127, "step": 12846 }, { "epoch": 3.86, "learning_rate": 2.5882662259615386e-05, "loss": 0.141, "step": 12847 }, { "epoch": 3.86, "learning_rate": 2.5880784254807694e-05, "loss": 0.1615, "step": 12848 }, { "epoch": 3.86, "learning_rate": 2.5878906250000003e-05, "loss": 0.1497, "step": 12849 }, { "epoch": 3.86, "learning_rate": 2.5877028245192307e-05, "loss": 0.2103, "step": 12850 }, { "epoch": 3.86, "learning_rate": 2.5875150240384616e-05, "loss": 0.1699, "step": 12851 }, { "epoch": 3.86, "learning_rate": 2.5873272235576924e-05, "loss": 0.1695, "step": 12852 }, { "epoch": 3.86, "learning_rate": 2.5871394230769232e-05, "loss": 0.1645, "step": 12853 }, { "epoch": 3.86, "learning_rate": 2.5869516225961537e-05, "loss": 0.1257, "step": 12854 }, { "epoch": 3.86, "learning_rate": 2.5867638221153845e-05, "loss": 0.1636, "step": 12855 }, { "epoch": 3.86, "learning_rate": 2.5865760216346156e-05, "loss": 0.1544, "step": 12856 }, { "epoch": 3.86, "learning_rate": 2.5863882211538465e-05, "loss": 0.2112, "step": 12857 }, { "epoch": 3.86, "learning_rate": 2.5862004206730773e-05, "loss": 0.159, "step": 12858 }, { "epoch": 3.86, "learning_rate": 2.586012620192308e-05, "loss": 0.0959, "step": 12859 }, { "epoch": 3.86, "learning_rate": 2.5858248197115386e-05, "loss": 0.1779, "step": 12860 }, { "epoch": 3.86, "learning_rate": 2.5856370192307694e-05, "loss": 0.099, "step": 12861 }, { "epoch": 3.86, "learning_rate": 2.5854492187500002e-05, "loss": 0.1972, "step": 12862 }, { "epoch": 3.87, "learning_rate": 2.585261418269231e-05, "loss": 0.1461, "step": 12863 }, { "epoch": 3.87, "learning_rate": 2.5850736177884615e-05, "loss": 0.1373, "step": 12864 }, { "epoch": 3.87, "learning_rate": 2.5848858173076923e-05, "loss": 0.1182, "step": 12865 }, { "epoch": 3.87, "learning_rate": 2.584698016826923e-05, "loss": 0.1471, "step": 12866 }, { "epoch": 3.87, "learning_rate": 2.5845102163461536e-05, "loss": 0.1355, "step": 12867 }, { "epoch": 3.87, "learning_rate": 2.5843224158653844e-05, "loss": 0.1899, "step": 12868 }, { "epoch": 3.87, "learning_rate": 2.584134615384616e-05, "loss": 0.1293, "step": 12869 }, { "epoch": 3.87, "learning_rate": 2.5839468149038464e-05, "loss": 0.1837, "step": 12870 }, { "epoch": 3.87, "learning_rate": 2.5837590144230772e-05, "loss": 0.1213, "step": 12871 }, { "epoch": 3.87, "learning_rate": 2.583571213942308e-05, "loss": 0.1163, "step": 12872 }, { "epoch": 3.87, "learning_rate": 2.583383413461539e-05, "loss": 0.2074, "step": 12873 }, { "epoch": 3.87, "learning_rate": 2.5831956129807693e-05, "loss": 0.1164, "step": 12874 }, { "epoch": 3.87, "learning_rate": 2.5830078125e-05, "loss": 0.1406, "step": 12875 }, { "epoch": 3.87, "learning_rate": 2.582820012019231e-05, "loss": 0.1481, "step": 12876 }, { "epoch": 3.87, "learning_rate": 2.5826322115384614e-05, "loss": 0.156, "step": 12877 }, { "epoch": 3.87, "learning_rate": 2.5824444110576922e-05, "loss": 0.1129, "step": 12878 }, { "epoch": 3.87, "learning_rate": 2.582256610576923e-05, "loss": 0.153, "step": 12879 }, { "epoch": 3.87, "learning_rate": 2.582068810096154e-05, "loss": 0.0963, "step": 12880 }, { "epoch": 3.87, "learning_rate": 2.5818810096153843e-05, "loss": 0.1533, "step": 12881 }, { "epoch": 3.87, "learning_rate": 2.581693209134616e-05, "loss": 0.1902, "step": 12882 }, { "epoch": 3.87, "learning_rate": 2.5815054086538467e-05, "loss": 0.13, "step": 12883 }, { "epoch": 3.87, "learning_rate": 2.581317608173077e-05, "loss": 0.1576, "step": 12884 }, { "epoch": 3.87, "learning_rate": 2.581129807692308e-05, "loss": 0.2756, "step": 12885 }, { "epoch": 3.87, "learning_rate": 2.5809420072115388e-05, "loss": 0.078, "step": 12886 }, { "epoch": 3.87, "learning_rate": 2.5807542067307692e-05, "loss": 0.1884, "step": 12887 }, { "epoch": 3.87, "learning_rate": 2.58056640625e-05, "loss": 0.1404, "step": 12888 }, { "epoch": 3.87, "learning_rate": 2.580378605769231e-05, "loss": 0.1273, "step": 12889 }, { "epoch": 3.87, "learning_rate": 2.5801908052884617e-05, "loss": 0.1365, "step": 12890 }, { "epoch": 3.87, "learning_rate": 2.580003004807692e-05, "loss": 0.1209, "step": 12891 }, { "epoch": 3.87, "learning_rate": 2.579815204326923e-05, "loss": 0.1521, "step": 12892 }, { "epoch": 3.87, "learning_rate": 2.5796274038461538e-05, "loss": 0.1119, "step": 12893 }, { "epoch": 3.87, "learning_rate": 2.5794396033653846e-05, "loss": 0.1438, "step": 12894 }, { "epoch": 3.87, "learning_rate": 2.5792518028846158e-05, "loss": 0.1288, "step": 12895 }, { "epoch": 3.88, "learning_rate": 2.5790640024038466e-05, "loss": 0.1268, "step": 12896 }, { "epoch": 3.88, "learning_rate": 2.578876201923077e-05, "loss": 0.1599, "step": 12897 }, { "epoch": 3.88, "learning_rate": 2.578688401442308e-05, "loss": 0.1587, "step": 12898 }, { "epoch": 3.88, "learning_rate": 2.5785006009615387e-05, "loss": 0.1252, "step": 12899 }, { "epoch": 3.88, "learning_rate": 2.5783128004807695e-05, "loss": 0.1922, "step": 12900 }, { "epoch": 3.88, "learning_rate": 2.578125e-05, "loss": 0.1784, "step": 12901 }, { "epoch": 3.88, "learning_rate": 2.5779371995192308e-05, "loss": 0.1653, "step": 12902 }, { "epoch": 3.88, "learning_rate": 2.5777493990384616e-05, "loss": 0.1167, "step": 12903 }, { "epoch": 3.88, "learning_rate": 2.5775615985576924e-05, "loss": 0.1229, "step": 12904 }, { "epoch": 3.88, "learning_rate": 2.577373798076923e-05, "loss": 0.2054, "step": 12905 }, { "epoch": 3.88, "learning_rate": 2.5771859975961537e-05, "loss": 0.1235, "step": 12906 }, { "epoch": 3.88, "learning_rate": 2.5769981971153845e-05, "loss": 0.1198, "step": 12907 }, { "epoch": 3.88, "learning_rate": 2.5768103966346157e-05, "loss": 0.1236, "step": 12908 }, { "epoch": 3.88, "learning_rate": 2.5766225961538465e-05, "loss": 0.156, "step": 12909 }, { "epoch": 3.88, "learning_rate": 2.5764347956730773e-05, "loss": 0.2584, "step": 12910 }, { "epoch": 3.88, "learning_rate": 2.5762469951923078e-05, "loss": 0.1264, "step": 12911 }, { "epoch": 3.88, "learning_rate": 2.5760591947115386e-05, "loss": 0.1702, "step": 12912 }, { "epoch": 3.88, "learning_rate": 2.5758713942307694e-05, "loss": 0.1925, "step": 12913 }, { "epoch": 3.88, "learning_rate": 2.5756835937500003e-05, "loss": 0.1523, "step": 12914 }, { "epoch": 3.88, "learning_rate": 2.5754957932692307e-05, "loss": 0.1238, "step": 12915 }, { "epoch": 3.88, "learning_rate": 2.5753079927884615e-05, "loss": 0.1351, "step": 12916 }, { "epoch": 3.88, "learning_rate": 2.5751201923076924e-05, "loss": 0.1537, "step": 12917 }, { "epoch": 3.88, "learning_rate": 2.5749323918269232e-05, "loss": 0.2365, "step": 12918 }, { "epoch": 3.88, "learning_rate": 2.5747445913461537e-05, "loss": 0.1052, "step": 12919 }, { "epoch": 3.88, "learning_rate": 2.5745567908653845e-05, "loss": 0.1496, "step": 12920 }, { "epoch": 3.88, "learning_rate": 2.5743689903846156e-05, "loss": 0.1383, "step": 12921 }, { "epoch": 3.88, "learning_rate": 2.5741811899038464e-05, "loss": 0.1583, "step": 12922 }, { "epoch": 3.88, "learning_rate": 2.5739933894230773e-05, "loss": 0.1302, "step": 12923 }, { "epoch": 3.88, "learning_rate": 2.573805588942308e-05, "loss": 0.1292, "step": 12924 }, { "epoch": 3.88, "learning_rate": 2.5736177884615386e-05, "loss": 0.1957, "step": 12925 }, { "epoch": 3.88, "learning_rate": 2.5734299879807694e-05, "loss": 0.1264, "step": 12926 }, { "epoch": 3.88, "learning_rate": 2.5732421875000002e-05, "loss": 0.1258, "step": 12927 }, { "epoch": 3.88, "learning_rate": 2.573054387019231e-05, "loss": 0.1126, "step": 12928 }, { "epoch": 3.88, "learning_rate": 2.5728665865384615e-05, "loss": 0.106, "step": 12929 }, { "epoch": 3.89, "learning_rate": 2.5726787860576923e-05, "loss": 0.1341, "step": 12930 }, { "epoch": 3.89, "learning_rate": 2.572490985576923e-05, "loss": 0.1227, "step": 12931 }, { "epoch": 3.89, "learning_rate": 2.5723031850961536e-05, "loss": 0.2005, "step": 12932 }, { "epoch": 3.89, "learning_rate": 2.5721153846153844e-05, "loss": 0.1405, "step": 12933 }, { "epoch": 3.89, "learning_rate": 2.571927584134616e-05, "loss": 0.1616, "step": 12934 }, { "epoch": 3.89, "learning_rate": 2.5717397836538464e-05, "loss": 0.1452, "step": 12935 }, { "epoch": 3.89, "learning_rate": 2.5715519831730772e-05, "loss": 0.1218, "step": 12936 }, { "epoch": 3.89, "learning_rate": 2.571364182692308e-05, "loss": 0.1428, "step": 12937 }, { "epoch": 3.89, "learning_rate": 2.5711763822115388e-05, "loss": 0.1058, "step": 12938 }, { "epoch": 3.89, "learning_rate": 2.5709885817307693e-05, "loss": 0.214, "step": 12939 }, { "epoch": 3.89, "learning_rate": 2.57080078125e-05, "loss": 0.1273, "step": 12940 }, { "epoch": 3.89, "learning_rate": 2.570612980769231e-05, "loss": 0.1032, "step": 12941 }, { "epoch": 3.89, "learning_rate": 2.5704251802884614e-05, "loss": 0.1764, "step": 12942 }, { "epoch": 3.89, "learning_rate": 2.5702373798076922e-05, "loss": 0.1322, "step": 12943 }, { "epoch": 3.89, "learning_rate": 2.570049579326923e-05, "loss": 0.1561, "step": 12944 }, { "epoch": 3.89, "learning_rate": 2.569861778846154e-05, "loss": 0.1149, "step": 12945 }, { "epoch": 3.89, "learning_rate": 2.5696739783653843e-05, "loss": 0.1627, "step": 12946 }, { "epoch": 3.89, "learning_rate": 2.5694861778846158e-05, "loss": 0.157, "step": 12947 }, { "epoch": 3.89, "learning_rate": 2.5692983774038466e-05, "loss": 0.0951, "step": 12948 }, { "epoch": 3.89, "learning_rate": 2.569110576923077e-05, "loss": 0.1328, "step": 12949 }, { "epoch": 3.89, "learning_rate": 2.568922776442308e-05, "loss": 0.1214, "step": 12950 }, { "epoch": 3.89, "learning_rate": 2.5687349759615388e-05, "loss": 0.1982, "step": 12951 }, { "epoch": 3.89, "learning_rate": 2.5685471754807692e-05, "loss": 0.1875, "step": 12952 }, { "epoch": 3.89, "learning_rate": 2.568359375e-05, "loss": 0.1899, "step": 12953 }, { "epoch": 3.89, "learning_rate": 2.568171574519231e-05, "loss": 0.1158, "step": 12954 }, { "epoch": 3.89, "learning_rate": 2.5679837740384617e-05, "loss": 0.1355, "step": 12955 }, { "epoch": 3.89, "learning_rate": 2.567795973557692e-05, "loss": 0.0939, "step": 12956 }, { "epoch": 3.89, "learning_rate": 2.567608173076923e-05, "loss": 0.1491, "step": 12957 }, { "epoch": 3.89, "learning_rate": 2.5674203725961538e-05, "loss": 0.1241, "step": 12958 }, { "epoch": 3.89, "learning_rate": 2.5672325721153846e-05, "loss": 0.1627, "step": 12959 }, { "epoch": 3.89, "learning_rate": 2.5670447716346158e-05, "loss": 0.1264, "step": 12960 }, { "epoch": 3.89, "learning_rate": 2.5668569711538466e-05, "loss": 0.2135, "step": 12961 }, { "epoch": 3.89, "learning_rate": 2.566669170673077e-05, "loss": 0.1755, "step": 12962 }, { "epoch": 3.9, "learning_rate": 2.566481370192308e-05, "loss": 0.1576, "step": 12963 }, { "epoch": 3.9, "learning_rate": 2.5662935697115387e-05, "loss": 0.199, "step": 12964 }, { "epoch": 3.9, "learning_rate": 2.5661057692307695e-05, "loss": 0.1575, "step": 12965 }, { "epoch": 3.9, "learning_rate": 2.56591796875e-05, "loss": 0.2038, "step": 12966 }, { "epoch": 3.9, "learning_rate": 2.5657301682692308e-05, "loss": 0.2051, "step": 12967 }, { "epoch": 3.9, "learning_rate": 2.5655423677884616e-05, "loss": 0.1267, "step": 12968 }, { "epoch": 3.9, "learning_rate": 2.5653545673076924e-05, "loss": 0.165, "step": 12969 }, { "epoch": 3.9, "learning_rate": 2.565166766826923e-05, "loss": 0.163, "step": 12970 }, { "epoch": 3.9, "learning_rate": 2.5649789663461537e-05, "loss": 0.1478, "step": 12971 }, { "epoch": 3.9, "learning_rate": 2.5647911658653845e-05, "loss": 0.1462, "step": 12972 }, { "epoch": 3.9, "learning_rate": 2.5646033653846157e-05, "loss": 0.1613, "step": 12973 }, { "epoch": 3.9, "learning_rate": 2.5644155649038465e-05, "loss": 0.1814, "step": 12974 }, { "epoch": 3.9, "learning_rate": 2.5642277644230773e-05, "loss": 0.219, "step": 12975 }, { "epoch": 3.9, "learning_rate": 2.5640399639423078e-05, "loss": 0.1481, "step": 12976 }, { "epoch": 3.9, "learning_rate": 2.5638521634615386e-05, "loss": 0.1123, "step": 12977 }, { "epoch": 3.9, "learning_rate": 2.5636643629807694e-05, "loss": 0.1907, "step": 12978 }, { "epoch": 3.9, "learning_rate": 2.5634765625000002e-05, "loss": 0.1227, "step": 12979 }, { "epoch": 3.9, "learning_rate": 2.5632887620192307e-05, "loss": 0.2133, "step": 12980 }, { "epoch": 3.9, "learning_rate": 2.5631009615384615e-05, "loss": 0.1657, "step": 12981 }, { "epoch": 3.9, "learning_rate": 2.5629131610576924e-05, "loss": 0.2186, "step": 12982 }, { "epoch": 3.9, "learning_rate": 2.562725360576923e-05, "loss": 0.1842, "step": 12983 }, { "epoch": 3.9, "learning_rate": 2.5625375600961536e-05, "loss": 0.1784, "step": 12984 }, { "epoch": 3.9, "learning_rate": 2.5623497596153845e-05, "loss": 0.2363, "step": 12985 }, { "epoch": 3.9, "learning_rate": 2.5621619591346156e-05, "loss": 0.1666, "step": 12986 }, { "epoch": 3.9, "learning_rate": 2.5619741586538464e-05, "loss": 0.1691, "step": 12987 }, { "epoch": 3.9, "learning_rate": 2.5617863581730772e-05, "loss": 0.1872, "step": 12988 }, { "epoch": 3.9, "learning_rate": 2.561598557692308e-05, "loss": 0.1808, "step": 12989 }, { "epoch": 3.9, "learning_rate": 2.5614107572115385e-05, "loss": 0.1569, "step": 12990 }, { "epoch": 3.9, "learning_rate": 2.5612229567307694e-05, "loss": 0.1476, "step": 12991 }, { "epoch": 3.9, "learning_rate": 2.5610351562500002e-05, "loss": 0.3272, "step": 12992 }, { "epoch": 3.9, "learning_rate": 2.560847355769231e-05, "loss": 0.1317, "step": 12993 }, { "epoch": 3.9, "learning_rate": 2.5606595552884615e-05, "loss": 0.1663, "step": 12994 }, { "epoch": 3.9, "learning_rate": 2.5604717548076923e-05, "loss": 0.1902, "step": 12995 }, { "epoch": 3.91, "learning_rate": 2.560283954326923e-05, "loss": 0.1301, "step": 12996 }, { "epoch": 3.91, "learning_rate": 2.560096153846154e-05, "loss": 0.1592, "step": 12997 }, { "epoch": 3.91, "learning_rate": 2.5599083533653844e-05, "loss": 0.1257, "step": 12998 }, { "epoch": 3.91, "learning_rate": 2.559720552884616e-05, "loss": 0.1416, "step": 12999 }, { "epoch": 3.91, "learning_rate": 2.5595327524038464e-05, "loss": 0.156, "step": 13000 }, { "epoch": 3.91, "learning_rate": 2.5593449519230772e-05, "loss": 0.1795, "step": 13001 }, { "epoch": 3.91, "learning_rate": 2.559157151442308e-05, "loss": 0.0847, "step": 13002 }, { "epoch": 3.91, "learning_rate": 2.5589693509615388e-05, "loss": 0.1394, "step": 13003 }, { "epoch": 3.91, "learning_rate": 2.5587815504807693e-05, "loss": 0.1745, "step": 13004 }, { "epoch": 3.91, "learning_rate": 2.55859375e-05, "loss": 0.1815, "step": 13005 }, { "epoch": 3.91, "learning_rate": 2.558405949519231e-05, "loss": 0.145, "step": 13006 }, { "epoch": 3.91, "learning_rate": 2.5582181490384617e-05, "loss": 0.238, "step": 13007 }, { "epoch": 3.91, "learning_rate": 2.5580303485576922e-05, "loss": 0.1349, "step": 13008 }, { "epoch": 3.91, "learning_rate": 2.557842548076923e-05, "loss": 0.0939, "step": 13009 }, { "epoch": 3.91, "learning_rate": 2.557654747596154e-05, "loss": 0.2053, "step": 13010 }, { "epoch": 3.91, "learning_rate": 2.5574669471153843e-05, "loss": 0.1619, "step": 13011 }, { "epoch": 3.91, "learning_rate": 2.5572791466346158e-05, "loss": 0.1368, "step": 13012 }, { "epoch": 3.91, "learning_rate": 2.5570913461538466e-05, "loss": 0.1329, "step": 13013 }, { "epoch": 3.91, "learning_rate": 2.556903545673077e-05, "loss": 0.1978, "step": 13014 }, { "epoch": 3.91, "learning_rate": 2.556715745192308e-05, "loss": 0.1286, "step": 13015 }, { "epoch": 3.91, "learning_rate": 2.5565279447115387e-05, "loss": 0.0977, "step": 13016 }, { "epoch": 3.91, "learning_rate": 2.5563401442307696e-05, "loss": 0.1589, "step": 13017 }, { "epoch": 3.91, "learning_rate": 2.55615234375e-05, "loss": 0.1344, "step": 13018 }, { "epoch": 3.91, "learning_rate": 2.555964543269231e-05, "loss": 0.1072, "step": 13019 }, { "epoch": 3.91, "learning_rate": 2.5557767427884617e-05, "loss": 0.147, "step": 13020 }, { "epoch": 3.91, "learning_rate": 2.555588942307692e-05, "loss": 0.1112, "step": 13021 }, { "epoch": 3.91, "learning_rate": 2.555401141826923e-05, "loss": 0.1628, "step": 13022 }, { "epoch": 3.91, "learning_rate": 2.5552133413461538e-05, "loss": 0.1579, "step": 13023 }, { "epoch": 3.91, "learning_rate": 2.5550255408653846e-05, "loss": 0.1954, "step": 13024 }, { "epoch": 3.91, "learning_rate": 2.5548377403846157e-05, "loss": 0.1367, "step": 13025 }, { "epoch": 3.91, "learning_rate": 2.5546499399038466e-05, "loss": 0.1668, "step": 13026 }, { "epoch": 3.91, "learning_rate": 2.5544621394230774e-05, "loss": 0.1675, "step": 13027 }, { "epoch": 3.91, "learning_rate": 2.554274338942308e-05, "loss": 0.1595, "step": 13028 }, { "epoch": 3.91, "learning_rate": 2.5540865384615387e-05, "loss": 0.128, "step": 13029 }, { "epoch": 3.92, "learning_rate": 2.5538987379807695e-05, "loss": 0.2388, "step": 13030 }, { "epoch": 3.92, "learning_rate": 2.5537109375e-05, "loss": 0.1038, "step": 13031 }, { "epoch": 3.92, "learning_rate": 2.5535231370192308e-05, "loss": 0.1432, "step": 13032 }, { "epoch": 3.92, "learning_rate": 2.5533353365384616e-05, "loss": 0.1276, "step": 13033 }, { "epoch": 3.92, "learning_rate": 2.5531475360576924e-05, "loss": 0.0986, "step": 13034 }, { "epoch": 3.92, "learning_rate": 2.552959735576923e-05, "loss": 0.1179, "step": 13035 }, { "epoch": 3.92, "learning_rate": 2.5527719350961537e-05, "loss": 0.1629, "step": 13036 }, { "epoch": 3.92, "learning_rate": 2.5525841346153845e-05, "loss": 0.1315, "step": 13037 }, { "epoch": 3.92, "learning_rate": 2.5523963341346157e-05, "loss": 0.1827, "step": 13038 }, { "epoch": 3.92, "learning_rate": 2.5522085336538465e-05, "loss": 0.1131, "step": 13039 }, { "epoch": 3.92, "learning_rate": 2.5520207331730773e-05, "loss": 0.1531, "step": 13040 }, { "epoch": 3.92, "learning_rate": 2.5518329326923078e-05, "loss": 0.1352, "step": 13041 }, { "epoch": 3.92, "learning_rate": 2.5516451322115386e-05, "loss": 0.1592, "step": 13042 }, { "epoch": 3.92, "learning_rate": 2.5514573317307694e-05, "loss": 0.1598, "step": 13043 }, { "epoch": 3.92, "learning_rate": 2.5512695312500002e-05, "loss": 0.2373, "step": 13044 }, { "epoch": 3.92, "learning_rate": 2.5510817307692307e-05, "loss": 0.1122, "step": 13045 }, { "epoch": 3.92, "learning_rate": 2.5508939302884615e-05, "loss": 0.1322, "step": 13046 }, { "epoch": 3.92, "learning_rate": 2.5507061298076923e-05, "loss": 0.0787, "step": 13047 }, { "epoch": 3.92, "learning_rate": 2.550518329326923e-05, "loss": 0.1392, "step": 13048 }, { "epoch": 3.92, "learning_rate": 2.5503305288461536e-05, "loss": 0.1105, "step": 13049 }, { "epoch": 3.92, "learning_rate": 2.5501427283653845e-05, "loss": 0.1125, "step": 13050 }, { "epoch": 3.92, "learning_rate": 2.5499549278846156e-05, "loss": 0.1388, "step": 13051 }, { "epoch": 3.92, "learning_rate": 2.5497671274038464e-05, "loss": 0.1532, "step": 13052 }, { "epoch": 3.92, "learning_rate": 2.5495793269230772e-05, "loss": 0.1954, "step": 13053 }, { "epoch": 3.92, "learning_rate": 2.549391526442308e-05, "loss": 0.1353, "step": 13054 }, { "epoch": 3.92, "learning_rate": 2.5492037259615385e-05, "loss": 0.1326, "step": 13055 }, { "epoch": 3.92, "learning_rate": 2.5490159254807693e-05, "loss": 0.1686, "step": 13056 }, { "epoch": 3.92, "learning_rate": 2.548828125e-05, "loss": 0.1933, "step": 13057 }, { "epoch": 3.92, "learning_rate": 2.548640324519231e-05, "loss": 0.1986, "step": 13058 }, { "epoch": 3.92, "learning_rate": 2.5484525240384615e-05, "loss": 0.1446, "step": 13059 }, { "epoch": 3.92, "learning_rate": 2.5482647235576923e-05, "loss": 0.1738, "step": 13060 }, { "epoch": 3.92, "learning_rate": 2.548076923076923e-05, "loss": 0.1079, "step": 13061 }, { "epoch": 3.92, "learning_rate": 2.547889122596154e-05, "loss": 0.2152, "step": 13062 }, { "epoch": 3.93, "learning_rate": 2.5477013221153844e-05, "loss": 0.1227, "step": 13063 }, { "epoch": 3.93, "learning_rate": 2.547513521634616e-05, "loss": 0.1032, "step": 13064 }, { "epoch": 3.93, "learning_rate": 2.5473257211538464e-05, "loss": 0.2256, "step": 13065 }, { "epoch": 3.93, "learning_rate": 2.547137920673077e-05, "loss": 0.1224, "step": 13066 }, { "epoch": 3.93, "learning_rate": 2.546950120192308e-05, "loss": 0.1305, "step": 13067 }, { "epoch": 3.93, "learning_rate": 2.5467623197115388e-05, "loss": 0.1533, "step": 13068 }, { "epoch": 3.93, "learning_rate": 2.5465745192307693e-05, "loss": 0.0963, "step": 13069 }, { "epoch": 3.93, "learning_rate": 2.54638671875e-05, "loss": 0.1592, "step": 13070 }, { "epoch": 3.93, "learning_rate": 2.546198918269231e-05, "loss": 0.1329, "step": 13071 }, { "epoch": 3.93, "learning_rate": 2.5460111177884617e-05, "loss": 0.1732, "step": 13072 }, { "epoch": 3.93, "learning_rate": 2.5458233173076922e-05, "loss": 0.1353, "step": 13073 }, { "epoch": 3.93, "learning_rate": 2.545635516826923e-05, "loss": 0.155, "step": 13074 }, { "epoch": 3.93, "learning_rate": 2.545447716346154e-05, "loss": 0.1781, "step": 13075 }, { "epoch": 3.93, "learning_rate": 2.5452599158653847e-05, "loss": 0.1548, "step": 13076 }, { "epoch": 3.93, "learning_rate": 2.5450721153846158e-05, "loss": 0.1678, "step": 13077 }, { "epoch": 3.93, "learning_rate": 2.5448843149038466e-05, "loss": 0.1105, "step": 13078 }, { "epoch": 3.93, "learning_rate": 2.544696514423077e-05, "loss": 0.1845, "step": 13079 }, { "epoch": 3.93, "learning_rate": 2.544696514423077e-05, "loss": 0.1473, "step": 13080 }, { "epoch": 3.93, "learning_rate": 2.544508713942308e-05, "loss": 0.1359, "step": 13081 }, { "epoch": 3.93, "learning_rate": 2.5443209134615387e-05, "loss": 0.1502, "step": 13082 }, { "epoch": 3.93, "learning_rate": 2.5441331129807695e-05, "loss": 0.1633, "step": 13083 }, { "epoch": 3.93, "learning_rate": 2.5439453125e-05, "loss": 0.0807, "step": 13084 }, { "epoch": 3.93, "learning_rate": 2.543757512019231e-05, "loss": 0.1341, "step": 13085 }, { "epoch": 3.93, "learning_rate": 2.5435697115384617e-05, "loss": 0.1959, "step": 13086 }, { "epoch": 3.93, "learning_rate": 2.543381911057692e-05, "loss": 0.2085, "step": 13087 }, { "epoch": 3.93, "learning_rate": 2.543194110576923e-05, "loss": 0.1683, "step": 13088 }, { "epoch": 3.93, "learning_rate": 2.5430063100961538e-05, "loss": 0.1486, "step": 13089 }, { "epoch": 3.93, "learning_rate": 2.5428185096153846e-05, "loss": 0.1246, "step": 13090 }, { "epoch": 3.93, "learning_rate": 2.5426307091346157e-05, "loss": 0.0861, "step": 13091 }, { "epoch": 3.93, "learning_rate": 2.5424429086538466e-05, "loss": 0.1716, "step": 13092 }, { "epoch": 3.93, "learning_rate": 2.5422551081730774e-05, "loss": 0.1359, "step": 13093 }, { "epoch": 3.93, "learning_rate": 2.542067307692308e-05, "loss": 0.1033, "step": 13094 }, { "epoch": 3.93, "learning_rate": 2.5418795072115387e-05, "loss": 0.1436, "step": 13095 }, { "epoch": 3.94, "learning_rate": 2.5416917067307695e-05, "loss": 0.1638, "step": 13096 }, { "epoch": 3.94, "learning_rate": 2.54150390625e-05, "loss": 0.1307, "step": 13097 }, { "epoch": 3.94, "learning_rate": 2.5413161057692308e-05, "loss": 0.1006, "step": 13098 }, { "epoch": 3.94, "learning_rate": 2.5411283052884616e-05, "loss": 0.1641, "step": 13099 }, { "epoch": 3.94, "learning_rate": 2.5409405048076924e-05, "loss": 0.1705, "step": 13100 }, { "epoch": 3.94, "learning_rate": 2.540752704326923e-05, "loss": 0.1119, "step": 13101 }, { "epoch": 3.94, "learning_rate": 2.5405649038461537e-05, "loss": 0.1806, "step": 13102 }, { "epoch": 3.94, "learning_rate": 2.5403771033653845e-05, "loss": 0.1444, "step": 13103 }, { "epoch": 3.94, "learning_rate": 2.5401893028846157e-05, "loss": 0.1537, "step": 13104 }, { "epoch": 3.94, "learning_rate": 2.5400015024038465e-05, "loss": 0.1493, "step": 13105 }, { "epoch": 3.94, "learning_rate": 2.5398137019230773e-05, "loss": 0.0975, "step": 13106 }, { "epoch": 3.94, "learning_rate": 2.5396259014423078e-05, "loss": 0.1573, "step": 13107 }, { "epoch": 3.94, "learning_rate": 2.5394381009615386e-05, "loss": 0.1268, "step": 13108 }, { "epoch": 3.94, "learning_rate": 2.5392503004807694e-05, "loss": 0.1959, "step": 13109 }, { "epoch": 3.94, "learning_rate": 2.5390625000000002e-05, "loss": 0.1566, "step": 13110 }, { "epoch": 3.94, "learning_rate": 2.5388746995192307e-05, "loss": 0.2025, "step": 13111 }, { "epoch": 3.94, "learning_rate": 2.5386868990384615e-05, "loss": 0.1215, "step": 13112 }, { "epoch": 3.94, "learning_rate": 2.5384990985576923e-05, "loss": 0.1454, "step": 13113 }, { "epoch": 3.94, "learning_rate": 2.538311298076923e-05, "loss": 0.1187, "step": 13114 }, { "epoch": 3.94, "learning_rate": 2.5381234975961536e-05, "loss": 0.1603, "step": 13115 }, { "epoch": 3.94, "learning_rate": 2.5379356971153844e-05, "loss": 0.1333, "step": 13116 }, { "epoch": 3.94, "learning_rate": 2.5377478966346156e-05, "loss": 0.1165, "step": 13117 }, { "epoch": 3.94, "learning_rate": 2.5375600961538464e-05, "loss": 0.1439, "step": 13118 }, { "epoch": 3.94, "learning_rate": 2.5373722956730772e-05, "loss": 0.1158, "step": 13119 }, { "epoch": 3.94, "learning_rate": 2.537184495192308e-05, "loss": 0.1378, "step": 13120 }, { "epoch": 3.94, "learning_rate": 2.5369966947115385e-05, "loss": 0.2142, "step": 13121 }, { "epoch": 3.94, "learning_rate": 2.5368088942307693e-05, "loss": 0.2091, "step": 13122 }, { "epoch": 3.94, "learning_rate": 2.53662109375e-05, "loss": 0.1671, "step": 13123 }, { "epoch": 3.94, "learning_rate": 2.536433293269231e-05, "loss": 0.1805, "step": 13124 }, { "epoch": 3.94, "learning_rate": 2.5362454927884614e-05, "loss": 0.1733, "step": 13125 }, { "epoch": 3.94, "learning_rate": 2.5360576923076923e-05, "loss": 0.1919, "step": 13126 }, { "epoch": 3.94, "learning_rate": 2.535869891826923e-05, "loss": 0.1949, "step": 13127 }, { "epoch": 3.94, "learning_rate": 2.535682091346154e-05, "loss": 0.1612, "step": 13128 }, { "epoch": 3.95, "learning_rate": 2.5354942908653844e-05, "loss": 0.1325, "step": 13129 }, { "epoch": 3.95, "learning_rate": 2.535306490384616e-05, "loss": 0.3141, "step": 13130 }, { "epoch": 3.95, "learning_rate": 2.5351186899038463e-05, "loss": 0.1153, "step": 13131 }, { "epoch": 3.95, "learning_rate": 2.534930889423077e-05, "loss": 0.1902, "step": 13132 }, { "epoch": 3.95, "learning_rate": 2.534743088942308e-05, "loss": 0.1436, "step": 13133 }, { "epoch": 3.95, "learning_rate": 2.5345552884615388e-05, "loss": 0.1252, "step": 13134 }, { "epoch": 3.95, "learning_rate": 2.5343674879807693e-05, "loss": 0.1862, "step": 13135 }, { "epoch": 3.95, "learning_rate": 2.5341796875e-05, "loss": 0.0927, "step": 13136 }, { "epoch": 3.95, "learning_rate": 2.533991887019231e-05, "loss": 0.1151, "step": 13137 }, { "epoch": 3.95, "learning_rate": 2.5338040865384617e-05, "loss": 0.1398, "step": 13138 }, { "epoch": 3.95, "learning_rate": 2.5336162860576922e-05, "loss": 0.2, "step": 13139 }, { "epoch": 3.95, "learning_rate": 2.533428485576923e-05, "loss": 0.1143, "step": 13140 }, { "epoch": 3.95, "learning_rate": 2.5332406850961538e-05, "loss": 0.1153, "step": 13141 }, { "epoch": 3.95, "learning_rate": 2.5330528846153846e-05, "loss": 0.0945, "step": 13142 }, { "epoch": 3.95, "learning_rate": 2.5328650841346158e-05, "loss": 0.145, "step": 13143 }, { "epoch": 3.95, "learning_rate": 2.5326772836538466e-05, "loss": 0.2124, "step": 13144 }, { "epoch": 3.95, "learning_rate": 2.532489483173077e-05, "loss": 0.1932, "step": 13145 }, { "epoch": 3.95, "learning_rate": 2.532301682692308e-05, "loss": 0.2171, "step": 13146 }, { "epoch": 3.95, "learning_rate": 2.5321138822115387e-05, "loss": 0.1159, "step": 13147 }, { "epoch": 3.95, "learning_rate": 2.5319260817307695e-05, "loss": 0.1346, "step": 13148 }, { "epoch": 3.95, "learning_rate": 2.53173828125e-05, "loss": 0.1551, "step": 13149 }, { "epoch": 3.95, "learning_rate": 2.5315504807692308e-05, "loss": 0.1249, "step": 13150 }, { "epoch": 3.95, "learning_rate": 2.5313626802884616e-05, "loss": 0.1598, "step": 13151 }, { "epoch": 3.95, "learning_rate": 2.5311748798076925e-05, "loss": 0.1141, "step": 13152 }, { "epoch": 3.95, "learning_rate": 2.530987079326923e-05, "loss": 0.1842, "step": 13153 }, { "epoch": 3.95, "learning_rate": 2.5307992788461538e-05, "loss": 0.1905, "step": 13154 }, { "epoch": 3.95, "learning_rate": 2.5306114783653846e-05, "loss": 0.1664, "step": 13155 }, { "epoch": 3.95, "learning_rate": 2.5304236778846157e-05, "loss": 0.1688, "step": 13156 }, { "epoch": 3.95, "learning_rate": 2.5302358774038465e-05, "loss": 0.0857, "step": 13157 }, { "epoch": 3.95, "learning_rate": 2.5300480769230774e-05, "loss": 0.0963, "step": 13158 }, { "epoch": 3.95, "learning_rate": 2.529860276442308e-05, "loss": 0.092, "step": 13159 }, { "epoch": 3.95, "learning_rate": 2.5296724759615386e-05, "loss": 0.1661, "step": 13160 }, { "epoch": 3.95, "learning_rate": 2.5294846754807695e-05, "loss": 0.2499, "step": 13161 }, { "epoch": 3.95, "learning_rate": 2.5292968750000003e-05, "loss": 0.1173, "step": 13162 }, { "epoch": 3.96, "learning_rate": 2.5291090745192308e-05, "loss": 0.1244, "step": 13163 }, { "epoch": 3.96, "learning_rate": 2.5289212740384616e-05, "loss": 0.1961, "step": 13164 }, { "epoch": 3.96, "learning_rate": 2.5287334735576924e-05, "loss": 0.1782, "step": 13165 }, { "epoch": 3.96, "learning_rate": 2.528545673076923e-05, "loss": 0.143, "step": 13166 }, { "epoch": 3.96, "learning_rate": 2.5283578725961537e-05, "loss": 0.0915, "step": 13167 }, { "epoch": 3.96, "learning_rate": 2.5281700721153845e-05, "loss": 0.1385, "step": 13168 }, { "epoch": 3.96, "learning_rate": 2.5279822716346157e-05, "loss": 0.2562, "step": 13169 }, { "epoch": 3.96, "learning_rate": 2.5277944711538465e-05, "loss": 0.1457, "step": 13170 }, { "epoch": 3.96, "learning_rate": 2.5276066706730773e-05, "loss": 0.1465, "step": 13171 }, { "epoch": 3.96, "learning_rate": 2.527418870192308e-05, "loss": 0.144, "step": 13172 }, { "epoch": 3.96, "learning_rate": 2.5272310697115386e-05, "loss": 0.1559, "step": 13173 }, { "epoch": 3.96, "learning_rate": 2.5270432692307694e-05, "loss": 0.1289, "step": 13174 }, { "epoch": 3.96, "learning_rate": 2.5268554687500002e-05, "loss": 0.1354, "step": 13175 }, { "epoch": 3.96, "learning_rate": 2.5266676682692307e-05, "loss": 0.1826, "step": 13176 }, { "epoch": 3.96, "learning_rate": 2.5264798677884615e-05, "loss": 0.1081, "step": 13177 }, { "epoch": 3.96, "learning_rate": 2.5262920673076923e-05, "loss": 0.1269, "step": 13178 }, { "epoch": 3.96, "learning_rate": 2.526104266826923e-05, "loss": 0.0882, "step": 13179 }, { "epoch": 3.96, "learning_rate": 2.5259164663461536e-05, "loss": 0.1778, "step": 13180 }, { "epoch": 3.96, "learning_rate": 2.5257286658653844e-05, "loss": 0.135, "step": 13181 }, { "epoch": 3.96, "learning_rate": 2.525540865384616e-05, "loss": 0.1207, "step": 13182 }, { "epoch": 3.96, "learning_rate": 2.5253530649038464e-05, "loss": 0.1636, "step": 13183 }, { "epoch": 3.96, "learning_rate": 2.5251652644230772e-05, "loss": 0.1628, "step": 13184 }, { "epoch": 3.96, "learning_rate": 2.524977463942308e-05, "loss": 0.115, "step": 13185 }, { "epoch": 3.96, "learning_rate": 2.5247896634615385e-05, "loss": 0.1683, "step": 13186 }, { "epoch": 3.96, "learning_rate": 2.5246018629807693e-05, "loss": 0.199, "step": 13187 }, { "epoch": 3.96, "learning_rate": 2.5244140625e-05, "loss": 0.1547, "step": 13188 }, { "epoch": 3.96, "learning_rate": 2.524226262019231e-05, "loss": 0.1798, "step": 13189 }, { "epoch": 3.96, "learning_rate": 2.5240384615384614e-05, "loss": 0.1643, "step": 13190 }, { "epoch": 3.96, "learning_rate": 2.5238506610576923e-05, "loss": 0.1545, "step": 13191 }, { "epoch": 3.96, "learning_rate": 2.523662860576923e-05, "loss": 0.153, "step": 13192 }, { "epoch": 3.96, "learning_rate": 2.523475060096154e-05, "loss": 0.2032, "step": 13193 }, { "epoch": 3.96, "learning_rate": 2.5232872596153844e-05, "loss": 0.2317, "step": 13194 }, { "epoch": 3.96, "learning_rate": 2.523099459134616e-05, "loss": 0.1309, "step": 13195 }, { "epoch": 3.97, "learning_rate": 2.5229116586538463e-05, "loss": 0.1243, "step": 13196 }, { "epoch": 3.97, "learning_rate": 2.522723858173077e-05, "loss": 0.1388, "step": 13197 }, { "epoch": 3.97, "learning_rate": 2.522536057692308e-05, "loss": 0.1613, "step": 13198 }, { "epoch": 3.97, "learning_rate": 2.5223482572115388e-05, "loss": 0.1466, "step": 13199 }, { "epoch": 3.97, "learning_rate": 2.5221604567307693e-05, "loss": 0.215, "step": 13200 }, { "epoch": 3.97, "learning_rate": 2.52197265625e-05, "loss": 0.1958, "step": 13201 }, { "epoch": 3.97, "learning_rate": 2.521784855769231e-05, "loss": 0.1211, "step": 13202 }, { "epoch": 3.97, "learning_rate": 2.5215970552884617e-05, "loss": 0.2113, "step": 13203 }, { "epoch": 3.97, "learning_rate": 2.5214092548076922e-05, "loss": 0.1105, "step": 13204 }, { "epoch": 3.97, "learning_rate": 2.521221454326923e-05, "loss": 0.148, "step": 13205 }, { "epoch": 3.97, "learning_rate": 2.5210336538461538e-05, "loss": 0.1299, "step": 13206 }, { "epoch": 3.97, "learning_rate": 2.5208458533653846e-05, "loss": 0.0976, "step": 13207 }, { "epoch": 3.97, "learning_rate": 2.5206580528846158e-05, "loss": 0.1633, "step": 13208 }, { "epoch": 3.97, "learning_rate": 2.5204702524038466e-05, "loss": 0.1891, "step": 13209 }, { "epoch": 3.97, "learning_rate": 2.520282451923077e-05, "loss": 0.107, "step": 13210 }, { "epoch": 3.97, "learning_rate": 2.520094651442308e-05, "loss": 0.1585, "step": 13211 }, { "epoch": 3.97, "learning_rate": 2.5199068509615387e-05, "loss": 0.1139, "step": 13212 }, { "epoch": 3.97, "learning_rate": 2.5197190504807695e-05, "loss": 0.0872, "step": 13213 }, { "epoch": 3.97, "learning_rate": 2.51953125e-05, "loss": 0.2128, "step": 13214 }, { "epoch": 3.97, "learning_rate": 2.5193434495192308e-05, "loss": 0.1609, "step": 13215 }, { "epoch": 3.97, "learning_rate": 2.5191556490384616e-05, "loss": 0.1465, "step": 13216 }, { "epoch": 3.97, "learning_rate": 2.5189678485576924e-05, "loss": 0.1184, "step": 13217 }, { "epoch": 3.97, "learning_rate": 2.518780048076923e-05, "loss": 0.2168, "step": 13218 }, { "epoch": 3.97, "learning_rate": 2.5185922475961537e-05, "loss": 0.1724, "step": 13219 }, { "epoch": 3.97, "learning_rate": 2.5184044471153846e-05, "loss": 0.1444, "step": 13220 }, { "epoch": 3.97, "learning_rate": 2.5182166466346157e-05, "loss": 0.2898, "step": 13221 }, { "epoch": 3.97, "learning_rate": 2.5180288461538465e-05, "loss": 0.2151, "step": 13222 }, { "epoch": 3.97, "learning_rate": 2.5178410456730773e-05, "loss": 0.1681, "step": 13223 }, { "epoch": 3.97, "learning_rate": 2.5176532451923078e-05, "loss": 0.1403, "step": 13224 }, { "epoch": 3.97, "learning_rate": 2.5174654447115386e-05, "loss": 0.1564, "step": 13225 }, { "epoch": 3.97, "learning_rate": 2.5172776442307695e-05, "loss": 0.1377, "step": 13226 }, { "epoch": 3.97, "learning_rate": 2.5170898437500003e-05, "loss": 0.1409, "step": 13227 }, { "epoch": 3.97, "learning_rate": 2.5169020432692307e-05, "loss": 0.1452, "step": 13228 }, { "epoch": 3.98, "learning_rate": 2.5167142427884616e-05, "loss": 0.1955, "step": 13229 }, { "epoch": 3.98, "learning_rate": 2.5165264423076924e-05, "loss": 0.2048, "step": 13230 }, { "epoch": 3.98, "learning_rate": 2.5163386418269232e-05, "loss": 0.1488, "step": 13231 }, { "epoch": 3.98, "learning_rate": 2.5161508413461537e-05, "loss": 0.1339, "step": 13232 }, { "epoch": 3.98, "learning_rate": 2.5159630408653845e-05, "loss": 0.137, "step": 13233 }, { "epoch": 3.98, "learning_rate": 2.5157752403846156e-05, "loss": 0.1263, "step": 13234 }, { "epoch": 3.98, "learning_rate": 2.5155874399038465e-05, "loss": 0.1145, "step": 13235 }, { "epoch": 3.98, "learning_rate": 2.5153996394230773e-05, "loss": 0.1107, "step": 13236 }, { "epoch": 3.98, "learning_rate": 2.515211838942308e-05, "loss": 0.1484, "step": 13237 }, { "epoch": 3.98, "learning_rate": 2.5150240384615386e-05, "loss": 0.1981, "step": 13238 }, { "epoch": 3.98, "learning_rate": 2.5148362379807694e-05, "loss": 0.1433, "step": 13239 }, { "epoch": 3.98, "learning_rate": 2.5146484375000002e-05, "loss": 0.1442, "step": 13240 }, { "epoch": 3.98, "learning_rate": 2.5144606370192307e-05, "loss": 0.1122, "step": 13241 }, { "epoch": 3.98, "learning_rate": 2.5142728365384615e-05, "loss": 0.1358, "step": 13242 }, { "epoch": 3.98, "learning_rate": 2.5140850360576923e-05, "loss": 0.1211, "step": 13243 }, { "epoch": 3.98, "learning_rate": 2.513897235576923e-05, "loss": 0.1086, "step": 13244 }, { "epoch": 3.98, "learning_rate": 2.5137094350961536e-05, "loss": 0.1453, "step": 13245 }, { "epoch": 3.98, "learning_rate": 2.5135216346153844e-05, "loss": 0.1498, "step": 13246 }, { "epoch": 3.98, "learning_rate": 2.513333834134616e-05, "loss": 0.1502, "step": 13247 }, { "epoch": 3.98, "learning_rate": 2.5131460336538464e-05, "loss": 0.1462, "step": 13248 }, { "epoch": 3.98, "learning_rate": 2.5129582331730772e-05, "loss": 0.1641, "step": 13249 }, { "epoch": 3.98, "learning_rate": 2.512770432692308e-05, "loss": 0.133, "step": 13250 }, { "epoch": 3.98, "learning_rate": 2.5125826322115385e-05, "loss": 0.1776, "step": 13251 }, { "epoch": 3.98, "learning_rate": 2.5123948317307693e-05, "loss": 0.1395, "step": 13252 }, { "epoch": 3.98, "learning_rate": 2.51220703125e-05, "loss": 0.1516, "step": 13253 }, { "epoch": 3.98, "learning_rate": 2.512019230769231e-05, "loss": 0.1347, "step": 13254 }, { "epoch": 3.98, "learning_rate": 2.5118314302884614e-05, "loss": 0.1182, "step": 13255 }, { "epoch": 3.98, "learning_rate": 2.5116436298076922e-05, "loss": 0.1358, "step": 13256 }, { "epoch": 3.98, "learning_rate": 2.511455829326923e-05, "loss": 0.1222, "step": 13257 }, { "epoch": 3.98, "learning_rate": 2.511268028846154e-05, "loss": 0.1177, "step": 13258 }, { "epoch": 3.98, "learning_rate": 2.5110802283653843e-05, "loss": 0.0913, "step": 13259 }, { "epoch": 3.98, "learning_rate": 2.510892427884616e-05, "loss": 0.1148, "step": 13260 }, { "epoch": 3.98, "learning_rate": 2.5107046274038463e-05, "loss": 0.1867, "step": 13261 }, { "epoch": 3.98, "learning_rate": 2.510516826923077e-05, "loss": 0.1418, "step": 13262 }, { "epoch": 3.99, "learning_rate": 2.510329026442308e-05, "loss": 0.1762, "step": 13263 }, { "epoch": 3.99, "learning_rate": 2.5101412259615388e-05, "loss": 0.1205, "step": 13264 }, { "epoch": 3.99, "learning_rate": 2.5099534254807692e-05, "loss": 0.1912, "step": 13265 }, { "epoch": 3.99, "learning_rate": 2.509765625e-05, "loss": 0.1275, "step": 13266 }, { "epoch": 3.99, "learning_rate": 2.509577824519231e-05, "loss": 0.1516, "step": 13267 }, { "epoch": 3.99, "learning_rate": 2.5093900240384617e-05, "loss": 0.1095, "step": 13268 }, { "epoch": 3.99, "learning_rate": 2.5092022235576922e-05, "loss": 0.152, "step": 13269 }, { "epoch": 3.99, "learning_rate": 2.509014423076923e-05, "loss": 0.1631, "step": 13270 }, { "epoch": 3.99, "learning_rate": 2.5088266225961538e-05, "loss": 0.1275, "step": 13271 }, { "epoch": 3.99, "learning_rate": 2.5086388221153846e-05, "loss": 0.1341, "step": 13272 }, { "epoch": 3.99, "learning_rate": 2.5084510216346158e-05, "loss": 0.1589, "step": 13273 }, { "epoch": 3.99, "learning_rate": 2.5082632211538466e-05, "loss": 0.094, "step": 13274 }, { "epoch": 3.99, "learning_rate": 2.508075420673077e-05, "loss": 0.1062, "step": 13275 }, { "epoch": 3.99, "learning_rate": 2.507887620192308e-05, "loss": 0.138, "step": 13276 }, { "epoch": 3.99, "learning_rate": 2.5076998197115387e-05, "loss": 0.1337, "step": 13277 }, { "epoch": 3.99, "learning_rate": 2.5075120192307695e-05, "loss": 0.2026, "step": 13278 }, { "epoch": 3.99, "learning_rate": 2.50732421875e-05, "loss": 0.1585, "step": 13279 }, { "epoch": 3.99, "learning_rate": 2.5071364182692308e-05, "loss": 0.1749, "step": 13280 }, { "epoch": 3.99, "learning_rate": 2.5069486177884616e-05, "loss": 0.2005, "step": 13281 }, { "epoch": 3.99, "learning_rate": 2.5067608173076924e-05, "loss": 0.1308, "step": 13282 }, { "epoch": 3.99, "learning_rate": 2.506573016826923e-05, "loss": 0.1538, "step": 13283 }, { "epoch": 3.99, "learning_rate": 2.5063852163461537e-05, "loss": 0.1834, "step": 13284 }, { "epoch": 3.99, "learning_rate": 2.5061974158653845e-05, "loss": 0.1413, "step": 13285 }, { "epoch": 3.99, "learning_rate": 2.5060096153846157e-05, "loss": 0.1372, "step": 13286 }, { "epoch": 3.99, "learning_rate": 2.5058218149038465e-05, "loss": 0.1575, "step": 13287 }, { "epoch": 3.99, "learning_rate": 2.5056340144230773e-05, "loss": 0.1563, "step": 13288 }, { "epoch": 3.99, "learning_rate": 2.5054462139423078e-05, "loss": 0.1962, "step": 13289 }, { "epoch": 3.99, "learning_rate": 2.5052584134615386e-05, "loss": 0.1272, "step": 13290 }, { "epoch": 3.99, "learning_rate": 2.5050706129807694e-05, "loss": 0.2768, "step": 13291 }, { "epoch": 3.99, "learning_rate": 2.5048828125000003e-05, "loss": 0.1477, "step": 13292 }, { "epoch": 3.99, "learning_rate": 2.5046950120192307e-05, "loss": 0.2652, "step": 13293 }, { "epoch": 3.99, "learning_rate": 2.5045072115384616e-05, "loss": 0.2155, "step": 13294 }, { "epoch": 3.99, "learning_rate": 2.5043194110576924e-05, "loss": 0.143, "step": 13295 }, { "epoch": 4.0, "learning_rate": 2.5041316105769232e-05, "loss": 0.1868, "step": 13296 }, { "epoch": 4.0, "learning_rate": 2.5039438100961537e-05, "loss": 0.1328, "step": 13297 }, { "epoch": 4.0, "learning_rate": 2.5037560096153845e-05, "loss": 0.1184, "step": 13298 }, { "epoch": 4.0, "learning_rate": 2.5035682091346156e-05, "loss": 0.1521, "step": 13299 }, { "epoch": 4.0, "learning_rate": 2.5033804086538464e-05, "loss": 0.1691, "step": 13300 }, { "epoch": 4.0, "learning_rate": 2.5031926081730773e-05, "loss": 0.1161, "step": 13301 }, { "epoch": 4.0, "learning_rate": 2.503004807692308e-05, "loss": 0.1301, "step": 13302 }, { "epoch": 4.0, "learning_rate": 2.5028170072115386e-05, "loss": 0.2464, "step": 13303 }, { "epoch": 4.0, "learning_rate": 2.5026292067307694e-05, "loss": 0.1355, "step": 13304 }, { "epoch": 4.0, "learning_rate": 2.5024414062500002e-05, "loss": 0.1563, "step": 13305 }, { "epoch": 4.0, "learning_rate": 2.502253605769231e-05, "loss": 0.0928, "step": 13306 }, { "epoch": 4.0, "learning_rate": 2.5020658052884615e-05, "loss": 0.1453, "step": 13307 }, { "epoch": 4.0, "learning_rate": 2.5018780048076923e-05, "loss": 0.1291, "step": 13308 }, { "epoch": 4.0, "learning_rate": 2.501690204326923e-05, "loss": 0.1528, "step": 13309 }, { "epoch": 4.0, "learning_rate": 2.5015024038461536e-05, "loss": 0.1385, "step": 13310 }, { "epoch": 4.0, "learning_rate": 2.5013146033653844e-05, "loss": 0.1253, "step": 13311 }, { "epoch": 4.0, "learning_rate": 2.501126802884616e-05, "loss": 0.2276, "step": 13312 } ], "logging_steps": 1, "max_steps": 26624, "num_train_epochs": 8, "save_steps": 32, "total_flos": 4.806490731949916e+18, "trial_name": null, "trial_params": null }