{ "best_metric": null, "best_model_checkpoint": null, "epoch": 1.9997711682948225, "global_step": 11470, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 4.359197907585005e-08, "loss": 0.8314, "step": 1 }, { "epoch": 0.0, "learning_rate": 8.71839581517001e-08, "loss": 0.807, "step": 2 }, { "epoch": 0.0, "learning_rate": 1.3077593722755015e-07, "loss": 0.8166, "step": 3 }, { "epoch": 0.0, "learning_rate": 1.743679163034002e-07, "loss": 0.8021, "step": 4 }, { "epoch": 0.0, "learning_rate": 2.179598953792502e-07, "loss": 0.8298, "step": 5 }, { "epoch": 0.0, "learning_rate": 2.615518744551003e-07, "loss": 0.7972, "step": 6 }, { "epoch": 0.0, "learning_rate": 3.051438535309503e-07, "loss": 0.8178, "step": 7 }, { "epoch": 0.0, "learning_rate": 3.487358326068004e-07, "loss": 0.8188, "step": 8 }, { "epoch": 0.0, "learning_rate": 3.923278116826504e-07, "loss": 0.8183, "step": 9 }, { "epoch": 0.0, "learning_rate": 4.359197907585004e-07, "loss": 0.8108, "step": 10 }, { "epoch": 0.0, "learning_rate": 4.795117698343506e-07, "loss": 0.8066, "step": 11 }, { "epoch": 0.0, "learning_rate": 5.231037489102006e-07, "loss": 0.8021, "step": 12 }, { "epoch": 0.0, "learning_rate": 5.666957279860506e-07, "loss": 0.823, "step": 13 }, { "epoch": 0.0, "learning_rate": 6.102877070619006e-07, "loss": 0.7961, "step": 14 }, { "epoch": 0.0, "learning_rate": 6.538796861377506e-07, "loss": 0.8009, "step": 15 }, { "epoch": 0.0, "learning_rate": 6.974716652136008e-07, "loss": 0.7923, "step": 16 }, { "epoch": 0.0, "learning_rate": 7.410636442894508e-07, "loss": 0.8038, "step": 17 }, { "epoch": 0.0, "learning_rate": 7.846556233653008e-07, "loss": 0.7977, "step": 18 }, { "epoch": 0.0, "learning_rate": 8.282476024411508e-07, "loss": 0.7579, "step": 19 }, { "epoch": 0.0, "learning_rate": 8.718395815170008e-07, "loss": 0.7758, "step": 20 }, { "epoch": 0.0, "learning_rate": 9.154315605928509e-07, "loss": 0.753, "step": 21 }, { "epoch": 0.0, "learning_rate": 9.590235396687012e-07, "loss": 0.7557, "step": 22 }, { "epoch": 0.0, "learning_rate": 1.0026155187445512e-06, "loss": 0.7489, "step": 23 }, { "epoch": 0.0, "learning_rate": 1.0462074978204012e-06, "loss": 0.7563, "step": 24 }, { "epoch": 0.0, "learning_rate": 1.0897994768962512e-06, "loss": 0.7487, "step": 25 }, { "epoch": 0.0, "learning_rate": 1.1333914559721012e-06, "loss": 0.7416, "step": 26 }, { "epoch": 0.0, "learning_rate": 1.1769834350479512e-06, "loss": 0.7467, "step": 27 }, { "epoch": 0.0, "learning_rate": 1.2205754141238013e-06, "loss": 0.7204, "step": 28 }, { "epoch": 0.01, "learning_rate": 1.2641673931996513e-06, "loss": 0.6837, "step": 29 }, { "epoch": 0.01, "learning_rate": 1.3077593722755013e-06, "loss": 0.6919, "step": 30 }, { "epoch": 0.01, "learning_rate": 1.3513513513513515e-06, "loss": 0.6882, "step": 31 }, { "epoch": 0.01, "learning_rate": 1.3949433304272015e-06, "loss": 0.676, "step": 32 }, { "epoch": 0.01, "learning_rate": 1.4385353095030515e-06, "loss": 0.6599, "step": 33 }, { "epoch": 0.01, "learning_rate": 1.4821272885789016e-06, "loss": 0.6719, "step": 34 }, { "epoch": 0.01, "learning_rate": 1.5257192676547516e-06, "loss": 0.656, "step": 35 }, { "epoch": 0.01, "learning_rate": 1.5693112467306016e-06, "loss": 0.6539, "step": 36 }, { "epoch": 0.01, "learning_rate": 1.6129032258064516e-06, "loss": 0.6707, "step": 37 }, { "epoch": 0.01, "learning_rate": 1.6564952048823016e-06, "loss": 0.6502, "step": 38 }, { "epoch": 0.01, "learning_rate": 1.7000871839581516e-06, "loss": 0.6584, "step": 39 }, { "epoch": 0.01, "learning_rate": 1.7436791630340016e-06, "loss": 0.6717, "step": 40 }, { "epoch": 0.01, "learning_rate": 1.7872711421098517e-06, "loss": 0.6733, "step": 41 }, { "epoch": 0.01, "learning_rate": 1.8308631211857019e-06, "loss": 0.6502, "step": 42 }, { "epoch": 0.01, "learning_rate": 1.8744551002615519e-06, "loss": 0.6265, "step": 43 }, { "epoch": 0.01, "learning_rate": 1.9180470793374023e-06, "loss": 0.6359, "step": 44 }, { "epoch": 0.01, "learning_rate": 1.961639058413252e-06, "loss": 0.6547, "step": 45 }, { "epoch": 0.01, "learning_rate": 2.0052310374891024e-06, "loss": 0.6111, "step": 46 }, { "epoch": 0.01, "learning_rate": 2.048823016564952e-06, "loss": 0.6214, "step": 47 }, { "epoch": 0.01, "learning_rate": 2.0924149956408024e-06, "loss": 0.6542, "step": 48 }, { "epoch": 0.01, "learning_rate": 2.136006974716652e-06, "loss": 0.6249, "step": 49 }, { "epoch": 0.01, "learning_rate": 2.1795989537925024e-06, "loss": 0.6467, "step": 50 }, { "epoch": 0.01, "learning_rate": 2.2231909328683522e-06, "loss": 0.6467, "step": 51 }, { "epoch": 0.01, "learning_rate": 2.2667829119442024e-06, "loss": 0.619, "step": 52 }, { "epoch": 0.01, "learning_rate": 2.3103748910200522e-06, "loss": 0.6556, "step": 53 }, { "epoch": 0.01, "learning_rate": 2.3539668700959025e-06, "loss": 0.622, "step": 54 }, { "epoch": 0.01, "learning_rate": 2.3975588491717523e-06, "loss": 0.6318, "step": 55 }, { "epoch": 0.01, "learning_rate": 2.4411508282476025e-06, "loss": 0.6042, "step": 56 }, { "epoch": 0.01, "learning_rate": 2.4847428073234523e-06, "loss": 0.6452, "step": 57 }, { "epoch": 0.01, "learning_rate": 2.5283347863993025e-06, "loss": 0.6304, "step": 58 }, { "epoch": 0.01, "learning_rate": 2.5719267654751528e-06, "loss": 0.6053, "step": 59 }, { "epoch": 0.01, "learning_rate": 2.6155187445510026e-06, "loss": 0.6057, "step": 60 }, { "epoch": 0.01, "learning_rate": 2.659110723626853e-06, "loss": 0.6171, "step": 61 }, { "epoch": 0.01, "learning_rate": 2.702702702702703e-06, "loss": 0.6072, "step": 62 }, { "epoch": 0.01, "learning_rate": 2.746294681778553e-06, "loss": 0.6026, "step": 63 }, { "epoch": 0.01, "learning_rate": 2.789886660854403e-06, "loss": 0.6168, "step": 64 }, { "epoch": 0.01, "learning_rate": 2.8334786399302533e-06, "loss": 0.5994, "step": 65 }, { "epoch": 0.01, "learning_rate": 2.877070619006103e-06, "loss": 0.6142, "step": 66 }, { "epoch": 0.01, "learning_rate": 2.9206625980819533e-06, "loss": 0.6057, "step": 67 }, { "epoch": 0.01, "learning_rate": 2.964254577157803e-06, "loss": 0.6118, "step": 68 }, { "epoch": 0.01, "learning_rate": 3.0078465562336533e-06, "loss": 0.6124, "step": 69 }, { "epoch": 0.01, "learning_rate": 3.051438535309503e-06, "loss": 0.6117, "step": 70 }, { "epoch": 0.01, "learning_rate": 3.0950305143853534e-06, "loss": 0.595, "step": 71 }, { "epoch": 0.01, "learning_rate": 3.138622493461203e-06, "loss": 0.6007, "step": 72 }, { "epoch": 0.01, "learning_rate": 3.1822144725370534e-06, "loss": 0.5975, "step": 73 }, { "epoch": 0.01, "learning_rate": 3.225806451612903e-06, "loss": 0.608, "step": 74 }, { "epoch": 0.01, "learning_rate": 3.2693984306887534e-06, "loss": 0.6062, "step": 75 }, { "epoch": 0.01, "learning_rate": 3.3129904097646032e-06, "loss": 0.6107, "step": 76 }, { "epoch": 0.01, "learning_rate": 3.3565823888404534e-06, "loss": 0.599, "step": 77 }, { "epoch": 0.01, "learning_rate": 3.4001743679163032e-06, "loss": 0.6136, "step": 78 }, { "epoch": 0.01, "learning_rate": 3.4437663469921535e-06, "loss": 0.5922, "step": 79 }, { "epoch": 0.01, "learning_rate": 3.4873583260680033e-06, "loss": 0.5769, "step": 80 }, { "epoch": 0.01, "learning_rate": 3.5309503051438535e-06, "loss": 0.5936, "step": 81 }, { "epoch": 0.01, "learning_rate": 3.5745422842197033e-06, "loss": 0.594, "step": 82 }, { "epoch": 0.01, "learning_rate": 3.6181342632955535e-06, "loss": 0.5986, "step": 83 }, { "epoch": 0.01, "learning_rate": 3.6617262423714038e-06, "loss": 0.5715, "step": 84 }, { "epoch": 0.01, "learning_rate": 3.7053182214472536e-06, "loss": 0.574, "step": 85 }, { "epoch": 0.01, "learning_rate": 3.7489102005231038e-06, "loss": 0.5761, "step": 86 }, { "epoch": 0.02, "learning_rate": 3.7925021795989544e-06, "loss": 0.5891, "step": 87 }, { "epoch": 0.02, "learning_rate": 3.836094158674805e-06, "loss": 0.5985, "step": 88 }, { "epoch": 0.02, "learning_rate": 3.8796861377506545e-06, "loss": 0.5643, "step": 89 }, { "epoch": 0.02, "learning_rate": 3.923278116826504e-06, "loss": 0.5967, "step": 90 }, { "epoch": 0.02, "learning_rate": 3.966870095902354e-06, "loss": 0.5707, "step": 91 }, { "epoch": 0.02, "learning_rate": 4.010462074978205e-06, "loss": 0.577, "step": 92 }, { "epoch": 0.02, "learning_rate": 4.0540540540540545e-06, "loss": 0.5644, "step": 93 }, { "epoch": 0.02, "learning_rate": 4.097646033129904e-06, "loss": 0.5647, "step": 94 }, { "epoch": 0.02, "learning_rate": 4.141238012205754e-06, "loss": 0.5907, "step": 95 }, { "epoch": 0.02, "learning_rate": 4.184829991281605e-06, "loss": 0.5831, "step": 96 }, { "epoch": 0.02, "learning_rate": 4.228421970357455e-06, "loss": 0.595, "step": 97 }, { "epoch": 0.02, "learning_rate": 4.272013949433304e-06, "loss": 0.5734, "step": 98 }, { "epoch": 0.02, "learning_rate": 4.315605928509155e-06, "loss": 0.5832, "step": 99 }, { "epoch": 0.02, "learning_rate": 4.359197907585005e-06, "loss": 0.5661, "step": 100 }, { "epoch": 0.02, "learning_rate": 4.402789886660855e-06, "loss": 0.5705, "step": 101 }, { "epoch": 0.02, "learning_rate": 4.4463818657367044e-06, "loss": 0.5816, "step": 102 }, { "epoch": 0.02, "learning_rate": 4.489973844812555e-06, "loss": 0.563, "step": 103 }, { "epoch": 0.02, "learning_rate": 4.533565823888405e-06, "loss": 0.5874, "step": 104 }, { "epoch": 0.02, "learning_rate": 4.577157802964255e-06, "loss": 0.5764, "step": 105 }, { "epoch": 0.02, "learning_rate": 4.6207497820401045e-06, "loss": 0.5903, "step": 106 }, { "epoch": 0.02, "learning_rate": 4.664341761115955e-06, "loss": 0.5722, "step": 107 }, { "epoch": 0.02, "learning_rate": 4.707933740191805e-06, "loss": 0.5692, "step": 108 }, { "epoch": 0.02, "learning_rate": 4.751525719267655e-06, "loss": 0.585, "step": 109 }, { "epoch": 0.02, "learning_rate": 4.7951176983435046e-06, "loss": 0.6061, "step": 110 }, { "epoch": 0.02, "learning_rate": 4.838709677419355e-06, "loss": 0.5794, "step": 111 }, { "epoch": 0.02, "learning_rate": 4.882301656495205e-06, "loss": 0.5656, "step": 112 }, { "epoch": 0.02, "learning_rate": 4.925893635571055e-06, "loss": 0.5853, "step": 113 }, { "epoch": 0.02, "learning_rate": 4.969485614646905e-06, "loss": 0.5758, "step": 114 }, { "epoch": 0.02, "learning_rate": 5.013077593722755e-06, "loss": 0.5758, "step": 115 }, { "epoch": 0.02, "learning_rate": 5.056669572798605e-06, "loss": 0.5744, "step": 116 }, { "epoch": 0.02, "learning_rate": 5.100261551874455e-06, "loss": 0.5522, "step": 117 }, { "epoch": 0.02, "learning_rate": 5.1438535309503055e-06, "loss": 0.5656, "step": 118 }, { "epoch": 0.02, "learning_rate": 5.187445510026155e-06, "loss": 0.5543, "step": 119 }, { "epoch": 0.02, "learning_rate": 5.231037489102005e-06, "loss": 0.5491, "step": 120 }, { "epoch": 0.02, "learning_rate": 5.274629468177855e-06, "loss": 0.579, "step": 121 }, { "epoch": 0.02, "learning_rate": 5.318221447253706e-06, "loss": 0.5616, "step": 122 }, { "epoch": 0.02, "learning_rate": 5.361813426329556e-06, "loss": 0.5683, "step": 123 }, { "epoch": 0.02, "learning_rate": 5.405405405405406e-06, "loss": 0.5737, "step": 124 }, { "epoch": 0.02, "learning_rate": 5.448997384481256e-06, "loss": 0.5858, "step": 125 }, { "epoch": 0.02, "learning_rate": 5.492589363557106e-06, "loss": 0.5698, "step": 126 }, { "epoch": 0.02, "learning_rate": 5.536181342632956e-06, "loss": 0.5564, "step": 127 }, { "epoch": 0.02, "learning_rate": 5.579773321708806e-06, "loss": 0.5515, "step": 128 }, { "epoch": 0.02, "learning_rate": 5.623365300784656e-06, "loss": 0.5782, "step": 129 }, { "epoch": 0.02, "learning_rate": 5.6669572798605065e-06, "loss": 0.5698, "step": 130 }, { "epoch": 0.02, "learning_rate": 5.710549258936356e-06, "loss": 0.5534, "step": 131 }, { "epoch": 0.02, "learning_rate": 5.754141238012206e-06, "loss": 0.5401, "step": 132 }, { "epoch": 0.02, "learning_rate": 5.797733217088056e-06, "loss": 0.5582, "step": 133 }, { "epoch": 0.02, "learning_rate": 5.841325196163907e-06, "loss": 0.5902, "step": 134 }, { "epoch": 0.02, "learning_rate": 5.884917175239756e-06, "loss": 0.5442, "step": 135 }, { "epoch": 0.02, "learning_rate": 5.928509154315606e-06, "loss": 0.5705, "step": 136 }, { "epoch": 0.02, "learning_rate": 5.972101133391456e-06, "loss": 0.5486, "step": 137 }, { "epoch": 0.02, "learning_rate": 6.015693112467307e-06, "loss": 0.5671, "step": 138 }, { "epoch": 0.02, "learning_rate": 6.0592850915431565e-06, "loss": 0.567, "step": 139 }, { "epoch": 0.02, "learning_rate": 6.102877070619006e-06, "loss": 0.5739, "step": 140 }, { "epoch": 0.02, "learning_rate": 6.146469049694856e-06, "loss": 0.5435, "step": 141 }, { "epoch": 0.02, "learning_rate": 6.190061028770707e-06, "loss": 0.5607, "step": 142 }, { "epoch": 0.02, "learning_rate": 6.2336530078465565e-06, "loss": 0.5556, "step": 143 }, { "epoch": 0.03, "learning_rate": 6.277244986922406e-06, "loss": 0.554, "step": 144 }, { "epoch": 0.03, "learning_rate": 6.320836965998257e-06, "loss": 0.5596, "step": 145 }, { "epoch": 0.03, "learning_rate": 6.364428945074107e-06, "loss": 0.5558, "step": 146 }, { "epoch": 0.03, "learning_rate": 6.408020924149957e-06, "loss": 0.5681, "step": 147 }, { "epoch": 0.03, "learning_rate": 6.451612903225806e-06, "loss": 0.5804, "step": 148 }, { "epoch": 0.03, "learning_rate": 6.495204882301657e-06, "loss": 0.5726, "step": 149 }, { "epoch": 0.03, "learning_rate": 6.538796861377507e-06, "loss": 0.5653, "step": 150 }, { "epoch": 0.03, "learning_rate": 6.5823888404533575e-06, "loss": 0.5755, "step": 151 }, { "epoch": 0.03, "learning_rate": 6.6259808195292064e-06, "loss": 0.5683, "step": 152 }, { "epoch": 0.03, "learning_rate": 6.669572798605057e-06, "loss": 0.5622, "step": 153 }, { "epoch": 0.03, "learning_rate": 6.713164777680907e-06, "loss": 0.539, "step": 154 }, { "epoch": 0.03, "learning_rate": 6.7567567567567575e-06, "loss": 0.5531, "step": 155 }, { "epoch": 0.03, "learning_rate": 6.8003487358326065e-06, "loss": 0.5707, "step": 156 }, { "epoch": 0.03, "learning_rate": 6.843940714908457e-06, "loss": 0.546, "step": 157 }, { "epoch": 0.03, "learning_rate": 6.887532693984307e-06, "loss": 0.5555, "step": 158 }, { "epoch": 0.03, "learning_rate": 6.931124673060158e-06, "loss": 0.5691, "step": 159 }, { "epoch": 0.03, "learning_rate": 6.9747166521360066e-06, "loss": 0.5672, "step": 160 }, { "epoch": 0.03, "learning_rate": 7.018308631211857e-06, "loss": 0.5496, "step": 161 }, { "epoch": 0.03, "learning_rate": 7.061900610287707e-06, "loss": 0.5591, "step": 162 }, { "epoch": 0.03, "learning_rate": 7.105492589363558e-06, "loss": 0.5455, "step": 163 }, { "epoch": 0.03, "learning_rate": 7.149084568439407e-06, "loss": 0.5642, "step": 164 }, { "epoch": 0.03, "learning_rate": 7.192676547515258e-06, "loss": 0.5445, "step": 165 }, { "epoch": 0.03, "learning_rate": 7.236268526591107e-06, "loss": 0.5542, "step": 166 }, { "epoch": 0.03, "learning_rate": 7.279860505666958e-06, "loss": 0.5445, "step": 167 }, { "epoch": 0.03, "learning_rate": 7.3234524847428075e-06, "loss": 0.5382, "step": 168 }, { "epoch": 0.03, "learning_rate": 7.367044463818658e-06, "loss": 0.5318, "step": 169 }, { "epoch": 0.03, "learning_rate": 7.410636442894507e-06, "loss": 0.5551, "step": 170 }, { "epoch": 0.03, "learning_rate": 7.454228421970358e-06, "loss": 0.56, "step": 171 }, { "epoch": 0.03, "learning_rate": 7.4978204010462076e-06, "loss": 0.5526, "step": 172 }, { "epoch": 0.03, "learning_rate": 7.541412380122058e-06, "loss": 0.5665, "step": 173 }, { "epoch": 0.03, "learning_rate": 7.585004359197909e-06, "loss": 0.6194, "step": 174 }, { "epoch": 0.03, "learning_rate": 7.628596338273758e-06, "loss": 0.5626, "step": 175 }, { "epoch": 0.03, "learning_rate": 7.67218831734961e-06, "loss": 0.5474, "step": 176 }, { "epoch": 0.03, "learning_rate": 7.715780296425458e-06, "loss": 0.5465, "step": 177 }, { "epoch": 0.03, "learning_rate": 7.759372275501309e-06, "loss": 0.5558, "step": 178 }, { "epoch": 0.03, "learning_rate": 7.802964254577158e-06, "loss": 0.5326, "step": 179 }, { "epoch": 0.03, "learning_rate": 7.846556233653009e-06, "loss": 0.5369, "step": 180 }, { "epoch": 0.03, "learning_rate": 7.890148212728857e-06, "loss": 0.5821, "step": 181 }, { "epoch": 0.03, "learning_rate": 7.933740191804708e-06, "loss": 0.5383, "step": 182 }, { "epoch": 0.03, "learning_rate": 7.977332170880559e-06, "loss": 0.5318, "step": 183 }, { "epoch": 0.03, "learning_rate": 8.02092414995641e-06, "loss": 0.5576, "step": 184 }, { "epoch": 0.03, "learning_rate": 8.064516129032258e-06, "loss": 0.5488, "step": 185 }, { "epoch": 0.03, "learning_rate": 8.108108108108109e-06, "loss": 0.5665, "step": 186 }, { "epoch": 0.03, "learning_rate": 8.151700087183958e-06, "loss": 0.5405, "step": 187 }, { "epoch": 0.03, "learning_rate": 8.195292066259809e-06, "loss": 0.5483, "step": 188 }, { "epoch": 0.03, "learning_rate": 8.238884045335658e-06, "loss": 0.5517, "step": 189 }, { "epoch": 0.03, "learning_rate": 8.282476024411508e-06, "loss": 0.5392, "step": 190 }, { "epoch": 0.03, "learning_rate": 8.326068003487359e-06, "loss": 0.5572, "step": 191 }, { "epoch": 0.03, "learning_rate": 8.36965998256321e-06, "loss": 0.5462, "step": 192 }, { "epoch": 0.03, "learning_rate": 8.413251961639059e-06, "loss": 0.5532, "step": 193 }, { "epoch": 0.03, "learning_rate": 8.45684394071491e-06, "loss": 0.5536, "step": 194 }, { "epoch": 0.03, "learning_rate": 8.500435919790758e-06, "loss": 0.5457, "step": 195 }, { "epoch": 0.03, "learning_rate": 8.544027898866609e-06, "loss": 0.5463, "step": 196 }, { "epoch": 0.03, "learning_rate": 8.587619877942458e-06, "loss": 0.5449, "step": 197 }, { "epoch": 0.03, "learning_rate": 8.63121185701831e-06, "loss": 0.5499, "step": 198 }, { "epoch": 0.03, "learning_rate": 8.674803836094159e-06, "loss": 0.5402, "step": 199 }, { "epoch": 0.03, "learning_rate": 8.71839581517001e-06, "loss": 0.549, "step": 200 }, { "epoch": 0.04, "learning_rate": 8.761987794245859e-06, "loss": 0.5577, "step": 201 }, { "epoch": 0.04, "learning_rate": 8.80557977332171e-06, "loss": 0.5402, "step": 202 }, { "epoch": 0.04, "learning_rate": 8.849171752397558e-06, "loss": 0.5514, "step": 203 }, { "epoch": 0.04, "learning_rate": 8.892763731473409e-06, "loss": 0.5625, "step": 204 }, { "epoch": 0.04, "learning_rate": 8.93635571054926e-06, "loss": 0.5303, "step": 205 }, { "epoch": 0.04, "learning_rate": 8.97994768962511e-06, "loss": 0.5648, "step": 206 }, { "epoch": 0.04, "learning_rate": 9.023539668700959e-06, "loss": 0.5555, "step": 207 }, { "epoch": 0.04, "learning_rate": 9.06713164777681e-06, "loss": 0.5366, "step": 208 }, { "epoch": 0.04, "learning_rate": 9.11072362685266e-06, "loss": 0.5426, "step": 209 }, { "epoch": 0.04, "learning_rate": 9.15431560592851e-06, "loss": 0.5614, "step": 210 }, { "epoch": 0.04, "learning_rate": 9.19790758500436e-06, "loss": 0.5417, "step": 211 }, { "epoch": 0.04, "learning_rate": 9.241499564080209e-06, "loss": 0.54, "step": 212 }, { "epoch": 0.04, "learning_rate": 9.28509154315606e-06, "loss": 0.5602, "step": 213 }, { "epoch": 0.04, "learning_rate": 9.32868352223191e-06, "loss": 0.5464, "step": 214 }, { "epoch": 0.04, "learning_rate": 9.372275501307761e-06, "loss": 0.5296, "step": 215 }, { "epoch": 0.04, "learning_rate": 9.41586748038361e-06, "loss": 0.5365, "step": 216 }, { "epoch": 0.04, "learning_rate": 9.45945945945946e-06, "loss": 0.556, "step": 217 }, { "epoch": 0.04, "learning_rate": 9.50305143853531e-06, "loss": 0.5196, "step": 218 }, { "epoch": 0.04, "learning_rate": 9.54664341761116e-06, "loss": 0.5371, "step": 219 }, { "epoch": 0.04, "learning_rate": 9.590235396687009e-06, "loss": 0.5403, "step": 220 }, { "epoch": 0.04, "learning_rate": 9.63382737576286e-06, "loss": 0.5259, "step": 221 }, { "epoch": 0.04, "learning_rate": 9.67741935483871e-06, "loss": 0.5428, "step": 222 }, { "epoch": 0.04, "learning_rate": 9.721011333914561e-06, "loss": 0.529, "step": 223 }, { "epoch": 0.04, "learning_rate": 9.76460331299041e-06, "loss": 0.546, "step": 224 }, { "epoch": 0.04, "learning_rate": 9.80819529206626e-06, "loss": 0.5352, "step": 225 }, { "epoch": 0.04, "learning_rate": 9.85178727114211e-06, "loss": 0.5391, "step": 226 }, { "epoch": 0.04, "learning_rate": 9.89537925021796e-06, "loss": 0.5321, "step": 227 }, { "epoch": 0.04, "learning_rate": 9.93897122929381e-06, "loss": 0.575, "step": 228 }, { "epoch": 0.04, "learning_rate": 9.982563208369662e-06, "loss": 0.5386, "step": 229 }, { "epoch": 0.04, "learning_rate": 1.002615518744551e-05, "loss": 0.5239, "step": 230 }, { "epoch": 0.04, "learning_rate": 1.0069747166521361e-05, "loss": 0.5275, "step": 231 }, { "epoch": 0.04, "learning_rate": 1.011333914559721e-05, "loss": 0.5527, "step": 232 }, { "epoch": 0.04, "learning_rate": 1.015693112467306e-05, "loss": 0.554, "step": 233 }, { "epoch": 0.04, "learning_rate": 1.020052310374891e-05, "loss": 0.5283, "step": 234 }, { "epoch": 0.04, "learning_rate": 1.024411508282476e-05, "loss": 0.5401, "step": 235 }, { "epoch": 0.04, "learning_rate": 1.0287707061900611e-05, "loss": 0.5438, "step": 236 }, { "epoch": 0.04, "learning_rate": 1.0331299040976462e-05, "loss": 0.5128, "step": 237 }, { "epoch": 0.04, "learning_rate": 1.037489102005231e-05, "loss": 0.5396, "step": 238 }, { "epoch": 0.04, "learning_rate": 1.0418482999128161e-05, "loss": 0.5397, "step": 239 }, { "epoch": 0.04, "learning_rate": 1.046207497820401e-05, "loss": 0.5357, "step": 240 }, { "epoch": 0.04, "learning_rate": 1.0505666957279861e-05, "loss": 0.5523, "step": 241 }, { "epoch": 0.04, "learning_rate": 1.054925893635571e-05, "loss": 0.532, "step": 242 }, { "epoch": 0.04, "learning_rate": 1.059285091543156e-05, "loss": 0.5216, "step": 243 }, { "epoch": 0.04, "learning_rate": 1.0636442894507411e-05, "loss": 0.5289, "step": 244 }, { "epoch": 0.04, "learning_rate": 1.0680034873583262e-05, "loss": 0.5415, "step": 245 }, { "epoch": 0.04, "learning_rate": 1.0723626852659112e-05, "loss": 0.5304, "step": 246 }, { "epoch": 0.04, "learning_rate": 1.0767218831734961e-05, "loss": 0.5465, "step": 247 }, { "epoch": 0.04, "learning_rate": 1.0810810810810812e-05, "loss": 0.5386, "step": 248 }, { "epoch": 0.04, "learning_rate": 1.0854402789886661e-05, "loss": 0.5035, "step": 249 }, { "epoch": 0.04, "learning_rate": 1.0897994768962512e-05, "loss": 0.5252, "step": 250 }, { "epoch": 0.04, "learning_rate": 1.094158674803836e-05, "loss": 0.5376, "step": 251 }, { "epoch": 0.04, "learning_rate": 1.0985178727114211e-05, "loss": 0.523, "step": 252 }, { "epoch": 0.04, "learning_rate": 1.1028770706190062e-05, "loss": 0.5379, "step": 253 }, { "epoch": 0.04, "learning_rate": 1.1072362685265913e-05, "loss": 0.5446, "step": 254 }, { "epoch": 0.04, "learning_rate": 1.1115954664341762e-05, "loss": 0.5526, "step": 255 }, { "epoch": 0.04, "learning_rate": 1.1159546643417612e-05, "loss": 0.5462, "step": 256 }, { "epoch": 0.04, "learning_rate": 1.1203138622493461e-05, "loss": 0.5296, "step": 257 }, { "epoch": 0.04, "learning_rate": 1.1246730601569312e-05, "loss": 0.5434, "step": 258 }, { "epoch": 0.05, "learning_rate": 1.129032258064516e-05, "loss": 0.5301, "step": 259 }, { "epoch": 0.05, "learning_rate": 1.1333914559721013e-05, "loss": 0.5227, "step": 260 }, { "epoch": 0.05, "learning_rate": 1.1377506538796862e-05, "loss": 0.5366, "step": 261 }, { "epoch": 0.05, "learning_rate": 1.1421098517872713e-05, "loss": 0.5281, "step": 262 }, { "epoch": 0.05, "learning_rate": 1.1464690496948562e-05, "loss": 0.5105, "step": 263 }, { "epoch": 0.05, "learning_rate": 1.1508282476024412e-05, "loss": 0.5267, "step": 264 }, { "epoch": 0.05, "learning_rate": 1.1551874455100261e-05, "loss": 0.52, "step": 265 }, { "epoch": 0.05, "learning_rate": 1.1595466434176112e-05, "loss": 0.535, "step": 266 }, { "epoch": 0.05, "learning_rate": 1.1639058413251963e-05, "loss": 0.5449, "step": 267 }, { "epoch": 0.05, "learning_rate": 1.1682650392327813e-05, "loss": 0.5304, "step": 268 }, { "epoch": 0.05, "learning_rate": 1.1726242371403662e-05, "loss": 0.5413, "step": 269 }, { "epoch": 0.05, "learning_rate": 1.1769834350479513e-05, "loss": 0.5264, "step": 270 }, { "epoch": 0.05, "learning_rate": 1.1813426329555362e-05, "loss": 0.5263, "step": 271 }, { "epoch": 0.05, "learning_rate": 1.1857018308631212e-05, "loss": 0.5455, "step": 272 }, { "epoch": 0.05, "learning_rate": 1.1900610287707061e-05, "loss": 0.5109, "step": 273 }, { "epoch": 0.05, "learning_rate": 1.1944202266782912e-05, "loss": 0.5335, "step": 274 }, { "epoch": 0.05, "learning_rate": 1.1987794245858763e-05, "loss": 0.5315, "step": 275 }, { "epoch": 0.05, "learning_rate": 1.2031386224934613e-05, "loss": 0.5346, "step": 276 }, { "epoch": 0.05, "learning_rate": 1.2074978204010462e-05, "loss": 0.5376, "step": 277 }, { "epoch": 0.05, "learning_rate": 1.2118570183086313e-05, "loss": 0.5277, "step": 278 }, { "epoch": 0.05, "learning_rate": 1.2162162162162164e-05, "loss": 0.5377, "step": 279 }, { "epoch": 0.05, "learning_rate": 1.2205754141238013e-05, "loss": 0.5153, "step": 280 }, { "epoch": 0.05, "learning_rate": 1.2249346120313863e-05, "loss": 0.529, "step": 281 }, { "epoch": 0.05, "learning_rate": 1.2292938099389712e-05, "loss": 0.5312, "step": 282 }, { "epoch": 0.05, "learning_rate": 1.2336530078465563e-05, "loss": 0.5136, "step": 283 }, { "epoch": 0.05, "learning_rate": 1.2380122057541413e-05, "loss": 0.5209, "step": 284 }, { "epoch": 0.05, "learning_rate": 1.2423714036617264e-05, "loss": 0.5204, "step": 285 }, { "epoch": 0.05, "learning_rate": 1.2467306015693113e-05, "loss": 0.5278, "step": 286 }, { "epoch": 0.05, "learning_rate": 1.2510897994768964e-05, "loss": 0.5135, "step": 287 }, { "epoch": 0.05, "learning_rate": 1.2554489973844813e-05, "loss": 0.5112, "step": 288 }, { "epoch": 0.05, "learning_rate": 1.2598081952920662e-05, "loss": 0.5326, "step": 289 }, { "epoch": 0.05, "learning_rate": 1.2641673931996514e-05, "loss": 0.5401, "step": 290 }, { "epoch": 0.05, "learning_rate": 1.2685265911072363e-05, "loss": 0.5292, "step": 291 }, { "epoch": 0.05, "learning_rate": 1.2728857890148214e-05, "loss": 0.5415, "step": 292 }, { "epoch": 0.05, "learning_rate": 1.2772449869224063e-05, "loss": 0.5139, "step": 293 }, { "epoch": 0.05, "learning_rate": 1.2816041848299915e-05, "loss": 0.5148, "step": 294 }, { "epoch": 0.05, "learning_rate": 1.2859633827375764e-05, "loss": 0.5269, "step": 295 }, { "epoch": 0.05, "learning_rate": 1.2903225806451613e-05, "loss": 0.5404, "step": 296 }, { "epoch": 0.05, "learning_rate": 1.2946817785527462e-05, "loss": 0.533, "step": 297 }, { "epoch": 0.05, "learning_rate": 1.2990409764603314e-05, "loss": 0.5212, "step": 298 }, { "epoch": 0.05, "learning_rate": 1.3034001743679165e-05, "loss": 0.526, "step": 299 }, { "epoch": 0.05, "learning_rate": 1.3077593722755014e-05, "loss": 0.5191, "step": 300 }, { "epoch": 0.05, "learning_rate": 1.3121185701830863e-05, "loss": 0.5085, "step": 301 }, { "epoch": 0.05, "learning_rate": 1.3164777680906715e-05, "loss": 0.5212, "step": 302 }, { "epoch": 0.05, "learning_rate": 1.3208369659982564e-05, "loss": 0.5079, "step": 303 }, { "epoch": 0.05, "learning_rate": 1.3251961639058413e-05, "loss": 0.5343, "step": 304 }, { "epoch": 0.05, "learning_rate": 1.3295553618134264e-05, "loss": 0.5231, "step": 305 }, { "epoch": 0.05, "learning_rate": 1.3339145597210114e-05, "loss": 0.5307, "step": 306 }, { "epoch": 0.05, "learning_rate": 1.3382737576285965e-05, "loss": 0.5264, "step": 307 }, { "epoch": 0.05, "learning_rate": 1.3426329555361814e-05, "loss": 0.5167, "step": 308 }, { "epoch": 0.05, "learning_rate": 1.3469921534437663e-05, "loss": 0.5223, "step": 309 }, { "epoch": 0.05, "learning_rate": 1.3513513513513515e-05, "loss": 0.5215, "step": 310 }, { "epoch": 0.05, "learning_rate": 1.3557105492589364e-05, "loss": 0.5248, "step": 311 }, { "epoch": 0.05, "learning_rate": 1.3600697471665213e-05, "loss": 0.5125, "step": 312 }, { "epoch": 0.05, "learning_rate": 1.3644289450741065e-05, "loss": 0.5288, "step": 313 }, { "epoch": 0.05, "learning_rate": 1.3687881429816914e-05, "loss": 0.5365, "step": 314 }, { "epoch": 0.05, "learning_rate": 1.3731473408892765e-05, "loss": 0.5146, "step": 315 }, { "epoch": 0.06, "learning_rate": 1.3775065387968614e-05, "loss": 0.5183, "step": 316 }, { "epoch": 0.06, "learning_rate": 1.3818657367044466e-05, "loss": 0.505, "step": 317 }, { "epoch": 0.06, "learning_rate": 1.3862249346120315e-05, "loss": 0.5285, "step": 318 }, { "epoch": 0.06, "learning_rate": 1.3905841325196164e-05, "loss": 0.5351, "step": 319 }, { "epoch": 0.06, "learning_rate": 1.3949433304272013e-05, "loss": 0.5187, "step": 320 }, { "epoch": 0.06, "learning_rate": 1.3993025283347865e-05, "loss": 0.5073, "step": 321 }, { "epoch": 0.06, "learning_rate": 1.4036617262423714e-05, "loss": 0.5335, "step": 322 }, { "epoch": 0.06, "learning_rate": 1.4080209241499565e-05, "loss": 0.5194, "step": 323 }, { "epoch": 0.06, "learning_rate": 1.4123801220575414e-05, "loss": 0.5123, "step": 324 }, { "epoch": 0.06, "learning_rate": 1.4167393199651266e-05, "loss": 0.5319, "step": 325 }, { "epoch": 0.06, "learning_rate": 1.4210985178727115e-05, "loss": 0.5192, "step": 326 }, { "epoch": 0.06, "learning_rate": 1.4254577157802964e-05, "loss": 0.512, "step": 327 }, { "epoch": 0.06, "learning_rate": 1.4298169136878813e-05, "loss": 0.5222, "step": 328 }, { "epoch": 0.06, "learning_rate": 1.4341761115954666e-05, "loss": 0.5108, "step": 329 }, { "epoch": 0.06, "learning_rate": 1.4385353095030516e-05, "loss": 0.5157, "step": 330 }, { "epoch": 0.06, "learning_rate": 1.4428945074106365e-05, "loss": 0.5267, "step": 331 }, { "epoch": 0.06, "learning_rate": 1.4472537053182214e-05, "loss": 0.5077, "step": 332 }, { "epoch": 0.06, "learning_rate": 1.4516129032258066e-05, "loss": 0.5198, "step": 333 }, { "epoch": 0.06, "learning_rate": 1.4559721011333915e-05, "loss": 0.5133, "step": 334 }, { "epoch": 0.06, "learning_rate": 1.4603312990409764e-05, "loss": 0.513, "step": 335 }, { "epoch": 0.06, "learning_rate": 1.4646904969485615e-05, "loss": 0.5317, "step": 336 }, { "epoch": 0.06, "learning_rate": 1.4690496948561466e-05, "loss": 0.5325, "step": 337 }, { "epoch": 0.06, "learning_rate": 1.4734088927637316e-05, "loss": 0.5281, "step": 338 }, { "epoch": 0.06, "learning_rate": 1.4777680906713165e-05, "loss": 0.5152, "step": 339 }, { "epoch": 0.06, "learning_rate": 1.4821272885789014e-05, "loss": 0.5132, "step": 340 }, { "epoch": 0.06, "learning_rate": 1.4864864864864867e-05, "loss": 0.5027, "step": 341 }, { "epoch": 0.06, "learning_rate": 1.4908456843940716e-05, "loss": 0.512, "step": 342 }, { "epoch": 0.06, "learning_rate": 1.4952048823016564e-05, "loss": 0.5251, "step": 343 }, { "epoch": 0.06, "learning_rate": 1.4995640802092415e-05, "loss": 0.4993, "step": 344 }, { "epoch": 0.06, "learning_rate": 1.5039232781168266e-05, "loss": 0.5451, "step": 345 }, { "epoch": 0.06, "learning_rate": 1.5082824760244116e-05, "loss": 0.5295, "step": 346 }, { "epoch": 0.06, "learning_rate": 1.5126416739319965e-05, "loss": 0.539, "step": 347 }, { "epoch": 0.06, "learning_rate": 1.5170008718395818e-05, "loss": 0.5138, "step": 348 }, { "epoch": 0.06, "learning_rate": 1.5213600697471667e-05, "loss": 0.5188, "step": 349 }, { "epoch": 0.06, "learning_rate": 1.5257192676547516e-05, "loss": 0.51, "step": 350 }, { "epoch": 0.06, "learning_rate": 1.5300784655623366e-05, "loss": 0.5069, "step": 351 }, { "epoch": 0.06, "learning_rate": 1.534437663469922e-05, "loss": 0.5112, "step": 352 }, { "epoch": 0.06, "learning_rate": 1.5387968613775068e-05, "loss": 0.522, "step": 353 }, { "epoch": 0.06, "learning_rate": 1.5431560592850917e-05, "loss": 0.53, "step": 354 }, { "epoch": 0.06, "learning_rate": 1.5475152571926766e-05, "loss": 0.5079, "step": 355 }, { "epoch": 0.06, "learning_rate": 1.5518744551002618e-05, "loss": 0.5223, "step": 356 }, { "epoch": 0.06, "learning_rate": 1.5562336530078467e-05, "loss": 0.5235, "step": 357 }, { "epoch": 0.06, "learning_rate": 1.5605928509154316e-05, "loss": 0.5205, "step": 358 }, { "epoch": 0.06, "learning_rate": 1.5649520488230165e-05, "loss": 0.5215, "step": 359 }, { "epoch": 0.06, "learning_rate": 1.5693112467306017e-05, "loss": 0.518, "step": 360 }, { "epoch": 0.06, "learning_rate": 1.5736704446381866e-05, "loss": 0.4888, "step": 361 }, { "epoch": 0.06, "learning_rate": 1.5780296425457715e-05, "loss": 0.5113, "step": 362 }, { "epoch": 0.06, "learning_rate": 1.5823888404533564e-05, "loss": 0.5062, "step": 363 }, { "epoch": 0.06, "learning_rate": 1.5867480383609416e-05, "loss": 0.5136, "step": 364 }, { "epoch": 0.06, "learning_rate": 1.591107236268527e-05, "loss": 0.5218, "step": 365 }, { "epoch": 0.06, "learning_rate": 1.5954664341761118e-05, "loss": 0.52, "step": 366 }, { "epoch": 0.06, "learning_rate": 1.5998256320836967e-05, "loss": 0.5295, "step": 367 }, { "epoch": 0.06, "learning_rate": 1.604184829991282e-05, "loss": 0.5049, "step": 368 }, { "epoch": 0.06, "learning_rate": 1.6085440278988668e-05, "loss": 0.5382, "step": 369 }, { "epoch": 0.06, "learning_rate": 1.6129032258064517e-05, "loss": 0.5237, "step": 370 }, { "epoch": 0.06, "learning_rate": 1.6172624237140366e-05, "loss": 0.5182, "step": 371 }, { "epoch": 0.06, "learning_rate": 1.6216216216216218e-05, "loss": 0.5182, "step": 372 }, { "epoch": 0.07, "learning_rate": 1.6259808195292067e-05, "loss": 0.5005, "step": 373 }, { "epoch": 0.07, "learning_rate": 1.6303400174367916e-05, "loss": 0.5236, "step": 374 }, { "epoch": 0.07, "learning_rate": 1.6346992153443765e-05, "loss": 0.5305, "step": 375 }, { "epoch": 0.07, "learning_rate": 1.6390584132519617e-05, "loss": 0.4941, "step": 376 }, { "epoch": 0.07, "learning_rate": 1.6434176111595466e-05, "loss": 0.5171, "step": 377 }, { "epoch": 0.07, "learning_rate": 1.6477768090671315e-05, "loss": 0.5108, "step": 378 }, { "epoch": 0.07, "learning_rate": 1.6521360069747168e-05, "loss": 0.5305, "step": 379 }, { "epoch": 0.07, "learning_rate": 1.6564952048823017e-05, "loss": 0.5094, "step": 380 }, { "epoch": 0.07, "learning_rate": 1.660854402789887e-05, "loss": 0.5187, "step": 381 }, { "epoch": 0.07, "learning_rate": 1.6652136006974718e-05, "loss": 0.5371, "step": 382 }, { "epoch": 0.07, "learning_rate": 1.669572798605057e-05, "loss": 0.5146, "step": 383 }, { "epoch": 0.07, "learning_rate": 1.673931996512642e-05, "loss": 0.5311, "step": 384 }, { "epoch": 0.07, "learning_rate": 1.6782911944202268e-05, "loss": 0.5209, "step": 385 }, { "epoch": 0.07, "learning_rate": 1.6826503923278117e-05, "loss": 0.5167, "step": 386 }, { "epoch": 0.07, "learning_rate": 1.687009590235397e-05, "loss": 0.5194, "step": 387 }, { "epoch": 0.07, "learning_rate": 1.691368788142982e-05, "loss": 0.5067, "step": 388 }, { "epoch": 0.07, "learning_rate": 1.6957279860505667e-05, "loss": 0.5133, "step": 389 }, { "epoch": 0.07, "learning_rate": 1.7000871839581516e-05, "loss": 0.497, "step": 390 }, { "epoch": 0.07, "learning_rate": 1.704446381865737e-05, "loss": 0.5263, "step": 391 }, { "epoch": 0.07, "learning_rate": 1.7088055797733218e-05, "loss": 0.4983, "step": 392 }, { "epoch": 0.07, "learning_rate": 1.7131647776809066e-05, "loss": 0.5097, "step": 393 }, { "epoch": 0.07, "learning_rate": 1.7175239755884915e-05, "loss": 0.5188, "step": 394 }, { "epoch": 0.07, "learning_rate": 1.7218831734960768e-05, "loss": 0.5204, "step": 395 }, { "epoch": 0.07, "learning_rate": 1.726242371403662e-05, "loss": 0.5095, "step": 396 }, { "epoch": 0.07, "learning_rate": 1.730601569311247e-05, "loss": 0.5117, "step": 397 }, { "epoch": 0.07, "learning_rate": 1.7349607672188318e-05, "loss": 0.5301, "step": 398 }, { "epoch": 0.07, "learning_rate": 1.739319965126417e-05, "loss": 0.4945, "step": 399 }, { "epoch": 0.07, "learning_rate": 1.743679163034002e-05, "loss": 0.5139, "step": 400 }, { "epoch": 0.07, "learning_rate": 1.7480383609415868e-05, "loss": 0.5168, "step": 401 }, { "epoch": 0.07, "learning_rate": 1.7523975588491717e-05, "loss": 0.5224, "step": 402 }, { "epoch": 0.07, "learning_rate": 1.756756756756757e-05, "loss": 0.511, "step": 403 }, { "epoch": 0.07, "learning_rate": 1.761115954664342e-05, "loss": 0.5132, "step": 404 }, { "epoch": 0.07, "learning_rate": 1.7654751525719268e-05, "loss": 0.5416, "step": 405 }, { "epoch": 0.07, "learning_rate": 1.7698343504795116e-05, "loss": 0.4799, "step": 406 }, { "epoch": 0.07, "learning_rate": 1.774193548387097e-05, "loss": 0.521, "step": 407 }, { "epoch": 0.07, "learning_rate": 1.7785527462946818e-05, "loss": 0.5171, "step": 408 }, { "epoch": 0.07, "learning_rate": 1.7829119442022667e-05, "loss": 0.5155, "step": 409 }, { "epoch": 0.07, "learning_rate": 1.787271142109852e-05, "loss": 0.5159, "step": 410 }, { "epoch": 0.07, "learning_rate": 1.7916303400174368e-05, "loss": 0.5382, "step": 411 }, { "epoch": 0.07, "learning_rate": 1.795989537925022e-05, "loss": 0.5154, "step": 412 }, { "epoch": 0.07, "learning_rate": 1.800348735832607e-05, "loss": 0.5132, "step": 413 }, { "epoch": 0.07, "learning_rate": 1.8047079337401918e-05, "loss": 0.5025, "step": 414 }, { "epoch": 0.07, "learning_rate": 1.809067131647777e-05, "loss": 0.5164, "step": 415 }, { "epoch": 0.07, "learning_rate": 1.813426329555362e-05, "loss": 0.5055, "step": 416 }, { "epoch": 0.07, "learning_rate": 1.817785527462947e-05, "loss": 0.5022, "step": 417 }, { "epoch": 0.07, "learning_rate": 1.822144725370532e-05, "loss": 0.488, "step": 418 }, { "epoch": 0.07, "learning_rate": 1.826503923278117e-05, "loss": 0.4882, "step": 419 }, { "epoch": 0.07, "learning_rate": 1.830863121185702e-05, "loss": 0.5124, "step": 420 }, { "epoch": 0.07, "learning_rate": 1.8352223190932868e-05, "loss": 0.4973, "step": 421 }, { "epoch": 0.07, "learning_rate": 1.839581517000872e-05, "loss": 0.5199, "step": 422 }, { "epoch": 0.07, "learning_rate": 1.843940714908457e-05, "loss": 0.5198, "step": 423 }, { "epoch": 0.07, "learning_rate": 1.8482999128160418e-05, "loss": 0.4944, "step": 424 }, { "epoch": 0.07, "learning_rate": 1.8526591107236267e-05, "loss": 0.525, "step": 425 }, { "epoch": 0.07, "learning_rate": 1.857018308631212e-05, "loss": 0.4916, "step": 426 }, { "epoch": 0.07, "learning_rate": 1.861377506538797e-05, "loss": 0.5157, "step": 427 }, { "epoch": 0.07, "learning_rate": 1.865736704446382e-05, "loss": 0.5233, "step": 428 }, { "epoch": 0.07, "learning_rate": 1.870095902353967e-05, "loss": 0.4915, "step": 429 }, { "epoch": 0.07, "learning_rate": 1.8744551002615522e-05, "loss": 0.5108, "step": 430 }, { "epoch": 0.08, "learning_rate": 1.878814298169137e-05, "loss": 0.5169, "step": 431 }, { "epoch": 0.08, "learning_rate": 1.883173496076722e-05, "loss": 0.5124, "step": 432 }, { "epoch": 0.08, "learning_rate": 1.887532693984307e-05, "loss": 0.4991, "step": 433 }, { "epoch": 0.08, "learning_rate": 1.891891891891892e-05, "loss": 0.4811, "step": 434 }, { "epoch": 0.08, "learning_rate": 1.896251089799477e-05, "loss": 0.5088, "step": 435 }, { "epoch": 0.08, "learning_rate": 1.900610287707062e-05, "loss": 0.5321, "step": 436 }, { "epoch": 0.08, "learning_rate": 1.9049694856146468e-05, "loss": 0.5268, "step": 437 }, { "epoch": 0.08, "learning_rate": 1.909328683522232e-05, "loss": 0.5146, "step": 438 }, { "epoch": 0.08, "learning_rate": 1.913687881429817e-05, "loss": 0.505, "step": 439 }, { "epoch": 0.08, "learning_rate": 1.9180470793374018e-05, "loss": 0.5072, "step": 440 }, { "epoch": 0.08, "learning_rate": 1.922406277244987e-05, "loss": 0.486, "step": 441 }, { "epoch": 0.08, "learning_rate": 1.926765475152572e-05, "loss": 0.5035, "step": 442 }, { "epoch": 0.08, "learning_rate": 1.9311246730601572e-05, "loss": 0.5194, "step": 443 }, { "epoch": 0.08, "learning_rate": 1.935483870967742e-05, "loss": 0.5008, "step": 444 }, { "epoch": 0.08, "learning_rate": 1.939843068875327e-05, "loss": 0.5019, "step": 445 }, { "epoch": 0.08, "learning_rate": 1.9442022667829122e-05, "loss": 0.5236, "step": 446 }, { "epoch": 0.08, "learning_rate": 1.948561464690497e-05, "loss": 0.5271, "step": 447 }, { "epoch": 0.08, "learning_rate": 1.952920662598082e-05, "loss": 0.4935, "step": 448 }, { "epoch": 0.08, "learning_rate": 1.957279860505667e-05, "loss": 0.5183, "step": 449 }, { "epoch": 0.08, "learning_rate": 1.961639058413252e-05, "loss": 0.5064, "step": 450 }, { "epoch": 0.08, "learning_rate": 1.965998256320837e-05, "loss": 0.505, "step": 451 }, { "epoch": 0.08, "learning_rate": 1.970357454228422e-05, "loss": 0.5246, "step": 452 }, { "epoch": 0.08, "learning_rate": 1.974716652136007e-05, "loss": 0.4921, "step": 453 }, { "epoch": 0.08, "learning_rate": 1.979075850043592e-05, "loss": 0.5076, "step": 454 }, { "epoch": 0.08, "learning_rate": 1.983435047951177e-05, "loss": 0.5027, "step": 455 }, { "epoch": 0.08, "learning_rate": 1.987794245858762e-05, "loss": 0.5127, "step": 456 }, { "epoch": 0.08, "learning_rate": 1.992153443766347e-05, "loss": 0.5078, "step": 457 }, { "epoch": 0.08, "learning_rate": 1.9965126416739323e-05, "loss": 0.5128, "step": 458 }, { "epoch": 0.08, "learning_rate": 2.0008718395815172e-05, "loss": 0.5047, "step": 459 }, { "epoch": 0.08, "learning_rate": 2.005231037489102e-05, "loss": 0.5018, "step": 460 }, { "epoch": 0.08, "learning_rate": 2.0095902353966873e-05, "loss": 0.4989, "step": 461 }, { "epoch": 0.08, "learning_rate": 2.0139494333042722e-05, "loss": 0.528, "step": 462 }, { "epoch": 0.08, "learning_rate": 2.018308631211857e-05, "loss": 0.4807, "step": 463 }, { "epoch": 0.08, "learning_rate": 2.022667829119442e-05, "loss": 0.5229, "step": 464 }, { "epoch": 0.08, "learning_rate": 2.0270270270270273e-05, "loss": 0.5, "step": 465 }, { "epoch": 0.08, "learning_rate": 2.031386224934612e-05, "loss": 0.5207, "step": 466 }, { "epoch": 0.08, "learning_rate": 2.035745422842197e-05, "loss": 0.4954, "step": 467 }, { "epoch": 0.08, "learning_rate": 2.040104620749782e-05, "loss": 0.4987, "step": 468 }, { "epoch": 0.08, "learning_rate": 2.0444638186573672e-05, "loss": 0.5174, "step": 469 }, { "epoch": 0.08, "learning_rate": 2.048823016564952e-05, "loss": 0.4823, "step": 470 }, { "epoch": 0.08, "learning_rate": 2.053182214472537e-05, "loss": 0.4917, "step": 471 }, { "epoch": 0.08, "learning_rate": 2.0575414123801222e-05, "loss": 0.498, "step": 472 }, { "epoch": 0.08, "learning_rate": 2.061900610287707e-05, "loss": 0.5522, "step": 473 }, { "epoch": 0.08, "learning_rate": 2.0662598081952923e-05, "loss": 0.5163, "step": 474 }, { "epoch": 0.08, "learning_rate": 2.0706190061028772e-05, "loss": 0.5068, "step": 475 }, { "epoch": 0.08, "learning_rate": 2.074978204010462e-05, "loss": 0.4949, "step": 476 }, { "epoch": 0.08, "learning_rate": 2.0793374019180474e-05, "loss": 0.5047, "step": 477 }, { "epoch": 0.08, "learning_rate": 2.0836965998256323e-05, "loss": 0.5139, "step": 478 }, { "epoch": 0.08, "learning_rate": 2.088055797733217e-05, "loss": 0.5033, "step": 479 }, { "epoch": 0.08, "learning_rate": 2.092414995640802e-05, "loss": 0.5094, "step": 480 }, { "epoch": 0.08, "learning_rate": 2.0967741935483873e-05, "loss": 0.4882, "step": 481 }, { "epoch": 0.08, "learning_rate": 2.1011333914559722e-05, "loss": 0.5017, "step": 482 }, { "epoch": 0.08, "learning_rate": 2.105492589363557e-05, "loss": 0.5, "step": 483 }, { "epoch": 0.08, "learning_rate": 2.109851787271142e-05, "loss": 0.5026, "step": 484 }, { "epoch": 0.08, "learning_rate": 2.1142109851787272e-05, "loss": 0.5032, "step": 485 }, { "epoch": 0.08, "learning_rate": 2.118570183086312e-05, "loss": 0.4983, "step": 486 }, { "epoch": 0.08, "learning_rate": 2.122929380993897e-05, "loss": 0.5057, "step": 487 }, { "epoch": 0.09, "learning_rate": 2.1272885789014822e-05, "loss": 0.5078, "step": 488 }, { "epoch": 0.09, "learning_rate": 2.1316477768090675e-05, "loss": 0.5016, "step": 489 }, { "epoch": 0.09, "learning_rate": 2.1360069747166524e-05, "loss": 0.4966, "step": 490 }, { "epoch": 0.09, "learning_rate": 2.1403661726242373e-05, "loss": 0.496, "step": 491 }, { "epoch": 0.09, "learning_rate": 2.1447253705318225e-05, "loss": 0.5209, "step": 492 }, { "epoch": 0.09, "learning_rate": 2.1490845684394074e-05, "loss": 0.5051, "step": 493 }, { "epoch": 0.09, "learning_rate": 2.1534437663469923e-05, "loss": 0.5169, "step": 494 }, { "epoch": 0.09, "learning_rate": 2.1578029642545772e-05, "loss": 0.5066, "step": 495 }, { "epoch": 0.09, "learning_rate": 2.1621621621621624e-05, "loss": 0.4974, "step": 496 }, { "epoch": 0.09, "learning_rate": 2.1665213600697473e-05, "loss": 0.507, "step": 497 }, { "epoch": 0.09, "learning_rate": 2.1708805579773322e-05, "loss": 0.4981, "step": 498 }, { "epoch": 0.09, "learning_rate": 2.175239755884917e-05, "loss": 0.5028, "step": 499 }, { "epoch": 0.09, "learning_rate": 2.1795989537925023e-05, "loss": 0.5067, "step": 500 }, { "epoch": 0.09, "learning_rate": 2.1839581517000872e-05, "loss": 0.5168, "step": 501 }, { "epoch": 0.09, "learning_rate": 2.188317349607672e-05, "loss": 0.5146, "step": 502 }, { "epoch": 0.09, "learning_rate": 2.1926765475152574e-05, "loss": 0.5154, "step": 503 }, { "epoch": 0.09, "learning_rate": 2.1970357454228423e-05, "loss": 0.5018, "step": 504 }, { "epoch": 0.09, "learning_rate": 2.2013949433304275e-05, "loss": 0.5003, "step": 505 }, { "epoch": 0.09, "learning_rate": 2.2057541412380124e-05, "loss": 0.5261, "step": 506 }, { "epoch": 0.09, "learning_rate": 2.2101133391455973e-05, "loss": 0.4799, "step": 507 }, { "epoch": 0.09, "learning_rate": 2.2144725370531825e-05, "loss": 0.4962, "step": 508 }, { "epoch": 0.09, "learning_rate": 2.2188317349607674e-05, "loss": 0.5083, "step": 509 }, { "epoch": 0.09, "learning_rate": 2.2231909328683523e-05, "loss": 0.5101, "step": 510 }, { "epoch": 0.09, "learning_rate": 2.2275501307759372e-05, "loss": 0.5006, "step": 511 }, { "epoch": 0.09, "learning_rate": 2.2319093286835224e-05, "loss": 0.4693, "step": 512 }, { "epoch": 0.09, "learning_rate": 2.2362685265911073e-05, "loss": 0.4946, "step": 513 }, { "epoch": 0.09, "learning_rate": 2.2406277244986922e-05, "loss": 0.4842, "step": 514 }, { "epoch": 0.09, "learning_rate": 2.244986922406277e-05, "loss": 0.4933, "step": 515 }, { "epoch": 0.09, "learning_rate": 2.2493461203138624e-05, "loss": 0.5127, "step": 516 }, { "epoch": 0.09, "learning_rate": 2.2537053182214473e-05, "loss": 0.5138, "step": 517 }, { "epoch": 0.09, "learning_rate": 2.258064516129032e-05, "loss": 0.4934, "step": 518 }, { "epoch": 0.09, "learning_rate": 2.2624237140366174e-05, "loss": 0.4981, "step": 519 }, { "epoch": 0.09, "learning_rate": 2.2667829119442026e-05, "loss": 0.4838, "step": 520 }, { "epoch": 0.09, "learning_rate": 2.2711421098517875e-05, "loss": 0.4913, "step": 521 }, { "epoch": 0.09, "learning_rate": 2.2755013077593724e-05, "loss": 0.4985, "step": 522 }, { "epoch": 0.09, "learning_rate": 2.2798605056669576e-05, "loss": 0.5004, "step": 523 }, { "epoch": 0.09, "learning_rate": 2.2842197035745425e-05, "loss": 0.5026, "step": 524 }, { "epoch": 0.09, "learning_rate": 2.2885789014821274e-05, "loss": 0.4928, "step": 525 }, { "epoch": 0.09, "learning_rate": 2.2929380993897123e-05, "loss": 0.4982, "step": 526 }, { "epoch": 0.09, "learning_rate": 2.2972972972972976e-05, "loss": 0.5119, "step": 527 }, { "epoch": 0.09, "learning_rate": 2.3016564952048825e-05, "loss": 0.51, "step": 528 }, { "epoch": 0.09, "learning_rate": 2.3060156931124674e-05, "loss": 0.5111, "step": 529 }, { "epoch": 0.09, "learning_rate": 2.3103748910200522e-05, "loss": 0.4928, "step": 530 }, { "epoch": 0.09, "learning_rate": 2.3147340889276375e-05, "loss": 0.5134, "step": 531 }, { "epoch": 0.09, "learning_rate": 2.3190932868352224e-05, "loss": 0.4978, "step": 532 }, { "epoch": 0.09, "learning_rate": 2.3234524847428073e-05, "loss": 0.4751, "step": 533 }, { "epoch": 0.09, "learning_rate": 2.3278116826503925e-05, "loss": 0.5057, "step": 534 }, { "epoch": 0.09, "learning_rate": 2.3321708805579774e-05, "loss": 0.5046, "step": 535 }, { "epoch": 0.09, "learning_rate": 2.3365300784655626e-05, "loss": 0.5, "step": 536 }, { "epoch": 0.09, "learning_rate": 2.3408892763731475e-05, "loss": 0.5053, "step": 537 }, { "epoch": 0.09, "learning_rate": 2.3452484742807324e-05, "loss": 0.4929, "step": 538 }, { "epoch": 0.09, "learning_rate": 2.3496076721883177e-05, "loss": 0.5072, "step": 539 }, { "epoch": 0.09, "learning_rate": 2.3539668700959026e-05, "loss": 0.5115, "step": 540 }, { "epoch": 0.09, "learning_rate": 2.3583260680034875e-05, "loss": 0.4898, "step": 541 }, { "epoch": 0.09, "learning_rate": 2.3626852659110724e-05, "loss": 0.4886, "step": 542 }, { "epoch": 0.09, "learning_rate": 2.3670444638186576e-05, "loss": 0.5046, "step": 543 }, { "epoch": 0.09, "learning_rate": 2.3714036617262425e-05, "loss": 0.5032, "step": 544 }, { "epoch": 0.1, "learning_rate": 2.3757628596338274e-05, "loss": 0.5028, "step": 545 }, { "epoch": 0.1, "learning_rate": 2.3801220575414123e-05, "loss": 0.495, "step": 546 }, { "epoch": 0.1, "learning_rate": 2.3844812554489975e-05, "loss": 0.4966, "step": 547 }, { "epoch": 0.1, "learning_rate": 2.3888404533565824e-05, "loss": 0.5113, "step": 548 }, { "epoch": 0.1, "learning_rate": 2.3931996512641673e-05, "loss": 0.4784, "step": 549 }, { "epoch": 0.1, "learning_rate": 2.3975588491717525e-05, "loss": 0.5038, "step": 550 }, { "epoch": 0.1, "learning_rate": 2.4019180470793374e-05, "loss": 0.4878, "step": 551 }, { "epoch": 0.1, "learning_rate": 2.4062772449869227e-05, "loss": 0.481, "step": 552 }, { "epoch": 0.1, "learning_rate": 2.4106364428945076e-05, "loss": 0.5046, "step": 553 }, { "epoch": 0.1, "learning_rate": 2.4149956408020925e-05, "loss": 0.518, "step": 554 }, { "epoch": 0.1, "learning_rate": 2.4193548387096777e-05, "loss": 0.5045, "step": 555 }, { "epoch": 0.1, "learning_rate": 2.4237140366172626e-05, "loss": 0.4874, "step": 556 }, { "epoch": 0.1, "learning_rate": 2.4280732345248475e-05, "loss": 0.5083, "step": 557 }, { "epoch": 0.1, "learning_rate": 2.4324324324324327e-05, "loss": 0.491, "step": 558 }, { "epoch": 0.1, "learning_rate": 2.4367916303400176e-05, "loss": 0.5107, "step": 559 }, { "epoch": 0.1, "learning_rate": 2.4411508282476025e-05, "loss": 0.4946, "step": 560 }, { "epoch": 0.1, "learning_rate": 2.4455100261551874e-05, "loss": 0.4905, "step": 561 }, { "epoch": 0.1, "learning_rate": 2.4498692240627726e-05, "loss": 0.4937, "step": 562 }, { "epoch": 0.1, "learning_rate": 2.4542284219703575e-05, "loss": 0.4955, "step": 563 }, { "epoch": 0.1, "learning_rate": 2.4585876198779424e-05, "loss": 0.4861, "step": 564 }, { "epoch": 0.1, "learning_rate": 2.4629468177855277e-05, "loss": 0.495, "step": 565 }, { "epoch": 0.1, "learning_rate": 2.4673060156931126e-05, "loss": 0.5155, "step": 566 }, { "epoch": 0.1, "learning_rate": 2.4716652136006978e-05, "loss": 0.4882, "step": 567 }, { "epoch": 0.1, "learning_rate": 2.4760244115082827e-05, "loss": 0.4915, "step": 568 }, { "epoch": 0.1, "learning_rate": 2.4803836094158676e-05, "loss": 0.4942, "step": 569 }, { "epoch": 0.1, "learning_rate": 2.4847428073234528e-05, "loss": 0.5024, "step": 570 }, { "epoch": 0.1, "learning_rate": 2.4891020052310377e-05, "loss": 0.4971, "step": 571 }, { "epoch": 0.1, "learning_rate": 2.4934612031386226e-05, "loss": 0.4865, "step": 572 }, { "epoch": 0.1, "learning_rate": 2.4978204010462075e-05, "loss": 0.504, "step": 573 }, { "epoch": 0.1, "learning_rate": 2.5021795989537927e-05, "loss": 0.5027, "step": 574 }, { "epoch": 0.1, "learning_rate": 2.5065387968613773e-05, "loss": 0.4785, "step": 575 }, { "epoch": 0.1, "learning_rate": 2.5108979947689625e-05, "loss": 0.4974, "step": 576 }, { "epoch": 0.1, "learning_rate": 2.5152571926765478e-05, "loss": 0.5063, "step": 577 }, { "epoch": 0.1, "learning_rate": 2.5196163905841323e-05, "loss": 0.484, "step": 578 }, { "epoch": 0.1, "learning_rate": 2.5239755884917176e-05, "loss": 0.4705, "step": 579 }, { "epoch": 0.1, "learning_rate": 2.5283347863993028e-05, "loss": 0.498, "step": 580 }, { "epoch": 0.1, "learning_rate": 2.5326939843068877e-05, "loss": 0.495, "step": 581 }, { "epoch": 0.1, "learning_rate": 2.5370531822144726e-05, "loss": 0.5085, "step": 582 }, { "epoch": 0.1, "learning_rate": 2.5414123801220575e-05, "loss": 0.5047, "step": 583 }, { "epoch": 0.1, "learning_rate": 2.5457715780296427e-05, "loss": 0.5026, "step": 584 }, { "epoch": 0.1, "learning_rate": 2.550130775937228e-05, "loss": 0.488, "step": 585 }, { "epoch": 0.1, "learning_rate": 2.5544899738448125e-05, "loss": 0.4996, "step": 586 }, { "epoch": 0.1, "learning_rate": 2.5588491717523977e-05, "loss": 0.4854, "step": 587 }, { "epoch": 0.1, "learning_rate": 2.563208369659983e-05, "loss": 0.4855, "step": 588 }, { "epoch": 0.1, "learning_rate": 2.5675675675675675e-05, "loss": 0.5192, "step": 589 }, { "epoch": 0.1, "learning_rate": 2.5719267654751528e-05, "loss": 0.4956, "step": 590 }, { "epoch": 0.1, "learning_rate": 2.576285963382738e-05, "loss": 0.4908, "step": 591 }, { "epoch": 0.1, "learning_rate": 2.5806451612903226e-05, "loss": 0.4932, "step": 592 }, { "epoch": 0.1, "learning_rate": 2.5850043591979078e-05, "loss": 0.4993, "step": 593 }, { "epoch": 0.1, "learning_rate": 2.5893635571054923e-05, "loss": 0.4783, "step": 594 }, { "epoch": 0.1, "learning_rate": 2.5937227550130776e-05, "loss": 0.4882, "step": 595 }, { "epoch": 0.1, "learning_rate": 2.5980819529206628e-05, "loss": 0.489, "step": 596 }, { "epoch": 0.1, "learning_rate": 2.6024411508282477e-05, "loss": 0.4823, "step": 597 }, { "epoch": 0.1, "learning_rate": 2.606800348735833e-05, "loss": 0.4798, "step": 598 }, { "epoch": 0.1, "learning_rate": 2.611159546643418e-05, "loss": 0.5048, "step": 599 }, { "epoch": 0.1, "learning_rate": 2.6155187445510027e-05, "loss": 0.4863, "step": 600 }, { "epoch": 0.1, "learning_rate": 2.619877942458588e-05, "loss": 0.4847, "step": 601 }, { "epoch": 0.1, "learning_rate": 2.6242371403661725e-05, "loss": 0.4987, "step": 602 }, { "epoch": 0.11, "learning_rate": 2.6285963382737578e-05, "loss": 0.4708, "step": 603 }, { "epoch": 0.11, "learning_rate": 2.632955536181343e-05, "loss": 0.487, "step": 604 }, { "epoch": 0.11, "learning_rate": 2.6373147340889275e-05, "loss": 0.4899, "step": 605 }, { "epoch": 0.11, "learning_rate": 2.6416739319965128e-05, "loss": 0.5056, "step": 606 }, { "epoch": 0.11, "learning_rate": 2.646033129904098e-05, "loss": 0.5024, "step": 607 }, { "epoch": 0.11, "learning_rate": 2.6503923278116826e-05, "loss": 0.4916, "step": 608 }, { "epoch": 0.11, "learning_rate": 2.6547515257192678e-05, "loss": 0.4965, "step": 609 }, { "epoch": 0.11, "learning_rate": 2.6591107236268527e-05, "loss": 0.4939, "step": 610 }, { "epoch": 0.11, "learning_rate": 2.6634699215344376e-05, "loss": 0.4848, "step": 611 }, { "epoch": 0.11, "learning_rate": 2.667829119442023e-05, "loss": 0.4805, "step": 612 }, { "epoch": 0.11, "learning_rate": 2.6721883173496077e-05, "loss": 0.501, "step": 613 }, { "epoch": 0.11, "learning_rate": 2.676547515257193e-05, "loss": 0.4953, "step": 614 }, { "epoch": 0.11, "learning_rate": 2.6809067131647782e-05, "loss": 0.4845, "step": 615 }, { "epoch": 0.11, "learning_rate": 2.6852659110723628e-05, "loss": 0.5126, "step": 616 }, { "epoch": 0.11, "learning_rate": 2.689625108979948e-05, "loss": 0.4902, "step": 617 }, { "epoch": 0.11, "learning_rate": 2.6939843068875325e-05, "loss": 0.4855, "step": 618 }, { "epoch": 0.11, "learning_rate": 2.6983435047951178e-05, "loss": 0.4857, "step": 619 }, { "epoch": 0.11, "learning_rate": 2.702702702702703e-05, "loss": 0.5033, "step": 620 }, { "epoch": 0.11, "learning_rate": 2.7070619006102876e-05, "loss": 0.4952, "step": 621 }, { "epoch": 0.11, "learning_rate": 2.7114210985178728e-05, "loss": 0.482, "step": 622 }, { "epoch": 0.11, "learning_rate": 2.715780296425458e-05, "loss": 0.499, "step": 623 }, { "epoch": 0.11, "learning_rate": 2.7201394943330426e-05, "loss": 0.5029, "step": 624 }, { "epoch": 0.11, "learning_rate": 2.724498692240628e-05, "loss": 0.4856, "step": 625 }, { "epoch": 0.11, "learning_rate": 2.728857890148213e-05, "loss": 0.4929, "step": 626 }, { "epoch": 0.11, "learning_rate": 2.7332170880557976e-05, "loss": 0.5002, "step": 627 }, { "epoch": 0.11, "learning_rate": 2.737576285963383e-05, "loss": 0.4979, "step": 628 }, { "epoch": 0.11, "learning_rate": 2.7419354838709678e-05, "loss": 0.463, "step": 629 }, { "epoch": 0.11, "learning_rate": 2.746294681778553e-05, "loss": 0.4806, "step": 630 }, { "epoch": 0.11, "learning_rate": 2.7506538796861382e-05, "loss": 0.4866, "step": 631 }, { "epoch": 0.11, "learning_rate": 2.7550130775937228e-05, "loss": 0.4757, "step": 632 }, { "epoch": 0.11, "learning_rate": 2.759372275501308e-05, "loss": 0.4815, "step": 633 }, { "epoch": 0.11, "learning_rate": 2.7637314734088932e-05, "loss": 0.4987, "step": 634 }, { "epoch": 0.11, "learning_rate": 2.7680906713164778e-05, "loss": 0.4988, "step": 635 }, { "epoch": 0.11, "learning_rate": 2.772449869224063e-05, "loss": 0.4965, "step": 636 }, { "epoch": 0.11, "learning_rate": 2.7768090671316476e-05, "loss": 0.5015, "step": 637 }, { "epoch": 0.11, "learning_rate": 2.7811682650392328e-05, "loss": 0.4764, "step": 638 }, { "epoch": 0.11, "learning_rate": 2.785527462946818e-05, "loss": 0.488, "step": 639 }, { "epoch": 0.11, "learning_rate": 2.7898866608544026e-05, "loss": 0.4698, "step": 640 }, { "epoch": 0.11, "learning_rate": 2.794245858761988e-05, "loss": 0.4895, "step": 641 }, { "epoch": 0.11, "learning_rate": 2.798605056669573e-05, "loss": 0.4906, "step": 642 }, { "epoch": 0.11, "learning_rate": 2.802964254577158e-05, "loss": 0.5091, "step": 643 }, { "epoch": 0.11, "learning_rate": 2.807323452484743e-05, "loss": 0.4739, "step": 644 }, { "epoch": 0.11, "learning_rate": 2.8116826503923278e-05, "loss": 0.4815, "step": 645 }, { "epoch": 0.11, "learning_rate": 2.816041848299913e-05, "loss": 0.4922, "step": 646 }, { "epoch": 0.11, "learning_rate": 2.8204010462074982e-05, "loss": 0.5036, "step": 647 }, { "epoch": 0.11, "learning_rate": 2.8247602441150828e-05, "loss": 0.4759, "step": 648 }, { "epoch": 0.11, "learning_rate": 2.829119442022668e-05, "loss": 0.5055, "step": 649 }, { "epoch": 0.11, "learning_rate": 2.8334786399302533e-05, "loss": 0.4784, "step": 650 }, { "epoch": 0.11, "learning_rate": 2.8378378378378378e-05, "loss": 0.4925, "step": 651 }, { "epoch": 0.11, "learning_rate": 2.842197035745423e-05, "loss": 0.4838, "step": 652 }, { "epoch": 0.11, "learning_rate": 2.8465562336530076e-05, "loss": 0.4611, "step": 653 }, { "epoch": 0.11, "learning_rate": 2.850915431560593e-05, "loss": 0.5047, "step": 654 }, { "epoch": 0.11, "learning_rate": 2.855274629468178e-05, "loss": 0.4951, "step": 655 }, { "epoch": 0.11, "learning_rate": 2.8596338273757626e-05, "loss": 0.5003, "step": 656 }, { "epoch": 0.11, "learning_rate": 2.863993025283348e-05, "loss": 0.4964, "step": 657 }, { "epoch": 0.11, "learning_rate": 2.868352223190933e-05, "loss": 0.4781, "step": 658 }, { "epoch": 0.11, "learning_rate": 2.872711421098518e-05, "loss": 0.5199, "step": 659 }, { "epoch": 0.12, "learning_rate": 2.8770706190061032e-05, "loss": 0.4784, "step": 660 }, { "epoch": 0.12, "learning_rate": 2.881429816913688e-05, "loss": 0.4818, "step": 661 }, { "epoch": 0.12, "learning_rate": 2.885789014821273e-05, "loss": 0.4853, "step": 662 }, { "epoch": 0.12, "learning_rate": 2.8901482127288583e-05, "loss": 0.476, "step": 663 }, { "epoch": 0.12, "learning_rate": 2.8945074106364428e-05, "loss": 0.4994, "step": 664 }, { "epoch": 0.12, "learning_rate": 2.898866608544028e-05, "loss": 0.5029, "step": 665 }, { "epoch": 0.12, "learning_rate": 2.9032258064516133e-05, "loss": 0.4896, "step": 666 }, { "epoch": 0.12, "learning_rate": 2.907585004359198e-05, "loss": 0.4731, "step": 667 }, { "epoch": 0.12, "learning_rate": 2.911944202266783e-05, "loss": 0.5007, "step": 668 }, { "epoch": 0.12, "learning_rate": 2.9163034001743683e-05, "loss": 0.4937, "step": 669 }, { "epoch": 0.12, "learning_rate": 2.920662598081953e-05, "loss": 0.4754, "step": 670 }, { "epoch": 0.12, "learning_rate": 2.925021795989538e-05, "loss": 0.4788, "step": 671 }, { "epoch": 0.12, "learning_rate": 2.929380993897123e-05, "loss": 0.508, "step": 672 }, { "epoch": 0.12, "learning_rate": 2.933740191804708e-05, "loss": 0.4844, "step": 673 }, { "epoch": 0.12, "learning_rate": 2.938099389712293e-05, "loss": 0.4863, "step": 674 }, { "epoch": 0.12, "learning_rate": 2.942458587619878e-05, "loss": 0.4808, "step": 675 }, { "epoch": 0.12, "learning_rate": 2.9468177855274633e-05, "loss": 0.4693, "step": 676 }, { "epoch": 0.12, "learning_rate": 2.9511769834350485e-05, "loss": 0.4984, "step": 677 }, { "epoch": 0.12, "learning_rate": 2.955536181342633e-05, "loss": 0.4984, "step": 678 }, { "epoch": 0.12, "learning_rate": 2.9598953792502183e-05, "loss": 0.4838, "step": 679 }, { "epoch": 0.12, "learning_rate": 2.964254577157803e-05, "loss": 0.4758, "step": 680 }, { "epoch": 0.12, "learning_rate": 2.968613775065388e-05, "loss": 0.5045, "step": 681 }, { "epoch": 0.12, "learning_rate": 2.9729729729729733e-05, "loss": 0.4946, "step": 682 }, { "epoch": 0.12, "learning_rate": 2.977332170880558e-05, "loss": 0.4909, "step": 683 }, { "epoch": 0.12, "learning_rate": 2.981691368788143e-05, "loss": 0.4935, "step": 684 }, { "epoch": 0.12, "learning_rate": 2.9860505666957283e-05, "loss": 0.4858, "step": 685 }, { "epoch": 0.12, "learning_rate": 2.990409764603313e-05, "loss": 0.5065, "step": 686 }, { "epoch": 0.12, "learning_rate": 2.994768962510898e-05, "loss": 0.5004, "step": 687 }, { "epoch": 0.12, "learning_rate": 2.999128160418483e-05, "loss": 0.4895, "step": 688 }, { "epoch": 0.12, "learning_rate": 3.003487358326068e-05, "loss": 0.4757, "step": 689 }, { "epoch": 0.12, "learning_rate": 3.007846556233653e-05, "loss": 0.502, "step": 690 }, { "epoch": 0.12, "learning_rate": 3.012205754141238e-05, "loss": 0.4736, "step": 691 }, { "epoch": 0.12, "learning_rate": 3.0165649520488233e-05, "loss": 0.4952, "step": 692 }, { "epoch": 0.12, "learning_rate": 3.0209241499564085e-05, "loss": 0.4946, "step": 693 }, { "epoch": 0.12, "learning_rate": 3.025283347863993e-05, "loss": 0.476, "step": 694 }, { "epoch": 0.12, "learning_rate": 3.0296425457715783e-05, "loss": 0.4936, "step": 695 }, { "epoch": 0.12, "learning_rate": 3.0340017436791635e-05, "loss": 0.4805, "step": 696 }, { "epoch": 0.12, "learning_rate": 3.038360941586748e-05, "loss": 0.4766, "step": 697 }, { "epoch": 0.12, "learning_rate": 3.0427201394943333e-05, "loss": 0.4828, "step": 698 }, { "epoch": 0.12, "learning_rate": 3.047079337401918e-05, "loss": 0.5034, "step": 699 }, { "epoch": 0.12, "learning_rate": 3.051438535309503e-05, "loss": 0.4878, "step": 700 }, { "epoch": 0.12, "learning_rate": 3.055797733217089e-05, "loss": 0.4795, "step": 701 }, { "epoch": 0.12, "learning_rate": 3.060156931124673e-05, "loss": 0.4933, "step": 702 }, { "epoch": 0.12, "learning_rate": 3.0645161290322585e-05, "loss": 0.4933, "step": 703 }, { "epoch": 0.12, "learning_rate": 3.068875326939844e-05, "loss": 0.5029, "step": 704 }, { "epoch": 0.12, "learning_rate": 3.073234524847428e-05, "loss": 0.5005, "step": 705 }, { "epoch": 0.12, "learning_rate": 3.0775937227550135e-05, "loss": 0.4773, "step": 706 }, { "epoch": 0.12, "learning_rate": 3.081952920662598e-05, "loss": 0.5095, "step": 707 }, { "epoch": 0.12, "learning_rate": 3.086312118570183e-05, "loss": 0.5012, "step": 708 }, { "epoch": 0.12, "learning_rate": 3.0906713164777685e-05, "loss": 0.4908, "step": 709 }, { "epoch": 0.12, "learning_rate": 3.095030514385353e-05, "loss": 0.4749, "step": 710 }, { "epoch": 0.12, "learning_rate": 3.099389712292938e-05, "loss": 0.4931, "step": 711 }, { "epoch": 0.12, "learning_rate": 3.1037489102005236e-05, "loss": 0.4902, "step": 712 }, { "epoch": 0.12, "learning_rate": 3.108108108108108e-05, "loss": 0.4802, "step": 713 }, { "epoch": 0.12, "learning_rate": 3.1124673060156934e-05, "loss": 0.5025, "step": 714 }, { "epoch": 0.12, "learning_rate": 3.116826503923278e-05, "loss": 0.4884, "step": 715 }, { "epoch": 0.12, "learning_rate": 3.121185701830863e-05, "loss": 0.4751, "step": 716 }, { "epoch": 0.13, "learning_rate": 3.1255448997384484e-05, "loss": 0.4863, "step": 717 }, { "epoch": 0.13, "learning_rate": 3.129904097646033e-05, "loss": 0.4844, "step": 718 }, { "epoch": 0.13, "learning_rate": 3.134263295553618e-05, "loss": 0.4799, "step": 719 }, { "epoch": 0.13, "learning_rate": 3.1386224934612034e-05, "loss": 0.4844, "step": 720 }, { "epoch": 0.13, "learning_rate": 3.142981691368788e-05, "loss": 0.4992, "step": 721 }, { "epoch": 0.13, "learning_rate": 3.147340889276373e-05, "loss": 0.4921, "step": 722 }, { "epoch": 0.13, "learning_rate": 3.151700087183958e-05, "loss": 0.4856, "step": 723 }, { "epoch": 0.13, "learning_rate": 3.156059285091543e-05, "loss": 0.5064, "step": 724 }, { "epoch": 0.13, "learning_rate": 3.160418482999128e-05, "loss": 0.4793, "step": 725 }, { "epoch": 0.13, "learning_rate": 3.164777680906713e-05, "loss": 0.4914, "step": 726 }, { "epoch": 0.13, "learning_rate": 3.169136878814298e-05, "loss": 0.4941, "step": 727 }, { "epoch": 0.13, "learning_rate": 3.173496076721883e-05, "loss": 0.493, "step": 728 }, { "epoch": 0.13, "learning_rate": 3.1778552746294685e-05, "loss": 0.4942, "step": 729 }, { "epoch": 0.13, "learning_rate": 3.182214472537054e-05, "loss": 0.4589, "step": 730 }, { "epoch": 0.13, "learning_rate": 3.186573670444638e-05, "loss": 0.4992, "step": 731 }, { "epoch": 0.13, "learning_rate": 3.1909328683522235e-05, "loss": 0.5064, "step": 732 }, { "epoch": 0.13, "learning_rate": 3.195292066259809e-05, "loss": 0.4979, "step": 733 }, { "epoch": 0.13, "learning_rate": 3.199651264167393e-05, "loss": 0.4792, "step": 734 }, { "epoch": 0.13, "learning_rate": 3.2040104620749785e-05, "loss": 0.4939, "step": 735 }, { "epoch": 0.13, "learning_rate": 3.208369659982564e-05, "loss": 0.5045, "step": 736 }, { "epoch": 0.13, "learning_rate": 3.212728857890148e-05, "loss": 0.5015, "step": 737 }, { "epoch": 0.13, "learning_rate": 3.2170880557977336e-05, "loss": 0.4745, "step": 738 }, { "epoch": 0.13, "learning_rate": 3.221447253705319e-05, "loss": 0.4977, "step": 739 }, { "epoch": 0.13, "learning_rate": 3.2258064516129034e-05, "loss": 0.4683, "step": 740 }, { "epoch": 0.13, "learning_rate": 3.2301656495204886e-05, "loss": 0.4796, "step": 741 }, { "epoch": 0.13, "learning_rate": 3.234524847428073e-05, "loss": 0.5053, "step": 742 }, { "epoch": 0.13, "learning_rate": 3.2388840453356584e-05, "loss": 0.4982, "step": 743 }, { "epoch": 0.13, "learning_rate": 3.2432432432432436e-05, "loss": 0.487, "step": 744 }, { "epoch": 0.13, "learning_rate": 3.247602441150828e-05, "loss": 0.4928, "step": 745 }, { "epoch": 0.13, "learning_rate": 3.2519616390584134e-05, "loss": 0.5071, "step": 746 }, { "epoch": 0.13, "learning_rate": 3.2563208369659986e-05, "loss": 0.484, "step": 747 }, { "epoch": 0.13, "learning_rate": 3.260680034873583e-05, "loss": 0.4881, "step": 748 }, { "epoch": 0.13, "learning_rate": 3.2650392327811684e-05, "loss": 0.4856, "step": 749 }, { "epoch": 0.13, "learning_rate": 3.269398430688753e-05, "loss": 0.4813, "step": 750 }, { "epoch": 0.13, "learning_rate": 3.273757628596338e-05, "loss": 0.4803, "step": 751 }, { "epoch": 0.13, "learning_rate": 3.2781168265039235e-05, "loss": 0.4867, "step": 752 }, { "epoch": 0.13, "learning_rate": 3.282476024411508e-05, "loss": 0.4758, "step": 753 }, { "epoch": 0.13, "learning_rate": 3.286835222319093e-05, "loss": 0.4851, "step": 754 }, { "epoch": 0.13, "learning_rate": 3.2911944202266785e-05, "loss": 0.5059, "step": 755 }, { "epoch": 0.13, "learning_rate": 3.295553618134263e-05, "loss": 0.4994, "step": 756 }, { "epoch": 0.13, "learning_rate": 3.299912816041848e-05, "loss": 0.4935, "step": 757 }, { "epoch": 0.13, "learning_rate": 3.3042720139494335e-05, "loss": 0.5039, "step": 758 }, { "epoch": 0.13, "learning_rate": 3.308631211857019e-05, "loss": 0.5003, "step": 759 }, { "epoch": 0.13, "learning_rate": 3.312990409764603e-05, "loss": 0.4922, "step": 760 }, { "epoch": 0.13, "learning_rate": 3.3173496076721885e-05, "loss": 0.4867, "step": 761 }, { "epoch": 0.13, "learning_rate": 3.321708805579774e-05, "loss": 0.4948, "step": 762 }, { "epoch": 0.13, "learning_rate": 3.326068003487359e-05, "loss": 0.484, "step": 763 }, { "epoch": 0.13, "learning_rate": 3.3304272013949436e-05, "loss": 0.4786, "step": 764 }, { "epoch": 0.13, "learning_rate": 3.334786399302529e-05, "loss": 0.4839, "step": 765 }, { "epoch": 0.13, "learning_rate": 3.339145597210114e-05, "loss": 0.4902, "step": 766 }, { "epoch": 0.13, "learning_rate": 3.3435047951176986e-05, "loss": 0.4976, "step": 767 }, { "epoch": 0.13, "learning_rate": 3.347863993025284e-05, "loss": 0.4819, "step": 768 }, { "epoch": 0.13, "learning_rate": 3.3522231909328684e-05, "loss": 0.5012, "step": 769 }, { "epoch": 0.13, "learning_rate": 3.3565823888404536e-05, "loss": 0.4768, "step": 770 }, { "epoch": 0.13, "learning_rate": 3.360941586748039e-05, "loss": 0.4867, "step": 771 }, { "epoch": 0.13, "learning_rate": 3.3653007846556234e-05, "loss": 0.4969, "step": 772 }, { "epoch": 0.13, "learning_rate": 3.3696599825632086e-05, "loss": 0.4978, "step": 773 }, { "epoch": 0.13, "learning_rate": 3.374019180470794e-05, "loss": 0.4894, "step": 774 }, { "epoch": 0.14, "learning_rate": 3.3783783783783784e-05, "loss": 0.4826, "step": 775 }, { "epoch": 0.14, "learning_rate": 3.382737576285964e-05, "loss": 0.4864, "step": 776 }, { "epoch": 0.14, "learning_rate": 3.387096774193548e-05, "loss": 0.5014, "step": 777 }, { "epoch": 0.14, "learning_rate": 3.3914559721011335e-05, "loss": 0.4788, "step": 778 }, { "epoch": 0.14, "learning_rate": 3.395815170008719e-05, "loss": 0.4866, "step": 779 }, { "epoch": 0.14, "learning_rate": 3.400174367916303e-05, "loss": 0.4764, "step": 780 }, { "epoch": 0.14, "learning_rate": 3.4045335658238885e-05, "loss": 0.4779, "step": 781 }, { "epoch": 0.14, "learning_rate": 3.408892763731474e-05, "loss": 0.48, "step": 782 }, { "epoch": 0.14, "learning_rate": 3.413251961639058e-05, "loss": 0.4676, "step": 783 }, { "epoch": 0.14, "learning_rate": 3.4176111595466435e-05, "loss": 0.4967, "step": 784 }, { "epoch": 0.14, "learning_rate": 3.421970357454228e-05, "loss": 0.5115, "step": 785 }, { "epoch": 0.14, "learning_rate": 3.426329555361813e-05, "loss": 0.4942, "step": 786 }, { "epoch": 0.14, "learning_rate": 3.4306887532693985e-05, "loss": 0.5083, "step": 787 }, { "epoch": 0.14, "learning_rate": 3.435047951176983e-05, "loss": 0.4689, "step": 788 }, { "epoch": 0.14, "learning_rate": 3.439407149084568e-05, "loss": 0.4862, "step": 789 }, { "epoch": 0.14, "learning_rate": 3.4437663469921536e-05, "loss": 0.5002, "step": 790 }, { "epoch": 0.14, "learning_rate": 3.448125544899739e-05, "loss": 0.5027, "step": 791 }, { "epoch": 0.14, "learning_rate": 3.452484742807324e-05, "loss": 0.4682, "step": 792 }, { "epoch": 0.14, "learning_rate": 3.4568439407149086e-05, "loss": 0.4854, "step": 793 }, { "epoch": 0.14, "learning_rate": 3.461203138622494e-05, "loss": 0.4865, "step": 794 }, { "epoch": 0.14, "learning_rate": 3.465562336530079e-05, "loss": 0.4761, "step": 795 }, { "epoch": 0.14, "learning_rate": 3.4699215344376636e-05, "loss": 0.4595, "step": 796 }, { "epoch": 0.14, "learning_rate": 3.474280732345249e-05, "loss": 0.4948, "step": 797 }, { "epoch": 0.14, "learning_rate": 3.478639930252834e-05, "loss": 0.474, "step": 798 }, { "epoch": 0.14, "learning_rate": 3.4829991281604186e-05, "loss": 0.4809, "step": 799 }, { "epoch": 0.14, "learning_rate": 3.487358326068004e-05, "loss": 0.4972, "step": 800 }, { "epoch": 0.14, "learning_rate": 3.491717523975589e-05, "loss": 0.4514, "step": 801 }, { "epoch": 0.14, "learning_rate": 3.4960767218831737e-05, "loss": 0.464, "step": 802 }, { "epoch": 0.14, "learning_rate": 3.500435919790759e-05, "loss": 0.4865, "step": 803 }, { "epoch": 0.14, "learning_rate": 3.5047951176983435e-05, "loss": 0.4859, "step": 804 }, { "epoch": 0.14, "learning_rate": 3.509154315605929e-05, "loss": 0.4751, "step": 805 }, { "epoch": 0.14, "learning_rate": 3.513513513513514e-05, "loss": 0.4827, "step": 806 }, { "epoch": 0.14, "learning_rate": 3.5178727114210985e-05, "loss": 0.4687, "step": 807 }, { "epoch": 0.14, "learning_rate": 3.522231909328684e-05, "loss": 0.4846, "step": 808 }, { "epoch": 0.14, "learning_rate": 3.526591107236269e-05, "loss": 0.4958, "step": 809 }, { "epoch": 0.14, "learning_rate": 3.5309503051438535e-05, "loss": 0.4937, "step": 810 }, { "epoch": 0.14, "learning_rate": 3.535309503051439e-05, "loss": 0.4985, "step": 811 }, { "epoch": 0.14, "learning_rate": 3.539668700959023e-05, "loss": 0.4602, "step": 812 }, { "epoch": 0.14, "learning_rate": 3.5440278988666085e-05, "loss": 0.4662, "step": 813 }, { "epoch": 0.14, "learning_rate": 3.548387096774194e-05, "loss": 0.4816, "step": 814 }, { "epoch": 0.14, "learning_rate": 3.552746294681778e-05, "loss": 0.4701, "step": 815 }, { "epoch": 0.14, "learning_rate": 3.5571054925893636e-05, "loss": 0.4693, "step": 816 }, { "epoch": 0.14, "learning_rate": 3.561464690496949e-05, "loss": 0.5097, "step": 817 }, { "epoch": 0.14, "learning_rate": 3.5658238884045333e-05, "loss": 0.5003, "step": 818 }, { "epoch": 0.14, "learning_rate": 3.5701830863121186e-05, "loss": 0.4709, "step": 819 }, { "epoch": 0.14, "learning_rate": 3.574542284219704e-05, "loss": 0.4974, "step": 820 }, { "epoch": 0.14, "learning_rate": 3.5789014821272884e-05, "loss": 0.5082, "step": 821 }, { "epoch": 0.14, "learning_rate": 3.5832606800348736e-05, "loss": 0.4676, "step": 822 }, { "epoch": 0.14, "learning_rate": 3.587619877942459e-05, "loss": 0.5002, "step": 823 }, { "epoch": 0.14, "learning_rate": 3.591979075850044e-05, "loss": 0.4836, "step": 824 }, { "epoch": 0.14, "learning_rate": 3.596338273757629e-05, "loss": 0.4739, "step": 825 }, { "epoch": 0.14, "learning_rate": 3.600697471665214e-05, "loss": 0.4824, "step": 826 }, { "epoch": 0.14, "learning_rate": 3.605056669572799e-05, "loss": 0.4952, "step": 827 }, { "epoch": 0.14, "learning_rate": 3.6094158674803837e-05, "loss": 0.5088, "step": 828 }, { "epoch": 0.14, "learning_rate": 3.613775065387969e-05, "loss": 0.4931, "step": 829 }, { "epoch": 0.14, "learning_rate": 3.618134263295554e-05, "loss": 0.4935, "step": 830 }, { "epoch": 0.14, "learning_rate": 3.622493461203139e-05, "loss": 0.4889, "step": 831 }, { "epoch": 0.15, "learning_rate": 3.626852659110724e-05, "loss": 0.4881, "step": 832 }, { "epoch": 0.15, "learning_rate": 3.631211857018309e-05, "loss": 0.4977, "step": 833 }, { "epoch": 0.15, "learning_rate": 3.635571054925894e-05, "loss": 0.481, "step": 834 }, { "epoch": 0.15, "learning_rate": 3.639930252833479e-05, "loss": 0.4813, "step": 835 }, { "epoch": 0.15, "learning_rate": 3.644289450741064e-05, "loss": 0.4638, "step": 836 }, { "epoch": 0.15, "learning_rate": 3.648648648648649e-05, "loss": 0.4737, "step": 837 }, { "epoch": 0.15, "learning_rate": 3.653007846556234e-05, "loss": 0.5126, "step": 838 }, { "epoch": 0.15, "learning_rate": 3.6573670444638185e-05, "loss": 0.4865, "step": 839 }, { "epoch": 0.15, "learning_rate": 3.661726242371404e-05, "loss": 0.4993, "step": 840 }, { "epoch": 0.15, "learning_rate": 3.666085440278989e-05, "loss": 0.4937, "step": 841 }, { "epoch": 0.15, "learning_rate": 3.6704446381865735e-05, "loss": 0.4867, "step": 842 }, { "epoch": 0.15, "learning_rate": 3.674803836094159e-05, "loss": 0.4798, "step": 843 }, { "epoch": 0.15, "learning_rate": 3.679163034001744e-05, "loss": 0.4713, "step": 844 }, { "epoch": 0.15, "learning_rate": 3.6835222319093286e-05, "loss": 0.4861, "step": 845 }, { "epoch": 0.15, "learning_rate": 3.687881429816914e-05, "loss": 0.4841, "step": 846 }, { "epoch": 0.15, "learning_rate": 3.6922406277244984e-05, "loss": 0.4847, "step": 847 }, { "epoch": 0.15, "learning_rate": 3.6965998256320836e-05, "loss": 0.4636, "step": 848 }, { "epoch": 0.15, "learning_rate": 3.700959023539669e-05, "loss": 0.4721, "step": 849 }, { "epoch": 0.15, "learning_rate": 3.7053182214472534e-05, "loss": 0.4671, "step": 850 }, { "epoch": 0.15, "learning_rate": 3.7096774193548386e-05, "loss": 0.4785, "step": 851 }, { "epoch": 0.15, "learning_rate": 3.714036617262424e-05, "loss": 0.4684, "step": 852 }, { "epoch": 0.15, "learning_rate": 3.718395815170009e-05, "loss": 0.4841, "step": 853 }, { "epoch": 0.15, "learning_rate": 3.722755013077594e-05, "loss": 0.4685, "step": 854 }, { "epoch": 0.15, "learning_rate": 3.727114210985179e-05, "loss": 0.4829, "step": 855 }, { "epoch": 0.15, "learning_rate": 3.731473408892764e-05, "loss": 0.4632, "step": 856 }, { "epoch": 0.15, "learning_rate": 3.7358326068003494e-05, "loss": 0.4677, "step": 857 }, { "epoch": 0.15, "learning_rate": 3.740191804707934e-05, "loss": 0.4983, "step": 858 }, { "epoch": 0.15, "learning_rate": 3.744551002615519e-05, "loss": 0.4817, "step": 859 }, { "epoch": 0.15, "learning_rate": 3.7489102005231044e-05, "loss": 0.4784, "step": 860 }, { "epoch": 0.15, "learning_rate": 3.753269398430689e-05, "loss": 0.4627, "step": 861 }, { "epoch": 0.15, "learning_rate": 3.757628596338274e-05, "loss": 0.4923, "step": 862 }, { "epoch": 0.15, "learning_rate": 3.761987794245859e-05, "loss": 0.497, "step": 863 }, { "epoch": 0.15, "learning_rate": 3.766346992153444e-05, "loss": 0.4728, "step": 864 }, { "epoch": 0.15, "learning_rate": 3.770706190061029e-05, "loss": 0.4899, "step": 865 }, { "epoch": 0.15, "learning_rate": 3.775065387968614e-05, "loss": 0.4896, "step": 866 }, { "epoch": 0.15, "learning_rate": 3.779424585876199e-05, "loss": 0.4543, "step": 867 }, { "epoch": 0.15, "learning_rate": 3.783783783783784e-05, "loss": 0.4583, "step": 868 }, { "epoch": 0.15, "learning_rate": 3.788142981691369e-05, "loss": 0.4785, "step": 869 }, { "epoch": 0.15, "learning_rate": 3.792502179598954e-05, "loss": 0.4936, "step": 870 }, { "epoch": 0.15, "learning_rate": 3.796861377506539e-05, "loss": 0.4684, "step": 871 }, { "epoch": 0.15, "learning_rate": 3.801220575414124e-05, "loss": 0.4979, "step": 872 }, { "epoch": 0.15, "learning_rate": 3.805579773321709e-05, "loss": 0.4764, "step": 873 }, { "epoch": 0.15, "learning_rate": 3.8099389712292936e-05, "loss": 0.4633, "step": 874 }, { "epoch": 0.15, "learning_rate": 3.814298169136879e-05, "loss": 0.4734, "step": 875 }, { "epoch": 0.15, "learning_rate": 3.818657367044464e-05, "loss": 0.4898, "step": 876 }, { "epoch": 0.15, "learning_rate": 3.8230165649520486e-05, "loss": 0.5063, "step": 877 }, { "epoch": 0.15, "learning_rate": 3.827375762859634e-05, "loss": 0.4818, "step": 878 }, { "epoch": 0.15, "learning_rate": 3.831734960767219e-05, "loss": 0.4814, "step": 879 }, { "epoch": 0.15, "learning_rate": 3.8360941586748036e-05, "loss": 0.4916, "step": 880 }, { "epoch": 0.15, "learning_rate": 3.840453356582389e-05, "loss": 0.4835, "step": 881 }, { "epoch": 0.15, "learning_rate": 3.844812554489974e-05, "loss": 0.4708, "step": 882 }, { "epoch": 0.15, "learning_rate": 3.849171752397559e-05, "loss": 0.4753, "step": 883 }, { "epoch": 0.15, "learning_rate": 3.853530950305144e-05, "loss": 0.4914, "step": 884 }, { "epoch": 0.15, "learning_rate": 3.857890148212729e-05, "loss": 0.4868, "step": 885 }, { "epoch": 0.15, "learning_rate": 3.8622493461203144e-05, "loss": 0.478, "step": 886 }, { "epoch": 0.15, "learning_rate": 3.8666085440278996e-05, "loss": 0.4728, "step": 887 }, { "epoch": 0.15, "learning_rate": 3.870967741935484e-05, "loss": 0.456, "step": 888 }, { "epoch": 0.15, "learning_rate": 3.8753269398430694e-05, "loss": 0.5146, "step": 889 }, { "epoch": 0.16, "learning_rate": 3.879686137750654e-05, "loss": 0.494, "step": 890 }, { "epoch": 0.16, "learning_rate": 3.884045335658239e-05, "loss": 0.48, "step": 891 }, { "epoch": 0.16, "learning_rate": 3.8884045335658244e-05, "loss": 0.4707, "step": 892 }, { "epoch": 0.16, "learning_rate": 3.892763731473409e-05, "loss": 0.4891, "step": 893 }, { "epoch": 0.16, "learning_rate": 3.897122929380994e-05, "loss": 0.481, "step": 894 }, { "epoch": 0.16, "learning_rate": 3.9014821272885795e-05, "loss": 0.4737, "step": 895 }, { "epoch": 0.16, "learning_rate": 3.905841325196164e-05, "loss": 0.4817, "step": 896 }, { "epoch": 0.16, "learning_rate": 3.910200523103749e-05, "loss": 0.4778, "step": 897 }, { "epoch": 0.16, "learning_rate": 3.914559721011334e-05, "loss": 0.4752, "step": 898 }, { "epoch": 0.16, "learning_rate": 3.918918918918919e-05, "loss": 0.4713, "step": 899 }, { "epoch": 0.16, "learning_rate": 3.923278116826504e-05, "loss": 0.4794, "step": 900 }, { "epoch": 0.16, "learning_rate": 3.927637314734089e-05, "loss": 0.4854, "step": 901 }, { "epoch": 0.16, "learning_rate": 3.931996512641674e-05, "loss": 0.4664, "step": 902 }, { "epoch": 0.16, "learning_rate": 3.936355710549259e-05, "loss": 0.4819, "step": 903 }, { "epoch": 0.16, "learning_rate": 3.940714908456844e-05, "loss": 0.4785, "step": 904 }, { "epoch": 0.16, "learning_rate": 3.945074106364429e-05, "loss": 0.4824, "step": 905 }, { "epoch": 0.16, "learning_rate": 3.949433304272014e-05, "loss": 0.4906, "step": 906 }, { "epoch": 0.16, "learning_rate": 3.953792502179599e-05, "loss": 0.4731, "step": 907 }, { "epoch": 0.16, "learning_rate": 3.958151700087184e-05, "loss": 0.4656, "step": 908 }, { "epoch": 0.16, "learning_rate": 3.962510897994769e-05, "loss": 0.4803, "step": 909 }, { "epoch": 0.16, "learning_rate": 3.966870095902354e-05, "loss": 0.4682, "step": 910 }, { "epoch": 0.16, "learning_rate": 3.971229293809939e-05, "loss": 0.4811, "step": 911 }, { "epoch": 0.16, "learning_rate": 3.975588491717524e-05, "loss": 0.4786, "step": 912 }, { "epoch": 0.16, "learning_rate": 3.979947689625109e-05, "loss": 0.4889, "step": 913 }, { "epoch": 0.16, "learning_rate": 3.984306887532694e-05, "loss": 0.4871, "step": 914 }, { "epoch": 0.16, "learning_rate": 3.9886660854402794e-05, "loss": 0.4731, "step": 915 }, { "epoch": 0.16, "learning_rate": 3.9930252833478646e-05, "loss": 0.466, "step": 916 }, { "epoch": 0.16, "learning_rate": 3.997384481255449e-05, "loss": 0.497, "step": 917 }, { "epoch": 0.16, "learning_rate": 4.0017436791630344e-05, "loss": 0.4813, "step": 918 }, { "epoch": 0.16, "learning_rate": 4.0061028770706197e-05, "loss": 0.4978, "step": 919 }, { "epoch": 0.16, "learning_rate": 4.010462074978204e-05, "loss": 0.4762, "step": 920 }, { "epoch": 0.16, "learning_rate": 4.0148212728857894e-05, "loss": 0.5172, "step": 921 }, { "epoch": 0.16, "learning_rate": 4.019180470793375e-05, "loss": 0.4708, "step": 922 }, { "epoch": 0.16, "learning_rate": 4.023539668700959e-05, "loss": 0.4804, "step": 923 }, { "epoch": 0.16, "learning_rate": 4.0278988666085445e-05, "loss": 0.4716, "step": 924 }, { "epoch": 0.16, "learning_rate": 4.032258064516129e-05, "loss": 0.4739, "step": 925 }, { "epoch": 0.16, "learning_rate": 4.036617262423714e-05, "loss": 0.4777, "step": 926 }, { "epoch": 0.16, "learning_rate": 4.0409764603312995e-05, "loss": 0.4914, "step": 927 }, { "epoch": 0.16, "learning_rate": 4.045335658238884e-05, "loss": 0.5086, "step": 928 }, { "epoch": 0.16, "learning_rate": 4.049694856146469e-05, "loss": 0.4806, "step": 929 }, { "epoch": 0.16, "learning_rate": 4.0540540540540545e-05, "loss": 0.4838, "step": 930 }, { "epoch": 0.16, "learning_rate": 4.058413251961639e-05, "loss": 0.4775, "step": 931 }, { "epoch": 0.16, "learning_rate": 4.062772449869224e-05, "loss": 0.4872, "step": 932 }, { "epoch": 0.16, "learning_rate": 4.067131647776809e-05, "loss": 0.4935, "step": 933 }, { "epoch": 0.16, "learning_rate": 4.071490845684394e-05, "loss": 0.4912, "step": 934 }, { "epoch": 0.16, "learning_rate": 4.0758500435919793e-05, "loss": 0.4789, "step": 935 }, { "epoch": 0.16, "learning_rate": 4.080209241499564e-05, "loss": 0.4937, "step": 936 }, { "epoch": 0.16, "learning_rate": 4.084568439407149e-05, "loss": 0.479, "step": 937 }, { "epoch": 0.16, "learning_rate": 4.0889276373147344e-05, "loss": 0.4882, "step": 938 }, { "epoch": 0.16, "learning_rate": 4.093286835222319e-05, "loss": 0.4713, "step": 939 }, { "epoch": 0.16, "learning_rate": 4.097646033129904e-05, "loss": 0.4765, "step": 940 }, { "epoch": 0.16, "learning_rate": 4.1020052310374894e-05, "loss": 0.4602, "step": 941 }, { "epoch": 0.16, "learning_rate": 4.106364428945074e-05, "loss": 0.4818, "step": 942 }, { "epoch": 0.16, "learning_rate": 4.110723626852659e-05, "loss": 0.4879, "step": 943 }, { "epoch": 0.16, "learning_rate": 4.1150828247602444e-05, "loss": 0.4753, "step": 944 }, { "epoch": 0.16, "learning_rate": 4.119442022667829e-05, "loss": 0.4737, "step": 945 }, { "epoch": 0.16, "learning_rate": 4.123801220575414e-05, "loss": 0.4715, "step": 946 }, { "epoch": 0.17, "learning_rate": 4.1281604184829994e-05, "loss": 0.4767, "step": 947 }, { "epoch": 0.17, "learning_rate": 4.132519616390585e-05, "loss": 0.4794, "step": 948 }, { "epoch": 0.17, "learning_rate": 4.13687881429817e-05, "loss": 0.4624, "step": 949 }, { "epoch": 0.17, "learning_rate": 4.1412380122057545e-05, "loss": 0.4695, "step": 950 }, { "epoch": 0.17, "learning_rate": 4.14559721011334e-05, "loss": 0.4662, "step": 951 }, { "epoch": 0.17, "learning_rate": 4.149956408020924e-05, "loss": 0.481, "step": 952 }, { "epoch": 0.17, "learning_rate": 4.1543156059285095e-05, "loss": 0.4682, "step": 953 }, { "epoch": 0.17, "learning_rate": 4.158674803836095e-05, "loss": 0.4806, "step": 954 }, { "epoch": 0.17, "learning_rate": 4.163034001743679e-05, "loss": 0.4954, "step": 955 }, { "epoch": 0.17, "learning_rate": 4.1673931996512645e-05, "loss": 0.4791, "step": 956 }, { "epoch": 0.17, "learning_rate": 4.17175239755885e-05, "loss": 0.4666, "step": 957 }, { "epoch": 0.17, "learning_rate": 4.176111595466434e-05, "loss": 0.4562, "step": 958 }, { "epoch": 0.17, "learning_rate": 4.1804707933740195e-05, "loss": 0.4902, "step": 959 }, { "epoch": 0.17, "learning_rate": 4.184829991281604e-05, "loss": 0.4651, "step": 960 }, { "epoch": 0.17, "learning_rate": 4.189189189189189e-05, "loss": 0.4714, "step": 961 }, { "epoch": 0.17, "learning_rate": 4.1935483870967746e-05, "loss": 0.4758, "step": 962 }, { "epoch": 0.17, "learning_rate": 4.197907585004359e-05, "loss": 0.4799, "step": 963 }, { "epoch": 0.17, "learning_rate": 4.2022667829119444e-05, "loss": 0.4806, "step": 964 }, { "epoch": 0.17, "learning_rate": 4.2066259808195296e-05, "loss": 0.4841, "step": 965 }, { "epoch": 0.17, "learning_rate": 4.210985178727114e-05, "loss": 0.4715, "step": 966 }, { "epoch": 0.17, "learning_rate": 4.2153443766346994e-05, "loss": 0.4765, "step": 967 }, { "epoch": 0.17, "learning_rate": 4.219703574542284e-05, "loss": 0.4856, "step": 968 }, { "epoch": 0.17, "learning_rate": 4.224062772449869e-05, "loss": 0.4648, "step": 969 }, { "epoch": 0.17, "learning_rate": 4.2284219703574544e-05, "loss": 0.4808, "step": 970 }, { "epoch": 0.17, "learning_rate": 4.232781168265039e-05, "loss": 0.4864, "step": 971 }, { "epoch": 0.17, "learning_rate": 4.237140366172624e-05, "loss": 0.4895, "step": 972 }, { "epoch": 0.17, "learning_rate": 4.2414995640802094e-05, "loss": 0.4584, "step": 973 }, { "epoch": 0.17, "learning_rate": 4.245858761987794e-05, "loss": 0.4652, "step": 974 }, { "epoch": 0.17, "learning_rate": 4.250217959895379e-05, "loss": 0.4619, "step": 975 }, { "epoch": 0.17, "learning_rate": 4.2545771578029645e-05, "loss": 0.4783, "step": 976 }, { "epoch": 0.17, "learning_rate": 4.25893635571055e-05, "loss": 0.473, "step": 977 }, { "epoch": 0.17, "learning_rate": 4.263295553618135e-05, "loss": 0.4633, "step": 978 }, { "epoch": 0.17, "learning_rate": 4.2676547515257195e-05, "loss": 0.4824, "step": 979 }, { "epoch": 0.17, "learning_rate": 4.272013949433305e-05, "loss": 0.4875, "step": 980 }, { "epoch": 0.17, "learning_rate": 4.27637314734089e-05, "loss": 0.4809, "step": 981 }, { "epoch": 0.17, "learning_rate": 4.2807323452484745e-05, "loss": 0.4572, "step": 982 }, { "epoch": 0.17, "learning_rate": 4.28509154315606e-05, "loss": 0.4825, "step": 983 }, { "epoch": 0.17, "learning_rate": 4.289450741063645e-05, "loss": 0.4844, "step": 984 }, { "epoch": 0.17, "learning_rate": 4.2938099389712295e-05, "loss": 0.4854, "step": 985 }, { "epoch": 0.17, "learning_rate": 4.298169136878815e-05, "loss": 0.475, "step": 986 }, { "epoch": 0.17, "learning_rate": 4.302528334786399e-05, "loss": 0.4614, "step": 987 }, { "epoch": 0.17, "learning_rate": 4.3068875326939846e-05, "loss": 0.4568, "step": 988 }, { "epoch": 0.17, "learning_rate": 4.31124673060157e-05, "loss": 0.4645, "step": 989 }, { "epoch": 0.17, "learning_rate": 4.3156059285091544e-05, "loss": 0.469, "step": 990 }, { "epoch": 0.17, "learning_rate": 4.3199651264167396e-05, "loss": 0.4827, "step": 991 }, { "epoch": 0.17, "learning_rate": 4.324324324324325e-05, "loss": 0.4838, "step": 992 }, { "epoch": 0.17, "learning_rate": 4.3286835222319094e-05, "loss": 0.4616, "step": 993 }, { "epoch": 0.17, "learning_rate": 4.3330427201394946e-05, "loss": 0.4815, "step": 994 }, { "epoch": 0.17, "learning_rate": 4.337401918047079e-05, "loss": 0.4881, "step": 995 }, { "epoch": 0.17, "learning_rate": 4.3417611159546644e-05, "loss": 0.4752, "step": 996 }, { "epoch": 0.17, "learning_rate": 4.3461203138622496e-05, "loss": 0.4927, "step": 997 }, { "epoch": 0.17, "learning_rate": 4.350479511769834e-05, "loss": 0.481, "step": 998 }, { "epoch": 0.17, "learning_rate": 4.3548387096774194e-05, "loss": 0.4643, "step": 999 }, { "epoch": 0.17, "learning_rate": 4.359197907585005e-05, "loss": 0.4782, "step": 1000 }, { "epoch": 0.17, "learning_rate": 4.363557105492589e-05, "loss": 0.4833, "step": 1001 }, { "epoch": 0.17, "learning_rate": 4.3679163034001745e-05, "loss": 0.4721, "step": 1002 }, { "epoch": 0.17, "learning_rate": 4.372275501307759e-05, "loss": 0.477, "step": 1003 }, { "epoch": 0.18, "learning_rate": 4.376634699215344e-05, "loss": 0.4744, "step": 1004 }, { "epoch": 0.18, "learning_rate": 4.3809938971229295e-05, "loss": 0.4499, "step": 1005 }, { "epoch": 0.18, "learning_rate": 4.385353095030515e-05, "loss": 0.4785, "step": 1006 }, { "epoch": 0.18, "learning_rate": 4.389712292938099e-05, "loss": 0.4836, "step": 1007 }, { "epoch": 0.18, "learning_rate": 4.3940714908456845e-05, "loss": 0.4912, "step": 1008 }, { "epoch": 0.18, "learning_rate": 4.39843068875327e-05, "loss": 0.4673, "step": 1009 }, { "epoch": 0.18, "learning_rate": 4.402789886660855e-05, "loss": 0.4673, "step": 1010 }, { "epoch": 0.18, "learning_rate": 4.40714908456844e-05, "loss": 0.4784, "step": 1011 }, { "epoch": 0.18, "learning_rate": 4.411508282476025e-05, "loss": 0.4924, "step": 1012 }, { "epoch": 0.18, "learning_rate": 4.41586748038361e-05, "loss": 0.4784, "step": 1013 }, { "epoch": 0.18, "learning_rate": 4.4202266782911946e-05, "loss": 0.4561, "step": 1014 }, { "epoch": 0.18, "learning_rate": 4.42458587619878e-05, "loss": 0.4792, "step": 1015 }, { "epoch": 0.18, "learning_rate": 4.428945074106365e-05, "loss": 0.4518, "step": 1016 }, { "epoch": 0.18, "learning_rate": 4.4333042720139496e-05, "loss": 0.4758, "step": 1017 }, { "epoch": 0.18, "learning_rate": 4.437663469921535e-05, "loss": 0.4615, "step": 1018 }, { "epoch": 0.18, "learning_rate": 4.44202266782912e-05, "loss": 0.4793, "step": 1019 }, { "epoch": 0.18, "learning_rate": 4.4463818657367046e-05, "loss": 0.4693, "step": 1020 }, { "epoch": 0.18, "learning_rate": 4.45074106364429e-05, "loss": 0.4722, "step": 1021 }, { "epoch": 0.18, "learning_rate": 4.4551002615518744e-05, "loss": 0.4776, "step": 1022 }, { "epoch": 0.18, "learning_rate": 4.4594594594594596e-05, "loss": 0.475, "step": 1023 }, { "epoch": 0.18, "learning_rate": 4.463818657367045e-05, "loss": 0.4838, "step": 1024 }, { "epoch": 0.18, "learning_rate": 4.4681778552746294e-05, "loss": 0.4801, "step": 1025 }, { "epoch": 0.18, "learning_rate": 4.472537053182215e-05, "loss": 0.472, "step": 1026 }, { "epoch": 0.18, "learning_rate": 4.4768962510898e-05, "loss": 0.4793, "step": 1027 }, { "epoch": 0.18, "learning_rate": 4.4812554489973845e-05, "loss": 0.4697, "step": 1028 }, { "epoch": 0.18, "learning_rate": 4.48561464690497e-05, "loss": 0.472, "step": 1029 }, { "epoch": 0.18, "learning_rate": 4.489973844812554e-05, "loss": 0.4685, "step": 1030 }, { "epoch": 0.18, "learning_rate": 4.4943330427201395e-05, "loss": 0.486, "step": 1031 }, { "epoch": 0.18, "learning_rate": 4.498692240627725e-05, "loss": 0.4886, "step": 1032 }, { "epoch": 0.18, "learning_rate": 4.503051438535309e-05, "loss": 0.4746, "step": 1033 }, { "epoch": 0.18, "learning_rate": 4.5074106364428945e-05, "loss": 0.478, "step": 1034 }, { "epoch": 0.18, "learning_rate": 4.51176983435048e-05, "loss": 0.4788, "step": 1035 }, { "epoch": 0.18, "learning_rate": 4.516129032258064e-05, "loss": 0.4929, "step": 1036 }, { "epoch": 0.18, "learning_rate": 4.5204882301656495e-05, "loss": 0.474, "step": 1037 }, { "epoch": 0.18, "learning_rate": 4.524847428073235e-05, "loss": 0.4971, "step": 1038 }, { "epoch": 0.18, "learning_rate": 4.52920662598082e-05, "loss": 0.467, "step": 1039 }, { "epoch": 0.18, "learning_rate": 4.533565823888405e-05, "loss": 0.4746, "step": 1040 }, { "epoch": 0.18, "learning_rate": 4.53792502179599e-05, "loss": 0.4617, "step": 1041 }, { "epoch": 0.18, "learning_rate": 4.542284219703575e-05, "loss": 0.4733, "step": 1042 }, { "epoch": 0.18, "learning_rate": 4.54664341761116e-05, "loss": 0.4778, "step": 1043 }, { "epoch": 0.18, "learning_rate": 4.551002615518745e-05, "loss": 0.4762, "step": 1044 }, { "epoch": 0.18, "learning_rate": 4.55536181342633e-05, "loss": 0.4821, "step": 1045 }, { "epoch": 0.18, "learning_rate": 4.559721011333915e-05, "loss": 0.4734, "step": 1046 }, { "epoch": 0.18, "learning_rate": 4.5640802092415e-05, "loss": 0.4856, "step": 1047 }, { "epoch": 0.18, "learning_rate": 4.568439407149085e-05, "loss": 0.4579, "step": 1048 }, { "epoch": 0.18, "learning_rate": 4.5727986050566696e-05, "loss": 0.4856, "step": 1049 }, { "epoch": 0.18, "learning_rate": 4.577157802964255e-05, "loss": 0.4687, "step": 1050 }, { "epoch": 0.18, "learning_rate": 4.58151700087184e-05, "loss": 0.4739, "step": 1051 }, { "epoch": 0.18, "learning_rate": 4.5858761987794247e-05, "loss": 0.4846, "step": 1052 }, { "epoch": 0.18, "learning_rate": 4.59023539668701e-05, "loss": 0.4855, "step": 1053 }, { "epoch": 0.18, "learning_rate": 4.594594594594595e-05, "loss": 0.4919, "step": 1054 }, { "epoch": 0.18, "learning_rate": 4.59895379250218e-05, "loss": 0.4769, "step": 1055 }, { "epoch": 0.18, "learning_rate": 4.603312990409765e-05, "loss": 0.4691, "step": 1056 }, { "epoch": 0.18, "learning_rate": 4.6076721883173495e-05, "loss": 0.4745, "step": 1057 }, { "epoch": 0.18, "learning_rate": 4.612031386224935e-05, "loss": 0.4679, "step": 1058 }, { "epoch": 0.18, "learning_rate": 4.61639058413252e-05, "loss": 0.4703, "step": 1059 }, { "epoch": 0.18, "learning_rate": 4.6207497820401045e-05, "loss": 0.4792, "step": 1060 }, { "epoch": 0.18, "learning_rate": 4.62510897994769e-05, "loss": 0.4998, "step": 1061 }, { "epoch": 0.19, "learning_rate": 4.629468177855275e-05, "loss": 0.469, "step": 1062 }, { "epoch": 0.19, "learning_rate": 4.6338273757628595e-05, "loss": 0.4737, "step": 1063 }, { "epoch": 0.19, "learning_rate": 4.638186573670445e-05, "loss": 0.4811, "step": 1064 }, { "epoch": 0.19, "learning_rate": 4.642545771578029e-05, "loss": 0.492, "step": 1065 }, { "epoch": 0.19, "learning_rate": 4.6469049694856146e-05, "loss": 0.4816, "step": 1066 }, { "epoch": 0.19, "learning_rate": 4.6512641673932e-05, "loss": 0.4717, "step": 1067 }, { "epoch": 0.19, "learning_rate": 4.655623365300785e-05, "loss": 0.4712, "step": 1068 }, { "epoch": 0.19, "learning_rate": 4.6599825632083696e-05, "loss": 0.4613, "step": 1069 }, { "epoch": 0.19, "learning_rate": 4.664341761115955e-05, "loss": 0.4602, "step": 1070 }, { "epoch": 0.19, "learning_rate": 4.66870095902354e-05, "loss": 0.4729, "step": 1071 }, { "epoch": 0.19, "learning_rate": 4.673060156931125e-05, "loss": 0.4994, "step": 1072 }, { "epoch": 0.19, "learning_rate": 4.67741935483871e-05, "loss": 0.4755, "step": 1073 }, { "epoch": 0.19, "learning_rate": 4.681778552746295e-05, "loss": 0.4581, "step": 1074 }, { "epoch": 0.19, "learning_rate": 4.68613775065388e-05, "loss": 0.4795, "step": 1075 }, { "epoch": 0.19, "learning_rate": 4.690496948561465e-05, "loss": 0.4555, "step": 1076 }, { "epoch": 0.19, "learning_rate": 4.69485614646905e-05, "loss": 0.4828, "step": 1077 }, { "epoch": 0.19, "learning_rate": 4.699215344376635e-05, "loss": 0.4722, "step": 1078 }, { "epoch": 0.19, "learning_rate": 4.70357454228422e-05, "loss": 0.4778, "step": 1079 }, { "epoch": 0.19, "learning_rate": 4.707933740191805e-05, "loss": 0.4805, "step": 1080 }, { "epoch": 0.19, "learning_rate": 4.7122929380993904e-05, "loss": 0.4676, "step": 1081 }, { "epoch": 0.19, "learning_rate": 4.716652136006975e-05, "loss": 0.4713, "step": 1082 }, { "epoch": 0.19, "learning_rate": 4.72101133391456e-05, "loss": 0.469, "step": 1083 }, { "epoch": 0.19, "learning_rate": 4.725370531822145e-05, "loss": 0.4761, "step": 1084 }, { "epoch": 0.19, "learning_rate": 4.72972972972973e-05, "loss": 0.47, "step": 1085 }, { "epoch": 0.19, "learning_rate": 4.734088927637315e-05, "loss": 0.4781, "step": 1086 }, { "epoch": 0.19, "learning_rate": 4.7384481255449e-05, "loss": 0.4795, "step": 1087 }, { "epoch": 0.19, "learning_rate": 4.742807323452485e-05, "loss": 0.4792, "step": 1088 }, { "epoch": 0.19, "learning_rate": 4.74716652136007e-05, "loss": 0.4821, "step": 1089 }, { "epoch": 0.19, "learning_rate": 4.751525719267655e-05, "loss": 0.4623, "step": 1090 }, { "epoch": 0.19, "learning_rate": 4.75588491717524e-05, "loss": 0.4721, "step": 1091 }, { "epoch": 0.19, "learning_rate": 4.7602441150828245e-05, "loss": 0.491, "step": 1092 }, { "epoch": 0.19, "learning_rate": 4.76460331299041e-05, "loss": 0.4715, "step": 1093 }, { "epoch": 0.19, "learning_rate": 4.768962510897995e-05, "loss": 0.4691, "step": 1094 }, { "epoch": 0.19, "learning_rate": 4.7733217088055796e-05, "loss": 0.4721, "step": 1095 }, { "epoch": 0.19, "learning_rate": 4.777680906713165e-05, "loss": 0.4749, "step": 1096 }, { "epoch": 0.19, "learning_rate": 4.78204010462075e-05, "loss": 0.4754, "step": 1097 }, { "epoch": 0.19, "learning_rate": 4.7863993025283346e-05, "loss": 0.4888, "step": 1098 }, { "epoch": 0.19, "learning_rate": 4.79075850043592e-05, "loss": 0.473, "step": 1099 }, { "epoch": 0.19, "learning_rate": 4.795117698343505e-05, "loss": 0.4953, "step": 1100 }, { "epoch": 0.19, "learning_rate": 4.79947689625109e-05, "loss": 0.4847, "step": 1101 }, { "epoch": 0.19, "learning_rate": 4.803836094158675e-05, "loss": 0.4716, "step": 1102 }, { "epoch": 0.19, "learning_rate": 4.80819529206626e-05, "loss": 0.4923, "step": 1103 }, { "epoch": 0.19, "learning_rate": 4.812554489973845e-05, "loss": 0.471, "step": 1104 }, { "epoch": 0.19, "learning_rate": 4.8169136878814306e-05, "loss": 0.4936, "step": 1105 }, { "epoch": 0.19, "learning_rate": 4.821272885789015e-05, "loss": 0.4796, "step": 1106 }, { "epoch": 0.19, "learning_rate": 4.8256320836966004e-05, "loss": 0.4841, "step": 1107 }, { "epoch": 0.19, "learning_rate": 4.829991281604185e-05, "loss": 0.4527, "step": 1108 }, { "epoch": 0.19, "learning_rate": 4.83435047951177e-05, "loss": 0.4723, "step": 1109 }, { "epoch": 0.19, "learning_rate": 4.8387096774193554e-05, "loss": 0.4698, "step": 1110 }, { "epoch": 0.19, "learning_rate": 4.84306887532694e-05, "loss": 0.4701, "step": 1111 }, { "epoch": 0.19, "learning_rate": 4.847428073234525e-05, "loss": 0.4615, "step": 1112 }, { "epoch": 0.19, "learning_rate": 4.8517872711421104e-05, "loss": 0.451, "step": 1113 }, { "epoch": 0.19, "learning_rate": 4.856146469049695e-05, "loss": 0.4644, "step": 1114 }, { "epoch": 0.19, "learning_rate": 4.86050566695728e-05, "loss": 0.4771, "step": 1115 }, { "epoch": 0.19, "learning_rate": 4.8648648648648654e-05, "loss": 0.5001, "step": 1116 }, { "epoch": 0.19, "learning_rate": 4.86922406277245e-05, "loss": 0.4763, "step": 1117 }, { "epoch": 0.19, "learning_rate": 4.873583260680035e-05, "loss": 0.4689, "step": 1118 }, { "epoch": 0.2, "learning_rate": 4.87794245858762e-05, "loss": 0.4911, "step": 1119 }, { "epoch": 0.2, "learning_rate": 4.882301656495205e-05, "loss": 0.4982, "step": 1120 }, { "epoch": 0.2, "learning_rate": 4.88666085440279e-05, "loss": 0.4695, "step": 1121 }, { "epoch": 0.2, "learning_rate": 4.891020052310375e-05, "loss": 0.5039, "step": 1122 }, { "epoch": 0.2, "learning_rate": 4.89537925021796e-05, "loss": 0.4923, "step": 1123 }, { "epoch": 0.2, "learning_rate": 4.899738448125545e-05, "loss": 0.4694, "step": 1124 }, { "epoch": 0.2, "learning_rate": 4.90409764603313e-05, "loss": 0.4831, "step": 1125 }, { "epoch": 0.2, "learning_rate": 4.908456843940715e-05, "loss": 0.4661, "step": 1126 }, { "epoch": 0.2, "learning_rate": 4.9128160418482996e-05, "loss": 0.4896, "step": 1127 }, { "epoch": 0.2, "learning_rate": 4.917175239755885e-05, "loss": 0.489, "step": 1128 }, { "epoch": 0.2, "learning_rate": 4.92153443766347e-05, "loss": 0.4853, "step": 1129 }, { "epoch": 0.2, "learning_rate": 4.925893635571055e-05, "loss": 0.4717, "step": 1130 }, { "epoch": 0.2, "learning_rate": 4.93025283347864e-05, "loss": 0.4598, "step": 1131 }, { "epoch": 0.2, "learning_rate": 4.934612031386225e-05, "loss": 0.4872, "step": 1132 }, { "epoch": 0.2, "learning_rate": 4.9389712292938103e-05, "loss": 0.4761, "step": 1133 }, { "epoch": 0.2, "learning_rate": 4.9433304272013956e-05, "loss": 0.4767, "step": 1134 }, { "epoch": 0.2, "learning_rate": 4.94768962510898e-05, "loss": 0.456, "step": 1135 }, { "epoch": 0.2, "learning_rate": 4.9520488230165654e-05, "loss": 0.4684, "step": 1136 }, { "epoch": 0.2, "learning_rate": 4.9564080209241506e-05, "loss": 0.4692, "step": 1137 }, { "epoch": 0.2, "learning_rate": 4.960767218831735e-05, "loss": 0.482, "step": 1138 }, { "epoch": 0.2, "learning_rate": 4.9651264167393204e-05, "loss": 0.49, "step": 1139 }, { "epoch": 0.2, "learning_rate": 4.9694856146469056e-05, "loss": 0.4729, "step": 1140 }, { "epoch": 0.2, "learning_rate": 4.97384481255449e-05, "loss": 0.4692, "step": 1141 }, { "epoch": 0.2, "learning_rate": 4.9782040104620754e-05, "loss": 0.4901, "step": 1142 }, { "epoch": 0.2, "learning_rate": 4.98256320836966e-05, "loss": 0.4777, "step": 1143 }, { "epoch": 0.2, "learning_rate": 4.986922406277245e-05, "loss": 0.4647, "step": 1144 }, { "epoch": 0.2, "learning_rate": 4.9912816041848304e-05, "loss": 0.4689, "step": 1145 }, { "epoch": 0.2, "learning_rate": 4.995640802092415e-05, "loss": 0.4876, "step": 1146 }, { "epoch": 0.2, "learning_rate": 5e-05, "loss": 0.4623, "step": 1147 }, { "epoch": 0.2, "learning_rate": 4.9999998958065526e-05, "loss": 0.4753, "step": 1148 }, { "epoch": 0.2, "learning_rate": 4.999999583226218e-05, "loss": 0.4732, "step": 1149 }, { "epoch": 0.2, "learning_rate": 4.9999990622590265e-05, "loss": 0.4599, "step": 1150 }, { "epoch": 0.2, "learning_rate": 4.999998332905026e-05, "loss": 0.4878, "step": 1151 }, { "epoch": 0.2, "learning_rate": 4.999997395164284e-05, "loss": 0.4874, "step": 1152 }, { "epoch": 0.2, "learning_rate": 4.999996249036888e-05, "loss": 0.4938, "step": 1153 }, { "epoch": 0.2, "learning_rate": 4.999994894522943e-05, "loss": 0.4719, "step": 1154 }, { "epoch": 0.2, "learning_rate": 4.999993331622575e-05, "loss": 0.4622, "step": 1155 }, { "epoch": 0.2, "learning_rate": 4.9999915603359285e-05, "loss": 0.481, "step": 1156 }, { "epoch": 0.2, "learning_rate": 4.999989580663168e-05, "loss": 0.4888, "step": 1157 }, { "epoch": 0.2, "learning_rate": 4.9999873926044765e-05, "loss": 0.4692, "step": 1158 }, { "epoch": 0.2, "learning_rate": 4.999984996160057e-05, "loss": 0.4648, "step": 1159 }, { "epoch": 0.2, "learning_rate": 4.999982391330131e-05, "loss": 0.4932, "step": 1160 }, { "epoch": 0.2, "learning_rate": 4.99997957811494e-05, "loss": 0.4609, "step": 1161 }, { "epoch": 0.2, "learning_rate": 4.999976556514744e-05, "loss": 0.4553, "step": 1162 }, { "epoch": 0.2, "learning_rate": 4.9999733265298246e-05, "loss": 0.4935, "step": 1163 }, { "epoch": 0.2, "learning_rate": 4.9999698881604796e-05, "loss": 0.4666, "step": 1164 }, { "epoch": 0.2, "learning_rate": 4.9999662414070265e-05, "loss": 0.4574, "step": 1165 }, { "epoch": 0.2, "learning_rate": 4.999962386269805e-05, "loss": 0.472, "step": 1166 }, { "epoch": 0.2, "learning_rate": 4.9999583227491706e-05, "loss": 0.4853, "step": 1167 }, { "epoch": 0.2, "learning_rate": 4.999954050845501e-05, "loss": 0.4628, "step": 1168 }, { "epoch": 0.2, "learning_rate": 4.999949570559191e-05, "loss": 0.4675, "step": 1169 }, { "epoch": 0.2, "learning_rate": 4.999944881890656e-05, "loss": 0.4774, "step": 1170 }, { "epoch": 0.2, "learning_rate": 4.9999399848403296e-05, "loss": 0.474, "step": 1171 }, { "epoch": 0.2, "learning_rate": 4.9999348794086644e-05, "loss": 0.4526, "step": 1172 }, { "epoch": 0.2, "learning_rate": 4.999929565596137e-05, "loss": 0.4744, "step": 1173 }, { "epoch": 0.2, "learning_rate": 4.999924043403236e-05, "loss": 0.4713, "step": 1174 }, { "epoch": 0.2, "learning_rate": 4.999918312830474e-05, "loss": 0.4588, "step": 1175 }, { "epoch": 0.21, "learning_rate": 4.999912373878381e-05, "loss": 0.4681, "step": 1176 }, { "epoch": 0.21, "learning_rate": 4.9999062265475084e-05, "loss": 0.4949, "step": 1177 }, { "epoch": 0.21, "learning_rate": 4.999899870838425e-05, "loss": 0.4757, "step": 1178 }, { "epoch": 0.21, "learning_rate": 4.999893306751719e-05, "loss": 0.4758, "step": 1179 }, { "epoch": 0.21, "learning_rate": 4.999886534287999e-05, "loss": 0.4811, "step": 1180 }, { "epoch": 0.21, "learning_rate": 4.9998795534478914e-05, "loss": 0.4769, "step": 1181 }, { "epoch": 0.21, "learning_rate": 4.999872364232044e-05, "loss": 0.4689, "step": 1182 }, { "epoch": 0.21, "learning_rate": 4.9998649666411215e-05, "loss": 0.4733, "step": 1183 }, { "epoch": 0.21, "learning_rate": 4.999857360675809e-05, "loss": 0.4923, "step": 1184 }, { "epoch": 0.21, "learning_rate": 4.9998495463368126e-05, "loss": 0.4746, "step": 1185 }, { "epoch": 0.21, "learning_rate": 4.9998415236248534e-05, "loss": 0.4486, "step": 1186 }, { "epoch": 0.21, "learning_rate": 4.9998332925406764e-05, "loss": 0.4719, "step": 1187 }, { "epoch": 0.21, "learning_rate": 4.999824853085044e-05, "loss": 0.4672, "step": 1188 }, { "epoch": 0.21, "learning_rate": 4.999816205258736e-05, "loss": 0.4635, "step": 1189 }, { "epoch": 0.21, "learning_rate": 4.999807349062555e-05, "loss": 0.4777, "step": 1190 }, { "epoch": 0.21, "learning_rate": 4.999798284497321e-05, "loss": 0.4759, "step": 1191 }, { "epoch": 0.21, "learning_rate": 4.999789011563874e-05, "loss": 0.4628, "step": 1192 }, { "epoch": 0.21, "learning_rate": 4.999779530263071e-05, "loss": 0.4858, "step": 1193 }, { "epoch": 0.21, "learning_rate": 4.999769840595791e-05, "loss": 0.4717, "step": 1194 }, { "epoch": 0.21, "learning_rate": 4.999759942562933e-05, "loss": 0.4742, "step": 1195 }, { "epoch": 0.21, "learning_rate": 4.999749836165412e-05, "loss": 0.4648, "step": 1196 }, { "epoch": 0.21, "learning_rate": 4.999739521404164e-05, "loss": 0.4888, "step": 1197 }, { "epoch": 0.21, "learning_rate": 4.9997289982801446e-05, "loss": 0.4906, "step": 1198 }, { "epoch": 0.21, "learning_rate": 4.999718266794328e-05, "loss": 0.4737, "step": 1199 }, { "epoch": 0.21, "learning_rate": 4.99970732694771e-05, "loss": 0.4657, "step": 1200 }, { "epoch": 0.21, "learning_rate": 4.999696178741301e-05, "loss": 0.4597, "step": 1201 }, { "epoch": 0.21, "learning_rate": 4.9996848221761364e-05, "loss": 0.4709, "step": 1202 }, { "epoch": 0.21, "learning_rate": 4.999673257253265e-05, "loss": 0.468, "step": 1203 }, { "epoch": 0.21, "learning_rate": 4.999661483973761e-05, "loss": 0.4687, "step": 1204 }, { "epoch": 0.21, "learning_rate": 4.999649502338712e-05, "loss": 0.4772, "step": 1205 }, { "epoch": 0.21, "learning_rate": 4.99963731234923e-05, "loss": 0.4737, "step": 1206 }, { "epoch": 0.21, "learning_rate": 4.999624914006443e-05, "loss": 0.4569, "step": 1207 }, { "epoch": 0.21, "learning_rate": 4.9996123073114986e-05, "loss": 0.4658, "step": 1208 }, { "epoch": 0.21, "learning_rate": 4.999599492265565e-05, "loss": 0.4603, "step": 1209 }, { "epoch": 0.21, "learning_rate": 4.999586468869829e-05, "loss": 0.4777, "step": 1210 }, { "epoch": 0.21, "learning_rate": 4.9995732371254975e-05, "loss": 0.4663, "step": 1211 }, { "epoch": 0.21, "learning_rate": 4.9995597970337944e-05, "loss": 0.4807, "step": 1212 }, { "epoch": 0.21, "learning_rate": 4.999546148595967e-05, "loss": 0.4826, "step": 1213 }, { "epoch": 0.21, "learning_rate": 4.999532291813276e-05, "loss": 0.477, "step": 1214 }, { "epoch": 0.21, "learning_rate": 4.999518226687008e-05, "loss": 0.4611, "step": 1215 }, { "epoch": 0.21, "learning_rate": 4.999503953218463e-05, "loss": 0.4591, "step": 1216 }, { "epoch": 0.21, "learning_rate": 4.999489471408965e-05, "loss": 0.4646, "step": 1217 }, { "epoch": 0.21, "learning_rate": 4.999474781259855e-05, "loss": 0.4755, "step": 1218 }, { "epoch": 0.21, "learning_rate": 4.999459882772492e-05, "loss": 0.4796, "step": 1219 }, { "epoch": 0.21, "learning_rate": 4.999444775948258e-05, "loss": 0.4814, "step": 1220 }, { "epoch": 0.21, "learning_rate": 4.9994294607885494e-05, "loss": 0.4767, "step": 1221 }, { "epoch": 0.21, "learning_rate": 4.999413937294787e-05, "loss": 0.4821, "step": 1222 }, { "epoch": 0.21, "learning_rate": 4.999398205468408e-05, "loss": 0.4763, "step": 1223 }, { "epoch": 0.21, "learning_rate": 4.9993822653108694e-05, "loss": 0.4756, "step": 1224 }, { "epoch": 0.21, "learning_rate": 4.999366116823646e-05, "loss": 0.4778, "step": 1225 }, { "epoch": 0.21, "learning_rate": 4.9993497600082355e-05, "loss": 0.4633, "step": 1226 }, { "epoch": 0.21, "learning_rate": 4.9993331948661525e-05, "loss": 0.4715, "step": 1227 }, { "epoch": 0.21, "learning_rate": 4.99931642139893e-05, "loss": 0.4606, "step": 1228 }, { "epoch": 0.21, "learning_rate": 4.999299439608122e-05, "loss": 0.4706, "step": 1229 }, { "epoch": 0.21, "learning_rate": 4.9992822494953024e-05, "loss": 0.4565, "step": 1230 }, { "epoch": 0.21, "learning_rate": 4.9992648510620617e-05, "loss": 0.452, "step": 1231 }, { "epoch": 0.21, "learning_rate": 4.999247244310012e-05, "loss": 0.4746, "step": 1232 }, { "epoch": 0.21, "learning_rate": 4.9992294292407845e-05, "loss": 0.4811, "step": 1233 }, { "epoch": 0.22, "learning_rate": 4.999211405856028e-05, "loss": 0.4773, "step": 1234 }, { "epoch": 0.22, "learning_rate": 4.999193174157413e-05, "loss": 0.4789, "step": 1235 }, { "epoch": 0.22, "learning_rate": 4.999174734146627e-05, "loss": 0.4614, "step": 1236 }, { "epoch": 0.22, "learning_rate": 4.9991560858253784e-05, "loss": 0.4817, "step": 1237 }, { "epoch": 0.22, "learning_rate": 4.9991372291953934e-05, "loss": 0.4789, "step": 1238 }, { "epoch": 0.22, "learning_rate": 4.99911816425842e-05, "loss": 0.4441, "step": 1239 }, { "epoch": 0.22, "learning_rate": 4.999098891016224e-05, "loss": 0.461, "step": 1240 }, { "epoch": 0.22, "learning_rate": 4.999079409470589e-05, "loss": 0.4785, "step": 1241 }, { "epoch": 0.22, "learning_rate": 4.9990597196233194e-05, "loss": 0.4753, "step": 1242 }, { "epoch": 0.22, "learning_rate": 4.9990398214762395e-05, "loss": 0.4688, "step": 1243 }, { "epoch": 0.22, "learning_rate": 4.9990197150311925e-05, "loss": 0.4576, "step": 1244 }, { "epoch": 0.22, "learning_rate": 4.9989994002900394e-05, "loss": 0.4703, "step": 1245 }, { "epoch": 0.22, "learning_rate": 4.998978877254663e-05, "loss": 0.4698, "step": 1246 }, { "epoch": 0.22, "learning_rate": 4.998958145926963e-05, "loss": 0.4525, "step": 1247 }, { "epoch": 0.22, "learning_rate": 4.9989372063088596e-05, "loss": 0.4695, "step": 1248 }, { "epoch": 0.22, "learning_rate": 4.998916058402293e-05, "loss": 0.4674, "step": 1249 }, { "epoch": 0.22, "learning_rate": 4.998894702209221e-05, "loss": 0.4738, "step": 1250 }, { "epoch": 0.22, "learning_rate": 4.998873137731622e-05, "loss": 0.4649, "step": 1251 }, { "epoch": 0.22, "learning_rate": 4.9988513649714935e-05, "loss": 0.4717, "step": 1252 }, { "epoch": 0.22, "learning_rate": 4.9988293839308506e-05, "loss": 0.4512, "step": 1253 }, { "epoch": 0.22, "learning_rate": 4.998807194611731e-05, "loss": 0.4576, "step": 1254 }, { "epoch": 0.22, "learning_rate": 4.998784797016188e-05, "loss": 0.4811, "step": 1255 }, { "epoch": 0.22, "learning_rate": 4.9987621911462976e-05, "loss": 0.4697, "step": 1256 }, { "epoch": 0.22, "learning_rate": 4.9987393770041524e-05, "loss": 0.4917, "step": 1257 }, { "epoch": 0.22, "learning_rate": 4.998716354591865e-05, "loss": 0.4743, "step": 1258 }, { "epoch": 0.22, "learning_rate": 4.9986931239115706e-05, "loss": 0.4796, "step": 1259 }, { "epoch": 0.22, "learning_rate": 4.9986696849654165e-05, "loss": 0.4699, "step": 1260 }, { "epoch": 0.22, "learning_rate": 4.998646037755576e-05, "loss": 0.4694, "step": 1261 }, { "epoch": 0.22, "learning_rate": 4.998622182284238e-05, "loss": 0.4432, "step": 1262 }, { "epoch": 0.22, "learning_rate": 4.9985981185536137e-05, "loss": 0.4705, "step": 1263 }, { "epoch": 0.22, "learning_rate": 4.9985738465659306e-05, "loss": 0.4822, "step": 1264 }, { "epoch": 0.22, "learning_rate": 4.9985493663234365e-05, "loss": 0.4811, "step": 1265 }, { "epoch": 0.22, "learning_rate": 4.998524677828399e-05, "loss": 0.4654, "step": 1266 }, { "epoch": 0.22, "learning_rate": 4.9984997810831046e-05, "loss": 0.4898, "step": 1267 }, { "epoch": 0.22, "learning_rate": 4.9984746760898596e-05, "loss": 0.4756, "step": 1268 }, { "epoch": 0.22, "learning_rate": 4.998449362850989e-05, "loss": 0.4754, "step": 1269 }, { "epoch": 0.22, "learning_rate": 4.998423841368836e-05, "loss": 0.4701, "step": 1270 }, { "epoch": 0.22, "learning_rate": 4.998398111645765e-05, "loss": 0.4508, "step": 1271 }, { "epoch": 0.22, "learning_rate": 4.99837217368416e-05, "loss": 0.4607, "step": 1272 }, { "epoch": 0.22, "learning_rate": 4.998346027486423e-05, "loss": 0.4565, "step": 1273 }, { "epoch": 0.22, "learning_rate": 4.9983196730549744e-05, "loss": 0.4705, "step": 1274 }, { "epoch": 0.22, "learning_rate": 4.998293110392256e-05, "loss": 0.4625, "step": 1275 }, { "epoch": 0.22, "learning_rate": 4.998266339500727e-05, "loss": 0.4597, "step": 1276 }, { "epoch": 0.22, "learning_rate": 4.9982393603828685e-05, "loss": 0.4828, "step": 1277 }, { "epoch": 0.22, "learning_rate": 4.998212173041178e-05, "loss": 0.4636, "step": 1278 }, { "epoch": 0.22, "learning_rate": 4.998184777478173e-05, "loss": 0.455, "step": 1279 }, { "epoch": 0.22, "learning_rate": 4.9981571736963926e-05, "loss": 0.4729, "step": 1280 }, { "epoch": 0.22, "learning_rate": 4.998129361698392e-05, "loss": 0.4532, "step": 1281 }, { "epoch": 0.22, "learning_rate": 4.9981013414867465e-05, "loss": 0.4581, "step": 1282 }, { "epoch": 0.22, "learning_rate": 4.9980731130640526e-05, "loss": 0.4679, "step": 1283 }, { "epoch": 0.22, "learning_rate": 4.9980446764329246e-05, "loss": 0.4878, "step": 1284 }, { "epoch": 0.22, "learning_rate": 4.998016031595995e-05, "loss": 0.4682, "step": 1285 }, { "epoch": 0.22, "learning_rate": 4.997987178555918e-05, "loss": 0.4871, "step": 1286 }, { "epoch": 0.22, "learning_rate": 4.997958117315365e-05, "loss": 0.4938, "step": 1287 }, { "epoch": 0.22, "learning_rate": 4.997928847877029e-05, "loss": 0.4859, "step": 1288 }, { "epoch": 0.22, "learning_rate": 4.997899370243618e-05, "loss": 0.4657, "step": 1289 }, { "epoch": 0.22, "learning_rate": 4.9978696844178655e-05, "loss": 0.4786, "step": 1290 }, { "epoch": 0.23, "learning_rate": 4.997839790402519e-05, "loss": 0.4797, "step": 1291 }, { "epoch": 0.23, "learning_rate": 4.9978096882003466e-05, "loss": 0.4609, "step": 1292 }, { "epoch": 0.23, "learning_rate": 4.997779377814138e-05, "loss": 0.4615, "step": 1293 }, { "epoch": 0.23, "learning_rate": 4.997748859246699e-05, "loss": 0.4446, "step": 1294 }, { "epoch": 0.23, "learning_rate": 4.997718132500857e-05, "loss": 0.5032, "step": 1295 }, { "epoch": 0.23, "learning_rate": 4.997687197579457e-05, "loss": 0.4761, "step": 1296 }, { "epoch": 0.23, "learning_rate": 4.9976560544853654e-05, "loss": 0.488, "step": 1297 }, { "epoch": 0.23, "learning_rate": 4.9976247032214655e-05, "loss": 0.467, "step": 1298 }, { "epoch": 0.23, "learning_rate": 4.99759314379066e-05, "loss": 0.4822, "step": 1299 }, { "epoch": 0.23, "learning_rate": 4.9975613761958746e-05, "loss": 0.4698, "step": 1300 }, { "epoch": 0.23, "learning_rate": 4.9975294004400485e-05, "loss": 0.456, "step": 1301 }, { "epoch": 0.23, "learning_rate": 4.997497216526145e-05, "loss": 0.4712, "step": 1302 }, { "epoch": 0.23, "learning_rate": 4.9974648244571445e-05, "loss": 0.4699, "step": 1303 }, { "epoch": 0.23, "learning_rate": 4.997432224236048e-05, "loss": 0.4919, "step": 1304 }, { "epoch": 0.23, "learning_rate": 4.997399415865872e-05, "loss": 0.477, "step": 1305 }, { "epoch": 0.23, "learning_rate": 4.997366399349658e-05, "loss": 0.4744, "step": 1306 }, { "epoch": 0.23, "learning_rate": 4.997333174690462e-05, "loss": 0.4622, "step": 1307 }, { "epoch": 0.23, "learning_rate": 4.997299741891363e-05, "loss": 0.4637, "step": 1308 }, { "epoch": 0.23, "learning_rate": 4.997266100955455e-05, "loss": 0.4551, "step": 1309 }, { "epoch": 0.23, "learning_rate": 4.997232251885856e-05, "loss": 0.4585, "step": 1310 }, { "epoch": 0.23, "learning_rate": 4.9971981946856996e-05, "loss": 0.4655, "step": 1311 }, { "epoch": 0.23, "learning_rate": 4.9971639293581404e-05, "loss": 0.484, "step": 1312 }, { "epoch": 0.23, "learning_rate": 4.997129455906353e-05, "loss": 0.4511, "step": 1313 }, { "epoch": 0.23, "learning_rate": 4.997094774333528e-05, "loss": 0.4803, "step": 1314 }, { "epoch": 0.23, "learning_rate": 4.9970598846428776e-05, "loss": 0.4855, "step": 1315 }, { "epoch": 0.23, "learning_rate": 4.9970247868376355e-05, "loss": 0.4711, "step": 1316 }, { "epoch": 0.23, "learning_rate": 4.9969894809210515e-05, "loss": 0.4681, "step": 1317 }, { "epoch": 0.23, "learning_rate": 4.9969539668963945e-05, "loss": 0.4786, "step": 1318 }, { "epoch": 0.23, "learning_rate": 4.9969182447669544e-05, "loss": 0.4537, "step": 1319 }, { "epoch": 0.23, "learning_rate": 4.996882314536039e-05, "loss": 0.4619, "step": 1320 }, { "epoch": 0.23, "learning_rate": 4.996846176206978e-05, "loss": 0.4537, "step": 1321 }, { "epoch": 0.23, "learning_rate": 4.996809829783116e-05, "loss": 0.4668, "step": 1322 }, { "epoch": 0.23, "learning_rate": 4.99677327526782e-05, "loss": 0.4623, "step": 1323 }, { "epoch": 0.23, "learning_rate": 4.996736512664476e-05, "loss": 0.4699, "step": 1324 }, { "epoch": 0.23, "learning_rate": 4.996699541976488e-05, "loss": 0.4698, "step": 1325 }, { "epoch": 0.23, "learning_rate": 4.996662363207281e-05, "loss": 0.4656, "step": 1326 }, { "epoch": 0.23, "learning_rate": 4.9966249763602984e-05, "loss": 0.496, "step": 1327 }, { "epoch": 0.23, "learning_rate": 4.996587381439002e-05, "loss": 0.4654, "step": 1328 }, { "epoch": 0.23, "learning_rate": 4.996549578446874e-05, "loss": 0.4594, "step": 1329 }, { "epoch": 0.23, "learning_rate": 4.9965115673874166e-05, "loss": 0.4379, "step": 1330 }, { "epoch": 0.23, "learning_rate": 4.996473348264149e-05, "loss": 0.4848, "step": 1331 }, { "epoch": 0.23, "learning_rate": 4.996434921080611e-05, "loss": 0.4589, "step": 1332 }, { "epoch": 0.23, "learning_rate": 4.996396285840363e-05, "loss": 0.4671, "step": 1333 }, { "epoch": 0.23, "learning_rate": 4.9963574425469806e-05, "loss": 0.449, "step": 1334 }, { "epoch": 0.23, "learning_rate": 4.9963183912040635e-05, "loss": 0.472, "step": 1335 }, { "epoch": 0.23, "learning_rate": 4.996279131815228e-05, "loss": 0.4548, "step": 1336 }, { "epoch": 0.23, "learning_rate": 4.99623966438411e-05, "loss": 0.4595, "step": 1337 }, { "epoch": 0.23, "learning_rate": 4.996199988914365e-05, "loss": 0.4622, "step": 1338 }, { "epoch": 0.23, "learning_rate": 4.996160105409667e-05, "loss": 0.4508, "step": 1339 }, { "epoch": 0.23, "learning_rate": 4.9961200138737104e-05, "loss": 0.4662, "step": 1340 }, { "epoch": 0.23, "learning_rate": 4.996079714310209e-05, "loss": 0.4589, "step": 1341 }, { "epoch": 0.23, "learning_rate": 4.996039206722893e-05, "loss": 0.4494, "step": 1342 }, { "epoch": 0.23, "learning_rate": 4.995998491115517e-05, "loss": 0.4645, "step": 1343 }, { "epoch": 0.23, "learning_rate": 4.9959575674918495e-05, "loss": 0.4412, "step": 1344 }, { "epoch": 0.23, "learning_rate": 4.995916435855682e-05, "loss": 0.4647, "step": 1345 }, { "epoch": 0.23, "learning_rate": 4.9958750962108245e-05, "loss": 0.4585, "step": 1346 }, { "epoch": 0.23, "learning_rate": 4.995833548561104e-05, "loss": 0.4785, "step": 1347 }, { "epoch": 0.24, "learning_rate": 4.995791792910369e-05, "loss": 0.4457, "step": 1348 }, { "epoch": 0.24, "learning_rate": 4.995749829262488e-05, "loss": 0.4627, "step": 1349 }, { "epoch": 0.24, "learning_rate": 4.9957076576213465e-05, "loss": 0.462, "step": 1350 }, { "epoch": 0.24, "learning_rate": 4.995665277990849e-05, "loss": 0.4449, "step": 1351 }, { "epoch": 0.24, "learning_rate": 4.995622690374923e-05, "loss": 0.451, "step": 1352 }, { "epoch": 0.24, "learning_rate": 4.9955798947775124e-05, "loss": 0.4801, "step": 1353 }, { "epoch": 0.24, "learning_rate": 4.99553689120258e-05, "loss": 0.4351, "step": 1354 }, { "epoch": 0.24, "learning_rate": 4.995493679654108e-05, "loss": 0.4909, "step": 1355 }, { "epoch": 0.24, "learning_rate": 4.9954502601361005e-05, "loss": 0.4458, "step": 1356 }, { "epoch": 0.24, "learning_rate": 4.9954066326525764e-05, "loss": 0.4728, "step": 1357 }, { "epoch": 0.24, "learning_rate": 4.9953627972075776e-05, "loss": 0.4683, "step": 1358 }, { "epoch": 0.24, "learning_rate": 4.9953187538051644e-05, "loss": 0.4746, "step": 1359 }, { "epoch": 0.24, "learning_rate": 4.995274502449416e-05, "loss": 0.4662, "step": 1360 }, { "epoch": 0.24, "learning_rate": 4.995230043144429e-05, "loss": 0.4636, "step": 1361 }, { "epoch": 0.24, "learning_rate": 4.995185375894323e-05, "loss": 0.4736, "step": 1362 }, { "epoch": 0.24, "learning_rate": 4.995140500703234e-05, "loss": 0.4629, "step": 1363 }, { "epoch": 0.24, "learning_rate": 4.9950954175753195e-05, "loss": 0.4799, "step": 1364 }, { "epoch": 0.24, "learning_rate": 4.995050126514753e-05, "loss": 0.4612, "step": 1365 }, { "epoch": 0.24, "learning_rate": 4.99500462752573e-05, "loss": 0.478, "step": 1366 }, { "epoch": 0.24, "learning_rate": 4.9949589206124645e-05, "loss": 0.4783, "step": 1367 }, { "epoch": 0.24, "learning_rate": 4.994913005779191e-05, "loss": 0.4745, "step": 1368 }, { "epoch": 0.24, "learning_rate": 4.994866883030159e-05, "loss": 0.4785, "step": 1369 }, { "epoch": 0.24, "learning_rate": 4.9948205523696433e-05, "loss": 0.4648, "step": 1370 }, { "epoch": 0.24, "learning_rate": 4.994774013801933e-05, "loss": 0.4714, "step": 1371 }, { "epoch": 0.24, "learning_rate": 4.994727267331338e-05, "loss": 0.4707, "step": 1372 }, { "epoch": 0.24, "learning_rate": 4.9946803129621895e-05, "loss": 0.4575, "step": 1373 }, { "epoch": 0.24, "learning_rate": 4.9946331506988357e-05, "loss": 0.4689, "step": 1374 }, { "epoch": 0.24, "learning_rate": 4.994585780545644e-05, "loss": 0.4473, "step": 1375 }, { "epoch": 0.24, "learning_rate": 4.9945382025070016e-05, "loss": 0.4759, "step": 1376 }, { "epoch": 0.24, "learning_rate": 4.9944904165873155e-05, "loss": 0.452, "step": 1377 }, { "epoch": 0.24, "learning_rate": 4.9944424227910106e-05, "loss": 0.4734, "step": 1378 }, { "epoch": 0.24, "learning_rate": 4.994394221122534e-05, "loss": 0.4752, "step": 1379 }, { "epoch": 0.24, "learning_rate": 4.994345811586348e-05, "loss": 0.4581, "step": 1380 }, { "epoch": 0.24, "learning_rate": 4.994297194186936e-05, "loss": 0.4644, "step": 1381 }, { "epoch": 0.24, "learning_rate": 4.994248368928802e-05, "loss": 0.4484, "step": 1382 }, { "epoch": 0.24, "learning_rate": 4.994199335816468e-05, "loss": 0.4569, "step": 1383 }, { "epoch": 0.24, "learning_rate": 4.994150094854474e-05, "loss": 0.4808, "step": 1384 }, { "epoch": 0.24, "learning_rate": 4.9941006460473814e-05, "loss": 0.4501, "step": 1385 }, { "epoch": 0.24, "learning_rate": 4.9940509893997704e-05, "loss": 0.465, "step": 1386 }, { "epoch": 0.24, "learning_rate": 4.994001124916239e-05, "loss": 0.4682, "step": 1387 }, { "epoch": 0.24, "learning_rate": 4.993951052601405e-05, "loss": 0.4719, "step": 1388 }, { "epoch": 0.24, "learning_rate": 4.993900772459909e-05, "loss": 0.471, "step": 1389 }, { "epoch": 0.24, "learning_rate": 4.993850284496404e-05, "loss": 0.4386, "step": 1390 }, { "epoch": 0.24, "learning_rate": 4.993799588715568e-05, "loss": 0.4646, "step": 1391 }, { "epoch": 0.24, "learning_rate": 4.993748685122096e-05, "loss": 0.4677, "step": 1392 }, { "epoch": 0.24, "learning_rate": 4.9936975737207026e-05, "loss": 0.4609, "step": 1393 }, { "epoch": 0.24, "learning_rate": 4.9936462545161214e-05, "loss": 0.4734, "step": 1394 }, { "epoch": 0.24, "learning_rate": 4.9935947275131046e-05, "loss": 0.4625, "step": 1395 }, { "epoch": 0.24, "learning_rate": 4.993542992716426e-05, "loss": 0.4693, "step": 1396 }, { "epoch": 0.24, "learning_rate": 4.9934910501308755e-05, "loss": 0.4637, "step": 1397 }, { "epoch": 0.24, "learning_rate": 4.993438899761265e-05, "loss": 0.4754, "step": 1398 }, { "epoch": 0.24, "learning_rate": 4.9933865416124245e-05, "loss": 0.4824, "step": 1399 }, { "epoch": 0.24, "learning_rate": 4.9933339756892025e-05, "loss": 0.4402, "step": 1400 }, { "epoch": 0.24, "learning_rate": 4.993281201996468e-05, "loss": 0.4559, "step": 1401 }, { "epoch": 0.24, "learning_rate": 4.993228220539108e-05, "loss": 0.4605, "step": 1402 }, { "epoch": 0.24, "learning_rate": 4.99317503132203e-05, "loss": 0.4453, "step": 1403 }, { "epoch": 0.24, "learning_rate": 4.9931216343501606e-05, "loss": 0.4617, "step": 1404 }, { "epoch": 0.24, "learning_rate": 4.9930680296284446e-05, "loss": 0.46, "step": 1405 }, { "epoch": 0.25, "learning_rate": 4.993014217161846e-05, "loss": 0.437, "step": 1406 }, { "epoch": 0.25, "learning_rate": 4.992960196955351e-05, "loss": 0.4509, "step": 1407 }, { "epoch": 0.25, "learning_rate": 4.992905969013961e-05, "loss": 0.457, "step": 1408 }, { "epoch": 0.25, "learning_rate": 4.992851533342699e-05, "loss": 0.4536, "step": 1409 }, { "epoch": 0.25, "learning_rate": 4.992796889946605e-05, "loss": 0.4664, "step": 1410 }, { "epoch": 0.25, "learning_rate": 4.9927420388307415e-05, "loss": 0.4636, "step": 1411 }, { "epoch": 0.25, "learning_rate": 4.992686980000189e-05, "loss": 0.4694, "step": 1412 }, { "epoch": 0.25, "learning_rate": 4.992631713460046e-05, "loss": 0.4714, "step": 1413 }, { "epoch": 0.25, "learning_rate": 4.992576239215432e-05, "loss": 0.4608, "step": 1414 }, { "epoch": 0.25, "learning_rate": 4.992520557271484e-05, "loss": 0.4647, "step": 1415 }, { "epoch": 0.25, "learning_rate": 4.9924646676333574e-05, "loss": 0.4614, "step": 1416 }, { "epoch": 0.25, "learning_rate": 4.992408570306232e-05, "loss": 0.4601, "step": 1417 }, { "epoch": 0.25, "learning_rate": 4.9923522652953004e-05, "loss": 0.4528, "step": 1418 }, { "epoch": 0.25, "learning_rate": 4.992295752605779e-05, "loss": 0.4521, "step": 1419 }, { "epoch": 0.25, "learning_rate": 4.992239032242902e-05, "loss": 0.4583, "step": 1420 }, { "epoch": 0.25, "learning_rate": 4.992182104211921e-05, "loss": 0.4847, "step": 1421 }, { "epoch": 0.25, "learning_rate": 4.992124968518111e-05, "loss": 0.4436, "step": 1422 }, { "epoch": 0.25, "learning_rate": 4.9920676251667616e-05, "loss": 0.4602, "step": 1423 }, { "epoch": 0.25, "learning_rate": 4.992010074163184e-05, "loss": 0.4639, "step": 1424 }, { "epoch": 0.25, "learning_rate": 4.991952315512708e-05, "loss": 0.4516, "step": 1425 }, { "epoch": 0.25, "learning_rate": 4.9918943492206846e-05, "loss": 0.467, "step": 1426 }, { "epoch": 0.25, "learning_rate": 4.9918361752924815e-05, "loss": 0.4752, "step": 1427 }, { "epoch": 0.25, "learning_rate": 4.991777793733486e-05, "loss": 0.463, "step": 1428 }, { "epoch": 0.25, "learning_rate": 4.9917192045491065e-05, "loss": 0.4383, "step": 1429 }, { "epoch": 0.25, "learning_rate": 4.9916604077447685e-05, "loss": 0.4683, "step": 1430 }, { "epoch": 0.25, "learning_rate": 4.9916014033259175e-05, "loss": 0.4662, "step": 1431 }, { "epoch": 0.25, "learning_rate": 4.991542191298018e-05, "loss": 0.4699, "step": 1432 }, { "epoch": 0.25, "learning_rate": 4.991482771666555e-05, "loss": 0.4521, "step": 1433 }, { "epoch": 0.25, "learning_rate": 4.99142314443703e-05, "loss": 0.4869, "step": 1434 }, { "epoch": 0.25, "learning_rate": 4.991363309614968e-05, "loss": 0.4855, "step": 1435 }, { "epoch": 0.25, "learning_rate": 4.991303267205908e-05, "loss": 0.4595, "step": 1436 }, { "epoch": 0.25, "learning_rate": 4.991243017215413e-05, "loss": 0.4583, "step": 1437 }, { "epoch": 0.25, "learning_rate": 4.991182559649062e-05, "loss": 0.4631, "step": 1438 }, { "epoch": 0.25, "learning_rate": 4.9911218945124546e-05, "loss": 0.4748, "step": 1439 }, { "epoch": 0.25, "learning_rate": 4.9910610218112106e-05, "loss": 0.4614, "step": 1440 }, { "epoch": 0.25, "learning_rate": 4.990999941550965e-05, "loss": 0.4663, "step": 1441 }, { "epoch": 0.25, "learning_rate": 4.990938653737378e-05, "loss": 0.4576, "step": 1442 }, { "epoch": 0.25, "learning_rate": 4.990877158376123e-05, "loss": 0.4633, "step": 1443 }, { "epoch": 0.25, "learning_rate": 4.9908154554728984e-05, "loss": 0.4619, "step": 1444 }, { "epoch": 0.25, "learning_rate": 4.9907535450334165e-05, "loss": 0.4611, "step": 1445 }, { "epoch": 0.25, "learning_rate": 4.9906914270634114e-05, "loss": 0.4792, "step": 1446 }, { "epoch": 0.25, "learning_rate": 4.990629101568638e-05, "loss": 0.4607, "step": 1447 }, { "epoch": 0.25, "learning_rate": 4.9905665685548674e-05, "loss": 0.4514, "step": 1448 }, { "epoch": 0.25, "learning_rate": 4.990503828027891e-05, "loss": 0.4617, "step": 1449 }, { "epoch": 0.25, "learning_rate": 4.99044087999352e-05, "loss": 0.4733, "step": 1450 }, { "epoch": 0.25, "learning_rate": 4.990377724457584e-05, "loss": 0.4812, "step": 1451 }, { "epoch": 0.25, "learning_rate": 4.990314361425934e-05, "loss": 0.46, "step": 1452 }, { "epoch": 0.25, "learning_rate": 4.990250790904436e-05, "loss": 0.4467, "step": 1453 }, { "epoch": 0.25, "learning_rate": 4.990187012898979e-05, "loss": 0.4424, "step": 1454 }, { "epoch": 0.25, "learning_rate": 4.99012302741547e-05, "loss": 0.475, "step": 1455 }, { "epoch": 0.25, "learning_rate": 4.990058834459835e-05, "loss": 0.4562, "step": 1456 }, { "epoch": 0.25, "learning_rate": 4.989994434038018e-05, "loss": 0.449, "step": 1457 }, { "epoch": 0.25, "learning_rate": 4.989929826155985e-05, "loss": 0.5058, "step": 1458 }, { "epoch": 0.25, "learning_rate": 4.98986501081972e-05, "loss": 0.4498, "step": 1459 }, { "epoch": 0.25, "learning_rate": 4.989799988035224e-05, "loss": 0.4602, "step": 1460 }, { "epoch": 0.25, "learning_rate": 4.989734757808522e-05, "loss": 0.4758, "step": 1461 }, { "epoch": 0.25, "learning_rate": 4.989669320145653e-05, "loss": 0.4584, "step": 1462 }, { "epoch": 0.26, "learning_rate": 4.989603675052679e-05, "loss": 0.4578, "step": 1463 }, { "epoch": 0.26, "learning_rate": 4.9895378225356785e-05, "loss": 0.4725, "step": 1464 }, { "epoch": 0.26, "learning_rate": 4.989471762600752e-05, "loss": 0.4423, "step": 1465 }, { "epoch": 0.26, "learning_rate": 4.9894054952540165e-05, "loss": 0.4782, "step": 1466 }, { "epoch": 0.26, "learning_rate": 4.989339020501611e-05, "loss": 0.4633, "step": 1467 }, { "epoch": 0.26, "learning_rate": 4.9892723383496905e-05, "loss": 0.4701, "step": 1468 }, { "epoch": 0.26, "learning_rate": 4.989205448804431e-05, "loss": 0.4652, "step": 1469 }, { "epoch": 0.26, "learning_rate": 4.989138351872029e-05, "loss": 0.4718, "step": 1470 }, { "epoch": 0.26, "learning_rate": 4.989071047558697e-05, "loss": 0.4681, "step": 1471 }, { "epoch": 0.26, "learning_rate": 4.9890035358706704e-05, "loss": 0.4459, "step": 1472 }, { "epoch": 0.26, "learning_rate": 4.9889358168142e-05, "loss": 0.4719, "step": 1473 }, { "epoch": 0.26, "learning_rate": 4.988867890395559e-05, "loss": 0.4566, "step": 1474 }, { "epoch": 0.26, "learning_rate": 4.988799756621037e-05, "loss": 0.4505, "step": 1475 }, { "epoch": 0.26, "learning_rate": 4.9887314154969466e-05, "loss": 0.4382, "step": 1476 }, { "epoch": 0.26, "learning_rate": 4.9886628670296156e-05, "loss": 0.4604, "step": 1477 }, { "epoch": 0.26, "learning_rate": 4.9885941112253924e-05, "loss": 0.4783, "step": 1478 }, { "epoch": 0.26, "learning_rate": 4.988525148090647e-05, "loss": 0.4534, "step": 1479 }, { "epoch": 0.26, "learning_rate": 4.9884559776317644e-05, "loss": 0.4469, "step": 1480 }, { "epoch": 0.26, "learning_rate": 4.988386599855152e-05, "loss": 0.4634, "step": 1481 }, { "epoch": 0.26, "learning_rate": 4.988317014767235e-05, "loss": 0.4494, "step": 1482 }, { "epoch": 0.26, "learning_rate": 4.988247222374457e-05, "loss": 0.4622, "step": 1483 }, { "epoch": 0.26, "learning_rate": 4.9881772226832846e-05, "loss": 0.4625, "step": 1484 }, { "epoch": 0.26, "learning_rate": 4.988107015700199e-05, "loss": 0.4556, "step": 1485 }, { "epoch": 0.26, "learning_rate": 4.988036601431702e-05, "loss": 0.461, "step": 1486 }, { "epoch": 0.26, "learning_rate": 4.9879659798843173e-05, "loss": 0.4432, "step": 1487 }, { "epoch": 0.26, "learning_rate": 4.987895151064584e-05, "loss": 0.4607, "step": 1488 }, { "epoch": 0.26, "learning_rate": 4.987824114979062e-05, "loss": 0.4751, "step": 1489 }, { "epoch": 0.26, "learning_rate": 4.98775287163433e-05, "loss": 0.448, "step": 1490 }, { "epoch": 0.26, "learning_rate": 4.987681421036989e-05, "loss": 0.4549, "step": 1491 }, { "epoch": 0.26, "learning_rate": 4.9876097631936533e-05, "loss": 0.4632, "step": 1492 }, { "epoch": 0.26, "learning_rate": 4.98753789811096e-05, "loss": 0.4532, "step": 1493 }, { "epoch": 0.26, "learning_rate": 4.987465825795567e-05, "loss": 0.4578, "step": 1494 }, { "epoch": 0.26, "learning_rate": 4.9873935462541476e-05, "loss": 0.4667, "step": 1495 }, { "epoch": 0.26, "learning_rate": 4.9873210594933976e-05, "loss": 0.4608, "step": 1496 }, { "epoch": 0.26, "learning_rate": 4.9872483655200294e-05, "loss": 0.4514, "step": 1497 }, { "epoch": 0.26, "learning_rate": 4.9871754643407745e-05, "loss": 0.4728, "step": 1498 }, { "epoch": 0.26, "learning_rate": 4.9871023559623874e-05, "loss": 0.4586, "step": 1499 }, { "epoch": 0.26, "learning_rate": 4.987029040391638e-05, "loss": 0.4699, "step": 1500 }, { "epoch": 0.26, "learning_rate": 4.9869555176353145e-05, "loss": 0.4641, "step": 1501 }, { "epoch": 0.26, "learning_rate": 4.986881787700229e-05, "loss": 0.4604, "step": 1502 }, { "epoch": 0.26, "learning_rate": 4.98680785059321e-05, "loss": 0.4781, "step": 1503 }, { "epoch": 0.26, "learning_rate": 4.986733706321103e-05, "loss": 0.4541, "step": 1504 }, { "epoch": 0.26, "learning_rate": 4.986659354890778e-05, "loss": 0.4572, "step": 1505 }, { "epoch": 0.26, "learning_rate": 4.986584796309118e-05, "loss": 0.4551, "step": 1506 }, { "epoch": 0.26, "learning_rate": 4.986510030583032e-05, "loss": 0.4588, "step": 1507 }, { "epoch": 0.26, "learning_rate": 4.9864350577194406e-05, "loss": 0.4482, "step": 1508 }, { "epoch": 0.26, "learning_rate": 4.9863598777252905e-05, "loss": 0.4687, "step": 1509 }, { "epoch": 0.26, "learning_rate": 4.9862844906075426e-05, "loss": 0.4556, "step": 1510 }, { "epoch": 0.26, "learning_rate": 4.9862088963731806e-05, "loss": 0.464, "step": 1511 }, { "epoch": 0.26, "learning_rate": 4.986133095029205e-05, "loss": 0.4699, "step": 1512 }, { "epoch": 0.26, "learning_rate": 4.9860570865826365e-05, "loss": 0.4672, "step": 1513 }, { "epoch": 0.26, "learning_rate": 4.9859808710405146e-05, "loss": 0.4519, "step": 1514 }, { "epoch": 0.26, "learning_rate": 4.985904448409898e-05, "loss": 0.4405, "step": 1515 }, { "epoch": 0.26, "learning_rate": 4.9858278186978644e-05, "loss": 0.4273, "step": 1516 }, { "epoch": 0.26, "learning_rate": 4.9857509819115114e-05, "loss": 0.447, "step": 1517 }, { "epoch": 0.26, "learning_rate": 4.9856739380579554e-05, "loss": 0.4436, "step": 1518 }, { "epoch": 0.26, "learning_rate": 4.9855966871443324e-05, "loss": 0.4647, "step": 1519 }, { "epoch": 0.27, "learning_rate": 4.985519229177795e-05, "loss": 0.445, "step": 1520 }, { "epoch": 0.27, "learning_rate": 4.98544156416552e-05, "loss": 0.4431, "step": 1521 }, { "epoch": 0.27, "learning_rate": 4.985363692114699e-05, "loss": 0.4358, "step": 1522 }, { "epoch": 0.27, "learning_rate": 4.985285613032543e-05, "loss": 0.4786, "step": 1523 }, { "epoch": 0.27, "learning_rate": 4.985207326926285e-05, "loss": 0.4363, "step": 1524 }, { "epoch": 0.27, "learning_rate": 4.9851288338031756e-05, "loss": 0.4563, "step": 1525 }, { "epoch": 0.27, "learning_rate": 4.985050133670484e-05, "loss": 0.4565, "step": 1526 }, { "epoch": 0.27, "learning_rate": 4.984971226535499e-05, "loss": 0.4706, "step": 1527 }, { "epoch": 0.27, "learning_rate": 4.98489211240553e-05, "loss": 0.464, "step": 1528 }, { "epoch": 0.27, "learning_rate": 4.984812791287902e-05, "loss": 0.4559, "step": 1529 }, { "epoch": 0.27, "learning_rate": 4.984733263189964e-05, "loss": 0.4498, "step": 1530 }, { "epoch": 0.27, "learning_rate": 4.9846535281190795e-05, "loss": 0.4597, "step": 1531 }, { "epoch": 0.27, "learning_rate": 4.9845735860826335e-05, "loss": 0.4589, "step": 1532 }, { "epoch": 0.27, "learning_rate": 4.984493437088032e-05, "loss": 0.449, "step": 1533 }, { "epoch": 0.27, "learning_rate": 4.9844130811426957e-05, "loss": 0.4636, "step": 1534 }, { "epoch": 0.27, "learning_rate": 4.984332518254068e-05, "loss": 0.4453, "step": 1535 }, { "epoch": 0.27, "learning_rate": 4.9842517484296104e-05, "loss": 0.4651, "step": 1536 }, { "epoch": 0.27, "learning_rate": 4.984170771676803e-05, "loss": 0.4474, "step": 1537 }, { "epoch": 0.27, "learning_rate": 4.984089588003146e-05, "loss": 0.4675, "step": 1538 }, { "epoch": 0.27, "learning_rate": 4.984008197416158e-05, "loss": 0.4505, "step": 1539 }, { "epoch": 0.27, "learning_rate": 4.983926599923377e-05, "loss": 0.4596, "step": 1540 }, { "epoch": 0.27, "learning_rate": 4.983844795532361e-05, "loss": 0.4556, "step": 1541 }, { "epoch": 0.27, "learning_rate": 4.9837627842506856e-05, "loss": 0.4406, "step": 1542 }, { "epoch": 0.27, "learning_rate": 4.9836805660859473e-05, "loss": 0.4623, "step": 1543 }, { "epoch": 0.27, "learning_rate": 4.98359814104576e-05, "loss": 0.4518, "step": 1544 }, { "epoch": 0.27, "learning_rate": 4.9835155091377586e-05, "loss": 0.4753, "step": 1545 }, { "epoch": 0.27, "learning_rate": 4.983432670369594e-05, "loss": 0.454, "step": 1546 }, { "epoch": 0.27, "learning_rate": 4.9833496247489406e-05, "loss": 0.4575, "step": 1547 }, { "epoch": 0.27, "learning_rate": 4.983266372283489e-05, "loss": 0.4521, "step": 1548 }, { "epoch": 0.27, "learning_rate": 4.9831829129809506e-05, "loss": 0.4539, "step": 1549 }, { "epoch": 0.27, "learning_rate": 4.983099246849054e-05, "loss": 0.4727, "step": 1550 }, { "epoch": 0.27, "learning_rate": 4.983015373895548e-05, "loss": 0.4476, "step": 1551 }, { "epoch": 0.27, "learning_rate": 4.982931294128202e-05, "loss": 0.471, "step": 1552 }, { "epoch": 0.27, "learning_rate": 4.982847007554801e-05, "loss": 0.4511, "step": 1553 }, { "epoch": 0.27, "learning_rate": 4.982762514183153e-05, "loss": 0.4583, "step": 1554 }, { "epoch": 0.27, "learning_rate": 4.9826778140210836e-05, "loss": 0.4534, "step": 1555 }, { "epoch": 0.27, "learning_rate": 4.982592907076436e-05, "loss": 0.478, "step": 1556 }, { "epoch": 0.27, "learning_rate": 4.982507793357075e-05, "loss": 0.4545, "step": 1557 }, { "epoch": 0.27, "learning_rate": 4.982422472870883e-05, "loss": 0.4706, "step": 1558 }, { "epoch": 0.27, "learning_rate": 4.982336945625763e-05, "loss": 0.4641, "step": 1559 }, { "epoch": 0.27, "learning_rate": 4.982251211629635e-05, "loss": 0.4511, "step": 1560 }, { "epoch": 0.27, "learning_rate": 4.9821652708904404e-05, "loss": 0.4736, "step": 1561 }, { "epoch": 0.27, "learning_rate": 4.982079123416138e-05, "loss": 0.4554, "step": 1562 }, { "epoch": 0.27, "learning_rate": 4.9819927692147065e-05, "loss": 0.4753, "step": 1563 }, { "epoch": 0.27, "learning_rate": 4.981906208294145e-05, "loss": 0.469, "step": 1564 }, { "epoch": 0.27, "learning_rate": 4.9818194406624686e-05, "loss": 0.4711, "step": 1565 }, { "epoch": 0.27, "learning_rate": 4.9817324663277146e-05, "loss": 0.4641, "step": 1566 }, { "epoch": 0.27, "learning_rate": 4.981645285297938e-05, "loss": 0.4581, "step": 1567 }, { "epoch": 0.27, "learning_rate": 4.981557897581212e-05, "loss": 0.4585, "step": 1568 }, { "epoch": 0.27, "learning_rate": 4.9814703031856324e-05, "loss": 0.4719, "step": 1569 }, { "epoch": 0.27, "learning_rate": 4.981382502119311e-05, "loss": 0.4427, "step": 1570 }, { "epoch": 0.27, "learning_rate": 4.981294494390378e-05, "loss": 0.4486, "step": 1571 }, { "epoch": 0.27, "learning_rate": 4.981206280006986e-05, "loss": 0.4364, "step": 1572 }, { "epoch": 0.27, "learning_rate": 4.9811178589773064e-05, "loss": 0.4587, "step": 1573 }, { "epoch": 0.27, "learning_rate": 4.981029231309525e-05, "loss": 0.4343, "step": 1574 }, { "epoch": 0.27, "learning_rate": 4.980940397011853e-05, "loss": 0.4889, "step": 1575 }, { "epoch": 0.27, "learning_rate": 4.980851356092516e-05, "loss": 0.4499, "step": 1576 }, { "epoch": 0.27, "learning_rate": 4.9807621085597624e-05, "loss": 0.4723, "step": 1577 }, { "epoch": 0.28, "learning_rate": 4.980672654421857e-05, "loss": 0.441, "step": 1578 }, { "epoch": 0.28, "learning_rate": 4.980582993687084e-05, "loss": 0.4523, "step": 1579 }, { "epoch": 0.28, "learning_rate": 4.980493126363749e-05, "loss": 0.4537, "step": 1580 }, { "epoch": 0.28, "learning_rate": 4.9804030524601746e-05, "loss": 0.4695, "step": 1581 }, { "epoch": 0.28, "learning_rate": 4.980312771984703e-05, "loss": 0.4382, "step": 1582 }, { "epoch": 0.28, "learning_rate": 4.980222284945696e-05, "loss": 0.4511, "step": 1583 }, { "epoch": 0.28, "learning_rate": 4.980131591351533e-05, "loss": 0.4609, "step": 1584 }, { "epoch": 0.28, "learning_rate": 4.980040691210615e-05, "loss": 0.4542, "step": 1585 }, { "epoch": 0.28, "learning_rate": 4.979949584531361e-05, "loss": 0.4468, "step": 1586 }, { "epoch": 0.28, "learning_rate": 4.9798582713222076e-05, "loss": 0.4645, "step": 1587 }, { "epoch": 0.28, "learning_rate": 4.979766751591613e-05, "loss": 0.4474, "step": 1588 }, { "epoch": 0.28, "learning_rate": 4.9796750253480526e-05, "loss": 0.474, "step": 1589 }, { "epoch": 0.28, "learning_rate": 4.9795830926000234e-05, "loss": 0.4608, "step": 1590 }, { "epoch": 0.28, "learning_rate": 4.979490953356038e-05, "loss": 0.4571, "step": 1591 }, { "epoch": 0.28, "learning_rate": 4.97939860762463e-05, "loss": 0.4531, "step": 1592 }, { "epoch": 0.28, "learning_rate": 4.979306055414354e-05, "loss": 0.462, "step": 1593 }, { "epoch": 0.28, "learning_rate": 4.9792132967337804e-05, "loss": 0.4488, "step": 1594 }, { "epoch": 0.28, "learning_rate": 4.9791203315915005e-05, "loss": 0.4554, "step": 1595 }, { "epoch": 0.28, "learning_rate": 4.979027159996125e-05, "loss": 0.4447, "step": 1596 }, { "epoch": 0.28, "learning_rate": 4.978933781956282e-05, "loss": 0.4611, "step": 1597 }, { "epoch": 0.28, "learning_rate": 4.97884019748062e-05, "loss": 0.4605, "step": 1598 }, { "epoch": 0.28, "learning_rate": 4.978746406577807e-05, "loss": 0.4453, "step": 1599 }, { "epoch": 0.28, "learning_rate": 4.97865240925653e-05, "loss": 0.4611, "step": 1600 }, { "epoch": 0.28, "learning_rate": 4.978558205525494e-05, "loss": 0.4637, "step": 1601 }, { "epoch": 0.28, "learning_rate": 4.978463795393424e-05, "loss": 0.4342, "step": 1602 }, { "epoch": 0.28, "learning_rate": 4.978369178869063e-05, "loss": 0.432, "step": 1603 }, { "epoch": 0.28, "learning_rate": 4.978274355961174e-05, "loss": 0.4609, "step": 1604 }, { "epoch": 0.28, "learning_rate": 4.978179326678542e-05, "loss": 0.4657, "step": 1605 }, { "epoch": 0.28, "learning_rate": 4.978084091029966e-05, "loss": 0.4623, "step": 1606 }, { "epoch": 0.28, "learning_rate": 4.977988649024266e-05, "loss": 0.4655, "step": 1607 }, { "epoch": 0.28, "learning_rate": 4.977893000670282e-05, "loss": 0.4641, "step": 1608 }, { "epoch": 0.28, "learning_rate": 4.977797145976874e-05, "loss": 0.4443, "step": 1609 }, { "epoch": 0.28, "learning_rate": 4.9777010849529176e-05, "loss": 0.4334, "step": 1610 }, { "epoch": 0.28, "learning_rate": 4.97760481760731e-05, "loss": 0.4445, "step": 1611 }, { "epoch": 0.28, "learning_rate": 4.977508343948968e-05, "loss": 0.4485, "step": 1612 }, { "epoch": 0.28, "learning_rate": 4.977411663986827e-05, "loss": 0.4435, "step": 1613 }, { "epoch": 0.28, "learning_rate": 4.97731477772984e-05, "loss": 0.4592, "step": 1614 }, { "epoch": 0.28, "learning_rate": 4.977217685186981e-05, "loss": 0.4431, "step": 1615 }, { "epoch": 0.28, "learning_rate": 4.977120386367242e-05, "loss": 0.444, "step": 1616 }, { "epoch": 0.28, "learning_rate": 4.9770228812796346e-05, "loss": 0.4687, "step": 1617 }, { "epoch": 0.28, "learning_rate": 4.9769251699331886e-05, "loss": 0.4562, "step": 1618 }, { "epoch": 0.28, "learning_rate": 4.9768272523369555e-05, "loss": 0.4596, "step": 1619 }, { "epoch": 0.28, "learning_rate": 4.976729128500002e-05, "loss": 0.4464, "step": 1620 }, { "epoch": 0.28, "learning_rate": 4.976630798431417e-05, "loss": 0.4505, "step": 1621 }, { "epoch": 0.28, "learning_rate": 4.976532262140308e-05, "loss": 0.4592, "step": 1622 }, { "epoch": 0.28, "learning_rate": 4.9764335196357995e-05, "loss": 0.4404, "step": 1623 }, { "epoch": 0.28, "learning_rate": 4.9763345709270385e-05, "loss": 0.4487, "step": 1624 }, { "epoch": 0.28, "learning_rate": 4.9762354160231885e-05, "loss": 0.4361, "step": 1625 }, { "epoch": 0.28, "learning_rate": 4.9761360549334326e-05, "loss": 0.4757, "step": 1626 }, { "epoch": 0.28, "learning_rate": 4.976036487666973e-05, "loss": 0.4433, "step": 1627 }, { "epoch": 0.28, "learning_rate": 4.975936714233032e-05, "loss": 0.4466, "step": 1628 }, { "epoch": 0.28, "learning_rate": 4.97583673464085e-05, "loss": 0.4586, "step": 1629 }, { "epoch": 0.28, "learning_rate": 4.975736548899687e-05, "loss": 0.4601, "step": 1630 }, { "epoch": 0.28, "learning_rate": 4.9756361570188214e-05, "loss": 0.4479, "step": 1631 }, { "epoch": 0.28, "learning_rate": 4.9755355590075506e-05, "loss": 0.4482, "step": 1632 }, { "epoch": 0.28, "learning_rate": 4.975434754875194e-05, "loss": 0.4657, "step": 1633 }, { "epoch": 0.28, "learning_rate": 4.9753337446310847e-05, "loss": 0.4496, "step": 1634 }, { "epoch": 0.29, "learning_rate": 4.97523252828458e-05, "loss": 0.457, "step": 1635 }, { "epoch": 0.29, "learning_rate": 4.975131105845053e-05, "loss": 0.4553, "step": 1636 }, { "epoch": 0.29, "learning_rate": 4.9750294773218966e-05, "loss": 0.4587, "step": 1637 }, { "epoch": 0.29, "learning_rate": 4.9749276427245264e-05, "loss": 0.4527, "step": 1638 }, { "epoch": 0.29, "learning_rate": 4.97482560206237e-05, "loss": 0.4769, "step": 1639 }, { "epoch": 0.29, "learning_rate": 4.97472335534488e-05, "loss": 0.4582, "step": 1640 }, { "epoch": 0.29, "learning_rate": 4.9746209025815267e-05, "loss": 0.4504, "step": 1641 }, { "epoch": 0.29, "learning_rate": 4.974518243781797e-05, "loss": 0.4383, "step": 1642 }, { "epoch": 0.29, "learning_rate": 4.974415378955201e-05, "loss": 0.4481, "step": 1643 }, { "epoch": 0.29, "learning_rate": 4.974312308111264e-05, "loss": 0.4638, "step": 1644 }, { "epoch": 0.29, "learning_rate": 4.974209031259533e-05, "loss": 0.4382, "step": 1645 }, { "epoch": 0.29, "learning_rate": 4.974105548409573e-05, "loss": 0.4369, "step": 1646 }, { "epoch": 0.29, "learning_rate": 4.974001859570967e-05, "loss": 0.4688, "step": 1647 }, { "epoch": 0.29, "learning_rate": 4.9738979647533204e-05, "loss": 0.4657, "step": 1648 }, { "epoch": 0.29, "learning_rate": 4.973793863966253e-05, "loss": 0.4519, "step": 1649 }, { "epoch": 0.29, "learning_rate": 4.973689557219409e-05, "loss": 0.4613, "step": 1650 }, { "epoch": 0.29, "learning_rate": 4.973585044522447e-05, "loss": 0.4557, "step": 1651 }, { "epoch": 0.29, "learning_rate": 4.973480325885047e-05, "loss": 0.46, "step": 1652 }, { "epoch": 0.29, "learning_rate": 4.973375401316908e-05, "loss": 0.4633, "step": 1653 }, { "epoch": 0.29, "learning_rate": 4.9732702708277476e-05, "loss": 0.4342, "step": 1654 }, { "epoch": 0.29, "learning_rate": 4.973164934427302e-05, "loss": 0.4531, "step": 1655 }, { "epoch": 0.29, "learning_rate": 4.973059392125329e-05, "loss": 0.4449, "step": 1656 }, { "epoch": 0.29, "learning_rate": 4.9729536439316014e-05, "loss": 0.4297, "step": 1657 }, { "epoch": 0.29, "learning_rate": 4.972847689855913e-05, "loss": 0.4286, "step": 1658 }, { "epoch": 0.29, "learning_rate": 4.9727415299080797e-05, "loss": 0.434, "step": 1659 }, { "epoch": 0.29, "learning_rate": 4.972635164097931e-05, "loss": 0.4525, "step": 1660 }, { "epoch": 0.29, "learning_rate": 4.9725285924353186e-05, "loss": 0.4593, "step": 1661 }, { "epoch": 0.29, "learning_rate": 4.972421814930114e-05, "loss": 0.4612, "step": 1662 }, { "epoch": 0.29, "learning_rate": 4.9723148315922056e-05, "loss": 0.4453, "step": 1663 }, { "epoch": 0.29, "learning_rate": 4.9722076424315006e-05, "loss": 0.4344, "step": 1664 }, { "epoch": 0.29, "learning_rate": 4.9721002474579285e-05, "loss": 0.4684, "step": 1665 }, { "epoch": 0.29, "learning_rate": 4.9719926466814354e-05, "loss": 0.49, "step": 1666 }, { "epoch": 0.29, "learning_rate": 4.971884840111987e-05, "loss": 0.4488, "step": 1667 }, { "epoch": 0.29, "learning_rate": 4.971776827759567e-05, "loss": 0.4466, "step": 1668 }, { "epoch": 0.29, "learning_rate": 4.9716686096341805e-05, "loss": 0.4297, "step": 1669 }, { "epoch": 0.29, "learning_rate": 4.9715601857458484e-05, "loss": 0.4666, "step": 1670 }, { "epoch": 0.29, "learning_rate": 4.971451556104614e-05, "loss": 0.4457, "step": 1671 }, { "epoch": 0.29, "learning_rate": 4.971342720720537e-05, "loss": 0.453, "step": 1672 }, { "epoch": 0.29, "learning_rate": 4.9712336796037e-05, "loss": 0.4346, "step": 1673 }, { "epoch": 0.29, "learning_rate": 4.971124432764198e-05, "loss": 0.4425, "step": 1674 }, { "epoch": 0.29, "learning_rate": 4.971014980212152e-05, "loss": 0.4709, "step": 1675 }, { "epoch": 0.29, "learning_rate": 4.970905321957699e-05, "loss": 0.4414, "step": 1676 }, { "epoch": 0.29, "learning_rate": 4.9707954580109935e-05, "loss": 0.439, "step": 1677 }, { "epoch": 0.29, "learning_rate": 4.970685388382212e-05, "loss": 0.468, "step": 1678 }, { "epoch": 0.29, "learning_rate": 4.9705751130815484e-05, "loss": 0.459, "step": 1679 }, { "epoch": 0.29, "learning_rate": 4.9704646321192154e-05, "loss": 0.4512, "step": 1680 }, { "epoch": 0.29, "learning_rate": 4.970353945505446e-05, "loss": 0.4474, "step": 1681 }, { "epoch": 0.29, "learning_rate": 4.9702430532504924e-05, "loss": 0.436, "step": 1682 }, { "epoch": 0.29, "learning_rate": 4.970131955364624e-05, "loss": 0.4685, "step": 1683 }, { "epoch": 0.29, "learning_rate": 4.970020651858129e-05, "loss": 0.425, "step": 1684 }, { "epoch": 0.29, "learning_rate": 4.969909142741319e-05, "loss": 0.431, "step": 1685 }, { "epoch": 0.29, "learning_rate": 4.9697974280245194e-05, "loss": 0.4643, "step": 1686 }, { "epoch": 0.29, "learning_rate": 4.9696855077180774e-05, "loss": 0.4344, "step": 1687 }, { "epoch": 0.29, "learning_rate": 4.969573381832359e-05, "loss": 0.4329, "step": 1688 }, { "epoch": 0.29, "learning_rate": 4.969461050377748e-05, "loss": 0.4373, "step": 1689 }, { "epoch": 0.29, "learning_rate": 4.9693485133646485e-05, "loss": 0.4593, "step": 1690 }, { "epoch": 0.29, "learning_rate": 4.9692357708034835e-05, "loss": 0.4345, "step": 1691 }, { "epoch": 0.29, "learning_rate": 4.969122822704695e-05, "loss": 0.4592, "step": 1692 }, { "epoch": 0.3, "learning_rate": 4.969009669078743e-05, "loss": 0.4644, "step": 1693 }, { "epoch": 0.3, "learning_rate": 4.968896309936109e-05, "loss": 0.4454, "step": 1694 }, { "epoch": 0.3, "learning_rate": 4.96878274528729e-05, "loss": 0.4705, "step": 1695 }, { "epoch": 0.3, "learning_rate": 4.9686689751428043e-05, "loss": 0.4613, "step": 1696 }, { "epoch": 0.3, "learning_rate": 4.9685549995131906e-05, "loss": 0.4334, "step": 1697 }, { "epoch": 0.3, "learning_rate": 4.968440818409004e-05, "loss": 0.4575, "step": 1698 }, { "epoch": 0.3, "learning_rate": 4.9683264318408176e-05, "loss": 0.462, "step": 1699 }, { "epoch": 0.3, "learning_rate": 4.968211839819229e-05, "loss": 0.4853, "step": 1700 }, { "epoch": 0.3, "learning_rate": 4.9680970423548483e-05, "loss": 0.4488, "step": 1701 }, { "epoch": 0.3, "learning_rate": 4.967982039458309e-05, "loss": 0.4288, "step": 1702 }, { "epoch": 0.3, "learning_rate": 4.967866831140262e-05, "loss": 0.4448, "step": 1703 }, { "epoch": 0.3, "learning_rate": 4.9677514174113773e-05, "loss": 0.4631, "step": 1704 }, { "epoch": 0.3, "learning_rate": 4.967635798282344e-05, "loss": 0.4551, "step": 1705 }, { "epoch": 0.3, "learning_rate": 4.967519973763871e-05, "loss": 0.462, "step": 1706 }, { "epoch": 0.3, "learning_rate": 4.9674039438666856e-05, "loss": 0.4467, "step": 1707 }, { "epoch": 0.3, "learning_rate": 4.9672877086015336e-05, "loss": 0.4698, "step": 1708 }, { "epoch": 0.3, "learning_rate": 4.96717126797918e-05, "loss": 0.4479, "step": 1709 }, { "epoch": 0.3, "learning_rate": 4.967054622010409e-05, "loss": 0.4467, "step": 1710 }, { "epoch": 0.3, "learning_rate": 4.966937770706025e-05, "loss": 0.4421, "step": 1711 }, { "epoch": 0.3, "learning_rate": 4.966820714076849e-05, "loss": 0.4514, "step": 1712 }, { "epoch": 0.3, "learning_rate": 4.9667034521337236e-05, "loss": 0.4476, "step": 1713 }, { "epoch": 0.3, "learning_rate": 4.966585984887509e-05, "loss": 0.4606, "step": 1714 }, { "epoch": 0.3, "learning_rate": 4.9664683123490843e-05, "loss": 0.4495, "step": 1715 }, { "epoch": 0.3, "learning_rate": 4.966350434529347e-05, "loss": 0.4364, "step": 1716 }, { "epoch": 0.3, "learning_rate": 4.966232351439216e-05, "loss": 0.448, "step": 1717 }, { "epoch": 0.3, "learning_rate": 4.9661140630896276e-05, "loss": 0.4607, "step": 1718 }, { "epoch": 0.3, "learning_rate": 4.9659955694915356e-05, "loss": 0.474, "step": 1719 }, { "epoch": 0.3, "learning_rate": 4.9658768706559166e-05, "loss": 0.4341, "step": 1720 }, { "epoch": 0.3, "learning_rate": 4.965757966593763e-05, "loss": 0.4833, "step": 1721 }, { "epoch": 0.3, "learning_rate": 4.9656388573160865e-05, "loss": 0.4605, "step": 1722 }, { "epoch": 0.3, "learning_rate": 4.96551954283392e-05, "loss": 0.4455, "step": 1723 }, { "epoch": 0.3, "learning_rate": 4.965400023158313e-05, "loss": 0.4451, "step": 1724 }, { "epoch": 0.3, "learning_rate": 4.965280298300336e-05, "loss": 0.4379, "step": 1725 }, { "epoch": 0.3, "learning_rate": 4.965160368271076e-05, "loss": 0.45, "step": 1726 }, { "epoch": 0.3, "learning_rate": 4.965040233081642e-05, "loss": 0.4442, "step": 1727 }, { "epoch": 0.3, "learning_rate": 4.9649198927431594e-05, "loss": 0.4545, "step": 1728 }, { "epoch": 0.3, "learning_rate": 4.964799347266774e-05, "loss": 0.4539, "step": 1729 }, { "epoch": 0.3, "learning_rate": 4.9646785966636506e-05, "loss": 0.4403, "step": 1730 }, { "epoch": 0.3, "learning_rate": 4.964557640944972e-05, "loss": 0.4575, "step": 1731 }, { "epoch": 0.3, "learning_rate": 4.9644364801219426e-05, "loss": 0.4368, "step": 1732 }, { "epoch": 0.3, "learning_rate": 4.964315114205781e-05, "loss": 0.4507, "step": 1733 }, { "epoch": 0.3, "learning_rate": 4.96419354320773e-05, "loss": 0.4442, "step": 1734 }, { "epoch": 0.3, "learning_rate": 4.964071767139048e-05, "loss": 0.4559, "step": 1735 }, { "epoch": 0.3, "learning_rate": 4.9639497860110136e-05, "loss": 0.4373, "step": 1736 }, { "epoch": 0.3, "learning_rate": 4.9638275998349244e-05, "loss": 0.4433, "step": 1737 }, { "epoch": 0.3, "learning_rate": 4.963705208622096e-05, "loss": 0.4493, "step": 1738 }, { "epoch": 0.3, "learning_rate": 4.963582612383866e-05, "loss": 0.4463, "step": 1739 }, { "epoch": 0.3, "learning_rate": 4.963459811131585e-05, "loss": 0.4363, "step": 1740 }, { "epoch": 0.3, "learning_rate": 4.963336804876631e-05, "loss": 0.4469, "step": 1741 }, { "epoch": 0.3, "learning_rate": 4.963213593630394e-05, "loss": 0.4385, "step": 1742 }, { "epoch": 0.3, "learning_rate": 4.963090177404285e-05, "loss": 0.4557, "step": 1743 }, { "epoch": 0.3, "learning_rate": 4.962966556209735e-05, "loss": 0.454, "step": 1744 }, { "epoch": 0.3, "learning_rate": 4.9628427300581935e-05, "loss": 0.4568, "step": 1745 }, { "epoch": 0.3, "learning_rate": 4.9627186989611286e-05, "loss": 0.4525, "step": 1746 }, { "epoch": 0.3, "learning_rate": 4.962594462930028e-05, "loss": 0.4441, "step": 1747 }, { "epoch": 0.3, "learning_rate": 4.962470021976397e-05, "loss": 0.4443, "step": 1748 }, { "epoch": 0.3, "learning_rate": 4.962345376111762e-05, "loss": 0.4515, "step": 1749 }, { "epoch": 0.31, "learning_rate": 4.9622205253476664e-05, "loss": 0.4527, "step": 1750 }, { "epoch": 0.31, "learning_rate": 4.962095469695673e-05, "loss": 0.4669, "step": 1751 }, { "epoch": 0.31, "learning_rate": 4.9619702091673665e-05, "loss": 0.4416, "step": 1752 }, { "epoch": 0.31, "learning_rate": 4.961844743774346e-05, "loss": 0.4559, "step": 1753 }, { "epoch": 0.31, "learning_rate": 4.961719073528231e-05, "loss": 0.4399, "step": 1754 }, { "epoch": 0.31, "learning_rate": 4.9615931984406626e-05, "loss": 0.4475, "step": 1755 }, { "epoch": 0.31, "learning_rate": 4.961467118523297e-05, "loss": 0.4592, "step": 1756 }, { "epoch": 0.31, "learning_rate": 4.961340833787814e-05, "loss": 0.4383, "step": 1757 }, { "epoch": 0.31, "learning_rate": 4.961214344245906e-05, "loss": 0.461, "step": 1758 }, { "epoch": 0.31, "learning_rate": 4.96108764990929e-05, "loss": 0.4514, "step": 1759 }, { "epoch": 0.31, "learning_rate": 4.960960750789701e-05, "loss": 0.4534, "step": 1760 }, { "epoch": 0.31, "learning_rate": 4.9608336468988905e-05, "loss": 0.455, "step": 1761 }, { "epoch": 0.31, "learning_rate": 4.96070633824863e-05, "loss": 0.4312, "step": 1762 }, { "epoch": 0.31, "learning_rate": 4.960578824850711e-05, "loss": 0.4353, "step": 1763 }, { "epoch": 0.31, "learning_rate": 4.960451106716944e-05, "loss": 0.4416, "step": 1764 }, { "epoch": 0.31, "learning_rate": 4.960323183859158e-05, "loss": 0.4406, "step": 1765 }, { "epoch": 0.31, "learning_rate": 4.960195056289198e-05, "loss": 0.465, "step": 1766 }, { "epoch": 0.31, "learning_rate": 4.9600667240189335e-05, "loss": 0.429, "step": 1767 }, { "epoch": 0.31, "learning_rate": 4.959938187060249e-05, "loss": 0.4593, "step": 1768 }, { "epoch": 0.31, "learning_rate": 4.95980944542505e-05, "loss": 0.4385, "step": 1769 }, { "epoch": 0.31, "learning_rate": 4.9596804991252595e-05, "loss": 0.4547, "step": 1770 }, { "epoch": 0.31, "learning_rate": 4.95955134817282e-05, "loss": 0.443, "step": 1771 }, { "epoch": 0.31, "learning_rate": 4.9594219925796934e-05, "loss": 0.4598, "step": 1772 }, { "epoch": 0.31, "learning_rate": 4.959292432357858e-05, "loss": 0.4408, "step": 1773 }, { "epoch": 0.31, "learning_rate": 4.959162667519317e-05, "loss": 0.4569, "step": 1774 }, { "epoch": 0.31, "learning_rate": 4.959032698076086e-05, "loss": 0.4481, "step": 1775 }, { "epoch": 0.31, "learning_rate": 4.9589025240402036e-05, "loss": 0.4464, "step": 1776 }, { "epoch": 0.31, "learning_rate": 4.958772145423726e-05, "loss": 0.4185, "step": 1777 }, { "epoch": 0.31, "learning_rate": 4.9586415622387265e-05, "loss": 0.4448, "step": 1778 }, { "epoch": 0.31, "learning_rate": 4.9585107744973005e-05, "loss": 0.4455, "step": 1779 }, { "epoch": 0.31, "learning_rate": 4.9583797822115626e-05, "loss": 0.4317, "step": 1780 }, { "epoch": 0.31, "learning_rate": 4.958248585393643e-05, "loss": 0.4567, "step": 1781 }, { "epoch": 0.31, "learning_rate": 4.9581171840556926e-05, "loss": 0.4665, "step": 1782 }, { "epoch": 0.31, "learning_rate": 4.9579855782098825e-05, "loss": 0.4342, "step": 1783 }, { "epoch": 0.31, "learning_rate": 4.957853767868401e-05, "loss": 0.4392, "step": 1784 }, { "epoch": 0.31, "learning_rate": 4.957721753043455e-05, "loss": 0.4598, "step": 1785 }, { "epoch": 0.31, "learning_rate": 4.957589533747274e-05, "loss": 0.4589, "step": 1786 }, { "epoch": 0.31, "learning_rate": 4.9574571099921e-05, "loss": 0.4597, "step": 1787 }, { "epoch": 0.31, "learning_rate": 4.9573244817901995e-05, "loss": 0.4419, "step": 1788 }, { "epoch": 0.31, "learning_rate": 4.9571916491538575e-05, "loss": 0.4517, "step": 1789 }, { "epoch": 0.31, "learning_rate": 4.957058612095373e-05, "loss": 0.4641, "step": 1790 }, { "epoch": 0.31, "learning_rate": 4.95692537062707e-05, "loss": 0.4676, "step": 1791 }, { "epoch": 0.31, "learning_rate": 4.956791924761288e-05, "loss": 0.4646, "step": 1792 }, { "epoch": 0.31, "learning_rate": 4.956658274510386e-05, "loss": 0.4471, "step": 1793 }, { "epoch": 0.31, "learning_rate": 4.956524419886744e-05, "loss": 0.4394, "step": 1794 }, { "epoch": 0.31, "learning_rate": 4.956390360902757e-05, "loss": 0.4525, "step": 1795 }, { "epoch": 0.31, "learning_rate": 4.9562560975708415e-05, "loss": 0.4644, "step": 1796 }, { "epoch": 0.31, "learning_rate": 4.956121629903433e-05, "loss": 0.4641, "step": 1797 }, { "epoch": 0.31, "learning_rate": 4.955986957912985e-05, "loss": 0.4463, "step": 1798 }, { "epoch": 0.31, "learning_rate": 4.955852081611971e-05, "loss": 0.4409, "step": 1799 }, { "epoch": 0.31, "learning_rate": 4.9557170010128825e-05, "loss": 0.4439, "step": 1800 }, { "epoch": 0.31, "learning_rate": 4.95558171612823e-05, "loss": 0.4438, "step": 1801 }, { "epoch": 0.31, "learning_rate": 4.955446226970542e-05, "loss": 0.4499, "step": 1802 }, { "epoch": 0.31, "learning_rate": 4.9553105335523695e-05, "loss": 0.4264, "step": 1803 }, { "epoch": 0.31, "learning_rate": 4.9551746358862775e-05, "loss": 0.4544, "step": 1804 }, { "epoch": 0.31, "learning_rate": 4.955038533984854e-05, "loss": 0.4316, "step": 1805 }, { "epoch": 0.31, "learning_rate": 4.9549022278607035e-05, "loss": 0.453, "step": 1806 }, { "epoch": 0.32, "learning_rate": 4.954765717526451e-05, "loss": 0.4537, "step": 1807 }, { "epoch": 0.32, "learning_rate": 4.9546290029947375e-05, "loss": 0.4582, "step": 1808 }, { "epoch": 0.32, "learning_rate": 4.9544920842782274e-05, "loss": 0.4133, "step": 1809 }, { "epoch": 0.32, "learning_rate": 4.9543549613896004e-05, "loss": 0.4344, "step": 1810 }, { "epoch": 0.32, "learning_rate": 4.954217634341557e-05, "loss": 0.4472, "step": 1811 }, { "epoch": 0.32, "learning_rate": 4.9540801031468156e-05, "loss": 0.4586, "step": 1812 }, { "epoch": 0.32, "learning_rate": 4.953942367818113e-05, "loss": 0.4555, "step": 1813 }, { "epoch": 0.32, "learning_rate": 4.953804428368207e-05, "loss": 0.4619, "step": 1814 }, { "epoch": 0.32, "learning_rate": 4.9536662848098724e-05, "loss": 0.4267, "step": 1815 }, { "epoch": 0.32, "learning_rate": 4.953527937155904e-05, "loss": 0.4395, "step": 1816 }, { "epoch": 0.32, "learning_rate": 4.9533893854191146e-05, "loss": 0.4547, "step": 1817 }, { "epoch": 0.32, "learning_rate": 4.9532506296123366e-05, "loss": 0.4469, "step": 1818 }, { "epoch": 0.32, "learning_rate": 4.953111669748421e-05, "loss": 0.4463, "step": 1819 }, { "epoch": 0.32, "learning_rate": 4.952972505840238e-05, "loss": 0.4532, "step": 1820 }, { "epoch": 0.32, "learning_rate": 4.9528331379006765e-05, "loss": 0.46, "step": 1821 }, { "epoch": 0.32, "learning_rate": 4.952693565942643e-05, "loss": 0.4474, "step": 1822 }, { "epoch": 0.32, "learning_rate": 4.952553789979066e-05, "loss": 0.4245, "step": 1823 }, { "epoch": 0.32, "learning_rate": 4.95241381002289e-05, "loss": 0.4522, "step": 1824 }, { "epoch": 0.32, "learning_rate": 4.95227362608708e-05, "loss": 0.4241, "step": 1825 }, { "epoch": 0.32, "learning_rate": 4.9521332381846184e-05, "loss": 0.4137, "step": 1826 }, { "epoch": 0.32, "learning_rate": 4.9519926463285075e-05, "loss": 0.4438, "step": 1827 }, { "epoch": 0.32, "learning_rate": 4.9518518505317704e-05, "loss": 0.4344, "step": 1828 }, { "epoch": 0.32, "learning_rate": 4.951710850807444e-05, "loss": 0.4725, "step": 1829 }, { "epoch": 0.32, "learning_rate": 4.95156964716859e-05, "loss": 0.444, "step": 1830 }, { "epoch": 0.32, "learning_rate": 4.9514282396282845e-05, "loss": 0.4284, "step": 1831 }, { "epoch": 0.32, "learning_rate": 4.9512866281996254e-05, "loss": 0.4441, "step": 1832 }, { "epoch": 0.32, "learning_rate": 4.951144812895726e-05, "loss": 0.445, "step": 1833 }, { "epoch": 0.32, "learning_rate": 4.9510027937297236e-05, "loss": 0.4342, "step": 1834 }, { "epoch": 0.32, "learning_rate": 4.950860570714769e-05, "loss": 0.4312, "step": 1835 }, { "epoch": 0.32, "learning_rate": 4.950718143864036e-05, "loss": 0.4412, "step": 1836 }, { "epoch": 0.32, "learning_rate": 4.950575513190715e-05, "loss": 0.4452, "step": 1837 }, { "epoch": 0.32, "learning_rate": 4.950432678708016e-05, "loss": 0.4544, "step": 1838 }, { "epoch": 0.32, "learning_rate": 4.9502896404291684e-05, "loss": 0.4465, "step": 1839 }, { "epoch": 0.32, "learning_rate": 4.950146398367419e-05, "loss": 0.4438, "step": 1840 }, { "epoch": 0.32, "learning_rate": 4.950002952536035e-05, "loss": 0.4391, "step": 1841 }, { "epoch": 0.32, "learning_rate": 4.949859302948301e-05, "loss": 0.441, "step": 1842 }, { "epoch": 0.32, "learning_rate": 4.949715449617522e-05, "loss": 0.4586, "step": 1843 }, { "epoch": 0.32, "learning_rate": 4.949571392557021e-05, "loss": 0.4628, "step": 1844 }, { "epoch": 0.32, "learning_rate": 4.949427131780141e-05, "loss": 0.4404, "step": 1845 }, { "epoch": 0.32, "learning_rate": 4.949282667300241e-05, "loss": 0.4498, "step": 1846 }, { "epoch": 0.32, "learning_rate": 4.949137999130703e-05, "loss": 0.4266, "step": 1847 }, { "epoch": 0.32, "learning_rate": 4.948993127284924e-05, "loss": 0.4528, "step": 1848 }, { "epoch": 0.32, "learning_rate": 4.948848051776321e-05, "loss": 0.4498, "step": 1849 }, { "epoch": 0.32, "learning_rate": 4.948702772618332e-05, "loss": 0.4297, "step": 1850 }, { "epoch": 0.32, "learning_rate": 4.948557289824412e-05, "loss": 0.4449, "step": 1851 }, { "epoch": 0.32, "learning_rate": 4.948411603408034e-05, "loss": 0.4387, "step": 1852 }, { "epoch": 0.32, "learning_rate": 4.948265713382692e-05, "loss": 0.4462, "step": 1853 }, { "epoch": 0.32, "learning_rate": 4.948119619761897e-05, "loss": 0.4629, "step": 1854 }, { "epoch": 0.32, "learning_rate": 4.94797332255918e-05, "loss": 0.4376, "step": 1855 }, { "epoch": 0.32, "learning_rate": 4.947826821788091e-05, "loss": 0.4315, "step": 1856 }, { "epoch": 0.32, "learning_rate": 4.9476801174621985e-05, "loss": 0.4522, "step": 1857 }, { "epoch": 0.32, "learning_rate": 4.9475332095950874e-05, "loss": 0.447, "step": 1858 }, { "epoch": 0.32, "learning_rate": 4.947386098200368e-05, "loss": 0.4333, "step": 1859 }, { "epoch": 0.32, "learning_rate": 4.947238783291661e-05, "loss": 0.4587, "step": 1860 }, { "epoch": 0.32, "learning_rate": 4.947091264882613e-05, "loss": 0.4508, "step": 1861 }, { "epoch": 0.32, "learning_rate": 4.9469435429868855e-05, "loss": 0.4325, "step": 1862 }, { "epoch": 0.32, "learning_rate": 4.946795617618159e-05, "loss": 0.4441, "step": 1863 }, { "epoch": 0.32, "learning_rate": 4.946647488790136e-05, "loss": 0.4394, "step": 1864 }, { "epoch": 0.33, "learning_rate": 4.9464991565165336e-05, "loss": 0.4449, "step": 1865 }, { "epoch": 0.33, "learning_rate": 4.946350620811092e-05, "loss": 0.462, "step": 1866 }, { "epoch": 0.33, "learning_rate": 4.946201881687566e-05, "loss": 0.4592, "step": 1867 }, { "epoch": 0.33, "learning_rate": 4.946052939159732e-05, "loss": 0.4483, "step": 1868 }, { "epoch": 0.33, "learning_rate": 4.9459037932413844e-05, "loss": 0.4481, "step": 1869 }, { "epoch": 0.33, "learning_rate": 4.945754443946338e-05, "loss": 0.4668, "step": 1870 }, { "epoch": 0.33, "learning_rate": 4.945604891288422e-05, "loss": 0.4508, "step": 1871 }, { "epoch": 0.33, "learning_rate": 4.94545513528149e-05, "loss": 0.4729, "step": 1872 }, { "epoch": 0.33, "learning_rate": 4.9453051759394116e-05, "loss": 0.4338, "step": 1873 }, { "epoch": 0.33, "learning_rate": 4.945155013276074e-05, "loss": 0.4319, "step": 1874 }, { "epoch": 0.33, "learning_rate": 4.945004647305387e-05, "loss": 0.4451, "step": 1875 }, { "epoch": 0.33, "learning_rate": 4.9448540780412744e-05, "loss": 0.4438, "step": 1876 }, { "epoch": 0.33, "learning_rate": 4.944703305497683e-05, "loss": 0.4456, "step": 1877 }, { "epoch": 0.33, "learning_rate": 4.944552329688576e-05, "loss": 0.4296, "step": 1878 }, { "epoch": 0.33, "learning_rate": 4.9444011506279365e-05, "loss": 0.458, "step": 1879 }, { "epoch": 0.33, "learning_rate": 4.9442497683297675e-05, "loss": 0.4594, "step": 1880 }, { "epoch": 0.33, "learning_rate": 4.9440981828080874e-05, "loss": 0.4395, "step": 1881 }, { "epoch": 0.33, "learning_rate": 4.9439463940769366e-05, "loss": 0.4238, "step": 1882 }, { "epoch": 0.33, "learning_rate": 4.943794402150373e-05, "loss": 0.4503, "step": 1883 }, { "epoch": 0.33, "learning_rate": 4.943642207042473e-05, "loss": 0.4353, "step": 1884 }, { "epoch": 0.33, "learning_rate": 4.9434898087673336e-05, "loss": 0.43, "step": 1885 }, { "epoch": 0.33, "learning_rate": 4.9433372073390685e-05, "loss": 0.4319, "step": 1886 }, { "epoch": 0.33, "learning_rate": 4.943184402771811e-05, "loss": 0.4305, "step": 1887 }, { "epoch": 0.33, "learning_rate": 4.943031395079714e-05, "loss": 0.4594, "step": 1888 }, { "epoch": 0.33, "learning_rate": 4.9428781842769476e-05, "loss": 0.4448, "step": 1889 }, { "epoch": 0.33, "learning_rate": 4.942724770377703e-05, "loss": 0.4361, "step": 1890 }, { "epoch": 0.33, "learning_rate": 4.942571153396187e-05, "loss": 0.4665, "step": 1891 }, { "epoch": 0.33, "learning_rate": 4.9424173333466286e-05, "loss": 0.4513, "step": 1892 }, { "epoch": 0.33, "learning_rate": 4.942263310243274e-05, "loss": 0.4497, "step": 1893 }, { "epoch": 0.33, "learning_rate": 4.9421090841003875e-05, "loss": 0.4311, "step": 1894 }, { "epoch": 0.33, "learning_rate": 4.9419546549322526e-05, "loss": 0.4485, "step": 1895 }, { "epoch": 0.33, "learning_rate": 4.941800022753173e-05, "loss": 0.4632, "step": 1896 }, { "epoch": 0.33, "learning_rate": 4.94164518757747e-05, "loss": 0.4203, "step": 1897 }, { "epoch": 0.33, "learning_rate": 4.941490149419483e-05, "loss": 0.449, "step": 1898 }, { "epoch": 0.33, "learning_rate": 4.941334908293573e-05, "loss": 0.4428, "step": 1899 }, { "epoch": 0.33, "learning_rate": 4.9411794642141156e-05, "loss": 0.4328, "step": 1900 }, { "epoch": 0.33, "learning_rate": 4.9410238171955095e-05, "loss": 0.4491, "step": 1901 }, { "epoch": 0.33, "learning_rate": 4.940867967252168e-05, "loss": 0.4551, "step": 1902 }, { "epoch": 0.33, "learning_rate": 4.940711914398527e-05, "loss": 0.4317, "step": 1903 }, { "epoch": 0.33, "learning_rate": 4.940555658649039e-05, "loss": 0.4315, "step": 1904 }, { "epoch": 0.33, "learning_rate": 4.940399200018176e-05, "loss": 0.4506, "step": 1905 }, { "epoch": 0.33, "learning_rate": 4.940242538520428e-05, "loss": 0.4237, "step": 1906 }, { "epoch": 0.33, "learning_rate": 4.940085674170306e-05, "loss": 0.4501, "step": 1907 }, { "epoch": 0.33, "learning_rate": 4.939928606982337e-05, "loss": 0.4662, "step": 1908 }, { "epoch": 0.33, "learning_rate": 4.9397713369710676e-05, "loss": 0.4382, "step": 1909 }, { "epoch": 0.33, "learning_rate": 4.939613864151064e-05, "loss": 0.4287, "step": 1910 }, { "epoch": 0.33, "learning_rate": 4.939456188536911e-05, "loss": 0.4304, "step": 1911 }, { "epoch": 0.33, "learning_rate": 4.939298310143213e-05, "loss": 0.4496, "step": 1912 }, { "epoch": 0.33, "learning_rate": 4.9391402289845894e-05, "loss": 0.4393, "step": 1913 }, { "epoch": 0.33, "learning_rate": 4.9389819450756834e-05, "loss": 0.4218, "step": 1914 }, { "epoch": 0.33, "learning_rate": 4.9388234584311546e-05, "loss": 0.4388, "step": 1915 }, { "epoch": 0.33, "learning_rate": 4.938664769065681e-05, "loss": 0.4509, "step": 1916 }, { "epoch": 0.33, "learning_rate": 4.938505876993959e-05, "loss": 0.4377, "step": 1917 }, { "epoch": 0.33, "learning_rate": 4.938346782230705e-05, "loss": 0.4325, "step": 1918 }, { "epoch": 0.33, "learning_rate": 4.938187484790654e-05, "loss": 0.4553, "step": 1919 }, { "epoch": 0.33, "learning_rate": 4.93802798468856e-05, "loss": 0.4486, "step": 1920 }, { "epoch": 0.33, "learning_rate": 4.937868281939195e-05, "loss": 0.4498, "step": 1921 }, { "epoch": 0.34, "learning_rate": 4.937708376557351e-05, "loss": 0.4426, "step": 1922 }, { "epoch": 0.34, "learning_rate": 4.9375482685578357e-05, "loss": 0.4491, "step": 1923 }, { "epoch": 0.34, "learning_rate": 4.93738795795548e-05, "loss": 0.4429, "step": 1924 }, { "epoch": 0.34, "learning_rate": 4.937227444765129e-05, "loss": 0.4405, "step": 1925 }, { "epoch": 0.34, "learning_rate": 4.9370667290016506e-05, "loss": 0.4534, "step": 1926 }, { "epoch": 0.34, "learning_rate": 4.936905810679929e-05, "loss": 0.4505, "step": 1927 }, { "epoch": 0.34, "learning_rate": 4.9367446898148685e-05, "loss": 0.4513, "step": 1928 }, { "epoch": 0.34, "learning_rate": 4.9365833664213914e-05, "loss": 0.434, "step": 1929 }, { "epoch": 0.34, "learning_rate": 4.9364218405144376e-05, "loss": 0.4435, "step": 1930 }, { "epoch": 0.34, "learning_rate": 4.9362601121089686e-05, "loss": 0.4208, "step": 1931 }, { "epoch": 0.34, "learning_rate": 4.9360981812199624e-05, "loss": 0.433, "step": 1932 }, { "epoch": 0.34, "learning_rate": 4.935936047862417e-05, "loss": 0.4611, "step": 1933 }, { "epoch": 0.34, "learning_rate": 4.935773712051349e-05, "loss": 0.462, "step": 1934 }, { "epoch": 0.34, "learning_rate": 4.9356111738017906e-05, "loss": 0.4606, "step": 1935 }, { "epoch": 0.34, "learning_rate": 4.9354484331287995e-05, "loss": 0.4452, "step": 1936 }, { "epoch": 0.34, "learning_rate": 4.9352854900474445e-05, "loss": 0.4396, "step": 1937 }, { "epoch": 0.34, "learning_rate": 4.935122344572819e-05, "loss": 0.4474, "step": 1938 }, { "epoch": 0.34, "learning_rate": 4.934958996720033e-05, "loss": 0.4255, "step": 1939 }, { "epoch": 0.34, "learning_rate": 4.934795446504214e-05, "loss": 0.4651, "step": 1940 }, { "epoch": 0.34, "learning_rate": 4.9346316939405105e-05, "loss": 0.469, "step": 1941 }, { "epoch": 0.34, "learning_rate": 4.9344677390440866e-05, "loss": 0.4606, "step": 1942 }, { "epoch": 0.34, "learning_rate": 4.9343035818301306e-05, "loss": 0.4435, "step": 1943 }, { "epoch": 0.34, "learning_rate": 4.9341392223138436e-05, "loss": 0.4369, "step": 1944 }, { "epoch": 0.34, "learning_rate": 4.933974660510449e-05, "loss": 0.4324, "step": 1945 }, { "epoch": 0.34, "learning_rate": 4.933809896435188e-05, "loss": 0.4305, "step": 1946 }, { "epoch": 0.34, "learning_rate": 4.93364493010332e-05, "loss": 0.4472, "step": 1947 }, { "epoch": 0.34, "learning_rate": 4.933479761530123e-05, "loss": 0.4333, "step": 1948 }, { "epoch": 0.34, "learning_rate": 4.933314390730895e-05, "loss": 0.4403, "step": 1949 }, { "epoch": 0.34, "learning_rate": 4.9331488177209525e-05, "loss": 0.4491, "step": 1950 }, { "epoch": 0.34, "learning_rate": 4.93298304251563e-05, "loss": 0.4295, "step": 1951 }, { "epoch": 0.34, "learning_rate": 4.932817065130281e-05, "loss": 0.4289, "step": 1952 }, { "epoch": 0.34, "learning_rate": 4.9326508855802765e-05, "loss": 0.4395, "step": 1953 }, { "epoch": 0.34, "learning_rate": 4.9324845038810096e-05, "loss": 0.4417, "step": 1954 }, { "epoch": 0.34, "learning_rate": 4.9323179200478894e-05, "loss": 0.4633, "step": 1955 }, { "epoch": 0.34, "learning_rate": 4.9321511340963425e-05, "loss": 0.4508, "step": 1956 }, { "epoch": 0.34, "learning_rate": 4.931984146041817e-05, "loss": 0.4326, "step": 1957 }, { "epoch": 0.34, "learning_rate": 4.93181695589978e-05, "loss": 0.4549, "step": 1958 }, { "epoch": 0.34, "learning_rate": 4.931649563685715e-05, "loss": 0.4501, "step": 1959 }, { "epoch": 0.34, "learning_rate": 4.9314819694151255e-05, "loss": 0.4481, "step": 1960 }, { "epoch": 0.34, "learning_rate": 4.931314173103533e-05, "loss": 0.4327, "step": 1961 }, { "epoch": 0.34, "learning_rate": 4.931146174766479e-05, "loss": 0.4674, "step": 1962 }, { "epoch": 0.34, "learning_rate": 4.930977974419521e-05, "loss": 0.4326, "step": 1963 }, { "epoch": 0.34, "learning_rate": 4.93080957207824e-05, "loss": 0.4637, "step": 1964 }, { "epoch": 0.34, "learning_rate": 4.930640967758231e-05, "loss": 0.4521, "step": 1965 }, { "epoch": 0.34, "learning_rate": 4.9304721614751095e-05, "loss": 0.4361, "step": 1966 }, { "epoch": 0.34, "learning_rate": 4.930303153244511e-05, "loss": 0.4369, "step": 1967 }, { "epoch": 0.34, "learning_rate": 4.930133943082087e-05, "loss": 0.4304, "step": 1968 }, { "epoch": 0.34, "learning_rate": 4.9299645310035093e-05, "loss": 0.4451, "step": 1969 }, { "epoch": 0.34, "learning_rate": 4.9297949170244686e-05, "loss": 0.4515, "step": 1970 }, { "epoch": 0.34, "learning_rate": 4.929625101160675e-05, "loss": 0.4582, "step": 1971 }, { "epoch": 0.34, "learning_rate": 4.929455083427853e-05, "loss": 0.4325, "step": 1972 }, { "epoch": 0.34, "learning_rate": 4.929284863841753e-05, "loss": 0.4324, "step": 1973 }, { "epoch": 0.34, "learning_rate": 4.929114442418139e-05, "loss": 0.4536, "step": 1974 }, { "epoch": 0.34, "learning_rate": 4.928943819172792e-05, "loss": 0.4483, "step": 1975 }, { "epoch": 0.34, "learning_rate": 4.9287729941215176e-05, "loss": 0.4535, "step": 1976 }, { "epoch": 0.34, "learning_rate": 4.928601967280137e-05, "loss": 0.4341, "step": 1977 }, { "epoch": 0.34, "learning_rate": 4.9284307386644876e-05, "loss": 0.4235, "step": 1978 }, { "epoch": 0.35, "learning_rate": 4.928259308290431e-05, "loss": 0.4166, "step": 1979 }, { "epoch": 0.35, "learning_rate": 4.928087676173842e-05, "loss": 0.4364, "step": 1980 }, { "epoch": 0.35, "learning_rate": 4.927915842330618e-05, "loss": 0.4285, "step": 1981 }, { "epoch": 0.35, "learning_rate": 4.9277438067766736e-05, "loss": 0.4576, "step": 1982 }, { "epoch": 0.35, "learning_rate": 4.927571569527941e-05, "loss": 0.447, "step": 1983 }, { "epoch": 0.35, "learning_rate": 4.927399130600373e-05, "loss": 0.4487, "step": 1984 }, { "epoch": 0.35, "learning_rate": 4.92722649000994e-05, "loss": 0.4442, "step": 1985 }, { "epoch": 0.35, "learning_rate": 4.927053647772631e-05, "loss": 0.4339, "step": 1986 }, { "epoch": 0.35, "learning_rate": 4.9268806039044554e-05, "loss": 0.4345, "step": 1987 }, { "epoch": 0.35, "learning_rate": 4.926707358421439e-05, "loss": 0.4624, "step": 1988 }, { "epoch": 0.35, "learning_rate": 4.926533911339627e-05, "loss": 0.4294, "step": 1989 }, { "epoch": 0.35, "learning_rate": 4.926360262675084e-05, "loss": 0.4128, "step": 1990 }, { "epoch": 0.35, "learning_rate": 4.926186412443892e-05, "loss": 0.4288, "step": 1991 }, { "epoch": 0.35, "learning_rate": 4.9260123606621526e-05, "loss": 0.45, "step": 1992 }, { "epoch": 0.35, "learning_rate": 4.925838107345986e-05, "loss": 0.4555, "step": 1993 }, { "epoch": 0.35, "learning_rate": 4.9256636525115315e-05, "loss": 0.4429, "step": 1994 }, { "epoch": 0.35, "learning_rate": 4.925488996174945e-05, "loss": 0.4477, "step": 1995 }, { "epoch": 0.35, "learning_rate": 4.925314138352404e-05, "loss": 0.428, "step": 1996 }, { "epoch": 0.35, "learning_rate": 4.9251390790601024e-05, "loss": 0.4261, "step": 1997 }, { "epoch": 0.35, "learning_rate": 4.924963818314254e-05, "loss": 0.4503, "step": 1998 }, { "epoch": 0.35, "learning_rate": 4.9247883561310906e-05, "loss": 0.4574, "step": 1999 }, { "epoch": 0.35, "learning_rate": 4.9246126925268624e-05, "loss": 0.4513, "step": 2000 }, { "epoch": 0.35, "learning_rate": 4.92443682751784e-05, "loss": 0.4557, "step": 2001 }, { "epoch": 0.35, "learning_rate": 4.9242607611203106e-05, "loss": 0.4401, "step": 2002 }, { "epoch": 0.35, "learning_rate": 4.92408449335058e-05, "loss": 0.4457, "step": 2003 }, { "epoch": 0.35, "learning_rate": 4.923908024224975e-05, "loss": 0.4474, "step": 2004 }, { "epoch": 0.35, "learning_rate": 4.923731353759839e-05, "loss": 0.4593, "step": 2005 }, { "epoch": 0.35, "learning_rate": 4.9235544819715335e-05, "loss": 0.4443, "step": 2006 }, { "epoch": 0.35, "learning_rate": 4.923377408876441e-05, "loss": 0.4532, "step": 2007 }, { "epoch": 0.35, "learning_rate": 4.923200134490961e-05, "loss": 0.4392, "step": 2008 }, { "epoch": 0.35, "learning_rate": 4.923022658831513e-05, "loss": 0.4324, "step": 2009 }, { "epoch": 0.35, "learning_rate": 4.922844981914532e-05, "loss": 0.4442, "step": 2010 }, { "epoch": 0.35, "learning_rate": 4.9226671037564754e-05, "loss": 0.4421, "step": 2011 }, { "epoch": 0.35, "learning_rate": 4.922489024373817e-05, "loss": 0.4357, "step": 2012 }, { "epoch": 0.35, "learning_rate": 4.922310743783051e-05, "loss": 0.4425, "step": 2013 }, { "epoch": 0.35, "learning_rate": 4.9221322620006877e-05, "loss": 0.4504, "step": 2014 }, { "epoch": 0.35, "learning_rate": 4.9219535790432574e-05, "loss": 0.4497, "step": 2015 }, { "epoch": 0.35, "learning_rate": 4.9217746949273104e-05, "loss": 0.4198, "step": 2016 }, { "epoch": 0.35, "learning_rate": 4.9215956096694134e-05, "loss": 0.4289, "step": 2017 }, { "epoch": 0.35, "learning_rate": 4.921416323286153e-05, "loss": 0.4297, "step": 2018 }, { "epoch": 0.35, "learning_rate": 4.921236835794133e-05, "loss": 0.4418, "step": 2019 }, { "epoch": 0.35, "learning_rate": 4.921057147209979e-05, "loss": 0.4352, "step": 2020 }, { "epoch": 0.35, "learning_rate": 4.92087725755033e-05, "loss": 0.4258, "step": 2021 }, { "epoch": 0.35, "learning_rate": 4.9206971668318495e-05, "loss": 0.4523, "step": 2022 }, { "epoch": 0.35, "learning_rate": 4.920516875071217e-05, "loss": 0.4287, "step": 2023 }, { "epoch": 0.35, "learning_rate": 4.920336382285127e-05, "loss": 0.4296, "step": 2024 }, { "epoch": 0.35, "learning_rate": 4.9201556884903e-05, "loss": 0.4757, "step": 2025 }, { "epoch": 0.35, "learning_rate": 4.9199747937034695e-05, "loss": 0.4453, "step": 2026 }, { "epoch": 0.35, "learning_rate": 4.9197936979413894e-05, "loss": 0.4538, "step": 2027 }, { "epoch": 0.35, "learning_rate": 4.919612401220832e-05, "loss": 0.4422, "step": 2028 }, { "epoch": 0.35, "learning_rate": 4.919430903558589e-05, "loss": 0.4511, "step": 2029 }, { "epoch": 0.35, "learning_rate": 4.91924920497147e-05, "loss": 0.4273, "step": 2030 }, { "epoch": 0.35, "learning_rate": 4.919067305476302e-05, "loss": 0.4297, "step": 2031 }, { "epoch": 0.35, "learning_rate": 4.9188852050899336e-05, "loss": 0.4443, "step": 2032 }, { "epoch": 0.35, "learning_rate": 4.918702903829228e-05, "loss": 0.43, "step": 2033 }, { "epoch": 0.35, "learning_rate": 4.9185204017110725e-05, "loss": 0.4314, "step": 2034 }, { "epoch": 0.35, "learning_rate": 4.918337698752367e-05, "loss": 0.433, "step": 2035 }, { "epoch": 0.35, "learning_rate": 4.9181547949700346e-05, "loss": 0.4211, "step": 2036 }, { "epoch": 0.36, "learning_rate": 4.917971690381014e-05, "loss": 0.4497, "step": 2037 }, { "epoch": 0.36, "learning_rate": 4.917788385002264e-05, "loss": 0.4479, "step": 2038 }, { "epoch": 0.36, "learning_rate": 4.9176048788507615e-05, "loss": 0.4505, "step": 2039 }, { "epoch": 0.36, "learning_rate": 4.9174211719435034e-05, "loss": 0.449, "step": 2040 }, { "epoch": 0.36, "learning_rate": 4.9172372642975026e-05, "loss": 0.4583, "step": 2041 }, { "epoch": 0.36, "learning_rate": 4.917053155929792e-05, "loss": 0.4542, "step": 2042 }, { "epoch": 0.36, "learning_rate": 4.916868846857424e-05, "loss": 0.4424, "step": 2043 }, { "epoch": 0.36, "learning_rate": 4.916684337097468e-05, "loss": 0.4607, "step": 2044 }, { "epoch": 0.36, "learning_rate": 4.916499626667012e-05, "loss": 0.4455, "step": 2045 }, { "epoch": 0.36, "learning_rate": 4.916314715583165e-05, "loss": 0.4406, "step": 2046 }, { "epoch": 0.36, "learning_rate": 4.916129603863051e-05, "loss": 0.4193, "step": 2047 }, { "epoch": 0.36, "learning_rate": 4.915944291523816e-05, "loss": 0.4468, "step": 2048 }, { "epoch": 0.36, "learning_rate": 4.9157587785826214e-05, "loss": 0.4454, "step": 2049 }, { "epoch": 0.36, "learning_rate": 4.91557306505665e-05, "loss": 0.4402, "step": 2050 }, { "epoch": 0.36, "learning_rate": 4.9153871509631004e-05, "loss": 0.4481, "step": 2051 }, { "epoch": 0.36, "learning_rate": 4.9152010363191926e-05, "loss": 0.4455, "step": 2052 }, { "epoch": 0.36, "learning_rate": 4.9150147211421644e-05, "loss": 0.4403, "step": 2053 }, { "epoch": 0.36, "learning_rate": 4.91482820544927e-05, "loss": 0.4512, "step": 2054 }, { "epoch": 0.36, "learning_rate": 4.914641489257785e-05, "loss": 0.4244, "step": 2055 }, { "epoch": 0.36, "learning_rate": 4.914454572585001e-05, "loss": 0.451, "step": 2056 }, { "epoch": 0.36, "learning_rate": 4.914267455448231e-05, "loss": 0.4506, "step": 2057 }, { "epoch": 0.36, "learning_rate": 4.9140801378648046e-05, "loss": 0.431, "step": 2058 }, { "epoch": 0.36, "learning_rate": 4.913892619852071e-05, "loss": 0.4584, "step": 2059 }, { "epoch": 0.36, "learning_rate": 4.9137049014273963e-05, "loss": 0.4343, "step": 2060 }, { "epoch": 0.36, "learning_rate": 4.913516982608167e-05, "loss": 0.4502, "step": 2061 }, { "epoch": 0.36, "learning_rate": 4.913328863411787e-05, "loss": 0.4287, "step": 2062 }, { "epoch": 0.36, "learning_rate": 4.9131405438556806e-05, "loss": 0.4452, "step": 2063 }, { "epoch": 0.36, "learning_rate": 4.912952023957289e-05, "loss": 0.442, "step": 2064 }, { "epoch": 0.36, "learning_rate": 4.91276330373407e-05, "loss": 0.4347, "step": 2065 }, { "epoch": 0.36, "learning_rate": 4.912574383203503e-05, "loss": 0.4563, "step": 2066 }, { "epoch": 0.36, "learning_rate": 4.9123852623830876e-05, "loss": 0.4566, "step": 2067 }, { "epoch": 0.36, "learning_rate": 4.912195941290337e-05, "loss": 0.4341, "step": 2068 }, { "epoch": 0.36, "learning_rate": 4.912006419942787e-05, "loss": 0.4457, "step": 2069 }, { "epoch": 0.36, "learning_rate": 4.911816698357988e-05, "loss": 0.4319, "step": 2070 }, { "epoch": 0.36, "learning_rate": 4.911626776553514e-05, "loss": 0.4399, "step": 2071 }, { "epoch": 0.36, "learning_rate": 4.911436654546954e-05, "loss": 0.4557, "step": 2072 }, { "epoch": 0.36, "learning_rate": 4.911246332355915e-05, "loss": 0.4397, "step": 2073 }, { "epoch": 0.36, "learning_rate": 4.911055809998026e-05, "loss": 0.4549, "step": 2074 }, { "epoch": 0.36, "learning_rate": 4.910865087490932e-05, "loss": 0.4371, "step": 2075 }, { "epoch": 0.36, "learning_rate": 4.910674164852296e-05, "loss": 0.4671, "step": 2076 }, { "epoch": 0.36, "learning_rate": 4.9104830420998014e-05, "loss": 0.4516, "step": 2077 }, { "epoch": 0.36, "learning_rate": 4.9102917192511494e-05, "loss": 0.4402, "step": 2078 }, { "epoch": 0.36, "learning_rate": 4.910100196324059e-05, "loss": 0.4541, "step": 2079 }, { "epoch": 0.36, "learning_rate": 4.909908473336269e-05, "loss": 0.4233, "step": 2080 }, { "epoch": 0.36, "learning_rate": 4.909716550305537e-05, "loss": 0.4403, "step": 2081 }, { "epoch": 0.36, "learning_rate": 4.909524427249635e-05, "loss": 0.4629, "step": 2082 }, { "epoch": 0.36, "learning_rate": 4.909332104186359e-05, "loss": 0.4256, "step": 2083 }, { "epoch": 0.36, "learning_rate": 4.909139581133522e-05, "loss": 0.4546, "step": 2084 }, { "epoch": 0.36, "learning_rate": 4.908946858108953e-05, "loss": 0.4262, "step": 2085 }, { "epoch": 0.36, "learning_rate": 4.908753935130503e-05, "loss": 0.4441, "step": 2086 }, { "epoch": 0.36, "learning_rate": 4.908560812216039e-05, "loss": 0.4411, "step": 2087 }, { "epoch": 0.36, "learning_rate": 4.9083674893834467e-05, "loss": 0.4486, "step": 2088 }, { "epoch": 0.36, "learning_rate": 4.908173966650632e-05, "loss": 0.4515, "step": 2089 }, { "epoch": 0.36, "learning_rate": 4.907980244035518e-05, "loss": 0.436, "step": 2090 }, { "epoch": 0.36, "learning_rate": 4.907786321556046e-05, "loss": 0.4233, "step": 2091 }, { "epoch": 0.36, "learning_rate": 4.9075921992301774e-05, "loss": 0.4342, "step": 2092 }, { "epoch": 0.36, "learning_rate": 4.90739787707589e-05, "loss": 0.4237, "step": 2093 }, { "epoch": 0.37, "learning_rate": 4.907203355111183e-05, "loss": 0.4445, "step": 2094 }, { "epoch": 0.37, "learning_rate": 4.90700863335407e-05, "loss": 0.4447, "step": 2095 }, { "epoch": 0.37, "learning_rate": 4.906813711822588e-05, "loss": 0.4424, "step": 2096 }, { "epoch": 0.37, "learning_rate": 4.9066185905347866e-05, "loss": 0.4353, "step": 2097 }, { "epoch": 0.37, "learning_rate": 4.9064232695087404e-05, "loss": 0.4251, "step": 2098 }, { "epoch": 0.37, "learning_rate": 4.906227748762537e-05, "loss": 0.4264, "step": 2099 }, { "epoch": 0.37, "learning_rate": 4.906032028314287e-05, "loss": 0.4346, "step": 2100 }, { "epoch": 0.37, "learning_rate": 4.905836108182116e-05, "loss": 0.45, "step": 2101 }, { "epoch": 0.37, "learning_rate": 4.905639988384169e-05, "loss": 0.4462, "step": 2102 }, { "epoch": 0.37, "learning_rate": 4.905443668938611e-05, "loss": 0.4376, "step": 2103 }, { "epoch": 0.37, "learning_rate": 4.905247149863624e-05, "loss": 0.4717, "step": 2104 }, { "epoch": 0.37, "learning_rate": 4.905050431177409e-05, "loss": 0.4361, "step": 2105 }, { "epoch": 0.37, "learning_rate": 4.9048535128981846e-05, "loss": 0.4395, "step": 2106 }, { "epoch": 0.37, "learning_rate": 4.904656395044189e-05, "loss": 0.4503, "step": 2107 }, { "epoch": 0.37, "learning_rate": 4.904459077633679e-05, "loss": 0.4227, "step": 2108 }, { "epoch": 0.37, "learning_rate": 4.90426156068493e-05, "loss": 0.4403, "step": 2109 }, { "epoch": 0.37, "learning_rate": 4.904063844216235e-05, "loss": 0.4446, "step": 2110 }, { "epoch": 0.37, "learning_rate": 4.903865928245904e-05, "loss": 0.456, "step": 2111 }, { "epoch": 0.37, "learning_rate": 4.903667812792269e-05, "loss": 0.4325, "step": 2112 }, { "epoch": 0.37, "learning_rate": 4.903469497873678e-05, "loss": 0.4422, "step": 2113 }, { "epoch": 0.37, "learning_rate": 4.903270983508499e-05, "loss": 0.4172, "step": 2114 }, { "epoch": 0.37, "learning_rate": 4.903072269715118e-05, "loss": 0.4449, "step": 2115 }, { "epoch": 0.37, "learning_rate": 4.9028733565119366e-05, "loss": 0.4274, "step": 2116 }, { "epoch": 0.37, "learning_rate": 4.902674243917381e-05, "loss": 0.4411, "step": 2117 }, { "epoch": 0.37, "learning_rate": 4.902474931949889e-05, "loss": 0.4503, "step": 2118 }, { "epoch": 0.37, "learning_rate": 4.902275420627922e-05, "loss": 0.4632, "step": 2119 }, { "epoch": 0.37, "learning_rate": 4.9020757099699585e-05, "loss": 0.4475, "step": 2120 }, { "epoch": 0.37, "learning_rate": 4.9018757999944933e-05, "loss": 0.4332, "step": 2121 }, { "epoch": 0.37, "learning_rate": 4.901675690720043e-05, "loss": 0.4429, "step": 2122 }, { "epoch": 0.37, "learning_rate": 4.90147538216514e-05, "loss": 0.4407, "step": 2123 }, { "epoch": 0.37, "learning_rate": 4.901274874348336e-05, "loss": 0.4272, "step": 2124 }, { "epoch": 0.37, "learning_rate": 4.901074167288201e-05, "loss": 0.4281, "step": 2125 }, { "epoch": 0.37, "learning_rate": 4.900873261003326e-05, "loss": 0.4407, "step": 2126 }, { "epoch": 0.37, "learning_rate": 4.900672155512316e-05, "loss": 0.4487, "step": 2127 }, { "epoch": 0.37, "learning_rate": 4.9004708508337964e-05, "loss": 0.4426, "step": 2128 }, { "epoch": 0.37, "learning_rate": 4.900269346986413e-05, "loss": 0.4462, "step": 2129 }, { "epoch": 0.37, "learning_rate": 4.900067643988828e-05, "loss": 0.4314, "step": 2130 }, { "epoch": 0.37, "learning_rate": 4.8998657418597216e-05, "loss": 0.4524, "step": 2131 }, { "epoch": 0.37, "learning_rate": 4.8996636406177936e-05, "loss": 0.4405, "step": 2132 }, { "epoch": 0.37, "learning_rate": 4.899461340281762e-05, "loss": 0.4389, "step": 2133 }, { "epoch": 0.37, "learning_rate": 4.899258840870363e-05, "loss": 0.4473, "step": 2134 }, { "epoch": 0.37, "learning_rate": 4.8990561424023526e-05, "loss": 0.4499, "step": 2135 }, { "epoch": 0.37, "learning_rate": 4.898853244896501e-05, "loss": 0.4424, "step": 2136 }, { "epoch": 0.37, "learning_rate": 4.8986501483716025e-05, "loss": 0.44, "step": 2137 }, { "epoch": 0.37, "learning_rate": 4.898446852846466e-05, "loss": 0.4237, "step": 2138 }, { "epoch": 0.37, "learning_rate": 4.8982433583399205e-05, "loss": 0.4336, "step": 2139 }, { "epoch": 0.37, "learning_rate": 4.898039664870813e-05, "loss": 0.4404, "step": 2140 }, { "epoch": 0.37, "learning_rate": 4.8978357724580085e-05, "loss": 0.4497, "step": 2141 }, { "epoch": 0.37, "learning_rate": 4.897631681120391e-05, "loss": 0.4415, "step": 2142 }, { "epoch": 0.37, "learning_rate": 4.897427390876862e-05, "loss": 0.4505, "step": 2143 }, { "epoch": 0.37, "learning_rate": 4.897222901746343e-05, "loss": 0.4394, "step": 2144 }, { "epoch": 0.37, "learning_rate": 4.8970182137477735e-05, "loss": 0.4511, "step": 2145 }, { "epoch": 0.37, "learning_rate": 4.8968133269001096e-05, "loss": 0.429, "step": 2146 }, { "epoch": 0.37, "learning_rate": 4.896608241222329e-05, "loss": 0.4559, "step": 2147 }, { "epoch": 0.37, "learning_rate": 4.896402956733423e-05, "loss": 0.4268, "step": 2148 }, { "epoch": 0.37, "learning_rate": 4.8961974734524074e-05, "loss": 0.4155, "step": 2149 }, { "epoch": 0.37, "learning_rate": 4.8959917913983114e-05, "loss": 0.4468, "step": 2150 }, { "epoch": 0.38, "learning_rate": 4.8957859105901855e-05, "loss": 0.4446, "step": 2151 }, { "epoch": 0.38, "learning_rate": 4.895579831047097e-05, "loss": 0.4414, "step": 2152 }, { "epoch": 0.38, "learning_rate": 4.895373552788133e-05, "loss": 0.4172, "step": 2153 }, { "epoch": 0.38, "learning_rate": 4.895167075832398e-05, "loss": 0.416, "step": 2154 }, { "epoch": 0.38, "learning_rate": 4.894960400199015e-05, "loss": 0.4503, "step": 2155 }, { "epoch": 0.38, "learning_rate": 4.894753525907124e-05, "loss": 0.4273, "step": 2156 }, { "epoch": 0.38, "learning_rate": 4.8945464529758884e-05, "loss": 0.4412, "step": 2157 }, { "epoch": 0.38, "learning_rate": 4.8943391814244836e-05, "loss": 0.4428, "step": 2158 }, { "epoch": 0.38, "learning_rate": 4.8941317112721076e-05, "loss": 0.4299, "step": 2159 }, { "epoch": 0.38, "learning_rate": 4.893924042537975e-05, "loss": 0.4476, "step": 2160 }, { "epoch": 0.38, "learning_rate": 4.893716175241321e-05, "loss": 0.4311, "step": 2161 }, { "epoch": 0.38, "learning_rate": 4.893508109401395e-05, "loss": 0.438, "step": 2162 }, { "epoch": 0.38, "learning_rate": 4.893299845037468e-05, "loss": 0.4396, "step": 2163 }, { "epoch": 0.38, "learning_rate": 4.89309138216883e-05, "loss": 0.4593, "step": 2164 }, { "epoch": 0.38, "learning_rate": 4.892882720814786e-05, "loss": 0.4319, "step": 2165 }, { "epoch": 0.38, "learning_rate": 4.892673860994664e-05, "loss": 0.4405, "step": 2166 }, { "epoch": 0.38, "learning_rate": 4.8924648027278066e-05, "loss": 0.4522, "step": 2167 }, { "epoch": 0.38, "learning_rate": 4.892255546033576e-05, "loss": 0.4423, "step": 2168 }, { "epoch": 0.38, "learning_rate": 4.892046090931352e-05, "loss": 0.4554, "step": 2169 }, { "epoch": 0.38, "learning_rate": 4.8918364374405345e-05, "loss": 0.4261, "step": 2170 }, { "epoch": 0.38, "learning_rate": 4.891626585580541e-05, "loss": 0.4473, "step": 2171 }, { "epoch": 0.38, "learning_rate": 4.8914165353708075e-05, "loss": 0.4443, "step": 2172 }, { "epoch": 0.38, "learning_rate": 4.8912062868307865e-05, "loss": 0.4373, "step": 2173 }, { "epoch": 0.38, "learning_rate": 4.890995839979952e-05, "loss": 0.4261, "step": 2174 }, { "epoch": 0.38, "learning_rate": 4.8907851948377945e-05, "loss": 0.448, "step": 2175 }, { "epoch": 0.38, "learning_rate": 4.890574351423822e-05, "loss": 0.4333, "step": 2176 }, { "epoch": 0.38, "learning_rate": 4.890363309757565e-05, "loss": 0.4479, "step": 2177 }, { "epoch": 0.38, "learning_rate": 4.8901520698585653e-05, "loss": 0.4318, "step": 2178 }, { "epoch": 0.38, "learning_rate": 4.889940631746391e-05, "loss": 0.4553, "step": 2179 }, { "epoch": 0.38, "learning_rate": 4.889728995440622e-05, "loss": 0.4438, "step": 2180 }, { "epoch": 0.38, "learning_rate": 4.889517160960861e-05, "loss": 0.4533, "step": 2181 }, { "epoch": 0.38, "learning_rate": 4.8893051283267266e-05, "loss": 0.4267, "step": 2182 }, { "epoch": 0.38, "learning_rate": 4.889092897557856e-05, "loss": 0.4275, "step": 2183 }, { "epoch": 0.38, "learning_rate": 4.888880468673907e-05, "loss": 0.4336, "step": 2184 }, { "epoch": 0.38, "learning_rate": 4.888667841694552e-05, "loss": 0.4419, "step": 2185 }, { "epoch": 0.38, "learning_rate": 4.888455016639485e-05, "loss": 0.4459, "step": 2186 }, { "epoch": 0.38, "learning_rate": 4.888241993528417e-05, "loss": 0.4474, "step": 2187 }, { "epoch": 0.38, "learning_rate": 4.8880287723810766e-05, "loss": 0.4409, "step": 2188 }, { "epoch": 0.38, "learning_rate": 4.887815353217212e-05, "loss": 0.419, "step": 2189 }, { "epoch": 0.38, "learning_rate": 4.8876017360565904e-05, "loss": 0.4394, "step": 2190 }, { "epoch": 0.38, "learning_rate": 4.887387920918994e-05, "loss": 0.4282, "step": 2191 }, { "epoch": 0.38, "learning_rate": 4.8871739078242285e-05, "loss": 0.4163, "step": 2192 }, { "epoch": 0.38, "learning_rate": 4.886959696792112e-05, "loss": 0.4207, "step": 2193 }, { "epoch": 0.38, "learning_rate": 4.8867452878424864e-05, "loss": 0.4467, "step": 2194 }, { "epoch": 0.38, "learning_rate": 4.886530680995208e-05, "loss": 0.4456, "step": 2195 }, { "epoch": 0.38, "learning_rate": 4.8863158762701536e-05, "loss": 0.4444, "step": 2196 }, { "epoch": 0.38, "learning_rate": 4.8861008736872166e-05, "loss": 0.4258, "step": 2197 }, { "epoch": 0.38, "learning_rate": 4.88588567326631e-05, "loss": 0.4429, "step": 2198 }, { "epoch": 0.38, "learning_rate": 4.885670275027368e-05, "loss": 0.4219, "step": 2199 }, { "epoch": 0.38, "learning_rate": 4.885454678990336e-05, "loss": 0.4242, "step": 2200 }, { "epoch": 0.38, "learning_rate": 4.885238885175182e-05, "loss": 0.4345, "step": 2201 }, { "epoch": 0.38, "learning_rate": 4.885022893601894e-05, "loss": 0.4328, "step": 2202 }, { "epoch": 0.38, "learning_rate": 4.884806704290477e-05, "loss": 0.4125, "step": 2203 }, { "epoch": 0.38, "learning_rate": 4.8845903172609507e-05, "loss": 0.4531, "step": 2204 }, { "epoch": 0.38, "learning_rate": 4.884373732533358e-05, "loss": 0.4061, "step": 2205 }, { "epoch": 0.38, "learning_rate": 4.8841569501277587e-05, "loss": 0.441, "step": 2206 }, { "epoch": 0.38, "learning_rate": 4.8839399700642285e-05, "loss": 0.466, "step": 2207 }, { "epoch": 0.38, "learning_rate": 4.883722792362865e-05, "loss": 0.4453, "step": 2208 }, { "epoch": 0.39, "learning_rate": 4.883505417043782e-05, "loss": 0.4406, "step": 2209 }, { "epoch": 0.39, "learning_rate": 4.883287844127111e-05, "loss": 0.4452, "step": 2210 }, { "epoch": 0.39, "learning_rate": 4.883070073633004e-05, "loss": 0.4285, "step": 2211 }, { "epoch": 0.39, "learning_rate": 4.882852105581629e-05, "loss": 0.4254, "step": 2212 }, { "epoch": 0.39, "learning_rate": 4.882633939993175e-05, "loss": 0.4416, "step": 2213 }, { "epoch": 0.39, "learning_rate": 4.882415576887847e-05, "loss": 0.4385, "step": 2214 }, { "epoch": 0.39, "learning_rate": 4.8821970162858685e-05, "loss": 0.4235, "step": 2215 }, { "epoch": 0.39, "learning_rate": 4.881978258207482e-05, "loss": 0.4439, "step": 2216 }, { "epoch": 0.39, "learning_rate": 4.881759302672949e-05, "loss": 0.4272, "step": 2217 }, { "epoch": 0.39, "learning_rate": 4.881540149702548e-05, "loss": 0.4401, "step": 2218 }, { "epoch": 0.39, "learning_rate": 4.8813207993165736e-05, "loss": 0.427, "step": 2219 }, { "epoch": 0.39, "learning_rate": 4.8811012515353456e-05, "loss": 0.4306, "step": 2220 }, { "epoch": 0.39, "learning_rate": 4.880881506379195e-05, "loss": 0.4409, "step": 2221 }, { "epoch": 0.39, "learning_rate": 4.880661563868473e-05, "loss": 0.4417, "step": 2222 }, { "epoch": 0.39, "learning_rate": 4.8804414240235533e-05, "loss": 0.4171, "step": 2223 }, { "epoch": 0.39, "learning_rate": 4.880221086864821e-05, "loss": 0.4415, "step": 2224 }, { "epoch": 0.39, "learning_rate": 4.880000552412685e-05, "loss": 0.4473, "step": 2225 }, { "epoch": 0.39, "learning_rate": 4.879779820687569e-05, "loss": 0.4284, "step": 2226 }, { "epoch": 0.39, "learning_rate": 4.8795588917099185e-05, "loss": 0.4331, "step": 2227 }, { "epoch": 0.39, "learning_rate": 4.8793377655001935e-05, "loss": 0.419, "step": 2228 }, { "epoch": 0.39, "learning_rate": 4.879116442078873e-05, "loss": 0.4534, "step": 2229 }, { "epoch": 0.39, "learning_rate": 4.878894921466457e-05, "loss": 0.4148, "step": 2230 }, { "epoch": 0.39, "learning_rate": 4.8786732036834616e-05, "loss": 0.4433, "step": 2231 }, { "epoch": 0.39, "learning_rate": 4.878451288750421e-05, "loss": 0.426, "step": 2232 }, { "epoch": 0.39, "learning_rate": 4.878229176687888e-05, "loss": 0.4424, "step": 2233 }, { "epoch": 0.39, "learning_rate": 4.878006867516434e-05, "loss": 0.4437, "step": 2234 }, { "epoch": 0.39, "learning_rate": 4.87778436125665e-05, "loss": 0.4563, "step": 2235 }, { "epoch": 0.39, "learning_rate": 4.8775616579291414e-05, "loss": 0.4388, "step": 2236 }, { "epoch": 0.39, "learning_rate": 4.877338757554534e-05, "loss": 0.4431, "step": 2237 }, { "epoch": 0.39, "learning_rate": 4.877115660153476e-05, "loss": 0.4405, "step": 2238 }, { "epoch": 0.39, "learning_rate": 4.8768923657466245e-05, "loss": 0.4305, "step": 2239 }, { "epoch": 0.39, "learning_rate": 4.876668874354664e-05, "loss": 0.4243, "step": 2240 }, { "epoch": 0.39, "learning_rate": 4.876445185998292e-05, "loss": 0.4365, "step": 2241 }, { "epoch": 0.39, "learning_rate": 4.876221300698226e-05, "loss": 0.4303, "step": 2242 }, { "epoch": 0.39, "learning_rate": 4.875997218475202e-05, "loss": 0.4357, "step": 2243 }, { "epoch": 0.39, "learning_rate": 4.875772939349972e-05, "loss": 0.4521, "step": 2244 }, { "epoch": 0.39, "learning_rate": 4.87554846334331e-05, "loss": 0.4453, "step": 2245 }, { "epoch": 0.39, "learning_rate": 4.8753237904760046e-05, "loss": 0.4492, "step": 2246 }, { "epoch": 0.39, "learning_rate": 4.8750989207688655e-05, "loss": 0.4437, "step": 2247 }, { "epoch": 0.39, "learning_rate": 4.874873854242718e-05, "loss": 0.437, "step": 2248 }, { "epoch": 0.39, "learning_rate": 4.874648590918408e-05, "loss": 0.4402, "step": 2249 }, { "epoch": 0.39, "learning_rate": 4.874423130816798e-05, "loss": 0.4285, "step": 2250 }, { "epoch": 0.39, "learning_rate": 4.874197473958769e-05, "loss": 0.4328, "step": 2251 }, { "epoch": 0.39, "learning_rate": 4.873971620365221e-05, "loss": 0.4566, "step": 2252 }, { "epoch": 0.39, "learning_rate": 4.873745570057072e-05, "loss": 0.428, "step": 2253 }, { "epoch": 0.39, "learning_rate": 4.873519323055258e-05, "loss": 0.4372, "step": 2254 }, { "epoch": 0.39, "learning_rate": 4.873292879380733e-05, "loss": 0.4661, "step": 2255 }, { "epoch": 0.39, "learning_rate": 4.873066239054468e-05, "loss": 0.4462, "step": 2256 }, { "epoch": 0.39, "learning_rate": 4.872839402097456e-05, "loss": 0.4461, "step": 2257 }, { "epoch": 0.39, "learning_rate": 4.872612368530704e-05, "loss": 0.4374, "step": 2258 }, { "epoch": 0.39, "learning_rate": 4.87238513837524e-05, "loss": 0.418, "step": 2259 }, { "epoch": 0.39, "learning_rate": 4.8721577116521086e-05, "loss": 0.4311, "step": 2260 }, { "epoch": 0.39, "learning_rate": 4.871930088382375e-05, "loss": 0.453, "step": 2261 }, { "epoch": 0.39, "learning_rate": 4.871702268587118e-05, "loss": 0.4371, "step": 2262 }, { "epoch": 0.39, "learning_rate": 4.87147425228744e-05, "loss": 0.418, "step": 2263 }, { "epoch": 0.39, "learning_rate": 4.871246039504457e-05, "loss": 0.4438, "step": 2264 }, { "epoch": 0.39, "learning_rate": 4.871017630259307e-05, "loss": 0.4284, "step": 2265 }, { "epoch": 0.4, "learning_rate": 4.870789024573143e-05, "loss": 0.4469, "step": 2266 }, { "epoch": 0.4, "learning_rate": 4.8705602224671385e-05, "loss": 0.4409, "step": 2267 }, { "epoch": 0.4, "learning_rate": 4.870331223962484e-05, "loss": 0.4474, "step": 2268 }, { "epoch": 0.4, "learning_rate": 4.870102029080388e-05, "loss": 0.4466, "step": 2269 }, { "epoch": 0.4, "learning_rate": 4.86987263784208e-05, "loss": 0.446, "step": 2270 }, { "epoch": 0.4, "learning_rate": 4.8696430502688025e-05, "loss": 0.4214, "step": 2271 }, { "epoch": 0.4, "learning_rate": 4.8694132663818196e-05, "loss": 0.4235, "step": 2272 }, { "epoch": 0.4, "learning_rate": 4.869183286202416e-05, "loss": 0.4422, "step": 2273 }, { "epoch": 0.4, "learning_rate": 4.8689531097518876e-05, "loss": 0.4199, "step": 2274 }, { "epoch": 0.4, "learning_rate": 4.8687227370515546e-05, "loss": 0.4386, "step": 2275 }, { "epoch": 0.4, "learning_rate": 4.868492168122754e-05, "loss": 0.4347, "step": 2276 }, { "epoch": 0.4, "learning_rate": 4.868261402986837e-05, "loss": 0.4571, "step": 2277 }, { "epoch": 0.4, "learning_rate": 4.868030441665181e-05, "loss": 0.434, "step": 2278 }, { "epoch": 0.4, "learning_rate": 4.8677992841791726e-05, "loss": 0.4466, "step": 2279 }, { "epoch": 0.4, "learning_rate": 4.8675679305502234e-05, "loss": 0.4492, "step": 2280 }, { "epoch": 0.4, "learning_rate": 4.867336380799759e-05, "loss": 0.436, "step": 2281 }, { "epoch": 0.4, "learning_rate": 4.867104634949225e-05, "loss": 0.4365, "step": 2282 }, { "epoch": 0.4, "learning_rate": 4.866872693020086e-05, "loss": 0.4247, "step": 2283 }, { "epoch": 0.4, "learning_rate": 4.866640555033822e-05, "loss": 0.4499, "step": 2284 }, { "epoch": 0.4, "learning_rate": 4.866408221011934e-05, "loss": 0.4527, "step": 2285 }, { "epoch": 0.4, "learning_rate": 4.8661756909759405e-05, "loss": 0.4135, "step": 2286 }, { "epoch": 0.4, "learning_rate": 4.865942964947375e-05, "loss": 0.4374, "step": 2287 }, { "epoch": 0.4, "learning_rate": 4.8657100429477944e-05, "loss": 0.4463, "step": 2288 }, { "epoch": 0.4, "learning_rate": 4.8654769249987696e-05, "loss": 0.45, "step": 2289 }, { "epoch": 0.4, "learning_rate": 4.8652436111218916e-05, "loss": 0.4259, "step": 2290 }, { "epoch": 0.4, "learning_rate": 4.8650101013387694e-05, "loss": 0.4444, "step": 2291 }, { "epoch": 0.4, "learning_rate": 4.8647763956710294e-05, "loss": 0.4381, "step": 2292 }, { "epoch": 0.4, "learning_rate": 4.8645424941403164e-05, "loss": 0.4195, "step": 2293 }, { "epoch": 0.4, "learning_rate": 4.864308396768294e-05, "loss": 0.4445, "step": 2294 }, { "epoch": 0.4, "learning_rate": 4.8640741035766436e-05, "loss": 0.4511, "step": 2295 }, { "epoch": 0.4, "learning_rate": 4.863839614587064e-05, "loss": 0.452, "step": 2296 }, { "epoch": 0.4, "learning_rate": 4.8636049298212735e-05, "loss": 0.4457, "step": 2297 }, { "epoch": 0.4, "learning_rate": 4.863370049301006e-05, "loss": 0.4132, "step": 2298 }, { "epoch": 0.4, "learning_rate": 4.8631349730480174e-05, "loss": 0.451, "step": 2299 }, { "epoch": 0.4, "learning_rate": 4.8628997010840785e-05, "loss": 0.4241, "step": 2300 }, { "epoch": 0.4, "learning_rate": 4.8626642334309804e-05, "loss": 0.4328, "step": 2301 }, { "epoch": 0.4, "learning_rate": 4.862428570110529e-05, "loss": 0.4349, "step": 2302 }, { "epoch": 0.4, "learning_rate": 4.862192711144554e-05, "loss": 0.4402, "step": 2303 }, { "epoch": 0.4, "learning_rate": 4.861956656554896e-05, "loss": 0.4508, "step": 2304 }, { "epoch": 0.4, "learning_rate": 4.86172040636342e-05, "loss": 0.4177, "step": 2305 }, { "epoch": 0.4, "learning_rate": 4.8614839605920066e-05, "loss": 0.4282, "step": 2306 }, { "epoch": 0.4, "learning_rate": 4.861247319262553e-05, "loss": 0.4406, "step": 2307 }, { "epoch": 0.4, "learning_rate": 4.861010482396978e-05, "loss": 0.4373, "step": 2308 }, { "epoch": 0.4, "learning_rate": 4.8607734500172154e-05, "loss": 0.4253, "step": 2309 }, { "epoch": 0.4, "learning_rate": 4.860536222145218e-05, "loss": 0.4345, "step": 2310 }, { "epoch": 0.4, "learning_rate": 4.8602987988029584e-05, "loss": 0.4463, "step": 2311 }, { "epoch": 0.4, "learning_rate": 4.860061180012425e-05, "loss": 0.4156, "step": 2312 }, { "epoch": 0.4, "learning_rate": 4.8598233657956235e-05, "loss": 0.4186, "step": 2313 }, { "epoch": 0.4, "learning_rate": 4.8595853561745825e-05, "loss": 0.4577, "step": 2314 }, { "epoch": 0.4, "learning_rate": 4.859347151171344e-05, "loss": 0.4182, "step": 2315 }, { "epoch": 0.4, "learning_rate": 4.85910875080797e-05, "loss": 0.4415, "step": 2316 }, { "epoch": 0.4, "learning_rate": 4.85887015510654e-05, "loss": 0.4194, "step": 2317 }, { "epoch": 0.4, "learning_rate": 4.8586313640891516e-05, "loss": 0.4249, "step": 2318 }, { "epoch": 0.4, "learning_rate": 4.8583923777779226e-05, "loss": 0.4308, "step": 2319 }, { "epoch": 0.4, "learning_rate": 4.858153196194984e-05, "loss": 0.4445, "step": 2320 }, { "epoch": 0.4, "learning_rate": 4.85791381936249e-05, "loss": 0.4298, "step": 2321 }, { "epoch": 0.4, "learning_rate": 4.857674247302611e-05, "loss": 0.4422, "step": 2322 }, { "epoch": 0.41, "learning_rate": 4.8574344800375344e-05, "loss": 0.4205, "step": 2323 }, { "epoch": 0.41, "learning_rate": 4.857194517589467e-05, "loss": 0.4291, "step": 2324 }, { "epoch": 0.41, "learning_rate": 4.8569543599806325e-05, "loss": 0.4425, "step": 2325 }, { "epoch": 0.41, "learning_rate": 4.856714007233275e-05, "loss": 0.4333, "step": 2326 }, { "epoch": 0.41, "learning_rate": 4.856473459369654e-05, "loss": 0.4315, "step": 2327 }, { "epoch": 0.41, "learning_rate": 4.856232716412048e-05, "loss": 0.4577, "step": 2328 }, { "epoch": 0.41, "learning_rate": 4.855991778382754e-05, "loss": 0.4329, "step": 2329 }, { "epoch": 0.41, "learning_rate": 4.855750645304087e-05, "loss": 0.4305, "step": 2330 }, { "epoch": 0.41, "learning_rate": 4.8555093171983805e-05, "loss": 0.4241, "step": 2331 }, { "epoch": 0.41, "learning_rate": 4.8552677940879845e-05, "loss": 0.4553, "step": 2332 }, { "epoch": 0.41, "learning_rate": 4.8550260759952674e-05, "loss": 0.4469, "step": 2333 }, { "epoch": 0.41, "learning_rate": 4.854784162942618e-05, "loss": 0.4382, "step": 2334 }, { "epoch": 0.41, "learning_rate": 4.8545420549524394e-05, "loss": 0.4194, "step": 2335 }, { "epoch": 0.41, "learning_rate": 4.8542997520471576e-05, "loss": 0.4175, "step": 2336 }, { "epoch": 0.41, "learning_rate": 4.854057254249211e-05, "loss": 0.4287, "step": 2337 }, { "epoch": 0.41, "learning_rate": 4.8538145615810596e-05, "loss": 0.428, "step": 2338 }, { "epoch": 0.41, "learning_rate": 4.853571674065182e-05, "loss": 0.4315, "step": 2339 }, { "epoch": 0.41, "learning_rate": 4.853328591724072e-05, "loss": 0.432, "step": 2340 }, { "epoch": 0.41, "learning_rate": 4.853085314580244e-05, "loss": 0.4238, "step": 2341 }, { "epoch": 0.41, "learning_rate": 4.8528418426562286e-05, "loss": 0.4506, "step": 2342 }, { "epoch": 0.41, "learning_rate": 4.852598175974577e-05, "loss": 0.4519, "step": 2343 }, { "epoch": 0.41, "learning_rate": 4.852354314557855e-05, "loss": 0.424, "step": 2344 }, { "epoch": 0.41, "learning_rate": 4.852110258428648e-05, "loss": 0.4573, "step": 2345 }, { "epoch": 0.41, "learning_rate": 4.851866007609561e-05, "loss": 0.4561, "step": 2346 }, { "epoch": 0.41, "learning_rate": 4.8516215621232154e-05, "loss": 0.4428, "step": 2347 }, { "epoch": 0.41, "learning_rate": 4.851376921992249e-05, "loss": 0.4261, "step": 2348 }, { "epoch": 0.41, "learning_rate": 4.851132087239322e-05, "loss": 0.424, "step": 2349 }, { "epoch": 0.41, "learning_rate": 4.850887057887109e-05, "loss": 0.4369, "step": 2350 }, { "epoch": 0.41, "learning_rate": 4.850641833958303e-05, "loss": 0.4283, "step": 2351 }, { "epoch": 0.41, "learning_rate": 4.8503964154756175e-05, "loss": 0.4504, "step": 2352 }, { "epoch": 0.41, "learning_rate": 4.850150802461781e-05, "loss": 0.4534, "step": 2353 }, { "epoch": 0.41, "learning_rate": 4.849904994939541e-05, "loss": 0.4505, "step": 2354 }, { "epoch": 0.41, "learning_rate": 4.8496589929316644e-05, "loss": 0.4358, "step": 2355 }, { "epoch": 0.41, "learning_rate": 4.849412796460934e-05, "loss": 0.4195, "step": 2356 }, { "epoch": 0.41, "learning_rate": 4.8491664055501526e-05, "loss": 0.4209, "step": 2357 }, { "epoch": 0.41, "learning_rate": 4.8489198202221396e-05, "loss": 0.4599, "step": 2358 }, { "epoch": 0.41, "learning_rate": 4.8486730404997325e-05, "loss": 0.4381, "step": 2359 }, { "epoch": 0.41, "learning_rate": 4.8484260664057876e-05, "loss": 0.441, "step": 2360 }, { "epoch": 0.41, "learning_rate": 4.8481788979631796e-05, "loss": 0.4172, "step": 2361 }, { "epoch": 0.41, "learning_rate": 4.847931535194797e-05, "loss": 0.4214, "step": 2362 }, { "epoch": 0.41, "learning_rate": 4.847683978123554e-05, "loss": 0.4495, "step": 2363 }, { "epoch": 0.41, "learning_rate": 4.847436226772376e-05, "loss": 0.4128, "step": 2364 }, { "epoch": 0.41, "learning_rate": 4.847188281164209e-05, "loss": 0.4375, "step": 2365 }, { "epoch": 0.41, "learning_rate": 4.846940141322017e-05, "loss": 0.4394, "step": 2366 }, { "epoch": 0.41, "learning_rate": 4.8466918072687825e-05, "loss": 0.4359, "step": 2367 }, { "epoch": 0.41, "learning_rate": 4.8464432790275035e-05, "loss": 0.4338, "step": 2368 }, { "epoch": 0.41, "learning_rate": 4.8461945566212015e-05, "loss": 0.4296, "step": 2369 }, { "epoch": 0.41, "learning_rate": 4.8459456400729084e-05, "loss": 0.4327, "step": 2370 }, { "epoch": 0.41, "learning_rate": 4.845696529405679e-05, "loss": 0.4431, "step": 2371 }, { "epoch": 0.41, "learning_rate": 4.845447224642585e-05, "loss": 0.4312, "step": 2372 }, { "epoch": 0.41, "learning_rate": 4.8451977258067175e-05, "loss": 0.436, "step": 2373 }, { "epoch": 0.41, "learning_rate": 4.8449480329211826e-05, "loss": 0.4175, "step": 2374 }, { "epoch": 0.41, "learning_rate": 4.844698146009107e-05, "loss": 0.4462, "step": 2375 }, { "epoch": 0.41, "learning_rate": 4.844448065093633e-05, "loss": 0.4265, "step": 2376 }, { "epoch": 0.41, "learning_rate": 4.844197790197925e-05, "loss": 0.4478, "step": 2377 }, { "epoch": 0.41, "learning_rate": 4.8439473213451587e-05, "loss": 0.4256, "step": 2378 }, { "epoch": 0.41, "learning_rate": 4.843696658558534e-05, "loss": 0.4249, "step": 2379 }, { "epoch": 0.41, "learning_rate": 4.843445801861266e-05, "loss": 0.4431, "step": 2380 }, { "epoch": 0.42, "learning_rate": 4.8431947512765884e-05, "loss": 0.4404, "step": 2381 }, { "epoch": 0.42, "learning_rate": 4.842943506827753e-05, "loss": 0.4303, "step": 2382 }, { "epoch": 0.42, "learning_rate": 4.842692068538027e-05, "loss": 0.4161, "step": 2383 }, { "epoch": 0.42, "learning_rate": 4.8424404364306995e-05, "loss": 0.4274, "step": 2384 }, { "epoch": 0.42, "learning_rate": 4.842188610529076e-05, "loss": 0.4465, "step": 2385 }, { "epoch": 0.42, "learning_rate": 4.8419365908564785e-05, "loss": 0.4618, "step": 2386 }, { "epoch": 0.42, "learning_rate": 4.8416843774362485e-05, "loss": 0.4279, "step": 2387 }, { "epoch": 0.42, "learning_rate": 4.841431970291747e-05, "loss": 0.4299, "step": 2388 }, { "epoch": 0.42, "learning_rate": 4.841179369446347e-05, "loss": 0.4376, "step": 2389 }, { "epoch": 0.42, "learning_rate": 4.840926574923447e-05, "loss": 0.428, "step": 2390 }, { "epoch": 0.42, "learning_rate": 4.840673586746459e-05, "loss": 0.4419, "step": 2391 }, { "epoch": 0.42, "learning_rate": 4.8404204049388144e-05, "loss": 0.4294, "step": 2392 }, { "epoch": 0.42, "learning_rate": 4.84016702952396e-05, "loss": 0.4343, "step": 2393 }, { "epoch": 0.42, "learning_rate": 4.8399134605253635e-05, "loss": 0.4348, "step": 2394 }, { "epoch": 0.42, "learning_rate": 4.839659697966511e-05, "loss": 0.4374, "step": 2395 }, { "epoch": 0.42, "learning_rate": 4.8394057418709034e-05, "loss": 0.4436, "step": 2396 }, { "epoch": 0.42, "learning_rate": 4.8391515922620616e-05, "loss": 0.4308, "step": 2397 }, { "epoch": 0.42, "learning_rate": 4.838897249163524e-05, "loss": 0.4433, "step": 2398 }, { "epoch": 0.42, "learning_rate": 4.8386427125988475e-05, "loss": 0.428, "step": 2399 }, { "epoch": 0.42, "learning_rate": 4.838387982591605e-05, "loss": 0.4223, "step": 2400 }, { "epoch": 0.42, "learning_rate": 4.838133059165391e-05, "loss": 0.4156, "step": 2401 }, { "epoch": 0.42, "learning_rate": 4.8378779423438134e-05, "loss": 0.4362, "step": 2402 }, { "epoch": 0.42, "learning_rate": 4.8376226321505006e-05, "loss": 0.4246, "step": 2403 }, { "epoch": 0.42, "learning_rate": 4.837367128609099e-05, "loss": 0.4305, "step": 2404 }, { "epoch": 0.42, "learning_rate": 4.8371114317432726e-05, "loss": 0.4271, "step": 2405 }, { "epoch": 0.42, "learning_rate": 4.836855541576703e-05, "loss": 0.4314, "step": 2406 }, { "epoch": 0.42, "learning_rate": 4.836599458133089e-05, "loss": 0.458, "step": 2407 }, { "epoch": 0.42, "learning_rate": 4.8363431814361494e-05, "loss": 0.4312, "step": 2408 }, { "epoch": 0.42, "learning_rate": 4.836086711509619e-05, "loss": 0.4556, "step": 2409 }, { "epoch": 0.42, "learning_rate": 4.835830048377251e-05, "loss": 0.4283, "step": 2410 }, { "epoch": 0.42, "learning_rate": 4.8355731920628164e-05, "loss": 0.4247, "step": 2411 }, { "epoch": 0.42, "learning_rate": 4.8353161425901044e-05, "loss": 0.4275, "step": 2412 }, { "epoch": 0.42, "learning_rate": 4.8350588999829225e-05, "loss": 0.4261, "step": 2413 }, { "epoch": 0.42, "learning_rate": 4.8348014642650955e-05, "loss": 0.4336, "step": 2414 }, { "epoch": 0.42, "learning_rate": 4.834543835460466e-05, "loss": 0.4326, "step": 2415 }, { "epoch": 0.42, "learning_rate": 4.8342860135928945e-05, "loss": 0.435, "step": 2416 }, { "epoch": 0.42, "learning_rate": 4.83402799868626e-05, "loss": 0.4384, "step": 2417 }, { "epoch": 0.42, "learning_rate": 4.833769790764459e-05, "loss": 0.4521, "step": 2418 }, { "epoch": 0.42, "learning_rate": 4.8335113898514044e-05, "loss": 0.4249, "step": 2419 }, { "epoch": 0.42, "learning_rate": 4.83325279597103e-05, "loss": 0.4324, "step": 2420 }, { "epoch": 0.42, "learning_rate": 4.832994009147284e-05, "loss": 0.4239, "step": 2421 }, { "epoch": 0.42, "learning_rate": 4.8327350294041366e-05, "loss": 0.4342, "step": 2422 }, { "epoch": 0.42, "learning_rate": 4.8324758567655725e-05, "loss": 0.4333, "step": 2423 }, { "epoch": 0.42, "learning_rate": 4.8322164912555944e-05, "loss": 0.4155, "step": 2424 }, { "epoch": 0.42, "learning_rate": 4.831956932898226e-05, "loss": 0.4262, "step": 2425 }, { "epoch": 0.42, "learning_rate": 4.831697181717504e-05, "loss": 0.4303, "step": 2426 }, { "epoch": 0.42, "learning_rate": 4.831437237737487e-05, "loss": 0.4219, "step": 2427 }, { "epoch": 0.42, "learning_rate": 4.8311771009822514e-05, "loss": 0.4272, "step": 2428 }, { "epoch": 0.42, "learning_rate": 4.830916771475887e-05, "loss": 0.4491, "step": 2429 }, { "epoch": 0.42, "learning_rate": 4.830656249242508e-05, "loss": 0.4292, "step": 2430 }, { "epoch": 0.42, "learning_rate": 4.8303955343062405e-05, "loss": 0.4537, "step": 2431 }, { "epoch": 0.42, "learning_rate": 4.830134626691232e-05, "loss": 0.4351, "step": 2432 }, { "epoch": 0.42, "learning_rate": 4.8298735264216466e-05, "loss": 0.4243, "step": 2433 }, { "epoch": 0.42, "learning_rate": 4.8296122335216664e-05, "loss": 0.4352, "step": 2434 }, { "epoch": 0.42, "learning_rate": 4.829350748015492e-05, "loss": 0.4186, "step": 2435 }, { "epoch": 0.42, "learning_rate": 4.8290890699273404e-05, "loss": 0.4378, "step": 2436 }, { "epoch": 0.42, "learning_rate": 4.828827199281448e-05, "loss": 0.4451, "step": 2437 }, { "epoch": 0.43, "learning_rate": 4.828565136102068e-05, "loss": 0.4219, "step": 2438 }, { "epoch": 0.43, "learning_rate": 4.8283028804134715e-05, "loss": 0.4326, "step": 2439 }, { "epoch": 0.43, "learning_rate": 4.828040432239948e-05, "loss": 0.4207, "step": 2440 }, { "epoch": 0.43, "learning_rate": 4.8277777916058046e-05, "loss": 0.4251, "step": 2441 }, { "epoch": 0.43, "learning_rate": 4.827514958535366e-05, "loss": 0.4424, "step": 2442 }, { "epoch": 0.43, "learning_rate": 4.827251933052974e-05, "loss": 0.4336, "step": 2443 }, { "epoch": 0.43, "learning_rate": 4.8269887151829905e-05, "loss": 0.444, "step": 2444 }, { "epoch": 0.43, "learning_rate": 4.826725304949793e-05, "loss": 0.439, "step": 2445 }, { "epoch": 0.43, "learning_rate": 4.826461702377778e-05, "loss": 0.4365, "step": 2446 }, { "epoch": 0.43, "learning_rate": 4.8261979074913595e-05, "loss": 0.428, "step": 2447 }, { "epoch": 0.43, "learning_rate": 4.825933920314968e-05, "loss": 0.4421, "step": 2448 }, { "epoch": 0.43, "learning_rate": 4.825669740873055e-05, "loss": 0.4397, "step": 2449 }, { "epoch": 0.43, "learning_rate": 4.8254053691900856e-05, "loss": 0.4346, "step": 2450 }, { "epoch": 0.43, "learning_rate": 4.825140805290547e-05, "loss": 0.4323, "step": 2451 }, { "epoch": 0.43, "learning_rate": 4.824876049198942e-05, "loss": 0.4316, "step": 2452 }, { "epoch": 0.43, "learning_rate": 4.82461110093979e-05, "loss": 0.4276, "step": 2453 }, { "epoch": 0.43, "learning_rate": 4.82434596053763e-05, "loss": 0.4626, "step": 2454 }, { "epoch": 0.43, "learning_rate": 4.824080628017018e-05, "loss": 0.4292, "step": 2455 }, { "epoch": 0.43, "learning_rate": 4.8238151034025294e-05, "loss": 0.4297, "step": 2456 }, { "epoch": 0.43, "learning_rate": 4.823549386718755e-05, "loss": 0.4264, "step": 2457 }, { "epoch": 0.43, "learning_rate": 4.823283477990306e-05, "loss": 0.437, "step": 2458 }, { "epoch": 0.43, "learning_rate": 4.823017377241808e-05, "loss": 0.4342, "step": 2459 }, { "epoch": 0.43, "learning_rate": 4.822751084497908e-05, "loss": 0.4265, "step": 2460 }, { "epoch": 0.43, "learning_rate": 4.822484599783268e-05, "loss": 0.4353, "step": 2461 }, { "epoch": 0.43, "learning_rate": 4.8222179231225686e-05, "loss": 0.4445, "step": 2462 }, { "epoch": 0.43, "learning_rate": 4.82195105454051e-05, "loss": 0.4233, "step": 2463 }, { "epoch": 0.43, "learning_rate": 4.821683994061806e-05, "loss": 0.4328, "step": 2464 }, { "epoch": 0.43, "learning_rate": 4.8214167417111935e-05, "loss": 0.4297, "step": 2465 }, { "epoch": 0.43, "learning_rate": 4.8211492975134235e-05, "loss": 0.4507, "step": 2466 }, { "epoch": 0.43, "learning_rate": 4.820881661493265e-05, "loss": 0.4503, "step": 2467 }, { "epoch": 0.43, "learning_rate": 4.8206138336755055e-05, "loss": 0.4305, "step": 2468 }, { "epoch": 0.43, "learning_rate": 4.820345814084951e-05, "loss": 0.4257, "step": 2469 }, { "epoch": 0.43, "learning_rate": 4.820077602746425e-05, "loss": 0.4319, "step": 2470 }, { "epoch": 0.43, "learning_rate": 4.819809199684767e-05, "loss": 0.4023, "step": 2471 }, { "epoch": 0.43, "learning_rate": 4.819540604924835e-05, "loss": 0.4601, "step": 2472 }, { "epoch": 0.43, "learning_rate": 4.819271818491508e-05, "loss": 0.4419, "step": 2473 }, { "epoch": 0.43, "learning_rate": 4.8190028404096774e-05, "loss": 0.4314, "step": 2474 }, { "epoch": 0.43, "learning_rate": 4.818733670704256e-05, "loss": 0.4324, "step": 2475 }, { "epoch": 0.43, "learning_rate": 4.8184643094001735e-05, "loss": 0.4505, "step": 2476 }, { "epoch": 0.43, "learning_rate": 4.818194756522377e-05, "loss": 0.4348, "step": 2477 }, { "epoch": 0.43, "learning_rate": 4.8179250120958315e-05, "loss": 0.4186, "step": 2478 }, { "epoch": 0.43, "learning_rate": 4.8176550761455195e-05, "loss": 0.4186, "step": 2479 }, { "epoch": 0.43, "learning_rate": 4.8173849486964415e-05, "loss": 0.4324, "step": 2480 }, { "epoch": 0.43, "learning_rate": 4.8171146297736156e-05, "loss": 0.4393, "step": 2481 }, { "epoch": 0.43, "learning_rate": 4.816844119402079e-05, "loss": 0.4267, "step": 2482 }, { "epoch": 0.43, "learning_rate": 4.816573417606884e-05, "loss": 0.4444, "step": 2483 }, { "epoch": 0.43, "learning_rate": 4.8163025244131024e-05, "loss": 0.4294, "step": 2484 }, { "epoch": 0.43, "learning_rate": 4.8160314398458236e-05, "loss": 0.4598, "step": 2485 }, { "epoch": 0.43, "learning_rate": 4.815760163930155e-05, "loss": 0.4237, "step": 2486 }, { "epoch": 0.43, "learning_rate": 4.8154886966912196e-05, "loss": 0.439, "step": 2487 }, { "epoch": 0.43, "learning_rate": 4.8152170381541614e-05, "loss": 0.4392, "step": 2488 }, { "epoch": 0.43, "learning_rate": 4.8149451883441394e-05, "loss": 0.4349, "step": 2489 }, { "epoch": 0.43, "learning_rate": 4.814673147286331e-05, "loss": 0.4344, "step": 2490 }, { "epoch": 0.43, "learning_rate": 4.814400915005933e-05, "loss": 0.4092, "step": 2491 }, { "epoch": 0.43, "learning_rate": 4.814128491528158e-05, "loss": 0.4313, "step": 2492 }, { "epoch": 0.43, "learning_rate": 4.813855876878237e-05, "loss": 0.4032, "step": 2493 }, { "epoch": 0.43, "learning_rate": 4.8135830710814175e-05, "loss": 0.445, "step": 2494 }, { "epoch": 0.43, "learning_rate": 4.813310074162967e-05, "loss": 0.4238, "step": 2495 }, { "epoch": 0.44, "learning_rate": 4.813036886148169e-05, "loss": 0.4233, "step": 2496 }, { "epoch": 0.44, "learning_rate": 4.812763507062325e-05, "loss": 0.439, "step": 2497 }, { "epoch": 0.44, "learning_rate": 4.812489936930755e-05, "loss": 0.4208, "step": 2498 }, { "epoch": 0.44, "learning_rate": 4.812216175778796e-05, "loss": 0.4329, "step": 2499 }, { "epoch": 0.44, "learning_rate": 4.811942223631803e-05, "loss": 0.4215, "step": 2500 }, { "epoch": 0.44, "learning_rate": 4.8116680805151476e-05, "loss": 0.4274, "step": 2501 }, { "epoch": 0.44, "learning_rate": 4.8113937464542204e-05, "loss": 0.4289, "step": 2502 }, { "epoch": 0.44, "learning_rate": 4.811119221474429e-05, "loss": 0.4334, "step": 2503 }, { "epoch": 0.44, "learning_rate": 4.810844505601199e-05, "loss": 0.4353, "step": 2504 }, { "epoch": 0.44, "learning_rate": 4.810569598859974e-05, "loss": 0.4395, "step": 2505 }, { "epoch": 0.44, "learning_rate": 4.810294501276215e-05, "loss": 0.4295, "step": 2506 }, { "epoch": 0.44, "learning_rate": 4.810019212875399e-05, "loss": 0.4145, "step": 2507 }, { "epoch": 0.44, "learning_rate": 4.809743733683025e-05, "loss": 0.4427, "step": 2508 }, { "epoch": 0.44, "learning_rate": 4.8094680637246045e-05, "loss": 0.4263, "step": 2509 }, { "epoch": 0.44, "learning_rate": 4.8091922030256694e-05, "loss": 0.4194, "step": 2510 }, { "epoch": 0.44, "learning_rate": 4.808916151611769e-05, "loss": 0.4415, "step": 2511 }, { "epoch": 0.44, "learning_rate": 4.8086399095084705e-05, "loss": 0.425, "step": 2512 }, { "epoch": 0.44, "learning_rate": 4.808363476741359e-05, "loss": 0.4458, "step": 2513 }, { "epoch": 0.44, "learning_rate": 4.8080868533360365e-05, "loss": 0.4208, "step": 2514 }, { "epoch": 0.44, "learning_rate": 4.807810039318122e-05, "loss": 0.4503, "step": 2515 }, { "epoch": 0.44, "learning_rate": 4.807533034713254e-05, "loss": 0.4524, "step": 2516 }, { "epoch": 0.44, "learning_rate": 4.807255839547087e-05, "loss": 0.4266, "step": 2517 }, { "epoch": 0.44, "learning_rate": 4.806978453845293e-05, "loss": 0.4457, "step": 2518 }, { "epoch": 0.44, "learning_rate": 4.8067008776335655e-05, "loss": 0.4309, "step": 2519 }, { "epoch": 0.44, "learning_rate": 4.8064231109376095e-05, "loss": 0.4627, "step": 2520 }, { "epoch": 0.44, "learning_rate": 4.806145153783151e-05, "loss": 0.434, "step": 2521 }, { "epoch": 0.44, "learning_rate": 4.805867006195936e-05, "loss": 0.4533, "step": 2522 }, { "epoch": 0.44, "learning_rate": 4.805588668201723e-05, "loss": 0.4031, "step": 2523 }, { "epoch": 0.44, "learning_rate": 4.805310139826291e-05, "loss": 0.4396, "step": 2524 }, { "epoch": 0.44, "learning_rate": 4.805031421095437e-05, "loss": 0.4199, "step": 2525 }, { "epoch": 0.44, "learning_rate": 4.8047525120349753e-05, "loss": 0.4196, "step": 2526 }, { "epoch": 0.44, "learning_rate": 4.8044734126707364e-05, "loss": 0.4125, "step": 2527 }, { "epoch": 0.44, "learning_rate": 4.804194123028569e-05, "loss": 0.4248, "step": 2528 }, { "epoch": 0.44, "learning_rate": 4.803914643134342e-05, "loss": 0.4457, "step": 2529 }, { "epoch": 0.44, "learning_rate": 4.8036349730139385e-05, "loss": 0.4451, "step": 2530 }, { "epoch": 0.44, "learning_rate": 4.80335511269326e-05, "loss": 0.439, "step": 2531 }, { "epoch": 0.44, "learning_rate": 4.803075062198228e-05, "loss": 0.4225, "step": 2532 }, { "epoch": 0.44, "learning_rate": 4.802794821554777e-05, "loss": 0.4461, "step": 2533 }, { "epoch": 0.44, "learning_rate": 4.8025143907888646e-05, "loss": 0.4116, "step": 2534 }, { "epoch": 0.44, "learning_rate": 4.802233769926461e-05, "loss": 0.4436, "step": 2535 }, { "epoch": 0.44, "learning_rate": 4.8019529589935594e-05, "loss": 0.4289, "step": 2536 }, { "epoch": 0.44, "learning_rate": 4.801671958016163e-05, "loss": 0.4319, "step": 2537 }, { "epoch": 0.44, "learning_rate": 4.801390767020301e-05, "loss": 0.4133, "step": 2538 }, { "epoch": 0.44, "learning_rate": 4.8011093860320156e-05, "loss": 0.4458, "step": 2539 }, { "epoch": 0.44, "learning_rate": 4.800827815077365e-05, "loss": 0.4504, "step": 2540 }, { "epoch": 0.44, "learning_rate": 4.800546054182431e-05, "loss": 0.4266, "step": 2541 }, { "epoch": 0.44, "learning_rate": 4.800264103373306e-05, "loss": 0.4426, "step": 2542 }, { "epoch": 0.44, "learning_rate": 4.799981962676104e-05, "loss": 0.4392, "step": 2543 }, { "epoch": 0.44, "learning_rate": 4.799699632116957e-05, "loss": 0.432, "step": 2544 }, { "epoch": 0.44, "learning_rate": 4.799417111722012e-05, "loss": 0.4351, "step": 2545 }, { "epoch": 0.44, "learning_rate": 4.799134401517437e-05, "loss": 0.4175, "step": 2546 }, { "epoch": 0.44, "learning_rate": 4.798851501529413e-05, "loss": 0.4314, "step": 2547 }, { "epoch": 0.44, "learning_rate": 4.798568411784144e-05, "loss": 0.4285, "step": 2548 }, { "epoch": 0.44, "learning_rate": 4.798285132307846e-05, "loss": 0.4237, "step": 2549 }, { "epoch": 0.44, "learning_rate": 4.798001663126757e-05, "loss": 0.4369, "step": 2550 }, { "epoch": 0.44, "learning_rate": 4.797718004267131e-05, "loss": 0.4513, "step": 2551 }, { "epoch": 0.44, "learning_rate": 4.797434155755238e-05, "loss": 0.4113, "step": 2552 }, { "epoch": 0.45, "learning_rate": 4.797150117617369e-05, "loss": 0.4147, "step": 2553 }, { "epoch": 0.45, "learning_rate": 4.7968658898798294e-05, "loss": 0.4264, "step": 2554 }, { "epoch": 0.45, "learning_rate": 4.796581472568943e-05, "loss": 0.4421, "step": 2555 }, { "epoch": 0.45, "learning_rate": 4.796296865711053e-05, "loss": 0.4483, "step": 2556 }, { "epoch": 0.45, "learning_rate": 4.7960120693325175e-05, "loss": 0.4448, "step": 2557 }, { "epoch": 0.45, "learning_rate": 4.795727083459713e-05, "loss": 0.4355, "step": 2558 }, { "epoch": 0.45, "learning_rate": 4.795441908119035e-05, "loss": 0.4425, "step": 2559 }, { "epoch": 0.45, "learning_rate": 4.795156543336894e-05, "loss": 0.4226, "step": 2560 }, { "epoch": 0.45, "learning_rate": 4.79487098913972e-05, "loss": 0.4203, "step": 2561 }, { "epoch": 0.45, "learning_rate": 4.794585245553961e-05, "loss": 0.4347, "step": 2562 }, { "epoch": 0.45, "learning_rate": 4.794299312606082e-05, "loss": 0.4255, "step": 2563 }, { "epoch": 0.45, "learning_rate": 4.794013190322562e-05, "loss": 0.4187, "step": 2564 }, { "epoch": 0.45, "learning_rate": 4.793726878729903e-05, "loss": 0.438, "step": 2565 }, { "epoch": 0.45, "learning_rate": 4.793440377854623e-05, "loss": 0.4273, "step": 2566 }, { "epoch": 0.45, "learning_rate": 4.793153687723254e-05, "loss": 0.4264, "step": 2567 }, { "epoch": 0.45, "learning_rate": 4.792866808362349e-05, "loss": 0.4412, "step": 2568 }, { "epoch": 0.45, "learning_rate": 4.792579739798479e-05, "loss": 0.4422, "step": 2569 }, { "epoch": 0.45, "learning_rate": 4.792292482058231e-05, "loss": 0.4394, "step": 2570 }, { "epoch": 0.45, "learning_rate": 4.792005035168209e-05, "loss": 0.4372, "step": 2571 }, { "epoch": 0.45, "learning_rate": 4.791717399155035e-05, "loss": 0.4347, "step": 2572 }, { "epoch": 0.45, "learning_rate": 4.7914295740453505e-05, "loss": 0.4319, "step": 2573 }, { "epoch": 0.45, "learning_rate": 4.7911415598658104e-05, "loss": 0.4358, "step": 2574 }, { "epoch": 0.45, "learning_rate": 4.7908533566430916e-05, "loss": 0.4192, "step": 2575 }, { "epoch": 0.45, "learning_rate": 4.790564964403886e-05, "loss": 0.4392, "step": 2576 }, { "epoch": 0.45, "learning_rate": 4.790276383174903e-05, "loss": 0.4296, "step": 2577 }, { "epoch": 0.45, "learning_rate": 4.7899876129828704e-05, "loss": 0.4245, "step": 2578 }, { "epoch": 0.45, "learning_rate": 4.7896986538545324e-05, "loss": 0.4273, "step": 2579 }, { "epoch": 0.45, "learning_rate": 4.789409505816651e-05, "loss": 0.4465, "step": 2580 }, { "epoch": 0.45, "learning_rate": 4.789120168896008e-05, "loss": 0.4187, "step": 2581 }, { "epoch": 0.45, "learning_rate": 4.788830643119399e-05, "loss": 0.4438, "step": 2582 }, { "epoch": 0.45, "learning_rate": 4.7885409285136395e-05, "loss": 0.4355, "step": 2583 }, { "epoch": 0.45, "learning_rate": 4.7882510251055614e-05, "loss": 0.4239, "step": 2584 }, { "epoch": 0.45, "learning_rate": 4.787960932922015e-05, "loss": 0.4319, "step": 2585 }, { "epoch": 0.45, "learning_rate": 4.787670651989868e-05, "loss": 0.436, "step": 2586 }, { "epoch": 0.45, "learning_rate": 4.7873801823360035e-05, "loss": 0.451, "step": 2587 }, { "epoch": 0.45, "learning_rate": 4.7870895239873255e-05, "loss": 0.4179, "step": 2588 }, { "epoch": 0.45, "learning_rate": 4.786798676970753e-05, "loss": 0.4363, "step": 2589 }, { "epoch": 0.45, "learning_rate": 4.7865076413132234e-05, "loss": 0.4274, "step": 2590 }, { "epoch": 0.45, "learning_rate": 4.786216417041691e-05, "loss": 0.4508, "step": 2591 }, { "epoch": 0.45, "learning_rate": 4.785925004183128e-05, "loss": 0.4226, "step": 2592 }, { "epoch": 0.45, "learning_rate": 4.785633402764524e-05, "loss": 0.428, "step": 2593 }, { "epoch": 0.45, "learning_rate": 4.785341612812886e-05, "loss": 0.4294, "step": 2594 }, { "epoch": 0.45, "learning_rate": 4.785049634355239e-05, "loss": 0.4215, "step": 2595 }, { "epoch": 0.45, "learning_rate": 4.7847574674186253e-05, "loss": 0.43, "step": 2596 }, { "epoch": 0.45, "learning_rate": 4.7844651120301046e-05, "loss": 0.4252, "step": 2597 }, { "epoch": 0.45, "learning_rate": 4.784172568216751e-05, "loss": 0.4551, "step": 2598 }, { "epoch": 0.45, "learning_rate": 4.783879836005662e-05, "loss": 0.4424, "step": 2599 }, { "epoch": 0.45, "learning_rate": 4.7835869154239484e-05, "loss": 0.4318, "step": 2600 }, { "epoch": 0.45, "learning_rate": 4.78329380649874e-05, "loss": 0.4398, "step": 2601 }, { "epoch": 0.45, "learning_rate": 4.783000509257181e-05, "loss": 0.4076, "step": 2602 }, { "epoch": 0.45, "learning_rate": 4.782707023726439e-05, "loss": 0.4267, "step": 2603 }, { "epoch": 0.45, "learning_rate": 4.782413349933693e-05, "loss": 0.4215, "step": 2604 }, { "epoch": 0.45, "learning_rate": 4.782119487906143e-05, "loss": 0.4195, "step": 2605 }, { "epoch": 0.45, "learning_rate": 4.7818254376710067e-05, "loss": 0.4266, "step": 2606 }, { "epoch": 0.45, "learning_rate": 4.7815311992555145e-05, "loss": 0.427, "step": 2607 }, { "epoch": 0.45, "learning_rate": 4.7812367726869214e-05, "loss": 0.4297, "step": 2608 }, { "epoch": 0.45, "learning_rate": 4.780942157992494e-05, "loss": 0.4234, "step": 2609 }, { "epoch": 0.46, "learning_rate": 4.780647355199519e-05, "loss": 0.4158, "step": 2610 }, { "epoch": 0.46, "learning_rate": 4.7803523643353e-05, "loss": 0.4254, "step": 2611 }, { "epoch": 0.46, "learning_rate": 4.780057185427159e-05, "loss": 0.4364, "step": 2612 }, { "epoch": 0.46, "learning_rate": 4.779761818502432e-05, "loss": 0.4353, "step": 2613 }, { "epoch": 0.46, "learning_rate": 4.779466263588477e-05, "loss": 0.4153, "step": 2614 }, { "epoch": 0.46, "learning_rate": 4.7791705207126666e-05, "loss": 0.4313, "step": 2615 }, { "epoch": 0.46, "learning_rate": 4.7788745899023906e-05, "loss": 0.4307, "step": 2616 }, { "epoch": 0.46, "learning_rate": 4.7785784711850584e-05, "loss": 0.4305, "step": 2617 }, { "epoch": 0.46, "learning_rate": 4.778282164588094e-05, "loss": 0.4356, "step": 2618 }, { "epoch": 0.46, "learning_rate": 4.777985670138942e-05, "loss": 0.4305, "step": 2619 }, { "epoch": 0.46, "learning_rate": 4.777688987865062e-05, "loss": 0.44, "step": 2620 }, { "epoch": 0.46, "learning_rate": 4.77739211779393e-05, "loss": 0.4274, "step": 2621 }, { "epoch": 0.46, "learning_rate": 4.777095059953043e-05, "loss": 0.425, "step": 2622 }, { "epoch": 0.46, "learning_rate": 4.7767978143699126e-05, "loss": 0.4085, "step": 2623 }, { "epoch": 0.46, "learning_rate": 4.77650038107207e-05, "loss": 0.4246, "step": 2624 }, { "epoch": 0.46, "learning_rate": 4.77620276008706e-05, "loss": 0.4483, "step": 2625 }, { "epoch": 0.46, "learning_rate": 4.775904951442448e-05, "loss": 0.4358, "step": 2626 }, { "epoch": 0.46, "learning_rate": 4.7756069551658176e-05, "loss": 0.4292, "step": 2627 }, { "epoch": 0.46, "learning_rate": 4.775308771284767e-05, "loss": 0.4309, "step": 2628 }, { "epoch": 0.46, "learning_rate": 4.775010399826912e-05, "loss": 0.4033, "step": 2629 }, { "epoch": 0.46, "learning_rate": 4.774711840819888e-05, "loss": 0.4216, "step": 2630 }, { "epoch": 0.46, "learning_rate": 4.774413094291345e-05, "loss": 0.4379, "step": 2631 }, { "epoch": 0.46, "learning_rate": 4.774114160268954e-05, "loss": 0.457, "step": 2632 }, { "epoch": 0.46, "learning_rate": 4.773815038780399e-05, "loss": 0.4298, "step": 2633 }, { "epoch": 0.46, "learning_rate": 4.7735157298533854e-05, "loss": 0.4347, "step": 2634 }, { "epoch": 0.46, "learning_rate": 4.773216233515633e-05, "loss": 0.4188, "step": 2635 }, { "epoch": 0.46, "learning_rate": 4.7729165497948804e-05, "loss": 0.417, "step": 2636 }, { "epoch": 0.46, "learning_rate": 4.7726166787188834e-05, "loss": 0.4305, "step": 2637 }, { "epoch": 0.46, "learning_rate": 4.772316620315414e-05, "loss": 0.4323, "step": 2638 }, { "epoch": 0.46, "learning_rate": 4.7720163746122636e-05, "loss": 0.4309, "step": 2639 }, { "epoch": 0.46, "learning_rate": 4.77171594163724e-05, "loss": 0.4375, "step": 2640 }, { "epoch": 0.46, "learning_rate": 4.7714153214181675e-05, "loss": 0.4268, "step": 2641 }, { "epoch": 0.46, "learning_rate": 4.7711145139828884e-05, "loss": 0.4291, "step": 2642 }, { "epoch": 0.46, "learning_rate": 4.7708135193592627e-05, "loss": 0.4376, "step": 2643 }, { "epoch": 0.46, "learning_rate": 4.7705123375751684e-05, "loss": 0.4423, "step": 2644 }, { "epoch": 0.46, "learning_rate": 4.770210968658498e-05, "loss": 0.4449, "step": 2645 }, { "epoch": 0.46, "learning_rate": 4.769909412637164e-05, "loss": 0.4294, "step": 2646 }, { "epoch": 0.46, "learning_rate": 4.769607669539096e-05, "loss": 0.4272, "step": 2647 }, { "epoch": 0.46, "learning_rate": 4.7693057393922404e-05, "loss": 0.4261, "step": 2648 }, { "epoch": 0.46, "learning_rate": 4.76900362222456e-05, "loss": 0.4259, "step": 2649 }, { "epoch": 0.46, "learning_rate": 4.768701318064037e-05, "loss": 0.4147, "step": 2650 }, { "epoch": 0.46, "learning_rate": 4.768398826938668e-05, "loss": 0.4194, "step": 2651 }, { "epoch": 0.46, "learning_rate": 4.768096148876469e-05, "loss": 0.4209, "step": 2652 }, { "epoch": 0.46, "learning_rate": 4.7677932839054744e-05, "loss": 0.4305, "step": 2653 }, { "epoch": 0.46, "learning_rate": 4.767490232053734e-05, "loss": 0.4324, "step": 2654 }, { "epoch": 0.46, "learning_rate": 4.7671869933493146e-05, "loss": 0.4225, "step": 2655 }, { "epoch": 0.46, "learning_rate": 4.7668835678203014e-05, "loss": 0.4197, "step": 2656 }, { "epoch": 0.46, "learning_rate": 4.7665799554947965e-05, "loss": 0.4294, "step": 2657 }, { "epoch": 0.46, "learning_rate": 4.766276156400919e-05, "loss": 0.4013, "step": 2658 }, { "epoch": 0.46, "learning_rate": 4.765972170566807e-05, "loss": 0.4102, "step": 2659 }, { "epoch": 0.46, "learning_rate": 4.765667998020614e-05, "loss": 0.4158, "step": 2660 }, { "epoch": 0.46, "learning_rate": 4.76536363879051e-05, "loss": 0.4253, "step": 2661 }, { "epoch": 0.46, "learning_rate": 4.765059092904686e-05, "loss": 0.4169, "step": 2662 }, { "epoch": 0.46, "learning_rate": 4.7647543603913456e-05, "loss": 0.4078, "step": 2663 }, { "epoch": 0.46, "learning_rate": 4.764449441278714e-05, "loss": 0.4305, "step": 2664 }, { "epoch": 0.46, "learning_rate": 4.764144335595031e-05, "loss": 0.4398, "step": 2665 }, { "epoch": 0.46, "learning_rate": 4.7638390433685525e-05, "loss": 0.4249, "step": 2666 }, { "epoch": 0.46, "learning_rate": 4.763533564627557e-05, "loss": 0.4251, "step": 2667 }, { "epoch": 0.47, "learning_rate": 4.763227899400335e-05, "loss": 0.4105, "step": 2668 }, { "epoch": 0.47, "learning_rate": 4.7629220477151965e-05, "loss": 0.4192, "step": 2669 }, { "epoch": 0.47, "learning_rate": 4.762616009600467e-05, "loss": 0.4403, "step": 2670 }, { "epoch": 0.47, "learning_rate": 4.762309785084492e-05, "loss": 0.4106, "step": 2671 }, { "epoch": 0.47, "learning_rate": 4.762003374195634e-05, "loss": 0.4264, "step": 2672 }, { "epoch": 0.47, "learning_rate": 4.761696776962269e-05, "loss": 0.422, "step": 2673 }, { "epoch": 0.47, "learning_rate": 4.761389993412795e-05, "loss": 0.4244, "step": 2674 }, { "epoch": 0.47, "learning_rate": 4.761083023575625e-05, "loss": 0.4291, "step": 2675 }, { "epoch": 0.47, "learning_rate": 4.760775867479188e-05, "loss": 0.4116, "step": 2676 }, { "epoch": 0.47, "learning_rate": 4.760468525151933e-05, "loss": 0.4275, "step": 2677 }, { "epoch": 0.47, "learning_rate": 4.760160996622325e-05, "loss": 0.4297, "step": 2678 }, { "epoch": 0.47, "learning_rate": 4.759853281918845e-05, "loss": 0.4282, "step": 2679 }, { "epoch": 0.47, "learning_rate": 4.7595453810699937e-05, "loss": 0.4407, "step": 2680 }, { "epoch": 0.47, "learning_rate": 4.7592372941042876e-05, "loss": 0.4274, "step": 2681 }, { "epoch": 0.47, "learning_rate": 4.7589290210502596e-05, "loss": 0.4251, "step": 2682 }, { "epoch": 0.47, "learning_rate": 4.758620561936461e-05, "loss": 0.4332, "step": 2683 }, { "epoch": 0.47, "learning_rate": 4.758311916791461e-05, "loss": 0.453, "step": 2684 }, { "epoch": 0.47, "learning_rate": 4.758003085643846e-05, "loss": 0.4269, "step": 2685 }, { "epoch": 0.47, "learning_rate": 4.757694068522217e-05, "loss": 0.4189, "step": 2686 }, { "epoch": 0.47, "learning_rate": 4.757384865455195e-05, "loss": 0.4202, "step": 2687 }, { "epoch": 0.47, "learning_rate": 4.757075476471416e-05, "loss": 0.4153, "step": 2688 }, { "epoch": 0.47, "learning_rate": 4.7567659015995367e-05, "loss": 0.4435, "step": 2689 }, { "epoch": 0.47, "learning_rate": 4.7564561408682266e-05, "loss": 0.4429, "step": 2690 }, { "epoch": 0.47, "learning_rate": 4.7561461943061766e-05, "loss": 0.4324, "step": 2691 }, { "epoch": 0.47, "learning_rate": 4.755836061942091e-05, "loss": 0.4132, "step": 2692 }, { "epoch": 0.47, "learning_rate": 4.755525743804695e-05, "loss": 0.435, "step": 2693 }, { "epoch": 0.47, "learning_rate": 4.7552152399227266e-05, "loss": 0.4217, "step": 2694 }, { "epoch": 0.47, "learning_rate": 4.754904550324947e-05, "loss": 0.4487, "step": 2695 }, { "epoch": 0.47, "learning_rate": 4.754593675040128e-05, "loss": 0.4377, "step": 2696 }, { "epoch": 0.47, "learning_rate": 4.754282614097064e-05, "loss": 0.437, "step": 2697 }, { "epoch": 0.47, "learning_rate": 4.753971367524563e-05, "loss": 0.4326, "step": 2698 }, { "epoch": 0.47, "learning_rate": 4.753659935351452e-05, "loss": 0.4342, "step": 2699 }, { "epoch": 0.47, "learning_rate": 4.753348317606575e-05, "loss": 0.415, "step": 2700 }, { "epoch": 0.47, "learning_rate": 4.753036514318792e-05, "loss": 0.4085, "step": 2701 }, { "epoch": 0.47, "learning_rate": 4.752724525516981e-05, "loss": 0.4503, "step": 2702 }, { "epoch": 0.47, "learning_rate": 4.7524123512300384e-05, "loss": 0.4374, "step": 2703 }, { "epoch": 0.47, "learning_rate": 4.752099991486877e-05, "loss": 0.4228, "step": 2704 }, { "epoch": 0.47, "learning_rate": 4.751787446316424e-05, "loss": 0.4286, "step": 2705 }, { "epoch": 0.47, "learning_rate": 4.751474715747629e-05, "loss": 0.4229, "step": 2706 }, { "epoch": 0.47, "learning_rate": 4.751161799809455e-05, "loss": 0.4206, "step": 2707 }, { "epoch": 0.47, "learning_rate": 4.750848698530883e-05, "loss": 0.4319, "step": 2708 }, { "epoch": 0.47, "learning_rate": 4.750535411940911e-05, "loss": 0.4263, "step": 2709 }, { "epoch": 0.47, "learning_rate": 4.750221940068555e-05, "loss": 0.424, "step": 2710 }, { "epoch": 0.47, "learning_rate": 4.7499082829428456e-05, "loss": 0.4443, "step": 2711 }, { "epoch": 0.47, "learning_rate": 4.7495944405928364e-05, "loss": 0.4284, "step": 2712 }, { "epoch": 0.47, "learning_rate": 4.7492804130475916e-05, "loss": 0.4282, "step": 2713 }, { "epoch": 0.47, "learning_rate": 4.748966200336196e-05, "loss": 0.4425, "step": 2714 }, { "epoch": 0.47, "learning_rate": 4.7486518024877516e-05, "loss": 0.4138, "step": 2715 }, { "epoch": 0.47, "learning_rate": 4.7483372195313746e-05, "loss": 0.4249, "step": 2716 }, { "epoch": 0.47, "learning_rate": 4.748022451496203e-05, "loss": 0.4277, "step": 2717 }, { "epoch": 0.47, "learning_rate": 4.747707498411388e-05, "loss": 0.4151, "step": 2718 }, { "epoch": 0.47, "learning_rate": 4.7473923603061006e-05, "loss": 0.4439, "step": 2719 }, { "epoch": 0.47, "learning_rate": 4.747077037209527e-05, "loss": 0.4234, "step": 2720 }, { "epoch": 0.47, "learning_rate": 4.7467615291508713e-05, "loss": 0.4483, "step": 2721 }, { "epoch": 0.47, "learning_rate": 4.746445836159354e-05, "loss": 0.4279, "step": 2722 }, { "epoch": 0.47, "learning_rate": 4.7461299582642146e-05, "loss": 0.419, "step": 2723 }, { "epoch": 0.47, "learning_rate": 4.745813895494708e-05, "loss": 0.4158, "step": 2724 }, { "epoch": 0.48, "learning_rate": 4.745497647880108e-05, "loss": 0.4358, "step": 2725 }, { "epoch": 0.48, "learning_rate": 4.745181215449702e-05, "loss": 0.4174, "step": 2726 }, { "epoch": 0.48, "learning_rate": 4.744864598232799e-05, "loss": 0.4386, "step": 2727 }, { "epoch": 0.48, "learning_rate": 4.7445477962587224e-05, "loss": 0.4385, "step": 2728 }, { "epoch": 0.48, "learning_rate": 4.744230809556812e-05, "loss": 0.4352, "step": 2729 }, { "epoch": 0.48, "learning_rate": 4.7439136381564277e-05, "loss": 0.4485, "step": 2730 }, { "epoch": 0.48, "learning_rate": 4.743596282086944e-05, "loss": 0.4502, "step": 2731 }, { "epoch": 0.48, "learning_rate": 4.743278741377752e-05, "loss": 0.4255, "step": 2732 }, { "epoch": 0.48, "learning_rate": 4.742961016058263e-05, "loss": 0.4377, "step": 2733 }, { "epoch": 0.48, "learning_rate": 4.742643106157903e-05, "loss": 0.4234, "step": 2734 }, { "epoch": 0.48, "learning_rate": 4.7423250117061167e-05, "loss": 0.4317, "step": 2735 }, { "epoch": 0.48, "learning_rate": 4.742006732732363e-05, "loss": 0.4361, "step": 2736 }, { "epoch": 0.48, "learning_rate": 4.741688269266121e-05, "loss": 0.4178, "step": 2737 }, { "epoch": 0.48, "learning_rate": 4.7413696213368845e-05, "loss": 0.4237, "step": 2738 }, { "epoch": 0.48, "learning_rate": 4.741050788974167e-05, "loss": 0.4606, "step": 2739 }, { "epoch": 0.48, "learning_rate": 4.740731772207497e-05, "loss": 0.4399, "step": 2740 }, { "epoch": 0.48, "learning_rate": 4.74041257106642e-05, "loss": 0.4496, "step": 2741 }, { "epoch": 0.48, "learning_rate": 4.7400931855805e-05, "loss": 0.4118, "step": 2742 }, { "epoch": 0.48, "learning_rate": 4.739773615779317e-05, "loss": 0.4236, "step": 2743 }, { "epoch": 0.48, "learning_rate": 4.739453861692468e-05, "loss": 0.417, "step": 2744 }, { "epoch": 0.48, "learning_rate": 4.739133923349569e-05, "loss": 0.4286, "step": 2745 }, { "epoch": 0.48, "learning_rate": 4.73881380078025e-05, "loss": 0.4298, "step": 2746 }, { "epoch": 0.48, "learning_rate": 4.738493494014161e-05, "loss": 0.4509, "step": 2747 }, { "epoch": 0.48, "learning_rate": 4.7381730030809654e-05, "loss": 0.4344, "step": 2748 }, { "epoch": 0.48, "learning_rate": 4.737852328010348e-05, "loss": 0.4311, "step": 2749 }, { "epoch": 0.48, "learning_rate": 4.737531468832008e-05, "loss": 0.4317, "step": 2750 }, { "epoch": 0.48, "learning_rate": 4.737210425575662e-05, "loss": 0.4345, "step": 2751 }, { "epoch": 0.48, "learning_rate": 4.736889198271044e-05, "loss": 0.4087, "step": 2752 }, { "epoch": 0.48, "learning_rate": 4.7365677869479046e-05, "loss": 0.4465, "step": 2753 }, { "epoch": 0.48, "learning_rate": 4.736246191636013e-05, "loss": 0.4121, "step": 2754 }, { "epoch": 0.48, "learning_rate": 4.735924412365152e-05, "loss": 0.4368, "step": 2755 }, { "epoch": 0.48, "learning_rate": 4.735602449165126e-05, "loss": 0.4204, "step": 2756 }, { "epoch": 0.48, "learning_rate": 4.735280302065753e-05, "loss": 0.4203, "step": 2757 }, { "epoch": 0.48, "learning_rate": 4.734957971096868e-05, "loss": 0.42, "step": 2758 }, { "epoch": 0.48, "learning_rate": 4.7346354562883255e-05, "loss": 0.4219, "step": 2759 }, { "epoch": 0.48, "learning_rate": 4.734312757669996e-05, "loss": 0.421, "step": 2760 }, { "epoch": 0.48, "learning_rate": 4.7339898752717657e-05, "loss": 0.417, "step": 2761 }, { "epoch": 0.48, "learning_rate": 4.733666809123539e-05, "loss": 0.4428, "step": 2762 }, { "epoch": 0.48, "learning_rate": 4.733343559255237e-05, "loss": 0.4283, "step": 2763 }, { "epoch": 0.48, "learning_rate": 4.7330201256967996e-05, "loss": 0.4191, "step": 2764 }, { "epoch": 0.48, "learning_rate": 4.732696508478179e-05, "loss": 0.4261, "step": 2765 }, { "epoch": 0.48, "learning_rate": 4.73237270762935e-05, "loss": 0.4287, "step": 2766 }, { "epoch": 0.48, "learning_rate": 4.7320487231803014e-05, "loss": 0.4237, "step": 2767 }, { "epoch": 0.48, "learning_rate": 4.7317245551610384e-05, "loss": 0.4127, "step": 2768 }, { "epoch": 0.48, "learning_rate": 4.731400203601585e-05, "loss": 0.4226, "step": 2769 }, { "epoch": 0.48, "learning_rate": 4.7310756685319817e-05, "loss": 0.4217, "step": 2770 }, { "epoch": 0.48, "learning_rate": 4.730750949982285e-05, "loss": 0.4219, "step": 2771 }, { "epoch": 0.48, "learning_rate": 4.7304260479825704e-05, "loss": 0.4372, "step": 2772 }, { "epoch": 0.48, "learning_rate": 4.730100962562927e-05, "loss": 0.4206, "step": 2773 }, { "epoch": 0.48, "learning_rate": 4.729775693753466e-05, "loss": 0.4171, "step": 2774 }, { "epoch": 0.48, "learning_rate": 4.7294502415843105e-05, "loss": 0.425, "step": 2775 }, { "epoch": 0.48, "learning_rate": 4.729124606085603e-05, "loss": 0.4194, "step": 2776 }, { "epoch": 0.48, "learning_rate": 4.728798787287503e-05, "loss": 0.4251, "step": 2777 }, { "epoch": 0.48, "learning_rate": 4.728472785220187e-05, "loss": 0.4132, "step": 2778 }, { "epoch": 0.48, "learning_rate": 4.728146599913848e-05, "loss": 0.4096, "step": 2779 }, { "epoch": 0.48, "learning_rate": 4.7278202313986944e-05, "loss": 0.4319, "step": 2780 }, { "epoch": 0.48, "learning_rate": 4.727493679704956e-05, "loss": 0.4251, "step": 2781 }, { "epoch": 0.49, "learning_rate": 4.727166944862876e-05, "loss": 0.4347, "step": 2782 }, { "epoch": 0.49, "learning_rate": 4.726840026902714e-05, "loss": 0.4587, "step": 2783 }, { "epoch": 0.49, "learning_rate": 4.72651292585475e-05, "loss": 0.428, "step": 2784 }, { "epoch": 0.49, "learning_rate": 4.726185641749277e-05, "loss": 0.4187, "step": 2785 }, { "epoch": 0.49, "learning_rate": 4.725858174616608e-05, "loss": 0.4207, "step": 2786 }, { "epoch": 0.49, "learning_rate": 4.7255305244870715e-05, "loss": 0.4265, "step": 2787 }, { "epoch": 0.49, "learning_rate": 4.7252026913910134e-05, "loss": 0.4098, "step": 2788 }, { "epoch": 0.49, "learning_rate": 4.724874675358796e-05, "loss": 0.4336, "step": 2789 }, { "epoch": 0.49, "learning_rate": 4.7245464764207994e-05, "loss": 0.405, "step": 2790 }, { "epoch": 0.49, "learning_rate": 4.724218094607422e-05, "loss": 0.4373, "step": 2791 }, { "epoch": 0.49, "learning_rate": 4.7238895299490724e-05, "loss": 0.4315, "step": 2792 }, { "epoch": 0.49, "learning_rate": 4.723560782476187e-05, "loss": 0.434, "step": 2793 }, { "epoch": 0.49, "learning_rate": 4.7232318522192086e-05, "loss": 0.4266, "step": 2794 }, { "epoch": 0.49, "learning_rate": 4.7229027392086034e-05, "loss": 0.4297, "step": 2795 }, { "epoch": 0.49, "learning_rate": 4.7225734434748534e-05, "loss": 0.4286, "step": 2796 }, { "epoch": 0.49, "learning_rate": 4.722243965048455e-05, "loss": 0.4163, "step": 2797 }, { "epoch": 0.49, "learning_rate": 4.721914303959925e-05, "loss": 0.4222, "step": 2798 }, { "epoch": 0.49, "learning_rate": 4.721584460239794e-05, "loss": 0.4274, "step": 2799 }, { "epoch": 0.49, "learning_rate": 4.721254433918612e-05, "loss": 0.4335, "step": 2800 }, { "epoch": 0.49, "learning_rate": 4.720924225026945e-05, "loss": 0.4145, "step": 2801 }, { "epoch": 0.49, "learning_rate": 4.7205938335953743e-05, "loss": 0.4335, "step": 2802 }, { "epoch": 0.49, "learning_rate": 4.720263259654502e-05, "loss": 0.4504, "step": 2803 }, { "epoch": 0.49, "learning_rate": 4.719932503234941e-05, "loss": 0.4345, "step": 2804 }, { "epoch": 0.49, "learning_rate": 4.719601564367329e-05, "loss": 0.4433, "step": 2805 }, { "epoch": 0.49, "learning_rate": 4.7192704430823126e-05, "loss": 0.4219, "step": 2806 }, { "epoch": 0.49, "learning_rate": 4.7189391394105614e-05, "loss": 0.4203, "step": 2807 }, { "epoch": 0.49, "learning_rate": 4.718607653382759e-05, "loss": 0.4361, "step": 2808 }, { "epoch": 0.49, "learning_rate": 4.718275985029606e-05, "loss": 0.445, "step": 2809 }, { "epoch": 0.49, "learning_rate": 4.717944134381821e-05, "loss": 0.4258, "step": 2810 }, { "epoch": 0.49, "learning_rate": 4.717612101470138e-05, "loss": 0.4198, "step": 2811 }, { "epoch": 0.49, "learning_rate": 4.71727988632531e-05, "loss": 0.4181, "step": 2812 }, { "epoch": 0.49, "learning_rate": 4.716947488978104e-05, "loss": 0.4325, "step": 2813 }, { "epoch": 0.49, "learning_rate": 4.7166149094593065e-05, "loss": 0.4238, "step": 2814 }, { "epoch": 0.49, "learning_rate": 4.71628214779972e-05, "loss": 0.4068, "step": 2815 }, { "epoch": 0.49, "learning_rate": 4.715949204030163e-05, "loss": 0.4189, "step": 2816 }, { "epoch": 0.49, "learning_rate": 4.715616078181471e-05, "loss": 0.4156, "step": 2817 }, { "epoch": 0.49, "learning_rate": 4.715282770284498e-05, "loss": 0.4379, "step": 2818 }, { "epoch": 0.49, "learning_rate": 4.714949280370113e-05, "loss": 0.4143, "step": 2819 }, { "epoch": 0.49, "learning_rate": 4.714615608469204e-05, "loss": 0.4315, "step": 2820 }, { "epoch": 0.49, "learning_rate": 4.7142817546126734e-05, "loss": 0.4269, "step": 2821 }, { "epoch": 0.49, "learning_rate": 4.713947718831442e-05, "loss": 0.4112, "step": 2822 }, { "epoch": 0.49, "learning_rate": 4.7136135011564466e-05, "loss": 0.4325, "step": 2823 }, { "epoch": 0.49, "learning_rate": 4.71327910161864e-05, "loss": 0.4165, "step": 2824 }, { "epoch": 0.49, "learning_rate": 4.7129445202489954e-05, "loss": 0.4132, "step": 2825 }, { "epoch": 0.49, "learning_rate": 4.7126097570785e-05, "loss": 0.4371, "step": 2826 }, { "epoch": 0.49, "learning_rate": 4.712274812138157e-05, "loss": 0.4324, "step": 2827 }, { "epoch": 0.49, "learning_rate": 4.7119396854589895e-05, "loss": 0.4298, "step": 2828 }, { "epoch": 0.49, "learning_rate": 4.711604377072034e-05, "loss": 0.4144, "step": 2829 }, { "epoch": 0.49, "learning_rate": 4.711268887008347e-05, "loss": 0.4143, "step": 2830 }, { "epoch": 0.49, "learning_rate": 4.710933215299e-05, "loss": 0.4219, "step": 2831 }, { "epoch": 0.49, "learning_rate": 4.710597361975082e-05, "loss": 0.4261, "step": 2832 }, { "epoch": 0.49, "learning_rate": 4.710261327067696e-05, "loss": 0.4146, "step": 2833 }, { "epoch": 0.49, "learning_rate": 4.7099251106079686e-05, "loss": 0.4337, "step": 2834 }, { "epoch": 0.49, "learning_rate": 4.709588712627036e-05, "loss": 0.4127, "step": 2835 }, { "epoch": 0.49, "learning_rate": 4.7092521331560545e-05, "loss": 0.4269, "step": 2836 }, { "epoch": 0.49, "learning_rate": 4.708915372226199e-05, "loss": 0.4408, "step": 2837 }, { "epoch": 0.49, "learning_rate": 4.708578429868656e-05, "loss": 0.4272, "step": 2838 }, { "epoch": 0.49, "learning_rate": 4.708241306114633e-05, "loss": 0.4149, "step": 2839 }, { "epoch": 0.5, "learning_rate": 4.707904000995354e-05, "loss": 0.429, "step": 2840 }, { "epoch": 0.5, "learning_rate": 4.7075665145420586e-05, "loss": 0.4042, "step": 2841 }, { "epoch": 0.5, "learning_rate": 4.707228846786003e-05, "loss": 0.4342, "step": 2842 }, { "epoch": 0.5, "learning_rate": 4.706890997758462e-05, "loss": 0.4269, "step": 2843 }, { "epoch": 0.5, "learning_rate": 4.706552967490725e-05, "loss": 0.4281, "step": 2844 }, { "epoch": 0.5, "learning_rate": 4.7062147560140984e-05, "loss": 0.4159, "step": 2845 }, { "epoch": 0.5, "learning_rate": 4.7058763633599086e-05, "loss": 0.4329, "step": 2846 }, { "epoch": 0.5, "learning_rate": 4.705537789559494e-05, "loss": 0.4177, "step": 2847 }, { "epoch": 0.5, "learning_rate": 4.705199034644212e-05, "loss": 0.4028, "step": 2848 }, { "epoch": 0.5, "learning_rate": 4.704860098645439e-05, "loss": 0.4108, "step": 2849 }, { "epoch": 0.5, "learning_rate": 4.704520981594564e-05, "loss": 0.4155, "step": 2850 }, { "epoch": 0.5, "learning_rate": 4.7041816835229954e-05, "loss": 0.4376, "step": 2851 }, { "epoch": 0.5, "learning_rate": 4.703842204462159e-05, "loss": 0.432, "step": 2852 }, { "epoch": 0.5, "learning_rate": 4.703502544443494e-05, "loss": 0.4444, "step": 2853 }, { "epoch": 0.5, "learning_rate": 4.70316270349846e-05, "loss": 0.4311, "step": 2854 }, { "epoch": 0.5, "learning_rate": 4.702822681658531e-05, "loss": 0.4149, "step": 2855 }, { "epoch": 0.5, "learning_rate": 4.7024824789551994e-05, "loss": 0.4165, "step": 2856 }, { "epoch": 0.5, "learning_rate": 4.702142095419973e-05, "loss": 0.4377, "step": 2857 }, { "epoch": 0.5, "learning_rate": 4.701801531084377e-05, "loss": 0.4212, "step": 2858 }, { "epoch": 0.5, "learning_rate": 4.701460785979953e-05, "loss": 0.4222, "step": 2859 }, { "epoch": 0.5, "learning_rate": 4.701119860138261e-05, "loss": 0.4289, "step": 2860 }, { "epoch": 0.5, "learning_rate": 4.700778753590874e-05, "loss": 0.4525, "step": 2861 }, { "epoch": 0.5, "learning_rate": 4.7004374663693854e-05, "loss": 0.4253, "step": 2862 }, { "epoch": 0.5, "learning_rate": 4.700095998505403e-05, "loss": 0.4214, "step": 2863 }, { "epoch": 0.5, "learning_rate": 4.699754350030555e-05, "loss": 0.4204, "step": 2864 }, { "epoch": 0.5, "learning_rate": 4.69941252097648e-05, "loss": 0.4218, "step": 2865 }, { "epoch": 0.5, "learning_rate": 4.69907051137484e-05, "loss": 0.4223, "step": 2866 }, { "epoch": 0.5, "learning_rate": 4.698728321257308e-05, "loss": 0.4212, "step": 2867 }, { "epoch": 0.5, "learning_rate": 4.6983859506555786e-05, "loss": 0.4307, "step": 2868 }, { "epoch": 0.5, "learning_rate": 4.698043399601361e-05, "loss": 0.4274, "step": 2869 }, { "epoch": 0.5, "learning_rate": 4.69770066812638e-05, "loss": 0.4248, "step": 2870 }, { "epoch": 0.5, "learning_rate": 4.6973577562623784e-05, "loss": 0.4293, "step": 2871 }, { "epoch": 0.5, "learning_rate": 4.697014664041115e-05, "loss": 0.4227, "step": 2872 }, { "epoch": 0.5, "learning_rate": 4.6966713914943666e-05, "loss": 0.4405, "step": 2873 }, { "epoch": 0.5, "learning_rate": 4.696327938653926e-05, "loss": 0.4178, "step": 2874 }, { "epoch": 0.5, "learning_rate": 4.6959843055516005e-05, "loss": 0.4235, "step": 2875 }, { "epoch": 0.5, "learning_rate": 4.6956404922192194e-05, "loss": 0.4258, "step": 2876 }, { "epoch": 0.5, "learning_rate": 4.695296498688623e-05, "loss": 0.4264, "step": 2877 }, { "epoch": 0.5, "learning_rate": 4.694952324991672e-05, "loss": 0.4371, "step": 2878 }, { "epoch": 0.5, "learning_rate": 4.694607971160241e-05, "loss": 0.418, "step": 2879 }, { "epoch": 0.5, "learning_rate": 4.6942634372262246e-05, "loss": 0.4255, "step": 2880 }, { "epoch": 0.5, "learning_rate": 4.693918723221531e-05, "loss": 0.43, "step": 2881 }, { "epoch": 0.5, "learning_rate": 4.693573829178088e-05, "loss": 0.4093, "step": 2882 }, { "epoch": 0.5, "learning_rate": 4.6932287551278365e-05, "loss": 0.4258, "step": 2883 }, { "epoch": 0.5, "learning_rate": 4.692883501102737e-05, "loss": 0.4261, "step": 2884 }, { "epoch": 0.5, "learning_rate": 4.6925380671347657e-05, "loss": 0.4354, "step": 2885 }, { "epoch": 0.5, "learning_rate": 4.692192453255915e-05, "loss": 0.4179, "step": 2886 }, { "epoch": 0.5, "learning_rate": 4.691846659498196e-05, "loss": 0.4222, "step": 2887 }, { "epoch": 0.5, "learning_rate": 4.6915006858936325e-05, "loss": 0.4231, "step": 2888 }, { "epoch": 0.5, "learning_rate": 4.6911545324742684e-05, "loss": 0.4171, "step": 2889 }, { "epoch": 0.5, "learning_rate": 4.690808199272164e-05, "loss": 0.4226, "step": 2890 }, { "epoch": 0.5, "learning_rate": 4.690461686319394e-05, "loss": 0.4127, "step": 2891 }, { "epoch": 0.5, "learning_rate": 4.690114993648052e-05, "loss": 0.426, "step": 2892 }, { "epoch": 0.5, "learning_rate": 4.689768121290247e-05, "loss": 0.433, "step": 2893 }, { "epoch": 0.5, "learning_rate": 4.6894210692781054e-05, "loss": 0.4204, "step": 2894 }, { "epoch": 0.5, "learning_rate": 4.6890738376437704e-05, "loss": 0.4217, "step": 2895 }, { "epoch": 0.5, "learning_rate": 4.6887264264194e-05, "loss": 0.439, "step": 2896 }, { "epoch": 0.51, "learning_rate": 4.688378835637172e-05, "loss": 0.4267, "step": 2897 }, { "epoch": 0.51, "learning_rate": 4.688031065329276e-05, "loss": 0.4219, "step": 2898 }, { "epoch": 0.51, "learning_rate": 4.687683115527925e-05, "loss": 0.4288, "step": 2899 }, { "epoch": 0.51, "learning_rate": 4.6873349862653425e-05, "loss": 0.4318, "step": 2900 }, { "epoch": 0.51, "learning_rate": 4.686986677573772e-05, "loss": 0.4312, "step": 2901 }, { "epoch": 0.51, "learning_rate": 4.686638189485472e-05, "loss": 0.4314, "step": 2902 }, { "epoch": 0.51, "learning_rate": 4.6862895220327176e-05, "loss": 0.4199, "step": 2903 }, { "epoch": 0.51, "learning_rate": 4.685940675247803e-05, "loss": 0.417, "step": 2904 }, { "epoch": 0.51, "learning_rate": 4.685591649163035e-05, "loss": 0.4077, "step": 2905 }, { "epoch": 0.51, "learning_rate": 4.68524244381074e-05, "loss": 0.4354, "step": 2906 }, { "epoch": 0.51, "learning_rate": 4.684893059223261e-05, "loss": 0.4145, "step": 2907 }, { "epoch": 0.51, "learning_rate": 4.6845434954329555e-05, "loss": 0.4335, "step": 2908 }, { "epoch": 0.51, "learning_rate": 4.6841937524721994e-05, "loss": 0.4109, "step": 2909 }, { "epoch": 0.51, "learning_rate": 4.683843830373385e-05, "loss": 0.4313, "step": 2910 }, { "epoch": 0.51, "learning_rate": 4.68349372916892e-05, "loss": 0.4348, "step": 2911 }, { "epoch": 0.51, "learning_rate": 4.68314344889123e-05, "loss": 0.4467, "step": 2912 }, { "epoch": 0.51, "learning_rate": 4.6827929895727576e-05, "loss": 0.4123, "step": 2913 }, { "epoch": 0.51, "learning_rate": 4.682442351245959e-05, "loss": 0.4312, "step": 2914 }, { "epoch": 0.51, "learning_rate": 4.6820915339433104e-05, "loss": 0.4363, "step": 2915 }, { "epoch": 0.51, "learning_rate": 4.6817405376973033e-05, "loss": 0.4357, "step": 2916 }, { "epoch": 0.51, "learning_rate": 4.681389362540445e-05, "loss": 0.4238, "step": 2917 }, { "epoch": 0.51, "learning_rate": 4.6810380085052605e-05, "loss": 0.4302, "step": 2918 }, { "epoch": 0.51, "learning_rate": 4.680686475624291e-05, "loss": 0.4243, "step": 2919 }, { "epoch": 0.51, "learning_rate": 4.6803347639300944e-05, "loss": 0.4268, "step": 2920 }, { "epoch": 0.51, "learning_rate": 4.679982873455244e-05, "loss": 0.4219, "step": 2921 }, { "epoch": 0.51, "learning_rate": 4.679630804232332e-05, "loss": 0.4328, "step": 2922 }, { "epoch": 0.51, "learning_rate": 4.6792785562939644e-05, "loss": 0.4364, "step": 2923 }, { "epoch": 0.51, "learning_rate": 4.678926129672767e-05, "loss": 0.4317, "step": 2924 }, { "epoch": 0.51, "learning_rate": 4.6785735244013784e-05, "loss": 0.4281, "step": 2925 }, { "epoch": 0.51, "learning_rate": 4.678220740512457e-05, "loss": 0.4329, "step": 2926 }, { "epoch": 0.51, "learning_rate": 4.677867778038675e-05, "loss": 0.4146, "step": 2927 }, { "epoch": 0.51, "learning_rate": 4.677514637012724e-05, "loss": 0.434, "step": 2928 }, { "epoch": 0.51, "learning_rate": 4.677161317467309e-05, "loss": 0.4166, "step": 2929 }, { "epoch": 0.51, "learning_rate": 4.6768078194351544e-05, "loss": 0.4236, "step": 2930 }, { "epoch": 0.51, "learning_rate": 4.676454142949e-05, "loss": 0.4365, "step": 2931 }, { "epoch": 0.51, "learning_rate": 4.6761002880416014e-05, "loss": 0.4341, "step": 2932 }, { "epoch": 0.51, "learning_rate": 4.675746254745731e-05, "loss": 0.4371, "step": 2933 }, { "epoch": 0.51, "learning_rate": 4.67539204309418e-05, "loss": 0.4379, "step": 2934 }, { "epoch": 0.51, "learning_rate": 4.675037653119752e-05, "loss": 0.423, "step": 2935 }, { "epoch": 0.51, "learning_rate": 4.67468308485527e-05, "loss": 0.4095, "step": 2936 }, { "epoch": 0.51, "learning_rate": 4.674328338333574e-05, "loss": 0.4217, "step": 2937 }, { "epoch": 0.51, "learning_rate": 4.6739734135875186e-05, "loss": 0.4153, "step": 2938 }, { "epoch": 0.51, "learning_rate": 4.6736183106499745e-05, "loss": 0.4288, "step": 2939 }, { "epoch": 0.51, "learning_rate": 4.673263029553831e-05, "loss": 0.4354, "step": 2940 }, { "epoch": 0.51, "learning_rate": 4.6729075703319935e-05, "loss": 0.4199, "step": 2941 }, { "epoch": 0.51, "learning_rate": 4.672551933017382e-05, "loss": 0.4409, "step": 2942 }, { "epoch": 0.51, "learning_rate": 4.6721961176429364e-05, "loss": 0.4151, "step": 2943 }, { "epoch": 0.51, "learning_rate": 4.671840124241609e-05, "loss": 0.4238, "step": 2944 }, { "epoch": 0.51, "learning_rate": 4.671483952846372e-05, "loss": 0.4273, "step": 2945 }, { "epoch": 0.51, "learning_rate": 4.671127603490212e-05, "loss": 0.4224, "step": 2946 }, { "epoch": 0.51, "learning_rate": 4.670771076206133e-05, "loss": 0.4359, "step": 2947 }, { "epoch": 0.51, "learning_rate": 4.6704143710271546e-05, "loss": 0.4311, "step": 2948 }, { "epoch": 0.51, "learning_rate": 4.6700574879863145e-05, "loss": 0.4354, "step": 2949 }, { "epoch": 0.51, "learning_rate": 4.6697004271166656e-05, "loss": 0.4365, "step": 2950 }, { "epoch": 0.51, "learning_rate": 4.669343188451278e-05, "loss": 0.4187, "step": 2951 }, { "epoch": 0.51, "learning_rate": 4.668985772023237e-05, "loss": 0.4095, "step": 2952 }, { "epoch": 0.51, "learning_rate": 4.668628177865645e-05, "loss": 0.409, "step": 2953 }, { "epoch": 0.52, "learning_rate": 4.668270406011623e-05, "loss": 0.4281, "step": 2954 }, { "epoch": 0.52, "learning_rate": 4.667912456494305e-05, "loss": 0.4207, "step": 2955 }, { "epoch": 0.52, "learning_rate": 4.667554329346843e-05, "loss": 0.4301, "step": 2956 }, { "epoch": 0.52, "learning_rate": 4.667196024602406e-05, "loss": 0.4194, "step": 2957 }, { "epoch": 0.52, "learning_rate": 4.666837542294178e-05, "loss": 0.424, "step": 2958 }, { "epoch": 0.52, "learning_rate": 4.6664788824553616e-05, "loss": 0.4178, "step": 2959 }, { "epoch": 0.52, "learning_rate": 4.666120045119174e-05, "loss": 0.3983, "step": 2960 }, { "epoch": 0.52, "learning_rate": 4.665761030318849e-05, "loss": 0.3998, "step": 2961 }, { "epoch": 0.52, "learning_rate": 4.6654018380876374e-05, "loss": 0.4055, "step": 2962 }, { "epoch": 0.52, "learning_rate": 4.6650424684588076e-05, "loss": 0.4487, "step": 2963 }, { "epoch": 0.52, "learning_rate": 4.664682921465641e-05, "loss": 0.4321, "step": 2964 }, { "epoch": 0.52, "learning_rate": 4.6643231971414395e-05, "loss": 0.4248, "step": 2965 }, { "epoch": 0.52, "learning_rate": 4.663963295519517e-05, "loss": 0.421, "step": 2966 }, { "epoch": 0.52, "learning_rate": 4.663603216633209e-05, "loss": 0.4121, "step": 2967 }, { "epoch": 0.52, "learning_rate": 4.6632429605158634e-05, "loss": 0.4191, "step": 2968 }, { "epoch": 0.52, "learning_rate": 4.6628825272008454e-05, "loss": 0.4332, "step": 2969 }, { "epoch": 0.52, "learning_rate": 4.662521916721538e-05, "loss": 0.4109, "step": 2970 }, { "epoch": 0.52, "learning_rate": 4.66216112911134e-05, "loss": 0.4099, "step": 2971 }, { "epoch": 0.52, "learning_rate": 4.6618001644036646e-05, "loss": 0.4335, "step": 2972 }, { "epoch": 0.52, "learning_rate": 4.6614390226319436e-05, "loss": 0.4258, "step": 2973 }, { "epoch": 0.52, "learning_rate": 4.661077703829626e-05, "loss": 0.4195, "step": 2974 }, { "epoch": 0.52, "learning_rate": 4.6607162080301745e-05, "loss": 0.4167, "step": 2975 }, { "epoch": 0.52, "learning_rate": 4.6603545352670704e-05, "loss": 0.4317, "step": 2976 }, { "epoch": 0.52, "learning_rate": 4.659992685573809e-05, "loss": 0.4271, "step": 2977 }, { "epoch": 0.52, "learning_rate": 4.659630658983905e-05, "loss": 0.4201, "step": 2978 }, { "epoch": 0.52, "learning_rate": 4.659268455530888e-05, "loss": 0.4398, "step": 2979 }, { "epoch": 0.52, "learning_rate": 4.658906075248303e-05, "loss": 0.4176, "step": 2980 }, { "epoch": 0.52, "learning_rate": 4.658543518169714e-05, "loss": 0.4305, "step": 2981 }, { "epoch": 0.52, "learning_rate": 4.6581807843286975e-05, "loss": 0.4195, "step": 2982 }, { "epoch": 0.52, "learning_rate": 4.65781787375885e-05, "loss": 0.4176, "step": 2983 }, { "epoch": 0.52, "learning_rate": 4.6574547864937816e-05, "loss": 0.4214, "step": 2984 }, { "epoch": 0.52, "learning_rate": 4.6570915225671227e-05, "loss": 0.4263, "step": 2985 }, { "epoch": 0.52, "learning_rate": 4.6567280820125146e-05, "loss": 0.4264, "step": 2986 }, { "epoch": 0.52, "learning_rate": 4.6563644648636204e-05, "loss": 0.4248, "step": 2987 }, { "epoch": 0.52, "learning_rate": 4.656000671154116e-05, "loss": 0.4419, "step": 2988 }, { "epoch": 0.52, "learning_rate": 4.655636700917694e-05, "loss": 0.4348, "step": 2989 }, { "epoch": 0.52, "learning_rate": 4.655272554188064e-05, "loss": 0.4337, "step": 2990 }, { "epoch": 0.52, "learning_rate": 4.654908230998953e-05, "loss": 0.3986, "step": 2991 }, { "epoch": 0.52, "learning_rate": 4.654543731384104e-05, "loss": 0.4253, "step": 2992 }, { "epoch": 0.52, "learning_rate": 4.654179055377273e-05, "loss": 0.4334, "step": 2993 }, { "epoch": 0.52, "learning_rate": 4.6538142030122364e-05, "loss": 0.4172, "step": 2994 }, { "epoch": 0.52, "learning_rate": 4.653449174322786e-05, "loss": 0.4226, "step": 2995 }, { "epoch": 0.52, "learning_rate": 4.653083969342729e-05, "loss": 0.4208, "step": 2996 }, { "epoch": 0.52, "learning_rate": 4.65271858810589e-05, "loss": 0.4588, "step": 2997 }, { "epoch": 0.52, "learning_rate": 4.652353030646108e-05, "loss": 0.4059, "step": 2998 }, { "epoch": 0.52, "learning_rate": 4.65198729699724e-05, "loss": 0.4177, "step": 2999 }, { "epoch": 0.52, "learning_rate": 4.65162138719316e-05, "loss": 0.4473, "step": 3000 }, { "epoch": 0.52, "learning_rate": 4.651255301267756e-05, "loss": 0.4257, "step": 3001 }, { "epoch": 0.52, "learning_rate": 4.6508890392549326e-05, "loss": 0.4275, "step": 3002 }, { "epoch": 0.52, "learning_rate": 4.650522601188615e-05, "loss": 0.43, "step": 3003 }, { "epoch": 0.52, "learning_rate": 4.650155987102738e-05, "loss": 0.4298, "step": 3004 }, { "epoch": 0.52, "learning_rate": 4.6497891970312585e-05, "loss": 0.4272, "step": 3005 }, { "epoch": 0.52, "learning_rate": 4.649422231008146e-05, "loss": 0.4349, "step": 3006 }, { "epoch": 0.52, "learning_rate": 4.649055089067389e-05, "loss": 0.4161, "step": 3007 }, { "epoch": 0.52, "learning_rate": 4.648687771242988e-05, "loss": 0.4261, "step": 3008 }, { "epoch": 0.52, "learning_rate": 4.6483202775689654e-05, "loss": 0.4152, "step": 3009 }, { "epoch": 0.52, "learning_rate": 4.647952608079356e-05, "loss": 0.4311, "step": 3010 }, { "epoch": 0.52, "learning_rate": 4.6475847628082115e-05, "loss": 0.4191, "step": 3011 }, { "epoch": 0.53, "learning_rate": 4.647216741789602e-05, "loss": 0.4228, "step": 3012 }, { "epoch": 0.53, "learning_rate": 4.64684854505761e-05, "loss": 0.4264, "step": 3013 }, { "epoch": 0.53, "learning_rate": 4.6464801726463396e-05, "loss": 0.4327, "step": 3014 }, { "epoch": 0.53, "learning_rate": 4.646111624589906e-05, "loss": 0.4249, "step": 3015 }, { "epoch": 0.53, "learning_rate": 4.645742900922442e-05, "loss": 0.4314, "step": 3016 }, { "epoch": 0.53, "learning_rate": 4.645374001678099e-05, "loss": 0.4029, "step": 3017 }, { "epoch": 0.53, "learning_rate": 4.645004926891044e-05, "loss": 0.4113, "step": 3018 }, { "epoch": 0.53, "learning_rate": 4.644635676595457e-05, "loss": 0.4178, "step": 3019 }, { "epoch": 0.53, "learning_rate": 4.644266250825538e-05, "loss": 0.4379, "step": 3020 }, { "epoch": 0.53, "learning_rate": 4.6438966496155025e-05, "loss": 0.4143, "step": 3021 }, { "epoch": 0.53, "learning_rate": 4.64352687299958e-05, "loss": 0.4244, "step": 3022 }, { "epoch": 0.53, "learning_rate": 4.643156921012019e-05, "loss": 0.4547, "step": 3023 }, { "epoch": 0.53, "learning_rate": 4.642786793687082e-05, "loss": 0.4068, "step": 3024 }, { "epoch": 0.53, "learning_rate": 4.6424164910590515e-05, "loss": 0.4425, "step": 3025 }, { "epoch": 0.53, "learning_rate": 4.642046013162221e-05, "loss": 0.4117, "step": 3026 }, { "epoch": 0.53, "learning_rate": 4.641675360030902e-05, "loss": 0.4253, "step": 3027 }, { "epoch": 0.53, "learning_rate": 4.641304531699426e-05, "loss": 0.4324, "step": 3028 }, { "epoch": 0.53, "learning_rate": 4.640933528202137e-05, "loss": 0.4092, "step": 3029 }, { "epoch": 0.53, "learning_rate": 4.6405623495733945e-05, "loss": 0.4359, "step": 3030 }, { "epoch": 0.53, "learning_rate": 4.640190995847576e-05, "loss": 0.4268, "step": 3031 }, { "epoch": 0.53, "learning_rate": 4.639819467059077e-05, "loss": 0.4295, "step": 3032 }, { "epoch": 0.53, "learning_rate": 4.639447763242306e-05, "loss": 0.4159, "step": 3033 }, { "epoch": 0.53, "learning_rate": 4.639075884431688e-05, "loss": 0.4201, "step": 3034 }, { "epoch": 0.53, "learning_rate": 4.6387038306616656e-05, "loss": 0.4252, "step": 3035 }, { "epoch": 0.53, "learning_rate": 4.638331601966697e-05, "loss": 0.4295, "step": 3036 }, { "epoch": 0.53, "learning_rate": 4.637959198381257e-05, "loss": 0.4393, "step": 3037 }, { "epoch": 0.53, "learning_rate": 4.6375866199398364e-05, "loss": 0.4333, "step": 3038 }, { "epoch": 0.53, "learning_rate": 4.637213866676942e-05, "loss": 0.4161, "step": 3039 }, { "epoch": 0.53, "learning_rate": 4.6368409386270964e-05, "loss": 0.4274, "step": 3040 }, { "epoch": 0.53, "learning_rate": 4.63646783582484e-05, "loss": 0.4047, "step": 3041 }, { "epoch": 0.53, "learning_rate": 4.6360945583047264e-05, "loss": 0.4339, "step": 3042 }, { "epoch": 0.53, "learning_rate": 4.635721106101328e-05, "loss": 0.4248, "step": 3043 }, { "epoch": 0.53, "learning_rate": 4.635347479249234e-05, "loss": 0.4345, "step": 3044 }, { "epoch": 0.53, "learning_rate": 4.6349736777830463e-05, "loss": 0.417, "step": 3045 }, { "epoch": 0.53, "learning_rate": 4.634599701737386e-05, "loss": 0.4341, "step": 3046 }, { "epoch": 0.53, "learning_rate": 4.6342255511468894e-05, "loss": 0.4538, "step": 3047 }, { "epoch": 0.53, "learning_rate": 4.6338512260462086e-05, "loss": 0.4118, "step": 3048 }, { "epoch": 0.53, "learning_rate": 4.633476726470013e-05, "loss": 0.443, "step": 3049 }, { "epoch": 0.53, "learning_rate": 4.633102052452987e-05, "loss": 0.4302, "step": 3050 }, { "epoch": 0.53, "learning_rate": 4.6327272040298314e-05, "loss": 0.4457, "step": 3051 }, { "epoch": 0.53, "learning_rate": 4.6323521812352635e-05, "loss": 0.4311, "step": 3052 }, { "epoch": 0.53, "learning_rate": 4.631976984104016e-05, "loss": 0.4129, "step": 3053 }, { "epoch": 0.53, "learning_rate": 4.6316016126708396e-05, "loss": 0.4225, "step": 3054 }, { "epoch": 0.53, "learning_rate": 4.6312260669704985e-05, "loss": 0.436, "step": 3055 }, { "epoch": 0.53, "learning_rate": 4.630850347037776e-05, "loss": 0.4297, "step": 3056 }, { "epoch": 0.53, "learning_rate": 4.630474452907467e-05, "loss": 0.4147, "step": 3057 }, { "epoch": 0.53, "learning_rate": 4.6300983846143884e-05, "loss": 0.4175, "step": 3058 }, { "epoch": 0.53, "learning_rate": 4.629722142193369e-05, "loss": 0.4202, "step": 3059 }, { "epoch": 0.53, "learning_rate": 4.6293457256792555e-05, "loss": 0.4194, "step": 3060 }, { "epoch": 0.53, "learning_rate": 4.6289691351069096e-05, "loss": 0.4387, "step": 3061 }, { "epoch": 0.53, "learning_rate": 4.62859237051121e-05, "loss": 0.4282, "step": 3062 }, { "epoch": 0.53, "learning_rate": 4.6282154319270526e-05, "loss": 0.4221, "step": 3063 }, { "epoch": 0.53, "learning_rate": 4.627838319389346e-05, "loss": 0.4239, "step": 3064 }, { "epoch": 0.53, "learning_rate": 4.627461032933018e-05, "loss": 0.4283, "step": 3065 }, { "epoch": 0.53, "learning_rate": 4.627083572593012e-05, "loss": 0.4212, "step": 3066 }, { "epoch": 0.53, "learning_rate": 4.626705938404285e-05, "loss": 0.4346, "step": 3067 }, { "epoch": 0.53, "learning_rate": 4.626328130401815e-05, "loss": 0.418, "step": 3068 }, { "epoch": 0.54, "learning_rate": 4.625950148620591e-05, "loss": 0.4239, "step": 3069 }, { "epoch": 0.54, "learning_rate": 4.625571993095622e-05, "loss": 0.4089, "step": 3070 }, { "epoch": 0.54, "learning_rate": 4.62519366386193e-05, "loss": 0.42, "step": 3071 }, { "epoch": 0.54, "learning_rate": 4.6248151609545544e-05, "loss": 0.4107, "step": 3072 }, { "epoch": 0.54, "learning_rate": 4.624436484408552e-05, "loss": 0.4343, "step": 3073 }, { "epoch": 0.54, "learning_rate": 4.624057634258993e-05, "loss": 0.4334, "step": 3074 }, { "epoch": 0.54, "learning_rate": 4.623678610540967e-05, "loss": 0.4103, "step": 3075 }, { "epoch": 0.54, "learning_rate": 4.623299413289577e-05, "loss": 0.4297, "step": 3076 }, { "epoch": 0.54, "learning_rate": 4.6229200425399415e-05, "loss": 0.4297, "step": 3077 }, { "epoch": 0.54, "learning_rate": 4.622540498327198e-05, "loss": 0.4229, "step": 3078 }, { "epoch": 0.54, "learning_rate": 4.622160780686499e-05, "loss": 0.4072, "step": 3079 }, { "epoch": 0.54, "learning_rate": 4.621780889653011e-05, "loss": 0.4378, "step": 3080 }, { "epoch": 0.54, "learning_rate": 4.621400825261919e-05, "loss": 0.4127, "step": 3081 }, { "epoch": 0.54, "learning_rate": 4.621020587548422e-05, "loss": 0.4325, "step": 3082 }, { "epoch": 0.54, "learning_rate": 4.6206401765477384e-05, "loss": 0.4162, "step": 3083 }, { "epoch": 0.54, "learning_rate": 4.620259592295099e-05, "loss": 0.424, "step": 3084 }, { "epoch": 0.54, "learning_rate": 4.6198788348257525e-05, "loss": 0.4048, "step": 3085 }, { "epoch": 0.54, "learning_rate": 4.619497904174964e-05, "loss": 0.4211, "step": 3086 }, { "epoch": 0.54, "learning_rate": 4.619116800378013e-05, "loss": 0.431, "step": 3087 }, { "epoch": 0.54, "learning_rate": 4.6187355234701955e-05, "loss": 0.4055, "step": 3088 }, { "epoch": 0.54, "learning_rate": 4.6183540734868253e-05, "loss": 0.4284, "step": 3089 }, { "epoch": 0.54, "learning_rate": 4.61797245046323e-05, "loss": 0.4459, "step": 3090 }, { "epoch": 0.54, "learning_rate": 4.617590654434754e-05, "loss": 0.4204, "step": 3091 }, { "epoch": 0.54, "learning_rate": 4.617208685436759e-05, "loss": 0.4204, "step": 3092 }, { "epoch": 0.54, "learning_rate": 4.616826543504621e-05, "loss": 0.4251, "step": 3093 }, { "epoch": 0.54, "learning_rate": 4.6164442286737325e-05, "loss": 0.4298, "step": 3094 }, { "epoch": 0.54, "learning_rate": 4.616061740979503e-05, "loss": 0.4331, "step": 3095 }, { "epoch": 0.54, "learning_rate": 4.615679080457355e-05, "loss": 0.4086, "step": 3096 }, { "epoch": 0.54, "learning_rate": 4.61529624714273e-05, "loss": 0.4137, "step": 3097 }, { "epoch": 0.54, "learning_rate": 4.614913241071086e-05, "loss": 0.3944, "step": 3098 }, { "epoch": 0.54, "learning_rate": 4.614530062277895e-05, "loss": 0.4054, "step": 3099 }, { "epoch": 0.54, "learning_rate": 4.614146710798645e-05, "loss": 0.4293, "step": 3100 }, { "epoch": 0.54, "learning_rate": 4.613763186668841e-05, "loss": 0.43, "step": 3101 }, { "epoch": 0.54, "learning_rate": 4.613379489924003e-05, "loss": 0.4092, "step": 3102 }, { "epoch": 0.54, "learning_rate": 4.612995620599669e-05, "loss": 0.41, "step": 3103 }, { "epoch": 0.54, "learning_rate": 4.612611578731391e-05, "loss": 0.4061, "step": 3104 }, { "epoch": 0.54, "learning_rate": 4.612227364354737e-05, "loss": 0.3996, "step": 3105 }, { "epoch": 0.54, "learning_rate": 4.6118429775052916e-05, "loss": 0.4306, "step": 3106 }, { "epoch": 0.54, "learning_rate": 4.6114584182186565e-05, "loss": 0.4317, "step": 3107 }, { "epoch": 0.54, "learning_rate": 4.611073686530447e-05, "loss": 0.4197, "step": 3108 }, { "epoch": 0.54, "learning_rate": 4.6106887824762955e-05, "loss": 0.4076, "step": 3109 }, { "epoch": 0.54, "learning_rate": 4.610303706091851e-05, "loss": 0.4228, "step": 3110 }, { "epoch": 0.54, "learning_rate": 4.609918457412778e-05, "loss": 0.426, "step": 3111 }, { "epoch": 0.54, "learning_rate": 4.609533036474757e-05, "loss": 0.4073, "step": 3112 }, { "epoch": 0.54, "learning_rate": 4.609147443313483e-05, "loss": 0.4247, "step": 3113 }, { "epoch": 0.54, "learning_rate": 4.60876167796467e-05, "loss": 0.4044, "step": 3114 }, { "epoch": 0.54, "learning_rate": 4.608375740464045e-05, "loss": 0.4145, "step": 3115 }, { "epoch": 0.54, "learning_rate": 4.607989630847352e-05, "loss": 0.4023, "step": 3116 }, { "epoch": 0.54, "learning_rate": 4.607603349150352e-05, "loss": 0.4316, "step": 3117 }, { "epoch": 0.54, "learning_rate": 4.607216895408821e-05, "loss": 0.4133, "step": 3118 }, { "epoch": 0.54, "learning_rate": 4.6068302696585496e-05, "loss": 0.4289, "step": 3119 }, { "epoch": 0.54, "learning_rate": 4.606443471935347e-05, "loss": 0.4242, "step": 3120 }, { "epoch": 0.54, "learning_rate": 4.606056502275037e-05, "loss": 0.411, "step": 3121 }, { "epoch": 0.54, "learning_rate": 4.6056693607134585e-05, "loss": 0.4007, "step": 3122 }, { "epoch": 0.54, "learning_rate": 4.605282047286468e-05, "loss": 0.4198, "step": 3123 }, { "epoch": 0.54, "learning_rate": 4.6048945620299366e-05, "loss": 0.3983, "step": 3124 }, { "epoch": 0.54, "learning_rate": 4.604506904979752e-05, "loss": 0.4176, "step": 3125 }, { "epoch": 0.55, "learning_rate": 4.6041190761718164e-05, "loss": 0.4204, "step": 3126 }, { "epoch": 0.55, "learning_rate": 4.6037310756420515e-05, "loss": 0.4279, "step": 3127 }, { "epoch": 0.55, "learning_rate": 4.603342903426391e-05, "loss": 0.422, "step": 3128 }, { "epoch": 0.55, "learning_rate": 4.602954559560786e-05, "loss": 0.4085, "step": 3129 }, { "epoch": 0.55, "learning_rate": 4.602566044081204e-05, "loss": 0.4378, "step": 3130 }, { "epoch": 0.55, "learning_rate": 4.602177357023627e-05, "loss": 0.4188, "step": 3131 }, { "epoch": 0.55, "learning_rate": 4.601788498424055e-05, "loss": 0.4143, "step": 3132 }, { "epoch": 0.55, "learning_rate": 4.601399468318502e-05, "loss": 0.419, "step": 3133 }, { "epoch": 0.55, "learning_rate": 4.6010102667429986e-05, "loss": 0.4459, "step": 3134 }, { "epoch": 0.55, "learning_rate": 4.600620893733591e-05, "loss": 0.4268, "step": 3135 }, { "epoch": 0.55, "learning_rate": 4.6002313493263424e-05, "loss": 0.4161, "step": 3136 }, { "epoch": 0.55, "learning_rate": 4.5998416335573305e-05, "loss": 0.4182, "step": 3137 }, { "epoch": 0.55, "learning_rate": 4.599451746462649e-05, "loss": 0.3859, "step": 3138 }, { "epoch": 0.55, "learning_rate": 4.5990616880784076e-05, "loss": 0.4108, "step": 3139 }, { "epoch": 0.55, "learning_rate": 4.598671458440733e-05, "loss": 0.4072, "step": 3140 }, { "epoch": 0.55, "learning_rate": 4.598281057585767e-05, "loss": 0.431, "step": 3141 }, { "epoch": 0.55, "learning_rate": 4.597890485549667e-05, "loss": 0.4239, "step": 3142 }, { "epoch": 0.55, "learning_rate": 4.597499742368605e-05, "loss": 0.4077, "step": 3143 }, { "epoch": 0.55, "learning_rate": 4.5971088280787714e-05, "loss": 0.4338, "step": 3144 }, { "epoch": 0.55, "learning_rate": 4.596717742716372e-05, "loss": 0.4361, "step": 3145 }, { "epoch": 0.55, "learning_rate": 4.5963264863176266e-05, "loss": 0.4377, "step": 3146 }, { "epoch": 0.55, "learning_rate": 4.595935058918772e-05, "loss": 0.4164, "step": 3147 }, { "epoch": 0.55, "learning_rate": 4.59554346055606e-05, "loss": 0.4278, "step": 3148 }, { "epoch": 0.55, "learning_rate": 4.595151691265761e-05, "loss": 0.4093, "step": 3149 }, { "epoch": 0.55, "learning_rate": 4.5947597510841585e-05, "loss": 0.4211, "step": 3150 }, { "epoch": 0.55, "learning_rate": 4.5943676400475524e-05, "loss": 0.4264, "step": 3151 }, { "epoch": 0.55, "learning_rate": 4.593975358192258e-05, "loss": 0.4173, "step": 3152 }, { "epoch": 0.55, "learning_rate": 4.593582905554608e-05, "loss": 0.4435, "step": 3153 }, { "epoch": 0.55, "learning_rate": 4.5931902821709486e-05, "loss": 0.424, "step": 3154 }, { "epoch": 0.55, "learning_rate": 4.5927974880776457e-05, "loss": 0.4318, "step": 3155 }, { "epoch": 0.55, "learning_rate": 4.5924045233110764e-05, "loss": 0.4096, "step": 3156 }, { "epoch": 0.55, "learning_rate": 4.5920113879076356e-05, "loss": 0.4278, "step": 3157 }, { "epoch": 0.55, "learning_rate": 4.591618081903735e-05, "loss": 0.4228, "step": 3158 }, { "epoch": 0.55, "learning_rate": 4.591224605335801e-05, "loss": 0.4294, "step": 3159 }, { "epoch": 0.55, "learning_rate": 4.590830958240276e-05, "loss": 0.4243, "step": 3160 }, { "epoch": 0.55, "learning_rate": 4.5904371406536175e-05, "loss": 0.4252, "step": 3161 }, { "epoch": 0.55, "learning_rate": 4.5900431526123006e-05, "loss": 0.4232, "step": 3162 }, { "epoch": 0.55, "learning_rate": 4.589648994152813e-05, "loss": 0.4267, "step": 3163 }, { "epoch": 0.55, "learning_rate": 4.5892546653116633e-05, "loss": 0.4518, "step": 3164 }, { "epoch": 0.55, "learning_rate": 4.588860166125371e-05, "loss": 0.4384, "step": 3165 }, { "epoch": 0.55, "learning_rate": 4.588465496630473e-05, "loss": 0.4239, "step": 3166 }, { "epoch": 0.55, "learning_rate": 4.588070656863522e-05, "loss": 0.4188, "step": 3167 }, { "epoch": 0.55, "learning_rate": 4.587675646861088e-05, "loss": 0.4237, "step": 3168 }, { "epoch": 0.55, "learning_rate": 4.587280466659754e-05, "loss": 0.4204, "step": 3169 }, { "epoch": 0.55, "learning_rate": 4.5868851162961214e-05, "loss": 0.4318, "step": 3170 }, { "epoch": 0.55, "learning_rate": 4.586489595806805e-05, "loss": 0.4315, "step": 3171 }, { "epoch": 0.55, "learning_rate": 4.586093905228437e-05, "loss": 0.4162, "step": 3172 }, { "epoch": 0.55, "learning_rate": 4.5856980445976654e-05, "loss": 0.4275, "step": 3173 }, { "epoch": 0.55, "learning_rate": 4.585302013951152e-05, "loss": 0.4112, "step": 3174 }, { "epoch": 0.55, "learning_rate": 4.584905813325577e-05, "loss": 0.4167, "step": 3175 }, { "epoch": 0.55, "learning_rate": 4.5845094427576346e-05, "loss": 0.4116, "step": 3176 }, { "epoch": 0.55, "learning_rate": 4.584112902284035e-05, "loss": 0.4296, "step": 3177 }, { "epoch": 0.55, "learning_rate": 4.5837161919415045e-05, "loss": 0.4498, "step": 3178 }, { "epoch": 0.55, "learning_rate": 4.583319311766785e-05, "loss": 0.426, "step": 3179 }, { "epoch": 0.55, "learning_rate": 4.582922261796634e-05, "loss": 0.4104, "step": 3180 }, { "epoch": 0.55, "learning_rate": 4.582525042067826e-05, "loss": 0.4426, "step": 3181 }, { "epoch": 0.55, "learning_rate": 4.582127652617149e-05, "loss": 0.4167, "step": 3182 }, { "epoch": 0.55, "learning_rate": 4.581730093481407e-05, "loss": 0.4241, "step": 3183 }, { "epoch": 0.56, "learning_rate": 4.581332364697421e-05, "loss": 0.396, "step": 3184 }, { "epoch": 0.56, "learning_rate": 4.580934466302028e-05, "loss": 0.4231, "step": 3185 }, { "epoch": 0.56, "learning_rate": 4.5805363983320805e-05, "loss": 0.4413, "step": 3186 }, { "epoch": 0.56, "learning_rate": 4.580138160824444e-05, "loss": 0.4267, "step": 3187 }, { "epoch": 0.56, "learning_rate": 4.5797397538160026e-05, "loss": 0.4314, "step": 3188 }, { "epoch": 0.56, "learning_rate": 4.5793411773436556e-05, "loss": 0.4145, "step": 3189 }, { "epoch": 0.56, "learning_rate": 4.5789424314443185e-05, "loss": 0.4012, "step": 3190 }, { "epoch": 0.56, "learning_rate": 4.5785435161549206e-05, "loss": 0.4023, "step": 3191 }, { "epoch": 0.56, "learning_rate": 4.578144431512408e-05, "loss": 0.4104, "step": 3192 }, { "epoch": 0.56, "learning_rate": 4.5777451775537437e-05, "loss": 0.4308, "step": 3193 }, { "epoch": 0.56, "learning_rate": 4.577345754315904e-05, "loss": 0.4171, "step": 3194 }, { "epoch": 0.56, "learning_rate": 4.576946161835882e-05, "loss": 0.4007, "step": 3195 }, { "epoch": 0.56, "learning_rate": 4.576546400150687e-05, "loss": 0.4509, "step": 3196 }, { "epoch": 0.56, "learning_rate": 4.576146469297343e-05, "loss": 0.4181, "step": 3197 }, { "epoch": 0.56, "learning_rate": 4.5757463693128906e-05, "loss": 0.4225, "step": 3198 }, { "epoch": 0.56, "learning_rate": 4.5753461002343864e-05, "loss": 0.4071, "step": 3199 }, { "epoch": 0.56, "learning_rate": 4.5749456620989e-05, "loss": 0.4084, "step": 3200 }, { "epoch": 0.56, "learning_rate": 4.574545054943519e-05, "loss": 0.4257, "step": 3201 }, { "epoch": 0.56, "learning_rate": 4.574144278805348e-05, "loss": 0.4036, "step": 3202 }, { "epoch": 0.56, "learning_rate": 4.5737433337215035e-05, "loss": 0.4313, "step": 3203 }, { "epoch": 0.56, "learning_rate": 4.573342219729121e-05, "loss": 0.4186, "step": 3204 }, { "epoch": 0.56, "learning_rate": 4.5729409368653486e-05, "loss": 0.4102, "step": 3205 }, { "epoch": 0.56, "learning_rate": 4.5725394851673534e-05, "loss": 0.4129, "step": 3206 }, { "epoch": 0.56, "learning_rate": 4.5721378646723146e-05, "loss": 0.4151, "step": 3207 }, { "epoch": 0.56, "learning_rate": 4.5717360754174305e-05, "loss": 0.4065, "step": 3208 }, { "epoch": 0.56, "learning_rate": 4.5713341174399124e-05, "loss": 0.4061, "step": 3209 }, { "epoch": 0.56, "learning_rate": 4.5709319907769883e-05, "loss": 0.4203, "step": 3210 }, { "epoch": 0.56, "learning_rate": 4.570529695465903e-05, "loss": 0.4183, "step": 3211 }, { "epoch": 0.56, "learning_rate": 4.5701272315439134e-05, "loss": 0.439, "step": 3212 }, { "epoch": 0.56, "learning_rate": 4.5697245990482954e-05, "loss": 0.43, "step": 3213 }, { "epoch": 0.56, "learning_rate": 4.56932179801634e-05, "loss": 0.4438, "step": 3214 }, { "epoch": 0.56, "learning_rate": 4.568918828485352e-05, "loss": 0.403, "step": 3215 }, { "epoch": 0.56, "learning_rate": 4.568515690492653e-05, "loss": 0.4282, "step": 3216 }, { "epoch": 0.56, "learning_rate": 4.5681123840755825e-05, "loss": 0.4137, "step": 3217 }, { "epoch": 0.56, "learning_rate": 4.5677089092714905e-05, "loss": 0.4138, "step": 3218 }, { "epoch": 0.56, "learning_rate": 4.567305266117746e-05, "loss": 0.4069, "step": 3219 }, { "epoch": 0.56, "learning_rate": 4.5669014546517344e-05, "loss": 0.4314, "step": 3220 }, { "epoch": 0.56, "learning_rate": 4.566497474910854e-05, "loss": 0.4259, "step": 3221 }, { "epoch": 0.56, "learning_rate": 4.56609332693252e-05, "loss": 0.4097, "step": 3222 }, { "epoch": 0.56, "learning_rate": 4.5656890107541626e-05, "loss": 0.4242, "step": 3223 }, { "epoch": 0.56, "learning_rate": 4.5652845264132304e-05, "loss": 0.4306, "step": 3224 }, { "epoch": 0.56, "learning_rate": 4.564879873947182e-05, "loss": 0.424, "step": 3225 }, { "epoch": 0.56, "learning_rate": 4.5644750533934974e-05, "loss": 0.4219, "step": 3226 }, { "epoch": 0.56, "learning_rate": 4.564070064789669e-05, "loss": 0.4187, "step": 3227 }, { "epoch": 0.56, "learning_rate": 4.5636649081732056e-05, "loss": 0.4258, "step": 3228 }, { "epoch": 0.56, "learning_rate": 4.56325958358163e-05, "loss": 0.4022, "step": 3229 }, { "epoch": 0.56, "learning_rate": 4.562854091052483e-05, "loss": 0.4163, "step": 3230 }, { "epoch": 0.56, "learning_rate": 4.56244843062332e-05, "loss": 0.4078, "step": 3231 }, { "epoch": 0.56, "learning_rate": 4.562042602331711e-05, "loss": 0.4262, "step": 3232 }, { "epoch": 0.56, "learning_rate": 4.5616366062152424e-05, "loss": 0.411, "step": 3233 }, { "epoch": 0.56, "learning_rate": 4.561230442311518e-05, "loss": 0.4216, "step": 3234 }, { "epoch": 0.56, "learning_rate": 4.5608241106581526e-05, "loss": 0.4259, "step": 3235 }, { "epoch": 0.56, "learning_rate": 4.560417611292781e-05, "loss": 0.4419, "step": 3236 }, { "epoch": 0.56, "learning_rate": 4.5600109442530515e-05, "loss": 0.4315, "step": 3237 }, { "epoch": 0.56, "learning_rate": 4.559604109576627e-05, "loss": 0.4364, "step": 3238 }, { "epoch": 0.56, "learning_rate": 4.5591971073011876e-05, "loss": 0.4205, "step": 3239 }, { "epoch": 0.56, "learning_rate": 4.558789937464429e-05, "loss": 0.4203, "step": 3240 }, { "epoch": 0.57, "learning_rate": 4.558382600104061e-05, "loss": 0.4197, "step": 3241 }, { "epoch": 0.57, "learning_rate": 4.55797509525781e-05, "loss": 0.416, "step": 3242 }, { "epoch": 0.57, "learning_rate": 4.557567422963418e-05, "loss": 0.4103, "step": 3243 }, { "epoch": 0.57, "learning_rate": 4.5571595832586416e-05, "loss": 0.4241, "step": 3244 }, { "epoch": 0.57, "learning_rate": 4.5567515761812544e-05, "loss": 0.3903, "step": 3245 }, { "epoch": 0.57, "learning_rate": 4.556343401769042e-05, "loss": 0.4184, "step": 3246 }, { "epoch": 0.57, "learning_rate": 4.555935060059812e-05, "loss": 0.4118, "step": 3247 }, { "epoch": 0.57, "learning_rate": 4.5555265510913806e-05, "loss": 0.4105, "step": 3248 }, { "epoch": 0.57, "learning_rate": 4.555117874901583e-05, "loss": 0.4293, "step": 3249 }, { "epoch": 0.57, "learning_rate": 4.55470903152827e-05, "loss": 0.4059, "step": 3250 }, { "epoch": 0.57, "learning_rate": 4.5543000210093066e-05, "loss": 0.4176, "step": 3251 }, { "epoch": 0.57, "learning_rate": 4.553890843382574e-05, "loss": 0.4117, "step": 3252 }, { "epoch": 0.57, "learning_rate": 4.553481498685969e-05, "loss": 0.4217, "step": 3253 }, { "epoch": 0.57, "learning_rate": 4.553071986957403e-05, "loss": 0.4333, "step": 3254 }, { "epoch": 0.57, "learning_rate": 4.5526623082348043e-05, "loss": 0.4127, "step": 3255 }, { "epoch": 0.57, "learning_rate": 4.5522524625561156e-05, "loss": 0.435, "step": 3256 }, { "epoch": 0.57, "learning_rate": 4.5518424499592955e-05, "loss": 0.4374, "step": 3257 }, { "epoch": 0.57, "learning_rate": 4.551432270482316e-05, "loss": 0.4158, "step": 3258 }, { "epoch": 0.57, "learning_rate": 4.5510219241631695e-05, "loss": 0.4243, "step": 3259 }, { "epoch": 0.57, "learning_rate": 4.5506114110398594e-05, "loss": 0.4258, "step": 3260 }, { "epoch": 0.57, "learning_rate": 4.5502007311504056e-05, "loss": 0.4068, "step": 3261 }, { "epoch": 0.57, "learning_rate": 4.549789884532844e-05, "loss": 0.4257, "step": 3262 }, { "epoch": 0.57, "learning_rate": 4.549378871225226e-05, "loss": 0.4256, "step": 3263 }, { "epoch": 0.57, "learning_rate": 4.548967691265618e-05, "loss": 0.4278, "step": 3264 }, { "epoch": 0.57, "learning_rate": 4.548556344692102e-05, "loss": 0.4196, "step": 3265 }, { "epoch": 0.57, "learning_rate": 4.548144831542776e-05, "loss": 0.4219, "step": 3266 }, { "epoch": 0.57, "learning_rate": 4.547733151855751e-05, "loss": 0.4011, "step": 3267 }, { "epoch": 0.57, "learning_rate": 4.5473213056691575e-05, "loss": 0.4261, "step": 3268 }, { "epoch": 0.57, "learning_rate": 4.546909293021138e-05, "loss": 0.4433, "step": 3269 }, { "epoch": 0.57, "learning_rate": 4.546497113949853e-05, "loss": 0.4267, "step": 3270 }, { "epoch": 0.57, "learning_rate": 4.5460847684934747e-05, "loss": 0.4091, "step": 3271 }, { "epoch": 0.57, "learning_rate": 4.545672256690194e-05, "loss": 0.4136, "step": 3272 }, { "epoch": 0.57, "learning_rate": 4.5452595785782184e-05, "loss": 0.4201, "step": 3273 }, { "epoch": 0.57, "learning_rate": 4.544846734195766e-05, "loss": 0.4126, "step": 3274 }, { "epoch": 0.57, "learning_rate": 4.544433723581073e-05, "loss": 0.4254, "step": 3275 }, { "epoch": 0.57, "learning_rate": 4.544020546772392e-05, "loss": 0.4334, "step": 3276 }, { "epoch": 0.57, "learning_rate": 4.5436072038079905e-05, "loss": 0.4325, "step": 3277 }, { "epoch": 0.57, "learning_rate": 4.54319369472615e-05, "loss": 0.4259, "step": 3278 }, { "epoch": 0.57, "learning_rate": 4.542780019565168e-05, "loss": 0.4252, "step": 3279 }, { "epoch": 0.57, "learning_rate": 4.542366178363358e-05, "loss": 0.3938, "step": 3280 }, { "epoch": 0.57, "learning_rate": 4.541952171159047e-05, "loss": 0.4123, "step": 3281 }, { "epoch": 0.57, "learning_rate": 4.541537997990583e-05, "loss": 0.4335, "step": 3282 }, { "epoch": 0.57, "learning_rate": 4.5411236588963206e-05, "loss": 0.4183, "step": 3283 }, { "epoch": 0.57, "learning_rate": 4.540709153914636e-05, "loss": 0.407, "step": 3284 }, { "epoch": 0.57, "learning_rate": 4.54029448308392e-05, "loss": 0.4151, "step": 3285 }, { "epoch": 0.57, "learning_rate": 4.539879646442577e-05, "loss": 0.4171, "step": 3286 }, { "epoch": 0.57, "learning_rate": 4.539464644029029e-05, "loss": 0.4264, "step": 3287 }, { "epoch": 0.57, "learning_rate": 4.5390494758817104e-05, "loss": 0.4115, "step": 3288 }, { "epoch": 0.57, "learning_rate": 4.5386341420390734e-05, "loss": 0.406, "step": 3289 }, { "epoch": 0.57, "learning_rate": 4.538218642539583e-05, "loss": 0.4158, "step": 3290 }, { "epoch": 0.57, "learning_rate": 4.537802977421725e-05, "loss": 0.4414, "step": 3291 }, { "epoch": 0.57, "learning_rate": 4.537387146723994e-05, "loss": 0.4086, "step": 3292 }, { "epoch": 0.57, "learning_rate": 4.536971150484902e-05, "loss": 0.4071, "step": 3293 }, { "epoch": 0.57, "learning_rate": 4.536554988742979e-05, "loss": 0.4243, "step": 3294 }, { "epoch": 0.57, "learning_rate": 4.5361386615367685e-05, "loss": 0.4332, "step": 3295 }, { "epoch": 0.57, "learning_rate": 4.535722168904829e-05, "loss": 0.441, "step": 3296 }, { "epoch": 0.57, "learning_rate": 4.5353055108857335e-05, "loss": 0.4241, "step": 3297 }, { "epoch": 0.57, "learning_rate": 4.5348886875180717e-05, "loss": 0.4395, "step": 3298 }, { "epoch": 0.58, "learning_rate": 4.5344716988404484e-05, "loss": 0.4102, "step": 3299 }, { "epoch": 0.58, "learning_rate": 4.534054544891484e-05, "loss": 0.4262, "step": 3300 }, { "epoch": 0.58, "learning_rate": 4.5336372257098134e-05, "loss": 0.423, "step": 3301 }, { "epoch": 0.58, "learning_rate": 4.533219741334087e-05, "loss": 0.4241, "step": 3302 }, { "epoch": 0.58, "learning_rate": 4.5328020918029716e-05, "loss": 0.4322, "step": 3303 }, { "epoch": 0.58, "learning_rate": 4.532384277155148e-05, "loss": 0.4023, "step": 3304 }, { "epoch": 0.58, "learning_rate": 4.531966297429312e-05, "loss": 0.4108, "step": 3305 }, { "epoch": 0.58, "learning_rate": 4.531548152664176e-05, "loss": 0.431, "step": 3306 }, { "epoch": 0.58, "learning_rate": 4.531129842898467e-05, "loss": 0.4294, "step": 3307 }, { "epoch": 0.58, "learning_rate": 4.530711368170927e-05, "loss": 0.4143, "step": 3308 }, { "epoch": 0.58, "learning_rate": 4.5302927285203146e-05, "loss": 0.4364, "step": 3309 }, { "epoch": 0.58, "learning_rate": 4.5298739239854015e-05, "loss": 0.4214, "step": 3310 }, { "epoch": 0.58, "learning_rate": 4.529454954604976e-05, "loss": 0.4165, "step": 3311 }, { "epoch": 0.58, "learning_rate": 4.5290358204178425e-05, "loss": 0.4208, "step": 3312 }, { "epoch": 0.58, "learning_rate": 4.528616521462819e-05, "loss": 0.4038, "step": 3313 }, { "epoch": 0.58, "learning_rate": 4.528197057778739e-05, "loss": 0.4244, "step": 3314 }, { "epoch": 0.58, "learning_rate": 4.527777429404453e-05, "loss": 0.4294, "step": 3315 }, { "epoch": 0.58, "learning_rate": 4.5273576363788244e-05, "loss": 0.4282, "step": 3316 }, { "epoch": 0.58, "learning_rate": 4.5269376787407335e-05, "loss": 0.4261, "step": 3317 }, { "epoch": 0.58, "learning_rate": 4.5265175565290754e-05, "loss": 0.4178, "step": 3318 }, { "epoch": 0.58, "learning_rate": 4.5260972697827595e-05, "loss": 0.4157, "step": 3319 }, { "epoch": 0.58, "learning_rate": 4.525676818540711e-05, "loss": 0.4116, "step": 3320 }, { "epoch": 0.58, "learning_rate": 4.525256202841872e-05, "loss": 0.4204, "step": 3321 }, { "epoch": 0.58, "learning_rate": 4.524835422725198e-05, "loss": 0.4093, "step": 3322 }, { "epoch": 0.58, "learning_rate": 4.5244144782296594e-05, "loss": 0.4245, "step": 3323 }, { "epoch": 0.58, "learning_rate": 4.5239933693942425e-05, "loss": 0.4073, "step": 3324 }, { "epoch": 0.58, "learning_rate": 4.52357209625795e-05, "loss": 0.4125, "step": 3325 }, { "epoch": 0.58, "learning_rate": 4.5231506588597976e-05, "loss": 0.411, "step": 3326 }, { "epoch": 0.58, "learning_rate": 4.522729057238818e-05, "loss": 0.418, "step": 3327 }, { "epoch": 0.58, "learning_rate": 4.5223072914340585e-05, "loss": 0.4039, "step": 3328 }, { "epoch": 0.58, "learning_rate": 4.5218853614845814e-05, "loss": 0.4218, "step": 3329 }, { "epoch": 0.58, "learning_rate": 4.521463267429464e-05, "loss": 0.4267, "step": 3330 }, { "epoch": 0.58, "learning_rate": 4.521041009307798e-05, "loss": 0.4338, "step": 3331 }, { "epoch": 0.58, "learning_rate": 4.520618587158694e-05, "loss": 0.4118, "step": 3332 }, { "epoch": 0.58, "learning_rate": 4.520196001021274e-05, "loss": 0.413, "step": 3333 }, { "epoch": 0.58, "learning_rate": 4.5197732509346766e-05, "loss": 0.4255, "step": 3334 }, { "epoch": 0.58, "learning_rate": 4.519350336938053e-05, "loss": 0.4104, "step": 3335 }, { "epoch": 0.58, "learning_rate": 4.518927259070577e-05, "loss": 0.4026, "step": 3336 }, { "epoch": 0.58, "learning_rate": 4.518504017371428e-05, "loss": 0.422, "step": 3337 }, { "epoch": 0.58, "learning_rate": 4.518080611879806e-05, "loss": 0.4268, "step": 3338 }, { "epoch": 0.58, "learning_rate": 4.517657042634928e-05, "loss": 0.4194, "step": 3339 }, { "epoch": 0.58, "learning_rate": 4.51723330967602e-05, "loss": 0.4111, "step": 3340 }, { "epoch": 0.58, "learning_rate": 4.5168094130423296e-05, "loss": 0.4252, "step": 3341 }, { "epoch": 0.58, "learning_rate": 4.516385352773114e-05, "loss": 0.4455, "step": 3342 }, { "epoch": 0.58, "learning_rate": 4.515961128907649e-05, "loss": 0.4157, "step": 3343 }, { "epoch": 0.58, "learning_rate": 4.5155367414852256e-05, "loss": 0.412, "step": 3344 }, { "epoch": 0.58, "learning_rate": 4.515112190545148e-05, "loss": 0.4295, "step": 3345 }, { "epoch": 0.58, "learning_rate": 4.514687476126738e-05, "loss": 0.4244, "step": 3346 }, { "epoch": 0.58, "learning_rate": 4.5142625982693296e-05, "loss": 0.4319, "step": 3347 }, { "epoch": 0.58, "learning_rate": 4.513837557012273e-05, "loss": 0.4269, "step": 3348 }, { "epoch": 0.58, "learning_rate": 4.5134123523949355e-05, "loss": 0.4302, "step": 3349 }, { "epoch": 0.58, "learning_rate": 4.5129869844566975e-05, "loss": 0.4388, "step": 3350 }, { "epoch": 0.58, "learning_rate": 4.512561453236955e-05, "loss": 0.4209, "step": 3351 }, { "epoch": 0.58, "learning_rate": 4.5121357587751195e-05, "loss": 0.4157, "step": 3352 }, { "epoch": 0.58, "learning_rate": 4.511709901110616e-05, "loss": 0.4325, "step": 3353 }, { "epoch": 0.58, "learning_rate": 4.5112838802828876e-05, "loss": 0.4074, "step": 3354 }, { "epoch": 0.58, "learning_rate": 4.51085769633139e-05, "loss": 0.4082, "step": 3355 }, { "epoch": 0.59, "learning_rate": 4.510431349295595e-05, "loss": 0.4231, "step": 3356 }, { "epoch": 0.59, "learning_rate": 4.5100048392149885e-05, "loss": 0.4293, "step": 3357 }, { "epoch": 0.59, "learning_rate": 4.5095781661290735e-05, "loss": 0.4123, "step": 3358 }, { "epoch": 0.59, "learning_rate": 4.5091513300773664e-05, "loss": 0.4259, "step": 3359 }, { "epoch": 0.59, "learning_rate": 4.508724331099399e-05, "loss": 0.4469, "step": 3360 }, { "epoch": 0.59, "learning_rate": 4.508297169234719e-05, "loss": 0.4165, "step": 3361 }, { "epoch": 0.59, "learning_rate": 4.507869844522888e-05, "loss": 0.4142, "step": 3362 }, { "epoch": 0.59, "learning_rate": 4.507442357003484e-05, "loss": 0.4038, "step": 3363 }, { "epoch": 0.59, "learning_rate": 4.5070147067160994e-05, "loss": 0.4188, "step": 3364 }, { "epoch": 0.59, "learning_rate": 4.506586893700341e-05, "loss": 0.4238, "step": 3365 }, { "epoch": 0.59, "learning_rate": 4.50615891799583e-05, "loss": 0.4318, "step": 3366 }, { "epoch": 0.59, "learning_rate": 4.5057307796422063e-05, "loss": 0.4127, "step": 3367 }, { "epoch": 0.59, "learning_rate": 4.505302478679122e-05, "loss": 0.4134, "step": 3368 }, { "epoch": 0.59, "learning_rate": 4.5048740151462444e-05, "loss": 0.4315, "step": 3369 }, { "epoch": 0.59, "learning_rate": 4.504445389083256e-05, "loss": 0.4, "step": 3370 }, { "epoch": 0.59, "learning_rate": 4.5040166005298553e-05, "loss": 0.4283, "step": 3371 }, { "epoch": 0.59, "learning_rate": 4.5035876495257536e-05, "loss": 0.4304, "step": 3372 }, { "epoch": 0.59, "learning_rate": 4.503158536110681e-05, "loss": 0.414, "step": 3373 }, { "epoch": 0.59, "learning_rate": 4.5027292603243793e-05, "loss": 0.4267, "step": 3374 }, { "epoch": 0.59, "learning_rate": 4.502299822206606e-05, "loss": 0.4177, "step": 3375 }, { "epoch": 0.59, "learning_rate": 4.501870221797135e-05, "loss": 0.4185, "step": 3376 }, { "epoch": 0.59, "learning_rate": 4.501440459135754e-05, "loss": 0.4015, "step": 3377 }, { "epoch": 0.59, "learning_rate": 4.5010105342622664e-05, "loss": 0.4345, "step": 3378 }, { "epoch": 0.59, "learning_rate": 4.500580447216489e-05, "loss": 0.4304, "step": 3379 }, { "epoch": 0.59, "learning_rate": 4.5001501980382566e-05, "loss": 0.4217, "step": 3380 }, { "epoch": 0.59, "learning_rate": 4.4997197867674166e-05, "loss": 0.4085, "step": 3381 }, { "epoch": 0.59, "learning_rate": 4.499289213443832e-05, "loss": 0.4247, "step": 3382 }, { "epoch": 0.59, "learning_rate": 4.498858478107381e-05, "loss": 0.4124, "step": 3383 }, { "epoch": 0.59, "learning_rate": 4.498427580797957e-05, "loss": 0.4009, "step": 3384 }, { "epoch": 0.59, "learning_rate": 4.497996521555468e-05, "loss": 0.4049, "step": 3385 }, { "epoch": 0.59, "learning_rate": 4.497565300419837e-05, "loss": 0.4357, "step": 3386 }, { "epoch": 0.59, "learning_rate": 4.497133917431003e-05, "loss": 0.4137, "step": 3387 }, { "epoch": 0.59, "learning_rate": 4.4967023726289174e-05, "loss": 0.4422, "step": 3388 }, { "epoch": 0.59, "learning_rate": 4.4962706660535495e-05, "loss": 0.4437, "step": 3389 }, { "epoch": 0.59, "learning_rate": 4.495838797744883e-05, "loss": 0.4035, "step": 3390 }, { "epoch": 0.59, "learning_rate": 4.495406767742915e-05, "loss": 0.4187, "step": 3391 }, { "epoch": 0.59, "learning_rate": 4.494974576087659e-05, "loss": 0.4133, "step": 3392 }, { "epoch": 0.59, "learning_rate": 4.4945422228191433e-05, "loss": 0.412, "step": 3393 }, { "epoch": 0.59, "learning_rate": 4.494109707977409e-05, "loss": 0.4075, "step": 3394 }, { "epoch": 0.59, "learning_rate": 4.4936770316025165e-05, "loss": 0.4223, "step": 3395 }, { "epoch": 0.59, "learning_rate": 4.4932441937345385e-05, "loss": 0.4062, "step": 3396 }, { "epoch": 0.59, "learning_rate": 4.492811194413561e-05, "loss": 0.4257, "step": 3397 }, { "epoch": 0.59, "learning_rate": 4.4923780336796885e-05, "loss": 0.3938, "step": 3398 }, { "epoch": 0.59, "learning_rate": 4.491944711573038e-05, "loss": 0.4195, "step": 3399 }, { "epoch": 0.59, "learning_rate": 4.4915112281337425e-05, "loss": 0.4373, "step": 3400 }, { "epoch": 0.59, "learning_rate": 4.4910775834019494e-05, "loss": 0.4247, "step": 3401 }, { "epoch": 0.59, "learning_rate": 4.490643777417822e-05, "loss": 0.4545, "step": 3402 }, { "epoch": 0.59, "learning_rate": 4.490209810221538e-05, "loss": 0.4139, "step": 3403 }, { "epoch": 0.59, "learning_rate": 4.4897756818532886e-05, "loss": 0.4329, "step": 3404 }, { "epoch": 0.59, "learning_rate": 4.489341392353282e-05, "loss": 0.4295, "step": 3405 }, { "epoch": 0.59, "learning_rate": 4.488906941761741e-05, "loss": 0.4228, "step": 3406 }, { "epoch": 0.59, "learning_rate": 4.4884723301189016e-05, "loss": 0.436, "step": 3407 }, { "epoch": 0.59, "learning_rate": 4.488037557465017e-05, "loss": 0.407, "step": 3408 }, { "epoch": 0.59, "learning_rate": 4.4876026238403535e-05, "loss": 0.4037, "step": 3409 }, { "epoch": 0.59, "learning_rate": 4.487167529285194e-05, "loss": 0.4078, "step": 3410 }, { "epoch": 0.59, "learning_rate": 4.4867322738398346e-05, "loss": 0.4077, "step": 3411 }, { "epoch": 0.59, "learning_rate": 4.4862968575445886e-05, "loss": 0.4074, "step": 3412 }, { "epoch": 0.6, "learning_rate": 4.4858612804397805e-05, "loss": 0.4232, "step": 3413 }, { "epoch": 0.6, "learning_rate": 4.485425542565753e-05, "loss": 0.4278, "step": 3414 }, { "epoch": 0.6, "learning_rate": 4.484989643962862e-05, "loss": 0.398, "step": 3415 }, { "epoch": 0.6, "learning_rate": 4.48455358467148e-05, "loss": 0.4113, "step": 3416 }, { "epoch": 0.6, "learning_rate": 4.4841173647319925e-05, "loss": 0.4349, "step": 3417 }, { "epoch": 0.6, "learning_rate": 4.4836809841847996e-05, "loss": 0.4255, "step": 3418 }, { "epoch": 0.6, "learning_rate": 4.48324444307032e-05, "loss": 0.421, "step": 3419 }, { "epoch": 0.6, "learning_rate": 4.482807741428982e-05, "loss": 0.4179, "step": 3420 }, { "epoch": 0.6, "learning_rate": 4.482370879301232e-05, "loss": 0.4152, "step": 3421 }, { "epoch": 0.6, "learning_rate": 4.481933856727532e-05, "loss": 0.4102, "step": 3422 }, { "epoch": 0.6, "learning_rate": 4.481496673748356e-05, "loss": 0.4186, "step": 3423 }, { "epoch": 0.6, "learning_rate": 4.481059330404194e-05, "loss": 0.4068, "step": 3424 }, { "epoch": 0.6, "learning_rate": 4.4806218267355515e-05, "loss": 0.4384, "step": 3425 }, { "epoch": 0.6, "learning_rate": 4.48018416278295e-05, "loss": 0.4213, "step": 3426 }, { "epoch": 0.6, "learning_rate": 4.4797463385869225e-05, "loss": 0.4029, "step": 3427 }, { "epoch": 0.6, "learning_rate": 4.47930835418802e-05, "loss": 0.4156, "step": 3428 }, { "epoch": 0.6, "learning_rate": 4.478870209626805e-05, "loss": 0.4111, "step": 3429 }, { "epoch": 0.6, "learning_rate": 4.47843190494386e-05, "loss": 0.4318, "step": 3430 }, { "epoch": 0.6, "learning_rate": 4.477993440179777e-05, "loss": 0.4137, "step": 3431 }, { "epoch": 0.6, "learning_rate": 4.477554815375165e-05, "loss": 0.4305, "step": 3432 }, { "epoch": 0.6, "learning_rate": 4.477116030570649e-05, "loss": 0.4148, "step": 3433 }, { "epoch": 0.6, "learning_rate": 4.4766770858068666e-05, "loss": 0.4105, "step": 3434 }, { "epoch": 0.6, "learning_rate": 4.4762379811244714e-05, "loss": 0.4251, "step": 3435 }, { "epoch": 0.6, "learning_rate": 4.475798716564132e-05, "loss": 0.4191, "step": 3436 }, { "epoch": 0.6, "learning_rate": 4.475359292166532e-05, "loss": 0.4072, "step": 3437 }, { "epoch": 0.6, "learning_rate": 4.474919707972369e-05, "loss": 0.4234, "step": 3438 }, { "epoch": 0.6, "learning_rate": 4.474479964022355e-05, "loss": 0.4028, "step": 3439 }, { "epoch": 0.6, "learning_rate": 4.474040060357218e-05, "loss": 0.4339, "step": 3440 }, { "epoch": 0.6, "learning_rate": 4.473599997017701e-05, "loss": 0.4106, "step": 3441 }, { "epoch": 0.6, "learning_rate": 4.4731597740445596e-05, "loss": 0.4126, "step": 3442 }, { "epoch": 0.6, "learning_rate": 4.472719391478567e-05, "loss": 0.4184, "step": 3443 }, { "epoch": 0.6, "learning_rate": 4.472278849360509e-05, "loss": 0.4261, "step": 3444 }, { "epoch": 0.6, "learning_rate": 4.4718381477311885e-05, "loss": 0.4151, "step": 3445 }, { "epoch": 0.6, "learning_rate": 4.4713972866314184e-05, "loss": 0.411, "step": 3446 }, { "epoch": 0.6, "learning_rate": 4.470956266102032e-05, "loss": 0.4158, "step": 3447 }, { "epoch": 0.6, "learning_rate": 4.470515086183877e-05, "loss": 0.3993, "step": 3448 }, { "epoch": 0.6, "learning_rate": 4.47007374691781e-05, "loss": 0.4226, "step": 3449 }, { "epoch": 0.6, "learning_rate": 4.469632248344709e-05, "loss": 0.4164, "step": 3450 }, { "epoch": 0.6, "learning_rate": 4.469190590505463e-05, "loss": 0.3981, "step": 3451 }, { "epoch": 0.6, "learning_rate": 4.468748773440975e-05, "loss": 0.4148, "step": 3452 }, { "epoch": 0.6, "learning_rate": 4.4683067971921676e-05, "loss": 0.4227, "step": 3453 }, { "epoch": 0.6, "learning_rate": 4.467864661799974e-05, "loss": 0.426, "step": 3454 }, { "epoch": 0.6, "learning_rate": 4.467422367305342e-05, "loss": 0.4193, "step": 3455 }, { "epoch": 0.6, "learning_rate": 4.466979913749237e-05, "loss": 0.4445, "step": 3456 }, { "epoch": 0.6, "learning_rate": 4.466537301172636e-05, "loss": 0.4278, "step": 3457 }, { "epoch": 0.6, "learning_rate": 4.4660945296165335e-05, "loss": 0.4027, "step": 3458 }, { "epoch": 0.6, "learning_rate": 4.4656515991219364e-05, "loss": 0.4355, "step": 3459 }, { "epoch": 0.6, "learning_rate": 4.465208509729867e-05, "loss": 0.4241, "step": 3460 }, { "epoch": 0.6, "learning_rate": 4.4647652614813645e-05, "loss": 0.4111, "step": 3461 }, { "epoch": 0.6, "learning_rate": 4.4643218544174795e-05, "loss": 0.4291, "step": 3462 }, { "epoch": 0.6, "learning_rate": 4.463878288579279e-05, "loss": 0.4108, "step": 3463 }, { "epoch": 0.6, "learning_rate": 4.463434564007844e-05, "loss": 0.4248, "step": 3464 }, { "epoch": 0.6, "learning_rate": 4.462990680744271e-05, "loss": 0.4237, "step": 3465 }, { "epoch": 0.6, "learning_rate": 4.4625466388296724e-05, "loss": 0.3905, "step": 3466 }, { "epoch": 0.6, "learning_rate": 4.4621024383051715e-05, "loss": 0.3901, "step": 3467 }, { "epoch": 0.6, "learning_rate": 4.46165807921191e-05, "loss": 0.4096, "step": 3468 }, { "epoch": 0.6, "learning_rate": 4.4612135615910414e-05, "loss": 0.4229, "step": 3469 }, { "epoch": 0.6, "learning_rate": 4.460768885483737e-05, "loss": 0.4192, "step": 3470 }, { "epoch": 0.61, "learning_rate": 4.46032405093118e-05, "loss": 0.4125, "step": 3471 }, { "epoch": 0.61, "learning_rate": 4.4598790579745694e-05, "loss": 0.4205, "step": 3472 }, { "epoch": 0.61, "learning_rate": 4.45943390665512e-05, "loss": 0.4169, "step": 3473 }, { "epoch": 0.61, "learning_rate": 4.458988597014059e-05, "loss": 0.4342, "step": 3474 }, { "epoch": 0.61, "learning_rate": 4.458543129092629e-05, "loss": 0.414, "step": 3475 }, { "epoch": 0.61, "learning_rate": 4.458097502932089e-05, "loss": 0.4093, "step": 3476 }, { "epoch": 0.61, "learning_rate": 4.4576517185737096e-05, "loss": 0.4048, "step": 3477 }, { "epoch": 0.61, "learning_rate": 4.4572057760587796e-05, "loss": 0.426, "step": 3478 }, { "epoch": 0.61, "learning_rate": 4.4567596754285996e-05, "loss": 0.4245, "step": 3479 }, { "epoch": 0.61, "learning_rate": 4.456313416724486e-05, "loss": 0.4031, "step": 3480 }, { "epoch": 0.61, "learning_rate": 4.455866999987769e-05, "loss": 0.4047, "step": 3481 }, { "epoch": 0.61, "learning_rate": 4.4554204252597956e-05, "loss": 0.4303, "step": 3482 }, { "epoch": 0.61, "learning_rate": 4.4549736925819246e-05, "loss": 0.4208, "step": 3483 }, { "epoch": 0.61, "learning_rate": 4.4545268019955304e-05, "loss": 0.4226, "step": 3484 }, { "epoch": 0.61, "learning_rate": 4.454079753542005e-05, "loss": 0.3982, "step": 3485 }, { "epoch": 0.61, "learning_rate": 4.4536325472627496e-05, "loss": 0.423, "step": 3486 }, { "epoch": 0.61, "learning_rate": 4.4531851831991837e-05, "loss": 0.4003, "step": 3487 }, { "epoch": 0.61, "learning_rate": 4.452737661392741e-05, "loss": 0.413, "step": 3488 }, { "epoch": 0.61, "learning_rate": 4.452289981884869e-05, "loss": 0.4372, "step": 3489 }, { "epoch": 0.61, "learning_rate": 4.45184214471703e-05, "loss": 0.4274, "step": 3490 }, { "epoch": 0.61, "learning_rate": 4.4513941499307016e-05, "loss": 0.4051, "step": 3491 }, { "epoch": 0.61, "learning_rate": 4.450945997567375e-05, "loss": 0.4155, "step": 3492 }, { "epoch": 0.61, "learning_rate": 4.4504976876685574e-05, "loss": 0.413, "step": 3493 }, { "epoch": 0.61, "learning_rate": 4.450049220275768e-05, "loss": 0.4125, "step": 3494 }, { "epoch": 0.61, "learning_rate": 4.449600595430544e-05, "loss": 0.4322, "step": 3495 }, { "epoch": 0.61, "learning_rate": 4.449151813174433e-05, "loss": 0.4052, "step": 3496 }, { "epoch": 0.61, "learning_rate": 4.4487028735490024e-05, "loss": 0.4235, "step": 3497 }, { "epoch": 0.61, "learning_rate": 4.44825377659583e-05, "loss": 0.4177, "step": 3498 }, { "epoch": 0.61, "learning_rate": 4.447804522356509e-05, "loss": 0.4096, "step": 3499 }, { "epoch": 0.61, "learning_rate": 4.447355110872649e-05, "loss": 0.4266, "step": 3500 }, { "epoch": 0.61, "learning_rate": 4.4469055421858716e-05, "loss": 0.4006, "step": 3501 }, { "epoch": 0.61, "learning_rate": 4.4464558163378154e-05, "loss": 0.399, "step": 3502 }, { "epoch": 0.61, "learning_rate": 4.4460059333701314e-05, "loss": 0.4109, "step": 3503 }, { "epoch": 0.61, "learning_rate": 4.4455558933244876e-05, "loss": 0.418, "step": 3504 }, { "epoch": 0.61, "learning_rate": 4.4451056962425626e-05, "loss": 0.4188, "step": 3505 }, { "epoch": 0.61, "learning_rate": 4.444655342166055e-05, "loss": 0.4204, "step": 3506 }, { "epoch": 0.61, "learning_rate": 4.444204831136672e-05, "loss": 0.409, "step": 3507 }, { "epoch": 0.61, "learning_rate": 4.4437541631961404e-05, "loss": 0.417, "step": 3508 }, { "epoch": 0.61, "learning_rate": 4.4433033383861986e-05, "loss": 0.4202, "step": 3509 }, { "epoch": 0.61, "learning_rate": 4.442852356748601e-05, "loss": 0.4282, "step": 3510 }, { "epoch": 0.61, "learning_rate": 4.442401218325116e-05, "loss": 0.4263, "step": 3511 }, { "epoch": 0.61, "learning_rate": 4.441949923157525e-05, "loss": 0.4251, "step": 3512 }, { "epoch": 0.61, "learning_rate": 4.441498471287627e-05, "loss": 0.4036, "step": 3513 }, { "epoch": 0.61, "learning_rate": 4.441046862757232e-05, "loss": 0.418, "step": 3514 }, { "epoch": 0.61, "learning_rate": 4.440595097608168e-05, "loss": 0.4128, "step": 3515 }, { "epoch": 0.61, "learning_rate": 4.4401431758822756e-05, "loss": 0.4153, "step": 3516 }, { "epoch": 0.61, "learning_rate": 4.4396910976214094e-05, "loss": 0.4307, "step": 3517 }, { "epoch": 0.61, "learning_rate": 4.4392388628674405e-05, "loss": 0.3843, "step": 3518 }, { "epoch": 0.61, "learning_rate": 4.438786471662252e-05, "loss": 0.4051, "step": 3519 }, { "epoch": 0.61, "learning_rate": 4.438333924047743e-05, "loss": 0.439, "step": 3520 }, { "epoch": 0.61, "learning_rate": 4.437881220065827e-05, "loss": 0.4103, "step": 3521 }, { "epoch": 0.61, "learning_rate": 4.437428359758432e-05, "loss": 0.4063, "step": 3522 }, { "epoch": 0.61, "learning_rate": 4.4369753431675e-05, "loss": 0.4294, "step": 3523 }, { "epoch": 0.61, "learning_rate": 4.4365221703349885e-05, "loss": 0.4165, "step": 3524 }, { "epoch": 0.61, "learning_rate": 4.436068841302868e-05, "loss": 0.4051, "step": 3525 }, { "epoch": 0.61, "learning_rate": 4.435615356113124e-05, "loss": 0.4056, "step": 3526 }, { "epoch": 0.61, "learning_rate": 4.435161714807757e-05, "loss": 0.4072, "step": 3527 }, { "epoch": 0.62, "learning_rate": 4.434707917428783e-05, "loss": 0.4271, "step": 3528 }, { "epoch": 0.62, "learning_rate": 4.4342539640182275e-05, "loss": 0.413, "step": 3529 }, { "epoch": 0.62, "learning_rate": 4.4337998546181375e-05, "loss": 0.4168, "step": 3530 }, { "epoch": 0.62, "learning_rate": 4.43334558927057e-05, "loss": 0.421, "step": 3531 }, { "epoch": 0.62, "learning_rate": 4.4328911680175966e-05, "loss": 0.4043, "step": 3532 }, { "epoch": 0.62, "learning_rate": 4.432436590901305e-05, "loss": 0.3973, "step": 3533 }, { "epoch": 0.62, "learning_rate": 4.431981857963796e-05, "loss": 0.4121, "step": 3534 }, { "epoch": 0.62, "learning_rate": 4.431526969247186e-05, "loss": 0.4167, "step": 3535 }, { "epoch": 0.62, "learning_rate": 4.431071924793603e-05, "loss": 0.404, "step": 3536 }, { "epoch": 0.62, "learning_rate": 4.430616724645195e-05, "loss": 0.4128, "step": 3537 }, { "epoch": 0.62, "learning_rate": 4.430161368844119e-05, "loss": 0.4275, "step": 3538 }, { "epoch": 0.62, "learning_rate": 4.429705857432548e-05, "loss": 0.4193, "step": 3539 }, { "epoch": 0.62, "learning_rate": 4.429250190452672e-05, "loss": 0.4187, "step": 3540 }, { "epoch": 0.62, "learning_rate": 4.4287943679466904e-05, "loss": 0.4215, "step": 3541 }, { "epoch": 0.62, "learning_rate": 4.428338389956822e-05, "loss": 0.4174, "step": 3542 }, { "epoch": 0.62, "learning_rate": 4.427882256525296e-05, "loss": 0.4118, "step": 3543 }, { "epoch": 0.62, "learning_rate": 4.4274259676943593e-05, "loss": 0.436, "step": 3544 }, { "epoch": 0.62, "learning_rate": 4.4269695235062716e-05, "loss": 0.4161, "step": 3545 }, { "epoch": 0.62, "learning_rate": 4.426512924003307e-05, "loss": 0.4246, "step": 3546 }, { "epoch": 0.62, "learning_rate": 4.426056169227753e-05, "loss": 0.4112, "step": 3547 }, { "epoch": 0.62, "learning_rate": 4.4255992592219134e-05, "loss": 0.4025, "step": 3548 }, { "epoch": 0.62, "learning_rate": 4.425142194028106e-05, "loss": 0.4083, "step": 3549 }, { "epoch": 0.62, "learning_rate": 4.424684973688663e-05, "loss": 0.4148, "step": 3550 }, { "epoch": 0.62, "learning_rate": 4.4242275982459285e-05, "loss": 0.4435, "step": 3551 }, { "epoch": 0.62, "learning_rate": 4.423770067742265e-05, "loss": 0.4272, "step": 3552 }, { "epoch": 0.62, "learning_rate": 4.423312382220046e-05, "loss": 0.3942, "step": 3553 }, { "epoch": 0.62, "learning_rate": 4.4228545417216624e-05, "loss": 0.4124, "step": 3554 }, { "epoch": 0.62, "learning_rate": 4.4223965462895143e-05, "loss": 0.4019, "step": 3555 }, { "epoch": 0.62, "learning_rate": 4.4219383959660234e-05, "loss": 0.4189, "step": 3556 }, { "epoch": 0.62, "learning_rate": 4.42148009079362e-05, "loss": 0.4125, "step": 3557 }, { "epoch": 0.62, "learning_rate": 4.4210216308147506e-05, "loss": 0.4187, "step": 3558 }, { "epoch": 0.62, "learning_rate": 4.420563016071877e-05, "loss": 0.4105, "step": 3559 }, { "epoch": 0.62, "learning_rate": 4.420104246607473e-05, "loss": 0.3946, "step": 3560 }, { "epoch": 0.62, "learning_rate": 4.4196453224640294e-05, "loss": 0.4103, "step": 3561 }, { "epoch": 0.62, "learning_rate": 4.419186243684051e-05, "loss": 0.4055, "step": 3562 }, { "epoch": 0.62, "learning_rate": 4.418727010310054e-05, "loss": 0.4301, "step": 3563 }, { "epoch": 0.62, "learning_rate": 4.4182676223845716e-05, "loss": 0.4111, "step": 3564 }, { "epoch": 0.62, "learning_rate": 4.417808079950151e-05, "loss": 0.4169, "step": 3565 }, { "epoch": 0.62, "learning_rate": 4.417348383049353e-05, "loss": 0.4261, "step": 3566 }, { "epoch": 0.62, "learning_rate": 4.416888531724754e-05, "loss": 0.3993, "step": 3567 }, { "epoch": 0.62, "learning_rate": 4.416428526018942e-05, "loss": 0.3926, "step": 3568 }, { "epoch": 0.62, "learning_rate": 4.4159683659745225e-05, "loss": 0.3861, "step": 3569 }, { "epoch": 0.62, "learning_rate": 4.415508051634113e-05, "loss": 0.4205, "step": 3570 }, { "epoch": 0.62, "learning_rate": 4.4150475830403477e-05, "loss": 0.4076, "step": 3571 }, { "epoch": 0.62, "learning_rate": 4.414586960235873e-05, "loss": 0.4259, "step": 3572 }, { "epoch": 0.62, "learning_rate": 4.4141261832633486e-05, "loss": 0.4285, "step": 3573 }, { "epoch": 0.62, "learning_rate": 4.4136652521654506e-05, "loss": 0.4068, "step": 3574 }, { "epoch": 0.62, "learning_rate": 4.4132041669848694e-05, "loss": 0.4064, "step": 3575 }, { "epoch": 0.62, "learning_rate": 4.4127429277643095e-05, "loss": 0.4159, "step": 3576 }, { "epoch": 0.62, "learning_rate": 4.412281534546488e-05, "loss": 0.4075, "step": 3577 }, { "epoch": 0.62, "learning_rate": 4.4118199873741385e-05, "loss": 0.4051, "step": 3578 }, { "epoch": 0.62, "learning_rate": 4.411358286290007e-05, "loss": 0.4161, "step": 3579 }, { "epoch": 0.62, "learning_rate": 4.410896431336855e-05, "loss": 0.4213, "step": 3580 }, { "epoch": 0.62, "learning_rate": 4.410434422557458e-05, "loss": 0.4015, "step": 3581 }, { "epoch": 0.62, "learning_rate": 4.409972259994605e-05, "loss": 0.4197, "step": 3582 }, { "epoch": 0.62, "learning_rate": 4.4095099436911e-05, "loss": 0.4197, "step": 3583 }, { "epoch": 0.62, "learning_rate": 4.409047473689762e-05, "loss": 0.4105, "step": 3584 }, { "epoch": 0.63, "learning_rate": 4.4085848500334226e-05, "loss": 0.4194, "step": 3585 }, { "epoch": 0.63, "learning_rate": 4.408122072764928e-05, "loss": 0.3981, "step": 3586 }, { "epoch": 0.63, "learning_rate": 4.4076591419271395e-05, "loss": 0.4139, "step": 3587 }, { "epoch": 0.63, "learning_rate": 4.407196057562932e-05, "loss": 0.4137, "step": 3588 }, { "epoch": 0.63, "learning_rate": 4.4067328197151934e-05, "loss": 0.3871, "step": 3589 }, { "epoch": 0.63, "learning_rate": 4.40626942842683e-05, "loss": 0.4188, "step": 3590 }, { "epoch": 0.63, "learning_rate": 4.4058058837407566e-05, "loss": 0.4297, "step": 3591 }, { "epoch": 0.63, "learning_rate": 4.4053421856999074e-05, "loss": 0.4105, "step": 3592 }, { "epoch": 0.63, "learning_rate": 4.4048783343472264e-05, "loss": 0.4236, "step": 3593 }, { "epoch": 0.63, "learning_rate": 4.404414329725674e-05, "loss": 0.4005, "step": 3594 }, { "epoch": 0.63, "learning_rate": 4.4039501718782266e-05, "loss": 0.4257, "step": 3595 }, { "epoch": 0.63, "learning_rate": 4.4034858608478714e-05, "loss": 0.406, "step": 3596 }, { "epoch": 0.63, "learning_rate": 4.4030213966776104e-05, "loss": 0.4163, "step": 3597 }, { "epoch": 0.63, "learning_rate": 4.402556779410462e-05, "loss": 0.4217, "step": 3598 }, { "epoch": 0.63, "learning_rate": 4.4020920090894576e-05, "loss": 0.4103, "step": 3599 }, { "epoch": 0.63, "learning_rate": 4.4016270857576405e-05, "loss": 0.421, "step": 3600 }, { "epoch": 0.63, "learning_rate": 4.401162009458073e-05, "loss": 0.421, "step": 3601 }, { "epoch": 0.63, "learning_rate": 4.400696780233826e-05, "loss": 0.4021, "step": 3602 }, { "epoch": 0.63, "learning_rate": 4.40023139812799e-05, "loss": 0.4196, "step": 3603 }, { "epoch": 0.63, "learning_rate": 4.399765863183665e-05, "loss": 0.4333, "step": 3604 }, { "epoch": 0.63, "learning_rate": 4.399300175443968e-05, "loss": 0.4305, "step": 3605 }, { "epoch": 0.63, "learning_rate": 4.3988343349520303e-05, "loss": 0.4162, "step": 3606 }, { "epoch": 0.63, "learning_rate": 4.398368341750994e-05, "loss": 0.4111, "step": 3607 }, { "epoch": 0.63, "learning_rate": 4.39790219588402e-05, "loss": 0.4541, "step": 3608 }, { "epoch": 0.63, "learning_rate": 4.397435897394279e-05, "loss": 0.4064, "step": 3609 }, { "epoch": 0.63, "learning_rate": 4.39696944632496e-05, "loss": 0.4054, "step": 3610 }, { "epoch": 0.63, "learning_rate": 4.396502842719263e-05, "loss": 0.4281, "step": 3611 }, { "epoch": 0.63, "learning_rate": 4.396036086620403e-05, "loss": 0.4334, "step": 3612 }, { "epoch": 0.63, "learning_rate": 4.3955691780716094e-05, "loss": 0.4267, "step": 3613 }, { "epoch": 0.63, "learning_rate": 4.3951021171161246e-05, "loss": 0.3988, "step": 3614 }, { "epoch": 0.63, "learning_rate": 4.394634903797208e-05, "loss": 0.4022, "step": 3615 }, { "epoch": 0.63, "learning_rate": 4.39416753815813e-05, "loss": 0.4134, "step": 3616 }, { "epoch": 0.63, "learning_rate": 4.393700020242177e-05, "loss": 0.408, "step": 3617 }, { "epoch": 0.63, "learning_rate": 4.3932323500926485e-05, "loss": 0.4217, "step": 3618 }, { "epoch": 0.63, "learning_rate": 4.3927645277528576e-05, "loss": 0.4094, "step": 3619 }, { "epoch": 0.63, "learning_rate": 4.3922965532661335e-05, "loss": 0.4053, "step": 3620 }, { "epoch": 0.63, "learning_rate": 4.391828426675818e-05, "loss": 0.4159, "step": 3621 }, { "epoch": 0.63, "learning_rate": 4.3913601480252674e-05, "loss": 0.4216, "step": 3622 }, { "epoch": 0.63, "learning_rate": 4.390891717357852e-05, "loss": 0.4231, "step": 3623 }, { "epoch": 0.63, "learning_rate": 4.3904231347169556e-05, "loss": 0.4136, "step": 3624 }, { "epoch": 0.63, "learning_rate": 4.389954400145977e-05, "loss": 0.4347, "step": 3625 }, { "epoch": 0.63, "learning_rate": 4.389485513688329e-05, "loss": 0.4115, "step": 3626 }, { "epoch": 0.63, "learning_rate": 4.389016475387438e-05, "loss": 0.399, "step": 3627 }, { "epoch": 0.63, "learning_rate": 4.3885472852867434e-05, "loss": 0.4021, "step": 3628 }, { "epoch": 0.63, "learning_rate": 4.388077943429701e-05, "loss": 0.4082, "step": 3629 }, { "epoch": 0.63, "learning_rate": 4.387608449859781e-05, "loss": 0.4098, "step": 3630 }, { "epoch": 0.63, "learning_rate": 4.387138804620464e-05, "loss": 0.418, "step": 3631 }, { "epoch": 0.63, "learning_rate": 4.386669007755248e-05, "loss": 0.4185, "step": 3632 }, { "epoch": 0.63, "learning_rate": 4.386199059307643e-05, "loss": 0.3935, "step": 3633 }, { "epoch": 0.63, "learning_rate": 4.3857289593211746e-05, "loss": 0.4199, "step": 3634 }, { "epoch": 0.63, "learning_rate": 4.385258707839381e-05, "loss": 0.3956, "step": 3635 }, { "epoch": 0.63, "learning_rate": 4.384788304905817e-05, "loss": 0.4279, "step": 3636 }, { "epoch": 0.63, "learning_rate": 4.384317750564048e-05, "loss": 0.4369, "step": 3637 }, { "epoch": 0.63, "learning_rate": 4.3838470448576555e-05, "loss": 0.4027, "step": 3638 }, { "epoch": 0.63, "learning_rate": 4.383376187830234e-05, "loss": 0.4026, "step": 3639 }, { "epoch": 0.63, "learning_rate": 4.382905179525394e-05, "loss": 0.4144, "step": 3640 }, { "epoch": 0.63, "learning_rate": 4.382434019986758e-05, "loss": 0.4058, "step": 3641 }, { "epoch": 0.63, "learning_rate": 4.3819627092579614e-05, "loss": 0.4156, "step": 3642 }, { "epoch": 0.64, "learning_rate": 4.381491247382657e-05, "loss": 0.398, "step": 3643 }, { "epoch": 0.64, "learning_rate": 4.381019634404511e-05, "loss": 0.4262, "step": 3644 }, { "epoch": 0.64, "learning_rate": 4.3805478703671997e-05, "loss": 0.4147, "step": 3645 }, { "epoch": 0.64, "learning_rate": 4.3800759553144184e-05, "loss": 0.4087, "step": 3646 }, { "epoch": 0.64, "learning_rate": 4.379603889289873e-05, "loss": 0.4115, "step": 3647 }, { "epoch": 0.64, "learning_rate": 4.379131672337285e-05, "loss": 0.4051, "step": 3648 }, { "epoch": 0.64, "learning_rate": 4.378659304500389e-05, "loss": 0.4305, "step": 3649 }, { "epoch": 0.64, "learning_rate": 4.3781867858229345e-05, "loss": 0.4172, "step": 3650 }, { "epoch": 0.64, "learning_rate": 4.377714116348685e-05, "loss": 0.4186, "step": 3651 }, { "epoch": 0.64, "learning_rate": 4.377241296121417e-05, "loss": 0.395, "step": 3652 }, { "epoch": 0.64, "learning_rate": 4.3767683251849205e-05, "loss": 0.4133, "step": 3653 }, { "epoch": 0.64, "learning_rate": 4.376295203583001e-05, "loss": 0.4108, "step": 3654 }, { "epoch": 0.64, "learning_rate": 4.375821931359478e-05, "loss": 0.4209, "step": 3655 }, { "epoch": 0.64, "learning_rate": 4.375348508558182e-05, "loss": 0.4013, "step": 3656 }, { "epoch": 0.64, "learning_rate": 4.374874935222964e-05, "loss": 0.4133, "step": 3657 }, { "epoch": 0.64, "learning_rate": 4.37440121139768e-05, "loss": 0.4048, "step": 3658 }, { "epoch": 0.64, "learning_rate": 4.373927337126208e-05, "loss": 0.4, "step": 3659 }, { "epoch": 0.64, "learning_rate": 4.373453312452434e-05, "loss": 0.4187, "step": 3660 }, { "epoch": 0.64, "learning_rate": 4.372979137420262e-05, "loss": 0.4044, "step": 3661 }, { "epoch": 0.64, "learning_rate": 4.372504812073608e-05, "loss": 0.4133, "step": 3662 }, { "epoch": 0.64, "learning_rate": 4.372030336456403e-05, "loss": 0.4185, "step": 3663 }, { "epoch": 0.64, "learning_rate": 4.37155571061259e-05, "loss": 0.4107, "step": 3664 }, { "epoch": 0.64, "learning_rate": 4.371080934586127e-05, "loss": 0.421, "step": 3665 }, { "epoch": 0.64, "learning_rate": 4.370606008420987e-05, "loss": 0.4088, "step": 3666 }, { "epoch": 0.64, "learning_rate": 4.370130932161156e-05, "loss": 0.4138, "step": 3667 }, { "epoch": 0.64, "learning_rate": 4.3696557058506334e-05, "loss": 0.4111, "step": 3668 }, { "epoch": 0.64, "learning_rate": 4.369180329533433e-05, "loss": 0.4102, "step": 3669 }, { "epoch": 0.64, "learning_rate": 4.368704803253583e-05, "loss": 0.3916, "step": 3670 }, { "epoch": 0.64, "learning_rate": 4.3682291270551236e-05, "loss": 0.4317, "step": 3671 }, { "epoch": 0.64, "learning_rate": 4.367753300982112e-05, "loss": 0.445, "step": 3672 }, { "epoch": 0.64, "learning_rate": 4.367277325078615e-05, "loss": 0.4111, "step": 3673 }, { "epoch": 0.64, "learning_rate": 4.366801199388719e-05, "loss": 0.4132, "step": 3674 }, { "epoch": 0.64, "learning_rate": 4.3663249239565176e-05, "loss": 0.4166, "step": 3675 }, { "epoch": 0.64, "learning_rate": 4.365848498826125e-05, "loss": 0.4136, "step": 3676 }, { "epoch": 0.64, "learning_rate": 4.365371924041664e-05, "loss": 0.4073, "step": 3677 }, { "epoch": 0.64, "learning_rate": 4.3648951996472735e-05, "loss": 0.4377, "step": 3678 }, { "epoch": 0.64, "learning_rate": 4.364418325687107e-05, "loss": 0.4299, "step": 3679 }, { "epoch": 0.64, "learning_rate": 4.363941302205329e-05, "loss": 0.4204, "step": 3680 }, { "epoch": 0.64, "learning_rate": 4.363464129246122e-05, "loss": 0.3994, "step": 3681 }, { "epoch": 0.64, "learning_rate": 4.362986806853677e-05, "loss": 0.3936, "step": 3682 }, { "epoch": 0.64, "learning_rate": 4.3625093350722046e-05, "loss": 0.4078, "step": 3683 }, { "epoch": 0.64, "learning_rate": 4.362031713945926e-05, "loss": 0.4133, "step": 3684 }, { "epoch": 0.64, "learning_rate": 4.361553943519076e-05, "loss": 0.4252, "step": 3685 }, { "epoch": 0.64, "learning_rate": 4.3610760238359045e-05, "loss": 0.3951, "step": 3686 }, { "epoch": 0.64, "learning_rate": 4.360597954940674e-05, "loss": 0.3991, "step": 3687 }, { "epoch": 0.64, "learning_rate": 4.360119736877662e-05, "loss": 0.4285, "step": 3688 }, { "epoch": 0.64, "learning_rate": 4.35964136969116e-05, "loss": 0.4017, "step": 3689 }, { "epoch": 0.64, "learning_rate": 4.3591628534254716e-05, "loss": 0.4109, "step": 3690 }, { "epoch": 0.64, "learning_rate": 4.3586841881249155e-05, "loss": 0.4112, "step": 3691 }, { "epoch": 0.64, "learning_rate": 4.358205373833824e-05, "loss": 0.4101, "step": 3692 }, { "epoch": 0.64, "learning_rate": 4.3577264105965436e-05, "loss": 0.4325, "step": 3693 }, { "epoch": 0.64, "learning_rate": 4.3572472984574333e-05, "loss": 0.4015, "step": 3694 }, { "epoch": 0.64, "learning_rate": 4.3567680374608674e-05, "loss": 0.4023, "step": 3695 }, { "epoch": 0.64, "learning_rate": 4.3562886276512345e-05, "loss": 0.4054, "step": 3696 }, { "epoch": 0.64, "learning_rate": 4.355809069072933e-05, "loss": 0.4013, "step": 3697 }, { "epoch": 0.64, "learning_rate": 4.35532936177038e-05, "loss": 0.4264, "step": 3698 }, { "epoch": 0.64, "learning_rate": 4.354849505788003e-05, "loss": 0.3909, "step": 3699 }, { "epoch": 0.65, "learning_rate": 4.354369501170246e-05, "loss": 0.427, "step": 3700 }, { "epoch": 0.65, "learning_rate": 4.353889347961564e-05, "loss": 0.4213, "step": 3701 }, { "epoch": 0.65, "learning_rate": 4.353409046206427e-05, "loss": 0.4139, "step": 3702 }, { "epoch": 0.65, "learning_rate": 4.352928595949321e-05, "loss": 0.4113, "step": 3703 }, { "epoch": 0.65, "learning_rate": 4.352447997234741e-05, "loss": 0.3989, "step": 3704 }, { "epoch": 0.65, "learning_rate": 4.351967250107199e-05, "loss": 0.4302, "step": 3705 }, { "epoch": 0.65, "learning_rate": 4.3514863546112205e-05, "loss": 0.4107, "step": 3706 }, { "epoch": 0.65, "learning_rate": 4.351005310791344e-05, "loss": 0.4135, "step": 3707 }, { "epoch": 0.65, "learning_rate": 4.350524118692122e-05, "loss": 0.4025, "step": 3708 }, { "epoch": 0.65, "learning_rate": 4.3500427783581216e-05, "loss": 0.4019, "step": 3709 }, { "epoch": 0.65, "learning_rate": 4.349561289833922e-05, "loss": 0.4279, "step": 3710 }, { "epoch": 0.65, "learning_rate": 4.349079653164117e-05, "loss": 0.4249, "step": 3711 }, { "epoch": 0.65, "learning_rate": 4.348597868393314e-05, "loss": 0.4364, "step": 3712 }, { "epoch": 0.65, "learning_rate": 4.348115935566134e-05, "loss": 0.4097, "step": 3713 }, { "epoch": 0.65, "learning_rate": 4.347633854727212e-05, "loss": 0.3994, "step": 3714 }, { "epoch": 0.65, "learning_rate": 4.347151625921198e-05, "loss": 0.4098, "step": 3715 }, { "epoch": 0.65, "learning_rate": 4.346669249192752e-05, "loss": 0.4103, "step": 3716 }, { "epoch": 0.65, "learning_rate": 4.3461867245865524e-05, "loss": 0.4244, "step": 3717 }, { "epoch": 0.65, "learning_rate": 4.3457040521472856e-05, "loss": 0.4116, "step": 3718 }, { "epoch": 0.65, "learning_rate": 4.3452212319196584e-05, "loss": 0.4223, "step": 3719 }, { "epoch": 0.65, "learning_rate": 4.3447382639483865e-05, "loss": 0.4097, "step": 3720 }, { "epoch": 0.65, "learning_rate": 4.3442551482782e-05, "loss": 0.4165, "step": 3721 }, { "epoch": 0.65, "learning_rate": 4.343771884953844e-05, "loss": 0.4094, "step": 3722 }, { "epoch": 0.65, "learning_rate": 4.3432884740200766e-05, "loss": 0.4123, "step": 3723 }, { "epoch": 0.65, "learning_rate": 4.342804915521669e-05, "loss": 0.4187, "step": 3724 }, { "epoch": 0.65, "learning_rate": 4.342321209503408e-05, "loss": 0.4072, "step": 3725 }, { "epoch": 0.65, "learning_rate": 4.341837356010092e-05, "loss": 0.4297, "step": 3726 }, { "epoch": 0.65, "learning_rate": 4.3413533550865327e-05, "loss": 0.3972, "step": 3727 }, { "epoch": 0.65, "learning_rate": 4.340869206777558e-05, "loss": 0.4187, "step": 3728 }, { "epoch": 0.65, "learning_rate": 4.3403849111280065e-05, "loss": 0.4359, "step": 3729 }, { "epoch": 0.65, "learning_rate": 4.3399004681827336e-05, "loss": 0.4165, "step": 3730 }, { "epoch": 0.65, "learning_rate": 4.339415877986607e-05, "loss": 0.4118, "step": 3731 }, { "epoch": 0.65, "learning_rate": 4.338931140584505e-05, "loss": 0.4017, "step": 3732 }, { "epoch": 0.65, "learning_rate": 4.338446256021323e-05, "loss": 0.4352, "step": 3733 }, { "epoch": 0.65, "learning_rate": 4.3379612243419725e-05, "loss": 0.4018, "step": 3734 }, { "epoch": 0.65, "learning_rate": 4.337476045591372e-05, "loss": 0.4122, "step": 3735 }, { "epoch": 0.65, "learning_rate": 4.336990719814457e-05, "loss": 0.4267, "step": 3736 }, { "epoch": 0.65, "learning_rate": 4.336505247056179e-05, "loss": 0.3933, "step": 3737 }, { "epoch": 0.65, "learning_rate": 4.336019627361499e-05, "loss": 0.4237, "step": 3738 }, { "epoch": 0.65, "learning_rate": 4.335533860775393e-05, "loss": 0.4245, "step": 3739 }, { "epoch": 0.65, "learning_rate": 4.335047947342852e-05, "loss": 0.4185, "step": 3740 }, { "epoch": 0.65, "learning_rate": 4.3345618871088795e-05, "loss": 0.4182, "step": 3741 }, { "epoch": 0.65, "learning_rate": 4.334075680118492e-05, "loss": 0.4039, "step": 3742 }, { "epoch": 0.65, "learning_rate": 4.333589326416721e-05, "loss": 0.4108, "step": 3743 }, { "epoch": 0.65, "learning_rate": 4.33310282604861e-05, "loss": 0.4101, "step": 3744 }, { "epoch": 0.65, "learning_rate": 4.332616179059218e-05, "loss": 0.3935, "step": 3745 }, { "epoch": 0.65, "learning_rate": 4.3321293854936146e-05, "loss": 0.429, "step": 3746 }, { "epoch": 0.65, "learning_rate": 4.331642445396888e-05, "loss": 0.4124, "step": 3747 }, { "epoch": 0.65, "learning_rate": 4.331155358814133e-05, "loss": 0.4104, "step": 3748 }, { "epoch": 0.65, "learning_rate": 4.3306681257904644e-05, "loss": 0.4116, "step": 3749 }, { "epoch": 0.65, "learning_rate": 4.3301807463710076e-05, "loss": 0.4195, "step": 3750 }, { "epoch": 0.65, "learning_rate": 4.3296932206009014e-05, "loss": 0.429, "step": 3751 }, { "epoch": 0.65, "learning_rate": 4.329205548525299e-05, "loss": 0.4138, "step": 3752 }, { "epoch": 0.65, "learning_rate": 4.328717730189366e-05, "loss": 0.4009, "step": 3753 }, { "epoch": 0.65, "learning_rate": 4.3282297656382835e-05, "loss": 0.4265, "step": 3754 }, { "epoch": 0.65, "learning_rate": 4.327741654917245e-05, "loss": 0.4145, "step": 3755 }, { "epoch": 0.65, "learning_rate": 4.327253398071457e-05, "loss": 0.4057, "step": 3756 }, { "epoch": 0.66, "learning_rate": 4.3267649951461396e-05, "loss": 0.4235, "step": 3757 }, { "epoch": 0.66, "learning_rate": 4.326276446186529e-05, "loss": 0.3992, "step": 3758 }, { "epoch": 0.66, "learning_rate": 4.325787751237871e-05, "loss": 0.4219, "step": 3759 }, { "epoch": 0.66, "learning_rate": 4.3252989103454264e-05, "loss": 0.4204, "step": 3760 }, { "epoch": 0.66, "learning_rate": 4.324809923554471e-05, "loss": 0.4123, "step": 3761 }, { "epoch": 0.66, "learning_rate": 4.324320790910293e-05, "loss": 0.4248, "step": 3762 }, { "epoch": 0.66, "learning_rate": 4.3238315124581935e-05, "loss": 0.4133, "step": 3763 }, { "epoch": 0.66, "learning_rate": 4.3233420882434886e-05, "loss": 0.4059, "step": 3764 }, { "epoch": 0.66, "learning_rate": 4.322852518311506e-05, "loss": 0.4047, "step": 3765 }, { "epoch": 0.66, "learning_rate": 4.322362802707588e-05, "loss": 0.4307, "step": 3766 }, { "epoch": 0.66, "learning_rate": 4.321872941477091e-05, "loss": 0.4359, "step": 3767 }, { "epoch": 0.66, "learning_rate": 4.321382934665383e-05, "loss": 0.3993, "step": 3768 }, { "epoch": 0.66, "learning_rate": 4.3208927823178486e-05, "loss": 0.4228, "step": 3769 }, { "epoch": 0.66, "learning_rate": 4.3204024844798834e-05, "loss": 0.4209, "step": 3770 }, { "epoch": 0.66, "learning_rate": 4.319912041196895e-05, "loss": 0.4107, "step": 3771 }, { "epoch": 0.66, "learning_rate": 4.319421452514308e-05, "loss": 0.4122, "step": 3772 }, { "epoch": 0.66, "learning_rate": 4.318930718477559e-05, "loss": 0.41, "step": 3773 }, { "epoch": 0.66, "learning_rate": 4.318439839132098e-05, "loss": 0.4282, "step": 3774 }, { "epoch": 0.66, "learning_rate": 4.3179488145233885e-05, "loss": 0.4097, "step": 3775 }, { "epoch": 0.66, "learning_rate": 4.317457644696908e-05, "loss": 0.3939, "step": 3776 }, { "epoch": 0.66, "learning_rate": 4.316966329698145e-05, "loss": 0.4005, "step": 3777 }, { "epoch": 0.66, "learning_rate": 4.316474869572605e-05, "loss": 0.4149, "step": 3778 }, { "epoch": 0.66, "learning_rate": 4.315983264365805e-05, "loss": 0.3953, "step": 3779 }, { "epoch": 0.66, "learning_rate": 4.315491514123274e-05, "loss": 0.4206, "step": 3780 }, { "epoch": 0.66, "learning_rate": 4.314999618890559e-05, "loss": 0.3986, "step": 3781 }, { "epoch": 0.66, "learning_rate": 4.314507578713215e-05, "loss": 0.4106, "step": 3782 }, { "epoch": 0.66, "learning_rate": 4.314015393636815e-05, "loss": 0.4108, "step": 3783 }, { "epoch": 0.66, "learning_rate": 4.313523063706943e-05, "loss": 0.4159, "step": 3784 }, { "epoch": 0.66, "learning_rate": 4.3130305889691955e-05, "loss": 0.3873, "step": 3785 }, { "epoch": 0.66, "learning_rate": 4.312537969469185e-05, "loss": 0.4029, "step": 3786 }, { "epoch": 0.66, "learning_rate": 4.312045205252537e-05, "loss": 0.4209, "step": 3787 }, { "epoch": 0.66, "learning_rate": 4.311552296364887e-05, "loss": 0.4148, "step": 3788 }, { "epoch": 0.66, "learning_rate": 4.311059242851888e-05, "loss": 0.4143, "step": 3789 }, { "epoch": 0.66, "learning_rate": 4.310566044759204e-05, "loss": 0.396, "step": 3790 }, { "epoch": 0.66, "learning_rate": 4.3100727021325145e-05, "loss": 0.429, "step": 3791 }, { "epoch": 0.66, "learning_rate": 4.3095792150175106e-05, "loss": 0.4196, "step": 3792 }, { "epoch": 0.66, "learning_rate": 4.3090855834598966e-05, "loss": 0.4246, "step": 3793 }, { "epoch": 0.66, "learning_rate": 4.308591807505392e-05, "loss": 0.4741, "step": 3794 }, { "epoch": 0.66, "learning_rate": 4.308097887199727e-05, "loss": 0.5056, "step": 3795 }, { "epoch": 0.66, "learning_rate": 4.3076038225886485e-05, "loss": 0.424, "step": 3796 }, { "epoch": 0.66, "learning_rate": 4.307109613717914e-05, "loss": 0.4177, "step": 3797 }, { "epoch": 0.66, "learning_rate": 4.306615260633296e-05, "loss": 0.4265, "step": 3798 }, { "epoch": 0.66, "learning_rate": 4.3061207633805785e-05, "loss": 0.4142, "step": 3799 }, { "epoch": 0.66, "learning_rate": 4.305626122005561e-05, "loss": 0.3931, "step": 3800 }, { "epoch": 0.66, "learning_rate": 4.305131336554055e-05, "loss": 0.3976, "step": 3801 }, { "epoch": 0.66, "learning_rate": 4.3046364070718865e-05, "loss": 0.4102, "step": 3802 }, { "epoch": 0.66, "learning_rate": 4.304141333604894e-05, "loss": 0.4152, "step": 3803 }, { "epoch": 0.66, "learning_rate": 4.303646116198928e-05, "loss": 0.3998, "step": 3804 }, { "epoch": 0.66, "learning_rate": 4.3031507548998545e-05, "loss": 0.4083, "step": 3805 }, { "epoch": 0.66, "learning_rate": 4.3026552497535526e-05, "loss": 0.4117, "step": 3806 }, { "epoch": 0.66, "learning_rate": 4.3021596008059136e-05, "loss": 0.4086, "step": 3807 }, { "epoch": 0.66, "learning_rate": 4.301663808102844e-05, "loss": 0.4265, "step": 3808 }, { "epoch": 0.66, "learning_rate": 4.3011678716902605e-05, "loss": 0.4177, "step": 3809 }, { "epoch": 0.66, "learning_rate": 4.300671791614095e-05, "loss": 0.4213, "step": 3810 }, { "epoch": 0.66, "learning_rate": 4.3001755679202946e-05, "loss": 0.4113, "step": 3811 }, { "epoch": 0.66, "learning_rate": 4.299679200654817e-05, "loss": 0.4161, "step": 3812 }, { "epoch": 0.66, "learning_rate": 4.299182689863633e-05, "loss": 0.3997, "step": 3813 }, { "epoch": 0.66, "learning_rate": 4.298686035592728e-05, "loss": 0.4056, "step": 3814 }, { "epoch": 0.67, "learning_rate": 4.2981892378881e-05, "loss": 0.39, "step": 3815 }, { "epoch": 0.67, "learning_rate": 4.297692296795762e-05, "loss": 0.3999, "step": 3816 }, { "epoch": 0.67, "learning_rate": 4.2971952123617374e-05, "loss": 0.4291, "step": 3817 }, { "epoch": 0.67, "learning_rate": 4.2966979846320646e-05, "loss": 0.401, "step": 3818 }, { "epoch": 0.67, "learning_rate": 4.2962006136527965e-05, "loss": 0.4239, "step": 3819 }, { "epoch": 0.67, "learning_rate": 4.295703099469996e-05, "loss": 0.4115, "step": 3820 }, { "epoch": 0.67, "learning_rate": 4.295205442129742e-05, "loss": 0.4062, "step": 3821 }, { "epoch": 0.67, "learning_rate": 4.2947076416781256e-05, "loss": 0.4188, "step": 3822 }, { "epoch": 0.67, "learning_rate": 4.2942096981612516e-05, "loss": 0.4263, "step": 3823 }, { "epoch": 0.67, "learning_rate": 4.293711611625237e-05, "loss": 0.4226, "step": 3824 }, { "epoch": 0.67, "learning_rate": 4.293213382116213e-05, "loss": 0.4039, "step": 3825 }, { "epoch": 0.67, "learning_rate": 4.2927150096803236e-05, "loss": 0.4178, "step": 3826 }, { "epoch": 0.67, "learning_rate": 4.292216494363727e-05, "loss": 0.4114, "step": 3827 }, { "epoch": 0.67, "learning_rate": 4.2917178362125934e-05, "loss": 0.3995, "step": 3828 }, { "epoch": 0.67, "learning_rate": 4.291219035273107e-05, "loss": 0.4327, "step": 3829 }, { "epoch": 0.67, "learning_rate": 4.2907200915914644e-05, "loss": 0.4046, "step": 3830 }, { "epoch": 0.67, "learning_rate": 4.290221005213877e-05, "loss": 0.4067, "step": 3831 }, { "epoch": 0.67, "learning_rate": 4.289721776186567e-05, "loss": 0.4224, "step": 3832 }, { "epoch": 0.67, "learning_rate": 4.289222404555773e-05, "loss": 0.408, "step": 3833 }, { "epoch": 0.67, "learning_rate": 4.288722890367742e-05, "loss": 0.4012, "step": 3834 }, { "epoch": 0.67, "learning_rate": 4.28822323366874e-05, "loss": 0.4021, "step": 3835 }, { "epoch": 0.67, "learning_rate": 4.2877234345050434e-05, "loss": 0.4108, "step": 3836 }, { "epoch": 0.67, "learning_rate": 4.2872234929229396e-05, "loss": 0.4043, "step": 3837 }, { "epoch": 0.67, "learning_rate": 4.286723408968734e-05, "loss": 0.4209, "step": 3838 }, { "epoch": 0.67, "learning_rate": 4.286223182688741e-05, "loss": 0.4173, "step": 3839 }, { "epoch": 0.67, "learning_rate": 4.28572281412929e-05, "loss": 0.4135, "step": 3840 }, { "epoch": 0.67, "learning_rate": 4.285222303336723e-05, "loss": 0.4254, "step": 3841 }, { "epoch": 0.67, "learning_rate": 4.284721650357396e-05, "loss": 0.4169, "step": 3842 }, { "epoch": 0.67, "learning_rate": 4.2842208552376765e-05, "loss": 0.3845, "step": 3843 }, { "epoch": 0.67, "learning_rate": 4.283719918023949e-05, "loss": 0.4066, "step": 3844 }, { "epoch": 0.67, "learning_rate": 4.283218838762607e-05, "loss": 0.4104, "step": 3845 }, { "epoch": 0.67, "learning_rate": 4.2827176175000574e-05, "loss": 0.4101, "step": 3846 }, { "epoch": 0.67, "learning_rate": 4.2822162542827225e-05, "loss": 0.4125, "step": 3847 }, { "epoch": 0.67, "learning_rate": 4.281714749157038e-05, "loss": 0.3955, "step": 3848 }, { "epoch": 0.67, "learning_rate": 4.2812131021694494e-05, "loss": 0.4112, "step": 3849 }, { "epoch": 0.67, "learning_rate": 4.28071131336642e-05, "loss": 0.3985, "step": 3850 }, { "epoch": 0.67, "learning_rate": 4.2802093827944215e-05, "loss": 0.4115, "step": 3851 }, { "epoch": 0.67, "learning_rate": 4.279707310499941e-05, "loss": 0.3998, "step": 3852 }, { "epoch": 0.67, "learning_rate": 4.2792050965294796e-05, "loss": 0.4227, "step": 3853 }, { "epoch": 0.67, "learning_rate": 4.278702740929549e-05, "loss": 0.4072, "step": 3854 }, { "epoch": 0.67, "learning_rate": 4.278200243746678e-05, "loss": 0.4033, "step": 3855 }, { "epoch": 0.67, "learning_rate": 4.277697605027404e-05, "loss": 0.4024, "step": 3856 }, { "epoch": 0.67, "learning_rate": 4.277194824818281e-05, "loss": 0.4222, "step": 3857 }, { "epoch": 0.67, "learning_rate": 4.2766919031658735e-05, "loss": 0.4225, "step": 3858 }, { "epoch": 0.67, "learning_rate": 4.276188840116761e-05, "loss": 0.41, "step": 3859 }, { "epoch": 0.67, "learning_rate": 4.275685635717535e-05, "loss": 0.4062, "step": 3860 }, { "epoch": 0.67, "learning_rate": 4.275182290014801e-05, "loss": 0.4114, "step": 3861 }, { "epoch": 0.67, "learning_rate": 4.274678803055176e-05, "loss": 0.4204, "step": 3862 }, { "epoch": 0.67, "learning_rate": 4.274175174885293e-05, "loss": 0.4043, "step": 3863 }, { "epoch": 0.67, "learning_rate": 4.2736714055517946e-05, "loss": 0.4126, "step": 3864 }, { "epoch": 0.67, "learning_rate": 4.273167495101338e-05, "loss": 0.3815, "step": 3865 }, { "epoch": 0.67, "learning_rate": 4.272663443580596e-05, "loss": 0.4108, "step": 3866 }, { "epoch": 0.67, "learning_rate": 4.272159251036248e-05, "loss": 0.4375, "step": 3867 }, { "epoch": 0.67, "learning_rate": 4.271654917514994e-05, "loss": 0.4228, "step": 3868 }, { "epoch": 0.67, "learning_rate": 4.271150443063542e-05, "loss": 0.4047, "step": 3869 }, { "epoch": 0.67, "learning_rate": 4.270645827728616e-05, "loss": 0.4015, "step": 3870 }, { "epoch": 0.67, "learning_rate": 4.270141071556949e-05, "loss": 0.4192, "step": 3871 }, { "epoch": 0.68, "learning_rate": 4.269636174595292e-05, "loss": 0.4318, "step": 3872 }, { "epoch": 0.68, "learning_rate": 4.269131136890406e-05, "loss": 0.4238, "step": 3873 }, { "epoch": 0.68, "learning_rate": 4.2686259584890656e-05, "loss": 0.4231, "step": 3874 }, { "epoch": 0.68, "learning_rate": 4.2681206394380584e-05, "loss": 0.4124, "step": 3875 }, { "epoch": 0.68, "learning_rate": 4.267615179784186e-05, "loss": 0.4357, "step": 3876 }, { "epoch": 0.68, "learning_rate": 4.2671095795742614e-05, "loss": 0.3963, "step": 3877 }, { "epoch": 0.68, "learning_rate": 4.266603838855113e-05, "loss": 0.4166, "step": 3878 }, { "epoch": 0.68, "learning_rate": 4.266097957673578e-05, "loss": 0.4091, "step": 3879 }, { "epoch": 0.68, "learning_rate": 4.265591936076512e-05, "loss": 0.4195, "step": 3880 }, { "epoch": 0.68, "learning_rate": 4.26508577411078e-05, "loss": 0.42, "step": 3881 }, { "epoch": 0.68, "learning_rate": 4.2645794718232603e-05, "loss": 0.4232, "step": 3882 }, { "epoch": 0.68, "learning_rate": 4.2640730292608455e-05, "loss": 0.41, "step": 3883 }, { "epoch": 0.68, "learning_rate": 4.2635664464704397e-05, "loss": 0.406, "step": 3884 }, { "epoch": 0.68, "learning_rate": 4.263059723498961e-05, "loss": 0.4197, "step": 3885 }, { "epoch": 0.68, "learning_rate": 4.262552860393341e-05, "loss": 0.4165, "step": 3886 }, { "epoch": 0.68, "learning_rate": 4.262045857200522e-05, "loss": 0.4204, "step": 3887 }, { "epoch": 0.68, "learning_rate": 4.261538713967463e-05, "loss": 0.4169, "step": 3888 }, { "epoch": 0.68, "learning_rate": 4.261031430741132e-05, "loss": 0.4208, "step": 3889 }, { "epoch": 0.68, "learning_rate": 4.260524007568514e-05, "loss": 0.4129, "step": 3890 }, { "epoch": 0.68, "learning_rate": 4.260016444496601e-05, "loss": 0.4126, "step": 3891 }, { "epoch": 0.68, "learning_rate": 4.259508741572405e-05, "loss": 0.3931, "step": 3892 }, { "epoch": 0.68, "learning_rate": 4.259000898842946e-05, "loss": 0.3991, "step": 3893 }, { "epoch": 0.68, "learning_rate": 4.258492916355259e-05, "loss": 0.4046, "step": 3894 }, { "epoch": 0.68, "learning_rate": 4.257984794156391e-05, "loss": 0.4133, "step": 3895 }, { "epoch": 0.68, "learning_rate": 4.257476532293403e-05, "loss": 0.4063, "step": 3896 }, { "epoch": 0.68, "learning_rate": 4.256968130813369e-05, "loss": 0.4146, "step": 3897 }, { "epoch": 0.68, "learning_rate": 4.2564595897633735e-05, "loss": 0.4148, "step": 3898 }, { "epoch": 0.68, "learning_rate": 4.255950909190517e-05, "loss": 0.4108, "step": 3899 }, { "epoch": 0.68, "learning_rate": 4.255442089141912e-05, "loss": 0.3925, "step": 3900 }, { "epoch": 0.68, "learning_rate": 4.254933129664683e-05, "loss": 0.4024, "step": 3901 }, { "epoch": 0.68, "learning_rate": 4.254424030805968e-05, "loss": 0.4127, "step": 3902 }, { "epoch": 0.68, "learning_rate": 4.253914792612918e-05, "loss": 0.411, "step": 3903 }, { "epoch": 0.68, "learning_rate": 4.253405415132696e-05, "loss": 0.4098, "step": 3904 }, { "epoch": 0.68, "learning_rate": 4.25289589841248e-05, "loss": 0.4224, "step": 3905 }, { "epoch": 0.68, "learning_rate": 4.2523862424994584e-05, "loss": 0.4123, "step": 3906 }, { "epoch": 0.68, "learning_rate": 4.251876447440836e-05, "loss": 0.4043, "step": 3907 }, { "epoch": 0.68, "learning_rate": 4.2513665132838256e-05, "loss": 0.4119, "step": 3908 }, { "epoch": 0.68, "learning_rate": 4.250856440075656e-05, "loss": 0.3974, "step": 3909 }, { "epoch": 0.68, "learning_rate": 4.2503462278635696e-05, "loss": 0.4159, "step": 3910 }, { "epoch": 0.68, "learning_rate": 4.2498358766948196e-05, "loss": 0.4137, "step": 3911 }, { "epoch": 0.68, "learning_rate": 4.2493253866166724e-05, "loss": 0.4053, "step": 3912 }, { "epoch": 0.68, "learning_rate": 4.248814757676408e-05, "loss": 0.4156, "step": 3913 }, { "epoch": 0.68, "learning_rate": 4.248303989921319e-05, "loss": 0.3966, "step": 3914 }, { "epoch": 0.68, "learning_rate": 4.247793083398712e-05, "loss": 0.4062, "step": 3915 }, { "epoch": 0.68, "learning_rate": 4.247282038155904e-05, "loss": 0.3937, "step": 3916 }, { "epoch": 0.68, "learning_rate": 4.2467708542402266e-05, "loss": 0.3977, "step": 3917 }, { "epoch": 0.68, "learning_rate": 4.246259531699024e-05, "loss": 0.3996, "step": 3918 }, { "epoch": 0.68, "learning_rate": 4.245748070579653e-05, "loss": 0.4233, "step": 3919 }, { "epoch": 0.68, "learning_rate": 4.2452364709294814e-05, "loss": 0.4227, "step": 3920 }, { "epoch": 0.68, "learning_rate": 4.244724732795896e-05, "loss": 0.4103, "step": 3921 }, { "epoch": 0.68, "learning_rate": 4.2442128562262884e-05, "loss": 0.4007, "step": 3922 }, { "epoch": 0.68, "learning_rate": 4.243700841268068e-05, "loss": 0.4137, "step": 3923 }, { "epoch": 0.68, "learning_rate": 4.243188687968655e-05, "loss": 0.3965, "step": 3924 }, { "epoch": 0.68, "learning_rate": 4.242676396375485e-05, "loss": 0.4129, "step": 3925 }, { "epoch": 0.68, "learning_rate": 4.242163966536003e-05, "loss": 0.4175, "step": 3926 }, { "epoch": 0.68, "learning_rate": 4.24165139849767e-05, "loss": 0.4042, "step": 3927 }, { "epoch": 0.68, "learning_rate": 4.241138692307956e-05, "loss": 0.4105, "step": 3928 }, { "epoch": 0.69, "learning_rate": 4.240625848014347e-05, "loss": 0.394, "step": 3929 }, { "epoch": 0.69, "learning_rate": 4.240112865664342e-05, "loss": 0.4021, "step": 3930 }, { "epoch": 0.69, "learning_rate": 4.2395997453054495e-05, "loss": 0.4044, "step": 3931 }, { "epoch": 0.69, "learning_rate": 4.239086486985194e-05, "loss": 0.3992, "step": 3932 }, { "epoch": 0.69, "learning_rate": 4.238573090751112e-05, "loss": 0.4142, "step": 3933 }, { "epoch": 0.69, "learning_rate": 4.238059556650752e-05, "loss": 0.4289, "step": 3934 }, { "epoch": 0.69, "learning_rate": 4.237545884731675e-05, "loss": 0.4148, "step": 3935 }, { "epoch": 0.69, "learning_rate": 4.2370320750414573e-05, "loss": 0.4223, "step": 3936 }, { "epoch": 0.69, "learning_rate": 4.236518127627683e-05, "loss": 0.4105, "step": 3937 }, { "epoch": 0.69, "learning_rate": 4.236004042537956e-05, "loss": 0.4019, "step": 3938 }, { "epoch": 0.69, "learning_rate": 4.235489819819886e-05, "loss": 0.4168, "step": 3939 }, { "epoch": 0.69, "learning_rate": 4.234975459521098e-05, "loss": 0.4135, "step": 3940 }, { "epoch": 0.69, "learning_rate": 4.234460961689233e-05, "loss": 0.4148, "step": 3941 }, { "epoch": 0.69, "learning_rate": 4.23394632637194e-05, "loss": 0.4169, "step": 3942 }, { "epoch": 0.69, "learning_rate": 4.233431553616884e-05, "loss": 0.3903, "step": 3943 }, { "epoch": 0.69, "learning_rate": 4.23291664347174e-05, "loss": 0.4025, "step": 3944 }, { "epoch": 0.69, "learning_rate": 4.232401595984197e-05, "loss": 0.4145, "step": 3945 }, { "epoch": 0.69, "learning_rate": 4.231886411201959e-05, "loss": 0.4208, "step": 3946 }, { "epoch": 0.69, "learning_rate": 4.231371089172739e-05, "loss": 0.4108, "step": 3947 }, { "epoch": 0.69, "learning_rate": 4.230855629944263e-05, "loss": 0.4255, "step": 3948 }, { "epoch": 0.69, "learning_rate": 4.2303400335642733e-05, "loss": 0.431, "step": 3949 }, { "epoch": 0.69, "learning_rate": 4.229824300080521e-05, "loss": 0.3982, "step": 3950 }, { "epoch": 0.69, "learning_rate": 4.229308429540773e-05, "loss": 0.3963, "step": 3951 }, { "epoch": 0.69, "learning_rate": 4.228792421992806e-05, "loss": 0.4171, "step": 3952 }, { "epoch": 0.69, "learning_rate": 4.228276277484411e-05, "loss": 0.4172, "step": 3953 }, { "epoch": 0.69, "learning_rate": 4.227759996063392e-05, "loss": 0.4071, "step": 3954 }, { "epoch": 0.69, "learning_rate": 4.227243577777565e-05, "loss": 0.4189, "step": 3955 }, { "epoch": 0.69, "learning_rate": 4.226727022674758e-05, "loss": 0.4197, "step": 3956 }, { "epoch": 0.69, "learning_rate": 4.226210330802814e-05, "loss": 0.4284, "step": 3957 }, { "epoch": 0.69, "learning_rate": 4.225693502209585e-05, "loss": 0.4102, "step": 3958 }, { "epoch": 0.69, "learning_rate": 4.22517653694294e-05, "loss": 0.3914, "step": 3959 }, { "epoch": 0.69, "learning_rate": 4.224659435050757e-05, "loss": 0.4189, "step": 3960 }, { "epoch": 0.69, "learning_rate": 4.22414219658093e-05, "loss": 0.4189, "step": 3961 }, { "epoch": 0.69, "learning_rate": 4.22362482158136e-05, "loss": 0.4136, "step": 3962 }, { "epoch": 0.69, "learning_rate": 4.223107310099968e-05, "loss": 0.3912, "step": 3963 }, { "epoch": 0.69, "learning_rate": 4.222589662184683e-05, "loss": 0.3915, "step": 3964 }, { "epoch": 0.69, "learning_rate": 4.222071877883447e-05, "loss": 0.4078, "step": 3965 }, { "epoch": 0.69, "learning_rate": 4.221553957244216e-05, "loss": 0.4191, "step": 3966 }, { "epoch": 0.69, "learning_rate": 4.2210359003149586e-05, "loss": 0.4028, "step": 3967 }, { "epoch": 0.69, "learning_rate": 4.2205177071436534e-05, "loss": 0.3832, "step": 3968 }, { "epoch": 0.69, "learning_rate": 4.2199993777782945e-05, "loss": 0.4187, "step": 3969 }, { "epoch": 0.69, "learning_rate": 4.219480912266889e-05, "loss": 0.408, "step": 3970 }, { "epoch": 0.69, "learning_rate": 4.218962310657454e-05, "loss": 0.4057, "step": 3971 }, { "epoch": 0.69, "learning_rate": 4.218443572998021e-05, "loss": 0.402, "step": 3972 }, { "epoch": 0.69, "learning_rate": 4.217924699336632e-05, "loss": 0.4058, "step": 3973 }, { "epoch": 0.69, "learning_rate": 4.217405689721346e-05, "loss": 0.4244, "step": 3974 }, { "epoch": 0.69, "learning_rate": 4.21688654420023e-05, "loss": 0.4154, "step": 3975 }, { "epoch": 0.69, "learning_rate": 4.2163672628213655e-05, "loss": 0.3902, "step": 3976 }, { "epoch": 0.69, "learning_rate": 4.215847845632847e-05, "loss": 0.4129, "step": 3977 }, { "epoch": 0.69, "learning_rate": 4.21532829268278e-05, "loss": 0.4217, "step": 3978 }, { "epoch": 0.69, "learning_rate": 4.2148086040192864e-05, "loss": 0.41, "step": 3979 }, { "epoch": 0.69, "learning_rate": 4.2142887796904945e-05, "loss": 0.3999, "step": 3980 }, { "epoch": 0.69, "learning_rate": 4.213768819744549e-05, "loss": 0.3949, "step": 3981 }, { "epoch": 0.69, "learning_rate": 4.2132487242296087e-05, "loss": 0.4003, "step": 3982 }, { "epoch": 0.69, "learning_rate": 4.212728493193843e-05, "loss": 0.3908, "step": 3983 }, { "epoch": 0.69, "learning_rate": 4.212208126685431e-05, "loss": 0.4096, "step": 3984 }, { "epoch": 0.69, "learning_rate": 4.211687624752569e-05, "loss": 0.429, "step": 3985 }, { "epoch": 0.69, "learning_rate": 4.211166987443465e-05, "loss": 0.3965, "step": 3986 }, { "epoch": 0.7, "learning_rate": 4.2106462148063375e-05, "loss": 0.398, "step": 3987 }, { "epoch": 0.7, "learning_rate": 4.210125306889418e-05, "loss": 0.4008, "step": 3988 }, { "epoch": 0.7, "learning_rate": 4.209604263740951e-05, "loss": 0.4162, "step": 3989 }, { "epoch": 0.7, "learning_rate": 4.209083085409195e-05, "loss": 0.4022, "step": 3990 }, { "epoch": 0.7, "learning_rate": 4.20856177194242e-05, "loss": 0.396, "step": 3991 }, { "epoch": 0.7, "learning_rate": 4.2080403233889066e-05, "loss": 0.4124, "step": 3992 }, { "epoch": 0.7, "learning_rate": 4.20751873979695e-05, "loss": 0.4114, "step": 3993 }, { "epoch": 0.7, "learning_rate": 4.206997021214857e-05, "loss": 0.4082, "step": 3994 }, { "epoch": 0.7, "learning_rate": 4.206475167690947e-05, "loss": 0.4188, "step": 3995 }, { "epoch": 0.7, "learning_rate": 4.205953179273555e-05, "loss": 0.43, "step": 3996 }, { "epoch": 0.7, "learning_rate": 4.205431056011023e-05, "loss": 0.4133, "step": 3997 }, { "epoch": 0.7, "learning_rate": 4.204908797951708e-05, "loss": 0.4106, "step": 3998 }, { "epoch": 0.7, "learning_rate": 4.204386405143981e-05, "loss": 0.4197, "step": 3999 }, { "epoch": 0.7, "learning_rate": 4.2038638776362236e-05, "loss": 0.4009, "step": 4000 }, { "epoch": 0.7, "learning_rate": 4.2033412154768304e-05, "loss": 0.4245, "step": 4001 }, { "epoch": 0.7, "learning_rate": 4.2028184187142085e-05, "loss": 0.4238, "step": 4002 }, { "epoch": 0.7, "learning_rate": 4.202295487396777e-05, "loss": 0.408, "step": 4003 }, { "epoch": 0.7, "learning_rate": 4.201772421572969e-05, "loss": 0.4224, "step": 4004 }, { "epoch": 0.7, "learning_rate": 4.2012492212912285e-05, "loss": 0.401, "step": 4005 }, { "epoch": 0.7, "learning_rate": 4.200725886600012e-05, "loss": 0.4191, "step": 4006 }, { "epoch": 0.7, "learning_rate": 4.200202417547789e-05, "loss": 0.4397, "step": 4007 }, { "epoch": 0.7, "learning_rate": 4.199678814183042e-05, "loss": 0.3903, "step": 4008 }, { "epoch": 0.7, "learning_rate": 4.1991550765542634e-05, "loss": 0.3785, "step": 4009 }, { "epoch": 0.7, "learning_rate": 4.198631204709963e-05, "loss": 0.4219, "step": 4010 }, { "epoch": 0.7, "learning_rate": 4.198107198698657e-05, "loss": 0.4252, "step": 4011 }, { "epoch": 0.7, "learning_rate": 4.1975830585688786e-05, "loss": 0.4099, "step": 4012 }, { "epoch": 0.7, "learning_rate": 4.197058784369171e-05, "loss": 0.3962, "step": 4013 }, { "epoch": 0.7, "learning_rate": 4.196534376148091e-05, "loss": 0.4287, "step": 4014 }, { "epoch": 0.7, "learning_rate": 4.196009833954207e-05, "loss": 0.4128, "step": 4015 }, { "epoch": 0.7, "learning_rate": 4.195485157836101e-05, "loss": 0.4031, "step": 4016 }, { "epoch": 0.7, "learning_rate": 4.194960347842366e-05, "loss": 0.4114, "step": 4017 }, { "epoch": 0.7, "learning_rate": 4.194435404021608e-05, "loss": 0.3973, "step": 4018 }, { "epoch": 0.7, "learning_rate": 4.193910326422445e-05, "loss": 0.4049, "step": 4019 }, { "epoch": 0.7, "learning_rate": 4.1933851150935086e-05, "loss": 0.424, "step": 4020 }, { "epoch": 0.7, "learning_rate": 4.192859770083442e-05, "loss": 0.4098, "step": 4021 }, { "epoch": 0.7, "learning_rate": 4.1923342914409004e-05, "loss": 0.4076, "step": 4022 }, { "epoch": 0.7, "learning_rate": 4.191808679214551e-05, "loss": 0.4077, "step": 4023 }, { "epoch": 0.7, "learning_rate": 4.191282933453076e-05, "loss": 0.4241, "step": 4024 }, { "epoch": 0.7, "learning_rate": 4.190757054205166e-05, "loss": 0.4131, "step": 4025 }, { "epoch": 0.7, "learning_rate": 4.190231041519528e-05, "loss": 0.411, "step": 4026 }, { "epoch": 0.7, "learning_rate": 4.1897048954448767e-05, "loss": 0.4022, "step": 4027 }, { "epoch": 0.7, "learning_rate": 4.189178616029945e-05, "loss": 0.3983, "step": 4028 }, { "epoch": 0.7, "learning_rate": 4.188652203323473e-05, "loss": 0.41, "step": 4029 }, { "epoch": 0.7, "learning_rate": 4.188125657374216e-05, "loss": 0.4375, "step": 4030 }, { "epoch": 0.7, "learning_rate": 4.187598978230941e-05, "loss": 0.4198, "step": 4031 }, { "epoch": 0.7, "learning_rate": 4.187072165942425e-05, "loss": 0.4366, "step": 4032 }, { "epoch": 0.7, "learning_rate": 4.186545220557463e-05, "loss": 0.3825, "step": 4033 }, { "epoch": 0.7, "learning_rate": 4.1860181421248564e-05, "loss": 0.4112, "step": 4034 }, { "epoch": 0.7, "learning_rate": 4.1854909306934216e-05, "loss": 0.3849, "step": 4035 }, { "epoch": 0.7, "learning_rate": 4.184963586311987e-05, "loss": 0.4015, "step": 4036 }, { "epoch": 0.7, "learning_rate": 4.184436109029393e-05, "loss": 0.3985, "step": 4037 }, { "epoch": 0.7, "learning_rate": 4.1839084988944936e-05, "loss": 0.3997, "step": 4038 }, { "epoch": 0.7, "learning_rate": 4.183380755956154e-05, "loss": 0.4161, "step": 4039 }, { "epoch": 0.7, "learning_rate": 4.1828528802632515e-05, "loss": 0.4217, "step": 4040 }, { "epoch": 0.7, "learning_rate": 4.182324871864676e-05, "loss": 0.4116, "step": 4041 }, { "epoch": 0.7, "learning_rate": 4.1817967308093294e-05, "loss": 0.3979, "step": 4042 }, { "epoch": 0.7, "learning_rate": 4.181268457146128e-05, "loss": 0.4035, "step": 4043 }, { "epoch": 0.71, "learning_rate": 4.180740050923997e-05, "loss": 0.4033, "step": 4044 }, { "epoch": 0.71, "learning_rate": 4.180211512191875e-05, "loss": 0.4054, "step": 4045 }, { "epoch": 0.71, "learning_rate": 4.1796828409987144e-05, "loss": 0.4187, "step": 4046 }, { "epoch": 0.71, "learning_rate": 4.179154037393478e-05, "loss": 0.4093, "step": 4047 }, { "epoch": 0.71, "learning_rate": 4.1786251014251434e-05, "loss": 0.416, "step": 4048 }, { "epoch": 0.71, "learning_rate": 4.1780960331426966e-05, "loss": 0.4008, "step": 4049 }, { "epoch": 0.71, "learning_rate": 4.1775668325951397e-05, "loss": 0.4046, "step": 4050 }, { "epoch": 0.71, "learning_rate": 4.1770374998314846e-05, "loss": 0.4153, "step": 4051 }, { "epoch": 0.71, "learning_rate": 4.1765080349007555e-05, "loss": 0.4294, "step": 4052 }, { "epoch": 0.71, "learning_rate": 4.1759784378519904e-05, "loss": 0.4129, "step": 4053 }, { "epoch": 0.71, "learning_rate": 4.1754487087342384e-05, "loss": 0.4195, "step": 4054 }, { "epoch": 0.71, "learning_rate": 4.174918847596562e-05, "loss": 0.4143, "step": 4055 }, { "epoch": 0.71, "learning_rate": 4.1743888544880336e-05, "loss": 0.4088, "step": 4056 }, { "epoch": 0.71, "learning_rate": 4.17385872945774e-05, "loss": 0.3872, "step": 4057 }, { "epoch": 0.71, "learning_rate": 4.17332847255478e-05, "loss": 0.4139, "step": 4058 }, { "epoch": 0.71, "learning_rate": 4.172798083828263e-05, "loss": 0.4112, "step": 4059 }, { "epoch": 0.71, "learning_rate": 4.172267563327313e-05, "loss": 0.4069, "step": 4060 }, { "epoch": 0.71, "learning_rate": 4.1717369111010624e-05, "loss": 0.4189, "step": 4061 }, { "epoch": 0.71, "learning_rate": 4.1712061271986606e-05, "loss": 0.4091, "step": 4062 }, { "epoch": 0.71, "learning_rate": 4.1706752116692674e-05, "loss": 0.3967, "step": 4063 }, { "epoch": 0.71, "learning_rate": 4.170144164562051e-05, "loss": 0.3899, "step": 4064 }, { "epoch": 0.71, "learning_rate": 4.169612985926199e-05, "loss": 0.4018, "step": 4065 }, { "epoch": 0.71, "learning_rate": 4.1690816758109044e-05, "loss": 0.4235, "step": 4066 }, { "epoch": 0.71, "learning_rate": 4.1685502342653773e-05, "loss": 0.4062, "step": 4067 }, { "epoch": 0.71, "learning_rate": 4.1680186613388365e-05, "loss": 0.3903, "step": 4068 }, { "epoch": 0.71, "learning_rate": 4.167486957080514e-05, "loss": 0.4088, "step": 4069 }, { "epoch": 0.71, "learning_rate": 4.166955121539656e-05, "loss": 0.4099, "step": 4070 }, { "epoch": 0.71, "learning_rate": 4.1664231547655174e-05, "loss": 0.3736, "step": 4071 }, { "epoch": 0.71, "learning_rate": 4.165891056807368e-05, "loss": 0.4177, "step": 4072 }, { "epoch": 0.71, "learning_rate": 4.16535882771449e-05, "loss": 0.406, "step": 4073 }, { "epoch": 0.71, "learning_rate": 4.1648264675361745e-05, "loss": 0.414, "step": 4074 }, { "epoch": 0.71, "learning_rate": 4.164293976321727e-05, "loss": 0.3996, "step": 4075 }, { "epoch": 0.71, "learning_rate": 4.163761354120466e-05, "loss": 0.4077, "step": 4076 }, { "epoch": 0.71, "learning_rate": 4.1632286009817206e-05, "loss": 0.4151, "step": 4077 }, { "epoch": 0.71, "learning_rate": 4.162695716954832e-05, "loss": 0.4051, "step": 4078 }, { "epoch": 0.71, "learning_rate": 4.1621627020891555e-05, "loss": 0.4046, "step": 4079 }, { "epoch": 0.71, "learning_rate": 4.161629556434055e-05, "loss": 0.4041, "step": 4080 }, { "epoch": 0.71, "learning_rate": 4.161096280038909e-05, "loss": 0.414, "step": 4081 }, { "epoch": 0.71, "learning_rate": 4.160562872953109e-05, "loss": 0.4062, "step": 4082 }, { "epoch": 0.71, "learning_rate": 4.160029335226055e-05, "loss": 0.4273, "step": 4083 }, { "epoch": 0.71, "learning_rate": 4.159495666907164e-05, "loss": 0.4109, "step": 4084 }, { "epoch": 0.71, "learning_rate": 4.158961868045861e-05, "loss": 0.4215, "step": 4085 }, { "epoch": 0.71, "learning_rate": 4.1584279386915845e-05, "loss": 0.4051, "step": 4086 }, { "epoch": 0.71, "learning_rate": 4.1578938788937844e-05, "loss": 0.4033, "step": 4087 }, { "epoch": 0.71, "learning_rate": 4.157359688701925e-05, "loss": 0.3986, "step": 4088 }, { "epoch": 0.71, "learning_rate": 4.1568253681654804e-05, "loss": 0.4057, "step": 4089 }, { "epoch": 0.71, "learning_rate": 4.156290917333937e-05, "loss": 0.3993, "step": 4090 }, { "epoch": 0.71, "learning_rate": 4.155756336256794e-05, "loss": 0.4247, "step": 4091 }, { "epoch": 0.71, "learning_rate": 4.155221624983563e-05, "loss": 0.4179, "step": 4092 }, { "epoch": 0.71, "learning_rate": 4.1546867835637654e-05, "loss": 0.4163, "step": 4093 }, { "epoch": 0.71, "learning_rate": 4.154151812046939e-05, "loss": 0.4269, "step": 4094 }, { "epoch": 0.71, "learning_rate": 4.153616710482627e-05, "loss": 0.401, "step": 4095 }, { "epoch": 0.71, "learning_rate": 4.153081478920394e-05, "loss": 0.4229, "step": 4096 }, { "epoch": 0.71, "learning_rate": 4.1525461174098055e-05, "loss": 0.4204, "step": 4097 }, { "epoch": 0.71, "learning_rate": 4.1520106260004473e-05, "loss": 0.4001, "step": 4098 }, { "epoch": 0.71, "learning_rate": 4.151475004741917e-05, "loss": 0.4174, "step": 4099 }, { "epoch": 0.71, "learning_rate": 4.1509392536838177e-05, "loss": 0.4123, "step": 4100 }, { "epoch": 0.72, "learning_rate": 4.150403372875771e-05, "loss": 0.4069, "step": 4101 }, { "epoch": 0.72, "learning_rate": 4.1498673623674075e-05, "loss": 0.4023, "step": 4102 }, { "epoch": 0.72, "learning_rate": 4.149331222208372e-05, "loss": 0.4118, "step": 4103 }, { "epoch": 0.72, "learning_rate": 4.148794952448319e-05, "loss": 0.4011, "step": 4104 }, { "epoch": 0.72, "learning_rate": 4.148258553136915e-05, "loss": 0.4242, "step": 4105 }, { "epoch": 0.72, "learning_rate": 4.14772202432384e-05, "loss": 0.4125, "step": 4106 }, { "epoch": 0.72, "learning_rate": 4.147185366058786e-05, "loss": 0.415, "step": 4107 }, { "epoch": 0.72, "learning_rate": 4.146648578391455e-05, "loss": 0.4153, "step": 4108 }, { "epoch": 0.72, "learning_rate": 4.146111661371565e-05, "loss": 0.4123, "step": 4109 }, { "epoch": 0.72, "learning_rate": 4.1455746150488396e-05, "loss": 0.4211, "step": 4110 }, { "epoch": 0.72, "learning_rate": 4.145037439473021e-05, "loss": 0.3957, "step": 4111 }, { "epoch": 0.72, "learning_rate": 4.1445001346938586e-05, "loss": 0.4023, "step": 4112 }, { "epoch": 0.72, "learning_rate": 4.1439627007611176e-05, "loss": 0.3968, "step": 4113 }, { "epoch": 0.72, "learning_rate": 4.143425137724571e-05, "loss": 0.3879, "step": 4114 }, { "epoch": 0.72, "learning_rate": 4.142887445634008e-05, "loss": 0.3962, "step": 4115 }, { "epoch": 0.72, "learning_rate": 4.142349624539226e-05, "loss": 0.4066, "step": 4116 }, { "epoch": 0.72, "learning_rate": 4.1418116744900363e-05, "loss": 0.4182, "step": 4117 }, { "epoch": 0.72, "learning_rate": 4.1412735955362633e-05, "loss": 0.4271, "step": 4118 }, { "epoch": 0.72, "learning_rate": 4.140735387727741e-05, "loss": 0.4061, "step": 4119 }, { "epoch": 0.72, "learning_rate": 4.140197051114316e-05, "loss": 0.4154, "step": 4120 }, { "epoch": 0.72, "learning_rate": 4.1396585857458474e-05, "loss": 0.4198, "step": 4121 }, { "epoch": 0.72, "learning_rate": 4.1391199916722056e-05, "loss": 0.4127, "step": 4122 }, { "epoch": 0.72, "learning_rate": 4.138581268943275e-05, "loss": 0.4044, "step": 4123 }, { "epoch": 0.72, "learning_rate": 4.138042417608947e-05, "loss": 0.4005, "step": 4124 }, { "epoch": 0.72, "learning_rate": 4.137503437719131e-05, "loss": 0.3983, "step": 4125 }, { "epoch": 0.72, "learning_rate": 4.136964329323743e-05, "loss": 0.4159, "step": 4126 }, { "epoch": 0.72, "learning_rate": 4.1364250924727156e-05, "loss": 0.4067, "step": 4127 }, { "epoch": 0.72, "learning_rate": 4.135885727215989e-05, "loss": 0.4082, "step": 4128 }, { "epoch": 0.72, "learning_rate": 4.135346233603518e-05, "loss": 0.4201, "step": 4129 }, { "epoch": 0.72, "learning_rate": 4.134806611685269e-05, "loss": 0.4124, "step": 4130 }, { "epoch": 0.72, "learning_rate": 4.134266861511219e-05, "loss": 0.3898, "step": 4131 }, { "epoch": 0.72, "learning_rate": 4.133726983131359e-05, "loss": 0.3882, "step": 4132 }, { "epoch": 0.72, "learning_rate": 4.133186976595689e-05, "loss": 0.4033, "step": 4133 }, { "epoch": 0.72, "learning_rate": 4.132646841954224e-05, "loss": 0.4005, "step": 4134 }, { "epoch": 0.72, "learning_rate": 4.132106579256988e-05, "loss": 0.405, "step": 4135 }, { "epoch": 0.72, "learning_rate": 4.131566188554019e-05, "loss": 0.409, "step": 4136 }, { "epoch": 0.72, "learning_rate": 4.131025669895366e-05, "loss": 0.4001, "step": 4137 }, { "epoch": 0.72, "learning_rate": 4.130485023331089e-05, "loss": 0.395, "step": 4138 }, { "epoch": 0.72, "learning_rate": 4.129944248911262e-05, "loss": 0.408, "step": 4139 }, { "epoch": 0.72, "learning_rate": 4.129403346685968e-05, "loss": 0.4065, "step": 4140 }, { "epoch": 0.72, "learning_rate": 4.1288623167053064e-05, "loss": 0.4255, "step": 4141 }, { "epoch": 0.72, "learning_rate": 4.1283211590193816e-05, "loss": 0.4132, "step": 4142 }, { "epoch": 0.72, "learning_rate": 4.127779873678316e-05, "loss": 0.404, "step": 4143 }, { "epoch": 0.72, "learning_rate": 4.127238460732242e-05, "loss": 0.3868, "step": 4144 }, { "epoch": 0.72, "learning_rate": 4.126696920231302e-05, "loss": 0.3979, "step": 4145 }, { "epoch": 0.72, "learning_rate": 4.1261552522256514e-05, "loss": 0.4019, "step": 4146 }, { "epoch": 0.72, "learning_rate": 4.125613456765457e-05, "loss": 0.4135, "step": 4147 }, { "epoch": 0.72, "learning_rate": 4.1250715339009e-05, "loss": 0.4059, "step": 4148 }, { "epoch": 0.72, "learning_rate": 4.124529483682171e-05, "loss": 0.4134, "step": 4149 }, { "epoch": 0.72, "learning_rate": 4.123987306159471e-05, "loss": 0.4069, "step": 4150 }, { "epoch": 0.72, "learning_rate": 4.123445001383016e-05, "loss": 0.4159, "step": 4151 }, { "epoch": 0.72, "learning_rate": 4.122902569403032e-05, "loss": 0.4044, "step": 4152 }, { "epoch": 0.72, "learning_rate": 4.122360010269757e-05, "loss": 0.4042, "step": 4153 }, { "epoch": 0.72, "learning_rate": 4.1218173240334415e-05, "loss": 0.3704, "step": 4154 }, { "epoch": 0.72, "learning_rate": 4.121274510744345e-05, "loss": 0.4039, "step": 4155 }, { "epoch": 0.72, "learning_rate": 4.1207315704527435e-05, "loss": 0.4062, "step": 4156 }, { "epoch": 0.72, "learning_rate": 4.120188503208921e-05, "loss": 0.4178, "step": 4157 }, { "epoch": 0.72, "learning_rate": 4.1196453090631735e-05, "loss": 0.3945, "step": 4158 }, { "epoch": 0.73, "learning_rate": 4.119101988065811e-05, "loss": 0.4366, "step": 4159 }, { "epoch": 0.73, "learning_rate": 4.118558540267155e-05, "loss": 0.4326, "step": 4160 }, { "epoch": 0.73, "learning_rate": 4.1180149657175353e-05, "loss": 0.426, "step": 4161 }, { "epoch": 0.73, "learning_rate": 4.117471264467297e-05, "loss": 0.4089, "step": 4162 }, { "epoch": 0.73, "learning_rate": 4.116927436566796e-05, "loss": 0.4217, "step": 4163 }, { "epoch": 0.73, "learning_rate": 4.1163834820663985e-05, "loss": 0.4107, "step": 4164 }, { "epoch": 0.73, "learning_rate": 4.115839401016485e-05, "loss": 0.4204, "step": 4165 }, { "epoch": 0.73, "learning_rate": 4.115295193467445e-05, "loss": 0.432, "step": 4166 }, { "epoch": 0.73, "learning_rate": 4.1147508594696825e-05, "loss": 0.4217, "step": 4167 }, { "epoch": 0.73, "learning_rate": 4.1142063990736104e-05, "loss": 0.4121, "step": 4168 }, { "epoch": 0.73, "learning_rate": 4.113661812329655e-05, "loss": 0.4105, "step": 4169 }, { "epoch": 0.73, "learning_rate": 4.1131170992882556e-05, "loss": 0.4099, "step": 4170 }, { "epoch": 0.73, "learning_rate": 4.112572259999859e-05, "loss": 0.4031, "step": 4171 }, { "epoch": 0.73, "learning_rate": 4.112027294514928e-05, "loss": 0.4094, "step": 4172 }, { "epoch": 0.73, "learning_rate": 4.111482202883935e-05, "loss": 0.4135, "step": 4173 }, { "epoch": 0.73, "learning_rate": 4.110936985157364e-05, "loss": 0.401, "step": 4174 }, { "epoch": 0.73, "learning_rate": 4.110391641385711e-05, "loss": 0.3875, "step": 4175 }, { "epoch": 0.73, "learning_rate": 4.1098461716194844e-05, "loss": 0.4213, "step": 4176 }, { "epoch": 0.73, "learning_rate": 4.109300575909203e-05, "loss": 0.3997, "step": 4177 }, { "epoch": 0.73, "learning_rate": 4.108754854305398e-05, "loss": 0.3944, "step": 4178 }, { "epoch": 0.73, "learning_rate": 4.108209006858613e-05, "loss": 0.4064, "step": 4179 }, { "epoch": 0.73, "learning_rate": 4.107663033619401e-05, "loss": 0.3967, "step": 4180 }, { "epoch": 0.73, "learning_rate": 4.10711693463833e-05, "loss": 0.4067, "step": 4181 }, { "epoch": 0.73, "learning_rate": 4.1065707099659763e-05, "loss": 0.4032, "step": 4182 }, { "epoch": 0.73, "learning_rate": 4.1060243596529283e-05, "loss": 0.3965, "step": 4183 }, { "epoch": 0.73, "learning_rate": 4.10547788374979e-05, "loss": 0.4244, "step": 4184 }, { "epoch": 0.73, "learning_rate": 4.104931282307172e-05, "loss": 0.42, "step": 4185 }, { "epoch": 0.73, "learning_rate": 4.1043845553756976e-05, "loss": 0.3887, "step": 4186 }, { "epoch": 0.73, "learning_rate": 4.1038377030060055e-05, "loss": 0.3991, "step": 4187 }, { "epoch": 0.73, "learning_rate": 4.1032907252487413e-05, "loss": 0.4085, "step": 4188 }, { "epoch": 0.73, "learning_rate": 4.102743622154563e-05, "loss": 0.4219, "step": 4189 }, { "epoch": 0.73, "learning_rate": 4.102196393774145e-05, "loss": 0.4073, "step": 4190 }, { "epoch": 0.73, "learning_rate": 4.101649040158166e-05, "loss": 0.3977, "step": 4191 }, { "epoch": 0.73, "learning_rate": 4.1011015613573223e-05, "loss": 0.4034, "step": 4192 }, { "epoch": 0.73, "learning_rate": 4.100553957422318e-05, "loss": 0.3776, "step": 4193 }, { "epoch": 0.73, "learning_rate": 4.10000622840387e-05, "loss": 0.4048, "step": 4194 }, { "epoch": 0.73, "learning_rate": 4.099458374352708e-05, "loss": 0.4183, "step": 4195 }, { "epoch": 0.73, "learning_rate": 4.0989103953195725e-05, "loss": 0.4174, "step": 4196 }, { "epoch": 0.73, "learning_rate": 4.0983622913552144e-05, "loss": 0.4042, "step": 4197 }, { "epoch": 0.73, "learning_rate": 4.097814062510397e-05, "loss": 0.4199, "step": 4198 }, { "epoch": 0.73, "learning_rate": 4.097265708835896e-05, "loss": 0.4113, "step": 4199 }, { "epoch": 0.73, "learning_rate": 4.096717230382497e-05, "loss": 0.4145, "step": 4200 }, { "epoch": 0.73, "learning_rate": 4.0961686272009994e-05, "loss": 0.3972, "step": 4201 }, { "epoch": 0.73, "learning_rate": 4.095619899342212e-05, "loss": 0.3885, "step": 4202 }, { "epoch": 0.73, "learning_rate": 4.095071046856957e-05, "loss": 0.3988, "step": 4203 }, { "epoch": 0.73, "learning_rate": 4.094522069796066e-05, "loss": 0.3999, "step": 4204 }, { "epoch": 0.73, "learning_rate": 4.093972968210383e-05, "loss": 0.3959, "step": 4205 }, { "epoch": 0.73, "learning_rate": 4.093423742150766e-05, "loss": 0.4033, "step": 4206 }, { "epoch": 0.73, "learning_rate": 4.09287439166808e-05, "loss": 0.4039, "step": 4207 }, { "epoch": 0.73, "learning_rate": 4.092324916813204e-05, "loss": 0.4032, "step": 4208 }, { "epoch": 0.73, "learning_rate": 4.091775317637029e-05, "loss": 0.4088, "step": 4209 }, { "epoch": 0.73, "learning_rate": 4.0912255941904576e-05, "loss": 0.392, "step": 4210 }, { "epoch": 0.73, "learning_rate": 4.090675746524402e-05, "loss": 0.3986, "step": 4211 }, { "epoch": 0.73, "learning_rate": 4.090125774689788e-05, "loss": 0.3839, "step": 4212 }, { "epoch": 0.73, "learning_rate": 4.089575678737552e-05, "loss": 0.3955, "step": 4213 }, { "epoch": 0.73, "learning_rate": 4.089025458718641e-05, "loss": 0.3984, "step": 4214 }, { "epoch": 0.73, "learning_rate": 4.088475114684015e-05, "loss": 0.3913, "step": 4215 }, { "epoch": 0.74, "learning_rate": 4.087924646684646e-05, "loss": 0.4069, "step": 4216 }, { "epoch": 0.74, "learning_rate": 4.087374054771514e-05, "loss": 0.4185, "step": 4217 }, { "epoch": 0.74, "learning_rate": 4.086823338995614e-05, "loss": 0.4093, "step": 4218 }, { "epoch": 0.74, "learning_rate": 4.086272499407951e-05, "loss": 0.4168, "step": 4219 }, { "epoch": 0.74, "learning_rate": 4.085721536059544e-05, "loss": 0.4027, "step": 4220 }, { "epoch": 0.74, "learning_rate": 4.085170449001417e-05, "loss": 0.4147, "step": 4221 }, { "epoch": 0.74, "learning_rate": 4.084619238284613e-05, "loss": 0.4309, "step": 4222 }, { "epoch": 0.74, "learning_rate": 4.0840679039601825e-05, "loss": 0.403, "step": 4223 }, { "epoch": 0.74, "learning_rate": 4.083516446079187e-05, "loss": 0.4147, "step": 4224 }, { "epoch": 0.74, "learning_rate": 4.0829648646927035e-05, "loss": 0.3978, "step": 4225 }, { "epoch": 0.74, "learning_rate": 4.082413159851814e-05, "loss": 0.3999, "step": 4226 }, { "epoch": 0.74, "learning_rate": 4.0818613316076156e-05, "loss": 0.4118, "step": 4227 }, { "epoch": 0.74, "learning_rate": 4.081309380011219e-05, "loss": 0.3864, "step": 4228 }, { "epoch": 0.74, "learning_rate": 4.080757305113743e-05, "loss": 0.4298, "step": 4229 }, { "epoch": 0.74, "learning_rate": 4.080205106966318e-05, "loss": 0.4154, "step": 4230 }, { "epoch": 0.74, "learning_rate": 4.0796527856200875e-05, "loss": 0.4121, "step": 4231 }, { "epoch": 0.74, "learning_rate": 4.079100341126206e-05, "loss": 0.4079, "step": 4232 }, { "epoch": 0.74, "learning_rate": 4.078547773535837e-05, "loss": 0.416, "step": 4233 }, { "epoch": 0.74, "learning_rate": 4.077995082900159e-05, "loss": 0.386, "step": 4234 }, { "epoch": 0.74, "learning_rate": 4.077442269270359e-05, "loss": 0.4103, "step": 4235 }, { "epoch": 0.74, "learning_rate": 4.076889332697638e-05, "loss": 0.4095, "step": 4236 }, { "epoch": 0.74, "learning_rate": 4.076336273233205e-05, "loss": 0.4137, "step": 4237 }, { "epoch": 0.74, "learning_rate": 4.075783090928285e-05, "loss": 0.4117, "step": 4238 }, { "epoch": 0.74, "learning_rate": 4.07522978583411e-05, "loss": 0.4162, "step": 4239 }, { "epoch": 0.74, "learning_rate": 4.0746763580019254e-05, "loss": 0.3995, "step": 4240 }, { "epoch": 0.74, "learning_rate": 4.074122807482987e-05, "loss": 0.3903, "step": 4241 }, { "epoch": 0.74, "learning_rate": 4.073569134328565e-05, "loss": 0.4127, "step": 4242 }, { "epoch": 0.74, "learning_rate": 4.073015338589936e-05, "loss": 0.4051, "step": 4243 }, { "epoch": 0.74, "learning_rate": 4.072461420318392e-05, "loss": 0.3999, "step": 4244 }, { "epoch": 0.74, "learning_rate": 4.0719073795652336e-05, "loss": 0.4017, "step": 4245 }, { "epoch": 0.74, "learning_rate": 4.0713532163817754e-05, "loss": 0.3975, "step": 4246 }, { "epoch": 0.74, "learning_rate": 4.070798930819342e-05, "loss": 0.4145, "step": 4247 }, { "epoch": 0.74, "learning_rate": 4.0702445229292695e-05, "loss": 0.3782, "step": 4248 }, { "epoch": 0.74, "learning_rate": 4.069689992762904e-05, "loss": 0.3978, "step": 4249 }, { "epoch": 0.74, "learning_rate": 4.0691353403716034e-05, "loss": 0.4252, "step": 4250 }, { "epoch": 0.74, "learning_rate": 4.068580565806741e-05, "loss": 0.3945, "step": 4251 }, { "epoch": 0.74, "learning_rate": 4.068025669119694e-05, "loss": 0.4052, "step": 4252 }, { "epoch": 0.74, "learning_rate": 4.0674706503618586e-05, "loss": 0.3933, "step": 4253 }, { "epoch": 0.74, "learning_rate": 4.066915509584636e-05, "loss": 0.3949, "step": 4254 }, { "epoch": 0.74, "learning_rate": 4.066360246839442e-05, "loss": 0.4267, "step": 4255 }, { "epoch": 0.74, "learning_rate": 4.065804862177704e-05, "loss": 0.4213, "step": 4256 }, { "epoch": 0.74, "learning_rate": 4.0652493556508586e-05, "loss": 0.3921, "step": 4257 }, { "epoch": 0.74, "learning_rate": 4.0646937273103554e-05, "loss": 0.4092, "step": 4258 }, { "epoch": 0.74, "learning_rate": 4.064137977207654e-05, "loss": 0.4041, "step": 4259 }, { "epoch": 0.74, "learning_rate": 4.063582105394226e-05, "loss": 0.4012, "step": 4260 }, { "epoch": 0.74, "learning_rate": 4.063026111921556e-05, "loss": 0.4075, "step": 4261 }, { "epoch": 0.74, "learning_rate": 4.062469996841136e-05, "loss": 0.4028, "step": 4262 }, { "epoch": 0.74, "learning_rate": 4.061913760204473e-05, "loss": 0.402, "step": 4263 }, { "epoch": 0.74, "learning_rate": 4.061357402063082e-05, "loss": 0.4086, "step": 4264 }, { "epoch": 0.74, "learning_rate": 4.0608009224684914e-05, "loss": 0.3991, "step": 4265 }, { "epoch": 0.74, "learning_rate": 4.060244321472241e-05, "loss": 0.4166, "step": 4266 }, { "epoch": 0.74, "learning_rate": 4.0596875991258815e-05, "loss": 0.4129, "step": 4267 }, { "epoch": 0.74, "learning_rate": 4.059130755480973e-05, "loss": 0.4004, "step": 4268 }, { "epoch": 0.74, "learning_rate": 4.0585737905890896e-05, "loss": 0.4069, "step": 4269 }, { "epoch": 0.74, "learning_rate": 4.058016704501815e-05, "loss": 0.4022, "step": 4270 }, { "epoch": 0.74, "learning_rate": 4.057459497270744e-05, "loss": 0.3994, "step": 4271 }, { "epoch": 0.74, "learning_rate": 4.0569021689474846e-05, "loss": 0.4048, "step": 4272 }, { "epoch": 0.74, "learning_rate": 4.056344719583653e-05, "loss": 0.4072, "step": 4273 }, { "epoch": 0.75, "learning_rate": 4.0557871492308784e-05, "loss": 0.4078, "step": 4274 }, { "epoch": 0.75, "learning_rate": 4.055229457940802e-05, "loss": 0.3912, "step": 4275 }, { "epoch": 0.75, "learning_rate": 4.054671645765075e-05, "loss": 0.4115, "step": 4276 }, { "epoch": 0.75, "learning_rate": 4.054113712755357e-05, "loss": 0.3968, "step": 4277 }, { "epoch": 0.75, "learning_rate": 4.053555658963325e-05, "loss": 0.4087, "step": 4278 }, { "epoch": 0.75, "learning_rate": 4.0529974844406646e-05, "loss": 0.3805, "step": 4279 }, { "epoch": 0.75, "learning_rate": 4.052439189239069e-05, "loss": 0.4016, "step": 4280 }, { "epoch": 0.75, "learning_rate": 4.051880773410248e-05, "loss": 0.3851, "step": 4281 }, { "epoch": 0.75, "learning_rate": 4.051322237005918e-05, "loss": 0.414, "step": 4282 }, { "epoch": 0.75, "learning_rate": 4.050763580077809e-05, "loss": 0.3902, "step": 4283 }, { "epoch": 0.75, "learning_rate": 4.050204802677663e-05, "loss": 0.3907, "step": 4284 }, { "epoch": 0.75, "learning_rate": 4.0496459048572314e-05, "loss": 0.3917, "step": 4285 }, { "epoch": 0.75, "learning_rate": 4.049086886668277e-05, "loss": 0.4117, "step": 4286 }, { "epoch": 0.75, "learning_rate": 4.0485277481625747e-05, "loss": 0.4148, "step": 4287 }, { "epoch": 0.75, "learning_rate": 4.047968489391909e-05, "loss": 0.3916, "step": 4288 }, { "epoch": 0.75, "learning_rate": 4.0474091104080774e-05, "loss": 0.4004, "step": 4289 }, { "epoch": 0.75, "learning_rate": 4.046849611262886e-05, "loss": 0.4098, "step": 4290 }, { "epoch": 0.75, "learning_rate": 4.0462899920081556e-05, "loss": 0.4053, "step": 4291 }, { "epoch": 0.75, "learning_rate": 4.045730252695715e-05, "loss": 0.4148, "step": 4292 }, { "epoch": 0.75, "learning_rate": 4.0451703933774047e-05, "loss": 0.3875, "step": 4293 }, { "epoch": 0.75, "learning_rate": 4.044610414105078e-05, "loss": 0.4081, "step": 4294 }, { "epoch": 0.75, "learning_rate": 4.044050314930598e-05, "loss": 0.3782, "step": 4295 }, { "epoch": 0.75, "learning_rate": 4.043490095905839e-05, "loss": 0.4024, "step": 4296 }, { "epoch": 0.75, "learning_rate": 4.0429297570826855e-05, "loss": 0.4152, "step": 4297 }, { "epoch": 0.75, "learning_rate": 4.042369298513036e-05, "loss": 0.4071, "step": 4298 }, { "epoch": 0.75, "learning_rate": 4.041808720248796e-05, "loss": 0.4046, "step": 4299 }, { "epoch": 0.75, "learning_rate": 4.041248022341885e-05, "loss": 0.3987, "step": 4300 }, { "epoch": 0.75, "learning_rate": 4.0406872048442334e-05, "loss": 0.4081, "step": 4301 }, { "epoch": 0.75, "learning_rate": 4.040126267807782e-05, "loss": 0.3907, "step": 4302 }, { "epoch": 0.75, "learning_rate": 4.0395652112844825e-05, "loss": 0.4016, "step": 4303 }, { "epoch": 0.75, "learning_rate": 4.039004035326297e-05, "loss": 0.3974, "step": 4304 }, { "epoch": 0.75, "learning_rate": 4.038442739985201e-05, "loss": 0.3977, "step": 4305 }, { "epoch": 0.75, "learning_rate": 4.03788132531318e-05, "loss": 0.3994, "step": 4306 }, { "epoch": 0.75, "learning_rate": 4.037319791362229e-05, "loss": 0.4123, "step": 4307 }, { "epoch": 0.75, "learning_rate": 4.036758138184355e-05, "loss": 0.4023, "step": 4308 }, { "epoch": 0.75, "learning_rate": 4.036196365831577e-05, "loss": 0.4102, "step": 4309 }, { "epoch": 0.75, "learning_rate": 4.035634474355925e-05, "loss": 0.4073, "step": 4310 }, { "epoch": 0.75, "learning_rate": 4.035072463809439e-05, "loss": 0.3762, "step": 4311 }, { "epoch": 0.75, "learning_rate": 4.034510334244168e-05, "loss": 0.4145, "step": 4312 }, { "epoch": 0.75, "learning_rate": 4.033948085712178e-05, "loss": 0.4102, "step": 4313 }, { "epoch": 0.75, "learning_rate": 4.0333857182655404e-05, "loss": 0.4061, "step": 4314 }, { "epoch": 0.75, "learning_rate": 4.0328232319563405e-05, "loss": 0.4163, "step": 4315 }, { "epoch": 0.75, "learning_rate": 4.032260626836673e-05, "loss": 0.4232, "step": 4316 }, { "epoch": 0.75, "learning_rate": 4.031697902958645e-05, "loss": 0.3897, "step": 4317 }, { "epoch": 0.75, "learning_rate": 4.031135060374374e-05, "loss": 0.3949, "step": 4318 }, { "epoch": 0.75, "learning_rate": 4.030572099135987e-05, "loss": 0.4068, "step": 4319 }, { "epoch": 0.75, "learning_rate": 4.030009019295626e-05, "loss": 0.4009, "step": 4320 }, { "epoch": 0.75, "learning_rate": 4.029445820905439e-05, "loss": 0.419, "step": 4321 }, { "epoch": 0.75, "learning_rate": 4.028882504017588e-05, "loss": 0.4084, "step": 4322 }, { "epoch": 0.75, "learning_rate": 4.028319068684247e-05, "loss": 0.4118, "step": 4323 }, { "epoch": 0.75, "learning_rate": 4.0277555149575975e-05, "loss": 0.4038, "step": 4324 }, { "epoch": 0.75, "learning_rate": 4.027191842889835e-05, "loss": 0.4113, "step": 4325 }, { "epoch": 0.75, "learning_rate": 4.026628052533163e-05, "loss": 0.4034, "step": 4326 }, { "epoch": 0.75, "learning_rate": 4.026064143939801e-05, "loss": 0.4064, "step": 4327 }, { "epoch": 0.75, "learning_rate": 4.025500117161973e-05, "loss": 0.4161, "step": 4328 }, { "epoch": 0.75, "learning_rate": 4.0249359722519174e-05, "loss": 0.4223, "step": 4329 }, { "epoch": 0.75, "learning_rate": 4.0243717092618854e-05, "loss": 0.4075, "step": 4330 }, { "epoch": 0.76, "learning_rate": 4.023807328244136e-05, "loss": 0.3849, "step": 4331 }, { "epoch": 0.76, "learning_rate": 4.0232428292509385e-05, "loss": 0.4206, "step": 4332 }, { "epoch": 0.76, "learning_rate": 4.022678212334578e-05, "loss": 0.3974, "step": 4333 }, { "epoch": 0.76, "learning_rate": 4.022113477547344e-05, "loss": 0.4138, "step": 4334 }, { "epoch": 0.76, "learning_rate": 4.021548624941542e-05, "loss": 0.3917, "step": 4335 }, { "epoch": 0.76, "learning_rate": 4.0209836545694865e-05, "loss": 0.4004, "step": 4336 }, { "epoch": 0.76, "learning_rate": 4.020418566483503e-05, "loss": 0.4043, "step": 4337 }, { "epoch": 0.76, "learning_rate": 4.019853360735927e-05, "loss": 0.3917, "step": 4338 }, { "epoch": 0.76, "learning_rate": 4.019288037379107e-05, "loss": 0.4244, "step": 4339 }, { "epoch": 0.76, "learning_rate": 4.0187225964654004e-05, "loss": 0.3992, "step": 4340 }, { "epoch": 0.76, "learning_rate": 4.018157038047177e-05, "loss": 0.411, "step": 4341 }, { "epoch": 0.76, "learning_rate": 4.0175913621768165e-05, "loss": 0.4146, "step": 4342 }, { "epoch": 0.76, "learning_rate": 4.017025568906709e-05, "loss": 0.4255, "step": 4343 }, { "epoch": 0.76, "learning_rate": 4.0164596582892574e-05, "loss": 0.4176, "step": 4344 }, { "epoch": 0.76, "learning_rate": 4.015893630376874e-05, "loss": 0.3916, "step": 4345 }, { "epoch": 0.76, "learning_rate": 4.015327485221982e-05, "loss": 0.4098, "step": 4346 }, { "epoch": 0.76, "learning_rate": 4.014761222877016e-05, "loss": 0.4169, "step": 4347 }, { "epoch": 0.76, "learning_rate": 4.014194843394421e-05, "loss": 0.4081, "step": 4348 }, { "epoch": 0.76, "learning_rate": 4.013628346826652e-05, "loss": 0.3989, "step": 4349 }, { "epoch": 0.76, "learning_rate": 4.013061733226178e-05, "loss": 0.3896, "step": 4350 }, { "epoch": 0.76, "learning_rate": 4.012495002645475e-05, "loss": 0.4096, "step": 4351 }, { "epoch": 0.76, "learning_rate": 4.0119281551370326e-05, "loss": 0.4108, "step": 4352 }, { "epoch": 0.76, "learning_rate": 4.01136119075335e-05, "loss": 0.3888, "step": 4353 }, { "epoch": 0.76, "learning_rate": 4.010794109546936e-05, "loss": 0.4056, "step": 4354 }, { "epoch": 0.76, "learning_rate": 4.010226911570314e-05, "loss": 0.4234, "step": 4355 }, { "epoch": 0.76, "learning_rate": 4.009659596876014e-05, "loss": 0.4155, "step": 4356 }, { "epoch": 0.76, "learning_rate": 4.009092165516579e-05, "loss": 0.4023, "step": 4357 }, { "epoch": 0.76, "learning_rate": 4.0085246175445636e-05, "loss": 0.3976, "step": 4358 }, { "epoch": 0.76, "learning_rate": 4.00795695301253e-05, "loss": 0.4114, "step": 4359 }, { "epoch": 0.76, "learning_rate": 4.007389171973054e-05, "loss": 0.4146, "step": 4360 }, { "epoch": 0.76, "learning_rate": 4.006821274478723e-05, "loss": 0.3962, "step": 4361 }, { "epoch": 0.76, "learning_rate": 4.006253260582132e-05, "loss": 0.4061, "step": 4362 }, { "epoch": 0.76, "learning_rate": 4.0056851303358886e-05, "loss": 0.3819, "step": 4363 }, { "epoch": 0.76, "learning_rate": 4.005116883792612e-05, "loss": 0.3774, "step": 4364 }, { "epoch": 0.76, "learning_rate": 4.004548521004929e-05, "loss": 0.4005, "step": 4365 }, { "epoch": 0.76, "learning_rate": 4.0039800420254816e-05, "loss": 0.4141, "step": 4366 }, { "epoch": 0.76, "learning_rate": 4.00341144690692e-05, "loss": 0.4305, "step": 4367 }, { "epoch": 0.76, "learning_rate": 4.002842735701903e-05, "loss": 0.3995, "step": 4368 }, { "epoch": 0.76, "learning_rate": 4.002273908463106e-05, "loss": 0.4046, "step": 4369 }, { "epoch": 0.76, "learning_rate": 4.0017049652432095e-05, "loss": 0.4245, "step": 4370 }, { "epoch": 0.76, "learning_rate": 4.001135906094907e-05, "loss": 0.4049, "step": 4371 }, { "epoch": 0.76, "learning_rate": 4.000566731070904e-05, "loss": 0.4016, "step": 4372 }, { "epoch": 0.76, "learning_rate": 3.999997440223914e-05, "loss": 0.3984, "step": 4373 }, { "epoch": 0.76, "learning_rate": 3.9994280336066633e-05, "loss": 0.4057, "step": 4374 }, { "epoch": 0.76, "learning_rate": 3.9988585112718893e-05, "loss": 0.406, "step": 4375 }, { "epoch": 0.76, "learning_rate": 3.9982888732723385e-05, "loss": 0.3984, "step": 4376 }, { "epoch": 0.76, "learning_rate": 3.997719119660767e-05, "loss": 0.3961, "step": 4377 }, { "epoch": 0.76, "learning_rate": 3.997149250489945e-05, "loss": 0.3858, "step": 4378 }, { "epoch": 0.76, "learning_rate": 3.996579265812652e-05, "loss": 0.4045, "step": 4379 }, { "epoch": 0.76, "learning_rate": 3.996009165681677e-05, "loss": 0.398, "step": 4380 }, { "epoch": 0.76, "learning_rate": 3.995438950149821e-05, "loss": 0.4047, "step": 4381 }, { "epoch": 0.76, "learning_rate": 3.994868619269896e-05, "loss": 0.4061, "step": 4382 }, { "epoch": 0.76, "learning_rate": 3.994298173094722e-05, "loss": 0.4103, "step": 4383 }, { "epoch": 0.76, "learning_rate": 3.993727611677134e-05, "loss": 0.4075, "step": 4384 }, { "epoch": 0.76, "learning_rate": 3.9931569350699746e-05, "loss": 0.4144, "step": 4385 }, { "epoch": 0.76, "learning_rate": 3.992586143326097e-05, "loss": 0.4044, "step": 4386 }, { "epoch": 0.76, "learning_rate": 3.992015236498366e-05, "loss": 0.4146, "step": 4387 }, { "epoch": 0.77, "learning_rate": 3.9914442146396574e-05, "loss": 0.4312, "step": 4388 }, { "epoch": 0.77, "learning_rate": 3.9908730778028576e-05, "loss": 0.4102, "step": 4389 }, { "epoch": 0.77, "learning_rate": 3.990301826040862e-05, "loss": 0.4179, "step": 4390 }, { "epoch": 0.77, "learning_rate": 3.989730459406579e-05, "loss": 0.3911, "step": 4391 }, { "epoch": 0.77, "learning_rate": 3.9891589779529266e-05, "loss": 0.3934, "step": 4392 }, { "epoch": 0.77, "learning_rate": 3.988587381732833e-05, "loss": 0.4047, "step": 4393 }, { "epoch": 0.77, "learning_rate": 3.988015670799236e-05, "loss": 0.3988, "step": 4394 }, { "epoch": 0.77, "learning_rate": 3.987443845205088e-05, "loss": 0.3872, "step": 4395 }, { "epoch": 0.77, "learning_rate": 3.9868719050033476e-05, "loss": 0.3903, "step": 4396 }, { "epoch": 0.77, "learning_rate": 3.986299850246987e-05, "loss": 0.4254, "step": 4397 }, { "epoch": 0.77, "learning_rate": 3.985727680988986e-05, "loss": 0.425, "step": 4398 }, { "epoch": 0.77, "learning_rate": 3.98515539728234e-05, "loss": 0.4142, "step": 4399 }, { "epoch": 0.77, "learning_rate": 3.984582999180048e-05, "loss": 0.4051, "step": 4400 }, { "epoch": 0.77, "learning_rate": 3.9840104867351264e-05, "loss": 0.3995, "step": 4401 }, { "epoch": 0.77, "learning_rate": 3.983437860000597e-05, "loss": 0.3848, "step": 4402 }, { "epoch": 0.77, "learning_rate": 3.982865119029497e-05, "loss": 0.4067, "step": 4403 }, { "epoch": 0.77, "learning_rate": 3.982292263874869e-05, "loss": 0.4094, "step": 4404 }, { "epoch": 0.77, "learning_rate": 3.981719294589772e-05, "loss": 0.4054, "step": 4405 }, { "epoch": 0.77, "learning_rate": 3.9811462112272685e-05, "loss": 0.4084, "step": 4406 }, { "epoch": 0.77, "learning_rate": 3.980573013840438e-05, "loss": 0.4, "step": 4407 }, { "epoch": 0.77, "learning_rate": 3.979999702482367e-05, "loss": 0.3887, "step": 4408 }, { "epoch": 0.77, "learning_rate": 3.979426277206154e-05, "loss": 0.4271, "step": 4409 }, { "epoch": 0.77, "learning_rate": 3.978852738064908e-05, "loss": 0.4207, "step": 4410 }, { "epoch": 0.77, "learning_rate": 3.9782790851117476e-05, "loss": 0.4178, "step": 4411 }, { "epoch": 0.77, "learning_rate": 3.9777053183998006e-05, "loss": 0.4049, "step": 4412 }, { "epoch": 0.77, "learning_rate": 3.9771314379822104e-05, "loss": 0.4047, "step": 4413 }, { "epoch": 0.77, "learning_rate": 3.9765574439121267e-05, "loss": 0.3917, "step": 4414 }, { "epoch": 0.77, "learning_rate": 3.97598333624271e-05, "loss": 0.3888, "step": 4415 }, { "epoch": 0.77, "learning_rate": 3.9754091150271324e-05, "loss": 0.3993, "step": 4416 }, { "epoch": 0.77, "learning_rate": 3.974834780318576e-05, "loss": 0.3881, "step": 4417 }, { "epoch": 0.77, "learning_rate": 3.974260332170234e-05, "loss": 0.4113, "step": 4418 }, { "epoch": 0.77, "learning_rate": 3.97368577063531e-05, "loss": 0.406, "step": 4419 }, { "epoch": 0.77, "learning_rate": 3.9731110957670165e-05, "loss": 0.4028, "step": 4420 }, { "epoch": 0.77, "learning_rate": 3.97253630761858e-05, "loss": 0.411, "step": 4421 }, { "epoch": 0.77, "learning_rate": 3.971961406243233e-05, "loss": 0.4096, "step": 4422 }, { "epoch": 0.77, "learning_rate": 3.971386391694222e-05, "loss": 0.4112, "step": 4423 }, { "epoch": 0.77, "learning_rate": 3.970811264024803e-05, "loss": 0.4194, "step": 4424 }, { "epoch": 0.77, "learning_rate": 3.970236023288243e-05, "loss": 0.4201, "step": 4425 }, { "epoch": 0.77, "learning_rate": 3.9696606695378154e-05, "loss": 0.3897, "step": 4426 }, { "epoch": 0.77, "learning_rate": 3.969085202826812e-05, "loss": 0.4268, "step": 4427 }, { "epoch": 0.77, "learning_rate": 3.968509623208526e-05, "loss": 0.3981, "step": 4428 }, { "epoch": 0.77, "learning_rate": 3.967933930736268e-05, "loss": 0.4076, "step": 4429 }, { "epoch": 0.77, "learning_rate": 3.967358125463357e-05, "loss": 0.3986, "step": 4430 }, { "epoch": 0.77, "learning_rate": 3.9667822074431204e-05, "loss": 0.4067, "step": 4431 }, { "epoch": 0.77, "learning_rate": 3.9662061767288984e-05, "loss": 0.4126, "step": 4432 }, { "epoch": 0.77, "learning_rate": 3.965630033374041e-05, "loss": 0.4048, "step": 4433 }, { "epoch": 0.77, "learning_rate": 3.965053777431908e-05, "loss": 0.4219, "step": 4434 }, { "epoch": 0.77, "learning_rate": 3.964477408955872e-05, "loss": 0.4152, "step": 4435 }, { "epoch": 0.77, "learning_rate": 3.963900927999312e-05, "loss": 0.3955, "step": 4436 }, { "epoch": 0.77, "learning_rate": 3.963324334615619e-05, "loss": 0.3731, "step": 4437 }, { "epoch": 0.77, "learning_rate": 3.962747628858198e-05, "loss": 0.4145, "step": 4438 }, { "epoch": 0.77, "learning_rate": 3.962170810780459e-05, "loss": 0.3962, "step": 4439 }, { "epoch": 0.77, "learning_rate": 3.9615938804358254e-05, "loss": 0.4032, "step": 4440 }, { "epoch": 0.77, "learning_rate": 3.9610168378777306e-05, "loss": 0.4188, "step": 4441 }, { "epoch": 0.77, "learning_rate": 3.960439683159619e-05, "loss": 0.4072, "step": 4442 }, { "epoch": 0.77, "learning_rate": 3.9598624163349426e-05, "loss": 0.4164, "step": 4443 }, { "epoch": 0.77, "learning_rate": 3.959285037457168e-05, "loss": 0.4053, "step": 4444 }, { "epoch": 0.77, "learning_rate": 3.958707546579767e-05, "loss": 0.3976, "step": 4445 }, { "epoch": 0.78, "learning_rate": 3.958129943756228e-05, "loss": 0.4055, "step": 4446 }, { "epoch": 0.78, "learning_rate": 3.9575522290400456e-05, "loss": 0.3977, "step": 4447 }, { "epoch": 0.78, "learning_rate": 3.9569744024847246e-05, "loss": 0.4408, "step": 4448 }, { "epoch": 0.78, "learning_rate": 3.956396464143782e-05, "loss": 0.413, "step": 4449 }, { "epoch": 0.78, "learning_rate": 3.955818414070744e-05, "loss": 0.3993, "step": 4450 }, { "epoch": 0.78, "learning_rate": 3.955240252319147e-05, "loss": 0.3964, "step": 4451 }, { "epoch": 0.78, "learning_rate": 3.954661978942539e-05, "loss": 0.3978, "step": 4452 }, { "epoch": 0.78, "learning_rate": 3.9540835939944786e-05, "loss": 0.4154, "step": 4453 }, { "epoch": 0.78, "learning_rate": 3.953505097528532e-05, "loss": 0.407, "step": 4454 }, { "epoch": 0.78, "learning_rate": 3.9529264895982774e-05, "loss": 0.4084, "step": 4455 }, { "epoch": 0.78, "learning_rate": 3.952347770257305e-05, "loss": 0.4145, "step": 4456 }, { "epoch": 0.78, "learning_rate": 3.951768939559212e-05, "loss": 0.3964, "step": 4457 }, { "epoch": 0.78, "learning_rate": 3.951189997557609e-05, "loss": 0.4091, "step": 4458 }, { "epoch": 0.78, "learning_rate": 3.950610944306115e-05, "loss": 0.4065, "step": 4459 }, { "epoch": 0.78, "learning_rate": 3.9500317798583596e-05, "loss": 0.3963, "step": 4460 }, { "epoch": 0.78, "learning_rate": 3.949452504267983e-05, "loss": 0.4119, "step": 4461 }, { "epoch": 0.78, "learning_rate": 3.9488731175886356e-05, "loss": 0.4005, "step": 4462 }, { "epoch": 0.78, "learning_rate": 3.948293619873979e-05, "loss": 0.3996, "step": 4463 }, { "epoch": 0.78, "learning_rate": 3.9477140111776824e-05, "loss": 0.4159, "step": 4464 }, { "epoch": 0.78, "learning_rate": 3.947134291553429e-05, "loss": 0.3864, "step": 4465 }, { "epoch": 0.78, "learning_rate": 3.946554461054908e-05, "loss": 0.4016, "step": 4466 }, { "epoch": 0.78, "learning_rate": 3.945974519735823e-05, "loss": 0.4276, "step": 4467 }, { "epoch": 0.78, "learning_rate": 3.945394467649886e-05, "loss": 0.4215, "step": 4468 }, { "epoch": 0.78, "learning_rate": 3.94481430485082e-05, "loss": 0.4073, "step": 4469 }, { "epoch": 0.78, "learning_rate": 3.9442340313923546e-05, "loss": 0.3937, "step": 4470 }, { "epoch": 0.78, "learning_rate": 3.943653647328236e-05, "loss": 0.4043, "step": 4471 }, { "epoch": 0.78, "learning_rate": 3.9430731527122156e-05, "loss": 0.3891, "step": 4472 }, { "epoch": 0.78, "learning_rate": 3.9424925475980574e-05, "loss": 0.3928, "step": 4473 }, { "epoch": 0.78, "learning_rate": 3.9419118320395334e-05, "loss": 0.4109, "step": 4474 }, { "epoch": 0.78, "learning_rate": 3.94133100609043e-05, "loss": 0.4063, "step": 4475 }, { "epoch": 0.78, "learning_rate": 3.940750069804539e-05, "loss": 0.4022, "step": 4476 }, { "epoch": 0.78, "learning_rate": 3.9401690232356654e-05, "loss": 0.3686, "step": 4477 }, { "epoch": 0.78, "learning_rate": 3.939587866437624e-05, "loss": 0.4117, "step": 4478 }, { "epoch": 0.78, "learning_rate": 3.939006599464239e-05, "loss": 0.3954, "step": 4479 }, { "epoch": 0.78, "learning_rate": 3.9384252223693457e-05, "loss": 0.4067, "step": 4480 }, { "epoch": 0.78, "learning_rate": 3.937843735206788e-05, "loss": 0.404, "step": 4481 }, { "epoch": 0.78, "learning_rate": 3.937262138030423e-05, "loss": 0.4032, "step": 4482 }, { "epoch": 0.78, "learning_rate": 3.9366804308941154e-05, "loss": 0.4096, "step": 4483 }, { "epoch": 0.78, "learning_rate": 3.9360986138517385e-05, "loss": 0.418, "step": 4484 }, { "epoch": 0.78, "learning_rate": 3.935516686957182e-05, "loss": 0.4106, "step": 4485 }, { "epoch": 0.78, "learning_rate": 3.934934650264338e-05, "loss": 0.4087, "step": 4486 }, { "epoch": 0.78, "learning_rate": 3.9343525038271166e-05, "loss": 0.4224, "step": 4487 }, { "epoch": 0.78, "learning_rate": 3.933770247699431e-05, "loss": 0.4018, "step": 4488 }, { "epoch": 0.78, "learning_rate": 3.93318788193521e-05, "loss": 0.4106, "step": 4489 }, { "epoch": 0.78, "learning_rate": 3.932605406588388e-05, "loss": 0.4285, "step": 4490 }, { "epoch": 0.78, "learning_rate": 3.932022821712914e-05, "loss": 0.375, "step": 4491 }, { "epoch": 0.78, "learning_rate": 3.9314401273627414e-05, "loss": 0.4093, "step": 4492 }, { "epoch": 0.78, "learning_rate": 3.930857323591841e-05, "loss": 0.402, "step": 4493 }, { "epoch": 0.78, "learning_rate": 3.93027441045419e-05, "loss": 0.4181, "step": 4494 }, { "epoch": 0.78, "learning_rate": 3.929691388003772e-05, "loss": 0.3932, "step": 4495 }, { "epoch": 0.78, "learning_rate": 3.929108256294587e-05, "loss": 0.4022, "step": 4496 }, { "epoch": 0.78, "learning_rate": 3.928525015380643e-05, "loss": 0.4005, "step": 4497 }, { "epoch": 0.78, "learning_rate": 3.927941665315956e-05, "loss": 0.3984, "step": 4498 }, { "epoch": 0.78, "learning_rate": 3.9273582061545557e-05, "loss": 0.401, "step": 4499 }, { "epoch": 0.78, "learning_rate": 3.926774637950478e-05, "loss": 0.3937, "step": 4500 }, { "epoch": 0.78, "learning_rate": 3.926190960757773e-05, "loss": 0.3976, "step": 4501 }, { "epoch": 0.78, "learning_rate": 3.9256071746304974e-05, "loss": 0.3853, "step": 4502 }, { "epoch": 0.79, "learning_rate": 3.9250232796227186e-05, "loss": 0.4087, "step": 4503 }, { "epoch": 0.79, "learning_rate": 3.924439275788517e-05, "loss": 0.4154, "step": 4504 }, { "epoch": 0.79, "learning_rate": 3.923855163181979e-05, "loss": 0.4301, "step": 4505 }, { "epoch": 0.79, "learning_rate": 3.923270941857205e-05, "loss": 0.4113, "step": 4506 }, { "epoch": 0.79, "learning_rate": 3.922686611868301e-05, "loss": 0.3938, "step": 4507 }, { "epoch": 0.79, "learning_rate": 3.9221021732693874e-05, "loss": 0.4063, "step": 4508 }, { "epoch": 0.79, "learning_rate": 3.921517626114593e-05, "loss": 0.4142, "step": 4509 }, { "epoch": 0.79, "learning_rate": 3.920932970458054e-05, "loss": 0.4102, "step": 4510 }, { "epoch": 0.79, "learning_rate": 3.9203482063539224e-05, "loss": 0.4035, "step": 4511 }, { "epoch": 0.79, "learning_rate": 3.919763333856354e-05, "loss": 0.3993, "step": 4512 }, { "epoch": 0.79, "learning_rate": 3.91917835301952e-05, "loss": 0.4064, "step": 4513 }, { "epoch": 0.79, "learning_rate": 3.918593263897597e-05, "loss": 0.395, "step": 4514 }, { "epoch": 0.79, "learning_rate": 3.918008066544776e-05, "loss": 0.4137, "step": 4515 }, { "epoch": 0.79, "learning_rate": 3.917422761015255e-05, "loss": 0.3972, "step": 4516 }, { "epoch": 0.79, "learning_rate": 3.916837347363242e-05, "loss": 0.4076, "step": 4517 }, { "epoch": 0.79, "learning_rate": 3.916251825642957e-05, "loss": 0.3866, "step": 4518 }, { "epoch": 0.79, "learning_rate": 3.915666195908628e-05, "loss": 0.4012, "step": 4519 }, { "epoch": 0.79, "learning_rate": 3.915080458214495e-05, "loss": 0.4275, "step": 4520 }, { "epoch": 0.79, "learning_rate": 3.9144946126148066e-05, "loss": 0.3968, "step": 4521 }, { "epoch": 0.79, "learning_rate": 3.913908659163821e-05, "loss": 0.4106, "step": 4522 }, { "epoch": 0.79, "learning_rate": 3.913322597915808e-05, "loss": 0.3953, "step": 4523 }, { "epoch": 0.79, "learning_rate": 3.912736428925046e-05, "loss": 0.4146, "step": 4524 }, { "epoch": 0.79, "learning_rate": 3.9121501522458234e-05, "loss": 0.3989, "step": 4525 }, { "epoch": 0.79, "learning_rate": 3.9115637679324405e-05, "loss": 0.4088, "step": 4526 }, { "epoch": 0.79, "learning_rate": 3.910977276039204e-05, "loss": 0.4259, "step": 4527 }, { "epoch": 0.79, "learning_rate": 3.910390676620435e-05, "loss": 0.3827, "step": 4528 }, { "epoch": 0.79, "learning_rate": 3.90980396973046e-05, "loss": 0.415, "step": 4529 }, { "epoch": 0.79, "learning_rate": 3.9092171554236194e-05, "loss": 0.3803, "step": 4530 }, { "epoch": 0.79, "learning_rate": 3.908630233754261e-05, "loss": 0.408, "step": 4531 }, { "epoch": 0.79, "learning_rate": 3.908043204776744e-05, "loss": 0.3958, "step": 4532 }, { "epoch": 0.79, "learning_rate": 3.9074560685454356e-05, "loss": 0.407, "step": 4533 }, { "epoch": 0.79, "learning_rate": 3.9068688251147154e-05, "loss": 0.4028, "step": 4534 }, { "epoch": 0.79, "learning_rate": 3.906281474538972e-05, "loss": 0.3982, "step": 4535 }, { "epoch": 0.79, "learning_rate": 3.905694016872603e-05, "loss": 0.4084, "step": 4536 }, { "epoch": 0.79, "learning_rate": 3.905106452170017e-05, "loss": 0.4115, "step": 4537 }, { "epoch": 0.79, "learning_rate": 3.904518780485631e-05, "loss": 0.3996, "step": 4538 }, { "epoch": 0.79, "learning_rate": 3.903931001873874e-05, "loss": 0.3863, "step": 4539 }, { "epoch": 0.79, "learning_rate": 3.9033431163891845e-05, "loss": 0.4142, "step": 4540 }, { "epoch": 0.79, "learning_rate": 3.902755124086009e-05, "loss": 0.4082, "step": 4541 }, { "epoch": 0.79, "learning_rate": 3.902167025018807e-05, "loss": 0.4137, "step": 4542 }, { "epoch": 0.79, "learning_rate": 3.901578819242044e-05, "loss": 0.3888, "step": 4543 }, { "epoch": 0.79, "learning_rate": 3.900990506810199e-05, "loss": 0.3874, "step": 4544 }, { "epoch": 0.79, "learning_rate": 3.900402087777759e-05, "loss": 0.3886, "step": 4545 }, { "epoch": 0.79, "learning_rate": 3.8998135621992215e-05, "loss": 0.4062, "step": 4546 }, { "epoch": 0.79, "learning_rate": 3.899224930129093e-05, "loss": 0.3993, "step": 4547 }, { "epoch": 0.79, "learning_rate": 3.898636191621891e-05, "loss": 0.4139, "step": 4548 }, { "epoch": 0.79, "learning_rate": 3.898047346732142e-05, "loss": 0.4054, "step": 4549 }, { "epoch": 0.79, "learning_rate": 3.8974583955143826e-05, "loss": 0.4055, "step": 4550 }, { "epoch": 0.79, "learning_rate": 3.896869338023159e-05, "loss": 0.4007, "step": 4551 }, { "epoch": 0.79, "learning_rate": 3.896280174313029e-05, "loss": 0.4015, "step": 4552 }, { "epoch": 0.79, "learning_rate": 3.895690904438558e-05, "loss": 0.3897, "step": 4553 }, { "epoch": 0.79, "learning_rate": 3.895101528454322e-05, "loss": 0.4177, "step": 4554 }, { "epoch": 0.79, "learning_rate": 3.8945120464149067e-05, "loss": 0.3989, "step": 4555 }, { "epoch": 0.79, "learning_rate": 3.893922458374908e-05, "loss": 0.3882, "step": 4556 }, { "epoch": 0.79, "learning_rate": 3.893332764388931e-05, "loss": 0.4038, "step": 4557 }, { "epoch": 0.79, "learning_rate": 3.8927429645115914e-05, "loss": 0.4072, "step": 4558 }, { "epoch": 0.79, "learning_rate": 3.892153058797515e-05, "loss": 0.3974, "step": 4559 }, { "epoch": 0.8, "learning_rate": 3.8915630473013365e-05, "loss": 0.4073, "step": 4560 }, { "epoch": 0.8, "learning_rate": 3.8909729300777e-05, "loss": 0.3983, "step": 4561 }, { "epoch": 0.8, "learning_rate": 3.8903827071812604e-05, "loss": 0.3913, "step": 4562 }, { "epoch": 0.8, "learning_rate": 3.889792378666681e-05, "loss": 0.4071, "step": 4563 }, { "epoch": 0.8, "learning_rate": 3.8892019445886375e-05, "loss": 0.3946, "step": 4564 }, { "epoch": 0.8, "learning_rate": 3.8886114050018134e-05, "loss": 0.4058, "step": 4565 }, { "epoch": 0.8, "learning_rate": 3.888020759960902e-05, "loss": 0.4064, "step": 4566 }, { "epoch": 0.8, "learning_rate": 3.8874300095206064e-05, "loss": 0.3966, "step": 4567 }, { "epoch": 0.8, "learning_rate": 3.8868391537356405e-05, "loss": 0.4041, "step": 4568 }, { "epoch": 0.8, "learning_rate": 3.886248192660727e-05, "loss": 0.409, "step": 4569 }, { "epoch": 0.8, "learning_rate": 3.885657126350598e-05, "loss": 0.3986, "step": 4570 }, { "epoch": 0.8, "learning_rate": 3.885065954859996e-05, "loss": 0.4074, "step": 4571 }, { "epoch": 0.8, "learning_rate": 3.8844746782436734e-05, "loss": 0.3963, "step": 4572 }, { "epoch": 0.8, "learning_rate": 3.883883296556394e-05, "loss": 0.4062, "step": 4573 }, { "epoch": 0.8, "learning_rate": 3.8832918098529266e-05, "loss": 0.4147, "step": 4574 }, { "epoch": 0.8, "learning_rate": 3.882700218188054e-05, "loss": 0.4094, "step": 4575 }, { "epoch": 0.8, "learning_rate": 3.882108521616567e-05, "loss": 0.418, "step": 4576 }, { "epoch": 0.8, "learning_rate": 3.881516720193266e-05, "loss": 0.4097, "step": 4577 }, { "epoch": 0.8, "learning_rate": 3.880924813972962e-05, "loss": 0.4051, "step": 4578 }, { "epoch": 0.8, "learning_rate": 3.880332803010475e-05, "loss": 0.4036, "step": 4579 }, { "epoch": 0.8, "learning_rate": 3.8797406873606355e-05, "loss": 0.4076, "step": 4580 }, { "epoch": 0.8, "learning_rate": 3.8791484670782825e-05, "loss": 0.3998, "step": 4581 }, { "epoch": 0.8, "learning_rate": 3.878556142218265e-05, "loss": 0.3934, "step": 4582 }, { "epoch": 0.8, "learning_rate": 3.877963712835443e-05, "loss": 0.406, "step": 4583 }, { "epoch": 0.8, "learning_rate": 3.877371178984685e-05, "loss": 0.4, "step": 4584 }, { "epoch": 0.8, "learning_rate": 3.8767785407208695e-05, "loss": 0.3972, "step": 4585 }, { "epoch": 0.8, "learning_rate": 3.876185798098882e-05, "loss": 0.3889, "step": 4586 }, { "epoch": 0.8, "learning_rate": 3.875592951173624e-05, "loss": 0.3917, "step": 4587 }, { "epoch": 0.8, "learning_rate": 3.875e-05, "loss": 0.3612, "step": 4588 }, { "epoch": 0.8, "learning_rate": 3.874406944632928e-05, "loss": 0.4234, "step": 4589 }, { "epoch": 0.8, "learning_rate": 3.873813785127336e-05, "loss": 0.3899, "step": 4590 }, { "epoch": 0.8, "learning_rate": 3.873220521538158e-05, "loss": 0.3983, "step": 4591 }, { "epoch": 0.8, "learning_rate": 3.8726271539203417e-05, "loss": 0.4067, "step": 4592 }, { "epoch": 0.8, "learning_rate": 3.872033682328841e-05, "loss": 0.4013, "step": 4593 }, { "epoch": 0.8, "learning_rate": 3.8714401068186226e-05, "loss": 0.3857, "step": 4594 }, { "epoch": 0.8, "learning_rate": 3.87084642744466e-05, "loss": 0.3988, "step": 4595 }, { "epoch": 0.8, "learning_rate": 3.87025264426194e-05, "loss": 0.4007, "step": 4596 }, { "epoch": 0.8, "learning_rate": 3.8696587573254534e-05, "loss": 0.4028, "step": 4597 }, { "epoch": 0.8, "learning_rate": 3.869064766690206e-05, "loss": 0.4075, "step": 4598 }, { "epoch": 0.8, "learning_rate": 3.868470672411211e-05, "loss": 0.394, "step": 4599 }, { "epoch": 0.8, "learning_rate": 3.867876474543491e-05, "loss": 0.3906, "step": 4600 }, { "epoch": 0.8, "learning_rate": 3.8672821731420774e-05, "loss": 0.4115, "step": 4601 }, { "epoch": 0.8, "learning_rate": 3.866687768262014e-05, "loss": 0.3838, "step": 4602 }, { "epoch": 0.8, "learning_rate": 3.866093259958351e-05, "loss": 0.3991, "step": 4603 }, { "epoch": 0.8, "learning_rate": 3.865498648286151e-05, "loss": 0.4005, "step": 4604 }, { "epoch": 0.8, "learning_rate": 3.864903933300484e-05, "loss": 0.4167, "step": 4605 }, { "epoch": 0.8, "learning_rate": 3.86430911505643e-05, "loss": 0.403, "step": 4606 }, { "epoch": 0.8, "learning_rate": 3.863714193609079e-05, "loss": 0.4076, "step": 4607 }, { "epoch": 0.8, "learning_rate": 3.8631191690135306e-05, "loss": 0.3911, "step": 4608 }, { "epoch": 0.8, "learning_rate": 3.862524041324895e-05, "loss": 0.4006, "step": 4609 }, { "epoch": 0.8, "learning_rate": 3.8619288105982895e-05, "loss": 0.4075, "step": 4610 }, { "epoch": 0.8, "learning_rate": 3.861333476888842e-05, "loss": 0.4105, "step": 4611 }, { "epoch": 0.8, "learning_rate": 3.860738040251691e-05, "loss": 0.4034, "step": 4612 }, { "epoch": 0.8, "learning_rate": 3.860142500741985e-05, "loss": 0.4206, "step": 4613 }, { "epoch": 0.8, "learning_rate": 3.859546858414877e-05, "loss": 0.3854, "step": 4614 }, { "epoch": 0.8, "learning_rate": 3.8589511133255356e-05, "loss": 0.4171, "step": 4615 }, { "epoch": 0.8, "learning_rate": 3.8583552655291376e-05, "loss": 0.4018, "step": 4616 }, { "epoch": 0.8, "learning_rate": 3.8577593150808655e-05, "loss": 0.4079, "step": 4617 }, { "epoch": 0.81, "learning_rate": 3.8571632620359174e-05, "loss": 0.3946, "step": 4618 }, { "epoch": 0.81, "learning_rate": 3.856567106449494e-05, "loss": 0.407, "step": 4619 }, { "epoch": 0.81, "learning_rate": 3.8559708483768114e-05, "loss": 0.3876, "step": 4620 }, { "epoch": 0.81, "learning_rate": 3.855374487873093e-05, "loss": 0.3857, "step": 4621 }, { "epoch": 0.81, "learning_rate": 3.85477802499357e-05, "loss": 0.4229, "step": 4622 }, { "epoch": 0.81, "learning_rate": 3.8541814597934865e-05, "loss": 0.3933, "step": 4623 }, { "epoch": 0.81, "learning_rate": 3.853584792328093e-05, "loss": 0.3983, "step": 4624 }, { "epoch": 0.81, "learning_rate": 3.8529880226526504e-05, "loss": 0.421, "step": 4625 }, { "epoch": 0.81, "learning_rate": 3.8523911508224304e-05, "loss": 0.3947, "step": 4626 }, { "epoch": 0.81, "learning_rate": 3.851794176892713e-05, "loss": 0.3954, "step": 4627 }, { "epoch": 0.81, "learning_rate": 3.8511971009187855e-05, "loss": 0.3863, "step": 4628 }, { "epoch": 0.81, "learning_rate": 3.8505999229559506e-05, "loss": 0.3907, "step": 4629 }, { "epoch": 0.81, "learning_rate": 3.850002643059515e-05, "loss": 0.4058, "step": 4630 }, { "epoch": 0.81, "learning_rate": 3.8494052612847965e-05, "loss": 0.4057, "step": 4631 }, { "epoch": 0.81, "learning_rate": 3.848807777687122e-05, "loss": 0.4034, "step": 4632 }, { "epoch": 0.81, "learning_rate": 3.8482101923218294e-05, "loss": 0.4141, "step": 4633 }, { "epoch": 0.81, "learning_rate": 3.8476125052442646e-05, "loss": 0.3917, "step": 4634 }, { "epoch": 0.81, "learning_rate": 3.847014716509783e-05, "loss": 0.3788, "step": 4635 }, { "epoch": 0.81, "learning_rate": 3.846416826173748e-05, "loss": 0.4017, "step": 4636 }, { "epoch": 0.81, "learning_rate": 3.845818834291538e-05, "loss": 0.4142, "step": 4637 }, { "epoch": 0.81, "learning_rate": 3.845220740918534e-05, "loss": 0.4353, "step": 4638 }, { "epoch": 0.81, "learning_rate": 3.844622546110128e-05, "loss": 0.4113, "step": 4639 }, { "epoch": 0.81, "learning_rate": 3.844024249921726e-05, "loss": 0.3828, "step": 4640 }, { "epoch": 0.81, "learning_rate": 3.843425852408738e-05, "loss": 0.3902, "step": 4641 }, { "epoch": 0.81, "learning_rate": 3.8428273536265855e-05, "loss": 0.4112, "step": 4642 }, { "epoch": 0.81, "learning_rate": 3.8422287536307006e-05, "loss": 0.3857, "step": 4643 }, { "epoch": 0.81, "learning_rate": 3.841630052476522e-05, "loss": 0.4074, "step": 4644 }, { "epoch": 0.81, "learning_rate": 3.8410312502195e-05, "loss": 0.3962, "step": 4645 }, { "epoch": 0.81, "learning_rate": 3.840432346915093e-05, "loss": 0.4126, "step": 4646 }, { "epoch": 0.81, "learning_rate": 3.83983334261877e-05, "loss": 0.4042, "step": 4647 }, { "epoch": 0.81, "learning_rate": 3.839234237386008e-05, "loss": 0.4035, "step": 4648 }, { "epoch": 0.81, "learning_rate": 3.8386350312722943e-05, "loss": 0.3851, "step": 4649 }, { "epoch": 0.81, "learning_rate": 3.838035724333125e-05, "loss": 0.4069, "step": 4650 }, { "epoch": 0.81, "learning_rate": 3.837436316624006e-05, "loss": 0.4251, "step": 4651 }, { "epoch": 0.81, "learning_rate": 3.836836808200452e-05, "loss": 0.4001, "step": 4652 }, { "epoch": 0.81, "learning_rate": 3.8362371991179874e-05, "loss": 0.4044, "step": 4653 }, { "epoch": 0.81, "learning_rate": 3.835637489432146e-05, "loss": 0.4364, "step": 4654 }, { "epoch": 0.81, "learning_rate": 3.83503767919847e-05, "loss": 0.4101, "step": 4655 }, { "epoch": 0.81, "learning_rate": 3.834437768472513e-05, "loss": 0.4065, "step": 4656 }, { "epoch": 0.81, "learning_rate": 3.8338377573098354e-05, "loss": 0.3923, "step": 4657 }, { "epoch": 0.81, "learning_rate": 3.833237645766009e-05, "loss": 0.39, "step": 4658 }, { "epoch": 0.81, "learning_rate": 3.8326374338966134e-05, "loss": 0.3833, "step": 4659 }, { "epoch": 0.81, "learning_rate": 3.832037121757237e-05, "loss": 0.3842, "step": 4660 }, { "epoch": 0.81, "learning_rate": 3.8314367094034816e-05, "loss": 0.4142, "step": 4661 }, { "epoch": 0.81, "learning_rate": 3.8308361968909524e-05, "loss": 0.4183, "step": 4662 }, { "epoch": 0.81, "learning_rate": 3.830235584275268e-05, "loss": 0.3862, "step": 4663 }, { "epoch": 0.81, "learning_rate": 3.8296348716120535e-05, "loss": 0.3936, "step": 4664 }, { "epoch": 0.81, "learning_rate": 3.829034058956947e-05, "loss": 0.3867, "step": 4665 }, { "epoch": 0.81, "learning_rate": 3.828433146365592e-05, "loss": 0.4024, "step": 4666 }, { "epoch": 0.81, "learning_rate": 3.827832133893643e-05, "loss": 0.4043, "step": 4667 }, { "epoch": 0.81, "learning_rate": 3.827231021596764e-05, "loss": 0.3988, "step": 4668 }, { "epoch": 0.81, "learning_rate": 3.826629809530629e-05, "loss": 0.4102, "step": 4669 }, { "epoch": 0.81, "learning_rate": 3.826028497750918e-05, "loss": 0.4104, "step": 4670 }, { "epoch": 0.81, "learning_rate": 3.8254270863133236e-05, "loss": 0.4177, "step": 4671 }, { "epoch": 0.81, "learning_rate": 3.8248255752735454e-05, "loss": 0.4061, "step": 4672 }, { "epoch": 0.81, "learning_rate": 3.8242239646872934e-05, "loss": 0.3851, "step": 4673 }, { "epoch": 0.81, "learning_rate": 3.823622254610288e-05, "loss": 0.412, "step": 4674 }, { "epoch": 0.82, "learning_rate": 3.823020445098256e-05, "loss": 0.3806, "step": 4675 }, { "epoch": 0.82, "learning_rate": 3.822418536206935e-05, "loss": 0.4193, "step": 4676 }, { "epoch": 0.82, "learning_rate": 3.8218165279920717e-05, "loss": 0.4095, "step": 4677 }, { "epoch": 0.82, "learning_rate": 3.821214420509422e-05, "loss": 0.4082, "step": 4678 }, { "epoch": 0.82, "learning_rate": 3.820612213814752e-05, "loss": 0.4271, "step": 4679 }, { "epoch": 0.82, "learning_rate": 3.820009907963834e-05, "loss": 0.4126, "step": 4680 }, { "epoch": 0.82, "learning_rate": 3.819407503012453e-05, "loss": 0.419, "step": 4681 }, { "epoch": 0.82, "learning_rate": 3.8188049990164e-05, "loss": 0.4109, "step": 4682 }, { "epoch": 0.82, "learning_rate": 3.818202396031478e-05, "loss": 0.3826, "step": 4683 }, { "epoch": 0.82, "learning_rate": 3.817599694113498e-05, "loss": 0.3898, "step": 4684 }, { "epoch": 0.82, "learning_rate": 3.81699689331828e-05, "loss": 0.4068, "step": 4685 }, { "epoch": 0.82, "learning_rate": 3.816393993701651e-05, "loss": 0.3997, "step": 4686 }, { "epoch": 0.82, "learning_rate": 3.815790995319453e-05, "loss": 0.4162, "step": 4687 }, { "epoch": 0.82, "learning_rate": 3.815187898227531e-05, "loss": 0.3839, "step": 4688 }, { "epoch": 0.82, "learning_rate": 3.8145847024817434e-05, "loss": 0.4151, "step": 4689 }, { "epoch": 0.82, "learning_rate": 3.8139814081379546e-05, "loss": 0.4073, "step": 4690 }, { "epoch": 0.82, "learning_rate": 3.81337801525204e-05, "loss": 0.3928, "step": 4691 }, { "epoch": 0.82, "learning_rate": 3.812774523879884e-05, "loss": 0.407, "step": 4692 }, { "epoch": 0.82, "learning_rate": 3.812170934077379e-05, "loss": 0.3852, "step": 4693 }, { "epoch": 0.82, "learning_rate": 3.811567245900428e-05, "loss": 0.3999, "step": 4694 }, { "epoch": 0.82, "learning_rate": 3.8109634594049425e-05, "loss": 0.4037, "step": 4695 }, { "epoch": 0.82, "learning_rate": 3.810359574646843e-05, "loss": 0.4105, "step": 4696 }, { "epoch": 0.82, "learning_rate": 3.8097555916820584e-05, "loss": 0.4032, "step": 4697 }, { "epoch": 0.82, "learning_rate": 3.809151510566528e-05, "loss": 0.3927, "step": 4698 }, { "epoch": 0.82, "learning_rate": 3.8085473313562e-05, "loss": 0.3919, "step": 4699 }, { "epoch": 0.82, "learning_rate": 3.807943054107031e-05, "loss": 0.393, "step": 4700 }, { "epoch": 0.82, "learning_rate": 3.8073386788749866e-05, "loss": 0.3939, "step": 4701 }, { "epoch": 0.82, "learning_rate": 3.806734205716042e-05, "loss": 0.4036, "step": 4702 }, { "epoch": 0.82, "learning_rate": 3.806129634686181e-05, "loss": 0.4012, "step": 4703 }, { "epoch": 0.82, "learning_rate": 3.8055249658413985e-05, "loss": 0.417, "step": 4704 }, { "epoch": 0.82, "learning_rate": 3.804920199237694e-05, "loss": 0.4086, "step": 4705 }, { "epoch": 0.82, "learning_rate": 3.804315334931082e-05, "loss": 0.3924, "step": 4706 }, { "epoch": 0.82, "learning_rate": 3.80371037297758e-05, "loss": 0.426, "step": 4707 }, { "epoch": 0.82, "learning_rate": 3.803105313433218e-05, "loss": 0.3979, "step": 4708 }, { "epoch": 0.82, "learning_rate": 3.802500156354036e-05, "loss": 0.4143, "step": 4709 }, { "epoch": 0.82, "learning_rate": 3.801894901796081e-05, "loss": 0.412, "step": 4710 }, { "epoch": 0.82, "learning_rate": 3.801289549815408e-05, "loss": 0.4083, "step": 4711 }, { "epoch": 0.82, "learning_rate": 3.8006841004680835e-05, "loss": 0.4079, "step": 4712 }, { "epoch": 0.82, "learning_rate": 3.800078553810181e-05, "loss": 0.3923, "step": 4713 }, { "epoch": 0.82, "learning_rate": 3.799472909897787e-05, "loss": 0.3862, "step": 4714 }, { "epoch": 0.82, "learning_rate": 3.798867168786991e-05, "loss": 0.3953, "step": 4715 }, { "epoch": 0.82, "learning_rate": 3.798261330533896e-05, "loss": 0.393, "step": 4716 }, { "epoch": 0.82, "learning_rate": 3.7976553951946126e-05, "loss": 0.3978, "step": 4717 }, { "epoch": 0.82, "learning_rate": 3.797049362825259e-05, "loss": 0.4064, "step": 4718 }, { "epoch": 0.82, "learning_rate": 3.7964432334819655e-05, "loss": 0.3874, "step": 4719 }, { "epoch": 0.82, "learning_rate": 3.795837007220869e-05, "loss": 0.389, "step": 4720 }, { "epoch": 0.82, "learning_rate": 3.795230684098116e-05, "loss": 0.3873, "step": 4721 }, { "epoch": 0.82, "learning_rate": 3.79462426416986e-05, "loss": 0.4131, "step": 4722 }, { "epoch": 0.82, "learning_rate": 3.794017747492269e-05, "loss": 0.3835, "step": 4723 }, { "epoch": 0.82, "learning_rate": 3.793411134121513e-05, "loss": 0.4092, "step": 4724 }, { "epoch": 0.82, "learning_rate": 3.792804424113778e-05, "loss": 0.4086, "step": 4725 }, { "epoch": 0.82, "learning_rate": 3.792197617525252e-05, "loss": 0.4082, "step": 4726 }, { "epoch": 0.82, "learning_rate": 3.791590714412136e-05, "loss": 0.3979, "step": 4727 }, { "epoch": 0.82, "learning_rate": 3.790983714830642e-05, "loss": 0.3967, "step": 4728 }, { "epoch": 0.82, "learning_rate": 3.7903766188369836e-05, "loss": 0.4004, "step": 4729 }, { "epoch": 0.82, "learning_rate": 3.789769426487391e-05, "loss": 0.4174, "step": 4730 }, { "epoch": 0.82, "learning_rate": 3.789162137838098e-05, "loss": 0.3976, "step": 4731 }, { "epoch": 0.83, "learning_rate": 3.788554752945352e-05, "loss": 0.3746, "step": 4732 }, { "epoch": 0.83, "learning_rate": 3.7879472718654044e-05, "loss": 0.3909, "step": 4733 }, { "epoch": 0.83, "learning_rate": 3.7873396946545195e-05, "loss": 0.4057, "step": 4734 }, { "epoch": 0.83, "learning_rate": 3.7867320213689686e-05, "loss": 0.4178, "step": 4735 }, { "epoch": 0.83, "learning_rate": 3.786124252065031e-05, "loss": 0.3868, "step": 4736 }, { "epoch": 0.83, "learning_rate": 3.785516386798999e-05, "loss": 0.4019, "step": 4737 }, { "epoch": 0.83, "learning_rate": 3.7849084256271674e-05, "loss": 0.4127, "step": 4738 }, { "epoch": 0.83, "learning_rate": 3.784300368605846e-05, "loss": 0.4027, "step": 4739 }, { "epoch": 0.83, "learning_rate": 3.78369221579135e-05, "loss": 0.4147, "step": 4740 }, { "epoch": 0.83, "learning_rate": 3.783083967240003e-05, "loss": 0.4162, "step": 4741 }, { "epoch": 0.83, "learning_rate": 3.7824756230081406e-05, "loss": 0.4296, "step": 4742 }, { "epoch": 0.83, "learning_rate": 3.781867183152105e-05, "loss": 0.3917, "step": 4743 }, { "epoch": 0.83, "learning_rate": 3.781258647728247e-05, "loss": 0.3998, "step": 4744 }, { "epoch": 0.83, "learning_rate": 3.7806500167929274e-05, "loss": 0.3944, "step": 4745 }, { "epoch": 0.83, "learning_rate": 3.780041290402516e-05, "loss": 0.4046, "step": 4746 }, { "epoch": 0.83, "learning_rate": 3.77943246861339e-05, "loss": 0.3864, "step": 4747 }, { "epoch": 0.83, "learning_rate": 3.778823551481937e-05, "loss": 0.4131, "step": 4748 }, { "epoch": 0.83, "learning_rate": 3.778214539064552e-05, "loss": 0.3957, "step": 4749 }, { "epoch": 0.83, "learning_rate": 3.77760543141764e-05, "loss": 0.3993, "step": 4750 }, { "epoch": 0.83, "learning_rate": 3.776996228597614e-05, "loss": 0.4113, "step": 4751 }, { "epoch": 0.83, "learning_rate": 3.776386930660897e-05, "loss": 0.3999, "step": 4752 }, { "epoch": 0.83, "learning_rate": 3.775777537663918e-05, "loss": 0.4166, "step": 4753 }, { "epoch": 0.83, "learning_rate": 3.775168049663119e-05, "loss": 0.3849, "step": 4754 }, { "epoch": 0.83, "learning_rate": 3.774558466714948e-05, "loss": 0.4179, "step": 4755 }, { "epoch": 0.83, "learning_rate": 3.7739487888758615e-05, "loss": 0.41, "step": 4756 }, { "epoch": 0.83, "learning_rate": 3.7733390162023274e-05, "loss": 0.3977, "step": 4757 }, { "epoch": 0.83, "learning_rate": 3.772729148750818e-05, "loss": 0.3767, "step": 4758 }, { "epoch": 0.83, "learning_rate": 3.7721191865778196e-05, "loss": 0.3846, "step": 4759 }, { "epoch": 0.83, "learning_rate": 3.7715091297398224e-05, "loss": 0.4297, "step": 4760 }, { "epoch": 0.83, "learning_rate": 3.77089897829333e-05, "loss": 0.4134, "step": 4761 }, { "epoch": 0.83, "learning_rate": 3.7702887322948515e-05, "loss": 0.4033, "step": 4762 }, { "epoch": 0.83, "learning_rate": 3.769678391800905e-05, "loss": 0.4067, "step": 4763 }, { "epoch": 0.83, "learning_rate": 3.769067956868019e-05, "loss": 0.402, "step": 4764 }, { "epoch": 0.83, "learning_rate": 3.768457427552729e-05, "loss": 0.3989, "step": 4765 }, { "epoch": 0.83, "learning_rate": 3.7678468039115806e-05, "loss": 0.4119, "step": 4766 }, { "epoch": 0.83, "learning_rate": 3.767236086001127e-05, "loss": 0.3982, "step": 4767 }, { "epoch": 0.83, "learning_rate": 3.7666252738779315e-05, "loss": 0.3916, "step": 4768 }, { "epoch": 0.83, "learning_rate": 3.766014367598564e-05, "loss": 0.3934, "step": 4769 }, { "epoch": 0.83, "learning_rate": 3.765403367219606e-05, "loss": 0.4052, "step": 4770 }, { "epoch": 0.83, "learning_rate": 3.764792272797646e-05, "loss": 0.4228, "step": 4771 }, { "epoch": 0.83, "learning_rate": 3.7641810843892796e-05, "loss": 0.4112, "step": 4772 }, { "epoch": 0.83, "learning_rate": 3.7635698020511144e-05, "loss": 0.3958, "step": 4773 }, { "epoch": 0.83, "learning_rate": 3.7629584258397646e-05, "loss": 0.3916, "step": 4774 }, { "epoch": 0.83, "learning_rate": 3.762346955811855e-05, "loss": 0.382, "step": 4775 }, { "epoch": 0.83, "learning_rate": 3.761735392024016e-05, "loss": 0.4128, "step": 4776 }, { "epoch": 0.83, "learning_rate": 3.7611237345328886e-05, "loss": 0.3849, "step": 4777 }, { "epoch": 0.83, "learning_rate": 3.760511983395124e-05, "loss": 0.3951, "step": 4778 }, { "epoch": 0.83, "learning_rate": 3.759900138667378e-05, "loss": 0.3927, "step": 4779 }, { "epoch": 0.83, "learning_rate": 3.759288200406319e-05, "loss": 0.4028, "step": 4780 }, { "epoch": 0.83, "learning_rate": 3.7586761686686226e-05, "loss": 0.4028, "step": 4781 }, { "epoch": 0.83, "learning_rate": 3.758064043510972e-05, "loss": 0.41, "step": 4782 }, { "epoch": 0.83, "learning_rate": 3.757451824990061e-05, "loss": 0.3943, "step": 4783 }, { "epoch": 0.83, "learning_rate": 3.756839513162591e-05, "loss": 0.4068, "step": 4784 }, { "epoch": 0.83, "learning_rate": 3.756227108085271e-05, "loss": 0.4029, "step": 4785 }, { "epoch": 0.83, "learning_rate": 3.7556146098148205e-05, "loss": 0.4058, "step": 4786 }, { "epoch": 0.83, "learning_rate": 3.755002018407967e-05, "loss": 0.3951, "step": 4787 }, { "epoch": 0.83, "learning_rate": 3.754389333921446e-05, "loss": 0.4054, "step": 4788 }, { "epoch": 0.83, "learning_rate": 3.7537765564120036e-05, "loss": 0.4012, "step": 4789 }, { "epoch": 0.84, "learning_rate": 3.7531636859363914e-05, "loss": 0.4038, "step": 4790 }, { "epoch": 0.84, "learning_rate": 3.7525507225513724e-05, "loss": 0.3994, "step": 4791 }, { "epoch": 0.84, "learning_rate": 3.7519376663137154e-05, "loss": 0.3978, "step": 4792 }, { "epoch": 0.84, "learning_rate": 3.7513245172802014e-05, "loss": 0.3939, "step": 4793 }, { "epoch": 0.84, "learning_rate": 3.750711275507617e-05, "loss": 0.3886, "step": 4794 }, { "epoch": 0.84, "learning_rate": 3.7500979410527584e-05, "loss": 0.3772, "step": 4795 }, { "epoch": 0.84, "learning_rate": 3.749484513972431e-05, "loss": 0.3835, "step": 4796 }, { "epoch": 0.84, "learning_rate": 3.748870994323448e-05, "loss": 0.4045, "step": 4797 }, { "epoch": 0.84, "learning_rate": 3.748257382162631e-05, "loss": 0.4056, "step": 4798 }, { "epoch": 0.84, "learning_rate": 3.747643677546811e-05, "loss": 0.397, "step": 4799 }, { "epoch": 0.84, "learning_rate": 3.747029880532827e-05, "loss": 0.4078, "step": 4800 }, { "epoch": 0.84, "learning_rate": 3.746415991177526e-05, "loss": 0.3939, "step": 4801 }, { "epoch": 0.84, "learning_rate": 3.745802009537765e-05, "loss": 0.4116, "step": 4802 }, { "epoch": 0.84, "learning_rate": 3.745187935670409e-05, "loss": 0.3837, "step": 4803 }, { "epoch": 0.84, "learning_rate": 3.744573769632331e-05, "loss": 0.4095, "step": 4804 }, { "epoch": 0.84, "learning_rate": 3.743959511480412e-05, "loss": 0.4082, "step": 4805 }, { "epoch": 0.84, "learning_rate": 3.7433451612715444e-05, "loss": 0.3912, "step": 4806 }, { "epoch": 0.84, "learning_rate": 3.742730719062624e-05, "loss": 0.4142, "step": 4807 }, { "epoch": 0.84, "learning_rate": 3.742116184910562e-05, "loss": 0.3981, "step": 4808 }, { "epoch": 0.84, "learning_rate": 3.7415015588722716e-05, "loss": 0.3938, "step": 4809 }, { "epoch": 0.84, "learning_rate": 3.740886841004678e-05, "loss": 0.3809, "step": 4810 }, { "epoch": 0.84, "learning_rate": 3.7402720313647134e-05, "loss": 0.3929, "step": 4811 }, { "epoch": 0.84, "learning_rate": 3.739657130009322e-05, "loss": 0.3967, "step": 4812 }, { "epoch": 0.84, "learning_rate": 3.739042136995449e-05, "loss": 0.3903, "step": 4813 }, { "epoch": 0.84, "learning_rate": 3.738427052380058e-05, "loss": 0.3998, "step": 4814 }, { "epoch": 0.84, "learning_rate": 3.737811876220113e-05, "loss": 0.4022, "step": 4815 }, { "epoch": 0.84, "learning_rate": 3.737196608572589e-05, "loss": 0.3871, "step": 4816 }, { "epoch": 0.84, "learning_rate": 3.7365812494944716e-05, "loss": 0.3891, "step": 4817 }, { "epoch": 0.84, "learning_rate": 3.7359657990427524e-05, "loss": 0.401, "step": 4818 }, { "epoch": 0.84, "learning_rate": 3.7353502572744325e-05, "loss": 0.3894, "step": 4819 }, { "epoch": 0.84, "learning_rate": 3.73473462424652e-05, "loss": 0.3914, "step": 4820 }, { "epoch": 0.84, "learning_rate": 3.7341189000160344e-05, "loss": 0.4028, "step": 4821 }, { "epoch": 0.84, "learning_rate": 3.73350308464e-05, "loss": 0.4027, "step": 4822 }, { "epoch": 0.84, "learning_rate": 3.7328871781754525e-05, "loss": 0.384, "step": 4823 }, { "epoch": 0.84, "learning_rate": 3.732271180679436e-05, "loss": 0.4013, "step": 4824 }, { "epoch": 0.84, "learning_rate": 3.7316550922089984e-05, "loss": 0.4054, "step": 4825 }, { "epoch": 0.84, "learning_rate": 3.731038912821204e-05, "loss": 0.4063, "step": 4826 }, { "epoch": 0.84, "learning_rate": 3.730422642573118e-05, "loss": 0.396, "step": 4827 }, { "epoch": 0.84, "learning_rate": 3.7298062815218184e-05, "loss": 0.4211, "step": 4828 }, { "epoch": 0.84, "learning_rate": 3.72918982972439e-05, "loss": 0.3924, "step": 4829 }, { "epoch": 0.84, "learning_rate": 3.728573287237927e-05, "loss": 0.416, "step": 4830 }, { "epoch": 0.84, "learning_rate": 3.7279566541195305e-05, "loss": 0.4054, "step": 4831 }, { "epoch": 0.84, "learning_rate": 3.727339930426311e-05, "loss": 0.4097, "step": 4832 }, { "epoch": 0.84, "learning_rate": 3.726723116215387e-05, "loss": 0.4188, "step": 4833 }, { "epoch": 0.84, "learning_rate": 3.7261062115438865e-05, "loss": 0.3993, "step": 4834 }, { "epoch": 0.84, "learning_rate": 3.7254892164689435e-05, "loss": 0.4058, "step": 4835 }, { "epoch": 0.84, "learning_rate": 3.7248721310477046e-05, "loss": 0.4028, "step": 4836 }, { "epoch": 0.84, "learning_rate": 3.724254955337319e-05, "loss": 0.3976, "step": 4837 }, { "epoch": 0.84, "learning_rate": 3.723637689394949e-05, "loss": 0.408, "step": 4838 }, { "epoch": 0.84, "learning_rate": 3.723020333277763e-05, "loss": 0.4102, "step": 4839 }, { "epoch": 0.84, "learning_rate": 3.7224028870429384e-05, "loss": 0.3919, "step": 4840 }, { "epoch": 0.84, "learning_rate": 3.721785350747661e-05, "loss": 0.4038, "step": 4841 }, { "epoch": 0.84, "learning_rate": 3.721167724449124e-05, "loss": 0.4013, "step": 4842 }, { "epoch": 0.84, "learning_rate": 3.720550008204531e-05, "loss": 0.3995, "step": 4843 }, { "epoch": 0.84, "learning_rate": 3.7199322020710906e-05, "loss": 0.3909, "step": 4844 }, { "epoch": 0.84, "learning_rate": 3.719314306106025e-05, "loss": 0.4142, "step": 4845 }, { "epoch": 0.84, "learning_rate": 3.718696320366559e-05, "loss": 0.4066, "step": 4846 }, { "epoch": 0.85, "learning_rate": 3.718078244909928e-05, "loss": 0.3944, "step": 4847 }, { "epoch": 0.85, "learning_rate": 3.717460079793378e-05, "loss": 0.4091, "step": 4848 }, { "epoch": 0.85, "learning_rate": 3.71684182507416e-05, "loss": 0.4066, "step": 4849 }, { "epoch": 0.85, "learning_rate": 3.7162234808095345e-05, "loss": 0.4023, "step": 4850 }, { "epoch": 0.85, "learning_rate": 3.7156050470567695e-05, "loss": 0.4023, "step": 4851 }, { "epoch": 0.85, "learning_rate": 3.7149865238731444e-05, "loss": 0.3827, "step": 4852 }, { "epoch": 0.85, "learning_rate": 3.7143679113159425e-05, "loss": 0.4052, "step": 4853 }, { "epoch": 0.85, "learning_rate": 3.7137492094424576e-05, "loss": 0.418, "step": 4854 }, { "epoch": 0.85, "learning_rate": 3.713130418309993e-05, "loss": 0.4179, "step": 4855 }, { "epoch": 0.85, "learning_rate": 3.712511537975857e-05, "loss": 0.3777, "step": 4856 }, { "epoch": 0.85, "learning_rate": 3.711892568497371e-05, "loss": 0.3854, "step": 4857 }, { "epoch": 0.85, "learning_rate": 3.711273509931859e-05, "loss": 0.3875, "step": 4858 }, { "epoch": 0.85, "learning_rate": 3.710654362336656e-05, "loss": 0.3909, "step": 4859 }, { "epoch": 0.85, "learning_rate": 3.7100351257691066e-05, "loss": 0.4046, "step": 4860 }, { "epoch": 0.85, "learning_rate": 3.709415800286562e-05, "loss": 0.4192, "step": 4861 }, { "epoch": 0.85, "learning_rate": 3.708796385946382e-05, "loss": 0.4121, "step": 4862 }, { "epoch": 0.85, "learning_rate": 3.7081768828059335e-05, "loss": 0.3945, "step": 4863 }, { "epoch": 0.85, "learning_rate": 3.707557290922594e-05, "loss": 0.3915, "step": 4864 }, { "epoch": 0.85, "learning_rate": 3.7069376103537454e-05, "loss": 0.4055, "step": 4865 }, { "epoch": 0.85, "learning_rate": 3.7063178411567835e-05, "loss": 0.3996, "step": 4866 }, { "epoch": 0.85, "learning_rate": 3.705697983389108e-05, "loss": 0.3973, "step": 4867 }, { "epoch": 0.85, "learning_rate": 3.7050780371081265e-05, "loss": 0.385, "step": 4868 }, { "epoch": 0.85, "learning_rate": 3.7044580023712585e-05, "loss": 0.403, "step": 4869 }, { "epoch": 0.85, "learning_rate": 3.703837879235927e-05, "loss": 0.4038, "step": 4870 }, { "epoch": 0.85, "learning_rate": 3.703217667759567e-05, "loss": 0.3991, "step": 4871 }, { "epoch": 0.85, "learning_rate": 3.70259736799962e-05, "loss": 0.394, "step": 4872 }, { "epoch": 0.85, "learning_rate": 3.701976980013536e-05, "loss": 0.3924, "step": 4873 }, { "epoch": 0.85, "learning_rate": 3.701356503858773e-05, "loss": 0.4105, "step": 4874 }, { "epoch": 0.85, "learning_rate": 3.700735939592796e-05, "loss": 0.4086, "step": 4875 }, { "epoch": 0.85, "learning_rate": 3.700115287273082e-05, "loss": 0.402, "step": 4876 }, { "epoch": 0.85, "learning_rate": 3.699494546957112e-05, "loss": 0.415, "step": 4877 }, { "epoch": 0.85, "learning_rate": 3.698873718702378e-05, "loss": 0.3826, "step": 4878 }, { "epoch": 0.85, "learning_rate": 3.698252802566377e-05, "loss": 0.4012, "step": 4879 }, { "epoch": 0.85, "learning_rate": 3.697631798606616e-05, "loss": 0.3935, "step": 4880 }, { "epoch": 0.85, "learning_rate": 3.697010706880612e-05, "loss": 0.394, "step": 4881 }, { "epoch": 0.85, "learning_rate": 3.696389527445887e-05, "loss": 0.4189, "step": 4882 }, { "epoch": 0.85, "learning_rate": 3.695768260359973e-05, "loss": 0.4038, "step": 4883 }, { "epoch": 0.85, "learning_rate": 3.695146905680409e-05, "loss": 0.4096, "step": 4884 }, { "epoch": 0.85, "learning_rate": 3.6945254634647426e-05, "loss": 0.3776, "step": 4885 }, { "epoch": 0.85, "learning_rate": 3.69390393377053e-05, "loss": 0.4049, "step": 4886 }, { "epoch": 0.85, "learning_rate": 3.693282316655335e-05, "loss": 0.4035, "step": 4887 }, { "epoch": 0.85, "learning_rate": 3.692660612176729e-05, "loss": 0.3829, "step": 4888 }, { "epoch": 0.85, "learning_rate": 3.692038820392293e-05, "loss": 0.3871, "step": 4889 }, { "epoch": 0.85, "learning_rate": 3.691416941359614e-05, "loss": 0.4001, "step": 4890 }, { "epoch": 0.85, "learning_rate": 3.6907949751362894e-05, "loss": 0.3886, "step": 4891 }, { "epoch": 0.85, "learning_rate": 3.690172921779922e-05, "loss": 0.3959, "step": 4892 }, { "epoch": 0.85, "learning_rate": 3.689550781348125e-05, "loss": 0.4006, "step": 4893 }, { "epoch": 0.85, "learning_rate": 3.688928553898519e-05, "loss": 0.3979, "step": 4894 }, { "epoch": 0.85, "learning_rate": 3.6883062394887324e-05, "loss": 0.4063, "step": 4895 }, { "epoch": 0.85, "learning_rate": 3.6876838381764014e-05, "loss": 0.4048, "step": 4896 }, { "epoch": 0.85, "learning_rate": 3.68706135001917e-05, "loss": 0.4009, "step": 4897 }, { "epoch": 0.85, "learning_rate": 3.686438775074692e-05, "loss": 0.4201, "step": 4898 }, { "epoch": 0.85, "learning_rate": 3.685816113400628e-05, "loss": 0.391, "step": 4899 }, { "epoch": 0.85, "learning_rate": 3.685193365054646e-05, "loss": 0.3966, "step": 4900 }, { "epoch": 0.85, "learning_rate": 3.6845705300944225e-05, "loss": 0.3825, "step": 4901 }, { "epoch": 0.85, "learning_rate": 3.683947608577644e-05, "loss": 0.4042, "step": 4902 }, { "epoch": 0.85, "learning_rate": 3.683324600562001e-05, "loss": 0.3996, "step": 4903 }, { "epoch": 0.86, "learning_rate": 3.682701506105194e-05, "loss": 0.3975, "step": 4904 }, { "epoch": 0.86, "learning_rate": 3.682078325264935e-05, "loss": 0.3853, "step": 4905 }, { "epoch": 0.86, "learning_rate": 3.681455058098937e-05, "loss": 0.4021, "step": 4906 }, { "epoch": 0.86, "learning_rate": 3.6808317046649273e-05, "loss": 0.399, "step": 4907 }, { "epoch": 0.86, "learning_rate": 3.680208265020638e-05, "loss": 0.3928, "step": 4908 }, { "epoch": 0.86, "learning_rate": 3.6795847392238094e-05, "loss": 0.3904, "step": 4909 }, { "epoch": 0.86, "learning_rate": 3.67896112733219e-05, "loss": 0.4243, "step": 4910 }, { "epoch": 0.86, "learning_rate": 3.678337429403537e-05, "loss": 0.4072, "step": 4911 }, { "epoch": 0.86, "learning_rate": 3.6777136454956164e-05, "loss": 0.3902, "step": 4912 }, { "epoch": 0.86, "learning_rate": 3.6770897756661975e-05, "loss": 0.4044, "step": 4913 }, { "epoch": 0.86, "learning_rate": 3.676465819973064e-05, "loss": 0.3973, "step": 4914 }, { "epoch": 0.86, "learning_rate": 3.675841778474003e-05, "loss": 0.3848, "step": 4915 }, { "epoch": 0.86, "learning_rate": 3.675217651226811e-05, "loss": 0.4158, "step": 4916 }, { "epoch": 0.86, "learning_rate": 3.6745934382892926e-05, "loss": 0.3962, "step": 4917 }, { "epoch": 0.86, "learning_rate": 3.6739691397192606e-05, "loss": 0.3808, "step": 4918 }, { "epoch": 0.86, "learning_rate": 3.673344755574535e-05, "loss": 0.3899, "step": 4919 }, { "epoch": 0.86, "learning_rate": 3.672720285912943e-05, "loss": 0.3865, "step": 4920 }, { "epoch": 0.86, "learning_rate": 3.672095730792322e-05, "loss": 0.4103, "step": 4921 }, { "epoch": 0.86, "learning_rate": 3.671471090270516e-05, "loss": 0.4228, "step": 4922 }, { "epoch": 0.86, "learning_rate": 3.670846364405376e-05, "loss": 0.4167, "step": 4923 }, { "epoch": 0.86, "learning_rate": 3.6702215532547635e-05, "loss": 0.3809, "step": 4924 }, { "epoch": 0.86, "learning_rate": 3.669596656876544e-05, "loss": 0.4092, "step": 4925 }, { "epoch": 0.86, "learning_rate": 3.668971675328595e-05, "loss": 0.3992, "step": 4926 }, { "epoch": 0.86, "learning_rate": 3.6683466086688e-05, "loss": 0.4085, "step": 4927 }, { "epoch": 0.86, "learning_rate": 3.667721456955049e-05, "loss": 0.3972, "step": 4928 }, { "epoch": 0.86, "learning_rate": 3.667096220245242e-05, "loss": 0.3884, "step": 4929 }, { "epoch": 0.86, "learning_rate": 3.666470898597287e-05, "loss": 0.388, "step": 4930 }, { "epoch": 0.86, "learning_rate": 3.6658454920690975e-05, "loss": 0.3994, "step": 4931 }, { "epoch": 0.86, "learning_rate": 3.665220000718597e-05, "loss": 0.4111, "step": 4932 }, { "epoch": 0.86, "learning_rate": 3.6645944246037174e-05, "loss": 0.3999, "step": 4933 }, { "epoch": 0.86, "learning_rate": 3.663968763782396e-05, "loss": 0.3986, "step": 4934 }, { "epoch": 0.86, "learning_rate": 3.663343018312581e-05, "loss": 0.4144, "step": 4935 }, { "epoch": 0.86, "learning_rate": 3.662717188252224e-05, "loss": 0.3923, "step": 4936 }, { "epoch": 0.86, "learning_rate": 3.662091273659288e-05, "loss": 0.3959, "step": 4937 }, { "epoch": 0.86, "learning_rate": 3.661465274591744e-05, "loss": 0.4111, "step": 4938 }, { "epoch": 0.86, "learning_rate": 3.66083919110757e-05, "loss": 0.4116, "step": 4939 }, { "epoch": 0.86, "learning_rate": 3.6602130232647494e-05, "loss": 0.4018, "step": 4940 }, { "epoch": 0.86, "learning_rate": 3.659586771121278e-05, "loss": 0.4071, "step": 4941 }, { "epoch": 0.86, "learning_rate": 3.6589604347351566e-05, "loss": 0.4037, "step": 4942 }, { "epoch": 0.86, "learning_rate": 3.658334014164392e-05, "loss": 0.4033, "step": 4943 }, { "epoch": 0.86, "learning_rate": 3.6577075094670044e-05, "loss": 0.4005, "step": 4944 }, { "epoch": 0.86, "learning_rate": 3.657080920701016e-05, "loss": 0.4043, "step": 4945 }, { "epoch": 0.86, "learning_rate": 3.6564542479244594e-05, "loss": 0.4012, "step": 4946 }, { "epoch": 0.86, "learning_rate": 3.6558274911953754e-05, "loss": 0.3908, "step": 4947 }, { "epoch": 0.86, "learning_rate": 3.6552006505718116e-05, "loss": 0.4244, "step": 4948 }, { "epoch": 0.86, "learning_rate": 3.6545737261118254e-05, "loss": 0.3962, "step": 4949 }, { "epoch": 0.86, "learning_rate": 3.653946717873477e-05, "loss": 0.3988, "step": 4950 }, { "epoch": 0.86, "learning_rate": 3.65331962591484e-05, "loss": 0.4095, "step": 4951 }, { "epoch": 0.86, "learning_rate": 3.652692450293994e-05, "loss": 0.3781, "step": 4952 }, { "epoch": 0.86, "learning_rate": 3.652065191069023e-05, "loss": 0.3884, "step": 4953 }, { "epoch": 0.86, "learning_rate": 3.651437848298024e-05, "loss": 0.4052, "step": 4954 }, { "epoch": 0.86, "learning_rate": 3.650810422039098e-05, "loss": 0.4118, "step": 4955 }, { "epoch": 0.86, "learning_rate": 3.650182912350355e-05, "loss": 0.4069, "step": 4956 }, { "epoch": 0.86, "learning_rate": 3.649555319289914e-05, "loss": 0.3894, "step": 4957 }, { "epoch": 0.86, "learning_rate": 3.6489276429158984e-05, "loss": 0.4029, "step": 4958 }, { "epoch": 0.86, "learning_rate": 3.648299883286442e-05, "loss": 0.3926, "step": 4959 }, { "epoch": 0.86, "learning_rate": 3.647672040459687e-05, "loss": 0.4146, "step": 4960 }, { "epoch": 0.86, "learning_rate": 3.64704411449378e-05, "loss": 0.406, "step": 4961 }, { "epoch": 0.87, "learning_rate": 3.646416105446879e-05, "loss": 0.4199, "step": 4962 }, { "epoch": 0.87, "learning_rate": 3.645788013377147e-05, "loss": 0.4074, "step": 4963 }, { "epoch": 0.87, "learning_rate": 3.6451598383427546e-05, "loss": 0.4275, "step": 4964 }, { "epoch": 0.87, "learning_rate": 3.6445315804018835e-05, "loss": 0.3985, "step": 4965 }, { "epoch": 0.87, "learning_rate": 3.6439032396127184e-05, "loss": 0.3999, "step": 4966 }, { "epoch": 0.87, "learning_rate": 3.6432748160334554e-05, "loss": 0.4052, "step": 4967 }, { "epoch": 0.87, "learning_rate": 3.6426463097222956e-05, "loss": 0.3911, "step": 4968 }, { "epoch": 0.87, "learning_rate": 3.6420177207374514e-05, "loss": 0.4046, "step": 4969 }, { "epoch": 0.87, "learning_rate": 3.6413890491371375e-05, "loss": 0.4191, "step": 4970 }, { "epoch": 0.87, "learning_rate": 3.6407602949795797e-05, "loss": 0.3966, "step": 4971 }, { "epoch": 0.87, "learning_rate": 3.640131458323013e-05, "loss": 0.3863, "step": 4972 }, { "epoch": 0.87, "learning_rate": 3.639502539225677e-05, "loss": 0.4225, "step": 4973 }, { "epoch": 0.87, "learning_rate": 3.63887353774582e-05, "loss": 0.4048, "step": 4974 }, { "epoch": 0.87, "learning_rate": 3.638244453941697e-05, "loss": 0.4198, "step": 4975 }, { "epoch": 0.87, "learning_rate": 3.6376152878715714e-05, "loss": 0.3942, "step": 4976 }, { "epoch": 0.87, "learning_rate": 3.636986039593715e-05, "loss": 0.3849, "step": 4977 }, { "epoch": 0.87, "learning_rate": 3.6363567091664064e-05, "loss": 0.3929, "step": 4978 }, { "epoch": 0.87, "learning_rate": 3.635727296647932e-05, "loss": 0.3883, "step": 4979 }, { "epoch": 0.87, "learning_rate": 3.635097802096586e-05, "loss": 0.4311, "step": 4980 }, { "epoch": 0.87, "learning_rate": 3.634468225570669e-05, "loss": 0.3984, "step": 4981 }, { "epoch": 0.87, "learning_rate": 3.633838567128491e-05, "loss": 0.3793, "step": 4982 }, { "epoch": 0.87, "learning_rate": 3.6332088268283684e-05, "loss": 0.4029, "step": 4983 }, { "epoch": 0.87, "learning_rate": 3.6325790047286254e-05, "loss": 0.4092, "step": 4984 }, { "epoch": 0.87, "learning_rate": 3.6319491008875935e-05, "loss": 0.3929, "step": 4985 }, { "epoch": 0.87, "learning_rate": 3.6313191153636136e-05, "loss": 0.3993, "step": 4986 }, { "epoch": 0.87, "learning_rate": 3.6306890482150306e-05, "loss": 0.4023, "step": 4987 }, { "epoch": 0.87, "learning_rate": 3.6300588995002e-05, "loss": 0.3932, "step": 4988 }, { "epoch": 0.87, "learning_rate": 3.629428669277484e-05, "loss": 0.3904, "step": 4989 }, { "epoch": 0.87, "learning_rate": 3.628798357605253e-05, "loss": 0.381, "step": 4990 }, { "epoch": 0.87, "learning_rate": 3.628167964541883e-05, "loss": 0.398, "step": 4991 }, { "epoch": 0.87, "learning_rate": 3.627537490145758e-05, "loss": 0.4012, "step": 4992 }, { "epoch": 0.87, "learning_rate": 3.626906934475273e-05, "loss": 0.3911, "step": 4993 }, { "epoch": 0.87, "learning_rate": 3.626276297588826e-05, "loss": 0.3979, "step": 4994 }, { "epoch": 0.87, "learning_rate": 3.625645579544824e-05, "loss": 0.3929, "step": 4995 }, { "epoch": 0.87, "learning_rate": 3.625014780401682e-05, "loss": 0.3865, "step": 4996 }, { "epoch": 0.87, "learning_rate": 3.624383900217824e-05, "loss": 0.4091, "step": 4997 }, { "epoch": 0.87, "learning_rate": 3.623752939051677e-05, "loss": 0.3965, "step": 4998 }, { "epoch": 0.87, "learning_rate": 3.6231218969616804e-05, "loss": 0.4004, "step": 4999 }, { "epoch": 0.87, "learning_rate": 3.622490774006279e-05, "loss": 0.3994, "step": 5000 }, { "epoch": 0.87, "learning_rate": 3.6218595702439245e-05, "loss": 0.4109, "step": 5001 }, { "epoch": 0.87, "learning_rate": 3.621228285733077e-05, "loss": 0.4128, "step": 5002 }, { "epoch": 0.87, "learning_rate": 3.6205969205322034e-05, "loss": 0.3887, "step": 5003 }, { "epoch": 0.87, "learning_rate": 3.6199654746997786e-05, "loss": 0.406, "step": 5004 }, { "epoch": 0.87, "learning_rate": 3.619333948294285e-05, "loss": 0.4032, "step": 5005 }, { "epoch": 0.87, "learning_rate": 3.6187023413742134e-05, "loss": 0.3861, "step": 5006 }, { "epoch": 0.87, "learning_rate": 3.6180706539980584e-05, "loss": 0.3985, "step": 5007 }, { "epoch": 0.87, "learning_rate": 3.617438886224328e-05, "loss": 0.3917, "step": 5008 }, { "epoch": 0.87, "learning_rate": 3.61680703811153e-05, "loss": 0.4011, "step": 5009 }, { "epoch": 0.87, "learning_rate": 3.616175109718188e-05, "loss": 0.4003, "step": 5010 }, { "epoch": 0.87, "learning_rate": 3.615543101102826e-05, "loss": 0.395, "step": 5011 }, { "epoch": 0.87, "learning_rate": 3.61491101232398e-05, "loss": 0.3858, "step": 5012 }, { "epoch": 0.87, "learning_rate": 3.6142788434401914e-05, "loss": 0.4127, "step": 5013 }, { "epoch": 0.87, "learning_rate": 3.613646594510009e-05, "loss": 0.409, "step": 5014 }, { "epoch": 0.87, "learning_rate": 3.61301426559199e-05, "loss": 0.3881, "step": 5015 }, { "epoch": 0.87, "learning_rate": 3.612381856744698e-05, "loss": 0.4233, "step": 5016 }, { "epoch": 0.87, "learning_rate": 3.611749368026704e-05, "loss": 0.4073, "step": 5017 }, { "epoch": 0.87, "learning_rate": 3.6111167994965887e-05, "loss": 0.4093, "step": 5018 }, { "epoch": 0.88, "learning_rate": 3.610484151212936e-05, "loss": 0.4055, "step": 5019 }, { "epoch": 0.88, "learning_rate": 3.609851423234341e-05, "loss": 0.3825, "step": 5020 }, { "epoch": 0.88, "learning_rate": 3.609218615619405e-05, "loss": 0.3924, "step": 5021 }, { "epoch": 0.88, "learning_rate": 3.608585728426734e-05, "loss": 0.3924, "step": 5022 }, { "epoch": 0.88, "learning_rate": 3.607952761714946e-05, "loss": 0.3914, "step": 5023 }, { "epoch": 0.88, "learning_rate": 3.607319715542665e-05, "loss": 0.4056, "step": 5024 }, { "epoch": 0.88, "learning_rate": 3.606686589968517e-05, "loss": 0.3948, "step": 5025 }, { "epoch": 0.88, "learning_rate": 3.606053385051144e-05, "loss": 0.3937, "step": 5026 }, { "epoch": 0.88, "learning_rate": 3.60542010084919e-05, "loss": 0.3881, "step": 5027 }, { "epoch": 0.88, "learning_rate": 3.604786737421307e-05, "loss": 0.3941, "step": 5028 }, { "epoch": 0.88, "learning_rate": 3.604153294826156e-05, "loss": 0.4191, "step": 5029 }, { "epoch": 0.88, "learning_rate": 3.603519773122402e-05, "loss": 0.405, "step": 5030 }, { "epoch": 0.88, "learning_rate": 3.602886172368723e-05, "loss": 0.3995, "step": 5031 }, { "epoch": 0.88, "learning_rate": 3.602252492623796e-05, "loss": 0.3965, "step": 5032 }, { "epoch": 0.88, "learning_rate": 3.6016187339463145e-05, "loss": 0.3777, "step": 5033 }, { "epoch": 0.88, "learning_rate": 3.6009848963949725e-05, "loss": 0.4062, "step": 5034 }, { "epoch": 0.88, "learning_rate": 3.6003509800284745e-05, "loss": 0.3948, "step": 5035 }, { "epoch": 0.88, "learning_rate": 3.5997169849055327e-05, "loss": 0.4154, "step": 5036 }, { "epoch": 0.88, "learning_rate": 3.5990829110848634e-05, "loss": 0.3923, "step": 5037 }, { "epoch": 0.88, "learning_rate": 3.598448758625193e-05, "loss": 0.3866, "step": 5038 }, { "epoch": 0.88, "learning_rate": 3.5978145275852546e-05, "loss": 0.3913, "step": 5039 }, { "epoch": 0.88, "learning_rate": 3.597180218023788e-05, "loss": 0.3917, "step": 5040 }, { "epoch": 0.88, "learning_rate": 3.5965458299995426e-05, "loss": 0.4079, "step": 5041 }, { "epoch": 0.88, "learning_rate": 3.5959113635712696e-05, "loss": 0.4017, "step": 5042 }, { "epoch": 0.88, "learning_rate": 3.595276818797735e-05, "loss": 0.4032, "step": 5043 }, { "epoch": 0.88, "learning_rate": 3.594642195737704e-05, "loss": 0.4071, "step": 5044 }, { "epoch": 0.88, "learning_rate": 3.594007494449955e-05, "loss": 0.3879, "step": 5045 }, { "epoch": 0.88, "learning_rate": 3.593372714993274e-05, "loss": 0.4136, "step": 5046 }, { "epoch": 0.88, "learning_rate": 3.592737857426448e-05, "loss": 0.4117, "step": 5047 }, { "epoch": 0.88, "learning_rate": 3.592102921808278e-05, "loss": 0.3853, "step": 5048 }, { "epoch": 0.88, "learning_rate": 3.591467908197568e-05, "loss": 0.4058, "step": 5049 }, { "epoch": 0.88, "learning_rate": 3.5908328166531305e-05, "loss": 0.3821, "step": 5050 }, { "epoch": 0.88, "learning_rate": 3.590197647233787e-05, "loss": 0.3971, "step": 5051 }, { "epoch": 0.88, "learning_rate": 3.589562399998364e-05, "loss": 0.4073, "step": 5052 }, { "epoch": 0.88, "learning_rate": 3.5889270750056945e-05, "loss": 0.401, "step": 5053 }, { "epoch": 0.88, "learning_rate": 3.588291672314621e-05, "loss": 0.4038, "step": 5054 }, { "epoch": 0.88, "learning_rate": 3.587656191983993e-05, "loss": 0.43, "step": 5055 }, { "epoch": 0.88, "learning_rate": 3.587020634072665e-05, "loss": 0.3981, "step": 5056 }, { "epoch": 0.88, "learning_rate": 3.586384998639501e-05, "loss": 0.4035, "step": 5057 }, { "epoch": 0.88, "learning_rate": 3.585749285743371e-05, "loss": 0.3839, "step": 5058 }, { "epoch": 0.88, "learning_rate": 3.585113495443153e-05, "loss": 0.3928, "step": 5059 }, { "epoch": 0.88, "learning_rate": 3.58447762779773e-05, "loss": 0.3893, "step": 5060 }, { "epoch": 0.88, "learning_rate": 3.583841682865995e-05, "loss": 0.4035, "step": 5061 }, { "epoch": 0.88, "learning_rate": 3.5832056607068464e-05, "loss": 0.4034, "step": 5062 }, { "epoch": 0.88, "learning_rate": 3.582569561379191e-05, "loss": 0.3996, "step": 5063 }, { "epoch": 0.88, "learning_rate": 3.5819333849419406e-05, "loss": 0.3885, "step": 5064 }, { "epoch": 0.88, "learning_rate": 3.581297131454018e-05, "loss": 0.3964, "step": 5065 }, { "epoch": 0.88, "learning_rate": 3.5806608009743493e-05, "loss": 0.402, "step": 5066 }, { "epoch": 0.88, "learning_rate": 3.580024393561868e-05, "loss": 0.3904, "step": 5067 }, { "epoch": 0.88, "learning_rate": 3.5793879092755175e-05, "loss": 0.4009, "step": 5068 }, { "epoch": 0.88, "learning_rate": 3.5787513481742465e-05, "loss": 0.3941, "step": 5069 }, { "epoch": 0.88, "learning_rate": 3.57811471031701e-05, "loss": 0.3991, "step": 5070 }, { "epoch": 0.88, "learning_rate": 3.5774779957627726e-05, "loss": 0.4112, "step": 5071 }, { "epoch": 0.88, "learning_rate": 3.576841204570503e-05, "loss": 0.4023, "step": 5072 }, { "epoch": 0.88, "learning_rate": 3.576204336799179e-05, "loss": 0.3917, "step": 5073 }, { "epoch": 0.88, "learning_rate": 3.575567392507786e-05, "loss": 0.3916, "step": 5074 }, { "epoch": 0.88, "learning_rate": 3.574930371755314e-05, "loss": 0.4206, "step": 5075 }, { "epoch": 0.88, "learning_rate": 3.5742932746007626e-05, "loss": 0.4021, "step": 5076 }, { "epoch": 0.89, "learning_rate": 3.573656101103137e-05, "loss": 0.3828, "step": 5077 }, { "epoch": 0.89, "learning_rate": 3.57301885132145e-05, "loss": 0.3949, "step": 5078 }, { "epoch": 0.89, "learning_rate": 3.572381525314721e-05, "loss": 0.395, "step": 5079 }, { "epoch": 0.89, "learning_rate": 3.571744123141978e-05, "loss": 0.4174, "step": 5080 }, { "epoch": 0.89, "learning_rate": 3.5711066448622535e-05, "loss": 0.3861, "step": 5081 }, { "epoch": 0.89, "learning_rate": 3.5704690905345883e-05, "loss": 0.392, "step": 5082 }, { "epoch": 0.89, "learning_rate": 3.569831460218032e-05, "loss": 0.3831, "step": 5083 }, { "epoch": 0.89, "learning_rate": 3.569193753971639e-05, "loss": 0.4086, "step": 5084 }, { "epoch": 0.89, "learning_rate": 3.568555971854471e-05, "loss": 0.3894, "step": 5085 }, { "epoch": 0.89, "learning_rate": 3.567918113925597e-05, "loss": 0.3988, "step": 5086 }, { "epoch": 0.89, "learning_rate": 3.567280180244092e-05, "loss": 0.4141, "step": 5087 }, { "epoch": 0.89, "learning_rate": 3.566642170869042e-05, "loss": 0.3988, "step": 5088 }, { "epoch": 0.89, "learning_rate": 3.5660040858595364e-05, "loss": 0.4158, "step": 5089 }, { "epoch": 0.89, "learning_rate": 3.56536592527467e-05, "loss": 0.4095, "step": 5090 }, { "epoch": 0.89, "learning_rate": 3.5647276891735485e-05, "loss": 0.3912, "step": 5091 }, { "epoch": 0.89, "learning_rate": 3.564089377615283e-05, "loss": 0.4009, "step": 5092 }, { "epoch": 0.89, "learning_rate": 3.563450990658991e-05, "loss": 0.3957, "step": 5093 }, { "epoch": 0.89, "learning_rate": 3.5628125283637995e-05, "loss": 0.4285, "step": 5094 }, { "epoch": 0.89, "learning_rate": 3.562173990788839e-05, "loss": 0.4122, "step": 5095 }, { "epoch": 0.89, "learning_rate": 3.5615353779932484e-05, "loss": 0.4047, "step": 5096 }, { "epoch": 0.89, "learning_rate": 3.560896690036174e-05, "loss": 0.411, "step": 5097 }, { "epoch": 0.89, "learning_rate": 3.560257926976769e-05, "loss": 0.4092, "step": 5098 }, { "epoch": 0.89, "learning_rate": 3.5596190888741936e-05, "loss": 0.4049, "step": 5099 }, { "epoch": 0.89, "learning_rate": 3.558980175787613e-05, "loss": 0.407, "step": 5100 }, { "epoch": 0.89, "learning_rate": 3.558341187776204e-05, "loss": 0.3956, "step": 5101 }, { "epoch": 0.89, "learning_rate": 3.5577021248991446e-05, "loss": 0.399, "step": 5102 }, { "epoch": 0.89, "learning_rate": 3.557062987215624e-05, "loss": 0.4015, "step": 5103 }, { "epoch": 0.89, "learning_rate": 3.5564237747848366e-05, "loss": 0.4096, "step": 5104 }, { "epoch": 0.89, "learning_rate": 3.5557844876659836e-05, "loss": 0.4069, "step": 5105 }, { "epoch": 0.89, "learning_rate": 3.555145125918273e-05, "loss": 0.3919, "step": 5106 }, { "epoch": 0.89, "learning_rate": 3.554505689600922e-05, "loss": 0.4051, "step": 5107 }, { "epoch": 0.89, "learning_rate": 3.5538661787731506e-05, "loss": 0.4263, "step": 5108 }, { "epoch": 0.89, "learning_rate": 3.55322659349419e-05, "loss": 0.4125, "step": 5109 }, { "epoch": 0.89, "learning_rate": 3.552586933823276e-05, "loss": 0.3976, "step": 5110 }, { "epoch": 0.89, "learning_rate": 3.55194719981965e-05, "loss": 0.3954, "step": 5111 }, { "epoch": 0.89, "learning_rate": 3.551307391542562e-05, "loss": 0.4061, "step": 5112 }, { "epoch": 0.89, "learning_rate": 3.550667509051271e-05, "loss": 0.3887, "step": 5113 }, { "epoch": 0.89, "learning_rate": 3.550027552405038e-05, "loss": 0.4033, "step": 5114 }, { "epoch": 0.89, "learning_rate": 3.549387521663136e-05, "loss": 0.3963, "step": 5115 }, { "epoch": 0.89, "learning_rate": 3.5487474168848406e-05, "loss": 0.3859, "step": 5116 }, { "epoch": 0.89, "learning_rate": 3.548107238129435e-05, "loss": 0.3962, "step": 5117 }, { "epoch": 0.89, "learning_rate": 3.547466985456214e-05, "loss": 0.3876, "step": 5118 }, { "epoch": 0.89, "learning_rate": 3.546826658924471e-05, "loss": 0.3853, "step": 5119 }, { "epoch": 0.89, "learning_rate": 3.546186258593514e-05, "loss": 0.3916, "step": 5120 }, { "epoch": 0.89, "learning_rate": 3.545545784522654e-05, "loss": 0.3955, "step": 5121 }, { "epoch": 0.89, "learning_rate": 3.5449052367712074e-05, "loss": 0.398, "step": 5122 }, { "epoch": 0.89, "learning_rate": 3.5442646153985024e-05, "loss": 0.3855, "step": 5123 }, { "epoch": 0.89, "learning_rate": 3.5436239204638686e-05, "loss": 0.3838, "step": 5124 }, { "epoch": 0.89, "learning_rate": 3.542983152026646e-05, "loss": 0.3852, "step": 5125 }, { "epoch": 0.89, "learning_rate": 3.5423423101461794e-05, "loss": 0.3985, "step": 5126 }, { "epoch": 0.89, "learning_rate": 3.541701394881823e-05, "loss": 0.398, "step": 5127 }, { "epoch": 0.89, "learning_rate": 3.5410604062929344e-05, "loss": 0.3932, "step": 5128 }, { "epoch": 0.89, "learning_rate": 3.5404193444388805e-05, "loss": 0.389, "step": 5129 }, { "epoch": 0.89, "learning_rate": 3.539778209379034e-05, "loss": 0.3939, "step": 5130 }, { "epoch": 0.89, "learning_rate": 3.539137001172774e-05, "loss": 0.3942, "step": 5131 }, { "epoch": 0.89, "learning_rate": 3.538495719879487e-05, "loss": 0.4008, "step": 5132 }, { "epoch": 0.89, "learning_rate": 3.5378543655585664e-05, "loss": 0.3883, "step": 5133 }, { "epoch": 0.9, "learning_rate": 3.537212938269413e-05, "loss": 0.4183, "step": 5134 }, { "epoch": 0.9, "learning_rate": 3.5365714380714324e-05, "loss": 0.3903, "step": 5135 }, { "epoch": 0.9, "learning_rate": 3.535929865024039e-05, "loss": 0.4046, "step": 5136 }, { "epoch": 0.9, "learning_rate": 3.535288219186651e-05, "loss": 0.3964, "step": 5137 }, { "epoch": 0.9, "learning_rate": 3.5346465006186975e-05, "loss": 0.4016, "step": 5138 }, { "epoch": 0.9, "learning_rate": 3.534004709379611e-05, "loss": 0.4126, "step": 5139 }, { "epoch": 0.9, "learning_rate": 3.533362845528832e-05, "loss": 0.3828, "step": 5140 }, { "epoch": 0.9, "learning_rate": 3.532720909125808e-05, "loss": 0.4229, "step": 5141 }, { "epoch": 0.9, "learning_rate": 3.532078900229992e-05, "loss": 0.3834, "step": 5142 }, { "epoch": 0.9, "learning_rate": 3.531436818900846e-05, "loss": 0.4047, "step": 5143 }, { "epoch": 0.9, "learning_rate": 3.5307946651978356e-05, "loss": 0.3941, "step": 5144 }, { "epoch": 0.9, "learning_rate": 3.5301524391804366e-05, "loss": 0.379, "step": 5145 }, { "epoch": 0.9, "learning_rate": 3.529510140908129e-05, "loss": 0.4061, "step": 5146 }, { "epoch": 0.9, "learning_rate": 3.528867770440398e-05, "loss": 0.4121, "step": 5147 }, { "epoch": 0.9, "learning_rate": 3.528225327836741e-05, "loss": 0.3935, "step": 5148 }, { "epoch": 0.9, "learning_rate": 3.527582813156656e-05, "loss": 0.3805, "step": 5149 }, { "epoch": 0.9, "learning_rate": 3.526940226459652e-05, "loss": 0.4015, "step": 5150 }, { "epoch": 0.9, "learning_rate": 3.526297567805243e-05, "loss": 0.3962, "step": 5151 }, { "epoch": 0.9, "learning_rate": 3.5256548372529496e-05, "loss": 0.3906, "step": 5152 }, { "epoch": 0.9, "learning_rate": 3.5250120348622984e-05, "loss": 0.3855, "step": 5153 }, { "epoch": 0.9, "learning_rate": 3.524369160692825e-05, "loss": 0.3859, "step": 5154 }, { "epoch": 0.9, "learning_rate": 3.523726214804067e-05, "loss": 0.4196, "step": 5155 }, { "epoch": 0.9, "learning_rate": 3.5230831972555755e-05, "loss": 0.3944, "step": 5156 }, { "epoch": 0.9, "learning_rate": 3.5224401081069026e-05, "loss": 0.4014, "step": 5157 }, { "epoch": 0.9, "learning_rate": 3.521796947417609e-05, "loss": 0.3817, "step": 5158 }, { "epoch": 0.9, "learning_rate": 3.521153715247261e-05, "loss": 0.3905, "step": 5159 }, { "epoch": 0.9, "learning_rate": 3.5205104116554346e-05, "loss": 0.3979, "step": 5160 }, { "epoch": 0.9, "learning_rate": 3.519867036701709e-05, "loss": 0.3968, "step": 5161 }, { "epoch": 0.9, "learning_rate": 3.519223590445671e-05, "loss": 0.4035, "step": 5162 }, { "epoch": 0.9, "learning_rate": 3.5185800729469156e-05, "loss": 0.3858, "step": 5163 }, { "epoch": 0.9, "learning_rate": 3.517936484265041e-05, "loss": 0.4038, "step": 5164 }, { "epoch": 0.9, "learning_rate": 3.517292824459656e-05, "loss": 0.4144, "step": 5165 }, { "epoch": 0.9, "learning_rate": 3.516649093590372e-05, "loss": 0.3935, "step": 5166 }, { "epoch": 0.9, "learning_rate": 3.516005291716812e-05, "loss": 0.3955, "step": 5167 }, { "epoch": 0.9, "learning_rate": 3.5153614188986e-05, "loss": 0.3929, "step": 5168 }, { "epoch": 0.9, "learning_rate": 3.51471747519537e-05, "loss": 0.3859, "step": 5169 }, { "epoch": 0.9, "learning_rate": 3.514073460666762e-05, "loss": 0.4031, "step": 5170 }, { "epoch": 0.9, "learning_rate": 3.5134293753724226e-05, "loss": 0.3982, "step": 5171 }, { "epoch": 0.9, "learning_rate": 3.512785219372004e-05, "loss": 0.4092, "step": 5172 }, { "epoch": 0.9, "learning_rate": 3.512140992725165e-05, "loss": 0.4099, "step": 5173 }, { "epoch": 0.9, "learning_rate": 3.511496695491574e-05, "loss": 0.3737, "step": 5174 }, { "epoch": 0.9, "learning_rate": 3.5108523277309e-05, "loss": 0.3964, "step": 5175 }, { "epoch": 0.9, "learning_rate": 3.510207889502825e-05, "loss": 0.3959, "step": 5176 }, { "epoch": 0.9, "learning_rate": 3.509563380867034e-05, "loss": 0.4024, "step": 5177 }, { "epoch": 0.9, "learning_rate": 3.508918801883216e-05, "loss": 0.4141, "step": 5178 }, { "epoch": 0.9, "learning_rate": 3.508274152611074e-05, "loss": 0.3875, "step": 5179 }, { "epoch": 0.9, "learning_rate": 3.507629433110311e-05, "loss": 0.4181, "step": 5180 }, { "epoch": 0.9, "learning_rate": 3.506984643440638e-05, "loss": 0.3987, "step": 5181 }, { "epoch": 0.9, "learning_rate": 3.506339783661775e-05, "loss": 0.4026, "step": 5182 }, { "epoch": 0.9, "learning_rate": 3.5056948538334444e-05, "loss": 0.4016, "step": 5183 }, { "epoch": 0.9, "learning_rate": 3.505049854015379e-05, "loss": 0.3995, "step": 5184 }, { "epoch": 0.9, "learning_rate": 3.5044047842673154e-05, "loss": 0.4093, "step": 5185 }, { "epoch": 0.9, "learning_rate": 3.503759644648998e-05, "loss": 0.3875, "step": 5186 }, { "epoch": 0.9, "learning_rate": 3.503114435220177e-05, "loss": 0.3877, "step": 5187 }, { "epoch": 0.9, "learning_rate": 3.5024691560406104e-05, "loss": 0.4057, "step": 5188 }, { "epoch": 0.9, "learning_rate": 3.50182380717006e-05, "loss": 0.3968, "step": 5189 }, { "epoch": 0.9, "learning_rate": 3.501178388668297e-05, "loss": 0.4037, "step": 5190 }, { "epoch": 0.91, "learning_rate": 3.500532900595097e-05, "loss": 0.3996, "step": 5191 }, { "epoch": 0.91, "learning_rate": 3.499887343010243e-05, "loss": 0.4138, "step": 5192 }, { "epoch": 0.91, "learning_rate": 3.499241715973524e-05, "loss": 0.4087, "step": 5193 }, { "epoch": 0.91, "learning_rate": 3.498596019544737e-05, "loss": 0.3911, "step": 5194 }, { "epoch": 0.91, "learning_rate": 3.497950253783682e-05, "loss": 0.3954, "step": 5195 }, { "epoch": 0.91, "learning_rate": 3.497304418750169e-05, "loss": 0.384, "step": 5196 }, { "epoch": 0.91, "learning_rate": 3.496658514504012e-05, "loss": 0.406, "step": 5197 }, { "epoch": 0.91, "learning_rate": 3.496012541105032e-05, "loss": 0.3963, "step": 5198 }, { "epoch": 0.91, "learning_rate": 3.495366498613059e-05, "loss": 0.3884, "step": 5199 }, { "epoch": 0.91, "learning_rate": 3.494720387087925e-05, "loss": 0.3808, "step": 5200 }, { "epoch": 0.91, "learning_rate": 3.4940742065894715e-05, "loss": 0.4003, "step": 5201 }, { "epoch": 0.91, "learning_rate": 3.493427957177543e-05, "loss": 0.3902, "step": 5202 }, { "epoch": 0.91, "learning_rate": 3.4927816389119975e-05, "loss": 0.3656, "step": 5203 }, { "epoch": 0.91, "learning_rate": 3.492135251852691e-05, "loss": 0.4004, "step": 5204 }, { "epoch": 0.91, "learning_rate": 3.49148879605949e-05, "loss": 0.3929, "step": 5205 }, { "epoch": 0.91, "learning_rate": 3.4908422715922675e-05, "loss": 0.3861, "step": 5206 }, { "epoch": 0.91, "learning_rate": 3.490195678510902e-05, "loss": 0.3938, "step": 5207 }, { "epoch": 0.91, "learning_rate": 3.4895490168752806e-05, "loss": 0.411, "step": 5208 }, { "epoch": 0.91, "learning_rate": 3.488902286745292e-05, "loss": 0.4081, "step": 5209 }, { "epoch": 0.91, "learning_rate": 3.4882554881808344e-05, "loss": 0.3792, "step": 5210 }, { "epoch": 0.91, "learning_rate": 3.487608621241812e-05, "loss": 0.4129, "step": 5211 }, { "epoch": 0.91, "learning_rate": 3.486961685988137e-05, "loss": 0.4016, "step": 5212 }, { "epoch": 0.91, "learning_rate": 3.486314682479725e-05, "loss": 0.4072, "step": 5213 }, { "epoch": 0.91, "learning_rate": 3.485667610776498e-05, "loss": 0.4304, "step": 5214 }, { "epoch": 0.91, "learning_rate": 3.485020470938389e-05, "loss": 0.4074, "step": 5215 }, { "epoch": 0.91, "learning_rate": 3.484373263025329e-05, "loss": 0.3904, "step": 5216 }, { "epoch": 0.91, "learning_rate": 3.483725987097264e-05, "loss": 0.4122, "step": 5217 }, { "epoch": 0.91, "learning_rate": 3.4830786432141404e-05, "loss": 0.3937, "step": 5218 }, { "epoch": 0.91, "learning_rate": 3.4824312314359125e-05, "loss": 0.3907, "step": 5219 }, { "epoch": 0.91, "learning_rate": 3.4817837518225435e-05, "loss": 0.3872, "step": 5220 }, { "epoch": 0.91, "learning_rate": 3.4811362044339986e-05, "loss": 0.4062, "step": 5221 }, { "epoch": 0.91, "learning_rate": 3.480488589330252e-05, "loss": 0.4044, "step": 5222 }, { "epoch": 0.91, "learning_rate": 3.479840906571282e-05, "loss": 0.4133, "step": 5223 }, { "epoch": 0.91, "learning_rate": 3.4791931562170766e-05, "loss": 0.376, "step": 5224 }, { "epoch": 0.91, "learning_rate": 3.478545338327629e-05, "loss": 0.3841, "step": 5225 }, { "epoch": 0.91, "learning_rate": 3.477897452962935e-05, "loss": 0.3872, "step": 5226 }, { "epoch": 0.91, "learning_rate": 3.477249500183001e-05, "loss": 0.3932, "step": 5227 }, { "epoch": 0.91, "learning_rate": 3.4766014800478374e-05, "loss": 0.3975, "step": 5228 }, { "epoch": 0.91, "learning_rate": 3.475953392617462e-05, "loss": 0.4012, "step": 5229 }, { "epoch": 0.91, "learning_rate": 3.475305237951898e-05, "loss": 0.3835, "step": 5230 }, { "epoch": 0.91, "learning_rate": 3.474657016111175e-05, "loss": 0.3979, "step": 5231 }, { "epoch": 0.91, "learning_rate": 3.474008727155331e-05, "loss": 0.3892, "step": 5232 }, { "epoch": 0.91, "learning_rate": 3.4733603711444044e-05, "loss": 0.3993, "step": 5233 }, { "epoch": 0.91, "learning_rate": 3.472711948138447e-05, "loss": 0.3786, "step": 5234 }, { "epoch": 0.91, "learning_rate": 3.472063458197511e-05, "loss": 0.3924, "step": 5235 }, { "epoch": 0.91, "learning_rate": 3.4714149013816586e-05, "loss": 0.4057, "step": 5236 }, { "epoch": 0.91, "learning_rate": 3.470766277750957e-05, "loss": 0.3888, "step": 5237 }, { "epoch": 0.91, "learning_rate": 3.470117587365477e-05, "loss": 0.3787, "step": 5238 }, { "epoch": 0.91, "learning_rate": 3.4694688302853024e-05, "loss": 0.3773, "step": 5239 }, { "epoch": 0.91, "learning_rate": 3.468820006570514e-05, "loss": 0.4083, "step": 5240 }, { "epoch": 0.91, "learning_rate": 3.468171116281207e-05, "loss": 0.3839, "step": 5241 }, { "epoch": 0.91, "learning_rate": 3.467522159477478e-05, "loss": 0.3898, "step": 5242 }, { "epoch": 0.91, "learning_rate": 3.4668731362194304e-05, "loss": 0.4109, "step": 5243 }, { "epoch": 0.91, "learning_rate": 3.466224046567175e-05, "loss": 0.3995, "step": 5244 }, { "epoch": 0.91, "learning_rate": 3.4655748905808284e-05, "loss": 0.381, "step": 5245 }, { "epoch": 0.91, "learning_rate": 3.464925668320512e-05, "loss": 0.4055, "step": 5246 }, { "epoch": 0.91, "learning_rate": 3.464276379846356e-05, "loss": 0.4074, "step": 5247 }, { "epoch": 0.91, "learning_rate": 3.463627025218495e-05, "loss": 0.3921, "step": 5248 }, { "epoch": 0.92, "learning_rate": 3.462977604497068e-05, "loss": 0.4027, "step": 5249 }, { "epoch": 0.92, "learning_rate": 3.4623281177422244e-05, "loss": 0.4073, "step": 5250 }, { "epoch": 0.92, "learning_rate": 3.461678565014116e-05, "loss": 0.4073, "step": 5251 }, { "epoch": 0.92, "learning_rate": 3.461028946372901e-05, "loss": 0.4044, "step": 5252 }, { "epoch": 0.92, "learning_rate": 3.460379261878748e-05, "loss": 0.3968, "step": 5253 }, { "epoch": 0.92, "learning_rate": 3.4597295115918256e-05, "loss": 0.3884, "step": 5254 }, { "epoch": 0.92, "learning_rate": 3.459079695572311e-05, "loss": 0.4061, "step": 5255 }, { "epoch": 0.92, "learning_rate": 3.458429813880391e-05, "loss": 0.3901, "step": 5256 }, { "epoch": 0.92, "learning_rate": 3.457779866576252e-05, "loss": 0.372, "step": 5257 }, { "epoch": 0.92, "learning_rate": 3.457129853720091e-05, "loss": 0.3871, "step": 5258 }, { "epoch": 0.92, "learning_rate": 3.4564797753721104e-05, "loss": 0.3851, "step": 5259 }, { "epoch": 0.92, "learning_rate": 3.4558296315925176e-05, "loss": 0.3959, "step": 5260 }, { "epoch": 0.92, "learning_rate": 3.455179422441526e-05, "loss": 0.4092, "step": 5261 }, { "epoch": 0.92, "learning_rate": 3.454529147979357e-05, "loss": 0.4025, "step": 5262 }, { "epoch": 0.92, "learning_rate": 3.4538788082662355e-05, "loss": 0.402, "step": 5263 }, { "epoch": 0.92, "learning_rate": 3.453228403362393e-05, "loss": 0.3881, "step": 5264 }, { "epoch": 0.92, "learning_rate": 3.452577933328071e-05, "loss": 0.4095, "step": 5265 }, { "epoch": 0.92, "learning_rate": 3.45192739822351e-05, "loss": 0.408, "step": 5266 }, { "epoch": 0.92, "learning_rate": 3.4512767981089615e-05, "loss": 0.4012, "step": 5267 }, { "epoch": 0.92, "learning_rate": 3.450626133044682e-05, "loss": 0.3901, "step": 5268 }, { "epoch": 0.92, "learning_rate": 3.449975403090934e-05, "loss": 0.3956, "step": 5269 }, { "epoch": 0.92, "learning_rate": 3.4493246083079854e-05, "loss": 0.4125, "step": 5270 }, { "epoch": 0.92, "learning_rate": 3.44867374875611e-05, "loss": 0.4189, "step": 5271 }, { "epoch": 0.92, "learning_rate": 3.448022824495589e-05, "loss": 0.4023, "step": 5272 }, { "epoch": 0.92, "learning_rate": 3.4473718355867073e-05, "loss": 0.4056, "step": 5273 }, { "epoch": 0.92, "learning_rate": 3.4467207820897576e-05, "loss": 0.384, "step": 5274 }, { "epoch": 0.92, "learning_rate": 3.4460696640650404e-05, "loss": 0.4022, "step": 5275 }, { "epoch": 0.92, "learning_rate": 3.445418481572858e-05, "loss": 0.3982, "step": 5276 }, { "epoch": 0.92, "learning_rate": 3.444767234673519e-05, "loss": 0.3942, "step": 5277 }, { "epoch": 0.92, "learning_rate": 3.4441159234273415e-05, "loss": 0.394, "step": 5278 }, { "epoch": 0.92, "learning_rate": 3.443464547894647e-05, "loss": 0.3918, "step": 5279 }, { "epoch": 0.92, "learning_rate": 3.442813108135765e-05, "loss": 0.3716, "step": 5280 }, { "epoch": 0.92, "learning_rate": 3.442161604211028e-05, "loss": 0.4216, "step": 5281 }, { "epoch": 0.92, "learning_rate": 3.441510036180776e-05, "loss": 0.3888, "step": 5282 }, { "epoch": 0.92, "learning_rate": 3.4408584041053557e-05, "loss": 0.4054, "step": 5283 }, { "epoch": 0.92, "learning_rate": 3.440206708045117e-05, "loss": 0.4041, "step": 5284 }, { "epoch": 0.92, "learning_rate": 3.4395549480604204e-05, "loss": 0.3754, "step": 5285 }, { "epoch": 0.92, "learning_rate": 3.4389031242116274e-05, "loss": 0.3853, "step": 5286 }, { "epoch": 0.92, "learning_rate": 3.438251236559109e-05, "loss": 0.3952, "step": 5287 }, { "epoch": 0.92, "learning_rate": 3.4375992851632386e-05, "loss": 0.3809, "step": 5288 }, { "epoch": 0.92, "learning_rate": 3.4369472700844005e-05, "loss": 0.4134, "step": 5289 }, { "epoch": 0.92, "learning_rate": 3.4362951913829805e-05, "loss": 0.3909, "step": 5290 }, { "epoch": 0.92, "learning_rate": 3.4356430491193705e-05, "loss": 0.3972, "step": 5291 }, { "epoch": 0.92, "learning_rate": 3.434990843353973e-05, "loss": 0.3891, "step": 5292 }, { "epoch": 0.92, "learning_rate": 3.434338574147189e-05, "loss": 0.3765, "step": 5293 }, { "epoch": 0.92, "learning_rate": 3.433686241559432e-05, "loss": 0.3993, "step": 5294 }, { "epoch": 0.92, "learning_rate": 3.4330338456511175e-05, "loss": 0.3903, "step": 5295 }, { "epoch": 0.92, "learning_rate": 3.4323813864826684e-05, "loss": 0.3969, "step": 5296 }, { "epoch": 0.92, "learning_rate": 3.4317288641145135e-05, "loss": 0.389, "step": 5297 }, { "epoch": 0.92, "learning_rate": 3.431076278607086e-05, "loss": 0.3927, "step": 5298 }, { "epoch": 0.92, "learning_rate": 3.4304236300208264e-05, "loss": 0.4114, "step": 5299 }, { "epoch": 0.92, "learning_rate": 3.4297709184161815e-05, "loss": 0.4116, "step": 5300 }, { "epoch": 0.92, "learning_rate": 3.429118143853603e-05, "loss": 0.3935, "step": 5301 }, { "epoch": 0.92, "learning_rate": 3.4284653063935475e-05, "loss": 0.3871, "step": 5302 }, { "epoch": 0.92, "learning_rate": 3.4278124060964795e-05, "loss": 0.3827, "step": 5303 }, { "epoch": 0.92, "learning_rate": 3.427159443022868e-05, "loss": 0.4125, "step": 5304 }, { "epoch": 0.92, "learning_rate": 3.426506417233186e-05, "loss": 0.4035, "step": 5305 }, { "epoch": 0.93, "learning_rate": 3.4258533287879185e-05, "loss": 0.3876, "step": 5306 }, { "epoch": 0.93, "learning_rate": 3.425200177747549e-05, "loss": 0.3878, "step": 5307 }, { "epoch": 0.93, "learning_rate": 3.424546964172572e-05, "loss": 0.3862, "step": 5308 }, { "epoch": 0.93, "learning_rate": 3.423893688123484e-05, "loss": 0.4076, "step": 5309 }, { "epoch": 0.93, "learning_rate": 3.423240349660789e-05, "loss": 0.3896, "step": 5310 }, { "epoch": 0.93, "learning_rate": 3.422586948844999e-05, "loss": 0.4038, "step": 5311 }, { "epoch": 0.93, "learning_rate": 3.421933485736628e-05, "loss": 0.3867, "step": 5312 }, { "epoch": 0.93, "learning_rate": 3.421279960396198e-05, "loss": 0.3962, "step": 5313 }, { "epoch": 0.93, "learning_rate": 3.420626372884235e-05, "loss": 0.3985, "step": 5314 }, { "epoch": 0.93, "learning_rate": 3.419972723261274e-05, "loss": 0.3958, "step": 5315 }, { "epoch": 0.93, "learning_rate": 3.419319011587852e-05, "loss": 0.3859, "step": 5316 }, { "epoch": 0.93, "learning_rate": 3.418665237924513e-05, "loss": 0.3857, "step": 5317 }, { "epoch": 0.93, "learning_rate": 3.418011402331809e-05, "loss": 0.4049, "step": 5318 }, { "epoch": 0.93, "learning_rate": 3.417357504870296e-05, "loss": 0.4029, "step": 5319 }, { "epoch": 0.93, "learning_rate": 3.416703545600533e-05, "loss": 0.4039, "step": 5320 }, { "epoch": 0.93, "learning_rate": 3.4160495245830895e-05, "loss": 0.4003, "step": 5321 }, { "epoch": 0.93, "learning_rate": 3.415395441878538e-05, "loss": 0.3897, "step": 5322 }, { "epoch": 0.93, "learning_rate": 3.414741297547458e-05, "loss": 0.3953, "step": 5323 }, { "epoch": 0.93, "learning_rate": 3.414087091650432e-05, "loss": 0.4006, "step": 5324 }, { "epoch": 0.93, "learning_rate": 3.413432824248052e-05, "loss": 0.3864, "step": 5325 }, { "epoch": 0.93, "learning_rate": 3.412778495400915e-05, "loss": 0.3857, "step": 5326 }, { "epoch": 0.93, "learning_rate": 3.412124105169618e-05, "loss": 0.3848, "step": 5327 }, { "epoch": 0.93, "learning_rate": 3.4114696536147735e-05, "loss": 0.3792, "step": 5328 }, { "epoch": 0.93, "learning_rate": 3.410815140796991e-05, "loss": 0.4062, "step": 5329 }, { "epoch": 0.93, "learning_rate": 3.410160566776891e-05, "loss": 0.4001, "step": 5330 }, { "epoch": 0.93, "learning_rate": 3.409505931615097e-05, "loss": 0.3888, "step": 5331 }, { "epoch": 0.93, "learning_rate": 3.4088512353722385e-05, "loss": 0.3813, "step": 5332 }, { "epoch": 0.93, "learning_rate": 3.408196478108952e-05, "loss": 0.4049, "step": 5333 }, { "epoch": 0.93, "learning_rate": 3.4075416598858785e-05, "loss": 0.4133, "step": 5334 }, { "epoch": 0.93, "learning_rate": 3.406886780763665e-05, "loss": 0.4058, "step": 5335 }, { "epoch": 0.93, "learning_rate": 3.406231840802964e-05, "loss": 0.3812, "step": 5336 }, { "epoch": 0.93, "learning_rate": 3.4055768400644334e-05, "loss": 0.3905, "step": 5337 }, { "epoch": 0.93, "learning_rate": 3.404921778608736e-05, "loss": 0.4054, "step": 5338 }, { "epoch": 0.93, "learning_rate": 3.404266656496544e-05, "loss": 0.3906, "step": 5339 }, { "epoch": 0.93, "learning_rate": 3.40361147378853e-05, "loss": 0.3991, "step": 5340 }, { "epoch": 0.93, "learning_rate": 3.4029562305453755e-05, "loss": 0.3913, "step": 5341 }, { "epoch": 0.93, "learning_rate": 3.402300926827767e-05, "loss": 0.3841, "step": 5342 }, { "epoch": 0.93, "learning_rate": 3.4016455626963954e-05, "loss": 0.4032, "step": 5343 }, { "epoch": 0.93, "learning_rate": 3.40099013821196e-05, "loss": 0.4014, "step": 5344 }, { "epoch": 0.93, "learning_rate": 3.400334653435163e-05, "loss": 0.3833, "step": 5345 }, { "epoch": 0.93, "learning_rate": 3.3996791084267115e-05, "loss": 0.3989, "step": 5346 }, { "epoch": 0.93, "learning_rate": 3.399023503247322e-05, "loss": 0.3961, "step": 5347 }, { "epoch": 0.93, "learning_rate": 3.3983678379577125e-05, "loss": 0.399, "step": 5348 }, { "epoch": 0.93, "learning_rate": 3.3977121126186104e-05, "loss": 0.4095, "step": 5349 }, { "epoch": 0.93, "learning_rate": 3.397056327290744e-05, "loss": 0.4182, "step": 5350 }, { "epoch": 0.93, "learning_rate": 3.3964004820348516e-05, "loss": 0.3836, "step": 5351 }, { "epoch": 0.93, "learning_rate": 3.3957445769116756e-05, "loss": 0.4004, "step": 5352 }, { "epoch": 0.93, "learning_rate": 3.3950886119819616e-05, "loss": 0.3842, "step": 5353 }, { "epoch": 0.93, "learning_rate": 3.394432587306464e-05, "loss": 0.4007, "step": 5354 }, { "epoch": 0.93, "learning_rate": 3.3937765029459416e-05, "loss": 0.401, "step": 5355 }, { "epoch": 0.93, "learning_rate": 3.393120358961159e-05, "loss": 0.3819, "step": 5356 }, { "epoch": 0.93, "learning_rate": 3.392464155412884e-05, "loss": 0.3873, "step": 5357 }, { "epoch": 0.93, "learning_rate": 3.391807892361893e-05, "loss": 0.3888, "step": 5358 }, { "epoch": 0.93, "learning_rate": 3.391151569868968e-05, "loss": 0.3905, "step": 5359 }, { "epoch": 0.93, "learning_rate": 3.390495187994892e-05, "loss": 0.3929, "step": 5360 }, { "epoch": 0.93, "learning_rate": 3.389838746800461e-05, "loss": 0.3907, "step": 5361 }, { "epoch": 0.93, "learning_rate": 3.389182246346468e-05, "loss": 0.3992, "step": 5362 }, { "epoch": 0.94, "learning_rate": 3.388525686693719e-05, "loss": 0.3983, "step": 5363 }, { "epoch": 0.94, "learning_rate": 3.387869067903019e-05, "loss": 0.3922, "step": 5364 }, { "epoch": 0.94, "learning_rate": 3.3872123900351835e-05, "loss": 0.4028, "step": 5365 }, { "epoch": 0.94, "learning_rate": 3.386555653151033e-05, "loss": 0.3964, "step": 5366 }, { "epoch": 0.94, "learning_rate": 3.3858988573113906e-05, "loss": 0.3803, "step": 5367 }, { "epoch": 0.94, "learning_rate": 3.385242002577086e-05, "loss": 0.3831, "step": 5368 }, { "epoch": 0.94, "learning_rate": 3.384585089008955e-05, "loss": 0.3904, "step": 5369 }, { "epoch": 0.94, "learning_rate": 3.3839281166678386e-05, "loss": 0.3885, "step": 5370 }, { "epoch": 0.94, "learning_rate": 3.3832710856145835e-05, "loss": 0.3976, "step": 5371 }, { "epoch": 0.94, "learning_rate": 3.382613995910041e-05, "loss": 0.3885, "step": 5372 }, { "epoch": 0.94, "learning_rate": 3.38195684761507e-05, "loss": 0.3928, "step": 5373 }, { "epoch": 0.94, "learning_rate": 3.381299640790531e-05, "loss": 0.4158, "step": 5374 }, { "epoch": 0.94, "learning_rate": 3.380642375497293e-05, "loss": 0.3927, "step": 5375 }, { "epoch": 0.94, "learning_rate": 3.37998505179623e-05, "loss": 0.4035, "step": 5376 }, { "epoch": 0.94, "learning_rate": 3.37932766974822e-05, "loss": 0.3766, "step": 5377 }, { "epoch": 0.94, "learning_rate": 3.378670229414149e-05, "loss": 0.3898, "step": 5378 }, { "epoch": 0.94, "learning_rate": 3.378012730854905e-05, "loss": 0.3844, "step": 5379 }, { "epoch": 0.94, "learning_rate": 3.377355174131385e-05, "loss": 0.3964, "step": 5380 }, { "epoch": 0.94, "learning_rate": 3.376697559304487e-05, "loss": 0.3932, "step": 5381 }, { "epoch": 0.94, "learning_rate": 3.376039886435119e-05, "loss": 0.3821, "step": 5382 }, { "epoch": 0.94, "learning_rate": 3.3753821555841905e-05, "loss": 0.3865, "step": 5383 }, { "epoch": 0.94, "learning_rate": 3.3747243668126204e-05, "loss": 0.4052, "step": 5384 }, { "epoch": 0.94, "learning_rate": 3.3740665201813295e-05, "loss": 0.4034, "step": 5385 }, { "epoch": 0.94, "learning_rate": 3.373408615751245e-05, "loss": 0.389, "step": 5386 }, { "epoch": 0.94, "learning_rate": 3.372750653583301e-05, "loss": 0.3756, "step": 5387 }, { "epoch": 0.94, "learning_rate": 3.372092633738433e-05, "loss": 0.3899, "step": 5388 }, { "epoch": 0.94, "learning_rate": 3.3714345562775866e-05, "loss": 0.3921, "step": 5389 }, { "epoch": 0.94, "learning_rate": 3.370776421261709e-05, "loss": 0.3958, "step": 5390 }, { "epoch": 0.94, "learning_rate": 3.370118228751757e-05, "loss": 0.3886, "step": 5391 }, { "epoch": 0.94, "learning_rate": 3.369459978808687e-05, "loss": 0.3848, "step": 5392 }, { "epoch": 0.94, "learning_rate": 3.368801671493465e-05, "loss": 0.3885, "step": 5393 }, { "epoch": 0.94, "learning_rate": 3.368143306867062e-05, "loss": 0.4026, "step": 5394 }, { "epoch": 0.94, "learning_rate": 3.3674848849904514e-05, "loss": 0.3936, "step": 5395 }, { "epoch": 0.94, "learning_rate": 3.3668264059246145e-05, "loss": 0.3928, "step": 5396 }, { "epoch": 0.94, "learning_rate": 3.366167869730539e-05, "loss": 0.4172, "step": 5397 }, { "epoch": 0.94, "learning_rate": 3.3655092764692136e-05, "loss": 0.3882, "step": 5398 }, { "epoch": 0.94, "learning_rate": 3.364850626201636e-05, "loss": 0.3864, "step": 5399 }, { "epoch": 0.94, "learning_rate": 3.364191918988808e-05, "loss": 0.3877, "step": 5400 }, { "epoch": 0.94, "learning_rate": 3.363533154891737e-05, "loss": 0.4106, "step": 5401 }, { "epoch": 0.94, "learning_rate": 3.362874333971435e-05, "loss": 0.4052, "step": 5402 }, { "epoch": 0.94, "learning_rate": 3.36221545628892e-05, "loss": 0.3828, "step": 5403 }, { "epoch": 0.94, "learning_rate": 3.3615565219052134e-05, "loss": 0.4096, "step": 5404 }, { "epoch": 0.94, "learning_rate": 3.3608975308813455e-05, "loss": 0.3931, "step": 5405 }, { "epoch": 0.94, "learning_rate": 3.3602384832783486e-05, "loss": 0.4166, "step": 5406 }, { "epoch": 0.94, "learning_rate": 3.3595793791572606e-05, "loss": 0.4186, "step": 5407 }, { "epoch": 0.94, "learning_rate": 3.358920218579127e-05, "loss": 0.4026, "step": 5408 }, { "epoch": 0.94, "learning_rate": 3.358261001604996e-05, "loss": 0.3968, "step": 5409 }, { "epoch": 0.94, "learning_rate": 3.357601728295921e-05, "loss": 0.3938, "step": 5410 }, { "epoch": 0.94, "learning_rate": 3.3569423987129636e-05, "loss": 0.4039, "step": 5411 }, { "epoch": 0.94, "learning_rate": 3.356283012917186e-05, "loss": 0.3899, "step": 5412 }, { "epoch": 0.94, "learning_rate": 3.35562357096966e-05, "loss": 0.4094, "step": 5413 }, { "epoch": 0.94, "learning_rate": 3.354964072931461e-05, "loss": 0.3902, "step": 5414 }, { "epoch": 0.94, "learning_rate": 3.354304518863667e-05, "loss": 0.4019, "step": 5415 }, { "epoch": 0.94, "learning_rate": 3.353644908827366e-05, "loss": 0.3935, "step": 5416 }, { "epoch": 0.94, "learning_rate": 3.352985242883648e-05, "loss": 0.3936, "step": 5417 }, { "epoch": 0.94, "learning_rate": 3.3523255210936074e-05, "loss": 0.3979, "step": 5418 }, { "epoch": 0.94, "learning_rate": 3.3516657435183485e-05, "loss": 0.3773, "step": 5419 }, { "epoch": 0.94, "learning_rate": 3.3510059102189745e-05, "loss": 0.3766, "step": 5420 }, { "epoch": 0.95, "learning_rate": 3.3503460212565973e-05, "loss": 0.4133, "step": 5421 }, { "epoch": 0.95, "learning_rate": 3.349686076692335e-05, "loss": 0.4045, "step": 5422 }, { "epoch": 0.95, "learning_rate": 3.349026076587307e-05, "loss": 0.3902, "step": 5423 }, { "epoch": 0.95, "learning_rate": 3.3483660210026424e-05, "loss": 0.3839, "step": 5424 }, { "epoch": 0.95, "learning_rate": 3.347705909999472e-05, "loss": 0.4008, "step": 5425 }, { "epoch": 0.95, "learning_rate": 3.347045743638932e-05, "loss": 0.3784, "step": 5426 }, { "epoch": 0.95, "learning_rate": 3.346385521982167e-05, "loss": 0.4037, "step": 5427 }, { "epoch": 0.95, "learning_rate": 3.3457252450903235e-05, "loss": 0.405, "step": 5428 }, { "epoch": 0.95, "learning_rate": 3.345064913024552e-05, "loss": 0.3646, "step": 5429 }, { "epoch": 0.95, "learning_rate": 3.344404525846013e-05, "loss": 0.3942, "step": 5430 }, { "epoch": 0.95, "learning_rate": 3.343744083615866e-05, "loss": 0.3836, "step": 5431 }, { "epoch": 0.95, "learning_rate": 3.343083586395281e-05, "loss": 0.407, "step": 5432 }, { "epoch": 0.95, "learning_rate": 3.342423034245432e-05, "loss": 0.3972, "step": 5433 }, { "epoch": 0.95, "learning_rate": 3.341762427227494e-05, "loss": 0.4073, "step": 5434 }, { "epoch": 0.95, "learning_rate": 3.341101765402652e-05, "loss": 0.3851, "step": 5435 }, { "epoch": 0.95, "learning_rate": 3.3404410488320934e-05, "loss": 0.3956, "step": 5436 }, { "epoch": 0.95, "learning_rate": 3.339780277577011e-05, "loss": 0.3853, "step": 5437 }, { "epoch": 0.95, "learning_rate": 3.339119451698605e-05, "loss": 0.3767, "step": 5438 }, { "epoch": 0.95, "learning_rate": 3.3384585712580756e-05, "loss": 0.412, "step": 5439 }, { "epoch": 0.95, "learning_rate": 3.337797636316633e-05, "loss": 0.3775, "step": 5440 }, { "epoch": 0.95, "learning_rate": 3.3371366469354906e-05, "loss": 0.3826, "step": 5441 }, { "epoch": 0.95, "learning_rate": 3.336475603175868e-05, "loss": 0.4059, "step": 5442 }, { "epoch": 0.95, "learning_rate": 3.3358145050989864e-05, "loss": 0.3867, "step": 5443 }, { "epoch": 0.95, "learning_rate": 3.335153352766076e-05, "loss": 0.3937, "step": 5444 }, { "epoch": 0.95, "learning_rate": 3.3344921462383696e-05, "loss": 0.3786, "step": 5445 }, { "epoch": 0.95, "learning_rate": 3.333830885577104e-05, "loss": 0.4025, "step": 5446 }, { "epoch": 0.95, "learning_rate": 3.333169570843527e-05, "loss": 0.3827, "step": 5447 }, { "epoch": 0.95, "learning_rate": 3.3325082020988834e-05, "loss": 0.4077, "step": 5448 }, { "epoch": 0.95, "learning_rate": 3.331846779404429e-05, "loss": 0.3839, "step": 5449 }, { "epoch": 0.95, "learning_rate": 3.331185302821422e-05, "loss": 0.4174, "step": 5450 }, { "epoch": 0.95, "learning_rate": 3.330523772411124e-05, "loss": 0.3931, "step": 5451 }, { "epoch": 0.95, "learning_rate": 3.3298621882348066e-05, "loss": 0.3918, "step": 5452 }, { "epoch": 0.95, "learning_rate": 3.329200550353741e-05, "loss": 0.4083, "step": 5453 }, { "epoch": 0.95, "learning_rate": 3.3285388588292075e-05, "loss": 0.4002, "step": 5454 }, { "epoch": 0.95, "learning_rate": 3.327877113722488e-05, "loss": 0.3976, "step": 5455 }, { "epoch": 0.95, "learning_rate": 3.327215315094872e-05, "loss": 0.3916, "step": 5456 }, { "epoch": 0.95, "learning_rate": 3.326553463007652e-05, "loss": 0.4123, "step": 5457 }, { "epoch": 0.95, "learning_rate": 3.325891557522127e-05, "loss": 0.3777, "step": 5458 }, { "epoch": 0.95, "learning_rate": 3.325229598699601e-05, "loss": 0.3812, "step": 5459 }, { "epoch": 0.95, "learning_rate": 3.324567586601381e-05, "loss": 0.4004, "step": 5460 }, { "epoch": 0.95, "learning_rate": 3.32390552128878e-05, "loss": 0.3914, "step": 5461 }, { "epoch": 0.95, "learning_rate": 3.323243402823117e-05, "loss": 0.3771, "step": 5462 }, { "epoch": 0.95, "learning_rate": 3.3225812312657156e-05, "loss": 0.4173, "step": 5463 }, { "epoch": 0.95, "learning_rate": 3.321919006677903e-05, "loss": 0.3857, "step": 5464 }, { "epoch": 0.95, "learning_rate": 3.321256729121011e-05, "loss": 0.4057, "step": 5465 }, { "epoch": 0.95, "learning_rate": 3.320594398656379e-05, "loss": 0.3929, "step": 5466 }, { "epoch": 0.95, "learning_rate": 3.319932015345348e-05, "loss": 0.4065, "step": 5467 }, { "epoch": 0.95, "learning_rate": 3.319269579249267e-05, "loss": 0.4005, "step": 5468 }, { "epoch": 0.95, "learning_rate": 3.3186070904294884e-05, "loss": 0.3902, "step": 5469 }, { "epoch": 0.95, "learning_rate": 3.317944548947369e-05, "loss": 0.3968, "step": 5470 }, { "epoch": 0.95, "learning_rate": 3.317281954864271e-05, "loss": 0.3789, "step": 5471 }, { "epoch": 0.95, "learning_rate": 3.3166193082415616e-05, "loss": 0.3889, "step": 5472 }, { "epoch": 0.95, "learning_rate": 3.3159566091406124e-05, "loss": 0.406, "step": 5473 }, { "epoch": 0.95, "learning_rate": 3.315293857622801e-05, "loss": 0.3821, "step": 5474 }, { "epoch": 0.95, "learning_rate": 3.314631053749508e-05, "loss": 0.3892, "step": 5475 }, { "epoch": 0.95, "learning_rate": 3.3139681975821215e-05, "loss": 0.3992, "step": 5476 }, { "epoch": 0.95, "learning_rate": 3.313305289182031e-05, "loss": 0.3966, "step": 5477 }, { "epoch": 0.96, "learning_rate": 3.312642328610634e-05, "loss": 0.389, "step": 5478 }, { "epoch": 0.96, "learning_rate": 3.311979315929331e-05, "loss": 0.3934, "step": 5479 }, { "epoch": 0.96, "learning_rate": 3.311316251199528e-05, "loss": 0.3831, "step": 5480 }, { "epoch": 0.96, "learning_rate": 3.310653134482635e-05, "loss": 0.3953, "step": 5481 }, { "epoch": 0.96, "learning_rate": 3.309989965840069e-05, "loss": 0.4105, "step": 5482 }, { "epoch": 0.96, "learning_rate": 3.3093267453332494e-05, "loss": 0.4026, "step": 5483 }, { "epoch": 0.96, "learning_rate": 3.3086634730236e-05, "loss": 0.393, "step": 5484 }, { "epoch": 0.96, "learning_rate": 3.3080001489725534e-05, "loss": 0.3864, "step": 5485 }, { "epoch": 0.96, "learning_rate": 3.3073367732415414e-05, "loss": 0.4197, "step": 5486 }, { "epoch": 0.96, "learning_rate": 3.3066733458920055e-05, "loss": 0.3807, "step": 5487 }, { "epoch": 0.96, "learning_rate": 3.3060098669853895e-05, "loss": 0.394, "step": 5488 }, { "epoch": 0.96, "learning_rate": 3.305346336583142e-05, "loss": 0.3837, "step": 5489 }, { "epoch": 0.96, "learning_rate": 3.304682754746717e-05, "loss": 0.4029, "step": 5490 }, { "epoch": 0.96, "learning_rate": 3.3040191215375736e-05, "loss": 0.3862, "step": 5491 }, { "epoch": 0.96, "learning_rate": 3.303355437017175e-05, "loss": 0.3796, "step": 5492 }, { "epoch": 0.96, "learning_rate": 3.3026917012469884e-05, "loss": 0.3929, "step": 5493 }, { "epoch": 0.96, "learning_rate": 3.302027914288487e-05, "loss": 0.386, "step": 5494 }, { "epoch": 0.96, "learning_rate": 3.3013640762031485e-05, "loss": 0.4226, "step": 5495 }, { "epoch": 0.96, "learning_rate": 3.300700187052455e-05, "loss": 0.3973, "step": 5496 }, { "epoch": 0.96, "learning_rate": 3.300036246897895e-05, "loss": 0.4087, "step": 5497 }, { "epoch": 0.96, "learning_rate": 3.2993722558009574e-05, "loss": 0.385, "step": 5498 }, { "epoch": 0.96, "learning_rate": 3.29870821382314e-05, "loss": 0.3826, "step": 5499 }, { "epoch": 0.96, "learning_rate": 3.298044121025946e-05, "loss": 0.4043, "step": 5500 }, { "epoch": 0.96, "learning_rate": 3.297379977470878e-05, "loss": 0.4092, "step": 5501 }, { "epoch": 0.96, "learning_rate": 3.296715783219449e-05, "loss": 0.4001, "step": 5502 }, { "epoch": 0.96, "learning_rate": 3.2960515383331725e-05, "loss": 0.4113, "step": 5503 }, { "epoch": 0.96, "learning_rate": 3.29538724287357e-05, "loss": 0.4023, "step": 5504 }, { "epoch": 0.96, "learning_rate": 3.294722896902165e-05, "loss": 0.3921, "step": 5505 }, { "epoch": 0.96, "learning_rate": 3.294058500480487e-05, "loss": 0.3935, "step": 5506 }, { "epoch": 0.96, "learning_rate": 3.2933940536700705e-05, "loss": 0.402, "step": 5507 }, { "epoch": 0.96, "learning_rate": 3.2927295565324546e-05, "loss": 0.382, "step": 5508 }, { "epoch": 0.96, "learning_rate": 3.292065009129181e-05, "loss": 0.4184, "step": 5509 }, { "epoch": 0.96, "learning_rate": 3.2914004115218e-05, "loss": 0.4114, "step": 5510 }, { "epoch": 0.96, "learning_rate": 3.290735763771861e-05, "loss": 0.3943, "step": 5511 }, { "epoch": 0.96, "learning_rate": 3.290071065940924e-05, "loss": 0.396, "step": 5512 }, { "epoch": 0.96, "learning_rate": 3.2894063180905515e-05, "loss": 0.4072, "step": 5513 }, { "epoch": 0.96, "learning_rate": 3.288741520282308e-05, "loss": 0.4027, "step": 5514 }, { "epoch": 0.96, "learning_rate": 3.2880766725777646e-05, "loss": 0.3947, "step": 5515 }, { "epoch": 0.96, "learning_rate": 3.287411775038498e-05, "loss": 0.3809, "step": 5516 }, { "epoch": 0.96, "learning_rate": 3.286746827726089e-05, "loss": 0.4013, "step": 5517 }, { "epoch": 0.96, "learning_rate": 3.286081830702121e-05, "loss": 0.3961, "step": 5518 }, { "epoch": 0.96, "learning_rate": 3.285416784028186e-05, "loss": 0.3957, "step": 5519 }, { "epoch": 0.96, "learning_rate": 3.284751687765876e-05, "loss": 0.367, "step": 5520 }, { "epoch": 0.96, "learning_rate": 3.2840865419767916e-05, "loss": 0.4005, "step": 5521 }, { "epoch": 0.96, "learning_rate": 3.283421346722535e-05, "loss": 0.3912, "step": 5522 }, { "epoch": 0.96, "learning_rate": 3.2827561020647144e-05, "loss": 0.3922, "step": 5523 }, { "epoch": 0.96, "learning_rate": 3.282090808064943e-05, "loss": 0.3681, "step": 5524 }, { "epoch": 0.96, "learning_rate": 3.281425464784837e-05, "loss": 0.3814, "step": 5525 }, { "epoch": 0.96, "learning_rate": 3.280760072286019e-05, "loss": 0.3944, "step": 5526 }, { "epoch": 0.96, "learning_rate": 3.280094630630114e-05, "loss": 0.3781, "step": 5527 }, { "epoch": 0.96, "learning_rate": 3.2794291398787544e-05, "loss": 0.391, "step": 5528 }, { "epoch": 0.96, "learning_rate": 3.278763600093574e-05, "loss": 0.4065, "step": 5529 }, { "epoch": 0.96, "learning_rate": 3.278098011336214e-05, "loss": 0.3919, "step": 5530 }, { "epoch": 0.96, "learning_rate": 3.277432373668319e-05, "loss": 0.3947, "step": 5531 }, { "epoch": 0.96, "learning_rate": 3.2767666871515366e-05, "loss": 0.3887, "step": 5532 }, { "epoch": 0.96, "learning_rate": 3.276100951847521e-05, "loss": 0.4107, "step": 5533 }, { "epoch": 0.96, "learning_rate": 3.275435167817929e-05, "loss": 0.4174, "step": 5534 }, { "epoch": 0.97, "learning_rate": 3.274769335124426e-05, "loss": 0.4115, "step": 5535 }, { "epoch": 0.97, "learning_rate": 3.274103453828677e-05, "loss": 0.402, "step": 5536 }, { "epoch": 0.97, "learning_rate": 3.2734375239923534e-05, "loss": 0.4002, "step": 5537 }, { "epoch": 0.97, "learning_rate": 3.2727715456771315e-05, "loss": 0.4079, "step": 5538 }, { "epoch": 0.97, "learning_rate": 3.272105518944692e-05, "loss": 0.3938, "step": 5539 }, { "epoch": 0.97, "learning_rate": 3.27143944385672e-05, "loss": 0.4018, "step": 5540 }, { "epoch": 0.97, "learning_rate": 3.2707733204749055e-05, "loss": 0.3959, "step": 5541 }, { "epoch": 0.97, "learning_rate": 3.2701071488609416e-05, "loss": 0.4149, "step": 5542 }, { "epoch": 0.97, "learning_rate": 3.269440929076526e-05, "loss": 0.4086, "step": 5543 }, { "epoch": 0.97, "learning_rate": 3.268774661183364e-05, "loss": 0.4028, "step": 5544 }, { "epoch": 0.97, "learning_rate": 3.26810834524316e-05, "loss": 0.4073, "step": 5545 }, { "epoch": 0.97, "learning_rate": 3.267441981317628e-05, "loss": 0.401, "step": 5546 }, { "epoch": 0.97, "learning_rate": 3.266775569468483e-05, "loss": 0.3962, "step": 5547 }, { "epoch": 0.97, "learning_rate": 3.266109109757446e-05, "loss": 0.3929, "step": 5548 }, { "epoch": 0.97, "learning_rate": 3.265442602246243e-05, "loss": 0.3815, "step": 5549 }, { "epoch": 0.97, "learning_rate": 3.264776046996602e-05, "loss": 0.392, "step": 5550 }, { "epoch": 0.97, "learning_rate": 3.2641094440702574e-05, "loss": 0.3862, "step": 5551 }, { "epoch": 0.97, "learning_rate": 3.2634427935289485e-05, "loss": 0.393, "step": 5552 }, { "epoch": 0.97, "learning_rate": 3.2627760954344165e-05, "loss": 0.4077, "step": 5553 }, { "epoch": 0.97, "learning_rate": 3.262109349848411e-05, "loss": 0.3812, "step": 5554 }, { "epoch": 0.97, "learning_rate": 3.261442556832681e-05, "loss": 0.4048, "step": 5555 }, { "epoch": 0.97, "learning_rate": 3.260775716448983e-05, "loss": 0.4066, "step": 5556 }, { "epoch": 0.97, "learning_rate": 3.2601088287590795e-05, "loss": 0.3721, "step": 5557 }, { "epoch": 0.97, "learning_rate": 3.259441893824733e-05, "loss": 0.3881, "step": 5558 }, { "epoch": 0.97, "learning_rate": 3.2587749117077125e-05, "loss": 0.3868, "step": 5559 }, { "epoch": 0.97, "learning_rate": 3.258107882469793e-05, "loss": 0.3885, "step": 5560 }, { "epoch": 0.97, "learning_rate": 3.2574408061727505e-05, "loss": 0.3937, "step": 5561 }, { "epoch": 0.97, "learning_rate": 3.256773682878369e-05, "loss": 0.4022, "step": 5562 }, { "epoch": 0.97, "learning_rate": 3.256106512648434e-05, "loss": 0.3948, "step": 5563 }, { "epoch": 0.97, "learning_rate": 3.255439295544737e-05, "loss": 0.4073, "step": 5564 }, { "epoch": 0.97, "learning_rate": 3.254772031629073e-05, "loss": 0.3964, "step": 5565 }, { "epoch": 0.97, "learning_rate": 3.2541047209632415e-05, "loss": 0.4104, "step": 5566 }, { "epoch": 0.97, "learning_rate": 3.253437363609046e-05, "loss": 0.398, "step": 5567 }, { "epoch": 0.97, "learning_rate": 3.252769959628296e-05, "loss": 0.3901, "step": 5568 }, { "epoch": 0.97, "learning_rate": 3.2521025090828025e-05, "loss": 0.3817, "step": 5569 }, { "epoch": 0.97, "learning_rate": 3.251435012034384e-05, "loss": 0.4174, "step": 5570 }, { "epoch": 0.97, "learning_rate": 3.2507674685448595e-05, "loss": 0.394, "step": 5571 }, { "epoch": 0.97, "learning_rate": 3.250099878676057e-05, "loss": 0.3753, "step": 5572 }, { "epoch": 0.97, "learning_rate": 3.249432242489804e-05, "loss": 0.3669, "step": 5573 }, { "epoch": 0.97, "learning_rate": 3.248764560047936e-05, "loss": 0.41, "step": 5574 }, { "epoch": 0.97, "learning_rate": 3.24809683141229e-05, "loss": 0.3782, "step": 5575 }, { "epoch": 0.97, "learning_rate": 3.247429056644711e-05, "loss": 0.3996, "step": 5576 }, { "epoch": 0.97, "learning_rate": 3.246761235807044e-05, "loss": 0.3963, "step": 5577 }, { "epoch": 0.97, "learning_rate": 3.24609336896114e-05, "loss": 0.4052, "step": 5578 }, { "epoch": 0.97, "learning_rate": 3.245425456168856e-05, "loss": 0.4007, "step": 5579 }, { "epoch": 0.97, "learning_rate": 3.24475749749205e-05, "loss": 0.3757, "step": 5580 }, { "epoch": 0.97, "learning_rate": 3.244089492992586e-05, "loss": 0.4022, "step": 5581 }, { "epoch": 0.97, "learning_rate": 3.2434214427323345e-05, "loss": 0.3775, "step": 5582 }, { "epoch": 0.97, "learning_rate": 3.242753346773165e-05, "loss": 0.3787, "step": 5583 }, { "epoch": 0.97, "learning_rate": 3.242085205176955e-05, "loss": 0.3866, "step": 5584 }, { "epoch": 0.97, "learning_rate": 3.241417018005587e-05, "loss": 0.3928, "step": 5585 }, { "epoch": 0.97, "learning_rate": 3.240748785320944e-05, "loss": 0.3861, "step": 5586 }, { "epoch": 0.97, "learning_rate": 3.240080507184916e-05, "loss": 0.3793, "step": 5587 }, { "epoch": 0.97, "learning_rate": 3.239412183659397e-05, "loss": 0.3844, "step": 5588 }, { "epoch": 0.97, "learning_rate": 3.2387438148062835e-05, "loss": 0.4009, "step": 5589 }, { "epoch": 0.97, "learning_rate": 3.23807540068748e-05, "loss": 0.367, "step": 5590 }, { "epoch": 0.97, "learning_rate": 3.237406941364889e-05, "loss": 0.3933, "step": 5591 }, { "epoch": 0.97, "learning_rate": 3.236738436900424e-05, "loss": 0.389, "step": 5592 }, { "epoch": 0.98, "learning_rate": 3.236069887355998e-05, "loss": 0.3938, "step": 5593 }, { "epoch": 0.98, "learning_rate": 3.235401292793528e-05, "loss": 0.4161, "step": 5594 }, { "epoch": 0.98, "learning_rate": 3.234732653274941e-05, "loss": 0.3868, "step": 5595 }, { "epoch": 0.98, "learning_rate": 3.23406396886216e-05, "loss": 0.3956, "step": 5596 }, { "epoch": 0.98, "learning_rate": 3.233395239617118e-05, "loss": 0.3785, "step": 5597 }, { "epoch": 0.98, "learning_rate": 3.23272646560175e-05, "loss": 0.4182, "step": 5598 }, { "epoch": 0.98, "learning_rate": 3.2320576468779954e-05, "loss": 0.3849, "step": 5599 }, { "epoch": 0.98, "learning_rate": 3.2313887835077985e-05, "loss": 0.3909, "step": 5600 }, { "epoch": 0.98, "learning_rate": 3.230719875553106e-05, "loss": 0.3899, "step": 5601 }, { "epoch": 0.98, "learning_rate": 3.23005092307587e-05, "loss": 0.4074, "step": 5602 }, { "epoch": 0.98, "learning_rate": 3.229381926138047e-05, "loss": 0.4083, "step": 5603 }, { "epoch": 0.98, "learning_rate": 3.2287128848015954e-05, "loss": 0.4012, "step": 5604 }, { "epoch": 0.98, "learning_rate": 3.2280437991284825e-05, "loss": 0.3988, "step": 5605 }, { "epoch": 0.98, "learning_rate": 3.2273746691806734e-05, "loss": 0.3866, "step": 5606 }, { "epoch": 0.98, "learning_rate": 3.226705495020143e-05, "loss": 0.3907, "step": 5607 }, { "epoch": 0.98, "learning_rate": 3.226036276708865e-05, "loss": 0.3836, "step": 5608 }, { "epoch": 0.98, "learning_rate": 3.225367014308823e-05, "loss": 0.3953, "step": 5609 }, { "epoch": 0.98, "learning_rate": 3.2246977078819995e-05, "loss": 0.3917, "step": 5610 }, { "epoch": 0.98, "learning_rate": 3.224028357490384e-05, "loss": 0.3733, "step": 5611 }, { "epoch": 0.98, "learning_rate": 3.223358963195971e-05, "loss": 0.3979, "step": 5612 }, { "epoch": 0.98, "learning_rate": 3.222689525060753e-05, "loss": 0.4067, "step": 5613 }, { "epoch": 0.98, "learning_rate": 3.222020043146737e-05, "loss": 0.4019, "step": 5614 }, { "epoch": 0.98, "learning_rate": 3.221350517515922e-05, "loss": 0.3915, "step": 5615 }, { "epoch": 0.98, "learning_rate": 3.220680948230321e-05, "loss": 0.398, "step": 5616 }, { "epoch": 0.98, "learning_rate": 3.220011335351946e-05, "loss": 0.4003, "step": 5617 }, { "epoch": 0.98, "learning_rate": 3.219341678942814e-05, "loss": 0.402, "step": 5618 }, { "epoch": 0.98, "learning_rate": 3.218671979064946e-05, "loss": 0.4176, "step": 5619 }, { "epoch": 0.98, "learning_rate": 3.218002235780367e-05, "loss": 0.3768, "step": 5620 }, { "epoch": 0.98, "learning_rate": 3.217332449151108e-05, "loss": 0.4007, "step": 5621 }, { "epoch": 0.98, "learning_rate": 3.216662619239199e-05, "loss": 0.4033, "step": 5622 }, { "epoch": 0.98, "learning_rate": 3.21599274610668e-05, "loss": 0.4054, "step": 5623 }, { "epoch": 0.98, "learning_rate": 3.215322829815592e-05, "loss": 0.3947, "step": 5624 }, { "epoch": 0.98, "learning_rate": 3.214652870427978e-05, "loss": 0.3788, "step": 5625 }, { "epoch": 0.98, "learning_rate": 3.21398286800589e-05, "loss": 0.4105, "step": 5626 }, { "epoch": 0.98, "learning_rate": 3.213312822611379e-05, "loss": 0.3776, "step": 5627 }, { "epoch": 0.98, "learning_rate": 3.212642734306503e-05, "loss": 0.4088, "step": 5628 }, { "epoch": 0.98, "learning_rate": 3.211972603153324e-05, "loss": 0.3868, "step": 5629 }, { "epoch": 0.98, "learning_rate": 3.211302429213906e-05, "loss": 0.3911, "step": 5630 }, { "epoch": 0.98, "learning_rate": 3.210632212550319e-05, "loss": 0.4003, "step": 5631 }, { "epoch": 0.98, "learning_rate": 3.209961953224636e-05, "loss": 0.4071, "step": 5632 }, { "epoch": 0.98, "learning_rate": 3.2092916512989326e-05, "loss": 0.3766, "step": 5633 }, { "epoch": 0.98, "learning_rate": 3.208621306835291e-05, "loss": 0.3856, "step": 5634 }, { "epoch": 0.98, "learning_rate": 3.2079509198957964e-05, "loss": 0.3923, "step": 5635 }, { "epoch": 0.98, "learning_rate": 3.2072804905425366e-05, "loss": 0.404, "step": 5636 }, { "epoch": 0.98, "learning_rate": 3.206610018837605e-05, "loss": 0.3795, "step": 5637 }, { "epoch": 0.98, "learning_rate": 3.205939504843098e-05, "loss": 0.3845, "step": 5638 }, { "epoch": 0.98, "learning_rate": 3.205268948621116e-05, "loss": 0.3999, "step": 5639 }, { "epoch": 0.98, "learning_rate": 3.204598350233764e-05, "loss": 0.3964, "step": 5640 }, { "epoch": 0.98, "learning_rate": 3.203927709743151e-05, "loss": 0.4072, "step": 5641 }, { "epoch": 0.98, "learning_rate": 3.203257027211387e-05, "loss": 0.3958, "step": 5642 }, { "epoch": 0.98, "learning_rate": 3.2025863027005904e-05, "loss": 0.4009, "step": 5643 }, { "epoch": 0.98, "learning_rate": 3.2019155362728805e-05, "loss": 0.3853, "step": 5644 }, { "epoch": 0.98, "learning_rate": 3.201244727990382e-05, "loss": 0.4047, "step": 5645 }, { "epoch": 0.98, "learning_rate": 3.200573877915221e-05, "loss": 0.3769, "step": 5646 }, { "epoch": 0.98, "learning_rate": 3.199902986109531e-05, "loss": 0.4036, "step": 5647 }, { "epoch": 0.98, "learning_rate": 3.199232052635446e-05, "loss": 0.3927, "step": 5648 }, { "epoch": 0.98, "learning_rate": 3.198561077555108e-05, "loss": 0.3674, "step": 5649 }, { "epoch": 0.99, "learning_rate": 3.197890060930658e-05, "loss": 0.395, "step": 5650 }, { "epoch": 0.99, "learning_rate": 3.197219002824243e-05, "loss": 0.4212, "step": 5651 }, { "epoch": 0.99, "learning_rate": 3.1965479032980156e-05, "loss": 0.3874, "step": 5652 }, { "epoch": 0.99, "learning_rate": 3.195876762414129e-05, "loss": 0.3774, "step": 5653 }, { "epoch": 0.99, "learning_rate": 3.195205580234744e-05, "loss": 0.3891, "step": 5654 }, { "epoch": 0.99, "learning_rate": 3.194534356822021e-05, "loss": 0.3972, "step": 5655 }, { "epoch": 0.99, "learning_rate": 3.193863092238126e-05, "loss": 0.382, "step": 5656 }, { "epoch": 0.99, "learning_rate": 3.193191786545231e-05, "loss": 0.3963, "step": 5657 }, { "epoch": 0.99, "learning_rate": 3.192520439805509e-05, "loss": 0.3851, "step": 5658 }, { "epoch": 0.99, "learning_rate": 3.191849052081138e-05, "loss": 0.3888, "step": 5659 }, { "epoch": 0.99, "learning_rate": 3.191177623434299e-05, "loss": 0.4132, "step": 5660 }, { "epoch": 0.99, "learning_rate": 3.1905061539271766e-05, "loss": 0.4072, "step": 5661 }, { "epoch": 0.99, "learning_rate": 3.189834643621963e-05, "loss": 0.3892, "step": 5662 }, { "epoch": 0.99, "learning_rate": 3.1891630925808475e-05, "loss": 0.3795, "step": 5663 }, { "epoch": 0.99, "learning_rate": 3.1884915008660286e-05, "loss": 0.3952, "step": 5664 }, { "epoch": 0.99, "learning_rate": 3.187819868539706e-05, "loss": 0.3771, "step": 5665 }, { "epoch": 0.99, "learning_rate": 3.187148195664084e-05, "loss": 0.3841, "step": 5666 }, { "epoch": 0.99, "learning_rate": 3.186476482301371e-05, "loss": 0.3907, "step": 5667 }, { "epoch": 0.99, "learning_rate": 3.185804728513778e-05, "loss": 0.4061, "step": 5668 }, { "epoch": 0.99, "learning_rate": 3.185132934363521e-05, "loss": 0.3711, "step": 5669 }, { "epoch": 0.99, "learning_rate": 3.1844610999128186e-05, "loss": 0.396, "step": 5670 }, { "epoch": 0.99, "learning_rate": 3.1837892252238945e-05, "loss": 0.4093, "step": 5671 }, { "epoch": 0.99, "learning_rate": 3.1831173103589745e-05, "loss": 0.3899, "step": 5672 }, { "epoch": 0.99, "learning_rate": 3.182445355380289e-05, "loss": 0.3905, "step": 5673 }, { "epoch": 0.99, "learning_rate": 3.181773360350073e-05, "loss": 0.3767, "step": 5674 }, { "epoch": 0.99, "learning_rate": 3.181101325330562e-05, "loss": 0.3788, "step": 5675 }, { "epoch": 0.99, "learning_rate": 3.180429250384001e-05, "loss": 0.4057, "step": 5676 }, { "epoch": 0.99, "learning_rate": 3.179757135572632e-05, "loss": 0.3896, "step": 5677 }, { "epoch": 0.99, "learning_rate": 3.179084980958705e-05, "loss": 0.4256, "step": 5678 }, { "epoch": 0.99, "learning_rate": 3.178412786604473e-05, "loss": 0.3897, "step": 5679 }, { "epoch": 0.99, "learning_rate": 3.1777405525721914e-05, "loss": 0.3949, "step": 5680 }, { "epoch": 0.99, "learning_rate": 3.1770682789241205e-05, "loss": 0.411, "step": 5681 }, { "epoch": 0.99, "learning_rate": 3.176395965722525e-05, "loss": 0.4048, "step": 5682 }, { "epoch": 0.99, "learning_rate": 3.175723613029669e-05, "loss": 0.4117, "step": 5683 }, { "epoch": 0.99, "learning_rate": 3.175051220907827e-05, "loss": 0.4052, "step": 5684 }, { "epoch": 0.99, "learning_rate": 3.174378789419271e-05, "loss": 0.4092, "step": 5685 }, { "epoch": 0.99, "learning_rate": 3.1737063186262806e-05, "loss": 0.3893, "step": 5686 }, { "epoch": 0.99, "learning_rate": 3.173033808591136e-05, "loss": 0.3913, "step": 5687 }, { "epoch": 0.99, "learning_rate": 3.1723612593761255e-05, "loss": 0.3923, "step": 5688 }, { "epoch": 0.99, "learning_rate": 3.171688671043535e-05, "loss": 0.3982, "step": 5689 }, { "epoch": 0.99, "learning_rate": 3.1710160436556584e-05, "loss": 0.3909, "step": 5690 }, { "epoch": 0.99, "learning_rate": 3.170343377274793e-05, "loss": 0.3942, "step": 5691 }, { "epoch": 0.99, "learning_rate": 3.169670671963238e-05, "loss": 0.3959, "step": 5692 }, { "epoch": 0.99, "learning_rate": 3.168997927783297e-05, "loss": 0.3966, "step": 5693 }, { "epoch": 0.99, "learning_rate": 3.168325144797277e-05, "loss": 0.3783, "step": 5694 }, { "epoch": 0.99, "learning_rate": 3.1676523230674885e-05, "loss": 0.3827, "step": 5695 }, { "epoch": 0.99, "learning_rate": 3.166979462656247e-05, "loss": 0.3974, "step": 5696 }, { "epoch": 0.99, "learning_rate": 3.166306563625868e-05, "loss": 0.3809, "step": 5697 }, { "epoch": 0.99, "learning_rate": 3.1656336260386767e-05, "loss": 0.3866, "step": 5698 }, { "epoch": 0.99, "learning_rate": 3.164960649956994e-05, "loss": 0.3894, "step": 5699 }, { "epoch": 0.99, "learning_rate": 3.1642876354431514e-05, "loss": 0.4092, "step": 5700 }, { "epoch": 0.99, "learning_rate": 3.1636145825594804e-05, "loss": 0.3885, "step": 5701 }, { "epoch": 0.99, "learning_rate": 3.162941491368316e-05, "loss": 0.3736, "step": 5702 }, { "epoch": 0.99, "learning_rate": 3.162268361931999e-05, "loss": 0.3976, "step": 5703 }, { "epoch": 0.99, "learning_rate": 3.16159519431287e-05, "loss": 0.3865, "step": 5704 }, { "epoch": 0.99, "learning_rate": 3.160921988573277e-05, "loss": 0.376, "step": 5705 }, { "epoch": 0.99, "learning_rate": 3.16024874477557e-05, "loss": 0.3985, "step": 5706 }, { "epoch": 1.0, "learning_rate": 3.1595754629821025e-05, "loss": 0.4046, "step": 5707 }, { "epoch": 1.0, "learning_rate": 3.158902143255229e-05, "loss": 0.3858, "step": 5708 }, { "epoch": 1.0, "learning_rate": 3.158228785657312e-05, "loss": 0.3945, "step": 5709 }, { "epoch": 1.0, "learning_rate": 3.157555390250717e-05, "loss": 0.3887, "step": 5710 }, { "epoch": 1.0, "learning_rate": 3.156881957097808e-05, "loss": 0.3941, "step": 5711 }, { "epoch": 1.0, "learning_rate": 3.156208486260959e-05, "loss": 0.3883, "step": 5712 }, { "epoch": 1.0, "learning_rate": 3.1555349778025424e-05, "loss": 0.4114, "step": 5713 }, { "epoch": 1.0, "learning_rate": 3.154861431784937e-05, "loss": 0.3927, "step": 5714 }, { "epoch": 1.0, "learning_rate": 3.154187848270523e-05, "loss": 0.3948, "step": 5715 }, { "epoch": 1.0, "learning_rate": 3.1535142273216864e-05, "loss": 0.399, "step": 5716 }, { "epoch": 1.0, "learning_rate": 3.152840569000817e-05, "loss": 0.3832, "step": 5717 }, { "epoch": 1.0, "learning_rate": 3.152166873370304e-05, "loss": 0.3838, "step": 5718 }, { "epoch": 1.0, "learning_rate": 3.1514931404925446e-05, "loss": 0.3939, "step": 5719 }, { "epoch": 1.0, "learning_rate": 3.150819370429936e-05, "loss": 0.4114, "step": 5720 }, { "epoch": 1.0, "learning_rate": 3.1501455632448805e-05, "loss": 0.3969, "step": 5721 }, { "epoch": 1.0, "learning_rate": 3.1494717189997854e-05, "loss": 0.386, "step": 5722 }, { "epoch": 1.0, "learning_rate": 3.1487978377570575e-05, "loss": 0.4103, "step": 5723 }, { "epoch": 1.0, "learning_rate": 3.148123919579111e-05, "loss": 0.4001, "step": 5724 }, { "epoch": 1.0, "learning_rate": 3.1474499645283606e-05, "loss": 0.3808, "step": 5725 }, { "epoch": 1.0, "learning_rate": 3.146775972667226e-05, "loss": 0.385, "step": 5726 }, { "epoch": 1.0, "learning_rate": 3.1461019440581295e-05, "loss": 0.391, "step": 5727 }, { "epoch": 1.0, "learning_rate": 3.145427878763498e-05, "loss": 0.4064, "step": 5728 }, { "epoch": 1.0, "learning_rate": 3.144753776845761e-05, "loss": 0.3849, "step": 5729 }, { "epoch": 1.0, "learning_rate": 3.144079638367351e-05, "loss": 0.3972, "step": 5730 }, { "epoch": 1.0, "learning_rate": 3.143405463390704e-05, "loss": 0.3875, "step": 5731 }, { "epoch": 1.0, "learning_rate": 3.1427312519782604e-05, "loss": 0.399, "step": 5732 }, { "epoch": 1.0, "learning_rate": 3.142057004192462e-05, "loss": 0.4127, "step": 5733 }, { "epoch": 1.0, "learning_rate": 3.141382720095757e-05, "loss": 0.4056, "step": 5734 }, { "epoch": 1.0, "learning_rate": 3.140708399750594e-05, "loss": 0.3813, "step": 5735 }, { "epoch": 1.0, "learning_rate": 3.140034043219426e-05, "loss": 0.3757, "step": 5736 }, { "epoch": 1.0, "learning_rate": 3.139359650564709e-05, "loss": 0.3873, "step": 5737 }, { "epoch": 1.0, "learning_rate": 3.138685221848904e-05, "loss": 0.3539, "step": 5738 }, { "epoch": 1.0, "learning_rate": 3.138010757134474e-05, "loss": 0.3736, "step": 5739 }, { "epoch": 1.0, "learning_rate": 3.1373362564838855e-05, "loss": 0.3781, "step": 5740 }, { "epoch": 1.0, "learning_rate": 3.136661719959608e-05, "loss": 0.3759, "step": 5741 }, { "epoch": 1.0, "learning_rate": 3.135987147624114e-05, "loss": 0.3665, "step": 5742 }, { "epoch": 1.0, "learning_rate": 3.135312539539882e-05, "loss": 0.3763, "step": 5743 }, { "epoch": 1.0, "learning_rate": 3.134637895769389e-05, "loss": 0.3537, "step": 5744 }, { "epoch": 1.0, "learning_rate": 3.13396321637512e-05, "loss": 0.3631, "step": 5745 }, { "epoch": 1.0, "learning_rate": 3.1332885014195607e-05, "loss": 0.387, "step": 5746 }, { "epoch": 1.0, "learning_rate": 3.132613750965201e-05, "loss": 0.3695, "step": 5747 }, { "epoch": 1.0, "learning_rate": 3.131938965074535e-05, "loss": 0.3608, "step": 5748 }, { "epoch": 1.0, "learning_rate": 3.131264143810055e-05, "loss": 0.3564, "step": 5749 }, { "epoch": 1.0, "learning_rate": 3.1305892872342664e-05, "loss": 0.367, "step": 5750 }, { "epoch": 1.0, "learning_rate": 3.129914395409666e-05, "loss": 0.3611, "step": 5751 }, { "epoch": 1.0, "learning_rate": 3.1292394683987645e-05, "loss": 0.3677, "step": 5752 }, { "epoch": 1.0, "learning_rate": 3.128564506264069e-05, "loss": 0.3716, "step": 5753 }, { "epoch": 1.0, "learning_rate": 3.1278895090680924e-05, "loss": 0.3581, "step": 5754 }, { "epoch": 1.0, "learning_rate": 3.12721447687335e-05, "loss": 0.3751, "step": 5755 }, { "epoch": 1.0, "learning_rate": 3.1265394097423626e-05, "loss": 0.3861, "step": 5756 }, { "epoch": 1.0, "learning_rate": 3.1258643077376504e-05, "loss": 0.3596, "step": 5757 }, { "epoch": 1.0, "learning_rate": 3.12518917092174e-05, "loss": 0.3724, "step": 5758 }, { "epoch": 1.0, "learning_rate": 3.1245139993571595e-05, "loss": 0.3607, "step": 5759 }, { "epoch": 1.0, "learning_rate": 3.1238387931064436e-05, "loss": 0.3743, "step": 5760 }, { "epoch": 1.0, "learning_rate": 3.1231635522321224e-05, "loss": 0.3726, "step": 5761 }, { "epoch": 1.0, "learning_rate": 3.1224882767967386e-05, "loss": 0.364, "step": 5762 }, { "epoch": 1.0, "learning_rate": 3.121812966862832e-05, "loss": 0.3533, "step": 5763 }, { "epoch": 1.0, "learning_rate": 3.121137622492946e-05, "loss": 0.3719, "step": 5764 }, { "epoch": 1.01, "learning_rate": 3.120462243749632e-05, "loss": 0.3574, "step": 5765 }, { "epoch": 1.01, "learning_rate": 3.119786830695439e-05, "loss": 0.3595, "step": 5766 }, { "epoch": 1.01, "learning_rate": 3.119111383392922e-05, "loss": 0.3483, "step": 5767 }, { "epoch": 1.01, "learning_rate": 3.118435901904637e-05, "loss": 0.376, "step": 5768 }, { "epoch": 1.01, "learning_rate": 3.117760386293147e-05, "loss": 0.3616, "step": 5769 }, { "epoch": 1.01, "learning_rate": 3.117084836621014e-05, "loss": 0.3789, "step": 5770 }, { "epoch": 1.01, "learning_rate": 3.116409252950805e-05, "loss": 0.3682, "step": 5771 }, { "epoch": 1.01, "learning_rate": 3.1157336353450924e-05, "loss": 0.3613, "step": 5772 }, { "epoch": 1.01, "learning_rate": 3.115057983866447e-05, "loss": 0.3654, "step": 5773 }, { "epoch": 1.01, "learning_rate": 3.1143822985774455e-05, "loss": 0.3794, "step": 5774 }, { "epoch": 1.01, "learning_rate": 3.113706579540668e-05, "loss": 0.3668, "step": 5775 }, { "epoch": 1.01, "learning_rate": 3.113030826818698e-05, "loss": 0.3642, "step": 5776 }, { "epoch": 1.01, "learning_rate": 3.112355040474119e-05, "loss": 0.3723, "step": 5777 }, { "epoch": 1.01, "learning_rate": 3.111679220569522e-05, "loss": 0.3687, "step": 5778 }, { "epoch": 1.01, "learning_rate": 3.111003367167499e-05, "loss": 0.3665, "step": 5779 }, { "epoch": 1.01, "learning_rate": 3.1103274803306436e-05, "loss": 0.3711, "step": 5780 }, { "epoch": 1.01, "learning_rate": 3.1096515601215553e-05, "loss": 0.3743, "step": 5781 }, { "epoch": 1.01, "learning_rate": 3.1089756066028354e-05, "loss": 0.3685, "step": 5782 }, { "epoch": 1.01, "learning_rate": 3.1082996198370864e-05, "loss": 0.3651, "step": 5783 }, { "epoch": 1.01, "learning_rate": 3.107623599886918e-05, "loss": 0.3543, "step": 5784 }, { "epoch": 1.01, "learning_rate": 3.1069475468149394e-05, "loss": 0.3585, "step": 5785 }, { "epoch": 1.01, "learning_rate": 3.1062714606837654e-05, "loss": 0.3981, "step": 5786 }, { "epoch": 1.01, "learning_rate": 3.105595341556011e-05, "loss": 0.3807, "step": 5787 }, { "epoch": 1.01, "learning_rate": 3.104919189494298e-05, "loss": 0.3745, "step": 5788 }, { "epoch": 1.01, "learning_rate": 3.104243004561247e-05, "loss": 0.3801, "step": 5789 }, { "epoch": 1.01, "learning_rate": 3.1035667868194854e-05, "loss": 0.3749, "step": 5790 }, { "epoch": 1.01, "learning_rate": 3.102890536331641e-05, "loss": 0.3823, "step": 5791 }, { "epoch": 1.01, "learning_rate": 3.1022142531603465e-05, "loss": 0.3685, "step": 5792 }, { "epoch": 1.01, "learning_rate": 3.101537937368237e-05, "loss": 0.3757, "step": 5793 }, { "epoch": 1.01, "learning_rate": 3.1008615890179485e-05, "loss": 0.3886, "step": 5794 }, { "epoch": 1.01, "learning_rate": 3.1001852081721245e-05, "loss": 0.3575, "step": 5795 }, { "epoch": 1.01, "learning_rate": 3.0995087948934076e-05, "loss": 0.3726, "step": 5796 }, { "epoch": 1.01, "learning_rate": 3.0988323492444456e-05, "loss": 0.3585, "step": 5797 }, { "epoch": 1.01, "learning_rate": 3.098155871287887e-05, "loss": 0.3847, "step": 5798 }, { "epoch": 1.01, "learning_rate": 3.097479361086386e-05, "loss": 0.346, "step": 5799 }, { "epoch": 1.01, "learning_rate": 3.096802818702598e-05, "loss": 0.3831, "step": 5800 }, { "epoch": 1.01, "learning_rate": 3.096126244199182e-05, "loss": 0.3727, "step": 5801 }, { "epoch": 1.01, "learning_rate": 3.095449637638799e-05, "loss": 0.3635, "step": 5802 }, { "epoch": 1.01, "learning_rate": 3.0947729990841165e-05, "loss": 0.3706, "step": 5803 }, { "epoch": 1.01, "learning_rate": 3.0940963285978e-05, "loss": 0.3678, "step": 5804 }, { "epoch": 1.01, "learning_rate": 3.0934196262425215e-05, "loss": 0.3775, "step": 5805 }, { "epoch": 1.01, "learning_rate": 3.092742892080953e-05, "loss": 0.364, "step": 5806 }, { "epoch": 1.01, "learning_rate": 3.092066126175773e-05, "loss": 0.3648, "step": 5807 }, { "epoch": 1.01, "learning_rate": 3.09138932858966e-05, "loss": 0.3678, "step": 5808 }, { "epoch": 1.01, "learning_rate": 3.0907124993852976e-05, "loss": 0.3639, "step": 5809 }, { "epoch": 1.01, "learning_rate": 3.0900356386253705e-05, "loss": 0.3633, "step": 5810 }, { "epoch": 1.01, "learning_rate": 3.089358746372567e-05, "loss": 0.3807, "step": 5811 }, { "epoch": 1.01, "learning_rate": 3.08868182268958e-05, "loss": 0.3622, "step": 5812 }, { "epoch": 1.01, "learning_rate": 3.0880048676391005e-05, "loss": 0.3785, "step": 5813 }, { "epoch": 1.01, "learning_rate": 3.087327881283828e-05, "loss": 0.3583, "step": 5814 }, { "epoch": 1.01, "learning_rate": 3.086650863686464e-05, "loss": 0.3852, "step": 5815 }, { "epoch": 1.01, "learning_rate": 3.085973814909707e-05, "loss": 0.3819, "step": 5816 }, { "epoch": 1.01, "learning_rate": 3.0852967350162676e-05, "loss": 0.3599, "step": 5817 }, { "epoch": 1.01, "learning_rate": 3.084619624068851e-05, "loss": 0.3643, "step": 5818 }, { "epoch": 1.01, "learning_rate": 3.0839424821301706e-05, "loss": 0.3583, "step": 5819 }, { "epoch": 1.01, "learning_rate": 3.08326530926294e-05, "loss": 0.3577, "step": 5820 }, { "epoch": 1.01, "learning_rate": 3.082588105529876e-05, "loss": 0.3757, "step": 5821 }, { "epoch": 1.02, "learning_rate": 3.081910870993702e-05, "loss": 0.3679, "step": 5822 }, { "epoch": 1.02, "learning_rate": 3.0812336057171364e-05, "loss": 0.3362, "step": 5823 }, { "epoch": 1.02, "learning_rate": 3.080556309762908e-05, "loss": 0.3527, "step": 5824 }, { "epoch": 1.02, "learning_rate": 3.079878983193745e-05, "loss": 0.3706, "step": 5825 }, { "epoch": 1.02, "learning_rate": 3.0792016260723785e-05, "loss": 0.3692, "step": 5826 }, { "epoch": 1.02, "learning_rate": 3.078524238461543e-05, "loss": 0.3777, "step": 5827 }, { "epoch": 1.02, "learning_rate": 3.077846820423976e-05, "loss": 0.3675, "step": 5828 }, { "epoch": 1.02, "learning_rate": 3.0771693720224176e-05, "loss": 0.3671, "step": 5829 }, { "epoch": 1.02, "learning_rate": 3.0764918933196097e-05, "loss": 0.3748, "step": 5830 }, { "epoch": 1.02, "learning_rate": 3.075814384378299e-05, "loss": 0.369, "step": 5831 }, { "epoch": 1.02, "learning_rate": 3.075136845261234e-05, "loss": 0.3618, "step": 5832 }, { "epoch": 1.02, "learning_rate": 3.074459276031165e-05, "loss": 0.384, "step": 5833 }, { "epoch": 1.02, "learning_rate": 3.0737816767508466e-05, "loss": 0.361, "step": 5834 }, { "epoch": 1.02, "learning_rate": 3.073104047483035e-05, "loss": 0.3689, "step": 5835 }, { "epoch": 1.02, "learning_rate": 3.072426388290491e-05, "loss": 0.3731, "step": 5836 }, { "epoch": 1.02, "learning_rate": 3.071748699235976e-05, "loss": 0.3554, "step": 5837 }, { "epoch": 1.02, "learning_rate": 3.071070980382256e-05, "loss": 0.3677, "step": 5838 }, { "epoch": 1.02, "learning_rate": 3.070393231792097e-05, "loss": 0.3728, "step": 5839 }, { "epoch": 1.02, "learning_rate": 3.069715453528271e-05, "loss": 0.3642, "step": 5840 }, { "epoch": 1.02, "learning_rate": 3.069037645653552e-05, "loss": 0.3717, "step": 5841 }, { "epoch": 1.02, "learning_rate": 3.0683598082307154e-05, "loss": 0.3859, "step": 5842 }, { "epoch": 1.02, "learning_rate": 3.06768194132254e-05, "loss": 0.3738, "step": 5843 }, { "epoch": 1.02, "learning_rate": 3.067004044991806e-05, "loss": 0.3738, "step": 5844 }, { "epoch": 1.02, "learning_rate": 3.066326119301301e-05, "loss": 0.3773, "step": 5845 }, { "epoch": 1.02, "learning_rate": 3.06564816431381e-05, "loss": 0.3604, "step": 5846 }, { "epoch": 1.02, "learning_rate": 3.064970180092122e-05, "loss": 0.3588, "step": 5847 }, { "epoch": 1.02, "learning_rate": 3.064292166699031e-05, "loss": 0.3568, "step": 5848 }, { "epoch": 1.02, "learning_rate": 3.0636141241973307e-05, "loss": 0.3736, "step": 5849 }, { "epoch": 1.02, "learning_rate": 3.062936052649821e-05, "loss": 0.4059, "step": 5850 }, { "epoch": 1.02, "learning_rate": 3.062257952119301e-05, "loss": 0.3795, "step": 5851 }, { "epoch": 1.02, "learning_rate": 3.061579822668573e-05, "loss": 0.358, "step": 5852 }, { "epoch": 1.02, "learning_rate": 3.060901664360446e-05, "loss": 0.3794, "step": 5853 }, { "epoch": 1.02, "learning_rate": 3.060223477257725e-05, "loss": 0.3622, "step": 5854 }, { "epoch": 1.02, "learning_rate": 3.059545261423224e-05, "loss": 0.358, "step": 5855 }, { "epoch": 1.02, "learning_rate": 3.0588670169197554e-05, "loss": 0.3685, "step": 5856 }, { "epoch": 1.02, "learning_rate": 3.058188743810136e-05, "loss": 0.3694, "step": 5857 }, { "epoch": 1.02, "learning_rate": 3.057510442157185e-05, "loss": 0.3782, "step": 5858 }, { "epoch": 1.02, "learning_rate": 3.0568321120237256e-05, "loss": 0.3644, "step": 5859 }, { "epoch": 1.02, "learning_rate": 3.05615375347258e-05, "loss": 0.3781, "step": 5860 }, { "epoch": 1.02, "learning_rate": 3.0554753665665764e-05, "loss": 0.3697, "step": 5861 }, { "epoch": 1.02, "learning_rate": 3.0547969513685457e-05, "loss": 0.3796, "step": 5862 }, { "epoch": 1.02, "learning_rate": 3.0541185079413185e-05, "loss": 0.3702, "step": 5863 }, { "epoch": 1.02, "learning_rate": 3.05344003634773e-05, "loss": 0.3753, "step": 5864 }, { "epoch": 1.02, "learning_rate": 3.052761536650619e-05, "loss": 0.3773, "step": 5865 }, { "epoch": 1.02, "learning_rate": 3.0520830089128245e-05, "loss": 0.371, "step": 5866 }, { "epoch": 1.02, "learning_rate": 3.0514044531971904e-05, "loss": 0.3679, "step": 5867 }, { "epoch": 1.02, "learning_rate": 3.050725869566561e-05, "loss": 0.3715, "step": 5868 }, { "epoch": 1.02, "learning_rate": 3.050047258083784e-05, "loss": 0.3754, "step": 5869 }, { "epoch": 1.02, "learning_rate": 3.049368618811712e-05, "loss": 0.3717, "step": 5870 }, { "epoch": 1.02, "learning_rate": 3.0486899518131963e-05, "loss": 0.3968, "step": 5871 }, { "epoch": 1.02, "learning_rate": 3.0480112571510933e-05, "loss": 0.3555, "step": 5872 }, { "epoch": 1.02, "learning_rate": 3.047332534888261e-05, "loss": 0.3533, "step": 5873 }, { "epoch": 1.02, "learning_rate": 3.04665378508756e-05, "loss": 0.3904, "step": 5874 }, { "epoch": 1.02, "learning_rate": 3.0459750078118538e-05, "loss": 0.3862, "step": 5875 }, { "epoch": 1.02, "learning_rate": 3.0452962031240084e-05, "loss": 0.3543, "step": 5876 }, { "epoch": 1.02, "learning_rate": 3.0446173710868924e-05, "loss": 0.3703, "step": 5877 }, { "epoch": 1.02, "learning_rate": 3.043938511763377e-05, "loss": 0.3769, "step": 5878 }, { "epoch": 1.02, "learning_rate": 3.0432596252163355e-05, "loss": 0.356, "step": 5879 }, { "epoch": 1.03, "learning_rate": 3.042580711508643e-05, "loss": 0.3693, "step": 5880 }, { "epoch": 1.03, "learning_rate": 3.041901770703179e-05, "loss": 0.3711, "step": 5881 }, { "epoch": 1.03, "learning_rate": 3.0412228028628246e-05, "loss": 0.3885, "step": 5882 }, { "epoch": 1.03, "learning_rate": 3.0405438080504633e-05, "loss": 0.3592, "step": 5883 }, { "epoch": 1.03, "learning_rate": 3.0398647863289813e-05, "loss": 0.3685, "step": 5884 }, { "epoch": 1.03, "learning_rate": 3.0391857377612663e-05, "loss": 0.3756, "step": 5885 }, { "epoch": 1.03, "learning_rate": 3.0385066624102104e-05, "loss": 0.3637, "step": 5886 }, { "epoch": 1.03, "learning_rate": 3.0378275603387056e-05, "loss": 0.3893, "step": 5887 }, { "epoch": 1.03, "learning_rate": 3.0371484316096492e-05, "loss": 0.3517, "step": 5888 }, { "epoch": 1.03, "learning_rate": 3.03646927628594e-05, "loss": 0.3641, "step": 5889 }, { "epoch": 1.03, "learning_rate": 3.0357900944304774e-05, "loss": 0.3823, "step": 5890 }, { "epoch": 1.03, "learning_rate": 3.0351108861061663e-05, "loss": 0.3741, "step": 5891 }, { "epoch": 1.03, "learning_rate": 3.0344316513759114e-05, "loss": 0.3685, "step": 5892 }, { "epoch": 1.03, "learning_rate": 3.0337523903026214e-05, "loss": 0.3809, "step": 5893 }, { "epoch": 1.03, "learning_rate": 3.033073102949207e-05, "loss": 0.3813, "step": 5894 }, { "epoch": 1.03, "learning_rate": 3.032393789378582e-05, "loss": 0.3778, "step": 5895 }, { "epoch": 1.03, "learning_rate": 3.0317144496536608e-05, "loss": 0.3632, "step": 5896 }, { "epoch": 1.03, "learning_rate": 3.0310350838373618e-05, "loss": 0.375, "step": 5897 }, { "epoch": 1.03, "learning_rate": 3.0303556919926057e-05, "loss": 0.3624, "step": 5898 }, { "epoch": 1.03, "learning_rate": 3.0296762741823158e-05, "loss": 0.3731, "step": 5899 }, { "epoch": 1.03, "learning_rate": 3.0289968304694165e-05, "loss": 0.381, "step": 5900 }, { "epoch": 1.03, "learning_rate": 3.0283173609168353e-05, "loss": 0.3585, "step": 5901 }, { "epoch": 1.03, "learning_rate": 3.0276378655875026e-05, "loss": 0.349, "step": 5902 }, { "epoch": 1.03, "learning_rate": 3.0269583445443516e-05, "loss": 0.3799, "step": 5903 }, { "epoch": 1.03, "learning_rate": 3.0262787978503156e-05, "loss": 0.372, "step": 5904 }, { "epoch": 1.03, "learning_rate": 3.0255992255683324e-05, "loss": 0.3796, "step": 5905 }, { "epoch": 1.03, "learning_rate": 3.0249196277613416e-05, "loss": 0.3773, "step": 5906 }, { "epoch": 1.03, "learning_rate": 3.0242400044922847e-05, "loss": 0.3788, "step": 5907 }, { "epoch": 1.03, "learning_rate": 3.0235603558241073e-05, "loss": 0.377, "step": 5908 }, { "epoch": 1.03, "learning_rate": 3.022880681819754e-05, "loss": 0.3734, "step": 5909 }, { "epoch": 1.03, "learning_rate": 3.022200982542176e-05, "loss": 0.3709, "step": 5910 }, { "epoch": 1.03, "learning_rate": 3.0215212580543227e-05, "loss": 0.3797, "step": 5911 }, { "epoch": 1.03, "learning_rate": 3.020841508419148e-05, "loss": 0.3541, "step": 5912 }, { "epoch": 1.03, "learning_rate": 3.0201617336996096e-05, "loss": 0.381, "step": 5913 }, { "epoch": 1.03, "learning_rate": 3.0194819339586638e-05, "loss": 0.3698, "step": 5914 }, { "epoch": 1.03, "learning_rate": 3.0188021092592716e-05, "loss": 0.3705, "step": 5915 }, { "epoch": 1.03, "learning_rate": 3.0181222596643965e-05, "loss": 0.3695, "step": 5916 }, { "epoch": 1.03, "learning_rate": 3.0174423852370037e-05, "loss": 0.3557, "step": 5917 }, { "epoch": 1.03, "learning_rate": 3.0167624860400606e-05, "loss": 0.3751, "step": 5918 }, { "epoch": 1.03, "learning_rate": 3.016082562136536e-05, "loss": 0.3768, "step": 5919 }, { "epoch": 1.03, "learning_rate": 3.0154026135894043e-05, "loss": 0.3773, "step": 5920 }, { "epoch": 1.03, "learning_rate": 3.0147226404616363e-05, "loss": 0.3605, "step": 5921 }, { "epoch": 1.03, "learning_rate": 3.0140426428162132e-05, "loss": 0.3731, "step": 5922 }, { "epoch": 1.03, "learning_rate": 3.0133626207161104e-05, "loss": 0.3674, "step": 5923 }, { "epoch": 1.03, "learning_rate": 3.0126825742243102e-05, "loss": 0.3686, "step": 5924 }, { "epoch": 1.03, "learning_rate": 3.0120025034037963e-05, "loss": 0.3668, "step": 5925 }, { "epoch": 1.03, "learning_rate": 3.0113224083175533e-05, "loss": 0.38, "step": 5926 }, { "epoch": 1.03, "learning_rate": 3.010642289028572e-05, "loss": 0.3622, "step": 5927 }, { "epoch": 1.03, "learning_rate": 3.0099621455998395e-05, "loss": 0.3693, "step": 5928 }, { "epoch": 1.03, "learning_rate": 3.0092819780943492e-05, "loss": 0.3704, "step": 5929 }, { "epoch": 1.03, "learning_rate": 3.0086017865750965e-05, "loss": 0.3834, "step": 5930 }, { "epoch": 1.03, "learning_rate": 3.0079215711050772e-05, "loss": 0.3765, "step": 5931 }, { "epoch": 1.03, "learning_rate": 3.0072413317472915e-05, "loss": 0.3567, "step": 5932 }, { "epoch": 1.03, "learning_rate": 3.00656106856474e-05, "loss": 0.38, "step": 5933 }, { "epoch": 1.03, "learning_rate": 3.0058807816204272e-05, "loss": 0.3654, "step": 5934 }, { "epoch": 1.03, "learning_rate": 3.0052004709773563e-05, "loss": 0.3585, "step": 5935 }, { "epoch": 1.03, "learning_rate": 3.004520136698538e-05, "loss": 0.3771, "step": 5936 }, { "epoch": 1.04, "learning_rate": 3.0038397788469813e-05, "loss": 0.3772, "step": 5937 }, { "epoch": 1.04, "learning_rate": 3.003159397485699e-05, "loss": 0.369, "step": 5938 }, { "epoch": 1.04, "learning_rate": 3.0024789926777048e-05, "loss": 0.3785, "step": 5939 }, { "epoch": 1.04, "learning_rate": 3.0017985644860154e-05, "loss": 0.3585, "step": 5940 }, { "epoch": 1.04, "learning_rate": 3.001118112973651e-05, "loss": 0.3645, "step": 5941 }, { "epoch": 1.04, "learning_rate": 3.0004376382036302e-05, "loss": 0.3722, "step": 5942 }, { "epoch": 1.04, "learning_rate": 2.9997571402389774e-05, "loss": 0.3748, "step": 5943 }, { "epoch": 1.04, "learning_rate": 2.9990766191427187e-05, "loss": 0.3598, "step": 5944 }, { "epoch": 1.04, "learning_rate": 2.9983960749778806e-05, "loss": 0.3686, "step": 5945 }, { "epoch": 1.04, "learning_rate": 2.997715507807492e-05, "loss": 0.3561, "step": 5946 }, { "epoch": 1.04, "learning_rate": 2.9970349176945854e-05, "loss": 0.3756, "step": 5947 }, { "epoch": 1.04, "learning_rate": 2.9963543047021953e-05, "loss": 0.3813, "step": 5948 }, { "epoch": 1.04, "learning_rate": 2.9956736688933562e-05, "loss": 0.3755, "step": 5949 }, { "epoch": 1.04, "learning_rate": 2.9949930103311065e-05, "loss": 0.3672, "step": 5950 }, { "epoch": 1.04, "learning_rate": 2.994312329078487e-05, "loss": 0.3631, "step": 5951 }, { "epoch": 1.04, "learning_rate": 2.993631625198539e-05, "loss": 0.37, "step": 5952 }, { "epoch": 1.04, "learning_rate": 2.9929508987543086e-05, "loss": 0.3919, "step": 5953 }, { "epoch": 1.04, "learning_rate": 2.99227014980884e-05, "loss": 0.3551, "step": 5954 }, { "epoch": 1.04, "learning_rate": 2.991589378425183e-05, "loss": 0.3862, "step": 5955 }, { "epoch": 1.04, "learning_rate": 2.9909085846663885e-05, "loss": 0.3714, "step": 5956 }, { "epoch": 1.04, "learning_rate": 2.9902277685955077e-05, "loss": 0.3786, "step": 5957 }, { "epoch": 1.04, "learning_rate": 2.989546930275597e-05, "loss": 0.382, "step": 5958 }, { "epoch": 1.04, "learning_rate": 2.9888660697697113e-05, "loss": 0.3769, "step": 5959 }, { "epoch": 1.04, "learning_rate": 2.9881851871409117e-05, "loss": 0.3736, "step": 5960 }, { "epoch": 1.04, "learning_rate": 2.9875042824522575e-05, "loss": 0.3648, "step": 5961 }, { "epoch": 1.04, "learning_rate": 2.986823355766812e-05, "loss": 0.355, "step": 5962 }, { "epoch": 1.04, "learning_rate": 2.986142407147641e-05, "loss": 0.3592, "step": 5963 }, { "epoch": 1.04, "learning_rate": 2.9854614366578104e-05, "loss": 0.3743, "step": 5964 }, { "epoch": 1.04, "learning_rate": 2.9847804443603887e-05, "loss": 0.372, "step": 5965 }, { "epoch": 1.04, "learning_rate": 2.9840994303184494e-05, "loss": 0.3741, "step": 5966 }, { "epoch": 1.04, "learning_rate": 2.9834183945950634e-05, "loss": 0.3701, "step": 5967 }, { "epoch": 1.04, "learning_rate": 2.9827373372533064e-05, "loss": 0.3828, "step": 5968 }, { "epoch": 1.04, "learning_rate": 2.9820562583562556e-05, "loss": 0.3637, "step": 5969 }, { "epoch": 1.04, "learning_rate": 2.9813751579669896e-05, "loss": 0.3753, "step": 5970 }, { "epoch": 1.04, "learning_rate": 2.9806940361485903e-05, "loss": 0.3818, "step": 5971 }, { "epoch": 1.04, "learning_rate": 2.9800128929641404e-05, "loss": 0.39, "step": 5972 }, { "epoch": 1.04, "learning_rate": 2.9793317284767237e-05, "loss": 0.3591, "step": 5973 }, { "epoch": 1.04, "learning_rate": 2.9786505427494293e-05, "loss": 0.3831, "step": 5974 }, { "epoch": 1.04, "learning_rate": 2.977969335845346e-05, "loss": 0.3841, "step": 5975 }, { "epoch": 1.04, "learning_rate": 2.9772881078275615e-05, "loss": 0.3684, "step": 5976 }, { "epoch": 1.04, "learning_rate": 2.9766068587591735e-05, "loss": 0.3661, "step": 5977 }, { "epoch": 1.04, "learning_rate": 2.9759255887032738e-05, "loss": 0.3831, "step": 5978 }, { "epoch": 1.04, "learning_rate": 2.975244297722959e-05, "loss": 0.3982, "step": 5979 }, { "epoch": 1.04, "learning_rate": 2.974562985881329e-05, "loss": 0.3724, "step": 5980 }, { "epoch": 1.04, "learning_rate": 2.9738816532414842e-05, "loss": 0.3768, "step": 5981 }, { "epoch": 1.04, "learning_rate": 2.973200299866527e-05, "loss": 0.3696, "step": 5982 }, { "epoch": 1.04, "learning_rate": 2.9725189258195623e-05, "loss": 0.3786, "step": 5983 }, { "epoch": 1.04, "learning_rate": 2.971837531163696e-05, "loss": 0.3748, "step": 5984 }, { "epoch": 1.04, "learning_rate": 2.9711561159620366e-05, "loss": 0.398, "step": 5985 }, { "epoch": 1.04, "learning_rate": 2.970474680277694e-05, "loss": 0.3726, "step": 5986 }, { "epoch": 1.04, "learning_rate": 2.969793224173781e-05, "loss": 0.3747, "step": 5987 }, { "epoch": 1.04, "learning_rate": 2.9691117477134107e-05, "loss": 0.3543, "step": 5988 }, { "epoch": 1.04, "learning_rate": 2.9684302509597005e-05, "loss": 0.3832, "step": 5989 }, { "epoch": 1.04, "learning_rate": 2.9677487339757664e-05, "loss": 0.369, "step": 5990 }, { "epoch": 1.04, "learning_rate": 2.9670671968247293e-05, "loss": 0.3889, "step": 5991 }, { "epoch": 1.04, "learning_rate": 2.9663856395697105e-05, "loss": 0.3779, "step": 5992 }, { "epoch": 1.04, "learning_rate": 2.9657040622738318e-05, "loss": 0.3684, "step": 5993 }, { "epoch": 1.05, "learning_rate": 2.9650224650002213e-05, "loss": 0.3526, "step": 5994 }, { "epoch": 1.05, "learning_rate": 2.9643408478120032e-05, "loss": 0.3627, "step": 5995 }, { "epoch": 1.05, "learning_rate": 2.9636592107723094e-05, "loss": 0.3881, "step": 5996 }, { "epoch": 1.05, "learning_rate": 2.9629775539442684e-05, "loss": 0.3682, "step": 5997 }, { "epoch": 1.05, "learning_rate": 2.962295877391013e-05, "loss": 0.3626, "step": 5998 }, { "epoch": 1.05, "learning_rate": 2.961614181175678e-05, "loss": 0.3834, "step": 5999 }, { "epoch": 1.05, "learning_rate": 2.9609324653614e-05, "loss": 0.3792, "step": 6000 }, { "epoch": 1.05, "learning_rate": 2.9602507300113168e-05, "loss": 0.3759, "step": 6001 }, { "epoch": 1.05, "learning_rate": 2.9595689751885687e-05, "loss": 0.3731, "step": 6002 }, { "epoch": 1.05, "learning_rate": 2.9588872009562962e-05, "loss": 0.3578, "step": 6003 }, { "epoch": 1.05, "learning_rate": 2.9582054073776444e-05, "loss": 0.3662, "step": 6004 }, { "epoch": 1.05, "learning_rate": 2.9575235945157574e-05, "loss": 0.37, "step": 6005 }, { "epoch": 1.05, "learning_rate": 2.9568417624337824e-05, "loss": 0.378, "step": 6006 }, { "epoch": 1.05, "learning_rate": 2.9561599111948678e-05, "loss": 0.3661, "step": 6007 }, { "epoch": 1.05, "learning_rate": 2.9554780408621656e-05, "loss": 0.3732, "step": 6008 }, { "epoch": 1.05, "learning_rate": 2.9547961514988272e-05, "loss": 0.3634, "step": 6009 }, { "epoch": 1.05, "learning_rate": 2.9541142431680068e-05, "loss": 0.3725, "step": 6010 }, { "epoch": 1.05, "learning_rate": 2.9534323159328604e-05, "loss": 0.3663, "step": 6011 }, { "epoch": 1.05, "learning_rate": 2.9527503698565456e-05, "loss": 0.373, "step": 6012 }, { "epoch": 1.05, "learning_rate": 2.9520684050022225e-05, "loss": 0.3881, "step": 6013 }, { "epoch": 1.05, "learning_rate": 2.9513864214330507e-05, "loss": 0.364, "step": 6014 }, { "epoch": 1.05, "learning_rate": 2.9507044192121934e-05, "loss": 0.3812, "step": 6015 }, { "epoch": 1.05, "learning_rate": 2.9500223984028162e-05, "loss": 0.3706, "step": 6016 }, { "epoch": 1.05, "learning_rate": 2.9493403590680858e-05, "loss": 0.3697, "step": 6017 }, { "epoch": 1.05, "learning_rate": 2.948658301271168e-05, "loss": 0.3613, "step": 6018 }, { "epoch": 1.05, "learning_rate": 2.9479762250752345e-05, "loss": 0.3696, "step": 6019 }, { "epoch": 1.05, "learning_rate": 2.947294130543456e-05, "loss": 0.3634, "step": 6020 }, { "epoch": 1.05, "learning_rate": 2.9466120177390062e-05, "loss": 0.3631, "step": 6021 }, { "epoch": 1.05, "learning_rate": 2.945929886725059e-05, "loss": 0.3625, "step": 6022 }, { "epoch": 1.05, "learning_rate": 2.9452477375647918e-05, "loss": 0.3945, "step": 6023 }, { "epoch": 1.05, "learning_rate": 2.944565570321382e-05, "loss": 0.383, "step": 6024 }, { "epoch": 1.05, "learning_rate": 2.9438833850580105e-05, "loss": 0.3992, "step": 6025 }, { "epoch": 1.05, "learning_rate": 2.943201181837857e-05, "loss": 0.3543, "step": 6026 }, { "epoch": 1.05, "learning_rate": 2.942518960724107e-05, "loss": 0.3663, "step": 6027 }, { "epoch": 1.05, "learning_rate": 2.9418367217799438e-05, "loss": 0.3712, "step": 6028 }, { "epoch": 1.05, "learning_rate": 2.941154465068554e-05, "loss": 0.3654, "step": 6029 }, { "epoch": 1.05, "learning_rate": 2.940472190653127e-05, "loss": 0.3585, "step": 6030 }, { "epoch": 1.05, "learning_rate": 2.939789898596851e-05, "loss": 0.3807, "step": 6031 }, { "epoch": 1.05, "learning_rate": 2.9391075889629193e-05, "loss": 0.3748, "step": 6032 }, { "epoch": 1.05, "learning_rate": 2.9384252618145225e-05, "loss": 0.3718, "step": 6033 }, { "epoch": 1.05, "learning_rate": 2.9377429172148574e-05, "loss": 0.3591, "step": 6034 }, { "epoch": 1.05, "learning_rate": 2.9370605552271197e-05, "loss": 0.36, "step": 6035 }, { "epoch": 1.05, "learning_rate": 2.9363781759145066e-05, "loss": 0.371, "step": 6036 }, { "epoch": 1.05, "learning_rate": 2.9356957793402178e-05, "loss": 0.3617, "step": 6037 }, { "epoch": 1.05, "learning_rate": 2.935013365567456e-05, "loss": 0.3669, "step": 6038 }, { "epoch": 1.05, "learning_rate": 2.9343309346594222e-05, "loss": 0.3673, "step": 6039 }, { "epoch": 1.05, "learning_rate": 2.933648486679321e-05, "loss": 0.3632, "step": 6040 }, { "epoch": 1.05, "learning_rate": 2.9329660216903578e-05, "loss": 0.3651, "step": 6041 }, { "epoch": 1.05, "learning_rate": 2.9322835397557413e-05, "loss": 0.3838, "step": 6042 }, { "epoch": 1.05, "learning_rate": 2.9316010409386794e-05, "loss": 0.3803, "step": 6043 }, { "epoch": 1.05, "learning_rate": 2.9309185253023845e-05, "loss": 0.3594, "step": 6044 }, { "epoch": 1.05, "learning_rate": 2.930235992910066e-05, "loss": 0.3796, "step": 6045 }, { "epoch": 1.05, "learning_rate": 2.92955344382494e-05, "loss": 0.3726, "step": 6046 }, { "epoch": 1.05, "learning_rate": 2.9288708781102205e-05, "loss": 0.3723, "step": 6047 }, { "epoch": 1.05, "learning_rate": 2.9281882958291236e-05, "loss": 0.3558, "step": 6048 }, { "epoch": 1.05, "learning_rate": 2.92750569704487e-05, "loss": 0.3819, "step": 6049 }, { "epoch": 1.05, "learning_rate": 2.926823081820677e-05, "loss": 0.3646, "step": 6050 }, { "epoch": 1.05, "learning_rate": 2.926140450219768e-05, "loss": 0.3583, "step": 6051 }, { "epoch": 1.06, "learning_rate": 2.925457802305364e-05, "loss": 0.3822, "step": 6052 }, { "epoch": 1.06, "learning_rate": 2.9247751381406912e-05, "loss": 0.3714, "step": 6053 }, { "epoch": 1.06, "learning_rate": 2.9240924577889738e-05, "loss": 0.3723, "step": 6054 }, { "epoch": 1.06, "learning_rate": 2.9234097613134405e-05, "loss": 0.3891, "step": 6055 }, { "epoch": 1.06, "learning_rate": 2.9227270487773194e-05, "loss": 0.3522, "step": 6056 }, { "epoch": 1.06, "learning_rate": 2.9220443202438415e-05, "loss": 0.37, "step": 6057 }, { "epoch": 1.06, "learning_rate": 2.9213615757762382e-05, "loss": 0.3706, "step": 6058 }, { "epoch": 1.06, "learning_rate": 2.9206788154377423e-05, "loss": 0.37, "step": 6059 }, { "epoch": 1.06, "learning_rate": 2.91999603929159e-05, "loss": 0.362, "step": 6060 }, { "epoch": 1.06, "learning_rate": 2.9193132474010175e-05, "loss": 0.3692, "step": 6061 }, { "epoch": 1.06, "learning_rate": 2.91863043982926e-05, "loss": 0.3538, "step": 6062 }, { "epoch": 1.06, "learning_rate": 2.9179476166395603e-05, "loss": 0.3813, "step": 6063 }, { "epoch": 1.06, "learning_rate": 2.917264777895156e-05, "loss": 0.3868, "step": 6064 }, { "epoch": 1.06, "learning_rate": 2.9165819236592917e-05, "loss": 0.3795, "step": 6065 }, { "epoch": 1.06, "learning_rate": 2.915899053995209e-05, "loss": 0.3847, "step": 6066 }, { "epoch": 1.06, "learning_rate": 2.915216168966153e-05, "loss": 0.3663, "step": 6067 }, { "epoch": 1.06, "learning_rate": 2.9145332686353716e-05, "loss": 0.3611, "step": 6068 }, { "epoch": 1.06, "learning_rate": 2.9138503530661104e-05, "loss": 0.3644, "step": 6069 }, { "epoch": 1.06, "learning_rate": 2.9131674223216204e-05, "loss": 0.3729, "step": 6070 }, { "epoch": 1.06, "learning_rate": 2.912484476465151e-05, "loss": 0.3541, "step": 6071 }, { "epoch": 1.06, "learning_rate": 2.9118015155599548e-05, "loss": 0.3708, "step": 6072 }, { "epoch": 1.06, "learning_rate": 2.9111185396692847e-05, "loss": 0.3829, "step": 6073 }, { "epoch": 1.06, "learning_rate": 2.910435548856396e-05, "loss": 0.3829, "step": 6074 }, { "epoch": 1.06, "learning_rate": 2.9097525431845452e-05, "loss": 0.3757, "step": 6075 }, { "epoch": 1.06, "learning_rate": 2.9090695227169878e-05, "loss": 0.3804, "step": 6076 }, { "epoch": 1.06, "learning_rate": 2.908386487516986e-05, "loss": 0.3417, "step": 6077 }, { "epoch": 1.06, "learning_rate": 2.907703437647797e-05, "loss": 0.3453, "step": 6078 }, { "epoch": 1.06, "learning_rate": 2.9070203731726836e-05, "loss": 0.3744, "step": 6079 }, { "epoch": 1.06, "learning_rate": 2.90633729415491e-05, "loss": 0.3619, "step": 6080 }, { "epoch": 1.06, "learning_rate": 2.9056542006577374e-05, "loss": 0.3664, "step": 6081 }, { "epoch": 1.06, "learning_rate": 2.9049710927444344e-05, "loss": 0.3657, "step": 6082 }, { "epoch": 1.06, "learning_rate": 2.9042879704782677e-05, "loss": 0.3652, "step": 6083 }, { "epoch": 1.06, "learning_rate": 2.9036048339225043e-05, "loss": 0.3691, "step": 6084 }, { "epoch": 1.06, "learning_rate": 2.902921683140414e-05, "loss": 0.3578, "step": 6085 }, { "epoch": 1.06, "learning_rate": 2.9022385181952683e-05, "loss": 0.3783, "step": 6086 }, { "epoch": 1.06, "learning_rate": 2.9015553391503397e-05, "loss": 0.3612, "step": 6087 }, { "epoch": 1.06, "learning_rate": 2.9008721460689016e-05, "loss": 0.3745, "step": 6088 }, { "epoch": 1.06, "learning_rate": 2.9001889390142285e-05, "loss": 0.3468, "step": 6089 }, { "epoch": 1.06, "learning_rate": 2.899505718049596e-05, "loss": 0.38, "step": 6090 }, { "epoch": 1.06, "learning_rate": 2.8988224832382832e-05, "loss": 0.3515, "step": 6091 }, { "epoch": 1.06, "learning_rate": 2.898139234643568e-05, "loss": 0.364, "step": 6092 }, { "epoch": 1.06, "learning_rate": 2.897455972328731e-05, "loss": 0.3747, "step": 6093 }, { "epoch": 1.06, "learning_rate": 2.896772696357053e-05, "loss": 0.3731, "step": 6094 }, { "epoch": 1.06, "learning_rate": 2.8960894067918154e-05, "loss": 0.3686, "step": 6095 }, { "epoch": 1.06, "learning_rate": 2.895406103696304e-05, "loss": 0.3595, "step": 6096 }, { "epoch": 1.06, "learning_rate": 2.894722787133803e-05, "loss": 0.3716, "step": 6097 }, { "epoch": 1.06, "learning_rate": 2.894039457167598e-05, "loss": 0.3885, "step": 6098 }, { "epoch": 1.06, "learning_rate": 2.8933561138609783e-05, "loss": 0.3647, "step": 6099 }, { "epoch": 1.06, "learning_rate": 2.89267275727723e-05, "loss": 0.3649, "step": 6100 }, { "epoch": 1.06, "learning_rate": 2.8919893874796466e-05, "loss": 0.3869, "step": 6101 }, { "epoch": 1.06, "learning_rate": 2.8913060045315165e-05, "loss": 0.361, "step": 6102 }, { "epoch": 1.06, "learning_rate": 2.8906226084961334e-05, "loss": 0.3826, "step": 6103 }, { "epoch": 1.06, "learning_rate": 2.8899391994367913e-05, "loss": 0.3811, "step": 6104 }, { "epoch": 1.06, "learning_rate": 2.8892557774167843e-05, "loss": 0.3648, "step": 6105 }, { "epoch": 1.06, "learning_rate": 2.888572342499409e-05, "loss": 0.3682, "step": 6106 }, { "epoch": 1.06, "learning_rate": 2.887888894747962e-05, "loss": 0.3706, "step": 6107 }, { "epoch": 1.06, "learning_rate": 2.8872054342257427e-05, "loss": 0.3779, "step": 6108 }, { "epoch": 1.07, "learning_rate": 2.88652196099605e-05, "loss": 0.3538, "step": 6109 }, { "epoch": 1.07, "learning_rate": 2.885838475122185e-05, "loss": 0.3879, "step": 6110 }, { "epoch": 1.07, "learning_rate": 2.8851549766674495e-05, "loss": 0.3661, "step": 6111 }, { "epoch": 1.07, "learning_rate": 2.8844714656951473e-05, "loss": 0.3577, "step": 6112 }, { "epoch": 1.07, "learning_rate": 2.8837879422685822e-05, "loss": 0.371, "step": 6113 }, { "epoch": 1.07, "learning_rate": 2.8831044064510588e-05, "loss": 0.3555, "step": 6114 }, { "epoch": 1.07, "learning_rate": 2.882420858305886e-05, "loss": 0.367, "step": 6115 }, { "epoch": 1.07, "learning_rate": 2.8817372978963702e-05, "loss": 0.3668, "step": 6116 }, { "epoch": 1.07, "learning_rate": 2.8810537252858196e-05, "loss": 0.3734, "step": 6117 }, { "epoch": 1.07, "learning_rate": 2.8803701405375457e-05, "loss": 0.3672, "step": 6118 }, { "epoch": 1.07, "learning_rate": 2.8796865437148584e-05, "loss": 0.3718, "step": 6119 }, { "epoch": 1.07, "learning_rate": 2.8790029348810705e-05, "loss": 0.3721, "step": 6120 }, { "epoch": 1.07, "learning_rate": 2.8783193140994956e-05, "loss": 0.3872, "step": 6121 }, { "epoch": 1.07, "learning_rate": 2.877635681433448e-05, "loss": 0.3713, "step": 6122 }, { "epoch": 1.07, "learning_rate": 2.8769520369462433e-05, "loss": 0.3897, "step": 6123 }, { "epoch": 1.07, "learning_rate": 2.8762683807011987e-05, "loss": 0.3636, "step": 6124 }, { "epoch": 1.07, "learning_rate": 2.8755847127616303e-05, "loss": 0.3772, "step": 6125 }, { "epoch": 1.07, "learning_rate": 2.8749010331908593e-05, "loss": 0.3673, "step": 6126 }, { "epoch": 1.07, "learning_rate": 2.874217342052204e-05, "loss": 0.3697, "step": 6127 }, { "epoch": 1.07, "learning_rate": 2.8735336394089855e-05, "loss": 0.3727, "step": 6128 }, { "epoch": 1.07, "learning_rate": 2.8728499253245272e-05, "loss": 0.3702, "step": 6129 }, { "epoch": 1.07, "learning_rate": 2.8721661998621514e-05, "loss": 0.3824, "step": 6130 }, { "epoch": 1.07, "learning_rate": 2.871482463085181e-05, "loss": 0.3777, "step": 6131 }, { "epoch": 1.07, "learning_rate": 2.8707987150569432e-05, "loss": 0.3692, "step": 6132 }, { "epoch": 1.07, "learning_rate": 2.8701149558407635e-05, "loss": 0.3659, "step": 6133 }, { "epoch": 1.07, "learning_rate": 2.8694311854999683e-05, "loss": 0.3738, "step": 6134 }, { "epoch": 1.07, "learning_rate": 2.8687474040978883e-05, "loss": 0.363, "step": 6135 }, { "epoch": 1.07, "learning_rate": 2.8680636116978506e-05, "loss": 0.3566, "step": 6136 }, { "epoch": 1.07, "learning_rate": 2.867379808363187e-05, "loss": 0.3646, "step": 6137 }, { "epoch": 1.07, "learning_rate": 2.866695994157228e-05, "loss": 0.376, "step": 6138 }, { "epoch": 1.07, "learning_rate": 2.8660121691433057e-05, "loss": 0.377, "step": 6139 }, { "epoch": 1.07, "learning_rate": 2.8653283333847552e-05, "loss": 0.3739, "step": 6140 }, { "epoch": 1.07, "learning_rate": 2.8646444869449096e-05, "loss": 0.3699, "step": 6141 }, { "epoch": 1.07, "learning_rate": 2.863960629887104e-05, "loss": 0.3641, "step": 6142 }, { "epoch": 1.07, "learning_rate": 2.8632767622746752e-05, "loss": 0.3499, "step": 6143 }, { "epoch": 1.07, "learning_rate": 2.862592884170962e-05, "loss": 0.37, "step": 6144 }, { "epoch": 1.07, "learning_rate": 2.8619089956392997e-05, "loss": 0.375, "step": 6145 }, { "epoch": 1.07, "learning_rate": 2.8612250967430306e-05, "loss": 0.3679, "step": 6146 }, { "epoch": 1.07, "learning_rate": 2.860541187545493e-05, "loss": 0.3616, "step": 6147 }, { "epoch": 1.07, "learning_rate": 2.8598572681100295e-05, "loss": 0.3592, "step": 6148 }, { "epoch": 1.07, "learning_rate": 2.8591733384999808e-05, "loss": 0.3627, "step": 6149 }, { "epoch": 1.07, "learning_rate": 2.8584893987786908e-05, "loss": 0.3445, "step": 6150 }, { "epoch": 1.07, "learning_rate": 2.8578054490095036e-05, "loss": 0.3765, "step": 6151 }, { "epoch": 1.07, "learning_rate": 2.857121489255764e-05, "loss": 0.3561, "step": 6152 }, { "epoch": 1.07, "learning_rate": 2.8564375195808175e-05, "loss": 0.3789, "step": 6153 }, { "epoch": 1.07, "learning_rate": 2.8557535400480123e-05, "loss": 0.3692, "step": 6154 }, { "epoch": 1.07, "learning_rate": 2.855069550720694e-05, "loss": 0.3732, "step": 6155 }, { "epoch": 1.07, "learning_rate": 2.8543855516622132e-05, "loss": 0.3658, "step": 6156 }, { "epoch": 1.07, "learning_rate": 2.853701542935918e-05, "loss": 0.3629, "step": 6157 }, { "epoch": 1.07, "learning_rate": 2.8530175246051594e-05, "loss": 0.3628, "step": 6158 }, { "epoch": 1.07, "learning_rate": 2.8523334967332895e-05, "loss": 0.3846, "step": 6159 }, { "epoch": 1.07, "learning_rate": 2.851649459383659e-05, "loss": 0.3662, "step": 6160 }, { "epoch": 1.07, "learning_rate": 2.8509654126196217e-05, "loss": 0.3821, "step": 6161 }, { "epoch": 1.07, "learning_rate": 2.8502813565045318e-05, "loss": 0.3715, "step": 6162 }, { "epoch": 1.07, "learning_rate": 2.8495972911017436e-05, "loss": 0.374, "step": 6163 }, { "epoch": 1.07, "learning_rate": 2.8489132164746135e-05, "loss": 0.3712, "step": 6164 }, { "epoch": 1.07, "learning_rate": 2.8482291326864975e-05, "loss": 0.3634, "step": 6165 }, { "epoch": 1.08, "learning_rate": 2.847545039800753e-05, "loss": 0.3728, "step": 6166 }, { "epoch": 1.08, "learning_rate": 2.8468609378807376e-05, "loss": 0.3657, "step": 6167 }, { "epoch": 1.08, "learning_rate": 2.8461768269898116e-05, "loss": 0.3528, "step": 6168 }, { "epoch": 1.08, "learning_rate": 2.845492707191334e-05, "loss": 0.354, "step": 6169 }, { "epoch": 1.08, "learning_rate": 2.844808578548666e-05, "loss": 0.3614, "step": 6170 }, { "epoch": 1.08, "learning_rate": 2.844124441125169e-05, "loss": 0.3716, "step": 6171 }, { "epoch": 1.08, "learning_rate": 2.8434402949842048e-05, "loss": 0.3851, "step": 6172 }, { "epoch": 1.08, "learning_rate": 2.842756140189138e-05, "loss": 0.3639, "step": 6173 }, { "epoch": 1.08, "learning_rate": 2.842071976803331e-05, "loss": 0.3505, "step": 6174 }, { "epoch": 1.08, "learning_rate": 2.8413878048901488e-05, "loss": 0.3585, "step": 6175 }, { "epoch": 1.08, "learning_rate": 2.8407036245129576e-05, "loss": 0.3634, "step": 6176 }, { "epoch": 1.08, "learning_rate": 2.8400194357351233e-05, "loss": 0.3743, "step": 6177 }, { "epoch": 1.08, "learning_rate": 2.839335238620014e-05, "loss": 0.37, "step": 6178 }, { "epoch": 1.08, "learning_rate": 2.8386510332309952e-05, "loss": 0.3667, "step": 6179 }, { "epoch": 1.08, "learning_rate": 2.8379668196314385e-05, "loss": 0.3798, "step": 6180 }, { "epoch": 1.08, "learning_rate": 2.8372825978847102e-05, "loss": 0.3703, "step": 6181 }, { "epoch": 1.08, "learning_rate": 2.836598368054183e-05, "loss": 0.3661, "step": 6182 }, { "epoch": 1.08, "learning_rate": 2.835914130203227e-05, "loss": 0.3733, "step": 6183 }, { "epoch": 1.08, "learning_rate": 2.8352298843952137e-05, "loss": 0.3818, "step": 6184 }, { "epoch": 1.08, "learning_rate": 2.8345456306935153e-05, "loss": 0.3713, "step": 6185 }, { "epoch": 1.08, "learning_rate": 2.833861369161504e-05, "loss": 0.3763, "step": 6186 }, { "epoch": 1.08, "learning_rate": 2.8331770998625566e-05, "loss": 0.3718, "step": 6187 }, { "epoch": 1.08, "learning_rate": 2.832492822860045e-05, "loss": 0.3463, "step": 6188 }, { "epoch": 1.08, "learning_rate": 2.8318085382173454e-05, "loss": 0.3699, "step": 6189 }, { "epoch": 1.08, "learning_rate": 2.831124245997833e-05, "loss": 0.3743, "step": 6190 }, { "epoch": 1.08, "learning_rate": 2.830439946264886e-05, "loss": 0.379, "step": 6191 }, { "epoch": 1.08, "learning_rate": 2.8297556390818808e-05, "loss": 0.3678, "step": 6192 }, { "epoch": 1.08, "learning_rate": 2.8290713245121952e-05, "loss": 0.3554, "step": 6193 }, { "epoch": 1.08, "learning_rate": 2.828387002619208e-05, "loss": 0.3802, "step": 6194 }, { "epoch": 1.08, "learning_rate": 2.8277026734662993e-05, "loss": 0.3718, "step": 6195 }, { "epoch": 1.08, "learning_rate": 2.8270183371168495e-05, "loss": 0.3761, "step": 6196 }, { "epoch": 1.08, "learning_rate": 2.826333993634238e-05, "loss": 0.3878, "step": 6197 }, { "epoch": 1.08, "learning_rate": 2.8256496430818476e-05, "loss": 0.3764, "step": 6198 }, { "epoch": 1.08, "learning_rate": 2.8249652855230597e-05, "loss": 0.3707, "step": 6199 }, { "epoch": 1.08, "learning_rate": 2.8242809210212555e-05, "loss": 0.3629, "step": 6200 }, { "epoch": 1.08, "learning_rate": 2.8235965496398214e-05, "loss": 0.3689, "step": 6201 }, { "epoch": 1.08, "learning_rate": 2.8229121714421396e-05, "loss": 0.3582, "step": 6202 }, { "epoch": 1.08, "learning_rate": 2.8222277864915947e-05, "loss": 0.3662, "step": 6203 }, { "epoch": 1.08, "learning_rate": 2.8215433948515735e-05, "loss": 0.3702, "step": 6204 }, { "epoch": 1.08, "learning_rate": 2.8208589965854593e-05, "loss": 0.3568, "step": 6205 }, { "epoch": 1.08, "learning_rate": 2.820174591756642e-05, "loss": 0.386, "step": 6206 }, { "epoch": 1.08, "learning_rate": 2.819490180428505e-05, "loss": 0.3642, "step": 6207 }, { "epoch": 1.08, "learning_rate": 2.818805762664438e-05, "loss": 0.3784, "step": 6208 }, { "epoch": 1.08, "learning_rate": 2.81812133852783e-05, "loss": 0.3759, "step": 6209 }, { "epoch": 1.08, "learning_rate": 2.8174369080820683e-05, "loss": 0.362, "step": 6210 }, { "epoch": 1.08, "learning_rate": 2.816752471390543e-05, "loss": 0.3672, "step": 6211 }, { "epoch": 1.08, "learning_rate": 2.8160680285166447e-05, "loss": 0.3559, "step": 6212 }, { "epoch": 1.08, "learning_rate": 2.8153835795237632e-05, "loss": 0.3651, "step": 6213 }, { "epoch": 1.08, "learning_rate": 2.8146991244752903e-05, "loss": 0.3815, "step": 6214 }, { "epoch": 1.08, "learning_rate": 2.8140146634346177e-05, "loss": 0.3639, "step": 6215 }, { "epoch": 1.08, "learning_rate": 2.8133301964651366e-05, "loss": 0.3709, "step": 6216 }, { "epoch": 1.08, "learning_rate": 2.8126457236302417e-05, "loss": 0.3743, "step": 6217 }, { "epoch": 1.08, "learning_rate": 2.8119612449933256e-05, "loss": 0.3697, "step": 6218 }, { "epoch": 1.08, "learning_rate": 2.8112767606177813e-05, "loss": 0.3721, "step": 6219 }, { "epoch": 1.08, "learning_rate": 2.810592270567005e-05, "loss": 0.3813, "step": 6220 }, { "epoch": 1.08, "learning_rate": 2.8099077749043907e-05, "loss": 0.3703, "step": 6221 }, { "epoch": 1.08, "learning_rate": 2.8092232736933334e-05, "loss": 0.3802, "step": 6222 }, { "epoch": 1.08, "learning_rate": 2.808538766997231e-05, "loss": 0.369, "step": 6223 }, { "epoch": 1.09, "learning_rate": 2.8078542548794778e-05, "loss": 0.393, "step": 6224 }, { "epoch": 1.09, "learning_rate": 2.807169737403473e-05, "loss": 0.3567, "step": 6225 }, { "epoch": 1.09, "learning_rate": 2.8064852146326116e-05, "loss": 0.375, "step": 6226 }, { "epoch": 1.09, "learning_rate": 2.8058006866302934e-05, "loss": 0.3867, "step": 6227 }, { "epoch": 1.09, "learning_rate": 2.8051161534599173e-05, "loss": 0.3451, "step": 6228 }, { "epoch": 1.09, "learning_rate": 2.804431615184881e-05, "loss": 0.3594, "step": 6229 }, { "epoch": 1.09, "learning_rate": 2.8037470718685848e-05, "loss": 0.3795, "step": 6230 }, { "epoch": 1.09, "learning_rate": 2.8030625235744286e-05, "loss": 0.3678, "step": 6231 }, { "epoch": 1.09, "learning_rate": 2.802377970365812e-05, "loss": 0.363, "step": 6232 }, { "epoch": 1.09, "learning_rate": 2.8016934123061367e-05, "loss": 0.3614, "step": 6233 }, { "epoch": 1.09, "learning_rate": 2.801008849458804e-05, "loss": 0.3682, "step": 6234 }, { "epoch": 1.09, "learning_rate": 2.8003242818872156e-05, "loss": 0.3592, "step": 6235 }, { "epoch": 1.09, "learning_rate": 2.7996397096547726e-05, "loss": 0.3595, "step": 6236 }, { "epoch": 1.09, "learning_rate": 2.798955132824879e-05, "loss": 0.3778, "step": 6237 }, { "epoch": 1.09, "learning_rate": 2.7982705514609374e-05, "loss": 0.359, "step": 6238 }, { "epoch": 1.09, "learning_rate": 2.7975859656263508e-05, "loss": 0.365, "step": 6239 }, { "epoch": 1.09, "learning_rate": 2.7969013753845247e-05, "loss": 0.3738, "step": 6240 }, { "epoch": 1.09, "learning_rate": 2.79621678079886e-05, "loss": 0.3625, "step": 6241 }, { "epoch": 1.09, "learning_rate": 2.7955321819327656e-05, "loss": 0.3685, "step": 6242 }, { "epoch": 1.09, "learning_rate": 2.7948475788496435e-05, "loss": 0.3746, "step": 6243 }, { "epoch": 1.09, "learning_rate": 2.7941629716129002e-05, "loss": 0.3824, "step": 6244 }, { "epoch": 1.09, "learning_rate": 2.7934783602859416e-05, "loss": 0.3734, "step": 6245 }, { "epoch": 1.09, "learning_rate": 2.792793744932174e-05, "loss": 0.3658, "step": 6246 }, { "epoch": 1.09, "learning_rate": 2.792109125615004e-05, "loss": 0.3612, "step": 6247 }, { "epoch": 1.09, "learning_rate": 2.791424502397838e-05, "loss": 0.3799, "step": 6248 }, { "epoch": 1.09, "learning_rate": 2.7907398753440846e-05, "loss": 0.3572, "step": 6249 }, { "epoch": 1.09, "learning_rate": 2.7900552445171495e-05, "loss": 0.3739, "step": 6250 }, { "epoch": 1.09, "learning_rate": 2.7893706099804422e-05, "loss": 0.3636, "step": 6251 }, { "epoch": 1.09, "learning_rate": 2.7886859717973712e-05, "loss": 0.3817, "step": 6252 }, { "epoch": 1.09, "learning_rate": 2.788001330031345e-05, "loss": 0.3813, "step": 6253 }, { "epoch": 1.09, "learning_rate": 2.7873166847457727e-05, "loss": 0.3641, "step": 6254 }, { "epoch": 1.09, "learning_rate": 2.7866320360040627e-05, "loss": 0.3862, "step": 6255 }, { "epoch": 1.09, "learning_rate": 2.7859473838696265e-05, "loss": 0.3797, "step": 6256 }, { "epoch": 1.09, "learning_rate": 2.7852627284058718e-05, "loss": 0.3828, "step": 6257 }, { "epoch": 1.09, "learning_rate": 2.7845780696762108e-05, "loss": 0.3632, "step": 6258 }, { "epoch": 1.09, "learning_rate": 2.7838934077440538e-05, "loss": 0.3721, "step": 6259 }, { "epoch": 1.09, "learning_rate": 2.783208742672811e-05, "loss": 0.3669, "step": 6260 }, { "epoch": 1.09, "learning_rate": 2.7825240745258945e-05, "loss": 0.3633, "step": 6261 }, { "epoch": 1.09, "learning_rate": 2.781839403366715e-05, "loss": 0.3687, "step": 6262 }, { "epoch": 1.09, "learning_rate": 2.7811547292586852e-05, "loss": 0.3772, "step": 6263 }, { "epoch": 1.09, "learning_rate": 2.7804700522652154e-05, "loss": 0.3635, "step": 6264 }, { "epoch": 1.09, "learning_rate": 2.7797853724497202e-05, "loss": 0.3715, "step": 6265 }, { "epoch": 1.09, "learning_rate": 2.7791006898756105e-05, "loss": 0.3754, "step": 6266 }, { "epoch": 1.09, "learning_rate": 2.7784160046063e-05, "loss": 0.3753, "step": 6267 }, { "epoch": 1.09, "learning_rate": 2.7777313167052022e-05, "loss": 0.3591, "step": 6268 }, { "epoch": 1.09, "learning_rate": 2.7770466262357286e-05, "loss": 0.3907, "step": 6269 }, { "epoch": 1.09, "learning_rate": 2.776361933261294e-05, "loss": 0.3559, "step": 6270 }, { "epoch": 1.09, "learning_rate": 2.7756772378453133e-05, "loss": 0.3827, "step": 6271 }, { "epoch": 1.09, "learning_rate": 2.7749925400511972e-05, "loss": 0.3747, "step": 6272 }, { "epoch": 1.09, "learning_rate": 2.7743078399423643e-05, "loss": 0.354, "step": 6273 }, { "epoch": 1.09, "learning_rate": 2.7736231375822253e-05, "loss": 0.372, "step": 6274 }, { "epoch": 1.09, "learning_rate": 2.7729384330341967e-05, "loss": 0.3658, "step": 6275 }, { "epoch": 1.09, "learning_rate": 2.7722537263616937e-05, "loss": 0.3662, "step": 6276 }, { "epoch": 1.09, "learning_rate": 2.77156901762813e-05, "loss": 0.377, "step": 6277 }, { "epoch": 1.09, "learning_rate": 2.7708843068969216e-05, "loss": 0.3648, "step": 6278 }, { "epoch": 1.09, "learning_rate": 2.7701995942314846e-05, "loss": 0.3754, "step": 6279 }, { "epoch": 1.09, "learning_rate": 2.7695148796952338e-05, "loss": 0.3712, "step": 6280 }, { "epoch": 1.1, "learning_rate": 2.7688301633515846e-05, "loss": 0.3625, "step": 6281 }, { "epoch": 1.1, "learning_rate": 2.7681454452639533e-05, "loss": 0.3685, "step": 6282 }, { "epoch": 1.1, "learning_rate": 2.7674607254957568e-05, "loss": 0.3685, "step": 6283 }, { "epoch": 1.1, "learning_rate": 2.7667760041104107e-05, "loss": 0.371, "step": 6284 }, { "epoch": 1.1, "learning_rate": 2.7660912811713318e-05, "loss": 0.364, "step": 6285 }, { "epoch": 1.1, "learning_rate": 2.7654065567419356e-05, "loss": 0.3896, "step": 6286 }, { "epoch": 1.1, "learning_rate": 2.76472183088564e-05, "loss": 0.3618, "step": 6287 }, { "epoch": 1.1, "learning_rate": 2.7640371036658615e-05, "loss": 0.3746, "step": 6288 }, { "epoch": 1.1, "learning_rate": 2.7633523751460165e-05, "loss": 0.3725, "step": 6289 }, { "epoch": 1.1, "learning_rate": 2.7626676453895238e-05, "loss": 0.3526, "step": 6290 }, { "epoch": 1.1, "learning_rate": 2.761982914459798e-05, "loss": 0.3589, "step": 6291 }, { "epoch": 1.1, "learning_rate": 2.761298182420259e-05, "loss": 0.3601, "step": 6292 }, { "epoch": 1.1, "learning_rate": 2.7606134493343227e-05, "loss": 0.3629, "step": 6293 }, { "epoch": 1.1, "learning_rate": 2.7599287152654063e-05, "loss": 0.3848, "step": 6294 }, { "epoch": 1.1, "learning_rate": 2.7592439802769286e-05, "loss": 0.3667, "step": 6295 }, { "epoch": 1.1, "learning_rate": 2.7585592444323073e-05, "loss": 0.3621, "step": 6296 }, { "epoch": 1.1, "learning_rate": 2.757874507794959e-05, "loss": 0.3896, "step": 6297 }, { "epoch": 1.1, "learning_rate": 2.757189770428302e-05, "loss": 0.3545, "step": 6298 }, { "epoch": 1.1, "learning_rate": 2.756505032395754e-05, "loss": 0.3696, "step": 6299 }, { "epoch": 1.1, "learning_rate": 2.7558202937607348e-05, "loss": 0.3767, "step": 6300 }, { "epoch": 1.1, "learning_rate": 2.7551355545866597e-05, "loss": 0.3669, "step": 6301 }, { "epoch": 1.1, "learning_rate": 2.7544508149369485e-05, "loss": 0.3841, "step": 6302 }, { "epoch": 1.1, "learning_rate": 2.7537660748750182e-05, "loss": 0.3758, "step": 6303 }, { "epoch": 1.1, "learning_rate": 2.7530813344642893e-05, "loss": 0.3741, "step": 6304 }, { "epoch": 1.1, "learning_rate": 2.7523965937681763e-05, "loss": 0.376, "step": 6305 }, { "epoch": 1.1, "learning_rate": 2.7517118528501e-05, "loss": 0.359, "step": 6306 }, { "epoch": 1.1, "learning_rate": 2.7510271117734792e-05, "loss": 0.3704, "step": 6307 }, { "epoch": 1.1, "learning_rate": 2.7503423706017294e-05, "loss": 0.368, "step": 6308 }, { "epoch": 1.1, "learning_rate": 2.7496576293982718e-05, "loss": 0.3689, "step": 6309 }, { "epoch": 1.1, "learning_rate": 2.7489728882265227e-05, "loss": 0.3714, "step": 6310 }, { "epoch": 1.1, "learning_rate": 2.7482881471499005e-05, "loss": 0.363, "step": 6311 }, { "epoch": 1.1, "learning_rate": 2.747603406231824e-05, "loss": 0.3896, "step": 6312 }, { "epoch": 1.1, "learning_rate": 2.7469186655357123e-05, "loss": 0.3605, "step": 6313 }, { "epoch": 1.1, "learning_rate": 2.7462339251249823e-05, "loss": 0.3606, "step": 6314 }, { "epoch": 1.1, "learning_rate": 2.745549185063052e-05, "loss": 0.3556, "step": 6315 }, { "epoch": 1.1, "learning_rate": 2.7448644454133408e-05, "loss": 0.3765, "step": 6316 }, { "epoch": 1.1, "learning_rate": 2.744179706239266e-05, "loss": 0.3864, "step": 6317 }, { "epoch": 1.1, "learning_rate": 2.7434949676042458e-05, "loss": 0.375, "step": 6318 }, { "epoch": 1.1, "learning_rate": 2.742810229571699e-05, "loss": 0.3831, "step": 6319 }, { "epoch": 1.1, "learning_rate": 2.7421254922050416e-05, "loss": 0.368, "step": 6320 }, { "epoch": 1.1, "learning_rate": 2.741440755567694e-05, "loss": 0.3675, "step": 6321 }, { "epoch": 1.1, "learning_rate": 2.740756019723072e-05, "loss": 0.3533, "step": 6322 }, { "epoch": 1.1, "learning_rate": 2.740071284734594e-05, "loss": 0.3831, "step": 6323 }, { "epoch": 1.1, "learning_rate": 2.739386550665678e-05, "loss": 0.3881, "step": 6324 }, { "epoch": 1.1, "learning_rate": 2.7387018175797412e-05, "loss": 0.3637, "step": 6325 }, { "epoch": 1.1, "learning_rate": 2.7380170855402026e-05, "loss": 0.3725, "step": 6326 }, { "epoch": 1.1, "learning_rate": 2.7373323546104774e-05, "loss": 0.3911, "step": 6327 }, { "epoch": 1.1, "learning_rate": 2.7366476248539837e-05, "loss": 0.3568, "step": 6328 }, { "epoch": 1.1, "learning_rate": 2.7359628963341387e-05, "loss": 0.3799, "step": 6329 }, { "epoch": 1.1, "learning_rate": 2.735278169114361e-05, "loss": 0.3862, "step": 6330 }, { "epoch": 1.1, "learning_rate": 2.7345934432580656e-05, "loss": 0.3649, "step": 6331 }, { "epoch": 1.1, "learning_rate": 2.7339087188286698e-05, "loss": 0.3823, "step": 6332 }, { "epoch": 1.1, "learning_rate": 2.7332239958895912e-05, "loss": 0.3664, "step": 6333 }, { "epoch": 1.1, "learning_rate": 2.7325392745042434e-05, "loss": 0.3635, "step": 6334 }, { "epoch": 1.1, "learning_rate": 2.7318545547360476e-05, "loss": 0.3668, "step": 6335 }, { "epoch": 1.1, "learning_rate": 2.731169836648417e-05, "loss": 0.3701, "step": 6336 }, { "epoch": 1.1, "learning_rate": 2.7304851203047678e-05, "loss": 0.3667, "step": 6337 }, { "epoch": 1.11, "learning_rate": 2.7298004057685166e-05, "loss": 0.3769, "step": 6338 }, { "epoch": 1.11, "learning_rate": 2.7291156931030782e-05, "loss": 0.3694, "step": 6339 }, { "epoch": 1.11, "learning_rate": 2.7284309823718708e-05, "loss": 0.3655, "step": 6340 }, { "epoch": 1.11, "learning_rate": 2.727746273638307e-05, "loss": 0.3815, "step": 6341 }, { "epoch": 1.11, "learning_rate": 2.7270615669658038e-05, "loss": 0.3771, "step": 6342 }, { "epoch": 1.11, "learning_rate": 2.7263768624177756e-05, "loss": 0.3489, "step": 6343 }, { "epoch": 1.11, "learning_rate": 2.725692160057637e-05, "loss": 0.3668, "step": 6344 }, { "epoch": 1.11, "learning_rate": 2.7250074599488023e-05, "loss": 0.3683, "step": 6345 }, { "epoch": 1.11, "learning_rate": 2.7243227621546875e-05, "loss": 0.3898, "step": 6346 }, { "epoch": 1.11, "learning_rate": 2.723638066738707e-05, "loss": 0.3597, "step": 6347 }, { "epoch": 1.11, "learning_rate": 2.722953373764272e-05, "loss": 0.3564, "step": 6348 }, { "epoch": 1.11, "learning_rate": 2.722268683294799e-05, "loss": 0.3683, "step": 6349 }, { "epoch": 1.11, "learning_rate": 2.7215839953937e-05, "loss": 0.3725, "step": 6350 }, { "epoch": 1.11, "learning_rate": 2.720899310124389e-05, "loss": 0.3554, "step": 6351 }, { "epoch": 1.11, "learning_rate": 2.7202146275502803e-05, "loss": 0.3672, "step": 6352 }, { "epoch": 1.11, "learning_rate": 2.7195299477347848e-05, "loss": 0.361, "step": 6353 }, { "epoch": 1.11, "learning_rate": 2.718845270741316e-05, "loss": 0.3825, "step": 6354 }, { "epoch": 1.11, "learning_rate": 2.7181605966332857e-05, "loss": 0.3664, "step": 6355 }, { "epoch": 1.11, "learning_rate": 2.717475925474106e-05, "loss": 0.3569, "step": 6356 }, { "epoch": 1.11, "learning_rate": 2.7167912573271893e-05, "loss": 0.3915, "step": 6357 }, { "epoch": 1.11, "learning_rate": 2.716106592255947e-05, "loss": 0.3639, "step": 6358 }, { "epoch": 1.11, "learning_rate": 2.71542193032379e-05, "loss": 0.3602, "step": 6359 }, { "epoch": 1.11, "learning_rate": 2.7147372715941287e-05, "loss": 0.3552, "step": 6360 }, { "epoch": 1.11, "learning_rate": 2.7140526161303743e-05, "loss": 0.3797, "step": 6361 }, { "epoch": 1.11, "learning_rate": 2.7133679639959375e-05, "loss": 0.3577, "step": 6362 }, { "epoch": 1.11, "learning_rate": 2.7126833152542275e-05, "loss": 0.3617, "step": 6363 }, { "epoch": 1.11, "learning_rate": 2.7119986699686555e-05, "loss": 0.3711, "step": 6364 }, { "epoch": 1.11, "learning_rate": 2.711314028202629e-05, "loss": 0.3468, "step": 6365 }, { "epoch": 1.11, "learning_rate": 2.7106293900195584e-05, "loss": 0.3731, "step": 6366 }, { "epoch": 1.11, "learning_rate": 2.7099447554828504e-05, "loss": 0.3663, "step": 6367 }, { "epoch": 1.11, "learning_rate": 2.709260124655917e-05, "loss": 0.3849, "step": 6368 }, { "epoch": 1.11, "learning_rate": 2.7085754976021627e-05, "loss": 0.3533, "step": 6369 }, { "epoch": 1.11, "learning_rate": 2.7078908743849966e-05, "loss": 0.3711, "step": 6370 }, { "epoch": 1.11, "learning_rate": 2.7072062550678268e-05, "loss": 0.3678, "step": 6371 }, { "epoch": 1.11, "learning_rate": 2.7065216397140593e-05, "loss": 0.3651, "step": 6372 }, { "epoch": 1.11, "learning_rate": 2.7058370283871003e-05, "loss": 0.3591, "step": 6373 }, { "epoch": 1.11, "learning_rate": 2.705152421150357e-05, "loss": 0.3755, "step": 6374 }, { "epoch": 1.11, "learning_rate": 2.704467818067235e-05, "loss": 0.3651, "step": 6375 }, { "epoch": 1.11, "learning_rate": 2.7037832192011403e-05, "loss": 0.3641, "step": 6376 }, { "epoch": 1.11, "learning_rate": 2.7030986246154772e-05, "loss": 0.3584, "step": 6377 }, { "epoch": 1.11, "learning_rate": 2.7024140343736494e-05, "loss": 0.3828, "step": 6378 }, { "epoch": 1.11, "learning_rate": 2.701729448539063e-05, "loss": 0.3559, "step": 6379 }, { "epoch": 1.11, "learning_rate": 2.7010448671751208e-05, "loss": 0.3706, "step": 6380 }, { "epoch": 1.11, "learning_rate": 2.700360290345228e-05, "loss": 0.3705, "step": 6381 }, { "epoch": 1.11, "learning_rate": 2.6996757181127856e-05, "loss": 0.3481, "step": 6382 }, { "epoch": 1.11, "learning_rate": 2.698991150541197e-05, "loss": 0.3783, "step": 6383 }, { "epoch": 1.11, "learning_rate": 2.6983065876938635e-05, "loss": 0.3705, "step": 6384 }, { "epoch": 1.11, "learning_rate": 2.6976220296341885e-05, "loss": 0.3702, "step": 6385 }, { "epoch": 1.11, "learning_rate": 2.6969374764255723e-05, "loss": 0.3682, "step": 6386 }, { "epoch": 1.11, "learning_rate": 2.6962529281314157e-05, "loss": 0.3545, "step": 6387 }, { "epoch": 1.11, "learning_rate": 2.6955683848151204e-05, "loss": 0.3476, "step": 6388 }, { "epoch": 1.11, "learning_rate": 2.694883846540083e-05, "loss": 0.3723, "step": 6389 }, { "epoch": 1.11, "learning_rate": 2.6941993133697068e-05, "loss": 0.375, "step": 6390 }, { "epoch": 1.11, "learning_rate": 2.6935147853673893e-05, "loss": 0.3586, "step": 6391 }, { "epoch": 1.11, "learning_rate": 2.6928302625965286e-05, "loss": 0.3816, "step": 6392 }, { "epoch": 1.11, "learning_rate": 2.6921457451205234e-05, "loss": 0.363, "step": 6393 }, { "epoch": 1.11, "learning_rate": 2.6914612330027707e-05, "loss": 0.3782, "step": 6394 }, { "epoch": 1.11, "learning_rate": 2.6907767263066668e-05, "loss": 0.3698, "step": 6395 }, { "epoch": 1.12, "learning_rate": 2.69009222509561e-05, "loss": 0.3788, "step": 6396 }, { "epoch": 1.12, "learning_rate": 2.689407729432996e-05, "loss": 0.3635, "step": 6397 }, { "epoch": 1.12, "learning_rate": 2.6887232393822192e-05, "loss": 0.3525, "step": 6398 }, { "epoch": 1.12, "learning_rate": 2.6880387550066756e-05, "loss": 0.3683, "step": 6399 }, { "epoch": 1.12, "learning_rate": 2.6873542763697585e-05, "loss": 0.3821, "step": 6400 }, { "epoch": 1.12, "learning_rate": 2.6866698035348632e-05, "loss": 0.384, "step": 6401 }, { "epoch": 1.12, "learning_rate": 2.6859853365653832e-05, "loss": 0.3783, "step": 6402 }, { "epoch": 1.12, "learning_rate": 2.6853008755247106e-05, "loss": 0.3769, "step": 6403 }, { "epoch": 1.12, "learning_rate": 2.6846164204762374e-05, "loss": 0.3689, "step": 6404 }, { "epoch": 1.12, "learning_rate": 2.6839319714833565e-05, "loss": 0.3678, "step": 6405 }, { "epoch": 1.12, "learning_rate": 2.6832475286094566e-05, "loss": 0.3584, "step": 6406 }, { "epoch": 1.12, "learning_rate": 2.6825630919179322e-05, "loss": 0.3623, "step": 6407 }, { "epoch": 1.12, "learning_rate": 2.6818786614721707e-05, "loss": 0.3808, "step": 6408 }, { "epoch": 1.12, "learning_rate": 2.6811942373355624e-05, "loss": 0.3824, "step": 6409 }, { "epoch": 1.12, "learning_rate": 2.680509819571496e-05, "loss": 0.3761, "step": 6410 }, { "epoch": 1.12, "learning_rate": 2.6798254082433593e-05, "loss": 0.3662, "step": 6411 }, { "epoch": 1.12, "learning_rate": 2.6791410034145402e-05, "loss": 0.3565, "step": 6412 }, { "epoch": 1.12, "learning_rate": 2.6784566051484273e-05, "loss": 0.3754, "step": 6413 }, { "epoch": 1.12, "learning_rate": 2.6777722135084055e-05, "loss": 0.3544, "step": 6414 }, { "epoch": 1.12, "learning_rate": 2.6770878285578616e-05, "loss": 0.3537, "step": 6415 }, { "epoch": 1.12, "learning_rate": 2.6764034503601792e-05, "loss": 0.3785, "step": 6416 }, { "epoch": 1.12, "learning_rate": 2.6757190789787446e-05, "loss": 0.372, "step": 6417 }, { "epoch": 1.12, "learning_rate": 2.6750347144769415e-05, "loss": 0.3686, "step": 6418 }, { "epoch": 1.12, "learning_rate": 2.6743503569181533e-05, "loss": 0.3745, "step": 6419 }, { "epoch": 1.12, "learning_rate": 2.6736660063657624e-05, "loss": 0.3919, "step": 6420 }, { "epoch": 1.12, "learning_rate": 2.6729816628831517e-05, "loss": 0.3615, "step": 6421 }, { "epoch": 1.12, "learning_rate": 2.6722973265337002e-05, "loss": 0.3477, "step": 6422 }, { "epoch": 1.12, "learning_rate": 2.6716129973807924e-05, "loss": 0.3569, "step": 6423 }, { "epoch": 1.12, "learning_rate": 2.6709286754878053e-05, "loss": 0.3839, "step": 6424 }, { "epoch": 1.12, "learning_rate": 2.6702443609181198e-05, "loss": 0.3693, "step": 6425 }, { "epoch": 1.12, "learning_rate": 2.669560053735115e-05, "loss": 0.3707, "step": 6426 }, { "epoch": 1.12, "learning_rate": 2.6688757540021675e-05, "loss": 0.3746, "step": 6427 }, { "epoch": 1.12, "learning_rate": 2.6681914617826555e-05, "loss": 0.3433, "step": 6428 }, { "epoch": 1.12, "learning_rate": 2.6675071771399558e-05, "loss": 0.3692, "step": 6429 }, { "epoch": 1.12, "learning_rate": 2.666822900137444e-05, "loss": 0.3522, "step": 6430 }, { "epoch": 1.12, "learning_rate": 2.666138630838496e-05, "loss": 0.3702, "step": 6431 }, { "epoch": 1.12, "learning_rate": 2.665454369306486e-05, "loss": 0.3669, "step": 6432 }, { "epoch": 1.12, "learning_rate": 2.6647701156047865e-05, "loss": 0.3562, "step": 6433 }, { "epoch": 1.12, "learning_rate": 2.6640858697967735e-05, "loss": 0.3546, "step": 6434 }, { "epoch": 1.12, "learning_rate": 2.6634016319458178e-05, "loss": 0.3641, "step": 6435 }, { "epoch": 1.12, "learning_rate": 2.66271740211529e-05, "loss": 0.3731, "step": 6436 }, { "epoch": 1.12, "learning_rate": 2.6620331803685628e-05, "loss": 0.3666, "step": 6437 }, { "epoch": 1.12, "learning_rate": 2.6613489667690057e-05, "loss": 0.3701, "step": 6438 }, { "epoch": 1.12, "learning_rate": 2.6606647613799866e-05, "loss": 0.3717, "step": 6439 }, { "epoch": 1.12, "learning_rate": 2.6599805642648772e-05, "loss": 0.3539, "step": 6440 }, { "epoch": 1.12, "learning_rate": 2.6592963754870426e-05, "loss": 0.3689, "step": 6441 }, { "epoch": 1.12, "learning_rate": 2.6586121951098514e-05, "loss": 0.3706, "step": 6442 }, { "epoch": 1.12, "learning_rate": 2.65792802319667e-05, "loss": 0.3951, "step": 6443 }, { "epoch": 1.12, "learning_rate": 2.6572438598108623e-05, "loss": 0.3669, "step": 6444 }, { "epoch": 1.12, "learning_rate": 2.6565597050157954e-05, "loss": 0.367, "step": 6445 }, { "epoch": 1.12, "learning_rate": 2.6558755588748313e-05, "loss": 0.3635, "step": 6446 }, { "epoch": 1.12, "learning_rate": 2.6551914214513344e-05, "loss": 0.3583, "step": 6447 }, { "epoch": 1.12, "learning_rate": 2.6545072928086673e-05, "loss": 0.3658, "step": 6448 }, { "epoch": 1.12, "learning_rate": 2.6538231730101893e-05, "loss": 0.3706, "step": 6449 }, { "epoch": 1.12, "learning_rate": 2.6531390621192626e-05, "loss": 0.3888, "step": 6450 }, { "epoch": 1.12, "learning_rate": 2.6524549601992483e-05, "loss": 0.3682, "step": 6451 }, { "epoch": 1.12, "learning_rate": 2.6517708673135037e-05, "loss": 0.3668, "step": 6452 }, { "epoch": 1.13, "learning_rate": 2.6510867835253874e-05, "loss": 0.3761, "step": 6453 }, { "epoch": 1.13, "learning_rate": 2.650402708898257e-05, "loss": 0.37, "step": 6454 }, { "epoch": 1.13, "learning_rate": 2.6497186434954684e-05, "loss": 0.3652, "step": 6455 }, { "epoch": 1.13, "learning_rate": 2.6490345873803785e-05, "loss": 0.3924, "step": 6456 }, { "epoch": 1.13, "learning_rate": 2.6483505406163416e-05, "loss": 0.3621, "step": 6457 }, { "epoch": 1.13, "learning_rate": 2.6476665032667114e-05, "loss": 0.3829, "step": 6458 }, { "epoch": 1.13, "learning_rate": 2.6469824753948415e-05, "loss": 0.3675, "step": 6459 }, { "epoch": 1.13, "learning_rate": 2.646298457064083e-05, "loss": 0.3809, "step": 6460 }, { "epoch": 1.13, "learning_rate": 2.6456144483377877e-05, "loss": 0.3603, "step": 6461 }, { "epoch": 1.13, "learning_rate": 2.6449304492793064e-05, "loss": 0.3751, "step": 6462 }, { "epoch": 1.13, "learning_rate": 2.6442464599519882e-05, "loss": 0.3695, "step": 6463 }, { "epoch": 1.13, "learning_rate": 2.6435624804191834e-05, "loss": 0.3636, "step": 6464 }, { "epoch": 1.13, "learning_rate": 2.642878510744237e-05, "loss": 0.3685, "step": 6465 }, { "epoch": 1.13, "learning_rate": 2.6421945509904966e-05, "loss": 0.3672, "step": 6466 }, { "epoch": 1.13, "learning_rate": 2.6415106012213097e-05, "loss": 0.368, "step": 6467 }, { "epoch": 1.13, "learning_rate": 2.64082666150002e-05, "loss": 0.3682, "step": 6468 }, { "epoch": 1.13, "learning_rate": 2.6401427318899717e-05, "loss": 0.3535, "step": 6469 }, { "epoch": 1.13, "learning_rate": 2.639458812454508e-05, "loss": 0.377, "step": 6470 }, { "epoch": 1.13, "learning_rate": 2.6387749032569703e-05, "loss": 0.3733, "step": 6471 }, { "epoch": 1.13, "learning_rate": 2.6380910043607e-05, "loss": 0.3802, "step": 6472 }, { "epoch": 1.13, "learning_rate": 2.6374071158290386e-05, "loss": 0.3616, "step": 6473 }, { "epoch": 1.13, "learning_rate": 2.6367232377253247e-05, "loss": 0.3684, "step": 6474 }, { "epoch": 1.13, "learning_rate": 2.6360393701128968e-05, "loss": 0.391, "step": 6475 }, { "epoch": 1.13, "learning_rate": 2.6353555130550923e-05, "loss": 0.3748, "step": 6476 }, { "epoch": 1.13, "learning_rate": 2.634671666615246e-05, "loss": 0.3686, "step": 6477 }, { "epoch": 1.13, "learning_rate": 2.6339878308566945e-05, "loss": 0.3875, "step": 6478 }, { "epoch": 1.13, "learning_rate": 2.6333040058427726e-05, "loss": 0.3601, "step": 6479 }, { "epoch": 1.13, "learning_rate": 2.632620191636814e-05, "loss": 0.3584, "step": 6480 }, { "epoch": 1.13, "learning_rate": 2.6319363883021503e-05, "loss": 0.3684, "step": 6481 }, { "epoch": 1.13, "learning_rate": 2.631252595902113e-05, "loss": 0.362, "step": 6482 }, { "epoch": 1.13, "learning_rate": 2.6305688145000313e-05, "loss": 0.3696, "step": 6483 }, { "epoch": 1.13, "learning_rate": 2.629885044159237e-05, "loss": 0.3505, "step": 6484 }, { "epoch": 1.13, "learning_rate": 2.629201284943057e-05, "loss": 0.3602, "step": 6485 }, { "epoch": 1.13, "learning_rate": 2.6285175369148195e-05, "loss": 0.3878, "step": 6486 }, { "epoch": 1.13, "learning_rate": 2.6278338001378498e-05, "loss": 0.3912, "step": 6487 }, { "epoch": 1.13, "learning_rate": 2.6271500746754733e-05, "loss": 0.3588, "step": 6488 }, { "epoch": 1.13, "learning_rate": 2.626466360591014e-05, "loss": 0.3595, "step": 6489 }, { "epoch": 1.13, "learning_rate": 2.625782657947797e-05, "loss": 0.3629, "step": 6490 }, { "epoch": 1.13, "learning_rate": 2.6250989668091413e-05, "loss": 0.357, "step": 6491 }, { "epoch": 1.13, "learning_rate": 2.62441528723837e-05, "loss": 0.3722, "step": 6492 }, { "epoch": 1.13, "learning_rate": 2.6237316192988025e-05, "loss": 0.3562, "step": 6493 }, { "epoch": 1.13, "learning_rate": 2.623047963053757e-05, "loss": 0.3695, "step": 6494 }, { "epoch": 1.13, "learning_rate": 2.6223643185665524e-05, "loss": 0.3761, "step": 6495 }, { "epoch": 1.13, "learning_rate": 2.621680685900505e-05, "loss": 0.3753, "step": 6496 }, { "epoch": 1.13, "learning_rate": 2.6209970651189297e-05, "loss": 0.3699, "step": 6497 }, { "epoch": 1.13, "learning_rate": 2.6203134562851428e-05, "loss": 0.3618, "step": 6498 }, { "epoch": 1.13, "learning_rate": 2.619629859462456e-05, "loss": 0.3776, "step": 6499 }, { "epoch": 1.13, "learning_rate": 2.6189462747141806e-05, "loss": 0.3752, "step": 6500 }, { "epoch": 1.13, "learning_rate": 2.618262702103631e-05, "loss": 0.3536, "step": 6501 }, { "epoch": 1.13, "learning_rate": 2.6175791416941146e-05, "loss": 0.3838, "step": 6502 }, { "epoch": 1.13, "learning_rate": 2.616895593548942e-05, "loss": 0.3495, "step": 6503 }, { "epoch": 1.13, "learning_rate": 2.6162120577314194e-05, "loss": 0.387, "step": 6504 }, { "epoch": 1.13, "learning_rate": 2.6155285343048536e-05, "loss": 0.3672, "step": 6505 }, { "epoch": 1.13, "learning_rate": 2.614845023332551e-05, "loss": 0.3728, "step": 6506 }, { "epoch": 1.13, "learning_rate": 2.6141615248778162e-05, "loss": 0.3617, "step": 6507 }, { "epoch": 1.13, "learning_rate": 2.613478039003951e-05, "loss": 0.3741, "step": 6508 }, { "epoch": 1.13, "learning_rate": 2.6127945657742582e-05, "loss": 0.3664, "step": 6509 }, { "epoch": 1.14, "learning_rate": 2.6121111052520393e-05, "loss": 0.3692, "step": 6510 }, { "epoch": 1.14, "learning_rate": 2.6114276575005913e-05, "loss": 0.3674, "step": 6511 }, { "epoch": 1.14, "learning_rate": 2.6107442225832162e-05, "loss": 0.3785, "step": 6512 }, { "epoch": 1.14, "learning_rate": 2.6100608005632092e-05, "loss": 0.3733, "step": 6513 }, { "epoch": 1.14, "learning_rate": 2.6093773915038665e-05, "loss": 0.3642, "step": 6514 }, { "epoch": 1.14, "learning_rate": 2.608693995468484e-05, "loss": 0.3494, "step": 6515 }, { "epoch": 1.14, "learning_rate": 2.608010612520354e-05, "loss": 0.3542, "step": 6516 }, { "epoch": 1.14, "learning_rate": 2.60732724272277e-05, "loss": 0.3618, "step": 6517 }, { "epoch": 1.14, "learning_rate": 2.6066438861390226e-05, "loss": 0.3555, "step": 6518 }, { "epoch": 1.14, "learning_rate": 2.605960542832403e-05, "loss": 0.3538, "step": 6519 }, { "epoch": 1.14, "learning_rate": 2.6052772128661983e-05, "loss": 0.3706, "step": 6520 }, { "epoch": 1.14, "learning_rate": 2.604593896303697e-05, "loss": 0.3752, "step": 6521 }, { "epoch": 1.14, "learning_rate": 2.6039105932081848e-05, "loss": 0.3627, "step": 6522 }, { "epoch": 1.14, "learning_rate": 2.603227303642948e-05, "loss": 0.3774, "step": 6523 }, { "epoch": 1.14, "learning_rate": 2.6025440276712697e-05, "loss": 0.3431, "step": 6524 }, { "epoch": 1.14, "learning_rate": 2.601860765356432e-05, "loss": 0.3678, "step": 6525 }, { "epoch": 1.14, "learning_rate": 2.601177516761717e-05, "loss": 0.3794, "step": 6526 }, { "epoch": 1.14, "learning_rate": 2.6004942819504035e-05, "loss": 0.366, "step": 6527 }, { "epoch": 1.14, "learning_rate": 2.5998110609857723e-05, "loss": 0.3624, "step": 6528 }, { "epoch": 1.14, "learning_rate": 2.5991278539310993e-05, "loss": 0.3695, "step": 6529 }, { "epoch": 1.14, "learning_rate": 2.598444660849661e-05, "loss": 0.3557, "step": 6530 }, { "epoch": 1.14, "learning_rate": 2.5977614818047326e-05, "loss": 0.3709, "step": 6531 }, { "epoch": 1.14, "learning_rate": 2.5970783168595876e-05, "loss": 0.3634, "step": 6532 }, { "epoch": 1.14, "learning_rate": 2.596395166077497e-05, "loss": 0.3582, "step": 6533 }, { "epoch": 1.14, "learning_rate": 2.5957120295217336e-05, "loss": 0.3554, "step": 6534 }, { "epoch": 1.14, "learning_rate": 2.5950289072555654e-05, "loss": 0.3773, "step": 6535 }, { "epoch": 1.14, "learning_rate": 2.594345799342263e-05, "loss": 0.3682, "step": 6536 }, { "epoch": 1.14, "learning_rate": 2.593662705845092e-05, "loss": 0.3928, "step": 6537 }, { "epoch": 1.14, "learning_rate": 2.5929796268273166e-05, "loss": 0.3801, "step": 6538 }, { "epoch": 1.14, "learning_rate": 2.5922965623522033e-05, "loss": 0.3641, "step": 6539 }, { "epoch": 1.14, "learning_rate": 2.591613512483015e-05, "loss": 0.361, "step": 6540 }, { "epoch": 1.14, "learning_rate": 2.5909304772830124e-05, "loss": 0.3574, "step": 6541 }, { "epoch": 1.14, "learning_rate": 2.590247456815456e-05, "loss": 0.3691, "step": 6542 }, { "epoch": 1.14, "learning_rate": 2.589564451143605e-05, "loss": 0.3644, "step": 6543 }, { "epoch": 1.14, "learning_rate": 2.5888814603307155e-05, "loss": 0.3657, "step": 6544 }, { "epoch": 1.14, "learning_rate": 2.5881984844400457e-05, "loss": 0.3621, "step": 6545 }, { "epoch": 1.14, "learning_rate": 2.5875155235348497e-05, "loss": 0.365, "step": 6546 }, { "epoch": 1.14, "learning_rate": 2.586832577678381e-05, "loss": 0.3518, "step": 6547 }, { "epoch": 1.14, "learning_rate": 2.586149646933891e-05, "loss": 0.3668, "step": 6548 }, { "epoch": 1.14, "learning_rate": 2.5854667313646286e-05, "loss": 0.3779, "step": 6549 }, { "epoch": 1.14, "learning_rate": 2.5847838310338473e-05, "loss": 0.3641, "step": 6550 }, { "epoch": 1.14, "learning_rate": 2.5841009460047917e-05, "loss": 0.3785, "step": 6551 }, { "epoch": 1.14, "learning_rate": 2.583418076340709e-05, "loss": 0.3514, "step": 6552 }, { "epoch": 1.14, "learning_rate": 2.5827352221048445e-05, "loss": 0.3754, "step": 6553 }, { "epoch": 1.14, "learning_rate": 2.582052383360441e-05, "loss": 0.3754, "step": 6554 }, { "epoch": 1.14, "learning_rate": 2.58136956017074e-05, "loss": 0.3621, "step": 6555 }, { "epoch": 1.14, "learning_rate": 2.580686752598983e-05, "loss": 0.3717, "step": 6556 }, { "epoch": 1.14, "learning_rate": 2.5800039607084102e-05, "loss": 0.3651, "step": 6557 }, { "epoch": 1.14, "learning_rate": 2.5793211845622582e-05, "loss": 0.3636, "step": 6558 }, { "epoch": 1.14, "learning_rate": 2.5786384242237627e-05, "loss": 0.3666, "step": 6559 }, { "epoch": 1.14, "learning_rate": 2.5779556797561587e-05, "loss": 0.3527, "step": 6560 }, { "epoch": 1.14, "learning_rate": 2.577272951222681e-05, "loss": 0.364, "step": 6561 }, { "epoch": 1.14, "learning_rate": 2.5765902386865604e-05, "loss": 0.378, "step": 6562 }, { "epoch": 1.14, "learning_rate": 2.5759075422110267e-05, "loss": 0.3395, "step": 6563 }, { "epoch": 1.14, "learning_rate": 2.57522486185931e-05, "loss": 0.3828, "step": 6564 }, { "epoch": 1.14, "learning_rate": 2.574542197694637e-05, "loss": 0.3599, "step": 6565 }, { "epoch": 1.14, "learning_rate": 2.573859549780233e-05, "loss": 0.3736, "step": 6566 }, { "epoch": 1.14, "learning_rate": 2.5731769181793232e-05, "loss": 0.376, "step": 6567 }, { "epoch": 1.15, "learning_rate": 2.572494302955131e-05, "loss": 0.3575, "step": 6568 }, { "epoch": 1.15, "learning_rate": 2.5718117041708766e-05, "loss": 0.3637, "step": 6569 }, { "epoch": 1.15, "learning_rate": 2.5711291218897804e-05, "loss": 0.3634, "step": 6570 }, { "epoch": 1.15, "learning_rate": 2.5704465561750606e-05, "loss": 0.3532, "step": 6571 }, { "epoch": 1.15, "learning_rate": 2.5697640070899343e-05, "loss": 0.3758, "step": 6572 }, { "epoch": 1.15, "learning_rate": 2.569081474697616e-05, "loss": 0.3574, "step": 6573 }, { "epoch": 1.15, "learning_rate": 2.5683989590613204e-05, "loss": 0.3583, "step": 6574 }, { "epoch": 1.15, "learning_rate": 2.5677164602442592e-05, "loss": 0.3631, "step": 6575 }, { "epoch": 1.15, "learning_rate": 2.5670339783096424e-05, "loss": 0.3652, "step": 6576 }, { "epoch": 1.15, "learning_rate": 2.56635151332068e-05, "loss": 0.3922, "step": 6577 }, { "epoch": 1.15, "learning_rate": 2.5656690653405787e-05, "loss": 0.3669, "step": 6578 }, { "epoch": 1.15, "learning_rate": 2.5649866344325453e-05, "loss": 0.3691, "step": 6579 }, { "epoch": 1.15, "learning_rate": 2.5643042206597824e-05, "loss": 0.3522, "step": 6580 }, { "epoch": 1.15, "learning_rate": 2.5636218240854943e-05, "loss": 0.3597, "step": 6581 }, { "epoch": 1.15, "learning_rate": 2.5629394447728815e-05, "loss": 0.379, "step": 6582 }, { "epoch": 1.15, "learning_rate": 2.5622570827851434e-05, "loss": 0.3717, "step": 6583 }, { "epoch": 1.15, "learning_rate": 2.5615747381854783e-05, "loss": 0.3785, "step": 6584 }, { "epoch": 1.15, "learning_rate": 2.560892411037082e-05, "loss": 0.3765, "step": 6585 }, { "epoch": 1.15, "learning_rate": 2.56021010140315e-05, "loss": 0.3689, "step": 6586 }, { "epoch": 1.15, "learning_rate": 2.559527809346874e-05, "loss": 0.3724, "step": 6587 }, { "epoch": 1.15, "learning_rate": 2.558845534931446e-05, "loss": 0.3678, "step": 6588 }, { "epoch": 1.15, "learning_rate": 2.558163278220057e-05, "loss": 0.3711, "step": 6589 }, { "epoch": 1.15, "learning_rate": 2.5574810392758937e-05, "loss": 0.3702, "step": 6590 }, { "epoch": 1.15, "learning_rate": 2.556798818162144e-05, "loss": 0.3645, "step": 6591 }, { "epoch": 1.15, "learning_rate": 2.5561166149419907e-05, "loss": 0.3918, "step": 6592 }, { "epoch": 1.15, "learning_rate": 2.5554344296786186e-05, "loss": 0.3611, "step": 6593 }, { "epoch": 1.15, "learning_rate": 2.5547522624352088e-05, "loss": 0.3581, "step": 6594 }, { "epoch": 1.15, "learning_rate": 2.5540701132749416e-05, "loss": 0.362, "step": 6595 }, { "epoch": 1.15, "learning_rate": 2.5533879822609947e-05, "loss": 0.3568, "step": 6596 }, { "epoch": 1.15, "learning_rate": 2.552705869456544e-05, "loss": 0.3567, "step": 6597 }, { "epoch": 1.15, "learning_rate": 2.5520237749247667e-05, "loss": 0.3707, "step": 6598 }, { "epoch": 1.15, "learning_rate": 2.5513416987288318e-05, "loss": 0.3616, "step": 6599 }, { "epoch": 1.15, "learning_rate": 2.550659640931915e-05, "loss": 0.3771, "step": 6600 }, { "epoch": 1.15, "learning_rate": 2.5499776015971837e-05, "loss": 0.3687, "step": 6601 }, { "epoch": 1.15, "learning_rate": 2.549295580787807e-05, "loss": 0.3553, "step": 6602 }, { "epoch": 1.15, "learning_rate": 2.548613578566951e-05, "loss": 0.3683, "step": 6603 }, { "epoch": 1.15, "learning_rate": 2.5479315949977794e-05, "loss": 0.3611, "step": 6604 }, { "epoch": 1.15, "learning_rate": 2.547249630143455e-05, "loss": 0.3624, "step": 6605 }, { "epoch": 1.15, "learning_rate": 2.5465676840671398e-05, "loss": 0.3609, "step": 6606 }, { "epoch": 1.15, "learning_rate": 2.5458857568319934e-05, "loss": 0.3687, "step": 6607 }, { "epoch": 1.15, "learning_rate": 2.5452038485011737e-05, "loss": 0.3665, "step": 6608 }, { "epoch": 1.15, "learning_rate": 2.544521959137835e-05, "loss": 0.3512, "step": 6609 }, { "epoch": 1.15, "learning_rate": 2.543840088805132e-05, "loss": 0.3556, "step": 6610 }, { "epoch": 1.15, "learning_rate": 2.543158237566218e-05, "loss": 0.3588, "step": 6611 }, { "epoch": 1.15, "learning_rate": 2.5424764054842438e-05, "loss": 0.3516, "step": 6612 }, { "epoch": 1.15, "learning_rate": 2.5417945926223564e-05, "loss": 0.3799, "step": 6613 }, { "epoch": 1.15, "learning_rate": 2.541112799043704e-05, "loss": 0.3666, "step": 6614 }, { "epoch": 1.15, "learning_rate": 2.540431024811432e-05, "loss": 0.3633, "step": 6615 }, { "epoch": 1.15, "learning_rate": 2.539749269988683e-05, "loss": 0.3664, "step": 6616 }, { "epoch": 1.15, "learning_rate": 2.5390675346386005e-05, "loss": 0.3745, "step": 6617 }, { "epoch": 1.15, "learning_rate": 2.5383858188243224e-05, "loss": 0.3693, "step": 6618 }, { "epoch": 1.15, "learning_rate": 2.5377041226089882e-05, "loss": 0.3753, "step": 6619 }, { "epoch": 1.15, "learning_rate": 2.5370224460557335e-05, "loss": 0.3542, "step": 6620 }, { "epoch": 1.15, "learning_rate": 2.5363407892276915e-05, "loss": 0.3802, "step": 6621 }, { "epoch": 1.15, "learning_rate": 2.535659152187997e-05, "loss": 0.3673, "step": 6622 }, { "epoch": 1.15, "learning_rate": 2.5349775349997796e-05, "loss": 0.371, "step": 6623 }, { "epoch": 1.15, "learning_rate": 2.5342959377261687e-05, "loss": 0.3704, "step": 6624 }, { "epoch": 1.16, "learning_rate": 2.5336143604302907e-05, "loss": 0.3707, "step": 6625 }, { "epoch": 1.16, "learning_rate": 2.5329328031752712e-05, "loss": 0.3609, "step": 6626 }, { "epoch": 1.16, "learning_rate": 2.532251266024234e-05, "loss": 0.3472, "step": 6627 }, { "epoch": 1.16, "learning_rate": 2.5315697490403e-05, "loss": 0.3659, "step": 6628 }, { "epoch": 1.16, "learning_rate": 2.5308882522865895e-05, "loss": 0.3723, "step": 6629 }, { "epoch": 1.16, "learning_rate": 2.53020677582622e-05, "loss": 0.3584, "step": 6630 }, { "epoch": 1.16, "learning_rate": 2.5295253197223066e-05, "loss": 0.3503, "step": 6631 }, { "epoch": 1.16, "learning_rate": 2.5288438840379636e-05, "loss": 0.3745, "step": 6632 }, { "epoch": 1.16, "learning_rate": 2.5281624688363047e-05, "loss": 0.3729, "step": 6633 }, { "epoch": 1.16, "learning_rate": 2.5274810741804382e-05, "loss": 0.3579, "step": 6634 }, { "epoch": 1.16, "learning_rate": 2.5267997001334735e-05, "loss": 0.3878, "step": 6635 }, { "epoch": 1.16, "learning_rate": 2.5261183467585163e-05, "loss": 0.3578, "step": 6636 }, { "epoch": 1.16, "learning_rate": 2.5254370141186716e-05, "loss": 0.3746, "step": 6637 }, { "epoch": 1.16, "learning_rate": 2.524755702277042e-05, "loss": 0.3599, "step": 6638 }, { "epoch": 1.16, "learning_rate": 2.5240744112967274e-05, "loss": 0.3766, "step": 6639 }, { "epoch": 1.16, "learning_rate": 2.5233931412408274e-05, "loss": 0.3817, "step": 6640 }, { "epoch": 1.16, "learning_rate": 2.5227118921724384e-05, "loss": 0.3825, "step": 6641 }, { "epoch": 1.16, "learning_rate": 2.522030664154656e-05, "loss": 0.3786, "step": 6642 }, { "epoch": 1.16, "learning_rate": 2.521349457250571e-05, "loss": 0.3611, "step": 6643 }, { "epoch": 1.16, "learning_rate": 2.5206682715232765e-05, "loss": 0.3638, "step": 6644 }, { "epoch": 1.16, "learning_rate": 2.5199871070358605e-05, "loss": 0.3701, "step": 6645 }, { "epoch": 1.16, "learning_rate": 2.519305963851411e-05, "loss": 0.3621, "step": 6646 }, { "epoch": 1.16, "learning_rate": 2.518624842033011e-05, "loss": 0.3847, "step": 6647 }, { "epoch": 1.16, "learning_rate": 2.5179437416437456e-05, "loss": 0.3771, "step": 6648 }, { "epoch": 1.16, "learning_rate": 2.5172626627466938e-05, "loss": 0.3606, "step": 6649 }, { "epoch": 1.16, "learning_rate": 2.5165816054049375e-05, "loss": 0.3735, "step": 6650 }, { "epoch": 1.16, "learning_rate": 2.5159005696815514e-05, "loss": 0.353, "step": 6651 }, { "epoch": 1.16, "learning_rate": 2.515219555639611e-05, "loss": 0.3749, "step": 6652 }, { "epoch": 1.16, "learning_rate": 2.5145385633421908e-05, "loss": 0.3795, "step": 6653 }, { "epoch": 1.16, "learning_rate": 2.5138575928523594e-05, "loss": 0.3661, "step": 6654 }, { "epoch": 1.16, "learning_rate": 2.5131766442331883e-05, "loss": 0.3465, "step": 6655 }, { "epoch": 1.16, "learning_rate": 2.512495717547743e-05, "loss": 0.3736, "step": 6656 }, { "epoch": 1.16, "learning_rate": 2.511814812859089e-05, "loss": 0.3663, "step": 6657 }, { "epoch": 1.16, "learning_rate": 2.5111339302302896e-05, "loss": 0.371, "step": 6658 }, { "epoch": 1.16, "learning_rate": 2.5104530697244043e-05, "loss": 0.3619, "step": 6659 }, { "epoch": 1.16, "learning_rate": 2.509772231404493e-05, "loss": 0.388, "step": 6660 }, { "epoch": 1.16, "learning_rate": 2.509091415333612e-05, "loss": 0.3712, "step": 6661 }, { "epoch": 1.16, "learning_rate": 2.5084106215748178e-05, "loss": 0.3785, "step": 6662 }, { "epoch": 1.16, "learning_rate": 2.5077298501911605e-05, "loss": 0.3505, "step": 6663 }, { "epoch": 1.16, "learning_rate": 2.5070491012456926e-05, "loss": 0.3702, "step": 6664 }, { "epoch": 1.16, "learning_rate": 2.506368374801461e-05, "loss": 0.3585, "step": 6665 }, { "epoch": 1.16, "learning_rate": 2.5056876709215132e-05, "loss": 0.3545, "step": 6666 }, { "epoch": 1.16, "learning_rate": 2.5050069896688944e-05, "loss": 0.3724, "step": 6667 }, { "epoch": 1.16, "learning_rate": 2.504326331106645e-05, "loss": 0.3821, "step": 6668 }, { "epoch": 1.16, "learning_rate": 2.503645695297806e-05, "loss": 0.3697, "step": 6669 }, { "epoch": 1.16, "learning_rate": 2.5029650823054158e-05, "loss": 0.3599, "step": 6670 }, { "epoch": 1.16, "learning_rate": 2.502284492192508e-05, "loss": 0.3894, "step": 6671 }, { "epoch": 1.16, "learning_rate": 2.5016039250221202e-05, "loss": 0.3737, "step": 6672 }, { "epoch": 1.16, "learning_rate": 2.5009233808572818e-05, "loss": 0.3704, "step": 6673 }, { "epoch": 1.16, "learning_rate": 2.5002428597610235e-05, "loss": 0.3717, "step": 6674 }, { "epoch": 1.16, "learning_rate": 2.499562361796371e-05, "loss": 0.3772, "step": 6675 }, { "epoch": 1.16, "learning_rate": 2.4988818870263496e-05, "loss": 0.3628, "step": 6676 }, { "epoch": 1.16, "learning_rate": 2.4982014355139848e-05, "loss": 0.3813, "step": 6677 }, { "epoch": 1.16, "learning_rate": 2.4975210073222954e-05, "loss": 0.3921, "step": 6678 }, { "epoch": 1.16, "learning_rate": 2.496840602514302e-05, "loss": 0.361, "step": 6679 }, { "epoch": 1.16, "learning_rate": 2.496160221153019e-05, "loss": 0.3515, "step": 6680 }, { "epoch": 1.16, "learning_rate": 2.4954798633014625e-05, "loss": 0.3634, "step": 6681 }, { "epoch": 1.16, "learning_rate": 2.4947995290226433e-05, "loss": 0.363, "step": 6682 }, { "epoch": 1.17, "learning_rate": 2.4941192183795737e-05, "loss": 0.3887, "step": 6683 }, { "epoch": 1.17, "learning_rate": 2.493438931435261e-05, "loss": 0.3834, "step": 6684 }, { "epoch": 1.17, "learning_rate": 2.4927586682527094e-05, "loss": 0.3772, "step": 6685 }, { "epoch": 1.17, "learning_rate": 2.4920784288949233e-05, "loss": 0.3638, "step": 6686 }, { "epoch": 1.17, "learning_rate": 2.4913982134249047e-05, "loss": 0.372, "step": 6687 }, { "epoch": 1.17, "learning_rate": 2.4907180219056513e-05, "loss": 0.3586, "step": 6688 }, { "epoch": 1.17, "learning_rate": 2.490037854400161e-05, "loss": 0.3642, "step": 6689 }, { "epoch": 1.17, "learning_rate": 2.489357710971429e-05, "loss": 0.3701, "step": 6690 }, { "epoch": 1.17, "learning_rate": 2.4886775916824466e-05, "loss": 0.3671, "step": 6691 }, { "epoch": 1.17, "learning_rate": 2.487997496596205e-05, "loss": 0.3782, "step": 6692 }, { "epoch": 1.17, "learning_rate": 2.4873174257756903e-05, "loss": 0.3728, "step": 6693 }, { "epoch": 1.17, "learning_rate": 2.4866373792838904e-05, "loss": 0.3548, "step": 6694 }, { "epoch": 1.17, "learning_rate": 2.4859573571837873e-05, "loss": 0.364, "step": 6695 }, { "epoch": 1.17, "learning_rate": 2.485277359538364e-05, "loss": 0.3637, "step": 6696 }, { "epoch": 1.17, "learning_rate": 2.484597386410597e-05, "loss": 0.3629, "step": 6697 }, { "epoch": 1.17, "learning_rate": 2.4839174378634644e-05, "loss": 0.3741, "step": 6698 }, { "epoch": 1.17, "learning_rate": 2.48323751395994e-05, "loss": 0.3847, "step": 6699 }, { "epoch": 1.17, "learning_rate": 2.482557614762997e-05, "loss": 0.3607, "step": 6700 }, { "epoch": 1.17, "learning_rate": 2.4818777403356036e-05, "loss": 0.362, "step": 6701 }, { "epoch": 1.17, "learning_rate": 2.4811978907407286e-05, "loss": 0.3512, "step": 6702 }, { "epoch": 1.17, "learning_rate": 2.4805180660413374e-05, "loss": 0.379, "step": 6703 }, { "epoch": 1.17, "learning_rate": 2.4798382663003906e-05, "loss": 0.3576, "step": 6704 }, { "epoch": 1.17, "learning_rate": 2.479158491580852e-05, "loss": 0.3716, "step": 6705 }, { "epoch": 1.17, "learning_rate": 2.4784787419456778e-05, "loss": 0.3765, "step": 6706 }, { "epoch": 1.17, "learning_rate": 2.4777990174578246e-05, "loss": 0.3762, "step": 6707 }, { "epoch": 1.17, "learning_rate": 2.4771193181802464e-05, "loss": 0.3684, "step": 6708 }, { "epoch": 1.17, "learning_rate": 2.4764396441758936e-05, "loss": 0.3747, "step": 6709 }, { "epoch": 1.17, "learning_rate": 2.4757599955077155e-05, "loss": 0.3849, "step": 6710 }, { "epoch": 1.17, "learning_rate": 2.475080372238659e-05, "loss": 0.3698, "step": 6711 }, { "epoch": 1.17, "learning_rate": 2.4744007744316682e-05, "loss": 0.3621, "step": 6712 }, { "epoch": 1.17, "learning_rate": 2.473721202149686e-05, "loss": 0.3654, "step": 6713 }, { "epoch": 1.17, "learning_rate": 2.4730416554556496e-05, "loss": 0.3687, "step": 6714 }, { "epoch": 1.17, "learning_rate": 2.4723621344124976e-05, "loss": 0.3643, "step": 6715 }, { "epoch": 1.17, "learning_rate": 2.471682639083165e-05, "loss": 0.3701, "step": 6716 }, { "epoch": 1.17, "learning_rate": 2.4710031695305847e-05, "loss": 0.3696, "step": 6717 }, { "epoch": 1.17, "learning_rate": 2.4703237258176847e-05, "loss": 0.3717, "step": 6718 }, { "epoch": 1.17, "learning_rate": 2.4696443080073945e-05, "loss": 0.3609, "step": 6719 }, { "epoch": 1.17, "learning_rate": 2.4689649161626387e-05, "loss": 0.3569, "step": 6720 }, { "epoch": 1.17, "learning_rate": 2.468285550346339e-05, "loss": 0.3714, "step": 6721 }, { "epoch": 1.17, "learning_rate": 2.4676062106214186e-05, "loss": 0.3852, "step": 6722 }, { "epoch": 1.17, "learning_rate": 2.4669268970507932e-05, "loss": 0.3707, "step": 6723 }, { "epoch": 1.17, "learning_rate": 2.466247609697379e-05, "loss": 0.3579, "step": 6724 }, { "epoch": 1.17, "learning_rate": 2.46556834862409e-05, "loss": 0.3673, "step": 6725 }, { "epoch": 1.17, "learning_rate": 2.4648891138938346e-05, "loss": 0.3629, "step": 6726 }, { "epoch": 1.17, "learning_rate": 2.4642099055695228e-05, "loss": 0.3658, "step": 6727 }, { "epoch": 1.17, "learning_rate": 2.463530723714061e-05, "loss": 0.3766, "step": 6728 }, { "epoch": 1.17, "learning_rate": 2.4628515683903513e-05, "loss": 0.3621, "step": 6729 }, { "epoch": 1.17, "learning_rate": 2.4621724396612956e-05, "loss": 0.3627, "step": 6730 }, { "epoch": 1.17, "learning_rate": 2.4614933375897915e-05, "loss": 0.3862, "step": 6731 }, { "epoch": 1.17, "learning_rate": 2.4608142622387342e-05, "loss": 0.3605, "step": 6732 }, { "epoch": 1.17, "learning_rate": 2.4601352136710193e-05, "loss": 0.3694, "step": 6733 }, { "epoch": 1.17, "learning_rate": 2.4594561919495372e-05, "loss": 0.3478, "step": 6734 }, { "epoch": 1.17, "learning_rate": 2.458777197137176e-05, "loss": 0.3612, "step": 6735 }, { "epoch": 1.17, "learning_rate": 2.4580982292968213e-05, "loss": 0.3718, "step": 6736 }, { "epoch": 1.17, "learning_rate": 2.4574192884913573e-05, "loss": 0.3842, "step": 6737 }, { "epoch": 1.17, "learning_rate": 2.4567403747836647e-05, "loss": 0.3696, "step": 6738 }, { "epoch": 1.17, "learning_rate": 2.4560614882366234e-05, "loss": 0.3751, "step": 6739 }, { "epoch": 1.18, "learning_rate": 2.4553826289131078e-05, "loss": 0.3878, "step": 6740 }, { "epoch": 1.18, "learning_rate": 2.4547037968759922e-05, "loss": 0.3585, "step": 6741 }, { "epoch": 1.18, "learning_rate": 2.454024992188147e-05, "loss": 0.3615, "step": 6742 }, { "epoch": 1.18, "learning_rate": 2.453346214912441e-05, "loss": 0.3491, "step": 6743 }, { "epoch": 1.18, "learning_rate": 2.4526674651117397e-05, "loss": 0.3729, "step": 6744 }, { "epoch": 1.18, "learning_rate": 2.4519887428489075e-05, "loss": 0.3767, "step": 6745 }, { "epoch": 1.18, "learning_rate": 2.4513100481868046e-05, "loss": 0.3558, "step": 6746 }, { "epoch": 1.18, "learning_rate": 2.4506313811882888e-05, "loss": 0.3798, "step": 6747 }, { "epoch": 1.18, "learning_rate": 2.4499527419162154e-05, "loss": 0.3795, "step": 6748 }, { "epoch": 1.18, "learning_rate": 2.4492741304334393e-05, "loss": 0.3456, "step": 6749 }, { "epoch": 1.18, "learning_rate": 2.4485955468028098e-05, "loss": 0.3639, "step": 6750 }, { "epoch": 1.18, "learning_rate": 2.4479169910871763e-05, "loss": 0.3518, "step": 6751 }, { "epoch": 1.18, "learning_rate": 2.4472384633493816e-05, "loss": 0.377, "step": 6752 }, { "epoch": 1.18, "learning_rate": 2.4465599636522707e-05, "loss": 0.3346, "step": 6753 }, { "epoch": 1.18, "learning_rate": 2.445881492058682e-05, "loss": 0.3795, "step": 6754 }, { "epoch": 1.18, "learning_rate": 2.4452030486314552e-05, "loss": 0.3725, "step": 6755 }, { "epoch": 1.18, "learning_rate": 2.4445246334334238e-05, "loss": 0.3719, "step": 6756 }, { "epoch": 1.18, "learning_rate": 2.4438462465274208e-05, "loss": 0.3559, "step": 6757 }, { "epoch": 1.18, "learning_rate": 2.443167887976276e-05, "loss": 0.3631, "step": 6758 }, { "epoch": 1.18, "learning_rate": 2.4424895578428148e-05, "loss": 0.3756, "step": 6759 }, { "epoch": 1.18, "learning_rate": 2.4418112561898647e-05, "loss": 0.361, "step": 6760 }, { "epoch": 1.18, "learning_rate": 2.4411329830802455e-05, "loss": 0.4037, "step": 6761 }, { "epoch": 1.18, "learning_rate": 2.4404547385767768e-05, "loss": 0.364, "step": 6762 }, { "epoch": 1.18, "learning_rate": 2.439776522742276e-05, "loss": 0.3639, "step": 6763 }, { "epoch": 1.18, "learning_rate": 2.439098335639555e-05, "loss": 0.3712, "step": 6764 }, { "epoch": 1.18, "learning_rate": 2.438420177331427e-05, "loss": 0.3671, "step": 6765 }, { "epoch": 1.18, "learning_rate": 2.4377420478806996e-05, "loss": 0.3858, "step": 6766 }, { "epoch": 1.18, "learning_rate": 2.43706394735018e-05, "loss": 0.3744, "step": 6767 }, { "epoch": 1.18, "learning_rate": 2.43638587580267e-05, "loss": 0.3831, "step": 6768 }, { "epoch": 1.18, "learning_rate": 2.4357078333009696e-05, "loss": 0.3637, "step": 6769 }, { "epoch": 1.18, "learning_rate": 2.4350298199078785e-05, "loss": 0.3897, "step": 6770 }, { "epoch": 1.18, "learning_rate": 2.4343518356861903e-05, "loss": 0.3685, "step": 6771 }, { "epoch": 1.18, "learning_rate": 2.4336738806986996e-05, "loss": 0.371, "step": 6772 }, { "epoch": 1.18, "learning_rate": 2.4329959550081936e-05, "loss": 0.3668, "step": 6773 }, { "epoch": 1.18, "learning_rate": 2.4323180586774613e-05, "loss": 0.3643, "step": 6774 }, { "epoch": 1.18, "learning_rate": 2.4316401917692862e-05, "loss": 0.3613, "step": 6775 }, { "epoch": 1.18, "learning_rate": 2.430962354346448e-05, "loss": 0.3736, "step": 6776 }, { "epoch": 1.18, "learning_rate": 2.4302845464717295e-05, "loss": 0.3794, "step": 6777 }, { "epoch": 1.18, "learning_rate": 2.4296067682079036e-05, "loss": 0.3701, "step": 6778 }, { "epoch": 1.18, "learning_rate": 2.428929019617745e-05, "loss": 0.3635, "step": 6779 }, { "epoch": 1.18, "learning_rate": 2.428251300764025e-05, "loss": 0.3654, "step": 6780 }, { "epoch": 1.18, "learning_rate": 2.42757361170951e-05, "loss": 0.3605, "step": 6781 }, { "epoch": 1.18, "learning_rate": 2.4268959525169655e-05, "loss": 0.3673, "step": 6782 }, { "epoch": 1.18, "learning_rate": 2.4262183232491543e-05, "loss": 0.3747, "step": 6783 }, { "epoch": 1.18, "learning_rate": 2.425540723968836e-05, "loss": 0.3483, "step": 6784 }, { "epoch": 1.18, "learning_rate": 2.4248631547387675e-05, "loss": 0.3681, "step": 6785 }, { "epoch": 1.18, "learning_rate": 2.4241856156217018e-05, "loss": 0.3608, "step": 6786 }, { "epoch": 1.18, "learning_rate": 2.4235081066803905e-05, "loss": 0.3699, "step": 6787 }, { "epoch": 1.18, "learning_rate": 2.4228306279775832e-05, "loss": 0.3608, "step": 6788 }, { "epoch": 1.18, "learning_rate": 2.4221531795760244e-05, "loss": 0.3664, "step": 6789 }, { "epoch": 1.18, "learning_rate": 2.4214757615384574e-05, "loss": 0.3625, "step": 6790 }, { "epoch": 1.18, "learning_rate": 2.4207983739276224e-05, "loss": 0.3696, "step": 6791 }, { "epoch": 1.18, "learning_rate": 2.420121016806256e-05, "loss": 0.3663, "step": 6792 }, { "epoch": 1.18, "learning_rate": 2.4194436902370926e-05, "loss": 0.372, "step": 6793 }, { "epoch": 1.18, "learning_rate": 2.418766394282864e-05, "loss": 0.3685, "step": 6794 }, { "epoch": 1.18, "learning_rate": 2.4180891290062997e-05, "loss": 0.3753, "step": 6795 }, { "epoch": 1.18, "learning_rate": 2.4174118944701242e-05, "loss": 0.3551, "step": 6796 }, { "epoch": 1.19, "learning_rate": 2.4167346907370615e-05, "loss": 0.3646, "step": 6797 }, { "epoch": 1.19, "learning_rate": 2.4160575178698302e-05, "loss": 0.363, "step": 6798 }, { "epoch": 1.19, "learning_rate": 2.4153803759311495e-05, "loss": 0.3612, "step": 6799 }, { "epoch": 1.19, "learning_rate": 2.414703264983733e-05, "loss": 0.3668, "step": 6800 }, { "epoch": 1.19, "learning_rate": 2.4140261850902936e-05, "loss": 0.3676, "step": 6801 }, { "epoch": 1.19, "learning_rate": 2.4133491363135375e-05, "loss": 0.3602, "step": 6802 }, { "epoch": 1.19, "learning_rate": 2.4126721187161723e-05, "loss": 0.3498, "step": 6803 }, { "epoch": 1.19, "learning_rate": 2.4119951323608997e-05, "loss": 0.3758, "step": 6804 }, { "epoch": 1.19, "learning_rate": 2.411318177310421e-05, "loss": 0.3664, "step": 6805 }, { "epoch": 1.19, "learning_rate": 2.410641253627433e-05, "loss": 0.3889, "step": 6806 }, { "epoch": 1.19, "learning_rate": 2.4099643613746304e-05, "loss": 0.3783, "step": 6807 }, { "epoch": 1.19, "learning_rate": 2.4092875006147033e-05, "loss": 0.3627, "step": 6808 }, { "epoch": 1.19, "learning_rate": 2.4086106714103397e-05, "loss": 0.3776, "step": 6809 }, { "epoch": 1.19, "learning_rate": 2.4079338738242274e-05, "loss": 0.3802, "step": 6810 }, { "epoch": 1.19, "learning_rate": 2.4072571079190477e-05, "loss": 0.361, "step": 6811 }, { "epoch": 1.19, "learning_rate": 2.4065803737574794e-05, "loss": 0.3772, "step": 6812 }, { "epoch": 1.19, "learning_rate": 2.4059036714022008e-05, "loss": 0.3752, "step": 6813 }, { "epoch": 1.19, "learning_rate": 2.405227000915884e-05, "loss": 0.3565, "step": 6814 }, { "epoch": 1.19, "learning_rate": 2.4045503623612002e-05, "loss": 0.3602, "step": 6815 }, { "epoch": 1.19, "learning_rate": 2.4038737558008184e-05, "loss": 0.3537, "step": 6816 }, { "epoch": 1.19, "learning_rate": 2.4031971812974024e-05, "loss": 0.3748, "step": 6817 }, { "epoch": 1.19, "learning_rate": 2.402520638913615e-05, "loss": 0.3828, "step": 6818 }, { "epoch": 1.19, "learning_rate": 2.4018441287121136e-05, "loss": 0.3623, "step": 6819 }, { "epoch": 1.19, "learning_rate": 2.4011676507555546e-05, "loss": 0.3653, "step": 6820 }, { "epoch": 1.19, "learning_rate": 2.400491205106592e-05, "loss": 0.3524, "step": 6821 }, { "epoch": 1.19, "learning_rate": 2.3998147918278757e-05, "loss": 0.3538, "step": 6822 }, { "epoch": 1.19, "learning_rate": 2.3991384109820513e-05, "loss": 0.3597, "step": 6823 }, { "epoch": 1.19, "learning_rate": 2.3984620626317637e-05, "loss": 0.3651, "step": 6824 }, { "epoch": 1.19, "learning_rate": 2.3977857468396547e-05, "loss": 0.3721, "step": 6825 }, { "epoch": 1.19, "learning_rate": 2.397109463668359e-05, "loss": 0.363, "step": 6826 }, { "epoch": 1.19, "learning_rate": 2.3964332131805155e-05, "loss": 0.3536, "step": 6827 }, { "epoch": 1.19, "learning_rate": 2.3957569954387537e-05, "loss": 0.3705, "step": 6828 }, { "epoch": 1.19, "learning_rate": 2.3950808105057032e-05, "loss": 0.3859, "step": 6829 }, { "epoch": 1.19, "learning_rate": 2.3944046584439898e-05, "loss": 0.3712, "step": 6830 }, { "epoch": 1.19, "learning_rate": 2.3937285393162358e-05, "loss": 0.3696, "step": 6831 }, { "epoch": 1.19, "learning_rate": 2.393052453185061e-05, "loss": 0.3581, "step": 6832 }, { "epoch": 1.19, "learning_rate": 2.392376400113082e-05, "loss": 0.3621, "step": 6833 }, { "epoch": 1.19, "learning_rate": 2.3917003801629145e-05, "loss": 0.3628, "step": 6834 }, { "epoch": 1.19, "learning_rate": 2.391024393397166e-05, "loss": 0.3704, "step": 6835 }, { "epoch": 1.19, "learning_rate": 2.3903484398784455e-05, "loss": 0.3656, "step": 6836 }, { "epoch": 1.19, "learning_rate": 2.3896725196693566e-05, "loss": 0.3658, "step": 6837 }, { "epoch": 1.19, "learning_rate": 2.3889966328325007e-05, "loss": 0.361, "step": 6838 }, { "epoch": 1.19, "learning_rate": 2.3883207794304778e-05, "loss": 0.371, "step": 6839 }, { "epoch": 1.19, "learning_rate": 2.387644959525881e-05, "loss": 0.3757, "step": 6840 }, { "epoch": 1.19, "learning_rate": 2.3869691731813032e-05, "loss": 0.35, "step": 6841 }, { "epoch": 1.19, "learning_rate": 2.386293420459332e-05, "loss": 0.3694, "step": 6842 }, { "epoch": 1.19, "learning_rate": 2.385617701422555e-05, "loss": 0.3621, "step": 6843 }, { "epoch": 1.19, "learning_rate": 2.3849420161335538e-05, "loss": 0.357, "step": 6844 }, { "epoch": 1.19, "learning_rate": 2.3842663646549085e-05, "loss": 0.3654, "step": 6845 }, { "epoch": 1.19, "learning_rate": 2.3835907470491948e-05, "loss": 0.3648, "step": 6846 }, { "epoch": 1.19, "learning_rate": 2.382915163378987e-05, "loss": 0.3688, "step": 6847 }, { "epoch": 1.19, "learning_rate": 2.382239613706854e-05, "loss": 0.369, "step": 6848 }, { "epoch": 1.19, "learning_rate": 2.381564098095363e-05, "loss": 0.3707, "step": 6849 }, { "epoch": 1.19, "learning_rate": 2.3808886166070787e-05, "loss": 0.3658, "step": 6850 }, { "epoch": 1.19, "learning_rate": 2.3802131693045622e-05, "loss": 0.3624, "step": 6851 }, { "epoch": 1.19, "learning_rate": 2.3795377562503684e-05, "loss": 0.3533, "step": 6852 }, { "epoch": 1.19, "learning_rate": 2.3788623775070533e-05, "loss": 0.3616, "step": 6853 }, { "epoch": 1.19, "learning_rate": 2.378187033137169e-05, "loss": 0.3703, "step": 6854 }, { "epoch": 1.2, "learning_rate": 2.3775117232032623e-05, "loss": 0.3617, "step": 6855 }, { "epoch": 1.2, "learning_rate": 2.3768364477678785e-05, "loss": 0.3571, "step": 6856 }, { "epoch": 1.2, "learning_rate": 2.3761612068935586e-05, "loss": 0.3797, "step": 6857 }, { "epoch": 1.2, "learning_rate": 2.3754860006428403e-05, "loss": 0.3789, "step": 6858 }, { "epoch": 1.2, "learning_rate": 2.37481082907826e-05, "loss": 0.378, "step": 6859 }, { "epoch": 1.2, "learning_rate": 2.3741356922623498e-05, "loss": 0.356, "step": 6860 }, { "epoch": 1.2, "learning_rate": 2.3734605902576386e-05, "loss": 0.3595, "step": 6861 }, { "epoch": 1.2, "learning_rate": 2.37278552312665e-05, "loss": 0.3684, "step": 6862 }, { "epoch": 1.2, "learning_rate": 2.3721104909319085e-05, "loss": 0.3594, "step": 6863 }, { "epoch": 1.2, "learning_rate": 2.3714354937359313e-05, "loss": 0.3919, "step": 6864 }, { "epoch": 1.2, "learning_rate": 2.370760531601236e-05, "loss": 0.3526, "step": 6865 }, { "epoch": 1.2, "learning_rate": 2.3700856045903336e-05, "loss": 0.3813, "step": 6866 }, { "epoch": 1.2, "learning_rate": 2.3694107127657352e-05, "loss": 0.3697, "step": 6867 }, { "epoch": 1.2, "learning_rate": 2.3687358561899453e-05, "loss": 0.3598, "step": 6868 }, { "epoch": 1.2, "learning_rate": 2.3680610349254668e-05, "loss": 0.4222, "step": 6869 }, { "epoch": 1.2, "learning_rate": 2.3673862490347993e-05, "loss": 0.3691, "step": 6870 }, { "epoch": 1.2, "learning_rate": 2.3667114985804395e-05, "loss": 0.3822, "step": 6871 }, { "epoch": 1.2, "learning_rate": 2.3660367836248803e-05, "loss": 0.3734, "step": 6872 }, { "epoch": 1.2, "learning_rate": 2.3653621042306122e-05, "loss": 0.3572, "step": 6873 }, { "epoch": 1.2, "learning_rate": 2.3646874604601192e-05, "loss": 0.3814, "step": 6874 }, { "epoch": 1.2, "learning_rate": 2.364012852375886e-05, "loss": 0.3585, "step": 6875 }, { "epoch": 1.2, "learning_rate": 2.3633382800403926e-05, "loss": 0.3808, "step": 6876 }, { "epoch": 1.2, "learning_rate": 2.3626637435161154e-05, "loss": 0.3661, "step": 6877 }, { "epoch": 1.2, "learning_rate": 2.3619892428655265e-05, "loss": 0.3751, "step": 6878 }, { "epoch": 1.2, "learning_rate": 2.3613147781510962e-05, "loss": 0.3698, "step": 6879 }, { "epoch": 1.2, "learning_rate": 2.360640349435292e-05, "loss": 0.3706, "step": 6880 }, { "epoch": 1.2, "learning_rate": 2.359965956780575e-05, "loss": 0.3664, "step": 6881 }, { "epoch": 1.2, "learning_rate": 2.359291600249407e-05, "loss": 0.3596, "step": 6882 }, { "epoch": 1.2, "learning_rate": 2.3586172799042434e-05, "loss": 0.361, "step": 6883 }, { "epoch": 1.2, "learning_rate": 2.3579429958075385e-05, "loss": 0.3589, "step": 6884 }, { "epoch": 1.2, "learning_rate": 2.3572687480217408e-05, "loss": 0.3749, "step": 6885 }, { "epoch": 1.2, "learning_rate": 2.3565945366092963e-05, "loss": 0.3554, "step": 6886 }, { "epoch": 1.2, "learning_rate": 2.3559203616326493e-05, "loss": 0.3581, "step": 6887 }, { "epoch": 1.2, "learning_rate": 2.3552462231542393e-05, "loss": 0.3469, "step": 6888 }, { "epoch": 1.2, "learning_rate": 2.3545721212365025e-05, "loss": 0.3748, "step": 6889 }, { "epoch": 1.2, "learning_rate": 2.353898055941871e-05, "loss": 0.3623, "step": 6890 }, { "epoch": 1.2, "learning_rate": 2.3532240273327748e-05, "loss": 0.3834, "step": 6891 }, { "epoch": 1.2, "learning_rate": 2.3525500354716402e-05, "loss": 0.3536, "step": 6892 }, { "epoch": 1.2, "learning_rate": 2.3518760804208896e-05, "loss": 0.3792, "step": 6893 }, { "epoch": 1.2, "learning_rate": 2.351202162242943e-05, "loss": 0.3741, "step": 6894 }, { "epoch": 1.2, "learning_rate": 2.3505282810002155e-05, "loss": 0.3784, "step": 6895 }, { "epoch": 1.2, "learning_rate": 2.3498544367551197e-05, "loss": 0.3568, "step": 6896 }, { "epoch": 1.2, "learning_rate": 2.3491806295700653e-05, "loss": 0.3716, "step": 6897 }, { "epoch": 1.2, "learning_rate": 2.348506859507456e-05, "loss": 0.3602, "step": 6898 }, { "epoch": 1.2, "learning_rate": 2.347833126629696e-05, "loss": 0.3791, "step": 6899 }, { "epoch": 1.2, "learning_rate": 2.3471594309991834e-05, "loss": 0.3867, "step": 6900 }, { "epoch": 1.2, "learning_rate": 2.3464857726783134e-05, "loss": 0.3743, "step": 6901 }, { "epoch": 1.2, "learning_rate": 2.345812151729478e-05, "loss": 0.3697, "step": 6902 }, { "epoch": 1.2, "learning_rate": 2.3451385682150638e-05, "loss": 0.3669, "step": 6903 }, { "epoch": 1.2, "learning_rate": 2.3444650221974585e-05, "loss": 0.3599, "step": 6904 }, { "epoch": 1.2, "learning_rate": 2.3437915137390413e-05, "loss": 0.3757, "step": 6905 }, { "epoch": 1.2, "learning_rate": 2.3431180429021926e-05, "loss": 0.3586, "step": 6906 }, { "epoch": 1.2, "learning_rate": 2.3424446097492843e-05, "loss": 0.3724, "step": 6907 }, { "epoch": 1.2, "learning_rate": 2.341771214342688e-05, "loss": 0.3624, "step": 6908 }, { "epoch": 1.2, "learning_rate": 2.341097856744771e-05, "loss": 0.3929, "step": 6909 }, { "epoch": 1.2, "learning_rate": 2.3404245370178984e-05, "loss": 0.3708, "step": 6910 }, { "epoch": 1.2, "learning_rate": 2.339751255224431e-05, "loss": 0.3774, "step": 6911 }, { "epoch": 1.21, "learning_rate": 2.339078011426723e-05, "loss": 0.3722, "step": 6912 }, { "epoch": 1.21, "learning_rate": 2.3384048056871306e-05, "loss": 0.3766, "step": 6913 }, { "epoch": 1.21, "learning_rate": 2.3377316380680013e-05, "loss": 0.3669, "step": 6914 }, { "epoch": 1.21, "learning_rate": 2.3370585086316843e-05, "loss": 0.3786, "step": 6915 }, { "epoch": 1.21, "learning_rate": 2.3363854174405198e-05, "loss": 0.3602, "step": 6916 }, { "epoch": 1.21, "learning_rate": 2.3357123645568485e-05, "loss": 0.373, "step": 6917 }, { "epoch": 1.21, "learning_rate": 2.3350393500430067e-05, "loss": 0.3625, "step": 6918 }, { "epoch": 1.21, "learning_rate": 2.334366373961325e-05, "loss": 0.3639, "step": 6919 }, { "epoch": 1.21, "learning_rate": 2.333693436374132e-05, "loss": 0.3704, "step": 6920 }, { "epoch": 1.21, "learning_rate": 2.3330205373437537e-05, "loss": 0.37, "step": 6921 }, { "epoch": 1.21, "learning_rate": 2.3323476769325114e-05, "loss": 0.3597, "step": 6922 }, { "epoch": 1.21, "learning_rate": 2.331674855202724e-05, "loss": 0.3461, "step": 6923 }, { "epoch": 1.21, "learning_rate": 2.3310020722167037e-05, "loss": 0.3863, "step": 6924 }, { "epoch": 1.21, "learning_rate": 2.3303293280367623e-05, "loss": 0.3659, "step": 6925 }, { "epoch": 1.21, "learning_rate": 2.329656622725207e-05, "loss": 0.3761, "step": 6926 }, { "epoch": 1.21, "learning_rate": 2.328983956344342e-05, "loss": 0.3541, "step": 6927 }, { "epoch": 1.21, "learning_rate": 2.3283113289564664e-05, "loss": 0.3693, "step": 6928 }, { "epoch": 1.21, "learning_rate": 2.3276387406238757e-05, "loss": 0.3644, "step": 6929 }, { "epoch": 1.21, "learning_rate": 2.3269661914088648e-05, "loss": 0.3625, "step": 6930 }, { "epoch": 1.21, "learning_rate": 2.3262936813737203e-05, "loss": 0.3625, "step": 6931 }, { "epoch": 1.21, "learning_rate": 2.3256212105807296e-05, "loss": 0.3612, "step": 6932 }, { "epoch": 1.21, "learning_rate": 2.3249487790921744e-05, "loss": 0.366, "step": 6933 }, { "epoch": 1.21, "learning_rate": 2.324276386970331e-05, "loss": 0.3643, "step": 6934 }, { "epoch": 1.21, "learning_rate": 2.3236040342774767e-05, "loss": 0.3768, "step": 6935 }, { "epoch": 1.21, "learning_rate": 2.322931721075879e-05, "loss": 0.3597, "step": 6936 }, { "epoch": 1.21, "learning_rate": 2.3222594474278088e-05, "loss": 0.3795, "step": 6937 }, { "epoch": 1.21, "learning_rate": 2.3215872133955275e-05, "loss": 0.3547, "step": 6938 }, { "epoch": 1.21, "learning_rate": 2.3209150190412947e-05, "loss": 0.3685, "step": 6939 }, { "epoch": 1.21, "learning_rate": 2.3202428644273687e-05, "loss": 0.3536, "step": 6940 }, { "epoch": 1.21, "learning_rate": 2.3195707496159996e-05, "loss": 0.3705, "step": 6941 }, { "epoch": 1.21, "learning_rate": 2.3188986746694376e-05, "loss": 0.3637, "step": 6942 }, { "epoch": 1.21, "learning_rate": 2.3182266396499274e-05, "loss": 0.3537, "step": 6943 }, { "epoch": 1.21, "learning_rate": 2.3175546446197117e-05, "loss": 0.383, "step": 6944 }, { "epoch": 1.21, "learning_rate": 2.316882689641026e-05, "loss": 0.3779, "step": 6945 }, { "epoch": 1.21, "learning_rate": 2.3162107747761064e-05, "loss": 0.3578, "step": 6946 }, { "epoch": 1.21, "learning_rate": 2.3155389000871812e-05, "loss": 0.3723, "step": 6947 }, { "epoch": 1.21, "learning_rate": 2.314867065636479e-05, "loss": 0.3757, "step": 6948 }, { "epoch": 1.21, "learning_rate": 2.3141952714862226e-05, "loss": 0.3643, "step": 6949 }, { "epoch": 1.21, "learning_rate": 2.31352351769863e-05, "loss": 0.3715, "step": 6950 }, { "epoch": 1.21, "learning_rate": 2.3128518043359166e-05, "loss": 0.3839, "step": 6951 }, { "epoch": 1.21, "learning_rate": 2.3121801314602953e-05, "loss": 0.3693, "step": 6952 }, { "epoch": 1.21, "learning_rate": 2.3115084991339726e-05, "loss": 0.3653, "step": 6953 }, { "epoch": 1.21, "learning_rate": 2.3108369074191527e-05, "loss": 0.3624, "step": 6954 }, { "epoch": 1.21, "learning_rate": 2.310165356378038e-05, "loss": 0.3701, "step": 6955 }, { "epoch": 1.21, "learning_rate": 2.3094938460728232e-05, "loss": 0.3531, "step": 6956 }, { "epoch": 1.21, "learning_rate": 2.3088223765657018e-05, "loss": 0.3492, "step": 6957 }, { "epoch": 1.21, "learning_rate": 2.308150947918863e-05, "loss": 0.3685, "step": 6958 }, { "epoch": 1.21, "learning_rate": 2.3074795601944914e-05, "loss": 0.3658, "step": 6959 }, { "epoch": 1.21, "learning_rate": 2.306808213454769e-05, "loss": 0.3554, "step": 6960 }, { "epoch": 1.21, "learning_rate": 2.306136907761875e-05, "loss": 0.363, "step": 6961 }, { "epoch": 1.21, "learning_rate": 2.3054656431779803e-05, "loss": 0.3756, "step": 6962 }, { "epoch": 1.21, "learning_rate": 2.3047944197652577e-05, "loss": 0.3598, "step": 6963 }, { "epoch": 1.21, "learning_rate": 2.3041232375858707e-05, "loss": 0.3675, "step": 6964 }, { "epoch": 1.21, "learning_rate": 2.3034520967019856e-05, "loss": 0.3633, "step": 6965 }, { "epoch": 1.21, "learning_rate": 2.3027809971757574e-05, "loss": 0.3642, "step": 6966 }, { "epoch": 1.21, "learning_rate": 2.3021099390693428e-05, "loss": 0.3655, "step": 6967 }, { "epoch": 1.21, "learning_rate": 2.3014389224448936e-05, "loss": 0.3649, "step": 6968 }, { "epoch": 1.22, "learning_rate": 2.3007679473645533e-05, "loss": 0.3629, "step": 6969 }, { "epoch": 1.22, "learning_rate": 2.3000970138904693e-05, "loss": 0.3508, "step": 6970 }, { "epoch": 1.22, "learning_rate": 2.2994261220847796e-05, "loss": 0.3657, "step": 6971 }, { "epoch": 1.22, "learning_rate": 2.298755272009619e-05, "loss": 0.3756, "step": 6972 }, { "epoch": 1.22, "learning_rate": 2.2980844637271203e-05, "loss": 0.3558, "step": 6973 }, { "epoch": 1.22, "learning_rate": 2.2974136972994105e-05, "loss": 0.3736, "step": 6974 }, { "epoch": 1.22, "learning_rate": 2.2967429727886138e-05, "loss": 0.3552, "step": 6975 }, { "epoch": 1.22, "learning_rate": 2.2960722902568505e-05, "loss": 0.3731, "step": 6976 }, { "epoch": 1.22, "learning_rate": 2.295401649766236e-05, "loss": 0.3592, "step": 6977 }, { "epoch": 1.22, "learning_rate": 2.2947310513788844e-05, "loss": 0.3649, "step": 6978 }, { "epoch": 1.22, "learning_rate": 2.2940604951569027e-05, "loss": 0.3861, "step": 6979 }, { "epoch": 1.22, "learning_rate": 2.2933899811623956e-05, "loss": 0.3689, "step": 6980 }, { "epoch": 1.22, "learning_rate": 2.292719509457464e-05, "loss": 0.3431, "step": 6981 }, { "epoch": 1.22, "learning_rate": 2.2920490801042044e-05, "loss": 0.3648, "step": 6982 }, { "epoch": 1.22, "learning_rate": 2.2913786931647093e-05, "loss": 0.3573, "step": 6983 }, { "epoch": 1.22, "learning_rate": 2.2907083487010676e-05, "loss": 0.368, "step": 6984 }, { "epoch": 1.22, "learning_rate": 2.2900380467753657e-05, "loss": 0.3606, "step": 6985 }, { "epoch": 1.22, "learning_rate": 2.2893677874496806e-05, "loss": 0.3684, "step": 6986 }, { "epoch": 1.22, "learning_rate": 2.2886975707860943e-05, "loss": 0.379, "step": 6987 }, { "epoch": 1.22, "learning_rate": 2.288027396846676e-05, "loss": 0.3493, "step": 6988 }, { "epoch": 1.22, "learning_rate": 2.2873572656934973e-05, "loss": 0.3451, "step": 6989 }, { "epoch": 1.22, "learning_rate": 2.2866871773886218e-05, "loss": 0.3582, "step": 6990 }, { "epoch": 1.22, "learning_rate": 2.2860171319941108e-05, "loss": 0.3782, "step": 6991 }, { "epoch": 1.22, "learning_rate": 2.285347129572022e-05, "loss": 0.3556, "step": 6992 }, { "epoch": 1.22, "learning_rate": 2.2846771701844087e-05, "loss": 0.3687, "step": 6993 }, { "epoch": 1.22, "learning_rate": 2.2840072538933205e-05, "loss": 0.3764, "step": 6994 }, { "epoch": 1.22, "learning_rate": 2.2833373807608012e-05, "loss": 0.3719, "step": 6995 }, { "epoch": 1.22, "learning_rate": 2.2826675508488933e-05, "loss": 0.3649, "step": 6996 }, { "epoch": 1.22, "learning_rate": 2.2819977642196327e-05, "loss": 0.375, "step": 6997 }, { "epoch": 1.22, "learning_rate": 2.281328020935054e-05, "loss": 0.374, "step": 6998 }, { "epoch": 1.22, "learning_rate": 2.2806583210571865e-05, "loss": 0.3454, "step": 6999 }, { "epoch": 1.22, "learning_rate": 2.2799886646480546e-05, "loss": 0.371, "step": 7000 }, { "epoch": 1.22, "learning_rate": 2.2793190517696793e-05, "loss": 0.3838, "step": 7001 }, { "epoch": 1.22, "learning_rate": 2.278649482484079e-05, "loss": 0.3734, "step": 7002 }, { "epoch": 1.22, "learning_rate": 2.277979956853264e-05, "loss": 0.362, "step": 7003 }, { "epoch": 1.22, "learning_rate": 2.2773104749392466e-05, "loss": 0.3476, "step": 7004 }, { "epoch": 1.22, "learning_rate": 2.27664103680403e-05, "loss": 0.3693, "step": 7005 }, { "epoch": 1.22, "learning_rate": 2.275971642509617e-05, "loss": 0.3581, "step": 7006 }, { "epoch": 1.22, "learning_rate": 2.2753022921180017e-05, "loss": 0.3702, "step": 7007 }, { "epoch": 1.22, "learning_rate": 2.274632985691178e-05, "loss": 0.3673, "step": 7008 }, { "epoch": 1.22, "learning_rate": 2.2739637232911353e-05, "loss": 0.3619, "step": 7009 }, { "epoch": 1.22, "learning_rate": 2.2732945049798582e-05, "loss": 0.363, "step": 7010 }, { "epoch": 1.22, "learning_rate": 2.2726253308193278e-05, "loss": 0.3662, "step": 7011 }, { "epoch": 1.22, "learning_rate": 2.271956200871519e-05, "loss": 0.3556, "step": 7012 }, { "epoch": 1.22, "learning_rate": 2.271287115198405e-05, "loss": 0.3498, "step": 7013 }, { "epoch": 1.22, "learning_rate": 2.2706180738619538e-05, "loss": 0.3718, "step": 7014 }, { "epoch": 1.22, "learning_rate": 2.2699490769241304e-05, "loss": 0.3675, "step": 7015 }, { "epoch": 1.22, "learning_rate": 2.269280124446895e-05, "loss": 0.3682, "step": 7016 }, { "epoch": 1.22, "learning_rate": 2.2686112164922023e-05, "loss": 0.3912, "step": 7017 }, { "epoch": 1.22, "learning_rate": 2.267942353122005e-05, "loss": 0.37, "step": 7018 }, { "epoch": 1.22, "learning_rate": 2.26727353439825e-05, "loss": 0.3508, "step": 7019 }, { "epoch": 1.22, "learning_rate": 2.2666047603828827e-05, "loss": 0.3578, "step": 7020 }, { "epoch": 1.22, "learning_rate": 2.2659360311378408e-05, "loss": 0.3799, "step": 7021 }, { "epoch": 1.22, "learning_rate": 2.26526734672506e-05, "loss": 0.3679, "step": 7022 }, { "epoch": 1.22, "learning_rate": 2.2645987072064724e-05, "loss": 0.3541, "step": 7023 }, { "epoch": 1.22, "learning_rate": 2.2639301126440038e-05, "loss": 0.367, "step": 7024 }, { "epoch": 1.22, "learning_rate": 2.263261563099577e-05, "loss": 0.3681, "step": 7025 }, { "epoch": 1.22, "learning_rate": 2.2625930586351112e-05, "loss": 0.3659, "step": 7026 }, { "epoch": 1.23, "learning_rate": 2.261924599312521e-05, "loss": 0.3538, "step": 7027 }, { "epoch": 1.23, "learning_rate": 2.2612561851937164e-05, "loss": 0.383, "step": 7028 }, { "epoch": 1.23, "learning_rate": 2.2605878163406035e-05, "loss": 0.3519, "step": 7029 }, { "epoch": 1.23, "learning_rate": 2.2599194928150842e-05, "loss": 0.3685, "step": 7030 }, { "epoch": 1.23, "learning_rate": 2.2592512146790563e-05, "loss": 0.3667, "step": 7031 }, { "epoch": 1.23, "learning_rate": 2.2585829819944138e-05, "loss": 0.3737, "step": 7032 }, { "epoch": 1.23, "learning_rate": 2.257914794823045e-05, "loss": 0.375, "step": 7033 }, { "epoch": 1.23, "learning_rate": 2.2572466532268353e-05, "loss": 0.3809, "step": 7034 }, { "epoch": 1.23, "learning_rate": 2.2565785572676667e-05, "loss": 0.3625, "step": 7035 }, { "epoch": 1.23, "learning_rate": 2.2559105070074134e-05, "loss": 0.361, "step": 7036 }, { "epoch": 1.23, "learning_rate": 2.2552425025079503e-05, "loss": 0.3626, "step": 7037 }, { "epoch": 1.23, "learning_rate": 2.254574543831145e-05, "loss": 0.3703, "step": 7038 }, { "epoch": 1.23, "learning_rate": 2.2539066310388603e-05, "loss": 0.36, "step": 7039 }, { "epoch": 1.23, "learning_rate": 2.253238764192957e-05, "loss": 0.3562, "step": 7040 }, { "epoch": 1.23, "learning_rate": 2.2525709433552887e-05, "loss": 0.3658, "step": 7041 }, { "epoch": 1.23, "learning_rate": 2.2519031685877096e-05, "loss": 0.3509, "step": 7042 }, { "epoch": 1.23, "learning_rate": 2.2512354399520646e-05, "loss": 0.3671, "step": 7043 }, { "epoch": 1.23, "learning_rate": 2.250567757510196e-05, "loss": 0.3653, "step": 7044 }, { "epoch": 1.23, "learning_rate": 2.2499001213239445e-05, "loss": 0.372, "step": 7045 }, { "epoch": 1.23, "learning_rate": 2.2492325314551414e-05, "loss": 0.3677, "step": 7046 }, { "epoch": 1.23, "learning_rate": 2.2485649879656168e-05, "loss": 0.3673, "step": 7047 }, { "epoch": 1.23, "learning_rate": 2.247897490917198e-05, "loss": 0.3375, "step": 7048 }, { "epoch": 1.23, "learning_rate": 2.247230040371705e-05, "loss": 0.3732, "step": 7049 }, { "epoch": 1.23, "learning_rate": 2.2465626363909544e-05, "loss": 0.3708, "step": 7050 }, { "epoch": 1.23, "learning_rate": 2.2458952790367593e-05, "loss": 0.3434, "step": 7051 }, { "epoch": 1.23, "learning_rate": 2.2452279683709272e-05, "loss": 0.3858, "step": 7052 }, { "epoch": 1.23, "learning_rate": 2.244560704455263e-05, "loss": 0.3543, "step": 7053 }, { "epoch": 1.23, "learning_rate": 2.2438934873515665e-05, "loss": 0.3705, "step": 7054 }, { "epoch": 1.23, "learning_rate": 2.2432263171216317e-05, "loss": 0.374, "step": 7055 }, { "epoch": 1.23, "learning_rate": 2.24255919382725e-05, "loss": 0.3614, "step": 7056 }, { "epoch": 1.23, "learning_rate": 2.241892117530208e-05, "loss": 0.3588, "step": 7057 }, { "epoch": 1.23, "learning_rate": 2.241225088292288e-05, "loss": 0.3626, "step": 7058 }, { "epoch": 1.23, "learning_rate": 2.240558106175268e-05, "loss": 0.3785, "step": 7059 }, { "epoch": 1.23, "learning_rate": 2.239891171240921e-05, "loss": 0.3868, "step": 7060 }, { "epoch": 1.23, "learning_rate": 2.2392242835510176e-05, "loss": 0.3629, "step": 7061 }, { "epoch": 1.23, "learning_rate": 2.23855744316732e-05, "loss": 0.3573, "step": 7062 }, { "epoch": 1.23, "learning_rate": 2.2378906501515895e-05, "loss": 0.3664, "step": 7063 }, { "epoch": 1.23, "learning_rate": 2.237223904565583e-05, "loss": 0.3527, "step": 7064 }, { "epoch": 1.23, "learning_rate": 2.2365572064710517e-05, "loss": 0.3571, "step": 7065 }, { "epoch": 1.23, "learning_rate": 2.235890555929743e-05, "loss": 0.375, "step": 7066 }, { "epoch": 1.23, "learning_rate": 2.235223953003399e-05, "loss": 0.3788, "step": 7067 }, { "epoch": 1.23, "learning_rate": 2.234557397753758e-05, "loss": 0.3444, "step": 7068 }, { "epoch": 1.23, "learning_rate": 2.2338908902425543e-05, "loss": 0.3586, "step": 7069 }, { "epoch": 1.23, "learning_rate": 2.233224430531517e-05, "loss": 0.3657, "step": 7070 }, { "epoch": 1.23, "learning_rate": 2.232558018682373e-05, "loss": 0.364, "step": 7071 }, { "epoch": 1.23, "learning_rate": 2.2318916547568404e-05, "loss": 0.3763, "step": 7072 }, { "epoch": 1.23, "learning_rate": 2.2312253388166375e-05, "loss": 0.355, "step": 7073 }, { "epoch": 1.23, "learning_rate": 2.2305590709234735e-05, "loss": 0.3653, "step": 7074 }, { "epoch": 1.23, "learning_rate": 2.2298928511390586e-05, "loss": 0.3655, "step": 7075 }, { "epoch": 1.23, "learning_rate": 2.229226679525095e-05, "loss": 0.3688, "step": 7076 }, { "epoch": 1.23, "learning_rate": 2.22856055614328e-05, "loss": 0.3596, "step": 7077 }, { "epoch": 1.23, "learning_rate": 2.2278944810553083e-05, "loss": 0.36, "step": 7078 }, { "epoch": 1.23, "learning_rate": 2.227228454322869e-05, "loss": 0.3577, "step": 7079 }, { "epoch": 1.23, "learning_rate": 2.226562476007647e-05, "loss": 0.3733, "step": 7080 }, { "epoch": 1.23, "learning_rate": 2.2258965461713237e-05, "loss": 0.3704, "step": 7081 }, { "epoch": 1.23, "learning_rate": 2.2252306648755745e-05, "loss": 0.3533, "step": 7082 }, { "epoch": 1.23, "learning_rate": 2.2245648321820713e-05, "loss": 0.3872, "step": 7083 }, { "epoch": 1.24, "learning_rate": 2.22389904815248e-05, "loss": 0.3784, "step": 7084 }, { "epoch": 1.24, "learning_rate": 2.223233312848464e-05, "loss": 0.3867, "step": 7085 }, { "epoch": 1.24, "learning_rate": 2.2225676263316815e-05, "loss": 0.3615, "step": 7086 }, { "epoch": 1.24, "learning_rate": 2.2219019886637864e-05, "loss": 0.3801, "step": 7087 }, { "epoch": 1.24, "learning_rate": 2.221236399906426e-05, "loss": 0.375, "step": 7088 }, { "epoch": 1.24, "learning_rate": 2.220570860121246e-05, "loss": 0.3583, "step": 7089 }, { "epoch": 1.24, "learning_rate": 2.219905369369887e-05, "loss": 0.361, "step": 7090 }, { "epoch": 1.24, "learning_rate": 2.2192399277139814e-05, "loss": 0.3742, "step": 7091 }, { "epoch": 1.24, "learning_rate": 2.2185745352151637e-05, "loss": 0.3622, "step": 7092 }, { "epoch": 1.24, "learning_rate": 2.2179091919350575e-05, "loss": 0.3597, "step": 7093 }, { "epoch": 1.24, "learning_rate": 2.2172438979352862e-05, "loss": 0.3631, "step": 7094 }, { "epoch": 1.24, "learning_rate": 2.2165786532774656e-05, "loss": 0.3435, "step": 7095 }, { "epoch": 1.24, "learning_rate": 2.2159134580232083e-05, "loss": 0.3608, "step": 7096 }, { "epoch": 1.24, "learning_rate": 2.215248312234124e-05, "loss": 0.366, "step": 7097 }, { "epoch": 1.24, "learning_rate": 2.2145832159718143e-05, "loss": 0.37, "step": 7098 }, { "epoch": 1.24, "learning_rate": 2.2139181692978794e-05, "loss": 0.3619, "step": 7099 }, { "epoch": 1.24, "learning_rate": 2.213253172273912e-05, "loss": 0.3637, "step": 7100 }, { "epoch": 1.24, "learning_rate": 2.2125882249615028e-05, "loss": 0.3881, "step": 7101 }, { "epoch": 1.24, "learning_rate": 2.2119233274222356e-05, "loss": 0.3594, "step": 7102 }, { "epoch": 1.24, "learning_rate": 2.211258479717693e-05, "loss": 0.3669, "step": 7103 }, { "epoch": 1.24, "learning_rate": 2.210593681909449e-05, "loss": 0.3557, "step": 7104 }, { "epoch": 1.24, "learning_rate": 2.2099289340590756e-05, "loss": 0.3638, "step": 7105 }, { "epoch": 1.24, "learning_rate": 2.209264236228139e-05, "loss": 0.3519, "step": 7106 }, { "epoch": 1.24, "learning_rate": 2.208599588478202e-05, "loss": 0.37, "step": 7107 }, { "epoch": 1.24, "learning_rate": 2.2079349908708192e-05, "loss": 0.3504, "step": 7108 }, { "epoch": 1.24, "learning_rate": 2.2072704434675466e-05, "loss": 0.3643, "step": 7109 }, { "epoch": 1.24, "learning_rate": 2.20660594632993e-05, "loss": 0.3537, "step": 7110 }, { "epoch": 1.24, "learning_rate": 2.2059414995195136e-05, "loss": 0.3758, "step": 7111 }, { "epoch": 1.24, "learning_rate": 2.2052771030978364e-05, "loss": 0.3687, "step": 7112 }, { "epoch": 1.24, "learning_rate": 2.2046127571264308e-05, "loss": 0.3557, "step": 7113 }, { "epoch": 1.24, "learning_rate": 2.203948461666828e-05, "loss": 0.3582, "step": 7114 }, { "epoch": 1.24, "learning_rate": 2.2032842167805516e-05, "loss": 0.3658, "step": 7115 }, { "epoch": 1.24, "learning_rate": 2.2026200225291227e-05, "loss": 0.3544, "step": 7116 }, { "epoch": 1.24, "learning_rate": 2.2019558789740554e-05, "loss": 0.3658, "step": 7117 }, { "epoch": 1.24, "learning_rate": 2.20129178617686e-05, "loss": 0.3759, "step": 7118 }, { "epoch": 1.24, "learning_rate": 2.2006277441990432e-05, "loss": 0.3609, "step": 7119 }, { "epoch": 1.24, "learning_rate": 2.199963753102106e-05, "loss": 0.3674, "step": 7120 }, { "epoch": 1.24, "learning_rate": 2.1992998129475456e-05, "loss": 0.3543, "step": 7121 }, { "epoch": 1.24, "learning_rate": 2.1986359237968524e-05, "loss": 0.37, "step": 7122 }, { "epoch": 1.24, "learning_rate": 2.1979720857115137e-05, "loss": 0.3671, "step": 7123 }, { "epoch": 1.24, "learning_rate": 2.1973082987530118e-05, "loss": 0.3527, "step": 7124 }, { "epoch": 1.24, "learning_rate": 2.1966445629828258e-05, "loss": 0.3647, "step": 7125 }, { "epoch": 1.24, "learning_rate": 2.1959808784624262e-05, "loss": 0.3551, "step": 7126 }, { "epoch": 1.24, "learning_rate": 2.1953172452532826e-05, "loss": 0.3894, "step": 7127 }, { "epoch": 1.24, "learning_rate": 2.1946536634168586e-05, "loss": 0.3597, "step": 7128 }, { "epoch": 1.24, "learning_rate": 2.1939901330146113e-05, "loss": 0.368, "step": 7129 }, { "epoch": 1.24, "learning_rate": 2.193326654107995e-05, "loss": 0.3696, "step": 7130 }, { "epoch": 1.24, "learning_rate": 2.192663226758459e-05, "loss": 0.3629, "step": 7131 }, { "epoch": 1.24, "learning_rate": 2.1919998510274475e-05, "loss": 0.3666, "step": 7132 }, { "epoch": 1.24, "learning_rate": 2.191336526976401e-05, "loss": 0.36, "step": 7133 }, { "epoch": 1.24, "learning_rate": 2.1906732546667518e-05, "loss": 0.3465, "step": 7134 }, { "epoch": 1.24, "learning_rate": 2.1900100341599318e-05, "loss": 0.3683, "step": 7135 }, { "epoch": 1.24, "learning_rate": 2.1893468655173648e-05, "loss": 0.3659, "step": 7136 }, { "epoch": 1.24, "learning_rate": 2.1886837488004724e-05, "loss": 0.3923, "step": 7137 }, { "epoch": 1.24, "learning_rate": 2.1880206840706696e-05, "loss": 0.3763, "step": 7138 }, { "epoch": 1.24, "learning_rate": 2.1873576713893664e-05, "loss": 0.3787, "step": 7139 }, { "epoch": 1.24, "learning_rate": 2.1866947108179698e-05, "loss": 0.3476, "step": 7140 }, { "epoch": 1.25, "learning_rate": 2.186031802417879e-05, "loss": 0.3658, "step": 7141 }, { "epoch": 1.25, "learning_rate": 2.1853689462504923e-05, "loss": 0.3933, "step": 7142 }, { "epoch": 1.25, "learning_rate": 2.1847061423771996e-05, "loss": 0.3585, "step": 7143 }, { "epoch": 1.25, "learning_rate": 2.184043390859388e-05, "loss": 0.3616, "step": 7144 }, { "epoch": 1.25, "learning_rate": 2.18338069175844e-05, "loss": 0.3734, "step": 7145 }, { "epoch": 1.25, "learning_rate": 2.1827180451357293e-05, "loss": 0.3808, "step": 7146 }, { "epoch": 1.25, "learning_rate": 2.182055451052632e-05, "loss": 0.351, "step": 7147 }, { "epoch": 1.25, "learning_rate": 2.1813929095705122e-05, "loss": 0.3639, "step": 7148 }, { "epoch": 1.25, "learning_rate": 2.180730420750733e-05, "loss": 0.3643, "step": 7149 }, { "epoch": 1.25, "learning_rate": 2.180067984654653e-05, "loss": 0.3507, "step": 7150 }, { "epoch": 1.25, "learning_rate": 2.1794056013436224e-05, "loss": 0.3695, "step": 7151 }, { "epoch": 1.25, "learning_rate": 2.1787432708789894e-05, "loss": 0.3576, "step": 7152 }, { "epoch": 1.25, "learning_rate": 2.178080993322098e-05, "loss": 0.354, "step": 7153 }, { "epoch": 1.25, "learning_rate": 2.1774187687342852e-05, "loss": 0.3626, "step": 7154 }, { "epoch": 1.25, "learning_rate": 2.1767565971768827e-05, "loss": 0.3534, "step": 7155 }, { "epoch": 1.25, "learning_rate": 2.1760944787112204e-05, "loss": 0.3579, "step": 7156 }, { "epoch": 1.25, "learning_rate": 2.1754324133986194e-05, "loss": 0.3811, "step": 7157 }, { "epoch": 1.25, "learning_rate": 2.1747704013003995e-05, "loss": 0.362, "step": 7158 }, { "epoch": 1.25, "learning_rate": 2.1741084424778736e-05, "loss": 0.3651, "step": 7159 }, { "epoch": 1.25, "learning_rate": 2.1734465369923484e-05, "loss": 0.3548, "step": 7160 }, { "epoch": 1.25, "learning_rate": 2.1727846849051293e-05, "loss": 0.3616, "step": 7161 }, { "epoch": 1.25, "learning_rate": 2.1721228862775134e-05, "loss": 0.3811, "step": 7162 }, { "epoch": 1.25, "learning_rate": 2.1714611411707937e-05, "loss": 0.3712, "step": 7163 }, { "epoch": 1.25, "learning_rate": 2.170799449646259e-05, "loss": 0.3592, "step": 7164 }, { "epoch": 1.25, "learning_rate": 2.170137811765194e-05, "loss": 0.3699, "step": 7165 }, { "epoch": 1.25, "learning_rate": 2.1694762275888765e-05, "loss": 0.3602, "step": 7166 }, { "epoch": 1.25, "learning_rate": 2.16881469717858e-05, "loss": 0.3735, "step": 7167 }, { "epoch": 1.25, "learning_rate": 2.168153220595572e-05, "loss": 0.3724, "step": 7168 }, { "epoch": 1.25, "learning_rate": 2.1674917979011168e-05, "loss": 0.3599, "step": 7169 }, { "epoch": 1.25, "learning_rate": 2.1668304291564734e-05, "loss": 0.3633, "step": 7170 }, { "epoch": 1.25, "learning_rate": 2.1661691144228964e-05, "loss": 0.3583, "step": 7171 }, { "epoch": 1.25, "learning_rate": 2.165507853761632e-05, "loss": 0.3495, "step": 7172 }, { "epoch": 1.25, "learning_rate": 2.164846647233925e-05, "loss": 0.352, "step": 7173 }, { "epoch": 1.25, "learning_rate": 2.164185494901014e-05, "loss": 0.3475, "step": 7174 }, { "epoch": 1.25, "learning_rate": 2.163524396824132e-05, "loss": 0.3654, "step": 7175 }, { "epoch": 1.25, "learning_rate": 2.1628633530645093e-05, "loss": 0.3703, "step": 7176 }, { "epoch": 1.25, "learning_rate": 2.162202363683367e-05, "loss": 0.3515, "step": 7177 }, { "epoch": 1.25, "learning_rate": 2.161541428741925e-05, "loss": 0.366, "step": 7178 }, { "epoch": 1.25, "learning_rate": 2.160880548301396e-05, "loss": 0.3516, "step": 7179 }, { "epoch": 1.25, "learning_rate": 2.1602197224229892e-05, "loss": 0.3645, "step": 7180 }, { "epoch": 1.25, "learning_rate": 2.159558951167907e-05, "loss": 0.3662, "step": 7181 }, { "epoch": 1.25, "learning_rate": 2.1588982345973484e-05, "loss": 0.3538, "step": 7182 }, { "epoch": 1.25, "learning_rate": 2.1582375727725067e-05, "loss": 0.3465, "step": 7183 }, { "epoch": 1.25, "learning_rate": 2.1575769657545693e-05, "loss": 0.3677, "step": 7184 }, { "epoch": 1.25, "learning_rate": 2.156916413604718e-05, "loss": 0.357, "step": 7185 }, { "epoch": 1.25, "learning_rate": 2.156255916384134e-05, "loss": 0.3668, "step": 7186 }, { "epoch": 1.25, "learning_rate": 2.155595474153988e-05, "loss": 0.3597, "step": 7187 }, { "epoch": 1.25, "learning_rate": 2.1549350869754488e-05, "loss": 0.3675, "step": 7188 }, { "epoch": 1.25, "learning_rate": 2.1542747549096777e-05, "loss": 0.362, "step": 7189 }, { "epoch": 1.25, "learning_rate": 2.153614478017833e-05, "loss": 0.3715, "step": 7190 }, { "epoch": 1.25, "learning_rate": 2.152954256361067e-05, "loss": 0.3634, "step": 7191 }, { "epoch": 1.25, "learning_rate": 2.152294090000529e-05, "loss": 0.3619, "step": 7192 }, { "epoch": 1.25, "learning_rate": 2.1516339789973582e-05, "loss": 0.3592, "step": 7193 }, { "epoch": 1.25, "learning_rate": 2.1509739234126935e-05, "loss": 0.3587, "step": 7194 }, { "epoch": 1.25, "learning_rate": 2.1503139233076664e-05, "loss": 0.3847, "step": 7195 }, { "epoch": 1.25, "learning_rate": 2.149653978743403e-05, "loss": 0.3736, "step": 7196 }, { "epoch": 1.25, "learning_rate": 2.1489940897810267e-05, "loss": 0.3692, "step": 7197 }, { "epoch": 1.25, "learning_rate": 2.1483342564816528e-05, "loss": 0.3592, "step": 7198 }, { "epoch": 1.26, "learning_rate": 2.1476744789063924e-05, "loss": 0.3798, "step": 7199 }, { "epoch": 1.26, "learning_rate": 2.147014757116353e-05, "loss": 0.3608, "step": 7200 }, { "epoch": 1.26, "learning_rate": 2.1463550911726338e-05, "loss": 0.354, "step": 7201 }, { "epoch": 1.26, "learning_rate": 2.1456954811363335e-05, "loss": 0.3667, "step": 7202 }, { "epoch": 1.26, "learning_rate": 2.14503592706854e-05, "loss": 0.3615, "step": 7203 }, { "epoch": 1.26, "learning_rate": 2.1443764290303403e-05, "loss": 0.372, "step": 7204 }, { "epoch": 1.26, "learning_rate": 2.143716987082815e-05, "loss": 0.3538, "step": 7205 }, { "epoch": 1.26, "learning_rate": 2.143057601287038e-05, "loss": 0.3636, "step": 7206 }, { "epoch": 1.26, "learning_rate": 2.142398271704079e-05, "loss": 0.3665, "step": 7207 }, { "epoch": 1.26, "learning_rate": 2.141738998395005e-05, "loss": 0.3583, "step": 7208 }, { "epoch": 1.26, "learning_rate": 2.1410797814208737e-05, "loss": 0.3728, "step": 7209 }, { "epoch": 1.26, "learning_rate": 2.1404206208427396e-05, "loss": 0.3672, "step": 7210 }, { "epoch": 1.26, "learning_rate": 2.1397615167216522e-05, "loss": 0.3498, "step": 7211 }, { "epoch": 1.26, "learning_rate": 2.1391024691186557e-05, "loss": 0.3573, "step": 7212 }, { "epoch": 1.26, "learning_rate": 2.1384434780947858e-05, "loss": 0.3639, "step": 7213 }, { "epoch": 1.26, "learning_rate": 2.137784543711081e-05, "loss": 0.3774, "step": 7214 }, { "epoch": 1.26, "learning_rate": 2.1371256660285655e-05, "loss": 0.3559, "step": 7215 }, { "epoch": 1.26, "learning_rate": 2.136466845108263e-05, "loss": 0.3553, "step": 7216 }, { "epoch": 1.26, "learning_rate": 2.135808081011193e-05, "loss": 0.3677, "step": 7217 }, { "epoch": 1.26, "learning_rate": 2.1351493737983647e-05, "loss": 0.3516, "step": 7218 }, { "epoch": 1.26, "learning_rate": 2.134490723530787e-05, "loss": 0.3602, "step": 7219 }, { "epoch": 1.26, "learning_rate": 2.133832130269462e-05, "loss": 0.3486, "step": 7220 }, { "epoch": 1.26, "learning_rate": 2.133173594075386e-05, "loss": 0.3595, "step": 7221 }, { "epoch": 1.26, "learning_rate": 2.1325151150095495e-05, "loss": 0.3668, "step": 7222 }, { "epoch": 1.26, "learning_rate": 2.131856693132939e-05, "loss": 0.3646, "step": 7223 }, { "epoch": 1.26, "learning_rate": 2.1311983285065352e-05, "loss": 0.3575, "step": 7224 }, { "epoch": 1.26, "learning_rate": 2.130540021191313e-05, "loss": 0.3579, "step": 7225 }, { "epoch": 1.26, "learning_rate": 2.1298817712482438e-05, "loss": 0.3517, "step": 7226 }, { "epoch": 1.26, "learning_rate": 2.1292235787382907e-05, "loss": 0.366, "step": 7227 }, { "epoch": 1.26, "learning_rate": 2.128565443722414e-05, "loss": 0.3588, "step": 7228 }, { "epoch": 1.26, "learning_rate": 2.1279073662615674e-05, "loss": 0.3633, "step": 7229 }, { "epoch": 1.26, "learning_rate": 2.1272493464167005e-05, "loss": 0.3568, "step": 7230 }, { "epoch": 1.26, "learning_rate": 2.126591384248755e-05, "loss": 0.4018, "step": 7231 }, { "epoch": 1.26, "learning_rate": 2.125933479818671e-05, "loss": 0.3669, "step": 7232 }, { "epoch": 1.26, "learning_rate": 2.12527563318738e-05, "loss": 0.3461, "step": 7233 }, { "epoch": 1.26, "learning_rate": 2.12461784441581e-05, "loss": 0.389, "step": 7234 }, { "epoch": 1.26, "learning_rate": 2.1239601135648818e-05, "loss": 0.3556, "step": 7235 }, { "epoch": 1.26, "learning_rate": 2.1233024406955133e-05, "loss": 0.391, "step": 7236 }, { "epoch": 1.26, "learning_rate": 2.122644825868616e-05, "loss": 0.3581, "step": 7237 }, { "epoch": 1.26, "learning_rate": 2.1219872691450955e-05, "loss": 0.3655, "step": 7238 }, { "epoch": 1.26, "learning_rate": 2.1213297705858516e-05, "loss": 0.3566, "step": 7239 }, { "epoch": 1.26, "learning_rate": 2.12067233025178e-05, "loss": 0.3581, "step": 7240 }, { "epoch": 1.26, "learning_rate": 2.1200149482037707e-05, "loss": 0.3599, "step": 7241 }, { "epoch": 1.26, "learning_rate": 2.119357624502707e-05, "loss": 0.3729, "step": 7242 }, { "epoch": 1.26, "learning_rate": 2.1187003592094696e-05, "loss": 0.36, "step": 7243 }, { "epoch": 1.26, "learning_rate": 2.1180431523849308e-05, "loss": 0.3684, "step": 7244 }, { "epoch": 1.26, "learning_rate": 2.117386004089959e-05, "loss": 0.3577, "step": 7245 }, { "epoch": 1.26, "learning_rate": 2.116728914385417e-05, "loss": 0.361, "step": 7246 }, { "epoch": 1.26, "learning_rate": 2.116071883332162e-05, "loss": 0.3818, "step": 7247 }, { "epoch": 1.26, "learning_rate": 2.115414910991046e-05, "loss": 0.3675, "step": 7248 }, { "epoch": 1.26, "learning_rate": 2.1147579974229147e-05, "loss": 0.3519, "step": 7249 }, { "epoch": 1.26, "learning_rate": 2.114101142688611e-05, "loss": 0.3608, "step": 7250 }, { "epoch": 1.26, "learning_rate": 2.1134443468489667e-05, "loss": 0.3671, "step": 7251 }, { "epoch": 1.26, "learning_rate": 2.1127876099648157e-05, "loss": 0.3448, "step": 7252 }, { "epoch": 1.26, "learning_rate": 2.112130932096982e-05, "loss": 0.367, "step": 7253 }, { "epoch": 1.26, "learning_rate": 2.1114743133062825e-05, "loss": 0.3899, "step": 7254 }, { "epoch": 1.26, "learning_rate": 2.1108177536535327e-05, "loss": 0.3874, "step": 7255 }, { "epoch": 1.27, "learning_rate": 2.1101612531995408e-05, "loss": 0.3626, "step": 7256 }, { "epoch": 1.27, "learning_rate": 2.1095048120051076e-05, "loss": 0.3618, "step": 7257 }, { "epoch": 1.27, "learning_rate": 2.108848430131033e-05, "loss": 0.3578, "step": 7258 }, { "epoch": 1.27, "learning_rate": 2.1081921076381074e-05, "loss": 0.3582, "step": 7259 }, { "epoch": 1.27, "learning_rate": 2.1075358445871167e-05, "loss": 0.3756, "step": 7260 }, { "epoch": 1.27, "learning_rate": 2.1068796410388425e-05, "loss": 0.3749, "step": 7261 }, { "epoch": 1.27, "learning_rate": 2.1062234970540582e-05, "loss": 0.3691, "step": 7262 }, { "epoch": 1.27, "learning_rate": 2.105567412693536e-05, "loss": 0.3521, "step": 7263 }, { "epoch": 1.27, "learning_rate": 2.1049113880180393e-05, "loss": 0.3486, "step": 7264 }, { "epoch": 1.27, "learning_rate": 2.104255423088326e-05, "loss": 0.3697, "step": 7265 }, { "epoch": 1.27, "learning_rate": 2.103599517965149e-05, "loss": 0.3683, "step": 7266 }, { "epoch": 1.27, "learning_rate": 2.102943672709257e-05, "loss": 0.3438, "step": 7267 }, { "epoch": 1.27, "learning_rate": 2.1022878873813904e-05, "loss": 0.3683, "step": 7268 }, { "epoch": 1.27, "learning_rate": 2.101632162042288e-05, "loss": 0.3687, "step": 7269 }, { "epoch": 1.27, "learning_rate": 2.1009764967526784e-05, "loss": 0.3573, "step": 7270 }, { "epoch": 1.27, "learning_rate": 2.1003208915732887e-05, "loss": 0.3695, "step": 7271 }, { "epoch": 1.27, "learning_rate": 2.0996653465648387e-05, "loss": 0.3728, "step": 7272 }, { "epoch": 1.27, "learning_rate": 2.09900986178804e-05, "loss": 0.3554, "step": 7273 }, { "epoch": 1.27, "learning_rate": 2.0983544373036044e-05, "loss": 0.3561, "step": 7274 }, { "epoch": 1.27, "learning_rate": 2.0976990731722336e-05, "loss": 0.3631, "step": 7275 }, { "epoch": 1.27, "learning_rate": 2.0970437694546257e-05, "loss": 0.3571, "step": 7276 }, { "epoch": 1.27, "learning_rate": 2.096388526211471e-05, "loss": 0.3729, "step": 7277 }, { "epoch": 1.27, "learning_rate": 2.095733343503457e-05, "loss": 0.3619, "step": 7278 }, { "epoch": 1.27, "learning_rate": 2.0950782213912638e-05, "loss": 0.3643, "step": 7279 }, { "epoch": 1.27, "learning_rate": 2.0944231599355675e-05, "loss": 0.3561, "step": 7280 }, { "epoch": 1.27, "learning_rate": 2.093768159197037e-05, "loss": 0.3687, "step": 7281 }, { "epoch": 1.27, "learning_rate": 2.0931132192363355e-05, "loss": 0.3809, "step": 7282 }, { "epoch": 1.27, "learning_rate": 2.0924583401141217e-05, "loss": 0.359, "step": 7283 }, { "epoch": 1.27, "learning_rate": 2.0918035218910482e-05, "loss": 0.371, "step": 7284 }, { "epoch": 1.27, "learning_rate": 2.0911487646277623e-05, "loss": 0.3709, "step": 7285 }, { "epoch": 1.27, "learning_rate": 2.0904940683849038e-05, "loss": 0.3732, "step": 7286 }, { "epoch": 1.27, "learning_rate": 2.0898394332231096e-05, "loss": 0.3688, "step": 7287 }, { "epoch": 1.27, "learning_rate": 2.08918485920301e-05, "loss": 0.3519, "step": 7288 }, { "epoch": 1.27, "learning_rate": 2.088530346385228e-05, "loss": 0.3674, "step": 7289 }, { "epoch": 1.27, "learning_rate": 2.0878758948303823e-05, "loss": 0.3612, "step": 7290 }, { "epoch": 1.27, "learning_rate": 2.0872215045990868e-05, "loss": 0.3542, "step": 7291 }, { "epoch": 1.27, "learning_rate": 2.0865671757519478e-05, "loss": 0.356, "step": 7292 }, { "epoch": 1.27, "learning_rate": 2.0859129083495688e-05, "loss": 0.3696, "step": 7293 }, { "epoch": 1.27, "learning_rate": 2.0852587024525434e-05, "loss": 0.3613, "step": 7294 }, { "epoch": 1.27, "learning_rate": 2.0846045581214625e-05, "loss": 0.3717, "step": 7295 }, { "epoch": 1.27, "learning_rate": 2.0839504754169104e-05, "loss": 0.3678, "step": 7296 }, { "epoch": 1.27, "learning_rate": 2.0832964543994676e-05, "loss": 0.3448, "step": 7297 }, { "epoch": 1.27, "learning_rate": 2.082642495129705e-05, "loss": 0.3603, "step": 7298 }, { "epoch": 1.27, "learning_rate": 2.0819885976681913e-05, "loss": 0.3474, "step": 7299 }, { "epoch": 1.27, "learning_rate": 2.0813347620754876e-05, "loss": 0.3874, "step": 7300 }, { "epoch": 1.27, "learning_rate": 2.0806809884121488e-05, "loss": 0.3692, "step": 7301 }, { "epoch": 1.27, "learning_rate": 2.0800272767387264e-05, "loss": 0.3786, "step": 7302 }, { "epoch": 1.27, "learning_rate": 2.0793736271157648e-05, "loss": 0.3685, "step": 7303 }, { "epoch": 1.27, "learning_rate": 2.0787200396038027e-05, "loss": 0.3704, "step": 7304 }, { "epoch": 1.27, "learning_rate": 2.0780665142633725e-05, "loss": 0.3715, "step": 7305 }, { "epoch": 1.27, "learning_rate": 2.0774130511550004e-05, "loss": 0.3642, "step": 7306 }, { "epoch": 1.27, "learning_rate": 2.0767596503392107e-05, "loss": 0.389, "step": 7307 }, { "epoch": 1.27, "learning_rate": 2.076106311876517e-05, "loss": 0.3488, "step": 7308 }, { "epoch": 1.27, "learning_rate": 2.0754530358274287e-05, "loss": 0.3665, "step": 7309 }, { "epoch": 1.27, "learning_rate": 2.0747998222524517e-05, "loss": 0.3621, "step": 7310 }, { "epoch": 1.27, "learning_rate": 2.074146671212082e-05, "loss": 0.3545, "step": 7311 }, { "epoch": 1.27, "learning_rate": 2.0734935827668134e-05, "loss": 0.3702, "step": 7312 }, { "epoch": 1.28, "learning_rate": 2.0728405569771332e-05, "loss": 0.3631, "step": 7313 }, { "epoch": 1.28, "learning_rate": 2.0721875939035214e-05, "loss": 0.3448, "step": 7314 }, { "epoch": 1.28, "learning_rate": 2.0715346936064527e-05, "loss": 0.3631, "step": 7315 }, { "epoch": 1.28, "learning_rate": 2.0708818561463978e-05, "loss": 0.3784, "step": 7316 }, { "epoch": 1.28, "learning_rate": 2.0702290815838194e-05, "loss": 0.3611, "step": 7317 }, { "epoch": 1.28, "learning_rate": 2.0695763699791738e-05, "loss": 0.3742, "step": 7318 }, { "epoch": 1.28, "learning_rate": 2.068923721392915e-05, "loss": 0.3627, "step": 7319 }, { "epoch": 1.28, "learning_rate": 2.0682711358854877e-05, "loss": 0.354, "step": 7320 }, { "epoch": 1.28, "learning_rate": 2.0676186135173325e-05, "loss": 0.3699, "step": 7321 }, { "epoch": 1.28, "learning_rate": 2.0669661543488838e-05, "loss": 0.37, "step": 7322 }, { "epoch": 1.28, "learning_rate": 2.066313758440569e-05, "loss": 0.3639, "step": 7323 }, { "epoch": 1.28, "learning_rate": 2.0656614258528113e-05, "loss": 0.3678, "step": 7324 }, { "epoch": 1.28, "learning_rate": 2.065009156646028e-05, "loss": 0.3733, "step": 7325 }, { "epoch": 1.28, "learning_rate": 2.0643569508806293e-05, "loss": 0.3724, "step": 7326 }, { "epoch": 1.28, "learning_rate": 2.0637048086170207e-05, "loss": 0.3603, "step": 7327 }, { "epoch": 1.28, "learning_rate": 2.0630527299156e-05, "loss": 0.3549, "step": 7328 }, { "epoch": 1.28, "learning_rate": 2.062400714836761e-05, "loss": 0.3659, "step": 7329 }, { "epoch": 1.28, "learning_rate": 2.0617487634408916e-05, "loss": 0.3596, "step": 7330 }, { "epoch": 1.28, "learning_rate": 2.061096875788373e-05, "loss": 0.3645, "step": 7331 }, { "epoch": 1.28, "learning_rate": 2.06044505193958e-05, "loss": 0.3797, "step": 7332 }, { "epoch": 1.28, "learning_rate": 2.059793291954883e-05, "loss": 0.3519, "step": 7333 }, { "epoch": 1.28, "learning_rate": 2.0591415958946452e-05, "loss": 0.3729, "step": 7334 }, { "epoch": 1.28, "learning_rate": 2.058489963819224e-05, "loss": 0.3456, "step": 7335 }, { "epoch": 1.28, "learning_rate": 2.0578383957889725e-05, "loss": 0.361, "step": 7336 }, { "epoch": 1.28, "learning_rate": 2.057186891864235e-05, "loss": 0.3577, "step": 7337 }, { "epoch": 1.28, "learning_rate": 2.0565354521053533e-05, "loss": 0.3607, "step": 7338 }, { "epoch": 1.28, "learning_rate": 2.055884076572659e-05, "loss": 0.3633, "step": 7339 }, { "epoch": 1.28, "learning_rate": 2.0552327653264813e-05, "loss": 0.3619, "step": 7340 }, { "epoch": 1.28, "learning_rate": 2.054581518427143e-05, "loss": 0.3536, "step": 7341 }, { "epoch": 1.28, "learning_rate": 2.0539303359349598e-05, "loss": 0.3635, "step": 7342 }, { "epoch": 1.28, "learning_rate": 2.0532792179102422e-05, "loss": 0.3667, "step": 7343 }, { "epoch": 1.28, "learning_rate": 2.0526281644132932e-05, "loss": 0.3614, "step": 7344 }, { "epoch": 1.28, "learning_rate": 2.0519771755044116e-05, "loss": 0.3562, "step": 7345 }, { "epoch": 1.28, "learning_rate": 2.0513262512438902e-05, "loss": 0.3717, "step": 7346 }, { "epoch": 1.28, "learning_rate": 2.050675391692015e-05, "loss": 0.3764, "step": 7347 }, { "epoch": 1.28, "learning_rate": 2.0500245969090663e-05, "loss": 0.3496, "step": 7348 }, { "epoch": 1.28, "learning_rate": 2.0493738669553182e-05, "loss": 0.3616, "step": 7349 }, { "epoch": 1.28, "learning_rate": 2.048723201891039e-05, "loss": 0.3645, "step": 7350 }, { "epoch": 1.28, "learning_rate": 2.04807260177649e-05, "loss": 0.3671, "step": 7351 }, { "epoch": 1.28, "learning_rate": 2.0474220666719297e-05, "loss": 0.3676, "step": 7352 }, { "epoch": 1.28, "learning_rate": 2.0467715966376066e-05, "loss": 0.3702, "step": 7353 }, { "epoch": 1.28, "learning_rate": 2.0461211917337654e-05, "loss": 0.383, "step": 7354 }, { "epoch": 1.28, "learning_rate": 2.0454708520206444e-05, "loss": 0.3594, "step": 7355 }, { "epoch": 1.28, "learning_rate": 2.0448205775584738e-05, "loss": 0.3703, "step": 7356 }, { "epoch": 1.28, "learning_rate": 2.0441703684074836e-05, "loss": 0.3667, "step": 7357 }, { "epoch": 1.28, "learning_rate": 2.0435202246278905e-05, "loss": 0.364, "step": 7358 }, { "epoch": 1.28, "learning_rate": 2.0428701462799093e-05, "loss": 0.3577, "step": 7359 }, { "epoch": 1.28, "learning_rate": 2.042220133423749e-05, "loss": 0.3704, "step": 7360 }, { "epoch": 1.28, "learning_rate": 2.041570186119611e-05, "loss": 0.3569, "step": 7361 }, { "epoch": 1.28, "learning_rate": 2.0409203044276888e-05, "loss": 0.3506, "step": 7362 }, { "epoch": 1.28, "learning_rate": 2.0402704884081753e-05, "loss": 0.3604, "step": 7363 }, { "epoch": 1.28, "learning_rate": 2.0396207381212533e-05, "loss": 0.3733, "step": 7364 }, { "epoch": 1.28, "learning_rate": 2.038971053627099e-05, "loss": 0.3758, "step": 7365 }, { "epoch": 1.28, "learning_rate": 2.0383214349858854e-05, "loss": 0.36, "step": 7366 }, { "epoch": 1.28, "learning_rate": 2.037671882257776e-05, "loss": 0.3587, "step": 7367 }, { "epoch": 1.28, "learning_rate": 2.0370223955029316e-05, "loss": 0.3667, "step": 7368 }, { "epoch": 1.28, "learning_rate": 2.0363729747815057e-05, "loss": 0.3452, "step": 7369 }, { "epoch": 1.28, "learning_rate": 2.035723620153644e-05, "loss": 0.3639, "step": 7370 }, { "epoch": 1.29, "learning_rate": 2.035074331679488e-05, "loss": 0.355, "step": 7371 }, { "epoch": 1.29, "learning_rate": 2.034425109419173e-05, "loss": 0.3569, "step": 7372 }, { "epoch": 1.29, "learning_rate": 2.033775953432825e-05, "loss": 0.3635, "step": 7373 }, { "epoch": 1.29, "learning_rate": 2.03312686378057e-05, "loss": 0.3639, "step": 7374 }, { "epoch": 1.29, "learning_rate": 2.0324778405225226e-05, "loss": 0.3613, "step": 7375 }, { "epoch": 1.29, "learning_rate": 2.0318288837187933e-05, "loss": 0.3543, "step": 7376 }, { "epoch": 1.29, "learning_rate": 2.0311799934294862e-05, "loss": 0.3703, "step": 7377 }, { "epoch": 1.29, "learning_rate": 2.0305311697146985e-05, "loss": 0.3677, "step": 7378 }, { "epoch": 1.29, "learning_rate": 2.0298824126345222e-05, "loss": 0.3621, "step": 7379 }, { "epoch": 1.29, "learning_rate": 2.0292337222490436e-05, "loss": 0.3696, "step": 7380 }, { "epoch": 1.29, "learning_rate": 2.0285850986183423e-05, "loss": 0.3707, "step": 7381 }, { "epoch": 1.29, "learning_rate": 2.0279365418024897e-05, "loss": 0.3577, "step": 7382 }, { "epoch": 1.29, "learning_rate": 2.0272880518615545e-05, "loss": 0.353, "step": 7383 }, { "epoch": 1.29, "learning_rate": 2.0266396288555955e-05, "loss": 0.3716, "step": 7384 }, { "epoch": 1.29, "learning_rate": 2.02599127284467e-05, "loss": 0.376, "step": 7385 }, { "epoch": 1.29, "learning_rate": 2.0253429838888255e-05, "loss": 0.381, "step": 7386 }, { "epoch": 1.29, "learning_rate": 2.024694762048103e-05, "loss": 0.3671, "step": 7387 }, { "epoch": 1.29, "learning_rate": 2.0240466073825387e-05, "loss": 0.3462, "step": 7388 }, { "epoch": 1.29, "learning_rate": 2.023398519952163e-05, "loss": 0.3656, "step": 7389 }, { "epoch": 1.29, "learning_rate": 2.0227504998169998e-05, "loss": 0.376, "step": 7390 }, { "epoch": 1.29, "learning_rate": 2.0221025470370657e-05, "loss": 0.3567, "step": 7391 }, { "epoch": 1.29, "learning_rate": 2.0214546616723723e-05, "loss": 0.3675, "step": 7392 }, { "epoch": 1.29, "learning_rate": 2.0208068437829236e-05, "loss": 0.3526, "step": 7393 }, { "epoch": 1.29, "learning_rate": 2.0201590934287186e-05, "loss": 0.368, "step": 7394 }, { "epoch": 1.29, "learning_rate": 2.0195114106697492e-05, "loss": 0.3553, "step": 7395 }, { "epoch": 1.29, "learning_rate": 2.0188637955660023e-05, "loss": 0.3655, "step": 7396 }, { "epoch": 1.29, "learning_rate": 2.018216248177457e-05, "loss": 0.3559, "step": 7397 }, { "epoch": 1.29, "learning_rate": 2.0175687685640877e-05, "loss": 0.3497, "step": 7398 }, { "epoch": 1.29, "learning_rate": 2.0169213567858608e-05, "loss": 0.3692, "step": 7399 }, { "epoch": 1.29, "learning_rate": 2.016274012902737e-05, "loss": 0.3649, "step": 7400 }, { "epoch": 1.29, "learning_rate": 2.0156267369746705e-05, "loss": 0.3483, "step": 7401 }, { "epoch": 1.29, "learning_rate": 2.0149795290616118e-05, "loss": 0.3674, "step": 7402 }, { "epoch": 1.29, "learning_rate": 2.0143323892235017e-05, "loss": 0.3564, "step": 7403 }, { "epoch": 1.29, "learning_rate": 2.0136853175202756e-05, "loss": 0.3707, "step": 7404 }, { "epoch": 1.29, "learning_rate": 2.0130383140118643e-05, "loss": 0.3609, "step": 7405 }, { "epoch": 1.29, "learning_rate": 2.0123913787581878e-05, "loss": 0.3858, "step": 7406 }, { "epoch": 1.29, "learning_rate": 2.0117445118191662e-05, "loss": 0.3731, "step": 7407 }, { "epoch": 1.29, "learning_rate": 2.011097713254709e-05, "loss": 0.3717, "step": 7408 }, { "epoch": 1.29, "learning_rate": 2.0104509831247202e-05, "loss": 0.3754, "step": 7409 }, { "epoch": 1.29, "learning_rate": 2.0098043214890982e-05, "loss": 0.3689, "step": 7410 }, { "epoch": 1.29, "learning_rate": 2.0091577284077323e-05, "loss": 0.3723, "step": 7411 }, { "epoch": 1.29, "learning_rate": 2.0085112039405102e-05, "loss": 0.3576, "step": 7412 }, { "epoch": 1.29, "learning_rate": 2.0078647481473097e-05, "loss": 0.3661, "step": 7413 }, { "epoch": 1.29, "learning_rate": 2.007218361088003e-05, "loss": 0.3517, "step": 7414 }, { "epoch": 1.29, "learning_rate": 2.0065720428224565e-05, "loss": 0.3651, "step": 7415 }, { "epoch": 1.29, "learning_rate": 2.0059257934105304e-05, "loss": 0.3665, "step": 7416 }, { "epoch": 1.29, "learning_rate": 2.0052796129120755e-05, "loss": 0.3731, "step": 7417 }, { "epoch": 1.29, "learning_rate": 2.004633501386941e-05, "loss": 0.3579, "step": 7418 }, { "epoch": 1.29, "learning_rate": 2.0039874588949682e-05, "loss": 0.3508, "step": 7419 }, { "epoch": 1.29, "learning_rate": 2.0033414854959894e-05, "loss": 0.355, "step": 7420 }, { "epoch": 1.29, "learning_rate": 2.0026955812498323e-05, "loss": 0.3824, "step": 7421 }, { "epoch": 1.29, "learning_rate": 2.002049746216319e-05, "loss": 0.3695, "step": 7422 }, { "epoch": 1.29, "learning_rate": 2.0014039804552637e-05, "loss": 0.3948, "step": 7423 }, { "epoch": 1.29, "learning_rate": 2.000758284026476e-05, "loss": 0.3731, "step": 7424 }, { "epoch": 1.29, "learning_rate": 2.0001126569897575e-05, "loss": 0.3668, "step": 7425 }, { "epoch": 1.29, "learning_rate": 1.9994670994049035e-05, "loss": 0.3684, "step": 7426 }, { "epoch": 1.29, "learning_rate": 1.9988216113317043e-05, "loss": 0.3748, "step": 7427 }, { "epoch": 1.3, "learning_rate": 1.9981761928299398e-05, "loss": 0.366, "step": 7428 }, { "epoch": 1.3, "learning_rate": 1.9975308439593905e-05, "loss": 0.342, "step": 7429 }, { "epoch": 1.3, "learning_rate": 1.996885564779823e-05, "loss": 0.3502, "step": 7430 }, { "epoch": 1.3, "learning_rate": 1.996240355351003e-05, "loss": 0.3588, "step": 7431 }, { "epoch": 1.3, "learning_rate": 1.995595215732685e-05, "loss": 0.3598, "step": 7432 }, { "epoch": 1.3, "learning_rate": 1.9949501459846214e-05, "loss": 0.3653, "step": 7433 }, { "epoch": 1.3, "learning_rate": 1.994305146166556e-05, "loss": 0.3691, "step": 7434 }, { "epoch": 1.3, "learning_rate": 1.9936602163382258e-05, "loss": 0.3615, "step": 7435 }, { "epoch": 1.3, "learning_rate": 1.9930153565593622e-05, "loss": 0.3722, "step": 7436 }, { "epoch": 1.3, "learning_rate": 1.99237056688969e-05, "loss": 0.348, "step": 7437 }, { "epoch": 1.3, "learning_rate": 1.9917258473889267e-05, "loss": 0.3639, "step": 7438 }, { "epoch": 1.3, "learning_rate": 1.9910811981167837e-05, "loss": 0.3606, "step": 7439 }, { "epoch": 1.3, "learning_rate": 1.990436619132967e-05, "loss": 0.3572, "step": 7440 }, { "epoch": 1.3, "learning_rate": 1.9897921104971762e-05, "loss": 0.3474, "step": 7441 }, { "epoch": 1.3, "learning_rate": 1.9891476722691006e-05, "loss": 0.3541, "step": 7442 }, { "epoch": 1.3, "learning_rate": 1.988503304508427e-05, "loss": 0.3497, "step": 7443 }, { "epoch": 1.3, "learning_rate": 1.9878590072748355e-05, "loss": 0.3672, "step": 7444 }, { "epoch": 1.3, "learning_rate": 1.9872147806279967e-05, "loss": 0.3497, "step": 7445 }, { "epoch": 1.3, "learning_rate": 1.986570624627578e-05, "loss": 0.3711, "step": 7446 }, { "epoch": 1.3, "learning_rate": 1.9859265393332376e-05, "loss": 0.3586, "step": 7447 }, { "epoch": 1.3, "learning_rate": 1.985282524804631e-05, "loss": 0.3512, "step": 7448 }, { "epoch": 1.3, "learning_rate": 1.9846385811014005e-05, "loss": 0.3669, "step": 7449 }, { "epoch": 1.3, "learning_rate": 1.9839947082831882e-05, "loss": 0.3758, "step": 7450 }, { "epoch": 1.3, "learning_rate": 1.9833509064096274e-05, "loss": 0.3427, "step": 7451 }, { "epoch": 1.3, "learning_rate": 1.9827071755403443e-05, "loss": 0.3581, "step": 7452 }, { "epoch": 1.3, "learning_rate": 1.9820635157349594e-05, "loss": 0.346, "step": 7453 }, { "epoch": 1.3, "learning_rate": 1.9814199270530853e-05, "loss": 0.3709, "step": 7454 }, { "epoch": 1.3, "learning_rate": 1.9807764095543292e-05, "loss": 0.3613, "step": 7455 }, { "epoch": 1.3, "learning_rate": 1.9801329632982913e-05, "loss": 0.3639, "step": 7456 }, { "epoch": 1.3, "learning_rate": 1.979489588344566e-05, "loss": 0.3442, "step": 7457 }, { "epoch": 1.3, "learning_rate": 1.978846284752739e-05, "loss": 0.3765, "step": 7458 }, { "epoch": 1.3, "learning_rate": 1.978203052582392e-05, "loss": 0.3578, "step": 7459 }, { "epoch": 1.3, "learning_rate": 1.9775598918930986e-05, "loss": 0.3722, "step": 7460 }, { "epoch": 1.3, "learning_rate": 1.9769168027444246e-05, "loss": 0.366, "step": 7461 }, { "epoch": 1.3, "learning_rate": 1.976273785195933e-05, "loss": 0.3652, "step": 7462 }, { "epoch": 1.3, "learning_rate": 1.9756308393071766e-05, "loss": 0.3629, "step": 7463 }, { "epoch": 1.3, "learning_rate": 1.9749879651377018e-05, "loss": 0.3669, "step": 7464 }, { "epoch": 1.3, "learning_rate": 1.9743451627470516e-05, "loss": 0.3557, "step": 7465 }, { "epoch": 1.3, "learning_rate": 1.9737024321947574e-05, "loss": 0.3577, "step": 7466 }, { "epoch": 1.3, "learning_rate": 1.9730597735403477e-05, "loss": 0.3476, "step": 7467 }, { "epoch": 1.3, "learning_rate": 1.972417186843344e-05, "loss": 0.3761, "step": 7468 }, { "epoch": 1.3, "learning_rate": 1.9717746721632598e-05, "loss": 0.3578, "step": 7469 }, { "epoch": 1.3, "learning_rate": 1.9711322295596025e-05, "loss": 0.3581, "step": 7470 }, { "epoch": 1.3, "learning_rate": 1.9704898590918725e-05, "loss": 0.3745, "step": 7471 }, { "epoch": 1.3, "learning_rate": 1.969847560819564e-05, "loss": 0.3705, "step": 7472 }, { "epoch": 1.3, "learning_rate": 1.9692053348021646e-05, "loss": 0.3614, "step": 7473 }, { "epoch": 1.3, "learning_rate": 1.968563181099155e-05, "loss": 0.3668, "step": 7474 }, { "epoch": 1.3, "learning_rate": 1.967921099770009e-05, "loss": 0.3572, "step": 7475 }, { "epoch": 1.3, "learning_rate": 1.967279090874193e-05, "loss": 0.3649, "step": 7476 }, { "epoch": 1.3, "learning_rate": 1.9666371544711692e-05, "loss": 0.3513, "step": 7477 }, { "epoch": 1.3, "learning_rate": 1.9659952906203894e-05, "loss": 0.3588, "step": 7478 }, { "epoch": 1.3, "learning_rate": 1.9653534993813034e-05, "loss": 0.3533, "step": 7479 }, { "epoch": 1.3, "learning_rate": 1.964711780813349e-05, "loss": 0.3558, "step": 7480 }, { "epoch": 1.3, "learning_rate": 1.9640701349759622e-05, "loss": 0.3892, "step": 7481 }, { "epoch": 1.3, "learning_rate": 1.963428561928568e-05, "loss": 0.3523, "step": 7482 }, { "epoch": 1.3, "learning_rate": 1.9627870617305872e-05, "loss": 0.3494, "step": 7483 }, { "epoch": 1.3, "learning_rate": 1.9621456344414335e-05, "loss": 0.3527, "step": 7484 }, { "epoch": 1.3, "learning_rate": 1.961504280120513e-05, "loss": 0.3629, "step": 7485 }, { "epoch": 1.31, "learning_rate": 1.960862998827227e-05, "loss": 0.3708, "step": 7486 }, { "epoch": 1.31, "learning_rate": 1.9602217906209667e-05, "loss": 0.3522, "step": 7487 }, { "epoch": 1.31, "learning_rate": 1.95958065556112e-05, "loss": 0.3665, "step": 7488 }, { "epoch": 1.31, "learning_rate": 1.9589395937070658e-05, "loss": 0.3591, "step": 7489 }, { "epoch": 1.31, "learning_rate": 1.9582986051181773e-05, "loss": 0.3594, "step": 7490 }, { "epoch": 1.31, "learning_rate": 1.9576576898538208e-05, "loss": 0.354, "step": 7491 }, { "epoch": 1.31, "learning_rate": 1.9570168479733546e-05, "loss": 0.3524, "step": 7492 }, { "epoch": 1.31, "learning_rate": 1.9563760795361323e-05, "loss": 0.3659, "step": 7493 }, { "epoch": 1.31, "learning_rate": 1.955735384601498e-05, "loss": 0.3545, "step": 7494 }, { "epoch": 1.31, "learning_rate": 1.955094763228793e-05, "loss": 0.3735, "step": 7495 }, { "epoch": 1.31, "learning_rate": 1.9544542154773472e-05, "loss": 0.3667, "step": 7496 }, { "epoch": 1.31, "learning_rate": 1.953813741406486e-05, "loss": 0.3651, "step": 7497 }, { "epoch": 1.31, "learning_rate": 1.9531733410755294e-05, "loss": 0.3666, "step": 7498 }, { "epoch": 1.31, "learning_rate": 1.9525330145437874e-05, "loss": 0.3792, "step": 7499 }, { "epoch": 1.31, "learning_rate": 1.9518927618705646e-05, "loss": 0.3657, "step": 7500 }, { "epoch": 1.31, "learning_rate": 1.95125258311516e-05, "loss": 0.3651, "step": 7501 }, { "epoch": 1.31, "learning_rate": 1.9506124783368645e-05, "loss": 0.3581, "step": 7502 }, { "epoch": 1.31, "learning_rate": 1.949972447594962e-05, "loss": 0.3683, "step": 7503 }, { "epoch": 1.31, "learning_rate": 1.94933249094873e-05, "loss": 0.3594, "step": 7504 }, { "epoch": 1.31, "learning_rate": 1.9486926084574382e-05, "loss": 0.3471, "step": 7505 }, { "epoch": 1.31, "learning_rate": 1.948052800180351e-05, "loss": 0.3469, "step": 7506 }, { "epoch": 1.31, "learning_rate": 1.9474130661767247e-05, "loss": 0.3729, "step": 7507 }, { "epoch": 1.31, "learning_rate": 1.9467734065058103e-05, "loss": 0.3551, "step": 7508 }, { "epoch": 1.31, "learning_rate": 1.94613382122685e-05, "loss": 0.3645, "step": 7509 }, { "epoch": 1.31, "learning_rate": 1.9454943103990787e-05, "loss": 0.3606, "step": 7510 }, { "epoch": 1.31, "learning_rate": 1.944854874081727e-05, "loss": 0.3541, "step": 7511 }, { "epoch": 1.31, "learning_rate": 1.9442155123340173e-05, "loss": 0.3711, "step": 7512 }, { "epoch": 1.31, "learning_rate": 1.943576225215164e-05, "loss": 0.3625, "step": 7513 }, { "epoch": 1.31, "learning_rate": 1.9429370127843766e-05, "loss": 0.3463, "step": 7514 }, { "epoch": 1.31, "learning_rate": 1.9422978751008563e-05, "loss": 0.3766, "step": 7515 }, { "epoch": 1.31, "learning_rate": 1.9416588122237968e-05, "loss": 0.3472, "step": 7516 }, { "epoch": 1.31, "learning_rate": 1.9410198242123867e-05, "loss": 0.3457, "step": 7517 }, { "epoch": 1.31, "learning_rate": 1.940380911125807e-05, "loss": 0.3583, "step": 7518 }, { "epoch": 1.31, "learning_rate": 1.9397420730232314e-05, "loss": 0.355, "step": 7519 }, { "epoch": 1.31, "learning_rate": 1.9391033099638273e-05, "loss": 0.3694, "step": 7520 }, { "epoch": 1.31, "learning_rate": 1.9384646220067525e-05, "loss": 0.3657, "step": 7521 }, { "epoch": 1.31, "learning_rate": 1.9378260092111615e-05, "loss": 0.3701, "step": 7522 }, { "epoch": 1.31, "learning_rate": 1.937187471636201e-05, "loss": 0.3708, "step": 7523 }, { "epoch": 1.31, "learning_rate": 1.9365490093410087e-05, "loss": 0.355, "step": 7524 }, { "epoch": 1.31, "learning_rate": 1.9359106223847178e-05, "loss": 0.3598, "step": 7525 }, { "epoch": 1.31, "learning_rate": 1.9352723108264524e-05, "loss": 0.3587, "step": 7526 }, { "epoch": 1.31, "learning_rate": 1.9346340747253317e-05, "loss": 0.3713, "step": 7527 }, { "epoch": 1.31, "learning_rate": 1.9339959141404645e-05, "loss": 0.3668, "step": 7528 }, { "epoch": 1.31, "learning_rate": 1.933357829130958e-05, "loss": 0.3813, "step": 7529 }, { "epoch": 1.31, "learning_rate": 1.9327198197559076e-05, "loss": 0.3618, "step": 7530 }, { "epoch": 1.31, "learning_rate": 1.9320818860744045e-05, "loss": 0.3584, "step": 7531 }, { "epoch": 1.31, "learning_rate": 1.9314440281455305e-05, "loss": 0.3558, "step": 7532 }, { "epoch": 1.31, "learning_rate": 1.930806246028361e-05, "loss": 0.3565, "step": 7533 }, { "epoch": 1.31, "learning_rate": 1.9301685397819687e-05, "loss": 0.3619, "step": 7534 }, { "epoch": 1.31, "learning_rate": 1.929530909465412e-05, "loss": 0.3632, "step": 7535 }, { "epoch": 1.31, "learning_rate": 1.9288933551377474e-05, "loss": 0.3636, "step": 7536 }, { "epoch": 1.31, "learning_rate": 1.928255876858024e-05, "loss": 0.3695, "step": 7537 }, { "epoch": 1.31, "learning_rate": 1.92761847468528e-05, "loss": 0.3697, "step": 7538 }, { "epoch": 1.31, "learning_rate": 1.926981148678551e-05, "loss": 0.3592, "step": 7539 }, { "epoch": 1.31, "learning_rate": 1.9263438988968635e-05, "loss": 0.3618, "step": 7540 }, { "epoch": 1.31, "learning_rate": 1.9257067253992382e-05, "loss": 0.3691, "step": 7541 }, { "epoch": 1.31, "learning_rate": 1.9250696282446868e-05, "loss": 0.3676, "step": 7542 }, { "epoch": 1.32, "learning_rate": 1.9244326074922147e-05, "loss": 0.3593, "step": 7543 }, { "epoch": 1.32, "learning_rate": 1.923795663200821e-05, "loss": 0.3581, "step": 7544 }, { "epoch": 1.32, "learning_rate": 1.923158795429497e-05, "loss": 0.357, "step": 7545 }, { "epoch": 1.32, "learning_rate": 1.9225220042372283e-05, "loss": 0.3509, "step": 7546 }, { "epoch": 1.32, "learning_rate": 1.9218852896829904e-05, "loss": 0.3557, "step": 7547 }, { "epoch": 1.32, "learning_rate": 1.9212486518257544e-05, "loss": 0.3568, "step": 7548 }, { "epoch": 1.32, "learning_rate": 1.9206120907244834e-05, "loss": 0.3581, "step": 7549 }, { "epoch": 1.32, "learning_rate": 1.9199756064381326e-05, "loss": 0.3611, "step": 7550 }, { "epoch": 1.32, "learning_rate": 1.9193391990256515e-05, "loss": 0.3678, "step": 7551 }, { "epoch": 1.32, "learning_rate": 1.9187028685459825e-05, "loss": 0.3629, "step": 7552 }, { "epoch": 1.32, "learning_rate": 1.9180666150580596e-05, "loss": 0.3476, "step": 7553 }, { "epoch": 1.32, "learning_rate": 1.9174304386208102e-05, "loss": 0.3551, "step": 7554 }, { "epoch": 1.32, "learning_rate": 1.916794339293154e-05, "loss": 0.3736, "step": 7555 }, { "epoch": 1.32, "learning_rate": 1.9161583171340057e-05, "loss": 0.3739, "step": 7556 }, { "epoch": 1.32, "learning_rate": 1.9155223722022708e-05, "loss": 0.3653, "step": 7557 }, { "epoch": 1.32, "learning_rate": 1.9148865045568482e-05, "loss": 0.3705, "step": 7558 }, { "epoch": 1.32, "learning_rate": 1.9142507142566295e-05, "loss": 0.3641, "step": 7559 }, { "epoch": 1.32, "learning_rate": 1.913615001360499e-05, "loss": 0.3727, "step": 7560 }, { "epoch": 1.32, "learning_rate": 1.9129793659273348e-05, "loss": 0.355, "step": 7561 }, { "epoch": 1.32, "learning_rate": 1.912343808016008e-05, "loss": 0.3768, "step": 7562 }, { "epoch": 1.32, "learning_rate": 1.9117083276853793e-05, "loss": 0.3485, "step": 7563 }, { "epoch": 1.32, "learning_rate": 1.911072924994306e-05, "loss": 0.3615, "step": 7564 }, { "epoch": 1.32, "learning_rate": 1.9104376000016372e-05, "loss": 0.3574, "step": 7565 }, { "epoch": 1.32, "learning_rate": 1.9098023527662128e-05, "loss": 0.3614, "step": 7566 }, { "epoch": 1.32, "learning_rate": 1.9091671833468693e-05, "loss": 0.3684, "step": 7567 }, { "epoch": 1.32, "learning_rate": 1.908532091802433e-05, "loss": 0.3508, "step": 7568 }, { "epoch": 1.32, "learning_rate": 1.9078970781917228e-05, "loss": 0.3717, "step": 7569 }, { "epoch": 1.32, "learning_rate": 1.9072621425735526e-05, "loss": 0.3658, "step": 7570 }, { "epoch": 1.32, "learning_rate": 1.9066272850067274e-05, "loss": 0.3573, "step": 7571 }, { "epoch": 1.32, "learning_rate": 1.9059925055500443e-05, "loss": 0.3705, "step": 7572 }, { "epoch": 1.32, "learning_rate": 1.9053578042622965e-05, "loss": 0.3586, "step": 7573 }, { "epoch": 1.32, "learning_rate": 1.9047231812022664e-05, "loss": 0.3489, "step": 7574 }, { "epoch": 1.32, "learning_rate": 1.904088636428731e-05, "loss": 0.3641, "step": 7575 }, { "epoch": 1.32, "learning_rate": 1.9034541700004586e-05, "loss": 0.3593, "step": 7576 }, { "epoch": 1.32, "learning_rate": 1.9028197819762115e-05, "loss": 0.3579, "step": 7577 }, { "epoch": 1.32, "learning_rate": 1.9021854724147453e-05, "loss": 0.3698, "step": 7578 }, { "epoch": 1.32, "learning_rate": 1.9015512413748077e-05, "loss": 0.3587, "step": 7579 }, { "epoch": 1.32, "learning_rate": 1.9009170889151375e-05, "loss": 0.3631, "step": 7580 }, { "epoch": 1.32, "learning_rate": 1.9002830150944682e-05, "loss": 0.3763, "step": 7581 }, { "epoch": 1.32, "learning_rate": 1.899649019971526e-05, "loss": 0.356, "step": 7582 }, { "epoch": 1.32, "learning_rate": 1.8990151036050274e-05, "loss": 0.3637, "step": 7583 }, { "epoch": 1.32, "learning_rate": 1.898381266053686e-05, "loss": 0.3601, "step": 7584 }, { "epoch": 1.32, "learning_rate": 1.8977475073762042e-05, "loss": 0.3688, "step": 7585 }, { "epoch": 1.32, "learning_rate": 1.8971138276312782e-05, "loss": 0.3717, "step": 7586 }, { "epoch": 1.32, "learning_rate": 1.8964802268775982e-05, "loss": 0.3583, "step": 7587 }, { "epoch": 1.32, "learning_rate": 1.8958467051738444e-05, "loss": 0.354, "step": 7588 }, { "epoch": 1.32, "learning_rate": 1.8952132625786927e-05, "loss": 0.3655, "step": 7589 }, { "epoch": 1.32, "learning_rate": 1.89457989915081e-05, "loss": 0.364, "step": 7590 }, { "epoch": 1.32, "learning_rate": 1.8939466149488567e-05, "loss": 0.3663, "step": 7591 }, { "epoch": 1.32, "learning_rate": 1.8933134100314837e-05, "loss": 0.3573, "step": 7592 }, { "epoch": 1.32, "learning_rate": 1.8926802844573372e-05, "loss": 0.358, "step": 7593 }, { "epoch": 1.32, "learning_rate": 1.892047238285054e-05, "loss": 0.3498, "step": 7594 }, { "epoch": 1.32, "learning_rate": 1.8914142715732656e-05, "loss": 0.3676, "step": 7595 }, { "epoch": 1.32, "learning_rate": 1.890781384380596e-05, "loss": 0.347, "step": 7596 }, { "epoch": 1.32, "learning_rate": 1.890148576765659e-05, "loss": 0.3774, "step": 7597 }, { "epoch": 1.32, "learning_rate": 1.889515848787064e-05, "loss": 0.3593, "step": 7598 }, { "epoch": 1.32, "learning_rate": 1.8888832005034112e-05, "loss": 0.38, "step": 7599 }, { "epoch": 1.33, "learning_rate": 1.8882506319732953e-05, "loss": 0.383, "step": 7600 }, { "epoch": 1.33, "learning_rate": 1.8876181432553026e-05, "loss": 0.3519, "step": 7601 }, { "epoch": 1.33, "learning_rate": 1.8869857344080104e-05, "loss": 0.3466, "step": 7602 }, { "epoch": 1.33, "learning_rate": 1.886353405489991e-05, "loss": 0.3405, "step": 7603 }, { "epoch": 1.33, "learning_rate": 1.885721156559809e-05, "loss": 0.3772, "step": 7604 }, { "epoch": 1.33, "learning_rate": 1.88508898767602e-05, "loss": 0.3691, "step": 7605 }, { "epoch": 1.33, "learning_rate": 1.8844568988971744e-05, "loss": 0.3586, "step": 7606 }, { "epoch": 1.33, "learning_rate": 1.8838248902818127e-05, "loss": 0.3578, "step": 7607 }, { "epoch": 1.33, "learning_rate": 1.8831929618884702e-05, "loss": 0.3442, "step": 7608 }, { "epoch": 1.33, "learning_rate": 1.8825611137756737e-05, "loss": 0.3562, "step": 7609 }, { "epoch": 1.33, "learning_rate": 1.8819293460019415e-05, "loss": 0.3621, "step": 7610 }, { "epoch": 1.33, "learning_rate": 1.8812976586257868e-05, "loss": 0.3562, "step": 7611 }, { "epoch": 1.33, "learning_rate": 1.8806660517057146e-05, "loss": 0.3628, "step": 7612 }, { "epoch": 1.33, "learning_rate": 1.8800345253002216e-05, "loss": 0.3534, "step": 7613 }, { "epoch": 1.33, "learning_rate": 1.8794030794677975e-05, "loss": 0.3538, "step": 7614 }, { "epoch": 1.33, "learning_rate": 1.8787717142669235e-05, "loss": 0.3495, "step": 7615 }, { "epoch": 1.33, "learning_rate": 1.8781404297560757e-05, "loss": 0.3648, "step": 7616 }, { "epoch": 1.33, "learning_rate": 1.8775092259937217e-05, "loss": 0.3631, "step": 7617 }, { "epoch": 1.33, "learning_rate": 1.8768781030383197e-05, "loss": 0.3557, "step": 7618 }, { "epoch": 1.33, "learning_rate": 1.8762470609483235e-05, "loss": 0.3862, "step": 7619 }, { "epoch": 1.33, "learning_rate": 1.8756160997821775e-05, "loss": 0.3656, "step": 7620 }, { "epoch": 1.33, "learning_rate": 1.8749852195983177e-05, "loss": 0.3532, "step": 7621 }, { "epoch": 1.33, "learning_rate": 1.8743544204551773e-05, "loss": 0.3666, "step": 7622 }, { "epoch": 1.33, "learning_rate": 1.8737237024111747e-05, "loss": 0.361, "step": 7623 }, { "epoch": 1.33, "learning_rate": 1.8730930655247276e-05, "loss": 0.3351, "step": 7624 }, { "epoch": 1.33, "learning_rate": 1.8724625098542426e-05, "loss": 0.3552, "step": 7625 }, { "epoch": 1.33, "learning_rate": 1.871832035458118e-05, "loss": 0.3582, "step": 7626 }, { "epoch": 1.33, "learning_rate": 1.8712016423947475e-05, "loss": 0.362, "step": 7627 }, { "epoch": 1.33, "learning_rate": 1.870571330722516e-05, "loss": 0.3513, "step": 7628 }, { "epoch": 1.33, "learning_rate": 1.869941100499801e-05, "loss": 0.3606, "step": 7629 }, { "epoch": 1.33, "learning_rate": 1.8693109517849707e-05, "loss": 0.3543, "step": 7630 }, { "epoch": 1.33, "learning_rate": 1.8686808846363873e-05, "loss": 0.3668, "step": 7631 }, { "epoch": 1.33, "learning_rate": 1.8680508991124074e-05, "loss": 0.3402, "step": 7632 }, { "epoch": 1.33, "learning_rate": 1.867420995271375e-05, "loss": 0.3709, "step": 7633 }, { "epoch": 1.33, "learning_rate": 1.8667911731716325e-05, "loss": 0.3652, "step": 7634 }, { "epoch": 1.33, "learning_rate": 1.8661614328715092e-05, "loss": 0.3728, "step": 7635 }, { "epoch": 1.33, "learning_rate": 1.8655317744293316e-05, "loss": 0.3547, "step": 7636 }, { "epoch": 1.33, "learning_rate": 1.864902197903415e-05, "loss": 0.3666, "step": 7637 }, { "epoch": 1.33, "learning_rate": 1.864272703352068e-05, "loss": 0.3512, "step": 7638 }, { "epoch": 1.33, "learning_rate": 1.8636432908335938e-05, "loss": 0.3467, "step": 7639 }, { "epoch": 1.33, "learning_rate": 1.8630139604062857e-05, "loss": 0.3385, "step": 7640 }, { "epoch": 1.33, "learning_rate": 1.8623847121284294e-05, "loss": 0.3695, "step": 7641 }, { "epoch": 1.33, "learning_rate": 1.861755546058305e-05, "loss": 0.3722, "step": 7642 }, { "epoch": 1.33, "learning_rate": 1.8611264622541814e-05, "loss": 0.3667, "step": 7643 }, { "epoch": 1.33, "learning_rate": 1.860497460774323e-05, "loss": 0.3572, "step": 7644 }, { "epoch": 1.33, "learning_rate": 1.8598685416769867e-05, "loss": 0.348, "step": 7645 }, { "epoch": 1.33, "learning_rate": 1.8592397050204202e-05, "loss": 0.3623, "step": 7646 }, { "epoch": 1.33, "learning_rate": 1.8586109508628634e-05, "loss": 0.3616, "step": 7647 }, { "epoch": 1.33, "learning_rate": 1.8579822792625502e-05, "loss": 0.348, "step": 7648 }, { "epoch": 1.33, "learning_rate": 1.8573536902777046e-05, "loss": 0.3649, "step": 7649 }, { "epoch": 1.33, "learning_rate": 1.856725183966545e-05, "loss": 0.357, "step": 7650 }, { "epoch": 1.33, "learning_rate": 1.8560967603872824e-05, "loss": 0.3602, "step": 7651 }, { "epoch": 1.33, "learning_rate": 1.8554684195981174e-05, "loss": 0.379, "step": 7652 }, { "epoch": 1.33, "learning_rate": 1.854840161657246e-05, "loss": 0.3754, "step": 7653 }, { "epoch": 1.33, "learning_rate": 1.8542119866228545e-05, "loss": 0.3355, "step": 7654 }, { "epoch": 1.33, "learning_rate": 1.8535838945531214e-05, "loss": 0.3599, "step": 7655 }, { "epoch": 1.33, "learning_rate": 1.85295588550622e-05, "loss": 0.3691, "step": 7656 }, { "epoch": 1.33, "learning_rate": 1.8523279595403137e-05, "loss": 0.3556, "step": 7657 }, { "epoch": 1.34, "learning_rate": 1.8517001167135584e-05, "loss": 0.3648, "step": 7658 }, { "epoch": 1.34, "learning_rate": 1.8510723570841028e-05, "loss": 0.3572, "step": 7659 }, { "epoch": 1.34, "learning_rate": 1.8504446807100868e-05, "loss": 0.3421, "step": 7660 }, { "epoch": 1.34, "learning_rate": 1.8498170876496453e-05, "loss": 0.3515, "step": 7661 }, { "epoch": 1.34, "learning_rate": 1.8491895779609024e-05, "loss": 0.368, "step": 7662 }, { "epoch": 1.34, "learning_rate": 1.848562151701977e-05, "loss": 0.3526, "step": 7663 }, { "epoch": 1.34, "learning_rate": 1.8479348089309777e-05, "loss": 0.3522, "step": 7664 }, { "epoch": 1.34, "learning_rate": 1.847307549706007e-05, "loss": 0.3551, "step": 7665 }, { "epoch": 1.34, "learning_rate": 1.8466803740851603e-05, "loss": 0.3562, "step": 7666 }, { "epoch": 1.34, "learning_rate": 1.846053282126523e-05, "loss": 0.3776, "step": 7667 }, { "epoch": 1.34, "learning_rate": 1.8454262738881754e-05, "loss": 0.3599, "step": 7668 }, { "epoch": 1.34, "learning_rate": 1.8447993494281883e-05, "loss": 0.3707, "step": 7669 }, { "epoch": 1.34, "learning_rate": 1.8441725088046258e-05, "loss": 0.3539, "step": 7670 }, { "epoch": 1.34, "learning_rate": 1.843545752075541e-05, "loss": 0.3586, "step": 7671 }, { "epoch": 1.34, "learning_rate": 1.842919079298985e-05, "loss": 0.3498, "step": 7672 }, { "epoch": 1.34, "learning_rate": 1.8422924905329965e-05, "loss": 0.3515, "step": 7673 }, { "epoch": 1.34, "learning_rate": 1.8416659858356077e-05, "loss": 0.3613, "step": 7674 }, { "epoch": 1.34, "learning_rate": 1.841039565264845e-05, "loss": 0.3462, "step": 7675 }, { "epoch": 1.34, "learning_rate": 1.8404132288787223e-05, "loss": 0.3682, "step": 7676 }, { "epoch": 1.34, "learning_rate": 1.8397869767352504e-05, "loss": 0.3529, "step": 7677 }, { "epoch": 1.34, "learning_rate": 1.8391608088924304e-05, "loss": 0.3451, "step": 7678 }, { "epoch": 1.34, "learning_rate": 1.838534725408256e-05, "loss": 0.3572, "step": 7679 }, { "epoch": 1.34, "learning_rate": 1.8379087263407124e-05, "loss": 0.3609, "step": 7680 }, { "epoch": 1.34, "learning_rate": 1.837282811747777e-05, "loss": 0.3611, "step": 7681 }, { "epoch": 1.34, "learning_rate": 1.83665698168742e-05, "loss": 0.3693, "step": 7682 }, { "epoch": 1.34, "learning_rate": 1.8360312362176035e-05, "loss": 0.3445, "step": 7683 }, { "epoch": 1.34, "learning_rate": 1.835405575396283e-05, "loss": 0.37, "step": 7684 }, { "epoch": 1.34, "learning_rate": 1.834779999281403e-05, "loss": 0.3758, "step": 7685 }, { "epoch": 1.34, "learning_rate": 1.8341545079309037e-05, "loss": 0.3391, "step": 7686 }, { "epoch": 1.34, "learning_rate": 1.833529101402715e-05, "loss": 0.3692, "step": 7687 }, { "epoch": 1.34, "learning_rate": 1.8329037797547583e-05, "loss": 0.3608, "step": 7688 }, { "epoch": 1.34, "learning_rate": 1.832278543044952e-05, "loss": 0.3528, "step": 7689 }, { "epoch": 1.34, "learning_rate": 1.8316533913312012e-05, "loss": 0.3702, "step": 7690 }, { "epoch": 1.34, "learning_rate": 1.8310283246714057e-05, "loss": 0.3562, "step": 7691 }, { "epoch": 1.34, "learning_rate": 1.830403343123457e-05, "loss": 0.3573, "step": 7692 }, { "epoch": 1.34, "learning_rate": 1.8297784467452373e-05, "loss": 0.3739, "step": 7693 }, { "epoch": 1.34, "learning_rate": 1.8291536355946244e-05, "loss": 0.3675, "step": 7694 }, { "epoch": 1.34, "learning_rate": 1.8285289097294845e-05, "loss": 0.3472, "step": 7695 }, { "epoch": 1.34, "learning_rate": 1.8279042692076788e-05, "loss": 0.3462, "step": 7696 }, { "epoch": 1.34, "learning_rate": 1.8272797140870572e-05, "loss": 0.367, "step": 7697 }, { "epoch": 1.34, "learning_rate": 1.8266552444254663e-05, "loss": 0.3444, "step": 7698 }, { "epoch": 1.34, "learning_rate": 1.8260308602807396e-05, "loss": 0.3649, "step": 7699 }, { "epoch": 1.34, "learning_rate": 1.8254065617107073e-05, "loss": 0.3793, "step": 7700 }, { "epoch": 1.34, "learning_rate": 1.8247823487731896e-05, "loss": 0.3618, "step": 7701 }, { "epoch": 1.34, "learning_rate": 1.824158221525998e-05, "loss": 0.3618, "step": 7702 }, { "epoch": 1.34, "learning_rate": 1.8235341800269366e-05, "loss": 0.3727, "step": 7703 }, { "epoch": 1.34, "learning_rate": 1.8229102243338024e-05, "loss": 0.3682, "step": 7704 }, { "epoch": 1.34, "learning_rate": 1.8222863545043845e-05, "loss": 0.3623, "step": 7705 }, { "epoch": 1.34, "learning_rate": 1.8216625705964634e-05, "loss": 0.3564, "step": 7706 }, { "epoch": 1.34, "learning_rate": 1.82103887266781e-05, "loss": 0.3591, "step": 7707 }, { "epoch": 1.34, "learning_rate": 1.8204152607761914e-05, "loss": 0.359, "step": 7708 }, { "epoch": 1.34, "learning_rate": 1.8197917349793637e-05, "loss": 0.3729, "step": 7709 }, { "epoch": 1.34, "learning_rate": 1.8191682953350732e-05, "loss": 0.3672, "step": 7710 }, { "epoch": 1.34, "learning_rate": 1.8185449419010636e-05, "loss": 0.3797, "step": 7711 }, { "epoch": 1.34, "learning_rate": 1.8179216747350665e-05, "loss": 0.3645, "step": 7712 }, { "epoch": 1.34, "learning_rate": 1.8172984938948063e-05, "loss": 0.3426, "step": 7713 }, { "epoch": 1.34, "learning_rate": 1.8166753994380002e-05, "loss": 0.3685, "step": 7714 }, { "epoch": 1.35, "learning_rate": 1.816052391422357e-05, "loss": 0.3495, "step": 7715 }, { "epoch": 1.35, "learning_rate": 1.8154294699055773e-05, "loss": 0.3543, "step": 7716 }, { "epoch": 1.35, "learning_rate": 1.814806634945354e-05, "loss": 0.3629, "step": 7717 }, { "epoch": 1.35, "learning_rate": 1.8141838865993726e-05, "loss": 0.3565, "step": 7718 }, { "epoch": 1.35, "learning_rate": 1.813561224925308e-05, "loss": 0.3611, "step": 7719 }, { "epoch": 1.35, "learning_rate": 1.812938649980831e-05, "loss": 0.3577, "step": 7720 }, { "epoch": 1.35, "learning_rate": 1.8123161618235995e-05, "loss": 0.3544, "step": 7721 }, { "epoch": 1.35, "learning_rate": 1.8116937605112684e-05, "loss": 0.3544, "step": 7722 }, { "epoch": 1.35, "learning_rate": 1.8110714461014816e-05, "loss": 0.3509, "step": 7723 }, { "epoch": 1.35, "learning_rate": 1.8104492186518756e-05, "loss": 0.3563, "step": 7724 }, { "epoch": 1.35, "learning_rate": 1.809827078220078e-05, "loss": 0.3634, "step": 7725 }, { "epoch": 1.35, "learning_rate": 1.8092050248637112e-05, "loss": 0.3672, "step": 7726 }, { "epoch": 1.35, "learning_rate": 1.808583058640386e-05, "loss": 0.3661, "step": 7727 }, { "epoch": 1.35, "learning_rate": 1.8079611796077084e-05, "loss": 0.3573, "step": 7728 }, { "epoch": 1.35, "learning_rate": 1.8073393878232714e-05, "loss": 0.39, "step": 7729 }, { "epoch": 1.35, "learning_rate": 1.8067176833446657e-05, "loss": 0.3674, "step": 7730 }, { "epoch": 1.35, "learning_rate": 1.8060960662294712e-05, "loss": 0.3704, "step": 7731 }, { "epoch": 1.35, "learning_rate": 1.8054745365352573e-05, "loss": 0.3601, "step": 7732 }, { "epoch": 1.35, "learning_rate": 1.8048530943195917e-05, "loss": 0.3527, "step": 7733 }, { "epoch": 1.35, "learning_rate": 1.804231739640028e-05, "loss": 0.3582, "step": 7734 }, { "epoch": 1.35, "learning_rate": 1.803610472554113e-05, "loss": 0.349, "step": 7735 }, { "epoch": 1.35, "learning_rate": 1.8029892931193887e-05, "loss": 0.3729, "step": 7736 }, { "epoch": 1.35, "learning_rate": 1.802368201393385e-05, "loss": 0.3604, "step": 7737 }, { "epoch": 1.35, "learning_rate": 1.801747197433624e-05, "loss": 0.3679, "step": 7738 }, { "epoch": 1.35, "learning_rate": 1.8011262812976227e-05, "loss": 0.3476, "step": 7739 }, { "epoch": 1.35, "learning_rate": 1.8005054530428883e-05, "loss": 0.3676, "step": 7740 }, { "epoch": 1.35, "learning_rate": 1.7998847127269187e-05, "loss": 0.3663, "step": 7741 }, { "epoch": 1.35, "learning_rate": 1.7992640604072037e-05, "loss": 0.3585, "step": 7742 }, { "epoch": 1.35, "learning_rate": 1.7986434961412272e-05, "loss": 0.3459, "step": 7743 }, { "epoch": 1.35, "learning_rate": 1.7980230199864644e-05, "loss": 0.3849, "step": 7744 }, { "epoch": 1.35, "learning_rate": 1.797402632000381e-05, "loss": 0.3674, "step": 7745 }, { "epoch": 1.35, "learning_rate": 1.7967823322404335e-05, "loss": 0.3529, "step": 7746 }, { "epoch": 1.35, "learning_rate": 1.796162120764074e-05, "loss": 0.3557, "step": 7747 }, { "epoch": 1.35, "learning_rate": 1.795541997628743e-05, "loss": 0.3718, "step": 7748 }, { "epoch": 1.35, "learning_rate": 1.7949219628918734e-05, "loss": 0.356, "step": 7749 }, { "epoch": 1.35, "learning_rate": 1.7943020166108926e-05, "loss": 0.3628, "step": 7750 }, { "epoch": 1.35, "learning_rate": 1.7936821588432167e-05, "loss": 0.3669, "step": 7751 }, { "epoch": 1.35, "learning_rate": 1.7930623896462545e-05, "loss": 0.3484, "step": 7752 }, { "epoch": 1.35, "learning_rate": 1.7924427090774076e-05, "loss": 0.371, "step": 7753 }, { "epoch": 1.35, "learning_rate": 1.791823117194067e-05, "loss": 0.3731, "step": 7754 }, { "epoch": 1.35, "learning_rate": 1.7912036140536192e-05, "loss": 0.3529, "step": 7755 }, { "epoch": 1.35, "learning_rate": 1.790584199713438e-05, "loss": 0.3712, "step": 7756 }, { "epoch": 1.35, "learning_rate": 1.789964874230894e-05, "loss": 0.3682, "step": 7757 }, { "epoch": 1.35, "learning_rate": 1.789345637663345e-05, "loss": 0.3477, "step": 7758 }, { "epoch": 1.35, "learning_rate": 1.788726490068142e-05, "loss": 0.367, "step": 7759 }, { "epoch": 1.35, "learning_rate": 1.7881074315026298e-05, "loss": 0.3699, "step": 7760 }, { "epoch": 1.35, "learning_rate": 1.7874884620241428e-05, "loss": 0.3435, "step": 7761 }, { "epoch": 1.35, "learning_rate": 1.7868695816900082e-05, "loss": 0.359, "step": 7762 }, { "epoch": 1.35, "learning_rate": 1.7862507905575426e-05, "loss": 0.3721, "step": 7763 }, { "epoch": 1.35, "learning_rate": 1.7856320886840584e-05, "loss": 0.3725, "step": 7764 }, { "epoch": 1.35, "learning_rate": 1.785013476126856e-05, "loss": 0.3419, "step": 7765 }, { "epoch": 1.35, "learning_rate": 1.7843949529432297e-05, "loss": 0.3663, "step": 7766 }, { "epoch": 1.35, "learning_rate": 1.783776519190466e-05, "loss": 0.3616, "step": 7767 }, { "epoch": 1.35, "learning_rate": 1.7831581749258408e-05, "loss": 0.3612, "step": 7768 }, { "epoch": 1.35, "learning_rate": 1.7825399202066216e-05, "loss": 0.3498, "step": 7769 }, { "epoch": 1.35, "learning_rate": 1.781921755090072e-05, "loss": 0.3577, "step": 7770 }, { "epoch": 1.35, "learning_rate": 1.7813036796334416e-05, "loss": 0.3576, "step": 7771 }, { "epoch": 1.36, "learning_rate": 1.780685693893976e-05, "loss": 0.3648, "step": 7772 }, { "epoch": 1.36, "learning_rate": 1.780067797928909e-05, "loss": 0.3598, "step": 7773 }, { "epoch": 1.36, "learning_rate": 1.7794499917954705e-05, "loss": 0.3592, "step": 7774 }, { "epoch": 1.36, "learning_rate": 1.7788322755508775e-05, "loss": 0.3475, "step": 7775 }, { "epoch": 1.36, "learning_rate": 1.77821464925234e-05, "loss": 0.3556, "step": 7776 }, { "epoch": 1.36, "learning_rate": 1.7775971129570618e-05, "loss": 0.3578, "step": 7777 }, { "epoch": 1.36, "learning_rate": 1.7769796667222377e-05, "loss": 0.3592, "step": 7778 }, { "epoch": 1.36, "learning_rate": 1.776362310605052e-05, "loss": 0.3653, "step": 7779 }, { "epoch": 1.36, "learning_rate": 1.7757450446626813e-05, "loss": 0.3685, "step": 7780 }, { "epoch": 1.36, "learning_rate": 1.7751278689522966e-05, "loss": 0.353, "step": 7781 }, { "epoch": 1.36, "learning_rate": 1.774510783531056e-05, "loss": 0.3537, "step": 7782 }, { "epoch": 1.36, "learning_rate": 1.773893788456114e-05, "loss": 0.3593, "step": 7783 }, { "epoch": 1.36, "learning_rate": 1.773276883784613e-05, "loss": 0.3787, "step": 7784 }, { "epoch": 1.36, "learning_rate": 1.77266006957369e-05, "loss": 0.3782, "step": 7785 }, { "epoch": 1.36, "learning_rate": 1.7720433458804707e-05, "loss": 0.362, "step": 7786 }, { "epoch": 1.36, "learning_rate": 1.7714267127620727e-05, "loss": 0.3706, "step": 7787 }, { "epoch": 1.36, "learning_rate": 1.77081017027561e-05, "loss": 0.3622, "step": 7788 }, { "epoch": 1.36, "learning_rate": 1.7701937184781818e-05, "loss": 0.3525, "step": 7789 }, { "epoch": 1.36, "learning_rate": 1.769577357426882e-05, "loss": 0.3748, "step": 7790 }, { "epoch": 1.36, "learning_rate": 1.7689610871787964e-05, "loss": 0.3522, "step": 7791 }, { "epoch": 1.36, "learning_rate": 1.768344907791002e-05, "loss": 0.3497, "step": 7792 }, { "epoch": 1.36, "learning_rate": 1.7677288193205655e-05, "loss": 0.3555, "step": 7793 }, { "epoch": 1.36, "learning_rate": 1.7671128218245477e-05, "loss": 0.3793, "step": 7794 }, { "epoch": 1.36, "learning_rate": 1.7664969153600002e-05, "loss": 0.3695, "step": 7795 }, { "epoch": 1.36, "learning_rate": 1.765881099983967e-05, "loss": 0.3537, "step": 7796 }, { "epoch": 1.36, "learning_rate": 1.7652653757534805e-05, "loss": 0.3648, "step": 7797 }, { "epoch": 1.36, "learning_rate": 1.7646497427255677e-05, "loss": 0.3704, "step": 7798 }, { "epoch": 1.36, "learning_rate": 1.764034200957248e-05, "loss": 0.3513, "step": 7799 }, { "epoch": 1.36, "learning_rate": 1.763418750505529e-05, "loss": 0.362, "step": 7800 }, { "epoch": 1.36, "learning_rate": 1.762803391427411e-05, "loss": 0.3564, "step": 7801 }, { "epoch": 1.36, "learning_rate": 1.7621881237798882e-05, "loss": 0.3644, "step": 7802 }, { "epoch": 1.36, "learning_rate": 1.7615729476199437e-05, "loss": 0.3755, "step": 7803 }, { "epoch": 1.36, "learning_rate": 1.7609578630045503e-05, "loss": 0.3479, "step": 7804 }, { "epoch": 1.36, "learning_rate": 1.7603428699906792e-05, "loss": 0.3653, "step": 7805 }, { "epoch": 1.36, "learning_rate": 1.7597279686352868e-05, "loss": 0.3737, "step": 7806 }, { "epoch": 1.36, "learning_rate": 1.759113158995323e-05, "loss": 0.3603, "step": 7807 }, { "epoch": 1.36, "learning_rate": 1.7584984411277293e-05, "loss": 0.3504, "step": 7808 }, { "epoch": 1.36, "learning_rate": 1.757883815089439e-05, "loss": 0.3788, "step": 7809 }, { "epoch": 1.36, "learning_rate": 1.7572692809373757e-05, "loss": 0.3633, "step": 7810 }, { "epoch": 1.36, "learning_rate": 1.7566548387284565e-05, "loss": 0.3752, "step": 7811 }, { "epoch": 1.36, "learning_rate": 1.756040488519588e-05, "loss": 0.3628, "step": 7812 }, { "epoch": 1.36, "learning_rate": 1.75542623036767e-05, "loss": 0.3515, "step": 7813 }, { "epoch": 1.36, "learning_rate": 1.7548120643295916e-05, "loss": 0.3717, "step": 7814 }, { "epoch": 1.36, "learning_rate": 1.7541979904622346e-05, "loss": 0.3741, "step": 7815 }, { "epoch": 1.36, "learning_rate": 1.7535840088224748e-05, "loss": 0.3477, "step": 7816 }, { "epoch": 1.36, "learning_rate": 1.7529701194671745e-05, "loss": 0.3576, "step": 7817 }, { "epoch": 1.36, "learning_rate": 1.75235632245319e-05, "loss": 0.3614, "step": 7818 }, { "epoch": 1.36, "learning_rate": 1.7517426178373704e-05, "loss": 0.3485, "step": 7819 }, { "epoch": 1.36, "learning_rate": 1.7511290056765528e-05, "loss": 0.351, "step": 7820 }, { "epoch": 1.36, "learning_rate": 1.7505154860275698e-05, "loss": 0.3482, "step": 7821 }, { "epoch": 1.36, "learning_rate": 1.7499020589472418e-05, "loss": 0.3559, "step": 7822 }, { "epoch": 1.36, "learning_rate": 1.749288724492384e-05, "loss": 0.375, "step": 7823 }, { "epoch": 1.36, "learning_rate": 1.7486754827197998e-05, "loss": 0.3552, "step": 7824 }, { "epoch": 1.36, "learning_rate": 1.748062333686285e-05, "loss": 0.3485, "step": 7825 }, { "epoch": 1.36, "learning_rate": 1.7474492774486285e-05, "loss": 0.3587, "step": 7826 }, { "epoch": 1.36, "learning_rate": 1.746836314063609e-05, "loss": 0.3564, "step": 7827 }, { "epoch": 1.36, "learning_rate": 1.7462234435879966e-05, "loss": 0.3741, "step": 7828 }, { "epoch": 1.36, "learning_rate": 1.745610666078554e-05, "loss": 0.3514, "step": 7829 }, { "epoch": 1.37, "learning_rate": 1.7449979815920343e-05, "loss": 0.3624, "step": 7830 }, { "epoch": 1.37, "learning_rate": 1.74438539018518e-05, "loss": 0.3601, "step": 7831 }, { "epoch": 1.37, "learning_rate": 1.7437728919147293e-05, "loss": 0.3493, "step": 7832 }, { "epoch": 1.37, "learning_rate": 1.74316048683741e-05, "loss": 0.3758, "step": 7833 }, { "epoch": 1.37, "learning_rate": 1.7425481750099397e-05, "loss": 0.3705, "step": 7834 }, { "epoch": 1.37, "learning_rate": 1.7419359564890283e-05, "loss": 0.3757, "step": 7835 }, { "epoch": 1.37, "learning_rate": 1.7413238313313786e-05, "loss": 0.3619, "step": 7836 }, { "epoch": 1.37, "learning_rate": 1.740711799593681e-05, "loss": 0.3578, "step": 7837 }, { "epoch": 1.37, "learning_rate": 1.7400998613326226e-05, "loss": 0.3388, "step": 7838 }, { "epoch": 1.37, "learning_rate": 1.7394880166048765e-05, "loss": 0.3827, "step": 7839 }, { "epoch": 1.37, "learning_rate": 1.738876265467112e-05, "loss": 0.3829, "step": 7840 }, { "epoch": 1.37, "learning_rate": 1.7382646079759852e-05, "loss": 0.3732, "step": 7841 }, { "epoch": 1.37, "learning_rate": 1.737653044188146e-05, "loss": 0.3773, "step": 7842 }, { "epoch": 1.37, "learning_rate": 1.7370415741602346e-05, "loss": 0.3624, "step": 7843 }, { "epoch": 1.37, "learning_rate": 1.736430197948886e-05, "loss": 0.3574, "step": 7844 }, { "epoch": 1.37, "learning_rate": 1.735818915610721e-05, "loss": 0.377, "step": 7845 }, { "epoch": 1.37, "learning_rate": 1.735207727202355e-05, "loss": 0.3537, "step": 7846 }, { "epoch": 1.37, "learning_rate": 1.7345966327803946e-05, "loss": 0.3678, "step": 7847 }, { "epoch": 1.37, "learning_rate": 1.733985632401436e-05, "loss": 0.3413, "step": 7848 }, { "epoch": 1.37, "learning_rate": 1.733374726122069e-05, "loss": 0.3703, "step": 7849 }, { "epoch": 1.37, "learning_rate": 1.7327639139988732e-05, "loss": 0.3669, "step": 7850 }, { "epoch": 1.37, "learning_rate": 1.7321531960884206e-05, "loss": 0.3629, "step": 7851 }, { "epoch": 1.37, "learning_rate": 1.7315425724472713e-05, "loss": 0.3572, "step": 7852 }, { "epoch": 1.37, "learning_rate": 1.7309320431319817e-05, "loss": 0.3669, "step": 7853 }, { "epoch": 1.37, "learning_rate": 1.730321608199095e-05, "loss": 0.3467, "step": 7854 }, { "epoch": 1.37, "learning_rate": 1.729711267705149e-05, "loss": 0.355, "step": 7855 }, { "epoch": 1.37, "learning_rate": 1.72910102170667e-05, "loss": 0.3623, "step": 7856 }, { "epoch": 1.37, "learning_rate": 1.7284908702601778e-05, "loss": 0.3624, "step": 7857 }, { "epoch": 1.37, "learning_rate": 1.727880813422182e-05, "loss": 0.3613, "step": 7858 }, { "epoch": 1.37, "learning_rate": 1.727270851249182e-05, "loss": 0.3561, "step": 7859 }, { "epoch": 1.37, "learning_rate": 1.7266609837976735e-05, "loss": 0.3575, "step": 7860 }, { "epoch": 1.37, "learning_rate": 1.726051211124139e-05, "loss": 0.3546, "step": 7861 }, { "epoch": 1.37, "learning_rate": 1.725441533285053e-05, "loss": 0.3603, "step": 7862 }, { "epoch": 1.37, "learning_rate": 1.724831950336881e-05, "loss": 0.3656, "step": 7863 }, { "epoch": 1.37, "learning_rate": 1.7242224623360824e-05, "loss": 0.3618, "step": 7864 }, { "epoch": 1.37, "learning_rate": 1.7236130693391038e-05, "loss": 0.3604, "step": 7865 }, { "epoch": 1.37, "learning_rate": 1.7230037714023862e-05, "loss": 0.372, "step": 7866 }, { "epoch": 1.37, "learning_rate": 1.7223945685823607e-05, "loss": 0.3719, "step": 7867 }, { "epoch": 1.37, "learning_rate": 1.721785460935449e-05, "loss": 0.3525, "step": 7868 }, { "epoch": 1.37, "learning_rate": 1.7211764485180633e-05, "loss": 0.367, "step": 7869 }, { "epoch": 1.37, "learning_rate": 1.72056753138661e-05, "loss": 0.3633, "step": 7870 }, { "epoch": 1.37, "learning_rate": 1.7199587095974845e-05, "loss": 0.3536, "step": 7871 }, { "epoch": 1.37, "learning_rate": 1.719349983207073e-05, "loss": 0.3306, "step": 7872 }, { "epoch": 1.37, "learning_rate": 1.7187413522717533e-05, "loss": 0.3599, "step": 7873 }, { "epoch": 1.37, "learning_rate": 1.718132816847896e-05, "loss": 0.3652, "step": 7874 }, { "epoch": 1.37, "learning_rate": 1.717524376991861e-05, "loss": 0.3558, "step": 7875 }, { "epoch": 1.37, "learning_rate": 1.716916032759998e-05, "loss": 0.339, "step": 7876 }, { "epoch": 1.37, "learning_rate": 1.7163077842086507e-05, "loss": 0.3616, "step": 7877 }, { "epoch": 1.37, "learning_rate": 1.7156996313941548e-05, "loss": 0.3587, "step": 7878 }, { "epoch": 1.37, "learning_rate": 1.7150915743728334e-05, "loss": 0.3625, "step": 7879 }, { "epoch": 1.37, "learning_rate": 1.7144836132010018e-05, "loss": 0.3643, "step": 7880 }, { "epoch": 1.37, "learning_rate": 1.7138757479349682e-05, "loss": 0.3621, "step": 7881 }, { "epoch": 1.37, "learning_rate": 1.713267978631032e-05, "loss": 0.3802, "step": 7882 }, { "epoch": 1.37, "learning_rate": 1.7126603053454804e-05, "loss": 0.3502, "step": 7883 }, { "epoch": 1.37, "learning_rate": 1.712052728134596e-05, "loss": 0.3594, "step": 7884 }, { "epoch": 1.37, "learning_rate": 1.7114452470546496e-05, "loss": 0.3511, "step": 7885 }, { "epoch": 1.37, "learning_rate": 1.7108378621619025e-05, "loss": 0.3646, "step": 7886 }, { "epoch": 1.38, "learning_rate": 1.71023057351261e-05, "loss": 0.3468, "step": 7887 }, { "epoch": 1.38, "learning_rate": 1.709623381163017e-05, "loss": 0.3692, "step": 7888 }, { "epoch": 1.38, "learning_rate": 1.7090162851693597e-05, "loss": 0.3628, "step": 7889 }, { "epoch": 1.38, "learning_rate": 1.7084092855878637e-05, "loss": 0.3584, "step": 7890 }, { "epoch": 1.38, "learning_rate": 1.707802382474749e-05, "loss": 0.3683, "step": 7891 }, { "epoch": 1.38, "learning_rate": 1.707195575886223e-05, "loss": 0.3775, "step": 7892 }, { "epoch": 1.38, "learning_rate": 1.7065888658784868e-05, "loss": 0.3533, "step": 7893 }, { "epoch": 1.38, "learning_rate": 1.7059822525077312e-05, "loss": 0.3581, "step": 7894 }, { "epoch": 1.38, "learning_rate": 1.7053757358301403e-05, "loss": 0.3582, "step": 7895 }, { "epoch": 1.38, "learning_rate": 1.704769315901886e-05, "loss": 0.3628, "step": 7896 }, { "epoch": 1.38, "learning_rate": 1.704162992779132e-05, "loss": 0.3566, "step": 7897 }, { "epoch": 1.38, "learning_rate": 1.7035567665180347e-05, "loss": 0.3483, "step": 7898 }, { "epoch": 1.38, "learning_rate": 1.7029506371747412e-05, "loss": 0.358, "step": 7899 }, { "epoch": 1.38, "learning_rate": 1.7023446048053876e-05, "loss": 0.359, "step": 7900 }, { "epoch": 1.38, "learning_rate": 1.7017386694661043e-05, "loss": 0.3567, "step": 7901 }, { "epoch": 1.38, "learning_rate": 1.7011328312130097e-05, "loss": 0.3651, "step": 7902 }, { "epoch": 1.38, "learning_rate": 1.7005270901022135e-05, "loss": 0.3603, "step": 7903 }, { "epoch": 1.38, "learning_rate": 1.6999214461898182e-05, "loss": 0.3453, "step": 7904 }, { "epoch": 1.38, "learning_rate": 1.699315899531917e-05, "loss": 0.3605, "step": 7905 }, { "epoch": 1.38, "learning_rate": 1.698710450184593e-05, "loss": 0.3623, "step": 7906 }, { "epoch": 1.38, "learning_rate": 1.6981050982039197e-05, "loss": 0.3534, "step": 7907 }, { "epoch": 1.38, "learning_rate": 1.6974998436459642e-05, "loss": 0.357, "step": 7908 }, { "epoch": 1.38, "learning_rate": 1.6968946865667816e-05, "loss": 0.3697, "step": 7909 }, { "epoch": 1.38, "learning_rate": 1.6962896270224207e-05, "loss": 0.3516, "step": 7910 }, { "epoch": 1.38, "learning_rate": 1.6956846650689188e-05, "loss": 0.3432, "step": 7911 }, { "epoch": 1.38, "learning_rate": 1.6950798007623058e-05, "loss": 0.3527, "step": 7912 }, { "epoch": 1.38, "learning_rate": 1.6944750341586023e-05, "loss": 0.3661, "step": 7913 }, { "epoch": 1.38, "learning_rate": 1.6938703653138187e-05, "loss": 0.3886, "step": 7914 }, { "epoch": 1.38, "learning_rate": 1.693265794283958e-05, "loss": 0.3559, "step": 7915 }, { "epoch": 1.38, "learning_rate": 1.6926613211250143e-05, "loss": 0.3596, "step": 7916 }, { "epoch": 1.38, "learning_rate": 1.69205694589297e-05, "loss": 0.3388, "step": 7917 }, { "epoch": 1.38, "learning_rate": 1.6914526686438004e-05, "loss": 0.3646, "step": 7918 }, { "epoch": 1.38, "learning_rate": 1.6908484894334725e-05, "loss": 0.3584, "step": 7919 }, { "epoch": 1.38, "learning_rate": 1.690244408317942e-05, "loss": 0.3678, "step": 7920 }, { "epoch": 1.38, "learning_rate": 1.689640425353157e-05, "loss": 0.3553, "step": 7921 }, { "epoch": 1.38, "learning_rate": 1.6890365405950577e-05, "loss": 0.3607, "step": 7922 }, { "epoch": 1.38, "learning_rate": 1.6884327540995726e-05, "loss": 0.3832, "step": 7923 }, { "epoch": 1.38, "learning_rate": 1.687829065922621e-05, "loss": 0.3537, "step": 7924 }, { "epoch": 1.38, "learning_rate": 1.687225476120116e-05, "loss": 0.3506, "step": 7925 }, { "epoch": 1.38, "learning_rate": 1.6866219847479605e-05, "loss": 0.35, "step": 7926 }, { "epoch": 1.38, "learning_rate": 1.686018591862046e-05, "loss": 0.3641, "step": 7927 }, { "epoch": 1.38, "learning_rate": 1.685415297518257e-05, "loss": 0.3718, "step": 7928 }, { "epoch": 1.38, "learning_rate": 1.6848121017724693e-05, "loss": 0.3592, "step": 7929 }, { "epoch": 1.38, "learning_rate": 1.684209004680548e-05, "loss": 0.3546, "step": 7930 }, { "epoch": 1.38, "learning_rate": 1.683606006298349e-05, "loss": 0.3526, "step": 7931 }, { "epoch": 1.38, "learning_rate": 1.683003106681721e-05, "loss": 0.3636, "step": 7932 }, { "epoch": 1.38, "learning_rate": 1.6824003058865027e-05, "loss": 0.3505, "step": 7933 }, { "epoch": 1.38, "learning_rate": 1.681797603968523e-05, "loss": 0.3489, "step": 7934 }, { "epoch": 1.38, "learning_rate": 1.6811950009836e-05, "loss": 0.3534, "step": 7935 }, { "epoch": 1.38, "learning_rate": 1.6805924969875476e-05, "loss": 0.3608, "step": 7936 }, { "epoch": 1.38, "learning_rate": 1.6799900920361665e-05, "loss": 0.3591, "step": 7937 }, { "epoch": 1.38, "learning_rate": 1.6793877861852482e-05, "loss": 0.3751, "step": 7938 }, { "epoch": 1.38, "learning_rate": 1.6787855794905783e-05, "loss": 0.3685, "step": 7939 }, { "epoch": 1.38, "learning_rate": 1.6781834720079292e-05, "loss": 0.3466, "step": 7940 }, { "epoch": 1.38, "learning_rate": 1.6775814637930658e-05, "loss": 0.3622, "step": 7941 }, { "epoch": 1.38, "learning_rate": 1.6769795549017442e-05, "loss": 0.3524, "step": 7942 }, { "epoch": 1.38, "learning_rate": 1.6763777453897127e-05, "loss": 0.3696, "step": 7943 }, { "epoch": 1.39, "learning_rate": 1.675776035312707e-05, "loss": 0.3458, "step": 7944 }, { "epoch": 1.39, "learning_rate": 1.675174424726455e-05, "loss": 0.3787, "step": 7945 }, { "epoch": 1.39, "learning_rate": 1.6745729136866776e-05, "loss": 0.357, "step": 7946 }, { "epoch": 1.39, "learning_rate": 1.673971502249083e-05, "loss": 0.3457, "step": 7947 }, { "epoch": 1.39, "learning_rate": 1.673370190469372e-05, "loss": 0.358, "step": 7948 }, { "epoch": 1.39, "learning_rate": 1.6727689784032357e-05, "loss": 0.3738, "step": 7949 }, { "epoch": 1.39, "learning_rate": 1.6721678661063573e-05, "loss": 0.3603, "step": 7950 }, { "epoch": 1.39, "learning_rate": 1.671566853634409e-05, "loss": 0.3733, "step": 7951 }, { "epoch": 1.39, "learning_rate": 1.6709659410430538e-05, "loss": 0.3541, "step": 7952 }, { "epoch": 1.39, "learning_rate": 1.6703651283879464e-05, "loss": 0.3633, "step": 7953 }, { "epoch": 1.39, "learning_rate": 1.669764415724733e-05, "loss": 0.3588, "step": 7954 }, { "epoch": 1.39, "learning_rate": 1.669163803109049e-05, "loss": 0.351, "step": 7955 }, { "epoch": 1.39, "learning_rate": 1.668563290596519e-05, "loss": 0.3654, "step": 7956 }, { "epoch": 1.39, "learning_rate": 1.6679628782427632e-05, "loss": 0.3732, "step": 7957 }, { "epoch": 1.39, "learning_rate": 1.667362566103388e-05, "loss": 0.3709, "step": 7958 }, { "epoch": 1.39, "learning_rate": 1.666762354233991e-05, "loss": 0.3562, "step": 7959 }, { "epoch": 1.39, "learning_rate": 1.6661622426901648e-05, "loss": 0.3559, "step": 7960 }, { "epoch": 1.39, "learning_rate": 1.6655622315274876e-05, "loss": 0.3498, "step": 7961 }, { "epoch": 1.39, "learning_rate": 1.6649623208015303e-05, "loss": 0.3583, "step": 7962 }, { "epoch": 1.39, "learning_rate": 1.6643625105678547e-05, "loss": 0.3523, "step": 7963 }, { "epoch": 1.39, "learning_rate": 1.6637628008820132e-05, "loss": 0.3435, "step": 7964 }, { "epoch": 1.39, "learning_rate": 1.6631631917995487e-05, "loss": 0.3451, "step": 7965 }, { "epoch": 1.39, "learning_rate": 1.662563683375994e-05, "loss": 0.3593, "step": 7966 }, { "epoch": 1.39, "learning_rate": 1.6619642756668756e-05, "loss": 0.3538, "step": 7967 }, { "epoch": 1.39, "learning_rate": 1.6613649687277065e-05, "loss": 0.3597, "step": 7968 }, { "epoch": 1.39, "learning_rate": 1.6607657626139927e-05, "loss": 0.3667, "step": 7969 }, { "epoch": 1.39, "learning_rate": 1.66016665738123e-05, "loss": 0.3509, "step": 7970 }, { "epoch": 1.39, "learning_rate": 1.6595676530849076e-05, "loss": 0.3612, "step": 7971 }, { "epoch": 1.39, "learning_rate": 1.658968749780501e-05, "loss": 0.3354, "step": 7972 }, { "epoch": 1.39, "learning_rate": 1.6583699475234783e-05, "loss": 0.3748, "step": 7973 }, { "epoch": 1.39, "learning_rate": 1.6577712463693003e-05, "loss": 0.3552, "step": 7974 }, { "epoch": 1.39, "learning_rate": 1.6571726463734144e-05, "loss": 0.3357, "step": 7975 }, { "epoch": 1.39, "learning_rate": 1.6565741475912622e-05, "loss": 0.3742, "step": 7976 }, { "epoch": 1.39, "learning_rate": 1.6559757500782747e-05, "loss": 0.3593, "step": 7977 }, { "epoch": 1.39, "learning_rate": 1.6553774538898725e-05, "loss": 0.3738, "step": 7978 }, { "epoch": 1.39, "learning_rate": 1.6547792590814672e-05, "loss": 0.3646, "step": 7979 }, { "epoch": 1.39, "learning_rate": 1.654181165708463e-05, "loss": 0.3619, "step": 7980 }, { "epoch": 1.39, "learning_rate": 1.6535831738262512e-05, "loss": 0.3778, "step": 7981 }, { "epoch": 1.39, "learning_rate": 1.6529852834902183e-05, "loss": 0.3488, "step": 7982 }, { "epoch": 1.39, "learning_rate": 1.6523874947557355e-05, "loss": 0.3731, "step": 7983 }, { "epoch": 1.39, "learning_rate": 1.6517898076781708e-05, "loss": 0.3671, "step": 7984 }, { "epoch": 1.39, "learning_rate": 1.6511922223128784e-05, "loss": 0.3509, "step": 7985 }, { "epoch": 1.39, "learning_rate": 1.6505947387152044e-05, "loss": 0.341, "step": 7986 }, { "epoch": 1.39, "learning_rate": 1.649997356940485e-05, "loss": 0.3716, "step": 7987 }, { "epoch": 1.39, "learning_rate": 1.6494000770440493e-05, "loss": 0.3628, "step": 7988 }, { "epoch": 1.39, "learning_rate": 1.648802899081215e-05, "loss": 0.3473, "step": 7989 }, { "epoch": 1.39, "learning_rate": 1.648205823107288e-05, "loss": 0.3628, "step": 7990 }, { "epoch": 1.39, "learning_rate": 1.6476088491775705e-05, "loss": 0.3763, "step": 7991 }, { "epoch": 1.39, "learning_rate": 1.64701197734735e-05, "loss": 0.3402, "step": 7992 }, { "epoch": 1.39, "learning_rate": 1.646415207671908e-05, "loss": 0.3479, "step": 7993 }, { "epoch": 1.39, "learning_rate": 1.645818540206514e-05, "loss": 0.3706, "step": 7994 }, { "epoch": 1.39, "learning_rate": 1.6452219750064307e-05, "loss": 0.3613, "step": 7995 }, { "epoch": 1.39, "learning_rate": 1.6446255121269083e-05, "loss": 0.3458, "step": 7996 }, { "epoch": 1.39, "learning_rate": 1.644029151623188e-05, "loss": 0.3674, "step": 7997 }, { "epoch": 1.39, "learning_rate": 1.643432893550506e-05, "loss": 0.3585, "step": 7998 }, { "epoch": 1.39, "learning_rate": 1.642836737964084e-05, "loss": 0.3554, "step": 7999 }, { "epoch": 1.39, "learning_rate": 1.6422406849191343e-05, "loss": 0.3743, "step": 8000 }, { "epoch": 1.39, "learning_rate": 1.6416447344708636e-05, "loss": 0.3616, "step": 8001 }, { "epoch": 1.4, "learning_rate": 1.641048886674465e-05, "loss": 0.3562, "step": 8002 }, { "epoch": 1.4, "learning_rate": 1.640453141585124e-05, "loss": 0.3706, "step": 8003 }, { "epoch": 1.4, "learning_rate": 1.639857499258016e-05, "loss": 0.3659, "step": 8004 }, { "epoch": 1.4, "learning_rate": 1.639261959748309e-05, "loss": 0.3498, "step": 8005 }, { "epoch": 1.4, "learning_rate": 1.6386665231111582e-05, "loss": 0.3529, "step": 8006 }, { "epoch": 1.4, "learning_rate": 1.638071189401711e-05, "loss": 0.3648, "step": 8007 }, { "epoch": 1.4, "learning_rate": 1.6374759586751048e-05, "loss": 0.3652, "step": 8008 }, { "epoch": 1.4, "learning_rate": 1.6368808309864693e-05, "loss": 0.3561, "step": 8009 }, { "epoch": 1.4, "learning_rate": 1.636285806390922e-05, "loss": 0.3611, "step": 8010 }, { "epoch": 1.4, "learning_rate": 1.635690884943571e-05, "loss": 0.3493, "step": 8011 }, { "epoch": 1.4, "learning_rate": 1.6350960666995173e-05, "loss": 0.3619, "step": 8012 }, { "epoch": 1.4, "learning_rate": 1.6345013517138503e-05, "loss": 0.3487, "step": 8013 }, { "epoch": 1.4, "learning_rate": 1.633906740041649e-05, "loss": 0.3408, "step": 8014 }, { "epoch": 1.4, "learning_rate": 1.6333122317379865e-05, "loss": 0.3416, "step": 8015 }, { "epoch": 1.4, "learning_rate": 1.632717826857923e-05, "loss": 0.3506, "step": 8016 }, { "epoch": 1.4, "learning_rate": 1.6321235254565095e-05, "loss": 0.3645, "step": 8017 }, { "epoch": 1.4, "learning_rate": 1.6315293275887898e-05, "loss": 0.3656, "step": 8018 }, { "epoch": 1.4, "learning_rate": 1.630935233309794e-05, "loss": 0.3674, "step": 8019 }, { "epoch": 1.4, "learning_rate": 1.6303412426745475e-05, "loss": 0.3693, "step": 8020 }, { "epoch": 1.4, "learning_rate": 1.6297473557380606e-05, "loss": 0.3602, "step": 8021 }, { "epoch": 1.4, "learning_rate": 1.6291535725553397e-05, "loss": 0.3572, "step": 8022 }, { "epoch": 1.4, "learning_rate": 1.6285598931813783e-05, "loss": 0.3549, "step": 8023 }, { "epoch": 1.4, "learning_rate": 1.6279663176711593e-05, "loss": 0.3534, "step": 8024 }, { "epoch": 1.4, "learning_rate": 1.6273728460796585e-05, "loss": 0.354, "step": 8025 }, { "epoch": 1.4, "learning_rate": 1.6267794784618424e-05, "loss": 0.3533, "step": 8026 }, { "epoch": 1.4, "learning_rate": 1.6261862148726646e-05, "loss": 0.3544, "step": 8027 }, { "epoch": 1.4, "learning_rate": 1.6255930553670716e-05, "loss": 0.3561, "step": 8028 }, { "epoch": 1.4, "learning_rate": 1.6250000000000005e-05, "loss": 0.3506, "step": 8029 }, { "epoch": 1.4, "learning_rate": 1.6244070488263764e-05, "loss": 0.3717, "step": 8030 }, { "epoch": 1.4, "learning_rate": 1.6238142019011175e-05, "loss": 0.3592, "step": 8031 }, { "epoch": 1.4, "learning_rate": 1.6232214592791314e-05, "loss": 0.3488, "step": 8032 }, { "epoch": 1.4, "learning_rate": 1.6226288210153158e-05, "loss": 0.3727, "step": 8033 }, { "epoch": 1.4, "learning_rate": 1.6220362871645567e-05, "loss": 0.3591, "step": 8034 }, { "epoch": 1.4, "learning_rate": 1.621443857781735e-05, "loss": 0.3754, "step": 8035 }, { "epoch": 1.4, "learning_rate": 1.6208515329217177e-05, "loss": 0.3683, "step": 8036 }, { "epoch": 1.4, "learning_rate": 1.620259312639365e-05, "loss": 0.3716, "step": 8037 }, { "epoch": 1.4, "learning_rate": 1.619667196989525e-05, "loss": 0.3552, "step": 8038 }, { "epoch": 1.4, "learning_rate": 1.6190751860270385e-05, "loss": 0.3583, "step": 8039 }, { "epoch": 1.4, "learning_rate": 1.6184832798067354e-05, "loss": 0.3388, "step": 8040 }, { "epoch": 1.4, "learning_rate": 1.617891478383434e-05, "loss": 0.3627, "step": 8041 }, { "epoch": 1.4, "learning_rate": 1.6172997818119465e-05, "loss": 0.3643, "step": 8042 }, { "epoch": 1.4, "learning_rate": 1.616708190147074e-05, "loss": 0.3673, "step": 8043 }, { "epoch": 1.4, "learning_rate": 1.6161167034436074e-05, "loss": 0.3651, "step": 8044 }, { "epoch": 1.4, "learning_rate": 1.6155253217563264e-05, "loss": 0.3533, "step": 8045 }, { "epoch": 1.4, "learning_rate": 1.6149340451400047e-05, "loss": 0.3659, "step": 8046 }, { "epoch": 1.4, "learning_rate": 1.6143428736494028e-05, "loss": 0.3517, "step": 8047 }, { "epoch": 1.4, "learning_rate": 1.6137518073392747e-05, "loss": 0.3728, "step": 8048 }, { "epoch": 1.4, "learning_rate": 1.6131608462643604e-05, "loss": 0.3687, "step": 8049 }, { "epoch": 1.4, "learning_rate": 1.6125699904793944e-05, "loss": 0.3542, "step": 8050 }, { "epoch": 1.4, "learning_rate": 1.6119792400390996e-05, "loss": 0.3507, "step": 8051 }, { "epoch": 1.4, "learning_rate": 1.6113885949981874e-05, "loss": 0.3661, "step": 8052 }, { "epoch": 1.4, "learning_rate": 1.6107980554113624e-05, "loss": 0.3565, "step": 8053 }, { "epoch": 1.4, "learning_rate": 1.6102076213333194e-05, "loss": 0.3563, "step": 8054 }, { "epoch": 1.4, "learning_rate": 1.6096172928187405e-05, "loss": 0.3728, "step": 8055 }, { "epoch": 1.4, "learning_rate": 1.609027069922301e-05, "loss": 0.3511, "step": 8056 }, { "epoch": 1.4, "learning_rate": 1.6084369526986644e-05, "loss": 0.3441, "step": 8057 }, { "epoch": 1.4, "learning_rate": 1.607846941202485e-05, "loss": 0.3579, "step": 8058 }, { "epoch": 1.41, "learning_rate": 1.607257035488408e-05, "loss": 0.3638, "step": 8059 }, { "epoch": 1.41, "learning_rate": 1.6066672356110696e-05, "loss": 0.3688, "step": 8060 }, { "epoch": 1.41, "learning_rate": 1.6060775416250935e-05, "loss": 0.3551, "step": 8061 }, { "epoch": 1.41, "learning_rate": 1.6054879535850942e-05, "loss": 0.3779, "step": 8062 }, { "epoch": 1.41, "learning_rate": 1.6048984715456795e-05, "loss": 0.3447, "step": 8063 }, { "epoch": 1.41, "learning_rate": 1.6043090955614428e-05, "loss": 0.3631, "step": 8064 }, { "epoch": 1.41, "learning_rate": 1.6037198256869718e-05, "loss": 0.367, "step": 8065 }, { "epoch": 1.41, "learning_rate": 1.603130661976841e-05, "loss": 0.3713, "step": 8066 }, { "epoch": 1.41, "learning_rate": 1.6025416044856183e-05, "loss": 0.3672, "step": 8067 }, { "epoch": 1.41, "learning_rate": 1.6019526532678596e-05, "loss": 0.3376, "step": 8068 }, { "epoch": 1.41, "learning_rate": 1.6013638083781092e-05, "loss": 0.3519, "step": 8069 }, { "epoch": 1.41, "learning_rate": 1.600775069870907e-05, "loss": 0.3591, "step": 8070 }, { "epoch": 1.41, "learning_rate": 1.6001864378007794e-05, "loss": 0.3638, "step": 8071 }, { "epoch": 1.41, "learning_rate": 1.599597912222241e-05, "loss": 0.3461, "step": 8072 }, { "epoch": 1.41, "learning_rate": 1.5990094931898013e-05, "loss": 0.3847, "step": 8073 }, { "epoch": 1.41, "learning_rate": 1.5984211807579567e-05, "loss": 0.3598, "step": 8074 }, { "epoch": 1.41, "learning_rate": 1.5978329749811937e-05, "loss": 0.3348, "step": 8075 }, { "epoch": 1.41, "learning_rate": 1.5972448759139907e-05, "loss": 0.3652, "step": 8076 }, { "epoch": 1.41, "learning_rate": 1.5966568836108163e-05, "loss": 0.3636, "step": 8077 }, { "epoch": 1.41, "learning_rate": 1.5960689981261267e-05, "loss": 0.3541, "step": 8078 }, { "epoch": 1.41, "learning_rate": 1.5954812195143697e-05, "loss": 0.3664, "step": 8079 }, { "epoch": 1.41, "learning_rate": 1.5948935478299835e-05, "loss": 0.342, "step": 8080 }, { "epoch": 1.41, "learning_rate": 1.5943059831273978e-05, "loss": 0.3647, "step": 8081 }, { "epoch": 1.41, "learning_rate": 1.5937185254610287e-05, "loss": 0.3554, "step": 8082 }, { "epoch": 1.41, "learning_rate": 1.5931311748852848e-05, "loss": 0.3686, "step": 8083 }, { "epoch": 1.41, "learning_rate": 1.5925439314545653e-05, "loss": 0.3576, "step": 8084 }, { "epoch": 1.41, "learning_rate": 1.5919567952232575e-05, "loss": 0.3633, "step": 8085 }, { "epoch": 1.41, "learning_rate": 1.5913697662457395e-05, "loss": 0.3658, "step": 8086 }, { "epoch": 1.41, "learning_rate": 1.590782844576381e-05, "loss": 0.3659, "step": 8087 }, { "epoch": 1.41, "learning_rate": 1.5901960302695408e-05, "loss": 0.352, "step": 8088 }, { "epoch": 1.41, "learning_rate": 1.5896093233795665e-05, "loss": 0.3608, "step": 8089 }, { "epoch": 1.41, "learning_rate": 1.5890227239607965e-05, "loss": 0.3541, "step": 8090 }, { "epoch": 1.41, "learning_rate": 1.58843623206756e-05, "loss": 0.3477, "step": 8091 }, { "epoch": 1.41, "learning_rate": 1.587849847754177e-05, "loss": 0.36, "step": 8092 }, { "epoch": 1.41, "learning_rate": 1.5872635710749544e-05, "loss": 0.3625, "step": 8093 }, { "epoch": 1.41, "learning_rate": 1.5866774020841928e-05, "loss": 0.3707, "step": 8094 }, { "epoch": 1.41, "learning_rate": 1.58609134083618e-05, "loss": 0.3675, "step": 8095 }, { "epoch": 1.41, "learning_rate": 1.585505387385194e-05, "loss": 0.364, "step": 8096 }, { "epoch": 1.41, "learning_rate": 1.584919541785505e-05, "loss": 0.3589, "step": 8097 }, { "epoch": 1.41, "learning_rate": 1.5843338040913723e-05, "loss": 0.352, "step": 8098 }, { "epoch": 1.41, "learning_rate": 1.583748174357044e-05, "loss": 0.3583, "step": 8099 }, { "epoch": 1.41, "learning_rate": 1.5831626526367585e-05, "loss": 0.3623, "step": 8100 }, { "epoch": 1.41, "learning_rate": 1.582577238984746e-05, "loss": 0.3504, "step": 8101 }, { "epoch": 1.41, "learning_rate": 1.581991933455224e-05, "loss": 0.3637, "step": 8102 }, { "epoch": 1.41, "learning_rate": 1.581406736102403e-05, "loss": 0.3544, "step": 8103 }, { "epoch": 1.41, "learning_rate": 1.5808216469804804e-05, "loss": 0.3597, "step": 8104 }, { "epoch": 1.41, "learning_rate": 1.580236666143646e-05, "loss": 0.3462, "step": 8105 }, { "epoch": 1.41, "learning_rate": 1.579651793646079e-05, "loss": 0.3562, "step": 8106 }, { "epoch": 1.41, "learning_rate": 1.5790670295419463e-05, "loss": 0.3568, "step": 8107 }, { "epoch": 1.41, "learning_rate": 1.5784823738854078e-05, "loss": 0.3509, "step": 8108 }, { "epoch": 1.41, "learning_rate": 1.5778978267306128e-05, "loss": 0.3505, "step": 8109 }, { "epoch": 1.41, "learning_rate": 1.577313388131699e-05, "loss": 0.3559, "step": 8110 }, { "epoch": 1.41, "learning_rate": 1.576729058142796e-05, "loss": 0.3604, "step": 8111 }, { "epoch": 1.41, "learning_rate": 1.5761448368180214e-05, "loss": 0.3536, "step": 8112 }, { "epoch": 1.41, "learning_rate": 1.5755607242114835e-05, "loss": 0.3573, "step": 8113 }, { "epoch": 1.41, "learning_rate": 1.574976720377281e-05, "loss": 0.3489, "step": 8114 }, { "epoch": 1.41, "learning_rate": 1.5743928253695035e-05, "loss": 0.3663, "step": 8115 }, { "epoch": 1.42, "learning_rate": 1.5738090392422277e-05, "loss": 0.3576, "step": 8116 }, { "epoch": 1.42, "learning_rate": 1.573225362049522e-05, "loss": 0.355, "step": 8117 }, { "epoch": 1.42, "learning_rate": 1.5726417938454452e-05, "loss": 0.3699, "step": 8118 }, { "epoch": 1.42, "learning_rate": 1.572058334684044e-05, "loss": 0.3583, "step": 8119 }, { "epoch": 1.42, "learning_rate": 1.5714749846193583e-05, "loss": 0.3708, "step": 8120 }, { "epoch": 1.42, "learning_rate": 1.570891743705413e-05, "loss": 0.3519, "step": 8121 }, { "epoch": 1.42, "learning_rate": 1.5703086119962292e-05, "loss": 0.3576, "step": 8122 }, { "epoch": 1.42, "learning_rate": 1.5697255895458122e-05, "loss": 0.3859, "step": 8123 }, { "epoch": 1.42, "learning_rate": 1.5691426764081586e-05, "loss": 0.3408, "step": 8124 }, { "epoch": 1.42, "learning_rate": 1.5685598726372584e-05, "loss": 0.3522, "step": 8125 }, { "epoch": 1.42, "learning_rate": 1.5679771782870875e-05, "loss": 0.3651, "step": 8126 }, { "epoch": 1.42, "learning_rate": 1.567394593411613e-05, "loss": 0.3717, "step": 8127 }, { "epoch": 1.42, "learning_rate": 1.5668121180647907e-05, "loss": 0.3483, "step": 8128 }, { "epoch": 1.42, "learning_rate": 1.5662297523005693e-05, "loss": 0.3466, "step": 8129 }, { "epoch": 1.42, "learning_rate": 1.565647496172884e-05, "loss": 0.3726, "step": 8130 }, { "epoch": 1.42, "learning_rate": 1.5650653497356615e-05, "loss": 0.3652, "step": 8131 }, { "epoch": 1.42, "learning_rate": 1.564483313042819e-05, "loss": 0.3603, "step": 8132 }, { "epoch": 1.42, "learning_rate": 1.563901386148262e-05, "loss": 0.3705, "step": 8133 }, { "epoch": 1.42, "learning_rate": 1.5633195691058858e-05, "loss": 0.3763, "step": 8134 }, { "epoch": 1.42, "learning_rate": 1.562737861969577e-05, "loss": 0.3707, "step": 8135 }, { "epoch": 1.42, "learning_rate": 1.562156264793212e-05, "loss": 0.3563, "step": 8136 }, { "epoch": 1.42, "learning_rate": 1.561574777630655e-05, "loss": 0.3643, "step": 8137 }, { "epoch": 1.42, "learning_rate": 1.560993400535761e-05, "loss": 0.3768, "step": 8138 }, { "epoch": 1.42, "learning_rate": 1.5604121335623764e-05, "loss": 0.3694, "step": 8139 }, { "epoch": 1.42, "learning_rate": 1.5598309767643355e-05, "loss": 0.3642, "step": 8140 }, { "epoch": 1.42, "learning_rate": 1.5592499301954616e-05, "loss": 0.3649, "step": 8141 }, { "epoch": 1.42, "learning_rate": 1.5586689939095702e-05, "loss": 0.3658, "step": 8142 }, { "epoch": 1.42, "learning_rate": 1.558088167960467e-05, "loss": 0.3484, "step": 8143 }, { "epoch": 1.42, "learning_rate": 1.5575074524019438e-05, "loss": 0.3557, "step": 8144 }, { "epoch": 1.42, "learning_rate": 1.556926847287785e-05, "loss": 0.3621, "step": 8145 }, { "epoch": 1.42, "learning_rate": 1.5563463526717643e-05, "loss": 0.3586, "step": 8146 }, { "epoch": 1.42, "learning_rate": 1.5557659686076453e-05, "loss": 0.3704, "step": 8147 }, { "epoch": 1.42, "learning_rate": 1.5551856951491807e-05, "loss": 0.3661, "step": 8148 }, { "epoch": 1.42, "learning_rate": 1.5546055323501144e-05, "loss": 0.3532, "step": 8149 }, { "epoch": 1.42, "learning_rate": 1.5540254802641775e-05, "loss": 0.3556, "step": 8150 }, { "epoch": 1.42, "learning_rate": 1.5534455389450925e-05, "loss": 0.3605, "step": 8151 }, { "epoch": 1.42, "learning_rate": 1.552865708446572e-05, "loss": 0.3625, "step": 8152 }, { "epoch": 1.42, "learning_rate": 1.552285988822318e-05, "loss": 0.3686, "step": 8153 }, { "epoch": 1.42, "learning_rate": 1.551706380126022e-05, "loss": 0.3612, "step": 8154 }, { "epoch": 1.42, "learning_rate": 1.5511268824113646e-05, "loss": 0.3571, "step": 8155 }, { "epoch": 1.42, "learning_rate": 1.5505474957320176e-05, "loss": 0.3553, "step": 8156 }, { "epoch": 1.42, "learning_rate": 1.5499682201416413e-05, "loss": 0.3628, "step": 8157 }, { "epoch": 1.42, "learning_rate": 1.5493890556938854e-05, "loss": 0.3538, "step": 8158 }, { "epoch": 1.42, "learning_rate": 1.548810002442391e-05, "loss": 0.3621, "step": 8159 }, { "epoch": 1.42, "learning_rate": 1.5482310604407878e-05, "loss": 0.3667, "step": 8160 }, { "epoch": 1.42, "learning_rate": 1.5476522297426956e-05, "loss": 0.3603, "step": 8161 }, { "epoch": 1.42, "learning_rate": 1.5470735104017225e-05, "loss": 0.3826, "step": 8162 }, { "epoch": 1.42, "learning_rate": 1.5464949024714682e-05, "loss": 0.3562, "step": 8163 }, { "epoch": 1.42, "learning_rate": 1.545916406005522e-05, "loss": 0.3673, "step": 8164 }, { "epoch": 1.42, "learning_rate": 1.5453380210574605e-05, "loss": 0.3629, "step": 8165 }, { "epoch": 1.42, "learning_rate": 1.5447597476808533e-05, "loss": 0.3645, "step": 8166 }, { "epoch": 1.42, "learning_rate": 1.5441815859292573e-05, "loss": 0.3621, "step": 8167 }, { "epoch": 1.42, "learning_rate": 1.5436035358562194e-05, "loss": 0.3664, "step": 8168 }, { "epoch": 1.42, "learning_rate": 1.5430255975152753e-05, "loss": 0.3427, "step": 8169 }, { "epoch": 1.42, "learning_rate": 1.5424477709599542e-05, "loss": 0.3439, "step": 8170 }, { "epoch": 1.42, "learning_rate": 1.5418700562437722e-05, "loss": 0.356, "step": 8171 }, { "epoch": 1.42, "learning_rate": 1.5412924534202326e-05, "loss": 0.3522, "step": 8172 }, { "epoch": 1.42, "learning_rate": 1.540714962542833e-05, "loss": 0.3514, "step": 8173 }, { "epoch": 1.43, "learning_rate": 1.540137583665058e-05, "loss": 0.3503, "step": 8174 }, { "epoch": 1.43, "learning_rate": 1.539560316840382e-05, "loss": 0.3598, "step": 8175 }, { "epoch": 1.43, "learning_rate": 1.5389831621222696e-05, "loss": 0.3333, "step": 8176 }, { "epoch": 1.43, "learning_rate": 1.5384061195641752e-05, "loss": 0.3453, "step": 8177 }, { "epoch": 1.43, "learning_rate": 1.5378291892195423e-05, "loss": 0.3624, "step": 8178 }, { "epoch": 1.43, "learning_rate": 1.537252371141803e-05, "loss": 0.3649, "step": 8179 }, { "epoch": 1.43, "learning_rate": 1.5366756653843807e-05, "loss": 0.3638, "step": 8180 }, { "epoch": 1.43, "learning_rate": 1.5360990720006892e-05, "loss": 0.3555, "step": 8181 }, { "epoch": 1.43, "learning_rate": 1.5355225910441295e-05, "loss": 0.3659, "step": 8182 }, { "epoch": 1.43, "learning_rate": 1.5349462225680923e-05, "loss": 0.3712, "step": 8183 }, { "epoch": 1.43, "learning_rate": 1.5343699666259597e-05, "loss": 0.3596, "step": 8184 }, { "epoch": 1.43, "learning_rate": 1.5337938232711018e-05, "loss": 0.368, "step": 8185 }, { "epoch": 1.43, "learning_rate": 1.5332177925568798e-05, "loss": 0.3704, "step": 8186 }, { "epoch": 1.43, "learning_rate": 1.5326418745366438e-05, "loss": 0.3542, "step": 8187 }, { "epoch": 1.43, "learning_rate": 1.5320660692637324e-05, "loss": 0.3654, "step": 8188 }, { "epoch": 1.43, "learning_rate": 1.5314903767914747e-05, "loss": 0.3595, "step": 8189 }, { "epoch": 1.43, "learning_rate": 1.5309147971731895e-05, "loss": 0.3645, "step": 8190 }, { "epoch": 1.43, "learning_rate": 1.5303393304621844e-05, "loss": 0.3575, "step": 8191 }, { "epoch": 1.43, "learning_rate": 1.529763976711758e-05, "loss": 0.3542, "step": 8192 }, { "epoch": 1.43, "learning_rate": 1.529188735975197e-05, "loss": 0.3719, "step": 8193 }, { "epoch": 1.43, "learning_rate": 1.5286136083057782e-05, "loss": 0.3684, "step": 8194 }, { "epoch": 1.43, "learning_rate": 1.5280385937567677e-05, "loss": 0.3482, "step": 8195 }, { "epoch": 1.43, "learning_rate": 1.5274636923814213e-05, "loss": 0.3723, "step": 8196 }, { "epoch": 1.43, "learning_rate": 1.5268889042329837e-05, "loss": 0.3627, "step": 8197 }, { "epoch": 1.43, "learning_rate": 1.526314229364691e-05, "loss": 0.3595, "step": 8198 }, { "epoch": 1.43, "learning_rate": 1.525739667829767e-05, "loss": 0.3646, "step": 8199 }, { "epoch": 1.43, "learning_rate": 1.5251652196814248e-05, "loss": 0.3591, "step": 8200 }, { "epoch": 1.43, "learning_rate": 1.524590884972869e-05, "loss": 0.3578, "step": 8201 }, { "epoch": 1.43, "learning_rate": 1.5240166637572911e-05, "loss": 0.3626, "step": 8202 }, { "epoch": 1.43, "learning_rate": 1.5234425560878734e-05, "loss": 0.3432, "step": 8203 }, { "epoch": 1.43, "learning_rate": 1.5228685620177896e-05, "loss": 0.3441, "step": 8204 }, { "epoch": 1.43, "learning_rate": 1.5222946816001999e-05, "loss": 0.3475, "step": 8205 }, { "epoch": 1.43, "learning_rate": 1.5217209148882532e-05, "loss": 0.3449, "step": 8206 }, { "epoch": 1.43, "learning_rate": 1.521147261935092e-05, "loss": 0.3678, "step": 8207 }, { "epoch": 1.43, "learning_rate": 1.520573722793846e-05, "loss": 0.3635, "step": 8208 }, { "epoch": 1.43, "learning_rate": 1.520000297517633e-05, "loss": 0.3648, "step": 8209 }, { "epoch": 1.43, "learning_rate": 1.5194269861595622e-05, "loss": 0.3526, "step": 8210 }, { "epoch": 1.43, "learning_rate": 1.5188537887727319e-05, "loss": 0.3525, "step": 8211 }, { "epoch": 1.43, "learning_rate": 1.5182807054102297e-05, "loss": 0.3594, "step": 8212 }, { "epoch": 1.43, "learning_rate": 1.5177077361251313e-05, "loss": 0.3555, "step": 8213 }, { "epoch": 1.43, "learning_rate": 1.5171348809705033e-05, "loss": 0.3565, "step": 8214 }, { "epoch": 1.43, "learning_rate": 1.5165621399994035e-05, "loss": 0.3694, "step": 8215 }, { "epoch": 1.43, "learning_rate": 1.5159895132648753e-05, "loss": 0.3716, "step": 8216 }, { "epoch": 1.43, "learning_rate": 1.5154170008199527e-05, "loss": 0.3644, "step": 8217 }, { "epoch": 1.43, "learning_rate": 1.514844602717661e-05, "loss": 0.3629, "step": 8218 }, { "epoch": 1.43, "learning_rate": 1.5142723190110143e-05, "loss": 0.3494, "step": 8219 }, { "epoch": 1.43, "learning_rate": 1.513700149753014e-05, "loss": 0.3606, "step": 8220 }, { "epoch": 1.43, "learning_rate": 1.5131280949966526e-05, "loss": 0.3485, "step": 8221 }, { "epoch": 1.43, "learning_rate": 1.512556154794913e-05, "loss": 0.3794, "step": 8222 }, { "epoch": 1.43, "learning_rate": 1.5119843292007646e-05, "loss": 0.3581, "step": 8223 }, { "epoch": 1.43, "learning_rate": 1.5114126182671674e-05, "loss": 0.37, "step": 8224 }, { "epoch": 1.43, "learning_rate": 1.5108410220470736e-05, "loss": 0.3558, "step": 8225 }, { "epoch": 1.43, "learning_rate": 1.5102695405934214e-05, "loss": 0.3527, "step": 8226 }, { "epoch": 1.43, "learning_rate": 1.5096981739591381e-05, "loss": 0.3606, "step": 8227 }, { "epoch": 1.43, "learning_rate": 1.5091269221971432e-05, "loss": 0.3733, "step": 8228 }, { "epoch": 1.43, "learning_rate": 1.508555785360343e-05, "loss": 0.3381, "step": 8229 }, { "epoch": 1.43, "learning_rate": 1.5079847635016348e-05, "loss": 0.3373, "step": 8230 }, { "epoch": 1.44, "learning_rate": 1.5074138566739038e-05, "loss": 0.3641, "step": 8231 }, { "epoch": 1.44, "learning_rate": 1.5068430649300263e-05, "loss": 0.3648, "step": 8232 }, { "epoch": 1.44, "learning_rate": 1.5062723883228666e-05, "loss": 0.3772, "step": 8233 }, { "epoch": 1.44, "learning_rate": 1.5057018269052781e-05, "loss": 0.355, "step": 8234 }, { "epoch": 1.44, "learning_rate": 1.5051313807301042e-05, "loss": 0.3605, "step": 8235 }, { "epoch": 1.44, "learning_rate": 1.504561049850179e-05, "loss": 0.3564, "step": 8236 }, { "epoch": 1.44, "learning_rate": 1.5039908343183234e-05, "loss": 0.3449, "step": 8237 }, { "epoch": 1.44, "learning_rate": 1.503420734187348e-05, "loss": 0.3583, "step": 8238 }, { "epoch": 1.44, "learning_rate": 1.5028507495100552e-05, "loss": 0.3705, "step": 8239 }, { "epoch": 1.44, "learning_rate": 1.5022808803392332e-05, "loss": 0.3678, "step": 8240 }, { "epoch": 1.44, "learning_rate": 1.5017111267276617e-05, "loss": 0.3617, "step": 8241 }, { "epoch": 1.44, "learning_rate": 1.5011414887281105e-05, "loss": 0.3569, "step": 8242 }, { "epoch": 1.44, "learning_rate": 1.5005719663933365e-05, "loss": 0.3541, "step": 8243 }, { "epoch": 1.44, "learning_rate": 1.5000025597760858e-05, "loss": 0.3597, "step": 8244 }, { "epoch": 1.44, "learning_rate": 1.4994332689290968e-05, "loss": 0.3614, "step": 8245 }, { "epoch": 1.44, "learning_rate": 1.4988640939050929e-05, "loss": 0.3604, "step": 8246 }, { "epoch": 1.44, "learning_rate": 1.4982950347567915e-05, "loss": 0.3511, "step": 8247 }, { "epoch": 1.44, "learning_rate": 1.4977260915368943e-05, "loss": 0.3496, "step": 8248 }, { "epoch": 1.44, "learning_rate": 1.497157264298097e-05, "loss": 0.3588, "step": 8249 }, { "epoch": 1.44, "learning_rate": 1.4965885530930813e-05, "loss": 0.3659, "step": 8250 }, { "epoch": 1.44, "learning_rate": 1.4960199579745182e-05, "loss": 0.3643, "step": 8251 }, { "epoch": 1.44, "learning_rate": 1.4954514789950703e-05, "loss": 0.3516, "step": 8252 }, { "epoch": 1.44, "learning_rate": 1.4948831162073888e-05, "loss": 0.3475, "step": 8253 }, { "epoch": 1.44, "learning_rate": 1.4943148696641118e-05, "loss": 0.3774, "step": 8254 }, { "epoch": 1.44, "learning_rate": 1.4937467394178683e-05, "loss": 0.3586, "step": 8255 }, { "epoch": 1.44, "learning_rate": 1.4931787255212776e-05, "loss": 0.3501, "step": 8256 }, { "epoch": 1.44, "learning_rate": 1.4926108280269458e-05, "loss": 0.3481, "step": 8257 }, { "epoch": 1.44, "learning_rate": 1.4920430469874703e-05, "loss": 0.3656, "step": 8258 }, { "epoch": 1.44, "learning_rate": 1.4914753824554373e-05, "loss": 0.357, "step": 8259 }, { "epoch": 1.44, "learning_rate": 1.4909078344834215e-05, "loss": 0.3413, "step": 8260 }, { "epoch": 1.44, "learning_rate": 1.490340403123987e-05, "loss": 0.3476, "step": 8261 }, { "epoch": 1.44, "learning_rate": 1.4897730884296868e-05, "loss": 0.3546, "step": 8262 }, { "epoch": 1.44, "learning_rate": 1.4892058904530637e-05, "loss": 0.3489, "step": 8263 }, { "epoch": 1.44, "learning_rate": 1.4886388092466508e-05, "loss": 0.3635, "step": 8264 }, { "epoch": 1.44, "learning_rate": 1.4880718448629676e-05, "loss": 0.3569, "step": 8265 }, { "epoch": 1.44, "learning_rate": 1.4875049973545253e-05, "loss": 0.3824, "step": 8266 }, { "epoch": 1.44, "learning_rate": 1.486938266773823e-05, "loss": 0.366, "step": 8267 }, { "epoch": 1.44, "learning_rate": 1.4863716531733483e-05, "loss": 0.3732, "step": 8268 }, { "epoch": 1.44, "learning_rate": 1.4858051566055795e-05, "loss": 0.376, "step": 8269 }, { "epoch": 1.44, "learning_rate": 1.4852387771229848e-05, "loss": 0.3586, "step": 8270 }, { "epoch": 1.44, "learning_rate": 1.4846725147780188e-05, "loss": 0.3608, "step": 8271 }, { "epoch": 1.44, "learning_rate": 1.4841063696231262e-05, "loss": 0.3609, "step": 8272 }, { "epoch": 1.44, "learning_rate": 1.483540341710743e-05, "loss": 0.3539, "step": 8273 }, { "epoch": 1.44, "learning_rate": 1.4829744310932913e-05, "loss": 0.3621, "step": 8274 }, { "epoch": 1.44, "learning_rate": 1.4824086378231847e-05, "loss": 0.3544, "step": 8275 }, { "epoch": 1.44, "learning_rate": 1.4818429619528235e-05, "loss": 0.3353, "step": 8276 }, { "epoch": 1.44, "learning_rate": 1.4812774035346006e-05, "loss": 0.3564, "step": 8277 }, { "epoch": 1.44, "learning_rate": 1.4807119626208943e-05, "loss": 0.3682, "step": 8278 }, { "epoch": 1.44, "learning_rate": 1.4801466392640728e-05, "loss": 0.371, "step": 8279 }, { "epoch": 1.44, "learning_rate": 1.4795814335164976e-05, "loss": 0.3629, "step": 8280 }, { "epoch": 1.44, "learning_rate": 1.4790163454305142e-05, "loss": 0.3832, "step": 8281 }, { "epoch": 1.44, "learning_rate": 1.4784513750584583e-05, "loss": 0.363, "step": 8282 }, { "epoch": 1.44, "learning_rate": 1.4778865224526567e-05, "loss": 0.3643, "step": 8283 }, { "epoch": 1.44, "learning_rate": 1.4773217876654236e-05, "loss": 0.362, "step": 8284 }, { "epoch": 1.44, "learning_rate": 1.4767571707490618e-05, "loss": 0.3622, "step": 8285 }, { "epoch": 1.44, "learning_rate": 1.4761926717558645e-05, "loss": 0.3691, "step": 8286 }, { "epoch": 1.44, "learning_rate": 1.475628290738115e-05, "loss": 0.372, "step": 8287 }, { "epoch": 1.44, "learning_rate": 1.4750640277480831e-05, "loss": 0.346, "step": 8288 }, { "epoch": 1.45, "learning_rate": 1.4744998828380277e-05, "loss": 0.3515, "step": 8289 }, { "epoch": 1.45, "learning_rate": 1.4739358560601995e-05, "loss": 0.3577, "step": 8290 }, { "epoch": 1.45, "learning_rate": 1.4733719474668368e-05, "loss": 0.3679, "step": 8291 }, { "epoch": 1.45, "learning_rate": 1.472808157110166e-05, "loss": 0.3631, "step": 8292 }, { "epoch": 1.45, "learning_rate": 1.4722444850424027e-05, "loss": 0.3764, "step": 8293 }, { "epoch": 1.45, "learning_rate": 1.4716809313157538e-05, "loss": 0.3674, "step": 8294 }, { "epoch": 1.45, "learning_rate": 1.4711174959824125e-05, "loss": 0.3618, "step": 8295 }, { "epoch": 1.45, "learning_rate": 1.470554179094561e-05, "loss": 0.3405, "step": 8296 }, { "epoch": 1.45, "learning_rate": 1.4699909807043749e-05, "loss": 0.3643, "step": 8297 }, { "epoch": 1.45, "learning_rate": 1.4694279008640136e-05, "loss": 0.3537, "step": 8298 }, { "epoch": 1.45, "learning_rate": 1.4688649396256266e-05, "loss": 0.3499, "step": 8299 }, { "epoch": 1.45, "learning_rate": 1.4683020970413558e-05, "loss": 0.347, "step": 8300 }, { "epoch": 1.45, "learning_rate": 1.467739373163327e-05, "loss": 0.3694, "step": 8301 }, { "epoch": 1.45, "learning_rate": 1.4671767680436598e-05, "loss": 0.3561, "step": 8302 }, { "epoch": 1.45, "learning_rate": 1.4666142817344597e-05, "loss": 0.3545, "step": 8303 }, { "epoch": 1.45, "learning_rate": 1.4660519142878226e-05, "loss": 0.3543, "step": 8304 }, { "epoch": 1.45, "learning_rate": 1.4654896657558326e-05, "loss": 0.3489, "step": 8305 }, { "epoch": 1.45, "learning_rate": 1.4649275361905623e-05, "loss": 0.3641, "step": 8306 }, { "epoch": 1.45, "learning_rate": 1.4643655256440753e-05, "loss": 0.3372, "step": 8307 }, { "epoch": 1.45, "learning_rate": 1.463803634168423e-05, "loss": 0.3594, "step": 8308 }, { "epoch": 1.45, "learning_rate": 1.4632418618156457e-05, "loss": 0.3427, "step": 8309 }, { "epoch": 1.45, "learning_rate": 1.4626802086377717e-05, "loss": 0.3507, "step": 8310 }, { "epoch": 1.45, "learning_rate": 1.462118674686821e-05, "loss": 0.3484, "step": 8311 }, { "epoch": 1.45, "learning_rate": 1.4615572600147986e-05, "loss": 0.3625, "step": 8312 }, { "epoch": 1.45, "learning_rate": 1.4609959646737037e-05, "loss": 0.3704, "step": 8313 }, { "epoch": 1.45, "learning_rate": 1.4604347887155184e-05, "loss": 0.3468, "step": 8314 }, { "epoch": 1.45, "learning_rate": 1.4598737321922187e-05, "loss": 0.3629, "step": 8315 }, { "epoch": 1.45, "learning_rate": 1.4593127951557678e-05, "loss": 0.354, "step": 8316 }, { "epoch": 1.45, "learning_rate": 1.4587519776581154e-05, "loss": 0.3265, "step": 8317 }, { "epoch": 1.45, "learning_rate": 1.4581912797512043e-05, "loss": 0.3682, "step": 8318 }, { "epoch": 1.45, "learning_rate": 1.457630701486965e-05, "loss": 0.3607, "step": 8319 }, { "epoch": 1.45, "learning_rate": 1.4570702429173144e-05, "loss": 0.3895, "step": 8320 }, { "epoch": 1.45, "learning_rate": 1.4565099040941615e-05, "loss": 0.3689, "step": 8321 }, { "epoch": 1.45, "learning_rate": 1.4559496850694024e-05, "loss": 0.3395, "step": 8322 }, { "epoch": 1.45, "learning_rate": 1.4553895858949218e-05, "loss": 0.3696, "step": 8323 }, { "epoch": 1.45, "learning_rate": 1.454829606622595e-05, "loss": 0.346, "step": 8324 }, { "epoch": 1.45, "learning_rate": 1.4542697473042855e-05, "loss": 0.3463, "step": 8325 }, { "epoch": 1.45, "learning_rate": 1.4537100079918453e-05, "loss": 0.368, "step": 8326 }, { "epoch": 1.45, "learning_rate": 1.453150388737114e-05, "loss": 0.3756, "step": 8327 }, { "epoch": 1.45, "learning_rate": 1.4525908895919238e-05, "loss": 0.345, "step": 8328 }, { "epoch": 1.45, "learning_rate": 1.4520315106080912e-05, "loss": 0.358, "step": 8329 }, { "epoch": 1.45, "learning_rate": 1.4514722518374262e-05, "loss": 0.3624, "step": 8330 }, { "epoch": 1.45, "learning_rate": 1.4509131133317232e-05, "loss": 0.3703, "step": 8331 }, { "epoch": 1.45, "learning_rate": 1.4503540951427692e-05, "loss": 0.3549, "step": 8332 }, { "epoch": 1.45, "learning_rate": 1.4497951973223379e-05, "loss": 0.335, "step": 8333 }, { "epoch": 1.45, "learning_rate": 1.4492364199221905e-05, "loss": 0.3508, "step": 8334 }, { "epoch": 1.45, "learning_rate": 1.4486777629940829e-05, "loss": 0.3413, "step": 8335 }, { "epoch": 1.45, "learning_rate": 1.4481192265897533e-05, "loss": 0.3559, "step": 8336 }, { "epoch": 1.45, "learning_rate": 1.4475608107609311e-05, "loss": 0.3561, "step": 8337 }, { "epoch": 1.45, "learning_rate": 1.4470025155593361e-05, "loss": 0.364, "step": 8338 }, { "epoch": 1.45, "learning_rate": 1.4464443410366752e-05, "loss": 0.3639, "step": 8339 }, { "epoch": 1.45, "learning_rate": 1.4458862872446432e-05, "loss": 0.349, "step": 8340 }, { "epoch": 1.45, "learning_rate": 1.4453283542349261e-05, "loss": 0.3662, "step": 8341 }, { "epoch": 1.45, "learning_rate": 1.4447705420591983e-05, "loss": 0.3733, "step": 8342 }, { "epoch": 1.45, "learning_rate": 1.4442128507691222e-05, "loss": 0.3495, "step": 8343 }, { "epoch": 1.45, "learning_rate": 1.4436552804163472e-05, "loss": 0.362, "step": 8344 }, { "epoch": 1.45, "learning_rate": 1.4430978310525153e-05, "loss": 0.347, "step": 8345 }, { "epoch": 1.46, "learning_rate": 1.442540502729256e-05, "loss": 0.351, "step": 8346 }, { "epoch": 1.46, "learning_rate": 1.4419832954981855e-05, "loss": 0.3508, "step": 8347 }, { "epoch": 1.46, "learning_rate": 1.4414262094109104e-05, "loss": 0.3548, "step": 8348 }, { "epoch": 1.46, "learning_rate": 1.4408692445190274e-05, "loss": 0.3674, "step": 8349 }, { "epoch": 1.46, "learning_rate": 1.4403124008741197e-05, "loss": 0.3581, "step": 8350 }, { "epoch": 1.46, "learning_rate": 1.4397556785277593e-05, "loss": 0.3562, "step": 8351 }, { "epoch": 1.46, "learning_rate": 1.4391990775315084e-05, "loss": 0.3594, "step": 8352 }, { "epoch": 1.46, "learning_rate": 1.4386425979369189e-05, "loss": 0.3577, "step": 8353 }, { "epoch": 1.46, "learning_rate": 1.4380862397955283e-05, "loss": 0.3492, "step": 8354 }, { "epoch": 1.46, "learning_rate": 1.4375300031588645e-05, "loss": 0.3548, "step": 8355 }, { "epoch": 1.46, "learning_rate": 1.4369738880784445e-05, "loss": 0.3534, "step": 8356 }, { "epoch": 1.46, "learning_rate": 1.4364178946057738e-05, "loss": 0.3666, "step": 8357 }, { "epoch": 1.46, "learning_rate": 1.435862022792346e-05, "loss": 0.3505, "step": 8358 }, { "epoch": 1.46, "learning_rate": 1.4353062726896454e-05, "loss": 0.3414, "step": 8359 }, { "epoch": 1.46, "learning_rate": 1.4347506443491424e-05, "loss": 0.3618, "step": 8360 }, { "epoch": 1.46, "learning_rate": 1.4341951378222964e-05, "loss": 0.3607, "step": 8361 }, { "epoch": 1.46, "learning_rate": 1.4336397531605578e-05, "loss": 0.3695, "step": 8362 }, { "epoch": 1.46, "learning_rate": 1.4330844904153644e-05, "loss": 0.3599, "step": 8363 }, { "epoch": 1.46, "learning_rate": 1.432529349638142e-05, "loss": 0.3585, "step": 8364 }, { "epoch": 1.46, "learning_rate": 1.4319743308803058e-05, "loss": 0.3608, "step": 8365 }, { "epoch": 1.46, "learning_rate": 1.43141943419326e-05, "loss": 0.353, "step": 8366 }, { "epoch": 1.46, "learning_rate": 1.430864659628397e-05, "loss": 0.3446, "step": 8367 }, { "epoch": 1.46, "learning_rate": 1.4303100072370973e-05, "loss": 0.3685, "step": 8368 }, { "epoch": 1.46, "learning_rate": 1.429755477070731e-05, "loss": 0.3751, "step": 8369 }, { "epoch": 1.46, "learning_rate": 1.4292010691806584e-05, "loss": 0.3442, "step": 8370 }, { "epoch": 1.46, "learning_rate": 1.4286467836182255e-05, "loss": 0.3756, "step": 8371 }, { "epoch": 1.46, "learning_rate": 1.4280926204347673e-05, "loss": 0.3549, "step": 8372 }, { "epoch": 1.46, "learning_rate": 1.427538579681609e-05, "loss": 0.3487, "step": 8373 }, { "epoch": 1.46, "learning_rate": 1.4269846614100651e-05, "loss": 0.3628, "step": 8374 }, { "epoch": 1.46, "learning_rate": 1.426430865671436e-05, "loss": 0.3543, "step": 8375 }, { "epoch": 1.46, "learning_rate": 1.4258771925170133e-05, "loss": 0.3524, "step": 8376 }, { "epoch": 1.46, "learning_rate": 1.425323641998076e-05, "loss": 0.3672, "step": 8377 }, { "epoch": 1.46, "learning_rate": 1.4247702141658908e-05, "loss": 0.3663, "step": 8378 }, { "epoch": 1.46, "learning_rate": 1.424216909071715e-05, "loss": 0.3619, "step": 8379 }, { "epoch": 1.46, "learning_rate": 1.4236637267667952e-05, "loss": 0.3614, "step": 8380 }, { "epoch": 1.46, "learning_rate": 1.4231106673023633e-05, "loss": 0.3565, "step": 8381 }, { "epoch": 1.46, "learning_rate": 1.4225577307296415e-05, "loss": 0.3406, "step": 8382 }, { "epoch": 1.46, "learning_rate": 1.4220049170998424e-05, "loss": 0.3611, "step": 8383 }, { "epoch": 1.46, "learning_rate": 1.4214522264641634e-05, "loss": 0.3625, "step": 8384 }, { "epoch": 1.46, "learning_rate": 1.420899658873795e-05, "loss": 0.3734, "step": 8385 }, { "epoch": 1.46, "learning_rate": 1.4203472143799124e-05, "loss": 0.3674, "step": 8386 }, { "epoch": 1.46, "learning_rate": 1.4197948930336821e-05, "loss": 0.3486, "step": 8387 }, { "epoch": 1.46, "learning_rate": 1.4192426948862578e-05, "loss": 0.3602, "step": 8388 }, { "epoch": 1.46, "learning_rate": 1.4186906199887811e-05, "loss": 0.3658, "step": 8389 }, { "epoch": 1.46, "learning_rate": 1.4181386683923839e-05, "loss": 0.3498, "step": 8390 }, { "epoch": 1.46, "learning_rate": 1.417586840148187e-05, "loss": 0.3546, "step": 8391 }, { "epoch": 1.46, "learning_rate": 1.417035135307298e-05, "loss": 0.3893, "step": 8392 }, { "epoch": 1.46, "learning_rate": 1.4164835539208126e-05, "loss": 0.3612, "step": 8393 }, { "epoch": 1.46, "learning_rate": 1.4159320960398182e-05, "loss": 0.3519, "step": 8394 }, { "epoch": 1.46, "learning_rate": 1.4153807617153874e-05, "loss": 0.3534, "step": 8395 }, { "epoch": 1.46, "learning_rate": 1.4148295509985831e-05, "loss": 0.3647, "step": 8396 }, { "epoch": 1.46, "learning_rate": 1.4142784639404574e-05, "loss": 0.368, "step": 8397 }, { "epoch": 1.46, "learning_rate": 1.4137275005920497e-05, "loss": 0.3522, "step": 8398 }, { "epoch": 1.46, "learning_rate": 1.4131766610043867e-05, "loss": 0.3477, "step": 8399 }, { "epoch": 1.46, "learning_rate": 1.4126259452284874e-05, "loss": 0.343, "step": 8400 }, { "epoch": 1.46, "learning_rate": 1.4120753533153552e-05, "loss": 0.3556, "step": 8401 }, { "epoch": 1.46, "learning_rate": 1.4115248853159855e-05, "loss": 0.351, "step": 8402 }, { "epoch": 1.47, "learning_rate": 1.4109745412813594e-05, "loss": 0.3914, "step": 8403 }, { "epoch": 1.47, "learning_rate": 1.4104243212624491e-05, "loss": 0.3759, "step": 8404 }, { "epoch": 1.47, "learning_rate": 1.409874225310213e-05, "loss": 0.3515, "step": 8405 }, { "epoch": 1.47, "learning_rate": 1.4093242534755985e-05, "loss": 0.3616, "step": 8406 }, { "epoch": 1.47, "learning_rate": 1.4087744058095426e-05, "loss": 0.3598, "step": 8407 }, { "epoch": 1.47, "learning_rate": 1.4082246823629713e-05, "loss": 0.3695, "step": 8408 }, { "epoch": 1.47, "learning_rate": 1.4076750831867971e-05, "loss": 0.3624, "step": 8409 }, { "epoch": 1.47, "learning_rate": 1.4071256083319212e-05, "loss": 0.3558, "step": 8410 }, { "epoch": 1.47, "learning_rate": 1.4065762578492354e-05, "loss": 0.3631, "step": 8411 }, { "epoch": 1.47, "learning_rate": 1.4060270317896167e-05, "loss": 0.3454, "step": 8412 }, { "epoch": 1.47, "learning_rate": 1.4054779302039339e-05, "loss": 0.3591, "step": 8413 }, { "epoch": 1.47, "learning_rate": 1.4049289531430432e-05, "loss": 0.362, "step": 8414 }, { "epoch": 1.47, "learning_rate": 1.4043801006577879e-05, "loss": 0.3706, "step": 8415 }, { "epoch": 1.47, "learning_rate": 1.4038313727990005e-05, "loss": 0.366, "step": 8416 }, { "epoch": 1.47, "learning_rate": 1.4032827696175027e-05, "loss": 0.3552, "step": 8417 }, { "epoch": 1.47, "learning_rate": 1.4027342911641045e-05, "loss": 0.3733, "step": 8418 }, { "epoch": 1.47, "learning_rate": 1.402185937489604e-05, "loss": 0.3553, "step": 8419 }, { "epoch": 1.47, "learning_rate": 1.4016377086447863e-05, "loss": 0.3764, "step": 8420 }, { "epoch": 1.47, "learning_rate": 1.4010896046804283e-05, "loss": 0.3391, "step": 8421 }, { "epoch": 1.47, "learning_rate": 1.400541625647293e-05, "loss": 0.3536, "step": 8422 }, { "epoch": 1.47, "learning_rate": 1.3999937715961303e-05, "loss": 0.3547, "step": 8423 }, { "epoch": 1.47, "learning_rate": 1.3994460425776829e-05, "loss": 0.3571, "step": 8424 }, { "epoch": 1.47, "learning_rate": 1.3988984386426784e-05, "loss": 0.3509, "step": 8425 }, { "epoch": 1.47, "learning_rate": 1.3983509598418346e-05, "loss": 0.3546, "step": 8426 }, { "epoch": 1.47, "learning_rate": 1.3978036062258554e-05, "loss": 0.361, "step": 8427 }, { "epoch": 1.47, "learning_rate": 1.3972563778454362e-05, "loss": 0.3607, "step": 8428 }, { "epoch": 1.47, "learning_rate": 1.3967092747512595e-05, "loss": 0.3685, "step": 8429 }, { "epoch": 1.47, "learning_rate": 1.3961622969939947e-05, "loss": 0.3827, "step": 8430 }, { "epoch": 1.47, "learning_rate": 1.3956154446243024e-05, "loss": 0.3646, "step": 8431 }, { "epoch": 1.47, "learning_rate": 1.395068717692829e-05, "loss": 0.3589, "step": 8432 }, { "epoch": 1.47, "learning_rate": 1.3945221162502103e-05, "loss": 0.3765, "step": 8433 }, { "epoch": 1.47, "learning_rate": 1.393975640347071e-05, "loss": 0.3423, "step": 8434 }, { "epoch": 1.47, "learning_rate": 1.3934292900340245e-05, "loss": 0.3641, "step": 8435 }, { "epoch": 1.47, "learning_rate": 1.3928830653616709e-05, "loss": 0.3471, "step": 8436 }, { "epoch": 1.47, "learning_rate": 1.392336966380599e-05, "loss": 0.3616, "step": 8437 }, { "epoch": 1.47, "learning_rate": 1.391790993141388e-05, "loss": 0.3665, "step": 8438 }, { "epoch": 1.47, "learning_rate": 1.3912451456946023e-05, "loss": 0.3616, "step": 8439 }, { "epoch": 1.47, "learning_rate": 1.390699424090798e-05, "loss": 0.3553, "step": 8440 }, { "epoch": 1.47, "learning_rate": 1.3901538283805165e-05, "loss": 0.353, "step": 8441 }, { "epoch": 1.47, "learning_rate": 1.3896083586142899e-05, "loss": 0.3468, "step": 8442 }, { "epoch": 1.47, "learning_rate": 1.3890630148426375e-05, "loss": 0.3423, "step": 8443 }, { "epoch": 1.47, "learning_rate": 1.3885177971160656e-05, "loss": 0.3612, "step": 8444 }, { "epoch": 1.47, "learning_rate": 1.3879727054850722e-05, "loss": 0.3533, "step": 8445 }, { "epoch": 1.47, "learning_rate": 1.3874277400001412e-05, "loss": 0.3487, "step": 8446 }, { "epoch": 1.47, "learning_rate": 1.3868829007117451e-05, "loss": 0.3471, "step": 8447 }, { "epoch": 1.47, "learning_rate": 1.3863381876703446e-05, "loss": 0.3698, "step": 8448 }, { "epoch": 1.47, "learning_rate": 1.38579360092639e-05, "loss": 0.3545, "step": 8449 }, { "epoch": 1.47, "learning_rate": 1.385249140530318e-05, "loss": 0.3701, "step": 8450 }, { "epoch": 1.47, "learning_rate": 1.3847048065325547e-05, "loss": 0.35, "step": 8451 }, { "epoch": 1.47, "learning_rate": 1.3841605989835155e-05, "loss": 0.3645, "step": 8452 }, { "epoch": 1.47, "learning_rate": 1.3836165179336019e-05, "loss": 0.35, "step": 8453 }, { "epoch": 1.47, "learning_rate": 1.3830725634332047e-05, "loss": 0.3578, "step": 8454 }, { "epoch": 1.47, "learning_rate": 1.3825287355327036e-05, "loss": 0.3554, "step": 8455 }, { "epoch": 1.47, "learning_rate": 1.3819850342824647e-05, "loss": 0.3691, "step": 8456 }, { "epoch": 1.47, "learning_rate": 1.3814414597328457e-05, "loss": 0.3581, "step": 8457 }, { "epoch": 1.47, "learning_rate": 1.3808980119341885e-05, "loss": 0.3575, "step": 8458 }, { "epoch": 1.47, "learning_rate": 1.380354690936827e-05, "loss": 0.3755, "step": 8459 }, { "epoch": 1.47, "learning_rate": 1.3798114967910803e-05, "loss": 0.3572, "step": 8460 }, { "epoch": 1.48, "learning_rate": 1.3792684295472574e-05, "loss": 0.3734, "step": 8461 }, { "epoch": 1.48, "learning_rate": 1.378725489255655e-05, "loss": 0.3502, "step": 8462 }, { "epoch": 1.48, "learning_rate": 1.3781826759665597e-05, "loss": 0.3498, "step": 8463 }, { "epoch": 1.48, "learning_rate": 1.3776399897302438e-05, "loss": 0.3647, "step": 8464 }, { "epoch": 1.48, "learning_rate": 1.3770974305969681e-05, "loss": 0.3486, "step": 8465 }, { "epoch": 1.48, "learning_rate": 1.3765549986169845e-05, "loss": 0.3572, "step": 8466 }, { "epoch": 1.48, "learning_rate": 1.376012693840529e-05, "loss": 0.3484, "step": 8467 }, { "epoch": 1.48, "learning_rate": 1.3754705163178291e-05, "loss": 0.3598, "step": 8468 }, { "epoch": 1.48, "learning_rate": 1.3749284660991002e-05, "loss": 0.3693, "step": 8469 }, { "epoch": 1.48, "learning_rate": 1.3743865432345432e-05, "loss": 0.3546, "step": 8470 }, { "epoch": 1.48, "learning_rate": 1.3738447477743497e-05, "loss": 0.3483, "step": 8471 }, { "epoch": 1.48, "learning_rate": 1.3733030797686994e-05, "loss": 0.3636, "step": 8472 }, { "epoch": 1.48, "learning_rate": 1.3727615392677584e-05, "loss": 0.3579, "step": 8473 }, { "epoch": 1.48, "learning_rate": 1.3722201263216842e-05, "loss": 0.3566, "step": 8474 }, { "epoch": 1.48, "learning_rate": 1.3716788409806186e-05, "loss": 0.3483, "step": 8475 }, { "epoch": 1.48, "learning_rate": 1.3711376832946945e-05, "loss": 0.3595, "step": 8476 }, { "epoch": 1.48, "learning_rate": 1.3705966533140324e-05, "loss": 0.3373, "step": 8477 }, { "epoch": 1.48, "learning_rate": 1.3700557510887385e-05, "loss": 0.3566, "step": 8478 }, { "epoch": 1.48, "learning_rate": 1.3695149766689109e-05, "loss": 0.367, "step": 8479 }, { "epoch": 1.48, "learning_rate": 1.3689743301046348e-05, "loss": 0.3566, "step": 8480 }, { "epoch": 1.48, "learning_rate": 1.3684338114459816e-05, "loss": 0.3541, "step": 8481 }, { "epoch": 1.48, "learning_rate": 1.3678934207430124e-05, "loss": 0.3739, "step": 8482 }, { "epoch": 1.48, "learning_rate": 1.367353158045777e-05, "loss": 0.35, "step": 8483 }, { "epoch": 1.48, "learning_rate": 1.3668130234043114e-05, "loss": 0.3579, "step": 8484 }, { "epoch": 1.48, "learning_rate": 1.366273016868642e-05, "loss": 0.3457, "step": 8485 }, { "epoch": 1.48, "learning_rate": 1.365733138488781e-05, "loss": 0.3611, "step": 8486 }, { "epoch": 1.48, "learning_rate": 1.3651933883147316e-05, "loss": 0.3372, "step": 8487 }, { "epoch": 1.48, "learning_rate": 1.3646537663964831e-05, "loss": 0.3375, "step": 8488 }, { "epoch": 1.48, "learning_rate": 1.3641142727840111e-05, "loss": 0.3621, "step": 8489 }, { "epoch": 1.48, "learning_rate": 1.3635749075272853e-05, "loss": 0.3693, "step": 8490 }, { "epoch": 1.48, "learning_rate": 1.3630356706762576e-05, "loss": 0.3599, "step": 8491 }, { "epoch": 1.48, "learning_rate": 1.3624965622808696e-05, "loss": 0.3511, "step": 8492 }, { "epoch": 1.48, "learning_rate": 1.3619575823910536e-05, "loss": 0.3578, "step": 8493 }, { "epoch": 1.48, "learning_rate": 1.3614187310567266e-05, "loss": 0.3467, "step": 8494 }, { "epoch": 1.48, "learning_rate": 1.3608800083277944e-05, "loss": 0.3651, "step": 8495 }, { "epoch": 1.48, "learning_rate": 1.3603414142541526e-05, "loss": 0.3502, "step": 8496 }, { "epoch": 1.48, "learning_rate": 1.3598029488856844e-05, "loss": 0.345, "step": 8497 }, { "epoch": 1.48, "learning_rate": 1.3592646122722599e-05, "loss": 0.3612, "step": 8498 }, { "epoch": 1.48, "learning_rate": 1.3587264044637372e-05, "loss": 0.3655, "step": 8499 }, { "epoch": 1.48, "learning_rate": 1.3581883255099633e-05, "loss": 0.3539, "step": 8500 }, { "epoch": 1.48, "learning_rate": 1.3576503754607747e-05, "loss": 0.366, "step": 8501 }, { "epoch": 1.48, "learning_rate": 1.3571125543659935e-05, "loss": 0.3453, "step": 8502 }, { "epoch": 1.48, "learning_rate": 1.3565748622754292e-05, "loss": 0.3734, "step": 8503 }, { "epoch": 1.48, "learning_rate": 1.3560372992388837e-05, "loss": 0.3524, "step": 8504 }, { "epoch": 1.48, "learning_rate": 1.3554998653061424e-05, "loss": 0.3504, "step": 8505 }, { "epoch": 1.48, "learning_rate": 1.3549625605269795e-05, "loss": 0.3616, "step": 8506 }, { "epoch": 1.48, "learning_rate": 1.3544253849511606e-05, "loss": 0.366, "step": 8507 }, { "epoch": 1.48, "learning_rate": 1.3538883386284361e-05, "loss": 0.353, "step": 8508 }, { "epoch": 1.48, "learning_rate": 1.3533514216085447e-05, "loss": 0.3437, "step": 8509 }, { "epoch": 1.48, "learning_rate": 1.3528146339412146e-05, "loss": 0.3702, "step": 8510 }, { "epoch": 1.48, "learning_rate": 1.35227797567616e-05, "loss": 0.3648, "step": 8511 }, { "epoch": 1.48, "learning_rate": 1.3517414468630857e-05, "loss": 0.3526, "step": 8512 }, { "epoch": 1.48, "learning_rate": 1.3512050475516813e-05, "loss": 0.3723, "step": 8513 }, { "epoch": 1.48, "learning_rate": 1.3506687777916282e-05, "loss": 0.3601, "step": 8514 }, { "epoch": 1.48, "learning_rate": 1.3501326376325929e-05, "loss": 0.3484, "step": 8515 }, { "epoch": 1.48, "learning_rate": 1.3495966271242294e-05, "loss": 0.3442, "step": 8516 }, { "epoch": 1.48, "learning_rate": 1.3490607463161825e-05, "loss": 0.3657, "step": 8517 }, { "epoch": 1.49, "learning_rate": 1.3485249952580842e-05, "loss": 0.3702, "step": 8518 }, { "epoch": 1.49, "learning_rate": 1.3479893739995525e-05, "loss": 0.3541, "step": 8519 }, { "epoch": 1.49, "learning_rate": 1.347453882590195e-05, "loss": 0.3607, "step": 8520 }, { "epoch": 1.49, "learning_rate": 1.3469185210796076e-05, "loss": 0.3448, "step": 8521 }, { "epoch": 1.49, "learning_rate": 1.3463832895173728e-05, "loss": 0.3549, "step": 8522 }, { "epoch": 1.49, "learning_rate": 1.3458481879530616e-05, "loss": 0.3708, "step": 8523 }, { "epoch": 1.49, "learning_rate": 1.3453132164362345e-05, "loss": 0.3578, "step": 8524 }, { "epoch": 1.49, "learning_rate": 1.3447783750164378e-05, "loss": 0.3638, "step": 8525 }, { "epoch": 1.49, "learning_rate": 1.3442436637432072e-05, "loss": 0.3527, "step": 8526 }, { "epoch": 1.49, "learning_rate": 1.3437090826660636e-05, "loss": 0.358, "step": 8527 }, { "epoch": 1.49, "learning_rate": 1.34317463183452e-05, "loss": 0.3654, "step": 8528 }, { "epoch": 1.49, "learning_rate": 1.3426403112980757e-05, "loss": 0.3601, "step": 8529 }, { "epoch": 1.49, "learning_rate": 1.3421061211062158e-05, "loss": 0.3556, "step": 8530 }, { "epoch": 1.49, "learning_rate": 1.3415720613084166e-05, "loss": 0.3645, "step": 8531 }, { "epoch": 1.49, "learning_rate": 1.3410381319541402e-05, "loss": 0.3686, "step": 8532 }, { "epoch": 1.49, "learning_rate": 1.3405043330928363e-05, "loss": 0.3467, "step": 8533 }, { "epoch": 1.49, "learning_rate": 1.3399706647739447e-05, "loss": 0.3563, "step": 8534 }, { "epoch": 1.49, "learning_rate": 1.339437127046892e-05, "loss": 0.3621, "step": 8535 }, { "epoch": 1.49, "learning_rate": 1.3389037199610915e-05, "loss": 0.3468, "step": 8536 }, { "epoch": 1.49, "learning_rate": 1.3383704435659458e-05, "loss": 0.3438, "step": 8537 }, { "epoch": 1.49, "learning_rate": 1.3378372979108456e-05, "loss": 0.3462, "step": 8538 }, { "epoch": 1.49, "learning_rate": 1.3373042830451681e-05, "loss": 0.3499, "step": 8539 }, { "epoch": 1.49, "learning_rate": 1.33677139901828e-05, "loss": 0.3542, "step": 8540 }, { "epoch": 1.49, "learning_rate": 1.336238645879534e-05, "loss": 0.3564, "step": 8541 }, { "epoch": 1.49, "learning_rate": 1.3357060236782736e-05, "loss": 0.3638, "step": 8542 }, { "epoch": 1.49, "learning_rate": 1.3351735324638268e-05, "loss": 0.362, "step": 8543 }, { "epoch": 1.49, "learning_rate": 1.3346411722855102e-05, "loss": 0.363, "step": 8544 }, { "epoch": 1.49, "learning_rate": 1.3341089431926315e-05, "loss": 0.3534, "step": 8545 }, { "epoch": 1.49, "learning_rate": 1.3335768452344826e-05, "loss": 0.3625, "step": 8546 }, { "epoch": 1.49, "learning_rate": 1.3330448784603444e-05, "loss": 0.3542, "step": 8547 }, { "epoch": 1.49, "learning_rate": 1.3325130429194865e-05, "loss": 0.3746, "step": 8548 }, { "epoch": 1.49, "learning_rate": 1.331981338661165e-05, "loss": 0.3592, "step": 8549 }, { "epoch": 1.49, "learning_rate": 1.3314497657346234e-05, "loss": 0.3592, "step": 8550 }, { "epoch": 1.49, "learning_rate": 1.3309183241890954e-05, "loss": 0.3472, "step": 8551 }, { "epoch": 1.49, "learning_rate": 1.3303870140738015e-05, "loss": 0.3571, "step": 8552 }, { "epoch": 1.49, "learning_rate": 1.3298558354379491e-05, "loss": 0.3492, "step": 8553 }, { "epoch": 1.49, "learning_rate": 1.3293247883307335e-05, "loss": 0.3654, "step": 8554 }, { "epoch": 1.49, "learning_rate": 1.3287938728013391e-05, "loss": 0.3397, "step": 8555 }, { "epoch": 1.49, "learning_rate": 1.3282630888989378e-05, "loss": 0.3574, "step": 8556 }, { "epoch": 1.49, "learning_rate": 1.3277324366726885e-05, "loss": 0.3535, "step": 8557 }, { "epoch": 1.49, "learning_rate": 1.327201916171737e-05, "loss": 0.3511, "step": 8558 }, { "epoch": 1.49, "learning_rate": 1.3266715274452208e-05, "loss": 0.3471, "step": 8559 }, { "epoch": 1.49, "learning_rate": 1.3261412705422604e-05, "loss": 0.3496, "step": 8560 }, { "epoch": 1.49, "learning_rate": 1.3256111455119661e-05, "loss": 0.3631, "step": 8561 }, { "epoch": 1.49, "learning_rate": 1.3250811524034382e-05, "loss": 0.356, "step": 8562 }, { "epoch": 1.49, "learning_rate": 1.3245512912657618e-05, "loss": 0.3582, "step": 8563 }, { "epoch": 1.49, "learning_rate": 1.3240215621480095e-05, "loss": 0.3732, "step": 8564 }, { "epoch": 1.49, "learning_rate": 1.323491965099245e-05, "loss": 0.35, "step": 8565 }, { "epoch": 1.49, "learning_rate": 1.3229625001685158e-05, "loss": 0.3541, "step": 8566 }, { "epoch": 1.49, "learning_rate": 1.3224331674048609e-05, "loss": 0.3504, "step": 8567 }, { "epoch": 1.49, "learning_rate": 1.3219039668573036e-05, "loss": 0.3617, "step": 8568 }, { "epoch": 1.49, "learning_rate": 1.3213748985748575e-05, "loss": 0.3632, "step": 8569 }, { "epoch": 1.49, "learning_rate": 1.3208459626065223e-05, "loss": 0.3477, "step": 8570 }, { "epoch": 1.49, "learning_rate": 1.3203171590012864e-05, "loss": 0.3579, "step": 8571 }, { "epoch": 1.49, "learning_rate": 1.319788487808125e-05, "loss": 0.3412, "step": 8572 }, { "epoch": 1.49, "learning_rate": 1.3192599490760041e-05, "loss": 0.3503, "step": 8573 }, { "epoch": 1.49, "learning_rate": 1.3187315428538726e-05, "loss": 0.3634, "step": 8574 }, { "epoch": 1.5, "learning_rate": 1.3182032691906704e-05, "loss": 0.3291, "step": 8575 }, { "epoch": 1.5, "learning_rate": 1.3176751281353245e-05, "loss": 0.3457, "step": 8576 }, { "epoch": 1.5, "learning_rate": 1.3171471197367496e-05, "loss": 0.3397, "step": 8577 }, { "epoch": 1.5, "learning_rate": 1.3166192440438465e-05, "loss": 0.3521, "step": 8578 }, { "epoch": 1.5, "learning_rate": 1.3160915011055063e-05, "loss": 0.3652, "step": 8579 }, { "epoch": 1.5, "learning_rate": 1.3155638909706074e-05, "loss": 0.3467, "step": 8580 }, { "epoch": 1.5, "learning_rate": 1.3150364136880145e-05, "loss": 0.3675, "step": 8581 }, { "epoch": 1.5, "learning_rate": 1.3145090693065795e-05, "loss": 0.3532, "step": 8582 }, { "epoch": 1.5, "learning_rate": 1.3139818578751443e-05, "loss": 0.3707, "step": 8583 }, { "epoch": 1.5, "learning_rate": 1.3134547794425372e-05, "loss": 0.3574, "step": 8584 }, { "epoch": 1.5, "learning_rate": 1.312927834057574e-05, "loss": 0.3379, "step": 8585 }, { "epoch": 1.5, "learning_rate": 1.3124010217690597e-05, "loss": 0.367, "step": 8586 }, { "epoch": 1.5, "learning_rate": 1.3118743426257846e-05, "loss": 0.3616, "step": 8587 }, { "epoch": 1.5, "learning_rate": 1.311347796676527e-05, "loss": 0.3565, "step": 8588 }, { "epoch": 1.5, "learning_rate": 1.3108213839700547e-05, "loss": 0.3622, "step": 8589 }, { "epoch": 1.5, "learning_rate": 1.3102951045551232e-05, "loss": 0.3554, "step": 8590 }, { "epoch": 1.5, "learning_rate": 1.3097689584804735e-05, "loss": 0.3483, "step": 8591 }, { "epoch": 1.5, "learning_rate": 1.3092429457948344e-05, "loss": 0.3725, "step": 8592 }, { "epoch": 1.5, "learning_rate": 1.308717066546925e-05, "loss": 0.3433, "step": 8593 }, { "epoch": 1.5, "learning_rate": 1.308191320785449e-05, "loss": 0.3615, "step": 8594 }, { "epoch": 1.5, "learning_rate": 1.3076657085591008e-05, "loss": 0.3657, "step": 8595 }, { "epoch": 1.5, "learning_rate": 1.3071402299165583e-05, "loss": 0.3609, "step": 8596 }, { "epoch": 1.5, "learning_rate": 1.3066148849064918e-05, "loss": 0.3344, "step": 8597 }, { "epoch": 1.5, "learning_rate": 1.3060896735775557e-05, "loss": 0.3579, "step": 8598 }, { "epoch": 1.5, "learning_rate": 1.305564595978393e-05, "loss": 0.3569, "step": 8599 }, { "epoch": 1.5, "learning_rate": 1.3050396521576343e-05, "loss": 0.3472, "step": 8600 }, { "epoch": 1.5, "learning_rate": 1.3045148421638994e-05, "loss": 0.3564, "step": 8601 }, { "epoch": 1.5, "learning_rate": 1.3039901660457926e-05, "loss": 0.3505, "step": 8602 }, { "epoch": 1.5, "learning_rate": 1.3034656238519095e-05, "loss": 0.3677, "step": 8603 }, { "epoch": 1.5, "learning_rate": 1.3029412156308296e-05, "loss": 0.4002, "step": 8604 }, { "epoch": 1.5, "learning_rate": 1.3024169414311216e-05, "loss": 0.3492, "step": 8605 }, { "epoch": 1.5, "learning_rate": 1.3018928013013426e-05, "loss": 0.3514, "step": 8606 }, { "epoch": 1.5, "learning_rate": 1.3013687952900375e-05, "loss": 0.3615, "step": 8607 }, { "epoch": 1.5, "learning_rate": 1.3008449234457364e-05, "loss": 0.3621, "step": 8608 }, { "epoch": 1.5, "learning_rate": 1.3003211858169589e-05, "loss": 0.3569, "step": 8609 }, { "epoch": 1.5, "learning_rate": 1.2997975824522115e-05, "loss": 0.3462, "step": 8610 }, { "epoch": 1.5, "learning_rate": 1.2992741133999883e-05, "loss": 0.3651, "step": 8611 }, { "epoch": 1.5, "learning_rate": 1.2987507787087722e-05, "loss": 0.3621, "step": 8612 }, { "epoch": 1.5, "learning_rate": 1.2982275784270309e-05, "loss": 0.355, "step": 8613 }, { "epoch": 1.5, "learning_rate": 1.2977045126032234e-05, "loss": 0.3677, "step": 8614 }, { "epoch": 1.5, "learning_rate": 1.2971815812857927e-05, "loss": 0.3604, "step": 8615 }, { "epoch": 1.5, "learning_rate": 1.2966587845231703e-05, "loss": 0.3534, "step": 8616 }, { "epoch": 1.5, "learning_rate": 1.2961361223637766e-05, "loss": 0.3676, "step": 8617 }, { "epoch": 1.5, "learning_rate": 1.2956135948560194e-05, "loss": 0.3657, "step": 8618 }, { "epoch": 1.5, "learning_rate": 1.295091202048293e-05, "loss": 0.3555, "step": 8619 }, { "epoch": 1.5, "learning_rate": 1.294568943988978e-05, "loss": 0.3505, "step": 8620 }, { "epoch": 1.5, "learning_rate": 1.2940468207264458e-05, "loss": 0.3639, "step": 8621 }, { "epoch": 1.5, "learning_rate": 1.2935248323090527e-05, "loss": 0.3562, "step": 8622 }, { "epoch": 1.5, "learning_rate": 1.2930029787851434e-05, "loss": 0.3477, "step": 8623 }, { "epoch": 1.5, "learning_rate": 1.2924812602030509e-05, "loss": 0.359, "step": 8624 }, { "epoch": 1.5, "learning_rate": 1.2919596766110945e-05, "loss": 0.3512, "step": 8625 }, { "epoch": 1.5, "learning_rate": 1.2914382280575803e-05, "loss": 0.3714, "step": 8626 }, { "epoch": 1.5, "learning_rate": 1.2909169145908046e-05, "loss": 0.3648, "step": 8627 }, { "epoch": 1.5, "learning_rate": 1.2903957362590488e-05, "loss": 0.359, "step": 8628 }, { "epoch": 1.5, "learning_rate": 1.2898746931105832e-05, "loss": 0.358, "step": 8629 }, { "epoch": 1.5, "learning_rate": 1.2893537851936632e-05, "loss": 0.3528, "step": 8630 }, { "epoch": 1.5, "learning_rate": 1.2888330125565357e-05, "loss": 0.3578, "step": 8631 }, { "epoch": 1.5, "learning_rate": 1.2883123752474316e-05, "loss": 0.3529, "step": 8632 }, { "epoch": 1.51, "learning_rate": 1.2877918733145697e-05, "loss": 0.3515, "step": 8633 }, { "epoch": 1.51, "learning_rate": 1.287271506806158e-05, "loss": 0.3541, "step": 8634 }, { "epoch": 1.51, "learning_rate": 1.2867512757703915e-05, "loss": 0.3841, "step": 8635 }, { "epoch": 1.51, "learning_rate": 1.2862311802554516e-05, "loss": 0.3601, "step": 8636 }, { "epoch": 1.51, "learning_rate": 1.2857112203095067e-05, "loss": 0.3514, "step": 8637 }, { "epoch": 1.51, "learning_rate": 1.2851913959807143e-05, "loss": 0.3704, "step": 8638 }, { "epoch": 1.51, "learning_rate": 1.2846717073172198e-05, "loss": 0.3733, "step": 8639 }, { "epoch": 1.51, "learning_rate": 1.284152154367153e-05, "loss": 0.3567, "step": 8640 }, { "epoch": 1.51, "learning_rate": 1.2836327371786347e-05, "loss": 0.3423, "step": 8641 }, { "epoch": 1.51, "learning_rate": 1.2831134557997707e-05, "loss": 0.3578, "step": 8642 }, { "epoch": 1.51, "learning_rate": 1.2825943102786541e-05, "loss": 0.3475, "step": 8643 }, { "epoch": 1.51, "learning_rate": 1.2820753006633673e-05, "loss": 0.369, "step": 8644 }, { "epoch": 1.51, "learning_rate": 1.2815564270019797e-05, "loss": 0.3426, "step": 8645 }, { "epoch": 1.51, "learning_rate": 1.2810376893425466e-05, "loss": 0.3522, "step": 8646 }, { "epoch": 1.51, "learning_rate": 1.280519087733111e-05, "loss": 0.3546, "step": 8647 }, { "epoch": 1.51, "learning_rate": 1.2800006222217053e-05, "loss": 0.3698, "step": 8648 }, { "epoch": 1.51, "learning_rate": 1.2794822928563465e-05, "loss": 0.34, "step": 8649 }, { "epoch": 1.51, "learning_rate": 1.2789640996850422e-05, "loss": 0.357, "step": 8650 }, { "epoch": 1.51, "learning_rate": 1.2784460427557837e-05, "loss": 0.356, "step": 8651 }, { "epoch": 1.51, "learning_rate": 1.2779281221165535e-05, "loss": 0.3574, "step": 8652 }, { "epoch": 1.51, "learning_rate": 1.2774103378153179e-05, "loss": 0.3652, "step": 8653 }, { "epoch": 1.51, "learning_rate": 1.2768926899000325e-05, "loss": 0.3517, "step": 8654 }, { "epoch": 1.51, "learning_rate": 1.2763751784186402e-05, "loss": 0.3495, "step": 8655 }, { "epoch": 1.51, "learning_rate": 1.2758578034190714e-05, "loss": 0.3692, "step": 8656 }, { "epoch": 1.51, "learning_rate": 1.2753405649492439e-05, "loss": 0.3438, "step": 8657 }, { "epoch": 1.51, "learning_rate": 1.2748234630570605e-05, "loss": 0.3574, "step": 8658 }, { "epoch": 1.51, "learning_rate": 1.2743064977904157e-05, "loss": 0.3476, "step": 8659 }, { "epoch": 1.51, "learning_rate": 1.2737896691971868e-05, "loss": 0.3635, "step": 8660 }, { "epoch": 1.51, "learning_rate": 1.2732729773252419e-05, "loss": 0.365, "step": 8661 }, { "epoch": 1.51, "learning_rate": 1.2727564222224354e-05, "loss": 0.3509, "step": 8662 }, { "epoch": 1.51, "learning_rate": 1.2722400039366089e-05, "loss": 0.3527, "step": 8663 }, { "epoch": 1.51, "learning_rate": 1.2717237225155893e-05, "loss": 0.3549, "step": 8664 }, { "epoch": 1.51, "learning_rate": 1.271207578007195e-05, "loss": 0.3704, "step": 8665 }, { "epoch": 1.51, "learning_rate": 1.2706915704592276e-05, "loss": 0.355, "step": 8666 }, { "epoch": 1.51, "learning_rate": 1.2701756999194792e-05, "loss": 0.3678, "step": 8667 }, { "epoch": 1.51, "learning_rate": 1.2696599664357272e-05, "loss": 0.3585, "step": 8668 }, { "epoch": 1.51, "learning_rate": 1.2691443700557375e-05, "loss": 0.3664, "step": 8669 }, { "epoch": 1.51, "learning_rate": 1.2686289108272626e-05, "loss": 0.366, "step": 8670 }, { "epoch": 1.51, "learning_rate": 1.2681135887980416e-05, "loss": 0.3536, "step": 8671 }, { "epoch": 1.51, "learning_rate": 1.2675984040158027e-05, "loss": 0.3524, "step": 8672 }, { "epoch": 1.51, "learning_rate": 1.2670833565282609e-05, "loss": 0.3479, "step": 8673 }, { "epoch": 1.51, "learning_rate": 1.2665684463831169e-05, "loss": 0.3521, "step": 8674 }, { "epoch": 1.51, "learning_rate": 1.2660536736280603e-05, "loss": 0.3667, "step": 8675 }, { "epoch": 1.51, "learning_rate": 1.2655390383107676e-05, "loss": 0.3532, "step": 8676 }, { "epoch": 1.51, "learning_rate": 1.2650245404789018e-05, "loss": 0.3759, "step": 8677 }, { "epoch": 1.51, "learning_rate": 1.2645101801801149e-05, "loss": 0.3506, "step": 8678 }, { "epoch": 1.51, "learning_rate": 1.2639959574620452e-05, "loss": 0.3473, "step": 8679 }, { "epoch": 1.51, "learning_rate": 1.2634818723723174e-05, "loss": 0.3426, "step": 8680 }, { "epoch": 1.51, "learning_rate": 1.2629679249585435e-05, "loss": 0.3452, "step": 8681 }, { "epoch": 1.51, "learning_rate": 1.2624541152683255e-05, "loss": 0.3561, "step": 8682 }, { "epoch": 1.51, "learning_rate": 1.2619404433492483e-05, "loss": 0.3432, "step": 8683 }, { "epoch": 1.51, "learning_rate": 1.2614269092488883e-05, "loss": 0.3368, "step": 8684 }, { "epoch": 1.51, "learning_rate": 1.2609135130148059e-05, "loss": 0.3786, "step": 8685 }, { "epoch": 1.51, "learning_rate": 1.2604002546945509e-05, "loss": 0.3434, "step": 8686 }, { "epoch": 1.51, "learning_rate": 1.2598871343356594e-05, "loss": 0.3526, "step": 8687 }, { "epoch": 1.51, "learning_rate": 1.2593741519856534e-05, "loss": 0.3483, "step": 8688 }, { "epoch": 1.51, "learning_rate": 1.2588613076920445e-05, "loss": 0.3529, "step": 8689 }, { "epoch": 1.52, "learning_rate": 1.2583486015023311e-05, "loss": 0.368, "step": 8690 }, { "epoch": 1.52, "learning_rate": 1.2578360334639977e-05, "loss": 0.3595, "step": 8691 }, { "epoch": 1.52, "learning_rate": 1.2573236036245157e-05, "loss": 0.341, "step": 8692 }, { "epoch": 1.52, "learning_rate": 1.2568113120313454e-05, "loss": 0.373, "step": 8693 }, { "epoch": 1.52, "learning_rate": 1.2562991587319329e-05, "loss": 0.3642, "step": 8694 }, { "epoch": 1.52, "learning_rate": 1.2557871437737118e-05, "loss": 0.3478, "step": 8695 }, { "epoch": 1.52, "learning_rate": 1.255275267204105e-05, "loss": 0.3417, "step": 8696 }, { "epoch": 1.52, "learning_rate": 1.2547635290705184e-05, "loss": 0.3548, "step": 8697 }, { "epoch": 1.52, "learning_rate": 1.2542519294203482e-05, "loss": 0.3591, "step": 8698 }, { "epoch": 1.52, "learning_rate": 1.2537404683009763e-05, "loss": 0.371, "step": 8699 }, { "epoch": 1.52, "learning_rate": 1.253229145759774e-05, "loss": 0.3519, "step": 8700 }, { "epoch": 1.52, "learning_rate": 1.2527179618440968e-05, "loss": 0.348, "step": 8701 }, { "epoch": 1.52, "learning_rate": 1.2522069166012884e-05, "loss": 0.3595, "step": 8702 }, { "epoch": 1.52, "learning_rate": 1.2516960100786812e-05, "loss": 0.3666, "step": 8703 }, { "epoch": 1.52, "learning_rate": 1.2511852423235928e-05, "loss": 0.3662, "step": 8704 }, { "epoch": 1.52, "learning_rate": 1.2506746133833285e-05, "loss": 0.3413, "step": 8705 }, { "epoch": 1.52, "learning_rate": 1.2501641233051808e-05, "loss": 0.3578, "step": 8706 }, { "epoch": 1.52, "learning_rate": 1.2496537721364308e-05, "loss": 0.3561, "step": 8707 }, { "epoch": 1.52, "learning_rate": 1.2491435599243444e-05, "loss": 0.3533, "step": 8708 }, { "epoch": 1.52, "learning_rate": 1.2486334867161747e-05, "loss": 0.3624, "step": 8709 }, { "epoch": 1.52, "learning_rate": 1.2481235525591642e-05, "loss": 0.3583, "step": 8710 }, { "epoch": 1.52, "learning_rate": 1.2476137575005413e-05, "loss": 0.3575, "step": 8711 }, { "epoch": 1.52, "learning_rate": 1.2471041015875205e-05, "loss": 0.3541, "step": 8712 }, { "epoch": 1.52, "learning_rate": 1.2465945848673042e-05, "loss": 0.3547, "step": 8713 }, { "epoch": 1.52, "learning_rate": 1.2460852073870833e-05, "loss": 0.3558, "step": 8714 }, { "epoch": 1.52, "learning_rate": 1.2455759691940335e-05, "loss": 0.3573, "step": 8715 }, { "epoch": 1.52, "learning_rate": 1.2450668703353171e-05, "loss": 0.351, "step": 8716 }, { "epoch": 1.52, "learning_rate": 1.2445579108580883e-05, "loss": 0.3691, "step": 8717 }, { "epoch": 1.52, "learning_rate": 1.2440490908094832e-05, "loss": 0.3635, "step": 8718 }, { "epoch": 1.52, "learning_rate": 1.2435404102366267e-05, "loss": 0.3435, "step": 8719 }, { "epoch": 1.52, "learning_rate": 1.2430318691866321e-05, "loss": 0.3654, "step": 8720 }, { "epoch": 1.52, "learning_rate": 1.2425234677065972e-05, "loss": 0.3786, "step": 8721 }, { "epoch": 1.52, "learning_rate": 1.2420152058436095e-05, "loss": 0.3543, "step": 8722 }, { "epoch": 1.52, "learning_rate": 1.2415070836447414e-05, "loss": 0.3417, "step": 8723 }, { "epoch": 1.52, "learning_rate": 1.2409991011570546e-05, "loss": 0.3597, "step": 8724 }, { "epoch": 1.52, "learning_rate": 1.240491258427596e-05, "loss": 0.3701, "step": 8725 }, { "epoch": 1.52, "learning_rate": 1.239983555503399e-05, "loss": 0.3714, "step": 8726 }, { "epoch": 1.52, "learning_rate": 1.2394759924314864e-05, "loss": 0.3594, "step": 8727 }, { "epoch": 1.52, "learning_rate": 1.2389685692588676e-05, "loss": 0.3505, "step": 8728 }, { "epoch": 1.52, "learning_rate": 1.2384612860325374e-05, "loss": 0.3491, "step": 8729 }, { "epoch": 1.52, "learning_rate": 1.2379541427994777e-05, "loss": 0.353, "step": 8730 }, { "epoch": 1.52, "learning_rate": 1.23744713960666e-05, "loss": 0.3567, "step": 8731 }, { "epoch": 1.52, "learning_rate": 1.2369402765010393e-05, "loss": 0.3381, "step": 8732 }, { "epoch": 1.52, "learning_rate": 1.2364335535295605e-05, "loss": 0.3676, "step": 8733 }, { "epoch": 1.52, "learning_rate": 1.2359269707391554e-05, "loss": 0.3523, "step": 8734 }, { "epoch": 1.52, "learning_rate": 1.2354205281767405e-05, "loss": 0.3515, "step": 8735 }, { "epoch": 1.52, "learning_rate": 1.2349142258892203e-05, "loss": 0.3538, "step": 8736 }, { "epoch": 1.52, "learning_rate": 1.2344080639234884e-05, "loss": 0.358, "step": 8737 }, { "epoch": 1.52, "learning_rate": 1.2339020423264217e-05, "loss": 0.3658, "step": 8738 }, { "epoch": 1.52, "learning_rate": 1.2333961611448879e-05, "loss": 0.3542, "step": 8739 }, { "epoch": 1.52, "learning_rate": 1.2328904204257385e-05, "loss": 0.3432, "step": 8740 }, { "epoch": 1.52, "learning_rate": 1.2323848202158147e-05, "loss": 0.3502, "step": 8741 }, { "epoch": 1.52, "learning_rate": 1.2318793605619423e-05, "loss": 0.3611, "step": 8742 }, { "epoch": 1.52, "learning_rate": 1.2313740415109353e-05, "loss": 0.3682, "step": 8743 }, { "epoch": 1.52, "learning_rate": 1.2308688631095945e-05, "loss": 0.3544, "step": 8744 }, { "epoch": 1.52, "learning_rate": 1.2303638254047084e-05, "loss": 0.3742, "step": 8745 }, { "epoch": 1.52, "learning_rate": 1.2298589284430517e-05, "loss": 0.3645, "step": 8746 }, { "epoch": 1.53, "learning_rate": 1.229354172271385e-05, "loss": 0.3715, "step": 8747 }, { "epoch": 1.53, "learning_rate": 1.2288495569364582e-05, "loss": 0.356, "step": 8748 }, { "epoch": 1.53, "learning_rate": 1.228345082485006e-05, "loss": 0.3552, "step": 8749 }, { "epoch": 1.53, "learning_rate": 1.2278407489637522e-05, "loss": 0.3485, "step": 8750 }, { "epoch": 1.53, "learning_rate": 1.227336556419405e-05, "loss": 0.3609, "step": 8751 }, { "epoch": 1.53, "learning_rate": 1.226832504898662e-05, "loss": 0.3656, "step": 8752 }, { "epoch": 1.53, "learning_rate": 1.2263285944482066e-05, "loss": 0.3423, "step": 8753 }, { "epoch": 1.53, "learning_rate": 1.225824825114707e-05, "loss": 0.3708, "step": 8754 }, { "epoch": 1.53, "learning_rate": 1.225321196944824e-05, "loss": 0.3565, "step": 8755 }, { "epoch": 1.53, "learning_rate": 1.2248177099851999e-05, "loss": 0.3667, "step": 8756 }, { "epoch": 1.53, "learning_rate": 1.2243143642824651e-05, "loss": 0.3671, "step": 8757 }, { "epoch": 1.53, "learning_rate": 1.2238111598832398e-05, "loss": 0.3745, "step": 8758 }, { "epoch": 1.53, "learning_rate": 1.2233080968341276e-05, "loss": 0.3518, "step": 8759 }, { "epoch": 1.53, "learning_rate": 1.2228051751817198e-05, "loss": 0.3591, "step": 8760 }, { "epoch": 1.53, "learning_rate": 1.222302394972596e-05, "loss": 0.3462, "step": 8761 }, { "epoch": 1.53, "learning_rate": 1.2217997562533228e-05, "loss": 0.3592, "step": 8762 }, { "epoch": 1.53, "learning_rate": 1.2212972590704514e-05, "loss": 0.3636, "step": 8763 }, { "epoch": 1.53, "learning_rate": 1.2207949034705213e-05, "loss": 0.3499, "step": 8764 }, { "epoch": 1.53, "learning_rate": 1.2202926895000593e-05, "loss": 0.3402, "step": 8765 }, { "epoch": 1.53, "learning_rate": 1.2197906172055792e-05, "loss": 0.3527, "step": 8766 }, { "epoch": 1.53, "learning_rate": 1.219288686633581e-05, "loss": 0.3529, "step": 8767 }, { "epoch": 1.53, "learning_rate": 1.2187868978305503e-05, "loss": 0.3584, "step": 8768 }, { "epoch": 1.53, "learning_rate": 1.2182852508429624e-05, "loss": 0.3642, "step": 8769 }, { "epoch": 1.53, "learning_rate": 1.217783745717278e-05, "loss": 0.3602, "step": 8770 }, { "epoch": 1.53, "learning_rate": 1.2172823824999426e-05, "loss": 0.3454, "step": 8771 }, { "epoch": 1.53, "learning_rate": 1.2167811612373939e-05, "loss": 0.3768, "step": 8772 }, { "epoch": 1.53, "learning_rate": 1.2162800819760516e-05, "loss": 0.366, "step": 8773 }, { "epoch": 1.53, "learning_rate": 1.2157791447623232e-05, "loss": 0.3585, "step": 8774 }, { "epoch": 1.53, "learning_rate": 1.2152783496426051e-05, "loss": 0.3534, "step": 8775 }, { "epoch": 1.53, "learning_rate": 1.2147776966632776e-05, "loss": 0.3579, "step": 8776 }, { "epoch": 1.53, "learning_rate": 1.214277185870711e-05, "loss": 0.355, "step": 8777 }, { "epoch": 1.53, "learning_rate": 1.2137768173112594e-05, "loss": 0.3593, "step": 8778 }, { "epoch": 1.53, "learning_rate": 1.2132765910312664e-05, "loss": 0.3538, "step": 8779 }, { "epoch": 1.53, "learning_rate": 1.2127765070770606e-05, "loss": 0.3609, "step": 8780 }, { "epoch": 1.53, "learning_rate": 1.2122765654949571e-05, "loss": 0.3397, "step": 8781 }, { "epoch": 1.53, "learning_rate": 1.2117767663312596e-05, "loss": 0.3564, "step": 8782 }, { "epoch": 1.53, "learning_rate": 1.2112771096322582e-05, "loss": 0.3513, "step": 8783 }, { "epoch": 1.53, "learning_rate": 1.2107775954442286e-05, "loss": 0.3344, "step": 8784 }, { "epoch": 1.53, "learning_rate": 1.2102782238134334e-05, "loss": 0.3714, "step": 8785 }, { "epoch": 1.53, "learning_rate": 1.209778994786124e-05, "loss": 0.3444, "step": 8786 }, { "epoch": 1.53, "learning_rate": 1.2092799084085365e-05, "loss": 0.3621, "step": 8787 }, { "epoch": 1.53, "learning_rate": 1.208780964726893e-05, "loss": 0.3531, "step": 8788 }, { "epoch": 1.53, "learning_rate": 1.2082821637874068e-05, "loss": 0.3576, "step": 8789 }, { "epoch": 1.53, "learning_rate": 1.2077835056362737e-05, "loss": 0.3399, "step": 8790 }, { "epoch": 1.53, "learning_rate": 1.2072849903196776e-05, "loss": 0.3624, "step": 8791 }, { "epoch": 1.53, "learning_rate": 1.206786617883788e-05, "loss": 0.36, "step": 8792 }, { "epoch": 1.53, "learning_rate": 1.2062883883747636e-05, "loss": 0.3598, "step": 8793 }, { "epoch": 1.53, "learning_rate": 1.2057903018387495e-05, "loss": 0.3622, "step": 8794 }, { "epoch": 1.53, "learning_rate": 1.2052923583218743e-05, "loss": 0.3435, "step": 8795 }, { "epoch": 1.53, "learning_rate": 1.2047945578702583e-05, "loss": 0.3562, "step": 8796 }, { "epoch": 1.53, "learning_rate": 1.2042969005300046e-05, "loss": 0.3812, "step": 8797 }, { "epoch": 1.53, "learning_rate": 1.203799386347204e-05, "loss": 0.358, "step": 8798 }, { "epoch": 1.53, "learning_rate": 1.203302015367935e-05, "loss": 0.3712, "step": 8799 }, { "epoch": 1.53, "learning_rate": 1.2028047876382631e-05, "loss": 0.3542, "step": 8800 }, { "epoch": 1.53, "learning_rate": 1.2023077032042391e-05, "loss": 0.3761, "step": 8801 }, { "epoch": 1.53, "learning_rate": 1.2018107621119003e-05, "loss": 0.3436, "step": 8802 }, { "epoch": 1.53, "learning_rate": 1.2013139644072732e-05, "loss": 0.3701, "step": 8803 }, { "epoch": 1.53, "learning_rate": 1.2008173101363676e-05, "loss": 0.3503, "step": 8804 }, { "epoch": 1.54, "learning_rate": 1.200320799345184e-05, "loss": 0.3672, "step": 8805 }, { "epoch": 1.54, "learning_rate": 1.199824432079705e-05, "loss": 0.3801, "step": 8806 }, { "epoch": 1.54, "learning_rate": 1.1993282083859048e-05, "loss": 0.3472, "step": 8807 }, { "epoch": 1.54, "learning_rate": 1.1988321283097406e-05, "loss": 0.3476, "step": 8808 }, { "epoch": 1.54, "learning_rate": 1.198336191897157e-05, "loss": 0.3478, "step": 8809 }, { "epoch": 1.54, "learning_rate": 1.1978403991940862e-05, "loss": 0.3476, "step": 8810 }, { "epoch": 1.54, "learning_rate": 1.1973447502464479e-05, "loss": 0.3606, "step": 8811 }, { "epoch": 1.54, "learning_rate": 1.1968492451001458e-05, "loss": 0.3473, "step": 8812 }, { "epoch": 1.54, "learning_rate": 1.1963538838010729e-05, "loss": 0.3573, "step": 8813 }, { "epoch": 1.54, "learning_rate": 1.1958586663951075e-05, "loss": 0.3667, "step": 8814 }, { "epoch": 1.54, "learning_rate": 1.1953635929281137e-05, "loss": 0.3562, "step": 8815 }, { "epoch": 1.54, "learning_rate": 1.1948686634459446e-05, "loss": 0.3664, "step": 8816 }, { "epoch": 1.54, "learning_rate": 1.1943738779944393e-05, "loss": 0.375, "step": 8817 }, { "epoch": 1.54, "learning_rate": 1.1938792366194222e-05, "loss": 0.3424, "step": 8818 }, { "epoch": 1.54, "learning_rate": 1.1933847393667045e-05, "loss": 0.3609, "step": 8819 }, { "epoch": 1.54, "learning_rate": 1.1928903862820866e-05, "loss": 0.3735, "step": 8820 }, { "epoch": 1.54, "learning_rate": 1.1923961774113517e-05, "loss": 0.3721, "step": 8821 }, { "epoch": 1.54, "learning_rate": 1.1919021128002734e-05, "loss": 0.3649, "step": 8822 }, { "epoch": 1.54, "learning_rate": 1.1914081924946087e-05, "loss": 0.3332, "step": 8823 }, { "epoch": 1.54, "learning_rate": 1.1909144165401041e-05, "loss": 0.3683, "step": 8824 }, { "epoch": 1.54, "learning_rate": 1.1904207849824908e-05, "loss": 0.3564, "step": 8825 }, { "epoch": 1.54, "learning_rate": 1.1899272978674857e-05, "loss": 0.3626, "step": 8826 }, { "epoch": 1.54, "learning_rate": 1.1894339552407963e-05, "loss": 0.3562, "step": 8827 }, { "epoch": 1.54, "learning_rate": 1.188940757148113e-05, "loss": 0.3606, "step": 8828 }, { "epoch": 1.54, "learning_rate": 1.1884477036351137e-05, "loss": 0.3616, "step": 8829 }, { "epoch": 1.54, "learning_rate": 1.1879547947474642e-05, "loss": 0.3612, "step": 8830 }, { "epoch": 1.54, "learning_rate": 1.1874620305308153e-05, "loss": 0.3379, "step": 8831 }, { "epoch": 1.54, "learning_rate": 1.1869694110308045e-05, "loss": 0.3474, "step": 8832 }, { "epoch": 1.54, "learning_rate": 1.186476936293057e-05, "loss": 0.3465, "step": 8833 }, { "epoch": 1.54, "learning_rate": 1.1859846063631848e-05, "loss": 0.3577, "step": 8834 }, { "epoch": 1.54, "learning_rate": 1.1854924212867853e-05, "loss": 0.3413, "step": 8835 }, { "epoch": 1.54, "learning_rate": 1.1850003811094418e-05, "loss": 0.356, "step": 8836 }, { "epoch": 1.54, "learning_rate": 1.1845084858767261e-05, "loss": 0.3612, "step": 8837 }, { "epoch": 1.54, "learning_rate": 1.1840167356341965e-05, "loss": 0.3672, "step": 8838 }, { "epoch": 1.54, "learning_rate": 1.1835251304273963e-05, "loss": 0.3724, "step": 8839 }, { "epoch": 1.54, "learning_rate": 1.1830336703018557e-05, "loss": 0.3782, "step": 8840 }, { "epoch": 1.54, "learning_rate": 1.1825423553030935e-05, "loss": 0.3465, "step": 8841 }, { "epoch": 1.54, "learning_rate": 1.1820511854766126e-05, "loss": 0.3584, "step": 8842 }, { "epoch": 1.54, "learning_rate": 1.1815601608679025e-05, "loss": 0.342, "step": 8843 }, { "epoch": 1.54, "learning_rate": 1.181069281522441e-05, "loss": 0.3464, "step": 8844 }, { "epoch": 1.54, "learning_rate": 1.1805785474856925e-05, "loss": 0.3513, "step": 8845 }, { "epoch": 1.54, "learning_rate": 1.1800879588031062e-05, "loss": 0.3612, "step": 8846 }, { "epoch": 1.54, "learning_rate": 1.1795975155201178e-05, "loss": 0.3553, "step": 8847 }, { "epoch": 1.54, "learning_rate": 1.179107217682151e-05, "loss": 0.3376, "step": 8848 }, { "epoch": 1.54, "learning_rate": 1.1786170653346166e-05, "loss": 0.3496, "step": 8849 }, { "epoch": 1.54, "learning_rate": 1.1781270585229088e-05, "loss": 0.3572, "step": 8850 }, { "epoch": 1.54, "learning_rate": 1.177637197292412e-05, "loss": 0.3504, "step": 8851 }, { "epoch": 1.54, "learning_rate": 1.1771474816884947e-05, "loss": 0.3627, "step": 8852 }, { "epoch": 1.54, "learning_rate": 1.1766579117565116e-05, "loss": 0.3633, "step": 8853 }, { "epoch": 1.54, "learning_rate": 1.1761684875418059e-05, "loss": 0.3467, "step": 8854 }, { "epoch": 1.54, "learning_rate": 1.1756792090897071e-05, "loss": 0.3632, "step": 8855 }, { "epoch": 1.54, "learning_rate": 1.1751900764455294e-05, "loss": 0.3714, "step": 8856 }, { "epoch": 1.54, "learning_rate": 1.1747010896545738e-05, "loss": 0.3461, "step": 8857 }, { "epoch": 1.54, "learning_rate": 1.17421224876213e-05, "loss": 0.3607, "step": 8858 }, { "epoch": 1.54, "learning_rate": 1.1737235538134713e-05, "loss": 0.3573, "step": 8859 }, { "epoch": 1.54, "learning_rate": 1.1732350048538604e-05, "loss": 0.3626, "step": 8860 }, { "epoch": 1.54, "learning_rate": 1.1727466019285432e-05, "loss": 0.3502, "step": 8861 }, { "epoch": 1.55, "learning_rate": 1.1722583450827556e-05, "loss": 0.3444, "step": 8862 }, { "epoch": 1.55, "learning_rate": 1.1717702343617174e-05, "loss": 0.3308, "step": 8863 }, { "epoch": 1.55, "learning_rate": 1.1712822698106345e-05, "loss": 0.3652, "step": 8864 }, { "epoch": 1.55, "learning_rate": 1.1707944514747017e-05, "loss": 0.3561, "step": 8865 }, { "epoch": 1.55, "learning_rate": 1.1703067793990995e-05, "loss": 0.3445, "step": 8866 }, { "epoch": 1.55, "learning_rate": 1.169819253628993e-05, "loss": 0.3527, "step": 8867 }, { "epoch": 1.55, "learning_rate": 1.169331874209536e-05, "loss": 0.3556, "step": 8868 }, { "epoch": 1.55, "learning_rate": 1.1688446411858678e-05, "loss": 0.3395, "step": 8869 }, { "epoch": 1.55, "learning_rate": 1.1683575546031136e-05, "loss": 0.3454, "step": 8870 }, { "epoch": 1.55, "learning_rate": 1.1678706145063852e-05, "loss": 0.3506, "step": 8871 }, { "epoch": 1.55, "learning_rate": 1.1673838209407828e-05, "loss": 0.3701, "step": 8872 }, { "epoch": 1.55, "learning_rate": 1.1668971739513906e-05, "loss": 0.3555, "step": 8873 }, { "epoch": 1.55, "learning_rate": 1.1664106735832794e-05, "loss": 0.3713, "step": 8874 }, { "epoch": 1.55, "learning_rate": 1.1659243198815085e-05, "loss": 0.3692, "step": 8875 }, { "epoch": 1.55, "learning_rate": 1.1654381128911206e-05, "loss": 0.3451, "step": 8876 }, { "epoch": 1.55, "learning_rate": 1.1649520526571483e-05, "loss": 0.3536, "step": 8877 }, { "epoch": 1.55, "learning_rate": 1.1644661392246071e-05, "loss": 0.3549, "step": 8878 }, { "epoch": 1.55, "learning_rate": 1.1639803726385021e-05, "loss": 0.3463, "step": 8879 }, { "epoch": 1.55, "learning_rate": 1.163494752943822e-05, "loss": 0.3518, "step": 8880 }, { "epoch": 1.55, "learning_rate": 1.163009280185543e-05, "loss": 0.3476, "step": 8881 }, { "epoch": 1.55, "learning_rate": 1.1625239544086284e-05, "loss": 0.3613, "step": 8882 }, { "epoch": 1.55, "learning_rate": 1.1620387756580282e-05, "loss": 0.3551, "step": 8883 }, { "epoch": 1.55, "learning_rate": 1.1615537439786767e-05, "loss": 0.3588, "step": 8884 }, { "epoch": 1.55, "learning_rate": 1.1610688594154957e-05, "loss": 0.3459, "step": 8885 }, { "epoch": 1.55, "learning_rate": 1.1605841220133945e-05, "loss": 0.3401, "step": 8886 }, { "epoch": 1.55, "learning_rate": 1.1600995318172664e-05, "loss": 0.3412, "step": 8887 }, { "epoch": 1.55, "learning_rate": 1.159615088871993e-05, "loss": 0.3497, "step": 8888 }, { "epoch": 1.55, "learning_rate": 1.1591307932224427e-05, "loss": 0.3732, "step": 8889 }, { "epoch": 1.55, "learning_rate": 1.1586466449134682e-05, "loss": 0.3519, "step": 8890 }, { "epoch": 1.55, "learning_rate": 1.1581626439899087e-05, "loss": 0.3547, "step": 8891 }, { "epoch": 1.55, "learning_rate": 1.1576787904965927e-05, "loss": 0.3501, "step": 8892 }, { "epoch": 1.55, "learning_rate": 1.1571950844783308e-05, "loss": 0.3656, "step": 8893 }, { "epoch": 1.55, "learning_rate": 1.156711525979924e-05, "loss": 0.3602, "step": 8894 }, { "epoch": 1.55, "learning_rate": 1.1562281150461563e-05, "loss": 0.3295, "step": 8895 }, { "epoch": 1.55, "learning_rate": 1.1557448517218008e-05, "loss": 0.3769, "step": 8896 }, { "epoch": 1.55, "learning_rate": 1.1552617360516145e-05, "loss": 0.3601, "step": 8897 }, { "epoch": 1.55, "learning_rate": 1.1547787680803418e-05, "loss": 0.3501, "step": 8898 }, { "epoch": 1.55, "learning_rate": 1.154295947852714e-05, "loss": 0.3565, "step": 8899 }, { "epoch": 1.55, "learning_rate": 1.1538132754134485e-05, "loss": 0.3526, "step": 8900 }, { "epoch": 1.55, "learning_rate": 1.1533307508072482e-05, "loss": 0.3645, "step": 8901 }, { "epoch": 1.55, "learning_rate": 1.1528483740788026e-05, "loss": 0.3603, "step": 8902 }, { "epoch": 1.55, "learning_rate": 1.1523661452727881e-05, "loss": 0.3574, "step": 8903 }, { "epoch": 1.55, "learning_rate": 1.1518840644338663e-05, "loss": 0.3512, "step": 8904 }, { "epoch": 1.55, "learning_rate": 1.1514021316066864e-05, "loss": 0.3524, "step": 8905 }, { "epoch": 1.55, "learning_rate": 1.1509203468358835e-05, "loss": 0.3536, "step": 8906 }, { "epoch": 1.55, "learning_rate": 1.150438710166079e-05, "loss": 0.3634, "step": 8907 }, { "epoch": 1.55, "learning_rate": 1.1499572216418787e-05, "loss": 0.3547, "step": 8908 }, { "epoch": 1.55, "learning_rate": 1.1494758813078775e-05, "loss": 0.3556, "step": 8909 }, { "epoch": 1.55, "learning_rate": 1.1489946892086563e-05, "loss": 0.3647, "step": 8910 }, { "epoch": 1.55, "learning_rate": 1.1485136453887802e-05, "loss": 0.3486, "step": 8911 }, { "epoch": 1.55, "learning_rate": 1.1480327498928013e-05, "loss": 0.3425, "step": 8912 }, { "epoch": 1.55, "learning_rate": 1.1475520027652599e-05, "loss": 0.3526, "step": 8913 }, { "epoch": 1.55, "learning_rate": 1.1470714040506801e-05, "loss": 0.3617, "step": 8914 }, { "epoch": 1.55, "learning_rate": 1.1465909537935728e-05, "loss": 0.355, "step": 8915 }, { "epoch": 1.55, "learning_rate": 1.146110652038436e-05, "loss": 0.3595, "step": 8916 }, { "epoch": 1.55, "learning_rate": 1.1456304988297546e-05, "loss": 0.3698, "step": 8917 }, { "epoch": 1.55, "learning_rate": 1.1451504942119976e-05, "loss": 0.3549, "step": 8918 }, { "epoch": 1.56, "learning_rate": 1.1446706382296208e-05, "loss": 0.3679, "step": 8919 }, { "epoch": 1.56, "learning_rate": 1.1441909309270671e-05, "loss": 0.3387, "step": 8920 }, { "epoch": 1.56, "learning_rate": 1.1437113723487666e-05, "loss": 0.3334, "step": 8921 }, { "epoch": 1.56, "learning_rate": 1.1432319625391321e-05, "loss": 0.349, "step": 8922 }, { "epoch": 1.56, "learning_rate": 1.142752701542567e-05, "loss": 0.365, "step": 8923 }, { "epoch": 1.56, "learning_rate": 1.1422735894034574e-05, "loss": 0.3627, "step": 8924 }, { "epoch": 1.56, "learning_rate": 1.1417946261661771e-05, "loss": 0.3333, "step": 8925 }, { "epoch": 1.56, "learning_rate": 1.1413158118750847e-05, "loss": 0.3534, "step": 8926 }, { "epoch": 1.56, "learning_rate": 1.140837146574529e-05, "loss": 0.3585, "step": 8927 }, { "epoch": 1.56, "learning_rate": 1.1403586303088411e-05, "loss": 0.3451, "step": 8928 }, { "epoch": 1.56, "learning_rate": 1.1398802631223383e-05, "loss": 0.3584, "step": 8929 }, { "epoch": 1.56, "learning_rate": 1.1394020450593269e-05, "loss": 0.3687, "step": 8930 }, { "epoch": 1.56, "learning_rate": 1.1389239761640962e-05, "loss": 0.3452, "step": 8931 }, { "epoch": 1.56, "learning_rate": 1.1384460564809248e-05, "loss": 0.3659, "step": 8932 }, { "epoch": 1.56, "learning_rate": 1.1379682860540742e-05, "loss": 0.3519, "step": 8933 }, { "epoch": 1.56, "learning_rate": 1.1374906649277956e-05, "loss": 0.3499, "step": 8934 }, { "epoch": 1.56, "learning_rate": 1.1370131931463234e-05, "loss": 0.3622, "step": 8935 }, { "epoch": 1.56, "learning_rate": 1.1365358707538789e-05, "loss": 0.3619, "step": 8936 }, { "epoch": 1.56, "learning_rate": 1.1360586977946708e-05, "loss": 0.3591, "step": 8937 }, { "epoch": 1.56, "learning_rate": 1.1355816743128936e-05, "loss": 0.3677, "step": 8938 }, { "epoch": 1.56, "learning_rate": 1.1351048003527269e-05, "loss": 0.3576, "step": 8939 }, { "epoch": 1.56, "learning_rate": 1.134628075958336e-05, "loss": 0.3514, "step": 8940 }, { "epoch": 1.56, "learning_rate": 1.1341515011738753e-05, "loss": 0.3675, "step": 8941 }, { "epoch": 1.56, "learning_rate": 1.133675076043482e-05, "loss": 0.3334, "step": 8942 }, { "epoch": 1.56, "learning_rate": 1.1331988006112813e-05, "loss": 0.3636, "step": 8943 }, { "epoch": 1.56, "learning_rate": 1.1327226749213849e-05, "loss": 0.362, "step": 8944 }, { "epoch": 1.56, "learning_rate": 1.1322466990178892e-05, "loss": 0.3797, "step": 8945 }, { "epoch": 1.56, "learning_rate": 1.1317708729448767e-05, "loss": 0.3499, "step": 8946 }, { "epoch": 1.56, "learning_rate": 1.131295196746418e-05, "loss": 0.3624, "step": 8947 }, { "epoch": 1.56, "learning_rate": 1.1308196704665673e-05, "loss": 0.3476, "step": 8948 }, { "epoch": 1.56, "learning_rate": 1.1303442941493673e-05, "loss": 0.3511, "step": 8949 }, { "epoch": 1.56, "learning_rate": 1.1298690678388444e-05, "loss": 0.3569, "step": 8950 }, { "epoch": 1.56, "learning_rate": 1.1293939915790134e-05, "loss": 0.3652, "step": 8951 }, { "epoch": 1.56, "learning_rate": 1.1289190654138742e-05, "loss": 0.3578, "step": 8952 }, { "epoch": 1.56, "learning_rate": 1.1284442893874111e-05, "loss": 0.3649, "step": 8953 }, { "epoch": 1.56, "learning_rate": 1.1279696635435978e-05, "loss": 0.3567, "step": 8954 }, { "epoch": 1.56, "learning_rate": 1.1274951879263924e-05, "loss": 0.3609, "step": 8955 }, { "epoch": 1.56, "learning_rate": 1.1270208625797387e-05, "loss": 0.3342, "step": 8956 }, { "epoch": 1.56, "learning_rate": 1.1265466875475664e-05, "loss": 0.3522, "step": 8957 }, { "epoch": 1.56, "learning_rate": 1.1260726628737932e-05, "loss": 0.3463, "step": 8958 }, { "epoch": 1.56, "learning_rate": 1.1255987886023202e-05, "loss": 0.3532, "step": 8959 }, { "epoch": 1.56, "learning_rate": 1.1251250647770364e-05, "loss": 0.3449, "step": 8960 }, { "epoch": 1.56, "learning_rate": 1.1246514914418174e-05, "loss": 0.3571, "step": 8961 }, { "epoch": 1.56, "learning_rate": 1.124178068640523e-05, "loss": 0.3622, "step": 8962 }, { "epoch": 1.56, "learning_rate": 1.1237047964169992e-05, "loss": 0.3478, "step": 8963 }, { "epoch": 1.56, "learning_rate": 1.1232316748150797e-05, "loss": 0.3604, "step": 8964 }, { "epoch": 1.56, "learning_rate": 1.1227587038785835e-05, "loss": 0.353, "step": 8965 }, { "epoch": 1.56, "learning_rate": 1.1222858836513154e-05, "loss": 0.3555, "step": 8966 }, { "epoch": 1.56, "learning_rate": 1.1218132141770652e-05, "loss": 0.3525, "step": 8967 }, { "epoch": 1.56, "learning_rate": 1.1213406954996112e-05, "loss": 0.3636, "step": 8968 }, { "epoch": 1.56, "learning_rate": 1.1208683276627159e-05, "loss": 0.3568, "step": 8969 }, { "epoch": 1.56, "learning_rate": 1.1203961107101275e-05, "loss": 0.3552, "step": 8970 }, { "epoch": 1.56, "learning_rate": 1.119924044685582e-05, "loss": 0.3515, "step": 8971 }, { "epoch": 1.56, "learning_rate": 1.1194521296328009e-05, "loss": 0.3504, "step": 8972 }, { "epoch": 1.56, "learning_rate": 1.1189803655954904e-05, "loss": 0.3451, "step": 8973 }, { "epoch": 1.56, "learning_rate": 1.118508752617343e-05, "loss": 0.3554, "step": 8974 }, { "epoch": 1.56, "learning_rate": 1.1180372907420388e-05, "loss": 0.3483, "step": 8975 }, { "epoch": 1.56, "learning_rate": 1.1175659800132431e-05, "loss": 0.3476, "step": 8976 }, { "epoch": 1.57, "learning_rate": 1.1170948204746067e-05, "loss": 0.3546, "step": 8977 }, { "epoch": 1.57, "learning_rate": 1.116623812169766e-05, "loss": 0.3533, "step": 8978 }, { "epoch": 1.57, "learning_rate": 1.1161529551423452e-05, "loss": 0.354, "step": 8979 }, { "epoch": 1.57, "learning_rate": 1.1156822494359531e-05, "loss": 0.3589, "step": 8980 }, { "epoch": 1.57, "learning_rate": 1.1152116950941826e-05, "loss": 0.353, "step": 8981 }, { "epoch": 1.57, "learning_rate": 1.1147412921606186e-05, "loss": 0.365, "step": 8982 }, { "epoch": 1.57, "learning_rate": 1.1142710406788259e-05, "loss": 0.3528, "step": 8983 }, { "epoch": 1.57, "learning_rate": 1.1138009406923571e-05, "loss": 0.3466, "step": 8984 }, { "epoch": 1.57, "learning_rate": 1.1133309922447526e-05, "loss": 0.3601, "step": 8985 }, { "epoch": 1.57, "learning_rate": 1.1128611953795362e-05, "loss": 0.3515, "step": 8986 }, { "epoch": 1.57, "learning_rate": 1.1123915501402196e-05, "loss": 0.3578, "step": 8987 }, { "epoch": 1.57, "learning_rate": 1.1119220565702983e-05, "loss": 0.3615, "step": 8988 }, { "epoch": 1.57, "learning_rate": 1.1114527147132573e-05, "loss": 0.3663, "step": 8989 }, { "epoch": 1.57, "learning_rate": 1.1109835246125635e-05, "loss": 0.3526, "step": 8990 }, { "epoch": 1.57, "learning_rate": 1.1105144863116717e-05, "loss": 0.3601, "step": 8991 }, { "epoch": 1.57, "learning_rate": 1.1100455998540233e-05, "loss": 0.3489, "step": 8992 }, { "epoch": 1.57, "learning_rate": 1.109576865283045e-05, "loss": 0.3627, "step": 8993 }, { "epoch": 1.57, "learning_rate": 1.1091082826421487e-05, "loss": 0.3529, "step": 8994 }, { "epoch": 1.57, "learning_rate": 1.1086398519747328e-05, "loss": 0.3609, "step": 8995 }, { "epoch": 1.57, "learning_rate": 1.1081715733241823e-05, "loss": 0.3474, "step": 8996 }, { "epoch": 1.57, "learning_rate": 1.1077034467338674e-05, "loss": 0.3782, "step": 8997 }, { "epoch": 1.57, "learning_rate": 1.1072354722471423e-05, "loss": 0.3334, "step": 8998 }, { "epoch": 1.57, "learning_rate": 1.1067676499073524e-05, "loss": 0.3634, "step": 8999 }, { "epoch": 1.57, "learning_rate": 1.1062999797578239e-05, "loss": 0.3454, "step": 9000 }, { "epoch": 1.57, "learning_rate": 1.1058324618418703e-05, "loss": 0.3514, "step": 9001 }, { "epoch": 1.57, "learning_rate": 1.1053650962027926e-05, "loss": 0.3597, "step": 9002 }, { "epoch": 1.57, "learning_rate": 1.1048978828838758e-05, "loss": 0.3417, "step": 9003 }, { "epoch": 1.57, "learning_rate": 1.104430821928392e-05, "loss": 0.3595, "step": 9004 }, { "epoch": 1.57, "learning_rate": 1.103963913379598e-05, "loss": 0.3723, "step": 9005 }, { "epoch": 1.57, "learning_rate": 1.1034971572807377e-05, "loss": 0.3512, "step": 9006 }, { "epoch": 1.57, "learning_rate": 1.1030305536750407e-05, "loss": 0.3544, "step": 9007 }, { "epoch": 1.57, "learning_rate": 1.1025641026057209e-05, "loss": 0.3547, "step": 9008 }, { "epoch": 1.57, "learning_rate": 1.1020978041159802e-05, "loss": 0.3647, "step": 9009 }, { "epoch": 1.57, "learning_rate": 1.101631658249006e-05, "loss": 0.3494, "step": 9010 }, { "epoch": 1.57, "learning_rate": 1.1011656650479705e-05, "loss": 0.3652, "step": 9011 }, { "epoch": 1.57, "learning_rate": 1.1006998245560317e-05, "loss": 0.348, "step": 9012 }, { "epoch": 1.57, "learning_rate": 1.1002341368163352e-05, "loss": 0.3409, "step": 9013 }, { "epoch": 1.57, "learning_rate": 1.0997686018720104e-05, "loss": 0.3537, "step": 9014 }, { "epoch": 1.57, "learning_rate": 1.099303219766174e-05, "loss": 0.3514, "step": 9015 }, { "epoch": 1.57, "learning_rate": 1.0988379905419278e-05, "loss": 0.3519, "step": 9016 }, { "epoch": 1.57, "learning_rate": 1.0983729142423597e-05, "loss": 0.3489, "step": 9017 }, { "epoch": 1.57, "learning_rate": 1.0979079909105436e-05, "loss": 0.3768, "step": 9018 }, { "epoch": 1.57, "learning_rate": 1.0974432205895385e-05, "loss": 0.3687, "step": 9019 }, { "epoch": 1.57, "learning_rate": 1.0969786033223898e-05, "loss": 0.3578, "step": 9020 }, { "epoch": 1.57, "learning_rate": 1.0965141391521296e-05, "loss": 0.3386, "step": 9021 }, { "epoch": 1.57, "learning_rate": 1.0960498281217738e-05, "loss": 0.3494, "step": 9022 }, { "epoch": 1.57, "learning_rate": 1.095585670274326e-05, "loss": 0.3524, "step": 9023 }, { "epoch": 1.57, "learning_rate": 1.0951216656527746e-05, "loss": 0.3299, "step": 9024 }, { "epoch": 1.57, "learning_rate": 1.0946578143000933e-05, "loss": 0.3541, "step": 9025 }, { "epoch": 1.57, "learning_rate": 1.0941941162592431e-05, "loss": 0.3516, "step": 9026 }, { "epoch": 1.57, "learning_rate": 1.0937305715731705e-05, "loss": 0.3502, "step": 9027 }, { "epoch": 1.57, "learning_rate": 1.0932671802848065e-05, "loss": 0.3614, "step": 9028 }, { "epoch": 1.57, "learning_rate": 1.0928039424370685e-05, "loss": 0.3593, "step": 9029 }, { "epoch": 1.57, "learning_rate": 1.0923408580728612e-05, "loss": 0.3616, "step": 9030 }, { "epoch": 1.57, "learning_rate": 1.0918779272350724e-05, "loss": 0.3583, "step": 9031 }, { "epoch": 1.57, "learning_rate": 1.0914151499665781e-05, "loss": 0.3569, "step": 9032 }, { "epoch": 1.57, "learning_rate": 1.0909525263102382e-05, "loss": 0.3582, "step": 9033 }, { "epoch": 1.58, "learning_rate": 1.0904900563089002e-05, "loss": 0.3618, "step": 9034 }, { "epoch": 1.58, "learning_rate": 1.090027740005396e-05, "loss": 0.3544, "step": 9035 }, { "epoch": 1.58, "learning_rate": 1.089565577442542e-05, "loss": 0.3492, "step": 9036 }, { "epoch": 1.58, "learning_rate": 1.0891035686631452e-05, "loss": 0.3447, "step": 9037 }, { "epoch": 1.58, "learning_rate": 1.0886417137099936e-05, "loss": 0.3423, "step": 9038 }, { "epoch": 1.58, "learning_rate": 1.0881800126258617e-05, "loss": 0.3736, "step": 9039 }, { "epoch": 1.58, "learning_rate": 1.0877184654535124e-05, "loss": 0.3588, "step": 9040 }, { "epoch": 1.58, "learning_rate": 1.0872570722356915e-05, "loss": 0.3523, "step": 9041 }, { "epoch": 1.58, "learning_rate": 1.0867958330151307e-05, "loss": 0.3656, "step": 9042 }, { "epoch": 1.58, "learning_rate": 1.0863347478345496e-05, "loss": 0.3485, "step": 9043 }, { "epoch": 1.58, "learning_rate": 1.0858738167366523e-05, "loss": 0.353, "step": 9044 }, { "epoch": 1.58, "learning_rate": 1.0854130397641284e-05, "loss": 0.3552, "step": 9045 }, { "epoch": 1.58, "learning_rate": 1.0849524169596525e-05, "loss": 0.3573, "step": 9046 }, { "epoch": 1.58, "learning_rate": 1.0844919483658864e-05, "loss": 0.3559, "step": 9047 }, { "epoch": 1.58, "learning_rate": 1.084031634025478e-05, "loss": 0.3396, "step": 9048 }, { "epoch": 1.58, "learning_rate": 1.0835714739810587e-05, "loss": 0.3651, "step": 9049 }, { "epoch": 1.58, "learning_rate": 1.083111468275247e-05, "loss": 0.3643, "step": 9050 }, { "epoch": 1.58, "learning_rate": 1.082651616950648e-05, "loss": 0.3477, "step": 9051 }, { "epoch": 1.58, "learning_rate": 1.0821919200498503e-05, "loss": 0.3679, "step": 9052 }, { "epoch": 1.58, "learning_rate": 1.0817323776154288e-05, "loss": 0.3646, "step": 9053 }, { "epoch": 1.58, "learning_rate": 1.0812729896899468e-05, "loss": 0.3498, "step": 9054 }, { "epoch": 1.58, "learning_rate": 1.0808137563159501e-05, "loss": 0.3561, "step": 9055 }, { "epoch": 1.58, "learning_rate": 1.0803546775359711e-05, "loss": 0.3534, "step": 9056 }, { "epoch": 1.58, "learning_rate": 1.0798957533925275e-05, "loss": 0.3543, "step": 9057 }, { "epoch": 1.58, "learning_rate": 1.0794369839281239e-05, "loss": 0.3556, "step": 9058 }, { "epoch": 1.58, "learning_rate": 1.0789783691852501e-05, "loss": 0.3519, "step": 9059 }, { "epoch": 1.58, "learning_rate": 1.0785199092063808e-05, "loss": 0.3592, "step": 9060 }, { "epoch": 1.58, "learning_rate": 1.0780616040339772e-05, "loss": 0.3407, "step": 9061 }, { "epoch": 1.58, "learning_rate": 1.0776034537104862e-05, "loss": 0.3577, "step": 9062 }, { "epoch": 1.58, "learning_rate": 1.0771454582783388e-05, "loss": 0.3543, "step": 9063 }, { "epoch": 1.58, "learning_rate": 1.076687617779954e-05, "loss": 0.3574, "step": 9064 }, { "epoch": 1.58, "learning_rate": 1.0762299322577352e-05, "loss": 0.348, "step": 9065 }, { "epoch": 1.58, "learning_rate": 1.0757724017540717e-05, "loss": 0.3504, "step": 9066 }, { "epoch": 1.58, "learning_rate": 1.0753150263113376e-05, "loss": 0.3783, "step": 9067 }, { "epoch": 1.58, "learning_rate": 1.0748578059718942e-05, "loss": 0.3486, "step": 9068 }, { "epoch": 1.58, "learning_rate": 1.0744007407780868e-05, "loss": 0.3637, "step": 9069 }, { "epoch": 1.58, "learning_rate": 1.073943830772248e-05, "loss": 0.354, "step": 9070 }, { "epoch": 1.58, "learning_rate": 1.0734870759966941e-05, "loss": 0.3335, "step": 9071 }, { "epoch": 1.58, "learning_rate": 1.0730304764937293e-05, "loss": 0.3571, "step": 9072 }, { "epoch": 1.58, "learning_rate": 1.0725740323056419e-05, "loss": 0.3516, "step": 9073 }, { "epoch": 1.58, "learning_rate": 1.0721177434747048e-05, "loss": 0.3573, "step": 9074 }, { "epoch": 1.58, "learning_rate": 1.0716616100431789e-05, "loss": 0.3443, "step": 9075 }, { "epoch": 1.58, "learning_rate": 1.0712056320533103e-05, "loss": 0.351, "step": 9076 }, { "epoch": 1.58, "learning_rate": 1.0707498095473288e-05, "loss": 0.3514, "step": 9077 }, { "epoch": 1.58, "learning_rate": 1.070294142567452e-05, "loss": 0.3493, "step": 9078 }, { "epoch": 1.58, "learning_rate": 1.0698386311558819e-05, "loss": 0.356, "step": 9079 }, { "epoch": 1.58, "learning_rate": 1.0693832753548052e-05, "loss": 0.366, "step": 9080 }, { "epoch": 1.58, "learning_rate": 1.0689280752063965e-05, "loss": 0.3484, "step": 9081 }, { "epoch": 1.58, "learning_rate": 1.068473030752815e-05, "loss": 0.3513, "step": 9082 }, { "epoch": 1.58, "learning_rate": 1.068018142036205e-05, "loss": 0.3327, "step": 9083 }, { "epoch": 1.58, "learning_rate": 1.0675634090986956e-05, "loss": 0.3452, "step": 9084 }, { "epoch": 1.58, "learning_rate": 1.0671088319824041e-05, "loss": 0.3671, "step": 9085 }, { "epoch": 1.58, "learning_rate": 1.0666544107294308e-05, "loss": 0.369, "step": 9086 }, { "epoch": 1.58, "learning_rate": 1.0662001453818632e-05, "loss": 0.3588, "step": 9087 }, { "epoch": 1.58, "learning_rate": 1.0657460359817726e-05, "loss": 0.3313, "step": 9088 }, { "epoch": 1.58, "learning_rate": 1.0652920825712185e-05, "loss": 0.3389, "step": 9089 }, { "epoch": 1.58, "learning_rate": 1.0648382851922437e-05, "loss": 0.3323, "step": 9090 }, { "epoch": 1.58, "learning_rate": 1.0643846438868757e-05, "loss": 0.3431, "step": 9091 }, { "epoch": 1.59, "learning_rate": 1.0639311586971322e-05, "loss": 0.3444, "step": 9092 }, { "epoch": 1.59, "learning_rate": 1.063477829665012e-05, "loss": 0.352, "step": 9093 }, { "epoch": 1.59, "learning_rate": 1.0630246568324995e-05, "loss": 0.3552, "step": 9094 }, { "epoch": 1.59, "learning_rate": 1.0625716402415683e-05, "loss": 0.3472, "step": 9095 }, { "epoch": 1.59, "learning_rate": 1.0621187799341737e-05, "loss": 0.365, "step": 9096 }, { "epoch": 1.59, "learning_rate": 1.0616660759522575e-05, "loss": 0.3644, "step": 9097 }, { "epoch": 1.59, "learning_rate": 1.0612135283377483e-05, "loss": 0.3612, "step": 9098 }, { "epoch": 1.59, "learning_rate": 1.0607611371325602e-05, "loss": 0.3631, "step": 9099 }, { "epoch": 1.59, "learning_rate": 1.0603089023785908e-05, "loss": 0.3435, "step": 9100 }, { "epoch": 1.59, "learning_rate": 1.0598568241177246e-05, "loss": 0.3637, "step": 9101 }, { "epoch": 1.59, "learning_rate": 1.0594049023918326e-05, "loss": 0.3411, "step": 9102 }, { "epoch": 1.59, "learning_rate": 1.0589531372427682e-05, "loss": 0.3596, "step": 9103 }, { "epoch": 1.59, "learning_rate": 1.058501528712374e-05, "loss": 0.3543, "step": 9104 }, { "epoch": 1.59, "learning_rate": 1.0580500768424752e-05, "loss": 0.3614, "step": 9105 }, { "epoch": 1.59, "learning_rate": 1.057598781674885e-05, "loss": 0.3465, "step": 9106 }, { "epoch": 1.59, "learning_rate": 1.0571476432513994e-05, "loss": 0.3597, "step": 9107 }, { "epoch": 1.59, "learning_rate": 1.0566966616138016e-05, "loss": 0.3475, "step": 9108 }, { "epoch": 1.59, "learning_rate": 1.0562458368038597e-05, "loss": 0.3241, "step": 9109 }, { "epoch": 1.59, "learning_rate": 1.0557951688633284e-05, "loss": 0.3567, "step": 9110 }, { "epoch": 1.59, "learning_rate": 1.0553446578339465e-05, "loss": 0.3557, "step": 9111 }, { "epoch": 1.59, "learning_rate": 1.0548943037574379e-05, "loss": 0.3421, "step": 9112 }, { "epoch": 1.59, "learning_rate": 1.054444106675514e-05, "loss": 0.3535, "step": 9113 }, { "epoch": 1.59, "learning_rate": 1.053994066629869e-05, "loss": 0.3618, "step": 9114 }, { "epoch": 1.59, "learning_rate": 1.0535441836621848e-05, "loss": 0.3659, "step": 9115 }, { "epoch": 1.59, "learning_rate": 1.0530944578141288e-05, "loss": 0.3436, "step": 9116 }, { "epoch": 1.59, "learning_rate": 1.0526448891273518e-05, "loss": 0.3509, "step": 9117 }, { "epoch": 1.59, "learning_rate": 1.0521954776434912e-05, "loss": 0.3582, "step": 9118 }, { "epoch": 1.59, "learning_rate": 1.0517462234041703e-05, "loss": 0.3458, "step": 9119 }, { "epoch": 1.59, "learning_rate": 1.0512971264509983e-05, "loss": 0.3586, "step": 9120 }, { "epoch": 1.59, "learning_rate": 1.0508481868255674e-05, "loss": 0.3526, "step": 9121 }, { "epoch": 1.59, "learning_rate": 1.0503994045694568e-05, "loss": 0.3492, "step": 9122 }, { "epoch": 1.59, "learning_rate": 1.0499507797242327e-05, "loss": 0.3455, "step": 9123 }, { "epoch": 1.59, "learning_rate": 1.0495023123314441e-05, "loss": 0.3506, "step": 9124 }, { "epoch": 1.59, "learning_rate": 1.0490540024326253e-05, "loss": 0.3505, "step": 9125 }, { "epoch": 1.59, "learning_rate": 1.0486058500692984e-05, "loss": 0.3777, "step": 9126 }, { "epoch": 1.59, "learning_rate": 1.0481578552829705e-05, "loss": 0.3681, "step": 9127 }, { "epoch": 1.59, "learning_rate": 1.047710018115132e-05, "loss": 0.3616, "step": 9128 }, { "epoch": 1.59, "learning_rate": 1.0472623386072598e-05, "loss": 0.3519, "step": 9129 }, { "epoch": 1.59, "learning_rate": 1.0468148168008167e-05, "loss": 0.3535, "step": 9130 }, { "epoch": 1.59, "learning_rate": 1.0463674527372513e-05, "loss": 0.3531, "step": 9131 }, { "epoch": 1.59, "learning_rate": 1.0459202464579957e-05, "loss": 0.3482, "step": 9132 }, { "epoch": 1.59, "learning_rate": 1.0454731980044694e-05, "loss": 0.359, "step": 9133 }, { "epoch": 1.59, "learning_rate": 1.0450263074180763e-05, "loss": 0.3668, "step": 9134 }, { "epoch": 1.59, "learning_rate": 1.044579574740205e-05, "loss": 0.3611, "step": 9135 }, { "epoch": 1.59, "learning_rate": 1.0441330000122306e-05, "loss": 0.3578, "step": 9136 }, { "epoch": 1.59, "learning_rate": 1.0436865832755144e-05, "loss": 0.3503, "step": 9137 }, { "epoch": 1.59, "learning_rate": 1.043240324571401e-05, "loss": 0.3678, "step": 9138 }, { "epoch": 1.59, "learning_rate": 1.0427942239412206e-05, "loss": 0.3703, "step": 9139 }, { "epoch": 1.59, "learning_rate": 1.0423482814262905e-05, "loss": 0.3484, "step": 9140 }, { "epoch": 1.59, "learning_rate": 1.0419024970679114e-05, "loss": 0.3627, "step": 9141 }, { "epoch": 1.59, "learning_rate": 1.0414568709073716e-05, "loss": 0.3544, "step": 9142 }, { "epoch": 1.59, "learning_rate": 1.0410114029859416e-05, "loss": 0.3553, "step": 9143 }, { "epoch": 1.59, "learning_rate": 1.0405660933448809e-05, "loss": 0.3424, "step": 9144 }, { "epoch": 1.59, "learning_rate": 1.0401209420254312e-05, "loss": 0.3488, "step": 9145 }, { "epoch": 1.59, "learning_rate": 1.0396759490688207e-05, "loss": 0.3494, "step": 9146 }, { "epoch": 1.59, "learning_rate": 1.0392311145162637e-05, "loss": 0.3414, "step": 9147 }, { "epoch": 1.59, "learning_rate": 1.0387864384089591e-05, "loss": 0.353, "step": 9148 }, { "epoch": 1.6, "learning_rate": 1.0383419207880913e-05, "loss": 0.372, "step": 9149 }, { "epoch": 1.6, "learning_rate": 1.037897561694829e-05, "loss": 0.3493, "step": 9150 }, { "epoch": 1.6, "learning_rate": 1.0374533611703287e-05, "loss": 0.3537, "step": 9151 }, { "epoch": 1.6, "learning_rate": 1.0370093192557287e-05, "loss": 0.3683, "step": 9152 }, { "epoch": 1.6, "learning_rate": 1.0365654359921562e-05, "loss": 0.3465, "step": 9153 }, { "epoch": 1.6, "learning_rate": 1.0361217114207217e-05, "loss": 0.3694, "step": 9154 }, { "epoch": 1.6, "learning_rate": 1.0356781455825216e-05, "loss": 0.3558, "step": 9155 }, { "epoch": 1.6, "learning_rate": 1.0352347385186357e-05, "loss": 0.3601, "step": 9156 }, { "epoch": 1.6, "learning_rate": 1.0347914902701331e-05, "loss": 0.3714, "step": 9157 }, { "epoch": 1.6, "learning_rate": 1.0343484008780641e-05, "loss": 0.3548, "step": 9158 }, { "epoch": 1.6, "learning_rate": 1.0339054703834672e-05, "loss": 0.3366, "step": 9159 }, { "epoch": 1.6, "learning_rate": 1.033462698827364e-05, "loss": 0.3494, "step": 9160 }, { "epoch": 1.6, "learning_rate": 1.033020086250764e-05, "loss": 0.3547, "step": 9161 }, { "epoch": 1.6, "learning_rate": 1.0325776326946587e-05, "loss": 0.3452, "step": 9162 }, { "epoch": 1.6, "learning_rate": 1.0321353382000267e-05, "loss": 0.3617, "step": 9163 }, { "epoch": 1.6, "learning_rate": 1.0316932028078323e-05, "loss": 0.3449, "step": 9164 }, { "epoch": 1.6, "learning_rate": 1.031251226559025e-05, "loss": 0.3702, "step": 9165 }, { "epoch": 1.6, "learning_rate": 1.0308094094945386e-05, "loss": 0.3799, "step": 9166 }, { "epoch": 1.6, "learning_rate": 1.0303677516552917e-05, "loss": 0.3466, "step": 9167 }, { "epoch": 1.6, "learning_rate": 1.0299262530821905e-05, "loss": 0.3531, "step": 9168 }, { "epoch": 1.6, "learning_rate": 1.0294849138161236e-05, "loss": 0.3514, "step": 9169 }, { "epoch": 1.6, "learning_rate": 1.029043733897967e-05, "loss": 0.3365, "step": 9170 }, { "epoch": 1.6, "learning_rate": 1.0286027133685816e-05, "loss": 0.3607, "step": 9171 }, { "epoch": 1.6, "learning_rate": 1.028161852268813e-05, "loss": 0.3718, "step": 9172 }, { "epoch": 1.6, "learning_rate": 1.0277211506394913e-05, "loss": 0.341, "step": 9173 }, { "epoch": 1.6, "learning_rate": 1.0272806085214329e-05, "loss": 0.3314, "step": 9174 }, { "epoch": 1.6, "learning_rate": 1.0268402259554406e-05, "loss": 0.3395, "step": 9175 }, { "epoch": 1.6, "learning_rate": 1.0264000029822999e-05, "loss": 0.3508, "step": 9176 }, { "epoch": 1.6, "learning_rate": 1.0259599396427821e-05, "loss": 0.3424, "step": 9177 }, { "epoch": 1.6, "learning_rate": 1.0255200359776455e-05, "loss": 0.3509, "step": 9178 }, { "epoch": 1.6, "learning_rate": 1.025080292027632e-05, "loss": 0.3453, "step": 9179 }, { "epoch": 1.6, "learning_rate": 1.0246407078334686e-05, "loss": 0.3461, "step": 9180 }, { "epoch": 1.6, "learning_rate": 1.024201283435868e-05, "loss": 0.3674, "step": 9181 }, { "epoch": 1.6, "learning_rate": 1.0237620188755292e-05, "loss": 0.3531, "step": 9182 }, { "epoch": 1.6, "learning_rate": 1.0233229141931348e-05, "loss": 0.3442, "step": 9183 }, { "epoch": 1.6, "learning_rate": 1.0228839694293521e-05, "loss": 0.3693, "step": 9184 }, { "epoch": 1.6, "learning_rate": 1.0224451846248354e-05, "loss": 0.353, "step": 9185 }, { "epoch": 1.6, "learning_rate": 1.0220065598202238e-05, "loss": 0.3566, "step": 9186 }, { "epoch": 1.6, "learning_rate": 1.0215680950561405e-05, "loss": 0.3413, "step": 9187 }, { "epoch": 1.6, "learning_rate": 1.0211297903731947e-05, "loss": 0.3411, "step": 9188 }, { "epoch": 1.6, "learning_rate": 1.0206916458119808e-05, "loss": 0.3617, "step": 9189 }, { "epoch": 1.6, "learning_rate": 1.0202536614130784e-05, "loss": 0.3682, "step": 9190 }, { "epoch": 1.6, "learning_rate": 1.0198158372170499e-05, "loss": 0.3645, "step": 9191 }, { "epoch": 1.6, "learning_rate": 1.0193781732644484e-05, "loss": 0.3588, "step": 9192 }, { "epoch": 1.6, "learning_rate": 1.0189406695958068e-05, "loss": 0.3507, "step": 9193 }, { "epoch": 1.6, "learning_rate": 1.0185033262516447e-05, "loss": 0.3602, "step": 9194 }, { "epoch": 1.6, "learning_rate": 1.0180661432724686e-05, "loss": 0.3651, "step": 9195 }, { "epoch": 1.6, "learning_rate": 1.017629120698768e-05, "loss": 0.3619, "step": 9196 }, { "epoch": 1.6, "learning_rate": 1.0171922585710185e-05, "loss": 0.3721, "step": 9197 }, { "epoch": 1.6, "learning_rate": 1.0167555569296805e-05, "loss": 0.3386, "step": 9198 }, { "epoch": 1.6, "learning_rate": 1.0163190158152004e-05, "loss": 0.3507, "step": 9199 }, { "epoch": 1.6, "learning_rate": 1.0158826352680087e-05, "loss": 0.3677, "step": 9200 }, { "epoch": 1.6, "learning_rate": 1.0154464153285207e-05, "loss": 0.3529, "step": 9201 }, { "epoch": 1.6, "learning_rate": 1.0150103560371383e-05, "loss": 0.3446, "step": 9202 }, { "epoch": 1.6, "learning_rate": 1.0145744574342478e-05, "loss": 0.3521, "step": 9203 }, { "epoch": 1.6, "learning_rate": 1.0141387195602207e-05, "loss": 0.3465, "step": 9204 }, { "epoch": 1.6, "learning_rate": 1.0137031424554122e-05, "loss": 0.3589, "step": 9205 }, { "epoch": 1.61, "learning_rate": 1.0132677261601656e-05, "loss": 0.3706, "step": 9206 }, { "epoch": 1.61, "learning_rate": 1.012832470714807e-05, "loss": 0.363, "step": 9207 }, { "epoch": 1.61, "learning_rate": 1.0123973761596464e-05, "loss": 0.3521, "step": 9208 }, { "epoch": 1.61, "learning_rate": 1.0119624425349835e-05, "loss": 0.3449, "step": 9209 }, { "epoch": 1.61, "learning_rate": 1.0115276698810991e-05, "loss": 0.3442, "step": 9210 }, { "epoch": 1.61, "learning_rate": 1.0110930582382597e-05, "loss": 0.3508, "step": 9211 }, { "epoch": 1.61, "learning_rate": 1.0106586076467186e-05, "loss": 0.3538, "step": 9212 }, { "epoch": 1.61, "learning_rate": 1.0102243181467118e-05, "loss": 0.3332, "step": 9213 }, { "epoch": 1.61, "learning_rate": 1.0097901897784627e-05, "loss": 0.3518, "step": 9214 }, { "epoch": 1.61, "learning_rate": 1.009356222582178e-05, "loss": 0.3626, "step": 9215 }, { "epoch": 1.61, "learning_rate": 1.0089224165980508e-05, "loss": 0.3596, "step": 9216 }, { "epoch": 1.61, "learning_rate": 1.0084887718662583e-05, "loss": 0.3846, "step": 9217 }, { "epoch": 1.61, "learning_rate": 1.0080552884269627e-05, "loss": 0.348, "step": 9218 }, { "epoch": 1.61, "learning_rate": 1.0076219663203119e-05, "loss": 0.3497, "step": 9219 }, { "epoch": 1.61, "learning_rate": 1.0071888055864396e-05, "loss": 0.3451, "step": 9220 }, { "epoch": 1.61, "learning_rate": 1.0067558062654627e-05, "loss": 0.3582, "step": 9221 }, { "epoch": 1.61, "learning_rate": 1.0063229683974836e-05, "loss": 0.3444, "step": 9222 }, { "epoch": 1.61, "learning_rate": 1.0058902920225913e-05, "loss": 0.3411, "step": 9223 }, { "epoch": 1.61, "learning_rate": 1.0054577771808574e-05, "loss": 0.3701, "step": 9224 }, { "epoch": 1.61, "learning_rate": 1.0050254239123408e-05, "loss": 0.3649, "step": 9225 }, { "epoch": 1.61, "learning_rate": 1.0045932322570849e-05, "loss": 0.3619, "step": 9226 }, { "epoch": 1.61, "learning_rate": 1.0041612022551171e-05, "loss": 0.3496, "step": 9227 }, { "epoch": 1.61, "learning_rate": 1.0037293339464502e-05, "loss": 0.3427, "step": 9228 }, { "epoch": 1.61, "learning_rate": 1.003297627371083e-05, "loss": 0.3466, "step": 9229 }, { "epoch": 1.61, "learning_rate": 1.0028660825689975e-05, "loss": 0.3599, "step": 9230 }, { "epoch": 1.61, "learning_rate": 1.0024346995801632e-05, "loss": 0.3508, "step": 9231 }, { "epoch": 1.61, "learning_rate": 1.002003478444532e-05, "loss": 0.3503, "step": 9232 }, { "epoch": 1.61, "learning_rate": 1.0015724192020435e-05, "loss": 0.3543, "step": 9233 }, { "epoch": 1.61, "learning_rate": 1.0011415218926197e-05, "loss": 0.3571, "step": 9234 }, { "epoch": 1.61, "learning_rate": 1.0007107865561687e-05, "loss": 0.3688, "step": 9235 }, { "epoch": 1.61, "learning_rate": 1.0002802132325836e-05, "loss": 0.3545, "step": 9236 }, { "epoch": 1.61, "learning_rate": 9.99849801961744e-06, "loss": 0.3659, "step": 9237 }, { "epoch": 1.61, "learning_rate": 9.994195527835116e-06, "loss": 0.3447, "step": 9238 }, { "epoch": 1.61, "learning_rate": 9.989894657377345e-06, "loss": 0.3697, "step": 9239 }, { "epoch": 1.61, "learning_rate": 9.985595408642468e-06, "loss": 0.3513, "step": 9240 }, { "epoch": 1.61, "learning_rate": 9.981297782028653e-06, "loss": 0.3581, "step": 9241 }, { "epoch": 1.61, "learning_rate": 9.977001777933948e-06, "loss": 0.357, "step": 9242 }, { "epoch": 1.61, "learning_rate": 9.972707396756215e-06, "loss": 0.3433, "step": 9243 }, { "epoch": 1.61, "learning_rate": 9.968414638893195e-06, "loss": 0.3606, "step": 9244 }, { "epoch": 1.61, "learning_rate": 9.96412350474247e-06, "loss": 0.363, "step": 9245 }, { "epoch": 1.61, "learning_rate": 9.95983399470145e-06, "loss": 0.3412, "step": 9246 }, { "epoch": 1.61, "learning_rate": 9.955546109167443e-06, "loss": 0.3547, "step": 9247 }, { "epoch": 1.61, "learning_rate": 9.951259848537561e-06, "loss": 0.3671, "step": 9248 }, { "epoch": 1.61, "learning_rate": 9.94697521320878e-06, "loss": 0.364, "step": 9249 }, { "epoch": 1.61, "learning_rate": 9.942692203577937e-06, "loss": 0.3451, "step": 9250 }, { "epoch": 1.61, "learning_rate": 9.938410820041704e-06, "loss": 0.3727, "step": 9251 }, { "epoch": 1.61, "learning_rate": 9.934131062996602e-06, "loss": 0.3593, "step": 9252 }, { "epoch": 1.61, "learning_rate": 9.929852932839006e-06, "loss": 0.362, "step": 9253 }, { "epoch": 1.61, "learning_rate": 9.925576429965159e-06, "loss": 0.343, "step": 9254 }, { "epoch": 1.61, "learning_rate": 9.921301554771119e-06, "loss": 0.3481, "step": 9255 }, { "epoch": 1.61, "learning_rate": 9.91702830765281e-06, "loss": 0.356, "step": 9256 }, { "epoch": 1.61, "learning_rate": 9.912756689006009e-06, "loss": 0.3628, "step": 9257 }, { "epoch": 1.61, "learning_rate": 9.908486699226338e-06, "loss": 0.3658, "step": 9258 }, { "epoch": 1.61, "learning_rate": 9.90421833870927e-06, "loss": 0.3573, "step": 9259 }, { "epoch": 1.61, "learning_rate": 9.899951607850117e-06, "loss": 0.3633, "step": 9260 }, { "epoch": 1.61, "learning_rate": 9.89568650704406e-06, "loss": 0.3579, "step": 9261 }, { "epoch": 1.61, "learning_rate": 9.891423036686113e-06, "loss": 0.364, "step": 9262 }, { "epoch": 1.61, "learning_rate": 9.887161197171124e-06, "loss": 0.3376, "step": 9263 }, { "epoch": 1.62, "learning_rate": 9.882900988893842e-06, "loss": 0.3441, "step": 9264 }, { "epoch": 1.62, "learning_rate": 9.878642412248815e-06, "loss": 0.3526, "step": 9265 }, { "epoch": 1.62, "learning_rate": 9.874385467630451e-06, "loss": 0.367, "step": 9266 }, { "epoch": 1.62, "learning_rate": 9.87013015543303e-06, "loss": 0.3258, "step": 9267 }, { "epoch": 1.62, "learning_rate": 9.865876476050645e-06, "loss": 0.3503, "step": 9268 }, { "epoch": 1.62, "learning_rate": 9.861624429877276e-06, "loss": 0.3503, "step": 9269 }, { "epoch": 1.62, "learning_rate": 9.857374017306711e-06, "loss": 0.3575, "step": 9270 }, { "epoch": 1.62, "learning_rate": 9.853125238732627e-06, "loss": 0.3546, "step": 9271 }, { "epoch": 1.62, "learning_rate": 9.848878094548523e-06, "loss": 0.3481, "step": 9272 }, { "epoch": 1.62, "learning_rate": 9.844632585147744e-06, "loss": 0.3549, "step": 9273 }, { "epoch": 1.62, "learning_rate": 9.840388710923506e-06, "loss": 0.3608, "step": 9274 }, { "epoch": 1.62, "learning_rate": 9.836146472268864e-06, "loss": 0.3549, "step": 9275 }, { "epoch": 1.62, "learning_rate": 9.831905869576716e-06, "loss": 0.3469, "step": 9276 }, { "epoch": 1.62, "learning_rate": 9.827666903239798e-06, "loss": 0.3425, "step": 9277 }, { "epoch": 1.62, "learning_rate": 9.823429573650728e-06, "loss": 0.3516, "step": 9278 }, { "epoch": 1.62, "learning_rate": 9.819193881201938e-06, "loss": 0.3667, "step": 9279 }, { "epoch": 1.62, "learning_rate": 9.81495982628573e-06, "loss": 0.3567, "step": 9280 }, { "epoch": 1.62, "learning_rate": 9.81072740929424e-06, "loss": 0.3551, "step": 9281 }, { "epoch": 1.62, "learning_rate": 9.806496630619467e-06, "loss": 0.3633, "step": 9282 }, { "epoch": 1.62, "learning_rate": 9.80226749065325e-06, "loss": 0.3672, "step": 9283 }, { "epoch": 1.62, "learning_rate": 9.798039989787267e-06, "loss": 0.35, "step": 9284 }, { "epoch": 1.62, "learning_rate": 9.79381412841306e-06, "loss": 0.3435, "step": 9285 }, { "epoch": 1.62, "learning_rate": 9.78958990692202e-06, "loss": 0.3448, "step": 9286 }, { "epoch": 1.62, "learning_rate": 9.785367325705367e-06, "loss": 0.3457, "step": 9287 }, { "epoch": 1.62, "learning_rate": 9.781146385154193e-06, "loss": 0.3589, "step": 9288 }, { "epoch": 1.62, "learning_rate": 9.77692708565942e-06, "loss": 0.3715, "step": 9289 }, { "epoch": 1.62, "learning_rate": 9.772709427611819e-06, "loss": 0.3386, "step": 9290 }, { "epoch": 1.62, "learning_rate": 9.768493411402021e-06, "loss": 0.3417, "step": 9291 }, { "epoch": 1.62, "learning_rate": 9.764279037420504e-06, "loss": 0.3444, "step": 9292 }, { "epoch": 1.62, "learning_rate": 9.760066306057578e-06, "loss": 0.3457, "step": 9293 }, { "epoch": 1.62, "learning_rate": 9.755855217703411e-06, "loss": 0.3534, "step": 9294 }, { "epoch": 1.62, "learning_rate": 9.751645772748029e-06, "loss": 0.3672, "step": 9295 }, { "epoch": 1.62, "learning_rate": 9.74743797158128e-06, "loss": 0.3435, "step": 9296 }, { "epoch": 1.62, "learning_rate": 9.743231814592893e-06, "loss": 0.3372, "step": 9297 }, { "epoch": 1.62, "learning_rate": 9.739027302172412e-06, "loss": 0.3463, "step": 9298 }, { "epoch": 1.62, "learning_rate": 9.734824434709256e-06, "loss": 0.3733, "step": 9299 }, { "epoch": 1.62, "learning_rate": 9.730623212592672e-06, "loss": 0.3463, "step": 9300 }, { "epoch": 1.62, "learning_rate": 9.726423636211751e-06, "loss": 0.3489, "step": 9301 }, { "epoch": 1.62, "learning_rate": 9.722225705955471e-06, "loss": 0.3665, "step": 9302 }, { "epoch": 1.62, "learning_rate": 9.71802942221261e-06, "loss": 0.3388, "step": 9303 }, { "epoch": 1.62, "learning_rate": 9.713834785371811e-06, "loss": 0.3653, "step": 9304 }, { "epoch": 1.62, "learning_rate": 9.70964179582158e-06, "loss": 0.3504, "step": 9305 }, { "epoch": 1.62, "learning_rate": 9.705450453950246e-06, "loss": 0.363, "step": 9306 }, { "epoch": 1.62, "learning_rate": 9.701260760145993e-06, "loss": 0.3607, "step": 9307 }, { "epoch": 1.62, "learning_rate": 9.697072714796858e-06, "loss": 0.3519, "step": 9308 }, { "epoch": 1.62, "learning_rate": 9.692886318290734e-06, "loss": 0.3588, "step": 9309 }, { "epoch": 1.62, "learning_rate": 9.68870157101534e-06, "loss": 0.3647, "step": 9310 }, { "epoch": 1.62, "learning_rate": 9.684518473358245e-06, "loss": 0.3587, "step": 9311 }, { "epoch": 1.62, "learning_rate": 9.680337025706887e-06, "loss": 0.3504, "step": 9312 }, { "epoch": 1.62, "learning_rate": 9.676157228448526e-06, "loss": 0.3331, "step": 9313 }, { "epoch": 1.62, "learning_rate": 9.67197908197029e-06, "loss": 0.3507, "step": 9314 }, { "epoch": 1.62, "learning_rate": 9.66780258665913e-06, "loss": 0.343, "step": 9315 }, { "epoch": 1.62, "learning_rate": 9.663627742901873e-06, "loss": 0.3549, "step": 9316 }, { "epoch": 1.62, "learning_rate": 9.659454551085172e-06, "loss": 0.3526, "step": 9317 }, { "epoch": 1.62, "learning_rate": 9.655283011595518e-06, "loss": 0.3476, "step": 9318 }, { "epoch": 1.62, "learning_rate": 9.651113124819289e-06, "loss": 0.3619, "step": 9319 }, { "epoch": 1.62, "learning_rate": 9.646944891142675e-06, "loss": 0.3712, "step": 9320 }, { "epoch": 1.63, "learning_rate": 9.642778310951723e-06, "loss": 0.3523, "step": 9321 }, { "epoch": 1.63, "learning_rate": 9.638613384632317e-06, "loss": 0.3458, "step": 9322 }, { "epoch": 1.63, "learning_rate": 9.634450112570212e-06, "loss": 0.3697, "step": 9323 }, { "epoch": 1.63, "learning_rate": 9.63028849515098e-06, "loss": 0.3397, "step": 9324 }, { "epoch": 1.63, "learning_rate": 9.626128532760069e-06, "loss": 0.3383, "step": 9325 }, { "epoch": 1.63, "learning_rate": 9.62197022578276e-06, "loss": 0.3403, "step": 9326 }, { "epoch": 1.63, "learning_rate": 9.61781357460417e-06, "loss": 0.3496, "step": 9327 }, { "epoch": 1.63, "learning_rate": 9.613658579609275e-06, "loss": 0.3581, "step": 9328 }, { "epoch": 1.63, "learning_rate": 9.609505241182901e-06, "loss": 0.355, "step": 9329 }, { "epoch": 1.63, "learning_rate": 9.605353559709715e-06, "loss": 0.3632, "step": 9330 }, { "epoch": 1.63, "learning_rate": 9.601203535574232e-06, "loss": 0.3493, "step": 9331 }, { "epoch": 1.63, "learning_rate": 9.597055169160801e-06, "loss": 0.3485, "step": 9332 }, { "epoch": 1.63, "learning_rate": 9.592908460853641e-06, "loss": 0.3818, "step": 9333 }, { "epoch": 1.63, "learning_rate": 9.588763411036804e-06, "loss": 0.3516, "step": 9334 }, { "epoch": 1.63, "learning_rate": 9.584620020094182e-06, "loss": 0.3487, "step": 9335 }, { "epoch": 1.63, "learning_rate": 9.580478288409522e-06, "loss": 0.3465, "step": 9336 }, { "epoch": 1.63, "learning_rate": 9.576338216366425e-06, "loss": 0.3486, "step": 9337 }, { "epoch": 1.63, "learning_rate": 9.572199804348328e-06, "loss": 0.3534, "step": 9338 }, { "epoch": 1.63, "learning_rate": 9.568063052738505e-06, "loss": 0.3409, "step": 9339 }, { "epoch": 1.63, "learning_rate": 9.563927961920097e-06, "loss": 0.3593, "step": 9340 }, { "epoch": 1.63, "learning_rate": 9.559794532276078e-06, "loss": 0.3553, "step": 9341 }, { "epoch": 1.63, "learning_rate": 9.55566276418927e-06, "loss": 0.3559, "step": 9342 }, { "epoch": 1.63, "learning_rate": 9.55153265804235e-06, "loss": 0.3467, "step": 9343 }, { "epoch": 1.63, "learning_rate": 9.547404214217828e-06, "loss": 0.3647, "step": 9344 }, { "epoch": 1.63, "learning_rate": 9.543277433098057e-06, "loss": 0.358, "step": 9345 }, { "epoch": 1.63, "learning_rate": 9.539152315065255e-06, "loss": 0.3606, "step": 9346 }, { "epoch": 1.63, "learning_rate": 9.535028860501477e-06, "loss": 0.3631, "step": 9347 }, { "epoch": 1.63, "learning_rate": 9.530907069788621e-06, "loss": 0.3459, "step": 9348 }, { "epoch": 1.63, "learning_rate": 9.526786943308426e-06, "loss": 0.3546, "step": 9349 }, { "epoch": 1.63, "learning_rate": 9.522668481442492e-06, "loss": 0.3583, "step": 9350 }, { "epoch": 1.63, "learning_rate": 9.518551684572248e-06, "loss": 0.3663, "step": 9351 }, { "epoch": 1.63, "learning_rate": 9.514436553078986e-06, "loss": 0.352, "step": 9352 }, { "epoch": 1.63, "learning_rate": 9.510323087343822e-06, "loss": 0.3474, "step": 9353 }, { "epoch": 1.63, "learning_rate": 9.506211287747745e-06, "loss": 0.3528, "step": 9354 }, { "epoch": 1.63, "learning_rate": 9.502101154671566e-06, "loss": 0.3488, "step": 9355 }, { "epoch": 1.63, "learning_rate": 9.497992688495946e-06, "loss": 0.3406, "step": 9356 }, { "epoch": 1.63, "learning_rate": 9.49388588960141e-06, "loss": 0.3578, "step": 9357 }, { "epoch": 1.63, "learning_rate": 9.489780758368305e-06, "loss": 0.3467, "step": 9358 }, { "epoch": 1.63, "learning_rate": 9.485677295176833e-06, "loss": 0.3645, "step": 9359 }, { "epoch": 1.63, "learning_rate": 9.481575500407056e-06, "loss": 0.3622, "step": 9360 }, { "epoch": 1.63, "learning_rate": 9.477475374438851e-06, "loss": 0.3575, "step": 9361 }, { "epoch": 1.63, "learning_rate": 9.473376917651957e-06, "loss": 0.3561, "step": 9362 }, { "epoch": 1.63, "learning_rate": 9.469280130425967e-06, "loss": 0.3494, "step": 9363 }, { "epoch": 1.63, "learning_rate": 9.465185013140313e-06, "loss": 0.3684, "step": 9364 }, { "epoch": 1.63, "learning_rate": 9.461091566174266e-06, "loss": 0.341, "step": 9365 }, { "epoch": 1.63, "learning_rate": 9.456999789906935e-06, "loss": 0.3679, "step": 9366 }, { "epoch": 1.63, "learning_rate": 9.452909684717304e-06, "loss": 0.3456, "step": 9367 }, { "epoch": 1.63, "learning_rate": 9.448821250984169e-06, "loss": 0.3621, "step": 9368 }, { "epoch": 1.63, "learning_rate": 9.444734489086196e-06, "loss": 0.3518, "step": 9369 }, { "epoch": 1.63, "learning_rate": 9.440649399401882e-06, "loss": 0.3532, "step": 9370 }, { "epoch": 1.63, "learning_rate": 9.436565982309575e-06, "loss": 0.3562, "step": 9371 }, { "epoch": 1.63, "learning_rate": 9.432484238187467e-06, "loss": 0.3681, "step": 9372 }, { "epoch": 1.63, "learning_rate": 9.428404167413587e-06, "loss": 0.356, "step": 9373 }, { "epoch": 1.63, "learning_rate": 9.424325770365822e-06, "loss": 0.3565, "step": 9374 }, { "epoch": 1.63, "learning_rate": 9.420249047421903e-06, "loss": 0.3676, "step": 9375 }, { "epoch": 1.63, "learning_rate": 9.416173998959398e-06, "loss": 0.3537, "step": 9376 }, { "epoch": 1.63, "learning_rate": 9.412100625355718e-06, "loss": 0.3599, "step": 9377 }, { "epoch": 1.64, "learning_rate": 9.408028926988135e-06, "loss": 0.3645, "step": 9378 }, { "epoch": 1.64, "learning_rate": 9.403958904233738e-06, "loss": 0.3531, "step": 9379 }, { "epoch": 1.64, "learning_rate": 9.39989055746949e-06, "loss": 0.3558, "step": 9380 }, { "epoch": 1.64, "learning_rate": 9.395823887072192e-06, "loss": 0.3388, "step": 9381 }, { "epoch": 1.64, "learning_rate": 9.391758893418478e-06, "loss": 0.3784, "step": 9382 }, { "epoch": 1.64, "learning_rate": 9.387695576884828e-06, "loss": 0.3624, "step": 9383 }, { "epoch": 1.64, "learning_rate": 9.383633937847573e-06, "loss": 0.3489, "step": 9384 }, { "epoch": 1.64, "learning_rate": 9.379573976682897e-06, "loss": 0.3522, "step": 9385 }, { "epoch": 1.64, "learning_rate": 9.375515693766812e-06, "loss": 0.334, "step": 9386 }, { "epoch": 1.64, "learning_rate": 9.371459089475176e-06, "loss": 0.3553, "step": 9387 }, { "epoch": 1.64, "learning_rate": 9.36740416418371e-06, "loss": 0.3337, "step": 9388 }, { "epoch": 1.64, "learning_rate": 9.36335091826796e-06, "loss": 0.3417, "step": 9389 }, { "epoch": 1.64, "learning_rate": 9.359299352103315e-06, "loss": 0.3493, "step": 9390 }, { "epoch": 1.64, "learning_rate": 9.355249466065028e-06, "loss": 0.3613, "step": 9391 }, { "epoch": 1.64, "learning_rate": 9.351201260528184e-06, "loss": 0.3551, "step": 9392 }, { "epoch": 1.64, "learning_rate": 9.347154735867711e-06, "loss": 0.3486, "step": 9393 }, { "epoch": 1.64, "learning_rate": 9.343109892458377e-06, "loss": 0.3466, "step": 9394 }, { "epoch": 1.64, "learning_rate": 9.339066730674805e-06, "loss": 0.3501, "step": 9395 }, { "epoch": 1.64, "learning_rate": 9.335025250891469e-06, "loss": 0.3533, "step": 9396 }, { "epoch": 1.64, "learning_rate": 9.33098545348266e-06, "loss": 0.3493, "step": 9397 }, { "epoch": 1.64, "learning_rate": 9.32694733882254e-06, "loss": 0.3632, "step": 9398 }, { "epoch": 1.64, "learning_rate": 9.322910907285102e-06, "loss": 0.3674, "step": 9399 }, { "epoch": 1.64, "learning_rate": 9.31887615924418e-06, "loss": 0.3471, "step": 9400 }, { "epoch": 1.64, "learning_rate": 9.314843095073463e-06, "loss": 0.3424, "step": 9401 }, { "epoch": 1.64, "learning_rate": 9.310811715146483e-06, "loss": 0.3452, "step": 9402 }, { "epoch": 1.64, "learning_rate": 9.30678201983661e-06, "loss": 0.3548, "step": 9403 }, { "epoch": 1.64, "learning_rate": 9.302754009517051e-06, "loss": 0.3531, "step": 9404 }, { "epoch": 1.64, "learning_rate": 9.298727684560873e-06, "loss": 0.3471, "step": 9405 }, { "epoch": 1.64, "learning_rate": 9.294703045340977e-06, "loss": 0.3658, "step": 9406 }, { "epoch": 1.64, "learning_rate": 9.29068009223012e-06, "loss": 0.3713, "step": 9407 }, { "epoch": 1.64, "learning_rate": 9.286658825600878e-06, "loss": 0.3388, "step": 9408 }, { "epoch": 1.64, "learning_rate": 9.282639245825702e-06, "loss": 0.3751, "step": 9409 }, { "epoch": 1.64, "learning_rate": 9.27862135327686e-06, "loss": 0.3645, "step": 9410 }, { "epoch": 1.64, "learning_rate": 9.274605148326475e-06, "loss": 0.3639, "step": 9411 }, { "epoch": 1.64, "learning_rate": 9.270590631346515e-06, "loss": 0.3474, "step": 9412 }, { "epoch": 1.64, "learning_rate": 9.266577802708798e-06, "loss": 0.3692, "step": 9413 }, { "epoch": 1.64, "learning_rate": 9.262566662784969e-06, "loss": 0.3458, "step": 9414 }, { "epoch": 1.64, "learning_rate": 9.258557211946523e-06, "loss": 0.3704, "step": 9415 }, { "epoch": 1.64, "learning_rate": 9.254549450564812e-06, "loss": 0.3528, "step": 9416 }, { "epoch": 1.64, "learning_rate": 9.250543379011011e-06, "loss": 0.3634, "step": 9417 }, { "epoch": 1.64, "learning_rate": 9.246538997656143e-06, "loss": 0.3406, "step": 9418 }, { "epoch": 1.64, "learning_rate": 9.242536306871097e-06, "loss": 0.3669, "step": 9419 }, { "epoch": 1.64, "learning_rate": 9.238535307026574e-06, "loss": 0.3539, "step": 9420 }, { "epoch": 1.64, "learning_rate": 9.234535998493135e-06, "loss": 0.3621, "step": 9421 }, { "epoch": 1.64, "learning_rate": 9.230538381641187e-06, "loss": 0.3579, "step": 9422 }, { "epoch": 1.64, "learning_rate": 9.226542456840968e-06, "loss": 0.3408, "step": 9423 }, { "epoch": 1.64, "learning_rate": 9.22254822446257e-06, "loss": 0.3671, "step": 9424 }, { "epoch": 1.64, "learning_rate": 9.21855568487592e-06, "loss": 0.3551, "step": 9425 }, { "epoch": 1.64, "learning_rate": 9.214564838450799e-06, "loss": 0.3492, "step": 9426 }, { "epoch": 1.64, "learning_rate": 9.210575685556824e-06, "loss": 0.3688, "step": 9427 }, { "epoch": 1.64, "learning_rate": 9.206588226563446e-06, "loss": 0.3637, "step": 9428 }, { "epoch": 1.64, "learning_rate": 9.202602461839976e-06, "loss": 0.3622, "step": 9429 }, { "epoch": 1.64, "learning_rate": 9.19861839175557e-06, "loss": 0.3539, "step": 9430 }, { "epoch": 1.64, "learning_rate": 9.194636016679209e-06, "loss": 0.3494, "step": 9431 }, { "epoch": 1.64, "learning_rate": 9.190655336979722e-06, "loss": 0.3441, "step": 9432 }, { "epoch": 1.64, "learning_rate": 9.186676353025794e-06, "loss": 0.3599, "step": 9433 }, { "epoch": 1.64, "learning_rate": 9.182699065185938e-06, "loss": 0.3585, "step": 9434 }, { "epoch": 1.64, "learning_rate": 9.178723473828517e-06, "loss": 0.3454, "step": 9435 }, { "epoch": 1.65, "learning_rate": 9.174749579321743e-06, "loss": 0.352, "step": 9436 }, { "epoch": 1.65, "learning_rate": 9.170777382033661e-06, "loss": 0.3448, "step": 9437 }, { "epoch": 1.65, "learning_rate": 9.166806882332149e-06, "loss": 0.3547, "step": 9438 }, { "epoch": 1.65, "learning_rate": 9.16283808058496e-06, "loss": 0.3576, "step": 9439 }, { "epoch": 1.65, "learning_rate": 9.158870977159654e-06, "loss": 0.3456, "step": 9440 }, { "epoch": 1.65, "learning_rate": 9.15490557242366e-06, "loss": 0.362, "step": 9441 }, { "epoch": 1.65, "learning_rate": 9.150941866744232e-06, "loss": 0.3488, "step": 9442 }, { "epoch": 1.65, "learning_rate": 9.146979860488482e-06, "loss": 0.3417, "step": 9443 }, { "epoch": 1.65, "learning_rate": 9.143019554023355e-06, "loss": 0.3615, "step": 9444 }, { "epoch": 1.65, "learning_rate": 9.139060947715633e-06, "loss": 0.3548, "step": 9445 }, { "epoch": 1.65, "learning_rate": 9.13510404193195e-06, "loss": 0.3413, "step": 9446 }, { "epoch": 1.65, "learning_rate": 9.13114883703879e-06, "loss": 0.3424, "step": 9447 }, { "epoch": 1.65, "learning_rate": 9.127195333402466e-06, "loss": 0.3406, "step": 9448 }, { "epoch": 1.65, "learning_rate": 9.123243531389125e-06, "loss": 0.3545, "step": 9449 }, { "epoch": 1.65, "learning_rate": 9.119293431364786e-06, "loss": 0.3422, "step": 9450 }, { "epoch": 1.65, "learning_rate": 9.115345033695278e-06, "loss": 0.3623, "step": 9451 }, { "epoch": 1.65, "learning_rate": 9.111398338746296e-06, "loss": 0.3786, "step": 9452 }, { "epoch": 1.65, "learning_rate": 9.107453346883369e-06, "loss": 0.3543, "step": 9453 }, { "epoch": 1.65, "learning_rate": 9.103510058471871e-06, "loss": 0.3525, "step": 9454 }, { "epoch": 1.65, "learning_rate": 9.099568473877006e-06, "loss": 0.348, "step": 9455 }, { "epoch": 1.65, "learning_rate": 9.095628593463826e-06, "loss": 0.3467, "step": 9456 }, { "epoch": 1.65, "learning_rate": 9.091690417597245e-06, "loss": 0.3628, "step": 9457 }, { "epoch": 1.65, "learning_rate": 9.087753946641993e-06, "loss": 0.3384, "step": 9458 }, { "epoch": 1.65, "learning_rate": 9.083819180962653e-06, "loss": 0.3557, "step": 9459 }, { "epoch": 1.65, "learning_rate": 9.079886120923651e-06, "loss": 0.3396, "step": 9460 }, { "epoch": 1.65, "learning_rate": 9.075954766889246e-06, "loss": 0.3557, "step": 9461 }, { "epoch": 1.65, "learning_rate": 9.07202511922355e-06, "loss": 0.3592, "step": 9462 }, { "epoch": 1.65, "learning_rate": 9.06809717829051e-06, "loss": 0.3478, "step": 9463 }, { "epoch": 1.65, "learning_rate": 9.064170944453925e-06, "loss": 0.3568, "step": 9464 }, { "epoch": 1.65, "learning_rate": 9.060246418077429e-06, "loss": 0.3562, "step": 9465 }, { "epoch": 1.65, "learning_rate": 9.056323599524483e-06, "loss": 0.3419, "step": 9466 }, { "epoch": 1.65, "learning_rate": 9.052402489158419e-06, "loss": 0.3517, "step": 9467 }, { "epoch": 1.65, "learning_rate": 9.04848308734239e-06, "loss": 0.3617, "step": 9468 }, { "epoch": 1.65, "learning_rate": 9.044565394439402e-06, "loss": 0.3628, "step": 9469 }, { "epoch": 1.65, "learning_rate": 9.040649410812289e-06, "loss": 0.3569, "step": 9470 }, { "epoch": 1.65, "learning_rate": 9.036735136823743e-06, "loss": 0.3429, "step": 9471 }, { "epoch": 1.65, "learning_rate": 9.03282257283629e-06, "loss": 0.3456, "step": 9472 }, { "epoch": 1.65, "learning_rate": 9.028911719212282e-06, "loss": 0.3632, "step": 9473 }, { "epoch": 1.65, "learning_rate": 9.025002576313951e-06, "loss": 0.3534, "step": 9474 }, { "epoch": 1.65, "learning_rate": 9.021095144503338e-06, "loss": 0.367, "step": 9475 }, { "epoch": 1.65, "learning_rate": 9.01718942414233e-06, "loss": 0.3386, "step": 9476 }, { "epoch": 1.65, "learning_rate": 9.01328541559267e-06, "loss": 0.3599, "step": 9477 }, { "epoch": 1.65, "learning_rate": 9.009383119215926e-06, "loss": 0.3476, "step": 9478 }, { "epoch": 1.65, "learning_rate": 9.00548253537352e-06, "loss": 0.3514, "step": 9479 }, { "epoch": 1.65, "learning_rate": 9.0015836644267e-06, "loss": 0.3401, "step": 9480 }, { "epoch": 1.65, "learning_rate": 8.997686506736583e-06, "loss": 0.3597, "step": 9481 }, { "epoch": 1.65, "learning_rate": 8.993791062664098e-06, "loss": 0.3552, "step": 9482 }, { "epoch": 1.65, "learning_rate": 8.989897332570021e-06, "loss": 0.3447, "step": 9483 }, { "epoch": 1.65, "learning_rate": 8.986005316814983e-06, "loss": 0.3498, "step": 9484 }, { "epoch": 1.65, "learning_rate": 8.982115015759454e-06, "loss": 0.357, "step": 9485 }, { "epoch": 1.65, "learning_rate": 8.978226429763736e-06, "loss": 0.3608, "step": 9486 }, { "epoch": 1.65, "learning_rate": 8.974339559187967e-06, "loss": 0.3461, "step": 9487 }, { "epoch": 1.65, "learning_rate": 8.970454404392146e-06, "loss": 0.347, "step": 9488 }, { "epoch": 1.65, "learning_rate": 8.966570965736095e-06, "loss": 0.3448, "step": 9489 }, { "epoch": 1.65, "learning_rate": 8.962689243579486e-06, "loss": 0.3737, "step": 9490 }, { "epoch": 1.65, "learning_rate": 8.958809238281834e-06, "loss": 0.3515, "step": 9491 }, { "epoch": 1.65, "learning_rate": 8.954930950202489e-06, "loss": 0.3631, "step": 9492 }, { "epoch": 1.66, "learning_rate": 8.951054379700637e-06, "loss": 0.3576, "step": 9493 }, { "epoch": 1.66, "learning_rate": 8.947179527135328e-06, "loss": 0.3671, "step": 9494 }, { "epoch": 1.66, "learning_rate": 8.943306392865417e-06, "loss": 0.3491, "step": 9495 }, { "epoch": 1.66, "learning_rate": 8.939434977249635e-06, "loss": 0.3446, "step": 9496 }, { "epoch": 1.66, "learning_rate": 8.935565280646529e-06, "loss": 0.3327, "step": 9497 }, { "epoch": 1.66, "learning_rate": 8.931697303414508e-06, "loss": 0.3531, "step": 9498 }, { "epoch": 1.66, "learning_rate": 8.927831045911802e-06, "loss": 0.3629, "step": 9499 }, { "epoch": 1.66, "learning_rate": 8.923966508496482e-06, "loss": 0.3541, "step": 9500 }, { "epoch": 1.66, "learning_rate": 8.92010369152648e-06, "loss": 0.3423, "step": 9501 }, { "epoch": 1.66, "learning_rate": 8.916242595359556e-06, "loss": 0.3694, "step": 9502 }, { "epoch": 1.66, "learning_rate": 8.912383220353307e-06, "loss": 0.3413, "step": 9503 }, { "epoch": 1.66, "learning_rate": 8.908525566865173e-06, "loss": 0.357, "step": 9504 }, { "epoch": 1.66, "learning_rate": 8.904669635252438e-06, "loss": 0.3572, "step": 9505 }, { "epoch": 1.66, "learning_rate": 8.90081542587222e-06, "loss": 0.3555, "step": 9506 }, { "epoch": 1.66, "learning_rate": 8.896962939081492e-06, "loss": 0.348, "step": 9507 }, { "epoch": 1.66, "learning_rate": 8.893112175237048e-06, "loss": 0.3484, "step": 9508 }, { "epoch": 1.66, "learning_rate": 8.889263134695535e-06, "loss": 0.3532, "step": 9509 }, { "epoch": 1.66, "learning_rate": 8.885415817813444e-06, "loss": 0.3581, "step": 9510 }, { "epoch": 1.66, "learning_rate": 8.881570224947081e-06, "loss": 0.347, "step": 9511 }, { "epoch": 1.66, "learning_rate": 8.877726356452632e-06, "loss": 0.3521, "step": 9512 }, { "epoch": 1.66, "learning_rate": 8.873884212686095e-06, "loss": 0.3701, "step": 9513 }, { "epoch": 1.66, "learning_rate": 8.870043794003307e-06, "loss": 0.3499, "step": 9514 }, { "epoch": 1.66, "learning_rate": 8.86620510075997e-06, "loss": 0.3613, "step": 9515 }, { "epoch": 1.66, "learning_rate": 8.8623681333116e-06, "loss": 0.3384, "step": 9516 }, { "epoch": 1.66, "learning_rate": 8.858532892013555e-06, "loss": 0.3429, "step": 9517 }, { "epoch": 1.66, "learning_rate": 8.854699377221052e-06, "loss": 0.3406, "step": 9518 }, { "epoch": 1.66, "learning_rate": 8.85086758928914e-06, "loss": 0.3558, "step": 9519 }, { "epoch": 1.66, "learning_rate": 8.847037528572701e-06, "loss": 0.3641, "step": 9520 }, { "epoch": 1.66, "learning_rate": 8.843209195426457e-06, "loss": 0.3673, "step": 9521 }, { "epoch": 1.66, "learning_rate": 8.839382590204983e-06, "loss": 0.3543, "step": 9522 }, { "epoch": 1.66, "learning_rate": 8.835557713262673e-06, "loss": 0.3589, "step": 9523 }, { "epoch": 1.66, "learning_rate": 8.831734564953791e-06, "loss": 0.3457, "step": 9524 }, { "epoch": 1.66, "learning_rate": 8.827913145632407e-06, "loss": 0.345, "step": 9525 }, { "epoch": 1.66, "learning_rate": 8.82409345565246e-06, "loss": 0.3545, "step": 9526 }, { "epoch": 1.66, "learning_rate": 8.820275495367708e-06, "loss": 0.3404, "step": 9527 }, { "epoch": 1.66, "learning_rate": 8.816459265131747e-06, "loss": 0.3599, "step": 9528 }, { "epoch": 1.66, "learning_rate": 8.812644765298047e-06, "loss": 0.3399, "step": 9529 }, { "epoch": 1.66, "learning_rate": 8.80883199621988e-06, "loss": 0.3306, "step": 9530 }, { "epoch": 1.66, "learning_rate": 8.805020958250363e-06, "loss": 0.3702, "step": 9531 }, { "epoch": 1.66, "learning_rate": 8.801211651742477e-06, "loss": 0.3543, "step": 9532 }, { "epoch": 1.66, "learning_rate": 8.797404077049015e-06, "loss": 0.348, "step": 9533 }, { "epoch": 1.66, "learning_rate": 8.793598234522618e-06, "loss": 0.3556, "step": 9534 }, { "epoch": 1.66, "learning_rate": 8.78979412451578e-06, "loss": 0.385, "step": 9535 }, { "epoch": 1.66, "learning_rate": 8.78599174738082e-06, "loss": 0.3616, "step": 9536 }, { "epoch": 1.66, "learning_rate": 8.7821911034699e-06, "loss": 0.3487, "step": 9537 }, { "epoch": 1.66, "learning_rate": 8.778392193135018e-06, "loss": 0.3547, "step": 9538 }, { "epoch": 1.66, "learning_rate": 8.774595016728018e-06, "loss": 0.3592, "step": 9539 }, { "epoch": 1.66, "learning_rate": 8.770799574600588e-06, "loss": 0.3533, "step": 9540 }, { "epoch": 1.66, "learning_rate": 8.767005867104241e-06, "loss": 0.3467, "step": 9541 }, { "epoch": 1.66, "learning_rate": 8.763213894590334e-06, "loss": 0.3579, "step": 9542 }, { "epoch": 1.66, "learning_rate": 8.75942365741007e-06, "loss": 0.3525, "step": 9543 }, { "epoch": 1.66, "learning_rate": 8.755635155914488e-06, "loss": 0.3582, "step": 9544 }, { "epoch": 1.66, "learning_rate": 8.75184839045446e-06, "loss": 0.3527, "step": 9545 }, { "epoch": 1.66, "learning_rate": 8.748063361380707e-06, "loss": 0.3567, "step": 9546 }, { "epoch": 1.66, "learning_rate": 8.744280069043789e-06, "loss": 0.3522, "step": 9547 }, { "epoch": 1.66, "learning_rate": 8.740498513794094e-06, "loss": 0.3418, "step": 9548 }, { "epoch": 1.66, "learning_rate": 8.736718695981855e-06, "loss": 0.3723, "step": 9549 }, { "epoch": 1.67, "learning_rate": 8.732940615957147e-06, "loss": 0.3529, "step": 9550 }, { "epoch": 1.67, "learning_rate": 8.72916427406989e-06, "loss": 0.3579, "step": 9551 }, { "epoch": 1.67, "learning_rate": 8.72538967066982e-06, "loss": 0.3484, "step": 9552 }, { "epoch": 1.67, "learning_rate": 8.721616806106545e-06, "loss": 0.355, "step": 9553 }, { "epoch": 1.67, "learning_rate": 8.717845680729483e-06, "loss": 0.3616, "step": 9554 }, { "epoch": 1.67, "learning_rate": 8.714076294887896e-06, "loss": 0.3394, "step": 9555 }, { "epoch": 1.67, "learning_rate": 8.710308648930903e-06, "loss": 0.3498, "step": 9556 }, { "epoch": 1.67, "learning_rate": 8.706542743207449e-06, "loss": 0.3437, "step": 9557 }, { "epoch": 1.67, "learning_rate": 8.702778578066316e-06, "loss": 0.3496, "step": 9558 }, { "epoch": 1.67, "learning_rate": 8.699016153856118e-06, "loss": 0.3583, "step": 9559 }, { "epoch": 1.67, "learning_rate": 8.695255470925338e-06, "loss": 0.365, "step": 9560 }, { "epoch": 1.67, "learning_rate": 8.69149652962225e-06, "loss": 0.3436, "step": 9561 }, { "epoch": 1.67, "learning_rate": 8.687739330295019e-06, "loss": 0.3608, "step": 9562 }, { "epoch": 1.67, "learning_rate": 8.68398387329161e-06, "loss": 0.3504, "step": 9563 }, { "epoch": 1.67, "learning_rate": 8.680230158959845e-06, "loss": 0.3556, "step": 9564 }, { "epoch": 1.67, "learning_rate": 8.676478187647377e-06, "loss": 0.3547, "step": 9565 }, { "epoch": 1.67, "learning_rate": 8.672727959701693e-06, "loss": 0.3526, "step": 9566 }, { "epoch": 1.67, "learning_rate": 8.668979475470133e-06, "loss": 0.3772, "step": 9567 }, { "epoch": 1.67, "learning_rate": 8.665232735299875e-06, "loss": 0.3582, "step": 9568 }, { "epoch": 1.67, "learning_rate": 8.661487739537915e-06, "loss": 0.329, "step": 9569 }, { "epoch": 1.67, "learning_rate": 8.657744488531113e-06, "loss": 0.3638, "step": 9570 }, { "epoch": 1.67, "learning_rate": 8.65400298262615e-06, "loss": 0.343, "step": 9571 }, { "epoch": 1.67, "learning_rate": 8.650263222169545e-06, "loss": 0.3577, "step": 9572 }, { "epoch": 1.67, "learning_rate": 8.646525207507667e-06, "loss": 0.3255, "step": 9573 }, { "epoch": 1.67, "learning_rate": 8.642788938986722e-06, "loss": 0.3329, "step": 9574 }, { "epoch": 1.67, "learning_rate": 8.639054416952747e-06, "loss": 0.3514, "step": 9575 }, { "epoch": 1.67, "learning_rate": 8.63532164175161e-06, "loss": 0.342, "step": 9576 }, { "epoch": 1.67, "learning_rate": 8.63159061372904e-06, "loss": 0.3558, "step": 9577 }, { "epoch": 1.67, "learning_rate": 8.627861333230584e-06, "loss": 0.3559, "step": 9578 }, { "epoch": 1.67, "learning_rate": 8.624133800601638e-06, "loss": 0.3534, "step": 9579 }, { "epoch": 1.67, "learning_rate": 8.620408016187428e-06, "loss": 0.3461, "step": 9580 }, { "epoch": 1.67, "learning_rate": 8.616683980333033e-06, "loss": 0.3673, "step": 9581 }, { "epoch": 1.67, "learning_rate": 8.612961693383353e-06, "loss": 0.3292, "step": 9582 }, { "epoch": 1.67, "learning_rate": 8.609241155683124e-06, "loss": 0.3486, "step": 9583 }, { "epoch": 1.67, "learning_rate": 8.605522367576945e-06, "loss": 0.3468, "step": 9584 }, { "epoch": 1.67, "learning_rate": 8.601805329409233e-06, "loss": 0.3494, "step": 9585 }, { "epoch": 1.67, "learning_rate": 8.598090041524234e-06, "loss": 0.3533, "step": 9586 }, { "epoch": 1.67, "learning_rate": 8.59437650426606e-06, "loss": 0.3597, "step": 9587 }, { "epoch": 1.67, "learning_rate": 8.590664717978642e-06, "loss": 0.3477, "step": 9588 }, { "epoch": 1.67, "learning_rate": 8.58695468300574e-06, "loss": 0.3607, "step": 9589 }, { "epoch": 1.67, "learning_rate": 8.583246399690979e-06, "loss": 0.3569, "step": 9590 }, { "epoch": 1.67, "learning_rate": 8.579539868377803e-06, "loss": 0.3749, "step": 9591 }, { "epoch": 1.67, "learning_rate": 8.575835089409499e-06, "loss": 0.3415, "step": 9592 }, { "epoch": 1.67, "learning_rate": 8.572132063129177e-06, "loss": 0.3428, "step": 9593 }, { "epoch": 1.67, "learning_rate": 8.568430789879812e-06, "loss": 0.3372, "step": 9594 }, { "epoch": 1.67, "learning_rate": 8.564731270004203e-06, "loss": 0.36, "step": 9595 }, { "epoch": 1.67, "learning_rate": 8.561033503844984e-06, "loss": 0.3425, "step": 9596 }, { "epoch": 1.67, "learning_rate": 8.557337491744622e-06, "loss": 0.3429, "step": 9597 }, { "epoch": 1.67, "learning_rate": 8.553643234045434e-06, "loss": 0.3581, "step": 9598 }, { "epoch": 1.67, "learning_rate": 8.54995073108957e-06, "loss": 0.3362, "step": 9599 }, { "epoch": 1.67, "learning_rate": 8.54625998321901e-06, "loss": 0.3525, "step": 9600 }, { "epoch": 1.67, "learning_rate": 8.542570990775583e-06, "loss": 0.3382, "step": 9601 }, { "epoch": 1.67, "learning_rate": 8.538883754100952e-06, "loss": 0.3445, "step": 9602 }, { "epoch": 1.67, "learning_rate": 8.535198273536616e-06, "loss": 0.3465, "step": 9603 }, { "epoch": 1.67, "learning_rate": 8.531514549423899e-06, "loss": 0.3311, "step": 9604 }, { "epoch": 1.67, "learning_rate": 8.527832582103985e-06, "loss": 0.3635, "step": 9605 }, { "epoch": 1.67, "learning_rate": 8.524152371917887e-06, "loss": 0.3773, "step": 9606 }, { "epoch": 1.67, "learning_rate": 8.520473919206445e-06, "loss": 0.3637, "step": 9607 }, { "epoch": 1.68, "learning_rate": 8.51679722431035e-06, "loss": 0.356, "step": 9608 }, { "epoch": 1.68, "learning_rate": 8.513122287570125e-06, "loss": 0.369, "step": 9609 }, { "epoch": 1.68, "learning_rate": 8.509449109326118e-06, "loss": 0.3573, "step": 9610 }, { "epoch": 1.68, "learning_rate": 8.505777689918537e-06, "loss": 0.3522, "step": 9611 }, { "epoch": 1.68, "learning_rate": 8.502108029687413e-06, "loss": 0.3606, "step": 9612 }, { "epoch": 1.68, "learning_rate": 8.49844012897262e-06, "loss": 0.3542, "step": 9613 }, { "epoch": 1.68, "learning_rate": 8.494773988113855e-06, "loss": 0.3552, "step": 9614 }, { "epoch": 1.68, "learning_rate": 8.491109607450676e-06, "loss": 0.3537, "step": 9615 }, { "epoch": 1.68, "learning_rate": 8.48744698732245e-06, "loss": 0.3578, "step": 9616 }, { "epoch": 1.68, "learning_rate": 8.48378612806841e-06, "loss": 0.3613, "step": 9617 }, { "epoch": 1.68, "learning_rate": 8.480127030027603e-06, "loss": 0.3497, "step": 9618 }, { "epoch": 1.68, "learning_rate": 8.476469693538928e-06, "loss": 0.3504, "step": 9619 }, { "epoch": 1.68, "learning_rate": 8.472814118941111e-06, "loss": 0.3527, "step": 9620 }, { "epoch": 1.68, "learning_rate": 8.469160306572712e-06, "loss": 0.3387, "step": 9621 }, { "epoch": 1.68, "learning_rate": 8.465508256772138e-06, "loss": 0.3456, "step": 9622 }, { "epoch": 1.68, "learning_rate": 8.461857969877636e-06, "loss": 0.3526, "step": 9623 }, { "epoch": 1.68, "learning_rate": 8.458209446227273e-06, "loss": 0.3514, "step": 9624 }, { "epoch": 1.68, "learning_rate": 8.454562686158968e-06, "loss": 0.3416, "step": 9625 }, { "epoch": 1.68, "learning_rate": 8.450917690010471e-06, "loss": 0.3522, "step": 9626 }, { "epoch": 1.68, "learning_rate": 8.44727445811936e-06, "loss": 0.333, "step": 9627 }, { "epoch": 1.68, "learning_rate": 8.44363299082306e-06, "loss": 0.3633, "step": 9628 }, { "epoch": 1.68, "learning_rate": 8.439993288458847e-06, "loss": 0.3542, "step": 9629 }, { "epoch": 1.68, "learning_rate": 8.436355351363801e-06, "loss": 0.3593, "step": 9630 }, { "epoch": 1.68, "learning_rate": 8.432719179874855e-06, "loss": 0.3551, "step": 9631 }, { "epoch": 1.68, "learning_rate": 8.429084774328782e-06, "loss": 0.3463, "step": 9632 }, { "epoch": 1.68, "learning_rate": 8.425452135062185e-06, "loss": 0.3511, "step": 9633 }, { "epoch": 1.68, "learning_rate": 8.42182126241151e-06, "loss": 0.3509, "step": 9634 }, { "epoch": 1.68, "learning_rate": 8.418192156713032e-06, "loss": 0.3488, "step": 9635 }, { "epoch": 1.68, "learning_rate": 8.41456481830287e-06, "loss": 0.3509, "step": 9636 }, { "epoch": 1.68, "learning_rate": 8.410939247516974e-06, "loss": 0.3563, "step": 9637 }, { "epoch": 1.68, "learning_rate": 8.407315444691121e-06, "loss": 0.3489, "step": 9638 }, { "epoch": 1.68, "learning_rate": 8.403693410160946e-06, "loss": 0.3642, "step": 9639 }, { "epoch": 1.68, "learning_rate": 8.40007314426191e-06, "loss": 0.3661, "step": 9640 }, { "epoch": 1.68, "learning_rate": 8.396454647329303e-06, "loss": 0.3567, "step": 9641 }, { "epoch": 1.68, "learning_rate": 8.392837919698254e-06, "loss": 0.3574, "step": 9642 }, { "epoch": 1.68, "learning_rate": 8.389222961703744e-06, "loss": 0.3406, "step": 9643 }, { "epoch": 1.68, "learning_rate": 8.385609773680561e-06, "loss": 0.3456, "step": 9644 }, { "epoch": 1.68, "learning_rate": 8.381998355963355e-06, "loss": 0.3698, "step": 9645 }, { "epoch": 1.68, "learning_rate": 8.378388708886607e-06, "loss": 0.3534, "step": 9646 }, { "epoch": 1.68, "learning_rate": 8.374780832784623e-06, "loss": 0.3504, "step": 9647 }, { "epoch": 1.68, "learning_rate": 8.371174727991546e-06, "loss": 0.3422, "step": 9648 }, { "epoch": 1.68, "learning_rate": 8.367570394841373e-06, "loss": 0.3564, "step": 9649 }, { "epoch": 1.68, "learning_rate": 8.363967833667912e-06, "loss": 0.3611, "step": 9650 }, { "epoch": 1.68, "learning_rate": 8.360367044804835e-06, "loss": 0.353, "step": 9651 }, { "epoch": 1.68, "learning_rate": 8.356768028585616e-06, "loss": 0.3601, "step": 9652 }, { "epoch": 1.68, "learning_rate": 8.353170785343598e-06, "loss": 0.3458, "step": 9653 }, { "epoch": 1.68, "learning_rate": 8.349575315411936e-06, "loss": 0.3658, "step": 9654 }, { "epoch": 1.68, "learning_rate": 8.345981619123628e-06, "loss": 0.3332, "step": 9655 }, { "epoch": 1.68, "learning_rate": 8.342389696811512e-06, "loss": 0.3412, "step": 9656 }, { "epoch": 1.68, "learning_rate": 8.338799548808269e-06, "loss": 0.3441, "step": 9657 }, { "epoch": 1.68, "learning_rate": 8.335211175446391e-06, "loss": 0.3539, "step": 9658 }, { "epoch": 1.68, "learning_rate": 8.331624577058223e-06, "loss": 0.353, "step": 9659 }, { "epoch": 1.68, "learning_rate": 8.32803975397595e-06, "loss": 0.3479, "step": 9660 }, { "epoch": 1.68, "learning_rate": 8.324456706531575e-06, "loss": 0.3419, "step": 9661 }, { "epoch": 1.68, "learning_rate": 8.320875435056952e-06, "loss": 0.3447, "step": 9662 }, { "epoch": 1.68, "learning_rate": 8.317295939883773e-06, "loss": 0.3539, "step": 9663 }, { "epoch": 1.68, "learning_rate": 8.31371822134355e-06, "loss": 0.347, "step": 9664 }, { "epoch": 1.69, "learning_rate": 8.310142279767635e-06, "loss": 0.3543, "step": 9665 }, { "epoch": 1.69, "learning_rate": 8.306568115487223e-06, "loss": 0.3582, "step": 9666 }, { "epoch": 1.69, "learning_rate": 8.302995728833346e-06, "loss": 0.3594, "step": 9667 }, { "epoch": 1.69, "learning_rate": 8.29942512013686e-06, "loss": 0.3571, "step": 9668 }, { "epoch": 1.69, "learning_rate": 8.295856289728457e-06, "loss": 0.3434, "step": 9669 }, { "epoch": 1.69, "learning_rate": 8.29228923793868e-06, "loss": 0.3701, "step": 9670 }, { "epoch": 1.69, "learning_rate": 8.288723965097889e-06, "loss": 0.3686, "step": 9671 }, { "epoch": 1.69, "learning_rate": 8.285160471536285e-06, "loss": 0.3648, "step": 9672 }, { "epoch": 1.69, "learning_rate": 8.281598757583908e-06, "loss": 0.3453, "step": 9673 }, { "epoch": 1.69, "learning_rate": 8.278038823570641e-06, "loss": 0.3608, "step": 9674 }, { "epoch": 1.69, "learning_rate": 8.274480669826181e-06, "loss": 0.3567, "step": 9675 }, { "epoch": 1.69, "learning_rate": 8.270924296680069e-06, "loss": 0.3486, "step": 9676 }, { "epoch": 1.69, "learning_rate": 8.26736970446169e-06, "loss": 0.3416, "step": 9677 }, { "epoch": 1.69, "learning_rate": 8.263816893500262e-06, "loss": 0.3645, "step": 9678 }, { "epoch": 1.69, "learning_rate": 8.260265864124824e-06, "loss": 0.3633, "step": 9679 }, { "epoch": 1.69, "learning_rate": 8.256716616664262e-06, "loss": 0.3519, "step": 9680 }, { "epoch": 1.69, "learning_rate": 8.253169151447302e-06, "loss": 0.3558, "step": 9681 }, { "epoch": 1.69, "learning_rate": 8.249623468802488e-06, "loss": 0.3516, "step": 9682 }, { "epoch": 1.69, "learning_rate": 8.246079569058203e-06, "loss": 0.3528, "step": 9683 }, { "epoch": 1.69, "learning_rate": 8.242537452542687e-06, "loss": 0.37, "step": 9684 }, { "epoch": 1.69, "learning_rate": 8.238997119583992e-06, "loss": 0.3503, "step": 9685 }, { "epoch": 1.69, "learning_rate": 8.235458570510003e-06, "loss": 0.3492, "step": 9686 }, { "epoch": 1.69, "learning_rate": 8.23192180564846e-06, "loss": 0.3573, "step": 9687 }, { "epoch": 1.69, "learning_rate": 8.228386825326912e-06, "loss": 0.3459, "step": 9688 }, { "epoch": 1.69, "learning_rate": 8.22485362987277e-06, "loss": 0.3486, "step": 9689 }, { "epoch": 1.69, "learning_rate": 8.221322219613253e-06, "loss": 0.35, "step": 9690 }, { "epoch": 1.69, "learning_rate": 8.217792594875438e-06, "loss": 0.3496, "step": 9691 }, { "epoch": 1.69, "learning_rate": 8.21426475598622e-06, "loss": 0.3601, "step": 9692 }, { "epoch": 1.69, "learning_rate": 8.210738703272333e-06, "loss": 0.3516, "step": 9693 }, { "epoch": 1.69, "learning_rate": 8.20721443706035e-06, "loss": 0.3536, "step": 9694 }, { "epoch": 1.69, "learning_rate": 8.203691957676683e-06, "loss": 0.3455, "step": 9695 }, { "epoch": 1.69, "learning_rate": 8.200171265447565e-06, "loss": 0.3426, "step": 9696 }, { "epoch": 1.69, "learning_rate": 8.196652360699065e-06, "loss": 0.3391, "step": 9697 }, { "epoch": 1.69, "learning_rate": 8.193135243757097e-06, "loss": 0.3457, "step": 9698 }, { "epoch": 1.69, "learning_rate": 8.189619914947399e-06, "loss": 0.3432, "step": 9699 }, { "epoch": 1.69, "learning_rate": 8.186106374595553e-06, "loss": 0.3479, "step": 9700 }, { "epoch": 1.69, "learning_rate": 8.182594623026974e-06, "loss": 0.351, "step": 9701 }, { "epoch": 1.69, "learning_rate": 8.179084660566903e-06, "loss": 0.3605, "step": 9702 }, { "epoch": 1.69, "learning_rate": 8.175576487540416e-06, "loss": 0.3505, "step": 9703 }, { "epoch": 1.69, "learning_rate": 8.172070104272433e-06, "loss": 0.3511, "step": 9704 }, { "epoch": 1.69, "learning_rate": 8.168565511087701e-06, "loss": 0.3482, "step": 9705 }, { "epoch": 1.69, "learning_rate": 8.1650627083108e-06, "loss": 0.3478, "step": 9706 }, { "epoch": 1.69, "learning_rate": 8.161561696266153e-06, "loss": 0.346, "step": 9707 }, { "epoch": 1.69, "learning_rate": 8.158062475278007e-06, "loss": 0.3687, "step": 9708 }, { "epoch": 1.69, "learning_rate": 8.15456504567045e-06, "loss": 0.3752, "step": 9709 }, { "epoch": 1.69, "learning_rate": 8.151069407767396e-06, "loss": 0.3658, "step": 9710 }, { "epoch": 1.69, "learning_rate": 8.1475755618926e-06, "loss": 0.3449, "step": 9711 }, { "epoch": 1.69, "learning_rate": 8.144083508369657e-06, "loss": 0.366, "step": 9712 }, { "epoch": 1.69, "learning_rate": 8.140593247521983e-06, "loss": 0.3558, "step": 9713 }, { "epoch": 1.69, "learning_rate": 8.137104779672828e-06, "loss": 0.3574, "step": 9714 }, { "epoch": 1.69, "learning_rate": 8.133618105145288e-06, "loss": 0.3527, "step": 9715 }, { "epoch": 1.69, "learning_rate": 8.130133224262285e-06, "loss": 0.355, "step": 9716 }, { "epoch": 1.69, "learning_rate": 8.126650137346573e-06, "loss": 0.3551, "step": 9717 }, { "epoch": 1.69, "learning_rate": 8.12316884472075e-06, "loss": 0.3678, "step": 9718 }, { "epoch": 1.69, "learning_rate": 8.119689346707237e-06, "loss": 0.3614, "step": 9719 }, { "epoch": 1.69, "learning_rate": 8.116211643628294e-06, "loss": 0.331, "step": 9720 }, { "epoch": 1.69, "learning_rate": 8.112735735805998e-06, "loss": 0.3534, "step": 9721 }, { "epoch": 1.7, "learning_rate": 8.109261623562298e-06, "loss": 0.3726, "step": 9722 }, { "epoch": 1.7, "learning_rate": 8.10578930721895e-06, "loss": 0.3567, "step": 9723 }, { "epoch": 1.7, "learning_rate": 8.10231878709753e-06, "loss": 0.337, "step": 9724 }, { "epoch": 1.7, "learning_rate": 8.098850063519485e-06, "loss": 0.347, "step": 9725 }, { "epoch": 1.7, "learning_rate": 8.095383136806067e-06, "loss": 0.3531, "step": 9726 }, { "epoch": 1.7, "learning_rate": 8.091918007278367e-06, "loss": 0.3532, "step": 9727 }, { "epoch": 1.7, "learning_rate": 8.088454675257313e-06, "loss": 0.3595, "step": 9728 }, { "epoch": 1.7, "learning_rate": 8.08499314106368e-06, "loss": 0.3435, "step": 9729 }, { "epoch": 1.7, "learning_rate": 8.081533405018049e-06, "loss": 0.3601, "step": 9730 }, { "epoch": 1.7, "learning_rate": 8.078075467440849e-06, "loss": 0.3533, "step": 9731 }, { "epoch": 1.7, "learning_rate": 8.074619328652345e-06, "loss": 0.3526, "step": 9732 }, { "epoch": 1.7, "learning_rate": 8.071164988972631e-06, "loss": 0.3495, "step": 9733 }, { "epoch": 1.7, "learning_rate": 8.06771244872164e-06, "loss": 0.3421, "step": 9734 }, { "epoch": 1.7, "learning_rate": 8.064261708219126e-06, "loss": 0.347, "step": 9735 }, { "epoch": 1.7, "learning_rate": 8.060812767784694e-06, "loss": 0.3525, "step": 9736 }, { "epoch": 1.7, "learning_rate": 8.057365627737762e-06, "loss": 0.3601, "step": 9737 }, { "epoch": 1.7, "learning_rate": 8.05392028839759e-06, "loss": 0.3315, "step": 9738 }, { "epoch": 1.7, "learning_rate": 8.05047675008329e-06, "loss": 0.3406, "step": 9739 }, { "epoch": 1.7, "learning_rate": 8.04703501311378e-06, "loss": 0.3509, "step": 9740 }, { "epoch": 1.7, "learning_rate": 8.043595077807813e-06, "loss": 0.3624, "step": 9741 }, { "epoch": 1.7, "learning_rate": 8.040156944483997e-06, "loss": 0.355, "step": 9742 }, { "epoch": 1.7, "learning_rate": 8.036720613460754e-06, "loss": 0.3618, "step": 9743 }, { "epoch": 1.7, "learning_rate": 8.033286085056341e-06, "loss": 0.3422, "step": 9744 }, { "epoch": 1.7, "learning_rate": 8.029853359588853e-06, "loss": 0.3563, "step": 9745 }, { "epoch": 1.7, "learning_rate": 8.026422437376223e-06, "loss": 0.3336, "step": 9746 }, { "epoch": 1.7, "learning_rate": 8.022993318736206e-06, "loss": 0.3423, "step": 9747 }, { "epoch": 1.7, "learning_rate": 8.019566003986392e-06, "loss": 0.3737, "step": 9748 }, { "epoch": 1.7, "learning_rate": 8.016140493444209e-06, "loss": 0.3305, "step": 9749 }, { "epoch": 1.7, "learning_rate": 8.01271678742692e-06, "loss": 0.3744, "step": 9750 }, { "epoch": 1.7, "learning_rate": 8.009294886251611e-06, "loss": 0.3634, "step": 9751 }, { "epoch": 1.7, "learning_rate": 8.005874790235205e-06, "loss": 0.3332, "step": 9752 }, { "epoch": 1.7, "learning_rate": 8.002456499694463e-06, "loss": 0.3329, "step": 9753 }, { "epoch": 1.7, "learning_rate": 7.999040014945978e-06, "loss": 0.3376, "step": 9754 }, { "epoch": 1.7, "learning_rate": 7.995625336306155e-06, "loss": 0.3561, "step": 9755 }, { "epoch": 1.7, "learning_rate": 7.99221246409127e-06, "loss": 0.3515, "step": 9756 }, { "epoch": 1.7, "learning_rate": 7.988801398617405e-06, "loss": 0.3617, "step": 9757 }, { "epoch": 1.7, "learning_rate": 7.985392140200473e-06, "loss": 0.3418, "step": 9758 }, { "epoch": 1.7, "learning_rate": 7.981984689156237e-06, "loss": 0.3551, "step": 9759 }, { "epoch": 1.7, "learning_rate": 7.978579045800275e-06, "loss": 0.3693, "step": 9760 }, { "epoch": 1.7, "learning_rate": 7.975175210448012e-06, "loss": 0.343, "step": 9761 }, { "epoch": 1.7, "learning_rate": 7.971773183414692e-06, "loss": 0.3446, "step": 9762 }, { "epoch": 1.7, "learning_rate": 7.968372965015404e-06, "loss": 0.3667, "step": 9763 }, { "epoch": 1.7, "learning_rate": 7.964974555565066e-06, "loss": 0.3602, "step": 9764 }, { "epoch": 1.7, "learning_rate": 7.961577955378418e-06, "loss": 0.3551, "step": 9765 }, { "epoch": 1.7, "learning_rate": 7.958183164770043e-06, "loss": 0.3525, "step": 9766 }, { "epoch": 1.7, "learning_rate": 7.954790184054362e-06, "loss": 0.3794, "step": 9767 }, { "epoch": 1.7, "learning_rate": 7.95139901354562e-06, "loss": 0.3631, "step": 9768 }, { "epoch": 1.7, "learning_rate": 7.94800965355788e-06, "loss": 0.356, "step": 9769 }, { "epoch": 1.7, "learning_rate": 7.944622104405071e-06, "loss": 0.3473, "step": 9770 }, { "epoch": 1.7, "learning_rate": 7.94123636640092e-06, "loss": 0.3446, "step": 9771 }, { "epoch": 1.7, "learning_rate": 7.937852439859017e-06, "loss": 0.3573, "step": 9772 }, { "epoch": 1.7, "learning_rate": 7.934470325092754e-06, "loss": 0.3509, "step": 9773 }, { "epoch": 1.7, "learning_rate": 7.931090022415386e-06, "loss": 0.3324, "step": 9774 }, { "epoch": 1.7, "learning_rate": 7.927711532139974e-06, "loss": 0.3524, "step": 9775 }, { "epoch": 1.7, "learning_rate": 7.924334854579419e-06, "loss": 0.3559, "step": 9776 }, { "epoch": 1.7, "learning_rate": 7.920959990046461e-06, "loss": 0.3556, "step": 9777 }, { "epoch": 1.7, "learning_rate": 7.917586938853674e-06, "loss": 0.3495, "step": 9778 }, { "epoch": 1.7, "learning_rate": 7.914215701313447e-06, "loss": 0.3481, "step": 9779 }, { "epoch": 1.71, "learning_rate": 7.910846277738022e-06, "loss": 0.3561, "step": 9780 }, { "epoch": 1.71, "learning_rate": 7.907478668439455e-06, "loss": 0.3688, "step": 9781 }, { "epoch": 1.71, "learning_rate": 7.904112873729644e-06, "loss": 0.3606, "step": 9782 }, { "epoch": 1.71, "learning_rate": 7.900748893920317e-06, "loss": 0.3601, "step": 9783 }, { "epoch": 1.71, "learning_rate": 7.897386729323038e-06, "loss": 0.3661, "step": 9784 }, { "epoch": 1.71, "learning_rate": 7.894026380249193e-06, "loss": 0.3405, "step": 9785 }, { "epoch": 1.71, "learning_rate": 7.890667847010005e-06, "loss": 0.3444, "step": 9786 }, { "epoch": 1.71, "learning_rate": 7.887311129916534e-06, "loss": 0.3528, "step": 9787 }, { "epoch": 1.71, "learning_rate": 7.883956229279661e-06, "loss": 0.3498, "step": 9788 }, { "epoch": 1.71, "learning_rate": 7.880603145410114e-06, "loss": 0.3473, "step": 9789 }, { "epoch": 1.71, "learning_rate": 7.877251878618432e-06, "loss": 0.3526, "step": 9790 }, { "epoch": 1.71, "learning_rate": 7.873902429215008e-06, "loss": 0.339, "step": 9791 }, { "epoch": 1.71, "learning_rate": 7.870554797510051e-06, "loss": 0.3505, "step": 9792 }, { "epoch": 1.71, "learning_rate": 7.867208983813598e-06, "loss": 0.3462, "step": 9793 }, { "epoch": 1.71, "learning_rate": 7.863864988435542e-06, "loss": 0.3506, "step": 9794 }, { "epoch": 1.71, "learning_rate": 7.860522811685589e-06, "loss": 0.3314, "step": 9795 }, { "epoch": 1.71, "learning_rate": 7.857182453873268e-06, "loss": 0.352, "step": 9796 }, { "epoch": 1.71, "learning_rate": 7.853843915307962e-06, "loss": 0.3645, "step": 9797 }, { "epoch": 1.71, "learning_rate": 7.850507196298869e-06, "loss": 0.3716, "step": 9798 }, { "epoch": 1.71, "learning_rate": 7.847172297155023e-06, "loss": 0.348, "step": 9799 }, { "epoch": 1.71, "learning_rate": 7.84383921818529e-06, "loss": 0.3508, "step": 9800 }, { "epoch": 1.71, "learning_rate": 7.840507959698378e-06, "loss": 0.3472, "step": 9801 }, { "epoch": 1.71, "learning_rate": 7.837178522002807e-06, "loss": 0.3646, "step": 9802 }, { "epoch": 1.71, "learning_rate": 7.833850905406936e-06, "loss": 0.3399, "step": 9803 }, { "epoch": 1.71, "learning_rate": 7.830525110218959e-06, "loss": 0.3617, "step": 9804 }, { "epoch": 1.71, "learning_rate": 7.827201136746903e-06, "loss": 0.3506, "step": 9805 }, { "epoch": 1.71, "learning_rate": 7.823878985298622e-06, "loss": 0.3616, "step": 9806 }, { "epoch": 1.71, "learning_rate": 7.820558656181792e-06, "loss": 0.3388, "step": 9807 }, { "epoch": 1.71, "learning_rate": 7.817240149703946e-06, "loss": 0.3576, "step": 9808 }, { "epoch": 1.71, "learning_rate": 7.81392346617242e-06, "loss": 0.3728, "step": 9809 }, { "epoch": 1.71, "learning_rate": 7.810608605894393e-06, "loss": 0.3458, "step": 9810 }, { "epoch": 1.71, "learning_rate": 7.80729556917688e-06, "loss": 0.3456, "step": 9811 }, { "epoch": 1.71, "learning_rate": 7.803984356326723e-06, "loss": 0.3535, "step": 9812 }, { "epoch": 1.71, "learning_rate": 7.800674967650595e-06, "loss": 0.3454, "step": 9813 }, { "epoch": 1.71, "learning_rate": 7.797367403454994e-06, "loss": 0.3754, "step": 9814 }, { "epoch": 1.71, "learning_rate": 7.794061664046257e-06, "loss": 0.3408, "step": 9815 }, { "epoch": 1.71, "learning_rate": 7.790757749730555e-06, "loss": 0.3559, "step": 9816 }, { "epoch": 1.71, "learning_rate": 7.787455660813878e-06, "loss": 0.3504, "step": 9817 }, { "epoch": 1.71, "learning_rate": 7.78415539760206e-06, "loss": 0.3532, "step": 9818 }, { "epoch": 1.71, "learning_rate": 7.780856960400757e-06, "loss": 0.34, "step": 9819 }, { "epoch": 1.71, "learning_rate": 7.777560349515453e-06, "loss": 0.349, "step": 9820 }, { "epoch": 1.71, "learning_rate": 7.774265565251471e-06, "loss": 0.3413, "step": 9821 }, { "epoch": 1.71, "learning_rate": 7.77097260791397e-06, "loss": 0.3505, "step": 9822 }, { "epoch": 1.71, "learning_rate": 7.767681477807923e-06, "loss": 0.3476, "step": 9823 }, { "epoch": 1.71, "learning_rate": 7.764392175238138e-06, "loss": 0.3467, "step": 9824 }, { "epoch": 1.71, "learning_rate": 7.761104700509275e-06, "loss": 0.3569, "step": 9825 }, { "epoch": 1.71, "learning_rate": 7.757819053925791e-06, "loss": 0.341, "step": 9826 }, { "epoch": 1.71, "learning_rate": 7.754535235792005e-06, "loss": 0.3553, "step": 9827 }, { "epoch": 1.71, "learning_rate": 7.751253246412042e-06, "loss": 0.3493, "step": 9828 }, { "epoch": 1.71, "learning_rate": 7.747973086089873e-06, "loss": 0.3556, "step": 9829 }, { "epoch": 1.71, "learning_rate": 7.744694755129294e-06, "loss": 0.3676, "step": 9830 }, { "epoch": 1.71, "learning_rate": 7.741418253833927e-06, "loss": 0.3477, "step": 9831 }, { "epoch": 1.71, "learning_rate": 7.738143582507233e-06, "loss": 0.3651, "step": 9832 }, { "epoch": 1.71, "learning_rate": 7.73487074145251e-06, "loss": 0.3639, "step": 9833 }, { "epoch": 1.71, "learning_rate": 7.73159973097286e-06, "loss": 0.3613, "step": 9834 }, { "epoch": 1.71, "learning_rate": 7.728330551371246e-06, "loss": 0.3412, "step": 9835 }, { "epoch": 1.71, "learning_rate": 7.725063202950443e-06, "loss": 0.351, "step": 9836 }, { "epoch": 1.72, "learning_rate": 7.721797686013054e-06, "loss": 0.3424, "step": 9837 }, { "epoch": 1.72, "learning_rate": 7.718534000861526e-06, "loss": 0.3548, "step": 9838 }, { "epoch": 1.72, "learning_rate": 7.715272147798132e-06, "loss": 0.3565, "step": 9839 }, { "epoch": 1.72, "learning_rate": 7.712012127124974e-06, "loss": 0.3542, "step": 9840 }, { "epoch": 1.72, "learning_rate": 7.708753939143972e-06, "loss": 0.3485, "step": 9841 }, { "epoch": 1.72, "learning_rate": 7.705497584156902e-06, "loss": 0.359, "step": 9842 }, { "epoch": 1.72, "learning_rate": 7.702243062465344e-06, "loss": 0.347, "step": 9843 }, { "epoch": 1.72, "learning_rate": 7.69899037437073e-06, "loss": 0.3661, "step": 9844 }, { "epoch": 1.72, "learning_rate": 7.695739520174303e-06, "loss": 0.3376, "step": 9845 }, { "epoch": 1.72, "learning_rate": 7.692490500177156e-06, "loss": 0.3511, "step": 9846 }, { "epoch": 1.72, "learning_rate": 7.689243314680192e-06, "loss": 0.3606, "step": 9847 }, { "epoch": 1.72, "learning_rate": 7.685997963984156e-06, "loss": 0.3404, "step": 9848 }, { "epoch": 1.72, "learning_rate": 7.68275444838962e-06, "loss": 0.334, "step": 9849 }, { "epoch": 1.72, "learning_rate": 7.679512768196993e-06, "loss": 0.3643, "step": 9850 }, { "epoch": 1.72, "learning_rate": 7.676272923706498e-06, "loss": 0.3479, "step": 9851 }, { "epoch": 1.72, "learning_rate": 7.67303491521821e-06, "loss": 0.3356, "step": 9852 }, { "epoch": 1.72, "learning_rate": 7.669798743032013e-06, "loss": 0.3532, "step": 9853 }, { "epoch": 1.72, "learning_rate": 7.666564407447627e-06, "loss": 0.3591, "step": 9854 }, { "epoch": 1.72, "learning_rate": 7.663331908764608e-06, "loss": 0.3519, "step": 9855 }, { "epoch": 1.72, "learning_rate": 7.660101247282345e-06, "loss": 0.3625, "step": 9856 }, { "epoch": 1.72, "learning_rate": 7.656872423300044e-06, "loss": 0.3638, "step": 9857 }, { "epoch": 1.72, "learning_rate": 7.653645437116745e-06, "loss": 0.3333, "step": 9858 }, { "epoch": 1.72, "learning_rate": 7.650420289031325e-06, "loss": 0.354, "step": 9859 }, { "epoch": 1.72, "learning_rate": 7.647196979342478e-06, "loss": 0.3548, "step": 9860 }, { "epoch": 1.72, "learning_rate": 7.643975508348748e-06, "loss": 0.3517, "step": 9861 }, { "epoch": 1.72, "learning_rate": 7.64075587634848e-06, "loss": 0.3552, "step": 9862 }, { "epoch": 1.72, "learning_rate": 7.637538083639879e-06, "loss": 0.3459, "step": 9863 }, { "epoch": 1.72, "learning_rate": 7.634322130520958e-06, "loss": 0.3463, "step": 9864 }, { "epoch": 1.72, "learning_rate": 7.631108017289563e-06, "loss": 0.3503, "step": 9865 }, { "epoch": 1.72, "learning_rate": 7.627895744243384e-06, "loss": 0.35, "step": 9866 }, { "epoch": 1.72, "learning_rate": 7.624685311679924e-06, "loss": 0.3314, "step": 9867 }, { "epoch": 1.72, "learning_rate": 7.621476719896526e-06, "loss": 0.3437, "step": 9868 }, { "epoch": 1.72, "learning_rate": 7.6182699691903505e-06, "loss": 0.3388, "step": 9869 }, { "epoch": 1.72, "learning_rate": 7.615065059858403e-06, "loss": 0.3442, "step": 9870 }, { "epoch": 1.72, "learning_rate": 7.611861992197503e-06, "loss": 0.3503, "step": 9871 }, { "epoch": 1.72, "learning_rate": 7.608660766504312e-06, "loss": 0.3433, "step": 9872 }, { "epoch": 1.72, "learning_rate": 7.605461383075321e-06, "loss": 0.3466, "step": 9873 }, { "epoch": 1.72, "learning_rate": 7.602263842206839e-06, "loss": 0.3445, "step": 9874 }, { "epoch": 1.72, "learning_rate": 7.5990681441950084e-06, "loss": 0.3533, "step": 9875 }, { "epoch": 1.72, "learning_rate": 7.595874289335803e-06, "loss": 0.3503, "step": 9876 }, { "epoch": 1.72, "learning_rate": 7.592682277925036e-06, "loss": 0.3264, "step": 9877 }, { "epoch": 1.72, "learning_rate": 7.5894921102583335e-06, "loss": 0.3524, "step": 9878 }, { "epoch": 1.72, "learning_rate": 7.5863037866311536e-06, "loss": 0.3547, "step": 9879 }, { "epoch": 1.72, "learning_rate": 7.583117307338797e-06, "loss": 0.3506, "step": 9880 }, { "epoch": 1.72, "learning_rate": 7.579932672676375e-06, "loss": 0.3579, "step": 9881 }, { "epoch": 1.72, "learning_rate": 7.576749882938837e-06, "loss": 0.348, "step": 9882 }, { "epoch": 1.72, "learning_rate": 7.573568938420963e-06, "loss": 0.3544, "step": 9883 }, { "epoch": 1.72, "learning_rate": 7.570389839417369e-06, "loss": 0.3569, "step": 9884 }, { "epoch": 1.72, "learning_rate": 7.567212586222484e-06, "loss": 0.3412, "step": 9885 }, { "epoch": 1.72, "learning_rate": 7.56403717913057e-06, "loss": 0.3412, "step": 9886 }, { "epoch": 1.72, "learning_rate": 7.560863618435728e-06, "loss": 0.3565, "step": 9887 }, { "epoch": 1.72, "learning_rate": 7.5576919044318826e-06, "loss": 0.3481, "step": 9888 }, { "epoch": 1.72, "learning_rate": 7.554522037412778e-06, "loss": 0.353, "step": 9889 }, { "epoch": 1.72, "learning_rate": 7.55135401767201e-06, "loss": 0.3581, "step": 9890 }, { "epoch": 1.72, "learning_rate": 7.548187845502981e-06, "loss": 0.354, "step": 9891 }, { "epoch": 1.72, "learning_rate": 7.5450235211989255e-06, "loss": 0.3559, "step": 9892 }, { "epoch": 1.72, "learning_rate": 7.541861045052916e-06, "loss": 0.3519, "step": 9893 }, { "epoch": 1.72, "learning_rate": 7.538700417357857e-06, "loss": 0.3589, "step": 9894 }, { "epoch": 1.73, "learning_rate": 7.5355416384064654e-06, "loss": 0.3431, "step": 9895 }, { "epoch": 1.73, "learning_rate": 7.532384708491295e-06, "loss": 0.3503, "step": 9896 }, { "epoch": 1.73, "learning_rate": 7.529229627904738e-06, "loss": 0.3627, "step": 9897 }, { "epoch": 1.73, "learning_rate": 7.526076396938997e-06, "loss": 0.344, "step": 9898 }, { "epoch": 1.73, "learning_rate": 7.522925015886122e-06, "loss": 0.3667, "step": 9899 }, { "epoch": 1.73, "learning_rate": 7.519775485037972e-06, "loss": 0.3611, "step": 9900 }, { "epoch": 1.73, "learning_rate": 7.516627804686257e-06, "loss": 0.3466, "step": 9901 }, { "epoch": 1.73, "learning_rate": 7.513481975122497e-06, "loss": 0.3436, "step": 9902 }, { "epoch": 1.73, "learning_rate": 7.510337996638045e-06, "loss": 0.3396, "step": 9903 }, { "epoch": 1.73, "learning_rate": 7.507195869524087e-06, "loss": 0.3472, "step": 9904 }, { "epoch": 1.73, "learning_rate": 7.504055594071642e-06, "loss": 0.3463, "step": 9905 }, { "epoch": 1.73, "learning_rate": 7.500917170571545e-06, "loss": 0.3589, "step": 9906 }, { "epoch": 1.73, "learning_rate": 7.497780599314462e-06, "loss": 0.3447, "step": 9907 }, { "epoch": 1.73, "learning_rate": 7.4946458805909e-06, "loss": 0.3633, "step": 9908 }, { "epoch": 1.73, "learning_rate": 7.491513014691179e-06, "loss": 0.3433, "step": 9909 }, { "epoch": 1.73, "learning_rate": 7.4883820019054515e-06, "loss": 0.3532, "step": 9910 }, { "epoch": 1.73, "learning_rate": 7.485252842523709e-06, "loss": 0.3433, "step": 9911 }, { "epoch": 1.73, "learning_rate": 7.48212553683576e-06, "loss": 0.3374, "step": 9912 }, { "epoch": 1.73, "learning_rate": 7.479000085131238e-06, "loss": 0.3527, "step": 9913 }, { "epoch": 1.73, "learning_rate": 7.475876487699619e-06, "loss": 0.356, "step": 9914 }, { "epoch": 1.73, "learning_rate": 7.472754744830195e-06, "loss": 0.3637, "step": 9915 }, { "epoch": 1.73, "learning_rate": 7.469634856812091e-06, "loss": 0.3546, "step": 9916 }, { "epoch": 1.73, "learning_rate": 7.4665168239342605e-06, "loss": 0.3488, "step": 9917 }, { "epoch": 1.73, "learning_rate": 7.463400646485487e-06, "loss": 0.3235, "step": 9918 }, { "epoch": 1.73, "learning_rate": 7.460286324754379e-06, "loss": 0.3588, "step": 9919 }, { "epoch": 1.73, "learning_rate": 7.457173859029364e-06, "loss": 0.3615, "step": 9920 }, { "epoch": 1.73, "learning_rate": 7.454063249598717e-06, "loss": 0.336, "step": 9921 }, { "epoch": 1.73, "learning_rate": 7.450954496750534e-06, "loss": 0.3606, "step": 9922 }, { "epoch": 1.73, "learning_rate": 7.447847600772734e-06, "loss": 0.3386, "step": 9923 }, { "epoch": 1.73, "learning_rate": 7.444742561953057e-06, "loss": 0.3452, "step": 9924 }, { "epoch": 1.73, "learning_rate": 7.441639380579094e-06, "loss": 0.3629, "step": 9925 }, { "epoch": 1.73, "learning_rate": 7.438538056938239e-06, "loss": 0.349, "step": 9926 }, { "epoch": 1.73, "learning_rate": 7.4354385913177355e-06, "loss": 0.3597, "step": 9927 }, { "epoch": 1.73, "learning_rate": 7.4323409840046405e-06, "loss": 0.3364, "step": 9928 }, { "epoch": 1.73, "learning_rate": 7.429245235285845e-06, "loss": 0.3322, "step": 9929 }, { "epoch": 1.73, "learning_rate": 7.4261513454480586e-06, "loss": 0.3396, "step": 9930 }, { "epoch": 1.73, "learning_rate": 7.4230593147778335e-06, "loss": 0.3493, "step": 9931 }, { "epoch": 1.73, "learning_rate": 7.419969143561545e-06, "loss": 0.3425, "step": 9932 }, { "epoch": 1.73, "learning_rate": 7.416880832085389e-06, "loss": 0.3289, "step": 9933 }, { "epoch": 1.73, "learning_rate": 7.413794380635389e-06, "loss": 0.3785, "step": 9934 }, { "epoch": 1.73, "learning_rate": 7.410709789497411e-06, "loss": 0.3569, "step": 9935 }, { "epoch": 1.73, "learning_rate": 7.407627058957137e-06, "loss": 0.3343, "step": 9936 }, { "epoch": 1.73, "learning_rate": 7.404546189300069e-06, "loss": 0.3516, "step": 9937 }, { "epoch": 1.73, "learning_rate": 7.40146718081155e-06, "loss": 0.3564, "step": 9938 }, { "epoch": 1.73, "learning_rate": 7.398390033776756e-06, "loss": 0.3394, "step": 9939 }, { "epoch": 1.73, "learning_rate": 7.3953147484806724e-06, "loss": 0.3535, "step": 9940 }, { "epoch": 1.73, "learning_rate": 7.392241325208121e-06, "loss": 0.3596, "step": 9941 }, { "epoch": 1.73, "learning_rate": 7.389169764243752e-06, "loss": 0.3547, "step": 9942 }, { "epoch": 1.73, "learning_rate": 7.386100065872048e-06, "loss": 0.3401, "step": 9943 }, { "epoch": 1.73, "learning_rate": 7.38303223037731e-06, "loss": 0.348, "step": 9944 }, { "epoch": 1.73, "learning_rate": 7.3799662580436636e-06, "loss": 0.3625, "step": 9945 }, { "epoch": 1.73, "learning_rate": 7.3769021491550796e-06, "loss": 0.3576, "step": 9946 }, { "epoch": 1.73, "learning_rate": 7.373839903995335e-06, "loss": 0.3436, "step": 9947 }, { "epoch": 1.73, "learning_rate": 7.370779522848041e-06, "loss": 0.3438, "step": 9948 }, { "epoch": 1.73, "learning_rate": 7.367721005996654e-06, "loss": 0.3556, "step": 9949 }, { "epoch": 1.73, "learning_rate": 7.364664353724433e-06, "loss": 0.3441, "step": 9950 }, { "epoch": 1.73, "learning_rate": 7.361609566314473e-06, "loss": 0.3525, "step": 9951 }, { "epoch": 1.74, "learning_rate": 7.358556644049702e-06, "loss": 0.3584, "step": 9952 }, { "epoch": 1.74, "learning_rate": 7.355505587212869e-06, "loss": 0.3391, "step": 9953 }, { "epoch": 1.74, "learning_rate": 7.35245639608655e-06, "loss": 0.3453, "step": 9954 }, { "epoch": 1.74, "learning_rate": 7.349409070953148e-06, "loss": 0.3535, "step": 9955 }, { "epoch": 1.74, "learning_rate": 7.3463636120949035e-06, "loss": 0.3514, "step": 9956 }, { "epoch": 1.74, "learning_rate": 7.343320019793873e-06, "loss": 0.3573, "step": 9957 }, { "epoch": 1.74, "learning_rate": 7.340278294331937e-06, "loss": 0.3485, "step": 9958 }, { "epoch": 1.74, "learning_rate": 7.337238435990812e-06, "loss": 0.3698, "step": 9959 }, { "epoch": 1.74, "learning_rate": 7.3342004450520415e-06, "loss": 0.3537, "step": 9960 }, { "epoch": 1.74, "learning_rate": 7.3311643217969986e-06, "loss": 0.365, "step": 9961 }, { "epoch": 1.74, "learning_rate": 7.328130066506861e-06, "loss": 0.36, "step": 9962 }, { "epoch": 1.74, "learning_rate": 7.32509767946267e-06, "loss": 0.357, "step": 9963 }, { "epoch": 1.74, "learning_rate": 7.322067160945262e-06, "loss": 0.3612, "step": 9964 }, { "epoch": 1.74, "learning_rate": 7.3190385112353085e-06, "loss": 0.3334, "step": 9965 }, { "epoch": 1.74, "learning_rate": 7.316011730613329e-06, "loss": 0.3577, "step": 9966 }, { "epoch": 1.74, "learning_rate": 7.312986819359642e-06, "loss": 0.3528, "step": 9967 }, { "epoch": 1.74, "learning_rate": 7.309963777754403e-06, "loss": 0.3423, "step": 9968 }, { "epoch": 1.74, "learning_rate": 7.306942606077599e-06, "loss": 0.3487, "step": 9969 }, { "epoch": 1.74, "learning_rate": 7.303923304609039e-06, "loss": 0.3681, "step": 9970 }, { "epoch": 1.74, "learning_rate": 7.300905873628361e-06, "loss": 0.3629, "step": 9971 }, { "epoch": 1.74, "learning_rate": 7.297890313415023e-06, "loss": 0.3593, "step": 9972 }, { "epoch": 1.74, "learning_rate": 7.294876624248323e-06, "loss": 0.3619, "step": 9973 }, { "epoch": 1.74, "learning_rate": 7.291864806407378e-06, "loss": 0.3515, "step": 9974 }, { "epoch": 1.74, "learning_rate": 7.288854860171121e-06, "loss": 0.3492, "step": 9975 }, { "epoch": 1.74, "learning_rate": 7.285846785818329e-06, "loss": 0.3442, "step": 9976 }, { "epoch": 1.74, "learning_rate": 7.282840583627607e-06, "loss": 0.3767, "step": 9977 }, { "epoch": 1.74, "learning_rate": 7.279836253877369e-06, "loss": 0.3506, "step": 9978 }, { "epoch": 1.74, "learning_rate": 7.276833796845862e-06, "loss": 0.3487, "step": 9979 }, { "epoch": 1.74, "learning_rate": 7.273833212811175e-06, "loss": 0.3612, "step": 9980 }, { "epoch": 1.74, "learning_rate": 7.2708345020512005e-06, "loss": 0.3382, "step": 9981 }, { "epoch": 1.74, "learning_rate": 7.2678376648436714e-06, "loss": 0.3635, "step": 9982 }, { "epoch": 1.74, "learning_rate": 7.264842701466149e-06, "loss": 0.3613, "step": 9983 }, { "epoch": 1.74, "learning_rate": 7.2618496121960115e-06, "loss": 0.3548, "step": 9984 }, { "epoch": 1.74, "learning_rate": 7.25885839731047e-06, "loss": 0.3394, "step": 9985 }, { "epoch": 1.74, "learning_rate": 7.255869057086552e-06, "loss": 0.3678, "step": 9986 }, { "epoch": 1.74, "learning_rate": 7.252881591801127e-06, "loss": 0.349, "step": 9987 }, { "epoch": 1.74, "learning_rate": 7.249896001730887e-06, "loss": 0.3493, "step": 9988 }, { "epoch": 1.74, "learning_rate": 7.246912287152338e-06, "loss": 0.3565, "step": 9989 }, { "epoch": 1.74, "learning_rate": 7.2439304483418275e-06, "loss": 0.3585, "step": 9990 }, { "epoch": 1.74, "learning_rate": 7.24095048557552e-06, "loss": 0.3526, "step": 9991 }, { "epoch": 1.74, "learning_rate": 7.237972399129406e-06, "loss": 0.35, "step": 9992 }, { "epoch": 1.74, "learning_rate": 7.2349961892793055e-06, "loss": 0.3528, "step": 9993 }, { "epoch": 1.74, "learning_rate": 7.232021856300875e-06, "loss": 0.3398, "step": 9994 }, { "epoch": 1.74, "learning_rate": 7.229049400469576e-06, "loss": 0.3521, "step": 9995 }, { "epoch": 1.74, "learning_rate": 7.2260788220607015e-06, "loss": 0.3582, "step": 9996 }, { "epoch": 1.74, "learning_rate": 7.223110121349393e-06, "loss": 0.3307, "step": 9997 }, { "epoch": 1.74, "learning_rate": 7.2201432986105836e-06, "loss": 0.3606, "step": 9998 }, { "epoch": 1.74, "learning_rate": 7.21717835411906e-06, "loss": 0.3481, "step": 9999 }, { "epoch": 1.74, "learning_rate": 7.21421528814942e-06, "loss": 0.3458, "step": 10000 }, { "epoch": 1.74, "learning_rate": 7.211254100976098e-06, "loss": 0.344, "step": 10001 }, { "epoch": 1.74, "learning_rate": 7.208294792873345e-06, "loss": 0.3496, "step": 10002 }, { "epoch": 1.74, "learning_rate": 7.205337364115232e-06, "loss": 0.3362, "step": 10003 }, { "epoch": 1.74, "learning_rate": 7.202381814975681e-06, "loss": 0.3537, "step": 10004 }, { "epoch": 1.74, "learning_rate": 7.1994281457284195e-06, "loss": 0.3583, "step": 10005 }, { "epoch": 1.74, "learning_rate": 7.196476356646998e-06, "loss": 0.3383, "step": 10006 }, { "epoch": 1.74, "learning_rate": 7.193526448004815e-06, "loss": 0.3368, "step": 10007 }, { "epoch": 1.74, "learning_rate": 7.190578420075069e-06, "loss": 0.3514, "step": 10008 }, { "epoch": 1.75, "learning_rate": 7.187632273130791e-06, "loss": 0.3497, "step": 10009 }, { "epoch": 1.75, "learning_rate": 7.184688007444856e-06, "loss": 0.3484, "step": 10010 }, { "epoch": 1.75, "learning_rate": 7.1817456232899455e-06, "loss": 0.3492, "step": 10011 }, { "epoch": 1.75, "learning_rate": 7.178805120938574e-06, "loss": 0.3564, "step": 10012 }, { "epoch": 1.75, "learning_rate": 7.175866500663075e-06, "loss": 0.3636, "step": 10013 }, { "epoch": 1.75, "learning_rate": 7.172929762735614e-06, "loss": 0.3451, "step": 10014 }, { "epoch": 1.75, "learning_rate": 7.169994907428189e-06, "loss": 0.3435, "step": 10015 }, { "epoch": 1.75, "learning_rate": 7.1670619350126115e-06, "loss": 0.3507, "step": 10016 }, { "epoch": 1.75, "learning_rate": 7.164130845760518e-06, "loss": 0.3456, "step": 10017 }, { "epoch": 1.75, "learning_rate": 7.161201639943382e-06, "loss": 0.3642, "step": 10018 }, { "epoch": 1.75, "learning_rate": 7.1582743178324916e-06, "loss": 0.3534, "step": 10019 }, { "epoch": 1.75, "learning_rate": 7.1553488796989605e-06, "loss": 0.3447, "step": 10020 }, { "epoch": 1.75, "learning_rate": 7.152425325813747e-06, "loss": 0.3535, "step": 10021 }, { "epoch": 1.75, "learning_rate": 7.149503656447609e-06, "loss": 0.3433, "step": 10022 }, { "epoch": 1.75, "learning_rate": 7.14658387187114e-06, "loss": 0.3523, "step": 10023 }, { "epoch": 1.75, "learning_rate": 7.143665972354767e-06, "loss": 0.3588, "step": 10024 }, { "epoch": 1.75, "learning_rate": 7.140749958168728e-06, "loss": 0.3466, "step": 10025 }, { "epoch": 1.75, "learning_rate": 7.137835829583101e-06, "loss": 0.362, "step": 10026 }, { "epoch": 1.75, "learning_rate": 7.134923586867772e-06, "loss": 0.3502, "step": 10027 }, { "epoch": 1.75, "learning_rate": 7.132013230292475e-06, "loss": 0.3383, "step": 10028 }, { "epoch": 1.75, "learning_rate": 7.12910476012675e-06, "loss": 0.3501, "step": 10029 }, { "epoch": 1.75, "learning_rate": 7.126198176639969e-06, "loss": 0.3508, "step": 10030 }, { "epoch": 1.75, "learning_rate": 7.123293480101326e-06, "loss": 0.3391, "step": 10031 }, { "epoch": 1.75, "learning_rate": 7.120390670779851e-06, "loss": 0.3555, "step": 10032 }, { "epoch": 1.75, "learning_rate": 7.117489748944389e-06, "loss": 0.3433, "step": 10033 }, { "epoch": 1.75, "learning_rate": 7.114590714863607e-06, "loss": 0.3716, "step": 10034 }, { "epoch": 1.75, "learning_rate": 7.111693568806016e-06, "loss": 0.3393, "step": 10035 }, { "epoch": 1.75, "learning_rate": 7.108798311039923e-06, "loss": 0.3577, "step": 10036 }, { "epoch": 1.75, "learning_rate": 7.1059049418334915e-06, "loss": 0.3398, "step": 10037 }, { "epoch": 1.75, "learning_rate": 7.103013461454683e-06, "loss": 0.3513, "step": 10038 }, { "epoch": 1.75, "learning_rate": 7.100123870171304e-06, "loss": 0.3601, "step": 10039 }, { "epoch": 1.75, "learning_rate": 7.097236168250977e-06, "loss": 0.3335, "step": 10040 }, { "epoch": 1.75, "learning_rate": 7.094350355961145e-06, "loss": 0.3331, "step": 10041 }, { "epoch": 1.75, "learning_rate": 7.0914664335690845e-06, "loss": 0.3436, "step": 10042 }, { "epoch": 1.75, "learning_rate": 7.088584401341897e-06, "loss": 0.3597, "step": 10043 }, { "epoch": 1.75, "learning_rate": 7.085704259546501e-06, "loss": 0.3513, "step": 10044 }, { "epoch": 1.75, "learning_rate": 7.082826008449653e-06, "loss": 0.3574, "step": 10045 }, { "epoch": 1.75, "learning_rate": 7.07994964831792e-06, "loss": 0.3454, "step": 10046 }, { "epoch": 1.75, "learning_rate": 7.077075179417697e-06, "loss": 0.3594, "step": 10047 }, { "epoch": 1.75, "learning_rate": 7.074202602015209e-06, "loss": 0.3543, "step": 10048 }, { "epoch": 1.75, "learning_rate": 7.071331916376513e-06, "loss": 0.3336, "step": 10049 }, { "epoch": 1.75, "learning_rate": 7.06846312276747e-06, "loss": 0.3588, "step": 10050 }, { "epoch": 1.75, "learning_rate": 7.06559622145378e-06, "loss": 0.3425, "step": 10051 }, { "epoch": 1.75, "learning_rate": 7.062731212700971e-06, "loss": 0.3577, "step": 10052 }, { "epoch": 1.75, "learning_rate": 7.059868096774383e-06, "loss": 0.369, "step": 10053 }, { "epoch": 1.75, "learning_rate": 7.0570068739391895e-06, "loss": 0.3563, "step": 10054 }, { "epoch": 1.75, "learning_rate": 7.0541475444603874e-06, "loss": 0.3591, "step": 10055 }, { "epoch": 1.75, "learning_rate": 7.0512901086027985e-06, "loss": 0.342, "step": 10056 }, { "epoch": 1.75, "learning_rate": 7.048434566631068e-06, "loss": 0.3516, "step": 10057 }, { "epoch": 1.75, "learning_rate": 7.0455809188096605e-06, "loss": 0.3825, "step": 10058 }, { "epoch": 1.75, "learning_rate": 7.0427291654028744e-06, "loss": 0.3597, "step": 10059 }, { "epoch": 1.75, "learning_rate": 7.039879306674835e-06, "loss": 0.3501, "step": 10060 }, { "epoch": 1.75, "learning_rate": 7.037031342889472e-06, "loss": 0.3629, "step": 10061 }, { "epoch": 1.75, "learning_rate": 7.03418527431057e-06, "loss": 0.3551, "step": 10062 }, { "epoch": 1.75, "learning_rate": 7.031341101201713e-06, "loss": 0.3429, "step": 10063 }, { "epoch": 1.75, "learning_rate": 7.028498823826313e-06, "loss": 0.3702, "step": 10064 }, { "epoch": 1.75, "learning_rate": 7.025658442447617e-06, "loss": 0.3461, "step": 10065 }, { "epoch": 1.75, "learning_rate": 7.0228199573286956e-06, "loss": 0.357, "step": 10066 }, { "epoch": 1.76, "learning_rate": 7.019983368732434e-06, "loss": 0.3453, "step": 10067 }, { "epoch": 1.76, "learning_rate": 7.017148676921545e-06, "loss": 0.3691, "step": 10068 }, { "epoch": 1.76, "learning_rate": 7.01431588215857e-06, "loss": 0.3486, "step": 10069 }, { "epoch": 1.76, "learning_rate": 7.011484984705871e-06, "loss": 0.3422, "step": 10070 }, { "epoch": 1.76, "learning_rate": 7.00865598482564e-06, "loss": 0.3493, "step": 10071 }, { "epoch": 1.76, "learning_rate": 7.0058288827798805e-06, "loss": 0.3649, "step": 10072 }, { "epoch": 1.76, "learning_rate": 7.003003678830436e-06, "loss": 0.3466, "step": 10073 }, { "epoch": 1.76, "learning_rate": 7.000180373238965e-06, "loss": 0.352, "step": 10074 }, { "epoch": 1.76, "learning_rate": 6.9973589662669455e-06, "loss": 0.3574, "step": 10075 }, { "epoch": 1.76, "learning_rate": 6.994539458175697e-06, "loss": 0.3594, "step": 10076 }, { "epoch": 1.76, "learning_rate": 6.991721849226347e-06, "loss": 0.3643, "step": 10077 }, { "epoch": 1.76, "learning_rate": 6.988906139679853e-06, "loss": 0.3593, "step": 10078 }, { "epoch": 1.76, "learning_rate": 6.986092329796986e-06, "loss": 0.3562, "step": 10079 }, { "epoch": 1.76, "learning_rate": 6.98328041983837e-06, "loss": 0.3422, "step": 10080 }, { "epoch": 1.76, "learning_rate": 6.9804704100644155e-06, "loss": 0.3535, "step": 10081 }, { "epoch": 1.76, "learning_rate": 6.977662300735388e-06, "loss": 0.3379, "step": 10082 }, { "epoch": 1.76, "learning_rate": 6.974856092111359e-06, "loss": 0.3492, "step": 10083 }, { "epoch": 1.76, "learning_rate": 6.972051784452234e-06, "loss": 0.3379, "step": 10084 }, { "epoch": 1.76, "learning_rate": 6.969249378017728e-06, "loss": 0.3531, "step": 10085 }, { "epoch": 1.76, "learning_rate": 6.9664488730674e-06, "loss": 0.3574, "step": 10086 }, { "epoch": 1.76, "learning_rate": 6.963650269860619e-06, "loss": 0.3423, "step": 10087 }, { "epoch": 1.76, "learning_rate": 6.960853568656582e-06, "loss": 0.3402, "step": 10088 }, { "epoch": 1.76, "learning_rate": 6.958058769714306e-06, "loss": 0.3619, "step": 10089 }, { "epoch": 1.76, "learning_rate": 6.9552658732926434e-06, "loss": 0.3611, "step": 10090 }, { "epoch": 1.76, "learning_rate": 6.9524748796502545e-06, "loss": 0.3598, "step": 10091 }, { "epoch": 1.76, "learning_rate": 6.94968578904563e-06, "loss": 0.3348, "step": 10092 }, { "epoch": 1.76, "learning_rate": 6.94689860173709e-06, "loss": 0.3465, "step": 10093 }, { "epoch": 1.76, "learning_rate": 6.944113317982775e-06, "loss": 0.3478, "step": 10094 }, { "epoch": 1.76, "learning_rate": 6.941329938040647e-06, "loss": 0.3452, "step": 10095 }, { "epoch": 1.76, "learning_rate": 6.938548462168485e-06, "loss": 0.3614, "step": 10096 }, { "epoch": 1.76, "learning_rate": 6.93576889062391e-06, "loss": 0.347, "step": 10097 }, { "epoch": 1.76, "learning_rate": 6.93299122366435e-06, "loss": 0.3655, "step": 10098 }, { "epoch": 1.76, "learning_rate": 6.930215461547064e-06, "loss": 0.3581, "step": 10099 }, { "epoch": 1.76, "learning_rate": 6.927441604529136e-06, "loss": 0.3655, "step": 10100 }, { "epoch": 1.76, "learning_rate": 6.924669652867469e-06, "loss": 0.3524, "step": 10101 }, { "epoch": 1.76, "learning_rate": 6.921899606818781e-06, "loss": 0.3653, "step": 10102 }, { "epoch": 1.76, "learning_rate": 6.919131466639638e-06, "loss": 0.3518, "step": 10103 }, { "epoch": 1.76, "learning_rate": 6.91636523258641e-06, "loss": 0.354, "step": 10104 }, { "epoch": 1.76, "learning_rate": 6.913600904915297e-06, "loss": 0.3519, "step": 10105 }, { "epoch": 1.76, "learning_rate": 6.910838483882315e-06, "loss": 0.3635, "step": 10106 }, { "epoch": 1.76, "learning_rate": 6.908077969743318e-06, "loss": 0.3525, "step": 10107 }, { "epoch": 1.76, "learning_rate": 6.905319362753965e-06, "loss": 0.3351, "step": 10108 }, { "epoch": 1.76, "learning_rate": 6.902562663169759e-06, "loss": 0.3496, "step": 10109 }, { "epoch": 1.76, "learning_rate": 6.899807871246009e-06, "loss": 0.359, "step": 10110 }, { "epoch": 1.76, "learning_rate": 6.897054987237857e-06, "loss": 0.3549, "step": 10111 }, { "epoch": 1.76, "learning_rate": 6.8943040114002625e-06, "loss": 0.3477, "step": 10112 }, { "epoch": 1.76, "learning_rate": 6.891554943988011e-06, "loss": 0.3484, "step": 10113 }, { "epoch": 1.76, "learning_rate": 6.888807785255711e-06, "loss": 0.3544, "step": 10114 }, { "epoch": 1.76, "learning_rate": 6.8860625354578e-06, "loss": 0.3533, "step": 10115 }, { "epoch": 1.76, "learning_rate": 6.883319194848527e-06, "loss": 0.3527, "step": 10116 }, { "epoch": 1.76, "learning_rate": 6.880577763681974e-06, "loss": 0.345, "step": 10117 }, { "epoch": 1.76, "learning_rate": 6.8778382422120395e-06, "loss": 0.3588, "step": 10118 }, { "epoch": 1.76, "learning_rate": 6.875100630692448e-06, "loss": 0.3534, "step": 10119 }, { "epoch": 1.76, "learning_rate": 6.872364929376747e-06, "loss": 0.359, "step": 10120 }, { "epoch": 1.76, "learning_rate": 6.869631138518315e-06, "loss": 0.3439, "step": 10121 }, { "epoch": 1.76, "learning_rate": 6.866899258370338e-06, "loss": 0.3629, "step": 10122 }, { "epoch": 1.76, "learning_rate": 6.8641692891858316e-06, "loss": 0.3659, "step": 10123 }, { "epoch": 1.77, "learning_rate": 6.86144123121764e-06, "loss": 0.3549, "step": 10124 }, { "epoch": 1.77, "learning_rate": 6.858715084718424e-06, "loss": 0.3575, "step": 10125 }, { "epoch": 1.77, "learning_rate": 6.855990849940674e-06, "loss": 0.35, "step": 10126 }, { "epoch": 1.77, "learning_rate": 6.853268527136692e-06, "loss": 0.3477, "step": 10127 }, { "epoch": 1.77, "learning_rate": 6.8505481165586135e-06, "loss": 0.3457, "step": 10128 }, { "epoch": 1.77, "learning_rate": 6.847829618458394e-06, "loss": 0.3742, "step": 10129 }, { "epoch": 1.77, "learning_rate": 6.845113033087808e-06, "loss": 0.3478, "step": 10130 }, { "epoch": 1.77, "learning_rate": 6.842398360698456e-06, "loss": 0.3421, "step": 10131 }, { "epoch": 1.77, "learning_rate": 6.8396856015417645e-06, "loss": 0.3514, "step": 10132 }, { "epoch": 1.77, "learning_rate": 6.83697475586898e-06, "loss": 0.3546, "step": 10133 }, { "epoch": 1.77, "learning_rate": 6.834265823931165e-06, "loss": 0.3564, "step": 10134 }, { "epoch": 1.77, "learning_rate": 6.831558805979217e-06, "loss": 0.3612, "step": 10135 }, { "epoch": 1.77, "learning_rate": 6.828853702263845e-06, "loss": 0.3577, "step": 10136 }, { "epoch": 1.77, "learning_rate": 6.8261505130355884e-06, "loss": 0.3557, "step": 10137 }, { "epoch": 1.77, "learning_rate": 6.823449238544813e-06, "loss": 0.3519, "step": 10138 }, { "epoch": 1.77, "learning_rate": 6.820749879041692e-06, "loss": 0.3415, "step": 10139 }, { "epoch": 1.77, "learning_rate": 6.8180524347762346e-06, "loss": 0.347, "step": 10140 }, { "epoch": 1.77, "learning_rate": 6.8153569059982654e-06, "loss": 0.3357, "step": 10141 }, { "epoch": 1.77, "learning_rate": 6.812663292957443e-06, "loss": 0.3638, "step": 10142 }, { "epoch": 1.77, "learning_rate": 6.809971595903231e-06, "loss": 0.3392, "step": 10143 }, { "epoch": 1.77, "learning_rate": 6.807281815084923e-06, "loss": 0.3454, "step": 10144 }, { "epoch": 1.77, "learning_rate": 6.804593950751651e-06, "loss": 0.3499, "step": 10145 }, { "epoch": 1.77, "learning_rate": 6.801908003152341e-06, "loss": 0.3427, "step": 10146 }, { "epoch": 1.77, "learning_rate": 6.7992239725357595e-06, "loss": 0.3447, "step": 10147 }, { "epoch": 1.77, "learning_rate": 6.79654185915049e-06, "loss": 0.3594, "step": 10148 }, { "epoch": 1.77, "learning_rate": 6.793861663244949e-06, "loss": 0.3468, "step": 10149 }, { "epoch": 1.77, "learning_rate": 6.79118338506736e-06, "loss": 0.3359, "step": 10150 }, { "epoch": 1.77, "learning_rate": 6.788507024865773e-06, "loss": 0.3496, "step": 10151 }, { "epoch": 1.77, "learning_rate": 6.785832582888064e-06, "loss": 0.3385, "step": 10152 }, { "epoch": 1.77, "learning_rate": 6.783160059381939e-06, "loss": 0.3484, "step": 10153 }, { "epoch": 1.77, "learning_rate": 6.780489454594906e-06, "loss": 0.3439, "step": 10154 }, { "epoch": 1.77, "learning_rate": 6.777820768774315e-06, "loss": 0.3438, "step": 10155 }, { "epoch": 1.77, "learning_rate": 6.775154002167325e-06, "loss": 0.351, "step": 10156 }, { "epoch": 1.77, "learning_rate": 6.77248915502092e-06, "loss": 0.3574, "step": 10157 }, { "epoch": 1.77, "learning_rate": 6.769826227581915e-06, "loss": 0.3561, "step": 10158 }, { "epoch": 1.77, "learning_rate": 6.767165220096939e-06, "loss": 0.3537, "step": 10159 }, { "epoch": 1.77, "learning_rate": 6.764506132812448e-06, "loss": 0.3706, "step": 10160 }, { "epoch": 1.77, "learning_rate": 6.761848965974707e-06, "loss": 0.3552, "step": 10161 }, { "epoch": 1.77, "learning_rate": 6.759193719829821e-06, "loss": 0.3595, "step": 10162 }, { "epoch": 1.77, "learning_rate": 6.75654039462371e-06, "loss": 0.3597, "step": 10163 }, { "epoch": 1.77, "learning_rate": 6.753888990602108e-06, "loss": 0.3528, "step": 10164 }, { "epoch": 1.77, "learning_rate": 6.7512395080105875e-06, "loss": 0.3442, "step": 10165 }, { "epoch": 1.77, "learning_rate": 6.74859194709453e-06, "loss": 0.3414, "step": 10166 }, { "epoch": 1.77, "learning_rate": 6.745946308099145e-06, "loss": 0.3514, "step": 10167 }, { "epoch": 1.77, "learning_rate": 6.743302591269457e-06, "loss": 0.3534, "step": 10168 }, { "epoch": 1.77, "learning_rate": 6.740660796850319e-06, "loss": 0.3444, "step": 10169 }, { "epoch": 1.77, "learning_rate": 6.738020925086412e-06, "loss": 0.3572, "step": 10170 }, { "epoch": 1.77, "learning_rate": 6.735382976222223e-06, "loss": 0.3493, "step": 10171 }, { "epoch": 1.77, "learning_rate": 6.732746950502072e-06, "loss": 0.3612, "step": 10172 }, { "epoch": 1.77, "learning_rate": 6.730112848170099e-06, "loss": 0.3401, "step": 10173 }, { "epoch": 1.77, "learning_rate": 6.7274806694702655e-06, "loss": 0.3503, "step": 10174 }, { "epoch": 1.77, "learning_rate": 6.724850414646344e-06, "loss": 0.3578, "step": 10175 }, { "epoch": 1.77, "learning_rate": 6.722222083941958e-06, "loss": 0.3512, "step": 10176 }, { "epoch": 1.77, "learning_rate": 6.719595677600524e-06, "loss": 0.335, "step": 10177 }, { "epoch": 1.77, "learning_rate": 6.716971195865287e-06, "loss": 0.3501, "step": 10178 }, { "epoch": 1.77, "learning_rate": 6.714348638979324e-06, "loss": 0.3538, "step": 10179 }, { "epoch": 1.77, "learning_rate": 6.711728007185522e-06, "loss": 0.366, "step": 10180 }, { "epoch": 1.78, "learning_rate": 6.709109300726598e-06, "loss": 0.3611, "step": 10181 }, { "epoch": 1.78, "learning_rate": 6.706492519845081e-06, "loss": 0.3522, "step": 10182 }, { "epoch": 1.78, "learning_rate": 6.703877664783337e-06, "loss": 0.3477, "step": 10183 }, { "epoch": 1.78, "learning_rate": 6.701264735783539e-06, "loss": 0.3512, "step": 10184 }, { "epoch": 1.78, "learning_rate": 6.698653733087681e-06, "loss": 0.3497, "step": 10185 }, { "epoch": 1.78, "learning_rate": 6.696044656937596e-06, "loss": 0.3454, "step": 10186 }, { "epoch": 1.78, "learning_rate": 6.693437507574924e-06, "loss": 0.3352, "step": 10187 }, { "epoch": 1.78, "learning_rate": 6.690832285241129e-06, "loss": 0.3521, "step": 10188 }, { "epoch": 1.78, "learning_rate": 6.688228990177491e-06, "loss": 0.3623, "step": 10189 }, { "epoch": 1.78, "learning_rate": 6.685627622625129e-06, "loss": 0.3618, "step": 10190 }, { "epoch": 1.78, "learning_rate": 6.683028182824962e-06, "loss": 0.3527, "step": 10191 }, { "epoch": 1.78, "learning_rate": 6.680430671017746e-06, "loss": 0.3651, "step": 10192 }, { "epoch": 1.78, "learning_rate": 6.677835087444057e-06, "loss": 0.3595, "step": 10193 }, { "epoch": 1.78, "learning_rate": 6.675241432344281e-06, "loss": 0.3546, "step": 10194 }, { "epoch": 1.78, "learning_rate": 6.672649705958636e-06, "loss": 0.332, "step": 10195 }, { "epoch": 1.78, "learning_rate": 6.670059908527161e-06, "loss": 0.3273, "step": 10196 }, { "epoch": 1.78, "learning_rate": 6.667472040289708e-06, "loss": 0.3529, "step": 10197 }, { "epoch": 1.78, "learning_rate": 6.664886101485962e-06, "loss": 0.3379, "step": 10198 }, { "epoch": 1.78, "learning_rate": 6.6623020923554185e-06, "loss": 0.3538, "step": 10199 }, { "epoch": 1.78, "learning_rate": 6.659720013137402e-06, "loss": 0.3642, "step": 10200 }, { "epoch": 1.78, "learning_rate": 6.6571398640710575e-06, "loss": 0.3514, "step": 10201 }, { "epoch": 1.78, "learning_rate": 6.654561645395342e-06, "loss": 0.367, "step": 10202 }, { "epoch": 1.78, "learning_rate": 6.6519853573490455e-06, "loss": 0.3544, "step": 10203 }, { "epoch": 1.78, "learning_rate": 6.649411000170777e-06, "loss": 0.3408, "step": 10204 }, { "epoch": 1.78, "learning_rate": 6.64683857409896e-06, "loss": 0.353, "step": 10205 }, { "epoch": 1.78, "learning_rate": 6.644268079371843e-06, "loss": 0.3419, "step": 10206 }, { "epoch": 1.78, "learning_rate": 6.641699516227499e-06, "loss": 0.3449, "step": 10207 }, { "epoch": 1.78, "learning_rate": 6.639132884903818e-06, "loss": 0.3406, "step": 10208 }, { "epoch": 1.78, "learning_rate": 6.636568185638513e-06, "loss": 0.353, "step": 10209 }, { "epoch": 1.78, "learning_rate": 6.634005418669112e-06, "loss": 0.3545, "step": 10210 }, { "epoch": 1.78, "learning_rate": 6.631444584232978e-06, "loss": 0.3414, "step": 10211 }, { "epoch": 1.78, "learning_rate": 6.628885682567282e-06, "loss": 0.3448, "step": 10212 }, { "epoch": 1.78, "learning_rate": 6.626328713909009e-06, "loss": 0.3536, "step": 10213 }, { "epoch": 1.78, "learning_rate": 6.623773678494996e-06, "loss": 0.3459, "step": 10214 }, { "epoch": 1.78, "learning_rate": 6.6212205765618746e-06, "loss": 0.3684, "step": 10215 }, { "epoch": 1.78, "learning_rate": 6.618669408346096e-06, "loss": 0.3451, "step": 10216 }, { "epoch": 1.78, "learning_rate": 6.616120174083951e-06, "loss": 0.3552, "step": 10217 }, { "epoch": 1.78, "learning_rate": 6.6135728740115325e-06, "loss": 0.3516, "step": 10218 }, { "epoch": 1.78, "learning_rate": 6.611027508364761e-06, "loss": 0.3327, "step": 10219 }, { "epoch": 1.78, "learning_rate": 6.6084840773793844e-06, "loss": 0.357, "step": 10220 }, { "epoch": 1.78, "learning_rate": 6.605942581290971e-06, "loss": 0.3661, "step": 10221 }, { "epoch": 1.78, "learning_rate": 6.603403020334896e-06, "loss": 0.3595, "step": 10222 }, { "epoch": 1.78, "learning_rate": 6.600865394746361e-06, "loss": 0.3427, "step": 10223 }, { "epoch": 1.78, "learning_rate": 6.598329704760402e-06, "loss": 0.3521, "step": 10224 }, { "epoch": 1.78, "learning_rate": 6.595795950611863e-06, "loss": 0.3467, "step": 10225 }, { "epoch": 1.78, "learning_rate": 6.593264132535411e-06, "loss": 0.3384, "step": 10226 }, { "epoch": 1.78, "learning_rate": 6.590734250765529e-06, "loss": 0.3493, "step": 10227 }, { "epoch": 1.78, "learning_rate": 6.588206305536532e-06, "loss": 0.3518, "step": 10228 }, { "epoch": 1.78, "learning_rate": 6.5856802970825445e-06, "loss": 0.35, "step": 10229 }, { "epoch": 1.78, "learning_rate": 6.583156225637513e-06, "loss": 0.3498, "step": 10230 }, { "epoch": 1.78, "learning_rate": 6.5806340914352185e-06, "loss": 0.3504, "step": 10231 }, { "epoch": 1.78, "learning_rate": 6.578113894709248e-06, "loss": 0.3483, "step": 10232 }, { "epoch": 1.78, "learning_rate": 6.575595635693007e-06, "loss": 0.3503, "step": 10233 }, { "epoch": 1.78, "learning_rate": 6.573079314619736e-06, "loss": 0.3589, "step": 10234 }, { "epoch": 1.78, "learning_rate": 6.570564931722479e-06, "loss": 0.3416, "step": 10235 }, { "epoch": 1.78, "learning_rate": 6.568052487234118e-06, "loss": 0.343, "step": 10236 }, { "epoch": 1.78, "learning_rate": 6.565541981387338e-06, "loss": 0.3662, "step": 10237 }, { "epoch": 1.78, "learning_rate": 6.5630334144146616e-06, "loss": 0.3586, "step": 10238 }, { "epoch": 1.79, "learning_rate": 6.560526786548418e-06, "loss": 0.3559, "step": 10239 }, { "epoch": 1.79, "learning_rate": 6.558022098020761e-06, "loss": 0.3432, "step": 10240 }, { "epoch": 1.79, "learning_rate": 6.5555193490636655e-06, "loss": 0.3518, "step": 10241 }, { "epoch": 1.79, "learning_rate": 6.553018539908935e-06, "loss": 0.371, "step": 10242 }, { "epoch": 1.79, "learning_rate": 6.55051967078818e-06, "loss": 0.3444, "step": 10243 }, { "epoch": 1.79, "learning_rate": 6.54802274193283e-06, "loss": 0.3559, "step": 10244 }, { "epoch": 1.79, "learning_rate": 6.545527753574154e-06, "loss": 0.3409, "step": 10245 }, { "epoch": 1.79, "learning_rate": 6.5430347059432185e-06, "loss": 0.3455, "step": 10246 }, { "epoch": 1.79, "learning_rate": 6.540543599270925e-06, "loss": 0.3421, "step": 10247 }, { "epoch": 1.79, "learning_rate": 6.538054433787993e-06, "loss": 0.3459, "step": 10248 }, { "epoch": 1.79, "learning_rate": 6.535567209724961e-06, "loss": 0.3885, "step": 10249 }, { "epoch": 1.79, "learning_rate": 6.5330819273121794e-06, "loss": 0.3349, "step": 10250 }, { "epoch": 1.79, "learning_rate": 6.530598586779832e-06, "loss": 0.3562, "step": 10251 }, { "epoch": 1.79, "learning_rate": 6.528117188357914e-06, "loss": 0.3514, "step": 10252 }, { "epoch": 1.79, "learning_rate": 6.525637732276246e-06, "loss": 0.3334, "step": 10253 }, { "epoch": 1.79, "learning_rate": 6.523160218764464e-06, "loss": 0.3618, "step": 10254 }, { "epoch": 1.79, "learning_rate": 6.520684648052028e-06, "loss": 0.3469, "step": 10255 }, { "epoch": 1.79, "learning_rate": 6.518211020368217e-06, "loss": 0.3465, "step": 10256 }, { "epoch": 1.79, "learning_rate": 6.515739335942127e-06, "loss": 0.3586, "step": 10257 }, { "epoch": 1.79, "learning_rate": 6.513269595002675e-06, "loss": 0.3469, "step": 10258 }, { "epoch": 1.79, "learning_rate": 6.510801797778607e-06, "loss": 0.3501, "step": 10259 }, { "epoch": 1.79, "learning_rate": 6.508335944498477e-06, "loss": 0.3433, "step": 10260 }, { "epoch": 1.79, "learning_rate": 6.50587203539066e-06, "loss": 0.3498, "step": 10261 }, { "epoch": 1.79, "learning_rate": 6.503410070683362e-06, "loss": 0.3483, "step": 10262 }, { "epoch": 1.79, "learning_rate": 6.500950050604592e-06, "loss": 0.3281, "step": 10263 }, { "epoch": 1.79, "learning_rate": 6.498491975382198e-06, "loss": 0.351, "step": 10264 }, { "epoch": 1.79, "learning_rate": 6.4960358452438265e-06, "loss": 0.3645, "step": 10265 }, { "epoch": 1.79, "learning_rate": 6.493581660416971e-06, "loss": 0.3407, "step": 10266 }, { "epoch": 1.79, "learning_rate": 6.491129421128918e-06, "loss": 0.377, "step": 10267 }, { "epoch": 1.79, "learning_rate": 6.488679127606783e-06, "loss": 0.3451, "step": 10268 }, { "epoch": 1.79, "learning_rate": 6.4862307800775095e-06, "loss": 0.3614, "step": 10269 }, { "epoch": 1.79, "learning_rate": 6.483784378767854e-06, "loss": 0.3563, "step": 10270 }, { "epoch": 1.79, "learning_rate": 6.481339923904391e-06, "loss": 0.35, "step": 10271 }, { "epoch": 1.79, "learning_rate": 6.478897415713522e-06, "loss": 0.3427, "step": 10272 }, { "epoch": 1.79, "learning_rate": 6.476456854421459e-06, "loss": 0.3565, "step": 10273 }, { "epoch": 1.79, "learning_rate": 6.474018240254235e-06, "loss": 0.3489, "step": 10274 }, { "epoch": 1.79, "learning_rate": 6.47158157343771e-06, "loss": 0.3474, "step": 10275 }, { "epoch": 1.79, "learning_rate": 6.469146854197562e-06, "loss": 0.3518, "step": 10276 }, { "epoch": 1.79, "learning_rate": 6.466714082759283e-06, "loss": 0.346, "step": 10277 }, { "epoch": 1.79, "learning_rate": 6.4642832593481835e-06, "loss": 0.3545, "step": 10278 }, { "epoch": 1.79, "learning_rate": 6.461854384189407e-06, "loss": 0.3514, "step": 10279 }, { "epoch": 1.79, "learning_rate": 6.4594274575078965e-06, "loss": 0.3469, "step": 10280 }, { "epoch": 1.79, "learning_rate": 6.457002479528433e-06, "loss": 0.353, "step": 10281 }, { "epoch": 1.79, "learning_rate": 6.454579450475604e-06, "loss": 0.3469, "step": 10282 }, { "epoch": 1.79, "learning_rate": 6.452158370573825e-06, "loss": 0.3573, "step": 10283 }, { "epoch": 1.79, "learning_rate": 6.4497392400473325e-06, "loss": 0.3622, "step": 10284 }, { "epoch": 1.79, "learning_rate": 6.44732205912016e-06, "loss": 0.3575, "step": 10285 }, { "epoch": 1.79, "learning_rate": 6.444906828016199e-06, "loss": 0.332, "step": 10286 }, { "epoch": 1.79, "learning_rate": 6.442493546959129e-06, "loss": 0.3497, "step": 10287 }, { "epoch": 1.79, "learning_rate": 6.4400822161724596e-06, "loss": 0.3549, "step": 10288 }, { "epoch": 1.79, "learning_rate": 6.437672835879524e-06, "loss": 0.3423, "step": 10289 }, { "epoch": 1.79, "learning_rate": 6.435265406303466e-06, "loss": 0.3593, "step": 10290 }, { "epoch": 1.79, "learning_rate": 6.4328599276672535e-06, "loss": 0.3569, "step": 10291 }, { "epoch": 1.79, "learning_rate": 6.4304564001936715e-06, "loss": 0.3471, "step": 10292 }, { "epoch": 1.79, "learning_rate": 6.428054824105334e-06, "loss": 0.3588, "step": 10293 }, { "epoch": 1.79, "learning_rate": 6.425655199624661e-06, "loss": 0.3318, "step": 10294 }, { "epoch": 1.79, "learning_rate": 6.423257526973892e-06, "loss": 0.3373, "step": 10295 }, { "epoch": 1.8, "learning_rate": 6.420861806375096e-06, "loss": 0.3657, "step": 10296 }, { "epoch": 1.8, "learning_rate": 6.418468038050161e-06, "loss": 0.3612, "step": 10297 }, { "epoch": 1.8, "learning_rate": 6.416076222220785e-06, "loss": 0.3613, "step": 10298 }, { "epoch": 1.8, "learning_rate": 6.413686359108482e-06, "loss": 0.3321, "step": 10299 }, { "epoch": 1.8, "learning_rate": 6.411298448934605e-06, "loss": 0.3522, "step": 10300 }, { "epoch": 1.8, "learning_rate": 6.408912491920306e-06, "loss": 0.3554, "step": 10301 }, { "epoch": 1.8, "learning_rate": 6.406528488286562e-06, "loss": 0.34, "step": 10302 }, { "epoch": 1.8, "learning_rate": 6.4041464382541754e-06, "loss": 0.3417, "step": 10303 }, { "epoch": 1.8, "learning_rate": 6.401766342043765e-06, "loss": 0.3636, "step": 10304 }, { "epoch": 1.8, "learning_rate": 6.399388199875763e-06, "loss": 0.3461, "step": 10305 }, { "epoch": 1.8, "learning_rate": 6.3970120119704215e-06, "loss": 0.3345, "step": 10306 }, { "epoch": 1.8, "learning_rate": 6.39463777854782e-06, "loss": 0.3662, "step": 10307 }, { "epoch": 1.8, "learning_rate": 6.39226549982785e-06, "loss": 0.3653, "step": 10308 }, { "epoch": 1.8, "learning_rate": 6.3898951760302195e-06, "loss": 0.3524, "step": 10309 }, { "epoch": 1.8, "learning_rate": 6.387526807374468e-06, "loss": 0.3546, "step": 10310 }, { "epoch": 1.8, "learning_rate": 6.385160394079939e-06, "loss": 0.3438, "step": 10311 }, { "epoch": 1.8, "learning_rate": 6.3827959363658e-06, "loss": 0.3581, "step": 10312 }, { "epoch": 1.8, "learning_rate": 6.380433434451039e-06, "loss": 0.3487, "step": 10313 }, { "epoch": 1.8, "learning_rate": 6.37807288855447e-06, "loss": 0.3326, "step": 10314 }, { "epoch": 1.8, "learning_rate": 6.375714298894708e-06, "loss": 0.3473, "step": 10315 }, { "epoch": 1.8, "learning_rate": 6.373357665690202e-06, "loss": 0.3702, "step": 10316 }, { "epoch": 1.8, "learning_rate": 6.371002989159216e-06, "loss": 0.3353, "step": 10317 }, { "epoch": 1.8, "learning_rate": 6.368650269519828e-06, "loss": 0.3382, "step": 10318 }, { "epoch": 1.8, "learning_rate": 6.3662995069899435e-06, "loss": 0.3595, "step": 10319 }, { "epoch": 1.8, "learning_rate": 6.363950701787275e-06, "loss": 0.3521, "step": 10320 }, { "epoch": 1.8, "learning_rate": 6.3616038541293665e-06, "loss": 0.3457, "step": 10321 }, { "epoch": 1.8, "learning_rate": 6.35925896423357e-06, "loss": 0.3522, "step": 10322 }, { "epoch": 1.8, "learning_rate": 6.356916032317063e-06, "loss": 0.3506, "step": 10323 }, { "epoch": 1.8, "learning_rate": 6.354575058596839e-06, "loss": 0.3597, "step": 10324 }, { "epoch": 1.8, "learning_rate": 6.352236043289711e-06, "loss": 0.3338, "step": 10325 }, { "epoch": 1.8, "learning_rate": 6.34989898661231e-06, "loss": 0.3307, "step": 10326 }, { "epoch": 1.8, "learning_rate": 6.347563888781088e-06, "loss": 0.3284, "step": 10327 }, { "epoch": 1.8, "learning_rate": 6.34523075001231e-06, "loss": 0.3612, "step": 10328 }, { "epoch": 1.8, "learning_rate": 6.342899570522063e-06, "loss": 0.339, "step": 10329 }, { "epoch": 1.8, "learning_rate": 6.340570350526251e-06, "loss": 0.3274, "step": 10330 }, { "epoch": 1.8, "learning_rate": 6.338243090240605e-06, "loss": 0.3522, "step": 10331 }, { "epoch": 1.8, "learning_rate": 6.335917789880661e-06, "loss": 0.342, "step": 10332 }, { "epoch": 1.8, "learning_rate": 6.3335944496617795e-06, "loss": 0.3561, "step": 10333 }, { "epoch": 1.8, "learning_rate": 6.331273069799147e-06, "loss": 0.3433, "step": 10334 }, { "epoch": 1.8, "learning_rate": 6.3289536505077515e-06, "loss": 0.3542, "step": 10335 }, { "epoch": 1.8, "learning_rate": 6.326636192002418e-06, "loss": 0.3738, "step": 10336 }, { "epoch": 1.8, "learning_rate": 6.324320694497774e-06, "loss": 0.3617, "step": 10337 }, { "epoch": 1.8, "learning_rate": 6.322007158208279e-06, "loss": 0.355, "step": 10338 }, { "epoch": 1.8, "learning_rate": 6.319695583348204e-06, "loss": 0.3461, "step": 10339 }, { "epoch": 1.8, "learning_rate": 6.317385970131626e-06, "loss": 0.339, "step": 10340 }, { "epoch": 1.8, "learning_rate": 6.315078318772471e-06, "loss": 0.3484, "step": 10341 }, { "epoch": 1.8, "learning_rate": 6.312772629484459e-06, "loss": 0.338, "step": 10342 }, { "epoch": 1.8, "learning_rate": 6.31046890248113e-06, "loss": 0.3522, "step": 10343 }, { "epoch": 1.8, "learning_rate": 6.308167137975851e-06, "loss": 0.3458, "step": 10344 }, { "epoch": 1.8, "learning_rate": 6.305867336181802e-06, "loss": 0.3496, "step": 10345 }, { "epoch": 1.8, "learning_rate": 6.30356949731198e-06, "loss": 0.3579, "step": 10346 }, { "epoch": 1.8, "learning_rate": 6.301273621579207e-06, "loss": 0.3376, "step": 10347 }, { "epoch": 1.8, "learning_rate": 6.298979709196118e-06, "loss": 0.3485, "step": 10348 }, { "epoch": 1.8, "learning_rate": 6.296687760375167e-06, "loss": 0.3487, "step": 10349 }, { "epoch": 1.8, "learning_rate": 6.2943977753286195e-06, "loss": 0.3466, "step": 10350 }, { "epoch": 1.8, "learning_rate": 6.292109754268575e-06, "loss": 0.3637, "step": 10351 }, { "epoch": 1.8, "learning_rate": 6.289823697406937e-06, "loss": 0.3623, "step": 10352 }, { "epoch": 1.81, "learning_rate": 6.287539604955435e-06, "loss": 0.3613, "step": 10353 }, { "epoch": 1.81, "learning_rate": 6.285257477125605e-06, "loss": 0.3637, "step": 10354 }, { "epoch": 1.81, "learning_rate": 6.282977314128824e-06, "loss": 0.3505, "step": 10355 }, { "epoch": 1.81, "learning_rate": 6.280699116176259e-06, "loss": 0.3484, "step": 10356 }, { "epoch": 1.81, "learning_rate": 6.278422883478912e-06, "loss": 0.3396, "step": 10357 }, { "epoch": 1.81, "learning_rate": 6.2761486162476e-06, "loss": 0.3486, "step": 10358 }, { "epoch": 1.81, "learning_rate": 6.273876314692962e-06, "loss": 0.3327, "step": 10359 }, { "epoch": 1.81, "learning_rate": 6.271605979025448e-06, "loss": 0.3475, "step": 10360 }, { "epoch": 1.81, "learning_rate": 6.269337609455322e-06, "loss": 0.3492, "step": 10361 }, { "epoch": 1.81, "learning_rate": 6.26707120619268e-06, "loss": 0.3631, "step": 10362 }, { "epoch": 1.81, "learning_rate": 6.264806769447426e-06, "loss": 0.3598, "step": 10363 }, { "epoch": 1.81, "learning_rate": 6.262544299429281e-06, "loss": 0.3587, "step": 10364 }, { "epoch": 1.81, "learning_rate": 6.2602837963477935e-06, "loss": 0.3534, "step": 10365 }, { "epoch": 1.81, "learning_rate": 6.258025260412318e-06, "loss": 0.3507, "step": 10366 }, { "epoch": 1.81, "learning_rate": 6.255768691832028e-06, "loss": 0.3651, "step": 10367 }, { "epoch": 1.81, "learning_rate": 6.253514090815926e-06, "loss": 0.3467, "step": 10368 }, { "epoch": 1.81, "learning_rate": 6.2512614575728255e-06, "loss": 0.3601, "step": 10369 }, { "epoch": 1.81, "learning_rate": 6.2490107923113515e-06, "loss": 0.3549, "step": 10370 }, { "epoch": 1.81, "learning_rate": 6.246762095239955e-06, "loss": 0.3588, "step": 10371 }, { "epoch": 1.81, "learning_rate": 6.244515366566904e-06, "loss": 0.3532, "step": 10372 }, { "epoch": 1.81, "learning_rate": 6.242270606500284e-06, "loss": 0.3465, "step": 10373 }, { "epoch": 1.81, "learning_rate": 6.2400278152479886e-06, "loss": 0.3476, "step": 10374 }, { "epoch": 1.81, "learning_rate": 6.23778699301774e-06, "loss": 0.3478, "step": 10375 }, { "epoch": 1.81, "learning_rate": 6.235548140017083e-06, "loss": 0.3445, "step": 10376 }, { "epoch": 1.81, "learning_rate": 6.233311256453366e-06, "loss": 0.3444, "step": 10377 }, { "epoch": 1.81, "learning_rate": 6.231076342533756e-06, "loss": 0.3587, "step": 10378 }, { "epoch": 1.81, "learning_rate": 6.228843398465251e-06, "loss": 0.3429, "step": 10379 }, { "epoch": 1.81, "learning_rate": 6.226612424454655e-06, "loss": 0.3488, "step": 10380 }, { "epoch": 1.81, "learning_rate": 6.2243834207085935e-06, "loss": 0.3479, "step": 10381 }, { "epoch": 1.81, "learning_rate": 6.22215638743351e-06, "loss": 0.3523, "step": 10382 }, { "epoch": 1.81, "learning_rate": 6.219931324835662e-06, "loss": 0.3461, "step": 10383 }, { "epoch": 1.81, "learning_rate": 6.217708233121129e-06, "loss": 0.3666, "step": 10384 }, { "epoch": 1.81, "learning_rate": 6.215487112495796e-06, "loss": 0.336, "step": 10385 }, { "epoch": 1.81, "learning_rate": 6.21326796316539e-06, "loss": 0.347, "step": 10386 }, { "epoch": 1.81, "learning_rate": 6.2110507853354354e-06, "loss": 0.3498, "step": 10387 }, { "epoch": 1.81, "learning_rate": 6.208835579211273e-06, "loss": 0.3491, "step": 10388 }, { "epoch": 1.81, "learning_rate": 6.2066223449980765e-06, "loss": 0.3431, "step": 10389 }, { "epoch": 1.81, "learning_rate": 6.204411082900819e-06, "loss": 0.3509, "step": 10390 }, { "epoch": 1.81, "learning_rate": 6.202201793124307e-06, "loss": 0.3526, "step": 10391 }, { "epoch": 1.81, "learning_rate": 6.199994475873152e-06, "loss": 0.348, "step": 10392 }, { "epoch": 1.81, "learning_rate": 6.19778913135179e-06, "loss": 0.3355, "step": 10393 }, { "epoch": 1.81, "learning_rate": 6.195585759764474e-06, "loss": 0.3592, "step": 10394 }, { "epoch": 1.81, "learning_rate": 6.193384361315265e-06, "loss": 0.3422, "step": 10395 }, { "epoch": 1.81, "learning_rate": 6.191184936208056e-06, "loss": 0.3449, "step": 10396 }, { "epoch": 1.81, "learning_rate": 6.18898748464655e-06, "loss": 0.356, "step": 10397 }, { "epoch": 1.81, "learning_rate": 6.186792006834263e-06, "loss": 0.3641, "step": 10398 }, { "epoch": 1.81, "learning_rate": 6.184598502974532e-06, "loss": 0.357, "step": 10399 }, { "epoch": 1.81, "learning_rate": 6.182406973270514e-06, "loss": 0.3517, "step": 10400 }, { "epoch": 1.81, "learning_rate": 6.180217417925178e-06, "loss": 0.3441, "step": 10401 }, { "epoch": 1.81, "learning_rate": 6.178029837141314e-06, "loss": 0.3359, "step": 10402 }, { "epoch": 1.81, "learning_rate": 6.1758442311215326e-06, "loss": 0.366, "step": 10403 }, { "epoch": 1.81, "learning_rate": 6.1736606000682515e-06, "loss": 0.3688, "step": 10404 }, { "epoch": 1.81, "learning_rate": 6.171478944183708e-06, "loss": 0.3633, "step": 10405 }, { "epoch": 1.81, "learning_rate": 6.169299263669967e-06, "loss": 0.338, "step": 10406 }, { "epoch": 1.81, "learning_rate": 6.167121558728894e-06, "loss": 0.3498, "step": 10407 }, { "epoch": 1.81, "learning_rate": 6.164945829562191e-06, "loss": 0.3393, "step": 10408 }, { "epoch": 1.81, "learning_rate": 6.162772076371355e-06, "loss": 0.3532, "step": 10409 }, { "epoch": 1.81, "learning_rate": 6.160600299357719e-06, "loss": 0.3507, "step": 10410 }, { "epoch": 1.82, "learning_rate": 6.1584304987224205e-06, "loss": 0.3306, "step": 10411 }, { "epoch": 1.82, "learning_rate": 6.15626267466642e-06, "loss": 0.3323, "step": 10412 }, { "epoch": 1.82, "learning_rate": 6.154096827390492e-06, "loss": 0.3639, "step": 10413 }, { "epoch": 1.82, "learning_rate": 6.151932957095236e-06, "loss": 0.3464, "step": 10414 }, { "epoch": 1.82, "learning_rate": 6.149771063981058e-06, "loss": 0.3544, "step": 10415 }, { "epoch": 1.82, "learning_rate": 6.147611148248179e-06, "loss": 0.3551, "step": 10416 }, { "epoch": 1.82, "learning_rate": 6.145453210096652e-06, "loss": 0.3535, "step": 10417 }, { "epoch": 1.82, "learning_rate": 6.1432972497263305e-06, "loss": 0.3604, "step": 10418 }, { "epoch": 1.82, "learning_rate": 6.141143267336893e-06, "loss": 0.3686, "step": 10419 }, { "epoch": 1.82, "learning_rate": 6.138991263127838e-06, "loss": 0.3453, "step": 10420 }, { "epoch": 1.82, "learning_rate": 6.136841237298475e-06, "loss": 0.3566, "step": 10421 }, { "epoch": 1.82, "learning_rate": 6.134693190047926e-06, "loss": 0.376, "step": 10422 }, { "epoch": 1.82, "learning_rate": 6.13254712157514e-06, "loss": 0.337, "step": 10423 }, { "epoch": 1.82, "learning_rate": 6.13040303207888e-06, "loss": 0.3535, "step": 10424 }, { "epoch": 1.82, "learning_rate": 6.128260921757721e-06, "loss": 0.3461, "step": 10425 }, { "epoch": 1.82, "learning_rate": 6.126120790810055e-06, "loss": 0.3378, "step": 10426 }, { "epoch": 1.82, "learning_rate": 6.1239826394341005e-06, "loss": 0.3482, "step": 10427 }, { "epoch": 1.82, "learning_rate": 6.121846467827881e-06, "loss": 0.3463, "step": 10428 }, { "epoch": 1.82, "learning_rate": 6.1197122761892374e-06, "loss": 0.3572, "step": 10429 }, { "epoch": 1.82, "learning_rate": 6.1175800647158345e-06, "loss": 0.3259, "step": 10430 }, { "epoch": 1.82, "learning_rate": 6.115449833605154e-06, "loss": 0.3537, "step": 10431 }, { "epoch": 1.82, "learning_rate": 6.113321583054483e-06, "loss": 0.3404, "step": 10432 }, { "epoch": 1.82, "learning_rate": 6.111195313260936e-06, "loss": 0.3606, "step": 10433 }, { "epoch": 1.82, "learning_rate": 6.109071024421438e-06, "loss": 0.3362, "step": 10434 }, { "epoch": 1.82, "learning_rate": 6.106948716732741e-06, "loss": 0.3551, "step": 10435 }, { "epoch": 1.82, "learning_rate": 6.104828390391397e-06, "loss": 0.3484, "step": 10436 }, { "epoch": 1.82, "learning_rate": 6.1027100455937844e-06, "loss": 0.3287, "step": 10437 }, { "epoch": 1.82, "learning_rate": 6.100593682536099e-06, "loss": 0.3515, "step": 10438 }, { "epoch": 1.82, "learning_rate": 6.098479301414353e-06, "loss": 0.3406, "step": 10439 }, { "epoch": 1.82, "learning_rate": 6.096366902424359e-06, "loss": 0.3448, "step": 10440 }, { "epoch": 1.82, "learning_rate": 6.094256485761777e-06, "loss": 0.3541, "step": 10441 }, { "epoch": 1.82, "learning_rate": 6.092148051622062e-06, "loss": 0.3364, "step": 10442 }, { "epoch": 1.82, "learning_rate": 6.090041600200481e-06, "loss": 0.3592, "step": 10443 }, { "epoch": 1.82, "learning_rate": 6.087937131692137e-06, "loss": 0.3387, "step": 10444 }, { "epoch": 1.82, "learning_rate": 6.08583464629193e-06, "loss": 0.3522, "step": 10445 }, { "epoch": 1.82, "learning_rate": 6.083734144194591e-06, "loss": 0.3275, "step": 10446 }, { "epoch": 1.82, "learning_rate": 6.081635625594654e-06, "loss": 0.3437, "step": 10447 }, { "epoch": 1.82, "learning_rate": 6.079539090686483e-06, "loss": 0.354, "step": 10448 }, { "epoch": 1.82, "learning_rate": 6.077444539664249e-06, "loss": 0.3422, "step": 10449 }, { "epoch": 1.82, "learning_rate": 6.075351972721937e-06, "loss": 0.3377, "step": 10450 }, { "epoch": 1.82, "learning_rate": 6.073261390053358e-06, "loss": 0.3493, "step": 10451 }, { "epoch": 1.82, "learning_rate": 6.071172791852137e-06, "loss": 0.3555, "step": 10452 }, { "epoch": 1.82, "learning_rate": 6.069086178311709e-06, "loss": 0.3625, "step": 10453 }, { "epoch": 1.82, "learning_rate": 6.067001549625325e-06, "loss": 0.337, "step": 10454 }, { "epoch": 1.82, "learning_rate": 6.064918905986062e-06, "loss": 0.3453, "step": 10455 }, { "epoch": 1.82, "learning_rate": 6.062838247586802e-06, "loss": 0.3454, "step": 10456 }, { "epoch": 1.82, "learning_rate": 6.060759574620249e-06, "loss": 0.3622, "step": 10457 }, { "epoch": 1.82, "learning_rate": 6.058682887278927e-06, "loss": 0.3515, "step": 10458 }, { "epoch": 1.82, "learning_rate": 6.056608185755169e-06, "loss": 0.3436, "step": 10459 }, { "epoch": 1.82, "learning_rate": 6.054535470241121e-06, "loss": 0.3506, "step": 10460 }, { "epoch": 1.82, "learning_rate": 6.0524647409287586e-06, "loss": 0.3397, "step": 10461 }, { "epoch": 1.82, "learning_rate": 6.050395998009858e-06, "loss": 0.3535, "step": 10462 }, { "epoch": 1.82, "learning_rate": 6.048329241676027e-06, "loss": 0.3386, "step": 10463 }, { "epoch": 1.82, "learning_rate": 6.046264472118672e-06, "loss": 0.336, "step": 10464 }, { "epoch": 1.82, "learning_rate": 6.0442016895290315e-06, "loss": 0.351, "step": 10465 }, { "epoch": 1.82, "learning_rate": 6.042140894098149e-06, "loss": 0.3637, "step": 10466 }, { "epoch": 1.82, "learning_rate": 6.040082086016888e-06, "loss": 0.3461, "step": 10467 }, { "epoch": 1.83, "learning_rate": 6.038025265475928e-06, "loss": 0.3534, "step": 10468 }, { "epoch": 1.83, "learning_rate": 6.035970432665769e-06, "loss": 0.344, "step": 10469 }, { "epoch": 1.83, "learning_rate": 6.033917587776719e-06, "loss": 0.3561, "step": 10470 }, { "epoch": 1.83, "learning_rate": 6.031866730998903e-06, "loss": 0.352, "step": 10471 }, { "epoch": 1.83, "learning_rate": 6.029817862522266e-06, "loss": 0.3292, "step": 10472 }, { "epoch": 1.83, "learning_rate": 6.027770982536564e-06, "loss": 0.3557, "step": 10473 }, { "epoch": 1.83, "learning_rate": 6.0257260912313785e-06, "loss": 0.3553, "step": 10474 }, { "epoch": 1.83, "learning_rate": 6.023683188796093e-06, "loss": 0.3518, "step": 10475 }, { "epoch": 1.83, "learning_rate": 6.0216422754199185e-06, "loss": 0.3643, "step": 10476 }, { "epoch": 1.83, "learning_rate": 6.019603351291875e-06, "loss": 0.3487, "step": 10477 }, { "epoch": 1.83, "learning_rate": 6.017566416600795e-06, "loss": 0.3649, "step": 10478 }, { "epoch": 1.83, "learning_rate": 6.0155314715353415e-06, "loss": 0.3399, "step": 10479 }, { "epoch": 1.83, "learning_rate": 6.01349851628398e-06, "loss": 0.3496, "step": 10480 }, { "epoch": 1.83, "learning_rate": 6.011467551034992e-06, "loss": 0.3522, "step": 10481 }, { "epoch": 1.83, "learning_rate": 6.009438575976486e-06, "loss": 0.3515, "step": 10482 }, { "epoch": 1.83, "learning_rate": 6.007411591296373e-06, "loss": 0.3754, "step": 10483 }, { "epoch": 1.83, "learning_rate": 6.005386597182382e-06, "loss": 0.3521, "step": 10484 }, { "epoch": 1.83, "learning_rate": 6.0033635938220645e-06, "loss": 0.3478, "step": 10485 }, { "epoch": 1.83, "learning_rate": 6.001342581402787e-06, "loss": 0.3399, "step": 10486 }, { "epoch": 1.83, "learning_rate": 5.999323560111726e-06, "loss": 0.3377, "step": 10487 }, { "epoch": 1.83, "learning_rate": 5.997306530135871e-06, "loss": 0.3425, "step": 10488 }, { "epoch": 1.83, "learning_rate": 5.9952914916620365e-06, "loss": 0.3517, "step": 10489 }, { "epoch": 1.83, "learning_rate": 5.993278444876849e-06, "loss": 0.34, "step": 10490 }, { "epoch": 1.83, "learning_rate": 5.991267389966749e-06, "loss": 0.3451, "step": 10491 }, { "epoch": 1.83, "learning_rate": 5.98925832711799e-06, "loss": 0.3431, "step": 10492 }, { "epoch": 1.83, "learning_rate": 5.987251256516649e-06, "loss": 0.3416, "step": 10493 }, { "epoch": 1.83, "learning_rate": 5.985246178348612e-06, "loss": 0.3469, "step": 10494 }, { "epoch": 1.83, "learning_rate": 5.983243092799576e-06, "loss": 0.3565, "step": 10495 }, { "epoch": 1.83, "learning_rate": 5.981242000055068e-06, "loss": 0.349, "step": 10496 }, { "epoch": 1.83, "learning_rate": 5.979242900300423e-06, "loss": 0.3413, "step": 10497 }, { "epoch": 1.83, "learning_rate": 5.97724579372078e-06, "loss": 0.3408, "step": 10498 }, { "epoch": 1.83, "learning_rate": 5.975250680501116e-06, "loss": 0.3612, "step": 10499 }, { "epoch": 1.83, "learning_rate": 5.973257560826201e-06, "loss": 0.383, "step": 10500 }, { "epoch": 1.83, "learning_rate": 5.971266434880635e-06, "loss": 0.3525, "step": 10501 }, { "epoch": 1.83, "learning_rate": 5.969277302848829e-06, "loss": 0.3471, "step": 10502 }, { "epoch": 1.83, "learning_rate": 5.967290164915014e-06, "loss": 0.3468, "step": 10503 }, { "epoch": 1.83, "learning_rate": 5.965305021263222e-06, "loss": 0.3425, "step": 10504 }, { "epoch": 1.83, "learning_rate": 5.963321872077315e-06, "loss": 0.3402, "step": 10505 }, { "epoch": 1.83, "learning_rate": 5.9613407175409646e-06, "loss": 0.3606, "step": 10506 }, { "epoch": 1.83, "learning_rate": 5.95936155783766e-06, "loss": 0.3486, "step": 10507 }, { "epoch": 1.83, "learning_rate": 5.957384393150702e-06, "loss": 0.3611, "step": 10508 }, { "epoch": 1.83, "learning_rate": 5.9554092236632055e-06, "loss": 0.3521, "step": 10509 }, { "epoch": 1.83, "learning_rate": 5.95343604955811e-06, "loss": 0.3566, "step": 10510 }, { "epoch": 1.83, "learning_rate": 5.951464871018161e-06, "loss": 0.3407, "step": 10511 }, { "epoch": 1.83, "learning_rate": 5.9494956882259155e-06, "loss": 0.3611, "step": 10512 }, { "epoch": 1.83, "learning_rate": 5.947528501363764e-06, "loss": 0.363, "step": 10513 }, { "epoch": 1.83, "learning_rate": 5.9455633106138935e-06, "loss": 0.3531, "step": 10514 }, { "epoch": 1.83, "learning_rate": 5.94360011615831e-06, "loss": 0.3515, "step": 10515 }, { "epoch": 1.83, "learning_rate": 5.941638918178846e-06, "loss": 0.3553, "step": 10516 }, { "epoch": 1.83, "learning_rate": 5.939679716857131e-06, "loss": 0.3331, "step": 10517 }, { "epoch": 1.83, "learning_rate": 5.937722512374628e-06, "loss": 0.3526, "step": 10518 }, { "epoch": 1.83, "learning_rate": 5.935767304912598e-06, "loss": 0.3606, "step": 10519 }, { "epoch": 1.83, "learning_rate": 5.9338140946521336e-06, "loss": 0.343, "step": 10520 }, { "epoch": 1.83, "learning_rate": 5.93186288177413e-06, "loss": 0.3565, "step": 10521 }, { "epoch": 1.83, "learning_rate": 5.929913666459299e-06, "loss": 0.3449, "step": 10522 }, { "epoch": 1.83, "learning_rate": 5.927966448888171e-06, "loss": 0.3405, "step": 10523 }, { "epoch": 1.83, "learning_rate": 5.926021229241095e-06, "loss": 0.3514, "step": 10524 }, { "epoch": 1.84, "learning_rate": 5.924078007698227e-06, "loss": 0.3372, "step": 10525 }, { "epoch": 1.84, "learning_rate": 5.922136784439538e-06, "loss": 0.3439, "step": 10526 }, { "epoch": 1.84, "learning_rate": 5.920197559644823e-06, "loss": 0.3685, "step": 10527 }, { "epoch": 1.84, "learning_rate": 5.918260333493679e-06, "loss": 0.3298, "step": 10528 }, { "epoch": 1.84, "learning_rate": 5.9163251061655355e-06, "loss": 0.345, "step": 10529 }, { "epoch": 1.84, "learning_rate": 5.9143918778396135e-06, "loss": 0.3546, "step": 10530 }, { "epoch": 1.84, "learning_rate": 5.912460648694971e-06, "loss": 0.3428, "step": 10531 }, { "epoch": 1.84, "learning_rate": 5.910531418910471e-06, "loss": 0.3444, "step": 10532 }, { "epoch": 1.84, "learning_rate": 5.908604188664785e-06, "loss": 0.3578, "step": 10533 }, { "epoch": 1.84, "learning_rate": 5.9066789581364085e-06, "loss": 0.3508, "step": 10534 }, { "epoch": 1.84, "learning_rate": 5.904755727503657e-06, "loss": 0.3531, "step": 10535 }, { "epoch": 1.84, "learning_rate": 5.9028344969446436e-06, "loss": 0.3412, "step": 10536 }, { "epoch": 1.84, "learning_rate": 5.900915266637312e-06, "loss": 0.335, "step": 10537 }, { "epoch": 1.84, "learning_rate": 5.898998036759415e-06, "loss": 0.3521, "step": 10538 }, { "epoch": 1.84, "learning_rate": 5.89708280748851e-06, "loss": 0.3409, "step": 10539 }, { "epoch": 1.84, "learning_rate": 5.895169579001987e-06, "loss": 0.3496, "step": 10540 }, { "epoch": 1.84, "learning_rate": 5.893258351477042e-06, "loss": 0.361, "step": 10541 }, { "epoch": 1.84, "learning_rate": 5.891349125090688e-06, "loss": 0.3342, "step": 10542 }, { "epoch": 1.84, "learning_rate": 5.889441900019741e-06, "loss": 0.3389, "step": 10543 }, { "epoch": 1.84, "learning_rate": 5.887536676440851e-06, "loss": 0.3607, "step": 10544 }, { "epoch": 1.84, "learning_rate": 5.8856334545304676e-06, "loss": 0.3525, "step": 10545 }, { "epoch": 1.84, "learning_rate": 5.883732234464864e-06, "loss": 0.349, "step": 10546 }, { "epoch": 1.84, "learning_rate": 5.8818330164201195e-06, "loss": 0.3485, "step": 10547 }, { "epoch": 1.84, "learning_rate": 5.879935800572142e-06, "loss": 0.3644, "step": 10548 }, { "epoch": 1.84, "learning_rate": 5.878040587096636e-06, "loss": 0.3448, "step": 10549 }, { "epoch": 1.84, "learning_rate": 5.876147376169127e-06, "loss": 0.365, "step": 10550 }, { "epoch": 1.84, "learning_rate": 5.8742561679649685e-06, "loss": 0.3341, "step": 10551 }, { "epoch": 1.84, "learning_rate": 5.87236696265931e-06, "loss": 0.3541, "step": 10552 }, { "epoch": 1.84, "learning_rate": 5.8704797604271225e-06, "loss": 0.3297, "step": 10553 }, { "epoch": 1.84, "learning_rate": 5.868594561443198e-06, "loss": 0.3515, "step": 10554 }, { "epoch": 1.84, "learning_rate": 5.866711365882129e-06, "loss": 0.355, "step": 10555 }, { "epoch": 1.84, "learning_rate": 5.864830173918334e-06, "loss": 0.3583, "step": 10556 }, { "epoch": 1.84, "learning_rate": 5.862950985726039e-06, "loss": 0.3514, "step": 10557 }, { "epoch": 1.84, "learning_rate": 5.861073801479296e-06, "loss": 0.3368, "step": 10558 }, { "epoch": 1.84, "learning_rate": 5.859198621351958e-06, "loss": 0.3501, "step": 10559 }, { "epoch": 1.84, "learning_rate": 5.857325445517692e-06, "loss": 0.3429, "step": 10560 }, { "epoch": 1.84, "learning_rate": 5.8554542741499935e-06, "loss": 0.336, "step": 10561 }, { "epoch": 1.84, "learning_rate": 5.853585107422159e-06, "loss": 0.3848, "step": 10562 }, { "epoch": 1.84, "learning_rate": 5.851717945507307e-06, "loss": 0.3623, "step": 10563 }, { "epoch": 1.84, "learning_rate": 5.849852788578361e-06, "loss": 0.3408, "step": 10564 }, { "epoch": 1.84, "learning_rate": 5.8479896368080735e-06, "loss": 0.3497, "step": 10565 }, { "epoch": 1.84, "learning_rate": 5.846128490368999e-06, "loss": 0.3446, "step": 10566 }, { "epoch": 1.84, "learning_rate": 5.844269349433509e-06, "loss": 0.3574, "step": 10567 }, { "epoch": 1.84, "learning_rate": 5.842412214173788e-06, "loss": 0.3364, "step": 10568 }, { "epoch": 1.84, "learning_rate": 5.840557084761846e-06, "loss": 0.372, "step": 10569 }, { "epoch": 1.84, "learning_rate": 5.838703961369492e-06, "loss": 0.3594, "step": 10570 }, { "epoch": 1.84, "learning_rate": 5.8368528441683525e-06, "loss": 0.3538, "step": 10571 }, { "epoch": 1.84, "learning_rate": 5.835003733329878e-06, "loss": 0.3579, "step": 10572 }, { "epoch": 1.84, "learning_rate": 5.833156629025327e-06, "loss": 0.3609, "step": 10573 }, { "epoch": 1.84, "learning_rate": 5.831311531425763e-06, "loss": 0.3765, "step": 10574 }, { "epoch": 1.84, "learning_rate": 5.829468440702084e-06, "loss": 0.3307, "step": 10575 }, { "epoch": 1.84, "learning_rate": 5.8276273570249815e-06, "loss": 0.346, "step": 10576 }, { "epoch": 1.84, "learning_rate": 5.825788280564973e-06, "loss": 0.3587, "step": 10577 }, { "epoch": 1.84, "learning_rate": 5.823951211492384e-06, "loss": 0.3406, "step": 10578 }, { "epoch": 1.84, "learning_rate": 5.822116149977364e-06, "loss": 0.335, "step": 10579 }, { "epoch": 1.84, "learning_rate": 5.820283096189867e-06, "loss": 0.3487, "step": 10580 }, { "epoch": 1.84, "learning_rate": 5.81845205029966e-06, "loss": 0.3433, "step": 10581 }, { "epoch": 1.84, "learning_rate": 5.816623012476331e-06, "loss": 0.3509, "step": 10582 }, { "epoch": 1.85, "learning_rate": 5.814795982889281e-06, "loss": 0.346, "step": 10583 }, { "epoch": 1.85, "learning_rate": 5.812970961707718e-06, "loss": 0.352, "step": 10584 }, { "epoch": 1.85, "learning_rate": 5.811147949100671e-06, "loss": 0.3467, "step": 10585 }, { "epoch": 1.85, "learning_rate": 5.809326945236986e-06, "loss": 0.3657, "step": 10586 }, { "epoch": 1.85, "learning_rate": 5.80750795028531e-06, "loss": 0.3703, "step": 10587 }, { "epoch": 1.85, "learning_rate": 5.8056909644141136e-06, "loss": 0.3452, "step": 10588 }, { "epoch": 1.85, "learning_rate": 5.803875987791679e-06, "loss": 0.3479, "step": 10589 }, { "epoch": 1.85, "learning_rate": 5.802063020586111e-06, "loss": 0.3619, "step": 10590 }, { "epoch": 1.85, "learning_rate": 5.8002520629653085e-06, "loss": 0.3473, "step": 10591 }, { "epoch": 1.85, "learning_rate": 5.798443115097002e-06, "loss": 0.354, "step": 10592 }, { "epoch": 1.85, "learning_rate": 5.796636177148733e-06, "loss": 0.3508, "step": 10593 }, { "epoch": 1.85, "learning_rate": 5.794831249287844e-06, "loss": 0.3424, "step": 10594 }, { "epoch": 1.85, "learning_rate": 5.793028331681506e-06, "loss": 0.3433, "step": 10595 }, { "epoch": 1.85, "learning_rate": 5.791227424496701e-06, "loss": 0.3525, "step": 10596 }, { "epoch": 1.85, "learning_rate": 5.789428527900223e-06, "loss": 0.3486, "step": 10597 }, { "epoch": 1.85, "learning_rate": 5.787631642058673e-06, "loss": 0.3424, "step": 10598 }, { "epoch": 1.85, "learning_rate": 5.7858367671384795e-06, "loss": 0.3474, "step": 10599 }, { "epoch": 1.85, "learning_rate": 5.784043903305869e-06, "loss": 0.3582, "step": 10600 }, { "epoch": 1.85, "learning_rate": 5.7822530507269e-06, "loss": 0.3454, "step": 10601 }, { "epoch": 1.85, "learning_rate": 5.780464209567425e-06, "loss": 0.359, "step": 10602 }, { "epoch": 1.85, "learning_rate": 5.778677379993128e-06, "loss": 0.35, "step": 10603 }, { "epoch": 1.85, "learning_rate": 5.776892562169495e-06, "loss": 0.3636, "step": 10604 }, { "epoch": 1.85, "learning_rate": 5.775109756261828e-06, "loss": 0.3646, "step": 10605 }, { "epoch": 1.85, "learning_rate": 5.773328962435248e-06, "loss": 0.3461, "step": 10606 }, { "epoch": 1.85, "learning_rate": 5.771550180854683e-06, "loss": 0.3928, "step": 10607 }, { "epoch": 1.85, "learning_rate": 5.769773411684881e-06, "loss": 0.3559, "step": 10608 }, { "epoch": 1.85, "learning_rate": 5.76799865509039e-06, "loss": 0.3529, "step": 10609 }, { "epoch": 1.85, "learning_rate": 5.766225911235594e-06, "loss": 0.3543, "step": 10610 }, { "epoch": 1.85, "learning_rate": 5.764455180284669e-06, "loss": 0.356, "step": 10611 }, { "epoch": 1.85, "learning_rate": 5.762686462401619e-06, "loss": 0.3451, "step": 10612 }, { "epoch": 1.85, "learning_rate": 5.760919757750255e-06, "loss": 0.3602, "step": 10613 }, { "epoch": 1.85, "learning_rate": 5.759155066494205e-06, "loss": 0.3695, "step": 10614 }, { "epoch": 1.85, "learning_rate": 5.757392388796901e-06, "loss": 0.3538, "step": 10615 }, { "epoch": 1.85, "learning_rate": 5.755631724821604e-06, "loss": 0.3494, "step": 10616 }, { "epoch": 1.85, "learning_rate": 5.753873074731376e-06, "loss": 0.3429, "step": 10617 }, { "epoch": 1.85, "learning_rate": 5.7521164386890995e-06, "loss": 0.3466, "step": 10618 }, { "epoch": 1.85, "learning_rate": 5.750361816857463e-06, "loss": 0.3453, "step": 10619 }, { "epoch": 1.85, "learning_rate": 5.748609209398979e-06, "loss": 0.335, "step": 10620 }, { "epoch": 1.85, "learning_rate": 5.7468586164759645e-06, "loss": 0.3374, "step": 10621 }, { "epoch": 1.85, "learning_rate": 5.745110038250551e-06, "loss": 0.3567, "step": 10622 }, { "epoch": 1.85, "learning_rate": 5.74336347488469e-06, "loss": 0.3536, "step": 10623 }, { "epoch": 1.85, "learning_rate": 5.741618926540143e-06, "loss": 0.3678, "step": 10624 }, { "epoch": 1.85, "learning_rate": 5.739876393378479e-06, "loss": 0.3253, "step": 10625 }, { "epoch": 1.85, "learning_rate": 5.7381358755610865e-06, "loss": 0.3515, "step": 10626 }, { "epoch": 1.85, "learning_rate": 5.736397373249167e-06, "loss": 0.3493, "step": 10627 }, { "epoch": 1.85, "learning_rate": 5.734660886603734e-06, "loss": 0.3387, "step": 10628 }, { "epoch": 1.85, "learning_rate": 5.7329264157856125e-06, "loss": 0.3578, "step": 10629 }, { "epoch": 1.85, "learning_rate": 5.731193960955448e-06, "loss": 0.3609, "step": 10630 }, { "epoch": 1.85, "learning_rate": 5.7294635222736894e-06, "loss": 0.358, "step": 10631 }, { "epoch": 1.85, "learning_rate": 5.727735099900605e-06, "loss": 0.3537, "step": 10632 }, { "epoch": 1.85, "learning_rate": 5.726008693996276e-06, "loss": 0.3437, "step": 10633 }, { "epoch": 1.85, "learning_rate": 5.724284304720596e-06, "loss": 0.3541, "step": 10634 }, { "epoch": 1.85, "learning_rate": 5.722561932233274e-06, "loss": 0.3507, "step": 10635 }, { "epoch": 1.85, "learning_rate": 5.720841576693821e-06, "loss": 0.3364, "step": 10636 }, { "epoch": 1.85, "learning_rate": 5.719123238261582e-06, "loss": 0.3481, "step": 10637 }, { "epoch": 1.85, "learning_rate": 5.717406917095695e-06, "loss": 0.3423, "step": 10638 }, { "epoch": 1.85, "learning_rate": 5.715692613355122e-06, "loss": 0.3369, "step": 10639 }, { "epoch": 1.86, "learning_rate": 5.7139803271986345e-06, "loss": 0.3585, "step": 10640 }, { "epoch": 1.86, "learning_rate": 5.712270058784821e-06, "loss": 0.351, "step": 10641 }, { "epoch": 1.86, "learning_rate": 5.710561808272082e-06, "loss": 0.353, "step": 10642 }, { "epoch": 1.86, "learning_rate": 5.708855575818619e-06, "loss": 0.3549, "step": 10643 }, { "epoch": 1.86, "learning_rate": 5.707151361582469e-06, "loss": 0.3464, "step": 10644 }, { "epoch": 1.86, "learning_rate": 5.705449165721467e-06, "loss": 0.3404, "step": 10645 }, { "epoch": 1.86, "learning_rate": 5.703748988393258e-06, "loss": 0.3404, "step": 10646 }, { "epoch": 1.86, "learning_rate": 5.702050829755316e-06, "loss": 0.3584, "step": 10647 }, { "epoch": 1.86, "learning_rate": 5.700354689964914e-06, "loss": 0.364, "step": 10648 }, { "epoch": 1.86, "learning_rate": 5.698660569179138e-06, "loss": 0.3277, "step": 10649 }, { "epoch": 1.86, "learning_rate": 5.696968467554895e-06, "loss": 0.3537, "step": 10650 }, { "epoch": 1.86, "learning_rate": 5.695278385248905e-06, "loss": 0.3349, "step": 10651 }, { "epoch": 1.86, "learning_rate": 5.693590322417695e-06, "loss": 0.3467, "step": 10652 }, { "epoch": 1.86, "learning_rate": 5.691904279217602e-06, "loss": 0.3393, "step": 10653 }, { "epoch": 1.86, "learning_rate": 5.690220255804789e-06, "loss": 0.3656, "step": 10654 }, { "epoch": 1.86, "learning_rate": 5.688538252335217e-06, "loss": 0.3605, "step": 10655 }, { "epoch": 1.86, "learning_rate": 5.686858268964674e-06, "loss": 0.341, "step": 10656 }, { "epoch": 1.86, "learning_rate": 5.6851803058487484e-06, "loss": 0.3575, "step": 10657 }, { "epoch": 1.86, "learning_rate": 5.683504363142851e-06, "loss": 0.3602, "step": 10658 }, { "epoch": 1.86, "learning_rate": 5.681830441002202e-06, "loss": 0.3574, "step": 10659 }, { "epoch": 1.86, "learning_rate": 5.680158539581829e-06, "loss": 0.3438, "step": 10660 }, { "epoch": 1.86, "learning_rate": 5.67848865903658e-06, "loss": 0.3385, "step": 10661 }, { "epoch": 1.86, "learning_rate": 5.676820799521116e-06, "loss": 0.3408, "step": 10662 }, { "epoch": 1.86, "learning_rate": 5.675154961189906e-06, "loss": 0.3542, "step": 10663 }, { "epoch": 1.86, "learning_rate": 5.6734911441972325e-06, "loss": 0.3585, "step": 10664 }, { "epoch": 1.86, "learning_rate": 5.671829348697198e-06, "loss": 0.3497, "step": 10665 }, { "epoch": 1.86, "learning_rate": 5.670169574843703e-06, "loss": 0.3573, "step": 10666 }, { "epoch": 1.86, "learning_rate": 5.668511822790476e-06, "loss": 0.3466, "step": 10667 }, { "epoch": 1.86, "learning_rate": 5.66685609269105e-06, "loss": 0.3593, "step": 10668 }, { "epoch": 1.86, "learning_rate": 5.665202384698776e-06, "loss": 0.3447, "step": 10669 }, { "epoch": 1.86, "learning_rate": 5.6635506989668084e-06, "loss": 0.3691, "step": 10670 }, { "epoch": 1.86, "learning_rate": 5.661901035648125e-06, "loss": 0.3472, "step": 10671 }, { "epoch": 1.86, "learning_rate": 5.6602533948955095e-06, "loss": 0.3375, "step": 10672 }, { "epoch": 1.86, "learning_rate": 5.658607776861563e-06, "loss": 0.3339, "step": 10673 }, { "epoch": 1.86, "learning_rate": 5.656964181698696e-06, "loss": 0.338, "step": 10674 }, { "epoch": 1.86, "learning_rate": 5.655322609559129e-06, "loss": 0.3576, "step": 10675 }, { "epoch": 1.86, "learning_rate": 5.653683060594906e-06, "loss": 0.3518, "step": 10676 }, { "epoch": 1.86, "learning_rate": 5.652045534957864e-06, "loss": 0.3551, "step": 10677 }, { "epoch": 1.86, "learning_rate": 5.650410032799676e-06, "loss": 0.3406, "step": 10678 }, { "epoch": 1.86, "learning_rate": 5.648776554271811e-06, "loss": 0.3356, "step": 10679 }, { "epoch": 1.86, "learning_rate": 5.647145099525559e-06, "loss": 0.3549, "step": 10680 }, { "epoch": 1.86, "learning_rate": 5.645515668712015e-06, "loss": 0.3635, "step": 10681 }, { "epoch": 1.86, "learning_rate": 5.643888261982094e-06, "loss": 0.3377, "step": 10682 }, { "epoch": 1.86, "learning_rate": 5.642262879486521e-06, "loss": 0.3566, "step": 10683 }, { "epoch": 1.86, "learning_rate": 5.6406395213758306e-06, "loss": 0.3466, "step": 10684 }, { "epoch": 1.86, "learning_rate": 5.639018187800376e-06, "loss": 0.351, "step": 10685 }, { "epoch": 1.86, "learning_rate": 5.6373988789103164e-06, "loss": 0.3413, "step": 10686 }, { "epoch": 1.86, "learning_rate": 5.6357815948556255e-06, "loss": 0.3376, "step": 10687 }, { "epoch": 1.86, "learning_rate": 5.634166335786095e-06, "loss": 0.3523, "step": 10688 }, { "epoch": 1.86, "learning_rate": 5.632553101851318e-06, "loss": 0.3613, "step": 10689 }, { "epoch": 1.86, "learning_rate": 5.630941893200713e-06, "loss": 0.339, "step": 10690 }, { "epoch": 1.86, "learning_rate": 5.629332709983498e-06, "loss": 0.3466, "step": 10691 }, { "epoch": 1.86, "learning_rate": 5.627725552348717e-06, "loss": 0.349, "step": 10692 }, { "epoch": 1.86, "learning_rate": 5.6261204204452126e-06, "loss": 0.3417, "step": 10693 }, { "epoch": 1.86, "learning_rate": 5.624517314421649e-06, "loss": 0.3568, "step": 10694 }, { "epoch": 1.86, "learning_rate": 5.622916234426498e-06, "loss": 0.345, "step": 10695 }, { "epoch": 1.86, "learning_rate": 5.62131718060805e-06, "loss": 0.3683, "step": 10696 }, { "epoch": 1.87, "learning_rate": 5.619720153114404e-06, "loss": 0.3474, "step": 10697 }, { "epoch": 1.87, "learning_rate": 5.618125152093461e-06, "loss": 0.3513, "step": 10698 }, { "epoch": 1.87, "learning_rate": 5.616532177692957e-06, "loss": 0.3426, "step": 10699 }, { "epoch": 1.87, "learning_rate": 5.614941230060421e-06, "loss": 0.3566, "step": 10700 }, { "epoch": 1.87, "learning_rate": 5.613352309343201e-06, "loss": 0.3536, "step": 10701 }, { "epoch": 1.87, "learning_rate": 5.61176541568846e-06, "loss": 0.3644, "step": 10702 }, { "epoch": 1.87, "learning_rate": 5.610180549243166e-06, "loss": 0.3515, "step": 10703 }, { "epoch": 1.87, "learning_rate": 5.608597710154108e-06, "loss": 0.3565, "step": 10704 }, { "epoch": 1.87, "learning_rate": 5.607016898567878e-06, "loss": 0.3593, "step": 10705 }, { "epoch": 1.87, "learning_rate": 5.605438114630891e-06, "loss": 0.349, "step": 10706 }, { "epoch": 1.87, "learning_rate": 5.603861358489364e-06, "loss": 0.3353, "step": 10707 }, { "epoch": 1.87, "learning_rate": 5.602286630289331e-06, "loss": 0.3558, "step": 10708 }, { "epoch": 1.87, "learning_rate": 5.600713930176641e-06, "loss": 0.35, "step": 10709 }, { "epoch": 1.87, "learning_rate": 5.599143258296948e-06, "loss": 0.353, "step": 10710 }, { "epoch": 1.87, "learning_rate": 5.59757461479572e-06, "loss": 0.3476, "step": 10711 }, { "epoch": 1.87, "learning_rate": 5.596007999818244e-06, "loss": 0.3381, "step": 10712 }, { "epoch": 1.87, "learning_rate": 5.594443413509615e-06, "loss": 0.3417, "step": 10713 }, { "epoch": 1.87, "learning_rate": 5.592880856014736e-06, "loss": 0.3466, "step": 10714 }, { "epoch": 1.87, "learning_rate": 5.591320327478325e-06, "loss": 0.3583, "step": 10715 }, { "epoch": 1.87, "learning_rate": 5.589761828044913e-06, "loss": 0.3514, "step": 10716 }, { "epoch": 1.87, "learning_rate": 5.588205357858846e-06, "loss": 0.3459, "step": 10717 }, { "epoch": 1.87, "learning_rate": 5.586650917064274e-06, "loss": 0.3527, "step": 10718 }, { "epoch": 1.87, "learning_rate": 5.585098505805166e-06, "loss": 0.3614, "step": 10719 }, { "epoch": 1.87, "learning_rate": 5.583548124225304e-06, "loss": 0.3582, "step": 10720 }, { "epoch": 1.87, "learning_rate": 5.5819997724682725e-06, "loss": 0.3518, "step": 10721 }, { "epoch": 1.87, "learning_rate": 5.580453450677473e-06, "loss": 0.3427, "step": 10722 }, { "epoch": 1.87, "learning_rate": 5.578909158996131e-06, "loss": 0.3546, "step": 10723 }, { "epoch": 1.87, "learning_rate": 5.577366897567264e-06, "loss": 0.3442, "step": 10724 }, { "epoch": 1.87, "learning_rate": 5.575826666533712e-06, "loss": 0.3582, "step": 10725 }, { "epoch": 1.87, "learning_rate": 5.574288466038128e-06, "loss": 0.3491, "step": 10726 }, { "epoch": 1.87, "learning_rate": 5.572752296222972e-06, "loss": 0.3537, "step": 10727 }, { "epoch": 1.87, "learning_rate": 5.571218157230525e-06, "loss": 0.3388, "step": 10728 }, { "epoch": 1.87, "learning_rate": 5.569686049202863e-06, "loss": 0.3409, "step": 10729 }, { "epoch": 1.87, "learning_rate": 5.568155972281892e-06, "loss": 0.3468, "step": 10730 }, { "epoch": 1.87, "learning_rate": 5.566627926609321e-06, "loss": 0.3458, "step": 10731 }, { "epoch": 1.87, "learning_rate": 5.56510191232667e-06, "loss": 0.3466, "step": 10732 }, { "epoch": 1.87, "learning_rate": 5.563577929575273e-06, "loss": 0.3396, "step": 10733 }, { "epoch": 1.87, "learning_rate": 5.562055978496278e-06, "loss": 0.3463, "step": 10734 }, { "epoch": 1.87, "learning_rate": 5.560536059230642e-06, "loss": 0.3595, "step": 10735 }, { "epoch": 1.87, "learning_rate": 5.559018171919132e-06, "loss": 0.3449, "step": 10736 }, { "epoch": 1.87, "learning_rate": 5.557502316702333e-06, "loss": 0.3496, "step": 10737 }, { "epoch": 1.87, "learning_rate": 5.555988493720634e-06, "loss": 0.3489, "step": 10738 }, { "epoch": 1.87, "learning_rate": 5.554476703114241e-06, "loss": 0.3349, "step": 10739 }, { "epoch": 1.87, "learning_rate": 5.5529669450231755e-06, "loss": 0.3548, "step": 10740 }, { "epoch": 1.87, "learning_rate": 5.551459219587262e-06, "loss": 0.3477, "step": 10741 }, { "epoch": 1.87, "learning_rate": 5.549953526946139e-06, "loss": 0.3661, "step": 10742 }, { "epoch": 1.87, "learning_rate": 5.548449867239259e-06, "loss": 0.3433, "step": 10743 }, { "epoch": 1.87, "learning_rate": 5.546948240605886e-06, "loss": 0.3604, "step": 10744 }, { "epoch": 1.87, "learning_rate": 5.545448647185099e-06, "loss": 0.3511, "step": 10745 }, { "epoch": 1.87, "learning_rate": 5.5439510871157774e-06, "loss": 0.3536, "step": 10746 }, { "epoch": 1.87, "learning_rate": 5.542455560536628e-06, "loss": 0.3737, "step": 10747 }, { "epoch": 1.87, "learning_rate": 5.5409620675861575e-06, "loss": 0.3661, "step": 10748 }, { "epoch": 1.87, "learning_rate": 5.539470608402685e-06, "loss": 0.3549, "step": 10749 }, { "epoch": 1.87, "learning_rate": 5.5379811831243455e-06, "loss": 0.3369, "step": 10750 }, { "epoch": 1.87, "learning_rate": 5.536493791889085e-06, "loss": 0.3572, "step": 10751 }, { "epoch": 1.87, "learning_rate": 5.535008434834664e-06, "loss": 0.3454, "step": 10752 }, { "epoch": 1.87, "learning_rate": 5.533525112098646e-06, "loss": 0.349, "step": 10753 }, { "epoch": 1.87, "learning_rate": 5.5320438238184125e-06, "loss": 0.3481, "step": 10754 }, { "epoch": 1.88, "learning_rate": 5.530564570131153e-06, "loss": 0.3469, "step": 10755 }, { "epoch": 1.88, "learning_rate": 5.529087351173875e-06, "loss": 0.3591, "step": 10756 }, { "epoch": 1.88, "learning_rate": 5.52761216708339e-06, "loss": 0.3609, "step": 10757 }, { "epoch": 1.88, "learning_rate": 5.526139017996329e-06, "loss": 0.3764, "step": 10758 }, { "epoch": 1.88, "learning_rate": 5.524667904049124e-06, "loss": 0.3673, "step": 10759 }, { "epoch": 1.88, "learning_rate": 5.523198825378021e-06, "loss": 0.3411, "step": 10760 }, { "epoch": 1.88, "learning_rate": 5.52173178211909e-06, "loss": 0.3597, "step": 10761 }, { "epoch": 1.88, "learning_rate": 5.5202667744082016e-06, "loss": 0.3464, "step": 10762 }, { "epoch": 1.88, "learning_rate": 5.518803802381032e-06, "loss": 0.3532, "step": 10763 }, { "epoch": 1.88, "learning_rate": 5.517342866173086e-06, "loss": 0.3428, "step": 10764 }, { "epoch": 1.88, "learning_rate": 5.5158839659196645e-06, "loss": 0.335, "step": 10765 }, { "epoch": 1.88, "learning_rate": 5.514427101755887e-06, "loss": 0.3571, "step": 10766 }, { "epoch": 1.88, "learning_rate": 5.512972273816682e-06, "loss": 0.3494, "step": 10767 }, { "epoch": 1.88, "learning_rate": 5.511519482236793e-06, "loss": 0.3476, "step": 10768 }, { "epoch": 1.88, "learning_rate": 5.51006872715077e-06, "loss": 0.3518, "step": 10769 }, { "epoch": 1.88, "learning_rate": 5.508620008692975e-06, "loss": 0.3629, "step": 10770 }, { "epoch": 1.88, "learning_rate": 5.507173326997587e-06, "loss": 0.3406, "step": 10771 }, { "epoch": 1.88, "learning_rate": 5.505728682198592e-06, "loss": 0.3638, "step": 10772 }, { "epoch": 1.88, "learning_rate": 5.504286074429787e-06, "loss": 0.363, "step": 10773 }, { "epoch": 1.88, "learning_rate": 5.50284550382478e-06, "loss": 0.3562, "step": 10774 }, { "epoch": 1.88, "learning_rate": 5.501406970516993e-06, "loss": 0.3605, "step": 10775 }, { "epoch": 1.88, "learning_rate": 5.4999704746396565e-06, "loss": 0.357, "step": 10776 }, { "epoch": 1.88, "learning_rate": 5.498536016325813e-06, "loss": 0.3477, "step": 10777 }, { "epoch": 1.88, "learning_rate": 5.497103595708321e-06, "loss": 0.337, "step": 10778 }, { "epoch": 1.88, "learning_rate": 5.495673212919841e-06, "loss": 0.3449, "step": 10779 }, { "epoch": 1.88, "learning_rate": 5.494244868092851e-06, "loss": 0.3542, "step": 10780 }, { "epoch": 1.88, "learning_rate": 5.4928185613596435e-06, "loss": 0.3435, "step": 10781 }, { "epoch": 1.88, "learning_rate": 5.491394292852313e-06, "loss": 0.3546, "step": 10782 }, { "epoch": 1.88, "learning_rate": 5.489972062702773e-06, "loss": 0.329, "step": 10783 }, { "epoch": 1.88, "learning_rate": 5.4885518710427394e-06, "loss": 0.355, "step": 10784 }, { "epoch": 1.88, "learning_rate": 5.487133718003756e-06, "loss": 0.3552, "step": 10785 }, { "epoch": 1.88, "learning_rate": 5.4857176037171585e-06, "loss": 0.3616, "step": 10786 }, { "epoch": 1.88, "learning_rate": 5.484303528314102e-06, "loss": 0.3456, "step": 10787 }, { "epoch": 1.88, "learning_rate": 5.482891491925559e-06, "loss": 0.3533, "step": 10788 }, { "epoch": 1.88, "learning_rate": 5.4814814946823045e-06, "loss": 0.3528, "step": 10789 }, { "epoch": 1.88, "learning_rate": 5.480073536714925e-06, "loss": 0.3467, "step": 10790 }, { "epoch": 1.88, "learning_rate": 5.478667618153823e-06, "loss": 0.347, "step": 10791 }, { "epoch": 1.88, "learning_rate": 5.477263739129209e-06, "loss": 0.3556, "step": 10792 }, { "epoch": 1.88, "learning_rate": 5.475861899771106e-06, "loss": 0.3442, "step": 10793 }, { "epoch": 1.88, "learning_rate": 5.474462100209346e-06, "loss": 0.3712, "step": 10794 }, { "epoch": 1.88, "learning_rate": 5.4730643405735734e-06, "loss": 0.3488, "step": 10795 }, { "epoch": 1.88, "learning_rate": 5.471668620993245e-06, "loss": 0.349, "step": 10796 }, { "epoch": 1.88, "learning_rate": 5.470274941597626e-06, "loss": 0.3603, "step": 10797 }, { "epoch": 1.88, "learning_rate": 5.468883302515795e-06, "loss": 0.3349, "step": 10798 }, { "epoch": 1.88, "learning_rate": 5.467493703876638e-06, "loss": 0.3523, "step": 10799 }, { "epoch": 1.88, "learning_rate": 5.466106145808858e-06, "loss": 0.3766, "step": 10800 }, { "epoch": 1.88, "learning_rate": 5.464720628440966e-06, "loss": 0.3544, "step": 10801 }, { "epoch": 1.88, "learning_rate": 5.463337151901281e-06, "loss": 0.3605, "step": 10802 }, { "epoch": 1.88, "learning_rate": 5.461955716317936e-06, "loss": 0.3545, "step": 10803 }, { "epoch": 1.88, "learning_rate": 5.460576321818875e-06, "loss": 0.3489, "step": 10804 }, { "epoch": 1.88, "learning_rate": 5.459198968531853e-06, "loss": 0.3519, "step": 10805 }, { "epoch": 1.88, "learning_rate": 5.4578236565844366e-06, "loss": 0.3419, "step": 10806 }, { "epoch": 1.88, "learning_rate": 5.456450386103999e-06, "loss": 0.3489, "step": 10807 }, { "epoch": 1.88, "learning_rate": 5.455079157217729e-06, "loss": 0.3567, "step": 10808 }, { "epoch": 1.88, "learning_rate": 5.453709970052628e-06, "loss": 0.3604, "step": 10809 }, { "epoch": 1.88, "learning_rate": 5.452342824735499e-06, "loss": 0.3461, "step": 10810 }, { "epoch": 1.88, "learning_rate": 5.450977721392968e-06, "loss": 0.349, "step": 10811 }, { "epoch": 1.89, "learning_rate": 5.449614660151464e-06, "loss": 0.3611, "step": 10812 }, { "epoch": 1.89, "learning_rate": 5.448253641137226e-06, "loss": 0.3729, "step": 10813 }, { "epoch": 1.89, "learning_rate": 5.4468946644763125e-06, "loss": 0.3446, "step": 10814 }, { "epoch": 1.89, "learning_rate": 5.44553773029458e-06, "loss": 0.3575, "step": 10815 }, { "epoch": 1.89, "learning_rate": 5.444182838717707e-06, "loss": 0.3492, "step": 10816 }, { "epoch": 1.89, "learning_rate": 5.44282998987118e-06, "loss": 0.3489, "step": 10817 }, { "epoch": 1.89, "learning_rate": 5.44147918388029e-06, "loss": 0.35, "step": 10818 }, { "epoch": 1.89, "learning_rate": 5.440130420870149e-06, "loss": 0.352, "step": 10819 }, { "epoch": 1.89, "learning_rate": 5.438783700965675e-06, "loss": 0.3433, "step": 10820 }, { "epoch": 1.89, "learning_rate": 5.437439024291589e-06, "loss": 0.3413, "step": 10821 }, { "epoch": 1.89, "learning_rate": 5.4360963909724355e-06, "loss": 0.3506, "step": 10822 }, { "epoch": 1.89, "learning_rate": 5.434755801132566e-06, "loss": 0.3516, "step": 10823 }, { "epoch": 1.89, "learning_rate": 5.433417254896139e-06, "loss": 0.3421, "step": 10824 }, { "epoch": 1.89, "learning_rate": 5.4320807523871225e-06, "loss": 0.3573, "step": 10825 }, { "epoch": 1.89, "learning_rate": 5.430746293729307e-06, "loss": 0.3557, "step": 10826 }, { "epoch": 1.89, "learning_rate": 5.429413879046276e-06, "loss": 0.3501, "step": 10827 }, { "epoch": 1.89, "learning_rate": 5.428083508461437e-06, "loss": 0.3483, "step": 10828 }, { "epoch": 1.89, "learning_rate": 5.426755182098004e-06, "loss": 0.3406, "step": 10829 }, { "epoch": 1.89, "learning_rate": 5.4254289000790044e-06, "loss": 0.3426, "step": 10830 }, { "epoch": 1.89, "learning_rate": 5.424104662527271e-06, "loss": 0.3484, "step": 10831 }, { "epoch": 1.89, "learning_rate": 5.422782469565446e-06, "loss": 0.3584, "step": 10832 }, { "epoch": 1.89, "learning_rate": 5.421462321315991e-06, "loss": 0.3517, "step": 10833 }, { "epoch": 1.89, "learning_rate": 5.420144217901176e-06, "loss": 0.3593, "step": 10834 }, { "epoch": 1.89, "learning_rate": 5.4188281594430735e-06, "loss": 0.3401, "step": 10835 }, { "epoch": 1.89, "learning_rate": 5.417514146063574e-06, "loss": 0.353, "step": 10836 }, { "epoch": 1.89, "learning_rate": 5.4162021778843774e-06, "loss": 0.3458, "step": 10837 }, { "epoch": 1.89, "learning_rate": 5.414892255026993e-06, "loss": 0.3442, "step": 10838 }, { "epoch": 1.89, "learning_rate": 5.413584377612738e-06, "loss": 0.3439, "step": 10839 }, { "epoch": 1.89, "learning_rate": 5.412278545762749e-06, "loss": 0.3535, "step": 10840 }, { "epoch": 1.89, "learning_rate": 5.410974759597967e-06, "loss": 0.3477, "step": 10841 }, { "epoch": 1.89, "learning_rate": 5.409673019239138e-06, "loss": 0.3582, "step": 10842 }, { "epoch": 1.89, "learning_rate": 5.4083733248068285e-06, "loss": 0.3581, "step": 10843 }, { "epoch": 1.89, "learning_rate": 5.4070756764214156e-06, "loss": 0.3554, "step": 10844 }, { "epoch": 1.89, "learning_rate": 5.4057800742030765e-06, "loss": 0.3529, "step": 10845 }, { "epoch": 1.89, "learning_rate": 5.404486518271804e-06, "loss": 0.3451, "step": 10846 }, { "epoch": 1.89, "learning_rate": 5.40319500874741e-06, "loss": 0.3641, "step": 10847 }, { "epoch": 1.89, "learning_rate": 5.401905545749503e-06, "loss": 0.3434, "step": 10848 }, { "epoch": 1.89, "learning_rate": 5.4006181293975105e-06, "loss": 0.3506, "step": 10849 }, { "epoch": 1.89, "learning_rate": 5.399332759810668e-06, "loss": 0.3478, "step": 10850 }, { "epoch": 1.89, "learning_rate": 5.398049437108025e-06, "loss": 0.3495, "step": 10851 }, { "epoch": 1.89, "learning_rate": 5.396768161408434e-06, "loss": 0.3364, "step": 10852 }, { "epoch": 1.89, "learning_rate": 5.395488932830563e-06, "loss": 0.3617, "step": 10853 }, { "epoch": 1.89, "learning_rate": 5.394211751492889e-06, "loss": 0.352, "step": 10854 }, { "epoch": 1.89, "learning_rate": 5.392936617513705e-06, "loss": 0.3405, "step": 10855 }, { "epoch": 1.89, "learning_rate": 5.391663531011102e-06, "loss": 0.3232, "step": 10856 }, { "epoch": 1.89, "learning_rate": 5.390392492102994e-06, "loss": 0.3287, "step": 10857 }, { "epoch": 1.89, "learning_rate": 5.389123500907099e-06, "loss": 0.35, "step": 10858 }, { "epoch": 1.89, "learning_rate": 5.387856557540944e-06, "loss": 0.3395, "step": 10859 }, { "epoch": 1.89, "learning_rate": 5.38659166212187e-06, "loss": 0.3515, "step": 10860 }, { "epoch": 1.89, "learning_rate": 5.3853288147670295e-06, "loss": 0.3317, "step": 10861 }, { "epoch": 1.89, "learning_rate": 5.384068015593381e-06, "loss": 0.3547, "step": 10862 }, { "epoch": 1.89, "learning_rate": 5.382809264717694e-06, "loss": 0.3495, "step": 10863 }, { "epoch": 1.89, "learning_rate": 5.381552562256549e-06, "loss": 0.3562, "step": 10864 }, { "epoch": 1.89, "learning_rate": 5.38029790832634e-06, "loss": 0.3356, "step": 10865 }, { "epoch": 1.89, "learning_rate": 5.3790453030432675e-06, "loss": 0.3614, "step": 10866 }, { "epoch": 1.89, "learning_rate": 5.377794746523342e-06, "loss": 0.3346, "step": 10867 }, { "epoch": 1.89, "learning_rate": 5.376546238882387e-06, "loss": 0.3454, "step": 10868 }, { "epoch": 1.89, "learning_rate": 5.375299780236035e-06, "loss": 0.3422, "step": 10869 }, { "epoch": 1.9, "learning_rate": 5.374055370699725e-06, "loss": 0.3619, "step": 10870 }, { "epoch": 1.9, "learning_rate": 5.372813010388716e-06, "loss": 0.3518, "step": 10871 }, { "epoch": 1.9, "learning_rate": 5.371572699418067e-06, "loss": 0.3591, "step": 10872 }, { "epoch": 1.9, "learning_rate": 5.37033443790265e-06, "loss": 0.3519, "step": 10873 }, { "epoch": 1.9, "learning_rate": 5.369098225957153e-06, "loss": 0.3557, "step": 10874 }, { "epoch": 1.9, "learning_rate": 5.367864063696065e-06, "loss": 0.3602, "step": 10875 }, { "epoch": 1.9, "learning_rate": 5.366631951233688e-06, "loss": 0.3394, "step": 10876 }, { "epoch": 1.9, "learning_rate": 5.3654018886841404e-06, "loss": 0.3545, "step": 10877 }, { "epoch": 1.9, "learning_rate": 5.364173876161349e-06, "loss": 0.36, "step": 10878 }, { "epoch": 1.9, "learning_rate": 5.3629479137790395e-06, "loss": 0.35, "step": 10879 }, { "epoch": 1.9, "learning_rate": 5.361724001650761e-06, "loss": 0.3544, "step": 10880 }, { "epoch": 1.9, "learning_rate": 5.3605021398898666e-06, "loss": 0.338, "step": 10881 }, { "epoch": 1.9, "learning_rate": 5.359282328609523e-06, "loss": 0.3381, "step": 10882 }, { "epoch": 1.9, "learning_rate": 5.358064567922701e-06, "loss": 0.3659, "step": 10883 }, { "epoch": 1.9, "learning_rate": 5.3568488579421876e-06, "loss": 0.3206, "step": 10884 }, { "epoch": 1.9, "learning_rate": 5.355635198780577e-06, "loss": 0.3454, "step": 10885 }, { "epoch": 1.9, "learning_rate": 5.354423590550278e-06, "loss": 0.3423, "step": 10886 }, { "epoch": 1.9, "learning_rate": 5.353214033363495e-06, "loss": 0.3503, "step": 10887 }, { "epoch": 1.9, "learning_rate": 5.352006527332261e-06, "loss": 0.3576, "step": 10888 }, { "epoch": 1.9, "learning_rate": 5.350801072568409e-06, "loss": 0.366, "step": 10889 }, { "epoch": 1.9, "learning_rate": 5.349597669183588e-06, "loss": 0.3438, "step": 10890 }, { "epoch": 1.9, "learning_rate": 5.3483963172892435e-06, "loss": 0.3469, "step": 10891 }, { "epoch": 1.9, "learning_rate": 5.34719701699665e-06, "loss": 0.3625, "step": 10892 }, { "epoch": 1.9, "learning_rate": 5.345999768416874e-06, "loss": 0.3319, "step": 10893 }, { "epoch": 1.9, "learning_rate": 5.344804571660806e-06, "loss": 0.3384, "step": 10894 }, { "epoch": 1.9, "learning_rate": 5.343611426839141e-06, "loss": 0.3546, "step": 10895 }, { "epoch": 1.9, "learning_rate": 5.342420334062379e-06, "loss": 0.347, "step": 10896 }, { "epoch": 1.9, "learning_rate": 5.341231293440839e-06, "loss": 0.3605, "step": 10897 }, { "epoch": 1.9, "learning_rate": 5.340044305084646e-06, "loss": 0.3363, "step": 10898 }, { "epoch": 1.9, "learning_rate": 5.3388593691037295e-06, "loss": 0.3564, "step": 10899 }, { "epoch": 1.9, "learning_rate": 5.337676485607841e-06, "loss": 0.3399, "step": 10900 }, { "epoch": 1.9, "learning_rate": 5.336495654706529e-06, "loss": 0.3414, "step": 10901 }, { "epoch": 1.9, "learning_rate": 5.335316876509162e-06, "loss": 0.3568, "step": 10902 }, { "epoch": 1.9, "learning_rate": 5.334140151124914e-06, "loss": 0.3546, "step": 10903 }, { "epoch": 1.9, "learning_rate": 5.3329654786627626e-06, "loss": 0.3539, "step": 10904 }, { "epoch": 1.9, "learning_rate": 5.331792859231507e-06, "loss": 0.3389, "step": 10905 }, { "epoch": 1.9, "learning_rate": 5.3306222929397545e-06, "loss": 0.3452, "step": 10906 }, { "epoch": 1.9, "learning_rate": 5.329453779895912e-06, "loss": 0.3459, "step": 10907 }, { "epoch": 1.9, "learning_rate": 5.328287320208207e-06, "loss": 0.3585, "step": 10908 }, { "epoch": 1.9, "learning_rate": 5.327122913984672e-06, "loss": 0.3564, "step": 10909 }, { "epoch": 1.9, "learning_rate": 5.325960561333148e-06, "loss": 0.3502, "step": 10910 }, { "epoch": 1.9, "learning_rate": 5.324800262361287e-06, "loss": 0.3445, "step": 10911 }, { "epoch": 1.9, "learning_rate": 5.32364201717656e-06, "loss": 0.3399, "step": 10912 }, { "epoch": 1.9, "learning_rate": 5.322485825886232e-06, "loss": 0.3436, "step": 10913 }, { "epoch": 1.9, "learning_rate": 5.321331688597385e-06, "loss": 0.3434, "step": 10914 }, { "epoch": 1.9, "learning_rate": 5.3201796054169155e-06, "loss": 0.346, "step": 10915 }, { "epoch": 1.9, "learning_rate": 5.319029576451523e-06, "loss": 0.3477, "step": 10916 }, { "epoch": 1.9, "learning_rate": 5.317881601807718e-06, "loss": 0.337, "step": 10917 }, { "epoch": 1.9, "learning_rate": 5.316735681591823e-06, "loss": 0.3497, "step": 10918 }, { "epoch": 1.9, "learning_rate": 5.315591815909969e-06, "loss": 0.3358, "step": 10919 }, { "epoch": 1.9, "learning_rate": 5.314450004868095e-06, "loss": 0.3521, "step": 10920 }, { "epoch": 1.9, "learning_rate": 5.313310248571953e-06, "loss": 0.3524, "step": 10921 }, { "epoch": 1.9, "learning_rate": 5.312172547127105e-06, "loss": 0.3657, "step": 10922 }, { "epoch": 1.9, "learning_rate": 5.311036900638917e-06, "loss": 0.3504, "step": 10923 }, { "epoch": 1.9, "learning_rate": 5.309903309212572e-06, "loss": 0.3338, "step": 10924 }, { "epoch": 1.9, "learning_rate": 5.308771772953056e-06, "loss": 0.3533, "step": 10925 }, { "epoch": 1.9, "learning_rate": 5.307642291965168e-06, "loss": 0.3512, "step": 10926 }, { "epoch": 1.91, "learning_rate": 5.3065148663535205e-06, "loss": 0.3457, "step": 10927 }, { "epoch": 1.91, "learning_rate": 5.3053894962225275e-06, "loss": 0.3422, "step": 10928 }, { "epoch": 1.91, "learning_rate": 5.304266181676416e-06, "loss": 0.3387, "step": 10929 }, { "epoch": 1.91, "learning_rate": 5.303144922819229e-06, "loss": 0.3498, "step": 10930 }, { "epoch": 1.91, "learning_rate": 5.302025719754809e-06, "loss": 0.3451, "step": 10931 }, { "epoch": 1.91, "learning_rate": 5.300908572586811e-06, "loss": 0.3528, "step": 10932 }, { "epoch": 1.91, "learning_rate": 5.299793481418706e-06, "loss": 0.3672, "step": 10933 }, { "epoch": 1.91, "learning_rate": 5.2986804463537694e-06, "loss": 0.3552, "step": 10934 }, { "epoch": 1.91, "learning_rate": 5.29756946749508e-06, "loss": 0.3503, "step": 10935 }, { "epoch": 1.91, "learning_rate": 5.296460544945539e-06, "loss": 0.3514, "step": 10936 }, { "epoch": 1.91, "learning_rate": 5.2953536788078476e-06, "loss": 0.3571, "step": 10937 }, { "epoch": 1.91, "learning_rate": 5.294248869184523e-06, "loss": 0.352, "step": 10938 }, { "epoch": 1.91, "learning_rate": 5.293146116177884e-06, "loss": 0.3402, "step": 10939 }, { "epoch": 1.91, "learning_rate": 5.292045419890069e-06, "loss": 0.3434, "step": 10940 }, { "epoch": 1.91, "learning_rate": 5.290946780423018e-06, "loss": 0.3556, "step": 10941 }, { "epoch": 1.91, "learning_rate": 5.2898501978784795e-06, "loss": 0.3662, "step": 10942 }, { "epoch": 1.91, "learning_rate": 5.288755672358022e-06, "loss": 0.3334, "step": 10943 }, { "epoch": 1.91, "learning_rate": 5.287663203963012e-06, "loss": 0.3535, "step": 10944 }, { "epoch": 1.91, "learning_rate": 5.28657279279463e-06, "loss": 0.3548, "step": 10945 }, { "epoch": 1.91, "learning_rate": 5.285484438953866e-06, "loss": 0.3587, "step": 10946 }, { "epoch": 1.91, "learning_rate": 5.284398142541523e-06, "loss": 0.3562, "step": 10947 }, { "epoch": 1.91, "learning_rate": 5.283313903658206e-06, "loss": 0.3477, "step": 10948 }, { "epoch": 1.91, "learning_rate": 5.282231722404332e-06, "loss": 0.3351, "step": 10949 }, { "epoch": 1.91, "learning_rate": 5.2811515988801345e-06, "loss": 0.3326, "step": 10950 }, { "epoch": 1.91, "learning_rate": 5.280073533185647e-06, "loss": 0.3379, "step": 10951 }, { "epoch": 1.91, "learning_rate": 5.278997525420714e-06, "loss": 0.338, "step": 10952 }, { "epoch": 1.91, "learning_rate": 5.277923575684994e-06, "loss": 0.3301, "step": 10953 }, { "epoch": 1.91, "learning_rate": 5.276851684077952e-06, "loss": 0.3475, "step": 10954 }, { "epoch": 1.91, "learning_rate": 5.275781850698867e-06, "loss": 0.352, "step": 10955 }, { "epoch": 1.91, "learning_rate": 5.274714075646815e-06, "loss": 0.3433, "step": 10956 }, { "epoch": 1.91, "learning_rate": 5.273648359020695e-06, "loss": 0.345, "step": 10957 }, { "epoch": 1.91, "learning_rate": 5.2725847009192104e-06, "loss": 0.3474, "step": 10958 }, { "epoch": 1.91, "learning_rate": 5.271523101440869e-06, "loss": 0.3569, "step": 10959 }, { "epoch": 1.91, "learning_rate": 5.270463560683992e-06, "loss": 0.3461, "step": 10960 }, { "epoch": 1.91, "learning_rate": 5.269406078746715e-06, "loss": 0.3494, "step": 10961 }, { "epoch": 1.91, "learning_rate": 5.2683506557269775e-06, "loss": 0.3472, "step": 10962 }, { "epoch": 1.91, "learning_rate": 5.267297291722528e-06, "loss": 0.3593, "step": 10963 }, { "epoch": 1.91, "learning_rate": 5.266245986830924e-06, "loss": 0.3617, "step": 10964 }, { "epoch": 1.91, "learning_rate": 5.265196741149532e-06, "loss": 0.3424, "step": 10965 }, { "epoch": 1.91, "learning_rate": 5.264149554775535e-06, "loss": 0.3218, "step": 10966 }, { "epoch": 1.91, "learning_rate": 5.263104427805915e-06, "loss": 0.3668, "step": 10967 }, { "epoch": 1.91, "learning_rate": 5.262061360337469e-06, "loss": 0.3558, "step": 10968 }, { "epoch": 1.91, "learning_rate": 5.2610203524668035e-06, "loss": 0.3572, "step": 10969 }, { "epoch": 1.91, "learning_rate": 5.25998140429033e-06, "loss": 0.3464, "step": 10970 }, { "epoch": 1.91, "learning_rate": 5.258944515904276e-06, "loss": 0.3567, "step": 10971 }, { "epoch": 1.91, "learning_rate": 5.257909687404673e-06, "loss": 0.338, "step": 10972 }, { "epoch": 1.91, "learning_rate": 5.256876918887358e-06, "loss": 0.3456, "step": 10973 }, { "epoch": 1.91, "learning_rate": 5.255846210447991e-06, "loss": 0.3539, "step": 10974 }, { "epoch": 1.91, "learning_rate": 5.254817562182027e-06, "loss": 0.3476, "step": 10975 }, { "epoch": 1.91, "learning_rate": 5.253790974184737e-06, "loss": 0.3365, "step": 10976 }, { "epoch": 1.91, "learning_rate": 5.252766446551199e-06, "loss": 0.3513, "step": 10977 }, { "epoch": 1.91, "learning_rate": 5.251743979376302e-06, "loss": 0.3524, "step": 10978 }, { "epoch": 1.91, "learning_rate": 5.250723572754744e-06, "loss": 0.3566, "step": 10979 }, { "epoch": 1.91, "learning_rate": 5.2497052267810305e-06, "loss": 0.332, "step": 10980 }, { "epoch": 1.91, "learning_rate": 5.248688941549476e-06, "loss": 0.3468, "step": 10981 }, { "epoch": 1.91, "learning_rate": 5.247674717154208e-06, "loss": 0.356, "step": 10982 }, { "epoch": 1.91, "learning_rate": 5.2466625536891554e-06, "loss": 0.3397, "step": 10983 }, { "epoch": 1.92, "learning_rate": 5.2456524512480655e-06, "loss": 0.3536, "step": 10984 }, { "epoch": 1.92, "learning_rate": 5.244644409924493e-06, "loss": 0.3507, "step": 10985 }, { "epoch": 1.92, "learning_rate": 5.24363842981179e-06, "loss": 0.3447, "step": 10986 }, { "epoch": 1.92, "learning_rate": 5.242634511003131e-06, "loss": 0.3727, "step": 10987 }, { "epoch": 1.92, "learning_rate": 5.241632653591501e-06, "loss": 0.3542, "step": 10988 }, { "epoch": 1.92, "learning_rate": 5.240632857669683e-06, "loss": 0.3553, "step": 10989 }, { "epoch": 1.92, "learning_rate": 5.239635123330274e-06, "loss": 0.3508, "step": 10990 }, { "epoch": 1.92, "learning_rate": 5.238639450665682e-06, "loss": 0.358, "step": 10991 }, { "epoch": 1.92, "learning_rate": 5.2376458397681195e-06, "loss": 0.3534, "step": 10992 }, { "epoch": 1.92, "learning_rate": 5.236654290729616e-06, "loss": 0.3759, "step": 10993 }, { "epoch": 1.92, "learning_rate": 5.235664803642005e-06, "loss": 0.3396, "step": 10994 }, { "epoch": 1.92, "learning_rate": 5.234677378596926e-06, "loss": 0.3568, "step": 10995 }, { "epoch": 1.92, "learning_rate": 5.233692015685833e-06, "loss": 0.3557, "step": 10996 }, { "epoch": 1.92, "learning_rate": 5.232708714999986e-06, "loss": 0.3494, "step": 10997 }, { "epoch": 1.92, "learning_rate": 5.23172747663045e-06, "loss": 0.3493, "step": 10998 }, { "epoch": 1.92, "learning_rate": 5.230748300668113e-06, "loss": 0.3351, "step": 10999 }, { "epoch": 1.92, "learning_rate": 5.229771187203661e-06, "loss": 0.3437, "step": 11000 }, { "epoch": 1.92, "learning_rate": 5.228796136327583e-06, "loss": 0.3335, "step": 11001 }, { "epoch": 1.92, "learning_rate": 5.227823148130192e-06, "loss": 0.3514, "step": 11002 }, { "epoch": 1.92, "learning_rate": 5.2268522227016e-06, "loss": 0.3617, "step": 11003 }, { "epoch": 1.92, "learning_rate": 5.225883360131728e-06, "loss": 0.3347, "step": 11004 }, { "epoch": 1.92, "learning_rate": 5.224916560510316e-06, "loss": 0.3467, "step": 11005 }, { "epoch": 1.92, "learning_rate": 5.223951823926898e-06, "loss": 0.3498, "step": 11006 }, { "epoch": 1.92, "learning_rate": 5.22298915047083e-06, "loss": 0.3344, "step": 11007 }, { "epoch": 1.92, "learning_rate": 5.222028540231267e-06, "loss": 0.3527, "step": 11008 }, { "epoch": 1.92, "learning_rate": 5.221069993297178e-06, "loss": 0.351, "step": 11009 }, { "epoch": 1.92, "learning_rate": 5.220113509757344e-06, "loss": 0.3309, "step": 11010 }, { "epoch": 1.92, "learning_rate": 5.219159089700345e-06, "loss": 0.3455, "step": 11011 }, { "epoch": 1.92, "learning_rate": 5.218206733214582e-06, "loss": 0.3484, "step": 11012 }, { "epoch": 1.92, "learning_rate": 5.217256440388255e-06, "loss": 0.3475, "step": 11013 }, { "epoch": 1.92, "learning_rate": 5.216308211309376e-06, "loss": 0.3531, "step": 11014 }, { "epoch": 1.92, "learning_rate": 5.21536204606577e-06, "loss": 0.3484, "step": 11015 }, { "epoch": 1.92, "learning_rate": 5.214417944745066e-06, "loss": 0.3494, "step": 11016 }, { "epoch": 1.92, "learning_rate": 5.213475907434705e-06, "loss": 0.3502, "step": 11017 }, { "epoch": 1.92, "learning_rate": 5.212535934221929e-06, "loss": 0.3388, "step": 11018 }, { "epoch": 1.92, "learning_rate": 5.211598025193802e-06, "loss": 0.3442, "step": 11019 }, { "epoch": 1.92, "learning_rate": 5.2106621804371854e-06, "loss": 0.3331, "step": 11020 }, { "epoch": 1.92, "learning_rate": 5.209728400038756e-06, "loss": 0.3601, "step": 11021 }, { "epoch": 1.92, "learning_rate": 5.208796684084995e-06, "loss": 0.3581, "step": 11022 }, { "epoch": 1.92, "learning_rate": 5.207867032662196e-06, "loss": 0.3547, "step": 11023 }, { "epoch": 1.92, "learning_rate": 5.206939445856461e-06, "loss": 0.3513, "step": 11024 }, { "epoch": 1.92, "learning_rate": 5.206013923753699e-06, "loss": 0.3676, "step": 11025 }, { "epoch": 1.92, "learning_rate": 5.2050904664396265e-06, "loss": 0.3312, "step": 11026 }, { "epoch": 1.92, "learning_rate": 5.204169073999773e-06, "loss": 0.3344, "step": 11027 }, { "epoch": 1.92, "learning_rate": 5.2032497465194745e-06, "loss": 0.3485, "step": 11028 }, { "epoch": 1.92, "learning_rate": 5.202332484083875e-06, "loss": 0.3589, "step": 11029 }, { "epoch": 1.92, "learning_rate": 5.201417286777928e-06, "loss": 0.3471, "step": 11030 }, { "epoch": 1.92, "learning_rate": 5.200504154686395e-06, "loss": 0.3502, "step": 11031 }, { "epoch": 1.92, "learning_rate": 5.199593087893849e-06, "loss": 0.3409, "step": 11032 }, { "epoch": 1.92, "learning_rate": 5.1986840864846705e-06, "loss": 0.3516, "step": 11033 }, { "epoch": 1.92, "learning_rate": 5.1977771505430455e-06, "loss": 0.3573, "step": 11034 }, { "epoch": 1.92, "learning_rate": 5.196872280152971e-06, "loss": 0.3383, "step": 11035 }, { "epoch": 1.92, "learning_rate": 5.195969475398255e-06, "loss": 0.3557, "step": 11036 }, { "epoch": 1.92, "learning_rate": 5.195068736362509e-06, "loss": 0.3699, "step": 11037 }, { "epoch": 1.92, "learning_rate": 5.1941700631291595e-06, "loss": 0.3529, "step": 11038 }, { "epoch": 1.92, "learning_rate": 5.193273455781434e-06, "loss": 0.3431, "step": 11039 }, { "epoch": 1.92, "learning_rate": 5.19237891440238e-06, "loss": 0.3536, "step": 11040 }, { "epoch": 1.92, "learning_rate": 5.19148643907484e-06, "loss": 0.3584, "step": 11041 }, { "epoch": 1.93, "learning_rate": 5.190596029881474e-06, "loss": 0.3453, "step": 11042 }, { "epoch": 1.93, "learning_rate": 5.189707686904752e-06, "loss": 0.3646, "step": 11043 }, { "epoch": 1.93, "learning_rate": 5.188821410226944e-06, "loss": 0.3624, "step": 11044 }, { "epoch": 1.93, "learning_rate": 5.187937199930136e-06, "loss": 0.3418, "step": 11045 }, { "epoch": 1.93, "learning_rate": 5.187055056096221e-06, "loss": 0.3724, "step": 11046 }, { "epoch": 1.93, "learning_rate": 5.186174978806899e-06, "loss": 0.3569, "step": 11047 }, { "epoch": 1.93, "learning_rate": 5.185296968143677e-06, "loss": 0.3737, "step": 11048 }, { "epoch": 1.93, "learning_rate": 5.184421024187878e-06, "loss": 0.3409, "step": 11049 }, { "epoch": 1.93, "learning_rate": 5.183547147020626e-06, "loss": 0.3412, "step": 11050 }, { "epoch": 1.93, "learning_rate": 5.18267533672286e-06, "loss": 0.3468, "step": 11051 }, { "epoch": 1.93, "learning_rate": 5.181805593375317e-06, "loss": 0.3416, "step": 11052 }, { "epoch": 1.93, "learning_rate": 5.1809379170585546e-06, "loss": 0.3571, "step": 11053 }, { "epoch": 1.93, "learning_rate": 5.180072307852936e-06, "loss": 0.3529, "step": 11054 }, { "epoch": 1.93, "learning_rate": 5.179208765838625e-06, "loss": 0.3545, "step": 11055 }, { "epoch": 1.93, "learning_rate": 5.178347291095602e-06, "loss": 0.3584, "step": 11056 }, { "epoch": 1.93, "learning_rate": 5.177487883703653e-06, "loss": 0.3587, "step": 11057 }, { "epoch": 1.93, "learning_rate": 5.176630543742376e-06, "loss": 0.3386, "step": 11058 }, { "epoch": 1.93, "learning_rate": 5.175775271291172e-06, "loss": 0.3621, "step": 11059 }, { "epoch": 1.93, "learning_rate": 5.1749220664292555e-06, "loss": 0.3297, "step": 11060 }, { "epoch": 1.93, "learning_rate": 5.174070929235646e-06, "loss": 0.3588, "step": 11061 }, { "epoch": 1.93, "learning_rate": 5.173221859789172e-06, "loss": 0.3364, "step": 11062 }, { "epoch": 1.93, "learning_rate": 5.172374858168471e-06, "loss": 0.3437, "step": 11063 }, { "epoch": 1.93, "learning_rate": 5.171529924451991e-06, "loss": 0.3497, "step": 11064 }, { "epoch": 1.93, "learning_rate": 5.170687058717986e-06, "loss": 0.3418, "step": 11065 }, { "epoch": 1.93, "learning_rate": 5.169846261044519e-06, "loss": 0.36, "step": 11066 }, { "epoch": 1.93, "learning_rate": 5.169007531509462e-06, "loss": 0.3561, "step": 11067 }, { "epoch": 1.93, "learning_rate": 5.168170870190496e-06, "loss": 0.3497, "step": 11068 }, { "epoch": 1.93, "learning_rate": 5.167336277165107e-06, "loss": 0.3428, "step": 11069 }, { "epoch": 1.93, "learning_rate": 5.1665037525105945e-06, "loss": 0.3702, "step": 11070 }, { "epoch": 1.93, "learning_rate": 5.165673296304063e-06, "loss": 0.349, "step": 11071 }, { "epoch": 1.93, "learning_rate": 5.164844908622424e-06, "loss": 0.3553, "step": 11072 }, { "epoch": 1.93, "learning_rate": 5.164018589542403e-06, "loss": 0.3587, "step": 11073 }, { "epoch": 1.93, "learning_rate": 5.163194339140532e-06, "loss": 0.3506, "step": 11074 }, { "epoch": 1.93, "learning_rate": 5.162372157493145e-06, "loss": 0.3289, "step": 11075 }, { "epoch": 1.93, "learning_rate": 5.161552044676393e-06, "loss": 0.3484, "step": 11076 }, { "epoch": 1.93, "learning_rate": 5.160734000766232e-06, "loss": 0.3394, "step": 11077 }, { "epoch": 1.93, "learning_rate": 5.159918025838425e-06, "loss": 0.3626, "step": 11078 }, { "epoch": 1.93, "learning_rate": 5.159104119968548e-06, "loss": 0.3455, "step": 11079 }, { "epoch": 1.93, "learning_rate": 5.158292283231974e-06, "loss": 0.3434, "step": 11080 }, { "epoch": 1.93, "learning_rate": 5.1574825157039e-06, "loss": 0.3447, "step": 11081 }, { "epoch": 1.93, "learning_rate": 5.156674817459323e-06, "loss": 0.3499, "step": 11082 }, { "epoch": 1.93, "learning_rate": 5.155869188573045e-06, "loss": 0.3455, "step": 11083 }, { "epoch": 1.93, "learning_rate": 5.155065629119683e-06, "loss": 0.3423, "step": 11084 }, { "epoch": 1.93, "learning_rate": 5.154264139173661e-06, "loss": 0.3532, "step": 11085 }, { "epoch": 1.93, "learning_rate": 5.153464718809208e-06, "loss": 0.3488, "step": 11086 }, { "epoch": 1.93, "learning_rate": 5.152667368100364e-06, "loss": 0.3589, "step": 11087 }, { "epoch": 1.93, "learning_rate": 5.151872087120978e-06, "loss": 0.3521, "step": 11088 }, { "epoch": 1.93, "learning_rate": 5.151078875944706e-06, "loss": 0.3455, "step": 11089 }, { "epoch": 1.93, "learning_rate": 5.15028773464501e-06, "loss": 0.3568, "step": 11090 }, { "epoch": 1.93, "learning_rate": 5.149498663295163e-06, "loss": 0.3524, "step": 11091 }, { "epoch": 1.93, "learning_rate": 5.148711661968247e-06, "loss": 0.3526, "step": 11092 }, { "epoch": 1.93, "learning_rate": 5.147926730737155e-06, "loss": 0.3477, "step": 11093 }, { "epoch": 1.93, "learning_rate": 5.147143869674575e-06, "loss": 0.3423, "step": 11094 }, { "epoch": 1.93, "learning_rate": 5.146363078853024e-06, "loss": 0.3376, "step": 11095 }, { "epoch": 1.93, "learning_rate": 5.1455843583448065e-06, "loss": 0.3614, "step": 11096 }, { "epoch": 1.93, "learning_rate": 5.14480770822205e-06, "loss": 0.3476, "step": 11097 }, { "epoch": 1.93, "learning_rate": 5.1440331285566846e-06, "loss": 0.3444, "step": 11098 }, { "epoch": 1.94, "learning_rate": 5.143260619420448e-06, "loss": 0.3434, "step": 11099 }, { "epoch": 1.94, "learning_rate": 5.142490180884889e-06, "loss": 0.3353, "step": 11100 }, { "epoch": 1.94, "learning_rate": 5.1417218130213584e-06, "loss": 0.3403, "step": 11101 }, { "epoch": 1.94, "learning_rate": 5.1409555159010255e-06, "loss": 0.356, "step": 11102 }, { "epoch": 1.94, "learning_rate": 5.140191289594855e-06, "loss": 0.3466, "step": 11103 }, { "epoch": 1.94, "learning_rate": 5.139429134173633e-06, "loss": 0.3526, "step": 11104 }, { "epoch": 1.94, "learning_rate": 5.138669049707948e-06, "loss": 0.3525, "step": 11105 }, { "epoch": 1.94, "learning_rate": 5.137911036268191e-06, "loss": 0.3493, "step": 11106 }, { "epoch": 1.94, "learning_rate": 5.137155093924572e-06, "loss": 0.3459, "step": 11107 }, { "epoch": 1.94, "learning_rate": 5.136401222747098e-06, "loss": 0.3469, "step": 11108 }, { "epoch": 1.94, "learning_rate": 5.135649422805596e-06, "loss": 0.3461, "step": 11109 }, { "epoch": 1.94, "learning_rate": 5.134899694169689e-06, "loss": 0.3483, "step": 11110 }, { "epoch": 1.94, "learning_rate": 5.134152036908816e-06, "loss": 0.3585, "step": 11111 }, { "epoch": 1.94, "learning_rate": 5.1334064510922254e-06, "loss": 0.3519, "step": 11112 }, { "epoch": 1.94, "learning_rate": 5.132662936788971e-06, "loss": 0.3578, "step": 11113 }, { "epoch": 1.94, "learning_rate": 5.131921494067908e-06, "loss": 0.3433, "step": 11114 }, { "epoch": 1.94, "learning_rate": 5.131182122997709e-06, "loss": 0.3359, "step": 11115 }, { "epoch": 1.94, "learning_rate": 5.1304448236468576e-06, "loss": 0.3512, "step": 11116 }, { "epoch": 1.94, "learning_rate": 5.129709596083631e-06, "loss": 0.3587, "step": 11117 }, { "epoch": 1.94, "learning_rate": 5.128976440376129e-06, "loss": 0.3577, "step": 11118 }, { "epoch": 1.94, "learning_rate": 5.128245356592253e-06, "loss": 0.3412, "step": 11119 }, { "epoch": 1.94, "learning_rate": 5.127516344799713e-06, "loss": 0.3446, "step": 11120 }, { "epoch": 1.94, "learning_rate": 5.126789405066026e-06, "loss": 0.3558, "step": 11121 }, { "epoch": 1.94, "learning_rate": 5.126064537458521e-06, "loss": 0.3684, "step": 11122 }, { "epoch": 1.94, "learning_rate": 5.12534174204433e-06, "loss": 0.3396, "step": 11123 }, { "epoch": 1.94, "learning_rate": 5.1246210188903975e-06, "loss": 0.3611, "step": 11124 }, { "epoch": 1.94, "learning_rate": 5.1239023680634736e-06, "loss": 0.3682, "step": 11125 }, { "epoch": 1.94, "learning_rate": 5.1231857896301175e-06, "loss": 0.3616, "step": 11126 }, { "epoch": 1.94, "learning_rate": 5.122471283656698e-06, "loss": 0.3511, "step": 11127 }, { "epoch": 1.94, "learning_rate": 5.121758850209384e-06, "loss": 0.3509, "step": 11128 }, { "epoch": 1.94, "learning_rate": 5.121048489354166e-06, "loss": 0.3554, "step": 11129 }, { "epoch": 1.94, "learning_rate": 5.120340201156833e-06, "loss": 0.3406, "step": 11130 }, { "epoch": 1.94, "learning_rate": 5.119633985682978e-06, "loss": 0.3641, "step": 11131 }, { "epoch": 1.94, "learning_rate": 5.118929842998016e-06, "loss": 0.3662, "step": 11132 }, { "epoch": 1.94, "learning_rate": 5.118227773167156e-06, "loss": 0.3357, "step": 11133 }, { "epoch": 1.94, "learning_rate": 5.117527776255429e-06, "loss": 0.3436, "step": 11134 }, { "epoch": 1.94, "learning_rate": 5.116829852327659e-06, "loss": 0.3484, "step": 11135 }, { "epoch": 1.94, "learning_rate": 5.116134001448486e-06, "loss": 0.3531, "step": 11136 }, { "epoch": 1.94, "learning_rate": 5.11544022368236e-06, "loss": 0.3515, "step": 11137 }, { "epoch": 1.94, "learning_rate": 5.114748519093535e-06, "loss": 0.3596, "step": 11138 }, { "epoch": 1.94, "learning_rate": 5.114058887746075e-06, "loss": 0.3581, "step": 11139 }, { "epoch": 1.94, "learning_rate": 5.113371329703849e-06, "loss": 0.362, "step": 11140 }, { "epoch": 1.94, "learning_rate": 5.11268584503054e-06, "loss": 0.3479, "step": 11141 }, { "epoch": 1.94, "learning_rate": 5.11200243378963e-06, "loss": 0.3505, "step": 11142 }, { "epoch": 1.94, "learning_rate": 5.111321096044417e-06, "loss": 0.3505, "step": 11143 }, { "epoch": 1.94, "learning_rate": 5.110641831858005e-06, "loss": 0.3477, "step": 11144 }, { "epoch": 1.94, "learning_rate": 5.109964641293301e-06, "loss": 0.3616, "step": 11145 }, { "epoch": 1.94, "learning_rate": 5.109289524413031e-06, "loss": 0.3499, "step": 11146 }, { "epoch": 1.94, "learning_rate": 5.108616481279714e-06, "loss": 0.3554, "step": 11147 }, { "epoch": 1.94, "learning_rate": 5.1079455119556915e-06, "loss": 0.351, "step": 11148 }, { "epoch": 1.94, "learning_rate": 5.1072766165031e-06, "loss": 0.3643, "step": 11149 }, { "epoch": 1.94, "learning_rate": 5.106609794983896e-06, "loss": 0.3685, "step": 11150 }, { "epoch": 1.94, "learning_rate": 5.105945047459835e-06, "loss": 0.3491, "step": 11151 }, { "epoch": 1.94, "learning_rate": 5.1052823739924824e-06, "loss": 0.3608, "step": 11152 }, { "epoch": 1.94, "learning_rate": 5.104621774643215e-06, "loss": 0.3462, "step": 11153 }, { "epoch": 1.94, "learning_rate": 5.103963249473216e-06, "loss": 0.3487, "step": 11154 }, { "epoch": 1.94, "learning_rate": 5.1033067985434735e-06, "loss": 0.353, "step": 11155 }, { "epoch": 1.95, "learning_rate": 5.102652421914784e-06, "loss": 0.3649, "step": 11156 }, { "epoch": 1.95, "learning_rate": 5.102000119647758e-06, "loss": 0.3514, "step": 11157 }, { "epoch": 1.95, "learning_rate": 5.101349891802806e-06, "loss": 0.3394, "step": 11158 }, { "epoch": 1.95, "learning_rate": 5.10070173844015e-06, "loss": 0.3319, "step": 11159 }, { "epoch": 1.95, "learning_rate": 5.100055659619822e-06, "loss": 0.3378, "step": 11160 }, { "epoch": 1.95, "learning_rate": 5.099411655401658e-06, "loss": 0.3577, "step": 11161 }, { "epoch": 1.95, "learning_rate": 5.098769725845302e-06, "loss": 0.3485, "step": 11162 }, { "epoch": 1.95, "learning_rate": 5.098129871010209e-06, "loss": 0.3433, "step": 11163 }, { "epoch": 1.95, "learning_rate": 5.09749209095564e-06, "loss": 0.3399, "step": 11164 }, { "epoch": 1.95, "learning_rate": 5.096856385740662e-06, "loss": 0.3324, "step": 11165 }, { "epoch": 1.95, "learning_rate": 5.0962227554241535e-06, "loss": 0.3457, "step": 11166 }, { "epoch": 1.95, "learning_rate": 5.095591200064802e-06, "loss": 0.3472, "step": 11167 }, { "epoch": 1.95, "learning_rate": 5.094961719721094e-06, "loss": 0.3503, "step": 11168 }, { "epoch": 1.95, "learning_rate": 5.094334314451332e-06, "loss": 0.3356, "step": 11169 }, { "epoch": 1.95, "learning_rate": 5.093708984313623e-06, "loss": 0.3468, "step": 11170 }, { "epoch": 1.95, "learning_rate": 5.093085729365887e-06, "loss": 0.361, "step": 11171 }, { "epoch": 1.95, "learning_rate": 5.092464549665842e-06, "loss": 0.3559, "step": 11172 }, { "epoch": 1.95, "learning_rate": 5.091845445271023e-06, "loss": 0.3564, "step": 11173 }, { "epoch": 1.95, "learning_rate": 5.09122841623877e-06, "loss": 0.3452, "step": 11174 }, { "epoch": 1.95, "learning_rate": 5.090613462626226e-06, "loss": 0.3401, "step": 11175 }, { "epoch": 1.95, "learning_rate": 5.09000058449035e-06, "loss": 0.3484, "step": 11176 }, { "epoch": 1.95, "learning_rate": 5.089389781887903e-06, "loss": 0.3551, "step": 11177 }, { "epoch": 1.95, "learning_rate": 5.088781054875454e-06, "loss": 0.3654, "step": 11178 }, { "epoch": 1.95, "learning_rate": 5.088174403509382e-06, "loss": 0.3504, "step": 11179 }, { "epoch": 1.95, "learning_rate": 5.087569827845872e-06, "loss": 0.3696, "step": 11180 }, { "epoch": 1.95, "learning_rate": 5.086967327940923e-06, "loss": 0.3402, "step": 11181 }, { "epoch": 1.95, "learning_rate": 5.086366903850329e-06, "loss": 0.3499, "step": 11182 }, { "epoch": 1.95, "learning_rate": 5.085768555629703e-06, "loss": 0.3623, "step": 11183 }, { "epoch": 1.95, "learning_rate": 5.085172283334459e-06, "loss": 0.3395, "step": 11184 }, { "epoch": 1.95, "learning_rate": 5.084578087019827e-06, "loss": 0.3432, "step": 11185 }, { "epoch": 1.95, "learning_rate": 5.083985966740834e-06, "loss": 0.3589, "step": 11186 }, { "epoch": 1.95, "learning_rate": 5.08339592255232e-06, "loss": 0.353, "step": 11187 }, { "epoch": 1.95, "learning_rate": 5.082807954508938e-06, "loss": 0.3486, "step": 11188 }, { "epoch": 1.95, "learning_rate": 5.082222062665139e-06, "loss": 0.3528, "step": 11189 }, { "epoch": 1.95, "learning_rate": 5.081638247075188e-06, "loss": 0.3455, "step": 11190 }, { "epoch": 1.95, "learning_rate": 5.081056507793154e-06, "loss": 0.3281, "step": 11191 }, { "epoch": 1.95, "learning_rate": 5.080476844872918e-06, "loss": 0.3508, "step": 11192 }, { "epoch": 1.95, "learning_rate": 5.079899258368166e-06, "loss": 0.3594, "step": 11193 }, { "epoch": 1.95, "learning_rate": 5.079323748332389e-06, "loss": 0.3383, "step": 11194 }, { "epoch": 1.95, "learning_rate": 5.078750314818892e-06, "loss": 0.3607, "step": 11195 }, { "epoch": 1.95, "learning_rate": 5.0781789578807835e-06, "loss": 0.3372, "step": 11196 }, { "epoch": 1.95, "learning_rate": 5.077609677570979e-06, "loss": 0.3583, "step": 11197 }, { "epoch": 1.95, "learning_rate": 5.0770424739422065e-06, "loss": 0.3405, "step": 11198 }, { "epoch": 1.95, "learning_rate": 5.0764773470469965e-06, "loss": 0.3446, "step": 11199 }, { "epoch": 1.95, "learning_rate": 5.075914296937687e-06, "loss": 0.3746, "step": 11200 }, { "epoch": 1.95, "learning_rate": 5.075353323666426e-06, "loss": 0.3529, "step": 11201 }, { "epoch": 1.95, "learning_rate": 5.074794427285174e-06, "loss": 0.3401, "step": 11202 }, { "epoch": 1.95, "learning_rate": 5.07423760784569e-06, "loss": 0.35, "step": 11203 }, { "epoch": 1.95, "learning_rate": 5.073682865399543e-06, "loss": 0.3283, "step": 11204 }, { "epoch": 1.95, "learning_rate": 5.073130199998112e-06, "loss": 0.3485, "step": 11205 }, { "epoch": 1.95, "learning_rate": 5.072579611692586e-06, "loss": 0.3532, "step": 11206 }, { "epoch": 1.95, "learning_rate": 5.072031100533956e-06, "loss": 0.3618, "step": 11207 }, { "epoch": 1.95, "learning_rate": 5.0714846665730234e-06, "loss": 0.3319, "step": 11208 }, { "epoch": 1.95, "learning_rate": 5.070940309860398e-06, "loss": 0.3548, "step": 11209 }, { "epoch": 1.95, "learning_rate": 5.070398030446494e-06, "loss": 0.3508, "step": 11210 }, { "epoch": 1.95, "learning_rate": 5.0698578283815395e-06, "loss": 0.347, "step": 11211 }, { "epoch": 1.95, "learning_rate": 5.0693197037155615e-06, "loss": 0.3509, "step": 11212 }, { "epoch": 1.95, "learning_rate": 5.068783656498399e-06, "loss": 0.3587, "step": 11213 }, { "epoch": 1.96, "learning_rate": 5.068249686779703e-06, "loss": 0.3508, "step": 11214 }, { "epoch": 1.96, "learning_rate": 5.067717794608926e-06, "loss": 0.3596, "step": 11215 }, { "epoch": 1.96, "learning_rate": 5.067187980035328e-06, "loss": 0.3509, "step": 11216 }, { "epoch": 1.96, "learning_rate": 5.066660243107979e-06, "loss": 0.3396, "step": 11217 }, { "epoch": 1.96, "learning_rate": 5.0661345838757595e-06, "loss": 0.3606, "step": 11218 }, { "epoch": 1.96, "learning_rate": 5.065611002387352e-06, "loss": 0.3396, "step": 11219 }, { "epoch": 1.96, "learning_rate": 5.065089498691247e-06, "loss": 0.3432, "step": 11220 }, { "epoch": 1.96, "learning_rate": 5.064570072835745e-06, "loss": 0.336, "step": 11221 }, { "epoch": 1.96, "learning_rate": 5.064052724868956e-06, "loss": 0.3535, "step": 11222 }, { "epoch": 1.96, "learning_rate": 5.063537454838794e-06, "loss": 0.3546, "step": 11223 }, { "epoch": 1.96, "learning_rate": 5.0630242627929785e-06, "loss": 0.3423, "step": 11224 }, { "epoch": 1.96, "learning_rate": 5.062513148779041e-06, "loss": 0.3554, "step": 11225 }, { "epoch": 1.96, "learning_rate": 5.062004112844322e-06, "loss": 0.3492, "step": 11226 }, { "epoch": 1.96, "learning_rate": 5.061497155035961e-06, "loss": 0.3604, "step": 11227 }, { "epoch": 1.96, "learning_rate": 5.0609922754009155e-06, "loss": 0.3645, "step": 11228 }, { "epoch": 1.96, "learning_rate": 5.060489473985944e-06, "loss": 0.3294, "step": 11229 }, { "epoch": 1.96, "learning_rate": 5.059988750837615e-06, "loss": 0.3452, "step": 11230 }, { "epoch": 1.96, "learning_rate": 5.059490106002302e-06, "loss": 0.3374, "step": 11231 }, { "epoch": 1.96, "learning_rate": 5.058993539526186e-06, "loss": 0.3437, "step": 11232 }, { "epoch": 1.96, "learning_rate": 5.0584990514552616e-06, "loss": 0.3525, "step": 11233 }, { "epoch": 1.96, "learning_rate": 5.058006641835325e-06, "loss": 0.3411, "step": 11234 }, { "epoch": 1.96, "learning_rate": 5.057516310711981e-06, "loss": 0.3349, "step": 11235 }, { "epoch": 1.96, "learning_rate": 5.05702805813064e-06, "loss": 0.3483, "step": 11236 }, { "epoch": 1.96, "learning_rate": 5.056541884136527e-06, "loss": 0.3536, "step": 11237 }, { "epoch": 1.96, "learning_rate": 5.056057788774666e-06, "loss": 0.3343, "step": 11238 }, { "epoch": 1.96, "learning_rate": 5.0555757720898915e-06, "loss": 0.3581, "step": 11239 }, { "epoch": 1.96, "learning_rate": 5.05509583412685e-06, "loss": 0.349, "step": 11240 }, { "epoch": 1.96, "learning_rate": 5.0546179749299906e-06, "loss": 0.3351, "step": 11241 }, { "epoch": 1.96, "learning_rate": 5.0541421945435664e-06, "loss": 0.3686, "step": 11242 }, { "epoch": 1.96, "learning_rate": 5.053668493011649e-06, "loss": 0.3535, "step": 11243 }, { "epoch": 1.96, "learning_rate": 5.053196870378105e-06, "loss": 0.3468, "step": 11244 }, { "epoch": 1.96, "learning_rate": 5.0527273266866195e-06, "loss": 0.3492, "step": 11245 }, { "epoch": 1.96, "learning_rate": 5.052259861980678e-06, "loss": 0.3393, "step": 11246 }, { "epoch": 1.96, "learning_rate": 5.051794476303572e-06, "loss": 0.3341, "step": 11247 }, { "epoch": 1.96, "learning_rate": 5.051331169698409e-06, "loss": 0.3523, "step": 11248 }, { "epoch": 1.96, "learning_rate": 5.050869942208096e-06, "loss": 0.3511, "step": 11249 }, { "epoch": 1.96, "learning_rate": 5.050410793875352e-06, "loss": 0.3713, "step": 11250 }, { "epoch": 1.96, "learning_rate": 5.0499537247427e-06, "loss": 0.3521, "step": 11251 }, { "epoch": 1.96, "learning_rate": 5.049498734852473e-06, "loss": 0.3312, "step": 11252 }, { "epoch": 1.96, "learning_rate": 5.049045824246809e-06, "loss": 0.3686, "step": 11253 }, { "epoch": 1.96, "learning_rate": 5.048594992967658e-06, "loss": 0.3497, "step": 11254 }, { "epoch": 1.96, "learning_rate": 5.04814624105677e-06, "loss": 0.364, "step": 11255 }, { "epoch": 1.96, "learning_rate": 5.0476995685557105e-06, "loss": 0.3506, "step": 11256 }, { "epoch": 1.96, "learning_rate": 5.04725497550585e-06, "loss": 0.3583, "step": 11257 }, { "epoch": 1.96, "learning_rate": 5.046812461948358e-06, "loss": 0.3403, "step": 11258 }, { "epoch": 1.96, "learning_rate": 5.046372027924226e-06, "loss": 0.3584, "step": 11259 }, { "epoch": 1.96, "learning_rate": 5.045933673474243e-06, "loss": 0.3407, "step": 11260 }, { "epoch": 1.96, "learning_rate": 5.045497398639007e-06, "loss": 0.3425, "step": 11261 }, { "epoch": 1.96, "learning_rate": 5.045063203458921e-06, "loss": 0.3548, "step": 11262 }, { "epoch": 1.96, "learning_rate": 5.044631087974205e-06, "loss": 0.3497, "step": 11263 }, { "epoch": 1.96, "learning_rate": 5.044201052224878e-06, "loss": 0.338, "step": 11264 }, { "epoch": 1.96, "learning_rate": 5.043773096250766e-06, "loss": 0.3594, "step": 11265 }, { "epoch": 1.96, "learning_rate": 5.043347220091509e-06, "loss": 0.3441, "step": 11266 }, { "epoch": 1.96, "learning_rate": 5.042923423786545e-06, "loss": 0.3378, "step": 11267 }, { "epoch": 1.96, "learning_rate": 5.0425017073751254e-06, "loss": 0.3503, "step": 11268 }, { "epoch": 1.96, "learning_rate": 5.04208207089631e-06, "loss": 0.3544, "step": 11269 }, { "epoch": 1.96, "learning_rate": 5.041664514388967e-06, "loss": 0.3504, "step": 11270 }, { "epoch": 1.97, "learning_rate": 5.041249037891762e-06, "loss": 0.3513, "step": 11271 }, { "epoch": 1.97, "learning_rate": 5.04083564144318e-06, "loss": 0.3543, "step": 11272 }, { "epoch": 1.97, "learning_rate": 5.040424325081505e-06, "loss": 0.3462, "step": 11273 }, { "epoch": 1.97, "learning_rate": 5.040015088844834e-06, "loss": 0.3329, "step": 11274 }, { "epoch": 1.97, "learning_rate": 5.039607932771071e-06, "loss": 0.3545, "step": 11275 }, { "epoch": 1.97, "learning_rate": 5.039202856897919e-06, "loss": 0.3566, "step": 11276 }, { "epoch": 1.97, "learning_rate": 5.0387998612629006e-06, "loss": 0.3623, "step": 11277 }, { "epoch": 1.97, "learning_rate": 5.038398945903334e-06, "loss": 0.3509, "step": 11278 }, { "epoch": 1.97, "learning_rate": 5.038000110856358e-06, "loss": 0.3493, "step": 11279 }, { "epoch": 1.97, "learning_rate": 5.037603356158906e-06, "loss": 0.3554, "step": 11280 }, { "epoch": 1.97, "learning_rate": 5.037208681847725e-06, "loss": 0.3469, "step": 11281 }, { "epoch": 1.97, "learning_rate": 5.036816087959368e-06, "loss": 0.3403, "step": 11282 }, { "epoch": 1.97, "learning_rate": 5.036425574530198e-06, "loss": 0.332, "step": 11283 }, { "epoch": 1.97, "learning_rate": 5.036037141596382e-06, "loss": 0.3461, "step": 11284 }, { "epoch": 1.97, "learning_rate": 5.035650789193893e-06, "loss": 0.341, "step": 11285 }, { "epoch": 1.97, "learning_rate": 5.0352665173585155e-06, "loss": 0.351, "step": 11286 }, { "epoch": 1.97, "learning_rate": 5.034884326125835e-06, "loss": 0.3533, "step": 11287 }, { "epoch": 1.97, "learning_rate": 5.034504215531259e-06, "loss": 0.3571, "step": 11288 }, { "epoch": 1.97, "learning_rate": 5.034126185609982e-06, "loss": 0.3671, "step": 11289 }, { "epoch": 1.97, "learning_rate": 5.033750236397019e-06, "loss": 0.3444, "step": 11290 }, { "epoch": 1.97, "learning_rate": 5.0333763679271905e-06, "loss": 0.3482, "step": 11291 }, { "epoch": 1.97, "learning_rate": 5.03300458023512e-06, "loss": 0.3562, "step": 11292 }, { "epoch": 1.97, "learning_rate": 5.032634873355245e-06, "loss": 0.3482, "step": 11293 }, { "epoch": 1.97, "learning_rate": 5.032267247321805e-06, "loss": 0.3454, "step": 11294 }, { "epoch": 1.97, "learning_rate": 5.0319017021688455e-06, "loss": 0.3456, "step": 11295 }, { "epoch": 1.97, "learning_rate": 5.031538237930224e-06, "loss": 0.3508, "step": 11296 }, { "epoch": 1.97, "learning_rate": 5.031176854639604e-06, "loss": 0.3415, "step": 11297 }, { "epoch": 1.97, "learning_rate": 5.030817552330457e-06, "loss": 0.3535, "step": 11298 }, { "epoch": 1.97, "learning_rate": 5.0304603310360545e-06, "loss": 0.3559, "step": 11299 }, { "epoch": 1.97, "learning_rate": 5.0301051907894854e-06, "loss": 0.3465, "step": 11300 }, { "epoch": 1.97, "learning_rate": 5.0297521316236405e-06, "loss": 0.3451, "step": 11301 }, { "epoch": 1.97, "learning_rate": 5.0294011535712195e-06, "loss": 0.3566, "step": 11302 }, { "epoch": 1.97, "learning_rate": 5.029052256664728e-06, "loss": 0.3479, "step": 11303 }, { "epoch": 1.97, "learning_rate": 5.02870544093648e-06, "loss": 0.3508, "step": 11304 }, { "epoch": 1.97, "learning_rate": 5.028360706418598e-06, "loss": 0.3414, "step": 11305 }, { "epoch": 1.97, "learning_rate": 5.028018053143007e-06, "loss": 0.35, "step": 11306 }, { "epoch": 1.97, "learning_rate": 5.027677481141442e-06, "loss": 0.3674, "step": 11307 }, { "epoch": 1.97, "learning_rate": 5.02733899044545e-06, "loss": 0.35, "step": 11308 }, { "epoch": 1.97, "learning_rate": 5.027002581086375e-06, "loss": 0.3478, "step": 11309 }, { "epoch": 1.97, "learning_rate": 5.026668253095378e-06, "loss": 0.3364, "step": 11310 }, { "epoch": 1.97, "learning_rate": 5.026336006503421e-06, "loss": 0.3363, "step": 11311 }, { "epoch": 1.97, "learning_rate": 5.026005841341279e-06, "loss": 0.3434, "step": 11312 }, { "epoch": 1.97, "learning_rate": 5.025677757639528e-06, "loss": 0.3434, "step": 11313 }, { "epoch": 1.97, "learning_rate": 5.025351755428554e-06, "loss": 0.3532, "step": 11314 }, { "epoch": 1.97, "learning_rate": 5.0250278347385485e-06, "loss": 0.3527, "step": 11315 }, { "epoch": 1.97, "learning_rate": 5.024705995599517e-06, "loss": 0.3444, "step": 11316 }, { "epoch": 1.97, "learning_rate": 5.024386238041261e-06, "loss": 0.3457, "step": 11317 }, { "epoch": 1.97, "learning_rate": 5.024068562093401e-06, "loss": 0.3427, "step": 11318 }, { "epoch": 1.97, "learning_rate": 5.023752967785354e-06, "loss": 0.3534, "step": 11319 }, { "epoch": 1.97, "learning_rate": 5.023439455146352e-06, "loss": 0.3351, "step": 11320 }, { "epoch": 1.97, "learning_rate": 5.02312802420543e-06, "loss": 0.3419, "step": 11321 }, { "epoch": 1.97, "learning_rate": 5.022818674991434e-06, "loss": 0.3554, "step": 11322 }, { "epoch": 1.97, "learning_rate": 5.0225114075330134e-06, "loss": 0.3674, "step": 11323 }, { "epoch": 1.97, "learning_rate": 5.022206221858625e-06, "loss": 0.3435, "step": 11324 }, { "epoch": 1.97, "learning_rate": 5.021903117996535e-06, "loss": 0.3506, "step": 11325 }, { "epoch": 1.97, "learning_rate": 5.0216020959748155e-06, "loss": 0.3473, "step": 11326 }, { "epoch": 1.97, "learning_rate": 5.021303155821347e-06, "loss": 0.3448, "step": 11327 }, { "epoch": 1.98, "learning_rate": 5.021006297563817e-06, "loss": 0.3556, "step": 11328 }, { "epoch": 1.98, "learning_rate": 5.020711521229717e-06, "loss": 0.3663, "step": 11329 }, { "epoch": 1.98, "learning_rate": 5.020418826846348e-06, "loss": 0.353, "step": 11330 }, { "epoch": 1.98, "learning_rate": 5.020128214440821e-06, "loss": 0.3396, "step": 11331 }, { "epoch": 1.98, "learning_rate": 5.0198396840400496e-06, "loss": 0.3428, "step": 11332 }, { "epoch": 1.98, "learning_rate": 5.019553235670758e-06, "loss": 0.3553, "step": 11333 }, { "epoch": 1.98, "learning_rate": 5.0192688693594715e-06, "loss": 0.3503, "step": 11334 }, { "epoch": 1.98, "learning_rate": 5.018986585132533e-06, "loss": 0.367, "step": 11335 }, { "epoch": 1.98, "learning_rate": 5.018706383016083e-06, "loss": 0.3514, "step": 11336 }, { "epoch": 1.98, "learning_rate": 5.018428263036075e-06, "loss": 0.337, "step": 11337 }, { "epoch": 1.98, "learning_rate": 5.018152225218266e-06, "loss": 0.3474, "step": 11338 }, { "epoch": 1.98, "learning_rate": 5.0178782695882226e-06, "loss": 0.3778, "step": 11339 }, { "epoch": 1.98, "learning_rate": 5.017606396171317e-06, "loss": 0.3595, "step": 11340 }, { "epoch": 1.98, "learning_rate": 5.017336604992728e-06, "loss": 0.3615, "step": 11341 }, { "epoch": 1.98, "learning_rate": 5.0170688960774455e-06, "loss": 0.3529, "step": 11342 }, { "epoch": 1.98, "learning_rate": 5.016803269450258e-06, "loss": 0.3509, "step": 11343 }, { "epoch": 1.98, "learning_rate": 5.0165397251357745e-06, "loss": 0.3531, "step": 11344 }, { "epoch": 1.98, "learning_rate": 5.0162782631584004e-06, "loss": 0.3386, "step": 11345 }, { "epoch": 1.98, "learning_rate": 5.016018883542351e-06, "loss": 0.3376, "step": 11346 }, { "epoch": 1.98, "learning_rate": 5.015761586311647e-06, "loss": 0.3468, "step": 11347 }, { "epoch": 1.98, "learning_rate": 5.01550637149012e-06, "loss": 0.3502, "step": 11348 }, { "epoch": 1.98, "learning_rate": 5.01525323910141e-06, "loss": 0.3389, "step": 11349 }, { "epoch": 1.98, "learning_rate": 5.015002189168957e-06, "loss": 0.3432, "step": 11350 }, { "epoch": 1.98, "learning_rate": 5.014753221716013e-06, "loss": 0.351, "step": 11351 }, { "epoch": 1.98, "learning_rate": 5.014506336765641e-06, "loss": 0.3427, "step": 11352 }, { "epoch": 1.98, "learning_rate": 5.014261534340699e-06, "loss": 0.3642, "step": 11353 }, { "epoch": 1.98, "learning_rate": 5.014018814463867e-06, "loss": 0.3515, "step": 11354 }, { "epoch": 1.98, "learning_rate": 5.01377817715762e-06, "loss": 0.3573, "step": 11355 }, { "epoch": 1.98, "learning_rate": 5.013539622444247e-06, "loss": 0.3388, "step": 11356 }, { "epoch": 1.98, "learning_rate": 5.01330315034584e-06, "loss": 0.3577, "step": 11357 }, { "epoch": 1.98, "learning_rate": 5.013068760884305e-06, "loss": 0.362, "step": 11358 }, { "epoch": 1.98, "learning_rate": 5.012836454081346e-06, "loss": 0.3582, "step": 11359 }, { "epoch": 1.98, "learning_rate": 5.012606229958478e-06, "loss": 0.3445, "step": 11360 }, { "epoch": 1.98, "learning_rate": 5.0123780885370254e-06, "loss": 0.3297, "step": 11361 }, { "epoch": 1.98, "learning_rate": 5.0121520298381175e-06, "loss": 0.3461, "step": 11362 }, { "epoch": 1.98, "learning_rate": 5.011928053882693e-06, "loss": 0.3507, "step": 11363 }, { "epoch": 1.98, "learning_rate": 5.011706160691494e-06, "loss": 0.3351, "step": 11364 }, { "epoch": 1.98, "learning_rate": 5.011486350285071e-06, "loss": 0.337, "step": 11365 }, { "epoch": 1.98, "learning_rate": 5.011268622683779e-06, "loss": 0.3596, "step": 11366 }, { "epoch": 1.98, "learning_rate": 5.011052977907791e-06, "loss": 0.3555, "step": 11367 }, { "epoch": 1.98, "learning_rate": 5.010839415977071e-06, "loss": 0.3514, "step": 11368 }, { "epoch": 1.98, "learning_rate": 5.0106279369114045e-06, "loss": 0.3506, "step": 11369 }, { "epoch": 1.98, "learning_rate": 5.010418540730375e-06, "loss": 0.3424, "step": 11370 }, { "epoch": 1.98, "learning_rate": 5.0102112274533745e-06, "loss": 0.3438, "step": 11371 }, { "epoch": 1.98, "learning_rate": 5.010005997099608e-06, "loss": 0.3374, "step": 11372 }, { "epoch": 1.98, "learning_rate": 5.009802849688081e-06, "loss": 0.3485, "step": 11373 }, { "epoch": 1.98, "learning_rate": 5.009601785237606e-06, "loss": 0.3352, "step": 11374 }, { "epoch": 1.98, "learning_rate": 5.009402803766809e-06, "loss": 0.3256, "step": 11375 }, { "epoch": 1.98, "learning_rate": 5.009205905294115e-06, "loss": 0.3588, "step": 11376 }, { "epoch": 1.98, "learning_rate": 5.009011089837765e-06, "loss": 0.3367, "step": 11377 }, { "epoch": 1.98, "learning_rate": 5.0088183574157956e-06, "loss": 0.3543, "step": 11378 }, { "epoch": 1.98, "learning_rate": 5.0086277080460616e-06, "loss": 0.3512, "step": 11379 }, { "epoch": 1.98, "learning_rate": 5.00843914174622e-06, "loss": 0.3471, "step": 11380 }, { "epoch": 1.98, "learning_rate": 5.008252658533732e-06, "loss": 0.3558, "step": 11381 }, { "epoch": 1.98, "learning_rate": 5.008068258425874e-06, "loss": 0.3402, "step": 11382 }, { "epoch": 1.98, "learning_rate": 5.0078859414397216e-06, "loss": 0.3458, "step": 11383 }, { "epoch": 1.98, "learning_rate": 5.00770570759216e-06, "loss": 0.3493, "step": 11384 }, { "epoch": 1.98, "learning_rate": 5.007527556899879e-06, "loss": 0.3527, "step": 11385 }, { "epoch": 1.99, "learning_rate": 5.007351489379386e-06, "loss": 0.3486, "step": 11386 }, { "epoch": 1.99, "learning_rate": 5.00717750504698e-06, "loss": 0.3424, "step": 11387 }, { "epoch": 1.99, "learning_rate": 5.007005603918781e-06, "loss": 0.3342, "step": 11388 }, { "epoch": 1.99, "learning_rate": 5.006835786010705e-06, "loss": 0.3588, "step": 11389 }, { "epoch": 1.99, "learning_rate": 5.00666805133848e-06, "loss": 0.3395, "step": 11390 }, { "epoch": 1.99, "learning_rate": 5.006502399917648e-06, "loss": 0.3363, "step": 11391 }, { "epoch": 1.99, "learning_rate": 5.0063388317635415e-06, "loss": 0.3374, "step": 11392 }, { "epoch": 1.99, "learning_rate": 5.006177346891315e-06, "loss": 0.3281, "step": 11393 }, { "epoch": 1.99, "learning_rate": 5.006017945315926e-06, "loss": 0.346, "step": 11394 }, { "epoch": 1.99, "learning_rate": 5.0058606270521326e-06, "loss": 0.3441, "step": 11395 }, { "epoch": 1.99, "learning_rate": 5.005705392114507e-06, "loss": 0.349, "step": 11396 }, { "epoch": 1.99, "learning_rate": 5.005552240517428e-06, "loss": 0.3411, "step": 11397 }, { "epoch": 1.99, "learning_rate": 5.00540117227508e-06, "loss": 0.3481, "step": 11398 }, { "epoch": 1.99, "learning_rate": 5.005252187401454e-06, "loss": 0.3521, "step": 11399 }, { "epoch": 1.99, "learning_rate": 5.005105285910348e-06, "loss": 0.3615, "step": 11400 }, { "epoch": 1.99, "learning_rate": 5.004960467815367e-06, "loss": 0.3422, "step": 11401 }, { "epoch": 1.99, "learning_rate": 5.004817733129924e-06, "loss": 0.3567, "step": 11402 }, { "epoch": 1.99, "learning_rate": 5.00467708186724e-06, "loss": 0.3448, "step": 11403 }, { "epoch": 1.99, "learning_rate": 5.004538514040337e-06, "loss": 0.3334, "step": 11404 }, { "epoch": 1.99, "learning_rate": 5.004402029662055e-06, "loss": 0.3685, "step": 11405 }, { "epoch": 1.99, "learning_rate": 5.004267628745029e-06, "loss": 0.3382, "step": 11406 }, { "epoch": 1.99, "learning_rate": 5.004135311301712e-06, "loss": 0.3432, "step": 11407 }, { "epoch": 1.99, "learning_rate": 5.004005077344354e-06, "loss": 0.3592, "step": 11408 }, { "epoch": 1.99, "learning_rate": 5.00387692688502e-06, "loss": 0.3508, "step": 11409 }, { "epoch": 1.99, "learning_rate": 5.0037508599355775e-06, "loss": 0.3678, "step": 11410 }, { "epoch": 1.99, "learning_rate": 5.003626876507703e-06, "loss": 0.3553, "step": 11411 }, { "epoch": 1.99, "learning_rate": 5.00350497661288e-06, "loss": 0.3434, "step": 11412 }, { "epoch": 1.99, "learning_rate": 5.003385160262396e-06, "loss": 0.352, "step": 11413 }, { "epoch": 1.99, "learning_rate": 5.0032674274673484e-06, "loss": 0.362, "step": 11414 }, { "epoch": 1.99, "learning_rate": 5.003151778238643e-06, "loss": 0.3427, "step": 11415 }, { "epoch": 1.99, "learning_rate": 5.00303821258699e-06, "loss": 0.3429, "step": 11416 }, { "epoch": 1.99, "learning_rate": 5.002926730522907e-06, "loss": 0.3512, "step": 11417 }, { "epoch": 1.99, "learning_rate": 5.002817332056721e-06, "loss": 0.3609, "step": 11418 }, { "epoch": 1.99, "learning_rate": 5.002710017198561e-06, "loss": 0.3463, "step": 11419 }, { "epoch": 1.99, "learning_rate": 5.0026047859583696e-06, "loss": 0.3623, "step": 11420 }, { "epoch": 1.99, "learning_rate": 5.002501638345889e-06, "loss": 0.3466, "step": 11421 }, { "epoch": 1.99, "learning_rate": 5.002400574370675e-06, "loss": 0.3453, "step": 11422 }, { "epoch": 1.99, "learning_rate": 5.002301594042088e-06, "loss": 0.3575, "step": 11423 }, { "epoch": 1.99, "learning_rate": 5.002204697369293e-06, "loss": 0.3215, "step": 11424 }, { "epoch": 1.99, "learning_rate": 5.002109884361266e-06, "loss": 0.3533, "step": 11425 }, { "epoch": 1.99, "learning_rate": 5.00201715502679e-06, "loss": 0.3671, "step": 11426 }, { "epoch": 1.99, "learning_rate": 5.0019265093744485e-06, "loss": 0.356, "step": 11427 }, { "epoch": 1.99, "learning_rate": 5.0018379474126395e-06, "loss": 0.3486, "step": 11428 }, { "epoch": 1.99, "learning_rate": 5.001751469149567e-06, "loss": 0.3637, "step": 11429 }, { "epoch": 1.99, "learning_rate": 5.001667074593236e-06, "loss": 0.3586, "step": 11430 }, { "epoch": 1.99, "learning_rate": 5.001584763751466e-06, "loss": 0.3465, "step": 11431 }, { "epoch": 1.99, "learning_rate": 5.0015045366318805e-06, "loss": 0.3432, "step": 11432 }, { "epoch": 1.99, "learning_rate": 5.001426393241909e-06, "loss": 0.3617, "step": 11433 }, { "epoch": 1.99, "learning_rate": 5.001350333588788e-06, "loss": 0.3551, "step": 11434 }, { "epoch": 1.99, "learning_rate": 5.001276357679563e-06, "loss": 0.3497, "step": 11435 }, { "epoch": 1.99, "learning_rate": 5.001204465521086e-06, "loss": 0.3649, "step": 11436 }, { "epoch": 1.99, "learning_rate": 5.0011346571200145e-06, "loss": 0.3521, "step": 11437 }, { "epoch": 1.99, "learning_rate": 5.001066932482811e-06, "loss": 0.3502, "step": 11438 }, { "epoch": 1.99, "learning_rate": 5.001001291615754e-06, "loss": 0.3454, "step": 11439 }, { "epoch": 1.99, "learning_rate": 5.000937734524919e-06, "loss": 0.3398, "step": 11440 }, { "epoch": 1.99, "learning_rate": 5.0008762612161925e-06, "loss": 0.3418, "step": 11441 }, { "epoch": 1.99, "learning_rate": 5.0008168716952696e-06, "loss": 0.3543, "step": 11442 }, { "epoch": 2.0, "learning_rate": 5.000759565967646e-06, "loss": 0.3395, "step": 11443 }, { "epoch": 2.0, "learning_rate": 5.0007043440386365e-06, "loss": 0.3447, "step": 11444 }, { "epoch": 2.0, "learning_rate": 5.0006512059133516e-06, "loss": 0.3617, "step": 11445 }, { "epoch": 2.0, "learning_rate": 5.000600151596712e-06, "loss": 0.3538, "step": 11446 }, { "epoch": 2.0, "learning_rate": 5.000551181093447e-06, "loss": 0.3487, "step": 11447 }, { "epoch": 2.0, "learning_rate": 5.000504294408096e-06, "loss": 0.3549, "step": 11448 }, { "epoch": 2.0, "learning_rate": 5.000459491544993e-06, "loss": 0.3627, "step": 11449 }, { "epoch": 2.0, "learning_rate": 5.000416772508296e-06, "loss": 0.3435, "step": 11450 }, { "epoch": 2.0, "learning_rate": 5.000376137301956e-06, "loss": 0.3532, "step": 11451 }, { "epoch": 2.0, "learning_rate": 5.000337585929736e-06, "loss": 0.3359, "step": 11452 }, { "epoch": 2.0, "learning_rate": 5.0003011183952114e-06, "loss": 0.3329, "step": 11453 }, { "epoch": 2.0, "learning_rate": 5.000266734701757e-06, "loss": 0.3362, "step": 11454 }, { "epoch": 2.0, "learning_rate": 5.000234434852557e-06, "loss": 0.3403, "step": 11455 }, { "epoch": 2.0, "learning_rate": 5.000204218850603e-06, "loss": 0.3502, "step": 11456 }, { "epoch": 2.0, "learning_rate": 5.000176086698693e-06, "loss": 0.3787, "step": 11457 }, { "epoch": 2.0, "learning_rate": 5.000150038399435e-06, "loss": 0.3481, "step": 11458 }, { "epoch": 2.0, "learning_rate": 5.000126073955239e-06, "loss": 0.3522, "step": 11459 }, { "epoch": 2.0, "learning_rate": 5.000104193368324e-06, "loss": 0.3468, "step": 11460 }, { "epoch": 2.0, "learning_rate": 5.000084396640719e-06, "loss": 0.3543, "step": 11461 }, { "epoch": 2.0, "learning_rate": 5.000066683774256e-06, "loss": 0.3389, "step": 11462 }, { "epoch": 2.0, "learning_rate": 5.000051054770574e-06, "loss": 0.3326, "step": 11463 }, { "epoch": 2.0, "learning_rate": 5.000037509631124e-06, "loss": 0.363, "step": 11464 }, { "epoch": 2.0, "learning_rate": 5.000026048357159e-06, "loss": 0.3547, "step": 11465 }, { "epoch": 2.0, "learning_rate": 5.000016670949742e-06, "loss": 0.3558, "step": 11466 }, { "epoch": 2.0, "learning_rate": 5.000009377409737e-06, "loss": 0.3613, "step": 11467 }, { "epoch": 2.0, "learning_rate": 5.000004167737821e-06, "loss": 0.3559, "step": 11468 }, { "epoch": 2.0, "learning_rate": 5.000001041934479e-06, "loss": 0.3496, "step": 11469 }, { "epoch": 2.0, "learning_rate": 5e-06, "loss": 0.3436, "step": 11470 }, { "epoch": 2.0, "step": 11470, "total_flos": 1.25014286028201e+19, "train_loss": 0.3992903973033559, "train_runtime": 92831.6636, "train_samples_per_second": 31.634, "train_steps_per_second": 0.124 } ], "max_steps": 11470, "num_train_epochs": 2, "total_flos": 1.25014286028201e+19, "trial_name": null, "trial_params": null }