{ "best_metric": null, "best_model_checkpoint": null, "epoch": 2.0, "global_step": 2060, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 3.2258064516129035e-07, "loss": 2.2715, "step": 1 }, { "epoch": 0.0, "learning_rate": 6.451612903225807e-07, "loss": 2.2695, "step": 2 }, { "epoch": 0.0, "learning_rate": 9.67741935483871e-07, "loss": 2.2461, "step": 3 }, { "epoch": 0.0, "learning_rate": 1.2903225806451614e-06, "loss": 2.2305, "step": 4 }, { "epoch": 0.0, "learning_rate": 1.6129032258064516e-06, "loss": 2.1895, "step": 5 }, { "epoch": 0.01, "learning_rate": 1.935483870967742e-06, "loss": 2.0859, "step": 6 }, { "epoch": 0.01, "learning_rate": 2.2580645161290324e-06, "loss": 1.7109, "step": 7 }, { "epoch": 0.01, "learning_rate": 2.580645161290323e-06, "loss": 1.5615, "step": 8 }, { "epoch": 0.01, "learning_rate": 2.903225806451613e-06, "loss": 0.7832, "step": 9 }, { "epoch": 0.01, "learning_rate": 3.225806451612903e-06, "loss": 0.7124, "step": 10 }, { "epoch": 0.01, "learning_rate": 3.548387096774194e-06, "loss": 0.5889, "step": 11 }, { "epoch": 0.01, "learning_rate": 3.870967741935484e-06, "loss": 0.2386, "step": 12 }, { "epoch": 0.01, "learning_rate": 4.193548387096774e-06, "loss": 0.2957, "step": 13 }, { "epoch": 0.01, "learning_rate": 4.516129032258065e-06, "loss": 0.2091, "step": 14 }, { "epoch": 0.01, "learning_rate": 4.838709677419355e-06, "loss": 0.162, "step": 15 }, { "epoch": 0.02, "learning_rate": 5.161290322580646e-06, "loss": 0.1678, "step": 16 }, { "epoch": 0.02, "learning_rate": 5.483870967741935e-06, "loss": 0.1461, "step": 17 }, { "epoch": 0.02, "learning_rate": 5.806451612903226e-06, "loss": 0.1058, "step": 18 }, { "epoch": 0.02, "learning_rate": 6.129032258064517e-06, "loss": 0.1252, "step": 19 }, { "epoch": 0.02, "learning_rate": 6.451612903225806e-06, "loss": 0.1209, "step": 20 }, { "epoch": 0.02, "learning_rate": 6.774193548387097e-06, "loss": 0.118, "step": 21 }, { "epoch": 0.02, "learning_rate": 7.096774193548388e-06, "loss": 0.1064, "step": 22 }, { "epoch": 0.02, "learning_rate": 7.4193548387096784e-06, "loss": 0.1092, "step": 23 }, { "epoch": 0.02, "learning_rate": 7.741935483870968e-06, "loss": 0.097, "step": 24 }, { "epoch": 0.02, "learning_rate": 8.064516129032258e-06, "loss": 0.1014, "step": 25 }, { "epoch": 0.03, "learning_rate": 8.387096774193549e-06, "loss": 0.116, "step": 26 }, { "epoch": 0.03, "learning_rate": 8.70967741935484e-06, "loss": 0.0974, "step": 27 }, { "epoch": 0.03, "learning_rate": 9.03225806451613e-06, "loss": 0.1018, "step": 28 }, { "epoch": 0.03, "learning_rate": 9.35483870967742e-06, "loss": 0.1072, "step": 29 }, { "epoch": 0.03, "learning_rate": 9.67741935483871e-06, "loss": 0.1057, "step": 30 }, { "epoch": 0.03, "learning_rate": 1e-05, "loss": 0.0978, "step": 31 }, { "epoch": 0.03, "learning_rate": 1.0322580645161291e-05, "loss": 0.0934, "step": 32 }, { "epoch": 0.03, "learning_rate": 1.0645161290322582e-05, "loss": 0.1044, "step": 33 }, { "epoch": 0.03, "learning_rate": 1.096774193548387e-05, "loss": 0.0984, "step": 34 }, { "epoch": 0.03, "learning_rate": 1.1290322580645164e-05, "loss": 0.1036, "step": 35 }, { "epoch": 0.03, "learning_rate": 1.1612903225806453e-05, "loss": 0.1044, "step": 36 }, { "epoch": 0.04, "learning_rate": 1.1935483870967743e-05, "loss": 0.1042, "step": 37 }, { "epoch": 0.04, "learning_rate": 1.2258064516129034e-05, "loss": 0.1068, "step": 38 }, { "epoch": 0.04, "learning_rate": 1.2580645161290324e-05, "loss": 0.1008, "step": 39 }, { "epoch": 0.04, "learning_rate": 1.2903225806451613e-05, "loss": 0.0981, "step": 40 }, { "epoch": 0.04, "learning_rate": 1.3225806451612903e-05, "loss": 0.0955, "step": 41 }, { "epoch": 0.04, "learning_rate": 1.3548387096774194e-05, "loss": 0.097, "step": 42 }, { "epoch": 0.04, "learning_rate": 1.3870967741935486e-05, "loss": 0.1028, "step": 43 }, { "epoch": 0.04, "learning_rate": 1.4193548387096776e-05, "loss": 0.0958, "step": 44 }, { "epoch": 0.04, "learning_rate": 1.4516129032258066e-05, "loss": 0.0918, "step": 45 }, { "epoch": 0.04, "learning_rate": 1.4838709677419357e-05, "loss": 0.0987, "step": 46 }, { "epoch": 0.05, "learning_rate": 1.5161290322580646e-05, "loss": 0.0977, "step": 47 }, { "epoch": 0.05, "learning_rate": 1.5483870967741936e-05, "loss": 0.0987, "step": 48 }, { "epoch": 0.05, "learning_rate": 1.5806451612903226e-05, "loss": 0.0953, "step": 49 }, { "epoch": 0.05, "learning_rate": 1.6129032258064517e-05, "loss": 0.0873, "step": 50 }, { "epoch": 0.05, "learning_rate": 1.6451612903225807e-05, "loss": 0.0874, "step": 51 }, { "epoch": 0.05, "learning_rate": 1.6774193548387098e-05, "loss": 0.0901, "step": 52 }, { "epoch": 0.05, "learning_rate": 1.7096774193548388e-05, "loss": 0.0883, "step": 53 }, { "epoch": 0.05, "learning_rate": 1.741935483870968e-05, "loss": 0.0809, "step": 54 }, { "epoch": 0.05, "learning_rate": 1.774193548387097e-05, "loss": 0.0865, "step": 55 }, { "epoch": 0.05, "learning_rate": 1.806451612903226e-05, "loss": 0.0885, "step": 56 }, { "epoch": 0.06, "learning_rate": 1.838709677419355e-05, "loss": 0.0854, "step": 57 }, { "epoch": 0.06, "learning_rate": 1.870967741935484e-05, "loss": 0.0876, "step": 58 }, { "epoch": 0.06, "learning_rate": 1.903225806451613e-05, "loss": 0.0832, "step": 59 }, { "epoch": 0.06, "learning_rate": 1.935483870967742e-05, "loss": 0.0831, "step": 60 }, { "epoch": 0.06, "learning_rate": 1.967741935483871e-05, "loss": 0.0847, "step": 61 }, { "epoch": 0.06, "learning_rate": 2e-05, "loss": 0.0893, "step": 62 }, { "epoch": 0.06, "learning_rate": 1.9999987638285976e-05, "loss": 0.0906, "step": 63 }, { "epoch": 0.06, "learning_rate": 1.999995055317446e-05, "loss": 0.0811, "step": 64 }, { "epoch": 0.06, "learning_rate": 1.9999888744757143e-05, "loss": 0.0903, "step": 65 }, { "epoch": 0.06, "learning_rate": 1.9999802213186836e-05, "loss": 0.0864, "step": 66 }, { "epoch": 0.07, "learning_rate": 1.999969095867747e-05, "loss": 0.0838, "step": 67 }, { "epoch": 0.07, "learning_rate": 1.999955498150411e-05, "loss": 0.0854, "step": 68 }, { "epoch": 0.07, "learning_rate": 1.999939428200294e-05, "loss": 0.0822, "step": 69 }, { "epoch": 0.07, "learning_rate": 1.9999208860571256e-05, "loss": 0.09, "step": 70 }, { "epoch": 0.07, "learning_rate": 1.999899871766749e-05, "loss": 0.0858, "step": 71 }, { "epoch": 0.07, "learning_rate": 1.9998763853811184e-05, "loss": 0.0792, "step": 72 }, { "epoch": 0.07, "learning_rate": 1.9998504269583007e-05, "loss": 0.0774, "step": 73 }, { "epoch": 0.07, "learning_rate": 1.9998219965624736e-05, "loss": 0.0995, "step": 74 }, { "epoch": 0.07, "learning_rate": 1.999791094263926e-05, "loss": 0.0857, "step": 75 }, { "epoch": 0.07, "learning_rate": 1.9997577201390608e-05, "loss": 0.0858, "step": 76 }, { "epoch": 0.07, "learning_rate": 1.9997218742703887e-05, "loss": 0.0872, "step": 77 }, { "epoch": 0.08, "learning_rate": 1.999683556746534e-05, "loss": 0.0897, "step": 78 }, { "epoch": 0.08, "learning_rate": 1.99964276766223e-05, "loss": 0.0801, "step": 79 }, { "epoch": 0.08, "learning_rate": 1.999599507118322e-05, "loss": 0.1039, "step": 80 }, { "epoch": 0.08, "learning_rate": 1.999553775221764e-05, "loss": 0.0798, "step": 81 }, { "epoch": 0.08, "learning_rate": 1.9995055720856217e-05, "loss": 0.0892, "step": 82 }, { "epoch": 0.08, "learning_rate": 1.9994548978290695e-05, "loss": 0.0756, "step": 83 }, { "epoch": 0.08, "learning_rate": 1.9994017525773915e-05, "loss": 0.0873, "step": 84 }, { "epoch": 0.08, "learning_rate": 1.999346136461981e-05, "loss": 0.0916, "step": 85 }, { "epoch": 0.08, "learning_rate": 1.99928804962034e-05, "loss": 0.0762, "step": 86 }, { "epoch": 0.08, "learning_rate": 1.9992274921960795e-05, "loss": 0.0919, "step": 87 }, { "epoch": 0.09, "learning_rate": 1.999164464338918e-05, "loss": 0.0775, "step": 88 }, { "epoch": 0.09, "learning_rate": 1.999098966204682e-05, "loss": 0.0829, "step": 89 }, { "epoch": 0.09, "learning_rate": 1.9990309979553046e-05, "loss": 0.0888, "step": 90 }, { "epoch": 0.09, "learning_rate": 1.9989605597588276e-05, "loss": 0.079, "step": 91 }, { "epoch": 0.09, "learning_rate": 1.998887651789398e-05, "loss": 0.0894, "step": 92 }, { "epoch": 0.09, "learning_rate": 1.9988122742272694e-05, "loss": 0.0847, "step": 93 }, { "epoch": 0.09, "learning_rate": 1.9987344272588007e-05, "loss": 0.0802, "step": 94 }, { "epoch": 0.09, "learning_rate": 1.9986541110764565e-05, "loss": 0.0795, "step": 95 }, { "epoch": 0.09, "learning_rate": 1.998571325878806e-05, "loss": 0.0775, "step": 96 }, { "epoch": 0.09, "learning_rate": 1.9984860718705226e-05, "loss": 0.0719, "step": 97 }, { "epoch": 0.1, "learning_rate": 1.9983983492623832e-05, "loss": 0.0778, "step": 98 }, { "epoch": 0.1, "learning_rate": 1.9983081582712684e-05, "loss": 0.0755, "step": 99 }, { "epoch": 0.1, "learning_rate": 1.998215499120161e-05, "loss": 0.0817, "step": 100 }, { "epoch": 0.1, "learning_rate": 1.9981203720381463e-05, "loss": 0.0794, "step": 101 }, { "epoch": 0.1, "learning_rate": 1.9980227772604114e-05, "loss": 0.0789, "step": 102 }, { "epoch": 0.1, "learning_rate": 1.9979227150282433e-05, "loss": 0.0788, "step": 103 }, { "epoch": 0.1, "learning_rate": 1.997820185589031e-05, "loss": 0.0844, "step": 104 }, { "epoch": 0.1, "learning_rate": 1.9977151891962616e-05, "loss": 0.0808, "step": 105 }, { "epoch": 0.1, "learning_rate": 1.9976077261095226e-05, "loss": 0.0935, "step": 106 }, { "epoch": 0.1, "learning_rate": 1.9974977965945e-05, "loss": 0.0778, "step": 107 }, { "epoch": 0.1, "learning_rate": 1.9973854009229762e-05, "loss": 0.0754, "step": 108 }, { "epoch": 0.11, "learning_rate": 1.997270539372833e-05, "loss": 0.0806, "step": 109 }, { "epoch": 0.11, "learning_rate": 1.9971532122280466e-05, "loss": 0.0791, "step": 110 }, { "epoch": 0.11, "learning_rate": 1.9970334197786903e-05, "loss": 0.0805, "step": 111 }, { "epoch": 0.11, "learning_rate": 1.9969111623209324e-05, "loss": 0.0703, "step": 112 }, { "epoch": 0.11, "learning_rate": 1.9967864401570345e-05, "loss": 0.075, "step": 113 }, { "epoch": 0.11, "learning_rate": 1.9966592535953532e-05, "loss": 0.0885, "step": 114 }, { "epoch": 0.11, "learning_rate": 1.996529602950337e-05, "loss": 0.0852, "step": 115 }, { "epoch": 0.11, "learning_rate": 1.9963974885425267e-05, "loss": 0.072, "step": 116 }, { "epoch": 0.11, "learning_rate": 1.9962629106985547e-05, "loss": 0.073, "step": 117 }, { "epoch": 0.11, "learning_rate": 1.996125869751143e-05, "loss": 0.0778, "step": 118 }, { "epoch": 0.12, "learning_rate": 1.9959863660391047e-05, "loss": 0.0712, "step": 119 }, { "epoch": 0.12, "learning_rate": 1.99584439990734e-05, "loss": 0.0745, "step": 120 }, { "epoch": 0.12, "learning_rate": 1.995699971706838e-05, "loss": 0.0816, "step": 121 }, { "epoch": 0.12, "learning_rate": 1.9955530817946748e-05, "loss": 0.0791, "step": 122 }, { "epoch": 0.12, "learning_rate": 1.9954037305340124e-05, "loss": 0.0781, "step": 123 }, { "epoch": 0.12, "learning_rate": 1.995251918294099e-05, "loss": 0.0735, "step": 124 }, { "epoch": 0.12, "learning_rate": 1.995097645450266e-05, "loss": 0.0763, "step": 125 }, { "epoch": 0.12, "learning_rate": 1.994940912383929e-05, "loss": 0.0726, "step": 126 }, { "epoch": 0.12, "learning_rate": 1.9947817194825854e-05, "loss": 0.0697, "step": 127 }, { "epoch": 0.12, "learning_rate": 1.994620067139815e-05, "loss": 0.0685, "step": 128 }, { "epoch": 0.13, "learning_rate": 1.9944559557552776e-05, "loss": 0.0722, "step": 129 }, { "epoch": 0.13, "learning_rate": 1.994289385734713e-05, "loss": 0.0851, "step": 130 }, { "epoch": 0.13, "learning_rate": 1.9941203574899394e-05, "loss": 0.0792, "step": 131 }, { "epoch": 0.13, "learning_rate": 1.9939488714388524e-05, "loss": 0.0814, "step": 132 }, { "epoch": 0.13, "learning_rate": 1.9937749280054244e-05, "loss": 0.0756, "step": 133 }, { "epoch": 0.13, "learning_rate": 1.9935985276197033e-05, "loss": 0.0733, "step": 134 }, { "epoch": 0.13, "learning_rate": 1.993419670717811e-05, "loss": 0.0776, "step": 135 }, { "epoch": 0.13, "learning_rate": 1.9932383577419432e-05, "loss": 0.0737, "step": 136 }, { "epoch": 0.13, "learning_rate": 1.993054589140368e-05, "loss": 0.0767, "step": 137 }, { "epoch": 0.13, "learning_rate": 1.992868365367424e-05, "loss": 0.0755, "step": 138 }, { "epoch": 0.13, "learning_rate": 1.9926796868835202e-05, "loss": 0.0771, "step": 139 }, { "epoch": 0.14, "learning_rate": 1.992488554155135e-05, "loss": 0.0792, "step": 140 }, { "epoch": 0.14, "learning_rate": 1.992294967654814e-05, "loss": 0.0827, "step": 141 }, { "epoch": 0.14, "learning_rate": 1.9920989278611685e-05, "loss": 0.0745, "step": 142 }, { "epoch": 0.14, "learning_rate": 1.9919004352588768e-05, "loss": 0.0729, "step": 143 }, { "epoch": 0.14, "learning_rate": 1.991699490338681e-05, "loss": 0.0748, "step": 144 }, { "epoch": 0.14, "learning_rate": 1.991496093597385e-05, "loss": 0.0782, "step": 145 }, { "epoch": 0.14, "learning_rate": 1.9912902455378556e-05, "loss": 0.0666, "step": 146 }, { "epoch": 0.14, "learning_rate": 1.9910819466690198e-05, "loss": 0.0727, "step": 147 }, { "epoch": 0.14, "learning_rate": 1.9908711975058638e-05, "loss": 0.0791, "step": 148 }, { "epoch": 0.14, "learning_rate": 1.990657998569432e-05, "loss": 0.0887, "step": 149 }, { "epoch": 0.15, "learning_rate": 1.990442350386825e-05, "loss": 0.0764, "step": 150 }, { "epoch": 0.15, "learning_rate": 1.9902242534911988e-05, "loss": 0.0756, "step": 151 }, { "epoch": 0.15, "learning_rate": 1.9900037084217637e-05, "loss": 0.0784, "step": 152 }, { "epoch": 0.15, "learning_rate": 1.9897807157237837e-05, "loss": 0.0799, "step": 153 }, { "epoch": 0.15, "learning_rate": 1.989555275948572e-05, "loss": 0.0778, "step": 154 }, { "epoch": 0.15, "learning_rate": 1.9893273896534936e-05, "loss": 0.0734, "step": 155 }, { "epoch": 0.15, "learning_rate": 1.9890970574019616e-05, "loss": 0.0711, "step": 156 }, { "epoch": 0.15, "learning_rate": 1.9888642797634358e-05, "loss": 0.079, "step": 157 }, { "epoch": 0.15, "learning_rate": 1.9886290573134228e-05, "loss": 0.0815, "step": 158 }, { "epoch": 0.15, "learning_rate": 1.988391390633473e-05, "loss": 0.0723, "step": 159 }, { "epoch": 0.16, "learning_rate": 1.9881512803111797e-05, "loss": 0.0744, "step": 160 }, { "epoch": 0.16, "learning_rate": 1.9879087269401782e-05, "loss": 0.0728, "step": 161 }, { "epoch": 0.16, "learning_rate": 1.9876637311201433e-05, "loss": 0.0703, "step": 162 }, { "epoch": 0.16, "learning_rate": 1.987416293456789e-05, "loss": 0.0632, "step": 163 }, { "epoch": 0.16, "learning_rate": 1.9871664145618658e-05, "loss": 0.076, "step": 164 }, { "epoch": 0.16, "learning_rate": 1.9869140950531604e-05, "loss": 0.07, "step": 165 }, { "epoch": 0.16, "learning_rate": 1.9866593355544924e-05, "loss": 0.0763, "step": 166 }, { "epoch": 0.16, "learning_rate": 1.9864021366957145e-05, "loss": 0.0692, "step": 167 }, { "epoch": 0.16, "learning_rate": 1.9861424991127117e-05, "loss": 0.0769, "step": 168 }, { "epoch": 0.16, "learning_rate": 1.9858804234473958e-05, "loss": 0.0721, "step": 169 }, { "epoch": 0.17, "learning_rate": 1.9856159103477085e-05, "loss": 0.0788, "step": 170 }, { "epoch": 0.17, "learning_rate": 1.9853489604676166e-05, "loss": 0.0786, "step": 171 }, { "epoch": 0.17, "learning_rate": 1.9850795744671116e-05, "loss": 0.0709, "step": 172 }, { "epoch": 0.17, "learning_rate": 1.9848077530122083e-05, "loss": 0.0831, "step": 173 }, { "epoch": 0.17, "learning_rate": 1.984533496774942e-05, "loss": 0.0757, "step": 174 }, { "epoch": 0.17, "learning_rate": 1.984256806433369e-05, "loss": 0.0698, "step": 175 }, { "epoch": 0.17, "learning_rate": 1.9839776826715616e-05, "loss": 0.0727, "step": 176 }, { "epoch": 0.17, "learning_rate": 1.98369612617961e-05, "loss": 0.0728, "step": 177 }, { "epoch": 0.17, "learning_rate": 1.983412137653619e-05, "loss": 0.0753, "step": 178 }, { "epoch": 0.17, "learning_rate": 1.9831257177957045e-05, "loss": 0.0757, "step": 179 }, { "epoch": 0.17, "learning_rate": 1.9828368673139947e-05, "loss": 0.0657, "step": 180 }, { "epoch": 0.18, "learning_rate": 1.982545586922628e-05, "loss": 0.0697, "step": 181 }, { "epoch": 0.18, "learning_rate": 1.9822518773417485e-05, "loss": 0.0706, "step": 182 }, { "epoch": 0.18, "learning_rate": 1.981955739297507e-05, "loss": 0.0617, "step": 183 }, { "epoch": 0.18, "learning_rate": 1.9816571735220583e-05, "loss": 0.0752, "step": 184 }, { "epoch": 0.18, "learning_rate": 1.9813561807535597e-05, "loss": 0.0703, "step": 185 }, { "epoch": 0.18, "learning_rate": 1.981052761736168e-05, "loss": 0.0718, "step": 186 }, { "epoch": 0.18, "learning_rate": 1.9807469172200397e-05, "loss": 0.0792, "step": 187 }, { "epoch": 0.18, "learning_rate": 1.9804386479613268e-05, "loss": 0.0792, "step": 188 }, { "epoch": 0.18, "learning_rate": 1.980127954722177e-05, "loss": 0.0796, "step": 189 }, { "epoch": 0.18, "learning_rate": 1.9798148382707298e-05, "loss": 0.0768, "step": 190 }, { "epoch": 0.19, "learning_rate": 1.9794992993811168e-05, "loss": 0.071, "step": 191 }, { "epoch": 0.19, "learning_rate": 1.9791813388334582e-05, "loss": 0.071, "step": 192 }, { "epoch": 0.19, "learning_rate": 1.9788609574138617e-05, "loss": 0.0737, "step": 193 }, { "epoch": 0.19, "learning_rate": 1.9785381559144196e-05, "loss": 0.077, "step": 194 }, { "epoch": 0.19, "learning_rate": 1.9782129351332083e-05, "loss": 0.0736, "step": 195 }, { "epoch": 0.19, "learning_rate": 1.9778852958742854e-05, "loss": 0.0765, "step": 196 }, { "epoch": 0.19, "learning_rate": 1.9775552389476865e-05, "loss": 0.0706, "step": 197 }, { "epoch": 0.19, "learning_rate": 1.977222765169426e-05, "loss": 0.0735, "step": 198 }, { "epoch": 0.19, "learning_rate": 1.976887875361493e-05, "loss": 0.0697, "step": 199 }, { "epoch": 0.19, "learning_rate": 1.9765505703518494e-05, "loss": 0.0664, "step": 200 }, { "epoch": 0.2, "learning_rate": 1.97621085097443e-05, "loss": 0.0714, "step": 201 }, { "epoch": 0.2, "learning_rate": 1.9758687180691364e-05, "loss": 0.0709, "step": 202 }, { "epoch": 0.2, "learning_rate": 1.9755241724818387e-05, "loss": 0.0767, "step": 203 }, { "epoch": 0.2, "learning_rate": 1.975177215064372e-05, "loss": 0.0731, "step": 204 }, { "epoch": 0.2, "learning_rate": 1.9748278466745344e-05, "loss": 0.0725, "step": 205 }, { "epoch": 0.2, "learning_rate": 1.9744760681760832e-05, "loss": 0.0703, "step": 206 }, { "epoch": 0.2, "learning_rate": 1.9741218804387365e-05, "loss": 0.0808, "step": 207 }, { "epoch": 0.2, "learning_rate": 1.973765284338167e-05, "loss": 0.0782, "step": 208 }, { "epoch": 0.2, "learning_rate": 1.9734062807560027e-05, "loss": 0.0734, "step": 209 }, { "epoch": 0.2, "learning_rate": 1.973044870579824e-05, "loss": 0.0838, "step": 210 }, { "epoch": 0.2, "learning_rate": 1.9726810547031604e-05, "loss": 0.0747, "step": 211 }, { "epoch": 0.21, "learning_rate": 1.9723148340254892e-05, "loss": 0.0817, "step": 212 }, { "epoch": 0.21, "learning_rate": 1.971946209452234e-05, "loss": 0.0728, "step": 213 }, { "epoch": 0.21, "learning_rate": 1.9715751818947603e-05, "loss": 0.0799, "step": 214 }, { "epoch": 0.21, "learning_rate": 1.9712017522703764e-05, "loss": 0.084, "step": 215 }, { "epoch": 0.21, "learning_rate": 1.970825921502328e-05, "loss": 0.076, "step": 216 }, { "epoch": 0.21, "learning_rate": 1.970447690519797e-05, "loss": 0.0778, "step": 217 }, { "epoch": 0.21, "learning_rate": 1.970067060257901e-05, "loss": 0.0767, "step": 218 }, { "epoch": 0.21, "learning_rate": 1.9696840316576877e-05, "loss": 0.0787, "step": 219 }, { "epoch": 0.21, "learning_rate": 1.9692986056661355e-05, "loss": 0.0807, "step": 220 }, { "epoch": 0.21, "learning_rate": 1.9689107832361496e-05, "loss": 0.0793, "step": 221 }, { "epoch": 0.22, "learning_rate": 1.9685205653265598e-05, "loss": 0.0846, "step": 222 }, { "epoch": 0.22, "learning_rate": 1.968127952902119e-05, "loss": 0.0789, "step": 223 }, { "epoch": 0.22, "learning_rate": 1.967732946933499e-05, "loss": 0.0765, "step": 224 }, { "epoch": 0.22, "learning_rate": 1.9673355483972902e-05, "loss": 0.0784, "step": 225 }, { "epoch": 0.22, "learning_rate": 1.9669357582759983e-05, "loss": 0.0745, "step": 226 }, { "epoch": 0.22, "learning_rate": 1.9665335775580418e-05, "loss": 0.074, "step": 227 }, { "epoch": 0.22, "learning_rate": 1.966129007237748e-05, "loss": 0.0735, "step": 228 }, { "epoch": 0.22, "learning_rate": 1.9657220483153546e-05, "loss": 0.0676, "step": 229 }, { "epoch": 0.22, "learning_rate": 1.9653127017970035e-05, "loss": 0.0638, "step": 230 }, { "epoch": 0.22, "learning_rate": 1.9649009686947392e-05, "loss": 0.082, "step": 231 }, { "epoch": 0.23, "learning_rate": 1.964486850026507e-05, "loss": 0.0706, "step": 232 }, { "epoch": 0.23, "learning_rate": 1.9640703468161508e-05, "loss": 0.0842, "step": 233 }, { "epoch": 0.23, "learning_rate": 1.963651460093409e-05, "loss": 0.0764, "step": 234 }, { "epoch": 0.23, "learning_rate": 1.9632301908939127e-05, "loss": 0.0713, "step": 235 }, { "epoch": 0.23, "learning_rate": 1.9628065402591846e-05, "loss": 0.0794, "step": 236 }, { "epoch": 0.23, "learning_rate": 1.9623805092366338e-05, "loss": 0.0746, "step": 237 }, { "epoch": 0.23, "learning_rate": 1.961952098879555e-05, "loss": 0.0845, "step": 238 }, { "epoch": 0.23, "learning_rate": 1.9615213102471258e-05, "loss": 0.0729, "step": 239 }, { "epoch": 0.23, "learning_rate": 1.961088144404403e-05, "loss": 0.0716, "step": 240 }, { "epoch": 0.23, "learning_rate": 1.960652602422321e-05, "loss": 0.0815, "step": 241 }, { "epoch": 0.23, "learning_rate": 1.9602146853776894e-05, "loss": 0.0764, "step": 242 }, { "epoch": 0.24, "learning_rate": 1.9597743943531892e-05, "loss": 0.0739, "step": 243 }, { "epoch": 0.24, "learning_rate": 1.9593317304373704e-05, "loss": 0.074, "step": 244 }, { "epoch": 0.24, "learning_rate": 1.9588866947246498e-05, "loss": 0.0721, "step": 245 }, { "epoch": 0.24, "learning_rate": 1.9584392883153088e-05, "loss": 0.0666, "step": 246 }, { "epoch": 0.24, "learning_rate": 1.957989512315489e-05, "loss": 0.0745, "step": 247 }, { "epoch": 0.24, "learning_rate": 1.9575373678371908e-05, "loss": 0.0729, "step": 248 }, { "epoch": 0.24, "learning_rate": 1.957082855998271e-05, "loss": 0.0683, "step": 249 }, { "epoch": 0.24, "learning_rate": 1.956625977922438e-05, "loss": 0.0707, "step": 250 }, { "epoch": 0.24, "learning_rate": 1.956166734739251e-05, "loss": 0.0688, "step": 251 }, { "epoch": 0.24, "learning_rate": 1.9557051275841174e-05, "loss": 0.0674, "step": 252 }, { "epoch": 0.25, "learning_rate": 1.955241157598287e-05, "loss": 0.0652, "step": 253 }, { "epoch": 0.25, "learning_rate": 1.9547748259288536e-05, "loss": 0.0703, "step": 254 }, { "epoch": 0.25, "learning_rate": 1.954306133728749e-05, "loss": 0.0727, "step": 255 }, { "epoch": 0.25, "learning_rate": 1.9538350821567406e-05, "loss": 0.0709, "step": 256 }, { "epoch": 0.25, "learning_rate": 1.9533616723774294e-05, "loss": 0.0728, "step": 257 }, { "epoch": 0.25, "learning_rate": 1.9528859055612466e-05, "loss": 0.0787, "step": 258 }, { "epoch": 0.25, "learning_rate": 1.9524077828844516e-05, "loss": 0.0693, "step": 259 }, { "epoch": 0.25, "learning_rate": 1.9519273055291266e-05, "loss": 0.0732, "step": 260 }, { "epoch": 0.25, "learning_rate": 1.951444474683177e-05, "loss": 0.0835, "step": 261 }, { "epoch": 0.25, "learning_rate": 1.9509592915403256e-05, "loss": 0.0782, "step": 262 }, { "epoch": 0.26, "learning_rate": 1.950471757300112e-05, "loss": 0.0733, "step": 263 }, { "epoch": 0.26, "learning_rate": 1.9499818731678873e-05, "loss": 0.0704, "step": 264 }, { "epoch": 0.26, "learning_rate": 1.9494896403548137e-05, "loss": 0.0776, "step": 265 }, { "epoch": 0.26, "learning_rate": 1.948995060077859e-05, "loss": 0.077, "step": 266 }, { "epoch": 0.26, "learning_rate": 1.9484981335597952e-05, "loss": 0.0733, "step": 267 }, { "epoch": 0.26, "learning_rate": 1.9479988620291955e-05, "loss": 0.0697, "step": 268 }, { "epoch": 0.26, "learning_rate": 1.9474972467204298e-05, "loss": 0.0729, "step": 269 }, { "epoch": 0.26, "learning_rate": 1.9469932888736634e-05, "loss": 0.0745, "step": 270 }, { "epoch": 0.26, "learning_rate": 1.9464869897348526e-05, "loss": 0.0719, "step": 271 }, { "epoch": 0.26, "learning_rate": 1.9459783505557425e-05, "loss": 0.0694, "step": 272 }, { "epoch": 0.27, "learning_rate": 1.9454673725938635e-05, "loss": 0.0743, "step": 273 }, { "epoch": 0.27, "learning_rate": 1.9449540571125284e-05, "loss": 0.0711, "step": 274 }, { "epoch": 0.27, "learning_rate": 1.944438405380829e-05, "loss": 0.0741, "step": 275 }, { "epoch": 0.27, "learning_rate": 1.943920418673633e-05, "loss": 0.0792, "step": 276 }, { "epoch": 0.27, "learning_rate": 1.943400098271581e-05, "loss": 0.071, "step": 277 }, { "epoch": 0.27, "learning_rate": 1.9428774454610845e-05, "loss": 0.0665, "step": 278 }, { "epoch": 0.27, "learning_rate": 1.942352461534319e-05, "loss": 0.0723, "step": 279 }, { "epoch": 0.27, "learning_rate": 1.941825147789225e-05, "loss": 0.0657, "step": 280 }, { "epoch": 0.27, "learning_rate": 1.9412955055295035e-05, "loss": 0.0779, "step": 281 }, { "epoch": 0.27, "learning_rate": 1.940763536064611e-05, "loss": 0.0803, "step": 282 }, { "epoch": 0.27, "learning_rate": 1.940229240709759e-05, "loss": 0.0729, "step": 283 }, { "epoch": 0.28, "learning_rate": 1.9396926207859085e-05, "loss": 0.0734, "step": 284 }, { "epoch": 0.28, "learning_rate": 1.939153677619768e-05, "loss": 0.0676, "step": 285 }, { "epoch": 0.28, "learning_rate": 1.9386124125437895e-05, "loss": 0.0667, "step": 286 }, { "epoch": 0.28, "learning_rate": 1.938068826896166e-05, "loss": 0.0766, "step": 287 }, { "epoch": 0.28, "learning_rate": 1.9375229220208275e-05, "loss": 0.0718, "step": 288 }, { "epoch": 0.28, "learning_rate": 1.9369746992674382e-05, "loss": 0.0797, "step": 289 }, { "epoch": 0.28, "learning_rate": 1.9364241599913923e-05, "loss": 0.0684, "step": 290 }, { "epoch": 0.28, "learning_rate": 1.9358713055538125e-05, "loss": 0.0704, "step": 291 }, { "epoch": 0.28, "learning_rate": 1.935316137321543e-05, "loss": 0.0698, "step": 292 }, { "epoch": 0.28, "learning_rate": 1.9347586566671512e-05, "loss": 0.0728, "step": 293 }, { "epoch": 0.29, "learning_rate": 1.93419886496892e-05, "loss": 0.072, "step": 294 }, { "epoch": 0.29, "learning_rate": 1.9336367636108462e-05, "loss": 0.0722, "step": 295 }, { "epoch": 0.29, "learning_rate": 1.9330723539826373e-05, "loss": 0.0653, "step": 296 }, { "epoch": 0.29, "learning_rate": 1.932505637479708e-05, "loss": 0.0649, "step": 297 }, { "epoch": 0.29, "learning_rate": 1.9319366155031743e-05, "loss": 0.0727, "step": 298 }, { "epoch": 0.29, "learning_rate": 1.9313652894598543e-05, "loss": 0.0662, "step": 299 }, { "epoch": 0.29, "learning_rate": 1.930791660762262e-05, "loss": 0.0721, "step": 300 }, { "epoch": 0.29, "learning_rate": 1.9302157308286046e-05, "loss": 0.0742, "step": 301 }, { "epoch": 0.29, "learning_rate": 1.9296375010827773e-05, "loss": 0.0746, "step": 302 }, { "epoch": 0.29, "learning_rate": 1.929056972954363e-05, "loss": 0.0711, "step": 303 }, { "epoch": 0.3, "learning_rate": 1.928474147878626e-05, "loss": 0.0701, "step": 304 }, { "epoch": 0.3, "learning_rate": 1.9278890272965097e-05, "loss": 0.0703, "step": 305 }, { "epoch": 0.3, "learning_rate": 1.9273016126546323e-05, "loss": 0.0746, "step": 306 }, { "epoch": 0.3, "learning_rate": 1.926711905405285e-05, "loss": 0.0696, "step": 307 }, { "epoch": 0.3, "learning_rate": 1.926119907006426e-05, "loss": 0.0683, "step": 308 }, { "epoch": 0.3, "learning_rate": 1.925525618921678e-05, "loss": 0.0699, "step": 309 }, { "epoch": 0.3, "learning_rate": 1.9249290426203253e-05, "loss": 0.0768, "step": 310 }, { "epoch": 0.3, "learning_rate": 1.9243301795773087e-05, "loss": 0.0705, "step": 311 }, { "epoch": 0.3, "learning_rate": 1.9237290312732227e-05, "loss": 0.0794, "step": 312 }, { "epoch": 0.3, "learning_rate": 1.9231255991943124e-05, "loss": 0.0686, "step": 313 }, { "epoch": 0.3, "learning_rate": 1.9225198848324687e-05, "loss": 0.0819, "step": 314 }, { "epoch": 0.31, "learning_rate": 1.9219118896852254e-05, "loss": 0.0685, "step": 315 }, { "epoch": 0.31, "learning_rate": 1.9213016152557543e-05, "loss": 0.0612, "step": 316 }, { "epoch": 0.31, "learning_rate": 1.9206890630528632e-05, "loss": 0.0812, "step": 317 }, { "epoch": 0.31, "learning_rate": 1.9200742345909915e-05, "loss": 0.0719, "step": 318 }, { "epoch": 0.31, "learning_rate": 1.9194571313902057e-05, "loss": 0.0728, "step": 319 }, { "epoch": 0.31, "learning_rate": 1.9188377549761962e-05, "loss": 0.0663, "step": 320 }, { "epoch": 0.31, "learning_rate": 1.9182161068802742e-05, "loss": 0.0789, "step": 321 }, { "epoch": 0.31, "learning_rate": 1.917592188639367e-05, "loss": 0.0702, "step": 322 }, { "epoch": 0.31, "learning_rate": 1.9169660017960135e-05, "loss": 0.0748, "step": 323 }, { "epoch": 0.31, "learning_rate": 1.9163375478983632e-05, "loss": 0.0728, "step": 324 }, { "epoch": 0.32, "learning_rate": 1.9157068285001693e-05, "loss": 0.0809, "step": 325 }, { "epoch": 0.32, "learning_rate": 1.915073845160786e-05, "loss": 0.0746, "step": 326 }, { "epoch": 0.32, "learning_rate": 1.9144385994451657e-05, "loss": 0.0721, "step": 327 }, { "epoch": 0.32, "learning_rate": 1.9138010929238534e-05, "loss": 0.0757, "step": 328 }, { "epoch": 0.32, "learning_rate": 1.9131613271729834e-05, "loss": 0.0724, "step": 329 }, { "epoch": 0.32, "learning_rate": 1.9125193037742764e-05, "loss": 0.065, "step": 330 }, { "epoch": 0.32, "learning_rate": 1.911875024315034e-05, "loss": 0.0751, "step": 331 }, { "epoch": 0.32, "learning_rate": 1.911228490388136e-05, "loss": 0.0673, "step": 332 }, { "epoch": 0.32, "learning_rate": 1.9105797035920358e-05, "loss": 0.0748, "step": 333 }, { "epoch": 0.32, "learning_rate": 1.909928665530757e-05, "loss": 0.0651, "step": 334 }, { "epoch": 0.33, "learning_rate": 1.9092753778138885e-05, "loss": 0.0704, "step": 335 }, { "epoch": 0.33, "learning_rate": 1.9086198420565822e-05, "loss": 0.0731, "step": 336 }, { "epoch": 0.33, "learning_rate": 1.9079620598795462e-05, "loss": 0.0777, "step": 337 }, { "epoch": 0.33, "learning_rate": 1.9073020329090447e-05, "loss": 0.0773, "step": 338 }, { "epoch": 0.33, "learning_rate": 1.9066397627768897e-05, "loss": 0.0711, "step": 339 }, { "epoch": 0.33, "learning_rate": 1.90597525112044e-05, "loss": 0.0764, "step": 340 }, { "epoch": 0.33, "learning_rate": 1.905308499582597e-05, "loss": 0.0726, "step": 341 }, { "epoch": 0.33, "learning_rate": 1.9046395098117983e-05, "loss": 0.0698, "step": 342 }, { "epoch": 0.33, "learning_rate": 1.903968283462016e-05, "loss": 0.0659, "step": 343 }, { "epoch": 0.33, "learning_rate": 1.9032948221927525e-05, "loss": 0.0708, "step": 344 }, { "epoch": 0.33, "learning_rate": 1.9026191276690342e-05, "loss": 0.076, "step": 345 }, { "epoch": 0.34, "learning_rate": 1.90194120156141e-05, "loss": 0.0659, "step": 346 }, { "epoch": 0.34, "learning_rate": 1.9012610455459447e-05, "loss": 0.079, "step": 347 }, { "epoch": 0.34, "learning_rate": 1.9005786613042185e-05, "loss": 0.0664, "step": 348 }, { "epoch": 0.34, "learning_rate": 1.8998940505233182e-05, "loss": 0.0697, "step": 349 }, { "epoch": 0.34, "learning_rate": 1.8992072148958368e-05, "loss": 0.0689, "step": 350 }, { "epoch": 0.34, "learning_rate": 1.8985181561198674e-05, "loss": 0.0714, "step": 351 }, { "epoch": 0.34, "learning_rate": 1.897826875898999e-05, "loss": 0.068, "step": 352 }, { "epoch": 0.34, "learning_rate": 1.8971333759423143e-05, "loss": 0.0704, "step": 353 }, { "epoch": 0.34, "learning_rate": 1.896437657964382e-05, "loss": 0.068, "step": 354 }, { "epoch": 0.34, "learning_rate": 1.8957397236852558e-05, "loss": 0.0671, "step": 355 }, { "epoch": 0.35, "learning_rate": 1.895039574830468e-05, "loss": 0.0642, "step": 356 }, { "epoch": 0.35, "learning_rate": 1.8943372131310272e-05, "loss": 0.0761, "step": 357 }, { "epoch": 0.35, "learning_rate": 1.8936326403234125e-05, "loss": 0.0699, "step": 358 }, { "epoch": 0.35, "learning_rate": 1.8929258581495688e-05, "loss": 0.066, "step": 359 }, { "epoch": 0.35, "learning_rate": 1.8922168683569038e-05, "loss": 0.0656, "step": 360 }, { "epoch": 0.35, "learning_rate": 1.891505672698284e-05, "loss": 0.0699, "step": 361 }, { "epoch": 0.35, "learning_rate": 1.8907922729320286e-05, "loss": 0.062, "step": 362 }, { "epoch": 0.35, "learning_rate": 1.8900766708219064e-05, "loss": 0.0662, "step": 363 }, { "epoch": 0.35, "learning_rate": 1.8893588681371306e-05, "loss": 0.0623, "step": 364 }, { "epoch": 0.35, "learning_rate": 1.888638866652356e-05, "loss": 0.066, "step": 365 }, { "epoch": 0.36, "learning_rate": 1.887916668147673e-05, "loss": 0.0723, "step": 366 }, { "epoch": 0.36, "learning_rate": 1.887192274408604e-05, "loss": 0.075, "step": 367 }, { "epoch": 0.36, "learning_rate": 1.8864656872260985e-05, "loss": 0.0756, "step": 368 }, { "epoch": 0.36, "learning_rate": 1.8857369083965297e-05, "loss": 0.079, "step": 369 }, { "epoch": 0.36, "learning_rate": 1.8850059397216875e-05, "loss": 0.0771, "step": 370 }, { "epoch": 0.36, "learning_rate": 1.8842727830087778e-05, "loss": 0.0652, "step": 371 }, { "epoch": 0.36, "learning_rate": 1.8835374400704153e-05, "loss": 0.0681, "step": 372 }, { "epoch": 0.36, "learning_rate": 1.88279991272462e-05, "loss": 0.0668, "step": 373 }, { "epoch": 0.36, "learning_rate": 1.8820602027948112e-05, "loss": 0.0721, "step": 374 }, { "epoch": 0.36, "learning_rate": 1.8813183121098073e-05, "loss": 0.0753, "step": 375 }, { "epoch": 0.37, "learning_rate": 1.8805742425038148e-05, "loss": 0.0754, "step": 376 }, { "epoch": 0.37, "learning_rate": 1.8798279958164295e-05, "loss": 0.0672, "step": 377 }, { "epoch": 0.37, "learning_rate": 1.879079573892629e-05, "loss": 0.0672, "step": 378 }, { "epoch": 0.37, "learning_rate": 1.878328978582769e-05, "loss": 0.0721, "step": 379 }, { "epoch": 0.37, "learning_rate": 1.8775762117425777e-05, "loss": 0.0714, "step": 380 }, { "epoch": 0.37, "learning_rate": 1.8768212752331537e-05, "loss": 0.0689, "step": 381 }, { "epoch": 0.37, "learning_rate": 1.8760641709209584e-05, "loss": 0.0743, "step": 382 }, { "epoch": 0.37, "learning_rate": 1.8753049006778133e-05, "loss": 0.0639, "step": 383 }, { "epoch": 0.37, "learning_rate": 1.8745434663808945e-05, "loss": 0.0713, "step": 384 }, { "epoch": 0.37, "learning_rate": 1.8737798699127286e-05, "loss": 0.0667, "step": 385 }, { "epoch": 0.37, "learning_rate": 1.8730141131611882e-05, "loss": 0.0686, "step": 386 }, { "epoch": 0.38, "learning_rate": 1.8722461980194864e-05, "loss": 0.0765, "step": 387 }, { "epoch": 0.38, "learning_rate": 1.8714761263861726e-05, "loss": 0.0755, "step": 388 }, { "epoch": 0.38, "learning_rate": 1.8707039001651276e-05, "loss": 0.0699, "step": 389 }, { "epoch": 0.38, "learning_rate": 1.8699295212655596e-05, "loss": 0.0703, "step": 390 }, { "epoch": 0.38, "learning_rate": 1.869152991601999e-05, "loss": 0.0646, "step": 391 }, { "epoch": 0.38, "learning_rate": 1.8683743130942928e-05, "loss": 0.0605, "step": 392 }, { "epoch": 0.38, "learning_rate": 1.8675934876676015e-05, "loss": 0.0695, "step": 393 }, { "epoch": 0.38, "learning_rate": 1.866810517252393e-05, "loss": 0.074, "step": 394 }, { "epoch": 0.38, "learning_rate": 1.866025403784439e-05, "loss": 0.0716, "step": 395 }, { "epoch": 0.38, "learning_rate": 1.8652381492048082e-05, "loss": 0.0725, "step": 396 }, { "epoch": 0.39, "learning_rate": 1.8644487554598653e-05, "loss": 0.0712, "step": 397 }, { "epoch": 0.39, "learning_rate": 1.863657224501261e-05, "loss": 0.0663, "step": 398 }, { "epoch": 0.39, "learning_rate": 1.8628635582859314e-05, "loss": 0.0752, "step": 399 }, { "epoch": 0.39, "learning_rate": 1.8620677587760916e-05, "loss": 0.0657, "step": 400 }, { "epoch": 0.39, "learning_rate": 1.861269827939231e-05, "loss": 0.0706, "step": 401 }, { "epoch": 0.39, "learning_rate": 1.8604697677481078e-05, "loss": 0.0667, "step": 402 }, { "epoch": 0.39, "learning_rate": 1.8596675801807452e-05, "loss": 0.0667, "step": 403 }, { "epoch": 0.39, "learning_rate": 1.8588632672204264e-05, "loss": 0.0696, "step": 404 }, { "epoch": 0.39, "learning_rate": 1.8580568308556876e-05, "loss": 0.0778, "step": 405 }, { "epoch": 0.39, "learning_rate": 1.857248273080317e-05, "loss": 0.0728, "step": 406 }, { "epoch": 0.4, "learning_rate": 1.856437595893346e-05, "loss": 0.0662, "step": 407 }, { "epoch": 0.4, "learning_rate": 1.8556248012990468e-05, "loss": 0.0746, "step": 408 }, { "epoch": 0.4, "learning_rate": 1.8548098913069263e-05, "loss": 0.0662, "step": 409 }, { "epoch": 0.4, "learning_rate": 1.853992867931721e-05, "loss": 0.0681, "step": 410 }, { "epoch": 0.4, "learning_rate": 1.853173733193393e-05, "loss": 0.0685, "step": 411 }, { "epoch": 0.4, "learning_rate": 1.852352489117124e-05, "loss": 0.0641, "step": 412 }, { "epoch": 0.4, "learning_rate": 1.8515291377333114e-05, "loss": 0.0732, "step": 413 }, { "epoch": 0.4, "learning_rate": 1.8507036810775617e-05, "loss": 0.0676, "step": 414 }, { "epoch": 0.4, "learning_rate": 1.8498761211906867e-05, "loss": 0.0764, "step": 415 }, { "epoch": 0.4, "learning_rate": 1.849046460118698e-05, "loss": 0.0748, "step": 416 }, { "epoch": 0.4, "learning_rate": 1.8482146999128026e-05, "loss": 0.0684, "step": 417 }, { "epoch": 0.41, "learning_rate": 1.8473808426293964e-05, "loss": 0.0767, "step": 418 }, { "epoch": 0.41, "learning_rate": 1.8465448903300606e-05, "loss": 0.0697, "step": 419 }, { "epoch": 0.41, "learning_rate": 1.8457068450815562e-05, "loss": 0.0701, "step": 420 }, { "epoch": 0.41, "learning_rate": 1.8448667089558177e-05, "loss": 0.0698, "step": 421 }, { "epoch": 0.41, "learning_rate": 1.8440244840299507e-05, "loss": 0.0679, "step": 422 }, { "epoch": 0.41, "learning_rate": 1.8431801723862225e-05, "loss": 0.0679, "step": 423 }, { "epoch": 0.41, "learning_rate": 1.842333776112062e-05, "loss": 0.0799, "step": 424 }, { "epoch": 0.41, "learning_rate": 1.8414852973000503e-05, "loss": 0.0732, "step": 425 }, { "epoch": 0.41, "learning_rate": 1.8406347380479182e-05, "loss": 0.0713, "step": 426 }, { "epoch": 0.41, "learning_rate": 1.83978210045854e-05, "loss": 0.0726, "step": 427 }, { "epoch": 0.42, "learning_rate": 1.8389273866399275e-05, "loss": 0.0647, "step": 428 }, { "epoch": 0.42, "learning_rate": 1.8380705987052266e-05, "loss": 0.0619, "step": 429 }, { "epoch": 0.42, "learning_rate": 1.837211738772711e-05, "loss": 0.0701, "step": 430 }, { "epoch": 0.42, "learning_rate": 1.8363508089657763e-05, "loss": 0.0812, "step": 431 }, { "epoch": 0.42, "learning_rate": 1.8354878114129368e-05, "loss": 0.0735, "step": 432 }, { "epoch": 0.42, "learning_rate": 1.8346227482478177e-05, "loss": 0.0718, "step": 433 }, { "epoch": 0.42, "learning_rate": 1.833755621609152e-05, "loss": 0.0653, "step": 434 }, { "epoch": 0.42, "learning_rate": 1.8328864336407736e-05, "loss": 0.0681, "step": 435 }, { "epoch": 0.42, "learning_rate": 1.8320151864916138e-05, "loss": 0.0694, "step": 436 }, { "epoch": 0.42, "learning_rate": 1.8311418823156936e-05, "loss": 0.0672, "step": 437 }, { "epoch": 0.43, "learning_rate": 1.830266523272121e-05, "loss": 0.0726, "step": 438 }, { "epoch": 0.43, "learning_rate": 1.829389111525083e-05, "loss": 0.0739, "step": 439 }, { "epoch": 0.43, "learning_rate": 1.8285096492438424e-05, "loss": 0.0731, "step": 440 }, { "epoch": 0.43, "learning_rate": 1.8276281386027315e-05, "loss": 0.0721, "step": 441 }, { "epoch": 0.43, "learning_rate": 1.8267445817811466e-05, "loss": 0.0694, "step": 442 }, { "epoch": 0.43, "learning_rate": 1.825858980963543e-05, "loss": 0.071, "step": 443 }, { "epoch": 0.43, "learning_rate": 1.8249713383394303e-05, "loss": 0.0671, "step": 444 }, { "epoch": 0.43, "learning_rate": 1.8240816561033645e-05, "loss": 0.0709, "step": 445 }, { "epoch": 0.43, "learning_rate": 1.8231899364549453e-05, "loss": 0.0733, "step": 446 }, { "epoch": 0.43, "learning_rate": 1.8222961815988097e-05, "loss": 0.0706, "step": 447 }, { "epoch": 0.43, "learning_rate": 1.8214003937446254e-05, "loss": 0.0652, "step": 448 }, { "epoch": 0.44, "learning_rate": 1.8205025751070878e-05, "loss": 0.0713, "step": 449 }, { "epoch": 0.44, "learning_rate": 1.8196027279059115e-05, "loss": 0.0662, "step": 450 }, { "epoch": 0.44, "learning_rate": 1.818700854365828e-05, "loss": 0.0714, "step": 451 }, { "epoch": 0.44, "learning_rate": 1.817796956716578e-05, "loss": 0.0718, "step": 452 }, { "epoch": 0.44, "learning_rate": 1.8168910371929052e-05, "loss": 0.0742, "step": 453 }, { "epoch": 0.44, "learning_rate": 1.815983098034555e-05, "loss": 0.0677, "step": 454 }, { "epoch": 0.44, "learning_rate": 1.8150731414862623e-05, "loss": 0.0746, "step": 455 }, { "epoch": 0.44, "learning_rate": 1.8141611697977526e-05, "loss": 0.0718, "step": 456 }, { "epoch": 0.44, "learning_rate": 1.813247185223733e-05, "loss": 0.0703, "step": 457 }, { "epoch": 0.44, "learning_rate": 1.812331190023886e-05, "loss": 0.0718, "step": 458 }, { "epoch": 0.45, "learning_rate": 1.811413186462866e-05, "loss": 0.0686, "step": 459 }, { "epoch": 0.45, "learning_rate": 1.810493176810292e-05, "loss": 0.0717, "step": 460 }, { "epoch": 0.45, "learning_rate": 1.8095711633407444e-05, "loss": 0.0745, "step": 461 }, { "epoch": 0.45, "learning_rate": 1.808647148333755e-05, "loss": 0.0736, "step": 462 }, { "epoch": 0.45, "learning_rate": 1.807721134073807e-05, "loss": 0.0747, "step": 463 }, { "epoch": 0.45, "learning_rate": 1.8067931228503246e-05, "loss": 0.0636, "step": 464 }, { "epoch": 0.45, "learning_rate": 1.8058631169576692e-05, "loss": 0.0706, "step": 465 }, { "epoch": 0.45, "learning_rate": 1.804931118695135e-05, "loss": 0.0786, "step": 466 }, { "epoch": 0.45, "learning_rate": 1.8039971303669407e-05, "loss": 0.0667, "step": 467 }, { "epoch": 0.45, "learning_rate": 1.8030611542822257e-05, "loss": 0.0688, "step": 468 }, { "epoch": 0.46, "learning_rate": 1.802123192755044e-05, "loss": 0.0659, "step": 469 }, { "epoch": 0.46, "learning_rate": 1.8011832481043577e-05, "loss": 0.0703, "step": 470 }, { "epoch": 0.46, "learning_rate": 1.800241322654032e-05, "loss": 0.0721, "step": 471 }, { "epoch": 0.46, "learning_rate": 1.7992974187328303e-05, "loss": 0.0762, "step": 472 }, { "epoch": 0.46, "learning_rate": 1.7983515386744062e-05, "loss": 0.071, "step": 473 }, { "epoch": 0.46, "learning_rate": 1.7974036848172992e-05, "loss": 0.0708, "step": 474 }, { "epoch": 0.46, "learning_rate": 1.7964538595049293e-05, "loss": 0.0687, "step": 475 }, { "epoch": 0.46, "learning_rate": 1.79550206508559e-05, "loss": 0.0687, "step": 476 }, { "epoch": 0.46, "learning_rate": 1.7945483039124438e-05, "loss": 0.0745, "step": 477 }, { "epoch": 0.46, "learning_rate": 1.793592578343515e-05, "loss": 0.0694, "step": 478 }, { "epoch": 0.47, "learning_rate": 1.792634890741685e-05, "loss": 0.0749, "step": 479 }, { "epoch": 0.47, "learning_rate": 1.7916752434746856e-05, "loss": 0.0748, "step": 480 }, { "epoch": 0.47, "learning_rate": 1.790713638915094e-05, "loss": 0.0676, "step": 481 }, { "epoch": 0.47, "learning_rate": 1.7897500794403263e-05, "loss": 0.0704, "step": 482 }, { "epoch": 0.47, "learning_rate": 1.7887845674326316e-05, "loss": 0.0721, "step": 483 }, { "epoch": 0.47, "learning_rate": 1.787817105279087e-05, "loss": 0.0697, "step": 484 }, { "epoch": 0.47, "learning_rate": 1.78684769537159e-05, "loss": 0.0697, "step": 485 }, { "epoch": 0.47, "learning_rate": 1.785876340106855e-05, "loss": 0.0779, "step": 486 }, { "epoch": 0.47, "learning_rate": 1.7849030418864047e-05, "loss": 0.0706, "step": 487 }, { "epoch": 0.47, "learning_rate": 1.783927803116566e-05, "loss": 0.0781, "step": 488 }, { "epoch": 0.47, "learning_rate": 1.7829506262084634e-05, "loss": 0.0723, "step": 489 }, { "epoch": 0.48, "learning_rate": 1.7819715135780133e-05, "loss": 0.0683, "step": 490 }, { "epoch": 0.48, "learning_rate": 1.7809904676459178e-05, "loss": 0.0609, "step": 491 }, { "epoch": 0.48, "learning_rate": 1.7800074908376586e-05, "loss": 0.0731, "step": 492 }, { "epoch": 0.48, "learning_rate": 1.7790225855834913e-05, "loss": 0.0746, "step": 493 }, { "epoch": 0.48, "learning_rate": 1.7780357543184396e-05, "loss": 0.0633, "step": 494 }, { "epoch": 0.48, "learning_rate": 1.7770469994822884e-05, "loss": 0.0749, "step": 495 }, { "epoch": 0.48, "learning_rate": 1.776056323519579e-05, "loss": 0.0619, "step": 496 }, { "epoch": 0.48, "learning_rate": 1.775063728879602e-05, "loss": 0.0663, "step": 497 }, { "epoch": 0.48, "learning_rate": 1.7740692180163907e-05, "loss": 0.0714, "step": 498 }, { "epoch": 0.48, "learning_rate": 1.773072793388718e-05, "loss": 0.065, "step": 499 }, { "epoch": 0.49, "learning_rate": 1.7720744574600865e-05, "loss": 0.0691, "step": 500 }, { "epoch": 0.49, "learning_rate": 1.7710742126987252e-05, "loss": 0.0809, "step": 501 }, { "epoch": 0.49, "learning_rate": 1.770072061577581e-05, "loss": 0.0672, "step": 502 }, { "epoch": 0.49, "learning_rate": 1.769068006574317e-05, "loss": 0.0728, "step": 503 }, { "epoch": 0.49, "learning_rate": 1.7680620501712996e-05, "loss": 0.0697, "step": 504 }, { "epoch": 0.49, "learning_rate": 1.7670541948555987e-05, "loss": 0.0729, "step": 505 }, { "epoch": 0.49, "learning_rate": 1.766044443118978e-05, "loss": 0.0728, "step": 506 }, { "epoch": 0.49, "learning_rate": 1.7650327974578904e-05, "loss": 0.0642, "step": 507 }, { "epoch": 0.49, "learning_rate": 1.7640192603734692e-05, "loss": 0.0677, "step": 508 }, { "epoch": 0.49, "learning_rate": 1.7630038343715273e-05, "loss": 0.0674, "step": 509 }, { "epoch": 0.5, "learning_rate": 1.7619865219625453e-05, "loss": 0.0743, "step": 510 }, { "epoch": 0.5, "learning_rate": 1.7609673256616676e-05, "loss": 0.0715, "step": 511 }, { "epoch": 0.5, "learning_rate": 1.7599462479886976e-05, "loss": 0.0663, "step": 512 }, { "epoch": 0.5, "learning_rate": 1.7589232914680893e-05, "loss": 0.0704, "step": 513 }, { "epoch": 0.5, "learning_rate": 1.757898458628941e-05, "loss": 0.0648, "step": 514 }, { "epoch": 0.5, "learning_rate": 1.756871752004992e-05, "loss": 0.0667, "step": 515 }, { "epoch": 0.5, "learning_rate": 1.755843174134612e-05, "loss": 0.077, "step": 516 }, { "epoch": 0.5, "learning_rate": 1.754812727560799e-05, "loss": 0.0633, "step": 517 }, { "epoch": 0.5, "learning_rate": 1.7537804148311696e-05, "loss": 0.0761, "step": 518 }, { "epoch": 0.5, "learning_rate": 1.752746238497955e-05, "loss": 0.0718, "step": 519 }, { "epoch": 0.5, "learning_rate": 1.7517102011179935e-05, "loss": 0.0681, "step": 520 }, { "epoch": 0.51, "learning_rate": 1.7506723052527243e-05, "loss": 0.0755, "step": 521 }, { "epoch": 0.51, "learning_rate": 1.7496325534681825e-05, "loss": 0.0705, "step": 522 }, { "epoch": 0.51, "learning_rate": 1.7485909483349908e-05, "loss": 0.0745, "step": 523 }, { "epoch": 0.51, "learning_rate": 1.747547492428354e-05, "loss": 0.0775, "step": 524 }, { "epoch": 0.51, "learning_rate": 1.7465021883280523e-05, "loss": 0.0736, "step": 525 }, { "epoch": 0.51, "learning_rate": 1.7454550386184364e-05, "loss": 0.0703, "step": 526 }, { "epoch": 0.51, "learning_rate": 1.744406045888419e-05, "loss": 0.075, "step": 527 }, { "epoch": 0.51, "learning_rate": 1.74335521273147e-05, "loss": 0.0699, "step": 528 }, { "epoch": 0.51, "learning_rate": 1.7423025417456093e-05, "loss": 0.075, "step": 529 }, { "epoch": 0.51, "learning_rate": 1.7412480355334006e-05, "loss": 0.0634, "step": 530 }, { "epoch": 0.52, "learning_rate": 1.7401916967019437e-05, "loss": 0.0706, "step": 531 }, { "epoch": 0.52, "learning_rate": 1.7391335278628713e-05, "loss": 0.081, "step": 532 }, { "epoch": 0.52, "learning_rate": 1.738073531632339e-05, "loss": 0.069, "step": 533 }, { "epoch": 0.52, "learning_rate": 1.7370117106310216e-05, "loss": 0.0667, "step": 534 }, { "epoch": 0.52, "learning_rate": 1.7359480674841036e-05, "loss": 0.0745, "step": 535 }, { "epoch": 0.52, "learning_rate": 1.734882604821276e-05, "loss": 0.0702, "step": 536 }, { "epoch": 0.52, "learning_rate": 1.7338153252767277e-05, "loss": 0.0691, "step": 537 }, { "epoch": 0.52, "learning_rate": 1.7327462314891402e-05, "loss": 0.0681, "step": 538 }, { "epoch": 0.52, "learning_rate": 1.7316753261016782e-05, "loss": 0.0656, "step": 539 }, { "epoch": 0.52, "learning_rate": 1.730602611761989e-05, "loss": 0.0675, "step": 540 }, { "epoch": 0.53, "learning_rate": 1.7295280911221885e-05, "loss": 0.0666, "step": 541 }, { "epoch": 0.53, "learning_rate": 1.7284517668388613e-05, "loss": 0.0685, "step": 542 }, { "epoch": 0.53, "learning_rate": 1.7273736415730488e-05, "loss": 0.0643, "step": 543 }, { "epoch": 0.53, "learning_rate": 1.726293717990247e-05, "loss": 0.0696, "step": 544 }, { "epoch": 0.53, "learning_rate": 1.7252119987603976e-05, "loss": 0.0723, "step": 545 }, { "epoch": 0.53, "learning_rate": 1.72412848655788e-05, "loss": 0.0692, "step": 546 }, { "epoch": 0.53, "learning_rate": 1.7230431840615096e-05, "loss": 0.069, "step": 547 }, { "epoch": 0.53, "learning_rate": 1.7219560939545246e-05, "loss": 0.0685, "step": 548 }, { "epoch": 0.53, "learning_rate": 1.7208672189245853e-05, "loss": 0.0706, "step": 549 }, { "epoch": 0.53, "learning_rate": 1.7197765616637635e-05, "loss": 0.0582, "step": 550 }, { "epoch": 0.53, "learning_rate": 1.7186841248685385e-05, "loss": 0.0666, "step": 551 }, { "epoch": 0.54, "learning_rate": 1.717589911239788e-05, "loss": 0.0707, "step": 552 }, { "epoch": 0.54, "learning_rate": 1.7164939234827832e-05, "loss": 0.0736, "step": 553 }, { "epoch": 0.54, "learning_rate": 1.715396164307182e-05, "loss": 0.0726, "step": 554 }, { "epoch": 0.54, "learning_rate": 1.714296636427021e-05, "loss": 0.0645, "step": 555 }, { "epoch": 0.54, "learning_rate": 1.7131953425607103e-05, "loss": 0.0716, "step": 556 }, { "epoch": 0.54, "learning_rate": 1.712092285431026e-05, "loss": 0.0635, "step": 557 }, { "epoch": 0.54, "learning_rate": 1.7109874677651023e-05, "loss": 0.076, "step": 558 }, { "epoch": 0.54, "learning_rate": 1.709880892294429e-05, "loss": 0.0688, "step": 559 }, { "epoch": 0.54, "learning_rate": 1.7087725617548385e-05, "loss": 0.0735, "step": 560 }, { "epoch": 0.54, "learning_rate": 1.7076624788865044e-05, "loss": 0.0651, "step": 561 }, { "epoch": 0.55, "learning_rate": 1.7065506464339326e-05, "loss": 0.0748, "step": 562 }, { "epoch": 0.55, "learning_rate": 1.7054370671459533e-05, "loss": 0.0681, "step": 563 }, { "epoch": 0.55, "learning_rate": 1.7043217437757164e-05, "loss": 0.0708, "step": 564 }, { "epoch": 0.55, "learning_rate": 1.703204679080684e-05, "loss": 0.0669, "step": 565 }, { "epoch": 0.55, "learning_rate": 1.702085875822623e-05, "loss": 0.0742, "step": 566 }, { "epoch": 0.55, "learning_rate": 1.7009653367675975e-05, "loss": 0.07, "step": 567 }, { "epoch": 0.55, "learning_rate": 1.6998430646859656e-05, "loss": 0.0642, "step": 568 }, { "epoch": 0.55, "learning_rate": 1.6987190623523675e-05, "loss": 0.0728, "step": 569 }, { "epoch": 0.55, "learning_rate": 1.697593332545723e-05, "loss": 0.071, "step": 570 }, { "epoch": 0.55, "learning_rate": 1.696465878049222e-05, "loss": 0.0697, "step": 571 }, { "epoch": 0.56, "learning_rate": 1.6953367016503182e-05, "loss": 0.0706, "step": 572 }, { "epoch": 0.56, "learning_rate": 1.6942058061407226e-05, "loss": 0.0685, "step": 573 }, { "epoch": 0.56, "learning_rate": 1.6930731943163975e-05, "loss": 0.072, "step": 574 }, { "epoch": 0.56, "learning_rate": 1.6919388689775463e-05, "loss": 0.0762, "step": 575 }, { "epoch": 0.56, "learning_rate": 1.6908028329286113e-05, "loss": 0.0673, "step": 576 }, { "epoch": 0.56, "learning_rate": 1.689665088978262e-05, "loss": 0.0648, "step": 577 }, { "epoch": 0.56, "learning_rate": 1.6885256399393923e-05, "loss": 0.0739, "step": 578 }, { "epoch": 0.56, "learning_rate": 1.6873844886291103e-05, "loss": 0.0651, "step": 579 }, { "epoch": 0.56, "learning_rate": 1.686241637868734e-05, "loss": 0.07, "step": 580 }, { "epoch": 0.56, "learning_rate": 1.6850970904837813e-05, "loss": 0.0633, "step": 581 }, { "epoch": 0.57, "learning_rate": 1.683950849303966e-05, "loss": 0.0687, "step": 582 }, { "epoch": 0.57, "learning_rate": 1.6828029171631892e-05, "loss": 0.0739, "step": 583 }, { "epoch": 0.57, "learning_rate": 1.681653296899533e-05, "loss": 0.0623, "step": 584 }, { "epoch": 0.57, "learning_rate": 1.6805019913552526e-05, "loss": 0.0599, "step": 585 }, { "epoch": 0.57, "learning_rate": 1.6793490033767697e-05, "loss": 0.0621, "step": 586 }, { "epoch": 0.57, "learning_rate": 1.6781943358146666e-05, "loss": 0.068, "step": 587 }, { "epoch": 0.57, "learning_rate": 1.6770379915236766e-05, "loss": 0.0661, "step": 588 }, { "epoch": 0.57, "learning_rate": 1.6758799733626793e-05, "loss": 0.0662, "step": 589 }, { "epoch": 0.57, "learning_rate": 1.6747202841946928e-05, "loss": 0.0629, "step": 590 }, { "epoch": 0.57, "learning_rate": 1.673558926886866e-05, "loss": 0.0629, "step": 591 }, { "epoch": 0.57, "learning_rate": 1.6723959043104726e-05, "loss": 0.071, "step": 592 }, { "epoch": 0.58, "learning_rate": 1.6712312193409032e-05, "loss": 0.0721, "step": 593 }, { "epoch": 0.58, "learning_rate": 1.6700648748576577e-05, "loss": 0.0747, "step": 594 }, { "epoch": 0.58, "learning_rate": 1.66889687374434e-05, "loss": 0.0771, "step": 595 }, { "epoch": 0.58, "learning_rate": 1.6677272188886484e-05, "loss": 0.0686, "step": 596 }, { "epoch": 0.58, "learning_rate": 1.666555913182372e-05, "loss": 0.0706, "step": 597 }, { "epoch": 0.58, "learning_rate": 1.6653829595213795e-05, "loss": 0.0738, "step": 598 }, { "epoch": 0.58, "learning_rate": 1.6642083608056145e-05, "loss": 0.0746, "step": 599 }, { "epoch": 0.58, "learning_rate": 1.6630321199390868e-05, "loss": 0.0716, "step": 600 }, { "epoch": 0.58, "learning_rate": 1.661854239829868e-05, "loss": 0.0724, "step": 601 }, { "epoch": 0.58, "learning_rate": 1.6606747233900816e-05, "loss": 0.0709, "step": 602 }, { "epoch": 0.59, "learning_rate": 1.659493573535896e-05, "loss": 0.0779, "step": 603 }, { "epoch": 0.59, "learning_rate": 1.6583107931875192e-05, "loss": 0.0663, "step": 604 }, { "epoch": 0.59, "learning_rate": 1.6571263852691887e-05, "loss": 0.0767, "step": 605 }, { "epoch": 0.59, "learning_rate": 1.6559403527091677e-05, "loss": 0.0675, "step": 606 }, { "epoch": 0.59, "learning_rate": 1.6547526984397352e-05, "loss": 0.0689, "step": 607 }, { "epoch": 0.59, "learning_rate": 1.6535634253971795e-05, "loss": 0.0711, "step": 608 }, { "epoch": 0.59, "learning_rate": 1.6523725365217914e-05, "loss": 0.0712, "step": 609 }, { "epoch": 0.59, "learning_rate": 1.651180034757856e-05, "loss": 0.0749, "step": 610 }, { "epoch": 0.59, "learning_rate": 1.6499859230536468e-05, "loss": 0.0619, "step": 611 }, { "epoch": 0.59, "learning_rate": 1.648790204361417e-05, "loss": 0.0689, "step": 612 }, { "epoch": 0.6, "learning_rate": 1.6475928816373942e-05, "loss": 0.0692, "step": 613 }, { "epoch": 0.6, "learning_rate": 1.6463939578417695e-05, "loss": 0.0695, "step": 614 }, { "epoch": 0.6, "learning_rate": 1.6451934359386935e-05, "loss": 0.0728, "step": 615 }, { "epoch": 0.6, "learning_rate": 1.6439913188962683e-05, "loss": 0.069, "step": 616 }, { "epoch": 0.6, "learning_rate": 1.6427876096865394e-05, "loss": 0.0721, "step": 617 }, { "epoch": 0.6, "learning_rate": 1.6415823112854885e-05, "loss": 0.0773, "step": 618 }, { "epoch": 0.6, "learning_rate": 1.640375426673026e-05, "loss": 0.0657, "step": 619 }, { "epoch": 0.6, "learning_rate": 1.639166958832985e-05, "loss": 0.0749, "step": 620 }, { "epoch": 0.6, "learning_rate": 1.637956910753112e-05, "loss": 0.0702, "step": 621 }, { "epoch": 0.6, "learning_rate": 1.6367452854250604e-05, "loss": 0.0708, "step": 622 }, { "epoch": 0.6, "learning_rate": 1.635532085844384e-05, "loss": 0.0745, "step": 623 }, { "epoch": 0.61, "learning_rate": 1.634317315010528e-05, "loss": 0.0729, "step": 624 }, { "epoch": 0.61, "learning_rate": 1.6331009759268216e-05, "loss": 0.0657, "step": 625 }, { "epoch": 0.61, "learning_rate": 1.6318830716004722e-05, "loss": 0.0643, "step": 626 }, { "epoch": 0.61, "learning_rate": 1.630663605042557e-05, "loss": 0.0656, "step": 627 }, { "epoch": 0.61, "learning_rate": 1.629442579268016e-05, "loss": 0.0646, "step": 628 }, { "epoch": 0.61, "learning_rate": 1.6282199972956425e-05, "loss": 0.0692, "step": 629 }, { "epoch": 0.61, "learning_rate": 1.626995862148079e-05, "loss": 0.0629, "step": 630 }, { "epoch": 0.61, "learning_rate": 1.625770176851806e-05, "loss": 0.0627, "step": 631 }, { "epoch": 0.61, "learning_rate": 1.6245429444371392e-05, "loss": 0.0696, "step": 632 }, { "epoch": 0.61, "learning_rate": 1.623314167938217e-05, "loss": 0.0719, "step": 633 }, { "epoch": 0.62, "learning_rate": 1.622083850392996e-05, "loss": 0.065, "step": 634 }, { "epoch": 0.62, "learning_rate": 1.6208519948432438e-05, "loss": 0.0694, "step": 635 }, { "epoch": 0.62, "learning_rate": 1.619618604334529e-05, "loss": 0.0632, "step": 636 }, { "epoch": 0.62, "learning_rate": 1.6183836819162152e-05, "loss": 0.0735, "step": 637 }, { "epoch": 0.62, "learning_rate": 1.6171472306414554e-05, "loss": 0.0654, "step": 638 }, { "epoch": 0.62, "learning_rate": 1.61590925356718e-05, "loss": 0.0716, "step": 639 }, { "epoch": 0.62, "learning_rate": 1.6146697537540926e-05, "loss": 0.0706, "step": 640 }, { "epoch": 0.62, "learning_rate": 1.613428734266662e-05, "loss": 0.068, "step": 641 }, { "epoch": 0.62, "learning_rate": 1.6121861981731137e-05, "loss": 0.0677, "step": 642 }, { "epoch": 0.62, "learning_rate": 1.610942148545423e-05, "loss": 0.0749, "step": 643 }, { "epoch": 0.63, "learning_rate": 1.609696588459307e-05, "loss": 0.0662, "step": 644 }, { "epoch": 0.63, "learning_rate": 1.6084495209942176e-05, "loss": 0.0695, "step": 645 }, { "epoch": 0.63, "learning_rate": 1.607200949233332e-05, "loss": 0.072, "step": 646 }, { "epoch": 0.63, "learning_rate": 1.6059508762635482e-05, "loss": 0.079, "step": 647 }, { "epoch": 0.63, "learning_rate": 1.6046993051754755e-05, "loss": 0.071, "step": 648 }, { "epoch": 0.63, "learning_rate": 1.6034462390634265e-05, "loss": 0.0656, "step": 649 }, { "epoch": 0.63, "learning_rate": 1.6021916810254096e-05, "loss": 0.0751, "step": 650 }, { "epoch": 0.63, "learning_rate": 1.600935634163123e-05, "loss": 0.0665, "step": 651 }, { "epoch": 0.63, "learning_rate": 1.599678101581945e-05, "loss": 0.0683, "step": 652 }, { "epoch": 0.63, "learning_rate": 1.598419086390927e-05, "loss": 0.0696, "step": 653 }, { "epoch": 0.63, "learning_rate": 1.5971585917027864e-05, "loss": 0.0651, "step": 654 }, { "epoch": 0.64, "learning_rate": 1.595896620633898e-05, "loss": 0.0723, "step": 655 }, { "epoch": 0.64, "learning_rate": 1.594633176304287e-05, "loss": 0.0757, "step": 656 }, { "epoch": 0.64, "learning_rate": 1.5933682618376204e-05, "loss": 0.0769, "step": 657 }, { "epoch": 0.64, "learning_rate": 1.5921018803612014e-05, "loss": 0.0656, "step": 658 }, { "epoch": 0.64, "learning_rate": 1.5908340350059582e-05, "loss": 0.0786, "step": 659 }, { "epoch": 0.64, "learning_rate": 1.5895647289064395e-05, "loss": 0.0694, "step": 660 }, { "epoch": 0.64, "learning_rate": 1.5882939652008053e-05, "loss": 0.0657, "step": 661 }, { "epoch": 0.64, "learning_rate": 1.5870217470308185e-05, "loss": 0.0777, "step": 662 }, { "epoch": 0.64, "learning_rate": 1.5857480775418396e-05, "loss": 0.0765, "step": 663 }, { "epoch": 0.64, "learning_rate": 1.584472959882815e-05, "loss": 0.0719, "step": 664 }, { "epoch": 0.65, "learning_rate": 1.5831963972062734e-05, "loss": 0.069, "step": 665 }, { "epoch": 0.65, "learning_rate": 1.581918392668315e-05, "loss": 0.0797, "step": 666 }, { "epoch": 0.65, "learning_rate": 1.5806389494286058e-05, "loss": 0.068, "step": 667 }, { "epoch": 0.65, "learning_rate": 1.5793580706503672e-05, "loss": 0.0708, "step": 668 }, { "epoch": 0.65, "learning_rate": 1.5780757595003707e-05, "loss": 0.0715, "step": 669 }, { "epoch": 0.65, "learning_rate": 1.5767920191489297e-05, "loss": 0.0652, "step": 670 }, { "epoch": 0.65, "learning_rate": 1.5755068527698904e-05, "loss": 0.0693, "step": 671 }, { "epoch": 0.65, "learning_rate": 1.5742202635406236e-05, "loss": 0.071, "step": 672 }, { "epoch": 0.65, "learning_rate": 1.5729322546420204e-05, "loss": 0.0655, "step": 673 }, { "epoch": 0.65, "learning_rate": 1.5716428292584788e-05, "loss": 0.0672, "step": 674 }, { "epoch": 0.66, "learning_rate": 1.5703519905779017e-05, "loss": 0.0622, "step": 675 }, { "epoch": 0.66, "learning_rate": 1.569059741791684e-05, "loss": 0.0765, "step": 676 }, { "epoch": 0.66, "learning_rate": 1.567766086094708e-05, "loss": 0.0805, "step": 677 }, { "epoch": 0.66, "learning_rate": 1.566471026685334e-05, "loss": 0.0671, "step": 678 }, { "epoch": 0.66, "learning_rate": 1.565174566765393e-05, "loss": 0.0667, "step": 679 }, { "epoch": 0.66, "learning_rate": 1.5638767095401778e-05, "loss": 0.0719, "step": 680 }, { "epoch": 0.66, "learning_rate": 1.5625774582184374e-05, "loss": 0.0698, "step": 681 }, { "epoch": 0.66, "learning_rate": 1.5612768160123654e-05, "loss": 0.075, "step": 682 }, { "epoch": 0.66, "learning_rate": 1.5599747861375957e-05, "loss": 0.0654, "step": 683 }, { "epoch": 0.66, "learning_rate": 1.5586713718131924e-05, "loss": 0.0704, "step": 684 }, { "epoch": 0.67, "learning_rate": 1.557366576261642e-05, "loss": 0.0704, "step": 685 }, { "epoch": 0.67, "learning_rate": 1.556060402708848e-05, "loss": 0.0695, "step": 686 }, { "epoch": 0.67, "learning_rate": 1.5547528543841176e-05, "loss": 0.0671, "step": 687 }, { "epoch": 0.67, "learning_rate": 1.5534439345201588e-05, "loss": 0.0653, "step": 688 }, { "epoch": 0.67, "learning_rate": 1.5521336463530705e-05, "loss": 0.0648, "step": 689 }, { "epoch": 0.67, "learning_rate": 1.550821993122334e-05, "loss": 0.0677, "step": 690 }, { "epoch": 0.67, "learning_rate": 1.5495089780708062e-05, "loss": 0.0639, "step": 691 }, { "epoch": 0.67, "learning_rate": 1.54819460444471e-05, "loss": 0.0638, "step": 692 }, { "epoch": 0.67, "learning_rate": 1.5468788754936275e-05, "loss": 0.0666, "step": 693 }, { "epoch": 0.67, "learning_rate": 1.545561794470492e-05, "loss": 0.0608, "step": 694 }, { "epoch": 0.67, "learning_rate": 1.5442433646315792e-05, "loss": 0.0608, "step": 695 }, { "epoch": 0.68, "learning_rate": 1.5429235892364996e-05, "loss": 0.0714, "step": 696 }, { "epoch": 0.68, "learning_rate": 1.5416024715481902e-05, "loss": 0.075, "step": 697 }, { "epoch": 0.68, "learning_rate": 1.540280014832907e-05, "loss": 0.0614, "step": 698 }, { "epoch": 0.68, "learning_rate": 1.5389562223602168e-05, "loss": 0.0775, "step": 699 }, { "epoch": 0.68, "learning_rate": 1.5376310974029872e-05, "loss": 0.0661, "step": 700 }, { "epoch": 0.68, "learning_rate": 1.5363046432373824e-05, "loss": 0.0695, "step": 701 }, { "epoch": 0.68, "learning_rate": 1.534976863142852e-05, "loss": 0.0768, "step": 702 }, { "epoch": 0.68, "learning_rate": 1.5336477604021226e-05, "loss": 0.0641, "step": 703 }, { "epoch": 0.68, "learning_rate": 1.5323173383011923e-05, "loss": 0.0732, "step": 704 }, { "epoch": 0.68, "learning_rate": 1.5309856001293203e-05, "loss": 0.0684, "step": 705 }, { "epoch": 0.69, "learning_rate": 1.5296525491790206e-05, "loss": 0.0639, "step": 706 }, { "epoch": 0.69, "learning_rate": 1.5283181887460516e-05, "loss": 0.0716, "step": 707 }, { "epoch": 0.69, "learning_rate": 1.52698252212941e-05, "loss": 0.074, "step": 708 }, { "epoch": 0.69, "learning_rate": 1.5256455526313208e-05, "loss": 0.0602, "step": 709 }, { "epoch": 0.69, "learning_rate": 1.5243072835572319e-05, "loss": 0.0706, "step": 710 }, { "epoch": 0.69, "learning_rate": 1.5229677182158026e-05, "loss": 0.0652, "step": 711 }, { "epoch": 0.69, "learning_rate": 1.5216268599188979e-05, "loss": 0.0648, "step": 712 }, { "epoch": 0.69, "learning_rate": 1.5202847119815792e-05, "loss": 0.0681, "step": 713 }, { "epoch": 0.69, "learning_rate": 1.518941277722096e-05, "loss": 0.0745, "step": 714 }, { "epoch": 0.69, "learning_rate": 1.5175965604618785e-05, "loss": 0.0733, "step": 715 }, { "epoch": 0.7, "learning_rate": 1.516250563525529e-05, "loss": 0.0668, "step": 716 }, { "epoch": 0.7, "learning_rate": 1.5149032902408129e-05, "loss": 0.066, "step": 717 }, { "epoch": 0.7, "learning_rate": 1.5135547439386518e-05, "loss": 0.072, "step": 718 }, { "epoch": 0.7, "learning_rate": 1.5122049279531143e-05, "loss": 0.0696, "step": 719 }, { "epoch": 0.7, "learning_rate": 1.5108538456214088e-05, "loss": 0.0672, "step": 720 }, { "epoch": 0.7, "learning_rate": 1.5095015002838734e-05, "loss": 0.0631, "step": 721 }, { "epoch": 0.7, "learning_rate": 1.5081478952839694e-05, "loss": 0.0699, "step": 722 }, { "epoch": 0.7, "learning_rate": 1.5067930339682728e-05, "loss": 0.0728, "step": 723 }, { "epoch": 0.7, "learning_rate": 1.5054369196864644e-05, "loss": 0.0719, "step": 724 }, { "epoch": 0.7, "learning_rate": 1.5040795557913246e-05, "loss": 0.0681, "step": 725 }, { "epoch": 0.7, "learning_rate": 1.5027209456387219e-05, "loss": 0.0628, "step": 726 }, { "epoch": 0.71, "learning_rate": 1.5013610925876058e-05, "loss": 0.0649, "step": 727 }, { "epoch": 0.71, "learning_rate": 1.5000000000000002e-05, "loss": 0.0626, "step": 728 }, { "epoch": 0.71, "learning_rate": 1.4986376712409915e-05, "loss": 0.0706, "step": 729 }, { "epoch": 0.71, "learning_rate": 1.4972741096787242e-05, "loss": 0.071, "step": 730 }, { "epoch": 0.71, "learning_rate": 1.4959093186843896e-05, "loss": 0.0697, "step": 731 }, { "epoch": 0.71, "learning_rate": 1.494543301632219e-05, "loss": 0.0632, "step": 732 }, { "epoch": 0.71, "learning_rate": 1.4931760618994746e-05, "loss": 0.0672, "step": 733 }, { "epoch": 0.71, "learning_rate": 1.491807602866442e-05, "loss": 0.0721, "step": 734 }, { "epoch": 0.71, "learning_rate": 1.4904379279164207e-05, "loss": 0.0785, "step": 735 }, { "epoch": 0.71, "learning_rate": 1.489067040435717e-05, "loss": 0.0711, "step": 736 }, { "epoch": 0.72, "learning_rate": 1.4876949438136348e-05, "loss": 0.0691, "step": 737 }, { "epoch": 0.72, "learning_rate": 1.486321641442467e-05, "loss": 0.0694, "step": 738 }, { "epoch": 0.72, "learning_rate": 1.4849471367174877e-05, "loss": 0.0637, "step": 739 }, { "epoch": 0.72, "learning_rate": 1.4835714330369445e-05, "loss": 0.0624, "step": 740 }, { "epoch": 0.72, "learning_rate": 1.482194533802048e-05, "loss": 0.0673, "step": 741 }, { "epoch": 0.72, "learning_rate": 1.480816442416965e-05, "loss": 0.0679, "step": 742 }, { "epoch": 0.72, "learning_rate": 1.47943716228881e-05, "loss": 0.0682, "step": 743 }, { "epoch": 0.72, "learning_rate": 1.4780566968276361e-05, "loss": 0.0689, "step": 744 }, { "epoch": 0.72, "learning_rate": 1.4766750494464274e-05, "loss": 0.0678, "step": 745 }, { "epoch": 0.72, "learning_rate": 1.47529222356109e-05, "loss": 0.0751, "step": 746 }, { "epoch": 0.73, "learning_rate": 1.4739082225904432e-05, "loss": 0.0641, "step": 747 }, { "epoch": 0.73, "learning_rate": 1.472523049956212e-05, "loss": 0.0703, "step": 748 }, { "epoch": 0.73, "learning_rate": 1.471136709083018e-05, "loss": 0.0639, "step": 749 }, { "epoch": 0.73, "learning_rate": 1.469749203398371e-05, "loss": 0.073, "step": 750 }, { "epoch": 0.73, "learning_rate": 1.4683605363326607e-05, "loss": 0.0656, "step": 751 }, { "epoch": 0.73, "learning_rate": 1.4669707113191483e-05, "loss": 0.0619, "step": 752 }, { "epoch": 0.73, "learning_rate": 1.4655797317939575e-05, "loss": 0.0665, "step": 753 }, { "epoch": 0.73, "learning_rate": 1.4641876011960661e-05, "loss": 0.0702, "step": 754 }, { "epoch": 0.73, "learning_rate": 1.4627943229672992e-05, "loss": 0.0751, "step": 755 }, { "epoch": 0.73, "learning_rate": 1.4613999005523173e-05, "loss": 0.0636, "step": 756 }, { "epoch": 0.73, "learning_rate": 1.460004337398611e-05, "loss": 0.0659, "step": 757 }, { "epoch": 0.74, "learning_rate": 1.458607636956491e-05, "loss": 0.0709, "step": 758 }, { "epoch": 0.74, "learning_rate": 1.4572098026790793e-05, "loss": 0.0634, "step": 759 }, { "epoch": 0.74, "learning_rate": 1.4558108380223013e-05, "loss": 0.0627, "step": 760 }, { "epoch": 0.74, "learning_rate": 1.4544107464448776e-05, "loss": 0.0715, "step": 761 }, { "epoch": 0.74, "learning_rate": 1.4530095314083144e-05, "loss": 0.0677, "step": 762 }, { "epoch": 0.74, "learning_rate": 1.4516071963768952e-05, "loss": 0.0697, "step": 763 }, { "epoch": 0.74, "learning_rate": 1.4502037448176734e-05, "loss": 0.0622, "step": 764 }, { "epoch": 0.74, "learning_rate": 1.4487991802004625e-05, "loss": 0.0715, "step": 765 }, { "epoch": 0.74, "learning_rate": 1.447393505997827e-05, "loss": 0.0664, "step": 766 }, { "epoch": 0.74, "learning_rate": 1.445986725685076e-05, "loss": 0.0692, "step": 767 }, { "epoch": 0.75, "learning_rate": 1.4445788427402527e-05, "loss": 0.0721, "step": 768 }, { "epoch": 0.75, "learning_rate": 1.4431698606441262e-05, "loss": 0.0651, "step": 769 }, { "epoch": 0.75, "learning_rate": 1.4417597828801833e-05, "loss": 0.0714, "step": 770 }, { "epoch": 0.75, "learning_rate": 1.4403486129346195e-05, "loss": 0.0675, "step": 771 }, { "epoch": 0.75, "learning_rate": 1.4389363542963307e-05, "loss": 0.0635, "step": 772 }, { "epoch": 0.75, "learning_rate": 1.4375230104569044e-05, "loss": 0.0619, "step": 773 }, { "epoch": 0.75, "learning_rate": 1.436108584910611e-05, "loss": 0.0764, "step": 774 }, { "epoch": 0.75, "learning_rate": 1.4346930811543958e-05, "loss": 0.0601, "step": 775 }, { "epoch": 0.75, "learning_rate": 1.4332765026878688e-05, "loss": 0.0703, "step": 776 }, { "epoch": 0.75, "learning_rate": 1.4318588530132976e-05, "loss": 0.0561, "step": 777 }, { "epoch": 0.76, "learning_rate": 1.4304401356355984e-05, "loss": 0.0649, "step": 778 }, { "epoch": 0.76, "learning_rate": 1.4290203540623267e-05, "loss": 0.0587, "step": 779 }, { "epoch": 0.76, "learning_rate": 1.4275995118036694e-05, "loss": 0.0651, "step": 780 }, { "epoch": 0.76, "learning_rate": 1.4261776123724355e-05, "loss": 0.0681, "step": 781 }, { "epoch": 0.76, "learning_rate": 1.424754659284048e-05, "loss": 0.0727, "step": 782 }, { "epoch": 0.76, "learning_rate": 1.4233306560565345e-05, "loss": 0.0688, "step": 783 }, { "epoch": 0.76, "learning_rate": 1.4219056062105195e-05, "loss": 0.0739, "step": 784 }, { "epoch": 0.76, "learning_rate": 1.4204795132692146e-05, "loss": 0.066, "step": 785 }, { "epoch": 0.76, "learning_rate": 1.41905238075841e-05, "loss": 0.0641, "step": 786 }, { "epoch": 0.76, "learning_rate": 1.4176242122064671e-05, "loss": 0.0692, "step": 787 }, { "epoch": 0.77, "learning_rate": 1.4161950111443077e-05, "loss": 0.0689, "step": 788 }, { "epoch": 0.77, "learning_rate": 1.4147647811054073e-05, "loss": 0.0657, "step": 789 }, { "epoch": 0.77, "learning_rate": 1.4133335256257841e-05, "loss": 0.0622, "step": 790 }, { "epoch": 0.77, "learning_rate": 1.4119012482439929e-05, "loss": 0.0746, "step": 791 }, { "epoch": 0.77, "learning_rate": 1.410467952501114e-05, "loss": 0.0684, "step": 792 }, { "epoch": 0.77, "learning_rate": 1.409033641940746e-05, "loss": 0.0636, "step": 793 }, { "epoch": 0.77, "learning_rate": 1.4075983201089964e-05, "loss": 0.081, "step": 794 }, { "epoch": 0.77, "learning_rate": 1.4061619905544726e-05, "loss": 0.0699, "step": 795 }, { "epoch": 0.77, "learning_rate": 1.4047246568282738e-05, "loss": 0.0709, "step": 796 }, { "epoch": 0.77, "learning_rate": 1.4032863224839815e-05, "loss": 0.0728, "step": 797 }, { "epoch": 0.77, "learning_rate": 1.4018469910776513e-05, "loss": 0.0706, "step": 798 }, { "epoch": 0.78, "learning_rate": 1.400406666167804e-05, "loss": 0.0729, "step": 799 }, { "epoch": 0.78, "learning_rate": 1.3989653513154165e-05, "loss": 0.067, "step": 800 }, { "epoch": 0.78, "learning_rate": 1.3975230500839132e-05, "loss": 0.0709, "step": 801 }, { "epoch": 0.78, "learning_rate": 1.396079766039157e-05, "loss": 0.0645, "step": 802 }, { "epoch": 0.78, "learning_rate": 1.394635502749441e-05, "loss": 0.07, "step": 803 }, { "epoch": 0.78, "learning_rate": 1.393190263785479e-05, "loss": 0.0688, "step": 804 }, { "epoch": 0.78, "learning_rate": 1.3917440527203977e-05, "loss": 0.0617, "step": 805 }, { "epoch": 0.78, "learning_rate": 1.3902968731297257e-05, "loss": 0.0723, "step": 806 }, { "epoch": 0.78, "learning_rate": 1.3888487285913876e-05, "loss": 0.0694, "step": 807 }, { "epoch": 0.78, "learning_rate": 1.3873996226856932e-05, "loss": 0.0651, "step": 808 }, { "epoch": 0.79, "learning_rate": 1.3859495589953289e-05, "loss": 0.0706, "step": 809 }, { "epoch": 0.79, "learning_rate": 1.3844985411053491e-05, "loss": 0.0714, "step": 810 }, { "epoch": 0.79, "learning_rate": 1.3830465726031679e-05, "loss": 0.0722, "step": 811 }, { "epoch": 0.79, "learning_rate": 1.3815936570785487e-05, "loss": 0.0612, "step": 812 }, { "epoch": 0.79, "learning_rate": 1.3801397981235966e-05, "loss": 0.074, "step": 813 }, { "epoch": 0.79, "learning_rate": 1.3786849993327503e-05, "loss": 0.0654, "step": 814 }, { "epoch": 0.79, "learning_rate": 1.37722926430277e-05, "loss": 0.0648, "step": 815 }, { "epoch": 0.79, "learning_rate": 1.3757725966327321e-05, "loss": 0.0784, "step": 816 }, { "epoch": 0.79, "learning_rate": 1.374314999924019e-05, "loss": 0.0678, "step": 817 }, { "epoch": 0.79, "learning_rate": 1.3728564777803089e-05, "loss": 0.0716, "step": 818 }, { "epoch": 0.8, "learning_rate": 1.3713970338075682e-05, "loss": 0.0704, "step": 819 }, { "epoch": 0.8, "learning_rate": 1.3699366716140434e-05, "loss": 0.0757, "step": 820 }, { "epoch": 0.8, "learning_rate": 1.36847539481025e-05, "loss": 0.067, "step": 821 }, { "epoch": 0.8, "learning_rate": 1.3670132070089654e-05, "loss": 0.0707, "step": 822 }, { "epoch": 0.8, "learning_rate": 1.365550111825219e-05, "loss": 0.0717, "step": 823 }, { "epoch": 0.8, "learning_rate": 1.3640861128762838e-05, "loss": 0.0768, "step": 824 }, { "epoch": 0.8, "learning_rate": 1.362621213781667e-05, "loss": 0.0681, "step": 825 }, { "epoch": 0.8, "learning_rate": 1.3611554181631012e-05, "loss": 0.0657, "step": 826 }, { "epoch": 0.8, "learning_rate": 1.359688729644536e-05, "loss": 0.0698, "step": 827 }, { "epoch": 0.8, "learning_rate": 1.3582211518521273e-05, "loss": 0.07, "step": 828 }, { "epoch": 0.8, "learning_rate": 1.3567526884142319e-05, "loss": 0.0677, "step": 829 }, { "epoch": 0.81, "learning_rate": 1.3552833429613939e-05, "loss": 0.0681, "step": 830 }, { "epoch": 0.81, "learning_rate": 1.3538131191263388e-05, "loss": 0.0619, "step": 831 }, { "epoch": 0.81, "learning_rate": 1.3523420205439645e-05, "loss": 0.0649, "step": 832 }, { "epoch": 0.81, "learning_rate": 1.3508700508513306e-05, "loss": 0.0697, "step": 833 }, { "epoch": 0.81, "learning_rate": 1.349397213687651e-05, "loss": 0.0685, "step": 834 }, { "epoch": 0.81, "learning_rate": 1.347923512694284e-05, "loss": 0.0656, "step": 835 }, { "epoch": 0.81, "learning_rate": 1.3464489515147239e-05, "loss": 0.0682, "step": 836 }, { "epoch": 0.81, "learning_rate": 1.3449735337945905e-05, "loss": 0.0705, "step": 837 }, { "epoch": 0.81, "learning_rate": 1.3434972631816234e-05, "loss": 0.0735, "step": 838 }, { "epoch": 0.81, "learning_rate": 1.342020143325669e-05, "loss": 0.0621, "step": 839 }, { "epoch": 0.82, "learning_rate": 1.3405421778786738e-05, "loss": 0.0668, "step": 840 }, { "epoch": 0.82, "learning_rate": 1.3390633704946754e-05, "loss": 0.0713, "step": 841 }, { "epoch": 0.82, "learning_rate": 1.3375837248297926e-05, "loss": 0.071, "step": 842 }, { "epoch": 0.82, "learning_rate": 1.3361032445422162e-05, "loss": 0.0623, "step": 843 }, { "epoch": 0.82, "learning_rate": 1.3346219332922017e-05, "loss": 0.0714, "step": 844 }, { "epoch": 0.82, "learning_rate": 1.3331397947420578e-05, "loss": 0.0625, "step": 845 }, { "epoch": 0.82, "learning_rate": 1.3316568325561392e-05, "loss": 0.0678, "step": 846 }, { "epoch": 0.82, "learning_rate": 1.3301730504008368e-05, "loss": 0.0687, "step": 847 }, { "epoch": 0.82, "learning_rate": 1.328688451944569e-05, "loss": 0.0715, "step": 848 }, { "epoch": 0.82, "learning_rate": 1.327203040857772e-05, "loss": 0.0669, "step": 849 }, { "epoch": 0.83, "learning_rate": 1.3257168208128907e-05, "loss": 0.0671, "step": 850 }, { "epoch": 0.83, "learning_rate": 1.3242297954843713e-05, "loss": 0.0636, "step": 851 }, { "epoch": 0.83, "learning_rate": 1.3227419685486493e-05, "loss": 0.067, "step": 852 }, { "epoch": 0.83, "learning_rate": 1.321253343684144e-05, "loss": 0.0638, "step": 853 }, { "epoch": 0.83, "learning_rate": 1.3197639245712454e-05, "loss": 0.0731, "step": 854 }, { "epoch": 0.83, "learning_rate": 1.3182737148923086e-05, "loss": 0.0652, "step": 855 }, { "epoch": 0.83, "learning_rate": 1.316782718331643e-05, "loss": 0.0681, "step": 856 }, { "epoch": 0.83, "learning_rate": 1.3152909385755027e-05, "loss": 0.0647, "step": 857 }, { "epoch": 0.83, "learning_rate": 1.3137983793120788e-05, "loss": 0.0724, "step": 858 }, { "epoch": 0.83, "learning_rate": 1.3123050442314896e-05, "loss": 0.0642, "step": 859 }, { "epoch": 0.83, "learning_rate": 1.3108109370257714e-05, "loss": 0.0667, "step": 860 }, { "epoch": 0.84, "learning_rate": 1.3093160613888691e-05, "loss": 0.0711, "step": 861 }, { "epoch": 0.84, "learning_rate": 1.3078204210166278e-05, "loss": 0.0706, "step": 862 }, { "epoch": 0.84, "learning_rate": 1.3063240196067837e-05, "loss": 0.0707, "step": 863 }, { "epoch": 0.84, "learning_rate": 1.3048268608589535e-05, "loss": 0.0754, "step": 864 }, { "epoch": 0.84, "learning_rate": 1.303328948474627e-05, "loss": 0.0671, "step": 865 }, { "epoch": 0.84, "learning_rate": 1.301830286157157e-05, "loss": 0.0707, "step": 866 }, { "epoch": 0.84, "learning_rate": 1.3003308776117504e-05, "loss": 0.0742, "step": 867 }, { "epoch": 0.84, "learning_rate": 1.2988307265454598e-05, "loss": 0.0656, "step": 868 }, { "epoch": 0.84, "learning_rate": 1.2973298366671724e-05, "loss": 0.0773, "step": 869 }, { "epoch": 0.84, "learning_rate": 1.2958282116876027e-05, "loss": 0.0731, "step": 870 }, { "epoch": 0.85, "learning_rate": 1.294325855319282e-05, "loss": 0.0654, "step": 871 }, { "epoch": 0.85, "learning_rate": 1.2928227712765504e-05, "loss": 0.0654, "step": 872 }, { "epoch": 0.85, "learning_rate": 1.2913189632755474e-05, "loss": 0.0743, "step": 873 }, { "epoch": 0.85, "learning_rate": 1.2898144350342015e-05, "loss": 0.0738, "step": 874 }, { "epoch": 0.85, "learning_rate": 1.288309190272222e-05, "loss": 0.066, "step": 875 }, { "epoch": 0.85, "learning_rate": 1.2868032327110904e-05, "loss": 0.0665, "step": 876 }, { "epoch": 0.85, "learning_rate": 1.28529656607405e-05, "loss": 0.08, "step": 877 }, { "epoch": 0.85, "learning_rate": 1.2837891940860972e-05, "loss": 0.0635, "step": 878 }, { "epoch": 0.85, "learning_rate": 1.2822811204739719e-05, "loss": 0.0653, "step": 879 }, { "epoch": 0.85, "learning_rate": 1.2807723489661497e-05, "loss": 0.0645, "step": 880 }, { "epoch": 0.86, "learning_rate": 1.2792628832928302e-05, "loss": 0.063, "step": 881 }, { "epoch": 0.86, "learning_rate": 1.2777527271859307e-05, "loss": 0.0652, "step": 882 }, { "epoch": 0.86, "learning_rate": 1.2762418843790745e-05, "loss": 0.0665, "step": 883 }, { "epoch": 0.86, "learning_rate": 1.2747303586075832e-05, "loss": 0.0703, "step": 884 }, { "epoch": 0.86, "learning_rate": 1.273218153608466e-05, "loss": 0.0736, "step": 885 }, { "epoch": 0.86, "learning_rate": 1.2717052731204129e-05, "loss": 0.0681, "step": 886 }, { "epoch": 0.86, "learning_rate": 1.2701917208837822e-05, "loss": 0.0645, "step": 887 }, { "epoch": 0.86, "learning_rate": 1.2686775006405948e-05, "loss": 0.0615, "step": 888 }, { "epoch": 0.86, "learning_rate": 1.2671626161345216e-05, "loss": 0.0672, "step": 889 }, { "epoch": 0.86, "learning_rate": 1.2656470711108763e-05, "loss": 0.0649, "step": 890 }, { "epoch": 0.87, "learning_rate": 1.2641308693166066e-05, "loss": 0.059, "step": 891 }, { "epoch": 0.87, "learning_rate": 1.262614014500282e-05, "loss": 0.0649, "step": 892 }, { "epoch": 0.87, "learning_rate": 1.2610965104120884e-05, "loss": 0.0711, "step": 893 }, { "epoch": 0.87, "learning_rate": 1.2595783608038157e-05, "loss": 0.0699, "step": 894 }, { "epoch": 0.87, "learning_rate": 1.2580595694288505e-05, "loss": 0.0675, "step": 895 }, { "epoch": 0.87, "learning_rate": 1.2565401400421652e-05, "loss": 0.0736, "step": 896 }, { "epoch": 0.87, "learning_rate": 1.2550200764003104e-05, "loss": 0.0668, "step": 897 }, { "epoch": 0.87, "learning_rate": 1.253499382261405e-05, "loss": 0.0625, "step": 898 }, { "epoch": 0.87, "learning_rate": 1.2519780613851254e-05, "loss": 0.0826, "step": 899 }, { "epoch": 0.87, "learning_rate": 1.2504561175326986e-05, "loss": 0.0709, "step": 900 }, { "epoch": 0.87, "learning_rate": 1.2489335544668919e-05, "loss": 0.0648, "step": 901 }, { "epoch": 0.88, "learning_rate": 1.2474103759520027e-05, "loss": 0.0663, "step": 902 }, { "epoch": 0.88, "learning_rate": 1.2458865857538505e-05, "loss": 0.0626, "step": 903 }, { "epoch": 0.88, "learning_rate": 1.2443621876397672e-05, "loss": 0.0736, "step": 904 }, { "epoch": 0.88, "learning_rate": 1.2428371853785872e-05, "loss": 0.0639, "step": 905 }, { "epoch": 0.88, "learning_rate": 1.2413115827406392e-05, "loss": 0.068, "step": 906 }, { "epoch": 0.88, "learning_rate": 1.239785383497736e-05, "loss": 0.0654, "step": 907 }, { "epoch": 0.88, "learning_rate": 1.238258591423165e-05, "loss": 0.0661, "step": 908 }, { "epoch": 0.88, "learning_rate": 1.2367312102916799e-05, "loss": 0.0665, "step": 909 }, { "epoch": 0.88, "learning_rate": 1.2352032438794904e-05, "loss": 0.0655, "step": 910 }, { "epoch": 0.88, "learning_rate": 1.2336746959642528e-05, "loss": 0.0655, "step": 911 }, { "epoch": 0.89, "learning_rate": 1.2321455703250615e-05, "loss": 0.073, "step": 912 }, { "epoch": 0.89, "learning_rate": 1.2306158707424402e-05, "loss": 0.0682, "step": 913 }, { "epoch": 0.89, "learning_rate": 1.22908560099833e-05, "loss": 0.0701, "step": 914 }, { "epoch": 0.89, "learning_rate": 1.2275547648760827e-05, "loss": 0.0657, "step": 915 }, { "epoch": 0.89, "learning_rate": 1.2260233661604493e-05, "loss": 0.0599, "step": 916 }, { "epoch": 0.89, "learning_rate": 1.2244914086375726e-05, "loss": 0.0659, "step": 917 }, { "epoch": 0.89, "learning_rate": 1.2229588960949772e-05, "loss": 0.0701, "step": 918 }, { "epoch": 0.89, "learning_rate": 1.2214258323215593e-05, "loss": 0.0638, "step": 919 }, { "epoch": 0.89, "learning_rate": 1.2198922211075779e-05, "loss": 0.0759, "step": 920 }, { "epoch": 0.89, "learning_rate": 1.2183580662446455e-05, "loss": 0.0653, "step": 921 }, { "epoch": 0.9, "learning_rate": 1.2168233715257194e-05, "loss": 0.0763, "step": 922 }, { "epoch": 0.9, "learning_rate": 1.2152881407450905e-05, "loss": 0.0653, "step": 923 }, { "epoch": 0.9, "learning_rate": 1.2137523776983758e-05, "loss": 0.0688, "step": 924 }, { "epoch": 0.9, "learning_rate": 1.2122160861825081e-05, "loss": 0.0705, "step": 925 }, { "epoch": 0.9, "learning_rate": 1.2106792699957264e-05, "loss": 0.0619, "step": 926 }, { "epoch": 0.9, "learning_rate": 1.2091419329375676e-05, "loss": 0.0641, "step": 927 }, { "epoch": 0.9, "learning_rate": 1.2076040788088555e-05, "loss": 0.0655, "step": 928 }, { "epoch": 0.9, "learning_rate": 1.2060657114116927e-05, "loss": 0.0637, "step": 929 }, { "epoch": 0.9, "learning_rate": 1.2045268345494513e-05, "loss": 0.0646, "step": 930 }, { "epoch": 0.9, "learning_rate": 1.2029874520267615e-05, "loss": 0.0697, "step": 931 }, { "epoch": 0.9, "learning_rate": 1.2014475676495052e-05, "loss": 0.0727, "step": 932 }, { "epoch": 0.91, "learning_rate": 1.1999071852248047e-05, "loss": 0.0614, "step": 933 }, { "epoch": 0.91, "learning_rate": 1.198366308561013e-05, "loss": 0.0664, "step": 934 }, { "epoch": 0.91, "learning_rate": 1.1968249414677055e-05, "loss": 0.0682, "step": 935 }, { "epoch": 0.91, "learning_rate": 1.19528308775567e-05, "loss": 0.0703, "step": 936 }, { "epoch": 0.91, "learning_rate": 1.1937407512368977e-05, "loss": 0.0702, "step": 937 }, { "epoch": 0.91, "learning_rate": 1.192197935724573e-05, "loss": 0.0672, "step": 938 }, { "epoch": 0.91, "learning_rate": 1.1906546450330648e-05, "loss": 0.064, "step": 939 }, { "epoch": 0.91, "learning_rate": 1.1891108829779166e-05, "loss": 0.0728, "step": 940 }, { "epoch": 0.91, "learning_rate": 1.1875666533758372e-05, "loss": 0.0656, "step": 941 }, { "epoch": 0.91, "learning_rate": 1.186021960044692e-05, "loss": 0.0671, "step": 942 }, { "epoch": 0.92, "learning_rate": 1.1844768068034926e-05, "loss": 0.0676, "step": 943 }, { "epoch": 0.92, "learning_rate": 1.1829311974723868e-05, "loss": 0.0688, "step": 944 }, { "epoch": 0.92, "learning_rate": 1.1813851358726513e-05, "loss": 0.0656, "step": 945 }, { "epoch": 0.92, "learning_rate": 1.17983862582668e-05, "loss": 0.062, "step": 946 }, { "epoch": 0.92, "learning_rate": 1.1782916711579759e-05, "loss": 0.0685, "step": 947 }, { "epoch": 0.92, "learning_rate": 1.1767442756911418e-05, "loss": 0.0764, "step": 948 }, { "epoch": 0.92, "learning_rate": 1.1751964432518693e-05, "loss": 0.0693, "step": 949 }, { "epoch": 0.92, "learning_rate": 1.1736481776669307e-05, "loss": 0.0669, "step": 950 }, { "epoch": 0.92, "learning_rate": 1.1720994827641692e-05, "loss": 0.0682, "step": 951 }, { "epoch": 0.92, "learning_rate": 1.17055036237249e-05, "loss": 0.0765, "step": 952 }, { "epoch": 0.93, "learning_rate": 1.1690008203218493e-05, "loss": 0.0653, "step": 953 }, { "epoch": 0.93, "learning_rate": 1.1674508604432464e-05, "loss": 0.0656, "step": 954 }, { "epoch": 0.93, "learning_rate": 1.1659004865687136e-05, "loss": 0.0716, "step": 955 }, { "epoch": 0.93, "learning_rate": 1.1643497025313063e-05, "loss": 0.0663, "step": 956 }, { "epoch": 0.93, "learning_rate": 1.1627985121650945e-05, "loss": 0.068, "step": 957 }, { "epoch": 0.93, "learning_rate": 1.1612469193051526e-05, "loss": 0.0726, "step": 958 }, { "epoch": 0.93, "learning_rate": 1.1596949277875496e-05, "loss": 0.0685, "step": 959 }, { "epoch": 0.93, "learning_rate": 1.158142541449341e-05, "loss": 0.0616, "step": 960 }, { "epoch": 0.93, "learning_rate": 1.1565897641285577e-05, "loss": 0.079, "step": 961 }, { "epoch": 0.93, "learning_rate": 1.155036599664198e-05, "loss": 0.0664, "step": 962 }, { "epoch": 0.93, "learning_rate": 1.1534830518962164e-05, "loss": 0.0665, "step": 963 }, { "epoch": 0.94, "learning_rate": 1.151929124665516e-05, "loss": 0.0675, "step": 964 }, { "epoch": 0.94, "learning_rate": 1.150374821813937e-05, "loss": 0.0644, "step": 965 }, { "epoch": 0.94, "learning_rate": 1.148820147184249e-05, "loss": 0.0618, "step": 966 }, { "epoch": 0.94, "learning_rate": 1.1472651046201413e-05, "loss": 0.0604, "step": 967 }, { "epoch": 0.94, "learning_rate": 1.1457096979662115e-05, "loss": 0.0653, "step": 968 }, { "epoch": 0.94, "learning_rate": 1.1441539310679587e-05, "loss": 0.0764, "step": 969 }, { "epoch": 0.94, "learning_rate": 1.142597807771771e-05, "loss": 0.0674, "step": 970 }, { "epoch": 0.94, "learning_rate": 1.1410413319249193e-05, "loss": 0.0715, "step": 971 }, { "epoch": 0.94, "learning_rate": 1.1394845073755456e-05, "loss": 0.07, "step": 972 }, { "epoch": 0.94, "learning_rate": 1.1379273379726538e-05, "loss": 0.0681, "step": 973 }, { "epoch": 0.95, "learning_rate": 1.1363698275661002e-05, "loss": 0.0674, "step": 974 }, { "epoch": 0.95, "learning_rate": 1.1348119800065842e-05, "loss": 0.0742, "step": 975 }, { "epoch": 0.95, "learning_rate": 1.1332537991456398e-05, "loss": 0.0679, "step": 976 }, { "epoch": 0.95, "learning_rate": 1.1316952888356236e-05, "loss": 0.0673, "step": 977 }, { "epoch": 0.95, "learning_rate": 1.1301364529297078e-05, "loss": 0.0671, "step": 978 }, { "epoch": 0.95, "learning_rate": 1.128577295281869e-05, "loss": 0.0693, "step": 979 }, { "epoch": 0.95, "learning_rate": 1.1270178197468788e-05, "loss": 0.0678, "step": 980 }, { "epoch": 0.95, "learning_rate": 1.1254580301802961e-05, "loss": 0.0628, "step": 981 }, { "epoch": 0.95, "learning_rate": 1.1238979304384554e-05, "loss": 0.0724, "step": 982 }, { "epoch": 0.95, "learning_rate": 1.1223375243784573e-05, "loss": 0.0691, "step": 983 }, { "epoch": 0.96, "learning_rate": 1.1207768158581614e-05, "loss": 0.0637, "step": 984 }, { "epoch": 0.96, "learning_rate": 1.1192158087361736e-05, "loss": 0.0706, "step": 985 }, { "epoch": 0.96, "learning_rate": 1.1176545068718386e-05, "loss": 0.061, "step": 986 }, { "epoch": 0.96, "learning_rate": 1.1160929141252303e-05, "loss": 0.0661, "step": 987 }, { "epoch": 0.96, "learning_rate": 1.1145310343571412e-05, "loss": 0.0671, "step": 988 }, { "epoch": 0.96, "learning_rate": 1.112968871429073e-05, "loss": 0.0634, "step": 989 }, { "epoch": 0.96, "learning_rate": 1.1114064292032283e-05, "loss": 0.0659, "step": 990 }, { "epoch": 0.96, "learning_rate": 1.1098437115425003e-05, "loss": 0.0633, "step": 991 }, { "epoch": 0.96, "learning_rate": 1.1082807223104618e-05, "loss": 0.0671, "step": 992 }, { "epoch": 0.96, "learning_rate": 1.106717465371359e-05, "loss": 0.0643, "step": 993 }, { "epoch": 0.97, "learning_rate": 1.1051539445900982e-05, "loss": 0.0739, "step": 994 }, { "epoch": 0.97, "learning_rate": 1.1035901638322392e-05, "loss": 0.0679, "step": 995 }, { "epoch": 0.97, "learning_rate": 1.1020261269639842e-05, "loss": 0.0723, "step": 996 }, { "epoch": 0.97, "learning_rate": 1.1004618378521683e-05, "loss": 0.0694, "step": 997 }, { "epoch": 0.97, "learning_rate": 1.09889730036425e-05, "loss": 0.069, "step": 998 }, { "epoch": 0.97, "learning_rate": 1.0973325183683029e-05, "loss": 0.0653, "step": 999 }, { "epoch": 0.97, "learning_rate": 1.0957674957330043e-05, "loss": 0.0687, "step": 1000 }, { "epoch": 0.97, "learning_rate": 1.0942022363276263e-05, "loss": 0.0595, "step": 1001 }, { "epoch": 0.97, "learning_rate": 1.0926367440220276e-05, "loss": 0.0636, "step": 1002 }, { "epoch": 0.97, "learning_rate": 1.091071022686641e-05, "loss": 0.0692, "step": 1003 }, { "epoch": 0.97, "learning_rate": 1.0895050761924667e-05, "loss": 0.058, "step": 1004 }, { "epoch": 0.98, "learning_rate": 1.0879389084110615e-05, "loss": 0.0631, "step": 1005 }, { "epoch": 0.98, "learning_rate": 1.0863725232145286e-05, "loss": 0.0725, "step": 1006 }, { "epoch": 0.98, "learning_rate": 1.0848059244755093e-05, "loss": 0.0573, "step": 1007 }, { "epoch": 0.98, "learning_rate": 1.0832391160671728e-05, "loss": 0.0718, "step": 1008 }, { "epoch": 0.98, "learning_rate": 1.0816721018632069e-05, "loss": 0.0695, "step": 1009 }, { "epoch": 0.98, "learning_rate": 1.080104885737807e-05, "loss": 0.0668, "step": 1010 }, { "epoch": 0.98, "learning_rate": 1.0785374715656698e-05, "loss": 0.0733, "step": 1011 }, { "epoch": 0.98, "learning_rate": 1.0769698632219794e-05, "loss": 0.067, "step": 1012 }, { "epoch": 0.98, "learning_rate": 1.0754020645824017e-05, "loss": 0.069, "step": 1013 }, { "epoch": 0.98, "learning_rate": 1.0738340795230722e-05, "loss": 0.063, "step": 1014 }, { "epoch": 0.99, "learning_rate": 1.0722659119205872e-05, "loss": 0.0577, "step": 1015 }, { "epoch": 0.99, "learning_rate": 1.0706975656519946e-05, "loss": 0.0614, "step": 1016 }, { "epoch": 0.99, "learning_rate": 1.0691290445947849e-05, "loss": 0.0622, "step": 1017 }, { "epoch": 0.99, "learning_rate": 1.0675603526268785e-05, "loss": 0.0678, "step": 1018 }, { "epoch": 0.99, "learning_rate": 1.0659914936266207e-05, "loss": 0.063, "step": 1019 }, { "epoch": 0.99, "learning_rate": 1.0644224714727683e-05, "loss": 0.059, "step": 1020 }, { "epoch": 0.99, "learning_rate": 1.062853290044482e-05, "loss": 0.0653, "step": 1021 }, { "epoch": 0.99, "learning_rate": 1.0612839532213164e-05, "loss": 0.0698, "step": 1022 }, { "epoch": 0.99, "learning_rate": 1.05971446488321e-05, "loss": 0.0675, "step": 1023 }, { "epoch": 0.99, "learning_rate": 1.0581448289104759e-05, "loss": 0.0641, "step": 1024 }, { "epoch": 1.0, "learning_rate": 1.0565750491837925e-05, "loss": 0.0598, "step": 1025 }, { "epoch": 1.0, "learning_rate": 1.0550051295841932e-05, "loss": 0.066, "step": 1026 }, { "epoch": 1.0, "learning_rate": 1.0534350739930575e-05, "loss": 0.0691, "step": 1027 }, { "epoch": 1.0, "learning_rate": 1.0518648862921013e-05, "loss": 0.0644, "step": 1028 }, { "epoch": 1.0, "learning_rate": 1.0502945703633665e-05, "loss": 0.068, "step": 1029 }, { "epoch": 1.0, "learning_rate": 1.0487241300892121e-05, "loss": 0.0706, "step": 1030 }, { "epoch": 1.0, "learning_rate": 1.047153569352306e-05, "loss": 0.0646, "step": 1031 }, { "epoch": 1.0, "learning_rate": 1.0455828920356114e-05, "loss": 0.0576, "step": 1032 }, { "epoch": 1.0, "learning_rate": 1.0440121020223817e-05, "loss": 0.0589, "step": 1033 }, { "epoch": 1.0, "learning_rate": 1.0424412031961485e-05, "loss": 0.0573, "step": 1034 }, { "epoch": 1.0, "learning_rate": 1.0408701994407117e-05, "loss": 0.0529, "step": 1035 }, { "epoch": 1.01, "learning_rate": 1.0392990946401313e-05, "loss": 0.0603, "step": 1036 }, { "epoch": 1.01, "learning_rate": 1.0377278926787174e-05, "loss": 0.055, "step": 1037 }, { "epoch": 1.01, "learning_rate": 1.0361565974410192e-05, "loss": 0.053, "step": 1038 }, { "epoch": 1.01, "learning_rate": 1.0345852128118174e-05, "loss": 0.0575, "step": 1039 }, { "epoch": 1.01, "learning_rate": 1.0330137426761136e-05, "loss": 0.0518, "step": 1040 }, { "epoch": 1.01, "learning_rate": 1.0314421909191208e-05, "loss": 0.0554, "step": 1041 }, { "epoch": 1.01, "learning_rate": 1.0298705614262532e-05, "loss": 0.0445, "step": 1042 }, { "epoch": 1.01, "learning_rate": 1.0282988580831183e-05, "loss": 0.0603, "step": 1043 }, { "epoch": 1.01, "learning_rate": 1.0267270847755048e-05, "loss": 0.0599, "step": 1044 }, { "epoch": 1.01, "learning_rate": 1.0251552453893759e-05, "loss": 0.0632, "step": 1045 }, { "epoch": 1.02, "learning_rate": 1.023583343810857e-05, "loss": 0.0564, "step": 1046 }, { "epoch": 1.02, "learning_rate": 1.0220113839262278e-05, "loss": 0.0611, "step": 1047 }, { "epoch": 1.02, "learning_rate": 1.0204393696219116e-05, "loss": 0.0604, "step": 1048 }, { "epoch": 1.02, "learning_rate": 1.018867304784467e-05, "loss": 0.0515, "step": 1049 }, { "epoch": 1.02, "learning_rate": 1.0172951933005775e-05, "loss": 0.0623, "step": 1050 }, { "epoch": 1.02, "learning_rate": 1.015723039057041e-05, "loss": 0.0501, "step": 1051 }, { "epoch": 1.02, "learning_rate": 1.0141508459407622e-05, "loss": 0.0617, "step": 1052 }, { "epoch": 1.02, "learning_rate": 1.0125786178387413e-05, "loss": 0.0519, "step": 1053 }, { "epoch": 1.02, "learning_rate": 1.0110063586380647e-05, "loss": 0.0613, "step": 1054 }, { "epoch": 1.02, "learning_rate": 1.0094340722258969e-05, "loss": 0.0493, "step": 1055 }, { "epoch": 1.03, "learning_rate": 1.0078617624894684e-05, "loss": 0.0507, "step": 1056 }, { "epoch": 1.03, "learning_rate": 1.0062894333160677e-05, "loss": 0.0577, "step": 1057 }, { "epoch": 1.03, "learning_rate": 1.0047170885930324e-05, "loss": 0.0577, "step": 1058 }, { "epoch": 1.03, "learning_rate": 1.0031447322077365e-05, "loss": 0.0575, "step": 1059 }, { "epoch": 1.03, "learning_rate": 1.0015723680475847e-05, "loss": 0.0542, "step": 1060 }, { "epoch": 1.03, "learning_rate": 1e-05, "loss": 0.0497, "step": 1061 }, { "epoch": 1.03, "learning_rate": 9.984276319524155e-06, "loss": 0.0538, "step": 1062 }, { "epoch": 1.03, "learning_rate": 9.968552677922638e-06, "loss": 0.059, "step": 1063 }, { "epoch": 1.03, "learning_rate": 9.95282911406968e-06, "loss": 0.054, "step": 1064 }, { "epoch": 1.03, "learning_rate": 9.937105666839324e-06, "loss": 0.051, "step": 1065 }, { "epoch": 1.03, "learning_rate": 9.921382375105319e-06, "loss": 0.0528, "step": 1066 }, { "epoch": 1.04, "learning_rate": 9.905659277741031e-06, "loss": 0.0572, "step": 1067 }, { "epoch": 1.04, "learning_rate": 9.889936413619356e-06, "loss": 0.0555, "step": 1068 }, { "epoch": 1.04, "learning_rate": 9.874213821612589e-06, "loss": 0.0535, "step": 1069 }, { "epoch": 1.04, "learning_rate": 9.858491540592383e-06, "loss": 0.0548, "step": 1070 }, { "epoch": 1.04, "learning_rate": 9.842769609429592e-06, "loss": 0.0551, "step": 1071 }, { "epoch": 1.04, "learning_rate": 9.827048066994225e-06, "loss": 0.0565, "step": 1072 }, { "epoch": 1.04, "learning_rate": 9.811326952155331e-06, "loss": 0.0534, "step": 1073 }, { "epoch": 1.04, "learning_rate": 9.795606303780885e-06, "loss": 0.0605, "step": 1074 }, { "epoch": 1.04, "learning_rate": 9.779886160737727e-06, "loss": 0.0555, "step": 1075 }, { "epoch": 1.04, "learning_rate": 9.764166561891432e-06, "loss": 0.0558, "step": 1076 }, { "epoch": 1.05, "learning_rate": 9.748447546106243e-06, "loss": 0.0593, "step": 1077 }, { "epoch": 1.05, "learning_rate": 9.732729152244953e-06, "loss": 0.0517, "step": 1078 }, { "epoch": 1.05, "learning_rate": 9.71701141916882e-06, "loss": 0.0588, "step": 1079 }, { "epoch": 1.05, "learning_rate": 9.701294385737471e-06, "loss": 0.0531, "step": 1080 }, { "epoch": 1.05, "learning_rate": 9.685578090808795e-06, "loss": 0.053, "step": 1081 }, { "epoch": 1.05, "learning_rate": 9.669862573238864e-06, "loss": 0.0609, "step": 1082 }, { "epoch": 1.05, "learning_rate": 9.654147871881828e-06, "loss": 0.0599, "step": 1083 }, { "epoch": 1.05, "learning_rate": 9.63843402558981e-06, "loss": 0.0566, "step": 1084 }, { "epoch": 1.05, "learning_rate": 9.622721073212831e-06, "loss": 0.0575, "step": 1085 }, { "epoch": 1.05, "learning_rate": 9.607009053598689e-06, "loss": 0.0572, "step": 1086 }, { "epoch": 1.06, "learning_rate": 9.591298005592884e-06, "loss": 0.0534, "step": 1087 }, { "epoch": 1.06, "learning_rate": 9.57558796803852e-06, "loss": 0.0609, "step": 1088 }, { "epoch": 1.06, "learning_rate": 9.559878979776184e-06, "loss": 0.056, "step": 1089 }, { "epoch": 1.06, "learning_rate": 9.54417107964389e-06, "loss": 0.0469, "step": 1090 }, { "epoch": 1.06, "learning_rate": 9.528464306476946e-06, "loss": 0.0532, "step": 1091 }, { "epoch": 1.06, "learning_rate": 9.512758699107879e-06, "loss": 0.0504, "step": 1092 }, { "epoch": 1.06, "learning_rate": 9.49705429636634e-06, "loss": 0.0575, "step": 1093 }, { "epoch": 1.06, "learning_rate": 9.48135113707899e-06, "loss": 0.0584, "step": 1094 }, { "epoch": 1.06, "learning_rate": 9.465649260069429e-06, "loss": 0.0479, "step": 1095 }, { "epoch": 1.06, "learning_rate": 9.449948704158071e-06, "loss": 0.0553, "step": 1096 }, { "epoch": 1.07, "learning_rate": 9.434249508162076e-06, "loss": 0.0592, "step": 1097 }, { "epoch": 1.07, "learning_rate": 9.418551710895243e-06, "loss": 0.0518, "step": 1098 }, { "epoch": 1.07, "learning_rate": 9.402855351167901e-06, "loss": 0.0473, "step": 1099 }, { "epoch": 1.07, "learning_rate": 9.38716046778684e-06, "loss": 0.0554, "step": 1100 }, { "epoch": 1.07, "learning_rate": 9.371467099555182e-06, "loss": 0.0479, "step": 1101 }, { "epoch": 1.07, "learning_rate": 9.355775285272319e-06, "loss": 0.0533, "step": 1102 }, { "epoch": 1.07, "learning_rate": 9.340085063733798e-06, "loss": 0.0546, "step": 1103 }, { "epoch": 1.07, "learning_rate": 9.324396473731218e-06, "loss": 0.0496, "step": 1104 }, { "epoch": 1.07, "learning_rate": 9.308709554052158e-06, "loss": 0.0494, "step": 1105 }, { "epoch": 1.07, "learning_rate": 9.293024343480056e-06, "loss": 0.0482, "step": 1106 }, { "epoch": 1.07, "learning_rate": 9.277340880794132e-06, "loss": 0.0516, "step": 1107 }, { "epoch": 1.08, "learning_rate": 9.261659204769283e-06, "loss": 0.0501, "step": 1108 }, { "epoch": 1.08, "learning_rate": 9.245979354175985e-06, "loss": 0.0584, "step": 1109 }, { "epoch": 1.08, "learning_rate": 9.23030136778021e-06, "loss": 0.0593, "step": 1110 }, { "epoch": 1.08, "learning_rate": 9.214625284343305e-06, "loss": 0.0535, "step": 1111 }, { "epoch": 1.08, "learning_rate": 9.19895114262193e-06, "loss": 0.0578, "step": 1112 }, { "epoch": 1.08, "learning_rate": 9.183278981367935e-06, "loss": 0.0623, "step": 1113 }, { "epoch": 1.08, "learning_rate": 9.167608839328273e-06, "loss": 0.0523, "step": 1114 }, { "epoch": 1.08, "learning_rate": 9.151940755244912e-06, "loss": 0.0539, "step": 1115 }, { "epoch": 1.08, "learning_rate": 9.136274767854718e-06, "loss": 0.0604, "step": 1116 }, { "epoch": 1.08, "learning_rate": 9.120610915889387e-06, "loss": 0.0547, "step": 1117 }, { "epoch": 1.09, "learning_rate": 9.104949238075338e-06, "loss": 0.0525, "step": 1118 }, { "epoch": 1.09, "learning_rate": 9.089289773133594e-06, "loss": 0.0501, "step": 1119 }, { "epoch": 1.09, "learning_rate": 9.073632559779731e-06, "loss": 0.0647, "step": 1120 }, { "epoch": 1.09, "learning_rate": 9.057977636723739e-06, "loss": 0.0578, "step": 1121 }, { "epoch": 1.09, "learning_rate": 9.042325042669962e-06, "loss": 0.0574, "step": 1122 }, { "epoch": 1.09, "learning_rate": 9.026674816316976e-06, "loss": 0.0571, "step": 1123 }, { "epoch": 1.09, "learning_rate": 9.011026996357504e-06, "loss": 0.0548, "step": 1124 }, { "epoch": 1.09, "learning_rate": 8.99538162147832e-06, "loss": 0.0567, "step": 1125 }, { "epoch": 1.09, "learning_rate": 8.97973873036016e-06, "loss": 0.0575, "step": 1126 }, { "epoch": 1.09, "learning_rate": 8.964098361677607e-06, "loss": 0.0554, "step": 1127 }, { "epoch": 1.1, "learning_rate": 8.94846055409902e-06, "loss": 0.0632, "step": 1128 }, { "epoch": 1.1, "learning_rate": 8.932825346286412e-06, "loss": 0.0477, "step": 1129 }, { "epoch": 1.1, "learning_rate": 8.917192776895382e-06, "loss": 0.0578, "step": 1130 }, { "epoch": 1.1, "learning_rate": 8.901562884575002e-06, "loss": 0.055, "step": 1131 }, { "epoch": 1.1, "learning_rate": 8.885935707967716e-06, "loss": 0.0562, "step": 1132 }, { "epoch": 1.1, "learning_rate": 8.870311285709274e-06, "loss": 0.0525, "step": 1133 }, { "epoch": 1.1, "learning_rate": 8.854689656428591e-06, "loss": 0.0599, "step": 1134 }, { "epoch": 1.1, "learning_rate": 8.839070858747697e-06, "loss": 0.0501, "step": 1135 }, { "epoch": 1.1, "learning_rate": 8.823454931281616e-06, "loss": 0.0532, "step": 1136 }, { "epoch": 1.1, "learning_rate": 8.807841912638267e-06, "loss": 0.0543, "step": 1137 }, { "epoch": 1.1, "learning_rate": 8.792231841418391e-06, "loss": 0.0517, "step": 1138 }, { "epoch": 1.11, "learning_rate": 8.776624756215429e-06, "loss": 0.0609, "step": 1139 }, { "epoch": 1.11, "learning_rate": 8.76102069561545e-06, "loss": 0.0534, "step": 1140 }, { "epoch": 1.11, "learning_rate": 8.745419698197042e-06, "loss": 0.0594, "step": 1141 }, { "epoch": 1.11, "learning_rate": 8.729821802531213e-06, "loss": 0.0585, "step": 1142 }, { "epoch": 1.11, "learning_rate": 8.714227047181315e-06, "loss": 0.0507, "step": 1143 }, { "epoch": 1.11, "learning_rate": 8.698635470702925e-06, "loss": 0.0482, "step": 1144 }, { "epoch": 1.11, "learning_rate": 8.683047111643764e-06, "loss": 0.0509, "step": 1145 }, { "epoch": 1.11, "learning_rate": 8.667462008543604e-06, "loss": 0.0552, "step": 1146 }, { "epoch": 1.11, "learning_rate": 8.651880199934158e-06, "loss": 0.0499, "step": 1147 }, { "epoch": 1.11, "learning_rate": 8.636301724339005e-06, "loss": 0.0518, "step": 1148 }, { "epoch": 1.12, "learning_rate": 8.620726620273464e-06, "loss": 0.0566, "step": 1149 }, { "epoch": 1.12, "learning_rate": 8.605154926244544e-06, "loss": 0.0551, "step": 1150 }, { "epoch": 1.12, "learning_rate": 8.58958668075081e-06, "loss": 0.0541, "step": 1151 }, { "epoch": 1.12, "learning_rate": 8.574021922282292e-06, "loss": 0.0596, "step": 1152 }, { "epoch": 1.12, "learning_rate": 8.55846068932042e-06, "loss": 0.0519, "step": 1153 }, { "epoch": 1.12, "learning_rate": 8.542903020337887e-06, "loss": 0.0526, "step": 1154 }, { "epoch": 1.12, "learning_rate": 8.527348953798589e-06, "loss": 0.0594, "step": 1155 }, { "epoch": 1.12, "learning_rate": 8.511798528157513e-06, "loss": 0.0625, "step": 1156 }, { "epoch": 1.12, "learning_rate": 8.496251781860634e-06, "loss": 0.0523, "step": 1157 }, { "epoch": 1.12, "learning_rate": 8.480708753344846e-06, "loss": 0.0491, "step": 1158 }, { "epoch": 1.13, "learning_rate": 8.465169481037837e-06, "loss": 0.0553, "step": 1159 }, { "epoch": 1.13, "learning_rate": 8.449634003358022e-06, "loss": 0.0578, "step": 1160 }, { "epoch": 1.13, "learning_rate": 8.434102358714425e-06, "loss": 0.0495, "step": 1161 }, { "epoch": 1.13, "learning_rate": 8.418574585506592e-06, "loss": 0.0498, "step": 1162 }, { "epoch": 1.13, "learning_rate": 8.40305072212451e-06, "loss": 0.0561, "step": 1163 }, { "epoch": 1.13, "learning_rate": 8.387530806948478e-06, "loss": 0.0597, "step": 1164 }, { "epoch": 1.13, "learning_rate": 8.372014878349057e-06, "loss": 0.0562, "step": 1165 }, { "epoch": 1.13, "learning_rate": 8.356502974686942e-06, "loss": 0.0542, "step": 1166 }, { "epoch": 1.13, "learning_rate": 8.340995134312866e-06, "loss": 0.0551, "step": 1167 }, { "epoch": 1.13, "learning_rate": 8.325491395567541e-06, "loss": 0.0551, "step": 1168 }, { "epoch": 1.13, "learning_rate": 8.309991796781512e-06, "loss": 0.0572, "step": 1169 }, { "epoch": 1.14, "learning_rate": 8.294496376275104e-06, "loss": 0.0464, "step": 1170 }, { "epoch": 1.14, "learning_rate": 8.279005172358311e-06, "loss": 0.0526, "step": 1171 }, { "epoch": 1.14, "learning_rate": 8.263518223330698e-06, "loss": 0.056, "step": 1172 }, { "epoch": 1.14, "learning_rate": 8.248035567481313e-06, "loss": 0.0593, "step": 1173 }, { "epoch": 1.14, "learning_rate": 8.232557243088585e-06, "loss": 0.0533, "step": 1174 }, { "epoch": 1.14, "learning_rate": 8.217083288420241e-06, "loss": 0.0543, "step": 1175 }, { "epoch": 1.14, "learning_rate": 8.201613741733203e-06, "loss": 0.0567, "step": 1176 }, { "epoch": 1.14, "learning_rate": 8.18614864127349e-06, "loss": 0.0519, "step": 1177 }, { "epoch": 1.14, "learning_rate": 8.170688025276134e-06, "loss": 0.0546, "step": 1178 }, { "epoch": 1.14, "learning_rate": 8.155231931965077e-06, "loss": 0.0529, "step": 1179 }, { "epoch": 1.15, "learning_rate": 8.13978039955308e-06, "loss": 0.0523, "step": 1180 }, { "epoch": 1.15, "learning_rate": 8.124333466241632e-06, "loss": 0.0568, "step": 1181 }, { "epoch": 1.15, "learning_rate": 8.108891170220835e-06, "loss": 0.0584, "step": 1182 }, { "epoch": 1.15, "learning_rate": 8.093453549669357e-06, "loss": 0.0466, "step": 1183 }, { "epoch": 1.15, "learning_rate": 8.078020642754273e-06, "loss": 0.0511, "step": 1184 }, { "epoch": 1.15, "learning_rate": 8.062592487631023e-06, "loss": 0.0633, "step": 1185 }, { "epoch": 1.15, "learning_rate": 8.047169122443302e-06, "loss": 0.0602, "step": 1186 }, { "epoch": 1.15, "learning_rate": 8.031750585322948e-06, "loss": 0.0557, "step": 1187 }, { "epoch": 1.15, "learning_rate": 8.016336914389875e-06, "loss": 0.0526, "step": 1188 }, { "epoch": 1.15, "learning_rate": 8.000928147751955e-06, "loss": 0.0582, "step": 1189 }, { "epoch": 1.16, "learning_rate": 7.985524323504948e-06, "loss": 0.0559, "step": 1190 }, { "epoch": 1.16, "learning_rate": 7.970125479732389e-06, "loss": 0.0517, "step": 1191 }, { "epoch": 1.16, "learning_rate": 7.954731654505492e-06, "loss": 0.0536, "step": 1192 }, { "epoch": 1.16, "learning_rate": 7.939342885883078e-06, "loss": 0.0531, "step": 1193 }, { "epoch": 1.16, "learning_rate": 7.923959211911448e-06, "loss": 0.0507, "step": 1194 }, { "epoch": 1.16, "learning_rate": 7.908580670624325e-06, "loss": 0.0543, "step": 1195 }, { "epoch": 1.16, "learning_rate": 7.89320730004274e-06, "loss": 0.0531, "step": 1196 }, { "epoch": 1.16, "learning_rate": 7.87783913817492e-06, "loss": 0.0547, "step": 1197 }, { "epoch": 1.16, "learning_rate": 7.862476223016247e-06, "loss": 0.0491, "step": 1198 }, { "epoch": 1.16, "learning_rate": 7.847118592549099e-06, "loss": 0.0583, "step": 1199 }, { "epoch": 1.17, "learning_rate": 7.831766284742807e-06, "loss": 0.0497, "step": 1200 }, { "epoch": 1.17, "learning_rate": 7.816419337553548e-06, "loss": 0.057, "step": 1201 }, { "epoch": 1.17, "learning_rate": 7.801077788924225e-06, "loss": 0.054, "step": 1202 }, { "epoch": 1.17, "learning_rate": 7.78574167678441e-06, "loss": 0.0616, "step": 1203 }, { "epoch": 1.17, "learning_rate": 7.770411039050229e-06, "loss": 0.0588, "step": 1204 }, { "epoch": 1.17, "learning_rate": 7.755085913624274e-06, "loss": 0.0491, "step": 1205 }, { "epoch": 1.17, "learning_rate": 7.73976633839551e-06, "loss": 0.0551, "step": 1206 }, { "epoch": 1.17, "learning_rate": 7.724452351239177e-06, "loss": 0.059, "step": 1207 }, { "epoch": 1.17, "learning_rate": 7.709143990016703e-06, "loss": 0.0549, "step": 1208 }, { "epoch": 1.17, "learning_rate": 7.6938412925756e-06, "loss": 0.0618, "step": 1209 }, { "epoch": 1.17, "learning_rate": 7.678544296749384e-06, "loss": 0.0547, "step": 1210 }, { "epoch": 1.18, "learning_rate": 7.663253040357477e-06, "loss": 0.0604, "step": 1211 }, { "epoch": 1.18, "learning_rate": 7.647967561205101e-06, "loss": 0.054, "step": 1212 }, { "epoch": 1.18, "learning_rate": 7.632687897083206e-06, "loss": 0.0557, "step": 1213 }, { "epoch": 1.18, "learning_rate": 7.617414085768352e-06, "loss": 0.0569, "step": 1214 }, { "epoch": 1.18, "learning_rate": 7.602146165022639e-06, "loss": 0.0548, "step": 1215 }, { "epoch": 1.18, "learning_rate": 7.586884172593609e-06, "loss": 0.0501, "step": 1216 }, { "epoch": 1.18, "learning_rate": 7.57162814621413e-06, "loss": 0.0491, "step": 1217 }, { "epoch": 1.18, "learning_rate": 7.556378123602333e-06, "loss": 0.0467, "step": 1218 }, { "epoch": 1.18, "learning_rate": 7.541134142461499e-06, "loss": 0.0497, "step": 1219 }, { "epoch": 1.18, "learning_rate": 7.525896240479977e-06, "loss": 0.0591, "step": 1220 }, { "epoch": 1.19, "learning_rate": 7.510664455331086e-06, "loss": 0.0534, "step": 1221 }, { "epoch": 1.19, "learning_rate": 7.495438824673016e-06, "loss": 0.0495, "step": 1222 }, { "epoch": 1.19, "learning_rate": 7.480219386148751e-06, "loss": 0.0545, "step": 1223 }, { "epoch": 1.19, "learning_rate": 7.4650061773859535e-06, "loss": 0.0539, "step": 1224 }, { "epoch": 1.19, "learning_rate": 7.449799235996895e-06, "loss": 0.0643, "step": 1225 }, { "epoch": 1.19, "learning_rate": 7.434598599578351e-06, "loss": 0.0538, "step": 1226 }, { "epoch": 1.19, "learning_rate": 7.419404305711499e-06, "loss": 0.051, "step": 1227 }, { "epoch": 1.19, "learning_rate": 7.4042163919618475e-06, "loss": 0.0488, "step": 1228 }, { "epoch": 1.19, "learning_rate": 7.389034895879118e-06, "loss": 0.0495, "step": 1229 }, { "epoch": 1.19, "learning_rate": 7.373859854997181e-06, "loss": 0.0538, "step": 1230 }, { "epoch": 1.2, "learning_rate": 7.35869130683394e-06, "loss": 0.0603, "step": 1231 }, { "epoch": 1.2, "learning_rate": 7.343529288891239e-06, "loss": 0.0591, "step": 1232 }, { "epoch": 1.2, "learning_rate": 7.32837383865479e-06, "loss": 0.0664, "step": 1233 }, { "epoch": 1.2, "learning_rate": 7.313224993594057e-06, "loss": 0.0506, "step": 1234 }, { "epoch": 1.2, "learning_rate": 7.29808279116218e-06, "loss": 0.0531, "step": 1235 }, { "epoch": 1.2, "learning_rate": 7.2829472687958765e-06, "loss": 0.0531, "step": 1236 }, { "epoch": 1.2, "learning_rate": 7.267818463915343e-06, "loss": 0.0613, "step": 1237 }, { "epoch": 1.2, "learning_rate": 7.252696413924174e-06, "loss": 0.0556, "step": 1238 }, { "epoch": 1.2, "learning_rate": 7.237581156209257e-06, "loss": 0.053, "step": 1239 }, { "epoch": 1.2, "learning_rate": 7.222472728140695e-06, "loss": 0.052, "step": 1240 }, { "epoch": 1.2, "learning_rate": 7.2073711670717e-06, "loss": 0.0559, "step": 1241 }, { "epoch": 1.21, "learning_rate": 7.192276510338508e-06, "loss": 0.0563, "step": 1242 }, { "epoch": 1.21, "learning_rate": 7.177188795260287e-06, "loss": 0.0488, "step": 1243 }, { "epoch": 1.21, "learning_rate": 7.162108059139032e-06, "loss": 0.0549, "step": 1244 }, { "epoch": 1.21, "learning_rate": 7.147034339259502e-06, "loss": 0.0546, "step": 1245 }, { "epoch": 1.21, "learning_rate": 7.131967672889101e-06, "loss": 0.0517, "step": 1246 }, { "epoch": 1.21, "learning_rate": 7.1169080972777815e-06, "loss": 0.0567, "step": 1247 }, { "epoch": 1.21, "learning_rate": 7.101855649657991e-06, "loss": 0.051, "step": 1248 }, { "epoch": 1.21, "learning_rate": 7.08681036724453e-06, "loss": 0.0513, "step": 1249 }, { "epoch": 1.21, "learning_rate": 7.071772287234497e-06, "loss": 0.0494, "step": 1250 }, { "epoch": 1.21, "learning_rate": 7.056741446807185e-06, "loss": 0.0628, "step": 1251 }, { "epoch": 1.22, "learning_rate": 7.041717883123977e-06, "loss": 0.0463, "step": 1252 }, { "epoch": 1.22, "learning_rate": 7.026701633328276e-06, "loss": 0.0506, "step": 1253 }, { "epoch": 1.22, "learning_rate": 7.011692734545403e-06, "loss": 0.0595, "step": 1254 }, { "epoch": 1.22, "learning_rate": 6.996691223882496e-06, "loss": 0.0528, "step": 1255 }, { "epoch": 1.22, "learning_rate": 6.9816971384284335e-06, "loss": 0.0525, "step": 1256 }, { "epoch": 1.22, "learning_rate": 6.966710515253734e-06, "loss": 0.0501, "step": 1257 }, { "epoch": 1.22, "learning_rate": 6.951731391410468e-06, "loss": 0.0587, "step": 1258 }, { "epoch": 1.22, "learning_rate": 6.936759803932167e-06, "loss": 0.046, "step": 1259 }, { "epoch": 1.22, "learning_rate": 6.921795789833723e-06, "loss": 0.05, "step": 1260 }, { "epoch": 1.22, "learning_rate": 6.9068393861113146e-06, "loss": 0.055, "step": 1261 }, { "epoch": 1.23, "learning_rate": 6.891890629742289e-06, "loss": 0.0463, "step": 1262 }, { "epoch": 1.23, "learning_rate": 6.876949557685104e-06, "loss": 0.0541, "step": 1263 }, { "epoch": 1.23, "learning_rate": 6.862016206879216e-06, "loss": 0.053, "step": 1264 }, { "epoch": 1.23, "learning_rate": 6.8470906142449764e-06, "loss": 0.0526, "step": 1265 }, { "epoch": 1.23, "learning_rate": 6.832172816683575e-06, "loss": 0.0511, "step": 1266 }, { "epoch": 1.23, "learning_rate": 6.817262851076916e-06, "loss": 0.0551, "step": 1267 }, { "epoch": 1.23, "learning_rate": 6.802360754287548e-06, "loss": 0.0477, "step": 1268 }, { "epoch": 1.23, "learning_rate": 6.787466563158564e-06, "loss": 0.0593, "step": 1269 }, { "epoch": 1.23, "learning_rate": 6.772580314513508e-06, "loss": 0.0694, "step": 1270 }, { "epoch": 1.23, "learning_rate": 6.757702045156292e-06, "loss": 0.0518, "step": 1271 }, { "epoch": 1.23, "learning_rate": 6.742831791871096e-06, "loss": 0.0591, "step": 1272 }, { "epoch": 1.24, "learning_rate": 6.727969591422283e-06, "loss": 0.0535, "step": 1273 }, { "epoch": 1.24, "learning_rate": 6.713115480554313e-06, "loss": 0.0554, "step": 1274 }, { "epoch": 1.24, "learning_rate": 6.6982694959916315e-06, "loss": 0.0547, "step": 1275 }, { "epoch": 1.24, "learning_rate": 6.683431674438613e-06, "loss": 0.0525, "step": 1276 }, { "epoch": 1.24, "learning_rate": 6.668602052579425e-06, "loss": 0.0565, "step": 1277 }, { "epoch": 1.24, "learning_rate": 6.653780667077985e-06, "loss": 0.0589, "step": 1278 }, { "epoch": 1.24, "learning_rate": 6.638967554577842e-06, "loss": 0.059, "step": 1279 }, { "epoch": 1.24, "learning_rate": 6.624162751702077e-06, "loss": 0.0541, "step": 1280 }, { "epoch": 1.24, "learning_rate": 6.60936629505325e-06, "loss": 0.0549, "step": 1281 }, { "epoch": 1.24, "learning_rate": 6.594578221213266e-06, "loss": 0.0556, "step": 1282 }, { "epoch": 1.25, "learning_rate": 6.579798566743314e-06, "loss": 0.0522, "step": 1283 }, { "epoch": 1.25, "learning_rate": 6.56502736818377e-06, "loss": 0.0532, "step": 1284 }, { "epoch": 1.25, "learning_rate": 6.5502646620540955e-06, "loss": 0.0542, "step": 1285 }, { "epoch": 1.25, "learning_rate": 6.535510484852767e-06, "loss": 0.0523, "step": 1286 }, { "epoch": 1.25, "learning_rate": 6.5207648730571614e-06, "loss": 0.0544, "step": 1287 }, { "epoch": 1.25, "learning_rate": 6.5060278631234915e-06, "loss": 0.0504, "step": 1288 }, { "epoch": 1.25, "learning_rate": 6.491299491486696e-06, "loss": 0.0518, "step": 1289 }, { "epoch": 1.25, "learning_rate": 6.476579794560357e-06, "loss": 0.056, "step": 1290 }, { "epoch": 1.25, "learning_rate": 6.461868808736614e-06, "loss": 0.0543, "step": 1291 }, { "epoch": 1.25, "learning_rate": 6.4471665703860644e-06, "loss": 0.0518, "step": 1292 }, { "epoch": 1.26, "learning_rate": 6.4324731158576825e-06, "loss": 0.0575, "step": 1293 }, { "epoch": 1.26, "learning_rate": 6.4177884814787284e-06, "loss": 0.0546, "step": 1294 }, { "epoch": 1.26, "learning_rate": 6.403112703554643e-06, "loss": 0.0545, "step": 1295 }, { "epoch": 1.26, "learning_rate": 6.388445818368991e-06, "loss": 0.0563, "step": 1296 }, { "epoch": 1.26, "learning_rate": 6.373787862183333e-06, "loss": 0.0526, "step": 1297 }, { "epoch": 1.26, "learning_rate": 6.359138871237161e-06, "loss": 0.0551, "step": 1298 }, { "epoch": 1.26, "learning_rate": 6.344498881747811e-06, "loss": 0.0527, "step": 1299 }, { "epoch": 1.26, "learning_rate": 6.329867929910347e-06, "loss": 0.061, "step": 1300 }, { "epoch": 1.26, "learning_rate": 6.315246051897504e-06, "loss": 0.0549, "step": 1301 }, { "epoch": 1.26, "learning_rate": 6.30063328385957e-06, "loss": 0.0572, "step": 1302 }, { "epoch": 1.27, "learning_rate": 6.286029661924319e-06, "loss": 0.0539, "step": 1303 }, { "epoch": 1.27, "learning_rate": 6.2714352221969155e-06, "loss": 0.0556, "step": 1304 }, { "epoch": 1.27, "learning_rate": 6.256850000759811e-06, "loss": 0.0572, "step": 1305 }, { "epoch": 1.27, "learning_rate": 6.242274033672679e-06, "loss": 0.0534, "step": 1306 }, { "epoch": 1.27, "learning_rate": 6.227707356972301e-06, "loss": 0.0579, "step": 1307 }, { "epoch": 1.27, "learning_rate": 6.2131500066724995e-06, "loss": 0.0556, "step": 1308 }, { "epoch": 1.27, "learning_rate": 6.198602018764035e-06, "loss": 0.0541, "step": 1309 }, { "epoch": 1.27, "learning_rate": 6.184063429214515e-06, "loss": 0.0634, "step": 1310 }, { "epoch": 1.27, "learning_rate": 6.1695342739683265e-06, "loss": 0.0581, "step": 1311 }, { "epoch": 1.27, "learning_rate": 6.1550145889465105e-06, "loss": 0.0463, "step": 1312 }, { "epoch": 1.27, "learning_rate": 6.140504410046712e-06, "loss": 0.0564, "step": 1313 }, { "epoch": 1.28, "learning_rate": 6.126003773143072e-06, "loss": 0.0532, "step": 1314 }, { "epoch": 1.28, "learning_rate": 6.111512714086126e-06, "loss": 0.058, "step": 1315 }, { "epoch": 1.28, "learning_rate": 6.0970312687027466e-06, "loss": 0.0526, "step": 1316 }, { "epoch": 1.28, "learning_rate": 6.082559472796028e-06, "loss": 0.0553, "step": 1317 }, { "epoch": 1.28, "learning_rate": 6.068097362145211e-06, "loss": 0.0522, "step": 1318 }, { "epoch": 1.28, "learning_rate": 6.053644972505593e-06, "loss": 0.0522, "step": 1319 }, { "epoch": 1.28, "learning_rate": 6.039202339608432e-06, "loss": 0.0485, "step": 1320 }, { "epoch": 1.28, "learning_rate": 6.024769499160872e-06, "loss": 0.0517, "step": 1321 }, { "epoch": 1.28, "learning_rate": 6.010346486845837e-06, "loss": 0.0531, "step": 1322 }, { "epoch": 1.28, "learning_rate": 5.995933338321962e-06, "loss": 0.0481, "step": 1323 }, { "epoch": 1.29, "learning_rate": 5.981530089223489e-06, "loss": 0.0497, "step": 1324 }, { "epoch": 1.29, "learning_rate": 5.967136775160188e-06, "loss": 0.0563, "step": 1325 }, { "epoch": 1.29, "learning_rate": 5.9527534317172685e-06, "loss": 0.0524, "step": 1326 }, { "epoch": 1.29, "learning_rate": 5.938380094455278e-06, "loss": 0.0534, "step": 1327 }, { "epoch": 1.29, "learning_rate": 5.9240167989100375e-06, "loss": 0.0457, "step": 1328 }, { "epoch": 1.29, "learning_rate": 5.9096635805925435e-06, "loss": 0.0591, "step": 1329 }, { "epoch": 1.29, "learning_rate": 5.895320474988864e-06, "loss": 0.0508, "step": 1330 }, { "epoch": 1.29, "learning_rate": 5.880987517560075e-06, "loss": 0.0535, "step": 1331 }, { "epoch": 1.29, "learning_rate": 5.866664743742162e-06, "loss": 0.0535, "step": 1332 }, { "epoch": 1.29, "learning_rate": 5.8523521889459315e-06, "loss": 0.0473, "step": 1333 }, { "epoch": 1.3, "learning_rate": 5.8380498885569246e-06, "loss": 0.0623, "step": 1334 }, { "epoch": 1.3, "learning_rate": 5.823757877935332e-06, "loss": 0.0567, "step": 1335 }, { "epoch": 1.3, "learning_rate": 5.809476192415905e-06, "loss": 0.0458, "step": 1336 }, { "epoch": 1.3, "learning_rate": 5.79520486730786e-06, "loss": 0.0583, "step": 1337 }, { "epoch": 1.3, "learning_rate": 5.780943937894805e-06, "loss": 0.0534, "step": 1338 }, { "epoch": 1.3, "learning_rate": 5.7666934394346565e-06, "loss": 0.0493, "step": 1339 }, { "epoch": 1.3, "learning_rate": 5.752453407159521e-06, "loss": 0.0584, "step": 1340 }, { "epoch": 1.3, "learning_rate": 5.73822387627565e-06, "loss": 0.0486, "step": 1341 }, { "epoch": 1.3, "learning_rate": 5.724004881963311e-06, "loss": 0.0538, "step": 1342 }, { "epoch": 1.3, "learning_rate": 5.709796459376737e-06, "loss": 0.0499, "step": 1343 }, { "epoch": 1.3, "learning_rate": 5.69559864364402e-06, "loss": 0.0544, "step": 1344 }, { "epoch": 1.31, "learning_rate": 5.681411469867027e-06, "loss": 0.0526, "step": 1345 }, { "epoch": 1.31, "learning_rate": 5.667234973121317e-06, "loss": 0.0544, "step": 1346 }, { "epoch": 1.31, "learning_rate": 5.6530691884560465e-06, "loss": 0.0571, "step": 1347 }, { "epoch": 1.31, "learning_rate": 5.63891415089389e-06, "loss": 0.0534, "step": 1348 }, { "epoch": 1.31, "learning_rate": 5.6247698954309616e-06, "loss": 0.0595, "step": 1349 }, { "epoch": 1.31, "learning_rate": 5.610636457036693e-06, "loss": 0.0552, "step": 1350 }, { "epoch": 1.31, "learning_rate": 5.596513870653808e-06, "loss": 0.0555, "step": 1351 }, { "epoch": 1.31, "learning_rate": 5.582402171198169e-06, "loss": 0.0532, "step": 1352 }, { "epoch": 1.31, "learning_rate": 5.568301393558739e-06, "loss": 0.052, "step": 1353 }, { "epoch": 1.31, "learning_rate": 5.554211572597477e-06, "loss": 0.0628, "step": 1354 }, { "epoch": 1.32, "learning_rate": 5.5401327431492425e-06, "loss": 0.0576, "step": 1355 }, { "epoch": 1.32, "learning_rate": 5.526064940021733e-06, "loss": 0.0593, "step": 1356 }, { "epoch": 1.32, "learning_rate": 5.512008197995379e-06, "loss": 0.0518, "step": 1357 }, { "epoch": 1.32, "learning_rate": 5.497962551823266e-06, "loss": 0.0539, "step": 1358 }, { "epoch": 1.32, "learning_rate": 5.483928036231052e-06, "loss": 0.0487, "step": 1359 }, { "epoch": 1.32, "learning_rate": 5.469904685916861e-06, "loss": 0.0554, "step": 1360 }, { "epoch": 1.32, "learning_rate": 5.455892535551226e-06, "loss": 0.0523, "step": 1361 }, { "epoch": 1.32, "learning_rate": 5.441891619776988e-06, "loss": 0.0541, "step": 1362 }, { "epoch": 1.32, "learning_rate": 5.427901973209208e-06, "loss": 0.056, "step": 1363 }, { "epoch": 1.32, "learning_rate": 5.413923630435094e-06, "loss": 0.0499, "step": 1364 }, { "epoch": 1.33, "learning_rate": 5.399956626013891e-06, "loss": 0.0566, "step": 1365 }, { "epoch": 1.33, "learning_rate": 5.386000994476832e-06, "loss": 0.0482, "step": 1366 }, { "epoch": 1.33, "learning_rate": 5.3720567703270135e-06, "loss": 0.0514, "step": 1367 }, { "epoch": 1.33, "learning_rate": 5.358123988039338e-06, "loss": 0.0506, "step": 1368 }, { "epoch": 1.33, "learning_rate": 5.3442026820604305e-06, "loss": 0.0472, "step": 1369 }, { "epoch": 1.33, "learning_rate": 5.33029288680852e-06, "loss": 0.0571, "step": 1370 }, { "epoch": 1.33, "learning_rate": 5.316394636673399e-06, "loss": 0.0572, "step": 1371 }, { "epoch": 1.33, "learning_rate": 5.302507966016295e-06, "loss": 0.0633, "step": 1372 }, { "epoch": 1.33, "learning_rate": 5.288632909169823e-06, "loss": 0.05, "step": 1373 }, { "epoch": 1.33, "learning_rate": 5.274769500437882e-06, "loss": 0.0509, "step": 1374 }, { "epoch": 1.33, "learning_rate": 5.260917774095568e-06, "loss": 0.0607, "step": 1375 }, { "epoch": 1.34, "learning_rate": 5.247077764389099e-06, "loss": 0.0586, "step": 1376 }, { "epoch": 1.34, "learning_rate": 5.233249505535727e-06, "loss": 0.0549, "step": 1377 }, { "epoch": 1.34, "learning_rate": 5.219433031723641e-06, "loss": 0.0601, "step": 1378 }, { "epoch": 1.34, "learning_rate": 5.205628377111902e-06, "loss": 0.0705, "step": 1379 }, { "epoch": 1.34, "learning_rate": 5.1918355758303515e-06, "loss": 0.0471, "step": 1380 }, { "epoch": 1.34, "learning_rate": 5.178054661979521e-06, "loss": 0.0504, "step": 1381 }, { "epoch": 1.34, "learning_rate": 5.164285669630558e-06, "loss": 0.0465, "step": 1382 }, { "epoch": 1.34, "learning_rate": 5.1505286328251225e-06, "loss": 0.0605, "step": 1383 }, { "epoch": 1.34, "learning_rate": 5.136783585575336e-06, "loss": 0.0573, "step": 1384 }, { "epoch": 1.34, "learning_rate": 5.1230505618636575e-06, "loss": 0.0509, "step": 1385 }, { "epoch": 1.35, "learning_rate": 5.109329595642829e-06, "loss": 0.0516, "step": 1386 }, { "epoch": 1.35, "learning_rate": 5.095620720835795e-06, "loss": 0.0515, "step": 1387 }, { "epoch": 1.35, "learning_rate": 5.081923971335583e-06, "loss": 0.0634, "step": 1388 }, { "epoch": 1.35, "learning_rate": 5.06823938100526e-06, "loss": 0.051, "step": 1389 }, { "epoch": 1.35, "learning_rate": 5.0545669836778144e-06, "loss": 0.0509, "step": 1390 }, { "epoch": 1.35, "learning_rate": 5.040906813156107e-06, "loss": 0.0476, "step": 1391 }, { "epoch": 1.35, "learning_rate": 5.027258903212759e-06, "loss": 0.0493, "step": 1392 }, { "epoch": 1.35, "learning_rate": 5.013623287590084e-06, "loss": 0.0526, "step": 1393 }, { "epoch": 1.35, "learning_rate": 5.000000000000003e-06, "loss": 0.0528, "step": 1394 }, { "epoch": 1.35, "learning_rate": 4.986389074123943e-06, "loss": 0.0588, "step": 1395 }, { "epoch": 1.36, "learning_rate": 4.972790543612784e-06, "loss": 0.0525, "step": 1396 }, { "epoch": 1.36, "learning_rate": 4.959204442086754e-06, "loss": 0.0508, "step": 1397 }, { "epoch": 1.36, "learning_rate": 4.945630803135354e-06, "loss": 0.0568, "step": 1398 }, { "epoch": 1.36, "learning_rate": 4.932069660317277e-06, "loss": 0.0468, "step": 1399 }, { "epoch": 1.36, "learning_rate": 4.918521047160309e-06, "loss": 0.0589, "step": 1400 }, { "epoch": 1.36, "learning_rate": 4.9049849971612684e-06, "loss": 0.0606, "step": 1401 }, { "epoch": 1.36, "learning_rate": 4.891461543785917e-06, "loss": 0.0496, "step": 1402 }, { "epoch": 1.36, "learning_rate": 4.8779507204688595e-06, "loss": 0.0503, "step": 1403 }, { "epoch": 1.36, "learning_rate": 4.864452560613485e-06, "loss": 0.0536, "step": 1404 }, { "epoch": 1.36, "learning_rate": 4.850967097591873e-06, "loss": 0.0521, "step": 1405 }, { "epoch": 1.37, "learning_rate": 4.8374943647447115e-06, "loss": 0.0527, "step": 1406 }, { "epoch": 1.37, "learning_rate": 4.8240343953812184e-06, "loss": 0.0489, "step": 1407 }, { "epoch": 1.37, "learning_rate": 4.810587222779043e-06, "loss": 0.0525, "step": 1408 }, { "epoch": 1.37, "learning_rate": 4.797152880184212e-06, "loss": 0.0556, "step": 1409 }, { "epoch": 1.37, "learning_rate": 4.783731400811022e-06, "loss": 0.0564, "step": 1410 }, { "epoch": 1.37, "learning_rate": 4.7703228178419744e-06, "loss": 0.0594, "step": 1411 }, { "epoch": 1.37, "learning_rate": 4.756927164427685e-06, "loss": 0.0534, "step": 1412 }, { "epoch": 1.37, "learning_rate": 4.743544473686794e-06, "loss": 0.0523, "step": 1413 }, { "epoch": 1.37, "learning_rate": 4.730174778705909e-06, "loss": 0.0516, "step": 1414 }, { "epoch": 1.37, "learning_rate": 4.716818112539485e-06, "loss": 0.0613, "step": 1415 }, { "epoch": 1.37, "learning_rate": 4.703474508209793e-06, "loss": 0.0462, "step": 1416 }, { "epoch": 1.38, "learning_rate": 4.690143998706797e-06, "loss": 0.0546, "step": 1417 }, { "epoch": 1.38, "learning_rate": 4.676826616988081e-06, "loss": 0.0592, "step": 1418 }, { "epoch": 1.38, "learning_rate": 4.66352239597878e-06, "loss": 0.0551, "step": 1419 }, { "epoch": 1.38, "learning_rate": 4.650231368571486e-06, "loss": 0.0503, "step": 1420 }, { "epoch": 1.38, "learning_rate": 4.636953567626176e-06, "loss": 0.0559, "step": 1421 }, { "epoch": 1.38, "learning_rate": 4.623689025970129e-06, "loss": 0.0598, "step": 1422 }, { "epoch": 1.38, "learning_rate": 4.610437776397836e-06, "loss": 0.0597, "step": 1423 }, { "epoch": 1.38, "learning_rate": 4.5971998516709325e-06, "loss": 0.0529, "step": 1424 }, { "epoch": 1.38, "learning_rate": 4.583975284518101e-06, "loss": 0.0506, "step": 1425 }, { "epoch": 1.38, "learning_rate": 4.570764107635008e-06, "loss": 0.0582, "step": 1426 }, { "epoch": 1.39, "learning_rate": 4.55756635368421e-06, "loss": 0.0431, "step": 1427 }, { "epoch": 1.39, "learning_rate": 4.54438205529508e-06, "loss": 0.0543, "step": 1428 }, { "epoch": 1.39, "learning_rate": 4.531211245063728e-06, "loss": 0.0539, "step": 1429 }, { "epoch": 1.39, "learning_rate": 4.518053955552903e-06, "loss": 0.0541, "step": 1430 }, { "epoch": 1.39, "learning_rate": 4.504910219291941e-06, "loss": 0.0544, "step": 1431 }, { "epoch": 1.39, "learning_rate": 4.491780068776663e-06, "loss": 0.0574, "step": 1432 }, { "epoch": 1.39, "learning_rate": 4.478663536469297e-06, "loss": 0.054, "step": 1433 }, { "epoch": 1.39, "learning_rate": 4.4655606547984165e-06, "loss": 0.0498, "step": 1434 }, { "epoch": 1.39, "learning_rate": 4.452471456158829e-06, "loss": 0.0523, "step": 1435 }, { "epoch": 1.39, "learning_rate": 4.439395972911525e-06, "loss": 0.044, "step": 1436 }, { "epoch": 1.4, "learning_rate": 4.426334237383581e-06, "loss": 0.0515, "step": 1437 }, { "epoch": 1.4, "learning_rate": 4.413286281868081e-06, "loss": 0.0476, "step": 1438 }, { "epoch": 1.4, "learning_rate": 4.400252138624047e-06, "loss": 0.0552, "step": 1439 }, { "epoch": 1.4, "learning_rate": 4.387231839876349e-06, "loss": 0.0496, "step": 1440 }, { "epoch": 1.4, "learning_rate": 4.374225417815628e-06, "loss": 0.0542, "step": 1441 }, { "epoch": 1.4, "learning_rate": 4.3612329045982236e-06, "loss": 0.0488, "step": 1442 }, { "epoch": 1.4, "learning_rate": 4.348254332346073e-06, "loss": 0.0442, "step": 1443 }, { "epoch": 1.4, "learning_rate": 4.335289733146665e-06, "loss": 0.0495, "step": 1444 }, { "epoch": 1.4, "learning_rate": 4.322339139052922e-06, "loss": 0.0453, "step": 1445 }, { "epoch": 1.4, "learning_rate": 4.309402582083161e-06, "loss": 0.0549, "step": 1446 }, { "epoch": 1.4, "learning_rate": 4.296480094220987e-06, "loss": 0.0508, "step": 1447 }, { "epoch": 1.41, "learning_rate": 4.283571707415214e-06, "loss": 0.0512, "step": 1448 }, { "epoch": 1.41, "learning_rate": 4.270677453579805e-06, "loss": 0.0572, "step": 1449 }, { "epoch": 1.41, "learning_rate": 4.257797364593768e-06, "loss": 0.0564, "step": 1450 }, { "epoch": 1.41, "learning_rate": 4.244931472301098e-06, "loss": 0.056, "step": 1451 }, { "epoch": 1.41, "learning_rate": 4.232079808510704e-06, "loss": 0.0504, "step": 1452 }, { "epoch": 1.41, "learning_rate": 4.219242404996293e-06, "loss": 0.0546, "step": 1453 }, { "epoch": 1.41, "learning_rate": 4.206419293496333e-06, "loss": 0.0591, "step": 1454 }, { "epoch": 1.41, "learning_rate": 4.1936105057139466e-06, "loss": 0.0575, "step": 1455 }, { "epoch": 1.41, "learning_rate": 4.1808160733168505e-06, "loss": 0.0532, "step": 1456 }, { "epoch": 1.41, "learning_rate": 4.168036027937267e-06, "loss": 0.0484, "step": 1457 }, { "epoch": 1.42, "learning_rate": 4.155270401171851e-06, "loss": 0.049, "step": 1458 }, { "epoch": 1.42, "learning_rate": 4.142519224581608e-06, "loss": 0.0565, "step": 1459 }, { "epoch": 1.42, "learning_rate": 4.1297825296918145e-06, "loss": 0.0486, "step": 1460 }, { "epoch": 1.42, "learning_rate": 4.1170603479919495e-06, "loss": 0.0519, "step": 1461 }, { "epoch": 1.42, "learning_rate": 4.1043527109356095e-06, "loss": 0.0521, "step": 1462 }, { "epoch": 1.42, "learning_rate": 4.091659649940419e-06, "loss": 0.0471, "step": 1463 }, { "epoch": 1.42, "learning_rate": 4.078981196387991e-06, "loss": 0.0521, "step": 1464 }, { "epoch": 1.42, "learning_rate": 4.066317381623798e-06, "loss": 0.0469, "step": 1465 }, { "epoch": 1.42, "learning_rate": 4.053668236957135e-06, "loss": 0.0583, "step": 1466 }, { "epoch": 1.42, "learning_rate": 4.041033793661025e-06, "loss": 0.0542, "step": 1467 }, { "epoch": 1.43, "learning_rate": 4.028414082972141e-06, "loss": 0.0532, "step": 1468 }, { "epoch": 1.43, "learning_rate": 4.015809136090732e-06, "loss": 0.0593, "step": 1469 }, { "epoch": 1.43, "learning_rate": 4.003218984180552e-06, "loss": 0.0573, "step": 1470 }, { "epoch": 1.43, "learning_rate": 3.99064365836877e-06, "loss": 0.0536, "step": 1471 }, { "epoch": 1.43, "learning_rate": 3.978083189745907e-06, "loss": 0.0541, "step": 1472 }, { "epoch": 1.43, "learning_rate": 3.9655376093657394e-06, "loss": 0.0567, "step": 1473 }, { "epoch": 1.43, "learning_rate": 3.953006948245247e-06, "loss": 0.0473, "step": 1474 }, { "epoch": 1.43, "learning_rate": 3.940491237364519e-06, "loss": 0.0598, "step": 1475 }, { "epoch": 1.43, "learning_rate": 3.927990507666683e-06, "loss": 0.0546, "step": 1476 }, { "epoch": 1.43, "learning_rate": 3.915504790057831e-06, "loss": 0.0524, "step": 1477 }, { "epoch": 1.43, "learning_rate": 3.9030341154069315e-06, "loss": 0.0511, "step": 1478 }, { "epoch": 1.44, "learning_rate": 3.890578514545774e-06, "loss": 0.0555, "step": 1479 }, { "epoch": 1.44, "learning_rate": 3.878138018268867e-06, "loss": 0.0581, "step": 1480 }, { "epoch": 1.44, "learning_rate": 3.86571265733338e-06, "loss": 0.0497, "step": 1481 }, { "epoch": 1.44, "learning_rate": 3.853302462459077e-06, "loss": 0.0619, "step": 1482 }, { "epoch": 1.44, "learning_rate": 3.840907464328204e-06, "loss": 0.0521, "step": 1483 }, { "epoch": 1.44, "learning_rate": 3.828527693585451e-06, "loss": 0.0447, "step": 1484 }, { "epoch": 1.44, "learning_rate": 3.816163180837848e-06, "loss": 0.052, "step": 1485 }, { "epoch": 1.44, "learning_rate": 3.803813956654715e-06, "loss": 0.0484, "step": 1486 }, { "epoch": 1.44, "learning_rate": 3.7914800515675644e-06, "loss": 0.0503, "step": 1487 }, { "epoch": 1.44, "learning_rate": 3.7791614960700397e-06, "loss": 0.0467, "step": 1488 }, { "epoch": 1.45, "learning_rate": 3.7668583206178355e-06, "loss": 0.0529, "step": 1489 }, { "epoch": 1.45, "learning_rate": 3.754570555628613e-06, "loss": 0.0539, "step": 1490 }, { "epoch": 1.45, "learning_rate": 3.7422982314819425e-06, "loss": 0.0563, "step": 1491 }, { "epoch": 1.45, "learning_rate": 3.7300413785192158e-06, "loss": 0.0576, "step": 1492 }, { "epoch": 1.45, "learning_rate": 3.7178000270435765e-06, "loss": 0.0494, "step": 1493 }, { "epoch": 1.45, "learning_rate": 3.705574207319844e-06, "loss": 0.0581, "step": 1494 }, { "epoch": 1.45, "learning_rate": 3.693363949574429e-06, "loss": 0.0576, "step": 1495 }, { "epoch": 1.45, "learning_rate": 3.6811692839952795e-06, "loss": 0.0553, "step": 1496 }, { "epoch": 1.45, "learning_rate": 3.66899024073179e-06, "loss": 0.0569, "step": 1497 }, { "epoch": 1.45, "learning_rate": 3.6568268498947257e-06, "loss": 0.0496, "step": 1498 }, { "epoch": 1.46, "learning_rate": 3.6446791415561577e-06, "loss": 0.0581, "step": 1499 }, { "epoch": 1.46, "learning_rate": 3.6325471457493956e-06, "loss": 0.0499, "step": 1500 }, { "epoch": 1.46, "learning_rate": 3.6204308924688814e-06, "loss": 0.0562, "step": 1501 }, { "epoch": 1.46, "learning_rate": 3.6083304116701535e-06, "loss": 0.0549, "step": 1502 }, { "epoch": 1.46, "learning_rate": 3.596245733269742e-06, "loss": 0.0524, "step": 1503 }, { "epoch": 1.46, "learning_rate": 3.584176887145119e-06, "loss": 0.0529, "step": 1504 }, { "epoch": 1.46, "learning_rate": 3.5721239031346067e-06, "loss": 0.0443, "step": 1505 }, { "epoch": 1.46, "learning_rate": 3.5600868110373166e-06, "loss": 0.0575, "step": 1506 }, { "epoch": 1.46, "learning_rate": 3.5480656406130675e-06, "loss": 0.0499, "step": 1507 }, { "epoch": 1.46, "learning_rate": 3.536060421582309e-06, "loss": 0.0556, "step": 1508 }, { "epoch": 1.47, "learning_rate": 3.52407118362606e-06, "loss": 0.0511, "step": 1509 }, { "epoch": 1.47, "learning_rate": 3.512097956385827e-06, "loss": 0.0607, "step": 1510 }, { "epoch": 1.47, "learning_rate": 3.5001407694635326e-06, "loss": 0.0534, "step": 1511 }, { "epoch": 1.47, "learning_rate": 3.4881996524214445e-06, "loss": 0.0495, "step": 1512 }, { "epoch": 1.47, "learning_rate": 3.4762746347820907e-06, "loss": 0.0511, "step": 1513 }, { "epoch": 1.47, "learning_rate": 3.4643657460282078e-06, "loss": 0.0495, "step": 1514 }, { "epoch": 1.47, "learning_rate": 3.452473015602652e-06, "loss": 0.0514, "step": 1515 }, { "epoch": 1.47, "learning_rate": 3.440596472908325e-06, "loss": 0.06, "step": 1516 }, { "epoch": 1.47, "learning_rate": 3.428736147308115e-06, "loss": 0.0488, "step": 1517 }, { "epoch": 1.47, "learning_rate": 3.416892068124812e-06, "loss": 0.0499, "step": 1518 }, { "epoch": 1.47, "learning_rate": 3.4050642646410402e-06, "loss": 0.0572, "step": 1519 }, { "epoch": 1.48, "learning_rate": 3.3932527660991877e-06, "loss": 0.0537, "step": 1520 }, { "epoch": 1.48, "learning_rate": 3.3814576017013212e-06, "loss": 0.0494, "step": 1521 }, { "epoch": 1.48, "learning_rate": 3.3696788006091343e-06, "loss": 0.0519, "step": 1522 }, { "epoch": 1.48, "learning_rate": 3.3579163919438597e-06, "loss": 0.0508, "step": 1523 }, { "epoch": 1.48, "learning_rate": 3.3461704047862054e-06, "loss": 0.0527, "step": 1524 }, { "epoch": 1.48, "learning_rate": 3.3344408681762818e-06, "loss": 0.0587, "step": 1525 }, { "epoch": 1.48, "learning_rate": 3.3227278111135166e-06, "loss": 0.0557, "step": 1526 }, { "epoch": 1.48, "learning_rate": 3.311031262556609e-06, "loss": 0.0565, "step": 1527 }, { "epoch": 1.48, "learning_rate": 3.299351251423426e-06, "loss": 0.046, "step": 1528 }, { "epoch": 1.48, "learning_rate": 3.2876878065909714e-06, "loss": 0.05, "step": 1529 }, { "epoch": 1.49, "learning_rate": 3.2760409568952765e-06, "loss": 0.0572, "step": 1530 }, { "epoch": 1.49, "learning_rate": 3.264410731131341e-06, "loss": 0.0485, "step": 1531 }, { "epoch": 1.49, "learning_rate": 3.252797158053077e-06, "loss": 0.0597, "step": 1532 }, { "epoch": 1.49, "learning_rate": 3.241200266373211e-06, "loss": 0.0529, "step": 1533 }, { "epoch": 1.49, "learning_rate": 3.2296200847632376e-06, "loss": 0.0532, "step": 1534 }, { "epoch": 1.49, "learning_rate": 3.2180566418533365e-06, "loss": 0.0515, "step": 1535 }, { "epoch": 1.49, "learning_rate": 3.206509966232302e-06, "loss": 0.0471, "step": 1536 }, { "epoch": 1.49, "learning_rate": 3.1949800864474776e-06, "loss": 0.0509, "step": 1537 }, { "epoch": 1.49, "learning_rate": 3.1834670310046735e-06, "loss": 0.0565, "step": 1538 }, { "epoch": 1.49, "learning_rate": 3.1719708283681107e-06, "loss": 0.061, "step": 1539 }, { "epoch": 1.5, "learning_rate": 3.160491506960344e-06, "loss": 0.0541, "step": 1540 }, { "epoch": 1.5, "learning_rate": 3.1490290951621903e-06, "loss": 0.0513, "step": 1541 }, { "epoch": 1.5, "learning_rate": 3.1375836213126653e-06, "loss": 0.0486, "step": 1542 }, { "epoch": 1.5, "learning_rate": 3.1261551137088988e-06, "loss": 0.062, "step": 1543 }, { "epoch": 1.5, "learning_rate": 3.114743600606078e-06, "loss": 0.0503, "step": 1544 }, { "epoch": 1.5, "learning_rate": 3.103349110217384e-06, "loss": 0.0564, "step": 1545 }, { "epoch": 1.5, "learning_rate": 3.0919716707138893e-06, "loss": 0.0495, "step": 1546 }, { "epoch": 1.5, "learning_rate": 3.0806113102245395e-06, "loss": 0.0477, "step": 1547 }, { "epoch": 1.5, "learning_rate": 3.06926805683603e-06, "loss": 0.0559, "step": 1548 }, { "epoch": 1.5, "learning_rate": 3.0579419385927743e-06, "loss": 0.0496, "step": 1549 }, { "epoch": 1.5, "learning_rate": 3.0466329834968234e-06, "loss": 0.0496, "step": 1550 }, { "epoch": 1.51, "learning_rate": 3.0353412195077836e-06, "loss": 0.0593, "step": 1551 }, { "epoch": 1.51, "learning_rate": 3.0240666745427715e-06, "loss": 0.0491, "step": 1552 }, { "epoch": 1.51, "learning_rate": 3.0128093764763255e-06, "loss": 0.0473, "step": 1553 }, { "epoch": 1.51, "learning_rate": 3.001569353140347e-06, "loss": 0.05, "step": 1554 }, { "epoch": 1.51, "learning_rate": 2.990346632324027e-06, "loss": 0.052, "step": 1555 }, { "epoch": 1.51, "learning_rate": 2.979141241773775e-06, "loss": 0.0495, "step": 1556 }, { "epoch": 1.51, "learning_rate": 2.967953209193164e-06, "loss": 0.0479, "step": 1557 }, { "epoch": 1.51, "learning_rate": 2.9567825622428357e-06, "loss": 0.0467, "step": 1558 }, { "epoch": 1.51, "learning_rate": 2.945629328540468e-06, "loss": 0.0485, "step": 1559 }, { "epoch": 1.51, "learning_rate": 2.934493535660677e-06, "loss": 0.0473, "step": 1560 }, { "epoch": 1.52, "learning_rate": 2.9233752111349567e-06, "loss": 0.0564, "step": 1561 }, { "epoch": 1.52, "learning_rate": 2.91227438245162e-06, "loss": 0.0593, "step": 1562 }, { "epoch": 1.52, "learning_rate": 2.901191077055716e-06, "loss": 0.0513, "step": 1563 }, { "epoch": 1.52, "learning_rate": 2.8901253223489757e-06, "loss": 0.043, "step": 1564 }, { "epoch": 1.52, "learning_rate": 2.879077145689746e-06, "loss": 0.0548, "step": 1565 }, { "epoch": 1.52, "learning_rate": 2.868046574392899e-06, "loss": 0.0478, "step": 1566 }, { "epoch": 1.52, "learning_rate": 2.8570336357297922e-06, "loss": 0.0411, "step": 1567 }, { "epoch": 1.52, "learning_rate": 2.8460383569281823e-06, "loss": 0.0476, "step": 1568 }, { "epoch": 1.52, "learning_rate": 2.8350607651721684e-06, "loss": 0.0487, "step": 1569 }, { "epoch": 1.52, "learning_rate": 2.8241008876021216e-06, "loss": 0.0506, "step": 1570 }, { "epoch": 1.53, "learning_rate": 2.8131587513146164e-06, "loss": 0.0491, "step": 1571 }, { "epoch": 1.53, "learning_rate": 2.8022343833623666e-06, "loss": 0.0563, "step": 1572 }, { "epoch": 1.53, "learning_rate": 2.791327810754151e-06, "loss": 0.0493, "step": 1573 }, { "epoch": 1.53, "learning_rate": 2.780439060454756e-06, "loss": 0.04, "step": 1574 }, { "epoch": 1.53, "learning_rate": 2.76956815938491e-06, "loss": 0.055, "step": 1575 }, { "epoch": 1.53, "learning_rate": 2.7587151344211973e-06, "loss": 0.0538, "step": 1576 }, { "epoch": 1.53, "learning_rate": 2.747880012396028e-06, "loss": 0.0537, "step": 1577 }, { "epoch": 1.53, "learning_rate": 2.73706282009753e-06, "loss": 0.0553, "step": 1578 }, { "epoch": 1.53, "learning_rate": 2.726263584269513e-06, "loss": 0.0513, "step": 1579 }, { "epoch": 1.53, "learning_rate": 2.715482331611393e-06, "loss": 0.0457, "step": 1580 }, { "epoch": 1.53, "learning_rate": 2.704719088778117e-06, "loss": 0.0604, "step": 1581 }, { "epoch": 1.54, "learning_rate": 2.6939738823801144e-06, "loss": 0.0526, "step": 1582 }, { "epoch": 1.54, "learning_rate": 2.683246738983217e-06, "loss": 0.0512, "step": 1583 }, { "epoch": 1.54, "learning_rate": 2.672537685108603e-06, "loss": 0.0534, "step": 1584 }, { "epoch": 1.54, "learning_rate": 2.6618467472327248e-06, "loss": 0.0463, "step": 1585 }, { "epoch": 1.54, "learning_rate": 2.6511739517872428e-06, "loss": 0.0495, "step": 1586 }, { "epoch": 1.54, "learning_rate": 2.6405193251589666e-06, "loss": 0.0534, "step": 1587 }, { "epoch": 1.54, "learning_rate": 2.629882893689787e-06, "loss": 0.0505, "step": 1588 }, { "epoch": 1.54, "learning_rate": 2.619264683676609e-06, "loss": 0.0513, "step": 1589 }, { "epoch": 1.54, "learning_rate": 2.60866472137129e-06, "loss": 0.0544, "step": 1590 }, { "epoch": 1.54, "learning_rate": 2.598083032980565e-06, "loss": 0.0544, "step": 1591 }, { "epoch": 1.55, "learning_rate": 2.587519644666001e-06, "loss": 0.0522, "step": 1592 }, { "epoch": 1.55, "learning_rate": 2.5769745825439095e-06, "loss": 0.054, "step": 1593 }, { "epoch": 1.55, "learning_rate": 2.566447872685298e-06, "loss": 0.0565, "step": 1594 }, { "epoch": 1.55, "learning_rate": 2.5559395411158116e-06, "loss": 0.0512, "step": 1595 }, { "epoch": 1.55, "learning_rate": 2.5454496138156395e-06, "loss": 0.0457, "step": 1596 }, { "epoch": 1.55, "learning_rate": 2.534978116719482e-06, "loss": 0.0587, "step": 1597 }, { "epoch": 1.55, "learning_rate": 2.5245250757164664e-06, "loss": 0.0544, "step": 1598 }, { "epoch": 1.55, "learning_rate": 2.514090516650095e-06, "loss": 0.0535, "step": 1599 }, { "epoch": 1.55, "learning_rate": 2.5036744653181755e-06, "loss": 0.0519, "step": 1600 }, { "epoch": 1.55, "learning_rate": 2.493276947472756e-06, "loss": 0.0508, "step": 1601 }, { "epoch": 1.56, "learning_rate": 2.48289798882007e-06, "loss": 0.0551, "step": 1602 }, { "epoch": 1.56, "learning_rate": 2.472537615020454e-06, "loss": 0.0521, "step": 1603 }, { "epoch": 1.56, "learning_rate": 2.462195851688306e-06, "loss": 0.048, "step": 1604 }, { "epoch": 1.56, "learning_rate": 2.4518727243920114e-06, "loss": 0.0536, "step": 1605 }, { "epoch": 1.56, "learning_rate": 2.4415682586538793e-06, "loss": 0.0584, "step": 1606 }, { "epoch": 1.56, "learning_rate": 2.4312824799500847e-06, "loss": 0.0488, "step": 1607 }, { "epoch": 1.56, "learning_rate": 2.4210154137105913e-06, "loss": 0.0501, "step": 1608 }, { "epoch": 1.56, "learning_rate": 2.4107670853191114e-06, "loss": 0.045, "step": 1609 }, { "epoch": 1.56, "learning_rate": 2.4005375201130275e-06, "loss": 0.0555, "step": 1610 }, { "epoch": 1.56, "learning_rate": 2.390326743383327e-06, "loss": 0.0506, "step": 1611 }, { "epoch": 1.57, "learning_rate": 2.380134780374551e-06, "loss": 0.0522, "step": 1612 }, { "epoch": 1.57, "learning_rate": 2.369961656284727e-06, "loss": 0.0533, "step": 1613 }, { "epoch": 1.57, "learning_rate": 2.359807396265307e-06, "loss": 0.0478, "step": 1614 }, { "epoch": 1.57, "learning_rate": 2.3496720254211026e-06, "loss": 0.0552, "step": 1615 }, { "epoch": 1.57, "learning_rate": 2.339555568810221e-06, "loss": 0.0476, "step": 1616 }, { "epoch": 1.57, "learning_rate": 2.3294580514440137e-06, "loss": 0.0484, "step": 1617 }, { "epoch": 1.57, "learning_rate": 2.3193794982870046e-06, "loss": 0.0521, "step": 1618 }, { "epoch": 1.57, "learning_rate": 2.3093199342568316e-06, "loss": 0.056, "step": 1619 }, { "epoch": 1.57, "learning_rate": 2.29927938422419e-06, "loss": 0.0565, "step": 1620 }, { "epoch": 1.57, "learning_rate": 2.2892578730127536e-06, "loss": 0.0462, "step": 1621 }, { "epoch": 1.57, "learning_rate": 2.2792554253991417e-06, "loss": 0.0425, "step": 1622 }, { "epoch": 1.58, "learning_rate": 2.2692720661128232e-06, "loss": 0.0536, "step": 1623 }, { "epoch": 1.58, "learning_rate": 2.259307819836093e-06, "loss": 0.0526, "step": 1624 }, { "epoch": 1.58, "learning_rate": 2.249362711203985e-06, "loss": 0.0505, "step": 1625 }, { "epoch": 1.58, "learning_rate": 2.2394367648042104e-06, "loss": 0.0616, "step": 1626 }, { "epoch": 1.58, "learning_rate": 2.2295300051771153e-06, "loss": 0.0504, "step": 1627 }, { "epoch": 1.58, "learning_rate": 2.2196424568156073e-06, "loss": 0.0538, "step": 1628 }, { "epoch": 1.58, "learning_rate": 2.20977414416509e-06, "loss": 0.0565, "step": 1629 }, { "epoch": 1.58, "learning_rate": 2.1999250916234183e-06, "loss": 0.0466, "step": 1630 }, { "epoch": 1.58, "learning_rate": 2.190095323540825e-06, "loss": 0.0514, "step": 1631 }, { "epoch": 1.58, "learning_rate": 2.1802848642198693e-06, "loss": 0.0464, "step": 1632 }, { "epoch": 1.59, "learning_rate": 2.17049373791537e-06, "loss": 0.0508, "step": 1633 }, { "epoch": 1.59, "learning_rate": 2.160721968834344e-06, "loss": 0.048, "step": 1634 }, { "epoch": 1.59, "learning_rate": 2.1509695811359555e-06, "loss": 0.0567, "step": 1635 }, { "epoch": 1.59, "learning_rate": 2.141236598931451e-06, "loss": 0.0496, "step": 1636 }, { "epoch": 1.59, "learning_rate": 2.1315230462840985e-06, "loss": 0.049, "step": 1637 }, { "epoch": 1.59, "learning_rate": 2.1218289472091335e-06, "loss": 0.0529, "step": 1638 }, { "epoch": 1.59, "learning_rate": 2.1121543256736865e-06, "loss": 0.0492, "step": 1639 }, { "epoch": 1.59, "learning_rate": 2.102499205596743e-06, "loss": 0.0441, "step": 1640 }, { "epoch": 1.59, "learning_rate": 2.092863610849062e-06, "loss": 0.0562, "step": 1641 }, { "epoch": 1.59, "learning_rate": 2.0832475652531446e-06, "loss": 0.0504, "step": 1642 }, { "epoch": 1.6, "learning_rate": 2.073651092583153e-06, "loss": 0.0463, "step": 1643 }, { "epoch": 1.6, "learning_rate": 2.064074216564852e-06, "loss": 0.0588, "step": 1644 }, { "epoch": 1.6, "learning_rate": 2.0545169608755653e-06, "loss": 0.0493, "step": 1645 }, { "epoch": 1.6, "learning_rate": 2.0449793491441026e-06, "loss": 0.0492, "step": 1646 }, { "epoch": 1.6, "learning_rate": 2.0354614049507106e-06, "loss": 0.0475, "step": 1647 }, { "epoch": 1.6, "learning_rate": 2.025963151827011e-06, "loss": 0.0509, "step": 1648 }, { "epoch": 1.6, "learning_rate": 2.01648461325594e-06, "loss": 0.0512, "step": 1649 }, { "epoch": 1.6, "learning_rate": 2.0070258126717e-06, "loss": 0.0505, "step": 1650 }, { "epoch": 1.6, "learning_rate": 1.997586773459681e-06, "loss": 0.0419, "step": 1651 }, { "epoch": 1.6, "learning_rate": 1.9881675189564255e-06, "loss": 0.0492, "step": 1652 }, { "epoch": 1.6, "learning_rate": 1.9787680724495617e-06, "loss": 0.0591, "step": 1653 }, { "epoch": 1.61, "learning_rate": 1.969388457177743e-06, "loss": 0.0494, "step": 1654 }, { "epoch": 1.61, "learning_rate": 1.960028696330596e-06, "loss": 0.0478, "step": 1655 }, { "epoch": 1.61, "learning_rate": 1.9506888130486525e-06, "loss": 0.0537, "step": 1656 }, { "epoch": 1.61, "learning_rate": 1.9413688304233093e-06, "loss": 0.0523, "step": 1657 }, { "epoch": 1.61, "learning_rate": 1.9320687714967603e-06, "loss": 0.0511, "step": 1658 }, { "epoch": 1.61, "learning_rate": 1.9227886592619306e-06, "loss": 0.054, "step": 1659 }, { "epoch": 1.61, "learning_rate": 1.913528516662452e-06, "loss": 0.0559, "step": 1660 }, { "epoch": 1.61, "learning_rate": 1.9042883665925604e-06, "loss": 0.048, "step": 1661 }, { "epoch": 1.61, "learning_rate": 1.8950682318970792e-06, "loss": 0.0557, "step": 1662 }, { "epoch": 1.61, "learning_rate": 1.8858681353713436e-06, "loss": 0.049, "step": 1663 }, { "epoch": 1.62, "learning_rate": 1.8766880997611424e-06, "loss": 0.0555, "step": 1664 }, { "epoch": 1.62, "learning_rate": 1.8675281477626706e-06, "loss": 0.0551, "step": 1665 }, { "epoch": 1.62, "learning_rate": 1.8583883020224724e-06, "loss": 0.0481, "step": 1666 }, { "epoch": 1.62, "learning_rate": 1.849268585137377e-06, "loss": 0.0445, "step": 1667 }, { "epoch": 1.62, "learning_rate": 1.8401690196544553e-06, "loss": 0.0536, "step": 1668 }, { "epoch": 1.62, "learning_rate": 1.8310896280709466e-06, "loss": 0.052, "step": 1669 }, { "epoch": 1.62, "learning_rate": 1.8220304328342253e-06, "loss": 0.0484, "step": 1670 }, { "epoch": 1.62, "learning_rate": 1.8129914563417183e-06, "loss": 0.0499, "step": 1671 }, { "epoch": 1.62, "learning_rate": 1.8039727209408841e-06, "loss": 0.0508, "step": 1672 }, { "epoch": 1.62, "learning_rate": 1.7949742489291256e-06, "loss": 0.0466, "step": 1673 }, { "epoch": 1.63, "learning_rate": 1.7859960625537476e-06, "loss": 0.0523, "step": 1674 }, { "epoch": 1.63, "learning_rate": 1.7770381840119077e-06, "loss": 0.0532, "step": 1675 }, { "epoch": 1.63, "learning_rate": 1.7681006354505492e-06, "loss": 0.0555, "step": 1676 }, { "epoch": 1.63, "learning_rate": 1.7591834389663541e-06, "loss": 0.0489, "step": 1677 }, { "epoch": 1.63, "learning_rate": 1.7502866166056986e-06, "loss": 0.0477, "step": 1678 }, { "epoch": 1.63, "learning_rate": 1.7414101903645686e-06, "loss": 0.0523, "step": 1679 }, { "epoch": 1.63, "learning_rate": 1.7325541821885383e-06, "loss": 0.0576, "step": 1680 }, { "epoch": 1.63, "learning_rate": 1.7237186139726892e-06, "loss": 0.0487, "step": 1681 }, { "epoch": 1.63, "learning_rate": 1.7149035075615795e-06, "loss": 0.0555, "step": 1682 }, { "epoch": 1.63, "learning_rate": 1.7061088847491724e-06, "loss": 0.0516, "step": 1683 }, { "epoch": 1.63, "learning_rate": 1.697334767278792e-06, "loss": 0.0489, "step": 1684 }, { "epoch": 1.64, "learning_rate": 1.688581176843066e-06, "loss": 0.0455, "step": 1685 }, { "epoch": 1.64, "learning_rate": 1.6798481350838647e-06, "loss": 0.0531, "step": 1686 }, { "epoch": 1.64, "learning_rate": 1.671135663592266e-06, "loss": 0.0516, "step": 1687 }, { "epoch": 1.64, "learning_rate": 1.662443783908486e-06, "loss": 0.0532, "step": 1688 }, { "epoch": 1.64, "learning_rate": 1.6537725175218255e-06, "loss": 0.0485, "step": 1689 }, { "epoch": 1.64, "learning_rate": 1.6451218858706374e-06, "loss": 0.0487, "step": 1690 }, { "epoch": 1.64, "learning_rate": 1.6364919103422394e-06, "loss": 0.0548, "step": 1691 }, { "epoch": 1.64, "learning_rate": 1.627882612272893e-06, "loss": 0.0501, "step": 1692 }, { "epoch": 1.64, "learning_rate": 1.6192940129477375e-06, "loss": 0.0499, "step": 1693 }, { "epoch": 1.64, "learning_rate": 1.6107261336007284e-06, "loss": 0.0504, "step": 1694 }, { "epoch": 1.65, "learning_rate": 1.6021789954146038e-06, "loss": 0.0502, "step": 1695 }, { "epoch": 1.65, "learning_rate": 1.593652619520819e-06, "loss": 0.0607, "step": 1696 }, { "epoch": 1.65, "learning_rate": 1.5851470269994984e-06, "loss": 0.0478, "step": 1697 }, { "epoch": 1.65, "learning_rate": 1.576662238879384e-06, "loss": 0.0489, "step": 1698 }, { "epoch": 1.65, "learning_rate": 1.5681982761377767e-06, "loss": 0.0479, "step": 1699 }, { "epoch": 1.65, "learning_rate": 1.5597551597004968e-06, "loss": 0.0498, "step": 1700 }, { "epoch": 1.65, "learning_rate": 1.5513329104418217e-06, "loss": 0.0589, "step": 1701 }, { "epoch": 1.65, "learning_rate": 1.5429315491844387e-06, "loss": 0.0454, "step": 1702 }, { "epoch": 1.65, "learning_rate": 1.5345510966993959e-06, "loss": 0.049, "step": 1703 }, { "epoch": 1.65, "learning_rate": 1.5261915737060385e-06, "loss": 0.0541, "step": 1704 }, { "epoch": 1.66, "learning_rate": 1.5178530008719794e-06, "loss": 0.0528, "step": 1705 }, { "epoch": 1.66, "learning_rate": 1.5095353988130235e-06, "loss": 0.0525, "step": 1706 }, { "epoch": 1.66, "learning_rate": 1.5012387880931345e-06, "loss": 0.0469, "step": 1707 }, { "epoch": 1.66, "learning_rate": 1.4929631892243857e-06, "loss": 0.0488, "step": 1708 }, { "epoch": 1.66, "learning_rate": 1.4847086226668871e-06, "loss": 0.0511, "step": 1709 }, { "epoch": 1.66, "learning_rate": 1.4764751088287621e-06, "loss": 0.0489, "step": 1710 }, { "epoch": 1.66, "learning_rate": 1.468262668066074e-06, "loss": 0.0464, "step": 1711 }, { "epoch": 1.66, "learning_rate": 1.4600713206827932e-06, "loss": 0.059, "step": 1712 }, { "epoch": 1.66, "learning_rate": 1.4519010869307403e-06, "loss": 0.05, "step": 1713 }, { "epoch": 1.66, "learning_rate": 1.443751987009533e-06, "loss": 0.0441, "step": 1714 }, { "epoch": 1.67, "learning_rate": 1.4356240410665435e-06, "loss": 0.0615, "step": 1715 }, { "epoch": 1.67, "learning_rate": 1.427517269196833e-06, "loss": 0.0436, "step": 1716 }, { "epoch": 1.67, "learning_rate": 1.419431691443125e-06, "loss": 0.053, "step": 1717 }, { "epoch": 1.67, "learning_rate": 1.4113673277957395e-06, "loss": 0.0516, "step": 1718 }, { "epoch": 1.67, "learning_rate": 1.4033241981925472e-06, "loss": 0.0519, "step": 1719 }, { "epoch": 1.67, "learning_rate": 1.3953023225189243e-06, "loss": 0.0451, "step": 1720 }, { "epoch": 1.67, "learning_rate": 1.3873017206076933e-06, "loss": 0.0587, "step": 1721 }, { "epoch": 1.67, "learning_rate": 1.379322412239086e-06, "loss": 0.0534, "step": 1722 }, { "epoch": 1.67, "learning_rate": 1.3713644171406904e-06, "loss": 0.0507, "step": 1723 }, { "epoch": 1.67, "learning_rate": 1.3634277549873952e-06, "loss": 0.0497, "step": 1724 }, { "epoch": 1.67, "learning_rate": 1.3555124454013512e-06, "loss": 0.056, "step": 1725 }, { "epoch": 1.68, "learning_rate": 1.3476185079519178e-06, "loss": 0.06, "step": 1726 }, { "epoch": 1.68, "learning_rate": 1.339745962155613e-06, "loss": 0.0487, "step": 1727 }, { "epoch": 1.68, "learning_rate": 1.3318948274760735e-06, "loss": 0.0575, "step": 1728 }, { "epoch": 1.68, "learning_rate": 1.3240651233239875e-06, "loss": 0.0622, "step": 1729 }, { "epoch": 1.68, "learning_rate": 1.3162568690570743e-06, "loss": 0.0506, "step": 1730 }, { "epoch": 1.68, "learning_rate": 1.3084700839800112e-06, "loss": 0.0481, "step": 1731 }, { "epoch": 1.68, "learning_rate": 1.3007047873444034e-06, "loss": 0.0466, "step": 1732 }, { "epoch": 1.68, "learning_rate": 1.2929609983487257e-06, "loss": 0.0511, "step": 1733 }, { "epoch": 1.68, "learning_rate": 1.2852387361382768e-06, "loss": 0.0531, "step": 1734 }, { "epoch": 1.68, "learning_rate": 1.2775380198051391e-06, "loss": 0.0524, "step": 1735 }, { "epoch": 1.69, "learning_rate": 1.2698588683881185e-06, "loss": 0.0444, "step": 1736 }, { "epoch": 1.69, "learning_rate": 1.2622013008727142e-06, "loss": 0.0477, "step": 1737 }, { "epoch": 1.69, "learning_rate": 1.2545653361910591e-06, "loss": 0.058, "step": 1738 }, { "epoch": 1.69, "learning_rate": 1.2469509932218704e-06, "loss": 0.0496, "step": 1739 }, { "epoch": 1.69, "learning_rate": 1.23935829079042e-06, "loss": 0.05, "step": 1740 }, { "epoch": 1.69, "learning_rate": 1.2317872476684655e-06, "loss": 0.0519, "step": 1741 }, { "epoch": 1.69, "learning_rate": 1.224237882574224e-06, "loss": 0.0546, "step": 1742 }, { "epoch": 1.69, "learning_rate": 1.2167102141723142e-06, "loss": 0.0525, "step": 1743 }, { "epoch": 1.69, "learning_rate": 1.2092042610737108e-06, "loss": 0.055, "step": 1744 }, { "epoch": 1.69, "learning_rate": 1.2017200418357077e-06, "loss": 0.0589, "step": 1745 }, { "epoch": 1.7, "learning_rate": 1.1942575749618546e-06, "loss": 0.0491, "step": 1746 }, { "epoch": 1.7, "learning_rate": 1.1868168789019307e-06, "loss": 0.047, "step": 1747 }, { "epoch": 1.7, "learning_rate": 1.1793979720518866e-06, "loss": 0.0494, "step": 1748 }, { "epoch": 1.7, "learning_rate": 1.1720008727538036e-06, "loss": 0.0536, "step": 1749 }, { "epoch": 1.7, "learning_rate": 1.1646255992958466e-06, "loss": 0.0479, "step": 1750 }, { "epoch": 1.7, "learning_rate": 1.1572721699122236e-06, "loss": 0.0551, "step": 1751 }, { "epoch": 1.7, "learning_rate": 1.1499406027831261e-06, "loss": 0.0516, "step": 1752 }, { "epoch": 1.7, "learning_rate": 1.1426309160347082e-06, "loss": 0.0496, "step": 1753 }, { "epoch": 1.7, "learning_rate": 1.1353431277390125e-06, "loss": 0.0549, "step": 1754 }, { "epoch": 1.7, "learning_rate": 1.1280772559139585e-06, "loss": 0.0464, "step": 1755 }, { "epoch": 1.7, "learning_rate": 1.1208333185232712e-06, "loss": 0.0535, "step": 1756 }, { "epoch": 1.71, "learning_rate": 1.113611333476442e-06, "loss": 0.0481, "step": 1757 }, { "epoch": 1.71, "learning_rate": 1.1064113186286984e-06, "loss": 0.0528, "step": 1758 }, { "epoch": 1.71, "learning_rate": 1.0992332917809423e-06, "loss": 0.0544, "step": 1759 }, { "epoch": 1.71, "learning_rate": 1.0920772706797166e-06, "loss": 0.0462, "step": 1760 }, { "epoch": 1.71, "learning_rate": 1.0849432730171617e-06, "loss": 0.0509, "step": 1761 }, { "epoch": 1.71, "learning_rate": 1.0778313164309616e-06, "loss": 0.0509, "step": 1762 }, { "epoch": 1.71, "learning_rate": 1.0707414185043163e-06, "loss": 0.0519, "step": 1763 }, { "epoch": 1.71, "learning_rate": 1.0636735967658785e-06, "loss": 0.0468, "step": 1764 }, { "epoch": 1.71, "learning_rate": 1.056627868689728e-06, "loss": 0.0524, "step": 1765 }, { "epoch": 1.71, "learning_rate": 1.0496042516953208e-06, "loss": 0.046, "step": 1766 }, { "epoch": 1.72, "learning_rate": 1.0426027631474455e-06, "loss": 0.043, "step": 1767 }, { "epoch": 1.72, "learning_rate": 1.0356234203561832e-06, "loss": 0.0506, "step": 1768 }, { "epoch": 1.72, "learning_rate": 1.0286662405768599e-06, "loss": 0.0522, "step": 1769 }, { "epoch": 1.72, "learning_rate": 1.0217312410100088e-06, "loss": 0.0501, "step": 1770 }, { "epoch": 1.72, "learning_rate": 1.0148184388013294e-06, "loss": 0.0523, "step": 1771 }, { "epoch": 1.72, "learning_rate": 1.0079278510416313e-06, "loss": 0.0579, "step": 1772 }, { "epoch": 1.72, "learning_rate": 1.0010594947668184e-06, "loss": 0.0547, "step": 1773 }, { "epoch": 1.72, "learning_rate": 9.942133869578164e-07, "loss": 0.0532, "step": 1774 }, { "epoch": 1.72, "learning_rate": 9.873895445405523e-07, "loss": 0.0496, "step": 1775 }, { "epoch": 1.72, "learning_rate": 9.805879843859057e-07, "loss": 0.0459, "step": 1776 }, { "epoch": 1.73, "learning_rate": 9.738087233096605e-07, "loss": 0.0539, "step": 1777 }, { "epoch": 1.73, "learning_rate": 9.670517780724774e-07, "loss": 0.0558, "step": 1778 }, { "epoch": 1.73, "learning_rate": 9.603171653798393e-07, "loss": 0.0543, "step": 1779 }, { "epoch": 1.73, "learning_rate": 9.536049018820193e-07, "loss": 0.0495, "step": 1780 }, { "epoch": 1.73, "learning_rate": 9.469150041740338e-07, "loss": 0.0561, "step": 1781 }, { "epoch": 1.73, "learning_rate": 9.402474887956014e-07, "loss": 0.0548, "step": 1782 }, { "epoch": 1.73, "learning_rate": 9.336023722311083e-07, "loss": 0.0491, "step": 1783 }, { "epoch": 1.73, "learning_rate": 9.269796709095558e-07, "loss": 0.055, "step": 1784 }, { "epoch": 1.73, "learning_rate": 9.203794012045375e-07, "loss": 0.0471, "step": 1785 }, { "epoch": 1.73, "learning_rate": 9.138015794341814e-07, "loss": 0.0511, "step": 1786 }, { "epoch": 1.73, "learning_rate": 9.072462218611155e-07, "loss": 0.0509, "step": 1787 }, { "epoch": 1.74, "learning_rate": 9.007133446924343e-07, "loss": 0.054, "step": 1788 }, { "epoch": 1.74, "learning_rate": 8.942029640796457e-07, "loss": 0.0486, "step": 1789 }, { "epoch": 1.74, "learning_rate": 8.87715096118642e-07, "loss": 0.0491, "step": 1790 }, { "epoch": 1.74, "learning_rate": 8.812497568496636e-07, "loss": 0.0566, "step": 1791 }, { "epoch": 1.74, "learning_rate": 8.748069622572386e-07, "loss": 0.0477, "step": 1792 }, { "epoch": 1.74, "learning_rate": 8.683867282701685e-07, "loss": 0.048, "step": 1793 }, { "epoch": 1.74, "learning_rate": 8.619890707614687e-07, "loss": 0.0525, "step": 1794 }, { "epoch": 1.74, "learning_rate": 8.556140055483431e-07, "loss": 0.0556, "step": 1795 }, { "epoch": 1.74, "learning_rate": 8.492615483921395e-07, "loss": 0.0508, "step": 1796 }, { "epoch": 1.74, "learning_rate": 8.429317149983085e-07, "loss": 0.0487, "step": 1797 }, { "epoch": 1.75, "learning_rate": 8.366245210163704e-07, "loss": 0.0518, "step": 1798 }, { "epoch": 1.75, "learning_rate": 8.303399820398672e-07, "loss": 0.0555, "step": 1799 }, { "epoch": 1.75, "learning_rate": 8.240781136063348e-07, "loss": 0.0498, "step": 1800 }, { "epoch": 1.75, "learning_rate": 8.178389311972612e-07, "loss": 0.0538, "step": 1801 }, { "epoch": 1.75, "learning_rate": 8.116224502380387e-07, "loss": 0.0483, "step": 1802 }, { "epoch": 1.75, "learning_rate": 8.054286860979466e-07, "loss": 0.0509, "step": 1803 }, { "epoch": 1.75, "learning_rate": 7.992576540900876e-07, "loss": 0.05, "step": 1804 }, { "epoch": 1.75, "learning_rate": 7.931093694713687e-07, "loss": 0.0539, "step": 1805 }, { "epoch": 1.75, "learning_rate": 7.869838474424607e-07, "loss": 0.058, "step": 1806 }, { "epoch": 1.75, "learning_rate": 7.808811031477504e-07, "loss": 0.0534, "step": 1807 }, { "epoch": 1.76, "learning_rate": 7.74801151675314e-07, "loss": 0.0469, "step": 1808 }, { "epoch": 1.76, "learning_rate": 7.687440080568765e-07, "loss": 0.0477, "step": 1809 }, { "epoch": 1.76, "learning_rate": 7.627096872677741e-07, "loss": 0.0503, "step": 1810 }, { "epoch": 1.76, "learning_rate": 7.566982042269177e-07, "loss": 0.0499, "step": 1811 }, { "epoch": 1.76, "learning_rate": 7.507095737967495e-07, "loss": 0.0548, "step": 1812 }, { "epoch": 1.76, "learning_rate": 7.447438107832206e-07, "loss": 0.0479, "step": 1813 }, { "epoch": 1.76, "learning_rate": 7.388009299357412e-07, "loss": 0.0508, "step": 1814 }, { "epoch": 1.76, "learning_rate": 7.328809459471497e-07, "loss": 0.0532, "step": 1815 }, { "epoch": 1.76, "learning_rate": 7.269838734536771e-07, "loss": 0.058, "step": 1816 }, { "epoch": 1.76, "learning_rate": 7.211097270349065e-07, "loss": 0.0538, "step": 1817 }, { "epoch": 1.77, "learning_rate": 7.152585212137442e-07, "loss": 0.054, "step": 1818 }, { "epoch": 1.77, "learning_rate": 7.094302704563738e-07, "loss": 0.0457, "step": 1819 }, { "epoch": 1.77, "learning_rate": 7.03624989172228e-07, "loss": 0.0556, "step": 1820 }, { "epoch": 1.77, "learning_rate": 6.97842691713958e-07, "loss": 0.0531, "step": 1821 }, { "epoch": 1.77, "learning_rate": 6.920833923773796e-07, "loss": 0.0445, "step": 1822 }, { "epoch": 1.77, "learning_rate": 6.863471054014592e-07, "loss": 0.0497, "step": 1823 }, { "epoch": 1.77, "learning_rate": 6.806338449682614e-07, "loss": 0.05, "step": 1824 }, { "epoch": 1.77, "learning_rate": 6.749436252029262e-07, "loss": 0.045, "step": 1825 }, { "epoch": 1.77, "learning_rate": 6.692764601736268e-07, "loss": 0.0539, "step": 1826 }, { "epoch": 1.77, "learning_rate": 6.636323638915388e-07, "loss": 0.0479, "step": 1827 }, { "epoch": 1.77, "learning_rate": 6.580113503108033e-07, "loss": 0.0507, "step": 1828 }, { "epoch": 1.78, "learning_rate": 6.524134333284904e-07, "loss": 0.0566, "step": 1829 }, { "epoch": 1.78, "learning_rate": 6.468386267845716e-07, "loss": 0.0506, "step": 1830 }, { "epoch": 1.78, "learning_rate": 6.412869444618796e-07, "loss": 0.0538, "step": 1831 }, { "epoch": 1.78, "learning_rate": 6.357584000860761e-07, "loss": 0.0518, "step": 1832 }, { "epoch": 1.78, "learning_rate": 6.302530073256197e-07, "loss": 0.0511, "step": 1833 }, { "epoch": 1.78, "learning_rate": 6.247707797917257e-07, "loss": 0.0544, "step": 1834 }, { "epoch": 1.78, "learning_rate": 6.193117310383412e-07, "loss": 0.0471, "step": 1835 }, { "epoch": 1.78, "learning_rate": 6.138758745621087e-07, "loss": 0.0533, "step": 1836 }, { "epoch": 1.78, "learning_rate": 6.084632238023236e-07, "loss": 0.0492, "step": 1837 }, { "epoch": 1.78, "learning_rate": 6.030737921409169e-07, "loss": 0.0488, "step": 1838 }, { "epoch": 1.79, "learning_rate": 5.977075929024112e-07, "loss": 0.0488, "step": 1839 }, { "epoch": 1.79, "learning_rate": 5.923646393538907e-07, "loss": 0.0475, "step": 1840 }, { "epoch": 1.79, "learning_rate": 5.870449447049686e-07, "loss": 0.0582, "step": 1841 }, { "epoch": 1.79, "learning_rate": 5.817485221077523e-07, "loss": 0.0457, "step": 1842 }, { "epoch": 1.79, "learning_rate": 5.764753846568139e-07, "loss": 0.0508, "step": 1843 }, { "epoch": 1.79, "learning_rate": 5.71225545389158e-07, "loss": 0.0492, "step": 1844 }, { "epoch": 1.79, "learning_rate": 5.659990172841867e-07, "loss": 0.0532, "step": 1845 }, { "epoch": 1.79, "learning_rate": 5.607958132636715e-07, "loss": 0.0453, "step": 1846 }, { "epoch": 1.79, "learning_rate": 5.556159461917121e-07, "loss": 0.0495, "step": 1847 }, { "epoch": 1.79, "learning_rate": 5.504594288747189e-07, "loss": 0.0602, "step": 1848 }, { "epoch": 1.8, "learning_rate": 5.453262740613652e-07, "loss": 0.0428, "step": 1849 }, { "epoch": 1.8, "learning_rate": 5.402164944425758e-07, "loss": 0.056, "step": 1850 }, { "epoch": 1.8, "learning_rate": 5.351301026514765e-07, "loss": 0.0485, "step": 1851 }, { "epoch": 1.8, "learning_rate": 5.30067111263367e-07, "loss": 0.0463, "step": 1852 }, { "epoch": 1.8, "learning_rate": 5.250275327957033e-07, "loss": 0.0493, "step": 1853 }, { "epoch": 1.8, "learning_rate": 5.200113797080464e-07, "loss": 0.0515, "step": 1854 }, { "epoch": 1.8, "learning_rate": 5.150186644020483e-07, "loss": 0.047, "step": 1855 }, { "epoch": 1.8, "learning_rate": 5.100493992214129e-07, "loss": 0.0483, "step": 1856 }, { "epoch": 1.8, "learning_rate": 5.051035964518669e-07, "loss": 0.0464, "step": 1857 }, { "epoch": 1.8, "learning_rate": 5.001812683211304e-07, "loss": 0.0472, "step": 1858 }, { "epoch": 1.8, "learning_rate": 4.952824269988855e-07, "loss": 0.0613, "step": 1859 }, { "epoch": 1.81, "learning_rate": 4.904070845967467e-07, "loss": 0.051, "step": 1860 }, { "epoch": 1.81, "learning_rate": 4.855552531682329e-07, "loss": 0.0495, "step": 1861 }, { "epoch": 1.81, "learning_rate": 4.807269447087348e-07, "loss": 0.0608, "step": 1862 }, { "epoch": 1.81, "learning_rate": 4.759221711554862e-07, "loss": 0.05, "step": 1863 }, { "epoch": 1.81, "learning_rate": 4.7114094438753256e-07, "loss": 0.0571, "step": 1864 }, { "epoch": 1.81, "learning_rate": 4.66383276225707e-07, "loss": 0.0513, "step": 1865 }, { "epoch": 1.81, "learning_rate": 4.6164917843259713e-07, "loss": 0.0479, "step": 1866 }, { "epoch": 1.81, "learning_rate": 4.5693866271251165e-07, "loss": 0.054, "step": 1867 }, { "epoch": 1.81, "learning_rate": 4.522517407114646e-07, "loss": 0.0467, "step": 1868 }, { "epoch": 1.81, "learning_rate": 4.475884240171324e-07, "loss": 0.0547, "step": 1869 }, { "epoch": 1.82, "learning_rate": 4.429487241588304e-07, "loss": 0.0559, "step": 1870 }, { "epoch": 1.82, "learning_rate": 4.3833265260749157e-07, "loss": 0.0461, "step": 1871 }, { "epoch": 1.82, "learning_rate": 4.337402207756236e-07, "loss": 0.048, "step": 1872 }, { "epoch": 1.82, "learning_rate": 4.291714400172919e-07, "loss": 0.052, "step": 1873 }, { "epoch": 1.82, "learning_rate": 4.2462632162809103e-07, "loss": 0.0482, "step": 1874 }, { "epoch": 1.82, "learning_rate": 4.2010487684511105e-07, "loss": 0.0583, "step": 1875 }, { "epoch": 1.82, "learning_rate": 4.1560711684691444e-07, "loss": 0.0536, "step": 1876 }, { "epoch": 1.82, "learning_rate": 4.111330527535029e-07, "loss": 0.048, "step": 1877 }, { "epoch": 1.82, "learning_rate": 4.0668269562629814e-07, "loss": 0.0474, "step": 1878 }, { "epoch": 1.82, "learning_rate": 4.0225605646810883e-07, "loss": 0.0493, "step": 1879 }, { "epoch": 1.83, "learning_rate": 3.97853146223105e-07, "loss": 0.0493, "step": 1880 }, { "epoch": 1.83, "learning_rate": 3.9347397577679026e-07, "loss": 0.0482, "step": 1881 }, { "epoch": 1.83, "learning_rate": 3.891185559559729e-07, "loss": 0.0478, "step": 1882 }, { "epoch": 1.83, "learning_rate": 3.84786897528745e-07, "loss": 0.0492, "step": 1883 }, { "epoch": 1.83, "learning_rate": 3.804790112044532e-07, "loss": 0.0456, "step": 1884 }, { "epoch": 1.83, "learning_rate": 3.761949076336646e-07, "loss": 0.0478, "step": 1885 }, { "epoch": 1.83, "learning_rate": 3.719345974081567e-07, "loss": 0.0556, "step": 1886 }, { "epoch": 1.83, "learning_rate": 3.676980910608741e-07, "loss": 0.0537, "step": 1887 }, { "epoch": 1.83, "learning_rate": 3.6348539906591263e-07, "loss": 0.0563, "step": 1888 }, { "epoch": 1.83, "learning_rate": 3.5929653183849444e-07, "loss": 0.0522, "step": 1889 }, { "epoch": 1.83, "learning_rate": 3.551314997349298e-07, "loss": 0.0405, "step": 1890 }, { "epoch": 1.84, "learning_rate": 3.5099031305261045e-07, "loss": 0.0565, "step": 1891 }, { "epoch": 1.84, "learning_rate": 3.4687298202996654e-07, "loss": 0.0457, "step": 1892 }, { "epoch": 1.84, "learning_rate": 3.4277951684645315e-07, "loss": 0.0527, "step": 1893 }, { "epoch": 1.84, "learning_rate": 3.387099276225214e-07, "loss": 0.0521, "step": 1894 }, { "epoch": 1.84, "learning_rate": 3.346642244195863e-07, "loss": 0.048, "step": 1895 }, { "epoch": 1.84, "learning_rate": 3.30642417240018e-07, "loss": 0.0533, "step": 1896 }, { "epoch": 1.84, "learning_rate": 3.266445160270981e-07, "loss": 0.0427, "step": 1897 }, { "epoch": 1.84, "learning_rate": 3.226705306650113e-07, "loss": 0.0477, "step": 1898 }, { "epoch": 1.84, "learning_rate": 3.1872047097881274e-07, "loss": 0.0486, "step": 1899 }, { "epoch": 1.84, "learning_rate": 3.1479434673440167e-07, "loss": 0.0503, "step": 1900 }, { "epoch": 1.85, "learning_rate": 3.108921676385057e-07, "loss": 0.0517, "step": 1901 }, { "epoch": 1.85, "learning_rate": 3.070139433386454e-07, "loss": 0.0469, "step": 1902 }, { "epoch": 1.85, "learning_rate": 3.0315968342312317e-07, "loss": 0.0468, "step": 1903 }, { "epoch": 1.85, "learning_rate": 2.993293974209921e-07, "loss": 0.0483, "step": 1904 }, { "epoch": 1.85, "learning_rate": 2.9552309480203045e-07, "loss": 0.058, "step": 1905 }, { "epoch": 1.85, "learning_rate": 2.91740784976724e-07, "loss": 0.0554, "step": 1906 }, { "epoch": 1.85, "learning_rate": 2.879824772962381e-07, "loss": 0.0463, "step": 1907 }, { "epoch": 1.85, "learning_rate": 2.842481810523978e-07, "loss": 0.0459, "step": 1908 }, { "epoch": 1.85, "learning_rate": 2.8053790547766355e-07, "loss": 0.0513, "step": 1909 }, { "epoch": 1.85, "learning_rate": 2.7685165974510987e-07, "loss": 0.0497, "step": 1910 }, { "epoch": 1.86, "learning_rate": 2.7318945296839985e-07, "loss": 0.054, "step": 1911 }, { "epoch": 1.86, "learning_rate": 2.6955129420176193e-07, "loss": 0.056, "step": 1912 }, { "epoch": 1.86, "learning_rate": 2.659371924399734e-07, "loss": 0.0511, "step": 1913 }, { "epoch": 1.86, "learning_rate": 2.623471566183322e-07, "loss": 0.0462, "step": 1914 }, { "epoch": 1.86, "learning_rate": 2.5878119561263737e-07, "loss": 0.0493, "step": 1915 }, { "epoch": 1.86, "learning_rate": 2.552393182391677e-07, "loss": 0.0483, "step": 1916 }, { "epoch": 1.86, "learning_rate": 2.5172153325465853e-07, "loss": 0.0625, "step": 1917 }, { "epoch": 1.86, "learning_rate": 2.4822784935627844e-07, "loss": 0.0457, "step": 1918 }, { "epoch": 1.86, "learning_rate": 2.4475827518161357e-07, "loss": 0.0541, "step": 1919 }, { "epoch": 1.86, "learning_rate": 2.4131281930864006e-07, "loss": 0.0458, "step": 1920 }, { "epoch": 1.87, "learning_rate": 2.37891490255705e-07, "loss": 0.0479, "step": 1921 }, { "epoch": 1.87, "learning_rate": 2.3449429648150668e-07, "loss": 0.0495, "step": 1922 }, { "epoch": 1.87, "learning_rate": 2.3112124638507427e-07, "loss": 0.0503, "step": 1923 }, { "epoch": 1.87, "learning_rate": 2.277723483057448e-07, "loss": 0.0439, "step": 1924 }, { "epoch": 1.87, "learning_rate": 2.2444761052313857e-07, "loss": 0.0423, "step": 1925 }, { "epoch": 1.87, "learning_rate": 2.2114704125714925e-07, "loss": 0.0447, "step": 1926 }, { "epoch": 1.87, "learning_rate": 2.1787064866791608e-07, "loss": 0.0577, "step": 1927 }, { "epoch": 1.87, "learning_rate": 2.1461844085580385e-07, "loss": 0.0538, "step": 1928 }, { "epoch": 1.87, "learning_rate": 2.1139042586138525e-07, "loss": 0.0472, "step": 1929 }, { "epoch": 1.87, "learning_rate": 2.0818661166542075e-07, "loss": 0.052, "step": 1930 }, { "epoch": 1.87, "learning_rate": 2.0500700618883652e-07, "loss": 0.0447, "step": 1931 }, { "epoch": 1.88, "learning_rate": 2.0185161729270652e-07, "loss": 0.0561, "step": 1932 }, { "epoch": 1.88, "learning_rate": 1.9872045277823383e-07, "loss": 0.05, "step": 1933 }, { "epoch": 1.88, "learning_rate": 1.9561352038673264e-07, "loss": 0.0517, "step": 1934 }, { "epoch": 1.88, "learning_rate": 1.9253082779960408e-07, "loss": 0.0536, "step": 1935 }, { "epoch": 1.88, "learning_rate": 1.8947238263832046e-07, "loss": 0.0553, "step": 1936 }, { "epoch": 1.88, "learning_rate": 1.8643819246440543e-07, "loss": 0.0487, "step": 1937 }, { "epoch": 1.88, "learning_rate": 1.8342826477941944e-07, "loss": 0.0465, "step": 1938 }, { "epoch": 1.88, "learning_rate": 1.804426070249332e-07, "loss": 0.0566, "step": 1939 }, { "epoch": 1.88, "learning_rate": 1.7748122658251877e-07, "loss": 0.0442, "step": 1940 }, { "epoch": 1.88, "learning_rate": 1.7454413077372277e-07, "loss": 0.0467, "step": 1941 }, { "epoch": 1.89, "learning_rate": 1.7163132686005224e-07, "loss": 0.0526, "step": 1942 }, { "epoch": 1.89, "learning_rate": 1.6874282204295765e-07, "loss": 0.0482, "step": 1943 }, { "epoch": 1.89, "learning_rate": 1.6587862346381323e-07, "loss": 0.0513, "step": 1944 }, { "epoch": 1.89, "learning_rate": 1.6303873820389892e-07, "loss": 0.052, "step": 1945 }, { "epoch": 1.89, "learning_rate": 1.6022317328438508e-07, "loss": 0.0479, "step": 1946 }, { "epoch": 1.89, "learning_rate": 1.5743193566631342e-07, "loss": 0.0508, "step": 1947 }, { "epoch": 1.89, "learning_rate": 1.5466503225058048e-07, "loss": 0.0466, "step": 1948 }, { "epoch": 1.89, "learning_rate": 1.519224698779198e-07, "loss": 0.0476, "step": 1949 }, { "epoch": 1.89, "learning_rate": 1.4920425532888528e-07, "loss": 0.0487, "step": 1950 }, { "epoch": 1.89, "learning_rate": 1.465103953238356e-07, "loss": 0.0493, "step": 1951 }, { "epoch": 1.9, "learning_rate": 1.4384089652291544e-07, "loss": 0.0505, "step": 1952 }, { "epoch": 1.9, "learning_rate": 1.411957655260421e-07, "loss": 0.0538, "step": 1953 }, { "epoch": 1.9, "learning_rate": 1.3857500887288544e-07, "loss": 0.0499, "step": 1954 }, { "epoch": 1.9, "learning_rate": 1.3597863304285475e-07, "loss": 0.0473, "step": 1955 }, { "epoch": 1.9, "learning_rate": 1.3340664445507968e-07, "loss": 0.0503, "step": 1956 }, { "epoch": 1.9, "learning_rate": 1.3085904946840034e-07, "loss": 0.0495, "step": 1957 }, { "epoch": 1.9, "learning_rate": 1.2833585438134288e-07, "loss": 0.0507, "step": 1958 }, { "epoch": 1.9, "learning_rate": 1.2583706543211062e-07, "loss": 0.0573, "step": 1959 }, { "epoch": 1.9, "learning_rate": 1.2336268879856728e-07, "loss": 0.0462, "step": 1960 }, { "epoch": 1.9, "learning_rate": 1.209127305982205e-07, "loss": 0.0451, "step": 1961 }, { "epoch": 1.9, "learning_rate": 1.1848719688820398e-07, "loss": 0.0512, "step": 1962 }, { "epoch": 1.91, "learning_rate": 1.1608609366527191e-07, "loss": 0.0443, "step": 1963 }, { "epoch": 1.91, "learning_rate": 1.1370942686577346e-07, "loss": 0.0535, "step": 1964 }, { "epoch": 1.91, "learning_rate": 1.1135720236564284e-07, "loss": 0.0521, "step": 1965 }, { "epoch": 1.91, "learning_rate": 1.0902942598038702e-07, "loss": 0.0517, "step": 1966 }, { "epoch": 1.91, "learning_rate": 1.0672610346506462e-07, "loss": 0.0551, "step": 1967 }, { "epoch": 1.91, "learning_rate": 1.0444724051428157e-07, "loss": 0.0474, "step": 1968 }, { "epoch": 1.91, "learning_rate": 1.0219284276216545e-07, "loss": 0.0467, "step": 1969 }, { "epoch": 1.91, "learning_rate": 9.996291578236228e-08, "loss": 0.0518, "step": 1970 }, { "epoch": 1.91, "learning_rate": 9.775746508801531e-08, "loss": 0.0458, "step": 1971 }, { "epoch": 1.91, "learning_rate": 9.557649613175401e-08, "loss": 0.0526, "step": 1972 }, { "epoch": 1.92, "learning_rate": 9.342001430568292e-08, "loss": 0.0491, "step": 1973 }, { "epoch": 1.92, "learning_rate": 9.12880249413628e-08, "loss": 0.048, "step": 1974 }, { "epoch": 1.92, "learning_rate": 8.918053330980392e-08, "loss": 0.0494, "step": 1975 }, { "epoch": 1.92, "learning_rate": 8.709754462144615e-08, "loss": 0.0504, "step": 1976 }, { "epoch": 1.92, "learning_rate": 8.503906402615225e-08, "loss": 0.0468, "step": 1977 }, { "epoch": 1.92, "learning_rate": 8.300509661319234e-08, "loss": 0.0415, "step": 1978 }, { "epoch": 1.92, "learning_rate": 8.099564741123167e-08, "loss": 0.0518, "step": 1979 }, { "epoch": 1.92, "learning_rate": 7.901072138831512e-08, "loss": 0.0534, "step": 1980 }, { "epoch": 1.92, "learning_rate": 7.70503234518638e-08, "loss": 0.0487, "step": 1981 }, { "epoch": 1.92, "learning_rate": 7.511445844864962e-08, "loss": 0.0562, "step": 1982 }, { "epoch": 1.93, "learning_rate": 7.320313116479738e-08, "loss": 0.0503, "step": 1983 }, { "epoch": 1.93, "learning_rate": 7.131634632576268e-08, "loss": 0.0518, "step": 1984 }, { "epoch": 1.93, "learning_rate": 6.945410859632295e-08, "loss": 0.0467, "step": 1985 }, { "epoch": 1.93, "learning_rate": 6.761642258056977e-08, "loss": 0.0439, "step": 1986 }, { "epoch": 1.93, "learning_rate": 6.580329282189325e-08, "loss": 0.0508, "step": 1987 }, { "epoch": 1.93, "learning_rate": 6.401472380297091e-08, "loss": 0.0453, "step": 1988 }, { "epoch": 1.93, "learning_rate": 6.225071994575782e-08, "loss": 0.0526, "step": 1989 }, { "epoch": 1.93, "learning_rate": 6.051128561147756e-08, "loss": 0.0461, "step": 1990 }, { "epoch": 1.93, "learning_rate": 5.87964251006079e-08, "loss": 0.0551, "step": 1991 }, { "epoch": 1.93, "learning_rate": 5.7106142652870736e-08, "loss": 0.0554, "step": 1992 }, { "epoch": 1.93, "learning_rate": 5.544044244722435e-08, "loss": 0.0462, "step": 1993 }, { "epoch": 1.94, "learning_rate": 5.379932860185122e-08, "loss": 0.0527, "step": 1994 }, { "epoch": 1.94, "learning_rate": 5.218280517414687e-08, "loss": 0.0459, "step": 1995 }, { "epoch": 1.94, "learning_rate": 5.059087616071212e-08, "loss": 0.054, "step": 1996 }, { "epoch": 1.94, "learning_rate": 4.902354549733979e-08, "loss": 0.0558, "step": 1997 }, { "epoch": 1.94, "learning_rate": 4.748081705900909e-08, "loss": 0.0504, "step": 1998 }, { "epoch": 1.94, "learning_rate": 4.596269465987457e-08, "loss": 0.0525, "step": 1999 }, { "epoch": 1.94, "learning_rate": 4.44691820532539e-08, "loss": 0.0474, "step": 2000 }, { "epoch": 1.94, "learning_rate": 4.300028293162339e-08, "loss": 0.0613, "step": 2001 }, { "epoch": 1.94, "learning_rate": 4.155600092660361e-08, "loss": 0.0493, "step": 2002 }, { "epoch": 1.94, "learning_rate": 4.0136339608957134e-08, "loss": 0.0469, "step": 2003 }, { "epoch": 1.95, "learning_rate": 3.8741302488570774e-08, "loss": 0.0556, "step": 2004 }, { "epoch": 1.95, "learning_rate": 3.737089301445562e-08, "loss": 0.0591, "step": 2005 }, { "epoch": 1.95, "learning_rate": 3.602511457473479e-08, "loss": 0.0399, "step": 2006 }, { "epoch": 1.95, "learning_rate": 3.470397049663343e-08, "loss": 0.0493, "step": 2007 }, { "epoch": 1.95, "learning_rate": 3.340746404647099e-08, "loss": 0.0514, "step": 2008 }, { "epoch": 1.95, "learning_rate": 3.2135598429657855e-08, "loss": 0.0422, "step": 2009 }, { "epoch": 1.95, "learning_rate": 3.08883767906798e-08, "loss": 0.0508, "step": 2010 }, { "epoch": 1.95, "learning_rate": 2.9665802213098004e-08, "loss": 0.0592, "step": 2011 }, { "epoch": 1.95, "learning_rate": 2.8467877719535743e-08, "loss": 0.049, "step": 2012 }, { "epoch": 1.95, "learning_rate": 2.7294606271672795e-08, "loss": 0.0453, "step": 2013 }, { "epoch": 1.96, "learning_rate": 2.6145990770238827e-08, "loss": 0.0536, "step": 2014 }, { "epoch": 1.96, "learning_rate": 2.5022034055003363e-08, "loss": 0.0514, "step": 2015 }, { "epoch": 1.96, "learning_rate": 2.392273890477359e-08, "loss": 0.049, "step": 2016 }, { "epoch": 1.96, "learning_rate": 2.2848108037385462e-08, "loss": 0.0461, "step": 2017 }, { "epoch": 1.96, "learning_rate": 2.179814410969261e-08, "loss": 0.0462, "step": 2018 }, { "epoch": 1.96, "learning_rate": 2.0772849717567434e-08, "loss": 0.0386, "step": 2019 }, { "epoch": 1.96, "learning_rate": 1.977222739588891e-08, "loss": 0.057, "step": 2020 }, { "epoch": 1.96, "learning_rate": 1.8796279618537026e-08, "loss": 0.0526, "step": 2021 }, { "epoch": 1.96, "learning_rate": 1.7845008798391684e-08, "loss": 0.0543, "step": 2022 }, { "epoch": 1.96, "learning_rate": 1.6918417287318245e-08, "loss": 0.047, "step": 2023 }, { "epoch": 1.97, "learning_rate": 1.6016507376169776e-08, "loss": 0.0467, "step": 2024 }, { "epoch": 1.97, "learning_rate": 1.513928129477593e-08, "loss": 0.0536, "step": 2025 }, { "epoch": 1.97, "learning_rate": 1.4286741211940735e-08, "loss": 0.0462, "step": 2026 }, { "epoch": 1.97, "learning_rate": 1.3458889235435924e-08, "loss": 0.0465, "step": 2027 }, { "epoch": 1.97, "learning_rate": 1.2655727411994278e-08, "loss": 0.052, "step": 2028 }, { "epoch": 1.97, "learning_rate": 1.1877257727308521e-08, "loss": 0.0464, "step": 2029 }, { "epoch": 1.97, "learning_rate": 1.1123482106021322e-08, "loss": 0.0535, "step": 2030 }, { "epoch": 1.97, "learning_rate": 1.0394402411725291e-08, "loss": 0.0489, "step": 2031 }, { "epoch": 1.97, "learning_rate": 9.690020446956328e-09, "loss": 0.0489, "step": 2032 }, { "epoch": 1.97, "learning_rate": 9.010337953185843e-09, "loss": 0.0494, "step": 2033 }, { "epoch": 1.97, "learning_rate": 8.355356610822984e-09, "loss": 0.0436, "step": 2034 }, { "epoch": 1.98, "learning_rate": 7.725078039205747e-09, "loss": 0.0462, "step": 2035 }, { "epoch": 1.98, "learning_rate": 7.119503796599869e-09, "loss": 0.0452, "step": 2036 }, { "epoch": 1.98, "learning_rate": 6.538635380191061e-09, "loss": 0.0524, "step": 2037 }, { "epoch": 1.98, "learning_rate": 5.98247422608722e-09, "loss": 0.0538, "step": 2038 }, { "epoch": 1.98, "learning_rate": 5.451021709307336e-09, "loss": 0.0482, "step": 2039 }, { "epoch": 1.98, "learning_rate": 4.944279143784814e-09, "loss": 0.0472, "step": 2040 }, { "epoch": 1.98, "learning_rate": 4.462247782360818e-09, "loss": 0.0485, "step": 2041 }, { "epoch": 1.98, "learning_rate": 4.00492881678427e-09, "loss": 0.0501, "step": 2042 }, { "epoch": 1.98, "learning_rate": 3.572323377700748e-09, "loss": 0.0451, "step": 2043 }, { "epoch": 1.98, "learning_rate": 3.1644325346624762e-09, "loss": 0.0493, "step": 2044 }, { "epoch": 1.99, "learning_rate": 2.7812572961127825e-09, "loss": 0.0544, "step": 2045 }, { "epoch": 1.99, "learning_rate": 2.422798609393873e-09, "loss": 0.0522, "step": 2046 }, { "epoch": 1.99, "learning_rate": 2.089057360737945e-09, "loss": 0.047, "step": 2047 }, { "epoch": 1.99, "learning_rate": 1.7800343752683025e-09, "loss": 0.0518, "step": 2048 }, { "epoch": 1.99, "learning_rate": 1.4957304169960218e-09, "loss": 0.0501, "step": 2049 }, { "epoch": 1.99, "learning_rate": 1.2361461888166227e-09, "loss": 0.0548, "step": 2050 }, { "epoch": 1.99, "learning_rate": 1.0012823325111776e-09, "loss": 0.0511, "step": 2051 }, { "epoch": 1.99, "learning_rate": 7.911394287452023e-10, "loss": 0.0468, "step": 2052 }, { "epoch": 1.99, "learning_rate": 6.057179970631044e-10, "loss": 0.049, "step": 2053 }, { "epoch": 1.99, "learning_rate": 4.4501849589040357e-10, "loss": 0.049, "step": 2054 }, { "epoch": 2.0, "learning_rate": 3.0904132253151144e-10, "loss": 0.044, "step": 2055 }, { "epoch": 2.0, "learning_rate": 1.977868131675109e-10, "loss": 0.0437, "step": 2056 }, { "epoch": 2.0, "learning_rate": 1.1125524285948707e-10, "loss": 0.0454, "step": 2057 }, { "epoch": 2.0, "learning_rate": 4.9446825540755416e-11, "loss": 0.051, "step": 2058 }, { "epoch": 2.0, "learning_rate": 1.236171402574371e-11, "loss": 0.0504, "step": 2059 }, { "epoch": 2.0, "learning_rate": 0.0, "loss": 0.0454, "step": 2060 }, { "epoch": 2.0, "step": 2060, "total_flos": 255360458293248.0, "train_loss": 0.07134663109640474, "train_runtime": 26413.6011, "train_samples_per_second": 19.964, "train_steps_per_second": 0.078 } ], "max_steps": 2060, "num_train_epochs": 2, "total_flos": 255360458293248.0, "trial_name": null, "trial_params": null }