{ "best_metric": null, "best_model_checkpoint": null, "epoch": 0.19141389066677833, "eval_steps": 500, "global_step": 8000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 5e-06, "loss": 2.7065, "step": 5 }, { "epoch": 0.0, "learning_rate": 1e-05, "loss": 2.7451, "step": 10 }, { "epoch": 0.0, "learning_rate": 1.5e-05, "loss": 2.7756, "step": 15 }, { "epoch": 0.0, "learning_rate": 2e-05, "loss": 2.7781, "step": 20 }, { "epoch": 0.0, "learning_rate": 2.5e-05, "loss": 2.7405, "step": 25 }, { "epoch": 0.0, "learning_rate": 3e-05, "loss": 2.7645, "step": 30 }, { "epoch": 0.0, "learning_rate": 3.5e-05, "loss": 2.7564, "step": 35 }, { "epoch": 0.0, "learning_rate": 4e-05, "loss": 2.692, "step": 40 }, { "epoch": 0.0, "learning_rate": 4.5e-05, "loss": 2.7092, "step": 45 }, { "epoch": 0.0, "learning_rate": 5e-05, "loss": 2.7576, "step": 50 }, { "epoch": 0.0, "learning_rate": 5.500000000000001e-05, "loss": 2.6894, "step": 55 }, { "epoch": 0.0, "learning_rate": 6e-05, "loss": 2.7501, "step": 60 }, { "epoch": 0.0, "learning_rate": 6.500000000000001e-05, "loss": 2.689, "step": 65 }, { "epoch": 0.0, "learning_rate": 7e-05, "loss": 2.7256, "step": 70 }, { "epoch": 0.0, "learning_rate": 7.500000000000001e-05, "loss": 2.7196, "step": 75 }, { "epoch": 0.0, "learning_rate": 8e-05, "loss": 2.6977, "step": 80 }, { "epoch": 0.0, "learning_rate": 8.5e-05, "loss": 2.6785, "step": 85 }, { "epoch": 0.0, "learning_rate": 9e-05, "loss": 2.6907, "step": 90 }, { "epoch": 0.0, "learning_rate": 9.5e-05, "loss": 2.7362, "step": 95 }, { "epoch": 0.0, "learning_rate": 0.0001, "loss": 2.6994, "step": 100 }, { "epoch": 0.0, "learning_rate": 9.99999964515995e-05, "loss": 2.7181, "step": 105 }, { "epoch": 0.0, "learning_rate": 9.999998580639846e-05, "loss": 2.7631, "step": 110 }, { "epoch": 0.0, "learning_rate": 9.99999680643984e-05, "loss": 2.7108, "step": 115 }, { "epoch": 0.0, "learning_rate": 9.999994322560186e-05, "loss": 2.6916, "step": 120 }, { "epoch": 0.0, "learning_rate": 9.999991129001234e-05, "loss": 2.6967, "step": 125 }, { "epoch": 0.0, "learning_rate": 9.999987225763438e-05, "loss": 2.7368, "step": 130 }, { "epoch": 0.0, "learning_rate": 9.999982612847353e-05, "loss": 2.7164, "step": 135 }, { "epoch": 0.0, "learning_rate": 9.999977290253635e-05, "loss": 2.701, "step": 140 }, { "epoch": 0.0, "learning_rate": 9.999971257983034e-05, "loss": 2.6644, "step": 145 }, { "epoch": 0.0, "learning_rate": 9.999964516036413e-05, "loss": 2.7154, "step": 150 }, { "epoch": 0.0, "learning_rate": 9.999957064414724e-05, "loss": 2.6617, "step": 155 }, { "epoch": 0.0, "learning_rate": 9.999948903119026e-05, "loss": 2.7067, "step": 160 }, { "epoch": 0.0, "learning_rate": 9.999940032150479e-05, "loss": 2.6513, "step": 165 }, { "epoch": 0.0, "learning_rate": 9.99993045151034e-05, "loss": 2.6859, "step": 170 }, { "epoch": 0.0, "learning_rate": 9.99992016119997e-05, "loss": 2.6848, "step": 175 }, { "epoch": 0.0, "learning_rate": 9.999909161220829e-05, "loss": 2.6185, "step": 180 }, { "epoch": 0.0, "learning_rate": 9.999897451574478e-05, "loss": 2.6641, "step": 185 }, { "epoch": 0.0, "learning_rate": 9.999885032262579e-05, "loss": 2.6396, "step": 190 }, { "epoch": 0.0, "learning_rate": 9.999871903286897e-05, "loss": 2.6739, "step": 195 }, { "epoch": 0.0, "learning_rate": 9.999858064649292e-05, "loss": 2.6407, "step": 200 }, { "epoch": 0.0, "learning_rate": 9.999843516351732e-05, "loss": 2.6133, "step": 205 }, { "epoch": 0.01, "learning_rate": 9.99982825839628e-05, "loss": 2.6816, "step": 210 }, { "epoch": 0.01, "learning_rate": 9.999812290785098e-05, "loss": 2.6944, "step": 215 }, { "epoch": 0.01, "learning_rate": 9.999795613520459e-05, "loss": 2.7011, "step": 220 }, { "epoch": 0.01, "learning_rate": 9.999778226604727e-05, "loss": 2.6659, "step": 225 }, { "epoch": 0.01, "learning_rate": 9.999760130040369e-05, "loss": 2.6351, "step": 230 }, { "epoch": 0.01, "learning_rate": 9.999741323829955e-05, "loss": 2.5926, "step": 235 }, { "epoch": 0.01, "learning_rate": 9.999721807976152e-05, "loss": 2.6288, "step": 240 }, { "epoch": 0.01, "learning_rate": 9.999701582481733e-05, "loss": 2.7157, "step": 245 }, { "epoch": 0.01, "learning_rate": 9.999680647349569e-05, "loss": 2.6409, "step": 250 }, { "epoch": 0.01, "learning_rate": 9.999659002582628e-05, "loss": 2.6672, "step": 255 }, { "epoch": 0.01, "learning_rate": 9.999636648183983e-05, "loss": 2.641, "step": 260 }, { "epoch": 0.01, "learning_rate": 9.99961358415681e-05, "loss": 2.6927, "step": 265 }, { "epoch": 0.01, "learning_rate": 9.999589810504379e-05, "loss": 2.6929, "step": 270 }, { "epoch": 0.01, "learning_rate": 9.999565327230067e-05, "loss": 2.7246, "step": 275 }, { "epoch": 0.01, "learning_rate": 9.999540134337349e-05, "loss": 2.6322, "step": 280 }, { "epoch": 0.01, "learning_rate": 9.999514231829798e-05, "loss": 2.6313, "step": 285 }, { "epoch": 0.01, "learning_rate": 9.999487619711092e-05, "loss": 2.6511, "step": 290 }, { "epoch": 0.01, "learning_rate": 9.99946029798501e-05, "loss": 2.6253, "step": 295 }, { "epoch": 0.01, "learning_rate": 9.999432266655426e-05, "loss": 2.6525, "step": 300 }, { "epoch": 0.01, "learning_rate": 9.999403525726323e-05, "loss": 2.618, "step": 305 }, { "epoch": 0.01, "learning_rate": 9.999374075201778e-05, "loss": 2.6965, "step": 310 }, { "epoch": 0.01, "learning_rate": 9.999343915085971e-05, "loss": 2.6172, "step": 315 }, { "epoch": 0.01, "learning_rate": 9.999313045383184e-05, "loss": 2.6348, "step": 320 }, { "epoch": 0.01, "learning_rate": 9.999281466097799e-05, "loss": 2.6008, "step": 325 }, { "epoch": 0.01, "learning_rate": 9.999249177234294e-05, "loss": 2.7078, "step": 330 }, { "epoch": 0.01, "learning_rate": 9.999216178797257e-05, "loss": 2.6293, "step": 335 }, { "epoch": 0.01, "learning_rate": 9.999182470791368e-05, "loss": 2.6481, "step": 340 }, { "epoch": 0.01, "learning_rate": 9.999148053221416e-05, "loss": 2.5894, "step": 345 }, { "epoch": 0.01, "learning_rate": 9.99911292609228e-05, "loss": 2.6599, "step": 350 }, { "epoch": 0.01, "learning_rate": 9.999077089408951e-05, "loss": 2.6062, "step": 355 }, { "epoch": 0.01, "learning_rate": 9.999040543176513e-05, "loss": 2.6659, "step": 360 }, { "epoch": 0.01, "learning_rate": 9.999003287400153e-05, "loss": 2.6465, "step": 365 }, { "epoch": 0.01, "learning_rate": 9.99896532208516e-05, "loss": 2.6539, "step": 370 }, { "epoch": 0.01, "learning_rate": 9.998926647236923e-05, "loss": 2.6109, "step": 375 }, { "epoch": 0.01, "learning_rate": 9.998887262860931e-05, "loss": 2.5978, "step": 380 }, { "epoch": 0.01, "learning_rate": 9.998847168962772e-05, "loss": 2.6841, "step": 385 }, { "epoch": 0.01, "learning_rate": 9.99880636554814e-05, "loss": 2.6343, "step": 390 }, { "epoch": 0.01, "learning_rate": 9.998764852622823e-05, "loss": 2.627, "step": 395 }, { "epoch": 0.01, "learning_rate": 9.998722630192718e-05, "loss": 2.6291, "step": 400 }, { "epoch": 0.01, "learning_rate": 9.998679698263813e-05, "loss": 2.5715, "step": 405 }, { "epoch": 0.01, "learning_rate": 9.998636056842204e-05, "loss": 2.5893, "step": 410 }, { "epoch": 0.01, "learning_rate": 9.998591705934086e-05, "loss": 2.6485, "step": 415 }, { "epoch": 0.01, "learning_rate": 9.998546645545751e-05, "loss": 2.6241, "step": 420 }, { "epoch": 0.01, "learning_rate": 9.998500875683597e-05, "loss": 2.6429, "step": 425 }, { "epoch": 0.01, "learning_rate": 9.998454396354121e-05, "loss": 2.6412, "step": 430 }, { "epoch": 0.01, "learning_rate": 9.998407207563918e-05, "loss": 2.6068, "step": 435 }, { "epoch": 0.01, "learning_rate": 9.998359309319686e-05, "loss": 2.564, "step": 440 }, { "epoch": 0.01, "learning_rate": 9.998310701628228e-05, "loss": 2.6616, "step": 445 }, { "epoch": 0.01, "learning_rate": 9.998261384496437e-05, "loss": 2.568, "step": 450 }, { "epoch": 0.01, "learning_rate": 9.998211357931316e-05, "loss": 2.5775, "step": 455 }, { "epoch": 0.01, "learning_rate": 9.998160621939962e-05, "loss": 2.6556, "step": 460 }, { "epoch": 0.01, "learning_rate": 9.998109176529582e-05, "loss": 2.6, "step": 465 }, { "epoch": 0.01, "learning_rate": 9.998057021707476e-05, "loss": 2.617, "step": 470 }, { "epoch": 0.01, "learning_rate": 9.998004157481045e-05, "loss": 2.6154, "step": 475 }, { "epoch": 0.01, "learning_rate": 9.997950583857792e-05, "loss": 2.6441, "step": 480 }, { "epoch": 0.01, "learning_rate": 9.997896300845322e-05, "loss": 2.6782, "step": 485 }, { "epoch": 0.01, "learning_rate": 9.997841308451342e-05, "loss": 2.6265, "step": 490 }, { "epoch": 0.01, "learning_rate": 9.997785606683654e-05, "loss": 2.5947, "step": 495 }, { "epoch": 0.01, "learning_rate": 9.997729195550167e-05, "loss": 2.655, "step": 500 }, { "epoch": 0.01, "learning_rate": 9.997672075058884e-05, "loss": 2.5909, "step": 505 }, { "epoch": 0.01, "learning_rate": 9.997614245217916e-05, "loss": 2.6315, "step": 510 }, { "epoch": 0.01, "learning_rate": 9.997555706035469e-05, "loss": 2.6214, "step": 515 }, { "epoch": 0.01, "learning_rate": 9.997496457519852e-05, "loss": 2.5717, "step": 520 }, { "epoch": 0.01, "learning_rate": 9.997436499679475e-05, "loss": 2.627, "step": 525 }, { "epoch": 0.01, "learning_rate": 9.997375832522851e-05, "loss": 2.6226, "step": 530 }, { "epoch": 0.01, "learning_rate": 9.997314456058586e-05, "loss": 2.6466, "step": 535 }, { "epoch": 0.01, "learning_rate": 9.997252370295394e-05, "loss": 2.5526, "step": 540 }, { "epoch": 0.01, "learning_rate": 9.997189575242087e-05, "loss": 2.5878, "step": 545 }, { "epoch": 0.01, "learning_rate": 9.997126070907577e-05, "loss": 2.613, "step": 550 }, { "epoch": 0.01, "learning_rate": 9.997061857300878e-05, "loss": 2.5949, "step": 555 }, { "epoch": 0.01, "learning_rate": 9.996996934431105e-05, "loss": 2.6208, "step": 560 }, { "epoch": 0.01, "learning_rate": 9.996931302307472e-05, "loss": 2.6017, "step": 565 }, { "epoch": 0.01, "learning_rate": 9.996864960939296e-05, "loss": 2.6222, "step": 570 }, { "epoch": 0.01, "learning_rate": 9.996797910335994e-05, "loss": 2.6141, "step": 575 }, { "epoch": 0.01, "learning_rate": 9.996730150507077e-05, "loss": 2.6224, "step": 580 }, { "epoch": 0.01, "learning_rate": 9.996661681462169e-05, "loss": 2.6199, "step": 585 }, { "epoch": 0.01, "learning_rate": 9.996592503210985e-05, "loss": 2.5872, "step": 590 }, { "epoch": 0.01, "learning_rate": 9.996522615763347e-05, "loss": 2.6344, "step": 595 }, { "epoch": 0.01, "learning_rate": 9.996452019129169e-05, "loss": 2.6582, "step": 600 }, { "epoch": 0.01, "learning_rate": 9.996380713318476e-05, "loss": 2.6449, "step": 605 }, { "epoch": 0.01, "learning_rate": 9.996308698341388e-05, "loss": 2.6177, "step": 610 }, { "epoch": 0.01, "learning_rate": 9.996235974208125e-05, "loss": 2.5663, "step": 615 }, { "epoch": 0.01, "learning_rate": 9.996162540929011e-05, "loss": 2.6485, "step": 620 }, { "epoch": 0.01, "learning_rate": 9.996088398514466e-05, "loss": 2.689, "step": 625 }, { "epoch": 0.02, "learning_rate": 9.996013546975016e-05, "loss": 2.5926, "step": 630 }, { "epoch": 0.02, "learning_rate": 9.995937986321285e-05, "loss": 2.6152, "step": 635 }, { "epoch": 0.02, "learning_rate": 9.995861716563998e-05, "loss": 2.5543, "step": 640 }, { "epoch": 0.02, "learning_rate": 9.995784737713979e-05, "loss": 2.6036, "step": 645 }, { "epoch": 0.02, "learning_rate": 9.995707049782155e-05, "loss": 2.6075, "step": 650 }, { "epoch": 0.02, "learning_rate": 9.995628652779551e-05, "loss": 2.5944, "step": 655 }, { "epoch": 0.02, "learning_rate": 9.995549546717297e-05, "loss": 2.6122, "step": 660 }, { "epoch": 0.02, "learning_rate": 9.995469731606622e-05, "loss": 2.6195, "step": 665 }, { "epoch": 0.02, "learning_rate": 9.99538920745885e-05, "loss": 2.6113, "step": 670 }, { "epoch": 0.02, "learning_rate": 9.995307974285412e-05, "loss": 2.6005, "step": 675 }, { "epoch": 0.02, "learning_rate": 9.99522603209784e-05, "loss": 2.6105, "step": 680 }, { "epoch": 0.02, "learning_rate": 9.995143380907764e-05, "loss": 2.6144, "step": 685 }, { "epoch": 0.02, "learning_rate": 9.995060020726914e-05, "loss": 2.6203, "step": 690 }, { "epoch": 0.02, "learning_rate": 9.994975951567121e-05, "loss": 2.6055, "step": 695 }, { "epoch": 0.02, "learning_rate": 9.99489117344032e-05, "loss": 2.606, "step": 700 }, { "epoch": 0.02, "learning_rate": 9.994805686358541e-05, "loss": 2.5767, "step": 705 }, { "epoch": 0.02, "learning_rate": 9.994719490333922e-05, "loss": 2.5316, "step": 710 }, { "epoch": 0.02, "learning_rate": 9.994632585378694e-05, "loss": 2.5989, "step": 715 }, { "epoch": 0.02, "learning_rate": 9.99454497150519e-05, "loss": 2.582, "step": 720 }, { "epoch": 0.02, "learning_rate": 9.994456648725851e-05, "loss": 2.5758, "step": 725 }, { "epoch": 0.02, "learning_rate": 9.994367617053211e-05, "loss": 2.582, "step": 730 }, { "epoch": 0.02, "learning_rate": 9.994277876499905e-05, "loss": 2.5533, "step": 735 }, { "epoch": 0.02, "learning_rate": 9.994187427078672e-05, "loss": 2.6215, "step": 740 }, { "epoch": 0.02, "learning_rate": 9.994096268802349e-05, "loss": 2.6408, "step": 745 }, { "epoch": 0.02, "learning_rate": 9.994004401683875e-05, "loss": 2.6268, "step": 750 }, { "epoch": 0.02, "learning_rate": 9.99391182573629e-05, "loss": 2.624, "step": 755 }, { "epoch": 0.02, "learning_rate": 9.993818540972734e-05, "loss": 2.6075, "step": 760 }, { "epoch": 0.02, "learning_rate": 9.993724547406447e-05, "loss": 2.6207, "step": 765 }, { "epoch": 0.02, "learning_rate": 9.993629845050769e-05, "loss": 2.6489, "step": 770 }, { "epoch": 0.02, "learning_rate": 9.993534433919144e-05, "loss": 2.6037, "step": 775 }, { "epoch": 0.02, "learning_rate": 9.993438314025112e-05, "loss": 2.6435, "step": 780 }, { "epoch": 0.02, "learning_rate": 9.993341485382318e-05, "loss": 2.6431, "step": 785 }, { "epoch": 0.02, "learning_rate": 9.993243948004503e-05, "loss": 2.5807, "step": 790 }, { "epoch": 0.02, "learning_rate": 9.993145701905513e-05, "loss": 2.5836, "step": 795 }, { "epoch": 0.02, "learning_rate": 9.99304674709929e-05, "loss": 2.6496, "step": 800 }, { "epoch": 0.02, "learning_rate": 9.992947083599884e-05, "loss": 2.5345, "step": 805 }, { "epoch": 0.02, "learning_rate": 9.992846711421437e-05, "loss": 2.6104, "step": 810 }, { "epoch": 0.02, "learning_rate": 9.992745630578197e-05, "loss": 2.6352, "step": 815 }, { "epoch": 0.02, "learning_rate": 9.992643841084511e-05, "loss": 2.5592, "step": 820 }, { "epoch": 0.02, "learning_rate": 9.992541342954825e-05, "loss": 2.6059, "step": 825 }, { "epoch": 0.02, "learning_rate": 9.992438136203688e-05, "loss": 2.6622, "step": 830 }, { "epoch": 0.02, "learning_rate": 9.99233422084575e-05, "loss": 2.5513, "step": 835 }, { "epoch": 0.02, "learning_rate": 9.992229596895759e-05, "loss": 2.6403, "step": 840 }, { "epoch": 0.02, "learning_rate": 9.992124264368566e-05, "loss": 2.5768, "step": 845 }, { "epoch": 0.02, "learning_rate": 9.99201822327912e-05, "loss": 2.6033, "step": 850 }, { "epoch": 0.02, "learning_rate": 9.991911473642474e-05, "loss": 2.6022, "step": 855 }, { "epoch": 0.02, "learning_rate": 9.991804015473777e-05, "loss": 2.5283, "step": 860 }, { "epoch": 0.02, "learning_rate": 9.991695848788282e-05, "loss": 2.5966, "step": 865 }, { "epoch": 0.02, "learning_rate": 9.991586973601344e-05, "loss": 2.5594, "step": 870 }, { "epoch": 0.02, "learning_rate": 9.991477389928415e-05, "loss": 2.6374, "step": 875 }, { "epoch": 0.02, "learning_rate": 9.991367097785049e-05, "loss": 2.6379, "step": 880 }, { "epoch": 0.02, "learning_rate": 9.991256097186899e-05, "loss": 2.6089, "step": 885 }, { "epoch": 0.02, "learning_rate": 9.99114438814972e-05, "loss": 2.5478, "step": 890 }, { "epoch": 0.02, "learning_rate": 9.99103197068937e-05, "loss": 2.6591, "step": 895 }, { "epoch": 0.02, "learning_rate": 9.990918844821803e-05, "loss": 2.6427, "step": 900 }, { "epoch": 0.02, "learning_rate": 9.990805010563075e-05, "loss": 2.6166, "step": 905 }, { "epoch": 0.02, "learning_rate": 9.990690467929346e-05, "loss": 2.6097, "step": 910 }, { "epoch": 0.02, "learning_rate": 9.990575216936873e-05, "loss": 2.5737, "step": 915 }, { "epoch": 0.02, "learning_rate": 9.990459257602011e-05, "loss": 2.6036, "step": 920 }, { "epoch": 0.02, "learning_rate": 9.990342589941224e-05, "loss": 2.5669, "step": 925 }, { "epoch": 0.02, "learning_rate": 9.990225213971067e-05, "loss": 2.5647, "step": 930 }, { "epoch": 0.02, "learning_rate": 9.990107129708201e-05, "loss": 2.6438, "step": 935 }, { "epoch": 0.02, "learning_rate": 9.989988337169387e-05, "loss": 2.6354, "step": 940 }, { "epoch": 0.02, "learning_rate": 9.989868836371488e-05, "loss": 2.6415, "step": 945 }, { "epoch": 0.02, "learning_rate": 9.98974862733146e-05, "loss": 2.5875, "step": 950 }, { "epoch": 0.02, "learning_rate": 9.989627710066371e-05, "loss": 2.565, "step": 955 }, { "epoch": 0.02, "learning_rate": 9.989506084593379e-05, "loss": 2.5929, "step": 960 }, { "epoch": 0.02, "learning_rate": 9.98938375092975e-05, "loss": 2.5621, "step": 965 }, { "epoch": 0.02, "learning_rate": 9.989260709092844e-05, "loss": 2.5521, "step": 970 }, { "epoch": 0.02, "learning_rate": 9.98913695910013e-05, "loss": 2.5617, "step": 975 }, { "epoch": 0.02, "learning_rate": 9.98901250096917e-05, "loss": 2.5455, "step": 980 }, { "epoch": 0.02, "learning_rate": 9.988887334717628e-05, "loss": 2.569, "step": 985 }, { "epoch": 0.02, "learning_rate": 9.988761460363272e-05, "loss": 2.5779, "step": 990 }, { "epoch": 0.02, "learning_rate": 9.988634877923966e-05, "loss": 2.5948, "step": 995 }, { "epoch": 0.02, "learning_rate": 9.988507587417677e-05, "loss": 2.5499, "step": 1000 }, { "epoch": 0.02, "learning_rate": 9.988379588862472e-05, "loss": 2.5395, "step": 1005 }, { "epoch": 0.02, "learning_rate": 9.98825088227652e-05, "loss": 2.63, "step": 1010 }, { "epoch": 0.02, "learning_rate": 9.988121467678088e-05, "loss": 2.5484, "step": 1015 }, { "epoch": 0.02, "learning_rate": 9.987991345085545e-05, "loss": 2.5774, "step": 1020 }, { "epoch": 0.02, "learning_rate": 9.98786051451736e-05, "loss": 2.5511, "step": 1025 }, { "epoch": 0.02, "learning_rate": 9.987728975992103e-05, "loss": 2.6132, "step": 1030 }, { "epoch": 0.02, "learning_rate": 9.987596729528444e-05, "loss": 2.6051, "step": 1035 }, { "epoch": 0.02, "learning_rate": 9.98746377514515e-05, "loss": 2.5601, "step": 1040 }, { "epoch": 0.03, "learning_rate": 9.987330112861098e-05, "loss": 2.5438, "step": 1045 }, { "epoch": 0.03, "learning_rate": 9.987195742695256e-05, "loss": 2.6182, "step": 1050 }, { "epoch": 0.03, "learning_rate": 9.987060664666696e-05, "loss": 2.5861, "step": 1055 }, { "epoch": 0.03, "learning_rate": 9.986924878794591e-05, "loss": 2.6084, "step": 1060 }, { "epoch": 0.03, "learning_rate": 9.986788385098213e-05, "loss": 2.5716, "step": 1065 }, { "epoch": 0.03, "learning_rate": 9.986651183596938e-05, "loss": 2.577, "step": 1070 }, { "epoch": 0.03, "learning_rate": 9.986513274310238e-05, "loss": 2.5746, "step": 1075 }, { "epoch": 0.03, "learning_rate": 9.986374657257686e-05, "loss": 2.6135, "step": 1080 }, { "epoch": 0.03, "learning_rate": 9.98623533245896e-05, "loss": 2.6016, "step": 1085 }, { "epoch": 0.03, "learning_rate": 9.986095299933833e-05, "loss": 2.6191, "step": 1090 }, { "epoch": 0.03, "learning_rate": 9.985954559702179e-05, "loss": 2.54, "step": 1095 }, { "epoch": 0.03, "learning_rate": 9.985813111783979e-05, "loss": 2.5716, "step": 1100 }, { "epoch": 0.03, "learning_rate": 9.985670956199304e-05, "loss": 2.5914, "step": 1105 }, { "epoch": 0.03, "learning_rate": 9.985528092968336e-05, "loss": 2.5325, "step": 1110 }, { "epoch": 0.03, "learning_rate": 9.98538452211135e-05, "loss": 2.4852, "step": 1115 }, { "epoch": 0.03, "learning_rate": 9.985240243648722e-05, "loss": 2.5271, "step": 1120 }, { "epoch": 0.03, "learning_rate": 9.985095257600934e-05, "loss": 2.5751, "step": 1125 }, { "epoch": 0.03, "learning_rate": 9.984949563988562e-05, "loss": 2.6089, "step": 1130 }, { "epoch": 0.03, "learning_rate": 9.984803162832286e-05, "loss": 2.5949, "step": 1135 }, { "epoch": 0.03, "learning_rate": 9.984656054152886e-05, "loss": 2.5795, "step": 1140 }, { "epoch": 0.03, "learning_rate": 9.984508237971243e-05, "loss": 2.6057, "step": 1145 }, { "epoch": 0.03, "learning_rate": 9.984359714308335e-05, "loss": 2.6413, "step": 1150 }, { "epoch": 0.03, "learning_rate": 9.984210483185245e-05, "loss": 2.6175, "step": 1155 }, { "epoch": 0.03, "learning_rate": 9.984060544623153e-05, "loss": 2.6401, "step": 1160 }, { "epoch": 0.03, "learning_rate": 9.983909898643341e-05, "loss": 2.6539, "step": 1165 }, { "epoch": 0.03, "learning_rate": 9.983758545267192e-05, "loss": 2.6025, "step": 1170 }, { "epoch": 0.03, "learning_rate": 9.983606484516186e-05, "loss": 2.5708, "step": 1175 }, { "epoch": 0.03, "learning_rate": 9.983453716411909e-05, "loss": 2.547, "step": 1180 }, { "epoch": 0.03, "learning_rate": 9.983300240976042e-05, "loss": 2.5071, "step": 1185 }, { "epoch": 0.03, "learning_rate": 9.98314605823037e-05, "loss": 2.5475, "step": 1190 }, { "epoch": 0.03, "learning_rate": 9.982991168196778e-05, "loss": 2.6066, "step": 1195 }, { "epoch": 0.03, "learning_rate": 9.982835570897247e-05, "loss": 2.5219, "step": 1200 }, { "epoch": 0.03, "learning_rate": 9.982679266353865e-05, "loss": 2.5757, "step": 1205 }, { "epoch": 0.03, "learning_rate": 9.982522254588816e-05, "loss": 2.5991, "step": 1210 }, { "epoch": 0.03, "learning_rate": 9.982364535624385e-05, "loss": 2.6027, "step": 1215 }, { "epoch": 0.03, "learning_rate": 9.982206109482962e-05, "loss": 2.5506, "step": 1220 }, { "epoch": 0.03, "learning_rate": 9.982046976187026e-05, "loss": 2.5499, "step": 1225 }, { "epoch": 0.03, "learning_rate": 9.98188713575917e-05, "loss": 2.61, "step": 1230 }, { "epoch": 0.03, "learning_rate": 9.981726588222079e-05, "loss": 2.6, "step": 1235 }, { "epoch": 0.03, "learning_rate": 9.981565333598542e-05, "loss": 2.5723, "step": 1240 }, { "epoch": 0.03, "learning_rate": 9.981403371911442e-05, "loss": 2.5462, "step": 1245 }, { "epoch": 0.03, "learning_rate": 9.981240703183775e-05, "loss": 2.5916, "step": 1250 }, { "epoch": 0.03, "learning_rate": 9.981077327438621e-05, "loss": 2.5988, "step": 1255 }, { "epoch": 0.03, "learning_rate": 9.980913244699175e-05, "loss": 2.6173, "step": 1260 }, { "epoch": 0.03, "learning_rate": 9.980748454988724e-05, "loss": 2.5795, "step": 1265 }, { "epoch": 0.03, "learning_rate": 9.980582958330657e-05, "loss": 2.5776, "step": 1270 }, { "epoch": 0.03, "learning_rate": 9.980416754748467e-05, "loss": 2.5629, "step": 1275 }, { "epoch": 0.03, "learning_rate": 9.98024984426574e-05, "loss": 2.5777, "step": 1280 }, { "epoch": 0.03, "learning_rate": 9.98008222690617e-05, "loss": 2.533, "step": 1285 }, { "epoch": 0.03, "learning_rate": 9.979913902693547e-05, "loss": 2.5228, "step": 1290 }, { "epoch": 0.03, "learning_rate": 9.979744871651762e-05, "loss": 2.6026, "step": 1295 }, { "epoch": 0.03, "learning_rate": 9.979575133804805e-05, "loss": 2.5699, "step": 1300 }, { "epoch": 0.03, "learning_rate": 9.979404689176772e-05, "loss": 2.581, "step": 1305 }, { "epoch": 0.03, "learning_rate": 9.97923353779185e-05, "loss": 2.5205, "step": 1310 }, { "epoch": 0.03, "learning_rate": 9.979061679674335e-05, "loss": 2.543, "step": 1315 }, { "epoch": 0.03, "learning_rate": 9.97888911484862e-05, "loss": 2.5982, "step": 1320 }, { "epoch": 0.03, "learning_rate": 9.978715843339197e-05, "loss": 2.5614, "step": 1325 }, { "epoch": 0.03, "learning_rate": 9.978541865170658e-05, "loss": 2.5814, "step": 1330 }, { "epoch": 0.03, "learning_rate": 9.978367180367701e-05, "loss": 2.5529, "step": 1335 }, { "epoch": 0.03, "learning_rate": 9.978191788955116e-05, "loss": 2.5603, "step": 1340 }, { "epoch": 0.03, "learning_rate": 9.978015690957799e-05, "loss": 2.5798, "step": 1345 }, { "epoch": 0.03, "learning_rate": 9.977838886400746e-05, "loss": 2.6101, "step": 1350 }, { "epoch": 0.03, "learning_rate": 9.97766137530905e-05, "loss": 2.5786, "step": 1355 }, { "epoch": 0.03, "learning_rate": 9.977483157707904e-05, "loss": 2.6207, "step": 1360 }, { "epoch": 0.03, "learning_rate": 9.977304233622609e-05, "loss": 2.5522, "step": 1365 }, { "epoch": 0.03, "learning_rate": 9.977124603078557e-05, "loss": 2.5775, "step": 1370 }, { "epoch": 0.03, "learning_rate": 9.976944266101245e-05, "loss": 2.6102, "step": 1375 }, { "epoch": 0.03, "learning_rate": 9.97676322271627e-05, "loss": 2.5801, "step": 1380 }, { "epoch": 0.03, "learning_rate": 9.976581472949327e-05, "loss": 2.5727, "step": 1385 }, { "epoch": 0.03, "learning_rate": 9.976399016826215e-05, "loss": 2.5263, "step": 1390 }, { "epoch": 0.03, "learning_rate": 9.976215854372829e-05, "loss": 2.5483, "step": 1395 }, { "epoch": 0.03, "learning_rate": 9.976031985615168e-05, "loss": 2.5617, "step": 1400 }, { "epoch": 0.03, "learning_rate": 9.975847410579327e-05, "loss": 2.581, "step": 1405 }, { "epoch": 0.03, "learning_rate": 9.975662129291508e-05, "loss": 2.5747, "step": 1410 }, { "epoch": 0.03, "learning_rate": 9.975476141778004e-05, "loss": 2.5531, "step": 1415 }, { "epoch": 0.03, "learning_rate": 9.975289448065219e-05, "loss": 2.5511, "step": 1420 }, { "epoch": 0.03, "learning_rate": 9.975102048179646e-05, "loss": 2.4988, "step": 1425 }, { "epoch": 0.03, "learning_rate": 9.974913942147888e-05, "loss": 2.5742, "step": 1430 }, { "epoch": 0.03, "learning_rate": 9.974725129996642e-05, "loss": 2.5213, "step": 1435 }, { "epoch": 0.03, "learning_rate": 9.974535611752708e-05, "loss": 2.5933, "step": 1440 }, { "epoch": 0.03, "learning_rate": 9.974345387442984e-05, "loss": 2.5618, "step": 1445 }, { "epoch": 0.03, "learning_rate": 9.974154457094473e-05, "loss": 2.5496, "step": 1450 }, { "epoch": 0.03, "learning_rate": 9.973962820734271e-05, "loss": 2.61, "step": 1455 }, { "epoch": 0.03, "learning_rate": 9.973770478389582e-05, "loss": 2.5269, "step": 1460 }, { "epoch": 0.04, "learning_rate": 9.973577430087702e-05, "loss": 2.5932, "step": 1465 }, { "epoch": 0.04, "learning_rate": 9.973383675856034e-05, "loss": 2.6412, "step": 1470 }, { "epoch": 0.04, "learning_rate": 9.973189215722079e-05, "loss": 2.5745, "step": 1475 }, { "epoch": 0.04, "learning_rate": 9.972994049713436e-05, "loss": 2.5935, "step": 1480 }, { "epoch": 0.04, "learning_rate": 9.972798177857809e-05, "loss": 2.5454, "step": 1485 }, { "epoch": 0.04, "learning_rate": 9.972601600182996e-05, "loss": 2.569, "step": 1490 }, { "epoch": 0.04, "learning_rate": 9.972404316716901e-05, "loss": 2.5291, "step": 1495 }, { "epoch": 0.04, "learning_rate": 9.972206327487524e-05, "loss": 2.5638, "step": 1500 }, { "epoch": 0.04, "learning_rate": 9.972007632522968e-05, "loss": 2.5524, "step": 1505 }, { "epoch": 0.04, "learning_rate": 9.971808231851435e-05, "loss": 2.5264, "step": 1510 }, { "epoch": 0.04, "learning_rate": 9.971608125501226e-05, "loss": 2.5615, "step": 1515 }, { "epoch": 0.04, "learning_rate": 9.971407313500744e-05, "loss": 2.5334, "step": 1520 }, { "epoch": 0.04, "learning_rate": 9.97120579587849e-05, "loss": 2.579, "step": 1525 }, { "epoch": 0.04, "learning_rate": 9.97100357266307e-05, "loss": 2.589, "step": 1530 }, { "epoch": 0.04, "learning_rate": 9.970800643883185e-05, "loss": 2.5283, "step": 1535 }, { "epoch": 0.04, "learning_rate": 9.970597009567636e-05, "loss": 2.6074, "step": 1540 }, { "epoch": 0.04, "learning_rate": 9.970392669745328e-05, "loss": 2.6433, "step": 1545 }, { "epoch": 0.04, "learning_rate": 9.970187624445265e-05, "loss": 2.4676, "step": 1550 }, { "epoch": 0.04, "learning_rate": 9.969981873696548e-05, "loss": 2.5183, "step": 1555 }, { "epoch": 0.04, "learning_rate": 9.969775417528382e-05, "loss": 2.5522, "step": 1560 }, { "epoch": 0.04, "learning_rate": 9.96956825597007e-05, "loss": 2.562, "step": 1565 }, { "epoch": 0.04, "learning_rate": 9.969360389051017e-05, "loss": 2.5264, "step": 1570 }, { "epoch": 0.04, "learning_rate": 9.969151816800723e-05, "loss": 2.5789, "step": 1575 }, { "epoch": 0.04, "learning_rate": 9.968942539248797e-05, "loss": 2.5245, "step": 1580 }, { "epoch": 0.04, "learning_rate": 9.96873255642494e-05, "loss": 2.5515, "step": 1585 }, { "epoch": 0.04, "learning_rate": 9.968521868358957e-05, "loss": 2.5461, "step": 1590 }, { "epoch": 0.04, "learning_rate": 9.968310475080752e-05, "loss": 2.5155, "step": 1595 }, { "epoch": 0.04, "learning_rate": 9.968098376620327e-05, "loss": 2.6256, "step": 1600 }, { "epoch": 0.04, "learning_rate": 9.967885573007792e-05, "loss": 2.5862, "step": 1605 }, { "epoch": 0.04, "learning_rate": 9.967672064273346e-05, "loss": 2.6179, "step": 1610 }, { "epoch": 0.04, "learning_rate": 9.967457850447297e-05, "loss": 2.6037, "step": 1615 }, { "epoch": 0.04, "learning_rate": 9.967242931560048e-05, "loss": 2.5508, "step": 1620 }, { "epoch": 0.04, "learning_rate": 9.967027307642104e-05, "loss": 2.5971, "step": 1625 }, { "epoch": 0.04, "learning_rate": 9.966810978724071e-05, "loss": 2.5963, "step": 1630 }, { "epoch": 0.04, "learning_rate": 9.966593944836651e-05, "loss": 2.5135, "step": 1635 }, { "epoch": 0.04, "learning_rate": 9.966376206010652e-05, "loss": 2.5467, "step": 1640 }, { "epoch": 0.04, "learning_rate": 9.966157762276977e-05, "loss": 2.5656, "step": 1645 }, { "epoch": 0.04, "learning_rate": 9.965938613666632e-05, "loss": 2.6222, "step": 1650 }, { "epoch": 0.04, "learning_rate": 9.965718760210723e-05, "loss": 2.5066, "step": 1655 }, { "epoch": 0.04, "learning_rate": 9.965498201940452e-05, "loss": 2.603, "step": 1660 }, { "epoch": 0.04, "learning_rate": 9.965276938887127e-05, "loss": 2.5233, "step": 1665 }, { "epoch": 0.04, "learning_rate": 9.965054971082152e-05, "loss": 2.5745, "step": 1670 }, { "epoch": 0.04, "learning_rate": 9.964832298557034e-05, "loss": 2.5136, "step": 1675 }, { "epoch": 0.04, "learning_rate": 9.964608921343376e-05, "loss": 2.5222, "step": 1680 }, { "epoch": 0.04, "learning_rate": 9.964384839472884e-05, "loss": 2.5257, "step": 1685 }, { "epoch": 0.04, "learning_rate": 9.964160052977363e-05, "loss": 2.5535, "step": 1690 }, { "epoch": 0.04, "learning_rate": 9.963934561888719e-05, "loss": 2.579, "step": 1695 }, { "epoch": 0.04, "learning_rate": 9.963708366238958e-05, "loss": 2.5894, "step": 1700 }, { "epoch": 0.04, "learning_rate": 9.963481466060182e-05, "loss": 2.561, "step": 1705 }, { "epoch": 0.04, "learning_rate": 9.9632538613846e-05, "loss": 2.5584, "step": 1710 }, { "epoch": 0.04, "learning_rate": 9.963025552244516e-05, "loss": 2.5562, "step": 1715 }, { "epoch": 0.04, "learning_rate": 9.962796538672335e-05, "loss": 2.5656, "step": 1720 }, { "epoch": 0.04, "learning_rate": 9.962566820700563e-05, "loss": 2.5795, "step": 1725 }, { "epoch": 0.04, "learning_rate": 9.962336398361804e-05, "loss": 2.6144, "step": 1730 }, { "epoch": 0.04, "learning_rate": 9.962105271688765e-05, "loss": 2.5956, "step": 1735 }, { "epoch": 0.04, "learning_rate": 9.961873440714249e-05, "loss": 2.5782, "step": 1740 }, { "epoch": 0.04, "learning_rate": 9.961640905471161e-05, "loss": 2.5686, "step": 1745 }, { "epoch": 0.04, "learning_rate": 9.961407665992511e-05, "loss": 2.5667, "step": 1750 }, { "epoch": 0.04, "learning_rate": 9.961173722311398e-05, "loss": 2.5701, "step": 1755 }, { "epoch": 0.04, "learning_rate": 9.96093907446103e-05, "loss": 2.6271, "step": 1760 }, { "epoch": 0.04, "learning_rate": 9.960703722474711e-05, "loss": 2.544, "step": 1765 }, { "epoch": 0.04, "learning_rate": 9.960467666385848e-05, "loss": 2.5294, "step": 1770 }, { "epoch": 0.04, "learning_rate": 9.960230906227943e-05, "loss": 2.6166, "step": 1775 }, { "epoch": 0.04, "learning_rate": 9.959993442034602e-05, "loss": 2.5506, "step": 1780 }, { "epoch": 0.04, "learning_rate": 9.959755273839531e-05, "loss": 2.5482, "step": 1785 }, { "epoch": 0.04, "learning_rate": 9.959516401676534e-05, "loss": 2.5574, "step": 1790 }, { "epoch": 0.04, "learning_rate": 9.959276825579514e-05, "loss": 2.5272, "step": 1795 }, { "epoch": 0.04, "learning_rate": 9.959036545582477e-05, "loss": 2.5388, "step": 1800 }, { "epoch": 0.04, "learning_rate": 9.958795561719527e-05, "loss": 2.5753, "step": 1805 }, { "epoch": 0.04, "learning_rate": 9.958553874024868e-05, "loss": 2.5809, "step": 1810 }, { "epoch": 0.04, "learning_rate": 9.958311482532802e-05, "loss": 2.5696, "step": 1815 }, { "epoch": 0.04, "learning_rate": 9.95806838727774e-05, "loss": 2.6201, "step": 1820 }, { "epoch": 0.04, "learning_rate": 9.957824588294178e-05, "loss": 2.5422, "step": 1825 }, { "epoch": 0.04, "learning_rate": 9.957580085616724e-05, "loss": 2.5813, "step": 1830 }, { "epoch": 0.04, "learning_rate": 9.957334879280082e-05, "loss": 2.624, "step": 1835 }, { "epoch": 0.04, "learning_rate": 9.957088969319055e-05, "loss": 2.5277, "step": 1840 }, { "epoch": 0.04, "learning_rate": 9.956842355768546e-05, "loss": 2.5431, "step": 1845 }, { "epoch": 0.04, "learning_rate": 9.956595038663557e-05, "loss": 2.5461, "step": 1850 }, { "epoch": 0.04, "learning_rate": 9.956347018039194e-05, "loss": 2.5604, "step": 1855 }, { "epoch": 0.04, "learning_rate": 9.956098293930658e-05, "loss": 2.6015, "step": 1860 }, { "epoch": 0.04, "learning_rate": 9.955848866373254e-05, "loss": 2.5407, "step": 1865 }, { "epoch": 0.04, "learning_rate": 9.955598735402382e-05, "loss": 2.5133, "step": 1870 }, { "epoch": 0.04, "learning_rate": 9.955347901053546e-05, "loss": 2.5567, "step": 1875 }, { "epoch": 0.04, "learning_rate": 9.955096363362349e-05, "loss": 2.5975, "step": 1880 }, { "epoch": 0.05, "learning_rate": 9.954844122364493e-05, "loss": 2.5245, "step": 1885 }, { "epoch": 0.05, "learning_rate": 9.95459117809578e-05, "loss": 2.5869, "step": 1890 }, { "epoch": 0.05, "learning_rate": 9.95433753059211e-05, "loss": 2.5571, "step": 1895 }, { "epoch": 0.05, "learning_rate": 9.954083179889488e-05, "loss": 2.5161, "step": 1900 }, { "epoch": 0.05, "learning_rate": 9.953828126024014e-05, "loss": 2.518, "step": 1905 }, { "epoch": 0.05, "learning_rate": 9.95357236903189e-05, "loss": 2.5158, "step": 1910 }, { "epoch": 0.05, "learning_rate": 9.953315908949414e-05, "loss": 2.5396, "step": 1915 }, { "epoch": 0.05, "learning_rate": 9.953058745812992e-05, "loss": 2.5143, "step": 1920 }, { "epoch": 0.05, "learning_rate": 9.952800879659118e-05, "loss": 2.6017, "step": 1925 }, { "epoch": 0.05, "learning_rate": 9.9525423105244e-05, "loss": 2.5373, "step": 1930 }, { "epoch": 0.05, "learning_rate": 9.952283038445534e-05, "loss": 2.5509, "step": 1935 }, { "epoch": 0.05, "learning_rate": 9.952023063459322e-05, "loss": 2.5445, "step": 1940 }, { "epoch": 0.05, "learning_rate": 9.951762385602661e-05, "loss": 2.5344, "step": 1945 }, { "epoch": 0.05, "learning_rate": 9.951501004912554e-05, "loss": 2.5706, "step": 1950 }, { "epoch": 0.05, "learning_rate": 9.951238921426096e-05, "loss": 2.5558, "step": 1955 }, { "epoch": 0.05, "learning_rate": 9.95097613518049e-05, "loss": 2.5378, "step": 1960 }, { "epoch": 0.05, "learning_rate": 9.950712646213034e-05, "loss": 2.5659, "step": 1965 }, { "epoch": 0.05, "learning_rate": 9.950448454561125e-05, "loss": 2.6083, "step": 1970 }, { "epoch": 0.05, "learning_rate": 9.950183560262263e-05, "loss": 2.5883, "step": 1975 }, { "epoch": 0.05, "learning_rate": 9.949917963354045e-05, "loss": 2.567, "step": 1980 }, { "epoch": 0.05, "learning_rate": 9.949651663874171e-05, "loss": 2.55, "step": 1985 }, { "epoch": 0.05, "learning_rate": 9.949384661860434e-05, "loss": 2.4813, "step": 1990 }, { "epoch": 0.05, "learning_rate": 9.949116957350736e-05, "loss": 2.5215, "step": 1995 }, { "epoch": 0.05, "learning_rate": 9.948848550383072e-05, "loss": 2.6191, "step": 2000 }, { "epoch": 0.05, "learning_rate": 9.948579440995537e-05, "loss": 2.5221, "step": 2005 }, { "epoch": 0.05, "learning_rate": 9.948309629226329e-05, "loss": 2.5704, "step": 2010 }, { "epoch": 0.05, "learning_rate": 9.948039115113743e-05, "loss": 2.583, "step": 2015 }, { "epoch": 0.05, "learning_rate": 9.947767898696176e-05, "loss": 2.5796, "step": 2020 }, { "epoch": 0.05, "learning_rate": 9.947495980012125e-05, "loss": 2.5651, "step": 2025 }, { "epoch": 0.05, "learning_rate": 9.94722335910018e-05, "loss": 2.5388, "step": 2030 }, { "epoch": 0.05, "learning_rate": 9.946950035999041e-05, "loss": 2.6016, "step": 2035 }, { "epoch": 0.05, "learning_rate": 9.946676010747498e-05, "loss": 2.6336, "step": 2040 }, { "epoch": 0.05, "learning_rate": 9.946401283384448e-05, "loss": 2.5699, "step": 2045 }, { "epoch": 0.05, "learning_rate": 9.946125853948884e-05, "loss": 2.5089, "step": 2050 }, { "epoch": 0.05, "learning_rate": 9.945849722479898e-05, "loss": 2.5343, "step": 2055 }, { "epoch": 0.05, "learning_rate": 9.945572889016685e-05, "loss": 2.5482, "step": 2060 }, { "epoch": 0.05, "learning_rate": 9.945295353598536e-05, "loss": 2.5884, "step": 2065 }, { "epoch": 0.05, "learning_rate": 9.945017116264846e-05, "loss": 2.5567, "step": 2070 }, { "epoch": 0.05, "learning_rate": 9.944738177055103e-05, "loss": 2.5686, "step": 2075 }, { "epoch": 0.05, "learning_rate": 9.9444585360089e-05, "loss": 2.6003, "step": 2080 }, { "epoch": 0.05, "learning_rate": 9.94417819316593e-05, "loss": 2.5659, "step": 2085 }, { "epoch": 0.05, "learning_rate": 9.943897148565981e-05, "loss": 2.5989, "step": 2090 }, { "epoch": 0.05, "learning_rate": 9.943615402248947e-05, "loss": 2.5121, "step": 2095 }, { "epoch": 0.05, "learning_rate": 9.943332954254814e-05, "loss": 2.527, "step": 2100 }, { "epoch": 0.05, "learning_rate": 9.943049804623674e-05, "loss": 2.5014, "step": 2105 }, { "epoch": 0.05, "learning_rate": 9.942765953395715e-05, "loss": 2.5588, "step": 2110 }, { "epoch": 0.05, "learning_rate": 9.942481400611226e-05, "loss": 2.6034, "step": 2115 }, { "epoch": 0.05, "learning_rate": 9.942196146310594e-05, "loss": 2.5535, "step": 2120 }, { "epoch": 0.05, "learning_rate": 9.94191019053431e-05, "loss": 2.5353, "step": 2125 }, { "epoch": 0.05, "learning_rate": 9.941623533322959e-05, "loss": 2.5251, "step": 2130 }, { "epoch": 0.05, "learning_rate": 9.941336174717229e-05, "loss": 2.5626, "step": 2135 }, { "epoch": 0.05, "learning_rate": 9.941048114757905e-05, "loss": 2.5925, "step": 2140 }, { "epoch": 0.05, "learning_rate": 9.940759353485876e-05, "loss": 2.5853, "step": 2145 }, { "epoch": 0.05, "learning_rate": 9.940469890942124e-05, "loss": 2.4877, "step": 2150 }, { "epoch": 0.05, "learning_rate": 9.940179727167737e-05, "loss": 2.5892, "step": 2155 }, { "epoch": 0.05, "learning_rate": 9.939888862203899e-05, "loss": 2.5564, "step": 2160 }, { "epoch": 0.05, "learning_rate": 9.939597296091892e-05, "loss": 2.5278, "step": 2165 }, { "epoch": 0.05, "learning_rate": 9.939305028873104e-05, "loss": 2.5704, "step": 2170 }, { "epoch": 0.05, "learning_rate": 9.939012060589015e-05, "loss": 2.5498, "step": 2175 }, { "epoch": 0.05, "learning_rate": 9.938718391281208e-05, "loss": 2.5453, "step": 2180 }, { "epoch": 0.05, "learning_rate": 9.938424020991366e-05, "loss": 2.6067, "step": 2185 }, { "epoch": 0.05, "learning_rate": 9.938128949761272e-05, "loss": 2.5097, "step": 2190 }, { "epoch": 0.05, "learning_rate": 9.937833177632805e-05, "loss": 2.5426, "step": 2195 }, { "epoch": 0.05, "learning_rate": 9.937536704647946e-05, "loss": 2.5383, "step": 2200 }, { "epoch": 0.05, "learning_rate": 9.937239530848776e-05, "loss": 2.5429, "step": 2205 }, { "epoch": 0.05, "learning_rate": 9.936941656277477e-05, "loss": 2.5045, "step": 2210 }, { "epoch": 0.05, "learning_rate": 9.936643080976323e-05, "loss": 2.534, "step": 2215 }, { "epoch": 0.05, "learning_rate": 9.936343804987695e-05, "loss": 2.5326, "step": 2220 }, { "epoch": 0.05, "learning_rate": 9.936043828354071e-05, "loss": 2.534, "step": 2225 }, { "epoch": 0.05, "learning_rate": 9.93574315111803e-05, "loss": 2.5683, "step": 2230 }, { "epoch": 0.05, "learning_rate": 9.935441773322247e-05, "loss": 2.5915, "step": 2235 }, { "epoch": 0.05, "learning_rate": 9.935139695009499e-05, "loss": 2.5619, "step": 2240 }, { "epoch": 0.05, "learning_rate": 9.934836916222662e-05, "loss": 2.5331, "step": 2245 }, { "epoch": 0.05, "learning_rate": 9.934533437004712e-05, "loss": 2.5151, "step": 2250 }, { "epoch": 0.05, "learning_rate": 9.934229257398721e-05, "loss": 2.5626, "step": 2255 }, { "epoch": 0.05, "learning_rate": 9.933924377447865e-05, "loss": 2.4831, "step": 2260 }, { "epoch": 0.05, "learning_rate": 9.933618797195419e-05, "loss": 2.5549, "step": 2265 }, { "epoch": 0.05, "learning_rate": 9.933312516684751e-05, "loss": 2.5117, "step": 2270 }, { "epoch": 0.05, "learning_rate": 9.933005535959339e-05, "loss": 2.589, "step": 2275 }, { "epoch": 0.05, "learning_rate": 9.932697855062751e-05, "loss": 2.6079, "step": 2280 }, { "epoch": 0.05, "learning_rate": 9.932389474038657e-05, "loss": 2.5842, "step": 2285 }, { "epoch": 0.05, "learning_rate": 9.932080392930831e-05, "loss": 2.5654, "step": 2290 }, { "epoch": 0.05, "learning_rate": 9.931770611783141e-05, "loss": 2.5667, "step": 2295 }, { "epoch": 0.06, "learning_rate": 9.931460130639556e-05, "loss": 2.5879, "step": 2300 }, { "epoch": 0.06, "learning_rate": 9.931148949544144e-05, "loss": 2.5875, "step": 2305 }, { "epoch": 0.06, "learning_rate": 9.930837068541073e-05, "loss": 2.5725, "step": 2310 }, { "epoch": 0.06, "learning_rate": 9.930524487674612e-05, "loss": 2.5529, "step": 2315 }, { "epoch": 0.06, "learning_rate": 9.930211206989124e-05, "loss": 2.5592, "step": 2320 }, { "epoch": 0.06, "learning_rate": 9.929897226529079e-05, "loss": 2.5506, "step": 2325 }, { "epoch": 0.06, "learning_rate": 9.929582546339038e-05, "loss": 2.5743, "step": 2330 }, { "epoch": 0.06, "learning_rate": 9.929267166463668e-05, "loss": 2.5075, "step": 2335 }, { "epoch": 0.06, "learning_rate": 9.928951086947733e-05, "loss": 2.5591, "step": 2340 }, { "epoch": 0.06, "learning_rate": 9.928634307836094e-05, "loss": 2.4979, "step": 2345 }, { "epoch": 0.06, "learning_rate": 9.928316829173715e-05, "loss": 2.5173, "step": 2350 }, { "epoch": 0.06, "learning_rate": 9.927998651005657e-05, "loss": 2.5102, "step": 2355 }, { "epoch": 0.06, "learning_rate": 9.927679773377082e-05, "loss": 2.5109, "step": 2360 }, { "epoch": 0.06, "learning_rate": 9.927360196333249e-05, "loss": 2.5413, "step": 2365 }, { "epoch": 0.06, "learning_rate": 9.927039919919517e-05, "loss": 2.494, "step": 2370 }, { "epoch": 0.06, "learning_rate": 9.926718944181347e-05, "loss": 2.5882, "step": 2375 }, { "epoch": 0.06, "learning_rate": 9.926397269164294e-05, "loss": 2.5441, "step": 2380 }, { "epoch": 0.06, "learning_rate": 9.92607489491402e-05, "loss": 2.5574, "step": 2385 }, { "epoch": 0.06, "learning_rate": 9.925751821476274e-05, "loss": 2.6071, "step": 2390 }, { "epoch": 0.06, "learning_rate": 9.92542804889692e-05, "loss": 2.574, "step": 2395 }, { "epoch": 0.06, "learning_rate": 9.925103577221908e-05, "loss": 2.5843, "step": 2400 }, { "epoch": 0.06, "learning_rate": 9.924778406497292e-05, "loss": 2.5534, "step": 2405 }, { "epoch": 0.06, "learning_rate": 9.924452536769229e-05, "loss": 2.5405, "step": 2410 }, { "epoch": 0.06, "learning_rate": 9.924125968083968e-05, "loss": 2.5305, "step": 2415 }, { "epoch": 0.06, "learning_rate": 9.923798700487862e-05, "loss": 2.5296, "step": 2420 }, { "epoch": 0.06, "learning_rate": 9.923470734027365e-05, "loss": 2.5519, "step": 2425 }, { "epoch": 0.06, "learning_rate": 9.923142068749023e-05, "loss": 2.5034, "step": 2430 }, { "epoch": 0.06, "learning_rate": 9.922812704699488e-05, "loss": 2.5222, "step": 2435 }, { "epoch": 0.06, "learning_rate": 9.922482641925507e-05, "loss": 2.5919, "step": 2440 }, { "epoch": 0.06, "learning_rate": 9.922151880473929e-05, "loss": 2.5527, "step": 2445 }, { "epoch": 0.06, "learning_rate": 9.921820420391698e-05, "loss": 2.557, "step": 2450 }, { "epoch": 0.06, "learning_rate": 9.921488261725866e-05, "loss": 2.486, "step": 2455 }, { "epoch": 0.06, "learning_rate": 9.921155404523574e-05, "loss": 2.5604, "step": 2460 }, { "epoch": 0.06, "learning_rate": 9.920821848832067e-05, "loss": 2.5583, "step": 2465 }, { "epoch": 0.06, "learning_rate": 9.920487594698688e-05, "loss": 2.5175, "step": 2470 }, { "epoch": 0.06, "learning_rate": 9.920152642170882e-05, "loss": 2.5927, "step": 2475 }, { "epoch": 0.06, "learning_rate": 9.91981699129619e-05, "loss": 2.5592, "step": 2480 }, { "epoch": 0.06, "learning_rate": 9.919480642122251e-05, "loss": 2.5851, "step": 2485 }, { "epoch": 0.06, "learning_rate": 9.919143594696807e-05, "loss": 2.4942, "step": 2490 }, { "epoch": 0.06, "learning_rate": 9.918805849067696e-05, "loss": 2.5237, "step": 2495 }, { "epoch": 0.06, "learning_rate": 9.918467405282857e-05, "loss": 2.5529, "step": 2500 }, { "epoch": 0.06, "learning_rate": 9.918128263390329e-05, "loss": 2.5199, "step": 2505 }, { "epoch": 0.06, "learning_rate": 9.917788423438244e-05, "loss": 2.4838, "step": 2510 }, { "epoch": 0.06, "learning_rate": 9.917447885474843e-05, "loss": 2.5553, "step": 2515 }, { "epoch": 0.06, "learning_rate": 9.917106649548455e-05, "loss": 2.5306, "step": 2520 }, { "epoch": 0.06, "learning_rate": 9.916764715707518e-05, "loss": 2.4718, "step": 2525 }, { "epoch": 0.06, "learning_rate": 9.916422084000564e-05, "loss": 2.5566, "step": 2530 }, { "epoch": 0.06, "learning_rate": 9.916078754476224e-05, "loss": 2.4997, "step": 2535 }, { "epoch": 0.06, "learning_rate": 9.915734727183226e-05, "loss": 2.5451, "step": 2540 }, { "epoch": 0.06, "learning_rate": 9.915390002170406e-05, "loss": 2.5514, "step": 2545 }, { "epoch": 0.06, "learning_rate": 9.915044579486688e-05, "loss": 2.5188, "step": 2550 }, { "epoch": 0.06, "learning_rate": 9.914698459181102e-05, "loss": 2.5706, "step": 2555 }, { "epoch": 0.06, "learning_rate": 9.914351641302775e-05, "loss": 2.539, "step": 2560 }, { "epoch": 0.06, "learning_rate": 9.914004125900931e-05, "loss": 2.6325, "step": 2565 }, { "epoch": 0.06, "learning_rate": 9.913655913024897e-05, "loss": 2.5591, "step": 2570 }, { "epoch": 0.06, "learning_rate": 9.913307002724097e-05, "loss": 2.5796, "step": 2575 }, { "epoch": 0.06, "learning_rate": 9.912957395048053e-05, "loss": 2.5619, "step": 2580 }, { "epoch": 0.06, "learning_rate": 9.912607090046386e-05, "loss": 2.5865, "step": 2585 }, { "epoch": 0.06, "learning_rate": 9.91225608776882e-05, "loss": 2.5812, "step": 2590 }, { "epoch": 0.06, "learning_rate": 9.911904388265171e-05, "loss": 2.5275, "step": 2595 }, { "epoch": 0.06, "learning_rate": 9.911551991585362e-05, "loss": 2.5455, "step": 2600 }, { "epoch": 0.06, "learning_rate": 9.911198897779407e-05, "loss": 2.55, "step": 2605 }, { "epoch": 0.06, "learning_rate": 9.910845106897425e-05, "loss": 2.541, "step": 2610 }, { "epoch": 0.06, "learning_rate": 9.91049061898963e-05, "loss": 2.573, "step": 2615 }, { "epoch": 0.06, "learning_rate": 9.910135434106338e-05, "loss": 2.5357, "step": 2620 }, { "epoch": 0.06, "learning_rate": 9.909779552297963e-05, "loss": 2.5604, "step": 2625 }, { "epoch": 0.06, "learning_rate": 9.909422973615016e-05, "loss": 2.4776, "step": 2630 }, { "epoch": 0.06, "learning_rate": 9.909065698108108e-05, "loss": 2.6125, "step": 2635 }, { "epoch": 0.06, "learning_rate": 9.908707725827952e-05, "loss": 2.5189, "step": 2640 }, { "epoch": 0.06, "learning_rate": 9.908349056825354e-05, "loss": 2.5451, "step": 2645 }, { "epoch": 0.06, "learning_rate": 9.907989691151223e-05, "loss": 2.5833, "step": 2650 }, { "epoch": 0.06, "learning_rate": 9.907629628856567e-05, "loss": 2.5599, "step": 2655 }, { "epoch": 0.06, "learning_rate": 9.907268869992492e-05, "loss": 2.4412, "step": 2660 }, { "epoch": 0.06, "learning_rate": 9.9069074146102e-05, "loss": 2.5885, "step": 2665 }, { "epoch": 0.06, "learning_rate": 9.906545262760998e-05, "loss": 2.4851, "step": 2670 }, { "epoch": 0.06, "learning_rate": 9.906182414496285e-05, "loss": 2.5352, "step": 2675 }, { "epoch": 0.06, "learning_rate": 9.905818869867565e-05, "loss": 2.489, "step": 2680 }, { "epoch": 0.06, "learning_rate": 9.905454628926436e-05, "loss": 2.5962, "step": 2685 }, { "epoch": 0.06, "learning_rate": 9.905089691724601e-05, "loss": 2.5835, "step": 2690 }, { "epoch": 0.06, "learning_rate": 9.904724058313851e-05, "loss": 2.5475, "step": 2695 }, { "epoch": 0.06, "learning_rate": 9.904357728746087e-05, "loss": 2.4859, "step": 2700 }, { "epoch": 0.06, "learning_rate": 9.903990703073304e-05, "loss": 2.5123, "step": 2705 }, { "epoch": 0.06, "learning_rate": 9.903622981347598e-05, "loss": 2.5023, "step": 2710 }, { "epoch": 0.06, "learning_rate": 9.903254563621158e-05, "loss": 2.5123, "step": 2715 }, { "epoch": 0.07, "learning_rate": 9.902885449946278e-05, "loss": 2.5608, "step": 2720 }, { "epoch": 0.07, "learning_rate": 9.902515640375345e-05, "loss": 2.5546, "step": 2725 }, { "epoch": 0.07, "learning_rate": 9.902145134960854e-05, "loss": 2.5552, "step": 2730 }, { "epoch": 0.07, "learning_rate": 9.901773933755391e-05, "loss": 2.5414, "step": 2735 }, { "epoch": 0.07, "learning_rate": 9.901402036811641e-05, "loss": 2.5138, "step": 2740 }, { "epoch": 0.07, "learning_rate": 9.90102944418239e-05, "loss": 2.5068, "step": 2745 }, { "epoch": 0.07, "learning_rate": 9.900656155920524e-05, "loss": 2.6296, "step": 2750 }, { "epoch": 0.07, "learning_rate": 9.900282172079026e-05, "loss": 2.5484, "step": 2755 }, { "epoch": 0.07, "learning_rate": 9.899907492710976e-05, "loss": 2.4755, "step": 2760 }, { "epoch": 0.07, "learning_rate": 9.899532117869556e-05, "loss": 2.5904, "step": 2765 }, { "epoch": 0.07, "learning_rate": 9.899156047608044e-05, "loss": 2.5496, "step": 2770 }, { "epoch": 0.07, "learning_rate": 9.898779281979818e-05, "loss": 2.5108, "step": 2775 }, { "epoch": 0.07, "learning_rate": 9.898401821038356e-05, "loss": 2.4951, "step": 2780 }, { "epoch": 0.07, "learning_rate": 9.898023664837232e-05, "loss": 2.4762, "step": 2785 }, { "epoch": 0.07, "learning_rate": 9.89764481343012e-05, "loss": 2.4661, "step": 2790 }, { "epoch": 0.07, "learning_rate": 9.897265266870794e-05, "loss": 2.4766, "step": 2795 }, { "epoch": 0.07, "learning_rate": 9.896885025213124e-05, "loss": 2.5487, "step": 2800 }, { "epoch": 0.07, "learning_rate": 9.89650408851108e-05, "loss": 2.5809, "step": 2805 }, { "epoch": 0.07, "learning_rate": 9.896122456818732e-05, "loss": 2.5066, "step": 2810 }, { "epoch": 0.07, "learning_rate": 9.895740130190246e-05, "loss": 2.5821, "step": 2815 }, { "epoch": 0.07, "learning_rate": 9.895357108679888e-05, "loss": 2.5071, "step": 2820 }, { "epoch": 0.07, "learning_rate": 9.894973392342022e-05, "loss": 2.4901, "step": 2825 }, { "epoch": 0.07, "learning_rate": 9.894588981231114e-05, "loss": 2.5222, "step": 2830 }, { "epoch": 0.07, "learning_rate": 9.894203875401722e-05, "loss": 2.4675, "step": 2835 }, { "epoch": 0.07, "learning_rate": 9.893818074908509e-05, "loss": 2.4734, "step": 2840 }, { "epoch": 0.07, "learning_rate": 9.893431579806233e-05, "loss": 2.5104, "step": 2845 }, { "epoch": 0.07, "learning_rate": 9.89304439014975e-05, "loss": 2.5244, "step": 2850 }, { "epoch": 0.07, "learning_rate": 9.89265650599402e-05, "loss": 2.567, "step": 2855 }, { "epoch": 0.07, "learning_rate": 9.892267927394095e-05, "loss": 2.5148, "step": 2860 }, { "epoch": 0.07, "learning_rate": 9.891878654405129e-05, "loss": 2.5549, "step": 2865 }, { "epoch": 0.07, "learning_rate": 9.891488687082374e-05, "loss": 2.5529, "step": 2870 }, { "epoch": 0.07, "learning_rate": 9.89109802548118e-05, "loss": 2.5439, "step": 2875 }, { "epoch": 0.07, "learning_rate": 9.890706669656995e-05, "loss": 2.4827, "step": 2880 }, { "epoch": 0.07, "learning_rate": 9.890314619665369e-05, "loss": 2.5411, "step": 2885 }, { "epoch": 0.07, "learning_rate": 9.889921875561946e-05, "loss": 2.5491, "step": 2890 }, { "epoch": 0.07, "learning_rate": 9.889528437402472e-05, "loss": 2.4911, "step": 2895 }, { "epoch": 0.07, "learning_rate": 9.889134305242788e-05, "loss": 2.5705, "step": 2900 }, { "epoch": 0.07, "learning_rate": 9.888739479138837e-05, "loss": 2.5408, "step": 2905 }, { "epoch": 0.07, "learning_rate": 9.88834395914666e-05, "loss": 2.4953, "step": 2910 }, { "epoch": 0.07, "learning_rate": 9.887947745322393e-05, "loss": 2.6056, "step": 2915 }, { "epoch": 0.07, "learning_rate": 9.887550837722274e-05, "loss": 2.5368, "step": 2920 }, { "epoch": 0.07, "learning_rate": 9.887153236402639e-05, "loss": 2.5624, "step": 2925 }, { "epoch": 0.07, "learning_rate": 9.886754941419924e-05, "loss": 2.5165, "step": 2930 }, { "epoch": 0.07, "learning_rate": 9.886355952830657e-05, "loss": 2.5558, "step": 2935 }, { "epoch": 0.07, "learning_rate": 9.885956270691472e-05, "loss": 2.522, "step": 2940 }, { "epoch": 0.07, "learning_rate": 9.885555895059096e-05, "loss": 2.5535, "step": 2945 }, { "epoch": 0.07, "learning_rate": 9.885154825990359e-05, "loss": 2.5465, "step": 2950 }, { "epoch": 0.07, "learning_rate": 9.884753063542187e-05, "loss": 2.5765, "step": 2955 }, { "epoch": 0.07, "learning_rate": 9.884350607771602e-05, "loss": 2.5548, "step": 2960 }, { "epoch": 0.07, "learning_rate": 9.88394745873573e-05, "loss": 2.5657, "step": 2965 }, { "epoch": 0.07, "learning_rate": 9.88354361649179e-05, "loss": 2.4385, "step": 2970 }, { "epoch": 0.07, "learning_rate": 9.883139081097103e-05, "loss": 2.612, "step": 2975 }, { "epoch": 0.07, "learning_rate": 9.88273385260909e-05, "loss": 2.5342, "step": 2980 }, { "epoch": 0.07, "learning_rate": 9.88232793108526e-05, "loss": 2.4992, "step": 2985 }, { "epoch": 0.07, "learning_rate": 9.881921316583236e-05, "loss": 2.4798, "step": 2990 }, { "epoch": 0.07, "learning_rate": 9.881514009160725e-05, "loss": 2.5829, "step": 2995 }, { "epoch": 0.07, "learning_rate": 9.881106008875542e-05, "loss": 2.5601, "step": 3000 }, { "epoch": 0.07, "learning_rate": 9.880697315785599e-05, "loss": 2.5164, "step": 3005 }, { "epoch": 0.07, "learning_rate": 9.880287929948898e-05, "loss": 2.5144, "step": 3010 }, { "epoch": 0.07, "learning_rate": 9.87987785142355e-05, "loss": 2.547, "step": 3015 }, { "epoch": 0.07, "learning_rate": 9.87946708026776e-05, "loss": 2.5448, "step": 3020 }, { "epoch": 0.07, "learning_rate": 9.879055616539829e-05, "loss": 2.554, "step": 3025 }, { "epoch": 0.07, "learning_rate": 9.87864346029816e-05, "loss": 2.4814, "step": 3030 }, { "epoch": 0.07, "learning_rate": 9.878230611601252e-05, "loss": 2.4792, "step": 3035 }, { "epoch": 0.07, "learning_rate": 9.877817070507705e-05, "loss": 2.5302, "step": 3040 }, { "epoch": 0.07, "learning_rate": 9.877402837076213e-05, "loss": 2.4758, "step": 3045 }, { "epoch": 0.07, "learning_rate": 9.876987911365572e-05, "loss": 2.4872, "step": 3050 }, { "epoch": 0.07, "learning_rate": 9.876572293434674e-05, "loss": 2.4894, "step": 3055 }, { "epoch": 0.07, "learning_rate": 9.876155983342513e-05, "loss": 2.5656, "step": 3060 }, { "epoch": 0.07, "learning_rate": 9.875738981148172e-05, "loss": 2.5523, "step": 3065 }, { "epoch": 0.07, "learning_rate": 9.875321286910846e-05, "loss": 2.4833, "step": 3070 }, { "epoch": 0.07, "learning_rate": 9.874902900689816e-05, "loss": 2.5185, "step": 3075 }, { "epoch": 0.07, "learning_rate": 9.874483822544469e-05, "loss": 2.5708, "step": 3080 }, { "epoch": 0.07, "learning_rate": 9.874064052534286e-05, "loss": 2.4973, "step": 3085 }, { "epoch": 0.07, "learning_rate": 9.873643590718846e-05, "loss": 2.4943, "step": 3090 }, { "epoch": 0.07, "learning_rate": 9.87322243715783e-05, "loss": 2.5955, "step": 3095 }, { "epoch": 0.07, "learning_rate": 9.872800591911014e-05, "loss": 2.5493, "step": 3100 }, { "epoch": 0.07, "learning_rate": 9.872378055038272e-05, "loss": 2.5365, "step": 3105 }, { "epoch": 0.07, "learning_rate": 9.87195482659958e-05, "loss": 2.5177, "step": 3110 }, { "epoch": 0.07, "learning_rate": 9.871530906655007e-05, "loss": 2.4962, "step": 3115 }, { "epoch": 0.07, "learning_rate": 9.871106295264723e-05, "loss": 2.5156, "step": 3120 }, { "epoch": 0.07, "learning_rate": 9.870680992488995e-05, "loss": 2.4814, "step": 3125 }, { "epoch": 0.07, "learning_rate": 9.87025499838819e-05, "loss": 2.4993, "step": 3130 }, { "epoch": 0.08, "learning_rate": 9.869828313022771e-05, "loss": 2.5813, "step": 3135 }, { "epoch": 0.08, "learning_rate": 9.869400936453301e-05, "loss": 2.5292, "step": 3140 }, { "epoch": 0.08, "learning_rate": 9.86897286874044e-05, "loss": 2.5746, "step": 3145 }, { "epoch": 0.08, "learning_rate": 9.868544109944945e-05, "loss": 2.5499, "step": 3150 }, { "epoch": 0.08, "learning_rate": 9.868114660127674e-05, "loss": 2.6051, "step": 3155 }, { "epoch": 0.08, "learning_rate": 9.86768451934958e-05, "loss": 2.4783, "step": 3160 }, { "epoch": 0.08, "learning_rate": 9.867253687671716e-05, "loss": 2.5654, "step": 3165 }, { "epoch": 0.08, "learning_rate": 9.866822165155233e-05, "loss": 2.5256, "step": 3170 }, { "epoch": 0.08, "learning_rate": 9.866389951861378e-05, "loss": 2.4993, "step": 3175 }, { "epoch": 0.08, "learning_rate": 9.865957047851499e-05, "loss": 2.4265, "step": 3180 }, { "epoch": 0.08, "learning_rate": 9.865523453187042e-05, "loss": 2.4994, "step": 3185 }, { "epoch": 0.08, "learning_rate": 9.865089167929546e-05, "loss": 2.5449, "step": 3190 }, { "epoch": 0.08, "learning_rate": 9.864654192140657e-05, "loss": 2.5337, "step": 3195 }, { "epoch": 0.08, "learning_rate": 9.864218525882107e-05, "loss": 2.5054, "step": 3200 }, { "epoch": 0.08, "learning_rate": 9.863782169215739e-05, "loss": 2.4538, "step": 3205 }, { "epoch": 0.08, "learning_rate": 9.863345122203484e-05, "loss": 2.5337, "step": 3210 }, { "epoch": 0.08, "learning_rate": 9.862907384907375e-05, "loss": 2.5397, "step": 3215 }, { "epoch": 0.08, "learning_rate": 9.862468957389543e-05, "loss": 2.5614, "step": 3220 }, { "epoch": 0.08, "learning_rate": 9.862029839712218e-05, "loss": 2.5792, "step": 3225 }, { "epoch": 0.08, "learning_rate": 9.861590031937727e-05, "loss": 2.5527, "step": 3230 }, { "epoch": 0.08, "learning_rate": 9.861149534128493e-05, "loss": 2.5374, "step": 3235 }, { "epoch": 0.08, "learning_rate": 9.860708346347037e-05, "loss": 2.5292, "step": 3240 }, { "epoch": 0.08, "learning_rate": 9.860266468655984e-05, "loss": 2.5056, "step": 3245 }, { "epoch": 0.08, "learning_rate": 9.859823901118046e-05, "loss": 2.6172, "step": 3250 }, { "epoch": 0.08, "learning_rate": 9.859380643796046e-05, "loss": 2.5362, "step": 3255 }, { "epoch": 0.08, "learning_rate": 9.858936696752894e-05, "loss": 2.5063, "step": 3260 }, { "epoch": 0.08, "learning_rate": 9.858492060051603e-05, "loss": 2.5464, "step": 3265 }, { "epoch": 0.08, "learning_rate": 9.858046733755282e-05, "loss": 2.4569, "step": 3270 }, { "epoch": 0.08, "learning_rate": 9.857600717927142e-05, "loss": 2.5114, "step": 3275 }, { "epoch": 0.08, "learning_rate": 9.857154012630485e-05, "loss": 2.5069, "step": 3280 }, { "epoch": 0.08, "learning_rate": 9.856706617928717e-05, "loss": 2.4202, "step": 3285 }, { "epoch": 0.08, "learning_rate": 9.856258533885339e-05, "loss": 2.5441, "step": 3290 }, { "epoch": 0.08, "learning_rate": 9.855809760563949e-05, "loss": 2.5346, "step": 3295 }, { "epoch": 0.08, "learning_rate": 9.855360298028245e-05, "loss": 2.506, "step": 3300 }, { "epoch": 0.08, "learning_rate": 9.854910146342022e-05, "loss": 2.4795, "step": 3305 }, { "epoch": 0.08, "learning_rate": 9.854459305569174e-05, "loss": 2.565, "step": 3310 }, { "epoch": 0.08, "learning_rate": 9.854007775773688e-05, "loss": 2.5237, "step": 3315 }, { "epoch": 0.08, "learning_rate": 9.853555557019656e-05, "loss": 2.5053, "step": 3320 }, { "epoch": 0.08, "learning_rate": 9.853102649371263e-05, "loss": 2.5045, "step": 3325 }, { "epoch": 0.08, "learning_rate": 9.852649052892793e-05, "loss": 2.46, "step": 3330 }, { "epoch": 0.08, "learning_rate": 9.852194767648626e-05, "loss": 2.4999, "step": 3335 }, { "epoch": 0.08, "learning_rate": 9.851739793703243e-05, "loss": 2.5059, "step": 3340 }, { "epoch": 0.08, "learning_rate": 9.85128413112122e-05, "loss": 2.5418, "step": 3345 }, { "epoch": 0.08, "learning_rate": 9.850827779967234e-05, "loss": 2.4936, "step": 3350 }, { "epoch": 0.08, "learning_rate": 9.850370740306057e-05, "loss": 2.5137, "step": 3355 }, { "epoch": 0.08, "learning_rate": 9.849913012202559e-05, "loss": 2.5352, "step": 3360 }, { "epoch": 0.08, "learning_rate": 9.849454595721707e-05, "loss": 2.4873, "step": 3365 }, { "epoch": 0.08, "learning_rate": 9.848995490928568e-05, "loss": 2.5539, "step": 3370 }, { "epoch": 0.08, "learning_rate": 9.848535697888305e-05, "loss": 2.5508, "step": 3375 }, { "epoch": 0.08, "learning_rate": 9.848075216666179e-05, "loss": 2.5277, "step": 3380 }, { "epoch": 0.08, "learning_rate": 9.847614047327552e-05, "loss": 2.5733, "step": 3385 }, { "epoch": 0.08, "learning_rate": 9.847152189937876e-05, "loss": 2.5196, "step": 3390 }, { "epoch": 0.08, "learning_rate": 9.846689644562708e-05, "loss": 2.5582, "step": 3395 }, { "epoch": 0.08, "learning_rate": 9.846226411267697e-05, "loss": 2.4872, "step": 3400 }, { "epoch": 0.08, "learning_rate": 9.845762490118597e-05, "loss": 2.4912, "step": 3405 }, { "epoch": 0.08, "learning_rate": 9.84529788118125e-05, "loss": 2.544, "step": 3410 }, { "epoch": 0.08, "learning_rate": 9.844832584521606e-05, "loss": 2.5045, "step": 3415 }, { "epoch": 0.08, "learning_rate": 9.844366600205703e-05, "loss": 2.5819, "step": 3420 }, { "epoch": 0.08, "learning_rate": 9.843899928299683e-05, "loss": 2.5118, "step": 3425 }, { "epoch": 0.08, "learning_rate": 9.843432568869784e-05, "loss": 2.5186, "step": 3430 }, { "epoch": 0.08, "learning_rate": 9.842964521982339e-05, "loss": 2.4721, "step": 3435 }, { "epoch": 0.08, "learning_rate": 9.842495787703783e-05, "loss": 2.5372, "step": 3440 }, { "epoch": 0.08, "learning_rate": 9.842026366100645e-05, "loss": 2.5449, "step": 3445 }, { "epoch": 0.08, "learning_rate": 9.841556257239553e-05, "loss": 2.5462, "step": 3450 }, { "epoch": 0.08, "learning_rate": 9.841085461187233e-05, "loss": 2.5486, "step": 3455 }, { "epoch": 0.08, "learning_rate": 9.840613978010507e-05, "loss": 2.4982, "step": 3460 }, { "epoch": 0.08, "learning_rate": 9.840141807776297e-05, "loss": 2.5339, "step": 3465 }, { "epoch": 0.08, "learning_rate": 9.83966895055162e-05, "loss": 2.4946, "step": 3470 }, { "epoch": 0.08, "learning_rate": 9.839195406403589e-05, "loss": 2.525, "step": 3475 }, { "epoch": 0.08, "learning_rate": 9.838721175399421e-05, "loss": 2.4835, "step": 3480 }, { "epoch": 0.08, "learning_rate": 9.838246257606425e-05, "loss": 2.4998, "step": 3485 }, { "epoch": 0.08, "learning_rate": 9.837770653092008e-05, "loss": 2.5228, "step": 3490 }, { "epoch": 0.08, "learning_rate": 9.837294361923677e-05, "loss": 2.5127, "step": 3495 }, { "epoch": 0.08, "learning_rate": 9.836817384169033e-05, "loss": 2.5759, "step": 3500 }, { "epoch": 0.08, "learning_rate": 9.836339719895777e-05, "loss": 2.517, "step": 3505 }, { "epoch": 0.08, "learning_rate": 9.835861369171708e-05, "loss": 2.5237, "step": 3510 }, { "epoch": 0.08, "learning_rate": 9.83538233206472e-05, "loss": 2.5359, "step": 3515 }, { "epoch": 0.08, "learning_rate": 9.834902608642806e-05, "loss": 2.5209, "step": 3520 }, { "epoch": 0.08, "learning_rate": 9.834422198974056e-05, "loss": 2.5426, "step": 3525 }, { "epoch": 0.08, "learning_rate": 9.833941103126658e-05, "loss": 2.483, "step": 3530 }, { "epoch": 0.08, "learning_rate": 9.833459321168897e-05, "loss": 2.5814, "step": 3535 }, { "epoch": 0.08, "learning_rate": 9.83297685316915e-05, "loss": 2.5006, "step": 3540 }, { "epoch": 0.08, "learning_rate": 9.832493699195905e-05, "loss": 2.5082, "step": 3545 }, { "epoch": 0.08, "learning_rate": 9.832009859317734e-05, "loss": 2.5162, "step": 3550 }, { "epoch": 0.09, "learning_rate": 9.831525333603314e-05, "loss": 2.581, "step": 3555 }, { "epoch": 0.09, "learning_rate": 9.831040122121413e-05, "loss": 2.5279, "step": 3560 }, { "epoch": 0.09, "learning_rate": 9.830554224940902e-05, "loss": 2.4689, "step": 3565 }, { "epoch": 0.09, "learning_rate": 9.830067642130747e-05, "loss": 2.5543, "step": 3570 }, { "epoch": 0.09, "learning_rate": 9.829580373760012e-05, "loss": 2.5199, "step": 3575 }, { "epoch": 0.09, "learning_rate": 9.829092419897859e-05, "loss": 2.549, "step": 3580 }, { "epoch": 0.09, "learning_rate": 9.828603780613543e-05, "loss": 2.5117, "step": 3585 }, { "epoch": 0.09, "learning_rate": 9.828114455976424e-05, "loss": 2.4981, "step": 3590 }, { "epoch": 0.09, "learning_rate": 9.827624446055951e-05, "loss": 2.5, "step": 3595 }, { "epoch": 0.09, "learning_rate": 9.827133750921675e-05, "loss": 2.423, "step": 3600 }, { "epoch": 0.09, "learning_rate": 9.826642370643245e-05, "loss": 2.495, "step": 3605 }, { "epoch": 0.09, "learning_rate": 9.826150305290402e-05, "loss": 2.5472, "step": 3610 }, { "epoch": 0.09, "learning_rate": 9.825657554932992e-05, "loss": 2.4996, "step": 3615 }, { "epoch": 0.09, "learning_rate": 9.825164119640952e-05, "loss": 2.4926, "step": 3620 }, { "epoch": 0.09, "learning_rate": 9.824669999484317e-05, "loss": 2.4693, "step": 3625 }, { "epoch": 0.09, "learning_rate": 9.824175194533222e-05, "loss": 2.4514, "step": 3630 }, { "epoch": 0.09, "learning_rate": 9.823679704857899e-05, "loss": 2.5301, "step": 3635 }, { "epoch": 0.09, "learning_rate": 9.823183530528673e-05, "loss": 2.4551, "step": 3640 }, { "epoch": 0.09, "learning_rate": 9.82268667161597e-05, "loss": 2.5631, "step": 3645 }, { "epoch": 0.09, "learning_rate": 9.822189128190315e-05, "loss": 2.4765, "step": 3650 }, { "epoch": 0.09, "learning_rate": 9.821690900322322e-05, "loss": 2.4427, "step": 3655 }, { "epoch": 0.09, "learning_rate": 9.821191988082712e-05, "loss": 2.5228, "step": 3660 }, { "epoch": 0.09, "learning_rate": 9.820692391542294e-05, "loss": 2.5197, "step": 3665 }, { "epoch": 0.09, "learning_rate": 9.820192110771985e-05, "loss": 2.5676, "step": 3670 }, { "epoch": 0.09, "learning_rate": 9.819691145842788e-05, "loss": 2.5032, "step": 3675 }, { "epoch": 0.09, "learning_rate": 9.81918949682581e-05, "loss": 2.4869, "step": 3680 }, { "epoch": 0.09, "learning_rate": 9.818687163792251e-05, "loss": 2.588, "step": 3685 }, { "epoch": 0.09, "learning_rate": 9.818184146813413e-05, "loss": 2.5427, "step": 3690 }, { "epoch": 0.09, "learning_rate": 9.817680445960691e-05, "loss": 2.5316, "step": 3695 }, { "epoch": 0.09, "learning_rate": 9.817176061305578e-05, "loss": 2.5258, "step": 3700 }, { "epoch": 0.09, "learning_rate": 9.816670992919663e-05, "loss": 2.5277, "step": 3705 }, { "epoch": 0.09, "learning_rate": 9.816165240874637e-05, "loss": 2.531, "step": 3710 }, { "epoch": 0.09, "learning_rate": 9.815658805242281e-05, "loss": 2.4925, "step": 3715 }, { "epoch": 0.09, "learning_rate": 9.81515168609448e-05, "loss": 2.5302, "step": 3720 }, { "epoch": 0.09, "learning_rate": 9.814643883503208e-05, "loss": 2.4612, "step": 3725 }, { "epoch": 0.09, "learning_rate": 9.814135397540543e-05, "loss": 2.6235, "step": 3730 }, { "epoch": 0.09, "learning_rate": 9.813626228278658e-05, "loss": 2.5166, "step": 3735 }, { "epoch": 0.09, "learning_rate": 9.813116375789822e-05, "loss": 2.5045, "step": 3740 }, { "epoch": 0.09, "learning_rate": 9.8126058401464e-05, "loss": 2.5204, "step": 3745 }, { "epoch": 0.09, "learning_rate": 9.812094621420858e-05, "loss": 2.4512, "step": 3750 }, { "epoch": 0.09, "learning_rate": 9.811582719685755e-05, "loss": 2.5948, "step": 3755 }, { "epoch": 0.09, "learning_rate": 9.811070135013747e-05, "loss": 2.5093, "step": 3760 }, { "epoch": 0.09, "learning_rate": 9.81055686747759e-05, "loss": 2.5127, "step": 3765 }, { "epoch": 0.09, "learning_rate": 9.810042917150133e-05, "loss": 2.5022, "step": 3770 }, { "epoch": 0.09, "learning_rate": 9.809528284104327e-05, "loss": 2.4629, "step": 3775 }, { "epoch": 0.09, "learning_rate": 9.809012968413216e-05, "loss": 2.4866, "step": 3780 }, { "epoch": 0.09, "learning_rate": 9.80849697014994e-05, "loss": 2.4523, "step": 3785 }, { "epoch": 0.09, "learning_rate": 9.807980289387742e-05, "loss": 2.5873, "step": 3790 }, { "epoch": 0.09, "learning_rate": 9.807462926199953e-05, "loss": 2.5091, "step": 3795 }, { "epoch": 0.09, "learning_rate": 9.806944880660008e-05, "loss": 2.514, "step": 3800 }, { "epoch": 0.09, "learning_rate": 9.806426152841434e-05, "loss": 2.5629, "step": 3805 }, { "epoch": 0.09, "learning_rate": 9.805906742817859e-05, "loss": 2.5192, "step": 3810 }, { "epoch": 0.09, "learning_rate": 9.805386650663007e-05, "loss": 2.5087, "step": 3815 }, { "epoch": 0.09, "learning_rate": 9.804865876450697e-05, "loss": 2.5288, "step": 3820 }, { "epoch": 0.09, "learning_rate": 9.804344420254842e-05, "loss": 2.531, "step": 3825 }, { "epoch": 0.09, "learning_rate": 9.803822282149461e-05, "loss": 2.4973, "step": 3830 }, { "epoch": 0.09, "learning_rate": 9.803299462208662e-05, "loss": 2.5229, "step": 3835 }, { "epoch": 0.09, "learning_rate": 9.802775960506651e-05, "loss": 2.4951, "step": 3840 }, { "epoch": 0.09, "learning_rate": 9.802251777117735e-05, "loss": 2.5755, "step": 3845 }, { "epoch": 0.09, "learning_rate": 9.801726912116309e-05, "loss": 2.5387, "step": 3850 }, { "epoch": 0.09, "learning_rate": 9.801201365576876e-05, "loss": 2.5056, "step": 3855 }, { "epoch": 0.09, "learning_rate": 9.800675137574024e-05, "loss": 2.4909, "step": 3860 }, { "epoch": 0.09, "learning_rate": 9.800148228182449e-05, "loss": 2.5262, "step": 3865 }, { "epoch": 0.09, "learning_rate": 9.799620637476937e-05, "loss": 2.516, "step": 3870 }, { "epoch": 0.09, "learning_rate": 9.79909236553237e-05, "loss": 2.4633, "step": 3875 }, { "epoch": 0.09, "learning_rate": 9.798563412423731e-05, "loss": 2.506, "step": 3880 }, { "epoch": 0.09, "learning_rate": 9.798033778226096e-05, "loss": 2.5407, "step": 3885 }, { "epoch": 0.09, "learning_rate": 9.797503463014642e-05, "loss": 2.5501, "step": 3890 }, { "epoch": 0.09, "learning_rate": 9.796972466864636e-05, "loss": 2.4753, "step": 3895 }, { "epoch": 0.09, "learning_rate": 9.796440789851448e-05, "loss": 2.5453, "step": 3900 }, { "epoch": 0.09, "learning_rate": 9.795908432050543e-05, "loss": 2.4683, "step": 3905 }, { "epoch": 0.09, "learning_rate": 9.795375393537479e-05, "loss": 2.5055, "step": 3910 }, { "epoch": 0.09, "learning_rate": 9.794841674387913e-05, "loss": 2.5359, "step": 3915 }, { "epoch": 0.09, "learning_rate": 9.7943072746776e-05, "loss": 2.5091, "step": 3920 }, { "epoch": 0.09, "learning_rate": 9.793772194482394e-05, "loss": 2.5023, "step": 3925 }, { "epoch": 0.09, "learning_rate": 9.79323643387824e-05, "loss": 2.4791, "step": 3930 }, { "epoch": 0.09, "learning_rate": 9.792699992941178e-05, "loss": 2.5284, "step": 3935 }, { "epoch": 0.09, "learning_rate": 9.792162871747352e-05, "loss": 2.5162, "step": 3940 }, { "epoch": 0.09, "learning_rate": 9.791625070372998e-05, "loss": 2.5192, "step": 3945 }, { "epoch": 0.09, "learning_rate": 9.791086588894452e-05, "loss": 2.5276, "step": 3950 }, { "epoch": 0.09, "learning_rate": 9.790547427388139e-05, "loss": 2.5237, "step": 3955 }, { "epoch": 0.09, "learning_rate": 9.790007585930588e-05, "loss": 2.5311, "step": 3960 }, { "epoch": 0.09, "learning_rate": 9.789467064598423e-05, "loss": 2.5392, "step": 3965 }, { "epoch": 0.09, "learning_rate": 9.788925863468361e-05, "loss": 2.5665, "step": 3970 }, { "epoch": 0.1, "learning_rate": 9.78838398261722e-05, "loss": 2.48, "step": 3975 }, { "epoch": 0.1, "learning_rate": 9.78784142212191e-05, "loss": 2.4786, "step": 3980 }, { "epoch": 0.1, "learning_rate": 9.787298182059443e-05, "loss": 2.4618, "step": 3985 }, { "epoch": 0.1, "learning_rate": 9.786754262506923e-05, "loss": 2.4913, "step": 3990 }, { "epoch": 0.1, "learning_rate": 9.786209663541552e-05, "loss": 2.5142, "step": 3995 }, { "epoch": 0.1, "learning_rate": 9.785664385240627e-05, "loss": 2.4711, "step": 4000 }, { "epoch": 0.1, "learning_rate": 9.785118427681543e-05, "loss": 2.4157, "step": 4005 }, { "epoch": 0.1, "learning_rate": 9.784571790941792e-05, "loss": 2.4926, "step": 4010 }, { "epoch": 0.1, "learning_rate": 9.784024475098962e-05, "loss": 2.4895, "step": 4015 }, { "epoch": 0.1, "learning_rate": 9.783476480230734e-05, "loss": 2.501, "step": 4020 }, { "epoch": 0.1, "learning_rate": 9.782927806414891e-05, "loss": 2.5086, "step": 4025 }, { "epoch": 0.1, "learning_rate": 9.782378453729308e-05, "loss": 2.5363, "step": 4030 }, { "epoch": 0.1, "learning_rate": 9.78182842225196e-05, "loss": 2.5742, "step": 4035 }, { "epoch": 0.1, "learning_rate": 9.781277712060912e-05, "loss": 2.4552, "step": 4040 }, { "epoch": 0.1, "learning_rate": 9.780726323234335e-05, "loss": 2.5331, "step": 4045 }, { "epoch": 0.1, "learning_rate": 9.780174255850487e-05, "loss": 2.5126, "step": 4050 }, { "epoch": 0.1, "learning_rate": 9.779621509987728e-05, "loss": 2.5134, "step": 4055 }, { "epoch": 0.1, "learning_rate": 9.779068085724512e-05, "loss": 2.557, "step": 4060 }, { "epoch": 0.1, "learning_rate": 9.778513983139389e-05, "loss": 2.5028, "step": 4065 }, { "epoch": 0.1, "learning_rate": 9.777959202311007e-05, "loss": 2.5, "step": 4070 }, { "epoch": 0.1, "learning_rate": 9.777403743318112e-05, "loss": 2.5161, "step": 4075 }, { "epoch": 0.1, "learning_rate": 9.776847606239539e-05, "loss": 2.5178, "step": 4080 }, { "epoch": 0.1, "learning_rate": 9.776290791154226e-05, "loss": 2.4553, "step": 4085 }, { "epoch": 0.1, "learning_rate": 9.775733298141205e-05, "loss": 2.5605, "step": 4090 }, { "epoch": 0.1, "learning_rate": 9.775175127279606e-05, "loss": 2.5309, "step": 4095 }, { "epoch": 0.1, "learning_rate": 9.77461627864865e-05, "loss": 2.5942, "step": 4100 }, { "epoch": 0.1, "learning_rate": 9.77405675232766e-05, "loss": 2.525, "step": 4105 }, { "epoch": 0.1, "learning_rate": 9.773496548396054e-05, "loss": 2.5034, "step": 4110 }, { "epoch": 0.1, "learning_rate": 9.772935666933343e-05, "loss": 2.5391, "step": 4115 }, { "epoch": 0.1, "learning_rate": 9.772374108019138e-05, "loss": 2.4787, "step": 4120 }, { "epoch": 0.1, "learning_rate": 9.771811871733142e-05, "loss": 2.518, "step": 4125 }, { "epoch": 0.1, "learning_rate": 9.77124895815516e-05, "loss": 2.5446, "step": 4130 }, { "epoch": 0.1, "learning_rate": 9.770685367365086e-05, "loss": 2.4777, "step": 4135 }, { "epoch": 0.1, "learning_rate": 9.770121099442915e-05, "loss": 2.572, "step": 4140 }, { "epoch": 0.1, "learning_rate": 9.76955615446874e-05, "loss": 2.4982, "step": 4145 }, { "epoch": 0.1, "learning_rate": 9.768990532522743e-05, "loss": 2.5778, "step": 4150 }, { "epoch": 0.1, "learning_rate": 9.768424233685208e-05, "loss": 2.4829, "step": 4155 }, { "epoch": 0.1, "learning_rate": 9.767857258036513e-05, "loss": 2.4968, "step": 4160 }, { "epoch": 0.1, "learning_rate": 9.767289605657133e-05, "loss": 2.5719, "step": 4165 }, { "epoch": 0.1, "learning_rate": 9.766721276627636e-05, "loss": 2.5586, "step": 4170 }, { "epoch": 0.1, "learning_rate": 9.76615227102869e-05, "loss": 2.5707, "step": 4175 }, { "epoch": 0.1, "learning_rate": 9.765582588941058e-05, "loss": 2.4757, "step": 4180 }, { "epoch": 0.1, "learning_rate": 9.765012230445598e-05, "loss": 2.5111, "step": 4185 }, { "epoch": 0.1, "learning_rate": 9.764441195623264e-05, "loss": 2.5504, "step": 4190 }, { "epoch": 0.1, "learning_rate": 9.763869484555105e-05, "loss": 2.5139, "step": 4195 }, { "epoch": 0.1, "learning_rate": 9.763297097322271e-05, "loss": 2.5337, "step": 4200 }, { "epoch": 0.1, "learning_rate": 9.762724034006002e-05, "loss": 2.5324, "step": 4205 }, { "epoch": 0.1, "learning_rate": 9.762150294687636e-05, "loss": 2.5713, "step": 4210 }, { "epoch": 0.1, "learning_rate": 9.761575879448607e-05, "loss": 2.5624, "step": 4215 }, { "epoch": 0.1, "learning_rate": 9.76100078837045e-05, "loss": 2.4865, "step": 4220 }, { "epoch": 0.1, "learning_rate": 9.760425021534783e-05, "loss": 2.538, "step": 4225 }, { "epoch": 0.1, "learning_rate": 9.759848579023334e-05, "loss": 2.5842, "step": 4230 }, { "epoch": 0.1, "learning_rate": 9.759271460917919e-05, "loss": 2.4978, "step": 4235 }, { "epoch": 0.1, "learning_rate": 9.758693667300451e-05, "loss": 2.5436, "step": 4240 }, { "epoch": 0.1, "learning_rate": 9.758115198252941e-05, "loss": 2.5284, "step": 4245 }, { "epoch": 0.1, "learning_rate": 9.757536053857496e-05, "loss": 2.4961, "step": 4250 }, { "epoch": 0.1, "learning_rate": 9.756956234196316e-05, "loss": 2.5359, "step": 4255 }, { "epoch": 0.1, "learning_rate": 9.756375739351698e-05, "loss": 2.5441, "step": 4260 }, { "epoch": 0.1, "learning_rate": 9.755794569406033e-05, "loss": 2.5635, "step": 4265 }, { "epoch": 0.1, "learning_rate": 9.755212724441813e-05, "loss": 2.5298, "step": 4270 }, { "epoch": 0.1, "learning_rate": 9.754630204541624e-05, "loss": 2.5039, "step": 4275 }, { "epoch": 0.1, "learning_rate": 9.754047009788143e-05, "loss": 2.543, "step": 4280 }, { "epoch": 0.1, "learning_rate": 9.753463140264148e-05, "loss": 2.4918, "step": 4285 }, { "epoch": 0.1, "learning_rate": 9.752878596052512e-05, "loss": 2.5689, "step": 4290 }, { "epoch": 0.1, "learning_rate": 9.752293377236201e-05, "loss": 2.5258, "step": 4295 }, { "epoch": 0.1, "learning_rate": 9.75170748389828e-05, "loss": 2.5137, "step": 4300 }, { "epoch": 0.1, "learning_rate": 9.751120916121907e-05, "loss": 2.4823, "step": 4305 }, { "epoch": 0.1, "learning_rate": 9.750533673990339e-05, "loss": 2.4307, "step": 4310 }, { "epoch": 0.1, "learning_rate": 9.749945757586924e-05, "loss": 2.5192, "step": 4315 }, { "epoch": 0.1, "learning_rate": 9.749357166995112e-05, "loss": 2.5557, "step": 4320 }, { "epoch": 0.1, "learning_rate": 9.748767902298443e-05, "loss": 2.5367, "step": 4325 }, { "epoch": 0.1, "learning_rate": 9.748177963580556e-05, "loss": 2.5113, "step": 4330 }, { "epoch": 0.1, "learning_rate": 9.747587350925184e-05, "loss": 2.493, "step": 4335 }, { "epoch": 0.1, "learning_rate": 9.746996064416155e-05, "loss": 2.4499, "step": 4340 }, { "epoch": 0.1, "learning_rate": 9.746404104137397e-05, "loss": 2.5046, "step": 4345 }, { "epoch": 0.1, "learning_rate": 9.745811470172926e-05, "loss": 2.4619, "step": 4350 }, { "epoch": 0.1, "learning_rate": 9.745218162606861e-05, "loss": 2.4923, "step": 4355 }, { "epoch": 0.1, "learning_rate": 9.744624181523415e-05, "loss": 2.5265, "step": 4360 }, { "epoch": 0.1, "learning_rate": 9.744029527006892e-05, "loss": 2.5323, "step": 4365 }, { "epoch": 0.1, "learning_rate": 9.743434199141699e-05, "loss": 2.5588, "step": 4370 }, { "epoch": 0.1, "learning_rate": 9.742838198012329e-05, "loss": 2.4998, "step": 4375 }, { "epoch": 0.1, "learning_rate": 9.742241523703382e-05, "loss": 2.5689, "step": 4380 }, { "epoch": 0.1, "learning_rate": 9.741644176299542e-05, "loss": 2.4732, "step": 4385 }, { "epoch": 0.11, "learning_rate": 9.741046155885599e-05, "loss": 2.5078, "step": 4390 }, { "epoch": 0.11, "learning_rate": 9.740447462546429e-05, "loss": 2.5486, "step": 4395 }, { "epoch": 0.11, "learning_rate": 9.739848096367013e-05, "loss": 2.4982, "step": 4400 }, { "epoch": 0.11, "learning_rate": 9.739248057432417e-05, "loss": 2.5244, "step": 4405 }, { "epoch": 0.11, "learning_rate": 9.738647345827814e-05, "loss": 2.5092, "step": 4410 }, { "epoch": 0.11, "learning_rate": 9.738045961638462e-05, "loss": 2.5569, "step": 4415 }, { "epoch": 0.11, "learning_rate": 9.737443904949723e-05, "loss": 2.5522, "step": 4420 }, { "epoch": 0.11, "learning_rate": 9.736841175847047e-05, "loss": 2.5417, "step": 4425 }, { "epoch": 0.11, "learning_rate": 9.736237774415987e-05, "loss": 2.5194, "step": 4430 }, { "epoch": 0.11, "learning_rate": 9.735633700742181e-05, "loss": 2.5233, "step": 4435 }, { "epoch": 0.11, "learning_rate": 9.735028954911375e-05, "loss": 2.5098, "step": 4440 }, { "epoch": 0.11, "learning_rate": 9.7344235370094e-05, "loss": 2.5626, "step": 4445 }, { "epoch": 0.11, "learning_rate": 9.733817447122192e-05, "loss": 2.4889, "step": 4450 }, { "epoch": 0.11, "learning_rate": 9.73321068533577e-05, "loss": 2.5091, "step": 4455 }, { "epoch": 0.11, "learning_rate": 9.73260325173626e-05, "loss": 2.5151, "step": 4460 }, { "epoch": 0.11, "learning_rate": 9.73199514640988e-05, "loss": 2.4723, "step": 4465 }, { "epoch": 0.11, "learning_rate": 9.731386369442938e-05, "loss": 2.4797, "step": 4470 }, { "epoch": 0.11, "learning_rate": 9.730776920921842e-05, "loss": 2.4504, "step": 4475 }, { "epoch": 0.11, "learning_rate": 9.730166800933097e-05, "loss": 2.5609, "step": 4480 }, { "epoch": 0.11, "learning_rate": 9.729556009563298e-05, "loss": 2.5239, "step": 4485 }, { "epoch": 0.11, "learning_rate": 9.728944546899142e-05, "loss": 2.5434, "step": 4490 }, { "epoch": 0.11, "learning_rate": 9.728332413027414e-05, "loss": 2.5213, "step": 4495 }, { "epoch": 0.11, "learning_rate": 9.727719608034999e-05, "loss": 2.4585, "step": 4500 }, { "epoch": 0.11, "learning_rate": 9.727106132008879e-05, "loss": 2.5306, "step": 4505 }, { "epoch": 0.11, "learning_rate": 9.726491985036124e-05, "loss": 2.508, "step": 4510 }, { "epoch": 0.11, "learning_rate": 9.725877167203908e-05, "loss": 2.552, "step": 4515 }, { "epoch": 0.11, "learning_rate": 9.72526167859949e-05, "loss": 2.5442, "step": 4520 }, { "epoch": 0.11, "learning_rate": 9.724645519310235e-05, "loss": 2.4691, "step": 4525 }, { "epoch": 0.11, "learning_rate": 9.724028689423596e-05, "loss": 2.4522, "step": 4530 }, { "epoch": 0.11, "learning_rate": 9.723411189027123e-05, "loss": 2.5046, "step": 4535 }, { "epoch": 0.11, "learning_rate": 9.722793018208463e-05, "loss": 2.5145, "step": 4540 }, { "epoch": 0.11, "learning_rate": 9.722174177055356e-05, "loss": 2.4991, "step": 4545 }, { "epoch": 0.11, "learning_rate": 9.721554665655639e-05, "loss": 2.4894, "step": 4550 }, { "epoch": 0.11, "learning_rate": 9.720934484097241e-05, "loss": 2.4823, "step": 4555 }, { "epoch": 0.11, "learning_rate": 9.72031363246819e-05, "loss": 2.4988, "step": 4560 }, { "epoch": 0.11, "learning_rate": 9.719692110856604e-05, "loss": 2.482, "step": 4565 }, { "epoch": 0.11, "learning_rate": 9.719069919350704e-05, "loss": 2.5394, "step": 4570 }, { "epoch": 0.11, "learning_rate": 9.718447058038798e-05, "loss": 2.4355, "step": 4575 }, { "epoch": 0.11, "learning_rate": 9.717823527009293e-05, "loss": 2.593, "step": 4580 }, { "epoch": 0.11, "learning_rate": 9.71719932635069e-05, "loss": 2.5236, "step": 4585 }, { "epoch": 0.11, "learning_rate": 9.716574456151589e-05, "loss": 2.5442, "step": 4590 }, { "epoch": 0.11, "learning_rate": 9.715948916500677e-05, "loss": 2.4942, "step": 4595 }, { "epoch": 0.11, "learning_rate": 9.715322707486744e-05, "loss": 2.5115, "step": 4600 }, { "epoch": 0.11, "learning_rate": 9.714695829198669e-05, "loss": 2.5855, "step": 4605 }, { "epoch": 0.11, "learning_rate": 9.71406828172543e-05, "loss": 2.4525, "step": 4610 }, { "epoch": 0.11, "learning_rate": 9.713440065156098e-05, "loss": 2.5366, "step": 4615 }, { "epoch": 0.11, "learning_rate": 9.712811179579842e-05, "loss": 2.513, "step": 4620 }, { "epoch": 0.11, "learning_rate": 9.712181625085922e-05, "loss": 2.4912, "step": 4625 }, { "epoch": 0.11, "learning_rate": 9.711551401763691e-05, "loss": 2.5313, "step": 4630 }, { "epoch": 0.11, "learning_rate": 9.710920509702604e-05, "loss": 2.4905, "step": 4635 }, { "epoch": 0.11, "learning_rate": 9.710288948992207e-05, "loss": 2.4938, "step": 4640 }, { "epoch": 0.11, "learning_rate": 9.709656719722142e-05, "loss": 2.4688, "step": 4645 }, { "epoch": 0.11, "learning_rate": 9.709023821982144e-05, "loss": 2.536, "step": 4650 }, { "epoch": 0.11, "learning_rate": 9.708390255862043e-05, "loss": 2.5225, "step": 4655 }, { "epoch": 0.11, "learning_rate": 9.707756021451767e-05, "loss": 2.4763, "step": 4660 }, { "epoch": 0.11, "learning_rate": 9.707121118841336e-05, "loss": 2.5139, "step": 4665 }, { "epoch": 0.11, "learning_rate": 9.706485548120862e-05, "loss": 2.5379, "step": 4670 }, { "epoch": 0.11, "learning_rate": 9.705849309380562e-05, "loss": 2.5135, "step": 4675 }, { "epoch": 0.11, "learning_rate": 9.705212402710736e-05, "loss": 2.5493, "step": 4680 }, { "epoch": 0.11, "learning_rate": 9.704574828201785e-05, "loss": 2.5715, "step": 4685 }, { "epoch": 0.11, "learning_rate": 9.703936585944207e-05, "loss": 2.5492, "step": 4690 }, { "epoch": 0.11, "learning_rate": 9.703297676028587e-05, "loss": 2.4626, "step": 4695 }, { "epoch": 0.11, "learning_rate": 9.702658098545612e-05, "loss": 2.5761, "step": 4700 }, { "epoch": 0.11, "learning_rate": 9.702017853586059e-05, "loss": 2.57, "step": 4705 }, { "epoch": 0.11, "learning_rate": 9.701376941240805e-05, "loss": 2.5162, "step": 4710 }, { "epoch": 0.11, "learning_rate": 9.700735361600817e-05, "loss": 2.5593, "step": 4715 }, { "epoch": 0.11, "learning_rate": 9.700093114757156e-05, "loss": 2.517, "step": 4720 }, { "epoch": 0.11, "learning_rate": 9.699450200800984e-05, "loss": 2.5297, "step": 4725 }, { "epoch": 0.11, "learning_rate": 9.698806619823552e-05, "loss": 2.511, "step": 4730 }, { "epoch": 0.11, "learning_rate": 9.698162371916206e-05, "loss": 2.5374, "step": 4735 }, { "epoch": 0.11, "learning_rate": 9.697517457170388e-05, "loss": 2.4986, "step": 4740 }, { "epoch": 0.11, "learning_rate": 9.696871875677638e-05, "loss": 2.5022, "step": 4745 }, { "epoch": 0.11, "learning_rate": 9.696225627529584e-05, "loss": 2.5827, "step": 4750 }, { "epoch": 0.11, "learning_rate": 9.695578712817953e-05, "loss": 2.5136, "step": 4755 }, { "epoch": 0.11, "learning_rate": 9.694931131634564e-05, "loss": 2.5549, "step": 4760 }, { "epoch": 0.11, "learning_rate": 9.694282884071336e-05, "loss": 2.4917, "step": 4765 }, { "epoch": 0.11, "learning_rate": 9.693633970220272e-05, "loss": 2.462, "step": 4770 }, { "epoch": 0.11, "learning_rate": 9.692984390173484e-05, "loss": 2.5567, "step": 4775 }, { "epoch": 0.11, "learning_rate": 9.692334144023165e-05, "loss": 2.4944, "step": 4780 }, { "epoch": 0.11, "learning_rate": 9.691683231861612e-05, "loss": 2.4969, "step": 4785 }, { "epoch": 0.11, "learning_rate": 9.691031653781211e-05, "loss": 2.5175, "step": 4790 }, { "epoch": 0.11, "learning_rate": 9.690379409874443e-05, "loss": 2.53, "step": 4795 }, { "epoch": 0.11, "learning_rate": 9.689726500233888e-05, "loss": 2.5338, "step": 4800 }, { "epoch": 0.11, "learning_rate": 9.689072924952217e-05, "loss": 2.5323, "step": 4805 }, { "epoch": 0.12, "learning_rate": 9.688418684122194e-05, "loss": 2.4505, "step": 4810 }, { "epoch": 0.12, "learning_rate": 9.687763777836681e-05, "loss": 2.4779, "step": 4815 }, { "epoch": 0.12, "learning_rate": 9.687108206188631e-05, "loss": 2.5095, "step": 4820 }, { "epoch": 0.12, "learning_rate": 9.686451969271094e-05, "loss": 2.4486, "step": 4825 }, { "epoch": 0.12, "learning_rate": 9.685795067177215e-05, "loss": 2.4563, "step": 4830 }, { "epoch": 0.12, "learning_rate": 9.685137500000231e-05, "loss": 2.4432, "step": 4835 }, { "epoch": 0.12, "learning_rate": 9.684479267833474e-05, "loss": 2.4887, "step": 4840 }, { "epoch": 0.12, "learning_rate": 9.683820370770371e-05, "loss": 2.5045, "step": 4845 }, { "epoch": 0.12, "learning_rate": 9.683160808904445e-05, "loss": 2.5438, "step": 4850 }, { "epoch": 0.12, "learning_rate": 9.68250058232931e-05, "loss": 2.5371, "step": 4855 }, { "epoch": 0.12, "learning_rate": 9.681839691138674e-05, "loss": 2.5178, "step": 4860 }, { "epoch": 0.12, "learning_rate": 9.681178135426345e-05, "loss": 2.5063, "step": 4865 }, { "epoch": 0.12, "learning_rate": 9.680515915286219e-05, "loss": 2.5273, "step": 4870 }, { "epoch": 0.12, "learning_rate": 9.679853030812291e-05, "loss": 2.4237, "step": 4875 }, { "epoch": 0.12, "learning_rate": 9.679189482098648e-05, "loss": 2.5902, "step": 4880 }, { "epoch": 0.12, "learning_rate": 9.678525269239468e-05, "loss": 2.5239, "step": 4885 }, { "epoch": 0.12, "learning_rate": 9.67786039232903e-05, "loss": 2.5502, "step": 4890 }, { "epoch": 0.12, "learning_rate": 9.677194851461704e-05, "loss": 2.5103, "step": 4895 }, { "epoch": 0.12, "learning_rate": 9.676528646731952e-05, "loss": 2.5, "step": 4900 }, { "epoch": 0.12, "learning_rate": 9.675861778234336e-05, "loss": 2.4793, "step": 4905 }, { "epoch": 0.12, "learning_rate": 9.675194246063504e-05, "loss": 2.5053, "step": 4910 }, { "epoch": 0.12, "learning_rate": 9.674526050314207e-05, "loss": 2.5459, "step": 4915 }, { "epoch": 0.12, "learning_rate": 9.673857191081284e-05, "loss": 2.5055, "step": 4920 }, { "epoch": 0.12, "learning_rate": 9.673187668459671e-05, "loss": 2.5135, "step": 4925 }, { "epoch": 0.12, "learning_rate": 9.672517482544396e-05, "loss": 2.5003, "step": 4930 }, { "epoch": 0.12, "learning_rate": 9.671846633430585e-05, "loss": 2.522, "step": 4935 }, { "epoch": 0.12, "learning_rate": 9.671175121213453e-05, "loss": 2.5354, "step": 4940 }, { "epoch": 0.12, "learning_rate": 9.670502945988313e-05, "loss": 2.533, "step": 4945 }, { "epoch": 0.12, "learning_rate": 9.669830107850573e-05, "loss": 2.5423, "step": 4950 }, { "epoch": 0.12, "learning_rate": 9.669156606895728e-05, "loss": 2.507, "step": 4955 }, { "epoch": 0.12, "learning_rate": 9.668482443219376e-05, "loss": 2.5094, "step": 4960 }, { "epoch": 0.12, "learning_rate": 9.667807616917206e-05, "loss": 2.5677, "step": 4965 }, { "epoch": 0.12, "learning_rate": 9.667132128084994e-05, "loss": 2.5034, "step": 4970 }, { "epoch": 0.12, "learning_rate": 9.666455976818624e-05, "loss": 2.5587, "step": 4975 }, { "epoch": 0.12, "learning_rate": 9.665779163214061e-05, "loss": 2.5119, "step": 4980 }, { "epoch": 0.12, "learning_rate": 9.66510168736737e-05, "loss": 2.562, "step": 4985 }, { "epoch": 0.12, "learning_rate": 9.664423549374711e-05, "loss": 2.5137, "step": 4990 }, { "epoch": 0.12, "learning_rate": 9.663744749332335e-05, "loss": 2.4616, "step": 4995 }, { "epoch": 0.12, "learning_rate": 9.663065287336589e-05, "loss": 2.4797, "step": 5000 }, { "epoch": 0.12, "learning_rate": 9.662385163483912e-05, "loss": 2.5138, "step": 5005 }, { "epoch": 0.12, "learning_rate": 9.661704377870839e-05, "loss": 2.5289, "step": 5010 }, { "epoch": 0.12, "learning_rate": 9.661022930593997e-05, "loss": 2.4949, "step": 5015 }, { "epoch": 0.12, "learning_rate": 9.660340821750109e-05, "loss": 2.487, "step": 5020 }, { "epoch": 0.12, "learning_rate": 9.659658051435988e-05, "loss": 2.508, "step": 5025 }, { "epoch": 0.12, "learning_rate": 9.658974619748548e-05, "loss": 2.5437, "step": 5030 }, { "epoch": 0.12, "learning_rate": 9.658290526784793e-05, "loss": 2.5158, "step": 5035 }, { "epoch": 0.12, "learning_rate": 9.657605772641815e-05, "loss": 2.4578, "step": 5040 }, { "epoch": 0.12, "learning_rate": 9.656920357416808e-05, "loss": 2.4798, "step": 5045 }, { "epoch": 0.12, "learning_rate": 9.656234281207059e-05, "loss": 2.4863, "step": 5050 }, { "epoch": 0.12, "learning_rate": 9.655547544109944e-05, "loss": 2.4837, "step": 5055 }, { "epoch": 0.12, "learning_rate": 9.65486014622294e-05, "loss": 2.5619, "step": 5060 }, { "epoch": 0.12, "learning_rate": 9.654172087643607e-05, "loss": 2.5252, "step": 5065 }, { "epoch": 0.12, "learning_rate": 9.65348336846961e-05, "loss": 2.5011, "step": 5070 }, { "epoch": 0.12, "learning_rate": 9.652793988798703e-05, "loss": 2.4922, "step": 5075 }, { "epoch": 0.12, "learning_rate": 9.652103948728732e-05, "loss": 2.5284, "step": 5080 }, { "epoch": 0.12, "learning_rate": 9.651413248357639e-05, "loss": 2.5095, "step": 5085 }, { "epoch": 0.12, "learning_rate": 9.65072188778346e-05, "loss": 2.5561, "step": 5090 }, { "epoch": 0.12, "learning_rate": 9.650029867104325e-05, "loss": 2.4634, "step": 5095 }, { "epoch": 0.12, "learning_rate": 9.649337186418452e-05, "loss": 2.5589, "step": 5100 }, { "epoch": 0.12, "learning_rate": 9.648643845824162e-05, "loss": 2.4704, "step": 5105 }, { "epoch": 0.12, "learning_rate": 9.647949845419863e-05, "loss": 2.5034, "step": 5110 }, { "epoch": 0.12, "learning_rate": 9.647255185304061e-05, "loss": 2.5025, "step": 5115 }, { "epoch": 0.12, "learning_rate": 9.64655986557535e-05, "loss": 2.4619, "step": 5120 }, { "epoch": 0.12, "learning_rate": 9.645863886332424e-05, "loss": 2.4804, "step": 5125 }, { "epoch": 0.12, "learning_rate": 9.645167247674065e-05, "loss": 2.5355, "step": 5130 }, { "epoch": 0.12, "learning_rate": 9.644469949699151e-05, "loss": 2.5079, "step": 5135 }, { "epoch": 0.12, "learning_rate": 9.643771992506657e-05, "loss": 2.5698, "step": 5140 }, { "epoch": 0.12, "learning_rate": 9.643073376195644e-05, "loss": 2.4668, "step": 5145 }, { "epoch": 0.12, "learning_rate": 9.642374100865272e-05, "loss": 2.4705, "step": 5150 }, { "epoch": 0.12, "learning_rate": 9.641674166614796e-05, "loss": 2.478, "step": 5155 }, { "epoch": 0.12, "learning_rate": 9.64097357354356e-05, "loss": 2.5377, "step": 5160 }, { "epoch": 0.12, "learning_rate": 9.640272321751002e-05, "loss": 2.5195, "step": 5165 }, { "epoch": 0.12, "learning_rate": 9.639570411336659e-05, "loss": 2.5134, "step": 5170 }, { "epoch": 0.12, "learning_rate": 9.638867842400153e-05, "loss": 2.4696, "step": 5175 }, { "epoch": 0.12, "learning_rate": 9.638164615041206e-05, "loss": 2.4959, "step": 5180 }, { "epoch": 0.12, "learning_rate": 9.63746072935963e-05, "loss": 2.5147, "step": 5185 }, { "epoch": 0.12, "learning_rate": 9.636756185455333e-05, "loss": 2.492, "step": 5190 }, { "epoch": 0.12, "learning_rate": 9.636050983428317e-05, "loss": 2.5051, "step": 5195 }, { "epoch": 0.12, "learning_rate": 9.635345123378671e-05, "loss": 2.5179, "step": 5200 }, { "epoch": 0.12, "learning_rate": 9.634638605406585e-05, "loss": 2.5566, "step": 5205 }, { "epoch": 0.12, "learning_rate": 9.633931429612337e-05, "loss": 2.5507, "step": 5210 }, { "epoch": 0.12, "learning_rate": 9.633223596096303e-05, "loss": 2.519, "step": 5215 }, { "epoch": 0.12, "learning_rate": 9.632515104958951e-05, "loss": 2.4932, "step": 5220 }, { "epoch": 0.13, "learning_rate": 9.631805956300838e-05, "loss": 2.4961, "step": 5225 }, { "epoch": 0.13, "learning_rate": 9.631096150222622e-05, "loss": 2.4866, "step": 5230 }, { "epoch": 0.13, "learning_rate": 9.630385686825046e-05, "loss": 2.5426, "step": 5235 }, { "epoch": 0.13, "learning_rate": 9.629674566208952e-05, "loss": 2.5492, "step": 5240 }, { "epoch": 0.13, "learning_rate": 9.628962788475273e-05, "loss": 2.4959, "step": 5245 }, { "epoch": 0.13, "learning_rate": 9.628250353725037e-05, "loss": 2.5149, "step": 5250 }, { "epoch": 0.13, "learning_rate": 9.627537262059365e-05, "loss": 2.5054, "step": 5255 }, { "epoch": 0.13, "learning_rate": 9.626823513579469e-05, "loss": 2.4764, "step": 5260 }, { "epoch": 0.13, "learning_rate": 9.626109108386654e-05, "loss": 2.4589, "step": 5265 }, { "epoch": 0.13, "learning_rate": 9.625394046582321e-05, "loss": 2.5125, "step": 5270 }, { "epoch": 0.13, "learning_rate": 9.624678328267964e-05, "loss": 2.5245, "step": 5275 }, { "epoch": 0.13, "learning_rate": 9.62396195354517e-05, "loss": 2.4724, "step": 5280 }, { "epoch": 0.13, "learning_rate": 9.623244922515617e-05, "loss": 2.5251, "step": 5285 }, { "epoch": 0.13, "learning_rate": 9.622527235281076e-05, "loss": 2.5089, "step": 5290 }, { "epoch": 0.13, "learning_rate": 9.621808891943416e-05, "loss": 2.5217, "step": 5295 }, { "epoch": 0.13, "learning_rate": 9.621089892604593e-05, "loss": 2.4118, "step": 5300 }, { "epoch": 0.13, "learning_rate": 9.62037023736666e-05, "loss": 2.5414, "step": 5305 }, { "epoch": 0.13, "learning_rate": 9.619649926331764e-05, "loss": 2.4881, "step": 5310 }, { "epoch": 0.13, "learning_rate": 9.61892895960214e-05, "loss": 2.512, "step": 5315 }, { "epoch": 0.13, "learning_rate": 9.618207337280118e-05, "loss": 2.457, "step": 5320 }, { "epoch": 0.13, "learning_rate": 9.617485059468128e-05, "loss": 2.429, "step": 5325 }, { "epoch": 0.13, "learning_rate": 9.616762126268682e-05, "loss": 2.5086, "step": 5330 }, { "epoch": 0.13, "learning_rate": 9.616038537784391e-05, "loss": 2.4819, "step": 5335 }, { "epoch": 0.13, "learning_rate": 9.61531429411796e-05, "loss": 2.5529, "step": 5340 }, { "epoch": 0.13, "learning_rate": 9.614589395372186e-05, "loss": 2.5205, "step": 5345 }, { "epoch": 0.13, "learning_rate": 9.613863841649955e-05, "loss": 2.4864, "step": 5350 }, { "epoch": 0.13, "learning_rate": 9.613137633054251e-05, "loss": 2.473, "step": 5355 }, { "epoch": 0.13, "learning_rate": 9.61241076968815e-05, "loss": 2.5598, "step": 5360 }, { "epoch": 0.13, "learning_rate": 9.611683251654818e-05, "loss": 2.4922, "step": 5365 }, { "epoch": 0.13, "learning_rate": 9.610955079057517e-05, "loss": 2.5117, "step": 5370 }, { "epoch": 0.13, "learning_rate": 9.610226251999604e-05, "loss": 2.4031, "step": 5375 }, { "epoch": 0.13, "learning_rate": 9.609496770584519e-05, "loss": 2.4669, "step": 5380 }, { "epoch": 0.13, "learning_rate": 9.608766634915809e-05, "loss": 2.5187, "step": 5385 }, { "epoch": 0.13, "learning_rate": 9.608035845097101e-05, "loss": 2.442, "step": 5390 }, { "epoch": 0.13, "learning_rate": 9.607304401232123e-05, "loss": 2.466, "step": 5395 }, { "epoch": 0.13, "learning_rate": 9.606572303424692e-05, "loss": 2.5454, "step": 5400 }, { "epoch": 0.13, "learning_rate": 9.605839551778722e-05, "loss": 2.452, "step": 5405 }, { "epoch": 0.13, "learning_rate": 9.605106146398213e-05, "loss": 2.4458, "step": 5410 }, { "epoch": 0.13, "learning_rate": 9.604372087387264e-05, "loss": 2.4868, "step": 5415 }, { "epoch": 0.13, "learning_rate": 9.603637374850064e-05, "loss": 2.5013, "step": 5420 }, { "epoch": 0.13, "learning_rate": 9.602902008890896e-05, "loss": 2.4962, "step": 5425 }, { "epoch": 0.13, "learning_rate": 9.602165989614132e-05, "loss": 2.5232, "step": 5430 }, { "epoch": 0.13, "learning_rate": 9.60142931712424e-05, "loss": 2.5135, "step": 5435 }, { "epoch": 0.13, "learning_rate": 9.600691991525785e-05, "loss": 2.4152, "step": 5440 }, { "epoch": 0.13, "learning_rate": 9.599954012923416e-05, "loss": 2.4767, "step": 5445 }, { "epoch": 0.13, "learning_rate": 9.599215381421879e-05, "loss": 2.4649, "step": 5450 }, { "epoch": 0.13, "learning_rate": 9.598476097126013e-05, "loss": 2.5194, "step": 5455 }, { "epoch": 0.13, "learning_rate": 9.59773616014075e-05, "loss": 2.5012, "step": 5460 }, { "epoch": 0.13, "learning_rate": 9.596995570571112e-05, "loss": 2.501, "step": 5465 }, { "epoch": 0.13, "learning_rate": 9.596254328522218e-05, "loss": 2.5597, "step": 5470 }, { "epoch": 0.13, "learning_rate": 9.595512434099273e-05, "loss": 2.5426, "step": 5475 }, { "epoch": 0.13, "learning_rate": 9.594769887407582e-05, "loss": 2.5596, "step": 5480 }, { "epoch": 0.13, "learning_rate": 9.594026688552538e-05, "loss": 2.4285, "step": 5485 }, { "epoch": 0.13, "learning_rate": 9.593282837639626e-05, "loss": 2.484, "step": 5490 }, { "epoch": 0.13, "learning_rate": 9.592538334774428e-05, "loss": 2.5189, "step": 5495 }, { "epoch": 0.13, "learning_rate": 9.591793180062614e-05, "loss": 2.5056, "step": 5500 }, { "epoch": 0.13, "learning_rate": 9.591047373609948e-05, "loss": 2.4894, "step": 5505 }, { "epoch": 0.13, "learning_rate": 9.590300915522289e-05, "loss": 2.4643, "step": 5510 }, { "epoch": 0.13, "learning_rate": 9.589553805905585e-05, "loss": 2.5084, "step": 5515 }, { "epoch": 0.13, "learning_rate": 9.588806044865876e-05, "loss": 2.4749, "step": 5520 }, { "epoch": 0.13, "learning_rate": 9.588057632509299e-05, "loss": 2.4408, "step": 5525 }, { "epoch": 0.13, "learning_rate": 9.587308568942079e-05, "loss": 2.4963, "step": 5530 }, { "epoch": 0.13, "learning_rate": 9.586558854270538e-05, "loss": 2.5413, "step": 5535 }, { "epoch": 0.13, "learning_rate": 9.585808488601082e-05, "loss": 2.4916, "step": 5540 }, { "epoch": 0.13, "learning_rate": 9.58505747204022e-05, "loss": 2.5349, "step": 5545 }, { "epoch": 0.13, "learning_rate": 9.584305804694546e-05, "loss": 2.5245, "step": 5550 }, { "epoch": 0.13, "learning_rate": 9.583553486670749e-05, "loss": 2.4861, "step": 5555 }, { "epoch": 0.13, "learning_rate": 9.58280051807561e-05, "loss": 2.4543, "step": 5560 }, { "epoch": 0.13, "learning_rate": 9.582046899016002e-05, "loss": 2.5029, "step": 5565 }, { "epoch": 0.13, "learning_rate": 9.581292629598892e-05, "loss": 2.5212, "step": 5570 }, { "epoch": 0.13, "learning_rate": 9.580537709931336e-05, "loss": 2.5027, "step": 5575 }, { "epoch": 0.13, "learning_rate": 9.579782140120487e-05, "loss": 2.4863, "step": 5580 }, { "epoch": 0.13, "learning_rate": 9.579025920273586e-05, "loss": 2.5093, "step": 5585 }, { "epoch": 0.13, "learning_rate": 9.578269050497966e-05, "loss": 2.5417, "step": 5590 }, { "epoch": 0.13, "learning_rate": 9.577511530901057e-05, "loss": 2.4736, "step": 5595 }, { "epoch": 0.13, "learning_rate": 9.576753361590378e-05, "loss": 2.383, "step": 5600 }, { "epoch": 0.13, "learning_rate": 9.575994542673538e-05, "loss": 2.533, "step": 5605 }, { "epoch": 0.13, "learning_rate": 9.575235074258244e-05, "loss": 2.4321, "step": 5610 }, { "epoch": 0.13, "learning_rate": 9.57447495645229e-05, "loss": 2.4459, "step": 5615 }, { "epoch": 0.13, "learning_rate": 9.573714189363566e-05, "loss": 2.5021, "step": 5620 }, { "epoch": 0.13, "learning_rate": 9.572952773100049e-05, "loss": 2.5083, "step": 5625 }, { "epoch": 0.13, "learning_rate": 9.572190707769814e-05, "loss": 2.3938, "step": 5630 }, { "epoch": 0.13, "learning_rate": 9.571427993481023e-05, "loss": 2.5193, "step": 5635 }, { "epoch": 0.13, "learning_rate": 9.570664630341937e-05, "loss": 2.4905, "step": 5640 }, { "epoch": 0.14, "learning_rate": 9.569900618460901e-05, "loss": 2.4944, "step": 5645 }, { "epoch": 0.14, "learning_rate": 9.569135957946355e-05, "loss": 2.5411, "step": 5650 }, { "epoch": 0.14, "learning_rate": 9.568370648906836e-05, "loss": 2.51, "step": 5655 }, { "epoch": 0.14, "learning_rate": 9.567604691450967e-05, "loss": 2.4446, "step": 5660 }, { "epoch": 0.14, "learning_rate": 9.566838085687464e-05, "loss": 2.5362, "step": 5665 }, { "epoch": 0.14, "learning_rate": 9.566070831725136e-05, "loss": 2.5414, "step": 5670 }, { "epoch": 0.14, "learning_rate": 9.565302929672885e-05, "loss": 2.513, "step": 5675 }, { "epoch": 0.14, "learning_rate": 9.564534379639703e-05, "loss": 2.5017, "step": 5680 }, { "epoch": 0.14, "learning_rate": 9.563765181734677e-05, "loss": 2.4287, "step": 5685 }, { "epoch": 0.14, "learning_rate": 9.56299533606698e-05, "loss": 2.5097, "step": 5690 }, { "epoch": 0.14, "learning_rate": 9.562224842745885e-05, "loss": 2.4886, "step": 5695 }, { "epoch": 0.14, "learning_rate": 9.56145370188075e-05, "loss": 2.5349, "step": 5700 }, { "epoch": 0.14, "learning_rate": 9.56068191358103e-05, "loss": 2.3825, "step": 5705 }, { "epoch": 0.14, "learning_rate": 9.559909477956266e-05, "loss": 2.426, "step": 5710 }, { "epoch": 0.14, "learning_rate": 9.559136395116098e-05, "loss": 2.4684, "step": 5715 }, { "epoch": 0.14, "learning_rate": 9.558362665170253e-05, "loss": 2.5706, "step": 5720 }, { "epoch": 0.14, "learning_rate": 9.55758828822855e-05, "loss": 2.4933, "step": 5725 }, { "epoch": 0.14, "learning_rate": 9.556813264400903e-05, "loss": 2.5349, "step": 5730 }, { "epoch": 0.14, "learning_rate": 9.556037593797315e-05, "loss": 2.4992, "step": 5735 }, { "epoch": 0.14, "learning_rate": 9.555261276527881e-05, "loss": 2.4724, "step": 5740 }, { "epoch": 0.14, "learning_rate": 9.554484312702789e-05, "loss": 2.4724, "step": 5745 }, { "epoch": 0.14, "learning_rate": 9.553706702432317e-05, "loss": 2.4747, "step": 5750 }, { "epoch": 0.14, "learning_rate": 9.552928445826838e-05, "loss": 2.47, "step": 5755 }, { "epoch": 0.14, "learning_rate": 9.552149542996814e-05, "loss": 2.5473, "step": 5760 }, { "epoch": 0.14, "learning_rate": 9.551369994052797e-05, "loss": 2.496, "step": 5765 }, { "epoch": 0.14, "learning_rate": 9.550589799105436e-05, "loss": 2.4924, "step": 5770 }, { "epoch": 0.14, "learning_rate": 9.549808958265469e-05, "loss": 2.4857, "step": 5775 }, { "epoch": 0.14, "learning_rate": 9.549027471643723e-05, "loss": 2.4852, "step": 5780 }, { "epoch": 0.14, "learning_rate": 9.54824533935112e-05, "loss": 2.4706, "step": 5785 }, { "epoch": 0.14, "learning_rate": 9.547462561498674e-05, "loss": 2.4418, "step": 5790 }, { "epoch": 0.14, "learning_rate": 9.546679138197487e-05, "loss": 2.4978, "step": 5795 }, { "epoch": 0.14, "learning_rate": 9.545895069558757e-05, "loss": 2.4918, "step": 5800 }, { "epoch": 0.14, "learning_rate": 9.545110355693772e-05, "loss": 2.564, "step": 5805 }, { "epoch": 0.14, "learning_rate": 9.544324996713908e-05, "loss": 2.5412, "step": 5810 }, { "epoch": 0.14, "learning_rate": 9.54353899273064e-05, "loss": 2.448, "step": 5815 }, { "epoch": 0.14, "learning_rate": 9.542752343855526e-05, "loss": 2.5136, "step": 5820 }, { "epoch": 0.14, "learning_rate": 9.541965050200223e-05, "loss": 2.4205, "step": 5825 }, { "epoch": 0.14, "learning_rate": 9.541177111876476e-05, "loss": 2.4635, "step": 5830 }, { "epoch": 0.14, "learning_rate": 9.54038852899612e-05, "loss": 2.4391, "step": 5835 }, { "epoch": 0.14, "learning_rate": 9.539599301671086e-05, "loss": 2.4964, "step": 5840 }, { "epoch": 0.14, "learning_rate": 9.53880943001339e-05, "loss": 2.4901, "step": 5845 }, { "epoch": 0.14, "learning_rate": 9.538018914135148e-05, "loss": 2.4872, "step": 5850 }, { "epoch": 0.14, "learning_rate": 9.537227754148558e-05, "loss": 2.4485, "step": 5855 }, { "epoch": 0.14, "learning_rate": 9.536435950165916e-05, "loss": 2.4598, "step": 5860 }, { "epoch": 0.14, "learning_rate": 9.535643502299609e-05, "loss": 2.4492, "step": 5865 }, { "epoch": 0.14, "learning_rate": 9.534850410662113e-05, "loss": 2.485, "step": 5870 }, { "epoch": 0.14, "learning_rate": 9.534056675365995e-05, "loss": 2.4192, "step": 5875 }, { "epoch": 0.14, "learning_rate": 9.533262296523914e-05, "loss": 2.4788, "step": 5880 }, { "epoch": 0.14, "learning_rate": 9.532467274248625e-05, "loss": 2.44, "step": 5885 }, { "epoch": 0.14, "learning_rate": 9.531671608652968e-05, "loss": 2.5145, "step": 5890 }, { "epoch": 0.14, "learning_rate": 9.530875299849875e-05, "loss": 2.523, "step": 5895 }, { "epoch": 0.14, "learning_rate": 9.530078347952373e-05, "loss": 2.4544, "step": 5900 }, { "epoch": 0.14, "learning_rate": 9.529280753073577e-05, "loss": 2.444, "step": 5905 }, { "epoch": 0.14, "learning_rate": 9.528482515326694e-05, "loss": 2.5115, "step": 5910 }, { "epoch": 0.14, "learning_rate": 9.527683634825026e-05, "loss": 2.4768, "step": 5915 }, { "epoch": 0.14, "learning_rate": 9.526884111681958e-05, "loss": 2.4672, "step": 5920 }, { "epoch": 0.14, "learning_rate": 9.526083946010975e-05, "loss": 2.4966, "step": 5925 }, { "epoch": 0.14, "learning_rate": 9.52528313792565e-05, "loss": 2.4944, "step": 5930 }, { "epoch": 0.14, "learning_rate": 9.524481687539641e-05, "loss": 2.5084, "step": 5935 }, { "epoch": 0.14, "learning_rate": 9.523679594966708e-05, "loss": 2.5155, "step": 5940 }, { "epoch": 0.14, "learning_rate": 9.522876860320694e-05, "loss": 2.4474, "step": 5945 }, { "epoch": 0.14, "learning_rate": 9.522073483715539e-05, "loss": 2.4946, "step": 5950 }, { "epoch": 0.14, "learning_rate": 9.521269465265266e-05, "loss": 2.5672, "step": 5955 }, { "epoch": 0.14, "learning_rate": 9.520464805083999e-05, "loss": 2.5532, "step": 5960 }, { "epoch": 0.14, "learning_rate": 9.519659503285947e-05, "loss": 2.5246, "step": 5965 }, { "epoch": 0.14, "learning_rate": 9.51885355998541e-05, "loss": 2.4172, "step": 5970 }, { "epoch": 0.14, "learning_rate": 9.518046975296781e-05, "loss": 2.5564, "step": 5975 }, { "epoch": 0.14, "learning_rate": 9.517239749334545e-05, "loss": 2.4784, "step": 5980 }, { "epoch": 0.14, "learning_rate": 9.516431882213274e-05, "loss": 2.4851, "step": 5985 }, { "epoch": 0.14, "learning_rate": 9.515623374047635e-05, "loss": 2.4477, "step": 5990 }, { "epoch": 0.14, "learning_rate": 9.514814224952384e-05, "loss": 2.4851, "step": 5995 }, { "epoch": 0.14, "learning_rate": 9.514004435042367e-05, "loss": 2.5293, "step": 6000 }, { "epoch": 0.14, "learning_rate": 9.513194004432525e-05, "loss": 2.484, "step": 6005 }, { "epoch": 0.14, "learning_rate": 9.512382933237887e-05, "loss": 2.4983, "step": 6010 }, { "epoch": 0.14, "learning_rate": 9.51157122157357e-05, "loss": 2.5162, "step": 6015 }, { "epoch": 0.14, "learning_rate": 9.510758869554789e-05, "loss": 2.5572, "step": 6020 }, { "epoch": 0.14, "learning_rate": 9.509945877296845e-05, "loss": 2.4935, "step": 6025 }, { "epoch": 0.14, "learning_rate": 9.509132244915127e-05, "loss": 2.4991, "step": 6030 }, { "epoch": 0.14, "learning_rate": 9.508317972525125e-05, "loss": 2.4512, "step": 6035 }, { "epoch": 0.14, "learning_rate": 9.507503060242408e-05, "loss": 2.4746, "step": 6040 }, { "epoch": 0.14, "learning_rate": 9.506687508182647e-05, "loss": 2.4746, "step": 6045 }, { "epoch": 0.14, "learning_rate": 9.505871316461594e-05, "loss": 2.5079, "step": 6050 }, { "epoch": 0.14, "learning_rate": 9.505054485195097e-05, "loss": 2.4658, "step": 6055 }, { "epoch": 0.14, "learning_rate": 9.504237014499093e-05, "loss": 2.5476, "step": 6060 }, { "epoch": 0.15, "learning_rate": 9.503418904489614e-05, "loss": 2.4722, "step": 6065 }, { "epoch": 0.15, "learning_rate": 9.502600155282775e-05, "loss": 2.4901, "step": 6070 }, { "epoch": 0.15, "learning_rate": 9.501780766994788e-05, "loss": 2.5171, "step": 6075 }, { "epoch": 0.15, "learning_rate": 9.500960739741954e-05, "loss": 2.5318, "step": 6080 }, { "epoch": 0.15, "learning_rate": 9.500140073640662e-05, "loss": 2.5132, "step": 6085 }, { "epoch": 0.15, "learning_rate": 9.499318768807398e-05, "loss": 2.4729, "step": 6090 }, { "epoch": 0.15, "learning_rate": 9.498496825358733e-05, "loss": 2.4804, "step": 6095 }, { "epoch": 0.15, "learning_rate": 9.49767424341133e-05, "loss": 2.4746, "step": 6100 }, { "epoch": 0.15, "learning_rate": 9.496851023081942e-05, "loss": 2.5349, "step": 6105 }, { "epoch": 0.15, "learning_rate": 9.496027164487416e-05, "loss": 2.5244, "step": 6110 }, { "epoch": 0.15, "learning_rate": 9.495202667744685e-05, "loss": 2.5159, "step": 6115 }, { "epoch": 0.15, "learning_rate": 9.494377532970775e-05, "loss": 2.4968, "step": 6120 }, { "epoch": 0.15, "learning_rate": 9.493551760282804e-05, "loss": 2.4821, "step": 6125 }, { "epoch": 0.15, "learning_rate": 9.492725349797979e-05, "loss": 2.5138, "step": 6130 }, { "epoch": 0.15, "learning_rate": 9.491898301633595e-05, "loss": 2.4752, "step": 6135 }, { "epoch": 0.15, "learning_rate": 9.491070615907041e-05, "loss": 2.5299, "step": 6140 }, { "epoch": 0.15, "learning_rate": 9.490242292735798e-05, "loss": 2.5218, "step": 6145 }, { "epoch": 0.15, "learning_rate": 9.489413332237432e-05, "loss": 2.4744, "step": 6150 }, { "epoch": 0.15, "learning_rate": 9.488583734529602e-05, "loss": 2.4523, "step": 6155 }, { "epoch": 0.15, "learning_rate": 9.487753499730059e-05, "loss": 2.5045, "step": 6160 }, { "epoch": 0.15, "learning_rate": 9.486922627956643e-05, "loss": 2.5287, "step": 6165 }, { "epoch": 0.15, "learning_rate": 9.486091119327284e-05, "loss": 2.4027, "step": 6170 }, { "epoch": 0.15, "learning_rate": 9.485258973960006e-05, "loss": 2.4539, "step": 6175 }, { "epoch": 0.15, "learning_rate": 9.484426191972915e-05, "loss": 2.4243, "step": 6180 }, { "epoch": 0.15, "learning_rate": 9.483592773484216e-05, "loss": 2.4588, "step": 6185 }, { "epoch": 0.15, "learning_rate": 9.482758718612204e-05, "loss": 2.4627, "step": 6190 }, { "epoch": 0.15, "learning_rate": 9.481924027475256e-05, "loss": 2.4752, "step": 6195 }, { "epoch": 0.15, "learning_rate": 9.481088700191847e-05, "loss": 2.5805, "step": 6200 }, { "epoch": 0.15, "learning_rate": 9.480252736880542e-05, "loss": 2.5548, "step": 6205 }, { "epoch": 0.15, "learning_rate": 9.479416137659991e-05, "loss": 2.5524, "step": 6210 }, { "epoch": 0.15, "learning_rate": 9.478578902648938e-05, "loss": 2.5022, "step": 6215 }, { "epoch": 0.15, "learning_rate": 9.477741031966219e-05, "loss": 2.5079, "step": 6220 }, { "epoch": 0.15, "learning_rate": 9.476902525730757e-05, "loss": 2.5035, "step": 6225 }, { "epoch": 0.15, "learning_rate": 9.476063384061566e-05, "loss": 2.477, "step": 6230 }, { "epoch": 0.15, "learning_rate": 9.475223607077751e-05, "loss": 2.5525, "step": 6235 }, { "epoch": 0.15, "learning_rate": 9.474383194898507e-05, "loss": 2.5591, "step": 6240 }, { "epoch": 0.15, "learning_rate": 9.473542147643116e-05, "loss": 2.4981, "step": 6245 }, { "epoch": 0.15, "learning_rate": 9.472700465430955e-05, "loss": 2.4895, "step": 6250 }, { "epoch": 0.15, "learning_rate": 9.47185814838149e-05, "loss": 2.4799, "step": 6255 }, { "epoch": 0.15, "learning_rate": 9.471015196614272e-05, "loss": 2.485, "step": 6260 }, { "epoch": 0.15, "learning_rate": 9.470171610248952e-05, "loss": 2.5017, "step": 6265 }, { "epoch": 0.15, "learning_rate": 9.469327389405262e-05, "loss": 2.47, "step": 6270 }, { "epoch": 0.15, "learning_rate": 9.468482534203026e-05, "loss": 2.4643, "step": 6275 }, { "epoch": 0.15, "learning_rate": 9.467637044762163e-05, "loss": 2.4563, "step": 6280 }, { "epoch": 0.15, "learning_rate": 9.466790921202677e-05, "loss": 2.4752, "step": 6285 }, { "epoch": 0.15, "learning_rate": 9.465944163644663e-05, "loss": 2.5128, "step": 6290 }, { "epoch": 0.15, "learning_rate": 9.465096772208303e-05, "loss": 2.4968, "step": 6295 }, { "epoch": 0.15, "learning_rate": 9.464248747013879e-05, "loss": 2.4854, "step": 6300 }, { "epoch": 0.15, "learning_rate": 9.463400088181753e-05, "loss": 2.5298, "step": 6305 }, { "epoch": 0.15, "learning_rate": 9.46255079583238e-05, "loss": 2.5429, "step": 6310 }, { "epoch": 0.15, "learning_rate": 9.461700870086305e-05, "loss": 2.552, "step": 6315 }, { "epoch": 0.15, "learning_rate": 9.460850311064165e-05, "loss": 2.4691, "step": 6320 }, { "epoch": 0.15, "learning_rate": 9.459999118886682e-05, "loss": 2.5227, "step": 6325 }, { "epoch": 0.15, "learning_rate": 9.459147293674673e-05, "loss": 2.4558, "step": 6330 }, { "epoch": 0.15, "learning_rate": 9.458294835549041e-05, "loss": 2.4902, "step": 6335 }, { "epoch": 0.15, "learning_rate": 9.457441744630782e-05, "loss": 2.4672, "step": 6340 }, { "epoch": 0.15, "learning_rate": 9.456588021040981e-05, "loss": 2.5164, "step": 6345 }, { "epoch": 0.15, "learning_rate": 9.45573366490081e-05, "loss": 2.4312, "step": 6350 }, { "epoch": 0.15, "learning_rate": 9.454878676331535e-05, "loss": 2.529, "step": 6355 }, { "epoch": 0.15, "learning_rate": 9.454023055454507e-05, "loss": 2.4967, "step": 6360 }, { "epoch": 0.15, "learning_rate": 9.453166802391172e-05, "loss": 2.4686, "step": 6365 }, { "epoch": 0.15, "learning_rate": 9.452309917263063e-05, "loss": 2.517, "step": 6370 }, { "epoch": 0.15, "learning_rate": 9.4514524001918e-05, "loss": 2.5041, "step": 6375 }, { "epoch": 0.15, "learning_rate": 9.450594251299098e-05, "loss": 2.5258, "step": 6380 }, { "epoch": 0.15, "learning_rate": 9.449735470706762e-05, "loss": 2.4855, "step": 6385 }, { "epoch": 0.15, "learning_rate": 9.448876058536678e-05, "loss": 2.5088, "step": 6390 }, { "epoch": 0.15, "learning_rate": 9.448016014910829e-05, "loss": 2.4753, "step": 6395 }, { "epoch": 0.15, "learning_rate": 9.44715533995129e-05, "loss": 2.5304, "step": 6400 }, { "epoch": 0.15, "learning_rate": 9.446294033780219e-05, "loss": 2.4342, "step": 6405 }, { "epoch": 0.15, "learning_rate": 9.445432096519866e-05, "loss": 2.4991, "step": 6410 }, { "epoch": 0.15, "learning_rate": 9.444569528292572e-05, "loss": 2.4978, "step": 6415 }, { "epoch": 0.15, "learning_rate": 9.443706329220765e-05, "loss": 2.5253, "step": 6420 }, { "epoch": 0.15, "learning_rate": 9.442842499426967e-05, "loss": 2.4935, "step": 6425 }, { "epoch": 0.15, "learning_rate": 9.441978039033783e-05, "loss": 2.5185, "step": 6430 }, { "epoch": 0.15, "learning_rate": 9.441112948163914e-05, "loss": 2.5608, "step": 6435 }, { "epoch": 0.15, "learning_rate": 9.440247226940145e-05, "loss": 2.5715, "step": 6440 }, { "epoch": 0.15, "learning_rate": 9.439380875485355e-05, "loss": 2.4395, "step": 6445 }, { "epoch": 0.15, "learning_rate": 9.438513893922509e-05, "loss": 2.5109, "step": 6450 }, { "epoch": 0.15, "learning_rate": 9.437646282374666e-05, "loss": 2.4972, "step": 6455 }, { "epoch": 0.15, "learning_rate": 9.436778040964966e-05, "loss": 2.4654, "step": 6460 }, { "epoch": 0.15, "learning_rate": 9.435909169816648e-05, "loss": 2.4791, "step": 6465 }, { "epoch": 0.15, "learning_rate": 9.435039669053036e-05, "loss": 2.5234, "step": 6470 }, { "epoch": 0.15, "learning_rate": 9.434169538797542e-05, "loss": 2.4827, "step": 6475 }, { "epoch": 0.16, "learning_rate": 9.433298779173669e-05, "loss": 2.506, "step": 6480 }, { "epoch": 0.16, "learning_rate": 9.432427390305009e-05, "loss": 2.591, "step": 6485 }, { "epoch": 0.16, "learning_rate": 9.431555372315243e-05, "loss": 2.505, "step": 6490 }, { "epoch": 0.16, "learning_rate": 9.430682725328144e-05, "loss": 2.5136, "step": 6495 }, { "epoch": 0.16, "learning_rate": 9.42980944946757e-05, "loss": 2.4271, "step": 6500 }, { "epoch": 0.16, "learning_rate": 9.428935544857471e-05, "loss": 2.508, "step": 6505 }, { "epoch": 0.16, "learning_rate": 9.428061011621886e-05, "loss": 2.5428, "step": 6510 }, { "epoch": 0.16, "learning_rate": 9.427185849884941e-05, "loss": 2.5106, "step": 6515 }, { "epoch": 0.16, "learning_rate": 9.426310059770855e-05, "loss": 2.4738, "step": 6520 }, { "epoch": 0.16, "learning_rate": 9.425433641403934e-05, "loss": 2.434, "step": 6525 }, { "epoch": 0.16, "learning_rate": 9.424556594908571e-05, "loss": 2.5146, "step": 6530 }, { "epoch": 0.16, "learning_rate": 9.423678920409254e-05, "loss": 2.4578, "step": 6535 }, { "epoch": 0.16, "learning_rate": 9.422800618030554e-05, "loss": 2.4686, "step": 6540 }, { "epoch": 0.16, "learning_rate": 9.421921687897135e-05, "loss": 2.5824, "step": 6545 }, { "epoch": 0.16, "learning_rate": 9.421042130133748e-05, "loss": 2.5007, "step": 6550 }, { "epoch": 0.16, "learning_rate": 9.420161944865234e-05, "loss": 2.4658, "step": 6555 }, { "epoch": 0.16, "learning_rate": 9.419281132216525e-05, "loss": 2.458, "step": 6560 }, { "epoch": 0.16, "learning_rate": 9.418399692312636e-05, "loss": 2.4873, "step": 6565 }, { "epoch": 0.16, "learning_rate": 9.417517625278678e-05, "loss": 2.4934, "step": 6570 }, { "epoch": 0.16, "learning_rate": 9.416634931239848e-05, "loss": 2.4897, "step": 6575 }, { "epoch": 0.16, "learning_rate": 9.415751610321432e-05, "loss": 2.4545, "step": 6580 }, { "epoch": 0.16, "learning_rate": 9.414867662648804e-05, "loss": 2.4414, "step": 6585 }, { "epoch": 0.16, "learning_rate": 9.413983088347429e-05, "loss": 2.3902, "step": 6590 }, { "epoch": 0.16, "learning_rate": 9.41309788754286e-05, "loss": 2.4552, "step": 6595 }, { "epoch": 0.16, "learning_rate": 9.412212060360739e-05, "loss": 2.5145, "step": 6600 }, { "epoch": 0.16, "learning_rate": 9.411325606926795e-05, "loss": 2.5541, "step": 6605 }, { "epoch": 0.16, "learning_rate": 9.41043852736685e-05, "loss": 2.4573, "step": 6610 }, { "epoch": 0.16, "learning_rate": 9.409550821806809e-05, "loss": 2.5357, "step": 6615 }, { "epoch": 0.16, "learning_rate": 9.408662490372675e-05, "loss": 2.5213, "step": 6620 }, { "epoch": 0.16, "learning_rate": 9.40777353319053e-05, "loss": 2.4156, "step": 6625 }, { "epoch": 0.16, "learning_rate": 9.406883950386551e-05, "loss": 2.5143, "step": 6630 }, { "epoch": 0.16, "learning_rate": 9.405993742087e-05, "loss": 2.4884, "step": 6635 }, { "epoch": 0.16, "learning_rate": 9.405102908418231e-05, "loss": 2.5532, "step": 6640 }, { "epoch": 0.16, "learning_rate": 9.404211449506685e-05, "loss": 2.4653, "step": 6645 }, { "epoch": 0.16, "learning_rate": 9.403319365478893e-05, "loss": 2.5058, "step": 6650 }, { "epoch": 0.16, "learning_rate": 9.402426656461472e-05, "loss": 2.4974, "step": 6655 }, { "epoch": 0.16, "learning_rate": 9.40153332258113e-05, "loss": 2.4202, "step": 6660 }, { "epoch": 0.16, "learning_rate": 9.400639363964666e-05, "loss": 2.4672, "step": 6665 }, { "epoch": 0.16, "learning_rate": 9.399744780738959e-05, "loss": 2.5259, "step": 6670 }, { "epoch": 0.16, "learning_rate": 9.39884957303099e-05, "loss": 2.4977, "step": 6675 }, { "epoch": 0.16, "learning_rate": 9.397953740967815e-05, "loss": 2.4714, "step": 6680 }, { "epoch": 0.16, "learning_rate": 9.39705728467659e-05, "loss": 2.4502, "step": 6685 }, { "epoch": 0.16, "learning_rate": 9.39616020428455e-05, "loss": 2.3898, "step": 6690 }, { "epoch": 0.16, "learning_rate": 9.395262499919026e-05, "loss": 2.4355, "step": 6695 }, { "epoch": 0.16, "learning_rate": 9.394364171707433e-05, "loss": 2.4401, "step": 6700 }, { "epoch": 0.16, "learning_rate": 9.393465219777276e-05, "loss": 2.5204, "step": 6705 }, { "epoch": 0.16, "learning_rate": 9.39256564425615e-05, "loss": 2.5013, "step": 6710 }, { "epoch": 0.16, "learning_rate": 9.391665445271736e-05, "loss": 2.4749, "step": 6715 }, { "epoch": 0.16, "learning_rate": 9.390764622951805e-05, "loss": 2.5599, "step": 6720 }, { "epoch": 0.16, "learning_rate": 9.389863177424216e-05, "loss": 2.4546, "step": 6725 }, { "epoch": 0.16, "learning_rate": 9.388961108816918e-05, "loss": 2.4744, "step": 6730 }, { "epoch": 0.16, "learning_rate": 9.388058417257945e-05, "loss": 2.5388, "step": 6735 }, { "epoch": 0.16, "learning_rate": 9.387155102875424e-05, "loss": 2.5111, "step": 6740 }, { "epoch": 0.16, "learning_rate": 9.386251165797564e-05, "loss": 2.4816, "step": 6745 }, { "epoch": 0.16, "learning_rate": 9.385346606152669e-05, "loss": 2.431, "step": 6750 }, { "epoch": 0.16, "learning_rate": 9.384441424069129e-05, "loss": 2.5305, "step": 6755 }, { "epoch": 0.16, "learning_rate": 9.383535619675419e-05, "loss": 2.4921, "step": 6760 }, { "epoch": 0.16, "learning_rate": 9.382629193100108e-05, "loss": 2.5046, "step": 6765 }, { "epoch": 0.16, "learning_rate": 9.381722144471849e-05, "loss": 2.5181, "step": 6770 }, { "epoch": 0.16, "learning_rate": 9.380814473919388e-05, "loss": 2.4643, "step": 6775 }, { "epoch": 0.16, "learning_rate": 9.379906181571551e-05, "loss": 2.4941, "step": 6780 }, { "epoch": 0.16, "learning_rate": 9.37899726755726e-05, "loss": 2.5717, "step": 6785 }, { "epoch": 0.16, "learning_rate": 9.378087732005525e-05, "loss": 2.4194, "step": 6790 }, { "epoch": 0.16, "learning_rate": 9.37717757504544e-05, "loss": 2.4298, "step": 6795 }, { "epoch": 0.16, "learning_rate": 9.376266796806187e-05, "loss": 2.4888, "step": 6800 }, { "epoch": 0.16, "learning_rate": 9.37535539741704e-05, "loss": 2.4854, "step": 6805 }, { "epoch": 0.16, "learning_rate": 9.37444337700736e-05, "loss": 2.5155, "step": 6810 }, { "epoch": 0.16, "learning_rate": 9.373530735706596e-05, "loss": 2.4731, "step": 6815 }, { "epoch": 0.16, "learning_rate": 9.372617473644283e-05, "loss": 2.4436, "step": 6820 }, { "epoch": 0.16, "learning_rate": 9.371703590950045e-05, "loss": 2.4534, "step": 6825 }, { "epoch": 0.16, "learning_rate": 9.370789087753597e-05, "loss": 2.4445, "step": 6830 }, { "epoch": 0.16, "learning_rate": 9.369873964184739e-05, "loss": 2.5737, "step": 6835 }, { "epoch": 0.16, "learning_rate": 9.36895822037336e-05, "loss": 2.5327, "step": 6840 }, { "epoch": 0.16, "learning_rate": 9.368041856449438e-05, "loss": 2.4861, "step": 6845 }, { "epoch": 0.16, "learning_rate": 9.367124872543038e-05, "loss": 2.5295, "step": 6850 }, { "epoch": 0.16, "learning_rate": 9.36620726878431e-05, "loss": 2.5064, "step": 6855 }, { "epoch": 0.16, "learning_rate": 9.3652890453035e-05, "loss": 2.4655, "step": 6860 }, { "epoch": 0.16, "learning_rate": 9.364370202230931e-05, "loss": 2.4859, "step": 6865 }, { "epoch": 0.16, "learning_rate": 9.363450739697027e-05, "loss": 2.4048, "step": 6870 }, { "epoch": 0.16, "learning_rate": 9.362530657832287e-05, "loss": 2.4709, "step": 6875 }, { "epoch": 0.16, "learning_rate": 9.361609956767307e-05, "loss": 2.4846, "step": 6880 }, { "epoch": 0.16, "learning_rate": 9.360688636632766e-05, "loss": 2.5427, "step": 6885 }, { "epoch": 0.16, "learning_rate": 9.359766697559433e-05, "loss": 2.4461, "step": 6890 }, { "epoch": 0.16, "learning_rate": 9.358844139678163e-05, "loss": 2.4344, "step": 6895 }, { "epoch": 0.17, "learning_rate": 9.357920963119904e-05, "loss": 2.4761, "step": 6900 }, { "epoch": 0.17, "learning_rate": 9.356997168015684e-05, "loss": 2.5248, "step": 6905 }, { "epoch": 0.17, "learning_rate": 9.356072754496625e-05, "loss": 2.5198, "step": 6910 }, { "epoch": 0.17, "learning_rate": 9.355147722693933e-05, "loss": 2.4903, "step": 6915 }, { "epoch": 0.17, "learning_rate": 9.354222072738905e-05, "loss": 2.4412, "step": 6920 }, { "epoch": 0.17, "learning_rate": 9.353295804762925e-05, "loss": 2.4672, "step": 6925 }, { "epoch": 0.17, "learning_rate": 9.35236891889746e-05, "loss": 2.4953, "step": 6930 }, { "epoch": 0.17, "learning_rate": 9.351441415274072e-05, "loss": 2.5089, "step": 6935 }, { "epoch": 0.17, "learning_rate": 9.350513294024404e-05, "loss": 2.3885, "step": 6940 }, { "epoch": 0.17, "learning_rate": 9.349584555280193e-05, "loss": 2.5169, "step": 6945 }, { "epoch": 0.17, "learning_rate": 9.34865519917326e-05, "loss": 2.5131, "step": 6950 }, { "epoch": 0.17, "learning_rate": 9.347725225835511e-05, "loss": 2.4516, "step": 6955 }, { "epoch": 0.17, "learning_rate": 9.346794635398947e-05, "loss": 2.415, "step": 6960 }, { "epoch": 0.17, "learning_rate": 9.345863427995649e-05, "loss": 2.5046, "step": 6965 }, { "epoch": 0.17, "learning_rate": 9.344931603757792e-05, "loss": 2.5092, "step": 6970 }, { "epoch": 0.17, "learning_rate": 9.343999162817631e-05, "loss": 2.4957, "step": 6975 }, { "epoch": 0.17, "learning_rate": 9.343066105307516e-05, "loss": 2.4719, "step": 6980 }, { "epoch": 0.17, "learning_rate": 9.342132431359883e-05, "loss": 2.4766, "step": 6985 }, { "epoch": 0.17, "learning_rate": 9.34119814110725e-05, "loss": 2.5376, "step": 6990 }, { "epoch": 0.17, "learning_rate": 9.34026323468223e-05, "loss": 2.5364, "step": 6995 }, { "epoch": 0.17, "learning_rate": 9.339327712217517e-05, "loss": 2.3671, "step": 7000 }, { "epoch": 0.17, "learning_rate": 9.338391573845898e-05, "loss": 2.4661, "step": 7005 }, { "epoch": 0.17, "learning_rate": 9.337454819700242e-05, "loss": 2.5002, "step": 7010 }, { "epoch": 0.17, "learning_rate": 9.33651744991351e-05, "loss": 2.4751, "step": 7015 }, { "epoch": 0.17, "learning_rate": 9.335579464618748e-05, "loss": 2.5136, "step": 7020 }, { "epoch": 0.17, "learning_rate": 9.33464086394909e-05, "loss": 2.5231, "step": 7025 }, { "epoch": 0.17, "learning_rate": 9.333701648037759e-05, "loss": 2.4055, "step": 7030 }, { "epoch": 0.17, "learning_rate": 9.332761817018058e-05, "loss": 2.4168, "step": 7035 }, { "epoch": 0.17, "learning_rate": 9.33182137102339e-05, "loss": 2.5616, "step": 7040 }, { "epoch": 0.17, "learning_rate": 9.330880310187234e-05, "loss": 2.5228, "step": 7045 }, { "epoch": 0.17, "learning_rate": 9.32993863464316e-05, "loss": 2.5169, "step": 7050 }, { "epoch": 0.17, "learning_rate": 9.328996344524827e-05, "loss": 2.49, "step": 7055 }, { "epoch": 0.17, "learning_rate": 9.328053439965982e-05, "loss": 2.4787, "step": 7060 }, { "epoch": 0.17, "learning_rate": 9.327109921100453e-05, "loss": 2.4787, "step": 7065 }, { "epoch": 0.17, "learning_rate": 9.326165788062162e-05, "loss": 2.4734, "step": 7070 }, { "epoch": 0.17, "learning_rate": 9.325221040985114e-05, "loss": 2.5227, "step": 7075 }, { "epoch": 0.17, "learning_rate": 9.324275680003404e-05, "loss": 2.5417, "step": 7080 }, { "epoch": 0.17, "learning_rate": 9.323329705251214e-05, "loss": 2.4619, "step": 7085 }, { "epoch": 0.17, "learning_rate": 9.322383116862809e-05, "loss": 2.441, "step": 7090 }, { "epoch": 0.17, "learning_rate": 9.321435914972544e-05, "loss": 2.4467, "step": 7095 }, { "epoch": 0.17, "learning_rate": 9.320488099714863e-05, "loss": 2.5007, "step": 7100 }, { "epoch": 0.17, "learning_rate": 9.319539671224294e-05, "loss": 2.5065, "step": 7105 }, { "epoch": 0.17, "learning_rate": 9.318590629635453e-05, "loss": 2.4478, "step": 7110 }, { "epoch": 0.17, "learning_rate": 9.317640975083044e-05, "loss": 2.4322, "step": 7115 }, { "epoch": 0.17, "learning_rate": 9.316690707701857e-05, "loss": 2.4505, "step": 7120 }, { "epoch": 0.17, "learning_rate": 9.315739827626768e-05, "loss": 2.4969, "step": 7125 }, { "epoch": 0.17, "learning_rate": 9.314788334992743e-05, "loss": 2.506, "step": 7130 }, { "epoch": 0.17, "learning_rate": 9.313836229934833e-05, "loss": 2.482, "step": 7135 }, { "epoch": 0.17, "learning_rate": 9.312883512588176e-05, "loss": 2.4459, "step": 7140 }, { "epoch": 0.17, "learning_rate": 9.311930183087994e-05, "loss": 2.4739, "step": 7145 }, { "epoch": 0.17, "learning_rate": 9.3109762415696e-05, "loss": 2.5147, "step": 7150 }, { "epoch": 0.17, "learning_rate": 9.310021688168396e-05, "loss": 2.5716, "step": 7155 }, { "epoch": 0.17, "learning_rate": 9.309066523019862e-05, "loss": 2.5161, "step": 7160 }, { "epoch": 0.17, "learning_rate": 9.308110746259577e-05, "loss": 2.494, "step": 7165 }, { "epoch": 0.17, "learning_rate": 9.307154358023195e-05, "loss": 2.4814, "step": 7170 }, { "epoch": 0.17, "learning_rate": 9.306197358446463e-05, "loss": 2.4402, "step": 7175 }, { "epoch": 0.17, "learning_rate": 9.305239747665215e-05, "loss": 2.4696, "step": 7180 }, { "epoch": 0.17, "learning_rate": 9.304281525815369e-05, "loss": 2.5326, "step": 7185 }, { "epoch": 0.17, "learning_rate": 9.303322693032932e-05, "loss": 2.4779, "step": 7190 }, { "epoch": 0.17, "learning_rate": 9.302363249453999e-05, "loss": 2.5568, "step": 7195 }, { "epoch": 0.17, "learning_rate": 9.301403195214744e-05, "loss": 2.4991, "step": 7200 }, { "epoch": 0.17, "learning_rate": 9.300442530451439e-05, "loss": 2.4326, "step": 7205 }, { "epoch": 0.17, "learning_rate": 9.299481255300433e-05, "loss": 2.4732, "step": 7210 }, { "epoch": 0.17, "learning_rate": 9.298519369898166e-05, "loss": 2.5086, "step": 7215 }, { "epoch": 0.17, "learning_rate": 9.297556874381169e-05, "loss": 2.5194, "step": 7220 }, { "epoch": 0.17, "learning_rate": 9.296593768886048e-05, "loss": 2.5358, "step": 7225 }, { "epoch": 0.17, "learning_rate": 9.295630053549507e-05, "loss": 2.4644, "step": 7230 }, { "epoch": 0.17, "learning_rate": 9.294665728508332e-05, "loss": 2.4562, "step": 7235 }, { "epoch": 0.17, "learning_rate": 9.293700793899391e-05, "loss": 2.4914, "step": 7240 }, { "epoch": 0.17, "learning_rate": 9.292735249859647e-05, "loss": 2.5238, "step": 7245 }, { "epoch": 0.17, "learning_rate": 9.291769096526144e-05, "loss": 2.4656, "step": 7250 }, { "epoch": 0.17, "learning_rate": 9.290802334036015e-05, "loss": 2.5143, "step": 7255 }, { "epoch": 0.17, "learning_rate": 9.289834962526478e-05, "loss": 2.4365, "step": 7260 }, { "epoch": 0.17, "learning_rate": 9.288866982134837e-05, "loss": 2.4832, "step": 7265 }, { "epoch": 0.17, "learning_rate": 9.287898392998484e-05, "loss": 2.4857, "step": 7270 }, { "epoch": 0.17, "learning_rate": 9.286929195254895e-05, "loss": 2.5078, "step": 7275 }, { "epoch": 0.17, "learning_rate": 9.285959389041638e-05, "loss": 2.4741, "step": 7280 }, { "epoch": 0.17, "learning_rate": 9.28498897449636e-05, "loss": 2.4569, "step": 7285 }, { "epoch": 0.17, "learning_rate": 9.284017951756798e-05, "loss": 2.4637, "step": 7290 }, { "epoch": 0.17, "learning_rate": 9.283046320960777e-05, "loss": 2.493, "step": 7295 }, { "epoch": 0.17, "learning_rate": 9.282074082246206e-05, "loss": 2.497, "step": 7300 }, { "epoch": 0.17, "learning_rate": 9.281101235751078e-05, "loss": 2.4544, "step": 7305 }, { "epoch": 0.17, "learning_rate": 9.28012778161348e-05, "loss": 2.469, "step": 7310 }, { "epoch": 0.18, "learning_rate": 9.279153719971574e-05, "loss": 2.4044, "step": 7315 }, { "epoch": 0.18, "learning_rate": 9.278179050963619e-05, "loss": 2.5324, "step": 7320 }, { "epoch": 0.18, "learning_rate": 9.277203774727954e-05, "loss": 2.5707, "step": 7325 }, { "epoch": 0.18, "learning_rate": 9.276227891403006e-05, "loss": 2.4679, "step": 7330 }, { "epoch": 0.18, "learning_rate": 9.27525140112729e-05, "loss": 2.4735, "step": 7335 }, { "epoch": 0.18, "learning_rate": 9.274274304039401e-05, "loss": 2.4855, "step": 7340 }, { "epoch": 0.18, "learning_rate": 9.273296600278027e-05, "loss": 2.4827, "step": 7345 }, { "epoch": 0.18, "learning_rate": 9.27231828998194e-05, "loss": 2.462, "step": 7350 }, { "epoch": 0.18, "learning_rate": 9.271339373289995e-05, "loss": 2.4776, "step": 7355 }, { "epoch": 0.18, "learning_rate": 9.270359850341136e-05, "loss": 2.4366, "step": 7360 }, { "epoch": 0.18, "learning_rate": 9.269379721274395e-05, "loss": 2.5186, "step": 7365 }, { "epoch": 0.18, "learning_rate": 9.268398986228888e-05, "loss": 2.5187, "step": 7370 }, { "epoch": 0.18, "learning_rate": 9.267417645343812e-05, "loss": 2.4465, "step": 7375 }, { "epoch": 0.18, "learning_rate": 9.266435698758459e-05, "loss": 2.5156, "step": 7380 }, { "epoch": 0.18, "learning_rate": 9.265453146612199e-05, "loss": 2.5864, "step": 7385 }, { "epoch": 0.18, "learning_rate": 9.264469989044495e-05, "loss": 2.4538, "step": 7390 }, { "epoch": 0.18, "learning_rate": 9.263486226194892e-05, "loss": 2.4928, "step": 7395 }, { "epoch": 0.18, "learning_rate": 9.262501858203019e-05, "loss": 2.5105, "step": 7400 }, { "epoch": 0.18, "learning_rate": 9.261516885208594e-05, "loss": 2.4471, "step": 7405 }, { "epoch": 0.18, "learning_rate": 9.260531307351421e-05, "loss": 2.4614, "step": 7410 }, { "epoch": 0.18, "learning_rate": 9.25954512477139e-05, "loss": 2.4913, "step": 7415 }, { "epoch": 0.18, "learning_rate": 9.258558337608474e-05, "loss": 2.5182, "step": 7420 }, { "epoch": 0.18, "learning_rate": 9.257570946002733e-05, "loss": 2.4831, "step": 7425 }, { "epoch": 0.18, "learning_rate": 9.256582950094317e-05, "loss": 2.4535, "step": 7430 }, { "epoch": 0.18, "learning_rate": 9.255594350023455e-05, "loss": 2.4815, "step": 7435 }, { "epoch": 0.18, "learning_rate": 9.254605145930466e-05, "loss": 2.5329, "step": 7440 }, { "epoch": 0.18, "learning_rate": 9.253615337955752e-05, "loss": 2.444, "step": 7445 }, { "epoch": 0.18, "learning_rate": 9.252624926239806e-05, "loss": 2.5284, "step": 7450 }, { "epoch": 0.18, "learning_rate": 9.2516339109232e-05, "loss": 2.5033, "step": 7455 }, { "epoch": 0.18, "learning_rate": 9.250642292146597e-05, "loss": 2.3948, "step": 7460 }, { "epoch": 0.18, "learning_rate": 9.24965007005074e-05, "loss": 2.5964, "step": 7465 }, { "epoch": 0.18, "learning_rate": 9.248657244776464e-05, "loss": 2.4273, "step": 7470 }, { "epoch": 0.18, "learning_rate": 9.247663816464687e-05, "loss": 2.5048, "step": 7475 }, { "epoch": 0.18, "learning_rate": 9.24666978525641e-05, "loss": 2.47, "step": 7480 }, { "epoch": 0.18, "learning_rate": 9.245675151292724e-05, "loss": 2.4436, "step": 7485 }, { "epoch": 0.18, "learning_rate": 9.244679914714802e-05, "loss": 2.5028, "step": 7490 }, { "epoch": 0.18, "learning_rate": 9.243684075663905e-05, "loss": 2.5564, "step": 7495 }, { "epoch": 0.18, "learning_rate": 9.242687634281377e-05, "loss": 2.4099, "step": 7500 }, { "epoch": 0.18, "learning_rate": 9.24169059070865e-05, "loss": 2.5329, "step": 7505 }, { "epoch": 0.18, "learning_rate": 9.240692945087241e-05, "loss": 2.4834, "step": 7510 }, { "epoch": 0.18, "learning_rate": 9.239694697558749e-05, "loss": 2.4643, "step": 7515 }, { "epoch": 0.18, "learning_rate": 9.238695848264866e-05, "loss": 2.5003, "step": 7520 }, { "epoch": 0.18, "learning_rate": 9.237696397347359e-05, "loss": 2.4671, "step": 7525 }, { "epoch": 0.18, "learning_rate": 9.236696344948091e-05, "loss": 2.511, "step": 7530 }, { "epoch": 0.18, "learning_rate": 9.235695691209003e-05, "loss": 2.5112, "step": 7535 }, { "epoch": 0.18, "learning_rate": 9.234694436272124e-05, "loss": 2.428, "step": 7540 }, { "epoch": 0.18, "learning_rate": 9.233692580279569e-05, "loss": 2.4195, "step": 7545 }, { "epoch": 0.18, "learning_rate": 9.232690123373537e-05, "loss": 2.4561, "step": 7550 }, { "epoch": 0.18, "learning_rate": 9.231687065696313e-05, "loss": 2.4407, "step": 7555 }, { "epoch": 0.18, "learning_rate": 9.230683407390266e-05, "loss": 2.4747, "step": 7560 }, { "epoch": 0.18, "learning_rate": 9.229679148597852e-05, "loss": 2.4944, "step": 7565 }, { "epoch": 0.18, "learning_rate": 9.228674289461611e-05, "loss": 2.5038, "step": 7570 }, { "epoch": 0.18, "learning_rate": 9.22766883012417e-05, "loss": 2.5131, "step": 7575 }, { "epoch": 0.18, "learning_rate": 9.226662770728238e-05, "loss": 2.4518, "step": 7580 }, { "epoch": 0.18, "learning_rate": 9.225656111416612e-05, "loss": 2.4879, "step": 7585 }, { "epoch": 0.18, "learning_rate": 9.224648852332174e-05, "loss": 2.4771, "step": 7590 }, { "epoch": 0.18, "learning_rate": 9.223640993617889e-05, "loss": 2.4611, "step": 7595 }, { "epoch": 0.18, "learning_rate": 9.22263253541681e-05, "loss": 2.4781, "step": 7600 }, { "epoch": 0.18, "learning_rate": 9.221623477872073e-05, "loss": 2.4722, "step": 7605 }, { "epoch": 0.18, "learning_rate": 9.220613821126897e-05, "loss": 2.4884, "step": 7610 }, { "epoch": 0.18, "learning_rate": 9.219603565324592e-05, "loss": 2.4741, "step": 7615 }, { "epoch": 0.18, "learning_rate": 9.218592710608552e-05, "loss": 2.3971, "step": 7620 }, { "epoch": 0.18, "learning_rate": 9.217581257122246e-05, "loss": 2.4845, "step": 7625 }, { "epoch": 0.18, "learning_rate": 9.216569205009241e-05, "loss": 2.4663, "step": 7630 }, { "epoch": 0.18, "learning_rate": 9.215556554413183e-05, "loss": 2.4711, "step": 7635 }, { "epoch": 0.18, "learning_rate": 9.214543305477802e-05, "loss": 2.4626, "step": 7640 }, { "epoch": 0.18, "learning_rate": 9.213529458346919e-05, "loss": 2.4969, "step": 7645 }, { "epoch": 0.18, "learning_rate": 9.21251501316443e-05, "loss": 2.4728, "step": 7650 }, { "epoch": 0.18, "learning_rate": 9.211499970074323e-05, "loss": 2.5057, "step": 7655 }, { "epoch": 0.18, "learning_rate": 9.21048432922067e-05, "loss": 2.4826, "step": 7660 }, { "epoch": 0.18, "learning_rate": 9.209468090747628e-05, "loss": 2.4777, "step": 7665 }, { "epoch": 0.18, "learning_rate": 9.208451254799436e-05, "loss": 2.4287, "step": 7670 }, { "epoch": 0.18, "learning_rate": 9.207433821520419e-05, "loss": 2.4894, "step": 7675 }, { "epoch": 0.18, "learning_rate": 9.20641579105499e-05, "loss": 2.5058, "step": 7680 }, { "epoch": 0.18, "learning_rate": 9.205397163547642e-05, "loss": 2.5383, "step": 7685 }, { "epoch": 0.18, "learning_rate": 9.204377939142957e-05, "loss": 2.424, "step": 7690 }, { "epoch": 0.18, "learning_rate": 9.203358117985597e-05, "loss": 2.4699, "step": 7695 }, { "epoch": 0.18, "learning_rate": 9.202337700220313e-05, "loss": 2.4972, "step": 7700 }, { "epoch": 0.18, "learning_rate": 9.201316685991938e-05, "loss": 2.5366, "step": 7705 }, { "epoch": 0.18, "learning_rate": 9.200295075445393e-05, "loss": 2.4646, "step": 7710 }, { "epoch": 0.18, "learning_rate": 9.199272868725678e-05, "loss": 2.4867, "step": 7715 }, { "epoch": 0.18, "learning_rate": 9.198250065977885e-05, "loss": 2.5465, "step": 7720 }, { "epoch": 0.18, "learning_rate": 9.197226667347184e-05, "loss": 2.5727, "step": 7725 }, { "epoch": 0.18, "learning_rate": 9.196202672978832e-05, "loss": 2.4845, "step": 7730 }, { "epoch": 0.19, "learning_rate": 9.195178083018171e-05, "loss": 2.5066, "step": 7735 }, { "epoch": 0.19, "learning_rate": 9.194152897610628e-05, "loss": 2.4791, "step": 7740 }, { "epoch": 0.19, "learning_rate": 9.193127116901712e-05, "loss": 2.5046, "step": 7745 }, { "epoch": 0.19, "learning_rate": 9.19210074103702e-05, "loss": 2.474, "step": 7750 }, { "epoch": 0.19, "learning_rate": 9.19107377016223e-05, "loss": 2.5034, "step": 7755 }, { "epoch": 0.19, "learning_rate": 9.190046204423108e-05, "loss": 2.4685, "step": 7760 }, { "epoch": 0.19, "learning_rate": 9.189018043965503e-05, "loss": 2.4882, "step": 7765 }, { "epoch": 0.19, "learning_rate": 9.187989288935346e-05, "loss": 2.5231, "step": 7770 }, { "epoch": 0.19, "learning_rate": 9.186959939478654e-05, "loss": 2.4408, "step": 7775 }, { "epoch": 0.19, "learning_rate": 9.185929995741531e-05, "loss": 2.5347, "step": 7780 }, { "epoch": 0.19, "learning_rate": 9.184899457870163e-05, "loss": 2.4725, "step": 7785 }, { "epoch": 0.19, "learning_rate": 9.183868326010818e-05, "loss": 2.4958, "step": 7790 }, { "epoch": 0.19, "learning_rate": 9.182836600309853e-05, "loss": 2.4448, "step": 7795 }, { "epoch": 0.19, "learning_rate": 9.181804280913706e-05, "loss": 2.4121, "step": 7800 }, { "epoch": 0.19, "learning_rate": 9.1807713679689e-05, "loss": 2.5278, "step": 7805 }, { "epoch": 0.19, "learning_rate": 9.179737861622044e-05, "loss": 2.4912, "step": 7810 }, { "epoch": 0.19, "learning_rate": 9.178703762019829e-05, "loss": 2.4844, "step": 7815 }, { "epoch": 0.19, "learning_rate": 9.177669069309031e-05, "loss": 2.4393, "step": 7820 }, { "epoch": 0.19, "learning_rate": 9.17663378363651e-05, "loss": 2.5147, "step": 7825 }, { "epoch": 0.19, "learning_rate": 9.175597905149212e-05, "loss": 2.4941, "step": 7830 }, { "epoch": 0.19, "learning_rate": 9.174561433994163e-05, "loss": 2.4804, "step": 7835 }, { "epoch": 0.19, "learning_rate": 9.173524370318474e-05, "loss": 2.4819, "step": 7840 }, { "epoch": 0.19, "learning_rate": 9.172486714269346e-05, "loss": 2.4625, "step": 7845 }, { "epoch": 0.19, "learning_rate": 9.171448465994058e-05, "loss": 2.5007, "step": 7850 }, { "epoch": 0.19, "learning_rate": 9.170409625639976e-05, "loss": 2.3824, "step": 7855 }, { "epoch": 0.19, "learning_rate": 9.169370193354545e-05, "loss": 2.5104, "step": 7860 }, { "epoch": 0.19, "learning_rate": 9.168330169285301e-05, "loss": 2.5349, "step": 7865 }, { "epoch": 0.19, "learning_rate": 9.167289553579862e-05, "loss": 2.4631, "step": 7870 }, { "epoch": 0.19, "learning_rate": 9.166248346385926e-05, "loss": 2.5399, "step": 7875 }, { "epoch": 0.19, "learning_rate": 9.16520654785128e-05, "loss": 2.5768, "step": 7880 }, { "epoch": 0.19, "learning_rate": 9.164164158123791e-05, "loss": 2.5193, "step": 7885 }, { "epoch": 0.19, "learning_rate": 9.163121177351413e-05, "loss": 2.4899, "step": 7890 }, { "epoch": 0.19, "learning_rate": 9.162077605682183e-05, "loss": 2.4725, "step": 7895 }, { "epoch": 0.19, "learning_rate": 9.161033443264218e-05, "loss": 2.4747, "step": 7900 }, { "epoch": 0.19, "learning_rate": 9.159988690245727e-05, "loss": 2.4257, "step": 7905 }, { "epoch": 0.19, "learning_rate": 9.158943346774994e-05, "loss": 2.5592, "step": 7910 }, { "epoch": 0.19, "learning_rate": 9.157897413000394e-05, "loss": 2.4989, "step": 7915 }, { "epoch": 0.19, "learning_rate": 9.15685088907038e-05, "loss": 2.4874, "step": 7920 }, { "epoch": 0.19, "learning_rate": 9.155803775133493e-05, "loss": 2.4862, "step": 7925 }, { "epoch": 0.19, "learning_rate": 9.154756071338357e-05, "loss": 2.4425, "step": 7930 }, { "epoch": 0.19, "learning_rate": 9.153707777833678e-05, "loss": 2.4905, "step": 7935 }, { "epoch": 0.19, "learning_rate": 9.152658894768244e-05, "loss": 2.5175, "step": 7940 }, { "epoch": 0.19, "learning_rate": 9.151609422290934e-05, "loss": 2.4737, "step": 7945 }, { "epoch": 0.19, "learning_rate": 9.150559360550703e-05, "loss": 2.5006, "step": 7950 }, { "epoch": 0.19, "learning_rate": 9.149508709696594e-05, "loss": 2.5297, "step": 7955 }, { "epoch": 0.19, "learning_rate": 9.148457469877732e-05, "loss": 2.4137, "step": 7960 }, { "epoch": 0.19, "learning_rate": 9.147405641243322e-05, "loss": 2.4842, "step": 7965 }, { "epoch": 0.19, "learning_rate": 9.146353223942663e-05, "loss": 2.5375, "step": 7970 }, { "epoch": 0.19, "learning_rate": 9.145300218125125e-05, "loss": 2.5029, "step": 7975 }, { "epoch": 0.19, "learning_rate": 9.144246623940172e-05, "loss": 2.4668, "step": 7980 }, { "epoch": 0.19, "learning_rate": 9.143192441537344e-05, "loss": 2.5034, "step": 7985 }, { "epoch": 0.19, "learning_rate": 9.14213767106627e-05, "loss": 2.4772, "step": 7990 }, { "epoch": 0.19, "learning_rate": 9.141082312676657e-05, "loss": 2.4039, "step": 7995 }, { "epoch": 0.19, "learning_rate": 9.1400263665183e-05, "loss": 2.4349, "step": 8000 } ], "logging_steps": 5, "max_steps": 41794, "num_train_epochs": 1, "save_steps": 1000, "total_flos": 1.905411560256307e+19, "trial_name": null, "trial_params": null }