{ "best_metric": null, "best_model_checkpoint": null, "epoch": 0.12208055790814965, "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": 2.0026, "step": 5 }, { "epoch": 0.0, "learning_rate": 1e-05, "loss": 2.0248, "step": 10 }, { "epoch": 0.0, "learning_rate": 1.5e-05, "loss": 2.0099, "step": 15 }, { "epoch": 0.0, "learning_rate": 2e-05, "loss": 2.0089, "step": 20 }, { "epoch": 0.0, "learning_rate": 2.5e-05, "loss": 2.0238, "step": 25 }, { "epoch": 0.0, "learning_rate": 3e-05, "loss": 1.9942, "step": 30 }, { "epoch": 0.0, "learning_rate": 3.5e-05, "loss": 1.9853, "step": 35 }, { "epoch": 0.0, "learning_rate": 4e-05, "loss": 1.9805, "step": 40 }, { "epoch": 0.0, "learning_rate": 4.5e-05, "loss": 1.9678, "step": 45 }, { "epoch": 0.0, "learning_rate": 5e-05, "loss": 2.012, "step": 50 }, { "epoch": 0.0, "learning_rate": 5.500000000000001e-05, "loss": 1.9667, "step": 55 }, { "epoch": 0.0, "learning_rate": 6e-05, "loss": 2.0015, "step": 60 }, { "epoch": 0.0, "learning_rate": 6.500000000000001e-05, "loss": 1.961, "step": 65 }, { "epoch": 0.0, "learning_rate": 7e-05, "loss": 1.9918, "step": 70 }, { "epoch": 0.0, "learning_rate": 7.500000000000001e-05, "loss": 1.964, "step": 75 }, { "epoch": 0.0, "learning_rate": 8e-05, "loss": 1.9581, "step": 80 }, { "epoch": 0.0, "learning_rate": 8.5e-05, "loss": 1.9984, "step": 85 }, { "epoch": 0.0, "learning_rate": 9e-05, "loss": 1.9594, "step": 90 }, { "epoch": 0.0, "learning_rate": 9.5e-05, "loss": 1.9505, "step": 95 }, { "epoch": 0.0, "learning_rate": 0.0001, "loss": 1.9277, "step": 100 }, { "epoch": 0.0, "learning_rate": 9.999999855912629e-05, "loss": 1.9239, "step": 105 }, { "epoch": 0.0, "learning_rate": 9.999999423650521e-05, "loss": 1.9636, "step": 110 }, { "epoch": 0.0, "learning_rate": 9.999998703213703e-05, "loss": 1.9473, "step": 115 }, { "epoch": 0.0, "learning_rate": 9.999997694602217e-05, "loss": 1.9514, "step": 120 }, { "epoch": 0.0, "learning_rate": 9.999996397816119e-05, "loss": 1.9276, "step": 125 }, { "epoch": 0.0, "learning_rate": 9.999994812855484e-05, "loss": 1.9395, "step": 130 }, { "epoch": 0.0, "learning_rate": 9.999992939720406e-05, "loss": 1.9439, "step": 135 }, { "epoch": 0.0, "learning_rate": 9.999990778410989e-05, "loss": 1.8595, "step": 140 }, { "epoch": 0.0, "learning_rate": 9.999988328927361e-05, "loss": 1.9043, "step": 145 }, { "epoch": 0.0, "learning_rate": 9.999985591269662e-05, "loss": 1.9247, "step": 150 }, { "epoch": 0.0, "learning_rate": 9.99998256543805e-05, "loss": 1.9607, "step": 155 }, { "epoch": 0.0, "learning_rate": 9.999979251432698e-05, "loss": 1.9351, "step": 160 }, { "epoch": 0.0, "learning_rate": 9.9999756492538e-05, "loss": 1.9118, "step": 165 }, { "epoch": 0.0, "learning_rate": 9.999971758901558e-05, "loss": 1.954, "step": 170 }, { "epoch": 0.0, "learning_rate": 9.999967580376202e-05, "loss": 1.8809, "step": 175 }, { "epoch": 0.0, "learning_rate": 9.999963113677971e-05, "loss": 1.8932, "step": 180 }, { "epoch": 0.0, "learning_rate": 9.999958358807122e-05, "loss": 1.8943, "step": 185 }, { "epoch": 0.0, "learning_rate": 9.99995331576393e-05, "loss": 1.9126, "step": 190 }, { "epoch": 0.0, "learning_rate": 9.999947984548683e-05, "loss": 1.8863, "step": 195 }, { "epoch": 0.0, "learning_rate": 9.999942365161692e-05, "loss": 1.8884, "step": 200 }, { "epoch": 0.0, "learning_rate": 9.999936457603278e-05, "loss": 1.9114, "step": 205 }, { "epoch": 0.0, "learning_rate": 9.999930261873782e-05, "loss": 1.851, "step": 210 }, { "epoch": 0.0, "learning_rate": 9.999923777973563e-05, "loss": 1.8973, "step": 215 }, { "epoch": 0.0, "learning_rate": 9.999917005902992e-05, "loss": 1.9325, "step": 220 }, { "epoch": 0.0, "learning_rate": 9.99990994566246e-05, "loss": 1.9195, "step": 225 }, { "epoch": 0.0, "learning_rate": 9.999902597252375e-05, "loss": 1.8705, "step": 230 }, { "epoch": 0.0, "learning_rate": 9.999894960673161e-05, "loss": 1.9104, "step": 235 }, { "epoch": 0.0, "learning_rate": 9.999887035925258e-05, "loss": 1.9437, "step": 240 }, { "epoch": 0.0, "learning_rate": 9.999878823009118e-05, "loss": 1.9284, "step": 245 }, { "epoch": 0.0, "learning_rate": 9.999870321925223e-05, "loss": 1.9093, "step": 250 }, { "epoch": 0.0, "learning_rate": 9.999861532674056e-05, "loss": 1.8879, "step": 255 }, { "epoch": 0.0, "learning_rate": 9.999852455256125e-05, "loss": 1.8551, "step": 260 }, { "epoch": 0.0, "learning_rate": 9.999843089671954e-05, "loss": 1.8613, "step": 265 }, { "epoch": 0.0, "learning_rate": 9.999833435922084e-05, "loss": 1.9014, "step": 270 }, { "epoch": 0.0, "learning_rate": 9.999823494007068e-05, "loss": 1.8903, "step": 275 }, { "epoch": 0.0, "learning_rate": 9.999813263927484e-05, "loss": 1.8855, "step": 280 }, { "epoch": 0.0, "learning_rate": 9.999802745683917e-05, "loss": 1.8983, "step": 285 }, { "epoch": 0.0, "learning_rate": 9.999791939276975e-05, "loss": 1.925, "step": 290 }, { "epoch": 0.0, "learning_rate": 9.999780844707281e-05, "loss": 1.892, "step": 295 }, { "epoch": 0.0, "learning_rate": 9.999769461975474e-05, "loss": 1.8924, "step": 300 }, { "epoch": 0.0, "learning_rate": 9.99975779108221e-05, "loss": 1.8417, "step": 305 }, { "epoch": 0.0, "learning_rate": 9.999745832028163e-05, "loss": 1.8955, "step": 310 }, { "epoch": 0.0, "learning_rate": 9.999733584814019e-05, "loss": 1.8836, "step": 315 }, { "epoch": 0.0, "learning_rate": 9.999721049440489e-05, "loss": 1.8587, "step": 320 }, { "epoch": 0.0, "learning_rate": 9.999708225908291e-05, "loss": 1.8588, "step": 325 }, { "epoch": 0.01, "learning_rate": 9.999695114218167e-05, "loss": 1.8219, "step": 330 }, { "epoch": 0.01, "learning_rate": 9.999681714370869e-05, "loss": 1.9077, "step": 335 }, { "epoch": 0.01, "learning_rate": 9.999668026367172e-05, "loss": 1.9094, "step": 340 }, { "epoch": 0.01, "learning_rate": 9.999654050207865e-05, "loss": 1.8561, "step": 345 }, { "epoch": 0.01, "learning_rate": 9.999639785893752e-05, "loss": 1.8445, "step": 350 }, { "epoch": 0.01, "learning_rate": 9.999625233425658e-05, "loss": 1.8797, "step": 355 }, { "epoch": 0.01, "learning_rate": 9.99961039280442e-05, "loss": 1.876, "step": 360 }, { "epoch": 0.01, "learning_rate": 9.999595264030892e-05, "loss": 1.879, "step": 365 }, { "epoch": 0.01, "learning_rate": 9.999579847105946e-05, "loss": 1.8653, "step": 370 }, { "epoch": 0.01, "learning_rate": 9.999564142030474e-05, "loss": 1.8526, "step": 375 }, { "epoch": 0.01, "learning_rate": 9.999548148805378e-05, "loss": 1.8672, "step": 380 }, { "epoch": 0.01, "learning_rate": 9.99953186743158e-05, "loss": 1.8718, "step": 385 }, { "epoch": 0.01, "learning_rate": 9.99951529791002e-05, "loss": 1.8202, "step": 390 }, { "epoch": 0.01, "learning_rate": 9.999498440241652e-05, "loss": 1.8562, "step": 395 }, { "epoch": 0.01, "learning_rate": 9.999481294427446e-05, "loss": 1.8406, "step": 400 }, { "epoch": 0.01, "learning_rate": 9.999463860468395e-05, "loss": 1.8673, "step": 405 }, { "epoch": 0.01, "learning_rate": 9.999446138365497e-05, "loss": 1.8501, "step": 410 }, { "epoch": 0.01, "learning_rate": 9.999428128119779e-05, "loss": 1.869, "step": 415 }, { "epoch": 0.01, "learning_rate": 9.999409829732276e-05, "loss": 1.8194, "step": 420 }, { "epoch": 0.01, "learning_rate": 9.999391243204043e-05, "loss": 1.8632, "step": 425 }, { "epoch": 0.01, "learning_rate": 9.999372368536154e-05, "loss": 1.8524, "step": 430 }, { "epoch": 0.01, "learning_rate": 9.999353205729693e-05, "loss": 1.8634, "step": 435 }, { "epoch": 0.01, "learning_rate": 9.999333754785768e-05, "loss": 1.8359, "step": 440 }, { "epoch": 0.01, "learning_rate": 9.999314015705496e-05, "loss": 1.8565, "step": 445 }, { "epoch": 0.01, "learning_rate": 9.999293988490019e-05, "loss": 1.8383, "step": 450 }, { "epoch": 0.01, "learning_rate": 9.999273673140487e-05, "loss": 1.8944, "step": 455 }, { "epoch": 0.01, "learning_rate": 9.999253069658074e-05, "loss": 1.8553, "step": 460 }, { "epoch": 0.01, "learning_rate": 9.999232178043968e-05, "loss": 1.8335, "step": 465 }, { "epoch": 0.01, "learning_rate": 9.999210998299371e-05, "loss": 1.8712, "step": 470 }, { "epoch": 0.01, "learning_rate": 9.999189530425503e-05, "loss": 1.869, "step": 475 }, { "epoch": 0.01, "learning_rate": 9.999167774423603e-05, "loss": 1.8389, "step": 480 }, { "epoch": 0.01, "learning_rate": 9.999145730294924e-05, "loss": 1.8071, "step": 485 }, { "epoch": 0.01, "learning_rate": 9.999123398040737e-05, "loss": 1.8467, "step": 490 }, { "epoch": 0.01, "learning_rate": 9.999100777662331e-05, "loss": 1.8413, "step": 495 }, { "epoch": 0.01, "learning_rate": 9.999077869161006e-05, "loss": 1.8473, "step": 500 }, { "epoch": 0.01, "learning_rate": 9.999054672538084e-05, "loss": 1.8359, "step": 505 }, { "epoch": 0.01, "learning_rate": 9.999031187794904e-05, "loss": 1.8344, "step": 510 }, { "epoch": 0.01, "learning_rate": 9.999007414932817e-05, "loss": 1.8353, "step": 515 }, { "epoch": 0.01, "learning_rate": 9.998983353953193e-05, "loss": 1.8779, "step": 520 }, { "epoch": 0.01, "learning_rate": 9.998959004857419e-05, "loss": 1.8406, "step": 525 }, { "epoch": 0.01, "learning_rate": 9.998934367646899e-05, "loss": 1.8451, "step": 530 }, { "epoch": 0.01, "learning_rate": 9.998909442323054e-05, "loss": 1.8654, "step": 535 }, { "epoch": 0.01, "learning_rate": 9.99888422888732e-05, "loss": 1.7941, "step": 540 }, { "epoch": 0.01, "learning_rate": 9.998858727341148e-05, "loss": 1.8708, "step": 545 }, { "epoch": 0.01, "learning_rate": 9.99883293768601e-05, "loss": 1.8135, "step": 550 }, { "epoch": 0.01, "learning_rate": 9.998806859923392e-05, "loss": 1.805, "step": 555 }, { "epoch": 0.01, "learning_rate": 9.998780494054796e-05, "loss": 1.7936, "step": 560 }, { "epoch": 0.01, "learning_rate": 9.998753840081744e-05, "loss": 1.8126, "step": 565 }, { "epoch": 0.01, "learning_rate": 9.998726898005769e-05, "loss": 1.8107, "step": 570 }, { "epoch": 0.01, "learning_rate": 9.998699667828426e-05, "loss": 1.8132, "step": 575 }, { "epoch": 0.01, "learning_rate": 9.998672149551283e-05, "loss": 1.8331, "step": 580 }, { "epoch": 0.01, "learning_rate": 9.998644343175929e-05, "loss": 1.8465, "step": 585 }, { "epoch": 0.01, "learning_rate": 9.998616248703963e-05, "loss": 1.7908, "step": 590 }, { "epoch": 0.01, "learning_rate": 9.998587866137005e-05, "loss": 1.827, "step": 595 }, { "epoch": 0.01, "learning_rate": 9.998559195476693e-05, "loss": 1.8395, "step": 600 }, { "epoch": 0.01, "learning_rate": 9.998530236724676e-05, "loss": 1.8134, "step": 605 }, { "epoch": 0.01, "learning_rate": 9.998500989882626e-05, "loss": 1.8203, "step": 610 }, { "epoch": 0.01, "learning_rate": 9.998471454952228e-05, "loss": 1.7981, "step": 615 }, { "epoch": 0.01, "learning_rate": 9.998441631935182e-05, "loss": 1.8174, "step": 620 }, { "epoch": 0.01, "learning_rate": 9.99841152083321e-05, "loss": 1.8507, "step": 625 }, { "epoch": 0.01, "learning_rate": 9.998381121648046e-05, "loss": 1.8094, "step": 630 }, { "epoch": 0.01, "learning_rate": 9.998350434381443e-05, "loss": 1.7808, "step": 635 }, { "epoch": 0.01, "learning_rate": 9.998319459035168e-05, "loss": 1.8114, "step": 640 }, { "epoch": 0.01, "learning_rate": 9.998288195611008e-05, "loss": 1.8296, "step": 645 }, { "epoch": 0.01, "learning_rate": 9.998256644110762e-05, "loss": 1.8146, "step": 650 }, { "epoch": 0.01, "learning_rate": 9.998224804536251e-05, "loss": 1.8058, "step": 655 }, { "epoch": 0.01, "learning_rate": 9.99819267688931e-05, "loss": 1.8578, "step": 660 }, { "epoch": 0.01, "learning_rate": 9.998160261171792e-05, "loss": 1.8045, "step": 665 }, { "epoch": 0.01, "learning_rate": 9.998127557385561e-05, "loss": 1.8048, "step": 670 }, { "epoch": 0.01, "learning_rate": 9.998094565532506e-05, "loss": 1.8444, "step": 675 }, { "epoch": 0.01, "learning_rate": 9.998061285614526e-05, "loss": 1.7897, "step": 680 }, { "epoch": 0.01, "learning_rate": 9.998027717633539e-05, "loss": 1.8103, "step": 685 }, { "epoch": 0.01, "learning_rate": 9.997993861591481e-05, "loss": 1.8417, "step": 690 }, { "epoch": 0.01, "learning_rate": 9.997959717490303e-05, "loss": 1.8088, "step": 695 }, { "epoch": 0.01, "learning_rate": 9.997925285331974e-05, "loss": 1.8247, "step": 700 }, { "epoch": 0.01, "learning_rate": 9.997890565118477e-05, "loss": 1.7862, "step": 705 }, { "epoch": 0.01, "learning_rate": 9.997855556851812e-05, "loss": 1.8314, "step": 710 }, { "epoch": 0.01, "learning_rate": 9.997820260533999e-05, "loss": 1.807, "step": 715 }, { "epoch": 0.01, "learning_rate": 9.997784676167072e-05, "loss": 1.7836, "step": 720 }, { "epoch": 0.01, "learning_rate": 9.99774880375308e-05, "loss": 1.7768, "step": 725 }, { "epoch": 0.01, "learning_rate": 9.997712643294092e-05, "loss": 1.7714, "step": 730 }, { "epoch": 0.01, "learning_rate": 9.997676194792192e-05, "loss": 1.7684, "step": 735 }, { "epoch": 0.01, "learning_rate": 9.997639458249482e-05, "loss": 1.8397, "step": 740 }, { "epoch": 0.01, "learning_rate": 9.997602433668076e-05, "loss": 1.8569, "step": 745 }, { "epoch": 0.01, "learning_rate": 9.99756512105011e-05, "loss": 1.8608, "step": 750 }, { "epoch": 0.01, "learning_rate": 9.997527520397734e-05, "loss": 1.7984, "step": 755 }, { "epoch": 0.01, "learning_rate": 9.997489631713117e-05, "loss": 1.8228, "step": 760 }, { "epoch": 0.01, "learning_rate": 9.997451454998441e-05, "loss": 1.7488, "step": 765 }, { "epoch": 0.01, "learning_rate": 9.997412990255905e-05, "loss": 1.8032, "step": 770 }, { "epoch": 0.01, "learning_rate": 9.997374237487728e-05, "loss": 1.7939, "step": 775 }, { "epoch": 0.01, "learning_rate": 9.997335196696145e-05, "loss": 1.7998, "step": 780 }, { "epoch": 0.01, "learning_rate": 9.997295867883401e-05, "loss": 1.8002, "step": 785 }, { "epoch": 0.01, "learning_rate": 9.997256251051767e-05, "loss": 1.831, "step": 790 }, { "epoch": 0.01, "learning_rate": 9.997216346203525e-05, "loss": 1.8173, "step": 795 }, { "epoch": 0.01, "learning_rate": 9.997176153340975e-05, "loss": 1.7784, "step": 800 }, { "epoch": 0.01, "learning_rate": 9.997135672466433e-05, "loss": 1.7927, "step": 805 }, { "epoch": 0.01, "learning_rate": 9.997094903582233e-05, "loss": 1.8172, "step": 810 }, { "epoch": 0.01, "learning_rate": 9.997053846690725e-05, "loss": 1.8248, "step": 815 }, { "epoch": 0.01, "learning_rate": 9.997012501794274e-05, "loss": 1.7619, "step": 820 }, { "epoch": 0.01, "learning_rate": 9.996970868895263e-05, "loss": 1.7507, "step": 825 }, { "epoch": 0.01, "learning_rate": 9.996928947996092e-05, "loss": 1.809, "step": 830 }, { "epoch": 0.01, "learning_rate": 9.996886739099178e-05, "loss": 1.8014, "step": 835 }, { "epoch": 0.01, "learning_rate": 9.996844242206954e-05, "loss": 1.7486, "step": 840 }, { "epoch": 0.01, "learning_rate": 9.996801457321865e-05, "loss": 1.8112, "step": 845 }, { "epoch": 0.01, "learning_rate": 9.996758384446381e-05, "loss": 1.7958, "step": 850 }, { "epoch": 0.01, "learning_rate": 9.996715023582985e-05, "loss": 1.827, "step": 855 }, { "epoch": 0.01, "learning_rate": 9.996671374734173e-05, "loss": 1.7797, "step": 860 }, { "epoch": 0.01, "learning_rate": 9.996627437902464e-05, "loss": 1.7728, "step": 865 }, { "epoch": 0.01, "learning_rate": 9.996583213090387e-05, "loss": 1.8023, "step": 870 }, { "epoch": 0.01, "learning_rate": 9.996538700300493e-05, "loss": 1.8094, "step": 875 }, { "epoch": 0.01, "learning_rate": 9.996493899535348e-05, "loss": 1.7858, "step": 880 }, { "epoch": 0.01, "learning_rate": 9.996448810797532e-05, "loss": 1.7385, "step": 885 }, { "epoch": 0.01, "learning_rate": 9.996403434089645e-05, "loss": 1.7984, "step": 890 }, { "epoch": 0.01, "learning_rate": 9.996357769414304e-05, "loss": 1.7462, "step": 895 }, { "epoch": 0.01, "learning_rate": 9.996311816774137e-05, "loss": 1.7083, "step": 900 }, { "epoch": 0.01, "learning_rate": 9.996265576171795e-05, "loss": 1.8292, "step": 905 }, { "epoch": 0.01, "learning_rate": 9.996219047609943e-05, "loss": 1.7827, "step": 910 }, { "epoch": 0.01, "learning_rate": 9.996172231091261e-05, "loss": 1.7965, "step": 915 }, { "epoch": 0.01, "learning_rate": 9.996125126618448e-05, "loss": 1.8119, "step": 920 }, { "epoch": 0.01, "learning_rate": 9.99607773419422e-05, "loss": 1.763, "step": 925 }, { "epoch": 0.01, "learning_rate": 9.996030053821309e-05, "loss": 1.796, "step": 930 }, { "epoch": 0.01, "learning_rate": 9.995982085502461e-05, "loss": 1.7422, "step": 935 }, { "epoch": 0.01, "learning_rate": 9.995933829240442e-05, "loss": 1.8098, "step": 940 }, { "epoch": 0.01, "learning_rate": 9.995885285038032e-05, "loss": 1.759, "step": 945 }, { "epoch": 0.01, "learning_rate": 9.995836452898029e-05, "loss": 1.8041, "step": 950 }, { "epoch": 0.01, "learning_rate": 9.99578733282325e-05, "loss": 1.8144, "step": 955 }, { "epoch": 0.01, "learning_rate": 9.995737924816523e-05, "loss": 1.7755, "step": 960 }, { "epoch": 0.01, "learning_rate": 9.995688228880696e-05, "loss": 1.8064, "step": 965 }, { "epoch": 0.01, "learning_rate": 9.995638245018635e-05, "loss": 1.7959, "step": 970 }, { "epoch": 0.01, "learning_rate": 9.99558797323322e-05, "loss": 1.7577, "step": 975 }, { "epoch": 0.01, "learning_rate": 9.995537413527348e-05, "loss": 1.7481, "step": 980 }, { "epoch": 0.02, "learning_rate": 9.995486565903931e-05, "loss": 1.7473, "step": 985 }, { "epoch": 0.02, "learning_rate": 9.995435430365904e-05, "loss": 1.8309, "step": 990 }, { "epoch": 0.02, "learning_rate": 9.995384006916211e-05, "loss": 1.7571, "step": 995 }, { "epoch": 0.02, "learning_rate": 9.995332295557817e-05, "loss": 1.7568, "step": 1000 }, { "epoch": 0.02, "learning_rate": 9.995280296293703e-05, "loss": 1.7838, "step": 1005 }, { "epoch": 0.02, "learning_rate": 9.995228009126864e-05, "loss": 1.8197, "step": 1010 }, { "epoch": 0.02, "learning_rate": 9.995175434060316e-05, "loss": 1.8234, "step": 1015 }, { "epoch": 0.02, "learning_rate": 9.995122571097085e-05, "loss": 1.732, "step": 1020 }, { "epoch": 0.02, "learning_rate": 9.995069420240223e-05, "loss": 1.7941, "step": 1025 }, { "epoch": 0.02, "learning_rate": 9.99501598149279e-05, "loss": 1.8115, "step": 1030 }, { "epoch": 0.02, "learning_rate": 9.994962254857868e-05, "loss": 1.7651, "step": 1035 }, { "epoch": 0.02, "learning_rate": 9.994908240338551e-05, "loss": 1.761, "step": 1040 }, { "epoch": 0.02, "learning_rate": 9.994853937937954e-05, "loss": 1.8317, "step": 1045 }, { "epoch": 0.02, "learning_rate": 9.994799347659205e-05, "loss": 1.8123, "step": 1050 }, { "epoch": 0.02, "learning_rate": 9.994744469505454e-05, "loss": 1.7769, "step": 1055 }, { "epoch": 0.02, "learning_rate": 9.99468930347986e-05, "loss": 1.8033, "step": 1060 }, { "epoch": 0.02, "learning_rate": 9.994633849585604e-05, "loss": 1.7868, "step": 1065 }, { "epoch": 0.02, "learning_rate": 9.994578107825883e-05, "loss": 1.7839, "step": 1070 }, { "epoch": 0.02, "learning_rate": 9.994522078203907e-05, "loss": 1.7749, "step": 1075 }, { "epoch": 0.02, "learning_rate": 9.994465760722908e-05, "loss": 1.7297, "step": 1080 }, { "epoch": 0.02, "learning_rate": 9.994409155386131e-05, "loss": 1.7604, "step": 1085 }, { "epoch": 0.02, "learning_rate": 9.994352262196838e-05, "loss": 1.79, "step": 1090 }, { "epoch": 0.02, "learning_rate": 9.994295081158307e-05, "loss": 1.7977, "step": 1095 }, { "epoch": 0.02, "learning_rate": 9.994237612273836e-05, "loss": 1.7235, "step": 1100 }, { "epoch": 0.02, "learning_rate": 9.994179855546737e-05, "loss": 1.7615, "step": 1105 }, { "epoch": 0.02, "learning_rate": 9.994121810980337e-05, "loss": 1.7778, "step": 1110 }, { "epoch": 0.02, "learning_rate": 9.994063478577982e-05, "loss": 1.7519, "step": 1115 }, { "epoch": 0.02, "learning_rate": 9.994004858343035e-05, "loss": 1.7954, "step": 1120 }, { "epoch": 0.02, "learning_rate": 9.993945950278873e-05, "loss": 1.7965, "step": 1125 }, { "epoch": 0.02, "learning_rate": 9.993886754388893e-05, "loss": 1.742, "step": 1130 }, { "epoch": 0.02, "learning_rate": 9.993827270676505e-05, "loss": 1.7712, "step": 1135 }, { "epoch": 0.02, "learning_rate": 9.993767499145139e-05, "loss": 1.7868, "step": 1140 }, { "epoch": 0.02, "learning_rate": 9.993707439798239e-05, "loss": 1.7902, "step": 1145 }, { "epoch": 0.02, "learning_rate": 9.993647092639266e-05, "loss": 1.7613, "step": 1150 }, { "epoch": 0.02, "learning_rate": 9.993586457671699e-05, "loss": 1.7468, "step": 1155 }, { "epoch": 0.02, "learning_rate": 9.993525534899034e-05, "loss": 1.767, "step": 1160 }, { "epoch": 0.02, "learning_rate": 9.993464324324777e-05, "loss": 1.7123, "step": 1165 }, { "epoch": 0.02, "learning_rate": 9.993402825952463e-05, "loss": 1.732, "step": 1170 }, { "epoch": 0.02, "learning_rate": 9.99334103978563e-05, "loss": 1.7992, "step": 1175 }, { "epoch": 0.02, "learning_rate": 9.993278965827844e-05, "loss": 1.7379, "step": 1180 }, { "epoch": 0.02, "learning_rate": 9.99321660408268e-05, "loss": 1.7381, "step": 1185 }, { "epoch": 0.02, "learning_rate": 9.993153954553733e-05, "loss": 1.8009, "step": 1190 }, { "epoch": 0.02, "learning_rate": 9.993091017244615e-05, "loss": 1.7996, "step": 1195 }, { "epoch": 0.02, "learning_rate": 9.993027792158951e-05, "loss": 1.7463, "step": 1200 }, { "epoch": 0.02, "learning_rate": 9.992964279300385e-05, "loss": 1.7356, "step": 1205 }, { "epoch": 0.02, "learning_rate": 9.99290047867258e-05, "loss": 1.7633, "step": 1210 }, { "epoch": 0.02, "learning_rate": 9.992836390279212e-05, "loss": 1.7475, "step": 1215 }, { "epoch": 0.02, "learning_rate": 9.992772014123974e-05, "loss": 1.7673, "step": 1220 }, { "epoch": 0.02, "learning_rate": 9.992707350210576e-05, "loss": 1.8037, "step": 1225 }, { "epoch": 0.02, "learning_rate": 9.992642398542746e-05, "loss": 1.7495, "step": 1230 }, { "epoch": 0.02, "learning_rate": 9.992577159124228e-05, "loss": 1.7967, "step": 1235 }, { "epoch": 0.02, "learning_rate": 9.992511631958781e-05, "loss": 1.8108, "step": 1240 }, { "epoch": 0.02, "learning_rate": 9.992445817050182e-05, "loss": 1.8023, "step": 1245 }, { "epoch": 0.02, "learning_rate": 9.992379714402223e-05, "loss": 1.7489, "step": 1250 }, { "epoch": 0.02, "learning_rate": 9.992313324018717e-05, "loss": 1.8224, "step": 1255 }, { "epoch": 0.02, "learning_rate": 9.992246645903487e-05, "loss": 1.7398, "step": 1260 }, { "epoch": 0.02, "learning_rate": 9.992179680060378e-05, "loss": 1.7663, "step": 1265 }, { "epoch": 0.02, "learning_rate": 9.992112426493247e-05, "loss": 1.7606, "step": 1270 }, { "epoch": 0.02, "learning_rate": 9.992044885205975e-05, "loss": 1.7961, "step": 1275 }, { "epoch": 0.02, "learning_rate": 9.991977056202451e-05, "loss": 1.7715, "step": 1280 }, { "epoch": 0.02, "learning_rate": 9.991908939486582e-05, "loss": 1.7413, "step": 1285 }, { "epoch": 0.02, "learning_rate": 9.991840535062301e-05, "loss": 1.7502, "step": 1290 }, { "epoch": 0.02, "learning_rate": 9.991771842933546e-05, "loss": 1.7688, "step": 1295 }, { "epoch": 0.02, "learning_rate": 9.991702863104275e-05, "loss": 1.7245, "step": 1300 }, { "epoch": 0.02, "learning_rate": 9.991633595578467e-05, "loss": 1.7578, "step": 1305 }, { "epoch": 0.02, "learning_rate": 9.99156404036011e-05, "loss": 1.7339, "step": 1310 }, { "epoch": 0.02, "learning_rate": 9.991494197453218e-05, "loss": 1.7325, "step": 1315 }, { "epoch": 0.02, "learning_rate": 9.991424066861812e-05, "loss": 1.7334, "step": 1320 }, { "epoch": 0.02, "learning_rate": 9.991353648589937e-05, "loss": 1.7388, "step": 1325 }, { "epoch": 0.02, "learning_rate": 9.99128294264165e-05, "loss": 1.7631, "step": 1330 }, { "epoch": 0.02, "learning_rate": 9.991211949021026e-05, "loss": 1.7914, "step": 1335 }, { "epoch": 0.02, "learning_rate": 9.991140667732158e-05, "loss": 1.723, "step": 1340 }, { "epoch": 0.02, "learning_rate": 9.99106909877915e-05, "loss": 1.7402, "step": 1345 }, { "epoch": 0.02, "learning_rate": 9.990997242166134e-05, "loss": 1.7559, "step": 1350 }, { "epoch": 0.02, "learning_rate": 9.990925097897246e-05, "loss": 1.7178, "step": 1355 }, { "epoch": 0.02, "learning_rate": 9.990852665976646e-05, "loss": 1.7576, "step": 1360 }, { "epoch": 0.02, "learning_rate": 9.990779946408508e-05, "loss": 1.7329, "step": 1365 }, { "epoch": 0.02, "learning_rate": 9.990706939197024e-05, "loss": 1.7708, "step": 1370 }, { "epoch": 0.02, "learning_rate": 9.9906336443464e-05, "loss": 1.7427, "step": 1375 }, { "epoch": 0.02, "learning_rate": 9.990560061860865e-05, "loss": 1.7934, "step": 1380 }, { "epoch": 0.02, "learning_rate": 9.990486191744655e-05, "loss": 1.767, "step": 1385 }, { "epoch": 0.02, "learning_rate": 9.990412034002028e-05, "loss": 1.7538, "step": 1390 }, { "epoch": 0.02, "learning_rate": 9.99033758863726e-05, "loss": 1.751, "step": 1395 }, { "epoch": 0.02, "learning_rate": 9.990262855654639e-05, "loss": 1.7478, "step": 1400 }, { "epoch": 0.02, "learning_rate": 9.990187835058475e-05, "loss": 1.7407, "step": 1405 }, { "epoch": 0.02, "learning_rate": 9.990112526853093e-05, "loss": 1.7729, "step": 1410 }, { "epoch": 0.02, "learning_rate": 9.990036931042828e-05, "loss": 1.7626, "step": 1415 }, { "epoch": 0.02, "learning_rate": 9.989961047632041e-05, "loss": 1.7433, "step": 1420 }, { "epoch": 0.02, "learning_rate": 9.989884876625106e-05, "loss": 1.7879, "step": 1425 }, { "epoch": 0.02, "learning_rate": 9.989808418026412e-05, "loss": 1.7748, "step": 1430 }, { "epoch": 0.02, "learning_rate": 9.989731671840364e-05, "loss": 1.7403, "step": 1435 }, { "epoch": 0.02, "learning_rate": 9.989654638071389e-05, "loss": 1.7504, "step": 1440 }, { "epoch": 0.02, "learning_rate": 9.989577316723922e-05, "loss": 1.7867, "step": 1445 }, { "epoch": 0.02, "learning_rate": 9.989499707802423e-05, "loss": 1.7723, "step": 1450 }, { "epoch": 0.02, "learning_rate": 9.989421811311363e-05, "loss": 1.7481, "step": 1455 }, { "epoch": 0.02, "learning_rate": 9.989343627255235e-05, "loss": 1.7435, "step": 1460 }, { "epoch": 0.02, "learning_rate": 9.989265155638541e-05, "loss": 1.7017, "step": 1465 }, { "epoch": 0.02, "learning_rate": 9.989186396465806e-05, "loss": 1.7606, "step": 1470 }, { "epoch": 0.02, "learning_rate": 9.989107349741568e-05, "loss": 1.7669, "step": 1475 }, { "epoch": 0.02, "learning_rate": 9.989028015470385e-05, "loss": 1.7676, "step": 1480 }, { "epoch": 0.02, "learning_rate": 9.988948393656826e-05, "loss": 1.6961, "step": 1485 }, { "epoch": 0.02, "learning_rate": 9.988868484305483e-05, "loss": 1.7805, "step": 1490 }, { "epoch": 0.02, "learning_rate": 9.98878828742096e-05, "loss": 1.7119, "step": 1495 }, { "epoch": 0.02, "learning_rate": 9.988707803007882e-05, "loss": 1.7851, "step": 1500 }, { "epoch": 0.02, "learning_rate": 9.988627031070883e-05, "loss": 1.7576, "step": 1505 }, { "epoch": 0.02, "learning_rate": 9.98854597161462e-05, "loss": 1.7474, "step": 1510 }, { "epoch": 0.02, "learning_rate": 9.988464624643768e-05, "loss": 1.7418, "step": 1515 }, { "epoch": 0.02, "learning_rate": 9.988382990163012e-05, "loss": 1.7104, "step": 1520 }, { "epoch": 0.02, "learning_rate": 9.988301068177058e-05, "loss": 1.6872, "step": 1525 }, { "epoch": 0.02, "learning_rate": 9.988218858690627e-05, "loss": 1.7528, "step": 1530 }, { "epoch": 0.02, "learning_rate": 9.988136361708458e-05, "loss": 1.716, "step": 1535 }, { "epoch": 0.02, "learning_rate": 9.988053577235305e-05, "loss": 1.7068, "step": 1540 }, { "epoch": 0.02, "learning_rate": 9.98797050527594e-05, "loss": 1.7535, "step": 1545 }, { "epoch": 0.02, "learning_rate": 9.987887145835151e-05, "loss": 1.76, "step": 1550 }, { "epoch": 0.02, "learning_rate": 9.987803498917742e-05, "loss": 1.765, "step": 1555 }, { "epoch": 0.02, "learning_rate": 9.987719564528532e-05, "loss": 1.7896, "step": 1560 }, { "epoch": 0.02, "learning_rate": 9.987635342672363e-05, "loss": 1.7245, "step": 1565 }, { "epoch": 0.02, "learning_rate": 9.987550833354084e-05, "loss": 1.7219, "step": 1570 }, { "epoch": 0.02, "learning_rate": 9.987466036578571e-05, "loss": 1.7698, "step": 1575 }, { "epoch": 0.02, "learning_rate": 9.987380952350708e-05, "loss": 1.7234, "step": 1580 }, { "epoch": 0.02, "learning_rate": 9.987295580675397e-05, "loss": 1.7523, "step": 1585 }, { "epoch": 0.02, "learning_rate": 9.987209921557564e-05, "loss": 1.7153, "step": 1590 }, { "epoch": 0.02, "learning_rate": 9.987123975002141e-05, "loss": 1.709, "step": 1595 }, { "epoch": 0.02, "learning_rate": 9.987037741014084e-05, "loss": 1.7034, "step": 1600 }, { "epoch": 0.02, "learning_rate": 9.986951219598361e-05, "loss": 1.7825, "step": 1605 }, { "epoch": 0.02, "learning_rate": 9.986864410759961e-05, "loss": 1.7801, "step": 1610 }, { "epoch": 0.02, "learning_rate": 9.986777314503886e-05, "loss": 1.7233, "step": 1615 }, { "epoch": 0.02, "learning_rate": 9.986689930835155e-05, "loss": 1.7521, "step": 1620 }, { "epoch": 0.02, "learning_rate": 9.986602259758808e-05, "loss": 1.7708, "step": 1625 }, { "epoch": 0.02, "learning_rate": 9.986514301279893e-05, "loss": 1.7704, "step": 1630 }, { "epoch": 0.02, "learning_rate": 9.986426055403483e-05, "loss": 1.7451, "step": 1635 }, { "epoch": 0.03, "learning_rate": 9.986337522134662e-05, "loss": 1.7435, "step": 1640 }, { "epoch": 0.03, "learning_rate": 9.986248701478534e-05, "loss": 1.7718, "step": 1645 }, { "epoch": 0.03, "learning_rate": 9.986159593440217e-05, "loss": 1.7026, "step": 1650 }, { "epoch": 0.03, "learning_rate": 9.986070198024849e-05, "loss": 1.767, "step": 1655 }, { "epoch": 0.03, "learning_rate": 9.985980515237578e-05, "loss": 1.7283, "step": 1660 }, { "epoch": 0.03, "learning_rate": 9.985890545083576e-05, "loss": 1.7558, "step": 1665 }, { "epoch": 0.03, "learning_rate": 9.98580028756803e-05, "loss": 1.7071, "step": 1670 }, { "epoch": 0.03, "learning_rate": 9.985709742696137e-05, "loss": 1.7652, "step": 1675 }, { "epoch": 0.03, "learning_rate": 9.98561891047312e-05, "loss": 1.7229, "step": 1680 }, { "epoch": 0.03, "learning_rate": 9.985527790904213e-05, "loss": 1.7481, "step": 1685 }, { "epoch": 0.03, "learning_rate": 9.985436383994668e-05, "loss": 1.7322, "step": 1690 }, { "epoch": 0.03, "learning_rate": 9.98534468974975e-05, "loss": 1.7264, "step": 1695 }, { "epoch": 0.03, "learning_rate": 9.985252708174749e-05, "loss": 1.7551, "step": 1700 }, { "epoch": 0.03, "learning_rate": 9.985160439274961e-05, "loss": 1.7344, "step": 1705 }, { "epoch": 0.03, "learning_rate": 9.985067883055707e-05, "loss": 1.7906, "step": 1710 }, { "epoch": 0.03, "learning_rate": 9.984975039522323e-05, "loss": 1.7745, "step": 1715 }, { "epoch": 0.03, "learning_rate": 9.984881908680157e-05, "loss": 1.7166, "step": 1720 }, { "epoch": 0.03, "learning_rate": 9.984788490534577e-05, "loss": 1.7139, "step": 1725 }, { "epoch": 0.03, "learning_rate": 9.984694785090968e-05, "loss": 1.7118, "step": 1730 }, { "epoch": 0.03, "learning_rate": 9.984600792354731e-05, "loss": 1.7221, "step": 1735 }, { "epoch": 0.03, "learning_rate": 9.984506512331282e-05, "loss": 1.7908, "step": 1740 }, { "epoch": 0.03, "learning_rate": 9.984411945026056e-05, "loss": 1.7416, "step": 1745 }, { "epoch": 0.03, "learning_rate": 9.984317090444503e-05, "loss": 1.7253, "step": 1750 }, { "epoch": 0.03, "learning_rate": 9.984221948592089e-05, "loss": 1.7253, "step": 1755 }, { "epoch": 0.03, "learning_rate": 9.984126519474299e-05, "loss": 1.6949, "step": 1760 }, { "epoch": 0.03, "learning_rate": 9.984030803096634e-05, "loss": 1.7186, "step": 1765 }, { "epoch": 0.03, "learning_rate": 9.983934799464607e-05, "loss": 1.7529, "step": 1770 }, { "epoch": 0.03, "learning_rate": 9.983838508583754e-05, "loss": 1.7525, "step": 1775 }, { "epoch": 0.03, "learning_rate": 9.983741930459624e-05, "loss": 1.6876, "step": 1780 }, { "epoch": 0.03, "learning_rate": 9.983645065097784e-05, "loss": 1.7548, "step": 1785 }, { "epoch": 0.03, "learning_rate": 9.983547912503814e-05, "loss": 1.713, "step": 1790 }, { "epoch": 0.03, "learning_rate": 9.983450472683317e-05, "loss": 1.6764, "step": 1795 }, { "epoch": 0.03, "learning_rate": 9.983352745641907e-05, "loss": 1.7504, "step": 1800 }, { "epoch": 0.03, "learning_rate": 9.983254731385216e-05, "loss": 1.7552, "step": 1805 }, { "epoch": 0.03, "learning_rate": 9.983156429918893e-05, "loss": 1.7231, "step": 1810 }, { "epoch": 0.03, "learning_rate": 9.983057841248609e-05, "loss": 1.73, "step": 1815 }, { "epoch": 0.03, "learning_rate": 9.982958965380038e-05, "loss": 1.7371, "step": 1820 }, { "epoch": 0.03, "learning_rate": 9.982859802318882e-05, "loss": 1.7292, "step": 1825 }, { "epoch": 0.03, "learning_rate": 9.982760352070859e-05, "loss": 1.7616, "step": 1830 }, { "epoch": 0.03, "learning_rate": 9.982660614641697e-05, "loss": 1.7172, "step": 1835 }, { "epoch": 0.03, "learning_rate": 9.982560590037146e-05, "loss": 1.7313, "step": 1840 }, { "epoch": 0.03, "learning_rate": 9.98246027826297e-05, "loss": 1.7057, "step": 1845 }, { "epoch": 0.03, "learning_rate": 9.982359679324953e-05, "loss": 1.7153, "step": 1850 }, { "epoch": 0.03, "learning_rate": 9.98225879322889e-05, "loss": 1.7823, "step": 1855 }, { "epoch": 0.03, "learning_rate": 9.982157619980597e-05, "loss": 1.7251, "step": 1860 }, { "epoch": 0.03, "learning_rate": 9.982056159585904e-05, "loss": 1.7261, "step": 1865 }, { "epoch": 0.03, "learning_rate": 9.98195441205066e-05, "loss": 1.7428, "step": 1870 }, { "epoch": 0.03, "learning_rate": 9.98185237738073e-05, "loss": 1.708, "step": 1875 }, { "epoch": 0.03, "learning_rate": 9.981750055581992e-05, "loss": 1.738, "step": 1880 }, { "epoch": 0.03, "learning_rate": 9.981647446660345e-05, "loss": 1.7194, "step": 1885 }, { "epoch": 0.03, "learning_rate": 9.981544550621705e-05, "loss": 1.7285, "step": 1890 }, { "epoch": 0.03, "learning_rate": 9.981441367471997e-05, "loss": 1.7114, "step": 1895 }, { "epoch": 0.03, "learning_rate": 9.98133789721717e-05, "loss": 1.7365, "step": 1900 }, { "epoch": 0.03, "learning_rate": 9.981234139863192e-05, "loss": 1.7444, "step": 1905 }, { "epoch": 0.03, "learning_rate": 9.981130095416038e-05, "loss": 1.7258, "step": 1910 }, { "epoch": 0.03, "learning_rate": 9.981025763881707e-05, "loss": 1.7282, "step": 1915 }, { "epoch": 0.03, "learning_rate": 9.980921145266211e-05, "loss": 1.755, "step": 1920 }, { "epoch": 0.03, "learning_rate": 9.98081623957558e-05, "loss": 1.6993, "step": 1925 }, { "epoch": 0.03, "learning_rate": 9.98071104681586e-05, "loss": 1.7183, "step": 1930 }, { "epoch": 0.03, "learning_rate": 9.980605566993113e-05, "loss": 1.6983, "step": 1935 }, { "epoch": 0.03, "learning_rate": 9.98049980011342e-05, "loss": 1.7123, "step": 1940 }, { "epoch": 0.03, "learning_rate": 9.980393746182879e-05, "loss": 1.7051, "step": 1945 }, { "epoch": 0.03, "learning_rate": 9.980287405207596e-05, "loss": 1.7093, "step": 1950 }, { "epoch": 0.03, "learning_rate": 9.980180777193706e-05, "loss": 1.7784, "step": 1955 }, { "epoch": 0.03, "learning_rate": 9.98007386214735e-05, "loss": 1.7098, "step": 1960 }, { "epoch": 0.03, "learning_rate": 9.979966660074693e-05, "loss": 1.6902, "step": 1965 }, { "epoch": 0.03, "learning_rate": 9.979859170981913e-05, "loss": 1.741, "step": 1970 }, { "epoch": 0.03, "learning_rate": 9.979751394875205e-05, "loss": 1.7103, "step": 1975 }, { "epoch": 0.03, "learning_rate": 9.97964333176078e-05, "loss": 1.7642, "step": 1980 }, { "epoch": 0.03, "learning_rate": 9.979534981644866e-05, "loss": 1.7136, "step": 1985 }, { "epoch": 0.03, "learning_rate": 9.97942634453371e-05, "loss": 1.7372, "step": 1990 }, { "epoch": 0.03, "learning_rate": 9.979317420433573e-05, "loss": 1.6814, "step": 1995 }, { "epoch": 0.03, "learning_rate": 9.97920820935073e-05, "loss": 1.7536, "step": 2000 }, { "epoch": 0.03, "learning_rate": 9.979098711291477e-05, "loss": 1.723, "step": 2005 }, { "epoch": 0.03, "learning_rate": 9.978988926262124e-05, "loss": 1.7024, "step": 2010 }, { "epoch": 0.03, "learning_rate": 9.978878854269001e-05, "loss": 1.7271, "step": 2015 }, { "epoch": 0.03, "learning_rate": 9.978768495318449e-05, "loss": 1.7288, "step": 2020 }, { "epoch": 0.03, "learning_rate": 9.978657849416831e-05, "loss": 1.7288, "step": 2025 }, { "epoch": 0.03, "learning_rate": 9.978546916570521e-05, "loss": 1.7217, "step": 2030 }, { "epoch": 0.03, "learning_rate": 9.978435696785917e-05, "loss": 1.6941, "step": 2035 }, { "epoch": 0.03, "learning_rate": 9.978324190069426e-05, "loss": 1.7067, "step": 2040 }, { "epoch": 0.03, "learning_rate": 9.978212396427475e-05, "loss": 1.745, "step": 2045 }, { "epoch": 0.03, "learning_rate": 9.978100315866507e-05, "loss": 1.7446, "step": 2050 }, { "epoch": 0.03, "learning_rate": 9.977987948392984e-05, "loss": 1.7553, "step": 2055 }, { "epoch": 0.03, "learning_rate": 9.977875294013379e-05, "loss": 1.7146, "step": 2060 }, { "epoch": 0.03, "learning_rate": 9.977762352734188e-05, "loss": 1.6872, "step": 2065 }, { "epoch": 0.03, "learning_rate": 9.977649124561918e-05, "loss": 1.7106, "step": 2070 }, { "epoch": 0.03, "learning_rate": 9.977535609503094e-05, "loss": 1.7296, "step": 2075 }, { "epoch": 0.03, "learning_rate": 9.977421807564265e-05, "loss": 1.7122, "step": 2080 }, { "epoch": 0.03, "learning_rate": 9.97730771875198e-05, "loss": 1.7436, "step": 2085 }, { "epoch": 0.03, "learning_rate": 9.977193343072824e-05, "loss": 1.7459, "step": 2090 }, { "epoch": 0.03, "learning_rate": 9.977078680533382e-05, "loss": 1.7288, "step": 2095 }, { "epoch": 0.03, "learning_rate": 9.976963731140265e-05, "loss": 1.698, "step": 2100 }, { "epoch": 0.03, "learning_rate": 9.9768484949001e-05, "loss": 1.7131, "step": 2105 }, { "epoch": 0.03, "learning_rate": 9.976732971819527e-05, "loss": 1.7367, "step": 2110 }, { "epoch": 0.03, "learning_rate": 9.976617161905202e-05, "loss": 1.6902, "step": 2115 }, { "epoch": 0.03, "learning_rate": 9.976501065163804e-05, "loss": 1.7042, "step": 2120 }, { "epoch": 0.03, "learning_rate": 9.976384681602022e-05, "loss": 1.7161, "step": 2125 }, { "epoch": 0.03, "learning_rate": 9.976268011226564e-05, "loss": 1.6988, "step": 2130 }, { "epoch": 0.03, "learning_rate": 9.976151054044153e-05, "loss": 1.7403, "step": 2135 }, { "epoch": 0.03, "learning_rate": 9.976033810061532e-05, "loss": 1.6963, "step": 2140 }, { "epoch": 0.03, "learning_rate": 9.975916279285456e-05, "loss": 1.7074, "step": 2145 }, { "epoch": 0.03, "learning_rate": 9.975798461722701e-05, "loss": 1.7081, "step": 2150 }, { "epoch": 0.03, "learning_rate": 9.975680357380057e-05, "loss": 1.697, "step": 2155 }, { "epoch": 0.03, "learning_rate": 9.975561966264331e-05, "loss": 1.7138, "step": 2160 }, { "epoch": 0.03, "learning_rate": 9.975443288382344e-05, "loss": 1.7649, "step": 2165 }, { "epoch": 0.03, "learning_rate": 9.97532432374094e-05, "loss": 1.7248, "step": 2170 }, { "epoch": 0.03, "learning_rate": 9.975205072346974e-05, "loss": 1.7275, "step": 2175 }, { "epoch": 0.03, "learning_rate": 9.975085534207317e-05, "loss": 1.7438, "step": 2180 }, { "epoch": 0.03, "learning_rate": 9.97496570932886e-05, "loss": 1.7334, "step": 2185 }, { "epoch": 0.03, "learning_rate": 9.97484559771851e-05, "loss": 1.7262, "step": 2190 }, { "epoch": 0.03, "learning_rate": 9.97472519938319e-05, "loss": 1.7386, "step": 2195 }, { "epoch": 0.03, "learning_rate": 9.974604514329836e-05, "loss": 1.7371, "step": 2200 }, { "epoch": 0.03, "learning_rate": 9.974483542565407e-05, "loss": 1.698, "step": 2205 }, { "epoch": 0.03, "learning_rate": 9.974362284096873e-05, "loss": 1.6978, "step": 2210 }, { "epoch": 0.03, "learning_rate": 9.974240738931225e-05, "loss": 1.7056, "step": 2215 }, { "epoch": 0.03, "learning_rate": 9.974118907075465e-05, "loss": 1.7132, "step": 2220 }, { "epoch": 0.03, "learning_rate": 9.973996788536618e-05, "loss": 1.765, "step": 2225 }, { "epoch": 0.03, "learning_rate": 9.97387438332172e-05, "loss": 1.7276, "step": 2230 }, { "epoch": 0.03, "learning_rate": 9.973751691437828e-05, "loss": 1.7298, "step": 2235 }, { "epoch": 0.03, "learning_rate": 9.973628712892011e-05, "loss": 1.72, "step": 2240 }, { "epoch": 0.03, "learning_rate": 9.97350544769136e-05, "loss": 1.7464, "step": 2245 }, { "epoch": 0.03, "learning_rate": 9.973381895842975e-05, "loss": 1.7341, "step": 2250 }, { "epoch": 0.03, "learning_rate": 9.97325805735398e-05, "loss": 1.6781, "step": 2255 }, { "epoch": 0.03, "learning_rate": 9.973133932231513e-05, "loss": 1.6844, "step": 2260 }, { "epoch": 0.03, "learning_rate": 9.973009520482725e-05, "loss": 1.7176, "step": 2265 }, { "epoch": 0.03, "learning_rate": 9.972884822114789e-05, "loss": 1.7223, "step": 2270 }, { "epoch": 0.03, "learning_rate": 9.97275983713489e-05, "loss": 1.7425, "step": 2275 }, { "epoch": 0.03, "learning_rate": 9.972634565550234e-05, "loss": 1.7471, "step": 2280 }, { "epoch": 0.03, "learning_rate": 9.972509007368039e-05, "loss": 1.6918, "step": 2285 }, { "epoch": 0.03, "learning_rate": 9.972383162595542e-05, "loss": 1.6859, "step": 2290 }, { "epoch": 0.04, "learning_rate": 9.972257031239997e-05, "loss": 1.7018, "step": 2295 }, { "epoch": 0.04, "learning_rate": 9.972130613308672e-05, "loss": 1.7504, "step": 2300 }, { "epoch": 0.04, "learning_rate": 9.972003908808855e-05, "loss": 1.7171, "step": 2305 }, { "epoch": 0.04, "learning_rate": 9.971876917747846e-05, "loss": 1.6689, "step": 2310 }, { "epoch": 0.04, "learning_rate": 9.971749640132966e-05, "loss": 1.7124, "step": 2315 }, { "epoch": 0.04, "learning_rate": 9.971622075971551e-05, "loss": 1.6856, "step": 2320 }, { "epoch": 0.04, "learning_rate": 9.971494225270952e-05, "loss": 1.7149, "step": 2325 }, { "epoch": 0.04, "learning_rate": 9.971366088038537e-05, "loss": 1.6954, "step": 2330 }, { "epoch": 0.04, "learning_rate": 9.971237664281693e-05, "loss": 1.692, "step": 2335 }, { "epoch": 0.04, "learning_rate": 9.971108954007821e-05, "loss": 1.7812, "step": 2340 }, { "epoch": 0.04, "learning_rate": 9.970979957224341e-05, "loss": 1.7724, "step": 2345 }, { "epoch": 0.04, "learning_rate": 9.970850673938684e-05, "loss": 1.6774, "step": 2350 }, { "epoch": 0.04, "learning_rate": 9.970721104158303e-05, "loss": 1.7167, "step": 2355 }, { "epoch": 0.04, "learning_rate": 9.970591247890666e-05, "loss": 1.7181, "step": 2360 }, { "epoch": 0.04, "learning_rate": 9.970461105143256e-05, "loss": 1.7536, "step": 2365 }, { "epoch": 0.04, "learning_rate": 9.970330675923577e-05, "loss": 1.6917, "step": 2370 }, { "epoch": 0.04, "learning_rate": 9.970199960239143e-05, "loss": 1.7046, "step": 2375 }, { "epoch": 0.04, "learning_rate": 9.970068958097489e-05, "loss": 1.6758, "step": 2380 }, { "epoch": 0.04, "learning_rate": 9.969937669506167e-05, "loss": 1.733, "step": 2385 }, { "epoch": 0.04, "learning_rate": 9.969806094472741e-05, "loss": 1.703, "step": 2390 }, { "epoch": 0.04, "learning_rate": 9.969674233004794e-05, "loss": 1.7325, "step": 2395 }, { "epoch": 0.04, "learning_rate": 9.969542085109929e-05, "loss": 1.7406, "step": 2400 }, { "epoch": 0.04, "learning_rate": 9.96940965079576e-05, "loss": 1.7265, "step": 2405 }, { "epoch": 0.04, "learning_rate": 9.96927693006992e-05, "loss": 1.6817, "step": 2410 }, { "epoch": 0.04, "learning_rate": 9.969143922940059e-05, "loss": 1.7034, "step": 2415 }, { "epoch": 0.04, "learning_rate": 9.969010629413842e-05, "loss": 1.7206, "step": 2420 }, { "epoch": 0.04, "learning_rate": 9.968877049498953e-05, "loss": 1.7356, "step": 2425 }, { "epoch": 0.04, "learning_rate": 9.96874318320309e-05, "loss": 1.6914, "step": 2430 }, { "epoch": 0.04, "learning_rate": 9.968609030533968e-05, "loss": 1.6914, "step": 2435 }, { "epoch": 0.04, "learning_rate": 9.96847459149932e-05, "loss": 1.7125, "step": 2440 }, { "epoch": 0.04, "learning_rate": 9.96833986610689e-05, "loss": 1.6795, "step": 2445 }, { "epoch": 0.04, "learning_rate": 9.96820485436445e-05, "loss": 1.7271, "step": 2450 }, { "epoch": 0.04, "learning_rate": 9.968069556279777e-05, "loss": 1.7323, "step": 2455 }, { "epoch": 0.04, "learning_rate": 9.967933971860669e-05, "loss": 1.754, "step": 2460 }, { "epoch": 0.04, "learning_rate": 9.967798101114942e-05, "loss": 1.6902, "step": 2465 }, { "epoch": 0.04, "learning_rate": 9.967661944050425e-05, "loss": 1.6974, "step": 2470 }, { "epoch": 0.04, "learning_rate": 9.967525500674967e-05, "loss": 1.7345, "step": 2475 }, { "epoch": 0.04, "learning_rate": 9.967388770996431e-05, "loss": 1.7518, "step": 2480 }, { "epoch": 0.04, "learning_rate": 9.967251755022698e-05, "loss": 1.6865, "step": 2485 }, { "epoch": 0.04, "learning_rate": 9.967114452761664e-05, "loss": 1.6828, "step": 2490 }, { "epoch": 0.04, "learning_rate": 9.966976864221244e-05, "loss": 1.6945, "step": 2495 }, { "epoch": 0.04, "learning_rate": 9.966838989409366e-05, "loss": 1.6993, "step": 2500 }, { "epoch": 0.04, "learning_rate": 9.966700828333979e-05, "loss": 1.698, "step": 2505 }, { "epoch": 0.04, "learning_rate": 9.966562381003043e-05, "loss": 1.6931, "step": 2510 }, { "epoch": 0.04, "learning_rate": 9.966423647424539e-05, "loss": 1.7247, "step": 2515 }, { "epoch": 0.04, "learning_rate": 9.966284627606462e-05, "loss": 1.7252, "step": 2520 }, { "epoch": 0.04, "learning_rate": 9.966145321556826e-05, "loss": 1.7253, "step": 2525 }, { "epoch": 0.04, "learning_rate": 9.966005729283658e-05, "loss": 1.7511, "step": 2530 }, { "epoch": 0.04, "learning_rate": 9.965865850795005e-05, "loss": 1.7219, "step": 2535 }, { "epoch": 0.04, "learning_rate": 9.965725686098925e-05, "loss": 1.695, "step": 2540 }, { "epoch": 0.04, "learning_rate": 9.965585235203503e-05, "loss": 1.7365, "step": 2545 }, { "epoch": 0.04, "learning_rate": 9.965444498116829e-05, "loss": 1.6781, "step": 2550 }, { "epoch": 0.04, "learning_rate": 9.965303474847018e-05, "loss": 1.6889, "step": 2555 }, { "epoch": 0.04, "learning_rate": 9.965162165402193e-05, "loss": 1.6713, "step": 2560 }, { "epoch": 0.04, "learning_rate": 9.965020569790502e-05, "loss": 1.6923, "step": 2565 }, { "epoch": 0.04, "learning_rate": 9.964878688020105e-05, "loss": 1.6821, "step": 2570 }, { "epoch": 0.04, "learning_rate": 9.964736520099179e-05, "loss": 1.7059, "step": 2575 }, { "epoch": 0.04, "learning_rate": 9.964594066035919e-05, "loss": 1.7268, "step": 2580 }, { "epoch": 0.04, "learning_rate": 9.964451325838533e-05, "loss": 1.7058, "step": 2585 }, { "epoch": 0.04, "learning_rate": 9.96430829951525e-05, "loss": 1.7271, "step": 2590 }, { "epoch": 0.04, "learning_rate": 9.964164987074312e-05, "loss": 1.7047, "step": 2595 }, { "epoch": 0.04, "learning_rate": 9.96402138852398e-05, "loss": 1.723, "step": 2600 }, { "epoch": 0.04, "learning_rate": 9.963877503872529e-05, "loss": 1.6721, "step": 2605 }, { "epoch": 0.04, "learning_rate": 9.963733333128253e-05, "loss": 1.6555, "step": 2610 }, { "epoch": 0.04, "learning_rate": 9.963588876299461e-05, "loss": 1.6772, "step": 2615 }, { "epoch": 0.04, "learning_rate": 9.963444133394477e-05, "loss": 1.704, "step": 2620 }, { "epoch": 0.04, "learning_rate": 9.963299104421646e-05, "loss": 1.714, "step": 2625 }, { "epoch": 0.04, "learning_rate": 9.963153789389325e-05, "loss": 1.705, "step": 2630 }, { "epoch": 0.04, "learning_rate": 9.963008188305889e-05, "loss": 1.686, "step": 2635 }, { "epoch": 0.04, "learning_rate": 9.962862301179732e-05, "loss": 1.7056, "step": 2640 }, { "epoch": 0.04, "learning_rate": 9.96271612801926e-05, "loss": 1.6684, "step": 2645 }, { "epoch": 0.04, "learning_rate": 9.962569668832897e-05, "loss": 1.7061, "step": 2650 }, { "epoch": 0.04, "learning_rate": 9.962422923629087e-05, "loss": 1.7065, "step": 2655 }, { "epoch": 0.04, "learning_rate": 9.962275892416286e-05, "loss": 1.6953, "step": 2660 }, { "epoch": 0.04, "learning_rate": 9.962128575202966e-05, "loss": 1.7078, "step": 2665 }, { "epoch": 0.04, "learning_rate": 9.961980971997622e-05, "loss": 1.6641, "step": 2670 }, { "epoch": 0.04, "learning_rate": 9.961833082808757e-05, "loss": 1.7319, "step": 2675 }, { "epoch": 0.04, "learning_rate": 9.961684907644899e-05, "loss": 1.6761, "step": 2680 }, { "epoch": 0.04, "learning_rate": 9.961536446514585e-05, "loss": 1.7116, "step": 2685 }, { "epoch": 0.04, "learning_rate": 9.961387699426372e-05, "loss": 1.6685, "step": 2690 }, { "epoch": 0.04, "learning_rate": 9.961238666388832e-05, "loss": 1.7076, "step": 2695 }, { "epoch": 0.04, "learning_rate": 9.961089347410558e-05, "loss": 1.7008, "step": 2700 }, { "epoch": 0.04, "learning_rate": 9.96093974250015e-05, "loss": 1.6677, "step": 2705 }, { "epoch": 0.04, "learning_rate": 9.960789851666237e-05, "loss": 1.7001, "step": 2710 }, { "epoch": 0.04, "learning_rate": 9.960639674917455e-05, "loss": 1.6582, "step": 2715 }, { "epoch": 0.04, "learning_rate": 9.960489212262458e-05, "loss": 1.7001, "step": 2720 }, { "epoch": 0.04, "learning_rate": 9.960338463709922e-05, "loss": 1.715, "step": 2725 }, { "epoch": 0.04, "learning_rate": 9.960187429268531e-05, "loss": 1.7339, "step": 2730 }, { "epoch": 0.04, "learning_rate": 9.960036108946991e-05, "loss": 1.6668, "step": 2735 }, { "epoch": 0.04, "learning_rate": 9.959884502754026e-05, "loss": 1.7064, "step": 2740 }, { "epoch": 0.04, "learning_rate": 9.95973261069837e-05, "loss": 1.682, "step": 2745 }, { "epoch": 0.04, "learning_rate": 9.959580432788779e-05, "loss": 1.6941, "step": 2750 }, { "epoch": 0.04, "learning_rate": 9.959427969034025e-05, "loss": 1.6937, "step": 2755 }, { "epoch": 0.04, "learning_rate": 9.959275219442894e-05, "loss": 1.7702, "step": 2760 }, { "epoch": 0.04, "learning_rate": 9.95912218402419e-05, "loss": 1.672, "step": 2765 }, { "epoch": 0.04, "learning_rate": 9.958968862786733e-05, "loss": 1.7075, "step": 2770 }, { "epoch": 0.04, "learning_rate": 9.95881525573936e-05, "loss": 1.6847, "step": 2775 }, { "epoch": 0.04, "learning_rate": 9.958661362890923e-05, "loss": 1.7181, "step": 2780 }, { "epoch": 0.04, "learning_rate": 9.958507184250292e-05, "loss": 1.71, "step": 2785 }, { "epoch": 0.04, "learning_rate": 9.958352719826355e-05, "loss": 1.6719, "step": 2790 }, { "epoch": 0.04, "learning_rate": 9.958197969628012e-05, "loss": 1.7391, "step": 2795 }, { "epoch": 0.04, "learning_rate": 9.958042933664185e-05, "loss": 1.6887, "step": 2800 }, { "epoch": 0.04, "learning_rate": 9.957887611943805e-05, "loss": 1.6857, "step": 2805 }, { "epoch": 0.04, "learning_rate": 9.957732004475827e-05, "loss": 1.7196, "step": 2810 }, { "epoch": 0.04, "learning_rate": 9.957576111269221e-05, "loss": 1.669, "step": 2815 }, { "epoch": 0.04, "learning_rate": 9.957419932332967e-05, "loss": 1.6594, "step": 2820 }, { "epoch": 0.04, "learning_rate": 9.95726346767607e-05, "loss": 1.7169, "step": 2825 }, { "epoch": 0.04, "learning_rate": 9.957106717307548e-05, "loss": 1.6874, "step": 2830 }, { "epoch": 0.04, "learning_rate": 9.956949681236434e-05, "loss": 1.6917, "step": 2835 }, { "epoch": 0.04, "learning_rate": 9.95679235947178e-05, "loss": 1.7001, "step": 2840 }, { "epoch": 0.04, "learning_rate": 9.95663475202265e-05, "loss": 1.7097, "step": 2845 }, { "epoch": 0.04, "learning_rate": 9.956476858898133e-05, "loss": 1.681, "step": 2850 }, { "epoch": 0.04, "learning_rate": 9.956318680107324e-05, "loss": 1.709, "step": 2855 }, { "epoch": 0.04, "learning_rate": 9.956160215659343e-05, "loss": 1.6965, "step": 2860 }, { "epoch": 0.04, "learning_rate": 9.956001465563322e-05, "loss": 1.7003, "step": 2865 }, { "epoch": 0.04, "learning_rate": 9.95584242982841e-05, "loss": 1.7076, "step": 2870 }, { "epoch": 0.04, "learning_rate": 9.955683108463773e-05, "loss": 1.6656, "step": 2875 }, { "epoch": 0.04, "learning_rate": 9.955523501478594e-05, "loss": 1.6667, "step": 2880 }, { "epoch": 0.04, "learning_rate": 9.955363608882074e-05, "loss": 1.738, "step": 2885 }, { "epoch": 0.04, "learning_rate": 9.955203430683424e-05, "loss": 1.7062, "step": 2890 }, { "epoch": 0.04, "learning_rate": 9.955042966891879e-05, "loss": 1.7355, "step": 2895 }, { "epoch": 0.04, "learning_rate": 9.954882217516686e-05, "loss": 1.6858, "step": 2900 }, { "epoch": 0.04, "learning_rate": 9.95472118256711e-05, "loss": 1.7302, "step": 2905 }, { "epoch": 0.04, "learning_rate": 9.954559862052434e-05, "loss": 1.6656, "step": 2910 }, { "epoch": 0.04, "learning_rate": 9.954398255981954e-05, "loss": 1.6588, "step": 2915 }, { "epoch": 0.04, "learning_rate": 9.954236364364984e-05, "loss": 1.7378, "step": 2920 }, { "epoch": 0.04, "learning_rate": 9.954074187210854e-05, "loss": 1.6916, "step": 2925 }, { "epoch": 0.04, "learning_rate": 9.953911724528913e-05, "loss": 1.6656, "step": 2930 }, { "epoch": 0.04, "learning_rate": 9.953748976328523e-05, "loss": 1.6864, "step": 2935 }, { "epoch": 0.04, "learning_rate": 9.953585942619066e-05, "loss": 1.7524, "step": 2940 }, { "epoch": 0.04, "learning_rate": 9.953422623409937e-05, "loss": 1.6887, "step": 2945 }, { "epoch": 0.05, "learning_rate": 9.953259018710548e-05, "loss": 1.6698, "step": 2950 }, { "epoch": 0.05, "learning_rate": 9.953095128530329e-05, "loss": 1.65, "step": 2955 }, { "epoch": 0.05, "learning_rate": 9.952930952878726e-05, "loss": 1.7161, "step": 2960 }, { "epoch": 0.05, "learning_rate": 9.952766491765202e-05, "loss": 1.6942, "step": 2965 }, { "epoch": 0.05, "learning_rate": 9.952601745199235e-05, "loss": 1.6952, "step": 2970 }, { "epoch": 0.05, "learning_rate": 9.95243671319032e-05, "loss": 1.6367, "step": 2975 }, { "epoch": 0.05, "learning_rate": 9.952271395747968e-05, "loss": 1.6685, "step": 2980 }, { "epoch": 0.05, "learning_rate": 9.95210579288171e-05, "loss": 1.6945, "step": 2985 }, { "epoch": 0.05, "learning_rate": 9.951939904601087e-05, "loss": 1.7303, "step": 2990 }, { "epoch": 0.05, "learning_rate": 9.951773730915662e-05, "loss": 1.6688, "step": 2995 }, { "epoch": 0.05, "learning_rate": 9.95160727183501e-05, "loss": 1.6688, "step": 3000 }, { "epoch": 0.05, "learning_rate": 9.951440527368727e-05, "loss": 1.6383, "step": 3005 }, { "epoch": 0.05, "learning_rate": 9.951273497526423e-05, "loss": 1.7344, "step": 3010 }, { "epoch": 0.05, "learning_rate": 9.951106182317726e-05, "loss": 1.6871, "step": 3015 }, { "epoch": 0.05, "learning_rate": 9.950938581752279e-05, "loss": 1.7137, "step": 3020 }, { "epoch": 0.05, "learning_rate": 9.950770695839738e-05, "loss": 1.6865, "step": 3025 }, { "epoch": 0.05, "learning_rate": 9.950602524589781e-05, "loss": 1.6926, "step": 3030 }, { "epoch": 0.05, "learning_rate": 9.950434068012104e-05, "loss": 1.7332, "step": 3035 }, { "epoch": 0.05, "learning_rate": 9.95026532611641e-05, "loss": 1.6519, "step": 3040 }, { "epoch": 0.05, "learning_rate": 9.950096298912431e-05, "loss": 1.7427, "step": 3045 }, { "epoch": 0.05, "learning_rate": 9.949926986409903e-05, "loss": 1.6905, "step": 3050 }, { "epoch": 0.05, "learning_rate": 9.949757388618587e-05, "loss": 1.6518, "step": 3055 }, { "epoch": 0.05, "learning_rate": 9.949587505548259e-05, "loss": 1.659, "step": 3060 }, { "epoch": 0.05, "learning_rate": 9.949417337208708e-05, "loss": 1.7144, "step": 3065 }, { "epoch": 0.05, "learning_rate": 9.949246883609742e-05, "loss": 1.6686, "step": 3070 }, { "epoch": 0.05, "learning_rate": 9.949076144761186e-05, "loss": 1.6847, "step": 3075 }, { "epoch": 0.05, "learning_rate": 9.94890512067288e-05, "loss": 1.6764, "step": 3080 }, { "epoch": 0.05, "learning_rate": 9.94873381135468e-05, "loss": 1.6367, "step": 3085 }, { "epoch": 0.05, "learning_rate": 9.948562216816462e-05, "loss": 1.6531, "step": 3090 }, { "epoch": 0.05, "learning_rate": 9.948390337068113e-05, "loss": 1.7004, "step": 3095 }, { "epoch": 0.05, "learning_rate": 9.94821817211954e-05, "loss": 1.7057, "step": 3100 }, { "epoch": 0.05, "learning_rate": 9.948045721980667e-05, "loss": 1.6806, "step": 3105 }, { "epoch": 0.05, "learning_rate": 9.947872986661433e-05, "loss": 1.7157, "step": 3110 }, { "epoch": 0.05, "learning_rate": 9.94769996617179e-05, "loss": 1.6557, "step": 3115 }, { "epoch": 0.05, "learning_rate": 9.947526660521717e-05, "loss": 1.722, "step": 3120 }, { "epoch": 0.05, "learning_rate": 9.947353069721197e-05, "loss": 1.665, "step": 3125 }, { "epoch": 0.05, "learning_rate": 9.947179193780235e-05, "loss": 1.721, "step": 3130 }, { "epoch": 0.05, "learning_rate": 9.947005032708855e-05, "loss": 1.6763, "step": 3135 }, { "epoch": 0.05, "learning_rate": 9.946830586517094e-05, "loss": 1.6952, "step": 3140 }, { "epoch": 0.05, "learning_rate": 9.946655855215003e-05, "loss": 1.6807, "step": 3145 }, { "epoch": 0.05, "learning_rate": 9.946480838812658e-05, "loss": 1.6736, "step": 3150 }, { "epoch": 0.05, "learning_rate": 9.946305537320142e-05, "loss": 1.7024, "step": 3155 }, { "epoch": 0.05, "learning_rate": 9.94612995074756e-05, "loss": 1.7278, "step": 3160 }, { "epoch": 0.05, "learning_rate": 9.945954079105031e-05, "loss": 1.6766, "step": 3165 }, { "epoch": 0.05, "learning_rate": 9.945777922402693e-05, "loss": 1.6952, "step": 3170 }, { "epoch": 0.05, "learning_rate": 9.945601480650699e-05, "loss": 1.6471, "step": 3175 }, { "epoch": 0.05, "learning_rate": 9.945424753859216e-05, "loss": 1.6607, "step": 3180 }, { "epoch": 0.05, "learning_rate": 9.945247742038431e-05, "loss": 1.6915, "step": 3185 }, { "epoch": 0.05, "learning_rate": 9.945070445198544e-05, "loss": 1.7151, "step": 3190 }, { "epoch": 0.05, "learning_rate": 9.944892863349778e-05, "loss": 1.725, "step": 3195 }, { "epoch": 0.05, "learning_rate": 9.944714996502362e-05, "loss": 1.6527, "step": 3200 }, { "epoch": 0.05, "learning_rate": 9.944536844666553e-05, "loss": 1.7062, "step": 3205 }, { "epoch": 0.05, "learning_rate": 9.944358407852616e-05, "loss": 1.6633, "step": 3210 }, { "epoch": 0.05, "learning_rate": 9.944179686070836e-05, "loss": 1.7105, "step": 3215 }, { "epoch": 0.05, "learning_rate": 9.944000679331511e-05, "loss": 1.6805, "step": 3220 }, { "epoch": 0.05, "learning_rate": 9.943821387644962e-05, "loss": 1.7224, "step": 3225 }, { "epoch": 0.05, "learning_rate": 9.943641811021521e-05, "loss": 1.6579, "step": 3230 }, { "epoch": 0.05, "learning_rate": 9.943461949471537e-05, "loss": 1.6923, "step": 3235 }, { "epoch": 0.05, "learning_rate": 9.943281803005375e-05, "loss": 1.6454, "step": 3240 }, { "epoch": 0.05, "learning_rate": 9.943101371633422e-05, "loss": 1.6774, "step": 3245 }, { "epoch": 0.05, "learning_rate": 9.942920655366074e-05, "loss": 1.6622, "step": 3250 }, { "epoch": 0.05, "learning_rate": 9.942739654213747e-05, "loss": 1.6672, "step": 3255 }, { "epoch": 0.05, "learning_rate": 9.942558368186874e-05, "loss": 1.623, "step": 3260 }, { "epoch": 0.05, "learning_rate": 9.942376797295903e-05, "loss": 1.643, "step": 3265 }, { "epoch": 0.05, "learning_rate": 9.942194941551297e-05, "loss": 1.6798, "step": 3270 }, { "epoch": 0.05, "learning_rate": 9.942012800963541e-05, "loss": 1.6725, "step": 3275 }, { "epoch": 0.05, "learning_rate": 9.941830375543129e-05, "loss": 1.6544, "step": 3280 }, { "epoch": 0.05, "learning_rate": 9.941647665300575e-05, "loss": 1.7155, "step": 3285 }, { "epoch": 0.05, "learning_rate": 9.941464670246413e-05, "loss": 1.6838, "step": 3290 }, { "epoch": 0.05, "learning_rate": 9.941281390391188e-05, "loss": 1.7001, "step": 3295 }, { "epoch": 0.05, "learning_rate": 9.941097825745463e-05, "loss": 1.6839, "step": 3300 }, { "epoch": 0.05, "learning_rate": 9.940913976319818e-05, "loss": 1.7264, "step": 3305 }, { "epoch": 0.05, "learning_rate": 9.940729842124848e-05, "loss": 1.6624, "step": 3310 }, { "epoch": 0.05, "learning_rate": 9.940545423171168e-05, "loss": 1.6681, "step": 3315 }, { "epoch": 0.05, "learning_rate": 9.940360719469406e-05, "loss": 1.6752, "step": 3320 }, { "epoch": 0.05, "learning_rate": 9.940175731030205e-05, "loss": 1.7043, "step": 3325 }, { "epoch": 0.05, "learning_rate": 9.939990457864233e-05, "loss": 1.7058, "step": 3330 }, { "epoch": 0.05, "learning_rate": 9.93980489998216e-05, "loss": 1.6826, "step": 3335 }, { "epoch": 0.05, "learning_rate": 9.939619057394686e-05, "loss": 1.6599, "step": 3340 }, { "epoch": 0.05, "learning_rate": 9.93943293011252e-05, "loss": 1.6739, "step": 3345 }, { "epoch": 0.05, "learning_rate": 9.939246518146392e-05, "loss": 1.657, "step": 3350 }, { "epoch": 0.05, "learning_rate": 9.939059821507044e-05, "loss": 1.6954, "step": 3355 }, { "epoch": 0.05, "learning_rate": 9.938872840205236e-05, "loss": 1.6843, "step": 3360 }, { "epoch": 0.05, "learning_rate": 9.938685574251743e-05, "loss": 1.7121, "step": 3365 }, { "epoch": 0.05, "learning_rate": 9.938498023657362e-05, "loss": 1.6835, "step": 3370 }, { "epoch": 0.05, "learning_rate": 9.9383101884329e-05, "loss": 1.6723, "step": 3375 }, { "epoch": 0.05, "learning_rate": 9.938122068589182e-05, "loss": 1.6933, "step": 3380 }, { "epoch": 0.05, "learning_rate": 9.937933664137054e-05, "loss": 1.6574, "step": 3385 }, { "epoch": 0.05, "learning_rate": 9.93774497508737e-05, "loss": 1.6897, "step": 3390 }, { "epoch": 0.05, "learning_rate": 9.93755600145101e-05, "loss": 1.673, "step": 3395 }, { "epoch": 0.05, "learning_rate": 9.937366743238861e-05, "loss": 1.6494, "step": 3400 }, { "epoch": 0.05, "learning_rate": 9.937177200461831e-05, "loss": 1.6713, "step": 3405 }, { "epoch": 0.05, "learning_rate": 9.936987373130849e-05, "loss": 1.6551, "step": 3410 }, { "epoch": 0.05, "learning_rate": 9.936797261256851e-05, "loss": 1.7471, "step": 3415 }, { "epoch": 0.05, "learning_rate": 9.936606864850798e-05, "loss": 1.699, "step": 3420 }, { "epoch": 0.05, "learning_rate": 9.936416183923658e-05, "loss": 1.7255, "step": 3425 }, { "epoch": 0.05, "learning_rate": 9.936225218486427e-05, "loss": 1.6414, "step": 3430 }, { "epoch": 0.05, "learning_rate": 9.936033968550108e-05, "loss": 1.7098, "step": 3435 }, { "epoch": 0.05, "learning_rate": 9.935842434125723e-05, "loss": 1.683, "step": 3440 }, { "epoch": 0.05, "learning_rate": 9.935650615224314e-05, "loss": 1.6582, "step": 3445 }, { "epoch": 0.05, "learning_rate": 9.935458511856933e-05, "loss": 1.7079, "step": 3450 }, { "epoch": 0.05, "learning_rate": 9.935266124034654e-05, "loss": 1.6433, "step": 3455 }, { "epoch": 0.05, "learning_rate": 9.935073451768566e-05, "loss": 1.7153, "step": 3460 }, { "epoch": 0.05, "learning_rate": 9.934880495069772e-05, "loss": 1.738, "step": 3465 }, { "epoch": 0.05, "learning_rate": 9.934687253949392e-05, "loss": 1.6697, "step": 3470 }, { "epoch": 0.05, "learning_rate": 9.934493728418565e-05, "loss": 1.6667, "step": 3475 }, { "epoch": 0.05, "learning_rate": 9.934299918488447e-05, "loss": 1.7222, "step": 3480 }, { "epoch": 0.05, "learning_rate": 9.934105824170205e-05, "loss": 1.6176, "step": 3485 }, { "epoch": 0.05, "learning_rate": 9.933911445475028e-05, "loss": 1.6533, "step": 3490 }, { "epoch": 0.05, "learning_rate": 9.933716782414115e-05, "loss": 1.7075, "step": 3495 }, { "epoch": 0.05, "learning_rate": 9.933521834998689e-05, "loss": 1.6901, "step": 3500 }, { "epoch": 0.05, "learning_rate": 9.933326603239985e-05, "loss": 1.6388, "step": 3505 }, { "epoch": 0.05, "learning_rate": 9.933131087149256e-05, "loss": 1.6378, "step": 3510 }, { "epoch": 0.05, "learning_rate": 9.932935286737768e-05, "loss": 1.6249, "step": 3515 }, { "epoch": 0.05, "learning_rate": 9.932739202016808e-05, "loss": 1.6181, "step": 3520 }, { "epoch": 0.05, "learning_rate": 9.932542832997678e-05, "loss": 1.684, "step": 3525 }, { "epoch": 0.05, "learning_rate": 9.932346179691693e-05, "loss": 1.6836, "step": 3530 }, { "epoch": 0.05, "learning_rate": 9.93214924211019e-05, "loss": 1.7247, "step": 3535 }, { "epoch": 0.05, "learning_rate": 9.931952020264516e-05, "loss": 1.7111, "step": 3540 }, { "epoch": 0.05, "learning_rate": 9.931754514166044e-05, "loss": 1.6466, "step": 3545 }, { "epoch": 0.05, "learning_rate": 9.93155672382615e-05, "loss": 1.6805, "step": 3550 }, { "epoch": 0.05, "learning_rate": 9.931358649256238e-05, "loss": 1.6585, "step": 3555 }, { "epoch": 0.05, "learning_rate": 9.931160290467723e-05, "loss": 1.6569, "step": 3560 }, { "epoch": 0.05, "learning_rate": 9.930961647472037e-05, "loss": 1.6909, "step": 3565 }, { "epoch": 0.05, "learning_rate": 9.930762720280628e-05, "loss": 1.7205, "step": 3570 }, { "epoch": 0.05, "learning_rate": 9.930563508904964e-05, "loss": 1.7042, "step": 3575 }, { "epoch": 0.05, "learning_rate": 9.930364013356524e-05, "loss": 1.6687, "step": 3580 }, { "epoch": 0.05, "learning_rate": 9.930164233646808e-05, "loss": 1.6748, "step": 3585 }, { "epoch": 0.05, "learning_rate": 9.929964169787326e-05, "loss": 1.6705, "step": 3590 }, { "epoch": 0.05, "learning_rate": 9.929763821789616e-05, "loss": 1.7015, "step": 3595 }, { "epoch": 0.05, "learning_rate": 9.929563189665217e-05, "loss": 1.6601, "step": 3600 }, { "epoch": 0.06, "learning_rate": 9.929362273425697e-05, "loss": 1.6903, "step": 3605 }, { "epoch": 0.06, "learning_rate": 9.929161073082635e-05, "loss": 1.7043, "step": 3610 }, { "epoch": 0.06, "learning_rate": 9.928959588647629e-05, "loss": 1.6833, "step": 3615 }, { "epoch": 0.06, "learning_rate": 9.928757820132288e-05, "loss": 1.6588, "step": 3620 }, { "epoch": 0.06, "learning_rate": 9.928555767548244e-05, "loss": 1.6883, "step": 3625 }, { "epoch": 0.06, "learning_rate": 9.928353430907138e-05, "loss": 1.684, "step": 3630 }, { "epoch": 0.06, "learning_rate": 9.928150810220637e-05, "loss": 1.7293, "step": 3635 }, { "epoch": 0.06, "learning_rate": 9.927947905500417e-05, "loss": 1.6588, "step": 3640 }, { "epoch": 0.06, "learning_rate": 9.92774471675817e-05, "loss": 1.6015, "step": 3645 }, { "epoch": 0.06, "learning_rate": 9.92754124400561e-05, "loss": 1.6737, "step": 3650 }, { "epoch": 0.06, "learning_rate": 9.927337487254462e-05, "loss": 1.674, "step": 3655 }, { "epoch": 0.06, "learning_rate": 9.927133446516472e-05, "loss": 1.6961, "step": 3660 }, { "epoch": 0.06, "learning_rate": 9.926929121803397e-05, "loss": 1.697, "step": 3665 }, { "epoch": 0.06, "learning_rate": 9.926724513127015e-05, "loss": 1.6265, "step": 3670 }, { "epoch": 0.06, "learning_rate": 9.926519620499117e-05, "loss": 1.682, "step": 3675 }, { "epoch": 0.06, "learning_rate": 9.926314443931515e-05, "loss": 1.6695, "step": 3680 }, { "epoch": 0.06, "learning_rate": 9.926108983436031e-05, "loss": 1.6512, "step": 3685 }, { "epoch": 0.06, "learning_rate": 9.925903239024509e-05, "loss": 1.664, "step": 3690 }, { "epoch": 0.06, "learning_rate": 9.925697210708806e-05, "loss": 1.625, "step": 3695 }, { "epoch": 0.06, "learning_rate": 9.925490898500796e-05, "loss": 1.6366, "step": 3700 }, { "epoch": 0.06, "learning_rate": 9.925284302412373e-05, "loss": 1.6829, "step": 3705 }, { "epoch": 0.06, "learning_rate": 9.925077422455438e-05, "loss": 1.7007, "step": 3710 }, { "epoch": 0.06, "learning_rate": 9.92487025864192e-05, "loss": 1.6757, "step": 3715 }, { "epoch": 0.06, "learning_rate": 9.924662810983756e-05, "loss": 1.6819, "step": 3720 }, { "epoch": 0.06, "learning_rate": 9.924455079492905e-05, "loss": 1.697, "step": 3725 }, { "epoch": 0.06, "learning_rate": 9.924247064181336e-05, "loss": 1.6831, "step": 3730 }, { "epoch": 0.06, "learning_rate": 9.924038765061042e-05, "loss": 1.6715, "step": 3735 }, { "epoch": 0.06, "learning_rate": 9.923830182144022e-05, "loss": 1.6923, "step": 3740 }, { "epoch": 0.06, "learning_rate": 9.923621315442306e-05, "loss": 1.7057, "step": 3745 }, { "epoch": 0.06, "learning_rate": 9.923412164967927e-05, "loss": 1.64, "step": 3750 }, { "epoch": 0.06, "learning_rate": 9.923202730732939e-05, "loss": 1.6264, "step": 3755 }, { "epoch": 0.06, "learning_rate": 9.922993012749413e-05, "loss": 1.6537, "step": 3760 }, { "epoch": 0.06, "learning_rate": 9.922783011029439e-05, "loss": 1.6642, "step": 3765 }, { "epoch": 0.06, "learning_rate": 9.922572725585118e-05, "loss": 1.6397, "step": 3770 }, { "epoch": 0.06, "learning_rate": 9.92236215642857e-05, "loss": 1.6598, "step": 3775 }, { "epoch": 0.06, "learning_rate": 9.922151303571931e-05, "loss": 1.6912, "step": 3780 }, { "epoch": 0.06, "learning_rate": 9.921940167027354e-05, "loss": 1.6747, "step": 3785 }, { "epoch": 0.06, "learning_rate": 9.921728746807007e-05, "loss": 1.6992, "step": 3790 }, { "epoch": 0.06, "learning_rate": 9.921517042923076e-05, "loss": 1.6504, "step": 3795 }, { "epoch": 0.06, "learning_rate": 9.921305055387764e-05, "loss": 1.6478, "step": 3800 }, { "epoch": 0.06, "learning_rate": 9.921092784213286e-05, "loss": 1.6769, "step": 3805 }, { "epoch": 0.06, "learning_rate": 9.920880229411878e-05, "loss": 1.7077, "step": 3810 }, { "epoch": 0.06, "learning_rate": 9.92066739099579e-05, "loss": 1.6987, "step": 3815 }, { "epoch": 0.06, "learning_rate": 9.920454268977289e-05, "loss": 1.6842, "step": 3820 }, { "epoch": 0.06, "learning_rate": 9.92024086336866e-05, "loss": 1.6365, "step": 3825 }, { "epoch": 0.06, "learning_rate": 9.920027174182198e-05, "loss": 1.6633, "step": 3830 }, { "epoch": 0.06, "learning_rate": 9.919813201430224e-05, "loss": 1.6855, "step": 3835 }, { "epoch": 0.06, "learning_rate": 9.919598945125067e-05, "loss": 1.7212, "step": 3840 }, { "epoch": 0.06, "learning_rate": 9.919384405279078e-05, "loss": 1.635, "step": 3845 }, { "epoch": 0.06, "learning_rate": 9.919169581904619e-05, "loss": 1.6229, "step": 3850 }, { "epoch": 0.06, "learning_rate": 9.918954475014075e-05, "loss": 1.666, "step": 3855 }, { "epoch": 0.06, "learning_rate": 9.918739084619841e-05, "loss": 1.6673, "step": 3860 }, { "epoch": 0.06, "learning_rate": 9.918523410734333e-05, "loss": 1.7062, "step": 3865 }, { "epoch": 0.06, "learning_rate": 9.91830745336998e-05, "loss": 1.6932, "step": 3870 }, { "epoch": 0.06, "learning_rate": 9.918091212539227e-05, "loss": 1.6628, "step": 3875 }, { "epoch": 0.06, "learning_rate": 9.917874688254541e-05, "loss": 1.6689, "step": 3880 }, { "epoch": 0.06, "learning_rate": 9.917657880528399e-05, "loss": 1.6542, "step": 3885 }, { "epoch": 0.06, "learning_rate": 9.917440789373297e-05, "loss": 1.6641, "step": 3890 }, { "epoch": 0.06, "learning_rate": 9.917223414801747e-05, "loss": 1.6795, "step": 3895 }, { "epoch": 0.06, "learning_rate": 9.917005756826277e-05, "loss": 1.7124, "step": 3900 }, { "epoch": 0.06, "learning_rate": 9.916787815459432e-05, "loss": 1.6954, "step": 3905 }, { "epoch": 0.06, "learning_rate": 9.916569590713775e-05, "loss": 1.6836, "step": 3910 }, { "epoch": 0.06, "learning_rate": 9.91635108260188e-05, "loss": 1.6803, "step": 3915 }, { "epoch": 0.06, "learning_rate": 9.916132291136341e-05, "loss": 1.5993, "step": 3920 }, { "epoch": 0.06, "learning_rate": 9.915913216329773e-05, "loss": 1.6598, "step": 3925 }, { "epoch": 0.06, "learning_rate": 9.915693858194796e-05, "loss": 1.6285, "step": 3930 }, { "epoch": 0.06, "learning_rate": 9.915474216744056e-05, "loss": 1.6284, "step": 3935 }, { "epoch": 0.06, "learning_rate": 9.915254291990211e-05, "loss": 1.6513, "step": 3940 }, { "epoch": 0.06, "learning_rate": 9.915034083945938e-05, "loss": 1.6885, "step": 3945 }, { "epoch": 0.06, "learning_rate": 9.914813592623927e-05, "loss": 1.6839, "step": 3950 }, { "epoch": 0.06, "learning_rate": 9.914592818036885e-05, "loss": 1.6203, "step": 3955 }, { "epoch": 0.06, "learning_rate": 9.914371760197538e-05, "loss": 1.6961, "step": 3960 }, { "epoch": 0.06, "learning_rate": 9.914150419118627e-05, "loss": 1.656, "step": 3965 }, { "epoch": 0.06, "learning_rate": 9.913928794812909e-05, "loss": 1.6482, "step": 3970 }, { "epoch": 0.06, "learning_rate": 9.913706887293156e-05, "loss": 1.7352, "step": 3975 }, { "epoch": 0.06, "learning_rate": 9.91348469657216e-05, "loss": 1.6514, "step": 3980 }, { "epoch": 0.06, "learning_rate": 9.913262222662722e-05, "loss": 1.6654, "step": 3985 }, { "epoch": 0.06, "learning_rate": 9.913039465577671e-05, "loss": 1.6103, "step": 3990 }, { "epoch": 0.06, "learning_rate": 9.91281642532984e-05, "loss": 1.6505, "step": 3995 }, { "epoch": 0.06, "learning_rate": 9.912593101932087e-05, "loss": 1.6038, "step": 4000 }, { "epoch": 0.06, "learning_rate": 9.91236949539728e-05, "loss": 1.7014, "step": 4005 }, { "epoch": 0.06, "learning_rate": 9.912145605738312e-05, "loss": 1.6518, "step": 4010 }, { "epoch": 0.06, "learning_rate": 9.911921432968083e-05, "loss": 1.6361, "step": 4015 }, { "epoch": 0.06, "learning_rate": 9.911696977099513e-05, "loss": 1.6887, "step": 4020 }, { "epoch": 0.06, "learning_rate": 9.911472238145541e-05, "loss": 1.6719, "step": 4025 }, { "epoch": 0.06, "learning_rate": 9.911247216119116e-05, "loss": 1.6399, "step": 4030 }, { "epoch": 0.06, "learning_rate": 9.91102191103321e-05, "loss": 1.69, "step": 4035 }, { "epoch": 0.06, "learning_rate": 9.910796322900808e-05, "loss": 1.668, "step": 4040 }, { "epoch": 0.06, "learning_rate": 9.910570451734913e-05, "loss": 1.6508, "step": 4045 }, { "epoch": 0.06, "learning_rate": 9.91034429754854e-05, "loss": 1.6726, "step": 4050 }, { "epoch": 0.06, "learning_rate": 9.910117860354725e-05, "loss": 1.695, "step": 4055 }, { "epoch": 0.06, "learning_rate": 9.909891140166519e-05, "loss": 1.6363, "step": 4060 }, { "epoch": 0.06, "learning_rate": 9.90966413699699e-05, "loss": 1.6609, "step": 4065 }, { "epoch": 0.06, "learning_rate": 9.909436850859218e-05, "loss": 1.6858, "step": 4070 }, { "epoch": 0.06, "learning_rate": 9.909209281766305e-05, "loss": 1.6845, "step": 4075 }, { "epoch": 0.06, "learning_rate": 9.908981429731367e-05, "loss": 1.6789, "step": 4080 }, { "epoch": 0.06, "learning_rate": 9.908753294767537e-05, "loss": 1.6949, "step": 4085 }, { "epoch": 0.06, "learning_rate": 9.908524876887962e-05, "loss": 1.6272, "step": 4090 }, { "epoch": 0.06, "learning_rate": 9.908296176105807e-05, "loss": 1.645, "step": 4095 }, { "epoch": 0.06, "learning_rate": 9.908067192434252e-05, "loss": 1.6858, "step": 4100 }, { "epoch": 0.06, "learning_rate": 9.907837925886497e-05, "loss": 1.7205, "step": 4105 }, { "epoch": 0.06, "learning_rate": 9.907608376475754e-05, "loss": 1.6412, "step": 4110 }, { "epoch": 0.06, "learning_rate": 9.907378544215255e-05, "loss": 1.6683, "step": 4115 }, { "epoch": 0.06, "learning_rate": 9.907148429118245e-05, "loss": 1.6499, "step": 4120 }, { "epoch": 0.06, "learning_rate": 9.906918031197985e-05, "loss": 1.6385, "step": 4125 }, { "epoch": 0.06, "learning_rate": 9.906687350467757e-05, "loss": 1.6879, "step": 4130 }, { "epoch": 0.06, "learning_rate": 9.906456386940855e-05, "loss": 1.6618, "step": 4135 }, { "epoch": 0.06, "learning_rate": 9.90622514063059e-05, "loss": 1.675, "step": 4140 }, { "epoch": 0.06, "learning_rate": 9.90599361155029e-05, "loss": 1.6305, "step": 4145 }, { "epoch": 0.06, "learning_rate": 9.905761799713301e-05, "loss": 1.718, "step": 4150 }, { "epoch": 0.06, "learning_rate": 9.905529705132982e-05, "loss": 1.6067, "step": 4155 }, { "epoch": 0.06, "learning_rate": 9.90529732782271e-05, "loss": 1.7341, "step": 4160 }, { "epoch": 0.06, "learning_rate": 9.905064667795876e-05, "loss": 1.6582, "step": 4165 }, { "epoch": 0.06, "learning_rate": 9.904831725065892e-05, "loss": 1.6532, "step": 4170 }, { "epoch": 0.06, "learning_rate": 9.904598499646182e-05, "loss": 1.7045, "step": 4175 }, { "epoch": 0.06, "learning_rate": 9.904364991550192e-05, "loss": 1.6815, "step": 4180 }, { "epoch": 0.06, "learning_rate": 9.904131200791373e-05, "loss": 1.6553, "step": 4185 }, { "epoch": 0.06, "learning_rate": 9.903897127383206e-05, "loss": 1.6549, "step": 4190 }, { "epoch": 0.06, "learning_rate": 9.90366277133918e-05, "loss": 1.6443, "step": 4195 }, { "epoch": 0.06, "learning_rate": 9.9034281326728e-05, "loss": 1.6661, "step": 4200 }, { "epoch": 0.06, "learning_rate": 9.90319321139759e-05, "loss": 1.6501, "step": 4205 }, { "epoch": 0.06, "learning_rate": 9.902958007527093e-05, "loss": 1.6233, "step": 4210 }, { "epoch": 0.06, "learning_rate": 9.902722521074861e-05, "loss": 1.6404, "step": 4215 }, { "epoch": 0.06, "learning_rate": 9.902486752054468e-05, "loss": 1.6618, "step": 4220 }, { "epoch": 0.06, "learning_rate": 9.9022507004795e-05, "loss": 1.6114, "step": 4225 }, { "epoch": 0.06, "learning_rate": 9.902014366363567e-05, "loss": 1.6374, "step": 4230 }, { "epoch": 0.06, "learning_rate": 9.901777749720287e-05, "loss": 1.6664, "step": 4235 }, { "epoch": 0.06, "learning_rate": 9.901540850563295e-05, "loss": 1.6498, "step": 4240 }, { "epoch": 0.06, "learning_rate": 9.90130366890625e-05, "loss": 1.6399, "step": 4245 }, { "epoch": 0.06, "learning_rate": 9.901066204762819e-05, "loss": 1.6877, "step": 4250 }, { "epoch": 0.06, "learning_rate": 9.900828458146687e-05, "loss": 1.6408, "step": 4255 }, { "epoch": 0.07, "learning_rate": 9.900590429071558e-05, "loss": 1.6574, "step": 4260 }, { "epoch": 0.07, "learning_rate": 9.900352117551152e-05, "loss": 1.6527, "step": 4265 }, { "epoch": 0.07, "learning_rate": 9.900113523599202e-05, "loss": 1.615, "step": 4270 }, { "epoch": 0.07, "learning_rate": 9.899874647229461e-05, "loss": 1.687, "step": 4275 }, { "epoch": 0.07, "learning_rate": 9.899635488455695e-05, "loss": 1.6789, "step": 4280 }, { "epoch": 0.07, "learning_rate": 9.899396047291689e-05, "loss": 1.6953, "step": 4285 }, { "epoch": 0.07, "learning_rate": 9.899156323751243e-05, "loss": 1.6587, "step": 4290 }, { "epoch": 0.07, "learning_rate": 9.898916317848172e-05, "loss": 1.6231, "step": 4295 }, { "epoch": 0.07, "learning_rate": 9.898676029596313e-05, "loss": 1.6964, "step": 4300 }, { "epoch": 0.07, "learning_rate": 9.898435459009509e-05, "loss": 1.7068, "step": 4305 }, { "epoch": 0.07, "learning_rate": 9.898194606101629e-05, "loss": 1.6352, "step": 4310 }, { "epoch": 0.07, "learning_rate": 9.897953470886555e-05, "loss": 1.5901, "step": 4315 }, { "epoch": 0.07, "learning_rate": 9.897712053378183e-05, "loss": 1.6731, "step": 4320 }, { "epoch": 0.07, "learning_rate": 9.897470353590428e-05, "loss": 1.6959, "step": 4325 }, { "epoch": 0.07, "learning_rate": 9.89722837153722e-05, "loss": 1.6076, "step": 4330 }, { "epoch": 0.07, "learning_rate": 9.896986107232505e-05, "loss": 1.6568, "step": 4335 }, { "epoch": 0.07, "learning_rate": 9.896743560690247e-05, "loss": 1.6412, "step": 4340 }, { "epoch": 0.07, "learning_rate": 9.896500731924425e-05, "loss": 1.6306, "step": 4345 }, { "epoch": 0.07, "learning_rate": 9.896257620949034e-05, "loss": 1.6383, "step": 4350 }, { "epoch": 0.07, "learning_rate": 9.896014227778088e-05, "loss": 1.6562, "step": 4355 }, { "epoch": 0.07, "learning_rate": 9.895770552425609e-05, "loss": 1.6415, "step": 4360 }, { "epoch": 0.07, "learning_rate": 9.895526594905648e-05, "loss": 1.6624, "step": 4365 }, { "epoch": 0.07, "learning_rate": 9.895282355232262e-05, "loss": 1.6551, "step": 4370 }, { "epoch": 0.07, "learning_rate": 9.895037833419528e-05, "loss": 1.6777, "step": 4375 }, { "epoch": 0.07, "learning_rate": 9.89479302948154e-05, "loss": 1.614, "step": 4380 }, { "epoch": 0.07, "learning_rate": 9.894547943432405e-05, "loss": 1.6629, "step": 4385 }, { "epoch": 0.07, "learning_rate": 9.894302575286252e-05, "loss": 1.6497, "step": 4390 }, { "epoch": 0.07, "learning_rate": 9.894056925057219e-05, "loss": 1.6131, "step": 4395 }, { "epoch": 0.07, "learning_rate": 9.893810992759466e-05, "loss": 1.6533, "step": 4400 }, { "epoch": 0.07, "learning_rate": 9.893564778407168e-05, "loss": 1.6538, "step": 4405 }, { "epoch": 0.07, "learning_rate": 9.893318282014515e-05, "loss": 1.6082, "step": 4410 }, { "epoch": 0.07, "learning_rate": 9.893071503595714e-05, "loss": 1.6678, "step": 4415 }, { "epoch": 0.07, "learning_rate": 9.892824443164986e-05, "loss": 1.6535, "step": 4420 }, { "epoch": 0.07, "learning_rate": 9.892577100736573e-05, "loss": 1.6098, "step": 4425 }, { "epoch": 0.07, "learning_rate": 9.892329476324729e-05, "loss": 1.6313, "step": 4430 }, { "epoch": 0.07, "learning_rate": 9.892081569943727e-05, "loss": 1.6337, "step": 4435 }, { "epoch": 0.07, "learning_rate": 9.891833381607853e-05, "loss": 1.6194, "step": 4440 }, { "epoch": 0.07, "learning_rate": 9.891584911331415e-05, "loss": 1.6581, "step": 4445 }, { "epoch": 0.07, "learning_rate": 9.89133615912873e-05, "loss": 1.6217, "step": 4450 }, { "epoch": 0.07, "learning_rate": 9.891087125014135e-05, "loss": 1.6524, "step": 4455 }, { "epoch": 0.07, "learning_rate": 9.890837809001986e-05, "loss": 1.6769, "step": 4460 }, { "epoch": 0.07, "learning_rate": 9.89058821110665e-05, "loss": 1.6944, "step": 4465 }, { "epoch": 0.07, "learning_rate": 9.890338331342512e-05, "loss": 1.6673, "step": 4470 }, { "epoch": 0.07, "learning_rate": 9.890088169723977e-05, "loss": 1.6319, "step": 4475 }, { "epoch": 0.07, "learning_rate": 9.889837726265459e-05, "loss": 1.6899, "step": 4480 }, { "epoch": 0.07, "learning_rate": 9.889587000981395e-05, "loss": 1.659, "step": 4485 }, { "epoch": 0.07, "learning_rate": 9.889335993886236e-05, "loss": 1.6988, "step": 4490 }, { "epoch": 0.07, "learning_rate": 9.889084704994445e-05, "loss": 1.6884, "step": 4495 }, { "epoch": 0.07, "learning_rate": 9.888833134320508e-05, "loss": 1.6476, "step": 4500 }, { "epoch": 0.07, "learning_rate": 9.888581281878924e-05, "loss": 1.6442, "step": 4505 }, { "epoch": 0.07, "learning_rate": 9.88832914768421e-05, "loss": 1.626, "step": 4510 }, { "epoch": 0.07, "learning_rate": 9.888076731750895e-05, "loss": 1.6633, "step": 4515 }, { "epoch": 0.07, "learning_rate": 9.887824034093528e-05, "loss": 1.657, "step": 4520 }, { "epoch": 0.07, "learning_rate": 9.887571054726671e-05, "loss": 1.6245, "step": 4525 }, { "epoch": 0.07, "learning_rate": 9.88731779366491e-05, "loss": 1.6059, "step": 4530 }, { "epoch": 0.07, "learning_rate": 9.887064250922838e-05, "loss": 1.6525, "step": 4535 }, { "epoch": 0.07, "learning_rate": 9.886810426515067e-05, "loss": 1.6355, "step": 4540 }, { "epoch": 0.07, "learning_rate": 9.886556320456229e-05, "loss": 1.6738, "step": 4545 }, { "epoch": 0.07, "learning_rate": 9.886301932760967e-05, "loss": 1.6832, "step": 4550 }, { "epoch": 0.07, "learning_rate": 9.886047263443942e-05, "loss": 1.6096, "step": 4555 }, { "epoch": 0.07, "learning_rate": 9.885792312519834e-05, "loss": 1.6781, "step": 4560 }, { "epoch": 0.07, "learning_rate": 9.885537080003338e-05, "loss": 1.6187, "step": 4565 }, { "epoch": 0.07, "learning_rate": 9.885281565909161e-05, "loss": 1.6352, "step": 4570 }, { "epoch": 0.07, "learning_rate": 9.885025770252032e-05, "loss": 1.6163, "step": 4575 }, { "epoch": 0.07, "learning_rate": 9.884769693046693e-05, "loss": 1.7097, "step": 4580 }, { "epoch": 0.07, "learning_rate": 9.884513334307904e-05, "loss": 1.6856, "step": 4585 }, { "epoch": 0.07, "learning_rate": 9.884256694050439e-05, "loss": 1.6707, "step": 4590 }, { "epoch": 0.07, "learning_rate": 9.88399977228909e-05, "loss": 1.6025, "step": 4595 }, { "epoch": 0.07, "learning_rate": 9.883742569038663e-05, "loss": 1.6671, "step": 4600 }, { "epoch": 0.07, "learning_rate": 9.883485084313984e-05, "loss": 1.6106, "step": 4605 }, { "epoch": 0.07, "learning_rate": 9.883227318129892e-05, "loss": 1.6619, "step": 4610 }, { "epoch": 0.07, "learning_rate": 9.882969270501244e-05, "loss": 1.692, "step": 4615 }, { "epoch": 0.07, "learning_rate": 9.882710941442912e-05, "loss": 1.6641, "step": 4620 }, { "epoch": 0.07, "learning_rate": 9.882452330969786e-05, "loss": 1.6666, "step": 4625 }, { "epoch": 0.07, "learning_rate": 9.88219343909677e-05, "loss": 1.6506, "step": 4630 }, { "epoch": 0.07, "learning_rate": 9.881934265838783e-05, "loss": 1.6282, "step": 4635 }, { "epoch": 0.07, "learning_rate": 9.881674811210769e-05, "loss": 1.6391, "step": 4640 }, { "epoch": 0.07, "learning_rate": 9.881415075227673e-05, "loss": 1.6871, "step": 4645 }, { "epoch": 0.07, "learning_rate": 9.881155057904472e-05, "loss": 1.644, "step": 4650 }, { "epoch": 0.07, "learning_rate": 9.880894759256149e-05, "loss": 1.6549, "step": 4655 }, { "epoch": 0.07, "learning_rate": 9.880634179297705e-05, "loss": 1.6379, "step": 4660 }, { "epoch": 0.07, "learning_rate": 9.880373318044162e-05, "loss": 1.5833, "step": 4665 }, { "epoch": 0.07, "learning_rate": 9.880112175510551e-05, "loss": 1.6268, "step": 4670 }, { "epoch": 0.07, "learning_rate": 9.879850751711927e-05, "loss": 1.6878, "step": 4675 }, { "epoch": 0.07, "learning_rate": 9.879589046663355e-05, "loss": 1.6722, "step": 4680 }, { "epoch": 0.07, "learning_rate": 9.879327060379918e-05, "loss": 1.6232, "step": 4685 }, { "epoch": 0.07, "learning_rate": 9.879064792876715e-05, "loss": 1.6832, "step": 4690 }, { "epoch": 0.07, "learning_rate": 9.878802244168863e-05, "loss": 1.6488, "step": 4695 }, { "epoch": 0.07, "learning_rate": 9.878539414271492e-05, "loss": 1.5943, "step": 4700 }, { "epoch": 0.07, "learning_rate": 9.878276303199755e-05, "loss": 1.6438, "step": 4705 }, { "epoch": 0.07, "learning_rate": 9.878012910968812e-05, "loss": 1.6378, "step": 4710 }, { "epoch": 0.07, "learning_rate": 9.877749237593844e-05, "loss": 1.6676, "step": 4715 }, { "epoch": 0.07, "learning_rate": 9.877485283090049e-05, "loss": 1.6606, "step": 4720 }, { "epoch": 0.07, "learning_rate": 9.87722104747264e-05, "loss": 1.627, "step": 4725 }, { "epoch": 0.07, "learning_rate": 9.876956530756847e-05, "loss": 1.6856, "step": 4730 }, { "epoch": 0.07, "learning_rate": 9.876691732957913e-05, "loss": 1.6721, "step": 4735 }, { "epoch": 0.07, "learning_rate": 9.8764266540911e-05, "loss": 1.6433, "step": 4740 }, { "epoch": 0.07, "learning_rate": 9.876161294171687e-05, "loss": 1.6665, "step": 4745 }, { "epoch": 0.07, "learning_rate": 9.875895653214968e-05, "loss": 1.662, "step": 4750 }, { "epoch": 0.07, "learning_rate": 9.875629731236253e-05, "loss": 1.6281, "step": 4755 }, { "epoch": 0.07, "learning_rate": 9.87536352825087e-05, "loss": 1.5997, "step": 4760 }, { "epoch": 0.07, "learning_rate": 9.875097044274158e-05, "loss": 1.7167, "step": 4765 }, { "epoch": 0.07, "learning_rate": 9.874830279321476e-05, "loss": 1.6679, "step": 4770 }, { "epoch": 0.07, "learning_rate": 9.874563233408203e-05, "loss": 1.6484, "step": 4775 }, { "epoch": 0.07, "learning_rate": 9.874295906549728e-05, "loss": 1.6013, "step": 4780 }, { "epoch": 0.07, "learning_rate": 9.874028298761458e-05, "loss": 1.6241, "step": 4785 }, { "epoch": 0.07, "learning_rate": 9.873760410058816e-05, "loss": 1.6632, "step": 4790 }, { "epoch": 0.07, "learning_rate": 9.873492240457242e-05, "loss": 1.653, "step": 4795 }, { "epoch": 0.07, "learning_rate": 9.873223789972194e-05, "loss": 1.6212, "step": 4800 }, { "epoch": 0.07, "learning_rate": 9.872955058619142e-05, "loss": 1.628, "step": 4805 }, { "epoch": 0.07, "learning_rate": 9.872686046413574e-05, "loss": 1.6372, "step": 4810 }, { "epoch": 0.07, "learning_rate": 9.872416753370997e-05, "loss": 1.6752, "step": 4815 }, { "epoch": 0.07, "learning_rate": 9.872147179506927e-05, "loss": 1.6835, "step": 4820 }, { "epoch": 0.07, "learning_rate": 9.871877324836906e-05, "loss": 1.6783, "step": 4825 }, { "epoch": 0.07, "learning_rate": 9.871607189376486e-05, "loss": 1.6618, "step": 4830 }, { "epoch": 0.07, "learning_rate": 9.871336773141235e-05, "loss": 1.623, "step": 4835 }, { "epoch": 0.07, "learning_rate": 9.871066076146737e-05, "loss": 1.6305, "step": 4840 }, { "epoch": 0.07, "learning_rate": 9.870795098408598e-05, "loss": 1.6262, "step": 4845 }, { "epoch": 0.07, "learning_rate": 9.870523839942432e-05, "loss": 1.659, "step": 4850 }, { "epoch": 0.07, "learning_rate": 9.870252300763872e-05, "loss": 1.66, "step": 4855 }, { "epoch": 0.07, "learning_rate": 9.869980480888572e-05, "loss": 1.6858, "step": 4860 }, { "epoch": 0.07, "learning_rate": 9.869708380332197e-05, "loss": 1.6116, "step": 4865 }, { "epoch": 0.07, "learning_rate": 9.869435999110428e-05, "loss": 1.6041, "step": 4870 }, { "epoch": 0.07, "learning_rate": 9.869163337238965e-05, "loss": 1.6423, "step": 4875 }, { "epoch": 0.07, "learning_rate": 9.868890394733523e-05, "loss": 1.6381, "step": 4880 }, { "epoch": 0.07, "learning_rate": 9.868617171609831e-05, "loss": 1.6229, "step": 4885 }, { "epoch": 0.07, "learning_rate": 9.868343667883639e-05, "loss": 1.5839, "step": 4890 }, { "epoch": 0.07, "learning_rate": 9.86806988357071e-05, "loss": 1.648, "step": 4895 }, { "epoch": 0.07, "learning_rate": 9.867795818686822e-05, "loss": 1.6629, "step": 4900 }, { "epoch": 0.07, "learning_rate": 9.867521473247772e-05, "loss": 1.6676, "step": 4905 }, { "epoch": 0.07, "learning_rate": 9.86724684726937e-05, "loss": 1.6168, "step": 4910 }, { "epoch": 0.08, "learning_rate": 9.866971940767446e-05, "loss": 1.6793, "step": 4915 }, { "epoch": 0.08, "learning_rate": 9.866696753757844e-05, "loss": 1.6235, "step": 4920 }, { "epoch": 0.08, "learning_rate": 9.866421286256423e-05, "loss": 1.6152, "step": 4925 }, { "epoch": 0.08, "learning_rate": 9.866145538279061e-05, "loss": 1.6575, "step": 4930 }, { "epoch": 0.08, "learning_rate": 9.86586950984165e-05, "loss": 1.657, "step": 4935 }, { "epoch": 0.08, "learning_rate": 9.865593200960099e-05, "loss": 1.6365, "step": 4940 }, { "epoch": 0.08, "learning_rate": 9.865316611650334e-05, "loss": 1.6597, "step": 4945 }, { "epoch": 0.08, "learning_rate": 9.865039741928295e-05, "loss": 1.6097, "step": 4950 }, { "epoch": 0.08, "learning_rate": 9.86476259180994e-05, "loss": 1.6014, "step": 4955 }, { "epoch": 0.08, "learning_rate": 9.864485161311243e-05, "loss": 1.6181, "step": 4960 }, { "epoch": 0.08, "learning_rate": 9.864207450448191e-05, "loss": 1.7086, "step": 4965 }, { "epoch": 0.08, "learning_rate": 9.863929459236793e-05, "loss": 1.7067, "step": 4970 }, { "epoch": 0.08, "learning_rate": 9.86365118769307e-05, "loss": 1.6881, "step": 4975 }, { "epoch": 0.08, "learning_rate": 9.86337263583306e-05, "loss": 1.6267, "step": 4980 }, { "epoch": 0.08, "learning_rate": 9.863093803672818e-05, "loss": 1.5794, "step": 4985 }, { "epoch": 0.08, "learning_rate": 9.862814691228413e-05, "loss": 1.6569, "step": 4990 }, { "epoch": 0.08, "learning_rate": 9.862535298515932e-05, "loss": 1.6418, "step": 4995 }, { "epoch": 0.08, "learning_rate": 9.862255625551477e-05, "loss": 1.7354, "step": 5000 }, { "epoch": 0.08, "learning_rate": 9.861975672351172e-05, "loss": 1.6434, "step": 5005 }, { "epoch": 0.08, "learning_rate": 9.861695438931144e-05, "loss": 1.5897, "step": 5010 }, { "epoch": 0.08, "learning_rate": 9.86141492530755e-05, "loss": 1.633, "step": 5015 }, { "epoch": 0.08, "learning_rate": 9.861134131496558e-05, "loss": 1.7328, "step": 5020 }, { "epoch": 0.08, "learning_rate": 9.860853057514347e-05, "loss": 1.6041, "step": 5025 }, { "epoch": 0.08, "learning_rate": 9.860571703377121e-05, "loss": 1.6809, "step": 5030 }, { "epoch": 0.08, "learning_rate": 9.860290069101093e-05, "loss": 1.6125, "step": 5035 }, { "epoch": 0.08, "learning_rate": 9.860008154702496e-05, "loss": 1.6245, "step": 5040 }, { "epoch": 0.08, "learning_rate": 9.859725960197577e-05, "loss": 1.6407, "step": 5045 }, { "epoch": 0.08, "learning_rate": 9.859443485602602e-05, "loss": 1.615, "step": 5050 }, { "epoch": 0.08, "learning_rate": 9.859160730933853e-05, "loss": 1.6601, "step": 5055 }, { "epoch": 0.08, "learning_rate": 9.85887769620762e-05, "loss": 1.6595, "step": 5060 }, { "epoch": 0.08, "learning_rate": 9.858594381440222e-05, "loss": 1.6639, "step": 5065 }, { "epoch": 0.08, "learning_rate": 9.858310786647987e-05, "loss": 1.6764, "step": 5070 }, { "epoch": 0.08, "learning_rate": 9.858026911847256e-05, "loss": 1.6418, "step": 5075 }, { "epoch": 0.08, "learning_rate": 9.857742757054395e-05, "loss": 1.6008, "step": 5080 }, { "epoch": 0.08, "learning_rate": 9.85745832228578e-05, "loss": 1.6603, "step": 5085 }, { "epoch": 0.08, "learning_rate": 9.857173607557801e-05, "loss": 1.6378, "step": 5090 }, { "epoch": 0.08, "learning_rate": 9.856888612886872e-05, "loss": 1.6634, "step": 5095 }, { "epoch": 0.08, "learning_rate": 9.856603338289415e-05, "loss": 1.6202, "step": 5100 }, { "epoch": 0.08, "learning_rate": 9.856317783781874e-05, "loss": 1.6418, "step": 5105 }, { "epoch": 0.08, "learning_rate": 9.856031949380707e-05, "loss": 1.6445, "step": 5110 }, { "epoch": 0.08, "learning_rate": 9.855745835102387e-05, "loss": 1.6202, "step": 5115 }, { "epoch": 0.08, "learning_rate": 9.855459440963405e-05, "loss": 1.6107, "step": 5120 }, { "epoch": 0.08, "learning_rate": 9.855172766980267e-05, "loss": 1.6495, "step": 5125 }, { "epoch": 0.08, "learning_rate": 9.854885813169494e-05, "loss": 1.6365, "step": 5130 }, { "epoch": 0.08, "learning_rate": 9.854598579547627e-05, "loss": 1.6552, "step": 5135 }, { "epoch": 0.08, "learning_rate": 9.85431106613122e-05, "loss": 1.6765, "step": 5140 }, { "epoch": 0.08, "learning_rate": 9.854023272936843e-05, "loss": 1.6508, "step": 5145 }, { "epoch": 0.08, "learning_rate": 9.853735199981082e-05, "loss": 1.6793, "step": 5150 }, { "epoch": 0.08, "learning_rate": 9.853446847280545e-05, "loss": 1.6259, "step": 5155 }, { "epoch": 0.08, "learning_rate": 9.853158214851845e-05, "loss": 1.5918, "step": 5160 }, { "epoch": 0.08, "learning_rate": 9.85286930271162e-05, "loss": 1.6572, "step": 5165 }, { "epoch": 0.08, "learning_rate": 9.852580110876522e-05, "loss": 1.6289, "step": 5170 }, { "epoch": 0.08, "learning_rate": 9.852290639363218e-05, "loss": 1.6308, "step": 5175 }, { "epoch": 0.08, "learning_rate": 9.85200088818839e-05, "loss": 1.6608, "step": 5180 }, { "epoch": 0.08, "learning_rate": 9.851710857368741e-05, "loss": 1.597, "step": 5185 }, { "epoch": 0.08, "learning_rate": 9.851420546920987e-05, "loss": 1.6574, "step": 5190 }, { "epoch": 0.08, "learning_rate": 9.851129956861855e-05, "loss": 1.5958, "step": 5195 }, { "epoch": 0.08, "learning_rate": 9.850839087208099e-05, "loss": 1.6346, "step": 5200 }, { "epoch": 0.08, "learning_rate": 9.85054793797648e-05, "loss": 1.6763, "step": 5205 }, { "epoch": 0.08, "learning_rate": 9.85025650918378e-05, "loss": 1.6605, "step": 5210 }, { "epoch": 0.08, "learning_rate": 9.849964800846792e-05, "loss": 1.6287, "step": 5215 }, { "epoch": 0.08, "learning_rate": 9.849672812982334e-05, "loss": 1.6449, "step": 5220 }, { "epoch": 0.08, "learning_rate": 9.84938054560723e-05, "loss": 1.6445, "step": 5225 }, { "epoch": 0.08, "learning_rate": 9.849087998738328e-05, "loss": 1.6056, "step": 5230 }, { "epoch": 0.08, "learning_rate": 9.848795172392487e-05, "loss": 1.6192, "step": 5235 }, { "epoch": 0.08, "learning_rate": 9.848502066586585e-05, "loss": 1.6118, "step": 5240 }, { "epoch": 0.08, "learning_rate": 9.848208681337514e-05, "loss": 1.6739, "step": 5245 }, { "epoch": 0.08, "learning_rate": 9.847915016662186e-05, "loss": 1.6271, "step": 5250 }, { "epoch": 0.08, "learning_rate": 9.847621072577522e-05, "loss": 1.6107, "step": 5255 }, { "epoch": 0.08, "learning_rate": 9.847326849100467e-05, "loss": 1.6223, "step": 5260 }, { "epoch": 0.08, "learning_rate": 9.847032346247978e-05, "loss": 1.6169, "step": 5265 }, { "epoch": 0.08, "learning_rate": 9.846737564037027e-05, "loss": 1.6398, "step": 5270 }, { "epoch": 0.08, "learning_rate": 9.846442502484607e-05, "loss": 1.6424, "step": 5275 }, { "epoch": 0.08, "learning_rate": 9.84614716160772e-05, "loss": 1.6073, "step": 5280 }, { "epoch": 0.08, "learning_rate": 9.84585154142339e-05, "loss": 1.5937, "step": 5285 }, { "epoch": 0.08, "learning_rate": 9.845555641948654e-05, "loss": 1.6846, "step": 5290 }, { "epoch": 0.08, "learning_rate": 9.845259463200568e-05, "loss": 1.6259, "step": 5295 }, { "epoch": 0.08, "learning_rate": 9.844963005196201e-05, "loss": 1.6352, "step": 5300 }, { "epoch": 0.08, "learning_rate": 9.844666267952638e-05, "loss": 1.6651, "step": 5305 }, { "epoch": 0.08, "learning_rate": 9.844369251486985e-05, "loss": 1.6341, "step": 5310 }, { "epoch": 0.08, "learning_rate": 9.844071955816357e-05, "loss": 1.6517, "step": 5315 }, { "epoch": 0.08, "learning_rate": 9.84377438095789e-05, "loss": 1.594, "step": 5320 }, { "epoch": 0.08, "learning_rate": 9.843476526928735e-05, "loss": 1.6152, "step": 5325 }, { "epoch": 0.08, "learning_rate": 9.84317839374606e-05, "loss": 1.6382, "step": 5330 }, { "epoch": 0.08, "learning_rate": 9.842879981427045e-05, "loss": 1.6129, "step": 5335 }, { "epoch": 0.08, "learning_rate": 9.84258128998889e-05, "loss": 1.638, "step": 5340 }, { "epoch": 0.08, "learning_rate": 9.842282319448811e-05, "loss": 1.6356, "step": 5345 }, { "epoch": 0.08, "learning_rate": 9.841983069824039e-05, "loss": 1.6717, "step": 5350 }, { "epoch": 0.08, "learning_rate": 9.841683541131821e-05, "loss": 1.6329, "step": 5355 }, { "epoch": 0.08, "learning_rate": 9.84138373338942e-05, "loss": 1.6026, "step": 5360 }, { "epoch": 0.08, "learning_rate": 9.841083646614116e-05, "loss": 1.6383, "step": 5365 }, { "epoch": 0.08, "learning_rate": 9.840783280823203e-05, "loss": 1.6617, "step": 5370 }, { "epoch": 0.08, "learning_rate": 9.840482636033994e-05, "loss": 1.6592, "step": 5375 }, { "epoch": 0.08, "learning_rate": 9.840181712263816e-05, "loss": 1.6243, "step": 5380 }, { "epoch": 0.08, "learning_rate": 9.839880509530014e-05, "loss": 1.6561, "step": 5385 }, { "epoch": 0.08, "learning_rate": 9.839579027849946e-05, "loss": 1.6562, "step": 5390 }, { "epoch": 0.08, "learning_rate": 9.839277267240989e-05, "loss": 1.6501, "step": 5395 }, { "epoch": 0.08, "learning_rate": 9.838975227720536e-05, "loss": 1.6299, "step": 5400 }, { "epoch": 0.08, "learning_rate": 9.838672909305991e-05, "loss": 1.6384, "step": 5405 }, { "epoch": 0.08, "learning_rate": 9.838370312014784e-05, "loss": 1.6416, "step": 5410 }, { "epoch": 0.08, "learning_rate": 9.838067435864349e-05, "loss": 1.5986, "step": 5415 }, { "epoch": 0.08, "learning_rate": 9.837764280872146e-05, "loss": 1.6584, "step": 5420 }, { "epoch": 0.08, "learning_rate": 9.837460847055648e-05, "loss": 1.5836, "step": 5425 }, { "epoch": 0.08, "learning_rate": 9.837157134432341e-05, "loss": 1.6403, "step": 5430 }, { "epoch": 0.08, "learning_rate": 9.836853143019731e-05, "loss": 1.597, "step": 5435 }, { "epoch": 0.08, "learning_rate": 9.836548872835336e-05, "loss": 1.6002, "step": 5440 }, { "epoch": 0.08, "learning_rate": 9.836244323896698e-05, "loss": 1.6218, "step": 5445 }, { "epoch": 0.08, "learning_rate": 9.835939496221362e-05, "loss": 1.6671, "step": 5450 }, { "epoch": 0.08, "learning_rate": 9.835634389826904e-05, "loss": 1.6704, "step": 5455 }, { "epoch": 0.08, "learning_rate": 9.835329004730904e-05, "loss": 1.6963, "step": 5460 }, { "epoch": 0.08, "learning_rate": 9.835023340950966e-05, "loss": 1.6383, "step": 5465 }, { "epoch": 0.08, "learning_rate": 9.834717398504704e-05, "loss": 1.6395, "step": 5470 }, { "epoch": 0.08, "learning_rate": 9.834411177409755e-05, "loss": 1.6683, "step": 5475 }, { "epoch": 0.08, "learning_rate": 9.834104677683762e-05, "loss": 1.6838, "step": 5480 }, { "epoch": 0.08, "learning_rate": 9.833797899344397e-05, "loss": 1.6655, "step": 5485 }, { "epoch": 0.08, "learning_rate": 9.833490842409337e-05, "loss": 1.6567, "step": 5490 }, { "epoch": 0.08, "learning_rate": 9.833183506896278e-05, "loss": 1.6297, "step": 5495 }, { "epoch": 0.08, "learning_rate": 9.832875892822937e-05, "loss": 1.6142, "step": 5500 }, { "epoch": 0.08, "learning_rate": 9.83256800020704e-05, "loss": 1.6587, "step": 5505 }, { "epoch": 0.08, "learning_rate": 9.832259829066336e-05, "loss": 1.6593, "step": 5510 }, { "epoch": 0.08, "learning_rate": 9.831951379418583e-05, "loss": 1.6473, "step": 5515 }, { "epoch": 0.08, "learning_rate": 9.831642651281561e-05, "loss": 1.6017, "step": 5520 }, { "epoch": 0.08, "learning_rate": 9.83133364467306e-05, "loss": 1.6732, "step": 5525 }, { "epoch": 0.08, "learning_rate": 9.831024359610894e-05, "loss": 1.6136, "step": 5530 }, { "epoch": 0.08, "learning_rate": 9.830714796112887e-05, "loss": 1.6377, "step": 5535 }, { "epoch": 0.08, "learning_rate": 9.83040495419688e-05, "loss": 1.6551, "step": 5540 }, { "epoch": 0.08, "learning_rate": 9.83009483388073e-05, "loss": 1.6591, "step": 5545 }, { "epoch": 0.08, "learning_rate": 9.829784435182312e-05, "loss": 1.7092, "step": 5550 }, { "epoch": 0.08, "learning_rate": 9.829473758119515e-05, "loss": 1.6178, "step": 5555 }, { "epoch": 0.08, "learning_rate": 9.829162802710246e-05, "loss": 1.6487, "step": 5560 }, { "epoch": 0.08, "learning_rate": 9.828851568972426e-05, "loss": 1.6137, "step": 5565 }, { "epoch": 0.08, "learning_rate": 9.828540056923994e-05, "loss": 1.6334, "step": 5570 }, { "epoch": 0.09, "learning_rate": 9.828228266582904e-05, "loss": 1.6607, "step": 5575 }, { "epoch": 0.09, "learning_rate": 9.827916197967124e-05, "loss": 1.5881, "step": 5580 }, { "epoch": 0.09, "learning_rate": 9.82760385109464e-05, "loss": 1.643, "step": 5585 }, { "epoch": 0.09, "learning_rate": 9.827291225983456e-05, "loss": 1.5851, "step": 5590 }, { "epoch": 0.09, "learning_rate": 9.826978322651591e-05, "loss": 1.5929, "step": 5595 }, { "epoch": 0.09, "learning_rate": 9.826665141117078e-05, "loss": 1.6831, "step": 5600 }, { "epoch": 0.09, "learning_rate": 9.826351681397968e-05, "loss": 1.6087, "step": 5605 }, { "epoch": 0.09, "learning_rate": 9.826037943512323e-05, "loss": 1.6206, "step": 5610 }, { "epoch": 0.09, "learning_rate": 9.82572392747823e-05, "loss": 1.6686, "step": 5615 }, { "epoch": 0.09, "learning_rate": 9.825409633313785e-05, "loss": 1.6375, "step": 5620 }, { "epoch": 0.09, "learning_rate": 9.825095061037104e-05, "loss": 1.6101, "step": 5625 }, { "epoch": 0.09, "learning_rate": 9.824780210666315e-05, "loss": 1.6764, "step": 5630 }, { "epoch": 0.09, "learning_rate": 9.824465082219567e-05, "loss": 1.6465, "step": 5635 }, { "epoch": 0.09, "learning_rate": 9.82414967571502e-05, "loss": 1.6271, "step": 5640 }, { "epoch": 0.09, "learning_rate": 9.823833991170854e-05, "loss": 1.6209, "step": 5645 }, { "epoch": 0.09, "learning_rate": 9.823518028605264e-05, "loss": 1.698, "step": 5650 }, { "epoch": 0.09, "learning_rate": 9.82320178803646e-05, "loss": 1.6315, "step": 5655 }, { "epoch": 0.09, "learning_rate": 9.822885269482666e-05, "loss": 1.6756, "step": 5660 }, { "epoch": 0.09, "learning_rate": 9.822568472962129e-05, "loss": 1.6604, "step": 5665 }, { "epoch": 0.09, "learning_rate": 9.822251398493103e-05, "loss": 1.6407, "step": 5670 }, { "epoch": 0.09, "learning_rate": 9.821934046093866e-05, "loss": 1.5958, "step": 5675 }, { "epoch": 0.09, "learning_rate": 9.821616415782707e-05, "loss": 1.6479, "step": 5680 }, { "epoch": 0.09, "learning_rate": 9.821298507577934e-05, "loss": 1.6747, "step": 5685 }, { "epoch": 0.09, "learning_rate": 9.820980321497868e-05, "loss": 1.627, "step": 5690 }, { "epoch": 0.09, "learning_rate": 9.820661857560846e-05, "loss": 1.6448, "step": 5695 }, { "epoch": 0.09, "learning_rate": 9.820343115785228e-05, "loss": 1.6187, "step": 5700 }, { "epoch": 0.09, "learning_rate": 9.820024096189379e-05, "loss": 1.6551, "step": 5705 }, { "epoch": 0.09, "learning_rate": 9.81970479879169e-05, "loss": 1.611, "step": 5710 }, { "epoch": 0.09, "learning_rate": 9.819385223610562e-05, "loss": 1.6596, "step": 5715 }, { "epoch": 0.09, "learning_rate": 9.819065370664413e-05, "loss": 1.6689, "step": 5720 }, { "epoch": 0.09, "learning_rate": 9.818745239971678e-05, "loss": 1.5995, "step": 5725 }, { "epoch": 0.09, "learning_rate": 9.818424831550808e-05, "loss": 1.6609, "step": 5730 }, { "epoch": 0.09, "learning_rate": 9.81810414542027e-05, "loss": 1.6941, "step": 5735 }, { "epoch": 0.09, "learning_rate": 9.817783181598545e-05, "loss": 1.6593, "step": 5740 }, { "epoch": 0.09, "learning_rate": 9.817461940104136e-05, "loss": 1.6105, "step": 5745 }, { "epoch": 0.09, "learning_rate": 9.817140420955554e-05, "loss": 1.5876, "step": 5750 }, { "epoch": 0.09, "learning_rate": 9.81681862417133e-05, "loss": 1.6236, "step": 5755 }, { "epoch": 0.09, "learning_rate": 9.816496549770012e-05, "loss": 1.6603, "step": 5760 }, { "epoch": 0.09, "learning_rate": 9.816174197770161e-05, "loss": 1.6621, "step": 5765 }, { "epoch": 0.09, "learning_rate": 9.815851568190357e-05, "loss": 1.6342, "step": 5770 }, { "epoch": 0.09, "learning_rate": 9.815528661049196e-05, "loss": 1.6117, "step": 5775 }, { "epoch": 0.09, "learning_rate": 9.815205476365287e-05, "loss": 1.665, "step": 5780 }, { "epoch": 0.09, "learning_rate": 9.814882014157256e-05, "loss": 1.6652, "step": 5785 }, { "epoch": 0.09, "learning_rate": 9.814558274443748e-05, "loss": 1.6398, "step": 5790 }, { "epoch": 0.09, "learning_rate": 9.81423425724342e-05, "loss": 1.5677, "step": 5795 }, { "epoch": 0.09, "learning_rate": 9.813909962574948e-05, "loss": 1.7031, "step": 5800 }, { "epoch": 0.09, "learning_rate": 9.81358539045702e-05, "loss": 1.6255, "step": 5805 }, { "epoch": 0.09, "learning_rate": 9.813260540908346e-05, "loss": 1.6512, "step": 5810 }, { "epoch": 0.09, "learning_rate": 9.812935413947649e-05, "loss": 1.6298, "step": 5815 }, { "epoch": 0.09, "learning_rate": 9.812610009593664e-05, "loss": 1.6457, "step": 5820 }, { "epoch": 0.09, "learning_rate": 9.812284327865149e-05, "loss": 1.596, "step": 5825 }, { "epoch": 0.09, "learning_rate": 9.811958368780873e-05, "loss": 1.6641, "step": 5830 }, { "epoch": 0.09, "learning_rate": 9.811632132359621e-05, "loss": 1.6171, "step": 5835 }, { "epoch": 0.09, "learning_rate": 9.8113056186202e-05, "loss": 1.6575, "step": 5840 }, { "epoch": 0.09, "learning_rate": 9.810978827581426e-05, "loss": 1.6176, "step": 5845 }, { "epoch": 0.09, "learning_rate": 9.810651759262132e-05, "loss": 1.6671, "step": 5850 }, { "epoch": 0.09, "learning_rate": 9.810324413681172e-05, "loss": 1.5773, "step": 5855 }, { "epoch": 0.09, "learning_rate": 9.809996790857411e-05, "loss": 1.6512, "step": 5860 }, { "epoch": 0.09, "learning_rate": 9.809668890809731e-05, "loss": 1.6671, "step": 5865 }, { "epoch": 0.09, "learning_rate": 9.80934071355703e-05, "loss": 1.6026, "step": 5870 }, { "epoch": 0.09, "learning_rate": 9.809012259118225e-05, "loss": 1.6667, "step": 5875 }, { "epoch": 0.09, "learning_rate": 9.808683527512243e-05, "loss": 1.5932, "step": 5880 }, { "epoch": 0.09, "learning_rate": 9.808354518758035e-05, "loss": 1.6418, "step": 5885 }, { "epoch": 0.09, "learning_rate": 9.808025232874558e-05, "loss": 1.5923, "step": 5890 }, { "epoch": 0.09, "learning_rate": 9.807695669880794e-05, "loss": 1.6142, "step": 5895 }, { "epoch": 0.09, "learning_rate": 9.807365829795737e-05, "loss": 1.6644, "step": 5900 }, { "epoch": 0.09, "learning_rate": 9.807035712638395e-05, "loss": 1.6882, "step": 5905 }, { "epoch": 0.09, "learning_rate": 9.806705318427798e-05, "loss": 1.6522, "step": 5910 }, { "epoch": 0.09, "learning_rate": 9.806374647182985e-05, "loss": 1.6467, "step": 5915 }, { "epoch": 0.09, "learning_rate": 9.806043698923018e-05, "loss": 1.6882, "step": 5920 }, { "epoch": 0.09, "learning_rate": 9.805712473666966e-05, "loss": 1.6654, "step": 5925 }, { "epoch": 0.09, "learning_rate": 9.805380971433923e-05, "loss": 1.6272, "step": 5930 }, { "epoch": 0.09, "learning_rate": 9.805049192242993e-05, "loss": 1.6472, "step": 5935 }, { "epoch": 0.09, "learning_rate": 9.8047171361133e-05, "loss": 1.6475, "step": 5940 }, { "epoch": 0.09, "learning_rate": 9.804384803063981e-05, "loss": 1.6465, "step": 5945 }, { "epoch": 0.09, "learning_rate": 9.804052193114189e-05, "loss": 1.6073, "step": 5950 }, { "epoch": 0.09, "learning_rate": 9.803719306283096e-05, "loss": 1.6447, "step": 5955 }, { "epoch": 0.09, "learning_rate": 9.803386142589886e-05, "loss": 1.6387, "step": 5960 }, { "epoch": 0.09, "learning_rate": 9.803052702053764e-05, "loss": 1.6106, "step": 5965 }, { "epoch": 0.09, "learning_rate": 9.802718984693944e-05, "loss": 1.6489, "step": 5970 }, { "epoch": 0.09, "learning_rate": 9.80238499052966e-05, "loss": 1.679, "step": 5975 }, { "epoch": 0.09, "learning_rate": 9.802050719580165e-05, "loss": 1.5957, "step": 5980 }, { "epoch": 0.09, "learning_rate": 9.801716171864723e-05, "loss": 1.6394, "step": 5985 }, { "epoch": 0.09, "learning_rate": 9.801381347402615e-05, "loss": 1.6555, "step": 5990 }, { "epoch": 0.09, "learning_rate": 9.80104624621314e-05, "loss": 1.6779, "step": 5995 }, { "epoch": 0.09, "learning_rate": 9.800710868315607e-05, "loss": 1.5972, "step": 6000 }, { "epoch": 0.09, "learning_rate": 9.800375213729353e-05, "loss": 1.6044, "step": 6005 }, { "epoch": 0.09, "learning_rate": 9.800039282473719e-05, "loss": 1.6009, "step": 6010 }, { "epoch": 0.09, "learning_rate": 9.799703074568066e-05, "loss": 1.7184, "step": 6015 }, { "epoch": 0.09, "learning_rate": 9.799366590031774e-05, "loss": 1.6392, "step": 6020 }, { "epoch": 0.09, "learning_rate": 9.799029828884232e-05, "loss": 1.6291, "step": 6025 }, { "epoch": 0.09, "learning_rate": 9.798692791144852e-05, "loss": 1.5951, "step": 6030 }, { "epoch": 0.09, "learning_rate": 9.79835547683306e-05, "loss": 1.609, "step": 6035 }, { "epoch": 0.09, "learning_rate": 9.798017885968296e-05, "loss": 1.6637, "step": 6040 }, { "epoch": 0.09, "learning_rate": 9.797680018570016e-05, "loss": 1.6956, "step": 6045 }, { "epoch": 0.09, "learning_rate": 9.797341874657696e-05, "loss": 1.6613, "step": 6050 }, { "epoch": 0.09, "learning_rate": 9.797003454250819e-05, "loss": 1.6214, "step": 6055 }, { "epoch": 0.09, "learning_rate": 9.796664757368896e-05, "loss": 1.608, "step": 6060 }, { "epoch": 0.09, "learning_rate": 9.796325784031446e-05, "loss": 1.6159, "step": 6065 }, { "epoch": 0.09, "learning_rate": 9.795986534258006e-05, "loss": 1.647, "step": 6070 }, { "epoch": 0.09, "learning_rate": 9.795647008068126e-05, "loss": 1.6605, "step": 6075 }, { "epoch": 0.09, "learning_rate": 9.795307205481376e-05, "loss": 1.6519, "step": 6080 }, { "epoch": 0.09, "learning_rate": 9.794967126517343e-05, "loss": 1.6427, "step": 6085 }, { "epoch": 0.09, "learning_rate": 9.794626771195625e-05, "loss": 1.5837, "step": 6090 }, { "epoch": 0.09, "learning_rate": 9.794286139535838e-05, "loss": 1.6263, "step": 6095 }, { "epoch": 0.09, "learning_rate": 9.793945231557615e-05, "loss": 1.6924, "step": 6100 }, { "epoch": 0.09, "learning_rate": 9.793604047280605e-05, "loss": 1.6107, "step": 6105 }, { "epoch": 0.09, "learning_rate": 9.79326258672447e-05, "loss": 1.6383, "step": 6110 }, { "epoch": 0.09, "learning_rate": 9.792920849908892e-05, "loss": 1.5415, "step": 6115 }, { "epoch": 0.09, "learning_rate": 9.792578836853566e-05, "loss": 1.5782, "step": 6120 }, { "epoch": 0.09, "learning_rate": 9.792236547578206e-05, "loss": 1.6246, "step": 6125 }, { "epoch": 0.09, "learning_rate": 9.791893982102537e-05, "loss": 1.6299, "step": 6130 }, { "epoch": 0.09, "learning_rate": 9.791551140446303e-05, "loss": 1.5985, "step": 6135 }, { "epoch": 0.09, "learning_rate": 9.791208022629266e-05, "loss": 1.6431, "step": 6140 }, { "epoch": 0.09, "learning_rate": 9.790864628671198e-05, "loss": 1.6003, "step": 6145 }, { "epoch": 0.09, "learning_rate": 9.790520958591895e-05, "loss": 1.6655, "step": 6150 }, { "epoch": 0.09, "learning_rate": 9.790177012411161e-05, "loss": 1.6112, "step": 6155 }, { "epoch": 0.09, "learning_rate": 9.789832790148822e-05, "loss": 1.6498, "step": 6160 }, { "epoch": 0.09, "learning_rate": 9.789488291824713e-05, "loss": 1.582, "step": 6165 }, { "epoch": 0.09, "learning_rate": 9.789143517458695e-05, "loss": 1.6288, "step": 6170 }, { "epoch": 0.09, "learning_rate": 9.788798467070632e-05, "loss": 1.6546, "step": 6175 }, { "epoch": 0.09, "learning_rate": 9.788453140680417e-05, "loss": 1.6044, "step": 6180 }, { "epoch": 0.09, "learning_rate": 9.788107538307949e-05, "loss": 1.5701, "step": 6185 }, { "epoch": 0.09, "learning_rate": 9.78776165997315e-05, "loss": 1.6276, "step": 6190 }, { "epoch": 0.09, "learning_rate": 9.787415505695953e-05, "loss": 1.6232, "step": 6195 }, { "epoch": 0.09, "learning_rate": 9.787069075496307e-05, "loss": 1.579, "step": 6200 }, { "epoch": 0.09, "learning_rate": 9.786722369394182e-05, "loss": 1.6686, "step": 6205 }, { "epoch": 0.09, "learning_rate": 9.786375387409558e-05, "loss": 1.6216, "step": 6210 }, { "epoch": 0.09, "learning_rate": 9.786028129562433e-05, "loss": 1.5875, "step": 6215 }, { "epoch": 0.09, "learning_rate": 9.785680595872823e-05, "loss": 1.6653, "step": 6220 }, { "epoch": 0.09, "learning_rate": 9.785332786360757e-05, "loss": 1.6249, "step": 6225 }, { "epoch": 0.1, "learning_rate": 9.78498470104628e-05, "loss": 1.6791, "step": 6230 }, { "epoch": 0.1, "learning_rate": 9.784636339949456e-05, "loss": 1.5764, "step": 6235 }, { "epoch": 0.1, "learning_rate": 9.78428770309036e-05, "loss": 1.6535, "step": 6240 }, { "epoch": 0.1, "learning_rate": 9.783938790489089e-05, "loss": 1.6783, "step": 6245 }, { "epoch": 0.1, "learning_rate": 9.783589602165752e-05, "loss": 1.6265, "step": 6250 }, { "epoch": 0.1, "learning_rate": 9.783240138140471e-05, "loss": 1.5828, "step": 6255 }, { "epoch": 0.1, "learning_rate": 9.782890398433391e-05, "loss": 1.6195, "step": 6260 }, { "epoch": 0.1, "learning_rate": 9.782540383064668e-05, "loss": 1.6529, "step": 6265 }, { "epoch": 0.1, "learning_rate": 9.782190092054475e-05, "loss": 1.5694, "step": 6270 }, { "epoch": 0.1, "learning_rate": 9.781839525423e-05, "loss": 1.6513, "step": 6275 }, { "epoch": 0.1, "learning_rate": 9.781488683190451e-05, "loss": 1.6257, "step": 6280 }, { "epoch": 0.1, "learning_rate": 9.781137565377045e-05, "loss": 1.6434, "step": 6285 }, { "epoch": 0.1, "learning_rate": 9.780786172003022e-05, "loss": 1.644, "step": 6290 }, { "epoch": 0.1, "learning_rate": 9.780434503088631e-05, "loss": 1.649, "step": 6295 }, { "epoch": 0.1, "learning_rate": 9.780082558654145e-05, "loss": 1.6176, "step": 6300 }, { "epoch": 0.1, "learning_rate": 9.779730338719843e-05, "loss": 1.5938, "step": 6305 }, { "epoch": 0.1, "learning_rate": 9.779377843306028e-05, "loss": 1.6157, "step": 6310 }, { "epoch": 0.1, "learning_rate": 9.779025072433018e-05, "loss": 1.5673, "step": 6315 }, { "epoch": 0.1, "learning_rate": 9.77867202612114e-05, "loss": 1.5925, "step": 6320 }, { "epoch": 0.1, "learning_rate": 9.778318704390747e-05, "loss": 1.5735, "step": 6325 }, { "epoch": 0.1, "learning_rate": 9.7779651072622e-05, "loss": 1.6046, "step": 6330 }, { "epoch": 0.1, "learning_rate": 9.77761123475588e-05, "loss": 1.6598, "step": 6335 }, { "epoch": 0.1, "learning_rate": 9.777257086892178e-05, "loss": 1.6237, "step": 6340 }, { "epoch": 0.1, "learning_rate": 9.776902663691512e-05, "loss": 1.654, "step": 6345 }, { "epoch": 0.1, "learning_rate": 9.776547965174303e-05, "loss": 1.6354, "step": 6350 }, { "epoch": 0.1, "learning_rate": 9.776192991360998e-05, "loss": 1.5989, "step": 6355 }, { "epoch": 0.1, "learning_rate": 9.775837742272055e-05, "loss": 1.6716, "step": 6360 }, { "epoch": 0.1, "learning_rate": 9.775482217927948e-05, "loss": 1.6083, "step": 6365 }, { "epoch": 0.1, "learning_rate": 9.775126418349167e-05, "loss": 1.6475, "step": 6370 }, { "epoch": 0.1, "learning_rate": 9.774770343556222e-05, "loss": 1.557, "step": 6375 }, { "epoch": 0.1, "learning_rate": 9.774413993569631e-05, "loss": 1.6317, "step": 6380 }, { "epoch": 0.1, "learning_rate": 9.774057368409934e-05, "loss": 1.6242, "step": 6385 }, { "epoch": 0.1, "learning_rate": 9.773700468097687e-05, "loss": 1.6311, "step": 6390 }, { "epoch": 0.1, "learning_rate": 9.773343292653457e-05, "loss": 1.6447, "step": 6395 }, { "epoch": 0.1, "learning_rate": 9.772985842097831e-05, "loss": 1.6125, "step": 6400 }, { "epoch": 0.1, "learning_rate": 9.772628116451411e-05, "loss": 1.6155, "step": 6405 }, { "epoch": 0.1, "learning_rate": 9.772270115734813e-05, "loss": 1.6333, "step": 6410 }, { "epoch": 0.1, "learning_rate": 9.771911839968673e-05, "loss": 1.6302, "step": 6415 }, { "epoch": 0.1, "learning_rate": 9.771553289173638e-05, "loss": 1.6716, "step": 6420 }, { "epoch": 0.1, "learning_rate": 9.771194463370372e-05, "loss": 1.5803, "step": 6425 }, { "epoch": 0.1, "learning_rate": 9.77083536257956e-05, "loss": 1.6564, "step": 6430 }, { "epoch": 0.1, "learning_rate": 9.770475986821894e-05, "loss": 1.6206, "step": 6435 }, { "epoch": 0.1, "learning_rate": 9.770116336118091e-05, "loss": 1.6428, "step": 6440 }, { "epoch": 0.1, "learning_rate": 9.769756410488877e-05, "loss": 1.6313, "step": 6445 }, { "epoch": 0.1, "learning_rate": 9.769396209954996e-05, "loss": 1.6111, "step": 6450 }, { "epoch": 0.1, "learning_rate": 9.769035734537209e-05, "loss": 1.634, "step": 6455 }, { "epoch": 0.1, "learning_rate": 9.768674984256292e-05, "loss": 1.6118, "step": 6460 }, { "epoch": 0.1, "learning_rate": 9.768313959133038e-05, "loss": 1.684, "step": 6465 }, { "epoch": 0.1, "learning_rate": 9.767952659188251e-05, "loss": 1.6112, "step": 6470 }, { "epoch": 0.1, "learning_rate": 9.767591084442758e-05, "loss": 1.576, "step": 6475 }, { "epoch": 0.1, "learning_rate": 9.767229234917397e-05, "loss": 1.6188, "step": 6480 }, { "epoch": 0.1, "learning_rate": 9.766867110633022e-05, "loss": 1.6101, "step": 6485 }, { "epoch": 0.1, "learning_rate": 9.766504711610507e-05, "loss": 1.6038, "step": 6490 }, { "epoch": 0.1, "learning_rate": 9.766142037870737e-05, "loss": 1.6146, "step": 6495 }, { "epoch": 0.1, "learning_rate": 9.765779089434615e-05, "loss": 1.6505, "step": 6500 }, { "epoch": 0.1, "learning_rate": 9.765415866323058e-05, "loss": 1.5733, "step": 6505 }, { "epoch": 0.1, "learning_rate": 9.765052368557002e-05, "loss": 1.6828, "step": 6510 }, { "epoch": 0.1, "learning_rate": 9.764688596157398e-05, "loss": 1.6111, "step": 6515 }, { "epoch": 0.1, "learning_rate": 9.764324549145212e-05, "loss": 1.641, "step": 6520 }, { "epoch": 0.1, "learning_rate": 9.763960227541422e-05, "loss": 1.6476, "step": 6525 }, { "epoch": 0.1, "learning_rate": 9.763595631367031e-05, "loss": 1.5797, "step": 6530 }, { "epoch": 0.1, "learning_rate": 9.763230760643047e-05, "loss": 1.6694, "step": 6535 }, { "epoch": 0.1, "learning_rate": 9.762865615390503e-05, "loss": 1.5993, "step": 6540 }, { "epoch": 0.1, "learning_rate": 9.762500195630445e-05, "loss": 1.6201, "step": 6545 }, { "epoch": 0.1, "learning_rate": 9.762134501383932e-05, "loss": 1.619, "step": 6550 }, { "epoch": 0.1, "learning_rate": 9.761768532672039e-05, "loss": 1.6263, "step": 6555 }, { "epoch": 0.1, "learning_rate": 9.761402289515863e-05, "loss": 1.5998, "step": 6560 }, { "epoch": 0.1, "learning_rate": 9.761035771936508e-05, "loss": 1.6034, "step": 6565 }, { "epoch": 0.1, "learning_rate": 9.760668979955102e-05, "loss": 1.6343, "step": 6570 }, { "epoch": 0.1, "learning_rate": 9.760301913592782e-05, "loss": 1.618, "step": 6575 }, { "epoch": 0.1, "learning_rate": 9.759934572870705e-05, "loss": 1.6602, "step": 6580 }, { "epoch": 0.1, "learning_rate": 9.759566957810045e-05, "loss": 1.6067, "step": 6585 }, { "epoch": 0.1, "learning_rate": 9.759199068431985e-05, "loss": 1.6449, "step": 6590 }, { "epoch": 0.1, "learning_rate": 9.758830904757733e-05, "loss": 1.6326, "step": 6595 }, { "epoch": 0.1, "learning_rate": 9.758462466808502e-05, "loss": 1.5927, "step": 6600 }, { "epoch": 0.1, "learning_rate": 9.758093754605533e-05, "loss": 1.671, "step": 6605 }, { "epoch": 0.1, "learning_rate": 9.757724768170073e-05, "loss": 1.5996, "step": 6610 }, { "epoch": 0.1, "learning_rate": 9.757355507523391e-05, "loss": 1.6442, "step": 6615 }, { "epoch": 0.1, "learning_rate": 9.756985972686767e-05, "loss": 1.6339, "step": 6620 }, { "epoch": 0.1, "learning_rate": 9.756616163681502e-05, "loss": 1.6021, "step": 6625 }, { "epoch": 0.1, "learning_rate": 9.756246080528907e-05, "loss": 1.6319, "step": 6630 }, { "epoch": 0.1, "learning_rate": 9.755875723250313e-05, "loss": 1.6864, "step": 6635 }, { "epoch": 0.1, "learning_rate": 9.755505091867066e-05, "loss": 1.6778, "step": 6640 }, { "epoch": 0.1, "learning_rate": 9.755134186400525e-05, "loss": 1.6203, "step": 6645 }, { "epoch": 0.1, "learning_rate": 9.75476300687207e-05, "loss": 1.6361, "step": 6650 }, { "epoch": 0.1, "learning_rate": 9.754391553303092e-05, "loss": 1.5788, "step": 6655 }, { "epoch": 0.1, "learning_rate": 9.754019825715001e-05, "loss": 1.6204, "step": 6660 }, { "epoch": 0.1, "learning_rate": 9.753647824129221e-05, "loss": 1.6072, "step": 6665 }, { "epoch": 0.1, "learning_rate": 9.753275548567192e-05, "loss": 1.6069, "step": 6670 }, { "epoch": 0.1, "learning_rate": 9.75290299905037e-05, "loss": 1.6468, "step": 6675 }, { "epoch": 0.1, "learning_rate": 9.752530175600229e-05, "loss": 1.6101, "step": 6680 }, { "epoch": 0.1, "learning_rate": 9.752157078238253e-05, "loss": 1.6423, "step": 6685 }, { "epoch": 0.1, "learning_rate": 9.751783706985948e-05, "loss": 1.6129, "step": 6690 }, { "epoch": 0.1, "learning_rate": 9.751410061864832e-05, "loss": 1.6076, "step": 6695 }, { "epoch": 0.1, "learning_rate": 9.751036142896441e-05, "loss": 1.6357, "step": 6700 }, { "epoch": 0.1, "learning_rate": 9.750661950102326e-05, "loss": 1.6191, "step": 6705 }, { "epoch": 0.1, "learning_rate": 9.750287483504053e-05, "loss": 1.7004, "step": 6710 }, { "epoch": 0.1, "learning_rate": 9.749912743123201e-05, "loss": 1.6206, "step": 6715 }, { "epoch": 0.1, "learning_rate": 9.749537728981374e-05, "loss": 1.5814, "step": 6720 }, { "epoch": 0.1, "learning_rate": 9.749162441100185e-05, "loss": 1.5783, "step": 6725 }, { "epoch": 0.1, "learning_rate": 9.74878687950126e-05, "loss": 1.5791, "step": 6730 }, { "epoch": 0.1, "learning_rate": 9.748411044206247e-05, "loss": 1.6362, "step": 6735 }, { "epoch": 0.1, "learning_rate": 9.748034935236807e-05, "loss": 1.6372, "step": 6740 }, { "epoch": 0.1, "learning_rate": 9.747658552614616e-05, "loss": 1.6037, "step": 6745 }, { "epoch": 0.1, "learning_rate": 9.747281896361369e-05, "loss": 1.5766, "step": 6750 }, { "epoch": 0.1, "learning_rate": 9.746904966498771e-05, "loss": 1.6062, "step": 6755 }, { "epoch": 0.1, "learning_rate": 9.74652776304855e-05, "loss": 1.6126, "step": 6760 }, { "epoch": 0.1, "learning_rate": 9.746150286032445e-05, "loss": 1.6205, "step": 6765 }, { "epoch": 0.1, "learning_rate": 9.745772535472211e-05, "loss": 1.658, "step": 6770 }, { "epoch": 0.1, "learning_rate": 9.745394511389619e-05, "loss": 1.6153, "step": 6775 }, { "epoch": 0.1, "learning_rate": 9.74501621380646e-05, "loss": 1.6088, "step": 6780 }, { "epoch": 0.1, "learning_rate": 9.744637642744531e-05, "loss": 1.6366, "step": 6785 }, { "epoch": 0.1, "learning_rate": 9.744258798225659e-05, "loss": 1.659, "step": 6790 }, { "epoch": 0.1, "learning_rate": 9.74387968027167e-05, "loss": 1.6404, "step": 6795 }, { "epoch": 0.1, "learning_rate": 9.743500288904422e-05, "loss": 1.68, "step": 6800 }, { "epoch": 0.1, "learning_rate": 9.743120624145776e-05, "loss": 1.6075, "step": 6805 }, { "epoch": 0.1, "learning_rate": 9.742740686017617e-05, "loss": 1.579, "step": 6810 }, { "epoch": 0.1, "learning_rate": 9.742360474541841e-05, "loss": 1.5612, "step": 6815 }, { "epoch": 0.1, "learning_rate": 9.741979989740361e-05, "loss": 1.6552, "step": 6820 }, { "epoch": 0.1, "learning_rate": 9.741599231635108e-05, "loss": 1.6268, "step": 6825 }, { "epoch": 0.1, "learning_rate": 9.741218200248027e-05, "loss": 1.6748, "step": 6830 }, { "epoch": 0.1, "learning_rate": 9.740836895601078e-05, "loss": 1.6059, "step": 6835 }, { "epoch": 0.1, "learning_rate": 9.740455317716237e-05, "loss": 1.6291, "step": 6840 }, { "epoch": 0.1, "learning_rate": 9.740073466615496e-05, "loss": 1.64, "step": 6845 }, { "epoch": 0.1, "learning_rate": 9.739691342320865e-05, "loss": 1.5851, "step": 6850 }, { "epoch": 0.1, "learning_rate": 9.739308944854366e-05, "loss": 1.6178, "step": 6855 }, { "epoch": 0.1, "learning_rate": 9.738926274238038e-05, "loss": 1.6323, "step": 6860 }, { "epoch": 0.1, "learning_rate": 9.738543330493938e-05, "loss": 1.64, "step": 6865 }, { "epoch": 0.1, "learning_rate": 9.738160113644135e-05, "loss": 1.6426, "step": 6870 }, { "epoch": 0.1, "learning_rate": 9.737776623710717e-05, "loss": 1.613, "step": 6875 }, { "epoch": 0.1, "learning_rate": 9.737392860715786e-05, "loss": 1.6207, "step": 6880 }, { "epoch": 0.11, "learning_rate": 9.737008824681461e-05, "loss": 1.6486, "step": 6885 }, { "epoch": 0.11, "learning_rate": 9.736624515629875e-05, "loss": 1.5562, "step": 6890 }, { "epoch": 0.11, "learning_rate": 9.736239933583177e-05, "loss": 1.6173, "step": 6895 }, { "epoch": 0.11, "learning_rate": 9.735855078563533e-05, "loss": 1.6251, "step": 6900 }, { "epoch": 0.11, "learning_rate": 9.735469950593125e-05, "loss": 1.5931, "step": 6905 }, { "epoch": 0.11, "learning_rate": 9.735084549694148e-05, "loss": 1.6333, "step": 6910 }, { "epoch": 0.11, "learning_rate": 9.734698875888817e-05, "loss": 1.6027, "step": 6915 }, { "epoch": 0.11, "learning_rate": 9.734312929199358e-05, "loss": 1.5839, "step": 6920 }, { "epoch": 0.11, "learning_rate": 9.733926709648016e-05, "loss": 1.6155, "step": 6925 }, { "epoch": 0.11, "learning_rate": 9.733540217257052e-05, "loss": 1.5798, "step": 6930 }, { "epoch": 0.11, "learning_rate": 9.733153452048738e-05, "loss": 1.6222, "step": 6935 }, { "epoch": 0.11, "learning_rate": 9.732766414045368e-05, "loss": 1.6642, "step": 6940 }, { "epoch": 0.11, "learning_rate": 9.732379103269248e-05, "loss": 1.6245, "step": 6945 }, { "epoch": 0.11, "learning_rate": 9.731991519742702e-05, "loss": 1.6, "step": 6950 }, { "epoch": 0.11, "learning_rate": 9.731603663488066e-05, "loss": 1.6218, "step": 6955 }, { "epoch": 0.11, "learning_rate": 9.731215534527696e-05, "loss": 1.6062, "step": 6960 }, { "epoch": 0.11, "learning_rate": 9.730827132883962e-05, "loss": 1.6855, "step": 6965 }, { "epoch": 0.11, "learning_rate": 9.730438458579247e-05, "loss": 1.6629, "step": 6970 }, { "epoch": 0.11, "learning_rate": 9.730049511635955e-05, "loss": 1.6542, "step": 6975 }, { "epoch": 0.11, "learning_rate": 9.7296602920765e-05, "loss": 1.5975, "step": 6980 }, { "epoch": 0.11, "learning_rate": 9.729270799923318e-05, "loss": 1.6246, "step": 6985 }, { "epoch": 0.11, "learning_rate": 9.728881035198856e-05, "loss": 1.6298, "step": 6990 }, { "epoch": 0.11, "learning_rate": 9.728490997925576e-05, "loss": 1.6436, "step": 6995 }, { "epoch": 0.11, "learning_rate": 9.72810068812596e-05, "loss": 1.633, "step": 7000 }, { "epoch": 0.11, "learning_rate": 9.727710105822504e-05, "loss": 1.6113, "step": 7005 }, { "epoch": 0.11, "learning_rate": 9.72731925103772e-05, "loss": 1.6692, "step": 7010 }, { "epoch": 0.11, "learning_rate": 9.726928123794131e-05, "loss": 1.619, "step": 7015 }, { "epoch": 0.11, "learning_rate": 9.726536724114282e-05, "loss": 1.6273, "step": 7020 }, { "epoch": 0.11, "learning_rate": 9.726145052020732e-05, "loss": 1.6532, "step": 7025 }, { "epoch": 0.11, "learning_rate": 9.725753107536053e-05, "loss": 1.627, "step": 7030 }, { "epoch": 0.11, "learning_rate": 9.725360890682836e-05, "loss": 1.6653, "step": 7035 }, { "epoch": 0.11, "learning_rate": 9.724968401483687e-05, "loss": 1.6461, "step": 7040 }, { "epoch": 0.11, "learning_rate": 9.724575639961226e-05, "loss": 1.6619, "step": 7045 }, { "epoch": 0.11, "learning_rate": 9.72418260613809e-05, "loss": 1.6284, "step": 7050 }, { "epoch": 0.11, "learning_rate": 9.723789300036932e-05, "loss": 1.6057, "step": 7055 }, { "epoch": 0.11, "learning_rate": 9.723395721680419e-05, "loss": 1.6519, "step": 7060 }, { "epoch": 0.11, "learning_rate": 9.723001871091235e-05, "loss": 1.626, "step": 7065 }, { "epoch": 0.11, "learning_rate": 9.72260774829208e-05, "loss": 1.6285, "step": 7070 }, { "epoch": 0.11, "learning_rate": 9.722213353305671e-05, "loss": 1.6332, "step": 7075 }, { "epoch": 0.11, "learning_rate": 9.721818686154738e-05, "loss": 1.6249, "step": 7080 }, { "epoch": 0.11, "learning_rate": 9.721423746862026e-05, "loss": 1.658, "step": 7085 }, { "epoch": 0.11, "learning_rate": 9.721028535450297e-05, "loss": 1.5701, "step": 7090 }, { "epoch": 0.11, "learning_rate": 9.720633051942332e-05, "loss": 1.6718, "step": 7095 }, { "epoch": 0.11, "learning_rate": 9.720237296360921e-05, "loss": 1.6068, "step": 7100 }, { "epoch": 0.11, "learning_rate": 9.719841268728877e-05, "loss": 1.6125, "step": 7105 }, { "epoch": 0.11, "learning_rate": 9.719444969069022e-05, "loss": 1.6383, "step": 7110 }, { "epoch": 0.11, "learning_rate": 9.7190483974042e-05, "loss": 1.6508, "step": 7115 }, { "epoch": 0.11, "learning_rate": 9.718651553757265e-05, "loss": 1.6181, "step": 7120 }, { "epoch": 0.11, "learning_rate": 9.718254438151089e-05, "loss": 1.5986, "step": 7125 }, { "epoch": 0.11, "learning_rate": 9.717857050608558e-05, "loss": 1.6473, "step": 7130 }, { "epoch": 0.11, "learning_rate": 9.717459391152581e-05, "loss": 1.5897, "step": 7135 }, { "epoch": 0.11, "learning_rate": 9.717061459806072e-05, "loss": 1.5961, "step": 7140 }, { "epoch": 0.11, "learning_rate": 9.716663256591969e-05, "loss": 1.6177, "step": 7145 }, { "epoch": 0.11, "learning_rate": 9.71626478153322e-05, "loss": 1.6376, "step": 7150 }, { "epoch": 0.11, "learning_rate": 9.715866034652791e-05, "loss": 1.6114, "step": 7155 }, { "epoch": 0.11, "learning_rate": 9.715467015973665e-05, "loss": 1.6065, "step": 7160 }, { "epoch": 0.11, "learning_rate": 9.71506772551884e-05, "loss": 1.6209, "step": 7165 }, { "epoch": 0.11, "learning_rate": 9.714668163311328e-05, "loss": 1.6387, "step": 7170 }, { "epoch": 0.11, "learning_rate": 9.714268329374159e-05, "loss": 1.6808, "step": 7175 }, { "epoch": 0.11, "learning_rate": 9.713868223730375e-05, "loss": 1.5841, "step": 7180 }, { "epoch": 0.11, "learning_rate": 9.713467846403037e-05, "loss": 1.6066, "step": 7185 }, { "epoch": 0.11, "learning_rate": 9.713067197415223e-05, "loss": 1.626, "step": 7190 }, { "epoch": 0.11, "learning_rate": 9.712666276790022e-05, "loss": 1.644, "step": 7195 }, { "epoch": 0.11, "learning_rate": 9.71226508455054e-05, "loss": 1.5899, "step": 7200 }, { "epoch": 0.11, "learning_rate": 9.711863620719904e-05, "loss": 1.6733, "step": 7205 }, { "epoch": 0.11, "learning_rate": 9.711461885321247e-05, "loss": 1.639, "step": 7210 }, { "epoch": 0.11, "learning_rate": 9.711059878377726e-05, "loss": 1.6218, "step": 7215 }, { "epoch": 0.11, "learning_rate": 9.710657599912512e-05, "loss": 1.5806, "step": 7220 }, { "epoch": 0.11, "learning_rate": 9.710255049948787e-05, "loss": 1.5732, "step": 7225 }, { "epoch": 0.11, "learning_rate": 9.709852228509755e-05, "loss": 1.5444, "step": 7230 }, { "epoch": 0.11, "learning_rate": 9.70944913561863e-05, "loss": 1.6679, "step": 7235 }, { "epoch": 0.11, "learning_rate": 9.709045771298646e-05, "loss": 1.5989, "step": 7240 }, { "epoch": 0.11, "learning_rate": 9.70864213557305e-05, "loss": 1.6303, "step": 7245 }, { "epoch": 0.11, "learning_rate": 9.708238228465107e-05, "loss": 1.559, "step": 7250 }, { "epoch": 0.11, "learning_rate": 9.707834049998092e-05, "loss": 1.6641, "step": 7255 }, { "epoch": 0.11, "learning_rate": 9.707429600195306e-05, "loss": 1.5897, "step": 7260 }, { "epoch": 0.11, "learning_rate": 9.707024879080055e-05, "loss": 1.647, "step": 7265 }, { "epoch": 0.11, "learning_rate": 9.706619886675665e-05, "loss": 1.6335, "step": 7270 }, { "epoch": 0.11, "learning_rate": 9.706214623005481e-05, "loss": 1.605, "step": 7275 }, { "epoch": 0.11, "learning_rate": 9.705809088092856e-05, "loss": 1.5959, "step": 7280 }, { "epoch": 0.11, "learning_rate": 9.705403281961168e-05, "loss": 1.6387, "step": 7285 }, { "epoch": 0.11, "learning_rate": 9.7049972046338e-05, "loss": 1.6481, "step": 7290 }, { "epoch": 0.11, "learning_rate": 9.704590856134163e-05, "loss": 1.5873, "step": 7295 }, { "epoch": 0.11, "learning_rate": 9.704184236485672e-05, "loss": 1.6665, "step": 7300 }, { "epoch": 0.11, "learning_rate": 9.703777345711762e-05, "loss": 1.6091, "step": 7305 }, { "epoch": 0.11, "learning_rate": 9.703370183835886e-05, "loss": 1.5645, "step": 7310 }, { "epoch": 0.11, "learning_rate": 9.70296275088151e-05, "loss": 1.5558, "step": 7315 }, { "epoch": 0.11, "learning_rate": 9.702555046872118e-05, "loss": 1.5654, "step": 7320 }, { "epoch": 0.11, "learning_rate": 9.702147071831207e-05, "loss": 1.6092, "step": 7325 }, { "epoch": 0.11, "learning_rate": 9.701738825782289e-05, "loss": 1.5981, "step": 7330 }, { "epoch": 0.11, "learning_rate": 9.701330308748897e-05, "loss": 1.6163, "step": 7335 }, { "epoch": 0.11, "learning_rate": 9.70092152075457e-05, "loss": 1.6145, "step": 7340 }, { "epoch": 0.11, "learning_rate": 9.700512461822876e-05, "loss": 1.6079, "step": 7345 }, { "epoch": 0.11, "learning_rate": 9.700103131977384e-05, "loss": 1.6232, "step": 7350 }, { "epoch": 0.11, "learning_rate": 9.699693531241692e-05, "loss": 1.6109, "step": 7355 }, { "epoch": 0.11, "learning_rate": 9.699283659639402e-05, "loss": 1.6088, "step": 7360 }, { "epoch": 0.11, "learning_rate": 9.698873517194139e-05, "loss": 1.5976, "step": 7365 }, { "epoch": 0.11, "learning_rate": 9.698463103929542e-05, "loss": 1.5661, "step": 7370 }, { "epoch": 0.11, "learning_rate": 9.698052419869266e-05, "loss": 1.5951, "step": 7375 }, { "epoch": 0.11, "learning_rate": 9.697641465036977e-05, "loss": 1.6193, "step": 7380 }, { "epoch": 0.11, "learning_rate": 9.697230239456365e-05, "loss": 1.6136, "step": 7385 }, { "epoch": 0.11, "learning_rate": 9.696818743151128e-05, "loss": 1.5853, "step": 7390 }, { "epoch": 0.11, "learning_rate": 9.696406976144983e-05, "loss": 1.6221, "step": 7395 }, { "epoch": 0.11, "learning_rate": 9.695994938461664e-05, "loss": 1.6147, "step": 7400 }, { "epoch": 0.11, "learning_rate": 9.695582630124915e-05, "loss": 1.5673, "step": 7405 }, { "epoch": 0.11, "learning_rate": 9.695170051158505e-05, "loss": 1.6057, "step": 7410 }, { "epoch": 0.11, "learning_rate": 9.694757201586206e-05, "loss": 1.6194, "step": 7415 }, { "epoch": 0.11, "learning_rate": 9.694344081431819e-05, "loss": 1.6089, "step": 7420 }, { "epoch": 0.11, "learning_rate": 9.69393069071915e-05, "loss": 1.6547, "step": 7425 }, { "epoch": 0.11, "learning_rate": 9.693517029472027e-05, "loss": 1.6268, "step": 7430 }, { "epoch": 0.11, "learning_rate": 9.69310309771429e-05, "loss": 1.5767, "step": 7435 }, { "epoch": 0.11, "learning_rate": 9.692688895469797e-05, "loss": 1.6297, "step": 7440 }, { "epoch": 0.11, "learning_rate": 9.69227442276242e-05, "loss": 1.6467, "step": 7445 }, { "epoch": 0.11, "learning_rate": 9.691859679616046e-05, "loss": 1.5753, "step": 7450 }, { "epoch": 0.11, "learning_rate": 9.69144466605458e-05, "loss": 1.6253, "step": 7455 }, { "epoch": 0.11, "learning_rate": 9.691029382101941e-05, "loss": 1.6713, "step": 7460 }, { "epoch": 0.11, "learning_rate": 9.690613827782064e-05, "loss": 1.6127, "step": 7465 }, { "epoch": 0.11, "learning_rate": 9.6901980031189e-05, "loss": 1.5985, "step": 7470 }, { "epoch": 0.11, "learning_rate": 9.689781908136415e-05, "loss": 1.6186, "step": 7475 }, { "epoch": 0.11, "learning_rate": 9.68936554285859e-05, "loss": 1.6039, "step": 7480 }, { "epoch": 0.11, "learning_rate": 9.688948907309421e-05, "loss": 1.6033, "step": 7485 }, { "epoch": 0.11, "learning_rate": 9.688532001512923e-05, "loss": 1.5756, "step": 7490 }, { "epoch": 0.11, "learning_rate": 9.688114825493122e-05, "loss": 1.5675, "step": 7495 }, { "epoch": 0.11, "learning_rate": 9.687697379274064e-05, "loss": 1.5617, "step": 7500 }, { "epoch": 0.11, "learning_rate": 9.687279662879807e-05, "loss": 1.6428, "step": 7505 }, { "epoch": 0.11, "learning_rate": 9.686861676334429e-05, "loss": 1.6454, "step": 7510 }, { "epoch": 0.11, "learning_rate": 9.686443419662017e-05, "loss": 1.6287, "step": 7515 }, { "epoch": 0.11, "learning_rate": 9.686024892886678e-05, "loss": 1.6224, "step": 7520 }, { "epoch": 0.11, "learning_rate": 9.685606096032535e-05, "loss": 1.5567, "step": 7525 }, { "epoch": 0.11, "learning_rate": 9.685187029123725e-05, "loss": 1.6552, "step": 7530 }, { "epoch": 0.11, "learning_rate": 9.684767692184401e-05, "loss": 1.6393, "step": 7535 }, { "epoch": 0.12, "learning_rate": 9.684348085238731e-05, "loss": 1.5943, "step": 7540 }, { "epoch": 0.12, "learning_rate": 9.6839282083109e-05, "loss": 1.6192, "step": 7545 }, { "epoch": 0.12, "learning_rate": 9.683508061425104e-05, "loss": 1.6128, "step": 7550 }, { "epoch": 0.12, "learning_rate": 9.683087644605563e-05, "loss": 1.6191, "step": 7555 }, { "epoch": 0.12, "learning_rate": 9.682666957876504e-05, "loss": 1.6192, "step": 7560 }, { "epoch": 0.12, "learning_rate": 9.682246001262176e-05, "loss": 1.5839, "step": 7565 }, { "epoch": 0.12, "learning_rate": 9.681824774786841e-05, "loss": 1.6477, "step": 7570 }, { "epoch": 0.12, "learning_rate": 9.681403278474772e-05, "loss": 1.6408, "step": 7575 }, { "epoch": 0.12, "learning_rate": 9.680981512350267e-05, "loss": 1.6099, "step": 7580 }, { "epoch": 0.12, "learning_rate": 9.680559476437632e-05, "loss": 1.6117, "step": 7585 }, { "epoch": 0.12, "learning_rate": 9.680137170761191e-05, "loss": 1.6, "step": 7590 }, { "epoch": 0.12, "learning_rate": 9.679714595345284e-05, "loss": 1.6039, "step": 7595 }, { "epoch": 0.12, "learning_rate": 9.679291750214267e-05, "loss": 1.6813, "step": 7600 }, { "epoch": 0.12, "learning_rate": 9.678868635392508e-05, "loss": 1.6116, "step": 7605 }, { "epoch": 0.12, "learning_rate": 9.678445250904396e-05, "loss": 1.5822, "step": 7610 }, { "epoch": 0.12, "learning_rate": 9.678021596774331e-05, "loss": 1.5977, "step": 7615 }, { "epoch": 0.12, "learning_rate": 9.677597673026733e-05, "loss": 1.617, "step": 7620 }, { "epoch": 0.12, "learning_rate": 9.677173479686032e-05, "loss": 1.606, "step": 7625 }, { "epoch": 0.12, "learning_rate": 9.676749016776676e-05, "loss": 1.6004, "step": 7630 }, { "epoch": 0.12, "learning_rate": 9.67632428432313e-05, "loss": 1.6532, "step": 7635 }, { "epoch": 0.12, "learning_rate": 9.675899282349874e-05, "loss": 1.5298, "step": 7640 }, { "epoch": 0.12, "learning_rate": 9.675474010881404e-05, "loss": 1.5748, "step": 7645 }, { "epoch": 0.12, "learning_rate": 9.675048469942227e-05, "loss": 1.6521, "step": 7650 }, { "epoch": 0.12, "learning_rate": 9.674622659556871e-05, "loss": 1.6279, "step": 7655 }, { "epoch": 0.12, "learning_rate": 9.674196579749879e-05, "loss": 1.6, "step": 7660 }, { "epoch": 0.12, "learning_rate": 9.673770230545806e-05, "loss": 1.5938, "step": 7665 }, { "epoch": 0.12, "learning_rate": 9.673343611969225e-05, "loss": 1.6041, "step": 7670 }, { "epoch": 0.12, "learning_rate": 9.672916724044725e-05, "loss": 1.6599, "step": 7675 }, { "epoch": 0.12, "learning_rate": 9.672489566796908e-05, "loss": 1.645, "step": 7680 }, { "epoch": 0.12, "learning_rate": 9.672062140250395e-05, "loss": 1.6294, "step": 7685 }, { "epoch": 0.12, "learning_rate": 9.671634444429818e-05, "loss": 1.6174, "step": 7690 }, { "epoch": 0.12, "learning_rate": 9.671206479359832e-05, "loss": 1.651, "step": 7695 }, { "epoch": 0.12, "learning_rate": 9.670778245065099e-05, "loss": 1.6639, "step": 7700 }, { "epoch": 0.12, "learning_rate": 9.670349741570301e-05, "loss": 1.5687, "step": 7705 }, { "epoch": 0.12, "learning_rate": 9.669920968900134e-05, "loss": 1.6014, "step": 7710 }, { "epoch": 0.12, "learning_rate": 9.669491927079313e-05, "loss": 1.5772, "step": 7715 }, { "epoch": 0.12, "learning_rate": 9.669062616132563e-05, "loss": 1.5983, "step": 7720 }, { "epoch": 0.12, "learning_rate": 9.668633036084629e-05, "loss": 1.6228, "step": 7725 }, { "epoch": 0.12, "learning_rate": 9.668203186960268e-05, "loss": 1.57, "step": 7730 }, { "epoch": 0.12, "learning_rate": 9.667773068784257e-05, "loss": 1.5745, "step": 7735 }, { "epoch": 0.12, "learning_rate": 9.667342681581384e-05, "loss": 1.6353, "step": 7740 }, { "epoch": 0.12, "learning_rate": 9.666912025376454e-05, "loss": 1.6234, "step": 7745 }, { "epoch": 0.12, "learning_rate": 9.666481100194288e-05, "loss": 1.5552, "step": 7750 }, { "epoch": 0.12, "learning_rate": 9.666049906059724e-05, "loss": 1.6343, "step": 7755 }, { "epoch": 0.12, "learning_rate": 9.665618442997614e-05, "loss": 1.6117, "step": 7760 }, { "epoch": 0.12, "learning_rate": 9.665186711032824e-05, "loss": 1.6222, "step": 7765 }, { "epoch": 0.12, "learning_rate": 9.664754710190234e-05, "loss": 1.6167, "step": 7770 }, { "epoch": 0.12, "learning_rate": 9.664322440494747e-05, "loss": 1.6486, "step": 7775 }, { "epoch": 0.12, "learning_rate": 9.663889901971275e-05, "loss": 1.573, "step": 7780 }, { "epoch": 0.12, "learning_rate": 9.663457094644748e-05, "loss": 1.6136, "step": 7785 }, { "epoch": 0.12, "learning_rate": 9.66302401854011e-05, "loss": 1.6335, "step": 7790 }, { "epoch": 0.12, "learning_rate": 9.662590673682322e-05, "loss": 1.6564, "step": 7795 }, { "epoch": 0.12, "learning_rate": 9.662157060096358e-05, "loss": 1.6149, "step": 7800 }, { "epoch": 0.12, "learning_rate": 9.661723177807212e-05, "loss": 1.6329, "step": 7805 }, { "epoch": 0.12, "learning_rate": 9.66128902683989e-05, "loss": 1.595, "step": 7810 }, { "epoch": 0.12, "learning_rate": 9.660854607219413e-05, "loss": 1.5928, "step": 7815 }, { "epoch": 0.12, "learning_rate": 9.66041991897082e-05, "loss": 1.5984, "step": 7820 }, { "epoch": 0.12, "learning_rate": 9.659984962119161e-05, "loss": 1.6303, "step": 7825 }, { "epoch": 0.12, "learning_rate": 9.659549736689511e-05, "loss": 1.6231, "step": 7830 }, { "epoch": 0.12, "learning_rate": 9.65911424270695e-05, "loss": 1.6611, "step": 7835 }, { "epoch": 0.12, "learning_rate": 9.658678480196578e-05, "loss": 1.6076, "step": 7840 }, { "epoch": 0.12, "learning_rate": 9.658242449183509e-05, "loss": 1.6481, "step": 7845 }, { "epoch": 0.12, "learning_rate": 9.657806149692879e-05, "loss": 1.6226, "step": 7850 }, { "epoch": 0.12, "learning_rate": 9.657369581749826e-05, "loss": 1.6502, "step": 7855 }, { "epoch": 0.12, "learning_rate": 9.656932745379519e-05, "loss": 1.622, "step": 7860 }, { "epoch": 0.12, "learning_rate": 9.65649564060713e-05, "loss": 1.6109, "step": 7865 }, { "epoch": 0.12, "learning_rate": 9.656058267457853e-05, "loss": 1.6654, "step": 7870 }, { "epoch": 0.12, "learning_rate": 9.655620625956898e-05, "loss": 1.633, "step": 7875 }, { "epoch": 0.12, "learning_rate": 9.655182716129487e-05, "loss": 1.6294, "step": 7880 }, { "epoch": 0.12, "learning_rate": 9.654744538000857e-05, "loss": 1.6206, "step": 7885 }, { "epoch": 0.12, "learning_rate": 9.654306091596264e-05, "loss": 1.6438, "step": 7890 }, { "epoch": 0.12, "learning_rate": 9.653867376940979e-05, "loss": 1.5972, "step": 7895 }, { "epoch": 0.12, "learning_rate": 9.653428394060286e-05, "loss": 1.6636, "step": 7900 }, { "epoch": 0.12, "learning_rate": 9.652989142979485e-05, "loss": 1.6485, "step": 7905 }, { "epoch": 0.12, "learning_rate": 9.652549623723894e-05, "loss": 1.629, "step": 7910 }, { "epoch": 0.12, "learning_rate": 9.652109836318845e-05, "loss": 1.5601, "step": 7915 }, { "epoch": 0.12, "learning_rate": 9.651669780789683e-05, "loss": 1.6247, "step": 7920 }, { "epoch": 0.12, "learning_rate": 9.65122945716177e-05, "loss": 1.5719, "step": 7925 }, { "epoch": 0.12, "learning_rate": 9.650788865460487e-05, "loss": 1.6249, "step": 7930 }, { "epoch": 0.12, "learning_rate": 9.650348005711226e-05, "loss": 1.6471, "step": 7935 }, { "epoch": 0.12, "learning_rate": 9.649906877939396e-05, "loss": 1.6096, "step": 7940 }, { "epoch": 0.12, "learning_rate": 9.649465482170421e-05, "loss": 1.5979, "step": 7945 }, { "epoch": 0.12, "learning_rate": 9.649023818429742e-05, "loss": 1.6237, "step": 7950 }, { "epoch": 0.12, "learning_rate": 9.648581886742814e-05, "loss": 1.6223, "step": 7955 }, { "epoch": 0.12, "learning_rate": 9.648139687135105e-05, "loss": 1.6445, "step": 7960 }, { "epoch": 0.12, "learning_rate": 9.647697219632104e-05, "loss": 1.6062, "step": 7965 }, { "epoch": 0.12, "learning_rate": 9.647254484259313e-05, "loss": 1.6181, "step": 7970 }, { "epoch": 0.12, "learning_rate": 9.646811481042247e-05, "loss": 1.6448, "step": 7975 }, { "epoch": 0.12, "learning_rate": 9.64636821000644e-05, "loss": 1.5937, "step": 7980 }, { "epoch": 0.12, "learning_rate": 9.645924671177437e-05, "loss": 1.6271, "step": 7985 }, { "epoch": 0.12, "learning_rate": 9.645480864580807e-05, "loss": 1.5989, "step": 7990 }, { "epoch": 0.12, "learning_rate": 9.645036790242123e-05, "loss": 1.6162, "step": 7995 }, { "epoch": 0.12, "learning_rate": 9.644592448186983e-05, "loss": 1.5771, "step": 8000 } ], "logging_steps": 5, "max_steps": 65530, "num_train_epochs": 1, "save_steps": 1000, "total_flos": 2.122051858150195e+19, "trial_name": null, "trial_params": null }