{ "best_metric": null, "best_model_checkpoint": null, "epoch": 0.11911940976332462, "global_step": 4000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 2.7443463702742214e-06, "loss": 0.9983, "step": 10 }, { "epoch": 0.0, "learning_rate": 3.6776215359677665e-06, "loss": 0.9472, "step": 20 }, { "epoch": 0.0, "learning_rate": 4.205772220091474e-06, "loss": 0.8982, "step": 30 }, { "epoch": 0.0, "learning_rate": 4.5758099602275054e-06, "loss": 0.9054, "step": 40 }, { "epoch": 0.0, "learning_rate": 4.860906353882052e-06, "loss": 0.8821, "step": 50 }, { "epoch": 0.0, "learning_rate": 5.071517109647391e-06, "loss": 0.8922, "step": 60 }, { "epoch": 0.0, "learning_rate": 5.270189845174036e-06, "loss": 0.9029, "step": 70 }, { "epoch": 0.0, "learning_rate": 5.441554849783425e-06, "loss": 0.9273, "step": 80 }, { "epoch": 0.0, "learning_rate": 5.592219913271158e-06, "loss": 0.8706, "step": 90 }, { "epoch": 0.0, "learning_rate": 5.726651243437972e-06, "loss": 0.8855, "step": 100 }, { "epoch": 0.0, "learning_rate": 5.848009334523169e-06, "loss": 0.8569, "step": 110 }, { "epoch": 0.0, "learning_rate": 5.958613046622581e-06, "loss": 0.87, "step": 120 }, { "epoch": 0.0, "learning_rate": 6.0602142268914325e-06, "loss": 0.8761, "step": 130 }, { "epoch": 0.0, "learning_rate": 6.1541687207389654e-06, "loss": 0.8512, "step": 140 }, { "epoch": 0.0, "learning_rate": 6.241547407784983e-06, "loss": 0.8554, "step": 150 }, { "epoch": 0.0, "learning_rate": 6.323210857171233e-06, "loss": 0.8594, "step": 160 }, { "epoch": 0.01, "learning_rate": 6.399861030745893e-06, "loss": 0.8442, "step": 170 }, { "epoch": 0.01, "learning_rate": 6.465041342203772e-06, "loss": 0.8634, "step": 180 }, { "epoch": 0.01, "learning_rate": 6.533685310665599e-06, "loss": 0.8689, "step": 190 }, { "epoch": 0.01, "learning_rate": 6.598752410624751e-06, "loss": 0.908, "step": 200 }, { "epoch": 0.01, "learning_rate": 6.660597016320156e-06, "loss": 0.8696, "step": 210 }, { "epoch": 0.01, "learning_rate": 6.719523314515253e-06, "loss": 0.862, "step": 220 }, { "epoch": 0.01, "learning_rate": 6.775794355733718e-06, "loss": 0.8188, "step": 230 }, { "epoch": 0.01, "learning_rate": 6.829639152752426e-06, "loss": 0.8459, "step": 240 }, { "epoch": 0.01, "learning_rate": 6.881258311058162e-06, "loss": 0.8276, "step": 250 }, { "epoch": 0.01, "learning_rate": 6.930828541168008e-06, "loss": 0.8474, "step": 260 }, { "epoch": 0.01, "learning_rate": 6.978506309122687e-06, "loss": 0.8767, "step": 270 }, { "epoch": 0.01, "learning_rate": 7.024430815431458e-06, "loss": 0.846, "step": 280 }, { "epoch": 0.01, "learning_rate": 7.064366909249716e-06, "loss": 0.8219, "step": 290 }, { "epoch": 0.01, "learning_rate": 7.107292297340902e-06, "loss": 0.85, "step": 300 }, { "epoch": 0.01, "learning_rate": 7.148791325892338e-06, "loss": 0.8484, "step": 310 }, { "epoch": 0.01, "learning_rate": 7.188955746727152e-06, "loss": 0.8553, "step": 320 }, { "epoch": 0.01, "learning_rate": 7.227868733817176e-06, "loss": 0.845, "step": 330 }, { "epoch": 0.01, "learning_rate": 7.265605920301813e-06, "loss": 0.8831, "step": 340 }, { "epoch": 0.01, "learning_rate": 7.302236283380594e-06, "loss": 0.8332, "step": 350 }, { "epoch": 0.01, "learning_rate": 7.337822903097415e-06, "loss": 0.8109, "step": 360 }, { "epoch": 0.01, "learning_rate": 7.3724236159865746e-06, "loss": 0.8536, "step": 370 }, { "epoch": 0.01, "learning_rate": 7.406091580591952e-06, "loss": 0.8525, "step": 380 }, { "epoch": 0.01, "learning_rate": 7.438875768743633e-06, "loss": 0.8468, "step": 390 }, { "epoch": 0.01, "learning_rate": 7.47082139398946e-06, "loss": 0.8792, "step": 400 }, { "epoch": 0.01, "learning_rate": 7.501970286588418e-06, "loss": 0.8434, "step": 410 }, { "epoch": 0.01, "learning_rate": 7.53236122286999e-06, "loss": 0.8485, "step": 420 }, { "epoch": 0.01, "learning_rate": 7.562030215465441e-06, "loss": 0.8467, "step": 430 }, { "epoch": 0.01, "learning_rate": 7.59101076985994e-06, "loss": 0.8311, "step": 440 }, { "epoch": 0.01, "learning_rate": 7.619334111849203e-06, "loss": 0.8438, "step": 450 }, { "epoch": 0.01, "learning_rate": 7.647029389772635e-06, "loss": 0.875, "step": 460 }, { "epoch": 0.01, "learning_rate": 7.674123854806827e-06, "loss": 0.8308, "step": 470 }, { "epoch": 0.01, "learning_rate": 7.700643022115063e-06, "loss": 0.8445, "step": 480 }, { "epoch": 0.01, "learning_rate": 7.72661081524147e-06, "loss": 0.8857, "step": 490 }, { "epoch": 0.01, "learning_rate": 7.752049695797903e-06, "loss": 0.8448, "step": 500 }, { "epoch": 0.02, "learning_rate": 7.776980780205255e-06, "loss": 0.827, "step": 510 }, { "epoch": 0.02, "learning_rate": 7.801423945009691e-06, "loss": 0.7976, "step": 520 }, { "epoch": 0.02, "learning_rate": 7.825397922089763e-06, "loss": 0.8469, "step": 530 }, { "epoch": 0.02, "learning_rate": 7.848920384896975e-06, "loss": 0.8477, "step": 540 }, { "epoch": 0.02, "learning_rate": 7.87200802672445e-06, "loss": 0.819, "step": 550 }, { "epoch": 0.02, "learning_rate": 7.894676631871945e-06, "loss": 0.822, "step": 560 }, { "epoch": 0.02, "learning_rate": 7.916941140467113e-06, "loss": 0.8273, "step": 570 }, { "epoch": 0.02, "learning_rate": 7.938815707609735e-06, "loss": 0.8637, "step": 580 }, { "epoch": 0.02, "learning_rate": 7.960313757425318e-06, "loss": 0.8359, "step": 590 }, { "epoch": 0.02, "learning_rate": 7.98144803254492e-06, "loss": 0.8353, "step": 600 }, { "epoch": 0.02, "learning_rate": 8.002230639467933e-06, "loss": 0.8416, "step": 610 }, { "epoch": 0.02, "learning_rate": 8.022673090212183e-06, "loss": 0.8389, "step": 620 }, { "epoch": 0.02, "learning_rate": 8.042786340610085e-06, "loss": 0.835, "step": 630 }, { "epoch": 0.02, "learning_rate": 8.062580825569854e-06, "loss": 0.8712, "step": 640 }, { "epoch": 0.02, "learning_rate": 8.082066491585883e-06, "loss": 0.8419, "step": 650 }, { "epoch": 0.02, "learning_rate": 8.101252826751851e-06, "loss": 0.8674, "step": 660 }, { "epoch": 0.02, "learning_rate": 8.120148888503286e-06, "loss": 0.8224, "step": 670 }, { "epoch": 0.02, "learning_rate": 8.13876332929263e-06, "loss": 0.828, "step": 680 }, { "epoch": 0.02, "learning_rate": 8.157104420378997e-06, "loss": 0.8568, "step": 690 }, { "epoch": 0.02, "learning_rate": 8.17518007389634e-06, "loss": 0.8216, "step": 700 }, { "epoch": 0.02, "learning_rate": 8.192997863347404e-06, "loss": 0.7905, "step": 710 }, { "epoch": 0.02, "learning_rate": 8.210565042656325e-06, "loss": 0.8285, "step": 720 }, { "epoch": 0.02, "learning_rate": 8.22788856389983e-06, "loss": 0.7934, "step": 730 }, { "epoch": 0.02, "learning_rate": 8.244975093825531e-06, "loss": 0.8617, "step": 740 }, { "epoch": 0.02, "learning_rate": 8.26183102925549e-06, "loss": 0.8155, "step": 750 }, { "epoch": 0.02, "learning_rate": 8.278462511464196e-06, "loss": 0.7782, "step": 760 }, { "epoch": 0.02, "learning_rate": 8.294875439611728e-06, "loss": 0.8354, "step": 770 }, { "epoch": 0.02, "learning_rate": 8.311075483305743e-06, "loss": 0.8133, "step": 780 }, { "epoch": 0.02, "learning_rate": 8.32706809435913e-06, "loss": 0.835, "step": 790 }, { "epoch": 0.02, "learning_rate": 8.342858517804411e-06, "loss": 0.8339, "step": 800 }, { "epoch": 0.02, "learning_rate": 8.358451802220541e-06, "loss": 0.8192, "step": 810 }, { "epoch": 0.02, "learning_rate": 8.373852809422983e-06, "loss": 0.8187, "step": 820 }, { "epoch": 0.02, "learning_rate": 8.389066223563605e-06, "loss": 0.8202, "step": 830 }, { "epoch": 0.03, "learning_rate": 8.404096559683005e-06, "loss": 0.8373, "step": 840 }, { "epoch": 0.03, "learning_rate": 8.418948171754334e-06, "loss": 0.8085, "step": 850 }, { "epoch": 0.03, "learning_rate": 8.433625260254492e-06, "loss": 0.8292, "step": 860 }, { "epoch": 0.03, "learning_rate": 8.448131879295593e-06, "loss": 0.8149, "step": 870 }, { "epoch": 0.03, "learning_rate": 8.462471943347044e-06, "loss": 0.8174, "step": 880 }, { "epoch": 0.03, "learning_rate": 8.476649233576078e-06, "loss": 0.822, "step": 890 }, { "epoch": 0.03, "learning_rate": 8.49066740383246e-06, "loss": 0.8188, "step": 900 }, { "epoch": 0.03, "learning_rate": 8.504529986301062e-06, "loss": 0.8184, "step": 910 }, { "epoch": 0.03, "learning_rate": 8.518240396844167e-06, "loss": 0.8405, "step": 920 }, { "epoch": 0.03, "learning_rate": 8.531801940053751e-06, "loss": 0.8127, "step": 930 }, { "epoch": 0.03, "learning_rate": 8.545217814032373e-06, "loss": 0.8087, "step": 940 }, { "epoch": 0.03, "learning_rate": 8.558491114920063e-06, "loss": 0.7969, "step": 950 }, { "epoch": 0.03, "learning_rate": 8.571624841183157e-06, "loss": 0.7868, "step": 960 }, { "epoch": 0.03, "learning_rate": 8.58462189767999e-06, "loss": 0.803, "step": 970 }, { "epoch": 0.03, "learning_rate": 8.597485099517221e-06, "loss": 0.8116, "step": 980 }, { "epoch": 0.03, "learning_rate": 8.61021717570959e-06, "loss": 0.8084, "step": 990 }, { "epoch": 0.03, "learning_rate": 8.622820772655021e-06, "loss": 0.8238, "step": 1000 }, { "epoch": 0.03, "learning_rate": 8.635298457436133e-06, "loss": 0.8182, "step": 1010 }, { "epoch": 0.03, "learning_rate": 8.647652720958468e-06, "loss": 0.8431, "step": 1020 }, { "epoch": 0.03, "learning_rate": 8.659885980935016e-06, "loss": 0.8601, "step": 1030 }, { "epoch": 0.03, "learning_rate": 8.67200058472601e-06, "loss": 0.8029, "step": 1040 }, { "epoch": 0.03, "learning_rate": 8.683998812042295e-06, "loss": 0.8336, "step": 1050 }, { "epoch": 0.03, "learning_rate": 8.695882877520106e-06, "loss": 0.8158, "step": 1060 }, { "epoch": 0.03, "learning_rate": 8.70765493317449e-06, "loss": 0.7954, "step": 1070 }, { "epoch": 0.03, "learning_rate": 8.719317070738179e-06, "loss": 0.8156, "step": 1080 }, { "epoch": 0.03, "learning_rate": 8.73087132389229e-06, "loss": 0.826, "step": 1090 }, { "epoch": 0.03, "learning_rate": 8.742319670394788e-06, "loss": 0.8407, "step": 1100 }, { "epoch": 0.03, "learning_rate": 8.753664034112258e-06, "loss": 0.8149, "step": 1110 }, { "epoch": 0.03, "learning_rate": 8.764906286960285e-06, "loss": 0.7951, "step": 1120 }, { "epoch": 0.03, "learning_rate": 8.776048250757231e-06, "loss": 0.8036, "step": 1130 }, { "epoch": 0.03, "learning_rate": 8.787091698996089e-06, "loss": 0.8013, "step": 1140 }, { "epoch": 0.03, "learning_rate": 8.798038358538676e-06, "loss": 0.8109, "step": 1150 }, { "epoch": 0.03, "learning_rate": 8.808889911236233e-06, "loss": 0.8091, "step": 1160 }, { "epoch": 0.03, "learning_rate": 8.819647995480226e-06, "loss": 0.799, "step": 1170 }, { "epoch": 0.04, "learning_rate": 8.83031420768692e-06, "loss": 0.8179, "step": 1180 }, { "epoch": 0.04, "learning_rate": 8.840890103719111e-06, "loss": 0.7954, "step": 1190 }, { "epoch": 0.04, "learning_rate": 8.851377200248169e-06, "loss": 0.8444, "step": 1200 }, { "epoch": 0.04, "learning_rate": 8.861776976059358e-06, "loss": 0.8211, "step": 1210 }, { "epoch": 0.04, "learning_rate": 8.872090873303281e-06, "loss": 0.8336, "step": 1220 }, { "epoch": 0.04, "learning_rate": 8.88232029869606e-06, "loss": 0.8114, "step": 1230 }, { "epoch": 0.04, "learning_rate": 8.892466624670776e-06, "loss": 0.7884, "step": 1240 }, { "epoch": 0.04, "learning_rate": 8.90253119048252e-06, "loss": 0.7746, "step": 1250 }, { "epoch": 0.04, "learning_rate": 8.912515303269253e-06, "loss": 0.8045, "step": 1260 }, { "epoch": 0.04, "learning_rate": 8.922420239070625e-06, "loss": 0.8118, "step": 1270 }, { "epoch": 0.04, "learning_rate": 8.932247243806712e-06, "loss": 0.8386, "step": 1280 }, { "epoch": 0.04, "learning_rate": 8.941997534218523e-06, "loss": 0.7903, "step": 1290 }, { "epoch": 0.04, "learning_rate": 8.951672298772117e-06, "loss": 0.803, "step": 1300 }, { "epoch": 0.04, "learning_rate": 8.961272698527948e-06, "loss": 0.7786, "step": 1310 }, { "epoch": 0.04, "learning_rate": 8.97079986797706e-06, "loss": 0.8029, "step": 1320 }, { "epoch": 0.04, "learning_rate": 8.980254915845623e-06, "loss": 0.8261, "step": 1330 }, { "epoch": 0.04, "learning_rate": 8.98963892586927e-06, "loss": 0.7883, "step": 1340 }, { "epoch": 0.04, "learning_rate": 8.99895295753853e-06, "loss": 0.82, "step": 1350 }, { "epoch": 0.04, "learning_rate": 9.008198046816687e-06, "loss": 0.8107, "step": 1360 }, { "epoch": 0.04, "learning_rate": 9.017375206831278e-06, "loss": 0.8036, "step": 1370 }, { "epoch": 0.04, "learning_rate": 9.026485428540338e-06, "loss": 0.7946, "step": 1380 }, { "epoch": 0.04, "learning_rate": 9.035529681374568e-06, "loss": 0.8063, "step": 1390 }, { "epoch": 0.04, "learning_rate": 9.044508913856369e-06, "loss": 0.8009, "step": 1400 }, { "epoch": 0.04, "learning_rate": 9.053424054196832e-06, "loss": 0.8255, "step": 1410 }, { "epoch": 0.04, "learning_rate": 9.06227601087153e-06, "loss": 0.831, "step": 1420 }, { "epoch": 0.04, "learning_rate": 9.07106567317608e-06, "loss": 0.8133, "step": 1430 }, { "epoch": 0.04, "learning_rate": 9.079793911762282e-06, "loss": 0.8305, "step": 1440 }, { "epoch": 0.04, "learning_rate": 9.088461579155646e-06, "loss": 0.8283, "step": 1450 }, { "epoch": 0.04, "learning_rate": 9.097069510255097e-06, "loss": 0.815, "step": 1460 }, { "epoch": 0.04, "learning_rate": 9.105618522815576e-06, "loss": 0.8059, "step": 1470 }, { "epoch": 0.04, "learning_rate": 9.114109417914234e-06, "loss": 0.7878, "step": 1480 }, { "epoch": 0.04, "learning_rate": 9.122542980400885e-06, "loss": 0.8045, "step": 1490 }, { "epoch": 0.04, "learning_rate": 9.130919979333354e-06, "loss": 0.8232, "step": 1500 }, { "epoch": 0.04, "learning_rate": 9.139241168398338e-06, "loss": 0.812, "step": 1510 }, { "epoch": 0.05, "learning_rate": 9.14668313197706e-06, "loss": 0.8241, "step": 1520 }, { "epoch": 0.05, "learning_rate": 9.154900305423636e-06, "loss": 0.8143, "step": 1530 }, { "epoch": 0.05, "learning_rate": 9.163063771470579e-06, "loss": 0.8081, "step": 1540 }, { "epoch": 0.05, "learning_rate": 9.171174227623984e-06, "loss": 0.8189, "step": 1550 }, { "epoch": 0.05, "learning_rate": 9.179232357889644e-06, "loss": 0.8298, "step": 1560 }, { "epoch": 0.05, "learning_rate": 9.187238833119211e-06, "loss": 0.7892, "step": 1570 }, { "epoch": 0.05, "learning_rate": 9.195194311345337e-06, "loss": 0.7882, "step": 1580 }, { "epoch": 0.05, "learning_rate": 9.203099438106217e-06, "loss": 0.8157, "step": 1590 }, { "epoch": 0.05, "learning_rate": 9.21095484675992e-06, "loss": 0.7796, "step": 1600 }, { "epoch": 0.05, "learning_rate": 9.21876115878889e-06, "loss": 0.8145, "step": 1610 }, { "epoch": 0.05, "learning_rate": 9.226518984095009e-06, "loss": 0.8507, "step": 1620 }, { "epoch": 0.05, "learning_rate": 9.234228921285531e-06, "loss": 0.8343, "step": 1630 }, { "epoch": 0.05, "learning_rate": 9.241891557950258e-06, "loss": 0.8324, "step": 1640 }, { "epoch": 0.05, "learning_rate": 9.249507470930265e-06, "loss": 0.8095, "step": 1650 }, { "epoch": 0.05, "learning_rate": 9.257077226578481e-06, "loss": 0.8039, "step": 1660 }, { "epoch": 0.05, "learning_rate": 9.264601381012412e-06, "loss": 0.8153, "step": 1670 }, { "epoch": 0.05, "learning_rate": 9.272080480359307e-06, "loss": 0.8062, "step": 1680 }, { "epoch": 0.05, "learning_rate": 9.279515060994034e-06, "loss": 0.8195, "step": 1690 }, { "epoch": 0.05, "learning_rate": 9.286905649769893e-06, "loss": 0.8274, "step": 1700 }, { "epoch": 0.05, "learning_rate": 9.294252764242674e-06, "loss": 0.8118, "step": 1710 }, { "epoch": 0.05, "learning_rate": 9.301556912888123e-06, "loss": 0.8065, "step": 1720 }, { "epoch": 0.05, "learning_rate": 9.308818595313135e-06, "loss": 0.8255, "step": 1730 }, { "epoch": 0.05, "learning_rate": 9.3160383024608e-06, "loss": 0.7981, "step": 1740 }, { "epoch": 0.05, "learning_rate": 9.323216516809576e-06, "loss": 0.7966, "step": 1750 }, { "epoch": 0.05, "learning_rate": 9.330353712566771e-06, "loss": 0.7948, "step": 1760 }, { "epoch": 0.05, "learning_rate": 9.33745035585653e-06, "loss": 0.8022, "step": 1770 }, { "epoch": 0.05, "learning_rate": 9.344506904902502e-06, "loss": 0.7971, "step": 1780 }, { "epoch": 0.05, "learning_rate": 9.35152381020538e-06, "loss": 0.83, "step": 1790 }, { "epoch": 0.05, "learning_rate": 9.358501514715491e-06, "loss": 0.8204, "step": 1800 }, { "epoch": 0.05, "learning_rate": 9.365440454000577e-06, "loss": 0.7879, "step": 1810 }, { "epoch": 0.05, "learning_rate": 9.372341056408958e-06, "loss": 0.8028, "step": 1820 }, { "epoch": 0.05, "learning_rate": 9.379203743228185e-06, "loss": 0.812, "step": 1830 }, { "epoch": 0.05, "learning_rate": 9.386028928839394e-06, "loss": 0.7885, "step": 1840 }, { "epoch": 0.06, "learning_rate": 9.39281702086744e-06, "loss": 0.8224, "step": 1850 }, { "epoch": 0.06, "learning_rate": 9.399568420326976e-06, "loss": 0.8072, "step": 1860 }, { "epoch": 0.06, "learning_rate": 9.406283521764617e-06, "loss": 0.8267, "step": 1870 }, { "epoch": 0.06, "learning_rate": 9.41296271339729e-06, "loss": 0.8174, "step": 1880 }, { "epoch": 0.06, "learning_rate": 9.419606377246914e-06, "loss": 0.7966, "step": 1890 }, { "epoch": 0.06, "learning_rate": 9.42621488927151e-06, "loss": 0.8085, "step": 1900 }, { "epoch": 0.06, "learning_rate": 9.432788619492866e-06, "loss": 0.7715, "step": 1910 }, { "epoch": 0.06, "learning_rate": 9.439327932120854e-06, "loss": 0.7899, "step": 1920 }, { "epoch": 0.06, "learning_rate": 9.445833185674515e-06, "loss": 0.8234, "step": 1930 }, { "epoch": 0.06, "learning_rate": 9.45230473310001e-06, "loss": 0.7594, "step": 1940 }, { "epoch": 0.06, "learning_rate": 9.45874292188551e-06, "loss": 0.7989, "step": 1950 }, { "epoch": 0.06, "learning_rate": 9.465148094173177e-06, "loss": 0.8325, "step": 1960 }, { "epoch": 0.06, "learning_rate": 9.471520586868255e-06, "loss": 0.784, "step": 1970 }, { "epoch": 0.06, "learning_rate": 9.477860731745404e-06, "loss": 0.7925, "step": 1980 }, { "epoch": 0.06, "learning_rate": 9.484168855552354e-06, "loss": 0.7905, "step": 1990 }, { "epoch": 0.06, "learning_rate": 9.490445280110947e-06, "loss": 0.839, "step": 2000 }, { "epoch": 0.06, "learning_rate": 9.496690322415662e-06, "loss": 0.7983, "step": 2010 }, { "epoch": 0.06, "learning_rate": 9.502904294729667e-06, "loss": 0.7627, "step": 2020 }, { "epoch": 0.06, "learning_rate": 9.509087504678532e-06, "loss": 0.7854, "step": 2030 }, { "epoch": 0.06, "learning_rate": 9.515240255341591e-06, "loss": 0.8019, "step": 2040 }, { "epoch": 0.06, "learning_rate": 9.521362845341097e-06, "loss": 0.8063, "step": 2050 }, { "epoch": 0.06, "learning_rate": 9.527455568929186e-06, "loss": 0.7965, "step": 2060 }, { "epoch": 0.06, "learning_rate": 9.533518716072731e-06, "loss": 0.7828, "step": 2070 }, { "epoch": 0.06, "learning_rate": 9.539552572536162e-06, "loss": 0.7979, "step": 2080 }, { "epoch": 0.06, "learning_rate": 9.545557419962264e-06, "loss": 0.7961, "step": 2090 }, { "epoch": 0.06, "learning_rate": 9.551533535951078e-06, "loss": 0.8017, "step": 2100 }, { "epoch": 0.06, "learning_rate": 9.5574811941369e-06, "loss": 0.7808, "step": 2110 }, { "epoch": 0.06, "learning_rate": 9.56340066426346e-06, "loss": 0.8078, "step": 2120 }, { "epoch": 0.06, "learning_rate": 9.569292212257332e-06, "loss": 0.7956, "step": 2130 }, { "epoch": 0.06, "learning_rate": 9.575156100299618e-06, "loss": 0.792, "step": 2140 }, { "epoch": 0.06, "learning_rate": 9.580992586895953e-06, "loss": 0.8122, "step": 2150 }, { "epoch": 0.06, "learning_rate": 9.586801926944894e-06, "loss": 0.7944, "step": 2160 }, { "epoch": 0.06, "learning_rate": 9.592584371804719e-06, "loss": 0.8075, "step": 2170 }, { "epoch": 0.06, "learning_rate": 9.598340169358675e-06, "loss": 0.8205, "step": 2180 }, { "epoch": 0.07, "learning_rate": 9.604069564078745e-06, "loss": 0.8142, "step": 2190 }, { "epoch": 0.07, "learning_rate": 9.609772797087972e-06, "loss": 0.8124, "step": 2200 }, { "epoch": 0.07, "learning_rate": 9.615450106221318e-06, "loss": 0.8008, "step": 2210 }, { "epoch": 0.07, "learning_rate": 9.621101726085202e-06, "loss": 0.8091, "step": 2220 }, { "epoch": 0.07, "learning_rate": 9.626727888115666e-06, "loss": 0.7764, "step": 2230 }, { "epoch": 0.07, "learning_rate": 9.632328820635236e-06, "loss": 0.8022, "step": 2240 }, { "epoch": 0.07, "learning_rate": 9.637904748908531e-06, "loss": 0.8228, "step": 2250 }, { "epoch": 0.07, "learning_rate": 9.643455895196622e-06, "loss": 0.7955, "step": 2260 }, { "epoch": 0.07, "learning_rate": 9.648982478810187e-06, "loss": 0.7859, "step": 2270 }, { "epoch": 0.07, "learning_rate": 9.654484716161512e-06, "loss": 0.7956, "step": 2280 }, { "epoch": 0.07, "learning_rate": 9.659962820815331e-06, "loss": 0.8379, "step": 2290 }, { "epoch": 0.07, "learning_rate": 9.665417003538574e-06, "loss": 0.7958, "step": 2300 }, { "epoch": 0.07, "learning_rate": 9.670847472349013e-06, "loss": 0.8078, "step": 2310 }, { "epoch": 0.07, "learning_rate": 9.676254432562877e-06, "loss": 0.7764, "step": 2320 }, { "epoch": 0.07, "learning_rate": 9.681638086841426e-06, "loss": 0.7853, "step": 2330 }, { "epoch": 0.07, "learning_rate": 9.686998635236535e-06, "loss": 0.7811, "step": 2340 }, { "epoch": 0.07, "learning_rate": 9.69233627523529e-06, "loss": 0.7867, "step": 2350 }, { "epoch": 0.07, "learning_rate": 9.697651201803651e-06, "loss": 0.8049, "step": 2360 }, { "epoch": 0.07, "learning_rate": 9.70294360742919e-06, "loss": 0.7858, "step": 2370 }, { "epoch": 0.07, "learning_rate": 9.708213682162902e-06, "loss": 0.8043, "step": 2380 }, { "epoch": 0.07, "learning_rate": 9.713461613660172e-06, "loss": 0.7544, "step": 2390 }, { "epoch": 0.07, "learning_rate": 9.718687587220866e-06, "loss": 0.7901, "step": 2400 }, { "epoch": 0.07, "learning_rate": 9.723891785828587e-06, "loss": 0.8242, "step": 2410 }, { "epoch": 0.07, "learning_rate": 9.729074390189116e-06, "loss": 0.8037, "step": 2420 }, { "epoch": 0.07, "learning_rate": 9.734235578768077e-06, "loss": 0.7992, "step": 2430 }, { "epoch": 0.07, "learning_rate": 9.739375527827789e-06, "loss": 0.8321, "step": 2440 }, { "epoch": 0.07, "learning_rate": 9.744494411463412e-06, "loss": 0.7946, "step": 2450 }, { "epoch": 0.07, "learning_rate": 9.749592401638308e-06, "loss": 0.8284, "step": 2460 }, { "epoch": 0.07, "learning_rate": 9.754669668218715e-06, "loss": 0.8193, "step": 2470 }, { "epoch": 0.07, "learning_rate": 9.759726379007711e-06, "loss": 0.7918, "step": 2480 }, { "epoch": 0.07, "learning_rate": 9.764762699778483e-06, "loss": 0.7733, "step": 2490 }, { "epoch": 0.07, "learning_rate": 9.76977879430694e-06, "loss": 0.7545, "step": 2500 }, { "epoch": 0.07, "learning_rate": 9.774774824403668e-06, "loss": 0.7995, "step": 2510 }, { "epoch": 0.08, "learning_rate": 9.779750949945259e-06, "loss": 0.822, "step": 2520 }, { "epoch": 0.08, "learning_rate": 9.784707328904993e-06, "loss": 0.7891, "step": 2530 }, { "epoch": 0.08, "learning_rate": 9.789644117382946e-06, "loss": 0.8081, "step": 2540 }, { "epoch": 0.08, "learning_rate": 9.794561469635479e-06, "loss": 0.7719, "step": 2550 }, { "epoch": 0.08, "learning_rate": 9.799459538104166e-06, "loss": 0.7964, "step": 2560 }, { "epoch": 0.08, "learning_rate": 9.804338473444142e-06, "loss": 0.7849, "step": 2570 }, { "epoch": 0.08, "learning_rate": 9.80919842455192e-06, "loss": 0.7943, "step": 2580 }, { "epoch": 0.08, "learning_rate": 9.814039538592641e-06, "loss": 0.8179, "step": 2590 }, { "epoch": 0.08, "learning_rate": 9.81886196102683e-06, "loss": 0.7809, "step": 2600 }, { "epoch": 0.08, "learning_rate": 9.823665835636607e-06, "loss": 0.8191, "step": 2610 }, { "epoch": 0.08, "learning_rate": 9.82845130455142e-06, "loss": 0.774, "step": 2620 }, { "epoch": 0.08, "learning_rate": 9.83321850827327e-06, "loss": 0.8062, "step": 2630 }, { "epoch": 0.08, "learning_rate": 9.83796758570147e-06, "loss": 0.7829, "step": 2640 }, { "epoch": 0.08, "learning_rate": 9.842698674156915e-06, "loss": 0.7831, "step": 2650 }, { "epoch": 0.08, "learning_rate": 9.847411909405928e-06, "loss": 0.7677, "step": 2660 }, { "epoch": 0.08, "learning_rate": 9.852107425683615e-06, "loss": 0.8183, "step": 2670 }, { "epoch": 0.08, "learning_rate": 9.856785355716825e-06, "loss": 0.7847, "step": 2680 }, { "epoch": 0.08, "learning_rate": 9.861445830746643e-06, "loss": 0.7636, "step": 2690 }, { "epoch": 0.08, "learning_rate": 9.866088980550497e-06, "loss": 0.8071, "step": 2700 }, { "epoch": 0.08, "learning_rate": 9.870714933463837e-06, "loss": 0.8048, "step": 2710 }, { "epoch": 0.08, "learning_rate": 9.875323816401418e-06, "loss": 0.7643, "step": 2720 }, { "epoch": 0.08, "learning_rate": 9.879915754878191e-06, "loss": 0.8177, "step": 2730 }, { "epoch": 0.08, "learning_rate": 9.884490873029821e-06, "loss": 0.8113, "step": 2740 }, { "epoch": 0.08, "learning_rate": 9.889049293632811e-06, "loss": 0.8018, "step": 2750 }, { "epoch": 0.08, "learning_rate": 9.893591138124284e-06, "loss": 0.7857, "step": 2760 }, { "epoch": 0.08, "learning_rate": 9.898116526621384e-06, "loss": 0.7321, "step": 2770 }, { "epoch": 0.08, "learning_rate": 9.902625577940345e-06, "loss": 0.7987, "step": 2780 }, { "epoch": 0.08, "learning_rate": 9.907118409615204e-06, "loss": 0.795, "step": 2790 }, { "epoch": 0.08, "learning_rate": 9.911595137916184e-06, "loss": 0.7824, "step": 2800 }, { "epoch": 0.08, "learning_rate": 9.916055877867754e-06, "loss": 0.7657, "step": 2810 }, { "epoch": 0.08, "learning_rate": 9.920500743266357e-06, "loss": 0.7808, "step": 2820 }, { "epoch": 0.08, "learning_rate": 9.924929846697829e-06, "loss": 0.7936, "step": 2830 }, { "epoch": 0.08, "learning_rate": 9.929343299554512e-06, "loss": 0.7835, "step": 2840 }, { "epoch": 0.08, "learning_rate": 9.933741212052067e-06, "loss": 0.7982, "step": 2850 }, { "epoch": 0.09, "learning_rate": 9.938123693245983e-06, "loss": 0.7795, "step": 2860 }, { "epoch": 0.09, "learning_rate": 9.9424908510478e-06, "loss": 0.7723, "step": 2870 }, { "epoch": 0.09, "learning_rate": 9.94684279224107e-06, "loss": 0.7899, "step": 2880 }, { "epoch": 0.09, "learning_rate": 9.951179622497001e-06, "loss": 0.7733, "step": 2890 }, { "epoch": 0.09, "learning_rate": 9.955501446389869e-06, "loss": 0.7995, "step": 2900 }, { "epoch": 0.09, "learning_rate": 9.959808367412143e-06, "loss": 0.8119, "step": 2910 }, { "epoch": 0.09, "learning_rate": 9.964100487989352e-06, "loss": 0.8068, "step": 2920 }, { "epoch": 0.09, "learning_rate": 9.968377909494707e-06, "loss": 0.7862, "step": 2930 }, { "epoch": 0.09, "learning_rate": 9.97264073226346e-06, "loss": 0.7889, "step": 2940 }, { "epoch": 0.09, "learning_rate": 9.976889055607038e-06, "loss": 0.7854, "step": 2950 }, { "epoch": 0.09, "learning_rate": 9.981122977826906e-06, "loss": 0.7908, "step": 2960 }, { "epoch": 0.09, "learning_rate": 9.985342596228237e-06, "loss": 0.761, "step": 2970 }, { "epoch": 0.09, "learning_rate": 9.989548007133313e-06, "loss": 0.806, "step": 2980 }, { "epoch": 0.09, "learning_rate": 9.993739305894728e-06, "loss": 0.7941, "step": 2990 }, { "epoch": 0.09, "learning_rate": 9.997916586908357e-06, "loss": 0.8242, "step": 3000 }, { "epoch": 0.09, "learning_rate": 1e-05, "loss": 0.8142, "step": 3010 }, { "epoch": 0.09, "learning_rate": 1e-05, "loss": 0.8038, "step": 3020 }, { "epoch": 0.09, "learning_rate": 1e-05, "loss": 0.7903, "step": 3030 }, { "epoch": 0.09, "learning_rate": 1e-05, "loss": 0.7369, "step": 3040 }, { "epoch": 0.09, "learning_rate": 1e-05, "loss": 0.7631, "step": 3050 }, { "epoch": 0.09, "learning_rate": 1e-05, "loss": 0.8063, "step": 3060 }, { "epoch": 0.09, "learning_rate": 1e-05, "loss": 0.7876, "step": 3070 }, { "epoch": 0.09, "learning_rate": 1e-05, "loss": 0.7775, "step": 3080 }, { "epoch": 0.09, "learning_rate": 1e-05, "loss": 0.8087, "step": 3090 }, { "epoch": 0.09, "learning_rate": 1e-05, "loss": 0.7963, "step": 3100 }, { "epoch": 0.09, "learning_rate": 1e-05, "loss": 0.7648, "step": 3110 }, { "epoch": 0.09, "learning_rate": 1e-05, "loss": 0.7854, "step": 3120 }, { "epoch": 0.09, "learning_rate": 1e-05, "loss": 0.8144, "step": 3130 }, { "epoch": 0.09, "learning_rate": 1e-05, "loss": 0.7879, "step": 3140 }, { "epoch": 0.09, "learning_rate": 1e-05, "loss": 0.7973, "step": 3150 }, { "epoch": 0.09, "learning_rate": 1e-05, "loss": 0.7661, "step": 3160 }, { "epoch": 0.09, "learning_rate": 1e-05, "loss": 0.8059, "step": 3170 }, { "epoch": 0.09, "learning_rate": 1e-05, "loss": 0.7926, "step": 3180 }, { "epoch": 0.09, "learning_rate": 1e-05, "loss": 0.8048, "step": 3190 }, { "epoch": 0.1, "learning_rate": 1e-05, "loss": 0.8069, "step": 3200 }, { "epoch": 0.1, "learning_rate": 1e-05, "loss": 0.7716, "step": 3210 }, { "epoch": 0.1, "learning_rate": 1e-05, "loss": 0.7904, "step": 3220 }, { "epoch": 0.1, "learning_rate": 1e-05, "loss": 0.7995, "step": 3230 }, { "epoch": 0.1, "learning_rate": 1e-05, "loss": 0.8062, "step": 3240 }, { "epoch": 0.1, "learning_rate": 1e-05, "loss": 0.7894, "step": 3250 }, { "epoch": 0.1, "learning_rate": 1e-05, "loss": 0.7912, "step": 3260 }, { "epoch": 0.1, "learning_rate": 1e-05, "loss": 0.7735, "step": 3270 }, { "epoch": 0.1, "learning_rate": 1e-05, "loss": 0.7838, "step": 3280 }, { "epoch": 0.1, "learning_rate": 1e-05, "loss": 0.7779, "step": 3290 }, { "epoch": 0.1, "learning_rate": 1e-05, "loss": 0.7471, "step": 3300 }, { "epoch": 0.1, "learning_rate": 1e-05, "loss": 0.7632, "step": 3310 }, { "epoch": 0.1, "learning_rate": 1e-05, "loss": 0.8012, "step": 3320 }, { "epoch": 0.1, "learning_rate": 1e-05, "loss": 0.7941, "step": 3330 }, { "epoch": 0.1, "learning_rate": 1e-05, "loss": 0.7791, "step": 3340 }, { "epoch": 0.1, "learning_rate": 1e-05, "loss": 0.8017, "step": 3350 }, { "epoch": 0.1, "learning_rate": 1e-05, "loss": 0.8195, "step": 3360 }, { "epoch": 0.1, "learning_rate": 1e-05, "loss": 0.7782, "step": 3370 }, { "epoch": 0.1, "learning_rate": 1e-05, "loss": 0.7576, "step": 3380 }, { "epoch": 0.1, "learning_rate": 1e-05, "loss": 0.7975, "step": 3390 }, { "epoch": 0.1, "learning_rate": 1e-05, "loss": 0.7682, "step": 3400 }, { "epoch": 0.1, "learning_rate": 1e-05, "loss": 0.7719, "step": 3410 }, { "epoch": 0.1, "learning_rate": 1e-05, "loss": 0.7625, "step": 3420 }, { "epoch": 0.1, "learning_rate": 1e-05, "loss": 0.7799, "step": 3430 }, { "epoch": 0.1, "learning_rate": 1e-05, "loss": 0.7617, "step": 3440 }, { "epoch": 0.1, "learning_rate": 1e-05, "loss": 0.762, "step": 3450 }, { "epoch": 0.1, "learning_rate": 1e-05, "loss": 0.8049, "step": 3460 }, { "epoch": 0.1, "learning_rate": 1e-05, "loss": 0.7971, "step": 3470 }, { "epoch": 0.1, "learning_rate": 1e-05, "loss": 0.8157, "step": 3480 }, { "epoch": 0.1, "learning_rate": 1e-05, "loss": 0.7716, "step": 3490 }, { "epoch": 0.1, "learning_rate": 1e-05, "loss": 0.7746, "step": 3500 }, { "epoch": 0.1, "learning_rate": 1e-05, "loss": 0.7614, "step": 3510 }, { "epoch": 0.1, "learning_rate": 1e-05, "loss": 0.7895, "step": 3520 }, { "epoch": 0.11, "learning_rate": 1e-05, "loss": 0.8076, "step": 3530 }, { "epoch": 0.11, "learning_rate": 1e-05, "loss": 0.7599, "step": 3540 }, { "epoch": 0.11, "learning_rate": 1e-05, "loss": 0.7956, "step": 3550 }, { "epoch": 0.11, "learning_rate": 1e-05, "loss": 0.7751, "step": 3560 }, { "epoch": 0.11, "learning_rate": 1e-05, "loss": 0.7697, "step": 3570 }, { "epoch": 0.11, "learning_rate": 1e-05, "loss": 0.7982, "step": 3580 }, { "epoch": 0.11, "learning_rate": 1e-05, "loss": 0.7622, "step": 3590 }, { "epoch": 0.11, "learning_rate": 1e-05, "loss": 0.7716, "step": 3600 }, { "epoch": 0.11, "learning_rate": 1e-05, "loss": 0.7761, "step": 3610 }, { "epoch": 0.11, "learning_rate": 1e-05, "loss": 0.8177, "step": 3620 }, { "epoch": 0.11, "learning_rate": 1e-05, "loss": 0.7722, "step": 3630 }, { "epoch": 0.11, "learning_rate": 1e-05, "loss": 0.7809, "step": 3640 }, { "epoch": 0.11, "learning_rate": 1e-05, "loss": 0.8001, "step": 3650 }, { "epoch": 0.11, "learning_rate": 1e-05, "loss": 0.7641, "step": 3660 }, { "epoch": 0.11, "learning_rate": 1e-05, "loss": 0.7582, "step": 3670 }, { "epoch": 0.11, "learning_rate": 1e-05, "loss": 0.7644, "step": 3680 }, { "epoch": 0.11, "learning_rate": 1e-05, "loss": 0.7709, "step": 3690 }, { "epoch": 0.11, "learning_rate": 1e-05, "loss": 0.7986, "step": 3700 }, { "epoch": 0.11, "learning_rate": 1e-05, "loss": 0.7853, "step": 3710 }, { "epoch": 0.11, "learning_rate": 1e-05, "loss": 0.7782, "step": 3720 }, { "epoch": 0.11, "learning_rate": 1e-05, "loss": 0.747, "step": 3730 }, { "epoch": 0.11, "learning_rate": 1e-05, "loss": 0.7427, "step": 3740 }, { "epoch": 0.11, "learning_rate": 1e-05, "loss": 0.7813, "step": 3750 }, { "epoch": 0.11, "learning_rate": 1e-05, "loss": 0.7691, "step": 3760 }, { "epoch": 0.11, "learning_rate": 1e-05, "loss": 0.7912, "step": 3770 }, { "epoch": 0.11, "learning_rate": 1e-05, "loss": 0.7849, "step": 3780 }, { "epoch": 0.11, "learning_rate": 1e-05, "loss": 0.7848, "step": 3790 }, { "epoch": 0.11, "learning_rate": 1e-05, "loss": 0.7946, "step": 3800 }, { "epoch": 0.11, "learning_rate": 1e-05, "loss": 0.7852, "step": 3810 }, { "epoch": 0.11, "learning_rate": 1e-05, "loss": 0.7828, "step": 3820 }, { "epoch": 0.11, "learning_rate": 1e-05, "loss": 0.7714, "step": 3830 }, { "epoch": 0.11, "learning_rate": 1e-05, "loss": 0.8218, "step": 3840 }, { "epoch": 0.11, "learning_rate": 1e-05, "loss": 0.8017, "step": 3850 }, { "epoch": 0.11, "learning_rate": 1e-05, "loss": 0.7707, "step": 3860 }, { "epoch": 0.12, "learning_rate": 1e-05, "loss": 0.7678, "step": 3870 }, { "epoch": 0.12, "learning_rate": 1e-05, "loss": 0.7662, "step": 3880 }, { "epoch": 0.12, "learning_rate": 1e-05, "loss": 0.7681, "step": 3890 }, { "epoch": 0.12, "learning_rate": 1e-05, "loss": 0.7641, "step": 3900 }, { "epoch": 0.12, "learning_rate": 1e-05, "loss": 0.7553, "step": 3910 }, { "epoch": 0.12, "learning_rate": 1e-05, "loss": 0.7591, "step": 3920 }, { "epoch": 0.12, "learning_rate": 1e-05, "loss": 0.7863, "step": 3930 }, { "epoch": 0.12, "learning_rate": 1e-05, "loss": 0.7862, "step": 3940 }, { "epoch": 0.12, "learning_rate": 1e-05, "loss": 0.7807, "step": 3950 }, { "epoch": 0.12, "learning_rate": 1e-05, "loss": 0.777, "step": 3960 }, { "epoch": 0.12, "learning_rate": 1e-05, "loss": 0.7829, "step": 3970 }, { "epoch": 0.12, "learning_rate": 1e-05, "loss": 0.7952, "step": 3980 }, { "epoch": 0.12, "learning_rate": 1e-05, "loss": 0.7829, "step": 3990 }, { "epoch": 0.12, "learning_rate": 1e-05, "loss": 0.7675, "step": 4000 } ], "max_steps": 33579, "num_train_epochs": 1, "total_flos": 8.143508134258803e+18, "trial_name": null, "trial_params": null }