{ "best_metric": 0.13444018363952637, "best_model_checkpoint": "/leonardo_work/IscrC_AGENT/PROFES2025/results/full_size/oss_similar/checkpoint-11908", "epoch": 8.0, "eval_steps": 500, "global_step": 95264, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 1.9998320456835744e-05, "loss": 0.3734, "step": 10 }, { "epoch": 0.0, "learning_rate": 1.9996640913671483e-05, "loss": 0.2665, "step": 20 }, { "epoch": 0.0, "learning_rate": 1.9994961370507225e-05, "loss": 0.2582, "step": 30 }, { "epoch": 0.0, "learning_rate": 1.9993281827342964e-05, "loss": 0.1689, "step": 40 }, { "epoch": 0.0, "learning_rate": 1.9991602284178706e-05, "loss": 0.3073, "step": 50 }, { "epoch": 0.01, "learning_rate": 1.9989922741014444e-05, "loss": 0.2732, "step": 60 }, { "epoch": 0.01, "learning_rate": 1.9988243197850187e-05, "loss": 0.2593, "step": 70 }, { "epoch": 0.01, "learning_rate": 1.9986563654685925e-05, "loss": 0.1632, "step": 80 }, { "epoch": 0.01, "learning_rate": 1.9984884111521668e-05, "loss": 0.2904, "step": 90 }, { "epoch": 0.01, "learning_rate": 1.998320456835741e-05, "loss": 0.1791, "step": 100 }, { "epoch": 0.01, "learning_rate": 1.998152502519315e-05, "loss": 0.3151, "step": 110 }, { "epoch": 0.01, "learning_rate": 1.997984548202889e-05, "loss": 0.3131, "step": 120 }, { "epoch": 0.01, "learning_rate": 1.997816593886463e-05, "loss": 0.1948, "step": 130 }, { "epoch": 0.01, "learning_rate": 1.9976486395700372e-05, "loss": 0.2393, "step": 140 }, { "epoch": 0.01, "learning_rate": 1.9974806852536114e-05, "loss": 0.2771, "step": 150 }, { "epoch": 0.01, "learning_rate": 1.9973127309371853e-05, "loss": 0.1512, "step": 160 }, { "epoch": 0.01, "learning_rate": 1.997144776620759e-05, "loss": 0.2116, "step": 170 }, { "epoch": 0.02, "learning_rate": 1.9969768223043334e-05, "loss": 0.2142, "step": 180 }, { "epoch": 0.02, "learning_rate": 1.9968088679879072e-05, "loss": 0.1224, "step": 190 }, { "epoch": 0.02, "learning_rate": 1.9966409136714815e-05, "loss": 0.1952, "step": 200 }, { "epoch": 0.02, "learning_rate": 1.9964729593550557e-05, "loss": 0.188, "step": 210 }, { "epoch": 0.02, "learning_rate": 1.9963050050386296e-05, "loss": 0.261, "step": 220 }, { "epoch": 0.02, "learning_rate": 1.9961370507222038e-05, "loss": 0.1795, "step": 230 }, { "epoch": 0.02, "learning_rate": 1.995969096405778e-05, "loss": 0.1414, "step": 240 }, { "epoch": 0.02, "learning_rate": 1.995801142089352e-05, "loss": 0.2058, "step": 250 }, { "epoch": 0.02, "learning_rate": 1.995633187772926e-05, "loss": 0.1295, "step": 260 }, { "epoch": 0.02, "learning_rate": 1.9954652334565e-05, "loss": 0.1755, "step": 270 }, { "epoch": 0.02, "learning_rate": 1.995297279140074e-05, "loss": 0.2155, "step": 280 }, { "epoch": 0.02, "learning_rate": 1.995129324823648e-05, "loss": 0.2709, "step": 290 }, { "epoch": 0.03, "learning_rate": 1.9949613705072223e-05, "loss": 0.2396, "step": 300 }, { "epoch": 0.03, "learning_rate": 1.994793416190796e-05, "loss": 0.2168, "step": 310 }, { "epoch": 0.03, "learning_rate": 1.9946254618743704e-05, "loss": 0.2152, "step": 320 }, { "epoch": 0.03, "learning_rate": 1.9944575075579446e-05, "loss": 0.2622, "step": 330 }, { "epoch": 0.03, "learning_rate": 1.9942895532415185e-05, "loss": 0.184, "step": 340 }, { "epoch": 0.03, "learning_rate": 1.9941215989250927e-05, "loss": 0.204, "step": 350 }, { "epoch": 0.03, "learning_rate": 1.9939536446086666e-05, "loss": 0.1393, "step": 360 }, { "epoch": 0.03, "learning_rate": 1.9937856902922408e-05, "loss": 0.2023, "step": 370 }, { "epoch": 0.03, "learning_rate": 1.9936177359758147e-05, "loss": 0.1735, "step": 380 }, { "epoch": 0.03, "learning_rate": 1.993449781659389e-05, "loss": 0.261, "step": 390 }, { "epoch": 0.03, "learning_rate": 1.9932818273429627e-05, "loss": 0.1819, "step": 400 }, { "epoch": 0.03, "learning_rate": 1.993113873026537e-05, "loss": 0.1835, "step": 410 }, { "epoch": 0.04, "learning_rate": 1.992945918710111e-05, "loss": 0.1788, "step": 420 }, { "epoch": 0.04, "learning_rate": 1.992777964393685e-05, "loss": 0.1627, "step": 430 }, { "epoch": 0.04, "learning_rate": 1.9926100100772593e-05, "loss": 0.19, "step": 440 }, { "epoch": 0.04, "learning_rate": 1.992442055760833e-05, "loss": 0.1066, "step": 450 }, { "epoch": 0.04, "learning_rate": 1.9922741014444074e-05, "loss": 0.3286, "step": 460 }, { "epoch": 0.04, "learning_rate": 1.9921061471279816e-05, "loss": 0.1822, "step": 470 }, { "epoch": 0.04, "learning_rate": 1.9919381928115555e-05, "loss": 0.2271, "step": 480 }, { "epoch": 0.04, "learning_rate": 1.9917702384951293e-05, "loss": 0.2205, "step": 490 }, { "epoch": 0.04, "learning_rate": 1.9916022841787036e-05, "loss": 0.1113, "step": 500 }, { "epoch": 0.04, "learning_rate": 1.9914343298622774e-05, "loss": 0.1813, "step": 510 }, { "epoch": 0.04, "learning_rate": 1.9912663755458517e-05, "loss": 0.2134, "step": 520 }, { "epoch": 0.04, "learning_rate": 1.991098421229426e-05, "loss": 0.1823, "step": 530 }, { "epoch": 0.05, "learning_rate": 1.9909304669129998e-05, "loss": 0.209, "step": 540 }, { "epoch": 0.05, "learning_rate": 1.990762512596574e-05, "loss": 0.1729, "step": 550 }, { "epoch": 0.05, "learning_rate": 1.9905945582801482e-05, "loss": 0.1367, "step": 560 }, { "epoch": 0.05, "learning_rate": 1.990426603963722e-05, "loss": 0.249, "step": 570 }, { "epoch": 0.05, "learning_rate": 1.9902586496472963e-05, "loss": 0.2173, "step": 580 }, { "epoch": 0.05, "learning_rate": 1.99009069533087e-05, "loss": 0.1913, "step": 590 }, { "epoch": 0.05, "learning_rate": 1.989922741014444e-05, "loss": 0.187, "step": 600 }, { "epoch": 0.05, "learning_rate": 1.9897547866980183e-05, "loss": 0.1559, "step": 610 }, { "epoch": 0.05, "learning_rate": 1.9895868323815925e-05, "loss": 0.3267, "step": 620 }, { "epoch": 0.05, "learning_rate": 1.9894188780651664e-05, "loss": 0.0811, "step": 630 }, { "epoch": 0.05, "learning_rate": 1.9892509237487406e-05, "loss": 0.243, "step": 640 }, { "epoch": 0.05, "learning_rate": 1.9890829694323144e-05, "loss": 0.1449, "step": 650 }, { "epoch": 0.06, "learning_rate": 1.9889150151158887e-05, "loss": 0.1905, "step": 660 }, { "epoch": 0.06, "learning_rate": 1.988747060799463e-05, "loss": 0.1523, "step": 670 }, { "epoch": 0.06, "learning_rate": 1.9885791064830368e-05, "loss": 0.303, "step": 680 }, { "epoch": 0.06, "learning_rate": 1.9884111521666106e-05, "loss": 0.1104, "step": 690 }, { "epoch": 0.06, "learning_rate": 1.988243197850185e-05, "loss": 0.2114, "step": 700 }, { "epoch": 0.06, "learning_rate": 1.988075243533759e-05, "loss": 0.1759, "step": 710 }, { "epoch": 0.06, "learning_rate": 1.987907289217333e-05, "loss": 0.1708, "step": 720 }, { "epoch": 0.06, "learning_rate": 1.987739334900907e-05, "loss": 0.1793, "step": 730 }, { "epoch": 0.06, "learning_rate": 1.987571380584481e-05, "loss": 0.1645, "step": 740 }, { "epoch": 0.06, "learning_rate": 1.9874034262680553e-05, "loss": 0.1704, "step": 750 }, { "epoch": 0.06, "learning_rate": 1.9872354719516295e-05, "loss": 0.1364, "step": 760 }, { "epoch": 0.06, "learning_rate": 1.9870675176352034e-05, "loss": 0.2427, "step": 770 }, { "epoch": 0.07, "learning_rate": 1.9868995633187776e-05, "loss": 0.2091, "step": 780 }, { "epoch": 0.07, "learning_rate": 1.9867316090023518e-05, "loss": 0.1954, "step": 790 }, { "epoch": 0.07, "learning_rate": 1.9865636546859253e-05, "loss": 0.1722, "step": 800 }, { "epoch": 0.07, "learning_rate": 1.9863957003694995e-05, "loss": 0.2068, "step": 810 }, { "epoch": 0.07, "learning_rate": 1.9862277460530738e-05, "loss": 0.151, "step": 820 }, { "epoch": 0.07, "learning_rate": 1.9860597917366476e-05, "loss": 0.1477, "step": 830 }, { "epoch": 0.07, "learning_rate": 1.985891837420222e-05, "loss": 0.3199, "step": 840 }, { "epoch": 0.07, "learning_rate": 1.985723883103796e-05, "loss": 0.1202, "step": 850 }, { "epoch": 0.07, "learning_rate": 1.98555592878737e-05, "loss": 0.1616, "step": 860 }, { "epoch": 0.07, "learning_rate": 1.9853879744709442e-05, "loss": 0.1548, "step": 870 }, { "epoch": 0.07, "learning_rate": 1.985220020154518e-05, "loss": 0.2007, "step": 880 }, { "epoch": 0.07, "learning_rate": 1.9850520658380923e-05, "loss": 0.2075, "step": 890 }, { "epoch": 0.08, "learning_rate": 1.9848841115216665e-05, "loss": 0.1345, "step": 900 }, { "epoch": 0.08, "learning_rate": 1.9847161572052404e-05, "loss": 0.2184, "step": 910 }, { "epoch": 0.08, "learning_rate": 1.9845482028888142e-05, "loss": 0.2463, "step": 920 }, { "epoch": 0.08, "learning_rate": 1.9843802485723885e-05, "loss": 0.1605, "step": 930 }, { "epoch": 0.08, "learning_rate": 1.9842122942559627e-05, "loss": 0.2139, "step": 940 }, { "epoch": 0.08, "learning_rate": 1.9840443399395366e-05, "loss": 0.1903, "step": 950 }, { "epoch": 0.08, "learning_rate": 1.9838763856231108e-05, "loss": 0.1684, "step": 960 }, { "epoch": 0.08, "learning_rate": 1.9837084313066847e-05, "loss": 0.2399, "step": 970 }, { "epoch": 0.08, "learning_rate": 1.983540476990259e-05, "loss": 0.2199, "step": 980 }, { "epoch": 0.08, "learning_rate": 1.983372522673833e-05, "loss": 0.1969, "step": 990 }, { "epoch": 0.08, "learning_rate": 1.983204568357407e-05, "loss": 0.1202, "step": 1000 }, { "epoch": 0.08, "learning_rate": 1.983036614040981e-05, "loss": 0.243, "step": 1010 }, { "epoch": 0.09, "learning_rate": 1.982868659724555e-05, "loss": 0.123, "step": 1020 }, { "epoch": 0.09, "learning_rate": 1.982700705408129e-05, "loss": 0.138, "step": 1030 }, { "epoch": 0.09, "learning_rate": 1.982532751091703e-05, "loss": 0.179, "step": 1040 }, { "epoch": 0.09, "learning_rate": 1.9823647967752774e-05, "loss": 0.2395, "step": 1050 }, { "epoch": 0.09, "learning_rate": 1.9821968424588512e-05, "loss": 0.141, "step": 1060 }, { "epoch": 0.09, "learning_rate": 1.9820288881424255e-05, "loss": 0.1523, "step": 1070 }, { "epoch": 0.09, "learning_rate": 1.9818609338259997e-05, "loss": 0.1223, "step": 1080 }, { "epoch": 0.09, "learning_rate": 1.9816929795095736e-05, "loss": 0.2265, "step": 1090 }, { "epoch": 0.09, "learning_rate": 1.9815250251931478e-05, "loss": 0.1166, "step": 1100 }, { "epoch": 0.09, "learning_rate": 1.9813570708767217e-05, "loss": 0.1658, "step": 1110 }, { "epoch": 0.09, "learning_rate": 1.9811891165602955e-05, "loss": 0.1816, "step": 1120 }, { "epoch": 0.09, "learning_rate": 1.9810211622438698e-05, "loss": 0.2481, "step": 1130 }, { "epoch": 0.1, "learning_rate": 1.980853207927444e-05, "loss": 0.3276, "step": 1140 }, { "epoch": 0.1, "learning_rate": 1.980685253611018e-05, "loss": 0.1324, "step": 1150 }, { "epoch": 0.1, "learning_rate": 1.980517299294592e-05, "loss": 0.17, "step": 1160 }, { "epoch": 0.1, "learning_rate": 1.980349344978166e-05, "loss": 0.2759, "step": 1170 }, { "epoch": 0.1, "learning_rate": 1.98018139066174e-05, "loss": 0.1523, "step": 1180 }, { "epoch": 0.1, "learning_rate": 1.9800134363453144e-05, "loss": 0.1774, "step": 1190 }, { "epoch": 0.1, "learning_rate": 1.9798454820288883e-05, "loss": 0.2931, "step": 1200 }, { "epoch": 0.1, "learning_rate": 1.9796775277124625e-05, "loss": 0.1394, "step": 1210 }, { "epoch": 0.1, "learning_rate": 1.9795095733960364e-05, "loss": 0.2006, "step": 1220 }, { "epoch": 0.1, "learning_rate": 1.9793416190796106e-05, "loss": 0.18, "step": 1230 }, { "epoch": 0.1, "learning_rate": 1.9791736647631844e-05, "loss": 0.112, "step": 1240 }, { "epoch": 0.1, "learning_rate": 1.9790057104467587e-05, "loss": 0.186, "step": 1250 }, { "epoch": 0.11, "learning_rate": 1.9788377561303325e-05, "loss": 0.1522, "step": 1260 }, { "epoch": 0.11, "learning_rate": 1.9786698018139068e-05, "loss": 0.1927, "step": 1270 }, { "epoch": 0.11, "learning_rate": 1.978501847497481e-05, "loss": 0.1536, "step": 1280 }, { "epoch": 0.11, "learning_rate": 1.978333893181055e-05, "loss": 0.1809, "step": 1290 }, { "epoch": 0.11, "learning_rate": 1.978165938864629e-05, "loss": 0.2122, "step": 1300 }, { "epoch": 0.11, "learning_rate": 1.9779979845482033e-05, "loss": 0.1101, "step": 1310 }, { "epoch": 0.11, "learning_rate": 1.977830030231777e-05, "loss": 0.1545, "step": 1320 }, { "epoch": 0.11, "learning_rate": 1.977662075915351e-05, "loss": 0.2584, "step": 1330 }, { "epoch": 0.11, "learning_rate": 1.9774941215989253e-05, "loss": 0.1062, "step": 1340 }, { "epoch": 0.11, "learning_rate": 1.977326167282499e-05, "loss": 0.1944, "step": 1350 }, { "epoch": 0.11, "learning_rate": 1.9771582129660734e-05, "loss": 0.1994, "step": 1360 }, { "epoch": 0.12, "learning_rate": 1.9769902586496476e-05, "loss": 0.1675, "step": 1370 }, { "epoch": 0.12, "learning_rate": 1.9768223043332215e-05, "loss": 0.2304, "step": 1380 }, { "epoch": 0.12, "learning_rate": 1.9766543500167957e-05, "loss": 0.1611, "step": 1390 }, { "epoch": 0.12, "learning_rate": 1.9764863957003695e-05, "loss": 0.1469, "step": 1400 }, { "epoch": 0.12, "learning_rate": 1.9763184413839438e-05, "loss": 0.1634, "step": 1410 }, { "epoch": 0.12, "learning_rate": 1.976150487067518e-05, "loss": 0.2208, "step": 1420 }, { "epoch": 0.12, "learning_rate": 1.975982532751092e-05, "loss": 0.1341, "step": 1430 }, { "epoch": 0.12, "learning_rate": 1.9758145784346657e-05, "loss": 0.1998, "step": 1440 }, { "epoch": 0.12, "learning_rate": 1.97564662411824e-05, "loss": 0.1436, "step": 1450 }, { "epoch": 0.12, "learning_rate": 1.9754786698018142e-05, "loss": 0.1778, "step": 1460 }, { "epoch": 0.12, "learning_rate": 1.975310715485388e-05, "loss": 0.1525, "step": 1470 }, { "epoch": 0.12, "learning_rate": 1.9751427611689623e-05, "loss": 0.1904, "step": 1480 }, { "epoch": 0.13, "learning_rate": 1.974974806852536e-05, "loss": 0.156, "step": 1490 }, { "epoch": 0.13, "learning_rate": 1.9748068525361104e-05, "loss": 0.1413, "step": 1500 }, { "epoch": 0.13, "learning_rate": 1.9746388982196846e-05, "loss": 0.183, "step": 1510 }, { "epoch": 0.13, "learning_rate": 1.9744709439032585e-05, "loss": 0.1373, "step": 1520 }, { "epoch": 0.13, "learning_rate": 1.9743029895868327e-05, "loss": 0.1643, "step": 1530 }, { "epoch": 0.13, "learning_rate": 1.9741350352704066e-05, "loss": 0.1705, "step": 1540 }, { "epoch": 0.13, "learning_rate": 1.9739670809539804e-05, "loss": 0.1305, "step": 1550 }, { "epoch": 0.13, "learning_rate": 1.9737991266375546e-05, "loss": 0.1829, "step": 1560 }, { "epoch": 0.13, "learning_rate": 1.973631172321129e-05, "loss": 0.2181, "step": 1570 }, { "epoch": 0.13, "learning_rate": 1.9734632180047027e-05, "loss": 0.1781, "step": 1580 }, { "epoch": 0.13, "learning_rate": 1.973295263688277e-05, "loss": 0.1571, "step": 1590 }, { "epoch": 0.13, "learning_rate": 1.9731273093718512e-05, "loss": 0.1141, "step": 1600 }, { "epoch": 0.14, "learning_rate": 1.972959355055425e-05, "loss": 0.1882, "step": 1610 }, { "epoch": 0.14, "learning_rate": 1.9727914007389993e-05, "loss": 0.2541, "step": 1620 }, { "epoch": 0.14, "learning_rate": 1.972623446422573e-05, "loss": 0.1052, "step": 1630 }, { "epoch": 0.14, "learning_rate": 1.9724554921061474e-05, "loss": 0.1317, "step": 1640 }, { "epoch": 0.14, "learning_rate": 1.9722875377897212e-05, "loss": 0.3064, "step": 1650 }, { "epoch": 0.14, "learning_rate": 1.9721195834732955e-05, "loss": 0.0923, "step": 1660 }, { "epoch": 0.14, "learning_rate": 1.9719516291568693e-05, "loss": 0.1202, "step": 1670 }, { "epoch": 0.14, "learning_rate": 1.9717836748404436e-05, "loss": 0.1853, "step": 1680 }, { "epoch": 0.14, "learning_rate": 1.9716157205240178e-05, "loss": 0.1054, "step": 1690 }, { "epoch": 0.14, "learning_rate": 1.9714477662075917e-05, "loss": 0.1348, "step": 1700 }, { "epoch": 0.14, "learning_rate": 1.971279811891166e-05, "loss": 0.2964, "step": 1710 }, { "epoch": 0.14, "learning_rate": 1.9711118575747398e-05, "loss": 0.1633, "step": 1720 }, { "epoch": 0.15, "learning_rate": 1.970943903258314e-05, "loss": 0.1817, "step": 1730 }, { "epoch": 0.15, "learning_rate": 1.9707759489418882e-05, "loss": 0.1266, "step": 1740 }, { "epoch": 0.15, "learning_rate": 1.970607994625462e-05, "loss": 0.161, "step": 1750 }, { "epoch": 0.15, "learning_rate": 1.970440040309036e-05, "loss": 0.1968, "step": 1760 }, { "epoch": 0.15, "learning_rate": 1.97027208599261e-05, "loss": 0.2023, "step": 1770 }, { "epoch": 0.15, "learning_rate": 1.970104131676184e-05, "loss": 0.1436, "step": 1780 }, { "epoch": 0.15, "learning_rate": 1.9699361773597583e-05, "loss": 0.1899, "step": 1790 }, { "epoch": 0.15, "learning_rate": 1.9697682230433325e-05, "loss": 0.1945, "step": 1800 }, { "epoch": 0.15, "learning_rate": 1.9696002687269063e-05, "loss": 0.2061, "step": 1810 }, { "epoch": 0.15, "learning_rate": 1.9694323144104806e-05, "loss": 0.1195, "step": 1820 }, { "epoch": 0.15, "learning_rate": 1.9692643600940548e-05, "loss": 0.1665, "step": 1830 }, { "epoch": 0.15, "learning_rate": 1.9690964057776287e-05, "loss": 0.236, "step": 1840 }, { "epoch": 0.16, "learning_rate": 1.968928451461203e-05, "loss": 0.1635, "step": 1850 }, { "epoch": 0.16, "learning_rate": 1.9687604971447768e-05, "loss": 0.236, "step": 1860 }, { "epoch": 0.16, "learning_rate": 1.9685925428283506e-05, "loss": 0.0859, "step": 1870 }, { "epoch": 0.16, "learning_rate": 1.968424588511925e-05, "loss": 0.1814, "step": 1880 }, { "epoch": 0.16, "learning_rate": 1.968256634195499e-05, "loss": 0.2057, "step": 1890 }, { "epoch": 0.16, "learning_rate": 1.968088679879073e-05, "loss": 0.1778, "step": 1900 }, { "epoch": 0.16, "learning_rate": 1.967920725562647e-05, "loss": 0.1789, "step": 1910 }, { "epoch": 0.16, "learning_rate": 1.9677527712462214e-05, "loss": 0.2285, "step": 1920 }, { "epoch": 0.16, "learning_rate": 1.9675848169297953e-05, "loss": 0.2492, "step": 1930 }, { "epoch": 0.16, "learning_rate": 1.9674168626133695e-05, "loss": 0.1555, "step": 1940 }, { "epoch": 0.16, "learning_rate": 1.9672489082969434e-05, "loss": 0.0866, "step": 1950 }, { "epoch": 0.16, "learning_rate": 1.9670809539805172e-05, "loss": 0.159, "step": 1960 }, { "epoch": 0.17, "learning_rate": 1.9669129996640915e-05, "loss": 0.2057, "step": 1970 }, { "epoch": 0.17, "learning_rate": 1.9667450453476657e-05, "loss": 0.245, "step": 1980 }, { "epoch": 0.17, "learning_rate": 1.9665770910312395e-05, "loss": 0.2571, "step": 1990 }, { "epoch": 0.17, "learning_rate": 1.9664091367148138e-05, "loss": 0.2657, "step": 2000 }, { "epoch": 0.17, "learning_rate": 1.9662411823983876e-05, "loss": 0.1311, "step": 2010 }, { "epoch": 0.17, "learning_rate": 1.966073228081962e-05, "loss": 0.21, "step": 2020 }, { "epoch": 0.17, "learning_rate": 1.965905273765536e-05, "loss": 0.258, "step": 2030 }, { "epoch": 0.17, "learning_rate": 1.96573731944911e-05, "loss": 0.1474, "step": 2040 }, { "epoch": 0.17, "learning_rate": 1.9655693651326842e-05, "loss": 0.1555, "step": 2050 }, { "epoch": 0.17, "learning_rate": 1.9654014108162584e-05, "loss": 0.2286, "step": 2060 }, { "epoch": 0.17, "learning_rate": 1.9652334564998323e-05, "loss": 0.1315, "step": 2070 }, { "epoch": 0.17, "learning_rate": 1.965065502183406e-05, "loss": 0.1436, "step": 2080 }, { "epoch": 0.18, "learning_rate": 1.9648975478669804e-05, "loss": 0.1569, "step": 2090 }, { "epoch": 0.18, "learning_rate": 1.9647295935505542e-05, "loss": 0.1457, "step": 2100 }, { "epoch": 0.18, "learning_rate": 1.9645616392341285e-05, "loss": 0.236, "step": 2110 }, { "epoch": 0.18, "learning_rate": 1.9643936849177027e-05, "loss": 0.1421, "step": 2120 }, { "epoch": 0.18, "learning_rate": 1.9642257306012766e-05, "loss": 0.1358, "step": 2130 }, { "epoch": 0.18, "learning_rate": 1.9640577762848508e-05, "loss": 0.1692, "step": 2140 }, { "epoch": 0.18, "learning_rate": 1.9638898219684246e-05, "loss": 0.159, "step": 2150 }, { "epoch": 0.18, "learning_rate": 1.963721867651999e-05, "loss": 0.1674, "step": 2160 }, { "epoch": 0.18, "learning_rate": 1.963553913335573e-05, "loss": 0.1151, "step": 2170 }, { "epoch": 0.18, "learning_rate": 1.963385959019147e-05, "loss": 0.2403, "step": 2180 }, { "epoch": 0.18, "learning_rate": 1.963218004702721e-05, "loss": 0.1458, "step": 2190 }, { "epoch": 0.18, "learning_rate": 1.963050050386295e-05, "loss": 0.1489, "step": 2200 }, { "epoch": 0.19, "learning_rate": 1.9628820960698693e-05, "loss": 0.1525, "step": 2210 }, { "epoch": 0.19, "learning_rate": 1.962714141753443e-05, "loss": 0.1696, "step": 2220 }, { "epoch": 0.19, "learning_rate": 1.9625461874370174e-05, "loss": 0.2279, "step": 2230 }, { "epoch": 0.19, "learning_rate": 1.9623782331205912e-05, "loss": 0.0859, "step": 2240 }, { "epoch": 0.19, "learning_rate": 1.9622102788041655e-05, "loss": 0.1106, "step": 2250 }, { "epoch": 0.19, "learning_rate": 1.9620423244877397e-05, "loss": 0.1516, "step": 2260 }, { "epoch": 0.19, "learning_rate": 1.9618743701713136e-05, "loss": 0.1377, "step": 2270 }, { "epoch": 0.19, "learning_rate": 1.9617064158548874e-05, "loss": 0.178, "step": 2280 }, { "epoch": 0.19, "learning_rate": 1.9615384615384617e-05, "loss": 0.1374, "step": 2290 }, { "epoch": 0.19, "learning_rate": 1.961370507222036e-05, "loss": 0.1643, "step": 2300 }, { "epoch": 0.19, "learning_rate": 1.9612025529056097e-05, "loss": 0.1557, "step": 2310 }, { "epoch": 0.19, "learning_rate": 1.961034598589184e-05, "loss": 0.16, "step": 2320 }, { "epoch": 0.2, "learning_rate": 1.960866644272758e-05, "loss": 0.1945, "step": 2330 }, { "epoch": 0.2, "learning_rate": 1.960698689956332e-05, "loss": 0.1278, "step": 2340 }, { "epoch": 0.2, "learning_rate": 1.9605307356399063e-05, "loss": 0.2002, "step": 2350 }, { "epoch": 0.2, "learning_rate": 1.96036278132348e-05, "loss": 0.1604, "step": 2360 }, { "epoch": 0.2, "learning_rate": 1.9601948270070544e-05, "loss": 0.1945, "step": 2370 }, { "epoch": 0.2, "learning_rate": 1.9600268726906283e-05, "loss": 0.1695, "step": 2380 }, { "epoch": 0.2, "learning_rate": 1.959858918374202e-05, "loss": 0.1357, "step": 2390 }, { "epoch": 0.2, "learning_rate": 1.9596909640577763e-05, "loss": 0.1322, "step": 2400 }, { "epoch": 0.2, "learning_rate": 1.9595230097413506e-05, "loss": 0.2783, "step": 2410 }, { "epoch": 0.2, "learning_rate": 1.9593550554249244e-05, "loss": 0.2085, "step": 2420 }, { "epoch": 0.2, "learning_rate": 1.9591871011084987e-05, "loss": 0.1341, "step": 2430 }, { "epoch": 0.2, "learning_rate": 1.959019146792073e-05, "loss": 0.1644, "step": 2440 }, { "epoch": 0.21, "learning_rate": 1.9588511924756468e-05, "loss": 0.1838, "step": 2450 }, { "epoch": 0.21, "learning_rate": 1.958683238159221e-05, "loss": 0.1259, "step": 2460 }, { "epoch": 0.21, "learning_rate": 1.958515283842795e-05, "loss": 0.115, "step": 2470 }, { "epoch": 0.21, "learning_rate": 1.958347329526369e-05, "loss": 0.1535, "step": 2480 }, { "epoch": 0.21, "learning_rate": 1.958179375209943e-05, "loss": 0.11, "step": 2490 }, { "epoch": 0.21, "learning_rate": 1.958011420893517e-05, "loss": 0.0898, "step": 2500 }, { "epoch": 0.21, "learning_rate": 1.957843466577091e-05, "loss": 0.2251, "step": 2510 }, { "epoch": 0.21, "learning_rate": 1.9576755122606653e-05, "loss": 0.186, "step": 2520 }, { "epoch": 0.21, "learning_rate": 1.957507557944239e-05, "loss": 0.1353, "step": 2530 }, { "epoch": 0.21, "learning_rate": 1.9573396036278134e-05, "loss": 0.1397, "step": 2540 }, { "epoch": 0.21, "learning_rate": 1.9571716493113876e-05, "loss": 0.079, "step": 2550 }, { "epoch": 0.21, "learning_rate": 1.9570036949949614e-05, "loss": 0.3429, "step": 2560 }, { "epoch": 0.22, "learning_rate": 1.9568357406785357e-05, "loss": 0.2005, "step": 2570 }, { "epoch": 0.22, "learning_rate": 1.95666778636211e-05, "loss": 0.1655, "step": 2580 }, { "epoch": 0.22, "learning_rate": 1.9564998320456838e-05, "loss": 0.1385, "step": 2590 }, { "epoch": 0.22, "learning_rate": 1.9563318777292576e-05, "loss": 0.2557, "step": 2600 }, { "epoch": 0.22, "learning_rate": 1.956163923412832e-05, "loss": 0.153, "step": 2610 }, { "epoch": 0.22, "learning_rate": 1.9559959690964057e-05, "loss": 0.1799, "step": 2620 }, { "epoch": 0.22, "learning_rate": 1.95582801477998e-05, "loss": 0.1576, "step": 2630 }, { "epoch": 0.22, "learning_rate": 1.955660060463554e-05, "loss": 0.1594, "step": 2640 }, { "epoch": 0.22, "learning_rate": 1.955492106147128e-05, "loss": 0.1686, "step": 2650 }, { "epoch": 0.22, "learning_rate": 1.9553241518307023e-05, "loss": 0.1473, "step": 2660 }, { "epoch": 0.22, "learning_rate": 1.9551561975142765e-05, "loss": 0.1511, "step": 2670 }, { "epoch": 0.23, "learning_rate": 1.9549882431978504e-05, "loss": 0.1451, "step": 2680 }, { "epoch": 0.23, "learning_rate": 1.9548202888814246e-05, "loss": 0.108, "step": 2690 }, { "epoch": 0.23, "learning_rate": 1.9546523345649985e-05, "loss": 0.1496, "step": 2700 }, { "epoch": 0.23, "learning_rate": 1.9544843802485723e-05, "loss": 0.111, "step": 2710 }, { "epoch": 0.23, "learning_rate": 1.9543164259321466e-05, "loss": 0.2093, "step": 2720 }, { "epoch": 0.23, "learning_rate": 1.9541484716157208e-05, "loss": 0.1807, "step": 2730 }, { "epoch": 0.23, "learning_rate": 1.9539805172992946e-05, "loss": 0.1566, "step": 2740 }, { "epoch": 0.23, "learning_rate": 1.953812562982869e-05, "loss": 0.1315, "step": 2750 }, { "epoch": 0.23, "learning_rate": 1.9536446086664427e-05, "loss": 0.1024, "step": 2760 }, { "epoch": 0.23, "learning_rate": 1.953476654350017e-05, "loss": 0.1114, "step": 2770 }, { "epoch": 0.23, "learning_rate": 1.9533087000335912e-05, "loss": 0.1693, "step": 2780 }, { "epoch": 0.23, "learning_rate": 1.953140745717165e-05, "loss": 0.1974, "step": 2790 }, { "epoch": 0.24, "learning_rate": 1.9529727914007393e-05, "loss": 0.1529, "step": 2800 }, { "epoch": 0.24, "learning_rate": 1.952804837084313e-05, "loss": 0.1497, "step": 2810 }, { "epoch": 0.24, "learning_rate": 1.9526368827678874e-05, "loss": 0.1457, "step": 2820 }, { "epoch": 0.24, "learning_rate": 1.9524689284514612e-05, "loss": 0.1661, "step": 2830 }, { "epoch": 0.24, "learning_rate": 1.9523009741350355e-05, "loss": 0.2202, "step": 2840 }, { "epoch": 0.24, "learning_rate": 1.9521330198186093e-05, "loss": 0.1602, "step": 2850 }, { "epoch": 0.24, "learning_rate": 1.9519650655021836e-05, "loss": 0.2567, "step": 2860 }, { "epoch": 0.24, "learning_rate": 1.9517971111857578e-05, "loss": 0.1082, "step": 2870 }, { "epoch": 0.24, "learning_rate": 1.9516291568693317e-05, "loss": 0.2053, "step": 2880 }, { "epoch": 0.24, "learning_rate": 1.951461202552906e-05, "loss": 0.1459, "step": 2890 }, { "epoch": 0.24, "learning_rate": 1.95129324823648e-05, "loss": 0.0999, "step": 2900 }, { "epoch": 0.24, "learning_rate": 1.951125293920054e-05, "loss": 0.1736, "step": 2910 }, { "epoch": 0.25, "learning_rate": 1.950957339603628e-05, "loss": 0.1457, "step": 2920 }, { "epoch": 0.25, "learning_rate": 1.950789385287202e-05, "loss": 0.1317, "step": 2930 }, { "epoch": 0.25, "learning_rate": 1.950621430970776e-05, "loss": 0.2469, "step": 2940 }, { "epoch": 0.25, "learning_rate": 1.95045347665435e-05, "loss": 0.1149, "step": 2950 }, { "epoch": 0.25, "learning_rate": 1.9502855223379244e-05, "loss": 0.1672, "step": 2960 }, { "epoch": 0.25, "learning_rate": 1.9501175680214982e-05, "loss": 0.1466, "step": 2970 }, { "epoch": 0.25, "learning_rate": 1.9499496137050725e-05, "loss": 0.1679, "step": 2980 }, { "epoch": 0.25, "learning_rate": 1.9497816593886463e-05, "loss": 0.2015, "step": 2990 }, { "epoch": 0.25, "learning_rate": 1.9496137050722206e-05, "loss": 0.1547, "step": 3000 }, { "epoch": 0.25, "learning_rate": 1.9494457507557948e-05, "loss": 0.1242, "step": 3010 }, { "epoch": 0.25, "learning_rate": 1.9492777964393687e-05, "loss": 0.1206, "step": 3020 }, { "epoch": 0.25, "learning_rate": 1.9491098421229425e-05, "loss": 0.1806, "step": 3030 }, { "epoch": 0.26, "learning_rate": 1.9489418878065168e-05, "loss": 0.1243, "step": 3040 }, { "epoch": 0.26, "learning_rate": 1.948773933490091e-05, "loss": 0.0524, "step": 3050 }, { "epoch": 0.26, "learning_rate": 1.948605979173665e-05, "loss": 0.1983, "step": 3060 }, { "epoch": 0.26, "learning_rate": 1.948438024857239e-05, "loss": 0.1714, "step": 3070 }, { "epoch": 0.26, "learning_rate": 1.948270070540813e-05, "loss": 0.1264, "step": 3080 }, { "epoch": 0.26, "learning_rate": 1.948102116224387e-05, "loss": 0.2442, "step": 3090 }, { "epoch": 0.26, "learning_rate": 1.9479341619079614e-05, "loss": 0.1444, "step": 3100 }, { "epoch": 0.26, "learning_rate": 1.9477662075915353e-05, "loss": 0.3009, "step": 3110 }, { "epoch": 0.26, "learning_rate": 1.9475982532751095e-05, "loss": 0.1253, "step": 3120 }, { "epoch": 0.26, "learning_rate": 1.9474302989586834e-05, "loss": 0.1791, "step": 3130 }, { "epoch": 0.26, "learning_rate": 1.9472623446422572e-05, "loss": 0.1462, "step": 3140 }, { "epoch": 0.26, "learning_rate": 1.9470943903258314e-05, "loss": 0.2484, "step": 3150 }, { "epoch": 0.27, "learning_rate": 1.9469264360094057e-05, "loss": 0.2536, "step": 3160 }, { "epoch": 0.27, "learning_rate": 1.9467584816929795e-05, "loss": 0.1864, "step": 3170 }, { "epoch": 0.27, "learning_rate": 1.9465905273765538e-05, "loss": 0.1885, "step": 3180 }, { "epoch": 0.27, "learning_rate": 1.946422573060128e-05, "loss": 0.171, "step": 3190 }, { "epoch": 0.27, "learning_rate": 1.946254618743702e-05, "loss": 0.1559, "step": 3200 }, { "epoch": 0.27, "learning_rate": 1.946086664427276e-05, "loss": 0.1985, "step": 3210 }, { "epoch": 0.27, "learning_rate": 1.94591871011085e-05, "loss": 0.0936, "step": 3220 }, { "epoch": 0.27, "learning_rate": 1.945750755794424e-05, "loss": 0.119, "step": 3230 }, { "epoch": 0.27, "learning_rate": 1.945582801477998e-05, "loss": 0.2116, "step": 3240 }, { "epoch": 0.27, "learning_rate": 1.9454148471615723e-05, "loss": 0.275, "step": 3250 }, { "epoch": 0.27, "learning_rate": 1.945246892845146e-05, "loss": 0.1147, "step": 3260 }, { "epoch": 0.27, "learning_rate": 1.9450789385287204e-05, "loss": 0.2131, "step": 3270 }, { "epoch": 0.28, "learning_rate": 1.9449109842122946e-05, "loss": 0.0928, "step": 3280 }, { "epoch": 0.28, "learning_rate": 1.9447430298958685e-05, "loss": 0.1817, "step": 3290 }, { "epoch": 0.28, "learning_rate": 1.9445750755794427e-05, "loss": 0.1636, "step": 3300 }, { "epoch": 0.28, "learning_rate": 1.9444071212630165e-05, "loss": 0.2522, "step": 3310 }, { "epoch": 0.28, "learning_rate": 1.9442391669465908e-05, "loss": 0.2372, "step": 3320 }, { "epoch": 0.28, "learning_rate": 1.944071212630165e-05, "loss": 0.1789, "step": 3330 }, { "epoch": 0.28, "learning_rate": 1.943903258313739e-05, "loss": 0.1699, "step": 3340 }, { "epoch": 0.28, "learning_rate": 1.9437353039973127e-05, "loss": 0.1754, "step": 3350 }, { "epoch": 0.28, "learning_rate": 1.943567349680887e-05, "loss": 0.1546, "step": 3360 }, { "epoch": 0.28, "learning_rate": 1.943399395364461e-05, "loss": 0.2049, "step": 3370 }, { "epoch": 0.28, "learning_rate": 1.943231441048035e-05, "loss": 0.1494, "step": 3380 }, { "epoch": 0.28, "learning_rate": 1.9430634867316093e-05, "loss": 0.1436, "step": 3390 }, { "epoch": 0.29, "learning_rate": 1.942895532415183e-05, "loss": 0.1644, "step": 3400 }, { "epoch": 0.29, "learning_rate": 1.9427275780987574e-05, "loss": 0.1102, "step": 3410 }, { "epoch": 0.29, "learning_rate": 1.9425596237823316e-05, "loss": 0.156, "step": 3420 }, { "epoch": 0.29, "learning_rate": 1.9423916694659055e-05, "loss": 0.0851, "step": 3430 }, { "epoch": 0.29, "learning_rate": 1.9422237151494797e-05, "loss": 0.0532, "step": 3440 }, { "epoch": 0.29, "learning_rate": 1.9420557608330536e-05, "loss": 0.1915, "step": 3450 }, { "epoch": 0.29, "learning_rate": 1.9418878065166274e-05, "loss": 0.1014, "step": 3460 }, { "epoch": 0.29, "learning_rate": 1.9417198522002016e-05, "loss": 0.1518, "step": 3470 }, { "epoch": 0.29, "learning_rate": 1.941551897883776e-05, "loss": 0.0734, "step": 3480 }, { "epoch": 0.29, "learning_rate": 1.9413839435673497e-05, "loss": 0.1521, "step": 3490 }, { "epoch": 0.29, "learning_rate": 1.941215989250924e-05, "loss": 0.159, "step": 3500 }, { "epoch": 0.29, "learning_rate": 1.941048034934498e-05, "loss": 0.1921, "step": 3510 }, { "epoch": 0.3, "learning_rate": 1.940880080618072e-05, "loss": 0.1669, "step": 3520 }, { "epoch": 0.3, "learning_rate": 1.9407121263016463e-05, "loss": 0.1503, "step": 3530 }, { "epoch": 0.3, "learning_rate": 1.94054417198522e-05, "loss": 0.1918, "step": 3540 }, { "epoch": 0.3, "learning_rate": 1.940376217668794e-05, "loss": 0.1365, "step": 3550 }, { "epoch": 0.3, "learning_rate": 1.9402082633523682e-05, "loss": 0.1548, "step": 3560 }, { "epoch": 0.3, "learning_rate": 1.9400403090359425e-05, "loss": 0.1372, "step": 3570 }, { "epoch": 0.3, "learning_rate": 1.9398723547195163e-05, "loss": 0.1452, "step": 3580 }, { "epoch": 0.3, "learning_rate": 1.9397044004030906e-05, "loss": 0.1475, "step": 3590 }, { "epoch": 0.3, "learning_rate": 1.9395364460866644e-05, "loss": 0.2006, "step": 3600 }, { "epoch": 0.3, "learning_rate": 1.9393684917702387e-05, "loss": 0.1665, "step": 3610 }, { "epoch": 0.3, "learning_rate": 1.939200537453813e-05, "loss": 0.1278, "step": 3620 }, { "epoch": 0.3, "learning_rate": 1.9390325831373868e-05, "loss": 0.0399, "step": 3630 }, { "epoch": 0.31, "learning_rate": 1.938864628820961e-05, "loss": 0.1193, "step": 3640 }, { "epoch": 0.31, "learning_rate": 1.9386966745045352e-05, "loss": 0.1565, "step": 3650 }, { "epoch": 0.31, "learning_rate": 1.9385287201881087e-05, "loss": 0.0575, "step": 3660 }, { "epoch": 0.31, "learning_rate": 1.938360765871683e-05, "loss": 0.1936, "step": 3670 }, { "epoch": 0.31, "learning_rate": 1.938192811555257e-05, "loss": 0.2479, "step": 3680 }, { "epoch": 0.31, "learning_rate": 1.938024857238831e-05, "loss": 0.1538, "step": 3690 }, { "epoch": 0.31, "learning_rate": 1.9378569029224053e-05, "loss": 0.1937, "step": 3700 }, { "epoch": 0.31, "learning_rate": 1.9376889486059795e-05, "loss": 0.1472, "step": 3710 }, { "epoch": 0.31, "learning_rate": 1.9375209942895533e-05, "loss": 0.1534, "step": 3720 }, { "epoch": 0.31, "learning_rate": 1.9373530399731276e-05, "loss": 0.2154, "step": 3730 }, { "epoch": 0.31, "learning_rate": 1.9371850856567014e-05, "loss": 0.2354, "step": 3740 }, { "epoch": 0.31, "learning_rate": 1.9370171313402757e-05, "loss": 0.1398, "step": 3750 }, { "epoch": 0.32, "learning_rate": 1.93684917702385e-05, "loss": 0.2258, "step": 3760 }, { "epoch": 0.32, "learning_rate": 1.9366812227074238e-05, "loss": 0.1034, "step": 3770 }, { "epoch": 0.32, "learning_rate": 1.9365132683909976e-05, "loss": 0.1472, "step": 3780 }, { "epoch": 0.32, "learning_rate": 1.936345314074572e-05, "loss": 0.2002, "step": 3790 }, { "epoch": 0.32, "learning_rate": 1.936177359758146e-05, "loss": 0.0692, "step": 3800 }, { "epoch": 0.32, "learning_rate": 1.93600940544172e-05, "loss": 0.2311, "step": 3810 }, { "epoch": 0.32, "learning_rate": 1.935841451125294e-05, "loss": 0.2038, "step": 3820 }, { "epoch": 0.32, "learning_rate": 1.935673496808868e-05, "loss": 0.1764, "step": 3830 }, { "epoch": 0.32, "learning_rate": 1.9355055424924423e-05, "loss": 0.1446, "step": 3840 }, { "epoch": 0.32, "learning_rate": 1.9353375881760165e-05, "loss": 0.1503, "step": 3850 }, { "epoch": 0.32, "learning_rate": 1.9351696338595904e-05, "loss": 0.0751, "step": 3860 }, { "epoch": 0.32, "learning_rate": 1.9350016795431642e-05, "loss": 0.138, "step": 3870 }, { "epoch": 0.33, "learning_rate": 1.9348337252267385e-05, "loss": 0.1876, "step": 3880 }, { "epoch": 0.33, "learning_rate": 1.9346657709103123e-05, "loss": 0.1893, "step": 3890 }, { "epoch": 0.33, "learning_rate": 1.9344978165938865e-05, "loss": 0.209, "step": 3900 }, { "epoch": 0.33, "learning_rate": 1.9343298622774608e-05, "loss": 0.1501, "step": 3910 }, { "epoch": 0.33, "learning_rate": 1.9341619079610346e-05, "loss": 0.1257, "step": 3920 }, { "epoch": 0.33, "learning_rate": 1.933993953644609e-05, "loss": 0.1683, "step": 3930 }, { "epoch": 0.33, "learning_rate": 1.933825999328183e-05, "loss": 0.1419, "step": 3940 }, { "epoch": 0.33, "learning_rate": 1.933658045011757e-05, "loss": 0.1715, "step": 3950 }, { "epoch": 0.33, "learning_rate": 1.9334900906953312e-05, "loss": 0.1383, "step": 3960 }, { "epoch": 0.33, "learning_rate": 1.933322136378905e-05, "loss": 0.1324, "step": 3970 }, { "epoch": 0.33, "learning_rate": 1.933154182062479e-05, "loss": 0.2321, "step": 3980 }, { "epoch": 0.34, "learning_rate": 1.932986227746053e-05, "loss": 0.163, "step": 3990 }, { "epoch": 0.34, "learning_rate": 1.9328182734296274e-05, "loss": 0.1512, "step": 4000 }, { "epoch": 0.34, "learning_rate": 1.9326503191132012e-05, "loss": 0.1747, "step": 4010 }, { "epoch": 0.34, "learning_rate": 1.9324823647967755e-05, "loss": 0.183, "step": 4020 }, { "epoch": 0.34, "learning_rate": 1.9323144104803497e-05, "loss": 0.1209, "step": 4030 }, { "epoch": 0.34, "learning_rate": 1.9321464561639236e-05, "loss": 0.1099, "step": 4040 }, { "epoch": 0.34, "learning_rate": 1.9319785018474978e-05, "loss": 0.164, "step": 4050 }, { "epoch": 0.34, "learning_rate": 1.9318105475310716e-05, "loss": 0.1858, "step": 4060 }, { "epoch": 0.34, "learning_rate": 1.931642593214646e-05, "loss": 0.1479, "step": 4070 }, { "epoch": 0.34, "learning_rate": 1.9314746388982197e-05, "loss": 0.2251, "step": 4080 }, { "epoch": 0.34, "learning_rate": 1.931306684581794e-05, "loss": 0.0943, "step": 4090 }, { "epoch": 0.34, "learning_rate": 1.931138730265368e-05, "loss": 0.1636, "step": 4100 }, { "epoch": 0.35, "learning_rate": 1.930970775948942e-05, "loss": 0.1188, "step": 4110 }, { "epoch": 0.35, "learning_rate": 1.930802821632516e-05, "loss": 0.1152, "step": 4120 }, { "epoch": 0.35, "learning_rate": 1.93063486731609e-05, "loss": 0.1245, "step": 4130 }, { "epoch": 0.35, "learning_rate": 1.9304669129996644e-05, "loss": 0.2023, "step": 4140 }, { "epoch": 0.35, "learning_rate": 1.9302989586832382e-05, "loss": 0.0928, "step": 4150 }, { "epoch": 0.35, "learning_rate": 1.9301310043668125e-05, "loss": 0.2239, "step": 4160 }, { "epoch": 0.35, "learning_rate": 1.9299630500503867e-05, "loss": 0.0714, "step": 4170 }, { "epoch": 0.35, "learning_rate": 1.9297950957339606e-05, "loss": 0.1044, "step": 4180 }, { "epoch": 0.35, "learning_rate": 1.9296271414175344e-05, "loss": 0.2008, "step": 4190 }, { "epoch": 0.35, "learning_rate": 1.9294591871011087e-05, "loss": 0.1327, "step": 4200 }, { "epoch": 0.35, "learning_rate": 1.9292912327846825e-05, "loss": 0.135, "step": 4210 }, { "epoch": 0.35, "learning_rate": 1.9291232784682567e-05, "loss": 0.1791, "step": 4220 }, { "epoch": 0.36, "learning_rate": 1.928955324151831e-05, "loss": 0.109, "step": 4230 }, { "epoch": 0.36, "learning_rate": 1.928787369835405e-05, "loss": 0.1856, "step": 4240 }, { "epoch": 0.36, "learning_rate": 1.928619415518979e-05, "loss": 0.1286, "step": 4250 }, { "epoch": 0.36, "learning_rate": 1.9284514612025533e-05, "loss": 0.2671, "step": 4260 }, { "epoch": 0.36, "learning_rate": 1.928283506886127e-05, "loss": 0.1819, "step": 4270 }, { "epoch": 0.36, "learning_rate": 1.9281155525697014e-05, "loss": 0.108, "step": 4280 }, { "epoch": 0.36, "learning_rate": 1.9279475982532753e-05, "loss": 0.1305, "step": 4290 }, { "epoch": 0.36, "learning_rate": 1.927779643936849e-05, "loss": 0.1593, "step": 4300 }, { "epoch": 0.36, "learning_rate": 1.9276116896204233e-05, "loss": 0.0948, "step": 4310 }, { "epoch": 0.36, "learning_rate": 1.9274437353039976e-05, "loss": 0.2051, "step": 4320 }, { "epoch": 0.36, "learning_rate": 1.9272757809875714e-05, "loss": 0.1186, "step": 4330 }, { "epoch": 0.36, "learning_rate": 1.9271078266711457e-05, "loss": 0.1554, "step": 4340 }, { "epoch": 0.37, "learning_rate": 1.9269398723547195e-05, "loss": 0.2427, "step": 4350 }, { "epoch": 0.37, "learning_rate": 1.9267719180382938e-05, "loss": 0.1904, "step": 4360 }, { "epoch": 0.37, "learning_rate": 1.926603963721868e-05, "loss": 0.1257, "step": 4370 }, { "epoch": 0.37, "learning_rate": 1.926436009405442e-05, "loss": 0.2028, "step": 4380 }, { "epoch": 0.37, "learning_rate": 1.926268055089016e-05, "loss": 0.1195, "step": 4390 }, { "epoch": 0.37, "learning_rate": 1.92610010077259e-05, "loss": 0.1655, "step": 4400 }, { "epoch": 0.37, "learning_rate": 1.925932146456164e-05, "loss": 0.1342, "step": 4410 }, { "epoch": 0.37, "learning_rate": 1.925764192139738e-05, "loss": 0.1432, "step": 4420 }, { "epoch": 0.37, "learning_rate": 1.9255962378233123e-05, "loss": 0.1337, "step": 4430 }, { "epoch": 0.37, "learning_rate": 1.925428283506886e-05, "loss": 0.138, "step": 4440 }, { "epoch": 0.37, "learning_rate": 1.9252603291904604e-05, "loss": 0.1423, "step": 4450 }, { "epoch": 0.37, "learning_rate": 1.9250923748740346e-05, "loss": 0.0725, "step": 4460 }, { "epoch": 0.38, "learning_rate": 1.9249244205576084e-05, "loss": 0.1757, "step": 4470 }, { "epoch": 0.38, "learning_rate": 1.9247564662411827e-05, "loss": 0.2311, "step": 4480 }, { "epoch": 0.38, "learning_rate": 1.9245885119247565e-05, "loss": 0.1313, "step": 4490 }, { "epoch": 0.38, "learning_rate": 1.9244205576083308e-05, "loss": 0.1183, "step": 4500 }, { "epoch": 0.38, "learning_rate": 1.9242526032919046e-05, "loss": 0.132, "step": 4510 }, { "epoch": 0.38, "learning_rate": 1.924084648975479e-05, "loss": 0.1848, "step": 4520 }, { "epoch": 0.38, "learning_rate": 1.9239166946590527e-05, "loss": 0.0848, "step": 4530 }, { "epoch": 0.38, "learning_rate": 1.923748740342627e-05, "loss": 0.12, "step": 4540 }, { "epoch": 0.38, "learning_rate": 1.9235807860262012e-05, "loss": 0.1796, "step": 4550 }, { "epoch": 0.38, "learning_rate": 1.923412831709775e-05, "loss": 0.1926, "step": 4560 }, { "epoch": 0.38, "learning_rate": 1.9232448773933493e-05, "loss": 0.1578, "step": 4570 }, { "epoch": 0.38, "learning_rate": 1.923076923076923e-05, "loss": 0.1322, "step": 4580 }, { "epoch": 0.39, "learning_rate": 1.9229089687604974e-05, "loss": 0.1746, "step": 4590 }, { "epoch": 0.39, "learning_rate": 1.9227410144440716e-05, "loss": 0.1175, "step": 4600 }, { "epoch": 0.39, "learning_rate": 1.9225730601276455e-05, "loss": 0.1106, "step": 4610 }, { "epoch": 0.39, "learning_rate": 1.9224051058112193e-05, "loss": 0.1361, "step": 4620 }, { "epoch": 0.39, "learning_rate": 1.9222371514947936e-05, "loss": 0.1486, "step": 4630 }, { "epoch": 0.39, "learning_rate": 1.9220691971783674e-05, "loss": 0.0986, "step": 4640 }, { "epoch": 0.39, "learning_rate": 1.9219012428619416e-05, "loss": 0.22, "step": 4650 }, { "epoch": 0.39, "learning_rate": 1.921733288545516e-05, "loss": 0.1632, "step": 4660 }, { "epoch": 0.39, "learning_rate": 1.9215653342290897e-05, "loss": 0.1943, "step": 4670 }, { "epoch": 0.39, "learning_rate": 1.921397379912664e-05, "loss": 0.1371, "step": 4680 }, { "epoch": 0.39, "learning_rate": 1.9212294255962382e-05, "loss": 0.2131, "step": 4690 }, { "epoch": 0.39, "learning_rate": 1.921061471279812e-05, "loss": 0.1185, "step": 4700 }, { "epoch": 0.4, "learning_rate": 1.9208935169633863e-05, "loss": 0.1252, "step": 4710 }, { "epoch": 0.4, "learning_rate": 1.92072556264696e-05, "loss": 0.0882, "step": 4720 }, { "epoch": 0.4, "learning_rate": 1.920557608330534e-05, "loss": 0.1718, "step": 4730 }, { "epoch": 0.4, "learning_rate": 1.9203896540141082e-05, "loss": 0.1422, "step": 4740 }, { "epoch": 0.4, "learning_rate": 1.9202216996976825e-05, "loss": 0.1337, "step": 4750 }, { "epoch": 0.4, "learning_rate": 1.9200537453812563e-05, "loss": 0.1449, "step": 4760 }, { "epoch": 0.4, "learning_rate": 1.9198857910648306e-05, "loss": 0.1036, "step": 4770 }, { "epoch": 0.4, "learning_rate": 1.9197178367484048e-05, "loss": 0.0878, "step": 4780 }, { "epoch": 0.4, "learning_rate": 1.9195498824319787e-05, "loss": 0.1865, "step": 4790 }, { "epoch": 0.4, "learning_rate": 1.919381928115553e-05, "loss": 0.134, "step": 4800 }, { "epoch": 0.4, "learning_rate": 1.9192139737991267e-05, "loss": 0.1031, "step": 4810 }, { "epoch": 0.4, "learning_rate": 1.919046019482701e-05, "loss": 0.104, "step": 4820 }, { "epoch": 0.41, "learning_rate": 1.918878065166275e-05, "loss": 0.1766, "step": 4830 }, { "epoch": 0.41, "learning_rate": 1.918710110849849e-05, "loss": 0.158, "step": 4840 }, { "epoch": 0.41, "learning_rate": 1.918542156533423e-05, "loss": 0.158, "step": 4850 }, { "epoch": 0.41, "learning_rate": 1.918374202216997e-05, "loss": 0.1824, "step": 4860 }, { "epoch": 0.41, "learning_rate": 1.918206247900571e-05, "loss": 0.1378, "step": 4870 }, { "epoch": 0.41, "learning_rate": 1.9180382935841453e-05, "loss": 0.1439, "step": 4880 }, { "epoch": 0.41, "learning_rate": 1.9178703392677195e-05, "loss": 0.1312, "step": 4890 }, { "epoch": 0.41, "learning_rate": 1.9177023849512933e-05, "loss": 0.1587, "step": 4900 }, { "epoch": 0.41, "learning_rate": 1.9175344306348676e-05, "loss": 0.2084, "step": 4910 }, { "epoch": 0.41, "learning_rate": 1.9173664763184418e-05, "loss": 0.0893, "step": 4920 }, { "epoch": 0.41, "learning_rate": 1.9171985220020157e-05, "loss": 0.1488, "step": 4930 }, { "epoch": 0.41, "learning_rate": 1.9170305676855895e-05, "loss": 0.1516, "step": 4940 }, { "epoch": 0.42, "learning_rate": 1.9168626133691638e-05, "loss": 0.186, "step": 4950 }, { "epoch": 0.42, "learning_rate": 1.9166946590527376e-05, "loss": 0.1648, "step": 4960 }, { "epoch": 0.42, "learning_rate": 1.916526704736312e-05, "loss": 0.0715, "step": 4970 }, { "epoch": 0.42, "learning_rate": 1.916358750419886e-05, "loss": 0.1661, "step": 4980 }, { "epoch": 0.42, "learning_rate": 1.91619079610346e-05, "loss": 0.193, "step": 4990 }, { "epoch": 0.42, "learning_rate": 1.916022841787034e-05, "loss": 0.1231, "step": 5000 }, { "epoch": 0.42, "learning_rate": 1.9158548874706084e-05, "loss": 0.1449, "step": 5010 }, { "epoch": 0.42, "learning_rate": 1.9156869331541823e-05, "loss": 0.2417, "step": 5020 }, { "epoch": 0.42, "learning_rate": 1.9155189788377565e-05, "loss": 0.2212, "step": 5030 }, { "epoch": 0.42, "learning_rate": 1.9153510245213304e-05, "loss": 0.1148, "step": 5040 }, { "epoch": 0.42, "learning_rate": 1.9151830702049042e-05, "loss": 0.1606, "step": 5050 }, { "epoch": 0.42, "learning_rate": 1.9150151158884784e-05, "loss": 0.1193, "step": 5060 }, { "epoch": 0.43, "learning_rate": 1.9148471615720527e-05, "loss": 0.1973, "step": 5070 }, { "epoch": 0.43, "learning_rate": 1.9146792072556265e-05, "loss": 0.133, "step": 5080 }, { "epoch": 0.43, "learning_rate": 1.9145112529392008e-05, "loss": 0.1536, "step": 5090 }, { "epoch": 0.43, "learning_rate": 1.9143432986227746e-05, "loss": 0.2167, "step": 5100 }, { "epoch": 0.43, "learning_rate": 1.914175344306349e-05, "loss": 0.1538, "step": 5110 }, { "epoch": 0.43, "learning_rate": 1.914007389989923e-05, "loss": 0.1616, "step": 5120 }, { "epoch": 0.43, "learning_rate": 1.913839435673497e-05, "loss": 0.1964, "step": 5130 }, { "epoch": 0.43, "learning_rate": 1.9136714813570708e-05, "loss": 0.137, "step": 5140 }, { "epoch": 0.43, "learning_rate": 1.913503527040645e-05, "loss": 0.1507, "step": 5150 }, { "epoch": 0.43, "learning_rate": 1.9133355727242193e-05, "loss": 0.2119, "step": 5160 }, { "epoch": 0.43, "learning_rate": 1.913167618407793e-05, "loss": 0.2252, "step": 5170 }, { "epoch": 0.44, "learning_rate": 1.9129996640913674e-05, "loss": 0.1607, "step": 5180 }, { "epoch": 0.44, "learning_rate": 1.9128317097749412e-05, "loss": 0.0805, "step": 5190 }, { "epoch": 0.44, "learning_rate": 1.9126637554585155e-05, "loss": 0.1856, "step": 5200 }, { "epoch": 0.44, "learning_rate": 1.9124958011420897e-05, "loss": 0.0889, "step": 5210 }, { "epoch": 0.44, "learning_rate": 1.9123278468256635e-05, "loss": 0.1561, "step": 5220 }, { "epoch": 0.44, "learning_rate": 1.9121598925092378e-05, "loss": 0.126, "step": 5230 }, { "epoch": 0.44, "learning_rate": 1.911991938192812e-05, "loss": 0.175, "step": 5240 }, { "epoch": 0.44, "learning_rate": 1.9118239838763855e-05, "loss": 0.1019, "step": 5250 }, { "epoch": 0.44, "learning_rate": 1.9116560295599597e-05, "loss": 0.2117, "step": 5260 }, { "epoch": 0.44, "learning_rate": 1.911488075243534e-05, "loss": 0.1817, "step": 5270 }, { "epoch": 0.44, "learning_rate": 1.911320120927108e-05, "loss": 0.1322, "step": 5280 }, { "epoch": 0.44, "learning_rate": 1.911152166610682e-05, "loss": 0.0827, "step": 5290 }, { "epoch": 0.45, "learning_rate": 1.9109842122942563e-05, "loss": 0.1181, "step": 5300 }, { "epoch": 0.45, "learning_rate": 1.91081625797783e-05, "loss": 0.1213, "step": 5310 }, { "epoch": 0.45, "learning_rate": 1.9106483036614044e-05, "loss": 0.1477, "step": 5320 }, { "epoch": 0.45, "learning_rate": 1.9104803493449782e-05, "loss": 0.1271, "step": 5330 }, { "epoch": 0.45, "learning_rate": 1.9103123950285525e-05, "loss": 0.1505, "step": 5340 }, { "epoch": 0.45, "learning_rate": 1.9101444407121267e-05, "loss": 0.1168, "step": 5350 }, { "epoch": 0.45, "learning_rate": 1.9099764863957006e-05, "loss": 0.159, "step": 5360 }, { "epoch": 0.45, "learning_rate": 1.9098085320792744e-05, "loss": 0.1248, "step": 5370 }, { "epoch": 0.45, "learning_rate": 1.9096405777628487e-05, "loss": 0.1399, "step": 5380 }, { "epoch": 0.45, "learning_rate": 1.909472623446423e-05, "loss": 0.17, "step": 5390 }, { "epoch": 0.45, "learning_rate": 1.9093046691299967e-05, "loss": 0.1902, "step": 5400 }, { "epoch": 0.45, "learning_rate": 1.909136714813571e-05, "loss": 0.1734, "step": 5410 }, { "epoch": 0.46, "learning_rate": 1.908968760497145e-05, "loss": 0.1177, "step": 5420 }, { "epoch": 0.46, "learning_rate": 1.908800806180719e-05, "loss": 0.1423, "step": 5430 }, { "epoch": 0.46, "learning_rate": 1.9086328518642933e-05, "loss": 0.144, "step": 5440 }, { "epoch": 0.46, "learning_rate": 1.908464897547867e-05, "loss": 0.1238, "step": 5450 }, { "epoch": 0.46, "learning_rate": 1.908296943231441e-05, "loss": 0.0819, "step": 5460 }, { "epoch": 0.46, "learning_rate": 1.9081289889150152e-05, "loss": 0.1869, "step": 5470 }, { "epoch": 0.46, "learning_rate": 1.907961034598589e-05, "loss": 0.1941, "step": 5480 }, { "epoch": 0.46, "learning_rate": 1.9077930802821633e-05, "loss": 0.1059, "step": 5490 }, { "epoch": 0.46, "learning_rate": 1.9076251259657376e-05, "loss": 0.1605, "step": 5500 }, { "epoch": 0.46, "learning_rate": 1.9074571716493114e-05, "loss": 0.178, "step": 5510 }, { "epoch": 0.46, "learning_rate": 1.9072892173328857e-05, "loss": 0.1384, "step": 5520 }, { "epoch": 0.46, "learning_rate": 1.90712126301646e-05, "loss": 0.1234, "step": 5530 }, { "epoch": 0.47, "learning_rate": 1.9069533087000338e-05, "loss": 0.104, "step": 5540 }, { "epoch": 0.47, "learning_rate": 1.906785354383608e-05, "loss": 0.1186, "step": 5550 }, { "epoch": 0.47, "learning_rate": 1.906617400067182e-05, "loss": 0.1532, "step": 5560 }, { "epoch": 0.47, "learning_rate": 1.9064494457507557e-05, "loss": 0.2436, "step": 5570 }, { "epoch": 0.47, "learning_rate": 1.90628149143433e-05, "loss": 0.0736, "step": 5580 }, { "epoch": 0.47, "learning_rate": 1.906113537117904e-05, "loss": 0.126, "step": 5590 }, { "epoch": 0.47, "learning_rate": 1.905945582801478e-05, "loss": 0.1004, "step": 5600 }, { "epoch": 0.47, "learning_rate": 1.9057776284850523e-05, "loss": 0.156, "step": 5610 }, { "epoch": 0.47, "learning_rate": 1.905609674168626e-05, "loss": 0.129, "step": 5620 }, { "epoch": 0.47, "learning_rate": 1.9054417198522004e-05, "loss": 0.1707, "step": 5630 }, { "epoch": 0.47, "learning_rate": 1.9052737655357746e-05, "loss": 0.128, "step": 5640 }, { "epoch": 0.47, "learning_rate": 1.9051058112193484e-05, "loss": 0.1616, "step": 5650 }, { "epoch": 0.48, "learning_rate": 1.9049378569029227e-05, "loss": 0.0966, "step": 5660 }, { "epoch": 0.48, "learning_rate": 1.9047699025864965e-05, "loss": 0.2517, "step": 5670 }, { "epoch": 0.48, "learning_rate": 1.9046019482700708e-05, "loss": 0.1216, "step": 5680 }, { "epoch": 0.48, "learning_rate": 1.9044339939536446e-05, "loss": 0.1799, "step": 5690 }, { "epoch": 0.48, "learning_rate": 1.904266039637219e-05, "loss": 0.1611, "step": 5700 }, { "epoch": 0.48, "learning_rate": 1.9040980853207927e-05, "loss": 0.0891, "step": 5710 }, { "epoch": 0.48, "learning_rate": 1.903930131004367e-05, "loss": 0.1458, "step": 5720 }, { "epoch": 0.48, "learning_rate": 1.903762176687941e-05, "loss": 0.1241, "step": 5730 }, { "epoch": 0.48, "learning_rate": 1.903594222371515e-05, "loss": 0.1175, "step": 5740 }, { "epoch": 0.48, "learning_rate": 1.9034262680550893e-05, "loss": 0.1161, "step": 5750 }, { "epoch": 0.48, "learning_rate": 1.9032583137386635e-05, "loss": 0.1361, "step": 5760 }, { "epoch": 0.48, "learning_rate": 1.9030903594222374e-05, "loss": 0.2429, "step": 5770 }, { "epoch": 0.49, "learning_rate": 1.9029224051058112e-05, "loss": 0.1602, "step": 5780 }, { "epoch": 0.49, "learning_rate": 1.9027544507893855e-05, "loss": 0.0632, "step": 5790 }, { "epoch": 0.49, "learning_rate": 1.9025864964729593e-05, "loss": 0.1526, "step": 5800 }, { "epoch": 0.49, "learning_rate": 1.9024185421565335e-05, "loss": 0.167, "step": 5810 }, { "epoch": 0.49, "learning_rate": 1.9022505878401078e-05, "loss": 0.1275, "step": 5820 }, { "epoch": 0.49, "learning_rate": 1.9020826335236816e-05, "loss": 0.1689, "step": 5830 }, { "epoch": 0.49, "learning_rate": 1.901914679207256e-05, "loss": 0.0933, "step": 5840 }, { "epoch": 0.49, "learning_rate": 1.9017467248908297e-05, "loss": 0.078, "step": 5850 }, { "epoch": 0.49, "learning_rate": 1.901578770574404e-05, "loss": 0.0863, "step": 5860 }, { "epoch": 0.49, "learning_rate": 1.9014108162579782e-05, "loss": 0.1911, "step": 5870 }, { "epoch": 0.49, "learning_rate": 1.901242861941552e-05, "loss": 0.1418, "step": 5880 }, { "epoch": 0.49, "learning_rate": 1.901074907625126e-05, "loss": 0.1426, "step": 5890 }, { "epoch": 0.5, "learning_rate": 1.9009069533087e-05, "loss": 0.1089, "step": 5900 }, { "epoch": 0.5, "learning_rate": 1.9007389989922744e-05, "loss": 0.1443, "step": 5910 }, { "epoch": 0.5, "learning_rate": 1.9005710446758482e-05, "loss": 0.1013, "step": 5920 }, { "epoch": 0.5, "learning_rate": 1.9004030903594225e-05, "loss": 0.1444, "step": 5930 }, { "epoch": 0.5, "learning_rate": 1.9002351360429963e-05, "loss": 0.1407, "step": 5940 }, { "epoch": 0.5, "learning_rate": 1.9000671817265706e-05, "loss": 0.1082, "step": 5950 }, { "epoch": 0.5, "learning_rate": 1.8998992274101448e-05, "loss": 0.1194, "step": 5960 }, { "epoch": 0.5, "learning_rate": 1.8997312730937186e-05, "loss": 0.1777, "step": 5970 }, { "epoch": 0.5, "learning_rate": 1.899563318777293e-05, "loss": 0.1417, "step": 5980 }, { "epoch": 0.5, "learning_rate": 1.8993953644608667e-05, "loss": 0.182, "step": 5990 }, { "epoch": 0.5, "learning_rate": 1.8992274101444406e-05, "loss": 0.1138, "step": 6000 }, { "epoch": 0.5, "learning_rate": 1.899059455828015e-05, "loss": 0.1042, "step": 6010 }, { "epoch": 0.51, "learning_rate": 1.898891501511589e-05, "loss": 0.1148, "step": 6020 }, { "epoch": 0.51, "learning_rate": 1.898723547195163e-05, "loss": 0.1438, "step": 6030 }, { "epoch": 0.51, "learning_rate": 1.898555592878737e-05, "loss": 0.1665, "step": 6040 }, { "epoch": 0.51, "learning_rate": 1.8983876385623114e-05, "loss": 0.1941, "step": 6050 }, { "epoch": 0.51, "learning_rate": 1.8982196842458852e-05, "loss": 0.1601, "step": 6060 }, { "epoch": 0.51, "learning_rate": 1.8980517299294595e-05, "loss": 0.0873, "step": 6070 }, { "epoch": 0.51, "learning_rate": 1.8978837756130333e-05, "loss": 0.0833, "step": 6080 }, { "epoch": 0.51, "learning_rate": 1.8977158212966076e-05, "loss": 0.1073, "step": 6090 }, { "epoch": 0.51, "learning_rate": 1.8975478669801814e-05, "loss": 0.1362, "step": 6100 }, { "epoch": 0.51, "learning_rate": 1.8973799126637557e-05, "loss": 0.0569, "step": 6110 }, { "epoch": 0.51, "learning_rate": 1.8972119583473295e-05, "loss": 0.1105, "step": 6120 }, { "epoch": 0.51, "learning_rate": 1.8970440040309038e-05, "loss": 0.3129, "step": 6130 }, { "epoch": 0.52, "learning_rate": 1.896876049714478e-05, "loss": 0.131, "step": 6140 }, { "epoch": 0.52, "learning_rate": 1.896708095398052e-05, "loss": 0.1999, "step": 6150 }, { "epoch": 0.52, "learning_rate": 1.896540141081626e-05, "loss": 0.1892, "step": 6160 }, { "epoch": 0.52, "learning_rate": 1.8963721867652e-05, "loss": 0.0845, "step": 6170 }, { "epoch": 0.52, "learning_rate": 1.896204232448774e-05, "loss": 0.092, "step": 6180 }, { "epoch": 0.52, "learning_rate": 1.8960362781323484e-05, "loss": 0.1676, "step": 6190 }, { "epoch": 0.52, "learning_rate": 1.8958683238159223e-05, "loss": 0.0921, "step": 6200 }, { "epoch": 0.52, "learning_rate": 1.895700369499496e-05, "loss": 0.1247, "step": 6210 }, { "epoch": 0.52, "learning_rate": 1.8955324151830703e-05, "loss": 0.1211, "step": 6220 }, { "epoch": 0.52, "learning_rate": 1.8953644608666442e-05, "loss": 0.2184, "step": 6230 }, { "epoch": 0.52, "learning_rate": 1.8951965065502184e-05, "loss": 0.1372, "step": 6240 }, { "epoch": 0.52, "learning_rate": 1.8950285522337927e-05, "loss": 0.2684, "step": 6250 }, { "epoch": 0.53, "learning_rate": 1.8948605979173665e-05, "loss": 0.1225, "step": 6260 }, { "epoch": 0.53, "learning_rate": 1.8946926436009408e-05, "loss": 0.1142, "step": 6270 }, { "epoch": 0.53, "learning_rate": 1.894524689284515e-05, "loss": 0.2412, "step": 6280 }, { "epoch": 0.53, "learning_rate": 1.894356734968089e-05, "loss": 0.1147, "step": 6290 }, { "epoch": 0.53, "learning_rate": 1.894188780651663e-05, "loss": 0.1314, "step": 6300 }, { "epoch": 0.53, "learning_rate": 1.894020826335237e-05, "loss": 0.0977, "step": 6310 }, { "epoch": 0.53, "learning_rate": 1.8938528720188108e-05, "loss": 0.2464, "step": 6320 }, { "epoch": 0.53, "learning_rate": 1.893684917702385e-05, "loss": 0.1764, "step": 6330 }, { "epoch": 0.53, "learning_rate": 1.8935169633859593e-05, "loss": 0.1318, "step": 6340 }, { "epoch": 0.53, "learning_rate": 1.893349009069533e-05, "loss": 0.1274, "step": 6350 }, { "epoch": 0.53, "learning_rate": 1.8931810547531074e-05, "loss": 0.1695, "step": 6360 }, { "epoch": 0.53, "learning_rate": 1.8930131004366816e-05, "loss": 0.143, "step": 6370 }, { "epoch": 0.54, "learning_rate": 1.8928451461202555e-05, "loss": 0.0751, "step": 6380 }, { "epoch": 0.54, "learning_rate": 1.8926771918038297e-05, "loss": 0.1153, "step": 6390 }, { "epoch": 0.54, "learning_rate": 1.8925092374874035e-05, "loss": 0.1249, "step": 6400 }, { "epoch": 0.54, "learning_rate": 1.8923412831709774e-05, "loss": 0.1513, "step": 6410 }, { "epoch": 0.54, "learning_rate": 1.8921733288545516e-05, "loss": 0.1354, "step": 6420 }, { "epoch": 0.54, "learning_rate": 1.892005374538126e-05, "loss": 0.1745, "step": 6430 }, { "epoch": 0.54, "learning_rate": 1.8918374202216997e-05, "loss": 0.0942, "step": 6440 }, { "epoch": 0.54, "learning_rate": 1.891669465905274e-05, "loss": 0.0594, "step": 6450 }, { "epoch": 0.54, "learning_rate": 1.891501511588848e-05, "loss": 0.1443, "step": 6460 }, { "epoch": 0.54, "learning_rate": 1.891333557272422e-05, "loss": 0.1173, "step": 6470 }, { "epoch": 0.54, "learning_rate": 1.8911656029559963e-05, "loss": 0.1051, "step": 6480 }, { "epoch": 0.55, "learning_rate": 1.89099764863957e-05, "loss": 0.1015, "step": 6490 }, { "epoch": 0.55, "learning_rate": 1.8908296943231444e-05, "loss": 0.1906, "step": 6500 }, { "epoch": 0.55, "learning_rate": 1.8906617400067186e-05, "loss": 0.0883, "step": 6510 }, { "epoch": 0.55, "learning_rate": 1.8904937856902925e-05, "loss": 0.0831, "step": 6520 }, { "epoch": 0.55, "learning_rate": 1.8903258313738663e-05, "loss": 0.2414, "step": 6530 }, { "epoch": 0.55, "learning_rate": 1.8901578770574406e-05, "loss": 0.1478, "step": 6540 }, { "epoch": 0.55, "learning_rate": 1.8899899227410144e-05, "loss": 0.1266, "step": 6550 }, { "epoch": 0.55, "learning_rate": 1.8898219684245886e-05, "loss": 0.128, "step": 6560 }, { "epoch": 0.55, "learning_rate": 1.889654014108163e-05, "loss": 0.1527, "step": 6570 }, { "epoch": 0.55, "learning_rate": 1.8894860597917367e-05, "loss": 0.1903, "step": 6580 }, { "epoch": 0.55, "learning_rate": 1.889318105475311e-05, "loss": 0.1263, "step": 6590 }, { "epoch": 0.55, "learning_rate": 1.889150151158885e-05, "loss": 0.116, "step": 6600 }, { "epoch": 0.56, "learning_rate": 1.888982196842459e-05, "loss": 0.1189, "step": 6610 }, { "epoch": 0.56, "learning_rate": 1.8888142425260333e-05, "loss": 0.1252, "step": 6620 }, { "epoch": 0.56, "learning_rate": 1.888646288209607e-05, "loss": 0.0513, "step": 6630 }, { "epoch": 0.56, "learning_rate": 1.888478333893181e-05, "loss": 0.0833, "step": 6640 }, { "epoch": 0.56, "learning_rate": 1.8883103795767552e-05, "loss": 0.2002, "step": 6650 }, { "epoch": 0.56, "learning_rate": 1.8881424252603295e-05, "loss": 0.1723, "step": 6660 }, { "epoch": 0.56, "learning_rate": 1.8879744709439033e-05, "loss": 0.0902, "step": 6670 }, { "epoch": 0.56, "learning_rate": 1.8878065166274776e-05, "loss": 0.1129, "step": 6680 }, { "epoch": 0.56, "learning_rate": 1.8876385623110514e-05, "loss": 0.1233, "step": 6690 }, { "epoch": 0.56, "learning_rate": 1.8874706079946257e-05, "loss": 0.2482, "step": 6700 }, { "epoch": 0.56, "learning_rate": 1.8873026536782e-05, "loss": 0.1041, "step": 6710 }, { "epoch": 0.56, "learning_rate": 1.8871346993617737e-05, "loss": 0.2242, "step": 6720 }, { "epoch": 0.57, "learning_rate": 1.8869667450453476e-05, "loss": 0.1274, "step": 6730 }, { "epoch": 0.57, "learning_rate": 1.886798790728922e-05, "loss": 0.171, "step": 6740 }, { "epoch": 0.57, "learning_rate": 1.886630836412496e-05, "loss": 0.1655, "step": 6750 }, { "epoch": 0.57, "learning_rate": 1.88646288209607e-05, "loss": 0.1158, "step": 6760 }, { "epoch": 0.57, "learning_rate": 1.886294927779644e-05, "loss": 0.1309, "step": 6770 }, { "epoch": 0.57, "learning_rate": 1.886126973463218e-05, "loss": 0.0907, "step": 6780 }, { "epoch": 0.57, "learning_rate": 1.8859590191467923e-05, "loss": 0.2247, "step": 6790 }, { "epoch": 0.57, "learning_rate": 1.8857910648303665e-05, "loss": 0.1907, "step": 6800 }, { "epoch": 0.57, "learning_rate": 1.8856231105139403e-05, "loss": 0.1716, "step": 6810 }, { "epoch": 0.57, "learning_rate": 1.8854551561975146e-05, "loss": 0.1648, "step": 6820 }, { "epoch": 0.57, "learning_rate": 1.8852872018810884e-05, "loss": 0.0629, "step": 6830 }, { "epoch": 0.57, "learning_rate": 1.8851192475646623e-05, "loss": 0.1381, "step": 6840 }, { "epoch": 0.58, "learning_rate": 1.8849512932482365e-05, "loss": 0.0864, "step": 6850 }, { "epoch": 0.58, "learning_rate": 1.8847833389318108e-05, "loss": 0.1788, "step": 6860 }, { "epoch": 0.58, "learning_rate": 1.8846153846153846e-05, "loss": 0.0956, "step": 6870 }, { "epoch": 0.58, "learning_rate": 1.884447430298959e-05, "loss": 0.1919, "step": 6880 }, { "epoch": 0.58, "learning_rate": 1.884279475982533e-05, "loss": 0.1437, "step": 6890 }, { "epoch": 0.58, "learning_rate": 1.884111521666107e-05, "loss": 0.1252, "step": 6900 }, { "epoch": 0.58, "learning_rate": 1.883943567349681e-05, "loss": 0.1259, "step": 6910 }, { "epoch": 0.58, "learning_rate": 1.883775613033255e-05, "loss": 0.174, "step": 6920 }, { "epoch": 0.58, "learning_rate": 1.8836076587168293e-05, "loss": 0.1034, "step": 6930 }, { "epoch": 0.58, "learning_rate": 1.883439704400403e-05, "loss": 0.163, "step": 6940 }, { "epoch": 0.58, "learning_rate": 1.8832717500839774e-05, "loss": 0.1515, "step": 6950 }, { "epoch": 0.58, "learning_rate": 1.8831037957675512e-05, "loss": 0.212, "step": 6960 }, { "epoch": 0.59, "learning_rate": 1.8829358414511254e-05, "loss": 0.2273, "step": 6970 }, { "epoch": 0.59, "learning_rate": 1.8827678871346993e-05, "loss": 0.1939, "step": 6980 }, { "epoch": 0.59, "learning_rate": 1.8825999328182735e-05, "loss": 0.121, "step": 6990 }, { "epoch": 0.59, "learning_rate": 1.8824319785018478e-05, "loss": 0.1977, "step": 7000 }, { "epoch": 0.59, "learning_rate": 1.8822640241854216e-05, "loss": 0.1391, "step": 7010 }, { "epoch": 0.59, "learning_rate": 1.882096069868996e-05, "loss": 0.1293, "step": 7020 }, { "epoch": 0.59, "learning_rate": 1.88192811555257e-05, "loss": 0.1391, "step": 7030 }, { "epoch": 0.59, "learning_rate": 1.881760161236144e-05, "loss": 0.1706, "step": 7040 }, { "epoch": 0.59, "learning_rate": 1.8815922069197178e-05, "loss": 0.1261, "step": 7050 }, { "epoch": 0.59, "learning_rate": 1.881424252603292e-05, "loss": 0.0935, "step": 7060 }, { "epoch": 0.59, "learning_rate": 1.881256298286866e-05, "loss": 0.1743, "step": 7070 }, { "epoch": 0.59, "learning_rate": 1.88108834397044e-05, "loss": 0.0916, "step": 7080 }, { "epoch": 0.6, "learning_rate": 1.8809203896540144e-05, "loss": 0.1275, "step": 7090 }, { "epoch": 0.6, "learning_rate": 1.8807524353375882e-05, "loss": 0.0913, "step": 7100 }, { "epoch": 0.6, "learning_rate": 1.8805844810211625e-05, "loss": 0.0859, "step": 7110 }, { "epoch": 0.6, "learning_rate": 1.8804165267047367e-05, "loss": 0.05, "step": 7120 }, { "epoch": 0.6, "learning_rate": 1.8802485723883105e-05, "loss": 0.1732, "step": 7130 }, { "epoch": 0.6, "learning_rate": 1.8800806180718848e-05, "loss": 0.1201, "step": 7140 }, { "epoch": 0.6, "learning_rate": 1.8799126637554586e-05, "loss": 0.1709, "step": 7150 }, { "epoch": 0.6, "learning_rate": 1.8797447094390325e-05, "loss": 0.0823, "step": 7160 }, { "epoch": 0.6, "learning_rate": 1.8795767551226067e-05, "loss": 0.2207, "step": 7170 }, { "epoch": 0.6, "learning_rate": 1.879408800806181e-05, "loss": 0.1777, "step": 7180 }, { "epoch": 0.6, "learning_rate": 1.879240846489755e-05, "loss": 0.2301, "step": 7190 }, { "epoch": 0.6, "learning_rate": 1.879072892173329e-05, "loss": 0.197, "step": 7200 }, { "epoch": 0.61, "learning_rate": 1.878904937856903e-05, "loss": 0.077, "step": 7210 }, { "epoch": 0.61, "learning_rate": 1.878736983540477e-05, "loss": 0.1408, "step": 7220 }, { "epoch": 0.61, "learning_rate": 1.8785690292240514e-05, "loss": 0.116, "step": 7230 }, { "epoch": 0.61, "learning_rate": 1.8784010749076252e-05, "loss": 0.2007, "step": 7240 }, { "epoch": 0.61, "learning_rate": 1.8782331205911995e-05, "loss": 0.0793, "step": 7250 }, { "epoch": 0.61, "learning_rate": 1.8780651662747733e-05, "loss": 0.1196, "step": 7260 }, { "epoch": 0.61, "learning_rate": 1.8778972119583476e-05, "loss": 0.0845, "step": 7270 }, { "epoch": 0.61, "learning_rate": 1.8777292576419214e-05, "loss": 0.1237, "step": 7280 }, { "epoch": 0.61, "learning_rate": 1.8775613033254957e-05, "loss": 0.1885, "step": 7290 }, { "epoch": 0.61, "learning_rate": 1.8773933490090695e-05, "loss": 0.1058, "step": 7300 }, { "epoch": 0.61, "learning_rate": 1.8772253946926437e-05, "loss": 0.1207, "step": 7310 }, { "epoch": 0.61, "learning_rate": 1.877057440376218e-05, "loss": 0.1148, "step": 7320 }, { "epoch": 0.62, "learning_rate": 1.876889486059792e-05, "loss": 0.1353, "step": 7330 }, { "epoch": 0.62, "learning_rate": 1.876721531743366e-05, "loss": 0.1341, "step": 7340 }, { "epoch": 0.62, "learning_rate": 1.8765535774269403e-05, "loss": 0.1935, "step": 7350 }, { "epoch": 0.62, "learning_rate": 1.876385623110514e-05, "loss": 0.1544, "step": 7360 }, { "epoch": 0.62, "learning_rate": 1.876217668794088e-05, "loss": 0.1048, "step": 7370 }, { "epoch": 0.62, "learning_rate": 1.8760497144776622e-05, "loss": 0.2349, "step": 7380 }, { "epoch": 0.62, "learning_rate": 1.875881760161236e-05, "loss": 0.1375, "step": 7390 }, { "epoch": 0.62, "learning_rate": 1.8757138058448103e-05, "loss": 0.1022, "step": 7400 }, { "epoch": 0.62, "learning_rate": 1.8755458515283846e-05, "loss": 0.1527, "step": 7410 }, { "epoch": 0.62, "learning_rate": 1.8753778972119584e-05, "loss": 0.1756, "step": 7420 }, { "epoch": 0.62, "learning_rate": 1.8752099428955327e-05, "loss": 0.1064, "step": 7430 }, { "epoch": 0.62, "learning_rate": 1.8750419885791065e-05, "loss": 0.14, "step": 7440 }, { "epoch": 0.63, "learning_rate": 1.8748740342626808e-05, "loss": 0.1507, "step": 7450 }, { "epoch": 0.63, "learning_rate": 1.874706079946255e-05, "loss": 0.2084, "step": 7460 }, { "epoch": 0.63, "learning_rate": 1.874538125629829e-05, "loss": 0.1473, "step": 7470 }, { "epoch": 0.63, "learning_rate": 1.8743701713134027e-05, "loss": 0.1939, "step": 7480 }, { "epoch": 0.63, "learning_rate": 1.874202216996977e-05, "loss": 0.0891, "step": 7490 }, { "epoch": 0.63, "learning_rate": 1.874034262680551e-05, "loss": 0.0829, "step": 7500 }, { "epoch": 0.63, "learning_rate": 1.873866308364125e-05, "loss": 0.1059, "step": 7510 }, { "epoch": 0.63, "learning_rate": 1.8736983540476993e-05, "loss": 0.214, "step": 7520 }, { "epoch": 0.63, "learning_rate": 1.873530399731273e-05, "loss": 0.179, "step": 7530 }, { "epoch": 0.63, "learning_rate": 1.8733624454148474e-05, "loss": 0.1355, "step": 7540 }, { "epoch": 0.63, "learning_rate": 1.8731944910984216e-05, "loss": 0.1064, "step": 7550 }, { "epoch": 0.63, "learning_rate": 1.8730265367819954e-05, "loss": 0.1287, "step": 7560 }, { "epoch": 0.64, "learning_rate": 1.8728585824655697e-05, "loss": 0.1124, "step": 7570 }, { "epoch": 0.64, "learning_rate": 1.8726906281491435e-05, "loss": 0.1454, "step": 7580 }, { "epoch": 0.64, "learning_rate": 1.8725226738327174e-05, "loss": 0.0791, "step": 7590 }, { "epoch": 0.64, "learning_rate": 1.8723547195162916e-05, "loss": 0.1629, "step": 7600 }, { "epoch": 0.64, "learning_rate": 1.872186765199866e-05, "loss": 0.1217, "step": 7610 }, { "epoch": 0.64, "learning_rate": 1.8720188108834397e-05, "loss": 0.1015, "step": 7620 }, { "epoch": 0.64, "learning_rate": 1.871850856567014e-05, "loss": 0.1152, "step": 7630 }, { "epoch": 0.64, "learning_rate": 1.871682902250588e-05, "loss": 0.2509, "step": 7640 }, { "epoch": 0.64, "learning_rate": 1.871514947934162e-05, "loss": 0.1506, "step": 7650 }, { "epoch": 0.64, "learning_rate": 1.8713469936177363e-05, "loss": 0.1578, "step": 7660 }, { "epoch": 0.64, "learning_rate": 1.87117903930131e-05, "loss": 0.0908, "step": 7670 }, { "epoch": 0.64, "learning_rate": 1.8710110849848844e-05, "loss": 0.0972, "step": 7680 }, { "epoch": 0.65, "learning_rate": 1.8708431306684582e-05, "loss": 0.2111, "step": 7690 }, { "epoch": 0.65, "learning_rate": 1.8706751763520325e-05, "loss": 0.0579, "step": 7700 }, { "epoch": 0.65, "learning_rate": 1.8705072220356063e-05, "loss": 0.1397, "step": 7710 }, { "epoch": 0.65, "learning_rate": 1.8703392677191805e-05, "loss": 0.0583, "step": 7720 }, { "epoch": 0.65, "learning_rate": 1.8701713134027548e-05, "loss": 0.0976, "step": 7730 }, { "epoch": 0.65, "learning_rate": 1.8700033590863286e-05, "loss": 0.1899, "step": 7740 }, { "epoch": 0.65, "learning_rate": 1.869835404769903e-05, "loss": 0.1328, "step": 7750 }, { "epoch": 0.65, "learning_rate": 1.8696674504534767e-05, "loss": 0.1943, "step": 7760 }, { "epoch": 0.65, "learning_rate": 1.869499496137051e-05, "loss": 0.1517, "step": 7770 }, { "epoch": 0.65, "learning_rate": 1.8693315418206252e-05, "loss": 0.0992, "step": 7780 }, { "epoch": 0.65, "learning_rate": 1.869163587504199e-05, "loss": 0.2316, "step": 7790 }, { "epoch": 0.66, "learning_rate": 1.868995633187773e-05, "loss": 0.0916, "step": 7800 }, { "epoch": 0.66, "learning_rate": 1.868827678871347e-05, "loss": 0.1473, "step": 7810 }, { "epoch": 0.66, "learning_rate": 1.868659724554921e-05, "loss": 0.1849, "step": 7820 }, { "epoch": 0.66, "learning_rate": 1.8684917702384952e-05, "loss": 0.1194, "step": 7830 }, { "epoch": 0.66, "learning_rate": 1.8683238159220695e-05, "loss": 0.2002, "step": 7840 }, { "epoch": 0.66, "learning_rate": 1.8681558616056433e-05, "loss": 0.0701, "step": 7850 }, { "epoch": 0.66, "learning_rate": 1.8679879072892176e-05, "loss": 0.1234, "step": 7860 }, { "epoch": 0.66, "learning_rate": 1.8678199529727918e-05, "loss": 0.2642, "step": 7870 }, { "epoch": 0.66, "learning_rate": 1.8676519986563656e-05, "loss": 0.1323, "step": 7880 }, { "epoch": 0.66, "learning_rate": 1.86748404433994e-05, "loss": 0.1519, "step": 7890 }, { "epoch": 0.66, "learning_rate": 1.8673160900235137e-05, "loss": 0.09, "step": 7900 }, { "epoch": 0.66, "learning_rate": 1.8671481357070876e-05, "loss": 0.1841, "step": 7910 }, { "epoch": 0.67, "learning_rate": 1.866980181390662e-05, "loss": 0.1359, "step": 7920 }, { "epoch": 0.67, "learning_rate": 1.866812227074236e-05, "loss": 0.124, "step": 7930 }, { "epoch": 0.67, "learning_rate": 1.86664427275781e-05, "loss": 0.1468, "step": 7940 }, { "epoch": 0.67, "learning_rate": 1.866476318441384e-05, "loss": 0.1581, "step": 7950 }, { "epoch": 0.67, "learning_rate": 1.866308364124958e-05, "loss": 0.2301, "step": 7960 }, { "epoch": 0.67, "learning_rate": 1.8661404098085322e-05, "loss": 0.1271, "step": 7970 }, { "epoch": 0.67, "learning_rate": 1.8659724554921065e-05, "loss": 0.1487, "step": 7980 }, { "epoch": 0.67, "learning_rate": 1.8658045011756803e-05, "loss": 0.1979, "step": 7990 }, { "epoch": 0.67, "learning_rate": 1.8656365468592542e-05, "loss": 0.2148, "step": 8000 }, { "epoch": 0.67, "learning_rate": 1.8654685925428284e-05, "loss": 0.1102, "step": 8010 }, { "epoch": 0.67, "learning_rate": 1.8653006382264027e-05, "loss": 0.1104, "step": 8020 }, { "epoch": 0.67, "learning_rate": 1.8651326839099765e-05, "loss": 0.1242, "step": 8030 }, { "epoch": 0.68, "learning_rate": 1.8649647295935508e-05, "loss": 0.1392, "step": 8040 }, { "epoch": 0.68, "learning_rate": 1.8647967752771246e-05, "loss": 0.11, "step": 8050 }, { "epoch": 0.68, "learning_rate": 1.864628820960699e-05, "loss": 0.1369, "step": 8060 }, { "epoch": 0.68, "learning_rate": 1.864460866644273e-05, "loss": 0.1146, "step": 8070 }, { "epoch": 0.68, "learning_rate": 1.864292912327847e-05, "loss": 0.1294, "step": 8080 }, { "epoch": 0.68, "learning_rate": 1.864124958011421e-05, "loss": 0.0961, "step": 8090 }, { "epoch": 0.68, "learning_rate": 1.8639570036949954e-05, "loss": 0.081, "step": 8100 }, { "epoch": 0.68, "learning_rate": 1.863789049378569e-05, "loss": 0.1947, "step": 8110 }, { "epoch": 0.68, "learning_rate": 1.863621095062143e-05, "loss": 0.1417, "step": 8120 }, { "epoch": 0.68, "learning_rate": 1.8634531407457173e-05, "loss": 0.178, "step": 8130 }, { "epoch": 0.68, "learning_rate": 1.8632851864292912e-05, "loss": 0.153, "step": 8140 }, { "epoch": 0.68, "learning_rate": 1.8631172321128654e-05, "loss": 0.1774, "step": 8150 }, { "epoch": 0.69, "learning_rate": 1.8629492777964397e-05, "loss": 0.1605, "step": 8160 }, { "epoch": 0.69, "learning_rate": 1.8627813234800135e-05, "loss": 0.1449, "step": 8170 }, { "epoch": 0.69, "learning_rate": 1.8626133691635878e-05, "loss": 0.1512, "step": 8180 }, { "epoch": 0.69, "learning_rate": 1.8624454148471616e-05, "loss": 0.0943, "step": 8190 }, { "epoch": 0.69, "learning_rate": 1.862277460530736e-05, "loss": 0.0932, "step": 8200 }, { "epoch": 0.69, "learning_rate": 1.86210950621431e-05, "loss": 0.1491, "step": 8210 }, { "epoch": 0.69, "learning_rate": 1.861941551897884e-05, "loss": 0.092, "step": 8220 }, { "epoch": 0.69, "learning_rate": 1.8617735975814578e-05, "loss": 0.2053, "step": 8230 }, { "epoch": 0.69, "learning_rate": 1.861605643265032e-05, "loss": 0.1246, "step": 8240 }, { "epoch": 0.69, "learning_rate": 1.8614376889486063e-05, "loss": 0.1737, "step": 8250 }, { "epoch": 0.69, "learning_rate": 1.86126973463218e-05, "loss": 0.0925, "step": 8260 }, { "epoch": 0.69, "learning_rate": 1.8611017803157544e-05, "loss": 0.2299, "step": 8270 }, { "epoch": 0.7, "learning_rate": 1.8609338259993282e-05, "loss": 0.2214, "step": 8280 }, { "epoch": 0.7, "learning_rate": 1.8607658716829025e-05, "loss": 0.1163, "step": 8290 }, { "epoch": 0.7, "learning_rate": 1.8605979173664767e-05, "loss": 0.1365, "step": 8300 }, { "epoch": 0.7, "learning_rate": 1.8604299630500505e-05, "loss": 0.182, "step": 8310 }, { "epoch": 0.7, "learning_rate": 1.8602620087336244e-05, "loss": 0.1681, "step": 8320 }, { "epoch": 0.7, "learning_rate": 1.8600940544171986e-05, "loss": 0.1392, "step": 8330 }, { "epoch": 0.7, "learning_rate": 1.8599261001007725e-05, "loss": 0.1243, "step": 8340 }, { "epoch": 0.7, "learning_rate": 1.8597581457843467e-05, "loss": 0.1147, "step": 8350 }, { "epoch": 0.7, "learning_rate": 1.859590191467921e-05, "loss": 0.06, "step": 8360 }, { "epoch": 0.7, "learning_rate": 1.859422237151495e-05, "loss": 0.1372, "step": 8370 }, { "epoch": 0.7, "learning_rate": 1.859254282835069e-05, "loss": 0.1668, "step": 8380 }, { "epoch": 0.7, "learning_rate": 1.8590863285186433e-05, "loss": 0.1884, "step": 8390 }, { "epoch": 0.71, "learning_rate": 1.858918374202217e-05, "loss": 0.1349, "step": 8400 }, { "epoch": 0.71, "learning_rate": 1.8587504198857914e-05, "loss": 0.0543, "step": 8410 }, { "epoch": 0.71, "learning_rate": 1.8585824655693652e-05, "loss": 0.1833, "step": 8420 }, { "epoch": 0.71, "learning_rate": 1.858414511252939e-05, "loss": 0.0887, "step": 8430 }, { "epoch": 0.71, "learning_rate": 1.8582465569365133e-05, "loss": 0.0997, "step": 8440 }, { "epoch": 0.71, "learning_rate": 1.8580786026200876e-05, "loss": 0.1276, "step": 8450 }, { "epoch": 0.71, "learning_rate": 1.8579106483036614e-05, "loss": 0.1775, "step": 8460 }, { "epoch": 0.71, "learning_rate": 1.8577426939872356e-05, "loss": 0.0249, "step": 8470 }, { "epoch": 0.71, "learning_rate": 1.85757473967081e-05, "loss": 0.3213, "step": 8480 }, { "epoch": 0.71, "learning_rate": 1.8574067853543837e-05, "loss": 0.2213, "step": 8490 }, { "epoch": 0.71, "learning_rate": 1.857238831037958e-05, "loss": 0.0578, "step": 8500 }, { "epoch": 0.71, "learning_rate": 1.857070876721532e-05, "loss": 0.1598, "step": 8510 }, { "epoch": 0.72, "learning_rate": 1.856902922405106e-05, "loss": 0.1539, "step": 8520 }, { "epoch": 0.72, "learning_rate": 1.85673496808868e-05, "loss": 0.2029, "step": 8530 }, { "epoch": 0.72, "learning_rate": 1.856567013772254e-05, "loss": 0.1245, "step": 8540 }, { "epoch": 0.72, "learning_rate": 1.856399059455828e-05, "loss": 0.1523, "step": 8550 }, { "epoch": 0.72, "learning_rate": 1.8562311051394022e-05, "loss": 0.1677, "step": 8560 }, { "epoch": 0.72, "learning_rate": 1.856063150822976e-05, "loss": 0.2043, "step": 8570 }, { "epoch": 0.72, "learning_rate": 1.8558951965065503e-05, "loss": 0.127, "step": 8580 }, { "epoch": 0.72, "learning_rate": 1.8557272421901246e-05, "loss": 0.1244, "step": 8590 }, { "epoch": 0.72, "learning_rate": 1.8555592878736984e-05, "loss": 0.135, "step": 8600 }, { "epoch": 0.72, "learning_rate": 1.8553913335572727e-05, "loss": 0.1162, "step": 8610 }, { "epoch": 0.72, "learning_rate": 1.855223379240847e-05, "loss": 0.1466, "step": 8620 }, { "epoch": 0.72, "learning_rate": 1.8550554249244207e-05, "loss": 0.2166, "step": 8630 }, { "epoch": 0.73, "learning_rate": 1.8548874706079946e-05, "loss": 0.0917, "step": 8640 }, { "epoch": 0.73, "learning_rate": 1.854719516291569e-05, "loss": 0.17, "step": 8650 }, { "epoch": 0.73, "learning_rate": 1.8545515619751427e-05, "loss": 0.243, "step": 8660 }, { "epoch": 0.73, "learning_rate": 1.854383607658717e-05, "loss": 0.1357, "step": 8670 }, { "epoch": 0.73, "learning_rate": 1.854215653342291e-05, "loss": 0.1381, "step": 8680 }, { "epoch": 0.73, "learning_rate": 1.854047699025865e-05, "loss": 0.2444, "step": 8690 }, { "epoch": 0.73, "learning_rate": 1.8538797447094393e-05, "loss": 0.1197, "step": 8700 }, { "epoch": 0.73, "learning_rate": 1.853711790393013e-05, "loss": 0.0971, "step": 8710 }, { "epoch": 0.73, "learning_rate": 1.8535438360765873e-05, "loss": 0.1406, "step": 8720 }, { "epoch": 0.73, "learning_rate": 1.8533758817601616e-05, "loss": 0.1282, "step": 8730 }, { "epoch": 0.73, "learning_rate": 1.8532079274437354e-05, "loss": 0.1994, "step": 8740 }, { "epoch": 0.73, "learning_rate": 1.8530399731273093e-05, "loss": 0.1797, "step": 8750 }, { "epoch": 0.74, "learning_rate": 1.8528720188108835e-05, "loss": 0.132, "step": 8760 }, { "epoch": 0.74, "learning_rate": 1.8527040644944578e-05, "loss": 0.1197, "step": 8770 }, { "epoch": 0.74, "learning_rate": 1.8525361101780316e-05, "loss": 0.1787, "step": 8780 }, { "epoch": 0.74, "learning_rate": 1.852368155861606e-05, "loss": 0.0802, "step": 8790 }, { "epoch": 0.74, "learning_rate": 1.8522002015451797e-05, "loss": 0.2523, "step": 8800 }, { "epoch": 0.74, "learning_rate": 1.852032247228754e-05, "loss": 0.1148, "step": 8810 }, { "epoch": 0.74, "learning_rate": 1.851864292912328e-05, "loss": 0.1565, "step": 8820 }, { "epoch": 0.74, "learning_rate": 1.851696338595902e-05, "loss": 0.1312, "step": 8830 }, { "epoch": 0.74, "learning_rate": 1.8515283842794763e-05, "loss": 0.1371, "step": 8840 }, { "epoch": 0.74, "learning_rate": 1.85136042996305e-05, "loss": 0.1005, "step": 8850 }, { "epoch": 0.74, "learning_rate": 1.8511924756466244e-05, "loss": 0.1004, "step": 8860 }, { "epoch": 0.74, "learning_rate": 1.8510245213301982e-05, "loss": 0.2107, "step": 8870 }, { "epoch": 0.75, "learning_rate": 1.8508565670137724e-05, "loss": 0.2135, "step": 8880 }, { "epoch": 0.75, "learning_rate": 1.8506886126973463e-05, "loss": 0.1368, "step": 8890 }, { "epoch": 0.75, "learning_rate": 1.8505206583809205e-05, "loss": 0.1117, "step": 8900 }, { "epoch": 0.75, "learning_rate": 1.8503527040644948e-05, "loss": 0.086, "step": 8910 }, { "epoch": 0.75, "learning_rate": 1.8501847497480686e-05, "loss": 0.1644, "step": 8920 }, { "epoch": 0.75, "learning_rate": 1.850016795431643e-05, "loss": 0.1447, "step": 8930 }, { "epoch": 0.75, "learning_rate": 1.8498488411152167e-05, "loss": 0.1435, "step": 8940 }, { "epoch": 0.75, "learning_rate": 1.849680886798791e-05, "loss": 0.1247, "step": 8950 }, { "epoch": 0.75, "learning_rate": 1.8495129324823648e-05, "loss": 0.2442, "step": 8960 }, { "epoch": 0.75, "learning_rate": 1.849344978165939e-05, "loss": 0.181, "step": 8970 }, { "epoch": 0.75, "learning_rate": 1.849177023849513e-05, "loss": 0.2208, "step": 8980 }, { "epoch": 0.75, "learning_rate": 1.849009069533087e-05, "loss": 0.1799, "step": 8990 }, { "epoch": 0.76, "learning_rate": 1.8488411152166614e-05, "loss": 0.077, "step": 9000 }, { "epoch": 0.76, "learning_rate": 1.8486731609002352e-05, "loss": 0.0979, "step": 9010 }, { "epoch": 0.76, "learning_rate": 1.8485052065838095e-05, "loss": 0.1567, "step": 9020 }, { "epoch": 0.76, "learning_rate": 1.8483372522673833e-05, "loss": 0.1616, "step": 9030 }, { "epoch": 0.76, "learning_rate": 1.8481692979509576e-05, "loss": 0.2019, "step": 9040 }, { "epoch": 0.76, "learning_rate": 1.8480013436345318e-05, "loss": 0.1262, "step": 9050 }, { "epoch": 0.76, "learning_rate": 1.8478333893181056e-05, "loss": 0.1096, "step": 9060 }, { "epoch": 0.76, "learning_rate": 1.8476654350016795e-05, "loss": 0.1507, "step": 9070 }, { "epoch": 0.76, "learning_rate": 1.8474974806852537e-05, "loss": 0.1831, "step": 9080 }, { "epoch": 0.76, "learning_rate": 1.8473295263688276e-05, "loss": 0.0688, "step": 9090 }, { "epoch": 0.76, "learning_rate": 1.847161572052402e-05, "loss": 0.1402, "step": 9100 }, { "epoch": 0.77, "learning_rate": 1.846993617735976e-05, "loss": 0.1111, "step": 9110 }, { "epoch": 0.77, "learning_rate": 1.84682566341955e-05, "loss": 0.1006, "step": 9120 }, { "epoch": 0.77, "learning_rate": 1.846657709103124e-05, "loss": 0.222, "step": 9130 }, { "epoch": 0.77, "learning_rate": 1.8464897547866984e-05, "loss": 0.123, "step": 9140 }, { "epoch": 0.77, "learning_rate": 1.8463218004702722e-05, "loss": 0.0943, "step": 9150 }, { "epoch": 0.77, "learning_rate": 1.8461538461538465e-05, "loss": 0.0773, "step": 9160 }, { "epoch": 0.77, "learning_rate": 1.8459858918374203e-05, "loss": 0.111, "step": 9170 }, { "epoch": 0.77, "learning_rate": 1.8458179375209942e-05, "loss": 0.0766, "step": 9180 }, { "epoch": 0.77, "learning_rate": 1.8456499832045684e-05, "loss": 0.1609, "step": 9190 }, { "epoch": 0.77, "learning_rate": 1.8454820288881427e-05, "loss": 0.133, "step": 9200 }, { "epoch": 0.77, "learning_rate": 1.8453140745717165e-05, "loss": 0.1822, "step": 9210 }, { "epoch": 0.77, "learning_rate": 1.8451461202552907e-05, "loss": 0.134, "step": 9220 }, { "epoch": 0.78, "learning_rate": 1.844978165938865e-05, "loss": 0.1854, "step": 9230 }, { "epoch": 0.78, "learning_rate": 1.844810211622439e-05, "loss": 0.1167, "step": 9240 }, { "epoch": 0.78, "learning_rate": 1.844642257306013e-05, "loss": 0.075, "step": 9250 }, { "epoch": 0.78, "learning_rate": 1.844474302989587e-05, "loss": 0.1819, "step": 9260 }, { "epoch": 0.78, "learning_rate": 1.844306348673161e-05, "loss": 0.1258, "step": 9270 }, { "epoch": 0.78, "learning_rate": 1.844138394356735e-05, "loss": 0.0921, "step": 9280 }, { "epoch": 0.78, "learning_rate": 1.8439704400403093e-05, "loss": 0.0988, "step": 9290 }, { "epoch": 0.78, "learning_rate": 1.843802485723883e-05, "loss": 0.1266, "step": 9300 }, { "epoch": 0.78, "learning_rate": 1.8436345314074573e-05, "loss": 0.1382, "step": 9310 }, { "epoch": 0.78, "learning_rate": 1.8434665770910312e-05, "loss": 0.1859, "step": 9320 }, { "epoch": 0.78, "learning_rate": 1.8432986227746054e-05, "loss": 0.1096, "step": 9330 }, { "epoch": 0.78, "learning_rate": 1.8431306684581797e-05, "loss": 0.0952, "step": 9340 }, { "epoch": 0.79, "learning_rate": 1.8429627141417535e-05, "loss": 0.0486, "step": 9350 }, { "epoch": 0.79, "learning_rate": 1.8427947598253278e-05, "loss": 0.2032, "step": 9360 }, { "epoch": 0.79, "learning_rate": 1.842626805508902e-05, "loss": 0.0899, "step": 9370 }, { "epoch": 0.79, "learning_rate": 1.842458851192476e-05, "loss": 0.1583, "step": 9380 }, { "epoch": 0.79, "learning_rate": 1.8422908968760497e-05, "loss": 0.1187, "step": 9390 }, { "epoch": 0.79, "learning_rate": 1.842122942559624e-05, "loss": 0.1432, "step": 9400 }, { "epoch": 0.79, "learning_rate": 1.8419549882431978e-05, "loss": 0.1375, "step": 9410 }, { "epoch": 0.79, "learning_rate": 1.841787033926772e-05, "loss": 0.1579, "step": 9420 }, { "epoch": 0.79, "learning_rate": 1.8416190796103463e-05, "loss": 0.1554, "step": 9430 }, { "epoch": 0.79, "learning_rate": 1.84145112529392e-05, "loss": 0.1399, "step": 9440 }, { "epoch": 0.79, "learning_rate": 1.8412831709774944e-05, "loss": 0.1784, "step": 9450 }, { "epoch": 0.79, "learning_rate": 1.8411152166610686e-05, "loss": 0.139, "step": 9460 }, { "epoch": 0.8, "learning_rate": 1.8409472623446424e-05, "loss": 0.1504, "step": 9470 }, { "epoch": 0.8, "learning_rate": 1.8407793080282167e-05, "loss": 0.1923, "step": 9480 }, { "epoch": 0.8, "learning_rate": 1.8406113537117905e-05, "loss": 0.0926, "step": 9490 }, { "epoch": 0.8, "learning_rate": 1.8404433993953644e-05, "loss": 0.1056, "step": 9500 }, { "epoch": 0.8, "learning_rate": 1.8402754450789386e-05, "loss": 0.1045, "step": 9510 }, { "epoch": 0.8, "learning_rate": 1.840107490762513e-05, "loss": 0.1847, "step": 9520 }, { "epoch": 0.8, "learning_rate": 1.8399395364460867e-05, "loss": 0.2008, "step": 9530 }, { "epoch": 0.8, "learning_rate": 1.839771582129661e-05, "loss": 0.1312, "step": 9540 }, { "epoch": 0.8, "learning_rate": 1.8396036278132348e-05, "loss": 0.1866, "step": 9550 }, { "epoch": 0.8, "learning_rate": 1.839435673496809e-05, "loss": 0.1224, "step": 9560 }, { "epoch": 0.8, "learning_rate": 1.8392677191803833e-05, "loss": 0.1974, "step": 9570 }, { "epoch": 0.8, "learning_rate": 1.839099764863957e-05, "loss": 0.2284, "step": 9580 }, { "epoch": 0.81, "learning_rate": 1.838931810547531e-05, "loss": 0.1504, "step": 9590 }, { "epoch": 0.81, "learning_rate": 1.8387638562311052e-05, "loss": 0.1318, "step": 9600 }, { "epoch": 0.81, "learning_rate": 1.8385959019146795e-05, "loss": 0.1518, "step": 9610 }, { "epoch": 0.81, "learning_rate": 1.8384279475982533e-05, "loss": 0.2138, "step": 9620 }, { "epoch": 0.81, "learning_rate": 1.8382599932818275e-05, "loss": 0.1471, "step": 9630 }, { "epoch": 0.81, "learning_rate": 1.8380920389654014e-05, "loss": 0.1238, "step": 9640 }, { "epoch": 0.81, "learning_rate": 1.8379240846489756e-05, "loss": 0.1493, "step": 9650 }, { "epoch": 0.81, "learning_rate": 1.83775613033255e-05, "loss": 0.1227, "step": 9660 }, { "epoch": 0.81, "learning_rate": 1.8375881760161237e-05, "loss": 0.1936, "step": 9670 }, { "epoch": 0.81, "learning_rate": 1.837420221699698e-05, "loss": 0.1609, "step": 9680 }, { "epoch": 0.81, "learning_rate": 1.837252267383272e-05, "loss": 0.1806, "step": 9690 }, { "epoch": 0.81, "learning_rate": 1.8370843130668457e-05, "loss": 0.0989, "step": 9700 }, { "epoch": 0.82, "learning_rate": 1.83691635875042e-05, "loss": 0.1395, "step": 9710 }, { "epoch": 0.82, "learning_rate": 1.836748404433994e-05, "loss": 0.139, "step": 9720 }, { "epoch": 0.82, "learning_rate": 1.836580450117568e-05, "loss": 0.1722, "step": 9730 }, { "epoch": 0.82, "learning_rate": 1.8364124958011422e-05, "loss": 0.0669, "step": 9740 }, { "epoch": 0.82, "learning_rate": 1.8362445414847165e-05, "loss": 0.1569, "step": 9750 }, { "epoch": 0.82, "learning_rate": 1.8360765871682903e-05, "loss": 0.0871, "step": 9760 }, { "epoch": 0.82, "learning_rate": 1.8359086328518646e-05, "loss": 0.2036, "step": 9770 }, { "epoch": 0.82, "learning_rate": 1.8357406785354384e-05, "loss": 0.1336, "step": 9780 }, { "epoch": 0.82, "learning_rate": 1.8355727242190127e-05, "loss": 0.2762, "step": 9790 }, { "epoch": 0.82, "learning_rate": 1.8354047699025865e-05, "loss": 0.1271, "step": 9800 }, { "epoch": 0.82, "learning_rate": 1.8352368155861607e-05, "loss": 0.1464, "step": 9810 }, { "epoch": 0.82, "learning_rate": 1.8350688612697346e-05, "loss": 0.1261, "step": 9820 }, { "epoch": 0.83, "learning_rate": 1.834900906953309e-05, "loss": 0.2157, "step": 9830 }, { "epoch": 0.83, "learning_rate": 1.834732952636883e-05, "loss": 0.1777, "step": 9840 }, { "epoch": 0.83, "learning_rate": 1.834564998320457e-05, "loss": 0.1306, "step": 9850 }, { "epoch": 0.83, "learning_rate": 1.834397044004031e-05, "loss": 0.1014, "step": 9860 }, { "epoch": 0.83, "learning_rate": 1.834229089687605e-05, "loss": 0.1005, "step": 9870 }, { "epoch": 0.83, "learning_rate": 1.8340611353711792e-05, "loss": 0.1733, "step": 9880 }, { "epoch": 0.83, "learning_rate": 1.8338931810547535e-05, "loss": 0.1352, "step": 9890 }, { "epoch": 0.83, "learning_rate": 1.8337252267383273e-05, "loss": 0.1013, "step": 9900 }, { "epoch": 0.83, "learning_rate": 1.8335572724219012e-05, "loss": 0.1708, "step": 9910 }, { "epoch": 0.83, "learning_rate": 1.8333893181054754e-05, "loss": 0.1421, "step": 9920 }, { "epoch": 0.83, "learning_rate": 1.8332213637890493e-05, "loss": 0.1036, "step": 9930 }, { "epoch": 0.83, "learning_rate": 1.8330534094726235e-05, "loss": 0.1068, "step": 9940 }, { "epoch": 0.84, "learning_rate": 1.8328854551561978e-05, "loss": 0.103, "step": 9950 }, { "epoch": 0.84, "learning_rate": 1.8327175008397716e-05, "loss": 0.1402, "step": 9960 }, { "epoch": 0.84, "learning_rate": 1.832549546523346e-05, "loss": 0.1083, "step": 9970 }, { "epoch": 0.84, "learning_rate": 1.83238159220692e-05, "loss": 0.102, "step": 9980 }, { "epoch": 0.84, "learning_rate": 1.832213637890494e-05, "loss": 0.1339, "step": 9990 }, { "epoch": 0.84, "learning_rate": 1.832045683574068e-05, "loss": 0.1786, "step": 10000 }, { "epoch": 0.84, "learning_rate": 1.831877729257642e-05, "loss": 0.2547, "step": 10010 }, { "epoch": 0.84, "learning_rate": 1.831709774941216e-05, "loss": 0.1532, "step": 10020 }, { "epoch": 0.84, "learning_rate": 1.83154182062479e-05, "loss": 0.2012, "step": 10030 }, { "epoch": 0.84, "learning_rate": 1.8313738663083643e-05, "loss": 0.1916, "step": 10040 }, { "epoch": 0.84, "learning_rate": 1.8312059119919382e-05, "loss": 0.1216, "step": 10050 }, { "epoch": 0.84, "learning_rate": 1.8310379576755124e-05, "loss": 0.0926, "step": 10060 }, { "epoch": 0.85, "learning_rate": 1.8308700033590863e-05, "loss": 0.1791, "step": 10070 }, { "epoch": 0.85, "learning_rate": 1.8307020490426605e-05, "loss": 0.1666, "step": 10080 }, { "epoch": 0.85, "learning_rate": 1.8305340947262348e-05, "loss": 0.0791, "step": 10090 }, { "epoch": 0.85, "learning_rate": 1.8303661404098086e-05, "loss": 0.1383, "step": 10100 }, { "epoch": 0.85, "learning_rate": 1.830198186093383e-05, "loss": 0.1534, "step": 10110 }, { "epoch": 0.85, "learning_rate": 1.8300302317769567e-05, "loss": 0.1227, "step": 10120 }, { "epoch": 0.85, "learning_rate": 1.829862277460531e-05, "loss": 0.1857, "step": 10130 }, { "epoch": 0.85, "learning_rate": 1.8296943231441048e-05, "loss": 0.1138, "step": 10140 }, { "epoch": 0.85, "learning_rate": 1.829526368827679e-05, "loss": 0.166, "step": 10150 }, { "epoch": 0.85, "learning_rate": 1.829358414511253e-05, "loss": 0.1169, "step": 10160 }, { "epoch": 0.85, "learning_rate": 1.829190460194827e-05, "loss": 0.1267, "step": 10170 }, { "epoch": 0.85, "learning_rate": 1.8290225058784014e-05, "loss": 0.1673, "step": 10180 }, { "epoch": 0.86, "learning_rate": 1.8288545515619752e-05, "loss": 0.1138, "step": 10190 }, { "epoch": 0.86, "learning_rate": 1.8286865972455495e-05, "loss": 0.0655, "step": 10200 }, { "epoch": 0.86, "learning_rate": 1.8285186429291237e-05, "loss": 0.1788, "step": 10210 }, { "epoch": 0.86, "learning_rate": 1.8283506886126975e-05, "loss": 0.094, "step": 10220 }, { "epoch": 0.86, "learning_rate": 1.8281827342962714e-05, "loss": 0.099, "step": 10230 }, { "epoch": 0.86, "learning_rate": 1.8280147799798456e-05, "loss": 0.1104, "step": 10240 }, { "epoch": 0.86, "learning_rate": 1.8278468256634195e-05, "loss": 0.1554, "step": 10250 }, { "epoch": 0.86, "learning_rate": 1.8276788713469937e-05, "loss": 0.1221, "step": 10260 }, { "epoch": 0.86, "learning_rate": 1.827510917030568e-05, "loss": 0.1436, "step": 10270 }, { "epoch": 0.86, "learning_rate": 1.827342962714142e-05, "loss": 0.0861, "step": 10280 }, { "epoch": 0.86, "learning_rate": 1.827175008397716e-05, "loss": 0.1111, "step": 10290 }, { "epoch": 0.86, "learning_rate": 1.82700705408129e-05, "loss": 0.1082, "step": 10300 }, { "epoch": 0.87, "learning_rate": 1.826839099764864e-05, "loss": 0.1855, "step": 10310 }, { "epoch": 0.87, "learning_rate": 1.8266711454484384e-05, "loss": 0.1529, "step": 10320 }, { "epoch": 0.87, "learning_rate": 1.8265031911320122e-05, "loss": 0.0959, "step": 10330 }, { "epoch": 0.87, "learning_rate": 1.826335236815586e-05, "loss": 0.158, "step": 10340 }, { "epoch": 0.87, "learning_rate": 1.8261672824991603e-05, "loss": 0.1552, "step": 10350 }, { "epoch": 0.87, "learning_rate": 1.8259993281827346e-05, "loss": 0.073, "step": 10360 }, { "epoch": 0.87, "learning_rate": 1.8258313738663084e-05, "loss": 0.0741, "step": 10370 }, { "epoch": 0.87, "learning_rate": 1.8256634195498826e-05, "loss": 0.1412, "step": 10380 }, { "epoch": 0.87, "learning_rate": 1.8254954652334565e-05, "loss": 0.1726, "step": 10390 }, { "epoch": 0.87, "learning_rate": 1.8253275109170307e-05, "loss": 0.218, "step": 10400 }, { "epoch": 0.87, "learning_rate": 1.825159556600605e-05, "loss": 0.1649, "step": 10410 }, { "epoch": 0.88, "learning_rate": 1.824991602284179e-05, "loss": 0.141, "step": 10420 }, { "epoch": 0.88, "learning_rate": 1.824823647967753e-05, "loss": 0.1362, "step": 10430 }, { "epoch": 0.88, "learning_rate": 1.824655693651327e-05, "loss": 0.1305, "step": 10440 }, { "epoch": 0.88, "learning_rate": 1.8244877393349008e-05, "loss": 0.0918, "step": 10450 }, { "epoch": 0.88, "learning_rate": 1.824319785018475e-05, "loss": 0.1115, "step": 10460 }, { "epoch": 0.88, "learning_rate": 1.8241518307020492e-05, "loss": 0.1856, "step": 10470 }, { "epoch": 0.88, "learning_rate": 1.823983876385623e-05, "loss": 0.1273, "step": 10480 }, { "epoch": 0.88, "learning_rate": 1.8238159220691973e-05, "loss": 0.0314, "step": 10490 }, { "epoch": 0.88, "learning_rate": 1.8236479677527716e-05, "loss": 0.1344, "step": 10500 }, { "epoch": 0.88, "learning_rate": 1.8234800134363454e-05, "loss": 0.17, "step": 10510 }, { "epoch": 0.88, "learning_rate": 1.8233120591199197e-05, "loss": 0.1636, "step": 10520 }, { "epoch": 0.88, "learning_rate": 1.8231441048034935e-05, "loss": 0.0869, "step": 10530 }, { "epoch": 0.89, "learning_rate": 1.8229761504870677e-05, "loss": 0.2065, "step": 10540 }, { "epoch": 0.89, "learning_rate": 1.8228081961706416e-05, "loss": 0.0795, "step": 10550 }, { "epoch": 0.89, "learning_rate": 1.822640241854216e-05, "loss": 0.1446, "step": 10560 }, { "epoch": 0.89, "learning_rate": 1.8224722875377897e-05, "loss": 0.1465, "step": 10570 }, { "epoch": 0.89, "learning_rate": 1.822304333221364e-05, "loss": 0.0663, "step": 10580 }, { "epoch": 0.89, "learning_rate": 1.822136378904938e-05, "loss": 0.0815, "step": 10590 }, { "epoch": 0.89, "learning_rate": 1.821968424588512e-05, "loss": 0.2696, "step": 10600 }, { "epoch": 0.89, "learning_rate": 1.8218004702720863e-05, "loss": 0.2116, "step": 10610 }, { "epoch": 0.89, "learning_rate": 1.82163251595566e-05, "loss": 0.1409, "step": 10620 }, { "epoch": 0.89, "learning_rate": 1.8214645616392343e-05, "loss": 0.0958, "step": 10630 }, { "epoch": 0.89, "learning_rate": 1.8212966073228086e-05, "loss": 0.1682, "step": 10640 }, { "epoch": 0.89, "learning_rate": 1.8211286530063824e-05, "loss": 0.1, "step": 10650 }, { "epoch": 0.9, "learning_rate": 1.8209606986899563e-05, "loss": 0.1701, "step": 10660 }, { "epoch": 0.9, "learning_rate": 1.8207927443735305e-05, "loss": 0.1569, "step": 10670 }, { "epoch": 0.9, "learning_rate": 1.8206247900571044e-05, "loss": 0.1526, "step": 10680 }, { "epoch": 0.9, "learning_rate": 1.8204568357406786e-05, "loss": 0.1619, "step": 10690 }, { "epoch": 0.9, "learning_rate": 1.820288881424253e-05, "loss": 0.1718, "step": 10700 }, { "epoch": 0.9, "learning_rate": 1.8201209271078267e-05, "loss": 0.1337, "step": 10710 }, { "epoch": 0.9, "learning_rate": 1.819952972791401e-05, "loss": 0.122, "step": 10720 }, { "epoch": 0.9, "learning_rate": 1.819785018474975e-05, "loss": 0.1953, "step": 10730 }, { "epoch": 0.9, "learning_rate": 1.819617064158549e-05, "loss": 0.2464, "step": 10740 }, { "epoch": 0.9, "learning_rate": 1.8194491098421233e-05, "loss": 0.2193, "step": 10750 }, { "epoch": 0.9, "learning_rate": 1.819281155525697e-05, "loss": 0.1814, "step": 10760 }, { "epoch": 0.9, "learning_rate": 1.819113201209271e-05, "loss": 0.1094, "step": 10770 }, { "epoch": 0.91, "learning_rate": 1.8189452468928452e-05, "loss": 0.1054, "step": 10780 }, { "epoch": 0.91, "learning_rate": 1.8187772925764194e-05, "loss": 0.0902, "step": 10790 }, { "epoch": 0.91, "learning_rate": 1.8186093382599933e-05, "loss": 0.1339, "step": 10800 }, { "epoch": 0.91, "learning_rate": 1.8184413839435675e-05, "loss": 0.0851, "step": 10810 }, { "epoch": 0.91, "learning_rate": 1.8182734296271418e-05, "loss": 0.3053, "step": 10820 }, { "epoch": 0.91, "learning_rate": 1.8181054753107156e-05, "loss": 0.1048, "step": 10830 }, { "epoch": 0.91, "learning_rate": 1.81793752099429e-05, "loss": 0.1427, "step": 10840 }, { "epoch": 0.91, "learning_rate": 1.8177695666778637e-05, "loss": 0.1211, "step": 10850 }, { "epoch": 0.91, "learning_rate": 1.8176016123614376e-05, "loss": 0.1603, "step": 10860 }, { "epoch": 0.91, "learning_rate": 1.817433658045012e-05, "loss": 0.1068, "step": 10870 }, { "epoch": 0.91, "learning_rate": 1.817265703728586e-05, "loss": 0.1175, "step": 10880 }, { "epoch": 0.91, "learning_rate": 1.81709774941216e-05, "loss": 0.1153, "step": 10890 }, { "epoch": 0.92, "learning_rate": 1.816929795095734e-05, "loss": 0.1896, "step": 10900 }, { "epoch": 0.92, "learning_rate": 1.816761840779308e-05, "loss": 0.1246, "step": 10910 }, { "epoch": 0.92, "learning_rate": 1.8165938864628822e-05, "loss": 0.1437, "step": 10920 }, { "epoch": 0.92, "learning_rate": 1.8164259321464565e-05, "loss": 0.0694, "step": 10930 }, { "epoch": 0.92, "learning_rate": 1.8162579778300303e-05, "loss": 0.091, "step": 10940 }, { "epoch": 0.92, "learning_rate": 1.8160900235136046e-05, "loss": 0.1666, "step": 10950 }, { "epoch": 0.92, "learning_rate": 1.8159220691971788e-05, "loss": 0.1063, "step": 10960 }, { "epoch": 0.92, "learning_rate": 1.8157541148807526e-05, "loss": 0.088, "step": 10970 }, { "epoch": 0.92, "learning_rate": 1.8155861605643265e-05, "loss": 0.1132, "step": 10980 }, { "epoch": 0.92, "learning_rate": 1.8154182062479007e-05, "loss": 0.1007, "step": 10990 }, { "epoch": 0.92, "learning_rate": 1.8152502519314746e-05, "loss": 0.15, "step": 11000 }, { "epoch": 0.92, "learning_rate": 1.815082297615049e-05, "loss": 0.1503, "step": 11010 }, { "epoch": 0.93, "learning_rate": 1.814914343298623e-05, "loss": 0.1282, "step": 11020 }, { "epoch": 0.93, "learning_rate": 1.814746388982197e-05, "loss": 0.1051, "step": 11030 }, { "epoch": 0.93, "learning_rate": 1.814578434665771e-05, "loss": 0.1247, "step": 11040 }, { "epoch": 0.93, "learning_rate": 1.814410480349345e-05, "loss": 0.1113, "step": 11050 }, { "epoch": 0.93, "learning_rate": 1.8142425260329192e-05, "loss": 0.1394, "step": 11060 }, { "epoch": 0.93, "learning_rate": 1.8140745717164935e-05, "loss": 0.1634, "step": 11070 }, { "epoch": 0.93, "learning_rate": 1.8139066174000673e-05, "loss": 0.0374, "step": 11080 }, { "epoch": 0.93, "learning_rate": 1.8137386630836412e-05, "loss": 0.1541, "step": 11090 }, { "epoch": 0.93, "learning_rate": 1.8135707087672154e-05, "loss": 0.1169, "step": 11100 }, { "epoch": 0.93, "learning_rate": 1.8134027544507897e-05, "loss": 0.1322, "step": 11110 }, { "epoch": 0.93, "learning_rate": 1.8132348001343635e-05, "loss": 0.1315, "step": 11120 }, { "epoch": 0.93, "learning_rate": 1.8130668458179377e-05, "loss": 0.119, "step": 11130 }, { "epoch": 0.94, "learning_rate": 1.8128988915015116e-05, "loss": 0.1275, "step": 11140 }, { "epoch": 0.94, "learning_rate": 1.812730937185086e-05, "loss": 0.1858, "step": 11150 }, { "epoch": 0.94, "learning_rate": 1.81256298286866e-05, "loss": 0.1291, "step": 11160 }, { "epoch": 0.94, "learning_rate": 1.812395028552234e-05, "loss": 0.1027, "step": 11170 }, { "epoch": 0.94, "learning_rate": 1.8122270742358078e-05, "loss": 0.0965, "step": 11180 }, { "epoch": 0.94, "learning_rate": 1.812059119919382e-05, "loss": 0.097, "step": 11190 }, { "epoch": 0.94, "learning_rate": 1.811891165602956e-05, "loss": 0.1918, "step": 11200 }, { "epoch": 0.94, "learning_rate": 1.81172321128653e-05, "loss": 0.2049, "step": 11210 }, { "epoch": 0.94, "learning_rate": 1.8115552569701043e-05, "loss": 0.1008, "step": 11220 }, { "epoch": 0.94, "learning_rate": 1.8113873026536782e-05, "loss": 0.102, "step": 11230 }, { "epoch": 0.94, "learning_rate": 1.8112193483372524e-05, "loss": 0.301, "step": 11240 }, { "epoch": 0.94, "learning_rate": 1.8110513940208267e-05, "loss": 0.145, "step": 11250 }, { "epoch": 0.95, "learning_rate": 1.8108834397044005e-05, "loss": 0.1314, "step": 11260 }, { "epoch": 0.95, "learning_rate": 1.8107154853879748e-05, "loss": 0.1062, "step": 11270 }, { "epoch": 0.95, "learning_rate": 1.8105475310715486e-05, "loss": 0.1212, "step": 11280 }, { "epoch": 0.95, "learning_rate": 1.8103795767551225e-05, "loss": 0.1524, "step": 11290 }, { "epoch": 0.95, "learning_rate": 1.8102116224386967e-05, "loss": 0.2749, "step": 11300 }, { "epoch": 0.95, "learning_rate": 1.810043668122271e-05, "loss": 0.2385, "step": 11310 }, { "epoch": 0.95, "learning_rate": 1.8098757138058448e-05, "loss": 0.1095, "step": 11320 }, { "epoch": 0.95, "learning_rate": 1.809707759489419e-05, "loss": 0.1317, "step": 11330 }, { "epoch": 0.95, "learning_rate": 1.8095398051729933e-05, "loss": 0.1329, "step": 11340 }, { "epoch": 0.95, "learning_rate": 1.809371850856567e-05, "loss": 0.1498, "step": 11350 }, { "epoch": 0.95, "learning_rate": 1.8092038965401414e-05, "loss": 0.1217, "step": 11360 }, { "epoch": 0.95, "learning_rate": 1.8090359422237152e-05, "loss": 0.157, "step": 11370 }, { "epoch": 0.96, "learning_rate": 1.8088679879072894e-05, "loss": 0.1488, "step": 11380 }, { "epoch": 0.96, "learning_rate": 1.8087000335908633e-05, "loss": 0.2257, "step": 11390 }, { "epoch": 0.96, "learning_rate": 1.8085320792744375e-05, "loss": 0.0734, "step": 11400 }, { "epoch": 0.96, "learning_rate": 1.8083641249580114e-05, "loss": 0.1534, "step": 11410 }, { "epoch": 0.96, "learning_rate": 1.8081961706415856e-05, "loss": 0.1842, "step": 11420 }, { "epoch": 0.96, "learning_rate": 1.8080282163251595e-05, "loss": 0.1199, "step": 11430 }, { "epoch": 0.96, "learning_rate": 1.8078602620087337e-05, "loss": 0.0602, "step": 11440 }, { "epoch": 0.96, "learning_rate": 1.807692307692308e-05, "loss": 0.0787, "step": 11450 }, { "epoch": 0.96, "learning_rate": 1.8075243533758818e-05, "loss": 0.2117, "step": 11460 }, { "epoch": 0.96, "learning_rate": 1.807356399059456e-05, "loss": 0.1092, "step": 11470 }, { "epoch": 0.96, "learning_rate": 1.8071884447430303e-05, "loss": 0.1346, "step": 11480 }, { "epoch": 0.96, "learning_rate": 1.807020490426604e-05, "loss": 0.1964, "step": 11490 }, { "epoch": 0.97, "learning_rate": 1.806852536110178e-05, "loss": 0.1069, "step": 11500 }, { "epoch": 0.97, "learning_rate": 1.8066845817937522e-05, "loss": 0.1434, "step": 11510 }, { "epoch": 0.97, "learning_rate": 1.806516627477326e-05, "loss": 0.1706, "step": 11520 }, { "epoch": 0.97, "learning_rate": 1.8063486731609003e-05, "loss": 0.1315, "step": 11530 }, { "epoch": 0.97, "learning_rate": 1.8061807188444745e-05, "loss": 0.131, "step": 11540 }, { "epoch": 0.97, "learning_rate": 1.8060127645280484e-05, "loss": 0.0652, "step": 11550 }, { "epoch": 0.97, "learning_rate": 1.8058448102116226e-05, "loss": 0.1735, "step": 11560 }, { "epoch": 0.97, "learning_rate": 1.805676855895197e-05, "loss": 0.1353, "step": 11570 }, { "epoch": 0.97, "learning_rate": 1.8055089015787707e-05, "loss": 0.0877, "step": 11580 }, { "epoch": 0.97, "learning_rate": 1.805340947262345e-05, "loss": 0.1001, "step": 11590 }, { "epoch": 0.97, "learning_rate": 1.805172992945919e-05, "loss": 0.1397, "step": 11600 }, { "epoch": 0.97, "learning_rate": 1.8050050386294927e-05, "loss": 0.187, "step": 11610 }, { "epoch": 0.98, "learning_rate": 1.804837084313067e-05, "loss": 0.1803, "step": 11620 }, { "epoch": 0.98, "learning_rate": 1.804669129996641e-05, "loss": 0.1237, "step": 11630 }, { "epoch": 0.98, "learning_rate": 1.804501175680215e-05, "loss": 0.1211, "step": 11640 }, { "epoch": 0.98, "learning_rate": 1.8043332213637892e-05, "loss": 0.1107, "step": 11650 }, { "epoch": 0.98, "learning_rate": 1.804165267047363e-05, "loss": 0.0782, "step": 11660 }, { "epoch": 0.98, "learning_rate": 1.8039973127309373e-05, "loss": 0.1254, "step": 11670 }, { "epoch": 0.98, "learning_rate": 1.8038293584145116e-05, "loss": 0.1313, "step": 11680 }, { "epoch": 0.98, "learning_rate": 1.8036614040980854e-05, "loss": 0.1192, "step": 11690 }, { "epoch": 0.98, "learning_rate": 1.8034934497816597e-05, "loss": 0.1771, "step": 11700 }, { "epoch": 0.98, "learning_rate": 1.8033254954652335e-05, "loss": 0.0841, "step": 11710 }, { "epoch": 0.98, "learning_rate": 1.8031575411488077e-05, "loss": 0.1544, "step": 11720 }, { "epoch": 0.99, "learning_rate": 1.8029895868323816e-05, "loss": 0.1096, "step": 11730 }, { "epoch": 0.99, "learning_rate": 1.802821632515956e-05, "loss": 0.1805, "step": 11740 }, { "epoch": 0.99, "learning_rate": 1.8026536781995297e-05, "loss": 0.0982, "step": 11750 }, { "epoch": 0.99, "learning_rate": 1.802485723883104e-05, "loss": 0.1091, "step": 11760 }, { "epoch": 0.99, "learning_rate": 1.802317769566678e-05, "loss": 0.1202, "step": 11770 }, { "epoch": 0.99, "learning_rate": 1.802149815250252e-05, "loss": 0.1138, "step": 11780 }, { "epoch": 0.99, "learning_rate": 1.8019818609338262e-05, "loss": 0.1333, "step": 11790 }, { "epoch": 0.99, "learning_rate": 1.8018139066174005e-05, "loss": 0.1329, "step": 11800 }, { "epoch": 0.99, "learning_rate": 1.8016459523009743e-05, "loss": 0.1109, "step": 11810 }, { "epoch": 0.99, "learning_rate": 1.8014779979845482e-05, "loss": 0.0545, "step": 11820 }, { "epoch": 0.99, "learning_rate": 1.8013100436681224e-05, "loss": 0.2042, "step": 11830 }, { "epoch": 0.99, "learning_rate": 1.8011420893516963e-05, "loss": 0.0859, "step": 11840 }, { "epoch": 1.0, "learning_rate": 1.8009741350352705e-05, "loss": 0.1375, "step": 11850 }, { "epoch": 1.0, "learning_rate": 1.8008061807188448e-05, "loss": 0.1403, "step": 11860 }, { "epoch": 1.0, "learning_rate": 1.8006382264024186e-05, "loss": 0.1062, "step": 11870 }, { "epoch": 1.0, "learning_rate": 1.800470272085993e-05, "loss": 0.2203, "step": 11880 }, { "epoch": 1.0, "learning_rate": 1.8003023177695667e-05, "loss": 0.1086, "step": 11890 }, { "epoch": 1.0, "learning_rate": 1.800134363453141e-05, "loss": 0.0972, "step": 11900 }, { "epoch": 1.0, "eval_FN": 654, "eval_FP": 512, "eval_TN": 21202, "eval_TP": 1448, "eval_accuracy": 0.9510413167618408, "eval_f1": 0.7129492860659773, "eval_loss": 0.13444018363952637, "eval_precision": 0.7387755102040816, "eval_recall": 0.6888677450047573, "eval_runtime": 163.7471, "eval_samples_per_second": 145.444, "eval_steps_per_second": 9.093, "step": 11908 }, { "epoch": 1.0, "learning_rate": 1.799966409136715e-05, "loss": 0.1523, "step": 11910 }, { "epoch": 1.0, "learning_rate": 1.799798454820289e-05, "loss": 0.0805, "step": 11920 }, { "epoch": 1.0, "learning_rate": 1.799630500503863e-05, "loss": 0.1259, "step": 11930 }, { "epoch": 1.0, "learning_rate": 1.799462546187437e-05, "loss": 0.1381, "step": 11940 }, { "epoch": 1.0, "learning_rate": 1.7992945918710114e-05, "loss": 0.2345, "step": 11950 }, { "epoch": 1.0, "learning_rate": 1.7991266375545852e-05, "loss": 0.1364, "step": 11960 }, { "epoch": 1.01, "learning_rate": 1.7989586832381594e-05, "loss": 0.1715, "step": 11970 }, { "epoch": 1.01, "learning_rate": 1.7987907289217333e-05, "loss": 0.0947, "step": 11980 }, { "epoch": 1.01, "learning_rate": 1.7986227746053075e-05, "loss": 0.1624, "step": 11990 }, { "epoch": 1.01, "learning_rate": 1.7984548202888818e-05, "loss": 0.1541, "step": 12000 }, { "epoch": 1.01, "learning_rate": 1.7982868659724556e-05, "loss": 0.0913, "step": 12010 }, { "epoch": 1.01, "learning_rate": 1.79811891165603e-05, "loss": 0.0847, "step": 12020 }, { "epoch": 1.01, "learning_rate": 1.7979509573396037e-05, "loss": 0.1504, "step": 12030 }, { "epoch": 1.01, "learning_rate": 1.7977830030231776e-05, "loss": 0.1258, "step": 12040 }, { "epoch": 1.01, "learning_rate": 1.7976150487067518e-05, "loss": 0.0947, "step": 12050 }, { "epoch": 1.01, "learning_rate": 1.797447094390326e-05, "loss": 0.0664, "step": 12060 }, { "epoch": 1.01, "learning_rate": 1.7972791400739e-05, "loss": 0.0483, "step": 12070 }, { "epoch": 1.01, "learning_rate": 1.797111185757474e-05, "loss": 0.1226, "step": 12080 }, { "epoch": 1.02, "learning_rate": 1.7969432314410484e-05, "loss": 0.1819, "step": 12090 }, { "epoch": 1.02, "learning_rate": 1.7967752771246222e-05, "loss": 0.0791, "step": 12100 }, { "epoch": 1.02, "learning_rate": 1.7966073228081965e-05, "loss": 0.058, "step": 12110 }, { "epoch": 1.02, "learning_rate": 1.7964393684917703e-05, "loss": 0.1645, "step": 12120 }, { "epoch": 1.02, "learning_rate": 1.7962714141753445e-05, "loss": 0.1644, "step": 12130 }, { "epoch": 1.02, "learning_rate": 1.7961034598589184e-05, "loss": 0.1194, "step": 12140 }, { "epoch": 1.02, "learning_rate": 1.7959355055424926e-05, "loss": 0.0335, "step": 12150 }, { "epoch": 1.02, "learning_rate": 1.7957675512260665e-05, "loss": 0.1914, "step": 12160 }, { "epoch": 1.02, "learning_rate": 1.7955995969096407e-05, "loss": 0.1561, "step": 12170 }, { "epoch": 1.02, "learning_rate": 1.7954316425932146e-05, "loss": 0.1745, "step": 12180 }, { "epoch": 1.02, "learning_rate": 1.795263688276789e-05, "loss": 0.1371, "step": 12190 }, { "epoch": 1.02, "learning_rate": 1.795095733960363e-05, "loss": 0.1496, "step": 12200 }, { "epoch": 1.03, "learning_rate": 1.794927779643937e-05, "loss": 0.0834, "step": 12210 }, { "epoch": 1.03, "learning_rate": 1.794759825327511e-05, "loss": 0.1644, "step": 12220 }, { "epoch": 1.03, "learning_rate": 1.7945918710110854e-05, "loss": 0.1877, "step": 12230 }, { "epoch": 1.03, "learning_rate": 1.7944239166946592e-05, "loss": 0.1126, "step": 12240 }, { "epoch": 1.03, "learning_rate": 1.794255962378233e-05, "loss": 0.0798, "step": 12250 }, { "epoch": 1.03, "learning_rate": 1.7940880080618073e-05, "loss": 0.149, "step": 12260 }, { "epoch": 1.03, "learning_rate": 1.7939200537453812e-05, "loss": 0.1165, "step": 12270 }, { "epoch": 1.03, "learning_rate": 1.7937520994289554e-05, "loss": 0.1227, "step": 12280 }, { "epoch": 1.03, "learning_rate": 1.7935841451125296e-05, "loss": 0.1301, "step": 12290 }, { "epoch": 1.03, "learning_rate": 1.7934161907961035e-05, "loss": 0.0562, "step": 12300 }, { "epoch": 1.03, "learning_rate": 1.7932482364796777e-05, "loss": 0.179, "step": 12310 }, { "epoch": 1.03, "learning_rate": 1.793080282163252e-05, "loss": 0.1179, "step": 12320 }, { "epoch": 1.04, "learning_rate": 1.792912327846826e-05, "loss": 0.1127, "step": 12330 }, { "epoch": 1.04, "learning_rate": 1.7927443735304e-05, "loss": 0.148, "step": 12340 }, { "epoch": 1.04, "learning_rate": 1.792576419213974e-05, "loss": 0.0959, "step": 12350 }, { "epoch": 1.04, "learning_rate": 1.7924084648975478e-05, "loss": 0.0964, "step": 12360 }, { "epoch": 1.04, "learning_rate": 1.792240510581122e-05, "loss": 0.0884, "step": 12370 }, { "epoch": 1.04, "learning_rate": 1.7920725562646962e-05, "loss": 0.1866, "step": 12380 }, { "epoch": 1.04, "learning_rate": 1.79190460194827e-05, "loss": 0.0907, "step": 12390 }, { "epoch": 1.04, "learning_rate": 1.7917366476318443e-05, "loss": 0.0966, "step": 12400 }, { "epoch": 1.04, "learning_rate": 1.7915686933154182e-05, "loss": 0.1798, "step": 12410 }, { "epoch": 1.04, "learning_rate": 1.7914007389989924e-05, "loss": 0.1162, "step": 12420 }, { "epoch": 1.04, "learning_rate": 1.7912327846825667e-05, "loss": 0.0797, "step": 12430 }, { "epoch": 1.04, "learning_rate": 1.7910648303661405e-05, "loss": 0.1333, "step": 12440 }, { "epoch": 1.05, "learning_rate": 1.7908968760497144e-05, "loss": 0.1544, "step": 12450 }, { "epoch": 1.05, "learning_rate": 1.7907289217332886e-05, "loss": 0.0873, "step": 12460 }, { "epoch": 1.05, "learning_rate": 1.790560967416863e-05, "loss": 0.1772, "step": 12470 }, { "epoch": 1.05, "learning_rate": 1.7903930131004367e-05, "loss": 0.1278, "step": 12480 }, { "epoch": 1.05, "learning_rate": 1.790225058784011e-05, "loss": 0.082, "step": 12490 }, { "epoch": 1.05, "learning_rate": 1.7900571044675848e-05, "loss": 0.1337, "step": 12500 }, { "epoch": 1.05, "learning_rate": 1.789889150151159e-05, "loss": 0.1343, "step": 12510 }, { "epoch": 1.05, "learning_rate": 1.7897211958347333e-05, "loss": 0.1282, "step": 12520 }, { "epoch": 1.05, "learning_rate": 1.789553241518307e-05, "loss": 0.07, "step": 12530 }, { "epoch": 1.05, "learning_rate": 1.7893852872018813e-05, "loss": 0.1516, "step": 12540 }, { "epoch": 1.05, "learning_rate": 1.7892173328854556e-05, "loss": 0.1256, "step": 12550 }, { "epoch": 1.05, "learning_rate": 1.789049378569029e-05, "loss": 0.2049, "step": 12560 }, { "epoch": 1.06, "learning_rate": 1.7888814242526033e-05, "loss": 0.1293, "step": 12570 }, { "epoch": 1.06, "learning_rate": 1.7887134699361775e-05, "loss": 0.1629, "step": 12580 }, { "epoch": 1.06, "learning_rate": 1.7885455156197514e-05, "loss": 0.0529, "step": 12590 }, { "epoch": 1.06, "learning_rate": 1.7883775613033256e-05, "loss": 0.0956, "step": 12600 }, { "epoch": 1.06, "learning_rate": 1.7882096069869e-05, "loss": 0.0992, "step": 12610 }, { "epoch": 1.06, "learning_rate": 1.7880416526704737e-05, "loss": 0.1336, "step": 12620 }, { "epoch": 1.06, "learning_rate": 1.787873698354048e-05, "loss": 0.1148, "step": 12630 }, { "epoch": 1.06, "learning_rate": 1.7877057440376218e-05, "loss": 0.1394, "step": 12640 }, { "epoch": 1.06, "learning_rate": 1.787537789721196e-05, "loss": 0.1432, "step": 12650 }, { "epoch": 1.06, "learning_rate": 1.7873698354047703e-05, "loss": 0.1077, "step": 12660 }, { "epoch": 1.06, "learning_rate": 1.787201881088344e-05, "loss": 0.1883, "step": 12670 }, { "epoch": 1.06, "learning_rate": 1.787033926771918e-05, "loss": 0.0648, "step": 12680 }, { "epoch": 1.07, "learning_rate": 1.7868659724554922e-05, "loss": 0.1979, "step": 12690 }, { "epoch": 1.07, "learning_rate": 1.7866980181390665e-05, "loss": 0.0819, "step": 12700 }, { "epoch": 1.07, "learning_rate": 1.7865300638226403e-05, "loss": 0.1096, "step": 12710 }, { "epoch": 1.07, "learning_rate": 1.7863621095062145e-05, "loss": 0.0934, "step": 12720 }, { "epoch": 1.07, "learning_rate": 1.7861941551897884e-05, "loss": 0.099, "step": 12730 }, { "epoch": 1.07, "learning_rate": 1.7860262008733626e-05, "loss": 0.0977, "step": 12740 }, { "epoch": 1.07, "learning_rate": 1.785858246556937e-05, "loss": 0.0674, "step": 12750 }, { "epoch": 1.07, "learning_rate": 1.7856902922405107e-05, "loss": 0.1585, "step": 12760 }, { "epoch": 1.07, "learning_rate": 1.7855223379240846e-05, "loss": 0.1005, "step": 12770 }, { "epoch": 1.07, "learning_rate": 1.785354383607659e-05, "loss": 0.0905, "step": 12780 }, { "epoch": 1.07, "learning_rate": 1.7851864292912327e-05, "loss": 0.0922, "step": 12790 }, { "epoch": 1.07, "learning_rate": 1.785018474974807e-05, "loss": 0.1135, "step": 12800 }, { "epoch": 1.08, "learning_rate": 1.784850520658381e-05, "loss": 0.084, "step": 12810 }, { "epoch": 1.08, "learning_rate": 1.784682566341955e-05, "loss": 0.1265, "step": 12820 }, { "epoch": 1.08, "learning_rate": 1.7845146120255292e-05, "loss": 0.1036, "step": 12830 }, { "epoch": 1.08, "learning_rate": 1.7843466577091035e-05, "loss": 0.1185, "step": 12840 }, { "epoch": 1.08, "learning_rate": 1.7841787033926773e-05, "loss": 0.1955, "step": 12850 }, { "epoch": 1.08, "learning_rate": 1.7840107490762516e-05, "loss": 0.0725, "step": 12860 }, { "epoch": 1.08, "learning_rate": 1.7838427947598254e-05, "loss": 0.1199, "step": 12870 }, { "epoch": 1.08, "learning_rate": 1.7836748404433993e-05, "loss": 0.0844, "step": 12880 }, { "epoch": 1.08, "learning_rate": 1.7835068861269735e-05, "loss": 0.0252, "step": 12890 }, { "epoch": 1.08, "learning_rate": 1.7833389318105477e-05, "loss": 0.2474, "step": 12900 }, { "epoch": 1.08, "learning_rate": 1.7831709774941216e-05, "loss": 0.1932, "step": 12910 }, { "epoch": 1.08, "learning_rate": 1.783003023177696e-05, "loss": 0.0953, "step": 12920 }, { "epoch": 1.09, "learning_rate": 1.78283506886127e-05, "loss": 0.0927, "step": 12930 }, { "epoch": 1.09, "learning_rate": 1.782667114544844e-05, "loss": 0.0908, "step": 12940 }, { "epoch": 1.09, "learning_rate": 1.782499160228418e-05, "loss": 0.0997, "step": 12950 }, { "epoch": 1.09, "learning_rate": 1.782331205911992e-05, "loss": 0.0336, "step": 12960 }, { "epoch": 1.09, "learning_rate": 1.7821632515955662e-05, "loss": 0.0716, "step": 12970 }, { "epoch": 1.09, "learning_rate": 1.78199529727914e-05, "loss": 0.1572, "step": 12980 }, { "epoch": 1.09, "learning_rate": 1.7818273429627143e-05, "loss": 0.1274, "step": 12990 }, { "epoch": 1.09, "learning_rate": 1.7816593886462882e-05, "loss": 0.2101, "step": 13000 }, { "epoch": 1.09, "learning_rate": 1.7814914343298624e-05, "loss": 0.1088, "step": 13010 }, { "epoch": 1.09, "learning_rate": 1.7813234800134363e-05, "loss": 0.0745, "step": 13020 }, { "epoch": 1.09, "learning_rate": 1.7811555256970105e-05, "loss": 0.1386, "step": 13030 }, { "epoch": 1.1, "learning_rate": 1.7809875713805847e-05, "loss": 0.1415, "step": 13040 }, { "epoch": 1.1, "learning_rate": 1.7808196170641586e-05, "loss": 0.1438, "step": 13050 }, { "epoch": 1.1, "learning_rate": 1.780651662747733e-05, "loss": 0.0838, "step": 13060 }, { "epoch": 1.1, "learning_rate": 1.780483708431307e-05, "loss": 0.136, "step": 13070 }, { "epoch": 1.1, "learning_rate": 1.780315754114881e-05, "loss": 0.0623, "step": 13080 }, { "epoch": 1.1, "learning_rate": 1.7801477997984548e-05, "loss": 0.1732, "step": 13090 }, { "epoch": 1.1, "learning_rate": 1.779979845482029e-05, "loss": 0.0931, "step": 13100 }, { "epoch": 1.1, "learning_rate": 1.779811891165603e-05, "loss": 0.1563, "step": 13110 }, { "epoch": 1.1, "learning_rate": 1.779643936849177e-05, "loss": 0.1095, "step": 13120 }, { "epoch": 1.1, "learning_rate": 1.7794759825327513e-05, "loss": 0.128, "step": 13130 }, { "epoch": 1.1, "learning_rate": 1.7793080282163252e-05, "loss": 0.151, "step": 13140 }, { "epoch": 1.1, "learning_rate": 1.7791400738998994e-05, "loss": 0.128, "step": 13150 }, { "epoch": 1.11, "learning_rate": 1.7789721195834733e-05, "loss": 0.1818, "step": 13160 }, { "epoch": 1.11, "learning_rate": 1.7788041652670475e-05, "loss": 0.1092, "step": 13170 }, { "epoch": 1.11, "learning_rate": 1.7786362109506218e-05, "loss": 0.1431, "step": 13180 }, { "epoch": 1.11, "learning_rate": 1.7784682566341956e-05, "loss": 0.0727, "step": 13190 }, { "epoch": 1.11, "learning_rate": 1.7783003023177695e-05, "loss": 0.0901, "step": 13200 }, { "epoch": 1.11, "learning_rate": 1.7781323480013437e-05, "loss": 0.1346, "step": 13210 }, { "epoch": 1.11, "learning_rate": 1.777964393684918e-05, "loss": 0.1126, "step": 13220 }, { "epoch": 1.11, "learning_rate": 1.7777964393684918e-05, "loss": 0.1432, "step": 13230 }, { "epoch": 1.11, "learning_rate": 1.777628485052066e-05, "loss": 0.0931, "step": 13240 }, { "epoch": 1.11, "learning_rate": 1.77746053073564e-05, "loss": 0.138, "step": 13250 }, { "epoch": 1.11, "learning_rate": 1.777292576419214e-05, "loss": 0.0868, "step": 13260 }, { "epoch": 1.11, "learning_rate": 1.7771246221027884e-05, "loss": 0.1094, "step": 13270 }, { "epoch": 1.12, "learning_rate": 1.7769566677863622e-05, "loss": 0.0848, "step": 13280 }, { "epoch": 1.12, "learning_rate": 1.7767887134699364e-05, "loss": 0.1693, "step": 13290 }, { "epoch": 1.12, "learning_rate": 1.7766207591535103e-05, "loss": 0.1243, "step": 13300 }, { "epoch": 1.12, "learning_rate": 1.7764528048370845e-05, "loss": 0.1482, "step": 13310 }, { "epoch": 1.12, "learning_rate": 1.7762848505206584e-05, "loss": 0.1428, "step": 13320 }, { "epoch": 1.12, "learning_rate": 1.7761168962042326e-05, "loss": 0.13, "step": 13330 }, { "epoch": 1.12, "learning_rate": 1.7759489418878065e-05, "loss": 0.0837, "step": 13340 }, { "epoch": 1.12, "learning_rate": 1.7757809875713807e-05, "loss": 0.0363, "step": 13350 }, { "epoch": 1.12, "learning_rate": 1.775613033254955e-05, "loss": 0.1406, "step": 13360 }, { "epoch": 1.12, "learning_rate": 1.7754450789385288e-05, "loss": 0.1284, "step": 13370 }, { "epoch": 1.12, "learning_rate": 1.775277124622103e-05, "loss": 0.2912, "step": 13380 }, { "epoch": 1.12, "learning_rate": 1.775109170305677e-05, "loss": 0.1089, "step": 13390 }, { "epoch": 1.13, "learning_rate": 1.774941215989251e-05, "loss": 0.1392, "step": 13400 }, { "epoch": 1.13, "learning_rate": 1.774773261672825e-05, "loss": 0.0815, "step": 13410 }, { "epoch": 1.13, "learning_rate": 1.7746053073563992e-05, "loss": 0.0854, "step": 13420 }, { "epoch": 1.13, "learning_rate": 1.774437353039973e-05, "loss": 0.1747, "step": 13430 }, { "epoch": 1.13, "learning_rate": 1.7742693987235473e-05, "loss": 0.1014, "step": 13440 }, { "epoch": 1.13, "learning_rate": 1.7741014444071216e-05, "loss": 0.0804, "step": 13450 }, { "epoch": 1.13, "learning_rate": 1.7739334900906954e-05, "loss": 0.0623, "step": 13460 }, { "epoch": 1.13, "learning_rate": 1.7737655357742696e-05, "loss": 0.1678, "step": 13470 }, { "epoch": 1.13, "learning_rate": 1.7735975814578435e-05, "loss": 0.1931, "step": 13480 }, { "epoch": 1.13, "learning_rate": 1.7734296271414177e-05, "loss": 0.0889, "step": 13490 }, { "epoch": 1.13, "learning_rate": 1.773261672824992e-05, "loss": 0.0872, "step": 13500 }, { "epoch": 1.13, "learning_rate": 1.773093718508566e-05, "loss": 0.0832, "step": 13510 }, { "epoch": 1.14, "learning_rate": 1.7729257641921397e-05, "loss": 0.0689, "step": 13520 }, { "epoch": 1.14, "learning_rate": 1.772757809875714e-05, "loss": 0.0547, "step": 13530 }, { "epoch": 1.14, "learning_rate": 1.7725898555592878e-05, "loss": 0.1494, "step": 13540 }, { "epoch": 1.14, "learning_rate": 1.772421901242862e-05, "loss": 0.0777, "step": 13550 }, { "epoch": 1.14, "learning_rate": 1.7722539469264362e-05, "loss": 0.1487, "step": 13560 }, { "epoch": 1.14, "learning_rate": 1.77208599261001e-05, "loss": 0.1258, "step": 13570 }, { "epoch": 1.14, "learning_rate": 1.7719180382935843e-05, "loss": 0.0785, "step": 13580 }, { "epoch": 1.14, "learning_rate": 1.7717500839771586e-05, "loss": 0.075, "step": 13590 }, { "epoch": 1.14, "learning_rate": 1.7715821296607324e-05, "loss": 0.1128, "step": 13600 }, { "epoch": 1.14, "learning_rate": 1.7714141753443067e-05, "loss": 0.1152, "step": 13610 }, { "epoch": 1.14, "learning_rate": 1.7712462210278805e-05, "loss": 0.1076, "step": 13620 }, { "epoch": 1.14, "learning_rate": 1.7710782667114544e-05, "loss": 0.0692, "step": 13630 }, { "epoch": 1.15, "learning_rate": 1.7709103123950286e-05, "loss": 0.1665, "step": 13640 }, { "epoch": 1.15, "learning_rate": 1.770742358078603e-05, "loss": 0.0778, "step": 13650 }, { "epoch": 1.15, "learning_rate": 1.7705744037621767e-05, "loss": 0.0915, "step": 13660 }, { "epoch": 1.15, "learning_rate": 1.770406449445751e-05, "loss": 0.1586, "step": 13670 }, { "epoch": 1.15, "learning_rate": 1.770238495129325e-05, "loss": 0.1534, "step": 13680 }, { "epoch": 1.15, "learning_rate": 1.770070540812899e-05, "loss": 0.0898, "step": 13690 }, { "epoch": 1.15, "learning_rate": 1.7699025864964732e-05, "loss": 0.1347, "step": 13700 }, { "epoch": 1.15, "learning_rate": 1.769734632180047e-05, "loss": 0.0615, "step": 13710 }, { "epoch": 1.15, "learning_rate": 1.769566677863621e-05, "loss": 0.1676, "step": 13720 }, { "epoch": 1.15, "learning_rate": 1.7693987235471952e-05, "loss": 0.1401, "step": 13730 }, { "epoch": 1.15, "learning_rate": 1.7692307692307694e-05, "loss": 0.2207, "step": 13740 }, { "epoch": 1.15, "learning_rate": 1.7690628149143433e-05, "loss": 0.1672, "step": 13750 }, { "epoch": 1.16, "learning_rate": 1.7688948605979175e-05, "loss": 0.115, "step": 13760 }, { "epoch": 1.16, "learning_rate": 1.7687269062814914e-05, "loss": 0.0568, "step": 13770 }, { "epoch": 1.16, "learning_rate": 1.7685589519650656e-05, "loss": 0.1138, "step": 13780 }, { "epoch": 1.16, "learning_rate": 1.76839099764864e-05, "loss": 0.0989, "step": 13790 }, { "epoch": 1.16, "learning_rate": 1.7682230433322137e-05, "loss": 0.0548, "step": 13800 }, { "epoch": 1.16, "learning_rate": 1.768055089015788e-05, "loss": 0.1321, "step": 13810 }, { "epoch": 1.16, "learning_rate": 1.767887134699362e-05, "loss": 0.1408, "step": 13820 }, { "epoch": 1.16, "learning_rate": 1.767719180382936e-05, "loss": 0.0971, "step": 13830 }, { "epoch": 1.16, "learning_rate": 1.76755122606651e-05, "loss": 0.1684, "step": 13840 }, { "epoch": 1.16, "learning_rate": 1.767383271750084e-05, "loss": 0.0448, "step": 13850 }, { "epoch": 1.16, "learning_rate": 1.767215317433658e-05, "loss": 0.0767, "step": 13860 }, { "epoch": 1.16, "learning_rate": 1.7670473631172322e-05, "loss": 0.0811, "step": 13870 }, { "epoch": 1.17, "learning_rate": 1.7668794088008064e-05, "loss": 0.0854, "step": 13880 }, { "epoch": 1.17, "learning_rate": 1.7667114544843803e-05, "loss": 0.2873, "step": 13890 }, { "epoch": 1.17, "learning_rate": 1.7665435001679545e-05, "loss": 0.1893, "step": 13900 }, { "epoch": 1.17, "learning_rate": 1.7663755458515288e-05, "loss": 0.2539, "step": 13910 }, { "epoch": 1.17, "learning_rate": 1.7662075915351026e-05, "loss": 0.1757, "step": 13920 }, { "epoch": 1.17, "learning_rate": 1.766039637218677e-05, "loss": 0.124, "step": 13930 }, { "epoch": 1.17, "learning_rate": 1.7658716829022507e-05, "loss": 0.1062, "step": 13940 }, { "epoch": 1.17, "learning_rate": 1.7657037285858246e-05, "loss": 0.146, "step": 13950 }, { "epoch": 1.17, "learning_rate": 1.7655357742693988e-05, "loss": 0.0593, "step": 13960 }, { "epoch": 1.17, "learning_rate": 1.765367819952973e-05, "loss": 0.2146, "step": 13970 }, { "epoch": 1.17, "learning_rate": 1.765199865636547e-05, "loss": 0.1359, "step": 13980 }, { "epoch": 1.17, "learning_rate": 1.765031911320121e-05, "loss": 0.0878, "step": 13990 }, { "epoch": 1.18, "learning_rate": 1.764863957003695e-05, "loss": 0.1149, "step": 14000 }, { "epoch": 1.18, "learning_rate": 1.7646960026872692e-05, "loss": 0.0955, "step": 14010 }, { "epoch": 1.18, "learning_rate": 1.7645280483708435e-05, "loss": 0.1055, "step": 14020 }, { "epoch": 1.18, "learning_rate": 1.7643600940544173e-05, "loss": 0.1575, "step": 14030 }, { "epoch": 1.18, "learning_rate": 1.7641921397379912e-05, "loss": 0.0627, "step": 14040 }, { "epoch": 1.18, "learning_rate": 1.7640241854215654e-05, "loss": 0.1149, "step": 14050 }, { "epoch": 1.18, "learning_rate": 1.7638562311051396e-05, "loss": 0.1003, "step": 14060 }, { "epoch": 1.18, "learning_rate": 1.7636882767887135e-05, "loss": 0.1455, "step": 14070 }, { "epoch": 1.18, "learning_rate": 1.7635203224722877e-05, "loss": 0.1134, "step": 14080 }, { "epoch": 1.18, "learning_rate": 1.7633523681558616e-05, "loss": 0.0967, "step": 14090 }, { "epoch": 1.18, "learning_rate": 1.763184413839436e-05, "loss": 0.1367, "step": 14100 }, { "epoch": 1.18, "learning_rate": 1.76301645952301e-05, "loss": 0.141, "step": 14110 }, { "epoch": 1.19, "learning_rate": 1.762848505206584e-05, "loss": 0.143, "step": 14120 }, { "epoch": 1.19, "learning_rate": 1.762680550890158e-05, "loss": 0.0787, "step": 14130 }, { "epoch": 1.19, "learning_rate": 1.762512596573732e-05, "loss": 0.1245, "step": 14140 }, { "epoch": 1.19, "learning_rate": 1.762344642257306e-05, "loss": 0.0909, "step": 14150 }, { "epoch": 1.19, "learning_rate": 1.76217668794088e-05, "loss": 0.0882, "step": 14160 }, { "epoch": 1.19, "learning_rate": 1.7620087336244543e-05, "loss": 0.1104, "step": 14170 }, { "epoch": 1.19, "learning_rate": 1.7618407793080282e-05, "loss": 0.0927, "step": 14180 }, { "epoch": 1.19, "learning_rate": 1.7616728249916024e-05, "loss": 0.1439, "step": 14190 }, { "epoch": 1.19, "learning_rate": 1.7615048706751766e-05, "loss": 0.052, "step": 14200 }, { "epoch": 1.19, "learning_rate": 1.7613369163587505e-05, "loss": 0.1277, "step": 14210 }, { "epoch": 1.19, "learning_rate": 1.7611689620423247e-05, "loss": 0.0885, "step": 14220 }, { "epoch": 1.19, "learning_rate": 1.7610010077258986e-05, "loss": 0.1502, "step": 14230 }, { "epoch": 1.2, "learning_rate": 1.760833053409473e-05, "loss": 0.0752, "step": 14240 }, { "epoch": 1.2, "learning_rate": 1.7606650990930467e-05, "loss": 0.1282, "step": 14250 }, { "epoch": 1.2, "learning_rate": 1.760497144776621e-05, "loss": 0.1584, "step": 14260 }, { "epoch": 1.2, "learning_rate": 1.7603291904601948e-05, "loss": 0.1165, "step": 14270 }, { "epoch": 1.2, "learning_rate": 1.760161236143769e-05, "loss": 0.1237, "step": 14280 }, { "epoch": 1.2, "learning_rate": 1.7599932818273432e-05, "loss": 0.1037, "step": 14290 }, { "epoch": 1.2, "learning_rate": 1.759825327510917e-05, "loss": 0.1065, "step": 14300 }, { "epoch": 1.2, "learning_rate": 1.7596573731944913e-05, "loss": 0.1143, "step": 14310 }, { "epoch": 1.2, "learning_rate": 1.7594894188780652e-05, "loss": 0.0519, "step": 14320 }, { "epoch": 1.2, "learning_rate": 1.7593214645616394e-05, "loss": 0.1926, "step": 14330 }, { "epoch": 1.2, "learning_rate": 1.7591535102452137e-05, "loss": 0.0878, "step": 14340 }, { "epoch": 1.21, "learning_rate": 1.7589855559287875e-05, "loss": 0.1004, "step": 14350 }, { "epoch": 1.21, "learning_rate": 1.7588176016123614e-05, "loss": 0.0908, "step": 14360 }, { "epoch": 1.21, "learning_rate": 1.7586496472959356e-05, "loss": 0.0658, "step": 14370 }, { "epoch": 1.21, "learning_rate": 1.7584816929795095e-05, "loss": 0.087, "step": 14380 }, { "epoch": 1.21, "learning_rate": 1.7583137386630837e-05, "loss": 0.1672, "step": 14390 }, { "epoch": 1.21, "learning_rate": 1.758145784346658e-05, "loss": 0.177, "step": 14400 }, { "epoch": 1.21, "learning_rate": 1.7579778300302318e-05, "loss": 0.0215, "step": 14410 }, { "epoch": 1.21, "learning_rate": 1.757809875713806e-05, "loss": 0.0692, "step": 14420 }, { "epoch": 1.21, "learning_rate": 1.7576419213973803e-05, "loss": 0.1354, "step": 14430 }, { "epoch": 1.21, "learning_rate": 1.757473967080954e-05, "loss": 0.1239, "step": 14440 }, { "epoch": 1.21, "learning_rate": 1.7573060127645283e-05, "loss": 0.0928, "step": 14450 }, { "epoch": 1.21, "learning_rate": 1.7571380584481022e-05, "loss": 0.098, "step": 14460 }, { "epoch": 1.22, "learning_rate": 1.756970104131676e-05, "loss": 0.165, "step": 14470 }, { "epoch": 1.22, "learning_rate": 1.7568021498152503e-05, "loss": 0.1116, "step": 14480 }, { "epoch": 1.22, "learning_rate": 1.7566341954988245e-05, "loss": 0.1695, "step": 14490 }, { "epoch": 1.22, "learning_rate": 1.7564662411823984e-05, "loss": 0.0903, "step": 14500 }, { "epoch": 1.22, "learning_rate": 1.7562982868659726e-05, "loss": 0.1309, "step": 14510 }, { "epoch": 1.22, "learning_rate": 1.7561303325495465e-05, "loss": 0.145, "step": 14520 }, { "epoch": 1.22, "learning_rate": 1.7559623782331207e-05, "loss": 0.1465, "step": 14530 }, { "epoch": 1.22, "learning_rate": 1.755794423916695e-05, "loss": 0.1539, "step": 14540 }, { "epoch": 1.22, "learning_rate": 1.7556264696002688e-05, "loss": 0.0414, "step": 14550 }, { "epoch": 1.22, "learning_rate": 1.755458515283843e-05, "loss": 0.0594, "step": 14560 }, { "epoch": 1.22, "learning_rate": 1.755290560967417e-05, "loss": 0.1275, "step": 14570 }, { "epoch": 1.22, "learning_rate": 1.755122606650991e-05, "loss": 0.1386, "step": 14580 }, { "epoch": 1.23, "learning_rate": 1.754954652334565e-05, "loss": 0.1291, "step": 14590 }, { "epoch": 1.23, "learning_rate": 1.7547866980181392e-05, "loss": 0.1014, "step": 14600 }, { "epoch": 1.23, "learning_rate": 1.754618743701713e-05, "loss": 0.1365, "step": 14610 }, { "epoch": 1.23, "learning_rate": 1.7544507893852873e-05, "loss": 0.1684, "step": 14620 }, { "epoch": 1.23, "learning_rate": 1.7542828350688615e-05, "loss": 0.1698, "step": 14630 }, { "epoch": 1.23, "learning_rate": 1.7541148807524354e-05, "loss": 0.1673, "step": 14640 }, { "epoch": 1.23, "learning_rate": 1.7539469264360096e-05, "loss": 0.1839, "step": 14650 }, { "epoch": 1.23, "learning_rate": 1.753778972119584e-05, "loss": 0.0694, "step": 14660 }, { "epoch": 1.23, "learning_rate": 1.7536110178031577e-05, "loss": 0.1373, "step": 14670 }, { "epoch": 1.23, "learning_rate": 1.7534430634867316e-05, "loss": 0.1151, "step": 14680 }, { "epoch": 1.23, "learning_rate": 1.753275109170306e-05, "loss": 0.1738, "step": 14690 }, { "epoch": 1.23, "learning_rate": 1.7531071548538797e-05, "loss": 0.1242, "step": 14700 }, { "epoch": 1.24, "learning_rate": 1.752939200537454e-05, "loss": 0.108, "step": 14710 }, { "epoch": 1.24, "learning_rate": 1.752771246221028e-05, "loss": 0.1449, "step": 14720 }, { "epoch": 1.24, "learning_rate": 1.752603291904602e-05, "loss": 0.1704, "step": 14730 }, { "epoch": 1.24, "learning_rate": 1.7524353375881762e-05, "loss": 0.118, "step": 14740 }, { "epoch": 1.24, "learning_rate": 1.75226738327175e-05, "loss": 0.1414, "step": 14750 }, { "epoch": 1.24, "learning_rate": 1.7520994289553243e-05, "loss": 0.144, "step": 14760 }, { "epoch": 1.24, "learning_rate": 1.7519314746388986e-05, "loss": 0.1605, "step": 14770 }, { "epoch": 1.24, "learning_rate": 1.7517635203224724e-05, "loss": 0.0907, "step": 14780 }, { "epoch": 1.24, "learning_rate": 1.7515955660060463e-05, "loss": 0.0536, "step": 14790 }, { "epoch": 1.24, "learning_rate": 1.7514276116896205e-05, "loss": 0.1948, "step": 14800 }, { "epoch": 1.24, "learning_rate": 1.7512596573731947e-05, "loss": 0.0954, "step": 14810 }, { "epoch": 1.24, "learning_rate": 1.7510917030567686e-05, "loss": 0.1213, "step": 14820 }, { "epoch": 1.25, "learning_rate": 1.750923748740343e-05, "loss": 0.106, "step": 14830 }, { "epoch": 1.25, "learning_rate": 1.7507557944239167e-05, "loss": 0.0684, "step": 14840 }, { "epoch": 1.25, "learning_rate": 1.750587840107491e-05, "loss": 0.0923, "step": 14850 }, { "epoch": 1.25, "learning_rate": 1.750419885791065e-05, "loss": 0.0675, "step": 14860 }, { "epoch": 1.25, "learning_rate": 1.750251931474639e-05, "loss": 0.1515, "step": 14870 }, { "epoch": 1.25, "learning_rate": 1.7500839771582132e-05, "loss": 0.1447, "step": 14880 }, { "epoch": 1.25, "learning_rate": 1.749916022841787e-05, "loss": 0.1727, "step": 14890 }, { "epoch": 1.25, "learning_rate": 1.749748068525361e-05, "loss": 0.0737, "step": 14900 }, { "epoch": 1.25, "learning_rate": 1.7495801142089352e-05, "loss": 0.1653, "step": 14910 }, { "epoch": 1.25, "learning_rate": 1.7494121598925094e-05, "loss": 0.068, "step": 14920 }, { "epoch": 1.25, "learning_rate": 1.7492442055760833e-05, "loss": 0.1273, "step": 14930 }, { "epoch": 1.25, "learning_rate": 1.7490762512596575e-05, "loss": 0.0693, "step": 14940 }, { "epoch": 1.26, "learning_rate": 1.7489082969432317e-05, "loss": 0.0569, "step": 14950 }, { "epoch": 1.26, "learning_rate": 1.7487403426268056e-05, "loss": 0.1686, "step": 14960 }, { "epoch": 1.26, "learning_rate": 1.74857238831038e-05, "loss": 0.0695, "step": 14970 }, { "epoch": 1.26, "learning_rate": 1.7484044339939537e-05, "loss": 0.1889, "step": 14980 }, { "epoch": 1.26, "learning_rate": 1.748236479677528e-05, "loss": 0.1045, "step": 14990 }, { "epoch": 1.26, "learning_rate": 1.7480685253611018e-05, "loss": 0.1024, "step": 15000 }, { "epoch": 1.26, "learning_rate": 1.747900571044676e-05, "loss": 0.1167, "step": 15010 }, { "epoch": 1.26, "learning_rate": 1.74773261672825e-05, "loss": 0.1253, "step": 15020 }, { "epoch": 1.26, "learning_rate": 1.747564662411824e-05, "loss": 0.0471, "step": 15030 }, { "epoch": 1.26, "learning_rate": 1.7473967080953983e-05, "loss": 0.1735, "step": 15040 }, { "epoch": 1.26, "learning_rate": 1.7472287537789722e-05, "loss": 0.1886, "step": 15050 }, { "epoch": 1.26, "learning_rate": 1.7470607994625464e-05, "loss": 0.1676, "step": 15060 }, { "epoch": 1.27, "learning_rate": 1.7468928451461203e-05, "loss": 0.0589, "step": 15070 }, { "epoch": 1.27, "learning_rate": 1.7467248908296945e-05, "loss": 0.0593, "step": 15080 }, { "epoch": 1.27, "learning_rate": 1.7465569365132688e-05, "loss": 0.1812, "step": 15090 }, { "epoch": 1.27, "learning_rate": 1.7463889821968426e-05, "loss": 0.1692, "step": 15100 }, { "epoch": 1.27, "learning_rate": 1.7462210278804165e-05, "loss": 0.1109, "step": 15110 }, { "epoch": 1.27, "learning_rate": 1.7460530735639907e-05, "loss": 0.1241, "step": 15120 }, { "epoch": 1.27, "learning_rate": 1.7458851192475646e-05, "loss": 0.0973, "step": 15130 }, { "epoch": 1.27, "learning_rate": 1.7457171649311388e-05, "loss": 0.2026, "step": 15140 }, { "epoch": 1.27, "learning_rate": 1.745549210614713e-05, "loss": 0.1017, "step": 15150 }, { "epoch": 1.27, "learning_rate": 1.745381256298287e-05, "loss": 0.1468, "step": 15160 }, { "epoch": 1.27, "learning_rate": 1.745213301981861e-05, "loss": 0.1153, "step": 15170 }, { "epoch": 1.27, "learning_rate": 1.7450453476654354e-05, "loss": 0.13, "step": 15180 }, { "epoch": 1.28, "learning_rate": 1.7448773933490092e-05, "loss": 0.0639, "step": 15190 }, { "epoch": 1.28, "learning_rate": 1.7447094390325834e-05, "loss": 0.1963, "step": 15200 }, { "epoch": 1.28, "learning_rate": 1.7445414847161573e-05, "loss": 0.1247, "step": 15210 }, { "epoch": 1.28, "learning_rate": 1.7443735303997312e-05, "loss": 0.1009, "step": 15220 }, { "epoch": 1.28, "learning_rate": 1.7442055760833054e-05, "loss": 0.0797, "step": 15230 }, { "epoch": 1.28, "learning_rate": 1.7440376217668796e-05, "loss": 0.0265, "step": 15240 }, { "epoch": 1.28, "learning_rate": 1.7438696674504535e-05, "loss": 0.1518, "step": 15250 }, { "epoch": 1.28, "learning_rate": 1.7437017131340277e-05, "loss": 0.1653, "step": 15260 }, { "epoch": 1.28, "learning_rate": 1.743533758817602e-05, "loss": 0.1617, "step": 15270 }, { "epoch": 1.28, "learning_rate": 1.743365804501176e-05, "loss": 0.1189, "step": 15280 }, { "epoch": 1.28, "learning_rate": 1.74319785018475e-05, "loss": 0.1792, "step": 15290 }, { "epoch": 1.28, "learning_rate": 1.743029895868324e-05, "loss": 0.0851, "step": 15300 }, { "epoch": 1.29, "learning_rate": 1.7428619415518978e-05, "loss": 0.1451, "step": 15310 }, { "epoch": 1.29, "learning_rate": 1.742693987235472e-05, "loss": 0.1243, "step": 15320 }, { "epoch": 1.29, "learning_rate": 1.7425260329190462e-05, "loss": 0.1408, "step": 15330 }, { "epoch": 1.29, "learning_rate": 1.74235807860262e-05, "loss": 0.1385, "step": 15340 }, { "epoch": 1.29, "learning_rate": 1.7421901242861943e-05, "loss": 0.051, "step": 15350 }, { "epoch": 1.29, "learning_rate": 1.7420221699697682e-05, "loss": 0.0935, "step": 15360 }, { "epoch": 1.29, "learning_rate": 1.7418542156533424e-05, "loss": 0.117, "step": 15370 }, { "epoch": 1.29, "learning_rate": 1.7416862613369166e-05, "loss": 0.0768, "step": 15380 }, { "epoch": 1.29, "learning_rate": 1.7415183070204905e-05, "loss": 0.107, "step": 15390 }, { "epoch": 1.29, "learning_rate": 1.7413503527040647e-05, "loss": 0.0886, "step": 15400 }, { "epoch": 1.29, "learning_rate": 1.741182398387639e-05, "loss": 0.2295, "step": 15410 }, { "epoch": 1.29, "learning_rate": 1.741014444071213e-05, "loss": 0.1736, "step": 15420 }, { "epoch": 1.3, "learning_rate": 1.7408464897547867e-05, "loss": 0.1308, "step": 15430 }, { "epoch": 1.3, "learning_rate": 1.740678535438361e-05, "loss": 0.186, "step": 15440 }, { "epoch": 1.3, "learning_rate": 1.7405105811219348e-05, "loss": 0.1331, "step": 15450 }, { "epoch": 1.3, "learning_rate": 1.740342626805509e-05, "loss": 0.1213, "step": 15460 }, { "epoch": 1.3, "learning_rate": 1.7401746724890832e-05, "loss": 0.0801, "step": 15470 }, { "epoch": 1.3, "learning_rate": 1.740006718172657e-05, "loss": 0.1575, "step": 15480 }, { "epoch": 1.3, "learning_rate": 1.7398387638562313e-05, "loss": 0.091, "step": 15490 }, { "epoch": 1.3, "learning_rate": 1.7396708095398052e-05, "loss": 0.1175, "step": 15500 }, { "epoch": 1.3, "learning_rate": 1.7395028552233794e-05, "loss": 0.0723, "step": 15510 }, { "epoch": 1.3, "learning_rate": 1.7393349009069537e-05, "loss": 0.109, "step": 15520 }, { "epoch": 1.3, "learning_rate": 1.7391669465905275e-05, "loss": 0.1538, "step": 15530 }, { "epoch": 1.31, "learning_rate": 1.7389989922741014e-05, "loss": 0.1049, "step": 15540 }, { "epoch": 1.31, "learning_rate": 1.7388310379576756e-05, "loss": 0.1385, "step": 15550 }, { "epoch": 1.31, "learning_rate": 1.73866308364125e-05, "loss": 0.137, "step": 15560 }, { "epoch": 1.31, "learning_rate": 1.7384951293248237e-05, "loss": 0.1097, "step": 15570 }, { "epoch": 1.31, "learning_rate": 1.738327175008398e-05, "loss": 0.1306, "step": 15580 }, { "epoch": 1.31, "learning_rate": 1.7381592206919718e-05, "loss": 0.1024, "step": 15590 }, { "epoch": 1.31, "learning_rate": 1.737991266375546e-05, "loss": 0.1672, "step": 15600 }, { "epoch": 1.31, "learning_rate": 1.7378233120591203e-05, "loss": 0.1472, "step": 15610 }, { "epoch": 1.31, "learning_rate": 1.737655357742694e-05, "loss": 0.1626, "step": 15620 }, { "epoch": 1.31, "learning_rate": 1.737487403426268e-05, "loss": 0.1567, "step": 15630 }, { "epoch": 1.31, "learning_rate": 1.7373194491098422e-05, "loss": 0.0768, "step": 15640 }, { "epoch": 1.31, "learning_rate": 1.737151494793416e-05, "loss": 0.032, "step": 15650 }, { "epoch": 1.32, "learning_rate": 1.7369835404769903e-05, "loss": 0.0882, "step": 15660 }, { "epoch": 1.32, "learning_rate": 1.7368155861605645e-05, "loss": 0.0851, "step": 15670 }, { "epoch": 1.32, "learning_rate": 1.7366476318441384e-05, "loss": 0.1066, "step": 15680 }, { "epoch": 1.32, "learning_rate": 1.7364796775277126e-05, "loss": 0.1106, "step": 15690 }, { "epoch": 1.32, "learning_rate": 1.736311723211287e-05, "loss": 0.1481, "step": 15700 }, { "epoch": 1.32, "learning_rate": 1.7361437688948607e-05, "loss": 0.1803, "step": 15710 }, { "epoch": 1.32, "learning_rate": 1.735975814578435e-05, "loss": 0.0926, "step": 15720 }, { "epoch": 1.32, "learning_rate": 1.7358078602620088e-05, "loss": 0.151, "step": 15730 }, { "epoch": 1.32, "learning_rate": 1.7356399059455827e-05, "loss": 0.0697, "step": 15740 }, { "epoch": 1.32, "learning_rate": 1.735471951629157e-05, "loss": 0.0871, "step": 15750 }, { "epoch": 1.32, "learning_rate": 1.735303997312731e-05, "loss": 0.2163, "step": 15760 }, { "epoch": 1.32, "learning_rate": 1.735136042996305e-05, "loss": 0.0936, "step": 15770 }, { "epoch": 1.33, "learning_rate": 1.7349680886798792e-05, "loss": 0.113, "step": 15780 }, { "epoch": 1.33, "learning_rate": 1.7348001343634534e-05, "loss": 0.0913, "step": 15790 }, { "epoch": 1.33, "learning_rate": 1.7346321800470273e-05, "loss": 0.1058, "step": 15800 }, { "epoch": 1.33, "learning_rate": 1.7344642257306015e-05, "loss": 0.1757, "step": 15810 }, { "epoch": 1.33, "learning_rate": 1.7342962714141754e-05, "loss": 0.1133, "step": 15820 }, { "epoch": 1.33, "learning_rate": 1.7341283170977496e-05, "loss": 0.1762, "step": 15830 }, { "epoch": 1.33, "learning_rate": 1.7339603627813235e-05, "loss": 0.1057, "step": 15840 }, { "epoch": 1.33, "learning_rate": 1.7337924084648977e-05, "loss": 0.1152, "step": 15850 }, { "epoch": 1.33, "learning_rate": 1.7336244541484716e-05, "loss": 0.0951, "step": 15860 }, { "epoch": 1.33, "learning_rate": 1.7334564998320458e-05, "loss": 0.1951, "step": 15870 }, { "epoch": 1.33, "learning_rate": 1.7332885455156197e-05, "loss": 0.1323, "step": 15880 }, { "epoch": 1.33, "learning_rate": 1.733120591199194e-05, "loss": 0.1073, "step": 15890 }, { "epoch": 1.34, "learning_rate": 1.732952636882768e-05, "loss": 0.0701, "step": 15900 }, { "epoch": 1.34, "learning_rate": 1.732784682566342e-05, "loss": 0.0758, "step": 15910 }, { "epoch": 1.34, "learning_rate": 1.7326167282499162e-05, "loss": 0.1487, "step": 15920 }, { "epoch": 1.34, "learning_rate": 1.7324487739334905e-05, "loss": 0.136, "step": 15930 }, { "epoch": 1.34, "learning_rate": 1.7322808196170643e-05, "loss": 0.0912, "step": 15940 }, { "epoch": 1.34, "learning_rate": 1.7321128653006382e-05, "loss": 0.0874, "step": 15950 }, { "epoch": 1.34, "learning_rate": 1.7319449109842124e-05, "loss": 0.1112, "step": 15960 }, { "epoch": 1.34, "learning_rate": 1.7317769566677863e-05, "loss": 0.0667, "step": 15970 }, { "epoch": 1.34, "learning_rate": 1.7316090023513605e-05, "loss": 0.0832, "step": 15980 }, { "epoch": 1.34, "learning_rate": 1.7314410480349347e-05, "loss": 0.0804, "step": 15990 }, { "epoch": 1.34, "learning_rate": 1.7312730937185086e-05, "loss": 0.1191, "step": 16000 }, { "epoch": 1.34, "learning_rate": 1.731105139402083e-05, "loss": 0.1091, "step": 16010 }, { "epoch": 1.35, "learning_rate": 1.730937185085657e-05, "loss": 0.141, "step": 16020 }, { "epoch": 1.35, "learning_rate": 1.730769230769231e-05, "loss": 0.1146, "step": 16030 }, { "epoch": 1.35, "learning_rate": 1.730601276452805e-05, "loss": 0.0346, "step": 16040 }, { "epoch": 1.35, "learning_rate": 1.730433322136379e-05, "loss": 0.1097, "step": 16050 }, { "epoch": 1.35, "learning_rate": 1.730265367819953e-05, "loss": 0.11, "step": 16060 }, { "epoch": 1.35, "learning_rate": 1.730097413503527e-05, "loss": 0.0896, "step": 16070 }, { "epoch": 1.35, "learning_rate": 1.7299294591871013e-05, "loss": 0.153, "step": 16080 }, { "epoch": 1.35, "learning_rate": 1.7297615048706752e-05, "loss": 0.093, "step": 16090 }, { "epoch": 1.35, "learning_rate": 1.7295935505542494e-05, "loss": 0.1514, "step": 16100 }, { "epoch": 1.35, "learning_rate": 1.7294255962378233e-05, "loss": 0.1515, "step": 16110 }, { "epoch": 1.35, "learning_rate": 1.7292576419213975e-05, "loss": 0.1344, "step": 16120 }, { "epoch": 1.35, "learning_rate": 1.7290896876049717e-05, "loss": 0.1472, "step": 16130 }, { "epoch": 1.36, "learning_rate": 1.7289217332885456e-05, "loss": 0.1564, "step": 16140 }, { "epoch": 1.36, "learning_rate": 1.72875377897212e-05, "loss": 0.1119, "step": 16150 }, { "epoch": 1.36, "learning_rate": 1.7285858246556937e-05, "loss": 0.0814, "step": 16160 }, { "epoch": 1.36, "learning_rate": 1.728417870339268e-05, "loss": 0.1885, "step": 16170 }, { "epoch": 1.36, "learning_rate": 1.7282499160228418e-05, "loss": 0.0846, "step": 16180 }, { "epoch": 1.36, "learning_rate": 1.728081961706416e-05, "loss": 0.1878, "step": 16190 }, { "epoch": 1.36, "learning_rate": 1.72791400738999e-05, "loss": 0.1669, "step": 16200 }, { "epoch": 1.36, "learning_rate": 1.727746053073564e-05, "loss": 0.1551, "step": 16210 }, { "epoch": 1.36, "learning_rate": 1.7275780987571383e-05, "loss": 0.0914, "step": 16220 }, { "epoch": 1.36, "learning_rate": 1.7274101444407122e-05, "loss": 0.1099, "step": 16230 }, { "epoch": 1.36, "learning_rate": 1.7272421901242864e-05, "loss": 0.1587, "step": 16240 }, { "epoch": 1.36, "learning_rate": 1.7270742358078607e-05, "loss": 0.0986, "step": 16250 }, { "epoch": 1.37, "learning_rate": 1.7269062814914345e-05, "loss": 0.2079, "step": 16260 }, { "epoch": 1.37, "learning_rate": 1.7267383271750084e-05, "loss": 0.0753, "step": 16270 }, { "epoch": 1.37, "learning_rate": 1.7265703728585826e-05, "loss": 0.1107, "step": 16280 }, { "epoch": 1.37, "learning_rate": 1.7264024185421565e-05, "loss": 0.1306, "step": 16290 }, { "epoch": 1.37, "learning_rate": 1.7262344642257307e-05, "loss": 0.1351, "step": 16300 }, { "epoch": 1.37, "learning_rate": 1.726066509909305e-05, "loss": 0.0865, "step": 16310 }, { "epoch": 1.37, "learning_rate": 1.7258985555928788e-05, "loss": 0.1163, "step": 16320 }, { "epoch": 1.37, "learning_rate": 1.725730601276453e-05, "loss": 0.1291, "step": 16330 }, { "epoch": 1.37, "learning_rate": 1.725562646960027e-05, "loss": 0.1334, "step": 16340 }, { "epoch": 1.37, "learning_rate": 1.725394692643601e-05, "loss": 0.0864, "step": 16350 }, { "epoch": 1.37, "learning_rate": 1.7252267383271754e-05, "loss": 0.0857, "step": 16360 }, { "epoch": 1.37, "learning_rate": 1.7250587840107492e-05, "loss": 0.1002, "step": 16370 }, { "epoch": 1.38, "learning_rate": 1.724890829694323e-05, "loss": 0.0559, "step": 16380 }, { "epoch": 1.38, "learning_rate": 1.7247228753778973e-05, "loss": 0.1565, "step": 16390 }, { "epoch": 1.38, "learning_rate": 1.7245549210614715e-05, "loss": 0.1514, "step": 16400 }, { "epoch": 1.38, "learning_rate": 1.7243869667450454e-05, "loss": 0.1222, "step": 16410 }, { "epoch": 1.38, "learning_rate": 1.7242190124286196e-05, "loss": 0.1343, "step": 16420 }, { "epoch": 1.38, "learning_rate": 1.7240510581121935e-05, "loss": 0.1067, "step": 16430 }, { "epoch": 1.38, "learning_rate": 1.7238831037957677e-05, "loss": 0.0595, "step": 16440 }, { "epoch": 1.38, "learning_rate": 1.723715149479342e-05, "loss": 0.0996, "step": 16450 }, { "epoch": 1.38, "learning_rate": 1.7235471951629158e-05, "loss": 0.0697, "step": 16460 }, { "epoch": 1.38, "learning_rate": 1.72337924084649e-05, "loss": 0.0247, "step": 16470 }, { "epoch": 1.38, "learning_rate": 1.723211286530064e-05, "loss": 0.1347, "step": 16480 }, { "epoch": 1.38, "learning_rate": 1.7230433322136378e-05, "loss": 0.1987, "step": 16490 }, { "epoch": 1.39, "learning_rate": 1.722875377897212e-05, "loss": 0.2018, "step": 16500 }, { "epoch": 1.39, "learning_rate": 1.7227074235807862e-05, "loss": 0.0768, "step": 16510 }, { "epoch": 1.39, "learning_rate": 1.72253946926436e-05, "loss": 0.1016, "step": 16520 }, { "epoch": 1.39, "learning_rate": 1.7223715149479343e-05, "loss": 0.1278, "step": 16530 }, { "epoch": 1.39, "learning_rate": 1.7222035606315085e-05, "loss": 0.1676, "step": 16540 }, { "epoch": 1.39, "learning_rate": 1.7220356063150824e-05, "loss": 0.0811, "step": 16550 }, { "epoch": 1.39, "learning_rate": 1.7218676519986566e-05, "loss": 0.0964, "step": 16560 }, { "epoch": 1.39, "learning_rate": 1.7216996976822305e-05, "loss": 0.127, "step": 16570 }, { "epoch": 1.39, "learning_rate": 1.7215317433658047e-05, "loss": 0.1183, "step": 16580 }, { "epoch": 1.39, "learning_rate": 1.7213637890493786e-05, "loss": 0.1896, "step": 16590 }, { "epoch": 1.39, "learning_rate": 1.721195834732953e-05, "loss": 0.128, "step": 16600 }, { "epoch": 1.39, "learning_rate": 1.7210278804165267e-05, "loss": 0.0977, "step": 16610 }, { "epoch": 1.4, "learning_rate": 1.720859926100101e-05, "loss": 0.0757, "step": 16620 }, { "epoch": 1.4, "learning_rate": 1.7206919717836748e-05, "loss": 0.0758, "step": 16630 }, { "epoch": 1.4, "learning_rate": 1.720524017467249e-05, "loss": 0.1066, "step": 16640 }, { "epoch": 1.4, "learning_rate": 1.7203560631508232e-05, "loss": 0.0849, "step": 16650 }, { "epoch": 1.4, "learning_rate": 1.720188108834397e-05, "loss": 0.1701, "step": 16660 }, { "epoch": 1.4, "learning_rate": 1.7200201545179713e-05, "loss": 0.1404, "step": 16670 }, { "epoch": 1.4, "learning_rate": 1.7198522002015456e-05, "loss": 0.1504, "step": 16680 }, { "epoch": 1.4, "learning_rate": 1.7196842458851194e-05, "loss": 0.1151, "step": 16690 }, { "epoch": 1.4, "learning_rate": 1.7195162915686933e-05, "loss": 0.1276, "step": 16700 }, { "epoch": 1.4, "learning_rate": 1.7193483372522675e-05, "loss": 0.0434, "step": 16710 }, { "epoch": 1.4, "learning_rate": 1.7191803829358414e-05, "loss": 0.1366, "step": 16720 }, { "epoch": 1.4, "learning_rate": 1.7190124286194156e-05, "loss": 0.1283, "step": 16730 }, { "epoch": 1.41, "learning_rate": 1.71884447430299e-05, "loss": 0.1401, "step": 16740 }, { "epoch": 1.41, "learning_rate": 1.7186765199865637e-05, "loss": 0.1102, "step": 16750 }, { "epoch": 1.41, "learning_rate": 1.718508565670138e-05, "loss": 0.1028, "step": 16760 }, { "epoch": 1.41, "learning_rate": 1.718340611353712e-05, "loss": 0.0663, "step": 16770 }, { "epoch": 1.41, "learning_rate": 1.718172657037286e-05, "loss": 0.0686, "step": 16780 }, { "epoch": 1.41, "learning_rate": 1.7180047027208602e-05, "loss": 0.1964, "step": 16790 }, { "epoch": 1.41, "learning_rate": 1.717836748404434e-05, "loss": 0.1322, "step": 16800 }, { "epoch": 1.41, "learning_rate": 1.717668794088008e-05, "loss": 0.1097, "step": 16810 }, { "epoch": 1.41, "learning_rate": 1.7175008397715822e-05, "loss": 0.0288, "step": 16820 }, { "epoch": 1.41, "learning_rate": 1.7173328854551564e-05, "loss": 0.0975, "step": 16830 }, { "epoch": 1.41, "learning_rate": 1.7171649311387303e-05, "loss": 0.0764, "step": 16840 }, { "epoch": 1.42, "learning_rate": 1.7169969768223045e-05, "loss": 0.1689, "step": 16850 }, { "epoch": 1.42, "learning_rate": 1.7168290225058784e-05, "loss": 0.0861, "step": 16860 }, { "epoch": 1.42, "learning_rate": 1.7166610681894526e-05, "loss": 0.1465, "step": 16870 }, { "epoch": 1.42, "learning_rate": 1.716493113873027e-05, "loss": 0.0589, "step": 16880 }, { "epoch": 1.42, "learning_rate": 1.7163251595566007e-05, "loss": 0.1124, "step": 16890 }, { "epoch": 1.42, "learning_rate": 1.7161572052401746e-05, "loss": 0.102, "step": 16900 }, { "epoch": 1.42, "learning_rate": 1.7159892509237488e-05, "loss": 0.1176, "step": 16910 }, { "epoch": 1.42, "learning_rate": 1.715821296607323e-05, "loss": 0.0796, "step": 16920 }, { "epoch": 1.42, "learning_rate": 1.715653342290897e-05, "loss": 0.1052, "step": 16930 }, { "epoch": 1.42, "learning_rate": 1.715485387974471e-05, "loss": 0.053, "step": 16940 }, { "epoch": 1.42, "learning_rate": 1.715317433658045e-05, "loss": 0.104, "step": 16950 }, { "epoch": 1.42, "learning_rate": 1.7151494793416192e-05, "loss": 0.1507, "step": 16960 }, { "epoch": 1.43, "learning_rate": 1.7149815250251934e-05, "loss": 0.109, "step": 16970 }, { "epoch": 1.43, "learning_rate": 1.7148135707087673e-05, "loss": 0.0951, "step": 16980 }, { "epoch": 1.43, "learning_rate": 1.7146456163923415e-05, "loss": 0.0531, "step": 16990 }, { "epoch": 1.43, "learning_rate": 1.7144776620759158e-05, "loss": 0.14, "step": 17000 }, { "epoch": 1.43, "learning_rate": 1.7143097077594893e-05, "loss": 0.1249, "step": 17010 }, { "epoch": 1.43, "learning_rate": 1.7141417534430635e-05, "loss": 0.0928, "step": 17020 }, { "epoch": 1.43, "learning_rate": 1.7139737991266377e-05, "loss": 0.1513, "step": 17030 }, { "epoch": 1.43, "learning_rate": 1.7138058448102116e-05, "loss": 0.0456, "step": 17040 }, { "epoch": 1.43, "learning_rate": 1.7136378904937858e-05, "loss": 0.0728, "step": 17050 }, { "epoch": 1.43, "learning_rate": 1.71346993617736e-05, "loss": 0.1063, "step": 17060 }, { "epoch": 1.43, "learning_rate": 1.713301981860934e-05, "loss": 0.0712, "step": 17070 }, { "epoch": 1.43, "learning_rate": 1.713134027544508e-05, "loss": 0.1216, "step": 17080 }, { "epoch": 1.44, "learning_rate": 1.712966073228082e-05, "loss": 0.1908, "step": 17090 }, { "epoch": 1.44, "learning_rate": 1.7127981189116562e-05, "loss": 0.19, "step": 17100 }, { "epoch": 1.44, "learning_rate": 1.71263016459523e-05, "loss": 0.1201, "step": 17110 }, { "epoch": 1.44, "learning_rate": 1.7124622102788043e-05, "loss": 0.1295, "step": 17120 }, { "epoch": 1.44, "learning_rate": 1.7122942559623782e-05, "loss": 0.0613, "step": 17130 }, { "epoch": 1.44, "learning_rate": 1.7121263016459524e-05, "loss": 0.12, "step": 17140 }, { "epoch": 1.44, "learning_rate": 1.7119583473295266e-05, "loss": 0.0922, "step": 17150 }, { "epoch": 1.44, "learning_rate": 1.7117903930131005e-05, "loss": 0.0766, "step": 17160 }, { "epoch": 1.44, "learning_rate": 1.7116224386966747e-05, "loss": 0.1024, "step": 17170 }, { "epoch": 1.44, "learning_rate": 1.7114544843802486e-05, "loss": 0.061, "step": 17180 }, { "epoch": 1.44, "learning_rate": 1.711286530063823e-05, "loss": 0.145, "step": 17190 }, { "epoch": 1.44, "learning_rate": 1.711118575747397e-05, "loss": 0.0801, "step": 17200 }, { "epoch": 1.45, "learning_rate": 1.710950621430971e-05, "loss": 0.1443, "step": 17210 }, { "epoch": 1.45, "learning_rate": 1.7107826671145448e-05, "loss": 0.1088, "step": 17220 }, { "epoch": 1.45, "learning_rate": 1.710614712798119e-05, "loss": 0.1965, "step": 17230 }, { "epoch": 1.45, "learning_rate": 1.710446758481693e-05, "loss": 0.1019, "step": 17240 }, { "epoch": 1.45, "learning_rate": 1.710278804165267e-05, "loss": 0.1356, "step": 17250 }, { "epoch": 1.45, "learning_rate": 1.7101108498488413e-05, "loss": 0.1032, "step": 17260 }, { "epoch": 1.45, "learning_rate": 1.7099428955324152e-05, "loss": 0.1209, "step": 17270 }, { "epoch": 1.45, "learning_rate": 1.7097749412159894e-05, "loss": 0.1451, "step": 17280 }, { "epoch": 1.45, "learning_rate": 1.7096069868995636e-05, "loss": 0.1522, "step": 17290 }, { "epoch": 1.45, "learning_rate": 1.7094390325831375e-05, "loss": 0.0792, "step": 17300 }, { "epoch": 1.45, "learning_rate": 1.7092710782667117e-05, "loss": 0.1292, "step": 17310 }, { "epoch": 1.45, "learning_rate": 1.7091031239502856e-05, "loss": 0.1694, "step": 17320 }, { "epoch": 1.46, "learning_rate": 1.7089351696338595e-05, "loss": 0.0913, "step": 17330 }, { "epoch": 1.46, "learning_rate": 1.7087672153174337e-05, "loss": 0.128, "step": 17340 }, { "epoch": 1.46, "learning_rate": 1.708599261001008e-05, "loss": 0.0516, "step": 17350 }, { "epoch": 1.46, "learning_rate": 1.7084313066845818e-05, "loss": 0.1314, "step": 17360 }, { "epoch": 1.46, "learning_rate": 1.708263352368156e-05, "loss": 0.0714, "step": 17370 }, { "epoch": 1.46, "learning_rate": 1.7080953980517302e-05, "loss": 0.0565, "step": 17380 }, { "epoch": 1.46, "learning_rate": 1.707927443735304e-05, "loss": 0.108, "step": 17390 }, { "epoch": 1.46, "learning_rate": 1.7077594894188783e-05, "loss": 0.0435, "step": 17400 }, { "epoch": 1.46, "learning_rate": 1.7075915351024522e-05, "loss": 0.1383, "step": 17410 }, { "epoch": 1.46, "learning_rate": 1.7074235807860264e-05, "loss": 0.0856, "step": 17420 }, { "epoch": 1.46, "learning_rate": 1.7072556264696003e-05, "loss": 0.1426, "step": 17430 }, { "epoch": 1.46, "learning_rate": 1.7070876721531745e-05, "loss": 0.0412, "step": 17440 }, { "epoch": 1.47, "learning_rate": 1.7069197178367484e-05, "loss": 0.1085, "step": 17450 }, { "epoch": 1.47, "learning_rate": 1.7067517635203226e-05, "loss": 0.1285, "step": 17460 }, { "epoch": 1.47, "learning_rate": 1.7065838092038965e-05, "loss": 0.0731, "step": 17470 }, { "epoch": 1.47, "learning_rate": 1.7064158548874707e-05, "loss": 0.1701, "step": 17480 }, { "epoch": 1.47, "learning_rate": 1.706247900571045e-05, "loss": 0.1345, "step": 17490 }, { "epoch": 1.47, "learning_rate": 1.7060799462546188e-05, "loss": 0.0914, "step": 17500 }, { "epoch": 1.47, "learning_rate": 1.705911991938193e-05, "loss": 0.1066, "step": 17510 }, { "epoch": 1.47, "learning_rate": 1.7057440376217673e-05, "loss": 0.0925, "step": 17520 }, { "epoch": 1.47, "learning_rate": 1.705576083305341e-05, "loss": 0.1154, "step": 17530 }, { "epoch": 1.47, "learning_rate": 1.705408128988915e-05, "loss": 0.1866, "step": 17540 }, { "epoch": 1.47, "learning_rate": 1.7052401746724892e-05, "loss": 0.1318, "step": 17550 }, { "epoch": 1.47, "learning_rate": 1.705072220356063e-05, "loss": 0.1276, "step": 17560 }, { "epoch": 1.48, "learning_rate": 1.7049042660396373e-05, "loss": 0.1314, "step": 17570 }, { "epoch": 1.48, "learning_rate": 1.7047363117232115e-05, "loss": 0.1323, "step": 17580 }, { "epoch": 1.48, "learning_rate": 1.7045683574067854e-05, "loss": 0.1213, "step": 17590 }, { "epoch": 1.48, "learning_rate": 1.7044004030903596e-05, "loss": 0.1379, "step": 17600 }, { "epoch": 1.48, "learning_rate": 1.7042324487739335e-05, "loss": 0.0355, "step": 17610 }, { "epoch": 1.48, "learning_rate": 1.7040644944575077e-05, "loss": 0.177, "step": 17620 }, { "epoch": 1.48, "learning_rate": 1.703896540141082e-05, "loss": 0.1419, "step": 17630 }, { "epoch": 1.48, "learning_rate": 1.7037285858246558e-05, "loss": 0.1143, "step": 17640 }, { "epoch": 1.48, "learning_rate": 1.7035606315082297e-05, "loss": 0.1, "step": 17650 }, { "epoch": 1.48, "learning_rate": 1.703392677191804e-05, "loss": 0.0725, "step": 17660 }, { "epoch": 1.48, "learning_rate": 1.703224722875378e-05, "loss": 0.1058, "step": 17670 }, { "epoch": 1.48, "learning_rate": 1.703056768558952e-05, "loss": 0.0855, "step": 17680 }, { "epoch": 1.49, "learning_rate": 1.7028888142425262e-05, "loss": 0.2245, "step": 17690 }, { "epoch": 1.49, "learning_rate": 1.7027208599261e-05, "loss": 0.0588, "step": 17700 }, { "epoch": 1.49, "learning_rate": 1.7025529056096743e-05, "loss": 0.085, "step": 17710 }, { "epoch": 1.49, "learning_rate": 1.7023849512932485e-05, "loss": 0.0703, "step": 17720 }, { "epoch": 1.49, "learning_rate": 1.7022169969768224e-05, "loss": 0.1152, "step": 17730 }, { "epoch": 1.49, "learning_rate": 1.7020490426603966e-05, "loss": 0.1007, "step": 17740 }, { "epoch": 1.49, "learning_rate": 1.7018810883439705e-05, "loss": 0.1238, "step": 17750 }, { "epoch": 1.49, "learning_rate": 1.7017131340275447e-05, "loss": 0.0808, "step": 17760 }, { "epoch": 1.49, "learning_rate": 1.7015451797111186e-05, "loss": 0.1, "step": 17770 }, { "epoch": 1.49, "learning_rate": 1.7013772253946928e-05, "loss": 0.2232, "step": 17780 }, { "epoch": 1.49, "learning_rate": 1.7012092710782667e-05, "loss": 0.1203, "step": 17790 }, { "epoch": 1.49, "learning_rate": 1.701041316761841e-05, "loss": 0.1299, "step": 17800 }, { "epoch": 1.5, "learning_rate": 1.700873362445415e-05, "loss": 0.0959, "step": 17810 }, { "epoch": 1.5, "learning_rate": 1.700705408128989e-05, "loss": 0.1156, "step": 17820 }, { "epoch": 1.5, "learning_rate": 1.7005374538125632e-05, "loss": 0.1081, "step": 17830 }, { "epoch": 1.5, "learning_rate": 1.700369499496137e-05, "loss": 0.0843, "step": 17840 }, { "epoch": 1.5, "learning_rate": 1.7002015451797113e-05, "loss": 0.1058, "step": 17850 }, { "epoch": 1.5, "learning_rate": 1.7000335908632852e-05, "loss": 0.0869, "step": 17860 }, { "epoch": 1.5, "learning_rate": 1.6998656365468594e-05, "loss": 0.0867, "step": 17870 }, { "epoch": 1.5, "learning_rate": 1.6996976822304333e-05, "loss": 0.117, "step": 17880 }, { "epoch": 1.5, "learning_rate": 1.6995297279140075e-05, "loss": 0.1043, "step": 17890 }, { "epoch": 1.5, "learning_rate": 1.6993617735975817e-05, "loss": 0.0563, "step": 17900 }, { "epoch": 1.5, "learning_rate": 1.6991938192811556e-05, "loss": 0.1211, "step": 17910 }, { "epoch": 1.5, "learning_rate": 1.69902586496473e-05, "loss": 0.0768, "step": 17920 }, { "epoch": 1.51, "learning_rate": 1.6988579106483037e-05, "loss": 0.1759, "step": 17930 }, { "epoch": 1.51, "learning_rate": 1.698689956331878e-05, "loss": 0.1628, "step": 17940 }, { "epoch": 1.51, "learning_rate": 1.698522002015452e-05, "loss": 0.1355, "step": 17950 }, { "epoch": 1.51, "learning_rate": 1.698354047699026e-05, "loss": 0.045, "step": 17960 }, { "epoch": 1.51, "learning_rate": 1.6981860933826e-05, "loss": 0.1065, "step": 17970 }, { "epoch": 1.51, "learning_rate": 1.698018139066174e-05, "loss": 0.143, "step": 17980 }, { "epoch": 1.51, "learning_rate": 1.697850184749748e-05, "loss": 0.1289, "step": 17990 }, { "epoch": 1.51, "learning_rate": 1.6976822304333222e-05, "loss": 0.0542, "step": 18000 }, { "epoch": 1.51, "learning_rate": 1.6975142761168964e-05, "loss": 0.1479, "step": 18010 }, { "epoch": 1.51, "learning_rate": 1.6973463218004703e-05, "loss": 0.0878, "step": 18020 }, { "epoch": 1.51, "learning_rate": 1.6971783674840445e-05, "loss": 0.0259, "step": 18030 }, { "epoch": 1.51, "learning_rate": 1.6970104131676187e-05, "loss": 0.1346, "step": 18040 }, { "epoch": 1.52, "learning_rate": 1.6968424588511926e-05, "loss": 0.1032, "step": 18050 }, { "epoch": 1.52, "learning_rate": 1.696674504534767e-05, "loss": 0.1034, "step": 18060 }, { "epoch": 1.52, "learning_rate": 1.6965065502183407e-05, "loss": 0.0914, "step": 18070 }, { "epoch": 1.52, "learning_rate": 1.6963385959019146e-05, "loss": 0.1011, "step": 18080 }, { "epoch": 1.52, "learning_rate": 1.6961706415854888e-05, "loss": 0.1821, "step": 18090 }, { "epoch": 1.52, "learning_rate": 1.696002687269063e-05, "loss": 0.1044, "step": 18100 }, { "epoch": 1.52, "learning_rate": 1.695834732952637e-05, "loss": 0.0662, "step": 18110 }, { "epoch": 1.52, "learning_rate": 1.695666778636211e-05, "loss": 0.0434, "step": 18120 }, { "epoch": 1.52, "learning_rate": 1.6954988243197853e-05, "loss": 0.1218, "step": 18130 }, { "epoch": 1.52, "learning_rate": 1.6953308700033592e-05, "loss": 0.0663, "step": 18140 }, { "epoch": 1.52, "learning_rate": 1.6951629156869334e-05, "loss": 0.0887, "step": 18150 }, { "epoch": 1.53, "learning_rate": 1.6949949613705073e-05, "loss": 0.0796, "step": 18160 }, { "epoch": 1.53, "learning_rate": 1.6948270070540812e-05, "loss": 0.1372, "step": 18170 }, { "epoch": 1.53, "learning_rate": 1.6946590527376554e-05, "loss": 0.0906, "step": 18180 }, { "epoch": 1.53, "learning_rate": 1.6944910984212296e-05, "loss": 0.1393, "step": 18190 }, { "epoch": 1.53, "learning_rate": 1.6943231441048035e-05, "loss": 0.0932, "step": 18200 }, { "epoch": 1.53, "learning_rate": 1.6941551897883777e-05, "loss": 0.1103, "step": 18210 }, { "epoch": 1.53, "learning_rate": 1.6939872354719516e-05, "loss": 0.0685, "step": 18220 }, { "epoch": 1.53, "learning_rate": 1.6938192811555258e-05, "loss": 0.1737, "step": 18230 }, { "epoch": 1.53, "learning_rate": 1.6936513268391e-05, "loss": 0.1792, "step": 18240 }, { "epoch": 1.53, "learning_rate": 1.693483372522674e-05, "loss": 0.1148, "step": 18250 }, { "epoch": 1.53, "learning_rate": 1.693315418206248e-05, "loss": 0.128, "step": 18260 }, { "epoch": 1.53, "learning_rate": 1.6931474638898224e-05, "loss": 0.0711, "step": 18270 }, { "epoch": 1.54, "learning_rate": 1.6929795095733962e-05, "loss": 0.1335, "step": 18280 }, { "epoch": 1.54, "learning_rate": 1.69281155525697e-05, "loss": 0.0433, "step": 18290 }, { "epoch": 1.54, "learning_rate": 1.6926436009405443e-05, "loss": 0.0907, "step": 18300 }, { "epoch": 1.54, "learning_rate": 1.6924756466241182e-05, "loss": 0.2389, "step": 18310 }, { "epoch": 1.54, "learning_rate": 1.6923076923076924e-05, "loss": 0.1322, "step": 18320 }, { "epoch": 1.54, "learning_rate": 1.6921397379912666e-05, "loss": 0.0571, "step": 18330 }, { "epoch": 1.54, "learning_rate": 1.6919717836748405e-05, "loss": 0.1787, "step": 18340 }, { "epoch": 1.54, "learning_rate": 1.6918038293584147e-05, "loss": 0.1355, "step": 18350 }, { "epoch": 1.54, "learning_rate": 1.691635875041989e-05, "loss": 0.1569, "step": 18360 }, { "epoch": 1.54, "learning_rate": 1.6914679207255628e-05, "loss": 0.096, "step": 18370 }, { "epoch": 1.54, "learning_rate": 1.691299966409137e-05, "loss": 0.0898, "step": 18380 }, { "epoch": 1.54, "learning_rate": 1.691132012092711e-05, "loss": 0.1664, "step": 18390 }, { "epoch": 1.55, "learning_rate": 1.6909640577762848e-05, "loss": 0.0739, "step": 18400 }, { "epoch": 1.55, "learning_rate": 1.690796103459859e-05, "loss": 0.0716, "step": 18410 }, { "epoch": 1.55, "learning_rate": 1.6906281491434332e-05, "loss": 0.0497, "step": 18420 }, { "epoch": 1.55, "learning_rate": 1.690460194827007e-05, "loss": 0.136, "step": 18430 }, { "epoch": 1.55, "learning_rate": 1.6902922405105813e-05, "loss": 0.111, "step": 18440 }, { "epoch": 1.55, "learning_rate": 1.6901242861941552e-05, "loss": 0.1593, "step": 18450 }, { "epoch": 1.55, "learning_rate": 1.6899563318777294e-05, "loss": 0.1467, "step": 18460 }, { "epoch": 1.55, "learning_rate": 1.6897883775613036e-05, "loss": 0.1952, "step": 18470 }, { "epoch": 1.55, "learning_rate": 1.6896204232448775e-05, "loss": 0.1144, "step": 18480 }, { "epoch": 1.55, "learning_rate": 1.6894524689284514e-05, "loss": 0.1086, "step": 18490 }, { "epoch": 1.55, "learning_rate": 1.6892845146120256e-05, "loss": 0.0685, "step": 18500 }, { "epoch": 1.55, "learning_rate": 1.6891165602956e-05, "loss": 0.0906, "step": 18510 }, { "epoch": 1.56, "learning_rate": 1.6889486059791737e-05, "loss": 0.1858, "step": 18520 }, { "epoch": 1.56, "learning_rate": 1.688780651662748e-05, "loss": 0.095, "step": 18530 }, { "epoch": 1.56, "learning_rate": 1.6886126973463218e-05, "loss": 0.154, "step": 18540 }, { "epoch": 1.56, "learning_rate": 1.688444743029896e-05, "loss": 0.1161, "step": 18550 }, { "epoch": 1.56, "learning_rate": 1.6882767887134702e-05, "loss": 0.1581, "step": 18560 }, { "epoch": 1.56, "learning_rate": 1.688108834397044e-05, "loss": 0.1557, "step": 18570 }, { "epoch": 1.56, "learning_rate": 1.6879408800806183e-05, "loss": 0.1071, "step": 18580 }, { "epoch": 1.56, "learning_rate": 1.6877729257641922e-05, "loss": 0.0998, "step": 18590 }, { "epoch": 1.56, "learning_rate": 1.687604971447766e-05, "loss": 0.0674, "step": 18600 }, { "epoch": 1.56, "learning_rate": 1.6874370171313403e-05, "loss": 0.0668, "step": 18610 }, { "epoch": 1.56, "learning_rate": 1.6872690628149145e-05, "loss": 0.0792, "step": 18620 }, { "epoch": 1.56, "learning_rate": 1.6871011084984884e-05, "loss": 0.0791, "step": 18630 }, { "epoch": 1.57, "learning_rate": 1.6869331541820626e-05, "loss": 0.1202, "step": 18640 }, { "epoch": 1.57, "learning_rate": 1.686765199865637e-05, "loss": 0.1371, "step": 18650 }, { "epoch": 1.57, "learning_rate": 1.6865972455492107e-05, "loss": 0.1336, "step": 18660 }, { "epoch": 1.57, "learning_rate": 1.686429291232785e-05, "loss": 0.1354, "step": 18670 }, { "epoch": 1.57, "learning_rate": 1.6862613369163588e-05, "loss": 0.1686, "step": 18680 }, { "epoch": 1.57, "learning_rate": 1.686093382599933e-05, "loss": 0.0919, "step": 18690 }, { "epoch": 1.57, "learning_rate": 1.685925428283507e-05, "loss": 0.0766, "step": 18700 }, { "epoch": 1.57, "learning_rate": 1.685757473967081e-05, "loss": 0.1002, "step": 18710 }, { "epoch": 1.57, "learning_rate": 1.685589519650655e-05, "loss": 0.1701, "step": 18720 }, { "epoch": 1.57, "learning_rate": 1.6854215653342292e-05, "loss": 0.0785, "step": 18730 }, { "epoch": 1.57, "learning_rate": 1.6852536110178034e-05, "loss": 0.0875, "step": 18740 }, { "epoch": 1.57, "learning_rate": 1.6850856567013773e-05, "loss": 0.1087, "step": 18750 }, { "epoch": 1.58, "learning_rate": 1.6849177023849515e-05, "loss": 0.1264, "step": 18760 }, { "epoch": 1.58, "learning_rate": 1.6847497480685254e-05, "loss": 0.0731, "step": 18770 }, { "epoch": 1.58, "learning_rate": 1.6845817937520996e-05, "loss": 0.1877, "step": 18780 }, { "epoch": 1.58, "learning_rate": 1.684413839435674e-05, "loss": 0.1129, "step": 18790 }, { "epoch": 1.58, "learning_rate": 1.6842458851192477e-05, "loss": 0.1716, "step": 18800 }, { "epoch": 1.58, "learning_rate": 1.6840779308028216e-05, "loss": 0.1247, "step": 18810 }, { "epoch": 1.58, "learning_rate": 1.6839099764863958e-05, "loss": 0.1398, "step": 18820 }, { "epoch": 1.58, "learning_rate": 1.6837420221699697e-05, "loss": 0.1523, "step": 18830 }, { "epoch": 1.58, "learning_rate": 1.683574067853544e-05, "loss": 0.1323, "step": 18840 }, { "epoch": 1.58, "learning_rate": 1.683406113537118e-05, "loss": 0.132, "step": 18850 }, { "epoch": 1.58, "learning_rate": 1.683238159220692e-05, "loss": 0.0585, "step": 18860 }, { "epoch": 1.58, "learning_rate": 1.6830702049042662e-05, "loss": 0.0447, "step": 18870 }, { "epoch": 1.59, "learning_rate": 1.6829022505878404e-05, "loss": 0.1754, "step": 18880 }, { "epoch": 1.59, "learning_rate": 1.6827342962714143e-05, "loss": 0.0599, "step": 18890 }, { "epoch": 1.59, "learning_rate": 1.6825663419549885e-05, "loss": 0.0966, "step": 18900 }, { "epoch": 1.59, "learning_rate": 1.6823983876385624e-05, "loss": 0.1077, "step": 18910 }, { "epoch": 1.59, "learning_rate": 1.6822304333221363e-05, "loss": 0.222, "step": 18920 }, { "epoch": 1.59, "learning_rate": 1.6820624790057105e-05, "loss": 0.1161, "step": 18930 }, { "epoch": 1.59, "learning_rate": 1.6818945246892847e-05, "loss": 0.1286, "step": 18940 }, { "epoch": 1.59, "learning_rate": 1.6817265703728586e-05, "loss": 0.146, "step": 18950 }, { "epoch": 1.59, "learning_rate": 1.6815586160564328e-05, "loss": 0.1322, "step": 18960 }, { "epoch": 1.59, "learning_rate": 1.6813906617400067e-05, "loss": 0.1258, "step": 18970 }, { "epoch": 1.59, "learning_rate": 1.681222707423581e-05, "loss": 0.0676, "step": 18980 }, { "epoch": 1.59, "learning_rate": 1.681054753107155e-05, "loss": 0.083, "step": 18990 }, { "epoch": 1.6, "learning_rate": 1.680886798790729e-05, "loss": 0.0891, "step": 19000 }, { "epoch": 1.6, "learning_rate": 1.6807188444743032e-05, "loss": 0.1513, "step": 19010 }, { "epoch": 1.6, "learning_rate": 1.680550890157877e-05, "loss": 0.0754, "step": 19020 }, { "epoch": 1.6, "learning_rate": 1.6803829358414513e-05, "loss": 0.1272, "step": 19030 }, { "epoch": 1.6, "learning_rate": 1.6802149815250252e-05, "loss": 0.0801, "step": 19040 }, { "epoch": 1.6, "learning_rate": 1.6800470272085994e-05, "loss": 0.1654, "step": 19050 }, { "epoch": 1.6, "learning_rate": 1.6798790728921733e-05, "loss": 0.1093, "step": 19060 }, { "epoch": 1.6, "learning_rate": 1.6797111185757475e-05, "loss": 0.0858, "step": 19070 }, { "epoch": 1.6, "learning_rate": 1.6795431642593217e-05, "loss": 0.1311, "step": 19080 }, { "epoch": 1.6, "learning_rate": 1.6793752099428956e-05, "loss": 0.1329, "step": 19090 }, { "epoch": 1.6, "learning_rate": 1.67920725562647e-05, "loss": 0.1361, "step": 19100 }, { "epoch": 1.6, "learning_rate": 1.679039301310044e-05, "loss": 0.0713, "step": 19110 }, { "epoch": 1.61, "learning_rate": 1.678871346993618e-05, "loss": 0.0829, "step": 19120 }, { "epoch": 1.61, "learning_rate": 1.6787033926771918e-05, "loss": 0.0538, "step": 19130 }, { "epoch": 1.61, "learning_rate": 1.678535438360766e-05, "loss": 0.056, "step": 19140 }, { "epoch": 1.61, "learning_rate": 1.67836748404434e-05, "loss": 0.1735, "step": 19150 }, { "epoch": 1.61, "learning_rate": 1.678199529727914e-05, "loss": 0.0743, "step": 19160 }, { "epoch": 1.61, "learning_rate": 1.6780315754114883e-05, "loss": 0.1531, "step": 19170 }, { "epoch": 1.61, "learning_rate": 1.6778636210950622e-05, "loss": 0.1075, "step": 19180 }, { "epoch": 1.61, "learning_rate": 1.6776956667786364e-05, "loss": 0.165, "step": 19190 }, { "epoch": 1.61, "learning_rate": 1.6775277124622103e-05, "loss": 0.0875, "step": 19200 }, { "epoch": 1.61, "learning_rate": 1.6773597581457845e-05, "loss": 0.1207, "step": 19210 }, { "epoch": 1.61, "learning_rate": 1.6771918038293587e-05, "loss": 0.0435, "step": 19220 }, { "epoch": 1.61, "learning_rate": 1.6770238495129326e-05, "loss": 0.0896, "step": 19230 }, { "epoch": 1.62, "learning_rate": 1.6768558951965065e-05, "loss": 0.0567, "step": 19240 }, { "epoch": 1.62, "learning_rate": 1.6766879408800807e-05, "loss": 0.1382, "step": 19250 }, { "epoch": 1.62, "learning_rate": 1.676519986563655e-05, "loss": 0.1188, "step": 19260 }, { "epoch": 1.62, "learning_rate": 1.6763520322472288e-05, "loss": 0.0546, "step": 19270 }, { "epoch": 1.62, "learning_rate": 1.676184077930803e-05, "loss": 0.1587, "step": 19280 }, { "epoch": 1.62, "learning_rate": 1.676016123614377e-05, "loss": 0.1011, "step": 19290 }, { "epoch": 1.62, "learning_rate": 1.675848169297951e-05, "loss": 0.0972, "step": 19300 }, { "epoch": 1.62, "learning_rate": 1.6756802149815253e-05, "loss": 0.1061, "step": 19310 }, { "epoch": 1.62, "learning_rate": 1.6755122606650992e-05, "loss": 0.221, "step": 19320 }, { "epoch": 1.62, "learning_rate": 1.6753443063486734e-05, "loss": 0.1108, "step": 19330 }, { "epoch": 1.62, "learning_rate": 1.6751763520322473e-05, "loss": 0.1492, "step": 19340 }, { "epoch": 1.62, "learning_rate": 1.6750083977158212e-05, "loss": 0.1335, "step": 19350 }, { "epoch": 1.63, "learning_rate": 1.6748404433993954e-05, "loss": 0.135, "step": 19360 }, { "epoch": 1.63, "learning_rate": 1.6746724890829696e-05, "loss": 0.0424, "step": 19370 }, { "epoch": 1.63, "learning_rate": 1.6745045347665435e-05, "loss": 0.1301, "step": 19380 }, { "epoch": 1.63, "learning_rate": 1.6743365804501177e-05, "loss": 0.0928, "step": 19390 }, { "epoch": 1.63, "learning_rate": 1.674168626133692e-05, "loss": 0.0582, "step": 19400 }, { "epoch": 1.63, "learning_rate": 1.6740006718172658e-05, "loss": 0.1087, "step": 19410 }, { "epoch": 1.63, "learning_rate": 1.67383271750084e-05, "loss": 0.0729, "step": 19420 }, { "epoch": 1.63, "learning_rate": 1.673664763184414e-05, "loss": 0.1418, "step": 19430 }, { "epoch": 1.63, "learning_rate": 1.673496808867988e-05, "loss": 0.1271, "step": 19440 }, { "epoch": 1.63, "learning_rate": 1.673328854551562e-05, "loss": 0.151, "step": 19450 }, { "epoch": 1.63, "learning_rate": 1.6731609002351362e-05, "loss": 0.0652, "step": 19460 }, { "epoch": 1.64, "learning_rate": 1.67299294591871e-05, "loss": 0.092, "step": 19470 }, { "epoch": 1.64, "learning_rate": 1.6728249916022843e-05, "loss": 0.0993, "step": 19480 }, { "epoch": 1.64, "learning_rate": 1.6726570372858585e-05, "loss": 0.122, "step": 19490 }, { "epoch": 1.64, "learning_rate": 1.6724890829694324e-05, "loss": 0.1678, "step": 19500 }, { "epoch": 1.64, "learning_rate": 1.6723211286530066e-05, "loss": 0.1451, "step": 19510 }, { "epoch": 1.64, "learning_rate": 1.6721531743365805e-05, "loss": 0.1656, "step": 19520 }, { "epoch": 1.64, "learning_rate": 1.6719852200201547e-05, "loss": 0.0918, "step": 19530 }, { "epoch": 1.64, "learning_rate": 1.671817265703729e-05, "loss": 0.0815, "step": 19540 }, { "epoch": 1.64, "learning_rate": 1.6716493113873028e-05, "loss": 0.1479, "step": 19550 }, { "epoch": 1.64, "learning_rate": 1.6714813570708767e-05, "loss": 0.1255, "step": 19560 }, { "epoch": 1.64, "learning_rate": 1.671313402754451e-05, "loss": 0.1182, "step": 19570 }, { "epoch": 1.64, "learning_rate": 1.6711454484380248e-05, "loss": 0.0982, "step": 19580 }, { "epoch": 1.65, "learning_rate": 1.670977494121599e-05, "loss": 0.1733, "step": 19590 }, { "epoch": 1.65, "learning_rate": 1.6708095398051732e-05, "loss": 0.1487, "step": 19600 }, { "epoch": 1.65, "learning_rate": 1.670641585488747e-05, "loss": 0.1463, "step": 19610 }, { "epoch": 1.65, "learning_rate": 1.6704736311723213e-05, "loss": 0.1273, "step": 19620 }, { "epoch": 1.65, "learning_rate": 1.6703056768558955e-05, "loss": 0.049, "step": 19630 }, { "epoch": 1.65, "learning_rate": 1.6701377225394694e-05, "loss": 0.074, "step": 19640 }, { "epoch": 1.65, "learning_rate": 1.6699697682230436e-05, "loss": 0.1446, "step": 19650 }, { "epoch": 1.65, "learning_rate": 1.6698018139066175e-05, "loss": 0.2061, "step": 19660 }, { "epoch": 1.65, "learning_rate": 1.6696338595901914e-05, "loss": 0.1221, "step": 19670 }, { "epoch": 1.65, "learning_rate": 1.6694659052737656e-05, "loss": 0.1357, "step": 19680 }, { "epoch": 1.65, "learning_rate": 1.6692979509573398e-05, "loss": 0.0582, "step": 19690 }, { "epoch": 1.65, "learning_rate": 1.6691299966409137e-05, "loss": 0.2027, "step": 19700 }, { "epoch": 1.66, "learning_rate": 1.668962042324488e-05, "loss": 0.1352, "step": 19710 }, { "epoch": 1.66, "learning_rate": 1.668794088008062e-05, "loss": 0.1403, "step": 19720 }, { "epoch": 1.66, "learning_rate": 1.668626133691636e-05, "loss": 0.1544, "step": 19730 }, { "epoch": 1.66, "learning_rate": 1.6684581793752102e-05, "loss": 0.1194, "step": 19740 }, { "epoch": 1.66, "learning_rate": 1.668290225058784e-05, "loss": 0.164, "step": 19750 }, { "epoch": 1.66, "learning_rate": 1.668122270742358e-05, "loss": 0.047, "step": 19760 }, { "epoch": 1.66, "learning_rate": 1.6679543164259322e-05, "loss": 0.1254, "step": 19770 }, { "epoch": 1.66, "learning_rate": 1.6677863621095064e-05, "loss": 0.0947, "step": 19780 }, { "epoch": 1.66, "learning_rate": 1.6676184077930803e-05, "loss": 0.1351, "step": 19790 }, { "epoch": 1.66, "learning_rate": 1.6674504534766545e-05, "loss": 0.1007, "step": 19800 }, { "epoch": 1.66, "learning_rate": 1.6672824991602284e-05, "loss": 0.1006, "step": 19810 }, { "epoch": 1.66, "learning_rate": 1.6671145448438026e-05, "loss": 0.0918, "step": 19820 }, { "epoch": 1.67, "learning_rate": 1.666946590527377e-05, "loss": 0.1197, "step": 19830 }, { "epoch": 1.67, "learning_rate": 1.6667786362109507e-05, "loss": 0.1169, "step": 19840 }, { "epoch": 1.67, "learning_rate": 1.666610681894525e-05, "loss": 0.1293, "step": 19850 }, { "epoch": 1.67, "learning_rate": 1.666442727578099e-05, "loss": 0.0827, "step": 19860 }, { "epoch": 1.67, "learning_rate": 1.666274773261673e-05, "loss": 0.0426, "step": 19870 }, { "epoch": 1.67, "learning_rate": 1.666106818945247e-05, "loss": 0.0569, "step": 19880 }, { "epoch": 1.67, "learning_rate": 1.665938864628821e-05, "loss": 0.1388, "step": 19890 }, { "epoch": 1.67, "learning_rate": 1.665770910312395e-05, "loss": 0.1409, "step": 19900 }, { "epoch": 1.67, "learning_rate": 1.6656029559959692e-05, "loss": 0.217, "step": 19910 }, { "epoch": 1.67, "learning_rate": 1.6654350016795434e-05, "loss": 0.1789, "step": 19920 }, { "epoch": 1.67, "learning_rate": 1.6652670473631173e-05, "loss": 0.115, "step": 19930 }, { "epoch": 1.67, "learning_rate": 1.6650990930466915e-05, "loss": 0.0957, "step": 19940 }, { "epoch": 1.68, "learning_rate": 1.6649311387302654e-05, "loss": 0.1213, "step": 19950 }, { "epoch": 1.68, "learning_rate": 1.6647631844138396e-05, "loss": 0.1055, "step": 19960 }, { "epoch": 1.68, "learning_rate": 1.664595230097414e-05, "loss": 0.0559, "step": 19970 }, { "epoch": 1.68, "learning_rate": 1.6644272757809877e-05, "loss": 0.1312, "step": 19980 }, { "epoch": 1.68, "learning_rate": 1.6642593214645616e-05, "loss": 0.1316, "step": 19990 }, { "epoch": 1.68, "learning_rate": 1.6640913671481358e-05, "loss": 0.1139, "step": 20000 }, { "epoch": 1.68, "learning_rate": 1.66392341283171e-05, "loss": 0.0936, "step": 20010 }, { "epoch": 1.68, "learning_rate": 1.663755458515284e-05, "loss": 0.1428, "step": 20020 }, { "epoch": 1.68, "learning_rate": 1.663587504198858e-05, "loss": 0.1741, "step": 20030 }, { "epoch": 1.68, "learning_rate": 1.663419549882432e-05, "loss": 0.0486, "step": 20040 }, { "epoch": 1.68, "learning_rate": 1.6632515955660062e-05, "loss": 0.0672, "step": 20050 }, { "epoch": 1.68, "learning_rate": 1.6630836412495804e-05, "loss": 0.1439, "step": 20060 }, { "epoch": 1.69, "learning_rate": 1.6629156869331543e-05, "loss": 0.1124, "step": 20070 }, { "epoch": 1.69, "learning_rate": 1.6627477326167282e-05, "loss": 0.1693, "step": 20080 }, { "epoch": 1.69, "learning_rate": 1.6625797783003024e-05, "loss": 0.0626, "step": 20090 }, { "epoch": 1.69, "learning_rate": 1.6624118239838763e-05, "loss": 0.1811, "step": 20100 }, { "epoch": 1.69, "learning_rate": 1.6622438696674505e-05, "loss": 0.1142, "step": 20110 }, { "epoch": 1.69, "learning_rate": 1.6620759153510247e-05, "loss": 0.0754, "step": 20120 }, { "epoch": 1.69, "learning_rate": 1.6619079610345986e-05, "loss": 0.1901, "step": 20130 }, { "epoch": 1.69, "learning_rate": 1.6617400067181728e-05, "loss": 0.0338, "step": 20140 }, { "epoch": 1.69, "learning_rate": 1.661572052401747e-05, "loss": 0.1176, "step": 20150 }, { "epoch": 1.69, "learning_rate": 1.661404098085321e-05, "loss": 0.2192, "step": 20160 }, { "epoch": 1.69, "learning_rate": 1.661236143768895e-05, "loss": 0.1348, "step": 20170 }, { "epoch": 1.69, "learning_rate": 1.661068189452469e-05, "loss": 0.0411, "step": 20180 }, { "epoch": 1.7, "learning_rate": 1.660900235136043e-05, "loss": 0.1356, "step": 20190 }, { "epoch": 1.7, "learning_rate": 1.660732280819617e-05, "loss": 0.115, "step": 20200 }, { "epoch": 1.7, "learning_rate": 1.6605643265031913e-05, "loss": 0.126, "step": 20210 }, { "epoch": 1.7, "learning_rate": 1.6603963721867652e-05, "loss": 0.0798, "step": 20220 }, { "epoch": 1.7, "learning_rate": 1.6602284178703394e-05, "loss": 0.1443, "step": 20230 }, { "epoch": 1.7, "learning_rate": 1.6600604635539136e-05, "loss": 0.0777, "step": 20240 }, { "epoch": 1.7, "learning_rate": 1.6598925092374875e-05, "loss": 0.1039, "step": 20250 }, { "epoch": 1.7, "learning_rate": 1.6597245549210617e-05, "loss": 0.1675, "step": 20260 }, { "epoch": 1.7, "learning_rate": 1.6595566006046356e-05, "loss": 0.0845, "step": 20270 }, { "epoch": 1.7, "learning_rate": 1.6593886462882098e-05, "loss": 0.1235, "step": 20280 }, { "epoch": 1.7, "learning_rate": 1.6592206919717837e-05, "loss": 0.0938, "step": 20290 }, { "epoch": 1.7, "learning_rate": 1.659052737655358e-05, "loss": 0.0611, "step": 20300 }, { "epoch": 1.71, "learning_rate": 1.6588847833389318e-05, "loss": 0.1216, "step": 20310 }, { "epoch": 1.71, "learning_rate": 1.658716829022506e-05, "loss": 0.0921, "step": 20320 }, { "epoch": 1.71, "learning_rate": 1.65854887470608e-05, "loss": 0.2778, "step": 20330 }, { "epoch": 1.71, "learning_rate": 1.658380920389654e-05, "loss": 0.2225, "step": 20340 }, { "epoch": 1.71, "learning_rate": 1.6582129660732283e-05, "loss": 0.1473, "step": 20350 }, { "epoch": 1.71, "learning_rate": 1.6580450117568022e-05, "loss": 0.1253, "step": 20360 }, { "epoch": 1.71, "learning_rate": 1.6578770574403764e-05, "loss": 0.154, "step": 20370 }, { "epoch": 1.71, "learning_rate": 1.6577091031239506e-05, "loss": 0.1673, "step": 20380 }, { "epoch": 1.71, "learning_rate": 1.6575411488075245e-05, "loss": 0.128, "step": 20390 }, { "epoch": 1.71, "learning_rate": 1.6573731944910984e-05, "loss": 0.1074, "step": 20400 }, { "epoch": 1.71, "learning_rate": 1.6572052401746726e-05, "loss": 0.097, "step": 20410 }, { "epoch": 1.71, "learning_rate": 1.6570372858582465e-05, "loss": 0.143, "step": 20420 }, { "epoch": 1.72, "learning_rate": 1.6568693315418207e-05, "loss": 0.0931, "step": 20430 }, { "epoch": 1.72, "learning_rate": 1.656701377225395e-05, "loss": 0.113, "step": 20440 }, { "epoch": 1.72, "learning_rate": 1.6565334229089688e-05, "loss": 0.0548, "step": 20450 }, { "epoch": 1.72, "learning_rate": 1.656365468592543e-05, "loss": 0.1569, "step": 20460 }, { "epoch": 1.72, "learning_rate": 1.6561975142761172e-05, "loss": 0.2075, "step": 20470 }, { "epoch": 1.72, "learning_rate": 1.656029559959691e-05, "loss": 0.1492, "step": 20480 }, { "epoch": 1.72, "learning_rate": 1.6558616056432653e-05, "loss": 0.1778, "step": 20490 }, { "epoch": 1.72, "learning_rate": 1.6556936513268392e-05, "loss": 0.1079, "step": 20500 }, { "epoch": 1.72, "learning_rate": 1.655525697010413e-05, "loss": 0.0682, "step": 20510 }, { "epoch": 1.72, "learning_rate": 1.6553577426939873e-05, "loss": 0.1014, "step": 20520 }, { "epoch": 1.72, "learning_rate": 1.6551897883775615e-05, "loss": 0.0709, "step": 20530 }, { "epoch": 1.72, "learning_rate": 1.6550218340611354e-05, "loss": 0.1018, "step": 20540 }, { "epoch": 1.73, "learning_rate": 1.6548538797447096e-05, "loss": 0.1897, "step": 20550 }, { "epoch": 1.73, "learning_rate": 1.6546859254282835e-05, "loss": 0.1227, "step": 20560 }, { "epoch": 1.73, "learning_rate": 1.6545179711118577e-05, "loss": 0.0553, "step": 20570 }, { "epoch": 1.73, "learning_rate": 1.654350016795432e-05, "loss": 0.2071, "step": 20580 }, { "epoch": 1.73, "learning_rate": 1.6541820624790058e-05, "loss": 0.0553, "step": 20590 }, { "epoch": 1.73, "learning_rate": 1.65401410816258e-05, "loss": 0.1241, "step": 20600 }, { "epoch": 1.73, "learning_rate": 1.653846153846154e-05, "loss": 0.0696, "step": 20610 }, { "epoch": 1.73, "learning_rate": 1.653678199529728e-05, "loss": 0.1184, "step": 20620 }, { "epoch": 1.73, "learning_rate": 1.653510245213302e-05, "loss": 0.1346, "step": 20630 }, { "epoch": 1.73, "learning_rate": 1.6533422908968762e-05, "loss": 0.1301, "step": 20640 }, { "epoch": 1.73, "learning_rate": 1.65317433658045e-05, "loss": 0.1231, "step": 20650 }, { "epoch": 1.73, "learning_rate": 1.6530063822640243e-05, "loss": 0.0796, "step": 20660 }, { "epoch": 1.74, "learning_rate": 1.6528384279475985e-05, "loss": 0.1147, "step": 20670 }, { "epoch": 1.74, "learning_rate": 1.6526704736311724e-05, "loss": 0.2065, "step": 20680 }, { "epoch": 1.74, "learning_rate": 1.6525025193147466e-05, "loss": 0.1037, "step": 20690 }, { "epoch": 1.74, "learning_rate": 1.652334564998321e-05, "loss": 0.1266, "step": 20700 }, { "epoch": 1.74, "learning_rate": 1.6521666106818947e-05, "loss": 0.0611, "step": 20710 }, { "epoch": 1.74, "learning_rate": 1.6519986563654686e-05, "loss": 0.0876, "step": 20720 }, { "epoch": 1.74, "learning_rate": 1.6518307020490428e-05, "loss": 0.1366, "step": 20730 }, { "epoch": 1.74, "learning_rate": 1.6516627477326167e-05, "loss": 0.1639, "step": 20740 }, { "epoch": 1.74, "learning_rate": 1.651494793416191e-05, "loss": 0.1028, "step": 20750 }, { "epoch": 1.74, "learning_rate": 1.651326839099765e-05, "loss": 0.1285, "step": 20760 }, { "epoch": 1.74, "learning_rate": 1.651158884783339e-05, "loss": 0.0692, "step": 20770 }, { "epoch": 1.75, "learning_rate": 1.6509909304669132e-05, "loss": 0.0948, "step": 20780 }, { "epoch": 1.75, "learning_rate": 1.650822976150487e-05, "loss": 0.2392, "step": 20790 }, { "epoch": 1.75, "learning_rate": 1.6506550218340613e-05, "loss": 0.1399, "step": 20800 }, { "epoch": 1.75, "learning_rate": 1.6504870675176355e-05, "loss": 0.1392, "step": 20810 }, { "epoch": 1.75, "learning_rate": 1.6503191132012094e-05, "loss": 0.0739, "step": 20820 }, { "epoch": 1.75, "learning_rate": 1.6501511588847833e-05, "loss": 0.1811, "step": 20830 }, { "epoch": 1.75, "learning_rate": 1.6499832045683575e-05, "loss": 0.0731, "step": 20840 }, { "epoch": 1.75, "learning_rate": 1.6498152502519317e-05, "loss": 0.0827, "step": 20850 }, { "epoch": 1.75, "learning_rate": 1.6496472959355056e-05, "loss": 0.104, "step": 20860 }, { "epoch": 1.75, "learning_rate": 1.6494793416190798e-05, "loss": 0.1059, "step": 20870 }, { "epoch": 1.75, "learning_rate": 1.6493113873026537e-05, "loss": 0.084, "step": 20880 }, { "epoch": 1.75, "learning_rate": 1.649143432986228e-05, "loss": 0.1715, "step": 20890 }, { "epoch": 1.76, "learning_rate": 1.648975478669802e-05, "loss": 0.1546, "step": 20900 }, { "epoch": 1.76, "learning_rate": 1.648807524353376e-05, "loss": 0.1056, "step": 20910 }, { "epoch": 1.76, "learning_rate": 1.6486395700369502e-05, "loss": 0.1279, "step": 20920 }, { "epoch": 1.76, "learning_rate": 1.648471615720524e-05, "loss": 0.1169, "step": 20930 }, { "epoch": 1.76, "learning_rate": 1.648303661404098e-05, "loss": 0.0738, "step": 20940 }, { "epoch": 1.76, "learning_rate": 1.6481357070876722e-05, "loss": 0.1439, "step": 20950 }, { "epoch": 1.76, "learning_rate": 1.6479677527712464e-05, "loss": 0.1449, "step": 20960 }, { "epoch": 1.76, "learning_rate": 1.6477997984548203e-05, "loss": 0.1533, "step": 20970 }, { "epoch": 1.76, "learning_rate": 1.6476318441383945e-05, "loss": 0.0658, "step": 20980 }, { "epoch": 1.76, "learning_rate": 1.6474638898219687e-05, "loss": 0.1181, "step": 20990 }, { "epoch": 1.76, "learning_rate": 1.6472959355055426e-05, "loss": 0.0676, "step": 21000 }, { "epoch": 1.76, "learning_rate": 1.647127981189117e-05, "loss": 0.1513, "step": 21010 }, { "epoch": 1.77, "learning_rate": 1.6469600268726907e-05, "loss": 0.168, "step": 21020 }, { "epoch": 1.77, "learning_rate": 1.6467920725562646e-05, "loss": 0.1369, "step": 21030 }, { "epoch": 1.77, "learning_rate": 1.6466241182398388e-05, "loss": 0.0974, "step": 21040 }, { "epoch": 1.77, "learning_rate": 1.646456163923413e-05, "loss": 0.1424, "step": 21050 }, { "epoch": 1.77, "learning_rate": 1.646288209606987e-05, "loss": 0.076, "step": 21060 }, { "epoch": 1.77, "learning_rate": 1.646120255290561e-05, "loss": 0.0517, "step": 21070 }, { "epoch": 1.77, "learning_rate": 1.645952300974135e-05, "loss": 0.1394, "step": 21080 }, { "epoch": 1.77, "learning_rate": 1.6457843466577092e-05, "loss": 0.0509, "step": 21090 }, { "epoch": 1.77, "learning_rate": 1.6456163923412834e-05, "loss": 0.1798, "step": 21100 }, { "epoch": 1.77, "learning_rate": 1.6454484380248573e-05, "loss": 0.0832, "step": 21110 }, { "epoch": 1.77, "learning_rate": 1.6452804837084315e-05, "loss": 0.1231, "step": 21120 }, { "epoch": 1.77, "learning_rate": 1.6451125293920057e-05, "loss": 0.1362, "step": 21130 }, { "epoch": 1.78, "learning_rate": 1.6449445750755796e-05, "loss": 0.1548, "step": 21140 }, { "epoch": 1.78, "learning_rate": 1.6447766207591535e-05, "loss": 0.2052, "step": 21150 }, { "epoch": 1.78, "learning_rate": 1.6446086664427277e-05, "loss": 0.1323, "step": 21160 }, { "epoch": 1.78, "learning_rate": 1.6444407121263016e-05, "loss": 0.0757, "step": 21170 }, { "epoch": 1.78, "learning_rate": 1.6442727578098758e-05, "loss": 0.1158, "step": 21180 }, { "epoch": 1.78, "learning_rate": 1.64410480349345e-05, "loss": 0.1886, "step": 21190 }, { "epoch": 1.78, "learning_rate": 1.643936849177024e-05, "loss": 0.1301, "step": 21200 }, { "epoch": 1.78, "learning_rate": 1.643768894860598e-05, "loss": 0.1103, "step": 21210 }, { "epoch": 1.78, "learning_rate": 1.6436009405441723e-05, "loss": 0.0945, "step": 21220 }, { "epoch": 1.78, "learning_rate": 1.6434329862277462e-05, "loss": 0.0646, "step": 21230 }, { "epoch": 1.78, "learning_rate": 1.6432650319113204e-05, "loss": 0.116, "step": 21240 }, { "epoch": 1.78, "learning_rate": 1.6430970775948943e-05, "loss": 0.119, "step": 21250 }, { "epoch": 1.79, "learning_rate": 1.6429291232784682e-05, "loss": 0.1067, "step": 21260 }, { "epoch": 1.79, "learning_rate": 1.6427611689620424e-05, "loss": 0.1669, "step": 21270 }, { "epoch": 1.79, "learning_rate": 1.6425932146456166e-05, "loss": 0.0698, "step": 21280 }, { "epoch": 1.79, "learning_rate": 1.6424252603291905e-05, "loss": 0.0912, "step": 21290 }, { "epoch": 1.79, "learning_rate": 1.6422573060127647e-05, "loss": 0.0315, "step": 21300 }, { "epoch": 1.79, "learning_rate": 1.6420893516963386e-05, "loss": 0.0875, "step": 21310 }, { "epoch": 1.79, "learning_rate": 1.6419213973799128e-05, "loss": 0.1464, "step": 21320 }, { "epoch": 1.79, "learning_rate": 1.641753443063487e-05, "loss": 0.171, "step": 21330 }, { "epoch": 1.79, "learning_rate": 1.641585488747061e-05, "loss": 0.0745, "step": 21340 }, { "epoch": 1.79, "learning_rate": 1.6414175344306348e-05, "loss": 0.1771, "step": 21350 }, { "epoch": 1.79, "learning_rate": 1.641249580114209e-05, "loss": 0.15, "step": 21360 }, { "epoch": 1.79, "learning_rate": 1.6410816257977832e-05, "loss": 0.1194, "step": 21370 }, { "epoch": 1.8, "learning_rate": 1.640913671481357e-05, "loss": 0.1331, "step": 21380 }, { "epoch": 1.8, "learning_rate": 1.6407457171649313e-05, "loss": 0.1018, "step": 21390 }, { "epoch": 1.8, "learning_rate": 1.6405777628485052e-05, "loss": 0.1132, "step": 21400 }, { "epoch": 1.8, "learning_rate": 1.6404098085320794e-05, "loss": 0.1343, "step": 21410 }, { "epoch": 1.8, "learning_rate": 1.6402418542156536e-05, "loss": 0.1824, "step": 21420 }, { "epoch": 1.8, "learning_rate": 1.6400738998992275e-05, "loss": 0.2098, "step": 21430 }, { "epoch": 1.8, "learning_rate": 1.6399059455828017e-05, "loss": 0.057, "step": 21440 }, { "epoch": 1.8, "learning_rate": 1.639737991266376e-05, "loss": 0.0927, "step": 21450 }, { "epoch": 1.8, "learning_rate": 1.6395700369499495e-05, "loss": 0.1515, "step": 21460 }, { "epoch": 1.8, "learning_rate": 1.6394020826335237e-05, "loss": 0.1283, "step": 21470 }, { "epoch": 1.8, "learning_rate": 1.639234128317098e-05, "loss": 0.0808, "step": 21480 }, { "epoch": 1.8, "learning_rate": 1.6390661740006718e-05, "loss": 0.1249, "step": 21490 }, { "epoch": 1.81, "learning_rate": 1.638898219684246e-05, "loss": 0.0932, "step": 21500 }, { "epoch": 1.81, "learning_rate": 1.6387302653678202e-05, "loss": 0.1469, "step": 21510 }, { "epoch": 1.81, "learning_rate": 1.638562311051394e-05, "loss": 0.043, "step": 21520 }, { "epoch": 1.81, "learning_rate": 1.6383943567349683e-05, "loss": 0.2176, "step": 21530 }, { "epoch": 1.81, "learning_rate": 1.6382264024185422e-05, "loss": 0.1079, "step": 21540 }, { "epoch": 1.81, "learning_rate": 1.6380584481021164e-05, "loss": 0.1136, "step": 21550 }, { "epoch": 1.81, "learning_rate": 1.6378904937856903e-05, "loss": 0.1906, "step": 21560 }, { "epoch": 1.81, "learning_rate": 1.6377225394692645e-05, "loss": 0.125, "step": 21570 }, { "epoch": 1.81, "learning_rate": 1.6375545851528384e-05, "loss": 0.1138, "step": 21580 }, { "epoch": 1.81, "learning_rate": 1.6373866308364126e-05, "loss": 0.0374, "step": 21590 }, { "epoch": 1.81, "learning_rate": 1.637218676519987e-05, "loss": 0.0472, "step": 21600 }, { "epoch": 1.81, "learning_rate": 1.6370507222035607e-05, "loss": 0.1454, "step": 21610 }, { "epoch": 1.82, "learning_rate": 1.636882767887135e-05, "loss": 0.1023, "step": 21620 }, { "epoch": 1.82, "learning_rate": 1.6367148135707088e-05, "loss": 0.1437, "step": 21630 }, { "epoch": 1.82, "learning_rate": 1.636546859254283e-05, "loss": 0.1539, "step": 21640 }, { "epoch": 1.82, "learning_rate": 1.6363789049378572e-05, "loss": 0.1186, "step": 21650 }, { "epoch": 1.82, "learning_rate": 1.636210950621431e-05, "loss": 0.1787, "step": 21660 }, { "epoch": 1.82, "learning_rate": 1.636042996305005e-05, "loss": 0.1352, "step": 21670 }, { "epoch": 1.82, "learning_rate": 1.6358750419885792e-05, "loss": 0.0545, "step": 21680 }, { "epoch": 1.82, "learning_rate": 1.635707087672153e-05, "loss": 0.076, "step": 21690 }, { "epoch": 1.82, "learning_rate": 1.6355391333557273e-05, "loss": 0.1287, "step": 21700 }, { "epoch": 1.82, "learning_rate": 1.6353711790393015e-05, "loss": 0.0606, "step": 21710 }, { "epoch": 1.82, "learning_rate": 1.6352032247228754e-05, "loss": 0.1085, "step": 21720 }, { "epoch": 1.82, "learning_rate": 1.6350352704064496e-05, "loss": 0.1498, "step": 21730 }, { "epoch": 1.83, "learning_rate": 1.634867316090024e-05, "loss": 0.145, "step": 21740 }, { "epoch": 1.83, "learning_rate": 1.6346993617735977e-05, "loss": 0.0891, "step": 21750 }, { "epoch": 1.83, "learning_rate": 1.634531407457172e-05, "loss": 0.1661, "step": 21760 }, { "epoch": 1.83, "learning_rate": 1.6343634531407458e-05, "loss": 0.158, "step": 21770 }, { "epoch": 1.83, "learning_rate": 1.6341954988243197e-05, "loss": 0.1129, "step": 21780 }, { "epoch": 1.83, "learning_rate": 1.634027544507894e-05, "loss": 0.104, "step": 21790 }, { "epoch": 1.83, "learning_rate": 1.633859590191468e-05, "loss": 0.1203, "step": 21800 }, { "epoch": 1.83, "learning_rate": 1.633691635875042e-05, "loss": 0.1593, "step": 21810 }, { "epoch": 1.83, "learning_rate": 1.6335236815586162e-05, "loss": 0.1018, "step": 21820 }, { "epoch": 1.83, "learning_rate": 1.6333557272421904e-05, "loss": 0.1525, "step": 21830 }, { "epoch": 1.83, "learning_rate": 1.6331877729257643e-05, "loss": 0.139, "step": 21840 }, { "epoch": 1.83, "learning_rate": 1.6330198186093385e-05, "loss": 0.1401, "step": 21850 }, { "epoch": 1.84, "learning_rate": 1.6328518642929124e-05, "loss": 0.0819, "step": 21860 }, { "epoch": 1.84, "learning_rate": 1.6326839099764866e-05, "loss": 0.0646, "step": 21870 }, { "epoch": 1.84, "learning_rate": 1.6325159556600605e-05, "loss": 0.0997, "step": 21880 }, { "epoch": 1.84, "learning_rate": 1.6323480013436347e-05, "loss": 0.0717, "step": 21890 }, { "epoch": 1.84, "learning_rate": 1.6321800470272086e-05, "loss": 0.0752, "step": 21900 }, { "epoch": 1.84, "learning_rate": 1.6320120927107828e-05, "loss": 0.1694, "step": 21910 }, { "epoch": 1.84, "learning_rate": 1.6318441383943567e-05, "loss": 0.104, "step": 21920 }, { "epoch": 1.84, "learning_rate": 1.631676184077931e-05, "loss": 0.0711, "step": 21930 }, { "epoch": 1.84, "learning_rate": 1.631508229761505e-05, "loss": 0.1046, "step": 21940 }, { "epoch": 1.84, "learning_rate": 1.631340275445079e-05, "loss": 0.1214, "step": 21950 }, { "epoch": 1.84, "learning_rate": 1.6311723211286532e-05, "loss": 0.1388, "step": 21960 }, { "epoch": 1.84, "learning_rate": 1.6310043668122274e-05, "loss": 0.1629, "step": 21970 }, { "epoch": 1.85, "learning_rate": 1.6308364124958013e-05, "loss": 0.1071, "step": 21980 }, { "epoch": 1.85, "learning_rate": 1.6306684581793752e-05, "loss": 0.1728, "step": 21990 }, { "epoch": 1.85, "learning_rate": 1.6305005038629494e-05, "loss": 0.0792, "step": 22000 }, { "epoch": 1.85, "learning_rate": 1.6303325495465233e-05, "loss": 0.0387, "step": 22010 }, { "epoch": 1.85, "learning_rate": 1.6301645952300975e-05, "loss": 0.0793, "step": 22020 }, { "epoch": 1.85, "learning_rate": 1.6299966409136717e-05, "loss": 0.0781, "step": 22030 }, { "epoch": 1.85, "learning_rate": 1.6298286865972456e-05, "loss": 0.1203, "step": 22040 }, { "epoch": 1.85, "learning_rate": 1.6296607322808198e-05, "loss": 0.1185, "step": 22050 }, { "epoch": 1.85, "learning_rate": 1.6294927779643937e-05, "loss": 0.1067, "step": 22060 }, { "epoch": 1.85, "learning_rate": 1.629324823647968e-05, "loss": 0.075, "step": 22070 }, { "epoch": 1.85, "learning_rate": 1.629156869331542e-05, "loss": 0.0255, "step": 22080 }, { "epoch": 1.86, "learning_rate": 1.628988915015116e-05, "loss": 0.0505, "step": 22090 }, { "epoch": 1.86, "learning_rate": 1.62882096069869e-05, "loss": 0.0977, "step": 22100 }, { "epoch": 1.86, "learning_rate": 1.628653006382264e-05, "loss": 0.1111, "step": 22110 }, { "epoch": 1.86, "learning_rate": 1.6284850520658383e-05, "loss": 0.1003, "step": 22120 }, { "epoch": 1.86, "learning_rate": 1.6283170977494122e-05, "loss": 0.1326, "step": 22130 }, { "epoch": 1.86, "learning_rate": 1.6281491434329864e-05, "loss": 0.1006, "step": 22140 }, { "epoch": 1.86, "learning_rate": 1.6279811891165603e-05, "loss": 0.1425, "step": 22150 }, { "epoch": 1.86, "learning_rate": 1.6278132348001345e-05, "loss": 0.086, "step": 22160 }, { "epoch": 1.86, "learning_rate": 1.6276452804837087e-05, "loss": 0.067, "step": 22170 }, { "epoch": 1.86, "learning_rate": 1.6274773261672826e-05, "loss": 0.101, "step": 22180 }, { "epoch": 1.86, "learning_rate": 1.6273093718508568e-05, "loss": 0.1742, "step": 22190 }, { "epoch": 1.86, "learning_rate": 1.6271414175344307e-05, "loss": 0.1009, "step": 22200 }, { "epoch": 1.87, "learning_rate": 1.626973463218005e-05, "loss": 0.2212, "step": 22210 }, { "epoch": 1.87, "learning_rate": 1.6268055089015788e-05, "loss": 0.0833, "step": 22220 }, { "epoch": 1.87, "learning_rate": 1.626637554585153e-05, "loss": 0.0915, "step": 22230 }, { "epoch": 1.87, "learning_rate": 1.626469600268727e-05, "loss": 0.0543, "step": 22240 }, { "epoch": 1.87, "learning_rate": 1.626301645952301e-05, "loss": 0.0451, "step": 22250 }, { "epoch": 1.87, "learning_rate": 1.6261336916358753e-05, "loss": 0.1231, "step": 22260 }, { "epoch": 1.87, "learning_rate": 1.6259657373194492e-05, "loss": 0.1964, "step": 22270 }, { "epoch": 1.87, "learning_rate": 1.6257977830030234e-05, "loss": 0.0681, "step": 22280 }, { "epoch": 1.87, "learning_rate": 1.6256298286865973e-05, "loss": 0.1704, "step": 22290 }, { "epoch": 1.87, "learning_rate": 1.6254618743701715e-05, "loss": 0.1453, "step": 22300 }, { "epoch": 1.87, "learning_rate": 1.6252939200537454e-05, "loss": 0.0802, "step": 22310 }, { "epoch": 1.87, "learning_rate": 1.6251259657373196e-05, "loss": 0.0693, "step": 22320 }, { "epoch": 1.88, "learning_rate": 1.6249580114208935e-05, "loss": 0.1345, "step": 22330 }, { "epoch": 1.88, "learning_rate": 1.6247900571044677e-05, "loss": 0.0753, "step": 22340 }, { "epoch": 1.88, "learning_rate": 1.624622102788042e-05, "loss": 0.1491, "step": 22350 }, { "epoch": 1.88, "learning_rate": 1.6244541484716158e-05, "loss": 0.19, "step": 22360 }, { "epoch": 1.88, "learning_rate": 1.62428619415519e-05, "loss": 0.0633, "step": 22370 }, { "epoch": 1.88, "learning_rate": 1.624118239838764e-05, "loss": 0.111, "step": 22380 }, { "epoch": 1.88, "learning_rate": 1.623950285522338e-05, "loss": 0.1109, "step": 22390 }, { "epoch": 1.88, "learning_rate": 1.6237823312059123e-05, "loss": 0.0696, "step": 22400 }, { "epoch": 1.88, "learning_rate": 1.6236143768894862e-05, "loss": 0.0344, "step": 22410 }, { "epoch": 1.88, "learning_rate": 1.62344642257306e-05, "loss": 0.1187, "step": 22420 }, { "epoch": 1.88, "learning_rate": 1.6232784682566343e-05, "loss": 0.0829, "step": 22430 }, { "epoch": 1.88, "learning_rate": 1.6231105139402082e-05, "loss": 0.0885, "step": 22440 }, { "epoch": 1.89, "learning_rate": 1.6229425596237824e-05, "loss": 0.145, "step": 22450 }, { "epoch": 1.89, "learning_rate": 1.6227746053073566e-05, "loss": 0.0788, "step": 22460 }, { "epoch": 1.89, "learning_rate": 1.6226066509909305e-05, "loss": 0.1781, "step": 22470 }, { "epoch": 1.89, "learning_rate": 1.6224386966745047e-05, "loss": 0.0304, "step": 22480 }, { "epoch": 1.89, "learning_rate": 1.622270742358079e-05, "loss": 0.0894, "step": 22490 }, { "epoch": 1.89, "learning_rate": 1.6221027880416528e-05, "loss": 0.1259, "step": 22500 }, { "epoch": 1.89, "learning_rate": 1.621934833725227e-05, "loss": 0.0872, "step": 22510 }, { "epoch": 1.89, "learning_rate": 1.621766879408801e-05, "loss": 0.0815, "step": 22520 }, { "epoch": 1.89, "learning_rate": 1.6215989250923748e-05, "loss": 0.1442, "step": 22530 }, { "epoch": 1.89, "learning_rate": 1.621430970775949e-05, "loss": 0.1681, "step": 22540 }, { "epoch": 1.89, "learning_rate": 1.6212630164595232e-05, "loss": 0.0697, "step": 22550 }, { "epoch": 1.89, "learning_rate": 1.621095062143097e-05, "loss": 0.173, "step": 22560 }, { "epoch": 1.9, "learning_rate": 1.6209271078266713e-05, "loss": 0.0925, "step": 22570 }, { "epoch": 1.9, "learning_rate": 1.6207591535102455e-05, "loss": 0.0807, "step": 22580 }, { "epoch": 1.9, "learning_rate": 1.6205911991938194e-05, "loss": 0.1292, "step": 22590 }, { "epoch": 1.9, "learning_rate": 1.6204232448773936e-05, "loss": 0.0886, "step": 22600 }, { "epoch": 1.9, "learning_rate": 1.6202552905609675e-05, "loss": 0.1494, "step": 22610 }, { "epoch": 1.9, "learning_rate": 1.6200873362445414e-05, "loss": 0.1343, "step": 22620 }, { "epoch": 1.9, "learning_rate": 1.6199193819281156e-05, "loss": 0.0984, "step": 22630 }, { "epoch": 1.9, "learning_rate": 1.6197514276116898e-05, "loss": 0.1132, "step": 22640 }, { "epoch": 1.9, "learning_rate": 1.6195834732952637e-05, "loss": 0.0838, "step": 22650 }, { "epoch": 1.9, "learning_rate": 1.619415518978838e-05, "loss": 0.1532, "step": 22660 }, { "epoch": 1.9, "learning_rate": 1.6192475646624118e-05, "loss": 0.1384, "step": 22670 }, { "epoch": 1.9, "learning_rate": 1.619079610345986e-05, "loss": 0.1132, "step": 22680 }, { "epoch": 1.91, "learning_rate": 1.6189116560295602e-05, "loss": 0.1566, "step": 22690 }, { "epoch": 1.91, "learning_rate": 1.618743701713134e-05, "loss": 0.1021, "step": 22700 }, { "epoch": 1.91, "learning_rate": 1.6185757473967083e-05, "loss": 0.0804, "step": 22710 }, { "epoch": 1.91, "learning_rate": 1.6184077930802825e-05, "loss": 0.0809, "step": 22720 }, { "epoch": 1.91, "learning_rate": 1.6182398387638564e-05, "loss": 0.1326, "step": 22730 }, { "epoch": 1.91, "learning_rate": 1.6180718844474303e-05, "loss": 0.1592, "step": 22740 }, { "epoch": 1.91, "learning_rate": 1.6179039301310045e-05, "loss": 0.0598, "step": 22750 }, { "epoch": 1.91, "learning_rate": 1.6177359758145784e-05, "loss": 0.1393, "step": 22760 }, { "epoch": 1.91, "learning_rate": 1.6175680214981526e-05, "loss": 0.0427, "step": 22770 }, { "epoch": 1.91, "learning_rate": 1.6174000671817268e-05, "loss": 0.1129, "step": 22780 }, { "epoch": 1.91, "learning_rate": 1.6172321128653007e-05, "loss": 0.072, "step": 22790 }, { "epoch": 1.91, "learning_rate": 1.617064158548875e-05, "loss": 0.1221, "step": 22800 }, { "epoch": 1.92, "learning_rate": 1.616896204232449e-05, "loss": 0.2185, "step": 22810 }, { "epoch": 1.92, "learning_rate": 1.616728249916023e-05, "loss": 0.1534, "step": 22820 }, { "epoch": 1.92, "learning_rate": 1.6165602955995972e-05, "loss": 0.1066, "step": 22830 }, { "epoch": 1.92, "learning_rate": 1.616392341283171e-05, "loss": 0.1363, "step": 22840 }, { "epoch": 1.92, "learning_rate": 1.616224386966745e-05, "loss": 0.125, "step": 22850 }, { "epoch": 1.92, "learning_rate": 1.6160564326503192e-05, "loss": 0.1398, "step": 22860 }, { "epoch": 1.92, "learning_rate": 1.6158884783338934e-05, "loss": 0.192, "step": 22870 }, { "epoch": 1.92, "learning_rate": 1.6157205240174673e-05, "loss": 0.1087, "step": 22880 }, { "epoch": 1.92, "learning_rate": 1.6155525697010415e-05, "loss": 0.1019, "step": 22890 }, { "epoch": 1.92, "learning_rate": 1.6153846153846154e-05, "loss": 0.1021, "step": 22900 }, { "epoch": 1.92, "learning_rate": 1.6152166610681896e-05, "loss": 0.0614, "step": 22910 }, { "epoch": 1.92, "learning_rate": 1.615048706751764e-05, "loss": 0.1716, "step": 22920 }, { "epoch": 1.93, "learning_rate": 1.6148807524353377e-05, "loss": 0.1356, "step": 22930 }, { "epoch": 1.93, "learning_rate": 1.6147127981189116e-05, "loss": 0.0378, "step": 22940 }, { "epoch": 1.93, "learning_rate": 1.6145448438024858e-05, "loss": 0.1134, "step": 22950 }, { "epoch": 1.93, "learning_rate": 1.61437688948606e-05, "loss": 0.1203, "step": 22960 }, { "epoch": 1.93, "learning_rate": 1.614208935169634e-05, "loss": 0.0521, "step": 22970 }, { "epoch": 1.93, "learning_rate": 1.614040980853208e-05, "loss": 0.1579, "step": 22980 }, { "epoch": 1.93, "learning_rate": 1.613873026536782e-05, "loss": 0.1944, "step": 22990 }, { "epoch": 1.93, "learning_rate": 1.6137050722203562e-05, "loss": 0.1436, "step": 23000 }, { "epoch": 1.93, "learning_rate": 1.6135371179039304e-05, "loss": 0.142, "step": 23010 }, { "epoch": 1.93, "learning_rate": 1.6133691635875043e-05, "loss": 0.1543, "step": 23020 }, { "epoch": 1.93, "learning_rate": 1.6132012092710785e-05, "loss": 0.0616, "step": 23030 }, { "epoch": 1.93, "learning_rate": 1.6130332549546524e-05, "loss": 0.0922, "step": 23040 }, { "epoch": 1.94, "learning_rate": 1.6128653006382263e-05, "loss": 0.064, "step": 23050 }, { "epoch": 1.94, "learning_rate": 1.6126973463218005e-05, "loss": 0.0996, "step": 23060 }, { "epoch": 1.94, "learning_rate": 1.6125293920053747e-05, "loss": 0.0358, "step": 23070 }, { "epoch": 1.94, "learning_rate": 1.6123614376889486e-05, "loss": 0.1597, "step": 23080 }, { "epoch": 1.94, "learning_rate": 1.6121934833725228e-05, "loss": 0.1433, "step": 23090 }, { "epoch": 1.94, "learning_rate": 1.612025529056097e-05, "loss": 0.1478, "step": 23100 }, { "epoch": 1.94, "learning_rate": 1.611857574739671e-05, "loss": 0.0873, "step": 23110 }, { "epoch": 1.94, "learning_rate": 1.611689620423245e-05, "loss": 0.1735, "step": 23120 }, { "epoch": 1.94, "learning_rate": 1.611521666106819e-05, "loss": 0.1421, "step": 23130 }, { "epoch": 1.94, "learning_rate": 1.6113537117903932e-05, "loss": 0.0935, "step": 23140 }, { "epoch": 1.94, "learning_rate": 1.611185757473967e-05, "loss": 0.1405, "step": 23150 }, { "epoch": 1.94, "learning_rate": 1.6110178031575413e-05, "loss": 0.0811, "step": 23160 }, { "epoch": 1.95, "learning_rate": 1.6108498488411152e-05, "loss": 0.0959, "step": 23170 }, { "epoch": 1.95, "learning_rate": 1.6106818945246894e-05, "loss": 0.0636, "step": 23180 }, { "epoch": 1.95, "learning_rate": 1.6105139402082636e-05, "loss": 0.1728, "step": 23190 }, { "epoch": 1.95, "learning_rate": 1.6103459858918375e-05, "loss": 0.2499, "step": 23200 }, { "epoch": 1.95, "learning_rate": 1.6101780315754117e-05, "loss": 0.0857, "step": 23210 }, { "epoch": 1.95, "learning_rate": 1.6100100772589856e-05, "loss": 0.0975, "step": 23220 }, { "epoch": 1.95, "learning_rate": 1.6098421229425598e-05, "loss": 0.0604, "step": 23230 }, { "epoch": 1.95, "learning_rate": 1.609674168626134e-05, "loss": 0.0902, "step": 23240 }, { "epoch": 1.95, "learning_rate": 1.609506214309708e-05, "loss": 0.1031, "step": 23250 }, { "epoch": 1.95, "learning_rate": 1.6093382599932818e-05, "loss": 0.1301, "step": 23260 }, { "epoch": 1.95, "learning_rate": 1.609170305676856e-05, "loss": 0.0955, "step": 23270 }, { "epoch": 1.95, "learning_rate": 1.60900235136043e-05, "loss": 0.1585, "step": 23280 }, { "epoch": 1.96, "learning_rate": 1.608834397044004e-05, "loss": 0.1305, "step": 23290 }, { "epoch": 1.96, "learning_rate": 1.6086664427275783e-05, "loss": 0.2352, "step": 23300 }, { "epoch": 1.96, "learning_rate": 1.6084984884111522e-05, "loss": 0.0721, "step": 23310 }, { "epoch": 1.96, "learning_rate": 1.6083305340947264e-05, "loss": 0.0844, "step": 23320 }, { "epoch": 1.96, "learning_rate": 1.6081625797783006e-05, "loss": 0.0742, "step": 23330 }, { "epoch": 1.96, "learning_rate": 1.6079946254618745e-05, "loss": 0.2, "step": 23340 }, { "epoch": 1.96, "learning_rate": 1.6078266711454487e-05, "loss": 0.1241, "step": 23350 }, { "epoch": 1.96, "learning_rate": 1.6076587168290226e-05, "loss": 0.1504, "step": 23360 }, { "epoch": 1.96, "learning_rate": 1.6074907625125965e-05, "loss": 0.1039, "step": 23370 }, { "epoch": 1.96, "learning_rate": 1.6073228081961707e-05, "loss": 0.1249, "step": 23380 }, { "epoch": 1.96, "learning_rate": 1.607154853879745e-05, "loss": 0.1538, "step": 23390 }, { "epoch": 1.97, "learning_rate": 1.6069868995633188e-05, "loss": 0.1485, "step": 23400 }, { "epoch": 1.97, "learning_rate": 1.606818945246893e-05, "loss": 0.1279, "step": 23410 }, { "epoch": 1.97, "learning_rate": 1.606650990930467e-05, "loss": 0.1029, "step": 23420 }, { "epoch": 1.97, "learning_rate": 1.606483036614041e-05, "loss": 0.0916, "step": 23430 }, { "epoch": 1.97, "learning_rate": 1.6063150822976153e-05, "loss": 0.0764, "step": 23440 }, { "epoch": 1.97, "learning_rate": 1.6061471279811892e-05, "loss": 0.0384, "step": 23450 }, { "epoch": 1.97, "learning_rate": 1.6059791736647634e-05, "loss": 0.1078, "step": 23460 }, { "epoch": 1.97, "learning_rate": 1.6058112193483373e-05, "loss": 0.1361, "step": 23470 }, { "epoch": 1.97, "learning_rate": 1.6056432650319115e-05, "loss": 0.1289, "step": 23480 }, { "epoch": 1.97, "learning_rate": 1.6054753107154854e-05, "loss": 0.0949, "step": 23490 }, { "epoch": 1.97, "learning_rate": 1.6053073563990596e-05, "loss": 0.0851, "step": 23500 }, { "epoch": 1.97, "learning_rate": 1.6051394020826335e-05, "loss": 0.1401, "step": 23510 }, { "epoch": 1.98, "learning_rate": 1.6049714477662077e-05, "loss": 0.0639, "step": 23520 }, { "epoch": 1.98, "learning_rate": 1.604803493449782e-05, "loss": 0.1217, "step": 23530 }, { "epoch": 1.98, "learning_rate": 1.6046355391333558e-05, "loss": 0.1473, "step": 23540 }, { "epoch": 1.98, "learning_rate": 1.60446758481693e-05, "loss": 0.1743, "step": 23550 }, { "epoch": 1.98, "learning_rate": 1.6042996305005042e-05, "loss": 0.1075, "step": 23560 }, { "epoch": 1.98, "learning_rate": 1.604131676184078e-05, "loss": 0.1137, "step": 23570 }, { "epoch": 1.98, "learning_rate": 1.603963721867652e-05, "loss": 0.1174, "step": 23580 }, { "epoch": 1.98, "learning_rate": 1.6037957675512262e-05, "loss": 0.0622, "step": 23590 }, { "epoch": 1.98, "learning_rate": 1.6036278132348e-05, "loss": 0.0946, "step": 23600 }, { "epoch": 1.98, "learning_rate": 1.6034598589183743e-05, "loss": 0.1616, "step": 23610 }, { "epoch": 1.98, "learning_rate": 1.6032919046019485e-05, "loss": 0.0682, "step": 23620 }, { "epoch": 1.98, "learning_rate": 1.6031239502855224e-05, "loss": 0.0777, "step": 23630 }, { "epoch": 1.99, "learning_rate": 1.6029559959690966e-05, "loss": 0.1106, "step": 23640 }, { "epoch": 1.99, "learning_rate": 1.6027880416526705e-05, "loss": 0.0788, "step": 23650 }, { "epoch": 1.99, "learning_rate": 1.6026200873362447e-05, "loss": 0.125, "step": 23660 }, { "epoch": 1.99, "learning_rate": 1.602452133019819e-05, "loss": 0.121, "step": 23670 }, { "epoch": 1.99, "learning_rate": 1.6022841787033928e-05, "loss": 0.1223, "step": 23680 }, { "epoch": 1.99, "learning_rate": 1.6021162243869667e-05, "loss": 0.1162, "step": 23690 }, { "epoch": 1.99, "learning_rate": 1.601948270070541e-05, "loss": 0.1162, "step": 23700 }, { "epoch": 1.99, "learning_rate": 1.601780315754115e-05, "loss": 0.0772, "step": 23710 }, { "epoch": 1.99, "learning_rate": 1.601612361437689e-05, "loss": 0.1491, "step": 23720 }, { "epoch": 1.99, "learning_rate": 1.6014444071212632e-05, "loss": 0.1541, "step": 23730 }, { "epoch": 1.99, "learning_rate": 1.601276452804837e-05, "loss": 0.1585, "step": 23740 }, { "epoch": 1.99, "learning_rate": 1.6011084984884113e-05, "loss": 0.1153, "step": 23750 }, { "epoch": 2.0, "learning_rate": 1.6009405441719855e-05, "loss": 0.1072, "step": 23760 }, { "epoch": 2.0, "learning_rate": 1.6007725898555594e-05, "loss": 0.1434, "step": 23770 }, { "epoch": 2.0, "learning_rate": 1.6006046355391336e-05, "loss": 0.0811, "step": 23780 }, { "epoch": 2.0, "learning_rate": 1.6004366812227075e-05, "loss": 0.1323, "step": 23790 }, { "epoch": 2.0, "learning_rate": 1.6002687269062814e-05, "loss": 0.0424, "step": 23800 }, { "epoch": 2.0, "learning_rate": 1.6001007725898556e-05, "loss": 0.0808, "step": 23810 }, { "epoch": 2.0, "eval_FN": 690, "eval_FP": 319, "eval_TN": 21395, "eval_TP": 1412, "eval_accuracy": 0.9576335236815586, "eval_f1": 0.7367597182363683, "eval_loss": 0.13676545023918152, "eval_precision": 0.8157134604274986, "eval_recall": 0.6717411988582302, "eval_runtime": 163.9732, "eval_samples_per_second": 145.243, "eval_steps_per_second": 9.081, "step": 23816 }, { "epoch": 2.0, "learning_rate": 1.5999328182734298e-05, "loss": 0.1814, "step": 23820 }, { "epoch": 2.0, "learning_rate": 1.5997648639570037e-05, "loss": 0.1165, "step": 23830 }, { "epoch": 2.0, "learning_rate": 1.599596909640578e-05, "loss": 0.0308, "step": 23840 }, { "epoch": 2.0, "learning_rate": 1.599428955324152e-05, "loss": 0.1406, "step": 23850 }, { "epoch": 2.0, "learning_rate": 1.599261001007726e-05, "loss": 0.0697, "step": 23860 }, { "epoch": 2.0, "learning_rate": 1.5990930466913002e-05, "loss": 0.097, "step": 23870 }, { "epoch": 2.01, "learning_rate": 1.598925092374874e-05, "loss": 0.1106, "step": 23880 }, { "epoch": 2.01, "learning_rate": 1.5987571380584483e-05, "loss": 0.1053, "step": 23890 }, { "epoch": 2.01, "learning_rate": 1.5985891837420222e-05, "loss": 0.0806, "step": 23900 }, { "epoch": 2.01, "learning_rate": 1.5984212294255964e-05, "loss": 0.0466, "step": 23910 }, { "epoch": 2.01, "learning_rate": 1.5982532751091703e-05, "loss": 0.0809, "step": 23920 }, { "epoch": 2.01, "learning_rate": 1.5980853207927445e-05, "loss": 0.1263, "step": 23930 }, { "epoch": 2.01, "learning_rate": 1.5979173664763187e-05, "loss": 0.0698, "step": 23940 }, { "epoch": 2.01, "learning_rate": 1.5977494121598926e-05, "loss": 0.0575, "step": 23950 }, { "epoch": 2.01, "learning_rate": 1.5975814578434668e-05, "loss": 0.1005, "step": 23960 }, { "epoch": 2.01, "learning_rate": 1.5974135035270407e-05, "loss": 0.1013, "step": 23970 }, { "epoch": 2.01, "learning_rate": 1.597245549210615e-05, "loss": 0.0954, "step": 23980 }, { "epoch": 2.01, "learning_rate": 1.597077594894189e-05, "loss": 0.0608, "step": 23990 }, { "epoch": 2.02, "learning_rate": 1.596909640577763e-05, "loss": 0.0697, "step": 24000 }, { "epoch": 2.02, "learning_rate": 1.596741686261337e-05, "loss": 0.0598, "step": 24010 }, { "epoch": 2.02, "learning_rate": 1.596573731944911e-05, "loss": 0.0492, "step": 24020 }, { "epoch": 2.02, "learning_rate": 1.596405777628485e-05, "loss": 0.0516, "step": 24030 }, { "epoch": 2.02, "learning_rate": 1.5962378233120592e-05, "loss": 0.1086, "step": 24040 }, { "epoch": 2.02, "learning_rate": 1.5960698689956334e-05, "loss": 0.1434, "step": 24050 }, { "epoch": 2.02, "learning_rate": 1.5959019146792073e-05, "loss": 0.036, "step": 24060 }, { "epoch": 2.02, "learning_rate": 1.5957339603627815e-05, "loss": 0.0418, "step": 24070 }, { "epoch": 2.02, "learning_rate": 1.5955660060463557e-05, "loss": 0.0833, "step": 24080 }, { "epoch": 2.02, "learning_rate": 1.5953980517299296e-05, "loss": 0.0817, "step": 24090 }, { "epoch": 2.02, "learning_rate": 1.5952300974135038e-05, "loss": 0.0759, "step": 24100 }, { "epoch": 2.02, "learning_rate": 1.5950621430970777e-05, "loss": 0.1041, "step": 24110 }, { "epoch": 2.03, "learning_rate": 1.5948941887806516e-05, "loss": 0.1543, "step": 24120 }, { "epoch": 2.03, "learning_rate": 1.5947262344642258e-05, "loss": 0.0533, "step": 24130 }, { "epoch": 2.03, "learning_rate": 1.5945582801478e-05, "loss": 0.1073, "step": 24140 }, { "epoch": 2.03, "learning_rate": 1.594390325831374e-05, "loss": 0.1228, "step": 24150 }, { "epoch": 2.03, "learning_rate": 1.594222371514948e-05, "loss": 0.0714, "step": 24160 }, { "epoch": 2.03, "learning_rate": 1.5940544171985223e-05, "loss": 0.0801, "step": 24170 }, { "epoch": 2.03, "learning_rate": 1.5938864628820962e-05, "loss": 0.0784, "step": 24180 }, { "epoch": 2.03, "learning_rate": 1.5937185085656704e-05, "loss": 0.0757, "step": 24190 }, { "epoch": 2.03, "learning_rate": 1.5935505542492443e-05, "loss": 0.0142, "step": 24200 }, { "epoch": 2.03, "learning_rate": 1.5933825999328182e-05, "loss": 0.1749, "step": 24210 }, { "epoch": 2.03, "learning_rate": 1.5932146456163924e-05, "loss": 0.1702, "step": 24220 }, { "epoch": 2.03, "learning_rate": 1.5930466912999666e-05, "loss": 0.1088, "step": 24230 }, { "epoch": 2.04, "learning_rate": 1.5928787369835405e-05, "loss": 0.0876, "step": 24240 }, { "epoch": 2.04, "learning_rate": 1.5927107826671147e-05, "loss": 0.1374, "step": 24250 }, { "epoch": 2.04, "learning_rate": 1.5925428283506886e-05, "loss": 0.0614, "step": 24260 }, { "epoch": 2.04, "learning_rate": 1.5923748740342628e-05, "loss": 0.0355, "step": 24270 }, { "epoch": 2.04, "learning_rate": 1.592206919717837e-05, "loss": 0.1065, "step": 24280 }, { "epoch": 2.04, "learning_rate": 1.592038965401411e-05, "loss": 0.0676, "step": 24290 }, { "epoch": 2.04, "learning_rate": 1.591871011084985e-05, "loss": 0.1229, "step": 24300 }, { "epoch": 2.04, "learning_rate": 1.5917030567685593e-05, "loss": 0.1281, "step": 24310 }, { "epoch": 2.04, "learning_rate": 1.5915351024521332e-05, "loss": 0.1337, "step": 24320 }, { "epoch": 2.04, "learning_rate": 1.591367148135707e-05, "loss": 0.0662, "step": 24330 }, { "epoch": 2.04, "learning_rate": 1.5911991938192813e-05, "loss": 0.0977, "step": 24340 }, { "epoch": 2.04, "learning_rate": 1.5910312395028552e-05, "loss": 0.0651, "step": 24350 }, { "epoch": 2.05, "learning_rate": 1.5908632851864294e-05, "loss": 0.1393, "step": 24360 }, { "epoch": 2.05, "learning_rate": 1.5906953308700036e-05, "loss": 0.0846, "step": 24370 }, { "epoch": 2.05, "learning_rate": 1.5905273765535775e-05, "loss": 0.1026, "step": 24380 }, { "epoch": 2.05, "learning_rate": 1.5903594222371517e-05, "loss": 0.1525, "step": 24390 }, { "epoch": 2.05, "learning_rate": 1.5901914679207256e-05, "loss": 0.0953, "step": 24400 }, { "epoch": 2.05, "learning_rate": 1.5900235136042998e-05, "loss": 0.0769, "step": 24410 }, { "epoch": 2.05, "learning_rate": 1.5898555592878737e-05, "loss": 0.0775, "step": 24420 }, { "epoch": 2.05, "learning_rate": 1.589687604971448e-05, "loss": 0.0956, "step": 24430 }, { "epoch": 2.05, "learning_rate": 1.5895196506550218e-05, "loss": 0.0409, "step": 24440 }, { "epoch": 2.05, "learning_rate": 1.589351696338596e-05, "loss": 0.0902, "step": 24450 }, { "epoch": 2.05, "learning_rate": 1.5891837420221702e-05, "loss": 0.1694, "step": 24460 }, { "epoch": 2.05, "learning_rate": 1.589015787705744e-05, "loss": 0.0812, "step": 24470 }, { "epoch": 2.06, "learning_rate": 1.5888478333893183e-05, "loss": 0.0719, "step": 24480 }, { "epoch": 2.06, "learning_rate": 1.5886798790728922e-05, "loss": 0.1167, "step": 24490 }, { "epoch": 2.06, "learning_rate": 1.5885119247564664e-05, "loss": 0.1311, "step": 24500 }, { "epoch": 2.06, "learning_rate": 1.5883439704400406e-05, "loss": 0.1347, "step": 24510 }, { "epoch": 2.06, "learning_rate": 1.5881760161236145e-05, "loss": 0.0635, "step": 24520 }, { "epoch": 2.06, "learning_rate": 1.5880080618071884e-05, "loss": 0.1456, "step": 24530 }, { "epoch": 2.06, "learning_rate": 1.5878401074907626e-05, "loss": 0.1135, "step": 24540 }, { "epoch": 2.06, "learning_rate": 1.5876721531743365e-05, "loss": 0.0744, "step": 24550 }, { "epoch": 2.06, "learning_rate": 1.5875041988579107e-05, "loss": 0.0747, "step": 24560 }, { "epoch": 2.06, "learning_rate": 1.587336244541485e-05, "loss": 0.1428, "step": 24570 }, { "epoch": 2.06, "learning_rate": 1.5871682902250588e-05, "loss": 0.0838, "step": 24580 }, { "epoch": 2.06, "learning_rate": 1.587000335908633e-05, "loss": 0.1236, "step": 24590 }, { "epoch": 2.07, "learning_rate": 1.5868323815922072e-05, "loss": 0.0724, "step": 24600 }, { "epoch": 2.07, "learning_rate": 1.586664427275781e-05, "loss": 0.2326, "step": 24610 }, { "epoch": 2.07, "learning_rate": 1.5864964729593553e-05, "loss": 0.0422, "step": 24620 }, { "epoch": 2.07, "learning_rate": 1.5863285186429292e-05, "loss": 0.0992, "step": 24630 }, { "epoch": 2.07, "learning_rate": 1.586160564326503e-05, "loss": 0.0782, "step": 24640 }, { "epoch": 2.07, "learning_rate": 1.5859926100100773e-05, "loss": 0.0676, "step": 24650 }, { "epoch": 2.07, "learning_rate": 1.5858246556936515e-05, "loss": 0.1897, "step": 24660 }, { "epoch": 2.07, "learning_rate": 1.5856567013772254e-05, "loss": 0.1101, "step": 24670 }, { "epoch": 2.07, "learning_rate": 1.5854887470607996e-05, "loss": 0.0549, "step": 24680 }, { "epoch": 2.07, "learning_rate": 1.5853207927443738e-05, "loss": 0.061, "step": 24690 }, { "epoch": 2.07, "learning_rate": 1.5851528384279477e-05, "loss": 0.126, "step": 24700 }, { "epoch": 2.08, "learning_rate": 1.584984884111522e-05, "loss": 0.1313, "step": 24710 }, { "epoch": 2.08, "learning_rate": 1.5848169297950958e-05, "loss": 0.0968, "step": 24720 }, { "epoch": 2.08, "learning_rate": 1.58464897547867e-05, "loss": 0.1545, "step": 24730 }, { "epoch": 2.08, "learning_rate": 1.584481021162244e-05, "loss": 0.104, "step": 24740 }, { "epoch": 2.08, "learning_rate": 1.584313066845818e-05, "loss": 0.0516, "step": 24750 }, { "epoch": 2.08, "learning_rate": 1.584145112529392e-05, "loss": 0.0818, "step": 24760 }, { "epoch": 2.08, "learning_rate": 1.5839771582129662e-05, "loss": 0.0713, "step": 24770 }, { "epoch": 2.08, "learning_rate": 1.58380920389654e-05, "loss": 0.1781, "step": 24780 }, { "epoch": 2.08, "learning_rate": 1.5836412495801143e-05, "loss": 0.0625, "step": 24790 }, { "epoch": 2.08, "learning_rate": 1.5834732952636885e-05, "loss": 0.0972, "step": 24800 }, { "epoch": 2.08, "learning_rate": 1.5833053409472624e-05, "loss": 0.0539, "step": 24810 }, { "epoch": 2.08, "learning_rate": 1.5831373866308366e-05, "loss": 0.1681, "step": 24820 }, { "epoch": 2.09, "learning_rate": 1.582969432314411e-05, "loss": 0.0949, "step": 24830 }, { "epoch": 2.09, "learning_rate": 1.5828014779979847e-05, "loss": 0.1459, "step": 24840 }, { "epoch": 2.09, "learning_rate": 1.5826335236815586e-05, "loss": 0.0935, "step": 24850 }, { "epoch": 2.09, "learning_rate": 1.5824655693651328e-05, "loss": 0.1037, "step": 24860 }, { "epoch": 2.09, "learning_rate": 1.5822976150487067e-05, "loss": 0.1287, "step": 24870 }, { "epoch": 2.09, "learning_rate": 1.582129660732281e-05, "loss": 0.0801, "step": 24880 }, { "epoch": 2.09, "learning_rate": 1.581961706415855e-05, "loss": 0.1396, "step": 24890 }, { "epoch": 2.09, "learning_rate": 1.581793752099429e-05, "loss": 0.0478, "step": 24900 }, { "epoch": 2.09, "learning_rate": 1.5816257977830032e-05, "loss": 0.1181, "step": 24910 }, { "epoch": 2.09, "learning_rate": 1.5814578434665774e-05, "loss": 0.1361, "step": 24920 }, { "epoch": 2.09, "learning_rate": 1.5812898891501513e-05, "loss": 0.0509, "step": 24930 }, { "epoch": 2.09, "learning_rate": 1.5811219348337255e-05, "loss": 0.0736, "step": 24940 }, { "epoch": 2.1, "learning_rate": 1.5809539805172994e-05, "loss": 0.0853, "step": 24950 }, { "epoch": 2.1, "learning_rate": 1.5807860262008733e-05, "loss": 0.1482, "step": 24960 }, { "epoch": 2.1, "learning_rate": 1.5806180718844475e-05, "loss": 0.0367, "step": 24970 }, { "epoch": 2.1, "learning_rate": 1.5804501175680217e-05, "loss": 0.1317, "step": 24980 }, { "epoch": 2.1, "learning_rate": 1.5802821632515956e-05, "loss": 0.0236, "step": 24990 }, { "epoch": 2.1, "learning_rate": 1.5801142089351698e-05, "loss": 0.1092, "step": 25000 }, { "epoch": 2.1, "learning_rate": 1.5799462546187437e-05, "loss": 0.0742, "step": 25010 }, { "epoch": 2.1, "learning_rate": 1.579778300302318e-05, "loss": 0.1084, "step": 25020 }, { "epoch": 2.1, "learning_rate": 1.579610345985892e-05, "loss": 0.1504, "step": 25030 }, { "epoch": 2.1, "learning_rate": 1.579442391669466e-05, "loss": 0.0673, "step": 25040 }, { "epoch": 2.1, "learning_rate": 1.5792744373530402e-05, "loss": 0.1532, "step": 25050 }, { "epoch": 2.1, "learning_rate": 1.579106483036614e-05, "loss": 0.0936, "step": 25060 }, { "epoch": 2.11, "learning_rate": 1.5789385287201883e-05, "loss": 0.0547, "step": 25070 }, { "epoch": 2.11, "learning_rate": 1.5787705744037622e-05, "loss": 0.0709, "step": 25080 }, { "epoch": 2.11, "learning_rate": 1.5786026200873364e-05, "loss": 0.0503, "step": 25090 }, { "epoch": 2.11, "learning_rate": 1.5784346657709103e-05, "loss": 0.0248, "step": 25100 }, { "epoch": 2.11, "learning_rate": 1.5782667114544845e-05, "loss": 0.2483, "step": 25110 }, { "epoch": 2.11, "learning_rate": 1.5780987571380587e-05, "loss": 0.0908, "step": 25120 }, { "epoch": 2.11, "learning_rate": 1.5779308028216326e-05, "loss": 0.0755, "step": 25130 }, { "epoch": 2.11, "learning_rate": 1.5777628485052068e-05, "loss": 0.1026, "step": 25140 }, { "epoch": 2.11, "learning_rate": 1.5775948941887807e-05, "loss": 0.1008, "step": 25150 }, { "epoch": 2.11, "learning_rate": 1.577426939872355e-05, "loss": 0.1337, "step": 25160 }, { "epoch": 2.11, "learning_rate": 1.5772589855559288e-05, "loss": 0.0664, "step": 25170 }, { "epoch": 2.11, "learning_rate": 1.577091031239503e-05, "loss": 0.0601, "step": 25180 }, { "epoch": 2.12, "learning_rate": 1.576923076923077e-05, "loss": 0.051, "step": 25190 }, { "epoch": 2.12, "learning_rate": 1.576755122606651e-05, "loss": 0.0964, "step": 25200 }, { "epoch": 2.12, "learning_rate": 1.5765871682902253e-05, "loss": 0.0683, "step": 25210 }, { "epoch": 2.12, "learning_rate": 1.5764192139737992e-05, "loss": 0.1931, "step": 25220 }, { "epoch": 2.12, "learning_rate": 1.5762512596573734e-05, "loss": 0.1336, "step": 25230 }, { "epoch": 2.12, "learning_rate": 1.5760833053409473e-05, "loss": 0.075, "step": 25240 }, { "epoch": 2.12, "learning_rate": 1.5759153510245215e-05, "loss": 0.1047, "step": 25250 }, { "epoch": 2.12, "learning_rate": 1.5757473967080957e-05, "loss": 0.168, "step": 25260 }, { "epoch": 2.12, "learning_rate": 1.5755794423916696e-05, "loss": 0.1044, "step": 25270 }, { "epoch": 2.12, "learning_rate": 1.5754114880752435e-05, "loss": 0.1125, "step": 25280 }, { "epoch": 2.12, "learning_rate": 1.5752435337588177e-05, "loss": 0.1073, "step": 25290 }, { "epoch": 2.12, "learning_rate": 1.575075579442392e-05, "loss": 0.0617, "step": 25300 }, { "epoch": 2.13, "learning_rate": 1.5749076251259658e-05, "loss": 0.1268, "step": 25310 }, { "epoch": 2.13, "learning_rate": 1.57473967080954e-05, "loss": 0.1305, "step": 25320 }, { "epoch": 2.13, "learning_rate": 1.574571716493114e-05, "loss": 0.088, "step": 25330 }, { "epoch": 2.13, "learning_rate": 1.574403762176688e-05, "loss": 0.0234, "step": 25340 }, { "epoch": 2.13, "learning_rate": 1.5742358078602623e-05, "loss": 0.1766, "step": 25350 }, { "epoch": 2.13, "learning_rate": 1.5740678535438362e-05, "loss": 0.1646, "step": 25360 }, { "epoch": 2.13, "learning_rate": 1.5738998992274104e-05, "loss": 0.1428, "step": 25370 }, { "epoch": 2.13, "learning_rate": 1.5737319449109843e-05, "loss": 0.045, "step": 25380 }, { "epoch": 2.13, "learning_rate": 1.5735639905945582e-05, "loss": 0.0722, "step": 25390 }, { "epoch": 2.13, "learning_rate": 1.5733960362781324e-05, "loss": 0.0277, "step": 25400 }, { "epoch": 2.13, "learning_rate": 1.5732280819617066e-05, "loss": 0.2469, "step": 25410 }, { "epoch": 2.13, "learning_rate": 1.5730601276452805e-05, "loss": 0.0911, "step": 25420 }, { "epoch": 2.14, "learning_rate": 1.5728921733288547e-05, "loss": 0.0454, "step": 25430 }, { "epoch": 2.14, "learning_rate": 1.572724219012429e-05, "loss": 0.0476, "step": 25440 }, { "epoch": 2.14, "learning_rate": 1.5725562646960028e-05, "loss": 0.0599, "step": 25450 }, { "epoch": 2.14, "learning_rate": 1.572388310379577e-05, "loss": 0.1325, "step": 25460 }, { "epoch": 2.14, "learning_rate": 1.572220356063151e-05, "loss": 0.1103, "step": 25470 }, { "epoch": 2.14, "learning_rate": 1.5720524017467248e-05, "loss": 0.0393, "step": 25480 }, { "epoch": 2.14, "learning_rate": 1.571884447430299e-05, "loss": 0.0646, "step": 25490 }, { "epoch": 2.14, "learning_rate": 1.5717164931138732e-05, "loss": 0.1447, "step": 25500 }, { "epoch": 2.14, "learning_rate": 1.571548538797447e-05, "loss": 0.076, "step": 25510 }, { "epoch": 2.14, "learning_rate": 1.5713805844810213e-05, "loss": 0.0834, "step": 25520 }, { "epoch": 2.14, "learning_rate": 1.5712126301645952e-05, "loss": 0.0714, "step": 25530 }, { "epoch": 2.14, "learning_rate": 1.5710446758481694e-05, "loss": 0.0192, "step": 25540 }, { "epoch": 2.15, "learning_rate": 1.5708767215317436e-05, "loss": 0.0477, "step": 25550 }, { "epoch": 2.15, "learning_rate": 1.5707087672153175e-05, "loss": 0.0639, "step": 25560 }, { "epoch": 2.15, "learning_rate": 1.5705408128988917e-05, "loss": 0.1312, "step": 25570 }, { "epoch": 2.15, "learning_rate": 1.570372858582466e-05, "loss": 0.1697, "step": 25580 }, { "epoch": 2.15, "learning_rate": 1.5702049042660398e-05, "loss": 0.1766, "step": 25590 }, { "epoch": 2.15, "learning_rate": 1.5700369499496137e-05, "loss": 0.0698, "step": 25600 }, { "epoch": 2.15, "learning_rate": 1.569868995633188e-05, "loss": 0.1224, "step": 25610 }, { "epoch": 2.15, "learning_rate": 1.5697010413167618e-05, "loss": 0.0503, "step": 25620 }, { "epoch": 2.15, "learning_rate": 1.569533087000336e-05, "loss": 0.1202, "step": 25630 }, { "epoch": 2.15, "learning_rate": 1.5693651326839102e-05, "loss": 0.089, "step": 25640 }, { "epoch": 2.15, "learning_rate": 1.569197178367484e-05, "loss": 0.1094, "step": 25650 }, { "epoch": 2.15, "learning_rate": 1.5690292240510583e-05, "loss": 0.1174, "step": 25660 }, { "epoch": 2.16, "learning_rate": 1.5688612697346325e-05, "loss": 0.049, "step": 25670 }, { "epoch": 2.16, "learning_rate": 1.5686933154182064e-05, "loss": 0.1723, "step": 25680 }, { "epoch": 2.16, "learning_rate": 1.5685253611017806e-05, "loss": 0.0932, "step": 25690 }, { "epoch": 2.16, "learning_rate": 1.5683574067853545e-05, "loss": 0.0756, "step": 25700 }, { "epoch": 2.16, "learning_rate": 1.5681894524689284e-05, "loss": 0.0871, "step": 25710 }, { "epoch": 2.16, "learning_rate": 1.5680214981525026e-05, "loss": 0.1159, "step": 25720 }, { "epoch": 2.16, "learning_rate": 1.5678535438360768e-05, "loss": 0.1628, "step": 25730 }, { "epoch": 2.16, "learning_rate": 1.5676855895196507e-05, "loss": 0.1048, "step": 25740 }, { "epoch": 2.16, "learning_rate": 1.567517635203225e-05, "loss": 0.1244, "step": 25750 }, { "epoch": 2.16, "learning_rate": 1.5673496808867988e-05, "loss": 0.1013, "step": 25760 }, { "epoch": 2.16, "learning_rate": 1.567181726570373e-05, "loss": 0.1045, "step": 25770 }, { "epoch": 2.16, "learning_rate": 1.5670137722539472e-05, "loss": 0.106, "step": 25780 }, { "epoch": 2.17, "learning_rate": 1.566845817937521e-05, "loss": 0.1022, "step": 25790 }, { "epoch": 2.17, "learning_rate": 1.566677863621095e-05, "loss": 0.0201, "step": 25800 }, { "epoch": 2.17, "learning_rate": 1.5665099093046692e-05, "loss": 0.0549, "step": 25810 }, { "epoch": 2.17, "learning_rate": 1.5663419549882434e-05, "loss": 0.0984, "step": 25820 }, { "epoch": 2.17, "learning_rate": 1.5661740006718173e-05, "loss": 0.1334, "step": 25830 }, { "epoch": 2.17, "learning_rate": 1.5660060463553915e-05, "loss": 0.1236, "step": 25840 }, { "epoch": 2.17, "learning_rate": 1.5658380920389654e-05, "loss": 0.1538, "step": 25850 }, { "epoch": 2.17, "learning_rate": 1.5656701377225396e-05, "loss": 0.2275, "step": 25860 }, { "epoch": 2.17, "learning_rate": 1.5655021834061138e-05, "loss": 0.1353, "step": 25870 }, { "epoch": 2.17, "learning_rate": 1.5653342290896877e-05, "loss": 0.0831, "step": 25880 }, { "epoch": 2.17, "learning_rate": 1.565166274773262e-05, "loss": 0.0834, "step": 25890 }, { "epoch": 2.18, "learning_rate": 1.564998320456836e-05, "loss": 0.0735, "step": 25900 }, { "epoch": 2.18, "learning_rate": 1.5648303661404097e-05, "loss": 0.0158, "step": 25910 }, { "epoch": 2.18, "learning_rate": 1.564662411823984e-05, "loss": 0.1243, "step": 25920 }, { "epoch": 2.18, "learning_rate": 1.564494457507558e-05, "loss": 0.0807, "step": 25930 }, { "epoch": 2.18, "learning_rate": 1.564326503191132e-05, "loss": 0.0888, "step": 25940 }, { "epoch": 2.18, "learning_rate": 1.5641585488747062e-05, "loss": 0.1187, "step": 25950 }, { "epoch": 2.18, "learning_rate": 1.5639905945582804e-05, "loss": 0.0334, "step": 25960 }, { "epoch": 2.18, "learning_rate": 1.5638226402418543e-05, "loss": 0.1097, "step": 25970 }, { "epoch": 2.18, "learning_rate": 1.5636546859254285e-05, "loss": 0.0753, "step": 25980 }, { "epoch": 2.18, "learning_rate": 1.5634867316090024e-05, "loss": 0.1136, "step": 25990 }, { "epoch": 2.18, "learning_rate": 1.5633187772925766e-05, "loss": 0.0808, "step": 26000 }, { "epoch": 2.18, "learning_rate": 1.5631508229761505e-05, "loss": 0.0669, "step": 26010 }, { "epoch": 2.19, "learning_rate": 1.5629828686597247e-05, "loss": 0.081, "step": 26020 }, { "epoch": 2.19, "learning_rate": 1.5628149143432986e-05, "loss": 0.1346, "step": 26030 }, { "epoch": 2.19, "learning_rate": 1.5626469600268728e-05, "loss": 0.0408, "step": 26040 }, { "epoch": 2.19, "learning_rate": 1.562479005710447e-05, "loss": 0.0605, "step": 26050 }, { "epoch": 2.19, "learning_rate": 1.562311051394021e-05, "loss": 0.1317, "step": 26060 }, { "epoch": 2.19, "learning_rate": 1.562143097077595e-05, "loss": 0.1835, "step": 26070 }, { "epoch": 2.19, "learning_rate": 1.561975142761169e-05, "loss": 0.1006, "step": 26080 }, { "epoch": 2.19, "learning_rate": 1.5618071884447432e-05, "loss": 0.1265, "step": 26090 }, { "epoch": 2.19, "learning_rate": 1.5616392341283174e-05, "loss": 0.1077, "step": 26100 }, { "epoch": 2.19, "learning_rate": 1.5614712798118913e-05, "loss": 0.0307, "step": 26110 }, { "epoch": 2.19, "learning_rate": 1.5613033254954652e-05, "loss": 0.1186, "step": 26120 }, { "epoch": 2.19, "learning_rate": 1.5611353711790394e-05, "loss": 0.0498, "step": 26130 }, { "epoch": 2.2, "learning_rate": 1.5609674168626133e-05, "loss": 0.1866, "step": 26140 }, { "epoch": 2.2, "learning_rate": 1.5607994625461875e-05, "loss": 0.125, "step": 26150 }, { "epoch": 2.2, "learning_rate": 1.5606315082297617e-05, "loss": 0.1256, "step": 26160 }, { "epoch": 2.2, "learning_rate": 1.5604635539133356e-05, "loss": 0.1213, "step": 26170 }, { "epoch": 2.2, "learning_rate": 1.5602955995969098e-05, "loss": 0.0911, "step": 26180 }, { "epoch": 2.2, "learning_rate": 1.560127645280484e-05, "loss": 0.1173, "step": 26190 }, { "epoch": 2.2, "learning_rate": 1.559959690964058e-05, "loss": 0.0765, "step": 26200 }, { "epoch": 2.2, "learning_rate": 1.559791736647632e-05, "loss": 0.0287, "step": 26210 }, { "epoch": 2.2, "learning_rate": 1.559623782331206e-05, "loss": 0.0814, "step": 26220 }, { "epoch": 2.2, "learning_rate": 1.55945582801478e-05, "loss": 0.05, "step": 26230 }, { "epoch": 2.2, "learning_rate": 1.559287873698354e-05, "loss": 0.0909, "step": 26240 }, { "epoch": 2.2, "learning_rate": 1.5591199193819283e-05, "loss": 0.0874, "step": 26250 }, { "epoch": 2.21, "learning_rate": 1.5589519650655022e-05, "loss": 0.1189, "step": 26260 }, { "epoch": 2.21, "learning_rate": 1.5587840107490764e-05, "loss": 0.2067, "step": 26270 }, { "epoch": 2.21, "learning_rate": 1.5586160564326506e-05, "loss": 0.0762, "step": 26280 }, { "epoch": 2.21, "learning_rate": 1.5584481021162245e-05, "loss": 0.0915, "step": 26290 }, { "epoch": 2.21, "learning_rate": 1.5582801477997987e-05, "loss": 0.0531, "step": 26300 }, { "epoch": 2.21, "learning_rate": 1.5581121934833726e-05, "loss": 0.0964, "step": 26310 }, { "epoch": 2.21, "learning_rate": 1.5579442391669468e-05, "loss": 0.0758, "step": 26320 }, { "epoch": 2.21, "learning_rate": 1.5577762848505207e-05, "loss": 0.0231, "step": 26330 }, { "epoch": 2.21, "learning_rate": 1.557608330534095e-05, "loss": 0.0959, "step": 26340 }, { "epoch": 2.21, "learning_rate": 1.5574403762176688e-05, "loss": 0.1788, "step": 26350 }, { "epoch": 2.21, "learning_rate": 1.557272421901243e-05, "loss": 0.1087, "step": 26360 }, { "epoch": 2.21, "learning_rate": 1.557104467584817e-05, "loss": 0.0918, "step": 26370 }, { "epoch": 2.22, "learning_rate": 1.556936513268391e-05, "loss": 0.0658, "step": 26380 }, { "epoch": 2.22, "learning_rate": 1.5567685589519653e-05, "loss": 0.0383, "step": 26390 }, { "epoch": 2.22, "learning_rate": 1.5566006046355392e-05, "loss": 0.1084, "step": 26400 }, { "epoch": 2.22, "learning_rate": 1.5564326503191134e-05, "loss": 0.0328, "step": 26410 }, { "epoch": 2.22, "learning_rate": 1.5562646960026876e-05, "loss": 0.1647, "step": 26420 }, { "epoch": 2.22, "learning_rate": 1.5560967416862615e-05, "loss": 0.0969, "step": 26430 }, { "epoch": 2.22, "learning_rate": 1.5559287873698354e-05, "loss": 0.132, "step": 26440 }, { "epoch": 2.22, "learning_rate": 1.5557608330534096e-05, "loss": 0.0691, "step": 26450 }, { "epoch": 2.22, "learning_rate": 1.5555928787369835e-05, "loss": 0.149, "step": 26460 }, { "epoch": 2.22, "learning_rate": 1.5554249244205577e-05, "loss": 0.0821, "step": 26470 }, { "epoch": 2.22, "learning_rate": 1.555256970104132e-05, "loss": 0.0507, "step": 26480 }, { "epoch": 2.22, "learning_rate": 1.5550890157877058e-05, "loss": 0.1864, "step": 26490 }, { "epoch": 2.23, "learning_rate": 1.55492106147128e-05, "loss": 0.1373, "step": 26500 }, { "epoch": 2.23, "learning_rate": 1.554753107154854e-05, "loss": 0.1304, "step": 26510 }, { "epoch": 2.23, "learning_rate": 1.554585152838428e-05, "loss": 0.1102, "step": 26520 }, { "epoch": 2.23, "learning_rate": 1.5544171985220023e-05, "loss": 0.0687, "step": 26530 }, { "epoch": 2.23, "learning_rate": 1.5542492442055762e-05, "loss": 0.092, "step": 26540 }, { "epoch": 2.23, "learning_rate": 1.55408128988915e-05, "loss": 0.0803, "step": 26550 }, { "epoch": 2.23, "learning_rate": 1.5539133355727243e-05, "loss": 0.1894, "step": 26560 }, { "epoch": 2.23, "learning_rate": 1.5537453812562985e-05, "loss": 0.0883, "step": 26570 }, { "epoch": 2.23, "learning_rate": 1.5535774269398724e-05, "loss": 0.06, "step": 26580 }, { "epoch": 2.23, "learning_rate": 1.5534094726234466e-05, "loss": 0.1137, "step": 26590 }, { "epoch": 2.23, "learning_rate": 1.5532415183070205e-05, "loss": 0.0595, "step": 26600 }, { "epoch": 2.23, "learning_rate": 1.5530735639905947e-05, "loss": 0.1739, "step": 26610 }, { "epoch": 2.24, "learning_rate": 1.552905609674169e-05, "loss": 0.0561, "step": 26620 }, { "epoch": 2.24, "learning_rate": 1.5527376553577428e-05, "loss": 0.0735, "step": 26630 }, { "epoch": 2.24, "learning_rate": 1.552569701041317e-05, "loss": 0.0652, "step": 26640 }, { "epoch": 2.24, "learning_rate": 1.552401746724891e-05, "loss": 0.0686, "step": 26650 }, { "epoch": 2.24, "learning_rate": 1.552233792408465e-05, "loss": 0.0763, "step": 26660 }, { "epoch": 2.24, "learning_rate": 1.552065838092039e-05, "loss": 0.05, "step": 26670 }, { "epoch": 2.24, "learning_rate": 1.5518978837756132e-05, "loss": 0.117, "step": 26680 }, { "epoch": 2.24, "learning_rate": 1.551729929459187e-05, "loss": 0.1798, "step": 26690 }, { "epoch": 2.24, "learning_rate": 1.5515619751427613e-05, "loss": 0.085, "step": 26700 }, { "epoch": 2.24, "learning_rate": 1.5513940208263355e-05, "loss": 0.0664, "step": 26710 }, { "epoch": 2.24, "learning_rate": 1.5512260665099094e-05, "loss": 0.1434, "step": 26720 }, { "epoch": 2.24, "learning_rate": 1.5510581121934836e-05, "loss": 0.0677, "step": 26730 }, { "epoch": 2.25, "learning_rate": 1.5508901578770575e-05, "loss": 0.0503, "step": 26740 }, { "epoch": 2.25, "learning_rate": 1.5507222035606317e-05, "loss": 0.1291, "step": 26750 }, { "epoch": 2.25, "learning_rate": 1.5505542492442056e-05, "loss": 0.111, "step": 26760 }, { "epoch": 2.25, "learning_rate": 1.5503862949277798e-05, "loss": 0.1191, "step": 26770 }, { "epoch": 2.25, "learning_rate": 1.5502183406113537e-05, "loss": 0.1121, "step": 26780 }, { "epoch": 2.25, "learning_rate": 1.550050386294928e-05, "loss": 0.1279, "step": 26790 }, { "epoch": 2.25, "learning_rate": 1.549882431978502e-05, "loss": 0.0196, "step": 26800 }, { "epoch": 2.25, "learning_rate": 1.549714477662076e-05, "loss": 0.1024, "step": 26810 }, { "epoch": 2.25, "learning_rate": 1.5495465233456502e-05, "loss": 0.1025, "step": 26820 }, { "epoch": 2.25, "learning_rate": 1.549378569029224e-05, "loss": 0.2521, "step": 26830 }, { "epoch": 2.25, "learning_rate": 1.5492106147127983e-05, "loss": 0.0567, "step": 26840 }, { "epoch": 2.25, "learning_rate": 1.5490426603963725e-05, "loss": 0.0592, "step": 26850 }, { "epoch": 2.26, "learning_rate": 1.5488747060799464e-05, "loss": 0.1173, "step": 26860 }, { "epoch": 2.26, "learning_rate": 1.5487067517635203e-05, "loss": 0.1605, "step": 26870 }, { "epoch": 2.26, "learning_rate": 1.5485387974470945e-05, "loss": 0.0867, "step": 26880 }, { "epoch": 2.26, "learning_rate": 1.5483708431306684e-05, "loss": 0.0411, "step": 26890 }, { "epoch": 2.26, "learning_rate": 1.5482028888142426e-05, "loss": 0.053, "step": 26900 }, { "epoch": 2.26, "learning_rate": 1.5480349344978168e-05, "loss": 0.0799, "step": 26910 }, { "epoch": 2.26, "learning_rate": 1.5478669801813907e-05, "loss": 0.1536, "step": 26920 }, { "epoch": 2.26, "learning_rate": 1.547699025864965e-05, "loss": 0.0702, "step": 26930 }, { "epoch": 2.26, "learning_rate": 1.547531071548539e-05, "loss": 0.0779, "step": 26940 }, { "epoch": 2.26, "learning_rate": 1.547363117232113e-05, "loss": 0.0638, "step": 26950 }, { "epoch": 2.26, "learning_rate": 1.5471951629156872e-05, "loss": 0.0795, "step": 26960 }, { "epoch": 2.26, "learning_rate": 1.547027208599261e-05, "loss": 0.0657, "step": 26970 }, { "epoch": 2.27, "learning_rate": 1.546859254282835e-05, "loss": 0.0592, "step": 26980 }, { "epoch": 2.27, "learning_rate": 1.5466912999664092e-05, "loss": 0.1144, "step": 26990 }, { "epoch": 2.27, "learning_rate": 1.5465233456499834e-05, "loss": 0.1459, "step": 27000 }, { "epoch": 2.27, "learning_rate": 1.5463553913335573e-05, "loss": 0.083, "step": 27010 }, { "epoch": 2.27, "learning_rate": 1.5461874370171315e-05, "loss": 0.131, "step": 27020 }, { "epoch": 2.27, "learning_rate": 1.5460194827007057e-05, "loss": 0.0844, "step": 27030 }, { "epoch": 2.27, "learning_rate": 1.5458515283842796e-05, "loss": 0.0824, "step": 27040 }, { "epoch": 2.27, "learning_rate": 1.5456835740678538e-05, "loss": 0.0409, "step": 27050 }, { "epoch": 2.27, "learning_rate": 1.5455156197514277e-05, "loss": 0.1775, "step": 27060 }, { "epoch": 2.27, "learning_rate": 1.5453476654350016e-05, "loss": 0.0647, "step": 27070 }, { "epoch": 2.27, "learning_rate": 1.5451797111185758e-05, "loss": 0.0839, "step": 27080 }, { "epoch": 2.27, "learning_rate": 1.54501175680215e-05, "loss": 0.1336, "step": 27090 }, { "epoch": 2.28, "learning_rate": 1.544843802485724e-05, "loss": 0.0576, "step": 27100 }, { "epoch": 2.28, "learning_rate": 1.544675848169298e-05, "loss": 0.0752, "step": 27110 }, { "epoch": 2.28, "learning_rate": 1.544507893852872e-05, "loss": 0.063, "step": 27120 }, { "epoch": 2.28, "learning_rate": 1.5443399395364462e-05, "loss": 0.1242, "step": 27130 }, { "epoch": 2.28, "learning_rate": 1.5441719852200204e-05, "loss": 0.0281, "step": 27140 }, { "epoch": 2.28, "learning_rate": 1.5440040309035943e-05, "loss": 0.1046, "step": 27150 }, { "epoch": 2.28, "learning_rate": 1.5438360765871685e-05, "loss": 0.0711, "step": 27160 }, { "epoch": 2.28, "learning_rate": 1.5436681222707427e-05, "loss": 0.0691, "step": 27170 }, { "epoch": 2.28, "learning_rate": 1.5435001679543166e-05, "loss": 0.0686, "step": 27180 }, { "epoch": 2.28, "learning_rate": 1.5433322136378905e-05, "loss": 0.049, "step": 27190 }, { "epoch": 2.28, "learning_rate": 1.5431642593214647e-05, "loss": 0.0392, "step": 27200 }, { "epoch": 2.29, "learning_rate": 1.5429963050050386e-05, "loss": 0.1275, "step": 27210 }, { "epoch": 2.29, "learning_rate": 1.5428283506886128e-05, "loss": 0.0706, "step": 27220 }, { "epoch": 2.29, "learning_rate": 1.542660396372187e-05, "loss": 0.2538, "step": 27230 }, { "epoch": 2.29, "learning_rate": 1.542492442055761e-05, "loss": 0.0744, "step": 27240 }, { "epoch": 2.29, "learning_rate": 1.542324487739335e-05, "loss": 0.139, "step": 27250 }, { "epoch": 2.29, "learning_rate": 1.5421565334229093e-05, "loss": 0.1429, "step": 27260 }, { "epoch": 2.29, "learning_rate": 1.5419885791064832e-05, "loss": 0.12, "step": 27270 }, { "epoch": 2.29, "learning_rate": 1.5418206247900574e-05, "loss": 0.1057, "step": 27280 }, { "epoch": 2.29, "learning_rate": 1.5416526704736313e-05, "loss": 0.0786, "step": 27290 }, { "epoch": 2.29, "learning_rate": 1.5414847161572052e-05, "loss": 0.0607, "step": 27300 }, { "epoch": 2.29, "learning_rate": 1.5413167618407794e-05, "loss": 0.1712, "step": 27310 }, { "epoch": 2.29, "learning_rate": 1.5411488075243536e-05, "loss": 0.1819, "step": 27320 }, { "epoch": 2.3, "learning_rate": 1.5409808532079275e-05, "loss": 0.1389, "step": 27330 }, { "epoch": 2.3, "learning_rate": 1.5408128988915017e-05, "loss": 0.0084, "step": 27340 }, { "epoch": 2.3, "learning_rate": 1.5406449445750756e-05, "loss": 0.0949, "step": 27350 }, { "epoch": 2.3, "learning_rate": 1.5404769902586498e-05, "loss": 0.1189, "step": 27360 }, { "epoch": 2.3, "learning_rate": 1.540309035942224e-05, "loss": 0.0781, "step": 27370 }, { "epoch": 2.3, "learning_rate": 1.540141081625798e-05, "loss": 0.1438, "step": 27380 }, { "epoch": 2.3, "learning_rate": 1.5399731273093718e-05, "loss": 0.1179, "step": 27390 }, { "epoch": 2.3, "learning_rate": 1.539805172992946e-05, "loss": 0.1054, "step": 27400 }, { "epoch": 2.3, "learning_rate": 1.5396372186765202e-05, "loss": 0.1186, "step": 27410 }, { "epoch": 2.3, "learning_rate": 1.539469264360094e-05, "loss": 0.1268, "step": 27420 }, { "epoch": 2.3, "learning_rate": 1.5393013100436683e-05, "loss": 0.0806, "step": 27430 }, { "epoch": 2.3, "learning_rate": 1.5391333557272422e-05, "loss": 0.0733, "step": 27440 }, { "epoch": 2.31, "learning_rate": 1.5389654014108164e-05, "loss": 0.0645, "step": 27450 }, { "epoch": 2.31, "learning_rate": 1.5387974470943906e-05, "loss": 0.0224, "step": 27460 }, { "epoch": 2.31, "learning_rate": 1.5386294927779645e-05, "loss": 0.0869, "step": 27470 }, { "epoch": 2.31, "learning_rate": 1.5384615384615387e-05, "loss": 0.2203, "step": 27480 }, { "epoch": 2.31, "learning_rate": 1.5382935841451126e-05, "loss": 0.1494, "step": 27490 }, { "epoch": 2.31, "learning_rate": 1.5381256298286865e-05, "loss": 0.0926, "step": 27500 }, { "epoch": 2.31, "learning_rate": 1.5379576755122607e-05, "loss": 0.1493, "step": 27510 }, { "epoch": 2.31, "learning_rate": 1.537789721195835e-05, "loss": 0.1105, "step": 27520 }, { "epoch": 2.31, "learning_rate": 1.5376217668794088e-05, "loss": 0.0404, "step": 27530 }, { "epoch": 2.31, "learning_rate": 1.537453812562983e-05, "loss": 0.0139, "step": 27540 }, { "epoch": 2.31, "learning_rate": 1.5372858582465572e-05, "loss": 0.0561, "step": 27550 }, { "epoch": 2.31, "learning_rate": 1.537117903930131e-05, "loss": 0.1362, "step": 27560 }, { "epoch": 2.32, "learning_rate": 1.5369499496137053e-05, "loss": 0.0612, "step": 27570 }, { "epoch": 2.32, "learning_rate": 1.5367819952972792e-05, "loss": 0.1513, "step": 27580 }, { "epoch": 2.32, "learning_rate": 1.5366140409808534e-05, "loss": 0.0869, "step": 27590 }, { "epoch": 2.32, "learning_rate": 1.5364460866644273e-05, "loss": 0.1419, "step": 27600 }, { "epoch": 2.32, "learning_rate": 1.5362781323480015e-05, "loss": 0.1159, "step": 27610 }, { "epoch": 2.32, "learning_rate": 1.5361101780315754e-05, "loss": 0.0849, "step": 27620 }, { "epoch": 2.32, "learning_rate": 1.5359422237151496e-05, "loss": 0.0321, "step": 27630 }, { "epoch": 2.32, "learning_rate": 1.5357742693987235e-05, "loss": 0.111, "step": 27640 }, { "epoch": 2.32, "learning_rate": 1.5356063150822977e-05, "loss": 0.1257, "step": 27650 }, { "epoch": 2.32, "learning_rate": 1.535438360765872e-05, "loss": 0.1068, "step": 27660 }, { "epoch": 2.32, "learning_rate": 1.5352704064494458e-05, "loss": 0.0555, "step": 27670 }, { "epoch": 2.32, "learning_rate": 1.53510245213302e-05, "loss": 0.0541, "step": 27680 }, { "epoch": 2.33, "learning_rate": 1.5349344978165942e-05, "loss": 0.1516, "step": 27690 }, { "epoch": 2.33, "learning_rate": 1.534766543500168e-05, "loss": 0.2129, "step": 27700 }, { "epoch": 2.33, "learning_rate": 1.534598589183742e-05, "loss": 0.0465, "step": 27710 }, { "epoch": 2.33, "learning_rate": 1.5344306348673162e-05, "loss": 0.067, "step": 27720 }, { "epoch": 2.33, "learning_rate": 1.53426268055089e-05, "loss": 0.1606, "step": 27730 }, { "epoch": 2.33, "learning_rate": 1.5340947262344643e-05, "loss": 0.042, "step": 27740 }, { "epoch": 2.33, "learning_rate": 1.5339267719180385e-05, "loss": 0.054, "step": 27750 }, { "epoch": 2.33, "learning_rate": 1.5337588176016124e-05, "loss": 0.1168, "step": 27760 }, { "epoch": 2.33, "learning_rate": 1.5335908632851866e-05, "loss": 0.0914, "step": 27770 }, { "epoch": 2.33, "learning_rate": 1.5334229089687608e-05, "loss": 0.1283, "step": 27780 }, { "epoch": 2.33, "learning_rate": 1.5332549546523347e-05, "loss": 0.0761, "step": 27790 }, { "epoch": 2.33, "learning_rate": 1.533087000335909e-05, "loss": 0.0474, "step": 27800 }, { "epoch": 2.34, "learning_rate": 1.5329190460194828e-05, "loss": 0.047, "step": 27810 }, { "epoch": 2.34, "learning_rate": 1.5327510917030567e-05, "loss": 0.1618, "step": 27820 }, { "epoch": 2.34, "learning_rate": 1.532583137386631e-05, "loss": 0.1276, "step": 27830 }, { "epoch": 2.34, "learning_rate": 1.532415183070205e-05, "loss": 0.0834, "step": 27840 }, { "epoch": 2.34, "learning_rate": 1.532247228753779e-05, "loss": 0.0176, "step": 27850 }, { "epoch": 2.34, "learning_rate": 1.5320792744373532e-05, "loss": 0.1223, "step": 27860 }, { "epoch": 2.34, "learning_rate": 1.531911320120927e-05, "loss": 0.1104, "step": 27870 }, { "epoch": 2.34, "learning_rate": 1.5317433658045013e-05, "loss": 0.0996, "step": 27880 }, { "epoch": 2.34, "learning_rate": 1.5315754114880755e-05, "loss": 0.082, "step": 27890 }, { "epoch": 2.34, "learning_rate": 1.5314074571716494e-05, "loss": 0.0862, "step": 27900 }, { "epoch": 2.34, "learning_rate": 1.5312395028552236e-05, "loss": 0.0995, "step": 27910 }, { "epoch": 2.34, "learning_rate": 1.5310715485387975e-05, "loss": 0.0676, "step": 27920 }, { "epoch": 2.35, "learning_rate": 1.5309035942223717e-05, "loss": 0.1003, "step": 27930 }, { "epoch": 2.35, "learning_rate": 1.5307356399059456e-05, "loss": 0.0538, "step": 27940 }, { "epoch": 2.35, "learning_rate": 1.5305676855895198e-05, "loss": 0.089, "step": 27950 }, { "epoch": 2.35, "learning_rate": 1.5303997312730937e-05, "loss": 0.1447, "step": 27960 }, { "epoch": 2.35, "learning_rate": 1.530231776956668e-05, "loss": 0.1422, "step": 27970 }, { "epoch": 2.35, "learning_rate": 1.530063822640242e-05, "loss": 0.0986, "step": 27980 }, { "epoch": 2.35, "learning_rate": 1.529895868323816e-05, "loss": 0.1069, "step": 27990 }, { "epoch": 2.35, "learning_rate": 1.5297279140073902e-05, "loss": 0.0821, "step": 28000 }, { "epoch": 2.35, "learning_rate": 1.5295599596909644e-05, "loss": 0.088, "step": 28010 }, { "epoch": 2.35, "learning_rate": 1.5293920053745383e-05, "loss": 0.0703, "step": 28020 }, { "epoch": 2.35, "learning_rate": 1.5292240510581122e-05, "loss": 0.0676, "step": 28030 }, { "epoch": 2.35, "learning_rate": 1.5290560967416864e-05, "loss": 0.1186, "step": 28040 }, { "epoch": 2.36, "learning_rate": 1.5288881424252603e-05, "loss": 0.087, "step": 28050 }, { "epoch": 2.36, "learning_rate": 1.5287201881088345e-05, "loss": 0.0733, "step": 28060 }, { "epoch": 2.36, "learning_rate": 1.5285522337924087e-05, "loss": 0.0289, "step": 28070 }, { "epoch": 2.36, "learning_rate": 1.5283842794759826e-05, "loss": 0.0743, "step": 28080 }, { "epoch": 2.36, "learning_rate": 1.5282163251595568e-05, "loss": 0.1198, "step": 28090 }, { "epoch": 2.36, "learning_rate": 1.5280483708431307e-05, "loss": 0.1498, "step": 28100 }, { "epoch": 2.36, "learning_rate": 1.527880416526705e-05, "loss": 0.0829, "step": 28110 }, { "epoch": 2.36, "learning_rate": 1.527712462210279e-05, "loss": 0.0821, "step": 28120 }, { "epoch": 2.36, "learning_rate": 1.527544507893853e-05, "loss": 0.1637, "step": 28130 }, { "epoch": 2.36, "learning_rate": 1.527376553577427e-05, "loss": 0.0674, "step": 28140 }, { "epoch": 2.36, "learning_rate": 1.527208599261001e-05, "loss": 0.0347, "step": 28150 }, { "epoch": 2.36, "learning_rate": 1.5270406449445753e-05, "loss": 0.0795, "step": 28160 }, { "epoch": 2.37, "learning_rate": 1.5268726906281492e-05, "loss": 0.1628, "step": 28170 }, { "epoch": 2.37, "learning_rate": 1.5267047363117234e-05, "loss": 0.2176, "step": 28180 }, { "epoch": 2.37, "learning_rate": 1.5265367819952973e-05, "loss": 0.0547, "step": 28190 }, { "epoch": 2.37, "learning_rate": 1.5263688276788715e-05, "loss": 0.0741, "step": 28200 }, { "epoch": 2.37, "learning_rate": 1.5262008733624457e-05, "loss": 0.0869, "step": 28210 }, { "epoch": 2.37, "learning_rate": 1.5260329190460196e-05, "loss": 0.0734, "step": 28220 }, { "epoch": 2.37, "learning_rate": 1.5258649647295938e-05, "loss": 0.1003, "step": 28230 }, { "epoch": 2.37, "learning_rate": 1.5256970104131677e-05, "loss": 0.1638, "step": 28240 }, { "epoch": 2.37, "learning_rate": 1.5255290560967417e-05, "loss": 0.0823, "step": 28250 }, { "epoch": 2.37, "learning_rate": 1.5253611017803158e-05, "loss": 0.1078, "step": 28260 }, { "epoch": 2.37, "learning_rate": 1.5251931474638898e-05, "loss": 0.0768, "step": 28270 }, { "epoch": 2.37, "learning_rate": 1.525025193147464e-05, "loss": 0.182, "step": 28280 }, { "epoch": 2.38, "learning_rate": 1.5248572388310381e-05, "loss": 0.0576, "step": 28290 }, { "epoch": 2.38, "learning_rate": 1.5246892845146122e-05, "loss": 0.0219, "step": 28300 }, { "epoch": 2.38, "learning_rate": 1.5245213301981862e-05, "loss": 0.155, "step": 28310 }, { "epoch": 2.38, "learning_rate": 1.5243533758817604e-05, "loss": 0.1853, "step": 28320 }, { "epoch": 2.38, "learning_rate": 1.5241854215653345e-05, "loss": 0.0876, "step": 28330 }, { "epoch": 2.38, "learning_rate": 1.5240174672489083e-05, "loss": 0.0516, "step": 28340 }, { "epoch": 2.38, "learning_rate": 1.5238495129324824e-05, "loss": 0.0424, "step": 28350 }, { "epoch": 2.38, "learning_rate": 1.5236815586160564e-05, "loss": 0.0609, "step": 28360 }, { "epoch": 2.38, "learning_rate": 1.5235136042996307e-05, "loss": 0.0661, "step": 28370 }, { "epoch": 2.38, "learning_rate": 1.5233456499832047e-05, "loss": 0.1011, "step": 28380 }, { "epoch": 2.38, "learning_rate": 1.5231776956667787e-05, "loss": 0.1267, "step": 28390 }, { "epoch": 2.38, "learning_rate": 1.5230097413503528e-05, "loss": 0.092, "step": 28400 }, { "epoch": 2.39, "learning_rate": 1.522841787033927e-05, "loss": 0.1088, "step": 28410 }, { "epoch": 2.39, "learning_rate": 1.522673832717501e-05, "loss": 0.1339, "step": 28420 }, { "epoch": 2.39, "learning_rate": 1.5225058784010751e-05, "loss": 0.1033, "step": 28430 }, { "epoch": 2.39, "learning_rate": 1.5223379240846492e-05, "loss": 0.0444, "step": 28440 }, { "epoch": 2.39, "learning_rate": 1.522169969768223e-05, "loss": 0.092, "step": 28450 }, { "epoch": 2.39, "learning_rate": 1.522002015451797e-05, "loss": 0.0507, "step": 28460 }, { "epoch": 2.39, "learning_rate": 1.5218340611353713e-05, "loss": 0.0186, "step": 28470 }, { "epoch": 2.39, "learning_rate": 1.5216661068189453e-05, "loss": 0.1462, "step": 28480 }, { "epoch": 2.39, "learning_rate": 1.5214981525025194e-05, "loss": 0.0535, "step": 28490 }, { "epoch": 2.39, "learning_rate": 1.5213301981860934e-05, "loss": 0.1353, "step": 28500 }, { "epoch": 2.39, "learning_rate": 1.5211622438696677e-05, "loss": 0.1071, "step": 28510 }, { "epoch": 2.4, "learning_rate": 1.5209942895532417e-05, "loss": 0.1365, "step": 28520 }, { "epoch": 2.4, "learning_rate": 1.5208263352368158e-05, "loss": 0.0694, "step": 28530 }, { "epoch": 2.4, "learning_rate": 1.5206583809203898e-05, "loss": 0.1253, "step": 28540 }, { "epoch": 2.4, "learning_rate": 1.520490426603964e-05, "loss": 0.091, "step": 28550 }, { "epoch": 2.4, "learning_rate": 1.5203224722875379e-05, "loss": 0.0836, "step": 28560 }, { "epoch": 2.4, "learning_rate": 1.520154517971112e-05, "loss": 0.0752, "step": 28570 }, { "epoch": 2.4, "learning_rate": 1.519986563654686e-05, "loss": 0.148, "step": 28580 }, { "epoch": 2.4, "learning_rate": 1.51981860933826e-05, "loss": 0.1097, "step": 28590 }, { "epoch": 2.4, "learning_rate": 1.5196506550218343e-05, "loss": 0.082, "step": 28600 }, { "epoch": 2.4, "learning_rate": 1.5194827007054083e-05, "loss": 0.1454, "step": 28610 }, { "epoch": 2.4, "learning_rate": 1.5193147463889824e-05, "loss": 0.138, "step": 28620 }, { "epoch": 2.4, "learning_rate": 1.5191467920725564e-05, "loss": 0.1032, "step": 28630 }, { "epoch": 2.41, "learning_rate": 1.5189788377561304e-05, "loss": 0.1747, "step": 28640 }, { "epoch": 2.41, "learning_rate": 1.5188108834397047e-05, "loss": 0.1294, "step": 28650 }, { "epoch": 2.41, "learning_rate": 1.5186429291232785e-05, "loss": 0.0592, "step": 28660 }, { "epoch": 2.41, "learning_rate": 1.5184749748068526e-05, "loss": 0.0961, "step": 28670 }, { "epoch": 2.41, "learning_rate": 1.5183070204904266e-05, "loss": 0.0483, "step": 28680 }, { "epoch": 2.41, "learning_rate": 1.5181390661740007e-05, "loss": 0.1218, "step": 28690 }, { "epoch": 2.41, "learning_rate": 1.5179711118575749e-05, "loss": 0.0829, "step": 28700 }, { "epoch": 2.41, "learning_rate": 1.517803157541149e-05, "loss": 0.0505, "step": 28710 }, { "epoch": 2.41, "learning_rate": 1.517635203224723e-05, "loss": 0.1653, "step": 28720 }, { "epoch": 2.41, "learning_rate": 1.517467248908297e-05, "loss": 0.0286, "step": 28730 }, { "epoch": 2.41, "learning_rate": 1.5172992945918713e-05, "loss": 0.0674, "step": 28740 }, { "epoch": 2.41, "learning_rate": 1.5171313402754453e-05, "loss": 0.0803, "step": 28750 }, { "epoch": 2.42, "learning_rate": 1.5169633859590194e-05, "loss": 0.0883, "step": 28760 }, { "epoch": 2.42, "learning_rate": 1.5167954316425932e-05, "loss": 0.1424, "step": 28770 }, { "epoch": 2.42, "learning_rate": 1.5166274773261673e-05, "loss": 0.1287, "step": 28780 }, { "epoch": 2.42, "learning_rate": 1.5164595230097415e-05, "loss": 0.0058, "step": 28790 }, { "epoch": 2.42, "learning_rate": 1.5162915686933156e-05, "loss": 0.0074, "step": 28800 }, { "epoch": 2.42, "learning_rate": 1.5161236143768896e-05, "loss": 0.0788, "step": 28810 }, { "epoch": 2.42, "learning_rate": 1.5159556600604636e-05, "loss": 0.0399, "step": 28820 }, { "epoch": 2.42, "learning_rate": 1.5157877057440377e-05, "loss": 0.142, "step": 28830 }, { "epoch": 2.42, "learning_rate": 1.5156197514276119e-05, "loss": 0.0459, "step": 28840 }, { "epoch": 2.42, "learning_rate": 1.515451797111186e-05, "loss": 0.1891, "step": 28850 }, { "epoch": 2.42, "learning_rate": 1.51528384279476e-05, "loss": 0.0045, "step": 28860 }, { "epoch": 2.42, "learning_rate": 1.5151158884783339e-05, "loss": 0.1082, "step": 28870 }, { "epoch": 2.43, "learning_rate": 1.514947934161908e-05, "loss": 0.0287, "step": 28880 }, { "epoch": 2.43, "learning_rate": 1.5147799798454821e-05, "loss": 0.1262, "step": 28890 }, { "epoch": 2.43, "learning_rate": 1.5146120255290562e-05, "loss": 0.0912, "step": 28900 }, { "epoch": 2.43, "learning_rate": 1.5144440712126302e-05, "loss": 0.0734, "step": 28910 }, { "epoch": 2.43, "learning_rate": 1.5142761168962043e-05, "loss": 0.1572, "step": 28920 }, { "epoch": 2.43, "learning_rate": 1.5141081625797785e-05, "loss": 0.098, "step": 28930 }, { "epoch": 2.43, "learning_rate": 1.5139402082633526e-05, "loss": 0.068, "step": 28940 }, { "epoch": 2.43, "learning_rate": 1.5137722539469266e-05, "loss": 0.1113, "step": 28950 }, { "epoch": 2.43, "learning_rate": 1.5136042996305007e-05, "loss": 0.0996, "step": 28960 }, { "epoch": 2.43, "learning_rate": 1.5134363453140749e-05, "loss": 0.1984, "step": 28970 }, { "epoch": 2.43, "learning_rate": 1.5132683909976486e-05, "loss": 0.0439, "step": 28980 }, { "epoch": 2.43, "learning_rate": 1.5131004366812228e-05, "loss": 0.064, "step": 28990 }, { "epoch": 2.44, "learning_rate": 1.5129324823647968e-05, "loss": 0.1258, "step": 29000 }, { "epoch": 2.44, "learning_rate": 1.5127645280483709e-05, "loss": 0.0756, "step": 29010 }, { "epoch": 2.44, "learning_rate": 1.512596573731945e-05, "loss": 0.0943, "step": 29020 }, { "epoch": 2.44, "learning_rate": 1.5124286194155192e-05, "loss": 0.0918, "step": 29030 }, { "epoch": 2.44, "learning_rate": 1.5122606650990932e-05, "loss": 0.1075, "step": 29040 }, { "epoch": 2.44, "learning_rate": 1.5120927107826673e-05, "loss": 0.0598, "step": 29050 }, { "epoch": 2.44, "learning_rate": 1.5119247564662413e-05, "loss": 0.077, "step": 29060 }, { "epoch": 2.44, "learning_rate": 1.5117568021498155e-05, "loss": 0.0358, "step": 29070 }, { "epoch": 2.44, "learning_rate": 1.5115888478333896e-05, "loss": 0.0641, "step": 29080 }, { "epoch": 2.44, "learning_rate": 1.5114208935169634e-05, "loss": 0.053, "step": 29090 }, { "epoch": 2.44, "learning_rate": 1.5112529392005375e-05, "loss": 0.0528, "step": 29100 }, { "epoch": 2.44, "learning_rate": 1.5110849848841115e-05, "loss": 0.0972, "step": 29110 }, { "epoch": 2.45, "learning_rate": 1.5109170305676858e-05, "loss": 0.139, "step": 29120 }, { "epoch": 2.45, "learning_rate": 1.5107490762512598e-05, "loss": 0.0888, "step": 29130 }, { "epoch": 2.45, "learning_rate": 1.5105811219348338e-05, "loss": 0.108, "step": 29140 }, { "epoch": 2.45, "learning_rate": 1.5104131676184079e-05, "loss": 0.0333, "step": 29150 }, { "epoch": 2.45, "learning_rate": 1.5102452133019821e-05, "loss": 0.1129, "step": 29160 }, { "epoch": 2.45, "learning_rate": 1.5100772589855562e-05, "loss": 0.0586, "step": 29170 }, { "epoch": 2.45, "learning_rate": 1.5099093046691302e-05, "loss": 0.1284, "step": 29180 }, { "epoch": 2.45, "learning_rate": 1.5097413503527041e-05, "loss": 0.0342, "step": 29190 }, { "epoch": 2.45, "learning_rate": 1.5095733960362781e-05, "loss": 0.2128, "step": 29200 }, { "epoch": 2.45, "learning_rate": 1.5094054417198522e-05, "loss": 0.0576, "step": 29210 }, { "epoch": 2.45, "learning_rate": 1.5092374874034264e-05, "loss": 0.02, "step": 29220 }, { "epoch": 2.45, "learning_rate": 1.5090695330870004e-05, "loss": 0.1962, "step": 29230 }, { "epoch": 2.46, "learning_rate": 1.5089015787705745e-05, "loss": 0.0626, "step": 29240 }, { "epoch": 2.46, "learning_rate": 1.5087336244541485e-05, "loss": 0.0929, "step": 29250 }, { "epoch": 2.46, "learning_rate": 1.5085656701377228e-05, "loss": 0.1137, "step": 29260 }, { "epoch": 2.46, "learning_rate": 1.5083977158212968e-05, "loss": 0.1307, "step": 29270 }, { "epoch": 2.46, "learning_rate": 1.5082297615048709e-05, "loss": 0.0495, "step": 29280 }, { "epoch": 2.46, "learning_rate": 1.5080618071884449e-05, "loss": 0.1293, "step": 29290 }, { "epoch": 2.46, "learning_rate": 1.5078938528720188e-05, "loss": 0.0874, "step": 29300 }, { "epoch": 2.46, "learning_rate": 1.507725898555593e-05, "loss": 0.1245, "step": 29310 }, { "epoch": 2.46, "learning_rate": 1.507557944239167e-05, "loss": 0.0591, "step": 29320 }, { "epoch": 2.46, "learning_rate": 1.5073899899227411e-05, "loss": 0.0966, "step": 29330 }, { "epoch": 2.46, "learning_rate": 1.5072220356063151e-05, "loss": 0.1505, "step": 29340 }, { "epoch": 2.46, "learning_rate": 1.5070540812898894e-05, "loss": 0.1776, "step": 29350 }, { "epoch": 2.47, "learning_rate": 1.5068861269734634e-05, "loss": 0.0662, "step": 29360 }, { "epoch": 2.47, "learning_rate": 1.5067181726570375e-05, "loss": 0.0953, "step": 29370 }, { "epoch": 2.47, "learning_rate": 1.5065502183406115e-05, "loss": 0.1247, "step": 29380 }, { "epoch": 2.47, "learning_rate": 1.5063822640241857e-05, "loss": 0.2076, "step": 29390 }, { "epoch": 2.47, "learning_rate": 1.5062143097077594e-05, "loss": 0.0933, "step": 29400 }, { "epoch": 2.47, "learning_rate": 1.5060463553913336e-05, "loss": 0.0803, "step": 29410 }, { "epoch": 2.47, "learning_rate": 1.5058784010749077e-05, "loss": 0.0504, "step": 29420 }, { "epoch": 2.47, "learning_rate": 1.5057104467584817e-05, "loss": 0.0995, "step": 29430 }, { "epoch": 2.47, "learning_rate": 1.5055424924420558e-05, "loss": 0.1416, "step": 29440 }, { "epoch": 2.47, "learning_rate": 1.50537453812563e-05, "loss": 0.0583, "step": 29450 }, { "epoch": 2.47, "learning_rate": 1.505206583809204e-05, "loss": 0.1832, "step": 29460 }, { "epoch": 2.47, "learning_rate": 1.5050386294927781e-05, "loss": 0.0866, "step": 29470 }, { "epoch": 2.48, "learning_rate": 1.5048706751763521e-05, "loss": 0.0626, "step": 29480 }, { "epoch": 2.48, "learning_rate": 1.5047027208599264e-05, "loss": 0.0957, "step": 29490 }, { "epoch": 2.48, "learning_rate": 1.5045347665435004e-05, "loss": 0.0415, "step": 29500 }, { "epoch": 2.48, "learning_rate": 1.5043668122270743e-05, "loss": 0.1387, "step": 29510 }, { "epoch": 2.48, "learning_rate": 1.5041988579106483e-05, "loss": 0.0436, "step": 29520 }, { "epoch": 2.48, "learning_rate": 1.5040309035942224e-05, "loss": 0.0527, "step": 29530 }, { "epoch": 2.48, "learning_rate": 1.5038629492777966e-05, "loss": 0.1798, "step": 29540 }, { "epoch": 2.48, "learning_rate": 1.5036949949613707e-05, "loss": 0.0337, "step": 29550 }, { "epoch": 2.48, "learning_rate": 1.5035270406449447e-05, "loss": 0.1175, "step": 29560 }, { "epoch": 2.48, "learning_rate": 1.5033590863285187e-05, "loss": 0.1397, "step": 29570 }, { "epoch": 2.48, "learning_rate": 1.503191132012093e-05, "loss": 0.1042, "step": 29580 }, { "epoch": 2.48, "learning_rate": 1.503023177695667e-05, "loss": 0.059, "step": 29590 }, { "epoch": 2.49, "learning_rate": 1.502855223379241e-05, "loss": 0.0416, "step": 29600 }, { "epoch": 2.49, "learning_rate": 1.5026872690628151e-05, "loss": 0.1383, "step": 29610 }, { "epoch": 2.49, "learning_rate": 1.502519314746389e-05, "loss": 0.0815, "step": 29620 }, { "epoch": 2.49, "learning_rate": 1.502351360429963e-05, "loss": 0.1181, "step": 29630 }, { "epoch": 2.49, "learning_rate": 1.5021834061135372e-05, "loss": 0.1057, "step": 29640 }, { "epoch": 2.49, "learning_rate": 1.5020154517971113e-05, "loss": 0.0469, "step": 29650 }, { "epoch": 2.49, "learning_rate": 1.5018474974806853e-05, "loss": 0.0667, "step": 29660 }, { "epoch": 2.49, "learning_rate": 1.5016795431642594e-05, "loss": 0.1426, "step": 29670 }, { "epoch": 2.49, "learning_rate": 1.5015115888478336e-05, "loss": 0.2009, "step": 29680 }, { "epoch": 2.49, "learning_rate": 1.5013436345314077e-05, "loss": 0.0688, "step": 29690 }, { "epoch": 2.49, "learning_rate": 1.5011756802149817e-05, "loss": 0.0683, "step": 29700 }, { "epoch": 2.49, "learning_rate": 1.5010077258985558e-05, "loss": 0.0218, "step": 29710 }, { "epoch": 2.5, "learning_rate": 1.5008397715821296e-05, "loss": 0.0463, "step": 29720 }, { "epoch": 2.5, "learning_rate": 1.5006718172657038e-05, "loss": 0.1219, "step": 29730 }, { "epoch": 2.5, "learning_rate": 1.5005038629492779e-05, "loss": 0.0967, "step": 29740 }, { "epoch": 2.5, "learning_rate": 1.500335908632852e-05, "loss": 0.0744, "step": 29750 }, { "epoch": 2.5, "learning_rate": 1.500167954316426e-05, "loss": 0.1787, "step": 29760 }, { "epoch": 2.5, "learning_rate": 1.5000000000000002e-05, "loss": 0.0607, "step": 29770 }, { "epoch": 2.5, "learning_rate": 1.4998320456835743e-05, "loss": 0.0643, "step": 29780 }, { "epoch": 2.5, "learning_rate": 1.4996640913671483e-05, "loss": 0.1646, "step": 29790 }, { "epoch": 2.5, "learning_rate": 1.4994961370507224e-05, "loss": 0.1679, "step": 29800 }, { "epoch": 2.5, "learning_rate": 1.4993281827342964e-05, "loss": 0.0875, "step": 29810 }, { "epoch": 2.5, "learning_rate": 1.4991602284178706e-05, "loss": 0.0493, "step": 29820 }, { "epoch": 2.51, "learning_rate": 1.4989922741014445e-05, "loss": 0.0543, "step": 29830 }, { "epoch": 2.51, "learning_rate": 1.4988243197850185e-05, "loss": 0.0812, "step": 29840 }, { "epoch": 2.51, "learning_rate": 1.4986563654685926e-05, "loss": 0.0592, "step": 29850 }, { "epoch": 2.51, "learning_rate": 1.4984884111521666e-05, "loss": 0.1093, "step": 29860 }, { "epoch": 2.51, "learning_rate": 1.4983204568357409e-05, "loss": 0.0829, "step": 29870 }, { "epoch": 2.51, "learning_rate": 1.4981525025193149e-05, "loss": 0.1917, "step": 29880 }, { "epoch": 2.51, "learning_rate": 1.497984548202889e-05, "loss": 0.0816, "step": 29890 }, { "epoch": 2.51, "learning_rate": 1.497816593886463e-05, "loss": 0.1338, "step": 29900 }, { "epoch": 2.51, "learning_rate": 1.4976486395700372e-05, "loss": 0.053, "step": 29910 }, { "epoch": 2.51, "learning_rate": 1.4974806852536113e-05, "loss": 0.1252, "step": 29920 }, { "epoch": 2.51, "learning_rate": 1.4973127309371851e-05, "loss": 0.1302, "step": 29930 }, { "epoch": 2.51, "learning_rate": 1.4971447766207592e-05, "loss": 0.1575, "step": 29940 }, { "epoch": 2.52, "learning_rate": 1.4969768223043332e-05, "loss": 0.1202, "step": 29950 }, { "epoch": 2.52, "learning_rate": 1.4968088679879073e-05, "loss": 0.0576, "step": 29960 }, { "epoch": 2.52, "learning_rate": 1.4966409136714815e-05, "loss": 0.0558, "step": 29970 }, { "epoch": 2.52, "learning_rate": 1.4964729593550555e-05, "loss": 0.0637, "step": 29980 }, { "epoch": 2.52, "learning_rate": 1.4963050050386296e-05, "loss": 0.1863, "step": 29990 }, { "epoch": 2.52, "learning_rate": 1.4961370507222036e-05, "loss": 0.0679, "step": 30000 }, { "epoch": 2.52, "learning_rate": 1.4959690964057779e-05, "loss": 0.0536, "step": 30010 }, { "epoch": 2.52, "learning_rate": 1.4958011420893519e-05, "loss": 0.0926, "step": 30020 }, { "epoch": 2.52, "learning_rate": 1.495633187772926e-05, "loss": 0.0697, "step": 30030 }, { "epoch": 2.52, "learning_rate": 1.4954652334564998e-05, "loss": 0.1193, "step": 30040 }, { "epoch": 2.52, "learning_rate": 1.4952972791400739e-05, "loss": 0.0843, "step": 30050 }, { "epoch": 2.52, "learning_rate": 1.4951293248236481e-05, "loss": 0.1708, "step": 30060 }, { "epoch": 2.53, "learning_rate": 1.4949613705072221e-05, "loss": 0.0441, "step": 30070 }, { "epoch": 2.53, "learning_rate": 1.4947934161907962e-05, "loss": 0.1146, "step": 30080 }, { "epoch": 2.53, "learning_rate": 1.4946254618743702e-05, "loss": 0.0407, "step": 30090 }, { "epoch": 2.53, "learning_rate": 1.4944575075579445e-05, "loss": 0.0759, "step": 30100 }, { "epoch": 2.53, "learning_rate": 1.4942895532415185e-05, "loss": 0.1247, "step": 30110 }, { "epoch": 2.53, "learning_rate": 1.4941215989250926e-05, "loss": 0.0768, "step": 30120 }, { "epoch": 2.53, "learning_rate": 1.4939536446086666e-05, "loss": 0.1186, "step": 30130 }, { "epoch": 2.53, "learning_rate": 1.4937856902922408e-05, "loss": 0.1391, "step": 30140 }, { "epoch": 2.53, "learning_rate": 1.4936177359758145e-05, "loss": 0.1065, "step": 30150 }, { "epoch": 2.53, "learning_rate": 1.4934497816593887e-05, "loss": 0.0662, "step": 30160 }, { "epoch": 2.53, "learning_rate": 1.4932818273429628e-05, "loss": 0.0926, "step": 30170 }, { "epoch": 2.53, "learning_rate": 1.4931138730265368e-05, "loss": 0.1851, "step": 30180 }, { "epoch": 2.54, "learning_rate": 1.4929459187101109e-05, "loss": 0.1099, "step": 30190 }, { "epoch": 2.54, "learning_rate": 1.4927779643936851e-05, "loss": 0.1561, "step": 30200 }, { "epoch": 2.54, "learning_rate": 1.4926100100772592e-05, "loss": 0.1492, "step": 30210 }, { "epoch": 2.54, "learning_rate": 1.4924420557608332e-05, "loss": 0.0728, "step": 30220 }, { "epoch": 2.54, "learning_rate": 1.4922741014444072e-05, "loss": 0.1649, "step": 30230 }, { "epoch": 2.54, "learning_rate": 1.4921061471279815e-05, "loss": 0.1399, "step": 30240 }, { "epoch": 2.54, "learning_rate": 1.4919381928115553e-05, "loss": 0.0842, "step": 30250 }, { "epoch": 2.54, "learning_rate": 1.4917702384951294e-05, "loss": 0.0218, "step": 30260 }, { "epoch": 2.54, "learning_rate": 1.4916022841787034e-05, "loss": 0.0529, "step": 30270 }, { "epoch": 2.54, "learning_rate": 1.4914343298622775e-05, "loss": 0.1144, "step": 30280 }, { "epoch": 2.54, "learning_rate": 1.4912663755458517e-05, "loss": 0.091, "step": 30290 }, { "epoch": 2.54, "learning_rate": 1.4910984212294258e-05, "loss": 0.0947, "step": 30300 }, { "epoch": 2.55, "learning_rate": 1.4909304669129998e-05, "loss": 0.1155, "step": 30310 }, { "epoch": 2.55, "learning_rate": 1.4907625125965738e-05, "loss": 0.1485, "step": 30320 }, { "epoch": 2.55, "learning_rate": 1.490594558280148e-05, "loss": 0.0462, "step": 30330 }, { "epoch": 2.55, "learning_rate": 1.4904266039637221e-05, "loss": 0.0467, "step": 30340 }, { "epoch": 2.55, "learning_rate": 1.4902586496472962e-05, "loss": 0.0862, "step": 30350 }, { "epoch": 2.55, "learning_rate": 1.49009069533087e-05, "loss": 0.0871, "step": 30360 }, { "epoch": 2.55, "learning_rate": 1.489922741014444e-05, "loss": 0.1172, "step": 30370 }, { "epoch": 2.55, "learning_rate": 1.4897547866980181e-05, "loss": 0.1882, "step": 30380 }, { "epoch": 2.55, "learning_rate": 1.4895868323815923e-05, "loss": 0.0562, "step": 30390 }, { "epoch": 2.55, "learning_rate": 1.4894188780651664e-05, "loss": 0.1463, "step": 30400 }, { "epoch": 2.55, "learning_rate": 1.4892509237487404e-05, "loss": 0.0262, "step": 30410 }, { "epoch": 2.55, "learning_rate": 1.4890829694323145e-05, "loss": 0.1976, "step": 30420 }, { "epoch": 2.56, "learning_rate": 1.4889150151158887e-05, "loss": 0.1039, "step": 30430 }, { "epoch": 2.56, "learning_rate": 1.4887470607994628e-05, "loss": 0.0419, "step": 30440 }, { "epoch": 2.56, "learning_rate": 1.4885791064830368e-05, "loss": 0.1003, "step": 30450 }, { "epoch": 2.56, "learning_rate": 1.4884111521666107e-05, "loss": 0.0989, "step": 30460 }, { "epoch": 2.56, "learning_rate": 1.4882431978501847e-05, "loss": 0.0947, "step": 30470 }, { "epoch": 2.56, "learning_rate": 1.488075243533759e-05, "loss": 0.078, "step": 30480 }, { "epoch": 2.56, "learning_rate": 1.487907289217333e-05, "loss": 0.1145, "step": 30490 }, { "epoch": 2.56, "learning_rate": 1.487739334900907e-05, "loss": 0.1329, "step": 30500 }, { "epoch": 2.56, "learning_rate": 1.4875713805844811e-05, "loss": 0.1005, "step": 30510 }, { "epoch": 2.56, "learning_rate": 1.4874034262680553e-05, "loss": 0.1053, "step": 30520 }, { "epoch": 2.56, "learning_rate": 1.4872354719516294e-05, "loss": 0.0637, "step": 30530 }, { "epoch": 2.56, "learning_rate": 1.4870675176352034e-05, "loss": 0.023, "step": 30540 }, { "epoch": 2.57, "learning_rate": 1.4868995633187775e-05, "loss": 0.0662, "step": 30550 }, { "epoch": 2.57, "learning_rate": 1.4867316090023517e-05, "loss": 0.1325, "step": 30560 }, { "epoch": 2.57, "learning_rate": 1.4865636546859254e-05, "loss": 0.0765, "step": 30570 }, { "epoch": 2.57, "learning_rate": 1.4863957003694996e-05, "loss": 0.1073, "step": 30580 }, { "epoch": 2.57, "learning_rate": 1.4862277460530736e-05, "loss": 0.1073, "step": 30590 }, { "epoch": 2.57, "learning_rate": 1.4860597917366477e-05, "loss": 0.1338, "step": 30600 }, { "epoch": 2.57, "learning_rate": 1.4858918374202217e-05, "loss": 0.1138, "step": 30610 }, { "epoch": 2.57, "learning_rate": 1.485723883103796e-05, "loss": 0.0828, "step": 30620 }, { "epoch": 2.57, "learning_rate": 1.48555592878737e-05, "loss": 0.1272, "step": 30630 }, { "epoch": 2.57, "learning_rate": 1.485387974470944e-05, "loss": 0.0757, "step": 30640 }, { "epoch": 2.57, "learning_rate": 1.4852200201545181e-05, "loss": 0.0628, "step": 30650 }, { "epoch": 2.57, "learning_rate": 1.4850520658380923e-05, "loss": 0.0515, "step": 30660 }, { "epoch": 2.58, "learning_rate": 1.4848841115216664e-05, "loss": 0.0893, "step": 30670 }, { "epoch": 2.58, "learning_rate": 1.4847161572052402e-05, "loss": 0.1513, "step": 30680 }, { "epoch": 2.58, "learning_rate": 1.4845482028888143e-05, "loss": 0.0542, "step": 30690 }, { "epoch": 2.58, "learning_rate": 1.4843802485723883e-05, "loss": 0.0795, "step": 30700 }, { "epoch": 2.58, "learning_rate": 1.4842122942559626e-05, "loss": 0.1292, "step": 30710 }, { "epoch": 2.58, "learning_rate": 1.4840443399395366e-05, "loss": 0.0543, "step": 30720 }, { "epoch": 2.58, "learning_rate": 1.4838763856231106e-05, "loss": 0.1089, "step": 30730 }, { "epoch": 2.58, "learning_rate": 1.4837084313066847e-05, "loss": 0.1657, "step": 30740 }, { "epoch": 2.58, "learning_rate": 1.4835404769902587e-05, "loss": 0.032, "step": 30750 }, { "epoch": 2.58, "learning_rate": 1.483372522673833e-05, "loss": 0.0043, "step": 30760 }, { "epoch": 2.58, "learning_rate": 1.483204568357407e-05, "loss": 0.0692, "step": 30770 }, { "epoch": 2.58, "learning_rate": 1.4830366140409809e-05, "loss": 0.2331, "step": 30780 }, { "epoch": 2.59, "learning_rate": 1.482868659724555e-05, "loss": 0.1193, "step": 30790 }, { "epoch": 2.59, "learning_rate": 1.482700705408129e-05, "loss": 0.0952, "step": 30800 }, { "epoch": 2.59, "learning_rate": 1.4825327510917032e-05, "loss": 0.1037, "step": 30810 }, { "epoch": 2.59, "learning_rate": 1.4823647967752772e-05, "loss": 0.0823, "step": 30820 }, { "epoch": 2.59, "learning_rate": 1.4821968424588513e-05, "loss": 0.1236, "step": 30830 }, { "epoch": 2.59, "learning_rate": 1.4820288881424253e-05, "loss": 0.067, "step": 30840 }, { "epoch": 2.59, "learning_rate": 1.4818609338259996e-05, "loss": 0.09, "step": 30850 }, { "epoch": 2.59, "learning_rate": 1.4816929795095736e-05, "loss": 0.0596, "step": 30860 }, { "epoch": 2.59, "learning_rate": 1.4815250251931477e-05, "loss": 0.047, "step": 30870 }, { "epoch": 2.59, "learning_rate": 1.4813570708767217e-05, "loss": 0.1098, "step": 30880 }, { "epoch": 2.59, "learning_rate": 1.4811891165602956e-05, "loss": 0.1295, "step": 30890 }, { "epoch": 2.59, "learning_rate": 1.4810211622438698e-05, "loss": 0.105, "step": 30900 }, { "epoch": 2.6, "learning_rate": 1.4808532079274438e-05, "loss": 0.0248, "step": 30910 }, { "epoch": 2.6, "learning_rate": 1.4806852536110179e-05, "loss": 0.0142, "step": 30920 }, { "epoch": 2.6, "learning_rate": 1.480517299294592e-05, "loss": 0.0538, "step": 30930 }, { "epoch": 2.6, "learning_rate": 1.480349344978166e-05, "loss": 0.0208, "step": 30940 }, { "epoch": 2.6, "learning_rate": 1.4801813906617402e-05, "loss": 0.106, "step": 30950 }, { "epoch": 2.6, "learning_rate": 1.4800134363453143e-05, "loss": 0.1081, "step": 30960 }, { "epoch": 2.6, "learning_rate": 1.4798454820288883e-05, "loss": 0.1034, "step": 30970 }, { "epoch": 2.6, "learning_rate": 1.4796775277124623e-05, "loss": 0.0516, "step": 30980 }, { "epoch": 2.6, "learning_rate": 1.4795095733960362e-05, "loss": 0.179, "step": 30990 }, { "epoch": 2.6, "learning_rate": 1.4793416190796104e-05, "loss": 0.1308, "step": 31000 }, { "epoch": 2.6, "learning_rate": 1.4791736647631845e-05, "loss": 0.0774, "step": 31010 }, { "epoch": 2.6, "learning_rate": 1.4790057104467585e-05, "loss": 0.1516, "step": 31020 }, { "epoch": 2.61, "learning_rate": 1.4788377561303326e-05, "loss": 0.103, "step": 31030 }, { "epoch": 2.61, "learning_rate": 1.4786698018139068e-05, "loss": 0.1051, "step": 31040 }, { "epoch": 2.61, "learning_rate": 1.4785018474974808e-05, "loss": 0.0702, "step": 31050 }, { "epoch": 2.61, "learning_rate": 1.4783338931810549e-05, "loss": 0.1328, "step": 31060 }, { "epoch": 2.61, "learning_rate": 1.478165938864629e-05, "loss": 0.0449, "step": 31070 }, { "epoch": 2.61, "learning_rate": 1.4779979845482032e-05, "loss": 0.1288, "step": 31080 }, { "epoch": 2.61, "learning_rate": 1.4778300302317772e-05, "loss": 0.0381, "step": 31090 }, { "epoch": 2.61, "learning_rate": 1.4776620759153511e-05, "loss": 0.0978, "step": 31100 }, { "epoch": 2.61, "learning_rate": 1.4774941215989251e-05, "loss": 0.0822, "step": 31110 }, { "epoch": 2.61, "learning_rate": 1.4773261672824992e-05, "loss": 0.0901, "step": 31120 }, { "epoch": 2.61, "learning_rate": 1.4771582129660732e-05, "loss": 0.0952, "step": 31130 }, { "epoch": 2.62, "learning_rate": 1.4769902586496474e-05, "loss": 0.064, "step": 31140 }, { "epoch": 2.62, "learning_rate": 1.4768223043332215e-05, "loss": 0.1006, "step": 31150 }, { "epoch": 2.62, "learning_rate": 1.4766543500167955e-05, "loss": 0.1383, "step": 31160 }, { "epoch": 2.62, "learning_rate": 1.4764863957003696e-05, "loss": 0.1333, "step": 31170 }, { "epoch": 2.62, "learning_rate": 1.4763184413839438e-05, "loss": 0.0639, "step": 31180 }, { "epoch": 2.62, "learning_rate": 1.4761504870675179e-05, "loss": 0.1086, "step": 31190 }, { "epoch": 2.62, "learning_rate": 1.4759825327510919e-05, "loss": 0.1413, "step": 31200 }, { "epoch": 2.62, "learning_rate": 1.4758145784346658e-05, "loss": 0.0897, "step": 31210 }, { "epoch": 2.62, "learning_rate": 1.4756466241182398e-05, "loss": 0.0576, "step": 31220 }, { "epoch": 2.62, "learning_rate": 1.475478669801814e-05, "loss": 0.0536, "step": 31230 }, { "epoch": 2.62, "learning_rate": 1.4753107154853881e-05, "loss": 0.1752, "step": 31240 }, { "epoch": 2.62, "learning_rate": 1.4751427611689621e-05, "loss": 0.1897, "step": 31250 }, { "epoch": 2.63, "learning_rate": 1.4749748068525362e-05, "loss": 0.1304, "step": 31260 }, { "epoch": 2.63, "learning_rate": 1.4748068525361104e-05, "loss": 0.0608, "step": 31270 }, { "epoch": 2.63, "learning_rate": 1.4746388982196845e-05, "loss": 0.0801, "step": 31280 }, { "epoch": 2.63, "learning_rate": 1.4744709439032585e-05, "loss": 0.0331, "step": 31290 }, { "epoch": 2.63, "learning_rate": 1.4743029895868325e-05, "loss": 0.0481, "step": 31300 }, { "epoch": 2.63, "learning_rate": 1.4741350352704064e-05, "loss": 0.1251, "step": 31310 }, { "epoch": 2.63, "learning_rate": 1.4739670809539805e-05, "loss": 0.0759, "step": 31320 }, { "epoch": 2.63, "learning_rate": 1.4737991266375547e-05, "loss": 0.1881, "step": 31330 }, { "epoch": 2.63, "learning_rate": 1.4736311723211287e-05, "loss": 0.0745, "step": 31340 }, { "epoch": 2.63, "learning_rate": 1.4734632180047028e-05, "loss": 0.0985, "step": 31350 }, { "epoch": 2.63, "learning_rate": 1.4732952636882768e-05, "loss": 0.0835, "step": 31360 }, { "epoch": 2.63, "learning_rate": 1.473127309371851e-05, "loss": 0.1067, "step": 31370 }, { "epoch": 2.64, "learning_rate": 1.4729593550554251e-05, "loss": 0.0618, "step": 31380 }, { "epoch": 2.64, "learning_rate": 1.4727914007389991e-05, "loss": 0.0651, "step": 31390 }, { "epoch": 2.64, "learning_rate": 1.4726234464225732e-05, "loss": 0.0535, "step": 31400 }, { "epoch": 2.64, "learning_rate": 1.4724554921061474e-05, "loss": 0.0699, "step": 31410 }, { "epoch": 2.64, "learning_rate": 1.4722875377897213e-05, "loss": 0.1096, "step": 31420 }, { "epoch": 2.64, "learning_rate": 1.4721195834732953e-05, "loss": 0.1058, "step": 31430 }, { "epoch": 2.64, "learning_rate": 1.4719516291568694e-05, "loss": 0.1233, "step": 31440 }, { "epoch": 2.64, "learning_rate": 1.4717836748404434e-05, "loss": 0.1697, "step": 31450 }, { "epoch": 2.64, "learning_rate": 1.4716157205240177e-05, "loss": 0.1291, "step": 31460 }, { "epoch": 2.64, "learning_rate": 1.4714477662075917e-05, "loss": 0.0885, "step": 31470 }, { "epoch": 2.64, "learning_rate": 1.4712798118911657e-05, "loss": 0.0883, "step": 31480 }, { "epoch": 2.64, "learning_rate": 1.4711118575747398e-05, "loss": 0.0779, "step": 31490 }, { "epoch": 2.65, "learning_rate": 1.470943903258314e-05, "loss": 0.1461, "step": 31500 }, { "epoch": 2.65, "learning_rate": 1.470775948941888e-05, "loss": 0.132, "step": 31510 }, { "epoch": 2.65, "learning_rate": 1.470607994625462e-05, "loss": 0.1204, "step": 31520 }, { "epoch": 2.65, "learning_rate": 1.470440040309036e-05, "loss": 0.0995, "step": 31530 }, { "epoch": 2.65, "learning_rate": 1.47027208599261e-05, "loss": 0.1273, "step": 31540 }, { "epoch": 2.65, "learning_rate": 1.470104131676184e-05, "loss": 0.0697, "step": 31550 }, { "epoch": 2.65, "learning_rate": 1.4699361773597583e-05, "loss": 0.1135, "step": 31560 }, { "epoch": 2.65, "learning_rate": 1.4697682230433323e-05, "loss": 0.1574, "step": 31570 }, { "epoch": 2.65, "learning_rate": 1.4696002687269064e-05, "loss": 0.0779, "step": 31580 }, { "epoch": 2.65, "learning_rate": 1.4694323144104804e-05, "loss": 0.1149, "step": 31590 }, { "epoch": 2.65, "learning_rate": 1.4692643600940547e-05, "loss": 0.0565, "step": 31600 }, { "epoch": 2.65, "learning_rate": 1.4690964057776287e-05, "loss": 0.0422, "step": 31610 }, { "epoch": 2.66, "learning_rate": 1.4689284514612028e-05, "loss": 0.0971, "step": 31620 }, { "epoch": 2.66, "learning_rate": 1.4687604971447766e-05, "loss": 0.0154, "step": 31630 }, { "epoch": 2.66, "learning_rate": 1.4685925428283507e-05, "loss": 0.0656, "step": 31640 }, { "epoch": 2.66, "learning_rate": 1.4684245885119249e-05, "loss": 0.145, "step": 31650 }, { "epoch": 2.66, "learning_rate": 1.468256634195499e-05, "loss": 0.0924, "step": 31660 }, { "epoch": 2.66, "learning_rate": 1.468088679879073e-05, "loss": 0.0762, "step": 31670 }, { "epoch": 2.66, "learning_rate": 1.467920725562647e-05, "loss": 0.0895, "step": 31680 }, { "epoch": 2.66, "learning_rate": 1.4677527712462213e-05, "loss": 0.0964, "step": 31690 }, { "epoch": 2.66, "learning_rate": 1.4675848169297953e-05, "loss": 0.1119, "step": 31700 }, { "epoch": 2.66, "learning_rate": 1.4674168626133694e-05, "loss": 0.1263, "step": 31710 }, { "epoch": 2.66, "learning_rate": 1.4672489082969434e-05, "loss": 0.0915, "step": 31720 }, { "epoch": 2.66, "learning_rate": 1.4670809539805173e-05, "loss": 0.0634, "step": 31730 }, { "epoch": 2.67, "learning_rate": 1.4669129996640913e-05, "loss": 0.0634, "step": 31740 }, { "epoch": 2.67, "learning_rate": 1.4667450453476655e-05, "loss": 0.273, "step": 31750 }, { "epoch": 2.67, "learning_rate": 1.4665770910312396e-05, "loss": 0.1243, "step": 31760 }, { "epoch": 2.67, "learning_rate": 1.4664091367148136e-05, "loss": 0.0869, "step": 31770 }, { "epoch": 2.67, "learning_rate": 1.4662411823983877e-05, "loss": 0.0609, "step": 31780 }, { "epoch": 2.67, "learning_rate": 1.4660732280819619e-05, "loss": 0.1127, "step": 31790 }, { "epoch": 2.67, "learning_rate": 1.465905273765536e-05, "loss": 0.0874, "step": 31800 }, { "epoch": 2.67, "learning_rate": 1.46573731944911e-05, "loss": 0.0621, "step": 31810 }, { "epoch": 2.67, "learning_rate": 1.465569365132684e-05, "loss": 0.0617, "step": 31820 }, { "epoch": 2.67, "learning_rate": 1.4654014108162583e-05, "loss": 0.0864, "step": 31830 }, { "epoch": 2.67, "learning_rate": 1.4652334564998321e-05, "loss": 0.0349, "step": 31840 }, { "epoch": 2.67, "learning_rate": 1.4650655021834062e-05, "loss": 0.1104, "step": 31850 }, { "epoch": 2.68, "learning_rate": 1.4648975478669802e-05, "loss": 0.0323, "step": 31860 }, { "epoch": 2.68, "learning_rate": 1.4647295935505543e-05, "loss": 0.1253, "step": 31870 }, { "epoch": 2.68, "learning_rate": 1.4645616392341285e-05, "loss": 0.1114, "step": 31880 }, { "epoch": 2.68, "learning_rate": 1.4643936849177025e-05, "loss": 0.1009, "step": 31890 }, { "epoch": 2.68, "learning_rate": 1.4642257306012766e-05, "loss": 0.0348, "step": 31900 }, { "epoch": 2.68, "learning_rate": 1.4640577762848506e-05, "loss": 0.0154, "step": 31910 }, { "epoch": 2.68, "learning_rate": 1.4638898219684247e-05, "loss": 0.1227, "step": 31920 }, { "epoch": 2.68, "learning_rate": 1.4637218676519989e-05, "loss": 0.0923, "step": 31930 }, { "epoch": 2.68, "learning_rate": 1.463553913335573e-05, "loss": 0.135, "step": 31940 }, { "epoch": 2.68, "learning_rate": 1.4633859590191468e-05, "loss": 0.0881, "step": 31950 }, { "epoch": 2.68, "learning_rate": 1.4632180047027209e-05, "loss": 0.0471, "step": 31960 }, { "epoch": 2.68, "learning_rate": 1.463050050386295e-05, "loss": 0.1114, "step": 31970 }, { "epoch": 2.69, "learning_rate": 1.4628820960698691e-05, "loss": 0.0915, "step": 31980 }, { "epoch": 2.69, "learning_rate": 1.4627141417534432e-05, "loss": 0.1675, "step": 31990 }, { "epoch": 2.69, "learning_rate": 1.4625461874370172e-05, "loss": 0.117, "step": 32000 }, { "epoch": 2.69, "learning_rate": 1.4623782331205913e-05, "loss": 0.0514, "step": 32010 }, { "epoch": 2.69, "learning_rate": 1.4622102788041655e-05, "loss": 0.0786, "step": 32020 }, { "epoch": 2.69, "learning_rate": 1.4620423244877396e-05, "loss": 0.0334, "step": 32030 }, { "epoch": 2.69, "learning_rate": 1.4618743701713136e-05, "loss": 0.1952, "step": 32040 }, { "epoch": 2.69, "learning_rate": 1.4617064158548875e-05, "loss": 0.0479, "step": 32050 }, { "epoch": 2.69, "learning_rate": 1.4615384615384615e-05, "loss": 0.1208, "step": 32060 }, { "epoch": 2.69, "learning_rate": 1.4613705072220357e-05, "loss": 0.1284, "step": 32070 }, { "epoch": 2.69, "learning_rate": 1.4612025529056098e-05, "loss": 0.0441, "step": 32080 }, { "epoch": 2.69, "learning_rate": 1.4610345985891838e-05, "loss": 0.1639, "step": 32090 }, { "epoch": 2.7, "learning_rate": 1.4608666442727579e-05, "loss": 0.1503, "step": 32100 }, { "epoch": 2.7, "learning_rate": 1.460698689956332e-05, "loss": 0.0761, "step": 32110 }, { "epoch": 2.7, "learning_rate": 1.4605307356399062e-05, "loss": 0.0778, "step": 32120 }, { "epoch": 2.7, "learning_rate": 1.4603627813234802e-05, "loss": 0.079, "step": 32130 }, { "epoch": 2.7, "learning_rate": 1.4601948270070542e-05, "loss": 0.1611, "step": 32140 }, { "epoch": 2.7, "learning_rate": 1.4600268726906283e-05, "loss": 0.0397, "step": 32150 }, { "epoch": 2.7, "learning_rate": 1.4598589183742022e-05, "loss": 0.1363, "step": 32160 }, { "epoch": 2.7, "learning_rate": 1.4596909640577764e-05, "loss": 0.0972, "step": 32170 }, { "epoch": 2.7, "learning_rate": 1.4595230097413504e-05, "loss": 0.0405, "step": 32180 }, { "epoch": 2.7, "learning_rate": 1.4593550554249245e-05, "loss": 0.112, "step": 32190 }, { "epoch": 2.7, "learning_rate": 1.4591871011084985e-05, "loss": 0.0277, "step": 32200 }, { "epoch": 2.7, "learning_rate": 1.4590191467920728e-05, "loss": 0.0591, "step": 32210 }, { "epoch": 2.71, "learning_rate": 1.4588511924756468e-05, "loss": 0.1279, "step": 32220 }, { "epoch": 2.71, "learning_rate": 1.4586832381592208e-05, "loss": 0.1057, "step": 32230 }, { "epoch": 2.71, "learning_rate": 1.4585152838427949e-05, "loss": 0.1601, "step": 32240 }, { "epoch": 2.71, "learning_rate": 1.4583473295263691e-05, "loss": 0.0414, "step": 32250 }, { "epoch": 2.71, "learning_rate": 1.4581793752099428e-05, "loss": 0.1471, "step": 32260 }, { "epoch": 2.71, "learning_rate": 1.458011420893517e-05, "loss": 0.1297, "step": 32270 }, { "epoch": 2.71, "learning_rate": 1.457843466577091e-05, "loss": 0.1605, "step": 32280 }, { "epoch": 2.71, "learning_rate": 1.4576755122606651e-05, "loss": 0.126, "step": 32290 }, { "epoch": 2.71, "learning_rate": 1.4575075579442392e-05, "loss": 0.048, "step": 32300 }, { "epoch": 2.71, "learning_rate": 1.4573396036278134e-05, "loss": 0.0215, "step": 32310 }, { "epoch": 2.71, "learning_rate": 1.4571716493113874e-05, "loss": 0.0994, "step": 32320 }, { "epoch": 2.71, "learning_rate": 1.4570036949949615e-05, "loss": 0.0345, "step": 32330 }, { "epoch": 2.72, "learning_rate": 1.4568357406785355e-05, "loss": 0.0902, "step": 32340 }, { "epoch": 2.72, "learning_rate": 1.4566677863621098e-05, "loss": 0.048, "step": 32350 }, { "epoch": 2.72, "learning_rate": 1.4564998320456838e-05, "loss": 0.1265, "step": 32360 }, { "epoch": 2.72, "learning_rate": 1.4563318777292577e-05, "loss": 0.1733, "step": 32370 }, { "epoch": 2.72, "learning_rate": 1.4561639234128317e-05, "loss": 0.0534, "step": 32380 }, { "epoch": 2.72, "learning_rate": 1.4559959690964058e-05, "loss": 0.1198, "step": 32390 }, { "epoch": 2.72, "learning_rate": 1.45582801477998e-05, "loss": 0.1139, "step": 32400 }, { "epoch": 2.72, "learning_rate": 1.455660060463554e-05, "loss": 0.1539, "step": 32410 }, { "epoch": 2.72, "learning_rate": 1.4554921061471281e-05, "loss": 0.1733, "step": 32420 }, { "epoch": 2.72, "learning_rate": 1.4553241518307021e-05, "loss": 0.0464, "step": 32430 }, { "epoch": 2.72, "learning_rate": 1.4551561975142764e-05, "loss": 0.0852, "step": 32440 }, { "epoch": 2.73, "learning_rate": 1.4549882431978504e-05, "loss": 0.1004, "step": 32450 }, { "epoch": 2.73, "learning_rate": 1.4548202888814245e-05, "loss": 0.0371, "step": 32460 }, { "epoch": 2.73, "learning_rate": 1.4546523345649985e-05, "loss": 0.0882, "step": 32470 }, { "epoch": 2.73, "learning_rate": 1.4544843802485724e-05, "loss": 0.1152, "step": 32480 }, { "epoch": 2.73, "learning_rate": 1.4543164259321464e-05, "loss": 0.1286, "step": 32490 }, { "epoch": 2.73, "learning_rate": 1.4541484716157206e-05, "loss": 0.1406, "step": 32500 }, { "epoch": 2.73, "learning_rate": 1.4539805172992947e-05, "loss": 0.0662, "step": 32510 }, { "epoch": 2.73, "learning_rate": 1.4538125629828687e-05, "loss": 0.1462, "step": 32520 }, { "epoch": 2.73, "learning_rate": 1.4536446086664428e-05, "loss": 0.0491, "step": 32530 }, { "epoch": 2.73, "learning_rate": 1.453476654350017e-05, "loss": 0.1404, "step": 32540 }, { "epoch": 2.73, "learning_rate": 1.453308700033591e-05, "loss": 0.1142, "step": 32550 }, { "epoch": 2.73, "learning_rate": 1.4531407457171651e-05, "loss": 0.0331, "step": 32560 }, { "epoch": 2.74, "learning_rate": 1.4529727914007391e-05, "loss": 0.0861, "step": 32570 }, { "epoch": 2.74, "learning_rate": 1.452804837084313e-05, "loss": 0.135, "step": 32580 }, { "epoch": 2.74, "learning_rate": 1.4526368827678872e-05, "loss": 0.0569, "step": 32590 }, { "epoch": 2.74, "learning_rate": 1.4524689284514613e-05, "loss": 0.0881, "step": 32600 }, { "epoch": 2.74, "learning_rate": 1.4523009741350353e-05, "loss": 0.0388, "step": 32610 }, { "epoch": 2.74, "learning_rate": 1.4521330198186094e-05, "loss": 0.0865, "step": 32620 }, { "epoch": 2.74, "learning_rate": 1.4519650655021836e-05, "loss": 0.1464, "step": 32630 }, { "epoch": 2.74, "learning_rate": 1.4517971111857576e-05, "loss": 0.0301, "step": 32640 }, { "epoch": 2.74, "learning_rate": 1.4516291568693317e-05, "loss": 0.041, "step": 32650 }, { "epoch": 2.74, "learning_rate": 1.4514612025529057e-05, "loss": 0.1348, "step": 32660 }, { "epoch": 2.74, "learning_rate": 1.45129324823648e-05, "loss": 0.0979, "step": 32670 }, { "epoch": 2.74, "learning_rate": 1.451125293920054e-05, "loss": 0.0954, "step": 32680 }, { "epoch": 2.75, "learning_rate": 1.4509573396036279e-05, "loss": 0.1698, "step": 32690 }, { "epoch": 2.75, "learning_rate": 1.450789385287202e-05, "loss": 0.0995, "step": 32700 }, { "epoch": 2.75, "learning_rate": 1.450621430970776e-05, "loss": 0.0709, "step": 32710 }, { "epoch": 2.75, "learning_rate": 1.45045347665435e-05, "loss": 0.1171, "step": 32720 }, { "epoch": 2.75, "learning_rate": 1.4502855223379242e-05, "loss": 0.1117, "step": 32730 }, { "epoch": 2.75, "learning_rate": 1.4501175680214983e-05, "loss": 0.0661, "step": 32740 }, { "epoch": 2.75, "learning_rate": 1.4499496137050723e-05, "loss": 0.0676, "step": 32750 }, { "epoch": 2.75, "learning_rate": 1.4497816593886464e-05, "loss": 0.1377, "step": 32760 }, { "epoch": 2.75, "learning_rate": 1.4496137050722206e-05, "loss": 0.097, "step": 32770 }, { "epoch": 2.75, "learning_rate": 1.4494457507557947e-05, "loss": 0.0665, "step": 32780 }, { "epoch": 2.75, "learning_rate": 1.4492777964393685e-05, "loss": 0.2229, "step": 32790 }, { "epoch": 2.75, "learning_rate": 1.4491098421229426e-05, "loss": 0.1463, "step": 32800 }, { "epoch": 2.76, "learning_rate": 1.4489418878065166e-05, "loss": 0.122, "step": 32810 }, { "epoch": 2.76, "learning_rate": 1.4487739334900908e-05, "loss": 0.1454, "step": 32820 }, { "epoch": 2.76, "learning_rate": 1.4486059791736649e-05, "loss": 0.0632, "step": 32830 }, { "epoch": 2.76, "learning_rate": 1.448438024857239e-05, "loss": 0.1, "step": 32840 }, { "epoch": 2.76, "learning_rate": 1.448270070540813e-05, "loss": 0.1105, "step": 32850 }, { "epoch": 2.76, "learning_rate": 1.4481021162243872e-05, "loss": 0.0727, "step": 32860 }, { "epoch": 2.76, "learning_rate": 1.4479341619079613e-05, "loss": 0.086, "step": 32870 }, { "epoch": 2.76, "learning_rate": 1.4477662075915353e-05, "loss": 0.1498, "step": 32880 }, { "epoch": 2.76, "learning_rate": 1.4475982532751093e-05, "loss": 0.1156, "step": 32890 }, { "epoch": 2.76, "learning_rate": 1.4474302989586832e-05, "loss": 0.0441, "step": 32900 }, { "epoch": 2.76, "learning_rate": 1.4472623446422573e-05, "loss": 0.1143, "step": 32910 }, { "epoch": 2.76, "learning_rate": 1.4470943903258315e-05, "loss": 0.0484, "step": 32920 }, { "epoch": 2.77, "learning_rate": 1.4469264360094055e-05, "loss": 0.1006, "step": 32930 }, { "epoch": 2.77, "learning_rate": 1.4467584816929796e-05, "loss": 0.124, "step": 32940 }, { "epoch": 2.77, "learning_rate": 1.4465905273765536e-05, "loss": 0.103, "step": 32950 }, { "epoch": 2.77, "learning_rate": 1.4464225730601279e-05, "loss": 0.1957, "step": 32960 }, { "epoch": 2.77, "learning_rate": 1.4462546187437019e-05, "loss": 0.1129, "step": 32970 }, { "epoch": 2.77, "learning_rate": 1.446086664427276e-05, "loss": 0.0604, "step": 32980 }, { "epoch": 2.77, "learning_rate": 1.44591871011085e-05, "loss": 0.0906, "step": 32990 }, { "epoch": 2.77, "learning_rate": 1.4457507557944242e-05, "loss": 0.1321, "step": 33000 }, { "epoch": 2.77, "learning_rate": 1.4455828014779981e-05, "loss": 0.0783, "step": 33010 }, { "epoch": 2.77, "learning_rate": 1.4454148471615721e-05, "loss": 0.0795, "step": 33020 }, { "epoch": 2.77, "learning_rate": 1.4452468928451462e-05, "loss": 0.1508, "step": 33030 }, { "epoch": 2.77, "learning_rate": 1.4450789385287202e-05, "loss": 0.0558, "step": 33040 }, { "epoch": 2.78, "learning_rate": 1.4449109842122944e-05, "loss": 0.0643, "step": 33050 }, { "epoch": 2.78, "learning_rate": 1.4447430298958685e-05, "loss": 0.1905, "step": 33060 }, { "epoch": 2.78, "learning_rate": 1.4445750755794425e-05, "loss": 0.0985, "step": 33070 }, { "epoch": 2.78, "learning_rate": 1.4444071212630166e-05, "loss": 0.0556, "step": 33080 }, { "epoch": 2.78, "learning_rate": 1.4442391669465906e-05, "loss": 0.1122, "step": 33090 }, { "epoch": 2.78, "learning_rate": 1.4440712126301649e-05, "loss": 0.1045, "step": 33100 }, { "epoch": 2.78, "learning_rate": 1.4439032583137387e-05, "loss": 0.0916, "step": 33110 }, { "epoch": 2.78, "learning_rate": 1.4437353039973128e-05, "loss": 0.1269, "step": 33120 }, { "epoch": 2.78, "learning_rate": 1.4435673496808868e-05, "loss": 0.0215, "step": 33130 }, { "epoch": 2.78, "learning_rate": 1.4433993953644609e-05, "loss": 0.0608, "step": 33140 }, { "epoch": 2.78, "learning_rate": 1.4432314410480351e-05, "loss": 0.062, "step": 33150 }, { "epoch": 2.78, "learning_rate": 1.4430634867316091e-05, "loss": 0.1234, "step": 33160 }, { "epoch": 2.79, "learning_rate": 1.4428955324151832e-05, "loss": 0.0406, "step": 33170 }, { "epoch": 2.79, "learning_rate": 1.4427275780987572e-05, "loss": 0.036, "step": 33180 }, { "epoch": 2.79, "learning_rate": 1.4425596237823315e-05, "loss": 0.0476, "step": 33190 }, { "epoch": 2.79, "learning_rate": 1.4423916694659055e-05, "loss": 0.0568, "step": 33200 }, { "epoch": 2.79, "learning_rate": 1.4422237151494796e-05, "loss": 0.1615, "step": 33210 }, { "epoch": 2.79, "learning_rate": 1.4420557608330534e-05, "loss": 0.1508, "step": 33220 }, { "epoch": 2.79, "learning_rate": 1.4418878065166275e-05, "loss": 0.079, "step": 33230 }, { "epoch": 2.79, "learning_rate": 1.4417198522002015e-05, "loss": 0.0917, "step": 33240 }, { "epoch": 2.79, "learning_rate": 1.4415518978837757e-05, "loss": 0.078, "step": 33250 }, { "epoch": 2.79, "learning_rate": 1.4413839435673498e-05, "loss": 0.075, "step": 33260 }, { "epoch": 2.79, "learning_rate": 1.4412159892509238e-05, "loss": 0.028, "step": 33270 }, { "epoch": 2.79, "learning_rate": 1.4410480349344979e-05, "loss": 0.1368, "step": 33280 }, { "epoch": 2.8, "learning_rate": 1.4408800806180721e-05, "loss": 0.0556, "step": 33290 }, { "epoch": 2.8, "learning_rate": 1.4407121263016461e-05, "loss": 0.0479, "step": 33300 }, { "epoch": 2.8, "learning_rate": 1.4405441719852202e-05, "loss": 0.0594, "step": 33310 }, { "epoch": 2.8, "learning_rate": 1.440376217668794e-05, "loss": 0.1411, "step": 33320 }, { "epoch": 2.8, "learning_rate": 1.4402082633523681e-05, "loss": 0.0784, "step": 33330 }, { "epoch": 2.8, "learning_rate": 1.4400403090359423e-05, "loss": 0.1308, "step": 33340 }, { "epoch": 2.8, "learning_rate": 1.4398723547195164e-05, "loss": 0.0771, "step": 33350 }, { "epoch": 2.8, "learning_rate": 1.4397044004030904e-05, "loss": 0.0837, "step": 33360 }, { "epoch": 2.8, "learning_rate": 1.4395364460866645e-05, "loss": 0.1, "step": 33370 }, { "epoch": 2.8, "learning_rate": 1.4393684917702387e-05, "loss": 0.1197, "step": 33380 }, { "epoch": 2.8, "learning_rate": 1.4392005374538127e-05, "loss": 0.0875, "step": 33390 }, { "epoch": 2.8, "learning_rate": 1.4390325831373868e-05, "loss": 0.116, "step": 33400 }, { "epoch": 2.81, "learning_rate": 1.4388646288209608e-05, "loss": 0.1062, "step": 33410 }, { "epoch": 2.81, "learning_rate": 1.438696674504535e-05, "loss": 0.1338, "step": 33420 }, { "epoch": 2.81, "learning_rate": 1.4385287201881088e-05, "loss": 0.0672, "step": 33430 }, { "epoch": 2.81, "learning_rate": 1.438360765871683e-05, "loss": 0.1018, "step": 33440 }, { "epoch": 2.81, "learning_rate": 1.438192811555257e-05, "loss": 0.0477, "step": 33450 }, { "epoch": 2.81, "learning_rate": 1.438024857238831e-05, "loss": 0.0828, "step": 33460 }, { "epoch": 2.81, "learning_rate": 1.4378569029224051e-05, "loss": 0.138, "step": 33470 }, { "epoch": 2.81, "learning_rate": 1.4376889486059793e-05, "loss": 0.0744, "step": 33480 }, { "epoch": 2.81, "learning_rate": 1.4375209942895534e-05, "loss": 0.0796, "step": 33490 }, { "epoch": 2.81, "learning_rate": 1.4373530399731274e-05, "loss": 0.0338, "step": 33500 }, { "epoch": 2.81, "learning_rate": 1.4371850856567015e-05, "loss": 0.0409, "step": 33510 }, { "epoch": 2.81, "learning_rate": 1.4370171313402757e-05, "loss": 0.058, "step": 33520 }, { "epoch": 2.82, "learning_rate": 1.4368491770238498e-05, "loss": 0.0831, "step": 33530 }, { "epoch": 2.82, "learning_rate": 1.4366812227074236e-05, "loss": 0.0495, "step": 33540 }, { "epoch": 2.82, "learning_rate": 1.4365132683909977e-05, "loss": 0.1383, "step": 33550 }, { "epoch": 2.82, "learning_rate": 1.4363453140745717e-05, "loss": 0.1477, "step": 33560 }, { "epoch": 2.82, "learning_rate": 1.436177359758146e-05, "loss": 0.1095, "step": 33570 }, { "epoch": 2.82, "learning_rate": 1.43600940544172e-05, "loss": 0.0944, "step": 33580 }, { "epoch": 2.82, "learning_rate": 1.435841451125294e-05, "loss": 0.1507, "step": 33590 }, { "epoch": 2.82, "learning_rate": 1.4356734968088681e-05, "loss": 0.209, "step": 33600 }, { "epoch": 2.82, "learning_rate": 1.4355055424924423e-05, "loss": 0.0861, "step": 33610 }, { "epoch": 2.82, "learning_rate": 1.4353375881760164e-05, "loss": 0.0771, "step": 33620 }, { "epoch": 2.82, "learning_rate": 1.4351696338595904e-05, "loss": 0.0633, "step": 33630 }, { "epoch": 2.82, "learning_rate": 1.4350016795431643e-05, "loss": 0.126, "step": 33640 }, { "epoch": 2.83, "learning_rate": 1.4348337252267383e-05, "loss": 0.077, "step": 33650 }, { "epoch": 2.83, "learning_rate": 1.4346657709103124e-05, "loss": 0.0268, "step": 33660 }, { "epoch": 2.83, "learning_rate": 1.4344978165938866e-05, "loss": 0.1602, "step": 33670 }, { "epoch": 2.83, "learning_rate": 1.4343298622774606e-05, "loss": 0.0459, "step": 33680 }, { "epoch": 2.83, "learning_rate": 1.4341619079610347e-05, "loss": 0.0517, "step": 33690 }, { "epoch": 2.83, "learning_rate": 1.4339939536446087e-05, "loss": 0.0832, "step": 33700 }, { "epoch": 2.83, "learning_rate": 1.433825999328183e-05, "loss": 0.1433, "step": 33710 }, { "epoch": 2.83, "learning_rate": 1.433658045011757e-05, "loss": 0.1028, "step": 33720 }, { "epoch": 2.83, "learning_rate": 1.433490090695331e-05, "loss": 0.1237, "step": 33730 }, { "epoch": 2.83, "learning_rate": 1.4333221363789051e-05, "loss": 0.1002, "step": 33740 }, { "epoch": 2.83, "learning_rate": 1.433154182062479e-05, "loss": 0.1571, "step": 33750 }, { "epoch": 2.84, "learning_rate": 1.4329862277460532e-05, "loss": 0.0141, "step": 33760 }, { "epoch": 2.84, "learning_rate": 1.4328182734296272e-05, "loss": 0.0617, "step": 33770 }, { "epoch": 2.84, "learning_rate": 1.4326503191132013e-05, "loss": 0.0697, "step": 33780 }, { "epoch": 2.84, "learning_rate": 1.4324823647967753e-05, "loss": 0.0601, "step": 33790 }, { "epoch": 2.84, "learning_rate": 1.4323144104803495e-05, "loss": 0.0857, "step": 33800 }, { "epoch": 2.84, "learning_rate": 1.4321464561639236e-05, "loss": 0.1646, "step": 33810 }, { "epoch": 2.84, "learning_rate": 1.4319785018474976e-05, "loss": 0.1571, "step": 33820 }, { "epoch": 2.84, "learning_rate": 1.4318105475310717e-05, "loss": 0.1155, "step": 33830 }, { "epoch": 2.84, "learning_rate": 1.4316425932146459e-05, "loss": 0.1005, "step": 33840 }, { "epoch": 2.84, "learning_rate": 1.4314746388982196e-05, "loss": 0.0873, "step": 33850 }, { "epoch": 2.84, "learning_rate": 1.4313066845817938e-05, "loss": 0.1143, "step": 33860 }, { "epoch": 2.84, "learning_rate": 1.4311387302653679e-05, "loss": 0.0631, "step": 33870 }, { "epoch": 2.85, "learning_rate": 1.430970775948942e-05, "loss": 0.0991, "step": 33880 }, { "epoch": 2.85, "learning_rate": 1.430802821632516e-05, "loss": 0.1463, "step": 33890 }, { "epoch": 2.85, "learning_rate": 1.4306348673160902e-05, "loss": 0.0854, "step": 33900 }, { "epoch": 2.85, "learning_rate": 1.4304669129996642e-05, "loss": 0.1545, "step": 33910 }, { "epoch": 2.85, "learning_rate": 1.4302989586832383e-05, "loss": 0.0715, "step": 33920 }, { "epoch": 2.85, "learning_rate": 1.4301310043668123e-05, "loss": 0.0369, "step": 33930 }, { "epoch": 2.85, "learning_rate": 1.4299630500503866e-05, "loss": 0.154, "step": 33940 }, { "epoch": 2.85, "learning_rate": 1.4297950957339606e-05, "loss": 0.0996, "step": 33950 }, { "epoch": 2.85, "learning_rate": 1.4296271414175345e-05, "loss": 0.0704, "step": 33960 }, { "epoch": 2.85, "learning_rate": 1.4294591871011085e-05, "loss": 0.1361, "step": 33970 }, { "epoch": 2.85, "learning_rate": 1.4292912327846826e-05, "loss": 0.1227, "step": 33980 }, { "epoch": 2.85, "learning_rate": 1.4291232784682568e-05, "loss": 0.0437, "step": 33990 }, { "epoch": 2.86, "learning_rate": 1.4289553241518308e-05, "loss": 0.0652, "step": 34000 }, { "epoch": 2.86, "learning_rate": 1.4287873698354049e-05, "loss": 0.0748, "step": 34010 }, { "epoch": 2.86, "learning_rate": 1.428619415518979e-05, "loss": 0.1125, "step": 34020 }, { "epoch": 2.86, "learning_rate": 1.4284514612025532e-05, "loss": 0.0487, "step": 34030 }, { "epoch": 2.86, "learning_rate": 1.4282835068861272e-05, "loss": 0.0405, "step": 34040 }, { "epoch": 2.86, "learning_rate": 1.4281155525697012e-05, "loss": 0.1152, "step": 34050 }, { "epoch": 2.86, "learning_rate": 1.4279475982532753e-05, "loss": 0.1266, "step": 34060 }, { "epoch": 2.86, "learning_rate": 1.4277796439368492e-05, "loss": 0.1001, "step": 34070 }, { "epoch": 2.86, "learning_rate": 1.4276116896204232e-05, "loss": 0.1402, "step": 34080 }, { "epoch": 2.86, "learning_rate": 1.4274437353039974e-05, "loss": 0.0675, "step": 34090 }, { "epoch": 2.86, "learning_rate": 1.4272757809875715e-05, "loss": 0.0643, "step": 34100 }, { "epoch": 2.86, "learning_rate": 1.4271078266711455e-05, "loss": 0.075, "step": 34110 }, { "epoch": 2.87, "learning_rate": 1.4269398723547196e-05, "loss": 0.0441, "step": 34120 }, { "epoch": 2.87, "learning_rate": 1.4267719180382938e-05, "loss": 0.0868, "step": 34130 }, { "epoch": 2.87, "learning_rate": 1.4266039637218678e-05, "loss": 0.1468, "step": 34140 }, { "epoch": 2.87, "learning_rate": 1.4264360094054419e-05, "loss": 0.1885, "step": 34150 }, { "epoch": 2.87, "learning_rate": 1.426268055089016e-05, "loss": 0.0896, "step": 34160 }, { "epoch": 2.87, "learning_rate": 1.4261001007725898e-05, "loss": 0.0235, "step": 34170 }, { "epoch": 2.87, "learning_rate": 1.425932146456164e-05, "loss": 0.0713, "step": 34180 }, { "epoch": 2.87, "learning_rate": 1.4257641921397381e-05, "loss": 0.0646, "step": 34190 }, { "epoch": 2.87, "learning_rate": 1.4255962378233121e-05, "loss": 0.0928, "step": 34200 }, { "epoch": 2.87, "learning_rate": 1.4254282835068862e-05, "loss": 0.1799, "step": 34210 }, { "epoch": 2.87, "learning_rate": 1.4252603291904602e-05, "loss": 0.0865, "step": 34220 }, { "epoch": 2.87, "learning_rate": 1.4250923748740344e-05, "loss": 0.1082, "step": 34230 }, { "epoch": 2.88, "learning_rate": 1.4249244205576085e-05, "loss": 0.0659, "step": 34240 }, { "epoch": 2.88, "learning_rate": 1.4247564662411825e-05, "loss": 0.1221, "step": 34250 }, { "epoch": 2.88, "learning_rate": 1.4245885119247566e-05, "loss": 0.1022, "step": 34260 }, { "epoch": 2.88, "learning_rate": 1.4244205576083308e-05, "loss": 0.0651, "step": 34270 }, { "epoch": 2.88, "learning_rate": 1.4242526032919047e-05, "loss": 0.0432, "step": 34280 }, { "epoch": 2.88, "learning_rate": 1.4240846489754787e-05, "loss": 0.0732, "step": 34290 }, { "epoch": 2.88, "learning_rate": 1.4239166946590528e-05, "loss": 0.1346, "step": 34300 }, { "epoch": 2.88, "learning_rate": 1.4237487403426268e-05, "loss": 0.0514, "step": 34310 }, { "epoch": 2.88, "learning_rate": 1.423580786026201e-05, "loss": 0.098, "step": 34320 }, { "epoch": 2.88, "learning_rate": 1.4234128317097751e-05, "loss": 0.0974, "step": 34330 }, { "epoch": 2.88, "learning_rate": 1.4232448773933491e-05, "loss": 0.0415, "step": 34340 }, { "epoch": 2.88, "learning_rate": 1.4230769230769232e-05, "loss": 0.0735, "step": 34350 }, { "epoch": 2.89, "learning_rate": 1.4229089687604974e-05, "loss": 0.1526, "step": 34360 }, { "epoch": 2.89, "learning_rate": 1.4227410144440715e-05, "loss": 0.0571, "step": 34370 }, { "epoch": 2.89, "learning_rate": 1.4225730601276453e-05, "loss": 0.1065, "step": 34380 }, { "epoch": 2.89, "learning_rate": 1.4224051058112194e-05, "loss": 0.1864, "step": 34390 }, { "epoch": 2.89, "learning_rate": 1.4222371514947934e-05, "loss": 0.1627, "step": 34400 }, { "epoch": 2.89, "learning_rate": 1.4220691971783675e-05, "loss": 0.028, "step": 34410 }, { "epoch": 2.89, "learning_rate": 1.4219012428619417e-05, "loss": 0.1801, "step": 34420 }, { "epoch": 2.89, "learning_rate": 1.4217332885455157e-05, "loss": 0.0507, "step": 34430 }, { "epoch": 2.89, "learning_rate": 1.4215653342290898e-05, "loss": 0.1078, "step": 34440 }, { "epoch": 2.89, "learning_rate": 1.4213973799126638e-05, "loss": 0.176, "step": 34450 }, { "epoch": 2.89, "learning_rate": 1.421229425596238e-05, "loss": 0.0931, "step": 34460 }, { "epoch": 2.89, "learning_rate": 1.4210614712798121e-05, "loss": 0.1034, "step": 34470 }, { "epoch": 2.9, "learning_rate": 1.4208935169633861e-05, "loss": 0.1408, "step": 34480 }, { "epoch": 2.9, "learning_rate": 1.42072556264696e-05, "loss": 0.1542, "step": 34490 }, { "epoch": 2.9, "learning_rate": 1.420557608330534e-05, "loss": 0.12, "step": 34500 }, { "epoch": 2.9, "learning_rate": 1.4203896540141083e-05, "loss": 0.0535, "step": 34510 }, { "epoch": 2.9, "learning_rate": 1.4202216996976823e-05, "loss": 0.1194, "step": 34520 }, { "epoch": 2.9, "learning_rate": 1.4200537453812564e-05, "loss": 0.1072, "step": 34530 }, { "epoch": 2.9, "learning_rate": 1.4198857910648304e-05, "loss": 0.06, "step": 34540 }, { "epoch": 2.9, "learning_rate": 1.4197178367484046e-05, "loss": 0.0546, "step": 34550 }, { "epoch": 2.9, "learning_rate": 1.4195498824319787e-05, "loss": 0.1084, "step": 34560 }, { "epoch": 2.9, "learning_rate": 1.4193819281155527e-05, "loss": 0.0384, "step": 34570 }, { "epoch": 2.9, "learning_rate": 1.4192139737991268e-05, "loss": 0.0123, "step": 34580 }, { "epoch": 2.9, "learning_rate": 1.419046019482701e-05, "loss": 0.1118, "step": 34590 }, { "epoch": 2.91, "learning_rate": 1.4188780651662747e-05, "loss": 0.1815, "step": 34600 }, { "epoch": 2.91, "learning_rate": 1.418710110849849e-05, "loss": 0.039, "step": 34610 }, { "epoch": 2.91, "learning_rate": 1.418542156533423e-05, "loss": 0.1196, "step": 34620 }, { "epoch": 2.91, "learning_rate": 1.418374202216997e-05, "loss": 0.0806, "step": 34630 }, { "epoch": 2.91, "learning_rate": 1.418206247900571e-05, "loss": 0.1219, "step": 34640 }, { "epoch": 2.91, "learning_rate": 1.4180382935841453e-05, "loss": 0.0481, "step": 34650 }, { "epoch": 2.91, "learning_rate": 1.4178703392677193e-05, "loss": 0.0587, "step": 34660 }, { "epoch": 2.91, "learning_rate": 1.4177023849512934e-05, "loss": 0.1235, "step": 34670 }, { "epoch": 2.91, "learning_rate": 1.4175344306348674e-05, "loss": 0.1523, "step": 34680 }, { "epoch": 2.91, "learning_rate": 1.4173664763184417e-05, "loss": 0.1527, "step": 34690 }, { "epoch": 2.91, "learning_rate": 1.4171985220020155e-05, "loss": 0.144, "step": 34700 }, { "epoch": 2.91, "learning_rate": 1.4170305676855896e-05, "loss": 0.0597, "step": 34710 }, { "epoch": 2.92, "learning_rate": 1.4168626133691636e-05, "loss": 0.0867, "step": 34720 }, { "epoch": 2.92, "learning_rate": 1.4166946590527377e-05, "loss": 0.068, "step": 34730 }, { "epoch": 2.92, "learning_rate": 1.4165267047363119e-05, "loss": 0.1368, "step": 34740 }, { "epoch": 2.92, "learning_rate": 1.416358750419886e-05, "loss": 0.0069, "step": 34750 }, { "epoch": 2.92, "learning_rate": 1.41619079610346e-05, "loss": 0.1353, "step": 34760 }, { "epoch": 2.92, "learning_rate": 1.416022841787034e-05, "loss": 0.1466, "step": 34770 }, { "epoch": 2.92, "learning_rate": 1.4158548874706083e-05, "loss": 0.0155, "step": 34780 }, { "epoch": 2.92, "learning_rate": 1.4156869331541823e-05, "loss": 0.0629, "step": 34790 }, { "epoch": 2.92, "learning_rate": 1.4155189788377563e-05, "loss": 0.0994, "step": 34800 }, { "epoch": 2.92, "learning_rate": 1.4153510245213302e-05, "loss": 0.0692, "step": 34810 }, { "epoch": 2.92, "learning_rate": 1.4151830702049043e-05, "loss": 0.032, "step": 34820 }, { "epoch": 2.92, "learning_rate": 1.4150151158884783e-05, "loss": 0.0609, "step": 34830 }, { "epoch": 2.93, "learning_rate": 1.4148471615720525e-05, "loss": 0.0959, "step": 34840 }, { "epoch": 2.93, "learning_rate": 1.4146792072556266e-05, "loss": 0.1456, "step": 34850 }, { "epoch": 2.93, "learning_rate": 1.4145112529392006e-05, "loss": 0.175, "step": 34860 }, { "epoch": 2.93, "learning_rate": 1.4143432986227747e-05, "loss": 0.0654, "step": 34870 }, { "epoch": 2.93, "learning_rate": 1.4141753443063489e-05, "loss": 0.0553, "step": 34880 }, { "epoch": 2.93, "learning_rate": 1.414007389989923e-05, "loss": 0.184, "step": 34890 }, { "epoch": 2.93, "learning_rate": 1.413839435673497e-05, "loss": 0.1089, "step": 34900 }, { "epoch": 2.93, "learning_rate": 1.4136714813570709e-05, "loss": 0.1096, "step": 34910 }, { "epoch": 2.93, "learning_rate": 1.413503527040645e-05, "loss": 0.1421, "step": 34920 }, { "epoch": 2.93, "learning_rate": 1.4133355727242191e-05, "loss": 0.0738, "step": 34930 }, { "epoch": 2.93, "learning_rate": 1.4131676184077932e-05, "loss": 0.1458, "step": 34940 }, { "epoch": 2.94, "learning_rate": 1.4129996640913672e-05, "loss": 0.0566, "step": 34950 }, { "epoch": 2.94, "learning_rate": 1.4128317097749413e-05, "loss": 0.0286, "step": 34960 }, { "epoch": 2.94, "learning_rate": 1.4126637554585155e-05, "loss": 0.1037, "step": 34970 }, { "epoch": 2.94, "learning_rate": 1.4124958011420895e-05, "loss": 0.0849, "step": 34980 }, { "epoch": 2.94, "learning_rate": 1.4123278468256636e-05, "loss": 0.094, "step": 34990 }, { "epoch": 2.94, "learning_rate": 1.4121598925092376e-05, "loss": 0.126, "step": 35000 }, { "epoch": 2.94, "learning_rate": 1.4119919381928119e-05, "loss": 0.1625, "step": 35010 }, { "epoch": 2.94, "learning_rate": 1.4118239838763856e-05, "loss": 0.031, "step": 35020 }, { "epoch": 2.94, "learning_rate": 1.4116560295599598e-05, "loss": 0.0809, "step": 35030 }, { "epoch": 2.94, "learning_rate": 1.4114880752435338e-05, "loss": 0.1006, "step": 35040 }, { "epoch": 2.94, "learning_rate": 1.4113201209271079e-05, "loss": 0.0403, "step": 35050 }, { "epoch": 2.94, "learning_rate": 1.411152166610682e-05, "loss": 0.0971, "step": 35060 }, { "epoch": 2.95, "learning_rate": 1.4109842122942561e-05, "loss": 0.0879, "step": 35070 }, { "epoch": 2.95, "learning_rate": 1.4108162579778302e-05, "loss": 0.1225, "step": 35080 }, { "epoch": 2.95, "learning_rate": 1.4106483036614042e-05, "loss": 0.2002, "step": 35090 }, { "epoch": 2.95, "learning_rate": 1.4104803493449783e-05, "loss": 0.1567, "step": 35100 }, { "epoch": 2.95, "learning_rate": 1.4103123950285525e-05, "loss": 0.0352, "step": 35110 }, { "epoch": 2.95, "learning_rate": 1.4101444407121266e-05, "loss": 0.1134, "step": 35120 }, { "epoch": 2.95, "learning_rate": 1.4099764863957004e-05, "loss": 0.1004, "step": 35130 }, { "epoch": 2.95, "learning_rate": 1.4098085320792745e-05, "loss": 0.11, "step": 35140 }, { "epoch": 2.95, "learning_rate": 1.4096405777628485e-05, "loss": 0.075, "step": 35150 }, { "epoch": 2.95, "learning_rate": 1.4094726234464227e-05, "loss": 0.0989, "step": 35160 }, { "epoch": 2.95, "learning_rate": 1.4093046691299968e-05, "loss": 0.1314, "step": 35170 }, { "epoch": 2.95, "learning_rate": 1.4091367148135708e-05, "loss": 0.0933, "step": 35180 }, { "epoch": 2.96, "learning_rate": 1.4089687604971449e-05, "loss": 0.147, "step": 35190 }, { "epoch": 2.96, "learning_rate": 1.408800806180719e-05, "loss": 0.215, "step": 35200 }, { "epoch": 2.96, "learning_rate": 1.4086328518642931e-05, "loss": 0.0865, "step": 35210 }, { "epoch": 2.96, "learning_rate": 1.4084648975478672e-05, "loss": 0.1255, "step": 35220 }, { "epoch": 2.96, "learning_rate": 1.408296943231441e-05, "loss": 0.1495, "step": 35230 }, { "epoch": 2.96, "learning_rate": 1.4081289889150151e-05, "loss": 0.1083, "step": 35240 }, { "epoch": 2.96, "learning_rate": 1.4079610345985892e-05, "loss": 0.0498, "step": 35250 }, { "epoch": 2.96, "learning_rate": 1.4077930802821634e-05, "loss": 0.0576, "step": 35260 }, { "epoch": 2.96, "learning_rate": 1.4076251259657374e-05, "loss": 0.0424, "step": 35270 }, { "epoch": 2.96, "learning_rate": 1.4074571716493115e-05, "loss": 0.0496, "step": 35280 }, { "epoch": 2.96, "learning_rate": 1.4072892173328855e-05, "loss": 0.0583, "step": 35290 }, { "epoch": 2.96, "learning_rate": 1.4071212630164597e-05, "loss": 0.0349, "step": 35300 }, { "epoch": 2.97, "learning_rate": 1.4069533087000338e-05, "loss": 0.1845, "step": 35310 }, { "epoch": 2.97, "learning_rate": 1.4067853543836078e-05, "loss": 0.0747, "step": 35320 }, { "epoch": 2.97, "learning_rate": 1.4066174000671819e-05, "loss": 0.063, "step": 35330 }, { "epoch": 2.97, "learning_rate": 1.4064494457507558e-05, "loss": 0.1256, "step": 35340 }, { "epoch": 2.97, "learning_rate": 1.40628149143433e-05, "loss": 0.0838, "step": 35350 }, { "epoch": 2.97, "learning_rate": 1.406113537117904e-05, "loss": 0.0842, "step": 35360 }, { "epoch": 2.97, "learning_rate": 1.405945582801478e-05, "loss": 0.125, "step": 35370 }, { "epoch": 2.97, "learning_rate": 1.4057776284850521e-05, "loss": 0.2926, "step": 35380 }, { "epoch": 2.97, "learning_rate": 1.4056096741686262e-05, "loss": 0.1022, "step": 35390 }, { "epoch": 2.97, "learning_rate": 1.4054417198522004e-05, "loss": 0.1264, "step": 35400 }, { "epoch": 2.97, "learning_rate": 1.4052737655357744e-05, "loss": 0.1041, "step": 35410 }, { "epoch": 2.97, "learning_rate": 1.4051058112193485e-05, "loss": 0.0475, "step": 35420 }, { "epoch": 2.98, "learning_rate": 1.4049378569029225e-05, "loss": 0.0856, "step": 35430 }, { "epoch": 2.98, "learning_rate": 1.4047699025864964e-05, "loss": 0.1262, "step": 35440 }, { "epoch": 2.98, "learning_rate": 1.4046019482700706e-05, "loss": 0.0966, "step": 35450 }, { "epoch": 2.98, "learning_rate": 1.4044339939536447e-05, "loss": 0.0611, "step": 35460 }, { "epoch": 2.98, "learning_rate": 1.4042660396372187e-05, "loss": 0.0619, "step": 35470 }, { "epoch": 2.98, "learning_rate": 1.4040980853207928e-05, "loss": 0.0934, "step": 35480 }, { "epoch": 2.98, "learning_rate": 1.403930131004367e-05, "loss": 0.0833, "step": 35490 }, { "epoch": 2.98, "learning_rate": 1.403762176687941e-05, "loss": 0.0873, "step": 35500 }, { "epoch": 2.98, "learning_rate": 1.4035942223715151e-05, "loss": 0.143, "step": 35510 }, { "epoch": 2.98, "learning_rate": 1.4034262680550891e-05, "loss": 0.1254, "step": 35520 }, { "epoch": 2.98, "learning_rate": 1.4032583137386634e-05, "loss": 0.1029, "step": 35530 }, { "epoch": 2.98, "learning_rate": 1.4030903594222374e-05, "loss": 0.0947, "step": 35540 }, { "epoch": 2.99, "learning_rate": 1.4029224051058113e-05, "loss": 0.0965, "step": 35550 }, { "epoch": 2.99, "learning_rate": 1.4027544507893853e-05, "loss": 0.1583, "step": 35560 }, { "epoch": 2.99, "learning_rate": 1.4025864964729594e-05, "loss": 0.0918, "step": 35570 }, { "epoch": 2.99, "learning_rate": 1.4024185421565334e-05, "loss": 0.0577, "step": 35580 }, { "epoch": 2.99, "learning_rate": 1.4022505878401076e-05, "loss": 0.1886, "step": 35590 }, { "epoch": 2.99, "learning_rate": 1.4020826335236817e-05, "loss": 0.0564, "step": 35600 }, { "epoch": 2.99, "learning_rate": 1.4019146792072557e-05, "loss": 0.0723, "step": 35610 }, { "epoch": 2.99, "learning_rate": 1.4017467248908298e-05, "loss": 0.0846, "step": 35620 }, { "epoch": 2.99, "learning_rate": 1.401578770574404e-05, "loss": 0.1606, "step": 35630 }, { "epoch": 2.99, "learning_rate": 1.401410816257978e-05, "loss": 0.0316, "step": 35640 }, { "epoch": 2.99, "learning_rate": 1.401242861941552e-05, "loss": 0.1604, "step": 35650 }, { "epoch": 2.99, "learning_rate": 1.401074907625126e-05, "loss": 0.0594, "step": 35660 }, { "epoch": 3.0, "learning_rate": 1.4009069533087e-05, "loss": 0.1187, "step": 35670 }, { "epoch": 3.0, "learning_rate": 1.4007389989922742e-05, "loss": 0.0555, "step": 35680 }, { "epoch": 3.0, "learning_rate": 1.4005710446758483e-05, "loss": 0.0608, "step": 35690 }, { "epoch": 3.0, "learning_rate": 1.4004030903594223e-05, "loss": 0.0427, "step": 35700 }, { "epoch": 3.0, "learning_rate": 1.4002351360429964e-05, "loss": 0.0919, "step": 35710 }, { "epoch": 3.0, "learning_rate": 1.4000671817265706e-05, "loss": 0.0552, "step": 35720 }, { "epoch": 3.0, "eval_FN": 561, "eval_FP": 425, "eval_TN": 21289, "eval_TP": 1541, "eval_accuracy": 0.9585992610010077, "eval_f1": 0.7576204523107177, "eval_loss": 0.15207499265670776, "eval_precision": 0.78382502543235, "eval_recall": 0.7331113225499524, "eval_runtime": 163.9489, "eval_samples_per_second": 145.265, "eval_steps_per_second": 9.082, "step": 35724 }, { "epoch": 3.0, "learning_rate": 1.3998992274101446e-05, "loss": 0.0648, "step": 35730 }, { "epoch": 3.0, "learning_rate": 1.3997312730937187e-05, "loss": 0.0769, "step": 35740 }, { "epoch": 3.0, "learning_rate": 1.3995633187772927e-05, "loss": 0.1051, "step": 35750 }, { "epoch": 3.0, "learning_rate": 1.3993953644608666e-05, "loss": 0.0464, "step": 35760 }, { "epoch": 3.0, "learning_rate": 1.3992274101444407e-05, "loss": 0.067, "step": 35770 }, { "epoch": 3.0, "learning_rate": 1.3990594558280149e-05, "loss": 0.1226, "step": 35780 }, { "epoch": 3.01, "learning_rate": 1.398891501511589e-05, "loss": 0.0236, "step": 35790 }, { "epoch": 3.01, "learning_rate": 1.398723547195163e-05, "loss": 0.081, "step": 35800 }, { "epoch": 3.01, "learning_rate": 1.398555592878737e-05, "loss": 0.0928, "step": 35810 }, { "epoch": 3.01, "learning_rate": 1.3983876385623112e-05, "loss": 0.0616, "step": 35820 }, { "epoch": 3.01, "learning_rate": 1.3982196842458853e-05, "loss": 0.0679, "step": 35830 }, { "epoch": 3.01, "learning_rate": 1.3980517299294593e-05, "loss": 0.0619, "step": 35840 }, { "epoch": 3.01, "learning_rate": 1.3978837756130334e-05, "loss": 0.0038, "step": 35850 }, { "epoch": 3.01, "learning_rate": 1.3977158212966076e-05, "loss": 0.0685, "step": 35860 }, { "epoch": 3.01, "learning_rate": 1.3975478669801815e-05, "loss": 0.0467, "step": 35870 }, { "epoch": 3.01, "learning_rate": 1.3973799126637555e-05, "loss": 0.0711, "step": 35880 }, { "epoch": 3.01, "learning_rate": 1.3972119583473296e-05, "loss": 0.0177, "step": 35890 }, { "epoch": 3.01, "learning_rate": 1.3970440040309036e-05, "loss": 0.0315, "step": 35900 }, { "epoch": 3.02, "learning_rate": 1.3968760497144778e-05, "loss": 0.1016, "step": 35910 }, { "epoch": 3.02, "learning_rate": 1.3967080953980519e-05, "loss": 0.0975, "step": 35920 }, { "epoch": 3.02, "learning_rate": 1.396540141081626e-05, "loss": 0.0144, "step": 35930 }, { "epoch": 3.02, "learning_rate": 1.3963721867652e-05, "loss": 0.0683, "step": 35940 }, { "epoch": 3.02, "learning_rate": 1.3962042324487742e-05, "loss": 0.0016, "step": 35950 }, { "epoch": 3.02, "learning_rate": 1.3960362781323482e-05, "loss": 0.0457, "step": 35960 }, { "epoch": 3.02, "learning_rate": 1.3958683238159221e-05, "loss": 0.0976, "step": 35970 }, { "epoch": 3.02, "learning_rate": 1.3957003694994962e-05, "loss": 0.0702, "step": 35980 }, { "epoch": 3.02, "learning_rate": 1.3955324151830702e-05, "loss": 0.0669, "step": 35990 }, { "epoch": 3.02, "learning_rate": 1.3953644608666443e-05, "loss": 0.0689, "step": 36000 }, { "epoch": 3.02, "learning_rate": 1.3951965065502185e-05, "loss": 0.0764, "step": 36010 }, { "epoch": 3.02, "learning_rate": 1.3950285522337925e-05, "loss": 0.0421, "step": 36020 }, { "epoch": 3.03, "learning_rate": 1.3948605979173666e-05, "loss": 0.0081, "step": 36030 }, { "epoch": 3.03, "learning_rate": 1.3946926436009406e-05, "loss": 0.0905, "step": 36040 }, { "epoch": 3.03, "learning_rate": 1.3945246892845148e-05, "loss": 0.0468, "step": 36050 }, { "epoch": 3.03, "learning_rate": 1.3943567349680889e-05, "loss": 0.0483, "step": 36060 }, { "epoch": 3.03, "learning_rate": 1.394188780651663e-05, "loss": 0.0875, "step": 36070 }, { "epoch": 3.03, "learning_rate": 1.3940208263352368e-05, "loss": 0.0796, "step": 36080 }, { "epoch": 3.03, "learning_rate": 1.3938528720188109e-05, "loss": 0.0221, "step": 36090 }, { "epoch": 3.03, "learning_rate": 1.3936849177023851e-05, "loss": 0.1007, "step": 36100 }, { "epoch": 3.03, "learning_rate": 1.3935169633859591e-05, "loss": 0.1028, "step": 36110 }, { "epoch": 3.03, "learning_rate": 1.3933490090695332e-05, "loss": 0.1172, "step": 36120 }, { "epoch": 3.03, "learning_rate": 1.3931810547531072e-05, "loss": 0.1181, "step": 36130 }, { "epoch": 3.03, "learning_rate": 1.3930131004366814e-05, "loss": 0.0545, "step": 36140 }, { "epoch": 3.04, "learning_rate": 1.3928451461202555e-05, "loss": 0.0868, "step": 36150 }, { "epoch": 3.04, "learning_rate": 1.3926771918038295e-05, "loss": 0.0149, "step": 36160 }, { "epoch": 3.04, "learning_rate": 1.3925092374874036e-05, "loss": 0.0394, "step": 36170 }, { "epoch": 3.04, "learning_rate": 1.3923412831709775e-05, "loss": 0.0552, "step": 36180 }, { "epoch": 3.04, "learning_rate": 1.3921733288545515e-05, "loss": 0.0412, "step": 36190 }, { "epoch": 3.04, "learning_rate": 1.3920053745381257e-05, "loss": 0.054, "step": 36200 }, { "epoch": 3.04, "learning_rate": 1.3918374202216998e-05, "loss": 0.0607, "step": 36210 }, { "epoch": 3.04, "learning_rate": 1.3916694659052738e-05, "loss": 0.1073, "step": 36220 }, { "epoch": 3.04, "learning_rate": 1.3915015115888479e-05, "loss": 0.1277, "step": 36230 }, { "epoch": 3.04, "learning_rate": 1.3913335572724221e-05, "loss": 0.0264, "step": 36240 }, { "epoch": 3.04, "learning_rate": 1.3911656029559961e-05, "loss": 0.0223, "step": 36250 }, { "epoch": 3.05, "learning_rate": 1.3909976486395702e-05, "loss": 0.081, "step": 36260 }, { "epoch": 3.05, "learning_rate": 1.3908296943231442e-05, "loss": 0.1523, "step": 36270 }, { "epoch": 3.05, "learning_rate": 1.3906617400067185e-05, "loss": 0.0267, "step": 36280 }, { "epoch": 3.05, "learning_rate": 1.3904937856902923e-05, "loss": 0.0759, "step": 36290 }, { "epoch": 3.05, "learning_rate": 1.3903258313738664e-05, "loss": 0.0984, "step": 36300 }, { "epoch": 3.05, "learning_rate": 1.3901578770574404e-05, "loss": 0.1074, "step": 36310 }, { "epoch": 3.05, "learning_rate": 1.3899899227410145e-05, "loss": 0.0865, "step": 36320 }, { "epoch": 3.05, "learning_rate": 1.3898219684245887e-05, "loss": 0.133, "step": 36330 }, { "epoch": 3.05, "learning_rate": 1.3896540141081627e-05, "loss": 0.0712, "step": 36340 }, { "epoch": 3.05, "learning_rate": 1.3894860597917368e-05, "loss": 0.1045, "step": 36350 }, { "epoch": 3.05, "learning_rate": 1.3893181054753108e-05, "loss": 0.0606, "step": 36360 }, { "epoch": 3.05, "learning_rate": 1.3891501511588849e-05, "loss": 0.1291, "step": 36370 }, { "epoch": 3.06, "learning_rate": 1.3889821968424591e-05, "loss": 0.0099, "step": 36380 }, { "epoch": 3.06, "learning_rate": 1.3888142425260331e-05, "loss": 0.0434, "step": 36390 }, { "epoch": 3.06, "learning_rate": 1.388646288209607e-05, "loss": 0.0051, "step": 36400 }, { "epoch": 3.06, "learning_rate": 1.388478333893181e-05, "loss": 0.1516, "step": 36410 }, { "epoch": 3.06, "learning_rate": 1.3883103795767551e-05, "loss": 0.1496, "step": 36420 }, { "epoch": 3.06, "learning_rate": 1.3881424252603293e-05, "loss": 0.0243, "step": 36430 }, { "epoch": 3.06, "learning_rate": 1.3879744709439034e-05, "loss": 0.0421, "step": 36440 }, { "epoch": 3.06, "learning_rate": 1.3878065166274774e-05, "loss": 0.0933, "step": 36450 }, { "epoch": 3.06, "learning_rate": 1.3876385623110515e-05, "loss": 0.166, "step": 36460 }, { "epoch": 3.06, "learning_rate": 1.3874706079946257e-05, "loss": 0.0877, "step": 36470 }, { "epoch": 3.06, "learning_rate": 1.3873026536781997e-05, "loss": 0.0857, "step": 36480 }, { "epoch": 3.06, "learning_rate": 1.3871346993617738e-05, "loss": 0.1795, "step": 36490 }, { "epoch": 3.07, "learning_rate": 1.3869667450453477e-05, "loss": 0.0623, "step": 36500 }, { "epoch": 3.07, "learning_rate": 1.3867987907289217e-05, "loss": 0.0689, "step": 36510 }, { "epoch": 3.07, "learning_rate": 1.386630836412496e-05, "loss": 0.0706, "step": 36520 }, { "epoch": 3.07, "learning_rate": 1.38646288209607e-05, "loss": 0.0376, "step": 36530 }, { "epoch": 3.07, "learning_rate": 1.386294927779644e-05, "loss": 0.0931, "step": 36540 }, { "epoch": 3.07, "learning_rate": 1.386126973463218e-05, "loss": 0.0888, "step": 36550 }, { "epoch": 3.07, "learning_rate": 1.3859590191467921e-05, "loss": 0.0408, "step": 36560 }, { "epoch": 3.07, "learning_rate": 1.3857910648303663e-05, "loss": 0.0349, "step": 36570 }, { "epoch": 3.07, "learning_rate": 1.3856231105139404e-05, "loss": 0.0682, "step": 36580 }, { "epoch": 3.07, "learning_rate": 1.3854551561975144e-05, "loss": 0.0394, "step": 36590 }, { "epoch": 3.07, "learning_rate": 1.3852872018810885e-05, "loss": 0.0319, "step": 36600 }, { "epoch": 3.07, "learning_rate": 1.3851192475646624e-05, "loss": 0.1083, "step": 36610 }, { "epoch": 3.08, "learning_rate": 1.3849512932482366e-05, "loss": 0.1306, "step": 36620 }, { "epoch": 3.08, "learning_rate": 1.3847833389318106e-05, "loss": 0.0685, "step": 36630 }, { "epoch": 3.08, "learning_rate": 1.3846153846153847e-05, "loss": 0.0833, "step": 36640 }, { "epoch": 3.08, "learning_rate": 1.3844474302989587e-05, "loss": 0.1309, "step": 36650 }, { "epoch": 3.08, "learning_rate": 1.384279475982533e-05, "loss": 0.1391, "step": 36660 }, { "epoch": 3.08, "learning_rate": 1.384111521666107e-05, "loss": 0.0509, "step": 36670 }, { "epoch": 3.08, "learning_rate": 1.383943567349681e-05, "loss": 0.1026, "step": 36680 }, { "epoch": 3.08, "learning_rate": 1.383775613033255e-05, "loss": 0.0568, "step": 36690 }, { "epoch": 3.08, "learning_rate": 1.3836076587168293e-05, "loss": 0.0392, "step": 36700 }, { "epoch": 3.08, "learning_rate": 1.383439704400403e-05, "loss": 0.0228, "step": 36710 }, { "epoch": 3.08, "learning_rate": 1.3832717500839772e-05, "loss": 0.0461, "step": 36720 }, { "epoch": 3.08, "learning_rate": 1.3831037957675513e-05, "loss": 0.107, "step": 36730 }, { "epoch": 3.09, "learning_rate": 1.3829358414511253e-05, "loss": 0.103, "step": 36740 }, { "epoch": 3.09, "learning_rate": 1.3827678871346994e-05, "loss": 0.0998, "step": 36750 }, { "epoch": 3.09, "learning_rate": 1.3825999328182736e-05, "loss": 0.1583, "step": 36760 }, { "epoch": 3.09, "learning_rate": 1.3824319785018476e-05, "loss": 0.0742, "step": 36770 }, { "epoch": 3.09, "learning_rate": 1.3822640241854217e-05, "loss": 0.0348, "step": 36780 }, { "epoch": 3.09, "learning_rate": 1.3820960698689957e-05, "loss": 0.074, "step": 36790 }, { "epoch": 3.09, "learning_rate": 1.38192811555257e-05, "loss": 0.0857, "step": 36800 }, { "epoch": 3.09, "learning_rate": 1.381760161236144e-05, "loss": 0.1396, "step": 36810 }, { "epoch": 3.09, "learning_rate": 1.3815922069197179e-05, "loss": 0.0034, "step": 36820 }, { "epoch": 3.09, "learning_rate": 1.381424252603292e-05, "loss": 0.1203, "step": 36830 }, { "epoch": 3.09, "learning_rate": 1.381256298286866e-05, "loss": 0.0602, "step": 36840 }, { "epoch": 3.09, "learning_rate": 1.3810883439704402e-05, "loss": 0.1355, "step": 36850 }, { "epoch": 3.1, "learning_rate": 1.3809203896540142e-05, "loss": 0.0286, "step": 36860 }, { "epoch": 3.1, "learning_rate": 1.3807524353375883e-05, "loss": 0.0447, "step": 36870 }, { "epoch": 3.1, "learning_rate": 1.3805844810211623e-05, "loss": 0.1268, "step": 36880 }, { "epoch": 3.1, "learning_rate": 1.3804165267047365e-05, "loss": 0.1644, "step": 36890 }, { "epoch": 3.1, "learning_rate": 1.3802485723883106e-05, "loss": 0.0391, "step": 36900 }, { "epoch": 3.1, "learning_rate": 1.3800806180718846e-05, "loss": 0.087, "step": 36910 }, { "epoch": 3.1, "learning_rate": 1.3799126637554587e-05, "loss": 0.0563, "step": 36920 }, { "epoch": 3.1, "learning_rate": 1.3797447094390326e-05, "loss": 0.1377, "step": 36930 }, { "epoch": 3.1, "learning_rate": 1.3795767551226066e-05, "loss": 0.0272, "step": 36940 }, { "epoch": 3.1, "learning_rate": 1.3794088008061808e-05, "loss": 0.0731, "step": 36950 }, { "epoch": 3.1, "learning_rate": 1.3792408464897549e-05, "loss": 0.0203, "step": 36960 }, { "epoch": 3.1, "learning_rate": 1.379072892173329e-05, "loss": 0.0502, "step": 36970 }, { "epoch": 3.11, "learning_rate": 1.378904937856903e-05, "loss": 0.0391, "step": 36980 }, { "epoch": 3.11, "learning_rate": 1.3787369835404772e-05, "loss": 0.034, "step": 36990 }, { "epoch": 3.11, "learning_rate": 1.3785690292240512e-05, "loss": 0.0527, "step": 37000 }, { "epoch": 3.11, "learning_rate": 1.3784010749076253e-05, "loss": 0.0674, "step": 37010 }, { "epoch": 3.11, "learning_rate": 1.3782331205911993e-05, "loss": 0.0653, "step": 37020 }, { "epoch": 3.11, "learning_rate": 1.3780651662747732e-05, "loss": 0.0621, "step": 37030 }, { "epoch": 3.11, "learning_rate": 1.3778972119583474e-05, "loss": 0.1005, "step": 37040 }, { "epoch": 3.11, "learning_rate": 1.3777292576419215e-05, "loss": 0.0911, "step": 37050 }, { "epoch": 3.11, "learning_rate": 1.3775613033254955e-05, "loss": 0.1417, "step": 37060 }, { "epoch": 3.11, "learning_rate": 1.3773933490090696e-05, "loss": 0.0617, "step": 37070 }, { "epoch": 3.11, "learning_rate": 1.3772253946926438e-05, "loss": 0.0281, "step": 37080 }, { "epoch": 3.11, "learning_rate": 1.3770574403762178e-05, "loss": 0.0495, "step": 37090 }, { "epoch": 3.12, "learning_rate": 1.3768894860597919e-05, "loss": 0.1076, "step": 37100 }, { "epoch": 3.12, "learning_rate": 1.376721531743366e-05, "loss": 0.048, "step": 37110 }, { "epoch": 3.12, "learning_rate": 1.3765535774269402e-05, "loss": 0.0619, "step": 37120 }, { "epoch": 3.12, "learning_rate": 1.3763856231105142e-05, "loss": 0.1149, "step": 37130 }, { "epoch": 3.12, "learning_rate": 1.376217668794088e-05, "loss": 0.0681, "step": 37140 }, { "epoch": 3.12, "learning_rate": 1.3760497144776621e-05, "loss": 0.0769, "step": 37150 }, { "epoch": 3.12, "learning_rate": 1.3758817601612362e-05, "loss": 0.0493, "step": 37160 }, { "epoch": 3.12, "learning_rate": 1.3757138058448102e-05, "loss": 0.0781, "step": 37170 }, { "epoch": 3.12, "learning_rate": 1.3755458515283844e-05, "loss": 0.0192, "step": 37180 }, { "epoch": 3.12, "learning_rate": 1.3753778972119585e-05, "loss": 0.0775, "step": 37190 }, { "epoch": 3.12, "learning_rate": 1.3752099428955325e-05, "loss": 0.0606, "step": 37200 }, { "epoch": 3.12, "learning_rate": 1.3750419885791066e-05, "loss": 0.1688, "step": 37210 }, { "epoch": 3.13, "learning_rate": 1.3748740342626808e-05, "loss": 0.0548, "step": 37220 }, { "epoch": 3.13, "learning_rate": 1.3747060799462548e-05, "loss": 0.0584, "step": 37230 }, { "epoch": 3.13, "learning_rate": 1.3745381256298287e-05, "loss": 0.0737, "step": 37240 }, { "epoch": 3.13, "learning_rate": 1.3743701713134028e-05, "loss": 0.0769, "step": 37250 }, { "epoch": 3.13, "learning_rate": 1.3742022169969768e-05, "loss": 0.0075, "step": 37260 }, { "epoch": 3.13, "learning_rate": 1.374034262680551e-05, "loss": 0.0754, "step": 37270 }, { "epoch": 3.13, "learning_rate": 1.373866308364125e-05, "loss": 0.0223, "step": 37280 }, { "epoch": 3.13, "learning_rate": 1.3736983540476991e-05, "loss": 0.112, "step": 37290 }, { "epoch": 3.13, "learning_rate": 1.3735303997312732e-05, "loss": 0.0517, "step": 37300 }, { "epoch": 3.13, "learning_rate": 1.3733624454148474e-05, "loss": 0.0829, "step": 37310 }, { "epoch": 3.13, "learning_rate": 1.3731944910984214e-05, "loss": 0.0613, "step": 37320 }, { "epoch": 3.13, "learning_rate": 1.3730265367819955e-05, "loss": 0.0625, "step": 37330 }, { "epoch": 3.14, "learning_rate": 1.3728585824655695e-05, "loss": 0.04, "step": 37340 }, { "epoch": 3.14, "learning_rate": 1.3726906281491434e-05, "loss": 0.0952, "step": 37350 }, { "epoch": 3.14, "learning_rate": 1.3725226738327175e-05, "loss": 0.0588, "step": 37360 }, { "epoch": 3.14, "learning_rate": 1.3723547195162917e-05, "loss": 0.0417, "step": 37370 }, { "epoch": 3.14, "learning_rate": 1.3721867651998657e-05, "loss": 0.1346, "step": 37380 }, { "epoch": 3.14, "learning_rate": 1.3720188108834398e-05, "loss": 0.0036, "step": 37390 }, { "epoch": 3.14, "learning_rate": 1.3718508565670138e-05, "loss": 0.0355, "step": 37400 }, { "epoch": 3.14, "learning_rate": 1.371682902250588e-05, "loss": 0.1085, "step": 37410 }, { "epoch": 3.14, "learning_rate": 1.3715149479341621e-05, "loss": 0.0784, "step": 37420 }, { "epoch": 3.14, "learning_rate": 1.3713469936177361e-05, "loss": 0.1371, "step": 37430 }, { "epoch": 3.14, "learning_rate": 1.3711790393013102e-05, "loss": 0.0625, "step": 37440 }, { "epoch": 3.14, "learning_rate": 1.3710110849848844e-05, "loss": 0.0232, "step": 37450 }, { "epoch": 3.15, "learning_rate": 1.3708431306684583e-05, "loss": 0.0704, "step": 37460 }, { "epoch": 3.15, "learning_rate": 1.3706751763520323e-05, "loss": 0.0715, "step": 37470 }, { "epoch": 3.15, "learning_rate": 1.3705072220356064e-05, "loss": 0.0451, "step": 37480 }, { "epoch": 3.15, "learning_rate": 1.3703392677191804e-05, "loss": 0.0723, "step": 37490 }, { "epoch": 3.15, "learning_rate": 1.3701713134027546e-05, "loss": 0.0253, "step": 37500 }, { "epoch": 3.15, "learning_rate": 1.3700033590863287e-05, "loss": 0.0216, "step": 37510 }, { "epoch": 3.15, "learning_rate": 1.3698354047699027e-05, "loss": 0.0762, "step": 37520 }, { "epoch": 3.15, "learning_rate": 1.3696674504534768e-05, "loss": 0.1012, "step": 37530 }, { "epoch": 3.15, "learning_rate": 1.3694994961370508e-05, "loss": 0.1455, "step": 37540 }, { "epoch": 3.15, "learning_rate": 1.369331541820625e-05, "loss": 0.1524, "step": 37550 }, { "epoch": 3.15, "learning_rate": 1.369163587504199e-05, "loss": 0.0607, "step": 37560 }, { "epoch": 3.16, "learning_rate": 1.368995633187773e-05, "loss": 0.0537, "step": 37570 }, { "epoch": 3.16, "learning_rate": 1.368827678871347e-05, "loss": 0.0273, "step": 37580 }, { "epoch": 3.16, "learning_rate": 1.368659724554921e-05, "loss": 0.03, "step": 37590 }, { "epoch": 3.16, "learning_rate": 1.3684917702384953e-05, "loss": 0.0566, "step": 37600 }, { "epoch": 3.16, "learning_rate": 1.3683238159220693e-05, "loss": 0.0376, "step": 37610 }, { "epoch": 3.16, "learning_rate": 1.3681558616056434e-05, "loss": 0.052, "step": 37620 }, { "epoch": 3.16, "learning_rate": 1.3679879072892174e-05, "loss": 0.0757, "step": 37630 }, { "epoch": 3.16, "learning_rate": 1.3678199529727916e-05, "loss": 0.06, "step": 37640 }, { "epoch": 3.16, "learning_rate": 1.3676519986563657e-05, "loss": 0.0369, "step": 37650 }, { "epoch": 3.16, "learning_rate": 1.3674840443399397e-05, "loss": 0.0578, "step": 37660 }, { "epoch": 3.16, "learning_rate": 1.3673160900235136e-05, "loss": 0.0681, "step": 37670 }, { "epoch": 3.16, "learning_rate": 1.3671481357070877e-05, "loss": 0.073, "step": 37680 }, { "epoch": 3.17, "learning_rate": 1.3669801813906617e-05, "loss": 0.0254, "step": 37690 }, { "epoch": 3.17, "learning_rate": 1.366812227074236e-05, "loss": 0.0724, "step": 37700 }, { "epoch": 3.17, "learning_rate": 1.36664427275781e-05, "loss": 0.0081, "step": 37710 }, { "epoch": 3.17, "learning_rate": 1.366476318441384e-05, "loss": 0.1059, "step": 37720 }, { "epoch": 3.17, "learning_rate": 1.366308364124958e-05, "loss": 0.0262, "step": 37730 }, { "epoch": 3.17, "learning_rate": 1.3661404098085323e-05, "loss": 0.2053, "step": 37740 }, { "epoch": 3.17, "learning_rate": 1.3659724554921063e-05, "loss": 0.0339, "step": 37750 }, { "epoch": 3.17, "learning_rate": 1.3658045011756804e-05, "loss": 0.0777, "step": 37760 }, { "epoch": 3.17, "learning_rate": 1.3656365468592543e-05, "loss": 0.0495, "step": 37770 }, { "epoch": 3.17, "learning_rate": 1.3654685925428283e-05, "loss": 0.1178, "step": 37780 }, { "epoch": 3.17, "learning_rate": 1.3653006382264025e-05, "loss": 0.0576, "step": 37790 }, { "epoch": 3.17, "learning_rate": 1.3651326839099766e-05, "loss": 0.06, "step": 37800 }, { "epoch": 3.18, "learning_rate": 1.3649647295935506e-05, "loss": 0.1276, "step": 37810 }, { "epoch": 3.18, "learning_rate": 1.3647967752771247e-05, "loss": 0.069, "step": 37820 }, { "epoch": 3.18, "learning_rate": 1.3646288209606989e-05, "loss": 0.0821, "step": 37830 }, { "epoch": 3.18, "learning_rate": 1.364460866644273e-05, "loss": 0.0765, "step": 37840 }, { "epoch": 3.18, "learning_rate": 1.364292912327847e-05, "loss": 0.0524, "step": 37850 }, { "epoch": 3.18, "learning_rate": 1.364124958011421e-05, "loss": 0.1864, "step": 37860 }, { "epoch": 3.18, "learning_rate": 1.3639570036949953e-05, "loss": 0.018, "step": 37870 }, { "epoch": 3.18, "learning_rate": 1.363789049378569e-05, "loss": 0.0672, "step": 37880 }, { "epoch": 3.18, "learning_rate": 1.3636210950621432e-05, "loss": 0.1102, "step": 37890 }, { "epoch": 3.18, "learning_rate": 1.3634531407457172e-05, "loss": 0.1618, "step": 37900 }, { "epoch": 3.18, "learning_rate": 1.3632851864292913e-05, "loss": 0.0945, "step": 37910 }, { "epoch": 3.18, "learning_rate": 1.3631172321128653e-05, "loss": 0.0079, "step": 37920 }, { "epoch": 3.19, "learning_rate": 1.3629492777964395e-05, "loss": 0.1265, "step": 37930 }, { "epoch": 3.19, "learning_rate": 1.3627813234800136e-05, "loss": 0.0044, "step": 37940 }, { "epoch": 3.19, "learning_rate": 1.3626133691635876e-05, "loss": 0.0038, "step": 37950 }, { "epoch": 3.19, "learning_rate": 1.3624454148471617e-05, "loss": 0.0305, "step": 37960 }, { "epoch": 3.19, "learning_rate": 1.3622774605307359e-05, "loss": 0.0818, "step": 37970 }, { "epoch": 3.19, "learning_rate": 1.36210950621431e-05, "loss": 0.029, "step": 37980 }, { "epoch": 3.19, "learning_rate": 1.3619415518978838e-05, "loss": 0.1, "step": 37990 }, { "epoch": 3.19, "learning_rate": 1.3617735975814579e-05, "loss": 0.1434, "step": 38000 }, { "epoch": 3.19, "learning_rate": 1.361605643265032e-05, "loss": 0.0473, "step": 38010 }, { "epoch": 3.19, "learning_rate": 1.3614376889486061e-05, "loss": 0.1797, "step": 38020 }, { "epoch": 3.19, "learning_rate": 1.3612697346321802e-05, "loss": 0.0481, "step": 38030 }, { "epoch": 3.19, "learning_rate": 1.3611017803157542e-05, "loss": 0.0344, "step": 38040 }, { "epoch": 3.2, "learning_rate": 1.3609338259993283e-05, "loss": 0.0325, "step": 38050 }, { "epoch": 3.2, "learning_rate": 1.3607658716829025e-05, "loss": 0.0483, "step": 38060 }, { "epoch": 3.2, "learning_rate": 1.3605979173664765e-05, "loss": 0.1718, "step": 38070 }, { "epoch": 3.2, "learning_rate": 1.3604299630500506e-05, "loss": 0.086, "step": 38080 }, { "epoch": 3.2, "learning_rate": 1.3602620087336245e-05, "loss": 0.0964, "step": 38090 }, { "epoch": 3.2, "learning_rate": 1.3600940544171985e-05, "loss": 0.0469, "step": 38100 }, { "epoch": 3.2, "learning_rate": 1.3599261001007726e-05, "loss": 0.1435, "step": 38110 }, { "epoch": 3.2, "learning_rate": 1.3597581457843468e-05, "loss": 0.0226, "step": 38120 }, { "epoch": 3.2, "learning_rate": 1.3595901914679208e-05, "loss": 0.0749, "step": 38130 }, { "epoch": 3.2, "learning_rate": 1.3594222371514949e-05, "loss": 0.0561, "step": 38140 }, { "epoch": 3.2, "learning_rate": 1.359254282835069e-05, "loss": 0.0476, "step": 38150 }, { "epoch": 3.2, "learning_rate": 1.3590863285186431e-05, "loss": 0.0741, "step": 38160 }, { "epoch": 3.21, "learning_rate": 1.3589183742022172e-05, "loss": 0.1364, "step": 38170 }, { "epoch": 3.21, "learning_rate": 1.3587504198857912e-05, "loss": 0.044, "step": 38180 }, { "epoch": 3.21, "learning_rate": 1.3585824655693653e-05, "loss": 0.0591, "step": 38190 }, { "epoch": 3.21, "learning_rate": 1.3584145112529392e-05, "loss": 0.0629, "step": 38200 }, { "epoch": 3.21, "learning_rate": 1.3582465569365134e-05, "loss": 0.037, "step": 38210 }, { "epoch": 3.21, "learning_rate": 1.3580786026200874e-05, "loss": 0.1017, "step": 38220 }, { "epoch": 3.21, "learning_rate": 1.3579106483036615e-05, "loss": 0.0535, "step": 38230 }, { "epoch": 3.21, "learning_rate": 1.3577426939872355e-05, "loss": 0.0449, "step": 38240 }, { "epoch": 3.21, "learning_rate": 1.3575747396708097e-05, "loss": 0.088, "step": 38250 }, { "epoch": 3.21, "learning_rate": 1.3574067853543838e-05, "loss": 0.0423, "step": 38260 }, { "epoch": 3.21, "learning_rate": 1.3572388310379578e-05, "loss": 0.1833, "step": 38270 }, { "epoch": 3.21, "learning_rate": 1.3570708767215319e-05, "loss": 0.0943, "step": 38280 }, { "epoch": 3.22, "learning_rate": 1.3569029224051061e-05, "loss": 0.0912, "step": 38290 }, { "epoch": 3.22, "learning_rate": 1.3567349680886798e-05, "loss": 0.0775, "step": 38300 }, { "epoch": 3.22, "learning_rate": 1.356567013772254e-05, "loss": 0.1074, "step": 38310 }, { "epoch": 3.22, "learning_rate": 1.356399059455828e-05, "loss": 0.0841, "step": 38320 }, { "epoch": 3.22, "learning_rate": 1.3562311051394021e-05, "loss": 0.0823, "step": 38330 }, { "epoch": 3.22, "learning_rate": 1.3560631508229762e-05, "loss": 0.0352, "step": 38340 }, { "epoch": 3.22, "learning_rate": 1.3558951965065504e-05, "loss": 0.0346, "step": 38350 }, { "epoch": 3.22, "learning_rate": 1.3557272421901244e-05, "loss": 0.1723, "step": 38360 }, { "epoch": 3.22, "learning_rate": 1.3555592878736985e-05, "loss": 0.0542, "step": 38370 }, { "epoch": 3.22, "learning_rate": 1.3553913335572725e-05, "loss": 0.1523, "step": 38380 }, { "epoch": 3.22, "learning_rate": 1.3552233792408467e-05, "loss": 0.1186, "step": 38390 }, { "epoch": 3.22, "learning_rate": 1.3550554249244208e-05, "loss": 0.0473, "step": 38400 }, { "epoch": 3.23, "learning_rate": 1.3548874706079947e-05, "loss": 0.03, "step": 38410 }, { "epoch": 3.23, "learning_rate": 1.3547195162915687e-05, "loss": 0.0189, "step": 38420 }, { "epoch": 3.23, "learning_rate": 1.3545515619751428e-05, "loss": 0.1463, "step": 38430 }, { "epoch": 3.23, "learning_rate": 1.354383607658717e-05, "loss": 0.0029, "step": 38440 }, { "epoch": 3.23, "learning_rate": 1.354215653342291e-05, "loss": 0.1024, "step": 38450 }, { "epoch": 3.23, "learning_rate": 1.354047699025865e-05, "loss": 0.1234, "step": 38460 }, { "epoch": 3.23, "learning_rate": 1.3538797447094391e-05, "loss": 0.0889, "step": 38470 }, { "epoch": 3.23, "learning_rate": 1.3537117903930132e-05, "loss": 0.1244, "step": 38480 }, { "epoch": 3.23, "learning_rate": 1.3535438360765874e-05, "loss": 0.0728, "step": 38490 }, { "epoch": 3.23, "learning_rate": 1.3533758817601614e-05, "loss": 0.1496, "step": 38500 }, { "epoch": 3.23, "learning_rate": 1.3532079274437355e-05, "loss": 0.0491, "step": 38510 }, { "epoch": 3.23, "learning_rate": 1.3530399731273094e-05, "loss": 0.0416, "step": 38520 }, { "epoch": 3.24, "learning_rate": 1.3528720188108834e-05, "loss": 0.0297, "step": 38530 }, { "epoch": 3.24, "learning_rate": 1.3527040644944576e-05, "loss": 0.1134, "step": 38540 }, { "epoch": 3.24, "learning_rate": 1.3525361101780317e-05, "loss": 0.0283, "step": 38550 }, { "epoch": 3.24, "learning_rate": 1.3523681558616057e-05, "loss": 0.0806, "step": 38560 }, { "epoch": 3.24, "learning_rate": 1.3522002015451798e-05, "loss": 0.0547, "step": 38570 }, { "epoch": 3.24, "learning_rate": 1.352032247228754e-05, "loss": 0.1035, "step": 38580 }, { "epoch": 3.24, "learning_rate": 1.351864292912328e-05, "loss": 0.0249, "step": 38590 }, { "epoch": 3.24, "learning_rate": 1.3516963385959021e-05, "loss": 0.1577, "step": 38600 }, { "epoch": 3.24, "learning_rate": 1.3515283842794761e-05, "loss": 0.0543, "step": 38610 }, { "epoch": 3.24, "learning_rate": 1.35136042996305e-05, "loss": 0.1397, "step": 38620 }, { "epoch": 3.24, "learning_rate": 1.3511924756466242e-05, "loss": 0.0407, "step": 38630 }, { "epoch": 3.24, "learning_rate": 1.3510245213301983e-05, "loss": 0.0982, "step": 38640 }, { "epoch": 3.25, "learning_rate": 1.3508565670137723e-05, "loss": 0.1044, "step": 38650 }, { "epoch": 3.25, "learning_rate": 1.3506886126973464e-05, "loss": 0.0388, "step": 38660 }, { "epoch": 3.25, "learning_rate": 1.3505206583809204e-05, "loss": 0.1542, "step": 38670 }, { "epoch": 3.25, "learning_rate": 1.3503527040644946e-05, "loss": 0.1002, "step": 38680 }, { "epoch": 3.25, "learning_rate": 1.3501847497480687e-05, "loss": 0.0257, "step": 38690 }, { "epoch": 3.25, "learning_rate": 1.3500167954316427e-05, "loss": 0.1223, "step": 38700 }, { "epoch": 3.25, "learning_rate": 1.3498488411152168e-05, "loss": 0.071, "step": 38710 }, { "epoch": 3.25, "learning_rate": 1.349680886798791e-05, "loss": 0.0461, "step": 38720 }, { "epoch": 3.25, "learning_rate": 1.3495129324823649e-05, "loss": 0.0761, "step": 38730 }, { "epoch": 3.25, "learning_rate": 1.349344978165939e-05, "loss": 0.135, "step": 38740 }, { "epoch": 3.25, "learning_rate": 1.349177023849513e-05, "loss": 0.1399, "step": 38750 }, { "epoch": 3.25, "learning_rate": 1.349009069533087e-05, "loss": 0.1425, "step": 38760 }, { "epoch": 3.26, "learning_rate": 1.3488411152166612e-05, "loss": 0.0374, "step": 38770 }, { "epoch": 3.26, "learning_rate": 1.3486731609002353e-05, "loss": 0.0821, "step": 38780 }, { "epoch": 3.26, "learning_rate": 1.3485052065838093e-05, "loss": 0.0445, "step": 38790 }, { "epoch": 3.26, "learning_rate": 1.3483372522673834e-05, "loss": 0.0905, "step": 38800 }, { "epoch": 3.26, "learning_rate": 1.3481692979509576e-05, "loss": 0.053, "step": 38810 }, { "epoch": 3.26, "learning_rate": 1.3480013436345316e-05, "loss": 0.0978, "step": 38820 }, { "epoch": 3.26, "learning_rate": 1.3478333893181055e-05, "loss": 0.1022, "step": 38830 }, { "epoch": 3.26, "learning_rate": 1.3476654350016796e-05, "loss": 0.0642, "step": 38840 }, { "epoch": 3.26, "learning_rate": 1.3474974806852536e-05, "loss": 0.055, "step": 38850 }, { "epoch": 3.26, "learning_rate": 1.3473295263688277e-05, "loss": 0.0308, "step": 38860 }, { "epoch": 3.26, "learning_rate": 1.3471615720524019e-05, "loss": 0.1146, "step": 38870 }, { "epoch": 3.27, "learning_rate": 1.346993617735976e-05, "loss": 0.035, "step": 38880 }, { "epoch": 3.27, "learning_rate": 1.34682566341955e-05, "loss": 0.0789, "step": 38890 }, { "epoch": 3.27, "learning_rate": 1.346657709103124e-05, "loss": 0.1276, "step": 38900 }, { "epoch": 3.27, "learning_rate": 1.3464897547866982e-05, "loss": 0.1016, "step": 38910 }, { "epoch": 3.27, "learning_rate": 1.3463218004702723e-05, "loss": 0.0575, "step": 38920 }, { "epoch": 3.27, "learning_rate": 1.3461538461538463e-05, "loss": 0.0689, "step": 38930 }, { "epoch": 3.27, "learning_rate": 1.3459858918374202e-05, "loss": 0.1242, "step": 38940 }, { "epoch": 3.27, "learning_rate": 1.3458179375209943e-05, "loss": 0.0804, "step": 38950 }, { "epoch": 3.27, "learning_rate": 1.3456499832045685e-05, "loss": 0.071, "step": 38960 }, { "epoch": 3.27, "learning_rate": 1.3454820288881425e-05, "loss": 0.0953, "step": 38970 }, { "epoch": 3.27, "learning_rate": 1.3453140745717166e-05, "loss": 0.0734, "step": 38980 }, { "epoch": 3.27, "learning_rate": 1.3451461202552906e-05, "loss": 0.0886, "step": 38990 }, { "epoch": 3.28, "learning_rate": 1.3449781659388648e-05, "loss": 0.0211, "step": 39000 }, { "epoch": 3.28, "learning_rate": 1.3448102116224389e-05, "loss": 0.1326, "step": 39010 }, { "epoch": 3.28, "learning_rate": 1.344642257306013e-05, "loss": 0.1257, "step": 39020 }, { "epoch": 3.28, "learning_rate": 1.344474302989587e-05, "loss": 0.0618, "step": 39030 }, { "epoch": 3.28, "learning_rate": 1.3443063486731612e-05, "loss": 0.0672, "step": 39040 }, { "epoch": 3.28, "learning_rate": 1.3441383943567349e-05, "loss": 0.0927, "step": 39050 }, { "epoch": 3.28, "learning_rate": 1.3439704400403091e-05, "loss": 0.0834, "step": 39060 }, { "epoch": 3.28, "learning_rate": 1.3438024857238832e-05, "loss": 0.1136, "step": 39070 }, { "epoch": 3.28, "learning_rate": 1.3436345314074572e-05, "loss": 0.115, "step": 39080 }, { "epoch": 3.28, "learning_rate": 1.3434665770910313e-05, "loss": 0.1078, "step": 39090 }, { "epoch": 3.28, "learning_rate": 1.3432986227746055e-05, "loss": 0.0791, "step": 39100 }, { "epoch": 3.28, "learning_rate": 1.3431306684581795e-05, "loss": 0.0584, "step": 39110 }, { "epoch": 3.29, "learning_rate": 1.3429627141417536e-05, "loss": 0.0687, "step": 39120 }, { "epoch": 3.29, "learning_rate": 1.3427947598253276e-05, "loss": 0.1405, "step": 39130 }, { "epoch": 3.29, "learning_rate": 1.3426268055089018e-05, "loss": 0.0525, "step": 39140 }, { "epoch": 3.29, "learning_rate": 1.3424588511924757e-05, "loss": 0.0726, "step": 39150 }, { "epoch": 3.29, "learning_rate": 1.3422908968760498e-05, "loss": 0.1046, "step": 39160 }, { "epoch": 3.29, "learning_rate": 1.3421229425596238e-05, "loss": 0.0858, "step": 39170 }, { "epoch": 3.29, "learning_rate": 1.3419549882431979e-05, "loss": 0.052, "step": 39180 }, { "epoch": 3.29, "learning_rate": 1.341787033926772e-05, "loss": 0.0286, "step": 39190 }, { "epoch": 3.29, "learning_rate": 1.3416190796103461e-05, "loss": 0.0398, "step": 39200 }, { "epoch": 3.29, "learning_rate": 1.3414511252939202e-05, "loss": 0.0928, "step": 39210 }, { "epoch": 3.29, "learning_rate": 1.3412831709774942e-05, "loss": 0.1571, "step": 39220 }, { "epoch": 3.29, "learning_rate": 1.3411152166610684e-05, "loss": 0.0621, "step": 39230 }, { "epoch": 3.3, "learning_rate": 1.3409472623446425e-05, "loss": 0.0915, "step": 39240 }, { "epoch": 3.3, "learning_rate": 1.3407793080282165e-05, "loss": 0.0694, "step": 39250 }, { "epoch": 3.3, "learning_rate": 1.3406113537117904e-05, "loss": 0.0465, "step": 39260 }, { "epoch": 3.3, "learning_rate": 1.3404433993953645e-05, "loss": 0.0305, "step": 39270 }, { "epoch": 3.3, "learning_rate": 1.3402754450789385e-05, "loss": 0.1365, "step": 39280 }, { "epoch": 3.3, "learning_rate": 1.3401074907625127e-05, "loss": 0.1555, "step": 39290 }, { "epoch": 3.3, "learning_rate": 1.3399395364460868e-05, "loss": 0.0597, "step": 39300 }, { "epoch": 3.3, "learning_rate": 1.3397715821296608e-05, "loss": 0.0689, "step": 39310 }, { "epoch": 3.3, "learning_rate": 1.3396036278132349e-05, "loss": 0.0122, "step": 39320 }, { "epoch": 3.3, "learning_rate": 1.3394356734968091e-05, "loss": 0.1101, "step": 39330 }, { "epoch": 3.3, "learning_rate": 1.3392677191803831e-05, "loss": 0.0703, "step": 39340 }, { "epoch": 3.3, "learning_rate": 1.3390997648639572e-05, "loss": 0.0407, "step": 39350 }, { "epoch": 3.31, "learning_rate": 1.338931810547531e-05, "loss": 0.0835, "step": 39360 }, { "epoch": 3.31, "learning_rate": 1.3387638562311051e-05, "loss": 0.057, "step": 39370 }, { "epoch": 3.31, "learning_rate": 1.3385959019146793e-05, "loss": 0.0592, "step": 39380 }, { "epoch": 3.31, "learning_rate": 1.3384279475982534e-05, "loss": 0.0537, "step": 39390 }, { "epoch": 3.31, "learning_rate": 1.3382599932818274e-05, "loss": 0.1376, "step": 39400 }, { "epoch": 3.31, "learning_rate": 1.3380920389654015e-05, "loss": 0.0845, "step": 39410 }, { "epoch": 3.31, "learning_rate": 1.3379240846489757e-05, "loss": 0.0836, "step": 39420 }, { "epoch": 3.31, "learning_rate": 1.3377561303325497e-05, "loss": 0.0194, "step": 39430 }, { "epoch": 3.31, "learning_rate": 1.3375881760161238e-05, "loss": 0.1322, "step": 39440 }, { "epoch": 3.31, "learning_rate": 1.3374202216996978e-05, "loss": 0.1055, "step": 39450 }, { "epoch": 3.31, "learning_rate": 1.3372522673832719e-05, "loss": 0.0612, "step": 39460 }, { "epoch": 3.31, "learning_rate": 1.3370843130668458e-05, "loss": 0.106, "step": 39470 }, { "epoch": 3.32, "learning_rate": 1.33691635875042e-05, "loss": 0.0448, "step": 39480 }, { "epoch": 3.32, "learning_rate": 1.336748404433994e-05, "loss": 0.0913, "step": 39490 }, { "epoch": 3.32, "learning_rate": 1.336580450117568e-05, "loss": 0.0516, "step": 39500 }, { "epoch": 3.32, "learning_rate": 1.3364124958011421e-05, "loss": 0.0773, "step": 39510 }, { "epoch": 3.32, "learning_rate": 1.3362445414847163e-05, "loss": 0.1308, "step": 39520 }, { "epoch": 3.32, "learning_rate": 1.3360765871682904e-05, "loss": 0.0595, "step": 39530 }, { "epoch": 3.32, "learning_rate": 1.3359086328518644e-05, "loss": 0.0753, "step": 39540 }, { "epoch": 3.32, "learning_rate": 1.3357406785354385e-05, "loss": 0.0758, "step": 39550 }, { "epoch": 3.32, "learning_rate": 1.3355727242190127e-05, "loss": 0.1054, "step": 39560 }, { "epoch": 3.32, "learning_rate": 1.3354047699025866e-05, "loss": 0.1687, "step": 39570 }, { "epoch": 3.32, "learning_rate": 1.3352368155861606e-05, "loss": 0.0496, "step": 39580 }, { "epoch": 3.32, "learning_rate": 1.3350688612697347e-05, "loss": 0.1258, "step": 39590 }, { "epoch": 3.33, "learning_rate": 1.3349009069533087e-05, "loss": 0.2023, "step": 39600 }, { "epoch": 3.33, "learning_rate": 1.334732952636883e-05, "loss": 0.0346, "step": 39610 }, { "epoch": 3.33, "learning_rate": 1.334564998320457e-05, "loss": 0.0335, "step": 39620 }, { "epoch": 3.33, "learning_rate": 1.334397044004031e-05, "loss": 0.0389, "step": 39630 }, { "epoch": 3.33, "learning_rate": 1.334229089687605e-05, "loss": 0.0024, "step": 39640 }, { "epoch": 3.33, "learning_rate": 1.3340611353711791e-05, "loss": 0.1527, "step": 39650 }, { "epoch": 3.33, "learning_rate": 1.3338931810547533e-05, "loss": 0.0341, "step": 39660 }, { "epoch": 3.33, "learning_rate": 1.3337252267383274e-05, "loss": 0.0483, "step": 39670 }, { "epoch": 3.33, "learning_rate": 1.3335572724219013e-05, "loss": 0.0611, "step": 39680 }, { "epoch": 3.33, "learning_rate": 1.3333893181054753e-05, "loss": 0.2377, "step": 39690 }, { "epoch": 3.33, "learning_rate": 1.3332213637890494e-05, "loss": 0.1372, "step": 39700 }, { "epoch": 3.33, "learning_rate": 1.3330534094726236e-05, "loss": 0.0327, "step": 39710 }, { "epoch": 3.34, "learning_rate": 1.3328854551561976e-05, "loss": 0.0957, "step": 39720 }, { "epoch": 3.34, "learning_rate": 1.3327175008397717e-05, "loss": 0.1214, "step": 39730 }, { "epoch": 3.34, "learning_rate": 1.3325495465233457e-05, "loss": 0.0905, "step": 39740 }, { "epoch": 3.34, "learning_rate": 1.33238159220692e-05, "loss": 0.1613, "step": 39750 }, { "epoch": 3.34, "learning_rate": 1.332213637890494e-05, "loss": 0.0634, "step": 39760 }, { "epoch": 3.34, "learning_rate": 1.332045683574068e-05, "loss": 0.0744, "step": 39770 }, { "epoch": 3.34, "learning_rate": 1.331877729257642e-05, "loss": 0.0821, "step": 39780 }, { "epoch": 3.34, "learning_rate": 1.331709774941216e-05, "loss": 0.0813, "step": 39790 }, { "epoch": 3.34, "learning_rate": 1.3315418206247902e-05, "loss": 0.062, "step": 39800 }, { "epoch": 3.34, "learning_rate": 1.3313738663083642e-05, "loss": 0.06, "step": 39810 }, { "epoch": 3.34, "learning_rate": 1.3312059119919383e-05, "loss": 0.1392, "step": 39820 }, { "epoch": 3.34, "learning_rate": 1.3310379576755123e-05, "loss": 0.0212, "step": 39830 }, { "epoch": 3.35, "learning_rate": 1.3308700033590864e-05, "loss": 0.0894, "step": 39840 }, { "epoch": 3.35, "learning_rate": 1.3307020490426606e-05, "loss": 0.0495, "step": 39850 }, { "epoch": 3.35, "learning_rate": 1.3305340947262346e-05, "loss": 0.0725, "step": 39860 }, { "epoch": 3.35, "learning_rate": 1.3303661404098087e-05, "loss": 0.1291, "step": 39870 }, { "epoch": 3.35, "learning_rate": 1.3301981860933827e-05, "loss": 0.0032, "step": 39880 }, { "epoch": 3.35, "learning_rate": 1.3300302317769566e-05, "loss": 0.0886, "step": 39890 }, { "epoch": 3.35, "learning_rate": 1.3298622774605308e-05, "loss": 0.0918, "step": 39900 }, { "epoch": 3.35, "learning_rate": 1.3296943231441049e-05, "loss": 0.1116, "step": 39910 }, { "epoch": 3.35, "learning_rate": 1.329526368827679e-05, "loss": 0.0531, "step": 39920 }, { "epoch": 3.35, "learning_rate": 1.329358414511253e-05, "loss": 0.1385, "step": 39930 }, { "epoch": 3.35, "learning_rate": 1.3291904601948272e-05, "loss": 0.0741, "step": 39940 }, { "epoch": 3.35, "learning_rate": 1.3290225058784012e-05, "loss": 0.0489, "step": 39950 }, { "epoch": 3.36, "learning_rate": 1.3288545515619753e-05, "loss": 0.0727, "step": 39960 }, { "epoch": 3.36, "learning_rate": 1.3286865972455493e-05, "loss": 0.0054, "step": 39970 }, { "epoch": 3.36, "learning_rate": 1.3285186429291235e-05, "loss": 0.0538, "step": 39980 }, { "epoch": 3.36, "learning_rate": 1.3283506886126976e-05, "loss": 0.0763, "step": 39990 }, { "epoch": 3.36, "learning_rate": 1.3281827342962715e-05, "loss": 0.1087, "step": 40000 }, { "epoch": 3.36, "learning_rate": 1.3280147799798455e-05, "loss": 0.1578, "step": 40010 }, { "epoch": 3.36, "learning_rate": 1.3278468256634196e-05, "loss": 0.0937, "step": 40020 }, { "epoch": 3.36, "learning_rate": 1.3276788713469936e-05, "loss": 0.0269, "step": 40030 }, { "epoch": 3.36, "learning_rate": 1.3275109170305678e-05, "loss": 0.0485, "step": 40040 }, { "epoch": 3.36, "learning_rate": 1.3273429627141419e-05, "loss": 0.055, "step": 40050 }, { "epoch": 3.36, "learning_rate": 1.327175008397716e-05, "loss": 0.1327, "step": 40060 }, { "epoch": 3.36, "learning_rate": 1.32700705408129e-05, "loss": 0.0854, "step": 40070 }, { "epoch": 3.37, "learning_rate": 1.3268390997648642e-05, "loss": 0.0662, "step": 40080 }, { "epoch": 3.37, "learning_rate": 1.3266711454484382e-05, "loss": 0.0495, "step": 40090 }, { "epoch": 3.37, "learning_rate": 1.3265031911320121e-05, "loss": 0.0762, "step": 40100 }, { "epoch": 3.37, "learning_rate": 1.3263352368155862e-05, "loss": 0.0925, "step": 40110 }, { "epoch": 3.37, "learning_rate": 1.3261672824991602e-05, "loss": 0.0723, "step": 40120 }, { "epoch": 3.37, "learning_rate": 1.3259993281827344e-05, "loss": 0.1113, "step": 40130 }, { "epoch": 3.37, "learning_rate": 1.3258313738663085e-05, "loss": 0.154, "step": 40140 }, { "epoch": 3.37, "learning_rate": 1.3256634195498825e-05, "loss": 0.132, "step": 40150 }, { "epoch": 3.37, "learning_rate": 1.3254954652334566e-05, "loss": 0.0505, "step": 40160 }, { "epoch": 3.37, "learning_rate": 1.3253275109170308e-05, "loss": 0.037, "step": 40170 }, { "epoch": 3.37, "learning_rate": 1.3251595566006048e-05, "loss": 0.0876, "step": 40180 }, { "epoch": 3.38, "learning_rate": 1.3249916022841789e-05, "loss": 0.0724, "step": 40190 }, { "epoch": 3.38, "learning_rate": 1.324823647967753e-05, "loss": 0.06, "step": 40200 }, { "epoch": 3.38, "learning_rate": 1.3246556936513268e-05, "loss": 0.0575, "step": 40210 }, { "epoch": 3.38, "learning_rate": 1.3244877393349009e-05, "loss": 0.1091, "step": 40220 }, { "epoch": 3.38, "learning_rate": 1.324319785018475e-05, "loss": 0.0093, "step": 40230 }, { "epoch": 3.38, "learning_rate": 1.3241518307020491e-05, "loss": 0.089, "step": 40240 }, { "epoch": 3.38, "learning_rate": 1.3239838763856232e-05, "loss": 0.0128, "step": 40250 }, { "epoch": 3.38, "learning_rate": 1.3238159220691972e-05, "loss": 0.1185, "step": 40260 }, { "epoch": 3.38, "learning_rate": 1.3236479677527714e-05, "loss": 0.1748, "step": 40270 }, { "epoch": 3.38, "learning_rate": 1.3234800134363455e-05, "loss": 0.111, "step": 40280 }, { "epoch": 3.38, "learning_rate": 1.3233120591199195e-05, "loss": 0.1025, "step": 40290 }, { "epoch": 3.38, "learning_rate": 1.3231441048034936e-05, "loss": 0.0621, "step": 40300 }, { "epoch": 3.39, "learning_rate": 1.3229761504870678e-05, "loss": 0.0607, "step": 40310 }, { "epoch": 3.39, "learning_rate": 1.3228081961706417e-05, "loss": 0.0828, "step": 40320 }, { "epoch": 3.39, "learning_rate": 1.3226402418542157e-05, "loss": 0.124, "step": 40330 }, { "epoch": 3.39, "learning_rate": 1.3224722875377898e-05, "loss": 0.0511, "step": 40340 }, { "epoch": 3.39, "learning_rate": 1.3223043332213638e-05, "loss": 0.1036, "step": 40350 }, { "epoch": 3.39, "learning_rate": 1.322136378904938e-05, "loss": 0.0349, "step": 40360 }, { "epoch": 3.39, "learning_rate": 1.321968424588512e-05, "loss": 0.0821, "step": 40370 }, { "epoch": 3.39, "learning_rate": 1.3218004702720861e-05, "loss": 0.0475, "step": 40380 }, { "epoch": 3.39, "learning_rate": 1.3216325159556602e-05, "loss": 0.0755, "step": 40390 }, { "epoch": 3.39, "learning_rate": 1.3214645616392344e-05, "loss": 0.063, "step": 40400 }, { "epoch": 3.39, "learning_rate": 1.3212966073228084e-05, "loss": 0.0984, "step": 40410 }, { "epoch": 3.39, "learning_rate": 1.3211286530063823e-05, "loss": 0.0411, "step": 40420 }, { "epoch": 3.4, "learning_rate": 1.3209606986899564e-05, "loss": 0.0981, "step": 40430 }, { "epoch": 3.4, "learning_rate": 1.3207927443735304e-05, "loss": 0.0964, "step": 40440 }, { "epoch": 3.4, "learning_rate": 1.3206247900571045e-05, "loss": 0.0783, "step": 40450 }, { "epoch": 3.4, "learning_rate": 1.3204568357406787e-05, "loss": 0.0873, "step": 40460 }, { "epoch": 3.4, "learning_rate": 1.3202888814242527e-05, "loss": 0.0574, "step": 40470 }, { "epoch": 3.4, "learning_rate": 1.3201209271078268e-05, "loss": 0.0379, "step": 40480 }, { "epoch": 3.4, "learning_rate": 1.3199529727914008e-05, "loss": 0.1, "step": 40490 }, { "epoch": 3.4, "learning_rate": 1.319785018474975e-05, "loss": 0.166, "step": 40500 }, { "epoch": 3.4, "learning_rate": 1.3196170641585491e-05, "loss": 0.0738, "step": 40510 }, { "epoch": 3.4, "learning_rate": 1.3194491098421231e-05, "loss": 0.1993, "step": 40520 }, { "epoch": 3.4, "learning_rate": 1.319281155525697e-05, "loss": 0.1209, "step": 40530 }, { "epoch": 3.4, "learning_rate": 1.319113201209271e-05, "loss": 0.0595, "step": 40540 }, { "epoch": 3.41, "learning_rate": 1.3189452468928453e-05, "loss": 0.1003, "step": 40550 }, { "epoch": 3.41, "learning_rate": 1.3187772925764193e-05, "loss": 0.0459, "step": 40560 }, { "epoch": 3.41, "learning_rate": 1.3186093382599934e-05, "loss": 0.0281, "step": 40570 }, { "epoch": 3.41, "learning_rate": 1.3184413839435674e-05, "loss": 0.1167, "step": 40580 }, { "epoch": 3.41, "learning_rate": 1.3182734296271416e-05, "loss": 0.1019, "step": 40590 }, { "epoch": 3.41, "learning_rate": 1.3181054753107157e-05, "loss": 0.0464, "step": 40600 }, { "epoch": 3.41, "learning_rate": 1.3179375209942897e-05, "loss": 0.0074, "step": 40610 }, { "epoch": 3.41, "learning_rate": 1.3177695666778638e-05, "loss": 0.0953, "step": 40620 }, { "epoch": 3.41, "learning_rate": 1.3176016123614377e-05, "loss": 0.0595, "step": 40630 }, { "epoch": 3.41, "learning_rate": 1.3174336580450117e-05, "loss": 0.067, "step": 40640 }, { "epoch": 3.41, "learning_rate": 1.317265703728586e-05, "loss": 0.0637, "step": 40650 }, { "epoch": 3.41, "learning_rate": 1.31709774941216e-05, "loss": 0.1543, "step": 40660 }, { "epoch": 3.42, "learning_rate": 1.316929795095734e-05, "loss": 0.0548, "step": 40670 }, { "epoch": 3.42, "learning_rate": 1.316761840779308e-05, "loss": 0.1775, "step": 40680 }, { "epoch": 3.42, "learning_rate": 1.3165938864628823e-05, "loss": 0.1506, "step": 40690 }, { "epoch": 3.42, "learning_rate": 1.3164259321464563e-05, "loss": 0.066, "step": 40700 }, { "epoch": 3.42, "learning_rate": 1.3162579778300304e-05, "loss": 0.0933, "step": 40710 }, { "epoch": 3.42, "learning_rate": 1.3160900235136044e-05, "loss": 0.0227, "step": 40720 }, { "epoch": 3.42, "learning_rate": 1.3159220691971786e-05, "loss": 0.1591, "step": 40730 }, { "epoch": 3.42, "learning_rate": 1.3157541148807525e-05, "loss": 0.0949, "step": 40740 }, { "epoch": 3.42, "learning_rate": 1.3155861605643266e-05, "loss": 0.0994, "step": 40750 }, { "epoch": 3.42, "learning_rate": 1.3154182062479006e-05, "loss": 0.1543, "step": 40760 }, { "epoch": 3.42, "learning_rate": 1.3152502519314747e-05, "loss": 0.0631, "step": 40770 }, { "epoch": 3.42, "learning_rate": 1.3150822976150489e-05, "loss": 0.0401, "step": 40780 }, { "epoch": 3.43, "learning_rate": 1.314914343298623e-05, "loss": 0.0753, "step": 40790 }, { "epoch": 3.43, "learning_rate": 1.314746388982197e-05, "loss": 0.0782, "step": 40800 }, { "epoch": 3.43, "learning_rate": 1.314578434665771e-05, "loss": 0.0765, "step": 40810 }, { "epoch": 3.43, "learning_rate": 1.314410480349345e-05, "loss": 0.2007, "step": 40820 }, { "epoch": 3.43, "learning_rate": 1.3142425260329193e-05, "loss": 0.0027, "step": 40830 }, { "epoch": 3.43, "learning_rate": 1.3140745717164933e-05, "loss": 0.1116, "step": 40840 }, { "epoch": 3.43, "learning_rate": 1.3139066174000672e-05, "loss": 0.0207, "step": 40850 }, { "epoch": 3.43, "learning_rate": 1.3137386630836413e-05, "loss": 0.0088, "step": 40860 }, { "epoch": 3.43, "learning_rate": 1.3135707087672153e-05, "loss": 0.0858, "step": 40870 }, { "epoch": 3.43, "learning_rate": 1.3134027544507895e-05, "loss": 0.0762, "step": 40880 }, { "epoch": 3.43, "learning_rate": 1.3132348001343636e-05, "loss": 0.0362, "step": 40890 }, { "epoch": 3.43, "learning_rate": 1.3130668458179376e-05, "loss": 0.1677, "step": 40900 }, { "epoch": 3.44, "learning_rate": 1.3128988915015117e-05, "loss": 0.1234, "step": 40910 }, { "epoch": 3.44, "learning_rate": 1.3127309371850859e-05, "loss": 0.116, "step": 40920 }, { "epoch": 3.44, "learning_rate": 1.31256298286866e-05, "loss": 0.0598, "step": 40930 }, { "epoch": 3.44, "learning_rate": 1.312395028552234e-05, "loss": 0.023, "step": 40940 }, { "epoch": 3.44, "learning_rate": 1.3122270742358079e-05, "loss": 0.0644, "step": 40950 }, { "epoch": 3.44, "learning_rate": 1.3120591199193819e-05, "loss": 0.0042, "step": 40960 }, { "epoch": 3.44, "learning_rate": 1.311891165602956e-05, "loss": 0.0077, "step": 40970 }, { "epoch": 3.44, "learning_rate": 1.3117232112865302e-05, "loss": 0.0698, "step": 40980 }, { "epoch": 3.44, "learning_rate": 1.3115552569701042e-05, "loss": 0.1197, "step": 40990 }, { "epoch": 3.44, "learning_rate": 1.3113873026536783e-05, "loss": 0.0863, "step": 41000 }, { "epoch": 3.44, "learning_rate": 1.3112193483372523e-05, "loss": 0.1187, "step": 41010 }, { "epoch": 3.44, "learning_rate": 1.3110513940208265e-05, "loss": 0.0686, "step": 41020 }, { "epoch": 3.45, "learning_rate": 1.3108834397044006e-05, "loss": 0.0554, "step": 41030 }, { "epoch": 3.45, "learning_rate": 1.3107154853879746e-05, "loss": 0.0698, "step": 41040 }, { "epoch": 3.45, "learning_rate": 1.3105475310715487e-05, "loss": 0.0667, "step": 41050 }, { "epoch": 3.45, "learning_rate": 1.3103795767551226e-05, "loss": 0.1298, "step": 41060 }, { "epoch": 3.45, "learning_rate": 1.3102116224386968e-05, "loss": 0.2065, "step": 41070 }, { "epoch": 3.45, "learning_rate": 1.3100436681222708e-05, "loss": 0.035, "step": 41080 }, { "epoch": 3.45, "learning_rate": 1.3098757138058449e-05, "loss": 0.0637, "step": 41090 }, { "epoch": 3.45, "learning_rate": 1.3097077594894189e-05, "loss": 0.0951, "step": 41100 }, { "epoch": 3.45, "learning_rate": 1.3095398051729931e-05, "loss": 0.1335, "step": 41110 }, { "epoch": 3.45, "learning_rate": 1.3093718508565672e-05, "loss": 0.0487, "step": 41120 }, { "epoch": 3.45, "learning_rate": 1.3092038965401412e-05, "loss": 0.1008, "step": 41130 }, { "epoch": 3.45, "learning_rate": 1.3090359422237153e-05, "loss": 0.038, "step": 41140 }, { "epoch": 3.46, "learning_rate": 1.3088679879072895e-05, "loss": 0.1335, "step": 41150 }, { "epoch": 3.46, "learning_rate": 1.3087000335908632e-05, "loss": 0.0948, "step": 41160 }, { "epoch": 3.46, "learning_rate": 1.3085320792744374e-05, "loss": 0.0497, "step": 41170 }, { "epoch": 3.46, "learning_rate": 1.3083641249580115e-05, "loss": 0.0693, "step": 41180 }, { "epoch": 3.46, "learning_rate": 1.3081961706415855e-05, "loss": 0.0453, "step": 41190 }, { "epoch": 3.46, "learning_rate": 1.3080282163251596e-05, "loss": 0.0106, "step": 41200 }, { "epoch": 3.46, "learning_rate": 1.3078602620087338e-05, "loss": 0.0522, "step": 41210 }, { "epoch": 3.46, "learning_rate": 1.3076923076923078e-05, "loss": 0.0829, "step": 41220 }, { "epoch": 3.46, "learning_rate": 1.3075243533758819e-05, "loss": 0.1096, "step": 41230 }, { "epoch": 3.46, "learning_rate": 1.307356399059456e-05, "loss": 0.0203, "step": 41240 }, { "epoch": 3.46, "learning_rate": 1.3071884447430301e-05, "loss": 0.0535, "step": 41250 }, { "epoch": 3.46, "learning_rate": 1.3070204904266042e-05, "loss": 0.1165, "step": 41260 }, { "epoch": 3.47, "learning_rate": 1.306852536110178e-05, "loss": 0.1305, "step": 41270 }, { "epoch": 3.47, "learning_rate": 1.3066845817937521e-05, "loss": 0.111, "step": 41280 }, { "epoch": 3.47, "learning_rate": 1.3065166274773262e-05, "loss": 0.1131, "step": 41290 }, { "epoch": 3.47, "learning_rate": 1.3063486731609004e-05, "loss": 0.0787, "step": 41300 }, { "epoch": 3.47, "learning_rate": 1.3061807188444744e-05, "loss": 0.0851, "step": 41310 }, { "epoch": 3.47, "learning_rate": 1.3060127645280485e-05, "loss": 0.0364, "step": 41320 }, { "epoch": 3.47, "learning_rate": 1.3058448102116225e-05, "loss": 0.1306, "step": 41330 }, { "epoch": 3.47, "learning_rate": 1.3056768558951967e-05, "loss": 0.0399, "step": 41340 }, { "epoch": 3.47, "learning_rate": 1.3055089015787708e-05, "loss": 0.0473, "step": 41350 }, { "epoch": 3.47, "learning_rate": 1.3053409472623448e-05, "loss": 0.0405, "step": 41360 }, { "epoch": 3.47, "learning_rate": 1.3051729929459189e-05, "loss": 0.1345, "step": 41370 }, { "epoch": 3.47, "learning_rate": 1.3050050386294928e-05, "loss": 0.0366, "step": 41380 }, { "epoch": 3.48, "learning_rate": 1.3048370843130668e-05, "loss": 0.0321, "step": 41390 }, { "epoch": 3.48, "learning_rate": 1.304669129996641e-05, "loss": 0.0609, "step": 41400 }, { "epoch": 3.48, "learning_rate": 1.304501175680215e-05, "loss": 0.1207, "step": 41410 }, { "epoch": 3.48, "learning_rate": 1.3043332213637891e-05, "loss": 0.0823, "step": 41420 }, { "epoch": 3.48, "learning_rate": 1.3041652670473632e-05, "loss": 0.1629, "step": 41430 }, { "epoch": 3.48, "learning_rate": 1.3039973127309374e-05, "loss": 0.0359, "step": 41440 }, { "epoch": 3.48, "learning_rate": 1.3038293584145114e-05, "loss": 0.1391, "step": 41450 }, { "epoch": 3.48, "learning_rate": 1.3036614040980855e-05, "loss": 0.1332, "step": 41460 }, { "epoch": 3.48, "learning_rate": 1.3034934497816595e-05, "loss": 0.0353, "step": 41470 }, { "epoch": 3.48, "learning_rate": 1.3033254954652334e-05, "loss": 0.0798, "step": 41480 }, { "epoch": 3.48, "learning_rate": 1.3031575411488076e-05, "loss": 0.1004, "step": 41490 }, { "epoch": 3.49, "learning_rate": 1.3029895868323817e-05, "loss": 0.0509, "step": 41500 }, { "epoch": 3.49, "learning_rate": 1.3028216325159557e-05, "loss": 0.0819, "step": 41510 }, { "epoch": 3.49, "learning_rate": 1.3026536781995298e-05, "loss": 0.0567, "step": 41520 }, { "epoch": 3.49, "learning_rate": 1.302485723883104e-05, "loss": 0.0093, "step": 41530 }, { "epoch": 3.49, "learning_rate": 1.302317769566678e-05, "loss": 0.1061, "step": 41540 }, { "epoch": 3.49, "learning_rate": 1.302149815250252e-05, "loss": 0.0231, "step": 41550 }, { "epoch": 3.49, "learning_rate": 1.3019818609338261e-05, "loss": 0.1178, "step": 41560 }, { "epoch": 3.49, "learning_rate": 1.3018139066174003e-05, "loss": 0.0335, "step": 41570 }, { "epoch": 3.49, "learning_rate": 1.3016459523009744e-05, "loss": 0.0751, "step": 41580 }, { "epoch": 3.49, "learning_rate": 1.3014779979845483e-05, "loss": 0.049, "step": 41590 }, { "epoch": 3.49, "learning_rate": 1.3013100436681223e-05, "loss": 0.0522, "step": 41600 }, { "epoch": 3.49, "learning_rate": 1.3011420893516964e-05, "loss": 0.227, "step": 41610 }, { "epoch": 3.5, "learning_rate": 1.3009741350352704e-05, "loss": 0.0748, "step": 41620 }, { "epoch": 3.5, "learning_rate": 1.3008061807188446e-05, "loss": 0.0356, "step": 41630 }, { "epoch": 3.5, "learning_rate": 1.3006382264024187e-05, "loss": 0.0458, "step": 41640 }, { "epoch": 3.5, "learning_rate": 1.3004702720859927e-05, "loss": 0.1322, "step": 41650 }, { "epoch": 3.5, "learning_rate": 1.3003023177695668e-05, "loss": 0.117, "step": 41660 }, { "epoch": 3.5, "learning_rate": 1.300134363453141e-05, "loss": 0.0891, "step": 41670 }, { "epoch": 3.5, "learning_rate": 1.299966409136715e-05, "loss": 0.0753, "step": 41680 }, { "epoch": 3.5, "learning_rate": 1.2997984548202889e-05, "loss": 0.0441, "step": 41690 }, { "epoch": 3.5, "learning_rate": 1.299630500503863e-05, "loss": 0.0723, "step": 41700 }, { "epoch": 3.5, "learning_rate": 1.299462546187437e-05, "loss": 0.1245, "step": 41710 }, { "epoch": 3.5, "learning_rate": 1.2992945918710112e-05, "loss": 0.0599, "step": 41720 }, { "epoch": 3.5, "learning_rate": 1.2991266375545853e-05, "loss": 0.0516, "step": 41730 }, { "epoch": 3.51, "learning_rate": 1.2989586832381593e-05, "loss": 0.1295, "step": 41740 }, { "epoch": 3.51, "learning_rate": 1.2987907289217334e-05, "loss": 0.0283, "step": 41750 }, { "epoch": 3.51, "learning_rate": 1.2986227746053076e-05, "loss": 0.0517, "step": 41760 }, { "epoch": 3.51, "learning_rate": 1.2984548202888816e-05, "loss": 0.0583, "step": 41770 }, { "epoch": 3.51, "learning_rate": 1.2982868659724557e-05, "loss": 0.0715, "step": 41780 }, { "epoch": 3.51, "learning_rate": 1.2981189116560297e-05, "loss": 0.0019, "step": 41790 }, { "epoch": 3.51, "learning_rate": 1.2979509573396036e-05, "loss": 0.0097, "step": 41800 }, { "epoch": 3.51, "learning_rate": 1.2977830030231777e-05, "loss": 0.0445, "step": 41810 }, { "epoch": 3.51, "learning_rate": 1.2976150487067519e-05, "loss": 0.1048, "step": 41820 }, { "epoch": 3.51, "learning_rate": 1.297447094390326e-05, "loss": 0.0752, "step": 41830 }, { "epoch": 3.51, "learning_rate": 1.2972791400739e-05, "loss": 0.0439, "step": 41840 }, { "epoch": 3.51, "learning_rate": 1.297111185757474e-05, "loss": 0.0727, "step": 41850 }, { "epoch": 3.52, "learning_rate": 1.2969432314410482e-05, "loss": 0.0707, "step": 41860 }, { "epoch": 3.52, "learning_rate": 1.2967752771246223e-05, "loss": 0.11, "step": 41870 }, { "epoch": 3.52, "learning_rate": 1.2966073228081963e-05, "loss": 0.0427, "step": 41880 }, { "epoch": 3.52, "learning_rate": 1.2964393684917704e-05, "loss": 0.0456, "step": 41890 }, { "epoch": 3.52, "learning_rate": 1.2962714141753446e-05, "loss": 0.1041, "step": 41900 }, { "epoch": 3.52, "learning_rate": 1.2961034598589185e-05, "loss": 0.1705, "step": 41910 }, { "epoch": 3.52, "learning_rate": 1.2959355055424925e-05, "loss": 0.2036, "step": 41920 }, { "epoch": 3.52, "learning_rate": 1.2957675512260666e-05, "loss": 0.0755, "step": 41930 }, { "epoch": 3.52, "learning_rate": 1.2955995969096406e-05, "loss": 0.0575, "step": 41940 }, { "epoch": 3.52, "learning_rate": 1.2954316425932147e-05, "loss": 0.0914, "step": 41950 }, { "epoch": 3.52, "learning_rate": 1.2952636882767889e-05, "loss": 0.1909, "step": 41960 }, { "epoch": 3.52, "learning_rate": 1.295095733960363e-05, "loss": 0.0624, "step": 41970 }, { "epoch": 3.53, "learning_rate": 1.294927779643937e-05, "loss": 0.0441, "step": 41980 }, { "epoch": 3.53, "learning_rate": 1.294759825327511e-05, "loss": 0.1436, "step": 41990 }, { "epoch": 3.53, "learning_rate": 1.2945918710110852e-05, "loss": 0.1173, "step": 42000 }, { "epoch": 3.53, "learning_rate": 1.2944239166946591e-05, "loss": 0.0242, "step": 42010 }, { "epoch": 3.53, "learning_rate": 1.2942559623782332e-05, "loss": 0.0246, "step": 42020 }, { "epoch": 3.53, "learning_rate": 1.2940880080618072e-05, "loss": 0.0748, "step": 42030 }, { "epoch": 3.53, "learning_rate": 1.2939200537453813e-05, "loss": 0.1156, "step": 42040 }, { "epoch": 3.53, "learning_rate": 1.2937520994289555e-05, "loss": 0.0746, "step": 42050 }, { "epoch": 3.53, "learning_rate": 1.2935841451125295e-05, "loss": 0.0555, "step": 42060 }, { "epoch": 3.53, "learning_rate": 1.2934161907961036e-05, "loss": 0.0471, "step": 42070 }, { "epoch": 3.53, "learning_rate": 1.2932482364796776e-05, "loss": 0.0835, "step": 42080 }, { "epoch": 3.53, "learning_rate": 1.2930802821632518e-05, "loss": 0.0519, "step": 42090 }, { "epoch": 3.54, "learning_rate": 1.2929123278468259e-05, "loss": 0.1394, "step": 42100 }, { "epoch": 3.54, "learning_rate": 1.2927443735304e-05, "loss": 0.0915, "step": 42110 }, { "epoch": 3.54, "learning_rate": 1.2925764192139738e-05, "loss": 0.0394, "step": 42120 }, { "epoch": 3.54, "learning_rate": 1.2924084648975479e-05, "loss": 0.0733, "step": 42130 }, { "epoch": 3.54, "learning_rate": 1.2922405105811219e-05, "loss": 0.004, "step": 42140 }, { "epoch": 3.54, "learning_rate": 1.2920725562646961e-05, "loss": 0.0977, "step": 42150 }, { "epoch": 3.54, "learning_rate": 1.2919046019482702e-05, "loss": 0.0361, "step": 42160 }, { "epoch": 3.54, "learning_rate": 1.2917366476318442e-05, "loss": 0.1324, "step": 42170 }, { "epoch": 3.54, "learning_rate": 1.2915686933154183e-05, "loss": 0.0964, "step": 42180 }, { "epoch": 3.54, "learning_rate": 1.2914007389989925e-05, "loss": 0.1171, "step": 42190 }, { "epoch": 3.54, "learning_rate": 1.2912327846825665e-05, "loss": 0.0237, "step": 42200 }, { "epoch": 3.54, "learning_rate": 1.2910648303661406e-05, "loss": 0.0504, "step": 42210 }, { "epoch": 3.55, "learning_rate": 1.2908968760497145e-05, "loss": 0.0047, "step": 42220 }, { "epoch": 3.55, "learning_rate": 1.2907289217332885e-05, "loss": 0.002, "step": 42230 }, { "epoch": 3.55, "learning_rate": 1.2905609674168627e-05, "loss": 0.0467, "step": 42240 }, { "epoch": 3.55, "learning_rate": 1.2903930131004368e-05, "loss": 0.1177, "step": 42250 }, { "epoch": 3.55, "learning_rate": 1.2902250587840108e-05, "loss": 0.1492, "step": 42260 }, { "epoch": 3.55, "learning_rate": 1.2900571044675849e-05, "loss": 0.1023, "step": 42270 }, { "epoch": 3.55, "learning_rate": 1.289889150151159e-05, "loss": 0.1843, "step": 42280 }, { "epoch": 3.55, "learning_rate": 1.2897211958347331e-05, "loss": 0.0636, "step": 42290 }, { "epoch": 3.55, "learning_rate": 1.2895532415183072e-05, "loss": 0.1871, "step": 42300 }, { "epoch": 3.55, "learning_rate": 1.2893852872018812e-05, "loss": 0.1521, "step": 42310 }, { "epoch": 3.55, "learning_rate": 1.2892173328854554e-05, "loss": 0.0851, "step": 42320 }, { "epoch": 3.55, "learning_rate": 1.2890493785690291e-05, "loss": 0.1022, "step": 42330 }, { "epoch": 3.56, "learning_rate": 1.2888814242526034e-05, "loss": 0.0698, "step": 42340 }, { "epoch": 3.56, "learning_rate": 1.2887134699361774e-05, "loss": 0.1032, "step": 42350 }, { "epoch": 3.56, "learning_rate": 1.2885455156197515e-05, "loss": 0.1291, "step": 42360 }, { "epoch": 3.56, "learning_rate": 1.2883775613033255e-05, "loss": 0.0205, "step": 42370 }, { "epoch": 3.56, "learning_rate": 1.2882096069868997e-05, "loss": 0.0794, "step": 42380 }, { "epoch": 3.56, "learning_rate": 1.2880416526704738e-05, "loss": 0.0426, "step": 42390 }, { "epoch": 3.56, "learning_rate": 1.2878736983540478e-05, "loss": 0.0402, "step": 42400 }, { "epoch": 3.56, "learning_rate": 1.2877057440376219e-05, "loss": 0.0666, "step": 42410 }, { "epoch": 3.56, "learning_rate": 1.2875377897211961e-05, "loss": 0.0964, "step": 42420 }, { "epoch": 3.56, "learning_rate": 1.2873698354047701e-05, "loss": 0.0621, "step": 42430 }, { "epoch": 3.56, "learning_rate": 1.287201881088344e-05, "loss": 0.0736, "step": 42440 }, { "epoch": 3.56, "learning_rate": 1.287033926771918e-05, "loss": 0.1248, "step": 42450 }, { "epoch": 3.57, "learning_rate": 1.2868659724554921e-05, "loss": 0.1331, "step": 42460 }, { "epoch": 3.57, "learning_rate": 1.2866980181390663e-05, "loss": 0.0038, "step": 42470 }, { "epoch": 3.57, "learning_rate": 1.2865300638226404e-05, "loss": 0.0391, "step": 42480 }, { "epoch": 3.57, "learning_rate": 1.2863621095062144e-05, "loss": 0.0808, "step": 42490 }, { "epoch": 3.57, "learning_rate": 1.2861941551897885e-05, "loss": 0.0723, "step": 42500 }, { "epoch": 3.57, "learning_rate": 1.2860262008733627e-05, "loss": 0.0172, "step": 42510 }, { "epoch": 3.57, "learning_rate": 1.2858582465569367e-05, "loss": 0.0714, "step": 42520 }, { "epoch": 3.57, "learning_rate": 1.2856902922405108e-05, "loss": 0.0986, "step": 42530 }, { "epoch": 3.57, "learning_rate": 1.2855223379240847e-05, "loss": 0.0664, "step": 42540 }, { "epoch": 3.57, "learning_rate": 1.2853543836076587e-05, "loss": 0.0693, "step": 42550 }, { "epoch": 3.57, "learning_rate": 1.2851864292912328e-05, "loss": 0.0715, "step": 42560 }, { "epoch": 3.57, "learning_rate": 1.285018474974807e-05, "loss": 0.0679, "step": 42570 }, { "epoch": 3.58, "learning_rate": 1.284850520658381e-05, "loss": 0.0957, "step": 42580 }, { "epoch": 3.58, "learning_rate": 1.284682566341955e-05, "loss": 0.1408, "step": 42590 }, { "epoch": 3.58, "learning_rate": 1.2845146120255291e-05, "loss": 0.0738, "step": 42600 }, { "epoch": 3.58, "learning_rate": 1.2843466577091033e-05, "loss": 0.0943, "step": 42610 }, { "epoch": 3.58, "learning_rate": 1.2841787033926774e-05, "loss": 0.1009, "step": 42620 }, { "epoch": 3.58, "learning_rate": 1.2840107490762514e-05, "loss": 0.1279, "step": 42630 }, { "epoch": 3.58, "learning_rate": 1.2838427947598255e-05, "loss": 0.0644, "step": 42640 }, { "epoch": 3.58, "learning_rate": 1.2836748404433994e-05, "loss": 0.0467, "step": 42650 }, { "epoch": 3.58, "learning_rate": 1.2835068861269736e-05, "loss": 0.0845, "step": 42660 }, { "epoch": 3.58, "learning_rate": 1.2833389318105476e-05, "loss": 0.0951, "step": 42670 }, { "epoch": 3.58, "learning_rate": 1.2831709774941217e-05, "loss": 0.0238, "step": 42680 }, { "epoch": 3.58, "learning_rate": 1.2830030231776957e-05, "loss": 0.0701, "step": 42690 }, { "epoch": 3.59, "learning_rate": 1.28283506886127e-05, "loss": 0.0343, "step": 42700 }, { "epoch": 3.59, "learning_rate": 1.282667114544844e-05, "loss": 0.1094, "step": 42710 }, { "epoch": 3.59, "learning_rate": 1.282499160228418e-05, "loss": 0.1395, "step": 42720 }, { "epoch": 3.59, "learning_rate": 1.282331205911992e-05, "loss": 0.0594, "step": 42730 }, { "epoch": 3.59, "learning_rate": 1.2821632515955663e-05, "loss": 0.0457, "step": 42740 }, { "epoch": 3.59, "learning_rate": 1.28199529727914e-05, "loss": 0.0866, "step": 42750 }, { "epoch": 3.59, "learning_rate": 1.2818273429627142e-05, "loss": 0.0975, "step": 42760 }, { "epoch": 3.59, "learning_rate": 1.2816593886462883e-05, "loss": 0.0802, "step": 42770 }, { "epoch": 3.59, "learning_rate": 1.2814914343298623e-05, "loss": 0.0544, "step": 42780 }, { "epoch": 3.59, "learning_rate": 1.2813234800134364e-05, "loss": 0.021, "step": 42790 }, { "epoch": 3.59, "learning_rate": 1.2811555256970106e-05, "loss": 0.1058, "step": 42800 }, { "epoch": 3.6, "learning_rate": 1.2809875713805846e-05, "loss": 0.1025, "step": 42810 }, { "epoch": 3.6, "learning_rate": 1.2808196170641587e-05, "loss": 0.1012, "step": 42820 }, { "epoch": 3.6, "learning_rate": 1.2806516627477327e-05, "loss": 0.0733, "step": 42830 }, { "epoch": 3.6, "learning_rate": 1.280483708431307e-05, "loss": 0.0461, "step": 42840 }, { "epoch": 3.6, "learning_rate": 1.280315754114881e-05, "loss": 0.0221, "step": 42850 }, { "epoch": 3.6, "learning_rate": 1.2801477997984549e-05, "loss": 0.0461, "step": 42860 }, { "epoch": 3.6, "learning_rate": 1.2799798454820289e-05, "loss": 0.0763, "step": 42870 }, { "epoch": 3.6, "learning_rate": 1.279811891165603e-05, "loss": 0.0663, "step": 42880 }, { "epoch": 3.6, "learning_rate": 1.2796439368491772e-05, "loss": 0.1123, "step": 42890 }, { "epoch": 3.6, "learning_rate": 1.2794759825327512e-05, "loss": 0.0768, "step": 42900 }, { "epoch": 3.6, "learning_rate": 1.2793080282163253e-05, "loss": 0.117, "step": 42910 }, { "epoch": 3.6, "learning_rate": 1.2791400738998993e-05, "loss": 0.0407, "step": 42920 }, { "epoch": 3.61, "learning_rate": 1.2789721195834734e-05, "loss": 0.0662, "step": 42930 }, { "epoch": 3.61, "learning_rate": 1.2788041652670476e-05, "loss": 0.0639, "step": 42940 }, { "epoch": 3.61, "learning_rate": 1.2786362109506216e-05, "loss": 0.0889, "step": 42950 }, { "epoch": 3.61, "learning_rate": 1.2784682566341955e-05, "loss": 0.0301, "step": 42960 }, { "epoch": 3.61, "learning_rate": 1.2783003023177696e-05, "loss": 0.0743, "step": 42970 }, { "epoch": 3.61, "learning_rate": 1.2781323480013436e-05, "loss": 0.1122, "step": 42980 }, { "epoch": 3.61, "learning_rate": 1.2779643936849178e-05, "loss": 0.0777, "step": 42990 }, { "epoch": 3.61, "learning_rate": 1.2777964393684919e-05, "loss": 0.0332, "step": 43000 }, { "epoch": 3.61, "learning_rate": 1.2776284850520659e-05, "loss": 0.0844, "step": 43010 }, { "epoch": 3.61, "learning_rate": 1.27746053073564e-05, "loss": 0.1078, "step": 43020 }, { "epoch": 3.61, "learning_rate": 1.2772925764192142e-05, "loss": 0.0329, "step": 43030 }, { "epoch": 3.61, "learning_rate": 1.2771246221027882e-05, "loss": 0.086, "step": 43040 }, { "epoch": 3.62, "learning_rate": 1.2769566677863623e-05, "loss": 0.127, "step": 43050 }, { "epoch": 3.62, "learning_rate": 1.2767887134699363e-05, "loss": 0.0839, "step": 43060 }, { "epoch": 3.62, "learning_rate": 1.2766207591535102e-05, "loss": 0.041, "step": 43070 }, { "epoch": 3.62, "learning_rate": 1.2764528048370844e-05, "loss": 0.0774, "step": 43080 }, { "epoch": 3.62, "learning_rate": 1.2762848505206585e-05, "loss": 0.0978, "step": 43090 }, { "epoch": 3.62, "learning_rate": 1.2761168962042325e-05, "loss": 0.0613, "step": 43100 }, { "epoch": 3.62, "learning_rate": 1.2759489418878066e-05, "loss": 0.1376, "step": 43110 }, { "epoch": 3.62, "learning_rate": 1.2757809875713806e-05, "loss": 0.1181, "step": 43120 }, { "epoch": 3.62, "learning_rate": 1.2756130332549548e-05, "loss": 0.0876, "step": 43130 }, { "epoch": 3.62, "learning_rate": 1.2754450789385289e-05, "loss": 0.0868, "step": 43140 }, { "epoch": 3.62, "learning_rate": 1.275277124622103e-05, "loss": 0.091, "step": 43150 }, { "epoch": 3.62, "learning_rate": 1.275109170305677e-05, "loss": 0.1265, "step": 43160 }, { "epoch": 3.63, "learning_rate": 1.2749412159892512e-05, "loss": 0.0569, "step": 43170 }, { "epoch": 3.63, "learning_rate": 1.274773261672825e-05, "loss": 0.0959, "step": 43180 }, { "epoch": 3.63, "learning_rate": 1.2746053073563991e-05, "loss": 0.1106, "step": 43190 }, { "epoch": 3.63, "learning_rate": 1.2744373530399732e-05, "loss": 0.0941, "step": 43200 }, { "epoch": 3.63, "learning_rate": 1.2742693987235472e-05, "loss": 0.1246, "step": 43210 }, { "epoch": 3.63, "learning_rate": 1.2741014444071214e-05, "loss": 0.0588, "step": 43220 }, { "epoch": 3.63, "learning_rate": 1.2739334900906955e-05, "loss": 0.0974, "step": 43230 }, { "epoch": 3.63, "learning_rate": 1.2737655357742695e-05, "loss": 0.1197, "step": 43240 }, { "epoch": 3.63, "learning_rate": 1.2735975814578436e-05, "loss": 0.0943, "step": 43250 }, { "epoch": 3.63, "learning_rate": 1.2734296271414178e-05, "loss": 0.1009, "step": 43260 }, { "epoch": 3.63, "learning_rate": 1.2732616728249918e-05, "loss": 0.1186, "step": 43270 }, { "epoch": 3.63, "learning_rate": 1.2730937185085657e-05, "loss": 0.0402, "step": 43280 }, { "epoch": 3.64, "learning_rate": 1.2729257641921398e-05, "loss": 0.0833, "step": 43290 }, { "epoch": 3.64, "learning_rate": 1.2727578098757138e-05, "loss": 0.0628, "step": 43300 }, { "epoch": 3.64, "learning_rate": 1.2725898555592879e-05, "loss": 0.0647, "step": 43310 }, { "epoch": 3.64, "learning_rate": 1.272421901242862e-05, "loss": 0.129, "step": 43320 }, { "epoch": 3.64, "learning_rate": 1.2722539469264361e-05, "loss": 0.0756, "step": 43330 }, { "epoch": 3.64, "learning_rate": 1.2720859926100102e-05, "loss": 0.1307, "step": 43340 }, { "epoch": 3.64, "learning_rate": 1.2719180382935842e-05, "loss": 0.0824, "step": 43350 }, { "epoch": 3.64, "learning_rate": 1.2717500839771584e-05, "loss": 0.0948, "step": 43360 }, { "epoch": 3.64, "learning_rate": 1.2715821296607325e-05, "loss": 0.0784, "step": 43370 }, { "epoch": 3.64, "learning_rate": 1.2714141753443065e-05, "loss": 0.1002, "step": 43380 }, { "epoch": 3.64, "learning_rate": 1.2712462210278804e-05, "loss": 0.0909, "step": 43390 }, { "epoch": 3.64, "learning_rate": 1.2710782667114545e-05, "loss": 0.0837, "step": 43400 }, { "epoch": 3.65, "learning_rate": 1.2709103123950287e-05, "loss": 0.0206, "step": 43410 }, { "epoch": 3.65, "learning_rate": 1.2707423580786027e-05, "loss": 0.1847, "step": 43420 }, { "epoch": 3.65, "learning_rate": 1.2705744037621768e-05, "loss": 0.0273, "step": 43430 }, { "epoch": 3.65, "learning_rate": 1.2704064494457508e-05, "loss": 0.1239, "step": 43440 }, { "epoch": 3.65, "learning_rate": 1.270238495129325e-05, "loss": 0.0675, "step": 43450 }, { "epoch": 3.65, "learning_rate": 1.270070540812899e-05, "loss": 0.1073, "step": 43460 }, { "epoch": 3.65, "learning_rate": 1.2699025864964731e-05, "loss": 0.0948, "step": 43470 }, { "epoch": 3.65, "learning_rate": 1.2697346321800472e-05, "loss": 0.0533, "step": 43480 }, { "epoch": 3.65, "learning_rate": 1.269566677863621e-05, "loss": 0.1521, "step": 43490 }, { "epoch": 3.65, "learning_rate": 1.2693987235471951e-05, "loss": 0.0141, "step": 43500 }, { "epoch": 3.65, "learning_rate": 1.2692307692307693e-05, "loss": 0.0747, "step": 43510 }, { "epoch": 3.65, "learning_rate": 1.2690628149143434e-05, "loss": 0.1068, "step": 43520 }, { "epoch": 3.66, "learning_rate": 1.2688948605979174e-05, "loss": 0.0662, "step": 43530 }, { "epoch": 3.66, "learning_rate": 1.2687269062814915e-05, "loss": 0.005, "step": 43540 }, { "epoch": 3.66, "learning_rate": 1.2685589519650657e-05, "loss": 0.1242, "step": 43550 }, { "epoch": 3.66, "learning_rate": 1.2683909976486397e-05, "loss": 0.0864, "step": 43560 }, { "epoch": 3.66, "learning_rate": 1.2682230433322138e-05, "loss": 0.0429, "step": 43570 }, { "epoch": 3.66, "learning_rate": 1.2680550890157878e-05, "loss": 0.0838, "step": 43580 }, { "epoch": 3.66, "learning_rate": 1.267887134699362e-05, "loss": 0.1123, "step": 43590 }, { "epoch": 3.66, "learning_rate": 1.2677191803829359e-05, "loss": 0.0972, "step": 43600 }, { "epoch": 3.66, "learning_rate": 1.26755122606651e-05, "loss": 0.0561, "step": 43610 }, { "epoch": 3.66, "learning_rate": 1.267383271750084e-05, "loss": 0.0925, "step": 43620 }, { "epoch": 3.66, "learning_rate": 1.267215317433658e-05, "loss": 0.0267, "step": 43630 }, { "epoch": 3.66, "learning_rate": 1.2670473631172323e-05, "loss": 0.1587, "step": 43640 }, { "epoch": 3.67, "learning_rate": 1.2668794088008063e-05, "loss": 0.0707, "step": 43650 }, { "epoch": 3.67, "learning_rate": 1.2667114544843804e-05, "loss": 0.0149, "step": 43660 }, { "epoch": 3.67, "learning_rate": 1.2665435001679544e-05, "loss": 0.1041, "step": 43670 }, { "epoch": 3.67, "learning_rate": 1.2663755458515286e-05, "loss": 0.0569, "step": 43680 }, { "epoch": 3.67, "learning_rate": 1.2662075915351027e-05, "loss": 0.0577, "step": 43690 }, { "epoch": 3.67, "learning_rate": 1.2660396372186767e-05, "loss": 0.1037, "step": 43700 }, { "epoch": 3.67, "learning_rate": 1.2658716829022506e-05, "loss": 0.1314, "step": 43710 }, { "epoch": 3.67, "learning_rate": 1.2657037285858247e-05, "loss": 0.048, "step": 43720 }, { "epoch": 3.67, "learning_rate": 1.2655357742693987e-05, "loss": 0.0871, "step": 43730 }, { "epoch": 3.67, "learning_rate": 1.265367819952973e-05, "loss": 0.0387, "step": 43740 }, { "epoch": 3.67, "learning_rate": 1.265199865636547e-05, "loss": 0.0968, "step": 43750 }, { "epoch": 3.67, "learning_rate": 1.265031911320121e-05, "loss": 0.1634, "step": 43760 }, { "epoch": 3.68, "learning_rate": 1.264863957003695e-05, "loss": 0.1757, "step": 43770 }, { "epoch": 3.68, "learning_rate": 1.2646960026872693e-05, "loss": 0.0763, "step": 43780 }, { "epoch": 3.68, "learning_rate": 1.2645280483708433e-05, "loss": 0.0317, "step": 43790 }, { "epoch": 3.68, "learning_rate": 1.2643600940544174e-05, "loss": 0.0725, "step": 43800 }, { "epoch": 3.68, "learning_rate": 1.2641921397379913e-05, "loss": 0.1445, "step": 43810 }, { "epoch": 3.68, "learning_rate": 1.2640241854215653e-05, "loss": 0.0964, "step": 43820 }, { "epoch": 3.68, "learning_rate": 1.2638562311051395e-05, "loss": 0.0773, "step": 43830 }, { "epoch": 3.68, "learning_rate": 1.2636882767887136e-05, "loss": 0.0624, "step": 43840 }, { "epoch": 3.68, "learning_rate": 1.2635203224722876e-05, "loss": 0.069, "step": 43850 }, { "epoch": 3.68, "learning_rate": 1.2633523681558617e-05, "loss": 0.0843, "step": 43860 }, { "epoch": 3.68, "learning_rate": 1.2631844138394359e-05, "loss": 0.1099, "step": 43870 }, { "epoch": 3.68, "learning_rate": 1.26301645952301e-05, "loss": 0.0593, "step": 43880 }, { "epoch": 3.69, "learning_rate": 1.262848505206584e-05, "loss": 0.067, "step": 43890 }, { "epoch": 3.69, "learning_rate": 1.262680550890158e-05, "loss": 0.0701, "step": 43900 }, { "epoch": 3.69, "learning_rate": 1.262512596573732e-05, "loss": 0.0561, "step": 43910 }, { "epoch": 3.69, "learning_rate": 1.262344642257306e-05, "loss": 0.1642, "step": 43920 }, { "epoch": 3.69, "learning_rate": 1.2621766879408802e-05, "loss": 0.0867, "step": 43930 }, { "epoch": 3.69, "learning_rate": 1.2620087336244542e-05, "loss": 0.0744, "step": 43940 }, { "epoch": 3.69, "learning_rate": 1.2618407793080283e-05, "loss": 0.0733, "step": 43950 }, { "epoch": 3.69, "learning_rate": 1.2616728249916023e-05, "loss": 0.1385, "step": 43960 }, { "epoch": 3.69, "learning_rate": 1.2615048706751765e-05, "loss": 0.091, "step": 43970 }, { "epoch": 3.69, "learning_rate": 1.2613369163587506e-05, "loss": 0.0534, "step": 43980 }, { "epoch": 3.69, "learning_rate": 1.2611689620423246e-05, "loss": 0.0523, "step": 43990 }, { "epoch": 3.69, "learning_rate": 1.2610010077258987e-05, "loss": 0.0453, "step": 44000 }, { "epoch": 3.7, "learning_rate": 1.2608330534094729e-05, "loss": 0.0421, "step": 44010 }, { "epoch": 3.7, "learning_rate": 1.2606650990930468e-05, "loss": 0.1307, "step": 44020 }, { "epoch": 3.7, "learning_rate": 1.2604971447766208e-05, "loss": 0.0717, "step": 44030 }, { "epoch": 3.7, "learning_rate": 1.2603291904601949e-05, "loss": 0.0651, "step": 44040 }, { "epoch": 3.7, "learning_rate": 1.2601612361437689e-05, "loss": 0.1611, "step": 44050 }, { "epoch": 3.7, "learning_rate": 1.2599932818273431e-05, "loss": 0.031, "step": 44060 }, { "epoch": 3.7, "learning_rate": 1.2598253275109172e-05, "loss": 0.0702, "step": 44070 }, { "epoch": 3.7, "learning_rate": 1.2596573731944912e-05, "loss": 0.0821, "step": 44080 }, { "epoch": 3.7, "learning_rate": 1.2594894188780653e-05, "loss": 0.1111, "step": 44090 }, { "epoch": 3.7, "learning_rate": 1.2593214645616393e-05, "loss": 0.0542, "step": 44100 }, { "epoch": 3.7, "learning_rate": 1.2591535102452135e-05, "loss": 0.12, "step": 44110 }, { "epoch": 3.71, "learning_rate": 1.2589855559287876e-05, "loss": 0.1127, "step": 44120 }, { "epoch": 3.71, "learning_rate": 1.2588176016123615e-05, "loss": 0.0698, "step": 44130 }, { "epoch": 3.71, "learning_rate": 1.2586496472959355e-05, "loss": 0.0646, "step": 44140 }, { "epoch": 3.71, "learning_rate": 1.2584816929795096e-05, "loss": 0.1085, "step": 44150 }, { "epoch": 3.71, "learning_rate": 1.2583137386630838e-05, "loss": 0.0322, "step": 44160 }, { "epoch": 3.71, "learning_rate": 1.2581457843466578e-05, "loss": 0.0393, "step": 44170 }, { "epoch": 3.71, "learning_rate": 1.2579778300302319e-05, "loss": 0.0364, "step": 44180 }, { "epoch": 3.71, "learning_rate": 1.2578098757138059e-05, "loss": 0.1199, "step": 44190 }, { "epoch": 3.71, "learning_rate": 1.2576419213973801e-05, "loss": 0.0611, "step": 44200 }, { "epoch": 3.71, "learning_rate": 1.2574739670809542e-05, "loss": 0.0959, "step": 44210 }, { "epoch": 3.71, "learning_rate": 1.2573060127645282e-05, "loss": 0.0532, "step": 44220 }, { "epoch": 3.71, "learning_rate": 1.2571380584481023e-05, "loss": 0.0487, "step": 44230 }, { "epoch": 3.72, "learning_rate": 1.2569701041316761e-05, "loss": 0.0702, "step": 44240 }, { "epoch": 3.72, "learning_rate": 1.2568021498152504e-05, "loss": 0.068, "step": 44250 }, { "epoch": 3.72, "learning_rate": 1.2566341954988244e-05, "loss": 0.0802, "step": 44260 }, { "epoch": 3.72, "learning_rate": 1.2564662411823985e-05, "loss": 0.0388, "step": 44270 }, { "epoch": 3.72, "learning_rate": 1.2562982868659725e-05, "loss": 0.0418, "step": 44280 }, { "epoch": 3.72, "learning_rate": 1.2561303325495466e-05, "loss": 0.0362, "step": 44290 }, { "epoch": 3.72, "learning_rate": 1.2559623782331208e-05, "loss": 0.0461, "step": 44300 }, { "epoch": 3.72, "learning_rate": 1.2557944239166948e-05, "loss": 0.0684, "step": 44310 }, { "epoch": 3.72, "learning_rate": 1.2556264696002689e-05, "loss": 0.1927, "step": 44320 }, { "epoch": 3.72, "learning_rate": 1.255458515283843e-05, "loss": 0.1139, "step": 44330 }, { "epoch": 3.72, "learning_rate": 1.2552905609674168e-05, "loss": 0.1012, "step": 44340 }, { "epoch": 3.72, "learning_rate": 1.255122606650991e-05, "loss": 0.023, "step": 44350 }, { "epoch": 3.73, "learning_rate": 1.254954652334565e-05, "loss": 0.1563, "step": 44360 }, { "epoch": 3.73, "learning_rate": 1.2547866980181391e-05, "loss": 0.1743, "step": 44370 }, { "epoch": 3.73, "learning_rate": 1.2546187437017132e-05, "loss": 0.1563, "step": 44380 }, { "epoch": 3.73, "learning_rate": 1.2544507893852874e-05, "loss": 0.0682, "step": 44390 }, { "epoch": 3.73, "learning_rate": 1.2542828350688614e-05, "loss": 0.0435, "step": 44400 }, { "epoch": 3.73, "learning_rate": 1.2541148807524355e-05, "loss": 0.045, "step": 44410 }, { "epoch": 3.73, "learning_rate": 1.2539469264360095e-05, "loss": 0.1195, "step": 44420 }, { "epoch": 3.73, "learning_rate": 1.2537789721195837e-05, "loss": 0.0739, "step": 44430 }, { "epoch": 3.73, "learning_rate": 1.2536110178031578e-05, "loss": 0.0309, "step": 44440 }, { "epoch": 3.73, "learning_rate": 1.2534430634867317e-05, "loss": 0.0576, "step": 44450 }, { "epoch": 3.73, "learning_rate": 1.2532751091703057e-05, "loss": 0.1491, "step": 44460 }, { "epoch": 3.73, "learning_rate": 1.2531071548538798e-05, "loss": 0.0071, "step": 44470 }, { "epoch": 3.74, "learning_rate": 1.2529392005374538e-05, "loss": 0.0596, "step": 44480 }, { "epoch": 3.74, "learning_rate": 1.252771246221028e-05, "loss": 0.0366, "step": 44490 }, { "epoch": 3.74, "learning_rate": 1.252603291904602e-05, "loss": 0.0474, "step": 44500 }, { "epoch": 3.74, "learning_rate": 1.2524353375881761e-05, "loss": 0.0692, "step": 44510 }, { "epoch": 3.74, "learning_rate": 1.2522673832717502e-05, "loss": 0.2427, "step": 44520 }, { "epoch": 3.74, "learning_rate": 1.2520994289553244e-05, "loss": 0.0185, "step": 44530 }, { "epoch": 3.74, "learning_rate": 1.2519314746388984e-05, "loss": 0.0702, "step": 44540 }, { "epoch": 3.74, "learning_rate": 1.2517635203224723e-05, "loss": 0.0816, "step": 44550 }, { "epoch": 3.74, "learning_rate": 1.2515955660060464e-05, "loss": 0.0715, "step": 44560 }, { "epoch": 3.74, "learning_rate": 1.2514276116896204e-05, "loss": 0.0985, "step": 44570 }, { "epoch": 3.74, "learning_rate": 1.2512596573731946e-05, "loss": 0.1109, "step": 44580 }, { "epoch": 3.74, "learning_rate": 1.2510917030567687e-05, "loss": 0.1072, "step": 44590 }, { "epoch": 3.75, "learning_rate": 1.2509237487403427e-05, "loss": 0.0723, "step": 44600 }, { "epoch": 3.75, "learning_rate": 1.2507557944239168e-05, "loss": 0.0389, "step": 44610 }, { "epoch": 3.75, "learning_rate": 1.250587840107491e-05, "loss": 0.1591, "step": 44620 }, { "epoch": 3.75, "learning_rate": 1.250419885791065e-05, "loss": 0.0259, "step": 44630 }, { "epoch": 3.75, "learning_rate": 1.250251931474639e-05, "loss": 0.0878, "step": 44640 }, { "epoch": 3.75, "learning_rate": 1.2500839771582131e-05, "loss": 0.1746, "step": 44650 }, { "epoch": 3.75, "learning_rate": 1.249916022841787e-05, "loss": 0.1006, "step": 44660 }, { "epoch": 3.75, "learning_rate": 1.249748068525361e-05, "loss": 0.0502, "step": 44670 }, { "epoch": 3.75, "learning_rate": 1.2495801142089353e-05, "loss": 0.0366, "step": 44680 }, { "epoch": 3.75, "learning_rate": 1.2494121598925093e-05, "loss": 0.0754, "step": 44690 }, { "epoch": 3.75, "learning_rate": 1.2492442055760834e-05, "loss": 0.0319, "step": 44700 }, { "epoch": 3.75, "learning_rate": 1.2490762512596574e-05, "loss": 0.066, "step": 44710 }, { "epoch": 3.76, "learning_rate": 1.2489082969432316e-05, "loss": 0.1085, "step": 44720 }, { "epoch": 3.76, "learning_rate": 1.2487403426268057e-05, "loss": 0.0806, "step": 44730 }, { "epoch": 3.76, "learning_rate": 1.2485723883103797e-05, "loss": 0.0579, "step": 44740 }, { "epoch": 3.76, "learning_rate": 1.2484044339939538e-05, "loss": 0.1543, "step": 44750 }, { "epoch": 3.76, "learning_rate": 1.248236479677528e-05, "loss": 0.0248, "step": 44760 }, { "epoch": 3.76, "learning_rate": 1.2480685253611019e-05, "loss": 0.1053, "step": 44770 }, { "epoch": 3.76, "learning_rate": 1.2479005710446759e-05, "loss": 0.0634, "step": 44780 }, { "epoch": 3.76, "learning_rate": 1.24773261672825e-05, "loss": 0.0674, "step": 44790 }, { "epoch": 3.76, "learning_rate": 1.247564662411824e-05, "loss": 0.1367, "step": 44800 }, { "epoch": 3.76, "learning_rate": 1.2473967080953982e-05, "loss": 0.0297, "step": 44810 }, { "epoch": 3.76, "learning_rate": 1.2472287537789723e-05, "loss": 0.1082, "step": 44820 }, { "epoch": 3.76, "learning_rate": 1.2470607994625463e-05, "loss": 0.0576, "step": 44830 }, { "epoch": 3.77, "learning_rate": 1.2468928451461204e-05, "loss": 0.0984, "step": 44840 }, { "epoch": 3.77, "learning_rate": 1.2467248908296946e-05, "loss": 0.0547, "step": 44850 }, { "epoch": 3.77, "learning_rate": 1.2465569365132686e-05, "loss": 0.1011, "step": 44860 }, { "epoch": 3.77, "learning_rate": 1.2463889821968425e-05, "loss": 0.1093, "step": 44870 }, { "epoch": 3.77, "learning_rate": 1.2462210278804166e-05, "loss": 0.1033, "step": 44880 }, { "epoch": 3.77, "learning_rate": 1.2460530735639906e-05, "loss": 0.0734, "step": 44890 }, { "epoch": 3.77, "learning_rate": 1.2458851192475647e-05, "loss": 0.0946, "step": 44900 }, { "epoch": 3.77, "learning_rate": 1.2457171649311389e-05, "loss": 0.0797, "step": 44910 }, { "epoch": 3.77, "learning_rate": 1.245549210614713e-05, "loss": 0.115, "step": 44920 }, { "epoch": 3.77, "learning_rate": 1.245381256298287e-05, "loss": 0.101, "step": 44930 }, { "epoch": 3.77, "learning_rate": 1.245213301981861e-05, "loss": 0.0755, "step": 44940 }, { "epoch": 3.77, "learning_rate": 1.2450453476654352e-05, "loss": 0.0321, "step": 44950 }, { "epoch": 3.78, "learning_rate": 1.2448773933490093e-05, "loss": 0.0441, "step": 44960 }, { "epoch": 3.78, "learning_rate": 1.2447094390325833e-05, "loss": 0.1354, "step": 44970 }, { "epoch": 3.78, "learning_rate": 1.2445414847161572e-05, "loss": 0.0475, "step": 44980 }, { "epoch": 3.78, "learning_rate": 1.2443735303997312e-05, "loss": 0.0942, "step": 44990 }, { "epoch": 3.78, "learning_rate": 1.2442055760833055e-05, "loss": 0.0793, "step": 45000 }, { "epoch": 3.78, "learning_rate": 1.2440376217668795e-05, "loss": 0.0369, "step": 45010 }, { "epoch": 3.78, "learning_rate": 1.2438696674504536e-05, "loss": 0.1446, "step": 45020 }, { "epoch": 3.78, "learning_rate": 1.2437017131340276e-05, "loss": 0.0646, "step": 45030 }, { "epoch": 3.78, "learning_rate": 1.2435337588176018e-05, "loss": 0.2171, "step": 45040 }, { "epoch": 3.78, "learning_rate": 1.2433658045011759e-05, "loss": 0.0623, "step": 45050 }, { "epoch": 3.78, "learning_rate": 1.24319785018475e-05, "loss": 0.0315, "step": 45060 }, { "epoch": 3.78, "learning_rate": 1.243029895868324e-05, "loss": 0.071, "step": 45070 }, { "epoch": 3.79, "learning_rate": 1.2428619415518978e-05, "loss": 0.1325, "step": 45080 }, { "epoch": 3.79, "learning_rate": 1.2426939872354719e-05, "loss": 0.1025, "step": 45090 }, { "epoch": 3.79, "learning_rate": 1.2425260329190461e-05, "loss": 0.096, "step": 45100 }, { "epoch": 3.79, "learning_rate": 1.2423580786026202e-05, "loss": 0.0331, "step": 45110 }, { "epoch": 3.79, "learning_rate": 1.2421901242861942e-05, "loss": 0.1998, "step": 45120 }, { "epoch": 3.79, "learning_rate": 1.2420221699697683e-05, "loss": 0.1235, "step": 45130 }, { "epoch": 3.79, "learning_rate": 1.2418542156533425e-05, "loss": 0.1302, "step": 45140 }, { "epoch": 3.79, "learning_rate": 1.2416862613369165e-05, "loss": 0.0245, "step": 45150 }, { "epoch": 3.79, "learning_rate": 1.2415183070204906e-05, "loss": 0.0618, "step": 45160 }, { "epoch": 3.79, "learning_rate": 1.2413503527040646e-05, "loss": 0.0254, "step": 45170 }, { "epoch": 3.79, "learning_rate": 1.2411823983876388e-05, "loss": 0.088, "step": 45180 }, { "epoch": 3.79, "learning_rate": 1.2410144440712127e-05, "loss": 0.0833, "step": 45190 }, { "epoch": 3.8, "learning_rate": 1.2408464897547868e-05, "loss": 0.0609, "step": 45200 }, { "epoch": 3.8, "learning_rate": 1.2406785354383608e-05, "loss": 0.0282, "step": 45210 }, { "epoch": 3.8, "learning_rate": 1.2405105811219349e-05, "loss": 0.1686, "step": 45220 }, { "epoch": 3.8, "learning_rate": 1.240342626805509e-05, "loss": 0.0998, "step": 45230 }, { "epoch": 3.8, "learning_rate": 1.2401746724890831e-05, "loss": 0.0407, "step": 45240 }, { "epoch": 3.8, "learning_rate": 1.2400067181726572e-05, "loss": 0.0913, "step": 45250 }, { "epoch": 3.8, "learning_rate": 1.2398387638562312e-05, "loss": 0.0501, "step": 45260 }, { "epoch": 3.8, "learning_rate": 1.2396708095398053e-05, "loss": 0.106, "step": 45270 }, { "epoch": 3.8, "learning_rate": 1.2395028552233795e-05, "loss": 0.19, "step": 45280 }, { "epoch": 3.8, "learning_rate": 1.2393349009069535e-05, "loss": 0.0626, "step": 45290 }, { "epoch": 3.8, "learning_rate": 1.2391669465905274e-05, "loss": 0.0571, "step": 45300 }, { "epoch": 3.81, "learning_rate": 1.2389989922741015e-05, "loss": 0.0954, "step": 45310 }, { "epoch": 3.81, "learning_rate": 1.2388310379576755e-05, "loss": 0.0668, "step": 45320 }, { "epoch": 3.81, "learning_rate": 1.2386630836412497e-05, "loss": 0.1113, "step": 45330 }, { "epoch": 3.81, "learning_rate": 1.2384951293248238e-05, "loss": 0.0836, "step": 45340 }, { "epoch": 3.81, "learning_rate": 1.2383271750083978e-05, "loss": 0.0904, "step": 45350 }, { "epoch": 3.81, "learning_rate": 1.2381592206919719e-05, "loss": 0.0378, "step": 45360 }, { "epoch": 3.81, "learning_rate": 1.237991266375546e-05, "loss": 0.0456, "step": 45370 }, { "epoch": 3.81, "learning_rate": 1.2378233120591201e-05, "loss": 0.165, "step": 45380 }, { "epoch": 3.81, "learning_rate": 1.2376553577426942e-05, "loss": 0.0955, "step": 45390 }, { "epoch": 3.81, "learning_rate": 1.237487403426268e-05, "loss": 0.0807, "step": 45400 }, { "epoch": 3.81, "learning_rate": 1.2373194491098421e-05, "loss": 0.0514, "step": 45410 }, { "epoch": 3.81, "learning_rate": 1.2371514947934161e-05, "loss": 0.0566, "step": 45420 }, { "epoch": 3.82, "learning_rate": 1.2369835404769904e-05, "loss": 0.074, "step": 45430 }, { "epoch": 3.82, "learning_rate": 1.2368155861605644e-05, "loss": 0.0838, "step": 45440 }, { "epoch": 3.82, "learning_rate": 1.2366476318441385e-05, "loss": 0.1089, "step": 45450 }, { "epoch": 3.82, "learning_rate": 1.2364796775277125e-05, "loss": 0.1819, "step": 45460 }, { "epoch": 3.82, "learning_rate": 1.2363117232112867e-05, "loss": 0.0679, "step": 45470 }, { "epoch": 3.82, "learning_rate": 1.2361437688948608e-05, "loss": 0.0329, "step": 45480 }, { "epoch": 3.82, "learning_rate": 1.2359758145784348e-05, "loss": 0.0202, "step": 45490 }, { "epoch": 3.82, "learning_rate": 1.2358078602620089e-05, "loss": 0.0996, "step": 45500 }, { "epoch": 3.82, "learning_rate": 1.2356399059455827e-05, "loss": 0.2084, "step": 45510 }, { "epoch": 3.82, "learning_rate": 1.235471951629157e-05, "loss": 0.04, "step": 45520 }, { "epoch": 3.82, "learning_rate": 1.235303997312731e-05, "loss": 0.0572, "step": 45530 }, { "epoch": 3.82, "learning_rate": 1.235136042996305e-05, "loss": 0.1528, "step": 45540 }, { "epoch": 3.83, "learning_rate": 1.2349680886798791e-05, "loss": 0.1801, "step": 45550 }, { "epoch": 3.83, "learning_rate": 1.2348001343634533e-05, "loss": 0.1244, "step": 45560 }, { "epoch": 3.83, "learning_rate": 1.2346321800470274e-05, "loss": 0.0655, "step": 45570 }, { "epoch": 3.83, "learning_rate": 1.2344642257306014e-05, "loss": 0.094, "step": 45580 }, { "epoch": 3.83, "learning_rate": 1.2342962714141755e-05, "loss": 0.0214, "step": 45590 }, { "epoch": 3.83, "learning_rate": 1.2341283170977497e-05, "loss": 0.0478, "step": 45600 }, { "epoch": 3.83, "learning_rate": 1.2339603627813234e-05, "loss": 0.0395, "step": 45610 }, { "epoch": 3.83, "learning_rate": 1.2337924084648976e-05, "loss": 0.0711, "step": 45620 }, { "epoch": 3.83, "learning_rate": 1.2336244541484717e-05, "loss": 0.063, "step": 45630 }, { "epoch": 3.83, "learning_rate": 1.2334564998320457e-05, "loss": 0.0491, "step": 45640 }, { "epoch": 3.83, "learning_rate": 1.2332885455156198e-05, "loss": 0.1791, "step": 45650 }, { "epoch": 3.83, "learning_rate": 1.233120591199194e-05, "loss": 0.1068, "step": 45660 }, { "epoch": 3.84, "learning_rate": 1.232952636882768e-05, "loss": 0.0782, "step": 45670 }, { "epoch": 3.84, "learning_rate": 1.232784682566342e-05, "loss": 0.0485, "step": 45680 }, { "epoch": 3.84, "learning_rate": 1.2326167282499161e-05, "loss": 0.0434, "step": 45690 }, { "epoch": 3.84, "learning_rate": 1.2324487739334903e-05, "loss": 0.0983, "step": 45700 }, { "epoch": 3.84, "learning_rate": 1.2322808196170644e-05, "loss": 0.0407, "step": 45710 }, { "epoch": 3.84, "learning_rate": 1.2321128653006383e-05, "loss": 0.052, "step": 45720 }, { "epoch": 3.84, "learning_rate": 1.2319449109842123e-05, "loss": 0.0899, "step": 45730 }, { "epoch": 3.84, "learning_rate": 1.2317769566677863e-05, "loss": 0.0612, "step": 45740 }, { "epoch": 3.84, "learning_rate": 1.2316090023513606e-05, "loss": 0.0709, "step": 45750 }, { "epoch": 3.84, "learning_rate": 1.2314410480349346e-05, "loss": 0.02, "step": 45760 }, { "epoch": 3.84, "learning_rate": 1.2312730937185087e-05, "loss": 0.0751, "step": 45770 }, { "epoch": 3.84, "learning_rate": 1.2311051394020827e-05, "loss": 0.0994, "step": 45780 }, { "epoch": 3.85, "learning_rate": 1.230937185085657e-05, "loss": 0.0414, "step": 45790 }, { "epoch": 3.85, "learning_rate": 1.230769230769231e-05, "loss": 0.1097, "step": 45800 }, { "epoch": 3.85, "learning_rate": 1.230601276452805e-05, "loss": 0.0739, "step": 45810 }, { "epoch": 3.85, "learning_rate": 1.230433322136379e-05, "loss": 0.0031, "step": 45820 }, { "epoch": 3.85, "learning_rate": 1.230265367819953e-05, "loss": 0.1234, "step": 45830 }, { "epoch": 3.85, "learning_rate": 1.230097413503527e-05, "loss": 0.0539, "step": 45840 }, { "epoch": 3.85, "learning_rate": 1.2299294591871012e-05, "loss": 0.0932, "step": 45850 }, { "epoch": 3.85, "learning_rate": 1.2297615048706753e-05, "loss": 0.1011, "step": 45860 }, { "epoch": 3.85, "learning_rate": 1.2295935505542493e-05, "loss": 0.1129, "step": 45870 }, { "epoch": 3.85, "learning_rate": 1.2294255962378234e-05, "loss": 0.0746, "step": 45880 }, { "epoch": 3.85, "learning_rate": 1.2292576419213976e-05, "loss": 0.0713, "step": 45890 }, { "epoch": 3.85, "learning_rate": 1.2290896876049716e-05, "loss": 0.0376, "step": 45900 }, { "epoch": 3.86, "learning_rate": 1.2289217332885457e-05, "loss": 0.1093, "step": 45910 }, { "epoch": 3.86, "learning_rate": 1.2287537789721197e-05, "loss": 0.1622, "step": 45920 }, { "epoch": 3.86, "learning_rate": 1.2285858246556936e-05, "loss": 0.0679, "step": 45930 }, { "epoch": 3.86, "learning_rate": 1.2284178703392678e-05, "loss": 0.0556, "step": 45940 }, { "epoch": 3.86, "learning_rate": 1.2282499160228419e-05, "loss": 0.0945, "step": 45950 }, { "epoch": 3.86, "learning_rate": 1.2280819617064159e-05, "loss": 0.0666, "step": 45960 }, { "epoch": 3.86, "learning_rate": 1.22791400738999e-05, "loss": 0.059, "step": 45970 }, { "epoch": 3.86, "learning_rate": 1.2277460530735642e-05, "loss": 0.0582, "step": 45980 }, { "epoch": 3.86, "learning_rate": 1.2275780987571382e-05, "loss": 0.0945, "step": 45990 }, { "epoch": 3.86, "learning_rate": 1.2274101444407123e-05, "loss": 0.0484, "step": 46000 }, { "epoch": 3.86, "learning_rate": 1.2272421901242863e-05, "loss": 0.0393, "step": 46010 }, { "epoch": 3.86, "learning_rate": 1.2270742358078605e-05, "loss": 0.1091, "step": 46020 }, { "epoch": 3.87, "learning_rate": 1.2269062814914346e-05, "loss": 0.1059, "step": 46030 }, { "epoch": 3.87, "learning_rate": 1.2267383271750085e-05, "loss": 0.1221, "step": 46040 }, { "epoch": 3.87, "learning_rate": 1.2265703728585825e-05, "loss": 0.0688, "step": 46050 }, { "epoch": 3.87, "learning_rate": 1.2264024185421566e-05, "loss": 0.1298, "step": 46060 }, { "epoch": 3.87, "learning_rate": 1.2262344642257306e-05, "loss": 0.0422, "step": 46070 }, { "epoch": 3.87, "learning_rate": 1.2260665099093048e-05, "loss": 0.1263, "step": 46080 }, { "epoch": 3.87, "learning_rate": 1.2258985555928789e-05, "loss": 0.013, "step": 46090 }, { "epoch": 3.87, "learning_rate": 1.2257306012764529e-05, "loss": 0.0143, "step": 46100 }, { "epoch": 3.87, "learning_rate": 1.225562646960027e-05, "loss": 0.0771, "step": 46110 }, { "epoch": 3.87, "learning_rate": 1.2253946926436012e-05, "loss": 0.0945, "step": 46120 }, { "epoch": 3.87, "learning_rate": 1.2252267383271752e-05, "loss": 0.0734, "step": 46130 }, { "epoch": 3.87, "learning_rate": 1.2250587840107491e-05, "loss": 0.1292, "step": 46140 }, { "epoch": 3.88, "learning_rate": 1.2248908296943232e-05, "loss": 0.0974, "step": 46150 }, { "epoch": 3.88, "learning_rate": 1.2247228753778972e-05, "loss": 0.0035, "step": 46160 }, { "epoch": 3.88, "learning_rate": 1.2245549210614714e-05, "loss": 0.0686, "step": 46170 }, { "epoch": 3.88, "learning_rate": 1.2243869667450455e-05, "loss": 0.1727, "step": 46180 }, { "epoch": 3.88, "learning_rate": 1.2242190124286195e-05, "loss": 0.0153, "step": 46190 }, { "epoch": 3.88, "learning_rate": 1.2240510581121936e-05, "loss": 0.0835, "step": 46200 }, { "epoch": 3.88, "learning_rate": 1.2238831037957676e-05, "loss": 0.0909, "step": 46210 }, { "epoch": 3.88, "learning_rate": 1.2237151494793418e-05, "loss": 0.1087, "step": 46220 }, { "epoch": 3.88, "learning_rate": 1.2235471951629159e-05, "loss": 0.064, "step": 46230 }, { "epoch": 3.88, "learning_rate": 1.22337924084649e-05, "loss": 0.1106, "step": 46240 }, { "epoch": 3.88, "learning_rate": 1.2232112865300638e-05, "loss": 0.0864, "step": 46250 }, { "epoch": 3.88, "learning_rate": 1.2230433322136378e-05, "loss": 0.065, "step": 46260 }, { "epoch": 3.89, "learning_rate": 1.222875377897212e-05, "loss": 0.0685, "step": 46270 }, { "epoch": 3.89, "learning_rate": 1.2227074235807861e-05, "loss": 0.1107, "step": 46280 }, { "epoch": 3.89, "learning_rate": 1.2225394692643602e-05, "loss": 0.0992, "step": 46290 }, { "epoch": 3.89, "learning_rate": 1.2223715149479342e-05, "loss": 0.1605, "step": 46300 }, { "epoch": 3.89, "learning_rate": 1.2222035606315084e-05, "loss": 0.0587, "step": 46310 }, { "epoch": 3.89, "learning_rate": 1.2220356063150825e-05, "loss": 0.1054, "step": 46320 }, { "epoch": 3.89, "learning_rate": 1.2218676519986565e-05, "loss": 0.137, "step": 46330 }, { "epoch": 3.89, "learning_rate": 1.2216996976822306e-05, "loss": 0.0224, "step": 46340 }, { "epoch": 3.89, "learning_rate": 1.2215317433658048e-05, "loss": 0.055, "step": 46350 }, { "epoch": 3.89, "learning_rate": 1.2213637890493787e-05, "loss": 0.036, "step": 46360 }, { "epoch": 3.89, "learning_rate": 1.2211958347329527e-05, "loss": 0.0498, "step": 46370 }, { "epoch": 3.89, "learning_rate": 1.2210278804165268e-05, "loss": 0.0944, "step": 46380 }, { "epoch": 3.9, "learning_rate": 1.2208599261001008e-05, "loss": 0.0409, "step": 46390 }, { "epoch": 3.9, "learning_rate": 1.2206919717836749e-05, "loss": 0.0709, "step": 46400 }, { "epoch": 3.9, "learning_rate": 1.220524017467249e-05, "loss": 0.0526, "step": 46410 }, { "epoch": 3.9, "learning_rate": 1.2203560631508231e-05, "loss": 0.1014, "step": 46420 }, { "epoch": 3.9, "learning_rate": 1.2201881088343972e-05, "loss": 0.0661, "step": 46430 }, { "epoch": 3.9, "learning_rate": 1.2200201545179712e-05, "loss": 0.0996, "step": 46440 }, { "epoch": 3.9, "learning_rate": 1.2198522002015454e-05, "loss": 0.0594, "step": 46450 }, { "epoch": 3.9, "learning_rate": 1.2196842458851193e-05, "loss": 0.093, "step": 46460 }, { "epoch": 3.9, "learning_rate": 1.2195162915686934e-05, "loss": 0.0866, "step": 46470 }, { "epoch": 3.9, "learning_rate": 1.2193483372522674e-05, "loss": 0.0623, "step": 46480 }, { "epoch": 3.9, "learning_rate": 1.2191803829358414e-05, "loss": 0.1337, "step": 46490 }, { "epoch": 3.9, "learning_rate": 1.2190124286194157e-05, "loss": 0.1039, "step": 46500 }, { "epoch": 3.91, "learning_rate": 1.2188444743029897e-05, "loss": 0.1436, "step": 46510 }, { "epoch": 3.91, "learning_rate": 1.2186765199865638e-05, "loss": 0.0436, "step": 46520 }, { "epoch": 3.91, "learning_rate": 1.2185085656701378e-05, "loss": 0.1271, "step": 46530 }, { "epoch": 3.91, "learning_rate": 1.218340611353712e-05, "loss": 0.1306, "step": 46540 }, { "epoch": 3.91, "learning_rate": 1.218172657037286e-05, "loss": 0.0981, "step": 46550 }, { "epoch": 3.91, "learning_rate": 1.2180047027208601e-05, "loss": 0.1505, "step": 46560 }, { "epoch": 3.91, "learning_rate": 1.217836748404434e-05, "loss": 0.1068, "step": 46570 }, { "epoch": 3.91, "learning_rate": 1.217668794088008e-05, "loss": 0.0876, "step": 46580 }, { "epoch": 3.91, "learning_rate": 1.2175008397715821e-05, "loss": 0.0899, "step": 46590 }, { "epoch": 3.91, "learning_rate": 1.2173328854551563e-05, "loss": 0.0825, "step": 46600 }, { "epoch": 3.91, "learning_rate": 1.2171649311387304e-05, "loss": 0.0563, "step": 46610 }, { "epoch": 3.92, "learning_rate": 1.2169969768223044e-05, "loss": 0.0949, "step": 46620 }, { "epoch": 3.92, "learning_rate": 1.2168290225058785e-05, "loss": 0.1212, "step": 46630 }, { "epoch": 3.92, "learning_rate": 1.2166610681894527e-05, "loss": 0.0796, "step": 46640 }, { "epoch": 3.92, "learning_rate": 1.2164931138730267e-05, "loss": 0.0592, "step": 46650 }, { "epoch": 3.92, "learning_rate": 1.2163251595566008e-05, "loss": 0.0233, "step": 46660 }, { "epoch": 3.92, "learning_rate": 1.2161572052401746e-05, "loss": 0.1181, "step": 46670 }, { "epoch": 3.92, "learning_rate": 1.2159892509237487e-05, "loss": 0.1465, "step": 46680 }, { "epoch": 3.92, "learning_rate": 1.2158212966073229e-05, "loss": 0.0385, "step": 46690 }, { "epoch": 3.92, "learning_rate": 1.215653342290897e-05, "loss": 0.0636, "step": 46700 }, { "epoch": 3.92, "learning_rate": 1.215485387974471e-05, "loss": 0.129, "step": 46710 }, { "epoch": 3.92, "learning_rate": 1.215317433658045e-05, "loss": 0.0894, "step": 46720 }, { "epoch": 3.92, "learning_rate": 1.2151494793416193e-05, "loss": 0.0341, "step": 46730 }, { "epoch": 3.93, "learning_rate": 1.2149815250251933e-05, "loss": 0.1113, "step": 46740 }, { "epoch": 3.93, "learning_rate": 1.2148135707087674e-05, "loss": 0.1318, "step": 46750 }, { "epoch": 3.93, "learning_rate": 1.2146456163923414e-05, "loss": 0.0586, "step": 46760 }, { "epoch": 3.93, "learning_rate": 1.2144776620759156e-05, "loss": 0.0254, "step": 46770 }, { "epoch": 3.93, "learning_rate": 1.2143097077594893e-05, "loss": 0.1758, "step": 46780 }, { "epoch": 3.93, "learning_rate": 1.2141417534430636e-05, "loss": 0.0136, "step": 46790 }, { "epoch": 3.93, "learning_rate": 1.2139737991266376e-05, "loss": 0.0518, "step": 46800 }, { "epoch": 3.93, "learning_rate": 1.2138058448102117e-05, "loss": 0.0721, "step": 46810 }, { "epoch": 3.93, "learning_rate": 1.2136378904937857e-05, "loss": 0.0855, "step": 46820 }, { "epoch": 3.93, "learning_rate": 1.21346993617736e-05, "loss": 0.0765, "step": 46830 }, { "epoch": 3.93, "learning_rate": 1.213301981860934e-05, "loss": 0.1547, "step": 46840 }, { "epoch": 3.93, "learning_rate": 1.213134027544508e-05, "loss": 0.1635, "step": 46850 }, { "epoch": 3.94, "learning_rate": 1.212966073228082e-05, "loss": 0.1146, "step": 46860 }, { "epoch": 3.94, "learning_rate": 1.2127981189116563e-05, "loss": 0.1228, "step": 46870 }, { "epoch": 3.94, "learning_rate": 1.2126301645952302e-05, "loss": 0.0371, "step": 46880 }, { "epoch": 3.94, "learning_rate": 1.2124622102788042e-05, "loss": 0.0375, "step": 46890 }, { "epoch": 3.94, "learning_rate": 1.2122942559623783e-05, "loss": 0.0755, "step": 46900 }, { "epoch": 3.94, "learning_rate": 1.2121263016459523e-05, "loss": 0.0314, "step": 46910 }, { "epoch": 3.94, "learning_rate": 1.2119583473295265e-05, "loss": 0.1504, "step": 46920 }, { "epoch": 3.94, "learning_rate": 1.2117903930131006e-05, "loss": 0.1399, "step": 46930 }, { "epoch": 3.94, "learning_rate": 1.2116224386966746e-05, "loss": 0.0719, "step": 46940 }, { "epoch": 3.94, "learning_rate": 1.2114544843802487e-05, "loss": 0.0616, "step": 46950 }, { "epoch": 3.94, "learning_rate": 1.2112865300638229e-05, "loss": 0.1277, "step": 46960 }, { "epoch": 3.94, "learning_rate": 1.211118575747397e-05, "loss": 0.0831, "step": 46970 }, { "epoch": 3.95, "learning_rate": 1.210950621430971e-05, "loss": 0.0618, "step": 46980 }, { "epoch": 3.95, "learning_rate": 1.2107826671145448e-05, "loss": 0.0742, "step": 46990 }, { "epoch": 3.95, "learning_rate": 1.2106147127981189e-05, "loss": 0.0452, "step": 47000 }, { "epoch": 3.95, "learning_rate": 1.210446758481693e-05, "loss": 0.1432, "step": 47010 }, { "epoch": 3.95, "learning_rate": 1.2102788041652672e-05, "loss": 0.0605, "step": 47020 }, { "epoch": 3.95, "learning_rate": 1.2101108498488412e-05, "loss": 0.0499, "step": 47030 }, { "epoch": 3.95, "learning_rate": 1.2099428955324153e-05, "loss": 0.0261, "step": 47040 }, { "epoch": 3.95, "learning_rate": 1.2097749412159893e-05, "loss": 0.0572, "step": 47050 }, { "epoch": 3.95, "learning_rate": 1.2096069868995635e-05, "loss": 0.0859, "step": 47060 }, { "epoch": 3.95, "learning_rate": 1.2094390325831376e-05, "loss": 0.104, "step": 47070 }, { "epoch": 3.95, "learning_rate": 1.2092710782667116e-05, "loss": 0.104, "step": 47080 }, { "epoch": 3.95, "learning_rate": 1.2091031239502857e-05, "loss": 0.0478, "step": 47090 }, { "epoch": 3.96, "learning_rate": 1.2089351696338595e-05, "loss": 0.0446, "step": 47100 }, { "epoch": 3.96, "learning_rate": 1.2087672153174338e-05, "loss": 0.0283, "step": 47110 }, { "epoch": 3.96, "learning_rate": 1.2085992610010078e-05, "loss": 0.1919, "step": 47120 }, { "epoch": 3.96, "learning_rate": 1.2084313066845819e-05, "loss": 0.0991, "step": 47130 }, { "epoch": 3.96, "learning_rate": 1.2082633523681559e-05, "loss": 0.042, "step": 47140 }, { "epoch": 3.96, "learning_rate": 1.2080953980517301e-05, "loss": 0.0714, "step": 47150 }, { "epoch": 3.96, "learning_rate": 1.2079274437353042e-05, "loss": 0.0288, "step": 47160 }, { "epoch": 3.96, "learning_rate": 1.2077594894188782e-05, "loss": 0.0376, "step": 47170 }, { "epoch": 3.96, "learning_rate": 1.2075915351024523e-05, "loss": 0.0762, "step": 47180 }, { "epoch": 3.96, "learning_rate": 1.2074235807860263e-05, "loss": 0.0627, "step": 47190 }, { "epoch": 3.96, "learning_rate": 1.2072556264696002e-05, "loss": 0.1187, "step": 47200 }, { "epoch": 3.96, "learning_rate": 1.2070876721531744e-05, "loss": 0.0681, "step": 47210 }, { "epoch": 3.97, "learning_rate": 1.2069197178367485e-05, "loss": 0.0831, "step": 47220 }, { "epoch": 3.97, "learning_rate": 1.2067517635203225e-05, "loss": 0.0925, "step": 47230 }, { "epoch": 3.97, "learning_rate": 1.2065838092038965e-05, "loss": 0.1354, "step": 47240 }, { "epoch": 3.97, "learning_rate": 1.2064158548874708e-05, "loss": 0.0797, "step": 47250 }, { "epoch": 3.97, "learning_rate": 1.2062479005710448e-05, "loss": 0.0697, "step": 47260 }, { "epoch": 3.97, "learning_rate": 1.2060799462546189e-05, "loss": 0.091, "step": 47270 }, { "epoch": 3.97, "learning_rate": 1.2059119919381929e-05, "loss": 0.0861, "step": 47280 }, { "epoch": 3.97, "learning_rate": 1.2057440376217671e-05, "loss": 0.098, "step": 47290 }, { "epoch": 3.97, "learning_rate": 1.2055760833053412e-05, "loss": 0.089, "step": 47300 }, { "epoch": 3.97, "learning_rate": 1.205408128988915e-05, "loss": 0.0749, "step": 47310 }, { "epoch": 3.97, "learning_rate": 1.2052401746724891e-05, "loss": 0.0854, "step": 47320 }, { "epoch": 3.97, "learning_rate": 1.2050722203560631e-05, "loss": 0.1012, "step": 47330 }, { "epoch": 3.98, "learning_rate": 1.2049042660396374e-05, "loss": 0.071, "step": 47340 }, { "epoch": 3.98, "learning_rate": 1.2047363117232114e-05, "loss": 0.1517, "step": 47350 }, { "epoch": 3.98, "learning_rate": 1.2045683574067855e-05, "loss": 0.1516, "step": 47360 }, { "epoch": 3.98, "learning_rate": 1.2044004030903595e-05, "loss": 0.1165, "step": 47370 }, { "epoch": 3.98, "learning_rate": 1.2042324487739336e-05, "loss": 0.069, "step": 47380 }, { "epoch": 3.98, "learning_rate": 1.2040644944575078e-05, "loss": 0.1088, "step": 47390 }, { "epoch": 3.98, "learning_rate": 1.2038965401410818e-05, "loss": 0.0788, "step": 47400 }, { "epoch": 3.98, "learning_rate": 1.2037285858246557e-05, "loss": 0.0517, "step": 47410 }, { "epoch": 3.98, "learning_rate": 1.2035606315082297e-05, "loss": 0.0594, "step": 47420 }, { "epoch": 3.98, "learning_rate": 1.2033926771918038e-05, "loss": 0.0226, "step": 47430 }, { "epoch": 3.98, "learning_rate": 1.203224722875378e-05, "loss": 0.0354, "step": 47440 }, { "epoch": 3.98, "learning_rate": 1.203056768558952e-05, "loss": 0.1225, "step": 47450 }, { "epoch": 3.99, "learning_rate": 1.2028888142425261e-05, "loss": 0.0086, "step": 47460 }, { "epoch": 3.99, "learning_rate": 1.2027208599261002e-05, "loss": 0.0105, "step": 47470 }, { "epoch": 3.99, "learning_rate": 1.2025529056096744e-05, "loss": 0.1965, "step": 47480 }, { "epoch": 3.99, "learning_rate": 1.2023849512932484e-05, "loss": 0.048, "step": 47490 }, { "epoch": 3.99, "learning_rate": 1.2022169969768225e-05, "loss": 0.0273, "step": 47500 }, { "epoch": 3.99, "learning_rate": 1.2020490426603965e-05, "loss": 0.0957, "step": 47510 }, { "epoch": 3.99, "learning_rate": 1.2018810883439704e-05, "loss": 0.102, "step": 47520 }, { "epoch": 3.99, "learning_rate": 1.2017131340275446e-05, "loss": 0.1514, "step": 47530 }, { "epoch": 3.99, "learning_rate": 1.2015451797111187e-05, "loss": 0.1033, "step": 47540 }, { "epoch": 3.99, "learning_rate": 1.2013772253946927e-05, "loss": 0.0671, "step": 47550 }, { "epoch": 3.99, "learning_rate": 1.2012092710782668e-05, "loss": 0.1719, "step": 47560 }, { "epoch": 3.99, "learning_rate": 1.2010413167618408e-05, "loss": 0.034, "step": 47570 }, { "epoch": 4.0, "learning_rate": 1.200873362445415e-05, "loss": 0.0661, "step": 47580 }, { "epoch": 4.0, "learning_rate": 1.200705408128989e-05, "loss": 0.1248, "step": 47590 }, { "epoch": 4.0, "learning_rate": 1.2005374538125631e-05, "loss": 0.058, "step": 47600 }, { "epoch": 4.0, "learning_rate": 1.2003694994961372e-05, "loss": 0.0183, "step": 47610 }, { "epoch": 4.0, "learning_rate": 1.2002015451797114e-05, "loss": 0.0517, "step": 47620 }, { "epoch": 4.0, "learning_rate": 1.2000335908632853e-05, "loss": 0.1743, "step": 47630 }, { "epoch": 4.0, "eval_FN": 583, "eval_FP": 424, "eval_TN": 21290, "eval_TP": 1519, "eval_accuracy": 0.9577175008397716, "eval_f1": 0.7510506798516687, "eval_loss": 0.16761387884616852, "eval_precision": 0.7817807514153371, "eval_recall": 0.7226450999048525, "eval_runtime": 163.0122, "eval_samples_per_second": 146.1, "eval_steps_per_second": 9.134, "step": 47632 }, { "epoch": 4.0, "learning_rate": 1.1998656365468593e-05, "loss": 0.0406, "step": 47640 }, { "epoch": 4.0, "learning_rate": 1.1996976822304334e-05, "loss": 0.009, "step": 47650 }, { "epoch": 4.0, "learning_rate": 1.1995297279140074e-05, "loss": 0.0453, "step": 47660 }, { "epoch": 4.0, "learning_rate": 1.1993617735975816e-05, "loss": 0.003, "step": 47670 }, { "epoch": 4.0, "learning_rate": 1.1991938192811557e-05, "loss": 0.0313, "step": 47680 }, { "epoch": 4.0, "learning_rate": 1.1990258649647297e-05, "loss": 0.015, "step": 47690 }, { "epoch": 4.01, "learning_rate": 1.1988579106483038e-05, "loss": 0.096, "step": 47700 }, { "epoch": 4.01, "learning_rate": 1.198689956331878e-05, "loss": 0.0836, "step": 47710 }, { "epoch": 4.01, "learning_rate": 1.198522002015452e-05, "loss": 0.0268, "step": 47720 }, { "epoch": 4.01, "learning_rate": 1.1983540476990259e-05, "loss": 0.0809, "step": 47730 }, { "epoch": 4.01, "learning_rate": 1.1981860933826e-05, "loss": 0.1148, "step": 47740 }, { "epoch": 4.01, "learning_rate": 1.198018139066174e-05, "loss": 0.0171, "step": 47750 }, { "epoch": 4.01, "learning_rate": 1.197850184749748e-05, "loss": 0.0499, "step": 47760 }, { "epoch": 4.01, "learning_rate": 1.1976822304333223e-05, "loss": 0.0723, "step": 47770 }, { "epoch": 4.01, "learning_rate": 1.1975142761168963e-05, "loss": 0.0564, "step": 47780 }, { "epoch": 4.01, "learning_rate": 1.1973463218004704e-05, "loss": 0.0307, "step": 47790 }, { "epoch": 4.01, "learning_rate": 1.1971783674840444e-05, "loss": 0.0466, "step": 47800 }, { "epoch": 4.01, "learning_rate": 1.1970104131676186e-05, "loss": 0.1192, "step": 47810 }, { "epoch": 4.02, "learning_rate": 1.1968424588511927e-05, "loss": 0.037, "step": 47820 }, { "epoch": 4.02, "learning_rate": 1.1966745045347667e-05, "loss": 0.047, "step": 47830 }, { "epoch": 4.02, "learning_rate": 1.1965065502183406e-05, "loss": 0.0469, "step": 47840 }, { "epoch": 4.02, "learning_rate": 1.1963385959019146e-05, "loss": 0.0574, "step": 47850 }, { "epoch": 4.02, "learning_rate": 1.1961706415854889e-05, "loss": 0.0725, "step": 47860 }, { "epoch": 4.02, "learning_rate": 1.1960026872690629e-05, "loss": 0.0502, "step": 47870 }, { "epoch": 4.02, "learning_rate": 1.195834732952637e-05, "loss": 0.0175, "step": 47880 }, { "epoch": 4.02, "learning_rate": 1.195666778636211e-05, "loss": 0.0475, "step": 47890 }, { "epoch": 4.02, "learning_rate": 1.1954988243197852e-05, "loss": 0.0116, "step": 47900 }, { "epoch": 4.02, "learning_rate": 1.1953308700033593e-05, "loss": 0.1207, "step": 47910 }, { "epoch": 4.02, "learning_rate": 1.1951629156869333e-05, "loss": 0.049, "step": 47920 }, { "epoch": 4.03, "learning_rate": 1.1949949613705074e-05, "loss": 0.0777, "step": 47930 }, { "epoch": 4.03, "learning_rate": 1.1948270070540812e-05, "loss": 0.0607, "step": 47940 }, { "epoch": 4.03, "learning_rate": 1.1946590527376553e-05, "loss": 0.0279, "step": 47950 }, { "epoch": 4.03, "learning_rate": 1.1944910984212295e-05, "loss": 0.1341, "step": 47960 }, { "epoch": 4.03, "learning_rate": 1.1943231441048036e-05, "loss": 0.0445, "step": 47970 }, { "epoch": 4.03, "learning_rate": 1.1941551897883776e-05, "loss": 0.1484, "step": 47980 }, { "epoch": 4.03, "learning_rate": 1.1939872354719516e-05, "loss": 0.0661, "step": 47990 }, { "epoch": 4.03, "learning_rate": 1.1938192811555259e-05, "loss": 0.033, "step": 48000 }, { "epoch": 4.03, "learning_rate": 1.1936513268390999e-05, "loss": 0.0556, "step": 48010 }, { "epoch": 4.03, "learning_rate": 1.193483372522674e-05, "loss": 0.0951, "step": 48020 }, { "epoch": 4.03, "learning_rate": 1.193315418206248e-05, "loss": 0.0461, "step": 48030 }, { "epoch": 4.03, "learning_rate": 1.1931474638898222e-05, "loss": 0.0908, "step": 48040 }, { "epoch": 4.04, "learning_rate": 1.1929795095733961e-05, "loss": 0.0489, "step": 48050 }, { "epoch": 4.04, "learning_rate": 1.1928115552569702e-05, "loss": 0.0362, "step": 48060 }, { "epoch": 4.04, "learning_rate": 1.1926436009405442e-05, "loss": 0.0571, "step": 48070 }, { "epoch": 4.04, "learning_rate": 1.1924756466241182e-05, "loss": 0.1488, "step": 48080 }, { "epoch": 4.04, "learning_rate": 1.1923076923076925e-05, "loss": 0.0614, "step": 48090 }, { "epoch": 4.04, "learning_rate": 1.1921397379912665e-05, "loss": 0.0185, "step": 48100 }, { "epoch": 4.04, "learning_rate": 1.1919717836748406e-05, "loss": 0.0385, "step": 48110 }, { "epoch": 4.04, "learning_rate": 1.1918038293584146e-05, "loss": 0.0059, "step": 48120 }, { "epoch": 4.04, "learning_rate": 1.1916358750419888e-05, "loss": 0.1092, "step": 48130 }, { "epoch": 4.04, "learning_rate": 1.1914679207255629e-05, "loss": 0.0534, "step": 48140 }, { "epoch": 4.04, "learning_rate": 1.191299966409137e-05, "loss": 0.0161, "step": 48150 }, { "epoch": 4.04, "learning_rate": 1.1911320120927108e-05, "loss": 0.023, "step": 48160 }, { "epoch": 4.05, "learning_rate": 1.1909640577762848e-05, "loss": 0.1152, "step": 48170 }, { "epoch": 4.05, "learning_rate": 1.1907961034598589e-05, "loss": 0.1002, "step": 48180 }, { "epoch": 4.05, "learning_rate": 1.1906281491434331e-05, "loss": 0.001, "step": 48190 }, { "epoch": 4.05, "learning_rate": 1.1904601948270072e-05, "loss": 0.0671, "step": 48200 }, { "epoch": 4.05, "learning_rate": 1.1902922405105812e-05, "loss": 0.0672, "step": 48210 }, { "epoch": 4.05, "learning_rate": 1.1901242861941553e-05, "loss": 0.0203, "step": 48220 }, { "epoch": 4.05, "learning_rate": 1.1899563318777295e-05, "loss": 0.0224, "step": 48230 }, { "epoch": 4.05, "learning_rate": 1.1897883775613035e-05, "loss": 0.0751, "step": 48240 }, { "epoch": 4.05, "learning_rate": 1.1896204232448776e-05, "loss": 0.1014, "step": 48250 }, { "epoch": 4.05, "learning_rate": 1.1894524689284514e-05, "loss": 0.0392, "step": 48260 }, { "epoch": 4.05, "learning_rate": 1.1892845146120255e-05, "loss": 0.0641, "step": 48270 }, { "epoch": 4.05, "learning_rate": 1.1891165602955997e-05, "loss": 0.1196, "step": 48280 }, { "epoch": 4.06, "learning_rate": 1.1889486059791738e-05, "loss": 0.1308, "step": 48290 }, { "epoch": 4.06, "learning_rate": 1.1887806516627478e-05, "loss": 0.0495, "step": 48300 }, { "epoch": 4.06, "learning_rate": 1.1886126973463219e-05, "loss": 0.1104, "step": 48310 }, { "epoch": 4.06, "learning_rate": 1.188444743029896e-05, "loss": 0.0865, "step": 48320 }, { "epoch": 4.06, "learning_rate": 1.1882767887134701e-05, "loss": 0.023, "step": 48330 }, { "epoch": 4.06, "learning_rate": 1.1881088343970442e-05, "loss": 0.0863, "step": 48340 }, { "epoch": 4.06, "learning_rate": 1.1879408800806182e-05, "loss": 0.0034, "step": 48350 }, { "epoch": 4.06, "learning_rate": 1.1877729257641923e-05, "loss": 0.1374, "step": 48360 }, { "epoch": 4.06, "learning_rate": 1.1876049714477661e-05, "loss": 0.0357, "step": 48370 }, { "epoch": 4.06, "learning_rate": 1.1874370171313404e-05, "loss": 0.0435, "step": 48380 }, { "epoch": 4.06, "learning_rate": 1.1872690628149144e-05, "loss": 0.077, "step": 48390 }, { "epoch": 4.06, "learning_rate": 1.1871011084984884e-05, "loss": 0.006, "step": 48400 }, { "epoch": 4.07, "learning_rate": 1.1869331541820625e-05, "loss": 0.0498, "step": 48410 }, { "epoch": 4.07, "learning_rate": 1.1867651998656367e-05, "loss": 0.0332, "step": 48420 }, { "epoch": 4.07, "learning_rate": 1.1865972455492108e-05, "loss": 0.0366, "step": 48430 }, { "epoch": 4.07, "learning_rate": 1.1864292912327848e-05, "loss": 0.0417, "step": 48440 }, { "epoch": 4.07, "learning_rate": 1.1862613369163589e-05, "loss": 0.0216, "step": 48450 }, { "epoch": 4.07, "learning_rate": 1.186093382599933e-05, "loss": 0.1397, "step": 48460 }, { "epoch": 4.07, "learning_rate": 1.185925428283507e-05, "loss": 0.0584, "step": 48470 }, { "epoch": 4.07, "learning_rate": 1.185757473967081e-05, "loss": 0.0207, "step": 48480 }, { "epoch": 4.07, "learning_rate": 1.185589519650655e-05, "loss": 0.0896, "step": 48490 }, { "epoch": 4.07, "learning_rate": 1.1854215653342291e-05, "loss": 0.0726, "step": 48500 }, { "epoch": 4.07, "learning_rate": 1.1852536110178033e-05, "loss": 0.1332, "step": 48510 }, { "epoch": 4.07, "learning_rate": 1.1850856567013774e-05, "loss": 0.0596, "step": 48520 }, { "epoch": 4.08, "learning_rate": 1.1849177023849514e-05, "loss": 0.0664, "step": 48530 }, { "epoch": 4.08, "learning_rate": 1.1847497480685255e-05, "loss": 0.0852, "step": 48540 }, { "epoch": 4.08, "learning_rate": 1.1845817937520995e-05, "loss": 0.0199, "step": 48550 }, { "epoch": 4.08, "learning_rate": 1.1844138394356737e-05, "loss": 0.0264, "step": 48560 }, { "epoch": 4.08, "learning_rate": 1.1842458851192478e-05, "loss": 0.0288, "step": 48570 }, { "epoch": 4.08, "learning_rate": 1.1840779308028216e-05, "loss": 0.0712, "step": 48580 }, { "epoch": 4.08, "learning_rate": 1.1839099764863957e-05, "loss": 0.0936, "step": 48590 }, { "epoch": 4.08, "learning_rate": 1.1837420221699697e-05, "loss": 0.0735, "step": 48600 }, { "epoch": 4.08, "learning_rate": 1.183574067853544e-05, "loss": 0.036, "step": 48610 }, { "epoch": 4.08, "learning_rate": 1.183406113537118e-05, "loss": 0.019, "step": 48620 }, { "epoch": 4.08, "learning_rate": 1.183238159220692e-05, "loss": 0.1226, "step": 48630 }, { "epoch": 4.08, "learning_rate": 1.1830702049042661e-05, "loss": 0.0313, "step": 48640 }, { "epoch": 4.09, "learning_rate": 1.1829022505878403e-05, "loss": 0.0714, "step": 48650 }, { "epoch": 4.09, "learning_rate": 1.1827342962714144e-05, "loss": 0.086, "step": 48660 }, { "epoch": 4.09, "learning_rate": 1.1825663419549884e-05, "loss": 0.0617, "step": 48670 }, { "epoch": 4.09, "learning_rate": 1.1823983876385625e-05, "loss": 0.0623, "step": 48680 }, { "epoch": 4.09, "learning_rate": 1.1822304333221363e-05, "loss": 0.0757, "step": 48690 }, { "epoch": 4.09, "learning_rate": 1.1820624790057104e-05, "loss": 0.0634, "step": 48700 }, { "epoch": 4.09, "learning_rate": 1.1818945246892846e-05, "loss": 0.0797, "step": 48710 }, { "epoch": 4.09, "learning_rate": 1.1817265703728587e-05, "loss": 0.0422, "step": 48720 }, { "epoch": 4.09, "learning_rate": 1.1815586160564327e-05, "loss": 0.0549, "step": 48730 }, { "epoch": 4.09, "learning_rate": 1.1813906617400067e-05, "loss": 0.045, "step": 48740 }, { "epoch": 4.09, "learning_rate": 1.181222707423581e-05, "loss": 0.1201, "step": 48750 }, { "epoch": 4.09, "learning_rate": 1.181054753107155e-05, "loss": 0.0559, "step": 48760 }, { "epoch": 4.1, "learning_rate": 1.180886798790729e-05, "loss": 0.0108, "step": 48770 }, { "epoch": 4.1, "learning_rate": 1.1807188444743031e-05, "loss": 0.0556, "step": 48780 }, { "epoch": 4.1, "learning_rate": 1.180550890157877e-05, "loss": 0.1445, "step": 48790 }, { "epoch": 4.1, "learning_rate": 1.1803829358414512e-05, "loss": 0.0875, "step": 48800 }, { "epoch": 4.1, "learning_rate": 1.1802149815250253e-05, "loss": 0.0528, "step": 48810 }, { "epoch": 4.1, "learning_rate": 1.1800470272085993e-05, "loss": 0.0629, "step": 48820 }, { "epoch": 4.1, "learning_rate": 1.1798790728921733e-05, "loss": 0.0485, "step": 48830 }, { "epoch": 4.1, "learning_rate": 1.1797111185757476e-05, "loss": 0.0638, "step": 48840 }, { "epoch": 4.1, "learning_rate": 1.1795431642593216e-05, "loss": 0.0718, "step": 48850 }, { "epoch": 4.1, "learning_rate": 1.1793752099428957e-05, "loss": 0.0374, "step": 48860 }, { "epoch": 4.1, "learning_rate": 1.1792072556264697e-05, "loss": 0.1148, "step": 48870 }, { "epoch": 4.1, "learning_rate": 1.179039301310044e-05, "loss": 0.0227, "step": 48880 }, { "epoch": 4.11, "learning_rate": 1.178871346993618e-05, "loss": 0.0249, "step": 48890 }, { "epoch": 4.11, "learning_rate": 1.1787033926771918e-05, "loss": 0.0981, "step": 48900 }, { "epoch": 4.11, "learning_rate": 1.1785354383607659e-05, "loss": 0.2062, "step": 48910 }, { "epoch": 4.11, "learning_rate": 1.17836748404434e-05, "loss": 0.1522, "step": 48920 }, { "epoch": 4.11, "learning_rate": 1.178199529727914e-05, "loss": 0.0594, "step": 48930 }, { "epoch": 4.11, "learning_rate": 1.1780315754114882e-05, "loss": 0.0772, "step": 48940 }, { "epoch": 4.11, "learning_rate": 1.1778636210950623e-05, "loss": 0.1001, "step": 48950 }, { "epoch": 4.11, "learning_rate": 1.1776956667786363e-05, "loss": 0.0417, "step": 48960 }, { "epoch": 4.11, "learning_rate": 1.1775277124622104e-05, "loss": 0.0508, "step": 48970 }, { "epoch": 4.11, "learning_rate": 1.1773597581457846e-05, "loss": 0.0458, "step": 48980 }, { "epoch": 4.11, "learning_rate": 1.1771918038293586e-05, "loss": 0.0879, "step": 48990 }, { "epoch": 4.11, "learning_rate": 1.1770238495129325e-05, "loss": 0.1012, "step": 49000 }, { "epoch": 4.12, "learning_rate": 1.1768558951965065e-05, "loss": 0.0503, "step": 49010 }, { "epoch": 4.12, "learning_rate": 1.1766879408800806e-05, "loss": 0.0939, "step": 49020 }, { "epoch": 4.12, "learning_rate": 1.1765199865636548e-05, "loss": 0.0775, "step": 49030 }, { "epoch": 4.12, "learning_rate": 1.1763520322472289e-05, "loss": 0.0051, "step": 49040 }, { "epoch": 4.12, "learning_rate": 1.1761840779308029e-05, "loss": 0.0362, "step": 49050 }, { "epoch": 4.12, "learning_rate": 1.176016123614377e-05, "loss": 0.0418, "step": 49060 }, { "epoch": 4.12, "learning_rate": 1.1758481692979512e-05, "loss": 0.0337, "step": 49070 }, { "epoch": 4.12, "learning_rate": 1.1756802149815252e-05, "loss": 0.0208, "step": 49080 }, { "epoch": 4.12, "learning_rate": 1.1755122606650993e-05, "loss": 0.016, "step": 49090 }, { "epoch": 4.12, "learning_rate": 1.1753443063486733e-05, "loss": 0.226, "step": 49100 }, { "epoch": 4.12, "learning_rate": 1.1751763520322472e-05, "loss": 0.0014, "step": 49110 }, { "epoch": 4.12, "learning_rate": 1.1750083977158212e-05, "loss": 0.0745, "step": 49120 }, { "epoch": 4.13, "learning_rate": 1.1748404433993955e-05, "loss": 0.045, "step": 49130 }, { "epoch": 4.13, "learning_rate": 1.1746724890829695e-05, "loss": 0.0687, "step": 49140 }, { "epoch": 4.13, "learning_rate": 1.1745045347665435e-05, "loss": 0.1702, "step": 49150 }, { "epoch": 4.13, "learning_rate": 1.1743365804501176e-05, "loss": 0.0343, "step": 49160 }, { "epoch": 4.13, "learning_rate": 1.1741686261336918e-05, "loss": 0.0456, "step": 49170 }, { "epoch": 4.13, "learning_rate": 1.1740006718172659e-05, "loss": 0.0629, "step": 49180 }, { "epoch": 4.13, "learning_rate": 1.1738327175008399e-05, "loss": 0.0802, "step": 49190 }, { "epoch": 4.13, "learning_rate": 1.173664763184414e-05, "loss": 0.1549, "step": 49200 }, { "epoch": 4.13, "learning_rate": 1.1734968088679882e-05, "loss": 0.0711, "step": 49210 }, { "epoch": 4.13, "learning_rate": 1.173328854551562e-05, "loss": 0.1109, "step": 49220 }, { "epoch": 4.13, "learning_rate": 1.1731609002351361e-05, "loss": 0.0314, "step": 49230 }, { "epoch": 4.14, "learning_rate": 1.1729929459187101e-05, "loss": 0.1566, "step": 49240 }, { "epoch": 4.14, "learning_rate": 1.1728249916022842e-05, "loss": 0.0583, "step": 49250 }, { "epoch": 4.14, "learning_rate": 1.1726570372858584e-05, "loss": 0.0683, "step": 49260 }, { "epoch": 4.14, "learning_rate": 1.1724890829694325e-05, "loss": 0.0879, "step": 49270 }, { "epoch": 4.14, "learning_rate": 1.1723211286530065e-05, "loss": 0.0048, "step": 49280 }, { "epoch": 4.14, "learning_rate": 1.1721531743365806e-05, "loss": 0.0541, "step": 49290 }, { "epoch": 4.14, "learning_rate": 1.1719852200201548e-05, "loss": 0.1086, "step": 49300 }, { "epoch": 4.14, "learning_rate": 1.1718172657037288e-05, "loss": 0.1341, "step": 49310 }, { "epoch": 4.14, "learning_rate": 1.1716493113873027e-05, "loss": 0.0736, "step": 49320 }, { "epoch": 4.14, "learning_rate": 1.1714813570708767e-05, "loss": 0.0434, "step": 49330 }, { "epoch": 4.14, "learning_rate": 1.1713134027544508e-05, "loss": 0.1031, "step": 49340 }, { "epoch": 4.14, "learning_rate": 1.1711454484380248e-05, "loss": 0.0421, "step": 49350 }, { "epoch": 4.15, "learning_rate": 1.170977494121599e-05, "loss": 0.0804, "step": 49360 }, { "epoch": 4.15, "learning_rate": 1.1708095398051731e-05, "loss": 0.0024, "step": 49370 }, { "epoch": 4.15, "learning_rate": 1.1706415854887472e-05, "loss": 0.0287, "step": 49380 }, { "epoch": 4.15, "learning_rate": 1.1704736311723212e-05, "loss": 0.1058, "step": 49390 }, { "epoch": 4.15, "learning_rate": 1.1703056768558954e-05, "loss": 0.0438, "step": 49400 }, { "epoch": 4.15, "learning_rate": 1.1701377225394695e-05, "loss": 0.092, "step": 49410 }, { "epoch": 4.15, "learning_rate": 1.1699697682230435e-05, "loss": 0.0721, "step": 49420 }, { "epoch": 4.15, "learning_rate": 1.1698018139066174e-05, "loss": 0.1231, "step": 49430 }, { "epoch": 4.15, "learning_rate": 1.1696338595901914e-05, "loss": 0.0978, "step": 49440 }, { "epoch": 4.15, "learning_rate": 1.1694659052737657e-05, "loss": 0.0942, "step": 49450 }, { "epoch": 4.15, "learning_rate": 1.1692979509573397e-05, "loss": 0.0987, "step": 49460 }, { "epoch": 4.15, "learning_rate": 1.1691299966409138e-05, "loss": 0.0317, "step": 49470 }, { "epoch": 4.16, "learning_rate": 1.1689620423244878e-05, "loss": 0.0691, "step": 49480 }, { "epoch": 4.16, "learning_rate": 1.168794088008062e-05, "loss": 0.03, "step": 49490 }, { "epoch": 4.16, "learning_rate": 1.168626133691636e-05, "loss": 0.0529, "step": 49500 }, { "epoch": 4.16, "learning_rate": 1.1684581793752101e-05, "loss": 0.0845, "step": 49510 }, { "epoch": 4.16, "learning_rate": 1.1682902250587842e-05, "loss": 0.0443, "step": 49520 }, { "epoch": 4.16, "learning_rate": 1.168122270742358e-05, "loss": 0.0186, "step": 49530 }, { "epoch": 4.16, "learning_rate": 1.1679543164259321e-05, "loss": 0.107, "step": 49540 }, { "epoch": 4.16, "learning_rate": 1.1677863621095063e-05, "loss": 0.0657, "step": 49550 }, { "epoch": 4.16, "learning_rate": 1.1676184077930804e-05, "loss": 0.0197, "step": 49560 }, { "epoch": 4.16, "learning_rate": 1.1674504534766544e-05, "loss": 0.0133, "step": 49570 }, { "epoch": 4.16, "learning_rate": 1.1672824991602284e-05, "loss": 0.1621, "step": 49580 }, { "epoch": 4.16, "learning_rate": 1.1671145448438027e-05, "loss": 0.0114, "step": 49590 }, { "epoch": 4.17, "learning_rate": 1.1669465905273767e-05, "loss": 0.1316, "step": 49600 }, { "epoch": 4.17, "learning_rate": 1.1667786362109508e-05, "loss": 0.1266, "step": 49610 }, { "epoch": 4.17, "learning_rate": 1.1666106818945248e-05, "loss": 0.0459, "step": 49620 }, { "epoch": 4.17, "learning_rate": 1.166442727578099e-05, "loss": 0.0996, "step": 49630 }, { "epoch": 4.17, "learning_rate": 1.1662747732616729e-05, "loss": 0.0533, "step": 49640 }, { "epoch": 4.17, "learning_rate": 1.166106818945247e-05, "loss": 0.0014, "step": 49650 }, { "epoch": 4.17, "learning_rate": 1.165938864628821e-05, "loss": 0.0228, "step": 49660 }, { "epoch": 4.17, "learning_rate": 1.165770910312395e-05, "loss": 0.0575, "step": 49670 }, { "epoch": 4.17, "learning_rate": 1.1656029559959691e-05, "loss": 0.0864, "step": 49680 }, { "epoch": 4.17, "learning_rate": 1.1654350016795433e-05, "loss": 0.1173, "step": 49690 }, { "epoch": 4.17, "learning_rate": 1.1652670473631174e-05, "loss": 0.0418, "step": 49700 }, { "epoch": 4.17, "learning_rate": 1.1650990930466914e-05, "loss": 0.0578, "step": 49710 }, { "epoch": 4.18, "learning_rate": 1.1649311387302655e-05, "loss": 0.1018, "step": 49720 }, { "epoch": 4.18, "learning_rate": 1.1647631844138397e-05, "loss": 0.0876, "step": 49730 }, { "epoch": 4.18, "learning_rate": 1.1645952300974137e-05, "loss": 0.0118, "step": 49740 }, { "epoch": 4.18, "learning_rate": 1.1644272757809876e-05, "loss": 0.0526, "step": 49750 }, { "epoch": 4.18, "learning_rate": 1.1642593214645616e-05, "loss": 0.0715, "step": 49760 }, { "epoch": 4.18, "learning_rate": 1.1640913671481357e-05, "loss": 0.0173, "step": 49770 }, { "epoch": 4.18, "learning_rate": 1.1639234128317099e-05, "loss": 0.0857, "step": 49780 }, { "epoch": 4.18, "learning_rate": 1.163755458515284e-05, "loss": 0.0617, "step": 49790 }, { "epoch": 4.18, "learning_rate": 1.163587504198858e-05, "loss": 0.0654, "step": 49800 }, { "epoch": 4.18, "learning_rate": 1.163419549882432e-05, "loss": 0.0465, "step": 49810 }, { "epoch": 4.18, "learning_rate": 1.1632515955660063e-05, "loss": 0.059, "step": 49820 }, { "epoch": 4.18, "learning_rate": 1.1630836412495803e-05, "loss": 0.0773, "step": 49830 }, { "epoch": 4.19, "learning_rate": 1.1629156869331544e-05, "loss": 0.1342, "step": 49840 }, { "epoch": 4.19, "learning_rate": 1.1627477326167282e-05, "loss": 0.061, "step": 49850 }, { "epoch": 4.19, "learning_rate": 1.1625797783003023e-05, "loss": 0.0285, "step": 49860 }, { "epoch": 4.19, "learning_rate": 1.1624118239838763e-05, "loss": 0.082, "step": 49870 }, { "epoch": 4.19, "learning_rate": 1.1622438696674506e-05, "loss": 0.0644, "step": 49880 }, { "epoch": 4.19, "learning_rate": 1.1620759153510246e-05, "loss": 0.0634, "step": 49890 }, { "epoch": 4.19, "learning_rate": 1.1619079610345986e-05, "loss": 0.0708, "step": 49900 }, { "epoch": 4.19, "learning_rate": 1.1617400067181727e-05, "loss": 0.0308, "step": 49910 }, { "epoch": 4.19, "learning_rate": 1.1615720524017469e-05, "loss": 0.1529, "step": 49920 }, { "epoch": 4.19, "learning_rate": 1.161404098085321e-05, "loss": 0.0251, "step": 49930 }, { "epoch": 4.19, "learning_rate": 1.161236143768895e-05, "loss": 0.1113, "step": 49940 }, { "epoch": 4.19, "learning_rate": 1.161068189452469e-05, "loss": 0.0392, "step": 49950 }, { "epoch": 4.2, "learning_rate": 1.160900235136043e-05, "loss": 0.0238, "step": 49960 }, { "epoch": 4.2, "learning_rate": 1.1607322808196172e-05, "loss": 0.0193, "step": 49970 }, { "epoch": 4.2, "learning_rate": 1.1605643265031912e-05, "loss": 0.0216, "step": 49980 }, { "epoch": 4.2, "learning_rate": 1.1603963721867652e-05, "loss": 0.0396, "step": 49990 }, { "epoch": 4.2, "learning_rate": 1.1602284178703393e-05, "loss": 0.1312, "step": 50000 }, { "epoch": 4.2, "learning_rate": 1.1600604635539135e-05, "loss": 0.0709, "step": 50010 }, { "epoch": 4.2, "learning_rate": 1.1598925092374876e-05, "loss": 0.0743, "step": 50020 }, { "epoch": 4.2, "learning_rate": 1.1597245549210616e-05, "loss": 0.0022, "step": 50030 }, { "epoch": 4.2, "learning_rate": 1.1595566006046357e-05, "loss": 0.0186, "step": 50040 }, { "epoch": 4.2, "learning_rate": 1.1593886462882099e-05, "loss": 0.0836, "step": 50050 }, { "epoch": 4.2, "learning_rate": 1.1592206919717836e-05, "loss": 0.165, "step": 50060 }, { "epoch": 4.2, "learning_rate": 1.1590527376553578e-05, "loss": 0.1252, "step": 50070 }, { "epoch": 4.21, "learning_rate": 1.1588847833389318e-05, "loss": 0.058, "step": 50080 }, { "epoch": 4.21, "learning_rate": 1.1587168290225059e-05, "loss": 0.0721, "step": 50090 }, { "epoch": 4.21, "learning_rate": 1.15854887470608e-05, "loss": 0.0622, "step": 50100 }, { "epoch": 4.21, "learning_rate": 1.1583809203896542e-05, "loss": 0.0982, "step": 50110 }, { "epoch": 4.21, "learning_rate": 1.1582129660732282e-05, "loss": 0.087, "step": 50120 }, { "epoch": 4.21, "learning_rate": 1.1580450117568023e-05, "loss": 0.0373, "step": 50130 }, { "epoch": 4.21, "learning_rate": 1.1578770574403763e-05, "loss": 0.1368, "step": 50140 }, { "epoch": 4.21, "learning_rate": 1.1577091031239505e-05, "loss": 0.0592, "step": 50150 }, { "epoch": 4.21, "learning_rate": 1.1575411488075246e-05, "loss": 0.0879, "step": 50160 }, { "epoch": 4.21, "learning_rate": 1.1573731944910984e-05, "loss": 0.1232, "step": 50170 }, { "epoch": 4.21, "learning_rate": 1.1572052401746725e-05, "loss": 0.105, "step": 50180 }, { "epoch": 4.21, "learning_rate": 1.1570372858582465e-05, "loss": 0.0542, "step": 50190 }, { "epoch": 4.22, "learning_rate": 1.1568693315418208e-05, "loss": 0.1233, "step": 50200 }, { "epoch": 4.22, "learning_rate": 1.1567013772253948e-05, "loss": 0.037, "step": 50210 }, { "epoch": 4.22, "learning_rate": 1.1565334229089689e-05, "loss": 0.0628, "step": 50220 }, { "epoch": 4.22, "learning_rate": 1.1563654685925429e-05, "loss": 0.0588, "step": 50230 }, { "epoch": 4.22, "learning_rate": 1.1561975142761171e-05, "loss": 0.0694, "step": 50240 }, { "epoch": 4.22, "learning_rate": 1.1560295599596912e-05, "loss": 0.0308, "step": 50250 }, { "epoch": 4.22, "learning_rate": 1.1558616056432652e-05, "loss": 0.048, "step": 50260 }, { "epoch": 4.22, "learning_rate": 1.1556936513268391e-05, "loss": 0.1149, "step": 50270 }, { "epoch": 4.22, "learning_rate": 1.1555256970104131e-05, "loss": 0.0767, "step": 50280 }, { "epoch": 4.22, "learning_rate": 1.1553577426939872e-05, "loss": 0.0232, "step": 50290 }, { "epoch": 4.22, "learning_rate": 1.1551897883775614e-05, "loss": 0.0377, "step": 50300 }, { "epoch": 4.22, "learning_rate": 1.1550218340611355e-05, "loss": 0.0139, "step": 50310 }, { "epoch": 4.23, "learning_rate": 1.1548538797447095e-05, "loss": 0.0781, "step": 50320 }, { "epoch": 4.23, "learning_rate": 1.1546859254282835e-05, "loss": 0.0349, "step": 50330 }, { "epoch": 4.23, "learning_rate": 1.1545179711118578e-05, "loss": 0.0219, "step": 50340 }, { "epoch": 4.23, "learning_rate": 1.1543500167954318e-05, "loss": 0.1067, "step": 50350 }, { "epoch": 4.23, "learning_rate": 1.1541820624790059e-05, "loss": 0.051, "step": 50360 }, { "epoch": 4.23, "learning_rate": 1.1540141081625799e-05, "loss": 0.0407, "step": 50370 }, { "epoch": 4.23, "learning_rate": 1.1538461538461538e-05, "loss": 0.0139, "step": 50380 }, { "epoch": 4.23, "learning_rate": 1.153678199529728e-05, "loss": 0.0578, "step": 50390 }, { "epoch": 4.23, "learning_rate": 1.153510245213302e-05, "loss": 0.0281, "step": 50400 }, { "epoch": 4.23, "learning_rate": 1.1533422908968761e-05, "loss": 0.0007, "step": 50410 }, { "epoch": 4.23, "learning_rate": 1.1531743365804501e-05, "loss": 0.1251, "step": 50420 }, { "epoch": 4.23, "learning_rate": 1.1530063822640244e-05, "loss": 0.1149, "step": 50430 }, { "epoch": 4.24, "learning_rate": 1.1528384279475984e-05, "loss": 0.1022, "step": 50440 }, { "epoch": 4.24, "learning_rate": 1.1526704736311725e-05, "loss": 0.0328, "step": 50450 }, { "epoch": 4.24, "learning_rate": 1.1525025193147465e-05, "loss": 0.053, "step": 50460 }, { "epoch": 4.24, "learning_rate": 1.1523345649983207e-05, "loss": 0.0254, "step": 50470 }, { "epoch": 4.24, "learning_rate": 1.1521666106818948e-05, "loss": 0.0302, "step": 50480 }, { "epoch": 4.24, "learning_rate": 1.1519986563654686e-05, "loss": 0.0743, "step": 50490 }, { "epoch": 4.24, "learning_rate": 1.1518307020490427e-05, "loss": 0.1076, "step": 50500 }, { "epoch": 4.24, "learning_rate": 1.1516627477326167e-05, "loss": 0.0691, "step": 50510 }, { "epoch": 4.24, "learning_rate": 1.1514947934161908e-05, "loss": 0.1192, "step": 50520 }, { "epoch": 4.24, "learning_rate": 1.151326839099765e-05, "loss": 0.016, "step": 50530 }, { "epoch": 4.24, "learning_rate": 1.151158884783339e-05, "loss": 0.0423, "step": 50540 }, { "epoch": 4.25, "learning_rate": 1.1509909304669131e-05, "loss": 0.0592, "step": 50550 }, { "epoch": 4.25, "learning_rate": 1.1508229761504872e-05, "loss": 0.0246, "step": 50560 }, { "epoch": 4.25, "learning_rate": 1.1506550218340614e-05, "loss": 0.0558, "step": 50570 }, { "epoch": 4.25, "learning_rate": 1.1504870675176354e-05, "loss": 0.007, "step": 50580 }, { "epoch": 4.25, "learning_rate": 1.1503191132012093e-05, "loss": 0.0127, "step": 50590 }, { "epoch": 4.25, "learning_rate": 1.1501511588847833e-05, "loss": 0.0916, "step": 50600 }, { "epoch": 4.25, "learning_rate": 1.1499832045683574e-05, "loss": 0.0508, "step": 50610 }, { "epoch": 4.25, "learning_rate": 1.1498152502519316e-05, "loss": 0.1017, "step": 50620 }, { "epoch": 4.25, "learning_rate": 1.1496472959355057e-05, "loss": 0.0825, "step": 50630 }, { "epoch": 4.25, "learning_rate": 1.1494793416190797e-05, "loss": 0.1066, "step": 50640 }, { "epoch": 4.25, "learning_rate": 1.1493113873026537e-05, "loss": 0.0491, "step": 50650 }, { "epoch": 4.25, "learning_rate": 1.1491434329862278e-05, "loss": 0.0358, "step": 50660 }, { "epoch": 4.26, "learning_rate": 1.148975478669802e-05, "loss": 0.0957, "step": 50670 }, { "epoch": 4.26, "learning_rate": 1.148807524353376e-05, "loss": 0.0363, "step": 50680 }, { "epoch": 4.26, "learning_rate": 1.1486395700369501e-05, "loss": 0.0414, "step": 50690 }, { "epoch": 4.26, "learning_rate": 1.148471615720524e-05, "loss": 0.0701, "step": 50700 }, { "epoch": 4.26, "learning_rate": 1.148303661404098e-05, "loss": 0.0411, "step": 50710 }, { "epoch": 4.26, "learning_rate": 1.1481357070876723e-05, "loss": 0.0622, "step": 50720 }, { "epoch": 4.26, "learning_rate": 1.1479677527712463e-05, "loss": 0.0025, "step": 50730 }, { "epoch": 4.26, "learning_rate": 1.1477997984548203e-05, "loss": 0.0989, "step": 50740 }, { "epoch": 4.26, "learning_rate": 1.1476318441383944e-05, "loss": 0.1239, "step": 50750 }, { "epoch": 4.26, "learning_rate": 1.1474638898219686e-05, "loss": 0.0176, "step": 50760 }, { "epoch": 4.26, "learning_rate": 1.1472959355055427e-05, "loss": 0.0865, "step": 50770 }, { "epoch": 4.26, "learning_rate": 1.1471279811891167e-05, "loss": 0.0664, "step": 50780 }, { "epoch": 4.27, "learning_rate": 1.1469600268726908e-05, "loss": 0.0659, "step": 50790 }, { "epoch": 4.27, "learning_rate": 1.1467920725562646e-05, "loss": 0.0583, "step": 50800 }, { "epoch": 4.27, "learning_rate": 1.1466241182398389e-05, "loss": 0.133, "step": 50810 }, { "epoch": 4.27, "learning_rate": 1.1464561639234129e-05, "loss": 0.0526, "step": 50820 }, { "epoch": 4.27, "learning_rate": 1.146288209606987e-05, "loss": 0.0383, "step": 50830 }, { "epoch": 4.27, "learning_rate": 1.146120255290561e-05, "loss": 0.0223, "step": 50840 }, { "epoch": 4.27, "learning_rate": 1.145952300974135e-05, "loss": 0.0209, "step": 50850 }, { "epoch": 4.27, "learning_rate": 1.1457843466577093e-05, "loss": 0.0529, "step": 50860 }, { "epoch": 4.27, "learning_rate": 1.1456163923412833e-05, "loss": 0.0657, "step": 50870 }, { "epoch": 4.27, "learning_rate": 1.1454484380248574e-05, "loss": 0.0846, "step": 50880 }, { "epoch": 4.27, "learning_rate": 1.1452804837084314e-05, "loss": 0.0736, "step": 50890 }, { "epoch": 4.27, "learning_rate": 1.1451125293920056e-05, "loss": 0.0584, "step": 50900 }, { "epoch": 4.28, "learning_rate": 1.1449445750755795e-05, "loss": 0.0318, "step": 50910 }, { "epoch": 4.28, "learning_rate": 1.1447766207591535e-05, "loss": 0.1282, "step": 50920 }, { "epoch": 4.28, "learning_rate": 1.1446086664427276e-05, "loss": 0.093, "step": 50930 }, { "epoch": 4.28, "learning_rate": 1.1444407121263016e-05, "loss": 0.0676, "step": 50940 }, { "epoch": 4.28, "learning_rate": 1.1442727578098759e-05, "loss": 0.0045, "step": 50950 }, { "epoch": 4.28, "learning_rate": 1.1441048034934499e-05, "loss": 0.0364, "step": 50960 }, { "epoch": 4.28, "learning_rate": 1.143936849177024e-05, "loss": 0.0558, "step": 50970 }, { "epoch": 4.28, "learning_rate": 1.143768894860598e-05, "loss": 0.0938, "step": 50980 }, { "epoch": 4.28, "learning_rate": 1.1436009405441722e-05, "loss": 0.0648, "step": 50990 }, { "epoch": 4.28, "learning_rate": 1.1434329862277463e-05, "loss": 0.0166, "step": 51000 }, { "epoch": 4.28, "learning_rate": 1.1432650319113203e-05, "loss": 0.0129, "step": 51010 }, { "epoch": 4.28, "learning_rate": 1.1430970775948942e-05, "loss": 0.1031, "step": 51020 }, { "epoch": 4.29, "learning_rate": 1.1429291232784682e-05, "loss": 0.0279, "step": 51030 }, { "epoch": 4.29, "learning_rate": 1.1427611689620423e-05, "loss": 0.0659, "step": 51040 }, { "epoch": 4.29, "learning_rate": 1.1425932146456165e-05, "loss": 0.0656, "step": 51050 }, { "epoch": 4.29, "learning_rate": 1.1424252603291906e-05, "loss": 0.0416, "step": 51060 }, { "epoch": 4.29, "learning_rate": 1.1422573060127646e-05, "loss": 0.0372, "step": 51070 }, { "epoch": 4.29, "learning_rate": 1.1420893516963386e-05, "loss": 0.1378, "step": 51080 }, { "epoch": 4.29, "learning_rate": 1.1419213973799129e-05, "loss": 0.1311, "step": 51090 }, { "epoch": 4.29, "learning_rate": 1.1417534430634869e-05, "loss": 0.0073, "step": 51100 }, { "epoch": 4.29, "learning_rate": 1.141585488747061e-05, "loss": 0.0986, "step": 51110 }, { "epoch": 4.29, "learning_rate": 1.1414175344306348e-05, "loss": 0.0715, "step": 51120 }, { "epoch": 4.29, "learning_rate": 1.1412495801142089e-05, "loss": 0.0354, "step": 51130 }, { "epoch": 4.29, "learning_rate": 1.1410816257977831e-05, "loss": 0.1235, "step": 51140 }, { "epoch": 4.3, "learning_rate": 1.1409136714813571e-05, "loss": 0.0328, "step": 51150 }, { "epoch": 4.3, "learning_rate": 1.1407457171649312e-05, "loss": 0.0562, "step": 51160 }, { "epoch": 4.3, "learning_rate": 1.1405777628485052e-05, "loss": 0.0898, "step": 51170 }, { "epoch": 4.3, "learning_rate": 1.1404098085320795e-05, "loss": 0.0297, "step": 51180 }, { "epoch": 4.3, "learning_rate": 1.1402418542156535e-05, "loss": 0.0507, "step": 51190 }, { "epoch": 4.3, "learning_rate": 1.1400738998992276e-05, "loss": 0.0289, "step": 51200 }, { "epoch": 4.3, "learning_rate": 1.1399059455828016e-05, "loss": 0.0615, "step": 51210 }, { "epoch": 4.3, "learning_rate": 1.1397379912663758e-05, "loss": 0.0866, "step": 51220 }, { "epoch": 4.3, "learning_rate": 1.1395700369499495e-05, "loss": 0.0581, "step": 51230 }, { "epoch": 4.3, "learning_rate": 1.1394020826335237e-05, "loss": 0.0272, "step": 51240 }, { "epoch": 4.3, "learning_rate": 1.1392341283170978e-05, "loss": 0.0474, "step": 51250 }, { "epoch": 4.3, "learning_rate": 1.1390661740006718e-05, "loss": 0.086, "step": 51260 }, { "epoch": 4.31, "learning_rate": 1.1388982196842459e-05, "loss": 0.1531, "step": 51270 }, { "epoch": 4.31, "learning_rate": 1.1387302653678201e-05, "loss": 0.0562, "step": 51280 }, { "epoch": 4.31, "learning_rate": 1.1385623110513942e-05, "loss": 0.0556, "step": 51290 }, { "epoch": 4.31, "learning_rate": 1.1383943567349682e-05, "loss": 0.0581, "step": 51300 }, { "epoch": 4.31, "learning_rate": 1.1382264024185423e-05, "loss": 0.0123, "step": 51310 }, { "epoch": 4.31, "learning_rate": 1.1380584481021165e-05, "loss": 0.0413, "step": 51320 }, { "epoch": 4.31, "learning_rate": 1.1378904937856903e-05, "loss": 0.0398, "step": 51330 }, { "epoch": 4.31, "learning_rate": 1.1377225394692644e-05, "loss": 0.0778, "step": 51340 }, { "epoch": 4.31, "learning_rate": 1.1375545851528384e-05, "loss": 0.0169, "step": 51350 }, { "epoch": 4.31, "learning_rate": 1.1373866308364125e-05, "loss": 0.0791, "step": 51360 }, { "epoch": 4.31, "learning_rate": 1.1372186765199867e-05, "loss": 0.104, "step": 51370 }, { "epoch": 4.31, "learning_rate": 1.1370507222035608e-05, "loss": 0.0021, "step": 51380 }, { "epoch": 4.32, "learning_rate": 1.1368827678871348e-05, "loss": 0.0213, "step": 51390 }, { "epoch": 4.32, "learning_rate": 1.1367148135707088e-05, "loss": 0.0323, "step": 51400 }, { "epoch": 4.32, "learning_rate": 1.136546859254283e-05, "loss": 0.1478, "step": 51410 }, { "epoch": 4.32, "learning_rate": 1.1363789049378571e-05, "loss": 0.0296, "step": 51420 }, { "epoch": 4.32, "learning_rate": 1.1362109506214312e-05, "loss": 0.0312, "step": 51430 }, { "epoch": 4.32, "learning_rate": 1.136042996305005e-05, "loss": 0.0867, "step": 51440 }, { "epoch": 4.32, "learning_rate": 1.1358750419885791e-05, "loss": 0.0294, "step": 51450 }, { "epoch": 4.32, "learning_rate": 1.1357070876721531e-05, "loss": 0.1162, "step": 51460 }, { "epoch": 4.32, "learning_rate": 1.1355391333557274e-05, "loss": 0.0444, "step": 51470 }, { "epoch": 4.32, "learning_rate": 1.1353711790393014e-05, "loss": 0.034, "step": 51480 }, { "epoch": 4.32, "learning_rate": 1.1352032247228754e-05, "loss": 0.0503, "step": 51490 }, { "epoch": 4.32, "learning_rate": 1.1350352704064495e-05, "loss": 0.0545, "step": 51500 }, { "epoch": 4.33, "learning_rate": 1.1348673160900237e-05, "loss": 0.0352, "step": 51510 }, { "epoch": 4.33, "learning_rate": 1.1346993617735978e-05, "loss": 0.0338, "step": 51520 }, { "epoch": 4.33, "learning_rate": 1.1345314074571718e-05, "loss": 0.0964, "step": 51530 }, { "epoch": 4.33, "learning_rate": 1.1343634531407459e-05, "loss": 0.0779, "step": 51540 }, { "epoch": 4.33, "learning_rate": 1.1341954988243197e-05, "loss": 0.0649, "step": 51550 }, { "epoch": 4.33, "learning_rate": 1.134027544507894e-05, "loss": 0.039, "step": 51560 }, { "epoch": 4.33, "learning_rate": 1.133859590191468e-05, "loss": 0.0702, "step": 51570 }, { "epoch": 4.33, "learning_rate": 1.133691635875042e-05, "loss": 0.0569, "step": 51580 }, { "epoch": 4.33, "learning_rate": 1.1335236815586161e-05, "loss": 0.0487, "step": 51590 }, { "epoch": 4.33, "learning_rate": 1.1333557272421903e-05, "loss": 0.0821, "step": 51600 }, { "epoch": 4.33, "learning_rate": 1.1331877729257644e-05, "loss": 0.0878, "step": 51610 }, { "epoch": 4.33, "learning_rate": 1.1330198186093384e-05, "loss": 0.0229, "step": 51620 }, { "epoch": 4.34, "learning_rate": 1.1328518642929125e-05, "loss": 0.0355, "step": 51630 }, { "epoch": 4.34, "learning_rate": 1.1326839099764865e-05, "loss": 0.0437, "step": 51640 }, { "epoch": 4.34, "learning_rate": 1.1325159556600604e-05, "loss": 0.0036, "step": 51650 }, { "epoch": 4.34, "learning_rate": 1.1323480013436346e-05, "loss": 0.0193, "step": 51660 }, { "epoch": 4.34, "learning_rate": 1.1321800470272086e-05, "loss": 0.1326, "step": 51670 }, { "epoch": 4.34, "learning_rate": 1.1320120927107827e-05, "loss": 0.0535, "step": 51680 }, { "epoch": 4.34, "learning_rate": 1.1318441383943567e-05, "loss": 0.04, "step": 51690 }, { "epoch": 4.34, "learning_rate": 1.131676184077931e-05, "loss": 0.0682, "step": 51700 }, { "epoch": 4.34, "learning_rate": 1.131508229761505e-05, "loss": 0.0457, "step": 51710 }, { "epoch": 4.34, "learning_rate": 1.131340275445079e-05, "loss": 0.1004, "step": 51720 }, { "epoch": 4.34, "learning_rate": 1.1311723211286531e-05, "loss": 0.1271, "step": 51730 }, { "epoch": 4.34, "learning_rate": 1.1310043668122273e-05, "loss": 0.0983, "step": 51740 }, { "epoch": 4.35, "learning_rate": 1.1308364124958014e-05, "loss": 0.0074, "step": 51750 }, { "epoch": 4.35, "learning_rate": 1.1306684581793752e-05, "loss": 0.0718, "step": 51760 }, { "epoch": 4.35, "learning_rate": 1.1305005038629493e-05, "loss": 0.0032, "step": 51770 }, { "epoch": 4.35, "learning_rate": 1.1303325495465233e-05, "loss": 0.1054, "step": 51780 }, { "epoch": 4.35, "learning_rate": 1.1301645952300976e-05, "loss": 0.099, "step": 51790 }, { "epoch": 4.35, "learning_rate": 1.1299966409136716e-05, "loss": 0.124, "step": 51800 }, { "epoch": 4.35, "learning_rate": 1.1298286865972456e-05, "loss": 0.116, "step": 51810 }, { "epoch": 4.35, "learning_rate": 1.1296607322808197e-05, "loss": 0.056, "step": 51820 }, { "epoch": 4.35, "learning_rate": 1.1294927779643937e-05, "loss": 0.0349, "step": 51830 }, { "epoch": 4.35, "learning_rate": 1.129324823647968e-05, "loss": 0.0671, "step": 51840 }, { "epoch": 4.35, "learning_rate": 1.129156869331542e-05, "loss": 0.04, "step": 51850 }, { "epoch": 4.36, "learning_rate": 1.1289889150151159e-05, "loss": 0.0435, "step": 51860 }, { "epoch": 4.36, "learning_rate": 1.12882096069869e-05, "loss": 0.1619, "step": 51870 }, { "epoch": 4.36, "learning_rate": 1.128653006382264e-05, "loss": 0.027, "step": 51880 }, { "epoch": 4.36, "learning_rate": 1.1284850520658382e-05, "loss": 0.0759, "step": 51890 }, { "epoch": 4.36, "learning_rate": 1.1283170977494122e-05, "loss": 0.1111, "step": 51900 }, { "epoch": 4.36, "learning_rate": 1.1281491434329863e-05, "loss": 0.0299, "step": 51910 }, { "epoch": 4.36, "learning_rate": 1.1279811891165603e-05, "loss": 0.1104, "step": 51920 }, { "epoch": 4.36, "learning_rate": 1.1278132348001346e-05, "loss": 0.047, "step": 51930 }, { "epoch": 4.36, "learning_rate": 1.1276452804837086e-05, "loss": 0.0784, "step": 51940 }, { "epoch": 4.36, "learning_rate": 1.1274773261672827e-05, "loss": 0.043, "step": 51950 }, { "epoch": 4.36, "learning_rate": 1.1273093718508567e-05, "loss": 0.101, "step": 51960 }, { "epoch": 4.36, "learning_rate": 1.1271414175344306e-05, "loss": 0.0736, "step": 51970 }, { "epoch": 4.37, "learning_rate": 1.1269734632180048e-05, "loss": 0.0952, "step": 51980 }, { "epoch": 4.37, "learning_rate": 1.1268055089015788e-05, "loss": 0.0689, "step": 51990 }, { "epoch": 4.37, "learning_rate": 1.1266375545851529e-05, "loss": 0.0436, "step": 52000 }, { "epoch": 4.37, "learning_rate": 1.126469600268727e-05, "loss": 0.0918, "step": 52010 }, { "epoch": 4.37, "learning_rate": 1.126301645952301e-05, "loss": 0.0377, "step": 52020 }, { "epoch": 4.37, "learning_rate": 1.1261336916358752e-05, "loss": 0.0803, "step": 52030 }, { "epoch": 4.37, "learning_rate": 1.1259657373194493e-05, "loss": 0.0668, "step": 52040 }, { "epoch": 4.37, "learning_rate": 1.1257977830030233e-05, "loss": 0.0104, "step": 52050 }, { "epoch": 4.37, "learning_rate": 1.1256298286865973e-05, "loss": 0.0393, "step": 52060 }, { "epoch": 4.37, "learning_rate": 1.1254618743701716e-05, "loss": 0.018, "step": 52070 }, { "epoch": 4.37, "learning_rate": 1.1252939200537454e-05, "loss": 0.0049, "step": 52080 }, { "epoch": 4.37, "learning_rate": 1.1251259657373195e-05, "loss": 0.0302, "step": 52090 }, { "epoch": 4.38, "learning_rate": 1.1249580114208935e-05, "loss": 0.0466, "step": 52100 }, { "epoch": 4.38, "learning_rate": 1.1247900571044676e-05, "loss": 0.0492, "step": 52110 }, { "epoch": 4.38, "learning_rate": 1.1246221027880418e-05, "loss": 0.1105, "step": 52120 }, { "epoch": 4.38, "learning_rate": 1.1244541484716159e-05, "loss": 0.1238, "step": 52130 }, { "epoch": 4.38, "learning_rate": 1.1242861941551899e-05, "loss": 0.1286, "step": 52140 }, { "epoch": 4.38, "learning_rate": 1.124118239838764e-05, "loss": 0.0444, "step": 52150 }, { "epoch": 4.38, "learning_rate": 1.1239502855223382e-05, "loss": 0.0754, "step": 52160 }, { "epoch": 4.38, "learning_rate": 1.1237823312059122e-05, "loss": 0.049, "step": 52170 }, { "epoch": 4.38, "learning_rate": 1.1236143768894861e-05, "loss": 0.078, "step": 52180 }, { "epoch": 4.38, "learning_rate": 1.1234464225730601e-05, "loss": 0.051, "step": 52190 }, { "epoch": 4.38, "learning_rate": 1.1232784682566342e-05, "loss": 0.0125, "step": 52200 }, { "epoch": 4.38, "learning_rate": 1.1231105139402082e-05, "loss": 0.076, "step": 52210 }, { "epoch": 4.39, "learning_rate": 1.1229425596237825e-05, "loss": 0.0306, "step": 52220 }, { "epoch": 4.39, "learning_rate": 1.1227746053073565e-05, "loss": 0.0753, "step": 52230 }, { "epoch": 4.39, "learning_rate": 1.1226066509909305e-05, "loss": 0.0407, "step": 52240 }, { "epoch": 4.39, "learning_rate": 1.1224386966745046e-05, "loss": 0.1134, "step": 52250 }, { "epoch": 4.39, "learning_rate": 1.1222707423580788e-05, "loss": 0.0374, "step": 52260 }, { "epoch": 4.39, "learning_rate": 1.1221027880416529e-05, "loss": 0.0658, "step": 52270 }, { "epoch": 4.39, "learning_rate": 1.1219348337252269e-05, "loss": 0.0635, "step": 52280 }, { "epoch": 4.39, "learning_rate": 1.1217668794088008e-05, "loss": 0.0613, "step": 52290 }, { "epoch": 4.39, "learning_rate": 1.1215989250923748e-05, "loss": 0.0529, "step": 52300 }, { "epoch": 4.39, "learning_rate": 1.121430970775949e-05, "loss": 0.0602, "step": 52310 }, { "epoch": 4.39, "learning_rate": 1.1212630164595231e-05, "loss": 0.03, "step": 52320 }, { "epoch": 4.39, "learning_rate": 1.1210950621430971e-05, "loss": 0.0609, "step": 52330 }, { "epoch": 4.4, "learning_rate": 1.1209271078266712e-05, "loss": 0.0583, "step": 52340 }, { "epoch": 4.4, "learning_rate": 1.1207591535102454e-05, "loss": 0.1262, "step": 52350 }, { "epoch": 4.4, "learning_rate": 1.1205911991938195e-05, "loss": 0.0688, "step": 52360 }, { "epoch": 4.4, "learning_rate": 1.1204232448773935e-05, "loss": 0.0211, "step": 52370 }, { "epoch": 4.4, "learning_rate": 1.1202552905609676e-05, "loss": 0.0172, "step": 52380 }, { "epoch": 4.4, "learning_rate": 1.1200873362445414e-05, "loss": 0.0663, "step": 52390 }, { "epoch": 4.4, "learning_rate": 1.1199193819281155e-05, "loss": 0.1009, "step": 52400 }, { "epoch": 4.4, "learning_rate": 1.1197514276116897e-05, "loss": 0.0685, "step": 52410 }, { "epoch": 4.4, "learning_rate": 1.1195834732952637e-05, "loss": 0.0329, "step": 52420 }, { "epoch": 4.4, "learning_rate": 1.1194155189788378e-05, "loss": 0.0664, "step": 52430 }, { "epoch": 4.4, "learning_rate": 1.1192475646624118e-05, "loss": 0.0718, "step": 52440 }, { "epoch": 4.4, "learning_rate": 1.119079610345986e-05, "loss": 0.0838, "step": 52450 }, { "epoch": 4.41, "learning_rate": 1.1189116560295601e-05, "loss": 0.0802, "step": 52460 }, { "epoch": 4.41, "learning_rate": 1.1187437017131342e-05, "loss": 0.0551, "step": 52470 }, { "epoch": 4.41, "learning_rate": 1.1185757473967082e-05, "loss": 0.0487, "step": 52480 }, { "epoch": 4.41, "learning_rate": 1.1184077930802824e-05, "loss": 0.1147, "step": 52490 }, { "epoch": 4.41, "learning_rate": 1.1182398387638563e-05, "loss": 0.1228, "step": 52500 }, { "epoch": 4.41, "learning_rate": 1.1180718844474303e-05, "loss": 0.0743, "step": 52510 }, { "epoch": 4.41, "learning_rate": 1.1179039301310044e-05, "loss": 0.0378, "step": 52520 }, { "epoch": 4.41, "learning_rate": 1.1177359758145784e-05, "loss": 0.0775, "step": 52530 }, { "epoch": 4.41, "learning_rate": 1.1175680214981527e-05, "loss": 0.0575, "step": 52540 }, { "epoch": 4.41, "learning_rate": 1.1174000671817267e-05, "loss": 0.068, "step": 52550 }, { "epoch": 4.41, "learning_rate": 1.1172321128653007e-05, "loss": 0.0996, "step": 52560 }, { "epoch": 4.41, "learning_rate": 1.1170641585488748e-05, "loss": 0.0282, "step": 52570 }, { "epoch": 4.42, "learning_rate": 1.116896204232449e-05, "loss": 0.0425, "step": 52580 }, { "epoch": 4.42, "learning_rate": 1.116728249916023e-05, "loss": 0.0546, "step": 52590 }, { "epoch": 4.42, "learning_rate": 1.1165602955995971e-05, "loss": 0.0862, "step": 52600 }, { "epoch": 4.42, "learning_rate": 1.116392341283171e-05, "loss": 0.0664, "step": 52610 }, { "epoch": 4.42, "learning_rate": 1.116224386966745e-05, "loss": 0.1176, "step": 52620 }, { "epoch": 4.42, "learning_rate": 1.116056432650319e-05, "loss": 0.0598, "step": 52630 }, { "epoch": 4.42, "learning_rate": 1.1158884783338933e-05, "loss": 0.0467, "step": 52640 }, { "epoch": 4.42, "learning_rate": 1.1157205240174673e-05, "loss": 0.1258, "step": 52650 }, { "epoch": 4.42, "learning_rate": 1.1155525697010414e-05, "loss": 0.1027, "step": 52660 }, { "epoch": 4.42, "learning_rate": 1.1153846153846154e-05, "loss": 0.0172, "step": 52670 }, { "epoch": 4.42, "learning_rate": 1.1152166610681897e-05, "loss": 0.0394, "step": 52680 }, { "epoch": 4.42, "learning_rate": 1.1150487067517637e-05, "loss": 0.0847, "step": 52690 }, { "epoch": 4.43, "learning_rate": 1.1148807524353378e-05, "loss": 0.0336, "step": 52700 }, { "epoch": 4.43, "learning_rate": 1.1147127981189116e-05, "loss": 0.0232, "step": 52710 }, { "epoch": 4.43, "learning_rate": 1.1145448438024857e-05, "loss": 0.0282, "step": 52720 }, { "epoch": 4.43, "learning_rate": 1.1143768894860599e-05, "loss": 0.0596, "step": 52730 }, { "epoch": 4.43, "learning_rate": 1.114208935169634e-05, "loss": 0.0938, "step": 52740 }, { "epoch": 4.43, "learning_rate": 1.114040980853208e-05, "loss": 0.1391, "step": 52750 }, { "epoch": 4.43, "learning_rate": 1.113873026536782e-05, "loss": 0.1418, "step": 52760 }, { "epoch": 4.43, "learning_rate": 1.1137050722203563e-05, "loss": 0.1177, "step": 52770 }, { "epoch": 4.43, "learning_rate": 1.1135371179039303e-05, "loss": 0.0237, "step": 52780 }, { "epoch": 4.43, "learning_rate": 1.1133691635875044e-05, "loss": 0.0325, "step": 52790 }, { "epoch": 4.43, "learning_rate": 1.1132012092710784e-05, "loss": 0.0434, "step": 52800 }, { "epoch": 4.43, "learning_rate": 1.1130332549546524e-05, "loss": 0.0245, "step": 52810 }, { "epoch": 4.44, "learning_rate": 1.1128653006382263e-05, "loss": 0.0561, "step": 52820 }, { "epoch": 4.44, "learning_rate": 1.1126973463218005e-05, "loss": 0.0128, "step": 52830 }, { "epoch": 4.44, "learning_rate": 1.1125293920053746e-05, "loss": 0.0429, "step": 52840 }, { "epoch": 4.44, "learning_rate": 1.1123614376889486e-05, "loss": 0.0702, "step": 52850 }, { "epoch": 4.44, "learning_rate": 1.1121934833725227e-05, "loss": 0.0755, "step": 52860 }, { "epoch": 4.44, "learning_rate": 1.1120255290560969e-05, "loss": 0.0502, "step": 52870 }, { "epoch": 4.44, "learning_rate": 1.111857574739671e-05, "loss": 0.0595, "step": 52880 }, { "epoch": 4.44, "learning_rate": 1.111689620423245e-05, "loss": 0.1526, "step": 52890 }, { "epoch": 4.44, "learning_rate": 1.111521666106819e-05, "loss": 0.1182, "step": 52900 }, { "epoch": 4.44, "learning_rate": 1.1113537117903933e-05, "loss": 0.15, "step": 52910 }, { "epoch": 4.44, "learning_rate": 1.1111857574739671e-05, "loss": 0.0626, "step": 52920 }, { "epoch": 4.44, "learning_rate": 1.1110178031575412e-05, "loss": 0.0881, "step": 52930 }, { "epoch": 4.45, "learning_rate": 1.1108498488411152e-05, "loss": 0.066, "step": 52940 }, { "epoch": 4.45, "learning_rate": 1.1106818945246893e-05, "loss": 0.0886, "step": 52950 }, { "epoch": 4.45, "learning_rate": 1.1105139402082635e-05, "loss": 0.033, "step": 52960 }, { "epoch": 4.45, "learning_rate": 1.1103459858918376e-05, "loss": 0.0036, "step": 52970 }, { "epoch": 4.45, "learning_rate": 1.1101780315754116e-05, "loss": 0.0332, "step": 52980 }, { "epoch": 4.45, "learning_rate": 1.1100100772589856e-05, "loss": 0.0636, "step": 52990 }, { "epoch": 4.45, "learning_rate": 1.1098421229425597e-05, "loss": 0.0366, "step": 53000 }, { "epoch": 4.45, "learning_rate": 1.1096741686261339e-05, "loss": 0.1146, "step": 53010 }, { "epoch": 4.45, "learning_rate": 1.109506214309708e-05, "loss": 0.0384, "step": 53020 }, { "epoch": 4.45, "learning_rate": 1.1093382599932818e-05, "loss": 0.0606, "step": 53030 }, { "epoch": 4.45, "learning_rate": 1.1091703056768559e-05, "loss": 0.0907, "step": 53040 }, { "epoch": 4.45, "learning_rate": 1.10900235136043e-05, "loss": 0.0267, "step": 53050 }, { "epoch": 4.46, "learning_rate": 1.1088343970440041e-05, "loss": 0.11, "step": 53060 }, { "epoch": 4.46, "learning_rate": 1.1086664427275782e-05, "loss": 0.0629, "step": 53070 }, { "epoch": 4.46, "learning_rate": 1.1084984884111522e-05, "loss": 0.0379, "step": 53080 }, { "epoch": 4.46, "learning_rate": 1.1083305340947263e-05, "loss": 0.0183, "step": 53090 }, { "epoch": 4.46, "learning_rate": 1.1081625797783005e-05, "loss": 0.0258, "step": 53100 }, { "epoch": 4.46, "learning_rate": 1.1079946254618746e-05, "loss": 0.1195, "step": 53110 }, { "epoch": 4.46, "learning_rate": 1.1078266711454486e-05, "loss": 0.0172, "step": 53120 }, { "epoch": 4.46, "learning_rate": 1.1076587168290227e-05, "loss": 0.1105, "step": 53130 }, { "epoch": 4.46, "learning_rate": 1.1074907625125965e-05, "loss": 0.0902, "step": 53140 }, { "epoch": 4.46, "learning_rate": 1.1073228081961706e-05, "loss": 0.028, "step": 53150 }, { "epoch": 4.46, "learning_rate": 1.1071548538797448e-05, "loss": 0.0009, "step": 53160 }, { "epoch": 4.47, "learning_rate": 1.1069868995633188e-05, "loss": 0.1033, "step": 53170 }, { "epoch": 4.47, "learning_rate": 1.1068189452468929e-05, "loss": 0.0946, "step": 53180 }, { "epoch": 4.47, "learning_rate": 1.106650990930467e-05, "loss": 0.1177, "step": 53190 }, { "epoch": 4.47, "learning_rate": 1.1064830366140412e-05, "loss": 0.0633, "step": 53200 }, { "epoch": 4.47, "learning_rate": 1.1063150822976152e-05, "loss": 0.0209, "step": 53210 }, { "epoch": 4.47, "learning_rate": 1.1061471279811893e-05, "loss": 0.0342, "step": 53220 }, { "epoch": 4.47, "learning_rate": 1.1059791736647633e-05, "loss": 0.1167, "step": 53230 }, { "epoch": 4.47, "learning_rate": 1.1058112193483372e-05, "loss": 0.0529, "step": 53240 }, { "epoch": 4.47, "learning_rate": 1.1056432650319114e-05, "loss": 0.0306, "step": 53250 }, { "epoch": 4.47, "learning_rate": 1.1054753107154854e-05, "loss": 0.0367, "step": 53260 }, { "epoch": 4.47, "learning_rate": 1.1053073563990595e-05, "loss": 0.1239, "step": 53270 }, { "epoch": 4.47, "learning_rate": 1.1051394020826335e-05, "loss": 0.1008, "step": 53280 }, { "epoch": 4.48, "learning_rate": 1.1049714477662078e-05, "loss": 0.0577, "step": 53290 }, { "epoch": 4.48, "learning_rate": 1.1048034934497818e-05, "loss": 0.0268, "step": 53300 }, { "epoch": 4.48, "learning_rate": 1.1046355391333558e-05, "loss": 0.0767, "step": 53310 }, { "epoch": 4.48, "learning_rate": 1.1044675848169299e-05, "loss": 0.0841, "step": 53320 }, { "epoch": 4.48, "learning_rate": 1.1042996305005041e-05, "loss": 0.0393, "step": 53330 }, { "epoch": 4.48, "learning_rate": 1.1041316761840782e-05, "loss": 0.0427, "step": 53340 }, { "epoch": 4.48, "learning_rate": 1.103963721867652e-05, "loss": 0.0585, "step": 53350 }, { "epoch": 4.48, "learning_rate": 1.1037957675512261e-05, "loss": 0.0478, "step": 53360 }, { "epoch": 4.48, "learning_rate": 1.1036278132348001e-05, "loss": 0.0571, "step": 53370 }, { "epoch": 4.48, "learning_rate": 1.1034598589183742e-05, "loss": 0.0626, "step": 53380 }, { "epoch": 4.48, "learning_rate": 1.1032919046019484e-05, "loss": 0.0582, "step": 53390 }, { "epoch": 4.48, "learning_rate": 1.1031239502855224e-05, "loss": 0.1033, "step": 53400 }, { "epoch": 4.49, "learning_rate": 1.1029559959690965e-05, "loss": 0.0133, "step": 53410 }, { "epoch": 4.49, "learning_rate": 1.1027880416526705e-05, "loss": 0.064, "step": 53420 }, { "epoch": 4.49, "learning_rate": 1.1026200873362448e-05, "loss": 0.1548, "step": 53430 }, { "epoch": 4.49, "learning_rate": 1.1024521330198188e-05, "loss": 0.1492, "step": 53440 }, { "epoch": 4.49, "learning_rate": 1.1022841787033927e-05, "loss": 0.0287, "step": 53450 }, { "epoch": 4.49, "learning_rate": 1.1021162243869667e-05, "loss": 0.038, "step": 53460 }, { "epoch": 4.49, "learning_rate": 1.1019482700705408e-05, "loss": 0.0327, "step": 53470 }, { "epoch": 4.49, "learning_rate": 1.101780315754115e-05, "loss": 0.0161, "step": 53480 }, { "epoch": 4.49, "learning_rate": 1.101612361437689e-05, "loss": 0.1294, "step": 53490 }, { "epoch": 4.49, "learning_rate": 1.1014444071212631e-05, "loss": 0.0408, "step": 53500 }, { "epoch": 4.49, "learning_rate": 1.1012764528048371e-05, "loss": 0.0606, "step": 53510 }, { "epoch": 4.49, "learning_rate": 1.1011084984884114e-05, "loss": 0.073, "step": 53520 }, { "epoch": 4.5, "learning_rate": 1.1009405441719854e-05, "loss": 0.1167, "step": 53530 }, { "epoch": 4.5, "learning_rate": 1.1007725898555595e-05, "loss": 0.0896, "step": 53540 }, { "epoch": 4.5, "learning_rate": 1.1006046355391335e-05, "loss": 0.1209, "step": 53550 }, { "epoch": 4.5, "learning_rate": 1.1004366812227074e-05, "loss": 0.0485, "step": 53560 }, { "epoch": 4.5, "learning_rate": 1.1002687269062814e-05, "loss": 0.0201, "step": 53570 }, { "epoch": 4.5, "learning_rate": 1.1001007725898556e-05, "loss": 0.1158, "step": 53580 }, { "epoch": 4.5, "learning_rate": 1.0999328182734297e-05, "loss": 0.0234, "step": 53590 }, { "epoch": 4.5, "learning_rate": 1.0997648639570037e-05, "loss": 0.0539, "step": 53600 }, { "epoch": 4.5, "learning_rate": 1.0995969096405778e-05, "loss": 0.0962, "step": 53610 }, { "epoch": 4.5, "learning_rate": 1.099428955324152e-05, "loss": 0.1182, "step": 53620 }, { "epoch": 4.5, "learning_rate": 1.099261001007726e-05, "loss": 0.0819, "step": 53630 }, { "epoch": 4.5, "learning_rate": 1.0990930466913001e-05, "loss": 0.1121, "step": 53640 }, { "epoch": 4.51, "learning_rate": 1.0989250923748741e-05, "loss": 0.0494, "step": 53650 }, { "epoch": 4.51, "learning_rate": 1.0987571380584484e-05, "loss": 0.0057, "step": 53660 }, { "epoch": 4.51, "learning_rate": 1.0985891837420222e-05, "loss": 0.0296, "step": 53670 }, { "epoch": 4.51, "learning_rate": 1.0984212294255963e-05, "loss": 0.052, "step": 53680 }, { "epoch": 4.51, "learning_rate": 1.0982532751091703e-05, "loss": 0.0212, "step": 53690 }, { "epoch": 4.51, "learning_rate": 1.0980853207927444e-05, "loss": 0.001, "step": 53700 }, { "epoch": 4.51, "learning_rate": 1.0979173664763186e-05, "loss": 0.1072, "step": 53710 }, { "epoch": 4.51, "learning_rate": 1.0977494121598927e-05, "loss": 0.0378, "step": 53720 }, { "epoch": 4.51, "learning_rate": 1.0975814578434667e-05, "loss": 0.1051, "step": 53730 }, { "epoch": 4.51, "learning_rate": 1.0974135035270407e-05, "loss": 0.0252, "step": 53740 }, { "epoch": 4.51, "learning_rate": 1.097245549210615e-05, "loss": 0.0823, "step": 53750 }, { "epoch": 4.51, "learning_rate": 1.097077594894189e-05, "loss": 0.1296, "step": 53760 }, { "epoch": 4.52, "learning_rate": 1.0969096405777629e-05, "loss": 0.1626, "step": 53770 }, { "epoch": 4.52, "learning_rate": 1.096741686261337e-05, "loss": 0.0818, "step": 53780 }, { "epoch": 4.52, "learning_rate": 1.096573731944911e-05, "loss": 0.025, "step": 53790 }, { "epoch": 4.52, "learning_rate": 1.096405777628485e-05, "loss": 0.0616, "step": 53800 }, { "epoch": 4.52, "learning_rate": 1.0962378233120592e-05, "loss": 0.0864, "step": 53810 }, { "epoch": 4.52, "learning_rate": 1.0960698689956333e-05, "loss": 0.0648, "step": 53820 }, { "epoch": 4.52, "learning_rate": 1.0959019146792073e-05, "loss": 0.155, "step": 53830 }, { "epoch": 4.52, "learning_rate": 1.0957339603627814e-05, "loss": 0.0561, "step": 53840 }, { "epoch": 4.52, "learning_rate": 1.0955660060463556e-05, "loss": 0.1029, "step": 53850 }, { "epoch": 4.52, "learning_rate": 1.0953980517299297e-05, "loss": 0.0014, "step": 53860 }, { "epoch": 4.52, "learning_rate": 1.0952300974135037e-05, "loss": 0.1383, "step": 53870 }, { "epoch": 4.52, "learning_rate": 1.0950621430970776e-05, "loss": 0.0972, "step": 53880 }, { "epoch": 4.53, "learning_rate": 1.0948941887806516e-05, "loss": 0.0987, "step": 53890 }, { "epoch": 4.53, "learning_rate": 1.0947262344642258e-05, "loss": 0.0402, "step": 53900 }, { "epoch": 4.53, "learning_rate": 1.0945582801477999e-05, "loss": 0.0385, "step": 53910 }, { "epoch": 4.53, "learning_rate": 1.094390325831374e-05, "loss": 0.1013, "step": 53920 }, { "epoch": 4.53, "learning_rate": 1.094222371514948e-05, "loss": 0.0249, "step": 53930 }, { "epoch": 4.53, "learning_rate": 1.0940544171985222e-05, "loss": 0.0225, "step": 53940 }, { "epoch": 4.53, "learning_rate": 1.0938864628820963e-05, "loss": 0.0838, "step": 53950 }, { "epoch": 4.53, "learning_rate": 1.0937185085656703e-05, "loss": 0.1613, "step": 53960 }, { "epoch": 4.53, "learning_rate": 1.0935505542492444e-05, "loss": 0.1334, "step": 53970 }, { "epoch": 4.53, "learning_rate": 1.0933825999328182e-05, "loss": 0.0785, "step": 53980 }, { "epoch": 4.53, "learning_rate": 1.0932146456163923e-05, "loss": 0.102, "step": 53990 }, { "epoch": 4.53, "learning_rate": 1.0930466912999665e-05, "loss": 0.0796, "step": 54000 }, { "epoch": 4.54, "learning_rate": 1.0928787369835405e-05, "loss": 0.0939, "step": 54010 }, { "epoch": 4.54, "learning_rate": 1.0927107826671146e-05, "loss": 0.0034, "step": 54020 }, { "epoch": 4.54, "learning_rate": 1.0925428283506886e-05, "loss": 0.0414, "step": 54030 }, { "epoch": 4.54, "learning_rate": 1.0923748740342629e-05, "loss": 0.0179, "step": 54040 }, { "epoch": 4.54, "learning_rate": 1.0922069197178369e-05, "loss": 0.0424, "step": 54050 }, { "epoch": 4.54, "learning_rate": 1.092038965401411e-05, "loss": 0.0915, "step": 54060 }, { "epoch": 4.54, "learning_rate": 1.091871011084985e-05, "loss": 0.1294, "step": 54070 }, { "epoch": 4.54, "learning_rate": 1.0917030567685592e-05, "loss": 0.0369, "step": 54080 }, { "epoch": 4.54, "learning_rate": 1.0915351024521331e-05, "loss": 0.0761, "step": 54090 }, { "epoch": 4.54, "learning_rate": 1.0913671481357071e-05, "loss": 0.0676, "step": 54100 }, { "epoch": 4.54, "learning_rate": 1.0911991938192812e-05, "loss": 0.0033, "step": 54110 }, { "epoch": 4.54, "learning_rate": 1.0910312395028552e-05, "loss": 0.106, "step": 54120 }, { "epoch": 4.55, "learning_rate": 1.0908632851864293e-05, "loss": 0.0855, "step": 54130 }, { "epoch": 4.55, "learning_rate": 1.0906953308700035e-05, "loss": 0.0199, "step": 54140 }, { "epoch": 4.55, "learning_rate": 1.0905273765535775e-05, "loss": 0.0998, "step": 54150 }, { "epoch": 4.55, "learning_rate": 1.0903594222371516e-05, "loss": 0.0927, "step": 54160 }, { "epoch": 4.55, "learning_rate": 1.0901914679207256e-05, "loss": 0.0527, "step": 54170 }, { "epoch": 4.55, "learning_rate": 1.0900235136042999e-05, "loss": 0.1346, "step": 54180 }, { "epoch": 4.55, "learning_rate": 1.0898555592878737e-05, "loss": 0.0475, "step": 54190 }, { "epoch": 4.55, "learning_rate": 1.0896876049714478e-05, "loss": 0.1036, "step": 54200 }, { "epoch": 4.55, "learning_rate": 1.0895196506550218e-05, "loss": 0.0337, "step": 54210 }, { "epoch": 4.55, "learning_rate": 1.0893516963385959e-05, "loss": 0.0174, "step": 54220 }, { "epoch": 4.55, "learning_rate": 1.0891837420221701e-05, "loss": 0.1008, "step": 54230 }, { "epoch": 4.55, "learning_rate": 1.0890157877057441e-05, "loss": 0.0655, "step": 54240 }, { "epoch": 4.56, "learning_rate": 1.0888478333893182e-05, "loss": 0.0406, "step": 54250 }, { "epoch": 4.56, "learning_rate": 1.0886798790728922e-05, "loss": 0.1275, "step": 54260 }, { "epoch": 4.56, "learning_rate": 1.0885119247564665e-05, "loss": 0.0573, "step": 54270 }, { "epoch": 4.56, "learning_rate": 1.0883439704400405e-05, "loss": 0.0651, "step": 54280 }, { "epoch": 4.56, "learning_rate": 1.0881760161236146e-05, "loss": 0.0124, "step": 54290 }, { "epoch": 4.56, "learning_rate": 1.0880080618071884e-05, "loss": 0.085, "step": 54300 }, { "epoch": 4.56, "learning_rate": 1.0878401074907625e-05, "loss": 0.082, "step": 54310 }, { "epoch": 4.56, "learning_rate": 1.0876721531743365e-05, "loss": 0.0465, "step": 54320 }, { "epoch": 4.56, "learning_rate": 1.0875041988579107e-05, "loss": 0.0309, "step": 54330 }, { "epoch": 4.56, "learning_rate": 1.0873362445414848e-05, "loss": 0.0981, "step": 54340 }, { "epoch": 4.56, "learning_rate": 1.0871682902250588e-05, "loss": 0.0784, "step": 54350 }, { "epoch": 4.56, "learning_rate": 1.0870003359086329e-05, "loss": 0.1551, "step": 54360 }, { "epoch": 4.57, "learning_rate": 1.0868323815922071e-05, "loss": 0.1028, "step": 54370 }, { "epoch": 4.57, "learning_rate": 1.0866644272757812e-05, "loss": 0.0386, "step": 54380 }, { "epoch": 4.57, "learning_rate": 1.0864964729593552e-05, "loss": 0.1049, "step": 54390 }, { "epoch": 4.57, "learning_rate": 1.0863285186429292e-05, "loss": 0.069, "step": 54400 }, { "epoch": 4.57, "learning_rate": 1.0861605643265031e-05, "loss": 0.0958, "step": 54410 }, { "epoch": 4.57, "learning_rate": 1.0859926100100773e-05, "loss": 0.08, "step": 54420 }, { "epoch": 4.57, "learning_rate": 1.0858246556936514e-05, "loss": 0.0873, "step": 54430 }, { "epoch": 4.57, "learning_rate": 1.0856567013772254e-05, "loss": 0.0252, "step": 54440 }, { "epoch": 4.57, "learning_rate": 1.0854887470607995e-05, "loss": 0.0252, "step": 54450 }, { "epoch": 4.57, "learning_rate": 1.0853207927443737e-05, "loss": 0.0295, "step": 54460 }, { "epoch": 4.57, "learning_rate": 1.0851528384279478e-05, "loss": 0.0627, "step": 54470 }, { "epoch": 4.58, "learning_rate": 1.0849848841115218e-05, "loss": 0.0157, "step": 54480 }, { "epoch": 4.58, "learning_rate": 1.0848169297950958e-05, "loss": 0.0572, "step": 54490 }, { "epoch": 4.58, "learning_rate": 1.08464897547867e-05, "loss": 0.0274, "step": 54500 }, { "epoch": 4.58, "learning_rate": 1.0844810211622438e-05, "loss": 0.0692, "step": 54510 }, { "epoch": 4.58, "learning_rate": 1.084313066845818e-05, "loss": 0.057, "step": 54520 }, { "epoch": 4.58, "learning_rate": 1.084145112529392e-05, "loss": 0.0798, "step": 54530 }, { "epoch": 4.58, "learning_rate": 1.083977158212966e-05, "loss": 0.05, "step": 54540 }, { "epoch": 4.58, "learning_rate": 1.0838092038965401e-05, "loss": 0.1229, "step": 54550 }, { "epoch": 4.58, "learning_rate": 1.0836412495801143e-05, "loss": 0.0522, "step": 54560 }, { "epoch": 4.58, "learning_rate": 1.0834732952636884e-05, "loss": 0.0225, "step": 54570 }, { "epoch": 4.58, "learning_rate": 1.0833053409472624e-05, "loss": 0.0049, "step": 54580 }, { "epoch": 4.58, "learning_rate": 1.0831373866308365e-05, "loss": 0.1172, "step": 54590 }, { "epoch": 4.59, "learning_rate": 1.0829694323144107e-05, "loss": 0.1178, "step": 54600 }, { "epoch": 4.59, "learning_rate": 1.0828014779979848e-05, "loss": 0.0985, "step": 54610 }, { "epoch": 4.59, "learning_rate": 1.0826335236815586e-05, "loss": 0.0515, "step": 54620 }, { "epoch": 4.59, "learning_rate": 1.0824655693651327e-05, "loss": 0.1198, "step": 54630 }, { "epoch": 4.59, "learning_rate": 1.0822976150487067e-05, "loss": 0.0581, "step": 54640 }, { "epoch": 4.59, "learning_rate": 1.082129660732281e-05, "loss": 0.1078, "step": 54650 }, { "epoch": 4.59, "learning_rate": 1.081961706415855e-05, "loss": 0.056, "step": 54660 }, { "epoch": 4.59, "learning_rate": 1.081793752099429e-05, "loss": 0.0126, "step": 54670 }, { "epoch": 4.59, "learning_rate": 1.0816257977830031e-05, "loss": 0.0321, "step": 54680 }, { "epoch": 4.59, "learning_rate": 1.0814578434665773e-05, "loss": 0.0472, "step": 54690 }, { "epoch": 4.59, "learning_rate": 1.0812898891501514e-05, "loss": 0.059, "step": 54700 }, { "epoch": 4.59, "learning_rate": 1.0811219348337254e-05, "loss": 0.0652, "step": 54710 }, { "epoch": 4.6, "learning_rate": 1.0809539805172993e-05, "loss": 0.0507, "step": 54720 }, { "epoch": 4.6, "learning_rate": 1.0807860262008733e-05, "loss": 0.0498, "step": 54730 }, { "epoch": 4.6, "learning_rate": 1.0806180718844474e-05, "loss": 0.034, "step": 54740 }, { "epoch": 4.6, "learning_rate": 1.0804501175680216e-05, "loss": 0.0626, "step": 54750 }, { "epoch": 4.6, "learning_rate": 1.0802821632515956e-05, "loss": 0.0355, "step": 54760 }, { "epoch": 4.6, "learning_rate": 1.0801142089351697e-05, "loss": 0.0369, "step": 54770 }, { "epoch": 4.6, "learning_rate": 1.0799462546187437e-05, "loss": 0.0017, "step": 54780 }, { "epoch": 4.6, "learning_rate": 1.079778300302318e-05, "loss": 0.1164, "step": 54790 }, { "epoch": 4.6, "learning_rate": 1.079610345985892e-05, "loss": 0.1049, "step": 54800 }, { "epoch": 4.6, "learning_rate": 1.079442391669466e-05, "loss": 0.0276, "step": 54810 }, { "epoch": 4.6, "learning_rate": 1.0792744373530401e-05, "loss": 0.0787, "step": 54820 }, { "epoch": 4.6, "learning_rate": 1.079106483036614e-05, "loss": 0.1413, "step": 54830 }, { "epoch": 4.61, "learning_rate": 1.0789385287201882e-05, "loss": 0.0518, "step": 54840 }, { "epoch": 4.61, "learning_rate": 1.0787705744037622e-05, "loss": 0.0612, "step": 54850 }, { "epoch": 4.61, "learning_rate": 1.0786026200873363e-05, "loss": 0.0831, "step": 54860 }, { "epoch": 4.61, "learning_rate": 1.0784346657709103e-05, "loss": 0.0497, "step": 54870 }, { "epoch": 4.61, "learning_rate": 1.0782667114544846e-05, "loss": 0.0557, "step": 54880 }, { "epoch": 4.61, "learning_rate": 1.0780987571380586e-05, "loss": 0.1129, "step": 54890 }, { "epoch": 4.61, "learning_rate": 1.0779308028216326e-05, "loss": 0.058, "step": 54900 }, { "epoch": 4.61, "learning_rate": 1.0777628485052067e-05, "loss": 0.0315, "step": 54910 }, { "epoch": 4.61, "learning_rate": 1.0775948941887807e-05, "loss": 0.0604, "step": 54920 }, { "epoch": 4.61, "learning_rate": 1.077426939872355e-05, "loss": 0.0663, "step": 54930 }, { "epoch": 4.61, "learning_rate": 1.0772589855559288e-05, "loss": 0.0203, "step": 54940 }, { "epoch": 4.61, "learning_rate": 1.0770910312395029e-05, "loss": 0.0216, "step": 54950 }, { "epoch": 4.62, "learning_rate": 1.076923076923077e-05, "loss": 0.0026, "step": 54960 }, { "epoch": 4.62, "learning_rate": 1.076755122606651e-05, "loss": 0.0079, "step": 54970 }, { "epoch": 4.62, "learning_rate": 1.0765871682902252e-05, "loss": 0.0536, "step": 54980 }, { "epoch": 4.62, "learning_rate": 1.0764192139737992e-05, "loss": 0.1123, "step": 54990 }, { "epoch": 4.62, "learning_rate": 1.0762512596573733e-05, "loss": 0.0204, "step": 55000 }, { "epoch": 4.62, "learning_rate": 1.0760833053409473e-05, "loss": 0.0683, "step": 55010 }, { "epoch": 4.62, "learning_rate": 1.0759153510245216e-05, "loss": 0.0006, "step": 55020 }, { "epoch": 4.62, "learning_rate": 1.0757473967080956e-05, "loss": 0.0044, "step": 55030 }, { "epoch": 4.62, "learning_rate": 1.0755794423916695e-05, "loss": 0.0534, "step": 55040 }, { "epoch": 4.62, "learning_rate": 1.0754114880752435e-05, "loss": 0.032, "step": 55050 }, { "epoch": 4.62, "learning_rate": 1.0752435337588176e-05, "loss": 0.0566, "step": 55060 }, { "epoch": 4.62, "learning_rate": 1.0750755794423918e-05, "loss": 0.0421, "step": 55070 }, { "epoch": 4.63, "learning_rate": 1.0749076251259658e-05, "loss": 0.0462, "step": 55080 }, { "epoch": 4.63, "learning_rate": 1.0747396708095399e-05, "loss": 0.0578, "step": 55090 }, { "epoch": 4.63, "learning_rate": 1.074571716493114e-05, "loss": 0.1123, "step": 55100 }, { "epoch": 4.63, "learning_rate": 1.074403762176688e-05, "loss": 0.0487, "step": 55110 }, { "epoch": 4.63, "learning_rate": 1.0742358078602622e-05, "loss": 0.105, "step": 55120 }, { "epoch": 4.63, "learning_rate": 1.0740678535438363e-05, "loss": 0.1294, "step": 55130 }, { "epoch": 4.63, "learning_rate": 1.0738998992274103e-05, "loss": 0.0525, "step": 55140 }, { "epoch": 4.63, "learning_rate": 1.0737319449109842e-05, "loss": 0.1389, "step": 55150 }, { "epoch": 4.63, "learning_rate": 1.0735639905945582e-05, "loss": 0.051, "step": 55160 }, { "epoch": 4.63, "learning_rate": 1.0733960362781324e-05, "loss": 0.1589, "step": 55170 }, { "epoch": 4.63, "learning_rate": 1.0732280819617065e-05, "loss": 0.0416, "step": 55180 }, { "epoch": 4.63, "learning_rate": 1.0730601276452805e-05, "loss": 0.2088, "step": 55190 }, { "epoch": 4.64, "learning_rate": 1.0728921733288546e-05, "loss": 0.0411, "step": 55200 }, { "epoch": 4.64, "learning_rate": 1.0727242190124288e-05, "loss": 0.0674, "step": 55210 }, { "epoch": 4.64, "learning_rate": 1.0725562646960028e-05, "loss": 0.088, "step": 55220 }, { "epoch": 4.64, "learning_rate": 1.0723883103795769e-05, "loss": 0.1579, "step": 55230 }, { "epoch": 4.64, "learning_rate": 1.072220356063151e-05, "loss": 0.0767, "step": 55240 }, { "epoch": 4.64, "learning_rate": 1.0720524017467248e-05, "loss": 0.0637, "step": 55250 }, { "epoch": 4.64, "learning_rate": 1.071884447430299e-05, "loss": 0.0449, "step": 55260 }, { "epoch": 4.64, "learning_rate": 1.0717164931138731e-05, "loss": 0.0504, "step": 55270 }, { "epoch": 4.64, "learning_rate": 1.0715485387974471e-05, "loss": 0.0538, "step": 55280 }, { "epoch": 4.64, "learning_rate": 1.0713805844810212e-05, "loss": 0.0344, "step": 55290 }, { "epoch": 4.64, "learning_rate": 1.0712126301645952e-05, "loss": 0.0379, "step": 55300 }, { "epoch": 4.64, "learning_rate": 1.0710446758481694e-05, "loss": 0.0201, "step": 55310 }, { "epoch": 4.65, "learning_rate": 1.0708767215317435e-05, "loss": 0.0578, "step": 55320 }, { "epoch": 4.65, "learning_rate": 1.0707087672153175e-05, "loss": 0.0199, "step": 55330 }, { "epoch": 4.65, "learning_rate": 1.0705408128988916e-05, "loss": 0.0598, "step": 55340 }, { "epoch": 4.65, "learning_rate": 1.0703728585824658e-05, "loss": 0.1011, "step": 55350 }, { "epoch": 4.65, "learning_rate": 1.0702049042660397e-05, "loss": 0.0021, "step": 55360 }, { "epoch": 4.65, "learning_rate": 1.0700369499496137e-05, "loss": 0.0244, "step": 55370 }, { "epoch": 4.65, "learning_rate": 1.0698689956331878e-05, "loss": 0.1473, "step": 55380 }, { "epoch": 4.65, "learning_rate": 1.0697010413167618e-05, "loss": 0.019, "step": 55390 }, { "epoch": 4.65, "learning_rate": 1.069533087000336e-05, "loss": 0.0888, "step": 55400 }, { "epoch": 4.65, "learning_rate": 1.0693651326839101e-05, "loss": 0.0429, "step": 55410 }, { "epoch": 4.65, "learning_rate": 1.0691971783674841e-05, "loss": 0.0707, "step": 55420 }, { "epoch": 4.65, "learning_rate": 1.0690292240510582e-05, "loss": 0.0544, "step": 55430 }, { "epoch": 4.66, "learning_rate": 1.0688612697346324e-05, "loss": 0.0899, "step": 55440 }, { "epoch": 4.66, "learning_rate": 1.0686933154182065e-05, "loss": 0.024, "step": 55450 }, { "epoch": 4.66, "learning_rate": 1.0685253611017805e-05, "loss": 0.0369, "step": 55460 }, { "epoch": 4.66, "learning_rate": 1.0683574067853544e-05, "loss": 0.0023, "step": 55470 }, { "epoch": 4.66, "learning_rate": 1.0681894524689284e-05, "loss": 0.1568, "step": 55480 }, { "epoch": 4.66, "learning_rate": 1.0680214981525025e-05, "loss": 0.0444, "step": 55490 }, { "epoch": 4.66, "learning_rate": 1.0678535438360767e-05, "loss": 0.0022, "step": 55500 }, { "epoch": 4.66, "learning_rate": 1.0676855895196507e-05, "loss": 0.0016, "step": 55510 }, { "epoch": 4.66, "learning_rate": 1.0675176352032248e-05, "loss": 0.1532, "step": 55520 }, { "epoch": 4.66, "learning_rate": 1.0673496808867988e-05, "loss": 0.0485, "step": 55530 }, { "epoch": 4.66, "learning_rate": 1.067181726570373e-05, "loss": 0.0268, "step": 55540 }, { "epoch": 4.66, "learning_rate": 1.0670137722539471e-05, "loss": 0.107, "step": 55550 }, { "epoch": 4.67, "learning_rate": 1.0668458179375211e-05, "loss": 0.102, "step": 55560 }, { "epoch": 4.67, "learning_rate": 1.066677863621095e-05, "loss": 0.0377, "step": 55570 }, { "epoch": 4.67, "learning_rate": 1.066509909304669e-05, "loss": 0.0812, "step": 55580 }, { "epoch": 4.67, "learning_rate": 1.0663419549882433e-05, "loss": 0.0972, "step": 55590 }, { "epoch": 4.67, "learning_rate": 1.0661740006718173e-05, "loss": 0.0113, "step": 55600 }, { "epoch": 4.67, "learning_rate": 1.0660060463553914e-05, "loss": 0.0892, "step": 55610 }, { "epoch": 4.67, "learning_rate": 1.0658380920389654e-05, "loss": 0.0457, "step": 55620 }, { "epoch": 4.67, "learning_rate": 1.0656701377225397e-05, "loss": 0.0465, "step": 55630 }, { "epoch": 4.67, "learning_rate": 1.0655021834061137e-05, "loss": 0.0021, "step": 55640 }, { "epoch": 4.67, "learning_rate": 1.0653342290896877e-05, "loss": 0.1711, "step": 55650 }, { "epoch": 4.67, "learning_rate": 1.0651662747732618e-05, "loss": 0.0672, "step": 55660 }, { "epoch": 4.68, "learning_rate": 1.064998320456836e-05, "loss": 0.0336, "step": 55670 }, { "epoch": 4.68, "learning_rate": 1.0648303661404097e-05, "loss": 0.0667, "step": 55680 }, { "epoch": 4.68, "learning_rate": 1.064662411823984e-05, "loss": 0.0469, "step": 55690 }, { "epoch": 4.68, "learning_rate": 1.064494457507558e-05, "loss": 0.0477, "step": 55700 }, { "epoch": 4.68, "learning_rate": 1.064326503191132e-05, "loss": 0.114, "step": 55710 }, { "epoch": 4.68, "learning_rate": 1.064158548874706e-05, "loss": 0.0772, "step": 55720 }, { "epoch": 4.68, "learning_rate": 1.0639905945582803e-05, "loss": 0.0874, "step": 55730 }, { "epoch": 4.68, "learning_rate": 1.0638226402418543e-05, "loss": 0.0358, "step": 55740 }, { "epoch": 4.68, "learning_rate": 1.0636546859254284e-05, "loss": 0.064, "step": 55750 }, { "epoch": 4.68, "learning_rate": 1.0634867316090024e-05, "loss": 0.0507, "step": 55760 }, { "epoch": 4.68, "learning_rate": 1.0633187772925767e-05, "loss": 0.0589, "step": 55770 }, { "epoch": 4.68, "learning_rate": 1.0631508229761505e-05, "loss": 0.0605, "step": 55780 }, { "epoch": 4.69, "learning_rate": 1.0629828686597246e-05, "loss": 0.0483, "step": 55790 }, { "epoch": 4.69, "learning_rate": 1.0628149143432986e-05, "loss": 0.0931, "step": 55800 }, { "epoch": 4.69, "learning_rate": 1.0626469600268727e-05, "loss": 0.0648, "step": 55810 }, { "epoch": 4.69, "learning_rate": 1.0624790057104469e-05, "loss": 0.066, "step": 55820 }, { "epoch": 4.69, "learning_rate": 1.062311051394021e-05, "loss": 0.144, "step": 55830 }, { "epoch": 4.69, "learning_rate": 1.062143097077595e-05, "loss": 0.065, "step": 55840 }, { "epoch": 4.69, "learning_rate": 1.061975142761169e-05, "loss": 0.0372, "step": 55850 }, { "epoch": 4.69, "learning_rate": 1.0618071884447433e-05, "loss": 0.0623, "step": 55860 }, { "epoch": 4.69, "learning_rate": 1.0616392341283173e-05, "loss": 0.0664, "step": 55870 }, { "epoch": 4.69, "learning_rate": 1.0614712798118914e-05, "loss": 0.0348, "step": 55880 }, { "epoch": 4.69, "learning_rate": 1.0613033254954652e-05, "loss": 0.0018, "step": 55890 }, { "epoch": 4.69, "learning_rate": 1.0611353711790393e-05, "loss": 0.0226, "step": 55900 }, { "epoch": 4.7, "learning_rate": 1.0609674168626133e-05, "loss": 0.0894, "step": 55910 }, { "epoch": 4.7, "learning_rate": 1.0607994625461875e-05, "loss": 0.0308, "step": 55920 }, { "epoch": 4.7, "learning_rate": 1.0606315082297616e-05, "loss": 0.1023, "step": 55930 }, { "epoch": 4.7, "learning_rate": 1.0604635539133356e-05, "loss": 0.0025, "step": 55940 }, { "epoch": 4.7, "learning_rate": 1.0602955995969097e-05, "loss": 0.0396, "step": 55950 }, { "epoch": 4.7, "learning_rate": 1.0601276452804839e-05, "loss": 0.0238, "step": 55960 }, { "epoch": 4.7, "learning_rate": 1.059959690964058e-05, "loss": 0.0179, "step": 55970 }, { "epoch": 4.7, "learning_rate": 1.059791736647632e-05, "loss": 0.081, "step": 55980 }, { "epoch": 4.7, "learning_rate": 1.059623782331206e-05, "loss": 0.0022, "step": 55990 }, { "epoch": 4.7, "learning_rate": 1.05945582801478e-05, "loss": 0.0695, "step": 56000 }, { "epoch": 4.7, "learning_rate": 1.0592878736983541e-05, "loss": 0.0007, "step": 56010 }, { "epoch": 4.7, "learning_rate": 1.0591199193819282e-05, "loss": 0.0553, "step": 56020 }, { "epoch": 4.71, "learning_rate": 1.0589519650655022e-05, "loss": 0.0669, "step": 56030 }, { "epoch": 4.71, "learning_rate": 1.0587840107490763e-05, "loss": 0.0387, "step": 56040 }, { "epoch": 4.71, "learning_rate": 1.0586160564326505e-05, "loss": 0.003, "step": 56050 }, { "epoch": 4.71, "learning_rate": 1.0584481021162245e-05, "loss": 0.0518, "step": 56060 }, { "epoch": 4.71, "learning_rate": 1.0582801477997986e-05, "loss": 0.0346, "step": 56070 }, { "epoch": 4.71, "learning_rate": 1.0581121934833726e-05, "loss": 0.0254, "step": 56080 }, { "epoch": 4.71, "learning_rate": 1.0579442391669467e-05, "loss": 0.1041, "step": 56090 }, { "epoch": 4.71, "learning_rate": 1.0577762848505206e-05, "loss": 0.0556, "step": 56100 }, { "epoch": 4.71, "learning_rate": 1.0576083305340948e-05, "loss": 0.0533, "step": 56110 }, { "epoch": 4.71, "learning_rate": 1.0574403762176688e-05, "loss": 0.0891, "step": 56120 }, { "epoch": 4.71, "learning_rate": 1.0572724219012429e-05, "loss": 0.1182, "step": 56130 }, { "epoch": 4.71, "learning_rate": 1.057104467584817e-05, "loss": 0.0651, "step": 56140 }, { "epoch": 4.72, "learning_rate": 1.0569365132683911e-05, "loss": 0.002, "step": 56150 }, { "epoch": 4.72, "learning_rate": 1.0567685589519652e-05, "loss": 0.0708, "step": 56160 }, { "epoch": 4.72, "learning_rate": 1.0566006046355392e-05, "loss": 0.0111, "step": 56170 }, { "epoch": 4.72, "learning_rate": 1.0564326503191133e-05, "loss": 0.0396, "step": 56180 }, { "epoch": 4.72, "learning_rate": 1.0562646960026875e-05, "loss": 0.096, "step": 56190 }, { "epoch": 4.72, "learning_rate": 1.0560967416862616e-05, "loss": 0.1012, "step": 56200 }, { "epoch": 4.72, "learning_rate": 1.0559287873698354e-05, "loss": 0.0611, "step": 56210 }, { "epoch": 4.72, "learning_rate": 1.0557608330534095e-05, "loss": 0.1431, "step": 56220 }, { "epoch": 4.72, "learning_rate": 1.0555928787369835e-05, "loss": 0.0944, "step": 56230 }, { "epoch": 4.72, "learning_rate": 1.0554249244205577e-05, "loss": 0.0377, "step": 56240 }, { "epoch": 4.72, "learning_rate": 1.0552569701041318e-05, "loss": 0.0982, "step": 56250 }, { "epoch": 4.72, "learning_rate": 1.0550890157877058e-05, "loss": 0.113, "step": 56260 }, { "epoch": 4.73, "learning_rate": 1.0549210614712799e-05, "loss": 0.007, "step": 56270 }, { "epoch": 4.73, "learning_rate": 1.054753107154854e-05, "loss": 0.1557, "step": 56280 }, { "epoch": 4.73, "learning_rate": 1.0545851528384282e-05, "loss": 0.1019, "step": 56290 }, { "epoch": 4.73, "learning_rate": 1.0544171985220022e-05, "loss": 0.0866, "step": 56300 }, { "epoch": 4.73, "learning_rate": 1.054249244205576e-05, "loss": 0.0595, "step": 56310 }, { "epoch": 4.73, "learning_rate": 1.0540812898891501e-05, "loss": 0.1197, "step": 56320 }, { "epoch": 4.73, "learning_rate": 1.0539133355727242e-05, "loss": 0.0662, "step": 56330 }, { "epoch": 4.73, "learning_rate": 1.0537453812562984e-05, "loss": 0.0323, "step": 56340 }, { "epoch": 4.73, "learning_rate": 1.0535774269398724e-05, "loss": 0.0813, "step": 56350 }, { "epoch": 4.73, "learning_rate": 1.0534094726234465e-05, "loss": 0.1334, "step": 56360 }, { "epoch": 4.73, "learning_rate": 1.0532415183070205e-05, "loss": 0.051, "step": 56370 }, { "epoch": 4.73, "learning_rate": 1.0530735639905948e-05, "loss": 0.0364, "step": 56380 }, { "epoch": 4.74, "learning_rate": 1.0529056096741688e-05, "loss": 0.0578, "step": 56390 }, { "epoch": 4.74, "learning_rate": 1.0527376553577428e-05, "loss": 0.0772, "step": 56400 }, { "epoch": 4.74, "learning_rate": 1.0525697010413169e-05, "loss": 0.0356, "step": 56410 }, { "epoch": 4.74, "learning_rate": 1.0524017467248908e-05, "loss": 0.0406, "step": 56420 }, { "epoch": 4.74, "learning_rate": 1.052233792408465e-05, "loss": 0.1302, "step": 56430 }, { "epoch": 4.74, "learning_rate": 1.052065838092039e-05, "loss": 0.0586, "step": 56440 }, { "epoch": 4.74, "learning_rate": 1.051897883775613e-05, "loss": 0.0183, "step": 56450 }, { "epoch": 4.74, "learning_rate": 1.0517299294591871e-05, "loss": 0.113, "step": 56460 }, { "epoch": 4.74, "learning_rate": 1.0515619751427612e-05, "loss": 0.0563, "step": 56470 }, { "epoch": 4.74, "learning_rate": 1.0513940208263354e-05, "loss": 0.0403, "step": 56480 }, { "epoch": 4.74, "learning_rate": 1.0512260665099094e-05, "loss": 0.02, "step": 56490 }, { "epoch": 4.74, "learning_rate": 1.0510581121934835e-05, "loss": 0.1167, "step": 56500 }, { "epoch": 4.75, "learning_rate": 1.0508901578770575e-05, "loss": 0.0735, "step": 56510 }, { "epoch": 4.75, "learning_rate": 1.0507222035606318e-05, "loss": 0.0221, "step": 56520 }, { "epoch": 4.75, "learning_rate": 1.0505542492442056e-05, "loss": 0.0602, "step": 56530 }, { "epoch": 4.75, "learning_rate": 1.0503862949277797e-05, "loss": 0.0253, "step": 56540 }, { "epoch": 4.75, "learning_rate": 1.0502183406113537e-05, "loss": 0.036, "step": 56550 }, { "epoch": 4.75, "learning_rate": 1.0500503862949278e-05, "loss": 0.0651, "step": 56560 }, { "epoch": 4.75, "learning_rate": 1.049882431978502e-05, "loss": 0.1103, "step": 56570 }, { "epoch": 4.75, "learning_rate": 1.049714477662076e-05, "loss": 0.0258, "step": 56580 }, { "epoch": 4.75, "learning_rate": 1.0495465233456501e-05, "loss": 0.0448, "step": 56590 }, { "epoch": 4.75, "learning_rate": 1.0493785690292241e-05, "loss": 0.0384, "step": 56600 }, { "epoch": 4.75, "learning_rate": 1.0492106147127984e-05, "loss": 0.0417, "step": 56610 }, { "epoch": 4.75, "learning_rate": 1.0490426603963724e-05, "loss": 0.064, "step": 56620 }, { "epoch": 4.76, "learning_rate": 1.0488747060799463e-05, "loss": 0.0449, "step": 56630 }, { "epoch": 4.76, "learning_rate": 1.0487067517635203e-05, "loss": 0.0881, "step": 56640 }, { "epoch": 4.76, "learning_rate": 1.0485387974470944e-05, "loss": 0.1205, "step": 56650 }, { "epoch": 4.76, "learning_rate": 1.0483708431306684e-05, "loss": 0.0847, "step": 56660 }, { "epoch": 4.76, "learning_rate": 1.0482028888142426e-05, "loss": 0.0445, "step": 56670 }, { "epoch": 4.76, "learning_rate": 1.0480349344978167e-05, "loss": 0.0687, "step": 56680 }, { "epoch": 4.76, "learning_rate": 1.0478669801813907e-05, "loss": 0.015, "step": 56690 }, { "epoch": 4.76, "learning_rate": 1.0476990258649648e-05, "loss": 0.0507, "step": 56700 }, { "epoch": 4.76, "learning_rate": 1.047531071548539e-05, "loss": 0.1529, "step": 56710 }, { "epoch": 4.76, "learning_rate": 1.047363117232113e-05, "loss": 0.0622, "step": 56720 }, { "epoch": 4.76, "learning_rate": 1.0471951629156871e-05, "loss": 0.0732, "step": 56730 }, { "epoch": 4.76, "learning_rate": 1.047027208599261e-05, "loss": 0.1462, "step": 56740 }, { "epoch": 4.77, "learning_rate": 1.046859254282835e-05, "loss": 0.12, "step": 56750 }, { "epoch": 4.77, "learning_rate": 1.0466912999664092e-05, "loss": 0.1402, "step": 56760 }, { "epoch": 4.77, "learning_rate": 1.0465233456499833e-05, "loss": 0.0386, "step": 56770 }, { "epoch": 4.77, "learning_rate": 1.0463553913335573e-05, "loss": 0.0154, "step": 56780 }, { "epoch": 4.77, "learning_rate": 1.0461874370171314e-05, "loss": 0.1039, "step": 56790 }, { "epoch": 4.77, "learning_rate": 1.0460194827007056e-05, "loss": 0.0344, "step": 56800 }, { "epoch": 4.77, "learning_rate": 1.0458515283842796e-05, "loss": 0.0811, "step": 56810 }, { "epoch": 4.77, "learning_rate": 1.0456835740678537e-05, "loss": 0.0949, "step": 56820 }, { "epoch": 4.77, "learning_rate": 1.0455156197514277e-05, "loss": 0.002, "step": 56830 }, { "epoch": 4.77, "learning_rate": 1.0453476654350016e-05, "loss": 0.1526, "step": 56840 }, { "epoch": 4.77, "learning_rate": 1.0451797111185757e-05, "loss": 0.0872, "step": 56850 }, { "epoch": 4.77, "learning_rate": 1.0450117568021499e-05, "loss": 0.0559, "step": 56860 }, { "epoch": 4.78, "learning_rate": 1.044843802485724e-05, "loss": 0.0474, "step": 56870 }, { "epoch": 4.78, "learning_rate": 1.044675848169298e-05, "loss": 0.0448, "step": 56880 }, { "epoch": 4.78, "learning_rate": 1.044507893852872e-05, "loss": 0.0368, "step": 56890 }, { "epoch": 4.78, "learning_rate": 1.0443399395364462e-05, "loss": 0.0989, "step": 56900 }, { "epoch": 4.78, "learning_rate": 1.0441719852200203e-05, "loss": 0.0317, "step": 56910 }, { "epoch": 4.78, "learning_rate": 1.0440040309035943e-05, "loss": 0.0627, "step": 56920 }, { "epoch": 4.78, "learning_rate": 1.0438360765871684e-05, "loss": 0.0315, "step": 56930 }, { "epoch": 4.78, "learning_rate": 1.0436681222707426e-05, "loss": 0.0741, "step": 56940 }, { "epoch": 4.78, "learning_rate": 1.0435001679543165e-05, "loss": 0.0148, "step": 56950 }, { "epoch": 4.78, "learning_rate": 1.0433322136378905e-05, "loss": 0.0022, "step": 56960 }, { "epoch": 4.78, "learning_rate": 1.0431642593214646e-05, "loss": 0.1153, "step": 56970 }, { "epoch": 4.79, "learning_rate": 1.0429963050050386e-05, "loss": 0.1028, "step": 56980 }, { "epoch": 4.79, "learning_rate": 1.0428283506886128e-05, "loss": 0.0939, "step": 56990 }, { "epoch": 4.79, "learning_rate": 1.0426603963721869e-05, "loss": 0.1305, "step": 57000 }, { "epoch": 4.79, "learning_rate": 1.042492442055761e-05, "loss": 0.0744, "step": 57010 }, { "epoch": 4.79, "learning_rate": 1.042324487739335e-05, "loss": 0.0551, "step": 57020 }, { "epoch": 4.79, "learning_rate": 1.0421565334229092e-05, "loss": 0.1841, "step": 57030 }, { "epoch": 4.79, "learning_rate": 1.0419885791064833e-05, "loss": 0.0721, "step": 57040 }, { "epoch": 4.79, "learning_rate": 1.0418206247900573e-05, "loss": 0.068, "step": 57050 }, { "epoch": 4.79, "learning_rate": 1.0416526704736312e-05, "loss": 0.052, "step": 57060 }, { "epoch": 4.79, "learning_rate": 1.0414847161572052e-05, "loss": 0.0843, "step": 57070 }, { "epoch": 4.79, "learning_rate": 1.0413167618407793e-05, "loss": 0.0259, "step": 57080 }, { "epoch": 4.79, "learning_rate": 1.0411488075243535e-05, "loss": 0.0056, "step": 57090 }, { "epoch": 4.8, "learning_rate": 1.0409808532079275e-05, "loss": 0.1039, "step": 57100 }, { "epoch": 4.8, "learning_rate": 1.0408128988915016e-05, "loss": 0.0232, "step": 57110 }, { "epoch": 4.8, "learning_rate": 1.0406449445750756e-05, "loss": 0.1304, "step": 57120 }, { "epoch": 4.8, "learning_rate": 1.0404769902586499e-05, "loss": 0.0567, "step": 57130 }, { "epoch": 4.8, "learning_rate": 1.0403090359422239e-05, "loss": 0.072, "step": 57140 }, { "epoch": 4.8, "learning_rate": 1.040141081625798e-05, "loss": 0.0555, "step": 57150 }, { "epoch": 4.8, "learning_rate": 1.0399731273093718e-05, "loss": 0.0864, "step": 57160 }, { "epoch": 4.8, "learning_rate": 1.0398051729929459e-05, "loss": 0.0501, "step": 57170 }, { "epoch": 4.8, "learning_rate": 1.0396372186765201e-05, "loss": 0.0998, "step": 57180 }, { "epoch": 4.8, "learning_rate": 1.0394692643600941e-05, "loss": 0.1296, "step": 57190 }, { "epoch": 4.8, "learning_rate": 1.0393013100436682e-05, "loss": 0.0463, "step": 57200 }, { "epoch": 4.8, "learning_rate": 1.0391333557272422e-05, "loss": 0.0436, "step": 57210 }, { "epoch": 4.81, "learning_rate": 1.0389654014108164e-05, "loss": 0.0664, "step": 57220 }, { "epoch": 4.81, "learning_rate": 1.0387974470943905e-05, "loss": 0.1069, "step": 57230 }, { "epoch": 4.81, "learning_rate": 1.0386294927779645e-05, "loss": 0.0174, "step": 57240 }, { "epoch": 4.81, "learning_rate": 1.0384615384615386e-05, "loss": 0.1184, "step": 57250 }, { "epoch": 4.81, "learning_rate": 1.0382935841451126e-05, "loss": 0.0606, "step": 57260 }, { "epoch": 4.81, "learning_rate": 1.0381256298286865e-05, "loss": 0.1084, "step": 57270 }, { "epoch": 4.81, "learning_rate": 1.0379576755122607e-05, "loss": 0.0881, "step": 57280 }, { "epoch": 4.81, "learning_rate": 1.0377897211958348e-05, "loss": 0.0655, "step": 57290 }, { "epoch": 4.81, "learning_rate": 1.0376217668794088e-05, "loss": 0.0044, "step": 57300 }, { "epoch": 4.81, "learning_rate": 1.0374538125629829e-05, "loss": 0.038, "step": 57310 }, { "epoch": 4.81, "learning_rate": 1.0372858582465571e-05, "loss": 0.0713, "step": 57320 }, { "epoch": 4.81, "learning_rate": 1.0371179039301311e-05, "loss": 0.037, "step": 57330 }, { "epoch": 4.82, "learning_rate": 1.0369499496137052e-05, "loss": 0.0892, "step": 57340 }, { "epoch": 4.82, "learning_rate": 1.0367819952972792e-05, "loss": 0.0324, "step": 57350 }, { "epoch": 4.82, "learning_rate": 1.0366140409808535e-05, "loss": 0.0474, "step": 57360 }, { "epoch": 4.82, "learning_rate": 1.0364460866644273e-05, "loss": 0.0556, "step": 57370 }, { "epoch": 4.82, "learning_rate": 1.0362781323480014e-05, "loss": 0.0127, "step": 57380 }, { "epoch": 4.82, "learning_rate": 1.0361101780315754e-05, "loss": 0.0684, "step": 57390 }, { "epoch": 4.82, "learning_rate": 1.0359422237151495e-05, "loss": 0.0291, "step": 57400 }, { "epoch": 4.82, "learning_rate": 1.0357742693987235e-05, "loss": 0.0656, "step": 57410 }, { "epoch": 4.82, "learning_rate": 1.0356063150822977e-05, "loss": 0.0835, "step": 57420 }, { "epoch": 4.82, "learning_rate": 1.0354383607658718e-05, "loss": 0.116, "step": 57430 }, { "epoch": 4.82, "learning_rate": 1.0352704064494458e-05, "loss": 0.1006, "step": 57440 }, { "epoch": 4.82, "learning_rate": 1.0351024521330199e-05, "loss": 0.0621, "step": 57450 }, { "epoch": 4.83, "learning_rate": 1.0349344978165941e-05, "loss": 0.1014, "step": 57460 }, { "epoch": 4.83, "learning_rate": 1.0347665435001681e-05, "loss": 0.0347, "step": 57470 }, { "epoch": 4.83, "learning_rate": 1.034598589183742e-05, "loss": 0.0987, "step": 57480 }, { "epoch": 4.83, "learning_rate": 1.034430634867316e-05, "loss": 0.0857, "step": 57490 }, { "epoch": 4.83, "learning_rate": 1.0342626805508901e-05, "loss": 0.0548, "step": 57500 }, { "epoch": 4.83, "learning_rate": 1.0340947262344643e-05, "loss": 0.0742, "step": 57510 }, { "epoch": 4.83, "learning_rate": 1.0339267719180384e-05, "loss": 0.1307, "step": 57520 }, { "epoch": 4.83, "learning_rate": 1.0337588176016124e-05, "loss": 0.0708, "step": 57530 }, { "epoch": 4.83, "learning_rate": 1.0335908632851865e-05, "loss": 0.0072, "step": 57540 }, { "epoch": 4.83, "learning_rate": 1.0334229089687607e-05, "loss": 0.1204, "step": 57550 }, { "epoch": 4.83, "learning_rate": 1.0332549546523347e-05, "loss": 0.012, "step": 57560 }, { "epoch": 4.83, "learning_rate": 1.0330870003359088e-05, "loss": 0.0828, "step": 57570 }, { "epoch": 4.84, "learning_rate": 1.0329190460194828e-05, "loss": 0.0933, "step": 57580 }, { "epoch": 4.84, "learning_rate": 1.0327510917030567e-05, "loss": 0.0028, "step": 57590 }, { "epoch": 4.84, "learning_rate": 1.0325831373866308e-05, "loss": 0.0343, "step": 57600 }, { "epoch": 4.84, "learning_rate": 1.032415183070205e-05, "loss": 0.049, "step": 57610 }, { "epoch": 4.84, "learning_rate": 1.032247228753779e-05, "loss": 0.0678, "step": 57620 }, { "epoch": 4.84, "learning_rate": 1.032079274437353e-05, "loss": 0.0547, "step": 57630 }, { "epoch": 4.84, "learning_rate": 1.0319113201209271e-05, "loss": 0.0648, "step": 57640 }, { "epoch": 4.84, "learning_rate": 1.0317433658045013e-05, "loss": 0.0283, "step": 57650 }, { "epoch": 4.84, "learning_rate": 1.0315754114880754e-05, "loss": 0.0498, "step": 57660 }, { "epoch": 4.84, "learning_rate": 1.0314074571716494e-05, "loss": 0.0289, "step": 57670 }, { "epoch": 4.84, "learning_rate": 1.0312395028552235e-05, "loss": 0.0185, "step": 57680 }, { "epoch": 4.84, "learning_rate": 1.0310715485387974e-05, "loss": 0.0778, "step": 57690 }, { "epoch": 4.85, "learning_rate": 1.0309035942223716e-05, "loss": 0.0254, "step": 57700 }, { "epoch": 4.85, "learning_rate": 1.0307356399059456e-05, "loss": 0.0018, "step": 57710 }, { "epoch": 4.85, "learning_rate": 1.0305676855895197e-05, "loss": 0.1335, "step": 57720 }, { "epoch": 4.85, "learning_rate": 1.0303997312730937e-05, "loss": 0.0558, "step": 57730 }, { "epoch": 4.85, "learning_rate": 1.030231776956668e-05, "loss": 0.012, "step": 57740 }, { "epoch": 4.85, "learning_rate": 1.030063822640242e-05, "loss": 0.1143, "step": 57750 }, { "epoch": 4.85, "learning_rate": 1.029895868323816e-05, "loss": 0.025, "step": 57760 }, { "epoch": 4.85, "learning_rate": 1.0297279140073901e-05, "loss": 0.0531, "step": 57770 }, { "epoch": 4.85, "learning_rate": 1.0295599596909643e-05, "loss": 0.0878, "step": 57780 }, { "epoch": 4.85, "learning_rate": 1.0293920053745384e-05, "loss": 0.0307, "step": 57790 }, { "epoch": 4.85, "learning_rate": 1.0292240510581122e-05, "loss": 0.0437, "step": 57800 }, { "epoch": 4.85, "learning_rate": 1.0290560967416863e-05, "loss": 0.0566, "step": 57810 }, { "epoch": 4.86, "learning_rate": 1.0288881424252603e-05, "loss": 0.1313, "step": 57820 }, { "epoch": 4.86, "learning_rate": 1.0287201881088344e-05, "loss": 0.1316, "step": 57830 }, { "epoch": 4.86, "learning_rate": 1.0285522337924086e-05, "loss": 0.1514, "step": 57840 }, { "epoch": 4.86, "learning_rate": 1.0283842794759826e-05, "loss": 0.1264, "step": 57850 }, { "epoch": 4.86, "learning_rate": 1.0282163251595567e-05, "loss": 0.0628, "step": 57860 }, { "epoch": 4.86, "learning_rate": 1.0280483708431307e-05, "loss": 0.0486, "step": 57870 }, { "epoch": 4.86, "learning_rate": 1.027880416526705e-05, "loss": 0.0201, "step": 57880 }, { "epoch": 4.86, "learning_rate": 1.027712462210279e-05, "loss": 0.0027, "step": 57890 }, { "epoch": 4.86, "learning_rate": 1.0275445078938529e-05, "loss": 0.0857, "step": 57900 }, { "epoch": 4.86, "learning_rate": 1.027376553577427e-05, "loss": 0.0043, "step": 57910 }, { "epoch": 4.86, "learning_rate": 1.027208599261001e-05, "loss": 0.0773, "step": 57920 }, { "epoch": 4.86, "learning_rate": 1.0270406449445752e-05, "loss": 0.0505, "step": 57930 }, { "epoch": 4.87, "learning_rate": 1.0268726906281492e-05, "loss": 0.0768, "step": 57940 }, { "epoch": 4.87, "learning_rate": 1.0267047363117233e-05, "loss": 0.0525, "step": 57950 }, { "epoch": 4.87, "learning_rate": 1.0265367819952973e-05, "loss": 0.1716, "step": 57960 }, { "epoch": 4.87, "learning_rate": 1.0263688276788715e-05, "loss": 0.111, "step": 57970 }, { "epoch": 4.87, "learning_rate": 1.0262008733624456e-05, "loss": 0.0914, "step": 57980 }, { "epoch": 4.87, "learning_rate": 1.0260329190460196e-05, "loss": 0.0573, "step": 57990 }, { "epoch": 4.87, "learning_rate": 1.0258649647295937e-05, "loss": 0.0541, "step": 58000 }, { "epoch": 4.87, "learning_rate": 1.0256970104131676e-05, "loss": 0.0496, "step": 58010 }, { "epoch": 4.87, "learning_rate": 1.0255290560967416e-05, "loss": 0.0282, "step": 58020 }, { "epoch": 4.87, "learning_rate": 1.0253611017803158e-05, "loss": 0.08, "step": 58030 }, { "epoch": 4.87, "learning_rate": 1.0251931474638899e-05, "loss": 0.0419, "step": 58040 }, { "epoch": 4.87, "learning_rate": 1.025025193147464e-05, "loss": 0.0063, "step": 58050 }, { "epoch": 4.88, "learning_rate": 1.024857238831038e-05, "loss": 0.0522, "step": 58060 }, { "epoch": 4.88, "learning_rate": 1.0246892845146122e-05, "loss": 0.0726, "step": 58070 }, { "epoch": 4.88, "learning_rate": 1.0245213301981862e-05, "loss": 0.0656, "step": 58080 }, { "epoch": 4.88, "learning_rate": 1.0243533758817603e-05, "loss": 0.068, "step": 58090 }, { "epoch": 4.88, "learning_rate": 1.0241854215653343e-05, "loss": 0.037, "step": 58100 }, { "epoch": 4.88, "learning_rate": 1.0240174672489082e-05, "loss": 0.0825, "step": 58110 }, { "epoch": 4.88, "learning_rate": 1.0238495129324824e-05, "loss": 0.1831, "step": 58120 }, { "epoch": 4.88, "learning_rate": 1.0236815586160565e-05, "loss": 0.0274, "step": 58130 }, { "epoch": 4.88, "learning_rate": 1.0235136042996305e-05, "loss": 0.0486, "step": 58140 }, { "epoch": 4.88, "learning_rate": 1.0233456499832046e-05, "loss": 0.0727, "step": 58150 }, { "epoch": 4.88, "learning_rate": 1.0231776956667788e-05, "loss": 0.0405, "step": 58160 }, { "epoch": 4.88, "learning_rate": 1.0230097413503528e-05, "loss": 0.0873, "step": 58170 }, { "epoch": 4.89, "learning_rate": 1.0228417870339269e-05, "loss": 0.0241, "step": 58180 }, { "epoch": 4.89, "learning_rate": 1.022673832717501e-05, "loss": 0.0617, "step": 58190 }, { "epoch": 4.89, "learning_rate": 1.0225058784010752e-05, "loss": 0.035, "step": 58200 }, { "epoch": 4.89, "learning_rate": 1.0223379240846492e-05, "loss": 0.0013, "step": 58210 }, { "epoch": 4.89, "learning_rate": 1.022169969768223e-05, "loss": 0.0037, "step": 58220 }, { "epoch": 4.89, "learning_rate": 1.0220020154517971e-05, "loss": 0.073, "step": 58230 }, { "epoch": 4.89, "learning_rate": 1.0218340611353712e-05, "loss": 0.0727, "step": 58240 }, { "epoch": 4.89, "learning_rate": 1.0216661068189452e-05, "loss": 0.0226, "step": 58250 }, { "epoch": 4.89, "learning_rate": 1.0214981525025194e-05, "loss": 0.0516, "step": 58260 }, { "epoch": 4.89, "learning_rate": 1.0213301981860935e-05, "loss": 0.0632, "step": 58270 }, { "epoch": 4.89, "learning_rate": 1.0211622438696675e-05, "loss": 0.0406, "step": 58280 }, { "epoch": 4.9, "learning_rate": 1.0209942895532416e-05, "loss": 0.0631, "step": 58290 }, { "epoch": 4.9, "learning_rate": 1.0208263352368158e-05, "loss": 0.0654, "step": 58300 }, { "epoch": 4.9, "learning_rate": 1.0206583809203898e-05, "loss": 0.1448, "step": 58310 }, { "epoch": 4.9, "learning_rate": 1.0204904266039639e-05, "loss": 0.02, "step": 58320 }, { "epoch": 4.9, "learning_rate": 1.0203224722875378e-05, "loss": 0.0521, "step": 58330 }, { "epoch": 4.9, "learning_rate": 1.0201545179711118e-05, "loss": 0.0345, "step": 58340 }, { "epoch": 4.9, "learning_rate": 1.019986563654686e-05, "loss": 0.1774, "step": 58350 }, { "epoch": 4.9, "learning_rate": 1.0198186093382601e-05, "loss": 0.0542, "step": 58360 }, { "epoch": 4.9, "learning_rate": 1.0196506550218341e-05, "loss": 0.0475, "step": 58370 }, { "epoch": 4.9, "learning_rate": 1.0194827007054082e-05, "loss": 0.0028, "step": 58380 }, { "epoch": 4.9, "learning_rate": 1.0193147463889822e-05, "loss": 0.0306, "step": 58390 }, { "epoch": 4.9, "learning_rate": 1.0191467920725564e-05, "loss": 0.046, "step": 58400 }, { "epoch": 4.91, "learning_rate": 1.0189788377561305e-05, "loss": 0.0428, "step": 58410 }, { "epoch": 4.91, "learning_rate": 1.0188108834397045e-05, "loss": 0.0632, "step": 58420 }, { "epoch": 4.91, "learning_rate": 1.0186429291232784e-05, "loss": 0.0855, "step": 58430 }, { "epoch": 4.91, "learning_rate": 1.0184749748068525e-05, "loss": 0.1388, "step": 58440 }, { "epoch": 4.91, "learning_rate": 1.0183070204904267e-05, "loss": 0.0268, "step": 58450 }, { "epoch": 4.91, "learning_rate": 1.0181390661740007e-05, "loss": 0.0751, "step": 58460 }, { "epoch": 4.91, "learning_rate": 1.0179711118575748e-05, "loss": 0.0021, "step": 58470 }, { "epoch": 4.91, "learning_rate": 1.0178031575411488e-05, "loss": 0.1933, "step": 58480 }, { "epoch": 4.91, "learning_rate": 1.017635203224723e-05, "loss": 0.018, "step": 58490 }, { "epoch": 4.91, "learning_rate": 1.0174672489082971e-05, "loss": 0.002, "step": 58500 }, { "epoch": 4.91, "learning_rate": 1.0172992945918711e-05, "loss": 0.0849, "step": 58510 }, { "epoch": 4.91, "learning_rate": 1.0171313402754452e-05, "loss": 0.074, "step": 58520 }, { "epoch": 4.92, "learning_rate": 1.0169633859590194e-05, "loss": 0.0507, "step": 58530 }, { "epoch": 4.92, "learning_rate": 1.0167954316425933e-05, "loss": 0.1912, "step": 58540 }, { "epoch": 4.92, "learning_rate": 1.0166274773261673e-05, "loss": 0.1108, "step": 58550 }, { "epoch": 4.92, "learning_rate": 1.0164595230097414e-05, "loss": 0.0518, "step": 58560 }, { "epoch": 4.92, "learning_rate": 1.0162915686933154e-05, "loss": 0.1193, "step": 58570 }, { "epoch": 4.92, "learning_rate": 1.0161236143768895e-05, "loss": 0.0542, "step": 58580 }, { "epoch": 4.92, "learning_rate": 1.0159556600604637e-05, "loss": 0.0644, "step": 58590 }, { "epoch": 4.92, "learning_rate": 1.0157877057440377e-05, "loss": 0.0401, "step": 58600 }, { "epoch": 4.92, "learning_rate": 1.0156197514276118e-05, "loss": 0.051, "step": 58610 }, { "epoch": 4.92, "learning_rate": 1.0154517971111858e-05, "loss": 0.0024, "step": 58620 }, { "epoch": 4.92, "learning_rate": 1.01528384279476e-05, "loss": 0.0496, "step": 58630 }, { "epoch": 4.92, "learning_rate": 1.015115888478334e-05, "loss": 0.1265, "step": 58640 }, { "epoch": 4.93, "learning_rate": 1.014947934161908e-05, "loss": 0.0597, "step": 58650 }, { "epoch": 4.93, "learning_rate": 1.014779979845482e-05, "loss": 0.05, "step": 58660 }, { "epoch": 4.93, "learning_rate": 1.014612025529056e-05, "loss": 0.0957, "step": 58670 }, { "epoch": 4.93, "learning_rate": 1.0144440712126303e-05, "loss": 0.0381, "step": 58680 }, { "epoch": 4.93, "learning_rate": 1.0142761168962043e-05, "loss": 0.0738, "step": 58690 }, { "epoch": 4.93, "learning_rate": 1.0141081625797784e-05, "loss": 0.0497, "step": 58700 }, { "epoch": 4.93, "learning_rate": 1.0139402082633524e-05, "loss": 0.0532, "step": 58710 }, { "epoch": 4.93, "learning_rate": 1.0137722539469266e-05, "loss": 0.0985, "step": 58720 }, { "epoch": 4.93, "learning_rate": 1.0136042996305007e-05, "loss": 0.0614, "step": 58730 }, { "epoch": 4.93, "learning_rate": 1.0134363453140747e-05, "loss": 0.0725, "step": 58740 }, { "epoch": 4.93, "learning_rate": 1.0132683909976486e-05, "loss": 0.0739, "step": 58750 }, { "epoch": 4.93, "learning_rate": 1.0131004366812227e-05, "loss": 0.085, "step": 58760 }, { "epoch": 4.94, "learning_rate": 1.0129324823647967e-05, "loss": 0.0622, "step": 58770 }, { "epoch": 4.94, "learning_rate": 1.012764528048371e-05, "loss": 0.0311, "step": 58780 }, { "epoch": 4.94, "learning_rate": 1.012596573731945e-05, "loss": 0.0575, "step": 58790 }, { "epoch": 4.94, "learning_rate": 1.012428619415519e-05, "loss": 0.0513, "step": 58800 }, { "epoch": 4.94, "learning_rate": 1.012260665099093e-05, "loss": 0.1242, "step": 58810 }, { "epoch": 4.94, "learning_rate": 1.0120927107826673e-05, "loss": 0.0284, "step": 58820 }, { "epoch": 4.94, "learning_rate": 1.0119247564662413e-05, "loss": 0.0615, "step": 58830 }, { "epoch": 4.94, "learning_rate": 1.0117568021498154e-05, "loss": 0.0566, "step": 58840 }, { "epoch": 4.94, "learning_rate": 1.0115888478333894e-05, "loss": 0.1006, "step": 58850 }, { "epoch": 4.94, "learning_rate": 1.0114208935169633e-05, "loss": 0.0432, "step": 58860 }, { "epoch": 4.94, "learning_rate": 1.0112529392005375e-05, "loss": 0.0943, "step": 58870 }, { "epoch": 4.94, "learning_rate": 1.0110849848841116e-05, "loss": 0.0238, "step": 58880 }, { "epoch": 4.95, "learning_rate": 1.0109170305676856e-05, "loss": 0.0388, "step": 58890 }, { "epoch": 4.95, "learning_rate": 1.0107490762512597e-05, "loss": 0.0799, "step": 58900 }, { "epoch": 4.95, "learning_rate": 1.0105811219348339e-05, "loss": 0.0323, "step": 58910 }, { "epoch": 4.95, "learning_rate": 1.010413167618408e-05, "loss": 0.117, "step": 58920 }, { "epoch": 4.95, "learning_rate": 1.010245213301982e-05, "loss": 0.0927, "step": 58930 }, { "epoch": 4.95, "learning_rate": 1.010077258985556e-05, "loss": 0.071, "step": 58940 }, { "epoch": 4.95, "learning_rate": 1.0099093046691303e-05, "loss": 0.0996, "step": 58950 }, { "epoch": 4.95, "learning_rate": 1.009741350352704e-05, "loss": 0.0643, "step": 58960 }, { "epoch": 4.95, "learning_rate": 1.0095733960362782e-05, "loss": 0.0188, "step": 58970 }, { "epoch": 4.95, "learning_rate": 1.0094054417198522e-05, "loss": 0.0972, "step": 58980 }, { "epoch": 4.95, "learning_rate": 1.0092374874034263e-05, "loss": 0.0953, "step": 58990 }, { "epoch": 4.95, "learning_rate": 1.0090695330870003e-05, "loss": 0.0445, "step": 59000 }, { "epoch": 4.96, "learning_rate": 1.0089015787705745e-05, "loss": 0.0664, "step": 59010 }, { "epoch": 4.96, "learning_rate": 1.0087336244541486e-05, "loss": 0.0632, "step": 59020 }, { "epoch": 4.96, "learning_rate": 1.0085656701377226e-05, "loss": 0.0862, "step": 59030 }, { "epoch": 4.96, "learning_rate": 1.0083977158212967e-05, "loss": 0.0467, "step": 59040 }, { "epoch": 4.96, "learning_rate": 1.0082297615048709e-05, "loss": 0.1212, "step": 59050 }, { "epoch": 4.96, "learning_rate": 1.008061807188445e-05, "loss": 0.0205, "step": 59060 }, { "epoch": 4.96, "learning_rate": 1.0078938528720188e-05, "loss": 0.0486, "step": 59070 }, { "epoch": 4.96, "learning_rate": 1.0077258985555929e-05, "loss": 0.057, "step": 59080 }, { "epoch": 4.96, "learning_rate": 1.007557944239167e-05, "loss": 0.0228, "step": 59090 }, { "epoch": 4.96, "learning_rate": 1.0073899899227411e-05, "loss": 0.0227, "step": 59100 }, { "epoch": 4.96, "learning_rate": 1.0072220356063152e-05, "loss": 0.1024, "step": 59110 }, { "epoch": 4.96, "learning_rate": 1.0070540812898892e-05, "loss": 0.171, "step": 59120 }, { "epoch": 4.97, "learning_rate": 1.0068861269734633e-05, "loss": 0.0543, "step": 59130 }, { "epoch": 4.97, "learning_rate": 1.0067181726570375e-05, "loss": 0.1004, "step": 59140 }, { "epoch": 4.97, "learning_rate": 1.0065502183406115e-05, "loss": 0.0369, "step": 59150 }, { "epoch": 4.97, "learning_rate": 1.0063822640241856e-05, "loss": 0.0534, "step": 59160 }, { "epoch": 4.97, "learning_rate": 1.0062143097077595e-05, "loss": 0.0181, "step": 59170 }, { "epoch": 4.97, "learning_rate": 1.0060463553913335e-05, "loss": 0.0727, "step": 59180 }, { "epoch": 4.97, "learning_rate": 1.0058784010749076e-05, "loss": 0.0646, "step": 59190 }, { "epoch": 4.97, "learning_rate": 1.0057104467584818e-05, "loss": 0.0228, "step": 59200 }, { "epoch": 4.97, "learning_rate": 1.0055424924420558e-05, "loss": 0.0242, "step": 59210 }, { "epoch": 4.97, "learning_rate": 1.0053745381256299e-05, "loss": 0.1163, "step": 59220 }, { "epoch": 4.97, "learning_rate": 1.005206583809204e-05, "loss": 0.0784, "step": 59230 }, { "epoch": 4.97, "learning_rate": 1.0050386294927781e-05, "loss": 0.0516, "step": 59240 }, { "epoch": 4.98, "learning_rate": 1.0048706751763522e-05, "loss": 0.0206, "step": 59250 }, { "epoch": 4.98, "learning_rate": 1.0047027208599262e-05, "loss": 0.0763, "step": 59260 }, { "epoch": 4.98, "learning_rate": 1.0045347665435003e-05, "loss": 0.0012, "step": 59270 }, { "epoch": 4.98, "learning_rate": 1.0043668122270742e-05, "loss": 0.0534, "step": 59280 }, { "epoch": 4.98, "learning_rate": 1.0041988579106484e-05, "loss": 0.0154, "step": 59290 }, { "epoch": 4.98, "learning_rate": 1.0040309035942224e-05, "loss": 0.056, "step": 59300 }, { "epoch": 4.98, "learning_rate": 1.0038629492777965e-05, "loss": 0.061, "step": 59310 }, { "epoch": 4.98, "learning_rate": 1.0036949949613705e-05, "loss": 0.1254, "step": 59320 }, { "epoch": 4.98, "learning_rate": 1.0035270406449447e-05, "loss": 0.0367, "step": 59330 }, { "epoch": 4.98, "learning_rate": 1.0033590863285188e-05, "loss": 0.0543, "step": 59340 }, { "epoch": 4.98, "learning_rate": 1.0031911320120928e-05, "loss": 0.0534, "step": 59350 }, { "epoch": 4.98, "learning_rate": 1.0030231776956669e-05, "loss": 0.0511, "step": 59360 }, { "epoch": 4.99, "learning_rate": 1.002855223379241e-05, "loss": 0.0723, "step": 59370 }, { "epoch": 4.99, "learning_rate": 1.0026872690628151e-05, "loss": 0.1163, "step": 59380 }, { "epoch": 4.99, "learning_rate": 1.002519314746389e-05, "loss": 0.0617, "step": 59390 }, { "epoch": 4.99, "learning_rate": 1.002351360429963e-05, "loss": 0.0716, "step": 59400 }, { "epoch": 4.99, "learning_rate": 1.0021834061135371e-05, "loss": 0.1011, "step": 59410 }, { "epoch": 4.99, "learning_rate": 1.0020154517971112e-05, "loss": 0.0974, "step": 59420 }, { "epoch": 4.99, "learning_rate": 1.0018474974806854e-05, "loss": 0.0157, "step": 59430 }, { "epoch": 4.99, "learning_rate": 1.0016795431642594e-05, "loss": 0.066, "step": 59440 }, { "epoch": 4.99, "learning_rate": 1.0015115888478335e-05, "loss": 0.017, "step": 59450 }, { "epoch": 4.99, "learning_rate": 1.0013436345314075e-05, "loss": 0.0286, "step": 59460 }, { "epoch": 4.99, "learning_rate": 1.0011756802149817e-05, "loss": 0.1256, "step": 59470 }, { "epoch": 4.99, "learning_rate": 1.0010077258985558e-05, "loss": 0.0507, "step": 59480 }, { "epoch": 5.0, "learning_rate": 1.0008397715821297e-05, "loss": 0.1489, "step": 59490 }, { "epoch": 5.0, "learning_rate": 1.0006718172657037e-05, "loss": 0.0845, "step": 59500 }, { "epoch": 5.0, "learning_rate": 1.0005038629492778e-05, "loss": 0.0426, "step": 59510 }, { "epoch": 5.0, "learning_rate": 1.000335908632852e-05, "loss": 0.0324, "step": 59520 }, { "epoch": 5.0, "learning_rate": 1.000167954316426e-05, "loss": 0.1934, "step": 59530 }, { "epoch": 5.0, "learning_rate": 1e-05, "loss": 0.0574, "step": 59540 }, { "epoch": 5.0, "eval_FN": 625, "eval_FP": 347, "eval_TN": 21367, "eval_TP": 1477, "eval_accuracy": 0.9591871011084985, "eval_f1": 0.7524197656647987, "eval_loss": 0.16406774520874023, "eval_precision": 0.8097587719298246, "eval_recall": 0.7026641294005709, "eval_runtime": 163.9444, "eval_samples_per_second": 145.269, "eval_steps_per_second": 9.082, "step": 59540 }, { "epoch": 5.0, "learning_rate": 9.998320456835741e-06, "loss": 0.003, "step": 59550 }, { "epoch": 5.0, "learning_rate": 9.996640913671482e-06, "loss": 0.0735, "step": 59560 }, { "epoch": 5.0, "learning_rate": 9.994961370507222e-06, "loss": 0.0561, "step": 59570 }, { "epoch": 5.0, "learning_rate": 9.993281827342963e-06, "loss": 0.0524, "step": 59580 }, { "epoch": 5.0, "learning_rate": 9.991602284178705e-06, "loss": 0.0262, "step": 59590 }, { "epoch": 5.01, "learning_rate": 9.989922741014445e-06, "loss": 0.0495, "step": 59600 }, { "epoch": 5.01, "learning_rate": 9.988243197850186e-06, "loss": 0.0349, "step": 59610 }, { "epoch": 5.01, "learning_rate": 9.986563654685926e-06, "loss": 0.0379, "step": 59620 }, { "epoch": 5.01, "learning_rate": 9.984884111521667e-06, "loss": 0.0672, "step": 59630 }, { "epoch": 5.01, "learning_rate": 9.983204568357407e-06, "loss": 0.0632, "step": 59640 }, { "epoch": 5.01, "learning_rate": 9.981525025193148e-06, "loss": 0.0947, "step": 59650 }, { "epoch": 5.01, "learning_rate": 9.97984548202889e-06, "loss": 0.0587, "step": 59660 }, { "epoch": 5.01, "learning_rate": 9.97816593886463e-06, "loss": 0.0411, "step": 59670 }, { "epoch": 5.01, "learning_rate": 9.97648639570037e-06, "loss": 0.0433, "step": 59680 }, { "epoch": 5.01, "learning_rate": 9.974806852536111e-06, "loss": 0.0414, "step": 59690 }, { "epoch": 5.01, "learning_rate": 9.973127309371852e-06, "loss": 0.004, "step": 59700 }, { "epoch": 5.01, "learning_rate": 9.971447766207592e-06, "loss": 0.0098, "step": 59710 }, { "epoch": 5.02, "learning_rate": 9.969768223043333e-06, "loss": 0.0173, "step": 59720 }, { "epoch": 5.02, "learning_rate": 9.968088679879073e-06, "loss": 0.0188, "step": 59730 }, { "epoch": 5.02, "learning_rate": 9.966409136714814e-06, "loss": 0.0124, "step": 59740 }, { "epoch": 5.02, "learning_rate": 9.964729593550554e-06, "loss": 0.2028, "step": 59750 }, { "epoch": 5.02, "learning_rate": 9.963050050386296e-06, "loss": 0.001, "step": 59760 }, { "epoch": 5.02, "learning_rate": 9.961370507222037e-06, "loss": 0.0901, "step": 59770 }, { "epoch": 5.02, "learning_rate": 9.959690964057777e-06, "loss": 0.0179, "step": 59780 }, { "epoch": 5.02, "learning_rate": 9.958011420893518e-06, "loss": 0.0035, "step": 59790 }, { "epoch": 5.02, "learning_rate": 9.956331877729258e-06, "loss": 0.0428, "step": 59800 }, { "epoch": 5.02, "learning_rate": 9.954652334564999e-06, "loss": 0.0604, "step": 59810 }, { "epoch": 5.02, "learning_rate": 9.952972791400741e-06, "loss": 0.0029, "step": 59820 }, { "epoch": 5.02, "learning_rate": 9.951293248236481e-06, "loss": 0.0009, "step": 59830 }, { "epoch": 5.03, "learning_rate": 9.94961370507222e-06, "loss": 0.0234, "step": 59840 }, { "epoch": 5.03, "learning_rate": 9.947934161907962e-06, "loss": 0.0031, "step": 59850 }, { "epoch": 5.03, "learning_rate": 9.946254618743703e-06, "loss": 0.0245, "step": 59860 }, { "epoch": 5.03, "learning_rate": 9.944575075579443e-06, "loss": 0.0011, "step": 59870 }, { "epoch": 5.03, "learning_rate": 9.942895532415184e-06, "loss": 0.0006, "step": 59880 }, { "epoch": 5.03, "learning_rate": 9.941215989250924e-06, "loss": 0.0326, "step": 59890 }, { "epoch": 5.03, "learning_rate": 9.939536446086665e-06, "loss": 0.0167, "step": 59900 }, { "epoch": 5.03, "learning_rate": 9.937856902922405e-06, "loss": 0.0373, "step": 59910 }, { "epoch": 5.03, "learning_rate": 9.936177359758147e-06, "loss": 0.0971, "step": 59920 }, { "epoch": 5.03, "learning_rate": 9.934497816593888e-06, "loss": 0.1318, "step": 59930 }, { "epoch": 5.03, "learning_rate": 9.932818273429627e-06, "loss": 0.0004, "step": 59940 }, { "epoch": 5.03, "learning_rate": 9.931138730265369e-06, "loss": 0.0788, "step": 59950 }, { "epoch": 5.04, "learning_rate": 9.92945918710111e-06, "loss": 0.0677, "step": 59960 }, { "epoch": 5.04, "learning_rate": 9.92777964393685e-06, "loss": 0.0323, "step": 59970 }, { "epoch": 5.04, "learning_rate": 9.92610010077259e-06, "loss": 0.0807, "step": 59980 }, { "epoch": 5.04, "learning_rate": 9.924420557608332e-06, "loss": 0.0049, "step": 59990 }, { "epoch": 5.04, "learning_rate": 9.922741014444071e-06, "loss": 0.0952, "step": 60000 }, { "epoch": 5.04, "learning_rate": 9.921061471279813e-06, "loss": 0.0596, "step": 60010 }, { "epoch": 5.04, "learning_rate": 9.919381928115554e-06, "loss": 0.067, "step": 60020 }, { "epoch": 5.04, "learning_rate": 9.917702384951294e-06, "loss": 0.1255, "step": 60030 }, { "epoch": 5.04, "learning_rate": 9.916022841787035e-06, "loss": 0.0176, "step": 60040 }, { "epoch": 5.04, "learning_rate": 9.914343298622775e-06, "loss": 0.0226, "step": 60050 }, { "epoch": 5.04, "learning_rate": 9.912663755458516e-06, "loss": 0.0181, "step": 60060 }, { "epoch": 5.04, "learning_rate": 9.910984212294256e-06, "loss": 0.0689, "step": 60070 }, { "epoch": 5.05, "learning_rate": 9.909304669129998e-06, "loss": 0.0384, "step": 60080 }, { "epoch": 5.05, "learning_rate": 9.907625125965739e-06, "loss": 0.0015, "step": 60090 }, { "epoch": 5.05, "learning_rate": 9.905945582801478e-06, "loss": 0.0432, "step": 60100 }, { "epoch": 5.05, "learning_rate": 9.90426603963722e-06, "loss": 0.042, "step": 60110 }, { "epoch": 5.05, "learning_rate": 9.90258649647296e-06, "loss": 0.0312, "step": 60120 }, { "epoch": 5.05, "learning_rate": 9.9009069533087e-06, "loss": 0.0899, "step": 60130 }, { "epoch": 5.05, "learning_rate": 9.899227410144441e-06, "loss": 0.1547, "step": 60140 }, { "epoch": 5.05, "learning_rate": 9.897547866980182e-06, "loss": 0.0332, "step": 60150 }, { "epoch": 5.05, "learning_rate": 9.895868323815922e-06, "loss": 0.0298, "step": 60160 }, { "epoch": 5.05, "learning_rate": 9.894188780651663e-06, "loss": 0.0098, "step": 60170 }, { "epoch": 5.05, "learning_rate": 9.892509237487405e-06, "loss": 0.0041, "step": 60180 }, { "epoch": 5.05, "learning_rate": 9.890829694323145e-06, "loss": 0.0315, "step": 60190 }, { "epoch": 5.06, "learning_rate": 9.889150151158886e-06, "loss": 0.047, "step": 60200 }, { "epoch": 5.06, "learning_rate": 9.887470607994626e-06, "loss": 0.0441, "step": 60210 }, { "epoch": 5.06, "learning_rate": 9.885791064830367e-06, "loss": 0.1022, "step": 60220 }, { "epoch": 5.06, "learning_rate": 9.884111521666107e-06, "loss": 0.1045, "step": 60230 }, { "epoch": 5.06, "learning_rate": 9.882431978501848e-06, "loss": 0.0424, "step": 60240 }, { "epoch": 5.06, "learning_rate": 9.88075243533759e-06, "loss": 0.0668, "step": 60250 }, { "epoch": 5.06, "learning_rate": 9.879072892173329e-06, "loss": 0.0238, "step": 60260 }, { "epoch": 5.06, "learning_rate": 9.877393349009071e-06, "loss": 0.0334, "step": 60270 }, { "epoch": 5.06, "learning_rate": 9.875713805844811e-06, "loss": 0.0427, "step": 60280 }, { "epoch": 5.06, "learning_rate": 9.874034262680552e-06, "loss": 0.053, "step": 60290 }, { "epoch": 5.06, "learning_rate": 9.872354719516292e-06, "loss": 0.0773, "step": 60300 }, { "epoch": 5.06, "learning_rate": 9.870675176352033e-06, "loss": 0.0733, "step": 60310 }, { "epoch": 5.07, "learning_rate": 9.868995633187773e-06, "loss": 0.0493, "step": 60320 }, { "epoch": 5.07, "learning_rate": 9.867316090023514e-06, "loss": 0.024, "step": 60330 }, { "epoch": 5.07, "learning_rate": 9.865636546859256e-06, "loss": 0.0685, "step": 60340 }, { "epoch": 5.07, "learning_rate": 9.863957003694996e-06, "loss": 0.0451, "step": 60350 }, { "epoch": 5.07, "learning_rate": 9.862277460530737e-06, "loss": 0.0448, "step": 60360 }, { "epoch": 5.07, "learning_rate": 9.860597917366477e-06, "loss": 0.0353, "step": 60370 }, { "epoch": 5.07, "learning_rate": 9.858918374202218e-06, "loss": 0.0235, "step": 60380 }, { "epoch": 5.07, "learning_rate": 9.857238831037958e-06, "loss": 0.0373, "step": 60390 }, { "epoch": 5.07, "learning_rate": 9.855559287873699e-06, "loss": 0.0492, "step": 60400 }, { "epoch": 5.07, "learning_rate": 9.853879744709441e-06, "loss": 0.0687, "step": 60410 }, { "epoch": 5.07, "learning_rate": 9.85220020154518e-06, "loss": 0.0055, "step": 60420 }, { "epoch": 5.07, "learning_rate": 9.85052065838092e-06, "loss": 0.035, "step": 60430 }, { "epoch": 5.08, "learning_rate": 9.848841115216662e-06, "loss": 0.1254, "step": 60440 }, { "epoch": 5.08, "learning_rate": 9.847161572052403e-06, "loss": 0.0736, "step": 60450 }, { "epoch": 5.08, "learning_rate": 9.845482028888143e-06, "loss": 0.0932, "step": 60460 }, { "epoch": 5.08, "learning_rate": 9.843802485723884e-06, "loss": 0.1435, "step": 60470 }, { "epoch": 5.08, "learning_rate": 9.842122942559624e-06, "loss": 0.0818, "step": 60480 }, { "epoch": 5.08, "learning_rate": 9.840443399395365e-06, "loss": 0.0026, "step": 60490 }, { "epoch": 5.08, "learning_rate": 9.838763856231107e-06, "loss": 0.0448, "step": 60500 }, { "epoch": 5.08, "learning_rate": 9.837084313066847e-06, "loss": 0.0307, "step": 60510 }, { "epoch": 5.08, "learning_rate": 9.835404769902586e-06, "loss": 0.0022, "step": 60520 }, { "epoch": 5.08, "learning_rate": 9.833725226738328e-06, "loss": 0.0085, "step": 60530 }, { "epoch": 5.08, "learning_rate": 9.832045683574069e-06, "loss": 0.0526, "step": 60540 }, { "epoch": 5.08, "learning_rate": 9.83036614040981e-06, "loss": 0.1362, "step": 60550 }, { "epoch": 5.09, "learning_rate": 9.82868659724555e-06, "loss": 0.0543, "step": 60560 }, { "epoch": 5.09, "learning_rate": 9.827007054081292e-06, "loss": 0.0827, "step": 60570 }, { "epoch": 5.09, "learning_rate": 9.82532751091703e-06, "loss": 0.012, "step": 60580 }, { "epoch": 5.09, "learning_rate": 9.823647967752771e-06, "loss": 0.098, "step": 60590 }, { "epoch": 5.09, "learning_rate": 9.821968424588513e-06, "loss": 0.0399, "step": 60600 }, { "epoch": 5.09, "learning_rate": 9.820288881424254e-06, "loss": 0.0242, "step": 60610 }, { "epoch": 5.09, "learning_rate": 9.818609338259994e-06, "loss": 0.0519, "step": 60620 }, { "epoch": 5.09, "learning_rate": 9.816929795095735e-06, "loss": 0.0107, "step": 60630 }, { "epoch": 5.09, "learning_rate": 9.815250251931475e-06, "loss": 0.0024, "step": 60640 }, { "epoch": 5.09, "learning_rate": 9.813570708767216e-06, "loss": 0.0349, "step": 60650 }, { "epoch": 5.09, "learning_rate": 9.811891165602956e-06, "loss": 0.0204, "step": 60660 }, { "epoch": 5.09, "learning_rate": 9.810211622438698e-06, "loss": 0.0265, "step": 60670 }, { "epoch": 5.1, "learning_rate": 9.808532079274437e-06, "loss": 0.0404, "step": 60680 }, { "epoch": 5.1, "learning_rate": 9.80685253611018e-06, "loss": 0.0892, "step": 60690 }, { "epoch": 5.1, "learning_rate": 9.80517299294592e-06, "loss": 0.0009, "step": 60700 }, { "epoch": 5.1, "learning_rate": 9.80349344978166e-06, "loss": 0.0007, "step": 60710 }, { "epoch": 5.1, "learning_rate": 9.8018139066174e-06, "loss": 0.0751, "step": 60720 }, { "epoch": 5.1, "learning_rate": 9.800134363453141e-06, "loss": 0.0261, "step": 60730 }, { "epoch": 5.1, "learning_rate": 9.798454820288882e-06, "loss": 0.08, "step": 60740 }, { "epoch": 5.1, "learning_rate": 9.796775277124622e-06, "loss": 0.0595, "step": 60750 }, { "epoch": 5.1, "learning_rate": 9.795095733960364e-06, "loss": 0.0739, "step": 60760 }, { "epoch": 5.1, "learning_rate": 9.793416190796105e-06, "loss": 0.093, "step": 60770 }, { "epoch": 5.1, "learning_rate": 9.791736647631845e-06, "loss": 0.0339, "step": 60780 }, { "epoch": 5.1, "learning_rate": 9.790057104467586e-06, "loss": 0.0278, "step": 60790 }, { "epoch": 5.11, "learning_rate": 9.788377561303326e-06, "loss": 0.04, "step": 60800 }, { "epoch": 5.11, "learning_rate": 9.786698018139067e-06, "loss": 0.0435, "step": 60810 }, { "epoch": 5.11, "learning_rate": 9.785018474974807e-06, "loss": 0.0151, "step": 60820 }, { "epoch": 5.11, "learning_rate": 9.78333893181055e-06, "loss": 0.0818, "step": 60830 }, { "epoch": 5.11, "learning_rate": 9.781659388646288e-06, "loss": 0.041, "step": 60840 }, { "epoch": 5.11, "learning_rate": 9.779979845482029e-06, "loss": 0.0355, "step": 60850 }, { "epoch": 5.11, "learning_rate": 9.77830030231777e-06, "loss": 0.0721, "step": 60860 }, { "epoch": 5.11, "learning_rate": 9.776620759153511e-06, "loss": 0.0275, "step": 60870 }, { "epoch": 5.11, "learning_rate": 9.774941215989252e-06, "loss": 0.0011, "step": 60880 }, { "epoch": 5.11, "learning_rate": 9.773261672824992e-06, "loss": 0.0324, "step": 60890 }, { "epoch": 5.11, "learning_rate": 9.771582129660733e-06, "loss": 0.0427, "step": 60900 }, { "epoch": 5.12, "learning_rate": 9.769902586496473e-06, "loss": 0.0885, "step": 60910 }, { "epoch": 5.12, "learning_rate": 9.768223043332214e-06, "loss": 0.0179, "step": 60920 }, { "epoch": 5.12, "learning_rate": 9.766543500167956e-06, "loss": 0.0251, "step": 60930 }, { "epoch": 5.12, "learning_rate": 9.764863957003696e-06, "loss": 0.0697, "step": 60940 }, { "epoch": 5.12, "learning_rate": 9.763184413839437e-06, "loss": 0.1026, "step": 60950 }, { "epoch": 5.12, "learning_rate": 9.761504870675177e-06, "loss": 0.052, "step": 60960 }, { "epoch": 5.12, "learning_rate": 9.759825327510918e-06, "loss": 0.0015, "step": 60970 }, { "epoch": 5.12, "learning_rate": 9.758145784346658e-06, "loss": 0.0259, "step": 60980 }, { "epoch": 5.12, "learning_rate": 9.7564662411824e-06, "loss": 0.0603, "step": 60990 }, { "epoch": 5.12, "learning_rate": 9.75478669801814e-06, "loss": 0.0217, "step": 61000 }, { "epoch": 5.12, "learning_rate": 9.75310715485388e-06, "loss": 0.0551, "step": 61010 }, { "epoch": 5.12, "learning_rate": 9.751427611689622e-06, "loss": 0.0217, "step": 61020 }, { "epoch": 5.13, "learning_rate": 9.749748068525362e-06, "loss": 0.0571, "step": 61030 }, { "epoch": 5.13, "learning_rate": 9.748068525361103e-06, "loss": 0.0003, "step": 61040 }, { "epoch": 5.13, "learning_rate": 9.746388982196843e-06, "loss": 0.0074, "step": 61050 }, { "epoch": 5.13, "learning_rate": 9.744709439032584e-06, "loss": 0.0185, "step": 61060 }, { "epoch": 5.13, "learning_rate": 9.743029895868324e-06, "loss": 0.0197, "step": 61070 }, { "epoch": 5.13, "learning_rate": 9.741350352704065e-06, "loss": 0.0005, "step": 61080 }, { "epoch": 5.13, "learning_rate": 9.739670809539807e-06, "loss": 0.0002, "step": 61090 }, { "epoch": 5.13, "learning_rate": 9.737991266375547e-06, "loss": 0.0401, "step": 61100 }, { "epoch": 5.13, "learning_rate": 9.736311723211286e-06, "loss": 0.0787, "step": 61110 }, { "epoch": 5.13, "learning_rate": 9.734632180047028e-06, "loss": 0.0008, "step": 61120 }, { "epoch": 5.13, "learning_rate": 9.732952636882769e-06, "loss": 0.0583, "step": 61130 }, { "epoch": 5.13, "learning_rate": 9.73127309371851e-06, "loss": 0.0667, "step": 61140 }, { "epoch": 5.14, "learning_rate": 9.72959355055425e-06, "loss": 0.0005, "step": 61150 }, { "epoch": 5.14, "learning_rate": 9.72791400738999e-06, "loss": 0.0683, "step": 61160 }, { "epoch": 5.14, "learning_rate": 9.72623446422573e-06, "loss": 0.0619, "step": 61170 }, { "epoch": 5.14, "learning_rate": 9.724554921061473e-06, "loss": 0.0004, "step": 61180 }, { "epoch": 5.14, "learning_rate": 9.722875377897213e-06, "loss": 0.1081, "step": 61190 }, { "epoch": 5.14, "learning_rate": 9.721195834732954e-06, "loss": 0.0489, "step": 61200 }, { "epoch": 5.14, "learning_rate": 9.719516291568694e-06, "loss": 0.0015, "step": 61210 }, { "epoch": 5.14, "learning_rate": 9.717836748404435e-06, "loss": 0.021, "step": 61220 }, { "epoch": 5.14, "learning_rate": 9.716157205240175e-06, "loss": 0.0229, "step": 61230 }, { "epoch": 5.14, "learning_rate": 9.714477662075916e-06, "loss": 0.0473, "step": 61240 }, { "epoch": 5.14, "learning_rate": 9.712798118911658e-06, "loss": 0.0253, "step": 61250 }, { "epoch": 5.14, "learning_rate": 9.711118575747398e-06, "loss": 0.0554, "step": 61260 }, { "epoch": 5.15, "learning_rate": 9.709439032583137e-06, "loss": 0.0034, "step": 61270 }, { "epoch": 5.15, "learning_rate": 9.70775948941888e-06, "loss": 0.1339, "step": 61280 }, { "epoch": 5.15, "learning_rate": 9.70607994625462e-06, "loss": 0.0672, "step": 61290 }, { "epoch": 5.15, "learning_rate": 9.70440040309036e-06, "loss": 0.0299, "step": 61300 }, { "epoch": 5.15, "learning_rate": 9.7027208599261e-06, "loss": 0.0004, "step": 61310 }, { "epoch": 5.15, "learning_rate": 9.701041316761841e-06, "loss": 0.0239, "step": 61320 }, { "epoch": 5.15, "learning_rate": 9.699361773597582e-06, "loss": 0.0317, "step": 61330 }, { "epoch": 5.15, "learning_rate": 9.697682230433322e-06, "loss": 0.0004, "step": 61340 }, { "epoch": 5.15, "learning_rate": 9.696002687269064e-06, "loss": 0.0972, "step": 61350 }, { "epoch": 5.15, "learning_rate": 9.694323144104805e-06, "loss": 0.0199, "step": 61360 }, { "epoch": 5.15, "learning_rate": 9.692643600940544e-06, "loss": 0.0629, "step": 61370 }, { "epoch": 5.15, "learning_rate": 9.690964057776286e-06, "loss": 0.048, "step": 61380 }, { "epoch": 5.16, "learning_rate": 9.689284514612026e-06, "loss": 0.0665, "step": 61390 }, { "epoch": 5.16, "learning_rate": 9.687604971447767e-06, "loss": 0.0517, "step": 61400 }, { "epoch": 5.16, "learning_rate": 9.685925428283507e-06, "loss": 0.0845, "step": 61410 }, { "epoch": 5.16, "learning_rate": 9.68424588511925e-06, "loss": 0.0609, "step": 61420 }, { "epoch": 5.16, "learning_rate": 9.682566341954988e-06, "loss": 0.0022, "step": 61430 }, { "epoch": 5.16, "learning_rate": 9.68088679879073e-06, "loss": 0.0903, "step": 61440 }, { "epoch": 5.16, "learning_rate": 9.67920725562647e-06, "loss": 0.0222, "step": 61450 }, { "epoch": 5.16, "learning_rate": 9.677527712462211e-06, "loss": 0.0492, "step": 61460 }, { "epoch": 5.16, "learning_rate": 9.675848169297952e-06, "loss": 0.0312, "step": 61470 }, { "epoch": 5.16, "learning_rate": 9.674168626133692e-06, "loss": 0.0747, "step": 61480 }, { "epoch": 5.16, "learning_rate": 9.672489082969433e-06, "loss": 0.0834, "step": 61490 }, { "epoch": 5.16, "learning_rate": 9.670809539805173e-06, "loss": 0.0853, "step": 61500 }, { "epoch": 5.17, "learning_rate": 9.669129996640915e-06, "loss": 0.0506, "step": 61510 }, { "epoch": 5.17, "learning_rate": 9.667450453476656e-06, "loss": 0.0234, "step": 61520 }, { "epoch": 5.17, "learning_rate": 9.665770910312395e-06, "loss": 0.0697, "step": 61530 }, { "epoch": 5.17, "learning_rate": 9.664091367148137e-06, "loss": 0.0757, "step": 61540 }, { "epoch": 5.17, "learning_rate": 9.662411823983877e-06, "loss": 0.0258, "step": 61550 }, { "epoch": 5.17, "learning_rate": 9.660732280819618e-06, "loss": 0.0301, "step": 61560 }, { "epoch": 5.17, "learning_rate": 9.659052737655358e-06, "loss": 0.0201, "step": 61570 }, { "epoch": 5.17, "learning_rate": 9.657373194491099e-06, "loss": 0.0342, "step": 61580 }, { "epoch": 5.17, "learning_rate": 9.65569365132684e-06, "loss": 0.0221, "step": 61590 }, { "epoch": 5.17, "learning_rate": 9.65401410816258e-06, "loss": 0.0018, "step": 61600 }, { "epoch": 5.17, "learning_rate": 9.652334564998322e-06, "loss": 0.0009, "step": 61610 }, { "epoch": 5.17, "learning_rate": 9.650655021834062e-06, "loss": 0.0925, "step": 61620 }, { "epoch": 5.18, "learning_rate": 9.648975478669803e-06, "loss": 0.0853, "step": 61630 }, { "epoch": 5.18, "learning_rate": 9.647295935505543e-06, "loss": 0.0658, "step": 61640 }, { "epoch": 5.18, "learning_rate": 9.645616392341284e-06, "loss": 0.0571, "step": 61650 }, { "epoch": 5.18, "learning_rate": 9.643936849177024e-06, "loss": 0.0549, "step": 61660 }, { "epoch": 5.18, "learning_rate": 9.642257306012766e-06, "loss": 0.0262, "step": 61670 }, { "epoch": 5.18, "learning_rate": 9.640577762848507e-06, "loss": 0.0166, "step": 61680 }, { "epoch": 5.18, "learning_rate": 9.638898219684246e-06, "loss": 0.0619, "step": 61690 }, { "epoch": 5.18, "learning_rate": 9.637218676519988e-06, "loss": 0.0611, "step": 61700 }, { "epoch": 5.18, "learning_rate": 9.635539133355728e-06, "loss": 0.0737, "step": 61710 }, { "epoch": 5.18, "learning_rate": 9.633859590191469e-06, "loss": 0.0477, "step": 61720 }, { "epoch": 5.18, "learning_rate": 9.63218004702721e-06, "loss": 0.0228, "step": 61730 }, { "epoch": 5.18, "learning_rate": 9.63050050386295e-06, "loss": 0.0322, "step": 61740 }, { "epoch": 5.19, "learning_rate": 9.62882096069869e-06, "loss": 0.0025, "step": 61750 }, { "epoch": 5.19, "learning_rate": 9.62714141753443e-06, "loss": 0.0565, "step": 61760 }, { "epoch": 5.19, "learning_rate": 9.625461874370173e-06, "loss": 0.038, "step": 61770 }, { "epoch": 5.19, "learning_rate": 9.623782331205913e-06, "loss": 0.0515, "step": 61780 }, { "epoch": 5.19, "learning_rate": 9.622102788041654e-06, "loss": 0.0235, "step": 61790 }, { "epoch": 5.19, "learning_rate": 9.620423244877394e-06, "loss": 0.0845, "step": 61800 }, { "epoch": 5.19, "learning_rate": 9.618743701713135e-06, "loss": 0.1186, "step": 61810 }, { "epoch": 5.19, "learning_rate": 9.617064158548875e-06, "loss": 0.0915, "step": 61820 }, { "epoch": 5.19, "learning_rate": 9.615384615384616e-06, "loss": 0.0032, "step": 61830 }, { "epoch": 5.19, "learning_rate": 9.613705072220358e-06, "loss": 0.0695, "step": 61840 }, { "epoch": 5.19, "learning_rate": 9.612025529056097e-06, "loss": 0.0291, "step": 61850 }, { "epoch": 5.19, "learning_rate": 9.610345985891837e-06, "loss": 0.065, "step": 61860 }, { "epoch": 5.2, "learning_rate": 9.60866644272758e-06, "loss": 0.0495, "step": 61870 }, { "epoch": 5.2, "learning_rate": 9.60698689956332e-06, "loss": 0.035, "step": 61880 }, { "epoch": 5.2, "learning_rate": 9.60530735639906e-06, "loss": 0.0435, "step": 61890 }, { "epoch": 5.2, "learning_rate": 9.6036278132348e-06, "loss": 0.0805, "step": 61900 }, { "epoch": 5.2, "learning_rate": 9.601948270070541e-06, "loss": 0.0025, "step": 61910 }, { "epoch": 5.2, "learning_rate": 9.600268726906282e-06, "loss": 0.046, "step": 61920 }, { "epoch": 5.2, "learning_rate": 9.598589183742024e-06, "loss": 0.0451, "step": 61930 }, { "epoch": 5.2, "learning_rate": 9.596909640577764e-06, "loss": 0.0858, "step": 61940 }, { "epoch": 5.2, "learning_rate": 9.595230097413505e-06, "loss": 0.0656, "step": 61950 }, { "epoch": 5.2, "learning_rate": 9.593550554249245e-06, "loss": 0.0833, "step": 61960 }, { "epoch": 5.2, "learning_rate": 9.591871011084986e-06, "loss": 0.0944, "step": 61970 }, { "epoch": 5.2, "learning_rate": 9.590191467920726e-06, "loss": 0.0236, "step": 61980 }, { "epoch": 5.21, "learning_rate": 9.588511924756467e-06, "loss": 0.064, "step": 61990 }, { "epoch": 5.21, "learning_rate": 9.586832381592209e-06, "loss": 0.0625, "step": 62000 }, { "epoch": 5.21, "learning_rate": 9.585152838427948e-06, "loss": 0.021, "step": 62010 }, { "epoch": 5.21, "learning_rate": 9.583473295263688e-06, "loss": 0.035, "step": 62020 }, { "epoch": 5.21, "learning_rate": 9.58179375209943e-06, "loss": 0.0016, "step": 62030 }, { "epoch": 5.21, "learning_rate": 9.58011420893517e-06, "loss": 0.0016, "step": 62040 }, { "epoch": 5.21, "learning_rate": 9.578434665770911e-06, "loss": 0.16, "step": 62050 }, { "epoch": 5.21, "learning_rate": 9.576755122606652e-06, "loss": 0.018, "step": 62060 }, { "epoch": 5.21, "learning_rate": 9.575075579442392e-06, "loss": 0.0425, "step": 62070 }, { "epoch": 5.21, "learning_rate": 9.573396036278133e-06, "loss": 0.0269, "step": 62080 }, { "epoch": 5.21, "learning_rate": 9.571716493113873e-06, "loss": 0.0345, "step": 62090 }, { "epoch": 5.21, "learning_rate": 9.570036949949615e-06, "loss": 0.0217, "step": 62100 }, { "epoch": 5.22, "learning_rate": 9.568357406785354e-06, "loss": 0.0415, "step": 62110 }, { "epoch": 5.22, "learning_rate": 9.566677863621096e-06, "loss": 0.0538, "step": 62120 }, { "epoch": 5.22, "learning_rate": 9.564998320456837e-06, "loss": 0.0078, "step": 62130 }, { "epoch": 5.22, "learning_rate": 9.563318777292577e-06, "loss": 0.1657, "step": 62140 }, { "epoch": 5.22, "learning_rate": 9.561639234128318e-06, "loss": 0.138, "step": 62150 }, { "epoch": 5.22, "learning_rate": 9.55995969096406e-06, "loss": 0.0583, "step": 62160 }, { "epoch": 5.22, "learning_rate": 9.558280147799799e-06, "loss": 0.0301, "step": 62170 }, { "epoch": 5.22, "learning_rate": 9.55660060463554e-06, "loss": 0.0144, "step": 62180 }, { "epoch": 5.22, "learning_rate": 9.554921061471281e-06, "loss": 0.0823, "step": 62190 }, { "epoch": 5.22, "learning_rate": 9.553241518307022e-06, "loss": 0.0562, "step": 62200 }, { "epoch": 5.22, "learning_rate": 9.551561975142762e-06, "loss": 0.0597, "step": 62210 }, { "epoch": 5.23, "learning_rate": 9.549882431978503e-06, "loss": 0.0462, "step": 62220 }, { "epoch": 5.23, "learning_rate": 9.548202888814243e-06, "loss": 0.0275, "step": 62230 }, { "epoch": 5.23, "learning_rate": 9.546523345649984e-06, "loss": 0.0319, "step": 62240 }, { "epoch": 5.23, "learning_rate": 9.544843802485724e-06, "loss": 0.0483, "step": 62250 }, { "epoch": 5.23, "learning_rate": 9.543164259321466e-06, "loss": 0.0288, "step": 62260 }, { "epoch": 5.23, "learning_rate": 9.541484716157205e-06, "loss": 0.0416, "step": 62270 }, { "epoch": 5.23, "learning_rate": 9.539805172992946e-06, "loss": 0.0514, "step": 62280 }, { "epoch": 5.23, "learning_rate": 9.538125629828688e-06, "loss": 0.0365, "step": 62290 }, { "epoch": 5.23, "learning_rate": 9.536446086664428e-06, "loss": 0.0751, "step": 62300 }, { "epoch": 5.23, "learning_rate": 9.534766543500169e-06, "loss": 0.0565, "step": 62310 }, { "epoch": 5.23, "learning_rate": 9.53308700033591e-06, "loss": 0.0707, "step": 62320 }, { "epoch": 5.23, "learning_rate": 9.53140745717165e-06, "loss": 0.0838, "step": 62330 }, { "epoch": 5.24, "learning_rate": 9.52972791400739e-06, "loss": 0.0745, "step": 62340 }, { "epoch": 5.24, "learning_rate": 9.52804837084313e-06, "loss": 0.0257, "step": 62350 }, { "epoch": 5.24, "learning_rate": 9.526368827678873e-06, "loss": 0.0561, "step": 62360 }, { "epoch": 5.24, "learning_rate": 9.524689284514613e-06, "loss": 0.0603, "step": 62370 }, { "epoch": 5.24, "learning_rate": 9.523009741350354e-06, "loss": 0.0754, "step": 62380 }, { "epoch": 5.24, "learning_rate": 9.521330198186094e-06, "loss": 0.0491, "step": 62390 }, { "epoch": 5.24, "learning_rate": 9.519650655021835e-06, "loss": 0.0196, "step": 62400 }, { "epoch": 5.24, "learning_rate": 9.517971111857575e-06, "loss": 0.0015, "step": 62410 }, { "epoch": 5.24, "learning_rate": 9.516291568693317e-06, "loss": 0.0415, "step": 62420 }, { "epoch": 5.24, "learning_rate": 9.514612025529056e-06, "loss": 0.0239, "step": 62430 }, { "epoch": 5.24, "learning_rate": 9.512932482364797e-06, "loss": 0.0479, "step": 62440 }, { "epoch": 5.24, "learning_rate": 9.511252939200539e-06, "loss": 0.0006, "step": 62450 }, { "epoch": 5.25, "learning_rate": 9.50957339603628e-06, "loss": 0.0694, "step": 62460 }, { "epoch": 5.25, "learning_rate": 9.50789385287202e-06, "loss": 0.0012, "step": 62470 }, { "epoch": 5.25, "learning_rate": 9.50621430970776e-06, "loss": 0.049, "step": 62480 }, { "epoch": 5.25, "learning_rate": 9.5045347665435e-06, "loss": 0.0855, "step": 62490 }, { "epoch": 5.25, "learning_rate": 9.502855223379241e-06, "loss": 0.0367, "step": 62500 }, { "epoch": 5.25, "learning_rate": 9.501175680214982e-06, "loss": 0.0064, "step": 62510 }, { "epoch": 5.25, "learning_rate": 9.499496137050724e-06, "loss": 0.0463, "step": 62520 }, { "epoch": 5.25, "learning_rate": 9.497816593886464e-06, "loss": 0.0016, "step": 62530 }, { "epoch": 5.25, "learning_rate": 9.496137050722203e-06, "loss": 0.0434, "step": 62540 }, { "epoch": 5.25, "learning_rate": 9.494457507557945e-06, "loss": 0.0549, "step": 62550 }, { "epoch": 5.25, "learning_rate": 9.492777964393686e-06, "loss": 0.0346, "step": 62560 }, { "epoch": 5.25, "learning_rate": 9.491098421229426e-06, "loss": 0.0446, "step": 62570 }, { "epoch": 5.26, "learning_rate": 9.489418878065167e-06, "loss": 0.0989, "step": 62580 }, { "epoch": 5.26, "learning_rate": 9.487739334900907e-06, "loss": 0.0229, "step": 62590 }, { "epoch": 5.26, "learning_rate": 9.486059791736648e-06, "loss": 0.0495, "step": 62600 }, { "epoch": 5.26, "learning_rate": 9.48438024857239e-06, "loss": 0.068, "step": 62610 }, { "epoch": 5.26, "learning_rate": 9.48270070540813e-06, "loss": 0.0068, "step": 62620 }, { "epoch": 5.26, "learning_rate": 9.48102116224387e-06, "loss": 0.0908, "step": 62630 }, { "epoch": 5.26, "learning_rate": 9.479341619079611e-06, "loss": 0.047, "step": 62640 }, { "epoch": 5.26, "learning_rate": 9.477662075915352e-06, "loss": 0.0328, "step": 62650 }, { "epoch": 5.26, "learning_rate": 9.475982532751092e-06, "loss": 0.0851, "step": 62660 }, { "epoch": 5.26, "learning_rate": 9.474302989586833e-06, "loss": 0.0007, "step": 62670 }, { "epoch": 5.26, "learning_rate": 9.472623446422575e-06, "loss": 0.0393, "step": 62680 }, { "epoch": 5.26, "learning_rate": 9.470943903258315e-06, "loss": 0.0218, "step": 62690 }, { "epoch": 5.27, "learning_rate": 9.469264360094054e-06, "loss": 0.0235, "step": 62700 }, { "epoch": 5.27, "learning_rate": 9.467584816929796e-06, "loss": 0.0777, "step": 62710 }, { "epoch": 5.27, "learning_rate": 9.465905273765537e-06, "loss": 0.0301, "step": 62720 }, { "epoch": 5.27, "learning_rate": 9.464225730601277e-06, "loss": 0.0702, "step": 62730 }, { "epoch": 5.27, "learning_rate": 9.462546187437018e-06, "loss": 0.1158, "step": 62740 }, { "epoch": 5.27, "learning_rate": 9.460866644272758e-06, "loss": 0.0778, "step": 62750 }, { "epoch": 5.27, "learning_rate": 9.459187101108499e-06, "loss": 0.2147, "step": 62760 }, { "epoch": 5.27, "learning_rate": 9.45750755794424e-06, "loss": 0.0708, "step": 62770 }, { "epoch": 5.27, "learning_rate": 9.455828014779981e-06, "loss": 0.0273, "step": 62780 }, { "epoch": 5.27, "learning_rate": 9.454148471615722e-06, "loss": 0.0754, "step": 62790 }, { "epoch": 5.27, "learning_rate": 9.452468928451462e-06, "loss": 0.0118, "step": 62800 }, { "epoch": 5.27, "learning_rate": 9.450789385287203e-06, "loss": 0.0473, "step": 62810 }, { "epoch": 5.28, "learning_rate": 9.449109842122943e-06, "loss": 0.1118, "step": 62820 }, { "epoch": 5.28, "learning_rate": 9.447430298958684e-06, "loss": 0.0397, "step": 62830 }, { "epoch": 5.28, "learning_rate": 9.445750755794424e-06, "loss": 0.1844, "step": 62840 }, { "epoch": 5.28, "learning_rate": 9.444071212630166e-06, "loss": 0.0834, "step": 62850 }, { "epoch": 5.28, "learning_rate": 9.442391669465905e-06, "loss": 0.0895, "step": 62860 }, { "epoch": 5.28, "learning_rate": 9.440712126301647e-06, "loss": 0.0195, "step": 62870 }, { "epoch": 5.28, "learning_rate": 9.439032583137388e-06, "loss": 0.0465, "step": 62880 }, { "epoch": 5.28, "learning_rate": 9.437353039973128e-06, "loss": 0.0185, "step": 62890 }, { "epoch": 5.28, "learning_rate": 9.435673496808869e-06, "loss": 0.0462, "step": 62900 }, { "epoch": 5.28, "learning_rate": 9.43399395364461e-06, "loss": 0.0718, "step": 62910 }, { "epoch": 5.28, "learning_rate": 9.43231441048035e-06, "loss": 0.0465, "step": 62920 }, { "epoch": 5.28, "learning_rate": 9.43063486731609e-06, "loss": 0.0015, "step": 62930 }, { "epoch": 5.29, "learning_rate": 9.428955324151832e-06, "loss": 0.0965, "step": 62940 }, { "epoch": 5.29, "learning_rate": 9.427275780987573e-06, "loss": 0.1025, "step": 62950 }, { "epoch": 5.29, "learning_rate": 9.425596237823312e-06, "loss": 0.0043, "step": 62960 }, { "epoch": 5.29, "learning_rate": 9.423916694659054e-06, "loss": 0.0223, "step": 62970 }, { "epoch": 5.29, "learning_rate": 9.422237151494794e-06, "loss": 0.0019, "step": 62980 }, { "epoch": 5.29, "learning_rate": 9.420557608330535e-06, "loss": 0.0968, "step": 62990 }, { "epoch": 5.29, "learning_rate": 9.418878065166275e-06, "loss": 0.0888, "step": 63000 }, { "epoch": 5.29, "learning_rate": 9.417198522002016e-06, "loss": 0.0148, "step": 63010 }, { "epoch": 5.29, "learning_rate": 9.415518978837756e-06, "loss": 0.0009, "step": 63020 }, { "epoch": 5.29, "learning_rate": 9.413839435673497e-06, "loss": 0.0713, "step": 63030 }, { "epoch": 5.29, "learning_rate": 9.412159892509239e-06, "loss": 0.0865, "step": 63040 }, { "epoch": 5.29, "learning_rate": 9.41048034934498e-06, "loss": 0.0082, "step": 63050 }, { "epoch": 5.3, "learning_rate": 9.40880080618072e-06, "loss": 0.042, "step": 63060 }, { "epoch": 5.3, "learning_rate": 9.40712126301646e-06, "loss": 0.006, "step": 63070 }, { "epoch": 5.3, "learning_rate": 9.4054417198522e-06, "loss": 0.1523, "step": 63080 }, { "epoch": 5.3, "learning_rate": 9.403762176687941e-06, "loss": 0.0232, "step": 63090 }, { "epoch": 5.3, "learning_rate": 9.402082633523683e-06, "loss": 0.0539, "step": 63100 }, { "epoch": 5.3, "learning_rate": 9.400403090359424e-06, "loss": 0.1274, "step": 63110 }, { "epoch": 5.3, "learning_rate": 9.398723547195163e-06, "loss": 0.0375, "step": 63120 }, { "epoch": 5.3, "learning_rate": 9.397044004030905e-06, "loss": 0.0367, "step": 63130 }, { "epoch": 5.3, "learning_rate": 9.395364460866645e-06, "loss": 0.1235, "step": 63140 }, { "epoch": 5.3, "learning_rate": 9.393684917702386e-06, "loss": 0.0124, "step": 63150 }, { "epoch": 5.3, "learning_rate": 9.392005374538126e-06, "loss": 0.0455, "step": 63160 }, { "epoch": 5.3, "learning_rate": 9.390325831373867e-06, "loss": 0.0991, "step": 63170 }, { "epoch": 5.31, "learning_rate": 9.388646288209607e-06, "loss": 0.0621, "step": 63180 }, { "epoch": 5.31, "learning_rate": 9.386966745045348e-06, "loss": 0.0075, "step": 63190 }, { "epoch": 5.31, "learning_rate": 9.38528720188109e-06, "loss": 0.0675, "step": 63200 }, { "epoch": 5.31, "learning_rate": 9.38360765871683e-06, "loss": 0.1405, "step": 63210 }, { "epoch": 5.31, "learning_rate": 9.38192811555257e-06, "loss": 0.0887, "step": 63220 }, { "epoch": 5.31, "learning_rate": 9.380248572388311e-06, "loss": 0.0261, "step": 63230 }, { "epoch": 5.31, "learning_rate": 9.378569029224052e-06, "loss": 0.0382, "step": 63240 }, { "epoch": 5.31, "learning_rate": 9.376889486059792e-06, "loss": 0.0782, "step": 63250 }, { "epoch": 5.31, "learning_rate": 9.375209942895533e-06, "loss": 0.0379, "step": 63260 }, { "epoch": 5.31, "learning_rate": 9.373530399731275e-06, "loss": 0.0425, "step": 63270 }, { "epoch": 5.31, "learning_rate": 9.371850856567014e-06, "loss": 0.0322, "step": 63280 }, { "epoch": 5.31, "learning_rate": 9.370171313402756e-06, "loss": 0.0385, "step": 63290 }, { "epoch": 5.32, "learning_rate": 9.368491770238496e-06, "loss": 0.0579, "step": 63300 }, { "epoch": 5.32, "learning_rate": 9.366812227074237e-06, "loss": 0.0976, "step": 63310 }, { "epoch": 5.32, "learning_rate": 9.365132683909977e-06, "loss": 0.0452, "step": 63320 }, { "epoch": 5.32, "learning_rate": 9.363453140745718e-06, "loss": 0.0888, "step": 63330 }, { "epoch": 5.32, "learning_rate": 9.361773597581458e-06, "loss": 0.0433, "step": 63340 }, { "epoch": 5.32, "learning_rate": 9.360094054417199e-06, "loss": 0.0642, "step": 63350 }, { "epoch": 5.32, "learning_rate": 9.35841451125294e-06, "loss": 0.0228, "step": 63360 }, { "epoch": 5.32, "learning_rate": 9.356734968088681e-06, "loss": 0.0676, "step": 63370 }, { "epoch": 5.32, "learning_rate": 9.355055424924422e-06, "loss": 0.0334, "step": 63380 }, { "epoch": 5.32, "learning_rate": 9.353375881760162e-06, "loss": 0.0094, "step": 63390 }, { "epoch": 5.32, "learning_rate": 9.351696338595903e-06, "loss": 0.0309, "step": 63400 }, { "epoch": 5.32, "learning_rate": 9.350016795431643e-06, "loss": 0.0291, "step": 63410 }, { "epoch": 5.33, "learning_rate": 9.348337252267384e-06, "loss": 0.017, "step": 63420 }, { "epoch": 5.33, "learning_rate": 9.346657709103126e-06, "loss": 0.0824, "step": 63430 }, { "epoch": 5.33, "learning_rate": 9.344978165938865e-06, "loss": 0.0044, "step": 63440 }, { "epoch": 5.33, "learning_rate": 9.343298622774605e-06, "loss": 0.0689, "step": 63450 }, { "epoch": 5.33, "learning_rate": 9.341619079610347e-06, "loss": 0.0484, "step": 63460 }, { "epoch": 5.33, "learning_rate": 9.339939536446088e-06, "loss": 0.0017, "step": 63470 }, { "epoch": 5.33, "learning_rate": 9.338259993281828e-06, "loss": 0.0242, "step": 63480 }, { "epoch": 5.33, "learning_rate": 9.336580450117569e-06, "loss": 0.0112, "step": 63490 }, { "epoch": 5.33, "learning_rate": 9.33490090695331e-06, "loss": 0.0796, "step": 63500 }, { "epoch": 5.33, "learning_rate": 9.33322136378905e-06, "loss": 0.0032, "step": 63510 }, { "epoch": 5.33, "learning_rate": 9.33154182062479e-06, "loss": 0.0002, "step": 63520 }, { "epoch": 5.34, "learning_rate": 9.329862277460532e-06, "loss": 0.0395, "step": 63530 }, { "epoch": 5.34, "learning_rate": 9.328182734296271e-06, "loss": 0.1735, "step": 63540 }, { "epoch": 5.34, "learning_rate": 9.326503191132013e-06, "loss": 0.0856, "step": 63550 }, { "epoch": 5.34, "learning_rate": 9.324823647967754e-06, "loss": 0.0753, "step": 63560 }, { "epoch": 5.34, "learning_rate": 9.323144104803494e-06, "loss": 0.0247, "step": 63570 }, { "epoch": 5.34, "learning_rate": 9.321464561639235e-06, "loss": 0.0739, "step": 63580 }, { "epoch": 5.34, "learning_rate": 9.319785018474977e-06, "loss": 0.0297, "step": 63590 }, { "epoch": 5.34, "learning_rate": 9.318105475310716e-06, "loss": 0.0348, "step": 63600 }, { "epoch": 5.34, "learning_rate": 9.316425932146456e-06, "loss": 0.0213, "step": 63610 }, { "epoch": 5.34, "learning_rate": 9.314746388982198e-06, "loss": 0.039, "step": 63620 }, { "epoch": 5.34, "learning_rate": 9.313066845817939e-06, "loss": 0.0694, "step": 63630 }, { "epoch": 5.34, "learning_rate": 9.31138730265368e-06, "loss": 0.0381, "step": 63640 }, { "epoch": 5.35, "learning_rate": 9.30970775948942e-06, "loss": 0.0781, "step": 63650 }, { "epoch": 5.35, "learning_rate": 9.30802821632516e-06, "loss": 0.0717, "step": 63660 }, { "epoch": 5.35, "learning_rate": 9.3063486731609e-06, "loss": 0.0018, "step": 63670 }, { "epoch": 5.35, "learning_rate": 9.304669129996641e-06, "loss": 0.0006, "step": 63680 }, { "epoch": 5.35, "learning_rate": 9.302989586832383e-06, "loss": 0.0214, "step": 63690 }, { "epoch": 5.35, "learning_rate": 9.301310043668122e-06, "loss": 0.0977, "step": 63700 }, { "epoch": 5.35, "learning_rate": 9.299630500503863e-06, "loss": 0.0824, "step": 63710 }, { "epoch": 5.35, "learning_rate": 9.297950957339605e-06, "loss": 0.0394, "step": 63720 }, { "epoch": 5.35, "learning_rate": 9.296271414175345e-06, "loss": 0.0274, "step": 63730 }, { "epoch": 5.35, "learning_rate": 9.294591871011086e-06, "loss": 0.0222, "step": 63740 }, { "epoch": 5.35, "learning_rate": 9.292912327846826e-06, "loss": 0.0461, "step": 63750 }, { "epoch": 5.35, "learning_rate": 9.291232784682567e-06, "loss": 0.0509, "step": 63760 }, { "epoch": 5.36, "learning_rate": 9.289553241518307e-06, "loss": 0.1289, "step": 63770 }, { "epoch": 5.36, "learning_rate": 9.28787369835405e-06, "loss": 0.0372, "step": 63780 }, { "epoch": 5.36, "learning_rate": 9.28619415518979e-06, "loss": 0.0153, "step": 63790 }, { "epoch": 5.36, "learning_rate": 9.28451461202553e-06, "loss": 0.0226, "step": 63800 }, { "epoch": 5.36, "learning_rate": 9.28283506886127e-06, "loss": 0.002, "step": 63810 }, { "epoch": 5.36, "learning_rate": 9.281155525697011e-06, "loss": 0.0016, "step": 63820 }, { "epoch": 5.36, "learning_rate": 9.279475982532752e-06, "loss": 0.0109, "step": 63830 }, { "epoch": 5.36, "learning_rate": 9.277796439368492e-06, "loss": 0.0006, "step": 63840 }, { "epoch": 5.36, "learning_rate": 9.276116896204234e-06, "loss": 0.0984, "step": 63850 }, { "epoch": 5.36, "learning_rate": 9.274437353039973e-06, "loss": 0.0206, "step": 63860 }, { "epoch": 5.36, "learning_rate": 9.272757809875714e-06, "loss": 0.0016, "step": 63870 }, { "epoch": 5.36, "learning_rate": 9.271078266711456e-06, "loss": 0.0409, "step": 63880 }, { "epoch": 5.37, "learning_rate": 9.269398723547196e-06, "loss": 0.1677, "step": 63890 }, { "epoch": 5.37, "learning_rate": 9.267719180382937e-06, "loss": 0.0655, "step": 63900 }, { "epoch": 5.37, "learning_rate": 9.266039637218677e-06, "loss": 0.0608, "step": 63910 }, { "epoch": 5.37, "learning_rate": 9.264360094054418e-06, "loss": 0.0177, "step": 63920 }, { "epoch": 5.37, "learning_rate": 9.262680550890158e-06, "loss": 0.0392, "step": 63930 }, { "epoch": 5.37, "learning_rate": 9.261001007725899e-06, "loss": 0.0592, "step": 63940 }, { "epoch": 5.37, "learning_rate": 9.25932146456164e-06, "loss": 0.027, "step": 63950 }, { "epoch": 5.37, "learning_rate": 9.257641921397381e-06, "loss": 0.0584, "step": 63960 }, { "epoch": 5.37, "learning_rate": 9.255962378233122e-06, "loss": 0.0595, "step": 63970 }, { "epoch": 5.37, "learning_rate": 9.254282835068862e-06, "loss": 0.0855, "step": 63980 }, { "epoch": 5.37, "learning_rate": 9.252603291904603e-06, "loss": 0.0493, "step": 63990 }, { "epoch": 5.37, "learning_rate": 9.250923748740343e-06, "loss": 0.0444, "step": 64000 }, { "epoch": 5.38, "learning_rate": 9.249244205576084e-06, "loss": 0.0353, "step": 64010 }, { "epoch": 5.38, "learning_rate": 9.247564662411824e-06, "loss": 0.0014, "step": 64020 }, { "epoch": 5.38, "learning_rate": 9.245885119247565e-06, "loss": 0.061, "step": 64030 }, { "epoch": 5.38, "learning_rate": 9.244205576083307e-06, "loss": 0.0343, "step": 64040 }, { "epoch": 5.38, "learning_rate": 9.242526032919047e-06, "loss": 0.0535, "step": 64050 }, { "epoch": 5.38, "learning_rate": 9.240846489754788e-06, "loss": 0.1001, "step": 64060 }, { "epoch": 5.38, "learning_rate": 9.239166946590528e-06, "loss": 0.0862, "step": 64070 }, { "epoch": 5.38, "learning_rate": 9.237487403426269e-06, "loss": 0.1231, "step": 64080 }, { "epoch": 5.38, "learning_rate": 9.23580786026201e-06, "loss": 0.0289, "step": 64090 }, { "epoch": 5.38, "learning_rate": 9.23412831709775e-06, "loss": 0.1183, "step": 64100 }, { "epoch": 5.38, "learning_rate": 9.232448773933492e-06, "loss": 0.1009, "step": 64110 }, { "epoch": 5.38, "learning_rate": 9.230769230769232e-06, "loss": 0.0417, "step": 64120 }, { "epoch": 5.39, "learning_rate": 9.229089687604971e-06, "loss": 0.0566, "step": 64130 }, { "epoch": 5.39, "learning_rate": 9.227410144440713e-06, "loss": 0.0927, "step": 64140 }, { "epoch": 5.39, "learning_rate": 9.225730601276454e-06, "loss": 0.0254, "step": 64150 }, { "epoch": 5.39, "learning_rate": 9.224051058112194e-06, "loss": 0.0213, "step": 64160 }, { "epoch": 5.39, "learning_rate": 9.222371514947935e-06, "loss": 0.0012, "step": 64170 }, { "epoch": 5.39, "learning_rate": 9.220691971783675e-06, "loss": 0.0015, "step": 64180 }, { "epoch": 5.39, "learning_rate": 9.219012428619416e-06, "loss": 0.0822, "step": 64190 }, { "epoch": 5.39, "learning_rate": 9.217332885455156e-06, "loss": 0.0587, "step": 64200 }, { "epoch": 5.39, "learning_rate": 9.215653342290898e-06, "loss": 0.0027, "step": 64210 }, { "epoch": 5.39, "learning_rate": 9.213973799126639e-06, "loss": 0.0974, "step": 64220 }, { "epoch": 5.39, "learning_rate": 9.21229425596238e-06, "loss": 0.0657, "step": 64230 }, { "epoch": 5.39, "learning_rate": 9.21061471279812e-06, "loss": 0.028, "step": 64240 }, { "epoch": 5.4, "learning_rate": 9.20893516963386e-06, "loss": 0.1115, "step": 64250 }, { "epoch": 5.4, "learning_rate": 9.2072556264696e-06, "loss": 0.0923, "step": 64260 }, { "epoch": 5.4, "learning_rate": 9.205576083305343e-06, "loss": 0.0598, "step": 64270 }, { "epoch": 5.4, "learning_rate": 9.203896540141083e-06, "loss": 0.0453, "step": 64280 }, { "epoch": 5.4, "learning_rate": 9.202216996976822e-06, "loss": 0.0243, "step": 64290 }, { "epoch": 5.4, "learning_rate": 9.200537453812564e-06, "loss": 0.0466, "step": 64300 }, { "epoch": 5.4, "learning_rate": 9.198857910648305e-06, "loss": 0.0583, "step": 64310 }, { "epoch": 5.4, "learning_rate": 9.197178367484045e-06, "loss": 0.0143, "step": 64320 }, { "epoch": 5.4, "learning_rate": 9.195498824319786e-06, "loss": 0.0019, "step": 64330 }, { "epoch": 5.4, "learning_rate": 9.193819281155526e-06, "loss": 0.0611, "step": 64340 }, { "epoch": 5.4, "learning_rate": 9.192139737991267e-06, "loss": 0.0445, "step": 64350 }, { "epoch": 5.4, "learning_rate": 9.190460194827007e-06, "loss": 0.0389, "step": 64360 }, { "epoch": 5.41, "learning_rate": 9.18878065166275e-06, "loss": 0.0827, "step": 64370 }, { "epoch": 5.41, "learning_rate": 9.18710110849849e-06, "loss": 0.035, "step": 64380 }, { "epoch": 5.41, "learning_rate": 9.185421565334229e-06, "loss": 0.0079, "step": 64390 }, { "epoch": 5.41, "learning_rate": 9.18374202216997e-06, "loss": 0.0527, "step": 64400 }, { "epoch": 5.41, "learning_rate": 9.182062479005711e-06, "loss": 0.0614, "step": 64410 }, { "epoch": 5.41, "learning_rate": 9.180382935841452e-06, "loss": 0.1242, "step": 64420 }, { "epoch": 5.41, "learning_rate": 9.178703392677192e-06, "loss": 0.0272, "step": 64430 }, { "epoch": 5.41, "learning_rate": 9.177023849512933e-06, "loss": 0.0153, "step": 64440 }, { "epoch": 5.41, "learning_rate": 9.175344306348673e-06, "loss": 0.0012, "step": 64450 }, { "epoch": 5.41, "learning_rate": 9.173664763184415e-06, "loss": 0.0257, "step": 64460 }, { "epoch": 5.41, "learning_rate": 9.171985220020156e-06, "loss": 0.0491, "step": 64470 }, { "epoch": 5.41, "learning_rate": 9.170305676855896e-06, "loss": 0.0044, "step": 64480 }, { "epoch": 5.42, "learning_rate": 9.168626133691637e-06, "loss": 0.0424, "step": 64490 }, { "epoch": 5.42, "learning_rate": 9.166946590527377e-06, "loss": 0.0861, "step": 64500 }, { "epoch": 5.42, "learning_rate": 9.165267047363118e-06, "loss": 0.0508, "step": 64510 }, { "epoch": 5.42, "learning_rate": 9.163587504198858e-06, "loss": 0.0445, "step": 64520 }, { "epoch": 5.42, "learning_rate": 9.1619079610346e-06, "loss": 0.0309, "step": 64530 }, { "epoch": 5.42, "learning_rate": 9.16022841787034e-06, "loss": 0.0764, "step": 64540 }, { "epoch": 5.42, "learning_rate": 9.15854887470608e-06, "loss": 0.1507, "step": 64550 }, { "epoch": 5.42, "learning_rate": 9.156869331541822e-06, "loss": 0.0437, "step": 64560 }, { "epoch": 5.42, "learning_rate": 9.155189788377562e-06, "loss": 0.054, "step": 64570 }, { "epoch": 5.42, "learning_rate": 9.153510245213303e-06, "loss": 0.0284, "step": 64580 }, { "epoch": 5.42, "learning_rate": 9.151830702049043e-06, "loss": 0.0238, "step": 64590 }, { "epoch": 5.42, "learning_rate": 9.150151158884784e-06, "loss": 0.1557, "step": 64600 }, { "epoch": 5.43, "learning_rate": 9.148471615720524e-06, "loss": 0.0804, "step": 64610 }, { "epoch": 5.43, "learning_rate": 9.146792072556265e-06, "loss": 0.0822, "step": 64620 }, { "epoch": 5.43, "learning_rate": 9.145112529392007e-06, "loss": 0.0858, "step": 64630 }, { "epoch": 5.43, "learning_rate": 9.143432986227747e-06, "loss": 0.0132, "step": 64640 }, { "epoch": 5.43, "learning_rate": 9.141753443063488e-06, "loss": 0.043, "step": 64650 }, { "epoch": 5.43, "learning_rate": 9.140073899899228e-06, "loss": 0.0103, "step": 64660 }, { "epoch": 5.43, "learning_rate": 9.138394356734969e-06, "loss": 0.0328, "step": 64670 }, { "epoch": 5.43, "learning_rate": 9.13671481357071e-06, "loss": 0.0326, "step": 64680 }, { "epoch": 5.43, "learning_rate": 9.13503527040645e-06, "loss": 0.0014, "step": 64690 }, { "epoch": 5.43, "learning_rate": 9.133355727242192e-06, "loss": 0.0207, "step": 64700 }, { "epoch": 5.43, "learning_rate": 9.13167618407793e-06, "loss": 0.0754, "step": 64710 }, { "epoch": 5.44, "learning_rate": 9.129996640913673e-06, "loss": 0.0394, "step": 64720 }, { "epoch": 5.44, "learning_rate": 9.128317097749413e-06, "loss": 0.0582, "step": 64730 }, { "epoch": 5.44, "learning_rate": 9.126637554585154e-06, "loss": 0.0412, "step": 64740 }, { "epoch": 5.44, "learning_rate": 9.124958011420894e-06, "loss": 0.0734, "step": 64750 }, { "epoch": 5.44, "learning_rate": 9.123278468256635e-06, "loss": 0.0043, "step": 64760 }, { "epoch": 5.44, "learning_rate": 9.121598925092375e-06, "loss": 0.0069, "step": 64770 }, { "epoch": 5.44, "learning_rate": 9.119919381928116e-06, "loss": 0.0612, "step": 64780 }, { "epoch": 5.44, "learning_rate": 9.118239838763858e-06, "loss": 0.0975, "step": 64790 }, { "epoch": 5.44, "learning_rate": 9.116560295599598e-06, "loss": 0.0845, "step": 64800 }, { "epoch": 5.44, "learning_rate": 9.114880752435339e-06, "loss": 0.0409, "step": 64810 }, { "epoch": 5.44, "learning_rate": 9.11320120927108e-06, "loss": 0.0372, "step": 64820 }, { "epoch": 5.44, "learning_rate": 9.11152166610682e-06, "loss": 0.0348, "step": 64830 }, { "epoch": 5.45, "learning_rate": 9.10984212294256e-06, "loss": 0.048, "step": 64840 }, { "epoch": 5.45, "learning_rate": 9.1081625797783e-06, "loss": 0.079, "step": 64850 }, { "epoch": 5.45, "learning_rate": 9.106483036614043e-06, "loss": 0.0374, "step": 64860 }, { "epoch": 5.45, "learning_rate": 9.104803493449782e-06, "loss": 0.0993, "step": 64870 }, { "epoch": 5.45, "learning_rate": 9.103123950285522e-06, "loss": 0.0015, "step": 64880 }, { "epoch": 5.45, "learning_rate": 9.101444407121264e-06, "loss": 0.0708, "step": 64890 }, { "epoch": 5.45, "learning_rate": 9.099764863957005e-06, "loss": 0.0957, "step": 64900 }, { "epoch": 5.45, "learning_rate": 9.098085320792745e-06, "loss": 0.0775, "step": 64910 }, { "epoch": 5.45, "learning_rate": 9.096405777628486e-06, "loss": 0.0322, "step": 64920 }, { "epoch": 5.45, "learning_rate": 9.094726234464226e-06, "loss": 0.0434, "step": 64930 }, { "epoch": 5.45, "learning_rate": 9.093046691299967e-06, "loss": 0.0815, "step": 64940 }, { "epoch": 5.45, "learning_rate": 9.091367148135709e-06, "loss": 0.0728, "step": 64950 }, { "epoch": 5.46, "learning_rate": 9.08968760497145e-06, "loss": 0.0684, "step": 64960 }, { "epoch": 5.46, "learning_rate": 9.088008061807188e-06, "loss": 0.0689, "step": 64970 }, { "epoch": 5.46, "learning_rate": 9.08632851864293e-06, "loss": 0.0009, "step": 64980 }, { "epoch": 5.46, "learning_rate": 9.08464897547867e-06, "loss": 0.0035, "step": 64990 }, { "epoch": 5.46, "learning_rate": 9.082969432314411e-06, "loss": 0.1503, "step": 65000 }, { "epoch": 5.46, "learning_rate": 9.081289889150152e-06, "loss": 0.0018, "step": 65010 }, { "epoch": 5.46, "learning_rate": 9.079610345985894e-06, "loss": 0.0098, "step": 65020 }, { "epoch": 5.46, "learning_rate": 9.077930802821633e-06, "loss": 0.0016, "step": 65030 }, { "epoch": 5.46, "learning_rate": 9.076251259657373e-06, "loss": 0.087, "step": 65040 }, { "epoch": 5.46, "learning_rate": 9.074571716493115e-06, "loss": 0.0617, "step": 65050 }, { "epoch": 5.46, "learning_rate": 9.072892173328856e-06, "loss": 0.0255, "step": 65060 }, { "epoch": 5.46, "learning_rate": 9.071212630164596e-06, "loss": 0.0737, "step": 65070 }, { "epoch": 5.47, "learning_rate": 9.069533087000337e-06, "loss": 0.0314, "step": 65080 }, { "epoch": 5.47, "learning_rate": 9.067853543836077e-06, "loss": 0.0258, "step": 65090 }, { "epoch": 5.47, "learning_rate": 9.066174000671818e-06, "loss": 0.0269, "step": 65100 }, { "epoch": 5.47, "learning_rate": 9.064494457507558e-06, "loss": 0.0229, "step": 65110 }, { "epoch": 5.47, "learning_rate": 9.0628149143433e-06, "loss": 0.0318, "step": 65120 }, { "epoch": 5.47, "learning_rate": 9.061135371179039e-06, "loss": 0.1001, "step": 65130 }, { "epoch": 5.47, "learning_rate": 9.05945582801478e-06, "loss": 0.0945, "step": 65140 }, { "epoch": 5.47, "learning_rate": 9.057776284850522e-06, "loss": 0.0672, "step": 65150 }, { "epoch": 5.47, "learning_rate": 9.056096741686262e-06, "loss": 0.0471, "step": 65160 }, { "epoch": 5.47, "learning_rate": 9.054417198522003e-06, "loss": 0.1285, "step": 65170 }, { "epoch": 5.47, "learning_rate": 9.052737655357743e-06, "loss": 0.0191, "step": 65180 }, { "epoch": 5.47, "learning_rate": 9.051058112193484e-06, "loss": 0.0229, "step": 65190 }, { "epoch": 5.48, "learning_rate": 9.049378569029224e-06, "loss": 0.0029, "step": 65200 }, { "epoch": 5.48, "learning_rate": 9.047699025864966e-06, "loss": 0.105, "step": 65210 }, { "epoch": 5.48, "learning_rate": 9.046019482700707e-06, "loss": 0.0742, "step": 65220 }, { "epoch": 5.48, "learning_rate": 9.044339939536447e-06, "loss": 0.0273, "step": 65230 }, { "epoch": 5.48, "learning_rate": 9.042660396372188e-06, "loss": 0.0018, "step": 65240 }, { "epoch": 5.48, "learning_rate": 9.040980853207928e-06, "loss": 0.0241, "step": 65250 }, { "epoch": 5.48, "learning_rate": 9.039301310043669e-06, "loss": 0.1106, "step": 65260 }, { "epoch": 5.48, "learning_rate": 9.037621766879409e-06, "loss": 0.0322, "step": 65270 }, { "epoch": 5.48, "learning_rate": 9.035942223715151e-06, "loss": 0.0656, "step": 65280 }, { "epoch": 5.48, "learning_rate": 9.03426268055089e-06, "loss": 0.0404, "step": 65290 }, { "epoch": 5.48, "learning_rate": 9.03258313738663e-06, "loss": 0.0518, "step": 65300 }, { "epoch": 5.48, "learning_rate": 9.030903594222373e-06, "loss": 0.0737, "step": 65310 }, { "epoch": 5.49, "learning_rate": 9.029224051058113e-06, "loss": 0.0702, "step": 65320 }, { "epoch": 5.49, "learning_rate": 9.027544507893854e-06, "loss": 0.1343, "step": 65330 }, { "epoch": 5.49, "learning_rate": 9.025864964729594e-06, "loss": 0.0483, "step": 65340 }, { "epoch": 5.49, "learning_rate": 9.024185421565335e-06, "loss": 0.0683, "step": 65350 }, { "epoch": 5.49, "learning_rate": 9.022505878401075e-06, "loss": 0.003, "step": 65360 }, { "epoch": 5.49, "learning_rate": 9.020826335236816e-06, "loss": 0.0794, "step": 65370 }, { "epoch": 5.49, "learning_rate": 9.019146792072558e-06, "loss": 0.0318, "step": 65380 }, { "epoch": 5.49, "learning_rate": 9.017467248908298e-06, "loss": 0.0207, "step": 65390 }, { "epoch": 5.49, "learning_rate": 9.015787705744039e-06, "loss": 0.0582, "step": 65400 }, { "epoch": 5.49, "learning_rate": 9.01410816257978e-06, "loss": 0.063, "step": 65410 }, { "epoch": 5.49, "learning_rate": 9.01242861941552e-06, "loss": 0.0532, "step": 65420 }, { "epoch": 5.49, "learning_rate": 9.01074907625126e-06, "loss": 0.0287, "step": 65430 }, { "epoch": 5.5, "learning_rate": 9.009069533087002e-06, "loss": 0.0896, "step": 65440 }, { "epoch": 5.5, "learning_rate": 9.007389989922741e-06, "loss": 0.0012, "step": 65450 }, { "epoch": 5.5, "learning_rate": 9.005710446758482e-06, "loss": 0.0417, "step": 65460 }, { "epoch": 5.5, "learning_rate": 9.004030903594224e-06, "loss": 0.0109, "step": 65470 }, { "epoch": 5.5, "learning_rate": 9.002351360429964e-06, "loss": 0.0848, "step": 65480 }, { "epoch": 5.5, "learning_rate": 9.000671817265705e-06, "loss": 0.0414, "step": 65490 }, { "epoch": 5.5, "learning_rate": 8.998992274101445e-06, "loss": 0.0057, "step": 65500 }, { "epoch": 5.5, "learning_rate": 8.997312730937186e-06, "loss": 0.0245, "step": 65510 }, { "epoch": 5.5, "learning_rate": 8.995633187772926e-06, "loss": 0.0263, "step": 65520 }, { "epoch": 5.5, "learning_rate": 8.993953644608667e-06, "loss": 0.0643, "step": 65530 }, { "epoch": 5.5, "learning_rate": 8.992274101444409e-06, "loss": 0.0362, "step": 65540 }, { "epoch": 5.5, "learning_rate": 8.99059455828015e-06, "loss": 0.0042, "step": 65550 }, { "epoch": 5.51, "learning_rate": 8.988915015115888e-06, "loss": 0.0441, "step": 65560 }, { "epoch": 5.51, "learning_rate": 8.98723547195163e-06, "loss": 0.0017, "step": 65570 }, { "epoch": 5.51, "learning_rate": 8.98555592878737e-06, "loss": 0.0585, "step": 65580 }, { "epoch": 5.51, "learning_rate": 8.983876385623111e-06, "loss": 0.1367, "step": 65590 }, { "epoch": 5.51, "learning_rate": 8.982196842458852e-06, "loss": 0.1003, "step": 65600 }, { "epoch": 5.51, "learning_rate": 8.980517299294592e-06, "loss": 0.0588, "step": 65610 }, { "epoch": 5.51, "learning_rate": 8.978837756130333e-06, "loss": 0.1296, "step": 65620 }, { "epoch": 5.51, "learning_rate": 8.977158212966073e-06, "loss": 0.0472, "step": 65630 }, { "epoch": 5.51, "learning_rate": 8.975478669801815e-06, "loss": 0.0762, "step": 65640 }, { "epoch": 5.51, "learning_rate": 8.973799126637556e-06, "loss": 0.0169, "step": 65650 }, { "epoch": 5.51, "learning_rate": 8.972119583473296e-06, "loss": 0.1507, "step": 65660 }, { "epoch": 5.51, "learning_rate": 8.970440040309037e-06, "loss": 0.004, "step": 65670 }, { "epoch": 5.52, "learning_rate": 8.968760497144777e-06, "loss": 0.0478, "step": 65680 }, { "epoch": 5.52, "learning_rate": 8.967080953980518e-06, "loss": 0.0237, "step": 65690 }, { "epoch": 5.52, "learning_rate": 8.96540141081626e-06, "loss": 0.0006, "step": 65700 }, { "epoch": 5.52, "learning_rate": 8.963721867652e-06, "loss": 0.0276, "step": 65710 }, { "epoch": 5.52, "learning_rate": 8.962042324487739e-06, "loss": 0.0757, "step": 65720 }, { "epoch": 5.52, "learning_rate": 8.960362781323481e-06, "loss": 0.0724, "step": 65730 }, { "epoch": 5.52, "learning_rate": 8.958683238159222e-06, "loss": 0.1037, "step": 65740 }, { "epoch": 5.52, "learning_rate": 8.957003694994962e-06, "loss": 0.0404, "step": 65750 }, { "epoch": 5.52, "learning_rate": 8.955324151830703e-06, "loss": 0.054, "step": 65760 }, { "epoch": 5.52, "learning_rate": 8.953644608666443e-06, "loss": 0.1086, "step": 65770 }, { "epoch": 5.52, "learning_rate": 8.951965065502184e-06, "loss": 0.0604, "step": 65780 }, { "epoch": 5.52, "learning_rate": 8.950285522337924e-06, "loss": 0.0492, "step": 65790 }, { "epoch": 5.53, "learning_rate": 8.948605979173666e-06, "loss": 0.0958, "step": 65800 }, { "epoch": 5.53, "learning_rate": 8.946926436009407e-06, "loss": 0.0401, "step": 65810 }, { "epoch": 5.53, "learning_rate": 8.945246892845146e-06, "loss": 0.0628, "step": 65820 }, { "epoch": 5.53, "learning_rate": 8.943567349680888e-06, "loss": 0.0503, "step": 65830 }, { "epoch": 5.53, "learning_rate": 8.941887806516628e-06, "loss": 0.0542, "step": 65840 }, { "epoch": 5.53, "learning_rate": 8.940208263352369e-06, "loss": 0.026, "step": 65850 }, { "epoch": 5.53, "learning_rate": 8.938528720188109e-06, "loss": 0.0359, "step": 65860 }, { "epoch": 5.53, "learning_rate": 8.936849177023851e-06, "loss": 0.0839, "step": 65870 }, { "epoch": 5.53, "learning_rate": 8.93516963385959e-06, "loss": 0.0024, "step": 65880 }, { "epoch": 5.53, "learning_rate": 8.933490090695332e-06, "loss": 0.0802, "step": 65890 }, { "epoch": 5.53, "learning_rate": 8.931810547531073e-06, "loss": 0.0957, "step": 65900 }, { "epoch": 5.53, "learning_rate": 8.930131004366813e-06, "loss": 0.0568, "step": 65910 }, { "epoch": 5.54, "learning_rate": 8.928451461202554e-06, "loss": 0.0451, "step": 65920 }, { "epoch": 5.54, "learning_rate": 8.926771918038294e-06, "loss": 0.0356, "step": 65930 }, { "epoch": 5.54, "learning_rate": 8.925092374874035e-06, "loss": 0.0848, "step": 65940 }, { "epoch": 5.54, "learning_rate": 8.923412831709775e-06, "loss": 0.0778, "step": 65950 }, { "epoch": 5.54, "learning_rate": 8.921733288545517e-06, "loss": 0.1422, "step": 65960 }, { "epoch": 5.54, "learning_rate": 8.920053745381258e-06, "loss": 0.0491, "step": 65970 }, { "epoch": 5.54, "learning_rate": 8.918374202216997e-06, "loss": 0.0401, "step": 65980 }, { "epoch": 5.54, "learning_rate": 8.916694659052739e-06, "loss": 0.0976, "step": 65990 }, { "epoch": 5.54, "learning_rate": 8.91501511588848e-06, "loss": 0.1038, "step": 66000 }, { "epoch": 5.54, "learning_rate": 8.91333557272422e-06, "loss": 0.0173, "step": 66010 }, { "epoch": 5.54, "learning_rate": 8.91165602955996e-06, "loss": 0.0269, "step": 66020 }, { "epoch": 5.55, "learning_rate": 8.9099764863957e-06, "loss": 0.0113, "step": 66030 }, { "epoch": 5.55, "learning_rate": 8.908296943231441e-06, "loss": 0.0393, "step": 66040 }, { "epoch": 5.55, "learning_rate": 8.906617400067182e-06, "loss": 0.0565, "step": 66050 }, { "epoch": 5.55, "learning_rate": 8.904937856902924e-06, "loss": 0.0494, "step": 66060 }, { "epoch": 5.55, "learning_rate": 8.903258313738664e-06, "loss": 0.0314, "step": 66070 }, { "epoch": 5.55, "learning_rate": 8.901578770574405e-06, "loss": 0.0009, "step": 66080 }, { "epoch": 5.55, "learning_rate": 8.899899227410145e-06, "loss": 0.0039, "step": 66090 }, { "epoch": 5.55, "learning_rate": 8.898219684245886e-06, "loss": 0.0633, "step": 66100 }, { "epoch": 5.55, "learning_rate": 8.896540141081626e-06, "loss": 0.0789, "step": 66110 }, { "epoch": 5.55, "learning_rate": 8.894860597917367e-06, "loss": 0.0691, "step": 66120 }, { "epoch": 5.55, "learning_rate": 8.893181054753109e-06, "loss": 0.0101, "step": 66130 }, { "epoch": 5.55, "learning_rate": 8.891501511588848e-06, "loss": 0.0583, "step": 66140 }, { "epoch": 5.56, "learning_rate": 8.88982196842459e-06, "loss": 0.037, "step": 66150 }, { "epoch": 5.56, "learning_rate": 8.88814242526033e-06, "loss": 0.0019, "step": 66160 }, { "epoch": 5.56, "learning_rate": 8.88646288209607e-06, "loss": 0.0121, "step": 66170 }, { "epoch": 5.56, "learning_rate": 8.884783338931811e-06, "loss": 0.0102, "step": 66180 }, { "epoch": 5.56, "learning_rate": 8.883103795767552e-06, "loss": 0.1007, "step": 66190 }, { "epoch": 5.56, "learning_rate": 8.881424252603292e-06, "loss": 0.0495, "step": 66200 }, { "epoch": 5.56, "learning_rate": 8.879744709439033e-06, "loss": 0.0712, "step": 66210 }, { "epoch": 5.56, "learning_rate": 8.878065166274775e-06, "loss": 0.0656, "step": 66220 }, { "epoch": 5.56, "learning_rate": 8.876385623110515e-06, "loss": 0.0033, "step": 66230 }, { "epoch": 5.56, "learning_rate": 8.874706079946256e-06, "loss": 0.0262, "step": 66240 }, { "epoch": 5.56, "learning_rate": 8.873026536781996e-06, "loss": 0.0286, "step": 66250 }, { "epoch": 5.56, "learning_rate": 8.871346993617737e-06, "loss": 0.0209, "step": 66260 }, { "epoch": 5.57, "learning_rate": 8.869667450453477e-06, "loss": 0.112, "step": 66270 }, { "epoch": 5.57, "learning_rate": 8.867987907289218e-06, "loss": 0.1299, "step": 66280 }, { "epoch": 5.57, "learning_rate": 8.86630836412496e-06, "loss": 0.0687, "step": 66290 }, { "epoch": 5.57, "learning_rate": 8.864628820960699e-06, "loss": 0.0531, "step": 66300 }, { "epoch": 5.57, "learning_rate": 8.862949277796439e-06, "loss": 0.0359, "step": 66310 }, { "epoch": 5.57, "learning_rate": 8.861269734632181e-06, "loss": 0.0705, "step": 66320 }, { "epoch": 5.57, "learning_rate": 8.859590191467922e-06, "loss": 0.0435, "step": 66330 }, { "epoch": 5.57, "learning_rate": 8.857910648303662e-06, "loss": 0.0019, "step": 66340 }, { "epoch": 5.57, "learning_rate": 8.856231105139403e-06, "loss": 0.0029, "step": 66350 }, { "epoch": 5.57, "learning_rate": 8.854551561975143e-06, "loss": 0.1144, "step": 66360 }, { "epoch": 5.57, "learning_rate": 8.852872018810884e-06, "loss": 0.0623, "step": 66370 }, { "epoch": 5.57, "learning_rate": 8.851192475646626e-06, "loss": 0.0741, "step": 66380 }, { "epoch": 5.58, "learning_rate": 8.849512932482366e-06, "loss": 0.0272, "step": 66390 }, { "epoch": 5.58, "learning_rate": 8.847833389318105e-06, "loss": 0.0877, "step": 66400 }, { "epoch": 5.58, "learning_rate": 8.846153846153847e-06, "loss": 0.0022, "step": 66410 }, { "epoch": 5.58, "learning_rate": 8.844474302989588e-06, "loss": 0.0237, "step": 66420 }, { "epoch": 5.58, "learning_rate": 8.842794759825328e-06, "loss": 0.0349, "step": 66430 }, { "epoch": 5.58, "learning_rate": 8.841115216661069e-06, "loss": 0.084, "step": 66440 }, { "epoch": 5.58, "learning_rate": 8.83943567349681e-06, "loss": 0.0648, "step": 66450 }, { "epoch": 5.58, "learning_rate": 8.83775613033255e-06, "loss": 0.0434, "step": 66460 }, { "epoch": 5.58, "learning_rate": 8.83607658716829e-06, "loss": 0.004, "step": 66470 }, { "epoch": 5.58, "learning_rate": 8.834397044004032e-06, "loss": 0.0905, "step": 66480 }, { "epoch": 5.58, "learning_rate": 8.832717500839773e-06, "loss": 0.0149, "step": 66490 }, { "epoch": 5.58, "learning_rate": 8.831037957675513e-06, "loss": 0.0023, "step": 66500 }, { "epoch": 5.59, "learning_rate": 8.829358414511254e-06, "loss": 0.048, "step": 66510 }, { "epoch": 5.59, "learning_rate": 8.827678871346994e-06, "loss": 0.1185, "step": 66520 }, { "epoch": 5.59, "learning_rate": 8.825999328182735e-06, "loss": 0.0304, "step": 66530 }, { "epoch": 5.59, "learning_rate": 8.824319785018475e-06, "loss": 0.0011, "step": 66540 }, { "epoch": 5.59, "learning_rate": 8.822640241854217e-06, "loss": 0.0014, "step": 66550 }, { "epoch": 5.59, "learning_rate": 8.820960698689956e-06, "loss": 0.0618, "step": 66560 }, { "epoch": 5.59, "learning_rate": 8.819281155525698e-06, "loss": 0.1125, "step": 66570 }, { "epoch": 5.59, "learning_rate": 8.817601612361439e-06, "loss": 0.0338, "step": 66580 }, { "epoch": 5.59, "learning_rate": 8.81592206919718e-06, "loss": 0.0712, "step": 66590 }, { "epoch": 5.59, "learning_rate": 8.81424252603292e-06, "loss": 0.0005, "step": 66600 }, { "epoch": 5.59, "learning_rate": 8.81256298286866e-06, "loss": 0.0713, "step": 66610 }, { "epoch": 5.59, "learning_rate": 8.8108834397044e-06, "loss": 0.039, "step": 66620 }, { "epoch": 5.6, "learning_rate": 8.809203896540141e-06, "loss": 0.0397, "step": 66630 }, { "epoch": 5.6, "learning_rate": 8.807524353375883e-06, "loss": 0.0318, "step": 66640 }, { "epoch": 5.6, "learning_rate": 8.805844810211624e-06, "loss": 0.0572, "step": 66650 }, { "epoch": 5.6, "learning_rate": 8.804165267047364e-06, "loss": 0.1213, "step": 66660 }, { "epoch": 5.6, "learning_rate": 8.802485723883105e-06, "loss": 0.0362, "step": 66670 }, { "epoch": 5.6, "learning_rate": 8.800806180718845e-06, "loss": 0.0932, "step": 66680 }, { "epoch": 5.6, "learning_rate": 8.799126637554586e-06, "loss": 0.0627, "step": 66690 }, { "epoch": 5.6, "learning_rate": 8.797447094390326e-06, "loss": 0.0429, "step": 66700 }, { "epoch": 5.6, "learning_rate": 8.795767551226068e-06, "loss": 0.0187, "step": 66710 }, { "epoch": 5.6, "learning_rate": 8.794088008061807e-06, "loss": 0.0231, "step": 66720 }, { "epoch": 5.6, "learning_rate": 8.792408464897548e-06, "loss": 0.0597, "step": 66730 }, { "epoch": 5.6, "learning_rate": 8.79072892173329e-06, "loss": 0.0568, "step": 66740 }, { "epoch": 5.61, "learning_rate": 8.78904937856903e-06, "loss": 0.136, "step": 66750 }, { "epoch": 5.61, "learning_rate": 8.78736983540477e-06, "loss": 0.0364, "step": 66760 }, { "epoch": 5.61, "learning_rate": 8.785690292240511e-06, "loss": 0.0811, "step": 66770 }, { "epoch": 5.61, "learning_rate": 8.784010749076252e-06, "loss": 0.0468, "step": 66780 }, { "epoch": 5.61, "learning_rate": 8.782331205911992e-06, "loss": 0.0896, "step": 66790 }, { "epoch": 5.61, "learning_rate": 8.780651662747733e-06, "loss": 0.0682, "step": 66800 }, { "epoch": 5.61, "learning_rate": 8.778972119583475e-06, "loss": 0.0271, "step": 66810 }, { "epoch": 5.61, "learning_rate": 8.777292576419215e-06, "loss": 0.0445, "step": 66820 }, { "epoch": 5.61, "learning_rate": 8.775613033254956e-06, "loss": 0.043, "step": 66830 }, { "epoch": 5.61, "learning_rate": 8.773933490090696e-06, "loss": 0.0689, "step": 66840 }, { "epoch": 5.61, "learning_rate": 8.772253946926437e-06, "loss": 0.0301, "step": 66850 }, { "epoch": 5.61, "learning_rate": 8.770574403762177e-06, "loss": 0.0465, "step": 66860 }, { "epoch": 5.62, "learning_rate": 8.76889486059792e-06, "loss": 0.0023, "step": 66870 }, { "epoch": 5.62, "learning_rate": 8.767215317433658e-06, "loss": 0.0034, "step": 66880 }, { "epoch": 5.62, "learning_rate": 8.765535774269399e-06, "loss": 0.0996, "step": 66890 }, { "epoch": 5.62, "learning_rate": 8.76385623110514e-06, "loss": 0.0278, "step": 66900 }, { "epoch": 5.62, "learning_rate": 8.762176687940881e-06, "loss": 0.0226, "step": 66910 }, { "epoch": 5.62, "learning_rate": 8.760497144776622e-06, "loss": 0.1497, "step": 66920 }, { "epoch": 5.62, "learning_rate": 8.758817601612362e-06, "loss": 0.0559, "step": 66930 }, { "epoch": 5.62, "learning_rate": 8.757138058448103e-06, "loss": 0.062, "step": 66940 }, { "epoch": 5.62, "learning_rate": 8.755458515283843e-06, "loss": 0.0184, "step": 66950 }, { "epoch": 5.62, "learning_rate": 8.753778972119584e-06, "loss": 0.0329, "step": 66960 }, { "epoch": 5.62, "learning_rate": 8.752099428955326e-06, "loss": 0.0113, "step": 66970 }, { "epoch": 5.62, "learning_rate": 8.750419885791066e-06, "loss": 0.0522, "step": 66980 }, { "epoch": 5.63, "learning_rate": 8.748740342626805e-06, "loss": 0.0983, "step": 66990 }, { "epoch": 5.63, "learning_rate": 8.747060799462547e-06, "loss": 0.0307, "step": 67000 }, { "epoch": 5.63, "learning_rate": 8.745381256298288e-06, "loss": 0.0313, "step": 67010 }, { "epoch": 5.63, "learning_rate": 8.743701713134028e-06, "loss": 0.0238, "step": 67020 }, { "epoch": 5.63, "learning_rate": 8.742022169969769e-06, "loss": 0.0391, "step": 67030 }, { "epoch": 5.63, "learning_rate": 8.740342626805509e-06, "loss": 0.0595, "step": 67040 }, { "epoch": 5.63, "learning_rate": 8.73866308364125e-06, "loss": 0.0475, "step": 67050 }, { "epoch": 5.63, "learning_rate": 8.736983540476992e-06, "loss": 0.0431, "step": 67060 }, { "epoch": 5.63, "learning_rate": 8.735303997312732e-06, "loss": 0.1291, "step": 67070 }, { "epoch": 5.63, "learning_rate": 8.733624454148473e-06, "loss": 0.1662, "step": 67080 }, { "epoch": 5.63, "learning_rate": 8.731944910984213e-06, "loss": 0.0887, "step": 67090 }, { "epoch": 5.63, "learning_rate": 8.730265367819954e-06, "loss": 0.0746, "step": 67100 }, { "epoch": 5.64, "learning_rate": 8.728585824655694e-06, "loss": 0.0488, "step": 67110 }, { "epoch": 5.64, "learning_rate": 8.726906281491435e-06, "loss": 0.0675, "step": 67120 }, { "epoch": 5.64, "learning_rate": 8.725226738327177e-06, "loss": 0.0297, "step": 67130 }, { "epoch": 5.64, "learning_rate": 8.723547195162917e-06, "loss": 0.0766, "step": 67140 }, { "epoch": 5.64, "learning_rate": 8.721867651998656e-06, "loss": 0.1092, "step": 67150 }, { "epoch": 5.64, "learning_rate": 8.720188108834398e-06, "loss": 0.0838, "step": 67160 }, { "epoch": 5.64, "learning_rate": 8.718508565670139e-06, "loss": 0.0023, "step": 67170 }, { "epoch": 5.64, "learning_rate": 8.71682902250588e-06, "loss": 0.1171, "step": 67180 }, { "epoch": 5.64, "learning_rate": 8.71514947934162e-06, "loss": 0.0619, "step": 67190 }, { "epoch": 5.64, "learning_rate": 8.71346993617736e-06, "loss": 0.0303, "step": 67200 }, { "epoch": 5.64, "learning_rate": 8.7117903930131e-06, "loss": 0.0036, "step": 67210 }, { "epoch": 5.64, "learning_rate": 8.710110849848841e-06, "loss": 0.0234, "step": 67220 }, { "epoch": 5.65, "learning_rate": 8.708431306684583e-06, "loss": 0.001, "step": 67230 }, { "epoch": 5.65, "learning_rate": 8.706751763520324e-06, "loss": 0.0425, "step": 67240 }, { "epoch": 5.65, "learning_rate": 8.705072220356064e-06, "loss": 0.0414, "step": 67250 }, { "epoch": 5.65, "learning_rate": 8.703392677191805e-06, "loss": 0.1071, "step": 67260 }, { "epoch": 5.65, "learning_rate": 8.701713134027545e-06, "loss": 0.0511, "step": 67270 }, { "epoch": 5.65, "learning_rate": 8.700033590863286e-06, "loss": 0.0286, "step": 67280 }, { "epoch": 5.65, "learning_rate": 8.698354047699026e-06, "loss": 0.0769, "step": 67290 }, { "epoch": 5.65, "learning_rate": 8.696674504534768e-06, "loss": 0.0014, "step": 67300 }, { "epoch": 5.65, "learning_rate": 8.694994961370507e-06, "loss": 0.0359, "step": 67310 }, { "epoch": 5.65, "learning_rate": 8.69331541820625e-06, "loss": 0.0354, "step": 67320 }, { "epoch": 5.65, "learning_rate": 8.69163587504199e-06, "loss": 0.0556, "step": 67330 }, { "epoch": 5.66, "learning_rate": 8.68995633187773e-06, "loss": 0.0548, "step": 67340 }, { "epoch": 5.66, "learning_rate": 8.68827678871347e-06, "loss": 0.0424, "step": 67350 }, { "epoch": 5.66, "learning_rate": 8.686597245549211e-06, "loss": 0.0265, "step": 67360 }, { "epoch": 5.66, "learning_rate": 8.684917702384952e-06, "loss": 0.0396, "step": 67370 }, { "epoch": 5.66, "learning_rate": 8.683238159220692e-06, "loss": 0.0964, "step": 67380 }, { "epoch": 5.66, "learning_rate": 8.681558616056434e-06, "loss": 0.0215, "step": 67390 }, { "epoch": 5.66, "learning_rate": 8.679879072892175e-06, "loss": 0.0357, "step": 67400 }, { "epoch": 5.66, "learning_rate": 8.678199529727914e-06, "loss": 0.0433, "step": 67410 }, { "epoch": 5.66, "learning_rate": 8.676519986563656e-06, "loss": 0.0549, "step": 67420 }, { "epoch": 5.66, "learning_rate": 8.674840443399396e-06, "loss": 0.0559, "step": 67430 }, { "epoch": 5.66, "learning_rate": 8.673160900235137e-06, "loss": 0.0519, "step": 67440 }, { "epoch": 5.66, "learning_rate": 8.671481357070877e-06, "loss": 0.025, "step": 67450 }, { "epoch": 5.67, "learning_rate": 8.669801813906618e-06, "loss": 0.0749, "step": 67460 }, { "epoch": 5.67, "learning_rate": 8.668122270742358e-06, "loss": 0.0009, "step": 67470 }, { "epoch": 5.67, "learning_rate": 8.666442727578099e-06, "loss": 0.1049, "step": 67480 }, { "epoch": 5.67, "learning_rate": 8.66476318441384e-06, "loss": 0.0295, "step": 67490 }, { "epoch": 5.67, "learning_rate": 8.663083641249581e-06, "loss": 0.0714, "step": 67500 }, { "epoch": 5.67, "learning_rate": 8.661404098085322e-06, "loss": 0.0009, "step": 67510 }, { "epoch": 5.67, "learning_rate": 8.659724554921062e-06, "loss": 0.0972, "step": 67520 }, { "epoch": 5.67, "learning_rate": 8.658045011756803e-06, "loss": 0.043, "step": 67530 }, { "epoch": 5.67, "learning_rate": 8.656365468592543e-06, "loss": 0.0786, "step": 67540 }, { "epoch": 5.67, "learning_rate": 8.654685925428285e-06, "loss": 0.031, "step": 67550 }, { "epoch": 5.67, "learning_rate": 8.653006382264026e-06, "loss": 0.0782, "step": 67560 }, { "epoch": 5.67, "learning_rate": 8.651326839099765e-06, "loss": 0.022, "step": 67570 }, { "epoch": 5.68, "learning_rate": 8.649647295935507e-06, "loss": 0.0146, "step": 67580 }, { "epoch": 5.68, "learning_rate": 8.647967752771247e-06, "loss": 0.0456, "step": 67590 }, { "epoch": 5.68, "learning_rate": 8.646288209606988e-06, "loss": 0.0982, "step": 67600 }, { "epoch": 5.68, "learning_rate": 8.644608666442728e-06, "loss": 0.0745, "step": 67610 }, { "epoch": 5.68, "learning_rate": 8.642929123278469e-06, "loss": 0.0643, "step": 67620 }, { "epoch": 5.68, "learning_rate": 8.641249580114209e-06, "loss": 0.0251, "step": 67630 }, { "epoch": 5.68, "learning_rate": 8.63957003694995e-06, "loss": 0.002, "step": 67640 }, { "epoch": 5.68, "learning_rate": 8.637890493785692e-06, "loss": 0.0535, "step": 67650 }, { "epoch": 5.68, "learning_rate": 8.636210950621432e-06, "loss": 0.1172, "step": 67660 }, { "epoch": 5.68, "learning_rate": 8.634531407457173e-06, "loss": 0.0681, "step": 67670 }, { "epoch": 5.68, "learning_rate": 8.632851864292913e-06, "loss": 0.1279, "step": 67680 }, { "epoch": 5.68, "learning_rate": 8.631172321128654e-06, "loss": 0.0239, "step": 67690 }, { "epoch": 5.69, "learning_rate": 8.629492777964394e-06, "loss": 0.0153, "step": 67700 }, { "epoch": 5.69, "learning_rate": 8.627813234800135e-06, "loss": 0.0214, "step": 67710 }, { "epoch": 5.69, "learning_rate": 8.626133691635877e-06, "loss": 0.0583, "step": 67720 }, { "epoch": 5.69, "learning_rate": 8.624454148471616e-06, "loss": 0.1265, "step": 67730 }, { "epoch": 5.69, "learning_rate": 8.622774605307358e-06, "loss": 0.0909, "step": 67740 }, { "epoch": 5.69, "learning_rate": 8.621095062143098e-06, "loss": 0.0517, "step": 67750 }, { "epoch": 5.69, "learning_rate": 8.619415518978839e-06, "loss": 0.0241, "step": 67760 }, { "epoch": 5.69, "learning_rate": 8.617735975814579e-06, "loss": 0.0489, "step": 67770 }, { "epoch": 5.69, "learning_rate": 8.61605643265032e-06, "loss": 0.0171, "step": 67780 }, { "epoch": 5.69, "learning_rate": 8.61437688948606e-06, "loss": 0.0427, "step": 67790 }, { "epoch": 5.69, "learning_rate": 8.6126973463218e-06, "loss": 0.0513, "step": 67800 }, { "epoch": 5.69, "learning_rate": 8.611017803157543e-06, "loss": 0.0472, "step": 67810 }, { "epoch": 5.7, "learning_rate": 8.609338259993283e-06, "loss": 0.0286, "step": 67820 }, { "epoch": 5.7, "learning_rate": 8.607658716829024e-06, "loss": 0.0403, "step": 67830 }, { "epoch": 5.7, "learning_rate": 8.605979173664764e-06, "loss": 0.1095, "step": 67840 }, { "epoch": 5.7, "learning_rate": 8.604299630500505e-06, "loss": 0.0674, "step": 67850 }, { "epoch": 5.7, "learning_rate": 8.602620087336245e-06, "loss": 0.033, "step": 67860 }, { "epoch": 5.7, "learning_rate": 8.600940544171986e-06, "loss": 0.0378, "step": 67870 }, { "epoch": 5.7, "learning_rate": 8.599261001007728e-06, "loss": 0.0767, "step": 67880 }, { "epoch": 5.7, "learning_rate": 8.597581457843467e-06, "loss": 0.0843, "step": 67890 }, { "epoch": 5.7, "learning_rate": 8.595901914679207e-06, "loss": 0.0485, "step": 67900 }, { "epoch": 5.7, "learning_rate": 8.59422237151495e-06, "loss": 0.0009, "step": 67910 }, { "epoch": 5.7, "learning_rate": 8.59254282835069e-06, "loss": 0.0332, "step": 67920 }, { "epoch": 5.7, "learning_rate": 8.59086328518643e-06, "loss": 0.0638, "step": 67930 }, { "epoch": 5.71, "learning_rate": 8.58918374202217e-06, "loss": 0.1001, "step": 67940 }, { "epoch": 5.71, "learning_rate": 8.587504198857911e-06, "loss": 0.0527, "step": 67950 }, { "epoch": 5.71, "learning_rate": 8.585824655693652e-06, "loss": 0.0691, "step": 67960 }, { "epoch": 5.71, "learning_rate": 8.584145112529392e-06, "loss": 0.0636, "step": 67970 }, { "epoch": 5.71, "learning_rate": 8.582465569365134e-06, "loss": 0.0194, "step": 67980 }, { "epoch": 5.71, "learning_rate": 8.580786026200873e-06, "loss": 0.0012, "step": 67990 }, { "epoch": 5.71, "learning_rate": 8.579106483036615e-06, "loss": 0.0386, "step": 68000 }, { "epoch": 5.71, "learning_rate": 8.577426939872356e-06, "loss": 0.0009, "step": 68010 }, { "epoch": 5.71, "learning_rate": 8.575747396708096e-06, "loss": 0.0166, "step": 68020 }, { "epoch": 5.71, "learning_rate": 8.574067853543837e-06, "loss": 0.0416, "step": 68030 }, { "epoch": 5.71, "learning_rate": 8.572388310379579e-06, "loss": 0.0277, "step": 68040 }, { "epoch": 5.71, "learning_rate": 8.570708767215318e-06, "loss": 0.1114, "step": 68050 }, { "epoch": 5.72, "learning_rate": 8.569029224051058e-06, "loss": 0.0262, "step": 68060 }, { "epoch": 5.72, "learning_rate": 8.5673496808868e-06, "loss": 0.002, "step": 68070 }, { "epoch": 5.72, "learning_rate": 8.56567013772254e-06, "loss": 0.0427, "step": 68080 }, { "epoch": 5.72, "learning_rate": 8.563990594558281e-06, "loss": 0.0295, "step": 68090 }, { "epoch": 5.72, "learning_rate": 8.562311051394022e-06, "loss": 0.1201, "step": 68100 }, { "epoch": 5.72, "learning_rate": 8.560631508229762e-06, "loss": 0.078, "step": 68110 }, { "epoch": 5.72, "learning_rate": 8.558951965065503e-06, "loss": 0.024, "step": 68120 }, { "epoch": 5.72, "learning_rate": 8.557272421901243e-06, "loss": 0.0216, "step": 68130 }, { "epoch": 5.72, "learning_rate": 8.555592878736985e-06, "loss": 0.059, "step": 68140 }, { "epoch": 5.72, "learning_rate": 8.553913335572724e-06, "loss": 0.0017, "step": 68150 }, { "epoch": 5.72, "learning_rate": 8.552233792408465e-06, "loss": 0.0215, "step": 68160 }, { "epoch": 5.72, "learning_rate": 8.550554249244207e-06, "loss": 0.0693, "step": 68170 }, { "epoch": 5.73, "learning_rate": 8.548874706079947e-06, "loss": 0.046, "step": 68180 }, { "epoch": 5.73, "learning_rate": 8.547195162915688e-06, "loss": 0.0353, "step": 68190 }, { "epoch": 5.73, "learning_rate": 8.545515619751428e-06, "loss": 0.0434, "step": 68200 }, { "epoch": 5.73, "learning_rate": 8.543836076587169e-06, "loss": 0.0428, "step": 68210 }, { "epoch": 5.73, "learning_rate": 8.542156533422909e-06, "loss": 0.0901, "step": 68220 }, { "epoch": 5.73, "learning_rate": 8.540476990258651e-06, "loss": 0.0309, "step": 68230 }, { "epoch": 5.73, "learning_rate": 8.538797447094392e-06, "loss": 0.1673, "step": 68240 }, { "epoch": 5.73, "learning_rate": 8.537117903930132e-06, "loss": 0.0006, "step": 68250 }, { "epoch": 5.73, "learning_rate": 8.535438360765873e-06, "loss": 0.0152, "step": 68260 }, { "epoch": 5.73, "learning_rate": 8.533758817601613e-06, "loss": 0.0496, "step": 68270 }, { "epoch": 5.73, "learning_rate": 8.532079274437354e-06, "loss": 0.002, "step": 68280 }, { "epoch": 5.73, "learning_rate": 8.530399731273094e-06, "loss": 0.0506, "step": 68290 }, { "epoch": 5.74, "learning_rate": 8.528720188108836e-06, "loss": 0.0362, "step": 68300 }, { "epoch": 5.74, "learning_rate": 8.527040644944575e-06, "loss": 0.079, "step": 68310 }, { "epoch": 5.74, "learning_rate": 8.525361101780316e-06, "loss": 0.0431, "step": 68320 }, { "epoch": 5.74, "learning_rate": 8.523681558616058e-06, "loss": 0.0215, "step": 68330 }, { "epoch": 5.74, "learning_rate": 8.522002015451798e-06, "loss": 0.1016, "step": 68340 }, { "epoch": 5.74, "learning_rate": 8.520322472287539e-06, "loss": 0.0216, "step": 68350 }, { "epoch": 5.74, "learning_rate": 8.518642929123279e-06, "loss": 0.0359, "step": 68360 }, { "epoch": 5.74, "learning_rate": 8.51696338595902e-06, "loss": 0.0018, "step": 68370 }, { "epoch": 5.74, "learning_rate": 8.51528384279476e-06, "loss": 0.0047, "step": 68380 }, { "epoch": 5.74, "learning_rate": 8.5136042996305e-06, "loss": 0.0447, "step": 68390 }, { "epoch": 5.74, "learning_rate": 8.511924756466243e-06, "loss": 0.0648, "step": 68400 }, { "epoch": 5.74, "learning_rate": 8.510245213301983e-06, "loss": 0.0394, "step": 68410 }, { "epoch": 5.75, "learning_rate": 8.508565670137724e-06, "loss": 0.0389, "step": 68420 }, { "epoch": 5.75, "learning_rate": 8.506886126973464e-06, "loss": 0.0628, "step": 68430 }, { "epoch": 5.75, "learning_rate": 8.505206583809205e-06, "loss": 0.0747, "step": 68440 }, { "epoch": 5.75, "learning_rate": 8.503527040644945e-06, "loss": 0.1643, "step": 68450 }, { "epoch": 5.75, "learning_rate": 8.501847497480686e-06, "loss": 0.0671, "step": 68460 }, { "epoch": 5.75, "learning_rate": 8.500167954316426e-06, "loss": 0.1148, "step": 68470 }, { "epoch": 5.75, "learning_rate": 8.498488411152167e-06, "loss": 0.0226, "step": 68480 }, { "epoch": 5.75, "learning_rate": 8.496808867987909e-06, "loss": 0.0262, "step": 68490 }, { "epoch": 5.75, "learning_rate": 8.49512932482365e-06, "loss": 0.0269, "step": 68500 }, { "epoch": 5.75, "learning_rate": 8.49344978165939e-06, "loss": 0.1233, "step": 68510 }, { "epoch": 5.75, "learning_rate": 8.49177023849513e-06, "loss": 0.0556, "step": 68520 }, { "epoch": 5.75, "learning_rate": 8.49009069533087e-06, "loss": 0.0513, "step": 68530 }, { "epoch": 5.76, "learning_rate": 8.488411152166611e-06, "loss": 0.0542, "step": 68540 }, { "epoch": 5.76, "learning_rate": 8.486731609002352e-06, "loss": 0.0555, "step": 68550 }, { "epoch": 5.76, "learning_rate": 8.485052065838094e-06, "loss": 0.048, "step": 68560 }, { "epoch": 5.76, "learning_rate": 8.483372522673834e-06, "loss": 0.0678, "step": 68570 }, { "epoch": 5.76, "learning_rate": 8.481692979509573e-06, "loss": 0.1016, "step": 68580 }, { "epoch": 5.76, "learning_rate": 8.480013436345315e-06, "loss": 0.0515, "step": 68590 }, { "epoch": 5.76, "learning_rate": 8.478333893181056e-06, "loss": 0.1011, "step": 68600 }, { "epoch": 5.76, "learning_rate": 8.476654350016796e-06, "loss": 0.0225, "step": 68610 }, { "epoch": 5.76, "learning_rate": 8.474974806852537e-06, "loss": 0.0907, "step": 68620 }, { "epoch": 5.76, "learning_rate": 8.473295263688277e-06, "loss": 0.0781, "step": 68630 }, { "epoch": 5.76, "learning_rate": 8.471615720524018e-06, "loss": 0.1237, "step": 68640 }, { "epoch": 5.77, "learning_rate": 8.469936177359758e-06, "loss": 0.0223, "step": 68650 }, { "epoch": 5.77, "learning_rate": 8.4682566341955e-06, "loss": 0.0795, "step": 68660 }, { "epoch": 5.77, "learning_rate": 8.46657709103124e-06, "loss": 0.003, "step": 68670 }, { "epoch": 5.77, "learning_rate": 8.464897547866981e-06, "loss": 0.0483, "step": 68680 }, { "epoch": 5.77, "learning_rate": 8.463218004702722e-06, "loss": 0.0264, "step": 68690 }, { "epoch": 5.77, "learning_rate": 8.461538461538462e-06, "loss": 0.0534, "step": 68700 }, { "epoch": 5.77, "learning_rate": 8.459858918374203e-06, "loss": 0.0014, "step": 68710 }, { "epoch": 5.77, "learning_rate": 8.458179375209945e-06, "loss": 0.0222, "step": 68720 }, { "epoch": 5.77, "learning_rate": 8.456499832045685e-06, "loss": 0.095, "step": 68730 }, { "epoch": 5.77, "learning_rate": 8.454820288881424e-06, "loss": 0.106, "step": 68740 }, { "epoch": 5.77, "learning_rate": 8.453140745717166e-06, "loss": 0.1482, "step": 68750 }, { "epoch": 5.77, "learning_rate": 8.451461202552907e-06, "loss": 0.0771, "step": 68760 }, { "epoch": 5.78, "learning_rate": 8.449781659388647e-06, "loss": 0.0079, "step": 68770 }, { "epoch": 5.78, "learning_rate": 8.448102116224388e-06, "loss": 0.0231, "step": 68780 }, { "epoch": 5.78, "learning_rate": 8.446422573060128e-06, "loss": 0.0265, "step": 68790 }, { "epoch": 5.78, "learning_rate": 8.444743029895869e-06, "loss": 0.0039, "step": 68800 }, { "epoch": 5.78, "learning_rate": 8.443063486731609e-06, "loss": 0.047, "step": 68810 }, { "epoch": 5.78, "learning_rate": 8.441383943567351e-06, "loss": 0.0085, "step": 68820 }, { "epoch": 5.78, "learning_rate": 8.439704400403092e-06, "loss": 0.1296, "step": 68830 }, { "epoch": 5.78, "learning_rate": 8.43802485723883e-06, "loss": 0.1181, "step": 68840 }, { "epoch": 5.78, "learning_rate": 8.436345314074573e-06, "loss": 0.0249, "step": 68850 }, { "epoch": 5.78, "learning_rate": 8.434665770910313e-06, "loss": 0.0529, "step": 68860 }, { "epoch": 5.78, "learning_rate": 8.432986227746054e-06, "loss": 0.0611, "step": 68870 }, { "epoch": 5.78, "learning_rate": 8.431306684581794e-06, "loss": 0.0546, "step": 68880 }, { "epoch": 5.79, "learning_rate": 8.429627141417535e-06, "loss": 0.1067, "step": 68890 }, { "epoch": 5.79, "learning_rate": 8.427947598253275e-06, "loss": 0.0213, "step": 68900 }, { "epoch": 5.79, "learning_rate": 8.426268055089017e-06, "loss": 0.0219, "step": 68910 }, { "epoch": 5.79, "learning_rate": 8.424588511924758e-06, "loss": 0.0933, "step": 68920 }, { "epoch": 5.79, "learning_rate": 8.422908968760498e-06, "loss": 0.0711, "step": 68930 }, { "epoch": 5.79, "learning_rate": 8.421229425596239e-06, "loss": 0.0523, "step": 68940 }, { "epoch": 5.79, "learning_rate": 8.419549882431979e-06, "loss": 0.0678, "step": 68950 }, { "epoch": 5.79, "learning_rate": 8.41787033926772e-06, "loss": 0.0159, "step": 68960 }, { "epoch": 5.79, "learning_rate": 8.41619079610346e-06, "loss": 0.0431, "step": 68970 }, { "epoch": 5.79, "learning_rate": 8.414511252939202e-06, "loss": 0.0436, "step": 68980 }, { "epoch": 5.79, "learning_rate": 8.412831709774943e-06, "loss": 0.0137, "step": 68990 }, { "epoch": 5.79, "learning_rate": 8.411152166610681e-06, "loss": 0.031, "step": 69000 }, { "epoch": 5.8, "learning_rate": 8.409472623446424e-06, "loss": 0.0536, "step": 69010 }, { "epoch": 5.8, "learning_rate": 8.407793080282164e-06, "loss": 0.0329, "step": 69020 }, { "epoch": 5.8, "learning_rate": 8.406113537117905e-06, "loss": 0.0832, "step": 69030 }, { "epoch": 5.8, "learning_rate": 8.404433993953645e-06, "loss": 0.0433, "step": 69040 }, { "epoch": 5.8, "learning_rate": 8.402754450789386e-06, "loss": 0.0719, "step": 69050 }, { "epoch": 5.8, "learning_rate": 8.401074907625126e-06, "loss": 0.0748, "step": 69060 }, { "epoch": 5.8, "learning_rate": 8.399395364460867e-06, "loss": 0.0327, "step": 69070 }, { "epoch": 5.8, "learning_rate": 8.397715821296609e-06, "loss": 0.0017, "step": 69080 }, { "epoch": 5.8, "learning_rate": 8.39603627813235e-06, "loss": 0.0988, "step": 69090 }, { "epoch": 5.8, "learning_rate": 8.39435673496809e-06, "loss": 0.0917, "step": 69100 }, { "epoch": 5.8, "learning_rate": 8.39267719180383e-06, "loss": 0.0939, "step": 69110 }, { "epoch": 5.8, "learning_rate": 8.39099764863957e-06, "loss": 0.0157, "step": 69120 }, { "epoch": 5.81, "learning_rate": 8.389318105475311e-06, "loss": 0.0503, "step": 69130 }, { "epoch": 5.81, "learning_rate": 8.387638562311052e-06, "loss": 0.1781, "step": 69140 }, { "epoch": 5.81, "learning_rate": 8.385959019146794e-06, "loss": 0.1543, "step": 69150 }, { "epoch": 5.81, "learning_rate": 8.384279475982532e-06, "loss": 0.063, "step": 69160 }, { "epoch": 5.81, "learning_rate": 8.382599932818275e-06, "loss": 0.0906, "step": 69170 }, { "epoch": 5.81, "learning_rate": 8.380920389654015e-06, "loss": 0.0241, "step": 69180 }, { "epoch": 5.81, "learning_rate": 8.379240846489756e-06, "loss": 0.0298, "step": 69190 }, { "epoch": 5.81, "learning_rate": 8.377561303325496e-06, "loss": 0.0157, "step": 69200 }, { "epoch": 5.81, "learning_rate": 8.375881760161237e-06, "loss": 0.0006, "step": 69210 }, { "epoch": 5.81, "learning_rate": 8.374202216996977e-06, "loss": 0.0208, "step": 69220 }, { "epoch": 5.81, "learning_rate": 8.372522673832718e-06, "loss": 0.0101, "step": 69230 }, { "epoch": 5.81, "learning_rate": 8.37084313066846e-06, "loss": 0.0596, "step": 69240 }, { "epoch": 5.82, "learning_rate": 8.3691635875042e-06, "loss": 0.0794, "step": 69250 }, { "epoch": 5.82, "learning_rate": 8.36748404433994e-06, "loss": 0.0007, "step": 69260 }, { "epoch": 5.82, "learning_rate": 8.365804501175681e-06, "loss": 0.0365, "step": 69270 }, { "epoch": 5.82, "learning_rate": 8.364124958011422e-06, "loss": 0.0931, "step": 69280 }, { "epoch": 5.82, "learning_rate": 8.362445414847162e-06, "loss": 0.0292, "step": 69290 }, { "epoch": 5.82, "learning_rate": 8.360765871682903e-06, "loss": 0.0277, "step": 69300 }, { "epoch": 5.82, "learning_rate": 8.359086328518645e-06, "loss": 0.0366, "step": 69310 }, { "epoch": 5.82, "learning_rate": 8.357406785354384e-06, "loss": 0.0769, "step": 69320 }, { "epoch": 5.82, "learning_rate": 8.355727242190124e-06, "loss": 0.0739, "step": 69330 }, { "epoch": 5.82, "learning_rate": 8.354047699025866e-06, "loss": 0.0598, "step": 69340 }, { "epoch": 5.82, "learning_rate": 8.352368155861607e-06, "loss": 0.0359, "step": 69350 }, { "epoch": 5.82, "learning_rate": 8.350688612697347e-06, "loss": 0.0523, "step": 69360 }, { "epoch": 5.83, "learning_rate": 8.349009069533088e-06, "loss": 0.0912, "step": 69370 }, { "epoch": 5.83, "learning_rate": 8.347329526368828e-06, "loss": 0.107, "step": 69380 }, { "epoch": 5.83, "learning_rate": 8.345649983204569e-06, "loss": 0.04, "step": 69390 }, { "epoch": 5.83, "learning_rate": 8.34397044004031e-06, "loss": 0.0266, "step": 69400 }, { "epoch": 5.83, "learning_rate": 8.342290896876051e-06, "loss": 0.0805, "step": 69410 }, { "epoch": 5.83, "learning_rate": 8.34061135371179e-06, "loss": 0.066, "step": 69420 }, { "epoch": 5.83, "learning_rate": 8.338931810547532e-06, "loss": 0.0373, "step": 69430 }, { "epoch": 5.83, "learning_rate": 8.337252267383273e-06, "loss": 0.0237, "step": 69440 }, { "epoch": 5.83, "learning_rate": 8.335572724219013e-06, "loss": 0.0375, "step": 69450 }, { "epoch": 5.83, "learning_rate": 8.333893181054754e-06, "loss": 0.1242, "step": 69460 }, { "epoch": 5.83, "learning_rate": 8.332213637890496e-06, "loss": 0.0033, "step": 69470 }, { "epoch": 5.83, "learning_rate": 8.330534094726235e-06, "loss": 0.0842, "step": 69480 }, { "epoch": 5.84, "learning_rate": 8.328854551561975e-06, "loss": 0.0031, "step": 69490 }, { "epoch": 5.84, "learning_rate": 8.327175008397717e-06, "loss": 0.0009, "step": 69500 }, { "epoch": 5.84, "learning_rate": 8.325495465233458e-06, "loss": 0.0203, "step": 69510 }, { "epoch": 5.84, "learning_rate": 8.323815922069198e-06, "loss": 0.1464, "step": 69520 }, { "epoch": 5.84, "learning_rate": 8.322136378904939e-06, "loss": 0.0457, "step": 69530 }, { "epoch": 5.84, "learning_rate": 8.320456835740679e-06, "loss": 0.071, "step": 69540 }, { "epoch": 5.84, "learning_rate": 8.31877729257642e-06, "loss": 0.0664, "step": 69550 }, { "epoch": 5.84, "learning_rate": 8.31709774941216e-06, "loss": 0.0335, "step": 69560 }, { "epoch": 5.84, "learning_rate": 8.315418206247902e-06, "loss": 0.0964, "step": 69570 }, { "epoch": 5.84, "learning_rate": 8.313738663083641e-06, "loss": 0.0155, "step": 69580 }, { "epoch": 5.84, "learning_rate": 8.312059119919381e-06, "loss": 0.0328, "step": 69590 }, { "epoch": 5.84, "learning_rate": 8.310379576755124e-06, "loss": 0.0034, "step": 69600 }, { "epoch": 5.85, "learning_rate": 8.308700033590864e-06, "loss": 0.0314, "step": 69610 }, { "epoch": 5.85, "learning_rate": 8.307020490426605e-06, "loss": 0.0368, "step": 69620 }, { "epoch": 5.85, "learning_rate": 8.305340947262345e-06, "loss": 0.0601, "step": 69630 }, { "epoch": 5.85, "learning_rate": 8.303661404098086e-06, "loss": 0.061, "step": 69640 }, { "epoch": 5.85, "learning_rate": 8.301981860933826e-06, "loss": 0.0006, "step": 69650 }, { "epoch": 5.85, "learning_rate": 8.300302317769568e-06, "loss": 0.1439, "step": 69660 }, { "epoch": 5.85, "learning_rate": 8.298622774605309e-06, "loss": 0.0532, "step": 69670 }, { "epoch": 5.85, "learning_rate": 8.296943231441049e-06, "loss": 0.0196, "step": 69680 }, { "epoch": 5.85, "learning_rate": 8.29526368827679e-06, "loss": 0.0712, "step": 69690 }, { "epoch": 5.85, "learning_rate": 8.29358414511253e-06, "loss": 0.078, "step": 69700 }, { "epoch": 5.85, "learning_rate": 8.29190460194827e-06, "loss": 0.0394, "step": 69710 }, { "epoch": 5.85, "learning_rate": 8.290225058784011e-06, "loss": 0.0017, "step": 69720 }, { "epoch": 5.86, "learning_rate": 8.288545515619753e-06, "loss": 0.0688, "step": 69730 }, { "epoch": 5.86, "learning_rate": 8.286865972455492e-06, "loss": 0.0599, "step": 69740 }, { "epoch": 5.86, "learning_rate": 8.285186429291232e-06, "loss": 0.0492, "step": 69750 }, { "epoch": 5.86, "learning_rate": 8.283506886126975e-06, "loss": 0.1256, "step": 69760 }, { "epoch": 5.86, "learning_rate": 8.281827342962715e-06, "loss": 0.0008, "step": 69770 }, { "epoch": 5.86, "learning_rate": 8.280147799798456e-06, "loss": 0.0211, "step": 69780 }, { "epoch": 5.86, "learning_rate": 8.278468256634196e-06, "loss": 0.055, "step": 69790 }, { "epoch": 5.86, "learning_rate": 8.276788713469937e-06, "loss": 0.055, "step": 69800 }, { "epoch": 5.86, "learning_rate": 8.275109170305677e-06, "loss": 0.0904, "step": 69810 }, { "epoch": 5.86, "learning_rate": 8.273429627141418e-06, "loss": 0.0496, "step": 69820 }, { "epoch": 5.86, "learning_rate": 8.27175008397716e-06, "loss": 0.123, "step": 69830 }, { "epoch": 5.86, "learning_rate": 8.2700705408129e-06, "loss": 0.0961, "step": 69840 }, { "epoch": 5.87, "learning_rate": 8.26839099764864e-06, "loss": 0.0497, "step": 69850 }, { "epoch": 5.87, "learning_rate": 8.266711454484381e-06, "loss": 0.0203, "step": 69860 }, { "epoch": 5.87, "learning_rate": 8.265031911320122e-06, "loss": 0.0604, "step": 69870 }, { "epoch": 5.87, "learning_rate": 8.263352368155862e-06, "loss": 0.103, "step": 69880 }, { "epoch": 5.87, "learning_rate": 8.261672824991604e-06, "loss": 0.0867, "step": 69890 }, { "epoch": 5.87, "learning_rate": 8.259993281827343e-06, "loss": 0.051, "step": 69900 }, { "epoch": 5.87, "learning_rate": 8.258313738663083e-06, "loss": 0.0605, "step": 69910 }, { "epoch": 5.87, "learning_rate": 8.256634195498826e-06, "loss": 0.0058, "step": 69920 }, { "epoch": 5.87, "learning_rate": 8.254954652334566e-06, "loss": 0.0519, "step": 69930 }, { "epoch": 5.87, "learning_rate": 8.253275109170307e-06, "loss": 0.0249, "step": 69940 }, { "epoch": 5.87, "learning_rate": 8.251595566006047e-06, "loss": 0.0515, "step": 69950 }, { "epoch": 5.88, "learning_rate": 8.249916022841788e-06, "loss": 0.0554, "step": 69960 }, { "epoch": 5.88, "learning_rate": 8.248236479677528e-06, "loss": 0.1011, "step": 69970 }, { "epoch": 5.88, "learning_rate": 8.246556936513269e-06, "loss": 0.0117, "step": 69980 }, { "epoch": 5.88, "learning_rate": 8.24487739334901e-06, "loss": 0.0484, "step": 69990 }, { "epoch": 5.88, "learning_rate": 8.243197850184751e-06, "loss": 0.0265, "step": 70000 }, { "epoch": 5.88, "learning_rate": 8.24151830702049e-06, "loss": 0.0022, "step": 70010 }, { "epoch": 5.88, "learning_rate": 8.239838763856232e-06, "loss": 0.0327, "step": 70020 }, { "epoch": 5.88, "learning_rate": 8.238159220691973e-06, "loss": 0.0264, "step": 70030 }, { "epoch": 5.88, "learning_rate": 8.236479677527713e-06, "loss": 0.0401, "step": 70040 }, { "epoch": 5.88, "learning_rate": 8.234800134363454e-06, "loss": 0.0585, "step": 70050 }, { "epoch": 5.88, "learning_rate": 8.233120591199194e-06, "loss": 0.0195, "step": 70060 }, { "epoch": 5.88, "learning_rate": 8.231441048034935e-06, "loss": 0.045, "step": 70070 }, { "epoch": 5.89, "learning_rate": 8.229761504870675e-06, "loss": 0.0355, "step": 70080 }, { "epoch": 5.89, "learning_rate": 8.228081961706417e-06, "loss": 0.0586, "step": 70090 }, { "epoch": 5.89, "learning_rate": 8.226402418542158e-06, "loss": 0.0754, "step": 70100 }, { "epoch": 5.89, "learning_rate": 8.224722875377898e-06, "loss": 0.0558, "step": 70110 }, { "epoch": 5.89, "learning_rate": 8.223043332213639e-06, "loss": 0.0215, "step": 70120 }, { "epoch": 5.89, "learning_rate": 8.221363789049379e-06, "loss": 0.0603, "step": 70130 }, { "epoch": 5.89, "learning_rate": 8.21968424588512e-06, "loss": 0.0485, "step": 70140 }, { "epoch": 5.89, "learning_rate": 8.218004702720862e-06, "loss": 0.1048, "step": 70150 }, { "epoch": 5.89, "learning_rate": 8.216325159556602e-06, "loss": 0.0681, "step": 70160 }, { "epoch": 5.89, "learning_rate": 8.214645616392341e-06, "loss": 0.0194, "step": 70170 }, { "epoch": 5.89, "learning_rate": 8.212966073228083e-06, "loss": 0.0125, "step": 70180 }, { "epoch": 5.89, "learning_rate": 8.211286530063824e-06, "loss": 0.0017, "step": 70190 }, { "epoch": 5.9, "learning_rate": 8.209606986899564e-06, "loss": 0.0199, "step": 70200 }, { "epoch": 5.9, "learning_rate": 8.207927443735305e-06, "loss": 0.0005, "step": 70210 }, { "epoch": 5.9, "learning_rate": 8.206247900571045e-06, "loss": 0.0542, "step": 70220 }, { "epoch": 5.9, "learning_rate": 8.204568357406786e-06, "loss": 0.054, "step": 70230 }, { "epoch": 5.9, "learning_rate": 8.202888814242526e-06, "loss": 0.0387, "step": 70240 }, { "epoch": 5.9, "learning_rate": 8.201209271078268e-06, "loss": 0.1247, "step": 70250 }, { "epoch": 5.9, "learning_rate": 8.199529727914009e-06, "loss": 0.0622, "step": 70260 }, { "epoch": 5.9, "learning_rate": 8.197850184749747e-06, "loss": 0.0328, "step": 70270 }, { "epoch": 5.9, "learning_rate": 8.19617064158549e-06, "loss": 0.0269, "step": 70280 }, { "epoch": 5.9, "learning_rate": 8.19449109842123e-06, "loss": 0.0007, "step": 70290 }, { "epoch": 5.9, "learning_rate": 8.19281155525697e-06, "loss": 0.0503, "step": 70300 }, { "epoch": 5.9, "learning_rate": 8.191132012092711e-06, "loss": 0.0666, "step": 70310 }, { "epoch": 5.91, "learning_rate": 8.189452468928452e-06, "loss": 0.0132, "step": 70320 }, { "epoch": 5.91, "learning_rate": 8.187772925764192e-06, "loss": 0.0659, "step": 70330 }, { "epoch": 5.91, "learning_rate": 8.186093382599934e-06, "loss": 0.0739, "step": 70340 }, { "epoch": 5.91, "learning_rate": 8.184413839435675e-06, "loss": 0.0343, "step": 70350 }, { "epoch": 5.91, "learning_rate": 8.182734296271415e-06, "loss": 0.0205, "step": 70360 }, { "epoch": 5.91, "learning_rate": 8.181054753107156e-06, "loss": 0.091, "step": 70370 }, { "epoch": 5.91, "learning_rate": 8.179375209942896e-06, "loss": 0.0278, "step": 70380 }, { "epoch": 5.91, "learning_rate": 8.177695666778637e-06, "loss": 0.0754, "step": 70390 }, { "epoch": 5.91, "learning_rate": 8.176016123614377e-06, "loss": 0.0094, "step": 70400 }, { "epoch": 5.91, "learning_rate": 8.17433658045012e-06, "loss": 0.0379, "step": 70410 }, { "epoch": 5.91, "learning_rate": 8.17265703728586e-06, "loss": 0.0655, "step": 70420 }, { "epoch": 5.91, "learning_rate": 8.170977494121598e-06, "loss": 0.0975, "step": 70430 }, { "epoch": 5.92, "learning_rate": 8.16929795095734e-06, "loss": 0.0351, "step": 70440 }, { "epoch": 5.92, "learning_rate": 8.167618407793081e-06, "loss": 0.0732, "step": 70450 }, { "epoch": 5.92, "learning_rate": 8.165938864628822e-06, "loss": 0.2281, "step": 70460 }, { "epoch": 5.92, "learning_rate": 8.164259321464562e-06, "loss": 0.0989, "step": 70470 }, { "epoch": 5.92, "learning_rate": 8.162579778300303e-06, "loss": 0.0407, "step": 70480 }, { "epoch": 5.92, "learning_rate": 8.160900235136043e-06, "loss": 0.0227, "step": 70490 }, { "epoch": 5.92, "learning_rate": 8.159220691971783e-06, "loss": 0.0295, "step": 70500 }, { "epoch": 5.92, "learning_rate": 8.157541148807526e-06, "loss": 0.0679, "step": 70510 }, { "epoch": 5.92, "learning_rate": 8.155861605643266e-06, "loss": 0.1011, "step": 70520 }, { "epoch": 5.92, "learning_rate": 8.154182062479007e-06, "loss": 0.0551, "step": 70530 }, { "epoch": 5.92, "learning_rate": 8.152502519314747e-06, "loss": 0.061, "step": 70540 }, { "epoch": 5.92, "learning_rate": 8.150822976150488e-06, "loss": 0.0917, "step": 70550 }, { "epoch": 5.93, "learning_rate": 8.149143432986228e-06, "loss": 0.0892, "step": 70560 }, { "epoch": 5.93, "learning_rate": 8.147463889821969e-06, "loss": 0.0516, "step": 70570 }, { "epoch": 5.93, "learning_rate": 8.14578434665771e-06, "loss": 0.0992, "step": 70580 }, { "epoch": 5.93, "learning_rate": 8.14410480349345e-06, "loss": 0.0243, "step": 70590 }, { "epoch": 5.93, "learning_rate": 8.142425260329192e-06, "loss": 0.0739, "step": 70600 }, { "epoch": 5.93, "learning_rate": 8.140745717164932e-06, "loss": 0.0362, "step": 70610 }, { "epoch": 5.93, "learning_rate": 8.139066174000673e-06, "loss": 0.0474, "step": 70620 }, { "epoch": 5.93, "learning_rate": 8.137386630836413e-06, "loss": 0.0855, "step": 70630 }, { "epoch": 5.93, "learning_rate": 8.135707087672154e-06, "loss": 0.0013, "step": 70640 }, { "epoch": 5.93, "learning_rate": 8.134027544507894e-06, "loss": 0.0353, "step": 70650 }, { "epoch": 5.93, "learning_rate": 8.132348001343634e-06, "loss": 0.0004, "step": 70660 }, { "epoch": 5.93, "learning_rate": 8.130668458179377e-06, "loss": 0.016, "step": 70670 }, { "epoch": 5.94, "learning_rate": 8.128988915015117e-06, "loss": 0.1092, "step": 70680 }, { "epoch": 5.94, "learning_rate": 8.127309371850858e-06, "loss": 0.0686, "step": 70690 }, { "epoch": 5.94, "learning_rate": 8.125629828686598e-06, "loss": 0.0403, "step": 70700 }, { "epoch": 5.94, "learning_rate": 8.123950285522339e-06, "loss": 0.0451, "step": 70710 }, { "epoch": 5.94, "learning_rate": 8.122270742358079e-06, "loss": 0.0546, "step": 70720 }, { "epoch": 5.94, "learning_rate": 8.12059119919382e-06, "loss": 0.0237, "step": 70730 }, { "epoch": 5.94, "learning_rate": 8.118911656029562e-06, "loss": 0.0703, "step": 70740 }, { "epoch": 5.94, "learning_rate": 8.1172321128653e-06, "loss": 0.0192, "step": 70750 }, { "epoch": 5.94, "learning_rate": 8.115552569701041e-06, "loss": 0.135, "step": 70760 }, { "epoch": 5.94, "learning_rate": 8.113873026536783e-06, "loss": 0.0008, "step": 70770 }, { "epoch": 5.94, "learning_rate": 8.112193483372524e-06, "loss": 0.0645, "step": 70780 }, { "epoch": 5.94, "learning_rate": 8.110513940208264e-06, "loss": 0.0355, "step": 70790 }, { "epoch": 5.95, "learning_rate": 8.108834397044005e-06, "loss": 0.1057, "step": 70800 }, { "epoch": 5.95, "learning_rate": 8.107154853879745e-06, "loss": 0.0009, "step": 70810 }, { "epoch": 5.95, "learning_rate": 8.105475310715486e-06, "loss": 0.0964, "step": 70820 }, { "epoch": 5.95, "learning_rate": 8.103795767551228e-06, "loss": 0.0661, "step": 70830 }, { "epoch": 5.95, "learning_rate": 8.102116224386968e-06, "loss": 0.07, "step": 70840 }, { "epoch": 5.95, "learning_rate": 8.100436681222707e-06, "loss": 0.056, "step": 70850 }, { "epoch": 5.95, "learning_rate": 8.098757138058449e-06, "loss": 0.0573, "step": 70860 }, { "epoch": 5.95, "learning_rate": 8.09707759489419e-06, "loss": 0.0125, "step": 70870 }, { "epoch": 5.95, "learning_rate": 8.09539805172993e-06, "loss": 0.0287, "step": 70880 }, { "epoch": 5.95, "learning_rate": 8.09371850856567e-06, "loss": 0.1359, "step": 70890 }, { "epoch": 5.95, "learning_rate": 8.092038965401413e-06, "loss": 0.0359, "step": 70900 }, { "epoch": 5.95, "learning_rate": 8.090359422237151e-06, "loss": 0.0489, "step": 70910 }, { "epoch": 5.96, "learning_rate": 8.088679879072892e-06, "loss": 0.0815, "step": 70920 }, { "epoch": 5.96, "learning_rate": 8.087000335908634e-06, "loss": 0.162, "step": 70930 }, { "epoch": 5.96, "learning_rate": 8.085320792744375e-06, "loss": 0.0754, "step": 70940 }, { "epoch": 5.96, "learning_rate": 8.083641249580115e-06, "loss": 0.0224, "step": 70950 }, { "epoch": 5.96, "learning_rate": 8.081961706415856e-06, "loss": 0.1185, "step": 70960 }, { "epoch": 5.96, "learning_rate": 8.080282163251596e-06, "loss": 0.0489, "step": 70970 }, { "epoch": 5.96, "learning_rate": 8.078602620087337e-06, "loss": 0.022, "step": 70980 }, { "epoch": 5.96, "learning_rate": 8.076923076923077e-06, "loss": 0.0328, "step": 70990 }, { "epoch": 5.96, "learning_rate": 8.07524353375882e-06, "loss": 0.0573, "step": 71000 }, { "epoch": 5.96, "learning_rate": 8.073563990594558e-06, "loss": 0.0431, "step": 71010 }, { "epoch": 5.96, "learning_rate": 8.0718844474303e-06, "loss": 0.0822, "step": 71020 }, { "epoch": 5.96, "learning_rate": 8.07020490426604e-06, "loss": 0.0707, "step": 71030 }, { "epoch": 5.97, "learning_rate": 8.068525361101781e-06, "loss": 0.066, "step": 71040 }, { "epoch": 5.97, "learning_rate": 8.066845817937522e-06, "loss": 0.0826, "step": 71050 }, { "epoch": 5.97, "learning_rate": 8.065166274773262e-06, "loss": 0.0023, "step": 71060 }, { "epoch": 5.97, "learning_rate": 8.063486731609003e-06, "loss": 0.0745, "step": 71070 }, { "epoch": 5.97, "learning_rate": 8.061807188444743e-06, "loss": 0.0392, "step": 71080 }, { "epoch": 5.97, "learning_rate": 8.060127645280485e-06, "loss": 0.0341, "step": 71090 }, { "epoch": 5.97, "learning_rate": 8.058448102116226e-06, "loss": 0.0006, "step": 71100 }, { "epoch": 5.97, "learning_rate": 8.056768558951966e-06, "loss": 0.0313, "step": 71110 }, { "epoch": 5.97, "learning_rate": 8.055089015787707e-06, "loss": 0.0017, "step": 71120 }, { "epoch": 5.97, "learning_rate": 8.053409472623447e-06, "loss": 0.0378, "step": 71130 }, { "epoch": 5.97, "learning_rate": 8.051729929459188e-06, "loss": 0.0861, "step": 71140 }, { "epoch": 5.97, "learning_rate": 8.050050386294928e-06, "loss": 0.0584, "step": 71150 }, { "epoch": 5.98, "learning_rate": 8.04837084313067e-06, "loss": 0.0875, "step": 71160 }, { "epoch": 5.98, "learning_rate": 8.046691299966409e-06, "loss": 0.0176, "step": 71170 }, { "epoch": 5.98, "learning_rate": 8.04501175680215e-06, "loss": 0.0027, "step": 71180 }, { "epoch": 5.98, "learning_rate": 8.043332213637892e-06, "loss": 0.0396, "step": 71190 }, { "epoch": 5.98, "learning_rate": 8.041652670473632e-06, "loss": 0.0531, "step": 71200 }, { "epoch": 5.98, "learning_rate": 8.039973127309373e-06, "loss": 0.0163, "step": 71210 }, { "epoch": 5.98, "learning_rate": 8.038293584145113e-06, "loss": 0.0364, "step": 71220 }, { "epoch": 5.98, "learning_rate": 8.036614040980854e-06, "loss": 0.0469, "step": 71230 }, { "epoch": 5.98, "learning_rate": 8.034934497816594e-06, "loss": 0.0801, "step": 71240 }, { "epoch": 5.98, "learning_rate": 8.033254954652334e-06, "loss": 0.0587, "step": 71250 }, { "epoch": 5.98, "learning_rate": 8.031575411488077e-06, "loss": 0.048, "step": 71260 }, { "epoch": 5.99, "learning_rate": 8.029895868323817e-06, "loss": 0.0244, "step": 71270 }, { "epoch": 5.99, "learning_rate": 8.028216325159558e-06, "loss": 0.0904, "step": 71280 }, { "epoch": 5.99, "learning_rate": 8.026536781995298e-06, "loss": 0.0537, "step": 71290 }, { "epoch": 5.99, "learning_rate": 8.024857238831039e-06, "loss": 0.0452, "step": 71300 }, { "epoch": 5.99, "learning_rate": 8.023177695666779e-06, "loss": 0.031, "step": 71310 }, { "epoch": 5.99, "learning_rate": 8.021498152502521e-06, "loss": 0.0528, "step": 71320 }, { "epoch": 5.99, "learning_rate": 8.01981860933826e-06, "loss": 0.0216, "step": 71330 }, { "epoch": 5.99, "learning_rate": 8.018139066174e-06, "loss": 0.0097, "step": 71340 }, { "epoch": 5.99, "learning_rate": 8.016459523009743e-06, "loss": 0.0219, "step": 71350 }, { "epoch": 5.99, "learning_rate": 8.014779979845483e-06, "loss": 0.1035, "step": 71360 }, { "epoch": 5.99, "learning_rate": 8.013100436681224e-06, "loss": 0.0527, "step": 71370 }, { "epoch": 5.99, "learning_rate": 8.011420893516964e-06, "loss": 0.0004, "step": 71380 }, { "epoch": 6.0, "learning_rate": 8.009741350352705e-06, "loss": 0.0755, "step": 71390 }, { "epoch": 6.0, "learning_rate": 8.008061807188445e-06, "loss": 0.0296, "step": 71400 }, { "epoch": 6.0, "learning_rate": 8.006382264024185e-06, "loss": 0.0941, "step": 71410 }, { "epoch": 6.0, "learning_rate": 8.004702720859928e-06, "loss": 0.0383, "step": 71420 }, { "epoch": 6.0, "learning_rate": 8.003023177695668e-06, "loss": 0.0235, "step": 71430 }, { "epoch": 6.0, "learning_rate": 8.001343634531407e-06, "loss": 0.0637, "step": 71440 }, { "epoch": 6.0, "eval_FN": 596, "eval_FP": 341, "eval_TN": 21373, "eval_TP": 1506, "eval_accuracy": 0.9606567013772254, "eval_f1": 0.7627247404406178, "eval_loss": 0.21069449186325073, "eval_precision": 0.8153762858689767, "eval_recall": 0.7164605137963844, "eval_runtime": 163.2558, "eval_samples_per_second": 145.881, "eval_steps_per_second": 9.121, "step": 71448 }, { "epoch": 6.0, "learning_rate": 7.999664091367149e-06, "loss": 0.1517, "step": 71450 }, { "epoch": 6.0, "learning_rate": 7.99798454820289e-06, "loss": 0.0027, "step": 71460 }, { "epoch": 6.0, "learning_rate": 7.99630500503863e-06, "loss": 0.094, "step": 71470 }, { "epoch": 6.0, "learning_rate": 7.99462546187437e-06, "loss": 0.0005, "step": 71480 }, { "epoch": 6.0, "learning_rate": 7.992945918710111e-06, "loss": 0.0454, "step": 71490 }, { "epoch": 6.0, "learning_rate": 7.991266375545851e-06, "loss": 0.0367, "step": 71500 }, { "epoch": 6.01, "learning_rate": 7.989586832381594e-06, "loss": 0.0416, "step": 71510 }, { "epoch": 6.01, "learning_rate": 7.987907289217334e-06, "loss": 0.0419, "step": 71520 }, { "epoch": 6.01, "learning_rate": 7.986227746053075e-06, "loss": 0.0303, "step": 71530 }, { "epoch": 6.01, "learning_rate": 7.984548202888815e-06, "loss": 0.0428, "step": 71540 }, { "epoch": 6.01, "learning_rate": 7.982868659724556e-06, "loss": 0.0036, "step": 71550 }, { "epoch": 6.01, "learning_rate": 7.981189116560296e-06, "loss": 0.0651, "step": 71560 }, { "epoch": 6.01, "learning_rate": 7.979509573396037e-06, "loss": 0.0439, "step": 71570 }, { "epoch": 6.01, "learning_rate": 7.977830030231779e-06, "loss": 0.0257, "step": 71580 }, { "epoch": 6.01, "learning_rate": 7.976150487067519e-06, "loss": 0.0087, "step": 71590 }, { "epoch": 6.01, "learning_rate": 7.974470943903258e-06, "loss": 0.0282, "step": 71600 }, { "epoch": 6.01, "learning_rate": 7.972791400739e-06, "loss": 0.0538, "step": 71610 }, { "epoch": 6.01, "learning_rate": 7.97111185757474e-06, "loss": 0.0784, "step": 71620 }, { "epoch": 6.02, "learning_rate": 7.969432314410481e-06, "loss": 0.027, "step": 71630 }, { "epoch": 6.02, "learning_rate": 7.967752771246222e-06, "loss": 0.0259, "step": 71640 }, { "epoch": 6.02, "learning_rate": 7.966073228081962e-06, "loss": 0.0163, "step": 71650 }, { "epoch": 6.02, "learning_rate": 7.964393684917702e-06, "loss": 0.0225, "step": 71660 }, { "epoch": 6.02, "learning_rate": 7.962714141753443e-06, "loss": 0.001, "step": 71670 }, { "epoch": 6.02, "learning_rate": 7.961034598589185e-06, "loss": 0.0622, "step": 71680 }, { "epoch": 6.02, "learning_rate": 7.959355055424926e-06, "loss": 0.001, "step": 71690 }, { "epoch": 6.02, "learning_rate": 7.957675512260666e-06, "loss": 0.0177, "step": 71700 }, { "epoch": 6.02, "learning_rate": 7.955995969096407e-06, "loss": 0.0207, "step": 71710 }, { "epoch": 6.02, "learning_rate": 7.954316425932147e-06, "loss": 0.0332, "step": 71720 }, { "epoch": 6.02, "learning_rate": 7.952636882767888e-06, "loss": 0.0074, "step": 71730 }, { "epoch": 6.02, "learning_rate": 7.950957339603628e-06, "loss": 0.0785, "step": 71740 }, { "epoch": 6.03, "learning_rate": 7.949277796439368e-06, "loss": 0.001, "step": 71750 }, { "epoch": 6.03, "learning_rate": 7.947598253275109e-06, "loss": 0.056, "step": 71760 }, { "epoch": 6.03, "learning_rate": 7.945918710110851e-06, "loss": 0.0326, "step": 71770 }, { "epoch": 6.03, "learning_rate": 7.944239166946592e-06, "loss": 0.0858, "step": 71780 }, { "epoch": 6.03, "learning_rate": 7.942559623782332e-06, "loss": 0.0628, "step": 71790 }, { "epoch": 6.03, "learning_rate": 7.940880080618073e-06, "loss": 0.0565, "step": 71800 }, { "epoch": 6.03, "learning_rate": 7.939200537453813e-06, "loss": 0.0385, "step": 71810 }, { "epoch": 6.03, "learning_rate": 7.937520994289554e-06, "loss": 0.0013, "step": 71820 }, { "epoch": 6.03, "learning_rate": 7.935841451125294e-06, "loss": 0.0014, "step": 71830 }, { "epoch": 6.03, "learning_rate": 7.934161907961036e-06, "loss": 0.0041, "step": 71840 }, { "epoch": 6.03, "learning_rate": 7.932482364796777e-06, "loss": 0.0936, "step": 71850 }, { "epoch": 6.03, "learning_rate": 7.930802821632515e-06, "loss": 0.0146, "step": 71860 }, { "epoch": 6.04, "learning_rate": 7.929123278468258e-06, "loss": 0.0356, "step": 71870 }, { "epoch": 6.04, "learning_rate": 7.927443735303998e-06, "loss": 0.0207, "step": 71880 }, { "epoch": 6.04, "learning_rate": 7.925764192139739e-06, "loss": 0.1131, "step": 71890 }, { "epoch": 6.04, "learning_rate": 7.924084648975479e-06, "loss": 0.0406, "step": 71900 }, { "epoch": 6.04, "learning_rate": 7.92240510581122e-06, "loss": 0.0431, "step": 71910 }, { "epoch": 6.04, "learning_rate": 7.92072556264696e-06, "loss": 0.0422, "step": 71920 }, { "epoch": 6.04, "learning_rate": 7.9190460194827e-06, "loss": 0.0504, "step": 71930 }, { "epoch": 6.04, "learning_rate": 7.917366476318443e-06, "loss": 0.0237, "step": 71940 }, { "epoch": 6.04, "learning_rate": 7.915686933154183e-06, "loss": 0.0567, "step": 71950 }, { "epoch": 6.04, "learning_rate": 7.914007389989924e-06, "loss": 0.0132, "step": 71960 }, { "epoch": 6.04, "learning_rate": 7.912327846825664e-06, "loss": 0.0297, "step": 71970 }, { "epoch": 6.04, "learning_rate": 7.910648303661405e-06, "loss": 0.0412, "step": 71980 }, { "epoch": 6.05, "learning_rate": 7.908968760497145e-06, "loss": 0.0008, "step": 71990 }, { "epoch": 6.05, "learning_rate": 7.907289217332887e-06, "loss": 0.0184, "step": 72000 }, { "epoch": 6.05, "learning_rate": 7.905609674168628e-06, "loss": 0.0481, "step": 72010 }, { "epoch": 6.05, "learning_rate": 7.903930131004366e-06, "loss": 0.0366, "step": 72020 }, { "epoch": 6.05, "learning_rate": 7.902250587840109e-06, "loss": 0.0002, "step": 72030 }, { "epoch": 6.05, "learning_rate": 7.900571044675849e-06, "loss": 0.0128, "step": 72040 }, { "epoch": 6.05, "learning_rate": 7.89889150151159e-06, "loss": 0.0452, "step": 72050 }, { "epoch": 6.05, "learning_rate": 7.89721195834733e-06, "loss": 0.0137, "step": 72060 }, { "epoch": 6.05, "learning_rate": 7.89553241518307e-06, "loss": 0.0055, "step": 72070 }, { "epoch": 6.05, "learning_rate": 7.893852872018811e-06, "loss": 0.0004, "step": 72080 }, { "epoch": 6.05, "learning_rate": 7.892173328854551e-06, "loss": 0.0013, "step": 72090 }, { "epoch": 6.05, "learning_rate": 7.890493785690294e-06, "loss": 0.0004, "step": 72100 }, { "epoch": 6.06, "learning_rate": 7.888814242526034e-06, "loss": 0.0387, "step": 72110 }, { "epoch": 6.06, "learning_rate": 7.887134699361775e-06, "loss": 0.0001, "step": 72120 }, { "epoch": 6.06, "learning_rate": 7.885455156197515e-06, "loss": 0.054, "step": 72130 }, { "epoch": 6.06, "learning_rate": 7.883775613033256e-06, "loss": 0.002, "step": 72140 }, { "epoch": 6.06, "learning_rate": 7.882096069868996e-06, "loss": 0.0555, "step": 72150 }, { "epoch": 6.06, "learning_rate": 7.880416526704736e-06, "loss": 0.0165, "step": 72160 }, { "epoch": 6.06, "learning_rate": 7.878736983540479e-06, "loss": 0.0177, "step": 72170 }, { "epoch": 6.06, "learning_rate": 7.877057440376217e-06, "loss": 0.0678, "step": 72180 }, { "epoch": 6.06, "learning_rate": 7.87537789721196e-06, "loss": 0.0475, "step": 72190 }, { "epoch": 6.06, "learning_rate": 7.8736983540477e-06, "loss": 0.0325, "step": 72200 }, { "epoch": 6.06, "learning_rate": 7.87201881088344e-06, "loss": 0.0002, "step": 72210 }, { "epoch": 6.06, "learning_rate": 7.870339267719181e-06, "loss": 0.0423, "step": 72220 }, { "epoch": 6.07, "learning_rate": 7.868659724554922e-06, "loss": 0.0005, "step": 72230 }, { "epoch": 6.07, "learning_rate": 7.866980181390662e-06, "loss": 0.0628, "step": 72240 }, { "epoch": 6.07, "learning_rate": 7.865300638226402e-06, "loss": 0.0376, "step": 72250 }, { "epoch": 6.07, "learning_rate": 7.863621095062145e-06, "loss": 0.0004, "step": 72260 }, { "epoch": 6.07, "learning_rate": 7.861941551897885e-06, "loss": 0.0381, "step": 72270 }, { "epoch": 6.07, "learning_rate": 7.860262008733624e-06, "loss": 0.0935, "step": 72280 }, { "epoch": 6.07, "learning_rate": 7.858582465569366e-06, "loss": 0.0334, "step": 72290 }, { "epoch": 6.07, "learning_rate": 7.856902922405107e-06, "loss": 0.0974, "step": 72300 }, { "epoch": 6.07, "learning_rate": 7.855223379240847e-06, "loss": 0.0311, "step": 72310 }, { "epoch": 6.07, "learning_rate": 7.853543836076587e-06, "loss": 0.0773, "step": 72320 }, { "epoch": 6.07, "learning_rate": 7.85186429291233e-06, "loss": 0.0004, "step": 72330 }, { "epoch": 6.07, "learning_rate": 7.850184749748068e-06, "loss": 0.0258, "step": 72340 }, { "epoch": 6.08, "learning_rate": 7.848505206583809e-06, "loss": 0.0673, "step": 72350 }, { "epoch": 6.08, "learning_rate": 7.846825663419551e-06, "loss": 0.0312, "step": 72360 }, { "epoch": 6.08, "learning_rate": 7.845146120255292e-06, "loss": 0.0004, "step": 72370 }, { "epoch": 6.08, "learning_rate": 7.843466577091032e-06, "loss": 0.0065, "step": 72380 }, { "epoch": 6.08, "learning_rate": 7.841787033926773e-06, "loss": 0.0153, "step": 72390 }, { "epoch": 6.08, "learning_rate": 7.840107490762513e-06, "loss": 0.0004, "step": 72400 }, { "epoch": 6.08, "learning_rate": 7.838427947598253e-06, "loss": 0.0394, "step": 72410 }, { "epoch": 6.08, "learning_rate": 7.836748404433994e-06, "loss": 0.0009, "step": 72420 }, { "epoch": 6.08, "learning_rate": 7.835068861269736e-06, "loss": 0.0214, "step": 72430 }, { "epoch": 6.08, "learning_rate": 7.833389318105475e-06, "loss": 0.0283, "step": 72440 }, { "epoch": 6.08, "learning_rate": 7.831709774941217e-06, "loss": 0.1145, "step": 72450 }, { "epoch": 6.08, "learning_rate": 7.830030231776958e-06, "loss": 0.0992, "step": 72460 }, { "epoch": 6.09, "learning_rate": 7.828350688612698e-06, "loss": 0.0364, "step": 72470 }, { "epoch": 6.09, "learning_rate": 7.826671145448439e-06, "loss": 0.0008, "step": 72480 }, { "epoch": 6.09, "learning_rate": 7.82499160228418e-06, "loss": 0.0222, "step": 72490 }, { "epoch": 6.09, "learning_rate": 7.82331205911992e-06, "loss": 0.0118, "step": 72500 }, { "epoch": 6.09, "learning_rate": 7.82163251595566e-06, "loss": 0.0088, "step": 72510 }, { "epoch": 6.09, "learning_rate": 7.819952972791402e-06, "loss": 0.056, "step": 72520 }, { "epoch": 6.09, "learning_rate": 7.818273429627143e-06, "loss": 0.0249, "step": 72530 }, { "epoch": 6.09, "learning_rate": 7.816593886462883e-06, "loss": 0.0003, "step": 72540 }, { "epoch": 6.09, "learning_rate": 7.814914343298624e-06, "loss": 0.001, "step": 72550 }, { "epoch": 6.09, "learning_rate": 7.813234800134364e-06, "loss": 0.061, "step": 72560 }, { "epoch": 6.09, "learning_rate": 7.811555256970104e-06, "loss": 0.0315, "step": 72570 }, { "epoch": 6.1, "learning_rate": 7.809875713805845e-06, "loss": 0.0669, "step": 72580 }, { "epoch": 6.1, "learning_rate": 7.808196170641587e-06, "loss": 0.0004, "step": 72590 }, { "epoch": 6.1, "learning_rate": 7.806516627477326e-06, "loss": 0.0007, "step": 72600 }, { "epoch": 6.1, "learning_rate": 7.804837084313066e-06, "loss": 0.0605, "step": 72610 }, { "epoch": 6.1, "learning_rate": 7.803157541148809e-06, "loss": 0.0514, "step": 72620 }, { "epoch": 6.1, "learning_rate": 7.801477997984549e-06, "loss": 0.0474, "step": 72630 }, { "epoch": 6.1, "learning_rate": 7.79979845482029e-06, "loss": 0.0771, "step": 72640 }, { "epoch": 6.1, "learning_rate": 7.79811891165603e-06, "loss": 0.0006, "step": 72650 }, { "epoch": 6.1, "learning_rate": 7.79643936849177e-06, "loss": 0.0089, "step": 72660 }, { "epoch": 6.1, "learning_rate": 7.794759825327511e-06, "loss": 0.0301, "step": 72670 }, { "epoch": 6.1, "learning_rate": 7.793080282163253e-06, "loss": 0.1255, "step": 72680 }, { "epoch": 6.1, "learning_rate": 7.791400738998994e-06, "loss": 0.0225, "step": 72690 }, { "epoch": 6.11, "learning_rate": 7.789721195834734e-06, "loss": 0.0028, "step": 72700 }, { "epoch": 6.11, "learning_rate": 7.788041652670475e-06, "loss": 0.0232, "step": 72710 }, { "epoch": 6.11, "learning_rate": 7.786362109506215e-06, "loss": 0.0186, "step": 72720 }, { "epoch": 6.11, "learning_rate": 7.784682566341956e-06, "loss": 0.138, "step": 72730 }, { "epoch": 6.11, "learning_rate": 7.783003023177696e-06, "loss": 0.0394, "step": 72740 }, { "epoch": 6.11, "learning_rate": 7.781323480013438e-06, "loss": 0.039, "step": 72750 }, { "epoch": 6.11, "learning_rate": 7.779643936849177e-06, "loss": 0.0008, "step": 72760 }, { "epoch": 6.11, "learning_rate": 7.777964393684917e-06, "loss": 0.1093, "step": 72770 }, { "epoch": 6.11, "learning_rate": 7.77628485052066e-06, "loss": 0.0385, "step": 72780 }, { "epoch": 6.11, "learning_rate": 7.7746053073564e-06, "loss": 0.0007, "step": 72790 }, { "epoch": 6.11, "learning_rate": 7.77292576419214e-06, "loss": 0.0004, "step": 72800 }, { "epoch": 6.11, "learning_rate": 7.771246221027881e-06, "loss": 0.1093, "step": 72810 }, { "epoch": 6.12, "learning_rate": 7.769566677863621e-06, "loss": 0.0158, "step": 72820 }, { "epoch": 6.12, "learning_rate": 7.767887134699362e-06, "loss": 0.0009, "step": 72830 }, { "epoch": 6.12, "learning_rate": 7.766207591535102e-06, "loss": 0.0479, "step": 72840 }, { "epoch": 6.12, "learning_rate": 7.764528048370845e-06, "loss": 0.0435, "step": 72850 }, { "epoch": 6.12, "learning_rate": 7.762848505206585e-06, "loss": 0.0109, "step": 72860 }, { "epoch": 6.12, "learning_rate": 7.761168962042326e-06, "loss": 0.0526, "step": 72870 }, { "epoch": 6.12, "learning_rate": 7.759489418878066e-06, "loss": 0.073, "step": 72880 }, { "epoch": 6.12, "learning_rate": 7.757809875713807e-06, "loss": 0.0035, "step": 72890 }, { "epoch": 6.12, "learning_rate": 7.756130332549547e-06, "loss": 0.1181, "step": 72900 }, { "epoch": 6.12, "learning_rate": 7.754450789385287e-06, "loss": 0.0009, "step": 72910 }, { "epoch": 6.12, "learning_rate": 7.752771246221028e-06, "loss": 0.0286, "step": 72920 }, { "epoch": 6.12, "learning_rate": 7.751091703056768e-06, "loss": 0.0373, "step": 72930 }, { "epoch": 6.13, "learning_rate": 7.74941215989251e-06, "loss": 0.0664, "step": 72940 }, { "epoch": 6.13, "learning_rate": 7.747732616728251e-06, "loss": 0.0168, "step": 72950 }, { "epoch": 6.13, "learning_rate": 7.746053073563992e-06, "loss": 0.0009, "step": 72960 }, { "epoch": 6.13, "learning_rate": 7.744373530399732e-06, "loss": 0.0549, "step": 72970 }, { "epoch": 6.13, "learning_rate": 7.742693987235473e-06, "loss": 0.0312, "step": 72980 }, { "epoch": 6.13, "learning_rate": 7.741014444071213e-06, "loss": 0.0097, "step": 72990 }, { "epoch": 6.13, "learning_rate": 7.739334900906953e-06, "loss": 0.0262, "step": 73000 }, { "epoch": 6.13, "learning_rate": 7.737655357742696e-06, "loss": 0.0003, "step": 73010 }, { "epoch": 6.13, "learning_rate": 7.735975814578436e-06, "loss": 0.0324, "step": 73020 }, { "epoch": 6.13, "learning_rate": 7.734296271414175e-06, "loss": 0.0564, "step": 73030 }, { "epoch": 6.13, "learning_rate": 7.732616728249917e-06, "loss": 0.0291, "step": 73040 }, { "epoch": 6.13, "learning_rate": 7.730937185085658e-06, "loss": 0.0129, "step": 73050 }, { "epoch": 6.14, "learning_rate": 7.729257641921398e-06, "loss": 0.0164, "step": 73060 }, { "epoch": 6.14, "learning_rate": 7.727578098757138e-06, "loss": 0.0541, "step": 73070 }, { "epoch": 6.14, "learning_rate": 7.725898555592879e-06, "loss": 0.0019, "step": 73080 }, { "epoch": 6.14, "learning_rate": 7.72421901242862e-06, "loss": 0.0234, "step": 73090 }, { "epoch": 6.14, "learning_rate": 7.72253946926436e-06, "loss": 0.0866, "step": 73100 }, { "epoch": 6.14, "learning_rate": 7.720859926100102e-06, "loss": 0.1198, "step": 73110 }, { "epoch": 6.14, "learning_rate": 7.719180382935843e-06, "loss": 0.04, "step": 73120 }, { "epoch": 6.14, "learning_rate": 7.717500839771583e-06, "loss": 0.0272, "step": 73130 }, { "epoch": 6.14, "learning_rate": 7.715821296607324e-06, "loss": 0.0371, "step": 73140 }, { "epoch": 6.14, "learning_rate": 7.714141753443064e-06, "loss": 0.0003, "step": 73150 }, { "epoch": 6.14, "learning_rate": 7.712462210278804e-06, "loss": 0.035, "step": 73160 }, { "epoch": 6.14, "learning_rate": 7.710782667114547e-06, "loss": 0.0256, "step": 73170 }, { "epoch": 6.15, "learning_rate": 7.709103123950287e-06, "loss": 0.0836, "step": 73180 }, { "epoch": 6.15, "learning_rate": 7.707423580786026e-06, "loss": 0.056, "step": 73190 }, { "epoch": 6.15, "learning_rate": 7.705744037621768e-06, "loss": 0.0194, "step": 73200 }, { "epoch": 6.15, "learning_rate": 7.704064494457509e-06, "loss": 0.0008, "step": 73210 }, { "epoch": 6.15, "learning_rate": 7.702384951293249e-06, "loss": 0.0365, "step": 73220 }, { "epoch": 6.15, "learning_rate": 7.70070540812899e-06, "loss": 0.0636, "step": 73230 }, { "epoch": 6.15, "learning_rate": 7.69902586496473e-06, "loss": 0.0439, "step": 73240 }, { "epoch": 6.15, "learning_rate": 7.69734632180047e-06, "loss": 0.0276, "step": 73250 }, { "epoch": 6.15, "learning_rate": 7.695666778636211e-06, "loss": 0.0675, "step": 73260 }, { "epoch": 6.15, "learning_rate": 7.693987235471953e-06, "loss": 0.0026, "step": 73270 }, { "epoch": 6.15, "learning_rate": 7.692307692307694e-06, "loss": 0.0726, "step": 73280 }, { "epoch": 6.15, "learning_rate": 7.690628149143432e-06, "loss": 0.0384, "step": 73290 }, { "epoch": 6.16, "learning_rate": 7.688948605979175e-06, "loss": 0.102, "step": 73300 }, { "epoch": 6.16, "learning_rate": 7.687269062814915e-06, "loss": 0.0014, "step": 73310 }, { "epoch": 6.16, "learning_rate": 7.685589519650655e-06, "loss": 0.0626, "step": 73320 }, { "epoch": 6.16, "learning_rate": 7.683909976486396e-06, "loss": 0.1093, "step": 73330 }, { "epoch": 6.16, "learning_rate": 7.682230433322136e-06, "loss": 0.0011, "step": 73340 }, { "epoch": 6.16, "learning_rate": 7.680550890157877e-06, "loss": 0.0645, "step": 73350 }, { "epoch": 6.16, "learning_rate": 7.678871346993617e-06, "loss": 0.0176, "step": 73360 }, { "epoch": 6.16, "learning_rate": 7.67719180382936e-06, "loss": 0.0234, "step": 73370 }, { "epoch": 6.16, "learning_rate": 7.6755122606651e-06, "loss": 0.0815, "step": 73380 }, { "epoch": 6.16, "learning_rate": 7.67383271750084e-06, "loss": 0.0423, "step": 73390 }, { "epoch": 6.16, "learning_rate": 7.672153174336581e-06, "loss": 0.052, "step": 73400 }, { "epoch": 6.16, "learning_rate": 7.670473631172321e-06, "loss": 0.0317, "step": 73410 }, { "epoch": 6.17, "learning_rate": 7.668794088008062e-06, "loss": 0.0239, "step": 73420 }, { "epoch": 6.17, "learning_rate": 7.667114544843804e-06, "loss": 0.0413, "step": 73430 }, { "epoch": 6.17, "learning_rate": 7.665435001679545e-06, "loss": 0.0011, "step": 73440 }, { "epoch": 6.17, "learning_rate": 7.663755458515283e-06, "loss": 0.0268, "step": 73450 }, { "epoch": 6.17, "learning_rate": 7.662075915351026e-06, "loss": 0.1031, "step": 73460 }, { "epoch": 6.17, "learning_rate": 7.660396372186766e-06, "loss": 0.0692, "step": 73470 }, { "epoch": 6.17, "learning_rate": 7.658716829022507e-06, "loss": 0.0284, "step": 73480 }, { "epoch": 6.17, "learning_rate": 7.657037285858247e-06, "loss": 0.0017, "step": 73490 }, { "epoch": 6.17, "learning_rate": 7.655357742693987e-06, "loss": 0.0472, "step": 73500 }, { "epoch": 6.17, "learning_rate": 7.653678199529728e-06, "loss": 0.0165, "step": 73510 }, { "epoch": 6.17, "learning_rate": 7.651998656365468e-06, "loss": 0.024, "step": 73520 }, { "epoch": 6.17, "learning_rate": 7.65031911320121e-06, "loss": 0.0532, "step": 73530 }, { "epoch": 6.18, "learning_rate": 7.648639570036951e-06, "loss": 0.0008, "step": 73540 }, { "epoch": 6.18, "learning_rate": 7.646960026872692e-06, "loss": 0.0468, "step": 73550 }, { "epoch": 6.18, "learning_rate": 7.645280483708432e-06, "loss": 0.0112, "step": 73560 }, { "epoch": 6.18, "learning_rate": 7.643600940544172e-06, "loss": 0.0816, "step": 73570 }, { "epoch": 6.18, "learning_rate": 7.641921397379913e-06, "loss": 0.067, "step": 73580 }, { "epoch": 6.18, "learning_rate": 7.640241854215653e-06, "loss": 0.0011, "step": 73590 }, { "epoch": 6.18, "learning_rate": 7.638562311051396e-06, "loss": 0.0506, "step": 73600 }, { "epoch": 6.18, "learning_rate": 7.636882767887134e-06, "loss": 0.0005, "step": 73610 }, { "epoch": 6.18, "learning_rate": 7.635203224722877e-06, "loss": 0.0326, "step": 73620 }, { "epoch": 6.18, "learning_rate": 7.633523681558617e-06, "loss": 0.0378, "step": 73630 }, { "epoch": 6.18, "learning_rate": 7.631844138394358e-06, "loss": 0.0747, "step": 73640 }, { "epoch": 6.18, "learning_rate": 7.630164595230098e-06, "loss": 0.001, "step": 73650 }, { "epoch": 6.19, "learning_rate": 7.6284850520658385e-06, "loss": 0.0008, "step": 73660 }, { "epoch": 6.19, "learning_rate": 7.626805508901579e-06, "loss": 0.024, "step": 73670 }, { "epoch": 6.19, "learning_rate": 7.62512596573732e-06, "loss": 0.0002, "step": 73680 }, { "epoch": 6.19, "learning_rate": 7.623446422573061e-06, "loss": 0.0906, "step": 73690 }, { "epoch": 6.19, "learning_rate": 7.621766879408802e-06, "loss": 0.1356, "step": 73700 }, { "epoch": 6.19, "learning_rate": 7.620087336244542e-06, "loss": 0.0306, "step": 73710 }, { "epoch": 6.19, "learning_rate": 7.618407793080282e-06, "loss": 0.0918, "step": 73720 }, { "epoch": 6.19, "learning_rate": 7.6167282499160235e-06, "loss": 0.0256, "step": 73730 }, { "epoch": 6.19, "learning_rate": 7.615048706751764e-06, "loss": 0.0557, "step": 73740 }, { "epoch": 6.19, "learning_rate": 7.613369163587505e-06, "loss": 0.1067, "step": 73750 }, { "epoch": 6.19, "learning_rate": 7.611689620423246e-06, "loss": 0.0018, "step": 73760 }, { "epoch": 6.19, "learning_rate": 7.610010077258985e-06, "loss": 0.0429, "step": 73770 }, { "epoch": 6.2, "learning_rate": 7.608330534094727e-06, "loss": 0.0008, "step": 73780 }, { "epoch": 6.2, "learning_rate": 7.606650990930467e-06, "loss": 0.0005, "step": 73790 }, { "epoch": 6.2, "learning_rate": 7.6049714477662085e-06, "loss": 0.0016, "step": 73800 }, { "epoch": 6.2, "learning_rate": 7.603291904601949e-06, "loss": 0.0498, "step": 73810 }, { "epoch": 6.2, "learning_rate": 7.6016123614376895e-06, "loss": 0.03, "step": 73820 }, { "epoch": 6.2, "learning_rate": 7.59993281827343e-06, "loss": 0.0286, "step": 73830 }, { "epoch": 6.2, "learning_rate": 7.598253275109171e-06, "loss": 0.0298, "step": 73840 }, { "epoch": 6.2, "learning_rate": 7.596573731944912e-06, "loss": 0.0227, "step": 73850 }, { "epoch": 6.2, "learning_rate": 7.594894188780652e-06, "loss": 0.0791, "step": 73860 }, { "epoch": 6.2, "learning_rate": 7.593214645616393e-06, "loss": 0.0011, "step": 73870 }, { "epoch": 6.2, "learning_rate": 7.591535102452133e-06, "loss": 0.0247, "step": 73880 }, { "epoch": 6.21, "learning_rate": 7.5898555592878745e-06, "loss": 0.0134, "step": 73890 }, { "epoch": 6.21, "learning_rate": 7.588176016123615e-06, "loss": 0.0029, "step": 73900 }, { "epoch": 6.21, "learning_rate": 7.586496472959356e-06, "loss": 0.0541, "step": 73910 }, { "epoch": 6.21, "learning_rate": 7.584816929795097e-06, "loss": 0.0004, "step": 73920 }, { "epoch": 6.21, "learning_rate": 7.5831373866308364e-06, "loss": 0.0375, "step": 73930 }, { "epoch": 6.21, "learning_rate": 7.581457843466578e-06, "loss": 0.0924, "step": 73940 }, { "epoch": 6.21, "learning_rate": 7.579778300302318e-06, "loss": 0.0257, "step": 73950 }, { "epoch": 6.21, "learning_rate": 7.5780987571380596e-06, "loss": 0.001, "step": 73960 }, { "epoch": 6.21, "learning_rate": 7.5764192139738e-06, "loss": 0.0908, "step": 73970 }, { "epoch": 6.21, "learning_rate": 7.57473967080954e-06, "loss": 0.0013, "step": 73980 }, { "epoch": 6.21, "learning_rate": 7.573060127645281e-06, "loss": 0.0216, "step": 73990 }, { "epoch": 6.21, "learning_rate": 7.5713805844810215e-06, "loss": 0.0212, "step": 74000 }, { "epoch": 6.22, "learning_rate": 7.569701041316763e-06, "loss": 0.0546, "step": 74010 }, { "epoch": 6.22, "learning_rate": 7.568021498152503e-06, "loss": 0.0005, "step": 74020 }, { "epoch": 6.22, "learning_rate": 7.566341954988243e-06, "loss": 0.0358, "step": 74030 }, { "epoch": 6.22, "learning_rate": 7.564662411823984e-06, "loss": 0.0016, "step": 74040 }, { "epoch": 6.22, "learning_rate": 7.562982868659725e-06, "loss": 0.0495, "step": 74050 }, { "epoch": 6.22, "learning_rate": 7.561303325495466e-06, "loss": 0.0006, "step": 74060 }, { "epoch": 6.22, "learning_rate": 7.5596237823312065e-06, "loss": 0.0429, "step": 74070 }, { "epoch": 6.22, "learning_rate": 7.557944239166948e-06, "loss": 0.0221, "step": 74080 }, { "epoch": 6.22, "learning_rate": 7.5562646960026874e-06, "loss": 0.031, "step": 74090 }, { "epoch": 6.22, "learning_rate": 7.554585152838429e-06, "loss": 0.0338, "step": 74100 }, { "epoch": 6.22, "learning_rate": 7.552905609674169e-06, "loss": 0.0412, "step": 74110 }, { "epoch": 6.22, "learning_rate": 7.5512260665099106e-06, "loss": 0.0008, "step": 74120 }, { "epoch": 6.23, "learning_rate": 7.549546523345651e-06, "loss": 0.0002, "step": 74130 }, { "epoch": 6.23, "learning_rate": 7.547866980181391e-06, "loss": 0.0172, "step": 74140 }, { "epoch": 6.23, "learning_rate": 7.546187437017132e-06, "loss": 0.0196, "step": 74150 }, { "epoch": 6.23, "learning_rate": 7.5445078938528725e-06, "loss": 0.0005, "step": 74160 }, { "epoch": 6.23, "learning_rate": 7.542828350688614e-06, "loss": 0.0012, "step": 74170 }, { "epoch": 6.23, "learning_rate": 7.541148807524354e-06, "loss": 0.0159, "step": 74180 }, { "epoch": 6.23, "learning_rate": 7.539469264360094e-06, "loss": 0.1195, "step": 74190 }, { "epoch": 6.23, "learning_rate": 7.537789721195835e-06, "loss": 0.0003, "step": 74200 }, { "epoch": 6.23, "learning_rate": 7.536110178031576e-06, "loss": 0.0495, "step": 74210 }, { "epoch": 6.23, "learning_rate": 7.534430634867317e-06, "loss": 0.0419, "step": 74220 }, { "epoch": 6.23, "learning_rate": 7.5327510917030575e-06, "loss": 0.0571, "step": 74230 }, { "epoch": 6.23, "learning_rate": 7.531071548538797e-06, "loss": 0.1609, "step": 74240 }, { "epoch": 6.24, "learning_rate": 7.5293920053745385e-06, "loss": 0.0006, "step": 74250 }, { "epoch": 6.24, "learning_rate": 7.527712462210279e-06, "loss": 0.0231, "step": 74260 }, { "epoch": 6.24, "learning_rate": 7.52603291904602e-06, "loss": 0.0032, "step": 74270 }, { "epoch": 6.24, "learning_rate": 7.524353375881761e-06, "loss": 0.0009, "step": 74280 }, { "epoch": 6.24, "learning_rate": 7.522673832717502e-06, "loss": 0.0003, "step": 74290 }, { "epoch": 6.24, "learning_rate": 7.520994289553242e-06, "loss": 0.0264, "step": 74300 }, { "epoch": 6.24, "learning_rate": 7.519314746388983e-06, "loss": 0.0006, "step": 74310 }, { "epoch": 6.24, "learning_rate": 7.5176352032247235e-06, "loss": 0.0238, "step": 74320 }, { "epoch": 6.24, "learning_rate": 7.515955660060465e-06, "loss": 0.0299, "step": 74330 }, { "epoch": 6.24, "learning_rate": 7.514276116896205e-06, "loss": 0.0277, "step": 74340 }, { "epoch": 6.24, "learning_rate": 7.512596573731945e-06, "loss": 0.0432, "step": 74350 }, { "epoch": 6.24, "learning_rate": 7.510917030567686e-06, "loss": 0.0767, "step": 74360 }, { "epoch": 6.25, "learning_rate": 7.509237487403427e-06, "loss": 0.018, "step": 74370 }, { "epoch": 6.25, "learning_rate": 7.507557944239168e-06, "loss": 0.0923, "step": 74380 }, { "epoch": 6.25, "learning_rate": 7.5058784010749085e-06, "loss": 0.0006, "step": 74390 }, { "epoch": 6.25, "learning_rate": 7.504198857910648e-06, "loss": 0.1427, "step": 74400 }, { "epoch": 6.25, "learning_rate": 7.5025193147463895e-06, "loss": 0.0004, "step": 74410 }, { "epoch": 6.25, "learning_rate": 7.50083977158213e-06, "loss": 0.0208, "step": 74420 }, { "epoch": 6.25, "learning_rate": 7.499160228417871e-06, "loss": 0.0012, "step": 74430 }, { "epoch": 6.25, "learning_rate": 7.497480685253612e-06, "loss": 0.004, "step": 74440 }, { "epoch": 6.25, "learning_rate": 7.495801142089353e-06, "loss": 0.0002, "step": 74450 }, { "epoch": 6.25, "learning_rate": 7.494121598925093e-06, "loss": 0.0028, "step": 74460 }, { "epoch": 6.25, "learning_rate": 7.492442055760833e-06, "loss": 0.0234, "step": 74470 }, { "epoch": 6.25, "learning_rate": 7.4907625125965745e-06, "loss": 0.0498, "step": 74480 }, { "epoch": 6.26, "learning_rate": 7.489082969432315e-06, "loss": 0.0061, "step": 74490 }, { "epoch": 6.26, "learning_rate": 7.487403426268056e-06, "loss": 0.0879, "step": 74500 }, { "epoch": 6.26, "learning_rate": 7.485723883103796e-06, "loss": 0.0917, "step": 74510 }, { "epoch": 6.26, "learning_rate": 7.484044339939536e-06, "loss": 0.0903, "step": 74520 }, { "epoch": 6.26, "learning_rate": 7.482364796775278e-06, "loss": 0.0347, "step": 74530 }, { "epoch": 6.26, "learning_rate": 7.480685253611018e-06, "loss": 0.0547, "step": 74540 }, { "epoch": 6.26, "learning_rate": 7.4790057104467595e-06, "loss": 0.007, "step": 74550 }, { "epoch": 6.26, "learning_rate": 7.477326167282499e-06, "loss": 0.0399, "step": 74560 }, { "epoch": 6.26, "learning_rate": 7.4756466241182405e-06, "loss": 0.0346, "step": 74570 }, { "epoch": 6.26, "learning_rate": 7.473967080953981e-06, "loss": 0.0514, "step": 74580 }, { "epoch": 6.26, "learning_rate": 7.472287537789722e-06, "loss": 0.0208, "step": 74590 }, { "epoch": 6.26, "learning_rate": 7.470607994625463e-06, "loss": 0.0809, "step": 74600 }, { "epoch": 6.27, "learning_rate": 7.468928451461204e-06, "loss": 0.0007, "step": 74610 }, { "epoch": 6.27, "learning_rate": 7.467248908296944e-06, "loss": 0.0226, "step": 74620 }, { "epoch": 6.27, "learning_rate": 7.465569365132684e-06, "loss": 0.0505, "step": 74630 }, { "epoch": 6.27, "learning_rate": 7.4638898219684255e-06, "loss": 0.0011, "step": 74640 }, { "epoch": 6.27, "learning_rate": 7.462210278804166e-06, "loss": 0.0308, "step": 74650 }, { "epoch": 6.27, "learning_rate": 7.460530735639907e-06, "loss": 0.0293, "step": 74660 }, { "epoch": 6.27, "learning_rate": 7.458851192475647e-06, "loss": 0.0007, "step": 74670 }, { "epoch": 6.27, "learning_rate": 7.4571716493113874e-06, "loss": 0.0018, "step": 74680 }, { "epoch": 6.27, "learning_rate": 7.455492106147129e-06, "loss": 0.0296, "step": 74690 }, { "epoch": 6.27, "learning_rate": 7.453812562982869e-06, "loss": 0.0613, "step": 74700 }, { "epoch": 6.27, "learning_rate": 7.4521330198186106e-06, "loss": 0.0269, "step": 74710 }, { "epoch": 6.27, "learning_rate": 7.45045347665435e-06, "loss": 0.0276, "step": 74720 }, { "epoch": 6.28, "learning_rate": 7.448773933490091e-06, "loss": 0.0226, "step": 74730 }, { "epoch": 6.28, "learning_rate": 7.447094390325832e-06, "loss": 0.0708, "step": 74740 }, { "epoch": 6.28, "learning_rate": 7.4454148471615725e-06, "loss": 0.0005, "step": 74750 }, { "epoch": 6.28, "learning_rate": 7.443735303997314e-06, "loss": 0.0013, "step": 74760 }, { "epoch": 6.28, "learning_rate": 7.442055760833053e-06, "loss": 0.0231, "step": 74770 }, { "epoch": 6.28, "learning_rate": 7.440376217668795e-06, "loss": 0.0296, "step": 74780 }, { "epoch": 6.28, "learning_rate": 7.438696674504535e-06, "loss": 0.028, "step": 74790 }, { "epoch": 6.28, "learning_rate": 7.4370171313402765e-06, "loss": 0.0688, "step": 74800 }, { "epoch": 6.28, "learning_rate": 7.435337588176017e-06, "loss": 0.0881, "step": 74810 }, { "epoch": 6.28, "learning_rate": 7.433658045011758e-06, "loss": 0.0233, "step": 74820 }, { "epoch": 6.28, "learning_rate": 7.431978501847498e-06, "loss": 0.0432, "step": 74830 }, { "epoch": 6.28, "learning_rate": 7.4302989586832384e-06, "loss": 0.0616, "step": 74840 }, { "epoch": 6.29, "learning_rate": 7.42861941551898e-06, "loss": 0.0653, "step": 74850 }, { "epoch": 6.29, "learning_rate": 7.42693987235472e-06, "loss": 0.0531, "step": 74860 }, { "epoch": 6.29, "learning_rate": 7.4252603291904616e-06, "loss": 0.0006, "step": 74870 }, { "epoch": 6.29, "learning_rate": 7.423580786026201e-06, "loss": 0.0008, "step": 74880 }, { "epoch": 6.29, "learning_rate": 7.421901242861942e-06, "loss": 0.0473, "step": 74890 }, { "epoch": 6.29, "learning_rate": 7.420221699697683e-06, "loss": 0.0767, "step": 74900 }, { "epoch": 6.29, "learning_rate": 7.4185421565334235e-06, "loss": 0.0405, "step": 74910 }, { "epoch": 6.29, "learning_rate": 7.416862613369165e-06, "loss": 0.0318, "step": 74920 }, { "epoch": 6.29, "learning_rate": 7.4151830702049044e-06, "loss": 0.001, "step": 74930 }, { "epoch": 6.29, "learning_rate": 7.413503527040645e-06, "loss": 0.0459, "step": 74940 }, { "epoch": 6.29, "learning_rate": 7.411823983876386e-06, "loss": 0.0005, "step": 74950 }, { "epoch": 6.29, "learning_rate": 7.410144440712127e-06, "loss": 0.0388, "step": 74960 }, { "epoch": 6.3, "learning_rate": 7.408464897547868e-06, "loss": 0.081, "step": 74970 }, { "epoch": 6.3, "learning_rate": 7.4067853543836085e-06, "loss": 0.0212, "step": 74980 }, { "epoch": 6.3, "learning_rate": 7.405105811219349e-06, "loss": 0.0122, "step": 74990 }, { "epoch": 6.3, "learning_rate": 7.4034262680550895e-06, "loss": 0.0001, "step": 75000 }, { "epoch": 6.3, "learning_rate": 7.40174672489083e-06, "loss": 0.0888, "step": 75010 }, { "epoch": 6.3, "learning_rate": 7.400067181726571e-06, "loss": 0.0477, "step": 75020 }, { "epoch": 6.3, "learning_rate": 7.398387638562312e-06, "loss": 0.0042, "step": 75030 }, { "epoch": 6.3, "learning_rate": 7.396708095398052e-06, "loss": 0.0002, "step": 75040 }, { "epoch": 6.3, "learning_rate": 7.395028552233793e-06, "loss": 0.0391, "step": 75050 }, { "epoch": 6.3, "learning_rate": 7.393349009069534e-06, "loss": 0.0003, "step": 75060 }, { "epoch": 6.3, "learning_rate": 7.3916694659052745e-06, "loss": 0.0563, "step": 75070 }, { "epoch": 6.31, "learning_rate": 7.389989922741016e-06, "loss": 0.0644, "step": 75080 }, { "epoch": 6.31, "learning_rate": 7.3883103795767554e-06, "loss": 0.0215, "step": 75090 }, { "epoch": 6.31, "learning_rate": 7.386630836412496e-06, "loss": 0.0002, "step": 75100 }, { "epoch": 6.31, "learning_rate": 7.384951293248237e-06, "loss": 0.0426, "step": 75110 }, { "epoch": 6.31, "learning_rate": 7.383271750083978e-06, "loss": 0.001, "step": 75120 }, { "epoch": 6.31, "learning_rate": 7.381592206919719e-06, "loss": 0.0877, "step": 75130 }, { "epoch": 6.31, "learning_rate": 7.3799126637554595e-06, "loss": 0.0036, "step": 75140 }, { "epoch": 6.31, "learning_rate": 7.378233120591199e-06, "loss": 0.0433, "step": 75150 }, { "epoch": 6.31, "learning_rate": 7.3765535774269405e-06, "loss": 0.0006, "step": 75160 }, { "epoch": 6.31, "learning_rate": 7.374874034262681e-06, "loss": 0.0292, "step": 75170 }, { "epoch": 6.31, "learning_rate": 7.373194491098422e-06, "loss": 0.0658, "step": 75180 }, { "epoch": 6.31, "learning_rate": 7.371514947934163e-06, "loss": 0.0228, "step": 75190 }, { "epoch": 6.32, "learning_rate": 7.369835404769902e-06, "loss": 0.0337, "step": 75200 }, { "epoch": 6.32, "learning_rate": 7.368155861605644e-06, "loss": 0.0207, "step": 75210 }, { "epoch": 6.32, "learning_rate": 7.366476318441384e-06, "loss": 0.0669, "step": 75220 }, { "epoch": 6.32, "learning_rate": 7.3647967752771255e-06, "loss": 0.0605, "step": 75230 }, { "epoch": 6.32, "learning_rate": 7.363117232112866e-06, "loss": 0.0583, "step": 75240 }, { "epoch": 6.32, "learning_rate": 7.3614376889486065e-06, "loss": 0.0702, "step": 75250 }, { "epoch": 6.32, "learning_rate": 7.359758145784347e-06, "loss": 0.0349, "step": 75260 }, { "epoch": 6.32, "learning_rate": 7.358078602620088e-06, "loss": 0.0267, "step": 75270 }, { "epoch": 6.32, "learning_rate": 7.356399059455829e-06, "loss": 0.0007, "step": 75280 }, { "epoch": 6.32, "learning_rate": 7.35471951629157e-06, "loss": 0.0145, "step": 75290 }, { "epoch": 6.32, "learning_rate": 7.35303997312731e-06, "loss": 0.0166, "step": 75300 }, { "epoch": 6.32, "learning_rate": 7.35136042996305e-06, "loss": 0.052, "step": 75310 }, { "epoch": 6.33, "learning_rate": 7.3496808867987915e-06, "loss": 0.0564, "step": 75320 }, { "epoch": 6.33, "learning_rate": 7.348001343634532e-06, "loss": 0.0011, "step": 75330 }, { "epoch": 6.33, "learning_rate": 7.346321800470273e-06, "loss": 0.1295, "step": 75340 }, { "epoch": 6.33, "learning_rate": 7.344642257306014e-06, "loss": 0.04, "step": 75350 }, { "epoch": 6.33, "learning_rate": 7.342962714141753e-06, "loss": 0.0006, "step": 75360 }, { "epoch": 6.33, "learning_rate": 7.341283170977495e-06, "loss": 0.0477, "step": 75370 }, { "epoch": 6.33, "learning_rate": 7.339603627813235e-06, "loss": 0.0359, "step": 75380 }, { "epoch": 6.33, "learning_rate": 7.3379240846489765e-06, "loss": 0.063, "step": 75390 }, { "epoch": 6.33, "learning_rate": 7.336244541484717e-06, "loss": 0.037, "step": 75400 }, { "epoch": 6.33, "learning_rate": 7.334564998320457e-06, "loss": 0.0217, "step": 75410 }, { "epoch": 6.33, "learning_rate": 7.332885455156198e-06, "loss": 0.0716, "step": 75420 }, { "epoch": 6.33, "learning_rate": 7.3312059119919384e-06, "loss": 0.1136, "step": 75430 }, { "epoch": 6.34, "learning_rate": 7.32952636882768e-06, "loss": 0.0347, "step": 75440 }, { "epoch": 6.34, "learning_rate": 7.32784682566342e-06, "loss": 0.0248, "step": 75450 }, { "epoch": 6.34, "learning_rate": 7.326167282499161e-06, "loss": 0.0541, "step": 75460 }, { "epoch": 6.34, "learning_rate": 7.324487739334901e-06, "loss": 0.0652, "step": 75470 }, { "epoch": 6.34, "learning_rate": 7.3228081961706425e-06, "loss": 0.0248, "step": 75480 }, { "epoch": 6.34, "learning_rate": 7.321128653006383e-06, "loss": 0.0386, "step": 75490 }, { "epoch": 6.34, "learning_rate": 7.3194491098421235e-06, "loss": 0.0213, "step": 75500 }, { "epoch": 6.34, "learning_rate": 7.317769566677865e-06, "loss": 0.0022, "step": 75510 }, { "epoch": 6.34, "learning_rate": 7.316090023513604e-06, "loss": 0.0236, "step": 75520 }, { "epoch": 6.34, "learning_rate": 7.314410480349346e-06, "loss": 0.0004, "step": 75530 }, { "epoch": 6.34, "learning_rate": 7.312730937185086e-06, "loss": 0.0231, "step": 75540 }, { "epoch": 6.34, "learning_rate": 7.3110513940208275e-06, "loss": 0.0021, "step": 75550 }, { "epoch": 6.35, "learning_rate": 7.309371850856568e-06, "loss": 0.0353, "step": 75560 }, { "epoch": 6.35, "learning_rate": 7.307692307692308e-06, "loss": 0.0269, "step": 75570 }, { "epoch": 6.35, "learning_rate": 7.306012764528049e-06, "loss": 0.0128, "step": 75580 }, { "epoch": 6.35, "learning_rate": 7.3043332213637894e-06, "loss": 0.0664, "step": 75590 }, { "epoch": 6.35, "learning_rate": 7.302653678199531e-06, "loss": 0.0324, "step": 75600 }, { "epoch": 6.35, "learning_rate": 7.300974135035271e-06, "loss": 0.0771, "step": 75610 }, { "epoch": 6.35, "learning_rate": 7.299294591871011e-06, "loss": 0.0617, "step": 75620 }, { "epoch": 6.35, "learning_rate": 7.297615048706752e-06, "loss": 0.0988, "step": 75630 }, { "epoch": 6.35, "learning_rate": 7.295935505542493e-06, "loss": 0.0362, "step": 75640 }, { "epoch": 6.35, "learning_rate": 7.294255962378234e-06, "loss": 0.0467, "step": 75650 }, { "epoch": 6.35, "learning_rate": 7.2925764192139745e-06, "loss": 0.0639, "step": 75660 }, { "epoch": 6.35, "learning_rate": 7.290896876049714e-06, "loss": 0.0468, "step": 75670 }, { "epoch": 6.36, "learning_rate": 7.289217332885455e-06, "loss": 0.0847, "step": 75680 }, { "epoch": 6.36, "learning_rate": 7.287537789721196e-06, "loss": 0.0598, "step": 75690 }, { "epoch": 6.36, "learning_rate": 7.285858246556937e-06, "loss": 0.052, "step": 75700 }, { "epoch": 6.36, "learning_rate": 7.284178703392678e-06, "loss": 0.0168, "step": 75710 }, { "epoch": 6.36, "learning_rate": 7.282499160228419e-06, "loss": 0.0284, "step": 75720 }, { "epoch": 6.36, "learning_rate": 7.280819617064159e-06, "loss": 0.037, "step": 75730 }, { "epoch": 6.36, "learning_rate": 7.2791400738999e-06, "loss": 0.0475, "step": 75740 }, { "epoch": 6.36, "learning_rate": 7.2774605307356405e-06, "loss": 0.0012, "step": 75750 }, { "epoch": 6.36, "learning_rate": 7.275780987571382e-06, "loss": 0.0541, "step": 75760 }, { "epoch": 6.36, "learning_rate": 7.274101444407122e-06, "loss": 0.1169, "step": 75770 }, { "epoch": 6.36, "learning_rate": 7.272421901242862e-06, "loss": 0.0792, "step": 75780 }, { "epoch": 6.36, "learning_rate": 7.270742358078603e-06, "loss": 0.0389, "step": 75790 }, { "epoch": 6.37, "learning_rate": 7.269062814914344e-06, "loss": 0.0123, "step": 75800 }, { "epoch": 6.37, "learning_rate": 7.267383271750085e-06, "loss": 0.066, "step": 75810 }, { "epoch": 6.37, "learning_rate": 7.2657037285858255e-06, "loss": 0.018, "step": 75820 }, { "epoch": 6.37, "learning_rate": 7.264024185421565e-06, "loss": 0.0546, "step": 75830 }, { "epoch": 6.37, "learning_rate": 7.2623446422573064e-06, "loss": 0.0224, "step": 75840 }, { "epoch": 6.37, "learning_rate": 7.260665099093047e-06, "loss": 0.0292, "step": 75850 }, { "epoch": 6.37, "learning_rate": 7.258985555928788e-06, "loss": 0.0109, "step": 75860 }, { "epoch": 6.37, "learning_rate": 7.257306012764529e-06, "loss": 0.0731, "step": 75870 }, { "epoch": 6.37, "learning_rate": 7.25562646960027e-06, "loss": 0.0479, "step": 75880 }, { "epoch": 6.37, "learning_rate": 7.25394692643601e-06, "loss": 0.063, "step": 75890 }, { "epoch": 6.37, "learning_rate": 7.25226738327175e-06, "loss": 0.0465, "step": 75900 }, { "epoch": 6.37, "learning_rate": 7.2505878401074915e-06, "loss": 0.0021, "step": 75910 }, { "epoch": 6.38, "learning_rate": 7.248908296943232e-06, "loss": 0.0429, "step": 75920 }, { "epoch": 6.38, "learning_rate": 7.247228753778973e-06, "loss": 0.0458, "step": 75930 }, { "epoch": 6.38, "learning_rate": 7.245549210614713e-06, "loss": 0.0745, "step": 75940 }, { "epoch": 6.38, "learning_rate": 7.243869667450454e-06, "loss": 0.003, "step": 75950 }, { "epoch": 6.38, "learning_rate": 7.242190124286195e-06, "loss": 0.0018, "step": 75960 }, { "epoch": 6.38, "learning_rate": 7.240510581121936e-06, "loss": 0.0825, "step": 75970 }, { "epoch": 6.38, "learning_rate": 7.2388310379576765e-06, "loss": 0.0294, "step": 75980 }, { "epoch": 6.38, "learning_rate": 7.237151494793416e-06, "loss": 0.0057, "step": 75990 }, { "epoch": 6.38, "learning_rate": 7.2354719516291575e-06, "loss": 0.0379, "step": 76000 }, { "epoch": 6.38, "learning_rate": 7.233792408464898e-06, "loss": 0.0229, "step": 76010 }, { "epoch": 6.38, "learning_rate": 7.232112865300639e-06, "loss": 0.1077, "step": 76020 }, { "epoch": 6.38, "learning_rate": 7.23043332213638e-06, "loss": 0.0004, "step": 76030 }, { "epoch": 6.39, "learning_rate": 7.228753778972121e-06, "loss": 0.0163, "step": 76040 }, { "epoch": 6.39, "learning_rate": 7.227074235807861e-06, "loss": 0.0022, "step": 76050 }, { "epoch": 6.39, "learning_rate": 7.225394692643601e-06, "loss": 0.0002, "step": 76060 }, { "epoch": 6.39, "learning_rate": 7.2237151494793425e-06, "loss": 0.0236, "step": 76070 }, { "epoch": 6.39, "learning_rate": 7.222035606315083e-06, "loss": 0.0542, "step": 76080 }, { "epoch": 6.39, "learning_rate": 7.220356063150824e-06, "loss": 0.0849, "step": 76090 }, { "epoch": 6.39, "learning_rate": 7.218676519986564e-06, "loss": 0.0012, "step": 76100 }, { "epoch": 6.39, "learning_rate": 7.216996976822304e-06, "loss": 0.0012, "step": 76110 }, { "epoch": 6.39, "learning_rate": 7.215317433658046e-06, "loss": 0.0471, "step": 76120 }, { "epoch": 6.39, "learning_rate": 7.213637890493786e-06, "loss": 0.0461, "step": 76130 }, { "epoch": 6.39, "learning_rate": 7.2119583473295275e-06, "loss": 0.0593, "step": 76140 }, { "epoch": 6.39, "learning_rate": 7.210278804165267e-06, "loss": 0.002, "step": 76150 }, { "epoch": 6.4, "learning_rate": 7.208599261001008e-06, "loss": 0.0228, "step": 76160 }, { "epoch": 6.4, "learning_rate": 7.206919717836749e-06, "loss": 0.0233, "step": 76170 }, { "epoch": 6.4, "learning_rate": 7.205240174672489e-06, "loss": 0.0009, "step": 76180 }, { "epoch": 6.4, "learning_rate": 7.203560631508231e-06, "loss": 0.0547, "step": 76190 }, { "epoch": 6.4, "learning_rate": 7.20188108834397e-06, "loss": 0.055, "step": 76200 }, { "epoch": 6.4, "learning_rate": 7.200201545179712e-06, "loss": 0.0223, "step": 76210 }, { "epoch": 6.4, "learning_rate": 7.198522002015452e-06, "loss": 0.001, "step": 76220 }, { "epoch": 6.4, "learning_rate": 7.1968424588511935e-06, "loss": 0.0256, "step": 76230 }, { "epoch": 6.4, "learning_rate": 7.195162915686934e-06, "loss": 0.0822, "step": 76240 }, { "epoch": 6.4, "learning_rate": 7.193483372522675e-06, "loss": 0.0741, "step": 76250 }, { "epoch": 6.4, "learning_rate": 7.191803829358415e-06, "loss": 0.0877, "step": 76260 }, { "epoch": 6.4, "learning_rate": 7.190124286194155e-06, "loss": 0.0257, "step": 76270 }, { "epoch": 6.41, "learning_rate": 7.188444743029897e-06, "loss": 0.0527, "step": 76280 }, { "epoch": 6.41, "learning_rate": 7.186765199865637e-06, "loss": 0.0006, "step": 76290 }, { "epoch": 6.41, "learning_rate": 7.1850856567013785e-06, "loss": 0.0027, "step": 76300 }, { "epoch": 6.41, "learning_rate": 7.183406113537118e-06, "loss": 0.0003, "step": 76310 }, { "epoch": 6.41, "learning_rate": 7.181726570372859e-06, "loss": 0.0011, "step": 76320 }, { "epoch": 6.41, "learning_rate": 7.1800470272086e-06, "loss": 0.0224, "step": 76330 }, { "epoch": 6.41, "learning_rate": 7.1783674840443404e-06, "loss": 0.0259, "step": 76340 }, { "epoch": 6.41, "learning_rate": 7.176687940880082e-06, "loss": 0.0169, "step": 76350 }, { "epoch": 6.41, "learning_rate": 7.175008397715821e-06, "loss": 0.1472, "step": 76360 }, { "epoch": 6.41, "learning_rate": 7.173328854551562e-06, "loss": 0.0031, "step": 76370 }, { "epoch": 6.41, "learning_rate": 7.171649311387303e-06, "loss": 0.0254, "step": 76380 }, { "epoch": 6.42, "learning_rate": 7.169969768223044e-06, "loss": 0.0004, "step": 76390 }, { "epoch": 6.42, "learning_rate": 7.168290225058785e-06, "loss": 0.0431, "step": 76400 }, { "epoch": 6.42, "learning_rate": 7.1666106818945255e-06, "loss": 0.0018, "step": 76410 }, { "epoch": 6.42, "learning_rate": 7.164931138730266e-06, "loss": 0.0678, "step": 76420 }, { "epoch": 6.42, "learning_rate": 7.163251595566006e-06, "loss": 0.0985, "step": 76430 }, { "epoch": 6.42, "learning_rate": 7.161572052401748e-06, "loss": 0.0257, "step": 76440 }, { "epoch": 6.42, "learning_rate": 7.159892509237488e-06, "loss": 0.0184, "step": 76450 }, { "epoch": 6.42, "learning_rate": 7.1582129660732295e-06, "loss": 0.0196, "step": 76460 }, { "epoch": 6.42, "learning_rate": 7.156533422908969e-06, "loss": 0.0007, "step": 76470 }, { "epoch": 6.42, "learning_rate": 7.15485387974471e-06, "loss": 0.0282, "step": 76480 }, { "epoch": 6.42, "learning_rate": 7.153174336580451e-06, "loss": 0.0987, "step": 76490 }, { "epoch": 6.42, "learning_rate": 7.1514947934161915e-06, "loss": 0.0008, "step": 76500 }, { "epoch": 6.43, "learning_rate": 7.149815250251933e-06, "loss": 0.0004, "step": 76510 }, { "epoch": 6.43, "learning_rate": 7.148135707087672e-06, "loss": 0.0167, "step": 76520 }, { "epoch": 6.43, "learning_rate": 7.146456163923413e-06, "loss": 0.0564, "step": 76530 }, { "epoch": 6.43, "learning_rate": 7.144776620759154e-06, "loss": 0.0011, "step": 76540 }, { "epoch": 6.43, "learning_rate": 7.143097077594895e-06, "loss": 0.1176, "step": 76550 }, { "epoch": 6.43, "learning_rate": 7.141417534430636e-06, "loss": 0.0327, "step": 76560 }, { "epoch": 6.43, "learning_rate": 7.1397379912663765e-06, "loss": 0.0425, "step": 76570 }, { "epoch": 6.43, "learning_rate": 7.138058448102116e-06, "loss": 0.0001, "step": 76580 }, { "epoch": 6.43, "learning_rate": 7.1363789049378574e-06, "loss": 0.0394, "step": 76590 }, { "epoch": 6.43, "learning_rate": 7.134699361773598e-06, "loss": 0.0597, "step": 76600 }, { "epoch": 6.43, "learning_rate": 7.133019818609339e-06, "loss": 0.044, "step": 76610 }, { "epoch": 6.43, "learning_rate": 7.13134027544508e-06, "loss": 0.036, "step": 76620 }, { "epoch": 6.44, "learning_rate": 7.12966073228082e-06, "loss": 0.022, "step": 76630 }, { "epoch": 6.44, "learning_rate": 7.127981189116561e-06, "loss": 0.0619, "step": 76640 }, { "epoch": 6.44, "learning_rate": 7.126301645952301e-06, "loss": 0.0591, "step": 76650 }, { "epoch": 6.44, "learning_rate": 7.1246221027880425e-06, "loss": 0.0763, "step": 76660 }, { "epoch": 6.44, "learning_rate": 7.122942559623783e-06, "loss": 0.0156, "step": 76670 }, { "epoch": 6.44, "learning_rate": 7.121263016459523e-06, "loss": 0.0934, "step": 76680 }, { "epoch": 6.44, "learning_rate": 7.119583473295264e-06, "loss": 0.0368, "step": 76690 }, { "epoch": 6.44, "learning_rate": 7.117903930131005e-06, "loss": 0.0342, "step": 76700 }, { "epoch": 6.44, "learning_rate": 7.116224386966746e-06, "loss": 0.001, "step": 76710 }, { "epoch": 6.44, "learning_rate": 7.114544843802487e-06, "loss": 0.001, "step": 76720 }, { "epoch": 6.44, "learning_rate": 7.112865300638227e-06, "loss": 0.017, "step": 76730 }, { "epoch": 6.44, "learning_rate": 7.111185757473967e-06, "loss": 0.0209, "step": 76740 }, { "epoch": 6.45, "learning_rate": 7.1095062143097084e-06, "loss": 0.0044, "step": 76750 }, { "epoch": 6.45, "learning_rate": 7.107826671145449e-06, "loss": 0.0203, "step": 76760 }, { "epoch": 6.45, "learning_rate": 7.10614712798119e-06, "loss": 0.0853, "step": 76770 }, { "epoch": 6.45, "learning_rate": 7.104467584816931e-06, "loss": 0.0018, "step": 76780 }, { "epoch": 6.45, "learning_rate": 7.10278804165267e-06, "loss": 0.0583, "step": 76790 }, { "epoch": 6.45, "learning_rate": 7.101108498488412e-06, "loss": 0.0898, "step": 76800 }, { "epoch": 6.45, "learning_rate": 7.099428955324152e-06, "loss": 0.0021, "step": 76810 }, { "epoch": 6.45, "learning_rate": 7.0977494121598935e-06, "loss": 0.0514, "step": 76820 }, { "epoch": 6.45, "learning_rate": 7.096069868995634e-06, "loss": 0.0004, "step": 76830 }, { "epoch": 6.45, "learning_rate": 7.094390325831374e-06, "loss": 0.0481, "step": 76840 }, { "epoch": 6.45, "learning_rate": 7.092710782667115e-06, "loss": 0.0148, "step": 76850 }, { "epoch": 6.45, "learning_rate": 7.091031239502855e-06, "loss": 0.004, "step": 76860 }, { "epoch": 6.46, "learning_rate": 7.089351696338597e-06, "loss": 0.0773, "step": 76870 }, { "epoch": 6.46, "learning_rate": 7.087672153174337e-06, "loss": 0.0236, "step": 76880 }, { "epoch": 6.46, "learning_rate": 7.085992610010078e-06, "loss": 0.0284, "step": 76890 }, { "epoch": 6.46, "learning_rate": 7.084313066845818e-06, "loss": 0.0545, "step": 76900 }, { "epoch": 6.46, "learning_rate": 7.0826335236815595e-06, "loss": 0.0003, "step": 76910 }, { "epoch": 6.46, "learning_rate": 7.0809539805173e-06, "loss": 0.0274, "step": 76920 }, { "epoch": 6.46, "learning_rate": 7.079274437353041e-06, "loss": 0.0924, "step": 76930 }, { "epoch": 6.46, "learning_rate": 7.077594894188782e-06, "loss": 0.0792, "step": 76940 }, { "epoch": 6.46, "learning_rate": 7.075915351024521e-06, "loss": 0.0625, "step": 76950 }, { "epoch": 6.46, "learning_rate": 7.074235807860263e-06, "loss": 0.0425, "step": 76960 }, { "epoch": 6.46, "learning_rate": 7.072556264696003e-06, "loss": 0.0887, "step": 76970 }, { "epoch": 6.46, "learning_rate": 7.0708767215317445e-06, "loss": 0.0015, "step": 76980 }, { "epoch": 6.47, "learning_rate": 7.069197178367485e-06, "loss": 0.0886, "step": 76990 }, { "epoch": 6.47, "learning_rate": 7.067517635203225e-06, "loss": 0.0923, "step": 77000 }, { "epoch": 6.47, "learning_rate": 7.065838092038966e-06, "loss": 0.103, "step": 77010 }, { "epoch": 6.47, "learning_rate": 7.064158548874706e-06, "loss": 0.0011, "step": 77020 }, { "epoch": 6.47, "learning_rate": 7.062479005710448e-06, "loss": 0.03, "step": 77030 }, { "epoch": 6.47, "learning_rate": 7.060799462546188e-06, "loss": 0.0617, "step": 77040 }, { "epoch": 6.47, "learning_rate": 7.059119919381928e-06, "loss": 0.0124, "step": 77050 }, { "epoch": 6.47, "learning_rate": 7.057440376217669e-06, "loss": 0.0062, "step": 77060 }, { "epoch": 6.47, "learning_rate": 7.05576083305341e-06, "loss": 0.0111, "step": 77070 }, { "epoch": 6.47, "learning_rate": 7.054081289889151e-06, "loss": 0.0008, "step": 77080 }, { "epoch": 6.47, "learning_rate": 7.0524017467248914e-06, "loss": 0.074, "step": 77090 }, { "epoch": 6.47, "learning_rate": 7.050722203560633e-06, "loss": 0.0304, "step": 77100 }, { "epoch": 6.48, "learning_rate": 7.049042660396372e-06, "loss": 0.1791, "step": 77110 }, { "epoch": 6.48, "learning_rate": 7.047363117232114e-06, "loss": 0.002, "step": 77120 }, { "epoch": 6.48, "learning_rate": 7.045683574067854e-06, "loss": 0.0562, "step": 77130 }, { "epoch": 6.48, "learning_rate": 7.044004030903595e-06, "loss": 0.0418, "step": 77140 }, { "epoch": 6.48, "learning_rate": 7.042324487739336e-06, "loss": 0.0383, "step": 77150 }, { "epoch": 6.48, "learning_rate": 7.040644944575076e-06, "loss": 0.0619, "step": 77160 }, { "epoch": 6.48, "learning_rate": 7.038965401410817e-06, "loss": 0.0004, "step": 77170 }, { "epoch": 6.48, "learning_rate": 7.037285858246557e-06, "loss": 0.0007, "step": 77180 }, { "epoch": 6.48, "learning_rate": 7.035606315082299e-06, "loss": 0.0017, "step": 77190 }, { "epoch": 6.48, "learning_rate": 7.033926771918039e-06, "loss": 0.0185, "step": 77200 }, { "epoch": 6.48, "learning_rate": 7.032247228753779e-06, "loss": 0.0312, "step": 77210 }, { "epoch": 6.48, "learning_rate": 7.03056768558952e-06, "loss": 0.0006, "step": 77220 }, { "epoch": 6.49, "learning_rate": 7.028888142425261e-06, "loss": 0.022, "step": 77230 }, { "epoch": 6.49, "learning_rate": 7.027208599261002e-06, "loss": 0.0239, "step": 77240 }, { "epoch": 6.49, "learning_rate": 7.0255290560967424e-06, "loss": 0.0027, "step": 77250 }, { "epoch": 6.49, "learning_rate": 7.023849512932482e-06, "loss": 0.0022, "step": 77260 }, { "epoch": 6.49, "learning_rate": 7.022169969768223e-06, "loss": 0.0561, "step": 77270 }, { "epoch": 6.49, "learning_rate": 7.020490426603964e-06, "loss": 0.011, "step": 77280 }, { "epoch": 6.49, "learning_rate": 7.018810883439705e-06, "loss": 0.0386, "step": 77290 }, { "epoch": 6.49, "learning_rate": 7.017131340275446e-06, "loss": 0.0276, "step": 77300 }, { "epoch": 6.49, "learning_rate": 7.015451797111187e-06, "loss": 0.0225, "step": 77310 }, { "epoch": 6.49, "learning_rate": 7.013772253946927e-06, "loss": 0.0476, "step": 77320 }, { "epoch": 6.49, "learning_rate": 7.012092710782667e-06, "loss": 0.0218, "step": 77330 }, { "epoch": 6.49, "learning_rate": 7.0104131676184084e-06, "loss": 0.0163, "step": 77340 }, { "epoch": 6.5, "learning_rate": 7.008733624454149e-06, "loss": 0.0549, "step": 77350 }, { "epoch": 6.5, "learning_rate": 7.00705408128989e-06, "loss": 0.0229, "step": 77360 }, { "epoch": 6.5, "learning_rate": 7.00537453812563e-06, "loss": 0.0221, "step": 77370 }, { "epoch": 6.5, "learning_rate": 7.003694994961371e-06, "loss": 0.0216, "step": 77380 }, { "epoch": 6.5, "learning_rate": 7.002015451797112e-06, "loss": 0.0657, "step": 77390 }, { "epoch": 6.5, "learning_rate": 7.000335908632853e-06, "loss": 0.0018, "step": 77400 }, { "epoch": 6.5, "learning_rate": 6.9986563654685935e-06, "loss": 0.0805, "step": 77410 }, { "epoch": 6.5, "learning_rate": 6.996976822304333e-06, "loss": 0.001, "step": 77420 }, { "epoch": 6.5, "learning_rate": 6.995297279140074e-06, "loss": 0.0721, "step": 77430 }, { "epoch": 6.5, "learning_rate": 6.993617735975815e-06, "loss": 0.051, "step": 77440 }, { "epoch": 6.5, "learning_rate": 6.991938192811556e-06, "loss": 0.0009, "step": 77450 }, { "epoch": 6.5, "learning_rate": 6.990258649647297e-06, "loss": 0.046, "step": 77460 }, { "epoch": 6.51, "learning_rate": 6.988579106483038e-06, "loss": 0.0605, "step": 77470 }, { "epoch": 6.51, "learning_rate": 6.986899563318778e-06, "loss": 0.0618, "step": 77480 }, { "epoch": 6.51, "learning_rate": 6.985220020154518e-06, "loss": 0.0425, "step": 77490 }, { "epoch": 6.51, "learning_rate": 6.9835404769902594e-06, "loss": 0.0007, "step": 77500 }, { "epoch": 6.51, "learning_rate": 6.981860933826e-06, "loss": 0.0011, "step": 77510 }, { "epoch": 6.51, "learning_rate": 6.980181390661741e-06, "loss": 0.0907, "step": 77520 }, { "epoch": 6.51, "learning_rate": 6.978501847497481e-06, "loss": 0.0055, "step": 77530 }, { "epoch": 6.51, "learning_rate": 6.976822304333221e-06, "loss": 0.0395, "step": 77540 }, { "epoch": 6.51, "learning_rate": 6.975142761168963e-06, "loss": 0.028, "step": 77550 }, { "epoch": 6.51, "learning_rate": 6.973463218004703e-06, "loss": 0.0965, "step": 77560 }, { "epoch": 6.51, "learning_rate": 6.9717836748404445e-06, "loss": 0.001, "step": 77570 }, { "epoch": 6.51, "learning_rate": 6.970104131676184e-06, "loss": 0.0615, "step": 77580 }, { "epoch": 6.52, "learning_rate": 6.9684245885119254e-06, "loss": 0.134, "step": 77590 }, { "epoch": 6.52, "learning_rate": 6.966745045347666e-06, "loss": 0.038, "step": 77600 }, { "epoch": 6.52, "learning_rate": 6.965065502183407e-06, "loss": 0.0384, "step": 77610 }, { "epoch": 6.52, "learning_rate": 6.963385959019148e-06, "loss": 0.0522, "step": 77620 }, { "epoch": 6.52, "learning_rate": 6.961706415854887e-06, "loss": 0.0006, "step": 77630 }, { "epoch": 6.52, "learning_rate": 6.960026872690629e-06, "loss": 0.0251, "step": 77640 }, { "epoch": 6.52, "learning_rate": 6.958347329526369e-06, "loss": 0.058, "step": 77650 }, { "epoch": 6.52, "learning_rate": 6.9566677863621105e-06, "loss": 0.0159, "step": 77660 }, { "epoch": 6.52, "learning_rate": 6.954988243197851e-06, "loss": 0.0541, "step": 77670 }, { "epoch": 6.52, "learning_rate": 6.953308700033592e-06, "loss": 0.0524, "step": 77680 }, { "epoch": 6.52, "learning_rate": 6.951629156869332e-06, "loss": 0.0272, "step": 77690 }, { "epoch": 6.53, "learning_rate": 6.949949613705072e-06, "loss": 0.0867, "step": 77700 }, { "epoch": 6.53, "learning_rate": 6.948270070540814e-06, "loss": 0.0376, "step": 77710 }, { "epoch": 6.53, "learning_rate": 6.946590527376554e-06, "loss": 0.0385, "step": 77720 }, { "epoch": 6.53, "learning_rate": 6.9449109842122955e-06, "loss": 0.0355, "step": 77730 }, { "epoch": 6.53, "learning_rate": 6.943231441048035e-06, "loss": 0.064, "step": 77740 }, { "epoch": 6.53, "learning_rate": 6.941551897883776e-06, "loss": 0.0006, "step": 77750 }, { "epoch": 6.53, "learning_rate": 6.939872354719517e-06, "loss": 0.0017, "step": 77760 }, { "epoch": 6.53, "learning_rate": 6.938192811555257e-06, "loss": 0.0545, "step": 77770 }, { "epoch": 6.53, "learning_rate": 6.936513268390999e-06, "loss": 0.0878, "step": 77780 }, { "epoch": 6.53, "learning_rate": 6.934833725226738e-06, "loss": 0.0003, "step": 77790 }, { "epoch": 6.53, "learning_rate": 6.93315418206248e-06, "loss": 0.0735, "step": 77800 }, { "epoch": 6.53, "learning_rate": 6.93147463889822e-06, "loss": 0.0993, "step": 77810 }, { "epoch": 6.54, "learning_rate": 6.929795095733961e-06, "loss": 0.003, "step": 77820 }, { "epoch": 6.54, "learning_rate": 6.928115552569702e-06, "loss": 0.0518, "step": 77830 }, { "epoch": 6.54, "learning_rate": 6.9264360094054424e-06, "loss": 0.0776, "step": 77840 }, { "epoch": 6.54, "learning_rate": 6.924756466241183e-06, "loss": 0.0888, "step": 77850 }, { "epoch": 6.54, "learning_rate": 6.923076923076923e-06, "loss": 0.0642, "step": 77860 }, { "epoch": 6.54, "learning_rate": 6.921397379912665e-06, "loss": 0.0005, "step": 77870 }, { "epoch": 6.54, "learning_rate": 6.919717836748405e-06, "loss": 0.001, "step": 77880 }, { "epoch": 6.54, "learning_rate": 6.9180382935841465e-06, "loss": 0.0026, "step": 77890 }, { "epoch": 6.54, "learning_rate": 6.916358750419886e-06, "loss": 0.0463, "step": 77900 }, { "epoch": 6.54, "learning_rate": 6.914679207255627e-06, "loss": 0.0575, "step": 77910 }, { "epoch": 6.54, "learning_rate": 6.912999664091368e-06, "loss": 0.0606, "step": 77920 }, { "epoch": 6.54, "learning_rate": 6.911320120927108e-06, "loss": 0.0221, "step": 77930 }, { "epoch": 6.55, "learning_rate": 6.90964057776285e-06, "loss": 0.076, "step": 77940 }, { "epoch": 6.55, "learning_rate": 6.907961034598589e-06, "loss": 0.0012, "step": 77950 }, { "epoch": 6.55, "learning_rate": 6.90628149143433e-06, "loss": 0.0768, "step": 77960 }, { "epoch": 6.55, "learning_rate": 6.904601948270071e-06, "loss": 0.0018, "step": 77970 }, { "epoch": 6.55, "learning_rate": 6.902922405105812e-06, "loss": 0.0551, "step": 77980 }, { "epoch": 6.55, "learning_rate": 6.901242861941553e-06, "loss": 0.0283, "step": 77990 }, { "epoch": 6.55, "learning_rate": 6.8995633187772934e-06, "loss": 0.0441, "step": 78000 }, { "epoch": 6.55, "learning_rate": 6.897883775613033e-06, "loss": 0.036, "step": 78010 }, { "epoch": 6.55, "learning_rate": 6.896204232448774e-06, "loss": 0.0895, "step": 78020 }, { "epoch": 6.55, "learning_rate": 6.894524689284515e-06, "loss": 0.0576, "step": 78030 }, { "epoch": 6.55, "learning_rate": 6.892845146120256e-06, "loss": 0.0143, "step": 78040 }, { "epoch": 6.55, "learning_rate": 6.891165602955997e-06, "loss": 0.0278, "step": 78050 }, { "epoch": 6.56, "learning_rate": 6.889486059791737e-06, "loss": 0.0009, "step": 78060 }, { "epoch": 6.56, "learning_rate": 6.887806516627478e-06, "loss": 0.0487, "step": 78070 }, { "epoch": 6.56, "learning_rate": 6.886126973463219e-06, "loss": 0.0134, "step": 78080 }, { "epoch": 6.56, "learning_rate": 6.8844474302989594e-06, "loss": 0.0253, "step": 78090 }, { "epoch": 6.56, "learning_rate": 6.882767887134701e-06, "loss": 0.0698, "step": 78100 }, { "epoch": 6.56, "learning_rate": 6.88108834397044e-06, "loss": 0.0624, "step": 78110 }, { "epoch": 6.56, "learning_rate": 6.879408800806181e-06, "loss": 0.0542, "step": 78120 }, { "epoch": 6.56, "learning_rate": 6.877729257641922e-06, "loss": 0.0503, "step": 78130 }, { "epoch": 6.56, "learning_rate": 6.876049714477663e-06, "loss": 0.0031, "step": 78140 }, { "epoch": 6.56, "learning_rate": 6.874370171313404e-06, "loss": 0.0237, "step": 78150 }, { "epoch": 6.56, "learning_rate": 6.872690628149144e-06, "loss": 0.0267, "step": 78160 }, { "epoch": 6.56, "learning_rate": 6.871011084984884e-06, "loss": 0.0656, "step": 78170 }, { "epoch": 6.57, "learning_rate": 6.869331541820625e-06, "loss": 0.1065, "step": 78180 }, { "epoch": 6.57, "learning_rate": 6.867651998656366e-06, "loss": 0.0362, "step": 78190 }, { "epoch": 6.57, "learning_rate": 6.865972455492107e-06, "loss": 0.0243, "step": 78200 }, { "epoch": 6.57, "learning_rate": 6.864292912327848e-06, "loss": 0.0178, "step": 78210 }, { "epoch": 6.57, "learning_rate": 6.862613369163587e-06, "loss": 0.0441, "step": 78220 }, { "epoch": 6.57, "learning_rate": 6.860933825999329e-06, "loss": 0.029, "step": 78230 }, { "epoch": 6.57, "learning_rate": 6.859254282835069e-06, "loss": 0.1056, "step": 78240 }, { "epoch": 6.57, "learning_rate": 6.8575747396708104e-06, "loss": 0.0009, "step": 78250 }, { "epoch": 6.57, "learning_rate": 6.855895196506551e-06, "loss": 0.1283, "step": 78260 }, { "epoch": 6.57, "learning_rate": 6.854215653342291e-06, "loss": 0.0242, "step": 78270 }, { "epoch": 6.57, "learning_rate": 6.852536110178032e-06, "loss": 0.0693, "step": 78280 }, { "epoch": 6.57, "learning_rate": 6.850856567013773e-06, "loss": 0.0013, "step": 78290 }, { "epoch": 6.58, "learning_rate": 6.849177023849514e-06, "loss": 0.0791, "step": 78300 }, { "epoch": 6.58, "learning_rate": 6.847497480685254e-06, "loss": 0.0386, "step": 78310 }, { "epoch": 6.58, "learning_rate": 6.845817937520995e-06, "loss": 0.0471, "step": 78320 }, { "epoch": 6.58, "learning_rate": 6.844138394356735e-06, "loss": 0.1315, "step": 78330 }, { "epoch": 6.58, "learning_rate": 6.8424588511924764e-06, "loss": 0.0451, "step": 78340 }, { "epoch": 6.58, "learning_rate": 6.840779308028217e-06, "loss": 0.0525, "step": 78350 }, { "epoch": 6.58, "learning_rate": 6.839099764863958e-06, "loss": 0.0405, "step": 78360 }, { "epoch": 6.58, "learning_rate": 6.837420221699699e-06, "loss": 0.0431, "step": 78370 }, { "epoch": 6.58, "learning_rate": 6.835740678535438e-06, "loss": 0.047, "step": 78380 }, { "epoch": 6.58, "learning_rate": 6.83406113537118e-06, "loss": 0.0011, "step": 78390 }, { "epoch": 6.58, "learning_rate": 6.83238159220692e-06, "loss": 0.0034, "step": 78400 }, { "epoch": 6.58, "learning_rate": 6.8307020490426615e-06, "loss": 0.0484, "step": 78410 }, { "epoch": 6.59, "learning_rate": 6.829022505878402e-06, "loss": 0.037, "step": 78420 }, { "epoch": 6.59, "learning_rate": 6.8273429627141416e-06, "loss": 0.0403, "step": 78430 }, { "epoch": 6.59, "learning_rate": 6.825663419549883e-06, "loss": 0.02, "step": 78440 }, { "epoch": 6.59, "learning_rate": 6.823983876385623e-06, "loss": 0.0437, "step": 78450 }, { "epoch": 6.59, "learning_rate": 6.822304333221365e-06, "loss": 0.0006, "step": 78460 }, { "epoch": 6.59, "learning_rate": 6.820624790057105e-06, "loss": 0.0005, "step": 78470 }, { "epoch": 6.59, "learning_rate": 6.818945246892845e-06, "loss": 0.048, "step": 78480 }, { "epoch": 6.59, "learning_rate": 6.817265703728586e-06, "loss": 0.0894, "step": 78490 }, { "epoch": 6.59, "learning_rate": 6.815586160564327e-06, "loss": 0.0511, "step": 78500 }, { "epoch": 6.59, "learning_rate": 6.813906617400068e-06, "loss": 0.0549, "step": 78510 }, { "epoch": 6.59, "learning_rate": 6.812227074235808e-06, "loss": 0.1279, "step": 78520 }, { "epoch": 6.59, "learning_rate": 6.81054753107155e-06, "loss": 0.0263, "step": 78530 }, { "epoch": 6.6, "learning_rate": 6.808867987907289e-06, "loss": 0.0587, "step": 78540 }, { "epoch": 6.6, "learning_rate": 6.807188444743031e-06, "loss": 0.0666, "step": 78550 }, { "epoch": 6.6, "learning_rate": 6.805508901578771e-06, "loss": 0.001, "step": 78560 }, { "epoch": 6.6, "learning_rate": 6.8038293584145125e-06, "loss": 0.037, "step": 78570 }, { "epoch": 6.6, "learning_rate": 6.802149815250253e-06, "loss": 0.0231, "step": 78580 }, { "epoch": 6.6, "learning_rate": 6.800470272085993e-06, "loss": 0.0555, "step": 78590 }, { "epoch": 6.6, "learning_rate": 6.798790728921734e-06, "loss": 0.0766, "step": 78600 }, { "epoch": 6.6, "learning_rate": 6.797111185757474e-06, "loss": 0.0501, "step": 78610 }, { "epoch": 6.6, "learning_rate": 6.795431642593216e-06, "loss": 0.0074, "step": 78620 }, { "epoch": 6.6, "learning_rate": 6.793752099428956e-06, "loss": 0.0266, "step": 78630 }, { "epoch": 6.6, "learning_rate": 6.792072556264696e-06, "loss": 0.0242, "step": 78640 }, { "epoch": 6.6, "learning_rate": 6.790393013100437e-06, "loss": 0.0013, "step": 78650 }, { "epoch": 6.61, "learning_rate": 6.788713469936178e-06, "loss": 0.0214, "step": 78660 }, { "epoch": 6.61, "learning_rate": 6.787033926771919e-06, "loss": 0.0252, "step": 78670 }, { "epoch": 6.61, "learning_rate": 6.785354383607659e-06, "loss": 0.0322, "step": 78680 }, { "epoch": 6.61, "learning_rate": 6.783674840443399e-06, "loss": 0.0684, "step": 78690 }, { "epoch": 6.61, "learning_rate": 6.78199529727914e-06, "loss": 0.036, "step": 78700 }, { "epoch": 6.61, "learning_rate": 6.780315754114881e-06, "loss": 0.0005, "step": 78710 }, { "epoch": 6.61, "learning_rate": 6.778636210950622e-06, "loss": 0.0226, "step": 78720 }, { "epoch": 6.61, "learning_rate": 6.776956667786363e-06, "loss": 0.0371, "step": 78730 }, { "epoch": 6.61, "learning_rate": 6.775277124622104e-06, "loss": 0.0003, "step": 78740 }, { "epoch": 6.61, "learning_rate": 6.773597581457844e-06, "loss": 0.0528, "step": 78750 }, { "epoch": 6.61, "learning_rate": 6.771918038293585e-06, "loss": 0.1053, "step": 78760 }, { "epoch": 6.61, "learning_rate": 6.770238495129325e-06, "loss": 0.058, "step": 78770 }, { "epoch": 6.62, "learning_rate": 6.768558951965066e-06, "loss": 0.0013, "step": 78780 }, { "epoch": 6.62, "learning_rate": 6.766879408800807e-06, "loss": 0.0153, "step": 78790 }, { "epoch": 6.62, "learning_rate": 6.765199865636547e-06, "loss": 0.0155, "step": 78800 }, { "epoch": 6.62, "learning_rate": 6.763520322472288e-06, "loss": 0.0009, "step": 78810 }, { "epoch": 6.62, "learning_rate": 6.761840779308029e-06, "loss": 0.0494, "step": 78820 }, { "epoch": 6.62, "learning_rate": 6.76016123614377e-06, "loss": 0.0732, "step": 78830 }, { "epoch": 6.62, "learning_rate": 6.7584816929795104e-06, "loss": 0.0968, "step": 78840 }, { "epoch": 6.62, "learning_rate": 6.75680214981525e-06, "loss": 0.1812, "step": 78850 }, { "epoch": 6.62, "learning_rate": 6.755122606650991e-06, "loss": 0.0656, "step": 78860 }, { "epoch": 6.62, "learning_rate": 6.753443063486732e-06, "loss": 0.0494, "step": 78870 }, { "epoch": 6.62, "learning_rate": 6.751763520322473e-06, "loss": 0.0016, "step": 78880 }, { "epoch": 6.62, "learning_rate": 6.750083977158214e-06, "loss": 0.0017, "step": 78890 }, { "epoch": 6.63, "learning_rate": 6.748404433993955e-06, "loss": 0.0544, "step": 78900 }, { "epoch": 6.63, "learning_rate": 6.746724890829695e-06, "loss": 0.1057, "step": 78910 }, { "epoch": 6.63, "learning_rate": 6.745045347665435e-06, "loss": 0.0681, "step": 78920 }, { "epoch": 6.63, "learning_rate": 6.743365804501176e-06, "loss": 0.0132, "step": 78930 }, { "epoch": 6.63, "learning_rate": 6.741686261336917e-06, "loss": 0.0023, "step": 78940 }, { "epoch": 6.63, "learning_rate": 6.740006718172658e-06, "loss": 0.1091, "step": 78950 }, { "epoch": 6.63, "learning_rate": 6.738327175008398e-06, "loss": 0.0279, "step": 78960 }, { "epoch": 6.63, "learning_rate": 6.736647631844138e-06, "loss": 0.0698, "step": 78970 }, { "epoch": 6.63, "learning_rate": 6.73496808867988e-06, "loss": 0.0833, "step": 78980 }, { "epoch": 6.63, "learning_rate": 6.73328854551562e-06, "loss": 0.0348, "step": 78990 }, { "epoch": 6.63, "learning_rate": 6.7316090023513614e-06, "loss": 0.0109, "step": 79000 }, { "epoch": 6.64, "learning_rate": 6.729929459187101e-06, "loss": 0.0343, "step": 79010 }, { "epoch": 6.64, "learning_rate": 6.728249916022842e-06, "loss": 0.0134, "step": 79020 }, { "epoch": 6.64, "learning_rate": 6.726570372858583e-06, "loss": 0.0303, "step": 79030 }, { "epoch": 6.64, "learning_rate": 6.724890829694324e-06, "loss": 0.0006, "step": 79040 }, { "epoch": 6.64, "learning_rate": 6.723211286530065e-06, "loss": 0.0269, "step": 79050 }, { "epoch": 6.64, "learning_rate": 6.721531743365806e-06, "loss": 0.0049, "step": 79060 }, { "epoch": 6.64, "learning_rate": 6.719852200201546e-06, "loss": 0.0368, "step": 79070 }, { "epoch": 6.64, "learning_rate": 6.718172657037286e-06, "loss": 0.0504, "step": 79080 }, { "epoch": 6.64, "learning_rate": 6.7164931138730274e-06, "loss": 0.063, "step": 79090 }, { "epoch": 6.64, "learning_rate": 6.714813570708768e-06, "loss": 0.0272, "step": 79100 }, { "epoch": 6.64, "learning_rate": 6.713134027544509e-06, "loss": 0.0006, "step": 79110 }, { "epoch": 6.64, "learning_rate": 6.711454484380249e-06, "loss": 0.03, "step": 79120 }, { "epoch": 6.65, "learning_rate": 6.709774941215989e-06, "loss": 0.067, "step": 79130 }, { "epoch": 6.65, "learning_rate": 6.708095398051731e-06, "loss": 0.0563, "step": 79140 }, { "epoch": 6.65, "learning_rate": 6.706415854887471e-06, "loss": 0.1039, "step": 79150 }, { "epoch": 6.65, "learning_rate": 6.7047363117232125e-06, "loss": 0.0654, "step": 79160 }, { "epoch": 6.65, "learning_rate": 6.703056768558952e-06, "loss": 0.0537, "step": 79170 }, { "epoch": 6.65, "learning_rate": 6.7013772253946926e-06, "loss": 0.055, "step": 79180 }, { "epoch": 6.65, "learning_rate": 6.699697682230434e-06, "loss": 0.086, "step": 79190 }, { "epoch": 6.65, "learning_rate": 6.698018139066174e-06, "loss": 0.0259, "step": 79200 }, { "epoch": 6.65, "learning_rate": 6.696338595901916e-06, "loss": 0.0514, "step": 79210 }, { "epoch": 6.65, "learning_rate": 6.694659052737655e-06, "loss": 0.0013, "step": 79220 }, { "epoch": 6.65, "learning_rate": 6.692979509573397e-06, "loss": 0.0658, "step": 79230 }, { "epoch": 6.65, "learning_rate": 6.691299966409137e-06, "loss": 0.046, "step": 79240 }, { "epoch": 6.66, "learning_rate": 6.6896204232448784e-06, "loss": 0.0007, "step": 79250 }, { "epoch": 6.66, "learning_rate": 6.687940880080619e-06, "loss": 0.0233, "step": 79260 }, { "epoch": 6.66, "learning_rate": 6.686261336916359e-06, "loss": 0.0372, "step": 79270 }, { "epoch": 6.66, "learning_rate": 6.6845817937521e-06, "loss": 0.0225, "step": 79280 }, { "epoch": 6.66, "learning_rate": 6.68290225058784e-06, "loss": 0.0403, "step": 79290 }, { "epoch": 6.66, "learning_rate": 6.681222707423582e-06, "loss": 0.0481, "step": 79300 }, { "epoch": 6.66, "learning_rate": 6.679543164259322e-06, "loss": 0.0285, "step": 79310 }, { "epoch": 6.66, "learning_rate": 6.6778636210950635e-06, "loss": 0.0905, "step": 79320 }, { "epoch": 6.66, "learning_rate": 6.676184077930803e-06, "loss": 0.0485, "step": 79330 }, { "epoch": 6.66, "learning_rate": 6.6745045347665436e-06, "loss": 0.048, "step": 79340 }, { "epoch": 6.66, "learning_rate": 6.672824991602285e-06, "loss": 0.037, "step": 79350 }, { "epoch": 6.66, "learning_rate": 6.671145448438025e-06, "loss": 0.0218, "step": 79360 }, { "epoch": 6.67, "learning_rate": 6.669465905273767e-06, "loss": 0.0371, "step": 79370 }, { "epoch": 6.67, "learning_rate": 6.667786362109506e-06, "loss": 0.1133, "step": 79380 }, { "epoch": 6.67, "learning_rate": 6.666106818945247e-06, "loss": 0.089, "step": 79390 }, { "epoch": 6.67, "learning_rate": 6.664427275780988e-06, "loss": 0.0138, "step": 79400 }, { "epoch": 6.67, "learning_rate": 6.662747732616729e-06, "loss": 0.0458, "step": 79410 }, { "epoch": 6.67, "learning_rate": 6.66106818945247e-06, "loss": 0.0144, "step": 79420 }, { "epoch": 6.67, "learning_rate": 6.65938864628821e-06, "loss": 0.0093, "step": 79430 }, { "epoch": 6.67, "learning_rate": 6.657709103123951e-06, "loss": 0.1211, "step": 79440 }, { "epoch": 6.67, "learning_rate": 6.656029559959691e-06, "loss": 0.0282, "step": 79450 }, { "epoch": 6.67, "learning_rate": 6.654350016795432e-06, "loss": 0.0493, "step": 79460 }, { "epoch": 6.67, "learning_rate": 6.652670473631173e-06, "loss": 0.0012, "step": 79470 }, { "epoch": 6.67, "learning_rate": 6.650990930466914e-06, "loss": 0.0398, "step": 79480 }, { "epoch": 6.68, "learning_rate": 6.649311387302654e-06, "loss": 0.025, "step": 79490 }, { "epoch": 6.68, "learning_rate": 6.647631844138395e-06, "loss": 0.0115, "step": 79500 }, { "epoch": 6.68, "learning_rate": 6.645952300974136e-06, "loss": 0.0756, "step": 79510 }, { "epoch": 6.68, "learning_rate": 6.644272757809876e-06, "loss": 0.061, "step": 79520 }, { "epoch": 6.68, "learning_rate": 6.642593214645618e-06, "loss": 0.047, "step": 79530 }, { "epoch": 6.68, "learning_rate": 6.640913671481357e-06, "loss": 0.0388, "step": 79540 }, { "epoch": 6.68, "learning_rate": 6.639234128317098e-06, "loss": 0.0023, "step": 79550 }, { "epoch": 6.68, "learning_rate": 6.637554585152839e-06, "loss": 0.0003, "step": 79560 }, { "epoch": 6.68, "learning_rate": 6.63587504198858e-06, "loss": 0.0322, "step": 79570 }, { "epoch": 6.68, "learning_rate": 6.634195498824321e-06, "loss": 0.0007, "step": 79580 }, { "epoch": 6.68, "learning_rate": 6.6325159556600606e-06, "loss": 0.0055, "step": 79590 }, { "epoch": 6.68, "learning_rate": 6.630836412495801e-06, "loss": 0.0434, "step": 79600 }, { "epoch": 6.69, "learning_rate": 6.629156869331542e-06, "loss": 0.0614, "step": 79610 }, { "epoch": 6.69, "learning_rate": 6.627477326167283e-06, "loss": 0.0004, "step": 79620 }, { "epoch": 6.69, "learning_rate": 6.625797783003024e-06, "loss": 0.0005, "step": 79630 }, { "epoch": 6.69, "learning_rate": 6.624118239838765e-06, "loss": 0.0568, "step": 79640 }, { "epoch": 6.69, "learning_rate": 6.622438696674504e-06, "loss": 0.0007, "step": 79650 }, { "epoch": 6.69, "learning_rate": 6.620759153510246e-06, "loss": 0.0912, "step": 79660 }, { "epoch": 6.69, "learning_rate": 6.619079610345986e-06, "loss": 0.0227, "step": 79670 }, { "epoch": 6.69, "learning_rate": 6.617400067181727e-06, "loss": 0.0683, "step": 79680 }, { "epoch": 6.69, "learning_rate": 6.615720524017468e-06, "loss": 0.0004, "step": 79690 }, { "epoch": 6.69, "learning_rate": 6.614040980853208e-06, "loss": 0.0524, "step": 79700 }, { "epoch": 6.69, "learning_rate": 6.612361437688949e-06, "loss": 0.0509, "step": 79710 }, { "epoch": 6.69, "learning_rate": 6.61068189452469e-06, "loss": 0.0641, "step": 79720 }, { "epoch": 6.7, "learning_rate": 6.609002351360431e-06, "loss": 0.0057, "step": 79730 }, { "epoch": 6.7, "learning_rate": 6.607322808196172e-06, "loss": 0.1433, "step": 79740 }, { "epoch": 6.7, "learning_rate": 6.605643265031912e-06, "loss": 0.0658, "step": 79750 }, { "epoch": 6.7, "learning_rate": 6.603963721867652e-06, "loss": 0.0472, "step": 79760 }, { "epoch": 6.7, "learning_rate": 6.602284178703393e-06, "loss": 0.0232, "step": 79770 }, { "epoch": 6.7, "learning_rate": 6.600604635539134e-06, "loss": 0.0319, "step": 79780 }, { "epoch": 6.7, "learning_rate": 6.598925092374875e-06, "loss": 0.0345, "step": 79790 }, { "epoch": 6.7, "learning_rate": 6.597245549210616e-06, "loss": 0.1018, "step": 79800 }, { "epoch": 6.7, "learning_rate": 6.595566006046355e-06, "loss": 0.0765, "step": 79810 }, { "epoch": 6.7, "learning_rate": 6.593886462882097e-06, "loss": 0.0761, "step": 79820 }, { "epoch": 6.7, "learning_rate": 6.592206919717837e-06, "loss": 0.0615, "step": 79830 }, { "epoch": 6.7, "learning_rate": 6.590527376553578e-06, "loss": 0.0389, "step": 79840 }, { "epoch": 6.71, "learning_rate": 6.588847833389319e-06, "loss": 0.0015, "step": 79850 }, { "epoch": 6.71, "learning_rate": 6.5871682902250585e-06, "loss": 0.0166, "step": 79860 }, { "epoch": 6.71, "learning_rate": 6.5854887470608e-06, "loss": 0.0401, "step": 79870 }, { "epoch": 6.71, "learning_rate": 6.58380920389654e-06, "loss": 0.0004, "step": 79880 }, { "epoch": 6.71, "learning_rate": 6.582129660732282e-06, "loss": 0.0273, "step": 79890 }, { "epoch": 6.71, "learning_rate": 6.580450117568022e-06, "loss": 0.0473, "step": 79900 }, { "epoch": 6.71, "learning_rate": 6.578770574403763e-06, "loss": 0.0023, "step": 79910 }, { "epoch": 6.71, "learning_rate": 6.577091031239503e-06, "loss": 0.0326, "step": 79920 }, { "epoch": 6.71, "learning_rate": 6.575411488075244e-06, "loss": 0.0005, "step": 79930 }, { "epoch": 6.71, "learning_rate": 6.573731944910985e-06, "loss": 0.0368, "step": 79940 }, { "epoch": 6.71, "learning_rate": 6.572052401746725e-06, "loss": 0.0446, "step": 79950 }, { "epoch": 6.71, "learning_rate": 6.570372858582467e-06, "loss": 0.0352, "step": 79960 }, { "epoch": 6.72, "learning_rate": 6.568693315418206e-06, "loss": 0.0334, "step": 79970 }, { "epoch": 6.72, "learning_rate": 6.567013772253948e-06, "loss": 0.0117, "step": 79980 }, { "epoch": 6.72, "learning_rate": 6.565334229089688e-06, "loss": 0.0636, "step": 79990 }, { "epoch": 6.72, "learning_rate": 6.5636546859254294e-06, "loss": 0.048, "step": 80000 }, { "epoch": 6.72, "learning_rate": 6.56197514276117e-06, "loss": 0.0333, "step": 80010 }, { "epoch": 6.72, "learning_rate": 6.5602955995969095e-06, "loss": 0.0368, "step": 80020 }, { "epoch": 6.72, "learning_rate": 6.558616056432651e-06, "loss": 0.0008, "step": 80030 }, { "epoch": 6.72, "learning_rate": 6.556936513268391e-06, "loss": 0.06, "step": 80040 }, { "epoch": 6.72, "learning_rate": 6.555256970104133e-06, "loss": 0.0401, "step": 80050 }, { "epoch": 6.72, "learning_rate": 6.553577426939873e-06, "loss": 0.072, "step": 80060 }, { "epoch": 6.72, "learning_rate": 6.551897883775613e-06, "loss": 0.0224, "step": 80070 }, { "epoch": 6.72, "learning_rate": 6.550218340611354e-06, "loss": 0.0833, "step": 80080 }, { "epoch": 6.73, "learning_rate": 6.5485387974470946e-06, "loss": 0.0512, "step": 80090 }, { "epoch": 6.73, "learning_rate": 6.546859254282836e-06, "loss": 0.0371, "step": 80100 }, { "epoch": 6.73, "learning_rate": 6.545179711118576e-06, "loss": 0.1019, "step": 80110 }, { "epoch": 6.73, "learning_rate": 6.543500167954316e-06, "loss": 0.008, "step": 80120 }, { "epoch": 6.73, "learning_rate": 6.541820624790057e-06, "loss": 0.1027, "step": 80130 }, { "epoch": 6.73, "learning_rate": 6.540141081625798e-06, "loss": 0.0621, "step": 80140 }, { "epoch": 6.73, "learning_rate": 6.538461538461539e-06, "loss": 0.0031, "step": 80150 }, { "epoch": 6.73, "learning_rate": 6.53678199529728e-06, "loss": 0.0345, "step": 80160 }, { "epoch": 6.73, "learning_rate": 6.535102452133021e-06, "loss": 0.0011, "step": 80170 }, { "epoch": 6.73, "learning_rate": 6.5334229089687606e-06, "loss": 0.0138, "step": 80180 }, { "epoch": 6.73, "learning_rate": 6.531743365804502e-06, "loss": 0.0166, "step": 80190 }, { "epoch": 6.73, "learning_rate": 6.530063822640242e-06, "loss": 0.0014, "step": 80200 }, { "epoch": 6.74, "learning_rate": 6.528384279475984e-06, "loss": 0.0002, "step": 80210 }, { "epoch": 6.74, "learning_rate": 6.526704736311724e-06, "loss": 0.0066, "step": 80220 }, { "epoch": 6.74, "learning_rate": 6.525025193147464e-06, "loss": 0.0073, "step": 80230 }, { "epoch": 6.74, "learning_rate": 6.523345649983205e-06, "loss": 0.1592, "step": 80240 }, { "epoch": 6.74, "learning_rate": 6.521666106818946e-06, "loss": 0.038, "step": 80250 }, { "epoch": 6.74, "learning_rate": 6.519986563654687e-06, "loss": 0.0193, "step": 80260 }, { "epoch": 6.74, "learning_rate": 6.518307020490427e-06, "loss": 0.0007, "step": 80270 }, { "epoch": 6.74, "learning_rate": 6.516627477326167e-06, "loss": 0.049, "step": 80280 }, { "epoch": 6.74, "learning_rate": 6.514947934161908e-06, "loss": 0.0007, "step": 80290 }, { "epoch": 6.74, "learning_rate": 6.513268390997649e-06, "loss": 0.0292, "step": 80300 }, { "epoch": 6.74, "learning_rate": 6.51158884783339e-06, "loss": 0.0007, "step": 80310 }, { "epoch": 6.75, "learning_rate": 6.509909304669131e-06, "loss": 0.0008, "step": 80320 }, { "epoch": 6.75, "learning_rate": 6.508229761504872e-06, "loss": 0.0348, "step": 80330 }, { "epoch": 6.75, "learning_rate": 6.5065502183406116e-06, "loss": 0.0753, "step": 80340 }, { "epoch": 6.75, "learning_rate": 6.504870675176352e-06, "loss": 0.0939, "step": 80350 }, { "epoch": 6.75, "learning_rate": 6.503191132012093e-06, "loss": 0.0317, "step": 80360 }, { "epoch": 6.75, "learning_rate": 6.501511588847834e-06, "loss": 0.0006, "step": 80370 }, { "epoch": 6.75, "learning_rate": 6.499832045683575e-06, "loss": 0.1283, "step": 80380 }, { "epoch": 6.75, "learning_rate": 6.498152502519315e-06, "loss": 0.0557, "step": 80390 }, { "epoch": 6.75, "learning_rate": 6.496472959355056e-06, "loss": 0.0176, "step": 80400 }, { "epoch": 6.75, "learning_rate": 6.494793416190797e-06, "loss": 0.0307, "step": 80410 }, { "epoch": 6.75, "learning_rate": 6.493113873026538e-06, "loss": 0.0527, "step": 80420 }, { "epoch": 6.75, "learning_rate": 6.491434329862278e-06, "loss": 0.023, "step": 80430 }, { "epoch": 6.76, "learning_rate": 6.489754786698018e-06, "loss": 0.075, "step": 80440 }, { "epoch": 6.76, "learning_rate": 6.488075243533759e-06, "loss": 0.0519, "step": 80450 }, { "epoch": 6.76, "learning_rate": 6.4863957003695e-06, "loss": 0.0536, "step": 80460 }, { "epoch": 6.76, "learning_rate": 6.484716157205241e-06, "loss": 0.0593, "step": 80470 }, { "epoch": 6.76, "learning_rate": 6.483036614040982e-06, "loss": 0.0475, "step": 80480 }, { "epoch": 6.76, "learning_rate": 6.481357070876723e-06, "loss": 0.0463, "step": 80490 }, { "epoch": 6.76, "learning_rate": 6.479677527712463e-06, "loss": 0.028, "step": 80500 }, { "epoch": 6.76, "learning_rate": 6.477997984548203e-06, "loss": 0.0599, "step": 80510 }, { "epoch": 6.76, "learning_rate": 6.476318441383944e-06, "loss": 0.0507, "step": 80520 }, { "epoch": 6.76, "learning_rate": 6.474638898219685e-06, "loss": 0.0007, "step": 80530 }, { "epoch": 6.76, "learning_rate": 6.472959355055426e-06, "loss": 0.0008, "step": 80540 }, { "epoch": 6.76, "learning_rate": 6.471279811891166e-06, "loss": 0.0004, "step": 80550 }, { "epoch": 6.77, "learning_rate": 6.469600268726906e-06, "loss": 0.0173, "step": 80560 }, { "epoch": 6.77, "learning_rate": 6.467920725562648e-06, "loss": 0.002, "step": 80570 }, { "epoch": 6.77, "learning_rate": 6.466241182398388e-06, "loss": 0.0559, "step": 80580 }, { "epoch": 6.77, "learning_rate": 6.464561639234129e-06, "loss": 0.0918, "step": 80590 }, { "epoch": 6.77, "learning_rate": 6.462882096069869e-06, "loss": 0.0828, "step": 80600 }, { "epoch": 6.77, "learning_rate": 6.4612025529056095e-06, "loss": 0.0432, "step": 80610 }, { "epoch": 6.77, "learning_rate": 6.459523009741351e-06, "loss": 0.0654, "step": 80620 }, { "epoch": 6.77, "learning_rate": 6.457843466577091e-06, "loss": 0.0409, "step": 80630 }, { "epoch": 6.77, "learning_rate": 6.456163923412833e-06, "loss": 0.0603, "step": 80640 }, { "epoch": 6.77, "learning_rate": 6.454484380248572e-06, "loss": 0.0381, "step": 80650 }, { "epoch": 6.77, "learning_rate": 6.452804837084314e-06, "loss": 0.0006, "step": 80660 }, { "epoch": 6.77, "learning_rate": 6.451125293920054e-06, "loss": 0.0756, "step": 80670 }, { "epoch": 6.78, "learning_rate": 6.449445750755795e-06, "loss": 0.0356, "step": 80680 }, { "epoch": 6.78, "learning_rate": 6.447766207591536e-06, "loss": 0.1265, "step": 80690 }, { "epoch": 6.78, "learning_rate": 6.446086664427277e-06, "loss": 0.0292, "step": 80700 }, { "epoch": 6.78, "learning_rate": 6.444407121263017e-06, "loss": 0.1141, "step": 80710 }, { "epoch": 6.78, "learning_rate": 6.442727578098757e-06, "loss": 0.0347, "step": 80720 }, { "epoch": 6.78, "learning_rate": 6.441048034934499e-06, "loss": 0.0864, "step": 80730 }, { "epoch": 6.78, "learning_rate": 6.439368491770239e-06, "loss": 0.0464, "step": 80740 }, { "epoch": 6.78, "learning_rate": 6.4376889486059804e-06, "loss": 0.0024, "step": 80750 }, { "epoch": 6.78, "learning_rate": 6.43600940544172e-06, "loss": 0.0005, "step": 80760 }, { "epoch": 6.78, "learning_rate": 6.4343298622774605e-06, "loss": 0.1325, "step": 80770 }, { "epoch": 6.78, "learning_rate": 6.432650319113202e-06, "loss": 0.0264, "step": 80780 }, { "epoch": 6.78, "learning_rate": 6.430970775948942e-06, "loss": 0.0361, "step": 80790 }, { "epoch": 6.79, "learning_rate": 6.429291232784684e-06, "loss": 0.0415, "step": 80800 }, { "epoch": 6.79, "learning_rate": 6.427611689620423e-06, "loss": 0.0162, "step": 80810 }, { "epoch": 6.79, "learning_rate": 6.425932146456164e-06, "loss": 0.051, "step": 80820 }, { "epoch": 6.79, "learning_rate": 6.424252603291905e-06, "loss": 0.013, "step": 80830 }, { "epoch": 6.79, "learning_rate": 6.4225730601276456e-06, "loss": 0.0646, "step": 80840 }, { "epoch": 6.79, "learning_rate": 6.420893516963387e-06, "loss": 0.0562, "step": 80850 }, { "epoch": 6.79, "learning_rate": 6.419213973799127e-06, "loss": 0.0007, "step": 80860 }, { "epoch": 6.79, "learning_rate": 6.417534430634868e-06, "loss": 0.0008, "step": 80870 }, { "epoch": 6.79, "learning_rate": 6.415854887470608e-06, "loss": 0.0005, "step": 80880 }, { "epoch": 6.79, "learning_rate": 6.41417534430635e-06, "loss": 0.002, "step": 80890 }, { "epoch": 6.79, "learning_rate": 6.41249580114209e-06, "loss": 0.0387, "step": 80900 }, { "epoch": 6.79, "learning_rate": 6.4108162579778314e-06, "loss": 0.0781, "step": 80910 }, { "epoch": 6.8, "learning_rate": 6.409136714813571e-06, "loss": 0.0222, "step": 80920 }, { "epoch": 6.8, "learning_rate": 6.4074571716493116e-06, "loss": 0.0003, "step": 80930 }, { "epoch": 6.8, "learning_rate": 6.405777628485053e-06, "loss": 0.0641, "step": 80940 }, { "epoch": 6.8, "learning_rate": 6.404098085320793e-06, "loss": 0.0014, "step": 80950 }, { "epoch": 6.8, "learning_rate": 6.402418542156535e-06, "loss": 0.0002, "step": 80960 }, { "epoch": 6.8, "learning_rate": 6.400738998992274e-06, "loss": 0.0839, "step": 80970 }, { "epoch": 6.8, "learning_rate": 6.399059455828015e-06, "loss": 0.0286, "step": 80980 }, { "epoch": 6.8, "learning_rate": 6.397379912663756e-06, "loss": 0.0295, "step": 80990 }, { "epoch": 6.8, "learning_rate": 6.395700369499497e-06, "loss": 0.0185, "step": 81000 }, { "epoch": 6.8, "learning_rate": 6.394020826335238e-06, "loss": 0.0007, "step": 81010 }, { "epoch": 6.8, "learning_rate": 6.3923412831709775e-06, "loss": 0.0729, "step": 81020 }, { "epoch": 6.8, "learning_rate": 6.390661740006718e-06, "loss": 0.0533, "step": 81030 }, { "epoch": 6.81, "learning_rate": 6.388982196842459e-06, "loss": 0.1119, "step": 81040 }, { "epoch": 6.81, "learning_rate": 6.3873026536782e-06, "loss": 0.0118, "step": 81050 }, { "epoch": 6.81, "learning_rate": 6.385623110513941e-06, "loss": 0.0003, "step": 81060 }, { "epoch": 6.81, "learning_rate": 6.383943567349682e-06, "loss": 0.0473, "step": 81070 }, { "epoch": 6.81, "learning_rate": 6.382264024185422e-06, "loss": 0.0282, "step": 81080 }, { "epoch": 6.81, "learning_rate": 6.3805844810211626e-06, "loss": 0.0786, "step": 81090 }, { "epoch": 6.81, "learning_rate": 6.378904937856903e-06, "loss": 0.0136, "step": 81100 }, { "epoch": 6.81, "learning_rate": 6.377225394692644e-06, "loss": 0.0009, "step": 81110 }, { "epoch": 6.81, "learning_rate": 6.375545851528385e-06, "loss": 0.0603, "step": 81120 }, { "epoch": 6.81, "learning_rate": 6.373866308364125e-06, "loss": 0.1681, "step": 81130 }, { "epoch": 6.81, "learning_rate": 6.372186765199866e-06, "loss": 0.0151, "step": 81140 }, { "epoch": 6.81, "learning_rate": 6.370507222035607e-06, "loss": 0.0026, "step": 81150 }, { "epoch": 6.82, "learning_rate": 6.368827678871348e-06, "loss": 0.0023, "step": 81160 }, { "epoch": 6.82, "learning_rate": 6.367148135707089e-06, "loss": 0.0558, "step": 81170 }, { "epoch": 6.82, "learning_rate": 6.3654685925428286e-06, "loss": 0.0016, "step": 81180 }, { "epoch": 6.82, "learning_rate": 6.363789049378569e-06, "loss": 0.0523, "step": 81190 }, { "epoch": 6.82, "learning_rate": 6.36210950621431e-06, "loss": 0.0685, "step": 81200 }, { "epoch": 6.82, "learning_rate": 6.360429963050051e-06, "loss": 0.0453, "step": 81210 }, { "epoch": 6.82, "learning_rate": 6.358750419885792e-06, "loss": 0.0799, "step": 81220 }, { "epoch": 6.82, "learning_rate": 6.357070876721533e-06, "loss": 0.1333, "step": 81230 }, { "epoch": 6.82, "learning_rate": 6.355391333557272e-06, "loss": 0.0013, "step": 81240 }, { "epoch": 6.82, "learning_rate": 6.353711790393014e-06, "loss": 0.0014, "step": 81250 }, { "epoch": 6.82, "learning_rate": 6.352032247228754e-06, "loss": 0.0962, "step": 81260 }, { "epoch": 6.82, "learning_rate": 6.350352704064495e-06, "loss": 0.0388, "step": 81270 }, { "epoch": 6.83, "learning_rate": 6.348673160900236e-06, "loss": 0.0704, "step": 81280 }, { "epoch": 6.83, "learning_rate": 6.3469936177359755e-06, "loss": 0.0801, "step": 81290 }, { "epoch": 6.83, "learning_rate": 6.345314074571717e-06, "loss": 0.0245, "step": 81300 }, { "epoch": 6.83, "learning_rate": 6.343634531407457e-06, "loss": 0.0008, "step": 81310 }, { "epoch": 6.83, "learning_rate": 6.341954988243199e-06, "loss": 0.001, "step": 81320 }, { "epoch": 6.83, "learning_rate": 6.340275445078939e-06, "loss": 0.0009, "step": 81330 }, { "epoch": 6.83, "learning_rate": 6.3385959019146796e-06, "loss": 0.0496, "step": 81340 }, { "epoch": 6.83, "learning_rate": 6.33691635875042e-06, "loss": 0.0322, "step": 81350 }, { "epoch": 6.83, "learning_rate": 6.335236815586161e-06, "loss": 0.0009, "step": 81360 }, { "epoch": 6.83, "learning_rate": 6.333557272421902e-06, "loss": 0.0023, "step": 81370 }, { "epoch": 6.83, "learning_rate": 6.331877729257643e-06, "loss": 0.0421, "step": 81380 }, { "epoch": 6.83, "learning_rate": 6.330198186093384e-06, "loss": 0.0322, "step": 81390 }, { "epoch": 6.84, "learning_rate": 6.328518642929123e-06, "loss": 0.0665, "step": 81400 }, { "epoch": 6.84, "learning_rate": 6.326839099764865e-06, "loss": 0.0006, "step": 81410 }, { "epoch": 6.84, "learning_rate": 6.325159556600605e-06, "loss": 0.0167, "step": 81420 }, { "epoch": 6.84, "learning_rate": 6.323480013436346e-06, "loss": 0.0625, "step": 81430 }, { "epoch": 6.84, "learning_rate": 6.321800470272087e-06, "loss": 0.0383, "step": 81440 }, { "epoch": 6.84, "learning_rate": 6.3201209271078265e-06, "loss": 0.0076, "step": 81450 }, { "epoch": 6.84, "learning_rate": 6.318441383943568e-06, "loss": 0.0647, "step": 81460 }, { "epoch": 6.84, "learning_rate": 6.316761840779308e-06, "loss": 0.0516, "step": 81470 }, { "epoch": 6.84, "learning_rate": 6.31508229761505e-06, "loss": 0.036, "step": 81480 }, { "epoch": 6.84, "learning_rate": 6.31340275445079e-06, "loss": 0.034, "step": 81490 }, { "epoch": 6.84, "learning_rate": 6.31172321128653e-06, "loss": 0.1001, "step": 81500 }, { "epoch": 6.84, "learning_rate": 6.310043668122271e-06, "loss": 0.0032, "step": 81510 }, { "epoch": 6.85, "learning_rate": 6.3083641249580115e-06, "loss": 0.0599, "step": 81520 }, { "epoch": 6.85, "learning_rate": 6.306684581793753e-06, "loss": 0.0238, "step": 81530 }, { "epoch": 6.85, "learning_rate": 6.305005038629493e-06, "loss": 0.0475, "step": 81540 }, { "epoch": 6.85, "learning_rate": 6.303325495465234e-06, "loss": 0.0268, "step": 81550 }, { "epoch": 6.85, "learning_rate": 6.301645952300974e-06, "loss": 0.0237, "step": 81560 }, { "epoch": 6.85, "learning_rate": 6.299966409136716e-06, "loss": 0.0156, "step": 81570 }, { "epoch": 6.85, "learning_rate": 6.298286865972456e-06, "loss": 0.031, "step": 81580 }, { "epoch": 6.85, "learning_rate": 6.2966073228081966e-06, "loss": 0.0565, "step": 81590 }, { "epoch": 6.85, "learning_rate": 6.294927779643938e-06, "loss": 0.0394, "step": 81600 }, { "epoch": 6.85, "learning_rate": 6.2932482364796775e-06, "loss": 0.0289, "step": 81610 }, { "epoch": 6.85, "learning_rate": 6.291568693315419e-06, "loss": 0.0017, "step": 81620 }, { "epoch": 6.86, "learning_rate": 6.289889150151159e-06, "loss": 0.1784, "step": 81630 }, { "epoch": 6.86, "learning_rate": 6.288209606986901e-06, "loss": 0.0484, "step": 81640 }, { "epoch": 6.86, "learning_rate": 6.286530063822641e-06, "loss": 0.0921, "step": 81650 }, { "epoch": 6.86, "learning_rate": 6.284850520658381e-06, "loss": 0.0318, "step": 81660 }, { "epoch": 6.86, "learning_rate": 6.283170977494122e-06, "loss": 0.0303, "step": 81670 }, { "epoch": 6.86, "learning_rate": 6.2814914343298626e-06, "loss": 0.0008, "step": 81680 }, { "epoch": 6.86, "learning_rate": 6.279811891165604e-06, "loss": 0.0524, "step": 81690 }, { "epoch": 6.86, "learning_rate": 6.278132348001344e-06, "loss": 0.0477, "step": 81700 }, { "epoch": 6.86, "learning_rate": 6.276452804837084e-06, "loss": 0.0128, "step": 81710 }, { "epoch": 6.86, "learning_rate": 6.274773261672825e-06, "loss": 0.0108, "step": 81720 }, { "epoch": 6.86, "learning_rate": 6.273093718508566e-06, "loss": 0.0009, "step": 81730 }, { "epoch": 6.86, "learning_rate": 6.271414175344307e-06, "loss": 0.0009, "step": 81740 }, { "epoch": 6.87, "learning_rate": 6.269734632180048e-06, "loss": 0.0625, "step": 81750 }, { "epoch": 6.87, "learning_rate": 6.268055089015789e-06, "loss": 0.0545, "step": 81760 }, { "epoch": 6.87, "learning_rate": 6.2663755458515285e-06, "loss": 0.1098, "step": 81770 }, { "epoch": 6.87, "learning_rate": 6.264696002687269e-06, "loss": 0.1175, "step": 81780 }, { "epoch": 6.87, "learning_rate": 6.26301645952301e-06, "loss": 0.041, "step": 81790 }, { "epoch": 6.87, "learning_rate": 6.261336916358751e-06, "loss": 0.0082, "step": 81800 }, { "epoch": 6.87, "learning_rate": 6.259657373194492e-06, "loss": 0.0033, "step": 81810 }, { "epoch": 6.87, "learning_rate": 6.257977830030232e-06, "loss": 0.0405, "step": 81820 }, { "epoch": 6.87, "learning_rate": 6.256298286865973e-06, "loss": 0.0182, "step": 81830 }, { "epoch": 6.87, "learning_rate": 6.2546187437017136e-06, "loss": 0.0718, "step": 81840 }, { "epoch": 6.87, "learning_rate": 6.252939200537455e-06, "loss": 0.0597, "step": 81850 }, { "epoch": 6.87, "learning_rate": 6.251259657373195e-06, "loss": 0.1213, "step": 81860 }, { "epoch": 6.88, "learning_rate": 6.249580114208935e-06, "loss": 0.065, "step": 81870 }, { "epoch": 6.88, "learning_rate": 6.247900571044676e-06, "loss": 0.0733, "step": 81880 }, { "epoch": 6.88, "learning_rate": 6.246221027880417e-06, "loss": 0.001, "step": 81890 }, { "epoch": 6.88, "learning_rate": 6.244541484716158e-06, "loss": 0.0092, "step": 81900 }, { "epoch": 6.88, "learning_rate": 6.242861941551899e-06, "loss": 0.0006, "step": 81910 }, { "epoch": 6.88, "learning_rate": 6.24118239838764e-06, "loss": 0.0028, "step": 81920 }, { "epoch": 6.88, "learning_rate": 6.2395028552233795e-06, "loss": 0.0073, "step": 81930 }, { "epoch": 6.88, "learning_rate": 6.23782331205912e-06, "loss": 0.056, "step": 81940 }, { "epoch": 6.88, "learning_rate": 6.236143768894861e-06, "loss": 0.0164, "step": 81950 }, { "epoch": 6.88, "learning_rate": 6.234464225730602e-06, "loss": 0.0235, "step": 81960 }, { "epoch": 6.88, "learning_rate": 6.232784682566343e-06, "loss": 0.0005, "step": 81970 }, { "epoch": 6.88, "learning_rate": 6.231105139402083e-06, "loss": 0.0002, "step": 81980 }, { "epoch": 6.89, "learning_rate": 6.229425596237823e-06, "loss": 0.0736, "step": 81990 }, { "epoch": 6.89, "learning_rate": 6.227746053073565e-06, "loss": 0.0007, "step": 82000 }, { "epoch": 6.89, "learning_rate": 6.226066509909305e-06, "loss": 0.0468, "step": 82010 }, { "epoch": 6.89, "learning_rate": 6.224386966745046e-06, "loss": 0.0061, "step": 82020 }, { "epoch": 6.89, "learning_rate": 6.222707423580786e-06, "loss": 0.0468, "step": 82030 }, { "epoch": 6.89, "learning_rate": 6.221027880416527e-06, "loss": 0.0892, "step": 82040 }, { "epoch": 6.89, "learning_rate": 6.219348337252268e-06, "loss": 0.067, "step": 82050 }, { "epoch": 6.89, "learning_rate": 6.217668794088009e-06, "loss": 0.0405, "step": 82060 }, { "epoch": 6.89, "learning_rate": 6.21598925092375e-06, "loss": 0.0009, "step": 82070 }, { "epoch": 6.89, "learning_rate": 6.214309707759489e-06, "loss": 0.019, "step": 82080 }, { "epoch": 6.89, "learning_rate": 6.2126301645952306e-06, "loss": 0.0476, "step": 82090 }, { "epoch": 6.89, "learning_rate": 6.210950621430971e-06, "loss": 0.0541, "step": 82100 }, { "epoch": 6.9, "learning_rate": 6.209271078266712e-06, "loss": 0.0618, "step": 82110 }, { "epoch": 6.9, "learning_rate": 6.207591535102453e-06, "loss": 0.022, "step": 82120 }, { "epoch": 6.9, "learning_rate": 6.205911991938194e-06, "loss": 0.0496, "step": 82130 }, { "epoch": 6.9, "learning_rate": 6.204232448773934e-06, "loss": 0.0207, "step": 82140 }, { "epoch": 6.9, "learning_rate": 6.202552905609674e-06, "loss": 0.0685, "step": 82150 }, { "epoch": 6.9, "learning_rate": 6.200873362445416e-06, "loss": 0.0823, "step": 82160 }, { "epoch": 6.9, "learning_rate": 6.199193819281156e-06, "loss": 0.0394, "step": 82170 }, { "epoch": 6.9, "learning_rate": 6.197514276116897e-06, "loss": 0.0158, "step": 82180 }, { "epoch": 6.9, "learning_rate": 6.195834732952637e-06, "loss": 0.0945, "step": 82190 }, { "epoch": 6.9, "learning_rate": 6.1941551897883775e-06, "loss": 0.0638, "step": 82200 }, { "epoch": 6.9, "learning_rate": 6.192475646624119e-06, "loss": 0.063, "step": 82210 }, { "epoch": 6.9, "learning_rate": 6.190796103459859e-06, "loss": 0.0229, "step": 82220 }, { "epoch": 6.91, "learning_rate": 6.189116560295601e-06, "loss": 0.0017, "step": 82230 }, { "epoch": 6.91, "learning_rate": 6.18743701713134e-06, "loss": 0.0225, "step": 82240 }, { "epoch": 6.91, "learning_rate": 6.185757473967081e-06, "loss": 0.0678, "step": 82250 }, { "epoch": 6.91, "learning_rate": 6.184077930802822e-06, "loss": 0.0292, "step": 82260 }, { "epoch": 6.91, "learning_rate": 6.1823983876385625e-06, "loss": 0.0156, "step": 82270 }, { "epoch": 6.91, "learning_rate": 6.180718844474304e-06, "loss": 0.0205, "step": 82280 }, { "epoch": 6.91, "learning_rate": 6.179039301310044e-06, "loss": 0.0882, "step": 82290 }, { "epoch": 6.91, "learning_rate": 6.177359758145785e-06, "loss": 0.057, "step": 82300 }, { "epoch": 6.91, "learning_rate": 6.175680214981525e-06, "loss": 0.0007, "step": 82310 }, { "epoch": 6.91, "learning_rate": 6.174000671817267e-06, "loss": 0.0799, "step": 82320 }, { "epoch": 6.91, "learning_rate": 6.172321128653007e-06, "loss": 0.001, "step": 82330 }, { "epoch": 6.91, "learning_rate": 6.170641585488748e-06, "loss": 0.0061, "step": 82340 }, { "epoch": 6.92, "learning_rate": 6.168962042324488e-06, "loss": 0.0446, "step": 82350 }, { "epoch": 6.92, "learning_rate": 6.1672824991602285e-06, "loss": 0.04, "step": 82360 }, { "epoch": 6.92, "learning_rate": 6.16560295599597e-06, "loss": 0.0531, "step": 82370 }, { "epoch": 6.92, "learning_rate": 6.16392341283171e-06, "loss": 0.0589, "step": 82380 }, { "epoch": 6.92, "learning_rate": 6.162243869667452e-06, "loss": 0.0006, "step": 82390 }, { "epoch": 6.92, "learning_rate": 6.160564326503191e-06, "loss": 0.0352, "step": 82400 }, { "epoch": 6.92, "learning_rate": 6.158884783338932e-06, "loss": 0.0362, "step": 82410 }, { "epoch": 6.92, "learning_rate": 6.157205240174673e-06, "loss": 0.022, "step": 82420 }, { "epoch": 6.92, "learning_rate": 6.1555256970104135e-06, "loss": 0.016, "step": 82430 }, { "epoch": 6.92, "learning_rate": 6.153846153846155e-06, "loss": 0.0717, "step": 82440 }, { "epoch": 6.92, "learning_rate": 6.152166610681895e-06, "loss": 0.0007, "step": 82450 }, { "epoch": 6.92, "learning_rate": 6.150487067517635e-06, "loss": 0.085, "step": 82460 }, { "epoch": 6.93, "learning_rate": 6.148807524353376e-06, "loss": 0.0335, "step": 82470 }, { "epoch": 6.93, "learning_rate": 6.147127981189117e-06, "loss": 0.0906, "step": 82480 }, { "epoch": 6.93, "learning_rate": 6.145448438024858e-06, "loss": 0.0865, "step": 82490 }, { "epoch": 6.93, "learning_rate": 6.143768894860599e-06, "loss": 0.0015, "step": 82500 }, { "epoch": 6.93, "learning_rate": 6.142089351696339e-06, "loss": 0.0482, "step": 82510 }, { "epoch": 6.93, "learning_rate": 6.1404098085320795e-06, "loss": 0.0783, "step": 82520 }, { "epoch": 6.93, "learning_rate": 6.138730265367821e-06, "loss": 0.0011, "step": 82530 }, { "epoch": 6.93, "learning_rate": 6.137050722203561e-06, "loss": 0.0196, "step": 82540 }, { "epoch": 6.93, "learning_rate": 6.135371179039303e-06, "loss": 0.0164, "step": 82550 }, { "epoch": 6.93, "learning_rate": 6.133691635875042e-06, "loss": 0.0322, "step": 82560 }, { "epoch": 6.93, "learning_rate": 6.132012092710783e-06, "loss": 0.0003, "step": 82570 }, { "epoch": 6.93, "learning_rate": 6.130332549546524e-06, "loss": 0.0977, "step": 82580 }, { "epoch": 6.94, "learning_rate": 6.1286530063822646e-06, "loss": 0.0017, "step": 82590 }, { "epoch": 6.94, "learning_rate": 6.126973463218006e-06, "loss": 0.0072, "step": 82600 }, { "epoch": 6.94, "learning_rate": 6.1252939200537455e-06, "loss": 0.0063, "step": 82610 }, { "epoch": 6.94, "learning_rate": 6.123614376889486e-06, "loss": 0.0016, "step": 82620 }, { "epoch": 6.94, "learning_rate": 6.121934833725227e-06, "loss": 0.0252, "step": 82630 }, { "epoch": 6.94, "learning_rate": 6.120255290560968e-06, "loss": 0.0349, "step": 82640 }, { "epoch": 6.94, "learning_rate": 6.118575747396709e-06, "loss": 0.047, "step": 82650 }, { "epoch": 6.94, "learning_rate": 6.11689620423245e-06, "loss": 0.0742, "step": 82660 }, { "epoch": 6.94, "learning_rate": 6.115216661068189e-06, "loss": 0.0096, "step": 82670 }, { "epoch": 6.94, "learning_rate": 6.1135371179039305e-06, "loss": 0.0875, "step": 82680 }, { "epoch": 6.94, "learning_rate": 6.111857574739671e-06, "loss": 0.1345, "step": 82690 }, { "epoch": 6.94, "learning_rate": 6.110178031575412e-06, "loss": 0.1576, "step": 82700 }, { "epoch": 6.95, "learning_rate": 6.108498488411153e-06, "loss": 0.0433, "step": 82710 }, { "epoch": 6.95, "learning_rate": 6.106818945246893e-06, "loss": 0.0373, "step": 82720 }, { "epoch": 6.95, "learning_rate": 6.105139402082634e-06, "loss": 0.001, "step": 82730 }, { "epoch": 6.95, "learning_rate": 6.103459858918374e-06, "loss": 0.0254, "step": 82740 }, { "epoch": 6.95, "learning_rate": 6.101780315754116e-06, "loss": 0.0014, "step": 82750 }, { "epoch": 6.95, "learning_rate": 6.100100772589856e-06, "loss": 0.0218, "step": 82760 }, { "epoch": 6.95, "learning_rate": 6.0984212294255965e-06, "loss": 0.0313, "step": 82770 }, { "epoch": 6.95, "learning_rate": 6.096741686261337e-06, "loss": 0.0362, "step": 82780 }, { "epoch": 6.95, "learning_rate": 6.095062143097078e-06, "loss": 0.039, "step": 82790 }, { "epoch": 6.95, "learning_rate": 6.093382599932819e-06, "loss": 0.0635, "step": 82800 }, { "epoch": 6.95, "learning_rate": 6.09170305676856e-06, "loss": 0.056, "step": 82810 }, { "epoch": 6.95, "learning_rate": 6.090023513604301e-06, "loss": 0.0844, "step": 82820 }, { "epoch": 6.96, "learning_rate": 6.08834397044004e-06, "loss": 0.0529, "step": 82830 }, { "epoch": 6.96, "learning_rate": 6.0866644272757816e-06, "loss": 0.0627, "step": 82840 }, { "epoch": 6.96, "learning_rate": 6.084984884111522e-06, "loss": 0.0754, "step": 82850 }, { "epoch": 6.96, "learning_rate": 6.083305340947263e-06, "loss": 0.0005, "step": 82860 }, { "epoch": 6.96, "learning_rate": 6.081625797783004e-06, "loss": 0.0012, "step": 82870 }, { "epoch": 6.96, "learning_rate": 6.0799462546187435e-06, "loss": 0.0294, "step": 82880 }, { "epoch": 6.96, "learning_rate": 6.078266711454485e-06, "loss": 0.0317, "step": 82890 }, { "epoch": 6.96, "learning_rate": 6.076587168290225e-06, "loss": 0.1471, "step": 82900 }, { "epoch": 6.96, "learning_rate": 6.074907625125967e-06, "loss": 0.0246, "step": 82910 }, { "epoch": 6.96, "learning_rate": 6.073228081961707e-06, "loss": 0.0037, "step": 82920 }, { "epoch": 6.96, "learning_rate": 6.071548538797447e-06, "loss": 0.0239, "step": 82930 }, { "epoch": 6.97, "learning_rate": 6.069868995633188e-06, "loss": 0.0052, "step": 82940 }, { "epoch": 6.97, "learning_rate": 6.0681894524689285e-06, "loss": 0.0264, "step": 82950 }, { "epoch": 6.97, "learning_rate": 6.06650990930467e-06, "loss": 0.0162, "step": 82960 }, { "epoch": 6.97, "learning_rate": 6.06483036614041e-06, "loss": 0.0166, "step": 82970 }, { "epoch": 6.97, "learning_rate": 6.063150822976151e-06, "loss": 0.0327, "step": 82980 }, { "epoch": 6.97, "learning_rate": 6.061471279811891e-06, "loss": 0.0009, "step": 82990 }, { "epoch": 6.97, "learning_rate": 6.0597917366476326e-06, "loss": 0.1105, "step": 83000 }, { "epoch": 6.97, "learning_rate": 6.058112193483373e-06, "loss": 0.0796, "step": 83010 }, { "epoch": 6.97, "learning_rate": 6.056432650319114e-06, "loss": 0.0816, "step": 83020 }, { "epoch": 6.97, "learning_rate": 6.054753107154855e-06, "loss": 0.0531, "step": 83030 }, { "epoch": 6.97, "learning_rate": 6.0530735639905945e-06, "loss": 0.0326, "step": 83040 }, { "epoch": 6.97, "learning_rate": 6.051394020826336e-06, "loss": 0.086, "step": 83050 }, { "epoch": 6.98, "learning_rate": 6.049714477662076e-06, "loss": 0.0415, "step": 83060 }, { "epoch": 6.98, "learning_rate": 6.048034934497818e-06, "loss": 0.042, "step": 83070 }, { "epoch": 6.98, "learning_rate": 6.046355391333558e-06, "loss": 0.063, "step": 83080 }, { "epoch": 6.98, "learning_rate": 6.044675848169298e-06, "loss": 0.003, "step": 83090 }, { "epoch": 6.98, "learning_rate": 6.042996305005039e-06, "loss": 0.0221, "step": 83100 }, { "epoch": 6.98, "learning_rate": 6.0413167618407795e-06, "loss": 0.0066, "step": 83110 }, { "epoch": 6.98, "learning_rate": 6.039637218676521e-06, "loss": 0.0858, "step": 83120 }, { "epoch": 6.98, "learning_rate": 6.037957675512261e-06, "loss": 0.0039, "step": 83130 }, { "epoch": 6.98, "learning_rate": 6.036278132348001e-06, "loss": 0.0016, "step": 83140 }, { "epoch": 6.98, "learning_rate": 6.034598589183742e-06, "loss": 0.0373, "step": 83150 }, { "epoch": 6.98, "learning_rate": 6.032919046019483e-06, "loss": 0.0514, "step": 83160 }, { "epoch": 6.98, "learning_rate": 6.031239502855224e-06, "loss": 0.0423, "step": 83170 }, { "epoch": 6.99, "learning_rate": 6.0295599596909645e-06, "loss": 0.0467, "step": 83180 }, { "epoch": 6.99, "learning_rate": 6.027880416526706e-06, "loss": 0.032, "step": 83190 }, { "epoch": 6.99, "learning_rate": 6.0262008733624455e-06, "loss": 0.0145, "step": 83200 }, { "epoch": 6.99, "learning_rate": 6.024521330198187e-06, "loss": 0.1082, "step": 83210 }, { "epoch": 6.99, "learning_rate": 6.022841787033927e-06, "loss": 0.08, "step": 83220 }, { "epoch": 6.99, "learning_rate": 6.021162243869668e-06, "loss": 0.2269, "step": 83230 }, { "epoch": 6.99, "learning_rate": 6.019482700705409e-06, "loss": 0.0024, "step": 83240 }, { "epoch": 6.99, "learning_rate": 6.017803157541149e-06, "loss": 0.0301, "step": 83250 }, { "epoch": 6.99, "learning_rate": 6.01612361437689e-06, "loss": 0.0008, "step": 83260 }, { "epoch": 6.99, "learning_rate": 6.0144440712126305e-06, "loss": 0.0906, "step": 83270 }, { "epoch": 6.99, "learning_rate": 6.012764528048372e-06, "loss": 0.0392, "step": 83280 }, { "epoch": 6.99, "learning_rate": 6.011084984884112e-06, "loss": 0.0576, "step": 83290 }, { "epoch": 7.0, "learning_rate": 6.009405441719852e-06, "loss": 0.0042, "step": 83300 }, { "epoch": 7.0, "learning_rate": 6.007725898555593e-06, "loss": 0.0325, "step": 83310 }, { "epoch": 7.0, "learning_rate": 6.006046355391334e-06, "loss": 0.0823, "step": 83320 }, { "epoch": 7.0, "learning_rate": 6.004366812227075e-06, "loss": 0.0213, "step": 83330 }, { "epoch": 7.0, "learning_rate": 6.0026872690628156e-06, "loss": 0.0279, "step": 83340 }, { "epoch": 7.0, "learning_rate": 6.001007725898557e-06, "loss": 0.0667, "step": 83350 }, { "epoch": 7.0, "eval_FN": 529, "eval_FP": 404, "eval_TN": 21310, "eval_TP": 1573, "eval_accuracy": 0.9608246556936513, "eval_f1": 0.7712674675165482, "eval_loss": 0.20476439595222473, "eval_precision": 0.7956499747091553, "eval_recall": 0.7483349191246432, "eval_runtime": 163.3015, "eval_samples_per_second": 145.841, "eval_steps_per_second": 9.118, "step": 83356 }, { "epoch": 7.0, "learning_rate": 5.9993281827342965e-06, "loss": 0.0979, "step": 83360 }, { "epoch": 7.0, "learning_rate": 5.997648639570037e-06, "loss": 0.0108, "step": 83370 }, { "epoch": 7.0, "learning_rate": 5.995969096405778e-06, "loss": 0.0012, "step": 83380 }, { "epoch": 7.0, "learning_rate": 5.994289553241519e-06, "loss": 0.0409, "step": 83390 }, { "epoch": 7.0, "learning_rate": 5.99261001007726e-06, "loss": 0.0049, "step": 83400 }, { "epoch": 7.0, "learning_rate": 5.990930466913e-06, "loss": 0.0231, "step": 83410 }, { "epoch": 7.01, "learning_rate": 5.98925092374874e-06, "loss": 0.0426, "step": 83420 }, { "epoch": 7.01, "learning_rate": 5.9875713805844815e-06, "loss": 0.0366, "step": 83430 }, { "epoch": 7.01, "learning_rate": 5.985891837420222e-06, "loss": 0.0009, "step": 83440 }, { "epoch": 7.01, "learning_rate": 5.984212294255963e-06, "loss": 0.0715, "step": 83450 }, { "epoch": 7.01, "learning_rate": 5.982532751091703e-06, "loss": 0.1429, "step": 83460 }, { "epoch": 7.01, "learning_rate": 5.980853207927444e-06, "loss": 0.0571, "step": 83470 }, { "epoch": 7.01, "learning_rate": 5.979173664763185e-06, "loss": 0.004, "step": 83480 }, { "epoch": 7.01, "learning_rate": 5.977494121598926e-06, "loss": 0.0064, "step": 83490 }, { "epoch": 7.01, "learning_rate": 5.9758145784346666e-06, "loss": 0.0216, "step": 83500 }, { "epoch": 7.01, "learning_rate": 5.974135035270406e-06, "loss": 0.0002, "step": 83510 }, { "epoch": 7.01, "learning_rate": 5.9724554921061475e-06, "loss": 0.0005, "step": 83520 }, { "epoch": 7.01, "learning_rate": 5.970775948941888e-06, "loss": 0.0807, "step": 83530 }, { "epoch": 7.02, "learning_rate": 5.969096405777629e-06, "loss": 0.0563, "step": 83540 }, { "epoch": 7.02, "learning_rate": 5.96741686261337e-06, "loss": 0.0006, "step": 83550 }, { "epoch": 7.02, "learning_rate": 5.965737319449111e-06, "loss": 0.0004, "step": 83560 }, { "epoch": 7.02, "learning_rate": 5.964057776284851e-06, "loss": 0.0604, "step": 83570 }, { "epoch": 7.02, "learning_rate": 5.962378233120591e-06, "loss": 0.0328, "step": 83580 }, { "epoch": 7.02, "learning_rate": 5.9606986899563326e-06, "loss": 0.0005, "step": 83590 }, { "epoch": 7.02, "learning_rate": 5.959019146792073e-06, "loss": 0.0161, "step": 83600 }, { "epoch": 7.02, "learning_rate": 5.957339603627814e-06, "loss": 0.0076, "step": 83610 }, { "epoch": 7.02, "learning_rate": 5.955660060463554e-06, "loss": 0.004, "step": 83620 }, { "epoch": 7.02, "learning_rate": 5.9539805172992945e-06, "loss": 0.0334, "step": 83630 }, { "epoch": 7.02, "learning_rate": 5.952300974135036e-06, "loss": 0.0594, "step": 83640 }, { "epoch": 7.02, "learning_rate": 5.950621430970776e-06, "loss": 0.031, "step": 83650 }, { "epoch": 7.03, "learning_rate": 5.948941887806518e-06, "loss": 0.0008, "step": 83660 }, { "epoch": 7.03, "learning_rate": 5.947262344642257e-06, "loss": 0.0363, "step": 83670 }, { "epoch": 7.03, "learning_rate": 5.9455828014779985e-06, "loss": 0.0005, "step": 83680 }, { "epoch": 7.03, "learning_rate": 5.943903258313739e-06, "loss": 0.0483, "step": 83690 }, { "epoch": 7.03, "learning_rate": 5.94222371514948e-06, "loss": 0.031, "step": 83700 }, { "epoch": 7.03, "learning_rate": 5.940544171985221e-06, "loss": 0.0216, "step": 83710 }, { "epoch": 7.03, "learning_rate": 5.938864628820961e-06, "loss": 0.0266, "step": 83720 }, { "epoch": 7.03, "learning_rate": 5.937185085656702e-06, "loss": 0.0834, "step": 83730 }, { "epoch": 7.03, "learning_rate": 5.935505542492442e-06, "loss": 0.1184, "step": 83740 }, { "epoch": 7.03, "learning_rate": 5.9338259993281836e-06, "loss": 0.0777, "step": 83750 }, { "epoch": 7.03, "learning_rate": 5.932146456163924e-06, "loss": 0.0162, "step": 83760 }, { "epoch": 7.03, "learning_rate": 5.930466912999665e-06, "loss": 0.0481, "step": 83770 }, { "epoch": 7.04, "learning_rate": 5.928787369835405e-06, "loss": 0.0019, "step": 83780 }, { "epoch": 7.04, "learning_rate": 5.9271078266711455e-06, "loss": 0.0346, "step": 83790 }, { "epoch": 7.04, "learning_rate": 5.925428283506887e-06, "loss": 0.0004, "step": 83800 }, { "epoch": 7.04, "learning_rate": 5.923748740342627e-06, "loss": 0.0592, "step": 83810 }, { "epoch": 7.04, "learning_rate": 5.922069197178369e-06, "loss": 0.0009, "step": 83820 }, { "epoch": 7.04, "learning_rate": 5.920389654014108e-06, "loss": 0.0226, "step": 83830 }, { "epoch": 7.04, "learning_rate": 5.918710110849849e-06, "loss": 0.0026, "step": 83840 }, { "epoch": 7.04, "learning_rate": 5.91703056768559e-06, "loss": 0.0771, "step": 83850 }, { "epoch": 7.04, "learning_rate": 5.9153510245213305e-06, "loss": 0.0291, "step": 83860 }, { "epoch": 7.04, "learning_rate": 5.913671481357072e-06, "loss": 0.0355, "step": 83870 }, { "epoch": 7.04, "learning_rate": 5.911991938192812e-06, "loss": 0.0006, "step": 83880 }, { "epoch": 7.04, "learning_rate": 5.910312395028552e-06, "loss": 0.0179, "step": 83890 }, { "epoch": 7.05, "learning_rate": 5.908632851864293e-06, "loss": 0.0179, "step": 83900 }, { "epoch": 7.05, "learning_rate": 5.906953308700034e-06, "loss": 0.0036, "step": 83910 }, { "epoch": 7.05, "learning_rate": 5.905273765535775e-06, "loss": 0.0003, "step": 83920 }, { "epoch": 7.05, "learning_rate": 5.9035942223715155e-06, "loss": 0.0004, "step": 83930 }, { "epoch": 7.05, "learning_rate": 5.901914679207256e-06, "loss": 0.0007, "step": 83940 }, { "epoch": 7.05, "learning_rate": 5.9002351360429965e-06, "loss": 0.0247, "step": 83950 }, { "epoch": 7.05, "learning_rate": 5.898555592878738e-06, "loss": 0.0823, "step": 83960 }, { "epoch": 7.05, "learning_rate": 5.896876049714478e-06, "loss": 0.0473, "step": 83970 }, { "epoch": 7.05, "learning_rate": 5.89519650655022e-06, "loss": 0.0337, "step": 83980 }, { "epoch": 7.05, "learning_rate": 5.893516963385959e-06, "loss": 0.0038, "step": 83990 }, { "epoch": 7.05, "learning_rate": 5.8918374202217e-06, "loss": 0.0359, "step": 84000 }, { "epoch": 7.05, "learning_rate": 5.890157877057441e-06, "loss": 0.0002, "step": 84010 }, { "epoch": 7.06, "learning_rate": 5.8884783338931815e-06, "loss": 0.0005, "step": 84020 }, { "epoch": 7.06, "learning_rate": 5.886798790728923e-06, "loss": 0.0674, "step": 84030 }, { "epoch": 7.06, "learning_rate": 5.8851192475646625e-06, "loss": 0.0007, "step": 84040 }, { "epoch": 7.06, "learning_rate": 5.883439704400403e-06, "loss": 0.0571, "step": 84050 }, { "epoch": 7.06, "learning_rate": 5.881760161236144e-06, "loss": 0.0754, "step": 84060 }, { "epoch": 7.06, "learning_rate": 5.880080618071885e-06, "loss": 0.0635, "step": 84070 }, { "epoch": 7.06, "learning_rate": 5.878401074907626e-06, "loss": 0.001, "step": 84080 }, { "epoch": 7.06, "learning_rate": 5.8767215317433666e-06, "loss": 0.0626, "step": 84090 }, { "epoch": 7.06, "learning_rate": 5.875041988579106e-06, "loss": 0.0011, "step": 84100 }, { "epoch": 7.06, "learning_rate": 5.8733624454148475e-06, "loss": 0.0062, "step": 84110 }, { "epoch": 7.06, "learning_rate": 5.871682902250588e-06, "loss": 0.0748, "step": 84120 }, { "epoch": 7.06, "learning_rate": 5.870003359086329e-06, "loss": 0.0114, "step": 84130 }, { "epoch": 7.07, "learning_rate": 5.86832381592207e-06, "loss": 0.0007, "step": 84140 }, { "epoch": 7.07, "learning_rate": 5.86664427275781e-06, "loss": 0.0042, "step": 84150 }, { "epoch": 7.07, "learning_rate": 5.864964729593551e-06, "loss": 0.0523, "step": 84160 }, { "epoch": 7.07, "learning_rate": 5.863285186429292e-06, "loss": 0.0736, "step": 84170 }, { "epoch": 7.07, "learning_rate": 5.8616056432650325e-06, "loss": 0.0061, "step": 84180 }, { "epoch": 7.07, "learning_rate": 5.859926100100774e-06, "loss": 0.0072, "step": 84190 }, { "epoch": 7.07, "learning_rate": 5.8582465569365135e-06, "loss": 0.0595, "step": 84200 }, { "epoch": 7.07, "learning_rate": 5.856567013772254e-06, "loss": 0.0004, "step": 84210 }, { "epoch": 7.07, "learning_rate": 5.854887470607995e-06, "loss": 0.0002, "step": 84220 }, { "epoch": 7.07, "learning_rate": 5.853207927443736e-06, "loss": 0.0015, "step": 84230 }, { "epoch": 7.07, "learning_rate": 5.851528384279477e-06, "loss": 0.0002, "step": 84240 }, { "epoch": 7.08, "learning_rate": 5.8498488411152176e-06, "loss": 0.0174, "step": 84250 }, { "epoch": 7.08, "learning_rate": 5.848169297950957e-06, "loss": 0.0213, "step": 84260 }, { "epoch": 7.08, "learning_rate": 5.8464897547866985e-06, "loss": 0.0089, "step": 84270 }, { "epoch": 7.08, "learning_rate": 5.844810211622439e-06, "loss": 0.0013, "step": 84280 }, { "epoch": 7.08, "learning_rate": 5.84313066845818e-06, "loss": 0.0002, "step": 84290 }, { "epoch": 7.08, "learning_rate": 5.841451125293921e-06, "loss": 0.0462, "step": 84300 }, { "epoch": 7.08, "learning_rate": 5.8397715821296604e-06, "loss": 0.0428, "step": 84310 }, { "epoch": 7.08, "learning_rate": 5.838092038965402e-06, "loss": 0.0034, "step": 84320 }, { "epoch": 7.08, "learning_rate": 5.836412495801142e-06, "loss": 0.0009, "step": 84330 }, { "epoch": 7.08, "learning_rate": 5.8347329526368836e-06, "loss": 0.0373, "step": 84340 }, { "epoch": 7.08, "learning_rate": 5.833053409472624e-06, "loss": 0.0519, "step": 84350 }, { "epoch": 7.08, "learning_rate": 5.8313738663083645e-06, "loss": 0.0398, "step": 84360 }, { "epoch": 7.09, "learning_rate": 5.829694323144105e-06, "loss": 0.0502, "step": 84370 }, { "epoch": 7.09, "learning_rate": 5.8280147799798455e-06, "loss": 0.0212, "step": 84380 }, { "epoch": 7.09, "learning_rate": 5.826335236815587e-06, "loss": 0.0003, "step": 84390 }, { "epoch": 7.09, "learning_rate": 5.824655693651327e-06, "loss": 0.1243, "step": 84400 }, { "epoch": 7.09, "learning_rate": 5.822976150487069e-06, "loss": 0.0174, "step": 84410 }, { "epoch": 7.09, "learning_rate": 5.821296607322808e-06, "loss": 0.0168, "step": 84420 }, { "epoch": 7.09, "learning_rate": 5.8196170641585495e-06, "loss": 0.0262, "step": 84430 }, { "epoch": 7.09, "learning_rate": 5.81793752099429e-06, "loss": 0.0327, "step": 84440 }, { "epoch": 7.09, "learning_rate": 5.816257977830031e-06, "loss": 0.0463, "step": 84450 }, { "epoch": 7.09, "learning_rate": 5.814578434665772e-06, "loss": 0.0002, "step": 84460 }, { "epoch": 7.09, "learning_rate": 5.8128988915015114e-06, "loss": 0.0006, "step": 84470 }, { "epoch": 7.09, "learning_rate": 5.811219348337253e-06, "loss": 0.1056, "step": 84480 }, { "epoch": 7.1, "learning_rate": 5.809539805172993e-06, "loss": 0.0002, "step": 84490 }, { "epoch": 7.1, "learning_rate": 5.8078602620087346e-06, "loss": 0.0275, "step": 84500 }, { "epoch": 7.1, "learning_rate": 5.806180718844475e-06, "loss": 0.0376, "step": 84510 }, { "epoch": 7.1, "learning_rate": 5.804501175680215e-06, "loss": 0.0025, "step": 84520 }, { "epoch": 7.1, "learning_rate": 5.802821632515956e-06, "loss": 0.002, "step": 84530 }, { "epoch": 7.1, "learning_rate": 5.8011420893516965e-06, "loss": 0.0001, "step": 84540 }, { "epoch": 7.1, "learning_rate": 5.799462546187438e-06, "loss": 0.023, "step": 84550 }, { "epoch": 7.1, "learning_rate": 5.797783003023178e-06, "loss": 0.0241, "step": 84560 }, { "epoch": 7.1, "learning_rate": 5.796103459858918e-06, "loss": 0.0284, "step": 84570 }, { "epoch": 7.1, "learning_rate": 5.794423916694659e-06, "loss": 0.0016, "step": 84580 }, { "epoch": 7.1, "learning_rate": 5.7927443735304e-06, "loss": 0.0473, "step": 84590 }, { "epoch": 7.1, "learning_rate": 5.791064830366141e-06, "loss": 0.0003, "step": 84600 }, { "epoch": 7.11, "learning_rate": 5.7893852872018815e-06, "loss": 0.0066, "step": 84610 }, { "epoch": 7.11, "learning_rate": 5.787705744037623e-06, "loss": 0.0036, "step": 84620 }, { "epoch": 7.11, "learning_rate": 5.7860262008733625e-06, "loss": 0.0005, "step": 84630 }, { "epoch": 7.11, "learning_rate": 5.784346657709104e-06, "loss": 0.0002, "step": 84640 }, { "epoch": 7.11, "learning_rate": 5.782667114544844e-06, "loss": 0.0002, "step": 84650 }, { "epoch": 7.11, "learning_rate": 5.780987571380586e-06, "loss": 0.0198, "step": 84660 }, { "epoch": 7.11, "learning_rate": 5.779308028216326e-06, "loss": 0.0219, "step": 84670 }, { "epoch": 7.11, "learning_rate": 5.777628485052066e-06, "loss": 0.0537, "step": 84680 }, { "epoch": 7.11, "learning_rate": 5.775948941887807e-06, "loss": 0.0444, "step": 84690 }, { "epoch": 7.11, "learning_rate": 5.7742693987235475e-06, "loss": 0.0722, "step": 84700 }, { "epoch": 7.11, "learning_rate": 5.772589855559289e-06, "loss": 0.0361, "step": 84710 }, { "epoch": 7.11, "learning_rate": 5.770910312395029e-06, "loss": 0.0378, "step": 84720 }, { "epoch": 7.12, "learning_rate": 5.769230769230769e-06, "loss": 0.0002, "step": 84730 }, { "epoch": 7.12, "learning_rate": 5.76755122606651e-06, "loss": 0.0004, "step": 84740 }, { "epoch": 7.12, "learning_rate": 5.765871682902251e-06, "loss": 0.0001, "step": 84750 }, { "epoch": 7.12, "learning_rate": 5.764192139737992e-06, "loss": 0.0985, "step": 84760 }, { "epoch": 7.12, "learning_rate": 5.7625125965737325e-06, "loss": 0.0367, "step": 84770 }, { "epoch": 7.12, "learning_rate": 5.760833053409474e-06, "loss": 0.0003, "step": 84780 }, { "epoch": 7.12, "learning_rate": 5.7591535102452135e-06, "loss": 0.0394, "step": 84790 }, { "epoch": 7.12, "learning_rate": 5.757473967080954e-06, "loss": 0.0007, "step": 84800 }, { "epoch": 7.12, "learning_rate": 5.755794423916695e-06, "loss": 0.0684, "step": 84810 }, { "epoch": 7.12, "learning_rate": 5.754114880752436e-06, "loss": 0.0412, "step": 84820 }, { "epoch": 7.12, "learning_rate": 5.752435337588177e-06, "loss": 0.0289, "step": 84830 }, { "epoch": 7.12, "learning_rate": 5.750755794423917e-06, "loss": 0.0005, "step": 84840 }, { "epoch": 7.13, "learning_rate": 5.749076251259658e-06, "loss": 0.0179, "step": 84850 }, { "epoch": 7.13, "learning_rate": 5.7473967080953985e-06, "loss": 0.0901, "step": 84860 }, { "epoch": 7.13, "learning_rate": 5.745717164931139e-06, "loss": 0.0202, "step": 84870 }, { "epoch": 7.13, "learning_rate": 5.74403762176688e-06, "loss": 0.0535, "step": 84880 }, { "epoch": 7.13, "learning_rate": 5.74235807860262e-06, "loss": 0.0321, "step": 84890 }, { "epoch": 7.13, "learning_rate": 5.740678535438361e-06, "loss": 0.0403, "step": 84900 }, { "epoch": 7.13, "learning_rate": 5.738998992274102e-06, "loss": 0.0636, "step": 84910 }, { "epoch": 7.13, "learning_rate": 5.737319449109843e-06, "loss": 0.0005, "step": 84920 }, { "epoch": 7.13, "learning_rate": 5.7356399059455835e-06, "loss": 0.0639, "step": 84930 }, { "epoch": 7.13, "learning_rate": 5.733960362781323e-06, "loss": 0.0021, "step": 84940 }, { "epoch": 7.13, "learning_rate": 5.7322808196170645e-06, "loss": 0.0324, "step": 84950 }, { "epoch": 7.13, "learning_rate": 5.730601276452805e-06, "loss": 0.0355, "step": 84960 }, { "epoch": 7.14, "learning_rate": 5.728921733288546e-06, "loss": 0.0001, "step": 84970 }, { "epoch": 7.14, "learning_rate": 5.727242190124287e-06, "loss": 0.0131, "step": 84980 }, { "epoch": 7.14, "learning_rate": 5.725562646960028e-06, "loss": 0.024, "step": 84990 }, { "epoch": 7.14, "learning_rate": 5.723883103795768e-06, "loss": 0.0457, "step": 85000 }, { "epoch": 7.14, "learning_rate": 5.722203560631508e-06, "loss": 0.0754, "step": 85010 }, { "epoch": 7.14, "learning_rate": 5.7205240174672495e-06, "loss": 0.0039, "step": 85020 }, { "epoch": 7.14, "learning_rate": 5.71884447430299e-06, "loss": 0.0005, "step": 85030 }, { "epoch": 7.14, "learning_rate": 5.717164931138731e-06, "loss": 0.0756, "step": 85040 }, { "epoch": 7.14, "learning_rate": 5.715485387974471e-06, "loss": 0.0639, "step": 85050 }, { "epoch": 7.14, "learning_rate": 5.7138058448102114e-06, "loss": 0.1138, "step": 85060 }, { "epoch": 7.14, "learning_rate": 5.712126301645953e-06, "loss": 0.0441, "step": 85070 }, { "epoch": 7.14, "learning_rate": 5.710446758481693e-06, "loss": 0.0083, "step": 85080 }, { "epoch": 7.15, "learning_rate": 5.7087672153174346e-06, "loss": 0.0247, "step": 85090 }, { "epoch": 7.15, "learning_rate": 5.707087672153174e-06, "loss": 0.021, "step": 85100 }, { "epoch": 7.15, "learning_rate": 5.7054081289889155e-06, "loss": 0.0353, "step": 85110 }, { "epoch": 7.15, "learning_rate": 5.703728585824656e-06, "loss": 0.0455, "step": 85120 }, { "epoch": 7.15, "learning_rate": 5.702049042660397e-06, "loss": 0.0476, "step": 85130 }, { "epoch": 7.15, "learning_rate": 5.700369499496138e-06, "loss": 0.0002, "step": 85140 }, { "epoch": 7.15, "learning_rate": 5.698689956331879e-06, "loss": 0.0078, "step": 85150 }, { "epoch": 7.15, "learning_rate": 5.697010413167619e-06, "loss": 0.0128, "step": 85160 }, { "epoch": 7.15, "learning_rate": 5.695330870003359e-06, "loss": 0.0004, "step": 85170 }, { "epoch": 7.15, "learning_rate": 5.6936513268391005e-06, "loss": 0.0004, "step": 85180 }, { "epoch": 7.15, "learning_rate": 5.691971783674841e-06, "loss": 0.0253, "step": 85190 }, { "epoch": 7.15, "learning_rate": 5.690292240510582e-06, "loss": 0.0006, "step": 85200 }, { "epoch": 7.16, "learning_rate": 5.688612697346322e-06, "loss": 0.0381, "step": 85210 }, { "epoch": 7.16, "learning_rate": 5.6869331541820624e-06, "loss": 0.0001, "step": 85220 }, { "epoch": 7.16, "learning_rate": 5.685253611017804e-06, "loss": 0.0378, "step": 85230 }, { "epoch": 7.16, "learning_rate": 5.683574067853544e-06, "loss": 0.0303, "step": 85240 }, { "epoch": 7.16, "learning_rate": 5.6818945246892856e-06, "loss": 0.0001, "step": 85250 }, { "epoch": 7.16, "learning_rate": 5.680214981525025e-06, "loss": 0.0672, "step": 85260 }, { "epoch": 7.16, "learning_rate": 5.678535438360766e-06, "loss": 0.0541, "step": 85270 }, { "epoch": 7.16, "learning_rate": 5.676855895196507e-06, "loss": 0.0011, "step": 85280 }, { "epoch": 7.16, "learning_rate": 5.6751763520322475e-06, "loss": 0.0528, "step": 85290 }, { "epoch": 7.16, "learning_rate": 5.673496808867989e-06, "loss": 0.0125, "step": 85300 }, { "epoch": 7.16, "learning_rate": 5.671817265703729e-06, "loss": 0.0283, "step": 85310 }, { "epoch": 7.16, "learning_rate": 5.67013772253947e-06, "loss": 0.0235, "step": 85320 }, { "epoch": 7.17, "learning_rate": 5.66845817937521e-06, "loss": 0.0753, "step": 85330 }, { "epoch": 7.17, "learning_rate": 5.6667786362109516e-06, "loss": 0.0004, "step": 85340 }, { "epoch": 7.17, "learning_rate": 5.665099093046692e-06, "loss": 0.1186, "step": 85350 }, { "epoch": 7.17, "learning_rate": 5.6634195498824325e-06, "loss": 0.0008, "step": 85360 }, { "epoch": 7.17, "learning_rate": 5.661740006718173e-06, "loss": 0.0273, "step": 85370 }, { "epoch": 7.17, "learning_rate": 5.6600604635539135e-06, "loss": 0.0188, "step": 85380 }, { "epoch": 7.17, "learning_rate": 5.658380920389655e-06, "loss": 0.0416, "step": 85390 }, { "epoch": 7.17, "learning_rate": 5.656701377225395e-06, "loss": 0.0107, "step": 85400 }, { "epoch": 7.17, "learning_rate": 5.655021834061137e-06, "loss": 0.0645, "step": 85410 }, { "epoch": 7.17, "learning_rate": 5.653342290896876e-06, "loss": 0.0245, "step": 85420 }, { "epoch": 7.17, "learning_rate": 5.651662747732617e-06, "loss": 0.0005, "step": 85430 }, { "epoch": 7.18, "learning_rate": 5.649983204568358e-06, "loss": 0.0231, "step": 85440 }, { "epoch": 7.18, "learning_rate": 5.6483036614040985e-06, "loss": 0.0026, "step": 85450 }, { "epoch": 7.18, "learning_rate": 5.64662411823984e-06, "loss": 0.026, "step": 85460 }, { "epoch": 7.18, "learning_rate": 5.6449445750755794e-06, "loss": 0.0665, "step": 85470 }, { "epoch": 7.18, "learning_rate": 5.64326503191132e-06, "loss": 0.0008, "step": 85480 }, { "epoch": 7.18, "learning_rate": 5.641585488747061e-06, "loss": 0.0184, "step": 85490 }, { "epoch": 7.18, "learning_rate": 5.639905945582802e-06, "loss": 0.0262, "step": 85500 }, { "epoch": 7.18, "learning_rate": 5.638226402418543e-06, "loss": 0.0068, "step": 85510 }, { "epoch": 7.18, "learning_rate": 5.6365468592542835e-06, "loss": 0.0002, "step": 85520 }, { "epoch": 7.18, "learning_rate": 5.634867316090024e-06, "loss": 0.0412, "step": 85530 }, { "epoch": 7.18, "learning_rate": 5.6331877729257645e-06, "loss": 0.0465, "step": 85540 }, { "epoch": 7.18, "learning_rate": 5.631508229761505e-06, "loss": 0.0002, "step": 85550 }, { "epoch": 7.19, "learning_rate": 5.629828686597246e-06, "loss": 0.0405, "step": 85560 }, { "epoch": 7.19, "learning_rate": 5.628149143432987e-06, "loss": 0.0381, "step": 85570 }, { "epoch": 7.19, "learning_rate": 5.626469600268727e-06, "loss": 0.0005, "step": 85580 }, { "epoch": 7.19, "learning_rate": 5.624790057104468e-06, "loss": 0.0187, "step": 85590 }, { "epoch": 7.19, "learning_rate": 5.623110513940209e-06, "loss": 0.001, "step": 85600 }, { "epoch": 7.19, "learning_rate": 5.6214309707759495e-06, "loss": 0.0003, "step": 85610 }, { "epoch": 7.19, "learning_rate": 5.619751427611691e-06, "loss": 0.0006, "step": 85620 }, { "epoch": 7.19, "learning_rate": 5.6180718844474305e-06, "loss": 0.0781, "step": 85630 }, { "epoch": 7.19, "learning_rate": 5.616392341283171e-06, "loss": 0.0228, "step": 85640 }, { "epoch": 7.19, "learning_rate": 5.614712798118912e-06, "loss": 0.0317, "step": 85650 }, { "epoch": 7.19, "learning_rate": 5.613033254954653e-06, "loss": 0.0507, "step": 85660 }, { "epoch": 7.19, "learning_rate": 5.611353711790394e-06, "loss": 0.0008, "step": 85670 }, { "epoch": 7.2, "learning_rate": 5.6096741686261345e-06, "loss": 0.0259, "step": 85680 }, { "epoch": 7.2, "learning_rate": 5.607994625461874e-06, "loss": 0.0318, "step": 85690 }, { "epoch": 7.2, "learning_rate": 5.6063150822976155e-06, "loss": 0.0471, "step": 85700 }, { "epoch": 7.2, "learning_rate": 5.604635539133356e-06, "loss": 0.0267, "step": 85710 }, { "epoch": 7.2, "learning_rate": 5.602955995969097e-06, "loss": 0.1221, "step": 85720 }, { "epoch": 7.2, "learning_rate": 5.601276452804838e-06, "loss": 0.0341, "step": 85730 }, { "epoch": 7.2, "learning_rate": 5.599596909640577e-06, "loss": 0.0005, "step": 85740 }, { "epoch": 7.2, "learning_rate": 5.597917366476319e-06, "loss": 0.0151, "step": 85750 }, { "epoch": 7.2, "learning_rate": 5.596237823312059e-06, "loss": 0.0006, "step": 85760 }, { "epoch": 7.2, "learning_rate": 5.5945582801478005e-06, "loss": 0.0561, "step": 85770 }, { "epoch": 7.2, "learning_rate": 5.592878736983541e-06, "loss": 0.0479, "step": 85780 }, { "epoch": 7.2, "learning_rate": 5.5911991938192815e-06, "loss": 0.0083, "step": 85790 }, { "epoch": 7.21, "learning_rate": 5.589519650655022e-06, "loss": 0.0271, "step": 85800 }, { "epoch": 7.21, "learning_rate": 5.587840107490763e-06, "loss": 0.0008, "step": 85810 }, { "epoch": 7.21, "learning_rate": 5.586160564326504e-06, "loss": 0.0571, "step": 85820 }, { "epoch": 7.21, "learning_rate": 5.584481021162245e-06, "loss": 0.0002, "step": 85830 }, { "epoch": 7.21, "learning_rate": 5.5828014779979855e-06, "loss": 0.0222, "step": 85840 }, { "epoch": 7.21, "learning_rate": 5.581121934833725e-06, "loss": 0.018, "step": 85850 }, { "epoch": 7.21, "learning_rate": 5.5794423916694665e-06, "loss": 0.008, "step": 85860 }, { "epoch": 7.21, "learning_rate": 5.577762848505207e-06, "loss": 0.0315, "step": 85870 }, { "epoch": 7.21, "learning_rate": 5.576083305340948e-06, "loss": 0.0124, "step": 85880 }, { "epoch": 7.21, "learning_rate": 5.574403762176689e-06, "loss": 0.0873, "step": 85890 }, { "epoch": 7.21, "learning_rate": 5.572724219012428e-06, "loss": 0.001, "step": 85900 }, { "epoch": 7.21, "learning_rate": 5.57104467584817e-06, "loss": 0.0321, "step": 85910 }, { "epoch": 7.22, "learning_rate": 5.56936513268391e-06, "loss": 0.0003, "step": 85920 }, { "epoch": 7.22, "learning_rate": 5.5676855895196515e-06, "loss": 0.0187, "step": 85930 }, { "epoch": 7.22, "learning_rate": 5.566006046355392e-06, "loss": 0.0112, "step": 85940 }, { "epoch": 7.22, "learning_rate": 5.564326503191132e-06, "loss": 0.0099, "step": 85950 }, { "epoch": 7.22, "learning_rate": 5.562646960026873e-06, "loss": 0.0024, "step": 85960 }, { "epoch": 7.22, "learning_rate": 5.5609674168626134e-06, "loss": 0.0321, "step": 85970 }, { "epoch": 7.22, "learning_rate": 5.559287873698355e-06, "loss": 0.0762, "step": 85980 }, { "epoch": 7.22, "learning_rate": 5.557608330534095e-06, "loss": 0.0609, "step": 85990 }, { "epoch": 7.22, "learning_rate": 5.555928787369836e-06, "loss": 0.0002, "step": 86000 }, { "epoch": 7.22, "learning_rate": 5.554249244205576e-06, "loss": 0.0009, "step": 86010 }, { "epoch": 7.22, "learning_rate": 5.5525697010413175e-06, "loss": 0.0297, "step": 86020 }, { "epoch": 7.22, "learning_rate": 5.550890157877058e-06, "loss": 0.0258, "step": 86030 }, { "epoch": 7.23, "learning_rate": 5.5492106147127985e-06, "loss": 0.0093, "step": 86040 }, { "epoch": 7.23, "learning_rate": 5.54753107154854e-06, "loss": 0.0075, "step": 86050 }, { "epoch": 7.23, "learning_rate": 5.545851528384279e-06, "loss": 0.0808, "step": 86060 }, { "epoch": 7.23, "learning_rate": 5.544171985220021e-06, "loss": 0.0008, "step": 86070 }, { "epoch": 7.23, "learning_rate": 5.542492442055761e-06, "loss": 0.0217, "step": 86080 }, { "epoch": 7.23, "learning_rate": 5.5408128988915025e-06, "loss": 0.0599, "step": 86090 }, { "epoch": 7.23, "learning_rate": 5.539133355727243e-06, "loss": 0.0581, "step": 86100 }, { "epoch": 7.23, "learning_rate": 5.537453812562983e-06, "loss": 0.0041, "step": 86110 }, { "epoch": 7.23, "learning_rate": 5.535774269398724e-06, "loss": 0.0541, "step": 86120 }, { "epoch": 7.23, "learning_rate": 5.5340947262344645e-06, "loss": 0.0005, "step": 86130 }, { "epoch": 7.23, "learning_rate": 5.532415183070206e-06, "loss": 0.0208, "step": 86140 }, { "epoch": 7.23, "learning_rate": 5.530735639905946e-06, "loss": 0.0521, "step": 86150 }, { "epoch": 7.24, "learning_rate": 5.529056096741686e-06, "loss": 0.0002, "step": 86160 }, { "epoch": 7.24, "learning_rate": 5.527376553577427e-06, "loss": 0.1055, "step": 86170 }, { "epoch": 7.24, "learning_rate": 5.525697010413168e-06, "loss": 0.0159, "step": 86180 }, { "epoch": 7.24, "learning_rate": 5.524017467248909e-06, "loss": 0.0275, "step": 86190 }, { "epoch": 7.24, "learning_rate": 5.5223379240846495e-06, "loss": 0.0945, "step": 86200 }, { "epoch": 7.24, "learning_rate": 5.520658380920391e-06, "loss": 0.0525, "step": 86210 }, { "epoch": 7.24, "learning_rate": 5.5189788377561304e-06, "loss": 0.001, "step": 86220 }, { "epoch": 7.24, "learning_rate": 5.517299294591871e-06, "loss": 0.028, "step": 86230 }, { "epoch": 7.24, "learning_rate": 5.515619751427612e-06, "loss": 0.0627, "step": 86240 }, { "epoch": 7.24, "learning_rate": 5.513940208263353e-06, "loss": 0.0531, "step": 86250 }, { "epoch": 7.24, "learning_rate": 5.512260665099094e-06, "loss": 0.0249, "step": 86260 }, { "epoch": 7.24, "learning_rate": 5.510581121934834e-06, "loss": 0.0168, "step": 86270 }, { "epoch": 7.25, "learning_rate": 5.508901578770575e-06, "loss": 0.0025, "step": 86280 }, { "epoch": 7.25, "learning_rate": 5.5072220356063155e-06, "loss": 0.0067, "step": 86290 }, { "epoch": 7.25, "learning_rate": 5.505542492442057e-06, "loss": 0.0005, "step": 86300 }, { "epoch": 7.25, "learning_rate": 5.503862949277797e-06, "loss": 0.0329, "step": 86310 }, { "epoch": 7.25, "learning_rate": 5.502183406113537e-06, "loss": 0.0294, "step": 86320 }, { "epoch": 7.25, "learning_rate": 5.500503862949278e-06, "loss": 0.0001, "step": 86330 }, { "epoch": 7.25, "learning_rate": 5.498824319785019e-06, "loss": 0.025, "step": 86340 }, { "epoch": 7.25, "learning_rate": 5.49714477662076e-06, "loss": 0.0265, "step": 86350 }, { "epoch": 7.25, "learning_rate": 5.4954652334565005e-06, "loss": 0.0028, "step": 86360 }, { "epoch": 7.25, "learning_rate": 5.493785690292242e-06, "loss": 0.0008, "step": 86370 }, { "epoch": 7.25, "learning_rate": 5.4921061471279815e-06, "loss": 0.0025, "step": 86380 }, { "epoch": 7.25, "learning_rate": 5.490426603963722e-06, "loss": 0.0388, "step": 86390 }, { "epoch": 7.26, "learning_rate": 5.488747060799463e-06, "loss": 0.0212, "step": 86400 }, { "epoch": 7.26, "learning_rate": 5.487067517635204e-06, "loss": 0.03, "step": 86410 }, { "epoch": 7.26, "learning_rate": 5.485387974470945e-06, "loss": 0.0486, "step": 86420 }, { "epoch": 7.26, "learning_rate": 5.483708431306685e-06, "loss": 0.0846, "step": 86430 }, { "epoch": 7.26, "learning_rate": 5.482028888142425e-06, "loss": 0.0001, "step": 86440 }, { "epoch": 7.26, "learning_rate": 5.4803493449781665e-06, "loss": 0.0473, "step": 86450 }, { "epoch": 7.26, "learning_rate": 5.478669801813907e-06, "loss": 0.0464, "step": 86460 }, { "epoch": 7.26, "learning_rate": 5.476990258649648e-06, "loss": 0.0231, "step": 86470 }, { "epoch": 7.26, "learning_rate": 5.475310715485388e-06, "loss": 0.0355, "step": 86480 }, { "epoch": 7.26, "learning_rate": 5.473631172321129e-06, "loss": 0.0292, "step": 86490 }, { "epoch": 7.26, "learning_rate": 5.47195162915687e-06, "loss": 0.0348, "step": 86500 }, { "epoch": 7.26, "learning_rate": 5.470272085992611e-06, "loss": 0.0104, "step": 86510 }, { "epoch": 7.27, "learning_rate": 5.4685925428283515e-06, "loss": 0.0016, "step": 86520 }, { "epoch": 7.27, "learning_rate": 5.466912999664091e-06, "loss": 0.0002, "step": 86530 }, { "epoch": 7.27, "learning_rate": 5.4652334564998325e-06, "loss": 0.0002, "step": 86540 }, { "epoch": 7.27, "learning_rate": 5.463553913335573e-06, "loss": 0.0242, "step": 86550 }, { "epoch": 7.27, "learning_rate": 5.461874370171314e-06, "loss": 0.0256, "step": 86560 }, { "epoch": 7.27, "learning_rate": 5.460194827007055e-06, "loss": 0.0638, "step": 86570 }, { "epoch": 7.27, "learning_rate": 5.458515283842796e-06, "loss": 0.0049, "step": 86580 }, { "epoch": 7.27, "learning_rate": 5.456835740678536e-06, "loss": 0.0014, "step": 86590 }, { "epoch": 7.27, "learning_rate": 5.455156197514276e-06, "loss": 0.0308, "step": 86600 }, { "epoch": 7.27, "learning_rate": 5.4534766543500175e-06, "loss": 0.0014, "step": 86610 }, { "epoch": 7.27, "learning_rate": 5.451797111185758e-06, "loss": 0.0001, "step": 86620 }, { "epoch": 7.27, "learning_rate": 5.450117568021499e-06, "loss": 0.0525, "step": 86630 }, { "epoch": 7.28, "learning_rate": 5.448438024857239e-06, "loss": 0.0623, "step": 86640 }, { "epoch": 7.28, "learning_rate": 5.446758481692979e-06, "loss": 0.0051, "step": 86650 }, { "epoch": 7.28, "learning_rate": 5.445078938528721e-06, "loss": 0.0305, "step": 86660 }, { "epoch": 7.28, "learning_rate": 5.443399395364461e-06, "loss": 0.0206, "step": 86670 }, { "epoch": 7.28, "learning_rate": 5.4417198522002025e-06, "loss": 0.0003, "step": 86680 }, { "epoch": 7.28, "learning_rate": 5.440040309035942e-06, "loss": 0.0002, "step": 86690 }, { "epoch": 7.28, "learning_rate": 5.438360765871683e-06, "loss": 0.0006, "step": 86700 }, { "epoch": 7.28, "learning_rate": 5.436681222707424e-06, "loss": 0.0004, "step": 86710 }, { "epoch": 7.28, "learning_rate": 5.4350016795431644e-06, "loss": 0.03, "step": 86720 }, { "epoch": 7.28, "learning_rate": 5.433322136378906e-06, "loss": 0.0351, "step": 86730 }, { "epoch": 7.28, "learning_rate": 5.431642593214646e-06, "loss": 0.0003, "step": 86740 }, { "epoch": 7.29, "learning_rate": 5.429963050050387e-06, "loss": 0.0451, "step": 86750 }, { "epoch": 7.29, "learning_rate": 5.428283506886127e-06, "loss": 0.0026, "step": 86760 }, { "epoch": 7.29, "learning_rate": 5.4266039637218685e-06, "loss": 0.0148, "step": 86770 }, { "epoch": 7.29, "learning_rate": 5.424924420557609e-06, "loss": 0.0007, "step": 86780 }, { "epoch": 7.29, "learning_rate": 5.42324487739335e-06, "loss": 0.0001, "step": 86790 }, { "epoch": 7.29, "learning_rate": 5.42156533422909e-06, "loss": 0.0001, "step": 86800 }, { "epoch": 7.29, "learning_rate": 5.41988579106483e-06, "loss": 0.0446, "step": 86810 }, { "epoch": 7.29, "learning_rate": 5.418206247900572e-06, "loss": 0.0001, "step": 86820 }, { "epoch": 7.29, "learning_rate": 5.416526704736312e-06, "loss": 0.0528, "step": 86830 }, { "epoch": 7.29, "learning_rate": 5.4148471615720535e-06, "loss": 0.0929, "step": 86840 }, { "epoch": 7.29, "learning_rate": 5.413167618407793e-06, "loss": 0.0012, "step": 86850 }, { "epoch": 7.29, "learning_rate": 5.411488075243534e-06, "loss": 0.0416, "step": 86860 }, { "epoch": 7.3, "learning_rate": 5.409808532079275e-06, "loss": 0.0718, "step": 86870 }, { "epoch": 7.3, "learning_rate": 5.4081289889150155e-06, "loss": 0.0007, "step": 86880 }, { "epoch": 7.3, "learning_rate": 5.406449445750757e-06, "loss": 0.0391, "step": 86890 }, { "epoch": 7.3, "learning_rate": 5.404769902586496e-06, "loss": 0.0925, "step": 86900 }, { "epoch": 7.3, "learning_rate": 5.403090359422237e-06, "loss": 0.0102, "step": 86910 }, { "epoch": 7.3, "learning_rate": 5.401410816257978e-06, "loss": 0.0277, "step": 86920 }, { "epoch": 7.3, "learning_rate": 5.399731273093719e-06, "loss": 0.0202, "step": 86930 }, { "epoch": 7.3, "learning_rate": 5.39805172992946e-06, "loss": 0.0274, "step": 86940 }, { "epoch": 7.3, "learning_rate": 5.3963721867652005e-06, "loss": 0.0021, "step": 86950 }, { "epoch": 7.3, "learning_rate": 5.394692643600941e-06, "loss": 0.0581, "step": 86960 }, { "epoch": 7.3, "learning_rate": 5.3930131004366814e-06, "loss": 0.001, "step": 86970 }, { "epoch": 7.3, "learning_rate": 5.391333557272423e-06, "loss": 0.0738, "step": 86980 }, { "epoch": 7.31, "learning_rate": 5.389654014108163e-06, "loss": 0.0438, "step": 86990 }, { "epoch": 7.31, "learning_rate": 5.387974470943904e-06, "loss": 0.0136, "step": 87000 }, { "epoch": 7.31, "learning_rate": 5.386294927779644e-06, "loss": 0.0006, "step": 87010 }, { "epoch": 7.31, "learning_rate": 5.384615384615385e-06, "loss": 0.0315, "step": 87020 }, { "epoch": 7.31, "learning_rate": 5.382935841451126e-06, "loss": 0.0263, "step": 87030 }, { "epoch": 7.31, "learning_rate": 5.3812562982868665e-06, "loss": 0.0001, "step": 87040 }, { "epoch": 7.31, "learning_rate": 5.379576755122608e-06, "loss": 0.0925, "step": 87050 }, { "epoch": 7.31, "learning_rate": 5.377897211958347e-06, "loss": 0.0005, "step": 87060 }, { "epoch": 7.31, "learning_rate": 5.376217668794088e-06, "loss": 0.0807, "step": 87070 }, { "epoch": 7.31, "learning_rate": 5.374538125629829e-06, "loss": 0.0033, "step": 87080 }, { "epoch": 7.31, "learning_rate": 5.37285858246557e-06, "loss": 0.0362, "step": 87090 }, { "epoch": 7.31, "learning_rate": 5.371179039301311e-06, "loss": 0.0121, "step": 87100 }, { "epoch": 7.32, "learning_rate": 5.3694994961370515e-06, "loss": 0.0005, "step": 87110 }, { "epoch": 7.32, "learning_rate": 5.367819952972791e-06, "loss": 0.0354, "step": 87120 }, { "epoch": 7.32, "learning_rate": 5.3661404098085324e-06, "loss": 0.034, "step": 87130 }, { "epoch": 7.32, "learning_rate": 5.364460866644273e-06, "loss": 0.071, "step": 87140 }, { "epoch": 7.32, "learning_rate": 5.362781323480014e-06, "loss": 0.0009, "step": 87150 }, { "epoch": 7.32, "learning_rate": 5.361101780315755e-06, "loss": 0.0026, "step": 87160 }, { "epoch": 7.32, "learning_rate": 5.359422237151495e-06, "loss": 0.0372, "step": 87170 }, { "epoch": 7.32, "learning_rate": 5.357742693987236e-06, "loss": 0.0016, "step": 87180 }, { "epoch": 7.32, "learning_rate": 5.356063150822976e-06, "loss": 0.0498, "step": 87190 }, { "epoch": 7.32, "learning_rate": 5.3543836076587175e-06, "loss": 0.048, "step": 87200 }, { "epoch": 7.32, "learning_rate": 5.352704064494458e-06, "loss": 0.0005, "step": 87210 }, { "epoch": 7.32, "learning_rate": 5.3510245213301984e-06, "loss": 0.0818, "step": 87220 }, { "epoch": 7.33, "learning_rate": 5.349344978165939e-06, "loss": 0.0584, "step": 87230 }, { "epoch": 7.33, "learning_rate": 5.34766543500168e-06, "loss": 0.0187, "step": 87240 }, { "epoch": 7.33, "learning_rate": 5.345985891837421e-06, "loss": 0.0336, "step": 87250 }, { "epoch": 7.33, "learning_rate": 5.344306348673162e-06, "loss": 0.0004, "step": 87260 }, { "epoch": 7.33, "learning_rate": 5.3426268055089025e-06, "loss": 0.0321, "step": 87270 }, { "epoch": 7.33, "learning_rate": 5.340947262344642e-06, "loss": 0.0143, "step": 87280 }, { "epoch": 7.33, "learning_rate": 5.3392677191803835e-06, "loss": 0.0473, "step": 87290 }, { "epoch": 7.33, "learning_rate": 5.337588176016124e-06, "loss": 0.0008, "step": 87300 }, { "epoch": 7.33, "learning_rate": 5.335908632851865e-06, "loss": 0.0467, "step": 87310 }, { "epoch": 7.33, "learning_rate": 5.334229089687606e-06, "loss": 0.0232, "step": 87320 }, { "epoch": 7.33, "learning_rate": 5.332549546523345e-06, "loss": 0.0792, "step": 87330 }, { "epoch": 7.33, "learning_rate": 5.330870003359087e-06, "loss": 0.0008, "step": 87340 }, { "epoch": 7.34, "learning_rate": 5.329190460194827e-06, "loss": 0.0029, "step": 87350 }, { "epoch": 7.34, "learning_rate": 5.3275109170305685e-06, "loss": 0.0517, "step": 87360 }, { "epoch": 7.34, "learning_rate": 5.325831373866309e-06, "loss": 0.0092, "step": 87370 }, { "epoch": 7.34, "learning_rate": 5.324151830702049e-06, "loss": 0.0001, "step": 87380 }, { "epoch": 7.34, "learning_rate": 5.32247228753779e-06, "loss": 0.0823, "step": 87390 }, { "epoch": 7.34, "learning_rate": 5.32079274437353e-06, "loss": 0.0589, "step": 87400 }, { "epoch": 7.34, "learning_rate": 5.319113201209272e-06, "loss": 0.0276, "step": 87410 }, { "epoch": 7.34, "learning_rate": 5.317433658045012e-06, "loss": 0.0529, "step": 87420 }, { "epoch": 7.34, "learning_rate": 5.315754114880753e-06, "loss": 0.0549, "step": 87430 }, { "epoch": 7.34, "learning_rate": 5.314074571716493e-06, "loss": 0.0137, "step": 87440 }, { "epoch": 7.34, "learning_rate": 5.3123950285522345e-06, "loss": 0.0245, "step": 87450 }, { "epoch": 7.34, "learning_rate": 5.310715485387975e-06, "loss": 0.0004, "step": 87460 }, { "epoch": 7.35, "learning_rate": 5.309035942223716e-06, "loss": 0.0469, "step": 87470 }, { "epoch": 7.35, "learning_rate": 5.307356399059457e-06, "loss": 0.0645, "step": 87480 }, { "epoch": 7.35, "learning_rate": 5.305676855895196e-06, "loss": 0.07, "step": 87490 }, { "epoch": 7.35, "learning_rate": 5.303997312730938e-06, "loss": 0.0114, "step": 87500 }, { "epoch": 7.35, "learning_rate": 5.302317769566678e-06, "loss": 0.03, "step": 87510 }, { "epoch": 7.35, "learning_rate": 5.3006382264024195e-06, "loss": 0.0183, "step": 87520 }, { "epoch": 7.35, "learning_rate": 5.29895868323816e-06, "loss": 0.0948, "step": 87530 }, { "epoch": 7.35, "learning_rate": 5.2972791400739e-06, "loss": 0.0241, "step": 87540 }, { "epoch": 7.35, "learning_rate": 5.295599596909641e-06, "loss": 0.0339, "step": 87550 }, { "epoch": 7.35, "learning_rate": 5.293920053745381e-06, "loss": 0.0004, "step": 87560 }, { "epoch": 7.35, "learning_rate": 5.292240510581123e-06, "loss": 0.0013, "step": 87570 }, { "epoch": 7.35, "learning_rate": 5.290560967416863e-06, "loss": 0.0224, "step": 87580 }, { "epoch": 7.36, "learning_rate": 5.288881424252603e-06, "loss": 0.0213, "step": 87590 }, { "epoch": 7.36, "learning_rate": 5.287201881088344e-06, "loss": 0.0394, "step": 87600 }, { "epoch": 7.36, "learning_rate": 5.285522337924085e-06, "loss": 0.0672, "step": 87610 }, { "epoch": 7.36, "learning_rate": 5.283842794759826e-06, "loss": 0.0146, "step": 87620 }, { "epoch": 7.36, "learning_rate": 5.2821632515955664e-06, "loss": 0.0294, "step": 87630 }, { "epoch": 7.36, "learning_rate": 5.280483708431308e-06, "loss": 0.0476, "step": 87640 }, { "epoch": 7.36, "learning_rate": 5.278804165267047e-06, "loss": 0.0231, "step": 87650 }, { "epoch": 7.36, "learning_rate": 5.277124622102789e-06, "loss": 0.0007, "step": 87660 }, { "epoch": 7.36, "learning_rate": 5.275445078938529e-06, "loss": 0.0005, "step": 87670 }, { "epoch": 7.36, "learning_rate": 5.27376553577427e-06, "loss": 0.0058, "step": 87680 }, { "epoch": 7.36, "learning_rate": 5.272085992610011e-06, "loss": 0.0058, "step": 87690 }, { "epoch": 7.36, "learning_rate": 5.270406449445751e-06, "loss": 0.0007, "step": 87700 }, { "epoch": 7.37, "learning_rate": 5.268726906281492e-06, "loss": 0.0005, "step": 87710 }, { "epoch": 7.37, "learning_rate": 5.2670473631172324e-06, "loss": 0.0006, "step": 87720 }, { "epoch": 7.37, "learning_rate": 5.265367819952974e-06, "loss": 0.1426, "step": 87730 }, { "epoch": 7.37, "learning_rate": 5.263688276788714e-06, "loss": 0.0474, "step": 87740 }, { "epoch": 7.37, "learning_rate": 5.262008733624454e-06, "loss": 0.0821, "step": 87750 }, { "epoch": 7.37, "learning_rate": 5.260329190460195e-06, "loss": 0.0227, "step": 87760 }, { "epoch": 7.37, "learning_rate": 5.258649647295936e-06, "loss": 0.0006, "step": 87770 }, { "epoch": 7.37, "learning_rate": 5.256970104131677e-06, "loss": 0.0677, "step": 87780 }, { "epoch": 7.37, "learning_rate": 5.2552905609674175e-06, "loss": 0.0046, "step": 87790 }, { "epoch": 7.37, "learning_rate": 5.253611017803159e-06, "loss": 0.0009, "step": 87800 }, { "epoch": 7.37, "learning_rate": 5.251931474638898e-06, "loss": 0.0006, "step": 87810 }, { "epoch": 7.37, "learning_rate": 5.250251931474639e-06, "loss": 0.0005, "step": 87820 }, { "epoch": 7.38, "learning_rate": 5.24857238831038e-06, "loss": 0.0427, "step": 87830 }, { "epoch": 7.38, "learning_rate": 5.246892845146121e-06, "loss": 0.007, "step": 87840 }, { "epoch": 7.38, "learning_rate": 5.245213301981862e-06, "loss": 0.0631, "step": 87850 }, { "epoch": 7.38, "learning_rate": 5.243533758817602e-06, "loss": 0.0818, "step": 87860 }, { "epoch": 7.38, "learning_rate": 5.241854215653342e-06, "loss": 0.0002, "step": 87870 }, { "epoch": 7.38, "learning_rate": 5.2401746724890834e-06, "loss": 0.0216, "step": 87880 }, { "epoch": 7.38, "learning_rate": 5.238495129324824e-06, "loss": 0.0401, "step": 87890 }, { "epoch": 7.38, "learning_rate": 5.236815586160565e-06, "loss": 0.015, "step": 87900 }, { "epoch": 7.38, "learning_rate": 5.235136042996305e-06, "loss": 0.0002, "step": 87910 }, { "epoch": 7.38, "learning_rate": 5.233456499832046e-06, "loss": 0.1006, "step": 87920 }, { "epoch": 7.38, "learning_rate": 5.231776956667787e-06, "loss": 0.0705, "step": 87930 }, { "epoch": 7.38, "learning_rate": 5.230097413503528e-06, "loss": 0.0421, "step": 87940 }, { "epoch": 7.39, "learning_rate": 5.2284178703392685e-06, "loss": 0.0333, "step": 87950 }, { "epoch": 7.39, "learning_rate": 5.226738327175008e-06, "loss": 0.0013, "step": 87960 }, { "epoch": 7.39, "learning_rate": 5.2250587840107494e-06, "loss": 0.0142, "step": 87970 }, { "epoch": 7.39, "learning_rate": 5.22337924084649e-06, "loss": 0.0003, "step": 87980 }, { "epoch": 7.39, "learning_rate": 5.221699697682231e-06, "loss": 0.0624, "step": 87990 }, { "epoch": 7.39, "learning_rate": 5.220020154517972e-06, "loss": 0.0336, "step": 88000 }, { "epoch": 7.39, "learning_rate": 5.218340611353713e-06, "loss": 0.0119, "step": 88010 }, { "epoch": 7.39, "learning_rate": 5.216661068189453e-06, "loss": 0.0448, "step": 88020 }, { "epoch": 7.39, "learning_rate": 5.214981525025193e-06, "loss": 0.0903, "step": 88030 }, { "epoch": 7.39, "learning_rate": 5.2133019818609345e-06, "loss": 0.0361, "step": 88040 }, { "epoch": 7.39, "learning_rate": 5.211622438696675e-06, "loss": 0.0248, "step": 88050 }, { "epoch": 7.4, "learning_rate": 5.209942895532416e-06, "loss": 0.0035, "step": 88060 }, { "epoch": 7.4, "learning_rate": 5.208263352368156e-06, "loss": 0.0389, "step": 88070 }, { "epoch": 7.4, "learning_rate": 5.206583809203896e-06, "loss": 0.0201, "step": 88080 }, { "epoch": 7.4, "learning_rate": 5.204904266039638e-06, "loss": 0.0014, "step": 88090 }, { "epoch": 7.4, "learning_rate": 5.203224722875378e-06, "loss": 0.0329, "step": 88100 }, { "epoch": 7.4, "learning_rate": 5.2015451797111195e-06, "loss": 0.0223, "step": 88110 }, { "epoch": 7.4, "learning_rate": 5.199865636546859e-06, "loss": 0.0221, "step": 88120 }, { "epoch": 7.4, "learning_rate": 5.1981860933826004e-06, "loss": 0.0799, "step": 88130 }, { "epoch": 7.4, "learning_rate": 5.196506550218341e-06, "loss": 0.0668, "step": 88140 }, { "epoch": 7.4, "learning_rate": 5.194827007054082e-06, "loss": 0.0304, "step": 88150 }, { "epoch": 7.4, "learning_rate": 5.193147463889823e-06, "loss": 0.0403, "step": 88160 }, { "epoch": 7.4, "learning_rate": 5.191467920725563e-06, "loss": 0.0013, "step": 88170 }, { "epoch": 7.41, "learning_rate": 5.189788377561304e-06, "loss": 0.1225, "step": 88180 }, { "epoch": 7.41, "learning_rate": 5.188108834397044e-06, "loss": 0.069, "step": 88190 }, { "epoch": 7.41, "learning_rate": 5.1864292912327855e-06, "loss": 0.0286, "step": 88200 }, { "epoch": 7.41, "learning_rate": 5.184749748068526e-06, "loss": 0.0209, "step": 88210 }, { "epoch": 7.41, "learning_rate": 5.183070204904267e-06, "loss": 0.0169, "step": 88220 }, { "epoch": 7.41, "learning_rate": 5.181390661740007e-06, "loss": 0.0584, "step": 88230 }, { "epoch": 7.41, "learning_rate": 5.179711118575747e-06, "loss": 0.0063, "step": 88240 }, { "epoch": 7.41, "learning_rate": 5.178031575411489e-06, "loss": 0.0151, "step": 88250 }, { "epoch": 7.41, "learning_rate": 5.176352032247229e-06, "loss": 0.0404, "step": 88260 }, { "epoch": 7.41, "learning_rate": 5.1746724890829705e-06, "loss": 0.0465, "step": 88270 }, { "epoch": 7.41, "learning_rate": 5.17299294591871e-06, "loss": 0.0368, "step": 88280 }, { "epoch": 7.41, "learning_rate": 5.171313402754451e-06, "loss": 0.0322, "step": 88290 }, { "epoch": 7.42, "learning_rate": 5.169633859590192e-06, "loss": 0.0002, "step": 88300 }, { "epoch": 7.42, "learning_rate": 5.167954316425932e-06, "loss": 0.0004, "step": 88310 }, { "epoch": 7.42, "learning_rate": 5.166274773261674e-06, "loss": 0.0006, "step": 88320 }, { "epoch": 7.42, "learning_rate": 5.164595230097414e-06, "loss": 0.0021, "step": 88330 }, { "epoch": 7.42, "learning_rate": 5.162915686933154e-06, "loss": 0.0343, "step": 88340 }, { "epoch": 7.42, "learning_rate": 5.161236143768895e-06, "loss": 0.0341, "step": 88350 }, { "epoch": 7.42, "learning_rate": 5.159556600604636e-06, "loss": 0.0002, "step": 88360 }, { "epoch": 7.42, "learning_rate": 5.157877057440377e-06, "loss": 0.038, "step": 88370 }, { "epoch": 7.42, "learning_rate": 5.1561975142761174e-06, "loss": 0.0377, "step": 88380 }, { "epoch": 7.42, "learning_rate": 5.154517971111858e-06, "loss": 0.1031, "step": 88390 }, { "epoch": 7.42, "learning_rate": 5.152838427947598e-06, "loss": 0.0393, "step": 88400 }, { "epoch": 7.42, "learning_rate": 5.15115888478334e-06, "loss": 0.0882, "step": 88410 }, { "epoch": 7.43, "learning_rate": 5.14947934161908e-06, "loss": 0.0005, "step": 88420 }, { "epoch": 7.43, "learning_rate": 5.1477997984548215e-06, "loss": 0.0158, "step": 88430 }, { "epoch": 7.43, "learning_rate": 5.146120255290561e-06, "loss": 0.0094, "step": 88440 }, { "epoch": 7.43, "learning_rate": 5.144440712126302e-06, "loss": 0.0011, "step": 88450 }, { "epoch": 7.43, "learning_rate": 5.142761168962043e-06, "loss": 0.0172, "step": 88460 }, { "epoch": 7.43, "learning_rate": 5.1410816257977834e-06, "loss": 0.034, "step": 88470 }, { "epoch": 7.43, "learning_rate": 5.139402082633525e-06, "loss": 0.0759, "step": 88480 }, { "epoch": 7.43, "learning_rate": 5.137722539469264e-06, "loss": 0.0268, "step": 88490 }, { "epoch": 7.43, "learning_rate": 5.136042996305005e-06, "loss": 0.0232, "step": 88500 }, { "epoch": 7.43, "learning_rate": 5.134363453140746e-06, "loss": 0.0005, "step": 88510 }, { "epoch": 7.43, "learning_rate": 5.132683909976487e-06, "loss": 0.0738, "step": 88520 }, { "epoch": 7.43, "learning_rate": 5.131004366812228e-06, "loss": 0.0076, "step": 88530 }, { "epoch": 7.44, "learning_rate": 5.1293248236479685e-06, "loss": 0.0687, "step": 88540 }, { "epoch": 7.44, "learning_rate": 5.127645280483708e-06, "loss": 0.0031, "step": 88550 }, { "epoch": 7.44, "learning_rate": 5.125965737319449e-06, "loss": 0.0163, "step": 88560 }, { "epoch": 7.44, "learning_rate": 5.12428619415519e-06, "loss": 0.0165, "step": 88570 }, { "epoch": 7.44, "learning_rate": 5.122606650990931e-06, "loss": 0.0338, "step": 88580 }, { "epoch": 7.44, "learning_rate": 5.120927107826672e-06, "loss": 0.0094, "step": 88590 }, { "epoch": 7.44, "learning_rate": 5.119247564662412e-06, "loss": 0.059, "step": 88600 }, { "epoch": 7.44, "learning_rate": 5.117568021498153e-06, "loss": 0.0006, "step": 88610 }, { "epoch": 7.44, "learning_rate": 5.115888478333894e-06, "loss": 0.0008, "step": 88620 }, { "epoch": 7.44, "learning_rate": 5.1142089351696344e-06, "loss": 0.0017, "step": 88630 }, { "epoch": 7.44, "learning_rate": 5.112529392005376e-06, "loss": 0.0256, "step": 88640 }, { "epoch": 7.44, "learning_rate": 5.110849848841115e-06, "loss": 0.0113, "step": 88650 }, { "epoch": 7.45, "learning_rate": 5.109170305676856e-06, "loss": 0.0296, "step": 88660 }, { "epoch": 7.45, "learning_rate": 5.107490762512597e-06, "loss": 0.0067, "step": 88670 }, { "epoch": 7.45, "learning_rate": 5.105811219348338e-06, "loss": 0.0024, "step": 88680 }, { "epoch": 7.45, "learning_rate": 5.104131676184079e-06, "loss": 0.0496, "step": 88690 }, { "epoch": 7.45, "learning_rate": 5.1024521330198195e-06, "loss": 0.0465, "step": 88700 }, { "epoch": 7.45, "learning_rate": 5.100772589855559e-06, "loss": 0.0027, "step": 88710 }, { "epoch": 7.45, "learning_rate": 5.0990930466913004e-06, "loss": 0.0612, "step": 88720 }, { "epoch": 7.45, "learning_rate": 5.097413503527041e-06, "loss": 0.0259, "step": 88730 }, { "epoch": 7.45, "learning_rate": 5.095733960362782e-06, "loss": 0.0342, "step": 88740 }, { "epoch": 7.45, "learning_rate": 5.094054417198523e-06, "loss": 0.034, "step": 88750 }, { "epoch": 7.45, "learning_rate": 5.092374874034262e-06, "loss": 0.0071, "step": 88760 }, { "epoch": 7.45, "learning_rate": 5.090695330870004e-06, "loss": 0.0012, "step": 88770 }, { "epoch": 7.46, "learning_rate": 5.089015787705744e-06, "loss": 0.0104, "step": 88780 }, { "epoch": 7.46, "learning_rate": 5.0873362445414855e-06, "loss": 0.0833, "step": 88790 }, { "epoch": 7.46, "learning_rate": 5.085656701377226e-06, "loss": 0.0017, "step": 88800 }, { "epoch": 7.46, "learning_rate": 5.083977158212966e-06, "loss": 0.0107, "step": 88810 }, { "epoch": 7.46, "learning_rate": 5.082297615048707e-06, "loss": 0.0398, "step": 88820 }, { "epoch": 7.46, "learning_rate": 5.080618071884447e-06, "loss": 0.0502, "step": 88830 }, { "epoch": 7.46, "learning_rate": 5.078938528720189e-06, "loss": 0.1174, "step": 88840 }, { "epoch": 7.46, "learning_rate": 5.077258985555929e-06, "loss": 0.0007, "step": 88850 }, { "epoch": 7.46, "learning_rate": 5.07557944239167e-06, "loss": 0.0458, "step": 88860 }, { "epoch": 7.46, "learning_rate": 5.07389989922741e-06, "loss": 0.0711, "step": 88870 }, { "epoch": 7.46, "learning_rate": 5.0722203560631514e-06, "loss": 0.0291, "step": 88880 }, { "epoch": 7.46, "learning_rate": 5.070540812898892e-06, "loss": 0.1296, "step": 88890 }, { "epoch": 7.47, "learning_rate": 5.068861269734633e-06, "loss": 0.0096, "step": 88900 }, { "epoch": 7.47, "learning_rate": 5.067181726570374e-06, "loss": 0.0195, "step": 88910 }, { "epoch": 7.47, "learning_rate": 5.065502183406113e-06, "loss": 0.0004, "step": 88920 }, { "epoch": 7.47, "learning_rate": 5.063822640241855e-06, "loss": 0.06, "step": 88930 }, { "epoch": 7.47, "learning_rate": 5.062143097077595e-06, "loss": 0.0002, "step": 88940 }, { "epoch": 7.47, "learning_rate": 5.0604635539133365e-06, "loss": 0.0323, "step": 88950 }, { "epoch": 7.47, "learning_rate": 5.058784010749077e-06, "loss": 0.0846, "step": 88960 }, { "epoch": 7.47, "learning_rate": 5.057104467584817e-06, "loss": 0.0092, "step": 88970 }, { "epoch": 7.47, "learning_rate": 5.055424924420558e-06, "loss": 0.0173, "step": 88980 }, { "epoch": 7.47, "learning_rate": 5.053745381256298e-06, "loss": 0.0165, "step": 88990 }, { "epoch": 7.47, "learning_rate": 5.05206583809204e-06, "loss": 0.0248, "step": 89000 }, { "epoch": 7.47, "learning_rate": 5.05038629492778e-06, "loss": 0.0501, "step": 89010 }, { "epoch": 7.48, "learning_rate": 5.04870675176352e-06, "loss": 0.0443, "step": 89020 }, { "epoch": 7.48, "learning_rate": 5.047027208599261e-06, "loss": 0.0004, "step": 89030 }, { "epoch": 7.48, "learning_rate": 5.045347665435002e-06, "loss": 0.035, "step": 89040 }, { "epoch": 7.48, "learning_rate": 5.043668122270743e-06, "loss": 0.0407, "step": 89050 }, { "epoch": 7.48, "learning_rate": 5.041988579106483e-06, "loss": 0.0027, "step": 89060 }, { "epoch": 7.48, "learning_rate": 5.040309035942225e-06, "loss": 0.0244, "step": 89070 }, { "epoch": 7.48, "learning_rate": 5.038629492777964e-06, "loss": 0.045, "step": 89080 }, { "epoch": 7.48, "learning_rate": 5.036949949613706e-06, "loss": 0.0001, "step": 89090 }, { "epoch": 7.48, "learning_rate": 5.035270406449446e-06, "loss": 0.0386, "step": 89100 }, { "epoch": 7.48, "learning_rate": 5.0335908632851875e-06, "loss": 0.0804, "step": 89110 }, { "epoch": 7.48, "learning_rate": 5.031911320120928e-06, "loss": 0.0001, "step": 89120 }, { "epoch": 7.48, "learning_rate": 5.030231776956668e-06, "loss": 0.1004, "step": 89130 }, { "epoch": 7.49, "learning_rate": 5.028552233792409e-06, "loss": 0.0713, "step": 89140 }, { "epoch": 7.49, "learning_rate": 5.026872690628149e-06, "loss": 0.0704, "step": 89150 }, { "epoch": 7.49, "learning_rate": 5.025193147463891e-06, "loss": 0.0008, "step": 89160 }, { "epoch": 7.49, "learning_rate": 5.023513604299631e-06, "loss": 0.0203, "step": 89170 }, { "epoch": 7.49, "learning_rate": 5.021834061135371e-06, "loss": 0.0005, "step": 89180 }, { "epoch": 7.49, "learning_rate": 5.020154517971112e-06, "loss": 0.0018, "step": 89190 }, { "epoch": 7.49, "learning_rate": 5.018474974806853e-06, "loss": 0.0694, "step": 89200 }, { "epoch": 7.49, "learning_rate": 5.016795431642594e-06, "loss": 0.0829, "step": 89210 }, { "epoch": 7.49, "learning_rate": 5.0151158884783344e-06, "loss": 0.0243, "step": 89220 }, { "epoch": 7.49, "learning_rate": 5.013436345314076e-06, "loss": 0.0475, "step": 89230 }, { "epoch": 7.49, "learning_rate": 5.011756802149815e-06, "loss": 0.0455, "step": 89240 }, { "epoch": 7.49, "learning_rate": 5.010077258985556e-06, "loss": 0.1098, "step": 89250 }, { "epoch": 7.5, "learning_rate": 5.008397715821297e-06, "loss": 0.0235, "step": 89260 }, { "epoch": 7.5, "learning_rate": 5.006718172657038e-06, "loss": 0.0324, "step": 89270 }, { "epoch": 7.5, "learning_rate": 5.005038629492779e-06, "loss": 0.0148, "step": 89280 }, { "epoch": 7.5, "learning_rate": 5.003359086328519e-06, "loss": 0.0077, "step": 89290 }, { "epoch": 7.5, "learning_rate": 5.00167954316426e-06, "loss": 0.0004, "step": 89300 }, { "epoch": 7.5, "learning_rate": 5e-06, "loss": 0.0766, "step": 89310 }, { "epoch": 7.5, "learning_rate": 4.998320456835741e-06, "loss": 0.0208, "step": 89320 }, { "epoch": 7.5, "learning_rate": 4.996640913671481e-06, "loss": 0.0004, "step": 89330 }, { "epoch": 7.5, "learning_rate": 4.994961370507223e-06, "loss": 0.0002, "step": 89340 }, { "epoch": 7.5, "learning_rate": 4.993281827342963e-06, "loss": 0.0211, "step": 89350 }, { "epoch": 7.5, "learning_rate": 4.991602284178704e-06, "loss": 0.0003, "step": 89360 }, { "epoch": 7.51, "learning_rate": 4.989922741014445e-06, "loss": 0.0441, "step": 89370 }, { "epoch": 7.51, "learning_rate": 4.988243197850185e-06, "loss": 0.0313, "step": 89380 }, { "epoch": 7.51, "learning_rate": 4.986563654685926e-06, "loss": 0.0002, "step": 89390 }, { "epoch": 7.51, "learning_rate": 4.984884111521666e-06, "loss": 0.0001, "step": 89400 }, { "epoch": 7.51, "learning_rate": 4.983204568357407e-06, "loss": 0.0007, "step": 89410 }, { "epoch": 7.51, "learning_rate": 4.981525025193148e-06, "loss": 0.0005, "step": 89420 }, { "epoch": 7.51, "learning_rate": 4.979845482028889e-06, "loss": 0.0002, "step": 89430 }, { "epoch": 7.51, "learning_rate": 4.978165938864629e-06, "loss": 0.0136, "step": 89440 }, { "epoch": 7.51, "learning_rate": 4.9764863957003705e-06, "loss": 0.0002, "step": 89450 }, { "epoch": 7.51, "learning_rate": 4.97480685253611e-06, "loss": 0.1156, "step": 89460 }, { "epoch": 7.51, "learning_rate": 4.9731273093718514e-06, "loss": 0.0736, "step": 89470 }, { "epoch": 7.51, "learning_rate": 4.971447766207592e-06, "loss": 0.0743, "step": 89480 }, { "epoch": 7.52, "learning_rate": 4.969768223043332e-06, "loss": 0.0002, "step": 89490 }, { "epoch": 7.52, "learning_rate": 4.968088679879074e-06, "loss": 0.0156, "step": 89500 }, { "epoch": 7.52, "learning_rate": 4.966409136714813e-06, "loss": 0.0049, "step": 89510 }, { "epoch": 7.52, "learning_rate": 4.964729593550555e-06, "loss": 0.0316, "step": 89520 }, { "epoch": 7.52, "learning_rate": 4.963050050386295e-06, "loss": 0.1047, "step": 89530 }, { "epoch": 7.52, "learning_rate": 4.961370507222036e-06, "loss": 0.0804, "step": 89540 }, { "epoch": 7.52, "learning_rate": 4.959690964057777e-06, "loss": 0.0993, "step": 89550 }, { "epoch": 7.52, "learning_rate": 4.958011420893517e-06, "loss": 0.0933, "step": 89560 }, { "epoch": 7.52, "learning_rate": 4.956331877729258e-06, "loss": 0.0019, "step": 89570 }, { "epoch": 7.52, "learning_rate": 4.954652334564999e-06, "loss": 0.0031, "step": 89580 }, { "epoch": 7.52, "learning_rate": 4.952972791400739e-06, "loss": 0.0252, "step": 89590 }, { "epoch": 7.52, "learning_rate": 4.95129324823648e-06, "loss": 0.0105, "step": 89600 }, { "epoch": 7.53, "learning_rate": 4.949613705072221e-06, "loss": 0.0246, "step": 89610 }, { "epoch": 7.53, "learning_rate": 4.947934161907961e-06, "loss": 0.0404, "step": 89620 }, { "epoch": 7.53, "learning_rate": 4.9462546187437024e-06, "loss": 0.0377, "step": 89630 }, { "epoch": 7.53, "learning_rate": 4.944575075579443e-06, "loss": 0.0101, "step": 89640 }, { "epoch": 7.53, "learning_rate": 4.942895532415183e-06, "loss": 0.0426, "step": 89650 }, { "epoch": 7.53, "learning_rate": 4.941215989250924e-06, "loss": 0.0134, "step": 89660 }, { "epoch": 7.53, "learning_rate": 4.939536446086664e-06, "loss": 0.0084, "step": 89670 }, { "epoch": 7.53, "learning_rate": 4.937856902922406e-06, "loss": 0.0011, "step": 89680 }, { "epoch": 7.53, "learning_rate": 4.936177359758146e-06, "loss": 0.0406, "step": 89690 }, { "epoch": 7.53, "learning_rate": 4.934497816593887e-06, "loss": 0.0133, "step": 89700 }, { "epoch": 7.53, "learning_rate": 4.932818273429628e-06, "loss": 0.0577, "step": 89710 }, { "epoch": 7.53, "learning_rate": 4.931138730265368e-06, "loss": 0.0623, "step": 89720 }, { "epoch": 7.54, "learning_rate": 4.929459187101109e-06, "loss": 0.0005, "step": 89730 }, { "epoch": 7.54, "learning_rate": 4.927779643936849e-06, "loss": 0.0004, "step": 89740 }, { "epoch": 7.54, "learning_rate": 4.92610010077259e-06, "loss": 0.095, "step": 89750 }, { "epoch": 7.54, "learning_rate": 4.924420557608331e-06, "loss": 0.0003, "step": 89760 }, { "epoch": 7.54, "learning_rate": 4.922741014444072e-06, "loss": 0.0297, "step": 89770 }, { "epoch": 7.54, "learning_rate": 4.921061471279812e-06, "loss": 0.0054, "step": 89780 }, { "epoch": 7.54, "learning_rate": 4.9193819281155535e-06, "loss": 0.006, "step": 89790 }, { "epoch": 7.54, "learning_rate": 4.917702384951293e-06, "loss": 0.0002, "step": 89800 }, { "epoch": 7.54, "learning_rate": 4.916022841787034e-06, "loss": 0.0059, "step": 89810 }, { "epoch": 7.54, "learning_rate": 4.914343298622775e-06, "loss": 0.0587, "step": 89820 }, { "epoch": 7.54, "learning_rate": 4.912663755458515e-06, "loss": 0.0004, "step": 89830 }, { "epoch": 7.54, "learning_rate": 4.910984212294257e-06, "loss": 0.0014, "step": 89840 }, { "epoch": 7.55, "learning_rate": 4.909304669129997e-06, "loss": 0.0445, "step": 89850 }, { "epoch": 7.55, "learning_rate": 4.907625125965738e-06, "loss": 0.0317, "step": 89860 }, { "epoch": 7.55, "learning_rate": 4.905945582801478e-06, "loss": 0.0011, "step": 89870 }, { "epoch": 7.55, "learning_rate": 4.904266039637219e-06, "loss": 0.0435, "step": 89880 }, { "epoch": 7.55, "learning_rate": 4.90258649647296e-06, "loss": 0.0603, "step": 89890 }, { "epoch": 7.55, "learning_rate": 4.9009069533087e-06, "loss": 0.0433, "step": 89900 }, { "epoch": 7.55, "learning_rate": 4.899227410144441e-06, "loss": 0.0347, "step": 89910 }, { "epoch": 7.55, "learning_rate": 4.897547866980182e-06, "loss": 0.0237, "step": 89920 }, { "epoch": 7.55, "learning_rate": 4.895868323815923e-06, "loss": 0.0056, "step": 89930 }, { "epoch": 7.55, "learning_rate": 4.894188780651663e-06, "loss": 0.0002, "step": 89940 }, { "epoch": 7.55, "learning_rate": 4.892509237487404e-06, "loss": 0.0002, "step": 89950 }, { "epoch": 7.55, "learning_rate": 4.890829694323144e-06, "loss": 0.0002, "step": 89960 }, { "epoch": 7.56, "learning_rate": 4.889150151158885e-06, "loss": 0.0113, "step": 89970 }, { "epoch": 7.56, "learning_rate": 4.887470607994626e-06, "loss": 0.011, "step": 89980 }, { "epoch": 7.56, "learning_rate": 4.885791064830366e-06, "loss": 0.0001, "step": 89990 }, { "epoch": 7.56, "learning_rate": 4.884111521666107e-06, "loss": 0.0004, "step": 90000 }, { "epoch": 7.56, "learning_rate": 4.882431978501848e-06, "loss": 0.1043, "step": 90010 }, { "epoch": 7.56, "learning_rate": 4.880752435337589e-06, "loss": 0.0057, "step": 90020 }, { "epoch": 7.56, "learning_rate": 4.879072892173329e-06, "loss": 0.0313, "step": 90030 }, { "epoch": 7.56, "learning_rate": 4.87739334900907e-06, "loss": 0.0729, "step": 90040 }, { "epoch": 7.56, "learning_rate": 4.875713805844811e-06, "loss": 0.044, "step": 90050 }, { "epoch": 7.56, "learning_rate": 4.874034262680551e-06, "loss": 0.0017, "step": 90060 }, { "epoch": 7.56, "learning_rate": 4.872354719516292e-06, "loss": 0.1197, "step": 90070 }, { "epoch": 7.56, "learning_rate": 4.870675176352032e-06, "loss": 0.0004, "step": 90080 }, { "epoch": 7.57, "learning_rate": 4.868995633187774e-06, "loss": 0.0071, "step": 90090 }, { "epoch": 7.57, "learning_rate": 4.867316090023514e-06, "loss": 0.0876, "step": 90100 }, { "epoch": 7.57, "learning_rate": 4.865636546859255e-06, "loss": 0.0501, "step": 90110 }, { "epoch": 7.57, "learning_rate": 4.863957003694995e-06, "loss": 0.0015, "step": 90120 }, { "epoch": 7.57, "learning_rate": 4.8622774605307364e-06, "loss": 0.0004, "step": 90130 }, { "epoch": 7.57, "learning_rate": 4.860597917366477e-06, "loss": 0.037, "step": 90140 }, { "epoch": 7.57, "learning_rate": 4.858918374202217e-06, "loss": 0.0396, "step": 90150 }, { "epoch": 7.57, "learning_rate": 4.857238831037958e-06, "loss": 0.039, "step": 90160 }, { "epoch": 7.57, "learning_rate": 4.855559287873699e-06, "loss": 0.0023, "step": 90170 }, { "epoch": 7.57, "learning_rate": 4.85387974470944e-06, "loss": 0.0005, "step": 90180 }, { "epoch": 7.57, "learning_rate": 4.85220020154518e-06, "loss": 0.0543, "step": 90190 }, { "epoch": 7.57, "learning_rate": 4.850520658380921e-06, "loss": 0.0211, "step": 90200 }, { "epoch": 7.58, "learning_rate": 4.848841115216661e-06, "loss": 0.0882, "step": 90210 }, { "epoch": 7.58, "learning_rate": 4.847161572052402e-06, "loss": 0.0137, "step": 90220 }, { "epoch": 7.58, "learning_rate": 4.845482028888143e-06, "loss": 0.036, "step": 90230 }, { "epoch": 7.58, "learning_rate": 4.843802485723883e-06, "loss": 0.0456, "step": 90240 }, { "epoch": 7.58, "learning_rate": 4.842122942559625e-06, "loss": 0.0298, "step": 90250 }, { "epoch": 7.58, "learning_rate": 4.840443399395365e-06, "loss": 0.0116, "step": 90260 }, { "epoch": 7.58, "learning_rate": 4.838763856231106e-06, "loss": 0.0571, "step": 90270 }, { "epoch": 7.58, "learning_rate": 4.837084313066846e-06, "loss": 0.017, "step": 90280 }, { "epoch": 7.58, "learning_rate": 4.835404769902587e-06, "loss": 0.0399, "step": 90290 }, { "epoch": 7.58, "learning_rate": 4.833725226738328e-06, "loss": 0.0928, "step": 90300 }, { "epoch": 7.58, "learning_rate": 4.832045683574068e-06, "loss": 0.0664, "step": 90310 }, { "epoch": 7.58, "learning_rate": 4.830366140409809e-06, "loss": 0.0375, "step": 90320 }, { "epoch": 7.59, "learning_rate": 4.828686597245549e-06, "loss": 0.0227, "step": 90330 }, { "epoch": 7.59, "learning_rate": 4.82700705408129e-06, "loss": 0.0024, "step": 90340 }, { "epoch": 7.59, "learning_rate": 4.825327510917031e-06, "loss": 0.0656, "step": 90350 }, { "epoch": 7.59, "learning_rate": 4.823647967752772e-06, "loss": 0.0367, "step": 90360 }, { "epoch": 7.59, "learning_rate": 4.821968424588512e-06, "loss": 0.0365, "step": 90370 }, { "epoch": 7.59, "learning_rate": 4.8202888814242534e-06, "loss": 0.0418, "step": 90380 }, { "epoch": 7.59, "learning_rate": 4.818609338259994e-06, "loss": 0.0015, "step": 90390 }, { "epoch": 7.59, "learning_rate": 4.816929795095734e-06, "loss": 0.034, "step": 90400 }, { "epoch": 7.59, "learning_rate": 4.815250251931475e-06, "loss": 0.001, "step": 90410 }, { "epoch": 7.59, "learning_rate": 4.813570708767215e-06, "loss": 0.0526, "step": 90420 }, { "epoch": 7.59, "learning_rate": 4.811891165602957e-06, "loss": 0.0374, "step": 90430 }, { "epoch": 7.59, "learning_rate": 4.810211622438697e-06, "loss": 0.0537, "step": 90440 }, { "epoch": 7.6, "learning_rate": 4.808532079274438e-06, "loss": 0.0151, "step": 90450 }, { "epoch": 7.6, "learning_rate": 4.806852536110179e-06, "loss": 0.0167, "step": 90460 }, { "epoch": 7.6, "learning_rate": 4.8051729929459186e-06, "loss": 0.028, "step": 90470 }, { "epoch": 7.6, "learning_rate": 4.80349344978166e-06, "loss": 0.0295, "step": 90480 }, { "epoch": 7.6, "learning_rate": 4.8018139066174e-06, "loss": 0.0006, "step": 90490 }, { "epoch": 7.6, "learning_rate": 4.800134363453141e-06, "loss": 0.0309, "step": 90500 }, { "epoch": 7.6, "learning_rate": 4.798454820288882e-06, "loss": 0.0789, "step": 90510 }, { "epoch": 7.6, "learning_rate": 4.796775277124623e-06, "loss": 0.0006, "step": 90520 }, { "epoch": 7.6, "learning_rate": 4.795095733960363e-06, "loss": 0.0528, "step": 90530 }, { "epoch": 7.6, "learning_rate": 4.7934161907961045e-06, "loss": 0.0256, "step": 90540 }, { "epoch": 7.6, "learning_rate": 4.791736647631844e-06, "loss": 0.034, "step": 90550 }, { "epoch": 7.6, "learning_rate": 4.790057104467585e-06, "loss": 0.0453, "step": 90560 }, { "epoch": 7.61, "learning_rate": 4.788377561303326e-06, "loss": 0.0229, "step": 90570 }, { "epoch": 7.61, "learning_rate": 4.786698018139066e-06, "loss": 0.001, "step": 90580 }, { "epoch": 7.61, "learning_rate": 4.785018474974808e-06, "loss": 0.0879, "step": 90590 }, { "epoch": 7.61, "learning_rate": 4.783338931810548e-06, "loss": 0.0158, "step": 90600 }, { "epoch": 7.61, "learning_rate": 4.781659388646289e-06, "loss": 0.0923, "step": 90610 }, { "epoch": 7.61, "learning_rate": 4.77997984548203e-06, "loss": 0.0087, "step": 90620 }, { "epoch": 7.61, "learning_rate": 4.77830030231777e-06, "loss": 0.0342, "step": 90630 }, { "epoch": 7.61, "learning_rate": 4.776620759153511e-06, "loss": 0.0207, "step": 90640 }, { "epoch": 7.61, "learning_rate": 4.774941215989251e-06, "loss": 0.0006, "step": 90650 }, { "epoch": 7.61, "learning_rate": 4.773261672824992e-06, "loss": 0.0167, "step": 90660 }, { "epoch": 7.61, "learning_rate": 4.771582129660733e-06, "loss": 0.0351, "step": 90670 }, { "epoch": 7.62, "learning_rate": 4.769902586496473e-06, "loss": 0.0294, "step": 90680 }, { "epoch": 7.62, "learning_rate": 4.768223043332214e-06, "loss": 0.0007, "step": 90690 }, { "epoch": 7.62, "learning_rate": 4.766543500167955e-06, "loss": 0.0016, "step": 90700 }, { "epoch": 7.62, "learning_rate": 4.764863957003695e-06, "loss": 0.0003, "step": 90710 }, { "epoch": 7.62, "learning_rate": 4.763184413839436e-06, "loss": 0.0531, "step": 90720 }, { "epoch": 7.62, "learning_rate": 4.761504870675177e-06, "loss": 0.04, "step": 90730 }, { "epoch": 7.62, "learning_rate": 4.759825327510917e-06, "loss": 0.0156, "step": 90740 }, { "epoch": 7.62, "learning_rate": 4.758145784346659e-06, "loss": 0.0526, "step": 90750 }, { "epoch": 7.62, "learning_rate": 4.756466241182398e-06, "loss": 0.1037, "step": 90760 }, { "epoch": 7.62, "learning_rate": 4.75478669801814e-06, "loss": 0.0384, "step": 90770 }, { "epoch": 7.62, "learning_rate": 4.75310715485388e-06, "loss": 0.0005, "step": 90780 }, { "epoch": 7.62, "learning_rate": 4.751427611689621e-06, "loss": 0.0008, "step": 90790 }, { "epoch": 7.63, "learning_rate": 4.749748068525362e-06, "loss": 0.003, "step": 90800 }, { "epoch": 7.63, "learning_rate": 4.7480685253611016e-06, "loss": 0.002, "step": 90810 }, { "epoch": 7.63, "learning_rate": 4.746388982196843e-06, "loss": 0.0004, "step": 90820 }, { "epoch": 7.63, "learning_rate": 4.744709439032583e-06, "loss": 0.0014, "step": 90830 }, { "epoch": 7.63, "learning_rate": 4.743029895868324e-06, "loss": 0.0186, "step": 90840 }, { "epoch": 7.63, "learning_rate": 4.741350352704065e-06, "loss": 0.064, "step": 90850 }, { "epoch": 7.63, "learning_rate": 4.739670809539806e-06, "loss": 0.071, "step": 90860 }, { "epoch": 7.63, "learning_rate": 4.737991266375546e-06, "loss": 0.0001, "step": 90870 }, { "epoch": 7.63, "learning_rate": 4.7363117232112874e-06, "loss": 0.0038, "step": 90880 }, { "epoch": 7.63, "learning_rate": 4.734632180047027e-06, "loss": 0.0734, "step": 90890 }, { "epoch": 7.63, "learning_rate": 4.732952636882768e-06, "loss": 0.0047, "step": 90900 }, { "epoch": 7.63, "learning_rate": 4.731273093718509e-06, "loss": 0.0009, "step": 90910 }, { "epoch": 7.64, "learning_rate": 4.729593550554249e-06, "loss": 0.0376, "step": 90920 }, { "epoch": 7.64, "learning_rate": 4.727914007389991e-06, "loss": 0.0321, "step": 90930 }, { "epoch": 7.64, "learning_rate": 4.726234464225731e-06, "loss": 0.0456, "step": 90940 }, { "epoch": 7.64, "learning_rate": 4.724554921061472e-06, "loss": 0.0011, "step": 90950 }, { "epoch": 7.64, "learning_rate": 4.722875377897212e-06, "loss": 0.0509, "step": 90960 }, { "epoch": 7.64, "learning_rate": 4.7211958347329526e-06, "loss": 0.036, "step": 90970 }, { "epoch": 7.64, "learning_rate": 4.719516291568694e-06, "loss": 0.0604, "step": 90980 }, { "epoch": 7.64, "learning_rate": 4.717836748404434e-06, "loss": 0.0341, "step": 90990 }, { "epoch": 7.64, "learning_rate": 4.716157205240175e-06, "loss": 0.0353, "step": 91000 }, { "epoch": 7.64, "learning_rate": 4.714477662075916e-06, "loss": 0.0862, "step": 91010 }, { "epoch": 7.64, "learning_rate": 4.712798118911656e-06, "loss": 0.0671, "step": 91020 }, { "epoch": 7.64, "learning_rate": 4.711118575747397e-06, "loss": 0.003, "step": 91030 }, { "epoch": 7.65, "learning_rate": 4.709439032583138e-06, "loss": 0.0641, "step": 91040 }, { "epoch": 7.65, "learning_rate": 4.707759489418878e-06, "loss": 0.0315, "step": 91050 }, { "epoch": 7.65, "learning_rate": 4.706079946254619e-06, "loss": 0.0295, "step": 91060 }, { "epoch": 7.65, "learning_rate": 4.70440040309036e-06, "loss": 0.0158, "step": 91070 }, { "epoch": 7.65, "learning_rate": 4.7027208599261e-06, "loss": 0.0078, "step": 91080 }, { "epoch": 7.65, "learning_rate": 4.701041316761842e-06, "loss": 0.0004, "step": 91090 }, { "epoch": 7.65, "learning_rate": 4.699361773597581e-06, "loss": 0.0008, "step": 91100 }, { "epoch": 7.65, "learning_rate": 4.697682230433323e-06, "loss": 0.0934, "step": 91110 }, { "epoch": 7.65, "learning_rate": 4.696002687269063e-06, "loss": 0.0003, "step": 91120 }, { "epoch": 7.65, "learning_rate": 4.694323144104804e-06, "loss": 0.0002, "step": 91130 }, { "epoch": 7.65, "learning_rate": 4.692643600940545e-06, "loss": 0.0679, "step": 91140 }, { "epoch": 7.65, "learning_rate": 4.690964057776285e-06, "loss": 0.0006, "step": 91150 }, { "epoch": 7.66, "learning_rate": 4.689284514612026e-06, "loss": 0.0202, "step": 91160 }, { "epoch": 7.66, "learning_rate": 4.687604971447766e-06, "loss": 0.0335, "step": 91170 }, { "epoch": 7.66, "learning_rate": 4.685925428283507e-06, "loss": 0.0626, "step": 91180 }, { "epoch": 7.66, "learning_rate": 4.684245885119248e-06, "loss": 0.0183, "step": 91190 }, { "epoch": 7.66, "learning_rate": 4.682566341954989e-06, "loss": 0.0592, "step": 91200 }, { "epoch": 7.66, "learning_rate": 4.680886798790729e-06, "loss": 0.0007, "step": 91210 }, { "epoch": 7.66, "learning_rate": 4.67920725562647e-06, "loss": 0.0021, "step": 91220 }, { "epoch": 7.66, "learning_rate": 4.677527712462211e-06, "loss": 0.026, "step": 91230 }, { "epoch": 7.66, "learning_rate": 4.675848169297951e-06, "loss": 0.0086, "step": 91240 }, { "epoch": 7.66, "learning_rate": 4.674168626133692e-06, "loss": 0.0454, "step": 91250 }, { "epoch": 7.66, "learning_rate": 4.672489082969432e-06, "loss": 0.1037, "step": 91260 }, { "epoch": 7.66, "learning_rate": 4.670809539805174e-06, "loss": 0.0002, "step": 91270 }, { "epoch": 7.67, "learning_rate": 4.669129996640914e-06, "loss": 0.114, "step": 91280 }, { "epoch": 7.67, "learning_rate": 4.667450453476655e-06, "loss": 0.0551, "step": 91290 }, { "epoch": 7.67, "learning_rate": 4.665770910312395e-06, "loss": 0.0006, "step": 91300 }, { "epoch": 7.67, "learning_rate": 4.6640913671481356e-06, "loss": 0.0026, "step": 91310 }, { "epoch": 7.67, "learning_rate": 4.662411823983877e-06, "loss": 0.0014, "step": 91320 }, { "epoch": 7.67, "learning_rate": 4.660732280819617e-06, "loss": 0.0002, "step": 91330 }, { "epoch": 7.67, "learning_rate": 4.659052737655358e-06, "loss": 0.0665, "step": 91340 }, { "epoch": 7.67, "learning_rate": 4.657373194491099e-06, "loss": 0.0003, "step": 91350 }, { "epoch": 7.67, "learning_rate": 4.65569365132684e-06, "loss": 0.0036, "step": 91360 }, { "epoch": 7.67, "learning_rate": 4.65401410816258e-06, "loss": 0.042, "step": 91370 }, { "epoch": 7.67, "learning_rate": 4.652334564998321e-06, "loss": 0.0666, "step": 91380 }, { "epoch": 7.67, "learning_rate": 4.650655021834061e-06, "loss": 0.0442, "step": 91390 }, { "epoch": 7.68, "learning_rate": 4.648975478669802e-06, "loss": 0.0005, "step": 91400 }, { "epoch": 7.68, "learning_rate": 4.647295935505543e-06, "loss": 0.0476, "step": 91410 }, { "epoch": 7.68, "learning_rate": 4.645616392341283e-06, "loss": 0.0479, "step": 91420 }, { "epoch": 7.68, "learning_rate": 4.643936849177025e-06, "loss": 0.0026, "step": 91430 }, { "epoch": 7.68, "learning_rate": 4.642257306012765e-06, "loss": 0.0035, "step": 91440 }, { "epoch": 7.68, "learning_rate": 4.640577762848506e-06, "loss": 0.1103, "step": 91450 }, { "epoch": 7.68, "learning_rate": 4.638898219684246e-06, "loss": 0.0004, "step": 91460 }, { "epoch": 7.68, "learning_rate": 4.6372186765199866e-06, "loss": 0.0565, "step": 91470 }, { "epoch": 7.68, "learning_rate": 4.635539133355728e-06, "loss": 0.0272, "step": 91480 }, { "epoch": 7.68, "learning_rate": 4.633859590191468e-06, "loss": 0.0419, "step": 91490 }, { "epoch": 7.68, "learning_rate": 4.632180047027209e-06, "loss": 0.0292, "step": 91500 }, { "epoch": 7.68, "learning_rate": 4.630500503862949e-06, "loss": 0.0005, "step": 91510 }, { "epoch": 7.69, "learning_rate": 4.628820960698691e-06, "loss": 0.0124, "step": 91520 }, { "epoch": 7.69, "learning_rate": 4.627141417534431e-06, "loss": 0.0254, "step": 91530 }, { "epoch": 7.69, "learning_rate": 4.625461874370172e-06, "loss": 0.0005, "step": 91540 }, { "epoch": 7.69, "learning_rate": 4.623782331205912e-06, "loss": 0.0717, "step": 91550 }, { "epoch": 7.69, "learning_rate": 4.622102788041653e-06, "loss": 0.0243, "step": 91560 }, { "epoch": 7.69, "learning_rate": 4.620423244877394e-06, "loss": 0.0156, "step": 91570 }, { "epoch": 7.69, "learning_rate": 4.618743701713134e-06, "loss": 0.0317, "step": 91580 }, { "epoch": 7.69, "learning_rate": 4.617064158548875e-06, "loss": 0.047, "step": 91590 }, { "epoch": 7.69, "learning_rate": 4.615384615384616e-06, "loss": 0.0002, "step": 91600 }, { "epoch": 7.69, "learning_rate": 4.613705072220357e-06, "loss": 0.0291, "step": 91610 }, { "epoch": 7.69, "learning_rate": 4.612025529056097e-06, "loss": 0.0004, "step": 91620 }, { "epoch": 7.69, "learning_rate": 4.610345985891838e-06, "loss": 0.0005, "step": 91630 }, { "epoch": 7.7, "learning_rate": 4.608666442727578e-06, "loss": 0.1237, "step": 91640 }, { "epoch": 7.7, "learning_rate": 4.606986899563319e-06, "loss": 0.0371, "step": 91650 }, { "epoch": 7.7, "learning_rate": 4.60530735639906e-06, "loss": 0.0318, "step": 91660 }, { "epoch": 7.7, "learning_rate": 4.6036278132348e-06, "loss": 0.0004, "step": 91670 }, { "epoch": 7.7, "learning_rate": 4.601948270070542e-06, "loss": 0.0006, "step": 91680 }, { "epoch": 7.7, "learning_rate": 4.600268726906282e-06, "loss": 0.1017, "step": 91690 }, { "epoch": 7.7, "learning_rate": 4.598589183742023e-06, "loss": 0.0891, "step": 91700 }, { "epoch": 7.7, "learning_rate": 4.596909640577763e-06, "loss": 0.0316, "step": 91710 }, { "epoch": 7.7, "learning_rate": 4.5952300974135036e-06, "loss": 0.0015, "step": 91720 }, { "epoch": 7.7, "learning_rate": 4.593550554249245e-06, "loss": 0.0749, "step": 91730 }, { "epoch": 7.7, "learning_rate": 4.591871011084985e-06, "loss": 0.0261, "step": 91740 }, { "epoch": 7.7, "learning_rate": 4.590191467920726e-06, "loss": 0.0006, "step": 91750 }, { "epoch": 7.71, "learning_rate": 4.588511924756466e-06, "loss": 0.0421, "step": 91760 }, { "epoch": 7.71, "learning_rate": 4.586832381592208e-06, "loss": 0.0272, "step": 91770 }, { "epoch": 7.71, "learning_rate": 4.585152838427948e-06, "loss": 0.0313, "step": 91780 }, { "epoch": 7.71, "learning_rate": 4.583473295263689e-06, "loss": 0.0005, "step": 91790 }, { "epoch": 7.71, "learning_rate": 4.581793752099429e-06, "loss": 0.0004, "step": 91800 }, { "epoch": 7.71, "learning_rate": 4.58011420893517e-06, "loss": 0.0089, "step": 91810 }, { "epoch": 7.71, "learning_rate": 4.578434665770911e-06, "loss": 0.0953, "step": 91820 }, { "epoch": 7.71, "learning_rate": 4.576755122606651e-06, "loss": 0.0005, "step": 91830 }, { "epoch": 7.71, "learning_rate": 4.575075579442392e-06, "loss": 0.0009, "step": 91840 }, { "epoch": 7.71, "learning_rate": 4.573396036278132e-06, "loss": 0.0061, "step": 91850 }, { "epoch": 7.71, "learning_rate": 4.571716493113874e-06, "loss": 0.0012, "step": 91860 }, { "epoch": 7.71, "learning_rate": 4.570036949949614e-06, "loss": 0.0616, "step": 91870 }, { "epoch": 7.72, "learning_rate": 4.568357406785355e-06, "loss": 0.0346, "step": 91880 }, { "epoch": 7.72, "learning_rate": 4.566677863621096e-06, "loss": 0.0261, "step": 91890 }, { "epoch": 7.72, "learning_rate": 4.564998320456836e-06, "loss": 0.0039, "step": 91900 }, { "epoch": 7.72, "learning_rate": 4.563318777292577e-06, "loss": 0.0007, "step": 91910 }, { "epoch": 7.72, "learning_rate": 4.561639234128317e-06, "loss": 0.0026, "step": 91920 }, { "epoch": 7.72, "learning_rate": 4.559959690964058e-06, "loss": 0.0549, "step": 91930 }, { "epoch": 7.72, "learning_rate": 4.558280147799799e-06, "loss": 0.0326, "step": 91940 }, { "epoch": 7.72, "learning_rate": 4.55660060463554e-06, "loss": 0.0195, "step": 91950 }, { "epoch": 7.72, "learning_rate": 4.55492106147128e-06, "loss": 0.0022, "step": 91960 }, { "epoch": 7.72, "learning_rate": 4.553241518307021e-06, "loss": 0.0409, "step": 91970 }, { "epoch": 7.72, "learning_rate": 4.551561975142761e-06, "loss": 0.0232, "step": 91980 }, { "epoch": 7.73, "learning_rate": 4.549882431978502e-06, "loss": 0.0107, "step": 91990 }, { "epoch": 7.73, "learning_rate": 4.548202888814243e-06, "loss": 0.0002, "step": 92000 }, { "epoch": 7.73, "learning_rate": 4.546523345649983e-06, "loss": 0.0687, "step": 92010 }, { "epoch": 7.73, "learning_rate": 4.544843802485725e-06, "loss": 0.0024, "step": 92020 }, { "epoch": 7.73, "learning_rate": 4.543164259321465e-06, "loss": 0.0113, "step": 92030 }, { "epoch": 7.73, "learning_rate": 4.541484716157206e-06, "loss": 0.0868, "step": 92040 }, { "epoch": 7.73, "learning_rate": 4.539805172992947e-06, "loss": 0.0597, "step": 92050 }, { "epoch": 7.73, "learning_rate": 4.5381256298286866e-06, "loss": 0.012, "step": 92060 }, { "epoch": 7.73, "learning_rate": 4.536446086664428e-06, "loss": 0.0217, "step": 92070 }, { "epoch": 7.73, "learning_rate": 4.534766543500168e-06, "loss": 0.0001, "step": 92080 }, { "epoch": 7.73, "learning_rate": 4.533087000335909e-06, "loss": 0.0554, "step": 92090 }, { "epoch": 7.73, "learning_rate": 4.53140745717165e-06, "loss": 0.0045, "step": 92100 }, { "epoch": 7.74, "learning_rate": 4.52972791400739e-06, "loss": 0.0374, "step": 92110 }, { "epoch": 7.74, "learning_rate": 4.528048370843131e-06, "loss": 0.0283, "step": 92120 }, { "epoch": 7.74, "learning_rate": 4.526368827678872e-06, "loss": 0.0002, "step": 92130 }, { "epoch": 7.74, "learning_rate": 4.524689284514612e-06, "loss": 0.0657, "step": 92140 }, { "epoch": 7.74, "learning_rate": 4.523009741350353e-06, "loss": 0.0023, "step": 92150 }, { "epoch": 7.74, "learning_rate": 4.521330198186094e-06, "loss": 0.0572, "step": 92160 }, { "epoch": 7.74, "learning_rate": 4.519650655021834e-06, "loss": 0.0018, "step": 92170 }, { "epoch": 7.74, "learning_rate": 4.517971111857576e-06, "loss": 0.0413, "step": 92180 }, { "epoch": 7.74, "learning_rate": 4.516291568693315e-06, "loss": 0.0017, "step": 92190 }, { "epoch": 7.74, "learning_rate": 4.514612025529057e-06, "loss": 0.0207, "step": 92200 }, { "epoch": 7.74, "learning_rate": 4.512932482364797e-06, "loss": 0.0341, "step": 92210 }, { "epoch": 7.74, "learning_rate": 4.5112529392005376e-06, "loss": 0.001, "step": 92220 }, { "epoch": 7.75, "learning_rate": 4.509573396036279e-06, "loss": 0.0501, "step": 92230 }, { "epoch": 7.75, "learning_rate": 4.507893852872019e-06, "loss": 0.0085, "step": 92240 }, { "epoch": 7.75, "learning_rate": 4.50621430970776e-06, "loss": 0.023, "step": 92250 }, { "epoch": 7.75, "learning_rate": 4.504534766543501e-06, "loss": 0.0051, "step": 92260 }, { "epoch": 7.75, "learning_rate": 4.502855223379241e-06, "loss": 0.0268, "step": 92270 }, { "epoch": 7.75, "learning_rate": 4.501175680214982e-06, "loss": 0.0004, "step": 92280 }, { "epoch": 7.75, "learning_rate": 4.499496137050723e-06, "loss": 0.0459, "step": 92290 }, { "epoch": 7.75, "learning_rate": 4.497816593886463e-06, "loss": 0.0006, "step": 92300 }, { "epoch": 7.75, "learning_rate": 4.496137050722204e-06, "loss": 0.0897, "step": 92310 }, { "epoch": 7.75, "learning_rate": 4.494457507557944e-06, "loss": 0.0259, "step": 92320 }, { "epoch": 7.75, "learning_rate": 4.492777964393685e-06, "loss": 0.0004, "step": 92330 }, { "epoch": 7.75, "learning_rate": 4.491098421229426e-06, "loss": 0.0706, "step": 92340 }, { "epoch": 7.76, "learning_rate": 4.489418878065166e-06, "loss": 0.0142, "step": 92350 }, { "epoch": 7.76, "learning_rate": 4.487739334900908e-06, "loss": 0.0606, "step": 92360 }, { "epoch": 7.76, "learning_rate": 4.486059791736648e-06, "loss": 0.0124, "step": 92370 }, { "epoch": 7.76, "learning_rate": 4.484380248572389e-06, "loss": 0.0182, "step": 92380 }, { "epoch": 7.76, "learning_rate": 4.48270070540813e-06, "loss": 0.0009, "step": 92390 }, { "epoch": 7.76, "learning_rate": 4.4810211622438695e-06, "loss": 0.0387, "step": 92400 }, { "epoch": 7.76, "learning_rate": 4.479341619079611e-06, "loss": 0.0158, "step": 92410 }, { "epoch": 7.76, "learning_rate": 4.477662075915351e-06, "loss": 0.0243, "step": 92420 }, { "epoch": 7.76, "learning_rate": 4.475982532751092e-06, "loss": 0.0847, "step": 92430 }, { "epoch": 7.76, "learning_rate": 4.474302989586833e-06, "loss": 0.0207, "step": 92440 }, { "epoch": 7.76, "learning_rate": 4.472623446422573e-06, "loss": 0.0002, "step": 92450 }, { "epoch": 7.76, "learning_rate": 4.470943903258314e-06, "loss": 0.0246, "step": 92460 }, { "epoch": 7.77, "learning_rate": 4.4692643600940546e-06, "loss": 0.0225, "step": 92470 }, { "epoch": 7.77, "learning_rate": 4.467584816929795e-06, "loss": 0.0477, "step": 92480 }, { "epoch": 7.77, "learning_rate": 4.465905273765536e-06, "loss": 0.055, "step": 92490 }, { "epoch": 7.77, "learning_rate": 4.464225730601277e-06, "loss": 0.0316, "step": 92500 }, { "epoch": 7.77, "learning_rate": 4.462546187437017e-06, "loss": 0.0265, "step": 92510 }, { "epoch": 7.77, "learning_rate": 4.460866644272759e-06, "loss": 0.0371, "step": 92520 }, { "epoch": 7.77, "learning_rate": 4.459187101108498e-06, "loss": 0.0164, "step": 92530 }, { "epoch": 7.77, "learning_rate": 4.45750755794424e-06, "loss": 0.0071, "step": 92540 }, { "epoch": 7.77, "learning_rate": 4.45582801477998e-06, "loss": 0.0003, "step": 92550 }, { "epoch": 7.77, "learning_rate": 4.4541484716157205e-06, "loss": 0.0108, "step": 92560 }, { "epoch": 7.77, "learning_rate": 4.452468928451462e-06, "loss": 0.0433, "step": 92570 }, { "epoch": 7.77, "learning_rate": 4.450789385287202e-06, "loss": 0.078, "step": 92580 }, { "epoch": 7.78, "learning_rate": 4.449109842122943e-06, "loss": 0.0003, "step": 92590 }, { "epoch": 7.78, "learning_rate": 4.447430298958683e-06, "loss": 0.0259, "step": 92600 }, { "epoch": 7.78, "learning_rate": 4.445750755794424e-06, "loss": 0.0006, "step": 92610 }, { "epoch": 7.78, "learning_rate": 4.444071212630165e-06, "loss": 0.017, "step": 92620 }, { "epoch": 7.78, "learning_rate": 4.442391669465906e-06, "loss": 0.0378, "step": 92630 }, { "epoch": 7.78, "learning_rate": 4.440712126301646e-06, "loss": 0.0007, "step": 92640 }, { "epoch": 7.78, "learning_rate": 4.439032583137387e-06, "loss": 0.0255, "step": 92650 }, { "epoch": 7.78, "learning_rate": 4.437353039973128e-06, "loss": 0.0533, "step": 92660 }, { "epoch": 7.78, "learning_rate": 4.435673496808868e-06, "loss": 0.1, "step": 92670 }, { "epoch": 7.78, "learning_rate": 4.433993953644609e-06, "loss": 0.029, "step": 92680 }, { "epoch": 7.78, "learning_rate": 4.432314410480349e-06, "loss": 0.0034, "step": 92690 }, { "epoch": 7.78, "learning_rate": 4.430634867316091e-06, "loss": 0.0502, "step": 92700 }, { "epoch": 7.79, "learning_rate": 4.428955324151831e-06, "loss": 0.0904, "step": 92710 }, { "epoch": 7.79, "learning_rate": 4.4272757809875716e-06, "loss": 0.0082, "step": 92720 }, { "epoch": 7.79, "learning_rate": 4.425596237823313e-06, "loss": 0.0003, "step": 92730 }, { "epoch": 7.79, "learning_rate": 4.4239166946590525e-06, "loss": 0.0118, "step": 92740 }, { "epoch": 7.79, "learning_rate": 4.422237151494794e-06, "loss": 0.0205, "step": 92750 }, { "epoch": 7.79, "learning_rate": 4.420557608330534e-06, "loss": 0.0405, "step": 92760 }, { "epoch": 7.79, "learning_rate": 4.418878065166275e-06, "loss": 0.0293, "step": 92770 }, { "epoch": 7.79, "learning_rate": 4.417198522002016e-06, "loss": 0.0343, "step": 92780 }, { "epoch": 7.79, "learning_rate": 4.415518978837757e-06, "loss": 0.0585, "step": 92790 }, { "epoch": 7.79, "learning_rate": 4.413839435673497e-06, "loss": 0.0015, "step": 92800 }, { "epoch": 7.79, "learning_rate": 4.4121598925092375e-06, "loss": 0.0243, "step": 92810 }, { "epoch": 7.79, "learning_rate": 4.410480349344978e-06, "loss": 0.0838, "step": 92820 }, { "epoch": 7.8, "learning_rate": 4.408800806180719e-06, "loss": 0.0303, "step": 92830 }, { "epoch": 7.8, "learning_rate": 4.40712126301646e-06, "loss": 0.0372, "step": 92840 }, { "epoch": 7.8, "learning_rate": 4.4054417198522e-06, "loss": 0.0006, "step": 92850 }, { "epoch": 7.8, "learning_rate": 4.403762176687942e-06, "loss": 0.0032, "step": 92860 }, { "epoch": 7.8, "learning_rate": 4.402082633523682e-06, "loss": 0.096, "step": 92870 }, { "epoch": 7.8, "learning_rate": 4.400403090359423e-06, "loss": 0.0488, "step": 92880 }, { "epoch": 7.8, "learning_rate": 4.398723547195163e-06, "loss": 0.0025, "step": 92890 }, { "epoch": 7.8, "learning_rate": 4.3970440040309035e-06, "loss": 0.0019, "step": 92900 }, { "epoch": 7.8, "learning_rate": 4.395364460866645e-06, "loss": 0.0002, "step": 92910 }, { "epoch": 7.8, "learning_rate": 4.393684917702385e-06, "loss": 0.0235, "step": 92920 }, { "epoch": 7.8, "learning_rate": 4.392005374538126e-06, "loss": 0.0006, "step": 92930 }, { "epoch": 7.8, "learning_rate": 4.390325831373866e-06, "loss": 0.0829, "step": 92940 }, { "epoch": 7.81, "learning_rate": 4.388646288209608e-06, "loss": 0.0828, "step": 92950 }, { "epoch": 7.81, "learning_rate": 4.386966745045348e-06, "loss": 0.0351, "step": 92960 }, { "epoch": 7.81, "learning_rate": 4.3852872018810886e-06, "loss": 0.0669, "step": 92970 }, { "epoch": 7.81, "learning_rate": 4.383607658716829e-06, "loss": 0.001, "step": 92980 }, { "epoch": 7.81, "learning_rate": 4.38192811555257e-06, "loss": 0.0003, "step": 92990 }, { "epoch": 7.81, "learning_rate": 4.380248572388311e-06, "loss": 0.0002, "step": 93000 }, { "epoch": 7.81, "learning_rate": 4.378569029224051e-06, "loss": 0.0139, "step": 93010 }, { "epoch": 7.81, "learning_rate": 4.376889486059792e-06, "loss": 0.0911, "step": 93020 }, { "epoch": 7.81, "learning_rate": 4.375209942895533e-06, "loss": 0.0011, "step": 93030 }, { "epoch": 7.81, "learning_rate": 4.373530399731274e-06, "loss": 0.0821, "step": 93040 }, { "epoch": 7.81, "learning_rate": 4.371850856567014e-06, "loss": 0.0002, "step": 93050 }, { "epoch": 7.81, "learning_rate": 4.3701713134027545e-06, "loss": 0.0368, "step": 93060 }, { "epoch": 7.82, "learning_rate": 4.368491770238496e-06, "loss": 0.0385, "step": 93070 }, { "epoch": 7.82, "learning_rate": 4.366812227074236e-06, "loss": 0.1521, "step": 93080 }, { "epoch": 7.82, "learning_rate": 4.365132683909977e-06, "loss": 0.0039, "step": 93090 }, { "epoch": 7.82, "learning_rate": 4.363453140745717e-06, "loss": 0.0037, "step": 93100 }, { "epoch": 7.82, "learning_rate": 4.361773597581459e-06, "loss": 0.0276, "step": 93110 }, { "epoch": 7.82, "learning_rate": 4.360094054417199e-06, "loss": 0.0008, "step": 93120 }, { "epoch": 7.82, "learning_rate": 4.35841451125294e-06, "loss": 0.0268, "step": 93130 }, { "epoch": 7.82, "learning_rate": 4.35673496808868e-06, "loss": 0.0016, "step": 93140 }, { "epoch": 7.82, "learning_rate": 4.3550554249244205e-06, "loss": 0.0279, "step": 93150 }, { "epoch": 7.82, "learning_rate": 4.353375881760162e-06, "loss": 0.0007, "step": 93160 }, { "epoch": 7.82, "learning_rate": 4.351696338595902e-06, "loss": 0.0015, "step": 93170 }, { "epoch": 7.82, "learning_rate": 4.350016795431643e-06, "loss": 0.0769, "step": 93180 }, { "epoch": 7.83, "learning_rate": 4.348337252267384e-06, "loss": 0.0011, "step": 93190 }, { "epoch": 7.83, "learning_rate": 4.346657709103125e-06, "loss": 0.0157, "step": 93200 }, { "epoch": 7.83, "learning_rate": 4.344978165938865e-06, "loss": 0.0892, "step": 93210 }, { "epoch": 7.83, "learning_rate": 4.3432986227746056e-06, "loss": 0.0086, "step": 93220 }, { "epoch": 7.83, "learning_rate": 4.341619079610346e-06, "loss": 0.0002, "step": 93230 }, { "epoch": 7.83, "learning_rate": 4.339939536446087e-06, "loss": 0.1909, "step": 93240 }, { "epoch": 7.83, "learning_rate": 4.338259993281828e-06, "loss": 0.0003, "step": 93250 }, { "epoch": 7.83, "learning_rate": 4.336580450117568e-06, "loss": 0.0398, "step": 93260 }, { "epoch": 7.83, "learning_rate": 4.334900906953309e-06, "loss": 0.0662, "step": 93270 }, { "epoch": 7.83, "learning_rate": 4.333221363789049e-06, "loss": 0.0006, "step": 93280 }, { "epoch": 7.83, "learning_rate": 4.331541820624791e-06, "loss": 0.0029, "step": 93290 }, { "epoch": 7.84, "learning_rate": 4.329862277460531e-06, "loss": 0.0003, "step": 93300 }, { "epoch": 7.84, "learning_rate": 4.3281827342962715e-06, "loss": 0.0766, "step": 93310 }, { "epoch": 7.84, "learning_rate": 4.326503191132013e-06, "loss": 0.0378, "step": 93320 }, { "epoch": 7.84, "learning_rate": 4.324823647967753e-06, "loss": 0.0002, "step": 93330 }, { "epoch": 7.84, "learning_rate": 4.323144104803494e-06, "loss": 0.0205, "step": 93340 }, { "epoch": 7.84, "learning_rate": 4.321464561639234e-06, "loss": 0.0129, "step": 93350 }, { "epoch": 7.84, "learning_rate": 4.319785018474975e-06, "loss": 0.0311, "step": 93360 }, { "epoch": 7.84, "learning_rate": 4.318105475310716e-06, "loss": 0.0107, "step": 93370 }, { "epoch": 7.84, "learning_rate": 4.3164259321464566e-06, "loss": 0.0432, "step": 93380 }, { "epoch": 7.84, "learning_rate": 4.314746388982197e-06, "loss": 0.0003, "step": 93390 }, { "epoch": 7.84, "learning_rate": 4.313066845817938e-06, "loss": 0.0315, "step": 93400 }, { "epoch": 7.84, "learning_rate": 4.311387302653679e-06, "loss": 0.0102, "step": 93410 }, { "epoch": 7.85, "learning_rate": 4.309707759489419e-06, "loss": 0.0364, "step": 93420 }, { "epoch": 7.85, "learning_rate": 4.30802821632516e-06, "loss": 0.0179, "step": 93430 }, { "epoch": 7.85, "learning_rate": 4.3063486731609e-06, "loss": 0.0001, "step": 93440 }, { "epoch": 7.85, "learning_rate": 4.304669129996642e-06, "loss": 0.0002, "step": 93450 }, { "epoch": 7.85, "learning_rate": 4.302989586832382e-06, "loss": 0.084, "step": 93460 }, { "epoch": 7.85, "learning_rate": 4.3013100436681226e-06, "loss": 0.061, "step": 93470 }, { "epoch": 7.85, "learning_rate": 4.299630500503864e-06, "loss": 0.0008, "step": 93480 }, { "epoch": 7.85, "learning_rate": 4.2979509573396035e-06, "loss": 0.0015, "step": 93490 }, { "epoch": 7.85, "learning_rate": 4.296271414175345e-06, "loss": 0.0624, "step": 93500 }, { "epoch": 7.85, "learning_rate": 4.294591871011085e-06, "loss": 0.0125, "step": 93510 }, { "epoch": 7.85, "learning_rate": 4.292912327846826e-06, "loss": 0.0301, "step": 93520 }, { "epoch": 7.85, "learning_rate": 4.291232784682567e-06, "loss": 0.0447, "step": 93530 }, { "epoch": 7.86, "learning_rate": 4.289553241518308e-06, "loss": 0.0623, "step": 93540 }, { "epoch": 7.86, "learning_rate": 4.287873698354048e-06, "loss": 0.068, "step": 93550 }, { "epoch": 7.86, "learning_rate": 4.286194155189789e-06, "loss": 0.0331, "step": 93560 }, { "epoch": 7.86, "learning_rate": 4.284514612025529e-06, "loss": 0.0026, "step": 93570 }, { "epoch": 7.86, "learning_rate": 4.28283506886127e-06, "loss": 0.0006, "step": 93580 }, { "epoch": 7.86, "learning_rate": 4.281155525697011e-06, "loss": 0.0222, "step": 93590 }, { "epoch": 7.86, "learning_rate": 4.279475982532751e-06, "loss": 0.0015, "step": 93600 }, { "epoch": 7.86, "learning_rate": 4.277796439368493e-06, "loss": 0.0278, "step": 93610 }, { "epoch": 7.86, "learning_rate": 4.276116896204232e-06, "loss": 0.0455, "step": 93620 }, { "epoch": 7.86, "learning_rate": 4.2744373530399736e-06, "loss": 0.0756, "step": 93630 }, { "epoch": 7.86, "learning_rate": 4.272757809875714e-06, "loss": 0.0385, "step": 93640 }, { "epoch": 7.86, "learning_rate": 4.2710782667114545e-06, "loss": 0.0005, "step": 93650 }, { "epoch": 7.87, "learning_rate": 4.269398723547196e-06, "loss": 0.03, "step": 93660 }, { "epoch": 7.87, "learning_rate": 4.267719180382936e-06, "loss": 0.0557, "step": 93670 }, { "epoch": 7.87, "learning_rate": 4.266039637218677e-06, "loss": 0.0667, "step": 93680 }, { "epoch": 7.87, "learning_rate": 4.264360094054418e-06, "loss": 0.0416, "step": 93690 }, { "epoch": 7.87, "learning_rate": 4.262680550890158e-06, "loss": 0.0003, "step": 93700 }, { "epoch": 7.87, "learning_rate": 4.261001007725899e-06, "loss": 0.0559, "step": 93710 }, { "epoch": 7.87, "learning_rate": 4.2593214645616396e-06, "loss": 0.0285, "step": 93720 }, { "epoch": 7.87, "learning_rate": 4.25764192139738e-06, "loss": 0.0248, "step": 93730 }, { "epoch": 7.87, "learning_rate": 4.255962378233121e-06, "loss": 0.0706, "step": 93740 }, { "epoch": 7.87, "learning_rate": 4.254282835068862e-06, "loss": 0.0229, "step": 93750 }, { "epoch": 7.87, "learning_rate": 4.252603291904602e-06, "loss": 0.0136, "step": 93760 }, { "epoch": 7.87, "learning_rate": 4.250923748740343e-06, "loss": 0.0747, "step": 93770 }, { "epoch": 7.88, "learning_rate": 4.249244205576083e-06, "loss": 0.0214, "step": 93780 }, { "epoch": 7.88, "learning_rate": 4.247564662411825e-06, "loss": 0.0374, "step": 93790 }, { "epoch": 7.88, "learning_rate": 4.245885119247565e-06, "loss": 0.0236, "step": 93800 }, { "epoch": 7.88, "learning_rate": 4.2442055760833055e-06, "loss": 0.0411, "step": 93810 }, { "epoch": 7.88, "learning_rate": 4.242526032919047e-06, "loss": 0.0568, "step": 93820 }, { "epoch": 7.88, "learning_rate": 4.2408464897547865e-06, "loss": 0.0531, "step": 93830 }, { "epoch": 7.88, "learning_rate": 4.239166946590528e-06, "loss": 0.0593, "step": 93840 }, { "epoch": 7.88, "learning_rate": 4.237487403426268e-06, "loss": 0.001, "step": 93850 }, { "epoch": 7.88, "learning_rate": 4.235807860262009e-06, "loss": 0.0631, "step": 93860 }, { "epoch": 7.88, "learning_rate": 4.23412831709775e-06, "loss": 0.0632, "step": 93870 }, { "epoch": 7.88, "learning_rate": 4.2324487739334906e-06, "loss": 0.0127, "step": 93880 }, { "epoch": 7.88, "learning_rate": 4.230769230769231e-06, "loss": 0.1223, "step": 93890 }, { "epoch": 7.89, "learning_rate": 4.229089687604972e-06, "loss": 0.0511, "step": 93900 }, { "epoch": 7.89, "learning_rate": 4.227410144440712e-06, "loss": 0.0011, "step": 93910 }, { "epoch": 7.89, "learning_rate": 4.225730601276453e-06, "loss": 0.0259, "step": 93920 }, { "epoch": 7.89, "learning_rate": 4.224051058112194e-06, "loss": 0.0034, "step": 93930 }, { "epoch": 7.89, "learning_rate": 4.222371514947934e-06, "loss": 0.0004, "step": 93940 }, { "epoch": 7.89, "learning_rate": 4.220691971783676e-06, "loss": 0.0199, "step": 93950 }, { "epoch": 7.89, "learning_rate": 4.219012428619415e-06, "loss": 0.0195, "step": 93960 }, { "epoch": 7.89, "learning_rate": 4.2173328854551566e-06, "loss": 0.0257, "step": 93970 }, { "epoch": 7.89, "learning_rate": 4.215653342290897e-06, "loss": 0.0008, "step": 93980 }, { "epoch": 7.89, "learning_rate": 4.2139737991266375e-06, "loss": 0.0553, "step": 93990 }, { "epoch": 7.89, "learning_rate": 4.212294255962379e-06, "loss": 0.0199, "step": 94000 }, { "epoch": 7.89, "learning_rate": 4.210614712798119e-06, "loss": 0.0002, "step": 94010 }, { "epoch": 7.9, "learning_rate": 4.20893516963386e-06, "loss": 0.0288, "step": 94020 }, { "epoch": 7.9, "learning_rate": 4.207255626469601e-06, "loss": 0.0005, "step": 94030 }, { "epoch": 7.9, "learning_rate": 4.205576083305341e-06, "loss": 0.0007, "step": 94040 }, { "epoch": 7.9, "learning_rate": 4.203896540141082e-06, "loss": 0.0017, "step": 94050 }, { "epoch": 7.9, "learning_rate": 4.2022169969768225e-06, "loss": 0.0297, "step": 94060 }, { "epoch": 7.9, "learning_rate": 4.200537453812563e-06, "loss": 0.011, "step": 94070 }, { "epoch": 7.9, "learning_rate": 4.198857910648304e-06, "loss": 0.0836, "step": 94080 }, { "epoch": 7.9, "learning_rate": 4.197178367484045e-06, "loss": 0.0025, "step": 94090 }, { "epoch": 7.9, "learning_rate": 4.195498824319785e-06, "loss": 0.0483, "step": 94100 }, { "epoch": 7.9, "learning_rate": 4.193819281155526e-06, "loss": 0.0775, "step": 94110 }, { "epoch": 7.9, "learning_rate": 4.192139737991266e-06, "loss": 0.042, "step": 94120 }, { "epoch": 7.9, "learning_rate": 4.1904601948270076e-06, "loss": 0.0209, "step": 94130 }, { "epoch": 7.91, "learning_rate": 4.188780651662748e-06, "loss": 0.0138, "step": 94140 }, { "epoch": 7.91, "learning_rate": 4.1871011084984885e-06, "loss": 0.0333, "step": 94150 }, { "epoch": 7.91, "learning_rate": 4.18542156533423e-06, "loss": 0.0231, "step": 94160 }, { "epoch": 7.91, "learning_rate": 4.18374202216997e-06, "loss": 0.029, "step": 94170 }, { "epoch": 7.91, "learning_rate": 4.182062479005711e-06, "loss": 0.0013, "step": 94180 }, { "epoch": 7.91, "learning_rate": 4.180382935841451e-06, "loss": 0.0469, "step": 94190 }, { "epoch": 7.91, "learning_rate": 4.178703392677192e-06, "loss": 0.0409, "step": 94200 }, { "epoch": 7.91, "learning_rate": 4.177023849512933e-06, "loss": 0.0358, "step": 94210 }, { "epoch": 7.91, "learning_rate": 4.1753443063486736e-06, "loss": 0.0545, "step": 94220 }, { "epoch": 7.91, "learning_rate": 4.173664763184414e-06, "loss": 0.0547, "step": 94230 }, { "epoch": 7.91, "learning_rate": 4.171985220020155e-06, "loss": 0.0026, "step": 94240 }, { "epoch": 7.91, "learning_rate": 4.170305676855895e-06, "loss": 0.03, "step": 94250 }, { "epoch": 7.92, "learning_rate": 4.168626133691636e-06, "loss": 0.0004, "step": 94260 }, { "epoch": 7.92, "learning_rate": 4.166946590527377e-06, "loss": 0.0332, "step": 94270 }, { "epoch": 7.92, "learning_rate": 4.165267047363117e-06, "loss": 0.0292, "step": 94280 }, { "epoch": 7.92, "learning_rate": 4.163587504198859e-06, "loss": 0.0373, "step": 94290 }, { "epoch": 7.92, "learning_rate": 4.161907961034599e-06, "loss": 0.0087, "step": 94300 }, { "epoch": 7.92, "learning_rate": 4.1602284178703395e-06, "loss": 0.0301, "step": 94310 }, { "epoch": 7.92, "learning_rate": 4.15854887470608e-06, "loss": 0.0004, "step": 94320 }, { "epoch": 7.92, "learning_rate": 4.1568693315418205e-06, "loss": 0.0006, "step": 94330 }, { "epoch": 7.92, "learning_rate": 4.155189788377562e-06, "loss": 0.0028, "step": 94340 }, { "epoch": 7.92, "learning_rate": 4.153510245213302e-06, "loss": 0.0028, "step": 94350 }, { "epoch": 7.92, "learning_rate": 4.151830702049043e-06, "loss": 0.0515, "step": 94360 }, { "epoch": 7.92, "learning_rate": 4.150151158884784e-06, "loss": 0.0553, "step": 94370 }, { "epoch": 7.93, "learning_rate": 4.1484716157205246e-06, "loss": 0.0353, "step": 94380 }, { "epoch": 7.93, "learning_rate": 4.146792072556265e-06, "loss": 0.0505, "step": 94390 }, { "epoch": 7.93, "learning_rate": 4.1451125293920055e-06, "loss": 0.0202, "step": 94400 }, { "epoch": 7.93, "learning_rate": 4.143432986227746e-06, "loss": 0.0003, "step": 94410 }, { "epoch": 7.93, "learning_rate": 4.141753443063487e-06, "loss": 0.0001, "step": 94420 }, { "epoch": 7.93, "learning_rate": 4.140073899899228e-06, "loss": 0.0007, "step": 94430 }, { "epoch": 7.93, "learning_rate": 4.138394356734968e-06, "loss": 0.0329, "step": 94440 }, { "epoch": 7.93, "learning_rate": 4.136714813570709e-06, "loss": 0.0391, "step": 94450 }, { "epoch": 7.93, "learning_rate": 4.13503527040645e-06, "loss": 0.0275, "step": 94460 }, { "epoch": 7.93, "learning_rate": 4.1333557272421906e-06, "loss": 0.0245, "step": 94470 }, { "epoch": 7.93, "learning_rate": 4.131676184077931e-06, "loss": 0.0315, "step": 94480 }, { "epoch": 7.94, "learning_rate": 4.1299966409136715e-06, "loss": 0.0014, "step": 94490 }, { "epoch": 7.94, "learning_rate": 4.128317097749413e-06, "loss": 0.0643, "step": 94500 }, { "epoch": 7.94, "learning_rate": 4.126637554585153e-06, "loss": 0.0614, "step": 94510 }, { "epoch": 7.94, "learning_rate": 4.124958011420894e-06, "loss": 0.0014, "step": 94520 }, { "epoch": 7.94, "learning_rate": 4.123278468256634e-06, "loss": 0.0523, "step": 94530 }, { "epoch": 7.94, "learning_rate": 4.121598925092376e-06, "loss": 0.0486, "step": 94540 }, { "epoch": 7.94, "learning_rate": 4.119919381928116e-06, "loss": 0.0066, "step": 94550 }, { "epoch": 7.94, "learning_rate": 4.1182398387638565e-06, "loss": 0.0257, "step": 94560 }, { "epoch": 7.94, "learning_rate": 4.116560295599597e-06, "loss": 0.0668, "step": 94570 }, { "epoch": 7.94, "learning_rate": 4.1148807524353375e-06, "loss": 0.0212, "step": 94580 }, { "epoch": 7.94, "learning_rate": 4.113201209271079e-06, "loss": 0.0647, "step": 94590 }, { "epoch": 7.94, "learning_rate": 4.111521666106819e-06, "loss": 0.0439, "step": 94600 }, { "epoch": 7.95, "learning_rate": 4.10984212294256e-06, "loss": 0.0013, "step": 94610 }, { "epoch": 7.95, "learning_rate": 4.108162579778301e-06, "loss": 0.1287, "step": 94620 }, { "epoch": 7.95, "learning_rate": 4.1064830366140416e-06, "loss": 0.0178, "step": 94630 }, { "epoch": 7.95, "learning_rate": 4.104803493449782e-06, "loss": 0.0108, "step": 94640 }, { "epoch": 7.95, "learning_rate": 4.1031239502855225e-06, "loss": 0.0824, "step": 94650 }, { "epoch": 7.95, "learning_rate": 4.101444407121263e-06, "loss": 0.0005, "step": 94660 }, { "epoch": 7.95, "learning_rate": 4.099764863957004e-06, "loss": 0.0308, "step": 94670 }, { "epoch": 7.95, "learning_rate": 4.098085320792745e-06, "loss": 0.048, "step": 94680 }, { "epoch": 7.95, "learning_rate": 4.096405777628485e-06, "loss": 0.0115, "step": 94690 }, { "epoch": 7.95, "learning_rate": 4.094726234464226e-06, "loss": 0.1073, "step": 94700 }, { "epoch": 7.95, "learning_rate": 4.093046691299967e-06, "loss": 0.0231, "step": 94710 }, { "epoch": 7.95, "learning_rate": 4.0913671481357076e-06, "loss": 0.028, "step": 94720 }, { "epoch": 7.96, "learning_rate": 4.089687604971448e-06, "loss": 0.0149, "step": 94730 }, { "epoch": 7.96, "learning_rate": 4.0880080618071885e-06, "loss": 0.0179, "step": 94740 }, { "epoch": 7.96, "learning_rate": 4.08632851864293e-06, "loss": 0.0353, "step": 94750 }, { "epoch": 7.96, "learning_rate": 4.08464897547867e-06, "loss": 0.0007, "step": 94760 }, { "epoch": 7.96, "learning_rate": 4.082969432314411e-06, "loss": 0.0206, "step": 94770 }, { "epoch": 7.96, "learning_rate": 4.081289889150151e-06, "loss": 0.0309, "step": 94780 }, { "epoch": 7.96, "learning_rate": 4.079610345985892e-06, "loss": 0.0512, "step": 94790 }, { "epoch": 7.96, "learning_rate": 4.077930802821633e-06, "loss": 0.0005, "step": 94800 }, { "epoch": 7.96, "learning_rate": 4.0762512596573735e-06, "loss": 0.019, "step": 94810 }, { "epoch": 7.96, "learning_rate": 4.074571716493114e-06, "loss": 0.0252, "step": 94820 }, { "epoch": 7.96, "learning_rate": 4.072892173328855e-06, "loss": 0.0011, "step": 94830 }, { "epoch": 7.96, "learning_rate": 4.071212630164596e-06, "loss": 0.0209, "step": 94840 }, { "epoch": 7.97, "learning_rate": 4.069533087000336e-06, "loss": 0.0004, "step": 94850 }, { "epoch": 7.97, "learning_rate": 4.067853543836077e-06, "loss": 0.0935, "step": 94860 }, { "epoch": 7.97, "learning_rate": 4.066174000671817e-06, "loss": 0.0299, "step": 94870 }, { "epoch": 7.97, "learning_rate": 4.0644944575075586e-06, "loss": 0.001, "step": 94880 }, { "epoch": 7.97, "learning_rate": 4.062814914343299e-06, "loss": 0.0333, "step": 94890 }, { "epoch": 7.97, "learning_rate": 4.0611353711790395e-06, "loss": 0.0934, "step": 94900 }, { "epoch": 7.97, "learning_rate": 4.059455828014781e-06, "loss": 0.0753, "step": 94910 }, { "epoch": 7.97, "learning_rate": 4.0577762848505205e-06, "loss": 0.0536, "step": 94920 }, { "epoch": 7.97, "learning_rate": 4.056096741686262e-06, "loss": 0.055, "step": 94930 }, { "epoch": 7.97, "learning_rate": 4.054417198522002e-06, "loss": 0.0018, "step": 94940 }, { "epoch": 7.97, "learning_rate": 4.052737655357743e-06, "loss": 0.0047, "step": 94950 }, { "epoch": 7.97, "learning_rate": 4.051058112193484e-06, "loss": 0.0315, "step": 94960 }, { "epoch": 7.98, "learning_rate": 4.0493785690292246e-06, "loss": 0.0063, "step": 94970 }, { "epoch": 7.98, "learning_rate": 4.047699025864965e-06, "loss": 0.0301, "step": 94980 }, { "epoch": 7.98, "learning_rate": 4.046019482700706e-06, "loss": 0.0202, "step": 94990 }, { "epoch": 7.98, "learning_rate": 4.044339939536446e-06, "loss": 0.0362, "step": 95000 }, { "epoch": 7.98, "learning_rate": 4.042660396372187e-06, "loss": 0.0007, "step": 95010 }, { "epoch": 7.98, "learning_rate": 4.040980853207928e-06, "loss": 0.0004, "step": 95020 }, { "epoch": 7.98, "learning_rate": 4.039301310043668e-06, "loss": 0.0299, "step": 95030 }, { "epoch": 7.98, "learning_rate": 4.03762176687941e-06, "loss": 0.0038, "step": 95040 }, { "epoch": 7.98, "learning_rate": 4.03594222371515e-06, "loss": 0.0429, "step": 95050 }, { "epoch": 7.98, "learning_rate": 4.0342626805508905e-06, "loss": 0.0002, "step": 95060 }, { "epoch": 7.98, "learning_rate": 4.032583137386631e-06, "loss": 0.0137, "step": 95070 }, { "epoch": 7.98, "learning_rate": 4.0309035942223715e-06, "loss": 0.0188, "step": 95080 }, { "epoch": 7.99, "learning_rate": 4.029224051058113e-06, "loss": 0.1435, "step": 95090 }, { "epoch": 7.99, "learning_rate": 4.027544507893853e-06, "loss": 0.0003, "step": 95100 }, { "epoch": 7.99, "learning_rate": 4.025864964729594e-06, "loss": 0.0015, "step": 95110 }, { "epoch": 7.99, "learning_rate": 4.024185421565335e-06, "loss": 0.0266, "step": 95120 }, { "epoch": 7.99, "learning_rate": 4.022505878401075e-06, "loss": 0.0019, "step": 95130 }, { "epoch": 7.99, "learning_rate": 4.020826335236816e-06, "loss": 0.0113, "step": 95140 }, { "epoch": 7.99, "learning_rate": 4.0191467920725565e-06, "loss": 0.0005, "step": 95150 }, { "epoch": 7.99, "learning_rate": 4.017467248908297e-06, "loss": 0.0399, "step": 95160 }, { "epoch": 7.99, "learning_rate": 4.015787705744038e-06, "loss": 0.0568, "step": 95170 }, { "epoch": 7.99, "learning_rate": 4.014108162579779e-06, "loss": 0.0036, "step": 95180 }, { "epoch": 7.99, "learning_rate": 4.012428619415519e-06, "loss": 0.0002, "step": 95190 }, { "epoch": 7.99, "learning_rate": 4.010749076251261e-06, "loss": 0.1284, "step": 95200 }, { "epoch": 8.0, "learning_rate": 4.009069533087e-06, "loss": 0.038, "step": 95210 }, { "epoch": 8.0, "learning_rate": 4.0073899899227416e-06, "loss": 0.049, "step": 95220 }, { "epoch": 8.0, "learning_rate": 4.005710446758482e-06, "loss": 0.0104, "step": 95230 }, { "epoch": 8.0, "learning_rate": 4.0040309035942225e-06, "loss": 0.0241, "step": 95240 }, { "epoch": 8.0, "learning_rate": 4.002351360429964e-06, "loss": 0.0225, "step": 95250 }, { "epoch": 8.0, "learning_rate": 4.0006718172657035e-06, "loss": 0.0564, "step": 95260 }, { "epoch": 8.0, "eval_FN": 492, "eval_FP": 422, "eval_TN": 21292, "eval_TP": 1610, "eval_accuracy": 0.9616224386966745, "eval_f1": 0.7789066279632317, "eval_loss": 0.2681463956832886, "eval_precision": 0.7923228346456693, "eval_recall": 0.7659372026641293, "eval_runtime": 163.867, "eval_samples_per_second": 145.337, "eval_steps_per_second": 9.087, "step": 95264 } ], "logging_steps": 10, "max_steps": 119080, "num_train_epochs": 10, "save_steps": 500, "total_flos": 4.0104018524504064e+17, "trial_name": null, "trial_params": null }