{ "best_metric": null, "best_model_checkpoint": null, "epoch": 1.9986675549633577, "global_step": 21000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 1e-05, "loss": 5.8577, "step": 10 }, { "epoch": 0.0, "learning_rate": 2e-05, "loss": 5.2479, "step": 20 }, { "epoch": 0.0, "learning_rate": 3e-05, "loss": 4.2581, "step": 30 }, { "epoch": 0.0, "learning_rate": 4e-05, "loss": 4.1526, "step": 40 }, { "epoch": 0.0, "learning_rate": 5e-05, "loss": 4.0411, "step": 50 }, { "epoch": 0.01, "learning_rate": 6e-05, "loss": 3.4283, "step": 60 }, { "epoch": 0.01, "learning_rate": 7e-05, "loss": 3.833, "step": 70 }, { "epoch": 0.01, "learning_rate": 8e-05, "loss": 3.7164, "step": 80 }, { "epoch": 0.01, "learning_rate": 9e-05, "loss": 3.715, "step": 90 }, { "epoch": 0.01, "learning_rate": 0.0001, "loss": 3.559, "step": 100 }, { "epoch": 0.01, "learning_rate": 9.996817415104547e-05, "loss": 2.9796, "step": 110 }, { "epoch": 0.01, "learning_rate": 9.993634830209096e-05, "loss": 3.2296, "step": 120 }, { "epoch": 0.01, "learning_rate": 9.990452245313644e-05, "loss": 2.6348, "step": 130 }, { "epoch": 0.01, "learning_rate": 9.987269660418193e-05, "loss": 3.2406, "step": 140 }, { "epoch": 0.01, "learning_rate": 9.98408707552274e-05, "loss": 2.8193, "step": 150 }, { "epoch": 0.02, "learning_rate": 9.980904490627288e-05, "loss": 3.0848, "step": 160 }, { "epoch": 0.02, "learning_rate": 9.977721905731836e-05, "loss": 2.8889, "step": 170 }, { "epoch": 0.02, "learning_rate": 9.974539320836383e-05, "loss": 3.2759, "step": 180 }, { "epoch": 0.02, "learning_rate": 9.971356735940932e-05, "loss": 2.7277, "step": 190 }, { "epoch": 0.02, "learning_rate": 9.96817415104548e-05, "loss": 2.5394, "step": 200 }, { "epoch": 0.02, "learning_rate": 9.964991566150027e-05, "loss": 2.9889, "step": 210 }, { "epoch": 0.02, "learning_rate": 9.961808981254575e-05, "loss": 2.8291, "step": 220 }, { "epoch": 0.02, "learning_rate": 9.958626396359122e-05, "loss": 2.9713, "step": 230 }, { "epoch": 0.02, "learning_rate": 9.955443811463672e-05, "loss": 2.8579, "step": 240 }, { "epoch": 0.02, "learning_rate": 9.952261226568219e-05, "loss": 2.6235, "step": 250 }, { "epoch": 0.02, "learning_rate": 9.949078641672766e-05, "loss": 2.7682, "step": 260 }, { "epoch": 0.03, "learning_rate": 9.945896056777315e-05, "loss": 2.5362, "step": 270 }, { "epoch": 0.03, "learning_rate": 9.942713471881863e-05, "loss": 2.5971, "step": 280 }, { "epoch": 0.03, "learning_rate": 9.939530886986411e-05, "loss": 2.89, "step": 290 }, { "epoch": 0.03, "learning_rate": 9.936348302090958e-05, "loss": 2.4912, "step": 300 }, { "epoch": 0.03, "learning_rate": 9.933165717195507e-05, "loss": 2.8739, "step": 310 }, { "epoch": 0.03, "learning_rate": 9.929983132300055e-05, "loss": 2.6042, "step": 320 }, { "epoch": 0.03, "learning_rate": 9.926800547404602e-05, "loss": 2.3941, "step": 330 }, { "epoch": 0.03, "learning_rate": 9.92361796250915e-05, "loss": 2.5959, "step": 340 }, { "epoch": 0.03, "learning_rate": 9.920435377613699e-05, "loss": 2.6849, "step": 350 }, { "epoch": 0.03, "learning_rate": 9.917252792718246e-05, "loss": 2.8489, "step": 360 }, { "epoch": 0.04, "learning_rate": 9.914070207822794e-05, "loss": 2.6433, "step": 370 }, { "epoch": 0.04, "learning_rate": 9.910887622927341e-05, "loss": 2.5534, "step": 380 }, { "epoch": 0.04, "learning_rate": 9.907705038031891e-05, "loss": 2.3736, "step": 390 }, { "epoch": 0.04, "learning_rate": 9.904522453136438e-05, "loss": 2.8226, "step": 400 }, { "epoch": 0.04, "learning_rate": 9.901339868240985e-05, "loss": 2.468, "step": 410 }, { "epoch": 0.04, "learning_rate": 9.898157283345533e-05, "loss": 2.298, "step": 420 }, { "epoch": 0.04, "learning_rate": 9.894974698450082e-05, "loss": 2.5418, "step": 430 }, { "epoch": 0.04, "learning_rate": 9.89179211355463e-05, "loss": 2.3139, "step": 440 }, { "epoch": 0.04, "learning_rate": 9.888609528659177e-05, "loss": 2.5035, "step": 450 }, { "epoch": 0.04, "learning_rate": 9.885426943763725e-05, "loss": 2.8296, "step": 460 }, { "epoch": 0.04, "learning_rate": 9.882244358868274e-05, "loss": 2.1434, "step": 470 }, { "epoch": 0.05, "learning_rate": 9.879061773972821e-05, "loss": 2.2989, "step": 480 }, { "epoch": 0.05, "learning_rate": 9.875879189077369e-05, "loss": 2.7385, "step": 490 }, { "epoch": 0.05, "learning_rate": 9.872696604181917e-05, "loss": 2.5565, "step": 500 }, { "epoch": 0.05, "learning_rate": 9.869514019286466e-05, "loss": 2.4335, "step": 510 }, { "epoch": 0.05, "learning_rate": 9.866331434391013e-05, "loss": 2.9056, "step": 520 }, { "epoch": 0.05, "learning_rate": 9.86314884949556e-05, "loss": 2.392, "step": 530 }, { "epoch": 0.05, "learning_rate": 9.85996626460011e-05, "loss": 2.8014, "step": 540 }, { "epoch": 0.05, "learning_rate": 9.856783679704657e-05, "loss": 2.1574, "step": 550 }, { "epoch": 0.05, "learning_rate": 9.853601094809205e-05, "loss": 2.3031, "step": 560 }, { "epoch": 0.05, "learning_rate": 9.850418509913752e-05, "loss": 2.3997, "step": 570 }, { "epoch": 0.06, "learning_rate": 9.8472359250183e-05, "loss": 2.4631, "step": 580 }, { "epoch": 0.06, "learning_rate": 9.844053340122849e-05, "loss": 2.228, "step": 590 }, { "epoch": 0.06, "learning_rate": 9.840870755227396e-05, "loss": 2.375, "step": 600 }, { "epoch": 0.06, "learning_rate": 9.837688170331944e-05, "loss": 2.4447, "step": 610 }, { "epoch": 0.06, "learning_rate": 9.834505585436492e-05, "loss": 2.587, "step": 620 }, { "epoch": 0.06, "learning_rate": 9.83132300054104e-05, "loss": 2.1196, "step": 630 }, { "epoch": 0.06, "learning_rate": 9.828140415645588e-05, "loss": 2.5909, "step": 640 }, { "epoch": 0.06, "learning_rate": 9.824957830750136e-05, "loss": 2.4393, "step": 650 }, { "epoch": 0.06, "learning_rate": 9.821775245854684e-05, "loss": 2.8212, "step": 660 }, { "epoch": 0.06, "learning_rate": 9.818592660959231e-05, "loss": 2.4206, "step": 670 }, { "epoch": 0.06, "learning_rate": 9.815410076063778e-05, "loss": 2.2936, "step": 680 }, { "epoch": 0.07, "learning_rate": 9.812227491168328e-05, "loss": 2.3386, "step": 690 }, { "epoch": 0.07, "learning_rate": 9.809044906272875e-05, "loss": 1.9638, "step": 700 }, { "epoch": 0.07, "learning_rate": 9.805862321377424e-05, "loss": 2.5946, "step": 710 }, { "epoch": 0.07, "learning_rate": 9.80267973648197e-05, "loss": 2.4551, "step": 720 }, { "epoch": 0.07, "learning_rate": 9.799497151586519e-05, "loss": 2.6737, "step": 730 }, { "epoch": 0.07, "learning_rate": 9.796314566691067e-05, "loss": 2.1145, "step": 740 }, { "epoch": 0.07, "learning_rate": 9.793131981795614e-05, "loss": 2.1091, "step": 750 }, { "epoch": 0.07, "learning_rate": 9.789949396900163e-05, "loss": 2.7712, "step": 760 }, { "epoch": 0.07, "learning_rate": 9.786766812004711e-05, "loss": 2.1544, "step": 770 }, { "epoch": 0.07, "learning_rate": 9.783584227109258e-05, "loss": 2.3773, "step": 780 }, { "epoch": 0.08, "learning_rate": 9.780401642213806e-05, "loss": 2.1907, "step": 790 }, { "epoch": 0.08, "learning_rate": 9.777219057318355e-05, "loss": 2.6661, "step": 800 }, { "epoch": 0.08, "learning_rate": 9.774036472422903e-05, "loss": 2.238, "step": 810 }, { "epoch": 0.08, "learning_rate": 9.77085388752745e-05, "loss": 1.7985, "step": 820 }, { "epoch": 0.08, "learning_rate": 9.767671302631997e-05, "loss": 2.1975, "step": 830 }, { "epoch": 0.08, "learning_rate": 9.764488717736547e-05, "loss": 2.1023, "step": 840 }, { "epoch": 0.08, "learning_rate": 9.761306132841094e-05, "loss": 2.4209, "step": 850 }, { "epoch": 0.08, "learning_rate": 9.758123547945642e-05, "loss": 2.4297, "step": 860 }, { "epoch": 0.08, "learning_rate": 9.754940963050189e-05, "loss": 1.9522, "step": 870 }, { "epoch": 0.08, "learning_rate": 9.751758378154738e-05, "loss": 2.3907, "step": 880 }, { "epoch": 0.08, "learning_rate": 9.748575793259286e-05, "loss": 2.206, "step": 890 }, { "epoch": 0.09, "learning_rate": 9.745393208363833e-05, "loss": 1.9964, "step": 900 }, { "epoch": 0.09, "learning_rate": 9.742210623468381e-05, "loss": 2.6844, "step": 910 }, { "epoch": 0.09, "learning_rate": 9.73902803857293e-05, "loss": 2.2864, "step": 920 }, { "epoch": 0.09, "learning_rate": 9.735845453677477e-05, "loss": 2.4212, "step": 930 }, { "epoch": 0.09, "learning_rate": 9.732662868782025e-05, "loss": 2.463, "step": 940 }, { "epoch": 0.09, "learning_rate": 9.729480283886573e-05, "loss": 2.2846, "step": 950 }, { "epoch": 0.09, "learning_rate": 9.726297698991122e-05, "loss": 2.3168, "step": 960 }, { "epoch": 0.09, "learning_rate": 9.723115114095669e-05, "loss": 2.205, "step": 970 }, { "epoch": 0.09, "learning_rate": 9.719932529200216e-05, "loss": 2.6097, "step": 980 }, { "epoch": 0.09, "learning_rate": 9.716749944304766e-05, "loss": 2.5665, "step": 990 }, { "epoch": 0.1, "learning_rate": 9.713567359409313e-05, "loss": 2.0707, "step": 1000 }, { "epoch": 0.1, "learning_rate": 9.710384774513861e-05, "loss": 2.3376, "step": 1010 }, { "epoch": 0.1, "learning_rate": 9.707202189618408e-05, "loss": 2.1752, "step": 1020 }, { "epoch": 0.1, "learning_rate": 9.704019604722956e-05, "loss": 1.9965, "step": 1030 }, { "epoch": 0.1, "learning_rate": 9.700837019827505e-05, "loss": 2.3768, "step": 1040 }, { "epoch": 0.1, "learning_rate": 9.697654434932052e-05, "loss": 1.8873, "step": 1050 }, { "epoch": 0.1, "learning_rate": 9.6944718500366e-05, "loss": 2.1792, "step": 1060 }, { "epoch": 0.1, "learning_rate": 9.691289265141148e-05, "loss": 2.0115, "step": 1070 }, { "epoch": 0.1, "learning_rate": 9.688106680245697e-05, "loss": 2.3, "step": 1080 }, { "epoch": 0.1, "learning_rate": 9.684924095350244e-05, "loss": 2.3574, "step": 1090 }, { "epoch": 0.1, "learning_rate": 9.681741510454792e-05, "loss": 2.2998, "step": 1100 }, { "epoch": 0.11, "learning_rate": 9.67855892555934e-05, "loss": 2.2051, "step": 1110 }, { "epoch": 0.11, "learning_rate": 9.675376340663888e-05, "loss": 2.5525, "step": 1120 }, { "epoch": 0.11, "learning_rate": 9.672193755768436e-05, "loss": 2.5324, "step": 1130 }, { "epoch": 0.11, "learning_rate": 9.669011170872984e-05, "loss": 2.4017, "step": 1140 }, { "epoch": 0.11, "learning_rate": 9.665828585977531e-05, "loss": 2.2123, "step": 1150 }, { "epoch": 0.11, "learning_rate": 9.66264600108208e-05, "loss": 2.0349, "step": 1160 }, { "epoch": 0.11, "learning_rate": 9.659463416186627e-05, "loss": 1.7593, "step": 1170 }, { "epoch": 0.11, "learning_rate": 9.656280831291175e-05, "loss": 2.4243, "step": 1180 }, { "epoch": 0.11, "learning_rate": 9.653098246395723e-05, "loss": 2.3423, "step": 1190 }, { "epoch": 0.11, "learning_rate": 9.64991566150027e-05, "loss": 1.8123, "step": 1200 }, { "epoch": 0.12, "learning_rate": 9.646733076604819e-05, "loss": 2.0992, "step": 1210 }, { "epoch": 0.12, "learning_rate": 9.643550491709367e-05, "loss": 2.5563, "step": 1220 }, { "epoch": 0.12, "learning_rate": 9.640367906813915e-05, "loss": 2.3245, "step": 1230 }, { "epoch": 0.12, "learning_rate": 9.637185321918462e-05, "loss": 1.7916, "step": 1240 }, { "epoch": 0.12, "learning_rate": 9.63400273702301e-05, "loss": 1.4747, "step": 1250 }, { "epoch": 0.12, "learning_rate": 9.630820152127559e-05, "loss": 2.0067, "step": 1260 }, { "epoch": 0.12, "learning_rate": 9.627637567232106e-05, "loss": 1.9419, "step": 1270 }, { "epoch": 0.12, "learning_rate": 9.624454982336655e-05, "loss": 2.2326, "step": 1280 }, { "epoch": 0.12, "learning_rate": 9.621272397441202e-05, "loss": 2.4721, "step": 1290 }, { "epoch": 0.12, "learning_rate": 9.61808981254575e-05, "loss": 1.9018, "step": 1300 }, { "epoch": 0.12, "learning_rate": 9.614907227650298e-05, "loss": 2.0016, "step": 1310 }, { "epoch": 0.13, "learning_rate": 9.611724642754845e-05, "loss": 1.8423, "step": 1320 }, { "epoch": 0.13, "learning_rate": 9.608542057859394e-05, "loss": 2.4551, "step": 1330 }, { "epoch": 0.13, "learning_rate": 9.605359472963942e-05, "loss": 2.1678, "step": 1340 }, { "epoch": 0.13, "learning_rate": 9.602176888068489e-05, "loss": 2.2086, "step": 1350 }, { "epoch": 0.13, "learning_rate": 9.598994303173037e-05, "loss": 2.5018, "step": 1360 }, { "epoch": 0.13, "learning_rate": 9.595811718277586e-05, "loss": 1.8025, "step": 1370 }, { "epoch": 0.13, "learning_rate": 9.592629133382134e-05, "loss": 2.0562, "step": 1380 }, { "epoch": 0.13, "learning_rate": 9.589446548486681e-05, "loss": 2.4341, "step": 1390 }, { "epoch": 0.13, "learning_rate": 9.586263963591228e-05, "loss": 2.0408, "step": 1400 }, { "epoch": 0.13, "learning_rate": 9.583081378695778e-05, "loss": 1.923, "step": 1410 }, { "epoch": 0.14, "learning_rate": 9.579898793800325e-05, "loss": 2.232, "step": 1420 }, { "epoch": 0.14, "learning_rate": 9.576716208904873e-05, "loss": 1.8952, "step": 1430 }, { "epoch": 0.14, "learning_rate": 9.57353362400942e-05, "loss": 2.0362, "step": 1440 }, { "epoch": 0.14, "learning_rate": 9.570351039113969e-05, "loss": 2.3579, "step": 1450 }, { "epoch": 0.14, "learning_rate": 9.567168454218517e-05, "loss": 2.3189, "step": 1460 }, { "epoch": 0.14, "learning_rate": 9.563985869323064e-05, "loss": 2.0279, "step": 1470 }, { "epoch": 0.14, "learning_rate": 9.560803284427612e-05, "loss": 2.206, "step": 1480 }, { "epoch": 0.14, "learning_rate": 9.557620699532161e-05, "loss": 2.0405, "step": 1490 }, { "epoch": 0.14, "learning_rate": 9.554438114636708e-05, "loss": 2.0556, "step": 1500 }, { "epoch": 0.14, "learning_rate": 9.551255529741256e-05, "loss": 2.1394, "step": 1510 }, { "epoch": 0.14, "learning_rate": 9.548072944845804e-05, "loss": 2.0859, "step": 1520 }, { "epoch": 0.15, "learning_rate": 9.544890359950353e-05, "loss": 1.9614, "step": 1530 }, { "epoch": 0.15, "learning_rate": 9.5417077750549e-05, "loss": 2.0116, "step": 1540 }, { "epoch": 0.15, "learning_rate": 9.538525190159447e-05, "loss": 1.694, "step": 1550 }, { "epoch": 0.15, "learning_rate": 9.535342605263997e-05, "loss": 2.0689, "step": 1560 }, { "epoch": 0.15, "learning_rate": 9.532160020368544e-05, "loss": 2.4515, "step": 1570 }, { "epoch": 0.15, "learning_rate": 9.528977435473092e-05, "loss": 2.0836, "step": 1580 }, { "epoch": 0.15, "learning_rate": 9.525794850577639e-05, "loss": 2.1763, "step": 1590 }, { "epoch": 0.15, "learning_rate": 9.522612265682187e-05, "loss": 2.3392, "step": 1600 }, { "epoch": 0.15, "learning_rate": 9.519429680786736e-05, "loss": 2.2552, "step": 1610 }, { "epoch": 0.15, "learning_rate": 9.516247095891283e-05, "loss": 1.7769, "step": 1620 }, { "epoch": 0.16, "learning_rate": 9.513064510995831e-05, "loss": 2.3103, "step": 1630 }, { "epoch": 0.16, "learning_rate": 9.50988192610038e-05, "loss": 2.1365, "step": 1640 }, { "epoch": 0.16, "learning_rate": 9.506699341204926e-05, "loss": 2.4357, "step": 1650 }, { "epoch": 0.16, "learning_rate": 9.503516756309475e-05, "loss": 1.8809, "step": 1660 }, { "epoch": 0.16, "learning_rate": 9.500334171414023e-05, "loss": 2.2216, "step": 1670 }, { "epoch": 0.16, "learning_rate": 9.497151586518572e-05, "loss": 2.1124, "step": 1680 }, { "epoch": 0.16, "learning_rate": 9.493969001623119e-05, "loss": 1.9944, "step": 1690 }, { "epoch": 0.16, "learning_rate": 9.490786416727667e-05, "loss": 2.2696, "step": 1700 }, { "epoch": 0.16, "learning_rate": 9.487603831832215e-05, "loss": 1.9576, "step": 1710 }, { "epoch": 0.16, "learning_rate": 9.484421246936762e-05, "loss": 1.8947, "step": 1720 }, { "epoch": 0.16, "learning_rate": 9.48123866204131e-05, "loss": 2.0351, "step": 1730 }, { "epoch": 0.17, "learning_rate": 9.478056077145858e-05, "loss": 2.291, "step": 1740 }, { "epoch": 0.17, "learning_rate": 9.474873492250407e-05, "loss": 1.915, "step": 1750 }, { "epoch": 0.17, "learning_rate": 9.471690907354954e-05, "loss": 2.1551, "step": 1760 }, { "epoch": 0.17, "learning_rate": 9.468508322459501e-05, "loss": 1.8664, "step": 1770 }, { "epoch": 0.17, "learning_rate": 9.46532573756405e-05, "loss": 2.5013, "step": 1780 }, { "epoch": 0.17, "learning_rate": 9.462143152668598e-05, "loss": 2.2391, "step": 1790 }, { "epoch": 0.17, "learning_rate": 9.458960567773146e-05, "loss": 1.9795, "step": 1800 }, { "epoch": 0.17, "learning_rate": 9.455777982877693e-05, "loss": 1.8689, "step": 1810 }, { "epoch": 0.17, "learning_rate": 9.452595397982242e-05, "loss": 2.4631, "step": 1820 }, { "epoch": 0.17, "learning_rate": 9.44941281308679e-05, "loss": 2.1215, "step": 1830 }, { "epoch": 0.18, "learning_rate": 9.446230228191337e-05, "loss": 2.169, "step": 1840 }, { "epoch": 0.18, "learning_rate": 9.443047643295886e-05, "loss": 1.9249, "step": 1850 }, { "epoch": 0.18, "learning_rate": 9.439865058400434e-05, "loss": 2.1031, "step": 1860 }, { "epoch": 0.18, "learning_rate": 9.436682473504981e-05, "loss": 2.0699, "step": 1870 }, { "epoch": 0.18, "learning_rate": 9.433499888609529e-05, "loss": 2.0062, "step": 1880 }, { "epoch": 0.18, "learning_rate": 9.430317303714076e-05, "loss": 1.988, "step": 1890 }, { "epoch": 0.18, "learning_rate": 9.427134718818626e-05, "loss": 1.7972, "step": 1900 }, { "epoch": 0.18, "learning_rate": 9.423952133923173e-05, "loss": 1.7263, "step": 1910 }, { "epoch": 0.18, "learning_rate": 9.42076954902772e-05, "loss": 1.7781, "step": 1920 }, { "epoch": 0.18, "learning_rate": 9.417586964132268e-05, "loss": 1.784, "step": 1930 }, { "epoch": 0.18, "learning_rate": 9.414404379236817e-05, "loss": 2.1332, "step": 1940 }, { "epoch": 0.19, "learning_rate": 9.411221794341365e-05, "loss": 2.1236, "step": 1950 }, { "epoch": 0.19, "learning_rate": 9.408039209445912e-05, "loss": 2.1588, "step": 1960 }, { "epoch": 0.19, "learning_rate": 9.40485662455046e-05, "loss": 2.117, "step": 1970 }, { "epoch": 0.19, "learning_rate": 9.401674039655009e-05, "loss": 1.7931, "step": 1980 }, { "epoch": 0.19, "learning_rate": 9.398491454759556e-05, "loss": 2.1229, "step": 1990 }, { "epoch": 0.19, "learning_rate": 9.395308869864104e-05, "loss": 1.772, "step": 2000 }, { "epoch": 0.19, "learning_rate": 9.392126284968653e-05, "loss": 1.8689, "step": 2010 }, { "epoch": 0.19, "learning_rate": 9.3889437000732e-05, "loss": 2.1535, "step": 2020 }, { "epoch": 0.19, "learning_rate": 9.385761115177748e-05, "loss": 1.9323, "step": 2030 }, { "epoch": 0.19, "learning_rate": 9.382578530282295e-05, "loss": 2.2067, "step": 2040 }, { "epoch": 0.2, "learning_rate": 9.379395945386845e-05, "loss": 2.1581, "step": 2050 }, { "epoch": 0.2, "learning_rate": 9.376213360491392e-05, "loss": 2.0287, "step": 2060 }, { "epoch": 0.2, "learning_rate": 9.373030775595939e-05, "loss": 2.1503, "step": 2070 }, { "epoch": 0.2, "learning_rate": 9.369848190700487e-05, "loss": 1.8953, "step": 2080 }, { "epoch": 0.2, "learning_rate": 9.366665605805035e-05, "loss": 2.4307, "step": 2090 }, { "epoch": 0.2, "learning_rate": 9.363483020909584e-05, "loss": 2.0636, "step": 2100 }, { "epoch": 0.2, "learning_rate": 9.360300436014131e-05, "loss": 2.0434, "step": 2110 }, { "epoch": 0.2, "learning_rate": 9.357117851118679e-05, "loss": 1.9077, "step": 2120 }, { "epoch": 0.2, "learning_rate": 9.353935266223228e-05, "loss": 2.2045, "step": 2130 }, { "epoch": 0.2, "learning_rate": 9.350752681327775e-05, "loss": 1.8585, "step": 2140 }, { "epoch": 0.2, "learning_rate": 9.347570096432323e-05, "loss": 1.6934, "step": 2150 }, { "epoch": 0.21, "learning_rate": 9.344387511536871e-05, "loss": 1.8759, "step": 2160 }, { "epoch": 0.21, "learning_rate": 9.341204926641418e-05, "loss": 2.2663, "step": 2170 }, { "epoch": 0.21, "learning_rate": 9.338022341745967e-05, "loss": 1.7734, "step": 2180 }, { "epoch": 0.21, "learning_rate": 9.334839756850514e-05, "loss": 1.941, "step": 2190 }, { "epoch": 0.21, "learning_rate": 9.331657171955063e-05, "loss": 2.2826, "step": 2200 }, { "epoch": 0.21, "learning_rate": 9.32847458705961e-05, "loss": 2.4277, "step": 2210 }, { "epoch": 0.21, "learning_rate": 9.325292002164157e-05, "loss": 2.0279, "step": 2220 }, { "epoch": 0.21, "learning_rate": 9.322109417268706e-05, "loss": 1.738, "step": 2230 }, { "epoch": 0.21, "learning_rate": 9.318926832373254e-05, "loss": 2.3177, "step": 2240 }, { "epoch": 0.21, "learning_rate": 9.315744247477803e-05, "loss": 2.0615, "step": 2250 }, { "epoch": 0.22, "learning_rate": 9.31256166258235e-05, "loss": 1.788, "step": 2260 }, { "epoch": 0.22, "learning_rate": 9.309379077686898e-05, "loss": 2.0254, "step": 2270 }, { "epoch": 0.22, "learning_rate": 9.306196492791446e-05, "loss": 1.9799, "step": 2280 }, { "epoch": 0.22, "learning_rate": 9.303013907895993e-05, "loss": 1.9327, "step": 2290 }, { "epoch": 0.22, "learning_rate": 9.299831323000542e-05, "loss": 2.0215, "step": 2300 }, { "epoch": 0.22, "learning_rate": 9.296648738105089e-05, "loss": 1.7893, "step": 2310 }, { "epoch": 0.22, "learning_rate": 9.293466153209637e-05, "loss": 1.5906, "step": 2320 }, { "epoch": 0.22, "learning_rate": 9.290283568314185e-05, "loss": 1.9828, "step": 2330 }, { "epoch": 0.22, "learning_rate": 9.287100983418732e-05, "loss": 1.8949, "step": 2340 }, { "epoch": 0.22, "learning_rate": 9.283918398523281e-05, "loss": 2.0422, "step": 2350 }, { "epoch": 0.22, "learning_rate": 9.280735813627829e-05, "loss": 1.9592, "step": 2360 }, { "epoch": 0.23, "learning_rate": 9.277553228732377e-05, "loss": 2.0711, "step": 2370 }, { "epoch": 0.23, "learning_rate": 9.274370643836924e-05, "loss": 1.9374, "step": 2380 }, { "epoch": 0.23, "learning_rate": 9.271188058941473e-05, "loss": 2.0565, "step": 2390 }, { "epoch": 0.23, "learning_rate": 9.268005474046021e-05, "loss": 2.1121, "step": 2400 }, { "epoch": 0.23, "learning_rate": 9.264822889150568e-05, "loss": 2.263, "step": 2410 }, { "epoch": 0.23, "learning_rate": 9.261640304255117e-05, "loss": 1.8936, "step": 2420 }, { "epoch": 0.23, "learning_rate": 9.258457719359665e-05, "loss": 2.4082, "step": 2430 }, { "epoch": 0.23, "learning_rate": 9.255275134464212e-05, "loss": 2.1083, "step": 2440 }, { "epoch": 0.23, "learning_rate": 9.25209254956876e-05, "loss": 1.7903, "step": 2450 }, { "epoch": 0.23, "learning_rate": 9.248909964673307e-05, "loss": 2.2809, "step": 2460 }, { "epoch": 0.24, "learning_rate": 9.245727379777857e-05, "loss": 1.9134, "step": 2470 }, { "epoch": 0.24, "learning_rate": 9.242544794882404e-05, "loss": 1.8962, "step": 2480 }, { "epoch": 0.24, "learning_rate": 9.239362209986951e-05, "loss": 1.9613, "step": 2490 }, { "epoch": 0.24, "learning_rate": 9.2361796250915e-05, "loss": 2.0229, "step": 2500 }, { "epoch": 0.24, "learning_rate": 9.232997040196048e-05, "loss": 2.4332, "step": 2510 }, { "epoch": 0.24, "learning_rate": 9.229814455300596e-05, "loss": 2.3423, "step": 2520 }, { "epoch": 0.24, "learning_rate": 9.226631870405143e-05, "loss": 1.9032, "step": 2530 }, { "epoch": 0.24, "learning_rate": 9.223449285509692e-05, "loss": 1.7481, "step": 2540 }, { "epoch": 0.24, "learning_rate": 9.22026670061424e-05, "loss": 1.9051, "step": 2550 }, { "epoch": 0.24, "learning_rate": 9.217084115718787e-05, "loss": 1.7612, "step": 2560 }, { "epoch": 0.24, "learning_rate": 9.213901530823335e-05, "loss": 2.3907, "step": 2570 }, { "epoch": 0.25, "learning_rate": 9.210718945927884e-05, "loss": 1.8452, "step": 2580 }, { "epoch": 0.25, "learning_rate": 9.20753636103243e-05, "loss": 2.0806, "step": 2590 }, { "epoch": 0.25, "learning_rate": 9.204353776136979e-05, "loss": 1.6906, "step": 2600 }, { "epoch": 0.25, "learning_rate": 9.201171191241526e-05, "loss": 2.0094, "step": 2610 }, { "epoch": 0.25, "learning_rate": 9.197988606346076e-05, "loss": 2.2129, "step": 2620 }, { "epoch": 0.25, "learning_rate": 9.194806021450623e-05, "loss": 2.297, "step": 2630 }, { "epoch": 0.25, "learning_rate": 9.19162343655517e-05, "loss": 1.969, "step": 2640 }, { "epoch": 0.25, "learning_rate": 9.188440851659718e-05, "loss": 1.837, "step": 2650 }, { "epoch": 0.25, "learning_rate": 9.185258266764266e-05, "loss": 1.9689, "step": 2660 }, { "epoch": 0.25, "learning_rate": 9.182075681868815e-05, "loss": 1.8176, "step": 2670 }, { "epoch": 0.26, "learning_rate": 9.178893096973362e-05, "loss": 2.0166, "step": 2680 }, { "epoch": 0.26, "learning_rate": 9.17571051207791e-05, "loss": 1.7734, "step": 2690 }, { "epoch": 0.26, "learning_rate": 9.172527927182459e-05, "loss": 1.6355, "step": 2700 }, { "epoch": 0.26, "learning_rate": 9.169345342287006e-05, "loss": 1.4228, "step": 2710 }, { "epoch": 0.26, "learning_rate": 9.166162757391554e-05, "loss": 1.9741, "step": 2720 }, { "epoch": 0.26, "learning_rate": 9.162980172496102e-05, "loss": 1.849, "step": 2730 }, { "epoch": 0.26, "learning_rate": 9.159797587600649e-05, "loss": 1.8072, "step": 2740 }, { "epoch": 0.26, "learning_rate": 9.156615002705198e-05, "loss": 1.8057, "step": 2750 }, { "epoch": 0.26, "learning_rate": 9.153432417809745e-05, "loss": 1.766, "step": 2760 }, { "epoch": 0.26, "learning_rate": 9.150249832914294e-05, "loss": 1.8951, "step": 2770 }, { "epoch": 0.26, "learning_rate": 9.147067248018841e-05, "loss": 1.5905, "step": 2780 }, { "epoch": 0.27, "learning_rate": 9.143884663123388e-05, "loss": 1.8121, "step": 2790 }, { "epoch": 0.27, "learning_rate": 9.140702078227937e-05, "loss": 2.4688, "step": 2800 }, { "epoch": 0.27, "learning_rate": 9.137519493332485e-05, "loss": 1.8452, "step": 2810 }, { "epoch": 0.27, "learning_rate": 9.134336908437034e-05, "loss": 2.1774, "step": 2820 }, { "epoch": 0.27, "learning_rate": 9.13115432354158e-05, "loss": 2.1027, "step": 2830 }, { "epoch": 0.27, "learning_rate": 9.127971738646129e-05, "loss": 1.7269, "step": 2840 }, { "epoch": 0.27, "learning_rate": 9.124789153750677e-05, "loss": 1.5584, "step": 2850 }, { "epoch": 0.27, "learning_rate": 9.121606568855224e-05, "loss": 1.9226, "step": 2860 }, { "epoch": 0.27, "learning_rate": 9.118423983959773e-05, "loss": 1.9731, "step": 2870 }, { "epoch": 0.27, "learning_rate": 9.115241399064321e-05, "loss": 1.9055, "step": 2880 }, { "epoch": 0.28, "learning_rate": 9.112058814168868e-05, "loss": 2.1257, "step": 2890 }, { "epoch": 0.28, "learning_rate": 9.108876229273416e-05, "loss": 2.1161, "step": 2900 }, { "epoch": 0.28, "learning_rate": 9.105693644377963e-05, "loss": 2.0938, "step": 2910 }, { "epoch": 0.28, "learning_rate": 9.102511059482513e-05, "loss": 1.5036, "step": 2920 }, { "epoch": 0.28, "learning_rate": 9.09932847458706e-05, "loss": 2.4857, "step": 2930 }, { "epoch": 0.28, "learning_rate": 9.096145889691607e-05, "loss": 1.5093, "step": 2940 }, { "epoch": 0.28, "learning_rate": 9.092963304796155e-05, "loss": 2.1482, "step": 2950 }, { "epoch": 0.28, "learning_rate": 9.089780719900704e-05, "loss": 2.0191, "step": 2960 }, { "epoch": 0.28, "learning_rate": 9.086598135005252e-05, "loss": 1.9049, "step": 2970 }, { "epoch": 0.28, "learning_rate": 9.083415550109799e-05, "loss": 1.6495, "step": 2980 }, { "epoch": 0.28, "learning_rate": 9.080232965214348e-05, "loss": 2.2134, "step": 2990 }, { "epoch": 0.29, "learning_rate": 9.077050380318896e-05, "loss": 2.6072, "step": 3000 }, { "epoch": 0.29, "learning_rate": 9.073867795423443e-05, "loss": 1.7293, "step": 3010 }, { "epoch": 0.29, "learning_rate": 9.070685210527991e-05, "loss": 1.7487, "step": 3020 }, { "epoch": 0.29, "learning_rate": 9.06750262563254e-05, "loss": 2.1817, "step": 3030 }, { "epoch": 0.29, "learning_rate": 9.064320040737088e-05, "loss": 2.0928, "step": 3040 }, { "epoch": 0.29, "learning_rate": 9.061137455841635e-05, "loss": 1.8595, "step": 3050 }, { "epoch": 0.29, "learning_rate": 9.057954870946182e-05, "loss": 2.2285, "step": 3060 }, { "epoch": 0.29, "learning_rate": 9.054772286050732e-05, "loss": 2.1055, "step": 3070 }, { "epoch": 0.29, "learning_rate": 9.051589701155279e-05, "loss": 1.9423, "step": 3080 }, { "epoch": 0.29, "learning_rate": 9.048407116259827e-05, "loss": 1.9643, "step": 3090 }, { "epoch": 0.3, "learning_rate": 9.045224531364374e-05, "loss": 2.5573, "step": 3100 }, { "epoch": 0.3, "learning_rate": 9.042041946468923e-05, "loss": 2.2847, "step": 3110 }, { "epoch": 0.3, "learning_rate": 9.038859361573471e-05, "loss": 1.9266, "step": 3120 }, { "epoch": 0.3, "learning_rate": 9.035676776678018e-05, "loss": 1.7437, "step": 3130 }, { "epoch": 0.3, "learning_rate": 9.032494191782566e-05, "loss": 1.5789, "step": 3140 }, { "epoch": 0.3, "learning_rate": 9.029311606887115e-05, "loss": 2.0244, "step": 3150 }, { "epoch": 0.3, "learning_rate": 9.026129021991662e-05, "loss": 2.1154, "step": 3160 }, { "epoch": 0.3, "learning_rate": 9.02294643709621e-05, "loss": 2.1877, "step": 3170 }, { "epoch": 0.3, "learning_rate": 9.019763852200758e-05, "loss": 2.0215, "step": 3180 }, { "epoch": 0.3, "learning_rate": 9.016581267305307e-05, "loss": 1.7053, "step": 3190 }, { "epoch": 0.3, "learning_rate": 9.013398682409854e-05, "loss": 2.1041, "step": 3200 }, { "epoch": 0.31, "learning_rate": 9.010216097514401e-05, "loss": 1.8648, "step": 3210 }, { "epoch": 0.31, "learning_rate": 9.00703351261895e-05, "loss": 1.8743, "step": 3220 }, { "epoch": 0.31, "learning_rate": 9.003850927723497e-05, "loss": 1.8577, "step": 3230 }, { "epoch": 0.31, "learning_rate": 9.000668342828046e-05, "loss": 2.1161, "step": 3240 }, { "epoch": 0.31, "learning_rate": 8.997485757932593e-05, "loss": 1.8529, "step": 3250 }, { "epoch": 0.31, "learning_rate": 8.994303173037141e-05, "loss": 1.8198, "step": 3260 }, { "epoch": 0.31, "learning_rate": 8.99112058814169e-05, "loss": 1.6714, "step": 3270 }, { "epoch": 0.31, "learning_rate": 8.987938003246237e-05, "loss": 1.9358, "step": 3280 }, { "epoch": 0.31, "learning_rate": 8.984755418350785e-05, "loss": 2.1932, "step": 3290 }, { "epoch": 0.31, "learning_rate": 8.981572833455333e-05, "loss": 1.5896, "step": 3300 }, { "epoch": 0.32, "learning_rate": 8.97839024855988e-05, "loss": 1.9534, "step": 3310 }, { "epoch": 0.32, "learning_rate": 8.975207663664429e-05, "loss": 1.8981, "step": 3320 }, { "epoch": 0.32, "learning_rate": 8.972025078768977e-05, "loss": 1.697, "step": 3330 }, { "epoch": 0.32, "learning_rate": 8.968842493873525e-05, "loss": 1.92, "step": 3340 }, { "epoch": 0.32, "learning_rate": 8.965659908978072e-05, "loss": 1.7039, "step": 3350 }, { "epoch": 0.32, "learning_rate": 8.96247732408262e-05, "loss": 1.7979, "step": 3360 }, { "epoch": 0.32, "learning_rate": 8.959294739187168e-05, "loss": 2.0006, "step": 3370 }, { "epoch": 0.32, "learning_rate": 8.956112154291716e-05, "loss": 1.8557, "step": 3380 }, { "epoch": 0.32, "learning_rate": 8.952929569396265e-05, "loss": 1.6347, "step": 3390 }, { "epoch": 0.32, "learning_rate": 8.949746984500812e-05, "loss": 1.667, "step": 3400 }, { "epoch": 0.32, "learning_rate": 8.94656439960536e-05, "loss": 1.9593, "step": 3410 }, { "epoch": 0.33, "learning_rate": 8.943381814709908e-05, "loss": 1.504, "step": 3420 }, { "epoch": 0.33, "learning_rate": 8.940199229814455e-05, "loss": 2.2819, "step": 3430 }, { "epoch": 0.33, "learning_rate": 8.937016644919004e-05, "loss": 1.7575, "step": 3440 }, { "epoch": 0.33, "learning_rate": 8.933834060023552e-05, "loss": 1.8409, "step": 3450 }, { "epoch": 0.33, "learning_rate": 8.930651475128099e-05, "loss": 1.9553, "step": 3460 }, { "epoch": 0.33, "learning_rate": 8.927468890232647e-05, "loss": 2.1456, "step": 3470 }, { "epoch": 0.33, "learning_rate": 8.924286305337194e-05, "loss": 1.6622, "step": 3480 }, { "epoch": 0.33, "learning_rate": 8.921103720441744e-05, "loss": 2.3896, "step": 3490 }, { "epoch": 0.33, "learning_rate": 8.917921135546291e-05, "loss": 1.6848, "step": 3500 }, { "epoch": 0.33, "learning_rate": 8.914738550650838e-05, "loss": 1.8145, "step": 3510 }, { "epoch": 0.34, "learning_rate": 8.911555965755386e-05, "loss": 1.8479, "step": 3520 }, { "epoch": 0.34, "learning_rate": 8.908373380859935e-05, "loss": 2.0713, "step": 3530 }, { "epoch": 0.34, "learning_rate": 8.905190795964483e-05, "loss": 1.6836, "step": 3540 }, { "epoch": 0.34, "learning_rate": 8.90200821106903e-05, "loss": 2.1469, "step": 3550 }, { "epoch": 0.34, "learning_rate": 8.898825626173579e-05, "loss": 1.9421, "step": 3560 }, { "epoch": 0.34, "learning_rate": 8.895643041278127e-05, "loss": 1.424, "step": 3570 }, { "epoch": 0.34, "learning_rate": 8.892460456382674e-05, "loss": 2.1441, "step": 3580 }, { "epoch": 0.34, "learning_rate": 8.889277871487222e-05, "loss": 1.9829, "step": 3590 }, { "epoch": 0.34, "learning_rate": 8.88609528659177e-05, "loss": 1.6689, "step": 3600 }, { "epoch": 0.34, "learning_rate": 8.882912701696319e-05, "loss": 1.9551, "step": 3610 }, { "epoch": 0.34, "learning_rate": 8.879730116800866e-05, "loss": 1.5118, "step": 3620 }, { "epoch": 0.35, "learning_rate": 8.876547531905413e-05, "loss": 1.8151, "step": 3630 }, { "epoch": 0.35, "learning_rate": 8.873364947009963e-05, "loss": 2.0585, "step": 3640 }, { "epoch": 0.35, "learning_rate": 8.87018236211451e-05, "loss": 1.9604, "step": 3650 }, { "epoch": 0.35, "learning_rate": 8.866999777219058e-05, "loss": 1.6825, "step": 3660 }, { "epoch": 0.35, "learning_rate": 8.863817192323605e-05, "loss": 2.1185, "step": 3670 }, { "epoch": 0.35, "learning_rate": 8.860634607428154e-05, "loss": 1.7995, "step": 3680 }, { "epoch": 0.35, "learning_rate": 8.857452022532702e-05, "loss": 1.7444, "step": 3690 }, { "epoch": 0.35, "learning_rate": 8.854269437637249e-05, "loss": 1.7131, "step": 3700 }, { "epoch": 0.35, "learning_rate": 8.851086852741797e-05, "loss": 1.5694, "step": 3710 }, { "epoch": 0.35, "learning_rate": 8.847904267846346e-05, "loss": 1.8997, "step": 3720 }, { "epoch": 0.36, "learning_rate": 8.844721682950893e-05, "loss": 1.7479, "step": 3730 }, { "epoch": 0.36, "learning_rate": 8.841539098055441e-05, "loss": 2.0054, "step": 3740 }, { "epoch": 0.36, "learning_rate": 8.83835651315999e-05, "loss": 1.783, "step": 3750 }, { "epoch": 0.36, "learning_rate": 8.835173928264538e-05, "loss": 1.6751, "step": 3760 }, { "epoch": 0.36, "learning_rate": 8.831991343369085e-05, "loss": 1.9195, "step": 3770 }, { "epoch": 0.36, "learning_rate": 8.828808758473632e-05, "loss": 1.6263, "step": 3780 }, { "epoch": 0.36, "learning_rate": 8.825626173578181e-05, "loss": 1.8235, "step": 3790 }, { "epoch": 0.36, "learning_rate": 8.822443588682728e-05, "loss": 1.8654, "step": 3800 }, { "epoch": 0.36, "learning_rate": 8.819261003787277e-05, "loss": 2.0507, "step": 3810 }, { "epoch": 0.36, "learning_rate": 8.816078418891824e-05, "loss": 1.7236, "step": 3820 }, { "epoch": 0.36, "learning_rate": 8.812895833996372e-05, "loss": 2.1645, "step": 3830 }, { "epoch": 0.37, "learning_rate": 8.80971324910092e-05, "loss": 2.1542, "step": 3840 }, { "epoch": 0.37, "learning_rate": 8.806530664205468e-05, "loss": 1.6501, "step": 3850 }, { "epoch": 0.37, "learning_rate": 8.803348079310016e-05, "loss": 1.4847, "step": 3860 }, { "epoch": 0.37, "learning_rate": 8.800165494414564e-05, "loss": 1.9785, "step": 3870 }, { "epoch": 0.37, "learning_rate": 8.796982909519111e-05, "loss": 1.8721, "step": 3880 }, { "epoch": 0.37, "learning_rate": 8.79380032462366e-05, "loss": 1.9387, "step": 3890 }, { "epoch": 0.37, "learning_rate": 8.790617739728208e-05, "loss": 1.6376, "step": 3900 }, { "epoch": 0.37, "learning_rate": 8.787435154832756e-05, "loss": 1.7417, "step": 3910 }, { "epoch": 0.37, "learning_rate": 8.784252569937303e-05, "loss": 1.4899, "step": 3920 }, { "epoch": 0.37, "learning_rate": 8.78106998504185e-05, "loss": 1.8071, "step": 3930 }, { "epoch": 0.37, "learning_rate": 8.7778874001464e-05, "loss": 1.9723, "step": 3940 }, { "epoch": 0.38, "learning_rate": 8.774704815250947e-05, "loss": 1.7802, "step": 3950 }, { "epoch": 0.38, "learning_rate": 8.771522230355495e-05, "loss": 1.7954, "step": 3960 }, { "epoch": 0.38, "learning_rate": 8.768339645460043e-05, "loss": 1.825, "step": 3970 }, { "epoch": 0.38, "learning_rate": 8.765157060564591e-05, "loss": 2.056, "step": 3980 }, { "epoch": 0.38, "learning_rate": 8.761974475669139e-05, "loss": 2.1008, "step": 3990 }, { "epoch": 0.38, "learning_rate": 8.758791890773686e-05, "loss": 1.9251, "step": 4000 }, { "epoch": 0.38, "learning_rate": 8.755609305878235e-05, "loss": 1.7959, "step": 4010 }, { "epoch": 0.38, "learning_rate": 8.752426720982783e-05, "loss": 1.9192, "step": 4020 }, { "epoch": 0.38, "learning_rate": 8.74924413608733e-05, "loss": 2.0, "step": 4030 }, { "epoch": 0.38, "learning_rate": 8.746061551191878e-05, "loss": 1.8604, "step": 4040 }, { "epoch": 0.39, "learning_rate": 8.742878966296427e-05, "loss": 1.7686, "step": 4050 }, { "epoch": 0.39, "learning_rate": 8.739696381400975e-05, "loss": 1.7677, "step": 4060 }, { "epoch": 0.39, "learning_rate": 8.736513796505522e-05, "loss": 1.7588, "step": 4070 }, { "epoch": 0.39, "learning_rate": 8.733331211610069e-05, "loss": 1.8295, "step": 4080 }, { "epoch": 0.39, "learning_rate": 8.730148626714619e-05, "loss": 1.9025, "step": 4090 }, { "epoch": 0.39, "learning_rate": 8.726966041819166e-05, "loss": 1.7978, "step": 4100 }, { "epoch": 0.39, "learning_rate": 8.723783456923714e-05, "loss": 1.7281, "step": 4110 }, { "epoch": 0.39, "learning_rate": 8.720600872028261e-05, "loss": 1.5019, "step": 4120 }, { "epoch": 0.39, "learning_rate": 8.71741828713281e-05, "loss": 1.604, "step": 4130 }, { "epoch": 0.39, "learning_rate": 8.714235702237358e-05, "loss": 1.6965, "step": 4140 }, { "epoch": 0.39, "learning_rate": 8.711053117341905e-05, "loss": 1.7006, "step": 4150 }, { "epoch": 0.4, "learning_rate": 8.707870532446453e-05, "loss": 1.8867, "step": 4160 }, { "epoch": 0.4, "learning_rate": 8.704687947551002e-05, "loss": 1.9758, "step": 4170 }, { "epoch": 0.4, "learning_rate": 8.701505362655549e-05, "loss": 1.7924, "step": 4180 }, { "epoch": 0.4, "learning_rate": 8.698322777760097e-05, "loss": 1.4875, "step": 4190 }, { "epoch": 0.4, "learning_rate": 8.695140192864645e-05, "loss": 1.7991, "step": 4200 }, { "epoch": 0.4, "learning_rate": 8.691957607969194e-05, "loss": 2.2652, "step": 4210 }, { "epoch": 0.4, "learning_rate": 8.688775023073741e-05, "loss": 1.8837, "step": 4220 }, { "epoch": 0.4, "learning_rate": 8.685592438178288e-05, "loss": 1.7792, "step": 4230 }, { "epoch": 0.4, "learning_rate": 8.682409853282837e-05, "loss": 2.4074, "step": 4240 }, { "epoch": 0.4, "learning_rate": 8.679227268387384e-05, "loss": 1.8309, "step": 4250 }, { "epoch": 0.41, "learning_rate": 8.676044683491933e-05, "loss": 1.6532, "step": 4260 }, { "epoch": 0.41, "learning_rate": 8.67286209859648e-05, "loss": 1.6922, "step": 4270 }, { "epoch": 0.41, "learning_rate": 8.66967951370103e-05, "loss": 1.5487, "step": 4280 }, { "epoch": 0.41, "learning_rate": 8.666496928805577e-05, "loss": 1.9437, "step": 4290 }, { "epoch": 0.41, "learning_rate": 8.663314343910124e-05, "loss": 1.8552, "step": 4300 }, { "epoch": 0.41, "learning_rate": 8.660131759014672e-05, "loss": 2.2202, "step": 4310 }, { "epoch": 0.41, "learning_rate": 8.65694917411922e-05, "loss": 1.7241, "step": 4320 }, { "epoch": 0.41, "learning_rate": 8.653766589223769e-05, "loss": 2.0968, "step": 4330 }, { "epoch": 0.41, "learning_rate": 8.650584004328316e-05, "loss": 2.2153, "step": 4340 }, { "epoch": 0.41, "learning_rate": 8.647401419432864e-05, "loss": 1.7501, "step": 4350 }, { "epoch": 0.41, "learning_rate": 8.644218834537412e-05, "loss": 2.2423, "step": 4360 }, { "epoch": 0.42, "learning_rate": 8.64103624964196e-05, "loss": 1.6641, "step": 4370 }, { "epoch": 0.42, "learning_rate": 8.637853664746508e-05, "loss": 2.1719, "step": 4380 }, { "epoch": 0.42, "learning_rate": 8.634671079851056e-05, "loss": 1.7141, "step": 4390 }, { "epoch": 0.42, "learning_rate": 8.631488494955603e-05, "loss": 1.8436, "step": 4400 }, { "epoch": 0.42, "learning_rate": 8.628305910060152e-05, "loss": 1.9603, "step": 4410 }, { "epoch": 0.42, "learning_rate": 8.625123325164699e-05, "loss": 1.7869, "step": 4420 }, { "epoch": 0.42, "learning_rate": 8.621940740269247e-05, "loss": 1.9844, "step": 4430 }, { "epoch": 0.42, "learning_rate": 8.618758155373795e-05, "loss": 1.9699, "step": 4440 }, { "epoch": 0.42, "learning_rate": 8.615575570478342e-05, "loss": 1.879, "step": 4450 }, { "epoch": 0.42, "learning_rate": 8.61239298558289e-05, "loss": 1.8398, "step": 4460 }, { "epoch": 0.43, "learning_rate": 8.609210400687439e-05, "loss": 2.1011, "step": 4470 }, { "epoch": 0.43, "learning_rate": 8.606027815791987e-05, "loss": 1.9542, "step": 4480 }, { "epoch": 0.43, "learning_rate": 8.602845230896534e-05, "loss": 2.2341, "step": 4490 }, { "epoch": 0.43, "learning_rate": 8.599662646001081e-05, "loss": 1.7047, "step": 4500 }, { "epoch": 0.43, "learning_rate": 8.596480061105631e-05, "loss": 1.8369, "step": 4510 }, { "epoch": 0.43, "learning_rate": 8.593297476210178e-05, "loss": 1.7466, "step": 4520 }, { "epoch": 0.43, "learning_rate": 8.590114891314726e-05, "loss": 1.736, "step": 4530 }, { "epoch": 0.43, "learning_rate": 8.586932306419274e-05, "loss": 1.6473, "step": 4540 }, { "epoch": 0.43, "learning_rate": 8.583749721523822e-05, "loss": 1.8275, "step": 4550 }, { "epoch": 0.43, "learning_rate": 8.58056713662837e-05, "loss": 1.8659, "step": 4560 }, { "epoch": 0.43, "learning_rate": 8.577384551732917e-05, "loss": 1.6171, "step": 4570 }, { "epoch": 0.44, "learning_rate": 8.574201966837466e-05, "loss": 1.5898, "step": 4580 }, { "epoch": 0.44, "learning_rate": 8.571019381942014e-05, "loss": 1.7374, "step": 4590 }, { "epoch": 0.44, "learning_rate": 8.567836797046561e-05, "loss": 1.7308, "step": 4600 }, { "epoch": 0.44, "learning_rate": 8.56465421215111e-05, "loss": 1.8294, "step": 4610 }, { "epoch": 0.44, "learning_rate": 8.561471627255658e-05, "loss": 1.5242, "step": 4620 }, { "epoch": 0.44, "learning_rate": 8.558289042360206e-05, "loss": 2.284, "step": 4630 }, { "epoch": 0.44, "learning_rate": 8.555106457464753e-05, "loss": 1.7441, "step": 4640 }, { "epoch": 0.44, "learning_rate": 8.5519238725693e-05, "loss": 2.1215, "step": 4650 }, { "epoch": 0.44, "learning_rate": 8.54874128767385e-05, "loss": 1.9204, "step": 4660 }, { "epoch": 0.44, "learning_rate": 8.545558702778397e-05, "loss": 1.8143, "step": 4670 }, { "epoch": 0.45, "learning_rate": 8.542376117882945e-05, "loss": 1.8353, "step": 4680 }, { "epoch": 0.45, "learning_rate": 8.539193532987492e-05, "loss": 1.9182, "step": 4690 }, { "epoch": 0.45, "learning_rate": 8.53601094809204e-05, "loss": 1.7577, "step": 4700 }, { "epoch": 0.45, "learning_rate": 8.532828363196589e-05, "loss": 1.5695, "step": 4710 }, { "epoch": 0.45, "learning_rate": 8.529645778301136e-05, "loss": 1.2831, "step": 4720 }, { "epoch": 0.45, "learning_rate": 8.526463193405684e-05, "loss": 1.7315, "step": 4730 }, { "epoch": 0.45, "learning_rate": 8.523280608510233e-05, "loss": 1.7665, "step": 4740 }, { "epoch": 0.45, "learning_rate": 8.52009802361478e-05, "loss": 1.8807, "step": 4750 }, { "epoch": 0.45, "learning_rate": 8.516915438719328e-05, "loss": 1.7689, "step": 4760 }, { "epoch": 0.45, "learning_rate": 8.513732853823876e-05, "loss": 1.6982, "step": 4770 }, { "epoch": 0.45, "learning_rate": 8.510550268928425e-05, "loss": 2.0575, "step": 4780 }, { "epoch": 0.46, "learning_rate": 8.507367684032972e-05, "loss": 1.8133, "step": 4790 }, { "epoch": 0.46, "learning_rate": 8.504185099137519e-05, "loss": 1.949, "step": 4800 }, { "epoch": 0.46, "learning_rate": 8.501002514242068e-05, "loss": 1.5134, "step": 4810 }, { "epoch": 0.46, "learning_rate": 8.497819929346615e-05, "loss": 1.9392, "step": 4820 }, { "epoch": 0.46, "learning_rate": 8.494637344451164e-05, "loss": 1.7536, "step": 4830 }, { "epoch": 0.46, "learning_rate": 8.491454759555711e-05, "loss": 2.1834, "step": 4840 }, { "epoch": 0.46, "learning_rate": 8.488272174660259e-05, "loss": 1.672, "step": 4850 }, { "epoch": 0.46, "learning_rate": 8.485089589764808e-05, "loss": 1.7865, "step": 4860 }, { "epoch": 0.46, "learning_rate": 8.481907004869355e-05, "loss": 1.6641, "step": 4870 }, { "epoch": 0.46, "learning_rate": 8.478724419973903e-05, "loss": 1.8628, "step": 4880 }, { "epoch": 0.47, "learning_rate": 8.475541835078451e-05, "loss": 1.7441, "step": 4890 }, { "epoch": 0.47, "learning_rate": 8.472359250183e-05, "loss": 1.4225, "step": 4900 }, { "epoch": 0.47, "learning_rate": 8.469176665287547e-05, "loss": 1.8587, "step": 4910 }, { "epoch": 0.47, "learning_rate": 8.465994080392095e-05, "loss": 1.7314, "step": 4920 }, { "epoch": 0.47, "learning_rate": 8.462811495496643e-05, "loss": 2.0339, "step": 4930 }, { "epoch": 0.47, "learning_rate": 8.45962891060119e-05, "loss": 1.9146, "step": 4940 }, { "epoch": 0.47, "learning_rate": 8.456446325705739e-05, "loss": 1.9494, "step": 4950 }, { "epoch": 0.47, "learning_rate": 8.453263740810287e-05, "loss": 1.5817, "step": 4960 }, { "epoch": 0.47, "learning_rate": 8.450081155914834e-05, "loss": 1.8283, "step": 4970 }, { "epoch": 0.47, "learning_rate": 8.446898571019383e-05, "loss": 1.5496, "step": 4980 }, { "epoch": 0.47, "learning_rate": 8.44371598612393e-05, "loss": 1.6279, "step": 4990 }, { "epoch": 0.48, "learning_rate": 8.440533401228479e-05, "loss": 2.1064, "step": 5000 }, { "epoch": 0.48, "learning_rate": 8.437350816333026e-05, "loss": 1.8572, "step": 5010 }, { "epoch": 0.48, "learning_rate": 8.434168231437573e-05, "loss": 1.9051, "step": 5020 }, { "epoch": 0.48, "learning_rate": 8.430985646542122e-05, "loss": 1.8268, "step": 5030 }, { "epoch": 0.48, "learning_rate": 8.42780306164667e-05, "loss": 2.0814, "step": 5040 }, { "epoch": 0.48, "learning_rate": 8.424620476751218e-05, "loss": 1.5997, "step": 5050 }, { "epoch": 0.48, "learning_rate": 8.421437891855765e-05, "loss": 2.0973, "step": 5060 }, { "epoch": 0.48, "learning_rate": 8.418255306960314e-05, "loss": 2.0325, "step": 5070 }, { "epoch": 0.48, "learning_rate": 8.415072722064862e-05, "loss": 1.9608, "step": 5080 }, { "epoch": 0.48, "learning_rate": 8.411890137169409e-05, "loss": 2.1482, "step": 5090 }, { "epoch": 0.49, "learning_rate": 8.408707552273957e-05, "loss": 1.7059, "step": 5100 }, { "epoch": 0.49, "learning_rate": 8.405524967378506e-05, "loss": 1.6616, "step": 5110 }, { "epoch": 0.49, "learning_rate": 8.402342382483053e-05, "loss": 1.331, "step": 5120 }, { "epoch": 0.49, "learning_rate": 8.399159797587601e-05, "loss": 1.9625, "step": 5130 }, { "epoch": 0.49, "learning_rate": 8.395977212692148e-05, "loss": 1.7034, "step": 5140 }, { "epoch": 0.49, "learning_rate": 8.392794627796698e-05, "loss": 2.0897, "step": 5150 }, { "epoch": 0.49, "learning_rate": 8.389612042901245e-05, "loss": 1.681, "step": 5160 }, { "epoch": 0.49, "learning_rate": 8.386429458005792e-05, "loss": 1.7284, "step": 5170 }, { "epoch": 0.49, "learning_rate": 8.38324687311034e-05, "loss": 1.7513, "step": 5180 }, { "epoch": 0.49, "learning_rate": 8.380064288214889e-05, "loss": 1.3583, "step": 5190 }, { "epoch": 0.49, "learning_rate": 8.376881703319437e-05, "loss": 1.7959, "step": 5200 }, { "epoch": 0.5, "learning_rate": 8.373699118423984e-05, "loss": 2.121, "step": 5210 }, { "epoch": 0.5, "learning_rate": 8.370516533528532e-05, "loss": 2.2103, "step": 5220 }, { "epoch": 0.5, "learning_rate": 8.367333948633081e-05, "loss": 1.9372, "step": 5230 }, { "epoch": 0.5, "learning_rate": 8.364151363737628e-05, "loss": 1.7694, "step": 5240 }, { "epoch": 0.5, "learning_rate": 8.360968778842176e-05, "loss": 1.8922, "step": 5250 }, { "epoch": 0.5, "learning_rate": 8.357786193946725e-05, "loss": 1.8685, "step": 5260 }, { "epoch": 0.5, "learning_rate": 8.354603609051272e-05, "loss": 1.7645, "step": 5270 }, { "epoch": 0.5, "learning_rate": 8.35142102415582e-05, "loss": 2.0918, "step": 5280 }, { "epoch": 0.5, "learning_rate": 8.348238439260367e-05, "loss": 1.6694, "step": 5290 }, { "epoch": 0.5, "learning_rate": 8.345055854364917e-05, "loss": 1.6349, "step": 5300 }, { "epoch": 0.51, "learning_rate": 8.341873269469464e-05, "loss": 1.7016, "step": 5310 }, { "epoch": 0.51, "learning_rate": 8.33869068457401e-05, "loss": 1.9029, "step": 5320 }, { "epoch": 0.51, "learning_rate": 8.335508099678559e-05, "loss": 1.7197, "step": 5330 }, { "epoch": 0.51, "learning_rate": 8.332325514783107e-05, "loss": 2.0524, "step": 5340 }, { "epoch": 0.51, "learning_rate": 8.329142929887656e-05, "loss": 1.7017, "step": 5350 }, { "epoch": 0.51, "learning_rate": 8.325960344992203e-05, "loss": 2.2195, "step": 5360 }, { "epoch": 0.51, "learning_rate": 8.322777760096751e-05, "loss": 1.7724, "step": 5370 }, { "epoch": 0.51, "learning_rate": 8.3195951752013e-05, "loss": 1.9593, "step": 5380 }, { "epoch": 0.51, "learning_rate": 8.316412590305846e-05, "loss": 1.8544, "step": 5390 }, { "epoch": 0.51, "learning_rate": 8.313230005410395e-05, "loss": 2.0837, "step": 5400 }, { "epoch": 0.51, "learning_rate": 8.310047420514943e-05, "loss": 1.97, "step": 5410 }, { "epoch": 0.52, "learning_rate": 8.30686483561949e-05, "loss": 1.4903, "step": 5420 }, { "epoch": 0.52, "learning_rate": 8.303682250724039e-05, "loss": 1.6233, "step": 5430 }, { "epoch": 0.52, "learning_rate": 8.300499665828586e-05, "loss": 1.5539, "step": 5440 }, { "epoch": 0.52, "learning_rate": 8.297317080933135e-05, "loss": 1.8161, "step": 5450 }, { "epoch": 0.52, "learning_rate": 8.294134496037682e-05, "loss": 1.752, "step": 5460 }, { "epoch": 0.52, "learning_rate": 8.29095191114223e-05, "loss": 2.069, "step": 5470 }, { "epoch": 0.52, "learning_rate": 8.287769326246778e-05, "loss": 2.061, "step": 5480 }, { "epoch": 0.52, "learning_rate": 8.284586741351326e-05, "loss": 1.6476, "step": 5490 }, { "epoch": 0.52, "learning_rate": 8.281404156455874e-05, "loss": 1.6349, "step": 5500 }, { "epoch": 0.52, "learning_rate": 8.278221571560421e-05, "loss": 1.4046, "step": 5510 }, { "epoch": 0.53, "learning_rate": 8.27503898666497e-05, "loss": 1.7181, "step": 5520 }, { "epoch": 0.53, "learning_rate": 8.271856401769518e-05, "loss": 1.8994, "step": 5530 }, { "epoch": 0.53, "learning_rate": 8.268673816874065e-05, "loss": 2.2467, "step": 5540 }, { "epoch": 0.53, "learning_rate": 8.265491231978614e-05, "loss": 1.7663, "step": 5550 }, { "epoch": 0.53, "learning_rate": 8.26230864708316e-05, "loss": 2.232, "step": 5560 }, { "epoch": 0.53, "learning_rate": 8.25912606218771e-05, "loss": 2.0191, "step": 5570 }, { "epoch": 0.53, "learning_rate": 8.255943477292257e-05, "loss": 2.1281, "step": 5580 }, { "epoch": 0.53, "learning_rate": 8.252760892396804e-05, "loss": 1.7599, "step": 5590 }, { "epoch": 0.53, "learning_rate": 8.249578307501353e-05, "loss": 1.9067, "step": 5600 }, { "epoch": 0.53, "learning_rate": 8.246395722605901e-05, "loss": 1.8876, "step": 5610 }, { "epoch": 0.53, "learning_rate": 8.24321313771045e-05, "loss": 2.0441, "step": 5620 }, { "epoch": 0.54, "learning_rate": 8.240030552814996e-05, "loss": 2.1636, "step": 5630 }, { "epoch": 0.54, "learning_rate": 8.236847967919545e-05, "loss": 1.6258, "step": 5640 }, { "epoch": 0.54, "learning_rate": 8.233665383024093e-05, "loss": 1.761, "step": 5650 }, { "epoch": 0.54, "learning_rate": 8.23048279812864e-05, "loss": 1.6266, "step": 5660 }, { "epoch": 0.54, "learning_rate": 8.227300213233188e-05, "loss": 1.6244, "step": 5670 }, { "epoch": 0.54, "learning_rate": 8.224117628337737e-05, "loss": 1.9169, "step": 5680 }, { "epoch": 0.54, "learning_rate": 8.220935043442284e-05, "loss": 1.3947, "step": 5690 }, { "epoch": 0.54, "learning_rate": 8.217752458546832e-05, "loss": 1.6006, "step": 5700 }, { "epoch": 0.54, "learning_rate": 8.214569873651379e-05, "loss": 1.4239, "step": 5710 }, { "epoch": 0.54, "learning_rate": 8.211387288755929e-05, "loss": 1.5619, "step": 5720 }, { "epoch": 0.55, "learning_rate": 8.208204703860476e-05, "loss": 1.5051, "step": 5730 }, { "epoch": 0.55, "learning_rate": 8.205022118965023e-05, "loss": 1.8509, "step": 5740 }, { "epoch": 0.55, "learning_rate": 8.201839534069571e-05, "loss": 1.7132, "step": 5750 }, { "epoch": 0.55, "learning_rate": 8.19865694917412e-05, "loss": 1.9296, "step": 5760 }, { "epoch": 0.55, "learning_rate": 8.195474364278668e-05, "loss": 1.8191, "step": 5770 }, { "epoch": 0.55, "learning_rate": 8.192291779383215e-05, "loss": 1.6306, "step": 5780 }, { "epoch": 0.55, "learning_rate": 8.189109194487763e-05, "loss": 1.5563, "step": 5790 }, { "epoch": 0.55, "learning_rate": 8.185926609592312e-05, "loss": 1.8907, "step": 5800 }, { "epoch": 0.55, "learning_rate": 8.182744024696859e-05, "loss": 1.6601, "step": 5810 }, { "epoch": 0.55, "learning_rate": 8.179561439801407e-05, "loss": 1.8245, "step": 5820 }, { "epoch": 0.55, "learning_rate": 8.176378854905956e-05, "loss": 1.8156, "step": 5830 }, { "epoch": 0.56, "learning_rate": 8.173196270010503e-05, "loss": 1.9084, "step": 5840 }, { "epoch": 0.56, "learning_rate": 8.170013685115051e-05, "loss": 1.5468, "step": 5850 }, { "epoch": 0.56, "learning_rate": 8.166831100219598e-05, "loss": 1.9778, "step": 5860 }, { "epoch": 0.56, "learning_rate": 8.163648515324148e-05, "loss": 1.701, "step": 5870 }, { "epoch": 0.56, "learning_rate": 8.160465930428695e-05, "loss": 1.6719, "step": 5880 }, { "epoch": 0.56, "learning_rate": 8.157283345533242e-05, "loss": 1.6509, "step": 5890 }, { "epoch": 0.56, "learning_rate": 8.15410076063779e-05, "loss": 1.673, "step": 5900 }, { "epoch": 0.56, "learning_rate": 8.150918175742338e-05, "loss": 1.3831, "step": 5910 }, { "epoch": 0.56, "learning_rate": 8.147735590846887e-05, "loss": 1.7159, "step": 5920 }, { "epoch": 0.56, "learning_rate": 8.144553005951434e-05, "loss": 1.832, "step": 5930 }, { "epoch": 0.57, "learning_rate": 8.141370421055982e-05, "loss": 1.93, "step": 5940 }, { "epoch": 0.57, "learning_rate": 8.13818783616053e-05, "loss": 1.3991, "step": 5950 }, { "epoch": 0.57, "learning_rate": 8.135005251265077e-05, "loss": 1.7181, "step": 5960 }, { "epoch": 0.57, "learning_rate": 8.131822666369626e-05, "loss": 1.7019, "step": 5970 }, { "epoch": 0.57, "learning_rate": 8.128640081474174e-05, "loss": 1.7486, "step": 5980 }, { "epoch": 0.57, "learning_rate": 8.125457496578721e-05, "loss": 1.8143, "step": 5990 }, { "epoch": 0.57, "learning_rate": 8.12227491168327e-05, "loss": 1.7242, "step": 6000 }, { "epoch": 0.57, "learning_rate": 8.119092326787817e-05, "loss": 1.3983, "step": 6010 }, { "epoch": 0.57, "learning_rate": 8.115909741892366e-05, "loss": 1.5508, "step": 6020 }, { "epoch": 0.57, "learning_rate": 8.112727156996913e-05, "loss": 2.059, "step": 6030 }, { "epoch": 0.57, "learning_rate": 8.10954457210146e-05, "loss": 2.0893, "step": 6040 }, { "epoch": 0.58, "learning_rate": 8.106361987206009e-05, "loss": 2.0184, "step": 6050 }, { "epoch": 0.58, "learning_rate": 8.103179402310557e-05, "loss": 1.8744, "step": 6060 }, { "epoch": 0.58, "learning_rate": 8.099996817415105e-05, "loss": 1.4619, "step": 6070 }, { "epoch": 0.58, "learning_rate": 8.096814232519652e-05, "loss": 1.7368, "step": 6080 }, { "epoch": 0.58, "learning_rate": 8.093631647624201e-05, "loss": 1.7933, "step": 6090 }, { "epoch": 0.58, "learning_rate": 8.090449062728749e-05, "loss": 1.5629, "step": 6100 }, { "epoch": 0.58, "learning_rate": 8.087266477833296e-05, "loss": 2.0451, "step": 6110 }, { "epoch": 0.58, "learning_rate": 8.084083892937845e-05, "loss": 1.8756, "step": 6120 }, { "epoch": 0.58, "learning_rate": 8.080901308042393e-05, "loss": 1.7944, "step": 6130 }, { "epoch": 0.58, "learning_rate": 8.07771872314694e-05, "loss": 1.6811, "step": 6140 }, { "epoch": 0.59, "learning_rate": 8.074536138251488e-05, "loss": 1.4877, "step": 6150 }, { "epoch": 0.59, "learning_rate": 8.071353553356035e-05, "loss": 2.0818, "step": 6160 }, { "epoch": 0.59, "learning_rate": 8.068170968460585e-05, "loss": 1.4355, "step": 6170 }, { "epoch": 0.59, "learning_rate": 8.064988383565132e-05, "loss": 1.6414, "step": 6180 }, { "epoch": 0.59, "learning_rate": 8.06180579866968e-05, "loss": 2.0349, "step": 6190 }, { "epoch": 0.59, "learning_rate": 8.058623213774227e-05, "loss": 1.6727, "step": 6200 }, { "epoch": 0.59, "learning_rate": 8.055440628878776e-05, "loss": 1.8855, "step": 6210 }, { "epoch": 0.59, "learning_rate": 8.052258043983324e-05, "loss": 1.6737, "step": 6220 }, { "epoch": 0.59, "learning_rate": 8.049075459087871e-05, "loss": 1.485, "step": 6230 }, { "epoch": 0.59, "learning_rate": 8.04589287419242e-05, "loss": 1.277, "step": 6240 }, { "epoch": 0.59, "learning_rate": 8.042710289296968e-05, "loss": 1.778, "step": 6250 }, { "epoch": 0.6, "learning_rate": 8.039527704401515e-05, "loss": 2.0159, "step": 6260 }, { "epoch": 0.6, "learning_rate": 8.036345119506063e-05, "loss": 1.7595, "step": 6270 }, { "epoch": 0.6, "learning_rate": 8.033162534610612e-05, "loss": 1.7878, "step": 6280 }, { "epoch": 0.6, "learning_rate": 8.02997994971516e-05, "loss": 2.0789, "step": 6290 }, { "epoch": 0.6, "learning_rate": 8.026797364819707e-05, "loss": 1.613, "step": 6300 }, { "epoch": 0.6, "learning_rate": 8.023614779924254e-05, "loss": 1.9158, "step": 6310 }, { "epoch": 0.6, "learning_rate": 8.020432195028804e-05, "loss": 1.7132, "step": 6320 }, { "epoch": 0.6, "learning_rate": 8.017249610133351e-05, "loss": 1.8105, "step": 6330 }, { "epoch": 0.6, "learning_rate": 8.014067025237899e-05, "loss": 1.4884, "step": 6340 }, { "epoch": 0.6, "learning_rate": 8.010884440342446e-05, "loss": 1.2167, "step": 6350 }, { "epoch": 0.61, "learning_rate": 8.007701855446994e-05, "loss": 1.8082, "step": 6360 }, { "epoch": 0.61, "learning_rate": 8.004519270551543e-05, "loss": 1.6277, "step": 6370 }, { "epoch": 0.61, "learning_rate": 8.00133668565609e-05, "loss": 1.7452, "step": 6380 }, { "epoch": 0.61, "learning_rate": 7.998154100760638e-05, "loss": 1.7174, "step": 6390 }, { "epoch": 0.61, "learning_rate": 7.994971515865187e-05, "loss": 2.2034, "step": 6400 }, { "epoch": 0.61, "learning_rate": 7.991788930969734e-05, "loss": 1.6318, "step": 6410 }, { "epoch": 0.61, "learning_rate": 7.988606346074282e-05, "loss": 1.6383, "step": 6420 }, { "epoch": 0.61, "learning_rate": 7.98542376117883e-05, "loss": 1.4466, "step": 6430 }, { "epoch": 0.61, "learning_rate": 7.982241176283379e-05, "loss": 1.9118, "step": 6440 }, { "epoch": 0.61, "learning_rate": 7.979058591387926e-05, "loss": 1.8459, "step": 6450 }, { "epoch": 0.61, "learning_rate": 7.975876006492473e-05, "loss": 1.7772, "step": 6460 }, { "epoch": 0.62, "learning_rate": 7.972693421597022e-05, "loss": 1.6248, "step": 6470 }, { "epoch": 0.62, "learning_rate": 7.96951083670157e-05, "loss": 1.5957, "step": 6480 }, { "epoch": 0.62, "learning_rate": 7.966328251806118e-05, "loss": 1.4671, "step": 6490 }, { "epoch": 0.62, "learning_rate": 7.963145666910665e-05, "loss": 1.569, "step": 6500 }, { "epoch": 0.62, "learning_rate": 7.959963082015213e-05, "loss": 1.9128, "step": 6510 }, { "epoch": 0.62, "learning_rate": 7.956780497119761e-05, "loss": 1.7862, "step": 6520 }, { "epoch": 0.62, "learning_rate": 7.953597912224308e-05, "loss": 1.7062, "step": 6530 }, { "epoch": 0.62, "learning_rate": 7.950415327328857e-05, "loss": 1.4484, "step": 6540 }, { "epoch": 0.62, "learning_rate": 7.947232742433405e-05, "loss": 1.9372, "step": 6550 }, { "epoch": 0.62, "learning_rate": 7.944050157537952e-05, "loss": 2.0335, "step": 6560 }, { "epoch": 0.63, "learning_rate": 7.9408675726425e-05, "loss": 1.8989, "step": 6570 }, { "epoch": 0.63, "learning_rate": 7.937684987747049e-05, "loss": 1.4778, "step": 6580 }, { "epoch": 0.63, "learning_rate": 7.934502402851597e-05, "loss": 1.5545, "step": 6590 }, { "epoch": 0.63, "learning_rate": 7.931319817956144e-05, "loss": 1.7245, "step": 6600 }, { "epoch": 0.63, "learning_rate": 7.928137233060691e-05, "loss": 1.7073, "step": 6610 }, { "epoch": 0.63, "learning_rate": 7.92495464816524e-05, "loss": 1.5795, "step": 6620 }, { "epoch": 0.63, "learning_rate": 7.921772063269788e-05, "loss": 1.7212, "step": 6630 }, { "epoch": 0.63, "learning_rate": 7.918589478374336e-05, "loss": 1.3638, "step": 6640 }, { "epoch": 0.63, "learning_rate": 7.915406893478883e-05, "loss": 1.482, "step": 6650 }, { "epoch": 0.63, "learning_rate": 7.912224308583432e-05, "loss": 1.6127, "step": 6660 }, { "epoch": 0.63, "learning_rate": 7.90904172368798e-05, "loss": 1.8949, "step": 6670 }, { "epoch": 0.64, "learning_rate": 7.905859138792527e-05, "loss": 1.6316, "step": 6680 }, { "epoch": 0.64, "learning_rate": 7.902676553897076e-05, "loss": 1.7575, "step": 6690 }, { "epoch": 0.64, "learning_rate": 7.899493969001624e-05, "loss": 1.6866, "step": 6700 }, { "epoch": 0.64, "learning_rate": 7.896311384106171e-05, "loss": 1.6949, "step": 6710 }, { "epoch": 0.64, "learning_rate": 7.893128799210719e-05, "loss": 2.1837, "step": 6720 }, { "epoch": 0.64, "learning_rate": 7.889946214315266e-05, "loss": 1.754, "step": 6730 }, { "epoch": 0.64, "learning_rate": 7.886763629419816e-05, "loss": 1.6275, "step": 6740 }, { "epoch": 0.64, "learning_rate": 7.883581044524363e-05, "loss": 1.6981, "step": 6750 }, { "epoch": 0.64, "learning_rate": 7.88039845962891e-05, "loss": 1.6664, "step": 6760 }, { "epoch": 0.64, "learning_rate": 7.877215874733458e-05, "loss": 1.7061, "step": 6770 }, { "epoch": 0.65, "learning_rate": 7.874033289838007e-05, "loss": 1.6414, "step": 6780 }, { "epoch": 0.65, "learning_rate": 7.870850704942555e-05, "loss": 1.7035, "step": 6790 }, { "epoch": 0.65, "learning_rate": 7.867668120047102e-05, "loss": 1.5274, "step": 6800 }, { "epoch": 0.65, "learning_rate": 7.86448553515165e-05, "loss": 2.1185, "step": 6810 }, { "epoch": 0.65, "learning_rate": 7.861302950256199e-05, "loss": 1.949, "step": 6820 }, { "epoch": 0.65, "learning_rate": 7.858120365360746e-05, "loss": 1.7355, "step": 6830 }, { "epoch": 0.65, "learning_rate": 7.854937780465294e-05, "loss": 2.224, "step": 6840 }, { "epoch": 0.65, "learning_rate": 7.851755195569843e-05, "loss": 1.7225, "step": 6850 }, { "epoch": 0.65, "learning_rate": 7.848572610674391e-05, "loss": 2.2467, "step": 6860 }, { "epoch": 0.65, "learning_rate": 7.845390025778938e-05, "loss": 1.4054, "step": 6870 }, { "epoch": 0.65, "learning_rate": 7.842207440883485e-05, "loss": 1.3582, "step": 6880 }, { "epoch": 0.66, "learning_rate": 7.839024855988035e-05, "loss": 1.356, "step": 6890 }, { "epoch": 0.66, "learning_rate": 7.835842271092582e-05, "loss": 1.4775, "step": 6900 }, { "epoch": 0.66, "learning_rate": 7.83265968619713e-05, "loss": 1.8954, "step": 6910 }, { "epoch": 0.66, "learning_rate": 7.829477101301677e-05, "loss": 1.7116, "step": 6920 }, { "epoch": 0.66, "learning_rate": 7.826294516406225e-05, "loss": 1.4697, "step": 6930 }, { "epoch": 0.66, "learning_rate": 7.823111931510774e-05, "loss": 1.5814, "step": 6940 }, { "epoch": 0.66, "learning_rate": 7.819929346615321e-05, "loss": 1.9475, "step": 6950 }, { "epoch": 0.66, "learning_rate": 7.816746761719869e-05, "loss": 1.9221, "step": 6960 }, { "epoch": 0.66, "learning_rate": 7.813564176824418e-05, "loss": 1.8046, "step": 6970 }, { "epoch": 0.66, "learning_rate": 7.810381591928965e-05, "loss": 1.4795, "step": 6980 }, { "epoch": 0.67, "learning_rate": 7.807199007033513e-05, "loss": 1.6698, "step": 6990 }, { "epoch": 0.67, "learning_rate": 7.804016422138061e-05, "loss": 1.4368, "step": 7000 }, { "epoch": 0.67, "learning_rate": 7.80083383724261e-05, "loss": 2.2295, "step": 7010 }, { "epoch": 0.67, "learning_rate": 7.797651252347157e-05, "loss": 1.7967, "step": 7020 }, { "epoch": 0.67, "learning_rate": 7.794468667451704e-05, "loss": 1.8404, "step": 7030 }, { "epoch": 0.67, "learning_rate": 7.791286082556253e-05, "loss": 1.7147, "step": 7040 }, { "epoch": 0.67, "learning_rate": 7.7881034976608e-05, "loss": 1.893, "step": 7050 }, { "epoch": 0.67, "learning_rate": 7.784920912765349e-05, "loss": 1.8299, "step": 7060 }, { "epoch": 0.67, "learning_rate": 7.781738327869896e-05, "loss": 1.9237, "step": 7070 }, { "epoch": 0.67, "learning_rate": 7.778555742974444e-05, "loss": 2.1464, "step": 7080 }, { "epoch": 0.67, "learning_rate": 7.775373158078992e-05, "loss": 1.4533, "step": 7090 }, { "epoch": 0.68, "learning_rate": 7.77219057318354e-05, "loss": 1.5716, "step": 7100 }, { "epoch": 0.68, "learning_rate": 7.769007988288088e-05, "loss": 1.9034, "step": 7110 }, { "epoch": 0.68, "learning_rate": 7.765825403392636e-05, "loss": 1.6687, "step": 7120 }, { "epoch": 0.68, "learning_rate": 7.762642818497183e-05, "loss": 1.5473, "step": 7130 }, { "epoch": 0.68, "learning_rate": 7.759460233601732e-05, "loss": 1.4395, "step": 7140 }, { "epoch": 0.68, "learning_rate": 7.75627764870628e-05, "loss": 1.9144, "step": 7150 }, { "epoch": 0.68, "learning_rate": 7.753095063810828e-05, "loss": 1.4831, "step": 7160 }, { "epoch": 0.68, "learning_rate": 7.749912478915375e-05, "loss": 1.9204, "step": 7170 }, { "epoch": 0.68, "learning_rate": 7.746729894019922e-05, "loss": 2.0492, "step": 7180 }, { "epoch": 0.68, "learning_rate": 7.743547309124472e-05, "loss": 1.5555, "step": 7190 }, { "epoch": 0.69, "learning_rate": 7.740364724229019e-05, "loss": 1.4942, "step": 7200 }, { "epoch": 0.69, "learning_rate": 7.737182139333567e-05, "loss": 1.6958, "step": 7210 }, { "epoch": 0.69, "learning_rate": 7.733999554438114e-05, "loss": 1.4307, "step": 7220 }, { "epoch": 0.69, "learning_rate": 7.730816969542663e-05, "loss": 1.9068, "step": 7230 }, { "epoch": 0.69, "learning_rate": 7.727634384647211e-05, "loss": 1.8229, "step": 7240 }, { "epoch": 0.69, "learning_rate": 7.724451799751758e-05, "loss": 1.6404, "step": 7250 }, { "epoch": 0.69, "learning_rate": 7.721269214856307e-05, "loss": 1.7036, "step": 7260 }, { "epoch": 0.69, "learning_rate": 7.718086629960855e-05, "loss": 1.7163, "step": 7270 }, { "epoch": 0.69, "learning_rate": 7.714904045065402e-05, "loss": 1.6215, "step": 7280 }, { "epoch": 0.69, "learning_rate": 7.71172146016995e-05, "loss": 2.0836, "step": 7290 }, { "epoch": 0.69, "learning_rate": 7.708538875274499e-05, "loss": 1.9861, "step": 7300 }, { "epoch": 0.7, "learning_rate": 7.705356290379047e-05, "loss": 1.8064, "step": 7310 }, { "epoch": 0.7, "learning_rate": 7.702173705483594e-05, "loss": 1.616, "step": 7320 }, { "epoch": 0.7, "learning_rate": 7.698991120588141e-05, "loss": 1.1965, "step": 7330 }, { "epoch": 0.7, "learning_rate": 7.695808535692691e-05, "loss": 1.4957, "step": 7340 }, { "epoch": 0.7, "learning_rate": 7.692625950797238e-05, "loss": 1.8038, "step": 7350 }, { "epoch": 0.7, "learning_rate": 7.689443365901786e-05, "loss": 1.8001, "step": 7360 }, { "epoch": 0.7, "learning_rate": 7.686260781006333e-05, "loss": 1.9065, "step": 7370 }, { "epoch": 0.7, "learning_rate": 7.683078196110881e-05, "loss": 1.8561, "step": 7380 }, { "epoch": 0.7, "learning_rate": 7.67989561121543e-05, "loss": 1.7497, "step": 7390 }, { "epoch": 0.7, "learning_rate": 7.676713026319977e-05, "loss": 1.8024, "step": 7400 }, { "epoch": 0.71, "learning_rate": 7.673530441424525e-05, "loss": 1.666, "step": 7410 }, { "epoch": 0.71, "learning_rate": 7.670347856529074e-05, "loss": 1.4647, "step": 7420 }, { "epoch": 0.71, "learning_rate": 7.667165271633622e-05, "loss": 1.522, "step": 7430 }, { "epoch": 0.71, "learning_rate": 7.663982686738169e-05, "loss": 1.5436, "step": 7440 }, { "epoch": 0.71, "learning_rate": 7.660800101842717e-05, "loss": 1.6896, "step": 7450 }, { "epoch": 0.71, "learning_rate": 7.657617516947266e-05, "loss": 1.4487, "step": 7460 }, { "epoch": 0.71, "learning_rate": 7.654434932051813e-05, "loss": 1.5451, "step": 7470 }, { "epoch": 0.71, "learning_rate": 7.651252347156361e-05, "loss": 1.5389, "step": 7480 }, { "epoch": 0.71, "learning_rate": 7.64806976226091e-05, "loss": 1.856, "step": 7490 }, { "epoch": 0.71, "learning_rate": 7.644887177365456e-05, "loss": 1.6362, "step": 7500 }, { "epoch": 0.71, "learning_rate": 7.641704592470005e-05, "loss": 1.7324, "step": 7510 }, { "epoch": 0.72, "learning_rate": 7.638522007574552e-05, "loss": 1.6882, "step": 7520 }, { "epoch": 0.72, "learning_rate": 7.635339422679102e-05, "loss": 1.7187, "step": 7530 }, { "epoch": 0.72, "learning_rate": 7.632156837783649e-05, "loss": 1.4399, "step": 7540 }, { "epoch": 0.72, "learning_rate": 7.628974252888196e-05, "loss": 1.4401, "step": 7550 }, { "epoch": 0.72, "learning_rate": 7.625791667992744e-05, "loss": 1.5833, "step": 7560 }, { "epoch": 0.72, "learning_rate": 7.622609083097292e-05, "loss": 1.4363, "step": 7570 }, { "epoch": 0.72, "learning_rate": 7.61942649820184e-05, "loss": 1.9323, "step": 7580 }, { "epoch": 0.72, "learning_rate": 7.616243913306388e-05, "loss": 1.4852, "step": 7590 }, { "epoch": 0.72, "learning_rate": 7.613061328410936e-05, "loss": 1.7514, "step": 7600 }, { "epoch": 0.72, "learning_rate": 7.609878743515484e-05, "loss": 1.4138, "step": 7610 }, { "epoch": 0.73, "learning_rate": 7.606696158620031e-05, "loss": 1.7007, "step": 7620 }, { "epoch": 0.73, "learning_rate": 7.60351357372458e-05, "loss": 1.505, "step": 7630 }, { "epoch": 0.73, "learning_rate": 7.600330988829128e-05, "loss": 1.5586, "step": 7640 }, { "epoch": 0.73, "learning_rate": 7.597148403933675e-05, "loss": 1.777, "step": 7650 }, { "epoch": 0.73, "learning_rate": 7.593965819038223e-05, "loss": 1.7, "step": 7660 }, { "epoch": 0.73, "learning_rate": 7.59078323414277e-05, "loss": 1.467, "step": 7670 }, { "epoch": 0.73, "learning_rate": 7.587600649247319e-05, "loss": 1.9871, "step": 7680 }, { "epoch": 0.73, "learning_rate": 7.584418064351867e-05, "loss": 1.7776, "step": 7690 }, { "epoch": 0.73, "learning_rate": 7.581235479456414e-05, "loss": 1.4482, "step": 7700 }, { "epoch": 0.73, "learning_rate": 7.578052894560963e-05, "loss": 1.6553, "step": 7710 }, { "epoch": 0.73, "learning_rate": 7.574870309665511e-05, "loss": 1.6254, "step": 7720 }, { "epoch": 0.74, "learning_rate": 7.571687724770059e-05, "loss": 1.5823, "step": 7730 }, { "epoch": 0.74, "learning_rate": 7.568505139874606e-05, "loss": 1.9779, "step": 7740 }, { "epoch": 0.74, "learning_rate": 7.565322554979153e-05, "loss": 1.6201, "step": 7750 }, { "epoch": 0.74, "learning_rate": 7.562139970083703e-05, "loss": 1.5673, "step": 7760 }, { "epoch": 0.74, "learning_rate": 7.55895738518825e-05, "loss": 1.4576, "step": 7770 }, { "epoch": 0.74, "learning_rate": 7.555774800292798e-05, "loss": 1.8544, "step": 7780 }, { "epoch": 0.74, "learning_rate": 7.552592215397345e-05, "loss": 1.6615, "step": 7790 }, { "epoch": 0.74, "learning_rate": 7.549409630501894e-05, "loss": 1.7856, "step": 7800 }, { "epoch": 0.74, "learning_rate": 7.546227045606442e-05, "loss": 1.7265, "step": 7810 }, { "epoch": 0.74, "learning_rate": 7.543044460710989e-05, "loss": 1.9182, "step": 7820 }, { "epoch": 0.75, "learning_rate": 7.539861875815538e-05, "loss": 2.053, "step": 7830 }, { "epoch": 0.75, "learning_rate": 7.536679290920086e-05, "loss": 1.6568, "step": 7840 }, { "epoch": 0.75, "learning_rate": 7.533496706024633e-05, "loss": 1.7705, "step": 7850 }, { "epoch": 0.75, "learning_rate": 7.530314121129181e-05, "loss": 1.8134, "step": 7860 }, { "epoch": 0.75, "learning_rate": 7.52713153623373e-05, "loss": 1.4194, "step": 7870 }, { "epoch": 0.75, "learning_rate": 7.523948951338278e-05, "loss": 2.0076, "step": 7880 }, { "epoch": 0.75, "learning_rate": 7.520766366442825e-05, "loss": 1.7716, "step": 7890 }, { "epoch": 0.75, "learning_rate": 7.517583781547372e-05, "loss": 1.7052, "step": 7900 }, { "epoch": 0.75, "learning_rate": 7.514401196651922e-05, "loss": 1.6623, "step": 7910 }, { "epoch": 0.75, "learning_rate": 7.511218611756469e-05, "loss": 1.6529, "step": 7920 }, { "epoch": 0.75, "learning_rate": 7.508036026861017e-05, "loss": 1.6233, "step": 7930 }, { "epoch": 0.76, "learning_rate": 7.504853441965564e-05, "loss": 1.7521, "step": 7940 }, { "epoch": 0.76, "learning_rate": 7.501670857070112e-05, "loss": 2.0434, "step": 7950 }, { "epoch": 0.76, "learning_rate": 7.498488272174661e-05, "loss": 1.4489, "step": 7960 }, { "epoch": 0.76, "learning_rate": 7.495305687279208e-05, "loss": 1.475, "step": 7970 }, { "epoch": 0.76, "learning_rate": 7.492123102383756e-05, "loss": 1.3923, "step": 7980 }, { "epoch": 0.76, "learning_rate": 7.488940517488305e-05, "loss": 1.6717, "step": 7990 }, { "epoch": 0.76, "learning_rate": 7.485757932592852e-05, "loss": 1.7205, "step": 8000 }, { "epoch": 0.76, "learning_rate": 7.4825753476974e-05, "loss": 1.5514, "step": 8010 }, { "epoch": 0.76, "learning_rate": 7.479392762801948e-05, "loss": 1.9066, "step": 8020 }, { "epoch": 0.76, "learning_rate": 7.476210177906497e-05, "loss": 1.8837, "step": 8030 }, { "epoch": 0.77, "learning_rate": 7.473027593011044e-05, "loss": 2.0779, "step": 8040 }, { "epoch": 0.77, "learning_rate": 7.469845008115591e-05, "loss": 1.6063, "step": 8050 }, { "epoch": 0.77, "learning_rate": 7.46666242322014e-05, "loss": 1.6868, "step": 8060 }, { "epoch": 0.77, "learning_rate": 7.463479838324687e-05, "loss": 1.6176, "step": 8070 }, { "epoch": 0.77, "learning_rate": 7.460297253429236e-05, "loss": 1.8274, "step": 8080 }, { "epoch": 0.77, "learning_rate": 7.457114668533783e-05, "loss": 1.7037, "step": 8090 }, { "epoch": 0.77, "learning_rate": 7.453932083638333e-05, "loss": 1.9955, "step": 8100 }, { "epoch": 0.77, "learning_rate": 7.45074949874288e-05, "loss": 1.7274, "step": 8110 }, { "epoch": 0.77, "learning_rate": 7.447566913847427e-05, "loss": 1.543, "step": 8120 }, { "epoch": 0.77, "learning_rate": 7.444384328951975e-05, "loss": 1.4292, "step": 8130 }, { "epoch": 0.77, "learning_rate": 7.441201744056523e-05, "loss": 1.6324, "step": 8140 }, { "epoch": 0.78, "learning_rate": 7.438019159161072e-05, "loss": 1.6815, "step": 8150 }, { "epoch": 0.78, "learning_rate": 7.434836574265619e-05, "loss": 1.6445, "step": 8160 }, { "epoch": 0.78, "learning_rate": 7.431653989370167e-05, "loss": 1.3703, "step": 8170 }, { "epoch": 0.78, "learning_rate": 7.428471404474715e-05, "loss": 1.7691, "step": 8180 }, { "epoch": 0.78, "learning_rate": 7.425288819579262e-05, "loss": 1.4671, "step": 8190 }, { "epoch": 0.78, "learning_rate": 7.422106234683811e-05, "loss": 1.7908, "step": 8200 }, { "epoch": 0.78, "learning_rate": 7.418923649788359e-05, "loss": 1.5641, "step": 8210 }, { "epoch": 0.78, "learning_rate": 7.415741064892906e-05, "loss": 1.4575, "step": 8220 }, { "epoch": 0.78, "learning_rate": 7.412558479997454e-05, "loss": 1.6646, "step": 8230 }, { "epoch": 0.78, "learning_rate": 7.409375895102001e-05, "loss": 1.7336, "step": 8240 }, { "epoch": 0.79, "learning_rate": 7.406193310206551e-05, "loss": 1.6668, "step": 8250 }, { "epoch": 0.79, "learning_rate": 7.403010725311098e-05, "loss": 1.589, "step": 8260 }, { "epoch": 0.79, "learning_rate": 7.399828140415645e-05, "loss": 1.8278, "step": 8270 }, { "epoch": 0.79, "learning_rate": 7.396645555520194e-05, "loss": 1.4182, "step": 8280 }, { "epoch": 0.79, "learning_rate": 7.393462970624742e-05, "loss": 1.4516, "step": 8290 }, { "epoch": 0.79, "learning_rate": 7.39028038572929e-05, "loss": 1.6442, "step": 8300 }, { "epoch": 0.79, "learning_rate": 7.387097800833837e-05, "loss": 1.6038, "step": 8310 }, { "epoch": 0.79, "learning_rate": 7.383915215938386e-05, "loss": 1.7188, "step": 8320 }, { "epoch": 0.79, "learning_rate": 7.380732631042934e-05, "loss": 1.856, "step": 8330 }, { "epoch": 0.79, "learning_rate": 7.377550046147481e-05, "loss": 1.4309, "step": 8340 }, { "epoch": 0.79, "learning_rate": 7.37436746125203e-05, "loss": 1.7427, "step": 8350 }, { "epoch": 0.8, "learning_rate": 7.371184876356578e-05, "loss": 1.523, "step": 8360 }, { "epoch": 0.8, "learning_rate": 7.368002291461125e-05, "loss": 1.7628, "step": 8370 }, { "epoch": 0.8, "learning_rate": 7.364819706565673e-05, "loss": 1.5589, "step": 8380 }, { "epoch": 0.8, "learning_rate": 7.36163712167022e-05, "loss": 1.9238, "step": 8390 }, { "epoch": 0.8, "learning_rate": 7.35845453677477e-05, "loss": 1.5053, "step": 8400 }, { "epoch": 0.8, "learning_rate": 7.355271951879317e-05, "loss": 1.6986, "step": 8410 }, { "epoch": 0.8, "learning_rate": 7.352089366983864e-05, "loss": 1.7976, "step": 8420 }, { "epoch": 0.8, "learning_rate": 7.348906782088412e-05, "loss": 2.1376, "step": 8430 }, { "epoch": 0.8, "learning_rate": 7.34572419719296e-05, "loss": 1.6317, "step": 8440 }, { "epoch": 0.8, "learning_rate": 7.342541612297509e-05, "loss": 1.4605, "step": 8450 }, { "epoch": 0.81, "learning_rate": 7.339359027402056e-05, "loss": 1.6641, "step": 8460 }, { "epoch": 0.81, "learning_rate": 7.336176442506604e-05, "loss": 1.4108, "step": 8470 }, { "epoch": 0.81, "learning_rate": 7.332993857611153e-05, "loss": 1.4088, "step": 8480 }, { "epoch": 0.81, "learning_rate": 7.3298112727157e-05, "loss": 1.7873, "step": 8490 }, { "epoch": 0.81, "learning_rate": 7.326628687820248e-05, "loss": 1.8185, "step": 8500 }, { "epoch": 0.81, "learning_rate": 7.323446102924796e-05, "loss": 1.5374, "step": 8510 }, { "epoch": 0.81, "learning_rate": 7.320263518029343e-05, "loss": 1.2006, "step": 8520 }, { "epoch": 0.81, "learning_rate": 7.317080933133892e-05, "loss": 1.452, "step": 8530 }, { "epoch": 0.81, "learning_rate": 7.313898348238439e-05, "loss": 1.5237, "step": 8540 }, { "epoch": 0.81, "learning_rate": 7.310715763342989e-05, "loss": 1.5874, "step": 8550 }, { "epoch": 0.81, "learning_rate": 7.307533178447536e-05, "loss": 1.5766, "step": 8560 }, { "epoch": 0.82, "learning_rate": 7.304350593552083e-05, "loss": 1.7485, "step": 8570 }, { "epoch": 0.82, "learning_rate": 7.301168008656631e-05, "loss": 1.9478, "step": 8580 }, { "epoch": 0.82, "learning_rate": 7.297985423761179e-05, "loss": 1.8084, "step": 8590 }, { "epoch": 0.82, "learning_rate": 7.294802838865728e-05, "loss": 2.0169, "step": 8600 }, { "epoch": 0.82, "learning_rate": 7.291620253970275e-05, "loss": 1.7916, "step": 8610 }, { "epoch": 0.82, "learning_rate": 7.288437669074823e-05, "loss": 1.6838, "step": 8620 }, { "epoch": 0.82, "learning_rate": 7.285255084179371e-05, "loss": 1.7549, "step": 8630 }, { "epoch": 0.82, "learning_rate": 7.282072499283918e-05, "loss": 1.8129, "step": 8640 }, { "epoch": 0.82, "learning_rate": 7.278889914388467e-05, "loss": 1.4067, "step": 8650 }, { "epoch": 0.82, "learning_rate": 7.275707329493015e-05, "loss": 1.4214, "step": 8660 }, { "epoch": 0.83, "learning_rate": 7.272524744597562e-05, "loss": 1.6211, "step": 8670 }, { "epoch": 0.83, "learning_rate": 7.26934215970211e-05, "loss": 1.5487, "step": 8680 }, { "epoch": 0.83, "learning_rate": 7.266159574806658e-05, "loss": 1.6237, "step": 8690 }, { "epoch": 0.83, "learning_rate": 7.262976989911207e-05, "loss": 1.7543, "step": 8700 }, { "epoch": 0.83, "learning_rate": 7.259794405015754e-05, "loss": 1.9502, "step": 8710 }, { "epoch": 0.83, "learning_rate": 7.256611820120303e-05, "loss": 1.3376, "step": 8720 }, { "epoch": 0.83, "learning_rate": 7.25342923522485e-05, "loss": 1.6009, "step": 8730 }, { "epoch": 0.83, "learning_rate": 7.250246650329398e-05, "loss": 2.0273, "step": 8740 }, { "epoch": 0.83, "learning_rate": 7.247064065433946e-05, "loss": 1.8337, "step": 8750 }, { "epoch": 0.83, "learning_rate": 7.243881480538493e-05, "loss": 1.7087, "step": 8760 }, { "epoch": 0.83, "learning_rate": 7.240698895643042e-05, "loss": 1.4726, "step": 8770 }, { "epoch": 0.84, "learning_rate": 7.23751631074759e-05, "loss": 1.3815, "step": 8780 }, { "epoch": 0.84, "learning_rate": 7.234333725852137e-05, "loss": 2.0258, "step": 8790 }, { "epoch": 0.84, "learning_rate": 7.231151140956685e-05, "loss": 1.6538, "step": 8800 }, { "epoch": 0.84, "learning_rate": 7.227968556061232e-05, "loss": 1.6996, "step": 8810 }, { "epoch": 0.84, "learning_rate": 7.224785971165782e-05, "loss": 1.6224, "step": 8820 }, { "epoch": 0.84, "learning_rate": 7.221603386270329e-05, "loss": 1.4107, "step": 8830 }, { "epoch": 0.84, "learning_rate": 7.218420801374876e-05, "loss": 1.6466, "step": 8840 }, { "epoch": 0.84, "learning_rate": 7.215238216479425e-05, "loss": 1.8746, "step": 8850 }, { "epoch": 0.84, "learning_rate": 7.212055631583973e-05, "loss": 1.8413, "step": 8860 }, { "epoch": 0.84, "learning_rate": 7.208873046688521e-05, "loss": 1.722, "step": 8870 }, { "epoch": 0.85, "learning_rate": 7.205690461793068e-05, "loss": 1.9583, "step": 8880 }, { "epoch": 0.85, "learning_rate": 7.202507876897617e-05, "loss": 1.7814, "step": 8890 }, { "epoch": 0.85, "learning_rate": 7.199325292002165e-05, "loss": 1.5631, "step": 8900 }, { "epoch": 0.85, "learning_rate": 7.196142707106712e-05, "loss": 1.2928, "step": 8910 }, { "epoch": 0.85, "learning_rate": 7.19296012221126e-05, "loss": 1.6888, "step": 8920 }, { "epoch": 0.85, "learning_rate": 7.189777537315809e-05, "loss": 1.415, "step": 8930 }, { "epoch": 0.85, "learning_rate": 7.186594952420356e-05, "loss": 1.8556, "step": 8940 }, { "epoch": 0.85, "learning_rate": 7.183412367524904e-05, "loss": 1.4755, "step": 8950 }, { "epoch": 0.85, "learning_rate": 7.180229782629451e-05, "loss": 1.8606, "step": 8960 }, { "epoch": 0.85, "learning_rate": 7.177047197734001e-05, "loss": 1.5918, "step": 8970 }, { "epoch": 0.85, "learning_rate": 7.173864612838548e-05, "loss": 1.6491, "step": 8980 }, { "epoch": 0.86, "learning_rate": 7.170682027943095e-05, "loss": 1.1818, "step": 8990 }, { "epoch": 0.86, "learning_rate": 7.167499443047643e-05, "loss": 1.6124, "step": 9000 }, { "epoch": 0.86, "learning_rate": 7.164316858152192e-05, "loss": 1.5302, "step": 9010 }, { "epoch": 0.86, "learning_rate": 7.16113427325674e-05, "loss": 1.6171, "step": 9020 }, { "epoch": 0.86, "learning_rate": 7.157951688361287e-05, "loss": 1.5863, "step": 9030 }, { "epoch": 0.86, "learning_rate": 7.154769103465835e-05, "loss": 1.6996, "step": 9040 }, { "epoch": 0.86, "learning_rate": 7.151586518570384e-05, "loss": 1.6777, "step": 9050 }, { "epoch": 0.86, "learning_rate": 7.148403933674931e-05, "loss": 1.2677, "step": 9060 }, { "epoch": 0.86, "learning_rate": 7.145221348779479e-05, "loss": 1.704, "step": 9070 }, { "epoch": 0.86, "learning_rate": 7.142038763884027e-05, "loss": 1.5171, "step": 9080 }, { "epoch": 0.87, "learning_rate": 7.138856178988574e-05, "loss": 1.4865, "step": 9090 }, { "epoch": 0.87, "learning_rate": 7.135673594093123e-05, "loss": 1.397, "step": 9100 }, { "epoch": 0.87, "learning_rate": 7.13249100919767e-05, "loss": 1.3799, "step": 9110 }, { "epoch": 0.87, "learning_rate": 7.12930842430222e-05, "loss": 1.7878, "step": 9120 }, { "epoch": 0.87, "learning_rate": 7.126125839406767e-05, "loss": 1.7783, "step": 9130 }, { "epoch": 0.87, "learning_rate": 7.122943254511314e-05, "loss": 1.7521, "step": 9140 }, { "epoch": 0.87, "learning_rate": 7.119760669615862e-05, "loss": 1.7474, "step": 9150 }, { "epoch": 0.87, "learning_rate": 7.11657808472041e-05, "loss": 1.7479, "step": 9160 }, { "epoch": 0.87, "learning_rate": 7.113395499824959e-05, "loss": 1.7436, "step": 9170 }, { "epoch": 0.87, "learning_rate": 7.110212914929506e-05, "loss": 1.9751, "step": 9180 }, { "epoch": 0.87, "learning_rate": 7.107030330034054e-05, "loss": 1.5624, "step": 9190 }, { "epoch": 0.88, "learning_rate": 7.103847745138602e-05, "loss": 1.6065, "step": 9200 }, { "epoch": 0.88, "learning_rate": 7.10066516024315e-05, "loss": 1.553, "step": 9210 }, { "epoch": 0.88, "learning_rate": 7.097482575347698e-05, "loss": 1.8031, "step": 9220 }, { "epoch": 0.88, "learning_rate": 7.094299990452246e-05, "loss": 1.385, "step": 9230 }, { "epoch": 0.88, "learning_rate": 7.091117405556793e-05, "loss": 1.7275, "step": 9240 }, { "epoch": 0.88, "learning_rate": 7.087934820661342e-05, "loss": 1.288, "step": 9250 }, { "epoch": 0.88, "learning_rate": 7.084752235765889e-05, "loss": 1.8881, "step": 9260 }, { "epoch": 0.88, "learning_rate": 7.081569650870438e-05, "loss": 1.4875, "step": 9270 }, { "epoch": 0.88, "learning_rate": 7.078387065974985e-05, "loss": 1.9456, "step": 9280 }, { "epoch": 0.88, "learning_rate": 7.075204481079532e-05, "loss": 1.7929, "step": 9290 }, { "epoch": 0.89, "learning_rate": 7.07202189618408e-05, "loss": 1.6181, "step": 9300 }, { "epoch": 0.89, "learning_rate": 7.068839311288629e-05, "loss": 1.6676, "step": 9310 }, { "epoch": 0.89, "learning_rate": 7.065656726393177e-05, "loss": 1.4524, "step": 9320 }, { "epoch": 0.89, "learning_rate": 7.062474141497724e-05, "loss": 1.5569, "step": 9330 }, { "epoch": 0.89, "learning_rate": 7.059291556602273e-05, "loss": 1.542, "step": 9340 }, { "epoch": 0.89, "learning_rate": 7.056108971706821e-05, "loss": 1.4374, "step": 9350 }, { "epoch": 0.89, "learning_rate": 7.052926386811368e-05, "loss": 1.6697, "step": 9360 }, { "epoch": 0.89, "learning_rate": 7.049743801915916e-05, "loss": 1.5604, "step": 9370 }, { "epoch": 0.89, "learning_rate": 7.046561217020465e-05, "loss": 1.8572, "step": 9380 }, { "epoch": 0.89, "learning_rate": 7.043378632125013e-05, "loss": 2.0608, "step": 9390 }, { "epoch": 0.89, "learning_rate": 7.04019604722956e-05, "loss": 1.6197, "step": 9400 }, { "epoch": 0.9, "learning_rate": 7.037013462334107e-05, "loss": 1.5616, "step": 9410 }, { "epoch": 0.9, "learning_rate": 7.033830877438657e-05, "loss": 1.1327, "step": 9420 }, { "epoch": 0.9, "learning_rate": 7.030648292543204e-05, "loss": 1.7424, "step": 9430 }, { "epoch": 0.9, "learning_rate": 7.027465707647752e-05, "loss": 1.4176, "step": 9440 }, { "epoch": 0.9, "learning_rate": 7.024283122752299e-05, "loss": 1.6296, "step": 9450 }, { "epoch": 0.9, "learning_rate": 7.021100537856848e-05, "loss": 1.7058, "step": 9460 }, { "epoch": 0.9, "learning_rate": 7.017917952961396e-05, "loss": 1.743, "step": 9470 }, { "epoch": 0.9, "learning_rate": 7.014735368065943e-05, "loss": 1.896, "step": 9480 }, { "epoch": 0.9, "learning_rate": 7.011552783170491e-05, "loss": 1.3877, "step": 9490 }, { "epoch": 0.9, "learning_rate": 7.00837019827504e-05, "loss": 1.7929, "step": 9500 }, { "epoch": 0.91, "learning_rate": 7.005187613379587e-05, "loss": 1.887, "step": 9510 }, { "epoch": 0.91, "learning_rate": 7.002005028484135e-05, "loss": 1.6872, "step": 9520 }, { "epoch": 0.91, "learning_rate": 6.998822443588684e-05, "loss": 1.4152, "step": 9530 }, { "epoch": 0.91, "learning_rate": 6.995639858693232e-05, "loss": 1.4784, "step": 9540 }, { "epoch": 0.91, "learning_rate": 6.992457273797779e-05, "loss": 1.8467, "step": 9550 }, { "epoch": 0.91, "learning_rate": 6.989274688902326e-05, "loss": 1.7269, "step": 9560 }, { "epoch": 0.91, "learning_rate": 6.986092104006876e-05, "loss": 1.7993, "step": 9570 }, { "epoch": 0.91, "learning_rate": 6.982909519111423e-05, "loss": 1.6533, "step": 9580 }, { "epoch": 0.91, "learning_rate": 6.979726934215971e-05, "loss": 1.8264, "step": 9590 }, { "epoch": 0.91, "learning_rate": 6.976544349320518e-05, "loss": 1.4738, "step": 9600 }, { "epoch": 0.91, "learning_rate": 6.973361764425066e-05, "loss": 1.515, "step": 9610 }, { "epoch": 0.92, "learning_rate": 6.970179179529615e-05, "loss": 1.9079, "step": 9620 }, { "epoch": 0.92, "learning_rate": 6.966996594634162e-05, "loss": 1.8671, "step": 9630 }, { "epoch": 0.92, "learning_rate": 6.96381400973871e-05, "loss": 1.6265, "step": 9640 }, { "epoch": 0.92, "learning_rate": 6.960631424843258e-05, "loss": 1.568, "step": 9650 }, { "epoch": 0.92, "learning_rate": 6.957448839947805e-05, "loss": 1.8229, "step": 9660 }, { "epoch": 0.92, "learning_rate": 6.954266255052354e-05, "loss": 1.5253, "step": 9670 }, { "epoch": 0.92, "learning_rate": 6.951083670156902e-05, "loss": 1.9236, "step": 9680 }, { "epoch": 0.92, "learning_rate": 6.94790108526145e-05, "loss": 2.1358, "step": 9690 }, { "epoch": 0.92, "learning_rate": 6.944718500365998e-05, "loss": 1.8796, "step": 9700 }, { "epoch": 0.92, "learning_rate": 6.941535915470545e-05, "loss": 1.5388, "step": 9710 }, { "epoch": 0.93, "learning_rate": 6.938353330575094e-05, "loss": 1.6827, "step": 9720 }, { "epoch": 0.93, "learning_rate": 6.935170745679641e-05, "loss": 1.7603, "step": 9730 }, { "epoch": 0.93, "learning_rate": 6.93198816078419e-05, "loss": 1.3614, "step": 9740 }, { "epoch": 0.93, "learning_rate": 6.928805575888737e-05, "loss": 1.7216, "step": 9750 }, { "epoch": 0.93, "learning_rate": 6.925622990993285e-05, "loss": 1.3782, "step": 9760 }, { "epoch": 0.93, "learning_rate": 6.922440406097833e-05, "loss": 1.6877, "step": 9770 }, { "epoch": 0.93, "learning_rate": 6.91925782120238e-05, "loss": 1.5273, "step": 9780 }, { "epoch": 0.93, "learning_rate": 6.916075236306929e-05, "loss": 1.4268, "step": 9790 }, { "epoch": 0.93, "learning_rate": 6.912892651411477e-05, "loss": 1.6359, "step": 9800 }, { "epoch": 0.93, "learning_rate": 6.909710066516024e-05, "loss": 1.6763, "step": 9810 }, { "epoch": 0.93, "learning_rate": 6.906527481620573e-05, "loss": 1.4216, "step": 9820 }, { "epoch": 0.94, "learning_rate": 6.903344896725121e-05, "loss": 1.4369, "step": 9830 }, { "epoch": 0.94, "learning_rate": 6.900162311829669e-05, "loss": 1.4794, "step": 9840 }, { "epoch": 0.94, "learning_rate": 6.896979726934216e-05, "loss": 1.7834, "step": 9850 }, { "epoch": 0.94, "learning_rate": 6.893797142038763e-05, "loss": 1.6101, "step": 9860 }, { "epoch": 0.94, "learning_rate": 6.890614557143312e-05, "loss": 1.9098, "step": 9870 }, { "epoch": 0.94, "learning_rate": 6.88743197224786e-05, "loss": 1.5039, "step": 9880 }, { "epoch": 0.94, "learning_rate": 6.884249387352408e-05, "loss": 1.6931, "step": 9890 }, { "epoch": 0.94, "learning_rate": 6.881066802456955e-05, "loss": 1.9016, "step": 9900 }, { "epoch": 0.94, "learning_rate": 6.877884217561504e-05, "loss": 1.6264, "step": 9910 }, { "epoch": 0.94, "learning_rate": 6.874701632666052e-05, "loss": 2.0439, "step": 9920 }, { "epoch": 0.95, "learning_rate": 6.871519047770599e-05, "loss": 1.8369, "step": 9930 }, { "epoch": 0.95, "learning_rate": 6.868336462875147e-05, "loss": 1.52, "step": 9940 }, { "epoch": 0.95, "learning_rate": 6.865153877979696e-05, "loss": 1.3288, "step": 9950 }, { "epoch": 0.95, "learning_rate": 6.861971293084243e-05, "loss": 1.3007, "step": 9960 }, { "epoch": 0.95, "learning_rate": 6.858788708188791e-05, "loss": 1.712, "step": 9970 }, { "epoch": 0.95, "learning_rate": 6.855606123293338e-05, "loss": 1.6743, "step": 9980 }, { "epoch": 0.95, "learning_rate": 6.852423538397888e-05, "loss": 1.7357, "step": 9990 }, { "epoch": 0.95, "learning_rate": 6.849240953502435e-05, "loss": 1.3981, "step": 10000 }, { "epoch": 0.95, "learning_rate": 6.846058368606983e-05, "loss": 1.6568, "step": 10010 }, { "epoch": 0.95, "learning_rate": 6.84287578371153e-05, "loss": 1.7871, "step": 10020 }, { "epoch": 0.95, "learning_rate": 6.839693198816079e-05, "loss": 1.5819, "step": 10030 }, { "epoch": 0.96, "learning_rate": 6.836510613920627e-05, "loss": 1.1312, "step": 10040 }, { "epoch": 0.96, "learning_rate": 6.833328029025174e-05, "loss": 1.7113, "step": 10050 }, { "epoch": 0.96, "learning_rate": 6.830145444129722e-05, "loss": 1.4394, "step": 10060 }, { "epoch": 0.96, "learning_rate": 6.826962859234271e-05, "loss": 1.2527, "step": 10070 }, { "epoch": 0.96, "learning_rate": 6.823780274338818e-05, "loss": 1.303, "step": 10080 }, { "epoch": 0.96, "learning_rate": 6.820597689443366e-05, "loss": 1.8516, "step": 10090 }, { "epoch": 0.96, "learning_rate": 6.817415104547914e-05, "loss": 1.6264, "step": 10100 }, { "epoch": 0.96, "learning_rate": 6.814232519652463e-05, "loss": 1.6503, "step": 10110 }, { "epoch": 0.96, "learning_rate": 6.81104993475701e-05, "loss": 1.6317, "step": 10120 }, { "epoch": 0.96, "learning_rate": 6.807867349861557e-05, "loss": 1.9518, "step": 10130 }, { "epoch": 0.97, "learning_rate": 6.804684764966107e-05, "loss": 1.7174, "step": 10140 }, { "epoch": 0.97, "learning_rate": 6.801502180070654e-05, "loss": 1.524, "step": 10150 }, { "epoch": 0.97, "learning_rate": 6.798319595175202e-05, "loss": 1.3066, "step": 10160 }, { "epoch": 0.97, "learning_rate": 6.795137010279749e-05, "loss": 1.5514, "step": 10170 }, { "epoch": 0.97, "learning_rate": 6.791954425384297e-05, "loss": 1.8562, "step": 10180 }, { "epoch": 0.97, "learning_rate": 6.788771840488846e-05, "loss": 1.5758, "step": 10190 }, { "epoch": 0.97, "learning_rate": 6.785589255593393e-05, "loss": 1.7807, "step": 10200 }, { "epoch": 0.97, "learning_rate": 6.782406670697941e-05, "loss": 1.5524, "step": 10210 }, { "epoch": 0.97, "learning_rate": 6.77922408580249e-05, "loss": 1.48, "step": 10220 }, { "epoch": 0.97, "learning_rate": 6.776041500907036e-05, "loss": 1.6014, "step": 10230 }, { "epoch": 0.97, "learning_rate": 6.772858916011585e-05, "loss": 1.6446, "step": 10240 }, { "epoch": 0.98, "learning_rate": 6.769676331116133e-05, "loss": 1.4452, "step": 10250 }, { "epoch": 0.98, "learning_rate": 6.766493746220682e-05, "loss": 1.4993, "step": 10260 }, { "epoch": 0.98, "learning_rate": 6.763311161325229e-05, "loss": 1.4982, "step": 10270 }, { "epoch": 0.98, "learning_rate": 6.760128576429776e-05, "loss": 1.651, "step": 10280 }, { "epoch": 0.98, "learning_rate": 6.756945991534325e-05, "loss": 1.6332, "step": 10290 }, { "epoch": 0.98, "learning_rate": 6.753763406638872e-05, "loss": 1.4433, "step": 10300 }, { "epoch": 0.98, "learning_rate": 6.75058082174342e-05, "loss": 1.485, "step": 10310 }, { "epoch": 0.98, "learning_rate": 6.747398236847968e-05, "loss": 1.6712, "step": 10320 }, { "epoch": 0.98, "learning_rate": 6.744215651952516e-05, "loss": 1.5196, "step": 10330 }, { "epoch": 0.98, "learning_rate": 6.741033067057064e-05, "loss": 1.586, "step": 10340 }, { "epoch": 0.99, "learning_rate": 6.737850482161611e-05, "loss": 1.4815, "step": 10350 }, { "epoch": 0.99, "learning_rate": 6.73466789726616e-05, "loss": 1.7372, "step": 10360 }, { "epoch": 0.99, "learning_rate": 6.731485312370708e-05, "loss": 1.2242, "step": 10370 }, { "epoch": 0.99, "learning_rate": 6.728302727475255e-05, "loss": 1.8329, "step": 10380 }, { "epoch": 0.99, "learning_rate": 6.725120142579803e-05, "loss": 1.6306, "step": 10390 }, { "epoch": 0.99, "learning_rate": 6.721937557684352e-05, "loss": 1.6169, "step": 10400 }, { "epoch": 0.99, "learning_rate": 6.7187549727889e-05, "loss": 1.8023, "step": 10410 }, { "epoch": 0.99, "learning_rate": 6.715572387893447e-05, "loss": 1.4764, "step": 10420 }, { "epoch": 0.99, "learning_rate": 6.712389802997994e-05, "loss": 1.718, "step": 10430 }, { "epoch": 0.99, "learning_rate": 6.709207218102544e-05, "loss": 1.6362, "step": 10440 }, { "epoch": 0.99, "learning_rate": 6.706024633207091e-05, "loss": 1.4818, "step": 10450 }, { "epoch": 1.0, "learning_rate": 6.70284204831164e-05, "loss": 1.8302, "step": 10460 }, { "epoch": 1.0, "learning_rate": 6.699659463416186e-05, "loss": 1.4897, "step": 10470 }, { "epoch": 1.0, "learning_rate": 6.696476878520735e-05, "loss": 1.3032, "step": 10480 }, { "epoch": 1.0, "learning_rate": 6.693294293625283e-05, "loss": 1.4236, "step": 10490 }, { "epoch": 1.0, "learning_rate": 6.69011170872983e-05, "loss": 1.5428, "step": 10500 }, { "epoch": 1.0, "eval_loss": 1.5751161575317383, "eval_runtime": 4381.6742, "eval_samples_per_second": 2.433, "step": 10507 }, { "epoch": 1.0, "learning_rate": 6.686929123834378e-05, "loss": 1.3253, "step": 10510 }, { "epoch": 1.0, "learning_rate": 6.683746538938927e-05, "loss": 1.2831, "step": 10520 }, { "epoch": 1.0, "learning_rate": 6.680563954043474e-05, "loss": 1.4592, "step": 10530 }, { "epoch": 1.0, "learning_rate": 6.677381369148022e-05, "loss": 1.3726, "step": 10540 }, { "epoch": 1.0, "learning_rate": 6.67419878425257e-05, "loss": 1.3083, "step": 10550 }, { "epoch": 1.01, "learning_rate": 6.671016199357119e-05, "loss": 1.1417, "step": 10560 }, { "epoch": 1.01, "learning_rate": 6.667833614461666e-05, "loss": 1.8066, "step": 10570 }, { "epoch": 1.01, "learning_rate": 6.664651029566213e-05, "loss": 1.3587, "step": 10580 }, { "epoch": 1.01, "learning_rate": 6.661468444670763e-05, "loss": 1.5318, "step": 10590 }, { "epoch": 1.01, "learning_rate": 6.65828585977531e-05, "loss": 1.3035, "step": 10600 }, { "epoch": 1.01, "learning_rate": 6.655103274879858e-05, "loss": 1.2573, "step": 10610 }, { "epoch": 1.01, "learning_rate": 6.651920689984405e-05, "loss": 1.2519, "step": 10620 }, { "epoch": 1.01, "learning_rate": 6.648738105088955e-05, "loss": 1.0242, "step": 10630 }, { "epoch": 1.01, "learning_rate": 6.645555520193502e-05, "loss": 1.0862, "step": 10640 }, { "epoch": 1.01, "learning_rate": 6.642372935298049e-05, "loss": 1.3527, "step": 10650 }, { "epoch": 1.01, "learning_rate": 6.639190350402597e-05, "loss": 1.16, "step": 10660 }, { "epoch": 1.02, "learning_rate": 6.636007765507145e-05, "loss": 1.5568, "step": 10670 }, { "epoch": 1.02, "learning_rate": 6.632825180611694e-05, "loss": 1.2472, "step": 10680 }, { "epoch": 1.02, "learning_rate": 6.629642595716241e-05, "loss": 1.3669, "step": 10690 }, { "epoch": 1.02, "learning_rate": 6.626460010820789e-05, "loss": 0.9845, "step": 10700 }, { "epoch": 1.02, "learning_rate": 6.623277425925338e-05, "loss": 1.1218, "step": 10710 }, { "epoch": 1.02, "learning_rate": 6.620094841029885e-05, "loss": 1.4078, "step": 10720 }, { "epoch": 1.02, "learning_rate": 6.616912256134433e-05, "loss": 1.2828, "step": 10730 }, { "epoch": 1.02, "learning_rate": 6.613729671238981e-05, "loss": 1.3693, "step": 10740 }, { "epoch": 1.02, "learning_rate": 6.610547086343528e-05, "loss": 1.4504, "step": 10750 }, { "epoch": 1.02, "learning_rate": 6.607364501448077e-05, "loss": 1.2069, "step": 10760 }, { "epoch": 1.03, "learning_rate": 6.604181916552624e-05, "loss": 1.3613, "step": 10770 }, { "epoch": 1.03, "learning_rate": 6.600999331657173e-05, "loss": 1.4412, "step": 10780 }, { "epoch": 1.03, "learning_rate": 6.59781674676172e-05, "loss": 1.4046, "step": 10790 }, { "epoch": 1.03, "learning_rate": 6.594634161866267e-05, "loss": 1.2768, "step": 10800 }, { "epoch": 1.03, "learning_rate": 6.591451576970816e-05, "loss": 1.3946, "step": 10810 }, { "epoch": 1.03, "learning_rate": 6.588268992075364e-05, "loss": 1.034, "step": 10820 }, { "epoch": 1.03, "learning_rate": 6.585086407179913e-05, "loss": 1.1968, "step": 10830 }, { "epoch": 1.03, "learning_rate": 6.58190382228446e-05, "loss": 1.0514, "step": 10840 }, { "epoch": 1.03, "learning_rate": 6.578721237389008e-05, "loss": 1.3141, "step": 10850 }, { "epoch": 1.03, "learning_rate": 6.575538652493556e-05, "loss": 1.0505, "step": 10860 }, { "epoch": 1.03, "learning_rate": 6.572356067598103e-05, "loss": 1.3938, "step": 10870 }, { "epoch": 1.04, "learning_rate": 6.569173482702652e-05, "loss": 1.2018, "step": 10880 }, { "epoch": 1.04, "learning_rate": 6.5659908978072e-05, "loss": 1.5693, "step": 10890 }, { "epoch": 1.04, "learning_rate": 6.562808312911747e-05, "loss": 1.089, "step": 10900 }, { "epoch": 1.04, "learning_rate": 6.559625728016295e-05, "loss": 1.2522, "step": 10910 }, { "epoch": 1.04, "learning_rate": 6.556443143120842e-05, "loss": 1.0244, "step": 10920 }, { "epoch": 1.04, "learning_rate": 6.553260558225391e-05, "loss": 1.6794, "step": 10930 }, { "epoch": 1.04, "learning_rate": 6.550077973329939e-05, "loss": 1.3156, "step": 10940 }, { "epoch": 1.04, "learning_rate": 6.546895388434486e-05, "loss": 1.3705, "step": 10950 }, { "epoch": 1.04, "learning_rate": 6.543712803539034e-05, "loss": 1.2222, "step": 10960 }, { "epoch": 1.04, "learning_rate": 6.540530218643583e-05, "loss": 1.2035, "step": 10970 }, { "epoch": 1.05, "learning_rate": 6.537347633748131e-05, "loss": 1.3753, "step": 10980 }, { "epoch": 1.05, "learning_rate": 6.534165048852678e-05, "loss": 1.3482, "step": 10990 }, { "epoch": 1.05, "learning_rate": 6.530982463957225e-05, "loss": 1.6365, "step": 11000 }, { "epoch": 1.05, "learning_rate": 6.527799879061775e-05, "loss": 1.4579, "step": 11010 }, { "epoch": 1.05, "learning_rate": 6.524617294166322e-05, "loss": 1.389, "step": 11020 }, { "epoch": 1.05, "learning_rate": 6.52143470927087e-05, "loss": 1.7417, "step": 11030 }, { "epoch": 1.05, "learning_rate": 6.518252124375417e-05, "loss": 1.1974, "step": 11040 }, { "epoch": 1.05, "learning_rate": 6.515069539479966e-05, "loss": 1.281, "step": 11050 }, { "epoch": 1.05, "learning_rate": 6.511886954584514e-05, "loss": 1.4675, "step": 11060 }, { "epoch": 1.05, "learning_rate": 6.508704369689061e-05, "loss": 1.2103, "step": 11070 }, { "epoch": 1.05, "learning_rate": 6.50552178479361e-05, "loss": 1.2186, "step": 11080 }, { "epoch": 1.06, "learning_rate": 6.502339199898158e-05, "loss": 1.4346, "step": 11090 }, { "epoch": 1.06, "learning_rate": 6.499156615002705e-05, "loss": 1.4353, "step": 11100 }, { "epoch": 1.06, "learning_rate": 6.495974030107253e-05, "loss": 0.9587, "step": 11110 }, { "epoch": 1.06, "learning_rate": 6.492791445211802e-05, "loss": 1.2586, "step": 11120 }, { "epoch": 1.06, "learning_rate": 6.48960886031635e-05, "loss": 1.3769, "step": 11130 }, { "epoch": 1.06, "learning_rate": 6.486426275420897e-05, "loss": 1.0935, "step": 11140 }, { "epoch": 1.06, "learning_rate": 6.483243690525444e-05, "loss": 1.1413, "step": 11150 }, { "epoch": 1.06, "learning_rate": 6.480061105629994e-05, "loss": 1.4108, "step": 11160 }, { "epoch": 1.06, "learning_rate": 6.47687852073454e-05, "loss": 1.5105, "step": 11170 }, { "epoch": 1.06, "learning_rate": 6.473695935839089e-05, "loss": 1.3528, "step": 11180 }, { "epoch": 1.07, "learning_rate": 6.470513350943636e-05, "loss": 1.1827, "step": 11190 }, { "epoch": 1.07, "learning_rate": 6.467330766048184e-05, "loss": 1.6177, "step": 11200 }, { "epoch": 1.07, "learning_rate": 6.464148181152733e-05, "loss": 1.6265, "step": 11210 }, { "epoch": 1.07, "learning_rate": 6.46096559625728e-05, "loss": 1.4097, "step": 11220 }, { "epoch": 1.07, "learning_rate": 6.457783011361828e-05, "loss": 1.5653, "step": 11230 }, { "epoch": 1.07, "learning_rate": 6.454600426466376e-05, "loss": 1.1577, "step": 11240 }, { "epoch": 1.07, "learning_rate": 6.451417841570923e-05, "loss": 1.1006, "step": 11250 }, { "epoch": 1.07, "learning_rate": 6.448235256675472e-05, "loss": 1.163, "step": 11260 }, { "epoch": 1.07, "learning_rate": 6.44505267178002e-05, "loss": 1.4222, "step": 11270 }, { "epoch": 1.07, "learning_rate": 6.441870086884569e-05, "loss": 1.6539, "step": 11280 }, { "epoch": 1.07, "learning_rate": 6.438687501989116e-05, "loss": 1.5317, "step": 11290 }, { "epoch": 1.08, "learning_rate": 6.435504917093664e-05, "loss": 1.4867, "step": 11300 }, { "epoch": 1.08, "learning_rate": 6.432322332198212e-05, "loss": 1.577, "step": 11310 }, { "epoch": 1.08, "learning_rate": 6.42913974730276e-05, "loss": 1.3068, "step": 11320 }, { "epoch": 1.08, "learning_rate": 6.425957162407308e-05, "loss": 1.3321, "step": 11330 }, { "epoch": 1.08, "learning_rate": 6.422774577511855e-05, "loss": 1.2002, "step": 11340 }, { "epoch": 1.08, "learning_rate": 6.419591992616404e-05, "loss": 1.3479, "step": 11350 }, { "epoch": 1.08, "learning_rate": 6.416409407720951e-05, "loss": 1.4076, "step": 11360 }, { "epoch": 1.08, "learning_rate": 6.413226822825498e-05, "loss": 1.4901, "step": 11370 }, { "epoch": 1.08, "learning_rate": 6.410044237930047e-05, "loss": 0.9651, "step": 11380 }, { "epoch": 1.08, "learning_rate": 6.406861653034595e-05, "loss": 1.462, "step": 11390 }, { "epoch": 1.08, "learning_rate": 6.403679068139144e-05, "loss": 1.3459, "step": 11400 }, { "epoch": 1.09, "learning_rate": 6.40049648324369e-05, "loss": 1.1993, "step": 11410 }, { "epoch": 1.09, "learning_rate": 6.397313898348239e-05, "loss": 1.4077, "step": 11420 }, { "epoch": 1.09, "learning_rate": 6.394131313452787e-05, "loss": 1.0938, "step": 11430 }, { "epoch": 1.09, "learning_rate": 6.390948728557334e-05, "loss": 1.4993, "step": 11440 }, { "epoch": 1.09, "learning_rate": 6.387766143661883e-05, "loss": 1.31, "step": 11450 }, { "epoch": 1.09, "learning_rate": 6.384583558766431e-05, "loss": 1.5857, "step": 11460 }, { "epoch": 1.09, "learning_rate": 6.381400973870978e-05, "loss": 1.3809, "step": 11470 }, { "epoch": 1.09, "learning_rate": 6.378218388975526e-05, "loss": 1.6283, "step": 11480 }, { "epoch": 1.09, "learning_rate": 6.375035804080073e-05, "loss": 1.2548, "step": 11490 }, { "epoch": 1.09, "learning_rate": 6.371853219184623e-05, "loss": 1.2578, "step": 11500 }, { "epoch": 1.1, "learning_rate": 6.36867063428917e-05, "loss": 1.3211, "step": 11510 }, { "epoch": 1.1, "learning_rate": 6.365488049393717e-05, "loss": 1.1977, "step": 11520 }, { "epoch": 1.1, "learning_rate": 6.362305464498265e-05, "loss": 1.2643, "step": 11530 }, { "epoch": 1.1, "learning_rate": 6.359122879602814e-05, "loss": 1.3753, "step": 11540 }, { "epoch": 1.1, "learning_rate": 6.355940294707362e-05, "loss": 1.3017, "step": 11550 }, { "epoch": 1.1, "learning_rate": 6.352757709811909e-05, "loss": 1.3721, "step": 11560 }, { "epoch": 1.1, "learning_rate": 6.349575124916458e-05, "loss": 1.6016, "step": 11570 }, { "epoch": 1.1, "learning_rate": 6.346392540021006e-05, "loss": 1.2166, "step": 11580 }, { "epoch": 1.1, "learning_rate": 6.343209955125553e-05, "loss": 1.5257, "step": 11590 }, { "epoch": 1.1, "learning_rate": 6.340027370230101e-05, "loss": 1.5632, "step": 11600 }, { "epoch": 1.1, "learning_rate": 6.33684478533465e-05, "loss": 1.3745, "step": 11610 }, { "epoch": 1.11, "learning_rate": 6.333662200439197e-05, "loss": 1.5288, "step": 11620 }, { "epoch": 1.11, "learning_rate": 6.330479615543745e-05, "loss": 1.2667, "step": 11630 }, { "epoch": 1.11, "learning_rate": 6.327297030648292e-05, "loss": 1.5822, "step": 11640 }, { "epoch": 1.11, "learning_rate": 6.324114445752842e-05, "loss": 1.0847, "step": 11650 }, { "epoch": 1.11, "learning_rate": 6.320931860857389e-05, "loss": 1.503, "step": 11660 }, { "epoch": 1.11, "learning_rate": 6.317749275961936e-05, "loss": 1.4796, "step": 11670 }, { "epoch": 1.11, "learning_rate": 6.314566691066484e-05, "loss": 1.1655, "step": 11680 }, { "epoch": 1.11, "learning_rate": 6.311384106171033e-05, "loss": 1.213, "step": 11690 }, { "epoch": 1.11, "learning_rate": 6.308201521275581e-05, "loss": 1.5796, "step": 11700 }, { "epoch": 1.11, "learning_rate": 6.305018936380128e-05, "loss": 1.1776, "step": 11710 }, { "epoch": 1.12, "learning_rate": 6.301836351484676e-05, "loss": 1.2745, "step": 11720 }, { "epoch": 1.12, "learning_rate": 6.298653766589225e-05, "loss": 1.5491, "step": 11730 }, { "epoch": 1.12, "learning_rate": 6.295471181693772e-05, "loss": 1.3805, "step": 11740 }, { "epoch": 1.12, "learning_rate": 6.29228859679832e-05, "loss": 1.4238, "step": 11750 }, { "epoch": 1.12, "learning_rate": 6.289106011902868e-05, "loss": 1.3267, "step": 11760 }, { "epoch": 1.12, "learning_rate": 6.285923427007415e-05, "loss": 1.2011, "step": 11770 }, { "epoch": 1.12, "learning_rate": 6.282740842111964e-05, "loss": 1.2998, "step": 11780 }, { "epoch": 1.12, "learning_rate": 6.279558257216511e-05, "loss": 1.0914, "step": 11790 }, { "epoch": 1.12, "learning_rate": 6.27637567232106e-05, "loss": 1.4221, "step": 11800 }, { "epoch": 1.12, "learning_rate": 6.273193087425607e-05, "loss": 1.4707, "step": 11810 }, { "epoch": 1.12, "learning_rate": 6.270010502530154e-05, "loss": 1.2843, "step": 11820 }, { "epoch": 1.13, "learning_rate": 6.266827917634703e-05, "loss": 1.5337, "step": 11830 }, { "epoch": 1.13, "learning_rate": 6.263645332739251e-05, "loss": 1.3396, "step": 11840 }, { "epoch": 1.13, "learning_rate": 6.2604627478438e-05, "loss": 1.0909, "step": 11850 }, { "epoch": 1.13, "learning_rate": 6.257280162948347e-05, "loss": 1.4662, "step": 11860 }, { "epoch": 1.13, "learning_rate": 6.254097578052895e-05, "loss": 1.1194, "step": 11870 }, { "epoch": 1.13, "learning_rate": 6.250914993157443e-05, "loss": 1.2887, "step": 11880 }, { "epoch": 1.13, "learning_rate": 6.24773240826199e-05, "loss": 1.4379, "step": 11890 }, { "epoch": 1.13, "learning_rate": 6.244549823366539e-05, "loss": 1.2255, "step": 11900 }, { "epoch": 1.13, "learning_rate": 6.241367238471087e-05, "loss": 1.5839, "step": 11910 }, { "epoch": 1.13, "learning_rate": 6.238184653575635e-05, "loss": 1.4361, "step": 11920 }, { "epoch": 1.14, "learning_rate": 6.235002068680182e-05, "loss": 1.1001, "step": 11930 }, { "epoch": 1.14, "learning_rate": 6.23181948378473e-05, "loss": 1.5917, "step": 11940 }, { "epoch": 1.14, "learning_rate": 6.228636898889279e-05, "loss": 1.1532, "step": 11950 }, { "epoch": 1.14, "learning_rate": 6.225454313993826e-05, "loss": 1.3314, "step": 11960 }, { "epoch": 1.14, "learning_rate": 6.222271729098375e-05, "loss": 1.1664, "step": 11970 }, { "epoch": 1.14, "learning_rate": 6.219089144202922e-05, "loss": 1.1507, "step": 11980 }, { "epoch": 1.14, "learning_rate": 6.21590655930747e-05, "loss": 1.3231, "step": 11990 }, { "epoch": 1.14, "learning_rate": 6.212723974412018e-05, "loss": 1.401, "step": 12000 }, { "epoch": 1.14, "learning_rate": 6.209541389516565e-05, "loss": 1.148, "step": 12010 }, { "epoch": 1.14, "learning_rate": 6.206358804621114e-05, "loss": 1.3955, "step": 12020 }, { "epoch": 1.14, "learning_rate": 6.203176219725662e-05, "loss": 1.714, "step": 12030 }, { "epoch": 1.15, "learning_rate": 6.199993634830209e-05, "loss": 1.3327, "step": 12040 }, { "epoch": 1.15, "learning_rate": 6.196811049934757e-05, "loss": 1.3942, "step": 12050 }, { "epoch": 1.15, "learning_rate": 6.193628465039304e-05, "loss": 1.2902, "step": 12060 }, { "epoch": 1.15, "learning_rate": 6.190445880143854e-05, "loss": 1.2024, "step": 12070 }, { "epoch": 1.15, "learning_rate": 6.187263295248401e-05, "loss": 1.6921, "step": 12080 }, { "epoch": 1.15, "learning_rate": 6.184080710352948e-05, "loss": 1.5338, "step": 12090 }, { "epoch": 1.15, "learning_rate": 6.180898125457496e-05, "loss": 1.3473, "step": 12100 }, { "epoch": 1.15, "learning_rate": 6.177715540562045e-05, "loss": 1.1758, "step": 12110 }, { "epoch": 1.15, "learning_rate": 6.174532955666593e-05, "loss": 0.9934, "step": 12120 }, { "epoch": 1.15, "learning_rate": 6.17135037077114e-05, "loss": 1.0722, "step": 12130 }, { "epoch": 1.16, "learning_rate": 6.168167785875689e-05, "loss": 1.5355, "step": 12140 }, { "epoch": 1.16, "learning_rate": 6.164985200980237e-05, "loss": 1.1969, "step": 12150 }, { "epoch": 1.16, "learning_rate": 6.161802616084784e-05, "loss": 1.2552, "step": 12160 }, { "epoch": 1.16, "learning_rate": 6.158620031189332e-05, "loss": 1.4347, "step": 12170 }, { "epoch": 1.16, "learning_rate": 6.155437446293881e-05, "loss": 1.2723, "step": 12180 }, { "epoch": 1.16, "learning_rate": 6.152254861398428e-05, "loss": 1.3817, "step": 12190 }, { "epoch": 1.16, "learning_rate": 6.149072276502976e-05, "loss": 1.4304, "step": 12200 }, { "epoch": 1.16, "learning_rate": 6.145889691607523e-05, "loss": 1.3085, "step": 12210 }, { "epoch": 1.16, "learning_rate": 6.142707106712073e-05, "loss": 1.4066, "step": 12220 }, { "epoch": 1.16, "learning_rate": 6.13952452181662e-05, "loss": 1.2759, "step": 12230 }, { "epoch": 1.16, "learning_rate": 6.136341936921167e-05, "loss": 1.3005, "step": 12240 }, { "epoch": 1.17, "learning_rate": 6.133159352025715e-05, "loss": 1.4923, "step": 12250 }, { "epoch": 1.17, "learning_rate": 6.129976767130264e-05, "loss": 1.2941, "step": 12260 }, { "epoch": 1.17, "learning_rate": 6.126794182234812e-05, "loss": 1.188, "step": 12270 }, { "epoch": 1.17, "learning_rate": 6.123611597339359e-05, "loss": 1.515, "step": 12280 }, { "epoch": 1.17, "learning_rate": 6.120429012443907e-05, "loss": 1.1593, "step": 12290 }, { "epoch": 1.17, "learning_rate": 6.117246427548456e-05, "loss": 1.1917, "step": 12300 }, { "epoch": 1.17, "learning_rate": 6.114063842653003e-05, "loss": 1.4694, "step": 12310 }, { "epoch": 1.17, "learning_rate": 6.110881257757551e-05, "loss": 1.3641, "step": 12320 }, { "epoch": 1.17, "learning_rate": 6.1076986728621e-05, "loss": 1.0053, "step": 12330 }, { "epoch": 1.17, "learning_rate": 6.104516087966646e-05, "loss": 1.3955, "step": 12340 }, { "epoch": 1.18, "learning_rate": 6.101333503071195e-05, "loss": 1.4972, "step": 12350 }, { "epoch": 1.18, "learning_rate": 6.0981509181757424e-05, "loss": 1.5457, "step": 12360 }, { "epoch": 1.18, "learning_rate": 6.094968333280291e-05, "loss": 1.295, "step": 12370 }, { "epoch": 1.18, "learning_rate": 6.0917857483848385e-05, "loss": 1.3848, "step": 12380 }, { "epoch": 1.18, "learning_rate": 6.088603163489386e-05, "loss": 1.3321, "step": 12390 }, { "epoch": 1.18, "learning_rate": 6.0854205785939345e-05, "loss": 1.5353, "step": 12400 }, { "epoch": 1.18, "learning_rate": 6.082237993698482e-05, "loss": 1.2588, "step": 12410 }, { "epoch": 1.18, "learning_rate": 6.0790554088030306e-05, "loss": 1.274, "step": 12420 }, { "epoch": 1.18, "learning_rate": 6.075872823907578e-05, "loss": 1.1487, "step": 12430 }, { "epoch": 1.18, "learning_rate": 6.072690239012125e-05, "loss": 1.2118, "step": 12440 }, { "epoch": 1.18, "learning_rate": 6.069507654116674e-05, "loss": 1.1819, "step": 12450 }, { "epoch": 1.19, "learning_rate": 6.066325069221221e-05, "loss": 1.4046, "step": 12460 }, { "epoch": 1.19, "learning_rate": 6.0631424843257704e-05, "loss": 1.4242, "step": 12470 }, { "epoch": 1.19, "learning_rate": 6.0599598994303174e-05, "loss": 1.4458, "step": 12480 }, { "epoch": 1.19, "learning_rate": 6.056777314534865e-05, "loss": 1.7952, "step": 12490 }, { "epoch": 1.19, "learning_rate": 6.0535947296394134e-05, "loss": 1.6093, "step": 12500 }, { "epoch": 1.19, "learning_rate": 6.050412144743961e-05, "loss": 1.3165, "step": 12510 }, { "epoch": 1.19, "learning_rate": 6.0472295598485095e-05, "loss": 1.1477, "step": 12520 }, { "epoch": 1.19, "learning_rate": 6.044046974953057e-05, "loss": 1.3023, "step": 12530 }, { "epoch": 1.19, "learning_rate": 6.0408643900576055e-05, "loss": 1.1291, "step": 12540 }, { "epoch": 1.19, "learning_rate": 6.037681805162153e-05, "loss": 1.1525, "step": 12550 }, { "epoch": 1.2, "learning_rate": 6.0344992202667e-05, "loss": 1.4392, "step": 12560 }, { "epoch": 1.2, "learning_rate": 6.031316635371249e-05, "loss": 1.2571, "step": 12570 }, { "epoch": 1.2, "learning_rate": 6.028134050475796e-05, "loss": 1.4243, "step": 12580 }, { "epoch": 1.2, "learning_rate": 6.024951465580345e-05, "loss": 1.3288, "step": 12590 }, { "epoch": 1.2, "learning_rate": 6.021768880684892e-05, "loss": 1.2786, "step": 12600 }, { "epoch": 1.2, "learning_rate": 6.01858629578944e-05, "loss": 1.4158, "step": 12610 }, { "epoch": 1.2, "learning_rate": 6.0154037108939884e-05, "loss": 1.6776, "step": 12620 }, { "epoch": 1.2, "learning_rate": 6.012221125998536e-05, "loss": 1.2973, "step": 12630 }, { "epoch": 1.2, "learning_rate": 6.0090385411030844e-05, "loss": 1.6589, "step": 12640 }, { "epoch": 1.2, "learning_rate": 6.005855956207632e-05, "loss": 1.4143, "step": 12650 }, { "epoch": 1.2, "learning_rate": 6.00267337131218e-05, "loss": 1.1845, "step": 12660 }, { "epoch": 1.21, "learning_rate": 5.999490786416728e-05, "loss": 1.379, "step": 12670 }, { "epoch": 1.21, "learning_rate": 5.996308201521276e-05, "loss": 1.1948, "step": 12680 }, { "epoch": 1.21, "learning_rate": 5.993125616625824e-05, "loss": 1.4856, "step": 12690 }, { "epoch": 1.21, "learning_rate": 5.989943031730372e-05, "loss": 1.0585, "step": 12700 }, { "epoch": 1.21, "learning_rate": 5.986760446834919e-05, "loss": 1.0362, "step": 12710 }, { "epoch": 1.21, "learning_rate": 5.983577861939468e-05, "loss": 1.5063, "step": 12720 }, { "epoch": 1.21, "learning_rate": 5.980395277044015e-05, "loss": 1.3054, "step": 12730 }, { "epoch": 1.21, "learning_rate": 5.977212692148564e-05, "loss": 1.2419, "step": 12740 }, { "epoch": 1.21, "learning_rate": 5.974030107253111e-05, "loss": 1.3128, "step": 12750 }, { "epoch": 1.21, "learning_rate": 5.970847522357659e-05, "loss": 1.3978, "step": 12760 }, { "epoch": 1.22, "learning_rate": 5.967664937462207e-05, "loss": 1.3241, "step": 12770 }, { "epoch": 1.22, "learning_rate": 5.964482352566755e-05, "loss": 1.3458, "step": 12780 }, { "epoch": 1.22, "learning_rate": 5.961299767671303e-05, "loss": 1.1913, "step": 12790 }, { "epoch": 1.22, "learning_rate": 5.958117182775851e-05, "loss": 1.207, "step": 12800 }, { "epoch": 1.22, "learning_rate": 5.9549345978803985e-05, "loss": 1.8335, "step": 12810 }, { "epoch": 1.22, "learning_rate": 5.951752012984947e-05, "loss": 1.2501, "step": 12820 }, { "epoch": 1.22, "learning_rate": 5.9485694280894945e-05, "loss": 1.3696, "step": 12830 }, { "epoch": 1.22, "learning_rate": 5.945386843194043e-05, "loss": 1.4096, "step": 12840 }, { "epoch": 1.22, "learning_rate": 5.9422042582985906e-05, "loss": 1.2826, "step": 12850 }, { "epoch": 1.22, "learning_rate": 5.9390216734031376e-05, "loss": 1.781, "step": 12860 }, { "epoch": 1.22, "learning_rate": 5.9358390885076866e-05, "loss": 1.2511, "step": 12870 }, { "epoch": 1.23, "learning_rate": 5.9326565036122336e-05, "loss": 1.2594, "step": 12880 }, { "epoch": 1.23, "learning_rate": 5.929473918716783e-05, "loss": 1.2174, "step": 12890 }, { "epoch": 1.23, "learning_rate": 5.92629133382133e-05, "loss": 1.3863, "step": 12900 }, { "epoch": 1.23, "learning_rate": 5.9231087489258774e-05, "loss": 1.5468, "step": 12910 }, { "epoch": 1.23, "learning_rate": 5.919926164030426e-05, "loss": 1.3738, "step": 12920 }, { "epoch": 1.23, "learning_rate": 5.9167435791349734e-05, "loss": 1.2275, "step": 12930 }, { "epoch": 1.23, "learning_rate": 5.913560994239522e-05, "loss": 1.2463, "step": 12940 }, { "epoch": 1.23, "learning_rate": 5.9103784093440695e-05, "loss": 1.5187, "step": 12950 }, { "epoch": 1.23, "learning_rate": 5.907195824448617e-05, "loss": 1.2007, "step": 12960 }, { "epoch": 1.23, "learning_rate": 5.9040132395531655e-05, "loss": 1.5909, "step": 12970 }, { "epoch": 1.24, "learning_rate": 5.900830654657713e-05, "loss": 1.1711, "step": 12980 }, { "epoch": 1.24, "learning_rate": 5.8976480697622616e-05, "loss": 1.498, "step": 12990 }, { "epoch": 1.24, "learning_rate": 5.894465484866809e-05, "loss": 1.2641, "step": 13000 }, { "epoch": 1.24, "learning_rate": 5.891282899971356e-05, "loss": 1.4552, "step": 13010 }, { "epoch": 1.24, "learning_rate": 5.888100315075905e-05, "loss": 1.4062, "step": 13020 }, { "epoch": 1.24, "learning_rate": 5.884917730180452e-05, "loss": 1.6072, "step": 13030 }, { "epoch": 1.24, "learning_rate": 5.8817351452850014e-05, "loss": 1.5844, "step": 13040 }, { "epoch": 1.24, "learning_rate": 5.8785525603895484e-05, "loss": 1.3093, "step": 13050 }, { "epoch": 1.24, "learning_rate": 5.875369975494096e-05, "loss": 1.3364, "step": 13060 }, { "epoch": 1.24, "learning_rate": 5.8721873905986444e-05, "loss": 1.2106, "step": 13070 }, { "epoch": 1.24, "learning_rate": 5.869004805703192e-05, "loss": 1.4542, "step": 13080 }, { "epoch": 1.25, "learning_rate": 5.8658222208077405e-05, "loss": 1.3747, "step": 13090 }, { "epoch": 1.25, "learning_rate": 5.862639635912288e-05, "loss": 1.4218, "step": 13100 }, { "epoch": 1.25, "learning_rate": 5.859457051016836e-05, "loss": 1.3271, "step": 13110 }, { "epoch": 1.25, "learning_rate": 5.856274466121384e-05, "loss": 1.3049, "step": 13120 }, { "epoch": 1.25, "learning_rate": 5.853091881225932e-05, "loss": 1.1618, "step": 13130 }, { "epoch": 1.25, "learning_rate": 5.84990929633048e-05, "loss": 1.5606, "step": 13140 }, { "epoch": 1.25, "learning_rate": 5.846726711435028e-05, "loss": 1.5192, "step": 13150 }, { "epoch": 1.25, "learning_rate": 5.843544126539575e-05, "loss": 1.2331, "step": 13160 }, { "epoch": 1.25, "learning_rate": 5.840361541644124e-05, "loss": 1.3765, "step": 13170 }, { "epoch": 1.25, "learning_rate": 5.837178956748671e-05, "loss": 1.3304, "step": 13180 }, { "epoch": 1.26, "learning_rate": 5.83399637185322e-05, "loss": 1.0399, "step": 13190 }, { "epoch": 1.26, "learning_rate": 5.830813786957767e-05, "loss": 1.4977, "step": 13200 }, { "epoch": 1.26, "learning_rate": 5.827631202062316e-05, "loss": 1.2552, "step": 13210 }, { "epoch": 1.26, "learning_rate": 5.824448617166863e-05, "loss": 1.1272, "step": 13220 }, { "epoch": 1.26, "learning_rate": 5.821266032271411e-05, "loss": 1.686, "step": 13230 }, { "epoch": 1.26, "learning_rate": 5.818083447375959e-05, "loss": 1.3021, "step": 13240 }, { "epoch": 1.26, "learning_rate": 5.814900862480507e-05, "loss": 1.3472, "step": 13250 }, { "epoch": 1.26, "learning_rate": 5.811718277585055e-05, "loss": 1.1018, "step": 13260 }, { "epoch": 1.26, "learning_rate": 5.808535692689603e-05, "loss": 1.2079, "step": 13270 }, { "epoch": 1.26, "learning_rate": 5.8053531077941506e-05, "loss": 1.205, "step": 13280 }, { "epoch": 1.26, "learning_rate": 5.802170522898699e-05, "loss": 1.4772, "step": 13290 }, { "epoch": 1.27, "learning_rate": 5.7989879380032466e-05, "loss": 1.1362, "step": 13300 }, { "epoch": 1.27, "learning_rate": 5.795805353107795e-05, "loss": 1.5549, "step": 13310 }, { "epoch": 1.27, "learning_rate": 5.792622768212343e-05, "loss": 1.2256, "step": 13320 }, { "epoch": 1.27, "learning_rate": 5.78944018331689e-05, "loss": 1.1792, "step": 13330 }, { "epoch": 1.27, "learning_rate": 5.786257598421439e-05, "loss": 1.444, "step": 13340 }, { "epoch": 1.27, "learning_rate": 5.783075013525986e-05, "loss": 1.3135, "step": 13350 }, { "epoch": 1.27, "learning_rate": 5.779892428630535e-05, "loss": 1.2954, "step": 13360 }, { "epoch": 1.27, "learning_rate": 5.776709843735082e-05, "loss": 1.2679, "step": 13370 }, { "epoch": 1.27, "learning_rate": 5.7735272588396295e-05, "loss": 1.3775, "step": 13380 }, { "epoch": 1.27, "learning_rate": 5.770344673944178e-05, "loss": 1.0079, "step": 13390 }, { "epoch": 1.28, "learning_rate": 5.7671620890487255e-05, "loss": 1.3256, "step": 13400 }, { "epoch": 1.28, "learning_rate": 5.763979504153274e-05, "loss": 1.2643, "step": 13410 }, { "epoch": 1.28, "learning_rate": 5.7607969192578216e-05, "loss": 1.0838, "step": 13420 }, { "epoch": 1.28, "learning_rate": 5.757614334362369e-05, "loss": 1.1447, "step": 13430 }, { "epoch": 1.28, "learning_rate": 5.7544317494669176e-05, "loss": 1.0348, "step": 13440 }, { "epoch": 1.28, "learning_rate": 5.751249164571465e-05, "loss": 1.4069, "step": 13450 }, { "epoch": 1.28, "learning_rate": 5.748066579676014e-05, "loss": 1.5608, "step": 13460 }, { "epoch": 1.28, "learning_rate": 5.7448839947805614e-05, "loss": 1.0844, "step": 13470 }, { "epoch": 1.28, "learning_rate": 5.7417014098851084e-05, "loss": 1.4903, "step": 13480 }, { "epoch": 1.28, "learning_rate": 5.7385188249896574e-05, "loss": 1.2309, "step": 13490 }, { "epoch": 1.28, "learning_rate": 5.7353362400942044e-05, "loss": 1.4884, "step": 13500 }, { "epoch": 1.29, "learning_rate": 5.7321536551987535e-05, "loss": 1.4128, "step": 13510 }, { "epoch": 1.29, "learning_rate": 5.7289710703033005e-05, "loss": 1.2433, "step": 13520 }, { "epoch": 1.29, "learning_rate": 5.725788485407848e-05, "loss": 1.4097, "step": 13530 }, { "epoch": 1.29, "learning_rate": 5.7226059005123965e-05, "loss": 1.2903, "step": 13540 }, { "epoch": 1.29, "learning_rate": 5.719423315616944e-05, "loss": 1.2305, "step": 13550 }, { "epoch": 1.29, "learning_rate": 5.7162407307214926e-05, "loss": 1.2742, "step": 13560 }, { "epoch": 1.29, "learning_rate": 5.71305814582604e-05, "loss": 1.1847, "step": 13570 }, { "epoch": 1.29, "learning_rate": 5.709875560930588e-05, "loss": 1.1794, "step": 13580 }, { "epoch": 1.29, "learning_rate": 5.706692976035136e-05, "loss": 1.4876, "step": 13590 }, { "epoch": 1.29, "learning_rate": 5.703510391139684e-05, "loss": 1.3414, "step": 13600 }, { "epoch": 1.3, "learning_rate": 5.7003278062442324e-05, "loss": 1.6806, "step": 13610 }, { "epoch": 1.3, "learning_rate": 5.6971452213487794e-05, "loss": 1.167, "step": 13620 }, { "epoch": 1.3, "learning_rate": 5.693962636453327e-05, "loss": 1.6015, "step": 13630 }, { "epoch": 1.3, "learning_rate": 5.6907800515578754e-05, "loss": 1.3552, "step": 13640 }, { "epoch": 1.3, "learning_rate": 5.687597466662423e-05, "loss": 1.4357, "step": 13650 }, { "epoch": 1.3, "learning_rate": 5.6844148817669715e-05, "loss": 1.2554, "step": 13660 }, { "epoch": 1.3, "learning_rate": 5.681232296871519e-05, "loss": 1.3044, "step": 13670 }, { "epoch": 1.3, "learning_rate": 5.678049711976067e-05, "loss": 1.0156, "step": 13680 }, { "epoch": 1.3, "learning_rate": 5.674867127080615e-05, "loss": 1.3119, "step": 13690 }, { "epoch": 1.3, "learning_rate": 5.671684542185163e-05, "loss": 1.4552, "step": 13700 }, { "epoch": 1.3, "learning_rate": 5.668501957289711e-05, "loss": 1.3558, "step": 13710 }, { "epoch": 1.31, "learning_rate": 5.665319372394259e-05, "loss": 1.3309, "step": 13720 }, { "epoch": 1.31, "learning_rate": 5.662136787498806e-05, "loss": 1.4877, "step": 13730 }, { "epoch": 1.31, "learning_rate": 5.658954202603355e-05, "loss": 1.2664, "step": 13740 }, { "epoch": 1.31, "learning_rate": 5.655771617707902e-05, "loss": 1.2099, "step": 13750 }, { "epoch": 1.31, "learning_rate": 5.652589032812451e-05, "loss": 1.2328, "step": 13760 }, { "epoch": 1.31, "learning_rate": 5.649406447916998e-05, "loss": 1.1204, "step": 13770 }, { "epoch": 1.31, "learning_rate": 5.646223863021546e-05, "loss": 1.4109, "step": 13780 }, { "epoch": 1.31, "learning_rate": 5.643041278126094e-05, "loss": 1.1479, "step": 13790 }, { "epoch": 1.31, "learning_rate": 5.639858693230642e-05, "loss": 1.2953, "step": 13800 }, { "epoch": 1.31, "learning_rate": 5.63667610833519e-05, "loss": 1.4705, "step": 13810 }, { "epoch": 1.32, "learning_rate": 5.633493523439738e-05, "loss": 1.5887, "step": 13820 }, { "epoch": 1.32, "learning_rate": 5.630310938544286e-05, "loss": 1.2498, "step": 13830 }, { "epoch": 1.32, "learning_rate": 5.627128353648834e-05, "loss": 1.3203, "step": 13840 }, { "epoch": 1.32, "learning_rate": 5.6239457687533816e-05, "loss": 1.6888, "step": 13850 }, { "epoch": 1.32, "learning_rate": 5.62076318385793e-05, "loss": 1.2531, "step": 13860 }, { "epoch": 1.32, "learning_rate": 5.6175805989624776e-05, "loss": 1.0861, "step": 13870 }, { "epoch": 1.32, "learning_rate": 5.614398014067026e-05, "loss": 1.4684, "step": 13880 }, { "epoch": 1.32, "learning_rate": 5.611215429171574e-05, "loss": 0.8508, "step": 13890 }, { "epoch": 1.32, "learning_rate": 5.608032844276121e-05, "loss": 1.4018, "step": 13900 }, { "epoch": 1.32, "learning_rate": 5.60485025938067e-05, "loss": 1.1384, "step": 13910 }, { "epoch": 1.32, "learning_rate": 5.601667674485217e-05, "loss": 1.3172, "step": 13920 }, { "epoch": 1.33, "learning_rate": 5.598485089589766e-05, "loss": 1.3609, "step": 13930 }, { "epoch": 1.33, "learning_rate": 5.595302504694313e-05, "loss": 1.3021, "step": 13940 }, { "epoch": 1.33, "learning_rate": 5.5921199197988605e-05, "loss": 1.4604, "step": 13950 }, { "epoch": 1.33, "learning_rate": 5.588937334903409e-05, "loss": 1.4656, "step": 13960 }, { "epoch": 1.33, "learning_rate": 5.5857547500079565e-05, "loss": 1.0608, "step": 13970 }, { "epoch": 1.33, "learning_rate": 5.582572165112505e-05, "loss": 1.4819, "step": 13980 }, { "epoch": 1.33, "learning_rate": 5.5793895802170526e-05, "loss": 1.3399, "step": 13990 }, { "epoch": 1.33, "learning_rate": 5.5762069953216e-05, "loss": 1.1616, "step": 14000 }, { "epoch": 1.33, "learning_rate": 5.5730244104261486e-05, "loss": 1.2001, "step": 14010 }, { "epoch": 1.33, "learning_rate": 5.569841825530696e-05, "loss": 1.5204, "step": 14020 }, { "epoch": 1.34, "learning_rate": 5.566659240635245e-05, "loss": 1.2357, "step": 14030 }, { "epoch": 1.34, "learning_rate": 5.5634766557397924e-05, "loss": 1.2601, "step": 14040 }, { "epoch": 1.34, "learning_rate": 5.5602940708443394e-05, "loss": 1.4038, "step": 14050 }, { "epoch": 1.34, "learning_rate": 5.5571114859488884e-05, "loss": 1.3067, "step": 14060 }, { "epoch": 1.34, "learning_rate": 5.5539289010534354e-05, "loss": 1.269, "step": 14070 }, { "epoch": 1.34, "learning_rate": 5.5507463161579845e-05, "loss": 1.3595, "step": 14080 }, { "epoch": 1.34, "learning_rate": 5.5475637312625315e-05, "loss": 1.2674, "step": 14090 }, { "epoch": 1.34, "learning_rate": 5.544381146367079e-05, "loss": 1.2473, "step": 14100 }, { "epoch": 1.34, "learning_rate": 5.5411985614716275e-05, "loss": 1.5753, "step": 14110 }, { "epoch": 1.34, "learning_rate": 5.538015976576175e-05, "loss": 1.4645, "step": 14120 }, { "epoch": 1.34, "learning_rate": 5.5348333916807236e-05, "loss": 1.4572, "step": 14130 }, { "epoch": 1.35, "learning_rate": 5.531650806785271e-05, "loss": 1.2548, "step": 14140 }, { "epoch": 1.35, "learning_rate": 5.528468221889819e-05, "loss": 1.3171, "step": 14150 }, { "epoch": 1.35, "learning_rate": 5.525285636994367e-05, "loss": 1.3876, "step": 14160 }, { "epoch": 1.35, "learning_rate": 5.522103052098915e-05, "loss": 1.4725, "step": 14170 }, { "epoch": 1.35, "learning_rate": 5.5189204672034634e-05, "loss": 1.2703, "step": 14180 }, { "epoch": 1.35, "learning_rate": 5.515737882308011e-05, "loss": 1.2428, "step": 14190 }, { "epoch": 1.35, "learning_rate": 5.512555297412558e-05, "loss": 1.5501, "step": 14200 }, { "epoch": 1.35, "learning_rate": 5.509372712517107e-05, "loss": 1.4164, "step": 14210 }, { "epoch": 1.35, "learning_rate": 5.506190127621654e-05, "loss": 1.4017, "step": 14220 }, { "epoch": 1.35, "learning_rate": 5.503007542726203e-05, "loss": 1.2215, "step": 14230 }, { "epoch": 1.36, "learning_rate": 5.49982495783075e-05, "loss": 1.1972, "step": 14240 }, { "epoch": 1.36, "learning_rate": 5.496642372935298e-05, "loss": 1.4031, "step": 14250 }, { "epoch": 1.36, "learning_rate": 5.493459788039846e-05, "loss": 1.1811, "step": 14260 }, { "epoch": 1.36, "learning_rate": 5.490277203144394e-05, "loss": 1.2467, "step": 14270 }, { "epoch": 1.36, "learning_rate": 5.487094618248942e-05, "loss": 1.4318, "step": 14280 }, { "epoch": 1.36, "learning_rate": 5.48391203335349e-05, "loss": 1.3557, "step": 14290 }, { "epoch": 1.36, "learning_rate": 5.4807294484580376e-05, "loss": 1.3901, "step": 14300 }, { "epoch": 1.36, "learning_rate": 5.477546863562586e-05, "loss": 1.5302, "step": 14310 }, { "epoch": 1.36, "learning_rate": 5.474364278667134e-05, "loss": 1.1885, "step": 14320 }, { "epoch": 1.36, "learning_rate": 5.471181693771682e-05, "loss": 1.1578, "step": 14330 }, { "epoch": 1.36, "learning_rate": 5.46799910887623e-05, "loss": 1.3881, "step": 14340 }, { "epoch": 1.37, "learning_rate": 5.464816523980777e-05, "loss": 1.1401, "step": 14350 }, { "epoch": 1.37, "learning_rate": 5.461633939085326e-05, "loss": 1.0311, "step": 14360 }, { "epoch": 1.37, "learning_rate": 5.458451354189873e-05, "loss": 1.1228, "step": 14370 }, { "epoch": 1.37, "learning_rate": 5.455268769294422e-05, "loss": 1.2884, "step": 14380 }, { "epoch": 1.37, "learning_rate": 5.452086184398969e-05, "loss": 1.3827, "step": 14390 }, { "epoch": 1.37, "learning_rate": 5.4489035995035165e-05, "loss": 1.239, "step": 14400 }, { "epoch": 1.37, "learning_rate": 5.445721014608065e-05, "loss": 1.4542, "step": 14410 }, { "epoch": 1.37, "learning_rate": 5.4425384297126126e-05, "loss": 1.0058, "step": 14420 }, { "epoch": 1.37, "learning_rate": 5.439355844817161e-05, "loss": 1.3152, "step": 14430 }, { "epoch": 1.37, "learning_rate": 5.4361732599217086e-05, "loss": 1.4563, "step": 14440 }, { "epoch": 1.38, "learning_rate": 5.432990675026257e-05, "loss": 1.2788, "step": 14450 }, { "epoch": 1.38, "learning_rate": 5.429808090130805e-05, "loss": 1.5673, "step": 14460 }, { "epoch": 1.38, "learning_rate": 5.4266255052353524e-05, "loss": 1.5005, "step": 14470 }, { "epoch": 1.38, "learning_rate": 5.423442920339901e-05, "loss": 1.1309, "step": 14480 }, { "epoch": 1.38, "learning_rate": 5.4202603354444484e-05, "loss": 1.4121, "step": 14490 }, { "epoch": 1.38, "learning_rate": 5.417077750548997e-05, "loss": 1.3966, "step": 14500 }, { "epoch": 1.38, "learning_rate": 5.4138951656535445e-05, "loss": 1.1501, "step": 14510 }, { "epoch": 1.38, "learning_rate": 5.4107125807580915e-05, "loss": 0.9098, "step": 14520 }, { "epoch": 1.38, "learning_rate": 5.4075299958626405e-05, "loss": 1.123, "step": 14530 }, { "epoch": 1.38, "learning_rate": 5.4043474109671875e-05, "loss": 1.2539, "step": 14540 }, { "epoch": 1.38, "learning_rate": 5.4011648260717366e-05, "loss": 1.4045, "step": 14550 }, { "epoch": 1.39, "learning_rate": 5.3979822411762836e-05, "loss": 1.6323, "step": 14560 }, { "epoch": 1.39, "learning_rate": 5.394799656280831e-05, "loss": 1.3246, "step": 14570 }, { "epoch": 1.39, "learning_rate": 5.3916170713853796e-05, "loss": 1.1879, "step": 14580 }, { "epoch": 1.39, "learning_rate": 5.388434486489927e-05, "loss": 1.3357, "step": 14590 }, { "epoch": 1.39, "learning_rate": 5.385251901594476e-05, "loss": 1.2795, "step": 14600 }, { "epoch": 1.39, "learning_rate": 5.3820693166990234e-05, "loss": 1.521, "step": 14610 }, { "epoch": 1.39, "learning_rate": 5.378886731803571e-05, "loss": 1.2611, "step": 14620 }, { "epoch": 1.39, "learning_rate": 5.3757041469081194e-05, "loss": 1.4852, "step": 14630 }, { "epoch": 1.39, "learning_rate": 5.372521562012667e-05, "loss": 0.9539, "step": 14640 }, { "epoch": 1.39, "learning_rate": 5.3693389771172155e-05, "loss": 1.1787, "step": 14650 }, { "epoch": 1.4, "learning_rate": 5.366156392221763e-05, "loss": 1.3916, "step": 14660 }, { "epoch": 1.4, "learning_rate": 5.36297380732631e-05, "loss": 1.0306, "step": 14670 }, { "epoch": 1.4, "learning_rate": 5.3597912224308585e-05, "loss": 1.3695, "step": 14680 }, { "epoch": 1.4, "learning_rate": 5.356608637535406e-05, "loss": 1.2022, "step": 14690 }, { "epoch": 1.4, "learning_rate": 5.3534260526399546e-05, "loss": 1.1744, "step": 14700 }, { "epoch": 1.4, "learning_rate": 5.350243467744502e-05, "loss": 1.4006, "step": 14710 }, { "epoch": 1.4, "learning_rate": 5.34706088284905e-05, "loss": 1.3549, "step": 14720 }, { "epoch": 1.4, "learning_rate": 5.343878297953598e-05, "loss": 1.2094, "step": 14730 }, { "epoch": 1.4, "learning_rate": 5.340695713058146e-05, "loss": 1.2815, "step": 14740 }, { "epoch": 1.4, "learning_rate": 5.3375131281626944e-05, "loss": 1.1214, "step": 14750 }, { "epoch": 1.4, "learning_rate": 5.334330543267242e-05, "loss": 1.5456, "step": 14760 }, { "epoch": 1.41, "learning_rate": 5.331147958371789e-05, "loss": 1.3289, "step": 14770 }, { "epoch": 1.41, "learning_rate": 5.327965373476338e-05, "loss": 1.5725, "step": 14780 }, { "epoch": 1.41, "learning_rate": 5.324782788580885e-05, "loss": 1.3129, "step": 14790 }, { "epoch": 1.41, "learning_rate": 5.321600203685434e-05, "loss": 1.1212, "step": 14800 }, { "epoch": 1.41, "learning_rate": 5.318417618789981e-05, "loss": 1.5021, "step": 14810 }, { "epoch": 1.41, "learning_rate": 5.315235033894529e-05, "loss": 1.2091, "step": 14820 }, { "epoch": 1.41, "learning_rate": 5.312052448999077e-05, "loss": 1.3694, "step": 14830 }, { "epoch": 1.41, "learning_rate": 5.308869864103625e-05, "loss": 1.2515, "step": 14840 }, { "epoch": 1.41, "learning_rate": 5.305687279208173e-05, "loss": 1.2529, "step": 14850 }, { "epoch": 1.41, "learning_rate": 5.302504694312721e-05, "loss": 1.2047, "step": 14860 }, { "epoch": 1.42, "learning_rate": 5.2993221094172686e-05, "loss": 1.3786, "step": 14870 }, { "epoch": 1.42, "learning_rate": 5.296139524521817e-05, "loss": 1.3992, "step": 14880 }, { "epoch": 1.42, "learning_rate": 5.292956939626365e-05, "loss": 1.3026, "step": 14890 }, { "epoch": 1.42, "learning_rate": 5.289774354730913e-05, "loss": 1.2121, "step": 14900 }, { "epoch": 1.42, "learning_rate": 5.286591769835461e-05, "loss": 1.4507, "step": 14910 }, { "epoch": 1.42, "learning_rate": 5.283409184940008e-05, "loss": 1.0681, "step": 14920 }, { "epoch": 1.42, "learning_rate": 5.280226600044557e-05, "loss": 1.4707, "step": 14930 }, { "epoch": 1.42, "learning_rate": 5.277044015149104e-05, "loss": 1.3271, "step": 14940 }, { "epoch": 1.42, "learning_rate": 5.273861430253653e-05, "loss": 1.1235, "step": 14950 }, { "epoch": 1.42, "learning_rate": 5.2706788453582e-05, "loss": 1.0133, "step": 14960 }, { "epoch": 1.42, "learning_rate": 5.2674962604627475e-05, "loss": 1.2553, "step": 14970 }, { "epoch": 1.43, "learning_rate": 5.264313675567296e-05, "loss": 1.2721, "step": 14980 }, { "epoch": 1.43, "learning_rate": 5.2611310906718436e-05, "loss": 1.2123, "step": 14990 }, { "epoch": 1.43, "learning_rate": 5.257948505776392e-05, "loss": 1.4009, "step": 15000 }, { "epoch": 1.43, "learning_rate": 5.2547659208809396e-05, "loss": 1.2326, "step": 15010 }, { "epoch": 1.43, "learning_rate": 5.251583335985487e-05, "loss": 1.3902, "step": 15020 }, { "epoch": 1.43, "learning_rate": 5.248400751090036e-05, "loss": 1.4369, "step": 15030 }, { "epoch": 1.43, "learning_rate": 5.2452181661945834e-05, "loss": 1.2699, "step": 15040 }, { "epoch": 1.43, "learning_rate": 5.242035581299132e-05, "loss": 1.2985, "step": 15050 }, { "epoch": 1.43, "learning_rate": 5.2388529964036794e-05, "loss": 1.5252, "step": 15060 }, { "epoch": 1.43, "learning_rate": 5.2356704115082264e-05, "loss": 1.5001, "step": 15070 }, { "epoch": 1.44, "learning_rate": 5.2324878266127755e-05, "loss": 1.4595, "step": 15080 }, { "epoch": 1.44, "learning_rate": 5.2293052417173225e-05, "loss": 1.1815, "step": 15090 }, { "epoch": 1.44, "learning_rate": 5.2261226568218715e-05, "loss": 1.4579, "step": 15100 }, { "epoch": 1.44, "learning_rate": 5.2229400719264185e-05, "loss": 1.1008, "step": 15110 }, { "epoch": 1.44, "learning_rate": 5.2197574870309676e-05, "loss": 1.2104, "step": 15120 }, { "epoch": 1.44, "learning_rate": 5.2165749021355146e-05, "loss": 1.0013, "step": 15130 }, { "epoch": 1.44, "learning_rate": 5.213392317240062e-05, "loss": 1.2941, "step": 15140 }, { "epoch": 1.44, "learning_rate": 5.2102097323446106e-05, "loss": 1.2361, "step": 15150 }, { "epoch": 1.44, "learning_rate": 5.207027147449158e-05, "loss": 1.1462, "step": 15160 }, { "epoch": 1.44, "learning_rate": 5.203844562553707e-05, "loss": 1.2805, "step": 15170 }, { "epoch": 1.44, "learning_rate": 5.2006619776582544e-05, "loss": 1.157, "step": 15180 }, { "epoch": 1.45, "learning_rate": 5.197479392762802e-05, "loss": 1.2395, "step": 15190 }, { "epoch": 1.45, "learning_rate": 5.1942968078673504e-05, "loss": 1.2716, "step": 15200 }, { "epoch": 1.45, "learning_rate": 5.191114222971898e-05, "loss": 1.1911, "step": 15210 }, { "epoch": 1.45, "learning_rate": 5.1879316380764465e-05, "loss": 1.2859, "step": 15220 }, { "epoch": 1.45, "learning_rate": 5.184749053180994e-05, "loss": 0.9757, "step": 15230 }, { "epoch": 1.45, "learning_rate": 5.181566468285541e-05, "loss": 1.1481, "step": 15240 }, { "epoch": 1.45, "learning_rate": 5.17838388339009e-05, "loss": 1.3675, "step": 15250 }, { "epoch": 1.45, "learning_rate": 5.175201298494637e-05, "loss": 1.1877, "step": 15260 }, { "epoch": 1.45, "learning_rate": 5.172018713599186e-05, "loss": 1.2303, "step": 15270 }, { "epoch": 1.45, "learning_rate": 5.168836128703733e-05, "loss": 1.6646, "step": 15280 }, { "epoch": 1.46, "learning_rate": 5.165653543808281e-05, "loss": 1.4336, "step": 15290 }, { "epoch": 1.46, "learning_rate": 5.162470958912829e-05, "loss": 1.612, "step": 15300 }, { "epoch": 1.46, "learning_rate": 5.159288374017377e-05, "loss": 1.2497, "step": 15310 }, { "epoch": 1.46, "learning_rate": 5.1561057891219253e-05, "loss": 1.2639, "step": 15320 }, { "epoch": 1.46, "learning_rate": 5.152923204226473e-05, "loss": 1.3976, "step": 15330 }, { "epoch": 1.46, "learning_rate": 5.149740619331021e-05, "loss": 1.5085, "step": 15340 }, { "epoch": 1.46, "learning_rate": 5.146558034435569e-05, "loss": 1.5096, "step": 15350 }, { "epoch": 1.46, "learning_rate": 5.143375449540117e-05, "loss": 1.6825, "step": 15360 }, { "epoch": 1.46, "learning_rate": 5.140192864644665e-05, "loss": 1.265, "step": 15370 }, { "epoch": 1.46, "learning_rate": 5.137010279749213e-05, "loss": 1.4147, "step": 15380 }, { "epoch": 1.46, "learning_rate": 5.13382769485376e-05, "loss": 1.3444, "step": 15390 }, { "epoch": 1.47, "learning_rate": 5.130645109958309e-05, "loss": 1.1047, "step": 15400 }, { "epoch": 1.47, "learning_rate": 5.127462525062856e-05, "loss": 1.6598, "step": 15410 }, { "epoch": 1.47, "learning_rate": 5.124279940167405e-05, "loss": 1.309, "step": 15420 }, { "epoch": 1.47, "learning_rate": 5.121097355271952e-05, "loss": 1.148, "step": 15430 }, { "epoch": 1.47, "learning_rate": 5.1179147703764996e-05, "loss": 1.2568, "step": 15440 }, { "epoch": 1.47, "learning_rate": 5.114732185481048e-05, "loss": 1.3135, "step": 15450 }, { "epoch": 1.47, "learning_rate": 5.111549600585596e-05, "loss": 1.2472, "step": 15460 }, { "epoch": 1.47, "learning_rate": 5.108367015690144e-05, "loss": 1.3674, "step": 15470 }, { "epoch": 1.47, "learning_rate": 5.105184430794692e-05, "loss": 1.3011, "step": 15480 }, { "epoch": 1.47, "learning_rate": 5.1020018458992394e-05, "loss": 1.1521, "step": 15490 }, { "epoch": 1.48, "learning_rate": 5.098819261003788e-05, "loss": 1.1709, "step": 15500 }, { "epoch": 1.48, "learning_rate": 5.0956366761083355e-05, "loss": 1.3494, "step": 15510 }, { "epoch": 1.48, "learning_rate": 5.092454091212884e-05, "loss": 1.2846, "step": 15520 }, { "epoch": 1.48, "learning_rate": 5.0892715063174315e-05, "loss": 1.3508, "step": 15530 }, { "epoch": 1.48, "learning_rate": 5.0860889214219785e-05, "loss": 1.3032, "step": 15540 }, { "epoch": 1.48, "learning_rate": 5.0829063365265276e-05, "loss": 1.1961, "step": 15550 }, { "epoch": 1.48, "learning_rate": 5.0797237516310746e-05, "loss": 1.4699, "step": 15560 }, { "epoch": 1.48, "learning_rate": 5.0765411667356236e-05, "loss": 1.2985, "step": 15570 }, { "epoch": 1.48, "learning_rate": 5.0733585818401706e-05, "loss": 1.2926, "step": 15580 }, { "epoch": 1.48, "learning_rate": 5.070175996944718e-05, "loss": 1.3092, "step": 15590 }, { "epoch": 1.48, "learning_rate": 5.066993412049267e-05, "loss": 1.4662, "step": 15600 }, { "epoch": 1.49, "learning_rate": 5.0638108271538144e-05, "loss": 1.1928, "step": 15610 }, { "epoch": 1.49, "learning_rate": 5.060628242258363e-05, "loss": 1.3266, "step": 15620 }, { "epoch": 1.49, "learning_rate": 5.0574456573629104e-05, "loss": 1.2545, "step": 15630 }, { "epoch": 1.49, "learning_rate": 5.054263072467458e-05, "loss": 1.3546, "step": 15640 }, { "epoch": 1.49, "learning_rate": 5.0510804875720065e-05, "loss": 1.4485, "step": 15650 }, { "epoch": 1.49, "learning_rate": 5.047897902676554e-05, "loss": 1.0811, "step": 15660 }, { "epoch": 1.49, "learning_rate": 5.0447153177811025e-05, "loss": 1.256, "step": 15670 }, { "epoch": 1.49, "learning_rate": 5.04153273288565e-05, "loss": 1.3061, "step": 15680 }, { "epoch": 1.49, "learning_rate": 5.038350147990197e-05, "loss": 1.1861, "step": 15690 }, { "epoch": 1.49, "learning_rate": 5.035167563094746e-05, "loss": 1.3347, "step": 15700 }, { "epoch": 1.5, "learning_rate": 5.031984978199293e-05, "loss": 1.105, "step": 15710 }, { "epoch": 1.5, "learning_rate": 5.028802393303842e-05, "loss": 1.2095, "step": 15720 }, { "epoch": 1.5, "learning_rate": 5.025619808408389e-05, "loss": 1.0656, "step": 15730 }, { "epoch": 1.5, "learning_rate": 5.022437223512938e-05, "loss": 1.291, "step": 15740 }, { "epoch": 1.5, "learning_rate": 5.0192546386174853e-05, "loss": 1.2578, "step": 15750 }, { "epoch": 1.5, "learning_rate": 5.016072053722033e-05, "loss": 1.2552, "step": 15760 }, { "epoch": 1.5, "learning_rate": 5.0128894688265814e-05, "loss": 1.0318, "step": 15770 }, { "epoch": 1.5, "learning_rate": 5.009706883931129e-05, "loss": 1.2062, "step": 15780 }, { "epoch": 1.5, "learning_rate": 5.0065242990356774e-05, "loss": 1.1256, "step": 15790 }, { "epoch": 1.5, "learning_rate": 5.003341714140225e-05, "loss": 1.0347, "step": 15800 }, { "epoch": 1.5, "learning_rate": 5.000159129244772e-05, "loss": 1.1673, "step": 15810 }, { "epoch": 1.51, "learning_rate": 4.9969765443493205e-05, "loss": 1.4957, "step": 15820 }, { "epoch": 1.51, "learning_rate": 4.993793959453868e-05, "loss": 1.3662, "step": 15830 }, { "epoch": 1.51, "learning_rate": 4.9906113745584166e-05, "loss": 1.2346, "step": 15840 }, { "epoch": 1.51, "learning_rate": 4.987428789662964e-05, "loss": 1.2201, "step": 15850 }, { "epoch": 1.51, "learning_rate": 4.9842462047675126e-05, "loss": 1.2533, "step": 15860 }, { "epoch": 1.51, "learning_rate": 4.98106361987206e-05, "loss": 1.2212, "step": 15870 }, { "epoch": 1.51, "learning_rate": 4.977881034976608e-05, "loss": 1.2752, "step": 15880 }, { "epoch": 1.51, "learning_rate": 4.9746984500811563e-05, "loss": 1.3381, "step": 15890 }, { "epoch": 1.51, "learning_rate": 4.971515865185704e-05, "loss": 1.0002, "step": 15900 }, { "epoch": 1.51, "learning_rate": 4.9683332802902524e-05, "loss": 1.0658, "step": 15910 }, { "epoch": 1.52, "learning_rate": 4.9651506953948e-05, "loss": 1.3124, "step": 15920 }, { "epoch": 1.52, "learning_rate": 4.961968110499348e-05, "loss": 1.4023, "step": 15930 }, { "epoch": 1.52, "learning_rate": 4.9587855256038955e-05, "loss": 1.426, "step": 15940 }, { "epoch": 1.52, "learning_rate": 4.955602940708444e-05, "loss": 1.2818, "step": 15950 }, { "epoch": 1.52, "learning_rate": 4.9524203558129915e-05, "loss": 1.5452, "step": 15960 }, { "epoch": 1.52, "learning_rate": 4.94923777091754e-05, "loss": 1.5367, "step": 15970 }, { "epoch": 1.52, "learning_rate": 4.946055186022087e-05, "loss": 1.2374, "step": 15980 }, { "epoch": 1.52, "learning_rate": 4.942872601126635e-05, "loss": 1.0215, "step": 15990 }, { "epoch": 1.52, "learning_rate": 4.939690016231183e-05, "loss": 1.4311, "step": 16000 }, { "epoch": 1.52, "learning_rate": 4.936507431335731e-05, "loss": 1.419, "step": 16010 }, { "epoch": 1.52, "learning_rate": 4.933324846440279e-05, "loss": 1.1318, "step": 16020 }, { "epoch": 1.53, "learning_rate": 4.930142261544827e-05, "loss": 0.9369, "step": 16030 }, { "epoch": 1.53, "learning_rate": 4.926959676649375e-05, "loss": 1.2709, "step": 16040 }, { "epoch": 1.53, "learning_rate": 4.923777091753923e-05, "loss": 1.0934, "step": 16050 }, { "epoch": 1.53, "learning_rate": 4.920594506858471e-05, "loss": 0.9907, "step": 16060 }, { "epoch": 1.53, "learning_rate": 4.917411921963019e-05, "loss": 1.4633, "step": 16070 }, { "epoch": 1.53, "learning_rate": 4.9142293370675665e-05, "loss": 1.0784, "step": 16080 }, { "epoch": 1.53, "learning_rate": 4.911046752172114e-05, "loss": 1.5323, "step": 16090 }, { "epoch": 1.53, "learning_rate": 4.9078641672766625e-05, "loss": 1.2013, "step": 16100 }, { "epoch": 1.53, "learning_rate": 4.90468158238121e-05, "loss": 1.261, "step": 16110 }, { "epoch": 1.53, "learning_rate": 4.9014989974857586e-05, "loss": 1.4675, "step": 16120 }, { "epoch": 1.54, "learning_rate": 4.8983164125903056e-05, "loss": 1.0573, "step": 16130 }, { "epoch": 1.54, "learning_rate": 4.895133827694854e-05, "loss": 1.0632, "step": 16140 }, { "epoch": 1.54, "learning_rate": 4.8919512427994016e-05, "loss": 1.3218, "step": 16150 }, { "epoch": 1.54, "learning_rate": 4.88876865790395e-05, "loss": 1.0788, "step": 16160 }, { "epoch": 1.54, "learning_rate": 4.8855860730084977e-05, "loss": 1.3465, "step": 16170 }, { "epoch": 1.54, "learning_rate": 4.8824034881130453e-05, "loss": 1.2767, "step": 16180 }, { "epoch": 1.54, "learning_rate": 4.879220903217594e-05, "loss": 1.6768, "step": 16190 }, { "epoch": 1.54, "learning_rate": 4.8760383183221414e-05, "loss": 1.2286, "step": 16200 }, { "epoch": 1.54, "learning_rate": 4.87285573342669e-05, "loss": 1.4228, "step": 16210 }, { "epoch": 1.54, "learning_rate": 4.8696731485312374e-05, "loss": 1.2702, "step": 16220 }, { "epoch": 1.54, "learning_rate": 4.866490563635786e-05, "loss": 1.3598, "step": 16230 }, { "epoch": 1.55, "learning_rate": 4.863307978740333e-05, "loss": 1.2043, "step": 16240 }, { "epoch": 1.55, "learning_rate": 4.860125393844881e-05, "loss": 1.4265, "step": 16250 }, { "epoch": 1.55, "learning_rate": 4.856942808949429e-05, "loss": 1.1054, "step": 16260 }, { "epoch": 1.55, "learning_rate": 4.853760224053977e-05, "loss": 1.5339, "step": 16270 }, { "epoch": 1.55, "learning_rate": 4.850577639158525e-05, "loss": 1.2976, "step": 16280 }, { "epoch": 1.55, "learning_rate": 4.8473950542630726e-05, "loss": 1.4514, "step": 16290 }, { "epoch": 1.55, "learning_rate": 4.84421246936762e-05, "loss": 1.2413, "step": 16300 }, { "epoch": 1.55, "learning_rate": 4.8410298844721687e-05, "loss": 1.0662, "step": 16310 }, { "epoch": 1.55, "learning_rate": 4.8378472995767163e-05, "loss": 1.3955, "step": 16320 }, { "epoch": 1.55, "learning_rate": 4.834664714681265e-05, "loss": 1.208, "step": 16330 }, { "epoch": 1.56, "learning_rate": 4.831482129785812e-05, "loss": 1.4724, "step": 16340 }, { "epoch": 1.56, "learning_rate": 4.82829954489036e-05, "loss": 1.2841, "step": 16350 }, { "epoch": 1.56, "learning_rate": 4.825116959994908e-05, "loss": 1.2628, "step": 16360 }, { "epoch": 1.56, "learning_rate": 4.821934375099456e-05, "loss": 0.989, "step": 16370 }, { "epoch": 1.56, "learning_rate": 4.818751790204004e-05, "loss": 0.9159, "step": 16380 }, { "epoch": 1.56, "learning_rate": 4.8155692053085515e-05, "loss": 1.2558, "step": 16390 }, { "epoch": 1.56, "learning_rate": 4.8123866204131e-05, "loss": 1.0551, "step": 16400 }, { "epoch": 1.56, "learning_rate": 4.8092040355176476e-05, "loss": 1.2573, "step": 16410 }, { "epoch": 1.56, "learning_rate": 4.806021450622196e-05, "loss": 1.4726, "step": 16420 }, { "epoch": 1.56, "learning_rate": 4.8028388657267436e-05, "loss": 1.3828, "step": 16430 }, { "epoch": 1.56, "learning_rate": 4.799656280831291e-05, "loss": 1.2052, "step": 16440 }, { "epoch": 1.57, "learning_rate": 4.796473695935839e-05, "loss": 1.149, "step": 16450 }, { "epoch": 1.57, "learning_rate": 4.7932911110403873e-05, "loss": 1.3104, "step": 16460 }, { "epoch": 1.57, "learning_rate": 4.790108526144935e-05, "loss": 1.3008, "step": 16470 }, { "epoch": 1.57, "learning_rate": 4.7869259412494834e-05, "loss": 1.314, "step": 16480 }, { "epoch": 1.57, "learning_rate": 4.7837433563540304e-05, "loss": 1.2192, "step": 16490 }, { "epoch": 1.57, "learning_rate": 4.780560771458579e-05, "loss": 1.6693, "step": 16500 }, { "epoch": 1.57, "learning_rate": 4.7773781865631265e-05, "loss": 1.2332, "step": 16510 }, { "epoch": 1.57, "learning_rate": 4.774195601667675e-05, "loss": 1.6067, "step": 16520 }, { "epoch": 1.57, "learning_rate": 4.7710130167722225e-05, "loss": 1.1059, "step": 16530 }, { "epoch": 1.57, "learning_rate": 4.767830431876771e-05, "loss": 1.3827, "step": 16540 }, { "epoch": 1.58, "learning_rate": 4.7646478469813186e-05, "loss": 1.2931, "step": 16550 }, { "epoch": 1.58, "learning_rate": 4.761465262085866e-05, "loss": 1.4552, "step": 16560 }, { "epoch": 1.58, "learning_rate": 4.7582826771904146e-05, "loss": 1.1972, "step": 16570 }, { "epoch": 1.58, "learning_rate": 4.755100092294962e-05, "loss": 0.8767, "step": 16580 }, { "epoch": 1.58, "learning_rate": 4.7519175073995107e-05, "loss": 0.9475, "step": 16590 }, { "epoch": 1.58, "learning_rate": 4.7487349225040577e-05, "loss": 1.5473, "step": 16600 }, { "epoch": 1.58, "learning_rate": 4.745552337608606e-05, "loss": 1.269, "step": 16610 }, { "epoch": 1.58, "learning_rate": 4.742369752713154e-05, "loss": 1.3995, "step": 16620 }, { "epoch": 1.58, "learning_rate": 4.739187167817702e-05, "loss": 1.4818, "step": 16630 }, { "epoch": 1.58, "learning_rate": 4.73600458292225e-05, "loss": 1.2699, "step": 16640 }, { "epoch": 1.58, "learning_rate": 4.7328219980267974e-05, "loss": 1.3036, "step": 16650 }, { "epoch": 1.59, "learning_rate": 4.729639413131345e-05, "loss": 1.3475, "step": 16660 }, { "epoch": 1.59, "learning_rate": 4.7264568282358935e-05, "loss": 1.1247, "step": 16670 }, { "epoch": 1.59, "learning_rate": 4.723274243340441e-05, "loss": 1.2363, "step": 16680 }, { "epoch": 1.59, "learning_rate": 4.7200916584449895e-05, "loss": 1.4974, "step": 16690 }, { "epoch": 1.59, "learning_rate": 4.716909073549537e-05, "loss": 1.3437, "step": 16700 }, { "epoch": 1.59, "learning_rate": 4.713726488654085e-05, "loss": 1.162, "step": 16710 }, { "epoch": 1.59, "learning_rate": 4.710543903758633e-05, "loss": 1.3693, "step": 16720 }, { "epoch": 1.59, "learning_rate": 4.707361318863181e-05, "loss": 1.42, "step": 16730 }, { "epoch": 1.59, "learning_rate": 4.704178733967729e-05, "loss": 1.3814, "step": 16740 }, { "epoch": 1.59, "learning_rate": 4.7009961490722763e-05, "loss": 1.1444, "step": 16750 }, { "epoch": 1.6, "learning_rate": 4.697813564176825e-05, "loss": 1.1105, "step": 16760 }, { "epoch": 1.6, "learning_rate": 4.6946309792813724e-05, "loss": 1.2547, "step": 16770 }, { "epoch": 1.6, "learning_rate": 4.691448394385921e-05, "loss": 1.2134, "step": 16780 }, { "epoch": 1.6, "learning_rate": 4.6882658094904684e-05, "loss": 1.3874, "step": 16790 }, { "epoch": 1.6, "learning_rate": 4.685083224595016e-05, "loss": 1.2903, "step": 16800 }, { "epoch": 1.6, "learning_rate": 4.681900639699564e-05, "loss": 1.1891, "step": 16810 }, { "epoch": 1.6, "learning_rate": 4.678718054804112e-05, "loss": 1.1899, "step": 16820 }, { "epoch": 1.6, "learning_rate": 4.67553546990866e-05, "loss": 1.3783, "step": 16830 }, { "epoch": 1.6, "learning_rate": 4.672352885013208e-05, "loss": 1.1659, "step": 16840 }, { "epoch": 1.6, "learning_rate": 4.669170300117756e-05, "loss": 1.4276, "step": 16850 }, { "epoch": 1.6, "learning_rate": 4.6659877152223036e-05, "loss": 1.2746, "step": 16860 }, { "epoch": 1.61, "learning_rate": 4.662805130326851e-05, "loss": 1.222, "step": 16870 }, { "epoch": 1.61, "learning_rate": 4.6596225454313997e-05, "loss": 1.157, "step": 16880 }, { "epoch": 1.61, "learning_rate": 4.6564399605359473e-05, "loss": 1.3028, "step": 16890 }, { "epoch": 1.61, "learning_rate": 4.653257375640496e-05, "loss": 1.4982, "step": 16900 }, { "epoch": 1.61, "learning_rate": 4.6500747907450434e-05, "loss": 1.3935, "step": 16910 }, { "epoch": 1.61, "learning_rate": 4.646892205849591e-05, "loss": 1.5659, "step": 16920 }, { "epoch": 1.61, "learning_rate": 4.6437096209541394e-05, "loss": 1.3223, "step": 16930 }, { "epoch": 1.61, "learning_rate": 4.640527036058687e-05, "loss": 1.1122, "step": 16940 }, { "epoch": 1.61, "learning_rate": 4.6373444511632355e-05, "loss": 1.0243, "step": 16950 }, { "epoch": 1.61, "learning_rate": 4.6341618662677825e-05, "loss": 1.1851, "step": 16960 }, { "epoch": 1.62, "learning_rate": 4.630979281372331e-05, "loss": 1.321, "step": 16970 }, { "epoch": 1.62, "learning_rate": 4.6277966964768786e-05, "loss": 1.5267, "step": 16980 }, { "epoch": 1.62, "learning_rate": 4.624614111581427e-05, "loss": 1.0554, "step": 16990 }, { "epoch": 1.62, "learning_rate": 4.6214315266859746e-05, "loss": 1.2804, "step": 17000 }, { "epoch": 1.62, "learning_rate": 4.618248941790522e-05, "loss": 1.2965, "step": 17010 }, { "epoch": 1.62, "learning_rate": 4.61506635689507e-05, "loss": 1.0722, "step": 17020 }, { "epoch": 1.62, "learning_rate": 4.611883771999618e-05, "loss": 1.5165, "step": 17030 }, { "epoch": 1.62, "learning_rate": 4.608701187104166e-05, "loss": 1.441, "step": 17040 }, { "epoch": 1.62, "learning_rate": 4.6055186022087144e-05, "loss": 0.9771, "step": 17050 }, { "epoch": 1.62, "learning_rate": 4.602336017313262e-05, "loss": 1.3015, "step": 17060 }, { "epoch": 1.62, "learning_rate": 4.59915343241781e-05, "loss": 1.1422, "step": 17070 }, { "epoch": 1.63, "learning_rate": 4.595970847522358e-05, "loss": 1.3455, "step": 17080 }, { "epoch": 1.63, "learning_rate": 4.592788262626906e-05, "loss": 1.004, "step": 17090 }, { "epoch": 1.63, "learning_rate": 4.589605677731454e-05, "loss": 1.0713, "step": 17100 }, { "epoch": 1.63, "learning_rate": 4.586423092836001e-05, "loss": 1.0479, "step": 17110 }, { "epoch": 1.63, "learning_rate": 4.5832405079405495e-05, "loss": 1.4771, "step": 17120 }, { "epoch": 1.63, "learning_rate": 4.580057923045097e-05, "loss": 1.2325, "step": 17130 }, { "epoch": 1.63, "learning_rate": 4.5768753381496456e-05, "loss": 1.262, "step": 17140 }, { "epoch": 1.63, "learning_rate": 4.573692753254193e-05, "loss": 1.4222, "step": 17150 }, { "epoch": 1.63, "learning_rate": 4.5705101683587416e-05, "loss": 1.2611, "step": 17160 }, { "epoch": 1.63, "learning_rate": 4.5673275834632887e-05, "loss": 1.2773, "step": 17170 }, { "epoch": 1.64, "learning_rate": 4.564144998567837e-05, "loss": 1.3417, "step": 17180 }, { "epoch": 1.64, "learning_rate": 4.560962413672385e-05, "loss": 1.2499, "step": 17190 }, { "epoch": 1.64, "learning_rate": 4.557779828776933e-05, "loss": 1.3159, "step": 17200 }, { "epoch": 1.64, "learning_rate": 4.554597243881481e-05, "loss": 1.2028, "step": 17210 }, { "epoch": 1.64, "learning_rate": 4.5514146589860284e-05, "loss": 1.2043, "step": 17220 }, { "epoch": 1.64, "learning_rate": 4.548232074090577e-05, "loss": 0.967, "step": 17230 }, { "epoch": 1.64, "learning_rate": 4.5450494891951245e-05, "loss": 1.2259, "step": 17240 }, { "epoch": 1.64, "learning_rate": 4.541866904299673e-05, "loss": 1.1098, "step": 17250 }, { "epoch": 1.64, "learning_rate": 4.5386843194042205e-05, "loss": 1.2959, "step": 17260 }, { "epoch": 1.64, "learning_rate": 4.535501734508768e-05, "loss": 1.0986, "step": 17270 }, { "epoch": 1.64, "learning_rate": 4.532319149613316e-05, "loss": 1.2249, "step": 17280 }, { "epoch": 1.65, "learning_rate": 4.529136564717864e-05, "loss": 1.3841, "step": 17290 }, { "epoch": 1.65, "learning_rate": 4.525953979822412e-05, "loss": 0.9871, "step": 17300 }, { "epoch": 1.65, "learning_rate": 4.52277139492696e-05, "loss": 1.2703, "step": 17310 }, { "epoch": 1.65, "learning_rate": 4.5195888100315073e-05, "loss": 1.224, "step": 17320 }, { "epoch": 1.65, "learning_rate": 4.516406225136056e-05, "loss": 1.1882, "step": 17330 }, { "epoch": 1.65, "learning_rate": 4.5132236402406034e-05, "loss": 1.0581, "step": 17340 }, { "epoch": 1.65, "learning_rate": 4.510041055345152e-05, "loss": 1.4191, "step": 17350 }, { "epoch": 1.65, "learning_rate": 4.5068584704496994e-05, "loss": 1.3129, "step": 17360 }, { "epoch": 1.65, "learning_rate": 4.503675885554247e-05, "loss": 1.4355, "step": 17370 }, { "epoch": 1.65, "learning_rate": 4.5004933006587955e-05, "loss": 1.4007, "step": 17380 }, { "epoch": 1.66, "learning_rate": 4.497310715763343e-05, "loss": 1.235, "step": 17390 }, { "epoch": 1.66, "learning_rate": 4.494128130867891e-05, "loss": 1.124, "step": 17400 }, { "epoch": 1.66, "learning_rate": 4.490945545972439e-05, "loss": 1.3663, "step": 17410 }, { "epoch": 1.66, "learning_rate": 4.487762961076987e-05, "loss": 1.2598, "step": 17420 }, { "epoch": 1.66, "learning_rate": 4.4845803761815346e-05, "loss": 1.2365, "step": 17430 }, { "epoch": 1.66, "learning_rate": 4.481397791286083e-05, "loss": 1.3799, "step": 17440 }, { "epoch": 1.66, "learning_rate": 4.4782152063906306e-05, "loss": 1.1697, "step": 17450 }, { "epoch": 1.66, "learning_rate": 4.475032621495179e-05, "loss": 1.6229, "step": 17460 }, { "epoch": 1.66, "learning_rate": 4.471850036599727e-05, "loss": 1.3189, "step": 17470 }, { "epoch": 1.66, "learning_rate": 4.4686674517042744e-05, "loss": 1.3392, "step": 17480 }, { "epoch": 1.66, "learning_rate": 4.465484866808822e-05, "loss": 1.5232, "step": 17490 }, { "epoch": 1.67, "learning_rate": 4.4623022819133704e-05, "loss": 1.3666, "step": 17500 }, { "epoch": 1.67, "learning_rate": 4.459119697017918e-05, "loss": 1.1053, "step": 17510 }, { "epoch": 1.67, "learning_rate": 4.4559371121224665e-05, "loss": 1.102, "step": 17520 }, { "epoch": 1.67, "learning_rate": 4.4527545272270135e-05, "loss": 0.7984, "step": 17530 }, { "epoch": 1.67, "learning_rate": 4.449571942331562e-05, "loss": 1.175, "step": 17540 }, { "epoch": 1.67, "learning_rate": 4.4463893574361095e-05, "loss": 1.2059, "step": 17550 }, { "epoch": 1.67, "learning_rate": 4.443206772540658e-05, "loss": 1.4837, "step": 17560 }, { "epoch": 1.67, "learning_rate": 4.4400241876452056e-05, "loss": 1.2114, "step": 17570 }, { "epoch": 1.67, "learning_rate": 4.436841602749753e-05, "loss": 1.278, "step": 17580 }, { "epoch": 1.67, "learning_rate": 4.4336590178543016e-05, "loss": 1.2685, "step": 17590 }, { "epoch": 1.68, "learning_rate": 4.430476432958849e-05, "loss": 1.3156, "step": 17600 }, { "epoch": 1.68, "learning_rate": 4.427293848063398e-05, "loss": 1.1934, "step": 17610 }, { "epoch": 1.68, "learning_rate": 4.4241112631679454e-05, "loss": 1.2918, "step": 17620 }, { "epoch": 1.68, "learning_rate": 4.420928678272493e-05, "loss": 1.1634, "step": 17630 }, { "epoch": 1.68, "learning_rate": 4.417746093377041e-05, "loss": 0.9987, "step": 17640 }, { "epoch": 1.68, "learning_rate": 4.414563508481589e-05, "loss": 1.4489, "step": 17650 }, { "epoch": 1.68, "learning_rate": 4.411380923586137e-05, "loss": 1.2199, "step": 17660 }, { "epoch": 1.68, "learning_rate": 4.408198338690685e-05, "loss": 1.1415, "step": 17670 }, { "epoch": 1.68, "learning_rate": 4.405015753795232e-05, "loss": 1.5274, "step": 17680 }, { "epoch": 1.68, "learning_rate": 4.4018331688997805e-05, "loss": 1.21, "step": 17690 }, { "epoch": 1.68, "learning_rate": 4.398650584004328e-05, "loss": 0.9351, "step": 17700 }, { "epoch": 1.69, "learning_rate": 4.3954679991088766e-05, "loss": 1.3924, "step": 17710 }, { "epoch": 1.69, "learning_rate": 4.392285414213424e-05, "loss": 1.0862, "step": 17720 }, { "epoch": 1.69, "learning_rate": 4.389102829317972e-05, "loss": 1.2006, "step": 17730 }, { "epoch": 1.69, "learning_rate": 4.38592024442252e-05, "loss": 1.3581, "step": 17740 }, { "epoch": 1.69, "learning_rate": 4.382737659527068e-05, "loss": 1.091, "step": 17750 }, { "epoch": 1.69, "learning_rate": 4.3795550746316164e-05, "loss": 1.537, "step": 17760 }, { "epoch": 1.69, "learning_rate": 4.376372489736164e-05, "loss": 1.29, "step": 17770 }, { "epoch": 1.69, "learning_rate": 4.373189904840712e-05, "loss": 1.2013, "step": 17780 }, { "epoch": 1.69, "learning_rate": 4.3700073199452594e-05, "loss": 1.23, "step": 17790 }, { "epoch": 1.69, "learning_rate": 4.366824735049808e-05, "loss": 1.4847, "step": 17800 }, { "epoch": 1.7, "learning_rate": 4.3636421501543555e-05, "loss": 1.3086, "step": 17810 }, { "epoch": 1.7, "learning_rate": 4.360459565258904e-05, "loss": 1.303, "step": 17820 }, { "epoch": 1.7, "learning_rate": 4.3572769803634515e-05, "loss": 1.271, "step": 17830 }, { "epoch": 1.7, "learning_rate": 4.354094395467999e-05, "loss": 1.2053, "step": 17840 }, { "epoch": 1.7, "learning_rate": 4.350911810572547e-05, "loss": 1.3765, "step": 17850 }, { "epoch": 1.7, "learning_rate": 4.347729225677095e-05, "loss": 1.2532, "step": 17860 }, { "epoch": 1.7, "learning_rate": 4.344546640781643e-05, "loss": 1.2865, "step": 17870 }, { "epoch": 1.7, "learning_rate": 4.341364055886191e-05, "loss": 1.2211, "step": 17880 }, { "epoch": 1.7, "learning_rate": 4.338181470990739e-05, "loss": 1.4564, "step": 17890 }, { "epoch": 1.7, "learning_rate": 4.334998886095287e-05, "loss": 1.0033, "step": 17900 }, { "epoch": 1.7, "learning_rate": 4.331816301199835e-05, "loss": 0.962, "step": 17910 }, { "epoch": 1.71, "learning_rate": 4.328633716304383e-05, "loss": 1.6116, "step": 17920 }, { "epoch": 1.71, "learning_rate": 4.3254511314089304e-05, "loss": 1.0693, "step": 17930 }, { "epoch": 1.71, "learning_rate": 4.322268546513478e-05, "loss": 1.4096, "step": 17940 }, { "epoch": 1.71, "learning_rate": 4.3190859616180265e-05, "loss": 1.226, "step": 17950 }, { "epoch": 1.71, "learning_rate": 4.315903376722574e-05, "loss": 1.0396, "step": 17960 }, { "epoch": 1.71, "learning_rate": 4.3127207918271225e-05, "loss": 1.2501, "step": 17970 }, { "epoch": 1.71, "learning_rate": 4.30953820693167e-05, "loss": 1.2864, "step": 17980 }, { "epoch": 1.71, "learning_rate": 4.306355622036218e-05, "loss": 1.2269, "step": 17990 }, { "epoch": 1.71, "learning_rate": 4.3031730371407656e-05, "loss": 1.116, "step": 18000 }, { "epoch": 1.71, "learning_rate": 4.299990452245314e-05, "loss": 1.3539, "step": 18010 }, { "epoch": 1.72, "learning_rate": 4.2968078673498616e-05, "loss": 1.5695, "step": 18020 }, { "epoch": 1.72, "learning_rate": 4.29362528245441e-05, "loss": 1.3146, "step": 18030 }, { "epoch": 1.72, "learning_rate": 4.290442697558957e-05, "loss": 1.1909, "step": 18040 }, { "epoch": 1.72, "learning_rate": 4.2872601126635054e-05, "loss": 1.3509, "step": 18050 }, { "epoch": 1.72, "learning_rate": 4.284077527768053e-05, "loss": 1.3447, "step": 18060 }, { "epoch": 1.72, "learning_rate": 4.2808949428726014e-05, "loss": 1.258, "step": 18070 }, { "epoch": 1.72, "learning_rate": 4.277712357977149e-05, "loss": 1.5396, "step": 18080 }, { "epoch": 1.72, "learning_rate": 4.274529773081697e-05, "loss": 1.5036, "step": 18090 }, { "epoch": 1.72, "learning_rate": 4.271347188186245e-05, "loss": 0.9903, "step": 18100 }, { "epoch": 1.72, "learning_rate": 4.268164603290793e-05, "loss": 1.188, "step": 18110 }, { "epoch": 1.72, "learning_rate": 4.264982018395341e-05, "loss": 1.4617, "step": 18120 }, { "epoch": 1.73, "learning_rate": 4.261799433499889e-05, "loss": 1.5529, "step": 18130 }, { "epoch": 1.73, "learning_rate": 4.258616848604437e-05, "loss": 0.9911, "step": 18140 }, { "epoch": 1.73, "learning_rate": 4.255434263708984e-05, "loss": 1.2402, "step": 18150 }, { "epoch": 1.73, "learning_rate": 4.2522516788135326e-05, "loss": 1.1673, "step": 18160 }, { "epoch": 1.73, "learning_rate": 4.24906909391808e-05, "loss": 1.5972, "step": 18170 }, { "epoch": 1.73, "learning_rate": 4.245886509022629e-05, "loss": 1.3765, "step": 18180 }, { "epoch": 1.73, "learning_rate": 4.2427039241271764e-05, "loss": 1.0896, "step": 18190 }, { "epoch": 1.73, "learning_rate": 4.239521339231724e-05, "loss": 1.0541, "step": 18200 }, { "epoch": 1.73, "learning_rate": 4.236338754336272e-05, "loss": 1.2994, "step": 18210 }, { "epoch": 1.73, "learning_rate": 4.23315616944082e-05, "loss": 1.5104, "step": 18220 }, { "epoch": 1.74, "learning_rate": 4.229973584545368e-05, "loss": 1.7087, "step": 18230 }, { "epoch": 1.74, "learning_rate": 4.226790999649916e-05, "loss": 1.5083, "step": 18240 }, { "epoch": 1.74, "learning_rate": 4.223608414754464e-05, "loss": 1.5388, "step": 18250 }, { "epoch": 1.74, "learning_rate": 4.2204258298590115e-05, "loss": 1.3103, "step": 18260 }, { "epoch": 1.74, "learning_rate": 4.21724324496356e-05, "loss": 1.2792, "step": 18270 }, { "epoch": 1.74, "learning_rate": 4.2140606600681076e-05, "loss": 1.1251, "step": 18280 }, { "epoch": 1.74, "learning_rate": 4.210878075172656e-05, "loss": 1.2299, "step": 18290 }, { "epoch": 1.74, "learning_rate": 4.207695490277203e-05, "loss": 1.2919, "step": 18300 }, { "epoch": 1.74, "learning_rate": 4.204512905381751e-05, "loss": 1.0468, "step": 18310 }, { "epoch": 1.74, "learning_rate": 4.201330320486299e-05, "loss": 1.4261, "step": 18320 }, { "epoch": 1.74, "learning_rate": 4.1981477355908474e-05, "loss": 1.2958, "step": 18330 }, { "epoch": 1.75, "learning_rate": 4.194965150695395e-05, "loss": 1.3843, "step": 18340 }, { "epoch": 1.75, "learning_rate": 4.191782565799943e-05, "loss": 1.0905, "step": 18350 }, { "epoch": 1.75, "learning_rate": 4.1885999809044904e-05, "loss": 1.3191, "step": 18360 }, { "epoch": 1.75, "learning_rate": 4.185417396009039e-05, "loss": 1.3327, "step": 18370 }, { "epoch": 1.75, "learning_rate": 4.1822348111135865e-05, "loss": 0.9465, "step": 18380 }, { "epoch": 1.75, "learning_rate": 4.179052226218135e-05, "loss": 1.1875, "step": 18390 }, { "epoch": 1.75, "learning_rate": 4.1758696413226825e-05, "loss": 0.936, "step": 18400 }, { "epoch": 1.75, "learning_rate": 4.17268705642723e-05, "loss": 1.3193, "step": 18410 }, { "epoch": 1.75, "learning_rate": 4.1695044715317786e-05, "loss": 1.1251, "step": 18420 }, { "epoch": 1.75, "learning_rate": 4.166321886636326e-05, "loss": 1.0732, "step": 18430 }, { "epoch": 1.76, "learning_rate": 4.1631393017408746e-05, "loss": 1.5964, "step": 18440 }, { "epoch": 1.76, "learning_rate": 4.159956716845422e-05, "loss": 1.3061, "step": 18450 }, { "epoch": 1.76, "learning_rate": 4.15677413194997e-05, "loss": 1.5171, "step": 18460 }, { "epoch": 1.76, "learning_rate": 4.153591547054518e-05, "loss": 1.2409, "step": 18470 }, { "epoch": 1.76, "learning_rate": 4.150408962159066e-05, "loss": 1.4052, "step": 18480 }, { "epoch": 1.76, "learning_rate": 4.147226377263614e-05, "loss": 1.053, "step": 18490 }, { "epoch": 1.76, "learning_rate": 4.144043792368162e-05, "loss": 1.3229, "step": 18500 }, { "epoch": 1.76, "learning_rate": 4.140861207472709e-05, "loss": 1.4538, "step": 18510 }, { "epoch": 1.76, "learning_rate": 4.1376786225772575e-05, "loss": 1.2916, "step": 18520 }, { "epoch": 1.76, "learning_rate": 4.134496037681805e-05, "loss": 0.9643, "step": 18530 }, { "epoch": 1.76, "learning_rate": 4.1313134527863535e-05, "loss": 1.4604, "step": 18540 }, { "epoch": 1.77, "learning_rate": 4.128130867890901e-05, "loss": 1.2127, "step": 18550 }, { "epoch": 1.77, "learning_rate": 4.124948282995449e-05, "loss": 1.0547, "step": 18560 }, { "epoch": 1.77, "learning_rate": 4.1217656980999966e-05, "loss": 1.2186, "step": 18570 }, { "epoch": 1.77, "learning_rate": 4.118583113204545e-05, "loss": 1.0003, "step": 18580 }, { "epoch": 1.77, "learning_rate": 4.1154005283090926e-05, "loss": 1.2351, "step": 18590 }, { "epoch": 1.77, "learning_rate": 4.112217943413641e-05, "loss": 1.4122, "step": 18600 }, { "epoch": 1.77, "learning_rate": 4.109035358518189e-05, "loss": 1.438, "step": 18610 }, { "epoch": 1.77, "learning_rate": 4.1058527736227364e-05, "loss": 1.196, "step": 18620 }, { "epoch": 1.77, "learning_rate": 4.102670188727285e-05, "loss": 1.062, "step": 18630 }, { "epoch": 1.77, "learning_rate": 4.0994876038318324e-05, "loss": 1.3738, "step": 18640 }, { "epoch": 1.78, "learning_rate": 4.096305018936381e-05, "loss": 1.1569, "step": 18650 }, { "epoch": 1.78, "learning_rate": 4.093122434040928e-05, "loss": 1.2354, "step": 18660 }, { "epoch": 1.78, "learning_rate": 4.089939849145476e-05, "loss": 0.9465, "step": 18670 }, { "epoch": 1.78, "learning_rate": 4.086757264250024e-05, "loss": 1.1662, "step": 18680 }, { "epoch": 1.78, "learning_rate": 4.083574679354572e-05, "loss": 1.3165, "step": 18690 }, { "epoch": 1.78, "learning_rate": 4.08039209445912e-05, "loss": 1.5689, "step": 18700 }, { "epoch": 1.78, "learning_rate": 4.0772095095636676e-05, "loss": 1.2426, "step": 18710 }, { "epoch": 1.78, "learning_rate": 4.074026924668215e-05, "loss": 1.2528, "step": 18720 }, { "epoch": 1.78, "learning_rate": 4.0708443397727636e-05, "loss": 1.4647, "step": 18730 }, { "epoch": 1.78, "learning_rate": 4.067661754877311e-05, "loss": 1.0062, "step": 18740 }, { "epoch": 1.78, "learning_rate": 4.06447916998186e-05, "loss": 1.1861, "step": 18750 }, { "epoch": 1.79, "learning_rate": 4.0612965850864074e-05, "loss": 1.3053, "step": 18760 }, { "epoch": 1.79, "learning_rate": 4.058114000190955e-05, "loss": 1.3605, "step": 18770 }, { "epoch": 1.79, "learning_rate": 4.0549314152955034e-05, "loss": 1.2673, "step": 18780 }, { "epoch": 1.79, "learning_rate": 4.051748830400051e-05, "loss": 1.2602, "step": 18790 }, { "epoch": 1.79, "learning_rate": 4.0485662455045995e-05, "loss": 1.3034, "step": 18800 }, { "epoch": 1.79, "learning_rate": 4.045383660609147e-05, "loss": 1.2683, "step": 18810 }, { "epoch": 1.79, "learning_rate": 4.042201075713695e-05, "loss": 1.3468, "step": 18820 }, { "epoch": 1.79, "learning_rate": 4.0390184908182425e-05, "loss": 1.1144, "step": 18830 }, { "epoch": 1.79, "learning_rate": 4.035835905922791e-05, "loss": 1.3821, "step": 18840 }, { "epoch": 1.79, "learning_rate": 4.0326533210273386e-05, "loss": 1.3033, "step": 18850 }, { "epoch": 1.79, "learning_rate": 4.029470736131887e-05, "loss": 1.4069, "step": 18860 }, { "epoch": 1.8, "learning_rate": 4.026288151236434e-05, "loss": 1.3087, "step": 18870 }, { "epoch": 1.8, "learning_rate": 4.023105566340982e-05, "loss": 1.3872, "step": 18880 }, { "epoch": 1.8, "learning_rate": 4.01992298144553e-05, "loss": 1.4077, "step": 18890 }, { "epoch": 1.8, "learning_rate": 4.0167403965500784e-05, "loss": 1.172, "step": 18900 }, { "epoch": 1.8, "learning_rate": 4.013557811654626e-05, "loss": 1.2941, "step": 18910 }, { "epoch": 1.8, "learning_rate": 4.010375226759174e-05, "loss": 1.4621, "step": 18920 }, { "epoch": 1.8, "learning_rate": 4.007192641863722e-05, "loss": 1.2763, "step": 18930 }, { "epoch": 1.8, "learning_rate": 4.00401005696827e-05, "loss": 1.392, "step": 18940 }, { "epoch": 1.8, "learning_rate": 4.000827472072818e-05, "loss": 1.4797, "step": 18950 }, { "epoch": 1.8, "learning_rate": 3.997644887177366e-05, "loss": 0.9922, "step": 18960 }, { "epoch": 1.81, "learning_rate": 3.9944623022819135e-05, "loss": 1.2553, "step": 18970 }, { "epoch": 1.81, "learning_rate": 3.991279717386461e-05, "loss": 1.1796, "step": 18980 }, { "epoch": 1.81, "learning_rate": 3.9880971324910096e-05, "loss": 1.098, "step": 18990 }, { "epoch": 1.81, "learning_rate": 3.984914547595557e-05, "loss": 1.3635, "step": 19000 }, { "epoch": 1.81, "learning_rate": 3.9817319627001056e-05, "loss": 1.5246, "step": 19010 }, { "epoch": 1.81, "learning_rate": 3.9785493778046526e-05, "loss": 1.3422, "step": 19020 }, { "epoch": 1.81, "learning_rate": 3.975366792909201e-05, "loss": 1.1719, "step": 19030 }, { "epoch": 1.81, "learning_rate": 3.972184208013749e-05, "loss": 1.3276, "step": 19040 }, { "epoch": 1.81, "learning_rate": 3.969001623118297e-05, "loss": 1.4109, "step": 19050 }, { "epoch": 1.81, "learning_rate": 3.965819038222845e-05, "loss": 1.3793, "step": 19060 }, { "epoch": 1.81, "learning_rate": 3.962636453327393e-05, "loss": 1.2927, "step": 19070 }, { "epoch": 1.82, "learning_rate": 3.95945386843194e-05, "loss": 1.3429, "step": 19080 }, { "epoch": 1.82, "learning_rate": 3.9562712835364885e-05, "loss": 0.9858, "step": 19090 }, { "epoch": 1.82, "learning_rate": 3.953088698641036e-05, "loss": 1.4387, "step": 19100 }, { "epoch": 1.82, "learning_rate": 3.9499061137455845e-05, "loss": 1.4169, "step": 19110 }, { "epoch": 1.82, "learning_rate": 3.946723528850132e-05, "loss": 1.1409, "step": 19120 }, { "epoch": 1.82, "learning_rate": 3.94354094395468e-05, "loss": 1.133, "step": 19130 }, { "epoch": 1.82, "learning_rate": 3.940358359059228e-05, "loss": 1.2668, "step": 19140 }, { "epoch": 1.82, "learning_rate": 3.937175774163776e-05, "loss": 1.2295, "step": 19150 }, { "epoch": 1.82, "learning_rate": 3.933993189268324e-05, "loss": 1.3709, "step": 19160 }, { "epoch": 1.82, "learning_rate": 3.930810604372872e-05, "loss": 1.4518, "step": 19170 }, { "epoch": 1.83, "learning_rate": 3.92762801947742e-05, "loss": 0.9845, "step": 19180 }, { "epoch": 1.83, "learning_rate": 3.9244454345819674e-05, "loss": 1.2617, "step": 19190 }, { "epoch": 1.83, "learning_rate": 3.921262849686516e-05, "loss": 1.1381, "step": 19200 }, { "epoch": 1.83, "learning_rate": 3.9180802647910634e-05, "loss": 1.007, "step": 19210 }, { "epoch": 1.83, "learning_rate": 3.914897679895612e-05, "loss": 1.2175, "step": 19220 }, { "epoch": 1.83, "learning_rate": 3.911715095000159e-05, "loss": 1.2663, "step": 19230 }, { "epoch": 1.83, "learning_rate": 3.908532510104707e-05, "loss": 1.1347, "step": 19240 }, { "epoch": 1.83, "learning_rate": 3.905349925209255e-05, "loss": 1.1037, "step": 19250 }, { "epoch": 1.83, "learning_rate": 3.902167340313803e-05, "loss": 1.2356, "step": 19260 }, { "epoch": 1.83, "learning_rate": 3.898984755418351e-05, "loss": 1.355, "step": 19270 }, { "epoch": 1.83, "learning_rate": 3.8958021705228986e-05, "loss": 1.2678, "step": 19280 }, { "epoch": 1.84, "learning_rate": 3.892619585627447e-05, "loss": 1.295, "step": 19290 }, { "epoch": 1.84, "learning_rate": 3.8894370007319946e-05, "loss": 1.1144, "step": 19300 }, { "epoch": 1.84, "learning_rate": 3.886254415836543e-05, "loss": 1.0753, "step": 19310 }, { "epoch": 1.84, "learning_rate": 3.883071830941091e-05, "loss": 1.2428, "step": 19320 }, { "epoch": 1.84, "learning_rate": 3.8798892460456384e-05, "loss": 1.2343, "step": 19330 }, { "epoch": 1.84, "learning_rate": 3.876706661150186e-05, "loss": 1.2299, "step": 19340 }, { "epoch": 1.84, "learning_rate": 3.8735240762547344e-05, "loss": 1.2575, "step": 19350 }, { "epoch": 1.84, "learning_rate": 3.870341491359282e-05, "loss": 1.3199, "step": 19360 }, { "epoch": 1.84, "learning_rate": 3.8671589064638305e-05, "loss": 1.2286, "step": 19370 }, { "epoch": 1.84, "learning_rate": 3.863976321568378e-05, "loss": 1.1311, "step": 19380 }, { "epoch": 1.85, "learning_rate": 3.860793736672926e-05, "loss": 1.1787, "step": 19390 }, { "epoch": 1.85, "learning_rate": 3.8576111517774735e-05, "loss": 1.3663, "step": 19400 }, { "epoch": 1.85, "learning_rate": 3.854428566882022e-05, "loss": 1.28, "step": 19410 }, { "epoch": 1.85, "learning_rate": 3.8512459819865696e-05, "loss": 1.488, "step": 19420 }, { "epoch": 1.85, "learning_rate": 3.848063397091118e-05, "loss": 1.0685, "step": 19430 }, { "epoch": 1.85, "learning_rate": 3.8448808121956656e-05, "loss": 1.1896, "step": 19440 }, { "epoch": 1.85, "learning_rate": 3.841698227300213e-05, "loss": 1.2928, "step": 19450 }, { "epoch": 1.85, "learning_rate": 3.838515642404762e-05, "loss": 1.1183, "step": 19460 }, { "epoch": 1.85, "learning_rate": 3.8353330575093094e-05, "loss": 1.6208, "step": 19470 }, { "epoch": 1.85, "learning_rate": 3.832150472613858e-05, "loss": 0.9104, "step": 19480 }, { "epoch": 1.85, "learning_rate": 3.828967887718405e-05, "loss": 1.4046, "step": 19490 }, { "epoch": 1.86, "learning_rate": 3.825785302822953e-05, "loss": 1.0457, "step": 19500 }, { "epoch": 1.86, "learning_rate": 3.822602717927501e-05, "loss": 1.1939, "step": 19510 }, { "epoch": 1.86, "learning_rate": 3.819420133032049e-05, "loss": 1.0983, "step": 19520 }, { "epoch": 1.86, "learning_rate": 3.816237548136597e-05, "loss": 1.5828, "step": 19530 }, { "epoch": 1.86, "learning_rate": 3.8130549632411445e-05, "loss": 1.2187, "step": 19540 }, { "epoch": 1.86, "learning_rate": 3.809872378345692e-05, "loss": 1.1794, "step": 19550 }, { "epoch": 1.86, "learning_rate": 3.8066897934502406e-05, "loss": 0.9364, "step": 19560 }, { "epoch": 1.86, "learning_rate": 3.803507208554788e-05, "loss": 1.4084, "step": 19570 }, { "epoch": 1.86, "learning_rate": 3.8003246236593366e-05, "loss": 1.4533, "step": 19580 }, { "epoch": 1.86, "learning_rate": 3.7971420387638836e-05, "loss": 1.0989, "step": 19590 }, { "epoch": 1.87, "learning_rate": 3.793959453868432e-05, "loss": 1.1185, "step": 19600 }, { "epoch": 1.87, "learning_rate": 3.79077686897298e-05, "loss": 1.0785, "step": 19610 }, { "epoch": 1.87, "learning_rate": 3.787594284077528e-05, "loss": 1.3841, "step": 19620 }, { "epoch": 1.87, "learning_rate": 3.784411699182076e-05, "loss": 1.2914, "step": 19630 }, { "epoch": 1.87, "learning_rate": 3.7812291142866234e-05, "loss": 1.2935, "step": 19640 }, { "epoch": 1.87, "learning_rate": 3.778046529391172e-05, "loss": 1.0084, "step": 19650 }, { "epoch": 1.87, "learning_rate": 3.7748639444957195e-05, "loss": 1.2559, "step": 19660 }, { "epoch": 1.87, "learning_rate": 3.771681359600268e-05, "loss": 1.5865, "step": 19670 }, { "epoch": 1.87, "learning_rate": 3.7684987747048155e-05, "loss": 1.2636, "step": 19680 }, { "epoch": 1.87, "learning_rate": 3.765316189809363e-05, "loss": 1.2656, "step": 19690 }, { "epoch": 1.87, "learning_rate": 3.762133604913911e-05, "loss": 0.9276, "step": 19700 }, { "epoch": 1.88, "learning_rate": 3.758951020018459e-05, "loss": 1.5347, "step": 19710 }, { "epoch": 1.88, "learning_rate": 3.755768435123007e-05, "loss": 1.053, "step": 19720 }, { "epoch": 1.88, "learning_rate": 3.752585850227555e-05, "loss": 1.3505, "step": 19730 }, { "epoch": 1.88, "learning_rate": 3.749403265332103e-05, "loss": 1.33, "step": 19740 }, { "epoch": 1.88, "learning_rate": 3.746220680436651e-05, "loss": 0.9352, "step": 19750 }, { "epoch": 1.88, "learning_rate": 3.7430380955411984e-05, "loss": 1.2561, "step": 19760 }, { "epoch": 1.88, "learning_rate": 3.739855510645747e-05, "loss": 1.0789, "step": 19770 }, { "epoch": 1.88, "learning_rate": 3.7366729257502944e-05, "loss": 1.2284, "step": 19780 }, { "epoch": 1.88, "learning_rate": 3.733490340854843e-05, "loss": 1.2263, "step": 19790 }, { "epoch": 1.88, "learning_rate": 3.7303077559593905e-05, "loss": 1.0512, "step": 19800 }, { "epoch": 1.89, "learning_rate": 3.727125171063938e-05, "loss": 1.0489, "step": 19810 }, { "epoch": 1.89, "learning_rate": 3.7239425861684865e-05, "loss": 1.4009, "step": 19820 }, { "epoch": 1.89, "learning_rate": 3.720760001273034e-05, "loss": 1.1195, "step": 19830 }, { "epoch": 1.89, "learning_rate": 3.7175774163775826e-05, "loss": 0.8015, "step": 19840 }, { "epoch": 1.89, "learning_rate": 3.7143948314821296e-05, "loss": 1.4234, "step": 19850 }, { "epoch": 1.89, "learning_rate": 3.711212246586678e-05, "loss": 1.3675, "step": 19860 }, { "epoch": 1.89, "learning_rate": 3.7080296616912256e-05, "loss": 0.9495, "step": 19870 }, { "epoch": 1.89, "learning_rate": 3.704847076795774e-05, "loss": 1.1224, "step": 19880 }, { "epoch": 1.89, "learning_rate": 3.701664491900322e-05, "loss": 1.409, "step": 19890 }, { "epoch": 1.89, "learning_rate": 3.6984819070048694e-05, "loss": 1.3952, "step": 19900 }, { "epoch": 1.89, "learning_rate": 3.695299322109417e-05, "loss": 1.2911, "step": 19910 }, { "epoch": 1.9, "learning_rate": 3.6921167372139654e-05, "loss": 1.1573, "step": 19920 }, { "epoch": 1.9, "learning_rate": 3.688934152318513e-05, "loss": 1.2936, "step": 19930 }, { "epoch": 1.9, "learning_rate": 3.6857515674230615e-05, "loss": 1.2558, "step": 19940 }, { "epoch": 1.9, "learning_rate": 3.682568982527609e-05, "loss": 1.1733, "step": 19950 }, { "epoch": 1.9, "learning_rate": 3.679386397632157e-05, "loss": 1.19, "step": 19960 }, { "epoch": 1.9, "learning_rate": 3.676203812736705e-05, "loss": 1.022, "step": 19970 }, { "epoch": 1.9, "learning_rate": 3.673021227841253e-05, "loss": 1.1534, "step": 19980 }, { "epoch": 1.9, "learning_rate": 3.669838642945801e-05, "loss": 1.1945, "step": 19990 }, { "epoch": 1.9, "learning_rate": 3.666656058050348e-05, "loss": 1.2391, "step": 20000 }, { "epoch": 1.9, "learning_rate": 3.6634734731548966e-05, "loss": 1.133, "step": 20010 }, { "epoch": 1.91, "learning_rate": 3.660290888259444e-05, "loss": 1.1127, "step": 20020 }, { "epoch": 1.91, "learning_rate": 3.657108303363993e-05, "loss": 1.2856, "step": 20030 }, { "epoch": 1.91, "learning_rate": 3.6539257184685404e-05, "loss": 1.2292, "step": 20040 }, { "epoch": 1.91, "learning_rate": 3.650743133573089e-05, "loss": 1.5068, "step": 20050 }, { "epoch": 1.91, "learning_rate": 3.647560548677636e-05, "loss": 1.3691, "step": 20060 }, { "epoch": 1.91, "learning_rate": 3.644377963782184e-05, "loss": 1.2278, "step": 20070 }, { "epoch": 1.91, "learning_rate": 3.641195378886732e-05, "loss": 1.142, "step": 20080 }, { "epoch": 1.91, "learning_rate": 3.63801279399128e-05, "loss": 1.4245, "step": 20090 }, { "epoch": 1.91, "learning_rate": 3.634830209095828e-05, "loss": 1.2538, "step": 20100 }, { "epoch": 1.91, "learning_rate": 3.6316476242003755e-05, "loss": 1.0865, "step": 20110 }, { "epoch": 1.91, "learning_rate": 3.628465039304923e-05, "loss": 1.5445, "step": 20120 }, { "epoch": 1.92, "learning_rate": 3.6252824544094716e-05, "loss": 1.1644, "step": 20130 }, { "epoch": 1.92, "learning_rate": 3.622099869514019e-05, "loss": 1.1228, "step": 20140 }, { "epoch": 1.92, "learning_rate": 3.6189172846185676e-05, "loss": 1.2033, "step": 20150 }, { "epoch": 1.92, "learning_rate": 3.615734699723115e-05, "loss": 1.2266, "step": 20160 }, { "epoch": 1.92, "learning_rate": 3.612552114827663e-05, "loss": 1.2489, "step": 20170 }, { "epoch": 1.92, "learning_rate": 3.6093695299322114e-05, "loss": 1.1645, "step": 20180 }, { "epoch": 1.92, "learning_rate": 3.606186945036759e-05, "loss": 1.1382, "step": 20190 }, { "epoch": 1.92, "learning_rate": 3.6030043601413074e-05, "loss": 1.0547, "step": 20200 }, { "epoch": 1.92, "learning_rate": 3.5998217752458544e-05, "loss": 1.0566, "step": 20210 }, { "epoch": 1.92, "learning_rate": 3.596639190350403e-05, "loss": 1.305, "step": 20220 }, { "epoch": 1.93, "learning_rate": 3.5934566054549505e-05, "loss": 1.3903, "step": 20230 }, { "epoch": 1.93, "learning_rate": 3.590274020559499e-05, "loss": 1.3047, "step": 20240 }, { "epoch": 1.93, "learning_rate": 3.5870914356640465e-05, "loss": 1.1542, "step": 20250 }, { "epoch": 1.93, "learning_rate": 3.583908850768594e-05, "loss": 1.1636, "step": 20260 }, { "epoch": 1.93, "learning_rate": 3.580726265873142e-05, "loss": 0.9835, "step": 20270 }, { "epoch": 1.93, "learning_rate": 3.57754368097769e-05, "loss": 1.1189, "step": 20280 }, { "epoch": 1.93, "learning_rate": 3.574361096082238e-05, "loss": 1.448, "step": 20290 }, { "epoch": 1.93, "learning_rate": 3.571178511186786e-05, "loss": 0.9613, "step": 20300 }, { "epoch": 1.93, "learning_rate": 3.567995926291334e-05, "loss": 1.092, "step": 20310 }, { "epoch": 1.93, "learning_rate": 3.564813341395882e-05, "loss": 1.2723, "step": 20320 }, { "epoch": 1.93, "learning_rate": 3.56163075650043e-05, "loss": 1.37, "step": 20330 }, { "epoch": 1.94, "learning_rate": 3.558448171604978e-05, "loss": 1.0441, "step": 20340 }, { "epoch": 1.94, "learning_rate": 3.555265586709526e-05, "loss": 0.9713, "step": 20350 }, { "epoch": 1.94, "learning_rate": 3.552083001814074e-05, "loss": 1.1469, "step": 20360 }, { "epoch": 1.94, "learning_rate": 3.5489004169186215e-05, "loss": 1.1932, "step": 20370 }, { "epoch": 1.94, "learning_rate": 3.545717832023169e-05, "loss": 1.4208, "step": 20380 }, { "epoch": 1.94, "learning_rate": 3.5425352471277175e-05, "loss": 1.3295, "step": 20390 }, { "epoch": 1.94, "learning_rate": 3.539352662232265e-05, "loss": 1.1444, "step": 20400 }, { "epoch": 1.94, "learning_rate": 3.5361700773368136e-05, "loss": 1.3685, "step": 20410 }, { "epoch": 1.94, "learning_rate": 3.5329874924413606e-05, "loss": 1.3868, "step": 20420 }, { "epoch": 1.94, "learning_rate": 3.529804907545909e-05, "loss": 1.1486, "step": 20430 }, { "epoch": 1.95, "learning_rate": 3.5266223226504566e-05, "loss": 1.1494, "step": 20440 }, { "epoch": 1.95, "learning_rate": 3.523439737755005e-05, "loss": 1.3013, "step": 20450 }, { "epoch": 1.95, "learning_rate": 3.520257152859553e-05, "loss": 1.0297, "step": 20460 }, { "epoch": 1.95, "learning_rate": 3.5170745679641004e-05, "loss": 1.2975, "step": 20470 }, { "epoch": 1.95, "learning_rate": 3.513891983068649e-05, "loss": 1.2635, "step": 20480 }, { "epoch": 1.95, "learning_rate": 3.5107093981731964e-05, "loss": 1.3708, "step": 20490 }, { "epoch": 1.95, "learning_rate": 3.507526813277745e-05, "loss": 1.0579, "step": 20500 }, { "epoch": 1.95, "learning_rate": 3.5043442283822925e-05, "loss": 1.4045, "step": 20510 }, { "epoch": 1.95, "learning_rate": 3.50116164348684e-05, "loss": 1.2214, "step": 20520 }, { "epoch": 1.95, "learning_rate": 3.497979058591388e-05, "loss": 1.3855, "step": 20530 }, { "epoch": 1.95, "learning_rate": 3.494796473695936e-05, "loss": 1.2537, "step": 20540 }, { "epoch": 1.96, "learning_rate": 3.491613888800484e-05, "loss": 1.1688, "step": 20550 }, { "epoch": 1.96, "learning_rate": 3.488431303905032e-05, "loss": 1.2166, "step": 20560 }, { "epoch": 1.96, "learning_rate": 3.485248719009579e-05, "loss": 1.0007, "step": 20570 }, { "epoch": 1.96, "learning_rate": 3.4820661341141276e-05, "loss": 1.5996, "step": 20580 }, { "epoch": 1.96, "learning_rate": 3.478883549218675e-05, "loss": 1.2712, "step": 20590 }, { "epoch": 1.96, "learning_rate": 3.475700964323224e-05, "loss": 1.0818, "step": 20600 }, { "epoch": 1.96, "learning_rate": 3.4725183794277714e-05, "loss": 1.3156, "step": 20610 }, { "epoch": 1.96, "learning_rate": 3.469335794532319e-05, "loss": 1.0122, "step": 20620 }, { "epoch": 1.96, "learning_rate": 3.466153209636867e-05, "loss": 1.098, "step": 20630 }, { "epoch": 1.96, "learning_rate": 3.462970624741415e-05, "loss": 1.158, "step": 20640 }, { "epoch": 1.97, "learning_rate": 3.459788039845963e-05, "loss": 1.1666, "step": 20650 }, { "epoch": 1.97, "learning_rate": 3.456605454950511e-05, "loss": 1.2442, "step": 20660 }, { "epoch": 1.97, "learning_rate": 3.453422870055059e-05, "loss": 1.3601, "step": 20670 }, { "epoch": 1.97, "learning_rate": 3.4502402851596065e-05, "loss": 1.0865, "step": 20680 }, { "epoch": 1.97, "learning_rate": 3.447057700264155e-05, "loss": 1.121, "step": 20690 }, { "epoch": 1.97, "learning_rate": 3.4438751153687026e-05, "loss": 1.148, "step": 20700 }, { "epoch": 1.97, "learning_rate": 3.440692530473251e-05, "loss": 1.2275, "step": 20710 }, { "epoch": 1.97, "learning_rate": 3.4375099455777986e-05, "loss": 0.8846, "step": 20720 }, { "epoch": 1.97, "learning_rate": 3.434327360682346e-05, "loss": 1.3629, "step": 20730 }, { "epoch": 1.97, "learning_rate": 3.431144775786894e-05, "loss": 1.3894, "step": 20740 }, { "epoch": 1.97, "learning_rate": 3.4279621908914424e-05, "loss": 1.1238, "step": 20750 }, { "epoch": 1.98, "learning_rate": 3.42477960599599e-05, "loss": 1.3383, "step": 20760 }, { "epoch": 1.98, "learning_rate": 3.4215970211005384e-05, "loss": 1.1081, "step": 20770 }, { "epoch": 1.98, "learning_rate": 3.4184144362050854e-05, "loss": 1.4528, "step": 20780 }, { "epoch": 1.98, "learning_rate": 3.415231851309634e-05, "loss": 1.2063, "step": 20790 }, { "epoch": 1.98, "learning_rate": 3.4120492664141815e-05, "loss": 0.9596, "step": 20800 }, { "epoch": 1.98, "learning_rate": 3.40886668151873e-05, "loss": 1.4973, "step": 20810 }, { "epoch": 1.98, "learning_rate": 3.4056840966232775e-05, "loss": 1.2717, "step": 20820 }, { "epoch": 1.98, "learning_rate": 3.402501511727825e-05, "loss": 1.4137, "step": 20830 }, { "epoch": 1.98, "learning_rate": 3.3993189268323736e-05, "loss": 1.1622, "step": 20840 }, { "epoch": 1.98, "learning_rate": 3.396136341936921e-05, "loss": 0.962, "step": 20850 }, { "epoch": 1.99, "learning_rate": 3.3929537570414696e-05, "loss": 1.1181, "step": 20860 }, { "epoch": 1.99, "learning_rate": 3.389771172146017e-05, "loss": 1.2876, "step": 20870 }, { "epoch": 1.99, "learning_rate": 3.386588587250565e-05, "loss": 1.5209, "step": 20880 }, { "epoch": 1.99, "learning_rate": 3.383406002355113e-05, "loss": 1.1866, "step": 20890 }, { "epoch": 1.99, "learning_rate": 3.380223417459661e-05, "loss": 1.1722, "step": 20900 }, { "epoch": 1.99, "learning_rate": 3.377040832564209e-05, "loss": 1.1426, "step": 20910 }, { "epoch": 1.99, "learning_rate": 3.373858247668757e-05, "loss": 1.0596, "step": 20920 }, { "epoch": 1.99, "learning_rate": 3.370675662773304e-05, "loss": 1.2358, "step": 20930 }, { "epoch": 1.99, "learning_rate": 3.3674930778778525e-05, "loss": 0.8627, "step": 20940 }, { "epoch": 1.99, "learning_rate": 3.3643104929824e-05, "loss": 1.1639, "step": 20950 }, { "epoch": 1.99, "learning_rate": 3.3611279080869485e-05, "loss": 1.3031, "step": 20960 }, { "epoch": 2.0, "learning_rate": 3.357945323191496e-05, "loss": 0.9835, "step": 20970 }, { "epoch": 2.0, "learning_rate": 3.3547627382960446e-05, "loss": 1.3002, "step": 20980 }, { "epoch": 2.0, "learning_rate": 3.351580153400592e-05, "loss": 1.5188, "step": 20990 }, { "epoch": 2.0, "learning_rate": 3.34839756850514e-05, "loss": 1.0337, "step": 21000 } ], "max_steps": 31521, "num_train_epochs": 3, "total_flos": 1.9910406818664284e+18, "trial_name": null, "trial_params": null }