{ "best_metric": null, "best_model_checkpoint": null, "epoch": 1.0451505016722409, "global_step": 5000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 9.992682416893165e-06, "loss": 1.2194, "step": 10 }, { "epoch": 0.0, "learning_rate": 9.98222872674054e-06, "loss": 1.149, "step": 20 }, { "epoch": 0.01, "learning_rate": 9.971775036587917e-06, "loss": 1.0958, "step": 30 }, { "epoch": 0.01, "learning_rate": 9.961321346435293e-06, "loss": 1.0791, "step": 40 }, { "epoch": 0.01, "learning_rate": 9.950867656282669e-06, "loss": 1.1471, "step": 50 }, { "epoch": 0.01, "learning_rate": 9.940413966130045e-06, "loss": 1.1, "step": 60 }, { "epoch": 0.01, "learning_rate": 9.92996027597742e-06, "loss": 1.0613, "step": 70 }, { "epoch": 0.02, "learning_rate": 9.919506585824797e-06, "loss": 0.9759, "step": 80 }, { "epoch": 0.02, "learning_rate": 9.909052895672173e-06, "loss": 1.0735, "step": 90 }, { "epoch": 0.02, "learning_rate": 9.898599205519549e-06, "loss": 1.0464, "step": 100 }, { "epoch": 0.02, "learning_rate": 9.888145515366925e-06, "loss": 1.032, "step": 110 }, { "epoch": 0.03, "learning_rate": 9.877691825214302e-06, "loss": 0.9523, "step": 120 }, { "epoch": 0.03, "learning_rate": 9.867238135061677e-06, "loss": 1.0126, "step": 130 }, { "epoch": 0.03, "learning_rate": 9.856784444909053e-06, "loss": 1.0132, "step": 140 }, { "epoch": 0.03, "learning_rate": 9.84633075475643e-06, "loss": 1.0023, "step": 150 }, { "epoch": 0.03, "learning_rate": 9.835877064603807e-06, "loss": 0.9862, "step": 160 }, { "epoch": 0.04, "learning_rate": 9.825423374451183e-06, "loss": 0.9808, "step": 170 }, { "epoch": 0.04, "learning_rate": 9.814969684298559e-06, "loss": 0.9914, "step": 180 }, { "epoch": 0.04, "learning_rate": 9.804515994145935e-06, "loss": 1.0308, "step": 190 }, { "epoch": 0.04, "learning_rate": 9.79406230399331e-06, "loss": 1.0149, "step": 200 }, { "epoch": 0.04, "learning_rate": 9.783608613840687e-06, "loss": 0.936, "step": 210 }, { "epoch": 0.05, "learning_rate": 9.773154923688063e-06, "loss": 0.984, "step": 220 }, { "epoch": 0.05, "learning_rate": 9.762701233535439e-06, "loss": 0.9522, "step": 230 }, { "epoch": 0.05, "learning_rate": 9.752247543382815e-06, "loss": 0.9451, "step": 240 }, { "epoch": 0.05, "learning_rate": 9.74179385323019e-06, "loss": 0.9503, "step": 250 }, { "epoch": 0.05, "learning_rate": 9.731340163077567e-06, "loss": 1.0045, "step": 260 }, { "epoch": 0.06, "learning_rate": 9.720886472924943e-06, "loss": 0.9055, "step": 270 }, { "epoch": 0.06, "learning_rate": 9.71043278277232e-06, "loss": 0.9534, "step": 280 }, { "epoch": 0.06, "learning_rate": 9.699979092619696e-06, "loss": 0.9923, "step": 290 }, { "epoch": 0.06, "learning_rate": 9.689525402467072e-06, "loss": 1.0029, "step": 300 }, { "epoch": 0.06, "learning_rate": 9.679071712314447e-06, "loss": 0.9809, "step": 310 }, { "epoch": 0.07, "learning_rate": 9.668618022161824e-06, "loss": 0.9893, "step": 320 }, { "epoch": 0.07, "learning_rate": 9.6581643320092e-06, "loss": 0.9599, "step": 330 }, { "epoch": 0.07, "learning_rate": 9.647710641856576e-06, "loss": 0.995, "step": 340 }, { "epoch": 0.07, "learning_rate": 9.637256951703952e-06, "loss": 0.9273, "step": 350 }, { "epoch": 0.08, "learning_rate": 9.626803261551328e-06, "loss": 0.9223, "step": 360 }, { "epoch": 0.08, "learning_rate": 9.616349571398704e-06, "loss": 0.9966, "step": 370 }, { "epoch": 0.08, "learning_rate": 9.60589588124608e-06, "loss": 1.0168, "step": 380 }, { "epoch": 0.08, "learning_rate": 9.595442191093456e-06, "loss": 0.9002, "step": 390 }, { "epoch": 0.08, "learning_rate": 9.584988500940832e-06, "loss": 0.9722, "step": 400 }, { "epoch": 0.09, "learning_rate": 9.574534810788208e-06, "loss": 0.8962, "step": 410 }, { "epoch": 0.09, "learning_rate": 9.564081120635586e-06, "loss": 0.925, "step": 420 }, { "epoch": 0.09, "learning_rate": 9.55362743048296e-06, "loss": 0.9328, "step": 430 }, { "epoch": 0.09, "learning_rate": 9.543173740330336e-06, "loss": 0.9643, "step": 440 }, { "epoch": 0.09, "learning_rate": 9.532720050177714e-06, "loss": 0.9741, "step": 450 }, { "epoch": 0.1, "learning_rate": 9.52226636002509e-06, "loss": 0.9868, "step": 460 }, { "epoch": 0.1, "learning_rate": 9.511812669872466e-06, "loss": 0.9064, "step": 470 }, { "epoch": 0.1, "learning_rate": 9.50135897971984e-06, "loss": 0.9556, "step": 480 }, { "epoch": 0.1, "learning_rate": 9.490905289567218e-06, "loss": 0.8907, "step": 490 }, { "epoch": 0.1, "learning_rate": 9.480451599414594e-06, "loss": 0.9452, "step": 500 }, { "epoch": 0.11, "learning_rate": 9.46999790926197e-06, "loss": 0.8982, "step": 510 }, { "epoch": 0.11, "learning_rate": 9.459544219109346e-06, "loss": 0.9202, "step": 520 }, { "epoch": 0.11, "learning_rate": 9.449090528956722e-06, "loss": 0.9163, "step": 530 }, { "epoch": 0.11, "learning_rate": 9.438636838804098e-06, "loss": 1.0045, "step": 540 }, { "epoch": 0.11, "learning_rate": 9.428183148651476e-06, "loss": 0.9638, "step": 550 }, { "epoch": 0.12, "learning_rate": 9.41772945849885e-06, "loss": 1.0104, "step": 560 }, { "epoch": 0.12, "learning_rate": 9.407275768346226e-06, "loss": 0.9684, "step": 570 }, { "epoch": 0.12, "learning_rate": 9.396822078193602e-06, "loss": 1.0572, "step": 580 }, { "epoch": 0.12, "learning_rate": 9.38636838804098e-06, "loss": 0.93, "step": 590 }, { "epoch": 0.13, "learning_rate": 9.375914697888356e-06, "loss": 0.8864, "step": 600 }, { "epoch": 0.13, "learning_rate": 9.36546100773573e-06, "loss": 0.9299, "step": 610 }, { "epoch": 0.13, "learning_rate": 9.355007317583108e-06, "loss": 0.9593, "step": 620 }, { "epoch": 0.13, "learning_rate": 9.344553627430484e-06, "loss": 1.0237, "step": 630 }, { "epoch": 0.13, "learning_rate": 9.33409993727786e-06, "loss": 0.8734, "step": 640 }, { "epoch": 0.14, "learning_rate": 9.323646247125236e-06, "loss": 0.9851, "step": 650 }, { "epoch": 0.14, "learning_rate": 9.313192556972612e-06, "loss": 0.9216, "step": 660 }, { "epoch": 0.14, "learning_rate": 9.302738866819988e-06, "loss": 1.0198, "step": 670 }, { "epoch": 0.14, "learning_rate": 9.292285176667364e-06, "loss": 0.905, "step": 680 }, { "epoch": 0.14, "learning_rate": 9.28183148651474e-06, "loss": 1.0569, "step": 690 }, { "epoch": 0.15, "learning_rate": 9.271377796362116e-06, "loss": 0.923, "step": 700 }, { "epoch": 0.15, "learning_rate": 9.260924106209492e-06, "loss": 0.9127, "step": 710 }, { "epoch": 0.15, "learning_rate": 9.25047041605687e-06, "loss": 0.9825, "step": 720 }, { "epoch": 0.15, "learning_rate": 9.240016725904246e-06, "loss": 0.9216, "step": 730 }, { "epoch": 0.15, "learning_rate": 9.22956303575162e-06, "loss": 0.977, "step": 740 }, { "epoch": 0.16, "learning_rate": 9.219109345598998e-06, "loss": 0.8867, "step": 750 }, { "epoch": 0.16, "learning_rate": 9.208655655446374e-06, "loss": 0.916, "step": 760 }, { "epoch": 0.16, "learning_rate": 9.19820196529375e-06, "loss": 0.9983, "step": 770 }, { "epoch": 0.16, "learning_rate": 9.187748275141126e-06, "loss": 0.9232, "step": 780 }, { "epoch": 0.17, "learning_rate": 9.177294584988502e-06, "loss": 0.9142, "step": 790 }, { "epoch": 0.17, "learning_rate": 9.166840894835878e-06, "loss": 0.9104, "step": 800 }, { "epoch": 0.17, "learning_rate": 9.156387204683254e-06, "loss": 0.9219, "step": 810 }, { "epoch": 0.17, "learning_rate": 9.14593351453063e-06, "loss": 0.9512, "step": 820 }, { "epoch": 0.17, "learning_rate": 9.135479824378006e-06, "loss": 0.9737, "step": 830 }, { "epoch": 0.18, "learning_rate": 9.125026134225382e-06, "loss": 0.9051, "step": 840 }, { "epoch": 0.18, "learning_rate": 9.114572444072758e-06, "loss": 0.9342, "step": 850 }, { "epoch": 0.18, "learning_rate": 9.104118753920134e-06, "loss": 0.9987, "step": 860 }, { "epoch": 0.18, "learning_rate": 9.09366506376751e-06, "loss": 0.9175, "step": 870 }, { "epoch": 0.18, "learning_rate": 9.083211373614886e-06, "loss": 0.966, "step": 880 }, { "epoch": 0.19, "learning_rate": 9.072757683462264e-06, "loss": 0.8929, "step": 890 }, { "epoch": 0.19, "learning_rate": 9.06230399330964e-06, "loss": 1.0035, "step": 900 }, { "epoch": 0.19, "learning_rate": 9.051850303157016e-06, "loss": 0.9159, "step": 910 }, { "epoch": 0.19, "learning_rate": 9.041396613004392e-06, "loss": 0.8713, "step": 920 }, { "epoch": 0.19, "learning_rate": 9.030942922851768e-06, "loss": 0.9428, "step": 930 }, { "epoch": 0.2, "learning_rate": 9.020489232699144e-06, "loss": 0.9068, "step": 940 }, { "epoch": 0.2, "learning_rate": 9.01003554254652e-06, "loss": 0.9257, "step": 950 }, { "epoch": 0.2, "learning_rate": 8.999581852393896e-06, "loss": 0.8979, "step": 960 }, { "epoch": 0.2, "learning_rate": 8.989128162241272e-06, "loss": 0.9692, "step": 970 }, { "epoch": 0.2, "learning_rate": 8.978674472088648e-06, "loss": 0.891, "step": 980 }, { "epoch": 0.21, "learning_rate": 8.968220781936024e-06, "loss": 0.9504, "step": 990 }, { "epoch": 0.21, "learning_rate": 8.9577670917834e-06, "loss": 0.8841, "step": 1000 }, { "epoch": 0.21, "learning_rate": 8.947313401630776e-06, "loss": 0.9707, "step": 1010 }, { "epoch": 0.21, "learning_rate": 8.936859711478153e-06, "loss": 0.9047, "step": 1020 }, { "epoch": 0.22, "learning_rate": 8.92640602132553e-06, "loss": 0.9127, "step": 1030 }, { "epoch": 0.22, "learning_rate": 8.915952331172904e-06, "loss": 0.9204, "step": 1040 }, { "epoch": 0.22, "learning_rate": 8.90549864102028e-06, "loss": 0.9721, "step": 1050 }, { "epoch": 0.22, "learning_rate": 8.895044950867658e-06, "loss": 0.8829, "step": 1060 }, { "epoch": 0.22, "learning_rate": 8.884591260715034e-06, "loss": 0.9252, "step": 1070 }, { "epoch": 0.23, "learning_rate": 8.87413757056241e-06, "loss": 0.9248, "step": 1080 }, { "epoch": 0.23, "learning_rate": 8.863683880409786e-06, "loss": 0.9494, "step": 1090 }, { "epoch": 0.23, "learning_rate": 8.853230190257162e-06, "loss": 0.9655, "step": 1100 }, { "epoch": 0.23, "learning_rate": 8.842776500104538e-06, "loss": 0.9307, "step": 1110 }, { "epoch": 0.23, "learning_rate": 8.832322809951914e-06, "loss": 0.888, "step": 1120 }, { "epoch": 0.24, "learning_rate": 8.82186911979929e-06, "loss": 0.931, "step": 1130 }, { "epoch": 0.24, "learning_rate": 8.811415429646666e-06, "loss": 0.9236, "step": 1140 }, { "epoch": 0.24, "learning_rate": 8.800961739494042e-06, "loss": 0.9462, "step": 1150 }, { "epoch": 0.24, "learning_rate": 8.79050804934142e-06, "loss": 0.9676, "step": 1160 }, { "epoch": 0.24, "learning_rate": 8.780054359188794e-06, "loss": 0.9015, "step": 1170 }, { "epoch": 0.25, "learning_rate": 8.76960066903617e-06, "loss": 0.9533, "step": 1180 }, { "epoch": 0.25, "learning_rate": 8.759146978883547e-06, "loss": 0.8977, "step": 1190 }, { "epoch": 0.25, "learning_rate": 8.748693288730923e-06, "loss": 0.861, "step": 1200 }, { "epoch": 0.25, "learning_rate": 8.7382395985783e-06, "loss": 0.9662, "step": 1210 }, { "epoch": 0.26, "learning_rate": 8.727785908425675e-06, "loss": 0.9749, "step": 1220 }, { "epoch": 0.26, "learning_rate": 8.717332218273051e-06, "loss": 0.8911, "step": 1230 }, { "epoch": 0.26, "learning_rate": 8.706878528120427e-06, "loss": 0.9127, "step": 1240 }, { "epoch": 0.26, "learning_rate": 8.696424837967803e-06, "loss": 0.8746, "step": 1250 }, { "epoch": 0.26, "learning_rate": 8.68597114781518e-06, "loss": 0.9181, "step": 1260 }, { "epoch": 0.27, "learning_rate": 8.675517457662555e-06, "loss": 0.9101, "step": 1270 }, { "epoch": 0.27, "learning_rate": 8.665063767509931e-06, "loss": 0.8764, "step": 1280 }, { "epoch": 0.27, "learning_rate": 8.654610077357309e-06, "loss": 0.8952, "step": 1290 }, { "epoch": 0.27, "learning_rate": 8.644156387204683e-06, "loss": 0.9548, "step": 1300 }, { "epoch": 0.27, "learning_rate": 8.63370269705206e-06, "loss": 0.955, "step": 1310 }, { "epoch": 0.28, "learning_rate": 8.623249006899435e-06, "loss": 0.9375, "step": 1320 }, { "epoch": 0.28, "learning_rate": 8.612795316746813e-06, "loss": 0.9587, "step": 1330 }, { "epoch": 0.28, "learning_rate": 8.60234162659419e-06, "loss": 0.9171, "step": 1340 }, { "epoch": 0.28, "learning_rate": 8.591887936441563e-06, "loss": 0.971, "step": 1350 }, { "epoch": 0.28, "learning_rate": 8.581434246288941e-06, "loss": 0.9503, "step": 1360 }, { "epoch": 0.29, "learning_rate": 8.570980556136317e-06, "loss": 0.9779, "step": 1370 }, { "epoch": 0.29, "learning_rate": 8.560526865983693e-06, "loss": 0.8802, "step": 1380 }, { "epoch": 0.29, "learning_rate": 8.55007317583107e-06, "loss": 0.9075, "step": 1390 }, { "epoch": 0.29, "learning_rate": 8.539619485678445e-06, "loss": 0.8784, "step": 1400 }, { "epoch": 0.29, "learning_rate": 8.529165795525821e-06, "loss": 0.9136, "step": 1410 }, { "epoch": 0.3, "learning_rate": 8.518712105373197e-06, "loss": 0.9078, "step": 1420 }, { "epoch": 0.3, "learning_rate": 8.508258415220573e-06, "loss": 0.981, "step": 1430 }, { "epoch": 0.3, "learning_rate": 8.49780472506795e-06, "loss": 0.9346, "step": 1440 }, { "epoch": 0.3, "learning_rate": 8.487351034915325e-06, "loss": 0.8989, "step": 1450 }, { "epoch": 0.31, "learning_rate": 8.476897344762703e-06, "loss": 0.9462, "step": 1460 }, { "epoch": 0.31, "learning_rate": 8.466443654610077e-06, "loss": 0.8866, "step": 1470 }, { "epoch": 0.31, "learning_rate": 8.455989964457453e-06, "loss": 0.8557, "step": 1480 }, { "epoch": 0.31, "learning_rate": 8.445536274304831e-06, "loss": 0.9636, "step": 1490 }, { "epoch": 0.31, "learning_rate": 8.435082584152207e-06, "loss": 0.8861, "step": 1500 }, { "epoch": 0.32, "learning_rate": 8.424628893999583e-06, "loss": 0.95, "step": 1510 }, { "epoch": 0.32, "learning_rate": 8.414175203846957e-06, "loss": 0.8743, "step": 1520 }, { "epoch": 0.32, "learning_rate": 8.403721513694335e-06, "loss": 0.8983, "step": 1530 }, { "epoch": 0.32, "learning_rate": 8.393267823541711e-06, "loss": 0.9549, "step": 1540 }, { "epoch": 0.32, "learning_rate": 8.382814133389087e-06, "loss": 1.0213, "step": 1550 }, { "epoch": 0.33, "learning_rate": 8.372360443236463e-06, "loss": 0.9463, "step": 1560 }, { "epoch": 0.33, "learning_rate": 8.361906753083839e-06, "loss": 1.0107, "step": 1570 }, { "epoch": 0.33, "learning_rate": 8.351453062931215e-06, "loss": 0.9134, "step": 1580 }, { "epoch": 0.33, "learning_rate": 8.340999372778593e-06, "loss": 0.9037, "step": 1590 }, { "epoch": 0.33, "learning_rate": 8.330545682625967e-06, "loss": 1.0, "step": 1600 }, { "epoch": 0.34, "learning_rate": 8.320091992473343e-06, "loss": 0.9319, "step": 1610 }, { "epoch": 0.34, "learning_rate": 8.309638302320719e-06, "loss": 0.8916, "step": 1620 }, { "epoch": 0.34, "learning_rate": 8.299184612168097e-06, "loss": 0.9546, "step": 1630 }, { "epoch": 0.34, "learning_rate": 8.288730922015473e-06, "loss": 0.9687, "step": 1640 }, { "epoch": 0.34, "learning_rate": 8.278277231862847e-06, "loss": 0.888, "step": 1650 }, { "epoch": 0.35, "learning_rate": 8.267823541710225e-06, "loss": 0.924, "step": 1660 }, { "epoch": 0.35, "learning_rate": 8.257369851557601e-06, "loss": 0.9409, "step": 1670 }, { "epoch": 0.35, "learning_rate": 8.246916161404977e-06, "loss": 0.9448, "step": 1680 }, { "epoch": 0.35, "learning_rate": 8.236462471252353e-06, "loss": 0.8798, "step": 1690 }, { "epoch": 0.36, "learning_rate": 8.226008781099729e-06, "loss": 0.8185, "step": 1700 }, { "epoch": 0.36, "learning_rate": 8.215555090947105e-06, "loss": 0.8991, "step": 1710 }, { "epoch": 0.36, "learning_rate": 8.205101400794481e-06, "loss": 0.9261, "step": 1720 }, { "epoch": 0.36, "learning_rate": 8.194647710641857e-06, "loss": 0.9167, "step": 1730 }, { "epoch": 0.36, "learning_rate": 8.184194020489233e-06, "loss": 0.9146, "step": 1740 }, { "epoch": 0.37, "learning_rate": 8.173740330336609e-06, "loss": 0.8829, "step": 1750 }, { "epoch": 0.37, "learning_rate": 8.163286640183987e-06, "loss": 0.8972, "step": 1760 }, { "epoch": 0.37, "learning_rate": 8.152832950031363e-06, "loss": 0.9056, "step": 1770 }, { "epoch": 0.37, "learning_rate": 8.142379259878737e-06, "loss": 0.8516, "step": 1780 }, { "epoch": 0.37, "learning_rate": 8.131925569726113e-06, "loss": 0.9755, "step": 1790 }, { "epoch": 0.38, "learning_rate": 8.12147187957349e-06, "loss": 0.9584, "step": 1800 }, { "epoch": 0.38, "learning_rate": 8.111018189420867e-06, "loss": 0.9286, "step": 1810 }, { "epoch": 0.38, "learning_rate": 8.100564499268243e-06, "loss": 0.9321, "step": 1820 }, { "epoch": 0.38, "learning_rate": 8.090110809115619e-06, "loss": 0.8993, "step": 1830 }, { "epoch": 0.38, "learning_rate": 8.079657118962995e-06, "loss": 0.8684, "step": 1840 }, { "epoch": 0.39, "learning_rate": 8.06920342881037e-06, "loss": 0.9466, "step": 1850 }, { "epoch": 0.39, "learning_rate": 8.058749738657747e-06, "loss": 0.8792, "step": 1860 }, { "epoch": 0.39, "learning_rate": 8.048296048505123e-06, "loss": 0.9253, "step": 1870 }, { "epoch": 0.39, "learning_rate": 8.037842358352499e-06, "loss": 0.9671, "step": 1880 }, { "epoch": 0.4, "learning_rate": 8.027388668199875e-06, "loss": 0.9155, "step": 1890 }, { "epoch": 0.4, "learning_rate": 8.016934978047252e-06, "loss": 0.9275, "step": 1900 }, { "epoch": 0.4, "learning_rate": 8.006481287894627e-06, "loss": 0.9119, "step": 1910 }, { "epoch": 0.4, "learning_rate": 7.996027597742003e-06, "loss": 0.8613, "step": 1920 }, { "epoch": 0.4, "learning_rate": 7.98557390758938e-06, "loss": 0.8958, "step": 1930 }, { "epoch": 0.41, "learning_rate": 7.975120217436757e-06, "loss": 0.9375, "step": 1940 }, { "epoch": 0.41, "learning_rate": 7.964666527284133e-06, "loss": 0.9341, "step": 1950 }, { "epoch": 0.41, "learning_rate": 7.954212837131509e-06, "loss": 0.8851, "step": 1960 }, { "epoch": 0.41, "learning_rate": 7.943759146978885e-06, "loss": 0.8582, "step": 1970 }, { "epoch": 0.41, "learning_rate": 7.93330545682626e-06, "loss": 0.9287, "step": 1980 }, { "epoch": 0.42, "learning_rate": 7.922851766673637e-06, "loss": 0.8866, "step": 1990 }, { "epoch": 0.42, "learning_rate": 7.912398076521013e-06, "loss": 0.9085, "step": 2000 }, { "epoch": 0.42, "learning_rate": 7.901944386368389e-06, "loss": 0.9195, "step": 2010 }, { "epoch": 0.42, "learning_rate": 7.891490696215765e-06, "loss": 0.8571, "step": 2020 }, { "epoch": 0.42, "learning_rate": 7.88103700606314e-06, "loss": 0.8742, "step": 2030 }, { "epoch": 0.43, "learning_rate": 7.870583315910517e-06, "loss": 0.9843, "step": 2040 }, { "epoch": 0.43, "learning_rate": 7.860129625757893e-06, "loss": 0.9768, "step": 2050 }, { "epoch": 0.43, "learning_rate": 7.849675935605269e-06, "loss": 0.9654, "step": 2060 }, { "epoch": 0.43, "learning_rate": 7.839222245452646e-06, "loss": 0.8788, "step": 2070 }, { "epoch": 0.43, "learning_rate": 7.82876855530002e-06, "loss": 0.8983, "step": 2080 }, { "epoch": 0.44, "learning_rate": 7.818314865147397e-06, "loss": 0.8889, "step": 2090 }, { "epoch": 0.44, "learning_rate": 7.807861174994774e-06, "loss": 0.9339, "step": 2100 }, { "epoch": 0.44, "learning_rate": 7.79740748484215e-06, "loss": 0.9302, "step": 2110 }, { "epoch": 0.44, "learning_rate": 7.786953794689526e-06, "loss": 0.8483, "step": 2120 }, { "epoch": 0.45, "learning_rate": 7.776500104536902e-06, "loss": 0.9208, "step": 2130 }, { "epoch": 0.45, "learning_rate": 7.766046414384278e-06, "loss": 0.9515, "step": 2140 }, { "epoch": 0.45, "learning_rate": 7.755592724231654e-06, "loss": 0.8451, "step": 2150 }, { "epoch": 0.45, "learning_rate": 7.74513903407903e-06, "loss": 0.9629, "step": 2160 }, { "epoch": 0.45, "learning_rate": 7.734685343926406e-06, "loss": 0.8991, "step": 2170 }, { "epoch": 0.46, "learning_rate": 7.724231653773782e-06, "loss": 0.8856, "step": 2180 }, { "epoch": 0.46, "learning_rate": 7.713777963621158e-06, "loss": 0.901, "step": 2190 }, { "epoch": 0.46, "learning_rate": 7.703324273468536e-06, "loss": 0.896, "step": 2200 }, { "epoch": 0.46, "learning_rate": 7.69287058331591e-06, "loss": 0.9733, "step": 2210 }, { "epoch": 0.46, "learning_rate": 7.682416893163286e-06, "loss": 0.9834, "step": 2220 }, { "epoch": 0.47, "learning_rate": 7.671963203010664e-06, "loss": 0.8649, "step": 2230 }, { "epoch": 0.47, "learning_rate": 7.66150951285804e-06, "loss": 0.9071, "step": 2240 }, { "epoch": 0.47, "learning_rate": 7.651055822705416e-06, "loss": 0.9331, "step": 2250 }, { "epoch": 0.47, "learning_rate": 7.64060213255279e-06, "loss": 0.9564, "step": 2260 }, { "epoch": 0.47, "learning_rate": 7.630148442400168e-06, "loss": 0.9174, "step": 2270 }, { "epoch": 0.48, "learning_rate": 7.619694752247544e-06, "loss": 0.9184, "step": 2280 }, { "epoch": 0.48, "learning_rate": 7.609241062094919e-06, "loss": 0.8832, "step": 2290 }, { "epoch": 0.48, "learning_rate": 7.598787371942297e-06, "loss": 1.0121, "step": 2300 }, { "epoch": 0.48, "learning_rate": 7.588333681789672e-06, "loss": 0.8368, "step": 2310 }, { "epoch": 0.48, "learning_rate": 7.577879991637048e-06, "loss": 0.938, "step": 2320 }, { "epoch": 0.49, "learning_rate": 7.567426301484425e-06, "loss": 0.8101, "step": 2330 }, { "epoch": 0.49, "learning_rate": 7.556972611331801e-06, "loss": 0.9026, "step": 2340 }, { "epoch": 0.49, "learning_rate": 7.546518921179177e-06, "loss": 0.9025, "step": 2350 }, { "epoch": 0.49, "learning_rate": 7.536065231026552e-06, "loss": 0.8843, "step": 2360 }, { "epoch": 0.5, "learning_rate": 7.525611540873929e-06, "loss": 0.9459, "step": 2370 }, { "epoch": 0.5, "learning_rate": 7.515157850721305e-06, "loss": 0.8736, "step": 2380 }, { "epoch": 0.5, "learning_rate": 7.504704160568681e-06, "loss": 0.9031, "step": 2390 }, { "epoch": 0.5, "learning_rate": 7.494250470416058e-06, "loss": 0.8856, "step": 2400 }, { "epoch": 0.5, "learning_rate": 7.483796780263434e-06, "loss": 0.9425, "step": 2410 }, { "epoch": 0.51, "learning_rate": 7.473343090110809e-06, "loss": 0.9604, "step": 2420 }, { "epoch": 0.51, "learning_rate": 7.462889399958186e-06, "loss": 0.9365, "step": 2430 }, { "epoch": 0.51, "learning_rate": 7.452435709805562e-06, "loss": 0.8897, "step": 2440 }, { "epoch": 0.51, "learning_rate": 7.441982019652938e-06, "loss": 0.9331, "step": 2450 }, { "epoch": 0.51, "learning_rate": 7.431528329500314e-06, "loss": 0.9166, "step": 2460 }, { "epoch": 0.52, "learning_rate": 7.421074639347691e-06, "loss": 0.9193, "step": 2470 }, { "epoch": 0.52, "learning_rate": 7.410620949195066e-06, "loss": 0.8827, "step": 2480 }, { "epoch": 0.52, "learning_rate": 7.400167259042442e-06, "loss": 0.9035, "step": 2490 }, { "epoch": 0.52, "learning_rate": 7.389713568889819e-06, "loss": 0.9055, "step": 2500 }, { "epoch": 0.52, "learning_rate": 7.379259878737195e-06, "loss": 0.8615, "step": 2510 }, { "epoch": 0.53, "learning_rate": 7.368806188584571e-06, "loss": 0.9127, "step": 2520 }, { "epoch": 0.53, "learning_rate": 7.358352498431946e-06, "loss": 0.9056, "step": 2530 }, { "epoch": 0.53, "learning_rate": 7.347898808279324e-06, "loss": 0.8745, "step": 2540 }, { "epoch": 0.53, "learning_rate": 7.337445118126699e-06, "loss": 0.8897, "step": 2550 }, { "epoch": 0.54, "learning_rate": 7.326991427974075e-06, "loss": 0.9375, "step": 2560 }, { "epoch": 0.54, "learning_rate": 7.316537737821452e-06, "loss": 0.8699, "step": 2570 }, { "epoch": 0.54, "learning_rate": 7.306084047668828e-06, "loss": 0.8904, "step": 2580 }, { "epoch": 0.54, "learning_rate": 7.295630357516204e-06, "loss": 0.8984, "step": 2590 }, { "epoch": 0.54, "learning_rate": 7.285176667363581e-06, "loss": 0.8784, "step": 2600 }, { "epoch": 0.55, "learning_rate": 7.274722977210956e-06, "loss": 0.8727, "step": 2610 }, { "epoch": 0.55, "learning_rate": 7.264269287058332e-06, "loss": 0.9417, "step": 2620 }, { "epoch": 0.55, "learning_rate": 7.253815596905708e-06, "loss": 0.865, "step": 2630 }, { "epoch": 0.55, "learning_rate": 7.243361906753085e-06, "loss": 0.9408, "step": 2640 }, { "epoch": 0.55, "learning_rate": 7.232908216600461e-06, "loss": 0.8789, "step": 2650 }, { "epoch": 0.56, "learning_rate": 7.222454526447836e-06, "loss": 0.9247, "step": 2660 }, { "epoch": 0.56, "learning_rate": 7.212000836295213e-06, "loss": 0.8336, "step": 2670 }, { "epoch": 0.56, "learning_rate": 7.201547146142589e-06, "loss": 0.9389, "step": 2680 }, { "epoch": 0.56, "learning_rate": 7.191093455989965e-06, "loss": 0.8932, "step": 2690 }, { "epoch": 0.56, "learning_rate": 7.180639765837342e-06, "loss": 0.8682, "step": 2700 }, { "epoch": 0.57, "learning_rate": 7.170186075684718e-06, "loss": 0.8646, "step": 2710 }, { "epoch": 0.57, "learning_rate": 7.159732385532093e-06, "loss": 0.8836, "step": 2720 }, { "epoch": 0.57, "learning_rate": 7.149278695379469e-06, "loss": 0.8928, "step": 2730 }, { "epoch": 0.57, "learning_rate": 7.138825005226846e-06, "loss": 0.8607, "step": 2740 }, { "epoch": 0.57, "learning_rate": 7.128371315074222e-06, "loss": 0.8347, "step": 2750 }, { "epoch": 0.58, "learning_rate": 7.117917624921598e-06, "loss": 0.9103, "step": 2760 }, { "epoch": 0.58, "learning_rate": 7.107463934768975e-06, "loss": 0.8894, "step": 2770 }, { "epoch": 0.58, "learning_rate": 7.097010244616351e-06, "loss": 0.8801, "step": 2780 }, { "epoch": 0.58, "learning_rate": 7.086556554463726e-06, "loss": 0.869, "step": 2790 }, { "epoch": 0.59, "learning_rate": 7.076102864311103e-06, "loss": 0.9117, "step": 2800 }, { "epoch": 0.59, "learning_rate": 7.065649174158479e-06, "loss": 0.9176, "step": 2810 }, { "epoch": 0.59, "learning_rate": 7.055195484005855e-06, "loss": 0.9837, "step": 2820 }, { "epoch": 0.59, "learning_rate": 7.044741793853231e-06, "loss": 0.8136, "step": 2830 }, { "epoch": 0.59, "learning_rate": 7.0342881037006075e-06, "loss": 0.9226, "step": 2840 }, { "epoch": 0.6, "learning_rate": 7.023834413547983e-06, "loss": 0.9151, "step": 2850 }, { "epoch": 0.6, "learning_rate": 7.013380723395359e-06, "loss": 0.9269, "step": 2860 }, { "epoch": 0.6, "learning_rate": 7.0029270332427355e-06, "loss": 0.8706, "step": 2870 }, { "epoch": 0.6, "learning_rate": 6.9924733430901115e-06, "loss": 0.9008, "step": 2880 }, { "epoch": 0.6, "learning_rate": 6.9820196529374875e-06, "loss": 0.9568, "step": 2890 }, { "epoch": 0.61, "learning_rate": 6.971565962784863e-06, "loss": 0.912, "step": 2900 }, { "epoch": 0.61, "learning_rate": 6.9611122726322396e-06, "loss": 0.9409, "step": 2910 }, { "epoch": 0.61, "learning_rate": 6.9506585824796156e-06, "loss": 0.9169, "step": 2920 }, { "epoch": 0.61, "learning_rate": 6.940204892326992e-06, "loss": 0.8665, "step": 2930 }, { "epoch": 0.61, "learning_rate": 6.9297512021743684e-06, "loss": 0.8812, "step": 2940 }, { "epoch": 0.62, "learning_rate": 6.9192975120217444e-06, "loss": 0.8699, "step": 2950 }, { "epoch": 0.62, "learning_rate": 6.9088438218691205e-06, "loss": 0.8409, "step": 2960 }, { "epoch": 0.62, "learning_rate": 6.898390131716497e-06, "loss": 0.9777, "step": 2970 }, { "epoch": 0.62, "learning_rate": 6.8879364415638725e-06, "loss": 0.8964, "step": 2980 }, { "epoch": 0.62, "learning_rate": 6.8774827514112485e-06, "loss": 0.9382, "step": 2990 }, { "epoch": 0.63, "learning_rate": 6.8670290612586245e-06, "loss": 0.8833, "step": 3000 }, { "epoch": 0.63, "learning_rate": 6.856575371106001e-06, "loss": 0.8316, "step": 3010 }, { "epoch": 0.63, "learning_rate": 6.846121680953377e-06, "loss": 0.9556, "step": 3020 }, { "epoch": 0.63, "learning_rate": 6.8356679908007525e-06, "loss": 0.9029, "step": 3030 }, { "epoch": 0.64, "learning_rate": 6.825214300648129e-06, "loss": 0.9594, "step": 3040 }, { "epoch": 0.64, "learning_rate": 6.814760610495505e-06, "loss": 0.9346, "step": 3050 }, { "epoch": 0.64, "learning_rate": 6.804306920342881e-06, "loss": 0.8544, "step": 3060 }, { "epoch": 0.64, "learning_rate": 6.793853230190258e-06, "loss": 0.9281, "step": 3070 }, { "epoch": 0.64, "learning_rate": 6.783399540037634e-06, "loss": 0.8431, "step": 3080 }, { "epoch": 0.65, "learning_rate": 6.7729458498850094e-06, "loss": 0.8402, "step": 3090 }, { "epoch": 0.65, "learning_rate": 6.7624921597323854e-06, "loss": 0.8558, "step": 3100 }, { "epoch": 0.65, "learning_rate": 6.752038469579762e-06, "loss": 0.8985, "step": 3110 }, { "epoch": 0.65, "learning_rate": 6.741584779427138e-06, "loss": 0.8913, "step": 3120 }, { "epoch": 0.65, "learning_rate": 6.731131089274514e-06, "loss": 0.9463, "step": 3130 }, { "epoch": 0.66, "learning_rate": 6.720677399121891e-06, "loss": 0.9098, "step": 3140 }, { "epoch": 0.66, "learning_rate": 6.710223708969267e-06, "loss": 0.8892, "step": 3150 }, { "epoch": 0.66, "learning_rate": 6.699770018816642e-06, "loss": 0.9503, "step": 3160 }, { "epoch": 0.66, "learning_rate": 6.689316328664019e-06, "loss": 0.9012, "step": 3170 }, { "epoch": 0.66, "learning_rate": 6.678862638511395e-06, "loss": 0.922, "step": 3180 }, { "epoch": 0.67, "learning_rate": 6.668408948358771e-06, "loss": 0.8312, "step": 3190 }, { "epoch": 0.67, "learning_rate": 6.657955258206147e-06, "loss": 0.8729, "step": 3200 }, { "epoch": 0.67, "learning_rate": 6.647501568053524e-06, "loss": 0.9533, "step": 3210 }, { "epoch": 0.67, "learning_rate": 6.637047877900899e-06, "loss": 0.9126, "step": 3220 }, { "epoch": 0.68, "learning_rate": 6.626594187748275e-06, "loss": 0.9615, "step": 3230 }, { "epoch": 0.68, "learning_rate": 6.616140497595652e-06, "loss": 0.888, "step": 3240 }, { "epoch": 0.68, "learning_rate": 6.605686807443028e-06, "loss": 0.9598, "step": 3250 }, { "epoch": 0.68, "learning_rate": 6.595233117290404e-06, "loss": 0.9514, "step": 3260 }, { "epoch": 0.68, "learning_rate": 6.584779427137779e-06, "loss": 0.8736, "step": 3270 }, { "epoch": 0.69, "learning_rate": 6.574325736985156e-06, "loss": 0.8775, "step": 3280 }, { "epoch": 0.69, "learning_rate": 6.563872046832532e-06, "loss": 0.9449, "step": 3290 }, { "epoch": 0.69, "learning_rate": 6.553418356679908e-06, "loss": 0.8699, "step": 3300 }, { "epoch": 0.69, "learning_rate": 6.542964666527285e-06, "loss": 0.8853, "step": 3310 }, { "epoch": 0.69, "learning_rate": 6.532510976374661e-06, "loss": 0.9349, "step": 3320 }, { "epoch": 0.7, "learning_rate": 6.522057286222036e-06, "loss": 0.9356, "step": 3330 }, { "epoch": 0.7, "learning_rate": 6.511603596069414e-06, "loss": 0.9341, "step": 3340 }, { "epoch": 0.7, "learning_rate": 6.501149905916789e-06, "loss": 0.8925, "step": 3350 }, { "epoch": 0.7, "learning_rate": 6.490696215764165e-06, "loss": 0.8894, "step": 3360 }, { "epoch": 0.7, "learning_rate": 6.480242525611541e-06, "loss": 0.8947, "step": 3370 }, { "epoch": 0.71, "learning_rate": 6.469788835458918e-06, "loss": 0.8991, "step": 3380 }, { "epoch": 0.71, "learning_rate": 6.459335145306294e-06, "loss": 0.9011, "step": 3390 }, { "epoch": 0.71, "learning_rate": 6.448881455153669e-06, "loss": 0.8659, "step": 3400 }, { "epoch": 0.71, "learning_rate": 6.438427765001046e-06, "loss": 0.9192, "step": 3410 }, { "epoch": 0.71, "learning_rate": 6.427974074848422e-06, "loss": 0.9916, "step": 3420 }, { "epoch": 0.72, "learning_rate": 6.417520384695798e-06, "loss": 0.9284, "step": 3430 }, { "epoch": 0.72, "learning_rate": 6.407066694543175e-06, "loss": 0.8701, "step": 3440 }, { "epoch": 0.72, "learning_rate": 6.396613004390551e-06, "loss": 0.9108, "step": 3450 }, { "epoch": 0.72, "learning_rate": 6.386159314237926e-06, "loss": 0.9475, "step": 3460 }, { "epoch": 0.73, "learning_rate": 6.375705624085302e-06, "loss": 0.8547, "step": 3470 }, { "epoch": 0.73, "learning_rate": 6.365251933932679e-06, "loss": 0.9119, "step": 3480 }, { "epoch": 0.73, "learning_rate": 6.354798243780055e-06, "loss": 0.861, "step": 3490 }, { "epoch": 0.73, "learning_rate": 6.344344553627431e-06, "loss": 0.8882, "step": 3500 }, { "epoch": 0.73, "learning_rate": 6.333890863474808e-06, "loss": 0.8947, "step": 3510 }, { "epoch": 0.74, "learning_rate": 6.323437173322183e-06, "loss": 0.9661, "step": 3520 }, { "epoch": 0.74, "learning_rate": 6.312983483169559e-06, "loss": 0.914, "step": 3530 }, { "epoch": 0.74, "learning_rate": 6.302529793016936e-06, "loss": 0.8673, "step": 3540 }, { "epoch": 0.74, "learning_rate": 6.292076102864312e-06, "loss": 0.9488, "step": 3550 }, { "epoch": 0.74, "learning_rate": 6.281622412711688e-06, "loss": 0.9209, "step": 3560 }, { "epoch": 0.75, "learning_rate": 6.271168722559064e-06, "loss": 0.848, "step": 3570 }, { "epoch": 0.75, "learning_rate": 6.260715032406441e-06, "loss": 0.8185, "step": 3580 }, { "epoch": 0.75, "learning_rate": 6.250261342253816e-06, "loss": 0.9759, "step": 3590 }, { "epoch": 0.75, "learning_rate": 6.239807652101192e-06, "loss": 0.8938, "step": 3600 }, { "epoch": 0.75, "learning_rate": 6.229353961948569e-06, "loss": 0.9568, "step": 3610 }, { "epoch": 0.76, "learning_rate": 6.218900271795945e-06, "loss": 0.877, "step": 3620 }, { "epoch": 0.76, "learning_rate": 6.208446581643321e-06, "loss": 0.9064, "step": 3630 }, { "epoch": 0.76, "learning_rate": 6.1979928914906976e-06, "loss": 0.8697, "step": 3640 }, { "epoch": 0.76, "learning_rate": 6.187539201338073e-06, "loss": 0.9132, "step": 3650 }, { "epoch": 0.77, "learning_rate": 6.177085511185449e-06, "loss": 0.8063, "step": 3660 }, { "epoch": 0.77, "learning_rate": 6.166631821032825e-06, "loss": 0.9361, "step": 3670 }, { "epoch": 0.77, "learning_rate": 6.156178130880202e-06, "loss": 0.7849, "step": 3680 }, { "epoch": 0.77, "learning_rate": 6.145724440727578e-06, "loss": 0.9113, "step": 3690 }, { "epoch": 0.77, "learning_rate": 6.135270750574953e-06, "loss": 0.9431, "step": 3700 }, { "epoch": 0.78, "learning_rate": 6.12481706042233e-06, "loss": 0.8772, "step": 3710 }, { "epoch": 0.78, "learning_rate": 6.114363370269706e-06, "loss": 0.8491, "step": 3720 }, { "epoch": 0.78, "learning_rate": 6.103909680117082e-06, "loss": 0.9514, "step": 3730 }, { "epoch": 0.78, "learning_rate": 6.093455989964458e-06, "loss": 0.9193, "step": 3740 }, { "epoch": 0.78, "learning_rate": 6.0830022998118345e-06, "loss": 0.8388, "step": 3750 }, { "epoch": 0.79, "learning_rate": 6.0725486096592105e-06, "loss": 0.8309, "step": 3760 }, { "epoch": 0.79, "learning_rate": 6.062094919506586e-06, "loss": 0.8447, "step": 3770 }, { "epoch": 0.79, "learning_rate": 6.0516412293539625e-06, "loss": 0.8534, "step": 3780 }, { "epoch": 0.79, "learning_rate": 6.0411875392013385e-06, "loss": 0.8391, "step": 3790 }, { "epoch": 0.79, "learning_rate": 6.0307338490487145e-06, "loss": 0.8798, "step": 3800 }, { "epoch": 0.8, "learning_rate": 6.020280158896091e-06, "loss": 0.9287, "step": 3810 }, { "epoch": 0.8, "learning_rate": 6.009826468743467e-06, "loss": 0.8622, "step": 3820 }, { "epoch": 0.8, "learning_rate": 5.999372778590843e-06, "loss": 0.9576, "step": 3830 }, { "epoch": 0.8, "learning_rate": 5.988919088438219e-06, "loss": 0.8898, "step": 3840 }, { "epoch": 0.8, "learning_rate": 5.9784653982855954e-06, "loss": 0.918, "step": 3850 }, { "epoch": 0.81, "learning_rate": 5.9680117081329715e-06, "loss": 0.9338, "step": 3860 }, { "epoch": 0.81, "learning_rate": 5.9575580179803475e-06, "loss": 0.8506, "step": 3870 }, { "epoch": 0.81, "learning_rate": 5.947104327827724e-06, "loss": 0.8822, "step": 3880 }, { "epoch": 0.81, "learning_rate": 5.9366506376750995e-06, "loss": 0.8945, "step": 3890 }, { "epoch": 0.82, "learning_rate": 5.9261969475224755e-06, "loss": 0.8592, "step": 3900 }, { "epoch": 0.82, "learning_rate": 5.915743257369852e-06, "loss": 0.9177, "step": 3910 }, { "epoch": 0.82, "learning_rate": 5.905289567217228e-06, "loss": 0.9195, "step": 3920 }, { "epoch": 0.82, "learning_rate": 5.894835877064604e-06, "loss": 0.8741, "step": 3930 }, { "epoch": 0.82, "learning_rate": 5.8843821869119795e-06, "loss": 0.8836, "step": 3940 }, { "epoch": 0.83, "learning_rate": 5.873928496759357e-06, "loss": 0.8813, "step": 3950 }, { "epoch": 0.83, "learning_rate": 5.863474806606732e-06, "loss": 0.8993, "step": 3960 }, { "epoch": 0.83, "learning_rate": 5.853021116454108e-06, "loss": 0.9148, "step": 3970 }, { "epoch": 0.83, "learning_rate": 5.842567426301485e-06, "loss": 0.8566, "step": 3980 }, { "epoch": 0.83, "learning_rate": 5.832113736148861e-06, "loss": 0.9285, "step": 3990 }, { "epoch": 0.84, "learning_rate": 5.821660045996237e-06, "loss": 0.8902, "step": 4000 }, { "epoch": 0.84, "learning_rate": 5.811206355843614e-06, "loss": 0.8892, "step": 4010 }, { "epoch": 0.84, "learning_rate": 5.800752665690989e-06, "loss": 0.9198, "step": 4020 }, { "epoch": 0.84, "learning_rate": 5.790298975538365e-06, "loss": 0.8891, "step": 4030 }, { "epoch": 0.84, "learning_rate": 5.779845285385741e-06, "loss": 0.853, "step": 4040 }, { "epoch": 0.85, "learning_rate": 5.769391595233118e-06, "loss": 0.9333, "step": 4050 }, { "epoch": 0.85, "learning_rate": 5.758937905080494e-06, "loss": 0.9174, "step": 4060 }, { "epoch": 0.85, "learning_rate": 5.748484214927869e-06, "loss": 0.8924, "step": 4070 }, { "epoch": 0.85, "learning_rate": 5.738030524775246e-06, "loss": 0.8677, "step": 4080 }, { "epoch": 0.85, "learning_rate": 5.727576834622622e-06, "loss": 0.8614, "step": 4090 }, { "epoch": 0.86, "learning_rate": 5.717123144469998e-06, "loss": 0.9326, "step": 4100 }, { "epoch": 0.86, "learning_rate": 5.706669454317374e-06, "loss": 0.8554, "step": 4110 }, { "epoch": 0.86, "learning_rate": 5.696215764164751e-06, "loss": 0.8976, "step": 4120 }, { "epoch": 0.86, "learning_rate": 5.685762074012126e-06, "loss": 0.8818, "step": 4130 }, { "epoch": 0.87, "learning_rate": 5.675308383859502e-06, "loss": 0.9545, "step": 4140 }, { "epoch": 0.87, "learning_rate": 5.664854693706879e-06, "loss": 0.8384, "step": 4150 }, { "epoch": 0.87, "learning_rate": 5.654401003554255e-06, "loss": 0.8158, "step": 4160 }, { "epoch": 0.87, "learning_rate": 5.643947313401631e-06, "loss": 0.8621, "step": 4170 }, { "epoch": 0.87, "learning_rate": 5.633493623249008e-06, "loss": 0.8706, "step": 4180 }, { "epoch": 0.88, "learning_rate": 5.623039933096384e-06, "loss": 0.8428, "step": 4190 }, { "epoch": 0.88, "learning_rate": 5.612586242943759e-06, "loss": 0.9087, "step": 4200 }, { "epoch": 0.88, "learning_rate": 5.602132552791135e-06, "loss": 0.8231, "step": 4210 }, { "epoch": 0.88, "learning_rate": 5.591678862638512e-06, "loss": 0.8731, "step": 4220 }, { "epoch": 0.88, "learning_rate": 5.581225172485888e-06, "loss": 0.9065, "step": 4230 }, { "epoch": 0.89, "learning_rate": 5.570771482333264e-06, "loss": 0.9498, "step": 4240 }, { "epoch": 0.89, "learning_rate": 5.560317792180641e-06, "loss": 0.904, "step": 4250 }, { "epoch": 0.89, "learning_rate": 5.549864102028016e-06, "loss": 0.8868, "step": 4260 }, { "epoch": 0.89, "learning_rate": 5.539410411875392e-06, "loss": 0.8935, "step": 4270 }, { "epoch": 0.89, "learning_rate": 5.528956721722769e-06, "loss": 0.9195, "step": 4280 }, { "epoch": 0.9, "learning_rate": 5.518503031570145e-06, "loss": 0.8788, "step": 4290 }, { "epoch": 0.9, "learning_rate": 5.508049341417521e-06, "loss": 0.8446, "step": 4300 }, { "epoch": 0.9, "learning_rate": 5.497595651264896e-06, "loss": 0.8088, "step": 4310 }, { "epoch": 0.9, "learning_rate": 5.487141961112273e-06, "loss": 0.8634, "step": 4320 }, { "epoch": 0.91, "learning_rate": 5.476688270959649e-06, "loss": 0.9249, "step": 4330 }, { "epoch": 0.91, "learning_rate": 5.466234580807025e-06, "loss": 0.8693, "step": 4340 }, { "epoch": 0.91, "learning_rate": 5.455780890654402e-06, "loss": 0.9334, "step": 4350 }, { "epoch": 0.91, "learning_rate": 5.445327200501778e-06, "loss": 0.869, "step": 4360 }, { "epoch": 0.91, "learning_rate": 5.434873510349154e-06, "loss": 0.9951, "step": 4370 }, { "epoch": 0.92, "learning_rate": 5.424419820196531e-06, "loss": 0.8824, "step": 4380 }, { "epoch": 0.92, "learning_rate": 5.413966130043906e-06, "loss": 0.9051, "step": 4390 }, { "epoch": 0.92, "learning_rate": 5.403512439891282e-06, "loss": 0.9867, "step": 4400 }, { "epoch": 0.92, "learning_rate": 5.393058749738658e-06, "loss": 0.9016, "step": 4410 }, { "epoch": 0.92, "learning_rate": 5.382605059586035e-06, "loss": 0.8548, "step": 4420 }, { "epoch": 0.93, "learning_rate": 5.372151369433411e-06, "loss": 0.8845, "step": 4430 }, { "epoch": 0.93, "learning_rate": 5.361697679280786e-06, "loss": 0.8756, "step": 4440 }, { "epoch": 0.93, "learning_rate": 5.351243989128163e-06, "loss": 0.8726, "step": 4450 }, { "epoch": 0.93, "learning_rate": 5.340790298975539e-06, "loss": 0.8493, "step": 4460 }, { "epoch": 0.93, "learning_rate": 5.330336608822915e-06, "loss": 0.7948, "step": 4470 }, { "epoch": 0.94, "learning_rate": 5.319882918670291e-06, "loss": 0.8606, "step": 4480 }, { "epoch": 0.94, "learning_rate": 5.309429228517668e-06, "loss": 0.8924, "step": 4490 }, { "epoch": 0.94, "learning_rate": 5.298975538365043e-06, "loss": 0.9294, "step": 4500 }, { "epoch": 0.94, "learning_rate": 5.288521848212419e-06, "loss": 0.9276, "step": 4510 }, { "epoch": 0.94, "learning_rate": 5.278068158059796e-06, "loss": 0.8315, "step": 4520 }, { "epoch": 0.95, "learning_rate": 5.267614467907172e-06, "loss": 0.8768, "step": 4530 }, { "epoch": 0.95, "learning_rate": 5.257160777754548e-06, "loss": 0.8714, "step": 4540 }, { "epoch": 0.95, "learning_rate": 5.2467070876019246e-06, "loss": 0.9204, "step": 4550 }, { "epoch": 0.95, "learning_rate": 5.2362533974493006e-06, "loss": 0.8387, "step": 4560 }, { "epoch": 0.96, "learning_rate": 5.225799707296676e-06, "loss": 0.908, "step": 4570 }, { "epoch": 0.96, "learning_rate": 5.215346017144052e-06, "loss": 0.9011, "step": 4580 }, { "epoch": 0.96, "learning_rate": 5.204892326991429e-06, "loss": 0.891, "step": 4590 }, { "epoch": 0.96, "learning_rate": 5.194438636838805e-06, "loss": 0.8301, "step": 4600 }, { "epoch": 0.96, "learning_rate": 5.183984946686181e-06, "loss": 0.9309, "step": 4610 }, { "epoch": 0.97, "learning_rate": 5.1735312565335575e-06, "loss": 0.9124, "step": 4620 }, { "epoch": 0.97, "learning_rate": 5.163077566380933e-06, "loss": 0.7651, "step": 4630 }, { "epoch": 0.97, "learning_rate": 5.152623876228309e-06, "loss": 0.8533, "step": 4640 }, { "epoch": 0.97, "learning_rate": 5.1421701860756855e-06, "loss": 0.9362, "step": 4650 }, { "epoch": 0.97, "learning_rate": 5.1317164959230615e-06, "loss": 0.9113, "step": 4660 }, { "epoch": 0.98, "learning_rate": 5.1212628057704375e-06, "loss": 0.8711, "step": 4670 }, { "epoch": 0.98, "learning_rate": 5.110809115617813e-06, "loss": 0.8862, "step": 4680 }, { "epoch": 0.98, "learning_rate": 5.1003554254651895e-06, "loss": 0.8542, "step": 4690 }, { "epoch": 0.98, "learning_rate": 5.0899017353125655e-06, "loss": 0.8945, "step": 4700 }, { "epoch": 0.98, "learning_rate": 5.0794480451599416e-06, "loss": 0.8488, "step": 4710 }, { "epoch": 0.99, "learning_rate": 5.068994355007318e-06, "loss": 0.9322, "step": 4720 }, { "epoch": 0.99, "learning_rate": 5.058540664854694e-06, "loss": 0.8453, "step": 4730 }, { "epoch": 0.99, "learning_rate": 5.04808697470207e-06, "loss": 0.8598, "step": 4740 }, { "epoch": 0.99, "learning_rate": 5.037633284549447e-06, "loss": 0.9067, "step": 4750 }, { "epoch": 0.99, "learning_rate": 5.0271795943968224e-06, "loss": 0.879, "step": 4760 }, { "epoch": 1.0, "learning_rate": 5.0167259042441985e-06, "loss": 0.8492, "step": 4770 }, { "epoch": 1.0, "learning_rate": 5.0062722140915745e-06, "loss": 0.9149, "step": 4780 }, { "epoch": 1.0, "learning_rate": 4.9958185239389505e-06, "loss": 0.774, "step": 4790 }, { "epoch": 1.0, "learning_rate": 4.985364833786327e-06, "loss": 0.6481, "step": 4800 }, { "epoch": 1.01, "learning_rate": 4.974911143633703e-06, "loss": 0.6638, "step": 4810 }, { "epoch": 1.01, "learning_rate": 4.964457453481079e-06, "loss": 0.6349, "step": 4820 }, { "epoch": 1.01, "learning_rate": 4.954003763328455e-06, "loss": 0.6898, "step": 4830 }, { "epoch": 1.01, "learning_rate": 4.943550073175831e-06, "loss": 0.6785, "step": 4840 }, { "epoch": 1.01, "learning_rate": 4.933096383023207e-06, "loss": 0.6888, "step": 4850 }, { "epoch": 1.02, "learning_rate": 4.922642692870584e-06, "loss": 0.706, "step": 4860 }, { "epoch": 1.02, "learning_rate": 4.912189002717959e-06, "loss": 0.6787, "step": 4870 }, { "epoch": 1.02, "learning_rate": 4.901735312565336e-06, "loss": 0.6352, "step": 4880 }, { "epoch": 1.02, "learning_rate": 4.891281622412712e-06, "loss": 0.7415, "step": 4890 }, { "epoch": 1.02, "learning_rate": 4.880827932260088e-06, "loss": 0.6576, "step": 4900 }, { "epoch": 1.03, "learning_rate": 4.870374242107464e-06, "loss": 0.7198, "step": 4910 }, { "epoch": 1.03, "learning_rate": 4.85992055195484e-06, "loss": 0.6811, "step": 4920 }, { "epoch": 1.03, "learning_rate": 4.849466861802216e-06, "loss": 0.6682, "step": 4930 }, { "epoch": 1.03, "learning_rate": 4.839013171649592e-06, "loss": 0.69, "step": 4940 }, { "epoch": 1.03, "learning_rate": 4.828559481496969e-06, "loss": 0.6408, "step": 4950 }, { "epoch": 1.04, "learning_rate": 4.818105791344345e-06, "loss": 0.6992, "step": 4960 }, { "epoch": 1.04, "learning_rate": 4.807652101191721e-06, "loss": 0.675, "step": 4970 }, { "epoch": 1.04, "learning_rate": 4.797198411039097e-06, "loss": 0.5974, "step": 4980 }, { "epoch": 1.04, "learning_rate": 4.786744720886473e-06, "loss": 0.7038, "step": 4990 }, { "epoch": 1.05, "learning_rate": 4.776291030733849e-06, "loss": 0.7172, "step": 5000 } ], "max_steps": 9568, "num_train_epochs": 2, "total_flos": 691733423194112.0, "trial_name": null, "trial_params": null }