{ "best_metric": null, "best_model_checkpoint": null, "epoch": 0.999974259311694, "eval_steps": 500, "global_step": 9712, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 4.5e-07, "loss": 2.976, "step": 10 }, { "epoch": 0.0, "learning_rate": 9.5e-07, "loss": 2.6462, "step": 20 }, { "epoch": 0.0, "learning_rate": 1.45e-06, "loss": 2.2432, "step": 30 }, { "epoch": 0.0, "learning_rate": 1.95e-06, "loss": 2.3321, "step": 40 }, { "epoch": 0.01, "learning_rate": 2.4500000000000003e-06, "loss": 2.2145, "step": 50 }, { "epoch": 0.01, "learning_rate": 2.95e-06, "loss": 2.408, "step": 60 }, { "epoch": 0.01, "learning_rate": 3.4500000000000004e-06, "loss": 2.3696, "step": 70 }, { "epoch": 0.01, "learning_rate": 3.95e-06, "loss": 2.3913, "step": 80 }, { "epoch": 0.01, "learning_rate": 4.45e-06, "loss": 2.1274, "step": 90 }, { "epoch": 0.01, "learning_rate": 4.950000000000001e-06, "loss": 2.3101, "step": 100 }, { "epoch": 0.01, "learning_rate": 5.45e-06, "loss": 2.3138, "step": 110 }, { "epoch": 0.01, "learning_rate": 5.95e-06, "loss": 2.1893, "step": 120 }, { "epoch": 0.01, "learning_rate": 6.45e-06, "loss": 2.1304, "step": 130 }, { "epoch": 0.01, "learning_rate": 6.950000000000001e-06, "loss": 2.2672, "step": 140 }, { "epoch": 0.02, "learning_rate": 7.45e-06, "loss": 2.115, "step": 150 }, { "epoch": 0.02, "learning_rate": 7.95e-06, "loss": 2.2899, "step": 160 }, { "epoch": 0.02, "learning_rate": 8.45e-06, "loss": 2.3333, "step": 170 }, { "epoch": 0.02, "learning_rate": 8.95e-06, "loss": 2.2231, "step": 180 }, { "epoch": 0.02, "learning_rate": 9.450000000000001e-06, "loss": 2.2459, "step": 190 }, { "epoch": 0.02, "learning_rate": 9.950000000000001e-06, "loss": 2.1321, "step": 200 }, { "epoch": 0.02, "learning_rate": 1.045e-05, "loss": 2.2164, "step": 210 }, { "epoch": 0.02, "learning_rate": 1.095e-05, "loss": 2.0555, "step": 220 }, { "epoch": 0.02, "learning_rate": 1.145e-05, "loss": 2.1948, "step": 230 }, { "epoch": 0.02, "learning_rate": 1.195e-05, "loss": 2.1423, "step": 240 }, { "epoch": 0.03, "learning_rate": 1.2450000000000001e-05, "loss": 2.0471, "step": 250 }, { "epoch": 0.03, "learning_rate": 1.2950000000000001e-05, "loss": 2.2886, "step": 260 }, { "epoch": 0.03, "learning_rate": 1.3450000000000002e-05, "loss": 2.3572, "step": 270 }, { "epoch": 0.03, "learning_rate": 1.3950000000000002e-05, "loss": 2.1849, "step": 280 }, { "epoch": 0.03, "learning_rate": 1.4449999999999999e-05, "loss": 2.0637, "step": 290 }, { "epoch": 0.03, "learning_rate": 1.4950000000000001e-05, "loss": 1.9557, "step": 300 }, { "epoch": 0.03, "learning_rate": 1.545e-05, "loss": 2.1387, "step": 310 }, { "epoch": 0.03, "learning_rate": 1.595e-05, "loss": 2.2507, "step": 320 }, { "epoch": 0.03, "learning_rate": 1.645e-05, "loss": 2.1553, "step": 330 }, { "epoch": 0.04, "learning_rate": 1.6950000000000002e-05, "loss": 2.1832, "step": 340 }, { "epoch": 0.04, "learning_rate": 1.745e-05, "loss": 2.0478, "step": 350 }, { "epoch": 0.04, "learning_rate": 1.795e-05, "loss": 1.9862, "step": 360 }, { "epoch": 0.04, "learning_rate": 1.845e-05, "loss": 2.3255, "step": 370 }, { "epoch": 0.04, "learning_rate": 1.895e-05, "loss": 2.2428, "step": 380 }, { "epoch": 0.04, "learning_rate": 1.9450000000000002e-05, "loss": 2.1448, "step": 390 }, { "epoch": 0.04, "learning_rate": 1.995e-05, "loss": 2.2396, "step": 400 }, { "epoch": 0.04, "learning_rate": 2.045e-05, "loss": 2.0933, "step": 410 }, { "epoch": 0.04, "learning_rate": 2.095e-05, "loss": 2.0978, "step": 420 }, { "epoch": 0.04, "learning_rate": 2.145e-05, "loss": 2.0961, "step": 430 }, { "epoch": 0.05, "learning_rate": 2.195e-05, "loss": 2.0171, "step": 440 }, { "epoch": 0.05, "learning_rate": 2.245e-05, "loss": 1.8757, "step": 450 }, { "epoch": 0.05, "learning_rate": 2.2950000000000002e-05, "loss": 2.2018, "step": 460 }, { "epoch": 0.05, "learning_rate": 2.345e-05, "loss": 2.0696, "step": 470 }, { "epoch": 0.05, "learning_rate": 2.395e-05, "loss": 2.2329, "step": 480 }, { "epoch": 0.05, "learning_rate": 2.445e-05, "loss": 2.2144, "step": 490 }, { "epoch": 0.05, "learning_rate": 2.495e-05, "loss": 1.9678, "step": 500 }, { "epoch": 0.05, "learning_rate": 2.5450000000000002e-05, "loss": 2.0717, "step": 510 }, { "epoch": 0.05, "learning_rate": 2.595e-05, "loss": 2.0088, "step": 520 }, { "epoch": 0.05, "learning_rate": 2.6450000000000003e-05, "loss": 2.1696, "step": 530 }, { "epoch": 0.06, "learning_rate": 2.6950000000000005e-05, "loss": 1.9404, "step": 540 }, { "epoch": 0.06, "learning_rate": 2.7450000000000003e-05, "loss": 2.0756, "step": 550 }, { "epoch": 0.06, "learning_rate": 2.7950000000000005e-05, "loss": 2.0034, "step": 560 }, { "epoch": 0.06, "learning_rate": 2.845e-05, "loss": 2.0446, "step": 570 }, { "epoch": 0.06, "learning_rate": 2.895e-05, "loss": 2.054, "step": 580 }, { "epoch": 0.06, "learning_rate": 2.945e-05, "loss": 2.086, "step": 590 }, { "epoch": 0.06, "learning_rate": 2.995e-05, "loss": 2.1475, "step": 600 }, { "epoch": 0.06, "learning_rate": 3.045e-05, "loss": 2.2484, "step": 610 }, { "epoch": 0.06, "learning_rate": 3.095e-05, "loss": 2.0034, "step": 620 }, { "epoch": 0.06, "learning_rate": 3.145e-05, "loss": 2.0188, "step": 630 }, { "epoch": 0.07, "learning_rate": 3.1950000000000004e-05, "loss": 1.9873, "step": 640 }, { "epoch": 0.07, "learning_rate": 3.245e-05, "loss": 2.1709, "step": 650 }, { "epoch": 0.07, "learning_rate": 3.295e-05, "loss": 2.1436, "step": 660 }, { "epoch": 0.07, "learning_rate": 3.345000000000001e-05, "loss": 2.0891, "step": 670 }, { "epoch": 0.07, "learning_rate": 3.3950000000000005e-05, "loss": 2.1962, "step": 680 }, { "epoch": 0.07, "learning_rate": 3.445e-05, "loss": 1.9627, "step": 690 }, { "epoch": 0.07, "learning_rate": 3.495e-05, "loss": 1.9947, "step": 700 }, { "epoch": 0.07, "learning_rate": 3.545e-05, "loss": 2.1952, "step": 710 }, { "epoch": 0.07, "learning_rate": 3.595e-05, "loss": 2.0161, "step": 720 }, { "epoch": 0.08, "learning_rate": 3.645e-05, "loss": 2.134, "step": 730 }, { "epoch": 0.08, "learning_rate": 3.6950000000000004e-05, "loss": 2.085, "step": 740 }, { "epoch": 0.08, "learning_rate": 3.745e-05, "loss": 2.2181, "step": 750 }, { "epoch": 0.08, "learning_rate": 3.795e-05, "loss": 2.0471, "step": 760 }, { "epoch": 0.08, "learning_rate": 3.845e-05, "loss": 2.0651, "step": 770 }, { "epoch": 0.08, "learning_rate": 3.8950000000000005e-05, "loss": 2.0187, "step": 780 }, { "epoch": 0.08, "learning_rate": 3.9450000000000003e-05, "loss": 2.0299, "step": 790 }, { "epoch": 0.08, "learning_rate": 3.995e-05, "loss": 2.1057, "step": 800 }, { "epoch": 0.08, "learning_rate": 4.045000000000001e-05, "loss": 1.995, "step": 810 }, { "epoch": 0.08, "learning_rate": 4.095e-05, "loss": 2.1075, "step": 820 }, { "epoch": 0.09, "learning_rate": 4.145e-05, "loss": 1.97, "step": 830 }, { "epoch": 0.09, "learning_rate": 4.195e-05, "loss": 2.0722, "step": 840 }, { "epoch": 0.09, "learning_rate": 4.245e-05, "loss": 2.1999, "step": 850 }, { "epoch": 0.09, "learning_rate": 4.295e-05, "loss": 1.9943, "step": 860 }, { "epoch": 0.09, "learning_rate": 4.345e-05, "loss": 2.0222, "step": 870 }, { "epoch": 0.09, "learning_rate": 4.3950000000000004e-05, "loss": 2.0142, "step": 880 }, { "epoch": 0.09, "learning_rate": 4.445e-05, "loss": 2.0544, "step": 890 }, { "epoch": 0.09, "learning_rate": 4.495e-05, "loss": 2.0566, "step": 900 }, { "epoch": 0.09, "learning_rate": 4.545000000000001e-05, "loss": 2.0461, "step": 910 }, { "epoch": 0.09, "learning_rate": 4.5950000000000006e-05, "loss": 2.0164, "step": 920 }, { "epoch": 0.1, "learning_rate": 4.6450000000000004e-05, "loss": 2.1995, "step": 930 }, { "epoch": 0.1, "learning_rate": 4.695e-05, "loss": 1.9895, "step": 940 }, { "epoch": 0.1, "learning_rate": 4.745e-05, "loss": 2.0993, "step": 950 }, { "epoch": 0.1, "learning_rate": 4.795e-05, "loss": 2.011, "step": 960 }, { "epoch": 0.1, "learning_rate": 4.845e-05, "loss": 2.193, "step": 970 }, { "epoch": 0.1, "learning_rate": 4.8950000000000004e-05, "loss": 2.0688, "step": 980 }, { "epoch": 0.1, "learning_rate": 4.945e-05, "loss": 2.0526, "step": 990 }, { "epoch": 0.1, "learning_rate": 4.995e-05, "loss": 2.023, "step": 1000 }, { "epoch": 0.1, "learning_rate": 4.999214275736835e-05, "loss": 2.0966, "step": 1010 }, { "epoch": 0.11, "learning_rate": 4.9983412487777624e-05, "loss": 2.1602, "step": 1020 }, { "epoch": 0.11, "learning_rate": 4.9974682218186904e-05, "loss": 2.1433, "step": 1030 }, { "epoch": 0.11, "learning_rate": 4.996595194859617e-05, "loss": 2.1597, "step": 1040 }, { "epoch": 0.11, "learning_rate": 4.995722167900545e-05, "loss": 2.1164, "step": 1050 }, { "epoch": 0.11, "learning_rate": 4.994849140941473e-05, "loss": 2.0517, "step": 1060 }, { "epoch": 0.11, "learning_rate": 4.9939761139824e-05, "loss": 1.9674, "step": 1070 }, { "epoch": 0.11, "learning_rate": 4.993103087023328e-05, "loss": 2.0646, "step": 1080 }, { "epoch": 0.11, "learning_rate": 4.992230060064255e-05, "loss": 2.0989, "step": 1090 }, { "epoch": 0.11, "learning_rate": 4.991357033105182e-05, "loss": 2.0109, "step": 1100 }, { "epoch": 0.11, "learning_rate": 4.9904840061461103e-05, "loss": 2.0324, "step": 1110 }, { "epoch": 0.12, "learning_rate": 4.989610979187038e-05, "loss": 2.0882, "step": 1120 }, { "epoch": 0.12, "learning_rate": 4.988737952227965e-05, "loss": 2.0099, "step": 1130 }, { "epoch": 0.12, "learning_rate": 4.987864925268892e-05, "loss": 2.0875, "step": 1140 }, { "epoch": 0.12, "learning_rate": 4.98699189830982e-05, "loss": 1.9643, "step": 1150 }, { "epoch": 0.12, "learning_rate": 4.986118871350747e-05, "loss": 2.2203, "step": 1160 }, { "epoch": 0.12, "learning_rate": 4.985245844391675e-05, "loss": 1.9917, "step": 1170 }, { "epoch": 0.12, "learning_rate": 4.984372817432603e-05, "loss": 1.971, "step": 1180 }, { "epoch": 0.12, "learning_rate": 4.98349979047353e-05, "loss": 2.0906, "step": 1190 }, { "epoch": 0.12, "learning_rate": 4.9826267635144576e-05, "loss": 2.0374, "step": 1200 }, { "epoch": 0.12, "learning_rate": 4.981753736555385e-05, "loss": 1.9202, "step": 1210 }, { "epoch": 0.13, "learning_rate": 4.980880709596313e-05, "loss": 1.9862, "step": 1220 }, { "epoch": 0.13, "learning_rate": 4.9800076826372396e-05, "loss": 2.0537, "step": 1230 }, { "epoch": 0.13, "learning_rate": 4.9791346556781676e-05, "loss": 1.958, "step": 1240 }, { "epoch": 0.13, "learning_rate": 4.9782616287190956e-05, "loss": 2.0852, "step": 1250 }, { "epoch": 0.13, "learning_rate": 4.977388601760022e-05, "loss": 2.0906, "step": 1260 }, { "epoch": 0.13, "learning_rate": 4.97651557480095e-05, "loss": 2.0058, "step": 1270 }, { "epoch": 0.13, "learning_rate": 4.9756425478418776e-05, "loss": 1.9265, "step": 1280 }, { "epoch": 0.13, "learning_rate": 4.974769520882805e-05, "loss": 1.9428, "step": 1290 }, { "epoch": 0.13, "learning_rate": 4.973896493923732e-05, "loss": 1.9286, "step": 1300 }, { "epoch": 0.13, "learning_rate": 4.97302346696466e-05, "loss": 2.0018, "step": 1310 }, { "epoch": 0.14, "learning_rate": 4.9721504400055876e-05, "loss": 1.9844, "step": 1320 }, { "epoch": 0.14, "learning_rate": 4.971277413046515e-05, "loss": 1.888, "step": 1330 }, { "epoch": 0.14, "learning_rate": 4.970404386087443e-05, "loss": 2.1434, "step": 1340 }, { "epoch": 0.14, "learning_rate": 4.9695313591283696e-05, "loss": 1.7986, "step": 1350 }, { "epoch": 0.14, "learning_rate": 4.9686583321692976e-05, "loss": 2.1599, "step": 1360 }, { "epoch": 0.14, "learning_rate": 4.967785305210225e-05, "loss": 2.1983, "step": 1370 }, { "epoch": 0.14, "learning_rate": 4.966912278251153e-05, "loss": 2.0426, "step": 1380 }, { "epoch": 0.14, "learning_rate": 4.96603925129208e-05, "loss": 2.0469, "step": 1390 }, { "epoch": 0.14, "learning_rate": 4.9651662243330076e-05, "loss": 2.1305, "step": 1400 }, { "epoch": 0.15, "learning_rate": 4.9642931973739356e-05, "loss": 1.8912, "step": 1410 }, { "epoch": 0.15, "learning_rate": 4.963420170414862e-05, "loss": 2.0398, "step": 1420 }, { "epoch": 0.15, "learning_rate": 4.96254714345579e-05, "loss": 1.9241, "step": 1430 }, { "epoch": 0.15, "learning_rate": 4.9616741164967176e-05, "loss": 2.0574, "step": 1440 }, { "epoch": 0.15, "learning_rate": 4.960801089537645e-05, "loss": 1.9888, "step": 1450 }, { "epoch": 0.15, "learning_rate": 4.959928062578573e-05, "loss": 2.0308, "step": 1460 }, { "epoch": 0.15, "learning_rate": 4.9590550356195e-05, "loss": 2.0775, "step": 1470 }, { "epoch": 0.15, "learning_rate": 4.9581820086604275e-05, "loss": 2.1822, "step": 1480 }, { "epoch": 0.15, "learning_rate": 4.957308981701355e-05, "loss": 2.1338, "step": 1490 }, { "epoch": 0.15, "learning_rate": 4.956435954742283e-05, "loss": 1.9209, "step": 1500 }, { "epoch": 0.16, "learning_rate": 4.95556292778321e-05, "loss": 2.1183, "step": 1510 }, { "epoch": 0.16, "learning_rate": 4.9546899008241375e-05, "loss": 2.1071, "step": 1520 }, { "epoch": 0.16, "learning_rate": 4.9538168738650655e-05, "loss": 1.9681, "step": 1530 }, { "epoch": 0.16, "learning_rate": 4.952943846905992e-05, "loss": 1.9383, "step": 1540 }, { "epoch": 0.16, "learning_rate": 4.95207081994692e-05, "loss": 2.0337, "step": 1550 }, { "epoch": 0.16, "learning_rate": 4.9511977929878475e-05, "loss": 1.994, "step": 1560 }, { "epoch": 0.16, "learning_rate": 4.950324766028775e-05, "loss": 1.8505, "step": 1570 }, { "epoch": 0.16, "learning_rate": 4.949451739069703e-05, "loss": 2.041, "step": 1580 }, { "epoch": 0.16, "learning_rate": 4.94857871211063e-05, "loss": 1.9524, "step": 1590 }, { "epoch": 0.16, "learning_rate": 4.947705685151558e-05, "loss": 1.9885, "step": 1600 }, { "epoch": 0.17, "learning_rate": 4.946832658192485e-05, "loss": 1.8919, "step": 1610 }, { "epoch": 0.17, "learning_rate": 4.945959631233413e-05, "loss": 1.9439, "step": 1620 }, { "epoch": 0.17, "learning_rate": 4.94508660427434e-05, "loss": 1.9724, "step": 1630 }, { "epoch": 0.17, "learning_rate": 4.9442135773152675e-05, "loss": 2.0511, "step": 1640 }, { "epoch": 0.17, "learning_rate": 4.9433405503561955e-05, "loss": 2.0607, "step": 1650 }, { "epoch": 0.17, "learning_rate": 4.942467523397123e-05, "loss": 1.9982, "step": 1660 }, { "epoch": 0.17, "learning_rate": 4.94159449643805e-05, "loss": 2.1207, "step": 1670 }, { "epoch": 0.17, "learning_rate": 4.9407214694789775e-05, "loss": 1.9295, "step": 1680 }, { "epoch": 0.17, "learning_rate": 4.9398484425199055e-05, "loss": 1.9922, "step": 1690 }, { "epoch": 0.18, "learning_rate": 4.938975415560833e-05, "loss": 2.0557, "step": 1700 }, { "epoch": 0.18, "learning_rate": 4.93810238860176e-05, "loss": 2.1227, "step": 1710 }, { "epoch": 0.18, "learning_rate": 4.937229361642688e-05, "loss": 1.9687, "step": 1720 }, { "epoch": 0.18, "learning_rate": 4.936356334683615e-05, "loss": 2.0497, "step": 1730 }, { "epoch": 0.18, "learning_rate": 4.935483307724543e-05, "loss": 1.9009, "step": 1740 }, { "epoch": 0.18, "learning_rate": 4.93461028076547e-05, "loss": 1.9465, "step": 1750 }, { "epoch": 0.18, "learning_rate": 4.9337372538063974e-05, "loss": 1.9086, "step": 1760 }, { "epoch": 0.18, "learning_rate": 4.9328642268473254e-05, "loss": 2.1419, "step": 1770 }, { "epoch": 0.18, "learning_rate": 4.931991199888253e-05, "loss": 1.9966, "step": 1780 }, { "epoch": 0.18, "learning_rate": 4.931118172929181e-05, "loss": 2.077, "step": 1790 }, { "epoch": 0.19, "learning_rate": 4.9302451459701074e-05, "loss": 1.8679, "step": 1800 }, { "epoch": 0.19, "learning_rate": 4.9293721190110354e-05, "loss": 2.1705, "step": 1810 }, { "epoch": 0.19, "learning_rate": 4.928499092051963e-05, "loss": 1.9652, "step": 1820 }, { "epoch": 0.19, "learning_rate": 4.92762606509289e-05, "loss": 1.8823, "step": 1830 }, { "epoch": 0.19, "learning_rate": 4.926753038133818e-05, "loss": 1.982, "step": 1840 }, { "epoch": 0.19, "learning_rate": 4.9258800111747454e-05, "loss": 2.0703, "step": 1850 }, { "epoch": 0.19, "learning_rate": 4.925006984215673e-05, "loss": 1.8744, "step": 1860 }, { "epoch": 0.19, "learning_rate": 4.9241339572566e-05, "loss": 1.9123, "step": 1870 }, { "epoch": 0.19, "learning_rate": 4.923260930297528e-05, "loss": 1.9953, "step": 1880 }, { "epoch": 0.19, "learning_rate": 4.9223879033384554e-05, "loss": 1.9946, "step": 1890 }, { "epoch": 0.2, "learning_rate": 4.921514876379383e-05, "loss": 2.0267, "step": 1900 }, { "epoch": 0.2, "learning_rate": 4.920641849420311e-05, "loss": 2.0987, "step": 1910 }, { "epoch": 0.2, "learning_rate": 4.9197688224612374e-05, "loss": 1.9375, "step": 1920 }, { "epoch": 0.2, "learning_rate": 4.9188957955021654e-05, "loss": 2.0028, "step": 1930 }, { "epoch": 0.2, "learning_rate": 4.918022768543093e-05, "loss": 1.9827, "step": 1940 }, { "epoch": 0.2, "learning_rate": 4.91714974158402e-05, "loss": 1.8602, "step": 1950 }, { "epoch": 0.2, "learning_rate": 4.916276714624948e-05, "loss": 1.8787, "step": 1960 }, { "epoch": 0.2, "learning_rate": 4.9154036876658754e-05, "loss": 1.9983, "step": 1970 }, { "epoch": 0.2, "learning_rate": 4.9145306607068034e-05, "loss": 2.0019, "step": 1980 }, { "epoch": 0.2, "learning_rate": 4.91365763374773e-05, "loss": 1.9988, "step": 1990 }, { "epoch": 0.21, "learning_rate": 4.912784606788658e-05, "loss": 1.9989, "step": 2000 }, { "epoch": 0.21, "learning_rate": 4.9119115798295854e-05, "loss": 1.9736, "step": 2010 }, { "epoch": 0.21, "learning_rate": 4.911038552870513e-05, "loss": 2.0263, "step": 2020 }, { "epoch": 0.21, "learning_rate": 4.910165525911441e-05, "loss": 1.956, "step": 2030 }, { "epoch": 0.21, "learning_rate": 4.909292498952368e-05, "loss": 1.8579, "step": 2040 }, { "epoch": 0.21, "learning_rate": 4.9084194719932954e-05, "loss": 1.9411, "step": 2050 }, { "epoch": 0.21, "learning_rate": 4.907546445034223e-05, "loss": 1.96, "step": 2060 }, { "epoch": 0.21, "learning_rate": 4.9067607207710575e-05, "loss": 2.0006, "step": 2070 }, { "epoch": 0.21, "learning_rate": 4.9058876938119855e-05, "loss": 1.8304, "step": 2080 }, { "epoch": 0.22, "learning_rate": 4.905014666852913e-05, "loss": 1.9305, "step": 2090 }, { "epoch": 0.22, "learning_rate": 4.90414163989384e-05, "loss": 1.9985, "step": 2100 }, { "epoch": 0.22, "learning_rate": 4.9032686129347675e-05, "loss": 2.0149, "step": 2110 }, { "epoch": 0.22, "learning_rate": 4.9023955859756955e-05, "loss": 2.0072, "step": 2120 }, { "epoch": 0.22, "learning_rate": 4.901522559016623e-05, "loss": 2.0938, "step": 2130 }, { "epoch": 0.22, "learning_rate": 4.90064953205755e-05, "loss": 1.9227, "step": 2140 }, { "epoch": 0.22, "learning_rate": 4.899776505098478e-05, "loss": 1.9047, "step": 2150 }, { "epoch": 0.22, "learning_rate": 4.898903478139405e-05, "loss": 2.0212, "step": 2160 }, { "epoch": 0.22, "learning_rate": 4.898030451180333e-05, "loss": 2.0709, "step": 2170 }, { "epoch": 0.22, "learning_rate": 4.89715742422126e-05, "loss": 2.0179, "step": 2180 }, { "epoch": 0.23, "learning_rate": 4.8962843972621874e-05, "loss": 2.0354, "step": 2190 }, { "epoch": 0.23, "learning_rate": 4.8954113703031154e-05, "loss": 1.9762, "step": 2200 }, { "epoch": 0.23, "learning_rate": 4.894538343344043e-05, "loss": 1.9248, "step": 2210 }, { "epoch": 0.23, "learning_rate": 4.89366531638497e-05, "loss": 1.9126, "step": 2220 }, { "epoch": 0.23, "learning_rate": 4.8927922894258974e-05, "loss": 1.9365, "step": 2230 }, { "epoch": 0.23, "learning_rate": 4.8919192624668254e-05, "loss": 2.0461, "step": 2240 }, { "epoch": 0.23, "learning_rate": 4.891046235507753e-05, "loss": 2.0511, "step": 2250 }, { "epoch": 0.23, "learning_rate": 4.89017320854868e-05, "loss": 1.8505, "step": 2260 }, { "epoch": 0.23, "learning_rate": 4.889300181589608e-05, "loss": 1.9558, "step": 2270 }, { "epoch": 0.23, "learning_rate": 4.8884271546305354e-05, "loss": 1.9439, "step": 2280 }, { "epoch": 0.24, "learning_rate": 4.887554127671463e-05, "loss": 2.008, "step": 2290 }, { "epoch": 0.24, "learning_rate": 4.88668110071239e-05, "loss": 1.8518, "step": 2300 }, { "epoch": 0.24, "learning_rate": 4.885808073753318e-05, "loss": 2.0675, "step": 2310 }, { "epoch": 0.24, "learning_rate": 4.8849350467942454e-05, "loss": 1.9116, "step": 2320 }, { "epoch": 0.24, "learning_rate": 4.884062019835173e-05, "loss": 1.8394, "step": 2330 }, { "epoch": 0.24, "learning_rate": 4.883188992876101e-05, "loss": 1.9337, "step": 2340 }, { "epoch": 0.24, "learning_rate": 4.8823159659170274e-05, "loss": 1.9895, "step": 2350 }, { "epoch": 0.24, "learning_rate": 4.8814429389579554e-05, "loss": 2.1444, "step": 2360 }, { "epoch": 0.24, "learning_rate": 4.880569911998883e-05, "loss": 1.9105, "step": 2370 }, { "epoch": 0.25, "learning_rate": 4.87969688503981e-05, "loss": 2.0462, "step": 2380 }, { "epoch": 0.25, "learning_rate": 4.878823858080738e-05, "loss": 2.0439, "step": 2390 }, { "epoch": 0.25, "learning_rate": 4.8779508311216654e-05, "loss": 1.9062, "step": 2400 }, { "epoch": 0.25, "learning_rate": 4.877077804162593e-05, "loss": 1.9375, "step": 2410 }, { "epoch": 0.25, "learning_rate": 4.87620477720352e-05, "loss": 1.9148, "step": 2420 }, { "epoch": 0.25, "learning_rate": 4.875331750244448e-05, "loss": 2.1124, "step": 2430 }, { "epoch": 0.25, "learning_rate": 4.8744587232853754e-05, "loss": 2.1599, "step": 2440 }, { "epoch": 0.25, "learning_rate": 4.873585696326303e-05, "loss": 1.7231, "step": 2450 }, { "epoch": 0.25, "learning_rate": 4.872712669367231e-05, "loss": 1.904, "step": 2460 }, { "epoch": 0.25, "learning_rate": 4.871839642408158e-05, "loss": 2.0292, "step": 2470 }, { "epoch": 0.26, "learning_rate": 4.8709666154490853e-05, "loss": 2.1222, "step": 2480 }, { "epoch": 0.26, "learning_rate": 4.870093588490013e-05, "loss": 2.0809, "step": 2490 }, { "epoch": 0.26, "learning_rate": 4.869220561530941e-05, "loss": 1.8857, "step": 2500 }, { "epoch": 0.26, "learning_rate": 4.868347534571867e-05, "loss": 1.9735, "step": 2510 }, { "epoch": 0.26, "learning_rate": 4.867474507612795e-05, "loss": 2.0159, "step": 2520 }, { "epoch": 0.26, "learning_rate": 4.866601480653723e-05, "loss": 2.0342, "step": 2530 }, { "epoch": 0.26, "learning_rate": 4.86572845369465e-05, "loss": 1.9589, "step": 2540 }, { "epoch": 0.26, "learning_rate": 4.864855426735578e-05, "loss": 1.9393, "step": 2550 }, { "epoch": 0.26, "learning_rate": 4.863982399776505e-05, "loss": 1.8396, "step": 2560 }, { "epoch": 0.26, "learning_rate": 4.8631093728174326e-05, "loss": 1.8402, "step": 2570 }, { "epoch": 0.27, "learning_rate": 4.86223634585836e-05, "loss": 1.9787, "step": 2580 }, { "epoch": 0.27, "learning_rate": 4.861363318899288e-05, "loss": 1.9899, "step": 2590 }, { "epoch": 0.27, "learning_rate": 4.860490291940215e-05, "loss": 1.9194, "step": 2600 }, { "epoch": 0.27, "learning_rate": 4.8596172649811426e-05, "loss": 1.9271, "step": 2610 }, { "epoch": 0.27, "learning_rate": 4.8587442380220706e-05, "loss": 2.0848, "step": 2620 }, { "epoch": 0.27, "learning_rate": 4.857871211062997e-05, "loss": 2.0946, "step": 2630 }, { "epoch": 0.27, "learning_rate": 4.856998184103925e-05, "loss": 2.1027, "step": 2640 }, { "epoch": 0.27, "learning_rate": 4.8561251571448526e-05, "loss": 2.091, "step": 2650 }, { "epoch": 0.27, "learning_rate": 4.8552521301857806e-05, "loss": 2.1318, "step": 2660 }, { "epoch": 0.27, "learning_rate": 4.854379103226708e-05, "loss": 2.0221, "step": 2670 }, { "epoch": 0.28, "learning_rate": 4.853506076267635e-05, "loss": 2.1148, "step": 2680 }, { "epoch": 0.28, "learning_rate": 4.852633049308563e-05, "loss": 1.7976, "step": 2690 }, { "epoch": 0.28, "learning_rate": 4.85176002234949e-05, "loss": 1.9151, "step": 2700 }, { "epoch": 0.28, "learning_rate": 4.850886995390418e-05, "loss": 1.8436, "step": 2710 }, { "epoch": 0.28, "learning_rate": 4.850013968431345e-05, "loss": 2.0596, "step": 2720 }, { "epoch": 0.28, "learning_rate": 4.8491409414722726e-05, "loss": 2.0695, "step": 2730 }, { "epoch": 0.28, "learning_rate": 4.8482679145132006e-05, "loss": 1.9452, "step": 2740 }, { "epoch": 0.28, "learning_rate": 4.847394887554128e-05, "loss": 2.0062, "step": 2750 }, { "epoch": 0.28, "learning_rate": 4.846521860595055e-05, "loss": 1.9714, "step": 2760 }, { "epoch": 0.29, "learning_rate": 4.8456488336359826e-05, "loss": 1.9842, "step": 2770 }, { "epoch": 0.29, "learning_rate": 4.8447758066769106e-05, "loss": 1.8406, "step": 2780 }, { "epoch": 0.29, "learning_rate": 4.843902779717838e-05, "loss": 2.0617, "step": 2790 }, { "epoch": 0.29, "learning_rate": 4.843029752758765e-05, "loss": 1.9513, "step": 2800 }, { "epoch": 0.29, "learning_rate": 4.842156725799693e-05, "loss": 2.1505, "step": 2810 }, { "epoch": 0.29, "learning_rate": 4.84128369884062e-05, "loss": 2.0696, "step": 2820 }, { "epoch": 0.29, "learning_rate": 4.840410671881548e-05, "loss": 2.006, "step": 2830 }, { "epoch": 0.29, "learning_rate": 4.839537644922475e-05, "loss": 1.8353, "step": 2840 }, { "epoch": 0.29, "learning_rate": 4.838664617963403e-05, "loss": 2.0312, "step": 2850 }, { "epoch": 0.29, "learning_rate": 4.8377915910043305e-05, "loss": 1.9733, "step": 2860 }, { "epoch": 0.3, "learning_rate": 4.836918564045258e-05, "loss": 1.931, "step": 2870 }, { "epoch": 0.3, "learning_rate": 4.836045537086186e-05, "loss": 1.9602, "step": 2880 }, { "epoch": 0.3, "learning_rate": 4.8351725101271125e-05, "loss": 2.1504, "step": 2890 }, { "epoch": 0.3, "learning_rate": 4.8342994831680405e-05, "loss": 2.0694, "step": 2900 }, { "epoch": 0.3, "learning_rate": 4.833426456208968e-05, "loss": 1.9285, "step": 2910 }, { "epoch": 0.3, "learning_rate": 4.832553429249895e-05, "loss": 2.1324, "step": 2920 }, { "epoch": 0.3, "learning_rate": 4.831680402290823e-05, "loss": 1.9514, "step": 2930 }, { "epoch": 0.3, "learning_rate": 4.8308073753317505e-05, "loss": 1.8926, "step": 2940 }, { "epoch": 0.3, "learning_rate": 4.829934348372678e-05, "loss": 1.7547, "step": 2950 }, { "epoch": 0.3, "learning_rate": 4.829061321413605e-05, "loss": 2.0432, "step": 2960 }, { "epoch": 0.31, "learning_rate": 4.828188294454533e-05, "loss": 2.1317, "step": 2970 }, { "epoch": 0.31, "learning_rate": 4.8273152674954605e-05, "loss": 1.9686, "step": 2980 }, { "epoch": 0.31, "learning_rate": 4.826442240536388e-05, "loss": 1.894, "step": 2990 }, { "epoch": 0.31, "learning_rate": 4.825569213577316e-05, "loss": 2.0452, "step": 3000 }, { "epoch": 0.31, "learning_rate": 4.8246961866182425e-05, "loss": 1.9332, "step": 3010 }, { "epoch": 0.31, "learning_rate": 4.8238231596591705e-05, "loss": 1.837, "step": 3020 }, { "epoch": 0.31, "learning_rate": 4.822950132700098e-05, "loss": 1.8098, "step": 3030 }, { "epoch": 0.31, "learning_rate": 4.822077105741026e-05, "loss": 1.9209, "step": 3040 }, { "epoch": 0.31, "learning_rate": 4.821204078781953e-05, "loss": 2.0261, "step": 3050 }, { "epoch": 0.32, "learning_rate": 4.8203310518228805e-05, "loss": 1.9942, "step": 3060 }, { "epoch": 0.32, "learning_rate": 4.8194580248638085e-05, "loss": 1.9699, "step": 3070 }, { "epoch": 0.32, "learning_rate": 4.818584997904735e-05, "loss": 1.9254, "step": 3080 }, { "epoch": 0.32, "learning_rate": 4.817711970945663e-05, "loss": 1.9911, "step": 3090 }, { "epoch": 0.32, "learning_rate": 4.8168389439865905e-05, "loss": 2.0407, "step": 3100 }, { "epoch": 0.32, "learning_rate": 4.815965917027518e-05, "loss": 1.9203, "step": 3110 }, { "epoch": 0.32, "learning_rate": 4.815092890068446e-05, "loss": 1.8973, "step": 3120 }, { "epoch": 0.32, "learning_rate": 4.814219863109373e-05, "loss": 1.8513, "step": 3130 }, { "epoch": 0.32, "learning_rate": 4.8133468361503005e-05, "loss": 1.901, "step": 3140 }, { "epoch": 0.32, "learning_rate": 4.812473809191228e-05, "loss": 1.9327, "step": 3150 }, { "epoch": 0.33, "learning_rate": 4.811600782232156e-05, "loss": 1.8996, "step": 3160 }, { "epoch": 0.33, "learning_rate": 4.810727755273083e-05, "loss": 1.9939, "step": 3170 }, { "epoch": 0.33, "learning_rate": 4.8098547283140104e-05, "loss": 1.8483, "step": 3180 }, { "epoch": 0.33, "learning_rate": 4.8089817013549384e-05, "loss": 2.0592, "step": 3190 }, { "epoch": 0.33, "learning_rate": 4.808108674395865e-05, "loss": 2.0302, "step": 3200 }, { "epoch": 0.33, "learning_rate": 4.807235647436793e-05, "loss": 1.9272, "step": 3210 }, { "epoch": 0.33, "learning_rate": 4.8063626204777204e-05, "loss": 2.0519, "step": 3220 }, { "epoch": 0.33, "learning_rate": 4.8054895935186484e-05, "loss": 1.9114, "step": 3230 }, { "epoch": 0.33, "learning_rate": 4.804616566559576e-05, "loss": 1.9389, "step": 3240 }, { "epoch": 0.33, "learning_rate": 4.803743539600503e-05, "loss": 1.9797, "step": 3250 }, { "epoch": 0.34, "learning_rate": 4.802870512641431e-05, "loss": 1.9191, "step": 3260 }, { "epoch": 0.34, "learning_rate": 4.801997485682358e-05, "loss": 2.0026, "step": 3270 }, { "epoch": 0.34, "learning_rate": 4.801124458723286e-05, "loss": 1.9302, "step": 3280 }, { "epoch": 0.34, "learning_rate": 4.800251431764213e-05, "loss": 1.8641, "step": 3290 }, { "epoch": 0.34, "learning_rate": 4.7993784048051404e-05, "loss": 1.9597, "step": 3300 }, { "epoch": 0.34, "learning_rate": 4.7985053778460684e-05, "loss": 1.9298, "step": 3310 }, { "epoch": 0.34, "learning_rate": 4.797632350886996e-05, "loss": 2.1863, "step": 3320 }, { "epoch": 0.34, "learning_rate": 4.796759323927923e-05, "loss": 2.1563, "step": 3330 }, { "epoch": 0.34, "learning_rate": 4.7958862969688504e-05, "loss": 2.0512, "step": 3340 }, { "epoch": 0.34, "learning_rate": 4.7950132700097784e-05, "loss": 1.9424, "step": 3350 }, { "epoch": 0.35, "learning_rate": 4.794140243050706e-05, "loss": 1.901, "step": 3360 }, { "epoch": 0.35, "learning_rate": 4.793267216091633e-05, "loss": 1.9661, "step": 3370 }, { "epoch": 0.35, "learning_rate": 4.792394189132561e-05, "loss": 1.904, "step": 3380 }, { "epoch": 0.35, "learning_rate": 4.791521162173488e-05, "loss": 1.9385, "step": 3390 }, { "epoch": 0.35, "learning_rate": 4.790648135214416e-05, "loss": 2.0139, "step": 3400 }, { "epoch": 0.35, "learning_rate": 4.789775108255343e-05, "loss": 1.9967, "step": 3410 }, { "epoch": 0.35, "learning_rate": 4.7889020812962704e-05, "loss": 2.0197, "step": 3420 }, { "epoch": 0.35, "learning_rate": 4.7880290543371984e-05, "loss": 2.0362, "step": 3430 }, { "epoch": 0.35, "learning_rate": 4.787156027378126e-05, "loss": 1.7917, "step": 3440 }, { "epoch": 0.36, "learning_rate": 4.786283000419054e-05, "loss": 1.8769, "step": 3450 }, { "epoch": 0.36, "learning_rate": 4.7854099734599803e-05, "loss": 1.9675, "step": 3460 }, { "epoch": 0.36, "learning_rate": 4.7845369465009083e-05, "loss": 1.9496, "step": 3470 }, { "epoch": 0.36, "learning_rate": 4.783663919541836e-05, "loss": 1.8427, "step": 3480 }, { "epoch": 0.36, "learning_rate": 4.782790892582763e-05, "loss": 2.0347, "step": 3490 }, { "epoch": 0.36, "learning_rate": 4.781917865623691e-05, "loss": 1.9731, "step": 3500 }, { "epoch": 0.36, "learning_rate": 4.781044838664618e-05, "loss": 2.1021, "step": 3510 }, { "epoch": 0.36, "learning_rate": 4.7801718117055457e-05, "loss": 1.8601, "step": 3520 }, { "epoch": 0.36, "learning_rate": 4.779298784746473e-05, "loss": 1.9404, "step": 3530 }, { "epoch": 0.36, "learning_rate": 4.778425757787401e-05, "loss": 2.1529, "step": 3540 }, { "epoch": 0.37, "learning_rate": 4.777552730828328e-05, "loss": 1.828, "step": 3550 }, { "epoch": 0.37, "learning_rate": 4.7766797038692556e-05, "loss": 2.0343, "step": 3560 }, { "epoch": 0.37, "learning_rate": 4.7758066769101836e-05, "loss": 1.9911, "step": 3570 }, { "epoch": 0.37, "learning_rate": 4.77493364995111e-05, "loss": 1.9486, "step": 3580 }, { "epoch": 0.37, "learning_rate": 4.774060622992038e-05, "loss": 2.0261, "step": 3590 }, { "epoch": 0.37, "learning_rate": 4.7731875960329656e-05, "loss": 1.8422, "step": 3600 }, { "epoch": 0.37, "learning_rate": 4.772314569073893e-05, "loss": 1.9237, "step": 3610 }, { "epoch": 0.37, "learning_rate": 4.771441542114821e-05, "loss": 1.833, "step": 3620 }, { "epoch": 0.37, "learning_rate": 4.770568515155748e-05, "loss": 1.9817, "step": 3630 }, { "epoch": 0.37, "learning_rate": 4.769695488196676e-05, "loss": 1.8873, "step": 3640 }, { "epoch": 0.38, "learning_rate": 4.768822461237603e-05, "loss": 1.9343, "step": 3650 }, { "epoch": 0.38, "learning_rate": 4.767949434278531e-05, "loss": 1.9811, "step": 3660 }, { "epoch": 0.38, "learning_rate": 4.767076407319458e-05, "loss": 1.9471, "step": 3670 }, { "epoch": 0.38, "learning_rate": 4.7662033803603856e-05, "loss": 2.0074, "step": 3680 }, { "epoch": 0.38, "learning_rate": 4.7653303534013136e-05, "loss": 1.9113, "step": 3690 }, { "epoch": 0.38, "learning_rate": 4.764457326442241e-05, "loss": 1.9931, "step": 3700 }, { "epoch": 0.38, "learning_rate": 4.763584299483168e-05, "loss": 1.8319, "step": 3710 }, { "epoch": 0.38, "learning_rate": 4.7627112725240956e-05, "loss": 1.9883, "step": 3720 }, { "epoch": 0.38, "learning_rate": 4.7618382455650236e-05, "loss": 2.0284, "step": 3730 }, { "epoch": 0.39, "learning_rate": 4.76096521860595e-05, "loss": 1.8626, "step": 3740 }, { "epoch": 0.39, "learning_rate": 4.760092191646878e-05, "loss": 1.926, "step": 3750 }, { "epoch": 0.39, "learning_rate": 4.759219164687806e-05, "loss": 1.9263, "step": 3760 }, { "epoch": 0.39, "learning_rate": 4.758346137728733e-05, "loss": 1.7766, "step": 3770 }, { "epoch": 0.39, "learning_rate": 4.757473110769661e-05, "loss": 1.8262, "step": 3780 }, { "epoch": 0.39, "learning_rate": 4.756600083810588e-05, "loss": 1.8861, "step": 3790 }, { "epoch": 0.39, "learning_rate": 4.7557270568515156e-05, "loss": 1.7907, "step": 3800 }, { "epoch": 0.39, "learning_rate": 4.754854029892443e-05, "loss": 1.9, "step": 3810 }, { "epoch": 0.39, "learning_rate": 4.753981002933371e-05, "loss": 1.9612, "step": 3820 }, { "epoch": 0.39, "learning_rate": 4.753107975974299e-05, "loss": 1.9975, "step": 3830 }, { "epoch": 0.4, "learning_rate": 4.7522349490152255e-05, "loss": 1.8367, "step": 3840 }, { "epoch": 0.4, "learning_rate": 4.7513619220561535e-05, "loss": 1.9552, "step": 3850 }, { "epoch": 0.4, "learning_rate": 4.750488895097081e-05, "loss": 1.9816, "step": 3860 }, { "epoch": 0.4, "learning_rate": 4.749615868138008e-05, "loss": 1.8382, "step": 3870 }, { "epoch": 0.4, "learning_rate": 4.7487428411789355e-05, "loss": 2.0723, "step": 3880 }, { "epoch": 0.4, "learning_rate": 4.7478698142198635e-05, "loss": 1.8815, "step": 3890 }, { "epoch": 0.4, "learning_rate": 4.746996787260791e-05, "loss": 1.9258, "step": 3900 }, { "epoch": 0.4, "learning_rate": 4.746123760301718e-05, "loss": 1.8927, "step": 3910 }, { "epoch": 0.4, "learning_rate": 4.745250733342646e-05, "loss": 1.8611, "step": 3920 }, { "epoch": 0.4, "learning_rate": 4.744377706383573e-05, "loss": 2.068, "step": 3930 }, { "epoch": 0.41, "learning_rate": 4.743504679424501e-05, "loss": 2.0035, "step": 3940 }, { "epoch": 0.41, "learning_rate": 4.742631652465428e-05, "loss": 1.8908, "step": 3950 }, { "epoch": 0.41, "learning_rate": 4.7417586255063555e-05, "loss": 2.0301, "step": 3960 }, { "epoch": 0.41, "learning_rate": 4.7408855985472835e-05, "loss": 1.9541, "step": 3970 }, { "epoch": 0.41, "learning_rate": 4.740012571588211e-05, "loss": 1.8775, "step": 3980 }, { "epoch": 0.41, "learning_rate": 4.739139544629138e-05, "loss": 1.9272, "step": 3990 }, { "epoch": 0.41, "learning_rate": 4.7382665176700655e-05, "loss": 1.9072, "step": 4000 }, { "epoch": 0.41, "learning_rate": 4.7373934907109935e-05, "loss": 2.0545, "step": 4010 }, { "epoch": 0.41, "learning_rate": 4.736520463751921e-05, "loss": 2.0361, "step": 4020 }, { "epoch": 0.41, "learning_rate": 4.735647436792848e-05, "loss": 2.0615, "step": 4030 }, { "epoch": 0.42, "learning_rate": 4.734774409833776e-05, "loss": 1.9838, "step": 4040 }, { "epoch": 0.42, "learning_rate": 4.7339013828747035e-05, "loss": 1.961, "step": 4050 }, { "epoch": 0.42, "learning_rate": 4.733028355915631e-05, "loss": 2.0249, "step": 4060 }, { "epoch": 0.42, "learning_rate": 4.7322426316524656e-05, "loss": 1.9382, "step": 4070 }, { "epoch": 0.42, "learning_rate": 4.731369604693393e-05, "loss": 1.7809, "step": 4080 }, { "epoch": 0.42, "learning_rate": 4.730496577734321e-05, "loss": 1.8983, "step": 4090 }, { "epoch": 0.42, "learning_rate": 4.729623550775248e-05, "loss": 1.9597, "step": 4100 }, { "epoch": 0.42, "learning_rate": 4.7287505238161756e-05, "loss": 1.9825, "step": 4110 }, { "epoch": 0.42, "learning_rate": 4.727877496857103e-05, "loss": 1.7884, "step": 4120 }, { "epoch": 0.43, "learning_rate": 4.727004469898031e-05, "loss": 1.8879, "step": 4130 }, { "epoch": 0.43, "learning_rate": 4.726131442938958e-05, "loss": 2.029, "step": 4140 }, { "epoch": 0.43, "learning_rate": 4.7252584159798856e-05, "loss": 1.8555, "step": 4150 }, { "epoch": 0.43, "learning_rate": 4.7243853890208136e-05, "loss": 2.1565, "step": 4160 }, { "epoch": 0.43, "learning_rate": 4.72351236206174e-05, "loss": 1.9317, "step": 4170 }, { "epoch": 0.43, "learning_rate": 4.722639335102668e-05, "loss": 1.9698, "step": 4180 }, { "epoch": 0.43, "learning_rate": 4.7217663081435956e-05, "loss": 1.9343, "step": 4190 }, { "epoch": 0.43, "learning_rate": 4.720893281184523e-05, "loss": 1.9083, "step": 4200 }, { "epoch": 0.43, "learning_rate": 4.720020254225451e-05, "loss": 1.9588, "step": 4210 }, { "epoch": 0.43, "learning_rate": 4.719147227266378e-05, "loss": 1.9084, "step": 4220 }, { "epoch": 0.44, "learning_rate": 4.7182742003073055e-05, "loss": 2.0141, "step": 4230 }, { "epoch": 0.44, "learning_rate": 4.717401173348233e-05, "loss": 1.9221, "step": 4240 }, { "epoch": 0.44, "learning_rate": 4.716528146389161e-05, "loss": 1.8738, "step": 4250 }, { "epoch": 0.44, "learning_rate": 4.715655119430088e-05, "loss": 1.9212, "step": 4260 }, { "epoch": 0.44, "learning_rate": 4.7147820924710155e-05, "loss": 1.8343, "step": 4270 }, { "epoch": 0.44, "learning_rate": 4.7139090655119435e-05, "loss": 1.9677, "step": 4280 }, { "epoch": 0.44, "learning_rate": 4.713036038552871e-05, "loss": 1.9636, "step": 4290 }, { "epoch": 0.44, "learning_rate": 4.712163011593798e-05, "loss": 2.0357, "step": 4300 }, { "epoch": 0.44, "learning_rate": 4.7112899846347255e-05, "loss": 1.9199, "step": 4310 }, { "epoch": 0.44, "learning_rate": 4.7104169576756535e-05, "loss": 1.9294, "step": 4320 }, { "epoch": 0.45, "learning_rate": 4.709543930716581e-05, "loss": 1.9129, "step": 4330 }, { "epoch": 0.45, "learning_rate": 4.708670903757508e-05, "loss": 2.0412, "step": 4340 }, { "epoch": 0.45, "learning_rate": 4.707797876798436e-05, "loss": 1.8468, "step": 4350 }, { "epoch": 0.45, "learning_rate": 4.706924849839363e-05, "loss": 2.0292, "step": 4360 }, { "epoch": 0.45, "learning_rate": 4.706051822880291e-05, "loss": 1.9925, "step": 4370 }, { "epoch": 0.45, "learning_rate": 4.705178795921218e-05, "loss": 2.1383, "step": 4380 }, { "epoch": 0.45, "learning_rate": 4.7043057689621455e-05, "loss": 1.8481, "step": 4390 }, { "epoch": 0.45, "learning_rate": 4.7034327420030735e-05, "loss": 1.8528, "step": 4400 }, { "epoch": 0.45, "learning_rate": 4.702559715044001e-05, "loss": 2.0173, "step": 4410 }, { "epoch": 0.46, "learning_rate": 4.701686688084928e-05, "loss": 1.7626, "step": 4420 }, { "epoch": 0.46, "learning_rate": 4.7008136611258555e-05, "loss": 1.8769, "step": 4430 }, { "epoch": 0.46, "learning_rate": 4.6999406341667835e-05, "loss": 1.8164, "step": 4440 }, { "epoch": 0.46, "learning_rate": 4.699067607207711e-05, "loss": 1.9763, "step": 4450 }, { "epoch": 0.46, "learning_rate": 4.698194580248638e-05, "loss": 1.7779, "step": 4460 }, { "epoch": 0.46, "learning_rate": 4.697321553289566e-05, "loss": 1.8596, "step": 4470 }, { "epoch": 0.46, "learning_rate": 4.696448526330493e-05, "loss": 1.8938, "step": 4480 }, { "epoch": 0.46, "learning_rate": 4.695575499371421e-05, "loss": 1.912, "step": 4490 }, { "epoch": 0.46, "learning_rate": 4.694702472412348e-05, "loss": 1.9153, "step": 4500 }, { "epoch": 0.46, "learning_rate": 4.693829445453276e-05, "loss": 1.8165, "step": 4510 }, { "epoch": 0.47, "learning_rate": 4.6929564184942035e-05, "loss": 1.8726, "step": 4520 }, { "epoch": 0.47, "learning_rate": 4.692083391535131e-05, "loss": 2.0008, "step": 4530 }, { "epoch": 0.47, "learning_rate": 4.691210364576059e-05, "loss": 1.8834, "step": 4540 }, { "epoch": 0.47, "learning_rate": 4.6903373376169854e-05, "loss": 2.0086, "step": 4550 }, { "epoch": 0.47, "learning_rate": 4.6894643106579134e-05, "loss": 1.9196, "step": 4560 }, { "epoch": 0.47, "learning_rate": 4.688591283698841e-05, "loss": 1.7789, "step": 4570 }, { "epoch": 0.47, "learning_rate": 4.687718256739768e-05, "loss": 1.8876, "step": 4580 }, { "epoch": 0.47, "learning_rate": 4.686845229780696e-05, "loss": 1.8964, "step": 4590 }, { "epoch": 0.47, "learning_rate": 4.6859722028216234e-05, "loss": 1.8953, "step": 4600 }, { "epoch": 0.47, "learning_rate": 4.685099175862551e-05, "loss": 2.1051, "step": 4610 }, { "epoch": 0.48, "learning_rate": 4.684226148903478e-05, "loss": 1.9136, "step": 4620 }, { "epoch": 0.48, "learning_rate": 4.683353121944406e-05, "loss": 1.8372, "step": 4630 }, { "epoch": 0.48, "learning_rate": 4.6824800949853334e-05, "loss": 1.8293, "step": 4640 }, { "epoch": 0.48, "learning_rate": 4.681607068026261e-05, "loss": 1.9932, "step": 4650 }, { "epoch": 0.48, "learning_rate": 4.680734041067189e-05, "loss": 1.9893, "step": 4660 }, { "epoch": 0.48, "learning_rate": 4.6798610141081154e-05, "loss": 1.8914, "step": 4670 }, { "epoch": 0.48, "learning_rate": 4.6789879871490434e-05, "loss": 2.1179, "step": 4680 }, { "epoch": 0.48, "learning_rate": 4.678114960189971e-05, "loss": 1.9395, "step": 4690 }, { "epoch": 0.48, "learning_rate": 4.677241933230899e-05, "loss": 1.8291, "step": 4700 }, { "epoch": 0.48, "learning_rate": 4.676368906271826e-05, "loss": 1.8711, "step": 4710 }, { "epoch": 0.49, "learning_rate": 4.6754958793127534e-05, "loss": 1.9817, "step": 4720 }, { "epoch": 0.49, "learning_rate": 4.6746228523536814e-05, "loss": 1.8458, "step": 4730 }, { "epoch": 0.49, "learning_rate": 4.673749825394608e-05, "loss": 1.8851, "step": 4740 }, { "epoch": 0.49, "learning_rate": 4.672876798435536e-05, "loss": 1.8997, "step": 4750 }, { "epoch": 0.49, "learning_rate": 4.6720037714764634e-05, "loss": 2.0368, "step": 4760 }, { "epoch": 0.49, "learning_rate": 4.671130744517391e-05, "loss": 1.8385, "step": 4770 }, { "epoch": 0.49, "learning_rate": 4.670257717558319e-05, "loss": 1.8494, "step": 4780 }, { "epoch": 0.49, "learning_rate": 4.669384690599246e-05, "loss": 1.9522, "step": 4790 }, { "epoch": 0.49, "learning_rate": 4.6685116636401734e-05, "loss": 1.9617, "step": 4800 }, { "epoch": 0.5, "learning_rate": 4.667638636681101e-05, "loss": 1.8706, "step": 4810 }, { "epoch": 0.5, "learning_rate": 4.666765609722029e-05, "loss": 1.7858, "step": 4820 }, { "epoch": 0.5, "learning_rate": 4.665892582762956e-05, "loss": 1.8736, "step": 4830 }, { "epoch": 0.5, "learning_rate": 4.6650195558038833e-05, "loss": 1.8902, "step": 4840 }, { "epoch": 0.5, "learning_rate": 4.6641465288448113e-05, "loss": 1.8929, "step": 4850 }, { "epoch": 0.5, "learning_rate": 4.663273501885738e-05, "loss": 1.9886, "step": 4860 }, { "epoch": 0.5, "learning_rate": 4.662400474926666e-05, "loss": 1.9129, "step": 4870 }, { "epoch": 0.5, "learning_rate": 4.661527447967593e-05, "loss": 1.8532, "step": 4880 }, { "epoch": 0.5, "learning_rate": 4.660654421008521e-05, "loss": 2.0415, "step": 4890 }, { "epoch": 0.5, "learning_rate": 4.659781394049449e-05, "loss": 1.9084, "step": 4900 }, { "epoch": 0.51, "learning_rate": 4.658908367090376e-05, "loss": 2.0404, "step": 4910 }, { "epoch": 0.51, "learning_rate": 4.658035340131304e-05, "loss": 1.8304, "step": 4920 }, { "epoch": 0.51, "learning_rate": 4.6571623131722306e-05, "loss": 1.9758, "step": 4930 }, { "epoch": 0.51, "learning_rate": 4.6562892862131586e-05, "loss": 1.944, "step": 4940 }, { "epoch": 0.51, "learning_rate": 4.655416259254086e-05, "loss": 2.0301, "step": 4950 }, { "epoch": 0.51, "learning_rate": 4.654543232295013e-05, "loss": 1.9225, "step": 4960 }, { "epoch": 0.51, "learning_rate": 4.653670205335941e-05, "loss": 2.0178, "step": 4970 }, { "epoch": 0.51, "learning_rate": 4.6527971783768686e-05, "loss": 1.84, "step": 4980 }, { "epoch": 0.51, "learning_rate": 4.651924151417796e-05, "loss": 1.8213, "step": 4990 }, { "epoch": 0.51, "learning_rate": 4.651051124458723e-05, "loss": 1.9907, "step": 5000 }, { "epoch": 0.52, "learning_rate": 4.650178097499651e-05, "loss": 2.1089, "step": 5010 }, { "epoch": 0.52, "learning_rate": 4.6493050705405786e-05, "loss": 1.8555, "step": 5020 }, { "epoch": 0.52, "learning_rate": 4.648432043581506e-05, "loss": 1.9852, "step": 5030 }, { "epoch": 0.52, "learning_rate": 4.647559016622434e-05, "loss": 1.7622, "step": 5040 }, { "epoch": 0.52, "learning_rate": 4.6466859896633606e-05, "loss": 1.9574, "step": 5050 }, { "epoch": 0.52, "learning_rate": 4.6458129627042886e-05, "loss": 1.8197, "step": 5060 }, { "epoch": 0.52, "learning_rate": 4.644939935745216e-05, "loss": 1.9539, "step": 5070 }, { "epoch": 0.52, "learning_rate": 4.644066908786144e-05, "loss": 1.8171, "step": 5080 }, { "epoch": 0.52, "learning_rate": 4.643193881827071e-05, "loss": 1.9054, "step": 5090 }, { "epoch": 0.53, "learning_rate": 4.6423208548679986e-05, "loss": 1.964, "step": 5100 }, { "epoch": 0.53, "learning_rate": 4.6414478279089266e-05, "loss": 1.7665, "step": 5110 }, { "epoch": 0.53, "learning_rate": 4.640574800949853e-05, "loss": 1.7917, "step": 5120 }, { "epoch": 0.53, "learning_rate": 4.639701773990781e-05, "loss": 1.9539, "step": 5130 }, { "epoch": 0.53, "learning_rate": 4.6388287470317086e-05, "loss": 2.0556, "step": 5140 }, { "epoch": 0.53, "learning_rate": 4.637955720072636e-05, "loss": 1.8934, "step": 5150 }, { "epoch": 0.53, "learning_rate": 4.637082693113564e-05, "loss": 1.8274, "step": 5160 }, { "epoch": 0.53, "learning_rate": 4.636209666154491e-05, "loss": 1.9965, "step": 5170 }, { "epoch": 0.53, "learning_rate": 4.6353366391954186e-05, "loss": 1.9861, "step": 5180 }, { "epoch": 0.53, "learning_rate": 4.634463612236346e-05, "loss": 1.9275, "step": 5190 }, { "epoch": 0.54, "learning_rate": 4.633590585277274e-05, "loss": 1.8608, "step": 5200 }, { "epoch": 0.54, "learning_rate": 4.6327175583182005e-05, "loss": 2.0487, "step": 5210 }, { "epoch": 0.54, "learning_rate": 4.6318445313591286e-05, "loss": 1.8333, "step": 5220 }, { "epoch": 0.54, "learning_rate": 4.6309715044000566e-05, "loss": 1.9095, "step": 5230 }, { "epoch": 0.54, "learning_rate": 4.630098477440983e-05, "loss": 1.7228, "step": 5240 }, { "epoch": 0.54, "learning_rate": 4.629225450481911e-05, "loss": 1.7879, "step": 5250 }, { "epoch": 0.54, "learning_rate": 4.6283524235228385e-05, "loss": 1.9049, "step": 5260 }, { "epoch": 0.54, "learning_rate": 4.627479396563766e-05, "loss": 2.1479, "step": 5270 }, { "epoch": 0.54, "learning_rate": 4.626606369604693e-05, "loss": 1.8826, "step": 5280 }, { "epoch": 0.54, "learning_rate": 4.625733342645621e-05, "loss": 1.86, "step": 5290 }, { "epoch": 0.55, "learning_rate": 4.624860315686549e-05, "loss": 1.8297, "step": 5300 }, { "epoch": 0.55, "learning_rate": 4.623987288727476e-05, "loss": 1.9474, "step": 5310 }, { "epoch": 0.55, "learning_rate": 4.623114261768404e-05, "loss": 1.9161, "step": 5320 }, { "epoch": 0.55, "learning_rate": 4.622241234809331e-05, "loss": 1.8701, "step": 5330 }, { "epoch": 0.55, "learning_rate": 4.6213682078502585e-05, "loss": 1.9483, "step": 5340 }, { "epoch": 0.55, "learning_rate": 4.620495180891186e-05, "loss": 1.8677, "step": 5350 }, { "epoch": 0.55, "learning_rate": 4.619622153932114e-05, "loss": 1.8863, "step": 5360 }, { "epoch": 0.55, "learning_rate": 4.618749126973041e-05, "loss": 2.0326, "step": 5370 }, { "epoch": 0.55, "learning_rate": 4.6178761000139685e-05, "loss": 1.9934, "step": 5380 }, { "epoch": 0.55, "learning_rate": 4.6170030730548965e-05, "loss": 1.9114, "step": 5390 }, { "epoch": 0.56, "learning_rate": 4.616130046095823e-05, "loss": 1.8575, "step": 5400 }, { "epoch": 0.56, "learning_rate": 4.615257019136751e-05, "loss": 1.881, "step": 5410 }, { "epoch": 0.56, "learning_rate": 4.6143839921776785e-05, "loss": 1.8793, "step": 5420 }, { "epoch": 0.56, "learning_rate": 4.613510965218606e-05, "loss": 1.8581, "step": 5430 }, { "epoch": 0.56, "learning_rate": 4.612637938259534e-05, "loss": 1.921, "step": 5440 }, { "epoch": 0.56, "learning_rate": 4.611764911300461e-05, "loss": 1.924, "step": 5450 }, { "epoch": 0.56, "learning_rate": 4.6108918843413885e-05, "loss": 1.8078, "step": 5460 }, { "epoch": 0.56, "learning_rate": 4.610018857382316e-05, "loss": 1.9087, "step": 5470 }, { "epoch": 0.56, "learning_rate": 4.609145830423244e-05, "loss": 1.9449, "step": 5480 }, { "epoch": 0.57, "learning_rate": 4.608272803464171e-05, "loss": 1.9525, "step": 5490 }, { "epoch": 0.57, "learning_rate": 4.6073997765050985e-05, "loss": 2.0435, "step": 5500 }, { "epoch": 0.57, "learning_rate": 4.6065267495460265e-05, "loss": 1.9521, "step": 5510 }, { "epoch": 0.57, "learning_rate": 4.605653722586954e-05, "loss": 1.8731, "step": 5520 }, { "epoch": 0.57, "learning_rate": 4.604780695627881e-05, "loss": 1.8189, "step": 5530 }, { "epoch": 0.57, "learning_rate": 4.6039076686688084e-05, "loss": 1.9498, "step": 5540 }, { "epoch": 0.57, "learning_rate": 4.6030346417097364e-05, "loss": 1.9245, "step": 5550 }, { "epoch": 0.57, "learning_rate": 4.602161614750664e-05, "loss": 1.8627, "step": 5560 }, { "epoch": 0.57, "learning_rate": 4.601288587791591e-05, "loss": 1.9171, "step": 5570 }, { "epoch": 0.57, "learning_rate": 4.600415560832519e-05, "loss": 1.8403, "step": 5580 }, { "epoch": 0.58, "learning_rate": 4.599542533873446e-05, "loss": 1.8587, "step": 5590 }, { "epoch": 0.58, "learning_rate": 4.598669506914374e-05, "loss": 1.8318, "step": 5600 }, { "epoch": 0.58, "learning_rate": 4.597796479955301e-05, "loss": 1.954, "step": 5610 }, { "epoch": 0.58, "learning_rate": 4.5969234529962284e-05, "loss": 1.9895, "step": 5620 }, { "epoch": 0.58, "learning_rate": 4.5960504260371564e-05, "loss": 2.0052, "step": 5630 }, { "epoch": 0.58, "learning_rate": 4.595177399078084e-05, "loss": 1.8374, "step": 5640 }, { "epoch": 0.58, "learning_rate": 4.594304372119011e-05, "loss": 1.9758, "step": 5650 }, { "epoch": 0.58, "learning_rate": 4.5934313451599384e-05, "loss": 2.0072, "step": 5660 }, { "epoch": 0.58, "learning_rate": 4.5925583182008664e-05, "loss": 1.8884, "step": 5670 }, { "epoch": 0.58, "learning_rate": 4.591685291241794e-05, "loss": 1.84, "step": 5680 }, { "epoch": 0.59, "learning_rate": 4.590812264282721e-05, "loss": 1.9712, "step": 5690 }, { "epoch": 0.59, "learning_rate": 4.589939237323649e-05, "loss": 1.9127, "step": 5700 }, { "epoch": 0.59, "learning_rate": 4.5890662103645764e-05, "loss": 2.0855, "step": 5710 }, { "epoch": 0.59, "learning_rate": 4.588193183405504e-05, "loss": 2.1225, "step": 5720 }, { "epoch": 0.59, "learning_rate": 4.587320156446431e-05, "loss": 1.8943, "step": 5730 }, { "epoch": 0.59, "learning_rate": 4.586447129487359e-05, "loss": 1.6987, "step": 5740 }, { "epoch": 0.59, "learning_rate": 4.5855741025282864e-05, "loss": 1.8871, "step": 5750 }, { "epoch": 0.59, "learning_rate": 4.584701075569214e-05, "loss": 1.8374, "step": 5760 }, { "epoch": 0.59, "learning_rate": 4.583828048610142e-05, "loss": 1.9954, "step": 5770 }, { "epoch": 0.6, "learning_rate": 4.5829550216510684e-05, "loss": 1.8718, "step": 5780 }, { "epoch": 0.6, "learning_rate": 4.5820819946919964e-05, "loss": 1.7798, "step": 5790 }, { "epoch": 0.6, "learning_rate": 4.581208967732924e-05, "loss": 1.8892, "step": 5800 }, { "epoch": 0.6, "learning_rate": 4.580335940773851e-05, "loss": 1.9298, "step": 5810 }, { "epoch": 0.6, "learning_rate": 4.579462913814779e-05, "loss": 1.8787, "step": 5820 }, { "epoch": 0.6, "learning_rate": 4.5785898868557063e-05, "loss": 1.7445, "step": 5830 }, { "epoch": 0.6, "learning_rate": 4.577716859896634e-05, "loss": 1.9633, "step": 5840 }, { "epoch": 0.6, "learning_rate": 4.576843832937561e-05, "loss": 1.8791, "step": 5850 }, { "epoch": 0.6, "learning_rate": 4.575970805978489e-05, "loss": 1.8588, "step": 5860 }, { "epoch": 0.6, "learning_rate": 4.575097779019416e-05, "loss": 2.0058, "step": 5870 }, { "epoch": 0.61, "learning_rate": 4.5742247520603437e-05, "loss": 2.0155, "step": 5880 }, { "epoch": 0.61, "learning_rate": 4.573351725101272e-05, "loss": 1.9881, "step": 5890 }, { "epoch": 0.61, "learning_rate": 4.572478698142199e-05, "loss": 1.975, "step": 5900 }, { "epoch": 0.61, "learning_rate": 4.571605671183126e-05, "loss": 1.9278, "step": 5910 }, { "epoch": 0.61, "learning_rate": 4.5707326442240536e-05, "loss": 1.9845, "step": 5920 }, { "epoch": 0.61, "learning_rate": 4.5698596172649817e-05, "loss": 1.8113, "step": 5930 }, { "epoch": 0.61, "learning_rate": 4.568986590305909e-05, "loss": 1.8022, "step": 5940 }, { "epoch": 0.61, "learning_rate": 4.568113563346836e-05, "loss": 1.9143, "step": 5950 }, { "epoch": 0.61, "learning_rate": 4.567240536387764e-05, "loss": 1.8843, "step": 5960 }, { "epoch": 0.61, "learning_rate": 4.566367509428691e-05, "loss": 1.9027, "step": 5970 }, { "epoch": 0.62, "learning_rate": 4.565494482469619e-05, "loss": 1.8579, "step": 5980 }, { "epoch": 0.62, "learning_rate": 4.564621455510546e-05, "loss": 1.9922, "step": 5990 }, { "epoch": 0.62, "learning_rate": 4.5637484285514736e-05, "loss": 1.8765, "step": 6000 }, { "epoch": 0.62, "learning_rate": 4.5628754015924016e-05, "loss": 1.9406, "step": 6010 }, { "epoch": 0.62, "learning_rate": 4.562002374633329e-05, "loss": 1.7855, "step": 6020 }, { "epoch": 0.62, "learning_rate": 4.561129347674256e-05, "loss": 1.9938, "step": 6030 }, { "epoch": 0.62, "learning_rate": 4.5602563207151836e-05, "loss": 2.0107, "step": 6040 }, { "epoch": 0.62, "learning_rate": 4.5593832937561116e-05, "loss": 1.9262, "step": 6050 }, { "epoch": 0.62, "learning_rate": 4.558510266797039e-05, "loss": 1.8582, "step": 6060 }, { "epoch": 0.62, "learning_rate": 4.557637239837966e-05, "loss": 1.8521, "step": 6070 }, { "epoch": 0.63, "learning_rate": 4.556851515574801e-05, "loss": 2.0212, "step": 6080 }, { "epoch": 0.63, "learning_rate": 4.5559784886157284e-05, "loss": 1.6967, "step": 6090 }, { "epoch": 0.63, "learning_rate": 4.5551054616566564e-05, "loss": 1.8419, "step": 6100 }, { "epoch": 0.63, "learning_rate": 4.554232434697584e-05, "loss": 1.8578, "step": 6110 }, { "epoch": 0.63, "learning_rate": 4.553359407738511e-05, "loss": 1.942, "step": 6120 }, { "epoch": 0.63, "learning_rate": 4.552486380779439e-05, "loss": 1.8642, "step": 6130 }, { "epoch": 0.63, "learning_rate": 4.5516133538203664e-05, "loss": 1.9283, "step": 6140 }, { "epoch": 0.63, "learning_rate": 4.550740326861294e-05, "loss": 1.9561, "step": 6150 }, { "epoch": 0.63, "learning_rate": 4.549867299902221e-05, "loss": 1.9378, "step": 6160 }, { "epoch": 0.64, "learning_rate": 4.548994272943149e-05, "loss": 2.0321, "step": 6170 }, { "epoch": 0.64, "learning_rate": 4.5481212459840764e-05, "loss": 1.9804, "step": 6180 }, { "epoch": 0.64, "learning_rate": 4.547248219025004e-05, "loss": 1.8869, "step": 6190 }, { "epoch": 0.64, "learning_rate": 4.546375192065932e-05, "loss": 1.9017, "step": 6200 }, { "epoch": 0.64, "learning_rate": 4.5455021651068583e-05, "loss": 1.9475, "step": 6210 }, { "epoch": 0.64, "learning_rate": 4.5446291381477864e-05, "loss": 1.8769, "step": 6220 }, { "epoch": 0.64, "learning_rate": 4.543756111188714e-05, "loss": 1.9344, "step": 6230 }, { "epoch": 0.64, "learning_rate": 4.542883084229641e-05, "loss": 1.841, "step": 6240 }, { "epoch": 0.64, "learning_rate": 4.542010057270569e-05, "loss": 2.077, "step": 6250 }, { "epoch": 0.64, "learning_rate": 4.541137030311496e-05, "loss": 2.0144, "step": 6260 }, { "epoch": 0.65, "learning_rate": 4.540264003352424e-05, "loss": 1.8897, "step": 6270 }, { "epoch": 0.65, "learning_rate": 4.539390976393351e-05, "loss": 1.8136, "step": 6280 }, { "epoch": 0.65, "learning_rate": 4.538517949434279e-05, "loss": 1.9073, "step": 6290 }, { "epoch": 0.65, "learning_rate": 4.537644922475206e-05, "loss": 1.9607, "step": 6300 }, { "epoch": 0.65, "learning_rate": 4.5367718955161336e-05, "loss": 1.8875, "step": 6310 }, { "epoch": 0.65, "learning_rate": 4.5358988685570617e-05, "loss": 1.9473, "step": 6320 }, { "epoch": 0.65, "learning_rate": 4.535025841597988e-05, "loss": 1.7896, "step": 6330 }, { "epoch": 0.65, "learning_rate": 4.534152814638916e-05, "loss": 1.8456, "step": 6340 }, { "epoch": 0.65, "learning_rate": 4.5332797876798436e-05, "loss": 1.8739, "step": 6350 }, { "epoch": 0.65, "learning_rate": 4.5324067607207716e-05, "loss": 2.0112, "step": 6360 }, { "epoch": 0.66, "learning_rate": 4.531533733761699e-05, "loss": 1.8687, "step": 6370 }, { "epoch": 0.66, "learning_rate": 4.530660706802626e-05, "loss": 1.7995, "step": 6380 }, { "epoch": 0.66, "learning_rate": 4.529787679843554e-05, "loss": 1.9376, "step": 6390 }, { "epoch": 0.66, "learning_rate": 4.528914652884481e-05, "loss": 1.9731, "step": 6400 }, { "epoch": 0.66, "learning_rate": 4.528041625925409e-05, "loss": 1.9968, "step": 6410 }, { "epoch": 0.66, "learning_rate": 4.527168598966336e-05, "loss": 1.8424, "step": 6420 }, { "epoch": 0.66, "learning_rate": 4.5262955720072636e-05, "loss": 1.8483, "step": 6430 }, { "epoch": 0.66, "learning_rate": 4.5254225450481916e-05, "loss": 1.9378, "step": 6440 }, { "epoch": 0.66, "learning_rate": 4.524549518089119e-05, "loss": 1.9041, "step": 6450 }, { "epoch": 0.67, "learning_rate": 4.523676491130046e-05, "loss": 1.943, "step": 6460 }, { "epoch": 0.67, "learning_rate": 4.5228034641709736e-05, "loss": 1.9822, "step": 6470 }, { "epoch": 0.67, "learning_rate": 4.5219304372119016e-05, "loss": 1.7001, "step": 6480 }, { "epoch": 0.67, "learning_rate": 4.521057410252828e-05, "loss": 1.9421, "step": 6490 }, { "epoch": 0.67, "learning_rate": 4.520184383293756e-05, "loss": 1.9232, "step": 6500 }, { "epoch": 0.67, "learning_rate": 4.519311356334684e-05, "loss": 1.9127, "step": 6510 }, { "epoch": 0.67, "learning_rate": 4.518438329375611e-05, "loss": 1.7644, "step": 6520 }, { "epoch": 0.67, "learning_rate": 4.517565302416539e-05, "loss": 1.8973, "step": 6530 }, { "epoch": 0.67, "learning_rate": 4.516692275457466e-05, "loss": 2.0772, "step": 6540 }, { "epoch": 0.67, "learning_rate": 4.515819248498394e-05, "loss": 1.9666, "step": 6550 }, { "epoch": 0.68, "learning_rate": 4.514946221539321e-05, "loss": 1.8494, "step": 6560 }, { "epoch": 0.68, "learning_rate": 4.514073194580249e-05, "loss": 1.8023, "step": 6570 }, { "epoch": 0.68, "learning_rate": 4.513200167621177e-05, "loss": 1.8861, "step": 6580 }, { "epoch": 0.68, "learning_rate": 4.5123271406621036e-05, "loss": 1.8539, "step": 6590 }, { "epoch": 0.68, "learning_rate": 4.5114541137030316e-05, "loss": 1.86, "step": 6600 }, { "epoch": 0.68, "learning_rate": 4.510581086743959e-05, "loss": 1.9367, "step": 6610 }, { "epoch": 0.68, "learning_rate": 4.509708059784886e-05, "loss": 1.847, "step": 6620 }, { "epoch": 0.68, "learning_rate": 4.5088350328258135e-05, "loss": 1.9183, "step": 6630 }, { "epoch": 0.68, "learning_rate": 4.5079620058667415e-05, "loss": 1.9808, "step": 6640 }, { "epoch": 0.68, "learning_rate": 4.507088978907669e-05, "loss": 1.8325, "step": 6650 }, { "epoch": 0.69, "learning_rate": 4.506215951948596e-05, "loss": 1.9137, "step": 6660 }, { "epoch": 0.69, "learning_rate": 4.505342924989524e-05, "loss": 1.81, "step": 6670 }, { "epoch": 0.69, "learning_rate": 4.504469898030451e-05, "loss": 1.7753, "step": 6680 }, { "epoch": 0.69, "learning_rate": 4.503596871071379e-05, "loss": 1.9381, "step": 6690 }, { "epoch": 0.69, "learning_rate": 4.502723844112307e-05, "loss": 1.9853, "step": 6700 }, { "epoch": 0.69, "learning_rate": 4.5018508171532335e-05, "loss": 1.9685, "step": 6710 }, { "epoch": 0.69, "learning_rate": 4.5009777901941615e-05, "loss": 2.0147, "step": 6720 }, { "epoch": 0.69, "learning_rate": 4.500104763235089e-05, "loss": 1.9566, "step": 6730 }, { "epoch": 0.69, "learning_rate": 4.499231736276017e-05, "loss": 1.9469, "step": 6740 }, { "epoch": 0.69, "learning_rate": 4.4983587093169435e-05, "loss": 1.9955, "step": 6750 }, { "epoch": 0.7, "learning_rate": 4.4974856823578715e-05, "loss": 1.88, "step": 6760 }, { "epoch": 0.7, "learning_rate": 4.4966126553987995e-05, "loss": 1.9952, "step": 6770 }, { "epoch": 0.7, "learning_rate": 4.495739628439726e-05, "loss": 1.8726, "step": 6780 }, { "epoch": 0.7, "learning_rate": 4.494866601480654e-05, "loss": 1.8484, "step": 6790 }, { "epoch": 0.7, "learning_rate": 4.4939935745215815e-05, "loss": 1.985, "step": 6800 }, { "epoch": 0.7, "learning_rate": 4.493120547562509e-05, "loss": 1.8963, "step": 6810 }, { "epoch": 0.7, "learning_rate": 4.492247520603436e-05, "loss": 1.9718, "step": 6820 }, { "epoch": 0.7, "learning_rate": 4.491374493644364e-05, "loss": 1.9229, "step": 6830 }, { "epoch": 0.7, "learning_rate": 4.4905014666852915e-05, "loss": 1.9079, "step": 6840 }, { "epoch": 0.71, "learning_rate": 4.489628439726219e-05, "loss": 1.9877, "step": 6850 }, { "epoch": 0.71, "learning_rate": 4.488755412767147e-05, "loss": 1.9899, "step": 6860 }, { "epoch": 0.71, "learning_rate": 4.4878823858080735e-05, "loss": 1.9247, "step": 6870 }, { "epoch": 0.71, "learning_rate": 4.4870093588490015e-05, "loss": 1.9661, "step": 6880 }, { "epoch": 0.71, "learning_rate": 4.486136331889929e-05, "loss": 1.941, "step": 6890 }, { "epoch": 0.71, "learning_rate": 4.485263304930856e-05, "loss": 1.9007, "step": 6900 }, { "epoch": 0.71, "learning_rate": 4.484390277971784e-05, "loss": 1.9399, "step": 6910 }, { "epoch": 0.71, "learning_rate": 4.4835172510127114e-05, "loss": 1.8359, "step": 6920 }, { "epoch": 0.71, "learning_rate": 4.4826442240536394e-05, "loss": 1.8088, "step": 6930 }, { "epoch": 0.71, "learning_rate": 4.481771197094566e-05, "loss": 2.1038, "step": 6940 }, { "epoch": 0.72, "learning_rate": 4.480898170135494e-05, "loss": 1.8658, "step": 6950 }, { "epoch": 0.72, "learning_rate": 4.4800251431764214e-05, "loss": 1.8927, "step": 6960 }, { "epoch": 0.72, "learning_rate": 4.479152116217349e-05, "loss": 1.8161, "step": 6970 }, { "epoch": 0.72, "learning_rate": 4.478279089258277e-05, "loss": 2.0503, "step": 6980 }, { "epoch": 0.72, "learning_rate": 4.477406062299204e-05, "loss": 1.8365, "step": 6990 }, { "epoch": 0.72, "learning_rate": 4.4765330353401314e-05, "loss": 1.8593, "step": 7000 }, { "epoch": 0.72, "learning_rate": 4.475660008381059e-05, "loss": 1.8986, "step": 7010 }, { "epoch": 0.72, "learning_rate": 4.474786981421987e-05, "loss": 1.8684, "step": 7020 }, { "epoch": 0.72, "learning_rate": 4.473913954462914e-05, "loss": 1.9206, "step": 7030 }, { "epoch": 0.72, "learning_rate": 4.4730409275038414e-05, "loss": 1.9081, "step": 7040 }, { "epoch": 0.73, "learning_rate": 4.4721679005447694e-05, "loss": 1.9756, "step": 7050 }, { "epoch": 0.73, "learning_rate": 4.471294873585696e-05, "loss": 1.8253, "step": 7060 }, { "epoch": 0.73, "learning_rate": 4.470421846626624e-05, "loss": 2.0947, "step": 7070 }, { "epoch": 0.73, "learning_rate": 4.4695488196675514e-05, "loss": 1.7639, "step": 7080 }, { "epoch": 0.73, "learning_rate": 4.468675792708479e-05, "loss": 1.8864, "step": 7090 }, { "epoch": 0.73, "learning_rate": 4.467802765749407e-05, "loss": 2.0049, "step": 7100 }, { "epoch": 0.73, "learning_rate": 4.466929738790334e-05, "loss": 1.8289, "step": 7110 }, { "epoch": 0.73, "learning_rate": 4.4660567118312614e-05, "loss": 1.9287, "step": 7120 }, { "epoch": 0.73, "learning_rate": 4.465183684872189e-05, "loss": 1.8943, "step": 7130 }, { "epoch": 0.74, "learning_rate": 4.464310657913117e-05, "loss": 1.8995, "step": 7140 }, { "epoch": 0.74, "learning_rate": 4.463437630954044e-05, "loss": 1.9107, "step": 7150 }, { "epoch": 0.74, "learning_rate": 4.4625646039949714e-05, "loss": 1.8897, "step": 7160 }, { "epoch": 0.74, "learning_rate": 4.4616915770358994e-05, "loss": 1.9495, "step": 7170 }, { "epoch": 0.74, "learning_rate": 4.460818550076827e-05, "loss": 1.8295, "step": 7180 }, { "epoch": 0.74, "learning_rate": 4.459945523117754e-05, "loss": 2.0628, "step": 7190 }, { "epoch": 0.74, "learning_rate": 4.4590724961586813e-05, "loss": 1.9066, "step": 7200 }, { "epoch": 0.74, "learning_rate": 4.4581994691996094e-05, "loss": 1.7927, "step": 7210 }, { "epoch": 0.74, "learning_rate": 4.457326442240537e-05, "loss": 1.8584, "step": 7220 }, { "epoch": 0.74, "learning_rate": 4.456453415281464e-05, "loss": 1.9142, "step": 7230 }, { "epoch": 0.75, "learning_rate": 4.455580388322392e-05, "loss": 1.9398, "step": 7240 }, { "epoch": 0.75, "learning_rate": 4.4547073613633187e-05, "loss": 1.8544, "step": 7250 }, { "epoch": 0.75, "learning_rate": 4.453834334404247e-05, "loss": 2.0312, "step": 7260 }, { "epoch": 0.75, "learning_rate": 4.452961307445174e-05, "loss": 1.9066, "step": 7270 }, { "epoch": 0.75, "learning_rate": 4.452088280486101e-05, "loss": 1.8878, "step": 7280 }, { "epoch": 0.75, "learning_rate": 4.451215253527029e-05, "loss": 1.9033, "step": 7290 }, { "epoch": 0.75, "learning_rate": 4.4503422265679567e-05, "loss": 1.8675, "step": 7300 }, { "epoch": 0.75, "learning_rate": 4.449469199608884e-05, "loss": 1.9868, "step": 7310 }, { "epoch": 0.75, "learning_rate": 4.448596172649811e-05, "loss": 1.8584, "step": 7320 }, { "epoch": 0.75, "learning_rate": 4.447723145690739e-05, "loss": 1.8565, "step": 7330 }, { "epoch": 0.76, "learning_rate": 4.4468501187316666e-05, "loss": 1.8517, "step": 7340 }, { "epoch": 0.76, "learning_rate": 4.445977091772594e-05, "loss": 2.0381, "step": 7350 }, { "epoch": 0.76, "learning_rate": 4.445104064813522e-05, "loss": 1.8618, "step": 7360 }, { "epoch": 0.76, "learning_rate": 4.444231037854449e-05, "loss": 1.9044, "step": 7370 }, { "epoch": 0.76, "learning_rate": 4.4433580108953766e-05, "loss": 1.9624, "step": 7380 }, { "epoch": 0.76, "learning_rate": 4.442484983936304e-05, "loss": 1.93, "step": 7390 }, { "epoch": 0.76, "learning_rate": 4.441611956977232e-05, "loss": 1.7296, "step": 7400 }, { "epoch": 0.76, "learning_rate": 4.440738930018159e-05, "loss": 1.8708, "step": 7410 }, { "epoch": 0.76, "learning_rate": 4.4398659030590866e-05, "loss": 1.9082, "step": 7420 }, { "epoch": 0.77, "learning_rate": 4.4389928761000146e-05, "loss": 1.9923, "step": 7430 }, { "epoch": 0.77, "learning_rate": 4.438119849140941e-05, "loss": 1.8932, "step": 7440 }, { "epoch": 0.77, "learning_rate": 4.437246822181869e-05, "loss": 1.955, "step": 7450 }, { "epoch": 0.77, "learning_rate": 4.4363737952227966e-05, "loss": 1.8237, "step": 7460 }, { "epoch": 0.77, "learning_rate": 4.435500768263724e-05, "loss": 1.9451, "step": 7470 }, { "epoch": 0.77, "learning_rate": 4.434627741304652e-05, "loss": 1.9676, "step": 7480 }, { "epoch": 0.77, "learning_rate": 4.433754714345579e-05, "loss": 1.9159, "step": 7490 }, { "epoch": 0.77, "learning_rate": 4.4328816873865066e-05, "loss": 1.9371, "step": 7500 }, { "epoch": 0.77, "learning_rate": 4.432008660427434e-05, "loss": 2.0315, "step": 7510 }, { "epoch": 0.77, "learning_rate": 4.431135633468362e-05, "loss": 1.9277, "step": 7520 }, { "epoch": 0.78, "learning_rate": 4.430262606509289e-05, "loss": 1.9546, "step": 7530 }, { "epoch": 0.78, "learning_rate": 4.4293895795502166e-05, "loss": 1.8681, "step": 7540 }, { "epoch": 0.78, "learning_rate": 4.4285165525911446e-05, "loss": 1.9917, "step": 7550 }, { "epoch": 0.78, "learning_rate": 4.427643525632072e-05, "loss": 1.9484, "step": 7560 }, { "epoch": 0.78, "learning_rate": 4.426770498672999e-05, "loss": 1.7766, "step": 7570 }, { "epoch": 0.78, "learning_rate": 4.4258974717139266e-05, "loss": 1.7735, "step": 7580 }, { "epoch": 0.78, "learning_rate": 4.4250244447548546e-05, "loss": 2.0253, "step": 7590 }, { "epoch": 0.78, "learning_rate": 4.424151417795782e-05, "loss": 1.9334, "step": 7600 }, { "epoch": 0.78, "learning_rate": 4.423278390836709e-05, "loss": 1.9026, "step": 7610 }, { "epoch": 0.78, "learning_rate": 4.422405363877637e-05, "loss": 1.8632, "step": 7620 }, { "epoch": 0.79, "learning_rate": 4.421532336918564e-05, "loss": 1.9665, "step": 7630 }, { "epoch": 0.79, "learning_rate": 4.420659309959492e-05, "loss": 1.9057, "step": 7640 }, { "epoch": 0.79, "learning_rate": 4.419786283000419e-05, "loss": 1.8698, "step": 7650 }, { "epoch": 0.79, "learning_rate": 4.4189132560413465e-05, "loss": 1.836, "step": 7660 }, { "epoch": 0.79, "learning_rate": 4.4180402290822745e-05, "loss": 1.9126, "step": 7670 }, { "epoch": 0.79, "learning_rate": 4.417167202123202e-05, "loss": 1.8486, "step": 7680 }, { "epoch": 0.79, "learning_rate": 4.416294175164129e-05, "loss": 1.9936, "step": 7690 }, { "epoch": 0.79, "learning_rate": 4.4154211482050565e-05, "loss": 1.8677, "step": 7700 }, { "epoch": 0.79, "learning_rate": 4.4145481212459845e-05, "loss": 1.9565, "step": 7710 }, { "epoch": 0.79, "learning_rate": 4.413675094286912e-05, "loss": 1.8159, "step": 7720 }, { "epoch": 0.8, "learning_rate": 4.412802067327839e-05, "loss": 1.8449, "step": 7730 }, { "epoch": 0.8, "learning_rate": 4.411929040368767e-05, "loss": 1.8823, "step": 7740 }, { "epoch": 0.8, "learning_rate": 4.4110560134096945e-05, "loss": 1.8765, "step": 7750 }, { "epoch": 0.8, "learning_rate": 4.410182986450622e-05, "loss": 1.9246, "step": 7760 }, { "epoch": 0.8, "learning_rate": 4.409309959491549e-05, "loss": 1.8965, "step": 7770 }, { "epoch": 0.8, "learning_rate": 4.408436932532477e-05, "loss": 2.0334, "step": 7780 }, { "epoch": 0.8, "learning_rate": 4.4075639055734045e-05, "loss": 1.95, "step": 7790 }, { "epoch": 0.8, "learning_rate": 4.406690878614332e-05, "loss": 1.8416, "step": 7800 }, { "epoch": 0.8, "learning_rate": 4.40581785165526e-05, "loss": 2.128, "step": 7810 }, { "epoch": 0.81, "learning_rate": 4.4049448246961865e-05, "loss": 2.0332, "step": 7820 }, { "epoch": 0.81, "learning_rate": 4.4040717977371145e-05, "loss": 1.9482, "step": 7830 }, { "epoch": 0.81, "learning_rate": 4.403198770778042e-05, "loss": 1.907, "step": 7840 }, { "epoch": 0.81, "learning_rate": 4.402325743818969e-05, "loss": 1.9139, "step": 7850 }, { "epoch": 0.81, "learning_rate": 4.401452716859897e-05, "loss": 1.9173, "step": 7860 }, { "epoch": 0.81, "learning_rate": 4.4005796899008245e-05, "loss": 1.8155, "step": 7870 }, { "epoch": 0.81, "learning_rate": 4.399793965637659e-05, "loss": 2.0311, "step": 7880 }, { "epoch": 0.81, "learning_rate": 4.3989209386785866e-05, "loss": 1.8039, "step": 7890 }, { "epoch": 0.81, "learning_rate": 4.398047911719514e-05, "loss": 1.9292, "step": 7900 }, { "epoch": 0.81, "learning_rate": 4.397174884760442e-05, "loss": 1.826, "step": 7910 }, { "epoch": 0.82, "learning_rate": 4.396301857801369e-05, "loss": 1.8845, "step": 7920 }, { "epoch": 0.82, "learning_rate": 4.3954288308422966e-05, "loss": 1.8567, "step": 7930 }, { "epoch": 0.82, "learning_rate": 4.394555803883224e-05, "loss": 1.8737, "step": 7940 }, { "epoch": 0.82, "learning_rate": 4.393682776924152e-05, "loss": 1.9106, "step": 7950 }, { "epoch": 0.82, "learning_rate": 4.3928097499650786e-05, "loss": 2.0107, "step": 7960 }, { "epoch": 0.82, "learning_rate": 4.3919367230060066e-05, "loss": 2.0222, "step": 7970 }, { "epoch": 0.82, "learning_rate": 4.3910636960469346e-05, "loss": 1.8418, "step": 7980 }, { "epoch": 0.82, "learning_rate": 4.390190669087862e-05, "loss": 1.8249, "step": 7990 }, { "epoch": 0.82, "learning_rate": 4.389317642128789e-05, "loss": 1.9456, "step": 8000 }, { "epoch": 0.82, "learning_rate": 4.3884446151697165e-05, "loss": 1.8933, "step": 8010 }, { "epoch": 0.83, "learning_rate": 4.3875715882106445e-05, "loss": 1.9091, "step": 8020 }, { "epoch": 0.83, "learning_rate": 4.386698561251571e-05, "loss": 1.837, "step": 8030 }, { "epoch": 0.83, "learning_rate": 4.385825534292499e-05, "loss": 1.8042, "step": 8040 }, { "epoch": 0.83, "learning_rate": 4.384952507333427e-05, "loss": 1.9415, "step": 8050 }, { "epoch": 0.83, "learning_rate": 4.384079480374354e-05, "loss": 1.8731, "step": 8060 }, { "epoch": 0.83, "learning_rate": 4.383206453415282e-05, "loss": 1.8279, "step": 8070 }, { "epoch": 0.83, "learning_rate": 4.382333426456209e-05, "loss": 1.9256, "step": 8080 }, { "epoch": 0.83, "learning_rate": 4.3814603994971365e-05, "loss": 1.967, "step": 8090 }, { "epoch": 0.83, "learning_rate": 4.380587372538064e-05, "loss": 1.6856, "step": 8100 }, { "epoch": 0.84, "learning_rate": 4.379714345578992e-05, "loss": 1.9908, "step": 8110 }, { "epoch": 0.84, "learning_rate": 4.378841318619919e-05, "loss": 1.9756, "step": 8120 }, { "epoch": 0.84, "learning_rate": 4.3779682916608465e-05, "loss": 1.8669, "step": 8130 }, { "epoch": 0.84, "learning_rate": 4.3770952647017745e-05, "loss": 1.8277, "step": 8140 }, { "epoch": 0.84, "learning_rate": 4.376222237742701e-05, "loss": 1.8518, "step": 8150 }, { "epoch": 0.84, "learning_rate": 4.375349210783629e-05, "loss": 1.9726, "step": 8160 }, { "epoch": 0.84, "learning_rate": 4.3744761838245565e-05, "loss": 1.766, "step": 8170 }, { "epoch": 0.84, "learning_rate": 4.373603156865484e-05, "loss": 1.8517, "step": 8180 }, { "epoch": 0.84, "learning_rate": 4.372730129906412e-05, "loss": 1.8879, "step": 8190 }, { "epoch": 0.84, "learning_rate": 4.371857102947339e-05, "loss": 2.0617, "step": 8200 }, { "epoch": 0.85, "learning_rate": 4.370984075988267e-05, "loss": 1.9032, "step": 8210 }, { "epoch": 0.85, "learning_rate": 4.370111049029194e-05, "loss": 2.014, "step": 8220 }, { "epoch": 0.85, "learning_rate": 4.369238022070122e-05, "loss": 1.9015, "step": 8230 }, { "epoch": 0.85, "learning_rate": 4.368364995111049e-05, "loss": 1.9622, "step": 8240 }, { "epoch": 0.85, "learning_rate": 4.3674919681519765e-05, "loss": 1.8902, "step": 8250 }, { "epoch": 0.85, "learning_rate": 4.3666189411929045e-05, "loss": 1.9431, "step": 8260 }, { "epoch": 0.85, "learning_rate": 4.365745914233832e-05, "loss": 1.9136, "step": 8270 }, { "epoch": 0.85, "learning_rate": 4.364872887274759e-05, "loss": 1.8425, "step": 8280 }, { "epoch": 0.85, "learning_rate": 4.3639998603156864e-05, "loss": 2.0659, "step": 8290 }, { "epoch": 0.85, "learning_rate": 4.3631268333566145e-05, "loss": 1.9682, "step": 8300 }, { "epoch": 0.86, "learning_rate": 4.362253806397542e-05, "loss": 1.7982, "step": 8310 }, { "epoch": 0.86, "learning_rate": 4.361380779438469e-05, "loss": 1.8961, "step": 8320 }, { "epoch": 0.86, "learning_rate": 4.360507752479397e-05, "loss": 2.1465, "step": 8330 }, { "epoch": 0.86, "learning_rate": 4.359634725520324e-05, "loss": 1.8898, "step": 8340 }, { "epoch": 0.86, "learning_rate": 4.358761698561252e-05, "loss": 1.8611, "step": 8350 }, { "epoch": 0.86, "learning_rate": 4.357888671602179e-05, "loss": 1.7984, "step": 8360 }, { "epoch": 0.86, "learning_rate": 4.3570156446431064e-05, "loss": 1.8857, "step": 8370 }, { "epoch": 0.86, "learning_rate": 4.3561426176840344e-05, "loss": 1.9614, "step": 8380 }, { "epoch": 0.86, "learning_rate": 4.355269590724962e-05, "loss": 1.8525, "step": 8390 }, { "epoch": 0.86, "learning_rate": 4.35439656376589e-05, "loss": 1.8673, "step": 8400 }, { "epoch": 0.87, "learning_rate": 4.3535235368068164e-05, "loss": 1.7879, "step": 8410 }, { "epoch": 0.87, "learning_rate": 4.3526505098477444e-05, "loss": 1.8082, "step": 8420 }, { "epoch": 0.87, "learning_rate": 4.351777482888672e-05, "loss": 1.7898, "step": 8430 }, { "epoch": 0.87, "learning_rate": 4.350904455929599e-05, "loss": 1.923, "step": 8440 }, { "epoch": 0.87, "learning_rate": 4.350031428970527e-05, "loss": 1.7423, "step": 8450 }, { "epoch": 0.87, "learning_rate": 4.3491584020114544e-05, "loss": 1.8416, "step": 8460 }, { "epoch": 0.87, "learning_rate": 4.348285375052382e-05, "loss": 1.9893, "step": 8470 }, { "epoch": 0.87, "learning_rate": 4.347412348093309e-05, "loss": 1.8634, "step": 8480 }, { "epoch": 0.87, "learning_rate": 4.346539321134237e-05, "loss": 1.9571, "step": 8490 }, { "epoch": 0.88, "learning_rate": 4.3456662941751644e-05, "loss": 1.901, "step": 8500 }, { "epoch": 0.88, "learning_rate": 4.344793267216092e-05, "loss": 1.8699, "step": 8510 }, { "epoch": 0.88, "learning_rate": 4.34392024025702e-05, "loss": 1.8812, "step": 8520 }, { "epoch": 0.88, "learning_rate": 4.3430472132979464e-05, "loss": 1.9219, "step": 8530 }, { "epoch": 0.88, "learning_rate": 4.3421741863388744e-05, "loss": 1.8457, "step": 8540 }, { "epoch": 0.88, "learning_rate": 4.341301159379802e-05, "loss": 1.9833, "step": 8550 }, { "epoch": 0.88, "learning_rate": 4.340428132420729e-05, "loss": 1.9811, "step": 8560 }, { "epoch": 0.88, "learning_rate": 4.339555105461657e-05, "loss": 1.8513, "step": 8570 }, { "epoch": 0.88, "learning_rate": 4.3386820785025844e-05, "loss": 1.9744, "step": 8580 }, { "epoch": 0.88, "learning_rate": 4.3378090515435124e-05, "loss": 1.84, "step": 8590 }, { "epoch": 0.89, "learning_rate": 4.336936024584439e-05, "loss": 1.5563, "step": 8600 }, { "epoch": 0.89, "learning_rate": 4.336062997625367e-05, "loss": 1.8902, "step": 8610 }, { "epoch": 0.89, "learning_rate": 4.335189970666294e-05, "loss": 1.8737, "step": 8620 }, { "epoch": 0.89, "learning_rate": 4.334316943707222e-05, "loss": 1.9445, "step": 8630 }, { "epoch": 0.89, "learning_rate": 4.33344391674815e-05, "loss": 1.9151, "step": 8640 }, { "epoch": 0.89, "learning_rate": 4.332570889789077e-05, "loss": 1.8131, "step": 8650 }, { "epoch": 0.89, "learning_rate": 4.331697862830004e-05, "loss": 1.8619, "step": 8660 }, { "epoch": 0.89, "learning_rate": 4.3308248358709317e-05, "loss": 1.9442, "step": 8670 }, { "epoch": 0.89, "learning_rate": 4.3299518089118597e-05, "loss": 1.8936, "step": 8680 }, { "epoch": 0.89, "learning_rate": 4.329078781952787e-05, "loss": 1.8226, "step": 8690 }, { "epoch": 0.9, "learning_rate": 4.328205754993714e-05, "loss": 2.0373, "step": 8700 }, { "epoch": 0.9, "learning_rate": 4.327332728034642e-05, "loss": 1.8952, "step": 8710 }, { "epoch": 0.9, "learning_rate": 4.326459701075569e-05, "loss": 1.9068, "step": 8720 }, { "epoch": 0.9, "learning_rate": 4.325586674116497e-05, "loss": 1.8326, "step": 8730 }, { "epoch": 0.9, "learning_rate": 4.324713647157424e-05, "loss": 1.9627, "step": 8740 }, { "epoch": 0.9, "learning_rate": 4.3238406201983516e-05, "loss": 1.919, "step": 8750 }, { "epoch": 0.9, "learning_rate": 4.3229675932392796e-05, "loss": 2.0294, "step": 8760 }, { "epoch": 0.9, "learning_rate": 4.322094566280207e-05, "loss": 1.8436, "step": 8770 }, { "epoch": 0.9, "learning_rate": 4.321221539321135e-05, "loss": 1.8034, "step": 8780 }, { "epoch": 0.91, "learning_rate": 4.3203485123620616e-05, "loss": 1.8162, "step": 8790 }, { "epoch": 0.91, "learning_rate": 4.3194754854029896e-05, "loss": 1.8468, "step": 8800 }, { "epoch": 0.91, "learning_rate": 4.318602458443917e-05, "loss": 1.7474, "step": 8810 }, { "epoch": 0.91, "learning_rate": 4.317729431484844e-05, "loss": 1.741, "step": 8820 }, { "epoch": 0.91, "learning_rate": 4.316856404525772e-05, "loss": 1.9141, "step": 8830 }, { "epoch": 0.91, "learning_rate": 4.3159833775666996e-05, "loss": 1.8607, "step": 8840 }, { "epoch": 0.91, "learning_rate": 4.315110350607627e-05, "loss": 1.8223, "step": 8850 }, { "epoch": 0.91, "learning_rate": 4.314237323648554e-05, "loss": 2.0047, "step": 8860 }, { "epoch": 0.91, "learning_rate": 4.313364296689482e-05, "loss": 1.8995, "step": 8870 }, { "epoch": 0.91, "learning_rate": 4.3124912697304096e-05, "loss": 1.8478, "step": 8880 }, { "epoch": 0.92, "learning_rate": 4.311618242771337e-05, "loss": 2.0053, "step": 8890 }, { "epoch": 0.92, "learning_rate": 4.310745215812265e-05, "loss": 1.7061, "step": 8900 }, { "epoch": 0.92, "learning_rate": 4.3098721888531916e-05, "loss": 1.8929, "step": 8910 }, { "epoch": 0.92, "learning_rate": 4.3089991618941196e-05, "loss": 1.7674, "step": 8920 }, { "epoch": 0.92, "learning_rate": 4.308126134935047e-05, "loss": 1.8123, "step": 8930 }, { "epoch": 0.92, "learning_rate": 4.307253107975974e-05, "loss": 1.8774, "step": 8940 }, { "epoch": 0.92, "learning_rate": 4.306380081016902e-05, "loss": 1.8441, "step": 8950 }, { "epoch": 0.92, "learning_rate": 4.3055070540578296e-05, "loss": 1.9632, "step": 8960 }, { "epoch": 0.92, "learning_rate": 4.304634027098757e-05, "loss": 1.8782, "step": 8970 }, { "epoch": 0.92, "learning_rate": 4.303761000139684e-05, "loss": 1.8817, "step": 8980 }, { "epoch": 0.93, "learning_rate": 4.302887973180612e-05, "loss": 2.05, "step": 8990 }, { "epoch": 0.93, "learning_rate": 4.3020149462215395e-05, "loss": 1.8287, "step": 9000 }, { "epoch": 0.93, "learning_rate": 4.301141919262467e-05, "loss": 2.0142, "step": 9010 }, { "epoch": 0.93, "learning_rate": 4.300268892303395e-05, "loss": 2.0055, "step": 9020 }, { "epoch": 0.93, "learning_rate": 4.299395865344322e-05, "loss": 1.6785, "step": 9030 }, { "epoch": 0.93, "learning_rate": 4.2985228383852495e-05, "loss": 1.8379, "step": 9040 }, { "epoch": 0.93, "learning_rate": 4.297649811426177e-05, "loss": 1.8608, "step": 9050 }, { "epoch": 0.93, "learning_rate": 4.296776784467105e-05, "loss": 1.887, "step": 9060 }, { "epoch": 0.93, "learning_rate": 4.295903757508032e-05, "loss": 1.867, "step": 9070 }, { "epoch": 0.93, "learning_rate": 4.2950307305489595e-05, "loss": 1.8419, "step": 9080 }, { "epoch": 0.94, "learning_rate": 4.2941577035898875e-05, "loss": 1.9248, "step": 9090 }, { "epoch": 0.94, "learning_rate": 4.293284676630814e-05, "loss": 1.8193, "step": 9100 }, { "epoch": 0.94, "learning_rate": 4.292411649671742e-05, "loss": 1.8485, "step": 9110 }, { "epoch": 0.94, "learning_rate": 4.2915386227126695e-05, "loss": 1.9335, "step": 9120 }, { "epoch": 0.94, "learning_rate": 4.290665595753597e-05, "loss": 1.8016, "step": 9130 }, { "epoch": 0.94, "learning_rate": 4.289792568794525e-05, "loss": 1.8523, "step": 9140 }, { "epoch": 0.94, "learning_rate": 4.288919541835452e-05, "loss": 1.9271, "step": 9150 }, { "epoch": 0.94, "learning_rate": 4.2880465148763795e-05, "loss": 1.8492, "step": 9160 }, { "epoch": 0.94, "learning_rate": 4.287173487917307e-05, "loss": 1.9555, "step": 9170 }, { "epoch": 0.95, "learning_rate": 4.286300460958235e-05, "loss": 1.8619, "step": 9180 }, { "epoch": 0.95, "learning_rate": 4.285427433999162e-05, "loss": 1.8998, "step": 9190 }, { "epoch": 0.95, "learning_rate": 4.2845544070400895e-05, "loss": 1.8354, "step": 9200 }, { "epoch": 0.95, "learning_rate": 4.2836813800810175e-05, "loss": 1.9187, "step": 9210 }, { "epoch": 0.95, "learning_rate": 4.282808353121945e-05, "loss": 1.8662, "step": 9220 }, { "epoch": 0.95, "learning_rate": 4.281935326162872e-05, "loss": 1.9838, "step": 9230 }, { "epoch": 0.95, "learning_rate": 4.2810622992037995e-05, "loss": 1.8133, "step": 9240 }, { "epoch": 0.95, "learning_rate": 4.2801892722447275e-05, "loss": 1.8465, "step": 9250 }, { "epoch": 0.95, "learning_rate": 4.279316245285654e-05, "loss": 1.7488, "step": 9260 }, { "epoch": 0.95, "learning_rate": 4.278443218326582e-05, "loss": 1.7855, "step": 9270 }, { "epoch": 0.96, "learning_rate": 4.27757019136751e-05, "loss": 1.8665, "step": 9280 }, { "epoch": 0.96, "learning_rate": 4.276697164408437e-05, "loss": 1.7769, "step": 9290 }, { "epoch": 0.96, "learning_rate": 4.275824137449365e-05, "loss": 1.996, "step": 9300 }, { "epoch": 0.96, "learning_rate": 4.274951110490292e-05, "loss": 1.8218, "step": 9310 }, { "epoch": 0.96, "learning_rate": 4.2740780835312194e-05, "loss": 2.0463, "step": 9320 }, { "epoch": 0.96, "learning_rate": 4.273205056572147e-05, "loss": 1.874, "step": 9330 }, { "epoch": 0.96, "learning_rate": 4.272332029613075e-05, "loss": 1.9101, "step": 9340 }, { "epoch": 0.96, "learning_rate": 4.271459002654002e-05, "loss": 1.8118, "step": 9350 }, { "epoch": 0.96, "learning_rate": 4.2705859756949294e-05, "loss": 1.7913, "step": 9360 }, { "epoch": 0.96, "learning_rate": 4.2697129487358574e-05, "loss": 2.0015, "step": 9370 }, { "epoch": 0.97, "learning_rate": 4.268839921776785e-05, "loss": 1.8567, "step": 9380 }, { "epoch": 0.97, "learning_rate": 4.267966894817712e-05, "loss": 1.9197, "step": 9390 }, { "epoch": 0.97, "learning_rate": 4.2670938678586394e-05, "loss": 2.0934, "step": 9400 }, { "epoch": 0.97, "learning_rate": 4.2662208408995674e-05, "loss": 1.8394, "step": 9410 }, { "epoch": 0.97, "learning_rate": 4.265347813940495e-05, "loss": 1.8115, "step": 9420 }, { "epoch": 0.97, "learning_rate": 4.264474786981422e-05, "loss": 1.8204, "step": 9430 }, { "epoch": 0.97, "learning_rate": 4.26360176002235e-05, "loss": 2.0474, "step": 9440 }, { "epoch": 0.97, "learning_rate": 4.262728733063277e-05, "loss": 1.9147, "step": 9450 }, { "epoch": 0.97, "learning_rate": 4.261855706104205e-05, "loss": 1.9168, "step": 9460 }, { "epoch": 0.98, "learning_rate": 4.260982679145132e-05, "loss": 1.7557, "step": 9470 }, { "epoch": 0.98, "learning_rate": 4.2601096521860594e-05, "loss": 2.0177, "step": 9480 }, { "epoch": 0.98, "learning_rate": 4.2592366252269874e-05, "loss": 1.8579, "step": 9490 }, { "epoch": 0.98, "learning_rate": 4.258363598267915e-05, "loss": 1.8935, "step": 9500 }, { "epoch": 0.98, "learning_rate": 4.257490571308842e-05, "loss": 1.8755, "step": 9510 }, { "epoch": 0.98, "learning_rate": 4.2566175443497694e-05, "loss": 2.0003, "step": 9520 }, { "epoch": 0.98, "learning_rate": 4.2557445173906974e-05, "loss": 1.8433, "step": 9530 }, { "epoch": 0.98, "learning_rate": 4.254871490431625e-05, "loss": 1.9238, "step": 9540 }, { "epoch": 0.98, "learning_rate": 4.253998463472552e-05, "loss": 1.8563, "step": 9550 }, { "epoch": 0.98, "learning_rate": 4.25312543651348e-05, "loss": 1.8431, "step": 9560 }, { "epoch": 0.99, "learning_rate": 4.2522524095544074e-05, "loss": 2.0862, "step": 9570 }, { "epoch": 0.99, "learning_rate": 4.251379382595335e-05, "loss": 1.8584, "step": 9580 }, { "epoch": 0.99, "learning_rate": 4.250506355636262e-05, "loss": 1.8315, "step": 9590 }, { "epoch": 0.99, "learning_rate": 4.24963332867719e-05, "loss": 1.8362, "step": 9600 }, { "epoch": 0.99, "learning_rate": 4.2487603017181173e-05, "loss": 1.8479, "step": 9610 }, { "epoch": 0.99, "learning_rate": 4.247887274759045e-05, "loss": 1.6951, "step": 9620 }, { "epoch": 0.99, "learning_rate": 4.247014247799973e-05, "loss": 1.8114, "step": 9630 }, { "epoch": 0.99, "learning_rate": 4.246141220840899e-05, "loss": 1.9353, "step": 9640 }, { "epoch": 0.99, "learning_rate": 4.245268193881827e-05, "loss": 1.8924, "step": 9650 }, { "epoch": 0.99, "learning_rate": 4.2443951669227547e-05, "loss": 1.8605, "step": 9660 }, { "epoch": 1.0, "learning_rate": 4.243522139963682e-05, "loss": 1.954, "step": 9670 }, { "epoch": 1.0, "learning_rate": 4.24264911300461e-05, "loss": 2.0013, "step": 9680 }, { "epoch": 1.0, "learning_rate": 4.241776086045537e-05, "loss": 1.7673, "step": 9690 }, { "epoch": 1.0, "learning_rate": 4.2409030590864646e-05, "loss": 1.8135, "step": 9700 }, { "epoch": 1.0, "learning_rate": 4.240030032127392e-05, "loss": 1.922, "step": 9710 }, { "epoch": 1.0, "eval_loss": 1.8737725019454956, "eval_runtime": 988.0923, "eval_samples_per_second": 52.423, "eval_steps_per_second": 8.738, "step": 9712 } ], "logging_steps": 10, "max_steps": 58272, "num_input_tokens_seen": 0, "num_train_epochs": 6, "save_steps": 500, "total_flos": 1.21809903353856e+17, "train_batch_size": 6, "trial_name": null, "trial_params": null }