{ "best_metric": 0.3743310570716858, "best_model_checkpoint": "./ckpts/data16w/checkpoint-50000", "epoch": 2.5, "global_step": 50000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 1.0000000000000002e-06, "loss": 0.4346, "step": 10 }, { "epoch": 0.0, "learning_rate": 2.0000000000000003e-06, "loss": 0.4462, "step": 20 }, { "epoch": 0.0, "learning_rate": 3e-06, "loss": 0.3209, "step": 30 }, { "epoch": 0.0, "learning_rate": 4.000000000000001e-06, "loss": 0.3884, "step": 40 }, { "epoch": 0.0, "learning_rate": 5e-06, "loss": 0.5262, "step": 50 }, { "epoch": 0.0, "learning_rate": 6e-06, "loss": 0.3986, "step": 60 }, { "epoch": 0.0, "learning_rate": 7e-06, "loss": 0.4484, "step": 70 }, { "epoch": 0.0, "learning_rate": 8.000000000000001e-06, "loss": 0.4513, "step": 80 }, { "epoch": 0.0, "learning_rate": 9e-06, "loss": 0.4256, "step": 90 }, { "epoch": 0.01, "learning_rate": 1e-05, "loss": 0.4378, "step": 100 }, { "epoch": 0.01, "learning_rate": 9.998330550918199e-06, "loss": 0.4124, "step": 110 }, { "epoch": 0.01, "learning_rate": 9.996661101836395e-06, "loss": 0.3757, "step": 120 }, { "epoch": 0.01, "learning_rate": 9.994991652754593e-06, "loss": 0.39, "step": 130 }, { "epoch": 0.01, "learning_rate": 9.993322203672789e-06, "loss": 0.4449, "step": 140 }, { "epoch": 0.01, "learning_rate": 9.991652754590987e-06, "loss": 0.4217, "step": 150 }, { "epoch": 0.01, "learning_rate": 9.989983305509183e-06, "loss": 0.3719, "step": 160 }, { "epoch": 0.01, "learning_rate": 9.988313856427379e-06, "loss": 0.3855, "step": 170 }, { "epoch": 0.01, "learning_rate": 9.986644407345577e-06, "loss": 0.3889, "step": 180 }, { "epoch": 0.01, "learning_rate": 9.984974958263773e-06, "loss": 0.398, "step": 190 }, { "epoch": 0.01, "learning_rate": 9.98330550918197e-06, "loss": 0.3607, "step": 200 }, { "epoch": 0.01, "learning_rate": 9.981636060100169e-06, "loss": 0.322, "step": 210 }, { "epoch": 0.01, "learning_rate": 9.979966611018365e-06, "loss": 0.3954, "step": 220 }, { "epoch": 0.01, "learning_rate": 9.978297161936563e-06, "loss": 0.3775, "step": 230 }, { "epoch": 0.01, "learning_rate": 9.976627712854759e-06, "loss": 0.3705, "step": 240 }, { "epoch": 0.01, "learning_rate": 9.974958263772957e-06, "loss": 0.4326, "step": 250 }, { "epoch": 0.01, "learning_rate": 9.973288814691153e-06, "loss": 0.4553, "step": 260 }, { "epoch": 0.01, "learning_rate": 9.971619365609349e-06, "loss": 0.3977, "step": 270 }, { "epoch": 0.01, "learning_rate": 9.969949916527547e-06, "loss": 0.3343, "step": 280 }, { "epoch": 0.01, "learning_rate": 9.968280467445743e-06, "loss": 0.3811, "step": 290 }, { "epoch": 0.01, "learning_rate": 9.96661101836394e-06, "loss": 0.3596, "step": 300 }, { "epoch": 0.02, "learning_rate": 9.964941569282137e-06, "loss": 0.4308, "step": 310 }, { "epoch": 0.02, "learning_rate": 9.963272120200335e-06, "loss": 0.469, "step": 320 }, { "epoch": 0.02, "learning_rate": 9.961602671118533e-06, "loss": 0.4217, "step": 330 }, { "epoch": 0.02, "learning_rate": 9.959933222036729e-06, "loss": 0.3621, "step": 340 }, { "epoch": 0.02, "learning_rate": 9.958263772954927e-06, "loss": 0.475, "step": 350 }, { "epoch": 0.02, "learning_rate": 9.956594323873123e-06, "loss": 0.4325, "step": 360 }, { "epoch": 0.02, "learning_rate": 9.954924874791319e-06, "loss": 0.4182, "step": 370 }, { "epoch": 0.02, "learning_rate": 9.953255425709517e-06, "loss": 0.5118, "step": 380 }, { "epoch": 0.02, "learning_rate": 9.951585976627713e-06, "loss": 0.4111, "step": 390 }, { "epoch": 0.02, "learning_rate": 9.94991652754591e-06, "loss": 0.4386, "step": 400 }, { "epoch": 0.02, "learning_rate": 9.948247078464107e-06, "loss": 0.3924, "step": 410 }, { "epoch": 0.02, "learning_rate": 9.946577629382305e-06, "loss": 0.3344, "step": 420 }, { "epoch": 0.02, "learning_rate": 9.944908180300503e-06, "loss": 0.4, "step": 430 }, { "epoch": 0.02, "learning_rate": 9.943238731218699e-06, "loss": 0.4807, "step": 440 }, { "epoch": 0.02, "learning_rate": 9.941569282136897e-06, "loss": 0.381, "step": 450 }, { "epoch": 0.02, "learning_rate": 9.939899833055093e-06, "loss": 0.3609, "step": 460 }, { "epoch": 0.02, "learning_rate": 9.938230383973289e-06, "loss": 0.3849, "step": 470 }, { "epoch": 0.02, "learning_rate": 9.936560934891487e-06, "loss": 0.4883, "step": 480 }, { "epoch": 0.02, "learning_rate": 9.934891485809683e-06, "loss": 0.4457, "step": 490 }, { "epoch": 0.03, "learning_rate": 9.93322203672788e-06, "loss": 0.3828, "step": 500 }, { "epoch": 0.03, "learning_rate": 9.931552587646077e-06, "loss": 0.3498, "step": 510 }, { "epoch": 0.03, "learning_rate": 9.929883138564275e-06, "loss": 0.4257, "step": 520 }, { "epoch": 0.03, "learning_rate": 9.928213689482473e-06, "loss": 0.3254, "step": 530 }, { "epoch": 0.03, "learning_rate": 9.926544240400669e-06, "loss": 0.4689, "step": 540 }, { "epoch": 0.03, "learning_rate": 9.924874791318867e-06, "loss": 0.3629, "step": 550 }, { "epoch": 0.03, "learning_rate": 9.923205342237063e-06, "loss": 0.337, "step": 560 }, { "epoch": 0.03, "learning_rate": 9.921535893155259e-06, "loss": 0.4516, "step": 570 }, { "epoch": 0.03, "learning_rate": 9.919866444073457e-06, "loss": 0.3497, "step": 580 }, { "epoch": 0.03, "learning_rate": 9.918196994991653e-06, "loss": 0.3964, "step": 590 }, { "epoch": 0.03, "learning_rate": 9.91652754590985e-06, "loss": 0.3621, "step": 600 }, { "epoch": 0.03, "learning_rate": 9.914858096828047e-06, "loss": 0.3481, "step": 610 }, { "epoch": 0.03, "learning_rate": 9.913188647746245e-06, "loss": 0.32, "step": 620 }, { "epoch": 0.03, "learning_rate": 9.911519198664441e-06, "loss": 0.4328, "step": 630 }, { "epoch": 0.03, "learning_rate": 9.909849749582639e-06, "loss": 0.3475, "step": 640 }, { "epoch": 0.03, "learning_rate": 9.908180300500837e-06, "loss": 0.2832, "step": 650 }, { "epoch": 0.03, "learning_rate": 9.906510851419033e-06, "loss": 0.3624, "step": 660 }, { "epoch": 0.03, "learning_rate": 9.904841402337229e-06, "loss": 0.3638, "step": 670 }, { "epoch": 0.03, "learning_rate": 9.903171953255427e-06, "loss": 0.3364, "step": 680 }, { "epoch": 0.03, "learning_rate": 9.901502504173623e-06, "loss": 0.3605, "step": 690 }, { "epoch": 0.04, "learning_rate": 9.89983305509182e-06, "loss": 0.4309, "step": 700 }, { "epoch": 0.04, "learning_rate": 9.898163606010017e-06, "loss": 0.445, "step": 710 }, { "epoch": 0.04, "learning_rate": 9.896494156928215e-06, "loss": 0.3697, "step": 720 }, { "epoch": 0.04, "learning_rate": 9.894824707846411e-06, "loss": 0.4142, "step": 730 }, { "epoch": 0.04, "learning_rate": 9.893155258764609e-06, "loss": 0.3743, "step": 740 }, { "epoch": 0.04, "learning_rate": 9.891485809682807e-06, "loss": 0.3666, "step": 750 }, { "epoch": 0.04, "learning_rate": 9.889816360601003e-06, "loss": 0.3419, "step": 760 }, { "epoch": 0.04, "learning_rate": 9.888146911519199e-06, "loss": 0.3657, "step": 770 }, { "epoch": 0.04, "learning_rate": 9.886477462437397e-06, "loss": 0.387, "step": 780 }, { "epoch": 0.04, "learning_rate": 9.884808013355593e-06, "loss": 0.4437, "step": 790 }, { "epoch": 0.04, "learning_rate": 9.88313856427379e-06, "loss": 0.3514, "step": 800 }, { "epoch": 0.04, "learning_rate": 9.881469115191987e-06, "loss": 0.3239, "step": 810 }, { "epoch": 0.04, "learning_rate": 9.879799666110185e-06, "loss": 0.3664, "step": 820 }, { "epoch": 0.04, "learning_rate": 9.87813021702838e-06, "loss": 0.404, "step": 830 }, { "epoch": 0.04, "learning_rate": 9.876460767946579e-06, "loss": 0.4315, "step": 840 }, { "epoch": 0.04, "learning_rate": 9.874791318864777e-06, "loss": 0.485, "step": 850 }, { "epoch": 0.04, "learning_rate": 9.873121869782973e-06, "loss": 0.331, "step": 860 }, { "epoch": 0.04, "learning_rate": 9.87145242070117e-06, "loss": 0.366, "step": 870 }, { "epoch": 0.04, "learning_rate": 9.869782971619367e-06, "loss": 0.405, "step": 880 }, { "epoch": 0.04, "learning_rate": 9.868113522537563e-06, "loss": 0.4084, "step": 890 }, { "epoch": 0.04, "learning_rate": 9.86644407345576e-06, "loss": 0.4118, "step": 900 }, { "epoch": 0.05, "learning_rate": 9.864774624373957e-06, "loss": 0.3917, "step": 910 }, { "epoch": 0.05, "learning_rate": 9.863105175292155e-06, "loss": 0.4873, "step": 920 }, { "epoch": 0.05, "learning_rate": 9.86143572621035e-06, "loss": 0.3567, "step": 930 }, { "epoch": 0.05, "learning_rate": 9.859766277128549e-06, "loss": 0.4795, "step": 940 }, { "epoch": 0.05, "learning_rate": 9.858096828046745e-06, "loss": 0.3196, "step": 950 }, { "epoch": 0.05, "learning_rate": 9.856427378964943e-06, "loss": 0.3385, "step": 960 }, { "epoch": 0.05, "learning_rate": 9.85475792988314e-06, "loss": 0.3848, "step": 970 }, { "epoch": 0.05, "learning_rate": 9.853088480801337e-06, "loss": 0.3151, "step": 980 }, { "epoch": 0.05, "learning_rate": 9.851419031719533e-06, "loss": 0.3538, "step": 990 }, { "epoch": 0.05, "learning_rate": 9.84974958263773e-06, "loss": 0.3775, "step": 1000 }, { "epoch": 0.05, "learning_rate": 9.848080133555927e-06, "loss": 0.378, "step": 1010 }, { "epoch": 0.05, "learning_rate": 9.846410684474125e-06, "loss": 0.3572, "step": 1020 }, { "epoch": 0.05, "learning_rate": 9.84474123539232e-06, "loss": 0.3251, "step": 1030 }, { "epoch": 0.05, "learning_rate": 9.843071786310519e-06, "loss": 0.5266, "step": 1040 }, { "epoch": 0.05, "learning_rate": 9.841402337228715e-06, "loss": 0.4361, "step": 1050 }, { "epoch": 0.05, "learning_rate": 9.839732888146913e-06, "loss": 0.3545, "step": 1060 }, { "epoch": 0.05, "learning_rate": 9.83806343906511e-06, "loss": 0.3398, "step": 1070 }, { "epoch": 0.05, "learning_rate": 9.836393989983307e-06, "loss": 0.3479, "step": 1080 }, { "epoch": 0.05, "learning_rate": 9.834724540901503e-06, "loss": 0.4108, "step": 1090 }, { "epoch": 0.06, "learning_rate": 9.8330550918197e-06, "loss": 0.4536, "step": 1100 }, { "epoch": 0.06, "learning_rate": 9.831385642737897e-06, "loss": 0.3986, "step": 1110 }, { "epoch": 0.06, "learning_rate": 9.829716193656095e-06, "loss": 0.3545, "step": 1120 }, { "epoch": 0.06, "learning_rate": 9.82804674457429e-06, "loss": 0.3376, "step": 1130 }, { "epoch": 0.06, "learning_rate": 9.826377295492489e-06, "loss": 0.4275, "step": 1140 }, { "epoch": 0.06, "learning_rate": 9.824707846410685e-06, "loss": 0.333, "step": 1150 }, { "epoch": 0.06, "learning_rate": 9.823038397328883e-06, "loss": 0.475, "step": 1160 }, { "epoch": 0.06, "learning_rate": 9.82136894824708e-06, "loss": 0.4489, "step": 1170 }, { "epoch": 0.06, "learning_rate": 9.819699499165277e-06, "loss": 0.3987, "step": 1180 }, { "epoch": 0.06, "learning_rate": 9.818030050083473e-06, "loss": 0.4106, "step": 1190 }, { "epoch": 0.06, "learning_rate": 9.81636060100167e-06, "loss": 0.4176, "step": 1200 }, { "epoch": 0.06, "learning_rate": 9.814691151919867e-06, "loss": 0.4274, "step": 1210 }, { "epoch": 0.06, "learning_rate": 9.813021702838065e-06, "loss": 0.4589, "step": 1220 }, { "epoch": 0.06, "learning_rate": 9.81135225375626e-06, "loss": 0.4216, "step": 1230 }, { "epoch": 0.06, "learning_rate": 9.809682804674459e-06, "loss": 0.4067, "step": 1240 }, { "epoch": 0.06, "learning_rate": 9.808013355592655e-06, "loss": 0.4365, "step": 1250 }, { "epoch": 0.06, "learning_rate": 9.806343906510853e-06, "loss": 0.3773, "step": 1260 }, { "epoch": 0.06, "learning_rate": 9.804674457429049e-06, "loss": 0.4406, "step": 1270 }, { "epoch": 0.06, "learning_rate": 9.803005008347247e-06, "loss": 0.3539, "step": 1280 }, { "epoch": 0.06, "learning_rate": 9.801335559265443e-06, "loss": 0.3742, "step": 1290 }, { "epoch": 0.07, "learning_rate": 9.79966611018364e-06, "loss": 0.4103, "step": 1300 }, { "epoch": 0.07, "learning_rate": 9.797996661101837e-06, "loss": 0.3172, "step": 1310 }, { "epoch": 0.07, "learning_rate": 9.796327212020033e-06, "loss": 0.315, "step": 1320 }, { "epoch": 0.07, "learning_rate": 9.79465776293823e-06, "loss": 0.3044, "step": 1330 }, { "epoch": 0.07, "learning_rate": 9.792988313856429e-06, "loss": 0.2941, "step": 1340 }, { "epoch": 0.07, "learning_rate": 9.791318864774625e-06, "loss": 0.3591, "step": 1350 }, { "epoch": 0.07, "learning_rate": 9.789649415692823e-06, "loss": 0.3842, "step": 1360 }, { "epoch": 0.07, "learning_rate": 9.787979966611019e-06, "loss": 0.3454, "step": 1370 }, { "epoch": 0.07, "learning_rate": 9.786310517529217e-06, "loss": 0.3384, "step": 1380 }, { "epoch": 0.07, "learning_rate": 9.784641068447413e-06, "loss": 0.4354, "step": 1390 }, { "epoch": 0.07, "learning_rate": 9.78297161936561e-06, "loss": 0.3785, "step": 1400 }, { "epoch": 0.07, "learning_rate": 9.781302170283807e-06, "loss": 0.4548, "step": 1410 }, { "epoch": 0.07, "learning_rate": 9.779632721202003e-06, "loss": 0.3692, "step": 1420 }, { "epoch": 0.07, "learning_rate": 9.7779632721202e-06, "loss": 0.4343, "step": 1430 }, { "epoch": 0.07, "learning_rate": 9.776293823038399e-06, "loss": 0.459, "step": 1440 }, { "epoch": 0.07, "learning_rate": 9.774624373956595e-06, "loss": 0.4532, "step": 1450 }, { "epoch": 0.07, "learning_rate": 9.772954924874793e-06, "loss": 0.4742, "step": 1460 }, { "epoch": 0.07, "learning_rate": 9.771285475792989e-06, "loss": 0.3373, "step": 1470 }, { "epoch": 0.07, "learning_rate": 9.769616026711187e-06, "loss": 0.3735, "step": 1480 }, { "epoch": 0.07, "learning_rate": 9.767946577629384e-06, "loss": 0.4415, "step": 1490 }, { "epoch": 0.07, "learning_rate": 9.76627712854758e-06, "loss": 0.305, "step": 1500 }, { "epoch": 0.08, "learning_rate": 9.764607679465777e-06, "loss": 0.4132, "step": 1510 }, { "epoch": 0.08, "learning_rate": 9.762938230383973e-06, "loss": 0.3876, "step": 1520 }, { "epoch": 0.08, "learning_rate": 9.76126878130217e-06, "loss": 0.3868, "step": 1530 }, { "epoch": 0.08, "learning_rate": 9.759599332220369e-06, "loss": 0.4031, "step": 1540 }, { "epoch": 0.08, "learning_rate": 9.757929883138565e-06, "loss": 0.3142, "step": 1550 }, { "epoch": 0.08, "learning_rate": 9.756260434056763e-06, "loss": 0.3824, "step": 1560 }, { "epoch": 0.08, "learning_rate": 9.754590984974959e-06, "loss": 0.3541, "step": 1570 }, { "epoch": 0.08, "learning_rate": 9.752921535893157e-06, "loss": 0.4137, "step": 1580 }, { "epoch": 0.08, "learning_rate": 9.751252086811354e-06, "loss": 0.3964, "step": 1590 }, { "epoch": 0.08, "learning_rate": 9.74958263772955e-06, "loss": 0.4046, "step": 1600 }, { "epoch": 0.08, "learning_rate": 9.747913188647747e-06, "loss": 0.4158, "step": 1610 }, { "epoch": 0.08, "learning_rate": 9.746243739565945e-06, "loss": 0.4556, "step": 1620 }, { "epoch": 0.08, "learning_rate": 9.74457429048414e-06, "loss": 0.4058, "step": 1630 }, { "epoch": 0.08, "learning_rate": 9.742904841402337e-06, "loss": 0.3957, "step": 1640 }, { "epoch": 0.08, "learning_rate": 9.741235392320535e-06, "loss": 0.3763, "step": 1650 }, { "epoch": 0.08, "learning_rate": 9.739565943238733e-06, "loss": 0.4035, "step": 1660 }, { "epoch": 0.08, "learning_rate": 9.737896494156929e-06, "loss": 0.4071, "step": 1670 }, { "epoch": 0.08, "learning_rate": 9.736227045075127e-06, "loss": 0.3333, "step": 1680 }, { "epoch": 0.08, "learning_rate": 9.734557595993323e-06, "loss": 0.4687, "step": 1690 }, { "epoch": 0.09, "learning_rate": 9.73288814691152e-06, "loss": 0.3769, "step": 1700 }, { "epoch": 0.09, "learning_rate": 9.731218697829717e-06, "loss": 0.4384, "step": 1710 }, { "epoch": 0.09, "learning_rate": 9.729549248747915e-06, "loss": 0.4467, "step": 1720 }, { "epoch": 0.09, "learning_rate": 9.72787979966611e-06, "loss": 0.3002, "step": 1730 }, { "epoch": 0.09, "learning_rate": 9.726210350584307e-06, "loss": 0.4834, "step": 1740 }, { "epoch": 0.09, "learning_rate": 9.724540901502505e-06, "loss": 0.3851, "step": 1750 }, { "epoch": 0.09, "learning_rate": 9.722871452420703e-06, "loss": 0.4071, "step": 1760 }, { "epoch": 0.09, "learning_rate": 9.721202003338899e-06, "loss": 0.3744, "step": 1770 }, { "epoch": 0.09, "learning_rate": 9.719532554257097e-06, "loss": 0.3327, "step": 1780 }, { "epoch": 0.09, "learning_rate": 9.717863105175293e-06, "loss": 0.466, "step": 1790 }, { "epoch": 0.09, "learning_rate": 9.71619365609349e-06, "loss": 0.2973, "step": 1800 }, { "epoch": 0.09, "learning_rate": 9.714524207011687e-06, "loss": 0.4529, "step": 1810 }, { "epoch": 0.09, "learning_rate": 9.712854757929885e-06, "loss": 0.4172, "step": 1820 }, { "epoch": 0.09, "learning_rate": 9.71118530884808e-06, "loss": 0.478, "step": 1830 }, { "epoch": 0.09, "learning_rate": 9.709515859766277e-06, "loss": 0.3613, "step": 1840 }, { "epoch": 0.09, "learning_rate": 9.707846410684475e-06, "loss": 0.4658, "step": 1850 }, { "epoch": 0.09, "learning_rate": 9.706176961602673e-06, "loss": 0.4392, "step": 1860 }, { "epoch": 0.09, "learning_rate": 9.704507512520869e-06, "loss": 0.3718, "step": 1870 }, { "epoch": 0.09, "learning_rate": 9.702838063439067e-06, "loss": 0.4501, "step": 1880 }, { "epoch": 0.09, "learning_rate": 9.701168614357263e-06, "loss": 0.3388, "step": 1890 }, { "epoch": 0.1, "learning_rate": 9.69949916527546e-06, "loss": 0.3285, "step": 1900 }, { "epoch": 0.1, "learning_rate": 9.697829716193657e-06, "loss": 0.3483, "step": 1910 }, { "epoch": 0.1, "learning_rate": 9.696160267111854e-06, "loss": 0.3549, "step": 1920 }, { "epoch": 0.1, "learning_rate": 9.69449081803005e-06, "loss": 0.4357, "step": 1930 }, { "epoch": 0.1, "learning_rate": 9.692821368948247e-06, "loss": 0.3721, "step": 1940 }, { "epoch": 0.1, "learning_rate": 9.691151919866445e-06, "loss": 0.2865, "step": 1950 }, { "epoch": 0.1, "learning_rate": 9.68948247078464e-06, "loss": 0.3578, "step": 1960 }, { "epoch": 0.1, "learning_rate": 9.687813021702839e-06, "loss": 0.4302, "step": 1970 }, { "epoch": 0.1, "learning_rate": 9.686143572621036e-06, "loss": 0.3371, "step": 1980 }, { "epoch": 0.1, "learning_rate": 9.684474123539233e-06, "loss": 0.424, "step": 1990 }, { "epoch": 0.1, "learning_rate": 9.68280467445743e-06, "loss": 0.4102, "step": 2000 }, { "epoch": 0.1, "learning_rate": 9.681135225375627e-06, "loss": 0.3641, "step": 2010 }, { "epoch": 0.1, "learning_rate": 9.679465776293824e-06, "loss": 0.3917, "step": 2020 }, { "epoch": 0.1, "learning_rate": 9.67779632721202e-06, "loss": 0.3775, "step": 2030 }, { "epoch": 0.1, "learning_rate": 9.676126878130217e-06, "loss": 0.3672, "step": 2040 }, { "epoch": 0.1, "learning_rate": 9.674457429048415e-06, "loss": 0.4315, "step": 2050 }, { "epoch": 0.1, "learning_rate": 9.67278797996661e-06, "loss": 0.3493, "step": 2060 }, { "epoch": 0.1, "learning_rate": 9.671118530884809e-06, "loss": 0.4483, "step": 2070 }, { "epoch": 0.1, "learning_rate": 9.669449081803006e-06, "loss": 0.4572, "step": 2080 }, { "epoch": 0.1, "learning_rate": 9.667779632721203e-06, "loss": 0.453, "step": 2090 }, { "epoch": 0.1, "learning_rate": 9.6661101836394e-06, "loss": 0.366, "step": 2100 }, { "epoch": 0.11, "learning_rate": 9.664440734557597e-06, "loss": 0.3141, "step": 2110 }, { "epoch": 0.11, "learning_rate": 9.662771285475794e-06, "loss": 0.3622, "step": 2120 }, { "epoch": 0.11, "learning_rate": 9.66110183639399e-06, "loss": 0.4209, "step": 2130 }, { "epoch": 0.11, "learning_rate": 9.659432387312187e-06, "loss": 0.4764, "step": 2140 }, { "epoch": 0.11, "learning_rate": 9.657762938230385e-06, "loss": 0.3859, "step": 2150 }, { "epoch": 0.11, "learning_rate": 9.65609348914858e-06, "loss": 0.3547, "step": 2160 }, { "epoch": 0.11, "learning_rate": 9.654424040066779e-06, "loss": 0.409, "step": 2170 }, { "epoch": 0.11, "learning_rate": 9.652754590984976e-06, "loss": 0.2936, "step": 2180 }, { "epoch": 0.11, "learning_rate": 9.651085141903173e-06, "loss": 0.3565, "step": 2190 }, { "epoch": 0.11, "learning_rate": 9.64941569282137e-06, "loss": 0.3161, "step": 2200 }, { "epoch": 0.11, "learning_rate": 9.647746243739567e-06, "loss": 0.3599, "step": 2210 }, { "epoch": 0.11, "learning_rate": 9.646076794657764e-06, "loss": 0.4074, "step": 2220 }, { "epoch": 0.11, "learning_rate": 9.64440734557596e-06, "loss": 0.3987, "step": 2230 }, { "epoch": 0.11, "learning_rate": 9.642737896494157e-06, "loss": 0.3839, "step": 2240 }, { "epoch": 0.11, "learning_rate": 9.641068447412355e-06, "loss": 0.3859, "step": 2250 }, { "epoch": 0.11, "learning_rate": 9.63939899833055e-06, "loss": 0.3318, "step": 2260 }, { "epoch": 0.11, "learning_rate": 9.637729549248749e-06, "loss": 0.3342, "step": 2270 }, { "epoch": 0.11, "learning_rate": 9.636060100166945e-06, "loss": 0.3628, "step": 2280 }, { "epoch": 0.11, "learning_rate": 9.634390651085143e-06, "loss": 0.3836, "step": 2290 }, { "epoch": 0.12, "learning_rate": 9.63272120200334e-06, "loss": 0.5687, "step": 2300 }, { "epoch": 0.12, "learning_rate": 9.631051752921537e-06, "loss": 0.3914, "step": 2310 }, { "epoch": 0.12, "learning_rate": 9.629382303839734e-06, "loss": 0.4327, "step": 2320 }, { "epoch": 0.12, "learning_rate": 9.62771285475793e-06, "loss": 0.4236, "step": 2330 }, { "epoch": 0.12, "learning_rate": 9.626043405676128e-06, "loss": 0.3729, "step": 2340 }, { "epoch": 0.12, "learning_rate": 9.624373956594325e-06, "loss": 0.3996, "step": 2350 }, { "epoch": 0.12, "learning_rate": 9.62270450751252e-06, "loss": 0.3428, "step": 2360 }, { "epoch": 0.12, "learning_rate": 9.621035058430719e-06, "loss": 0.3604, "step": 2370 }, { "epoch": 0.12, "learning_rate": 9.619365609348915e-06, "loss": 0.4302, "step": 2380 }, { "epoch": 0.12, "learning_rate": 9.617696160267113e-06, "loss": 0.3493, "step": 2390 }, { "epoch": 0.12, "learning_rate": 9.61602671118531e-06, "loss": 0.4407, "step": 2400 }, { "epoch": 0.12, "learning_rate": 9.614357262103507e-06, "loss": 0.4749, "step": 2410 }, { "epoch": 0.12, "learning_rate": 9.612687813021704e-06, "loss": 0.4103, "step": 2420 }, { "epoch": 0.12, "learning_rate": 9.6110183639399e-06, "loss": 0.4477, "step": 2430 }, { "epoch": 0.12, "learning_rate": 9.609348914858098e-06, "loss": 0.3393, "step": 2440 }, { "epoch": 0.12, "learning_rate": 9.607679465776295e-06, "loss": 0.3098, "step": 2450 }, { "epoch": 0.12, "learning_rate": 9.60601001669449e-06, "loss": 0.3985, "step": 2460 }, { "epoch": 0.12, "learning_rate": 9.604340567612689e-06, "loss": 0.473, "step": 2470 }, { "epoch": 0.12, "learning_rate": 9.602671118530885e-06, "loss": 0.3964, "step": 2480 }, { "epoch": 0.12, "learning_rate": 9.601001669449083e-06, "loss": 0.3468, "step": 2490 }, { "epoch": 0.12, "learning_rate": 9.59933222036728e-06, "loss": 0.3554, "step": 2500 }, { "epoch": 0.13, "learning_rate": 9.597662771285477e-06, "loss": 0.3723, "step": 2510 }, { "epoch": 0.13, "learning_rate": 9.595993322203674e-06, "loss": 0.3997, "step": 2520 }, { "epoch": 0.13, "learning_rate": 9.59432387312187e-06, "loss": 0.3704, "step": 2530 }, { "epoch": 0.13, "learning_rate": 9.592654424040068e-06, "loss": 0.4261, "step": 2540 }, { "epoch": 0.13, "learning_rate": 9.590984974958265e-06, "loss": 0.3671, "step": 2550 }, { "epoch": 0.13, "learning_rate": 9.58931552587646e-06, "loss": 0.4097, "step": 2560 }, { "epoch": 0.13, "learning_rate": 9.587646076794659e-06, "loss": 0.3356, "step": 2570 }, { "epoch": 0.13, "learning_rate": 9.585976627712855e-06, "loss": 0.4001, "step": 2580 }, { "epoch": 0.13, "learning_rate": 9.584307178631053e-06, "loss": 0.3468, "step": 2590 }, { "epoch": 0.13, "learning_rate": 9.582637729549249e-06, "loss": 0.4, "step": 2600 }, { "epoch": 0.13, "learning_rate": 9.580968280467447e-06, "loss": 0.3686, "step": 2610 }, { "epoch": 0.13, "learning_rate": 9.579298831385644e-06, "loss": 0.3061, "step": 2620 }, { "epoch": 0.13, "learning_rate": 9.57762938230384e-06, "loss": 0.4056, "step": 2630 }, { "epoch": 0.13, "learning_rate": 9.575959933222038e-06, "loss": 0.3628, "step": 2640 }, { "epoch": 0.13, "learning_rate": 9.574290484140235e-06, "loss": 0.375, "step": 2650 }, { "epoch": 0.13, "learning_rate": 9.57262103505843e-06, "loss": 0.4212, "step": 2660 }, { "epoch": 0.13, "learning_rate": 9.570951585976629e-06, "loss": 0.3766, "step": 2670 }, { "epoch": 0.13, "learning_rate": 9.569282136894825e-06, "loss": 0.3603, "step": 2680 }, { "epoch": 0.13, "learning_rate": 9.567612687813023e-06, "loss": 0.3263, "step": 2690 }, { "epoch": 0.14, "learning_rate": 9.565943238731219e-06, "loss": 0.3713, "step": 2700 }, { "epoch": 0.14, "learning_rate": 9.564273789649417e-06, "loss": 0.338, "step": 2710 }, { "epoch": 0.14, "learning_rate": 9.562604340567614e-06, "loss": 0.4392, "step": 2720 }, { "epoch": 0.14, "learning_rate": 9.56093489148581e-06, "loss": 0.4178, "step": 2730 }, { "epoch": 0.14, "learning_rate": 9.559265442404008e-06, "loss": 0.3607, "step": 2740 }, { "epoch": 0.14, "learning_rate": 9.557595993322205e-06, "loss": 0.4272, "step": 2750 }, { "epoch": 0.14, "learning_rate": 9.5559265442404e-06, "loss": 0.3745, "step": 2760 }, { "epoch": 0.14, "learning_rate": 9.554257095158599e-06, "loss": 0.2873, "step": 2770 }, { "epoch": 0.14, "learning_rate": 9.552587646076795e-06, "loss": 0.3651, "step": 2780 }, { "epoch": 0.14, "learning_rate": 9.550918196994992e-06, "loss": 0.3835, "step": 2790 }, { "epoch": 0.14, "learning_rate": 9.549248747913189e-06, "loss": 0.3861, "step": 2800 }, { "epoch": 0.14, "learning_rate": 9.547579298831386e-06, "loss": 0.4442, "step": 2810 }, { "epoch": 0.14, "learning_rate": 9.545909849749584e-06, "loss": 0.4129, "step": 2820 }, { "epoch": 0.14, "learning_rate": 9.54424040066778e-06, "loss": 0.4261, "step": 2830 }, { "epoch": 0.14, "learning_rate": 9.542570951585978e-06, "loss": 0.4369, "step": 2840 }, { "epoch": 0.14, "learning_rate": 9.540901502504174e-06, "loss": 0.3768, "step": 2850 }, { "epoch": 0.14, "learning_rate": 9.53923205342237e-06, "loss": 0.4903, "step": 2860 }, { "epoch": 0.14, "learning_rate": 9.537562604340568e-06, "loss": 0.3493, "step": 2870 }, { "epoch": 0.14, "learning_rate": 9.535893155258765e-06, "loss": 0.4546, "step": 2880 }, { "epoch": 0.14, "learning_rate": 9.534223706176962e-06, "loss": 0.3615, "step": 2890 }, { "epoch": 0.14, "learning_rate": 9.532554257095159e-06, "loss": 0.4213, "step": 2900 }, { "epoch": 0.15, "learning_rate": 9.530884808013356e-06, "loss": 0.3456, "step": 2910 }, { "epoch": 0.15, "learning_rate": 9.529215358931554e-06, "loss": 0.3085, "step": 2920 }, { "epoch": 0.15, "learning_rate": 9.52754590984975e-06, "loss": 0.3992, "step": 2930 }, { "epoch": 0.15, "learning_rate": 9.525876460767948e-06, "loss": 0.4202, "step": 2940 }, { "epoch": 0.15, "learning_rate": 9.524207011686144e-06, "loss": 0.4084, "step": 2950 }, { "epoch": 0.15, "learning_rate": 9.522537562604342e-06, "loss": 0.5437, "step": 2960 }, { "epoch": 0.15, "learning_rate": 9.520868113522538e-06, "loss": 0.4364, "step": 2970 }, { "epoch": 0.15, "learning_rate": 9.519198664440735e-06, "loss": 0.341, "step": 2980 }, { "epoch": 0.15, "learning_rate": 9.517529215358932e-06, "loss": 0.4135, "step": 2990 }, { "epoch": 0.15, "learning_rate": 9.515859766277129e-06, "loss": 0.3587, "step": 3000 }, { "epoch": 0.15, "learning_rate": 9.514190317195326e-06, "loss": 0.3784, "step": 3010 }, { "epoch": 0.15, "learning_rate": 9.512520868113523e-06, "loss": 0.3249, "step": 3020 }, { "epoch": 0.15, "learning_rate": 9.51085141903172e-06, "loss": 0.3764, "step": 3030 }, { "epoch": 0.15, "learning_rate": 9.509181969949918e-06, "loss": 0.4178, "step": 3040 }, { "epoch": 0.15, "learning_rate": 9.507512520868114e-06, "loss": 0.4343, "step": 3050 }, { "epoch": 0.15, "learning_rate": 9.505843071786312e-06, "loss": 0.3563, "step": 3060 }, { "epoch": 0.15, "learning_rate": 9.504173622704508e-06, "loss": 0.3605, "step": 3070 }, { "epoch": 0.15, "learning_rate": 9.502504173622705e-06, "loss": 0.3177, "step": 3080 }, { "epoch": 0.15, "learning_rate": 9.500834724540902e-06, "loss": 0.4098, "step": 3090 }, { "epoch": 0.15, "learning_rate": 9.499165275459099e-06, "loss": 0.3975, "step": 3100 }, { "epoch": 0.16, "learning_rate": 9.497495826377296e-06, "loss": 0.5106, "step": 3110 }, { "epoch": 0.16, "learning_rate": 9.495826377295493e-06, "loss": 0.4458, "step": 3120 }, { "epoch": 0.16, "learning_rate": 9.49415692821369e-06, "loss": 0.3493, "step": 3130 }, { "epoch": 0.16, "learning_rate": 9.492487479131888e-06, "loss": 0.3944, "step": 3140 }, { "epoch": 0.16, "learning_rate": 9.490818030050084e-06, "loss": 0.4057, "step": 3150 }, { "epoch": 0.16, "learning_rate": 9.489148580968282e-06, "loss": 0.4086, "step": 3160 }, { "epoch": 0.16, "learning_rate": 9.487479131886478e-06, "loss": 0.3395, "step": 3170 }, { "epoch": 0.16, "learning_rate": 9.485809682804675e-06, "loss": 0.3657, "step": 3180 }, { "epoch": 0.16, "learning_rate": 9.484140233722872e-06, "loss": 0.4398, "step": 3190 }, { "epoch": 0.16, "learning_rate": 9.482470784641069e-06, "loss": 0.3751, "step": 3200 }, { "epoch": 0.16, "learning_rate": 9.480801335559266e-06, "loss": 0.2973, "step": 3210 }, { "epoch": 0.16, "learning_rate": 9.479131886477463e-06, "loss": 0.3839, "step": 3220 }, { "epoch": 0.16, "learning_rate": 9.47746243739566e-06, "loss": 0.4686, "step": 3230 }, { "epoch": 0.16, "learning_rate": 9.475792988313858e-06, "loss": 0.4548, "step": 3240 }, { "epoch": 0.16, "learning_rate": 9.474123539232054e-06, "loss": 0.4277, "step": 3250 }, { "epoch": 0.16, "learning_rate": 9.472454090150252e-06, "loss": 0.3584, "step": 3260 }, { "epoch": 0.16, "learning_rate": 9.470784641068448e-06, "loss": 0.3089, "step": 3270 }, { "epoch": 0.16, "learning_rate": 9.469115191986645e-06, "loss": 0.396, "step": 3280 }, { "epoch": 0.16, "learning_rate": 9.467445742904842e-06, "loss": 0.3941, "step": 3290 }, { "epoch": 0.17, "learning_rate": 9.465776293823039e-06, "loss": 0.4473, "step": 3300 }, { "epoch": 0.17, "learning_rate": 9.464106844741236e-06, "loss": 0.3454, "step": 3310 }, { "epoch": 0.17, "learning_rate": 9.462437395659433e-06, "loss": 0.4387, "step": 3320 }, { "epoch": 0.17, "learning_rate": 9.46076794657763e-06, "loss": 0.4139, "step": 3330 }, { "epoch": 0.17, "learning_rate": 9.459098497495827e-06, "loss": 0.4143, "step": 3340 }, { "epoch": 0.17, "learning_rate": 9.457429048414024e-06, "loss": 0.3367, "step": 3350 }, { "epoch": 0.17, "learning_rate": 9.455759599332222e-06, "loss": 0.3632, "step": 3360 }, { "epoch": 0.17, "learning_rate": 9.454090150250418e-06, "loss": 0.4102, "step": 3370 }, { "epoch": 0.17, "learning_rate": 9.452420701168615e-06, "loss": 0.3728, "step": 3380 }, { "epoch": 0.17, "learning_rate": 9.450751252086812e-06, "loss": 0.3998, "step": 3390 }, { "epoch": 0.17, "learning_rate": 9.449081803005009e-06, "loss": 0.3586, "step": 3400 }, { "epoch": 0.17, "learning_rate": 9.447412353923206e-06, "loss": 0.4588, "step": 3410 }, { "epoch": 0.17, "learning_rate": 9.445742904841403e-06, "loss": 0.3215, "step": 3420 }, { "epoch": 0.17, "learning_rate": 9.4440734557596e-06, "loss": 0.3891, "step": 3430 }, { "epoch": 0.17, "learning_rate": 9.442404006677797e-06, "loss": 0.3069, "step": 3440 }, { "epoch": 0.17, "learning_rate": 9.440734557595994e-06, "loss": 0.3553, "step": 3450 }, { "epoch": 0.17, "learning_rate": 9.439065108514192e-06, "loss": 0.3733, "step": 3460 }, { "epoch": 0.17, "learning_rate": 9.437395659432388e-06, "loss": 0.482, "step": 3470 }, { "epoch": 0.17, "learning_rate": 9.435726210350585e-06, "loss": 0.4172, "step": 3480 }, { "epoch": 0.17, "learning_rate": 9.434056761268782e-06, "loss": 0.526, "step": 3490 }, { "epoch": 0.17, "learning_rate": 9.432387312186979e-06, "loss": 0.4318, "step": 3500 }, { "epoch": 0.18, "learning_rate": 9.430717863105176e-06, "loss": 0.3619, "step": 3510 }, { "epoch": 0.18, "learning_rate": 9.429048414023373e-06, "loss": 0.4083, "step": 3520 }, { "epoch": 0.18, "learning_rate": 9.42737896494157e-06, "loss": 0.3467, "step": 3530 }, { "epoch": 0.18, "learning_rate": 9.425709515859767e-06, "loss": 0.4622, "step": 3540 }, { "epoch": 0.18, "learning_rate": 9.424040066777964e-06, "loss": 0.4247, "step": 3550 }, { "epoch": 0.18, "learning_rate": 9.422370617696162e-06, "loss": 0.4011, "step": 3560 }, { "epoch": 0.18, "learning_rate": 9.420701168614358e-06, "loss": 0.3379, "step": 3570 }, { "epoch": 0.18, "learning_rate": 9.419031719532555e-06, "loss": 0.4564, "step": 3580 }, { "epoch": 0.18, "learning_rate": 9.417362270450752e-06, "loss": 0.3251, "step": 3590 }, { "epoch": 0.18, "learning_rate": 9.415692821368949e-06, "loss": 0.4567, "step": 3600 }, { "epoch": 0.18, "learning_rate": 9.414023372287145e-06, "loss": 0.4898, "step": 3610 }, { "epoch": 0.18, "learning_rate": 9.412353923205343e-06, "loss": 0.41, "step": 3620 }, { "epoch": 0.18, "learning_rate": 9.41068447412354e-06, "loss": 0.296, "step": 3630 }, { "epoch": 0.18, "learning_rate": 9.409015025041737e-06, "loss": 0.4059, "step": 3640 }, { "epoch": 0.18, "learning_rate": 9.407345575959934e-06, "loss": 0.4133, "step": 3650 }, { "epoch": 0.18, "learning_rate": 9.40567612687813e-06, "loss": 0.4807, "step": 3660 }, { "epoch": 0.18, "learning_rate": 9.404006677796328e-06, "loss": 0.4793, "step": 3670 }, { "epoch": 0.18, "learning_rate": 9.402337228714526e-06, "loss": 0.4148, "step": 3680 }, { "epoch": 0.18, "learning_rate": 9.400667779632722e-06, "loss": 0.4234, "step": 3690 }, { "epoch": 0.18, "learning_rate": 9.398998330550918e-06, "loss": 0.3576, "step": 3700 }, { "epoch": 0.19, "learning_rate": 9.397328881469115e-06, "loss": 0.4143, "step": 3710 }, { "epoch": 0.19, "learning_rate": 9.395659432387312e-06, "loss": 0.4759, "step": 3720 }, { "epoch": 0.19, "learning_rate": 9.39398998330551e-06, "loss": 0.3479, "step": 3730 }, { "epoch": 0.19, "learning_rate": 9.392320534223706e-06, "loss": 0.4433, "step": 3740 }, { "epoch": 0.19, "learning_rate": 9.390651085141904e-06, "loss": 0.3177, "step": 3750 }, { "epoch": 0.19, "learning_rate": 9.3889816360601e-06, "loss": 0.4676, "step": 3760 }, { "epoch": 0.19, "learning_rate": 9.387312186978298e-06, "loss": 0.3939, "step": 3770 }, { "epoch": 0.19, "learning_rate": 9.385642737896496e-06, "loss": 0.4276, "step": 3780 }, { "epoch": 0.19, "learning_rate": 9.383973288814692e-06, "loss": 0.4486, "step": 3790 }, { "epoch": 0.19, "learning_rate": 9.382303839732888e-06, "loss": 0.4387, "step": 3800 }, { "epoch": 0.19, "learning_rate": 9.380634390651086e-06, "loss": 0.316, "step": 3810 }, { "epoch": 0.19, "learning_rate": 9.378964941569282e-06, "loss": 0.3855, "step": 3820 }, { "epoch": 0.19, "learning_rate": 9.37729549248748e-06, "loss": 0.388, "step": 3830 }, { "epoch": 0.19, "learning_rate": 9.375626043405676e-06, "loss": 0.3671, "step": 3840 }, { "epoch": 0.19, "learning_rate": 9.373956594323874e-06, "loss": 0.3813, "step": 3850 }, { "epoch": 0.19, "learning_rate": 9.37228714524207e-06, "loss": 0.4307, "step": 3860 }, { "epoch": 0.19, "learning_rate": 9.370617696160268e-06, "loss": 0.3063, "step": 3870 }, { "epoch": 0.19, "learning_rate": 9.368948247078466e-06, "loss": 0.4377, "step": 3880 }, { "epoch": 0.19, "learning_rate": 9.367278797996662e-06, "loss": 0.5302, "step": 3890 }, { "epoch": 0.2, "learning_rate": 9.365609348914858e-06, "loss": 0.3092, "step": 3900 }, { "epoch": 0.2, "learning_rate": 9.363939899833056e-06, "loss": 0.4461, "step": 3910 }, { "epoch": 0.2, "learning_rate": 9.362270450751252e-06, "loss": 0.3861, "step": 3920 }, { "epoch": 0.2, "learning_rate": 9.360601001669449e-06, "loss": 0.3756, "step": 3930 }, { "epoch": 0.2, "learning_rate": 9.358931552587646e-06, "loss": 0.4534, "step": 3940 }, { "epoch": 0.2, "learning_rate": 9.357262103505844e-06, "loss": 0.3684, "step": 3950 }, { "epoch": 0.2, "learning_rate": 9.35559265442404e-06, "loss": 0.4235, "step": 3960 }, { "epoch": 0.2, "learning_rate": 9.353923205342238e-06, "loss": 0.462, "step": 3970 }, { "epoch": 0.2, "learning_rate": 9.352253756260434e-06, "loss": 0.3618, "step": 3980 }, { "epoch": 0.2, "learning_rate": 9.350584307178632e-06, "loss": 0.4829, "step": 3990 }, { "epoch": 0.2, "learning_rate": 9.348914858096828e-06, "loss": 0.3754, "step": 4000 }, { "epoch": 0.2, "learning_rate": 9.347245409015026e-06, "loss": 0.4127, "step": 4010 }, { "epoch": 0.2, "learning_rate": 9.345575959933222e-06, "loss": 0.4026, "step": 4020 }, { "epoch": 0.2, "learning_rate": 9.343906510851419e-06, "loss": 0.3005, "step": 4030 }, { "epoch": 0.2, "learning_rate": 9.342237061769616e-06, "loss": 0.3757, "step": 4040 }, { "epoch": 0.2, "learning_rate": 9.340567612687814e-06, "loss": 0.4869, "step": 4050 }, { "epoch": 0.2, "learning_rate": 9.33889816360601e-06, "loss": 0.324, "step": 4060 }, { "epoch": 0.2, "learning_rate": 9.337228714524208e-06, "loss": 0.4448, "step": 4070 }, { "epoch": 0.2, "learning_rate": 9.335559265442404e-06, "loss": 0.3272, "step": 4080 }, { "epoch": 0.2, "learning_rate": 9.333889816360602e-06, "loss": 0.3747, "step": 4090 }, { "epoch": 0.2, "learning_rate": 9.332220367278798e-06, "loss": 0.3268, "step": 4100 }, { "epoch": 0.21, "learning_rate": 9.330550918196996e-06, "loss": 0.3702, "step": 4110 }, { "epoch": 0.21, "learning_rate": 9.328881469115192e-06, "loss": 0.3374, "step": 4120 }, { "epoch": 0.21, "learning_rate": 9.327212020033389e-06, "loss": 0.3608, "step": 4130 }, { "epoch": 0.21, "learning_rate": 9.325542570951586e-06, "loss": 0.4347, "step": 4140 }, { "epoch": 0.21, "learning_rate": 9.323873121869784e-06, "loss": 0.3786, "step": 4150 }, { "epoch": 0.21, "learning_rate": 9.32220367278798e-06, "loss": 0.4807, "step": 4160 }, { "epoch": 0.21, "learning_rate": 9.320534223706178e-06, "loss": 0.394, "step": 4170 }, { "epoch": 0.21, "learning_rate": 9.318864774624374e-06, "loss": 0.4282, "step": 4180 }, { "epoch": 0.21, "learning_rate": 9.317195325542572e-06, "loss": 0.4116, "step": 4190 }, { "epoch": 0.21, "learning_rate": 9.315525876460768e-06, "loss": 0.5318, "step": 4200 }, { "epoch": 0.21, "learning_rate": 9.313856427378966e-06, "loss": 0.3414, "step": 4210 }, { "epoch": 0.21, "learning_rate": 9.312186978297162e-06, "loss": 0.3707, "step": 4220 }, { "epoch": 0.21, "learning_rate": 9.310517529215359e-06, "loss": 0.3849, "step": 4230 }, { "epoch": 0.21, "learning_rate": 9.308848080133556e-06, "loss": 0.4114, "step": 4240 }, { "epoch": 0.21, "learning_rate": 9.307178631051754e-06, "loss": 0.3718, "step": 4250 }, { "epoch": 0.21, "learning_rate": 9.30550918196995e-06, "loss": 0.434, "step": 4260 }, { "epoch": 0.21, "learning_rate": 9.303839732888148e-06, "loss": 0.3864, "step": 4270 }, { "epoch": 0.21, "learning_rate": 9.302170283806344e-06, "loss": 0.3876, "step": 4280 }, { "epoch": 0.21, "learning_rate": 9.300500834724542e-06, "loss": 0.3106, "step": 4290 }, { "epoch": 0.21, "learning_rate": 9.298831385642738e-06, "loss": 0.3659, "step": 4300 }, { "epoch": 0.22, "learning_rate": 9.297161936560936e-06, "loss": 0.4361, "step": 4310 }, { "epoch": 0.22, "learning_rate": 9.295492487479132e-06, "loss": 0.3095, "step": 4320 }, { "epoch": 0.22, "learning_rate": 9.293823038397329e-06, "loss": 0.3463, "step": 4330 }, { "epoch": 0.22, "learning_rate": 9.292153589315526e-06, "loss": 0.5455, "step": 4340 }, { "epoch": 0.22, "learning_rate": 9.290484140233723e-06, "loss": 0.3133, "step": 4350 }, { "epoch": 0.22, "learning_rate": 9.28881469115192e-06, "loss": 0.4017, "step": 4360 }, { "epoch": 0.22, "learning_rate": 9.287145242070118e-06, "loss": 0.4289, "step": 4370 }, { "epoch": 0.22, "learning_rate": 9.285475792988314e-06, "loss": 0.3581, "step": 4380 }, { "epoch": 0.22, "learning_rate": 9.283806343906512e-06, "loss": 0.4248, "step": 4390 }, { "epoch": 0.22, "learning_rate": 9.282136894824708e-06, "loss": 0.4316, "step": 4400 }, { "epoch": 0.22, "learning_rate": 9.280467445742906e-06, "loss": 0.379, "step": 4410 }, { "epoch": 0.22, "learning_rate": 9.278797996661102e-06, "loss": 0.3679, "step": 4420 }, { "epoch": 0.22, "learning_rate": 9.2771285475793e-06, "loss": 0.3809, "step": 4430 }, { "epoch": 0.22, "learning_rate": 9.275459098497496e-06, "loss": 0.4165, "step": 4440 }, { "epoch": 0.22, "learning_rate": 9.273789649415693e-06, "loss": 0.408, "step": 4450 }, { "epoch": 0.22, "learning_rate": 9.27212020033389e-06, "loss": 0.4073, "step": 4460 }, { "epoch": 0.22, "learning_rate": 9.270450751252088e-06, "loss": 0.4213, "step": 4470 }, { "epoch": 0.22, "learning_rate": 9.268781302170284e-06, "loss": 0.4194, "step": 4480 }, { "epoch": 0.22, "learning_rate": 9.267111853088482e-06, "loss": 0.3645, "step": 4490 }, { "epoch": 0.23, "learning_rate": 9.265442404006678e-06, "loss": 0.3778, "step": 4500 }, { "epoch": 0.23, "learning_rate": 9.263772954924876e-06, "loss": 0.3676, "step": 4510 }, { "epoch": 0.23, "learning_rate": 9.262103505843072e-06, "loss": 0.3798, "step": 4520 }, { "epoch": 0.23, "learning_rate": 9.26043405676127e-06, "loss": 0.475, "step": 4530 }, { "epoch": 0.23, "learning_rate": 9.258764607679466e-06, "loss": 0.2769, "step": 4540 }, { "epoch": 0.23, "learning_rate": 9.257095158597662e-06, "loss": 0.3899, "step": 4550 }, { "epoch": 0.23, "learning_rate": 9.25542570951586e-06, "loss": 0.4051, "step": 4560 }, { "epoch": 0.23, "learning_rate": 9.253756260434058e-06, "loss": 0.4379, "step": 4570 }, { "epoch": 0.23, "learning_rate": 9.252086811352254e-06, "loss": 0.3613, "step": 4580 }, { "epoch": 0.23, "learning_rate": 9.250417362270452e-06, "loss": 0.4704, "step": 4590 }, { "epoch": 0.23, "learning_rate": 9.248747913188648e-06, "loss": 0.5157, "step": 4600 }, { "epoch": 0.23, "learning_rate": 9.247078464106846e-06, "loss": 0.3678, "step": 4610 }, { "epoch": 0.23, "learning_rate": 9.245409015025042e-06, "loss": 0.4324, "step": 4620 }, { "epoch": 0.23, "learning_rate": 9.24373956594324e-06, "loss": 0.3806, "step": 4630 }, { "epoch": 0.23, "learning_rate": 9.242070116861436e-06, "loss": 0.4353, "step": 4640 }, { "epoch": 0.23, "learning_rate": 9.240400667779632e-06, "loss": 0.4278, "step": 4650 }, { "epoch": 0.23, "learning_rate": 9.23873121869783e-06, "loss": 0.4127, "step": 4660 }, { "epoch": 0.23, "learning_rate": 9.237061769616026e-06, "loss": 0.5162, "step": 4670 }, { "epoch": 0.23, "learning_rate": 9.235392320534224e-06, "loss": 0.3079, "step": 4680 }, { "epoch": 0.23, "learning_rate": 9.233722871452422e-06, "loss": 0.4205, "step": 4690 }, { "epoch": 0.23, "learning_rate": 9.232053422370618e-06, "loss": 0.5113, "step": 4700 }, { "epoch": 0.24, "learning_rate": 9.230383973288816e-06, "loss": 0.4706, "step": 4710 }, { "epoch": 0.24, "learning_rate": 9.228714524207012e-06, "loss": 0.4264, "step": 4720 }, { "epoch": 0.24, "learning_rate": 9.22704507512521e-06, "loss": 0.4563, "step": 4730 }, { "epoch": 0.24, "learning_rate": 9.225375626043406e-06, "loss": 0.4002, "step": 4740 }, { "epoch": 0.24, "learning_rate": 9.223706176961602e-06, "loss": 0.4761, "step": 4750 }, { "epoch": 0.24, "learning_rate": 9.2220367278798e-06, "loss": 0.329, "step": 4760 }, { "epoch": 0.24, "learning_rate": 9.220367278797996e-06, "loss": 0.3932, "step": 4770 }, { "epoch": 0.24, "learning_rate": 9.218697829716194e-06, "loss": 0.4374, "step": 4780 }, { "epoch": 0.24, "learning_rate": 9.217028380634392e-06, "loss": 0.3981, "step": 4790 }, { "epoch": 0.24, "learning_rate": 9.215358931552588e-06, "loss": 0.3942, "step": 4800 }, { "epoch": 0.24, "learning_rate": 9.213689482470786e-06, "loss": 0.4286, "step": 4810 }, { "epoch": 0.24, "learning_rate": 9.212020033388982e-06, "loss": 0.3436, "step": 4820 }, { "epoch": 0.24, "learning_rate": 9.21035058430718e-06, "loss": 0.4074, "step": 4830 }, { "epoch": 0.24, "learning_rate": 9.208681135225376e-06, "loss": 0.3741, "step": 4840 }, { "epoch": 0.24, "learning_rate": 9.207011686143572e-06, "loss": 0.513, "step": 4850 }, { "epoch": 0.24, "learning_rate": 9.20534223706177e-06, "loss": 0.3802, "step": 4860 }, { "epoch": 0.24, "learning_rate": 9.203672787979966e-06, "loss": 0.3856, "step": 4870 }, { "epoch": 0.24, "learning_rate": 9.202003338898164e-06, "loss": 0.4007, "step": 4880 }, { "epoch": 0.24, "learning_rate": 9.200333889816362e-06, "loss": 0.4212, "step": 4890 }, { "epoch": 0.24, "learning_rate": 9.198664440734558e-06, "loss": 0.4298, "step": 4900 }, { "epoch": 0.25, "learning_rate": 9.196994991652756e-06, "loss": 0.3938, "step": 4910 }, { "epoch": 0.25, "learning_rate": 9.195325542570952e-06, "loss": 0.4458, "step": 4920 }, { "epoch": 0.25, "learning_rate": 9.19365609348915e-06, "loss": 0.3419, "step": 4930 }, { "epoch": 0.25, "learning_rate": 9.191986644407346e-06, "loss": 0.3897, "step": 4940 }, { "epoch": 0.25, "learning_rate": 9.190317195325542e-06, "loss": 0.344, "step": 4950 }, { "epoch": 0.25, "learning_rate": 9.18864774624374e-06, "loss": 0.4935, "step": 4960 }, { "epoch": 0.25, "learning_rate": 9.186978297161936e-06, "loss": 0.3727, "step": 4970 }, { "epoch": 0.25, "learning_rate": 9.185308848080134e-06, "loss": 0.3857, "step": 4980 }, { "epoch": 0.25, "learning_rate": 9.18363939899833e-06, "loss": 0.5216, "step": 4990 }, { "epoch": 0.25, "learning_rate": 9.181969949916528e-06, "loss": 0.3755, "step": 5000 }, { "epoch": 0.25, "eval_loss": 0.3821474611759186, "eval_runtime": 185.8135, "eval_samples_per_second": 20.262, "eval_steps_per_second": 2.535, "step": 5000 }, { "epoch": 0.25, "learning_rate": 9.180300500834726e-06, "loss": 0.3991, "step": 5010 }, { "epoch": 0.25, "learning_rate": 9.178631051752922e-06, "loss": 0.3779, "step": 5020 }, { "epoch": 0.25, "learning_rate": 9.17696160267112e-06, "loss": 0.4003, "step": 5030 }, { "epoch": 0.25, "learning_rate": 9.175292153589316e-06, "loss": 0.3082, "step": 5040 }, { "epoch": 0.25, "learning_rate": 9.173622704507512e-06, "loss": 0.3554, "step": 5050 }, { "epoch": 0.25, "learning_rate": 9.17195325542571e-06, "loss": 0.3841, "step": 5060 }, { "epoch": 0.25, "learning_rate": 9.170283806343906e-06, "loss": 0.5133, "step": 5070 }, { "epoch": 0.25, "learning_rate": 9.168614357262104e-06, "loss": 0.3245, "step": 5080 }, { "epoch": 0.25, "learning_rate": 9.1669449081803e-06, "loss": 0.3656, "step": 5090 }, { "epoch": 0.26, "learning_rate": 9.165275459098498e-06, "loss": 0.4471, "step": 5100 }, { "epoch": 0.26, "learning_rate": 9.163606010016696e-06, "loss": 0.4613, "step": 5110 }, { "epoch": 0.26, "learning_rate": 9.161936560934892e-06, "loss": 0.2903, "step": 5120 }, { "epoch": 0.26, "learning_rate": 9.16026711185309e-06, "loss": 0.3658, "step": 5130 }, { "epoch": 0.26, "learning_rate": 9.158597662771286e-06, "loss": 0.3646, "step": 5140 }, { "epoch": 0.26, "learning_rate": 9.156928213689484e-06, "loss": 0.3984, "step": 5150 }, { "epoch": 0.26, "learning_rate": 9.15525876460768e-06, "loss": 0.4817, "step": 5160 }, { "epoch": 0.26, "learning_rate": 9.153589315525876e-06, "loss": 0.4178, "step": 5170 }, { "epoch": 0.26, "learning_rate": 9.151919866444074e-06, "loss": 0.3539, "step": 5180 }, { "epoch": 0.26, "learning_rate": 9.15025041736227e-06, "loss": 0.3577, "step": 5190 }, { "epoch": 0.26, "learning_rate": 9.148580968280468e-06, "loss": 0.3584, "step": 5200 }, { "epoch": 0.26, "learning_rate": 9.146911519198666e-06, "loss": 0.4204, "step": 5210 }, { "epoch": 0.26, "learning_rate": 9.145242070116862e-06, "loss": 0.5507, "step": 5220 }, { "epoch": 0.26, "learning_rate": 9.14357262103506e-06, "loss": 0.4994, "step": 5230 }, { "epoch": 0.26, "learning_rate": 9.141903171953256e-06, "loss": 0.3902, "step": 5240 }, { "epoch": 0.26, "learning_rate": 9.140233722871454e-06, "loss": 0.4573, "step": 5250 }, { "epoch": 0.26, "learning_rate": 9.13856427378965e-06, "loss": 0.4832, "step": 5260 }, { "epoch": 0.26, "learning_rate": 9.136894824707846e-06, "loss": 0.4846, "step": 5270 }, { "epoch": 0.26, "learning_rate": 9.135225375626044e-06, "loss": 0.363, "step": 5280 }, { "epoch": 0.26, "learning_rate": 9.13355592654424e-06, "loss": 0.3573, "step": 5290 }, { "epoch": 0.27, "learning_rate": 9.131886477462438e-06, "loss": 0.3823, "step": 5300 }, { "epoch": 0.27, "learning_rate": 9.130217028380634e-06, "loss": 0.3771, "step": 5310 }, { "epoch": 0.27, "learning_rate": 9.128547579298832e-06, "loss": 0.3223, "step": 5320 }, { "epoch": 0.27, "learning_rate": 9.12687813021703e-06, "loss": 0.5088, "step": 5330 }, { "epoch": 0.27, "learning_rate": 9.125208681135226e-06, "loss": 0.4316, "step": 5340 }, { "epoch": 0.27, "learning_rate": 9.123539232053424e-06, "loss": 0.3585, "step": 5350 }, { "epoch": 0.27, "learning_rate": 9.12186978297162e-06, "loss": 0.4012, "step": 5360 }, { "epoch": 0.27, "learning_rate": 9.120200333889816e-06, "loss": 0.3649, "step": 5370 }, { "epoch": 0.27, "learning_rate": 9.118530884808014e-06, "loss": 0.4087, "step": 5380 }, { "epoch": 0.27, "learning_rate": 9.11686143572621e-06, "loss": 0.3643, "step": 5390 }, { "epoch": 0.27, "learning_rate": 9.115191986644408e-06, "loss": 0.3796, "step": 5400 }, { "epoch": 0.27, "learning_rate": 9.113522537562604e-06, "loss": 0.4286, "step": 5410 }, { "epoch": 0.27, "learning_rate": 9.111853088480802e-06, "loss": 0.3962, "step": 5420 }, { "epoch": 0.27, "learning_rate": 9.110183639399e-06, "loss": 0.4233, "step": 5430 }, { "epoch": 0.27, "learning_rate": 9.108514190317196e-06, "loss": 0.3632, "step": 5440 }, { "epoch": 0.27, "learning_rate": 9.106844741235394e-06, "loss": 0.4475, "step": 5450 }, { "epoch": 0.27, "learning_rate": 9.10517529215359e-06, "loss": 0.3442, "step": 5460 }, { "epoch": 0.27, "learning_rate": 9.103505843071786e-06, "loss": 0.375, "step": 5470 }, { "epoch": 0.27, "learning_rate": 9.101836393989984e-06, "loss": 0.3247, "step": 5480 }, { "epoch": 0.27, "learning_rate": 9.10016694490818e-06, "loss": 0.4473, "step": 5490 }, { "epoch": 0.28, "learning_rate": 9.098497495826378e-06, "loss": 0.4412, "step": 5500 }, { "epoch": 0.28, "learning_rate": 9.096828046744574e-06, "loss": 0.3301, "step": 5510 }, { "epoch": 0.28, "learning_rate": 9.095158597662772e-06, "loss": 0.4139, "step": 5520 }, { "epoch": 0.28, "learning_rate": 9.09348914858097e-06, "loss": 0.3921, "step": 5530 }, { "epoch": 0.28, "learning_rate": 9.091819699499166e-06, "loss": 0.3559, "step": 5540 }, { "epoch": 0.28, "learning_rate": 9.090150250417364e-06, "loss": 0.4059, "step": 5550 }, { "epoch": 0.28, "learning_rate": 9.08848080133556e-06, "loss": 0.4352, "step": 5560 }, { "epoch": 0.28, "learning_rate": 9.086811352253756e-06, "loss": 0.452, "step": 5570 }, { "epoch": 0.28, "learning_rate": 9.085141903171954e-06, "loss": 0.3929, "step": 5580 }, { "epoch": 0.28, "learning_rate": 9.08347245409015e-06, "loss": 0.4196, "step": 5590 }, { "epoch": 0.28, "learning_rate": 9.081803005008348e-06, "loss": 0.3719, "step": 5600 }, { "epoch": 0.28, "learning_rate": 9.080133555926544e-06, "loss": 0.3668, "step": 5610 }, { "epoch": 0.28, "learning_rate": 9.078464106844742e-06, "loss": 0.4118, "step": 5620 }, { "epoch": 0.28, "learning_rate": 9.076794657762938e-06, "loss": 0.3758, "step": 5630 }, { "epoch": 0.28, "learning_rate": 9.075125208681136e-06, "loss": 0.3806, "step": 5640 }, { "epoch": 0.28, "learning_rate": 9.073455759599334e-06, "loss": 0.439, "step": 5650 }, { "epoch": 0.28, "learning_rate": 9.07178631051753e-06, "loss": 0.4133, "step": 5660 }, { "epoch": 0.28, "learning_rate": 9.070116861435726e-06, "loss": 0.405, "step": 5670 }, { "epoch": 0.28, "learning_rate": 9.068447412353924e-06, "loss": 0.404, "step": 5680 }, { "epoch": 0.28, "learning_rate": 9.06677796327212e-06, "loss": 0.4052, "step": 5690 }, { "epoch": 0.28, "learning_rate": 9.065108514190318e-06, "loss": 0.3608, "step": 5700 }, { "epoch": 0.29, "learning_rate": 9.063439065108514e-06, "loss": 0.418, "step": 5710 }, { "epoch": 0.29, "learning_rate": 9.061769616026712e-06, "loss": 0.3614, "step": 5720 }, { "epoch": 0.29, "learning_rate": 9.060100166944908e-06, "loss": 0.369, "step": 5730 }, { "epoch": 0.29, "learning_rate": 9.058430717863106e-06, "loss": 0.4054, "step": 5740 }, { "epoch": 0.29, "learning_rate": 9.056761268781304e-06, "loss": 0.4084, "step": 5750 }, { "epoch": 0.29, "learning_rate": 9.0550918196995e-06, "loss": 0.3116, "step": 5760 }, { "epoch": 0.29, "learning_rate": 9.053422370617698e-06, "loss": 0.4926, "step": 5770 }, { "epoch": 0.29, "learning_rate": 9.051752921535894e-06, "loss": 0.4504, "step": 5780 }, { "epoch": 0.29, "learning_rate": 9.05008347245409e-06, "loss": 0.4338, "step": 5790 }, { "epoch": 0.29, "learning_rate": 9.048414023372288e-06, "loss": 0.376, "step": 5800 }, { "epoch": 0.29, "learning_rate": 9.046744574290484e-06, "loss": 0.4256, "step": 5810 }, { "epoch": 0.29, "learning_rate": 9.045075125208682e-06, "loss": 0.3336, "step": 5820 }, { "epoch": 0.29, "learning_rate": 9.043405676126878e-06, "loss": 0.3909, "step": 5830 }, { "epoch": 0.29, "learning_rate": 9.041736227045076e-06, "loss": 0.4627, "step": 5840 }, { "epoch": 0.29, "learning_rate": 9.040066777963274e-06, "loss": 0.4297, "step": 5850 }, { "epoch": 0.29, "learning_rate": 9.03839732888147e-06, "loss": 0.4207, "step": 5860 }, { "epoch": 0.29, "learning_rate": 9.036727879799668e-06, "loss": 0.4262, "step": 5870 }, { "epoch": 0.29, "learning_rate": 9.035058430717864e-06, "loss": 0.3971, "step": 5880 }, { "epoch": 0.29, "learning_rate": 9.03338898163606e-06, "loss": 0.3918, "step": 5890 }, { "epoch": 0.29, "learning_rate": 9.031719532554258e-06, "loss": 0.4772, "step": 5900 }, { "epoch": 0.3, "learning_rate": 9.030050083472454e-06, "loss": 0.3476, "step": 5910 }, { "epoch": 0.3, "learning_rate": 9.028380634390652e-06, "loss": 0.3962, "step": 5920 }, { "epoch": 0.3, "learning_rate": 9.026711185308848e-06, "loss": 0.4349, "step": 5930 }, { "epoch": 0.3, "learning_rate": 9.025041736227046e-06, "loss": 0.4485, "step": 5940 }, { "epoch": 0.3, "learning_rate": 9.023372287145242e-06, "loss": 0.4139, "step": 5950 }, { "epoch": 0.3, "learning_rate": 9.02170283806344e-06, "loss": 0.4041, "step": 5960 }, { "epoch": 0.3, "learning_rate": 9.020033388981638e-06, "loss": 0.4669, "step": 5970 }, { "epoch": 0.3, "learning_rate": 9.018363939899834e-06, "loss": 0.4119, "step": 5980 }, { "epoch": 0.3, "learning_rate": 9.01669449081803e-06, "loss": 0.3947, "step": 5990 }, { "epoch": 0.3, "learning_rate": 9.015025041736228e-06, "loss": 0.3727, "step": 6000 }, { "epoch": 0.3, "learning_rate": 9.013355592654424e-06, "loss": 0.35, "step": 6010 }, { "epoch": 0.3, "learning_rate": 9.011686143572622e-06, "loss": 0.4159, "step": 6020 }, { "epoch": 0.3, "learning_rate": 9.010016694490818e-06, "loss": 0.3389, "step": 6030 }, { "epoch": 0.3, "learning_rate": 9.008347245409016e-06, "loss": 0.451, "step": 6040 }, { "epoch": 0.3, "learning_rate": 9.006677796327212e-06, "loss": 0.4327, "step": 6050 }, { "epoch": 0.3, "learning_rate": 9.00500834724541e-06, "loss": 0.4449, "step": 6060 }, { "epoch": 0.3, "learning_rate": 9.003338898163608e-06, "loss": 0.3869, "step": 6070 }, { "epoch": 0.3, "learning_rate": 9.001669449081804e-06, "loss": 0.3632, "step": 6080 }, { "epoch": 0.3, "learning_rate": 9e-06, "loss": 0.3641, "step": 6090 }, { "epoch": 0.3, "learning_rate": 8.998330550918198e-06, "loss": 0.4654, "step": 6100 }, { "epoch": 0.31, "learning_rate": 8.996661101836394e-06, "loss": 0.3334, "step": 6110 }, { "epoch": 0.31, "learning_rate": 8.994991652754592e-06, "loss": 0.3961, "step": 6120 }, { "epoch": 0.31, "learning_rate": 8.993322203672788e-06, "loss": 0.3067, "step": 6130 }, { "epoch": 0.31, "learning_rate": 8.991652754590986e-06, "loss": 0.4189, "step": 6140 }, { "epoch": 0.31, "learning_rate": 8.989983305509182e-06, "loss": 0.4905, "step": 6150 }, { "epoch": 0.31, "learning_rate": 8.98831385642738e-06, "loss": 0.4022, "step": 6160 }, { "epoch": 0.31, "learning_rate": 8.986644407345578e-06, "loss": 0.3552, "step": 6170 }, { "epoch": 0.31, "learning_rate": 8.984974958263774e-06, "loss": 0.3295, "step": 6180 }, { "epoch": 0.31, "learning_rate": 8.98330550918197e-06, "loss": 0.4552, "step": 6190 }, { "epoch": 0.31, "learning_rate": 8.981636060100168e-06, "loss": 0.4045, "step": 6200 }, { "epoch": 0.31, "learning_rate": 8.979966611018364e-06, "loss": 0.3818, "step": 6210 }, { "epoch": 0.31, "learning_rate": 8.978297161936562e-06, "loss": 0.3848, "step": 6220 }, { "epoch": 0.31, "learning_rate": 8.976627712854758e-06, "loss": 0.4121, "step": 6230 }, { "epoch": 0.31, "learning_rate": 8.974958263772956e-06, "loss": 0.3708, "step": 6240 }, { "epoch": 0.31, "learning_rate": 8.973288814691152e-06, "loss": 0.3477, "step": 6250 }, { "epoch": 0.31, "learning_rate": 8.97161936560935e-06, "loss": 0.4601, "step": 6260 }, { "epoch": 0.31, "learning_rate": 8.969949916527546e-06, "loss": 0.4092, "step": 6270 }, { "epoch": 0.31, "learning_rate": 8.968280467445744e-06, "loss": 0.3874, "step": 6280 }, { "epoch": 0.31, "learning_rate": 8.96661101836394e-06, "loss": 0.317, "step": 6290 }, { "epoch": 0.32, "learning_rate": 8.964941569282138e-06, "loss": 0.398, "step": 6300 }, { "epoch": 0.32, "learning_rate": 8.963272120200334e-06, "loss": 0.3413, "step": 6310 }, { "epoch": 0.32, "learning_rate": 8.96160267111853e-06, "loss": 0.3721, "step": 6320 }, { "epoch": 0.32, "learning_rate": 8.959933222036728e-06, "loss": 0.3975, "step": 6330 }, { "epoch": 0.32, "learning_rate": 8.958263772954926e-06, "loss": 0.4565, "step": 6340 }, { "epoch": 0.32, "learning_rate": 8.956594323873122e-06, "loss": 0.284, "step": 6350 }, { "epoch": 0.32, "learning_rate": 8.95492487479132e-06, "loss": 0.3554, "step": 6360 }, { "epoch": 0.32, "learning_rate": 8.953255425709516e-06, "loss": 0.3676, "step": 6370 }, { "epoch": 0.32, "learning_rate": 8.951585976627714e-06, "loss": 0.3926, "step": 6380 }, { "epoch": 0.32, "learning_rate": 8.94991652754591e-06, "loss": 0.3971, "step": 6390 }, { "epoch": 0.32, "learning_rate": 8.948247078464108e-06, "loss": 0.4799, "step": 6400 }, { "epoch": 0.32, "learning_rate": 8.946577629382304e-06, "loss": 0.3465, "step": 6410 }, { "epoch": 0.32, "learning_rate": 8.9449081803005e-06, "loss": 0.3945, "step": 6420 }, { "epoch": 0.32, "learning_rate": 8.943238731218698e-06, "loss": 0.3849, "step": 6430 }, { "epoch": 0.32, "learning_rate": 8.941569282136896e-06, "loss": 0.3515, "step": 6440 }, { "epoch": 0.32, "learning_rate": 8.939899833055092e-06, "loss": 0.4486, "step": 6450 }, { "epoch": 0.32, "learning_rate": 8.93823038397329e-06, "loss": 0.4035, "step": 6460 }, { "epoch": 0.32, "learning_rate": 8.936560934891486e-06, "loss": 0.4692, "step": 6470 }, { "epoch": 0.32, "learning_rate": 8.934891485809684e-06, "loss": 0.4509, "step": 6480 }, { "epoch": 0.32, "learning_rate": 8.933222036727882e-06, "loss": 0.3831, "step": 6490 }, { "epoch": 0.33, "learning_rate": 8.931552587646078e-06, "loss": 0.4114, "step": 6500 }, { "epoch": 0.33, "learning_rate": 8.929883138564274e-06, "loss": 0.3657, "step": 6510 }, { "epoch": 0.33, "learning_rate": 8.92821368948247e-06, "loss": 0.3895, "step": 6520 }, { "epoch": 0.33, "learning_rate": 8.926544240400668e-06, "loss": 0.3234, "step": 6530 }, { "epoch": 0.33, "learning_rate": 8.924874791318866e-06, "loss": 0.4571, "step": 6540 }, { "epoch": 0.33, "learning_rate": 8.923205342237062e-06, "loss": 0.3952, "step": 6550 }, { "epoch": 0.33, "learning_rate": 8.92153589315526e-06, "loss": 0.4192, "step": 6560 }, { "epoch": 0.33, "learning_rate": 8.919866444073456e-06, "loss": 0.3953, "step": 6570 }, { "epoch": 0.33, "learning_rate": 8.918196994991654e-06, "loss": 0.3502, "step": 6580 }, { "epoch": 0.33, "learning_rate": 8.91652754590985e-06, "loss": 0.3455, "step": 6590 }, { "epoch": 0.33, "learning_rate": 8.914858096828048e-06, "loss": 0.3388, "step": 6600 }, { "epoch": 0.33, "learning_rate": 8.913188647746244e-06, "loss": 0.3253, "step": 6610 }, { "epoch": 0.33, "learning_rate": 8.911519198664442e-06, "loss": 0.4133, "step": 6620 }, { "epoch": 0.33, "learning_rate": 8.909849749582638e-06, "loss": 0.4795, "step": 6630 }, { "epoch": 0.33, "learning_rate": 8.908180300500834e-06, "loss": 0.3632, "step": 6640 }, { "epoch": 0.33, "learning_rate": 8.906510851419032e-06, "loss": 0.4207, "step": 6650 }, { "epoch": 0.33, "learning_rate": 8.90484140233723e-06, "loss": 0.3256, "step": 6660 }, { "epoch": 0.33, "learning_rate": 8.903171953255426e-06, "loss": 0.391, "step": 6670 }, { "epoch": 0.33, "learning_rate": 8.901502504173624e-06, "loss": 0.4789, "step": 6680 }, { "epoch": 0.33, "learning_rate": 8.89983305509182e-06, "loss": 0.4104, "step": 6690 }, { "epoch": 0.34, "learning_rate": 8.898163606010018e-06, "loss": 0.3923, "step": 6700 }, { "epoch": 0.34, "learning_rate": 8.896494156928214e-06, "loss": 0.3939, "step": 6710 }, { "epoch": 0.34, "learning_rate": 8.894824707846412e-06, "loss": 0.3674, "step": 6720 }, { "epoch": 0.34, "learning_rate": 8.893155258764608e-06, "loss": 0.4794, "step": 6730 }, { "epoch": 0.34, "learning_rate": 8.891485809682804e-06, "loss": 0.3975, "step": 6740 }, { "epoch": 0.34, "learning_rate": 8.889816360601002e-06, "loss": 0.4446, "step": 6750 }, { "epoch": 0.34, "learning_rate": 8.8881469115192e-06, "loss": 0.3611, "step": 6760 }, { "epoch": 0.34, "learning_rate": 8.886477462437396e-06, "loss": 0.3531, "step": 6770 }, { "epoch": 0.34, "learning_rate": 8.884808013355594e-06, "loss": 0.3812, "step": 6780 }, { "epoch": 0.34, "learning_rate": 8.88313856427379e-06, "loss": 0.2769, "step": 6790 }, { "epoch": 0.34, "learning_rate": 8.881469115191988e-06, "loss": 0.3573, "step": 6800 }, { "epoch": 0.34, "learning_rate": 8.879799666110184e-06, "loss": 0.5278, "step": 6810 }, { "epoch": 0.34, "learning_rate": 8.878130217028382e-06, "loss": 0.399, "step": 6820 }, { "epoch": 0.34, "learning_rate": 8.876460767946578e-06, "loss": 0.3783, "step": 6830 }, { "epoch": 0.34, "learning_rate": 8.874791318864774e-06, "loss": 0.3924, "step": 6840 }, { "epoch": 0.34, "learning_rate": 8.873121869782972e-06, "loss": 0.3257, "step": 6850 }, { "epoch": 0.34, "learning_rate": 8.87145242070117e-06, "loss": 0.394, "step": 6860 }, { "epoch": 0.34, "learning_rate": 8.869782971619366e-06, "loss": 0.2963, "step": 6870 }, { "epoch": 0.34, "learning_rate": 8.868113522537564e-06, "loss": 0.3605, "step": 6880 }, { "epoch": 0.34, "learning_rate": 8.86644407345576e-06, "loss": 0.4208, "step": 6890 }, { "epoch": 0.34, "learning_rate": 8.864774624373958e-06, "loss": 0.3753, "step": 6900 }, { "epoch": 0.35, "learning_rate": 8.863105175292154e-06, "loss": 0.3781, "step": 6910 }, { "epoch": 0.35, "learning_rate": 8.861435726210352e-06, "loss": 0.2955, "step": 6920 }, { "epoch": 0.35, "learning_rate": 8.859766277128548e-06, "loss": 0.431, "step": 6930 }, { "epoch": 0.35, "learning_rate": 8.858096828046744e-06, "loss": 0.3932, "step": 6940 }, { "epoch": 0.35, "learning_rate": 8.856427378964942e-06, "loss": 0.4129, "step": 6950 }, { "epoch": 0.35, "learning_rate": 8.854757929883138e-06, "loss": 0.3637, "step": 6960 }, { "epoch": 0.35, "learning_rate": 8.853088480801336e-06, "loss": 0.3369, "step": 6970 }, { "epoch": 0.35, "learning_rate": 8.851419031719534e-06, "loss": 0.4323, "step": 6980 }, { "epoch": 0.35, "learning_rate": 8.84974958263773e-06, "loss": 0.5627, "step": 6990 }, { "epoch": 0.35, "learning_rate": 8.848080133555928e-06, "loss": 0.411, "step": 7000 }, { "epoch": 0.35, "learning_rate": 8.846410684474124e-06, "loss": 0.404, "step": 7010 }, { "epoch": 0.35, "learning_rate": 8.844741235392322e-06, "loss": 0.3066, "step": 7020 }, { "epoch": 0.35, "learning_rate": 8.843071786310518e-06, "loss": 0.411, "step": 7030 }, { "epoch": 0.35, "learning_rate": 8.841402337228714e-06, "loss": 0.3651, "step": 7040 }, { "epoch": 0.35, "learning_rate": 8.839732888146912e-06, "loss": 0.3285, "step": 7050 }, { "epoch": 0.35, "learning_rate": 8.838063439065108e-06, "loss": 0.354, "step": 7060 }, { "epoch": 0.35, "learning_rate": 8.836393989983306e-06, "loss": 0.6048, "step": 7070 }, { "epoch": 0.35, "learning_rate": 8.834724540901504e-06, "loss": 0.3834, "step": 7080 }, { "epoch": 0.35, "learning_rate": 8.8330550918197e-06, "loss": 0.4388, "step": 7090 }, { "epoch": 0.35, "learning_rate": 8.831385642737898e-06, "loss": 0.436, "step": 7100 }, { "epoch": 0.36, "learning_rate": 8.829716193656094e-06, "loss": 0.3499, "step": 7110 }, { "epoch": 0.36, "learning_rate": 8.828046744574292e-06, "loss": 0.3144, "step": 7120 }, { "epoch": 0.36, "learning_rate": 8.826377295492488e-06, "loss": 0.4116, "step": 7130 }, { "epoch": 0.36, "learning_rate": 8.824707846410684e-06, "loss": 0.4303, "step": 7140 }, { "epoch": 0.36, "learning_rate": 8.823038397328882e-06, "loss": 0.4588, "step": 7150 }, { "epoch": 0.36, "learning_rate": 8.821368948247078e-06, "loss": 0.4384, "step": 7160 }, { "epoch": 0.36, "learning_rate": 8.819699499165276e-06, "loss": 0.4324, "step": 7170 }, { "epoch": 0.36, "learning_rate": 8.818030050083474e-06, "loss": 0.419, "step": 7180 }, { "epoch": 0.36, "learning_rate": 8.81636060100167e-06, "loss": 0.4185, "step": 7190 }, { "epoch": 0.36, "learning_rate": 8.814691151919868e-06, "loss": 0.414, "step": 7200 }, { "epoch": 0.36, "learning_rate": 8.813021702838064e-06, "loss": 0.3528, "step": 7210 }, { "epoch": 0.36, "learning_rate": 8.811352253756262e-06, "loss": 0.4962, "step": 7220 }, { "epoch": 0.36, "learning_rate": 8.809682804674458e-06, "loss": 0.2988, "step": 7230 }, { "epoch": 0.36, "learning_rate": 8.808013355592654e-06, "loss": 0.436, "step": 7240 }, { "epoch": 0.36, "learning_rate": 8.806343906510852e-06, "loss": 0.3828, "step": 7250 }, { "epoch": 0.36, "learning_rate": 8.804674457429048e-06, "loss": 0.3408, "step": 7260 }, { "epoch": 0.36, "learning_rate": 8.803005008347246e-06, "loss": 0.3942, "step": 7270 }, { "epoch": 0.36, "learning_rate": 8.801335559265442e-06, "loss": 0.39, "step": 7280 }, { "epoch": 0.36, "learning_rate": 8.79966611018364e-06, "loss": 0.3821, "step": 7290 }, { "epoch": 0.36, "learning_rate": 8.797996661101838e-06, "loss": 0.4493, "step": 7300 }, { "epoch": 0.37, "learning_rate": 8.796327212020034e-06, "loss": 0.3815, "step": 7310 }, { "epoch": 0.37, "learning_rate": 8.794657762938232e-06, "loss": 0.4281, "step": 7320 }, { "epoch": 0.37, "learning_rate": 8.792988313856428e-06, "loss": 0.4499, "step": 7330 }, { "epoch": 0.37, "learning_rate": 8.791318864774626e-06, "loss": 0.3245, "step": 7340 }, { "epoch": 0.37, "learning_rate": 8.789649415692822e-06, "loss": 0.418, "step": 7350 }, { "epoch": 0.37, "learning_rate": 8.787979966611018e-06, "loss": 0.4375, "step": 7360 }, { "epoch": 0.37, "learning_rate": 8.786310517529216e-06, "loss": 0.4213, "step": 7370 }, { "epoch": 0.37, "learning_rate": 8.784641068447412e-06, "loss": 0.329, "step": 7380 }, { "epoch": 0.37, "learning_rate": 8.78297161936561e-06, "loss": 0.3497, "step": 7390 }, { "epoch": 0.37, "learning_rate": 8.781302170283808e-06, "loss": 0.4729, "step": 7400 }, { "epoch": 0.37, "learning_rate": 8.779632721202004e-06, "loss": 0.4009, "step": 7410 }, { "epoch": 0.37, "learning_rate": 8.777963272120202e-06, "loss": 0.398, "step": 7420 }, { "epoch": 0.37, "learning_rate": 8.776293823038398e-06, "loss": 0.4279, "step": 7430 }, { "epoch": 0.37, "learning_rate": 8.774624373956596e-06, "loss": 0.3723, "step": 7440 }, { "epoch": 0.37, "learning_rate": 8.772954924874792e-06, "loss": 0.3757, "step": 7450 }, { "epoch": 0.37, "learning_rate": 8.771285475792988e-06, "loss": 0.3495, "step": 7460 }, { "epoch": 0.37, "learning_rate": 8.769616026711186e-06, "loss": 0.3137, "step": 7470 }, { "epoch": 0.37, "learning_rate": 8.767946577629382e-06, "loss": 0.3287, "step": 7480 }, { "epoch": 0.37, "learning_rate": 8.76627712854758e-06, "loss": 0.3516, "step": 7490 }, { "epoch": 0.38, "learning_rate": 8.764607679465778e-06, "loss": 0.4655, "step": 7500 }, { "epoch": 0.38, "learning_rate": 8.762938230383974e-06, "loss": 0.4078, "step": 7510 }, { "epoch": 0.38, "learning_rate": 8.761268781302172e-06, "loss": 0.4237, "step": 7520 }, { "epoch": 0.38, "learning_rate": 8.759599332220368e-06, "loss": 0.3798, "step": 7530 }, { "epoch": 0.38, "learning_rate": 8.757929883138566e-06, "loss": 0.4175, "step": 7540 }, { "epoch": 0.38, "learning_rate": 8.756260434056762e-06, "loss": 0.3788, "step": 7550 }, { "epoch": 0.38, "learning_rate": 8.754590984974958e-06, "loss": 0.4782, "step": 7560 }, { "epoch": 0.38, "learning_rate": 8.752921535893156e-06, "loss": 0.3987, "step": 7570 }, { "epoch": 0.38, "learning_rate": 8.751252086811352e-06, "loss": 0.3033, "step": 7580 }, { "epoch": 0.38, "learning_rate": 8.74958263772955e-06, "loss": 0.3378, "step": 7590 }, { "epoch": 0.38, "learning_rate": 8.747913188647746e-06, "loss": 0.3666, "step": 7600 }, { "epoch": 0.38, "learning_rate": 8.746243739565944e-06, "loss": 0.3307, "step": 7610 }, { "epoch": 0.38, "learning_rate": 8.744574290484142e-06, "loss": 0.4073, "step": 7620 }, { "epoch": 0.38, "learning_rate": 8.742904841402338e-06, "loss": 0.507, "step": 7630 }, { "epoch": 0.38, "learning_rate": 8.741235392320536e-06, "loss": 0.3946, "step": 7640 }, { "epoch": 0.38, "learning_rate": 8.739565943238732e-06, "loss": 0.3398, "step": 7650 }, { "epoch": 0.38, "learning_rate": 8.737896494156928e-06, "loss": 0.4527, "step": 7660 }, { "epoch": 0.38, "learning_rate": 8.736227045075126e-06, "loss": 0.3792, "step": 7670 }, { "epoch": 0.38, "learning_rate": 8.734557595993322e-06, "loss": 0.4455, "step": 7680 }, { "epoch": 0.38, "learning_rate": 8.73288814691152e-06, "loss": 0.3722, "step": 7690 }, { "epoch": 0.39, "learning_rate": 8.731218697829716e-06, "loss": 0.387, "step": 7700 }, { "epoch": 0.39, "learning_rate": 8.729549248747914e-06, "loss": 0.3724, "step": 7710 }, { "epoch": 0.39, "learning_rate": 8.727879799666112e-06, "loss": 0.3864, "step": 7720 }, { "epoch": 0.39, "learning_rate": 8.726210350584308e-06, "loss": 0.3291, "step": 7730 }, { "epoch": 0.39, "learning_rate": 8.724540901502506e-06, "loss": 0.4691, "step": 7740 }, { "epoch": 0.39, "learning_rate": 8.722871452420702e-06, "loss": 0.3968, "step": 7750 }, { "epoch": 0.39, "learning_rate": 8.721202003338898e-06, "loss": 0.4483, "step": 7760 }, { "epoch": 0.39, "learning_rate": 8.719532554257096e-06, "loss": 0.3745, "step": 7770 }, { "epoch": 0.39, "learning_rate": 8.717863105175292e-06, "loss": 0.4669, "step": 7780 }, { "epoch": 0.39, "learning_rate": 8.71619365609349e-06, "loss": 0.3372, "step": 7790 }, { "epoch": 0.39, "learning_rate": 8.714524207011686e-06, "loss": 0.3867, "step": 7800 }, { "epoch": 0.39, "learning_rate": 8.712854757929884e-06, "loss": 0.4389, "step": 7810 }, { "epoch": 0.39, "learning_rate": 8.711185308848082e-06, "loss": 0.3767, "step": 7820 }, { "epoch": 0.39, "learning_rate": 8.709515859766278e-06, "loss": 0.36, "step": 7830 }, { "epoch": 0.39, "learning_rate": 8.707846410684476e-06, "loss": 0.3695, "step": 7840 }, { "epoch": 0.39, "learning_rate": 8.706176961602672e-06, "loss": 0.4513, "step": 7850 }, { "epoch": 0.39, "learning_rate": 8.704507512520868e-06, "loss": 0.3999, "step": 7860 }, { "epoch": 0.39, "learning_rate": 8.702838063439066e-06, "loss": 0.3885, "step": 7870 }, { "epoch": 0.39, "learning_rate": 8.701168614357262e-06, "loss": 0.3052, "step": 7880 }, { "epoch": 0.39, "learning_rate": 8.69949916527546e-06, "loss": 0.2895, "step": 7890 }, { "epoch": 0.4, "learning_rate": 8.697829716193656e-06, "loss": 0.3119, "step": 7900 }, { "epoch": 0.4, "learning_rate": 8.696160267111854e-06, "loss": 0.4263, "step": 7910 }, { "epoch": 0.4, "learning_rate": 8.694490818030052e-06, "loss": 0.4181, "step": 7920 }, { "epoch": 0.4, "learning_rate": 8.692821368948248e-06, "loss": 0.3854, "step": 7930 }, { "epoch": 0.4, "learning_rate": 8.691151919866446e-06, "loss": 0.4198, "step": 7940 }, { "epoch": 0.4, "learning_rate": 8.689482470784642e-06, "loss": 0.4125, "step": 7950 }, { "epoch": 0.4, "learning_rate": 8.68781302170284e-06, "loss": 0.4549, "step": 7960 }, { "epoch": 0.4, "learning_rate": 8.686143572621036e-06, "loss": 0.5206, "step": 7970 }, { "epoch": 0.4, "learning_rate": 8.684474123539232e-06, "loss": 0.4112, "step": 7980 }, { "epoch": 0.4, "learning_rate": 8.68280467445743e-06, "loss": 0.4307, "step": 7990 }, { "epoch": 0.4, "learning_rate": 8.681135225375626e-06, "loss": 0.4104, "step": 8000 }, { "epoch": 0.4, "learning_rate": 8.679465776293824e-06, "loss": 0.3965, "step": 8010 }, { "epoch": 0.4, "learning_rate": 8.67779632721202e-06, "loss": 0.4182, "step": 8020 }, { "epoch": 0.4, "learning_rate": 8.676126878130218e-06, "loss": 0.357, "step": 8030 }, { "epoch": 0.4, "learning_rate": 8.674457429048416e-06, "loss": 0.4258, "step": 8040 }, { "epoch": 0.4, "learning_rate": 8.672787979966612e-06, "loss": 0.3659, "step": 8050 }, { "epoch": 0.4, "learning_rate": 8.67111853088481e-06, "loss": 0.3829, "step": 8060 }, { "epoch": 0.4, "learning_rate": 8.669449081803006e-06, "loss": 0.3164, "step": 8070 }, { "epoch": 0.4, "learning_rate": 8.667779632721202e-06, "loss": 0.3313, "step": 8080 }, { "epoch": 0.4, "learning_rate": 8.6661101836394e-06, "loss": 0.4017, "step": 8090 }, { "epoch": 0.41, "learning_rate": 8.664440734557596e-06, "loss": 0.3375, "step": 8100 }, { "epoch": 0.41, "learning_rate": 8.662771285475794e-06, "loss": 0.4614, "step": 8110 }, { "epoch": 0.41, "learning_rate": 8.66110183639399e-06, "loss": 0.4593, "step": 8120 }, { "epoch": 0.41, "learning_rate": 8.659432387312188e-06, "loss": 0.5306, "step": 8130 }, { "epoch": 0.41, "learning_rate": 8.657762938230386e-06, "loss": 0.5205, "step": 8140 }, { "epoch": 0.41, "learning_rate": 8.656093489148582e-06, "loss": 0.3769, "step": 8150 }, { "epoch": 0.41, "learning_rate": 8.65442404006678e-06, "loss": 0.4141, "step": 8160 }, { "epoch": 0.41, "learning_rate": 8.652754590984976e-06, "loss": 0.3406, "step": 8170 }, { "epoch": 0.41, "learning_rate": 8.651085141903172e-06, "loss": 0.4357, "step": 8180 }, { "epoch": 0.41, "learning_rate": 8.64941569282137e-06, "loss": 0.3956, "step": 8190 }, { "epoch": 0.41, "learning_rate": 8.647746243739566e-06, "loss": 0.3099, "step": 8200 }, { "epoch": 0.41, "learning_rate": 8.646076794657764e-06, "loss": 0.4079, "step": 8210 }, { "epoch": 0.41, "learning_rate": 8.64440734557596e-06, "loss": 0.3834, "step": 8220 }, { "epoch": 0.41, "learning_rate": 8.642737896494158e-06, "loss": 0.4104, "step": 8230 }, { "epoch": 0.41, "learning_rate": 8.641068447412356e-06, "loss": 0.5021, "step": 8240 }, { "epoch": 0.41, "learning_rate": 8.639398998330552e-06, "loss": 0.367, "step": 8250 }, { "epoch": 0.41, "learning_rate": 8.63772954924875e-06, "loss": 0.3122, "step": 8260 }, { "epoch": 0.41, "learning_rate": 8.636060100166946e-06, "loss": 0.4096, "step": 8270 }, { "epoch": 0.41, "learning_rate": 8.634390651085142e-06, "loss": 0.3462, "step": 8280 }, { "epoch": 0.41, "learning_rate": 8.63272120200334e-06, "loss": 0.4132, "step": 8290 }, { "epoch": 0.41, "learning_rate": 8.631051752921536e-06, "loss": 0.3416, "step": 8300 }, { "epoch": 0.42, "learning_rate": 8.629382303839734e-06, "loss": 0.4683, "step": 8310 }, { "epoch": 0.42, "learning_rate": 8.62771285475793e-06, "loss": 0.3615, "step": 8320 }, { "epoch": 0.42, "learning_rate": 8.626043405676128e-06, "loss": 0.3942, "step": 8330 }, { "epoch": 0.42, "learning_rate": 8.624373956594324e-06, "loss": 0.3631, "step": 8340 }, { "epoch": 0.42, "learning_rate": 8.622704507512522e-06, "loss": 0.3689, "step": 8350 }, { "epoch": 0.42, "learning_rate": 8.62103505843072e-06, "loss": 0.3205, "step": 8360 }, { "epoch": 0.42, "learning_rate": 8.619365609348916e-06, "loss": 0.4595, "step": 8370 }, { "epoch": 0.42, "learning_rate": 8.617696160267112e-06, "loss": 0.4124, "step": 8380 }, { "epoch": 0.42, "learning_rate": 8.61602671118531e-06, "loss": 0.4547, "step": 8390 }, { "epoch": 0.42, "learning_rate": 8.614357262103506e-06, "loss": 0.3878, "step": 8400 }, { "epoch": 0.42, "learning_rate": 8.612687813021704e-06, "loss": 0.4714, "step": 8410 }, { "epoch": 0.42, "learning_rate": 8.6110183639399e-06, "loss": 0.4664, "step": 8420 }, { "epoch": 0.42, "learning_rate": 8.609348914858098e-06, "loss": 0.4172, "step": 8430 }, { "epoch": 0.42, "learning_rate": 8.607679465776294e-06, "loss": 0.3419, "step": 8440 }, { "epoch": 0.42, "learning_rate": 8.606010016694492e-06, "loss": 0.3824, "step": 8450 }, { "epoch": 0.42, "learning_rate": 8.60434056761269e-06, "loss": 0.3122, "step": 8460 }, { "epoch": 0.42, "learning_rate": 8.602671118530886e-06, "loss": 0.3254, "step": 8470 }, { "epoch": 0.42, "learning_rate": 8.601001669449082e-06, "loss": 0.3949, "step": 8480 }, { "epoch": 0.42, "learning_rate": 8.59933222036728e-06, "loss": 0.4877, "step": 8490 }, { "epoch": 0.42, "learning_rate": 8.597662771285476e-06, "loss": 0.5559, "step": 8500 }, { "epoch": 0.43, "learning_rate": 8.595993322203674e-06, "loss": 0.3993, "step": 8510 }, { "epoch": 0.43, "learning_rate": 8.59432387312187e-06, "loss": 0.3757, "step": 8520 }, { "epoch": 0.43, "learning_rate": 8.592654424040068e-06, "loss": 0.3854, "step": 8530 }, { "epoch": 0.43, "learning_rate": 8.590984974958264e-06, "loss": 0.3823, "step": 8540 }, { "epoch": 0.43, "learning_rate": 8.589315525876462e-06, "loss": 0.3657, "step": 8550 }, { "epoch": 0.43, "learning_rate": 8.58764607679466e-06, "loss": 0.3323, "step": 8560 }, { "epoch": 0.43, "learning_rate": 8.585976627712856e-06, "loss": 0.355, "step": 8570 }, { "epoch": 0.43, "learning_rate": 8.584307178631052e-06, "loss": 0.3422, "step": 8580 }, { "epoch": 0.43, "learning_rate": 8.58263772954925e-06, "loss": 0.3913, "step": 8590 }, { "epoch": 0.43, "learning_rate": 8.580968280467446e-06, "loss": 0.4488, "step": 8600 }, { "epoch": 0.43, "learning_rate": 8.579298831385642e-06, "loss": 0.3792, "step": 8610 }, { "epoch": 0.43, "learning_rate": 8.57762938230384e-06, "loss": 0.3671, "step": 8620 }, { "epoch": 0.43, "learning_rate": 8.575959933222038e-06, "loss": 0.3119, "step": 8630 }, { "epoch": 0.43, "learning_rate": 8.574290484140234e-06, "loss": 0.3802, "step": 8640 }, { "epoch": 0.43, "learning_rate": 8.572621035058432e-06, "loss": 0.3539, "step": 8650 }, { "epoch": 0.43, "learning_rate": 8.570951585976628e-06, "loss": 0.3294, "step": 8660 }, { "epoch": 0.43, "learning_rate": 8.569282136894826e-06, "loss": 0.4595, "step": 8670 }, { "epoch": 0.43, "learning_rate": 8.567612687813024e-06, "loss": 0.4227, "step": 8680 }, { "epoch": 0.43, "learning_rate": 8.56594323873122e-06, "loss": 0.4219, "step": 8690 }, { "epoch": 0.43, "learning_rate": 8.564273789649416e-06, "loss": 0.3996, "step": 8700 }, { "epoch": 0.44, "learning_rate": 8.562604340567612e-06, "loss": 0.493, "step": 8710 }, { "epoch": 0.44, "learning_rate": 8.56093489148581e-06, "loss": 0.3663, "step": 8720 }, { "epoch": 0.44, "learning_rate": 8.559265442404008e-06, "loss": 0.436, "step": 8730 }, { "epoch": 0.44, "learning_rate": 8.557595993322204e-06, "loss": 0.3075, "step": 8740 }, { "epoch": 0.44, "learning_rate": 8.555926544240402e-06, "loss": 0.4131, "step": 8750 }, { "epoch": 0.44, "learning_rate": 8.554257095158598e-06, "loss": 0.3427, "step": 8760 }, { "epoch": 0.44, "learning_rate": 8.552587646076796e-06, "loss": 0.4796, "step": 8770 }, { "epoch": 0.44, "learning_rate": 8.550918196994994e-06, "loss": 0.359, "step": 8780 }, { "epoch": 0.44, "learning_rate": 8.54924874791319e-06, "loss": 0.4203, "step": 8790 }, { "epoch": 0.44, "learning_rate": 8.547579298831386e-06, "loss": 0.4379, "step": 8800 }, { "epoch": 0.44, "learning_rate": 8.545909849749584e-06, "loss": 0.4198, "step": 8810 }, { "epoch": 0.44, "learning_rate": 8.54424040066778e-06, "loss": 0.3654, "step": 8820 }, { "epoch": 0.44, "learning_rate": 8.542570951585978e-06, "loss": 0.4226, "step": 8830 }, { "epoch": 0.44, "learning_rate": 8.540901502504174e-06, "loss": 0.3582, "step": 8840 }, { "epoch": 0.44, "learning_rate": 8.539232053422372e-06, "loss": 0.4135, "step": 8850 }, { "epoch": 0.44, "learning_rate": 8.537562604340568e-06, "loss": 0.3487, "step": 8860 }, { "epoch": 0.44, "learning_rate": 8.535893155258766e-06, "loss": 0.3044, "step": 8870 }, { "epoch": 0.44, "learning_rate": 8.534223706176964e-06, "loss": 0.4707, "step": 8880 }, { "epoch": 0.44, "learning_rate": 8.53255425709516e-06, "loss": 0.3398, "step": 8890 }, { "epoch": 0.45, "learning_rate": 8.530884808013356e-06, "loss": 0.3529, "step": 8900 }, { "epoch": 0.45, "learning_rate": 8.529215358931554e-06, "loss": 0.4646, "step": 8910 }, { "epoch": 0.45, "learning_rate": 8.52754590984975e-06, "loss": 0.3397, "step": 8920 }, { "epoch": 0.45, "learning_rate": 8.525876460767946e-06, "loss": 0.5569, "step": 8930 }, { "epoch": 0.45, "learning_rate": 8.524207011686144e-06, "loss": 0.3695, "step": 8940 }, { "epoch": 0.45, "learning_rate": 8.522537562604342e-06, "loss": 0.4254, "step": 8950 }, { "epoch": 0.45, "learning_rate": 8.520868113522538e-06, "loss": 0.3471, "step": 8960 }, { "epoch": 0.45, "learning_rate": 8.519198664440736e-06, "loss": 0.3735, "step": 8970 }, { "epoch": 0.45, "learning_rate": 8.517529215358932e-06, "loss": 0.3627, "step": 8980 }, { "epoch": 0.45, "learning_rate": 8.51585976627713e-06, "loss": 0.4514, "step": 8990 }, { "epoch": 0.45, "learning_rate": 8.514190317195326e-06, "loss": 0.3579, "step": 9000 }, { "epoch": 0.45, "learning_rate": 8.512520868113524e-06, "loss": 0.3508, "step": 9010 }, { "epoch": 0.45, "learning_rate": 8.51085141903172e-06, "loss": 0.4453, "step": 9020 }, { "epoch": 0.45, "learning_rate": 8.509181969949916e-06, "loss": 0.303, "step": 9030 }, { "epoch": 0.45, "learning_rate": 8.507512520868114e-06, "loss": 0.3313, "step": 9040 }, { "epoch": 0.45, "learning_rate": 8.505843071786312e-06, "loss": 0.599, "step": 9050 }, { "epoch": 0.45, "learning_rate": 8.504173622704508e-06, "loss": 0.3652, "step": 9060 }, { "epoch": 0.45, "learning_rate": 8.502504173622706e-06, "loss": 0.4152, "step": 9070 }, { "epoch": 0.45, "learning_rate": 8.500834724540902e-06, "loss": 0.3384, "step": 9080 }, { "epoch": 0.45, "learning_rate": 8.4991652754591e-06, "loss": 0.3947, "step": 9090 }, { "epoch": 0.46, "learning_rate": 8.497495826377296e-06, "loss": 0.3373, "step": 9100 }, { "epoch": 0.46, "learning_rate": 8.495826377295494e-06, "loss": 0.427, "step": 9110 }, { "epoch": 0.46, "learning_rate": 8.49415692821369e-06, "loss": 0.345, "step": 9120 }, { "epoch": 0.46, "learning_rate": 8.492487479131886e-06, "loss": 0.3779, "step": 9130 }, { "epoch": 0.46, "learning_rate": 8.490818030050084e-06, "loss": 0.3821, "step": 9140 }, { "epoch": 0.46, "learning_rate": 8.489148580968282e-06, "loss": 0.3678, "step": 9150 }, { "epoch": 0.46, "learning_rate": 8.487479131886478e-06, "loss": 0.408, "step": 9160 }, { "epoch": 0.46, "learning_rate": 8.485809682804676e-06, "loss": 0.452, "step": 9170 }, { "epoch": 0.46, "learning_rate": 8.484140233722872e-06, "loss": 0.3926, "step": 9180 }, { "epoch": 0.46, "learning_rate": 8.48247078464107e-06, "loss": 0.4906, "step": 9190 }, { "epoch": 0.46, "learning_rate": 8.480801335559266e-06, "loss": 0.4476, "step": 9200 }, { "epoch": 0.46, "learning_rate": 8.479131886477464e-06, "loss": 0.3875, "step": 9210 }, { "epoch": 0.46, "learning_rate": 8.47746243739566e-06, "loss": 0.3445, "step": 9220 }, { "epoch": 0.46, "learning_rate": 8.475792988313856e-06, "loss": 0.4153, "step": 9230 }, { "epoch": 0.46, "learning_rate": 8.474123539232054e-06, "loss": 0.423, "step": 9240 }, { "epoch": 0.46, "learning_rate": 8.472454090150252e-06, "loss": 0.344, "step": 9250 }, { "epoch": 0.46, "learning_rate": 8.470784641068448e-06, "loss": 0.4018, "step": 9260 }, { "epoch": 0.46, "learning_rate": 8.469115191986646e-06, "loss": 0.4611, "step": 9270 }, { "epoch": 0.46, "learning_rate": 8.467445742904842e-06, "loss": 0.3768, "step": 9280 }, { "epoch": 0.46, "learning_rate": 8.46577629382304e-06, "loss": 0.3956, "step": 9290 }, { "epoch": 0.47, "learning_rate": 8.464106844741236e-06, "loss": 0.3834, "step": 9300 }, { "epoch": 0.47, "learning_rate": 8.462437395659434e-06, "loss": 0.3594, "step": 9310 }, { "epoch": 0.47, "learning_rate": 8.46076794657763e-06, "loss": 0.3121, "step": 9320 }, { "epoch": 0.47, "learning_rate": 8.459098497495826e-06, "loss": 0.3833, "step": 9330 }, { "epoch": 0.47, "learning_rate": 8.457429048414024e-06, "loss": 0.3774, "step": 9340 }, { "epoch": 0.47, "learning_rate": 8.45575959933222e-06, "loss": 0.3995, "step": 9350 }, { "epoch": 0.47, "learning_rate": 8.454090150250418e-06, "loss": 0.426, "step": 9360 }, { "epoch": 0.47, "learning_rate": 8.452420701168616e-06, "loss": 0.4023, "step": 9370 }, { "epoch": 0.47, "learning_rate": 8.450751252086812e-06, "loss": 0.4103, "step": 9380 }, { "epoch": 0.47, "learning_rate": 8.44908180300501e-06, "loss": 0.3491, "step": 9390 }, { "epoch": 0.47, "learning_rate": 8.447412353923206e-06, "loss": 0.3569, "step": 9400 }, { "epoch": 0.47, "learning_rate": 8.445742904841404e-06, "loss": 0.3141, "step": 9410 }, { "epoch": 0.47, "learning_rate": 8.4440734557596e-06, "loss": 0.3564, "step": 9420 }, { "epoch": 0.47, "learning_rate": 8.442404006677798e-06, "loss": 0.4325, "step": 9430 }, { "epoch": 0.47, "learning_rate": 8.440734557595994e-06, "loss": 0.3758, "step": 9440 }, { "epoch": 0.47, "learning_rate": 8.43906510851419e-06, "loss": 0.3384, "step": 9450 }, { "epoch": 0.47, "learning_rate": 8.437395659432388e-06, "loss": 0.4235, "step": 9460 }, { "epoch": 0.47, "learning_rate": 8.435726210350586e-06, "loss": 0.414, "step": 9470 }, { "epoch": 0.47, "learning_rate": 8.434056761268782e-06, "loss": 0.4053, "step": 9480 }, { "epoch": 0.47, "learning_rate": 8.43238731218698e-06, "loss": 0.4203, "step": 9490 }, { "epoch": 0.47, "learning_rate": 8.430717863105176e-06, "loss": 0.412, "step": 9500 }, { "epoch": 0.48, "learning_rate": 8.429048414023374e-06, "loss": 0.4267, "step": 9510 }, { "epoch": 0.48, "learning_rate": 8.42737896494157e-06, "loss": 0.3941, "step": 9520 }, { "epoch": 0.48, "learning_rate": 8.425709515859768e-06, "loss": 0.4241, "step": 9530 }, { "epoch": 0.48, "learning_rate": 8.424040066777964e-06, "loss": 0.3224, "step": 9540 }, { "epoch": 0.48, "learning_rate": 8.42237061769616e-06, "loss": 0.44, "step": 9550 }, { "epoch": 0.48, "learning_rate": 8.420701168614358e-06, "loss": 0.4282, "step": 9560 }, { "epoch": 0.48, "learning_rate": 8.419031719532556e-06, "loss": 0.385, "step": 9570 }, { "epoch": 0.48, "learning_rate": 8.417362270450752e-06, "loss": 0.4217, "step": 9580 }, { "epoch": 0.48, "learning_rate": 8.41569282136895e-06, "loss": 0.4471, "step": 9590 }, { "epoch": 0.48, "learning_rate": 8.414023372287146e-06, "loss": 0.4494, "step": 9600 }, { "epoch": 0.48, "learning_rate": 8.412353923205344e-06, "loss": 0.3831, "step": 9610 }, { "epoch": 0.48, "learning_rate": 8.41068447412354e-06, "loss": 0.3045, "step": 9620 }, { "epoch": 0.48, "learning_rate": 8.409015025041738e-06, "loss": 0.3897, "step": 9630 }, { "epoch": 0.48, "learning_rate": 8.407345575959934e-06, "loss": 0.3383, "step": 9640 }, { "epoch": 0.48, "learning_rate": 8.40567612687813e-06, "loss": 0.4272, "step": 9650 }, { "epoch": 0.48, "learning_rate": 8.404006677796328e-06, "loss": 0.2665, "step": 9660 }, { "epoch": 0.48, "learning_rate": 8.402337228714524e-06, "loss": 0.4692, "step": 9670 }, { "epoch": 0.48, "learning_rate": 8.400667779632722e-06, "loss": 0.3741, "step": 9680 }, { "epoch": 0.48, "learning_rate": 8.39899833055092e-06, "loss": 0.418, "step": 9690 }, { "epoch": 0.48, "learning_rate": 8.397328881469116e-06, "loss": 0.3962, "step": 9700 }, { "epoch": 0.49, "learning_rate": 8.395659432387314e-06, "loss": 0.354, "step": 9710 }, { "epoch": 0.49, "learning_rate": 8.39398998330551e-06, "loss": 0.4196, "step": 9720 }, { "epoch": 0.49, "learning_rate": 8.392320534223708e-06, "loss": 0.395, "step": 9730 }, { "epoch": 0.49, "learning_rate": 8.390651085141904e-06, "loss": 0.3567, "step": 9740 }, { "epoch": 0.49, "learning_rate": 8.3889816360601e-06, "loss": 0.386, "step": 9750 }, { "epoch": 0.49, "learning_rate": 8.387312186978298e-06, "loss": 0.3807, "step": 9760 }, { "epoch": 0.49, "learning_rate": 8.385642737896494e-06, "loss": 0.4587, "step": 9770 }, { "epoch": 0.49, "learning_rate": 8.383973288814692e-06, "loss": 0.3454, "step": 9780 }, { "epoch": 0.49, "learning_rate": 8.38230383973289e-06, "loss": 0.3076, "step": 9790 }, { "epoch": 0.49, "learning_rate": 8.380634390651086e-06, "loss": 0.3335, "step": 9800 }, { "epoch": 0.49, "learning_rate": 8.378964941569284e-06, "loss": 0.4272, "step": 9810 }, { "epoch": 0.49, "learning_rate": 8.37729549248748e-06, "loss": 0.4162, "step": 9820 }, { "epoch": 0.49, "learning_rate": 8.375626043405678e-06, "loss": 0.3771, "step": 9830 }, { "epoch": 0.49, "learning_rate": 8.373956594323874e-06, "loss": 0.3341, "step": 9840 }, { "epoch": 0.49, "learning_rate": 8.37228714524207e-06, "loss": 0.3798, "step": 9850 }, { "epoch": 0.49, "learning_rate": 8.370617696160268e-06, "loss": 0.4319, "step": 9860 }, { "epoch": 0.49, "learning_rate": 8.368948247078464e-06, "loss": 0.317, "step": 9870 }, { "epoch": 0.49, "learning_rate": 8.367278797996662e-06, "loss": 0.4418, "step": 9880 }, { "epoch": 0.49, "learning_rate": 8.36560934891486e-06, "loss": 0.3748, "step": 9890 }, { "epoch": 0.49, "learning_rate": 8.363939899833056e-06, "loss": 0.3604, "step": 9900 }, { "epoch": 0.5, "learning_rate": 8.362270450751254e-06, "loss": 0.4267, "step": 9910 }, { "epoch": 0.5, "learning_rate": 8.36060100166945e-06, "loss": 0.4307, "step": 9920 }, { "epoch": 0.5, "learning_rate": 8.358931552587648e-06, "loss": 0.4506, "step": 9930 }, { "epoch": 0.5, "learning_rate": 8.357262103505844e-06, "loss": 0.4502, "step": 9940 }, { "epoch": 0.5, "learning_rate": 8.35559265442404e-06, "loss": 0.4088, "step": 9950 }, { "epoch": 0.5, "learning_rate": 8.353923205342238e-06, "loss": 0.3746, "step": 9960 }, { "epoch": 0.5, "learning_rate": 8.352253756260434e-06, "loss": 0.398, "step": 9970 }, { "epoch": 0.5, "learning_rate": 8.350584307178632e-06, "loss": 0.3632, "step": 9980 }, { "epoch": 0.5, "learning_rate": 8.348914858096828e-06, "loss": 0.5537, "step": 9990 }, { "epoch": 0.5, "learning_rate": 8.347245409015026e-06, "loss": 0.3458, "step": 10000 }, { "epoch": 0.5, "eval_loss": 0.38069474697113037, "eval_runtime": 184.4189, "eval_samples_per_second": 20.415, "eval_steps_per_second": 2.554, "step": 10000 }, { "epoch": 0.5, "learning_rate": 8.345575959933224e-06, "loss": 0.2902, "step": 10010 }, { "epoch": 0.5, "learning_rate": 8.34390651085142e-06, "loss": 0.3501, "step": 10020 }, { "epoch": 0.5, "learning_rate": 8.342237061769618e-06, "loss": 0.3571, "step": 10030 }, { "epoch": 0.5, "learning_rate": 8.340567612687814e-06, "loss": 0.5465, "step": 10040 }, { "epoch": 0.5, "learning_rate": 8.33889816360601e-06, "loss": 0.4729, "step": 10050 }, { "epoch": 0.5, "learning_rate": 8.337228714524208e-06, "loss": 0.3289, "step": 10060 }, { "epoch": 0.5, "learning_rate": 8.335559265442404e-06, "loss": 0.4584, "step": 10070 }, { "epoch": 0.5, "learning_rate": 8.333889816360602e-06, "loss": 0.3755, "step": 10080 }, { "epoch": 0.5, "learning_rate": 8.332220367278798e-06, "loss": 0.353, "step": 10090 }, { "epoch": 0.51, "learning_rate": 8.330550918196996e-06, "loss": 0.3474, "step": 10100 }, { "epoch": 0.51, "learning_rate": 8.328881469115194e-06, "loss": 0.3698, "step": 10110 }, { "epoch": 0.51, "learning_rate": 8.32721202003339e-06, "loss": 0.3941, "step": 10120 }, { "epoch": 0.51, "learning_rate": 8.325542570951588e-06, "loss": 0.298, "step": 10130 }, { "epoch": 0.51, "learning_rate": 8.323873121869784e-06, "loss": 0.3456, "step": 10140 }, { "epoch": 0.51, "learning_rate": 8.322203672787982e-06, "loss": 0.388, "step": 10150 }, { "epoch": 0.51, "learning_rate": 8.320534223706178e-06, "loss": 0.3277, "step": 10160 }, { "epoch": 0.51, "learning_rate": 8.318864774624374e-06, "loss": 0.2837, "step": 10170 }, { "epoch": 0.51, "learning_rate": 8.317195325542572e-06, "loss": 0.3969, "step": 10180 }, { "epoch": 0.51, "learning_rate": 8.315525876460768e-06, "loss": 0.3424, "step": 10190 }, { "epoch": 0.51, "learning_rate": 8.313856427378966e-06, "loss": 0.3854, "step": 10200 }, { "epoch": 0.51, "learning_rate": 8.312186978297164e-06, "loss": 0.4195, "step": 10210 }, { "epoch": 0.51, "learning_rate": 8.31051752921536e-06, "loss": 0.4197, "step": 10220 }, { "epoch": 0.51, "learning_rate": 8.308848080133558e-06, "loss": 0.4629, "step": 10230 }, { "epoch": 0.51, "learning_rate": 8.307178631051754e-06, "loss": 0.5038, "step": 10240 }, { "epoch": 0.51, "learning_rate": 8.305509181969951e-06, "loss": 0.3759, "step": 10250 }, { "epoch": 0.51, "learning_rate": 8.303839732888148e-06, "loss": 0.3454, "step": 10260 }, { "epoch": 0.51, "learning_rate": 8.302170283806344e-06, "loss": 0.4301, "step": 10270 }, { "epoch": 0.51, "learning_rate": 8.300500834724542e-06, "loss": 0.4046, "step": 10280 }, { "epoch": 0.51, "learning_rate": 8.298831385642738e-06, "loss": 0.358, "step": 10290 }, { "epoch": 0.52, "learning_rate": 8.297161936560936e-06, "loss": 0.5255, "step": 10300 }, { "epoch": 0.52, "learning_rate": 8.295492487479132e-06, "loss": 0.4464, "step": 10310 }, { "epoch": 0.52, "learning_rate": 8.29382303839733e-06, "loss": 0.3637, "step": 10320 }, { "epoch": 0.52, "learning_rate": 8.292153589315527e-06, "loss": 0.4166, "step": 10330 }, { "epoch": 0.52, "learning_rate": 8.290484140233724e-06, "loss": 0.345, "step": 10340 }, { "epoch": 0.52, "learning_rate": 8.288814691151921e-06, "loss": 0.4724, "step": 10350 }, { "epoch": 0.52, "learning_rate": 8.287145242070118e-06, "loss": 0.3677, "step": 10360 }, { "epoch": 0.52, "learning_rate": 8.285475792988314e-06, "loss": 0.3492, "step": 10370 }, { "epoch": 0.52, "learning_rate": 8.283806343906512e-06, "loss": 0.3335, "step": 10380 }, { "epoch": 0.52, "learning_rate": 8.282136894824708e-06, "loss": 0.3419, "step": 10390 }, { "epoch": 0.52, "learning_rate": 8.280467445742906e-06, "loss": 0.3968, "step": 10400 }, { "epoch": 0.52, "learning_rate": 8.278797996661102e-06, "loss": 0.3871, "step": 10410 }, { "epoch": 0.52, "learning_rate": 8.2771285475793e-06, "loss": 0.3227, "step": 10420 }, { "epoch": 0.52, "learning_rate": 8.275459098497497e-06, "loss": 0.4791, "step": 10430 }, { "epoch": 0.52, "learning_rate": 8.273789649415694e-06, "loss": 0.3959, "step": 10440 }, { "epoch": 0.52, "learning_rate": 8.272120200333891e-06, "loss": 0.3529, "step": 10450 }, { "epoch": 0.52, "learning_rate": 8.270450751252088e-06, "loss": 0.4229, "step": 10460 }, { "epoch": 0.52, "learning_rate": 8.268781302170284e-06, "loss": 0.3399, "step": 10470 }, { "epoch": 0.52, "learning_rate": 8.267111853088482e-06, "loss": 0.36, "step": 10480 }, { "epoch": 0.52, "learning_rate": 8.265442404006678e-06, "loss": 0.3815, "step": 10490 }, { "epoch": 0.53, "learning_rate": 8.263772954924876e-06, "loss": 0.3984, "step": 10500 }, { "epoch": 0.53, "learning_rate": 8.262103505843072e-06, "loss": 0.4764, "step": 10510 }, { "epoch": 0.53, "learning_rate": 8.26043405676127e-06, "loss": 0.3743, "step": 10520 }, { "epoch": 0.53, "learning_rate": 8.258764607679467e-06, "loss": 0.3669, "step": 10530 }, { "epoch": 0.53, "learning_rate": 8.257095158597664e-06, "loss": 0.4868, "step": 10540 }, { "epoch": 0.53, "learning_rate": 8.255425709515861e-06, "loss": 0.4815, "step": 10550 }, { "epoch": 0.53, "learning_rate": 8.253756260434058e-06, "loss": 0.299, "step": 10560 }, { "epoch": 0.53, "learning_rate": 8.252086811352254e-06, "loss": 0.3269, "step": 10570 }, { "epoch": 0.53, "learning_rate": 8.250417362270452e-06, "loss": 0.4607, "step": 10580 }, { "epoch": 0.53, "learning_rate": 8.248747913188648e-06, "loss": 0.4275, "step": 10590 }, { "epoch": 0.53, "learning_rate": 8.247078464106846e-06, "loss": 0.4445, "step": 10600 }, { "epoch": 0.53, "learning_rate": 8.245409015025042e-06, "loss": 0.3994, "step": 10610 }, { "epoch": 0.53, "learning_rate": 8.24373956594324e-06, "loss": 0.48, "step": 10620 }, { "epoch": 0.53, "learning_rate": 8.242070116861436e-06, "loss": 0.3805, "step": 10630 }, { "epoch": 0.53, "learning_rate": 8.240400667779634e-06, "loss": 0.3919, "step": 10640 }, { "epoch": 0.53, "learning_rate": 8.238731218697831e-06, "loss": 0.4744, "step": 10650 }, { "epoch": 0.53, "learning_rate": 8.237061769616028e-06, "loss": 0.4242, "step": 10660 }, { "epoch": 0.53, "learning_rate": 8.235392320534224e-06, "loss": 0.3636, "step": 10670 }, { "epoch": 0.53, "learning_rate": 8.233722871452422e-06, "loss": 0.4841, "step": 10680 }, { "epoch": 0.53, "learning_rate": 8.232053422370618e-06, "loss": 0.3323, "step": 10690 }, { "epoch": 0.54, "learning_rate": 8.230383973288816e-06, "loss": 0.3675, "step": 10700 }, { "epoch": 0.54, "learning_rate": 8.228714524207012e-06, "loss": 0.3769, "step": 10710 }, { "epoch": 0.54, "learning_rate": 8.22704507512521e-06, "loss": 0.3465, "step": 10720 }, { "epoch": 0.54, "learning_rate": 8.225375626043406e-06, "loss": 0.3431, "step": 10730 }, { "epoch": 0.54, "learning_rate": 8.223706176961604e-06, "loss": 0.5196, "step": 10740 }, { "epoch": 0.54, "learning_rate": 8.222036727879801e-06, "loss": 0.4042, "step": 10750 }, { "epoch": 0.54, "learning_rate": 8.220367278797998e-06, "loss": 0.3905, "step": 10760 }, { "epoch": 0.54, "learning_rate": 8.218697829716195e-06, "loss": 0.3951, "step": 10770 }, { "epoch": 0.54, "learning_rate": 8.217028380634392e-06, "loss": 0.3365, "step": 10780 }, { "epoch": 0.54, "learning_rate": 8.215358931552588e-06, "loss": 0.3524, "step": 10790 }, { "epoch": 0.54, "learning_rate": 8.213689482470786e-06, "loss": 0.4662, "step": 10800 }, { "epoch": 0.54, "learning_rate": 8.212020033388982e-06, "loss": 0.3024, "step": 10810 }, { "epoch": 0.54, "learning_rate": 8.21035058430718e-06, "loss": 0.4558, "step": 10820 }, { "epoch": 0.54, "learning_rate": 8.208681135225376e-06, "loss": 0.345, "step": 10830 }, { "epoch": 0.54, "learning_rate": 8.207011686143574e-06, "loss": 0.3682, "step": 10840 }, { "epoch": 0.54, "learning_rate": 8.205342237061771e-06, "loss": 0.3272, "step": 10850 }, { "epoch": 0.54, "learning_rate": 8.203672787979968e-06, "loss": 0.3249, "step": 10860 }, { "epoch": 0.54, "learning_rate": 8.202003338898165e-06, "loss": 0.4094, "step": 10870 }, { "epoch": 0.54, "learning_rate": 8.200333889816362e-06, "loss": 0.4097, "step": 10880 }, { "epoch": 0.54, "learning_rate": 8.198664440734558e-06, "loss": 0.431, "step": 10890 }, { "epoch": 0.55, "learning_rate": 8.196994991652756e-06, "loss": 0.3887, "step": 10900 }, { "epoch": 0.55, "learning_rate": 8.195325542570952e-06, "loss": 0.3914, "step": 10910 }, { "epoch": 0.55, "learning_rate": 8.19365609348915e-06, "loss": 0.4004, "step": 10920 }, { "epoch": 0.55, "learning_rate": 8.191986644407346e-06, "loss": 0.3223, "step": 10930 }, { "epoch": 0.55, "learning_rate": 8.190317195325544e-06, "loss": 0.4355, "step": 10940 }, { "epoch": 0.55, "learning_rate": 8.18864774624374e-06, "loss": 0.4683, "step": 10950 }, { "epoch": 0.55, "learning_rate": 8.186978297161938e-06, "loss": 0.3989, "step": 10960 }, { "epoch": 0.55, "learning_rate": 8.185308848080135e-06, "loss": 0.3403, "step": 10970 }, { "epoch": 0.55, "learning_rate": 8.183639398998332e-06, "loss": 0.4732, "step": 10980 }, { "epoch": 0.55, "learning_rate": 8.181969949916528e-06, "loss": 0.3301, "step": 10990 }, { "epoch": 0.55, "learning_rate": 8.180300500834726e-06, "loss": 0.4183, "step": 11000 }, { "epoch": 0.55, "learning_rate": 8.178631051752922e-06, "loss": 0.4286, "step": 11010 }, { "epoch": 0.55, "learning_rate": 8.17696160267112e-06, "loss": 0.3863, "step": 11020 }, { "epoch": 0.55, "learning_rate": 8.175292153589316e-06, "loss": 0.3469, "step": 11030 }, { "epoch": 0.55, "learning_rate": 8.173622704507514e-06, "loss": 0.4152, "step": 11040 }, { "epoch": 0.55, "learning_rate": 8.17195325542571e-06, "loss": 0.4069, "step": 11050 }, { "epoch": 0.55, "learning_rate": 8.170283806343908e-06, "loss": 0.3999, "step": 11060 }, { "epoch": 0.55, "learning_rate": 8.168614357262105e-06, "loss": 0.2964, "step": 11070 }, { "epoch": 0.55, "learning_rate": 8.166944908180302e-06, "loss": 0.5428, "step": 11080 }, { "epoch": 0.55, "learning_rate": 8.165275459098498e-06, "loss": 0.3311, "step": 11090 }, { "epoch": 0.56, "learning_rate": 8.163606010016696e-06, "loss": 0.4355, "step": 11100 }, { "epoch": 0.56, "learning_rate": 8.161936560934892e-06, "loss": 0.2923, "step": 11110 }, { "epoch": 0.56, "learning_rate": 8.16026711185309e-06, "loss": 0.3814, "step": 11120 }, { "epoch": 0.56, "learning_rate": 8.158597662771286e-06, "loss": 0.3842, "step": 11130 }, { "epoch": 0.56, "learning_rate": 8.156928213689483e-06, "loss": 0.4068, "step": 11140 }, { "epoch": 0.56, "learning_rate": 8.15525876460768e-06, "loss": 0.4342, "step": 11150 }, { "epoch": 0.56, "learning_rate": 8.153589315525877e-06, "loss": 0.441, "step": 11160 }, { "epoch": 0.56, "learning_rate": 8.151919866444075e-06, "loss": 0.4318, "step": 11170 }, { "epoch": 0.56, "learning_rate": 8.150250417362271e-06, "loss": 0.3502, "step": 11180 }, { "epoch": 0.56, "learning_rate": 8.148580968280468e-06, "loss": 0.4251, "step": 11190 }, { "epoch": 0.56, "learning_rate": 8.146911519198665e-06, "loss": 0.4455, "step": 11200 }, { "epoch": 0.56, "learning_rate": 8.145242070116862e-06, "loss": 0.3798, "step": 11210 }, { "epoch": 0.56, "learning_rate": 8.14357262103506e-06, "loss": 0.3855, "step": 11220 }, { "epoch": 0.56, "learning_rate": 8.141903171953256e-06, "loss": 0.4531, "step": 11230 }, { "epoch": 0.56, "learning_rate": 8.140233722871453e-06, "loss": 0.3879, "step": 11240 }, { "epoch": 0.56, "learning_rate": 8.13856427378965e-06, "loss": 0.4053, "step": 11250 }, { "epoch": 0.56, "learning_rate": 8.136894824707847e-06, "loss": 0.3926, "step": 11260 }, { "epoch": 0.56, "learning_rate": 8.135225375626044e-06, "loss": 0.376, "step": 11270 }, { "epoch": 0.56, "learning_rate": 8.133555926544241e-06, "loss": 0.4974, "step": 11280 }, { "epoch": 0.56, "learning_rate": 8.131886477462438e-06, "loss": 0.4052, "step": 11290 }, { "epoch": 0.56, "learning_rate": 8.130217028380635e-06, "loss": 0.4488, "step": 11300 }, { "epoch": 0.57, "learning_rate": 8.128547579298832e-06, "loss": 0.4021, "step": 11310 }, { "epoch": 0.57, "learning_rate": 8.126878130217028e-06, "loss": 0.4723, "step": 11320 }, { "epoch": 0.57, "learning_rate": 8.125208681135226e-06, "loss": 0.3869, "step": 11330 }, { "epoch": 0.57, "learning_rate": 8.123539232053423e-06, "loss": 0.5151, "step": 11340 }, { "epoch": 0.57, "learning_rate": 8.12186978297162e-06, "loss": 0.3616, "step": 11350 }, { "epoch": 0.57, "learning_rate": 8.120200333889817e-06, "loss": 0.31, "step": 11360 }, { "epoch": 0.57, "learning_rate": 8.118530884808014e-06, "loss": 0.3305, "step": 11370 }, { "epoch": 0.57, "learning_rate": 8.116861435726211e-06, "loss": 0.4077, "step": 11380 }, { "epoch": 0.57, "learning_rate": 8.115191986644408e-06, "loss": 0.5223, "step": 11390 }, { "epoch": 0.57, "learning_rate": 8.113522537562605e-06, "loss": 0.3532, "step": 11400 }, { "epoch": 0.57, "learning_rate": 8.111853088480802e-06, "loss": 0.3226, "step": 11410 }, { "epoch": 0.57, "learning_rate": 8.110183639398998e-06, "loss": 0.4349, "step": 11420 }, { "epoch": 0.57, "learning_rate": 8.108514190317196e-06, "loss": 0.4081, "step": 11430 }, { "epoch": 0.57, "learning_rate": 8.106844741235393e-06, "loss": 0.4267, "step": 11440 }, { "epoch": 0.57, "learning_rate": 8.10517529215359e-06, "loss": 0.3414, "step": 11450 }, { "epoch": 0.57, "learning_rate": 8.103505843071787e-06, "loss": 0.4323, "step": 11460 }, { "epoch": 0.57, "learning_rate": 8.101836393989984e-06, "loss": 0.2984, "step": 11470 }, { "epoch": 0.57, "learning_rate": 8.100166944908181e-06, "loss": 0.3889, "step": 11480 }, { "epoch": 0.57, "learning_rate": 8.09849749582638e-06, "loss": 0.4248, "step": 11490 }, { "epoch": 0.57, "learning_rate": 8.096828046744575e-06, "loss": 0.3328, "step": 11500 }, { "epoch": 0.58, "learning_rate": 8.095158597662772e-06, "loss": 0.4141, "step": 11510 }, { "epoch": 0.58, "learning_rate": 8.093489148580968e-06, "loss": 0.3295, "step": 11520 }, { "epoch": 0.58, "learning_rate": 8.091819699499166e-06, "loss": 0.3546, "step": 11530 }, { "epoch": 0.58, "learning_rate": 8.090150250417363e-06, "loss": 0.3916, "step": 11540 }, { "epoch": 0.58, "learning_rate": 8.08848080133556e-06, "loss": 0.4436, "step": 11550 }, { "epoch": 0.58, "learning_rate": 8.086811352253757e-06, "loss": 0.2931, "step": 11560 }, { "epoch": 0.58, "learning_rate": 8.085141903171954e-06, "loss": 0.3909, "step": 11570 }, { "epoch": 0.58, "learning_rate": 8.083472454090151e-06, "loss": 0.3863, "step": 11580 }, { "epoch": 0.58, "learning_rate": 8.081803005008348e-06, "loss": 0.4058, "step": 11590 }, { "epoch": 0.58, "learning_rate": 8.080133555926545e-06, "loss": 0.3971, "step": 11600 }, { "epoch": 0.58, "learning_rate": 8.078464106844742e-06, "loss": 0.3951, "step": 11610 }, { "epoch": 0.58, "learning_rate": 8.07679465776294e-06, "loss": 0.4471, "step": 11620 }, { "epoch": 0.58, "learning_rate": 8.075125208681136e-06, "loss": 0.3805, "step": 11630 }, { "epoch": 0.58, "learning_rate": 8.073455759599332e-06, "loss": 0.3132, "step": 11640 }, { "epoch": 0.58, "learning_rate": 8.07178631051753e-06, "loss": 0.5229, "step": 11650 }, { "epoch": 0.58, "learning_rate": 8.070116861435727e-06, "loss": 0.4572, "step": 11660 }, { "epoch": 0.58, "learning_rate": 8.068447412353924e-06, "loss": 0.3775, "step": 11670 }, { "epoch": 0.58, "learning_rate": 8.066777963272121e-06, "loss": 0.4769, "step": 11680 }, { "epoch": 0.58, "learning_rate": 8.065108514190318e-06, "loss": 0.4174, "step": 11690 }, { "epoch": 0.58, "learning_rate": 8.063439065108515e-06, "loss": 0.3669, "step": 11700 }, { "epoch": 0.59, "learning_rate": 8.061769616026712e-06, "loss": 0.3719, "step": 11710 }, { "epoch": 0.59, "learning_rate": 8.06010016694491e-06, "loss": 0.3634, "step": 11720 }, { "epoch": 0.59, "learning_rate": 8.058430717863106e-06, "loss": 0.4737, "step": 11730 }, { "epoch": 0.59, "learning_rate": 8.056761268781302e-06, "loss": 0.4148, "step": 11740 }, { "epoch": 0.59, "learning_rate": 8.0550918196995e-06, "loss": 0.3869, "step": 11750 }, { "epoch": 0.59, "learning_rate": 8.053422370617697e-06, "loss": 0.4322, "step": 11760 }, { "epoch": 0.59, "learning_rate": 8.051752921535894e-06, "loss": 0.3735, "step": 11770 }, { "epoch": 0.59, "learning_rate": 8.050083472454091e-06, "loss": 0.4328, "step": 11780 }, { "epoch": 0.59, "learning_rate": 8.048414023372288e-06, "loss": 0.344, "step": 11790 }, { "epoch": 0.59, "learning_rate": 8.046744574290485e-06, "loss": 0.4009, "step": 11800 }, { "epoch": 0.59, "learning_rate": 8.045075125208682e-06, "loss": 0.3384, "step": 11810 }, { "epoch": 0.59, "learning_rate": 8.04340567612688e-06, "loss": 0.3443, "step": 11820 }, { "epoch": 0.59, "learning_rate": 8.041736227045076e-06, "loss": 0.3361, "step": 11830 }, { "epoch": 0.59, "learning_rate": 8.040066777963272e-06, "loss": 0.4494, "step": 11840 }, { "epoch": 0.59, "learning_rate": 8.03839732888147e-06, "loss": 0.4197, "step": 11850 }, { "epoch": 0.59, "learning_rate": 8.036727879799667e-06, "loss": 0.3757, "step": 11860 }, { "epoch": 0.59, "learning_rate": 8.035058430717864e-06, "loss": 0.3861, "step": 11870 }, { "epoch": 0.59, "learning_rate": 8.033388981636061e-06, "loss": 0.4942, "step": 11880 }, { "epoch": 0.59, "learning_rate": 8.031719532554258e-06, "loss": 0.3938, "step": 11890 }, { "epoch": 0.59, "learning_rate": 8.030050083472455e-06, "loss": 0.3808, "step": 11900 }, { "epoch": 0.6, "learning_rate": 8.028380634390652e-06, "loss": 0.4464, "step": 11910 }, { "epoch": 0.6, "learning_rate": 8.02671118530885e-06, "loss": 0.3572, "step": 11920 }, { "epoch": 0.6, "learning_rate": 8.025041736227046e-06, "loss": 0.3903, "step": 11930 }, { "epoch": 0.6, "learning_rate": 8.023372287145242e-06, "loss": 0.3528, "step": 11940 }, { "epoch": 0.6, "learning_rate": 8.02170283806344e-06, "loss": 0.4904, "step": 11950 }, { "epoch": 0.6, "learning_rate": 8.020033388981636e-06, "loss": 0.4089, "step": 11960 }, { "epoch": 0.6, "learning_rate": 8.018363939899834e-06, "loss": 0.4206, "step": 11970 }, { "epoch": 0.6, "learning_rate": 8.016694490818031e-06, "loss": 0.3719, "step": 11980 }, { "epoch": 0.6, "learning_rate": 8.015025041736228e-06, "loss": 0.3863, "step": 11990 }, { "epoch": 0.6, "learning_rate": 8.013355592654425e-06, "loss": 0.3894, "step": 12000 }, { "epoch": 0.6, "learning_rate": 8.011686143572621e-06, "loss": 0.3916, "step": 12010 }, { "epoch": 0.6, "learning_rate": 8.01001669449082e-06, "loss": 0.4084, "step": 12020 }, { "epoch": 0.6, "learning_rate": 8.008347245409015e-06, "loss": 0.4018, "step": 12030 }, { "epoch": 0.6, "learning_rate": 8.006677796327212e-06, "loss": 0.3522, "step": 12040 }, { "epoch": 0.6, "learning_rate": 8.00500834724541e-06, "loss": 0.4606, "step": 12050 }, { "epoch": 0.6, "learning_rate": 8.003338898163606e-06, "loss": 0.4579, "step": 12060 }, { "epoch": 0.6, "learning_rate": 8.001669449081803e-06, "loss": 0.457, "step": 12070 }, { "epoch": 0.6, "learning_rate": 8.000000000000001e-06, "loss": 0.4589, "step": 12080 }, { "epoch": 0.6, "learning_rate": 7.998330550918197e-06, "loss": 0.412, "step": 12090 }, { "epoch": 0.6, "learning_rate": 7.996661101836395e-06, "loss": 0.3298, "step": 12100 }, { "epoch": 0.61, "learning_rate": 7.994991652754591e-06, "loss": 0.4738, "step": 12110 }, { "epoch": 0.61, "learning_rate": 7.99332220367279e-06, "loss": 0.4287, "step": 12120 }, { "epoch": 0.61, "learning_rate": 7.991652754590985e-06, "loss": 0.2963, "step": 12130 }, { "epoch": 0.61, "learning_rate": 7.989983305509182e-06, "loss": 0.349, "step": 12140 }, { "epoch": 0.61, "learning_rate": 7.98831385642738e-06, "loss": 0.4809, "step": 12150 }, { "epoch": 0.61, "learning_rate": 7.986644407345576e-06, "loss": 0.3669, "step": 12160 }, { "epoch": 0.61, "learning_rate": 7.984974958263773e-06, "loss": 0.3669, "step": 12170 }, { "epoch": 0.61, "learning_rate": 7.983305509181971e-06, "loss": 0.3894, "step": 12180 }, { "epoch": 0.61, "learning_rate": 7.981636060100167e-06, "loss": 0.372, "step": 12190 }, { "epoch": 0.61, "learning_rate": 7.979966611018365e-06, "loss": 0.4082, "step": 12200 }, { "epoch": 0.61, "learning_rate": 7.978297161936561e-06, "loss": 0.3798, "step": 12210 }, { "epoch": 0.61, "learning_rate": 7.97662771285476e-06, "loss": 0.3631, "step": 12220 }, { "epoch": 0.61, "learning_rate": 7.974958263772955e-06, "loss": 0.4092, "step": 12230 }, { "epoch": 0.61, "learning_rate": 7.973288814691152e-06, "loss": 0.4401, "step": 12240 }, { "epoch": 0.61, "learning_rate": 7.97161936560935e-06, "loss": 0.3894, "step": 12250 }, { "epoch": 0.61, "learning_rate": 7.969949916527546e-06, "loss": 0.4294, "step": 12260 }, { "epoch": 0.61, "learning_rate": 7.968280467445743e-06, "loss": 0.3992, "step": 12270 }, { "epoch": 0.61, "learning_rate": 7.96661101836394e-06, "loss": 0.3788, "step": 12280 }, { "epoch": 0.61, "learning_rate": 7.964941569282137e-06, "loss": 0.4239, "step": 12290 }, { "epoch": 0.61, "learning_rate": 7.963272120200335e-06, "loss": 0.4044, "step": 12300 }, { "epoch": 0.62, "learning_rate": 7.961602671118531e-06, "loss": 0.349, "step": 12310 }, { "epoch": 0.62, "learning_rate": 7.95993322203673e-06, "loss": 0.3912, "step": 12320 }, { "epoch": 0.62, "learning_rate": 7.958263772954925e-06, "loss": 0.3599, "step": 12330 }, { "epoch": 0.62, "learning_rate": 7.956594323873123e-06, "loss": 0.2876, "step": 12340 }, { "epoch": 0.62, "learning_rate": 7.95492487479132e-06, "loss": 0.3974, "step": 12350 }, { "epoch": 0.62, "learning_rate": 7.953255425709516e-06, "loss": 0.4894, "step": 12360 }, { "epoch": 0.62, "learning_rate": 7.951585976627713e-06, "loss": 0.3409, "step": 12370 }, { "epoch": 0.62, "learning_rate": 7.94991652754591e-06, "loss": 0.3898, "step": 12380 }, { "epoch": 0.62, "learning_rate": 7.948247078464107e-06, "loss": 0.3214, "step": 12390 }, { "epoch": 0.62, "learning_rate": 7.946577629382305e-06, "loss": 0.3594, "step": 12400 }, { "epoch": 0.62, "learning_rate": 7.944908180300501e-06, "loss": 0.3405, "step": 12410 }, { "epoch": 0.62, "learning_rate": 7.9432387312187e-06, "loss": 0.3534, "step": 12420 }, { "epoch": 0.62, "learning_rate": 7.941569282136895e-06, "loss": 0.4078, "step": 12430 }, { "epoch": 0.62, "learning_rate": 7.939899833055093e-06, "loss": 0.3627, "step": 12440 }, { "epoch": 0.62, "learning_rate": 7.93823038397329e-06, "loss": 0.3755, "step": 12450 }, { "epoch": 0.62, "learning_rate": 7.936560934891486e-06, "loss": 0.2935, "step": 12460 }, { "epoch": 0.62, "learning_rate": 7.934891485809683e-06, "loss": 0.3387, "step": 12470 }, { "epoch": 0.62, "learning_rate": 7.93322203672788e-06, "loss": 0.4726, "step": 12480 }, { "epoch": 0.62, "learning_rate": 7.931552587646077e-06, "loss": 0.4536, "step": 12490 }, { "epoch": 0.62, "learning_rate": 7.929883138564275e-06, "loss": 0.394, "step": 12500 }, { "epoch": 0.63, "learning_rate": 7.928213689482471e-06, "loss": 0.4016, "step": 12510 }, { "epoch": 0.63, "learning_rate": 7.92654424040067e-06, "loss": 0.3682, "step": 12520 }, { "epoch": 0.63, "learning_rate": 7.924874791318865e-06, "loss": 0.3176, "step": 12530 }, { "epoch": 0.63, "learning_rate": 7.923205342237063e-06, "loss": 0.3754, "step": 12540 }, { "epoch": 0.63, "learning_rate": 7.92153589315526e-06, "loss": 0.338, "step": 12550 }, { "epoch": 0.63, "learning_rate": 7.919866444073456e-06, "loss": 0.39, "step": 12560 }, { "epoch": 0.63, "learning_rate": 7.918196994991653e-06, "loss": 0.3368, "step": 12570 }, { "epoch": 0.63, "learning_rate": 7.91652754590985e-06, "loss": 0.3813, "step": 12580 }, { "epoch": 0.63, "learning_rate": 7.914858096828047e-06, "loss": 0.4432, "step": 12590 }, { "epoch": 0.63, "learning_rate": 7.913188647746244e-06, "loss": 0.446, "step": 12600 }, { "epoch": 0.63, "learning_rate": 7.911519198664441e-06, "loss": 0.3717, "step": 12610 }, { "epoch": 0.63, "learning_rate": 7.90984974958264e-06, "loss": 0.3784, "step": 12620 }, { "epoch": 0.63, "learning_rate": 7.908180300500835e-06, "loss": 0.3579, "step": 12630 }, { "epoch": 0.63, "learning_rate": 7.906510851419033e-06, "loss": 0.3976, "step": 12640 }, { "epoch": 0.63, "learning_rate": 7.90484140233723e-06, "loss": 0.3576, "step": 12650 }, { "epoch": 0.63, "learning_rate": 7.903171953255426e-06, "loss": 0.3513, "step": 12660 }, { "epoch": 0.63, "learning_rate": 7.901502504173623e-06, "loss": 0.4423, "step": 12670 }, { "epoch": 0.63, "learning_rate": 7.89983305509182e-06, "loss": 0.3565, "step": 12680 }, { "epoch": 0.63, "learning_rate": 7.898163606010017e-06, "loss": 0.3201, "step": 12690 }, { "epoch": 0.64, "learning_rate": 7.896494156928214e-06, "loss": 0.3805, "step": 12700 }, { "epoch": 0.64, "learning_rate": 7.894824707846411e-06, "loss": 0.4078, "step": 12710 }, { "epoch": 0.64, "learning_rate": 7.89315525876461e-06, "loss": 0.3573, "step": 12720 }, { "epoch": 0.64, "learning_rate": 7.891485809682805e-06, "loss": 0.3847, "step": 12730 }, { "epoch": 0.64, "learning_rate": 7.889816360601003e-06, "loss": 0.4121, "step": 12740 }, { "epoch": 0.64, "learning_rate": 7.8881469115192e-06, "loss": 0.353, "step": 12750 }, { "epoch": 0.64, "learning_rate": 7.886477462437396e-06, "loss": 0.371, "step": 12760 }, { "epoch": 0.64, "learning_rate": 7.884808013355593e-06, "loss": 0.4005, "step": 12770 }, { "epoch": 0.64, "learning_rate": 7.88313856427379e-06, "loss": 0.4261, "step": 12780 }, { "epoch": 0.64, "learning_rate": 7.881469115191987e-06, "loss": 0.3958, "step": 12790 }, { "epoch": 0.64, "learning_rate": 7.879799666110184e-06, "loss": 0.4767, "step": 12800 }, { "epoch": 0.64, "learning_rate": 7.878130217028381e-06, "loss": 0.399, "step": 12810 }, { "epoch": 0.64, "learning_rate": 7.87646076794658e-06, "loss": 0.3242, "step": 12820 }, { "epoch": 0.64, "learning_rate": 7.874791318864775e-06, "loss": 0.3949, "step": 12830 }, { "epoch": 0.64, "learning_rate": 7.873121869782973e-06, "loss": 0.3868, "step": 12840 }, { "epoch": 0.64, "learning_rate": 7.87145242070117e-06, "loss": 0.3766, "step": 12850 }, { "epoch": 0.64, "learning_rate": 7.869782971619366e-06, "loss": 0.4418, "step": 12860 }, { "epoch": 0.64, "learning_rate": 7.868113522537563e-06, "loss": 0.2976, "step": 12870 }, { "epoch": 0.64, "learning_rate": 7.86644407345576e-06, "loss": 0.4407, "step": 12880 }, { "epoch": 0.64, "learning_rate": 7.864774624373957e-06, "loss": 0.3324, "step": 12890 }, { "epoch": 0.65, "learning_rate": 7.863105175292153e-06, "loss": 0.3226, "step": 12900 }, { "epoch": 0.65, "learning_rate": 7.861435726210351e-06, "loss": 0.4269, "step": 12910 }, { "epoch": 0.65, "learning_rate": 7.859766277128547e-06, "loss": 0.35, "step": 12920 }, { "epoch": 0.65, "learning_rate": 7.858096828046745e-06, "loss": 0.3187, "step": 12930 }, { "epoch": 0.65, "learning_rate": 7.856427378964943e-06, "loss": 0.426, "step": 12940 }, { "epoch": 0.65, "learning_rate": 7.85475792988314e-06, "loss": 0.3124, "step": 12950 }, { "epoch": 0.65, "learning_rate": 7.853088480801337e-06, "loss": 0.5231, "step": 12960 }, { "epoch": 0.65, "learning_rate": 7.851419031719533e-06, "loss": 0.3809, "step": 12970 }, { "epoch": 0.65, "learning_rate": 7.84974958263773e-06, "loss": 0.3448, "step": 12980 }, { "epoch": 0.65, "learning_rate": 7.848080133555927e-06, "loss": 0.3561, "step": 12990 }, { "epoch": 0.65, "learning_rate": 7.846410684474123e-06, "loss": 0.3877, "step": 13000 }, { "epoch": 0.65, "learning_rate": 7.844741235392321e-06, "loss": 0.4077, "step": 13010 }, { "epoch": 0.65, "learning_rate": 7.843071786310517e-06, "loss": 0.4171, "step": 13020 }, { "epoch": 0.65, "learning_rate": 7.841402337228715e-06, "loss": 0.4205, "step": 13030 }, { "epoch": 0.65, "learning_rate": 7.839732888146913e-06, "loss": 0.3631, "step": 13040 }, { "epoch": 0.65, "learning_rate": 7.83806343906511e-06, "loss": 0.3848, "step": 13050 }, { "epoch": 0.65, "learning_rate": 7.836393989983307e-06, "loss": 0.3989, "step": 13060 }, { "epoch": 0.65, "learning_rate": 7.834724540901503e-06, "loss": 0.3409, "step": 13070 }, { "epoch": 0.65, "learning_rate": 7.8330550918197e-06, "loss": 0.4712, "step": 13080 }, { "epoch": 0.65, "learning_rate": 7.831385642737897e-06, "loss": 0.3694, "step": 13090 }, { "epoch": 0.66, "learning_rate": 7.829716193656093e-06, "loss": 0.3481, "step": 13100 }, { "epoch": 0.66, "learning_rate": 7.828046744574291e-06, "loss": 0.427, "step": 13110 }, { "epoch": 0.66, "learning_rate": 7.826377295492487e-06, "loss": 0.4211, "step": 13120 }, { "epoch": 0.66, "learning_rate": 7.824707846410685e-06, "loss": 0.3363, "step": 13130 }, { "epoch": 0.66, "learning_rate": 7.823038397328883e-06, "loss": 0.4434, "step": 13140 }, { "epoch": 0.66, "learning_rate": 7.82136894824708e-06, "loss": 0.4368, "step": 13150 }, { "epoch": 0.66, "learning_rate": 7.819699499165277e-06, "loss": 0.3965, "step": 13160 }, { "epoch": 0.66, "learning_rate": 7.818030050083473e-06, "loss": 0.4019, "step": 13170 }, { "epoch": 0.66, "learning_rate": 7.81636060100167e-06, "loss": 0.3089, "step": 13180 }, { "epoch": 0.66, "learning_rate": 7.814691151919867e-06, "loss": 0.4309, "step": 13190 }, { "epoch": 0.66, "learning_rate": 7.813021702838063e-06, "loss": 0.4065, "step": 13200 }, { "epoch": 0.66, "learning_rate": 7.811352253756261e-06, "loss": 0.369, "step": 13210 }, { "epoch": 0.66, "learning_rate": 7.809682804674457e-06, "loss": 0.3508, "step": 13220 }, { "epoch": 0.66, "learning_rate": 7.808013355592655e-06, "loss": 0.4474, "step": 13230 }, { "epoch": 0.66, "learning_rate": 7.806343906510853e-06, "loss": 0.3608, "step": 13240 }, { "epoch": 0.66, "learning_rate": 7.80467445742905e-06, "loss": 0.2977, "step": 13250 }, { "epoch": 0.66, "learning_rate": 7.803005008347247e-06, "loss": 0.507, "step": 13260 }, { "epoch": 0.66, "learning_rate": 7.801335559265443e-06, "loss": 0.4032, "step": 13270 }, { "epoch": 0.66, "learning_rate": 7.79966611018364e-06, "loss": 0.3416, "step": 13280 }, { "epoch": 0.66, "learning_rate": 7.797996661101837e-06, "loss": 0.433, "step": 13290 }, { "epoch": 0.67, "learning_rate": 7.796327212020033e-06, "loss": 0.348, "step": 13300 }, { "epoch": 0.67, "learning_rate": 7.794657762938231e-06, "loss": 0.4263, "step": 13310 }, { "epoch": 0.67, "learning_rate": 7.792988313856427e-06, "loss": 0.3841, "step": 13320 }, { "epoch": 0.67, "learning_rate": 7.791318864774625e-06, "loss": 0.3937, "step": 13330 }, { "epoch": 0.67, "learning_rate": 7.789649415692821e-06, "loss": 0.2953, "step": 13340 }, { "epoch": 0.67, "learning_rate": 7.78797996661102e-06, "loss": 0.4085, "step": 13350 }, { "epoch": 0.67, "learning_rate": 7.786310517529217e-06, "loss": 0.5142, "step": 13360 }, { "epoch": 0.67, "learning_rate": 7.784641068447413e-06, "loss": 0.364, "step": 13370 }, { "epoch": 0.67, "learning_rate": 7.78297161936561e-06, "loss": 0.3776, "step": 13380 }, { "epoch": 0.67, "learning_rate": 7.781302170283807e-06, "loss": 0.4024, "step": 13390 }, { "epoch": 0.67, "learning_rate": 7.779632721202003e-06, "loss": 0.3796, "step": 13400 }, { "epoch": 0.67, "learning_rate": 7.777963272120201e-06, "loss": 0.4538, "step": 13410 }, { "epoch": 0.67, "learning_rate": 7.776293823038397e-06, "loss": 0.3157, "step": 13420 }, { "epoch": 0.67, "learning_rate": 7.774624373956595e-06, "loss": 0.2965, "step": 13430 }, { "epoch": 0.67, "learning_rate": 7.772954924874791e-06, "loss": 0.4026, "step": 13440 }, { "epoch": 0.67, "learning_rate": 7.77128547579299e-06, "loss": 0.3322, "step": 13450 }, { "epoch": 0.67, "learning_rate": 7.769616026711187e-06, "loss": 0.3562, "step": 13460 }, { "epoch": 0.67, "learning_rate": 7.767946577629383e-06, "loss": 0.4377, "step": 13470 }, { "epoch": 0.67, "learning_rate": 7.76627712854758e-06, "loss": 0.461, "step": 13480 }, { "epoch": 0.67, "learning_rate": 7.764607679465777e-06, "loss": 0.4132, "step": 13490 }, { "epoch": 0.68, "learning_rate": 7.762938230383973e-06, "loss": 0.3444, "step": 13500 }, { "epoch": 0.68, "learning_rate": 7.761268781302171e-06, "loss": 0.2974, "step": 13510 }, { "epoch": 0.68, "learning_rate": 7.759599332220367e-06, "loss": 0.4139, "step": 13520 }, { "epoch": 0.68, "learning_rate": 7.757929883138565e-06, "loss": 0.4502, "step": 13530 }, { "epoch": 0.68, "learning_rate": 7.756260434056761e-06, "loss": 0.3415, "step": 13540 }, { "epoch": 0.68, "learning_rate": 7.75459098497496e-06, "loss": 0.2772, "step": 13550 }, { "epoch": 0.68, "learning_rate": 7.752921535893157e-06, "loss": 0.3941, "step": 13560 }, { "epoch": 0.68, "learning_rate": 7.751252086811353e-06, "loss": 0.3577, "step": 13570 }, { "epoch": 0.68, "learning_rate": 7.74958263772955e-06, "loss": 0.3931, "step": 13580 }, { "epoch": 0.68, "learning_rate": 7.747913188647747e-06, "loss": 0.4202, "step": 13590 }, { "epoch": 0.68, "learning_rate": 7.746243739565943e-06, "loss": 0.3749, "step": 13600 }, { "epoch": 0.68, "learning_rate": 7.74457429048414e-06, "loss": 0.379, "step": 13610 }, { "epoch": 0.68, "learning_rate": 7.742904841402337e-06, "loss": 0.4631, "step": 13620 }, { "epoch": 0.68, "learning_rate": 7.741235392320535e-06, "loss": 0.3803, "step": 13630 }, { "epoch": 0.68, "learning_rate": 7.739565943238731e-06, "loss": 0.3751, "step": 13640 }, { "epoch": 0.68, "learning_rate": 7.73789649415693e-06, "loss": 0.492, "step": 13650 }, { "epoch": 0.68, "learning_rate": 7.736227045075125e-06, "loss": 0.3785, "step": 13660 }, { "epoch": 0.68, "learning_rate": 7.734557595993323e-06, "loss": 0.2972, "step": 13670 }, { "epoch": 0.68, "learning_rate": 7.732888146911521e-06, "loss": 0.3385, "step": 13680 }, { "epoch": 0.68, "learning_rate": 7.731218697829717e-06, "loss": 0.471, "step": 13690 }, { "epoch": 0.69, "learning_rate": 7.729549248747913e-06, "loss": 0.4823, "step": 13700 }, { "epoch": 0.69, "learning_rate": 7.72787979966611e-06, "loss": 0.4057, "step": 13710 }, { "epoch": 0.69, "learning_rate": 7.726210350584307e-06, "loss": 0.3133, "step": 13720 }, { "epoch": 0.69, "learning_rate": 7.724540901502505e-06, "loss": 0.3879, "step": 13730 }, { "epoch": 0.69, "learning_rate": 7.722871452420701e-06, "loss": 0.3877, "step": 13740 }, { "epoch": 0.69, "learning_rate": 7.7212020033389e-06, "loss": 0.3671, "step": 13750 }, { "epoch": 0.69, "learning_rate": 7.719532554257095e-06, "loss": 0.5334, "step": 13760 }, { "epoch": 0.69, "learning_rate": 7.717863105175293e-06, "loss": 0.4814, "step": 13770 }, { "epoch": 0.69, "learning_rate": 7.716193656093491e-06, "loss": 0.375, "step": 13780 }, { "epoch": 0.69, "learning_rate": 7.714524207011687e-06, "loss": 0.4615, "step": 13790 }, { "epoch": 0.69, "learning_rate": 7.712854757929883e-06, "loss": 0.484, "step": 13800 }, { "epoch": 0.69, "learning_rate": 7.711185308848081e-06, "loss": 0.3841, "step": 13810 }, { "epoch": 0.69, "learning_rate": 7.709515859766277e-06, "loss": 0.4154, "step": 13820 }, { "epoch": 0.69, "learning_rate": 7.707846410684475e-06, "loss": 0.2951, "step": 13830 }, { "epoch": 0.69, "learning_rate": 7.706176961602671e-06, "loss": 0.4277, "step": 13840 }, { "epoch": 0.69, "learning_rate": 7.70450751252087e-06, "loss": 0.3575, "step": 13850 }, { "epoch": 0.69, "learning_rate": 7.702838063439065e-06, "loss": 0.3944, "step": 13860 }, { "epoch": 0.69, "learning_rate": 7.701168614357263e-06, "loss": 0.3936, "step": 13870 }, { "epoch": 0.69, "learning_rate": 7.699499165275461e-06, "loss": 0.3627, "step": 13880 }, { "epoch": 0.69, "learning_rate": 7.697829716193657e-06, "loss": 0.4787, "step": 13890 }, { "epoch": 0.69, "learning_rate": 7.696160267111853e-06, "loss": 0.3044, "step": 13900 }, { "epoch": 0.7, "learning_rate": 7.694490818030051e-06, "loss": 0.4487, "step": 13910 }, { "epoch": 0.7, "learning_rate": 7.692821368948247e-06, "loss": 0.3092, "step": 13920 }, { "epoch": 0.7, "learning_rate": 7.691151919866443e-06, "loss": 0.4705, "step": 13930 }, { "epoch": 0.7, "learning_rate": 7.689482470784641e-06, "loss": 0.383, "step": 13940 }, { "epoch": 0.7, "learning_rate": 7.68781302170284e-06, "loss": 0.3868, "step": 13950 }, { "epoch": 0.7, "learning_rate": 7.686143572621035e-06, "loss": 0.3551, "step": 13960 }, { "epoch": 0.7, "learning_rate": 7.684474123539233e-06, "loss": 0.5262, "step": 13970 }, { "epoch": 0.7, "learning_rate": 7.68280467445743e-06, "loss": 0.3406, "step": 13980 }, { "epoch": 0.7, "learning_rate": 7.681135225375627e-06, "loss": 0.3808, "step": 13990 }, { "epoch": 0.7, "learning_rate": 7.679465776293823e-06, "loss": 0.4656, "step": 14000 }, { "epoch": 0.7, "learning_rate": 7.677796327212021e-06, "loss": 0.3563, "step": 14010 }, { "epoch": 0.7, "learning_rate": 7.676126878130217e-06, "loss": 0.3454, "step": 14020 }, { "epoch": 0.7, "learning_rate": 7.674457429048413e-06, "loss": 0.4795, "step": 14030 }, { "epoch": 0.7, "learning_rate": 7.672787979966611e-06, "loss": 0.4164, "step": 14040 }, { "epoch": 0.7, "learning_rate": 7.671118530884809e-06, "loss": 0.3734, "step": 14050 }, { "epoch": 0.7, "learning_rate": 7.669449081803005e-06, "loss": 0.4503, "step": 14060 }, { "epoch": 0.7, "learning_rate": 7.667779632721203e-06, "loss": 0.3637, "step": 14070 }, { "epoch": 0.7, "learning_rate": 7.6661101836394e-06, "loss": 0.3994, "step": 14080 }, { "epoch": 0.7, "learning_rate": 7.664440734557597e-06, "loss": 0.3132, "step": 14090 }, { "epoch": 0.7, "learning_rate": 7.662771285475793e-06, "loss": 0.5049, "step": 14100 }, { "epoch": 0.71, "learning_rate": 7.661101836393991e-06, "loss": 0.3483, "step": 14110 }, { "epoch": 0.71, "learning_rate": 7.659432387312187e-06, "loss": 0.4276, "step": 14120 }, { "epoch": 0.71, "learning_rate": 7.657762938230383e-06, "loss": 0.3831, "step": 14130 }, { "epoch": 0.71, "learning_rate": 7.656093489148581e-06, "loss": 0.3884, "step": 14140 }, { "epoch": 0.71, "learning_rate": 7.654424040066779e-06, "loss": 0.3815, "step": 14150 }, { "epoch": 0.71, "learning_rate": 7.652754590984975e-06, "loss": 0.5004, "step": 14160 }, { "epoch": 0.71, "learning_rate": 7.651085141903173e-06, "loss": 0.4059, "step": 14170 }, { "epoch": 0.71, "learning_rate": 7.64941569282137e-06, "loss": 0.3354, "step": 14180 }, { "epoch": 0.71, "learning_rate": 7.647746243739567e-06, "loss": 0.3957, "step": 14190 }, { "epoch": 0.71, "learning_rate": 7.646076794657763e-06, "loss": 0.4054, "step": 14200 }, { "epoch": 0.71, "learning_rate": 7.644407345575961e-06, "loss": 0.4452, "step": 14210 }, { "epoch": 0.71, "learning_rate": 7.642737896494157e-06, "loss": 0.3459, "step": 14220 }, { "epoch": 0.71, "learning_rate": 7.641068447412353e-06, "loss": 0.2902, "step": 14230 }, { "epoch": 0.71, "learning_rate": 7.639398998330551e-06, "loss": 0.4002, "step": 14240 }, { "epoch": 0.71, "learning_rate": 7.637729549248747e-06, "loss": 0.3814, "step": 14250 }, { "epoch": 0.71, "learning_rate": 7.636060100166945e-06, "loss": 0.3189, "step": 14260 }, { "epoch": 0.71, "learning_rate": 7.634390651085143e-06, "loss": 0.3714, "step": 14270 }, { "epoch": 0.71, "learning_rate": 7.63272120200334e-06, "loss": 0.3602, "step": 14280 }, { "epoch": 0.71, "learning_rate": 7.631051752921537e-06, "loss": 0.3872, "step": 14290 }, { "epoch": 0.71, "learning_rate": 7.629382303839733e-06, "loss": 0.4177, "step": 14300 }, { "epoch": 0.72, "learning_rate": 7.62771285475793e-06, "loss": 0.4364, "step": 14310 }, { "epoch": 0.72, "learning_rate": 7.626043405676128e-06, "loss": 0.3294, "step": 14320 }, { "epoch": 0.72, "learning_rate": 7.624373956594324e-06, "loss": 0.2781, "step": 14330 }, { "epoch": 0.72, "learning_rate": 7.622704507512522e-06, "loss": 0.3583, "step": 14340 }, { "epoch": 0.72, "learning_rate": 7.621035058430718e-06, "loss": 0.3839, "step": 14350 }, { "epoch": 0.72, "learning_rate": 7.619365609348915e-06, "loss": 0.3525, "step": 14360 }, { "epoch": 0.72, "learning_rate": 7.617696160267113e-06, "loss": 0.4823, "step": 14370 }, { "epoch": 0.72, "learning_rate": 7.616026711185309e-06, "loss": 0.3763, "step": 14380 }, { "epoch": 0.72, "learning_rate": 7.614357262103507e-06, "loss": 0.5069, "step": 14390 }, { "epoch": 0.72, "learning_rate": 7.612687813021703e-06, "loss": 0.3747, "step": 14400 }, { "epoch": 0.72, "learning_rate": 7.6110183639399e-06, "loss": 0.4476, "step": 14410 }, { "epoch": 0.72, "learning_rate": 7.609348914858098e-06, "loss": 0.3839, "step": 14420 }, { "epoch": 0.72, "learning_rate": 7.607679465776294e-06, "loss": 0.3917, "step": 14430 }, { "epoch": 0.72, "learning_rate": 7.606010016694492e-06, "loss": 0.4181, "step": 14440 }, { "epoch": 0.72, "learning_rate": 7.604340567612688e-06, "loss": 0.3924, "step": 14450 }, { "epoch": 0.72, "learning_rate": 7.602671118530885e-06, "loss": 0.4387, "step": 14460 }, { "epoch": 0.72, "learning_rate": 7.601001669449083e-06, "loss": 0.4462, "step": 14470 }, { "epoch": 0.72, "learning_rate": 7.599332220367279e-06, "loss": 0.3239, "step": 14480 }, { "epoch": 0.72, "learning_rate": 7.597662771285477e-06, "loss": 0.4532, "step": 14490 }, { "epoch": 0.72, "learning_rate": 7.595993322203673e-06, "loss": 0.3908, "step": 14500 }, { "epoch": 0.73, "learning_rate": 7.59432387312187e-06, "loss": 0.3997, "step": 14510 }, { "epoch": 0.73, "learning_rate": 7.592654424040068e-06, "loss": 0.3916, "step": 14520 }, { "epoch": 0.73, "learning_rate": 7.590984974958264e-06, "loss": 0.4185, "step": 14530 }, { "epoch": 0.73, "learning_rate": 7.589315525876462e-06, "loss": 0.3769, "step": 14540 }, { "epoch": 0.73, "learning_rate": 7.587646076794658e-06, "loss": 0.482, "step": 14550 }, { "epoch": 0.73, "learning_rate": 7.585976627712855e-06, "loss": 0.4794, "step": 14560 }, { "epoch": 0.73, "learning_rate": 7.584307178631053e-06, "loss": 0.3834, "step": 14570 }, { "epoch": 0.73, "learning_rate": 7.582637729549249e-06, "loss": 0.3418, "step": 14580 }, { "epoch": 0.73, "learning_rate": 7.580968280467447e-06, "loss": 0.3825, "step": 14590 }, { "epoch": 0.73, "learning_rate": 7.579298831385643e-06, "loss": 0.4911, "step": 14600 }, { "epoch": 0.73, "learning_rate": 7.57762938230384e-06, "loss": 0.3071, "step": 14610 }, { "epoch": 0.73, "learning_rate": 7.575959933222037e-06, "loss": 0.3786, "step": 14620 }, { "epoch": 0.73, "learning_rate": 7.574290484140234e-06, "loss": 0.4412, "step": 14630 }, { "epoch": 0.73, "learning_rate": 7.572621035058432e-06, "loss": 0.361, "step": 14640 }, { "epoch": 0.73, "learning_rate": 7.570951585976628e-06, "loss": 0.3896, "step": 14650 }, { "epoch": 0.73, "learning_rate": 7.569282136894825e-06, "loss": 0.318, "step": 14660 }, { "epoch": 0.73, "learning_rate": 7.567612687813022e-06, "loss": 0.4112, "step": 14670 }, { "epoch": 0.73, "learning_rate": 7.565943238731219e-06, "loss": 0.369, "step": 14680 }, { "epoch": 0.73, "learning_rate": 7.564273789649417e-06, "loss": 0.3053, "step": 14690 }, { "epoch": 0.73, "learning_rate": 7.562604340567613e-06, "loss": 0.3404, "step": 14700 }, { "epoch": 0.74, "learning_rate": 7.56093489148581e-06, "loss": 0.4312, "step": 14710 }, { "epoch": 0.74, "learning_rate": 7.559265442404007e-06, "loss": 0.3724, "step": 14720 }, { "epoch": 0.74, "learning_rate": 7.557595993322204e-06, "loss": 0.3512, "step": 14730 }, { "epoch": 0.74, "learning_rate": 7.555926544240402e-06, "loss": 0.3706, "step": 14740 }, { "epoch": 0.74, "learning_rate": 7.554257095158598e-06, "loss": 0.4368, "step": 14750 }, { "epoch": 0.74, "learning_rate": 7.552587646076795e-06, "loss": 0.3115, "step": 14760 }, { "epoch": 0.74, "learning_rate": 7.550918196994992e-06, "loss": 0.4026, "step": 14770 }, { "epoch": 0.74, "learning_rate": 7.549248747913189e-06, "loss": 0.3404, "step": 14780 }, { "epoch": 0.74, "learning_rate": 7.547579298831387e-06, "loss": 0.3114, "step": 14790 }, { "epoch": 0.74, "learning_rate": 7.545909849749583e-06, "loss": 0.3473, "step": 14800 }, { "epoch": 0.74, "learning_rate": 7.54424040066778e-06, "loss": 0.4178, "step": 14810 }, { "epoch": 0.74, "learning_rate": 7.542570951585977e-06, "loss": 0.37, "step": 14820 }, { "epoch": 0.74, "learning_rate": 7.540901502504174e-06, "loss": 0.4108, "step": 14830 }, { "epoch": 0.74, "learning_rate": 7.539232053422372e-06, "loss": 0.5165, "step": 14840 }, { "epoch": 0.74, "learning_rate": 7.537562604340568e-06, "loss": 0.3791, "step": 14850 }, { "epoch": 0.74, "learning_rate": 7.535893155258765e-06, "loss": 0.4366, "step": 14860 }, { "epoch": 0.74, "learning_rate": 7.534223706176962e-06, "loss": 0.3985, "step": 14870 }, { "epoch": 0.74, "learning_rate": 7.532554257095159e-06, "loss": 0.4625, "step": 14880 }, { "epoch": 0.74, "learning_rate": 7.530884808013357e-06, "loss": 0.4019, "step": 14890 }, { "epoch": 0.74, "learning_rate": 7.529215358931553e-06, "loss": 0.3936, "step": 14900 }, { "epoch": 0.75, "learning_rate": 7.52754590984975e-06, "loss": 0.396, "step": 14910 }, { "epoch": 0.75, "learning_rate": 7.525876460767947e-06, "loss": 0.4337, "step": 14920 }, { "epoch": 0.75, "learning_rate": 7.524207011686144e-06, "loss": 0.3088, "step": 14930 }, { "epoch": 0.75, "learning_rate": 7.52253756260434e-06, "loss": 0.3837, "step": 14940 }, { "epoch": 0.75, "learning_rate": 7.520868113522538e-06, "loss": 0.4189, "step": 14950 }, { "epoch": 0.75, "learning_rate": 7.519198664440735e-06, "loss": 0.3277, "step": 14960 }, { "epoch": 0.75, "learning_rate": 7.517529215358932e-06, "loss": 0.3682, "step": 14970 }, { "epoch": 0.75, "learning_rate": 7.515859766277129e-06, "loss": 0.3541, "step": 14980 }, { "epoch": 0.75, "learning_rate": 7.514190317195325e-06, "loss": 0.4105, "step": 14990 }, { "epoch": 0.75, "learning_rate": 7.512520868113523e-06, "loss": 0.3438, "step": 15000 }, { "epoch": 0.75, "eval_loss": 0.37906408309936523, "eval_runtime": 182.8451, "eval_samples_per_second": 20.591, "eval_steps_per_second": 2.576, "step": 15000 }, { "epoch": 0.75, "learning_rate": 7.51085141903172e-06, "loss": 0.3749, "step": 15010 }, { "epoch": 0.75, "learning_rate": 7.509181969949917e-06, "loss": 0.4087, "step": 15020 }, { "epoch": 0.75, "learning_rate": 7.507512520868114e-06, "loss": 0.4665, "step": 15030 }, { "epoch": 0.75, "learning_rate": 7.50584307178631e-06, "loss": 0.4322, "step": 15040 }, { "epoch": 0.75, "learning_rate": 7.504173622704508e-06, "loss": 0.4155, "step": 15050 }, { "epoch": 0.75, "learning_rate": 7.502504173622706e-06, "loss": 0.4803, "step": 15060 }, { "epoch": 0.75, "learning_rate": 7.500834724540902e-06, "loss": 0.3769, "step": 15070 }, { "epoch": 0.75, "learning_rate": 7.499165275459099e-06, "loss": 0.3665, "step": 15080 }, { "epoch": 0.75, "learning_rate": 7.497495826377295e-06, "loss": 0.4106, "step": 15090 }, { "epoch": 0.76, "learning_rate": 7.495826377295493e-06, "loss": 0.4153, "step": 15100 }, { "epoch": 0.76, "learning_rate": 7.494156928213691e-06, "loss": 0.4248, "step": 15110 }, { "epoch": 0.76, "learning_rate": 7.492487479131887e-06, "loss": 0.4436, "step": 15120 }, { "epoch": 0.76, "learning_rate": 7.490818030050084e-06, "loss": 0.4089, "step": 15130 }, { "epoch": 0.76, "learning_rate": 7.489148580968281e-06, "loss": 0.3166, "step": 15140 }, { "epoch": 0.76, "learning_rate": 7.487479131886478e-06, "loss": 0.3951, "step": 15150 }, { "epoch": 0.76, "learning_rate": 7.485809682804676e-06, "loss": 0.3133, "step": 15160 }, { "epoch": 0.76, "learning_rate": 7.484140233722872e-06, "loss": 0.3862, "step": 15170 }, { "epoch": 0.76, "learning_rate": 7.482470784641069e-06, "loss": 0.359, "step": 15180 }, { "epoch": 0.76, "learning_rate": 7.480801335559266e-06, "loss": 0.4478, "step": 15190 }, { "epoch": 0.76, "learning_rate": 7.479131886477463e-06, "loss": 0.3424, "step": 15200 }, { "epoch": 0.76, "learning_rate": 7.477462437395661e-06, "loss": 0.3078, "step": 15210 }, { "epoch": 0.76, "learning_rate": 7.475792988313857e-06, "loss": 0.3438, "step": 15220 }, { "epoch": 0.76, "learning_rate": 7.474123539232054e-06, "loss": 0.4155, "step": 15230 }, { "epoch": 0.76, "learning_rate": 7.472454090150251e-06, "loss": 0.3518, "step": 15240 }, { "epoch": 0.76, "learning_rate": 7.470784641068448e-06, "loss": 0.3259, "step": 15250 }, { "epoch": 0.76, "learning_rate": 7.469115191986644e-06, "loss": 0.4215, "step": 15260 }, { "epoch": 0.76, "learning_rate": 7.467445742904842e-06, "loss": 0.4889, "step": 15270 }, { "epoch": 0.76, "learning_rate": 7.465776293823039e-06, "loss": 0.3695, "step": 15280 }, { "epoch": 0.76, "learning_rate": 7.464106844741236e-06, "loss": 0.3468, "step": 15290 }, { "epoch": 0.77, "learning_rate": 7.462437395659433e-06, "loss": 0.3705, "step": 15300 }, { "epoch": 0.77, "learning_rate": 7.460767946577629e-06, "loss": 0.399, "step": 15310 }, { "epoch": 0.77, "learning_rate": 7.459098497495827e-06, "loss": 0.4325, "step": 15320 }, { "epoch": 0.77, "learning_rate": 7.457429048414024e-06, "loss": 0.3788, "step": 15330 }, { "epoch": 0.77, "learning_rate": 7.455759599332221e-06, "loss": 0.4285, "step": 15340 }, { "epoch": 0.77, "learning_rate": 7.454090150250418e-06, "loss": 0.4002, "step": 15350 }, { "epoch": 0.77, "learning_rate": 7.452420701168614e-06, "loss": 0.4012, "step": 15360 }, { "epoch": 0.77, "learning_rate": 7.450751252086812e-06, "loss": 0.4545, "step": 15370 }, { "epoch": 0.77, "learning_rate": 7.449081803005009e-06, "loss": 0.3791, "step": 15380 }, { "epoch": 0.77, "learning_rate": 7.447412353923206e-06, "loss": 0.3596, "step": 15390 }, { "epoch": 0.77, "learning_rate": 7.445742904841403e-06, "loss": 0.4178, "step": 15400 }, { "epoch": 0.77, "learning_rate": 7.444073455759599e-06, "loss": 0.3857, "step": 15410 }, { "epoch": 0.77, "learning_rate": 7.442404006677797e-06, "loss": 0.3421, "step": 15420 }, { "epoch": 0.77, "learning_rate": 7.440734557595994e-06, "loss": 0.4462, "step": 15430 }, { "epoch": 0.77, "learning_rate": 7.439065108514191e-06, "loss": 0.4334, "step": 15440 }, { "epoch": 0.77, "learning_rate": 7.437395659432388e-06, "loss": 0.4754, "step": 15450 }, { "epoch": 0.77, "learning_rate": 7.435726210350584e-06, "loss": 0.417, "step": 15460 }, { "epoch": 0.77, "learning_rate": 7.434056761268782e-06, "loss": 0.4562, "step": 15470 }, { "epoch": 0.77, "learning_rate": 7.432387312186979e-06, "loss": 0.3445, "step": 15480 }, { "epoch": 0.77, "learning_rate": 7.430717863105176e-06, "loss": 0.4377, "step": 15490 }, { "epoch": 0.78, "learning_rate": 7.429048414023373e-06, "loss": 0.4248, "step": 15500 }, { "epoch": 0.78, "learning_rate": 7.427378964941569e-06, "loss": 0.3751, "step": 15510 }, { "epoch": 0.78, "learning_rate": 7.425709515859767e-06, "loss": 0.3643, "step": 15520 }, { "epoch": 0.78, "learning_rate": 7.424040066777964e-06, "loss": 0.2836, "step": 15530 }, { "epoch": 0.78, "learning_rate": 7.422370617696161e-06, "loss": 0.3436, "step": 15540 }, { "epoch": 0.78, "learning_rate": 7.420701168614358e-06, "loss": 0.3942, "step": 15550 }, { "epoch": 0.78, "learning_rate": 7.419031719532554e-06, "loss": 0.3962, "step": 15560 }, { "epoch": 0.78, "learning_rate": 7.417362270450752e-06, "loss": 0.396, "step": 15570 }, { "epoch": 0.78, "learning_rate": 7.415692821368948e-06, "loss": 0.3999, "step": 15580 }, { "epoch": 0.78, "learning_rate": 7.414023372287146e-06, "loss": 0.3248, "step": 15590 }, { "epoch": 0.78, "learning_rate": 7.412353923205343e-06, "loss": 0.4475, "step": 15600 }, { "epoch": 0.78, "learning_rate": 7.410684474123539e-06, "loss": 0.3715, "step": 15610 }, { "epoch": 0.78, "learning_rate": 7.409015025041737e-06, "loss": 0.459, "step": 15620 }, { "epoch": 0.78, "learning_rate": 7.407345575959933e-06, "loss": 0.3424, "step": 15630 }, { "epoch": 0.78, "learning_rate": 7.405676126878131e-06, "loss": 0.36, "step": 15640 }, { "epoch": 0.78, "learning_rate": 7.404006677796328e-06, "loss": 0.385, "step": 15650 }, { "epoch": 0.78, "learning_rate": 7.402337228714524e-06, "loss": 0.4067, "step": 15660 }, { "epoch": 0.78, "learning_rate": 7.400667779632722e-06, "loss": 0.4133, "step": 15670 }, { "epoch": 0.78, "learning_rate": 7.398998330550918e-06, "loss": 0.3695, "step": 15680 }, { "epoch": 0.78, "learning_rate": 7.397328881469116e-06, "loss": 0.3885, "step": 15690 }, { "epoch": 0.79, "learning_rate": 7.395659432387313e-06, "loss": 0.4041, "step": 15700 }, { "epoch": 0.79, "learning_rate": 7.393989983305509e-06, "loss": 0.4337, "step": 15710 }, { "epoch": 0.79, "learning_rate": 7.392320534223707e-06, "loss": 0.4224, "step": 15720 }, { "epoch": 0.79, "learning_rate": 7.390651085141903e-06, "loss": 0.3502, "step": 15730 }, { "epoch": 0.79, "learning_rate": 7.388981636060101e-06, "loss": 0.3727, "step": 15740 }, { "epoch": 0.79, "learning_rate": 7.387312186978298e-06, "loss": 0.3908, "step": 15750 }, { "epoch": 0.79, "learning_rate": 7.385642737896494e-06, "loss": 0.3463, "step": 15760 }, { "epoch": 0.79, "learning_rate": 7.383973288814692e-06, "loss": 0.3758, "step": 15770 }, { "epoch": 0.79, "learning_rate": 7.382303839732888e-06, "loss": 0.3396, "step": 15780 }, { "epoch": 0.79, "learning_rate": 7.380634390651086e-06, "loss": 0.359, "step": 15790 }, { "epoch": 0.79, "learning_rate": 7.378964941569283e-06, "loss": 0.485, "step": 15800 }, { "epoch": 0.79, "learning_rate": 7.377295492487479e-06, "loss": 0.3333, "step": 15810 }, { "epoch": 0.79, "learning_rate": 7.375626043405677e-06, "loss": 0.347, "step": 15820 }, { "epoch": 0.79, "learning_rate": 7.373956594323873e-06, "loss": 0.4146, "step": 15830 }, { "epoch": 0.79, "learning_rate": 7.372287145242071e-06, "loss": 0.3971, "step": 15840 }, { "epoch": 0.79, "learning_rate": 7.370617696160268e-06, "loss": 0.3355, "step": 15850 }, { "epoch": 0.79, "learning_rate": 7.368948247078465e-06, "loss": 0.4741, "step": 15860 }, { "epoch": 0.79, "learning_rate": 7.367278797996662e-06, "loss": 0.2863, "step": 15870 }, { "epoch": 0.79, "learning_rate": 7.365609348914858e-06, "loss": 0.3513, "step": 15880 }, { "epoch": 0.79, "learning_rate": 7.363939899833056e-06, "loss": 0.3965, "step": 15890 }, { "epoch": 0.8, "learning_rate": 7.362270450751253e-06, "loss": 0.3929, "step": 15900 }, { "epoch": 0.8, "learning_rate": 7.36060100166945e-06, "loss": 0.4381, "step": 15910 }, { "epoch": 0.8, "learning_rate": 7.358931552587647e-06, "loss": 0.4543, "step": 15920 }, { "epoch": 0.8, "learning_rate": 7.357262103505843e-06, "loss": 0.3406, "step": 15930 }, { "epoch": 0.8, "learning_rate": 7.355592654424041e-06, "loss": 0.4122, "step": 15940 }, { "epoch": 0.8, "learning_rate": 7.353923205342237e-06, "loss": 0.4582, "step": 15950 }, { "epoch": 0.8, "learning_rate": 7.352253756260435e-06, "loss": 0.4458, "step": 15960 }, { "epoch": 0.8, "learning_rate": 7.350584307178632e-06, "loss": 0.3682, "step": 15970 }, { "epoch": 0.8, "learning_rate": 7.348914858096828e-06, "loss": 0.4353, "step": 15980 }, { "epoch": 0.8, "learning_rate": 7.347245409015026e-06, "loss": 0.3337, "step": 15990 }, { "epoch": 0.8, "learning_rate": 7.345575959933222e-06, "loss": 0.319, "step": 16000 }, { "epoch": 0.8, "learning_rate": 7.34390651085142e-06, "loss": 0.5408, "step": 16010 }, { "epoch": 0.8, "learning_rate": 7.342237061769617e-06, "loss": 0.4273, "step": 16020 }, { "epoch": 0.8, "learning_rate": 7.340567612687813e-06, "loss": 0.4192, "step": 16030 }, { "epoch": 0.8, "learning_rate": 7.338898163606011e-06, "loss": 0.3672, "step": 16040 }, { "epoch": 0.8, "learning_rate": 7.337228714524207e-06, "loss": 0.4045, "step": 16050 }, { "epoch": 0.8, "learning_rate": 7.335559265442405e-06, "loss": 0.323, "step": 16060 }, { "epoch": 0.8, "learning_rate": 7.333889816360602e-06, "loss": 0.3792, "step": 16070 }, { "epoch": 0.8, "learning_rate": 7.332220367278798e-06, "loss": 0.4181, "step": 16080 }, { "epoch": 0.8, "learning_rate": 7.330550918196996e-06, "loss": 0.4309, "step": 16090 }, { "epoch": 0.81, "learning_rate": 7.328881469115192e-06, "loss": 0.316, "step": 16100 }, { "epoch": 0.81, "learning_rate": 7.32721202003339e-06, "loss": 0.3914, "step": 16110 }, { "epoch": 0.81, "learning_rate": 7.325542570951587e-06, "loss": 0.3755, "step": 16120 }, { "epoch": 0.81, "learning_rate": 7.323873121869783e-06, "loss": 0.3181, "step": 16130 }, { "epoch": 0.81, "learning_rate": 7.322203672787981e-06, "loss": 0.3251, "step": 16140 }, { "epoch": 0.81, "learning_rate": 7.320534223706177e-06, "loss": 0.423, "step": 16150 }, { "epoch": 0.81, "learning_rate": 7.318864774624375e-06, "loss": 0.341, "step": 16160 }, { "epoch": 0.81, "learning_rate": 7.317195325542572e-06, "loss": 0.3447, "step": 16170 }, { "epoch": 0.81, "learning_rate": 7.315525876460768e-06, "loss": 0.3511, "step": 16180 }, { "epoch": 0.81, "learning_rate": 7.313856427378966e-06, "loss": 0.3303, "step": 16190 }, { "epoch": 0.81, "learning_rate": 7.312186978297162e-06, "loss": 0.4211, "step": 16200 }, { "epoch": 0.81, "learning_rate": 7.31051752921536e-06, "loss": 0.3639, "step": 16210 }, { "epoch": 0.81, "learning_rate": 7.308848080133557e-06, "loss": 0.4535, "step": 16220 }, { "epoch": 0.81, "learning_rate": 7.307178631051753e-06, "loss": 0.4692, "step": 16230 }, { "epoch": 0.81, "learning_rate": 7.305509181969951e-06, "loss": 0.3608, "step": 16240 }, { "epoch": 0.81, "learning_rate": 7.303839732888147e-06, "loss": 0.4031, "step": 16250 }, { "epoch": 0.81, "learning_rate": 7.302170283806345e-06, "loss": 0.3327, "step": 16260 }, { "epoch": 0.81, "learning_rate": 7.300500834724541e-06, "loss": 0.3197, "step": 16270 }, { "epoch": 0.81, "learning_rate": 7.298831385642738e-06, "loss": 0.4126, "step": 16280 }, { "epoch": 0.81, "learning_rate": 7.297161936560936e-06, "loss": 0.4358, "step": 16290 }, { "epoch": 0.81, "learning_rate": 7.295492487479132e-06, "loss": 0.4012, "step": 16300 }, { "epoch": 0.82, "learning_rate": 7.29382303839733e-06, "loss": 0.5094, "step": 16310 }, { "epoch": 0.82, "learning_rate": 7.292153589315526e-06, "loss": 0.4188, "step": 16320 }, { "epoch": 0.82, "learning_rate": 7.290484140233723e-06, "loss": 0.601, "step": 16330 }, { "epoch": 0.82, "learning_rate": 7.288814691151921e-06, "loss": 0.3506, "step": 16340 }, { "epoch": 0.82, "learning_rate": 7.287145242070117e-06, "loss": 0.4195, "step": 16350 }, { "epoch": 0.82, "learning_rate": 7.285475792988315e-06, "loss": 0.3742, "step": 16360 }, { "epoch": 0.82, "learning_rate": 7.283806343906511e-06, "loss": 0.4225, "step": 16370 }, { "epoch": 0.82, "learning_rate": 7.282136894824708e-06, "loss": 0.4147, "step": 16380 }, { "epoch": 0.82, "learning_rate": 7.280467445742906e-06, "loss": 0.4327, "step": 16390 }, { "epoch": 0.82, "learning_rate": 7.278797996661102e-06, "loss": 0.3421, "step": 16400 }, { "epoch": 0.82, "learning_rate": 7.2771285475793e-06, "loss": 0.3708, "step": 16410 }, { "epoch": 0.82, "learning_rate": 7.275459098497496e-06, "loss": 0.3713, "step": 16420 }, { "epoch": 0.82, "learning_rate": 7.273789649415693e-06, "loss": 0.4599, "step": 16430 }, { "epoch": 0.82, "learning_rate": 7.272120200333891e-06, "loss": 0.3247, "step": 16440 }, { "epoch": 0.82, "learning_rate": 7.270450751252087e-06, "loss": 0.4287, "step": 16450 }, { "epoch": 0.82, "learning_rate": 7.268781302170285e-06, "loss": 0.37, "step": 16460 }, { "epoch": 0.82, "learning_rate": 7.267111853088481e-06, "loss": 0.4619, "step": 16470 }, { "epoch": 0.82, "learning_rate": 7.265442404006678e-06, "loss": 0.3606, "step": 16480 }, { "epoch": 0.82, "learning_rate": 7.263772954924876e-06, "loss": 0.4018, "step": 16490 }, { "epoch": 0.82, "learning_rate": 7.262103505843072e-06, "loss": 0.4256, "step": 16500 }, { "epoch": 0.83, "learning_rate": 7.26043405676127e-06, "loss": 0.2925, "step": 16510 }, { "epoch": 0.83, "learning_rate": 7.258764607679466e-06, "loss": 0.3646, "step": 16520 }, { "epoch": 0.83, "learning_rate": 7.257095158597664e-06, "loss": 0.3251, "step": 16530 }, { "epoch": 0.83, "learning_rate": 7.255425709515861e-06, "loss": 0.5023, "step": 16540 }, { "epoch": 0.83, "learning_rate": 7.253756260434057e-06, "loss": 0.3675, "step": 16550 }, { "epoch": 0.83, "learning_rate": 7.252086811352255e-06, "loss": 0.4154, "step": 16560 }, { "epoch": 0.83, "learning_rate": 7.250417362270451e-06, "loss": 0.3056, "step": 16570 }, { "epoch": 0.83, "learning_rate": 7.248747913188649e-06, "loss": 0.4704, "step": 16580 }, { "epoch": 0.83, "learning_rate": 7.247078464106845e-06, "loss": 0.5128, "step": 16590 }, { "epoch": 0.83, "learning_rate": 7.245409015025042e-06, "loss": 0.4049, "step": 16600 }, { "epoch": 0.83, "learning_rate": 7.24373956594324e-06, "loss": 0.4984, "step": 16610 }, { "epoch": 0.83, "learning_rate": 7.242070116861436e-06, "loss": 0.3347, "step": 16620 }, { "epoch": 0.83, "learning_rate": 7.240400667779634e-06, "loss": 0.4416, "step": 16630 }, { "epoch": 0.83, "learning_rate": 7.23873121869783e-06, "loss": 0.5527, "step": 16640 }, { "epoch": 0.83, "learning_rate": 7.237061769616027e-06, "loss": 0.4452, "step": 16650 }, { "epoch": 0.83, "learning_rate": 7.235392320534225e-06, "loss": 0.442, "step": 16660 }, { "epoch": 0.83, "learning_rate": 7.233722871452421e-06, "loss": 0.3509, "step": 16670 }, { "epoch": 0.83, "learning_rate": 7.232053422370619e-06, "loss": 0.4367, "step": 16680 }, { "epoch": 0.83, "learning_rate": 7.230383973288815e-06, "loss": 0.377, "step": 16690 }, { "epoch": 0.83, "learning_rate": 7.228714524207012e-06, "loss": 0.3698, "step": 16700 }, { "epoch": 0.84, "learning_rate": 7.22704507512521e-06, "loss": 0.3913, "step": 16710 }, { "epoch": 0.84, "learning_rate": 7.225375626043406e-06, "loss": 0.4257, "step": 16720 }, { "epoch": 0.84, "learning_rate": 7.223706176961604e-06, "loss": 0.3571, "step": 16730 }, { "epoch": 0.84, "learning_rate": 7.2220367278798e-06, "loss": 0.397, "step": 16740 }, { "epoch": 0.84, "learning_rate": 7.220367278797997e-06, "loss": 0.4797, "step": 16750 }, { "epoch": 0.84, "learning_rate": 7.218697829716195e-06, "loss": 0.4247, "step": 16760 }, { "epoch": 0.84, "learning_rate": 7.217028380634391e-06, "loss": 0.3314, "step": 16770 }, { "epoch": 0.84, "learning_rate": 7.215358931552589e-06, "loss": 0.3455, "step": 16780 }, { "epoch": 0.84, "learning_rate": 7.213689482470785e-06, "loss": 0.4087, "step": 16790 }, { "epoch": 0.84, "learning_rate": 7.212020033388982e-06, "loss": 0.4361, "step": 16800 }, { "epoch": 0.84, "learning_rate": 7.21035058430718e-06, "loss": 0.3058, "step": 16810 }, { "epoch": 0.84, "learning_rate": 7.208681135225376e-06, "loss": 0.347, "step": 16820 }, { "epoch": 0.84, "learning_rate": 7.207011686143574e-06, "loss": 0.4159, "step": 16830 }, { "epoch": 0.84, "learning_rate": 7.20534223706177e-06, "loss": 0.4736, "step": 16840 }, { "epoch": 0.84, "learning_rate": 7.203672787979967e-06, "loss": 0.455, "step": 16850 }, { "epoch": 0.84, "learning_rate": 7.202003338898165e-06, "loss": 0.3705, "step": 16860 }, { "epoch": 0.84, "learning_rate": 7.200333889816361e-06, "loss": 0.3461, "step": 16870 }, { "epoch": 0.84, "learning_rate": 7.198664440734559e-06, "loss": 0.4046, "step": 16880 }, { "epoch": 0.84, "learning_rate": 7.196994991652755e-06, "loss": 0.423, "step": 16890 }, { "epoch": 0.84, "learning_rate": 7.195325542570952e-06, "loss": 0.4568, "step": 16900 }, { "epoch": 0.85, "learning_rate": 7.193656093489149e-06, "loss": 0.4808, "step": 16910 }, { "epoch": 0.85, "learning_rate": 7.191986644407346e-06, "loss": 0.35, "step": 16920 }, { "epoch": 0.85, "learning_rate": 7.190317195325544e-06, "loss": 0.4514, "step": 16930 }, { "epoch": 0.85, "learning_rate": 7.18864774624374e-06, "loss": 0.4248, "step": 16940 }, { "epoch": 0.85, "learning_rate": 7.186978297161937e-06, "loss": 0.3894, "step": 16950 }, { "epoch": 0.85, "learning_rate": 7.185308848080134e-06, "loss": 0.3777, "step": 16960 }, { "epoch": 0.85, "learning_rate": 7.183639398998331e-06, "loss": 0.2891, "step": 16970 }, { "epoch": 0.85, "learning_rate": 7.181969949916529e-06, "loss": 0.362, "step": 16980 }, { "epoch": 0.85, "learning_rate": 7.180300500834725e-06, "loss": 0.3859, "step": 16990 }, { "epoch": 0.85, "learning_rate": 7.178631051752922e-06, "loss": 0.3904, "step": 17000 }, { "epoch": 0.85, "learning_rate": 7.176961602671119e-06, "loss": 0.3825, "step": 17010 }, { "epoch": 0.85, "learning_rate": 7.175292153589316e-06, "loss": 0.3299, "step": 17020 }, { "epoch": 0.85, "learning_rate": 7.173622704507514e-06, "loss": 0.4376, "step": 17030 }, { "epoch": 0.85, "learning_rate": 7.17195325542571e-06, "loss": 0.3255, "step": 17040 }, { "epoch": 0.85, "learning_rate": 7.170283806343907e-06, "loss": 0.3787, "step": 17050 }, { "epoch": 0.85, "learning_rate": 7.168614357262104e-06, "loss": 0.4056, "step": 17060 }, { "epoch": 0.85, "learning_rate": 7.166944908180301e-06, "loss": 0.3516, "step": 17070 }, { "epoch": 0.85, "learning_rate": 7.165275459098499e-06, "loss": 0.3724, "step": 17080 }, { "epoch": 0.85, "learning_rate": 7.163606010016695e-06, "loss": 0.3588, "step": 17090 }, { "epoch": 0.85, "learning_rate": 7.161936560934892e-06, "loss": 0.3245, "step": 17100 }, { "epoch": 0.86, "learning_rate": 7.160267111853089e-06, "loss": 0.4323, "step": 17110 }, { "epoch": 0.86, "learning_rate": 7.158597662771286e-06, "loss": 0.3495, "step": 17120 }, { "epoch": 0.86, "learning_rate": 7.156928213689484e-06, "loss": 0.3715, "step": 17130 }, { "epoch": 0.86, "learning_rate": 7.15525876460768e-06, "loss": 0.4014, "step": 17140 }, { "epoch": 0.86, "learning_rate": 7.153589315525877e-06, "loss": 0.4422, "step": 17150 }, { "epoch": 0.86, "learning_rate": 7.151919866444074e-06, "loss": 0.2917, "step": 17160 }, { "epoch": 0.86, "learning_rate": 7.150250417362271e-06, "loss": 0.3452, "step": 17170 }, { "epoch": 0.86, "learning_rate": 7.148580968280469e-06, "loss": 0.4135, "step": 17180 }, { "epoch": 0.86, "learning_rate": 7.146911519198665e-06, "loss": 0.3233, "step": 17190 }, { "epoch": 0.86, "learning_rate": 7.145242070116863e-06, "loss": 0.2995, "step": 17200 }, { "epoch": 0.86, "learning_rate": 7.143572621035059e-06, "loss": 0.3589, "step": 17210 }, { "epoch": 0.86, "learning_rate": 7.141903171953256e-06, "loss": 0.415, "step": 17220 }, { "epoch": 0.86, "learning_rate": 7.140233722871454e-06, "loss": 0.4421, "step": 17230 }, { "epoch": 0.86, "learning_rate": 7.13856427378965e-06, "loss": 0.3688, "step": 17240 }, { "epoch": 0.86, "learning_rate": 7.136894824707848e-06, "loss": 0.4466, "step": 17250 }, { "epoch": 0.86, "learning_rate": 7.135225375626044e-06, "loss": 0.3647, "step": 17260 }, { "epoch": 0.86, "learning_rate": 7.133555926544241e-06, "loss": 0.3752, "step": 17270 }, { "epoch": 0.86, "learning_rate": 7.131886477462437e-06, "loss": 0.343, "step": 17280 }, { "epoch": 0.86, "learning_rate": 7.130217028380635e-06, "loss": 0.3011, "step": 17290 }, { "epoch": 0.86, "learning_rate": 7.128547579298833e-06, "loss": 0.5109, "step": 17300 }, { "epoch": 0.87, "learning_rate": 7.126878130217029e-06, "loss": 0.3735, "step": 17310 }, { "epoch": 0.87, "learning_rate": 7.125208681135226e-06, "loss": 0.4108, "step": 17320 }, { "epoch": 0.87, "learning_rate": 7.123539232053423e-06, "loss": 0.4574, "step": 17330 }, { "epoch": 0.87, "learning_rate": 7.12186978297162e-06, "loss": 0.3983, "step": 17340 }, { "epoch": 0.87, "learning_rate": 7.120200333889818e-06, "loss": 0.4484, "step": 17350 }, { "epoch": 0.87, "learning_rate": 7.118530884808014e-06, "loss": 0.3937, "step": 17360 }, { "epoch": 0.87, "learning_rate": 7.116861435726211e-06, "loss": 0.429, "step": 17370 }, { "epoch": 0.87, "learning_rate": 7.115191986644408e-06, "loss": 0.3179, "step": 17380 }, { "epoch": 0.87, "learning_rate": 7.113522537562605e-06, "loss": 0.495, "step": 17390 }, { "epoch": 0.87, "learning_rate": 7.111853088480803e-06, "loss": 0.4099, "step": 17400 }, { "epoch": 0.87, "learning_rate": 7.110183639398999e-06, "loss": 0.3543, "step": 17410 }, { "epoch": 0.87, "learning_rate": 7.108514190317196e-06, "loss": 0.3405, "step": 17420 }, { "epoch": 0.87, "learning_rate": 7.106844741235393e-06, "loss": 0.3863, "step": 17430 }, { "epoch": 0.87, "learning_rate": 7.10517529215359e-06, "loss": 0.4154, "step": 17440 }, { "epoch": 0.87, "learning_rate": 7.103505843071788e-06, "loss": 0.4159, "step": 17450 }, { "epoch": 0.87, "learning_rate": 7.101836393989984e-06, "loss": 0.3486, "step": 17460 }, { "epoch": 0.87, "learning_rate": 7.100166944908181e-06, "loss": 0.4269, "step": 17470 }, { "epoch": 0.87, "learning_rate": 7.098497495826378e-06, "loss": 0.4284, "step": 17480 }, { "epoch": 0.87, "learning_rate": 7.096828046744575e-06, "loss": 0.4099, "step": 17490 }, { "epoch": 0.88, "learning_rate": 7.095158597662773e-06, "loss": 0.4173, "step": 17500 }, { "epoch": 0.88, "learning_rate": 7.093489148580969e-06, "loss": 0.3636, "step": 17510 }, { "epoch": 0.88, "learning_rate": 7.091819699499166e-06, "loss": 0.4339, "step": 17520 }, { "epoch": 0.88, "learning_rate": 7.090150250417363e-06, "loss": 0.3889, "step": 17530 }, { "epoch": 0.88, "learning_rate": 7.08848080133556e-06, "loss": 0.283, "step": 17540 }, { "epoch": 0.88, "learning_rate": 7.086811352253758e-06, "loss": 0.3529, "step": 17550 }, { "epoch": 0.88, "learning_rate": 7.085141903171954e-06, "loss": 0.4284, "step": 17560 }, { "epoch": 0.88, "learning_rate": 7.083472454090151e-06, "loss": 0.3998, "step": 17570 }, { "epoch": 0.88, "learning_rate": 7.081803005008348e-06, "loss": 0.3683, "step": 17580 }, { "epoch": 0.88, "learning_rate": 7.080133555926545e-06, "loss": 0.4195, "step": 17590 }, { "epoch": 0.88, "learning_rate": 7.078464106844741e-06, "loss": 0.5254, "step": 17600 }, { "epoch": 0.88, "learning_rate": 7.076794657762939e-06, "loss": 0.4564, "step": 17610 }, { "epoch": 0.88, "learning_rate": 7.075125208681136e-06, "loss": 0.404, "step": 17620 }, { "epoch": 0.88, "learning_rate": 7.073455759599333e-06, "loss": 0.4362, "step": 17630 }, { "epoch": 0.88, "learning_rate": 7.07178631051753e-06, "loss": 0.3848, "step": 17640 }, { "epoch": 0.88, "learning_rate": 7.070116861435726e-06, "loss": 0.3663, "step": 17650 }, { "epoch": 0.88, "learning_rate": 7.068447412353924e-06, "loss": 0.3218, "step": 17660 }, { "epoch": 0.88, "learning_rate": 7.066777963272121e-06, "loss": 0.3869, "step": 17670 }, { "epoch": 0.88, "learning_rate": 7.065108514190318e-06, "loss": 0.404, "step": 17680 }, { "epoch": 0.88, "learning_rate": 7.063439065108515e-06, "loss": 0.3207, "step": 17690 }, { "epoch": 0.89, "learning_rate": 7.061769616026711e-06, "loss": 0.3567, "step": 17700 }, { "epoch": 0.89, "learning_rate": 7.060100166944909e-06, "loss": 0.3293, "step": 17710 }, { "epoch": 0.89, "learning_rate": 7.058430717863106e-06, "loss": 0.3943, "step": 17720 }, { "epoch": 0.89, "learning_rate": 7.056761268781303e-06, "loss": 0.3559, "step": 17730 }, { "epoch": 0.89, "learning_rate": 7.0550918196995e-06, "loss": 0.3883, "step": 17740 }, { "epoch": 0.89, "learning_rate": 7.053422370617696e-06, "loss": 0.4156, "step": 17750 }, { "epoch": 0.89, "learning_rate": 7.051752921535894e-06, "loss": 0.5215, "step": 17760 }, { "epoch": 0.89, "learning_rate": 7.050083472454091e-06, "loss": 0.4579, "step": 17770 }, { "epoch": 0.89, "learning_rate": 7.048414023372288e-06, "loss": 0.3981, "step": 17780 }, { "epoch": 0.89, "learning_rate": 7.046744574290485e-06, "loss": 0.3337, "step": 17790 }, { "epoch": 0.89, "learning_rate": 7.045075125208681e-06, "loss": 0.4377, "step": 17800 }, { "epoch": 0.89, "learning_rate": 7.043405676126879e-06, "loss": 0.4489, "step": 17810 }, { "epoch": 0.89, "learning_rate": 7.041736227045076e-06, "loss": 0.3888, "step": 17820 }, { "epoch": 0.89, "learning_rate": 7.040066777963273e-06, "loss": 0.4112, "step": 17830 }, { "epoch": 0.89, "learning_rate": 7.03839732888147e-06, "loss": 0.3592, "step": 17840 }, { "epoch": 0.89, "learning_rate": 7.036727879799666e-06, "loss": 0.3675, "step": 17850 }, { "epoch": 0.89, "learning_rate": 7.035058430717864e-06, "loss": 0.346, "step": 17860 }, { "epoch": 0.89, "learning_rate": 7.033388981636062e-06, "loss": 0.4415, "step": 17870 }, { "epoch": 0.89, "learning_rate": 7.031719532554258e-06, "loss": 0.4063, "step": 17880 }, { "epoch": 0.89, "learning_rate": 7.030050083472455e-06, "loss": 0.4522, "step": 17890 }, { "epoch": 0.9, "learning_rate": 7.028380634390651e-06, "loss": 0.4796, "step": 17900 }, { "epoch": 0.9, "learning_rate": 7.026711185308849e-06, "loss": 0.3893, "step": 17910 }, { "epoch": 0.9, "learning_rate": 7.025041736227045e-06, "loss": 0.3483, "step": 17920 }, { "epoch": 0.9, "learning_rate": 7.023372287145243e-06, "loss": 0.3193, "step": 17930 }, { "epoch": 0.9, "learning_rate": 7.02170283806344e-06, "loss": 0.448, "step": 17940 }, { "epoch": 0.9, "learning_rate": 7.020033388981636e-06, "loss": 0.4038, "step": 17950 }, { "epoch": 0.9, "learning_rate": 7.018363939899834e-06, "loss": 0.4312, "step": 17960 }, { "epoch": 0.9, "learning_rate": 7.01669449081803e-06, "loss": 0.3661, "step": 17970 }, { "epoch": 0.9, "learning_rate": 7.015025041736228e-06, "loss": 0.3375, "step": 17980 }, { "epoch": 0.9, "learning_rate": 7.013355592654425e-06, "loss": 0.4305, "step": 17990 }, { "epoch": 0.9, "learning_rate": 7.011686143572622e-06, "loss": 0.4359, "step": 18000 }, { "epoch": 0.9, "learning_rate": 7.010016694490819e-06, "loss": 0.3995, "step": 18010 }, { "epoch": 0.9, "learning_rate": 7.008347245409015e-06, "loss": 0.4679, "step": 18020 }, { "epoch": 0.9, "learning_rate": 7.006677796327213e-06, "loss": 0.4099, "step": 18030 }, { "epoch": 0.9, "learning_rate": 7.00500834724541e-06, "loss": 0.4412, "step": 18040 }, { "epoch": 0.9, "learning_rate": 7.003338898163607e-06, "loss": 0.2916, "step": 18050 }, { "epoch": 0.9, "learning_rate": 7.001669449081804e-06, "loss": 0.3879, "step": 18060 }, { "epoch": 0.9, "learning_rate": 7e-06, "loss": 0.4082, "step": 18070 }, { "epoch": 0.9, "learning_rate": 6.998330550918198e-06, "loss": 0.4247, "step": 18080 }, { "epoch": 0.9, "learning_rate": 6.996661101836395e-06, "loss": 0.2866, "step": 18090 }, { "epoch": 0.91, "learning_rate": 6.994991652754592e-06, "loss": 0.4466, "step": 18100 }, { "epoch": 0.91, "learning_rate": 6.993322203672789e-06, "loss": 0.4228, "step": 18110 }, { "epoch": 0.91, "learning_rate": 6.991652754590985e-06, "loss": 0.3747, "step": 18120 }, { "epoch": 0.91, "learning_rate": 6.989983305509183e-06, "loss": 0.3126, "step": 18130 }, { "epoch": 0.91, "learning_rate": 6.98831385642738e-06, "loss": 0.3703, "step": 18140 }, { "epoch": 0.91, "learning_rate": 6.986644407345577e-06, "loss": 0.4161, "step": 18150 }, { "epoch": 0.91, "learning_rate": 6.984974958263774e-06, "loss": 0.4539, "step": 18160 }, { "epoch": 0.91, "learning_rate": 6.98330550918197e-06, "loss": 0.3973, "step": 18170 }, { "epoch": 0.91, "learning_rate": 6.981636060100168e-06, "loss": 0.4127, "step": 18180 }, { "epoch": 0.91, "learning_rate": 6.979966611018365e-06, "loss": 0.3914, "step": 18190 }, { "epoch": 0.91, "learning_rate": 6.978297161936562e-06, "loss": 0.3658, "step": 18200 }, { "epoch": 0.91, "learning_rate": 6.976627712854759e-06, "loss": 0.3487, "step": 18210 }, { "epoch": 0.91, "learning_rate": 6.974958263772955e-06, "loss": 0.346, "step": 18220 }, { "epoch": 0.91, "learning_rate": 6.973288814691153e-06, "loss": 0.3487, "step": 18230 }, { "epoch": 0.91, "learning_rate": 6.971619365609349e-06, "loss": 0.307, "step": 18240 }, { "epoch": 0.91, "learning_rate": 6.969949916527547e-06, "loss": 0.3564, "step": 18250 }, { "epoch": 0.91, "learning_rate": 6.968280467445744e-06, "loss": 0.3959, "step": 18260 }, { "epoch": 0.91, "learning_rate": 6.96661101836394e-06, "loss": 0.4187, "step": 18270 }, { "epoch": 0.91, "learning_rate": 6.964941569282138e-06, "loss": 0.4397, "step": 18280 }, { "epoch": 0.91, "learning_rate": 6.963272120200334e-06, "loss": 0.442, "step": 18290 }, { "epoch": 0.92, "learning_rate": 6.961602671118532e-06, "loss": 0.3824, "step": 18300 }, { "epoch": 0.92, "learning_rate": 6.959933222036729e-06, "loss": 0.3379, "step": 18310 }, { "epoch": 0.92, "learning_rate": 6.958263772954925e-06, "loss": 0.4586, "step": 18320 }, { "epoch": 0.92, "learning_rate": 6.956594323873123e-06, "loss": 0.4055, "step": 18330 }, { "epoch": 0.92, "learning_rate": 6.954924874791319e-06, "loss": 0.4163, "step": 18340 }, { "epoch": 0.92, "learning_rate": 6.953255425709517e-06, "loss": 0.3946, "step": 18350 }, { "epoch": 0.92, "learning_rate": 6.951585976627714e-06, "loss": 0.4184, "step": 18360 }, { "epoch": 0.92, "learning_rate": 6.94991652754591e-06, "loss": 0.3937, "step": 18370 }, { "epoch": 0.92, "learning_rate": 6.948247078464108e-06, "loss": 0.4113, "step": 18380 }, { "epoch": 0.92, "learning_rate": 6.946577629382304e-06, "loss": 0.4106, "step": 18390 }, { "epoch": 0.92, "learning_rate": 6.944908180300502e-06, "loss": 0.3659, "step": 18400 }, { "epoch": 0.92, "learning_rate": 6.943238731218699e-06, "loss": 0.3245, "step": 18410 }, { "epoch": 0.92, "learning_rate": 6.941569282136895e-06, "loss": 0.3686, "step": 18420 }, { "epoch": 0.92, "learning_rate": 6.939899833055093e-06, "loss": 0.4723, "step": 18430 }, { "epoch": 0.92, "learning_rate": 6.938230383973289e-06, "loss": 0.3815, "step": 18440 }, { "epoch": 0.92, "learning_rate": 6.936560934891487e-06, "loss": 0.3392, "step": 18450 }, { "epoch": 0.92, "learning_rate": 6.934891485809684e-06, "loss": 0.365, "step": 18460 }, { "epoch": 0.92, "learning_rate": 6.93322203672788e-06, "loss": 0.4544, "step": 18470 }, { "epoch": 0.92, "learning_rate": 6.931552587646078e-06, "loss": 0.4375, "step": 18480 }, { "epoch": 0.92, "learning_rate": 6.929883138564274e-06, "loss": 0.419, "step": 18490 }, { "epoch": 0.93, "learning_rate": 6.928213689482472e-06, "loss": 0.374, "step": 18500 }, { "epoch": 0.93, "learning_rate": 6.926544240400669e-06, "loss": 0.3127, "step": 18510 }, { "epoch": 0.93, "learning_rate": 6.924874791318865e-06, "loss": 0.3635, "step": 18520 }, { "epoch": 0.93, "learning_rate": 6.923205342237063e-06, "loss": 0.4312, "step": 18530 }, { "epoch": 0.93, "learning_rate": 6.921535893155259e-06, "loss": 0.3833, "step": 18540 }, { "epoch": 0.93, "learning_rate": 6.919866444073457e-06, "loss": 0.4404, "step": 18550 }, { "epoch": 0.93, "learning_rate": 6.918196994991654e-06, "loss": 0.3846, "step": 18560 }, { "epoch": 0.93, "learning_rate": 6.91652754590985e-06, "loss": 0.4017, "step": 18570 }, { "epoch": 0.93, "learning_rate": 6.914858096828048e-06, "loss": 0.3392, "step": 18580 }, { "epoch": 0.93, "learning_rate": 6.913188647746244e-06, "loss": 0.3631, "step": 18590 }, { "epoch": 0.93, "learning_rate": 6.911519198664442e-06, "loss": 0.3805, "step": 18600 }, { "epoch": 0.93, "learning_rate": 6.909849749582638e-06, "loss": 0.4181, "step": 18610 }, { "epoch": 0.93, "learning_rate": 6.908180300500835e-06, "loss": 0.374, "step": 18620 }, { "epoch": 0.93, "learning_rate": 6.906510851419033e-06, "loss": 0.3586, "step": 18630 }, { "epoch": 0.93, "learning_rate": 6.904841402337229e-06, "loss": 0.3267, "step": 18640 }, { "epoch": 0.93, "learning_rate": 6.903171953255427e-06, "loss": 0.4405, "step": 18650 }, { "epoch": 0.93, "learning_rate": 6.901502504173623e-06, "loss": 0.4061, "step": 18660 }, { "epoch": 0.93, "learning_rate": 6.899833055091821e-06, "loss": 0.3531, "step": 18670 }, { "epoch": 0.93, "learning_rate": 6.898163606010018e-06, "loss": 0.4017, "step": 18680 }, { "epoch": 0.93, "learning_rate": 6.896494156928214e-06, "loss": 0.356, "step": 18690 }, { "epoch": 0.94, "learning_rate": 6.894824707846412e-06, "loss": 0.3094, "step": 18700 }, { "epoch": 0.94, "learning_rate": 6.893155258764608e-06, "loss": 0.4431, "step": 18710 }, { "epoch": 0.94, "learning_rate": 6.891485809682806e-06, "loss": 0.4047, "step": 18720 }, { "epoch": 0.94, "learning_rate": 6.889816360601003e-06, "loss": 0.426, "step": 18730 }, { "epoch": 0.94, "learning_rate": 6.888146911519199e-06, "loss": 0.4227, "step": 18740 }, { "epoch": 0.94, "learning_rate": 6.886477462437397e-06, "loss": 0.4418, "step": 18750 }, { "epoch": 0.94, "learning_rate": 6.884808013355593e-06, "loss": 0.3833, "step": 18760 }, { "epoch": 0.94, "learning_rate": 6.883138564273791e-06, "loss": 0.4023, "step": 18770 }, { "epoch": 0.94, "learning_rate": 6.881469115191988e-06, "loss": 0.3641, "step": 18780 }, { "epoch": 0.94, "learning_rate": 6.879799666110184e-06, "loss": 0.3325, "step": 18790 }, { "epoch": 0.94, "learning_rate": 6.878130217028382e-06, "loss": 0.4297, "step": 18800 }, { "epoch": 0.94, "learning_rate": 6.876460767946578e-06, "loss": 0.4342, "step": 18810 }, { "epoch": 0.94, "learning_rate": 6.874791318864776e-06, "loss": 0.3898, "step": 18820 }, { "epoch": 0.94, "learning_rate": 6.873121869782973e-06, "loss": 0.443, "step": 18830 }, { "epoch": 0.94, "learning_rate": 6.871452420701169e-06, "loss": 0.3777, "step": 18840 }, { "epoch": 0.94, "learning_rate": 6.869782971619367e-06, "loss": 0.5374, "step": 18850 }, { "epoch": 0.94, "learning_rate": 6.868113522537563e-06, "loss": 0.4668, "step": 18860 }, { "epoch": 0.94, "learning_rate": 6.866444073455761e-06, "loss": 0.3532, "step": 18870 }, { "epoch": 0.94, "learning_rate": 6.864774624373958e-06, "loss": 0.3913, "step": 18880 }, { "epoch": 0.94, "learning_rate": 6.863105175292154e-06, "loss": 0.3992, "step": 18890 }, { "epoch": 0.94, "learning_rate": 6.861435726210352e-06, "loss": 0.4045, "step": 18900 }, { "epoch": 0.95, "learning_rate": 6.859766277128548e-06, "loss": 0.466, "step": 18910 }, { "epoch": 0.95, "learning_rate": 6.858096828046746e-06, "loss": 0.3221, "step": 18920 }, { "epoch": 0.95, "learning_rate": 6.856427378964942e-06, "loss": 0.3851, "step": 18930 }, { "epoch": 0.95, "learning_rate": 6.854757929883139e-06, "loss": 0.4202, "step": 18940 }, { "epoch": 0.95, "learning_rate": 6.853088480801337e-06, "loss": 0.3565, "step": 18950 }, { "epoch": 0.95, "learning_rate": 6.851419031719533e-06, "loss": 0.3707, "step": 18960 }, { "epoch": 0.95, "learning_rate": 6.849749582637731e-06, "loss": 0.4368, "step": 18970 }, { "epoch": 0.95, "learning_rate": 6.848080133555927e-06, "loss": 0.3524, "step": 18980 }, { "epoch": 0.95, "learning_rate": 6.846410684474124e-06, "loss": 0.3707, "step": 18990 }, { "epoch": 0.95, "learning_rate": 6.844741235392322e-06, "loss": 0.3571, "step": 19000 }, { "epoch": 0.95, "learning_rate": 6.843071786310518e-06, "loss": 0.3431, "step": 19010 }, { "epoch": 0.95, "learning_rate": 6.8414023372287156e-06, "loss": 0.3998, "step": 19020 }, { "epoch": 0.95, "learning_rate": 6.839732888146912e-06, "loss": 0.3418, "step": 19030 }, { "epoch": 0.95, "learning_rate": 6.838063439065109e-06, "loss": 0.4265, "step": 19040 }, { "epoch": 0.95, "learning_rate": 6.8363939899833066e-06, "loss": 0.4867, "step": 19050 }, { "epoch": 0.95, "learning_rate": 6.834724540901503e-06, "loss": 0.4319, "step": 19060 }, { "epoch": 0.95, "learning_rate": 6.8330550918197006e-06, "loss": 0.44, "step": 19070 }, { "epoch": 0.95, "learning_rate": 6.831385642737897e-06, "loss": 0.4743, "step": 19080 }, { "epoch": 0.95, "learning_rate": 6.829716193656094e-06, "loss": 0.3864, "step": 19090 }, { "epoch": 0.95, "learning_rate": 6.8280467445742916e-06, "loss": 0.3436, "step": 19100 }, { "epoch": 0.96, "learning_rate": 6.826377295492488e-06, "loss": 0.4089, "step": 19110 }, { "epoch": 0.96, "learning_rate": 6.8247078464106856e-06, "loss": 0.3688, "step": 19120 }, { "epoch": 0.96, "learning_rate": 6.823038397328882e-06, "loss": 0.3799, "step": 19130 }, { "epoch": 0.96, "learning_rate": 6.821368948247079e-06, "loss": 0.4003, "step": 19140 }, { "epoch": 0.96, "learning_rate": 6.8196994991652766e-06, "loss": 0.3513, "step": 19150 }, { "epoch": 0.96, "learning_rate": 6.818030050083473e-06, "loss": 0.3465, "step": 19160 }, { "epoch": 0.96, "learning_rate": 6.8163606010016705e-06, "loss": 0.4503, "step": 19170 }, { "epoch": 0.96, "learning_rate": 6.814691151919867e-06, "loss": 0.4441, "step": 19180 }, { "epoch": 0.96, "learning_rate": 6.813021702838064e-06, "loss": 0.4201, "step": 19190 }, { "epoch": 0.96, "learning_rate": 6.8113522537562615e-06, "loss": 0.4837, "step": 19200 }, { "epoch": 0.96, "learning_rate": 6.809682804674458e-06, "loss": 0.3903, "step": 19210 }, { "epoch": 0.96, "learning_rate": 6.8080133555926555e-06, "loss": 0.4297, "step": 19220 }, { "epoch": 0.96, "learning_rate": 6.806343906510852e-06, "loss": 0.3788, "step": 19230 }, { "epoch": 0.96, "learning_rate": 6.804674457429049e-06, "loss": 0.3059, "step": 19240 }, { "epoch": 0.96, "learning_rate": 6.803005008347246e-06, "loss": 0.4347, "step": 19250 }, { "epoch": 0.96, "learning_rate": 6.801335559265443e-06, "loss": 0.3083, "step": 19260 }, { "epoch": 0.96, "learning_rate": 6.7996661101836405e-06, "loss": 0.4137, "step": 19270 }, { "epoch": 0.96, "learning_rate": 6.797996661101837e-06, "loss": 0.408, "step": 19280 }, { "epoch": 0.96, "learning_rate": 6.796327212020034e-06, "loss": 0.423, "step": 19290 }, { "epoch": 0.96, "learning_rate": 6.794657762938231e-06, "loss": 0.3964, "step": 19300 }, { "epoch": 0.97, "learning_rate": 6.792988313856428e-06, "loss": 0.3851, "step": 19310 }, { "epoch": 0.97, "learning_rate": 6.7913188647746255e-06, "loss": 0.4496, "step": 19320 }, { "epoch": 0.97, "learning_rate": 6.789649415692822e-06, "loss": 0.337, "step": 19330 }, { "epoch": 0.97, "learning_rate": 6.7879799666110195e-06, "loss": 0.3808, "step": 19340 }, { "epoch": 0.97, "learning_rate": 6.786310517529216e-06, "loss": 0.4024, "step": 19350 }, { "epoch": 0.97, "learning_rate": 6.784641068447413e-06, "loss": 0.3411, "step": 19360 }, { "epoch": 0.97, "learning_rate": 6.7829716193656105e-06, "loss": 0.4612, "step": 19370 }, { "epoch": 0.97, "learning_rate": 6.781302170283807e-06, "loss": 0.3569, "step": 19380 }, { "epoch": 0.97, "learning_rate": 6.7796327212020045e-06, "loss": 0.3671, "step": 19390 }, { "epoch": 0.97, "learning_rate": 6.777963272120201e-06, "loss": 0.385, "step": 19400 }, { "epoch": 0.97, "learning_rate": 6.776293823038398e-06, "loss": 0.4024, "step": 19410 }, { "epoch": 0.97, "learning_rate": 6.7746243739565955e-06, "loss": 0.4722, "step": 19420 }, { "epoch": 0.97, "learning_rate": 6.772954924874792e-06, "loss": 0.4436, "step": 19430 }, { "epoch": 0.97, "learning_rate": 6.7712854757929895e-06, "loss": 0.3656, "step": 19440 }, { "epoch": 0.97, "learning_rate": 6.769616026711186e-06, "loss": 0.303, "step": 19450 }, { "epoch": 0.97, "learning_rate": 6.767946577629383e-06, "loss": 0.3393, "step": 19460 }, { "epoch": 0.97, "learning_rate": 6.7662771285475805e-06, "loss": 0.3297, "step": 19470 }, { "epoch": 0.97, "learning_rate": 6.764607679465777e-06, "loss": 0.4409, "step": 19480 }, { "epoch": 0.97, "learning_rate": 6.7629382303839745e-06, "loss": 0.3829, "step": 19490 }, { "epoch": 0.97, "learning_rate": 6.761268781302171e-06, "loss": 0.3491, "step": 19500 }, { "epoch": 0.98, "learning_rate": 6.759599332220368e-06, "loss": 0.3547, "step": 19510 }, { "epoch": 0.98, "learning_rate": 6.7579298831385655e-06, "loss": 0.3697, "step": 19520 }, { "epoch": 0.98, "learning_rate": 6.756260434056762e-06, "loss": 0.4228, "step": 19530 }, { "epoch": 0.98, "learning_rate": 6.7545909849749595e-06, "loss": 0.4058, "step": 19540 }, { "epoch": 0.98, "learning_rate": 6.752921535893156e-06, "loss": 0.3327, "step": 19550 }, { "epoch": 0.98, "learning_rate": 6.751252086811353e-06, "loss": 0.3114, "step": 19560 }, { "epoch": 0.98, "learning_rate": 6.74958263772955e-06, "loss": 0.3948, "step": 19570 }, { "epoch": 0.98, "learning_rate": 6.747913188647747e-06, "loss": 0.4635, "step": 19580 }, { "epoch": 0.98, "learning_rate": 6.7462437395659445e-06, "loss": 0.423, "step": 19590 }, { "epoch": 0.98, "learning_rate": 6.744574290484141e-06, "loss": 0.3904, "step": 19600 }, { "epoch": 0.98, "learning_rate": 6.742904841402338e-06, "loss": 0.4399, "step": 19610 }, { "epoch": 0.98, "learning_rate": 6.741235392320535e-06, "loss": 0.346, "step": 19620 }, { "epoch": 0.98, "learning_rate": 6.739565943238732e-06, "loss": 0.3841, "step": 19630 }, { "epoch": 0.98, "learning_rate": 6.7378964941569295e-06, "loss": 0.4012, "step": 19640 }, { "epoch": 0.98, "learning_rate": 6.736227045075126e-06, "loss": 0.3417, "step": 19650 }, { "epoch": 0.98, "learning_rate": 6.734557595993323e-06, "loss": 0.3722, "step": 19660 }, { "epoch": 0.98, "learning_rate": 6.73288814691152e-06, "loss": 0.4013, "step": 19670 }, { "epoch": 0.98, "learning_rate": 6.731218697829717e-06, "loss": 0.3228, "step": 19680 }, { "epoch": 0.98, "learning_rate": 6.7295492487479145e-06, "loss": 0.3782, "step": 19690 }, { "epoch": 0.98, "learning_rate": 6.727879799666111e-06, "loss": 0.4288, "step": 19700 }, { "epoch": 0.99, "learning_rate": 6.726210350584308e-06, "loss": 0.3755, "step": 19710 }, { "epoch": 0.99, "learning_rate": 6.724540901502505e-06, "loss": 0.4226, "step": 19720 }, { "epoch": 0.99, "learning_rate": 6.722871452420702e-06, "loss": 0.3559, "step": 19730 }, { "epoch": 0.99, "learning_rate": 6.7212020033388994e-06, "loss": 0.322, "step": 19740 }, { "epoch": 0.99, "learning_rate": 6.719532554257096e-06, "loss": 0.3291, "step": 19750 }, { "epoch": 0.99, "learning_rate": 6.717863105175293e-06, "loss": 0.3907, "step": 19760 }, { "epoch": 0.99, "learning_rate": 6.71619365609349e-06, "loss": 0.3853, "step": 19770 }, { "epoch": 0.99, "learning_rate": 6.714524207011687e-06, "loss": 0.43, "step": 19780 }, { "epoch": 0.99, "learning_rate": 6.7128547579298844e-06, "loss": 0.4879, "step": 19790 }, { "epoch": 0.99, "learning_rate": 6.711185308848081e-06, "loss": 0.3597, "step": 19800 }, { "epoch": 0.99, "learning_rate": 6.709515859766278e-06, "loss": 0.4202, "step": 19810 }, { "epoch": 0.99, "learning_rate": 6.707846410684475e-06, "loss": 0.4169, "step": 19820 }, { "epoch": 0.99, "learning_rate": 6.706176961602672e-06, "loss": 0.3303, "step": 19830 }, { "epoch": 0.99, "learning_rate": 6.7045075125208694e-06, "loss": 0.3365, "step": 19840 }, { "epoch": 0.99, "learning_rate": 6.702838063439066e-06, "loss": 0.3992, "step": 19850 }, { "epoch": 0.99, "learning_rate": 6.701168614357263e-06, "loss": 0.4226, "step": 19860 }, { "epoch": 0.99, "learning_rate": 6.69949916527546e-06, "loss": 0.3598, "step": 19870 }, { "epoch": 0.99, "learning_rate": 6.697829716193657e-06, "loss": 0.3774, "step": 19880 }, { "epoch": 0.99, "learning_rate": 6.696160267111854e-06, "loss": 0.3915, "step": 19890 }, { "epoch": 0.99, "learning_rate": 6.6944908180300506e-06, "loss": 0.3328, "step": 19900 }, { "epoch": 1.0, "learning_rate": 6.6928213689482476e-06, "loss": 0.4635, "step": 19910 }, { "epoch": 1.0, "learning_rate": 6.6911519198664446e-06, "loss": 0.3575, "step": 19920 }, { "epoch": 1.0, "learning_rate": 6.6894824707846416e-06, "loss": 0.5112, "step": 19930 }, { "epoch": 1.0, "learning_rate": 6.687813021702838e-06, "loss": 0.3931, "step": 19940 }, { "epoch": 1.0, "learning_rate": 6.6861435726210356e-06, "loss": 0.432, "step": 19950 }, { "epoch": 1.0, "learning_rate": 6.6844741235392326e-06, "loss": 0.3977, "step": 19960 }, { "epoch": 1.0, "learning_rate": 6.6828046744574296e-06, "loss": 0.4669, "step": 19970 }, { "epoch": 1.0, "learning_rate": 6.6811352253756266e-06, "loss": 0.3382, "step": 19980 }, { "epoch": 1.0, "learning_rate": 6.679465776293823e-06, "loss": 0.4909, "step": 19990 }, { "epoch": 1.0, "learning_rate": 6.6777963272120206e-06, "loss": 0.3874, "step": 20000 }, { "epoch": 1.0, "eval_loss": 0.377562016248703, "eval_runtime": 182.1287, "eval_samples_per_second": 20.672, "eval_steps_per_second": 2.586, "step": 20000 }, { "epoch": 1.0, "learning_rate": 6.676126878130218e-06, "loss": 0.4484, "step": 20010 }, { "epoch": 1.0, "learning_rate": 6.6744574290484146e-06, "loss": 0.3775, "step": 20020 }, { "epoch": 1.0, "learning_rate": 6.6727879799666116e-06, "loss": 0.3934, "step": 20030 }, { "epoch": 1.0, "learning_rate": 6.671118530884808e-06, "loss": 0.3812, "step": 20040 }, { "epoch": 1.0, "learning_rate": 6.6694490818030055e-06, "loss": 0.3175, "step": 20050 }, { "epoch": 1.0, "learning_rate": 6.667779632721203e-06, "loss": 0.3497, "step": 20060 }, { "epoch": 1.0, "learning_rate": 6.6661101836393995e-06, "loss": 0.3642, "step": 20070 }, { "epoch": 1.0, "learning_rate": 6.6644407345575965e-06, "loss": 0.4038, "step": 20080 }, { "epoch": 1.0, "learning_rate": 6.662771285475793e-06, "loss": 0.3439, "step": 20090 }, { "epoch": 1.0, "learning_rate": 6.6611018363939905e-06, "loss": 0.3539, "step": 20100 }, { "epoch": 1.01, "learning_rate": 6.659432387312188e-06, "loss": 0.3473, "step": 20110 }, { "epoch": 1.01, "learning_rate": 6.6577629382303845e-06, "loss": 0.3769, "step": 20120 }, { "epoch": 1.01, "learning_rate": 6.6560934891485815e-06, "loss": 0.3872, "step": 20130 }, { "epoch": 1.01, "learning_rate": 6.6544240400667785e-06, "loss": 0.2917, "step": 20140 }, { "epoch": 1.01, "learning_rate": 6.6527545909849755e-06, "loss": 0.4281, "step": 20150 }, { "epoch": 1.01, "learning_rate": 6.651085141903173e-06, "loss": 0.3603, "step": 20160 }, { "epoch": 1.01, "learning_rate": 6.6494156928213695e-06, "loss": 0.3884, "step": 20170 }, { "epoch": 1.01, "learning_rate": 6.6477462437395665e-06, "loss": 0.4022, "step": 20180 }, { "epoch": 1.01, "learning_rate": 6.6460767946577635e-06, "loss": 0.3955, "step": 20190 }, { "epoch": 1.01, "learning_rate": 6.6444073455759605e-06, "loss": 0.2965, "step": 20200 }, { "epoch": 1.01, "learning_rate": 6.642737896494158e-06, "loss": 0.3483, "step": 20210 }, { "epoch": 1.01, "learning_rate": 6.6410684474123545e-06, "loss": 0.4586, "step": 20220 }, { "epoch": 1.01, "learning_rate": 6.6393989983305515e-06, "loss": 0.3892, "step": 20230 }, { "epoch": 1.01, "learning_rate": 6.6377295492487485e-06, "loss": 0.3364, "step": 20240 }, { "epoch": 1.01, "learning_rate": 6.6360601001669455e-06, "loss": 0.3643, "step": 20250 }, { "epoch": 1.01, "learning_rate": 6.634390651085142e-06, "loss": 0.2987, "step": 20260 }, { "epoch": 1.01, "learning_rate": 6.6327212020033395e-06, "loss": 0.3529, "step": 20270 }, { "epoch": 1.01, "learning_rate": 6.6310517529215365e-06, "loss": 0.304, "step": 20280 }, { "epoch": 1.01, "learning_rate": 6.6293823038397335e-06, "loss": 0.332, "step": 20290 }, { "epoch": 1.01, "learning_rate": 6.6277128547579305e-06, "loss": 0.3875, "step": 20300 }, { "epoch": 1.02, "learning_rate": 6.626043405676127e-06, "loss": 0.4864, "step": 20310 }, { "epoch": 1.02, "learning_rate": 6.6243739565943245e-06, "loss": 0.373, "step": 20320 }, { "epoch": 1.02, "learning_rate": 6.6227045075125215e-06, "loss": 0.4272, "step": 20330 }, { "epoch": 1.02, "learning_rate": 6.6210350584307185e-06, "loss": 0.4005, "step": 20340 }, { "epoch": 1.02, "learning_rate": 6.6193656093489155e-06, "loss": 0.3275, "step": 20350 }, { "epoch": 1.02, "learning_rate": 6.617696160267112e-06, "loss": 0.3386, "step": 20360 }, { "epoch": 1.02, "learning_rate": 6.6160267111853095e-06, "loss": 0.367, "step": 20370 }, { "epoch": 1.02, "learning_rate": 6.6143572621035065e-06, "loss": 0.3376, "step": 20380 }, { "epoch": 1.02, "learning_rate": 6.6126878130217035e-06, "loss": 0.399, "step": 20390 }, { "epoch": 1.02, "learning_rate": 6.6110183639399005e-06, "loss": 0.4735, "step": 20400 }, { "epoch": 1.02, "learning_rate": 6.609348914858097e-06, "loss": 0.3385, "step": 20410 }, { "epoch": 1.02, "learning_rate": 6.6076794657762945e-06, "loss": 0.428, "step": 20420 }, { "epoch": 1.02, "learning_rate": 6.6060100166944915e-06, "loss": 0.3851, "step": 20430 }, { "epoch": 1.02, "learning_rate": 6.6043405676126885e-06, "loss": 0.3923, "step": 20440 }, { "epoch": 1.02, "learning_rate": 6.6026711185308855e-06, "loss": 0.3227, "step": 20450 }, { "epoch": 1.02, "learning_rate": 6.601001669449082e-06, "loss": 0.356, "step": 20460 }, { "epoch": 1.02, "learning_rate": 6.5993322203672795e-06, "loss": 0.4605, "step": 20470 }, { "epoch": 1.02, "learning_rate": 6.5976627712854765e-06, "loss": 0.3967, "step": 20480 }, { "epoch": 1.02, "learning_rate": 6.5959933222036735e-06, "loss": 0.4126, "step": 20490 }, { "epoch": 1.02, "learning_rate": 6.5943238731218705e-06, "loss": 0.4536, "step": 20500 }, { "epoch": 1.03, "learning_rate": 6.592654424040067e-06, "loss": 0.3539, "step": 20510 }, { "epoch": 1.03, "learning_rate": 6.5909849749582645e-06, "loss": 0.377, "step": 20520 }, { "epoch": 1.03, "learning_rate": 6.5893155258764615e-06, "loss": 0.3546, "step": 20530 }, { "epoch": 1.03, "learning_rate": 6.5876460767946585e-06, "loss": 0.317, "step": 20540 }, { "epoch": 1.03, "learning_rate": 6.5859766277128555e-06, "loss": 0.2862, "step": 20550 }, { "epoch": 1.03, "learning_rate": 6.584307178631052e-06, "loss": 0.3827, "step": 20560 }, { "epoch": 1.03, "learning_rate": 6.5826377295492495e-06, "loss": 0.3221, "step": 20570 }, { "epoch": 1.03, "learning_rate": 6.580968280467446e-06, "loss": 0.3878, "step": 20580 }, { "epoch": 1.03, "learning_rate": 6.5792988313856435e-06, "loss": 0.3808, "step": 20590 }, { "epoch": 1.03, "learning_rate": 6.5776293823038405e-06, "loss": 0.46, "step": 20600 }, { "epoch": 1.03, "learning_rate": 6.575959933222037e-06, "loss": 0.4291, "step": 20610 }, { "epoch": 1.03, "learning_rate": 6.5742904841402344e-06, "loss": 0.3501, "step": 20620 }, { "epoch": 1.03, "learning_rate": 6.572621035058431e-06, "loss": 0.3375, "step": 20630 }, { "epoch": 1.03, "learning_rate": 6.5709515859766284e-06, "loss": 0.3443, "step": 20640 }, { "epoch": 1.03, "learning_rate": 6.5692821368948254e-06, "loss": 0.3745, "step": 20650 }, { "epoch": 1.03, "learning_rate": 6.567612687813022e-06, "loss": 0.371, "step": 20660 }, { "epoch": 1.03, "learning_rate": 6.5659432387312194e-06, "loss": 0.3907, "step": 20670 }, { "epoch": 1.03, "learning_rate": 6.564273789649416e-06, "loss": 0.3411, "step": 20680 }, { "epoch": 1.03, "learning_rate": 6.5626043405676134e-06, "loss": 0.405, "step": 20690 }, { "epoch": 1.03, "learning_rate": 6.5609348914858104e-06, "loss": 0.3842, "step": 20700 }, { "epoch": 1.04, "learning_rate": 6.559265442404007e-06, "loss": 0.4004, "step": 20710 }, { "epoch": 1.04, "learning_rate": 6.5575959933222044e-06, "loss": 0.305, "step": 20720 }, { "epoch": 1.04, "learning_rate": 6.555926544240401e-06, "loss": 0.337, "step": 20730 }, { "epoch": 1.04, "learning_rate": 6.5542570951585984e-06, "loss": 0.2642, "step": 20740 }, { "epoch": 1.04, "learning_rate": 6.5525876460767954e-06, "loss": 0.3696, "step": 20750 }, { "epoch": 1.04, "learning_rate": 6.550918196994992e-06, "loss": 0.3268, "step": 20760 }, { "epoch": 1.04, "learning_rate": 6.549248747913189e-06, "loss": 0.3454, "step": 20770 }, { "epoch": 1.04, "learning_rate": 6.5475792988313856e-06, "loss": 0.3521, "step": 20780 }, { "epoch": 1.04, "learning_rate": 6.545909849749583e-06, "loss": 0.3286, "step": 20790 }, { "epoch": 1.04, "learning_rate": 6.54424040066778e-06, "loss": 0.3721, "step": 20800 }, { "epoch": 1.04, "learning_rate": 6.542570951585977e-06, "loss": 0.3505, "step": 20810 }, { "epoch": 1.04, "learning_rate": 6.540901502504174e-06, "loss": 0.3006, "step": 20820 }, { "epoch": 1.04, "learning_rate": 6.5392320534223706e-06, "loss": 0.3485, "step": 20830 }, { "epoch": 1.04, "learning_rate": 6.537562604340568e-06, "loss": 0.36, "step": 20840 }, { "epoch": 1.04, "learning_rate": 6.535893155258765e-06, "loss": 0.3341, "step": 20850 }, { "epoch": 1.04, "learning_rate": 6.534223706176962e-06, "loss": 0.3662, "step": 20860 }, { "epoch": 1.04, "learning_rate": 6.532554257095159e-06, "loss": 0.2821, "step": 20870 }, { "epoch": 1.04, "learning_rate": 6.5308848080133556e-06, "loss": 0.4335, "step": 20880 }, { "epoch": 1.04, "learning_rate": 6.529215358931553e-06, "loss": 0.3094, "step": 20890 }, { "epoch": 1.04, "learning_rate": 6.5275459098497496e-06, "loss": 0.3688, "step": 20900 }, { "epoch": 1.05, "learning_rate": 6.525876460767947e-06, "loss": 0.3435, "step": 20910 }, { "epoch": 1.05, "learning_rate": 6.524207011686144e-06, "loss": 0.329, "step": 20920 }, { "epoch": 1.05, "learning_rate": 6.5225375626043405e-06, "loss": 0.4203, "step": 20930 }, { "epoch": 1.05, "learning_rate": 6.520868113522538e-06, "loss": 0.3809, "step": 20940 }, { "epoch": 1.05, "learning_rate": 6.5191986644407345e-06, "loss": 0.2801, "step": 20950 }, { "epoch": 1.05, "learning_rate": 6.517529215358932e-06, "loss": 0.3398, "step": 20960 }, { "epoch": 1.05, "learning_rate": 6.515859766277129e-06, "loss": 0.4036, "step": 20970 }, { "epoch": 1.05, "learning_rate": 6.5141903171953255e-06, "loss": 0.4802, "step": 20980 }, { "epoch": 1.05, "learning_rate": 6.512520868113523e-06, "loss": 0.702, "step": 20990 }, { "epoch": 1.05, "learning_rate": 6.5108514190317195e-06, "loss": 0.4161, "step": 21000 }, { "epoch": 1.05, "learning_rate": 6.509181969949917e-06, "loss": 0.2946, "step": 21010 }, { "epoch": 1.05, "learning_rate": 6.507512520868114e-06, "loss": 0.4494, "step": 21020 }, { "epoch": 1.05, "learning_rate": 6.5058430717863105e-06, "loss": 0.4092, "step": 21030 }, { "epoch": 1.05, "learning_rate": 6.504173622704508e-06, "loss": 0.3839, "step": 21040 }, { "epoch": 1.05, "learning_rate": 6.5025041736227045e-06, "loss": 0.4582, "step": 21050 }, { "epoch": 1.05, "learning_rate": 6.500834724540902e-06, "loss": 0.3823, "step": 21060 }, { "epoch": 1.05, "learning_rate": 6.499165275459099e-06, "loss": 0.3169, "step": 21070 }, { "epoch": 1.05, "learning_rate": 6.4974958263772955e-06, "loss": 0.3904, "step": 21080 }, { "epoch": 1.05, "learning_rate": 6.495826377295493e-06, "loss": 0.349, "step": 21090 }, { "epoch": 1.05, "learning_rate": 6.4941569282136895e-06, "loss": 0.2828, "step": 21100 }, { "epoch": 1.06, "learning_rate": 6.492487479131887e-06, "loss": 0.3507, "step": 21110 }, { "epoch": 1.06, "learning_rate": 6.490818030050084e-06, "loss": 0.3875, "step": 21120 }, { "epoch": 1.06, "learning_rate": 6.4891485809682805e-06, "loss": 0.3326, "step": 21130 }, { "epoch": 1.06, "learning_rate": 6.487479131886478e-06, "loss": 0.3597, "step": 21140 }, { "epoch": 1.06, "learning_rate": 6.4858096828046745e-06, "loss": 0.3419, "step": 21150 }, { "epoch": 1.06, "learning_rate": 6.484140233722872e-06, "loss": 0.3254, "step": 21160 }, { "epoch": 1.06, "learning_rate": 6.482470784641069e-06, "loss": 0.4237, "step": 21170 }, { "epoch": 1.06, "learning_rate": 6.4808013355592655e-06, "loss": 0.3902, "step": 21180 }, { "epoch": 1.06, "learning_rate": 6.479131886477463e-06, "loss": 0.3838, "step": 21190 }, { "epoch": 1.06, "learning_rate": 6.4774624373956595e-06, "loss": 0.3472, "step": 21200 }, { "epoch": 1.06, "learning_rate": 6.475792988313857e-06, "loss": 0.4966, "step": 21210 }, { "epoch": 1.06, "learning_rate": 6.474123539232054e-06, "loss": 0.3583, "step": 21220 }, { "epoch": 1.06, "learning_rate": 6.4724540901502505e-06, "loss": 0.3674, "step": 21230 }, { "epoch": 1.06, "learning_rate": 6.470784641068448e-06, "loss": 0.368, "step": 21240 }, { "epoch": 1.06, "learning_rate": 6.4691151919866445e-06, "loss": 0.4297, "step": 21250 }, { "epoch": 1.06, "learning_rate": 6.467445742904842e-06, "loss": 0.416, "step": 21260 }, { "epoch": 1.06, "learning_rate": 6.4657762938230385e-06, "loss": 0.3939, "step": 21270 }, { "epoch": 1.06, "learning_rate": 6.4641068447412355e-06, "loss": 0.3385, "step": 21280 }, { "epoch": 1.06, "learning_rate": 6.462437395659433e-06, "loss": 0.3697, "step": 21290 }, { "epoch": 1.06, "learning_rate": 6.4607679465776295e-06, "loss": 0.4056, "step": 21300 }, { "epoch": 1.07, "learning_rate": 6.459098497495827e-06, "loss": 0.4531, "step": 21310 }, { "epoch": 1.07, "learning_rate": 6.4574290484140235e-06, "loss": 0.2558, "step": 21320 }, { "epoch": 1.07, "learning_rate": 6.4557595993322205e-06, "loss": 0.3642, "step": 21330 }, { "epoch": 1.07, "learning_rate": 6.454090150250418e-06, "loss": 0.3846, "step": 21340 }, { "epoch": 1.07, "learning_rate": 6.4524207011686145e-06, "loss": 0.3467, "step": 21350 }, { "epoch": 1.07, "learning_rate": 6.450751252086812e-06, "loss": 0.4307, "step": 21360 }, { "epoch": 1.07, "learning_rate": 6.4490818030050085e-06, "loss": 0.3857, "step": 21370 }, { "epoch": 1.07, "learning_rate": 6.4474123539232055e-06, "loss": 0.3853, "step": 21380 }, { "epoch": 1.07, "learning_rate": 6.445742904841403e-06, "loss": 0.478, "step": 21390 }, { "epoch": 1.07, "learning_rate": 6.4440734557595995e-06, "loss": 0.3877, "step": 21400 }, { "epoch": 1.07, "learning_rate": 6.442404006677797e-06, "loss": 0.3255, "step": 21410 }, { "epoch": 1.07, "learning_rate": 6.4407345575959935e-06, "loss": 0.3953, "step": 21420 }, { "epoch": 1.07, "learning_rate": 6.4390651085141905e-06, "loss": 0.402, "step": 21430 }, { "epoch": 1.07, "learning_rate": 6.437395659432388e-06, "loss": 0.4424, "step": 21440 }, { "epoch": 1.07, "learning_rate": 6.4357262103505845e-06, "loss": 0.3028, "step": 21450 }, { "epoch": 1.07, "learning_rate": 6.434056761268782e-06, "loss": 0.3953, "step": 21460 }, { "epoch": 1.07, "learning_rate": 6.4323873121869785e-06, "loss": 0.4621, "step": 21470 }, { "epoch": 1.07, "learning_rate": 6.4307178631051755e-06, "loss": 0.3059, "step": 21480 }, { "epoch": 1.07, "learning_rate": 6.429048414023373e-06, "loss": 0.4057, "step": 21490 }, { "epoch": 1.07, "learning_rate": 6.4273789649415694e-06, "loss": 0.4617, "step": 21500 }, { "epoch": 1.08, "learning_rate": 6.425709515859767e-06, "loss": 0.3513, "step": 21510 }, { "epoch": 1.08, "learning_rate": 6.4240400667779634e-06, "loss": 0.3636, "step": 21520 }, { "epoch": 1.08, "learning_rate": 6.422370617696161e-06, "loss": 0.5222, "step": 21530 }, { "epoch": 1.08, "learning_rate": 6.420701168614358e-06, "loss": 0.3454, "step": 21540 }, { "epoch": 1.08, "learning_rate": 6.4190317195325544e-06, "loss": 0.3591, "step": 21550 }, { "epoch": 1.08, "learning_rate": 6.417362270450752e-06, "loss": 0.2933, "step": 21560 }, { "epoch": 1.08, "learning_rate": 6.4156928213689484e-06, "loss": 0.4125, "step": 21570 }, { "epoch": 1.08, "learning_rate": 6.414023372287146e-06, "loss": 0.3851, "step": 21580 }, { "epoch": 1.08, "learning_rate": 6.4123539232053424e-06, "loss": 0.3867, "step": 21590 }, { "epoch": 1.08, "learning_rate": 6.4106844741235394e-06, "loss": 0.4627, "step": 21600 }, { "epoch": 1.08, "learning_rate": 6.409015025041737e-06, "loss": 0.4463, "step": 21610 }, { "epoch": 1.08, "learning_rate": 6.4073455759599334e-06, "loss": 0.3631, "step": 21620 }, { "epoch": 1.08, "learning_rate": 6.405676126878131e-06, "loss": 0.3805, "step": 21630 }, { "epoch": 1.08, "learning_rate": 6.404006677796327e-06, "loss": 0.3489, "step": 21640 }, { "epoch": 1.08, "learning_rate": 6.402337228714524e-06, "loss": 0.3811, "step": 21650 }, { "epoch": 1.08, "learning_rate": 6.400667779632722e-06, "loss": 0.4143, "step": 21660 }, { "epoch": 1.08, "learning_rate": 6.398998330550918e-06, "loss": 0.3213, "step": 21670 }, { "epoch": 1.08, "learning_rate": 6.397328881469116e-06, "loss": 0.4053, "step": 21680 }, { "epoch": 1.08, "learning_rate": 6.395659432387312e-06, "loss": 0.3087, "step": 21690 }, { "epoch": 1.08, "learning_rate": 6.393989983305509e-06, "loss": 0.3793, "step": 21700 }, { "epoch": 1.09, "learning_rate": 6.392320534223707e-06, "loss": 0.3528, "step": 21710 }, { "epoch": 1.09, "learning_rate": 6.390651085141903e-06, "loss": 0.3217, "step": 21720 }, { "epoch": 1.09, "learning_rate": 6.388981636060101e-06, "loss": 0.4259, "step": 21730 }, { "epoch": 1.09, "learning_rate": 6.387312186978297e-06, "loss": 0.3359, "step": 21740 }, { "epoch": 1.09, "learning_rate": 6.385642737896494e-06, "loss": 0.3287, "step": 21750 }, { "epoch": 1.09, "learning_rate": 6.383973288814692e-06, "loss": 0.395, "step": 21760 }, { "epoch": 1.09, "learning_rate": 6.382303839732888e-06, "loss": 0.3658, "step": 21770 }, { "epoch": 1.09, "learning_rate": 6.380634390651086e-06, "loss": 0.4007, "step": 21780 }, { "epoch": 1.09, "learning_rate": 6.378964941569282e-06, "loss": 0.3701, "step": 21790 }, { "epoch": 1.09, "learning_rate": 6.377295492487479e-06, "loss": 0.3889, "step": 21800 }, { "epoch": 1.09, "learning_rate": 6.375626043405677e-06, "loss": 0.3642, "step": 21810 }, { "epoch": 1.09, "learning_rate": 6.373956594323873e-06, "loss": 0.3464, "step": 21820 }, { "epoch": 1.09, "learning_rate": 6.372287145242071e-06, "loss": 0.344, "step": 21830 }, { "epoch": 1.09, "learning_rate": 6.370617696160267e-06, "loss": 0.3207, "step": 21840 }, { "epoch": 1.09, "learning_rate": 6.368948247078464e-06, "loss": 0.3564, "step": 21850 }, { "epoch": 1.09, "learning_rate": 6.367278797996662e-06, "loss": 0.451, "step": 21860 }, { "epoch": 1.09, "learning_rate": 6.365609348914858e-06, "loss": 0.5357, "step": 21870 }, { "epoch": 1.09, "learning_rate": 6.363939899833056e-06, "loss": 0.3414, "step": 21880 }, { "epoch": 1.09, "learning_rate": 6.362270450751252e-06, "loss": 0.3709, "step": 21890 }, { "epoch": 1.09, "learning_rate": 6.360601001669449e-06, "loss": 0.3109, "step": 21900 }, { "epoch": 1.1, "learning_rate": 6.358931552587646e-06, "loss": 0.2738, "step": 21910 }, { "epoch": 1.1, "learning_rate": 6.357262103505843e-06, "loss": 0.3933, "step": 21920 }, { "epoch": 1.1, "learning_rate": 6.355592654424041e-06, "loss": 0.4066, "step": 21930 }, { "epoch": 1.1, "learning_rate": 6.353923205342237e-06, "loss": 0.3106, "step": 21940 }, { "epoch": 1.1, "learning_rate": 6.352253756260434e-06, "loss": 0.3552, "step": 21950 }, { "epoch": 1.1, "learning_rate": 6.350584307178631e-06, "loss": 0.3409, "step": 21960 }, { "epoch": 1.1, "learning_rate": 6.348914858096828e-06, "loss": 0.3705, "step": 21970 }, { "epoch": 1.1, "learning_rate": 6.347245409015026e-06, "loss": 0.4591, "step": 21980 }, { "epoch": 1.1, "learning_rate": 6.345575959933222e-06, "loss": 0.3314, "step": 21990 }, { "epoch": 1.1, "learning_rate": 6.343906510851419e-06, "loss": 0.4033, "step": 22000 }, { "epoch": 1.1, "learning_rate": 6.342237061769616e-06, "loss": 0.3697, "step": 22010 }, { "epoch": 1.1, "learning_rate": 6.340567612687813e-06, "loss": 0.3866, "step": 22020 }, { "epoch": 1.1, "learning_rate": 6.338898163606011e-06, "loss": 0.3691, "step": 22030 }, { "epoch": 1.1, "learning_rate": 6.337228714524207e-06, "loss": 0.4156, "step": 22040 }, { "epoch": 1.1, "learning_rate": 6.335559265442404e-06, "loss": 0.3618, "step": 22050 }, { "epoch": 1.1, "learning_rate": 6.333889816360601e-06, "loss": 0.376, "step": 22060 }, { "epoch": 1.1, "learning_rate": 6.332220367278798e-06, "loss": 0.3393, "step": 22070 }, { "epoch": 1.1, "learning_rate": 6.330550918196996e-06, "loss": 0.2883, "step": 22080 }, { "epoch": 1.1, "learning_rate": 6.328881469115192e-06, "loss": 0.4053, "step": 22090 }, { "epoch": 1.1, "learning_rate": 6.327212020033389e-06, "loss": 0.4212, "step": 22100 }, { "epoch": 1.11, "learning_rate": 6.325542570951586e-06, "loss": 0.3555, "step": 22110 }, { "epoch": 1.11, "learning_rate": 6.323873121869783e-06, "loss": 0.2925, "step": 22120 }, { "epoch": 1.11, "learning_rate": 6.322203672787981e-06, "loss": 0.437, "step": 22130 }, { "epoch": 1.11, "learning_rate": 6.320534223706177e-06, "loss": 0.3682, "step": 22140 }, { "epoch": 1.11, "learning_rate": 6.318864774624374e-06, "loss": 0.3218, "step": 22150 }, { "epoch": 1.11, "learning_rate": 6.317195325542571e-06, "loss": 0.3632, "step": 22160 }, { "epoch": 1.11, "learning_rate": 6.315525876460768e-06, "loss": 0.4134, "step": 22170 }, { "epoch": 1.11, "learning_rate": 6.313856427378966e-06, "loss": 0.4248, "step": 22180 }, { "epoch": 1.11, "learning_rate": 6.312186978297162e-06, "loss": 0.3749, "step": 22190 }, { "epoch": 1.11, "learning_rate": 6.31051752921536e-06, "loss": 0.3972, "step": 22200 }, { "epoch": 1.11, "learning_rate": 6.308848080133556e-06, "loss": 0.3523, "step": 22210 }, { "epoch": 1.11, "learning_rate": 6.307178631051753e-06, "loss": 0.3111, "step": 22220 }, { "epoch": 1.11, "learning_rate": 6.3055091819699495e-06, "loss": 0.3327, "step": 22230 }, { "epoch": 1.11, "learning_rate": 6.303839732888147e-06, "loss": 0.3782, "step": 22240 }, { "epoch": 1.11, "learning_rate": 6.302170283806345e-06, "loss": 0.3509, "step": 22250 }, { "epoch": 1.11, "learning_rate": 6.300500834724541e-06, "loss": 0.3989, "step": 22260 }, { "epoch": 1.11, "learning_rate": 6.298831385642738e-06, "loss": 0.415, "step": 22270 }, { "epoch": 1.11, "learning_rate": 6.2971619365609345e-06, "loss": 0.3877, "step": 22280 }, { "epoch": 1.11, "learning_rate": 6.295492487479132e-06, "loss": 0.3493, "step": 22290 }, { "epoch": 1.11, "learning_rate": 6.29382303839733e-06, "loss": 0.39, "step": 22300 }, { "epoch": 1.12, "learning_rate": 6.292153589315526e-06, "loss": 0.4215, "step": 22310 }, { "epoch": 1.12, "learning_rate": 6.290484140233723e-06, "loss": 0.3496, "step": 22320 }, { "epoch": 1.12, "learning_rate": 6.28881469115192e-06, "loss": 0.3253, "step": 22330 }, { "epoch": 1.12, "learning_rate": 6.287145242070117e-06, "loss": 0.2953, "step": 22340 }, { "epoch": 1.12, "learning_rate": 6.285475792988315e-06, "loss": 0.3484, "step": 22350 }, { "epoch": 1.12, "learning_rate": 6.283806343906511e-06, "loss": 0.4235, "step": 22360 }, { "epoch": 1.12, "learning_rate": 6.282136894824708e-06, "loss": 0.4058, "step": 22370 }, { "epoch": 1.12, "learning_rate": 6.280467445742905e-06, "loss": 0.3651, "step": 22380 }, { "epoch": 1.12, "learning_rate": 6.278797996661102e-06, "loss": 0.3266, "step": 22390 }, { "epoch": 1.12, "learning_rate": 6.2771285475793e-06, "loss": 0.3724, "step": 22400 }, { "epoch": 1.12, "learning_rate": 6.275459098497496e-06, "loss": 0.393, "step": 22410 }, { "epoch": 1.12, "learning_rate": 6.273789649415693e-06, "loss": 0.4824, "step": 22420 }, { "epoch": 1.12, "learning_rate": 6.27212020033389e-06, "loss": 0.4139, "step": 22430 }, { "epoch": 1.12, "learning_rate": 6.270450751252087e-06, "loss": 0.3059, "step": 22440 }, { "epoch": 1.12, "learning_rate": 6.268781302170285e-06, "loss": 0.4498, "step": 22450 }, { "epoch": 1.12, "learning_rate": 6.267111853088481e-06, "loss": 0.316, "step": 22460 }, { "epoch": 1.12, "learning_rate": 6.265442404006678e-06, "loss": 0.3202, "step": 22470 }, { "epoch": 1.12, "learning_rate": 6.263772954924875e-06, "loss": 0.3704, "step": 22480 }, { "epoch": 1.12, "learning_rate": 6.262103505843072e-06, "loss": 0.4074, "step": 22490 }, { "epoch": 1.12, "learning_rate": 6.26043405676127e-06, "loss": 0.3473, "step": 22500 }, { "epoch": 1.13, "learning_rate": 6.258764607679466e-06, "loss": 0.3964, "step": 22510 }, { "epoch": 1.13, "learning_rate": 6.257095158597663e-06, "loss": 0.4412, "step": 22520 }, { "epoch": 1.13, "learning_rate": 6.25542570951586e-06, "loss": 0.4294, "step": 22530 }, { "epoch": 1.13, "learning_rate": 6.253756260434057e-06, "loss": 0.3818, "step": 22540 }, { "epoch": 1.13, "learning_rate": 6.252086811352255e-06, "loss": 0.3494, "step": 22550 }, { "epoch": 1.13, "learning_rate": 6.250417362270451e-06, "loss": 0.3389, "step": 22560 }, { "epoch": 1.13, "learning_rate": 6.248747913188648e-06, "loss": 0.4166, "step": 22570 }, { "epoch": 1.13, "learning_rate": 6.247078464106845e-06, "loss": 0.3252, "step": 22580 }, { "epoch": 1.13, "learning_rate": 6.245409015025042e-06, "loss": 0.3563, "step": 22590 }, { "epoch": 1.13, "learning_rate": 6.243739565943238e-06, "loss": 0.3567, "step": 22600 }, { "epoch": 1.13, "learning_rate": 6.242070116861436e-06, "loss": 0.3565, "step": 22610 }, { "epoch": 1.13, "learning_rate": 6.240400667779633e-06, "loss": 0.4104, "step": 22620 }, { "epoch": 1.13, "learning_rate": 6.23873121869783e-06, "loss": 0.4213, "step": 22630 }, { "epoch": 1.13, "learning_rate": 6.237061769616027e-06, "loss": 0.3126, "step": 22640 }, { "epoch": 1.13, "learning_rate": 6.235392320534223e-06, "loss": 0.3402, "step": 22650 }, { "epoch": 1.13, "learning_rate": 6.233722871452421e-06, "loss": 0.3226, "step": 22660 }, { "epoch": 1.13, "learning_rate": 6.232053422370618e-06, "loss": 0.4164, "step": 22670 }, { "epoch": 1.13, "learning_rate": 6.230383973288815e-06, "loss": 0.4597, "step": 22680 }, { "epoch": 1.13, "learning_rate": 6.228714524207012e-06, "loss": 0.2868, "step": 22690 }, { "epoch": 1.14, "learning_rate": 6.227045075125208e-06, "loss": 0.2944, "step": 22700 }, { "epoch": 1.14, "learning_rate": 6.225375626043406e-06, "loss": 0.4294, "step": 22710 }, { "epoch": 1.14, "learning_rate": 6.223706176961603e-06, "loss": 0.4838, "step": 22720 }, { "epoch": 1.14, "learning_rate": 6.2220367278798e-06, "loss": 0.3096, "step": 22730 }, { "epoch": 1.14, "learning_rate": 6.220367278797997e-06, "loss": 0.5275, "step": 22740 }, { "epoch": 1.14, "learning_rate": 6.218697829716193e-06, "loss": 0.3717, "step": 22750 }, { "epoch": 1.14, "learning_rate": 6.217028380634391e-06, "loss": 0.3362, "step": 22760 }, { "epoch": 1.14, "learning_rate": 6.215358931552588e-06, "loss": 0.3567, "step": 22770 }, { "epoch": 1.14, "learning_rate": 6.213689482470785e-06, "loss": 0.3784, "step": 22780 }, { "epoch": 1.14, "learning_rate": 6.212020033388982e-06, "loss": 0.4837, "step": 22790 }, { "epoch": 1.14, "learning_rate": 6.210350584307178e-06, "loss": 0.3718, "step": 22800 }, { "epoch": 1.14, "learning_rate": 6.208681135225376e-06, "loss": 0.4386, "step": 22810 }, { "epoch": 1.14, "learning_rate": 6.207011686143573e-06, "loss": 0.3875, "step": 22820 }, { "epoch": 1.14, "learning_rate": 6.20534223706177e-06, "loss": 0.3271, "step": 22830 }, { "epoch": 1.14, "learning_rate": 6.203672787979967e-06, "loss": 0.4329, "step": 22840 }, { "epoch": 1.14, "learning_rate": 6.202003338898163e-06, "loss": 0.3447, "step": 22850 }, { "epoch": 1.14, "learning_rate": 6.200333889816361e-06, "loss": 0.4764, "step": 22860 }, { "epoch": 1.14, "learning_rate": 6.198664440734559e-06, "loss": 0.4028, "step": 22870 }, { "epoch": 1.14, "learning_rate": 6.196994991652755e-06, "loss": 0.3731, "step": 22880 }, { "epoch": 1.14, "learning_rate": 6.195325542570952e-06, "loss": 0.4714, "step": 22890 }, { "epoch": 1.15, "learning_rate": 6.193656093489148e-06, "loss": 0.3144, "step": 22900 }, { "epoch": 1.15, "learning_rate": 6.191986644407346e-06, "loss": 0.3934, "step": 22910 }, { "epoch": 1.15, "learning_rate": 6.190317195325542e-06, "loss": 0.4237, "step": 22920 }, { "epoch": 1.15, "learning_rate": 6.18864774624374e-06, "loss": 0.4214, "step": 22930 }, { "epoch": 1.15, "learning_rate": 6.186978297161937e-06, "loss": 0.4059, "step": 22940 }, { "epoch": 1.15, "learning_rate": 6.185308848080133e-06, "loss": 0.3454, "step": 22950 }, { "epoch": 1.15, "learning_rate": 6.183639398998331e-06, "loss": 0.3808, "step": 22960 }, { "epoch": 1.15, "learning_rate": 6.181969949916527e-06, "loss": 0.3473, "step": 22970 }, { "epoch": 1.15, "learning_rate": 6.180300500834725e-06, "loss": 0.3805, "step": 22980 }, { "epoch": 1.15, "learning_rate": 6.178631051752922e-06, "loss": 0.4542, "step": 22990 }, { "epoch": 1.15, "learning_rate": 6.176961602671119e-06, "loss": 0.3716, "step": 23000 }, { "epoch": 1.15, "learning_rate": 6.175292153589316e-06, "loss": 0.3894, "step": 23010 }, { "epoch": 1.15, "learning_rate": 6.173622704507512e-06, "loss": 0.4461, "step": 23020 }, { "epoch": 1.15, "learning_rate": 6.17195325542571e-06, "loss": 0.3348, "step": 23030 }, { "epoch": 1.15, "learning_rate": 6.170283806343907e-06, "loss": 0.3617, "step": 23040 }, { "epoch": 1.15, "learning_rate": 6.168614357262104e-06, "loss": 0.3819, "step": 23050 }, { "epoch": 1.15, "learning_rate": 6.166944908180301e-06, "loss": 0.3636, "step": 23060 }, { "epoch": 1.15, "learning_rate": 6.165275459098497e-06, "loss": 0.4497, "step": 23070 }, { "epoch": 1.15, "learning_rate": 6.163606010016695e-06, "loss": 0.2931, "step": 23080 }, { "epoch": 1.15, "learning_rate": 6.161936560934892e-06, "loss": 0.3511, "step": 23090 }, { "epoch": 1.16, "learning_rate": 6.160267111853089e-06, "loss": 0.4602, "step": 23100 }, { "epoch": 1.16, "learning_rate": 6.158597662771286e-06, "loss": 0.3616, "step": 23110 }, { "epoch": 1.16, "learning_rate": 6.156928213689482e-06, "loss": 0.2713, "step": 23120 }, { "epoch": 1.16, "learning_rate": 6.15525876460768e-06, "loss": 0.357, "step": 23130 }, { "epoch": 1.16, "learning_rate": 6.153589315525877e-06, "loss": 0.4637, "step": 23140 }, { "epoch": 1.16, "learning_rate": 6.151919866444074e-06, "loss": 0.3847, "step": 23150 }, { "epoch": 1.16, "learning_rate": 6.150250417362271e-06, "loss": 0.3792, "step": 23160 }, { "epoch": 1.16, "learning_rate": 6.148580968280467e-06, "loss": 0.3998, "step": 23170 }, { "epoch": 1.16, "learning_rate": 6.146911519198665e-06, "loss": 0.4522, "step": 23180 }, { "epoch": 1.16, "learning_rate": 6.145242070116862e-06, "loss": 0.3224, "step": 23190 }, { "epoch": 1.16, "learning_rate": 6.143572621035059e-06, "loss": 0.3656, "step": 23200 }, { "epoch": 1.16, "learning_rate": 6.141903171953256e-06, "loss": 0.3615, "step": 23210 }, { "epoch": 1.16, "learning_rate": 6.140233722871452e-06, "loss": 0.4079, "step": 23220 }, { "epoch": 1.16, "learning_rate": 6.13856427378965e-06, "loss": 0.3906, "step": 23230 }, { "epoch": 1.16, "learning_rate": 6.136894824707846e-06, "loss": 0.3833, "step": 23240 }, { "epoch": 1.16, "learning_rate": 6.135225375626044e-06, "loss": 0.3814, "step": 23250 }, { "epoch": 1.16, "learning_rate": 6.133555926544241e-06, "loss": 0.278, "step": 23260 }, { "epoch": 1.16, "learning_rate": 6.131886477462437e-06, "loss": 0.3728, "step": 23270 }, { "epoch": 1.16, "learning_rate": 6.130217028380635e-06, "loss": 0.2988, "step": 23280 }, { "epoch": 1.16, "learning_rate": 6.128547579298831e-06, "loss": 0.4066, "step": 23290 }, { "epoch": 1.17, "learning_rate": 6.126878130217029e-06, "loss": 0.411, "step": 23300 }, { "epoch": 1.17, "learning_rate": 6.125208681135226e-06, "loss": 0.3497, "step": 23310 }, { "epoch": 1.17, "learning_rate": 6.123539232053422e-06, "loss": 0.38, "step": 23320 }, { "epoch": 1.17, "learning_rate": 6.12186978297162e-06, "loss": 0.3483, "step": 23330 }, { "epoch": 1.17, "learning_rate": 6.120200333889816e-06, "loss": 0.4122, "step": 23340 }, { "epoch": 1.17, "learning_rate": 6.118530884808014e-06, "loss": 0.4276, "step": 23350 }, { "epoch": 1.17, "learning_rate": 6.116861435726211e-06, "loss": 0.4272, "step": 23360 }, { "epoch": 1.17, "learning_rate": 6.115191986644407e-06, "loss": 0.4319, "step": 23370 }, { "epoch": 1.17, "learning_rate": 6.113522537562605e-06, "loss": 0.464, "step": 23380 }, { "epoch": 1.17, "learning_rate": 6.111853088480801e-06, "loss": 0.3112, "step": 23390 }, { "epoch": 1.17, "learning_rate": 6.110183639398999e-06, "loss": 0.417, "step": 23400 }, { "epoch": 1.17, "learning_rate": 6.108514190317196e-06, "loss": 0.3301, "step": 23410 }, { "epoch": 1.17, "learning_rate": 6.106844741235392e-06, "loss": 0.3116, "step": 23420 }, { "epoch": 1.17, "learning_rate": 6.10517529215359e-06, "loss": 0.3228, "step": 23430 }, { "epoch": 1.17, "learning_rate": 6.103505843071786e-06, "loss": 0.3708, "step": 23440 }, { "epoch": 1.17, "learning_rate": 6.101836393989984e-06, "loss": 0.3233, "step": 23450 }, { "epoch": 1.17, "learning_rate": 6.100166944908181e-06, "loss": 0.429, "step": 23460 }, { "epoch": 1.17, "learning_rate": 6.098497495826377e-06, "loss": 0.3717, "step": 23470 }, { "epoch": 1.17, "learning_rate": 6.096828046744575e-06, "loss": 0.4392, "step": 23480 }, { "epoch": 1.17, "learning_rate": 6.095158597662771e-06, "loss": 0.2778, "step": 23490 }, { "epoch": 1.18, "learning_rate": 6.093489148580969e-06, "loss": 0.4447, "step": 23500 }, { "epoch": 1.18, "learning_rate": 6.091819699499166e-06, "loss": 0.3972, "step": 23510 }, { "epoch": 1.18, "learning_rate": 6.090150250417362e-06, "loss": 0.4102, "step": 23520 }, { "epoch": 1.18, "learning_rate": 6.08848080133556e-06, "loss": 0.334, "step": 23530 }, { "epoch": 1.18, "learning_rate": 6.086811352253756e-06, "loss": 0.3516, "step": 23540 }, { "epoch": 1.18, "learning_rate": 6.085141903171954e-06, "loss": 0.3749, "step": 23550 }, { "epoch": 1.18, "learning_rate": 6.083472454090151e-06, "loss": 0.5155, "step": 23560 }, { "epoch": 1.18, "learning_rate": 6.081803005008347e-06, "loss": 0.3263, "step": 23570 }, { "epoch": 1.18, "learning_rate": 6.080133555926545e-06, "loss": 0.3412, "step": 23580 }, { "epoch": 1.18, "learning_rate": 6.078464106844741e-06, "loss": 0.3744, "step": 23590 }, { "epoch": 1.18, "learning_rate": 6.076794657762939e-06, "loss": 0.4098, "step": 23600 }, { "epoch": 1.18, "learning_rate": 6.075125208681135e-06, "loss": 0.362, "step": 23610 }, { "epoch": 1.18, "learning_rate": 6.073455759599332e-06, "loss": 0.3706, "step": 23620 }, { "epoch": 1.18, "learning_rate": 6.07178631051753e-06, "loss": 0.3311, "step": 23630 }, { "epoch": 1.18, "learning_rate": 6.070116861435726e-06, "loss": 0.3762, "step": 23640 }, { "epoch": 1.18, "learning_rate": 6.068447412353924e-06, "loss": 0.279, "step": 23650 }, { "epoch": 1.18, "learning_rate": 6.06677796327212e-06, "loss": 0.2931, "step": 23660 }, { "epoch": 1.18, "learning_rate": 6.065108514190318e-06, "loss": 0.3661, "step": 23670 }, { "epoch": 1.18, "learning_rate": 6.063439065108515e-06, "loss": 0.3426, "step": 23680 }, { "epoch": 1.18, "learning_rate": 6.061769616026711e-06, "loss": 0.3834, "step": 23690 }, { "epoch": 1.19, "learning_rate": 6.060100166944909e-06, "loss": 0.399, "step": 23700 }, { "epoch": 1.19, "learning_rate": 6.058430717863105e-06, "loss": 0.3792, "step": 23710 }, { "epoch": 1.19, "learning_rate": 6.056761268781303e-06, "loss": 0.4706, "step": 23720 }, { "epoch": 1.19, "learning_rate": 6.0550918196995e-06, "loss": 0.4713, "step": 23730 }, { "epoch": 1.19, "learning_rate": 6.053422370617696e-06, "loss": 0.3395, "step": 23740 }, { "epoch": 1.19, "learning_rate": 6.051752921535894e-06, "loss": 0.3046, "step": 23750 }, { "epoch": 1.19, "learning_rate": 6.05008347245409e-06, "loss": 0.3496, "step": 23760 }, { "epoch": 1.19, "learning_rate": 6.048414023372288e-06, "loss": 0.4148, "step": 23770 }, { "epoch": 1.19, "learning_rate": 6.046744574290485e-06, "loss": 0.3945, "step": 23780 }, { "epoch": 1.19, "learning_rate": 6.045075125208681e-06, "loss": 0.303, "step": 23790 }, { "epoch": 1.19, "learning_rate": 6.043405676126879e-06, "loss": 0.3836, "step": 23800 }, { "epoch": 1.19, "learning_rate": 6.041736227045075e-06, "loss": 0.3762, "step": 23810 }, { "epoch": 1.19, "learning_rate": 6.040066777963273e-06, "loss": 0.5108, "step": 23820 }, { "epoch": 1.19, "learning_rate": 6.03839732888147e-06, "loss": 0.3335, "step": 23830 }, { "epoch": 1.19, "learning_rate": 6.036727879799666e-06, "loss": 0.3332, "step": 23840 }, { "epoch": 1.19, "learning_rate": 6.035058430717864e-06, "loss": 0.3498, "step": 23850 }, { "epoch": 1.19, "learning_rate": 6.03338898163606e-06, "loss": 0.3561, "step": 23860 }, { "epoch": 1.19, "learning_rate": 6.031719532554258e-06, "loss": 0.3864, "step": 23870 }, { "epoch": 1.19, "learning_rate": 6.030050083472455e-06, "loss": 0.4, "step": 23880 }, { "epoch": 1.19, "learning_rate": 6.028380634390651e-06, "loss": 0.3211, "step": 23890 }, { "epoch": 1.2, "learning_rate": 6.026711185308849e-06, "loss": 0.3025, "step": 23900 }, { "epoch": 1.2, "learning_rate": 6.025041736227045e-06, "loss": 0.3558, "step": 23910 }, { "epoch": 1.2, "learning_rate": 6.023372287145243e-06, "loss": 0.364, "step": 23920 }, { "epoch": 1.2, "learning_rate": 6.021702838063439e-06, "loss": 0.3956, "step": 23930 }, { "epoch": 1.2, "learning_rate": 6.020033388981636e-06, "loss": 0.3995, "step": 23940 }, { "epoch": 1.2, "learning_rate": 6.018363939899834e-06, "loss": 0.3912, "step": 23950 }, { "epoch": 1.2, "learning_rate": 6.01669449081803e-06, "loss": 0.3431, "step": 23960 }, { "epoch": 1.2, "learning_rate": 6.015025041736228e-06, "loss": 0.3662, "step": 23970 }, { "epoch": 1.2, "learning_rate": 6.013355592654424e-06, "loss": 0.3617, "step": 23980 }, { "epoch": 1.2, "learning_rate": 6.011686143572621e-06, "loss": 0.4195, "step": 23990 }, { "epoch": 1.2, "learning_rate": 6.010016694490819e-06, "loss": 0.3396, "step": 24000 }, { "epoch": 1.2, "learning_rate": 6.008347245409015e-06, "loss": 0.3515, "step": 24010 }, { "epoch": 1.2, "learning_rate": 6.006677796327213e-06, "loss": 0.3908, "step": 24020 }, { "epoch": 1.2, "learning_rate": 6.005008347245409e-06, "loss": 0.3751, "step": 24030 }, { "epoch": 1.2, "learning_rate": 6.003338898163606e-06, "loss": 0.3537, "step": 24040 }, { "epoch": 1.2, "learning_rate": 6.001669449081804e-06, "loss": 0.444, "step": 24050 }, { "epoch": 1.2, "learning_rate": 6e-06, "loss": 0.2715, "step": 24060 }, { "epoch": 1.2, "learning_rate": 5.998330550918198e-06, "loss": 0.4333, "step": 24070 }, { "epoch": 1.2, "learning_rate": 5.996661101836394e-06, "loss": 0.3635, "step": 24080 }, { "epoch": 1.2, "learning_rate": 5.994991652754591e-06, "loss": 0.3992, "step": 24090 }, { "epoch": 1.21, "learning_rate": 5.993322203672789e-06, "loss": 0.3928, "step": 24100 }, { "epoch": 1.21, "learning_rate": 5.991652754590985e-06, "loss": 0.4618, "step": 24110 }, { "epoch": 1.21, "learning_rate": 5.989983305509183e-06, "loss": 0.4011, "step": 24120 }, { "epoch": 1.21, "learning_rate": 5.988313856427379e-06, "loss": 0.357, "step": 24130 }, { "epoch": 1.21, "learning_rate": 5.986644407345576e-06, "loss": 0.3455, "step": 24140 }, { "epoch": 1.21, "learning_rate": 5.984974958263774e-06, "loss": 0.3225, "step": 24150 }, { "epoch": 1.21, "learning_rate": 5.98330550918197e-06, "loss": 0.3761, "step": 24160 }, { "epoch": 1.21, "learning_rate": 5.981636060100168e-06, "loss": 0.3869, "step": 24170 }, { "epoch": 1.21, "learning_rate": 5.979966611018364e-06, "loss": 0.4198, "step": 24180 }, { "epoch": 1.21, "learning_rate": 5.978297161936561e-06, "loss": 0.3977, "step": 24190 }, { "epoch": 1.21, "learning_rate": 5.976627712854759e-06, "loss": 0.3916, "step": 24200 }, { "epoch": 1.21, "learning_rate": 5.974958263772955e-06, "loss": 0.3681, "step": 24210 }, { "epoch": 1.21, "learning_rate": 5.973288814691153e-06, "loss": 0.3274, "step": 24220 }, { "epoch": 1.21, "learning_rate": 5.971619365609349e-06, "loss": 0.3653, "step": 24230 }, { "epoch": 1.21, "learning_rate": 5.969949916527546e-06, "loss": 0.4178, "step": 24240 }, { "epoch": 1.21, "learning_rate": 5.968280467445743e-06, "loss": 0.4453, "step": 24250 }, { "epoch": 1.21, "learning_rate": 5.96661101836394e-06, "loss": 0.4903, "step": 24260 }, { "epoch": 1.21, "learning_rate": 5.964941569282138e-06, "loss": 0.3198, "step": 24270 }, { "epoch": 1.21, "learning_rate": 5.963272120200334e-06, "loss": 0.3479, "step": 24280 }, { "epoch": 1.21, "learning_rate": 5.961602671118531e-06, "loss": 0.4363, "step": 24290 }, { "epoch": 1.22, "learning_rate": 5.959933222036728e-06, "loss": 0.3857, "step": 24300 }, { "epoch": 1.22, "learning_rate": 5.958263772954925e-06, "loss": 0.2691, "step": 24310 }, { "epoch": 1.22, "learning_rate": 5.956594323873123e-06, "loss": 0.4581, "step": 24320 }, { "epoch": 1.22, "learning_rate": 5.954924874791319e-06, "loss": 0.4418, "step": 24330 }, { "epoch": 1.22, "learning_rate": 5.953255425709517e-06, "loss": 0.4313, "step": 24340 }, { "epoch": 1.22, "learning_rate": 5.951585976627713e-06, "loss": 0.3152, "step": 24350 }, { "epoch": 1.22, "learning_rate": 5.94991652754591e-06, "loss": 0.3883, "step": 24360 }, { "epoch": 1.22, "learning_rate": 5.948247078464108e-06, "loss": 0.4322, "step": 24370 }, { "epoch": 1.22, "learning_rate": 5.946577629382304e-06, "loss": 0.369, "step": 24380 }, { "epoch": 1.22, "learning_rate": 5.944908180300502e-06, "loss": 0.3439, "step": 24390 }, { "epoch": 1.22, "learning_rate": 5.943238731218698e-06, "loss": 0.3715, "step": 24400 }, { "epoch": 1.22, "learning_rate": 5.941569282136895e-06, "loss": 0.3401, "step": 24410 }, { "epoch": 1.22, "learning_rate": 5.939899833055093e-06, "loss": 0.3113, "step": 24420 }, { "epoch": 1.22, "learning_rate": 5.938230383973289e-06, "loss": 0.4307, "step": 24430 }, { "epoch": 1.22, "learning_rate": 5.936560934891487e-06, "loss": 0.3689, "step": 24440 }, { "epoch": 1.22, "learning_rate": 5.934891485809683e-06, "loss": 0.4229, "step": 24450 }, { "epoch": 1.22, "learning_rate": 5.93322203672788e-06, "loss": 0.3895, "step": 24460 }, { "epoch": 1.22, "learning_rate": 5.931552587646078e-06, "loss": 0.4433, "step": 24470 }, { "epoch": 1.22, "learning_rate": 5.929883138564274e-06, "loss": 0.412, "step": 24480 }, { "epoch": 1.22, "learning_rate": 5.928213689482472e-06, "loss": 0.3559, "step": 24490 }, { "epoch": 1.23, "learning_rate": 5.926544240400668e-06, "loss": 0.3712, "step": 24500 }, { "epoch": 1.23, "learning_rate": 5.924874791318865e-06, "loss": 0.3568, "step": 24510 }, { "epoch": 1.23, "learning_rate": 5.923205342237063e-06, "loss": 0.4039, "step": 24520 }, { "epoch": 1.23, "learning_rate": 5.921535893155259e-06, "loss": 0.4634, "step": 24530 }, { "epoch": 1.23, "learning_rate": 5.919866444073457e-06, "loss": 0.4136, "step": 24540 }, { "epoch": 1.23, "learning_rate": 5.918196994991653e-06, "loss": 0.3577, "step": 24550 }, { "epoch": 1.23, "learning_rate": 5.91652754590985e-06, "loss": 0.4132, "step": 24560 }, { "epoch": 1.23, "learning_rate": 5.914858096828047e-06, "loss": 0.3044, "step": 24570 }, { "epoch": 1.23, "learning_rate": 5.913188647746244e-06, "loss": 0.4617, "step": 24580 }, { "epoch": 1.23, "learning_rate": 5.911519198664442e-06, "loss": 0.4003, "step": 24590 }, { "epoch": 1.23, "learning_rate": 5.909849749582638e-06, "loss": 0.3626, "step": 24600 }, { "epoch": 1.23, "learning_rate": 5.908180300500835e-06, "loss": 0.2703, "step": 24610 }, { "epoch": 1.23, "learning_rate": 5.906510851419032e-06, "loss": 0.4329, "step": 24620 }, { "epoch": 1.23, "learning_rate": 5.904841402337229e-06, "loss": 0.3827, "step": 24630 }, { "epoch": 1.23, "learning_rate": 5.903171953255427e-06, "loss": 0.2938, "step": 24640 }, { "epoch": 1.23, "learning_rate": 5.901502504173623e-06, "loss": 0.4048, "step": 24650 }, { "epoch": 1.23, "learning_rate": 5.89983305509182e-06, "loss": 0.4766, "step": 24660 }, { "epoch": 1.23, "learning_rate": 5.898163606010017e-06, "loss": 0.3284, "step": 24670 }, { "epoch": 1.23, "learning_rate": 5.896494156928214e-06, "loss": 0.5343, "step": 24680 }, { "epoch": 1.23, "learning_rate": 5.894824707846412e-06, "loss": 0.3679, "step": 24690 }, { "epoch": 1.23, "learning_rate": 5.893155258764608e-06, "loss": 0.3623, "step": 24700 }, { "epoch": 1.24, "learning_rate": 5.891485809682805e-06, "loss": 0.3437, "step": 24710 }, { "epoch": 1.24, "learning_rate": 5.889816360601002e-06, "loss": 0.3176, "step": 24720 }, { "epoch": 1.24, "learning_rate": 5.888146911519199e-06, "loss": 0.3243, "step": 24730 }, { "epoch": 1.24, "learning_rate": 5.886477462437397e-06, "loss": 0.3339, "step": 24740 }, { "epoch": 1.24, "learning_rate": 5.884808013355593e-06, "loss": 0.4374, "step": 24750 }, { "epoch": 1.24, "learning_rate": 5.88313856427379e-06, "loss": 0.3724, "step": 24760 }, { "epoch": 1.24, "learning_rate": 5.881469115191987e-06, "loss": 0.3616, "step": 24770 }, { "epoch": 1.24, "learning_rate": 5.879799666110184e-06, "loss": 0.4011, "step": 24780 }, { "epoch": 1.24, "learning_rate": 5.878130217028382e-06, "loss": 0.4121, "step": 24790 }, { "epoch": 1.24, "learning_rate": 5.876460767946578e-06, "loss": 0.4295, "step": 24800 }, { "epoch": 1.24, "learning_rate": 5.874791318864775e-06, "loss": 0.3619, "step": 24810 }, { "epoch": 1.24, "learning_rate": 5.873121869782972e-06, "loss": 0.313, "step": 24820 }, { "epoch": 1.24, "learning_rate": 5.871452420701169e-06, "loss": 0.3728, "step": 24830 }, { "epoch": 1.24, "learning_rate": 5.869782971619367e-06, "loss": 0.4074, "step": 24840 }, { "epoch": 1.24, "learning_rate": 5.868113522537563e-06, "loss": 0.3663, "step": 24850 }, { "epoch": 1.24, "learning_rate": 5.86644407345576e-06, "loss": 0.3587, "step": 24860 }, { "epoch": 1.24, "learning_rate": 5.864774624373957e-06, "loss": 0.2629, "step": 24870 }, { "epoch": 1.24, "learning_rate": 5.863105175292154e-06, "loss": 0.3413, "step": 24880 }, { "epoch": 1.24, "learning_rate": 5.861435726210352e-06, "loss": 0.4239, "step": 24890 }, { "epoch": 1.25, "learning_rate": 5.859766277128548e-06, "loss": 0.379, "step": 24900 }, { "epoch": 1.25, "learning_rate": 5.858096828046745e-06, "loss": 0.4398, "step": 24910 }, { "epoch": 1.25, "learning_rate": 5.856427378964942e-06, "loss": 0.4592, "step": 24920 }, { "epoch": 1.25, "learning_rate": 5.854757929883139e-06, "loss": 0.4698, "step": 24930 }, { "epoch": 1.25, "learning_rate": 5.853088480801335e-06, "loss": 0.343, "step": 24940 }, { "epoch": 1.25, "learning_rate": 5.851419031719533e-06, "loss": 0.3938, "step": 24950 }, { "epoch": 1.25, "learning_rate": 5.84974958263773e-06, "loss": 0.3685, "step": 24960 }, { "epoch": 1.25, "learning_rate": 5.848080133555927e-06, "loss": 0.3358, "step": 24970 }, { "epoch": 1.25, "learning_rate": 5.846410684474124e-06, "loss": 0.3252, "step": 24980 }, { "epoch": 1.25, "learning_rate": 5.84474123539232e-06, "loss": 0.3691, "step": 24990 }, { "epoch": 1.25, "learning_rate": 5.843071786310518e-06, "loss": 0.409, "step": 25000 }, { "epoch": 1.25, "eval_loss": 0.3767710030078888, "eval_runtime": 181.8771, "eval_samples_per_second": 20.701, "eval_steps_per_second": 2.59, "step": 25000 }, { "epoch": 1.25, "learning_rate": 5.841402337228716e-06, "loss": 0.3703, "step": 25010 }, { "epoch": 1.25, "learning_rate": 5.839732888146912e-06, "loss": 0.2911, "step": 25020 }, { "epoch": 1.25, "learning_rate": 5.838063439065109e-06, "loss": 0.3382, "step": 25030 }, { "epoch": 1.25, "learning_rate": 5.836393989983305e-06, "loss": 0.3018, "step": 25040 }, { "epoch": 1.25, "learning_rate": 5.834724540901503e-06, "loss": 0.3766, "step": 25050 }, { "epoch": 1.25, "learning_rate": 5.833055091819701e-06, "loss": 0.4832, "step": 25060 }, { "epoch": 1.25, "learning_rate": 5.831385642737897e-06, "loss": 0.4644, "step": 25070 }, { "epoch": 1.25, "learning_rate": 5.829716193656094e-06, "loss": 0.3238, "step": 25080 }, { "epoch": 1.25, "learning_rate": 5.82804674457429e-06, "loss": 0.338, "step": 25090 }, { "epoch": 1.25, "learning_rate": 5.826377295492488e-06, "loss": 0.3692, "step": 25100 }, { "epoch": 1.26, "learning_rate": 5.824707846410686e-06, "loss": 0.4086, "step": 25110 }, { "epoch": 1.26, "learning_rate": 5.823038397328882e-06, "loss": 0.3086, "step": 25120 }, { "epoch": 1.26, "learning_rate": 5.821368948247079e-06, "loss": 0.4103, "step": 25130 }, { "epoch": 1.26, "learning_rate": 5.819699499165276e-06, "loss": 0.3753, "step": 25140 }, { "epoch": 1.26, "learning_rate": 5.818030050083473e-06, "loss": 0.4937, "step": 25150 }, { "epoch": 1.26, "learning_rate": 5.816360601001671e-06, "loss": 0.3067, "step": 25160 }, { "epoch": 1.26, "learning_rate": 5.814691151919867e-06, "loss": 0.402, "step": 25170 }, { "epoch": 1.26, "learning_rate": 5.813021702838064e-06, "loss": 0.3358, "step": 25180 }, { "epoch": 1.26, "learning_rate": 5.811352253756261e-06, "loss": 0.3704, "step": 25190 }, { "epoch": 1.26, "learning_rate": 5.809682804674458e-06, "loss": 0.2872, "step": 25200 }, { "epoch": 1.26, "learning_rate": 5.808013355592656e-06, "loss": 0.3084, "step": 25210 }, { "epoch": 1.26, "learning_rate": 5.806343906510852e-06, "loss": 0.5274, "step": 25220 }, { "epoch": 1.26, "learning_rate": 5.804674457429049e-06, "loss": 0.2879, "step": 25230 }, { "epoch": 1.26, "learning_rate": 5.803005008347246e-06, "loss": 0.3126, "step": 25240 }, { "epoch": 1.26, "learning_rate": 5.801335559265443e-06, "loss": 0.3936, "step": 25250 }, { "epoch": 1.26, "learning_rate": 5.799666110183639e-06, "loss": 0.3536, "step": 25260 }, { "epoch": 1.26, "learning_rate": 5.797996661101837e-06, "loss": 0.3471, "step": 25270 }, { "epoch": 1.26, "learning_rate": 5.796327212020034e-06, "loss": 0.3411, "step": 25280 }, { "epoch": 1.26, "learning_rate": 5.794657762938231e-06, "loss": 0.366, "step": 25290 }, { "epoch": 1.27, "learning_rate": 5.792988313856428e-06, "loss": 0.4119, "step": 25300 }, { "epoch": 1.27, "learning_rate": 5.791318864774624e-06, "loss": 0.3693, "step": 25310 }, { "epoch": 1.27, "learning_rate": 5.789649415692822e-06, "loss": 0.3504, "step": 25320 }, { "epoch": 1.27, "learning_rate": 5.787979966611019e-06, "loss": 0.3041, "step": 25330 }, { "epoch": 1.27, "learning_rate": 5.786310517529216e-06, "loss": 0.3463, "step": 25340 }, { "epoch": 1.27, "learning_rate": 5.784641068447413e-06, "loss": 0.439, "step": 25350 }, { "epoch": 1.27, "learning_rate": 5.782971619365609e-06, "loss": 0.3533, "step": 25360 }, { "epoch": 1.27, "learning_rate": 5.781302170283807e-06, "loss": 0.3825, "step": 25370 }, { "epoch": 1.27, "learning_rate": 5.779632721202004e-06, "loss": 0.3003, "step": 25380 }, { "epoch": 1.27, "learning_rate": 5.777963272120201e-06, "loss": 0.4671, "step": 25390 }, { "epoch": 1.27, "learning_rate": 5.776293823038398e-06, "loss": 0.4482, "step": 25400 }, { "epoch": 1.27, "learning_rate": 5.774624373956594e-06, "loss": 0.3347, "step": 25410 }, { "epoch": 1.27, "learning_rate": 5.772954924874792e-06, "loss": 0.4018, "step": 25420 }, { "epoch": 1.27, "learning_rate": 5.771285475792989e-06, "loss": 0.3596, "step": 25430 }, { "epoch": 1.27, "learning_rate": 5.769616026711186e-06, "loss": 0.4792, "step": 25440 }, { "epoch": 1.27, "learning_rate": 5.767946577629383e-06, "loss": 0.3606, "step": 25450 }, { "epoch": 1.27, "learning_rate": 5.766277128547579e-06, "loss": 0.4125, "step": 25460 }, { "epoch": 1.27, "learning_rate": 5.764607679465777e-06, "loss": 0.3609, "step": 25470 }, { "epoch": 1.27, "learning_rate": 5.762938230383974e-06, "loss": 0.3311, "step": 25480 }, { "epoch": 1.27, "learning_rate": 5.761268781302171e-06, "loss": 0.3251, "step": 25490 }, { "epoch": 1.27, "learning_rate": 5.759599332220368e-06, "loss": 0.2753, "step": 25500 }, { "epoch": 1.28, "learning_rate": 5.757929883138564e-06, "loss": 0.5479, "step": 25510 }, { "epoch": 1.28, "learning_rate": 5.756260434056762e-06, "loss": 0.3089, "step": 25520 }, { "epoch": 1.28, "learning_rate": 5.754590984974959e-06, "loss": 0.4078, "step": 25530 }, { "epoch": 1.28, "learning_rate": 5.752921535893156e-06, "loss": 0.3544, "step": 25540 }, { "epoch": 1.28, "learning_rate": 5.751252086811353e-06, "loss": 0.4436, "step": 25550 }, { "epoch": 1.28, "learning_rate": 5.749582637729549e-06, "loss": 0.3409, "step": 25560 }, { "epoch": 1.28, "learning_rate": 5.747913188647747e-06, "loss": 0.3227, "step": 25570 }, { "epoch": 1.28, "learning_rate": 5.746243739565943e-06, "loss": 0.3774, "step": 25580 }, { "epoch": 1.28, "learning_rate": 5.744574290484141e-06, "loss": 0.3269, "step": 25590 }, { "epoch": 1.28, "learning_rate": 5.742904841402338e-06, "loss": 0.3434, "step": 25600 }, { "epoch": 1.28, "learning_rate": 5.741235392320534e-06, "loss": 0.329, "step": 25610 }, { "epoch": 1.28, "learning_rate": 5.739565943238732e-06, "loss": 0.3555, "step": 25620 }, { "epoch": 1.28, "learning_rate": 5.737896494156928e-06, "loss": 0.2984, "step": 25630 }, { "epoch": 1.28, "learning_rate": 5.736227045075126e-06, "loss": 0.4475, "step": 25640 }, { "epoch": 1.28, "learning_rate": 5.734557595993323e-06, "loss": 0.3873, "step": 25650 }, { "epoch": 1.28, "learning_rate": 5.732888146911519e-06, "loss": 0.3848, "step": 25660 }, { "epoch": 1.28, "learning_rate": 5.731218697829717e-06, "loss": 0.3498, "step": 25670 }, { "epoch": 1.28, "learning_rate": 5.729549248747913e-06, "loss": 0.3514, "step": 25680 }, { "epoch": 1.28, "learning_rate": 5.727879799666111e-06, "loss": 0.4524, "step": 25690 }, { "epoch": 1.28, "learning_rate": 5.726210350584308e-06, "loss": 0.3618, "step": 25700 }, { "epoch": 1.29, "learning_rate": 5.724540901502504e-06, "loss": 0.4009, "step": 25710 }, { "epoch": 1.29, "learning_rate": 5.722871452420702e-06, "loss": 0.4289, "step": 25720 }, { "epoch": 1.29, "learning_rate": 5.721202003338898e-06, "loss": 0.3616, "step": 25730 }, { "epoch": 1.29, "learning_rate": 5.719532554257096e-06, "loss": 0.4254, "step": 25740 }, { "epoch": 1.29, "learning_rate": 5.717863105175293e-06, "loss": 0.3326, "step": 25750 }, { "epoch": 1.29, "learning_rate": 5.716193656093489e-06, "loss": 0.3223, "step": 25760 }, { "epoch": 1.29, "learning_rate": 5.714524207011687e-06, "loss": 0.342, "step": 25770 }, { "epoch": 1.29, "learning_rate": 5.712854757929883e-06, "loss": 0.3116, "step": 25780 }, { "epoch": 1.29, "learning_rate": 5.711185308848081e-06, "loss": 0.3851, "step": 25790 }, { "epoch": 1.29, "learning_rate": 5.709515859766278e-06, "loss": 0.4425, "step": 25800 }, { "epoch": 1.29, "learning_rate": 5.707846410684475e-06, "loss": 0.3672, "step": 25810 }, { "epoch": 1.29, "learning_rate": 5.706176961602672e-06, "loss": 0.4353, "step": 25820 }, { "epoch": 1.29, "learning_rate": 5.704507512520868e-06, "loss": 0.3797, "step": 25830 }, { "epoch": 1.29, "learning_rate": 5.702838063439066e-06, "loss": 0.4098, "step": 25840 }, { "epoch": 1.29, "learning_rate": 5.701168614357263e-06, "loss": 0.3595, "step": 25850 }, { "epoch": 1.29, "learning_rate": 5.69949916527546e-06, "loss": 0.3777, "step": 25860 }, { "epoch": 1.29, "learning_rate": 5.697829716193657e-06, "loss": 0.2924, "step": 25870 }, { "epoch": 1.29, "learning_rate": 5.696160267111853e-06, "loss": 0.3556, "step": 25880 }, { "epoch": 1.29, "learning_rate": 5.694490818030051e-06, "loss": 0.3438, "step": 25890 }, { "epoch": 1.29, "learning_rate": 5.692821368948247e-06, "loss": 0.4132, "step": 25900 }, { "epoch": 1.3, "learning_rate": 5.691151919866445e-06, "loss": 0.4212, "step": 25910 }, { "epoch": 1.3, "learning_rate": 5.689482470784642e-06, "loss": 0.4433, "step": 25920 }, { "epoch": 1.3, "learning_rate": 5.687813021702838e-06, "loss": 0.3814, "step": 25930 }, { "epoch": 1.3, "learning_rate": 5.686143572621036e-06, "loss": 0.3393, "step": 25940 }, { "epoch": 1.3, "learning_rate": 5.684474123539232e-06, "loss": 0.257, "step": 25950 }, { "epoch": 1.3, "learning_rate": 5.68280467445743e-06, "loss": 0.4191, "step": 25960 }, { "epoch": 1.3, "learning_rate": 5.681135225375627e-06, "loss": 0.4317, "step": 25970 }, { "epoch": 1.3, "learning_rate": 5.679465776293823e-06, "loss": 0.3511, "step": 25980 }, { "epoch": 1.3, "learning_rate": 5.677796327212021e-06, "loss": 0.377, "step": 25990 }, { "epoch": 1.3, "learning_rate": 5.676126878130217e-06, "loss": 0.3672, "step": 26000 }, { "epoch": 1.3, "learning_rate": 5.674457429048415e-06, "loss": 0.3311, "step": 26010 }, { "epoch": 1.3, "learning_rate": 5.672787979966612e-06, "loss": 0.4142, "step": 26020 }, { "epoch": 1.3, "learning_rate": 5.671118530884808e-06, "loss": 0.2775, "step": 26030 }, { "epoch": 1.3, "learning_rate": 5.669449081803006e-06, "loss": 0.3408, "step": 26040 }, { "epoch": 1.3, "learning_rate": 5.667779632721202e-06, "loss": 0.3657, "step": 26050 }, { "epoch": 1.3, "learning_rate": 5.6661101836394e-06, "loss": 0.4319, "step": 26060 }, { "epoch": 1.3, "learning_rate": 5.664440734557597e-06, "loss": 0.427, "step": 26070 }, { "epoch": 1.3, "learning_rate": 5.662771285475793e-06, "loss": 0.3351, "step": 26080 }, { "epoch": 1.3, "learning_rate": 5.661101836393991e-06, "loss": 0.3001, "step": 26090 }, { "epoch": 1.3, "learning_rate": 5.659432387312187e-06, "loss": 0.3663, "step": 26100 }, { "epoch": 1.31, "learning_rate": 5.657762938230385e-06, "loss": 0.4385, "step": 26110 }, { "epoch": 1.31, "learning_rate": 5.656093489148582e-06, "loss": 0.3188, "step": 26120 }, { "epoch": 1.31, "learning_rate": 5.654424040066778e-06, "loss": 0.3308, "step": 26130 }, { "epoch": 1.31, "learning_rate": 5.652754590984976e-06, "loss": 0.3856, "step": 26140 }, { "epoch": 1.31, "learning_rate": 5.651085141903172e-06, "loss": 0.3068, "step": 26150 }, { "epoch": 1.31, "learning_rate": 5.64941569282137e-06, "loss": 0.3983, "step": 26160 }, { "epoch": 1.31, "learning_rate": 5.647746243739567e-06, "loss": 0.3535, "step": 26170 }, { "epoch": 1.31, "learning_rate": 5.646076794657763e-06, "loss": 0.3756, "step": 26180 }, { "epoch": 1.31, "learning_rate": 5.644407345575961e-06, "loss": 0.3335, "step": 26190 }, { "epoch": 1.31, "learning_rate": 5.642737896494157e-06, "loss": 0.2949, "step": 26200 }, { "epoch": 1.31, "learning_rate": 5.641068447412355e-06, "loss": 0.2906, "step": 26210 }, { "epoch": 1.31, "learning_rate": 5.639398998330552e-06, "loss": 0.4197, "step": 26220 }, { "epoch": 1.31, "learning_rate": 5.637729549248748e-06, "loss": 0.3774, "step": 26230 }, { "epoch": 1.31, "learning_rate": 5.636060100166946e-06, "loss": 0.3484, "step": 26240 }, { "epoch": 1.31, "learning_rate": 5.634390651085142e-06, "loss": 0.4661, "step": 26250 }, { "epoch": 1.31, "learning_rate": 5.63272120200334e-06, "loss": 0.3503, "step": 26260 }, { "epoch": 1.31, "learning_rate": 5.631051752921536e-06, "loss": 0.3647, "step": 26270 }, { "epoch": 1.31, "learning_rate": 5.629382303839733e-06, "loss": 0.3854, "step": 26280 }, { "epoch": 1.31, "learning_rate": 5.627712854757931e-06, "loss": 0.3926, "step": 26290 }, { "epoch": 1.31, "learning_rate": 5.626043405676127e-06, "loss": 0.3755, "step": 26300 }, { "epoch": 1.32, "learning_rate": 5.624373956594325e-06, "loss": 0.3671, "step": 26310 }, { "epoch": 1.32, "learning_rate": 5.622704507512521e-06, "loss": 0.3255, "step": 26320 }, { "epoch": 1.32, "learning_rate": 5.621035058430718e-06, "loss": 0.4131, "step": 26330 }, { "epoch": 1.32, "learning_rate": 5.619365609348916e-06, "loss": 0.3764, "step": 26340 }, { "epoch": 1.32, "learning_rate": 5.617696160267112e-06, "loss": 0.369, "step": 26350 }, { "epoch": 1.32, "learning_rate": 5.61602671118531e-06, "loss": 0.4112, "step": 26360 }, { "epoch": 1.32, "learning_rate": 5.614357262103506e-06, "loss": 0.366, "step": 26370 }, { "epoch": 1.32, "learning_rate": 5.612687813021703e-06, "loss": 0.2982, "step": 26380 }, { "epoch": 1.32, "learning_rate": 5.611018363939901e-06, "loss": 0.339, "step": 26390 }, { "epoch": 1.32, "learning_rate": 5.609348914858097e-06, "loss": 0.4065, "step": 26400 }, { "epoch": 1.32, "learning_rate": 5.607679465776295e-06, "loss": 0.3739, "step": 26410 }, { "epoch": 1.32, "learning_rate": 5.606010016694491e-06, "loss": 0.3959, "step": 26420 }, { "epoch": 1.32, "learning_rate": 5.604340567612688e-06, "loss": 0.3323, "step": 26430 }, { "epoch": 1.32, "learning_rate": 5.602671118530886e-06, "loss": 0.3631, "step": 26440 }, { "epoch": 1.32, "learning_rate": 5.601001669449082e-06, "loss": 0.4628, "step": 26450 }, { "epoch": 1.32, "learning_rate": 5.59933222036728e-06, "loss": 0.4585, "step": 26460 }, { "epoch": 1.32, "learning_rate": 5.597662771285476e-06, "loss": 0.4326, "step": 26470 }, { "epoch": 1.32, "learning_rate": 5.595993322203674e-06, "loss": 0.4214, "step": 26480 }, { "epoch": 1.32, "learning_rate": 5.594323873121871e-06, "loss": 0.4691, "step": 26490 }, { "epoch": 1.32, "learning_rate": 5.592654424040067e-06, "loss": 0.3574, "step": 26500 }, { "epoch": 1.33, "learning_rate": 5.590984974958265e-06, "loss": 0.297, "step": 26510 }, { "epoch": 1.33, "learning_rate": 5.589315525876461e-06, "loss": 0.4302, "step": 26520 }, { "epoch": 1.33, "learning_rate": 5.587646076794659e-06, "loss": 0.3986, "step": 26530 }, { "epoch": 1.33, "learning_rate": 5.585976627712856e-06, "loss": 0.4422, "step": 26540 }, { "epoch": 1.33, "learning_rate": 5.584307178631052e-06, "loss": 0.3021, "step": 26550 }, { "epoch": 1.33, "learning_rate": 5.58263772954925e-06, "loss": 0.382, "step": 26560 }, { "epoch": 1.33, "learning_rate": 5.580968280467446e-06, "loss": 0.2552, "step": 26570 }, { "epoch": 1.33, "learning_rate": 5.579298831385644e-06, "loss": 0.4115, "step": 26580 }, { "epoch": 1.33, "learning_rate": 5.57762938230384e-06, "loss": 0.3538, "step": 26590 }, { "epoch": 1.33, "learning_rate": 5.575959933222037e-06, "loss": 0.4097, "step": 26600 }, { "epoch": 1.33, "learning_rate": 5.574290484140235e-06, "loss": 0.3768, "step": 26610 }, { "epoch": 1.33, "learning_rate": 5.572621035058431e-06, "loss": 0.327, "step": 26620 }, { "epoch": 1.33, "learning_rate": 5.570951585976629e-06, "loss": 0.37, "step": 26630 }, { "epoch": 1.33, "learning_rate": 5.569282136894825e-06, "loss": 0.4158, "step": 26640 }, { "epoch": 1.33, "learning_rate": 5.567612687813022e-06, "loss": 0.3909, "step": 26650 }, { "epoch": 1.33, "learning_rate": 5.56594323873122e-06, "loss": 0.3462, "step": 26660 }, { "epoch": 1.33, "learning_rate": 5.564273789649416e-06, "loss": 0.302, "step": 26670 }, { "epoch": 1.33, "learning_rate": 5.562604340567614e-06, "loss": 0.467, "step": 26680 }, { "epoch": 1.33, "learning_rate": 5.56093489148581e-06, "loss": 0.3778, "step": 26690 }, { "epoch": 1.33, "learning_rate": 5.559265442404007e-06, "loss": 0.3734, "step": 26700 }, { "epoch": 1.34, "learning_rate": 5.557595993322205e-06, "loss": 0.3733, "step": 26710 }, { "epoch": 1.34, "learning_rate": 5.555926544240401e-06, "loss": 0.4827, "step": 26720 }, { "epoch": 1.34, "learning_rate": 5.554257095158599e-06, "loss": 0.3501, "step": 26730 }, { "epoch": 1.34, "learning_rate": 5.552587646076795e-06, "loss": 0.3323, "step": 26740 }, { "epoch": 1.34, "learning_rate": 5.550918196994992e-06, "loss": 0.2864, "step": 26750 }, { "epoch": 1.34, "learning_rate": 5.54924874791319e-06, "loss": 0.4556, "step": 26760 }, { "epoch": 1.34, "learning_rate": 5.547579298831386e-06, "loss": 0.3258, "step": 26770 }, { "epoch": 1.34, "learning_rate": 5.545909849749584e-06, "loss": 0.344, "step": 26780 }, { "epoch": 1.34, "learning_rate": 5.54424040066778e-06, "loss": 0.4531, "step": 26790 }, { "epoch": 1.34, "learning_rate": 5.542570951585977e-06, "loss": 0.2933, "step": 26800 }, { "epoch": 1.34, "learning_rate": 5.540901502504175e-06, "loss": 0.3604, "step": 26810 }, { "epoch": 1.34, "learning_rate": 5.539232053422371e-06, "loss": 0.3374, "step": 26820 }, { "epoch": 1.34, "learning_rate": 5.537562604340569e-06, "loss": 0.4195, "step": 26830 }, { "epoch": 1.34, "learning_rate": 5.535893155258765e-06, "loss": 0.4178, "step": 26840 }, { "epoch": 1.34, "learning_rate": 5.534223706176962e-06, "loss": 0.3322, "step": 26850 }, { "epoch": 1.34, "learning_rate": 5.53255425709516e-06, "loss": 0.3897, "step": 26860 }, { "epoch": 1.34, "learning_rate": 5.530884808013356e-06, "loss": 0.3918, "step": 26870 }, { "epoch": 1.34, "learning_rate": 5.529215358931554e-06, "loss": 0.3669, "step": 26880 }, { "epoch": 1.34, "learning_rate": 5.52754590984975e-06, "loss": 0.3798, "step": 26890 }, { "epoch": 1.34, "learning_rate": 5.525876460767947e-06, "loss": 0.3271, "step": 26900 }, { "epoch": 1.35, "learning_rate": 5.524207011686144e-06, "loss": 0.3462, "step": 26910 }, { "epoch": 1.35, "learning_rate": 5.522537562604341e-06, "loss": 0.4147, "step": 26920 }, { "epoch": 1.35, "learning_rate": 5.520868113522539e-06, "loss": 0.4519, "step": 26930 }, { "epoch": 1.35, "learning_rate": 5.519198664440735e-06, "loss": 0.3794, "step": 26940 }, { "epoch": 1.35, "learning_rate": 5.517529215358932e-06, "loss": 0.3999, "step": 26950 }, { "epoch": 1.35, "learning_rate": 5.515859766277129e-06, "loss": 0.4514, "step": 26960 }, { "epoch": 1.35, "learning_rate": 5.514190317195326e-06, "loss": 0.3735, "step": 26970 }, { "epoch": 1.35, "learning_rate": 5.512520868113524e-06, "loss": 0.4935, "step": 26980 }, { "epoch": 1.35, "learning_rate": 5.51085141903172e-06, "loss": 0.3842, "step": 26990 }, { "epoch": 1.35, "learning_rate": 5.509181969949917e-06, "loss": 0.3551, "step": 27000 }, { "epoch": 1.35, "learning_rate": 5.507512520868114e-06, "loss": 0.371, "step": 27010 }, { "epoch": 1.35, "learning_rate": 5.505843071786311e-06, "loss": 0.4251, "step": 27020 }, { "epoch": 1.35, "learning_rate": 5.504173622704509e-06, "loss": 0.4277, "step": 27030 }, { "epoch": 1.35, "learning_rate": 5.502504173622705e-06, "loss": 0.3289, "step": 27040 }, { "epoch": 1.35, "learning_rate": 5.500834724540902e-06, "loss": 0.367, "step": 27050 }, { "epoch": 1.35, "learning_rate": 5.499165275459099e-06, "loss": 0.3142, "step": 27060 }, { "epoch": 1.35, "learning_rate": 5.497495826377296e-06, "loss": 0.4231, "step": 27070 }, { "epoch": 1.35, "learning_rate": 5.495826377295494e-06, "loss": 0.3807, "step": 27080 }, { "epoch": 1.35, "learning_rate": 5.49415692821369e-06, "loss": 0.3813, "step": 27090 }, { "epoch": 1.35, "learning_rate": 5.492487479131887e-06, "loss": 0.3769, "step": 27100 }, { "epoch": 1.36, "learning_rate": 5.490818030050084e-06, "loss": 0.3752, "step": 27110 }, { "epoch": 1.36, "learning_rate": 5.489148580968281e-06, "loss": 0.3927, "step": 27120 }, { "epoch": 1.36, "learning_rate": 5.487479131886479e-06, "loss": 0.334, "step": 27130 }, { "epoch": 1.36, "learning_rate": 5.485809682804675e-06, "loss": 0.4114, "step": 27140 }, { "epoch": 1.36, "learning_rate": 5.484140233722872e-06, "loss": 0.4258, "step": 27150 }, { "epoch": 1.36, "learning_rate": 5.482470784641069e-06, "loss": 0.3822, "step": 27160 }, { "epoch": 1.36, "learning_rate": 5.480801335559266e-06, "loss": 0.3974, "step": 27170 }, { "epoch": 1.36, "learning_rate": 5.479131886477464e-06, "loss": 0.2985, "step": 27180 }, { "epoch": 1.36, "learning_rate": 5.47746243739566e-06, "loss": 0.3293, "step": 27190 }, { "epoch": 1.36, "learning_rate": 5.475792988313858e-06, "loss": 0.4075, "step": 27200 }, { "epoch": 1.36, "learning_rate": 5.474123539232054e-06, "loss": 0.4994, "step": 27210 }, { "epoch": 1.36, "learning_rate": 5.472454090150251e-06, "loss": 0.4471, "step": 27220 }, { "epoch": 1.36, "learning_rate": 5.470784641068447e-06, "loss": 0.3915, "step": 27230 }, { "epoch": 1.36, "learning_rate": 5.469115191986645e-06, "loss": 0.3232, "step": 27240 }, { "epoch": 1.36, "learning_rate": 5.467445742904843e-06, "loss": 0.4257, "step": 27250 }, { "epoch": 1.36, "learning_rate": 5.465776293823039e-06, "loss": 0.417, "step": 27260 }, { "epoch": 1.36, "learning_rate": 5.464106844741236e-06, "loss": 0.382, "step": 27270 }, { "epoch": 1.36, "learning_rate": 5.462437395659433e-06, "loss": 0.3162, "step": 27280 }, { "epoch": 1.36, "learning_rate": 5.46076794657763e-06, "loss": 0.3348, "step": 27290 }, { "epoch": 1.36, "learning_rate": 5.459098497495828e-06, "loss": 0.3803, "step": 27300 }, { "epoch": 1.37, "learning_rate": 5.457429048414024e-06, "loss": 0.3475, "step": 27310 }, { "epoch": 1.37, "learning_rate": 5.455759599332221e-06, "loss": 0.3396, "step": 27320 }, { "epoch": 1.37, "learning_rate": 5.454090150250418e-06, "loss": 0.392, "step": 27330 }, { "epoch": 1.37, "learning_rate": 5.452420701168615e-06, "loss": 0.4389, "step": 27340 }, { "epoch": 1.37, "learning_rate": 5.4507512520868126e-06, "loss": 0.4176, "step": 27350 }, { "epoch": 1.37, "learning_rate": 5.449081803005009e-06, "loss": 0.3137, "step": 27360 }, { "epoch": 1.37, "learning_rate": 5.447412353923206e-06, "loss": 0.4365, "step": 27370 }, { "epoch": 1.37, "learning_rate": 5.445742904841403e-06, "loss": 0.4214, "step": 27380 }, { "epoch": 1.37, "learning_rate": 5.4440734557596e-06, "loss": 0.3505, "step": 27390 }, { "epoch": 1.37, "learning_rate": 5.4424040066777976e-06, "loss": 0.3796, "step": 27400 }, { "epoch": 1.37, "learning_rate": 5.440734557595994e-06, "loss": 0.4527, "step": 27410 }, { "epoch": 1.37, "learning_rate": 5.439065108514191e-06, "loss": 0.3463, "step": 27420 }, { "epoch": 1.37, "learning_rate": 5.437395659432388e-06, "loss": 0.3458, "step": 27430 }, { "epoch": 1.37, "learning_rate": 5.435726210350585e-06, "loss": 0.3033, "step": 27440 }, { "epoch": 1.37, "learning_rate": 5.4340567612687826e-06, "loss": 0.3893, "step": 27450 }, { "epoch": 1.37, "learning_rate": 5.432387312186979e-06, "loss": 0.3482, "step": 27460 }, { "epoch": 1.37, "learning_rate": 5.430717863105176e-06, "loss": 0.3527, "step": 27470 }, { "epoch": 1.37, "learning_rate": 5.429048414023373e-06, "loss": 0.3598, "step": 27480 }, { "epoch": 1.37, "learning_rate": 5.42737896494157e-06, "loss": 0.3331, "step": 27490 }, { "epoch": 1.38, "learning_rate": 5.4257095158597675e-06, "loss": 0.365, "step": 27500 }, { "epoch": 1.38, "learning_rate": 5.424040066777964e-06, "loss": 0.4343, "step": 27510 }, { "epoch": 1.38, "learning_rate": 5.422370617696161e-06, "loss": 0.3212, "step": 27520 }, { "epoch": 1.38, "learning_rate": 5.420701168614358e-06, "loss": 0.3628, "step": 27530 }, { "epoch": 1.38, "learning_rate": 5.419031719532555e-06, "loss": 0.4152, "step": 27540 }, { "epoch": 1.38, "learning_rate": 5.4173622704507525e-06, "loss": 0.3134, "step": 27550 }, { "epoch": 1.38, "learning_rate": 5.415692821368949e-06, "loss": 0.3237, "step": 27560 }, { "epoch": 1.38, "learning_rate": 5.414023372287146e-06, "loss": 0.4141, "step": 27570 }, { "epoch": 1.38, "learning_rate": 5.412353923205343e-06, "loss": 0.3826, "step": 27580 }, { "epoch": 1.38, "learning_rate": 5.41068447412354e-06, "loss": 0.3826, "step": 27590 }, { "epoch": 1.38, "learning_rate": 5.409015025041736e-06, "loss": 0.4423, "step": 27600 }, { "epoch": 1.38, "learning_rate": 5.407345575959934e-06, "loss": 0.3607, "step": 27610 }, { "epoch": 1.38, "learning_rate": 5.405676126878131e-06, "loss": 0.4032, "step": 27620 }, { "epoch": 1.38, "learning_rate": 5.404006677796328e-06, "loss": 0.4304, "step": 27630 }, { "epoch": 1.38, "learning_rate": 5.402337228714525e-06, "loss": 0.3514, "step": 27640 }, { "epoch": 1.38, "learning_rate": 5.400667779632721e-06, "loss": 0.3831, "step": 27650 }, { "epoch": 1.38, "learning_rate": 5.398998330550919e-06, "loss": 0.4101, "step": 27660 }, { "epoch": 1.38, "learning_rate": 5.397328881469116e-06, "loss": 0.3961, "step": 27670 }, { "epoch": 1.38, "learning_rate": 5.395659432387313e-06, "loss": 0.319, "step": 27680 }, { "epoch": 1.38, "learning_rate": 5.39398998330551e-06, "loss": 0.3916, "step": 27690 }, { "epoch": 1.39, "learning_rate": 5.392320534223706e-06, "loss": 0.443, "step": 27700 }, { "epoch": 1.39, "learning_rate": 5.390651085141904e-06, "loss": 0.3505, "step": 27710 }, { "epoch": 1.39, "learning_rate": 5.388981636060101e-06, "loss": 0.3699, "step": 27720 }, { "epoch": 1.39, "learning_rate": 5.387312186978298e-06, "loss": 0.3819, "step": 27730 }, { "epoch": 1.39, "learning_rate": 5.385642737896495e-06, "loss": 0.4413, "step": 27740 }, { "epoch": 1.39, "learning_rate": 5.383973288814691e-06, "loss": 0.401, "step": 27750 }, { "epoch": 1.39, "learning_rate": 5.382303839732889e-06, "loss": 0.4032, "step": 27760 }, { "epoch": 1.39, "learning_rate": 5.380634390651086e-06, "loss": 0.3827, "step": 27770 }, { "epoch": 1.39, "learning_rate": 5.378964941569283e-06, "loss": 0.3539, "step": 27780 }, { "epoch": 1.39, "learning_rate": 5.37729549248748e-06, "loss": 0.3543, "step": 27790 }, { "epoch": 1.39, "learning_rate": 5.375626043405676e-06, "loss": 0.402, "step": 27800 }, { "epoch": 1.39, "learning_rate": 5.373956594323874e-06, "loss": 0.3589, "step": 27810 }, { "epoch": 1.39, "learning_rate": 5.372287145242071e-06, "loss": 0.4233, "step": 27820 }, { "epoch": 1.39, "learning_rate": 5.370617696160268e-06, "loss": 0.3644, "step": 27830 }, { "epoch": 1.39, "learning_rate": 5.368948247078465e-06, "loss": 0.3944, "step": 27840 }, { "epoch": 1.39, "learning_rate": 5.367278797996661e-06, "loss": 0.3601, "step": 27850 }, { "epoch": 1.39, "learning_rate": 5.365609348914859e-06, "loss": 0.3871, "step": 27860 }, { "epoch": 1.39, "learning_rate": 5.3639398998330565e-06, "loss": 0.33, "step": 27870 }, { "epoch": 1.39, "learning_rate": 5.362270450751253e-06, "loss": 0.4, "step": 27880 }, { "epoch": 1.39, "learning_rate": 5.36060100166945e-06, "loss": 0.4188, "step": 27890 }, { "epoch": 1.4, "learning_rate": 5.358931552587646e-06, "loss": 0.3663, "step": 27900 }, { "epoch": 1.4, "learning_rate": 5.357262103505844e-06, "loss": 0.3675, "step": 27910 }, { "epoch": 1.4, "learning_rate": 5.35559265442404e-06, "loss": 0.4711, "step": 27920 }, { "epoch": 1.4, "learning_rate": 5.353923205342238e-06, "loss": 0.3092, "step": 27930 }, { "epoch": 1.4, "learning_rate": 5.352253756260435e-06, "loss": 0.3616, "step": 27940 }, { "epoch": 1.4, "learning_rate": 5.350584307178631e-06, "loss": 0.3329, "step": 27950 }, { "epoch": 1.4, "learning_rate": 5.348914858096829e-06, "loss": 0.3582, "step": 27960 }, { "epoch": 1.4, "learning_rate": 5.347245409015025e-06, "loss": 0.3476, "step": 27970 }, { "epoch": 1.4, "learning_rate": 5.345575959933223e-06, "loss": 0.4766, "step": 27980 }, { "epoch": 1.4, "learning_rate": 5.34390651085142e-06, "loss": 0.3894, "step": 27990 }, { "epoch": 1.4, "learning_rate": 5.342237061769617e-06, "loss": 0.3809, "step": 28000 }, { "epoch": 1.4, "learning_rate": 5.340567612687814e-06, "loss": 0.4249, "step": 28010 }, { "epoch": 1.4, "learning_rate": 5.33889816360601e-06, "loss": 0.3793, "step": 28020 }, { "epoch": 1.4, "learning_rate": 5.337228714524208e-06, "loss": 0.2973, "step": 28030 }, { "epoch": 1.4, "learning_rate": 5.335559265442405e-06, "loss": 0.441, "step": 28040 }, { "epoch": 1.4, "learning_rate": 5.333889816360602e-06, "loss": 0.4781, "step": 28050 }, { "epoch": 1.4, "learning_rate": 5.332220367278799e-06, "loss": 0.326, "step": 28060 }, { "epoch": 1.4, "learning_rate": 5.330550918196995e-06, "loss": 0.3634, "step": 28070 }, { "epoch": 1.4, "learning_rate": 5.328881469115193e-06, "loss": 0.317, "step": 28080 }, { "epoch": 1.4, "learning_rate": 5.32721202003339e-06, "loss": 0.4002, "step": 28090 }, { "epoch": 1.41, "learning_rate": 5.325542570951587e-06, "loss": 0.4348, "step": 28100 }, { "epoch": 1.41, "learning_rate": 5.323873121869784e-06, "loss": 0.436, "step": 28110 }, { "epoch": 1.41, "learning_rate": 5.32220367278798e-06, "loss": 0.3781, "step": 28120 }, { "epoch": 1.41, "learning_rate": 5.320534223706178e-06, "loss": 0.2913, "step": 28130 }, { "epoch": 1.41, "learning_rate": 5.318864774624375e-06, "loss": 0.4138, "step": 28140 }, { "epoch": 1.41, "learning_rate": 5.317195325542572e-06, "loss": 0.3907, "step": 28150 }, { "epoch": 1.41, "learning_rate": 5.315525876460769e-06, "loss": 0.4058, "step": 28160 }, { "epoch": 1.41, "learning_rate": 5.313856427378965e-06, "loss": 0.4146, "step": 28170 }, { "epoch": 1.41, "learning_rate": 5.312186978297163e-06, "loss": 0.3205, "step": 28180 }, { "epoch": 1.41, "learning_rate": 5.31051752921536e-06, "loss": 0.4742, "step": 28190 }, { "epoch": 1.41, "learning_rate": 5.308848080133557e-06, "loss": 0.3584, "step": 28200 }, { "epoch": 1.41, "learning_rate": 5.307178631051754e-06, "loss": 0.4321, "step": 28210 }, { "epoch": 1.41, "learning_rate": 5.30550918196995e-06, "loss": 0.3792, "step": 28220 }, { "epoch": 1.41, "learning_rate": 5.3038397328881476e-06, "loss": 0.4526, "step": 28230 }, { "epoch": 1.41, "learning_rate": 5.302170283806344e-06, "loss": 0.3498, "step": 28240 }, { "epoch": 1.41, "learning_rate": 5.3005008347245416e-06, "loss": 0.4047, "step": 28250 }, { "epoch": 1.41, "learning_rate": 5.2988313856427386e-06, "loss": 0.3009, "step": 28260 }, { "epoch": 1.41, "learning_rate": 5.297161936560935e-06, "loss": 0.3638, "step": 28270 }, { "epoch": 1.41, "learning_rate": 5.2954924874791326e-06, "loss": 0.3755, "step": 28280 }, { "epoch": 1.41, "learning_rate": 5.293823038397329e-06, "loss": 0.4403, "step": 28290 }, { "epoch": 1.42, "learning_rate": 5.2921535893155266e-06, "loss": 0.327, "step": 28300 }, { "epoch": 1.42, "learning_rate": 5.2904841402337236e-06, "loss": 0.4012, "step": 28310 }, { "epoch": 1.42, "learning_rate": 5.28881469115192e-06, "loss": 0.3323, "step": 28320 }, { "epoch": 1.42, "learning_rate": 5.2871452420701176e-06, "loss": 0.5119, "step": 28330 }, { "epoch": 1.42, "learning_rate": 5.285475792988314e-06, "loss": 0.4817, "step": 28340 }, { "epoch": 1.42, "learning_rate": 5.2838063439065116e-06, "loss": 0.2844, "step": 28350 }, { "epoch": 1.42, "learning_rate": 5.2821368948247086e-06, "loss": 0.4321, "step": 28360 }, { "epoch": 1.42, "learning_rate": 5.280467445742905e-06, "loss": 0.3973, "step": 28370 }, { "epoch": 1.42, "learning_rate": 5.2787979966611025e-06, "loss": 0.3434, "step": 28380 }, { "epoch": 1.42, "learning_rate": 5.277128547579299e-06, "loss": 0.3739, "step": 28390 }, { "epoch": 1.42, "learning_rate": 5.2754590984974965e-06, "loss": 0.4035, "step": 28400 }, { "epoch": 1.42, "learning_rate": 5.2737896494156935e-06, "loss": 0.4112, "step": 28410 }, { "epoch": 1.42, "learning_rate": 5.27212020033389e-06, "loss": 0.3199, "step": 28420 }, { "epoch": 1.42, "learning_rate": 5.2704507512520875e-06, "loss": 0.4779, "step": 28430 }, { "epoch": 1.42, "learning_rate": 5.268781302170284e-06, "loss": 0.4166, "step": 28440 }, { "epoch": 1.42, "learning_rate": 5.2671118530884815e-06, "loss": 0.4035, "step": 28450 }, { "epoch": 1.42, "learning_rate": 5.2654424040066785e-06, "loss": 0.3262, "step": 28460 }, { "epoch": 1.42, "learning_rate": 5.263772954924875e-06, "loss": 0.4025, "step": 28470 }, { "epoch": 1.42, "learning_rate": 5.2621035058430725e-06, "loss": 0.3103, "step": 28480 }, { "epoch": 1.42, "learning_rate": 5.260434056761269e-06, "loss": 0.4612, "step": 28490 }, { "epoch": 1.43, "learning_rate": 5.2587646076794665e-06, "loss": 0.4682, "step": 28500 }, { "epoch": 1.43, "learning_rate": 5.2570951585976635e-06, "loss": 0.5035, "step": 28510 }, { "epoch": 1.43, "learning_rate": 5.25542570951586e-06, "loss": 0.361, "step": 28520 }, { "epoch": 1.43, "learning_rate": 5.2537562604340575e-06, "loss": 0.3061, "step": 28530 }, { "epoch": 1.43, "learning_rate": 5.252086811352254e-06, "loss": 0.4055, "step": 28540 }, { "epoch": 1.43, "learning_rate": 5.2504173622704515e-06, "loss": 0.3744, "step": 28550 }, { "epoch": 1.43, "learning_rate": 5.248747913188648e-06, "loss": 0.4106, "step": 28560 }, { "epoch": 1.43, "learning_rate": 5.247078464106845e-06, "loss": 0.3208, "step": 28570 }, { "epoch": 1.43, "learning_rate": 5.2454090150250425e-06, "loss": 0.3924, "step": 28580 }, { "epoch": 1.43, "learning_rate": 5.243739565943239e-06, "loss": 0.4239, "step": 28590 }, { "epoch": 1.43, "learning_rate": 5.2420701168614365e-06, "loss": 0.3896, "step": 28600 }, { "epoch": 1.43, "learning_rate": 5.240400667779633e-06, "loss": 0.4306, "step": 28610 }, { "epoch": 1.43, "learning_rate": 5.23873121869783e-06, "loss": 0.3688, "step": 28620 }, { "epoch": 1.43, "learning_rate": 5.2370617696160275e-06, "loss": 0.2997, "step": 28630 }, { "epoch": 1.43, "learning_rate": 5.235392320534224e-06, "loss": 0.3672, "step": 28640 }, { "epoch": 1.43, "learning_rate": 5.2337228714524215e-06, "loss": 0.409, "step": 28650 }, { "epoch": 1.43, "learning_rate": 5.232053422370618e-06, "loss": 0.4428, "step": 28660 }, { "epoch": 1.43, "learning_rate": 5.2303839732888155e-06, "loss": 0.4142, "step": 28670 }, { "epoch": 1.43, "learning_rate": 5.2287145242070125e-06, "loss": 0.3634, "step": 28680 }, { "epoch": 1.43, "learning_rate": 5.227045075125209e-06, "loss": 0.353, "step": 28690 }, { "epoch": 1.44, "learning_rate": 5.2253756260434065e-06, "loss": 0.4037, "step": 28700 }, { "epoch": 1.44, "learning_rate": 5.223706176961603e-06, "loss": 0.3933, "step": 28710 }, { "epoch": 1.44, "learning_rate": 5.2220367278798005e-06, "loss": 0.4557, "step": 28720 }, { "epoch": 1.44, "learning_rate": 5.2203672787979975e-06, "loss": 0.4434, "step": 28730 }, { "epoch": 1.44, "learning_rate": 5.218697829716194e-06, "loss": 0.4118, "step": 28740 }, { "epoch": 1.44, "learning_rate": 5.2170283806343915e-06, "loss": 0.3131, "step": 28750 }, { "epoch": 1.44, "learning_rate": 5.215358931552588e-06, "loss": 0.3874, "step": 28760 }, { "epoch": 1.44, "learning_rate": 5.2136894824707855e-06, "loss": 0.3902, "step": 28770 }, { "epoch": 1.44, "learning_rate": 5.2120200333889825e-06, "loss": 0.3568, "step": 28780 }, { "epoch": 1.44, "learning_rate": 5.210350584307179e-06, "loss": 0.5024, "step": 28790 }, { "epoch": 1.44, "learning_rate": 5.2086811352253765e-06, "loss": 0.3603, "step": 28800 }, { "epoch": 1.44, "learning_rate": 5.207011686143573e-06, "loss": 0.3681, "step": 28810 }, { "epoch": 1.44, "learning_rate": 5.2053422370617705e-06, "loss": 0.3338, "step": 28820 }, { "epoch": 1.44, "learning_rate": 5.2036727879799675e-06, "loss": 0.39, "step": 28830 }, { "epoch": 1.44, "learning_rate": 5.202003338898164e-06, "loss": 0.3174, "step": 28840 }, { "epoch": 1.44, "learning_rate": 5.2003338898163615e-06, "loss": 0.3464, "step": 28850 }, { "epoch": 1.44, "learning_rate": 5.198664440734558e-06, "loss": 0.4233, "step": 28860 }, { "epoch": 1.44, "learning_rate": 5.1969949916527555e-06, "loss": 0.432, "step": 28870 }, { "epoch": 1.44, "learning_rate": 5.1953255425709525e-06, "loss": 0.4592, "step": 28880 }, { "epoch": 1.44, "learning_rate": 5.193656093489149e-06, "loss": 0.3727, "step": 28890 }, { "epoch": 1.45, "learning_rate": 5.1919866444073465e-06, "loss": 0.4224, "step": 28900 }, { "epoch": 1.45, "learning_rate": 5.190317195325543e-06, "loss": 0.3432, "step": 28910 }, { "epoch": 1.45, "learning_rate": 5.1886477462437405e-06, "loss": 0.3334, "step": 28920 }, { "epoch": 1.45, "learning_rate": 5.186978297161937e-06, "loss": 0.4173, "step": 28930 }, { "epoch": 1.45, "learning_rate": 5.185308848080134e-06, "loss": 0.3433, "step": 28940 }, { "epoch": 1.45, "learning_rate": 5.1836393989983314e-06, "loss": 0.4226, "step": 28950 }, { "epoch": 1.45, "learning_rate": 5.181969949916528e-06, "loss": 0.3992, "step": 28960 }, { "epoch": 1.45, "learning_rate": 5.1803005008347254e-06, "loss": 0.4737, "step": 28970 }, { "epoch": 1.45, "learning_rate": 5.178631051752922e-06, "loss": 0.3995, "step": 28980 }, { "epoch": 1.45, "learning_rate": 5.176961602671119e-06, "loss": 0.261, "step": 28990 }, { "epoch": 1.45, "learning_rate": 5.1752921535893164e-06, "loss": 0.347, "step": 29000 }, { "epoch": 1.45, "learning_rate": 5.173622704507513e-06, "loss": 0.3164, "step": 29010 }, { "epoch": 1.45, "learning_rate": 5.1719532554257104e-06, "loss": 0.3051, "step": 29020 }, { "epoch": 1.45, "learning_rate": 5.170283806343907e-06, "loss": 0.5165, "step": 29030 }, { "epoch": 1.45, "learning_rate": 5.168614357262104e-06, "loss": 0.3627, "step": 29040 }, { "epoch": 1.45, "learning_rate": 5.1669449081803014e-06, "loss": 0.4132, "step": 29050 }, { "epoch": 1.45, "learning_rate": 5.165275459098498e-06, "loss": 0.4022, "step": 29060 }, { "epoch": 1.45, "learning_rate": 5.1636060100166954e-06, "loss": 0.381, "step": 29070 }, { "epoch": 1.45, "learning_rate": 5.161936560934892e-06, "loss": 0.3862, "step": 29080 }, { "epoch": 1.45, "learning_rate": 5.160267111853089e-06, "loss": 0.4365, "step": 29090 }, { "epoch": 1.46, "learning_rate": 5.158597662771286e-06, "loss": 0.3907, "step": 29100 }, { "epoch": 1.46, "learning_rate": 5.1569282136894826e-06, "loss": 0.4029, "step": 29110 }, { "epoch": 1.46, "learning_rate": 5.15525876460768e-06, "loss": 0.4583, "step": 29120 }, { "epoch": 1.46, "learning_rate": 5.1535893155258766e-06, "loss": 0.4496, "step": 29130 }, { "epoch": 1.46, "learning_rate": 5.1519198664440736e-06, "loss": 0.3629, "step": 29140 }, { "epoch": 1.46, "learning_rate": 5.150250417362271e-06, "loss": 0.3568, "step": 29150 }, { "epoch": 1.46, "learning_rate": 5.1485809682804676e-06, "loss": 0.3095, "step": 29160 }, { "epoch": 1.46, "learning_rate": 5.146911519198665e-06, "loss": 0.425, "step": 29170 }, { "epoch": 1.46, "learning_rate": 5.1452420701168616e-06, "loss": 0.4046, "step": 29180 }, { "epoch": 1.46, "learning_rate": 5.1435726210350586e-06, "loss": 0.3851, "step": 29190 }, { "epoch": 1.46, "learning_rate": 5.141903171953256e-06, "loss": 0.3584, "step": 29200 }, { "epoch": 1.46, "learning_rate": 5.1402337228714526e-06, "loss": 0.3537, "step": 29210 }, { "epoch": 1.46, "learning_rate": 5.13856427378965e-06, "loss": 0.3306, "step": 29220 }, { "epoch": 1.46, "learning_rate": 5.1368948247078466e-06, "loss": 0.4159, "step": 29230 }, { "epoch": 1.46, "learning_rate": 5.1352253756260436e-06, "loss": 0.3052, "step": 29240 }, { "epoch": 1.46, "learning_rate": 5.1335559265442405e-06, "loss": 0.373, "step": 29250 }, { "epoch": 1.46, "learning_rate": 5.1318864774624375e-06, "loss": 0.3635, "step": 29260 }, { "epoch": 1.46, "learning_rate": 5.130217028380635e-06, "loss": 0.3162, "step": 29270 }, { "epoch": 1.46, "learning_rate": 5.1285475792988315e-06, "loss": 0.2938, "step": 29280 }, { "epoch": 1.46, "learning_rate": 5.1268781302170285e-06, "loss": 0.4283, "step": 29290 }, { "epoch": 1.47, "learning_rate": 5.1252086811352255e-06, "loss": 0.3889, "step": 29300 }, { "epoch": 1.47, "learning_rate": 5.1235392320534225e-06, "loss": 0.3066, "step": 29310 }, { "epoch": 1.47, "learning_rate": 5.12186978297162e-06, "loss": 0.4897, "step": 29320 }, { "epoch": 1.47, "learning_rate": 5.1202003338898165e-06, "loss": 0.3714, "step": 29330 }, { "epoch": 1.47, "learning_rate": 5.118530884808014e-06, "loss": 0.4163, "step": 29340 }, { "epoch": 1.47, "learning_rate": 5.1168614357262105e-06, "loss": 0.3674, "step": 29350 }, { "epoch": 1.47, "learning_rate": 5.1151919866444075e-06, "loss": 0.3835, "step": 29360 }, { "epoch": 1.47, "learning_rate": 5.113522537562605e-06, "loss": 0.3831, "step": 29370 }, { "epoch": 1.47, "learning_rate": 5.1118530884808015e-06, "loss": 0.358, "step": 29380 }, { "epoch": 1.47, "learning_rate": 5.110183639398999e-06, "loss": 0.3118, "step": 29390 }, { "epoch": 1.47, "learning_rate": 5.1085141903171955e-06, "loss": 0.3604, "step": 29400 }, { "epoch": 1.47, "learning_rate": 5.1068447412353925e-06, "loss": 0.333, "step": 29410 }, { "epoch": 1.47, "learning_rate": 5.10517529215359e-06, "loss": 0.3335, "step": 29420 }, { "epoch": 1.47, "learning_rate": 5.1035058430717865e-06, "loss": 0.361, "step": 29430 }, { "epoch": 1.47, "learning_rate": 5.101836393989984e-06, "loss": 0.3389, "step": 29440 }, { "epoch": 1.47, "learning_rate": 5.1001669449081805e-06, "loss": 0.3683, "step": 29450 }, { "epoch": 1.47, "learning_rate": 5.0984974958263775e-06, "loss": 0.3487, "step": 29460 }, { "epoch": 1.47, "learning_rate": 5.096828046744575e-06, "loss": 0.3767, "step": 29470 }, { "epoch": 1.47, "learning_rate": 5.0951585976627715e-06, "loss": 0.3224, "step": 29480 }, { "epoch": 1.47, "learning_rate": 5.093489148580969e-06, "loss": 0.381, "step": 29490 }, { "epoch": 1.48, "learning_rate": 5.0918196994991655e-06, "loss": 0.4094, "step": 29500 }, { "epoch": 1.48, "learning_rate": 5.0901502504173625e-06, "loss": 0.4002, "step": 29510 }, { "epoch": 1.48, "learning_rate": 5.08848080133556e-06, "loss": 0.3645, "step": 29520 }, { "epoch": 1.48, "learning_rate": 5.0868113522537565e-06, "loss": 0.3447, "step": 29530 }, { "epoch": 1.48, "learning_rate": 5.085141903171954e-06, "loss": 0.2922, "step": 29540 }, { "epoch": 1.48, "learning_rate": 5.0834724540901505e-06, "loss": 0.3541, "step": 29550 }, { "epoch": 1.48, "learning_rate": 5.0818030050083475e-06, "loss": 0.3593, "step": 29560 }, { "epoch": 1.48, "learning_rate": 5.0801335559265445e-06, "loss": 0.4056, "step": 29570 }, { "epoch": 1.48, "learning_rate": 5.0784641068447415e-06, "loss": 0.3341, "step": 29580 }, { "epoch": 1.48, "learning_rate": 5.076794657762939e-06, "loss": 0.3271, "step": 29590 }, { "epoch": 1.48, "learning_rate": 5.0751252086811355e-06, "loss": 0.3136, "step": 29600 }, { "epoch": 1.48, "learning_rate": 5.0734557595993325e-06, "loss": 0.3986, "step": 29610 }, { "epoch": 1.48, "learning_rate": 5.0717863105175295e-06, "loss": 0.3703, "step": 29620 }, { "epoch": 1.48, "learning_rate": 5.0701168614357265e-06, "loss": 0.4338, "step": 29630 }, { "epoch": 1.48, "learning_rate": 5.068447412353924e-06, "loss": 0.3527, "step": 29640 }, { "epoch": 1.48, "learning_rate": 5.0667779632721205e-06, "loss": 0.4424, "step": 29650 }, { "epoch": 1.48, "learning_rate": 5.0651085141903175e-06, "loss": 0.3473, "step": 29660 }, { "epoch": 1.48, "learning_rate": 5.0634390651085145e-06, "loss": 0.3561, "step": 29670 }, { "epoch": 1.48, "learning_rate": 5.0617696160267115e-06, "loss": 0.4663, "step": 29680 }, { "epoch": 1.48, "learning_rate": 5.060100166944909e-06, "loss": 0.3646, "step": 29690 }, { "epoch": 1.48, "learning_rate": 5.0584307178631055e-06, "loss": 0.4445, "step": 29700 }, { "epoch": 1.49, "learning_rate": 5.0567612687813025e-06, "loss": 0.392, "step": 29710 }, { "epoch": 1.49, "learning_rate": 5.0550918196994995e-06, "loss": 0.3765, "step": 29720 }, { "epoch": 1.49, "learning_rate": 5.0534223706176965e-06, "loss": 0.4036, "step": 29730 }, { "epoch": 1.49, "learning_rate": 5.051752921535894e-06, "loss": 0.3453, "step": 29740 }, { "epoch": 1.49, "learning_rate": 5.0500834724540905e-06, "loss": 0.355, "step": 29750 }, { "epoch": 1.49, "learning_rate": 5.0484140233722875e-06, "loss": 0.3699, "step": 29760 }, { "epoch": 1.49, "learning_rate": 5.0467445742904845e-06, "loss": 0.3574, "step": 29770 }, { "epoch": 1.49, "learning_rate": 5.0450751252086815e-06, "loss": 0.3955, "step": 29780 }, { "epoch": 1.49, "learning_rate": 5.043405676126879e-06, "loss": 0.3834, "step": 29790 }, { "epoch": 1.49, "learning_rate": 5.0417362270450755e-06, "loss": 0.3566, "step": 29800 }, { "epoch": 1.49, "learning_rate": 5.0400667779632725e-06, "loss": 0.3363, "step": 29810 }, { "epoch": 1.49, "learning_rate": 5.0383973288814694e-06, "loss": 0.3547, "step": 29820 }, { "epoch": 1.49, "learning_rate": 5.0367278797996664e-06, "loss": 0.3132, "step": 29830 }, { "epoch": 1.49, "learning_rate": 5.035058430717864e-06, "loss": 0.4343, "step": 29840 }, { "epoch": 1.49, "learning_rate": 5.0333889816360604e-06, "loss": 0.3451, "step": 29850 }, { "epoch": 1.49, "learning_rate": 5.0317195325542574e-06, "loss": 0.4631, "step": 29860 }, { "epoch": 1.49, "learning_rate": 5.0300500834724544e-06, "loss": 0.4467, "step": 29870 }, { "epoch": 1.49, "learning_rate": 5.0283806343906514e-06, "loss": 0.4178, "step": 29880 }, { "epoch": 1.49, "learning_rate": 5.026711185308848e-06, "loss": 0.3856, "step": 29890 }, { "epoch": 1.5, "learning_rate": 5.0250417362270454e-06, "loss": 0.4442, "step": 29900 }, { "epoch": 1.5, "learning_rate": 5.0233722871452424e-06, "loss": 0.2776, "step": 29910 }, { "epoch": 1.5, "learning_rate": 5.0217028380634394e-06, "loss": 0.2748, "step": 29920 }, { "epoch": 1.5, "learning_rate": 5.0200333889816364e-06, "loss": 0.3962, "step": 29930 }, { "epoch": 1.5, "learning_rate": 5.018363939899833e-06, "loss": 0.4074, "step": 29940 }, { "epoch": 1.5, "learning_rate": 5.0166944908180304e-06, "loss": 0.2761, "step": 29950 }, { "epoch": 1.5, "learning_rate": 5.0150250417362274e-06, "loss": 0.462, "step": 29960 }, { "epoch": 1.5, "learning_rate": 5.013355592654424e-06, "loss": 0.4202, "step": 29970 }, { "epoch": 1.5, "learning_rate": 5.011686143572621e-06, "loss": 0.3017, "step": 29980 }, { "epoch": 1.5, "learning_rate": 5.0100166944908176e-06, "loss": 0.3746, "step": 29990 }, { "epoch": 1.5, "learning_rate": 5.008347245409015e-06, "loss": 0.4151, "step": 30000 }, { "epoch": 1.5, "eval_loss": 0.3758208751678467, "eval_runtime": 180.1848, "eval_samples_per_second": 20.895, "eval_steps_per_second": 2.614, "step": 30000 }, { "epoch": 1.5, "learning_rate": 5.006677796327213e-06, "loss": 0.3126, "step": 30010 }, { "epoch": 1.5, "learning_rate": 5.005008347245409e-06, "loss": 0.269, "step": 30020 }, { "epoch": 1.5, "learning_rate": 5.003338898163606e-06, "loss": 0.3572, "step": 30030 }, { "epoch": 1.5, "learning_rate": 5.0016694490818026e-06, "loss": 0.3982, "step": 30040 }, { "epoch": 1.5, "learning_rate": 5e-06, "loss": 0.3852, "step": 30050 }, { "epoch": 1.5, "learning_rate": 4.998330550918197e-06, "loss": 0.411, "step": 30060 }, { "epoch": 1.5, "learning_rate": 4.996661101836394e-06, "loss": 0.3582, "step": 30070 }, { "epoch": 1.5, "learning_rate": 4.994991652754591e-06, "loss": 0.3584, "step": 30080 }, { "epoch": 1.5, "learning_rate": 4.993322203672788e-06, "loss": 0.3779, "step": 30090 }, { "epoch": 1.5, "learning_rate": 4.991652754590985e-06, "loss": 0.403, "step": 30100 }, { "epoch": 1.51, "learning_rate": 4.989983305509182e-06, "loss": 0.3428, "step": 30110 }, { "epoch": 1.51, "learning_rate": 4.988313856427379e-06, "loss": 0.3464, "step": 30120 }, { "epoch": 1.51, "learning_rate": 4.986644407345576e-06, "loss": 0.3757, "step": 30130 }, { "epoch": 1.51, "learning_rate": 4.984974958263773e-06, "loss": 0.3651, "step": 30140 }, { "epoch": 1.51, "learning_rate": 4.98330550918197e-06, "loss": 0.3705, "step": 30150 }, { "epoch": 1.51, "learning_rate": 4.981636060100167e-06, "loss": 0.303, "step": 30160 }, { "epoch": 1.51, "learning_rate": 4.979966611018364e-06, "loss": 0.3982, "step": 30170 }, { "epoch": 1.51, "learning_rate": 4.978297161936561e-06, "loss": 0.3163, "step": 30180 }, { "epoch": 1.51, "learning_rate": 4.976627712854758e-06, "loss": 0.4407, "step": 30190 }, { "epoch": 1.51, "learning_rate": 4.974958263772955e-06, "loss": 0.3619, "step": 30200 }, { "epoch": 1.51, "learning_rate": 4.973288814691152e-06, "loss": 0.3077, "step": 30210 }, { "epoch": 1.51, "learning_rate": 4.971619365609349e-06, "loss": 0.3369, "step": 30220 }, { "epoch": 1.51, "learning_rate": 4.969949916527546e-06, "loss": 0.4423, "step": 30230 }, { "epoch": 1.51, "learning_rate": 4.968280467445743e-06, "loss": 0.345, "step": 30240 }, { "epoch": 1.51, "learning_rate": 4.96661101836394e-06, "loss": 0.3061, "step": 30250 }, { "epoch": 1.51, "learning_rate": 4.964941569282137e-06, "loss": 0.3344, "step": 30260 }, { "epoch": 1.51, "learning_rate": 4.963272120200334e-06, "loss": 0.3914, "step": 30270 }, { "epoch": 1.51, "learning_rate": 4.961602671118531e-06, "loss": 0.3587, "step": 30280 }, { "epoch": 1.51, "learning_rate": 4.959933222036728e-06, "loss": 0.385, "step": 30290 }, { "epoch": 1.52, "learning_rate": 4.958263772954925e-06, "loss": 0.3475, "step": 30300 }, { "epoch": 1.52, "learning_rate": 4.956594323873122e-06, "loss": 0.4303, "step": 30310 }, { "epoch": 1.52, "learning_rate": 4.954924874791319e-06, "loss": 0.403, "step": 30320 }, { "epoch": 1.52, "learning_rate": 4.953255425709516e-06, "loss": 0.3548, "step": 30330 }, { "epoch": 1.52, "learning_rate": 4.951585976627713e-06, "loss": 0.3761, "step": 30340 }, { "epoch": 1.52, "learning_rate": 4.94991652754591e-06, "loss": 0.4056, "step": 30350 }, { "epoch": 1.52, "learning_rate": 4.948247078464107e-06, "loss": 0.3606, "step": 30360 }, { "epoch": 1.52, "learning_rate": 4.946577629382304e-06, "loss": 0.4297, "step": 30370 }, { "epoch": 1.52, "learning_rate": 4.944908180300501e-06, "loss": 0.3305, "step": 30380 }, { "epoch": 1.52, "learning_rate": 4.943238731218698e-06, "loss": 0.2907, "step": 30390 }, { "epoch": 1.52, "learning_rate": 4.941569282136895e-06, "loss": 0.3495, "step": 30400 }, { "epoch": 1.52, "learning_rate": 4.939899833055092e-06, "loss": 0.3407, "step": 30410 }, { "epoch": 1.52, "learning_rate": 4.938230383973289e-06, "loss": 0.4276, "step": 30420 }, { "epoch": 1.52, "learning_rate": 4.936560934891486e-06, "loss": 0.4424, "step": 30430 }, { "epoch": 1.52, "learning_rate": 4.934891485809683e-06, "loss": 0.3776, "step": 30440 }, { "epoch": 1.52, "learning_rate": 4.93322203672788e-06, "loss": 0.4512, "step": 30450 }, { "epoch": 1.52, "learning_rate": 4.931552587646077e-06, "loss": 0.4067, "step": 30460 }, { "epoch": 1.52, "learning_rate": 4.929883138564274e-06, "loss": 0.3341, "step": 30470 }, { "epoch": 1.52, "learning_rate": 4.928213689482471e-06, "loss": 0.4493, "step": 30480 }, { "epoch": 1.52, "learning_rate": 4.926544240400668e-06, "loss": 0.3655, "step": 30490 }, { "epoch": 1.52, "learning_rate": 4.924874791318865e-06, "loss": 0.377, "step": 30500 }, { "epoch": 1.53, "learning_rate": 4.923205342237062e-06, "loss": 0.3771, "step": 30510 }, { "epoch": 1.53, "learning_rate": 4.921535893155259e-06, "loss": 0.3659, "step": 30520 }, { "epoch": 1.53, "learning_rate": 4.919866444073456e-06, "loss": 0.3169, "step": 30530 }, { "epoch": 1.53, "learning_rate": 4.918196994991653e-06, "loss": 0.439, "step": 30540 }, { "epoch": 1.53, "learning_rate": 4.91652754590985e-06, "loss": 0.4433, "step": 30550 }, { "epoch": 1.53, "learning_rate": 4.914858096828047e-06, "loss": 0.3474, "step": 30560 }, { "epoch": 1.53, "learning_rate": 4.913188647746244e-06, "loss": 0.2982, "step": 30570 }, { "epoch": 1.53, "learning_rate": 4.911519198664441e-06, "loss": 0.4373, "step": 30580 }, { "epoch": 1.53, "learning_rate": 4.909849749582638e-06, "loss": 0.4277, "step": 30590 }, { "epoch": 1.53, "learning_rate": 4.908180300500835e-06, "loss": 0.3906, "step": 30600 }, { "epoch": 1.53, "learning_rate": 4.906510851419032e-06, "loss": 0.3325, "step": 30610 }, { "epoch": 1.53, "learning_rate": 4.904841402337229e-06, "loss": 0.3327, "step": 30620 }, { "epoch": 1.53, "learning_rate": 4.903171953255426e-06, "loss": 0.3849, "step": 30630 }, { "epoch": 1.53, "learning_rate": 4.901502504173623e-06, "loss": 0.2836, "step": 30640 }, { "epoch": 1.53, "learning_rate": 4.89983305509182e-06, "loss": 0.3939, "step": 30650 }, { "epoch": 1.53, "learning_rate": 4.8981636060100165e-06, "loss": 0.4045, "step": 30660 }, { "epoch": 1.53, "learning_rate": 4.896494156928214e-06, "loss": 0.4682, "step": 30670 }, { "epoch": 1.53, "learning_rate": 4.894824707846411e-06, "loss": 0.3656, "step": 30680 }, { "epoch": 1.53, "learning_rate": 4.893155258764608e-06, "loss": 0.3551, "step": 30690 }, { "epoch": 1.54, "learning_rate": 4.891485809682805e-06, "loss": 0.4702, "step": 30700 }, { "epoch": 1.54, "learning_rate": 4.8898163606010014e-06, "loss": 0.3605, "step": 30710 }, { "epoch": 1.54, "learning_rate": 4.888146911519199e-06, "loss": 0.3576, "step": 30720 }, { "epoch": 1.54, "learning_rate": 4.886477462437396e-06, "loss": 0.4167, "step": 30730 }, { "epoch": 1.54, "learning_rate": 4.884808013355593e-06, "loss": 0.3706, "step": 30740 }, { "epoch": 1.54, "learning_rate": 4.88313856427379e-06, "loss": 0.3242, "step": 30750 }, { "epoch": 1.54, "learning_rate": 4.8814691151919864e-06, "loss": 0.407, "step": 30760 }, { "epoch": 1.54, "learning_rate": 4.879799666110184e-06, "loss": 0.3503, "step": 30770 }, { "epoch": 1.54, "learning_rate": 4.878130217028381e-06, "loss": 0.3932, "step": 30780 }, { "epoch": 1.54, "learning_rate": 4.876460767946578e-06, "loss": 0.3978, "step": 30790 }, { "epoch": 1.54, "learning_rate": 4.874791318864775e-06, "loss": 0.3781, "step": 30800 }, { "epoch": 1.54, "learning_rate": 4.873121869782972e-06, "loss": 0.4507, "step": 30810 }, { "epoch": 1.54, "learning_rate": 4.8714524207011684e-06, "loss": 0.4037, "step": 30820 }, { "epoch": 1.54, "learning_rate": 4.869782971619366e-06, "loss": 0.4218, "step": 30830 }, { "epoch": 1.54, "learning_rate": 4.868113522537563e-06, "loss": 0.2973, "step": 30840 }, { "epoch": 1.54, "learning_rate": 4.86644407345576e-06, "loss": 0.3871, "step": 30850 }, { "epoch": 1.54, "learning_rate": 4.864774624373957e-06, "loss": 0.3587, "step": 30860 }, { "epoch": 1.54, "learning_rate": 4.863105175292153e-06, "loss": 0.4383, "step": 30870 }, { "epoch": 1.54, "learning_rate": 4.861435726210351e-06, "loss": 0.3882, "step": 30880 }, { "epoch": 1.54, "learning_rate": 4.859766277128548e-06, "loss": 0.3586, "step": 30890 }, { "epoch": 1.54, "learning_rate": 4.858096828046745e-06, "loss": 0.3611, "step": 30900 }, { "epoch": 1.55, "learning_rate": 4.856427378964942e-06, "loss": 0.3662, "step": 30910 }, { "epoch": 1.55, "learning_rate": 4.854757929883138e-06, "loss": 0.4175, "step": 30920 }, { "epoch": 1.55, "learning_rate": 4.853088480801336e-06, "loss": 0.4715, "step": 30930 }, { "epoch": 1.55, "learning_rate": 4.851419031719533e-06, "loss": 0.3578, "step": 30940 }, { "epoch": 1.55, "learning_rate": 4.84974958263773e-06, "loss": 0.3943, "step": 30950 }, { "epoch": 1.55, "learning_rate": 4.848080133555927e-06, "loss": 0.4905, "step": 30960 }, { "epoch": 1.55, "learning_rate": 4.846410684474123e-06, "loss": 0.4179, "step": 30970 }, { "epoch": 1.55, "learning_rate": 4.84474123539232e-06, "loss": 0.3551, "step": 30980 }, { "epoch": 1.55, "learning_rate": 4.843071786310518e-06, "loss": 0.4169, "step": 30990 }, { "epoch": 1.55, "learning_rate": 4.841402337228715e-06, "loss": 0.4208, "step": 31000 }, { "epoch": 1.55, "learning_rate": 4.839732888146912e-06, "loss": 0.3753, "step": 31010 }, { "epoch": 1.55, "learning_rate": 4.838063439065108e-06, "loss": 0.3686, "step": 31020 }, { "epoch": 1.55, "learning_rate": 4.836393989983305e-06, "loss": 0.3398, "step": 31030 }, { "epoch": 1.55, "learning_rate": 4.834724540901503e-06, "loss": 0.3247, "step": 31040 }, { "epoch": 1.55, "learning_rate": 4.8330550918197e-06, "loss": 0.3516, "step": 31050 }, { "epoch": 1.55, "learning_rate": 4.831385642737897e-06, "loss": 0.435, "step": 31060 }, { "epoch": 1.55, "learning_rate": 4.829716193656093e-06, "loss": 0.4148, "step": 31070 }, { "epoch": 1.55, "learning_rate": 4.82804674457429e-06, "loss": 0.4243, "step": 31080 }, { "epoch": 1.55, "learning_rate": 4.826377295492488e-06, "loss": 0.3879, "step": 31090 }, { "epoch": 1.56, "learning_rate": 4.824707846410685e-06, "loss": 0.3535, "step": 31100 }, { "epoch": 1.56, "learning_rate": 4.823038397328882e-06, "loss": 0.4066, "step": 31110 }, { "epoch": 1.56, "learning_rate": 4.821368948247078e-06, "loss": 0.4637, "step": 31120 }, { "epoch": 1.56, "learning_rate": 4.819699499165275e-06, "loss": 0.3315, "step": 31130 }, { "epoch": 1.56, "learning_rate": 4.818030050083472e-06, "loss": 0.4134, "step": 31140 }, { "epoch": 1.56, "learning_rate": 4.81636060100167e-06, "loss": 0.3603, "step": 31150 }, { "epoch": 1.56, "learning_rate": 4.814691151919867e-06, "loss": 0.3756, "step": 31160 }, { "epoch": 1.56, "learning_rate": 4.813021702838064e-06, "loss": 0.4551, "step": 31170 }, { "epoch": 1.56, "learning_rate": 4.81135225375626e-06, "loss": 0.2968, "step": 31180 }, { "epoch": 1.56, "learning_rate": 4.809682804674457e-06, "loss": 0.3883, "step": 31190 }, { "epoch": 1.56, "learning_rate": 4.808013355592655e-06, "loss": 0.4471, "step": 31200 }, { "epoch": 1.56, "learning_rate": 4.806343906510852e-06, "loss": 0.3466, "step": 31210 }, { "epoch": 1.56, "learning_rate": 4.804674457429049e-06, "loss": 0.357, "step": 31220 }, { "epoch": 1.56, "learning_rate": 4.803005008347245e-06, "loss": 0.3759, "step": 31230 }, { "epoch": 1.56, "learning_rate": 4.801335559265442e-06, "loss": 0.3993, "step": 31240 }, { "epoch": 1.56, "learning_rate": 4.79966611018364e-06, "loss": 0.4387, "step": 31250 }, { "epoch": 1.56, "learning_rate": 4.797996661101837e-06, "loss": 0.3886, "step": 31260 }, { "epoch": 1.56, "learning_rate": 4.796327212020034e-06, "loss": 0.4961, "step": 31270 }, { "epoch": 1.56, "learning_rate": 4.79465776293823e-06, "loss": 0.3866, "step": 31280 }, { "epoch": 1.56, "learning_rate": 4.792988313856427e-06, "loss": 0.3801, "step": 31290 }, { "epoch": 1.56, "learning_rate": 4.791318864774624e-06, "loss": 0.3614, "step": 31300 }, { "epoch": 1.57, "learning_rate": 4.789649415692822e-06, "loss": 0.3843, "step": 31310 }, { "epoch": 1.57, "learning_rate": 4.787979966611019e-06, "loss": 0.3356, "step": 31320 }, { "epoch": 1.57, "learning_rate": 4.786310517529215e-06, "loss": 0.3715, "step": 31330 }, { "epoch": 1.57, "learning_rate": 4.784641068447412e-06, "loss": 0.3533, "step": 31340 }, { "epoch": 1.57, "learning_rate": 4.782971619365609e-06, "loss": 0.4073, "step": 31350 }, { "epoch": 1.57, "learning_rate": 4.781302170283807e-06, "loss": 0.5262, "step": 31360 }, { "epoch": 1.57, "learning_rate": 4.779632721202004e-06, "loss": 0.3564, "step": 31370 }, { "epoch": 1.57, "learning_rate": 4.7779632721202e-06, "loss": 0.3679, "step": 31380 }, { "epoch": 1.57, "learning_rate": 4.776293823038397e-06, "loss": 0.276, "step": 31390 }, { "epoch": 1.57, "learning_rate": 4.774624373956594e-06, "loss": 0.3979, "step": 31400 }, { "epoch": 1.57, "learning_rate": 4.772954924874792e-06, "loss": 0.3857, "step": 31410 }, { "epoch": 1.57, "learning_rate": 4.771285475792989e-06, "loss": 0.4069, "step": 31420 }, { "epoch": 1.57, "learning_rate": 4.769616026711185e-06, "loss": 0.4402, "step": 31430 }, { "epoch": 1.57, "learning_rate": 4.767946577629382e-06, "loss": 0.3586, "step": 31440 }, { "epoch": 1.57, "learning_rate": 4.766277128547579e-06, "loss": 0.3632, "step": 31450 }, { "epoch": 1.57, "learning_rate": 4.764607679465777e-06, "loss": 0.2972, "step": 31460 }, { "epoch": 1.57, "learning_rate": 4.762938230383974e-06, "loss": 0.4302, "step": 31470 }, { "epoch": 1.57, "learning_rate": 4.761268781302171e-06, "loss": 0.3502, "step": 31480 }, { "epoch": 1.57, "learning_rate": 4.759599332220367e-06, "loss": 0.3932, "step": 31490 }, { "epoch": 1.57, "learning_rate": 4.757929883138564e-06, "loss": 0.3462, "step": 31500 }, { "epoch": 1.58, "learning_rate": 4.756260434056761e-06, "loss": 0.4618, "step": 31510 }, { "epoch": 1.58, "learning_rate": 4.754590984974959e-06, "loss": 0.439, "step": 31520 }, { "epoch": 1.58, "learning_rate": 4.752921535893156e-06, "loss": 0.44, "step": 31530 }, { "epoch": 1.58, "learning_rate": 4.751252086811352e-06, "loss": 0.3649, "step": 31540 }, { "epoch": 1.58, "learning_rate": 4.749582637729549e-06, "loss": 0.3548, "step": 31550 }, { "epoch": 1.58, "learning_rate": 4.747913188647746e-06, "loss": 0.4109, "step": 31560 }, { "epoch": 1.58, "learning_rate": 4.746243739565944e-06, "loss": 0.3632, "step": 31570 }, { "epoch": 1.58, "learning_rate": 4.744574290484141e-06, "loss": 0.4176, "step": 31580 }, { "epoch": 1.58, "learning_rate": 4.742904841402337e-06, "loss": 0.3278, "step": 31590 }, { "epoch": 1.58, "learning_rate": 4.741235392320534e-06, "loss": 0.4266, "step": 31600 }, { "epoch": 1.58, "learning_rate": 4.739565943238731e-06, "loss": 0.3929, "step": 31610 }, { "epoch": 1.58, "learning_rate": 4.737896494156929e-06, "loss": 0.3547, "step": 31620 }, { "epoch": 1.58, "learning_rate": 4.736227045075126e-06, "loss": 0.4086, "step": 31630 }, { "epoch": 1.58, "learning_rate": 4.734557595993322e-06, "loss": 0.3155, "step": 31640 }, { "epoch": 1.58, "learning_rate": 4.732888146911519e-06, "loss": 0.3409, "step": 31650 }, { "epoch": 1.58, "learning_rate": 4.731218697829716e-06, "loss": 0.4119, "step": 31660 }, { "epoch": 1.58, "learning_rate": 4.729549248747913e-06, "loss": 0.4019, "step": 31670 }, { "epoch": 1.58, "learning_rate": 4.727879799666111e-06, "loss": 0.3831, "step": 31680 }, { "epoch": 1.58, "learning_rate": 4.726210350584307e-06, "loss": 0.3994, "step": 31690 }, { "epoch": 1.58, "learning_rate": 4.724540901502504e-06, "loss": 0.403, "step": 31700 }, { "epoch": 1.59, "learning_rate": 4.722871452420701e-06, "loss": 0.4018, "step": 31710 }, { "epoch": 1.59, "learning_rate": 4.721202003338898e-06, "loss": 0.3946, "step": 31720 }, { "epoch": 1.59, "learning_rate": 4.719532554257096e-06, "loss": 0.4345, "step": 31730 }, { "epoch": 1.59, "learning_rate": 4.717863105175292e-06, "loss": 0.3446, "step": 31740 }, { "epoch": 1.59, "learning_rate": 4.716193656093489e-06, "loss": 0.4118, "step": 31750 }, { "epoch": 1.59, "learning_rate": 4.714524207011686e-06, "loss": 0.402, "step": 31760 }, { "epoch": 1.59, "learning_rate": 4.712854757929883e-06, "loss": 0.2837, "step": 31770 }, { "epoch": 1.59, "learning_rate": 4.711185308848081e-06, "loss": 0.334, "step": 31780 }, { "epoch": 1.59, "learning_rate": 4.709515859766277e-06, "loss": 0.4514, "step": 31790 }, { "epoch": 1.59, "learning_rate": 4.707846410684474e-06, "loss": 0.3679, "step": 31800 }, { "epoch": 1.59, "learning_rate": 4.706176961602671e-06, "loss": 0.3851, "step": 31810 }, { "epoch": 1.59, "learning_rate": 4.704507512520868e-06, "loss": 0.3521, "step": 31820 }, { "epoch": 1.59, "learning_rate": 4.702838063439065e-06, "loss": 0.274, "step": 31830 }, { "epoch": 1.59, "learning_rate": 4.701168614357263e-06, "loss": 0.3783, "step": 31840 }, { "epoch": 1.59, "learning_rate": 4.699499165275459e-06, "loss": 0.3112, "step": 31850 }, { "epoch": 1.59, "learning_rate": 4.697829716193656e-06, "loss": 0.295, "step": 31860 }, { "epoch": 1.59, "learning_rate": 4.696160267111853e-06, "loss": 0.325, "step": 31870 }, { "epoch": 1.59, "learning_rate": 4.69449081803005e-06, "loss": 0.3244, "step": 31880 }, { "epoch": 1.59, "learning_rate": 4.692821368948248e-06, "loss": 0.444, "step": 31890 }, { "epoch": 1.59, "learning_rate": 4.691151919866444e-06, "loss": 0.4203, "step": 31900 }, { "epoch": 1.6, "learning_rate": 4.689482470784641e-06, "loss": 0.3725, "step": 31910 }, { "epoch": 1.6, "learning_rate": 4.687813021702838e-06, "loss": 0.3014, "step": 31920 }, { "epoch": 1.6, "learning_rate": 4.686143572621035e-06, "loss": 0.3815, "step": 31930 }, { "epoch": 1.6, "learning_rate": 4.684474123539233e-06, "loss": 0.3628, "step": 31940 }, { "epoch": 1.6, "learning_rate": 4.682804674457429e-06, "loss": 0.4013, "step": 31950 }, { "epoch": 1.6, "learning_rate": 4.681135225375626e-06, "loss": 0.4072, "step": 31960 }, { "epoch": 1.6, "learning_rate": 4.679465776293823e-06, "loss": 0.3518, "step": 31970 }, { "epoch": 1.6, "learning_rate": 4.67779632721202e-06, "loss": 0.3668, "step": 31980 }, { "epoch": 1.6, "learning_rate": 4.676126878130217e-06, "loss": 0.3477, "step": 31990 }, { "epoch": 1.6, "learning_rate": 4.674457429048414e-06, "loss": 0.4316, "step": 32000 }, { "epoch": 1.6, "learning_rate": 4.672787979966611e-06, "loss": 0.3441, "step": 32010 }, { "epoch": 1.6, "learning_rate": 4.671118530884808e-06, "loss": 0.362, "step": 32020 }, { "epoch": 1.6, "learning_rate": 4.669449081803005e-06, "loss": 0.3951, "step": 32030 }, { "epoch": 1.6, "learning_rate": 4.667779632721202e-06, "loss": 0.2935, "step": 32040 }, { "epoch": 1.6, "learning_rate": 4.666110183639399e-06, "loss": 0.3416, "step": 32050 }, { "epoch": 1.6, "learning_rate": 4.664440734557596e-06, "loss": 0.3242, "step": 32060 }, { "epoch": 1.6, "learning_rate": 4.662771285475793e-06, "loss": 0.3585, "step": 32070 }, { "epoch": 1.6, "learning_rate": 4.66110183639399e-06, "loss": 0.4641, "step": 32080 }, { "epoch": 1.6, "learning_rate": 4.659432387312187e-06, "loss": 0.4288, "step": 32090 }, { "epoch": 1.6, "learning_rate": 4.657762938230384e-06, "loss": 0.2588, "step": 32100 }, { "epoch": 1.61, "learning_rate": 4.656093489148581e-06, "loss": 0.4395, "step": 32110 }, { "epoch": 1.61, "learning_rate": 4.654424040066778e-06, "loss": 0.4304, "step": 32120 }, { "epoch": 1.61, "learning_rate": 4.652754590984975e-06, "loss": 0.3212, "step": 32130 }, { "epoch": 1.61, "learning_rate": 4.651085141903172e-06, "loss": 0.3612, "step": 32140 }, { "epoch": 1.61, "learning_rate": 4.649415692821369e-06, "loss": 0.3594, "step": 32150 }, { "epoch": 1.61, "learning_rate": 4.647746243739566e-06, "loss": 0.3207, "step": 32160 }, { "epoch": 1.61, "learning_rate": 4.646076794657763e-06, "loss": 0.3544, "step": 32170 }, { "epoch": 1.61, "learning_rate": 4.64440734557596e-06, "loss": 0.3673, "step": 32180 }, { "epoch": 1.61, "learning_rate": 4.642737896494157e-06, "loss": 0.3014, "step": 32190 }, { "epoch": 1.61, "learning_rate": 4.641068447412354e-06, "loss": 0.3353, "step": 32200 }, { "epoch": 1.61, "learning_rate": 4.639398998330551e-06, "loss": 0.3739, "step": 32210 }, { "epoch": 1.61, "learning_rate": 4.637729549248748e-06, "loss": 0.3888, "step": 32220 }, { "epoch": 1.61, "learning_rate": 4.636060100166945e-06, "loss": 0.3274, "step": 32230 }, { "epoch": 1.61, "learning_rate": 4.634390651085142e-06, "loss": 0.4257, "step": 32240 }, { "epoch": 1.61, "learning_rate": 4.632721202003339e-06, "loss": 0.3291, "step": 32250 }, { "epoch": 1.61, "learning_rate": 4.631051752921536e-06, "loss": 0.3653, "step": 32260 }, { "epoch": 1.61, "learning_rate": 4.629382303839733e-06, "loss": 0.316, "step": 32270 }, { "epoch": 1.61, "learning_rate": 4.62771285475793e-06, "loss": 0.4185, "step": 32280 }, { "epoch": 1.61, "learning_rate": 4.626043405676127e-06, "loss": 0.2872, "step": 32290 }, { "epoch": 1.61, "learning_rate": 4.624373956594324e-06, "loss": 0.2806, "step": 32300 }, { "epoch": 1.62, "learning_rate": 4.622704507512521e-06, "loss": 0.437, "step": 32310 }, { "epoch": 1.62, "learning_rate": 4.621035058430718e-06, "loss": 0.3709, "step": 32320 }, { "epoch": 1.62, "learning_rate": 4.619365609348915e-06, "loss": 0.3217, "step": 32330 }, { "epoch": 1.62, "learning_rate": 4.617696160267112e-06, "loss": 0.4166, "step": 32340 }, { "epoch": 1.62, "learning_rate": 4.616026711185309e-06, "loss": 0.3763, "step": 32350 }, { "epoch": 1.62, "learning_rate": 4.614357262103506e-06, "loss": 0.3703, "step": 32360 }, { "epoch": 1.62, "learning_rate": 4.612687813021703e-06, "loss": 0.4467, "step": 32370 }, { "epoch": 1.62, "learning_rate": 4.6110183639399e-06, "loss": 0.4506, "step": 32380 }, { "epoch": 1.62, "learning_rate": 4.609348914858097e-06, "loss": 0.3288, "step": 32390 }, { "epoch": 1.62, "learning_rate": 4.607679465776294e-06, "loss": 0.4403, "step": 32400 }, { "epoch": 1.62, "learning_rate": 4.606010016694491e-06, "loss": 0.3412, "step": 32410 }, { "epoch": 1.62, "learning_rate": 4.604340567612688e-06, "loss": 0.3595, "step": 32420 }, { "epoch": 1.62, "learning_rate": 4.602671118530885e-06, "loss": 0.3651, "step": 32430 }, { "epoch": 1.62, "learning_rate": 4.601001669449082e-06, "loss": 0.3821, "step": 32440 }, { "epoch": 1.62, "learning_rate": 4.599332220367279e-06, "loss": 0.416, "step": 32450 }, { "epoch": 1.62, "learning_rate": 4.597662771285476e-06, "loss": 0.4219, "step": 32460 }, { "epoch": 1.62, "learning_rate": 4.595993322203673e-06, "loss": 0.2964, "step": 32470 }, { "epoch": 1.62, "learning_rate": 4.59432387312187e-06, "loss": 0.4119, "step": 32480 }, { "epoch": 1.62, "learning_rate": 4.592654424040067e-06, "loss": 0.3596, "step": 32490 }, { "epoch": 1.62, "learning_rate": 4.590984974958264e-06, "loss": 0.4108, "step": 32500 }, { "epoch": 1.63, "learning_rate": 4.589315525876461e-06, "loss": 0.4068, "step": 32510 }, { "epoch": 1.63, "learning_rate": 4.587646076794658e-06, "loss": 0.2806, "step": 32520 }, { "epoch": 1.63, "learning_rate": 4.585976627712855e-06, "loss": 0.3406, "step": 32530 }, { "epoch": 1.63, "learning_rate": 4.584307178631052e-06, "loss": 0.4551, "step": 32540 }, { "epoch": 1.63, "learning_rate": 4.582637729549249e-06, "loss": 0.4772, "step": 32550 }, { "epoch": 1.63, "learning_rate": 4.580968280467446e-06, "loss": 0.3849, "step": 32560 }, { "epoch": 1.63, "learning_rate": 4.579298831385643e-06, "loss": 0.4436, "step": 32570 }, { "epoch": 1.63, "learning_rate": 4.57762938230384e-06, "loss": 0.3297, "step": 32580 }, { "epoch": 1.63, "learning_rate": 4.575959933222037e-06, "loss": 0.3492, "step": 32590 }, { "epoch": 1.63, "learning_rate": 4.574290484140234e-06, "loss": 0.3273, "step": 32600 }, { "epoch": 1.63, "learning_rate": 4.572621035058431e-06, "loss": 0.3789, "step": 32610 }, { "epoch": 1.63, "learning_rate": 4.570951585976628e-06, "loss": 0.4109, "step": 32620 }, { "epoch": 1.63, "learning_rate": 4.569282136894825e-06, "loss": 0.3241, "step": 32630 }, { "epoch": 1.63, "learning_rate": 4.567612687813022e-06, "loss": 0.3603, "step": 32640 }, { "epoch": 1.63, "learning_rate": 4.565943238731219e-06, "loss": 0.3581, "step": 32650 }, { "epoch": 1.63, "learning_rate": 4.564273789649416e-06, "loss": 0.3905, "step": 32660 }, { "epoch": 1.63, "learning_rate": 4.562604340567613e-06, "loss": 0.3888, "step": 32670 }, { "epoch": 1.63, "learning_rate": 4.56093489148581e-06, "loss": 0.4313, "step": 32680 }, { "epoch": 1.63, "learning_rate": 4.559265442404007e-06, "loss": 0.4038, "step": 32690 }, { "epoch": 1.64, "learning_rate": 4.557595993322204e-06, "loss": 0.3784, "step": 32700 }, { "epoch": 1.64, "learning_rate": 4.555926544240401e-06, "loss": 0.4492, "step": 32710 }, { "epoch": 1.64, "learning_rate": 4.554257095158598e-06, "loss": 0.4274, "step": 32720 }, { "epoch": 1.64, "learning_rate": 4.552587646076795e-06, "loss": 0.3334, "step": 32730 }, { "epoch": 1.64, "learning_rate": 4.550918196994992e-06, "loss": 0.4119, "step": 32740 }, { "epoch": 1.64, "learning_rate": 4.549248747913189e-06, "loss": 0.3314, "step": 32750 }, { "epoch": 1.64, "learning_rate": 4.547579298831386e-06, "loss": 0.3635, "step": 32760 }, { "epoch": 1.64, "learning_rate": 4.545909849749583e-06, "loss": 0.4074, "step": 32770 }, { "epoch": 1.64, "learning_rate": 4.54424040066778e-06, "loss": 0.3953, "step": 32780 }, { "epoch": 1.64, "learning_rate": 4.542570951585977e-06, "loss": 0.3688, "step": 32790 }, { "epoch": 1.64, "learning_rate": 4.540901502504174e-06, "loss": 0.3836, "step": 32800 }, { "epoch": 1.64, "learning_rate": 4.539232053422371e-06, "loss": 0.3931, "step": 32810 }, { "epoch": 1.64, "learning_rate": 4.537562604340568e-06, "loss": 0.4094, "step": 32820 }, { "epoch": 1.64, "learning_rate": 4.535893155258765e-06, "loss": 0.3657, "step": 32830 }, { "epoch": 1.64, "learning_rate": 4.534223706176962e-06, "loss": 0.3698, "step": 32840 }, { "epoch": 1.64, "learning_rate": 4.532554257095159e-06, "loss": 0.3987, "step": 32850 }, { "epoch": 1.64, "learning_rate": 4.530884808013356e-06, "loss": 0.4062, "step": 32860 }, { "epoch": 1.64, "learning_rate": 4.529215358931553e-06, "loss": 0.3333, "step": 32870 }, { "epoch": 1.64, "learning_rate": 4.52754590984975e-06, "loss": 0.3951, "step": 32880 }, { "epoch": 1.64, "learning_rate": 4.525876460767947e-06, "loss": 0.3883, "step": 32890 }, { "epoch": 1.65, "learning_rate": 4.524207011686144e-06, "loss": 0.4531, "step": 32900 }, { "epoch": 1.65, "learning_rate": 4.522537562604341e-06, "loss": 0.3405, "step": 32910 }, { "epoch": 1.65, "learning_rate": 4.520868113522538e-06, "loss": 0.3938, "step": 32920 }, { "epoch": 1.65, "learning_rate": 4.519198664440735e-06, "loss": 0.4717, "step": 32930 }, { "epoch": 1.65, "learning_rate": 4.517529215358932e-06, "loss": 0.2949, "step": 32940 }, { "epoch": 1.65, "learning_rate": 4.515859766277129e-06, "loss": 0.404, "step": 32950 }, { "epoch": 1.65, "learning_rate": 4.514190317195326e-06, "loss": 0.4456, "step": 32960 }, { "epoch": 1.65, "learning_rate": 4.512520868113523e-06, "loss": 0.4015, "step": 32970 }, { "epoch": 1.65, "learning_rate": 4.51085141903172e-06, "loss": 0.346, "step": 32980 }, { "epoch": 1.65, "learning_rate": 4.509181969949917e-06, "loss": 0.4338, "step": 32990 }, { "epoch": 1.65, "learning_rate": 4.507512520868114e-06, "loss": 0.3047, "step": 33000 }, { "epoch": 1.65, "learning_rate": 4.505843071786311e-06, "loss": 0.3711, "step": 33010 }, { "epoch": 1.65, "learning_rate": 4.504173622704508e-06, "loss": 0.4182, "step": 33020 }, { "epoch": 1.65, "learning_rate": 4.502504173622705e-06, "loss": 0.4382, "step": 33030 }, { "epoch": 1.65, "learning_rate": 4.500834724540902e-06, "loss": 0.3597, "step": 33040 }, { "epoch": 1.65, "learning_rate": 4.499165275459099e-06, "loss": 0.5216, "step": 33050 }, { "epoch": 1.65, "learning_rate": 4.497495826377296e-06, "loss": 0.3793, "step": 33060 }, { "epoch": 1.65, "learning_rate": 4.495826377295493e-06, "loss": 0.3756, "step": 33070 }, { "epoch": 1.65, "learning_rate": 4.49415692821369e-06, "loss": 0.356, "step": 33080 }, { "epoch": 1.65, "learning_rate": 4.492487479131887e-06, "loss": 0.41, "step": 33090 }, { "epoch": 1.66, "learning_rate": 4.490818030050084e-06, "loss": 0.3639, "step": 33100 }, { "epoch": 1.66, "learning_rate": 4.489148580968281e-06, "loss": 0.3849, "step": 33110 }, { "epoch": 1.66, "learning_rate": 4.487479131886478e-06, "loss": 0.4244, "step": 33120 }, { "epoch": 1.66, "learning_rate": 4.485809682804675e-06, "loss": 0.3223, "step": 33130 }, { "epoch": 1.66, "learning_rate": 4.484140233722872e-06, "loss": 0.3393, "step": 33140 }, { "epoch": 1.66, "learning_rate": 4.482470784641069e-06, "loss": 0.4442, "step": 33150 }, { "epoch": 1.66, "learning_rate": 4.480801335559265e-06, "loss": 0.5016, "step": 33160 }, { "epoch": 1.66, "learning_rate": 4.479131886477463e-06, "loss": 0.3318, "step": 33170 }, { "epoch": 1.66, "learning_rate": 4.47746243739566e-06, "loss": 0.3543, "step": 33180 }, { "epoch": 1.66, "learning_rate": 4.475792988313857e-06, "loss": 0.318, "step": 33190 }, { "epoch": 1.66, "learning_rate": 4.474123539232054e-06, "loss": 0.4414, "step": 33200 }, { "epoch": 1.66, "learning_rate": 4.47245409015025e-06, "loss": 0.3412, "step": 33210 }, { "epoch": 1.66, "learning_rate": 4.470784641068448e-06, "loss": 0.4186, "step": 33220 }, { "epoch": 1.66, "learning_rate": 4.469115191986645e-06, "loss": 0.3768, "step": 33230 }, { "epoch": 1.66, "learning_rate": 4.467445742904842e-06, "loss": 0.367, "step": 33240 }, { "epoch": 1.66, "learning_rate": 4.465776293823039e-06, "loss": 0.4271, "step": 33250 }, { "epoch": 1.66, "learning_rate": 4.464106844741235e-06, "loss": 0.415, "step": 33260 }, { "epoch": 1.66, "learning_rate": 4.462437395659433e-06, "loss": 0.3262, "step": 33270 }, { "epoch": 1.66, "learning_rate": 4.46076794657763e-06, "loss": 0.4005, "step": 33280 }, { "epoch": 1.66, "learning_rate": 4.459098497495827e-06, "loss": 0.3037, "step": 33290 }, { "epoch": 1.67, "learning_rate": 4.457429048414024e-06, "loss": 0.4173, "step": 33300 }, { "epoch": 1.67, "learning_rate": 4.455759599332221e-06, "loss": 0.3373, "step": 33310 }, { "epoch": 1.67, "learning_rate": 4.454090150250417e-06, "loss": 0.4441, "step": 33320 }, { "epoch": 1.67, "learning_rate": 4.452420701168615e-06, "loss": 0.393, "step": 33330 }, { "epoch": 1.67, "learning_rate": 4.450751252086812e-06, "loss": 0.4668, "step": 33340 }, { "epoch": 1.67, "learning_rate": 4.449081803005009e-06, "loss": 0.3161, "step": 33350 }, { "epoch": 1.67, "learning_rate": 4.447412353923206e-06, "loss": 0.3687, "step": 33360 }, { "epoch": 1.67, "learning_rate": 4.445742904841402e-06, "loss": 0.3849, "step": 33370 }, { "epoch": 1.67, "learning_rate": 4.4440734557596e-06, "loss": 0.4292, "step": 33380 }, { "epoch": 1.67, "learning_rate": 4.442404006677797e-06, "loss": 0.4935, "step": 33390 }, { "epoch": 1.67, "learning_rate": 4.440734557595994e-06, "loss": 0.4012, "step": 33400 }, { "epoch": 1.67, "learning_rate": 4.439065108514191e-06, "loss": 0.4687, "step": 33410 }, { "epoch": 1.67, "learning_rate": 4.437395659432387e-06, "loss": 0.3655, "step": 33420 }, { "epoch": 1.67, "learning_rate": 4.435726210350585e-06, "loss": 0.3384, "step": 33430 }, { "epoch": 1.67, "learning_rate": 4.434056761268782e-06, "loss": 0.3834, "step": 33440 }, { "epoch": 1.67, "learning_rate": 4.432387312186979e-06, "loss": 0.3908, "step": 33450 }, { "epoch": 1.67, "learning_rate": 4.430717863105176e-06, "loss": 0.4129, "step": 33460 }, { "epoch": 1.67, "learning_rate": 4.429048414023372e-06, "loss": 0.3686, "step": 33470 }, { "epoch": 1.67, "learning_rate": 4.427378964941569e-06, "loss": 0.3075, "step": 33480 }, { "epoch": 1.67, "learning_rate": 4.425709515859767e-06, "loss": 0.3675, "step": 33490 }, { "epoch": 1.68, "learning_rate": 4.424040066777964e-06, "loss": 0.3509, "step": 33500 }, { "epoch": 1.68, "learning_rate": 4.422370617696161e-06, "loss": 0.4355, "step": 33510 }, { "epoch": 1.68, "learning_rate": 4.420701168614357e-06, "loss": 0.3674, "step": 33520 }, { "epoch": 1.68, "learning_rate": 4.419031719532554e-06, "loss": 0.4147, "step": 33530 }, { "epoch": 1.68, "learning_rate": 4.417362270450752e-06, "loss": 0.3709, "step": 33540 }, { "epoch": 1.68, "learning_rate": 4.415692821368949e-06, "loss": 0.3477, "step": 33550 }, { "epoch": 1.68, "learning_rate": 4.414023372287146e-06, "loss": 0.3811, "step": 33560 }, { "epoch": 1.68, "learning_rate": 4.412353923205342e-06, "loss": 0.3336, "step": 33570 }, { "epoch": 1.68, "learning_rate": 4.410684474123539e-06, "loss": 0.4146, "step": 33580 }, { "epoch": 1.68, "learning_rate": 4.409015025041737e-06, "loss": 0.3761, "step": 33590 }, { "epoch": 1.68, "learning_rate": 4.407345575959934e-06, "loss": 0.3794, "step": 33600 }, { "epoch": 1.68, "learning_rate": 4.405676126878131e-06, "loss": 0.3749, "step": 33610 }, { "epoch": 1.68, "learning_rate": 4.404006677796327e-06, "loss": 0.3277, "step": 33620 }, { "epoch": 1.68, "learning_rate": 4.402337228714524e-06, "loss": 0.3343, "step": 33630 }, { "epoch": 1.68, "learning_rate": 4.400667779632721e-06, "loss": 0.3489, "step": 33640 }, { "epoch": 1.68, "learning_rate": 4.398998330550919e-06, "loss": 0.4181, "step": 33650 }, { "epoch": 1.68, "learning_rate": 4.397328881469116e-06, "loss": 0.3205, "step": 33660 }, { "epoch": 1.68, "learning_rate": 4.395659432387313e-06, "loss": 0.4268, "step": 33670 }, { "epoch": 1.68, "learning_rate": 4.393989983305509e-06, "loss": 0.3886, "step": 33680 }, { "epoch": 1.68, "learning_rate": 4.392320534223706e-06, "loss": 0.3855, "step": 33690 }, { "epoch": 1.69, "learning_rate": 4.390651085141904e-06, "loss": 0.3232, "step": 33700 }, { "epoch": 1.69, "learning_rate": 4.388981636060101e-06, "loss": 0.4106, "step": 33710 }, { "epoch": 1.69, "learning_rate": 4.387312186978298e-06, "loss": 0.4495, "step": 33720 }, { "epoch": 1.69, "learning_rate": 4.385642737896494e-06, "loss": 0.3204, "step": 33730 }, { "epoch": 1.69, "learning_rate": 4.383973288814691e-06, "loss": 0.3644, "step": 33740 }, { "epoch": 1.69, "learning_rate": 4.382303839732889e-06, "loss": 0.4649, "step": 33750 }, { "epoch": 1.69, "learning_rate": 4.380634390651086e-06, "loss": 0.4428, "step": 33760 }, { "epoch": 1.69, "learning_rate": 4.378964941569283e-06, "loss": 0.4383, "step": 33770 }, { "epoch": 1.69, "learning_rate": 4.377295492487479e-06, "loss": 0.3616, "step": 33780 }, { "epoch": 1.69, "learning_rate": 4.375626043405676e-06, "loss": 0.3528, "step": 33790 }, { "epoch": 1.69, "learning_rate": 4.373956594323873e-06, "loss": 0.3435, "step": 33800 }, { "epoch": 1.69, "learning_rate": 4.372287145242071e-06, "loss": 0.3422, "step": 33810 }, { "epoch": 1.69, "learning_rate": 4.370617696160268e-06, "loss": 0.4319, "step": 33820 }, { "epoch": 1.69, "learning_rate": 4.368948247078464e-06, "loss": 0.3079, "step": 33830 }, { "epoch": 1.69, "learning_rate": 4.367278797996661e-06, "loss": 0.387, "step": 33840 }, { "epoch": 1.69, "learning_rate": 4.365609348914858e-06, "loss": 0.3912, "step": 33850 }, { "epoch": 1.69, "learning_rate": 4.363939899833056e-06, "loss": 0.3978, "step": 33860 }, { "epoch": 1.69, "learning_rate": 4.362270450751253e-06, "loss": 0.3748, "step": 33870 }, { "epoch": 1.69, "learning_rate": 4.360601001669449e-06, "loss": 0.3796, "step": 33880 }, { "epoch": 1.69, "learning_rate": 4.358931552587646e-06, "loss": 0.3584, "step": 33890 }, { "epoch": 1.69, "learning_rate": 4.357262103505843e-06, "loss": 0.4149, "step": 33900 }, { "epoch": 1.7, "learning_rate": 4.355592654424041e-06, "loss": 0.3534, "step": 33910 }, { "epoch": 1.7, "learning_rate": 4.353923205342238e-06, "loss": 0.4296, "step": 33920 }, { "epoch": 1.7, "learning_rate": 4.352253756260434e-06, "loss": 0.4139, "step": 33930 }, { "epoch": 1.7, "learning_rate": 4.350584307178631e-06, "loss": 0.413, "step": 33940 }, { "epoch": 1.7, "learning_rate": 4.348914858096828e-06, "loss": 0.4107, "step": 33950 }, { "epoch": 1.7, "learning_rate": 4.347245409015026e-06, "loss": 0.389, "step": 33960 }, { "epoch": 1.7, "learning_rate": 4.345575959933223e-06, "loss": 0.3055, "step": 33970 }, { "epoch": 1.7, "learning_rate": 4.34390651085142e-06, "loss": 0.3277, "step": 33980 }, { "epoch": 1.7, "learning_rate": 4.342237061769616e-06, "loss": 0.3612, "step": 33990 }, { "epoch": 1.7, "learning_rate": 4.340567612687813e-06, "loss": 0.2928, "step": 34000 }, { "epoch": 1.7, "learning_rate": 4.33889816360601e-06, "loss": 0.3938, "step": 34010 }, { "epoch": 1.7, "learning_rate": 4.337228714524208e-06, "loss": 0.3885, "step": 34020 }, { "epoch": 1.7, "learning_rate": 4.335559265442405e-06, "loss": 0.4212, "step": 34030 }, { "epoch": 1.7, "learning_rate": 4.333889816360601e-06, "loss": 0.4455, "step": 34040 }, { "epoch": 1.7, "learning_rate": 4.332220367278798e-06, "loss": 0.2695, "step": 34050 }, { "epoch": 1.7, "learning_rate": 4.330550918196995e-06, "loss": 0.3846, "step": 34060 }, { "epoch": 1.7, "learning_rate": 4.328881469115193e-06, "loss": 0.4396, "step": 34070 }, { "epoch": 1.7, "learning_rate": 4.32721202003339e-06, "loss": 0.393, "step": 34080 }, { "epoch": 1.7, "learning_rate": 4.325542570951586e-06, "loss": 0.4444, "step": 34090 }, { "epoch": 1.71, "learning_rate": 4.323873121869783e-06, "loss": 0.4295, "step": 34100 }, { "epoch": 1.71, "learning_rate": 4.32220367278798e-06, "loss": 0.2752, "step": 34110 }, { "epoch": 1.71, "learning_rate": 4.320534223706178e-06, "loss": 0.3578, "step": 34120 }, { "epoch": 1.71, "learning_rate": 4.318864774624375e-06, "loss": 0.3464, "step": 34130 }, { "epoch": 1.71, "learning_rate": 4.317195325542571e-06, "loss": 0.387, "step": 34140 }, { "epoch": 1.71, "learning_rate": 4.315525876460768e-06, "loss": 0.426, "step": 34150 }, { "epoch": 1.71, "learning_rate": 4.313856427378965e-06, "loss": 0.3559, "step": 34160 }, { "epoch": 1.71, "learning_rate": 4.312186978297162e-06, "loss": 0.3319, "step": 34170 }, { "epoch": 1.71, "learning_rate": 4.31051752921536e-06, "loss": 0.4187, "step": 34180 }, { "epoch": 1.71, "learning_rate": 4.308848080133556e-06, "loss": 0.3814, "step": 34190 }, { "epoch": 1.71, "learning_rate": 4.307178631051753e-06, "loss": 0.3902, "step": 34200 }, { "epoch": 1.71, "learning_rate": 4.30550918196995e-06, "loss": 0.4435, "step": 34210 }, { "epoch": 1.71, "learning_rate": 4.303839732888147e-06, "loss": 0.3403, "step": 34220 }, { "epoch": 1.71, "learning_rate": 4.302170283806345e-06, "loss": 0.2736, "step": 34230 }, { "epoch": 1.71, "learning_rate": 4.300500834724541e-06, "loss": 0.3205, "step": 34240 }, { "epoch": 1.71, "learning_rate": 4.298831385642738e-06, "loss": 0.3268, "step": 34250 }, { "epoch": 1.71, "learning_rate": 4.297161936560935e-06, "loss": 0.2753, "step": 34260 }, { "epoch": 1.71, "learning_rate": 4.295492487479132e-06, "loss": 0.4397, "step": 34270 }, { "epoch": 1.71, "learning_rate": 4.29382303839733e-06, "loss": 0.4089, "step": 34280 }, { "epoch": 1.71, "learning_rate": 4.292153589315526e-06, "loss": 0.3259, "step": 34290 }, { "epoch": 1.71, "learning_rate": 4.290484140233723e-06, "loss": 0.3476, "step": 34300 }, { "epoch": 1.72, "learning_rate": 4.28881469115192e-06, "loss": 0.3698, "step": 34310 }, { "epoch": 1.72, "learning_rate": 4.287145242070117e-06, "loss": 0.3967, "step": 34320 }, { "epoch": 1.72, "learning_rate": 4.285475792988314e-06, "loss": 0.415, "step": 34330 }, { "epoch": 1.72, "learning_rate": 4.283806343906512e-06, "loss": 0.3636, "step": 34340 }, { "epoch": 1.72, "learning_rate": 4.282136894824708e-06, "loss": 0.3028, "step": 34350 }, { "epoch": 1.72, "learning_rate": 4.280467445742905e-06, "loss": 0.3582, "step": 34360 }, { "epoch": 1.72, "learning_rate": 4.278797996661102e-06, "loss": 0.3979, "step": 34370 }, { "epoch": 1.72, "learning_rate": 4.277128547579299e-06, "loss": 0.3861, "step": 34380 }, { "epoch": 1.72, "learning_rate": 4.275459098497497e-06, "loss": 0.3671, "step": 34390 }, { "epoch": 1.72, "learning_rate": 4.273789649415693e-06, "loss": 0.4113, "step": 34400 }, { "epoch": 1.72, "learning_rate": 4.27212020033389e-06, "loss": 0.3851, "step": 34410 }, { "epoch": 1.72, "learning_rate": 4.270450751252087e-06, "loss": 0.3489, "step": 34420 }, { "epoch": 1.72, "learning_rate": 4.268781302170284e-06, "loss": 0.3633, "step": 34430 }, { "epoch": 1.72, "learning_rate": 4.267111853088482e-06, "loss": 0.4842, "step": 34440 }, { "epoch": 1.72, "learning_rate": 4.265442404006678e-06, "loss": 0.4282, "step": 34450 }, { "epoch": 1.72, "learning_rate": 4.263772954924875e-06, "loss": 0.3866, "step": 34460 }, { "epoch": 1.72, "learning_rate": 4.262103505843072e-06, "loss": 0.5203, "step": 34470 }, { "epoch": 1.72, "learning_rate": 4.260434056761269e-06, "loss": 0.3194, "step": 34480 }, { "epoch": 1.72, "learning_rate": 4.258764607679466e-06, "loss": 0.326, "step": 34490 }, { "epoch": 1.73, "learning_rate": 4.257095158597663e-06, "loss": 0.3087, "step": 34500 }, { "epoch": 1.73, "learning_rate": 4.25542570951586e-06, "loss": 0.4491, "step": 34510 }, { "epoch": 1.73, "learning_rate": 4.253756260434057e-06, "loss": 0.3563, "step": 34520 }, { "epoch": 1.73, "learning_rate": 4.252086811352254e-06, "loss": 0.3964, "step": 34530 }, { "epoch": 1.73, "learning_rate": 4.250417362270451e-06, "loss": 0.3921, "step": 34540 }, { "epoch": 1.73, "learning_rate": 4.248747913188648e-06, "loss": 0.4553, "step": 34550 }, { "epoch": 1.73, "learning_rate": 4.247078464106845e-06, "loss": 0.327, "step": 34560 }, { "epoch": 1.73, "learning_rate": 4.245409015025042e-06, "loss": 0.3396, "step": 34570 }, { "epoch": 1.73, "learning_rate": 4.243739565943239e-06, "loss": 0.3196, "step": 34580 }, { "epoch": 1.73, "learning_rate": 4.242070116861436e-06, "loss": 0.3337, "step": 34590 }, { "epoch": 1.73, "learning_rate": 4.240400667779633e-06, "loss": 0.3025, "step": 34600 }, { "epoch": 1.73, "learning_rate": 4.23873121869783e-06, "loss": 0.2823, "step": 34610 }, { "epoch": 1.73, "learning_rate": 4.237061769616027e-06, "loss": 0.3656, "step": 34620 }, { "epoch": 1.73, "learning_rate": 4.235392320534224e-06, "loss": 0.3745, "step": 34630 }, { "epoch": 1.73, "learning_rate": 4.233722871452421e-06, "loss": 0.3312, "step": 34640 }, { "epoch": 1.73, "learning_rate": 4.232053422370618e-06, "loss": 0.3565, "step": 34650 }, { "epoch": 1.73, "learning_rate": 4.230383973288815e-06, "loss": 0.3947, "step": 34660 }, { "epoch": 1.73, "learning_rate": 4.228714524207012e-06, "loss": 0.412, "step": 34670 }, { "epoch": 1.73, "learning_rate": 4.227045075125209e-06, "loss": 0.449, "step": 34680 }, { "epoch": 1.73, "learning_rate": 4.225375626043406e-06, "loss": 0.4397, "step": 34690 }, { "epoch": 1.73, "learning_rate": 4.223706176961603e-06, "loss": 0.4017, "step": 34700 }, { "epoch": 1.74, "learning_rate": 4.2220367278798e-06, "loss": 0.3675, "step": 34710 }, { "epoch": 1.74, "learning_rate": 4.220367278797997e-06, "loss": 0.3678, "step": 34720 }, { "epoch": 1.74, "learning_rate": 4.218697829716194e-06, "loss": 0.2918, "step": 34730 }, { "epoch": 1.74, "learning_rate": 4.217028380634391e-06, "loss": 0.3796, "step": 34740 }, { "epoch": 1.74, "learning_rate": 4.215358931552588e-06, "loss": 0.3967, "step": 34750 }, { "epoch": 1.74, "learning_rate": 4.213689482470785e-06, "loss": 0.3384, "step": 34760 }, { "epoch": 1.74, "learning_rate": 4.212020033388982e-06, "loss": 0.4066, "step": 34770 }, { "epoch": 1.74, "learning_rate": 4.210350584307179e-06, "loss": 0.4492, "step": 34780 }, { "epoch": 1.74, "learning_rate": 4.208681135225376e-06, "loss": 0.3883, "step": 34790 }, { "epoch": 1.74, "learning_rate": 4.207011686143573e-06, "loss": 0.3915, "step": 34800 }, { "epoch": 1.74, "learning_rate": 4.20534223706177e-06, "loss": 0.3818, "step": 34810 }, { "epoch": 1.74, "learning_rate": 4.203672787979967e-06, "loss": 0.3762, "step": 34820 }, { "epoch": 1.74, "learning_rate": 4.202003338898164e-06, "loss": 0.3813, "step": 34830 }, { "epoch": 1.74, "learning_rate": 4.200333889816361e-06, "loss": 0.3306, "step": 34840 }, { "epoch": 1.74, "learning_rate": 4.198664440734558e-06, "loss": 0.3155, "step": 34850 }, { "epoch": 1.74, "learning_rate": 4.196994991652755e-06, "loss": 0.2983, "step": 34860 }, { "epoch": 1.74, "learning_rate": 4.195325542570952e-06, "loss": 0.3321, "step": 34870 }, { "epoch": 1.74, "learning_rate": 4.193656093489149e-06, "loss": 0.4031, "step": 34880 }, { "epoch": 1.74, "learning_rate": 4.191986644407346e-06, "loss": 0.3732, "step": 34890 }, { "epoch": 1.75, "learning_rate": 4.190317195325543e-06, "loss": 0.3417, "step": 34900 }, { "epoch": 1.75, "learning_rate": 4.18864774624374e-06, "loss": 0.3599, "step": 34910 }, { "epoch": 1.75, "learning_rate": 4.186978297161937e-06, "loss": 0.3099, "step": 34920 }, { "epoch": 1.75, "learning_rate": 4.185308848080134e-06, "loss": 0.4387, "step": 34930 }, { "epoch": 1.75, "learning_rate": 4.183639398998331e-06, "loss": 0.3813, "step": 34940 }, { "epoch": 1.75, "learning_rate": 4.181969949916528e-06, "loss": 0.3262, "step": 34950 }, { "epoch": 1.75, "learning_rate": 4.180300500834725e-06, "loss": 0.4274, "step": 34960 }, { "epoch": 1.75, "learning_rate": 4.178631051752922e-06, "loss": 0.3602, "step": 34970 }, { "epoch": 1.75, "learning_rate": 4.176961602671119e-06, "loss": 0.3238, "step": 34980 }, { "epoch": 1.75, "learning_rate": 4.175292153589316e-06, "loss": 0.3235, "step": 34990 }, { "epoch": 1.75, "learning_rate": 4.173622704507513e-06, "loss": 0.3663, "step": 35000 }, { "epoch": 1.75, "eval_loss": 0.37505972385406494, "eval_runtime": 182.3336, "eval_samples_per_second": 20.649, "eval_steps_per_second": 2.583, "step": 35000 }, { "epoch": 1.75, "learning_rate": 4.17195325542571e-06, "loss": 0.3619, "step": 35010 }, { "epoch": 1.75, "learning_rate": 4.170283806343907e-06, "loss": 0.34, "step": 35020 }, { "epoch": 1.75, "learning_rate": 4.168614357262104e-06, "loss": 0.3868, "step": 35030 }, { "epoch": 1.75, "learning_rate": 4.166944908180301e-06, "loss": 0.3298, "step": 35040 }, { "epoch": 1.75, "learning_rate": 4.165275459098498e-06, "loss": 0.4197, "step": 35050 }, { "epoch": 1.75, "learning_rate": 4.163606010016695e-06, "loss": 0.3596, "step": 35060 }, { "epoch": 1.75, "learning_rate": 4.161936560934892e-06, "loss": 0.421, "step": 35070 }, { "epoch": 1.75, "learning_rate": 4.160267111853089e-06, "loss": 0.4343, "step": 35080 }, { "epoch": 1.75, "learning_rate": 4.158597662771286e-06, "loss": 0.3903, "step": 35090 }, { "epoch": 1.75, "learning_rate": 4.156928213689483e-06, "loss": 0.4177, "step": 35100 }, { "epoch": 1.76, "learning_rate": 4.15525876460768e-06, "loss": 0.2953, "step": 35110 }, { "epoch": 1.76, "learning_rate": 4.153589315525877e-06, "loss": 0.2987, "step": 35120 }, { "epoch": 1.76, "learning_rate": 4.151919866444074e-06, "loss": 0.3538, "step": 35130 }, { "epoch": 1.76, "learning_rate": 4.150250417362271e-06, "loss": 0.3915, "step": 35140 }, { "epoch": 1.76, "learning_rate": 4.148580968280468e-06, "loss": 0.3114, "step": 35150 }, { "epoch": 1.76, "learning_rate": 4.146911519198665e-06, "loss": 0.4029, "step": 35160 }, { "epoch": 1.76, "learning_rate": 4.145242070116862e-06, "loss": 0.4636, "step": 35170 }, { "epoch": 1.76, "learning_rate": 4.143572621035059e-06, "loss": 0.3864, "step": 35180 }, { "epoch": 1.76, "learning_rate": 4.141903171953256e-06, "loss": 0.4022, "step": 35190 }, { "epoch": 1.76, "learning_rate": 4.140233722871453e-06, "loss": 0.4293, "step": 35200 }, { "epoch": 1.76, "learning_rate": 4.13856427378965e-06, "loss": 0.4731, "step": 35210 }, { "epoch": 1.76, "learning_rate": 4.136894824707847e-06, "loss": 0.3915, "step": 35220 }, { "epoch": 1.76, "learning_rate": 4.135225375626044e-06, "loss": 0.4427, "step": 35230 }, { "epoch": 1.76, "learning_rate": 4.133555926544241e-06, "loss": 0.3445, "step": 35240 }, { "epoch": 1.76, "learning_rate": 4.131886477462438e-06, "loss": 0.4191, "step": 35250 }, { "epoch": 1.76, "learning_rate": 4.130217028380635e-06, "loss": 0.3316, "step": 35260 }, { "epoch": 1.76, "learning_rate": 4.128547579298832e-06, "loss": 0.3732, "step": 35270 }, { "epoch": 1.76, "learning_rate": 4.126878130217029e-06, "loss": 0.3113, "step": 35280 }, { "epoch": 1.76, "learning_rate": 4.125208681135226e-06, "loss": 0.3908, "step": 35290 }, { "epoch": 1.77, "learning_rate": 4.123539232053423e-06, "loss": 0.3104, "step": 35300 }, { "epoch": 1.77, "learning_rate": 4.12186978297162e-06, "loss": 0.4523, "step": 35310 }, { "epoch": 1.77, "learning_rate": 4.120200333889817e-06, "loss": 0.2942, "step": 35320 }, { "epoch": 1.77, "learning_rate": 4.118530884808014e-06, "loss": 0.3446, "step": 35330 }, { "epoch": 1.77, "learning_rate": 4.116861435726211e-06, "loss": 0.4261, "step": 35340 }, { "epoch": 1.77, "learning_rate": 4.115191986644408e-06, "loss": 0.3257, "step": 35350 }, { "epoch": 1.77, "learning_rate": 4.113522537562605e-06, "loss": 0.3578, "step": 35360 }, { "epoch": 1.77, "learning_rate": 4.111853088480802e-06, "loss": 0.2875, "step": 35370 }, { "epoch": 1.77, "learning_rate": 4.110183639398999e-06, "loss": 0.3754, "step": 35380 }, { "epoch": 1.77, "learning_rate": 4.108514190317196e-06, "loss": 0.3294, "step": 35390 }, { "epoch": 1.77, "learning_rate": 4.106844741235393e-06, "loss": 0.4171, "step": 35400 }, { "epoch": 1.77, "learning_rate": 4.10517529215359e-06, "loss": 0.3447, "step": 35410 }, { "epoch": 1.77, "learning_rate": 4.103505843071787e-06, "loss": 0.4005, "step": 35420 }, { "epoch": 1.77, "learning_rate": 4.101836393989984e-06, "loss": 0.3558, "step": 35430 }, { "epoch": 1.77, "learning_rate": 4.100166944908181e-06, "loss": 0.3044, "step": 35440 }, { "epoch": 1.77, "learning_rate": 4.098497495826378e-06, "loss": 0.3065, "step": 35450 }, { "epoch": 1.77, "learning_rate": 4.096828046744575e-06, "loss": 0.3289, "step": 35460 }, { "epoch": 1.77, "learning_rate": 4.095158597662772e-06, "loss": 0.3401, "step": 35470 }, { "epoch": 1.77, "learning_rate": 4.093489148580969e-06, "loss": 0.3696, "step": 35480 }, { "epoch": 1.77, "learning_rate": 4.091819699499166e-06, "loss": 0.3834, "step": 35490 }, { "epoch": 1.77, "learning_rate": 4.090150250417363e-06, "loss": 0.419, "step": 35500 }, { "epoch": 1.78, "learning_rate": 4.08848080133556e-06, "loss": 0.3545, "step": 35510 }, { "epoch": 1.78, "learning_rate": 4.086811352253757e-06, "loss": 0.3575, "step": 35520 }, { "epoch": 1.78, "learning_rate": 4.085141903171954e-06, "loss": 0.3826, "step": 35530 }, { "epoch": 1.78, "learning_rate": 4.083472454090151e-06, "loss": 0.3403, "step": 35540 }, { "epoch": 1.78, "learning_rate": 4.081803005008348e-06, "loss": 0.351, "step": 35550 }, { "epoch": 1.78, "learning_rate": 4.080133555926545e-06, "loss": 0.501, "step": 35560 }, { "epoch": 1.78, "learning_rate": 4.078464106844742e-06, "loss": 0.3788, "step": 35570 }, { "epoch": 1.78, "learning_rate": 4.076794657762939e-06, "loss": 0.4006, "step": 35580 }, { "epoch": 1.78, "learning_rate": 4.075125208681136e-06, "loss": 0.3615, "step": 35590 }, { "epoch": 1.78, "learning_rate": 4.073455759599333e-06, "loss": 0.3986, "step": 35600 }, { "epoch": 1.78, "learning_rate": 4.07178631051753e-06, "loss": 0.4071, "step": 35610 }, { "epoch": 1.78, "learning_rate": 4.070116861435727e-06, "loss": 0.3892, "step": 35620 }, { "epoch": 1.78, "learning_rate": 4.068447412353924e-06, "loss": 0.3226, "step": 35630 }, { "epoch": 1.78, "learning_rate": 4.066777963272121e-06, "loss": 0.3731, "step": 35640 }, { "epoch": 1.78, "learning_rate": 4.065108514190318e-06, "loss": 0.3704, "step": 35650 }, { "epoch": 1.78, "learning_rate": 4.063439065108514e-06, "loss": 0.3977, "step": 35660 }, { "epoch": 1.78, "learning_rate": 4.061769616026712e-06, "loss": 0.3866, "step": 35670 }, { "epoch": 1.78, "learning_rate": 4.060100166944909e-06, "loss": 0.3657, "step": 35680 }, { "epoch": 1.78, "learning_rate": 4.058430717863106e-06, "loss": 0.3125, "step": 35690 }, { "epoch": 1.79, "learning_rate": 4.056761268781303e-06, "loss": 0.5157, "step": 35700 }, { "epoch": 1.79, "learning_rate": 4.055091819699499e-06, "loss": 0.408, "step": 35710 }, { "epoch": 1.79, "learning_rate": 4.053422370617697e-06, "loss": 0.4055, "step": 35720 }, { "epoch": 1.79, "learning_rate": 4.051752921535894e-06, "loss": 0.3473, "step": 35730 }, { "epoch": 1.79, "learning_rate": 4.050083472454091e-06, "loss": 0.3557, "step": 35740 }, { "epoch": 1.79, "learning_rate": 4.048414023372288e-06, "loss": 0.3534, "step": 35750 }, { "epoch": 1.79, "learning_rate": 4.046744574290484e-06, "loss": 0.4585, "step": 35760 }, { "epoch": 1.79, "learning_rate": 4.045075125208682e-06, "loss": 0.3662, "step": 35770 }, { "epoch": 1.79, "learning_rate": 4.043405676126879e-06, "loss": 0.3679, "step": 35780 }, { "epoch": 1.79, "learning_rate": 4.041736227045076e-06, "loss": 0.4053, "step": 35790 }, { "epoch": 1.79, "learning_rate": 4.040066777963273e-06, "loss": 0.3684, "step": 35800 }, { "epoch": 1.79, "learning_rate": 4.03839732888147e-06, "loss": 0.3889, "step": 35810 }, { "epoch": 1.79, "learning_rate": 4.036727879799666e-06, "loss": 0.3166, "step": 35820 }, { "epoch": 1.79, "learning_rate": 4.035058430717864e-06, "loss": 0.3837, "step": 35830 }, { "epoch": 1.79, "learning_rate": 4.033388981636061e-06, "loss": 0.3757, "step": 35840 }, { "epoch": 1.79, "learning_rate": 4.031719532554258e-06, "loss": 0.4488, "step": 35850 }, { "epoch": 1.79, "learning_rate": 4.030050083472455e-06, "loss": 0.3718, "step": 35860 }, { "epoch": 1.79, "learning_rate": 4.028380634390651e-06, "loss": 0.3383, "step": 35870 }, { "epoch": 1.79, "learning_rate": 4.026711185308849e-06, "loss": 0.289, "step": 35880 }, { "epoch": 1.79, "learning_rate": 4.025041736227046e-06, "loss": 0.3579, "step": 35890 }, { "epoch": 1.79, "learning_rate": 4.023372287145243e-06, "loss": 0.3195, "step": 35900 }, { "epoch": 1.8, "learning_rate": 4.02170283806344e-06, "loss": 0.3498, "step": 35910 }, { "epoch": 1.8, "learning_rate": 4.020033388981636e-06, "loss": 0.4117, "step": 35920 }, { "epoch": 1.8, "learning_rate": 4.018363939899834e-06, "loss": 0.3455, "step": 35930 }, { "epoch": 1.8, "learning_rate": 4.016694490818031e-06, "loss": 0.3611, "step": 35940 }, { "epoch": 1.8, "learning_rate": 4.015025041736228e-06, "loss": 0.4239, "step": 35950 }, { "epoch": 1.8, "learning_rate": 4.013355592654425e-06, "loss": 0.3882, "step": 35960 }, { "epoch": 1.8, "learning_rate": 4.011686143572621e-06, "loss": 0.3813, "step": 35970 }, { "epoch": 1.8, "learning_rate": 4.010016694490818e-06, "loss": 0.5177, "step": 35980 }, { "epoch": 1.8, "learning_rate": 4.008347245409016e-06, "loss": 0.5381, "step": 35990 }, { "epoch": 1.8, "learning_rate": 4.006677796327213e-06, "loss": 0.3275, "step": 36000 }, { "epoch": 1.8, "learning_rate": 4.00500834724541e-06, "loss": 0.3247, "step": 36010 }, { "epoch": 1.8, "learning_rate": 4.003338898163606e-06, "loss": 0.3773, "step": 36020 }, { "epoch": 1.8, "learning_rate": 4.001669449081803e-06, "loss": 0.3411, "step": 36030 }, { "epoch": 1.8, "learning_rate": 4.000000000000001e-06, "loss": 0.3201, "step": 36040 }, { "epoch": 1.8, "learning_rate": 3.998330550918198e-06, "loss": 0.3023, "step": 36050 }, { "epoch": 1.8, "learning_rate": 3.996661101836395e-06, "loss": 0.4101, "step": 36060 }, { "epoch": 1.8, "learning_rate": 3.994991652754591e-06, "loss": 0.3167, "step": 36070 }, { "epoch": 1.8, "learning_rate": 3.993322203672788e-06, "loss": 0.4293, "step": 36080 }, { "epoch": 1.8, "learning_rate": 3.991652754590986e-06, "loss": 0.3214, "step": 36090 }, { "epoch": 1.81, "learning_rate": 3.989983305509183e-06, "loss": 0.3984, "step": 36100 }, { "epoch": 1.81, "learning_rate": 3.98831385642738e-06, "loss": 0.4147, "step": 36110 }, { "epoch": 1.81, "learning_rate": 3.986644407345576e-06, "loss": 0.4097, "step": 36120 }, { "epoch": 1.81, "learning_rate": 3.984974958263773e-06, "loss": 0.3928, "step": 36130 }, { "epoch": 1.81, "learning_rate": 3.98330550918197e-06, "loss": 0.3816, "step": 36140 }, { "epoch": 1.81, "learning_rate": 3.981636060100168e-06, "loss": 0.3717, "step": 36150 }, { "epoch": 1.81, "learning_rate": 3.979966611018365e-06, "loss": 0.4532, "step": 36160 }, { "epoch": 1.81, "learning_rate": 3.978297161936562e-06, "loss": 0.3703, "step": 36170 }, { "epoch": 1.81, "learning_rate": 3.976627712854758e-06, "loss": 0.3902, "step": 36180 }, { "epoch": 1.81, "learning_rate": 3.974958263772955e-06, "loss": 0.3286, "step": 36190 }, { "epoch": 1.81, "learning_rate": 3.973288814691153e-06, "loss": 0.3729, "step": 36200 }, { "epoch": 1.81, "learning_rate": 3.97161936560935e-06, "loss": 0.3475, "step": 36210 }, { "epoch": 1.81, "learning_rate": 3.969949916527547e-06, "loss": 0.3759, "step": 36220 }, { "epoch": 1.81, "learning_rate": 3.968280467445743e-06, "loss": 0.3502, "step": 36230 }, { "epoch": 1.81, "learning_rate": 3.96661101836394e-06, "loss": 0.3252, "step": 36240 }, { "epoch": 1.81, "learning_rate": 3.964941569282138e-06, "loss": 0.4146, "step": 36250 }, { "epoch": 1.81, "learning_rate": 3.963272120200335e-06, "loss": 0.3432, "step": 36260 }, { "epoch": 1.81, "learning_rate": 3.961602671118532e-06, "loss": 0.2754, "step": 36270 }, { "epoch": 1.81, "learning_rate": 3.959933222036728e-06, "loss": 0.2985, "step": 36280 }, { "epoch": 1.81, "learning_rate": 3.958263772954925e-06, "loss": 0.5214, "step": 36290 }, { "epoch": 1.81, "learning_rate": 3.956594323873122e-06, "loss": 0.3082, "step": 36300 }, { "epoch": 1.82, "learning_rate": 3.95492487479132e-06, "loss": 0.3521, "step": 36310 }, { "epoch": 1.82, "learning_rate": 3.953255425709517e-06, "loss": 0.3176, "step": 36320 }, { "epoch": 1.82, "learning_rate": 3.951585976627713e-06, "loss": 0.4268, "step": 36330 }, { "epoch": 1.82, "learning_rate": 3.94991652754591e-06, "loss": 0.4277, "step": 36340 }, { "epoch": 1.82, "learning_rate": 3.948247078464107e-06, "loss": 0.3588, "step": 36350 }, { "epoch": 1.82, "learning_rate": 3.946577629382305e-06, "loss": 0.3374, "step": 36360 }, { "epoch": 1.82, "learning_rate": 3.944908180300502e-06, "loss": 0.3031, "step": 36370 }, { "epoch": 1.82, "learning_rate": 3.943238731218698e-06, "loss": 0.3336, "step": 36380 }, { "epoch": 1.82, "learning_rate": 3.941569282136895e-06, "loss": 0.3801, "step": 36390 }, { "epoch": 1.82, "learning_rate": 3.939899833055092e-06, "loss": 0.3389, "step": 36400 }, { "epoch": 1.82, "learning_rate": 3.93823038397329e-06, "loss": 0.3544, "step": 36410 }, { "epoch": 1.82, "learning_rate": 3.936560934891487e-06, "loss": 0.3957, "step": 36420 }, { "epoch": 1.82, "learning_rate": 3.934891485809683e-06, "loss": 0.443, "step": 36430 }, { "epoch": 1.82, "learning_rate": 3.93322203672788e-06, "loss": 0.3656, "step": 36440 }, { "epoch": 1.82, "learning_rate": 3.931552587646077e-06, "loss": 0.4312, "step": 36450 }, { "epoch": 1.82, "learning_rate": 3.929883138564274e-06, "loss": 0.3897, "step": 36460 }, { "epoch": 1.82, "learning_rate": 3.928213689482472e-06, "loss": 0.3749, "step": 36470 }, { "epoch": 1.82, "learning_rate": 3.926544240400669e-06, "loss": 0.3814, "step": 36480 }, { "epoch": 1.82, "learning_rate": 3.924874791318865e-06, "loss": 0.3949, "step": 36490 }, { "epoch": 1.82, "learning_rate": 3.923205342237062e-06, "loss": 0.3056, "step": 36500 }, { "epoch": 1.83, "learning_rate": 3.921535893155259e-06, "loss": 0.3316, "step": 36510 }, { "epoch": 1.83, "learning_rate": 3.919866444073457e-06, "loss": 0.3073, "step": 36520 }, { "epoch": 1.83, "learning_rate": 3.918196994991654e-06, "loss": 0.4744, "step": 36530 }, { "epoch": 1.83, "learning_rate": 3.91652754590985e-06, "loss": 0.3589, "step": 36540 }, { "epoch": 1.83, "learning_rate": 3.914858096828047e-06, "loss": 0.3935, "step": 36550 }, { "epoch": 1.83, "learning_rate": 3.913188647746244e-06, "loss": 0.3657, "step": 36560 }, { "epoch": 1.83, "learning_rate": 3.9115191986644416e-06, "loss": 0.3668, "step": 36570 }, { "epoch": 1.83, "learning_rate": 3.9098497495826386e-06, "loss": 0.3821, "step": 36580 }, { "epoch": 1.83, "learning_rate": 3.908180300500835e-06, "loss": 0.3374, "step": 36590 }, { "epoch": 1.83, "learning_rate": 3.906510851419032e-06, "loss": 0.34, "step": 36600 }, { "epoch": 1.83, "learning_rate": 3.904841402337229e-06, "loss": 0.3798, "step": 36610 }, { "epoch": 1.83, "learning_rate": 3.9031719532554266e-06, "loss": 0.2999, "step": 36620 }, { "epoch": 1.83, "learning_rate": 3.9015025041736236e-06, "loss": 0.3386, "step": 36630 }, { "epoch": 1.83, "learning_rate": 3.89983305509182e-06, "loss": 0.4635, "step": 36640 }, { "epoch": 1.83, "learning_rate": 3.898163606010017e-06, "loss": 0.4621, "step": 36650 }, { "epoch": 1.83, "learning_rate": 3.896494156928214e-06, "loss": 0.3188, "step": 36660 }, { "epoch": 1.83, "learning_rate": 3.894824707846411e-06, "loss": 0.341, "step": 36670 }, { "epoch": 1.83, "learning_rate": 3.8931552587646086e-06, "loss": 0.2977, "step": 36680 }, { "epoch": 1.83, "learning_rate": 3.891485809682805e-06, "loss": 0.4226, "step": 36690 }, { "epoch": 1.83, "learning_rate": 3.889816360601002e-06, "loss": 0.3804, "step": 36700 }, { "epoch": 1.84, "learning_rate": 3.888146911519199e-06, "loss": 0.3944, "step": 36710 }, { "epoch": 1.84, "learning_rate": 3.886477462437396e-06, "loss": 0.4469, "step": 36720 }, { "epoch": 1.84, "learning_rate": 3.8848080133555935e-06, "loss": 0.5198, "step": 36730 }, { "epoch": 1.84, "learning_rate": 3.88313856427379e-06, "loss": 0.3302, "step": 36740 }, { "epoch": 1.84, "learning_rate": 3.881469115191987e-06, "loss": 0.4573, "step": 36750 }, { "epoch": 1.84, "learning_rate": 3.879799666110184e-06, "loss": 0.3473, "step": 36760 }, { "epoch": 1.84, "learning_rate": 3.878130217028381e-06, "loss": 0.5669, "step": 36770 }, { "epoch": 1.84, "learning_rate": 3.8764607679465785e-06, "loss": 0.3586, "step": 36780 }, { "epoch": 1.84, "learning_rate": 3.874791318864775e-06, "loss": 0.3442, "step": 36790 }, { "epoch": 1.84, "learning_rate": 3.873121869782972e-06, "loss": 0.2749, "step": 36800 }, { "epoch": 1.84, "learning_rate": 3.871452420701169e-06, "loss": 0.3087, "step": 36810 }, { "epoch": 1.84, "learning_rate": 3.869782971619366e-06, "loss": 0.379, "step": 36820 }, { "epoch": 1.84, "learning_rate": 3.868113522537563e-06, "loss": 0.4682, "step": 36830 }, { "epoch": 1.84, "learning_rate": 3.8664440734557605e-06, "loss": 0.35, "step": 36840 }, { "epoch": 1.84, "learning_rate": 3.864774624373957e-06, "loss": 0.4147, "step": 36850 }, { "epoch": 1.84, "learning_rate": 3.863105175292154e-06, "loss": 0.3293, "step": 36860 }, { "epoch": 1.84, "learning_rate": 3.861435726210351e-06, "loss": 0.4449, "step": 36870 }, { "epoch": 1.84, "learning_rate": 3.859766277128548e-06, "loss": 0.5222, "step": 36880 }, { "epoch": 1.84, "learning_rate": 3.8580968280467455e-06, "loss": 0.3675, "step": 36890 }, { "epoch": 1.84, "learning_rate": 3.856427378964942e-06, "loss": 0.3085, "step": 36900 }, { "epoch": 1.85, "learning_rate": 3.854757929883139e-06, "loss": 0.3317, "step": 36910 }, { "epoch": 1.85, "learning_rate": 3.853088480801336e-06, "loss": 0.3204, "step": 36920 }, { "epoch": 1.85, "learning_rate": 3.851419031719533e-06, "loss": 0.4312, "step": 36930 }, { "epoch": 1.85, "learning_rate": 3.8497495826377305e-06, "loss": 0.3456, "step": 36940 }, { "epoch": 1.85, "learning_rate": 3.848080133555927e-06, "loss": 0.339, "step": 36950 }, { "epoch": 1.85, "learning_rate": 3.846410684474124e-06, "loss": 0.3718, "step": 36960 }, { "epoch": 1.85, "learning_rate": 3.844741235392321e-06, "loss": 0.4179, "step": 36970 }, { "epoch": 1.85, "learning_rate": 3.843071786310518e-06, "loss": 0.3993, "step": 36980 }, { "epoch": 1.85, "learning_rate": 3.841402337228715e-06, "loss": 0.4254, "step": 36990 }, { "epoch": 1.85, "learning_rate": 3.839732888146912e-06, "loss": 0.485, "step": 37000 }, { "epoch": 1.85, "learning_rate": 3.838063439065109e-06, "loss": 0.3593, "step": 37010 }, { "epoch": 1.85, "learning_rate": 3.836393989983306e-06, "loss": 0.4009, "step": 37020 }, { "epoch": 1.85, "learning_rate": 3.834724540901503e-06, "loss": 0.4193, "step": 37030 }, { "epoch": 1.85, "learning_rate": 3.8330550918197e-06, "loss": 0.3388, "step": 37040 }, { "epoch": 1.85, "learning_rate": 3.831385642737897e-06, "loss": 0.4362, "step": 37050 }, { "epoch": 1.85, "learning_rate": 3.829716193656094e-06, "loss": 0.2854, "step": 37060 }, { "epoch": 1.85, "learning_rate": 3.828046744574291e-06, "loss": 0.3754, "step": 37070 }, { "epoch": 1.85, "learning_rate": 3.826377295492488e-06, "loss": 0.3929, "step": 37080 }, { "epoch": 1.85, "learning_rate": 3.824707846410685e-06, "loss": 0.3194, "step": 37090 }, { "epoch": 1.85, "learning_rate": 3.823038397328882e-06, "loss": 0.5165, "step": 37100 }, { "epoch": 1.86, "learning_rate": 3.821368948247079e-06, "loss": 0.5035, "step": 37110 }, { "epoch": 1.86, "learning_rate": 3.819699499165276e-06, "loss": 0.3192, "step": 37120 }, { "epoch": 1.86, "learning_rate": 3.818030050083473e-06, "loss": 0.3099, "step": 37130 }, { "epoch": 1.86, "learning_rate": 3.81636060100167e-06, "loss": 0.4256, "step": 37140 }, { "epoch": 1.86, "learning_rate": 3.8146911519198666e-06, "loss": 0.3838, "step": 37150 }, { "epoch": 1.86, "learning_rate": 3.813021702838064e-06, "loss": 0.3825, "step": 37160 }, { "epoch": 1.86, "learning_rate": 3.811352253756261e-06, "loss": 0.3518, "step": 37170 }, { "epoch": 1.86, "learning_rate": 3.8096828046744576e-06, "loss": 0.4095, "step": 37180 }, { "epoch": 1.86, "learning_rate": 3.8080133555926546e-06, "loss": 0.3684, "step": 37190 }, { "epoch": 1.86, "learning_rate": 3.8063439065108516e-06, "loss": 0.459, "step": 37200 }, { "epoch": 1.86, "learning_rate": 3.804674457429049e-06, "loss": 0.2733, "step": 37210 }, { "epoch": 1.86, "learning_rate": 3.803005008347246e-06, "loss": 0.3319, "step": 37220 }, { "epoch": 1.86, "learning_rate": 3.8013355592654426e-06, "loss": 0.3877, "step": 37230 }, { "epoch": 1.86, "learning_rate": 3.7996661101836396e-06, "loss": 0.3363, "step": 37240 }, { "epoch": 1.86, "learning_rate": 3.7979966611018366e-06, "loss": 0.4519, "step": 37250 }, { "epoch": 1.86, "learning_rate": 3.796327212020034e-06, "loss": 0.3648, "step": 37260 }, { "epoch": 1.86, "learning_rate": 3.794657762938231e-06, "loss": 0.4381, "step": 37270 }, { "epoch": 1.86, "learning_rate": 3.7929883138564276e-06, "loss": 0.366, "step": 37280 }, { "epoch": 1.86, "learning_rate": 3.7913188647746246e-06, "loss": 0.4123, "step": 37290 }, { "epoch": 1.86, "learning_rate": 3.7896494156928216e-06, "loss": 0.3888, "step": 37300 }, { "epoch": 1.87, "learning_rate": 3.7879799666110186e-06, "loss": 0.3971, "step": 37310 }, { "epoch": 1.87, "learning_rate": 3.786310517529216e-06, "loss": 0.3063, "step": 37320 }, { "epoch": 1.87, "learning_rate": 3.7846410684474126e-06, "loss": 0.3504, "step": 37330 }, { "epoch": 1.87, "learning_rate": 3.7829716193656096e-06, "loss": 0.3243, "step": 37340 }, { "epoch": 1.87, "learning_rate": 3.7813021702838066e-06, "loss": 0.421, "step": 37350 }, { "epoch": 1.87, "learning_rate": 3.7796327212020036e-06, "loss": 0.3661, "step": 37360 }, { "epoch": 1.87, "learning_rate": 3.777963272120201e-06, "loss": 0.3188, "step": 37370 }, { "epoch": 1.87, "learning_rate": 3.7762938230383976e-06, "loss": 0.3524, "step": 37380 }, { "epoch": 1.87, "learning_rate": 3.7746243739565946e-06, "loss": 0.3362, "step": 37390 }, { "epoch": 1.87, "learning_rate": 3.7729549248747916e-06, "loss": 0.4483, "step": 37400 }, { "epoch": 1.87, "learning_rate": 3.7712854757929886e-06, "loss": 0.3995, "step": 37410 }, { "epoch": 1.87, "learning_rate": 3.769616026711186e-06, "loss": 0.3611, "step": 37420 }, { "epoch": 1.87, "learning_rate": 3.7679465776293826e-06, "loss": 0.3111, "step": 37430 }, { "epoch": 1.87, "learning_rate": 3.7662771285475796e-06, "loss": 0.4122, "step": 37440 }, { "epoch": 1.87, "learning_rate": 3.7646076794657766e-06, "loss": 0.3645, "step": 37450 }, { "epoch": 1.87, "learning_rate": 3.7629382303839736e-06, "loss": 0.4244, "step": 37460 }, { "epoch": 1.87, "learning_rate": 3.76126878130217e-06, "loss": 0.3228, "step": 37470 }, { "epoch": 1.87, "learning_rate": 3.7595993322203676e-06, "loss": 0.3514, "step": 37480 }, { "epoch": 1.87, "learning_rate": 3.7579298831385646e-06, "loss": 0.2974, "step": 37490 }, { "epoch": 1.88, "learning_rate": 3.7562604340567616e-06, "loss": 0.2992, "step": 37500 }, { "epoch": 1.88, "learning_rate": 3.7545909849749586e-06, "loss": 0.3226, "step": 37510 }, { "epoch": 1.88, "learning_rate": 3.752921535893155e-06, "loss": 0.4208, "step": 37520 }, { "epoch": 1.88, "learning_rate": 3.751252086811353e-06, "loss": 0.3912, "step": 37530 }, { "epoch": 1.88, "learning_rate": 3.7495826377295496e-06, "loss": 0.4379, "step": 37540 }, { "epoch": 1.88, "learning_rate": 3.7479131886477466e-06, "loss": 0.4228, "step": 37550 }, { "epoch": 1.88, "learning_rate": 3.7462437395659436e-06, "loss": 0.2813, "step": 37560 }, { "epoch": 1.88, "learning_rate": 3.7445742904841406e-06, "loss": 0.3778, "step": 37570 }, { "epoch": 1.88, "learning_rate": 3.742904841402338e-06, "loss": 0.3975, "step": 37580 }, { "epoch": 1.88, "learning_rate": 3.7412353923205345e-06, "loss": 0.3093, "step": 37590 }, { "epoch": 1.88, "learning_rate": 3.7395659432387315e-06, "loss": 0.3821, "step": 37600 }, { "epoch": 1.88, "learning_rate": 3.7378964941569285e-06, "loss": 0.3873, "step": 37610 }, { "epoch": 1.88, "learning_rate": 3.7362270450751255e-06, "loss": 0.3705, "step": 37620 }, { "epoch": 1.88, "learning_rate": 3.734557595993322e-06, "loss": 0.3768, "step": 37630 }, { "epoch": 1.88, "learning_rate": 3.7328881469115195e-06, "loss": 0.3166, "step": 37640 }, { "epoch": 1.88, "learning_rate": 3.7312186978297165e-06, "loss": 0.4307, "step": 37650 }, { "epoch": 1.88, "learning_rate": 3.7295492487479135e-06, "loss": 0.4177, "step": 37660 }, { "epoch": 1.88, "learning_rate": 3.7278797996661105e-06, "loss": 0.313, "step": 37670 }, { "epoch": 1.88, "learning_rate": 3.726210350584307e-06, "loss": 0.3254, "step": 37680 }, { "epoch": 1.88, "learning_rate": 3.7245409015025045e-06, "loss": 0.3587, "step": 37690 }, { "epoch": 1.89, "learning_rate": 3.7228714524207015e-06, "loss": 0.3707, "step": 37700 }, { "epoch": 1.89, "learning_rate": 3.7212020033388985e-06, "loss": 0.326, "step": 37710 }, { "epoch": 1.89, "learning_rate": 3.7195325542570955e-06, "loss": 0.436, "step": 37720 }, { "epoch": 1.89, "learning_rate": 3.717863105175292e-06, "loss": 0.424, "step": 37730 }, { "epoch": 1.89, "learning_rate": 3.7161936560934895e-06, "loss": 0.3985, "step": 37740 }, { "epoch": 1.89, "learning_rate": 3.7145242070116865e-06, "loss": 0.3839, "step": 37750 }, { "epoch": 1.89, "learning_rate": 3.7128547579298835e-06, "loss": 0.347, "step": 37760 }, { "epoch": 1.89, "learning_rate": 3.7111853088480805e-06, "loss": 0.4355, "step": 37770 }, { "epoch": 1.89, "learning_rate": 3.709515859766277e-06, "loss": 0.3162, "step": 37780 }, { "epoch": 1.89, "learning_rate": 3.707846410684474e-06, "loss": 0.346, "step": 37790 }, { "epoch": 1.89, "learning_rate": 3.7061769616026715e-06, "loss": 0.4281, "step": 37800 }, { "epoch": 1.89, "learning_rate": 3.7045075125208685e-06, "loss": 0.3166, "step": 37810 }, { "epoch": 1.89, "learning_rate": 3.7028380634390655e-06, "loss": 0.3655, "step": 37820 }, { "epoch": 1.89, "learning_rate": 3.701168614357262e-06, "loss": 0.4123, "step": 37830 }, { "epoch": 1.89, "learning_rate": 3.699499165275459e-06, "loss": 0.4484, "step": 37840 }, { "epoch": 1.89, "learning_rate": 3.6978297161936565e-06, "loss": 0.3224, "step": 37850 }, { "epoch": 1.89, "learning_rate": 3.6961602671118535e-06, "loss": 0.5217, "step": 37860 }, { "epoch": 1.89, "learning_rate": 3.6944908180300505e-06, "loss": 0.3506, "step": 37870 }, { "epoch": 1.89, "learning_rate": 3.692821368948247e-06, "loss": 0.3631, "step": 37880 }, { "epoch": 1.89, "learning_rate": 3.691151919866444e-06, "loss": 0.391, "step": 37890 }, { "epoch": 1.9, "learning_rate": 3.6894824707846415e-06, "loss": 0.3587, "step": 37900 }, { "epoch": 1.9, "learning_rate": 3.6878130217028385e-06, "loss": 0.3862, "step": 37910 }, { "epoch": 1.9, "learning_rate": 3.6861435726210355e-06, "loss": 0.3458, "step": 37920 }, { "epoch": 1.9, "learning_rate": 3.6844741235392325e-06, "loss": 0.3304, "step": 37930 }, { "epoch": 1.9, "learning_rate": 3.682804674457429e-06, "loss": 0.4379, "step": 37940 }, { "epoch": 1.9, "learning_rate": 3.6811352253756265e-06, "loss": 0.3747, "step": 37950 }, { "epoch": 1.9, "learning_rate": 3.6794657762938235e-06, "loss": 0.2589, "step": 37960 }, { "epoch": 1.9, "learning_rate": 3.6777963272120205e-06, "loss": 0.3345, "step": 37970 }, { "epoch": 1.9, "learning_rate": 3.6761268781302175e-06, "loss": 0.4182, "step": 37980 }, { "epoch": 1.9, "learning_rate": 3.674457429048414e-06, "loss": 0.3233, "step": 37990 }, { "epoch": 1.9, "learning_rate": 3.672787979966611e-06, "loss": 0.3901, "step": 38000 }, { "epoch": 1.9, "learning_rate": 3.6711185308848085e-06, "loss": 0.3375, "step": 38010 }, { "epoch": 1.9, "learning_rate": 3.6694490818030055e-06, "loss": 0.3866, "step": 38020 }, { "epoch": 1.9, "learning_rate": 3.6677796327212025e-06, "loss": 0.3711, "step": 38030 }, { "epoch": 1.9, "learning_rate": 3.666110183639399e-06, "loss": 0.3666, "step": 38040 }, { "epoch": 1.9, "learning_rate": 3.664440734557596e-06, "loss": 0.4053, "step": 38050 }, { "epoch": 1.9, "learning_rate": 3.6627712854757935e-06, "loss": 0.3283, "step": 38060 }, { "epoch": 1.9, "learning_rate": 3.6611018363939905e-06, "loss": 0.3691, "step": 38070 }, { "epoch": 1.9, "learning_rate": 3.6594323873121875e-06, "loss": 0.3892, "step": 38080 }, { "epoch": 1.9, "learning_rate": 3.657762938230384e-06, "loss": 0.3357, "step": 38090 }, { "epoch": 1.91, "learning_rate": 3.656093489148581e-06, "loss": 0.2778, "step": 38100 }, { "epoch": 1.91, "learning_rate": 3.6544240400667785e-06, "loss": 0.3694, "step": 38110 }, { "epoch": 1.91, "learning_rate": 3.6527545909849755e-06, "loss": 0.4314, "step": 38120 }, { "epoch": 1.91, "learning_rate": 3.6510851419031725e-06, "loss": 0.3144, "step": 38130 }, { "epoch": 1.91, "learning_rate": 3.649415692821369e-06, "loss": 0.4517, "step": 38140 }, { "epoch": 1.91, "learning_rate": 3.647746243739566e-06, "loss": 0.3911, "step": 38150 }, { "epoch": 1.91, "learning_rate": 3.646076794657763e-06, "loss": 0.3704, "step": 38160 }, { "epoch": 1.91, "learning_rate": 3.6444073455759604e-06, "loss": 0.3908, "step": 38170 }, { "epoch": 1.91, "learning_rate": 3.6427378964941574e-06, "loss": 0.4285, "step": 38180 }, { "epoch": 1.91, "learning_rate": 3.641068447412354e-06, "loss": 0.3692, "step": 38190 }, { "epoch": 1.91, "learning_rate": 3.639398998330551e-06, "loss": 0.323, "step": 38200 }, { "epoch": 1.91, "learning_rate": 3.637729549248748e-06, "loss": 0.462, "step": 38210 }, { "epoch": 1.91, "learning_rate": 3.6360601001669454e-06, "loss": 0.3787, "step": 38220 }, { "epoch": 1.91, "learning_rate": 3.6343906510851424e-06, "loss": 0.3359, "step": 38230 }, { "epoch": 1.91, "learning_rate": 3.632721202003339e-06, "loss": 0.3956, "step": 38240 }, { "epoch": 1.91, "learning_rate": 3.631051752921536e-06, "loss": 0.4152, "step": 38250 }, { "epoch": 1.91, "learning_rate": 3.629382303839733e-06, "loss": 0.3767, "step": 38260 }, { "epoch": 1.91, "learning_rate": 3.6277128547579304e-06, "loss": 0.4341, "step": 38270 }, { "epoch": 1.91, "learning_rate": 3.6260434056761274e-06, "loss": 0.3857, "step": 38280 }, { "epoch": 1.91, "learning_rate": 3.6243739565943244e-06, "loss": 0.4127, "step": 38290 }, { "epoch": 1.92, "learning_rate": 3.622704507512521e-06, "loss": 0.3706, "step": 38300 }, { "epoch": 1.92, "learning_rate": 3.621035058430718e-06, "loss": 0.481, "step": 38310 }, { "epoch": 1.92, "learning_rate": 3.619365609348915e-06, "loss": 0.3446, "step": 38320 }, { "epoch": 1.92, "learning_rate": 3.6176961602671124e-06, "loss": 0.4616, "step": 38330 }, { "epoch": 1.92, "learning_rate": 3.6160267111853094e-06, "loss": 0.3896, "step": 38340 }, { "epoch": 1.92, "learning_rate": 3.614357262103506e-06, "loss": 0.4222, "step": 38350 }, { "epoch": 1.92, "learning_rate": 3.612687813021703e-06, "loss": 0.3357, "step": 38360 }, { "epoch": 1.92, "learning_rate": 3.6110183639399e-06, "loss": 0.3229, "step": 38370 }, { "epoch": 1.92, "learning_rate": 3.6093489148580974e-06, "loss": 0.3996, "step": 38380 }, { "epoch": 1.92, "learning_rate": 3.6076794657762944e-06, "loss": 0.3936, "step": 38390 }, { "epoch": 1.92, "learning_rate": 3.606010016694491e-06, "loss": 0.3727, "step": 38400 }, { "epoch": 1.92, "learning_rate": 3.604340567612688e-06, "loss": 0.387, "step": 38410 }, { "epoch": 1.92, "learning_rate": 3.602671118530885e-06, "loss": 0.3557, "step": 38420 }, { "epoch": 1.92, "learning_rate": 3.6010016694490824e-06, "loss": 0.4144, "step": 38430 }, { "epoch": 1.92, "learning_rate": 3.5993322203672794e-06, "loss": 0.4152, "step": 38440 }, { "epoch": 1.92, "learning_rate": 3.597662771285476e-06, "loss": 0.3301, "step": 38450 }, { "epoch": 1.92, "learning_rate": 3.595993322203673e-06, "loss": 0.3553, "step": 38460 }, { "epoch": 1.92, "learning_rate": 3.59432387312187e-06, "loss": 0.3681, "step": 38470 }, { "epoch": 1.92, "learning_rate": 3.592654424040067e-06, "loss": 0.4105, "step": 38480 }, { "epoch": 1.92, "learning_rate": 3.5909849749582644e-06, "loss": 0.3641, "step": 38490 }, { "epoch": 1.93, "learning_rate": 3.589315525876461e-06, "loss": 0.4038, "step": 38500 }, { "epoch": 1.93, "learning_rate": 3.587646076794658e-06, "loss": 0.4113, "step": 38510 }, { "epoch": 1.93, "learning_rate": 3.585976627712855e-06, "loss": 0.4323, "step": 38520 }, { "epoch": 1.93, "learning_rate": 3.584307178631052e-06, "loss": 0.3285, "step": 38530 }, { "epoch": 1.93, "learning_rate": 3.5826377295492494e-06, "loss": 0.3474, "step": 38540 }, { "epoch": 1.93, "learning_rate": 3.580968280467446e-06, "loss": 0.3494, "step": 38550 }, { "epoch": 1.93, "learning_rate": 3.579298831385643e-06, "loss": 0.3706, "step": 38560 }, { "epoch": 1.93, "learning_rate": 3.57762938230384e-06, "loss": 0.3266, "step": 38570 }, { "epoch": 1.93, "learning_rate": 3.575959933222037e-06, "loss": 0.3374, "step": 38580 }, { "epoch": 1.93, "learning_rate": 3.5742904841402344e-06, "loss": 0.4234, "step": 38590 }, { "epoch": 1.93, "learning_rate": 3.5726210350584314e-06, "loss": 0.3473, "step": 38600 }, { "epoch": 1.93, "learning_rate": 3.570951585976628e-06, "loss": 0.3232, "step": 38610 }, { "epoch": 1.93, "learning_rate": 3.569282136894825e-06, "loss": 0.4064, "step": 38620 }, { "epoch": 1.93, "learning_rate": 3.567612687813022e-06, "loss": 0.3921, "step": 38630 }, { "epoch": 1.93, "learning_rate": 3.5659432387312185e-06, "loss": 0.473, "step": 38640 }, { "epoch": 1.93, "learning_rate": 3.5642737896494164e-06, "loss": 0.4366, "step": 38650 }, { "epoch": 1.93, "learning_rate": 3.562604340567613e-06, "loss": 0.3551, "step": 38660 }, { "epoch": 1.93, "learning_rate": 3.56093489148581e-06, "loss": 0.4242, "step": 38670 }, { "epoch": 1.93, "learning_rate": 3.559265442404007e-06, "loss": 0.4149, "step": 38680 }, { "epoch": 1.93, "learning_rate": 3.557595993322204e-06, "loss": 0.3521, "step": 38690 }, { "epoch": 1.94, "learning_rate": 3.5559265442404014e-06, "loss": 0.4078, "step": 38700 }, { "epoch": 1.94, "learning_rate": 3.554257095158598e-06, "loss": 0.3172, "step": 38710 }, { "epoch": 1.94, "learning_rate": 3.552587646076795e-06, "loss": 0.4505, "step": 38720 }, { "epoch": 1.94, "learning_rate": 3.550918196994992e-06, "loss": 0.3724, "step": 38730 }, { "epoch": 1.94, "learning_rate": 3.549248747913189e-06, "loss": 0.3522, "step": 38740 }, { "epoch": 1.94, "learning_rate": 3.5475792988313863e-06, "loss": 0.3473, "step": 38750 }, { "epoch": 1.94, "learning_rate": 3.545909849749583e-06, "loss": 0.3383, "step": 38760 }, { "epoch": 1.94, "learning_rate": 3.54424040066778e-06, "loss": 0.4013, "step": 38770 }, { "epoch": 1.94, "learning_rate": 3.542570951585977e-06, "loss": 0.4339, "step": 38780 }, { "epoch": 1.94, "learning_rate": 3.540901502504174e-06, "loss": 0.3885, "step": 38790 }, { "epoch": 1.94, "learning_rate": 3.5392320534223705e-06, "loss": 0.3999, "step": 38800 }, { "epoch": 1.94, "learning_rate": 3.537562604340568e-06, "loss": 0.4099, "step": 38810 }, { "epoch": 1.94, "learning_rate": 3.535893155258765e-06, "loss": 0.4362, "step": 38820 }, { "epoch": 1.94, "learning_rate": 3.534223706176962e-06, "loss": 0.4304, "step": 38830 }, { "epoch": 1.94, "learning_rate": 3.532554257095159e-06, "loss": 0.317, "step": 38840 }, { "epoch": 1.94, "learning_rate": 3.5308848080133555e-06, "loss": 0.284, "step": 38850 }, { "epoch": 1.94, "learning_rate": 3.529215358931553e-06, "loss": 0.3363, "step": 38860 }, { "epoch": 1.94, "learning_rate": 3.52754590984975e-06, "loss": 0.3843, "step": 38870 }, { "epoch": 1.94, "learning_rate": 3.525876460767947e-06, "loss": 0.5343, "step": 38880 }, { "epoch": 1.94, "learning_rate": 3.524207011686144e-06, "loss": 0.3562, "step": 38890 }, { "epoch": 1.94, "learning_rate": 3.5225375626043405e-06, "loss": 0.3422, "step": 38900 }, { "epoch": 1.95, "learning_rate": 3.520868113522538e-06, "loss": 0.3352, "step": 38910 }, { "epoch": 1.95, "learning_rate": 3.519198664440735e-06, "loss": 0.3754, "step": 38920 }, { "epoch": 1.95, "learning_rate": 3.517529215358932e-06, "loss": 0.3362, "step": 38930 }, { "epoch": 1.95, "learning_rate": 3.515859766277129e-06, "loss": 0.3498, "step": 38940 }, { "epoch": 1.95, "learning_rate": 3.5141903171953255e-06, "loss": 0.3653, "step": 38950 }, { "epoch": 1.95, "learning_rate": 3.5125208681135225e-06, "loss": 0.3412, "step": 38960 }, { "epoch": 1.95, "learning_rate": 3.51085141903172e-06, "loss": 0.3627, "step": 38970 }, { "epoch": 1.95, "learning_rate": 3.509181969949917e-06, "loss": 0.3103, "step": 38980 }, { "epoch": 1.95, "learning_rate": 3.507512520868114e-06, "loss": 0.4054, "step": 38990 }, { "epoch": 1.95, "learning_rate": 3.505843071786311e-06, "loss": 0.3899, "step": 39000 }, { "epoch": 1.95, "learning_rate": 3.5041736227045075e-06, "loss": 0.3802, "step": 39010 }, { "epoch": 1.95, "learning_rate": 3.502504173622705e-06, "loss": 0.3094, "step": 39020 }, { "epoch": 1.95, "learning_rate": 3.500834724540902e-06, "loss": 0.3389, "step": 39030 }, { "epoch": 1.95, "learning_rate": 3.499165275459099e-06, "loss": 0.4133, "step": 39040 }, { "epoch": 1.95, "learning_rate": 3.497495826377296e-06, "loss": 0.3399, "step": 39050 }, { "epoch": 1.95, "learning_rate": 3.4958263772954924e-06, "loss": 0.3955, "step": 39060 }, { "epoch": 1.95, "learning_rate": 3.49415692821369e-06, "loss": 0.3211, "step": 39070 }, { "epoch": 1.95, "learning_rate": 3.492487479131887e-06, "loss": 0.2992, "step": 39080 }, { "epoch": 1.95, "learning_rate": 3.490818030050084e-06, "loss": 0.3239, "step": 39090 }, { "epoch": 1.96, "learning_rate": 3.489148580968281e-06, "loss": 0.4077, "step": 39100 }, { "epoch": 1.96, "learning_rate": 3.4874791318864774e-06, "loss": 0.3936, "step": 39110 }, { "epoch": 1.96, "learning_rate": 3.4858096828046744e-06, "loss": 0.2445, "step": 39120 }, { "epoch": 1.96, "learning_rate": 3.484140233722872e-06, "loss": 0.3246, "step": 39130 }, { "epoch": 1.96, "learning_rate": 3.482470784641069e-06, "loss": 0.4573, "step": 39140 }, { "epoch": 1.96, "learning_rate": 3.480801335559266e-06, "loss": 0.4868, "step": 39150 }, { "epoch": 1.96, "learning_rate": 3.4791318864774624e-06, "loss": 0.3679, "step": 39160 }, { "epoch": 1.96, "learning_rate": 3.4774624373956594e-06, "loss": 0.4097, "step": 39170 }, { "epoch": 1.96, "learning_rate": 3.475792988313857e-06, "loss": 0.3833, "step": 39180 }, { "epoch": 1.96, "learning_rate": 3.474123539232054e-06, "loss": 0.2986, "step": 39190 }, { "epoch": 1.96, "learning_rate": 3.472454090150251e-06, "loss": 0.364, "step": 39200 }, { "epoch": 1.96, "learning_rate": 3.4707846410684474e-06, "loss": 0.4303, "step": 39210 }, { "epoch": 1.96, "learning_rate": 3.4691151919866444e-06, "loss": 0.3132, "step": 39220 }, { "epoch": 1.96, "learning_rate": 3.467445742904842e-06, "loss": 0.4004, "step": 39230 }, { "epoch": 1.96, "learning_rate": 3.465776293823039e-06, "loss": 0.4213, "step": 39240 }, { "epoch": 1.96, "learning_rate": 3.464106844741236e-06, "loss": 0.3979, "step": 39250 }, { "epoch": 1.96, "learning_rate": 3.4624373956594324e-06, "loss": 0.4054, "step": 39260 }, { "epoch": 1.96, "learning_rate": 3.4607679465776294e-06, "loss": 0.3122, "step": 39270 }, { "epoch": 1.96, "learning_rate": 3.459098497495827e-06, "loss": 0.4145, "step": 39280 }, { "epoch": 1.96, "learning_rate": 3.457429048414024e-06, "loss": 0.3393, "step": 39290 }, { "epoch": 1.96, "learning_rate": 3.455759599332221e-06, "loss": 0.3384, "step": 39300 }, { "epoch": 1.97, "learning_rate": 3.4540901502504174e-06, "loss": 0.3922, "step": 39310 }, { "epoch": 1.97, "learning_rate": 3.4524207011686144e-06, "loss": 0.4454, "step": 39320 }, { "epoch": 1.97, "learning_rate": 3.4507512520868114e-06, "loss": 0.3558, "step": 39330 }, { "epoch": 1.97, "learning_rate": 3.449081803005009e-06, "loss": 0.4192, "step": 39340 }, { "epoch": 1.97, "learning_rate": 3.447412353923206e-06, "loss": 0.3895, "step": 39350 }, { "epoch": 1.97, "learning_rate": 3.445742904841403e-06, "loss": 0.3481, "step": 39360 }, { "epoch": 1.97, "learning_rate": 3.4440734557595994e-06, "loss": 0.3462, "step": 39370 }, { "epoch": 1.97, "learning_rate": 3.4424040066777964e-06, "loss": 0.5021, "step": 39380 }, { "epoch": 1.97, "learning_rate": 3.440734557595994e-06, "loss": 0.3553, "step": 39390 }, { "epoch": 1.97, "learning_rate": 3.439065108514191e-06, "loss": 0.3162, "step": 39400 }, { "epoch": 1.97, "learning_rate": 3.437395659432388e-06, "loss": 0.5258, "step": 39410 }, { "epoch": 1.97, "learning_rate": 3.4357262103505844e-06, "loss": 0.345, "step": 39420 }, { "epoch": 1.97, "learning_rate": 3.4340567612687814e-06, "loss": 0.4412, "step": 39430 }, { "epoch": 1.97, "learning_rate": 3.432387312186979e-06, "loss": 0.4637, "step": 39440 }, { "epoch": 1.97, "learning_rate": 3.430717863105176e-06, "loss": 0.4523, "step": 39450 }, { "epoch": 1.97, "learning_rate": 3.429048414023373e-06, "loss": 0.4302, "step": 39460 }, { "epoch": 1.97, "learning_rate": 3.4273789649415694e-06, "loss": 0.3918, "step": 39470 }, { "epoch": 1.97, "learning_rate": 3.4257095158597664e-06, "loss": 0.3511, "step": 39480 }, { "epoch": 1.97, "learning_rate": 3.4240400667779634e-06, "loss": 0.2936, "step": 39490 }, { "epoch": 1.98, "learning_rate": 3.422370617696161e-06, "loss": 0.4116, "step": 39500 }, { "epoch": 1.98, "learning_rate": 3.4207011686143578e-06, "loss": 0.39, "step": 39510 }, { "epoch": 1.98, "learning_rate": 3.4190317195325544e-06, "loss": 0.4175, "step": 39520 }, { "epoch": 1.98, "learning_rate": 3.4173622704507514e-06, "loss": 0.3408, "step": 39530 }, { "epoch": 1.98, "learning_rate": 3.4156928213689484e-06, "loss": 0.4242, "step": 39540 }, { "epoch": 1.98, "learning_rate": 3.4140233722871458e-06, "loss": 0.3462, "step": 39550 }, { "epoch": 1.98, "learning_rate": 3.4123539232053428e-06, "loss": 0.3957, "step": 39560 }, { "epoch": 1.98, "learning_rate": 3.4106844741235394e-06, "loss": 0.3922, "step": 39570 }, { "epoch": 1.98, "learning_rate": 3.4090150250417364e-06, "loss": 0.3445, "step": 39580 }, { "epoch": 1.98, "learning_rate": 3.4073455759599334e-06, "loss": 0.3839, "step": 39590 }, { "epoch": 1.98, "learning_rate": 3.4056761268781308e-06, "loss": 0.3831, "step": 39600 }, { "epoch": 1.98, "learning_rate": 3.4040066777963278e-06, "loss": 0.4052, "step": 39610 }, { "epoch": 1.98, "learning_rate": 3.4023372287145243e-06, "loss": 0.388, "step": 39620 }, { "epoch": 1.98, "learning_rate": 3.4006677796327213e-06, "loss": 0.3574, "step": 39630 }, { "epoch": 1.98, "learning_rate": 3.3989983305509183e-06, "loss": 0.365, "step": 39640 }, { "epoch": 1.98, "learning_rate": 3.3973288814691153e-06, "loss": 0.3825, "step": 39650 }, { "epoch": 1.98, "learning_rate": 3.3956594323873128e-06, "loss": 0.4516, "step": 39660 }, { "epoch": 1.98, "learning_rate": 3.3939899833055098e-06, "loss": 0.3778, "step": 39670 }, { "epoch": 1.98, "learning_rate": 3.3923205342237063e-06, "loss": 0.3057, "step": 39680 }, { "epoch": 1.98, "learning_rate": 3.3906510851419033e-06, "loss": 0.3528, "step": 39690 }, { "epoch": 1.98, "learning_rate": 3.3889816360601003e-06, "loss": 0.3793, "step": 39700 }, { "epoch": 1.99, "learning_rate": 3.3873121869782978e-06, "loss": 0.4548, "step": 39710 }, { "epoch": 1.99, "learning_rate": 3.3856427378964948e-06, "loss": 0.5032, "step": 39720 }, { "epoch": 1.99, "learning_rate": 3.3839732888146913e-06, "loss": 0.4151, "step": 39730 }, { "epoch": 1.99, "learning_rate": 3.3823038397328883e-06, "loss": 0.3632, "step": 39740 }, { "epoch": 1.99, "learning_rate": 3.3806343906510853e-06, "loss": 0.3166, "step": 39750 }, { "epoch": 1.99, "learning_rate": 3.3789649415692827e-06, "loss": 0.4052, "step": 39760 }, { "epoch": 1.99, "learning_rate": 3.3772954924874797e-06, "loss": 0.3508, "step": 39770 }, { "epoch": 1.99, "learning_rate": 3.3756260434056763e-06, "loss": 0.3644, "step": 39780 }, { "epoch": 1.99, "learning_rate": 3.3739565943238733e-06, "loss": 0.3094, "step": 39790 }, { "epoch": 1.99, "learning_rate": 3.3722871452420703e-06, "loss": 0.3689, "step": 39800 }, { "epoch": 1.99, "learning_rate": 3.3706176961602673e-06, "loss": 0.3295, "step": 39810 }, { "epoch": 1.99, "learning_rate": 3.3689482470784647e-06, "loss": 0.364, "step": 39820 }, { "epoch": 1.99, "learning_rate": 3.3672787979966613e-06, "loss": 0.3987, "step": 39830 }, { "epoch": 1.99, "learning_rate": 3.3656093489148583e-06, "loss": 0.4292, "step": 39840 }, { "epoch": 1.99, "learning_rate": 3.3639398998330553e-06, "loss": 0.4225, "step": 39850 }, { "epoch": 1.99, "learning_rate": 3.3622704507512523e-06, "loss": 0.3684, "step": 39860 }, { "epoch": 1.99, "learning_rate": 3.3606010016694497e-06, "loss": 0.3014, "step": 39870 }, { "epoch": 1.99, "learning_rate": 3.3589315525876463e-06, "loss": 0.3391, "step": 39880 }, { "epoch": 1.99, "learning_rate": 3.3572621035058433e-06, "loss": 0.3546, "step": 39890 }, { "epoch": 2.0, "learning_rate": 3.3555926544240403e-06, "loss": 0.3353, "step": 39900 }, { "epoch": 2.0, "learning_rate": 3.3539232053422373e-06, "loss": 0.3625, "step": 39910 }, { "epoch": 2.0, "learning_rate": 3.3522537562604347e-06, "loss": 0.448, "step": 39920 }, { "epoch": 2.0, "learning_rate": 3.3505843071786313e-06, "loss": 0.4188, "step": 39930 }, { "epoch": 2.0, "learning_rate": 3.3489148580968283e-06, "loss": 0.3248, "step": 39940 }, { "epoch": 2.0, "learning_rate": 3.3472454090150253e-06, "loss": 0.3436, "step": 39950 }, { "epoch": 2.0, "learning_rate": 3.3455759599332223e-06, "loss": 0.2813, "step": 39960 }, { "epoch": 2.0, "learning_rate": 3.343906510851419e-06, "loss": 0.459, "step": 39970 }, { "epoch": 2.0, "learning_rate": 3.3422370617696163e-06, "loss": 0.3181, "step": 39980 }, { "epoch": 2.0, "learning_rate": 3.3405676126878133e-06, "loss": 0.3881, "step": 39990 }, { "epoch": 2.0, "learning_rate": 3.3388981636060103e-06, "loss": 0.4442, "step": 40000 }, { "epoch": 2.0, "eval_loss": 0.37433475255966187, "eval_runtime": 182.7302, "eval_samples_per_second": 20.604, "eval_steps_per_second": 2.578, "step": 40000 }, { "epoch": 2.0, "learning_rate": 3.3372287145242073e-06, "loss": 0.312, "step": 40010 }, { "epoch": 2.0, "learning_rate": 3.335559265442404e-06, "loss": 0.3757, "step": 40020 }, { "epoch": 2.0, "learning_rate": 3.3338898163606017e-06, "loss": 0.427, "step": 40030 }, { "epoch": 2.0, "learning_rate": 3.3322203672787983e-06, "loss": 0.3031, "step": 40040 }, { "epoch": 2.0, "learning_rate": 3.3305509181969953e-06, "loss": 0.3431, "step": 40050 }, { "epoch": 2.0, "learning_rate": 3.3288814691151923e-06, "loss": 0.3903, "step": 40060 }, { "epoch": 2.0, "learning_rate": 3.3272120200333893e-06, "loss": 0.4305, "step": 40070 }, { "epoch": 2.0, "learning_rate": 3.3255425709515867e-06, "loss": 0.2968, "step": 40080 }, { "epoch": 2.0, "learning_rate": 3.3238731218697833e-06, "loss": 0.3524, "step": 40090 }, { "epoch": 2.0, "learning_rate": 3.3222036727879803e-06, "loss": 0.358, "step": 40100 }, { "epoch": 2.01, "learning_rate": 3.3205342237061773e-06, "loss": 0.3574, "step": 40110 }, { "epoch": 2.01, "learning_rate": 3.3188647746243743e-06, "loss": 0.4031, "step": 40120 }, { "epoch": 2.01, "learning_rate": 3.317195325542571e-06, "loss": 0.3137, "step": 40130 }, { "epoch": 2.01, "learning_rate": 3.3155258764607683e-06, "loss": 0.378, "step": 40140 }, { "epoch": 2.01, "learning_rate": 3.3138564273789653e-06, "loss": 0.3776, "step": 40150 }, { "epoch": 2.01, "learning_rate": 3.3121869782971623e-06, "loss": 0.3462, "step": 40160 }, { "epoch": 2.01, "learning_rate": 3.3105175292153592e-06, "loss": 0.3452, "step": 40170 }, { "epoch": 2.01, "learning_rate": 3.308848080133556e-06, "loss": 0.4054, "step": 40180 }, { "epoch": 2.01, "learning_rate": 3.3071786310517532e-06, "loss": 0.3541, "step": 40190 }, { "epoch": 2.01, "learning_rate": 3.3055091819699502e-06, "loss": 0.3866, "step": 40200 }, { "epoch": 2.01, "learning_rate": 3.3038397328881472e-06, "loss": 0.4276, "step": 40210 }, { "epoch": 2.01, "learning_rate": 3.3021702838063442e-06, "loss": 0.3319, "step": 40220 }, { "epoch": 2.01, "learning_rate": 3.300500834724541e-06, "loss": 0.3746, "step": 40230 }, { "epoch": 2.01, "learning_rate": 3.2988313856427382e-06, "loss": 0.3304, "step": 40240 }, { "epoch": 2.01, "learning_rate": 3.2971619365609352e-06, "loss": 0.3762, "step": 40250 }, { "epoch": 2.01, "learning_rate": 3.2954924874791322e-06, "loss": 0.5034, "step": 40260 }, { "epoch": 2.01, "learning_rate": 3.2938230383973292e-06, "loss": 0.4312, "step": 40270 }, { "epoch": 2.01, "learning_rate": 3.292153589315526e-06, "loss": 0.3312, "step": 40280 }, { "epoch": 2.01, "learning_rate": 3.290484140233723e-06, "loss": 0.3964, "step": 40290 }, { "epoch": 2.02, "learning_rate": 3.2888146911519202e-06, "loss": 0.3919, "step": 40300 }, { "epoch": 2.02, "learning_rate": 3.2871452420701172e-06, "loss": 0.3327, "step": 40310 }, { "epoch": 2.02, "learning_rate": 3.2854757929883142e-06, "loss": 0.3842, "step": 40320 }, { "epoch": 2.02, "learning_rate": 3.283806343906511e-06, "loss": 0.4282, "step": 40330 }, { "epoch": 2.02, "learning_rate": 3.282136894824708e-06, "loss": 0.3348, "step": 40340 }, { "epoch": 2.02, "learning_rate": 3.2804674457429052e-06, "loss": 0.2949, "step": 40350 }, { "epoch": 2.02, "learning_rate": 3.2787979966611022e-06, "loss": 0.3766, "step": 40360 }, { "epoch": 2.02, "learning_rate": 3.2771285475792992e-06, "loss": 0.3552, "step": 40370 }, { "epoch": 2.02, "learning_rate": 3.275459098497496e-06, "loss": 0.4679, "step": 40380 }, { "epoch": 2.02, "learning_rate": 3.2737896494156928e-06, "loss": 0.4829, "step": 40390 }, { "epoch": 2.02, "learning_rate": 3.27212020033389e-06, "loss": 0.3806, "step": 40400 }, { "epoch": 2.02, "learning_rate": 3.270450751252087e-06, "loss": 0.3067, "step": 40410 }, { "epoch": 2.02, "learning_rate": 3.268781302170284e-06, "loss": 0.3029, "step": 40420 }, { "epoch": 2.02, "learning_rate": 3.267111853088481e-06, "loss": 0.4291, "step": 40430 }, { "epoch": 2.02, "learning_rate": 3.2654424040066778e-06, "loss": 0.3304, "step": 40440 }, { "epoch": 2.02, "learning_rate": 3.2637729549248748e-06, "loss": 0.3054, "step": 40450 }, { "epoch": 2.02, "learning_rate": 3.262103505843072e-06, "loss": 0.3818, "step": 40460 }, { "epoch": 2.02, "learning_rate": 3.260434056761269e-06, "loss": 0.3762, "step": 40470 }, { "epoch": 2.02, "learning_rate": 3.258764607679466e-06, "loss": 0.3556, "step": 40480 }, { "epoch": 2.02, "learning_rate": 3.2570951585976628e-06, "loss": 0.4016, "step": 40490 }, { "epoch": 2.02, "learning_rate": 3.2554257095158598e-06, "loss": 0.4976, "step": 40500 }, { "epoch": 2.03, "learning_rate": 3.253756260434057e-06, "loss": 0.3351, "step": 40510 }, { "epoch": 2.03, "learning_rate": 3.252086811352254e-06, "loss": 0.361, "step": 40520 }, { "epoch": 2.03, "learning_rate": 3.250417362270451e-06, "loss": 0.3423, "step": 40530 }, { "epoch": 2.03, "learning_rate": 3.2487479131886478e-06, "loss": 0.4007, "step": 40540 }, { "epoch": 2.03, "learning_rate": 3.2470784641068448e-06, "loss": 0.341, "step": 40550 }, { "epoch": 2.03, "learning_rate": 3.245409015025042e-06, "loss": 0.3567, "step": 40560 }, { "epoch": 2.03, "learning_rate": 3.243739565943239e-06, "loss": 0.3806, "step": 40570 }, { "epoch": 2.03, "learning_rate": 3.242070116861436e-06, "loss": 0.404, "step": 40580 }, { "epoch": 2.03, "learning_rate": 3.2404006677796328e-06, "loss": 0.474, "step": 40590 }, { "epoch": 2.03, "learning_rate": 3.2387312186978297e-06, "loss": 0.3147, "step": 40600 }, { "epoch": 2.03, "learning_rate": 3.237061769616027e-06, "loss": 0.3711, "step": 40610 }, { "epoch": 2.03, "learning_rate": 3.235392320534224e-06, "loss": 0.4099, "step": 40620 }, { "epoch": 2.03, "learning_rate": 3.233722871452421e-06, "loss": 0.3845, "step": 40630 }, { "epoch": 2.03, "learning_rate": 3.2320534223706177e-06, "loss": 0.3163, "step": 40640 }, { "epoch": 2.03, "learning_rate": 3.2303839732888147e-06, "loss": 0.3654, "step": 40650 }, { "epoch": 2.03, "learning_rate": 3.2287145242070117e-06, "loss": 0.3881, "step": 40660 }, { "epoch": 2.03, "learning_rate": 3.227045075125209e-06, "loss": 0.3358, "step": 40670 }, { "epoch": 2.03, "learning_rate": 3.225375626043406e-06, "loss": 0.4836, "step": 40680 }, { "epoch": 2.03, "learning_rate": 3.2237061769616027e-06, "loss": 0.3763, "step": 40690 }, { "epoch": 2.04, "learning_rate": 3.2220367278797997e-06, "loss": 0.4527, "step": 40700 }, { "epoch": 2.04, "learning_rate": 3.2203672787979967e-06, "loss": 0.3355, "step": 40710 }, { "epoch": 2.04, "learning_rate": 3.218697829716194e-06, "loss": 0.325, "step": 40720 }, { "epoch": 2.04, "learning_rate": 3.217028380634391e-06, "loss": 0.332, "step": 40730 }, { "epoch": 2.04, "learning_rate": 3.2153589315525877e-06, "loss": 0.3918, "step": 40740 }, { "epoch": 2.04, "learning_rate": 3.2136894824707847e-06, "loss": 0.3941, "step": 40750 }, { "epoch": 2.04, "learning_rate": 3.2120200333889817e-06, "loss": 0.3402, "step": 40760 }, { "epoch": 2.04, "learning_rate": 3.210350584307179e-06, "loss": 0.3824, "step": 40770 }, { "epoch": 2.04, "learning_rate": 3.208681135225376e-06, "loss": 0.3855, "step": 40780 }, { "epoch": 2.04, "learning_rate": 3.207011686143573e-06, "loss": 0.3199, "step": 40790 }, { "epoch": 2.04, "learning_rate": 3.2053422370617697e-06, "loss": 0.3814, "step": 40800 }, { "epoch": 2.04, "learning_rate": 3.2036727879799667e-06, "loss": 0.3365, "step": 40810 }, { "epoch": 2.04, "learning_rate": 3.2020033388981637e-06, "loss": 0.3224, "step": 40820 }, { "epoch": 2.04, "learning_rate": 3.200333889816361e-06, "loss": 0.3625, "step": 40830 }, { "epoch": 2.04, "learning_rate": 3.198664440734558e-06, "loss": 0.3156, "step": 40840 }, { "epoch": 2.04, "learning_rate": 3.1969949916527547e-06, "loss": 0.4102, "step": 40850 }, { "epoch": 2.04, "learning_rate": 3.1953255425709517e-06, "loss": 0.3219, "step": 40860 }, { "epoch": 2.04, "learning_rate": 3.1936560934891487e-06, "loss": 0.3162, "step": 40870 }, { "epoch": 2.04, "learning_rate": 3.191986644407346e-06, "loss": 0.372, "step": 40880 }, { "epoch": 2.04, "learning_rate": 3.190317195325543e-06, "loss": 0.3274, "step": 40890 }, { "epoch": 2.04, "learning_rate": 3.1886477462437397e-06, "loss": 0.3994, "step": 40900 }, { "epoch": 2.05, "learning_rate": 3.1869782971619367e-06, "loss": 0.4334, "step": 40910 }, { "epoch": 2.05, "learning_rate": 3.1853088480801337e-06, "loss": 0.3286, "step": 40920 }, { "epoch": 2.05, "learning_rate": 3.183639398998331e-06, "loss": 0.3733, "step": 40930 }, { "epoch": 2.05, "learning_rate": 3.181969949916528e-06, "loss": 0.3882, "step": 40940 }, { "epoch": 2.05, "learning_rate": 3.1803005008347247e-06, "loss": 0.3401, "step": 40950 }, { "epoch": 2.05, "learning_rate": 3.1786310517529217e-06, "loss": 0.3361, "step": 40960 }, { "epoch": 2.05, "learning_rate": 3.1769616026711187e-06, "loss": 0.3617, "step": 40970 }, { "epoch": 2.05, "learning_rate": 3.1752921535893157e-06, "loss": 0.3559, "step": 40980 }, { "epoch": 2.05, "learning_rate": 3.173622704507513e-06, "loss": 0.4543, "step": 40990 }, { "epoch": 2.05, "learning_rate": 3.1719532554257097e-06, "loss": 0.4343, "step": 41000 }, { "epoch": 2.05, "learning_rate": 3.1702838063439067e-06, "loss": 0.394, "step": 41010 }, { "epoch": 2.05, "learning_rate": 3.1686143572621037e-06, "loss": 0.3125, "step": 41020 }, { "epoch": 2.05, "learning_rate": 3.1669449081803007e-06, "loss": 0.3494, "step": 41030 }, { "epoch": 2.05, "learning_rate": 3.165275459098498e-06, "loss": 0.3774, "step": 41040 }, { "epoch": 2.05, "learning_rate": 3.1636060100166947e-06, "loss": 0.2936, "step": 41050 }, { "epoch": 2.05, "learning_rate": 3.1619365609348917e-06, "loss": 0.3537, "step": 41060 }, { "epoch": 2.05, "learning_rate": 3.1602671118530887e-06, "loss": 0.4069, "step": 41070 }, { "epoch": 2.05, "learning_rate": 3.1585976627712857e-06, "loss": 0.3477, "step": 41080 }, { "epoch": 2.05, "learning_rate": 3.156928213689483e-06, "loss": 0.3922, "step": 41090 }, { "epoch": 2.06, "learning_rate": 3.15525876460768e-06, "loss": 0.329, "step": 41100 }, { "epoch": 2.06, "learning_rate": 3.1535893155258767e-06, "loss": 0.3496, "step": 41110 }, { "epoch": 2.06, "learning_rate": 3.1519198664440737e-06, "loss": 0.3933, "step": 41120 }, { "epoch": 2.06, "learning_rate": 3.1502504173622707e-06, "loss": 0.3836, "step": 41130 }, { "epoch": 2.06, "learning_rate": 3.1485809682804672e-06, "loss": 0.3516, "step": 41140 }, { "epoch": 2.06, "learning_rate": 3.146911519198665e-06, "loss": 0.3293, "step": 41150 }, { "epoch": 2.06, "learning_rate": 3.1452420701168617e-06, "loss": 0.3705, "step": 41160 }, { "epoch": 2.06, "learning_rate": 3.1435726210350587e-06, "loss": 0.3863, "step": 41170 }, { "epoch": 2.06, "learning_rate": 3.1419031719532556e-06, "loss": 0.4829, "step": 41180 }, { "epoch": 2.06, "learning_rate": 3.1402337228714526e-06, "loss": 0.3481, "step": 41190 }, { "epoch": 2.06, "learning_rate": 3.13856427378965e-06, "loss": 0.3336, "step": 41200 }, { "epoch": 2.06, "learning_rate": 3.1368948247078466e-06, "loss": 0.3368, "step": 41210 }, { "epoch": 2.06, "learning_rate": 3.1352253756260436e-06, "loss": 0.3708, "step": 41220 }, { "epoch": 2.06, "learning_rate": 3.1335559265442406e-06, "loss": 0.3394, "step": 41230 }, { "epoch": 2.06, "learning_rate": 3.1318864774624376e-06, "loss": 0.4799, "step": 41240 }, { "epoch": 2.06, "learning_rate": 3.130217028380635e-06, "loss": 0.3281, "step": 41250 }, { "epoch": 2.06, "learning_rate": 3.1285475792988316e-06, "loss": 0.403, "step": 41260 }, { "epoch": 2.06, "learning_rate": 3.1268781302170286e-06, "loss": 0.3646, "step": 41270 }, { "epoch": 2.06, "learning_rate": 3.1252086811352256e-06, "loss": 0.3731, "step": 41280 }, { "epoch": 2.06, "learning_rate": 3.1235392320534226e-06, "loss": 0.3618, "step": 41290 }, { "epoch": 2.06, "learning_rate": 3.121869782971619e-06, "loss": 0.529, "step": 41300 }, { "epoch": 2.07, "learning_rate": 3.1202003338898166e-06, "loss": 0.419, "step": 41310 }, { "epoch": 2.07, "learning_rate": 3.1185308848080136e-06, "loss": 0.336, "step": 41320 }, { "epoch": 2.07, "learning_rate": 3.1168614357262106e-06, "loss": 0.336, "step": 41330 }, { "epoch": 2.07, "learning_rate": 3.1151919866444076e-06, "loss": 0.3536, "step": 41340 }, { "epoch": 2.07, "learning_rate": 3.113522537562604e-06, "loss": 0.4374, "step": 41350 }, { "epoch": 2.07, "learning_rate": 3.1118530884808016e-06, "loss": 0.3931, "step": 41360 }, { "epoch": 2.07, "learning_rate": 3.1101836393989986e-06, "loss": 0.2857, "step": 41370 }, { "epoch": 2.07, "learning_rate": 3.1085141903171956e-06, "loss": 0.3498, "step": 41380 }, { "epoch": 2.07, "learning_rate": 3.1068447412353926e-06, "loss": 0.2866, "step": 41390 }, { "epoch": 2.07, "learning_rate": 3.105175292153589e-06, "loss": 0.321, "step": 41400 }, { "epoch": 2.07, "learning_rate": 3.1035058430717866e-06, "loss": 0.3291, "step": 41410 }, { "epoch": 2.07, "learning_rate": 3.1018363939899836e-06, "loss": 0.3873, "step": 41420 }, { "epoch": 2.07, "learning_rate": 3.1001669449081806e-06, "loss": 0.35, "step": 41430 }, { "epoch": 2.07, "learning_rate": 3.0984974958263776e-06, "loss": 0.36, "step": 41440 }, { "epoch": 2.07, "learning_rate": 3.096828046744574e-06, "loss": 0.3805, "step": 41450 }, { "epoch": 2.07, "learning_rate": 3.095158597662771e-06, "loss": 0.3703, "step": 41460 }, { "epoch": 2.07, "learning_rate": 3.0934891485809686e-06, "loss": 0.3841, "step": 41470 }, { "epoch": 2.07, "learning_rate": 3.0918196994991656e-06, "loss": 0.4154, "step": 41480 }, { "epoch": 2.07, "learning_rate": 3.0901502504173626e-06, "loss": 0.4299, "step": 41490 }, { "epoch": 2.08, "learning_rate": 3.0884808013355596e-06, "loss": 0.3196, "step": 41500 }, { "epoch": 2.08, "learning_rate": 3.086811352253756e-06, "loss": 0.321, "step": 41510 }, { "epoch": 2.08, "learning_rate": 3.0851419031719536e-06, "loss": 0.3383, "step": 41520 }, { "epoch": 2.08, "learning_rate": 3.0834724540901506e-06, "loss": 0.4072, "step": 41530 }, { "epoch": 2.08, "learning_rate": 3.0818030050083476e-06, "loss": 0.4126, "step": 41540 }, { "epoch": 2.08, "learning_rate": 3.0801335559265446e-06, "loss": 0.4147, "step": 41550 }, { "epoch": 2.08, "learning_rate": 3.078464106844741e-06, "loss": 0.4776, "step": 41560 }, { "epoch": 2.08, "learning_rate": 3.0767946577629386e-06, "loss": 0.3323, "step": 41570 }, { "epoch": 2.08, "learning_rate": 3.0751252086811356e-06, "loss": 0.4216, "step": 41580 }, { "epoch": 2.08, "learning_rate": 3.0734557595993326e-06, "loss": 0.3134, "step": 41590 }, { "epoch": 2.08, "learning_rate": 3.0717863105175296e-06, "loss": 0.3961, "step": 41600 }, { "epoch": 2.08, "learning_rate": 3.070116861435726e-06, "loss": 0.3078, "step": 41610 }, { "epoch": 2.08, "learning_rate": 3.068447412353923e-06, "loss": 0.2988, "step": 41620 }, { "epoch": 2.08, "learning_rate": 3.0667779632721206e-06, "loss": 0.3699, "step": 41630 }, { "epoch": 2.08, "learning_rate": 3.0651085141903176e-06, "loss": 0.4578, "step": 41640 }, { "epoch": 2.08, "learning_rate": 3.0634390651085146e-06, "loss": 0.3853, "step": 41650 }, { "epoch": 2.08, "learning_rate": 3.061769616026711e-06, "loss": 0.3166, "step": 41660 }, { "epoch": 2.08, "learning_rate": 3.060100166944908e-06, "loss": 0.3418, "step": 41670 }, { "epoch": 2.08, "learning_rate": 3.0584307178631056e-06, "loss": 0.3977, "step": 41680 }, { "epoch": 2.08, "learning_rate": 3.0567612687813026e-06, "loss": 0.4798, "step": 41690 }, { "epoch": 2.08, "learning_rate": 3.0550918196994996e-06, "loss": 0.3501, "step": 41700 }, { "epoch": 2.09, "learning_rate": 3.053422370617696e-06, "loss": 0.4036, "step": 41710 }, { "epoch": 2.09, "learning_rate": 3.051752921535893e-06, "loss": 0.3741, "step": 41720 }, { "epoch": 2.09, "learning_rate": 3.0500834724540906e-06, "loss": 0.3805, "step": 41730 }, { "epoch": 2.09, "learning_rate": 3.0484140233722876e-06, "loss": 0.4258, "step": 41740 }, { "epoch": 2.09, "learning_rate": 3.0467445742904845e-06, "loss": 0.3662, "step": 41750 }, { "epoch": 2.09, "learning_rate": 3.045075125208681e-06, "loss": 0.4793, "step": 41760 }, { "epoch": 2.09, "learning_rate": 3.043405676126878e-06, "loss": 0.4034, "step": 41770 }, { "epoch": 2.09, "learning_rate": 3.0417362270450755e-06, "loss": 0.3205, "step": 41780 }, { "epoch": 2.09, "learning_rate": 3.0400667779632725e-06, "loss": 0.4541, "step": 41790 }, { "epoch": 2.09, "learning_rate": 3.0383973288814695e-06, "loss": 0.4234, "step": 41800 }, { "epoch": 2.09, "learning_rate": 3.036727879799666e-06, "loss": 0.4781, "step": 41810 }, { "epoch": 2.09, "learning_rate": 3.035058430717863e-06, "loss": 0.2851, "step": 41820 }, { "epoch": 2.09, "learning_rate": 3.03338898163606e-06, "loss": 0.3802, "step": 41830 }, { "epoch": 2.09, "learning_rate": 3.0317195325542575e-06, "loss": 0.3089, "step": 41840 }, { "epoch": 2.09, "learning_rate": 3.0300500834724545e-06, "loss": 0.353, "step": 41850 }, { "epoch": 2.09, "learning_rate": 3.0283806343906515e-06, "loss": 0.4124, "step": 41860 }, { "epoch": 2.09, "learning_rate": 3.026711185308848e-06, "loss": 0.45, "step": 41870 }, { "epoch": 2.09, "learning_rate": 3.025041736227045e-06, "loss": 0.3824, "step": 41880 }, { "epoch": 2.09, "learning_rate": 3.0233722871452425e-06, "loss": 0.3495, "step": 41890 }, { "epoch": 2.1, "learning_rate": 3.0217028380634395e-06, "loss": 0.3456, "step": 41900 }, { "epoch": 2.1, "learning_rate": 3.0200333889816365e-06, "loss": 0.314, "step": 41910 }, { "epoch": 2.1, "learning_rate": 3.018363939899833e-06, "loss": 0.3398, "step": 41920 }, { "epoch": 2.1, "learning_rate": 3.01669449081803e-06, "loss": 0.3141, "step": 41930 }, { "epoch": 2.1, "learning_rate": 3.0150250417362275e-06, "loss": 0.3738, "step": 41940 }, { "epoch": 2.1, "learning_rate": 3.0133555926544245e-06, "loss": 0.4057, "step": 41950 }, { "epoch": 2.1, "learning_rate": 3.0116861435726215e-06, "loss": 0.4181, "step": 41960 }, { "epoch": 2.1, "learning_rate": 3.010016694490818e-06, "loss": 0.3867, "step": 41970 }, { "epoch": 2.1, "learning_rate": 3.008347245409015e-06, "loss": 0.3475, "step": 41980 }, { "epoch": 2.1, "learning_rate": 3.006677796327212e-06, "loss": 0.4001, "step": 41990 }, { "epoch": 2.1, "learning_rate": 3.0050083472454095e-06, "loss": 0.315, "step": 42000 }, { "epoch": 2.1, "learning_rate": 3.0033388981636065e-06, "loss": 0.3915, "step": 42010 }, { "epoch": 2.1, "learning_rate": 3.001669449081803e-06, "loss": 0.3564, "step": 42020 }, { "epoch": 2.1, "learning_rate": 3e-06, "loss": 0.3597, "step": 42030 }, { "epoch": 2.1, "learning_rate": 2.998330550918197e-06, "loss": 0.3952, "step": 42040 }, { "epoch": 2.1, "learning_rate": 2.9966611018363945e-06, "loss": 0.3467, "step": 42050 }, { "epoch": 2.1, "learning_rate": 2.9949916527545915e-06, "loss": 0.4337, "step": 42060 }, { "epoch": 2.1, "learning_rate": 2.993322203672788e-06, "loss": 0.3552, "step": 42070 }, { "epoch": 2.1, "learning_rate": 2.991652754590985e-06, "loss": 0.3732, "step": 42080 }, { "epoch": 2.1, "learning_rate": 2.989983305509182e-06, "loss": 0.3595, "step": 42090 }, { "epoch": 2.1, "learning_rate": 2.9883138564273795e-06, "loss": 0.4467, "step": 42100 }, { "epoch": 2.11, "learning_rate": 2.9866444073455765e-06, "loss": 0.3678, "step": 42110 }, { "epoch": 2.11, "learning_rate": 2.984974958263773e-06, "loss": 0.4614, "step": 42120 }, { "epoch": 2.11, "learning_rate": 2.98330550918197e-06, "loss": 0.3125, "step": 42130 }, { "epoch": 2.11, "learning_rate": 2.981636060100167e-06, "loss": 0.5229, "step": 42140 }, { "epoch": 2.11, "learning_rate": 2.979966611018364e-06, "loss": 0.3497, "step": 42150 }, { "epoch": 2.11, "learning_rate": 2.9782971619365615e-06, "loss": 0.3838, "step": 42160 }, { "epoch": 2.11, "learning_rate": 2.9766277128547585e-06, "loss": 0.4032, "step": 42170 }, { "epoch": 2.11, "learning_rate": 2.974958263772955e-06, "loss": 0.2927, "step": 42180 }, { "epoch": 2.11, "learning_rate": 2.973288814691152e-06, "loss": 0.4512, "step": 42190 }, { "epoch": 2.11, "learning_rate": 2.971619365609349e-06, "loss": 0.3317, "step": 42200 }, { "epoch": 2.11, "learning_rate": 2.9699499165275465e-06, "loss": 0.3784, "step": 42210 }, { "epoch": 2.11, "learning_rate": 2.9682804674457435e-06, "loss": 0.3425, "step": 42220 }, { "epoch": 2.11, "learning_rate": 2.96661101836394e-06, "loss": 0.2835, "step": 42230 }, { "epoch": 2.11, "learning_rate": 2.964941569282137e-06, "loss": 0.3022, "step": 42240 }, { "epoch": 2.11, "learning_rate": 2.963272120200334e-06, "loss": 0.4063, "step": 42250 }, { "epoch": 2.11, "learning_rate": 2.9616026711185315e-06, "loss": 0.3579, "step": 42260 }, { "epoch": 2.11, "learning_rate": 2.9599332220367285e-06, "loss": 0.4009, "step": 42270 }, { "epoch": 2.11, "learning_rate": 2.958263772954925e-06, "loss": 0.3587, "step": 42280 }, { "epoch": 2.11, "learning_rate": 2.956594323873122e-06, "loss": 0.3201, "step": 42290 }, { "epoch": 2.12, "learning_rate": 2.954924874791319e-06, "loss": 0.3954, "step": 42300 }, { "epoch": 2.12, "learning_rate": 2.953255425709516e-06, "loss": 0.4379, "step": 42310 }, { "epoch": 2.12, "learning_rate": 2.9515859766277134e-06, "loss": 0.3188, "step": 42320 }, { "epoch": 2.12, "learning_rate": 2.94991652754591e-06, "loss": 0.3499, "step": 42330 }, { "epoch": 2.12, "learning_rate": 2.948247078464107e-06, "loss": 0.424, "step": 42340 }, { "epoch": 2.12, "learning_rate": 2.946577629382304e-06, "loss": 0.4105, "step": 42350 }, { "epoch": 2.12, "learning_rate": 2.944908180300501e-06, "loss": 0.4014, "step": 42360 }, { "epoch": 2.12, "learning_rate": 2.9432387312186984e-06, "loss": 0.4484, "step": 42370 }, { "epoch": 2.12, "learning_rate": 2.941569282136895e-06, "loss": 0.3347, "step": 42380 }, { "epoch": 2.12, "learning_rate": 2.939899833055092e-06, "loss": 0.3686, "step": 42390 }, { "epoch": 2.12, "learning_rate": 2.938230383973289e-06, "loss": 0.492, "step": 42400 }, { "epoch": 2.12, "learning_rate": 2.936560934891486e-06, "loss": 0.4164, "step": 42410 }, { "epoch": 2.12, "learning_rate": 2.9348914858096834e-06, "loss": 0.3751, "step": 42420 }, { "epoch": 2.12, "learning_rate": 2.93322203672788e-06, "loss": 0.3346, "step": 42430 }, { "epoch": 2.12, "learning_rate": 2.931552587646077e-06, "loss": 0.2953, "step": 42440 }, { "epoch": 2.12, "learning_rate": 2.929883138564274e-06, "loss": 0.3608, "step": 42450 }, { "epoch": 2.12, "learning_rate": 2.928213689482471e-06, "loss": 0.4023, "step": 42460 }, { "epoch": 2.12, "learning_rate": 2.9265442404006676e-06, "loss": 0.308, "step": 42470 }, { "epoch": 2.12, "learning_rate": 2.924874791318865e-06, "loss": 0.4752, "step": 42480 }, { "epoch": 2.12, "learning_rate": 2.923205342237062e-06, "loss": 0.33, "step": 42490 }, { "epoch": 2.12, "learning_rate": 2.921535893155259e-06, "loss": 0.3335, "step": 42500 }, { "epoch": 2.13, "learning_rate": 2.919866444073456e-06, "loss": 0.3246, "step": 42510 }, { "epoch": 2.13, "learning_rate": 2.9181969949916526e-06, "loss": 0.2736, "step": 42520 }, { "epoch": 2.13, "learning_rate": 2.9165275459098504e-06, "loss": 0.475, "step": 42530 }, { "epoch": 2.13, "learning_rate": 2.914858096828047e-06, "loss": 0.3312, "step": 42540 }, { "epoch": 2.13, "learning_rate": 2.913188647746244e-06, "loss": 0.3498, "step": 42550 }, { "epoch": 2.13, "learning_rate": 2.911519198664441e-06, "loss": 0.292, "step": 42560 }, { "epoch": 2.13, "learning_rate": 2.909849749582638e-06, "loss": 0.3826, "step": 42570 }, { "epoch": 2.13, "learning_rate": 2.9081803005008354e-06, "loss": 0.3205, "step": 42580 }, { "epoch": 2.13, "learning_rate": 2.906510851419032e-06, "loss": 0.333, "step": 42590 }, { "epoch": 2.13, "learning_rate": 2.904841402337229e-06, "loss": 0.3761, "step": 42600 }, { "epoch": 2.13, "learning_rate": 2.903171953255426e-06, "loss": 0.4658, "step": 42610 }, { "epoch": 2.13, "learning_rate": 2.901502504173623e-06, "loss": 0.3467, "step": 42620 }, { "epoch": 2.13, "learning_rate": 2.8998330550918195e-06, "loss": 0.358, "step": 42630 }, { "epoch": 2.13, "learning_rate": 2.898163606010017e-06, "loss": 0.3515, "step": 42640 }, { "epoch": 2.13, "learning_rate": 2.896494156928214e-06, "loss": 0.4042, "step": 42650 }, { "epoch": 2.13, "learning_rate": 2.894824707846411e-06, "loss": 0.3751, "step": 42660 }, { "epoch": 2.13, "learning_rate": 2.893155258764608e-06, "loss": 0.3808, "step": 42670 }, { "epoch": 2.13, "learning_rate": 2.8914858096828045e-06, "loss": 0.3065, "step": 42680 }, { "epoch": 2.13, "learning_rate": 2.889816360601002e-06, "loss": 0.3914, "step": 42690 }, { "epoch": 2.13, "learning_rate": 2.888146911519199e-06, "loss": 0.3076, "step": 42700 }, { "epoch": 2.14, "learning_rate": 2.886477462437396e-06, "loss": 0.3707, "step": 42710 }, { "epoch": 2.14, "learning_rate": 2.884808013355593e-06, "loss": 0.3831, "step": 42720 }, { "epoch": 2.14, "learning_rate": 2.8831385642737895e-06, "loss": 0.3495, "step": 42730 }, { "epoch": 2.14, "learning_rate": 2.881469115191987e-06, "loss": 0.3624, "step": 42740 }, { "epoch": 2.14, "learning_rate": 2.879799666110184e-06, "loss": 0.3436, "step": 42750 }, { "epoch": 2.14, "learning_rate": 2.878130217028381e-06, "loss": 0.3476, "step": 42760 }, { "epoch": 2.14, "learning_rate": 2.876460767946578e-06, "loss": 0.3433, "step": 42770 }, { "epoch": 2.14, "learning_rate": 2.8747913188647745e-06, "loss": 0.3409, "step": 42780 }, { "epoch": 2.14, "learning_rate": 2.8731218697829715e-06, "loss": 0.3597, "step": 42790 }, { "epoch": 2.14, "learning_rate": 2.871452420701169e-06, "loss": 0.3509, "step": 42800 }, { "epoch": 2.14, "learning_rate": 2.869782971619366e-06, "loss": 0.3206, "step": 42810 }, { "epoch": 2.14, "learning_rate": 2.868113522537563e-06, "loss": 0.3572, "step": 42820 }, { "epoch": 2.14, "learning_rate": 2.8664440734557595e-06, "loss": 0.3846, "step": 42830 }, { "epoch": 2.14, "learning_rate": 2.8647746243739565e-06, "loss": 0.3208, "step": 42840 }, { "epoch": 2.14, "learning_rate": 2.863105175292154e-06, "loss": 0.3702, "step": 42850 }, { "epoch": 2.14, "learning_rate": 2.861435726210351e-06, "loss": 0.3049, "step": 42860 }, { "epoch": 2.14, "learning_rate": 2.859766277128548e-06, "loss": 0.2688, "step": 42870 }, { "epoch": 2.14, "learning_rate": 2.8580968280467445e-06, "loss": 0.3357, "step": 42880 }, { "epoch": 2.14, "learning_rate": 2.8564273789649415e-06, "loss": 0.3271, "step": 42890 }, { "epoch": 2.15, "learning_rate": 2.854757929883139e-06, "loss": 0.4478, "step": 42900 }, { "epoch": 2.15, "learning_rate": 2.853088480801336e-06, "loss": 0.3065, "step": 42910 }, { "epoch": 2.15, "learning_rate": 2.851419031719533e-06, "loss": 0.3409, "step": 42920 }, { "epoch": 2.15, "learning_rate": 2.84974958263773e-06, "loss": 0.3231, "step": 42930 }, { "epoch": 2.15, "learning_rate": 2.8480801335559265e-06, "loss": 0.3509, "step": 42940 }, { "epoch": 2.15, "learning_rate": 2.8464106844741235e-06, "loss": 0.3127, "step": 42950 }, { "epoch": 2.15, "learning_rate": 2.844741235392321e-06, "loss": 0.3739, "step": 42960 }, { "epoch": 2.15, "learning_rate": 2.843071786310518e-06, "loss": 0.3577, "step": 42970 }, { "epoch": 2.15, "learning_rate": 2.841402337228715e-06, "loss": 0.2734, "step": 42980 }, { "epoch": 2.15, "learning_rate": 2.8397328881469115e-06, "loss": 0.3539, "step": 42990 }, { "epoch": 2.15, "learning_rate": 2.8380634390651085e-06, "loss": 0.3947, "step": 43000 }, { "epoch": 2.15, "learning_rate": 2.836393989983306e-06, "loss": 0.3708, "step": 43010 }, { "epoch": 2.15, "learning_rate": 2.834724540901503e-06, "loss": 0.2929, "step": 43020 }, { "epoch": 2.15, "learning_rate": 2.8330550918197e-06, "loss": 0.321, "step": 43030 }, { "epoch": 2.15, "learning_rate": 2.8313856427378965e-06, "loss": 0.3953, "step": 43040 }, { "epoch": 2.15, "learning_rate": 2.8297161936560935e-06, "loss": 0.3313, "step": 43050 }, { "epoch": 2.15, "learning_rate": 2.828046744574291e-06, "loss": 0.4289, "step": 43060 }, { "epoch": 2.15, "learning_rate": 2.826377295492488e-06, "loss": 0.3821, "step": 43070 }, { "epoch": 2.15, "learning_rate": 2.824707846410685e-06, "loss": 0.4792, "step": 43080 }, { "epoch": 2.15, "learning_rate": 2.8230383973288815e-06, "loss": 0.4149, "step": 43090 }, { "epoch": 2.15, "learning_rate": 2.8213689482470785e-06, "loss": 0.4361, "step": 43100 }, { "epoch": 2.16, "learning_rate": 2.819699499165276e-06, "loss": 0.4311, "step": 43110 }, { "epoch": 2.16, "learning_rate": 2.818030050083473e-06, "loss": 0.2958, "step": 43120 }, { "epoch": 2.16, "learning_rate": 2.81636060100167e-06, "loss": 0.3585, "step": 43130 }, { "epoch": 2.16, "learning_rate": 2.8146911519198665e-06, "loss": 0.3514, "step": 43140 }, { "epoch": 2.16, "learning_rate": 2.8130217028380635e-06, "loss": 0.3547, "step": 43150 }, { "epoch": 2.16, "learning_rate": 2.8113522537562605e-06, "loss": 0.3934, "step": 43160 }, { "epoch": 2.16, "learning_rate": 2.809682804674458e-06, "loss": 0.362, "step": 43170 }, { "epoch": 2.16, "learning_rate": 2.808013355592655e-06, "loss": 0.286, "step": 43180 }, { "epoch": 2.16, "learning_rate": 2.8063439065108514e-06, "loss": 0.3724, "step": 43190 }, { "epoch": 2.16, "learning_rate": 2.8046744574290484e-06, "loss": 0.3535, "step": 43200 }, { "epoch": 2.16, "learning_rate": 2.8030050083472454e-06, "loss": 0.3778, "step": 43210 }, { "epoch": 2.16, "learning_rate": 2.801335559265443e-06, "loss": 0.4094, "step": 43220 }, { "epoch": 2.16, "learning_rate": 2.79966611018364e-06, "loss": 0.3784, "step": 43230 }, { "epoch": 2.16, "learning_rate": 2.797996661101837e-06, "loss": 0.4373, "step": 43240 }, { "epoch": 2.16, "learning_rate": 2.7963272120200334e-06, "loss": 0.4171, "step": 43250 }, { "epoch": 2.16, "learning_rate": 2.7946577629382304e-06, "loss": 0.4695, "step": 43260 }, { "epoch": 2.16, "learning_rate": 2.792988313856428e-06, "loss": 0.3775, "step": 43270 }, { "epoch": 2.16, "learning_rate": 2.791318864774625e-06, "loss": 0.4678, "step": 43280 }, { "epoch": 2.16, "learning_rate": 2.789649415692822e-06, "loss": 0.3277, "step": 43290 }, { "epoch": 2.17, "learning_rate": 2.7879799666110184e-06, "loss": 0.3714, "step": 43300 }, { "epoch": 2.17, "learning_rate": 2.7863105175292154e-06, "loss": 0.4015, "step": 43310 }, { "epoch": 2.17, "learning_rate": 2.7846410684474124e-06, "loss": 0.4573, "step": 43320 }, { "epoch": 2.17, "learning_rate": 2.78297161936561e-06, "loss": 0.4035, "step": 43330 }, { "epoch": 2.17, "learning_rate": 2.781302170283807e-06, "loss": 0.3209, "step": 43340 }, { "epoch": 2.17, "learning_rate": 2.7796327212020034e-06, "loss": 0.4, "step": 43350 }, { "epoch": 2.17, "learning_rate": 2.7779632721202004e-06, "loss": 0.3429, "step": 43360 }, { "epoch": 2.17, "learning_rate": 2.7762938230383974e-06, "loss": 0.3763, "step": 43370 }, { "epoch": 2.17, "learning_rate": 2.774624373956595e-06, "loss": 0.3594, "step": 43380 }, { "epoch": 2.17, "learning_rate": 2.772954924874792e-06, "loss": 0.3373, "step": 43390 }, { "epoch": 2.17, "learning_rate": 2.7712854757929884e-06, "loss": 0.3155, "step": 43400 }, { "epoch": 2.17, "learning_rate": 2.7696160267111854e-06, "loss": 0.408, "step": 43410 }, { "epoch": 2.17, "learning_rate": 2.7679465776293824e-06, "loss": 0.3566, "step": 43420 }, { "epoch": 2.17, "learning_rate": 2.76627712854758e-06, "loss": 0.3764, "step": 43430 }, { "epoch": 2.17, "learning_rate": 2.764607679465777e-06, "loss": 0.3427, "step": 43440 }, { "epoch": 2.17, "learning_rate": 2.7629382303839734e-06, "loss": 0.3303, "step": 43450 }, { "epoch": 2.17, "learning_rate": 2.7612687813021704e-06, "loss": 0.3466, "step": 43460 }, { "epoch": 2.17, "learning_rate": 2.7595993322203674e-06, "loss": 0.2836, "step": 43470 }, { "epoch": 2.17, "learning_rate": 2.7579298831385644e-06, "loss": 0.3404, "step": 43480 }, { "epoch": 2.17, "learning_rate": 2.756260434056762e-06, "loss": 0.4046, "step": 43490 }, { "epoch": 2.17, "learning_rate": 2.7545909849749584e-06, "loss": 0.409, "step": 43500 }, { "epoch": 2.18, "learning_rate": 2.7529215358931554e-06, "loss": 0.455, "step": 43510 }, { "epoch": 2.18, "learning_rate": 2.7512520868113524e-06, "loss": 0.4056, "step": 43520 }, { "epoch": 2.18, "learning_rate": 2.7495826377295494e-06, "loss": 0.3466, "step": 43530 }, { "epoch": 2.18, "learning_rate": 2.747913188647747e-06, "loss": 0.2737, "step": 43540 }, { "epoch": 2.18, "learning_rate": 2.7462437395659434e-06, "loss": 0.4022, "step": 43550 }, { "epoch": 2.18, "learning_rate": 2.7445742904841404e-06, "loss": 0.4121, "step": 43560 }, { "epoch": 2.18, "learning_rate": 2.7429048414023374e-06, "loss": 0.3284, "step": 43570 }, { "epoch": 2.18, "learning_rate": 2.7412353923205344e-06, "loss": 0.3682, "step": 43580 }, { "epoch": 2.18, "learning_rate": 2.739565943238732e-06, "loss": 0.4005, "step": 43590 }, { "epoch": 2.18, "learning_rate": 2.737896494156929e-06, "loss": 0.3829, "step": 43600 }, { "epoch": 2.18, "learning_rate": 2.7362270450751254e-06, "loss": 0.3486, "step": 43610 }, { "epoch": 2.18, "learning_rate": 2.7345575959933224e-06, "loss": 0.282, "step": 43620 }, { "epoch": 2.18, "learning_rate": 2.7328881469115194e-06, "loss": 0.3756, "step": 43630 }, { "epoch": 2.18, "learning_rate": 2.7312186978297164e-06, "loss": 0.3418, "step": 43640 }, { "epoch": 2.18, "learning_rate": 2.729549248747914e-06, "loss": 0.3819, "step": 43650 }, { "epoch": 2.18, "learning_rate": 2.7278797996661104e-06, "loss": 0.382, "step": 43660 }, { "epoch": 2.18, "learning_rate": 2.7262103505843074e-06, "loss": 0.3647, "step": 43670 }, { "epoch": 2.18, "learning_rate": 2.7245409015025044e-06, "loss": 0.4046, "step": 43680 }, { "epoch": 2.18, "learning_rate": 2.7228714524207014e-06, "loss": 0.4111, "step": 43690 }, { "epoch": 2.19, "learning_rate": 2.7212020033388988e-06, "loss": 0.3375, "step": 43700 }, { "epoch": 2.19, "learning_rate": 2.7195325542570954e-06, "loss": 0.4052, "step": 43710 }, { "epoch": 2.19, "learning_rate": 2.7178631051752924e-06, "loss": 0.3454, "step": 43720 }, { "epoch": 2.19, "learning_rate": 2.7161936560934894e-06, "loss": 0.3379, "step": 43730 }, { "epoch": 2.19, "learning_rate": 2.7145242070116864e-06, "loss": 0.3579, "step": 43740 }, { "epoch": 2.19, "learning_rate": 2.7128547579298838e-06, "loss": 0.5403, "step": 43750 }, { "epoch": 2.19, "learning_rate": 2.7111853088480803e-06, "loss": 0.3699, "step": 43760 }, { "epoch": 2.19, "learning_rate": 2.7095158597662773e-06, "loss": 0.3879, "step": 43770 }, { "epoch": 2.19, "learning_rate": 2.7078464106844743e-06, "loss": 0.3135, "step": 43780 }, { "epoch": 2.19, "learning_rate": 2.7061769616026713e-06, "loss": 0.3199, "step": 43790 }, { "epoch": 2.19, "learning_rate": 2.704507512520868e-06, "loss": 0.3729, "step": 43800 }, { "epoch": 2.19, "learning_rate": 2.7028380634390653e-06, "loss": 0.4129, "step": 43810 }, { "epoch": 2.19, "learning_rate": 2.7011686143572623e-06, "loss": 0.3556, "step": 43820 }, { "epoch": 2.19, "learning_rate": 2.6994991652754593e-06, "loss": 0.3579, "step": 43830 }, { "epoch": 2.19, "learning_rate": 2.6978297161936563e-06, "loss": 0.4062, "step": 43840 }, { "epoch": 2.19, "learning_rate": 2.696160267111853e-06, "loss": 0.438, "step": 43850 }, { "epoch": 2.19, "learning_rate": 2.6944908180300503e-06, "loss": 0.3392, "step": 43860 }, { "epoch": 2.19, "learning_rate": 2.6928213689482473e-06, "loss": 0.3833, "step": 43870 }, { "epoch": 2.19, "learning_rate": 2.6911519198664443e-06, "loss": 0.4605, "step": 43880 }, { "epoch": 2.19, "learning_rate": 2.6894824707846413e-06, "loss": 0.2752, "step": 43890 }, { "epoch": 2.19, "learning_rate": 2.687813021702838e-06, "loss": 0.3393, "step": 43900 }, { "epoch": 2.2, "learning_rate": 2.6861435726210353e-06, "loss": 0.3319, "step": 43910 }, { "epoch": 2.2, "learning_rate": 2.6844741235392323e-06, "loss": 0.3816, "step": 43920 }, { "epoch": 2.2, "learning_rate": 2.6828046744574293e-06, "loss": 0.3845, "step": 43930 }, { "epoch": 2.2, "learning_rate": 2.6811352253756263e-06, "loss": 0.3935, "step": 43940 }, { "epoch": 2.2, "learning_rate": 2.679465776293823e-06, "loss": 0.4032, "step": 43950 }, { "epoch": 2.2, "learning_rate": 2.67779632721202e-06, "loss": 0.4355, "step": 43960 }, { "epoch": 2.2, "learning_rate": 2.6761268781302173e-06, "loss": 0.3398, "step": 43970 }, { "epoch": 2.2, "learning_rate": 2.6744574290484143e-06, "loss": 0.3846, "step": 43980 }, { "epoch": 2.2, "learning_rate": 2.6727879799666113e-06, "loss": 0.3776, "step": 43990 }, { "epoch": 2.2, "learning_rate": 2.6711185308848083e-06, "loss": 0.3458, "step": 44000 }, { "epoch": 2.2, "learning_rate": 2.669449081803005e-06, "loss": 0.3879, "step": 44010 }, { "epoch": 2.2, "learning_rate": 2.6677796327212023e-06, "loss": 0.3985, "step": 44020 }, { "epoch": 2.2, "learning_rate": 2.6661101836393993e-06, "loss": 0.3804, "step": 44030 }, { "epoch": 2.2, "learning_rate": 2.6644407345575963e-06, "loss": 0.4344, "step": 44040 }, { "epoch": 2.2, "learning_rate": 2.6627712854757933e-06, "loss": 0.3398, "step": 44050 }, { "epoch": 2.2, "learning_rate": 2.66110183639399e-06, "loss": 0.3847, "step": 44060 }, { "epoch": 2.2, "learning_rate": 2.6594323873121873e-06, "loss": 0.3111, "step": 44070 }, { "epoch": 2.2, "learning_rate": 2.6577629382303843e-06, "loss": 0.4486, "step": 44080 }, { "epoch": 2.2, "learning_rate": 2.6560934891485813e-06, "loss": 0.4489, "step": 44090 }, { "epoch": 2.21, "learning_rate": 2.6544240400667783e-06, "loss": 0.3136, "step": 44100 }, { "epoch": 2.21, "learning_rate": 2.652754590984975e-06, "loss": 0.3028, "step": 44110 }, { "epoch": 2.21, "learning_rate": 2.651085141903172e-06, "loss": 0.2735, "step": 44120 }, { "epoch": 2.21, "learning_rate": 2.6494156928213693e-06, "loss": 0.2685, "step": 44130 }, { "epoch": 2.21, "learning_rate": 2.6477462437395663e-06, "loss": 0.3323, "step": 44140 }, { "epoch": 2.21, "learning_rate": 2.6460767946577633e-06, "loss": 0.3339, "step": 44150 }, { "epoch": 2.21, "learning_rate": 2.64440734557596e-06, "loss": 0.331, "step": 44160 }, { "epoch": 2.21, "learning_rate": 2.642737896494157e-06, "loss": 0.386, "step": 44170 }, { "epoch": 2.21, "learning_rate": 2.6410684474123543e-06, "loss": 0.3116, "step": 44180 }, { "epoch": 2.21, "learning_rate": 2.6393989983305513e-06, "loss": 0.3384, "step": 44190 }, { "epoch": 2.21, "learning_rate": 2.6377295492487483e-06, "loss": 0.314, "step": 44200 }, { "epoch": 2.21, "learning_rate": 2.636060100166945e-06, "loss": 0.3807, "step": 44210 }, { "epoch": 2.21, "learning_rate": 2.634390651085142e-06, "loss": 0.5073, "step": 44220 }, { "epoch": 2.21, "learning_rate": 2.6327212020033393e-06, "loss": 0.3721, "step": 44230 }, { "epoch": 2.21, "learning_rate": 2.6310517529215363e-06, "loss": 0.3297, "step": 44240 }, { "epoch": 2.21, "learning_rate": 2.6293823038397333e-06, "loss": 0.3026, "step": 44250 }, { "epoch": 2.21, "learning_rate": 2.62771285475793e-06, "loss": 0.3507, "step": 44260 }, { "epoch": 2.21, "learning_rate": 2.626043405676127e-06, "loss": 0.3063, "step": 44270 }, { "epoch": 2.21, "learning_rate": 2.624373956594324e-06, "loss": 0.334, "step": 44280 }, { "epoch": 2.21, "learning_rate": 2.6227045075125213e-06, "loss": 0.4393, "step": 44290 }, { "epoch": 2.21, "learning_rate": 2.6210350584307183e-06, "loss": 0.3275, "step": 44300 }, { "epoch": 2.22, "learning_rate": 2.619365609348915e-06, "loss": 0.4169, "step": 44310 }, { "epoch": 2.22, "learning_rate": 2.617696160267112e-06, "loss": 0.3667, "step": 44320 }, { "epoch": 2.22, "learning_rate": 2.616026711185309e-06, "loss": 0.3977, "step": 44330 }, { "epoch": 2.22, "learning_rate": 2.6143572621035062e-06, "loss": 0.4483, "step": 44340 }, { "epoch": 2.22, "learning_rate": 2.6126878130217032e-06, "loss": 0.2942, "step": 44350 }, { "epoch": 2.22, "learning_rate": 2.6110183639399002e-06, "loss": 0.3339, "step": 44360 }, { "epoch": 2.22, "learning_rate": 2.609348914858097e-06, "loss": 0.2281, "step": 44370 }, { "epoch": 2.22, "learning_rate": 2.607679465776294e-06, "loss": 0.3914, "step": 44380 }, { "epoch": 2.22, "learning_rate": 2.6060100166944912e-06, "loss": 0.4198, "step": 44390 }, { "epoch": 2.22, "learning_rate": 2.6043405676126882e-06, "loss": 0.44, "step": 44400 }, { "epoch": 2.22, "learning_rate": 2.6026711185308852e-06, "loss": 0.3425, "step": 44410 }, { "epoch": 2.22, "learning_rate": 2.601001669449082e-06, "loss": 0.4168, "step": 44420 }, { "epoch": 2.22, "learning_rate": 2.599332220367279e-06, "loss": 0.3329, "step": 44430 }, { "epoch": 2.22, "learning_rate": 2.5976627712854762e-06, "loss": 0.378, "step": 44440 }, { "epoch": 2.22, "learning_rate": 2.5959933222036732e-06, "loss": 0.3852, "step": 44450 }, { "epoch": 2.22, "learning_rate": 2.5943238731218702e-06, "loss": 0.3773, "step": 44460 }, { "epoch": 2.22, "learning_rate": 2.592654424040067e-06, "loss": 0.3843, "step": 44470 }, { "epoch": 2.22, "learning_rate": 2.590984974958264e-06, "loss": 0.344, "step": 44480 }, { "epoch": 2.22, "learning_rate": 2.589315525876461e-06, "loss": 0.3171, "step": 44490 }, { "epoch": 2.23, "learning_rate": 2.5876460767946582e-06, "loss": 0.32, "step": 44500 }, { "epoch": 2.23, "learning_rate": 2.5859766277128552e-06, "loss": 0.3203, "step": 44510 }, { "epoch": 2.23, "learning_rate": 2.584307178631052e-06, "loss": 0.3485, "step": 44520 }, { "epoch": 2.23, "learning_rate": 2.582637729549249e-06, "loss": 0.3979, "step": 44530 }, { "epoch": 2.23, "learning_rate": 2.580968280467446e-06, "loss": 0.4588, "step": 44540 }, { "epoch": 2.23, "learning_rate": 2.579298831385643e-06, "loss": 0.3395, "step": 44550 }, { "epoch": 2.23, "learning_rate": 2.57762938230384e-06, "loss": 0.3036, "step": 44560 }, { "epoch": 2.23, "learning_rate": 2.5759599332220368e-06, "loss": 0.3524, "step": 44570 }, { "epoch": 2.23, "learning_rate": 2.5742904841402338e-06, "loss": 0.4319, "step": 44580 }, { "epoch": 2.23, "learning_rate": 2.5726210350584308e-06, "loss": 0.2878, "step": 44590 }, { "epoch": 2.23, "learning_rate": 2.570951585976628e-06, "loss": 0.3638, "step": 44600 }, { "epoch": 2.23, "learning_rate": 2.569282136894825e-06, "loss": 0.3376, "step": 44610 }, { "epoch": 2.23, "learning_rate": 2.5676126878130218e-06, "loss": 0.3097, "step": 44620 }, { "epoch": 2.23, "learning_rate": 2.5659432387312188e-06, "loss": 0.3849, "step": 44630 }, { "epoch": 2.23, "learning_rate": 2.5642737896494158e-06, "loss": 0.3259, "step": 44640 }, { "epoch": 2.23, "learning_rate": 2.5626043405676128e-06, "loss": 0.3982, "step": 44650 }, { "epoch": 2.23, "learning_rate": 2.56093489148581e-06, "loss": 0.3247, "step": 44660 }, { "epoch": 2.23, "learning_rate": 2.559265442404007e-06, "loss": 0.3483, "step": 44670 }, { "epoch": 2.23, "learning_rate": 2.5575959933222038e-06, "loss": 0.3864, "step": 44680 }, { "epoch": 2.23, "learning_rate": 2.5559265442404008e-06, "loss": 0.339, "step": 44690 }, { "epoch": 2.23, "learning_rate": 2.5542570951585978e-06, "loss": 0.4737, "step": 44700 }, { "epoch": 2.24, "learning_rate": 2.552587646076795e-06, "loss": 0.3585, "step": 44710 }, { "epoch": 2.24, "learning_rate": 2.550918196994992e-06, "loss": 0.403, "step": 44720 }, { "epoch": 2.24, "learning_rate": 2.5492487479131888e-06, "loss": 0.3404, "step": 44730 }, { "epoch": 2.24, "learning_rate": 2.5475792988313858e-06, "loss": 0.3264, "step": 44740 }, { "epoch": 2.24, "learning_rate": 2.5459098497495828e-06, "loss": 0.4032, "step": 44750 }, { "epoch": 2.24, "learning_rate": 2.54424040066778e-06, "loss": 0.3543, "step": 44760 }, { "epoch": 2.24, "learning_rate": 2.542570951585977e-06, "loss": 0.2468, "step": 44770 }, { "epoch": 2.24, "learning_rate": 2.5409015025041737e-06, "loss": 0.3332, "step": 44780 }, { "epoch": 2.24, "learning_rate": 2.5392320534223707e-06, "loss": 0.3282, "step": 44790 }, { "epoch": 2.24, "learning_rate": 2.5375626043405677e-06, "loss": 0.3599, "step": 44800 }, { "epoch": 2.24, "learning_rate": 2.5358931552587647e-06, "loss": 0.3659, "step": 44810 }, { "epoch": 2.24, "learning_rate": 2.534223706176962e-06, "loss": 0.4105, "step": 44820 }, { "epoch": 2.24, "learning_rate": 2.5325542570951587e-06, "loss": 0.3453, "step": 44830 }, { "epoch": 2.24, "learning_rate": 2.5308848080133557e-06, "loss": 0.3754, "step": 44840 }, { "epoch": 2.24, "learning_rate": 2.5292153589315527e-06, "loss": 0.3806, "step": 44850 }, { "epoch": 2.24, "learning_rate": 2.5275459098497497e-06, "loss": 0.3307, "step": 44860 }, { "epoch": 2.24, "learning_rate": 2.525876460767947e-06, "loss": 0.4681, "step": 44870 }, { "epoch": 2.24, "learning_rate": 2.5242070116861437e-06, "loss": 0.3235, "step": 44880 }, { "epoch": 2.24, "learning_rate": 2.5225375626043407e-06, "loss": 0.3496, "step": 44890 }, { "epoch": 2.25, "learning_rate": 2.5208681135225377e-06, "loss": 0.3933, "step": 44900 }, { "epoch": 2.25, "learning_rate": 2.5191986644407347e-06, "loss": 0.3388, "step": 44910 }, { "epoch": 2.25, "learning_rate": 2.517529215358932e-06, "loss": 0.4104, "step": 44920 }, { "epoch": 2.25, "learning_rate": 2.5158597662771287e-06, "loss": 0.4028, "step": 44930 }, { "epoch": 2.25, "learning_rate": 2.5141903171953257e-06, "loss": 0.3842, "step": 44940 }, { "epoch": 2.25, "learning_rate": 2.5125208681135227e-06, "loss": 0.4117, "step": 44950 }, { "epoch": 2.25, "learning_rate": 2.5108514190317197e-06, "loss": 0.4396, "step": 44960 }, { "epoch": 2.25, "learning_rate": 2.5091819699499163e-06, "loss": 0.4664, "step": 44970 }, { "epoch": 2.25, "learning_rate": 2.5075125208681137e-06, "loss": 0.3772, "step": 44980 }, { "epoch": 2.25, "learning_rate": 2.5058430717863107e-06, "loss": 0.339, "step": 44990 }, { "epoch": 2.25, "learning_rate": 2.5041736227045077e-06, "loss": 0.4598, "step": 45000 }, { "epoch": 2.25, "eval_loss": 0.3745189607143402, "eval_runtime": 182.3652, "eval_samples_per_second": 20.645, "eval_steps_per_second": 2.583, "step": 45000 }, { "epoch": 2.25, "learning_rate": 2.5025041736227047e-06, "loss": 0.4506, "step": 45010 }, { "epoch": 2.25, "learning_rate": 2.5008347245409013e-06, "loss": 0.4668, "step": 45020 }, { "epoch": 2.25, "learning_rate": 2.4991652754590987e-06, "loss": 0.4595, "step": 45030 }, { "epoch": 2.25, "learning_rate": 2.4974958263772957e-06, "loss": 0.3181, "step": 45040 }, { "epoch": 2.25, "learning_rate": 2.4958263772954927e-06, "loss": 0.3924, "step": 45050 }, { "epoch": 2.25, "learning_rate": 2.4941569282136897e-06, "loss": 0.3771, "step": 45060 }, { "epoch": 2.25, "learning_rate": 2.4924874791318867e-06, "loss": 0.3715, "step": 45070 }, { "epoch": 2.25, "learning_rate": 2.4908180300500837e-06, "loss": 0.3399, "step": 45080 }, { "epoch": 2.25, "learning_rate": 2.4891485809682807e-06, "loss": 0.3927, "step": 45090 }, { "epoch": 2.25, "learning_rate": 2.4874791318864777e-06, "loss": 0.45, "step": 45100 }, { "epoch": 2.26, "learning_rate": 2.4858096828046747e-06, "loss": 0.3428, "step": 45110 }, { "epoch": 2.26, "learning_rate": 2.4841402337228717e-06, "loss": 0.3246, "step": 45120 }, { "epoch": 2.26, "learning_rate": 2.4824707846410687e-06, "loss": 0.2775, "step": 45130 }, { "epoch": 2.26, "learning_rate": 2.4808013355592657e-06, "loss": 0.3508, "step": 45140 }, { "epoch": 2.26, "learning_rate": 2.4791318864774627e-06, "loss": 0.3639, "step": 45150 }, { "epoch": 2.26, "learning_rate": 2.4774624373956597e-06, "loss": 0.459, "step": 45160 }, { "epoch": 2.26, "learning_rate": 2.4757929883138567e-06, "loss": 0.4408, "step": 45170 }, { "epoch": 2.26, "learning_rate": 2.4741235392320537e-06, "loss": 0.3167, "step": 45180 }, { "epoch": 2.26, "learning_rate": 2.4724540901502507e-06, "loss": 0.4196, "step": 45190 }, { "epoch": 2.26, "learning_rate": 2.4707846410684477e-06, "loss": 0.4277, "step": 45200 }, { "epoch": 2.26, "learning_rate": 2.4691151919866447e-06, "loss": 0.3444, "step": 45210 }, { "epoch": 2.26, "learning_rate": 2.4674457429048417e-06, "loss": 0.458, "step": 45220 }, { "epoch": 2.26, "learning_rate": 2.4657762938230387e-06, "loss": 0.3117, "step": 45230 }, { "epoch": 2.26, "learning_rate": 2.4641068447412357e-06, "loss": 0.3953, "step": 45240 }, { "epoch": 2.26, "learning_rate": 2.4624373956594327e-06, "loss": 0.4119, "step": 45250 }, { "epoch": 2.26, "learning_rate": 2.4607679465776297e-06, "loss": 0.2971, "step": 45260 }, { "epoch": 2.26, "learning_rate": 2.4590984974958267e-06, "loss": 0.2762, "step": 45270 }, { "epoch": 2.26, "learning_rate": 2.4574290484140237e-06, "loss": 0.3438, "step": 45280 }, { "epoch": 2.26, "learning_rate": 2.4557595993322207e-06, "loss": 0.3563, "step": 45290 }, { "epoch": 2.27, "learning_rate": 2.4540901502504177e-06, "loss": 0.3603, "step": 45300 }, { "epoch": 2.27, "learning_rate": 2.4524207011686147e-06, "loss": 0.3208, "step": 45310 }, { "epoch": 2.27, "learning_rate": 2.4507512520868117e-06, "loss": 0.3309, "step": 45320 }, { "epoch": 2.27, "learning_rate": 2.4490818030050082e-06, "loss": 0.2859, "step": 45330 }, { "epoch": 2.27, "learning_rate": 2.4474123539232056e-06, "loss": 0.3615, "step": 45340 }, { "epoch": 2.27, "learning_rate": 2.4457429048414026e-06, "loss": 0.3961, "step": 45350 }, { "epoch": 2.27, "learning_rate": 2.4440734557595996e-06, "loss": 0.3596, "step": 45360 }, { "epoch": 2.27, "learning_rate": 2.4424040066777966e-06, "loss": 0.2488, "step": 45370 }, { "epoch": 2.27, "learning_rate": 2.4407345575959932e-06, "loss": 0.3465, "step": 45380 }, { "epoch": 2.27, "learning_rate": 2.4390651085141906e-06, "loss": 0.3894, "step": 45390 }, { "epoch": 2.27, "learning_rate": 2.4373956594323876e-06, "loss": 0.3454, "step": 45400 }, { "epoch": 2.27, "learning_rate": 2.4357262103505842e-06, "loss": 0.4105, "step": 45410 }, { "epoch": 2.27, "learning_rate": 2.4340567612687816e-06, "loss": 0.2917, "step": 45420 }, { "epoch": 2.27, "learning_rate": 2.4323873121869786e-06, "loss": 0.3127, "step": 45430 }, { "epoch": 2.27, "learning_rate": 2.4307178631051756e-06, "loss": 0.3777, "step": 45440 }, { "epoch": 2.27, "learning_rate": 2.4290484140233726e-06, "loss": 0.33, "step": 45450 }, { "epoch": 2.27, "learning_rate": 2.427378964941569e-06, "loss": 0.3292, "step": 45460 }, { "epoch": 2.27, "learning_rate": 2.4257095158597666e-06, "loss": 0.4306, "step": 45470 }, { "epoch": 2.27, "learning_rate": 2.4240400667779636e-06, "loss": 0.3456, "step": 45480 }, { "epoch": 2.27, "learning_rate": 2.42237061769616e-06, "loss": 0.3718, "step": 45490 }, { "epoch": 2.27, "learning_rate": 2.4207011686143576e-06, "loss": 0.3505, "step": 45500 }, { "epoch": 2.28, "learning_rate": 2.419031719532554e-06, "loss": 0.4067, "step": 45510 }, { "epoch": 2.28, "learning_rate": 2.4173622704507516e-06, "loss": 0.304, "step": 45520 }, { "epoch": 2.28, "learning_rate": 2.4156928213689486e-06, "loss": 0.3209, "step": 45530 }, { "epoch": 2.28, "learning_rate": 2.414023372287145e-06, "loss": 0.4279, "step": 45540 }, { "epoch": 2.28, "learning_rate": 2.4123539232053426e-06, "loss": 0.3806, "step": 45550 }, { "epoch": 2.28, "learning_rate": 2.410684474123539e-06, "loss": 0.3328, "step": 45560 }, { "epoch": 2.28, "learning_rate": 2.409015025041736e-06, "loss": 0.378, "step": 45570 }, { "epoch": 2.28, "learning_rate": 2.4073455759599336e-06, "loss": 0.2745, "step": 45580 }, { "epoch": 2.28, "learning_rate": 2.40567612687813e-06, "loss": 0.3363, "step": 45590 }, { "epoch": 2.28, "learning_rate": 2.4040066777963276e-06, "loss": 0.432, "step": 45600 }, { "epoch": 2.28, "learning_rate": 2.4023372287145246e-06, "loss": 0.2734, "step": 45610 }, { "epoch": 2.28, "learning_rate": 2.400667779632721e-06, "loss": 0.4066, "step": 45620 }, { "epoch": 2.28, "learning_rate": 2.3989983305509186e-06, "loss": 0.3555, "step": 45630 }, { "epoch": 2.28, "learning_rate": 2.397328881469115e-06, "loss": 0.3117, "step": 45640 }, { "epoch": 2.28, "learning_rate": 2.395659432387312e-06, "loss": 0.3847, "step": 45650 }, { "epoch": 2.28, "learning_rate": 2.3939899833055096e-06, "loss": 0.4151, "step": 45660 }, { "epoch": 2.28, "learning_rate": 2.392320534223706e-06, "loss": 0.396, "step": 45670 }, { "epoch": 2.28, "learning_rate": 2.3906510851419036e-06, "loss": 0.4054, "step": 45680 }, { "epoch": 2.28, "learning_rate": 2.3889816360601e-06, "loss": 0.4271, "step": 45690 }, { "epoch": 2.29, "learning_rate": 2.387312186978297e-06, "loss": 0.3755, "step": 45700 }, { "epoch": 2.29, "learning_rate": 2.3856427378964946e-06, "loss": 0.3218, "step": 45710 }, { "epoch": 2.29, "learning_rate": 2.383973288814691e-06, "loss": 0.4212, "step": 45720 }, { "epoch": 2.29, "learning_rate": 2.3823038397328886e-06, "loss": 0.3017, "step": 45730 }, { "epoch": 2.29, "learning_rate": 2.3806343906510856e-06, "loss": 0.4015, "step": 45740 }, { "epoch": 2.29, "learning_rate": 2.378964941569282e-06, "loss": 0.3546, "step": 45750 }, { "epoch": 2.29, "learning_rate": 2.3772954924874796e-06, "loss": 0.3976, "step": 45760 }, { "epoch": 2.29, "learning_rate": 2.375626043405676e-06, "loss": 0.3829, "step": 45770 }, { "epoch": 2.29, "learning_rate": 2.373956594323873e-06, "loss": 0.3602, "step": 45780 }, { "epoch": 2.29, "learning_rate": 2.3722871452420706e-06, "loss": 0.3763, "step": 45790 }, { "epoch": 2.29, "learning_rate": 2.370617696160267e-06, "loss": 0.3251, "step": 45800 }, { "epoch": 2.29, "learning_rate": 2.3689482470784646e-06, "loss": 0.3741, "step": 45810 }, { "epoch": 2.29, "learning_rate": 2.367278797996661e-06, "loss": 0.4046, "step": 45820 }, { "epoch": 2.29, "learning_rate": 2.365609348914858e-06, "loss": 0.3936, "step": 45830 }, { "epoch": 2.29, "learning_rate": 2.3639398998330556e-06, "loss": 0.3904, "step": 45840 }, { "epoch": 2.29, "learning_rate": 2.362270450751252e-06, "loss": 0.4119, "step": 45850 }, { "epoch": 2.29, "learning_rate": 2.360601001669449e-06, "loss": 0.3211, "step": 45860 }, { "epoch": 2.29, "learning_rate": 2.358931552587646e-06, "loss": 0.3701, "step": 45870 }, { "epoch": 2.29, "learning_rate": 2.357262103505843e-06, "loss": 0.3528, "step": 45880 }, { "epoch": 2.29, "learning_rate": 2.3555926544240406e-06, "loss": 0.3311, "step": 45890 }, { "epoch": 2.29, "learning_rate": 2.353923205342237e-06, "loss": 0.3761, "step": 45900 }, { "epoch": 2.3, "learning_rate": 2.352253756260434e-06, "loss": 0.2954, "step": 45910 }, { "epoch": 2.3, "learning_rate": 2.3505843071786315e-06, "loss": 0.3141, "step": 45920 }, { "epoch": 2.3, "learning_rate": 2.348914858096828e-06, "loss": 0.3622, "step": 45930 }, { "epoch": 2.3, "learning_rate": 2.347245409015025e-06, "loss": 0.3317, "step": 45940 }, { "epoch": 2.3, "learning_rate": 2.345575959933222e-06, "loss": 0.366, "step": 45950 }, { "epoch": 2.3, "learning_rate": 2.343906510851419e-06, "loss": 0.2966, "step": 45960 }, { "epoch": 2.3, "learning_rate": 2.3422370617696165e-06, "loss": 0.2966, "step": 45970 }, { "epoch": 2.3, "learning_rate": 2.340567612687813e-06, "loss": 0.3902, "step": 45980 }, { "epoch": 2.3, "learning_rate": 2.33889816360601e-06, "loss": 0.4318, "step": 45990 }, { "epoch": 2.3, "learning_rate": 2.337228714524207e-06, "loss": 0.3801, "step": 46000 }, { "epoch": 2.3, "learning_rate": 2.335559265442404e-06, "loss": 0.4058, "step": 46010 }, { "epoch": 2.3, "learning_rate": 2.333889816360601e-06, "loss": 0.2955, "step": 46020 }, { "epoch": 2.3, "learning_rate": 2.332220367278798e-06, "loss": 0.4801, "step": 46030 }, { "epoch": 2.3, "learning_rate": 2.330550918196995e-06, "loss": 0.3733, "step": 46040 }, { "epoch": 2.3, "learning_rate": 2.328881469115192e-06, "loss": 0.3869, "step": 46050 }, { "epoch": 2.3, "learning_rate": 2.327212020033389e-06, "loss": 0.3492, "step": 46060 }, { "epoch": 2.3, "learning_rate": 2.325542570951586e-06, "loss": 0.5131, "step": 46070 }, { "epoch": 2.3, "learning_rate": 2.323873121869783e-06, "loss": 0.3577, "step": 46080 }, { "epoch": 2.3, "learning_rate": 2.32220367278798e-06, "loss": 0.3869, "step": 46090 }, { "epoch": 2.31, "learning_rate": 2.320534223706177e-06, "loss": 0.271, "step": 46100 }, { "epoch": 2.31, "learning_rate": 2.318864774624374e-06, "loss": 0.3289, "step": 46110 }, { "epoch": 2.31, "learning_rate": 2.317195325542571e-06, "loss": 0.4383, "step": 46120 }, { "epoch": 2.31, "learning_rate": 2.315525876460768e-06, "loss": 0.3284, "step": 46130 }, { "epoch": 2.31, "learning_rate": 2.313856427378965e-06, "loss": 0.4646, "step": 46140 }, { "epoch": 2.31, "learning_rate": 2.312186978297162e-06, "loss": 0.3274, "step": 46150 }, { "epoch": 2.31, "learning_rate": 2.310517529215359e-06, "loss": 0.3343, "step": 46160 }, { "epoch": 2.31, "learning_rate": 2.308848080133556e-06, "loss": 0.3602, "step": 46170 }, { "epoch": 2.31, "learning_rate": 2.307178631051753e-06, "loss": 0.4936, "step": 46180 }, { "epoch": 2.31, "learning_rate": 2.30550918196995e-06, "loss": 0.3467, "step": 46190 }, { "epoch": 2.31, "learning_rate": 2.303839732888147e-06, "loss": 0.3771, "step": 46200 }, { "epoch": 2.31, "learning_rate": 2.302170283806344e-06, "loss": 0.3746, "step": 46210 }, { "epoch": 2.31, "learning_rate": 2.300500834724541e-06, "loss": 0.3053, "step": 46220 }, { "epoch": 2.31, "learning_rate": 2.298831385642738e-06, "loss": 0.4525, "step": 46230 }, { "epoch": 2.31, "learning_rate": 2.297161936560935e-06, "loss": 0.3454, "step": 46240 }, { "epoch": 2.31, "learning_rate": 2.295492487479132e-06, "loss": 0.4618, "step": 46250 }, { "epoch": 2.31, "learning_rate": 2.293823038397329e-06, "loss": 0.4276, "step": 46260 }, { "epoch": 2.31, "learning_rate": 2.292153589315526e-06, "loss": 0.3289, "step": 46270 }, { "epoch": 2.31, "learning_rate": 2.290484140233723e-06, "loss": 0.3745, "step": 46280 }, { "epoch": 2.31, "learning_rate": 2.28881469115192e-06, "loss": 0.3714, "step": 46290 }, { "epoch": 2.31, "learning_rate": 2.287145242070117e-06, "loss": 0.5002, "step": 46300 }, { "epoch": 2.32, "learning_rate": 2.285475792988314e-06, "loss": 0.4286, "step": 46310 }, { "epoch": 2.32, "learning_rate": 2.283806343906511e-06, "loss": 0.3237, "step": 46320 }, { "epoch": 2.32, "learning_rate": 2.282136894824708e-06, "loss": 0.4209, "step": 46330 }, { "epoch": 2.32, "learning_rate": 2.280467445742905e-06, "loss": 0.3966, "step": 46340 }, { "epoch": 2.32, "learning_rate": 2.278797996661102e-06, "loss": 0.3827, "step": 46350 }, { "epoch": 2.32, "learning_rate": 2.277128547579299e-06, "loss": 0.3976, "step": 46360 }, { "epoch": 2.32, "learning_rate": 2.275459098497496e-06, "loss": 0.4236, "step": 46370 }, { "epoch": 2.32, "learning_rate": 2.273789649415693e-06, "loss": 0.4171, "step": 46380 }, { "epoch": 2.32, "learning_rate": 2.27212020033389e-06, "loss": 0.3714, "step": 46390 }, { "epoch": 2.32, "learning_rate": 2.270450751252087e-06, "loss": 0.4329, "step": 46400 }, { "epoch": 2.32, "learning_rate": 2.268781302170284e-06, "loss": 0.3887, "step": 46410 }, { "epoch": 2.32, "learning_rate": 2.267111853088481e-06, "loss": 0.3517, "step": 46420 }, { "epoch": 2.32, "learning_rate": 2.265442404006678e-06, "loss": 0.3202, "step": 46430 }, { "epoch": 2.32, "learning_rate": 2.263772954924875e-06, "loss": 0.4375, "step": 46440 }, { "epoch": 2.32, "learning_rate": 2.262103505843072e-06, "loss": 0.3599, "step": 46450 }, { "epoch": 2.32, "learning_rate": 2.260434056761269e-06, "loss": 0.3913, "step": 46460 }, { "epoch": 2.32, "learning_rate": 2.258764607679466e-06, "loss": 0.3367, "step": 46470 }, { "epoch": 2.32, "learning_rate": 2.257095158597663e-06, "loss": 0.3163, "step": 46480 }, { "epoch": 2.32, "learning_rate": 2.25542570951586e-06, "loss": 0.2752, "step": 46490 }, { "epoch": 2.33, "learning_rate": 2.253756260434057e-06, "loss": 0.3791, "step": 46500 }, { "epoch": 2.33, "learning_rate": 2.252086811352254e-06, "loss": 0.3539, "step": 46510 }, { "epoch": 2.33, "learning_rate": 2.250417362270451e-06, "loss": 0.349, "step": 46520 }, { "epoch": 2.33, "learning_rate": 2.248747913188648e-06, "loss": 0.3555, "step": 46530 }, { "epoch": 2.33, "learning_rate": 2.247078464106845e-06, "loss": 0.3692, "step": 46540 }, { "epoch": 2.33, "learning_rate": 2.245409015025042e-06, "loss": 0.3089, "step": 46550 }, { "epoch": 2.33, "learning_rate": 2.243739565943239e-06, "loss": 0.3463, "step": 46560 }, { "epoch": 2.33, "learning_rate": 2.242070116861436e-06, "loss": 0.4004, "step": 46570 }, { "epoch": 2.33, "learning_rate": 2.2404006677796326e-06, "loss": 0.3443, "step": 46580 }, { "epoch": 2.33, "learning_rate": 2.23873121869783e-06, "loss": 0.3719, "step": 46590 }, { "epoch": 2.33, "learning_rate": 2.237061769616027e-06, "loss": 0.2937, "step": 46600 }, { "epoch": 2.33, "learning_rate": 2.235392320534224e-06, "loss": 0.3445, "step": 46610 }, { "epoch": 2.33, "learning_rate": 2.233722871452421e-06, "loss": 0.3755, "step": 46620 }, { "epoch": 2.33, "learning_rate": 2.2320534223706176e-06, "loss": 0.2954, "step": 46630 }, { "epoch": 2.33, "learning_rate": 2.230383973288815e-06, "loss": 0.4737, "step": 46640 }, { "epoch": 2.33, "learning_rate": 2.228714524207012e-06, "loss": 0.3914, "step": 46650 }, { "epoch": 2.33, "learning_rate": 2.2270450751252086e-06, "loss": 0.4499, "step": 46660 }, { "epoch": 2.33, "learning_rate": 2.225375626043406e-06, "loss": 0.4929, "step": 46670 }, { "epoch": 2.33, "learning_rate": 2.223706176961603e-06, "loss": 0.3567, "step": 46680 }, { "epoch": 2.33, "learning_rate": 2.2220367278798e-06, "loss": 0.3065, "step": 46690 }, { "epoch": 2.33, "learning_rate": 2.220367278797997e-06, "loss": 0.3024, "step": 46700 }, { "epoch": 2.34, "learning_rate": 2.2186978297161936e-06, "loss": 0.3558, "step": 46710 }, { "epoch": 2.34, "learning_rate": 2.217028380634391e-06, "loss": 0.4319, "step": 46720 }, { "epoch": 2.34, "learning_rate": 2.215358931552588e-06, "loss": 0.3911, "step": 46730 }, { "epoch": 2.34, "learning_rate": 2.2136894824707846e-06, "loss": 0.3641, "step": 46740 }, { "epoch": 2.34, "learning_rate": 2.212020033388982e-06, "loss": 0.3084, "step": 46750 }, { "epoch": 2.34, "learning_rate": 2.2103505843071786e-06, "loss": 0.3903, "step": 46760 }, { "epoch": 2.34, "learning_rate": 2.208681135225376e-06, "loss": 0.3925, "step": 46770 }, { "epoch": 2.34, "learning_rate": 2.207011686143573e-06, "loss": 0.3889, "step": 46780 }, { "epoch": 2.34, "learning_rate": 2.2053422370617695e-06, "loss": 0.4174, "step": 46790 }, { "epoch": 2.34, "learning_rate": 2.203672787979967e-06, "loss": 0.357, "step": 46800 }, { "epoch": 2.34, "learning_rate": 2.2020033388981635e-06, "loss": 0.4327, "step": 46810 }, { "epoch": 2.34, "learning_rate": 2.2003338898163605e-06, "loss": 0.3216, "step": 46820 }, { "epoch": 2.34, "learning_rate": 2.198664440734558e-06, "loss": 0.413, "step": 46830 }, { "epoch": 2.34, "learning_rate": 2.1969949916527545e-06, "loss": 0.3883, "step": 46840 }, { "epoch": 2.34, "learning_rate": 2.195325542570952e-06, "loss": 0.339, "step": 46850 }, { "epoch": 2.34, "learning_rate": 2.193656093489149e-06, "loss": 0.3249, "step": 46860 }, { "epoch": 2.34, "learning_rate": 2.1919866444073455e-06, "loss": 0.3719, "step": 46870 }, { "epoch": 2.34, "learning_rate": 2.190317195325543e-06, "loss": 0.3622, "step": 46880 }, { "epoch": 2.34, "learning_rate": 2.1886477462437395e-06, "loss": 0.4143, "step": 46890 }, { "epoch": 2.34, "learning_rate": 2.1869782971619365e-06, "loss": 0.327, "step": 46900 }, { "epoch": 2.35, "learning_rate": 2.185308848080134e-06, "loss": 0.3981, "step": 46910 }, { "epoch": 2.35, "learning_rate": 2.1836393989983305e-06, "loss": 0.4406, "step": 46920 }, { "epoch": 2.35, "learning_rate": 2.181969949916528e-06, "loss": 0.4449, "step": 46930 }, { "epoch": 2.35, "learning_rate": 2.1803005008347245e-06, "loss": 0.2969, "step": 46940 }, { "epoch": 2.35, "learning_rate": 2.1786310517529215e-06, "loss": 0.3643, "step": 46950 }, { "epoch": 2.35, "learning_rate": 2.176961602671119e-06, "loss": 0.3833, "step": 46960 }, { "epoch": 2.35, "learning_rate": 2.1752921535893155e-06, "loss": 0.3402, "step": 46970 }, { "epoch": 2.35, "learning_rate": 2.173622704507513e-06, "loss": 0.3329, "step": 46980 }, { "epoch": 2.35, "learning_rate": 2.17195325542571e-06, "loss": 0.3076, "step": 46990 }, { "epoch": 2.35, "learning_rate": 2.1702838063439065e-06, "loss": 0.3206, "step": 47000 }, { "epoch": 2.35, "learning_rate": 2.168614357262104e-06, "loss": 0.3751, "step": 47010 }, { "epoch": 2.35, "learning_rate": 2.1669449081803005e-06, "loss": 0.3436, "step": 47020 }, { "epoch": 2.35, "learning_rate": 2.1652754590984975e-06, "loss": 0.4176, "step": 47030 }, { "epoch": 2.35, "learning_rate": 2.163606010016695e-06, "loss": 0.355, "step": 47040 }, { "epoch": 2.35, "learning_rate": 2.1619365609348915e-06, "loss": 0.3582, "step": 47050 }, { "epoch": 2.35, "learning_rate": 2.160267111853089e-06, "loss": 0.3407, "step": 47060 }, { "epoch": 2.35, "learning_rate": 2.1585976627712855e-06, "loss": 0.31, "step": 47070 }, { "epoch": 2.35, "learning_rate": 2.1569282136894825e-06, "loss": 0.4309, "step": 47080 }, { "epoch": 2.35, "learning_rate": 2.15525876460768e-06, "loss": 0.3823, "step": 47090 }, { "epoch": 2.35, "learning_rate": 2.1535893155258765e-06, "loss": 0.3275, "step": 47100 }, { "epoch": 2.36, "learning_rate": 2.1519198664440735e-06, "loss": 0.417, "step": 47110 }, { "epoch": 2.36, "learning_rate": 2.1502504173622705e-06, "loss": 0.4308, "step": 47120 }, { "epoch": 2.36, "learning_rate": 2.1485809682804675e-06, "loss": 0.3912, "step": 47130 }, { "epoch": 2.36, "learning_rate": 2.146911519198665e-06, "loss": 0.2994, "step": 47140 }, { "epoch": 2.36, "learning_rate": 2.1452420701168615e-06, "loss": 0.3701, "step": 47150 }, { "epoch": 2.36, "learning_rate": 2.1435726210350585e-06, "loss": 0.3465, "step": 47160 }, { "epoch": 2.36, "learning_rate": 2.141903171953256e-06, "loss": 0.35, "step": 47170 }, { "epoch": 2.36, "learning_rate": 2.1402337228714525e-06, "loss": 0.434, "step": 47180 }, { "epoch": 2.36, "learning_rate": 2.1385642737896495e-06, "loss": 0.4301, "step": 47190 }, { "epoch": 2.36, "learning_rate": 2.1368948247078465e-06, "loss": 0.3796, "step": 47200 }, { "epoch": 2.36, "learning_rate": 2.1352253756260435e-06, "loss": 0.4218, "step": 47210 }, { "epoch": 2.36, "learning_rate": 2.133555926544241e-06, "loss": 0.355, "step": 47220 }, { "epoch": 2.36, "learning_rate": 2.1318864774624375e-06, "loss": 0.349, "step": 47230 }, { "epoch": 2.36, "learning_rate": 2.1302170283806345e-06, "loss": 0.3618, "step": 47240 }, { "epoch": 2.36, "learning_rate": 2.1285475792988315e-06, "loss": 0.2367, "step": 47250 }, { "epoch": 2.36, "learning_rate": 2.1268781302170285e-06, "loss": 0.3687, "step": 47260 }, { "epoch": 2.36, "learning_rate": 2.1252086811352255e-06, "loss": 0.4233, "step": 47270 }, { "epoch": 2.36, "learning_rate": 2.1235392320534225e-06, "loss": 0.4217, "step": 47280 }, { "epoch": 2.36, "learning_rate": 2.1218697829716195e-06, "loss": 0.441, "step": 47290 }, { "epoch": 2.37, "learning_rate": 2.1202003338898165e-06, "loss": 0.399, "step": 47300 }, { "epoch": 2.37, "learning_rate": 2.1185308848080135e-06, "loss": 0.3412, "step": 47310 }, { "epoch": 2.37, "learning_rate": 2.1168614357262105e-06, "loss": 0.3414, "step": 47320 }, { "epoch": 2.37, "learning_rate": 2.1151919866444075e-06, "loss": 0.3068, "step": 47330 }, { "epoch": 2.37, "learning_rate": 2.1135225375626045e-06, "loss": 0.3083, "step": 47340 }, { "epoch": 2.37, "learning_rate": 2.1118530884808015e-06, "loss": 0.3816, "step": 47350 }, { "epoch": 2.37, "learning_rate": 2.1101836393989984e-06, "loss": 0.4362, "step": 47360 }, { "epoch": 2.37, "learning_rate": 2.1085141903171954e-06, "loss": 0.3604, "step": 47370 }, { "epoch": 2.37, "learning_rate": 2.1068447412353924e-06, "loss": 0.3624, "step": 47380 }, { "epoch": 2.37, "learning_rate": 2.1051752921535894e-06, "loss": 0.4097, "step": 47390 }, { "epoch": 2.37, "learning_rate": 2.1035058430717864e-06, "loss": 0.3825, "step": 47400 }, { "epoch": 2.37, "learning_rate": 2.1018363939899834e-06, "loss": 0.4005, "step": 47410 }, { "epoch": 2.37, "learning_rate": 2.1001669449081804e-06, "loss": 0.2725, "step": 47420 }, { "epoch": 2.37, "learning_rate": 2.0984974958263774e-06, "loss": 0.2825, "step": 47430 }, { "epoch": 2.37, "learning_rate": 2.0968280467445744e-06, "loss": 0.3526, "step": 47440 }, { "epoch": 2.37, "learning_rate": 2.0951585976627714e-06, "loss": 0.393, "step": 47450 }, { "epoch": 2.37, "learning_rate": 2.0934891485809684e-06, "loss": 0.3374, "step": 47460 }, { "epoch": 2.37, "learning_rate": 2.0918196994991654e-06, "loss": 0.3523, "step": 47470 }, { "epoch": 2.37, "learning_rate": 2.0901502504173624e-06, "loss": 0.4068, "step": 47480 }, { "epoch": 2.37, "learning_rate": 2.0884808013355594e-06, "loss": 0.3218, "step": 47490 }, { "epoch": 2.38, "learning_rate": 2.0868113522537564e-06, "loss": 0.3801, "step": 47500 }, { "epoch": 2.38, "learning_rate": 2.0851419031719534e-06, "loss": 0.3294, "step": 47510 }, { "epoch": 2.38, "learning_rate": 2.0834724540901504e-06, "loss": 0.4012, "step": 47520 }, { "epoch": 2.38, "learning_rate": 2.0818030050083474e-06, "loss": 0.3187, "step": 47530 }, { "epoch": 2.38, "learning_rate": 2.0801335559265444e-06, "loss": 0.4256, "step": 47540 }, { "epoch": 2.38, "learning_rate": 2.0784641068447414e-06, "loss": 0.3218, "step": 47550 }, { "epoch": 2.38, "learning_rate": 2.0767946577629384e-06, "loss": 0.3311, "step": 47560 }, { "epoch": 2.38, "learning_rate": 2.0751252086811354e-06, "loss": 0.3498, "step": 47570 }, { "epoch": 2.38, "learning_rate": 2.0734557595993324e-06, "loss": 0.3682, "step": 47580 }, { "epoch": 2.38, "learning_rate": 2.0717863105175294e-06, "loss": 0.3797, "step": 47590 }, { "epoch": 2.38, "learning_rate": 2.0701168614357264e-06, "loss": 0.3043, "step": 47600 }, { "epoch": 2.38, "learning_rate": 2.0684474123539234e-06, "loss": 0.4053, "step": 47610 }, { "epoch": 2.38, "learning_rate": 2.0667779632721204e-06, "loss": 0.359, "step": 47620 }, { "epoch": 2.38, "learning_rate": 2.0651085141903174e-06, "loss": 0.4948, "step": 47630 }, { "epoch": 2.38, "learning_rate": 2.0634390651085144e-06, "loss": 0.4224, "step": 47640 }, { "epoch": 2.38, "learning_rate": 2.0617696160267114e-06, "loss": 0.377, "step": 47650 }, { "epoch": 2.38, "learning_rate": 2.0601001669449084e-06, "loss": 0.3923, "step": 47660 }, { "epoch": 2.38, "learning_rate": 2.0584307178631054e-06, "loss": 0.3819, "step": 47670 }, { "epoch": 2.38, "learning_rate": 2.0567612687813024e-06, "loss": 0.3081, "step": 47680 }, { "epoch": 2.38, "learning_rate": 2.0550918196994994e-06, "loss": 0.3325, "step": 47690 }, { "epoch": 2.38, "learning_rate": 2.0534223706176964e-06, "loss": 0.3731, "step": 47700 }, { "epoch": 2.39, "learning_rate": 2.0517529215358934e-06, "loss": 0.4478, "step": 47710 }, { "epoch": 2.39, "learning_rate": 2.0500834724540904e-06, "loss": 0.3571, "step": 47720 }, { "epoch": 2.39, "learning_rate": 2.0484140233722874e-06, "loss": 0.4058, "step": 47730 }, { "epoch": 2.39, "learning_rate": 2.0467445742904844e-06, "loss": 0.3829, "step": 47740 }, { "epoch": 2.39, "learning_rate": 2.0450751252086814e-06, "loss": 0.3192, "step": 47750 }, { "epoch": 2.39, "learning_rate": 2.0434056761268784e-06, "loss": 0.3225, "step": 47760 }, { "epoch": 2.39, "learning_rate": 2.0417362270450754e-06, "loss": 0.3733, "step": 47770 }, { "epoch": 2.39, "learning_rate": 2.0400667779632724e-06, "loss": 0.3888, "step": 47780 }, { "epoch": 2.39, "learning_rate": 2.0383973288814694e-06, "loss": 0.3995, "step": 47790 }, { "epoch": 2.39, "learning_rate": 2.0367278797996664e-06, "loss": 0.3092, "step": 47800 }, { "epoch": 2.39, "learning_rate": 2.0350584307178634e-06, "loss": 0.3575, "step": 47810 }, { "epoch": 2.39, "learning_rate": 2.0333889816360604e-06, "loss": 0.3848, "step": 47820 }, { "epoch": 2.39, "learning_rate": 2.031719532554257e-06, "loss": 0.3711, "step": 47830 }, { "epoch": 2.39, "learning_rate": 2.0300500834724544e-06, "loss": 0.3178, "step": 47840 }, { "epoch": 2.39, "learning_rate": 2.0283806343906514e-06, "loss": 0.3257, "step": 47850 }, { "epoch": 2.39, "learning_rate": 2.0267111853088484e-06, "loss": 0.3531, "step": 47860 }, { "epoch": 2.39, "learning_rate": 2.0250417362270454e-06, "loss": 0.4005, "step": 47870 }, { "epoch": 2.39, "learning_rate": 2.023372287145242e-06, "loss": 0.4227, "step": 47880 }, { "epoch": 2.39, "learning_rate": 2.0217028380634394e-06, "loss": 0.3742, "step": 47890 }, { "epoch": 2.4, "learning_rate": 2.0200333889816364e-06, "loss": 0.3498, "step": 47900 }, { "epoch": 2.4, "learning_rate": 2.018363939899833e-06, "loss": 0.3902, "step": 47910 }, { "epoch": 2.4, "learning_rate": 2.0166944908180304e-06, "loss": 0.3463, "step": 47920 }, { "epoch": 2.4, "learning_rate": 2.0150250417362273e-06, "loss": 0.3442, "step": 47930 }, { "epoch": 2.4, "learning_rate": 2.0133555926544243e-06, "loss": 0.3299, "step": 47940 }, { "epoch": 2.4, "learning_rate": 2.0116861435726213e-06, "loss": 0.3877, "step": 47950 }, { "epoch": 2.4, "learning_rate": 2.010016694490818e-06, "loss": 0.3852, "step": 47960 }, { "epoch": 2.4, "learning_rate": 2.0083472454090153e-06, "loss": 0.3577, "step": 47970 }, { "epoch": 2.4, "learning_rate": 2.0066777963272123e-06, "loss": 0.4748, "step": 47980 }, { "epoch": 2.4, "learning_rate": 2.005008347245409e-06, "loss": 0.4588, "step": 47990 }, { "epoch": 2.4, "learning_rate": 2.0033388981636063e-06, "loss": 0.3782, "step": 48000 }, { "epoch": 2.4, "learning_rate": 2.001669449081803e-06, "loss": 0.4388, "step": 48010 }, { "epoch": 2.4, "learning_rate": 2.0000000000000003e-06, "loss": 0.2831, "step": 48020 }, { "epoch": 2.4, "learning_rate": 1.9983305509181973e-06, "loss": 0.4785, "step": 48030 }, { "epoch": 2.4, "learning_rate": 1.996661101836394e-06, "loss": 0.3224, "step": 48040 }, { "epoch": 2.4, "learning_rate": 1.9949916527545913e-06, "loss": 0.3347, "step": 48050 }, { "epoch": 2.4, "learning_rate": 1.993322203672788e-06, "loss": 0.3017, "step": 48060 }, { "epoch": 2.4, "learning_rate": 1.991652754590985e-06, "loss": 0.3503, "step": 48070 }, { "epoch": 2.4, "learning_rate": 1.9899833055091823e-06, "loss": 0.4232, "step": 48080 }, { "epoch": 2.4, "learning_rate": 1.988313856427379e-06, "loss": 0.3675, "step": 48090 }, { "epoch": 2.41, "learning_rate": 1.9866444073455763e-06, "loss": 0.3484, "step": 48100 }, { "epoch": 2.41, "learning_rate": 1.9849749582637733e-06, "loss": 0.393, "step": 48110 }, { "epoch": 2.41, "learning_rate": 1.98330550918197e-06, "loss": 0.3636, "step": 48120 }, { "epoch": 2.41, "learning_rate": 1.9816360601001673e-06, "loss": 0.2505, "step": 48130 }, { "epoch": 2.41, "learning_rate": 1.979966611018364e-06, "loss": 0.357, "step": 48140 }, { "epoch": 2.41, "learning_rate": 1.978297161936561e-06, "loss": 0.2776, "step": 48150 }, { "epoch": 2.41, "learning_rate": 1.9766277128547583e-06, "loss": 0.3204, "step": 48160 }, { "epoch": 2.41, "learning_rate": 1.974958263772955e-06, "loss": 0.4339, "step": 48170 }, { "epoch": 2.41, "learning_rate": 1.9732888146911523e-06, "loss": 0.3838, "step": 48180 }, { "epoch": 2.41, "learning_rate": 1.971619365609349e-06, "loss": 0.414, "step": 48190 }, { "epoch": 2.41, "learning_rate": 1.969949916527546e-06, "loss": 0.3896, "step": 48200 }, { "epoch": 2.41, "learning_rate": 1.9682804674457433e-06, "loss": 0.4654, "step": 48210 }, { "epoch": 2.41, "learning_rate": 1.96661101836394e-06, "loss": 0.3231, "step": 48220 }, { "epoch": 2.41, "learning_rate": 1.964941569282137e-06, "loss": 0.3809, "step": 48230 }, { "epoch": 2.41, "learning_rate": 1.9632721202003343e-06, "loss": 0.3872, "step": 48240 }, { "epoch": 2.41, "learning_rate": 1.961602671118531e-06, "loss": 0.3549, "step": 48250 }, { "epoch": 2.41, "learning_rate": 1.9599332220367283e-06, "loss": 0.3532, "step": 48260 }, { "epoch": 2.41, "learning_rate": 1.958263772954925e-06, "loss": 0.3206, "step": 48270 }, { "epoch": 2.41, "learning_rate": 1.956594323873122e-06, "loss": 0.3051, "step": 48280 }, { "epoch": 2.41, "learning_rate": 1.9549248747913193e-06, "loss": 0.4461, "step": 48290 }, { "epoch": 2.42, "learning_rate": 1.953255425709516e-06, "loss": 0.3146, "step": 48300 }, { "epoch": 2.42, "learning_rate": 1.9515859766277133e-06, "loss": 0.3279, "step": 48310 }, { "epoch": 2.42, "learning_rate": 1.94991652754591e-06, "loss": 0.3266, "step": 48320 }, { "epoch": 2.42, "learning_rate": 1.948247078464107e-06, "loss": 0.2914, "step": 48330 }, { "epoch": 2.42, "learning_rate": 1.9465776293823043e-06, "loss": 0.4122, "step": 48340 }, { "epoch": 2.42, "learning_rate": 1.944908180300501e-06, "loss": 0.3829, "step": 48350 }, { "epoch": 2.42, "learning_rate": 1.943238731218698e-06, "loss": 0.3788, "step": 48360 }, { "epoch": 2.42, "learning_rate": 1.941569282136895e-06, "loss": 0.3767, "step": 48370 }, { "epoch": 2.42, "learning_rate": 1.939899833055092e-06, "loss": 0.3421, "step": 48380 }, { "epoch": 2.42, "learning_rate": 1.9382303839732893e-06, "loss": 0.324, "step": 48390 }, { "epoch": 2.42, "learning_rate": 1.936560934891486e-06, "loss": 0.3939, "step": 48400 }, { "epoch": 2.42, "learning_rate": 1.934891485809683e-06, "loss": 0.3104, "step": 48410 }, { "epoch": 2.42, "learning_rate": 1.9332220367278803e-06, "loss": 0.3386, "step": 48420 }, { "epoch": 2.42, "learning_rate": 1.931552587646077e-06, "loss": 0.4073, "step": 48430 }, { "epoch": 2.42, "learning_rate": 1.929883138564274e-06, "loss": 0.3187, "step": 48440 }, { "epoch": 2.42, "learning_rate": 1.928213689482471e-06, "loss": 0.4059, "step": 48450 }, { "epoch": 2.42, "learning_rate": 1.926544240400668e-06, "loss": 0.3081, "step": 48460 }, { "epoch": 2.42, "learning_rate": 1.9248747913188653e-06, "loss": 0.2977, "step": 48470 }, { "epoch": 2.42, "learning_rate": 1.923205342237062e-06, "loss": 0.4478, "step": 48480 }, { "epoch": 2.42, "learning_rate": 1.921535893155259e-06, "loss": 0.4606, "step": 48490 }, { "epoch": 2.42, "learning_rate": 1.919866444073456e-06, "loss": 0.3482, "step": 48500 }, { "epoch": 2.43, "learning_rate": 1.918196994991653e-06, "loss": 0.3259, "step": 48510 }, { "epoch": 2.43, "learning_rate": 1.91652754590985e-06, "loss": 0.3759, "step": 48520 }, { "epoch": 2.43, "learning_rate": 1.914858096828047e-06, "loss": 0.3999, "step": 48530 }, { "epoch": 2.43, "learning_rate": 1.913188647746244e-06, "loss": 0.3696, "step": 48540 }, { "epoch": 2.43, "learning_rate": 1.911519198664441e-06, "loss": 0.3377, "step": 48550 }, { "epoch": 2.43, "learning_rate": 1.909849749582638e-06, "loss": 0.3734, "step": 48560 }, { "epoch": 2.43, "learning_rate": 1.908180300500835e-06, "loss": 0.3223, "step": 48570 }, { "epoch": 2.43, "learning_rate": 1.906510851419032e-06, "loss": 0.3867, "step": 48580 }, { "epoch": 2.43, "learning_rate": 1.9048414023372288e-06, "loss": 0.3492, "step": 48590 }, { "epoch": 2.43, "learning_rate": 1.9031719532554258e-06, "loss": 0.3827, "step": 48600 }, { "epoch": 2.43, "learning_rate": 1.901502504173623e-06, "loss": 0.3422, "step": 48610 }, { "epoch": 2.43, "learning_rate": 1.8998330550918198e-06, "loss": 0.3877, "step": 48620 }, { "epoch": 2.43, "learning_rate": 1.898163606010017e-06, "loss": 0.4443, "step": 48630 }, { "epoch": 2.43, "learning_rate": 1.8964941569282138e-06, "loss": 0.3656, "step": 48640 }, { "epoch": 2.43, "learning_rate": 1.8948247078464108e-06, "loss": 0.4167, "step": 48650 }, { "epoch": 2.43, "learning_rate": 1.893155258764608e-06, "loss": 0.3619, "step": 48660 }, { "epoch": 2.43, "learning_rate": 1.8914858096828048e-06, "loss": 0.3296, "step": 48670 }, { "epoch": 2.43, "learning_rate": 1.8898163606010018e-06, "loss": 0.4429, "step": 48680 }, { "epoch": 2.43, "learning_rate": 1.8881469115191988e-06, "loss": 0.4278, "step": 48690 }, { "epoch": 2.44, "learning_rate": 1.8864774624373958e-06, "loss": 0.3408, "step": 48700 }, { "epoch": 2.44, "learning_rate": 1.884808013355593e-06, "loss": 0.2866, "step": 48710 }, { "epoch": 2.44, "learning_rate": 1.8831385642737898e-06, "loss": 0.374, "step": 48720 }, { "epoch": 2.44, "learning_rate": 1.8814691151919868e-06, "loss": 0.3199, "step": 48730 }, { "epoch": 2.44, "learning_rate": 1.8797996661101838e-06, "loss": 0.4398, "step": 48740 }, { "epoch": 2.44, "learning_rate": 1.8781302170283808e-06, "loss": 0.4559, "step": 48750 }, { "epoch": 2.44, "learning_rate": 1.8764607679465776e-06, "loss": 0.2872, "step": 48760 }, { "epoch": 2.44, "learning_rate": 1.8747913188647748e-06, "loss": 0.3687, "step": 48770 }, { "epoch": 2.44, "learning_rate": 1.8731218697829718e-06, "loss": 0.2803, "step": 48780 }, { "epoch": 2.44, "learning_rate": 1.871452420701169e-06, "loss": 0.3646, "step": 48790 }, { "epoch": 2.44, "learning_rate": 1.8697829716193658e-06, "loss": 0.2667, "step": 48800 }, { "epoch": 2.44, "learning_rate": 1.8681135225375628e-06, "loss": 0.4243, "step": 48810 }, { "epoch": 2.44, "learning_rate": 1.8664440734557598e-06, "loss": 0.343, "step": 48820 }, { "epoch": 2.44, "learning_rate": 1.8647746243739568e-06, "loss": 0.4212, "step": 48830 }, { "epoch": 2.44, "learning_rate": 1.8631051752921536e-06, "loss": 0.3476, "step": 48840 }, { "epoch": 2.44, "learning_rate": 1.8614357262103508e-06, "loss": 0.3093, "step": 48850 }, { "epoch": 2.44, "learning_rate": 1.8597662771285478e-06, "loss": 0.3886, "step": 48860 }, { "epoch": 2.44, "learning_rate": 1.8580968280467448e-06, "loss": 0.3786, "step": 48870 }, { "epoch": 2.44, "learning_rate": 1.8564273789649418e-06, "loss": 0.3793, "step": 48880 }, { "epoch": 2.44, "learning_rate": 1.8547579298831385e-06, "loss": 0.3722, "step": 48890 }, { "epoch": 2.44, "learning_rate": 1.8530884808013358e-06, "loss": 0.3781, "step": 48900 }, { "epoch": 2.45, "learning_rate": 1.8514190317195328e-06, "loss": 0.404, "step": 48910 }, { "epoch": 2.45, "learning_rate": 1.8497495826377295e-06, "loss": 0.3541, "step": 48920 }, { "epoch": 2.45, "learning_rate": 1.8480801335559268e-06, "loss": 0.3175, "step": 48930 }, { "epoch": 2.45, "learning_rate": 1.8464106844741235e-06, "loss": 0.325, "step": 48940 }, { "epoch": 2.45, "learning_rate": 1.8447412353923207e-06, "loss": 0.4151, "step": 48950 }, { "epoch": 2.45, "learning_rate": 1.8430717863105177e-06, "loss": 0.3893, "step": 48960 }, { "epoch": 2.45, "learning_rate": 1.8414023372287145e-06, "loss": 0.4846, "step": 48970 }, { "epoch": 2.45, "learning_rate": 1.8397328881469117e-06, "loss": 0.3917, "step": 48980 }, { "epoch": 2.45, "learning_rate": 1.8380634390651087e-06, "loss": 0.3674, "step": 48990 }, { "epoch": 2.45, "learning_rate": 1.8363939899833055e-06, "loss": 0.3855, "step": 49000 }, { "epoch": 2.45, "learning_rate": 1.8347245409015027e-06, "loss": 0.3941, "step": 49010 }, { "epoch": 2.45, "learning_rate": 1.8330550918196995e-06, "loss": 0.3749, "step": 49020 }, { "epoch": 2.45, "learning_rate": 1.8313856427378967e-06, "loss": 0.3602, "step": 49030 }, { "epoch": 2.45, "learning_rate": 1.8297161936560937e-06, "loss": 0.4091, "step": 49040 }, { "epoch": 2.45, "learning_rate": 1.8280467445742905e-06, "loss": 0.3901, "step": 49050 }, { "epoch": 2.45, "learning_rate": 1.8263772954924877e-06, "loss": 0.3821, "step": 49060 }, { "epoch": 2.45, "learning_rate": 1.8247078464106845e-06, "loss": 0.2714, "step": 49070 }, { "epoch": 2.45, "learning_rate": 1.8230383973288815e-06, "loss": 0.2819, "step": 49080 }, { "epoch": 2.45, "learning_rate": 1.8213689482470787e-06, "loss": 0.3695, "step": 49090 }, { "epoch": 2.46, "learning_rate": 1.8196994991652755e-06, "loss": 0.3878, "step": 49100 }, { "epoch": 2.46, "learning_rate": 1.8180300500834727e-06, "loss": 0.3087, "step": 49110 }, { "epoch": 2.46, "learning_rate": 1.8163606010016695e-06, "loss": 0.3764, "step": 49120 }, { "epoch": 2.46, "learning_rate": 1.8146911519198665e-06, "loss": 0.3269, "step": 49130 }, { "epoch": 2.46, "learning_rate": 1.8130217028380637e-06, "loss": 0.3251, "step": 49140 }, { "epoch": 2.46, "learning_rate": 1.8113522537562605e-06, "loss": 0.3531, "step": 49150 }, { "epoch": 2.46, "learning_rate": 1.8096828046744575e-06, "loss": 0.4266, "step": 49160 }, { "epoch": 2.46, "learning_rate": 1.8080133555926547e-06, "loss": 0.3624, "step": 49170 }, { "epoch": 2.46, "learning_rate": 1.8063439065108515e-06, "loss": 0.2896, "step": 49180 }, { "epoch": 2.46, "learning_rate": 1.8046744574290487e-06, "loss": 0.3372, "step": 49190 }, { "epoch": 2.46, "learning_rate": 1.8030050083472455e-06, "loss": 0.4119, "step": 49200 }, { "epoch": 2.46, "learning_rate": 1.8013355592654425e-06, "loss": 0.3281, "step": 49210 }, { "epoch": 2.46, "learning_rate": 1.7996661101836397e-06, "loss": 0.3247, "step": 49220 }, { "epoch": 2.46, "learning_rate": 1.7979966611018365e-06, "loss": 0.3394, "step": 49230 }, { "epoch": 2.46, "learning_rate": 1.7963272120200335e-06, "loss": 0.3249, "step": 49240 }, { "epoch": 2.46, "learning_rate": 1.7946577629382305e-06, "loss": 0.287, "step": 49250 }, { "epoch": 2.46, "learning_rate": 1.7929883138564275e-06, "loss": 0.3175, "step": 49260 }, { "epoch": 2.46, "learning_rate": 1.7913188647746247e-06, "loss": 0.4063, "step": 49270 }, { "epoch": 2.46, "learning_rate": 1.7896494156928215e-06, "loss": 0.418, "step": 49280 }, { "epoch": 2.46, "learning_rate": 1.7879799666110185e-06, "loss": 0.3132, "step": 49290 }, { "epoch": 2.46, "learning_rate": 1.7863105175292157e-06, "loss": 0.2608, "step": 49300 }, { "epoch": 2.47, "learning_rate": 1.7846410684474125e-06, "loss": 0.4398, "step": 49310 }, { "epoch": 2.47, "learning_rate": 1.7829716193656093e-06, "loss": 0.3249, "step": 49320 }, { "epoch": 2.47, "learning_rate": 1.7813021702838065e-06, "loss": 0.309, "step": 49330 }, { "epoch": 2.47, "learning_rate": 1.7796327212020035e-06, "loss": 0.377, "step": 49340 }, { "epoch": 2.47, "learning_rate": 1.7779632721202007e-06, "loss": 0.3422, "step": 49350 }, { "epoch": 2.47, "learning_rate": 1.7762938230383975e-06, "loss": 0.4008, "step": 49360 }, { "epoch": 2.47, "learning_rate": 1.7746243739565945e-06, "loss": 0.3185, "step": 49370 }, { "epoch": 2.47, "learning_rate": 1.7729549248747915e-06, "loss": 0.3172, "step": 49380 }, { "epoch": 2.47, "learning_rate": 1.7712854757929885e-06, "loss": 0.3373, "step": 49390 }, { "epoch": 2.47, "learning_rate": 1.7696160267111852e-06, "loss": 0.3828, "step": 49400 }, { "epoch": 2.47, "learning_rate": 1.7679465776293825e-06, "loss": 0.2485, "step": 49410 }, { "epoch": 2.47, "learning_rate": 1.7662771285475795e-06, "loss": 0.3255, "step": 49420 }, { "epoch": 2.47, "learning_rate": 1.7646076794657765e-06, "loss": 0.3921, "step": 49430 }, { "epoch": 2.47, "learning_rate": 1.7629382303839734e-06, "loss": 0.4695, "step": 49440 }, { "epoch": 2.47, "learning_rate": 1.7612687813021702e-06, "loss": 0.5238, "step": 49450 }, { "epoch": 2.47, "learning_rate": 1.7595993322203674e-06, "loss": 0.2974, "step": 49460 }, { "epoch": 2.47, "learning_rate": 1.7579298831385644e-06, "loss": 0.2623, "step": 49470 }, { "epoch": 2.47, "learning_rate": 1.7562604340567612e-06, "loss": 0.2987, "step": 49480 }, { "epoch": 2.47, "learning_rate": 1.7545909849749584e-06, "loss": 0.3658, "step": 49490 }, { "epoch": 2.48, "learning_rate": 1.7529215358931554e-06, "loss": 0.4273, "step": 49500 }, { "epoch": 2.48, "learning_rate": 1.7512520868113524e-06, "loss": 0.3683, "step": 49510 }, { "epoch": 2.48, "learning_rate": 1.7495826377295494e-06, "loss": 0.4391, "step": 49520 }, { "epoch": 2.48, "learning_rate": 1.7479131886477462e-06, "loss": 0.3879, "step": 49530 }, { "epoch": 2.48, "learning_rate": 1.7462437395659434e-06, "loss": 0.436, "step": 49540 }, { "epoch": 2.48, "learning_rate": 1.7445742904841404e-06, "loss": 0.2855, "step": 49550 }, { "epoch": 2.48, "learning_rate": 1.7429048414023372e-06, "loss": 0.3974, "step": 49560 }, { "epoch": 2.48, "learning_rate": 1.7412353923205344e-06, "loss": 0.2895, "step": 49570 }, { "epoch": 2.48, "learning_rate": 1.7395659432387312e-06, "loss": 0.3588, "step": 49580 }, { "epoch": 2.48, "learning_rate": 1.7378964941569284e-06, "loss": 0.2481, "step": 49590 }, { "epoch": 2.48, "learning_rate": 1.7362270450751254e-06, "loss": 0.3021, "step": 49600 }, { "epoch": 2.48, "learning_rate": 1.7345575959933222e-06, "loss": 0.441, "step": 49610 }, { "epoch": 2.48, "learning_rate": 1.7328881469115194e-06, "loss": 0.3896, "step": 49620 }, { "epoch": 2.48, "learning_rate": 1.7312186978297162e-06, "loss": 0.2963, "step": 49630 }, { "epoch": 2.48, "learning_rate": 1.7295492487479134e-06, "loss": 0.3807, "step": 49640 }, { "epoch": 2.48, "learning_rate": 1.7278797996661104e-06, "loss": 0.2872, "step": 49650 }, { "epoch": 2.48, "learning_rate": 1.7262103505843072e-06, "loss": 0.3545, "step": 49660 }, { "epoch": 2.48, "learning_rate": 1.7245409015025044e-06, "loss": 0.3774, "step": 49670 }, { "epoch": 2.48, "learning_rate": 1.7228714524207014e-06, "loss": 0.429, "step": 49680 }, { "epoch": 2.48, "learning_rate": 1.7212020033388982e-06, "loss": 0.4297, "step": 49690 }, { "epoch": 2.48, "learning_rate": 1.7195325542570954e-06, "loss": 0.4035, "step": 49700 }, { "epoch": 2.49, "learning_rate": 1.7178631051752922e-06, "loss": 0.3453, "step": 49710 }, { "epoch": 2.49, "learning_rate": 1.7161936560934894e-06, "loss": 0.3653, "step": 49720 }, { "epoch": 2.49, "learning_rate": 1.7145242070116864e-06, "loss": 0.3513, "step": 49730 }, { "epoch": 2.49, "learning_rate": 1.7128547579298832e-06, "loss": 0.3728, "step": 49740 }, { "epoch": 2.49, "learning_rate": 1.7111853088480804e-06, "loss": 0.2853, "step": 49750 }, { "epoch": 2.49, "learning_rate": 1.7095158597662772e-06, "loss": 0.4999, "step": 49760 }, { "epoch": 2.49, "learning_rate": 1.7078464106844742e-06, "loss": 0.3584, "step": 49770 }, { "epoch": 2.49, "learning_rate": 1.7061769616026714e-06, "loss": 0.371, "step": 49780 }, { "epoch": 2.49, "learning_rate": 1.7045075125208682e-06, "loss": 0.4359, "step": 49790 }, { "epoch": 2.49, "learning_rate": 1.7028380634390654e-06, "loss": 0.3316, "step": 49800 }, { "epoch": 2.49, "learning_rate": 1.7011686143572622e-06, "loss": 0.3066, "step": 49810 }, { "epoch": 2.49, "learning_rate": 1.6994991652754592e-06, "loss": 0.3352, "step": 49820 }, { "epoch": 2.49, "learning_rate": 1.6978297161936564e-06, "loss": 0.3108, "step": 49830 }, { "epoch": 2.49, "learning_rate": 1.6961602671118532e-06, "loss": 0.3585, "step": 49840 }, { "epoch": 2.49, "learning_rate": 1.6944908180300502e-06, "loss": 0.3211, "step": 49850 }, { "epoch": 2.49, "learning_rate": 1.6928213689482474e-06, "loss": 0.4737, "step": 49860 }, { "epoch": 2.49, "learning_rate": 1.6911519198664442e-06, "loss": 0.4432, "step": 49870 }, { "epoch": 2.49, "learning_rate": 1.6894824707846414e-06, "loss": 0.3961, "step": 49880 }, { "epoch": 2.49, "learning_rate": 1.6878130217028382e-06, "loss": 0.3944, "step": 49890 }, { "epoch": 2.5, "learning_rate": 1.6861435726210352e-06, "loss": 0.3401, "step": 49900 }, { "epoch": 2.5, "learning_rate": 1.6844741235392324e-06, "loss": 0.3606, "step": 49910 }, { "epoch": 2.5, "learning_rate": 1.6828046744574292e-06, "loss": 0.3349, "step": 49920 }, { "epoch": 2.5, "learning_rate": 1.6811352253756262e-06, "loss": 0.3151, "step": 49930 }, { "epoch": 2.5, "learning_rate": 1.6794657762938231e-06, "loss": 0.3671, "step": 49940 }, { "epoch": 2.5, "learning_rate": 1.6777963272120201e-06, "loss": 0.4161, "step": 49950 }, { "epoch": 2.5, "learning_rate": 1.6761268781302174e-06, "loss": 0.3138, "step": 49960 }, { "epoch": 2.5, "learning_rate": 1.6744574290484141e-06, "loss": 0.3525, "step": 49970 }, { "epoch": 2.5, "learning_rate": 1.6727879799666111e-06, "loss": 0.3713, "step": 49980 }, { "epoch": 2.5, "learning_rate": 1.6711185308848081e-06, "loss": 0.2957, "step": 49990 }, { "epoch": 2.5, "learning_rate": 1.6694490818030051e-06, "loss": 0.4227, "step": 50000 }, { "epoch": 2.5, "eval_loss": 0.3743310570716858, "eval_runtime": 179.9929, "eval_samples_per_second": 20.917, "eval_steps_per_second": 2.617, "step": 50000 } ], "max_steps": 60000, "num_train_epochs": 3, "total_flos": 2.739029409792e+17, "trial_name": null, "trial_params": null }