{ "best_metric": null, "best_model_checkpoint": null, "epoch": 4.0, "eval_steps": 500, "global_step": 9824, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 4.999987217003186e-05, "loss": 0.734, "step": 10 }, { "epoch": 0.01, "learning_rate": 4.999948868143466e-05, "loss": 0.2527, "step": 20 }, { "epoch": 0.01, "learning_rate": 4.999884953813012e-05, "loss": 0.1508, "step": 30 }, { "epoch": 0.02, "learning_rate": 4.999795474665437e-05, "loss": 0.1364, "step": 40 }, { "epoch": 0.02, "learning_rate": 4.99968043161579e-05, "loss": 0.125, "step": 50 }, { "epoch": 0.02, "learning_rate": 4.999539825840548e-05, "loss": 0.102, "step": 60 }, { "epoch": 0.03, "learning_rate": 4.9993736587776e-05, "loss": 0.094, "step": 70 }, { "epoch": 0.03, "learning_rate": 4.9991819321262375e-05, "loss": 0.0937, "step": 80 }, { "epoch": 0.04, "learning_rate": 4.998964647847133e-05, "loss": 0.0814, "step": 90 }, { "epoch": 0.04, "learning_rate": 4.998721808162322e-05, "loss": 0.068, "step": 100 }, { "epoch": 0.04, "learning_rate": 4.9984534155551795e-05, "loss": 0.0713, "step": 110 }, { "epoch": 0.05, "learning_rate": 4.9981594727703955e-05, "loss": 0.0707, "step": 120 }, { "epoch": 0.05, "learning_rate": 4.997839982813946e-05, "loss": 0.0591, "step": 130 }, { "epoch": 0.06, "learning_rate": 4.9974949489530605e-05, "loss": 0.0613, "step": 140 }, { "epoch": 0.06, "learning_rate": 4.997124374716194e-05, "loss": 0.0544, "step": 150 }, { "epoch": 0.07, "learning_rate": 4.996728263892985e-05, "loss": 0.043, "step": 160 }, { "epoch": 0.07, "learning_rate": 4.9963066205342216e-05, "loss": 0.056, "step": 170 }, { "epoch": 0.07, "learning_rate": 4.995859448951795e-05, "loss": 0.0566, "step": 180 }, { "epoch": 0.08, "learning_rate": 4.99538675371866e-05, "loss": 0.0509, "step": 190 }, { "epoch": 0.08, "learning_rate": 4.994888539668786e-05, "loss": 0.0496, "step": 200 }, { "epoch": 0.09, "learning_rate": 4.9943648118971075e-05, "loss": 0.0578, "step": 210 }, { "epoch": 0.09, "learning_rate": 4.993815575759474e-05, "loss": 0.0509, "step": 220 }, { "epoch": 0.09, "learning_rate": 4.9932408368725915e-05, "loss": 0.0507, "step": 230 }, { "epoch": 0.1, "learning_rate": 4.992640601113968e-05, "loss": 0.0337, "step": 240 }, { "epoch": 0.1, "learning_rate": 4.992014874621853e-05, "loss": 0.0433, "step": 250 }, { "epoch": 0.11, "learning_rate": 4.9913636637951756e-05, "loss": 0.0433, "step": 260 }, { "epoch": 0.11, "learning_rate": 4.990686975293476e-05, "loss": 0.057, "step": 270 }, { "epoch": 0.11, "learning_rate": 4.9899848160368386e-05, "loss": 0.048, "step": 280 }, { "epoch": 0.12, "learning_rate": 4.989257193205824e-05, "loss": 0.046, "step": 290 }, { "epoch": 0.12, "learning_rate": 4.9885041142413926e-05, "loss": 0.0463, "step": 300 }, { "epoch": 0.13, "learning_rate": 4.987725586844829e-05, "loss": 0.047, "step": 310 }, { "epoch": 0.13, "learning_rate": 4.986921618977664e-05, "loss": 0.0434, "step": 320 }, { "epoch": 0.13, "learning_rate": 4.986092218861591e-05, "loss": 0.0363, "step": 330 }, { "epoch": 0.14, "learning_rate": 4.9852373949783875e-05, "loss": 0.0327, "step": 340 }, { "epoch": 0.14, "learning_rate": 4.984357156069821e-05, "loss": 0.0353, "step": 350 }, { "epoch": 0.15, "learning_rate": 4.983451511137565e-05, "loss": 0.0513, "step": 360 }, { "epoch": 0.15, "learning_rate": 4.982520469443104e-05, "loss": 0.0329, "step": 370 }, { "epoch": 0.15, "learning_rate": 4.981564040507641e-05, "loss": 0.0424, "step": 380 }, { "epoch": 0.16, "learning_rate": 4.980582234111998e-05, "loss": 0.0411, "step": 390 }, { "epoch": 0.16, "learning_rate": 4.979575060296517e-05, "loss": 0.0466, "step": 400 }, { "epoch": 0.17, "learning_rate": 4.9785425293609586e-05, "loss": 0.0312, "step": 410 }, { "epoch": 0.17, "learning_rate": 4.977484651864395e-05, "loss": 0.0397, "step": 420 }, { "epoch": 0.18, "learning_rate": 4.9764014386251e-05, "loss": 0.034, "step": 430 }, { "epoch": 0.18, "learning_rate": 4.975292900720444e-05, "loss": 0.0356, "step": 440 }, { "epoch": 0.18, "learning_rate": 4.974159049486776e-05, "loss": 0.0329, "step": 450 }, { "epoch": 0.19, "learning_rate": 4.9729998965193096e-05, "loss": 0.035, "step": 460 }, { "epoch": 0.19, "learning_rate": 4.9718154536720044e-05, "loss": 0.0383, "step": 470 }, { "epoch": 0.2, "learning_rate": 4.970605733057441e-05, "loss": 0.0369, "step": 480 }, { "epoch": 0.2, "learning_rate": 4.9693707470467063e-05, "loss": 0.0368, "step": 490 }, { "epoch": 0.2, "learning_rate": 4.968110508269257e-05, "loss": 0.036, "step": 500 }, { "epoch": 0.21, "learning_rate": 4.9668250296127946e-05, "loss": 0.0376, "step": 510 }, { "epoch": 0.21, "learning_rate": 4.965514324223137e-05, "loss": 0.0292, "step": 520 }, { "epoch": 0.22, "learning_rate": 4.964178405504076e-05, "loss": 0.0316, "step": 530 }, { "epoch": 0.22, "learning_rate": 4.9628172871172496e-05, "loss": 0.0297, "step": 540 }, { "epoch": 0.22, "learning_rate": 4.961430982981994e-05, "loss": 0.0417, "step": 550 }, { "epoch": 0.23, "learning_rate": 4.9600195072752066e-05, "loss": 0.0365, "step": 560 }, { "epoch": 0.23, "learning_rate": 4.9585828744311987e-05, "loss": 0.0355, "step": 570 }, { "epoch": 0.24, "learning_rate": 4.95712109914155e-05, "loss": 0.043, "step": 580 }, { "epoch": 0.24, "learning_rate": 4.9556341963549536e-05, "loss": 0.0261, "step": 590 }, { "epoch": 0.24, "learning_rate": 4.95412218127707e-05, "loss": 0.0376, "step": 600 }, { "epoch": 0.25, "learning_rate": 4.9525850693703656e-05, "loss": 0.0306, "step": 610 }, { "epoch": 0.25, "learning_rate": 4.951022876353958e-05, "loss": 0.0324, "step": 620 }, { "epoch": 0.26, "learning_rate": 4.9494356182034534e-05, "loss": 0.0383, "step": 630 }, { "epoch": 0.26, "learning_rate": 4.9478233111507856e-05, "loss": 0.0281, "step": 640 }, { "epoch": 0.26, "learning_rate": 4.9461859716840456e-05, "loss": 0.0385, "step": 650 }, { "epoch": 0.27, "learning_rate": 4.9445236165473196e-05, "loss": 0.0351, "step": 660 }, { "epoch": 0.27, "learning_rate": 4.94283626274051e-05, "loss": 0.0277, "step": 670 }, { "epoch": 0.28, "learning_rate": 4.941123927519169e-05, "loss": 0.0353, "step": 680 }, { "epoch": 0.28, "learning_rate": 4.9393866283943166e-05, "loss": 0.0295, "step": 690 }, { "epoch": 0.29, "learning_rate": 4.9376243831322635e-05, "loss": 0.034, "step": 700 }, { "epoch": 0.29, "learning_rate": 4.9358372097544316e-05, "loss": 0.0314, "step": 710 }, { "epoch": 0.29, "learning_rate": 4.934025126537164e-05, "loss": 0.031, "step": 720 }, { "epoch": 0.3, "learning_rate": 4.932188152011546e-05, "loss": 0.0292, "step": 730 }, { "epoch": 0.3, "learning_rate": 4.9303263049632083e-05, "loss": 0.027, "step": 740 }, { "epoch": 0.31, "learning_rate": 4.9284396044321386e-05, "loss": 0.0297, "step": 750 }, { "epoch": 0.31, "learning_rate": 4.926528069712487e-05, "loss": 0.0314, "step": 760 }, { "epoch": 0.31, "learning_rate": 4.924591720352367e-05, "loss": 0.0381, "step": 770 }, { "epoch": 0.32, "learning_rate": 4.922630576153656e-05, "loss": 0.0246, "step": 780 }, { "epoch": 0.32, "learning_rate": 4.9206446571717955e-05, "loss": 0.0337, "step": 790 }, { "epoch": 0.33, "learning_rate": 4.918633983715581e-05, "loss": 0.032, "step": 800 }, { "epoch": 0.33, "learning_rate": 4.9165985763469605e-05, "loss": 0.0256, "step": 810 }, { "epoch": 0.33, "learning_rate": 4.9145384558808156e-05, "loss": 0.0269, "step": 820 }, { "epoch": 0.34, "learning_rate": 4.912453643384759e-05, "loss": 0.0237, "step": 830 }, { "epoch": 0.34, "learning_rate": 4.910344160178911e-05, "loss": 0.0267, "step": 840 }, { "epoch": 0.35, "learning_rate": 4.908210027835687e-05, "loss": 0.0353, "step": 850 }, { "epoch": 0.35, "learning_rate": 4.906051268179571e-05, "loss": 0.0309, "step": 860 }, { "epoch": 0.35, "learning_rate": 4.9038679032868976e-05, "loss": 0.0337, "step": 870 }, { "epoch": 0.36, "learning_rate": 4.9016599554856235e-05, "loss": 0.0349, "step": 880 }, { "epoch": 0.36, "learning_rate": 4.899427447355101e-05, "loss": 0.0241, "step": 890 }, { "epoch": 0.37, "learning_rate": 4.897170401725846e-05, "loss": 0.0337, "step": 900 }, { "epoch": 0.37, "learning_rate": 4.894888841679303e-05, "loss": 0.0301, "step": 910 }, { "epoch": 0.37, "learning_rate": 4.892582790547613e-05, "loss": 0.0215, "step": 920 }, { "epoch": 0.38, "learning_rate": 4.8902522719133705e-05, "loss": 0.0288, "step": 930 }, { "epoch": 0.38, "learning_rate": 4.887897309609386e-05, "loss": 0.028, "step": 940 }, { "epoch": 0.39, "learning_rate": 4.88551792771844e-05, "loss": 0.025, "step": 950 }, { "epoch": 0.39, "learning_rate": 4.883114150573037e-05, "loss": 0.0293, "step": 960 }, { "epoch": 0.39, "learning_rate": 4.880686002755157e-05, "loss": 0.0273, "step": 970 }, { "epoch": 0.4, "learning_rate": 4.878233509096006e-05, "loss": 0.0248, "step": 980 }, { "epoch": 0.4, "learning_rate": 4.875756694675758e-05, "loss": 0.0291, "step": 990 }, { "epoch": 0.41, "learning_rate": 4.873255584823302e-05, "loss": 0.0364, "step": 1000 }, { "epoch": 0.41, "learning_rate": 4.8707302051159807e-05, "loss": 0.0285, "step": 1010 }, { "epoch": 0.42, "learning_rate": 4.868180581379333e-05, "loss": 0.0309, "step": 1020 }, { "epoch": 0.42, "learning_rate": 4.865606739686821e-05, "loss": 0.0233, "step": 1030 }, { "epoch": 0.42, "learning_rate": 4.8630087063595754e-05, "loss": 0.0313, "step": 1040 }, { "epoch": 0.43, "learning_rate": 4.8603865079661173e-05, "loss": 0.0281, "step": 1050 }, { "epoch": 0.43, "learning_rate": 4.857740171322088e-05, "loss": 0.025, "step": 1060 }, { "epoch": 0.44, "learning_rate": 4.85506972348998e-05, "loss": 0.029, "step": 1070 }, { "epoch": 0.44, "learning_rate": 4.852375191778853e-05, "loss": 0.0292, "step": 1080 }, { "epoch": 0.44, "learning_rate": 4.84965660374406e-05, "loss": 0.0304, "step": 1090 }, { "epoch": 0.45, "learning_rate": 4.846913987186962e-05, "loss": 0.0297, "step": 1100 }, { "epoch": 0.45, "learning_rate": 4.844147370154646e-05, "loss": 0.0298, "step": 1110 }, { "epoch": 0.46, "learning_rate": 4.8413567809396376e-05, "loss": 0.0251, "step": 1120 }, { "epoch": 0.46, "learning_rate": 4.8385422480796114e-05, "loss": 0.0307, "step": 1130 }, { "epoch": 0.46, "learning_rate": 4.8357038003570984e-05, "loss": 0.0259, "step": 1140 }, { "epoch": 0.47, "learning_rate": 4.832841466799195e-05, "loss": 0.0251, "step": 1150 }, { "epoch": 0.47, "learning_rate": 4.8299552766772594e-05, "loss": 0.0237, "step": 1160 }, { "epoch": 0.48, "learning_rate": 4.82704525950662e-05, "loss": 0.0263, "step": 1170 }, { "epoch": 0.48, "learning_rate": 4.824111445046269e-05, "loss": 0.024, "step": 1180 }, { "epoch": 0.48, "learning_rate": 4.8211538632985595e-05, "loss": 0.0321, "step": 1190 }, { "epoch": 0.49, "learning_rate": 4.818172544508897e-05, "loss": 0.027, "step": 1200 }, { "epoch": 0.49, "learning_rate": 4.815167519165433e-05, "loss": 0.0236, "step": 1210 }, { "epoch": 0.5, "learning_rate": 4.812138817998751e-05, "loss": 0.0271, "step": 1220 }, { "epoch": 0.5, "learning_rate": 4.809086471981553e-05, "loss": 0.0256, "step": 1230 }, { "epoch": 0.5, "learning_rate": 4.806010512328342e-05, "loss": 0.0202, "step": 1240 }, { "epoch": 0.51, "learning_rate": 4.8029109704951046e-05, "loss": 0.0287, "step": 1250 }, { "epoch": 0.51, "learning_rate": 4.7997878781789865e-05, "loss": 0.0284, "step": 1260 }, { "epoch": 0.52, "learning_rate": 4.796641267317973e-05, "loss": 0.0278, "step": 1270 }, { "epoch": 0.52, "learning_rate": 4.793471170090555e-05, "loss": 0.0272, "step": 1280 }, { "epoch": 0.53, "learning_rate": 4.790277618915408e-05, "loss": 0.0247, "step": 1290 }, { "epoch": 0.53, "learning_rate": 4.7870606464510556e-05, "loss": 0.0276, "step": 1300 }, { "epoch": 0.53, "learning_rate": 4.783820285595536e-05, "loss": 0.0336, "step": 1310 }, { "epoch": 0.54, "learning_rate": 4.7805565694860686e-05, "loss": 0.0339, "step": 1320 }, { "epoch": 0.54, "learning_rate": 4.77726953149871e-05, "loss": 0.0243, "step": 1330 }, { "epoch": 0.55, "learning_rate": 4.773959205248019e-05, "loss": 0.0296, "step": 1340 }, { "epoch": 0.55, "learning_rate": 4.770625624586705e-05, "loss": 0.0212, "step": 1350 }, { "epoch": 0.55, "learning_rate": 4.7672688236052906e-05, "loss": 0.0195, "step": 1360 }, { "epoch": 0.56, "learning_rate": 4.763888836631756e-05, "loss": 0.0253, "step": 1370 }, { "epoch": 0.56, "learning_rate": 4.760485698231192e-05, "loss": 0.0229, "step": 1380 }, { "epoch": 0.57, "learning_rate": 4.757059443205444e-05, "loss": 0.0259, "step": 1390 }, { "epoch": 0.57, "learning_rate": 4.7536101065927586e-05, "loss": 0.0204, "step": 1400 }, { "epoch": 0.57, "learning_rate": 4.7501377236674215e-05, "loss": 0.0217, "step": 1410 }, { "epoch": 0.58, "learning_rate": 4.7466423299394013e-05, "loss": 0.0295, "step": 1420 }, { "epoch": 0.58, "learning_rate": 4.743123961153983e-05, "loss": 0.0191, "step": 1430 }, { "epoch": 0.59, "learning_rate": 4.7395826532914054e-05, "loss": 0.027, "step": 1440 }, { "epoch": 0.59, "learning_rate": 4.736018442566489e-05, "loss": 0.0242, "step": 1450 }, { "epoch": 0.59, "learning_rate": 4.7324313654282705e-05, "loss": 0.0207, "step": 1460 }, { "epoch": 0.6, "learning_rate": 4.7288214585596246e-05, "loss": 0.026, "step": 1470 }, { "epoch": 0.6, "learning_rate": 4.725188758876895e-05, "loss": 0.0268, "step": 1480 }, { "epoch": 0.61, "learning_rate": 4.721533303529513e-05, "loss": 0.0209, "step": 1490 }, { "epoch": 0.61, "learning_rate": 4.717855129899617e-05, "loss": 0.0193, "step": 1500 }, { "epoch": 0.61, "learning_rate": 4.714154275601671e-05, "loss": 0.0327, "step": 1510 }, { "epoch": 0.62, "learning_rate": 4.7104307784820845e-05, "loss": 0.0311, "step": 1520 }, { "epoch": 0.62, "learning_rate": 4.706684676618818e-05, "loss": 0.0195, "step": 1530 }, { "epoch": 0.63, "learning_rate": 4.702916008320998e-05, "loss": 0.0353, "step": 1540 }, { "epoch": 0.63, "learning_rate": 4.699124812128525e-05, "loss": 0.027, "step": 1550 }, { "epoch": 0.64, "learning_rate": 4.695311126811676e-05, "loss": 0.0253, "step": 1560 }, { "epoch": 0.64, "learning_rate": 4.6914749913707146e-05, "loss": 0.0266, "step": 1570 }, { "epoch": 0.64, "learning_rate": 4.687616445035487e-05, "loss": 0.0222, "step": 1580 }, { "epoch": 0.65, "learning_rate": 4.68373552726502e-05, "loss": 0.0253, "step": 1590 }, { "epoch": 0.65, "learning_rate": 4.6798322777471216e-05, "loss": 0.0291, "step": 1600 }, { "epoch": 0.66, "learning_rate": 4.6759067363979735e-05, "loss": 0.0276, "step": 1610 }, { "epoch": 0.66, "learning_rate": 4.6719589433617215e-05, "loss": 0.0294, "step": 1620 }, { "epoch": 0.66, "learning_rate": 4.667988939010066e-05, "loss": 0.0236, "step": 1630 }, { "epoch": 0.67, "learning_rate": 4.6639967639418494e-05, "loss": 0.0227, "step": 1640 }, { "epoch": 0.67, "learning_rate": 4.6599824589826405e-05, "loss": 0.031, "step": 1650 }, { "epoch": 0.68, "learning_rate": 4.655946065184318e-05, "loss": 0.0229, "step": 1660 }, { "epoch": 0.68, "learning_rate": 4.6518876238246475e-05, "loss": 0.0203, "step": 1670 }, { "epoch": 0.68, "learning_rate": 4.647807176406865e-05, "loss": 0.0202, "step": 1680 }, { "epoch": 0.69, "learning_rate": 4.643704764659248e-05, "loss": 0.0216, "step": 1690 }, { "epoch": 0.69, "learning_rate": 4.639580430534688e-05, "loss": 0.0284, "step": 1700 }, { "epoch": 0.7, "learning_rate": 4.635434216210266e-05, "loss": 0.0268, "step": 1710 }, { "epoch": 0.7, "learning_rate": 4.631266164086816e-05, "loss": 0.0287, "step": 1720 }, { "epoch": 0.7, "learning_rate": 4.6270763167884976e-05, "loss": 0.0252, "step": 1730 }, { "epoch": 0.71, "learning_rate": 4.622864717162354e-05, "loss": 0.028, "step": 1740 }, { "epoch": 0.71, "learning_rate": 4.618631408277877e-05, "loss": 0.0243, "step": 1750 }, { "epoch": 0.72, "learning_rate": 4.614376433426565e-05, "loss": 0.021, "step": 1760 }, { "epoch": 0.72, "learning_rate": 4.610099836121482e-05, "loss": 0.0235, "step": 1770 }, { "epoch": 0.72, "learning_rate": 4.605801660096812e-05, "loss": 0.0274, "step": 1780 }, { "epoch": 0.73, "learning_rate": 4.6014819493074125e-05, "loss": 0.0228, "step": 1790 }, { "epoch": 0.73, "learning_rate": 4.597140747928361e-05, "loss": 0.0326, "step": 1800 }, { "epoch": 0.74, "learning_rate": 4.59277810035451e-05, "loss": 0.0319, "step": 1810 }, { "epoch": 0.74, "learning_rate": 4.5883940512000265e-05, "loss": 0.0232, "step": 1820 }, { "epoch": 0.75, "learning_rate": 4.58398864529794e-05, "loss": 0.021, "step": 1830 }, { "epoch": 0.75, "learning_rate": 4.5795619276996814e-05, "loss": 0.0242, "step": 1840 }, { "epoch": 0.75, "learning_rate": 4.575113943674625e-05, "loss": 0.0274, "step": 1850 }, { "epoch": 0.76, "learning_rate": 4.570644738709623e-05, "loss": 0.0227, "step": 1860 }, { "epoch": 0.76, "learning_rate": 4.566154358508542e-05, "loss": 0.022, "step": 1870 }, { "epoch": 0.77, "learning_rate": 4.561642848991795e-05, "loss": 0.0274, "step": 1880 }, { "epoch": 0.77, "learning_rate": 4.5571102562958704e-05, "loss": 0.0239, "step": 1890 }, { "epoch": 0.77, "learning_rate": 4.552556626772863e-05, "loss": 0.0228, "step": 1900 }, { "epoch": 0.78, "learning_rate": 4.5479820069899984e-05, "loss": 0.0213, "step": 1910 }, { "epoch": 0.78, "learning_rate": 4.543386443729157e-05, "loss": 0.0239, "step": 1920 }, { "epoch": 0.79, "learning_rate": 4.538769983986394e-05, "loss": 0.0259, "step": 1930 }, { "epoch": 0.79, "learning_rate": 4.5341326749714625e-05, "loss": 0.0274, "step": 1940 }, { "epoch": 0.79, "learning_rate": 4.5294745641073276e-05, "loss": 0.0209, "step": 1950 }, { "epoch": 0.8, "learning_rate": 4.524795699029681e-05, "loss": 0.0184, "step": 1960 }, { "epoch": 0.8, "learning_rate": 4.5200961275864594e-05, "loss": 0.0181, "step": 1970 }, { "epoch": 0.81, "learning_rate": 4.515375897837346e-05, "loss": 0.0208, "step": 1980 }, { "epoch": 0.81, "learning_rate": 4.510635058053287e-05, "loss": 0.0183, "step": 1990 }, { "epoch": 0.81, "learning_rate": 4.505873656715994e-05, "loss": 0.0245, "step": 2000 }, { "epoch": 0.82, "learning_rate": 4.501091742517449e-05, "loss": 0.0197, "step": 2010 }, { "epoch": 0.82, "learning_rate": 4.4962893643594096e-05, "loss": 0.0243, "step": 2020 }, { "epoch": 0.83, "learning_rate": 4.4914665713529004e-05, "loss": 0.0201, "step": 2030 }, { "epoch": 0.83, "learning_rate": 4.486623412817721e-05, "loss": 0.0186, "step": 2040 }, { "epoch": 0.83, "learning_rate": 4.481759938281935e-05, "loss": 0.0154, "step": 2050 }, { "epoch": 0.84, "learning_rate": 4.476876197481368e-05, "loss": 0.0233, "step": 2060 }, { "epoch": 0.84, "learning_rate": 4.4719722403590916e-05, "loss": 0.0245, "step": 2070 }, { "epoch": 0.85, "learning_rate": 4.4670481170649214e-05, "loss": 0.0307, "step": 2080 }, { "epoch": 0.85, "learning_rate": 4.462103877954901e-05, "loss": 0.0235, "step": 2090 }, { "epoch": 0.86, "learning_rate": 4.457139573590782e-05, "loss": 0.0279, "step": 2100 }, { "epoch": 0.86, "learning_rate": 4.452155254739516e-05, "loss": 0.0215, "step": 2110 }, { "epoch": 0.86, "learning_rate": 4.4471509723727265e-05, "loss": 0.0248, "step": 2120 }, { "epoch": 0.87, "learning_rate": 4.442126777666197e-05, "loss": 0.0298, "step": 2130 }, { "epoch": 0.87, "learning_rate": 4.437082721999336e-05, "loss": 0.0229, "step": 2140 }, { "epoch": 0.88, "learning_rate": 4.432018856954664e-05, "loss": 0.0252, "step": 2150 }, { "epoch": 0.88, "learning_rate": 4.4269352343172765e-05, "loss": 0.0242, "step": 2160 }, { "epoch": 0.88, "learning_rate": 4.421831906074319e-05, "loss": 0.0198, "step": 2170 }, { "epoch": 0.89, "learning_rate": 4.416708924414456e-05, "loss": 0.0222, "step": 2180 }, { "epoch": 0.89, "learning_rate": 4.411566341727331e-05, "loss": 0.0218, "step": 2190 }, { "epoch": 0.9, "learning_rate": 4.406404210603041e-05, "loss": 0.0238, "step": 2200 }, { "epoch": 0.9, "learning_rate": 4.401222583831591e-05, "loss": 0.0292, "step": 2210 }, { "epoch": 0.9, "learning_rate": 4.3960215144023534e-05, "loss": 0.0269, "step": 2220 }, { "epoch": 0.91, "learning_rate": 4.390801055503533e-05, "loss": 0.0219, "step": 2230 }, { "epoch": 0.91, "learning_rate": 4.3855612605216174e-05, "loss": 0.0264, "step": 2240 }, { "epoch": 0.92, "learning_rate": 4.380302183040832e-05, "loss": 0.0185, "step": 2250 }, { "epoch": 0.92, "learning_rate": 4.3750238768425936e-05, "loss": 0.0221, "step": 2260 }, { "epoch": 0.92, "learning_rate": 4.3697263959049597e-05, "loss": 0.0241, "step": 2270 }, { "epoch": 0.93, "learning_rate": 4.364409794402075e-05, "loss": 0.0266, "step": 2280 }, { "epoch": 0.93, "learning_rate": 4.3590741267036205e-05, "loss": 0.0224, "step": 2290 }, { "epoch": 0.94, "learning_rate": 4.3537194473742536e-05, "loss": 0.023, "step": 2300 }, { "epoch": 0.94, "learning_rate": 4.348345811173055e-05, "loss": 0.0204, "step": 2310 }, { "epoch": 0.94, "learning_rate": 4.3429532730529626e-05, "loss": 0.0251, "step": 2320 }, { "epoch": 0.95, "learning_rate": 4.337541888160215e-05, "loss": 0.0218, "step": 2330 }, { "epoch": 0.95, "learning_rate": 4.3321117118337856e-05, "loss": 0.0248, "step": 2340 }, { "epoch": 0.96, "learning_rate": 4.326662799604815e-05, "loss": 0.021, "step": 2350 }, { "epoch": 0.96, "learning_rate": 4.321195207196046e-05, "loss": 0.023, "step": 2360 }, { "epoch": 0.96, "learning_rate": 4.31570899052125e-05, "loss": 0.0208, "step": 2370 }, { "epoch": 0.97, "learning_rate": 4.3102042056846614e-05, "loss": 0.025, "step": 2380 }, { "epoch": 0.97, "learning_rate": 4.3046809089803965e-05, "loss": 0.0213, "step": 2390 }, { "epoch": 0.98, "learning_rate": 4.2991391568918825e-05, "loss": 0.0211, "step": 2400 }, { "epoch": 0.98, "learning_rate": 4.2935790060912805e-05, "loss": 0.0191, "step": 2410 }, { "epoch": 0.99, "learning_rate": 4.2880005134389004e-05, "loss": 0.0212, "step": 2420 }, { "epoch": 0.99, "learning_rate": 4.282403735982626e-05, "loss": 0.029, "step": 2430 }, { "epoch": 0.99, "learning_rate": 4.276788730957329e-05, "loss": 0.0208, "step": 2440 }, { "epoch": 1.0, "learning_rate": 4.271155555784282e-05, "loss": 0.0214, "step": 2450 }, { "epoch": 1.0, "learning_rate": 4.2655042680705715e-05, "loss": 0.0214, "step": 2460 }, { "epoch": 1.01, "learning_rate": 4.259834925608514e-05, "loss": 0.0203, "step": 2470 }, { "epoch": 1.01, "learning_rate": 4.2541475863750576e-05, "loss": 0.0148, "step": 2480 }, { "epoch": 1.01, "learning_rate": 4.2484423085311944e-05, "loss": 0.0163, "step": 2490 }, { "epoch": 1.02, "learning_rate": 4.242719150421363e-05, "loss": 0.0272, "step": 2500 }, { "epoch": 1.02, "learning_rate": 4.236978170572853e-05, "loss": 0.0174, "step": 2510 }, { "epoch": 1.03, "learning_rate": 4.231219427695205e-05, "loss": 0.0155, "step": 2520 }, { "epoch": 1.03, "learning_rate": 4.225442980679614e-05, "loss": 0.0194, "step": 2530 }, { "epoch": 1.03, "learning_rate": 4.2196488885983224e-05, "loss": 0.0131, "step": 2540 }, { "epoch": 1.04, "learning_rate": 4.2138372107040183e-05, "loss": 0.0217, "step": 2550 }, { "epoch": 1.04, "learning_rate": 4.2080080064292304e-05, "loss": 0.0151, "step": 2560 }, { "epoch": 1.05, "learning_rate": 4.202161335385718e-05, "loss": 0.0176, "step": 2570 }, { "epoch": 1.05, "learning_rate": 4.196297257363864e-05, "loss": 0.0204, "step": 2580 }, { "epoch": 1.05, "learning_rate": 4.19041583233206e-05, "loss": 0.0152, "step": 2590 }, { "epoch": 1.06, "learning_rate": 4.184517120436095e-05, "loss": 0.0137, "step": 2600 }, { "epoch": 1.06, "learning_rate": 4.178601181998543e-05, "loss": 0.0156, "step": 2610 }, { "epoch": 1.07, "learning_rate": 4.1726680775181406e-05, "loss": 0.0255, "step": 2620 }, { "epoch": 1.07, "learning_rate": 4.1667178676691734e-05, "loss": 0.0207, "step": 2630 }, { "epoch": 1.07, "learning_rate": 4.1607506133008514e-05, "loss": 0.0159, "step": 2640 }, { "epoch": 1.08, "learning_rate": 4.15476637543669e-05, "loss": 0.021, "step": 2650 }, { "epoch": 1.08, "learning_rate": 4.148765215273884e-05, "loss": 0.0206, "step": 2660 }, { "epoch": 1.09, "learning_rate": 4.142747194182682e-05, "loss": 0.0172, "step": 2670 }, { "epoch": 1.09, "learning_rate": 4.1367123737057586e-05, "loss": 0.0156, "step": 2680 }, { "epoch": 1.1, "learning_rate": 4.1306608155575894e-05, "loss": 0.0222, "step": 2690 }, { "epoch": 1.1, "learning_rate": 4.124592581623811e-05, "loss": 0.021, "step": 2700 }, { "epoch": 1.1, "learning_rate": 4.118507733960595e-05, "loss": 0.0138, "step": 2710 }, { "epoch": 1.11, "learning_rate": 4.1124063347940135e-05, "loss": 0.0178, "step": 2720 }, { "epoch": 1.11, "learning_rate": 4.1062884465194e-05, "loss": 0.0177, "step": 2730 }, { "epoch": 1.12, "learning_rate": 4.1001541317007084e-05, "loss": 0.017, "step": 2740 }, { "epoch": 1.12, "learning_rate": 4.0940034530698834e-05, "loss": 0.0209, "step": 2750 }, { "epoch": 1.12, "learning_rate": 4.087836473526208e-05, "loss": 0.016, "step": 2760 }, { "epoch": 1.13, "learning_rate": 4.0816532561356665e-05, "loss": 0.021, "step": 2770 }, { "epoch": 1.13, "learning_rate": 4.075453864130296e-05, "loss": 0.0184, "step": 2780 }, { "epoch": 1.14, "learning_rate": 4.069238360907544e-05, "loss": 0.0163, "step": 2790 }, { "epoch": 1.14, "learning_rate": 4.0630068100296184e-05, "loss": 0.02, "step": 2800 }, { "epoch": 1.14, "learning_rate": 4.0567592752228337e-05, "loss": 0.0201, "step": 2810 }, { "epoch": 1.15, "learning_rate": 4.050495820376963e-05, "loss": 0.0198, "step": 2820 }, { "epoch": 1.15, "learning_rate": 4.044216509544587e-05, "loss": 0.0212, "step": 2830 }, { "epoch": 1.16, "learning_rate": 4.037921406940433e-05, "loss": 0.0172, "step": 2840 }, { "epoch": 1.16, "learning_rate": 4.0316105769407215e-05, "loss": 0.0197, "step": 2850 }, { "epoch": 1.16, "learning_rate": 4.025284084082509e-05, "loss": 0.0144, "step": 2860 }, { "epoch": 1.17, "learning_rate": 4.0189419930630265e-05, "loss": 0.019, "step": 2870 }, { "epoch": 1.17, "learning_rate": 4.0125843687390166e-05, "loss": 0.0186, "step": 2880 }, { "epoch": 1.18, "learning_rate": 4.0062112761260734e-05, "loss": 0.0242, "step": 2890 }, { "epoch": 1.18, "learning_rate": 3.999822780397974e-05, "loss": 0.0182, "step": 2900 }, { "epoch": 1.18, "learning_rate": 3.993418946886016e-05, "loss": 0.0183, "step": 2910 }, { "epoch": 1.19, "learning_rate": 3.9869998410783445e-05, "loss": 0.0151, "step": 2920 }, { "epoch": 1.19, "learning_rate": 3.980565528619288e-05, "loss": 0.0168, "step": 2930 }, { "epoch": 1.2, "learning_rate": 3.974116075308682e-05, "loss": 0.0168, "step": 2940 }, { "epoch": 1.2, "learning_rate": 3.967651547101201e-05, "loss": 0.0198, "step": 2950 }, { "epoch": 1.21, "learning_rate": 3.9611720101056784e-05, "loss": 0.0201, "step": 2960 }, { "epoch": 1.21, "learning_rate": 3.954677530584435e-05, "loss": 0.0209, "step": 2970 }, { "epoch": 1.21, "learning_rate": 3.9481681749526e-05, "loss": 0.0127, "step": 2980 }, { "epoch": 1.22, "learning_rate": 3.9416440097774314e-05, "loss": 0.0178, "step": 2990 }, { "epoch": 1.22, "learning_rate": 3.935105101777635e-05, "loss": 0.0194, "step": 3000 }, { "epoch": 1.23, "learning_rate": 3.928551517822683e-05, "loss": 0.0168, "step": 3010 }, { "epoch": 1.23, "learning_rate": 3.921983324932129e-05, "loss": 0.0168, "step": 3020 }, { "epoch": 1.23, "learning_rate": 3.9154005902749255e-05, "loss": 0.0163, "step": 3030 }, { "epoch": 1.24, "learning_rate": 3.908803381168732e-05, "loss": 0.0166, "step": 3040 }, { "epoch": 1.24, "learning_rate": 3.902191765079232e-05, "loss": 0.0209, "step": 3050 }, { "epoch": 1.25, "learning_rate": 3.895565809619438e-05, "loss": 0.0229, "step": 3060 }, { "epoch": 1.25, "learning_rate": 3.888925582549006e-05, "loss": 0.0156, "step": 3070 }, { "epoch": 1.25, "learning_rate": 3.882271151773536e-05, "loss": 0.0119, "step": 3080 }, { "epoch": 1.26, "learning_rate": 3.875602585343881e-05, "loss": 0.0191, "step": 3090 }, { "epoch": 1.26, "learning_rate": 3.868919951455454e-05, "loss": 0.0183, "step": 3100 }, { "epoch": 1.27, "learning_rate": 3.862223318447523e-05, "loss": 0.0191, "step": 3110 }, { "epoch": 1.27, "learning_rate": 3.855512754802521e-05, "loss": 0.0206, "step": 3120 }, { "epoch": 1.27, "learning_rate": 3.848788329145335e-05, "loss": 0.0186, "step": 3130 }, { "epoch": 1.28, "learning_rate": 3.84205011024262e-05, "loss": 0.0178, "step": 3140 }, { "epoch": 1.28, "learning_rate": 3.835298167002076e-05, "loss": 0.0152, "step": 3150 }, { "epoch": 1.29, "learning_rate": 3.82853256847176e-05, "loss": 0.018, "step": 3160 }, { "epoch": 1.29, "learning_rate": 3.8217533838393714e-05, "loss": 0.016, "step": 3170 }, { "epoch": 1.29, "learning_rate": 3.8149606824315464e-05, "loss": 0.0169, "step": 3180 }, { "epoch": 1.3, "learning_rate": 3.8081545337131494e-05, "loss": 0.0192, "step": 3190 }, { "epoch": 1.3, "learning_rate": 3.801335007286564e-05, "loss": 0.0199, "step": 3200 }, { "epoch": 1.31, "learning_rate": 3.794502172890975e-05, "loss": 0.0181, "step": 3210 }, { "epoch": 1.31, "learning_rate": 3.787656100401664e-05, "loss": 0.0145, "step": 3220 }, { "epoch": 1.32, "learning_rate": 3.78079685982929e-05, "loss": 0.0213, "step": 3230 }, { "epoch": 1.32, "learning_rate": 3.7739245213191724e-05, "loss": 0.0198, "step": 3240 }, { "epoch": 1.32, "learning_rate": 3.7670391551505754e-05, "loss": 0.0137, "step": 3250 }, { "epoch": 1.33, "learning_rate": 3.760140831735993e-05, "loss": 0.0166, "step": 3260 }, { "epoch": 1.33, "learning_rate": 3.7532296216204186e-05, "loss": 0.0142, "step": 3270 }, { "epoch": 1.34, "learning_rate": 3.746305595480636e-05, "loss": 0.0174, "step": 3280 }, { "epoch": 1.34, "learning_rate": 3.739368824124488e-05, "loss": 0.0166, "step": 3290 }, { "epoch": 1.34, "learning_rate": 3.7324193784901554e-05, "loss": 0.0142, "step": 3300 }, { "epoch": 1.35, "learning_rate": 3.72545732964543e-05, "loss": 0.0127, "step": 3310 }, { "epoch": 1.35, "learning_rate": 3.718482748786993e-05, "loss": 0.0185, "step": 3320 }, { "epoch": 1.36, "learning_rate": 3.711495707239678e-05, "loss": 0.0159, "step": 3330 }, { "epoch": 1.36, "learning_rate": 3.70449627645575e-05, "loss": 0.0221, "step": 3340 }, { "epoch": 1.36, "learning_rate": 3.697484528014171e-05, "loss": 0.016, "step": 3350 }, { "epoch": 1.37, "learning_rate": 3.690460533619866e-05, "loss": 0.0219, "step": 3360 }, { "epoch": 1.37, "learning_rate": 3.6834243651029935e-05, "loss": 0.0182, "step": 3370 }, { "epoch": 1.38, "learning_rate": 3.676376094418208e-05, "loss": 0.0127, "step": 3380 }, { "epoch": 1.38, "learning_rate": 3.6693157936439306e-05, "loss": 0.0151, "step": 3390 }, { "epoch": 1.38, "learning_rate": 3.6622435349816e-05, "loss": 0.0161, "step": 3400 }, { "epoch": 1.39, "learning_rate": 3.6551593907549445e-05, "loss": 0.021, "step": 3410 }, { "epoch": 1.39, "learning_rate": 3.64806343340924e-05, "loss": 0.0148, "step": 3420 }, { "epoch": 1.4, "learning_rate": 3.640955735510566e-05, "loss": 0.0175, "step": 3430 }, { "epoch": 1.4, "learning_rate": 3.633836369745065e-05, "loss": 0.0182, "step": 3440 }, { "epoch": 1.4, "learning_rate": 3.626705408918202e-05, "loss": 0.0155, "step": 3450 }, { "epoch": 1.41, "learning_rate": 3.6195629259540174e-05, "loss": 0.0186, "step": 3460 }, { "epoch": 1.41, "learning_rate": 3.612408993894379e-05, "loss": 0.0136, "step": 3470 }, { "epoch": 1.42, "learning_rate": 3.6052436858982406e-05, "loss": 0.0196, "step": 3480 }, { "epoch": 1.42, "learning_rate": 3.5980670752408895e-05, "loss": 0.0148, "step": 3490 }, { "epoch": 1.43, "learning_rate": 3.590879235313199e-05, "loss": 0.0172, "step": 3500 }, { "epoch": 1.43, "learning_rate": 3.583680239620875e-05, "loss": 0.0171, "step": 3510 }, { "epoch": 1.43, "learning_rate": 3.576470161783712e-05, "loss": 0.02, "step": 3520 }, { "epoch": 1.44, "learning_rate": 3.569249075534828e-05, "loss": 0.0134, "step": 3530 }, { "epoch": 1.44, "learning_rate": 3.562017054719923e-05, "loss": 0.02, "step": 3540 }, { "epoch": 1.45, "learning_rate": 3.554774173296516e-05, "loss": 0.0147, "step": 3550 }, { "epoch": 1.45, "learning_rate": 3.547520505333192e-05, "loss": 0.0161, "step": 3560 }, { "epoch": 1.45, "learning_rate": 3.5402561250088414e-05, "loss": 0.0197, "step": 3570 }, { "epoch": 1.46, "learning_rate": 3.532981106611905e-05, "loss": 0.0214, "step": 3580 }, { "epoch": 1.46, "learning_rate": 3.5256955245396126e-05, "loss": 0.0166, "step": 3590 }, { "epoch": 1.47, "learning_rate": 3.518399453297221e-05, "loss": 0.0197, "step": 3600 }, { "epoch": 1.47, "learning_rate": 3.511092967497256e-05, "loss": 0.0155, "step": 3610 }, { "epoch": 1.47, "learning_rate": 3.503776141858745e-05, "loss": 0.0136, "step": 3620 }, { "epoch": 1.48, "learning_rate": 3.4964490512064545e-05, "loss": 0.0139, "step": 3630 }, { "epoch": 1.48, "learning_rate": 3.4891117704701266e-05, "loss": 0.0149, "step": 3640 }, { "epoch": 1.49, "learning_rate": 3.481764374683709e-05, "loss": 0.0147, "step": 3650 }, { "epoch": 1.49, "learning_rate": 3.474406938984593e-05, "loss": 0.0157, "step": 3660 }, { "epoch": 1.49, "learning_rate": 3.4670395386128384e-05, "loss": 0.0167, "step": 3670 }, { "epoch": 1.5, "learning_rate": 3.459662248910411e-05, "loss": 0.0171, "step": 3680 }, { "epoch": 1.5, "learning_rate": 3.452275145320407e-05, "loss": 0.0153, "step": 3690 }, { "epoch": 1.51, "learning_rate": 3.4448783033862834e-05, "loss": 0.017, "step": 3700 }, { "epoch": 1.51, "learning_rate": 3.437471798751086e-05, "loss": 0.017, "step": 3710 }, { "epoch": 1.51, "learning_rate": 3.4300557071566746e-05, "loss": 0.02, "step": 3720 }, { "epoch": 1.52, "learning_rate": 3.42263010444295e-05, "loss": 0.0209, "step": 3730 }, { "epoch": 1.52, "learning_rate": 3.415195066547077e-05, "loss": 0.0154, "step": 3740 }, { "epoch": 1.53, "learning_rate": 3.4077506695027074e-05, "loss": 0.0166, "step": 3750 }, { "epoch": 1.53, "learning_rate": 3.400296989439203e-05, "loss": 0.017, "step": 3760 }, { "epoch": 1.54, "learning_rate": 3.392834102580862e-05, "loss": 0.0151, "step": 3770 }, { "epoch": 1.54, "learning_rate": 3.385362085246129e-05, "loss": 0.0187, "step": 3780 }, { "epoch": 1.54, "learning_rate": 3.377881013846823e-05, "loss": 0.0163, "step": 3790 }, { "epoch": 1.55, "learning_rate": 3.370390964887355e-05, "loss": 0.0167, "step": 3800 }, { "epoch": 1.55, "learning_rate": 3.362892014963942e-05, "loss": 0.0196, "step": 3810 }, { "epoch": 1.56, "learning_rate": 3.355384240763826e-05, "loss": 0.0205, "step": 3820 }, { "epoch": 1.56, "learning_rate": 3.347867719064489e-05, "loss": 0.0157, "step": 3830 }, { "epoch": 1.56, "learning_rate": 3.3403425267328715e-05, "loss": 0.0176, "step": 3840 }, { "epoch": 1.57, "learning_rate": 3.33280874072458e-05, "loss": 0.0123, "step": 3850 }, { "epoch": 1.57, "learning_rate": 3.325266438083104e-05, "loss": 0.0139, "step": 3860 }, { "epoch": 1.58, "learning_rate": 3.31771569593903e-05, "loss": 0.0157, "step": 3870 }, { "epoch": 1.58, "learning_rate": 3.310156591509245e-05, "loss": 0.0178, "step": 3880 }, { "epoch": 1.58, "learning_rate": 3.302589202096159e-05, "loss": 0.015, "step": 3890 }, { "epoch": 1.59, "learning_rate": 3.2950136050869004e-05, "loss": 0.018, "step": 3900 }, { "epoch": 1.59, "learning_rate": 3.287429877952538e-05, "loss": 0.0174, "step": 3910 }, { "epoch": 1.6, "learning_rate": 3.2798380982472776e-05, "loss": 0.0141, "step": 3920 }, { "epoch": 1.6, "learning_rate": 3.272238343607677e-05, "loss": 0.0112, "step": 3930 }, { "epoch": 1.6, "learning_rate": 3.264630691751847e-05, "loss": 0.0163, "step": 3940 }, { "epoch": 1.61, "learning_rate": 3.25701522047866e-05, "loss": 0.0173, "step": 3950 }, { "epoch": 1.61, "learning_rate": 3.2493920076669504e-05, "loss": 0.0183, "step": 3960 }, { "epoch": 1.62, "learning_rate": 3.241761131274724e-05, "loss": 0.0244, "step": 3970 }, { "epoch": 1.62, "learning_rate": 3.2341226693383555e-05, "loss": 0.0208, "step": 3980 }, { "epoch": 1.62, "learning_rate": 3.226476699971791e-05, "loss": 0.017, "step": 3990 }, { "epoch": 1.63, "learning_rate": 3.218823301365754e-05, "loss": 0.0155, "step": 4000 }, { "epoch": 1.63, "learning_rate": 3.21116255178694e-05, "loss": 0.0174, "step": 4010 }, { "epoch": 1.64, "learning_rate": 3.203494529577218e-05, "loss": 0.0155, "step": 4020 }, { "epoch": 1.64, "learning_rate": 3.1958193131528315e-05, "loss": 0.0243, "step": 4030 }, { "epoch": 1.64, "learning_rate": 3.188136981003594e-05, "loss": 0.0149, "step": 4040 }, { "epoch": 1.65, "learning_rate": 3.180447611692088e-05, "loss": 0.0142, "step": 4050 }, { "epoch": 1.65, "learning_rate": 3.17275128385286e-05, "loss": 0.0212, "step": 4060 }, { "epoch": 1.66, "learning_rate": 3.1650480761916175e-05, "loss": 0.0133, "step": 4070 }, { "epoch": 1.66, "learning_rate": 3.1573380674844224e-05, "loss": 0.0158, "step": 4080 }, { "epoch": 1.67, "learning_rate": 3.14962133657689e-05, "loss": 0.0126, "step": 4090 }, { "epoch": 1.67, "learning_rate": 3.141897962383376e-05, "loss": 0.0171, "step": 4100 }, { "epoch": 1.67, "learning_rate": 3.134168023886176e-05, "loss": 0.0143, "step": 4110 }, { "epoch": 1.68, "learning_rate": 3.126431600134712e-05, "loss": 0.015, "step": 4120 }, { "epoch": 1.68, "learning_rate": 3.11868877024473e-05, "loss": 0.018, "step": 4130 }, { "epoch": 1.69, "learning_rate": 3.1109396133974835e-05, "loss": 0.0152, "step": 4140 }, { "epoch": 1.69, "learning_rate": 3.103184208838931e-05, "loss": 0.0185, "step": 4150 }, { "epoch": 1.69, "learning_rate": 3.095422635878923e-05, "loss": 0.0154, "step": 4160 }, { "epoch": 1.7, "learning_rate": 3.0876549738903895e-05, "loss": 0.0157, "step": 4170 }, { "epoch": 1.7, "learning_rate": 3.079881302308527e-05, "loss": 0.0152, "step": 4180 }, { "epoch": 1.71, "learning_rate": 3.072101700629993e-05, "loss": 0.0138, "step": 4190 }, { "epoch": 1.71, "learning_rate": 3.064316248412086e-05, "loss": 0.0181, "step": 4200 }, { "epoch": 1.71, "learning_rate": 3.056525025271934e-05, "loss": 0.0157, "step": 4210 }, { "epoch": 1.72, "learning_rate": 3.048728110885682e-05, "loss": 0.0163, "step": 4220 }, { "epoch": 1.72, "learning_rate": 3.0409255849876756e-05, "loss": 0.0176, "step": 4230 }, { "epoch": 1.73, "learning_rate": 3.0331175273696444e-05, "loss": 0.0151, "step": 4240 }, { "epoch": 1.73, "learning_rate": 3.0253040178798903e-05, "loss": 0.016, "step": 4250 }, { "epoch": 1.73, "learning_rate": 3.0174851364224666e-05, "loss": 0.0128, "step": 4260 }, { "epoch": 1.74, "learning_rate": 3.0096609629563622e-05, "loss": 0.0191, "step": 4270 }, { "epoch": 1.74, "learning_rate": 3.0018315774946847e-05, "loss": 0.0159, "step": 4280 }, { "epoch": 1.75, "learning_rate": 2.993997060103843e-05, "loss": 0.0174, "step": 4290 }, { "epoch": 1.75, "learning_rate": 2.986157490902724e-05, "loss": 0.016, "step": 4300 }, { "epoch": 1.75, "learning_rate": 2.9783129500618788e-05, "loss": 0.0161, "step": 4310 }, { "epoch": 1.76, "learning_rate": 2.9704635178027012e-05, "loss": 0.0143, "step": 4320 }, { "epoch": 1.76, "learning_rate": 2.9626092743966028e-05, "loss": 0.0186, "step": 4330 }, { "epoch": 1.77, "learning_rate": 2.9547503001642003e-05, "loss": 0.0176, "step": 4340 }, { "epoch": 1.77, "learning_rate": 2.9468866754744868e-05, "loss": 0.0161, "step": 4350 }, { "epoch": 1.78, "learning_rate": 2.9390184807440142e-05, "loss": 0.0209, "step": 4360 }, { "epoch": 1.78, "learning_rate": 2.931145796436068e-05, "loss": 0.0226, "step": 4370 }, { "epoch": 1.78, "learning_rate": 2.9232687030598493e-05, "loss": 0.0139, "step": 4380 }, { "epoch": 1.79, "learning_rate": 2.9153872811696432e-05, "loss": 0.0155, "step": 4390 }, { "epoch": 1.79, "learning_rate": 2.9075016113640037e-05, "loss": 0.0152, "step": 4400 }, { "epoch": 1.8, "learning_rate": 2.8996117742849245e-05, "loss": 0.0163, "step": 4410 }, { "epoch": 1.8, "learning_rate": 2.891717850617015e-05, "loss": 0.0174, "step": 4420 }, { "epoch": 1.8, "learning_rate": 2.8838199210866757e-05, "loss": 0.0151, "step": 4430 }, { "epoch": 1.81, "learning_rate": 2.8759180664612746e-05, "loss": 0.0218, "step": 4440 }, { "epoch": 1.81, "learning_rate": 2.8680123675483157e-05, "loss": 0.0154, "step": 4450 }, { "epoch": 1.82, "learning_rate": 2.860102905194619e-05, "loss": 0.0147, "step": 4460 }, { "epoch": 1.82, "learning_rate": 2.8521897602854908e-05, "loss": 0.0165, "step": 4470 }, { "epoch": 1.82, "learning_rate": 2.844273013743896e-05, "loss": 0.0139, "step": 4480 }, { "epoch": 1.83, "learning_rate": 2.8363527465296304e-05, "loss": 0.0259, "step": 4490 }, { "epoch": 1.83, "learning_rate": 2.8284290396384943e-05, "loss": 0.0153, "step": 4500 }, { "epoch": 1.84, "learning_rate": 2.8205019741014645e-05, "loss": 0.0142, "step": 4510 }, { "epoch": 1.84, "learning_rate": 2.812571630983864e-05, "loss": 0.013, "step": 4520 }, { "epoch": 1.84, "learning_rate": 2.8046380913845322e-05, "loss": 0.0148, "step": 4530 }, { "epoch": 1.85, "learning_rate": 2.7967014364349996e-05, "loss": 0.0134, "step": 4540 }, { "epoch": 1.85, "learning_rate": 2.7887617472986532e-05, "loss": 0.0177, "step": 4550 }, { "epoch": 1.86, "learning_rate": 2.78081910516991e-05, "loss": 0.0157, "step": 4560 }, { "epoch": 1.86, "learning_rate": 2.7728735912733854e-05, "loss": 0.0155, "step": 4570 }, { "epoch": 1.86, "learning_rate": 2.7649252868630626e-05, "loss": 0.0136, "step": 4580 }, { "epoch": 1.87, "learning_rate": 2.7569742732214615e-05, "loss": 0.0154, "step": 4590 }, { "epoch": 1.87, "learning_rate": 2.7490206316588067e-05, "loss": 0.019, "step": 4600 }, { "epoch": 1.88, "learning_rate": 2.7410644435121996e-05, "loss": 0.0136, "step": 4610 }, { "epoch": 1.88, "learning_rate": 2.7331057901447817e-05, "loss": 0.0148, "step": 4620 }, { "epoch": 1.89, "learning_rate": 2.7251447529449053e-05, "loss": 0.0181, "step": 4630 }, { "epoch": 1.89, "learning_rate": 2.7171814133253015e-05, "loss": 0.0174, "step": 4640 }, { "epoch": 1.89, "learning_rate": 2.7092158527222454e-05, "loss": 0.0169, "step": 4650 }, { "epoch": 1.9, "learning_rate": 2.7012481525947264e-05, "loss": 0.0153, "step": 4660 }, { "epoch": 1.9, "learning_rate": 2.6932783944236128e-05, "loss": 0.0133, "step": 4670 }, { "epoch": 1.91, "learning_rate": 2.685306659710819e-05, "loss": 0.0153, "step": 4680 }, { "epoch": 1.91, "learning_rate": 2.677333029978472e-05, "loss": 0.0206, "step": 4690 }, { "epoch": 1.91, "learning_rate": 2.6693575867680798e-05, "loss": 0.0151, "step": 4700 }, { "epoch": 1.92, "learning_rate": 2.6613804116396933e-05, "loss": 0.0167, "step": 4710 }, { "epoch": 1.92, "learning_rate": 2.653401586171077e-05, "loss": 0.0122, "step": 4720 }, { "epoch": 1.93, "learning_rate": 2.6454211919568707e-05, "loss": 0.0172, "step": 4730 }, { "epoch": 1.93, "learning_rate": 2.637439310607758e-05, "loss": 0.016, "step": 4740 }, { "epoch": 1.93, "learning_rate": 2.6294560237496286e-05, "loss": 0.0179, "step": 4750 }, { "epoch": 1.94, "learning_rate": 2.621471413022748e-05, "loss": 0.0173, "step": 4760 }, { "epoch": 1.94, "learning_rate": 2.6134855600809187e-05, "loss": 0.0144, "step": 4770 }, { "epoch": 1.95, "learning_rate": 2.6054985465906477e-05, "loss": 0.0175, "step": 4780 }, { "epoch": 1.95, "learning_rate": 2.5975104542303075e-05, "loss": 0.0139, "step": 4790 }, { "epoch": 1.95, "learning_rate": 2.589521364689308e-05, "loss": 0.0162, "step": 4800 }, { "epoch": 1.96, "learning_rate": 2.5815313596672518e-05, "loss": 0.0145, "step": 4810 }, { "epoch": 1.96, "learning_rate": 2.573540520873107e-05, "loss": 0.0144, "step": 4820 }, { "epoch": 1.97, "learning_rate": 2.5655489300243678e-05, "loss": 0.0164, "step": 4830 }, { "epoch": 1.97, "learning_rate": 2.5575566688462172e-05, "loss": 0.0128, "step": 4840 }, { "epoch": 1.97, "learning_rate": 2.5495638190706945e-05, "loss": 0.0217, "step": 4850 }, { "epoch": 1.98, "learning_rate": 2.5415704624358593e-05, "loss": 0.0127, "step": 4860 }, { "epoch": 1.98, "learning_rate": 2.5335766806849527e-05, "loss": 0.0114, "step": 4870 }, { "epoch": 1.99, "learning_rate": 2.5255825555655644e-05, "loss": 0.0174, "step": 4880 }, { "epoch": 1.99, "learning_rate": 2.517588168828795e-05, "loss": 0.012, "step": 4890 }, { "epoch": 2.0, "learning_rate": 2.509593602228421e-05, "loss": 0.0165, "step": 4900 }, { "epoch": 2.0, "learning_rate": 2.5015989375200578e-05, "loss": 0.0174, "step": 4910 }, { "epoch": 2.0, "learning_rate": 2.4936042564603233e-05, "loss": 0.0082, "step": 4920 }, { "epoch": 2.01, "learning_rate": 2.4856096408060047e-05, "loss": 0.01, "step": 4930 }, { "epoch": 2.01, "learning_rate": 2.4776151723132187e-05, "loss": 0.0097, "step": 4940 }, { "epoch": 2.02, "learning_rate": 2.4696209327365773e-05, "loss": 0.0141, "step": 4950 }, { "epoch": 2.02, "learning_rate": 2.461627003828352e-05, "loss": 0.0113, "step": 4960 }, { "epoch": 2.02, "learning_rate": 2.4536334673376367e-05, "loss": 0.0132, "step": 4970 }, { "epoch": 2.03, "learning_rate": 2.4456404050095137e-05, "loss": 0.0142, "step": 4980 }, { "epoch": 2.03, "learning_rate": 2.4376478985842134e-05, "loss": 0.0111, "step": 4990 }, { "epoch": 2.04, "learning_rate": 2.4296560297962844e-05, "loss": 0.0118, "step": 5000 }, { "epoch": 2.04, "learning_rate": 2.4216648803737528e-05, "loss": 0.0095, "step": 5010 }, { "epoch": 2.04, "learning_rate": 2.4136745320372904e-05, "loss": 0.0147, "step": 5020 }, { "epoch": 2.05, "learning_rate": 2.4056850664993722e-05, "loss": 0.0118, "step": 5030 }, { "epoch": 2.05, "learning_rate": 2.397696565463449e-05, "loss": 0.0101, "step": 5040 }, { "epoch": 2.06, "learning_rate": 2.389709110623109e-05, "loss": 0.0109, "step": 5050 }, { "epoch": 2.06, "learning_rate": 2.38172278366124e-05, "loss": 0.0127, "step": 5060 }, { "epoch": 2.06, "learning_rate": 2.3737376662491927e-05, "loss": 0.0117, "step": 5070 }, { "epoch": 2.07, "learning_rate": 2.3657538400459548e-05, "loss": 0.0127, "step": 5080 }, { "epoch": 2.07, "learning_rate": 2.3577713866973043e-05, "loss": 0.011, "step": 5090 }, { "epoch": 2.08, "learning_rate": 2.3497903878349835e-05, "loss": 0.011, "step": 5100 }, { "epoch": 2.08, "learning_rate": 2.3418109250758568e-05, "loss": 0.0114, "step": 5110 }, { "epoch": 2.08, "learning_rate": 2.3338330800210828e-05, "loss": 0.0093, "step": 5120 }, { "epoch": 2.09, "learning_rate": 2.325856934255276e-05, "loss": 0.0101, "step": 5130 }, { "epoch": 2.09, "learning_rate": 2.3178825693456734e-05, "loss": 0.012, "step": 5140 }, { "epoch": 2.1, "learning_rate": 2.3099100668412995e-05, "loss": 0.0116, "step": 5150 }, { "epoch": 2.1, "learning_rate": 2.3019395082721327e-05, "loss": 0.0139, "step": 5160 }, { "epoch": 2.11, "learning_rate": 2.293970975148274e-05, "loss": 0.0141, "step": 5170 }, { "epoch": 2.11, "learning_rate": 2.2860045489591113e-05, "loss": 0.0122, "step": 5180 }, { "epoch": 2.11, "learning_rate": 2.2780403111724826e-05, "loss": 0.0125, "step": 5190 }, { "epoch": 2.12, "learning_rate": 2.2700783432338496e-05, "loss": 0.0129, "step": 5200 }, { "epoch": 2.12, "learning_rate": 2.2621187265654622e-05, "loss": 0.011, "step": 5210 }, { "epoch": 2.13, "learning_rate": 2.254161542565524e-05, "loss": 0.0134, "step": 5220 }, { "epoch": 2.13, "learning_rate": 2.2462068726073592e-05, "loss": 0.0139, "step": 5230 }, { "epoch": 2.13, "learning_rate": 2.238254798038586e-05, "loss": 0.014, "step": 5240 }, { "epoch": 2.14, "learning_rate": 2.230305400180279e-05, "loss": 0.0092, "step": 5250 }, { "epoch": 2.14, "learning_rate": 2.2223587603261408e-05, "loss": 0.0126, "step": 5260 }, { "epoch": 2.15, "learning_rate": 2.214414959741668e-05, "loss": 0.0101, "step": 5270 }, { "epoch": 2.15, "learning_rate": 2.206474079663323e-05, "loss": 0.0102, "step": 5280 }, { "epoch": 2.15, "learning_rate": 2.198536201297702e-05, "loss": 0.0139, "step": 5290 }, { "epoch": 2.16, "learning_rate": 2.1906014058207042e-05, "loss": 0.0134, "step": 5300 }, { "epoch": 2.16, "learning_rate": 2.182669774376701e-05, "loss": 0.01, "step": 5310 }, { "epoch": 2.17, "learning_rate": 2.1747413880777083e-05, "loss": 0.01, "step": 5320 }, { "epoch": 2.17, "learning_rate": 2.1668163280025558e-05, "loss": 0.0124, "step": 5330 }, { "epoch": 2.17, "learning_rate": 2.1588946751960582e-05, "loss": 0.0116, "step": 5340 }, { "epoch": 2.18, "learning_rate": 2.150976510668184e-05, "loss": 0.0132, "step": 5350 }, { "epoch": 2.18, "learning_rate": 2.1430619153932314e-05, "loss": 0.0091, "step": 5360 }, { "epoch": 2.19, "learning_rate": 2.135150970308998e-05, "loss": 0.0103, "step": 5370 }, { "epoch": 2.19, "learning_rate": 2.1272437563159533e-05, "loss": 0.0111, "step": 5380 }, { "epoch": 2.19, "learning_rate": 2.1193403542764077e-05, "loss": 0.0114, "step": 5390 }, { "epoch": 2.2, "learning_rate": 2.111440845013693e-05, "loss": 0.0101, "step": 5400 }, { "epoch": 2.2, "learning_rate": 2.1035453093113312e-05, "loss": 0.0112, "step": 5410 }, { "epoch": 2.21, "learning_rate": 2.095653827912208e-05, "loss": 0.0155, "step": 5420 }, { "epoch": 2.21, "learning_rate": 2.0877664815177485e-05, "loss": 0.011, "step": 5430 }, { "epoch": 2.21, "learning_rate": 2.0798833507870918e-05, "loss": 0.0125, "step": 5440 }, { "epoch": 2.22, "learning_rate": 2.0720045163362658e-05, "loss": 0.0142, "step": 5450 }, { "epoch": 2.22, "learning_rate": 2.064130058737364e-05, "loss": 0.0104, "step": 5460 }, { "epoch": 2.23, "learning_rate": 2.056260058517718e-05, "loss": 0.0108, "step": 5470 }, { "epoch": 2.23, "learning_rate": 2.0483945961590788e-05, "loss": 0.0106, "step": 5480 }, { "epoch": 2.24, "learning_rate": 2.040533752096791e-05, "loss": 0.009, "step": 5490 }, { "epoch": 2.24, "learning_rate": 2.03267760671897e-05, "loss": 0.0128, "step": 5500 }, { "epoch": 2.24, "learning_rate": 2.0248262403656804e-05, "loss": 0.0122, "step": 5510 }, { "epoch": 2.25, "learning_rate": 2.0169797333281145e-05, "loss": 0.0119, "step": 5520 }, { "epoch": 2.25, "learning_rate": 2.009138165847775e-05, "loss": 0.0128, "step": 5530 }, { "epoch": 2.26, "learning_rate": 2.0013016181156437e-05, "loss": 0.0094, "step": 5540 }, { "epoch": 2.26, "learning_rate": 1.9934701702713736e-05, "loss": 0.0119, "step": 5550 }, { "epoch": 2.26, "learning_rate": 1.985643902402464e-05, "loss": 0.0084, "step": 5560 }, { "epoch": 2.27, "learning_rate": 1.9778228945434414e-05, "loss": 0.0093, "step": 5570 }, { "epoch": 2.27, "learning_rate": 1.970007226675038e-05, "loss": 0.0119, "step": 5580 }, { "epoch": 2.28, "learning_rate": 1.9621969787233812e-05, "loss": 0.0114, "step": 5590 }, { "epoch": 2.28, "learning_rate": 1.9543922305591704e-05, "loss": 0.0088, "step": 5600 }, { "epoch": 2.28, "learning_rate": 1.9465930619968637e-05, "loss": 0.0125, "step": 5610 }, { "epoch": 2.29, "learning_rate": 1.9387995527938568e-05, "loss": 0.0105, "step": 5620 }, { "epoch": 2.29, "learning_rate": 1.9310117826496728e-05, "loss": 0.0106, "step": 5630 }, { "epoch": 2.3, "learning_rate": 1.923229831205145e-05, "loss": 0.0087, "step": 5640 }, { "epoch": 2.3, "learning_rate": 1.915453778041602e-05, "loss": 0.0144, "step": 5650 }, { "epoch": 2.3, "learning_rate": 1.907683702680053e-05, "loss": 0.0118, "step": 5660 }, { "epoch": 2.31, "learning_rate": 1.899919684580378e-05, "loss": 0.0118, "step": 5670 }, { "epoch": 2.31, "learning_rate": 1.892161803140511e-05, "loss": 0.0106, "step": 5680 }, { "epoch": 2.32, "learning_rate": 1.8844101376956317e-05, "loss": 0.0126, "step": 5690 }, { "epoch": 2.32, "learning_rate": 1.876664767517351e-05, "loss": 0.0087, "step": 5700 }, { "epoch": 2.32, "learning_rate": 1.8689257718129027e-05, "loss": 0.0089, "step": 5710 }, { "epoch": 2.33, "learning_rate": 1.861193229724334e-05, "loss": 0.0107, "step": 5720 }, { "epoch": 2.33, "learning_rate": 1.8534672203276944e-05, "loss": 0.0122, "step": 5730 }, { "epoch": 2.34, "learning_rate": 1.845747822632223e-05, "loss": 0.0101, "step": 5740 }, { "epoch": 2.34, "learning_rate": 1.8380351155795516e-05, "loss": 0.0122, "step": 5750 }, { "epoch": 2.35, "learning_rate": 1.8303291780428876e-05, "loss": 0.0103, "step": 5760 }, { "epoch": 2.35, "learning_rate": 1.822630088826211e-05, "loss": 0.0104, "step": 5770 }, { "epoch": 2.35, "learning_rate": 1.8149379266634674e-05, "loss": 0.0104, "step": 5780 }, { "epoch": 2.36, "learning_rate": 1.807252770217765e-05, "loss": 0.013, "step": 5790 }, { "epoch": 2.36, "learning_rate": 1.7995746980805674e-05, "loss": 0.0095, "step": 5800 }, { "epoch": 2.37, "learning_rate": 1.7919037887708933e-05, "loss": 0.01, "step": 5810 }, { "epoch": 2.37, "learning_rate": 1.7842401207345087e-05, "loss": 0.0097, "step": 5820 }, { "epoch": 2.37, "learning_rate": 1.776583772343129e-05, "loss": 0.0103, "step": 5830 }, { "epoch": 2.38, "learning_rate": 1.7689348218936165e-05, "loss": 0.0091, "step": 5840 }, { "epoch": 2.38, "learning_rate": 1.761293347607179e-05, "loss": 0.0103, "step": 5850 }, { "epoch": 2.39, "learning_rate": 1.7536594276285686e-05, "loss": 0.0109, "step": 5860 }, { "epoch": 2.39, "learning_rate": 1.7460331400252854e-05, "loss": 0.0141, "step": 5870 }, { "epoch": 2.39, "learning_rate": 1.738414562786777e-05, "loss": 0.012, "step": 5880 }, { "epoch": 2.4, "learning_rate": 1.730803773823645e-05, "loss": 0.0115, "step": 5890 }, { "epoch": 2.4, "learning_rate": 1.723200850966838e-05, "loss": 0.0111, "step": 5900 }, { "epoch": 2.41, "learning_rate": 1.71560587196687e-05, "loss": 0.0094, "step": 5910 }, { "epoch": 2.41, "learning_rate": 1.7080189144930135e-05, "loss": 0.0131, "step": 5920 }, { "epoch": 2.41, "learning_rate": 1.700440056132513e-05, "loss": 0.0126, "step": 5930 }, { "epoch": 2.42, "learning_rate": 1.6928693743897833e-05, "loss": 0.0095, "step": 5940 }, { "epoch": 2.42, "learning_rate": 1.685306946685627e-05, "loss": 0.0102, "step": 5950 }, { "epoch": 2.43, "learning_rate": 1.6777528503564353e-05, "loss": 0.0104, "step": 5960 }, { "epoch": 2.43, "learning_rate": 1.6702071626533992e-05, "loss": 0.0094, "step": 5970 }, { "epoch": 2.43, "learning_rate": 1.6626699607417205e-05, "loss": 0.0113, "step": 5980 }, { "epoch": 2.44, "learning_rate": 1.6551413216998207e-05, "loss": 0.0114, "step": 5990 }, { "epoch": 2.44, "learning_rate": 1.647621322518556e-05, "loss": 0.0114, "step": 6000 }, { "epoch": 2.45, "learning_rate": 1.6401100401004274e-05, "loss": 0.0134, "step": 6010 }, { "epoch": 2.45, "learning_rate": 1.6326075512587925e-05, "loss": 0.0145, "step": 6020 }, { "epoch": 2.46, "learning_rate": 1.625113932717085e-05, "loss": 0.0124, "step": 6030 }, { "epoch": 2.46, "learning_rate": 1.617629261108026e-05, "loss": 0.0119, "step": 6040 }, { "epoch": 2.46, "learning_rate": 1.610153612972844e-05, "loss": 0.012, "step": 6050 }, { "epoch": 2.47, "learning_rate": 1.6026870647604852e-05, "loss": 0.0112, "step": 6060 }, { "epoch": 2.47, "learning_rate": 1.5952296928268413e-05, "loss": 0.0094, "step": 6070 }, { "epoch": 2.48, "learning_rate": 1.5877815734339608e-05, "loss": 0.0099, "step": 6080 }, { "epoch": 2.48, "learning_rate": 1.5803427827492736e-05, "loss": 0.0121, "step": 6090 }, { "epoch": 2.48, "learning_rate": 1.572913396844809e-05, "loss": 0.0091, "step": 6100 }, { "epoch": 2.49, "learning_rate": 1.56549349169642e-05, "loss": 0.0093, "step": 6110 }, { "epoch": 2.49, "learning_rate": 1.558083143183006e-05, "loss": 0.0124, "step": 6120 }, { "epoch": 2.5, "learning_rate": 1.550682427085737e-05, "loss": 0.014, "step": 6130 }, { "epoch": 2.5, "learning_rate": 1.5432914190872757e-05, "loss": 0.0118, "step": 6140 }, { "epoch": 2.5, "learning_rate": 1.5359101947710087e-05, "loss": 0.0099, "step": 6150 }, { "epoch": 2.51, "learning_rate": 1.528538829620269e-05, "loss": 0.0109, "step": 6160 }, { "epoch": 2.51, "learning_rate": 1.5211773990175659e-05, "loss": 0.0105, "step": 6170 }, { "epoch": 2.52, "learning_rate": 1.5138259782438152e-05, "loss": 0.0135, "step": 6180 }, { "epoch": 2.52, "learning_rate": 1.5064846424775677e-05, "loss": 0.0199, "step": 6190 }, { "epoch": 2.52, "learning_rate": 1.4991534667942408e-05, "loss": 0.0147, "step": 6200 }, { "epoch": 2.53, "learning_rate": 1.4918325261653505e-05, "loss": 0.0124, "step": 6210 }, { "epoch": 2.53, "learning_rate": 1.484521895457745e-05, "loss": 0.0109, "step": 6220 }, { "epoch": 2.54, "learning_rate": 1.4772216494328395e-05, "loss": 0.011, "step": 6230 }, { "epoch": 2.54, "learning_rate": 1.469931862745853e-05, "loss": 0.0081, "step": 6240 }, { "epoch": 2.54, "learning_rate": 1.4626526099450393e-05, "loss": 0.0112, "step": 6250 }, { "epoch": 2.55, "learning_rate": 1.4553839654709322e-05, "loss": 0.0124, "step": 6260 }, { "epoch": 2.55, "learning_rate": 1.4481260036555787e-05, "loss": 0.01, "step": 6270 }, { "epoch": 2.56, "learning_rate": 1.4408787987217814e-05, "loss": 0.0129, "step": 6280 }, { "epoch": 2.56, "learning_rate": 1.4336424247823366e-05, "loss": 0.0113, "step": 6290 }, { "epoch": 2.57, "learning_rate": 1.4264169558392826e-05, "loss": 0.0109, "step": 6300 }, { "epoch": 2.57, "learning_rate": 1.4192024657831343e-05, "loss": 0.0101, "step": 6310 }, { "epoch": 2.57, "learning_rate": 1.4119990283921366e-05, "loss": 0.0075, "step": 6320 }, { "epoch": 2.58, "learning_rate": 1.4048067173315016e-05, "loss": 0.0088, "step": 6330 }, { "epoch": 2.58, "learning_rate": 1.3976256061526622e-05, "loss": 0.0123, "step": 6340 }, { "epoch": 2.59, "learning_rate": 1.390455768292514e-05, "loss": 0.0101, "step": 6350 }, { "epoch": 2.59, "learning_rate": 1.3832972770726705e-05, "loss": 0.0084, "step": 6360 }, { "epoch": 2.59, "learning_rate": 1.376150205698707e-05, "loss": 0.0123, "step": 6370 }, { "epoch": 2.6, "learning_rate": 1.3690146272594146e-05, "loss": 0.0138, "step": 6380 }, { "epoch": 2.6, "learning_rate": 1.3618906147260568e-05, "loss": 0.013, "step": 6390 }, { "epoch": 2.61, "learning_rate": 1.354778240951617e-05, "loss": 0.0108, "step": 6400 }, { "epoch": 2.61, "learning_rate": 1.3476775786700535e-05, "loss": 0.0102, "step": 6410 }, { "epoch": 2.61, "learning_rate": 1.3405887004955647e-05, "loss": 0.0118, "step": 6420 }, { "epoch": 2.62, "learning_rate": 1.3335116789218333e-05, "loss": 0.0111, "step": 6430 }, { "epoch": 2.62, "learning_rate": 1.3264465863212968e-05, "loss": 0.0114, "step": 6440 }, { "epoch": 2.63, "learning_rate": 1.3193934949443993e-05, "loss": 0.0118, "step": 6450 }, { "epoch": 2.63, "learning_rate": 1.3123524769188556e-05, "loss": 0.0093, "step": 6460 }, { "epoch": 2.63, "learning_rate": 1.3053236042489154e-05, "loss": 0.0121, "step": 6470 }, { "epoch": 2.64, "learning_rate": 1.2983069488146263e-05, "loss": 0.0137, "step": 6480 }, { "epoch": 2.64, "learning_rate": 1.2913025823710915e-05, "loss": 0.0094, "step": 6490 }, { "epoch": 2.65, "learning_rate": 1.2843105765477479e-05, "loss": 0.0117, "step": 6500 }, { "epoch": 2.65, "learning_rate": 1.2773310028476273e-05, "loss": 0.0114, "step": 6510 }, { "epoch": 2.65, "learning_rate": 1.2703639326466235e-05, "loss": 0.0121, "step": 6520 }, { "epoch": 2.66, "learning_rate": 1.2634094371927644e-05, "loss": 0.014, "step": 6530 }, { "epoch": 2.66, "learning_rate": 1.2564675876054863e-05, "loss": 0.0126, "step": 6540 }, { "epoch": 2.67, "learning_rate": 1.2495384548749004e-05, "loss": 0.0102, "step": 6550 }, { "epoch": 2.67, "learning_rate": 1.2426221098610741e-05, "loss": 0.0125, "step": 6560 }, { "epoch": 2.68, "learning_rate": 1.235718623293299e-05, "loss": 0.0113, "step": 6570 }, { "epoch": 2.68, "learning_rate": 1.2288280657693718e-05, "loss": 0.009, "step": 6580 }, { "epoch": 2.68, "learning_rate": 1.2219505077548745e-05, "loss": 0.0108, "step": 6590 }, { "epoch": 2.69, "learning_rate": 1.215086019582448e-05, "loss": 0.0083, "step": 6600 }, { "epoch": 2.69, "learning_rate": 1.208234671451075e-05, "loss": 0.0122, "step": 6610 }, { "epoch": 2.7, "learning_rate": 1.2013965334253671e-05, "loss": 0.012, "step": 6620 }, { "epoch": 2.7, "learning_rate": 1.1945716754348397e-05, "loss": 0.0078, "step": 6630 }, { "epoch": 2.7, "learning_rate": 1.1877601672732048e-05, "loss": 0.0104, "step": 6640 }, { "epoch": 2.71, "learning_rate": 1.1809620785976511e-05, "loss": 0.0113, "step": 6650 }, { "epoch": 2.71, "learning_rate": 1.1741774789281345e-05, "loss": 0.0123, "step": 6660 }, { "epoch": 2.72, "learning_rate": 1.1674064376466687e-05, "loss": 0.0113, "step": 6670 }, { "epoch": 2.72, "learning_rate": 1.1606490239966139e-05, "loss": 0.0082, "step": 6680 }, { "epoch": 2.72, "learning_rate": 1.1539053070819672e-05, "loss": 0.01, "step": 6690 }, { "epoch": 2.73, "learning_rate": 1.1471753558666568e-05, "loss": 0.0127, "step": 6700 }, { "epoch": 2.73, "learning_rate": 1.1404592391738405e-05, "loss": 0.0082, "step": 6710 }, { "epoch": 2.74, "learning_rate": 1.1337570256851963e-05, "loss": 0.0092, "step": 6720 }, { "epoch": 2.74, "learning_rate": 1.1270687839402217e-05, "loss": 0.0093, "step": 6730 }, { "epoch": 2.74, "learning_rate": 1.1203945823355372e-05, "loss": 0.0098, "step": 6740 }, { "epoch": 2.75, "learning_rate": 1.1137344891241794e-05, "loss": 0.0092, "step": 6750 }, { "epoch": 2.75, "learning_rate": 1.1070885724149102e-05, "loss": 0.0136, "step": 6760 }, { "epoch": 2.76, "learning_rate": 1.1004569001715145e-05, "loss": 0.0101, "step": 6770 }, { "epoch": 2.76, "learning_rate": 1.0938395402121073e-05, "loss": 0.0116, "step": 6780 }, { "epoch": 2.76, "learning_rate": 1.0872365602084436e-05, "loss": 0.0073, "step": 6790 }, { "epoch": 2.77, "learning_rate": 1.0806480276852205e-05, "loss": 0.0119, "step": 6800 }, { "epoch": 2.77, "learning_rate": 1.074074010019389e-05, "loss": 0.012, "step": 6810 }, { "epoch": 2.78, "learning_rate": 1.0675145744394677e-05, "loss": 0.0087, "step": 6820 }, { "epoch": 2.78, "learning_rate": 1.0609697880248537e-05, "loss": 0.0095, "step": 6830 }, { "epoch": 2.79, "learning_rate": 1.0544397177051304e-05, "loss": 0.0113, "step": 6840 }, { "epoch": 2.79, "learning_rate": 1.0479244302593942e-05, "loss": 0.0109, "step": 6850 }, { "epoch": 2.79, "learning_rate": 1.0414239923155647e-05, "loss": 0.0091, "step": 6860 }, { "epoch": 2.8, "learning_rate": 1.0349384703497033e-05, "loss": 0.0128, "step": 6870 }, { "epoch": 2.8, "learning_rate": 1.0284679306853343e-05, "loss": 0.014, "step": 6880 }, { "epoch": 2.81, "learning_rate": 1.0220124394927698e-05, "loss": 0.0098, "step": 6890 }, { "epoch": 2.81, "learning_rate": 1.0155720627884268e-05, "loss": 0.0101, "step": 6900 }, { "epoch": 2.81, "learning_rate": 1.0091468664341589e-05, "loss": 0.0077, "step": 6910 }, { "epoch": 2.82, "learning_rate": 1.0027369161365768e-05, "loss": 0.0104, "step": 6920 }, { "epoch": 2.82, "learning_rate": 9.963422774463791e-06, "loss": 0.0145, "step": 6930 }, { "epoch": 2.83, "learning_rate": 9.899630157576844e-06, "loss": 0.0107, "step": 6940 }, { "epoch": 2.83, "learning_rate": 9.835991963073572e-06, "loss": 0.0119, "step": 6950 }, { "epoch": 2.83, "learning_rate": 9.772508841743435e-06, "loss": 0.0132, "step": 6960 }, { "epoch": 2.84, "learning_rate": 9.709181442790086e-06, "loss": 0.0108, "step": 6970 }, { "epoch": 2.84, "learning_rate": 9.646010413824653e-06, "loss": 0.0128, "step": 6980 }, { "epoch": 2.85, "learning_rate": 9.582996400859203e-06, "loss": 0.0099, "step": 6990 }, { "epoch": 2.85, "learning_rate": 9.520140048300072e-06, "loss": 0.0115, "step": 7000 }, { "epoch": 2.85, "learning_rate": 9.457441998941297e-06, "loss": 0.0102, "step": 7010 }, { "epoch": 2.86, "learning_rate": 9.394902893958052e-06, "loss": 0.0086, "step": 7020 }, { "epoch": 2.86, "learning_rate": 9.332523372900104e-06, "loss": 0.0117, "step": 7030 }, { "epoch": 2.87, "learning_rate": 9.270304073685193e-06, "loss": 0.0093, "step": 7040 }, { "epoch": 2.87, "learning_rate": 9.208245632592615e-06, "loss": 0.0119, "step": 7050 }, { "epoch": 2.87, "learning_rate": 9.146348684256666e-06, "loss": 0.0079, "step": 7060 }, { "epoch": 2.88, "learning_rate": 9.084613861660132e-06, "loss": 0.0152, "step": 7070 }, { "epoch": 2.88, "learning_rate": 9.023041796127838e-06, "loss": 0.0096, "step": 7080 }, { "epoch": 2.89, "learning_rate": 8.961633117320212e-06, "loss": 0.0098, "step": 7090 }, { "epoch": 2.89, "learning_rate": 8.900388453226794e-06, "loss": 0.01, "step": 7100 }, { "epoch": 2.89, "learning_rate": 8.839308430159879e-06, "loss": 0.0118, "step": 7110 }, { "epoch": 2.9, "learning_rate": 8.77839367274805e-06, "loss": 0.0105, "step": 7120 }, { "epoch": 2.9, "learning_rate": 8.717644803929822e-06, "loss": 0.0117, "step": 7130 }, { "epoch": 2.91, "learning_rate": 8.657062444947276e-06, "loss": 0.0094, "step": 7140 }, { "epoch": 2.91, "learning_rate": 8.59664721533971e-06, "loss": 0.0103, "step": 7150 }, { "epoch": 2.92, "learning_rate": 8.536399732937244e-06, "loss": 0.0094, "step": 7160 }, { "epoch": 2.92, "learning_rate": 8.476320613854594e-06, "loss": 0.0148, "step": 7170 }, { "epoch": 2.92, "learning_rate": 8.416410472484717e-06, "loss": 0.0098, "step": 7180 }, { "epoch": 2.93, "learning_rate": 8.356669921492525e-06, "loss": 0.01, "step": 7190 }, { "epoch": 2.93, "learning_rate": 8.297099571808625e-06, "loss": 0.012, "step": 7200 }, { "epoch": 2.94, "learning_rate": 8.237700032623109e-06, "loss": 0.0099, "step": 7210 }, { "epoch": 2.94, "learning_rate": 8.178471911379254e-06, "loss": 0.0087, "step": 7220 }, { "epoch": 2.94, "learning_rate": 8.119415813767383e-06, "loss": 0.0091, "step": 7230 }, { "epoch": 2.95, "learning_rate": 8.060532343718619e-06, "loss": 0.0101, "step": 7240 }, { "epoch": 2.95, "learning_rate": 8.001822103398718e-06, "loss": 0.0114, "step": 7250 }, { "epoch": 2.96, "learning_rate": 7.943285693201948e-06, "loss": 0.0069, "step": 7260 }, { "epoch": 2.96, "learning_rate": 7.884923711744898e-06, "loss": 0.0103, "step": 7270 }, { "epoch": 2.96, "learning_rate": 7.826736755860382e-06, "loss": 0.0123, "step": 7280 }, { "epoch": 2.97, "learning_rate": 7.768725420591346e-06, "loss": 0.0097, "step": 7290 }, { "epoch": 2.97, "learning_rate": 7.710890299184753e-06, "loss": 0.0111, "step": 7300 }, { "epoch": 2.98, "learning_rate": 7.65323198308555e-06, "loss": 0.0096, "step": 7310 }, { "epoch": 2.98, "learning_rate": 7.595751061930592e-06, "loss": 0.011, "step": 7320 }, { "epoch": 2.98, "learning_rate": 7.5384481235426126e-06, "loss": 0.0092, "step": 7330 }, { "epoch": 2.99, "learning_rate": 7.481323753924243e-06, "loss": 0.0078, "step": 7340 }, { "epoch": 2.99, "learning_rate": 7.424378537252011e-06, "loss": 0.0107, "step": 7350 }, { "epoch": 3.0, "learning_rate": 7.367613055870301e-06, "loss": 0.0073, "step": 7360 }, { "epoch": 3.0, "learning_rate": 7.311027890285499e-06, "loss": 0.0086, "step": 7370 }, { "epoch": 3.0, "learning_rate": 7.25462361916002e-06, "loss": 0.0106, "step": 7380 }, { "epoch": 3.01, "learning_rate": 7.198400819306325e-06, "loss": 0.0083, "step": 7390 }, { "epoch": 3.01, "learning_rate": 7.142360065681125e-06, "loss": 0.0076, "step": 7400 }, { "epoch": 3.02, "learning_rate": 7.08650193137945e-06, "loss": 0.0065, "step": 7410 }, { "epoch": 3.02, "learning_rate": 7.030826987628772e-06, "loss": 0.0091, "step": 7420 }, { "epoch": 3.03, "learning_rate": 6.97533580378319e-06, "loss": 0.0074, "step": 7430 }, { "epoch": 3.03, "learning_rate": 6.9200289473176195e-06, "loss": 0.007, "step": 7440 }, { "epoch": 3.03, "learning_rate": 6.864906983821937e-06, "loss": 0.0086, "step": 7450 }, { "epoch": 3.04, "learning_rate": 6.80997047699527e-06, "loss": 0.007, "step": 7460 }, { "epoch": 3.04, "learning_rate": 6.75521998864016e-06, "loss": 0.0092, "step": 7470 }, { "epoch": 3.05, "learning_rate": 6.700656078656858e-06, "loss": 0.0065, "step": 7480 }, { "epoch": 3.05, "learning_rate": 6.646279305037601e-06, "loss": 0.0084, "step": 7490 }, { "epoch": 3.05, "learning_rate": 6.5920902238609e-06, "loss": 0.011, "step": 7500 }, { "epoch": 3.06, "learning_rate": 6.538089389285809e-06, "loss": 0.0071, "step": 7510 }, { "epoch": 3.06, "learning_rate": 6.4842773535463416e-06, "loss": 0.0073, "step": 7520 }, { "epoch": 3.07, "learning_rate": 6.4306546669457654e-06, "loss": 0.0074, "step": 7530 }, { "epoch": 3.07, "learning_rate": 6.377221877850984e-06, "loss": 0.0078, "step": 7540 }, { "epoch": 3.07, "learning_rate": 6.323979532686924e-06, "loss": 0.0084, "step": 7550 }, { "epoch": 3.08, "learning_rate": 6.270928175930987e-06, "loss": 0.0103, "step": 7560 }, { "epoch": 3.08, "learning_rate": 6.218068350107414e-06, "loss": 0.0089, "step": 7570 }, { "epoch": 3.09, "learning_rate": 6.165400595781809e-06, "loss": 0.0061, "step": 7580 }, { "epoch": 3.09, "learning_rate": 6.112925451555557e-06, "loss": 0.0102, "step": 7590 }, { "epoch": 3.09, "learning_rate": 6.060643454060324e-06, "loss": 0.0077, "step": 7600 }, { "epoch": 3.1, "learning_rate": 6.00855513795262e-06, "loss": 0.007, "step": 7610 }, { "epoch": 3.1, "learning_rate": 5.956661035908253e-06, "loss": 0.009, "step": 7620 }, { "epoch": 3.11, "learning_rate": 5.9049616786169325e-06, "loss": 0.01, "step": 7630 }, { "epoch": 3.11, "learning_rate": 5.853457594776849e-06, "loss": 0.0077, "step": 7640 }, { "epoch": 3.11, "learning_rate": 5.802149311089217e-06, "loss": 0.0089, "step": 7650 }, { "epoch": 3.12, "learning_rate": 5.751037352252955e-06, "loss": 0.0066, "step": 7660 }, { "epoch": 3.12, "learning_rate": 5.700122240959261e-06, "loss": 0.0069, "step": 7670 }, { "epoch": 3.13, "learning_rate": 5.6494044978862906e-06, "loss": 0.0077, "step": 7680 }, { "epoch": 3.13, "learning_rate": 5.5988846416938525e-06, "loss": 0.0066, "step": 7690 }, { "epoch": 3.14, "learning_rate": 5.548563189018086e-06, "loss": 0.0072, "step": 7700 }, { "epoch": 3.14, "learning_rate": 5.498440654466138e-06, "loss": 0.0086, "step": 7710 }, { "epoch": 3.14, "learning_rate": 5.448517550610988e-06, "loss": 0.0075, "step": 7720 }, { "epoch": 3.15, "learning_rate": 5.3987943879861394e-06, "loss": 0.0091, "step": 7730 }, { "epoch": 3.15, "learning_rate": 5.3492716750804084e-06, "loss": 0.0091, "step": 7740 }, { "epoch": 3.16, "learning_rate": 5.299949918332739e-06, "loss": 0.0072, "step": 7750 }, { "epoch": 3.16, "learning_rate": 5.250829622127027e-06, "loss": 0.0093, "step": 7760 }, { "epoch": 3.16, "learning_rate": 5.201911288786931e-06, "loss": 0.0073, "step": 7770 }, { "epoch": 3.17, "learning_rate": 5.153195418570786e-06, "loss": 0.0099, "step": 7780 }, { "epoch": 3.17, "learning_rate": 5.104682509666437e-06, "loss": 0.0073, "step": 7790 }, { "epoch": 3.18, "learning_rate": 5.056373058186162e-06, "loss": 0.0065, "step": 7800 }, { "epoch": 3.18, "learning_rate": 5.008267558161623e-06, "loss": 0.0077, "step": 7810 }, { "epoch": 3.18, "learning_rate": 4.96036650153879e-06, "loss": 0.0095, "step": 7820 }, { "epoch": 3.19, "learning_rate": 4.91267037817289e-06, "loss": 0.0073, "step": 7830 }, { "epoch": 3.19, "learning_rate": 4.865179675823442e-06, "loss": 0.0069, "step": 7840 }, { "epoch": 3.2, "learning_rate": 4.8178948801492516e-06, "loss": 0.0077, "step": 7850 }, { "epoch": 3.2, "learning_rate": 4.77081647470343e-06, "loss": 0.007, "step": 7860 }, { "epoch": 3.2, "learning_rate": 4.723944940928451e-06, "loss": 0.0069, "step": 7870 }, { "epoch": 3.21, "learning_rate": 4.677280758151265e-06, "loss": 0.0073, "step": 7880 }, { "epoch": 3.21, "learning_rate": 4.6308244035783375e-06, "loss": 0.007, "step": 7890 }, { "epoch": 3.22, "learning_rate": 4.584576352290829e-06, "loss": 0.0081, "step": 7900 }, { "epoch": 3.22, "learning_rate": 4.538537077239683e-06, "loss": 0.0055, "step": 7910 }, { "epoch": 3.22, "learning_rate": 4.4927070492408215e-06, "loss": 0.0074, "step": 7920 }, { "epoch": 3.23, "learning_rate": 4.4470867369703376e-06, "loss": 0.0071, "step": 7930 }, { "epoch": 3.23, "learning_rate": 4.40167660695967e-06, "loss": 0.0107, "step": 7940 }, { "epoch": 3.24, "learning_rate": 4.356477123590852e-06, "loss": 0.0083, "step": 7950 }, { "epoch": 3.24, "learning_rate": 4.311488749091777e-06, "loss": 0.008, "step": 7960 }, { "epoch": 3.25, "learning_rate": 4.266711943531429e-06, "loss": 0.0076, "step": 7970 }, { "epoch": 3.25, "learning_rate": 4.222147164815232e-06, "loss": 0.0085, "step": 7980 }, { "epoch": 3.25, "learning_rate": 4.177794868680318e-06, "loss": 0.0076, "step": 7990 }, { "epoch": 3.26, "learning_rate": 4.13365550869089e-06, "loss": 0.0049, "step": 8000 }, { "epoch": 3.26, "learning_rate": 4.089729536233599e-06, "loss": 0.0081, "step": 8010 }, { "epoch": 3.27, "learning_rate": 4.046017400512894e-06, "loss": 0.0091, "step": 8020 }, { "epoch": 3.27, "learning_rate": 4.0025195485464374e-06, "loss": 0.0063, "step": 8030 }, { "epoch": 3.27, "learning_rate": 3.9592364251605615e-06, "loss": 0.0093, "step": 8040 }, { "epoch": 3.28, "learning_rate": 3.916168472985701e-06, "loss": 0.0094, "step": 8050 }, { "epoch": 3.28, "learning_rate": 3.873316132451826e-06, "loss": 0.0102, "step": 8060 }, { "epoch": 3.29, "learning_rate": 3.830679841784016e-06, "loss": 0.0069, "step": 8070 }, { "epoch": 3.29, "learning_rate": 3.7882600369979286e-06, "loss": 0.0075, "step": 8080 }, { "epoch": 3.29, "learning_rate": 3.746057151895346e-06, "loss": 0.0086, "step": 8090 }, { "epoch": 3.3, "learning_rate": 3.7040716180597345e-06, "loss": 0.0077, "step": 8100 }, { "epoch": 3.3, "learning_rate": 3.6623038648518654e-06, "loss": 0.0066, "step": 8110 }, { "epoch": 3.31, "learning_rate": 3.6248994424735423e-06, "loss": 0.0097, "step": 8120 }, { "epoch": 3.31, "learning_rate": 3.583546647367328e-06, "loss": 0.0079, "step": 8130 }, { "epoch": 3.31, "learning_rate": 3.542412865425107e-06, "loss": 0.0074, "step": 8140 }, { "epoch": 3.32, "learning_rate": 3.501498517297272e-06, "loss": 0.0074, "step": 8150 }, { "epoch": 3.32, "learning_rate": 3.460804021390221e-06, "loss": 0.0099, "step": 8160 }, { "epoch": 3.33, "learning_rate": 3.4203297938620344e-06, "loss": 0.0048, "step": 8170 }, { "epoch": 3.33, "learning_rate": 3.380076248618244e-06, "loss": 0.0055, "step": 8180 }, { "epoch": 3.33, "learning_rate": 3.3400437973076125e-06, "loss": 0.0069, "step": 8190 }, { "epoch": 3.34, "learning_rate": 3.3002328493179052e-06, "loss": 0.0106, "step": 8200 }, { "epoch": 3.34, "learning_rate": 3.260643811771674e-06, "loss": 0.0083, "step": 8210 }, { "epoch": 3.35, "learning_rate": 3.2212770895221706e-06, "loss": 0.0096, "step": 8220 }, { "epoch": 3.35, "learning_rate": 3.1821330851491414e-06, "loss": 0.0065, "step": 8230 }, { "epoch": 3.36, "learning_rate": 3.1432121989547313e-06, "loss": 0.0074, "step": 8240 }, { "epoch": 3.36, "learning_rate": 3.1045148289593843e-06, "loss": 0.0076, "step": 8250 }, { "epoch": 3.36, "learning_rate": 3.0660413708978014e-06, "loss": 0.008, "step": 8260 }, { "epoch": 3.37, "learning_rate": 3.0277922182148393e-06, "loss": 0.0083, "step": 8270 }, { "epoch": 3.37, "learning_rate": 2.9897677620615487e-06, "loss": 0.0047, "step": 8280 }, { "epoch": 3.38, "learning_rate": 2.9519683912911266e-06, "loss": 0.0081, "step": 8290 }, { "epoch": 3.38, "learning_rate": 2.9143944924549553e-06, "loss": 0.0133, "step": 8300 }, { "epoch": 3.38, "learning_rate": 2.8770464497986703e-06, "loss": 0.0076, "step": 8310 }, { "epoch": 3.39, "learning_rate": 2.8399246452581935e-06, "loss": 0.009, "step": 8320 }, { "epoch": 3.39, "learning_rate": 2.80302945845585e-06, "loss": 0.0071, "step": 8330 }, { "epoch": 3.4, "learning_rate": 2.766361266696488e-06, "loss": 0.008, "step": 8340 }, { "epoch": 3.4, "learning_rate": 2.72992044496361e-06, "loss": 0.0074, "step": 8350 }, { "epoch": 3.4, "learning_rate": 2.6937073659155347e-06, "loss": 0.0077, "step": 8360 }, { "epoch": 3.41, "learning_rate": 2.657722399881615e-06, "loss": 0.0068, "step": 8370 }, { "epoch": 3.41, "learning_rate": 2.6219659148584013e-06, "loss": 0.009, "step": 8380 }, { "epoch": 3.42, "learning_rate": 2.586438276505937e-06, "loss": 0.0089, "step": 8390 }, { "epoch": 3.42, "learning_rate": 2.5511398481439576e-06, "loss": 0.0094, "step": 8400 }, { "epoch": 3.42, "learning_rate": 2.5160709907482316e-06, "loss": 0.0062, "step": 8410 }, { "epoch": 3.43, "learning_rate": 2.481232062946823e-06, "loss": 0.0094, "step": 8420 }, { "epoch": 3.43, "learning_rate": 2.446623421016467e-06, "loss": 0.0084, "step": 8430 }, { "epoch": 3.44, "learning_rate": 2.4122454188788807e-06, "loss": 0.0097, "step": 8440 }, { "epoch": 3.44, "learning_rate": 2.378098408097179e-06, "loss": 0.0075, "step": 8450 }, { "epoch": 3.44, "learning_rate": 2.3441827378722693e-06, "loss": 0.0125, "step": 8460 }, { "epoch": 3.45, "learning_rate": 2.310498755039278e-06, "loss": 0.0074, "step": 8470 }, { "epoch": 3.45, "learning_rate": 2.27704680406399e-06, "loss": 0.0094, "step": 8480 }, { "epoch": 3.46, "learning_rate": 2.2438272270393635e-06, "loss": 0.0081, "step": 8490 }, { "epoch": 3.46, "learning_rate": 2.210840363681985e-06, "loss": 0.0085, "step": 8500 }, { "epoch": 3.46, "learning_rate": 2.1780865513286424e-06, "loss": 0.0063, "step": 8510 }, { "epoch": 3.47, "learning_rate": 2.1455661249328325e-06, "loss": 0.0062, "step": 8520 }, { "epoch": 3.47, "learning_rate": 2.113279417061356e-06, "loss": 0.0077, "step": 8530 }, { "epoch": 3.48, "learning_rate": 2.081226757890928e-06, "loss": 0.0068, "step": 8540 }, { "epoch": 3.48, "learning_rate": 2.049408475204784e-06, "loss": 0.0089, "step": 8550 }, { "epoch": 3.49, "learning_rate": 2.017824894389314e-06, "loss": 0.0083, "step": 8560 }, { "epoch": 3.49, "learning_rate": 1.9864763384307763e-06, "loss": 0.0062, "step": 8570 }, { "epoch": 3.49, "learning_rate": 1.9553631279119687e-06, "loss": 0.0084, "step": 8580 }, { "epoch": 3.5, "learning_rate": 1.9244855810089447e-06, "loss": 0.0059, "step": 8590 }, { "epoch": 3.5, "learning_rate": 1.8938440134877665e-06, "loss": 0.0082, "step": 8600 }, { "epoch": 3.51, "learning_rate": 1.8634387387012896e-06, "loss": 0.0074, "step": 8610 }, { "epoch": 3.51, "learning_rate": 1.8332700675859294e-06, "loss": 0.0057, "step": 8620 }, { "epoch": 3.51, "learning_rate": 1.803338308658517e-06, "loss": 0.0075, "step": 8630 }, { "epoch": 3.52, "learning_rate": 1.7736437680131112e-06, "loss": 0.0084, "step": 8640 }, { "epoch": 3.52, "learning_rate": 1.744186749317886e-06, "loss": 0.0101, "step": 8650 }, { "epoch": 3.53, "learning_rate": 1.7149675538120247e-06, "loss": 0.0062, "step": 8660 }, { "epoch": 3.53, "learning_rate": 1.6859864803026314e-06, "loss": 0.0086, "step": 8670 }, { "epoch": 3.53, "learning_rate": 1.6572438251616806e-06, "loss": 0.0088, "step": 8680 }, { "epoch": 3.54, "learning_rate": 1.6287398823229938e-06, "loss": 0.0083, "step": 8690 }, { "epoch": 3.54, "learning_rate": 1.6004749432792122e-06, "loss": 0.0058, "step": 8700 }, { "epoch": 3.55, "learning_rate": 1.5724492970788456e-06, "loss": 0.0078, "step": 8710 }, { "epoch": 3.55, "learning_rate": 1.5446632303232816e-06, "loss": 0.0069, "step": 8720 }, { "epoch": 3.55, "learning_rate": 1.5171170271638864e-06, "loss": 0.0066, "step": 8730 }, { "epoch": 3.56, "learning_rate": 1.4898109692990797e-06, "loss": 0.0066, "step": 8740 }, { "epoch": 3.56, "learning_rate": 1.462745335971466e-06, "loss": 0.0058, "step": 8750 }, { "epoch": 3.57, "learning_rate": 1.4359204039649688e-06, "loss": 0.0071, "step": 8760 }, { "epoch": 3.57, "learning_rate": 1.4093364476019998e-06, "loss": 0.006, "step": 8770 }, { "epoch": 3.57, "learning_rate": 1.3829937387406677e-06, "loss": 0.0071, "step": 8780 }, { "epoch": 3.58, "learning_rate": 1.3568925467719817e-06, "loss": 0.0059, "step": 8790 }, { "epoch": 3.58, "learning_rate": 1.3310331386170993e-06, "loss": 0.011, "step": 8800 }, { "epoch": 3.59, "learning_rate": 1.3054157787246141e-06, "loss": 0.01, "step": 8810 }, { "epoch": 3.59, "learning_rate": 1.2800407290678245e-06, "loss": 0.0066, "step": 8820 }, { "epoch": 3.6, "learning_rate": 1.254908249142081e-06, "loss": 0.0084, "step": 8830 }, { "epoch": 3.6, "learning_rate": 1.230018595962107e-06, "loss": 0.007, "step": 8840 }, { "epoch": 3.6, "learning_rate": 1.2053720240593813e-06, "loss": 0.0067, "step": 8850 }, { "epoch": 3.61, "learning_rate": 1.1809687854795553e-06, "loss": 0.0069, "step": 8860 }, { "epoch": 3.61, "learning_rate": 1.1568091297798483e-06, "loss": 0.0092, "step": 8870 }, { "epoch": 3.62, "learning_rate": 1.1328933040264861e-06, "loss": 0.0089, "step": 8880 }, { "epoch": 3.62, "learning_rate": 1.1092215527922195e-06, "loss": 0.0096, "step": 8890 }, { "epoch": 3.62, "learning_rate": 1.085794118153785e-06, "loss": 0.0071, "step": 8900 }, { "epoch": 3.63, "learning_rate": 1.06261123968944e-06, "loss": 0.0101, "step": 8910 }, { "epoch": 3.63, "learning_rate": 1.0396731544765088e-06, "loss": 0.0072, "step": 8920 }, { "epoch": 3.64, "learning_rate": 1.016980097088971e-06, "loss": 0.0072, "step": 8930 }, { "epoch": 3.64, "learning_rate": 9.945322995950552e-07, "loss": 0.0073, "step": 8940 }, { "epoch": 3.64, "learning_rate": 9.72329991554849e-07, "loss": 0.0092, "step": 8950 }, { "epoch": 3.65, "learning_rate": 9.503734000179926e-07, "loss": 0.0055, "step": 8960 }, { "epoch": 3.65, "learning_rate": 9.286627495213058e-07, "loss": 0.0071, "step": 8970 }, { "epoch": 3.66, "learning_rate": 9.071982620865394e-07, "loss": 0.006, "step": 8980 }, { "epoch": 3.66, "learning_rate": 8.859801572180665e-07, "loss": 0.007, "step": 8990 }, { "epoch": 3.66, "learning_rate": 8.650086519006589e-07, "loss": 0.0068, "step": 9000 }, { "epoch": 3.67, "learning_rate": 8.442839605972724e-07, "loss": 0.0066, "step": 9010 }, { "epoch": 3.67, "learning_rate": 8.238062952468345e-07, "loss": 0.0069, "step": 9020 }, { "epoch": 3.68, "learning_rate": 8.03575865262085e-07, "loss": 0.0096, "step": 9030 }, { "epoch": 3.68, "learning_rate": 7.835928775274504e-07, "loss": 0.008, "step": 9040 }, { "epoch": 3.68, "learning_rate": 7.638575363968975e-07, "loss": 0.0079, "step": 9050 }, { "epoch": 3.69, "learning_rate": 7.443700436918722e-07, "loss": 0.0065, "step": 9060 }, { "epoch": 3.69, "learning_rate": 7.251305986992224e-07, "loss": 0.0075, "step": 9070 }, { "epoch": 3.7, "learning_rate": 7.061393981691533e-07, "loss": 0.0073, "step": 9080 }, { "epoch": 3.7, "learning_rate": 6.873966363132339e-07, "loss": 0.0082, "step": 9090 }, { "epoch": 3.71, "learning_rate": 6.689025048023962e-07, "loss": 0.0093, "step": 9100 }, { "epoch": 3.71, "learning_rate": 6.506571927649808e-07, "loss": 0.0077, "step": 9110 }, { "epoch": 3.71, "learning_rate": 6.326608867847944e-07, "loss": 0.0075, "step": 9120 }, { "epoch": 3.72, "learning_rate": 6.149137708992197e-07, "loss": 0.0102, "step": 9130 }, { "epoch": 3.72, "learning_rate": 5.974160265973195e-07, "loss": 0.0059, "step": 9140 }, { "epoch": 3.73, "learning_rate": 5.801678328179739e-07, "loss": 0.0085, "step": 9150 }, { "epoch": 3.73, "learning_rate": 5.631693659480741e-07, "loss": 0.0073, "step": 9160 }, { "epoch": 3.73, "learning_rate": 5.464207998206933e-07, "loss": 0.0071, "step": 9170 }, { "epoch": 3.74, "learning_rate": 5.299223057133318e-07, "loss": 0.0063, "step": 9180 }, { "epoch": 3.74, "learning_rate": 5.136740523461442e-07, "loss": 0.0067, "step": 9190 }, { "epoch": 3.75, "learning_rate": 4.976762058802265e-07, "loss": 0.0088, "step": 9200 }, { "epoch": 3.75, "learning_rate": 4.819289299159146e-07, "loss": 0.0077, "step": 9210 }, { "epoch": 3.75, "learning_rate": 4.6643238549111955e-07, "loss": 0.007, "step": 9220 }, { "epoch": 3.76, "learning_rate": 4.5118673107965027e-07, "loss": 0.0094, "step": 9230 }, { "epoch": 3.76, "learning_rate": 4.3619212258963484e-07, "loss": 0.0059, "step": 9240 }, { "epoch": 3.77, "learning_rate": 4.214487133619022e-07, "loss": 0.0071, "step": 9250 }, { "epoch": 3.77, "learning_rate": 4.069566541684111e-07, "loss": 0.0085, "step": 9260 }, { "epoch": 3.77, "learning_rate": 3.9271609321071824e-07, "loss": 0.0061, "step": 9270 }, { "epoch": 3.78, "learning_rate": 3.787271761184652e-07, "loss": 0.0105, "step": 9280 }, { "epoch": 3.78, "learning_rate": 3.649900459478717e-07, "loss": 0.0079, "step": 9290 }, { "epoch": 3.79, "learning_rate": 3.5150484318029195e-07, "loss": 0.0068, "step": 9300 }, { "epoch": 3.79, "learning_rate": 3.382717057207746e-07, "loss": 0.0071, "step": 9310 }, { "epoch": 3.79, "learning_rate": 3.2529076889663557e-07, "loss": 0.0082, "step": 9320 }, { "epoch": 3.8, "learning_rate": 3.125621654560984e-07, "loss": 0.0079, "step": 9330 }, { "epoch": 3.8, "learning_rate": 3.000860255669202e-07, "loss": 0.0085, "step": 9340 }, { "epoch": 3.81, "learning_rate": 2.878624768150623e-07, "loss": 0.0075, "step": 9350 }, { "epoch": 3.81, "learning_rate": 2.758916442033993e-07, "loss": 0.0087, "step": 9360 }, { "epoch": 3.82, "learning_rate": 2.641736501504177e-07, "loss": 0.0085, "step": 9370 }, { "epoch": 3.82, "learning_rate": 2.527086144889862e-07, "loss": 0.0057, "step": 9380 }, { "epoch": 3.82, "learning_rate": 2.414966544651148e-07, "loss": 0.0083, "step": 9390 }, { "epoch": 3.83, "learning_rate": 2.3053788473676175e-07, "loss": 0.0077, "step": 9400 }, { "epoch": 3.83, "learning_rate": 2.198324173726618e-07, "loss": 0.0059, "step": 9410 }, { "epoch": 3.84, "learning_rate": 2.0938036185118026e-07, "loss": 0.0071, "step": 9420 }, { "epoch": 3.84, "learning_rate": 1.9918182505919147e-07, "loss": 0.0081, "step": 9430 }, { "epoch": 3.84, "learning_rate": 1.8923691129098254e-07, "loss": 0.0063, "step": 9440 }, { "epoch": 3.85, "learning_rate": 1.7954572224719855e-07, "loss": 0.0062, "step": 9450 }, { "epoch": 3.85, "learning_rate": 1.701083570337908e-07, "loss": 0.0079, "step": 9460 }, { "epoch": 3.86, "learning_rate": 1.6092491216100346e-07, "loss": 0.0085, "step": 9470 }, { "epoch": 3.86, "learning_rate": 1.5199548154239684e-07, "loss": 0.0069, "step": 9480 }, { "epoch": 3.86, "learning_rate": 1.43320156493873e-07, "loss": 0.0083, "step": 9490 }, { "epoch": 3.87, "learning_rate": 1.3489902573275982e-07, "loss": 0.0073, "step": 9500 }, { "epoch": 3.87, "learning_rate": 1.267321753768813e-07, "loss": 0.0084, "step": 9510 }, { "epoch": 3.88, "learning_rate": 1.1881968894369988e-07, "loss": 0.0057, "step": 9520 }, { "epoch": 3.88, "learning_rate": 1.1116164734944756e-07, "loss": 0.0078, "step": 9530 }, { "epoch": 3.88, "learning_rate": 1.0375812890829618e-07, "loss": 0.0106, "step": 9540 }, { "epoch": 3.89, "learning_rate": 9.660920933156903e-08, "loss": 0.0057, "step": 9550 }, { "epoch": 3.89, "learning_rate": 8.971496172696381e-08, "loss": 0.0084, "step": 9560 }, { "epoch": 3.9, "learning_rate": 8.307545659779203e-08, "loss": 0.0075, "step": 9570 }, { "epoch": 3.9, "learning_rate": 7.669076184227408e-08, "loss": 0.0068, "step": 9580 }, { "epoch": 3.9, "learning_rate": 7.056094275283421e-08, "loss": 0.0064, "step": 9590 }, { "epoch": 3.91, "learning_rate": 6.46860620154427e-08, "loss": 0.0078, "step": 9600 }, { "epoch": 3.91, "learning_rate": 5.906617970896089e-08, "loss": 0.008, "step": 9610 }, { "epoch": 3.92, "learning_rate": 5.3701353304538847e-08, "loss": 0.0084, "step": 9620 }, { "epoch": 3.92, "learning_rate": 4.859163766502417e-08, "loss": 0.0076, "step": 9630 }, { "epoch": 3.93, "learning_rate": 4.37370850444041e-08, "loss": 0.0088, "step": 9640 }, { "epoch": 3.93, "learning_rate": 3.913774508725876e-08, "loss": 0.0079, "step": 9650 }, { "epoch": 3.93, "learning_rate": 3.479366482826707e-08, "loss": 0.0065, "step": 9660 }, { "epoch": 3.94, "learning_rate": 3.070488869172106e-08, "loss": 0.0071, "step": 9670 }, { "epoch": 3.94, "learning_rate": 2.6871458491070643e-08, "loss": 0.0073, "step": 9680 }, { "epoch": 3.95, "learning_rate": 2.3293413428496225e-08, "loss": 0.0072, "step": 9690 }, { "epoch": 3.95, "learning_rate": 1.99707900945062e-08, "loss": 0.0085, "step": 9700 }, { "epoch": 3.95, "learning_rate": 1.6903622467573376e-08, "loss": 0.0058, "step": 9710 }, { "epoch": 3.96, "learning_rate": 1.4091941913768614e-08, "loss": 0.0074, "step": 9720 }, { "epoch": 3.96, "learning_rate": 1.1535777186455488e-08, "loss": 0.0078, "step": 9730 }, { "epoch": 3.97, "learning_rate": 9.23515442599332e-09, "loss": 0.0099, "step": 9740 }, { "epoch": 3.97, "learning_rate": 7.190097159459619e-09, "loss": 0.0067, "step": 9750 }, { "epoch": 3.97, "learning_rate": 5.40062630042526e-09, "loss": 0.0075, "step": 9760 }, { "epoch": 3.98, "learning_rate": 3.8667601487324405e-09, "loss": 0.0095, "step": 9770 }, { "epoch": 3.98, "learning_rate": 2.588514390303165e-09, "loss": 0.0084, "step": 9780 }, { "epoch": 3.99, "learning_rate": 1.565902096989369e-09, "loss": 0.0094, "step": 9790 }, { "epoch": 3.99, "learning_rate": 7.989337264258101e-10, "loss": 0.0065, "step": 9800 }, { "epoch": 3.99, "learning_rate": 2.876171219440282e-10, "loss": 0.0058, "step": 9810 }, { "epoch": 4.0, "learning_rate": 3.19575124613225e-11, "loss": 0.0093, "step": 9820 }, { "epoch": 4.0, "step": 9824, "total_flos": 2.0330723092781138e+18, "train_loss": 0.01796033810001285, "train_runtime": 18572.7384, "train_samples_per_second": 16.923, "train_steps_per_second": 0.529 } ], "logging_steps": 10, "max_steps": 9824, "num_train_epochs": 4, "save_steps": 200, "total_flos": 2.0330723092781138e+18, "trial_name": null, "trial_params": null }