{ "best_metric": null, "best_model_checkpoint": null, "epoch": 24.553571428571427, "global_step": 11000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.02, "learning_rate": 3.7174721189591085e-07, "loss": 5.4969, "step": 10 }, { "epoch": 0.04, "learning_rate": 7.434944237918217e-07, "loss": 5.7625, "step": 20 }, { "epoch": 0.07, "learning_rate": 1.1152416356877324e-06, "loss": 5.3281, "step": 30 }, { "epoch": 0.09, "learning_rate": 1.4869888475836434e-06, "loss": 4.8969, "step": 40 }, { "epoch": 0.11, "learning_rate": 1.858736059479554e-06, "loss": 4.0547, "step": 50 }, { "epoch": 0.13, "learning_rate": 2.2304832713754648e-06, "loss": 3.3922, "step": 60 }, { "epoch": 0.16, "learning_rate": 2.6022304832713758e-06, "loss": 3.0609, "step": 70 }, { "epoch": 0.18, "learning_rate": 2.973977695167287e-06, "loss": 2.8453, "step": 80 }, { "epoch": 0.2, "learning_rate": 3.3457249070631974e-06, "loss": 2.3508, "step": 90 }, { "epoch": 0.22, "learning_rate": 3.717472118959108e-06, "loss": 2.4211, "step": 100 }, { "epoch": 0.25, "learning_rate": 4.089219330855019e-06, "loss": 2.2891, "step": 110 }, { "epoch": 0.27, "learning_rate": 4.4609665427509296e-06, "loss": 2.0117, "step": 120 }, { "epoch": 0.29, "learning_rate": 4.83271375464684e-06, "loss": 1.7234, "step": 130 }, { "epoch": 0.31, "learning_rate": 5.2044609665427516e-06, "loss": 2.175, "step": 140 }, { "epoch": 0.33, "learning_rate": 5.576208178438662e-06, "loss": 1.2992, "step": 150 }, { "epoch": 0.36, "learning_rate": 5.947955390334574e-06, "loss": 1.2973, "step": 160 }, { "epoch": 0.38, "learning_rate": 6.319702602230484e-06, "loss": 1.5531, "step": 170 }, { "epoch": 0.4, "learning_rate": 6.691449814126395e-06, "loss": 1.25, "step": 180 }, { "epoch": 0.42, "learning_rate": 7.063197026022306e-06, "loss": 1.1742, "step": 190 }, { "epoch": 0.45, "learning_rate": 7.434944237918216e-06, "loss": 1.3242, "step": 200 }, { "epoch": 0.47, "learning_rate": 7.806691449814127e-06, "loss": 1.234, "step": 210 }, { "epoch": 0.49, "learning_rate": 8.178438661710038e-06, "loss": 1.1578, "step": 220 }, { "epoch": 0.51, "learning_rate": 8.550185873605949e-06, "loss": 1.2152, "step": 230 }, { "epoch": 0.54, "learning_rate": 8.921933085501859e-06, "loss": 1.2215, "step": 240 }, { "epoch": 0.56, "learning_rate": 9.29368029739777e-06, "loss": 1.4211, "step": 250 }, { "epoch": 0.58, "learning_rate": 9.66542750929368e-06, "loss": 1.0383, "step": 260 }, { "epoch": 0.6, "learning_rate": 1.0037174721189591e-05, "loss": 1.0629, "step": 270 }, { "epoch": 0.62, "learning_rate": 1.0408921933085503e-05, "loss": 1.3832, "step": 280 }, { "epoch": 0.65, "learning_rate": 1.0780669144981412e-05, "loss": 1.0078, "step": 290 }, { "epoch": 0.67, "learning_rate": 1.1152416356877324e-05, "loss": 1.8035, "step": 300 }, { "epoch": 0.69, "learning_rate": 1.1524163568773235e-05, "loss": 1.1969, "step": 310 }, { "epoch": 0.71, "learning_rate": 1.1895910780669147e-05, "loss": 0.7822, "step": 320 }, { "epoch": 0.74, "learning_rate": 1.2267657992565056e-05, "loss": 0.943, "step": 330 }, { "epoch": 0.76, "learning_rate": 1.2639405204460968e-05, "loss": 0.9402, "step": 340 }, { "epoch": 0.78, "learning_rate": 1.3011152416356879e-05, "loss": 1.0994, "step": 350 }, { "epoch": 0.8, "learning_rate": 1.338289962825279e-05, "loss": 1.0188, "step": 360 }, { "epoch": 0.83, "learning_rate": 1.37546468401487e-05, "loss": 1.2152, "step": 370 }, { "epoch": 0.85, "learning_rate": 1.4126394052044612e-05, "loss": 1.05, "step": 380 }, { "epoch": 0.87, "learning_rate": 1.4498141263940521e-05, "loss": 1.0902, "step": 390 }, { "epoch": 0.89, "learning_rate": 1.4869888475836432e-05, "loss": 0.9207, "step": 400 }, { "epoch": 0.92, "learning_rate": 1.5241635687732344e-05, "loss": 1.3006, "step": 410 }, { "epoch": 0.94, "learning_rate": 1.5613382899628255e-05, "loss": 0.7423, "step": 420 }, { "epoch": 0.96, "learning_rate": 1.5985130111524165e-05, "loss": 0.7264, "step": 430 }, { "epoch": 0.98, "learning_rate": 1.6356877323420076e-05, "loss": 0.7605, "step": 440 }, { "epoch": 1.0, "learning_rate": 1.6728624535315986e-05, "loss": 1.0223, "step": 450 }, { "epoch": 1.03, "learning_rate": 1.7100371747211897e-05, "loss": 0.6748, "step": 460 }, { "epoch": 1.05, "learning_rate": 1.7472118959107808e-05, "loss": 0.8947, "step": 470 }, { "epoch": 1.07, "learning_rate": 1.7843866171003718e-05, "loss": 0.6562, "step": 480 }, { "epoch": 1.09, "learning_rate": 1.8215613382899632e-05, "loss": 1.0439, "step": 490 }, { "epoch": 1.12, "learning_rate": 1.858736059479554e-05, "loss": 1.1691, "step": 500 }, { "epoch": 1.14, "learning_rate": 1.8959107806691453e-05, "loss": 0.8621, "step": 510 }, { "epoch": 1.16, "learning_rate": 1.933085501858736e-05, "loss": 0.818, "step": 520 }, { "epoch": 1.18, "learning_rate": 1.970260223048327e-05, "loss": 0.6332, "step": 530 }, { "epoch": 1.21, "learning_rate": 1.9999999346673545e-05, "loss": 0.8918, "step": 540 }, { "epoch": 1.23, "learning_rate": 1.9999976480256544e-05, "loss": 0.8881, "step": 550 }, { "epoch": 1.25, "learning_rate": 1.9999920947602093e-05, "loss": 0.8555, "step": 560 }, { "epoch": 1.27, "learning_rate": 1.9999832748891602e-05, "loss": 0.7271, "step": 570 }, { "epoch": 1.29, "learning_rate": 1.999971188441319e-05, "loss": 0.9385, "step": 580 }, { "epoch": 1.32, "learning_rate": 1.999955835456167e-05, "loss": 0.8719, "step": 590 }, { "epoch": 1.34, "learning_rate": 1.9999372159838563e-05, "loss": 0.703, "step": 600 }, { "epoch": 1.36, "learning_rate": 1.9999153300852108e-05, "loss": 0.6535, "step": 610 }, { "epoch": 1.38, "learning_rate": 1.999890177831723e-05, "loss": 0.8262, "step": 620 }, { "epoch": 1.41, "learning_rate": 1.9998617593055565e-05, "loss": 0.783, "step": 630 }, { "epoch": 1.43, "learning_rate": 1.9998300745995437e-05, "loss": 0.6609, "step": 640 }, { "epoch": 1.45, "learning_rate": 1.9997951238171875e-05, "loss": 0.7942, "step": 650 }, { "epoch": 1.47, "learning_rate": 1.9997569070726585e-05, "loss": 0.8227, "step": 660 }, { "epoch": 1.5, "learning_rate": 1.9997154244907972e-05, "loss": 1.009, "step": 670 }, { "epoch": 1.52, "learning_rate": 1.999670676207112e-05, "loss": 0.9586, "step": 680 }, { "epoch": 1.54, "learning_rate": 1.999622662367778e-05, "loss": 0.7836, "step": 690 }, { "epoch": 1.56, "learning_rate": 1.99957138312964e-05, "loss": 0.8104, "step": 700 }, { "epoch": 1.58, "learning_rate": 1.999516838660208e-05, "loss": 0.6854, "step": 710 }, { "epoch": 1.61, "learning_rate": 1.9994590291376585e-05, "loss": 0.7107, "step": 720 }, { "epoch": 1.63, "learning_rate": 1.999397954750834e-05, "loss": 0.6771, "step": 730 }, { "epoch": 1.65, "learning_rate": 1.999333615699242e-05, "loss": 0.8244, "step": 740 }, { "epoch": 1.67, "learning_rate": 1.999266012193054e-05, "loss": 0.6863, "step": 750 }, { "epoch": 1.7, "learning_rate": 1.9991951444531067e-05, "loss": 0.793, "step": 760 }, { "epoch": 1.72, "learning_rate": 1.999121012710898e-05, "loss": 0.8221, "step": 770 }, { "epoch": 1.74, "learning_rate": 1.9990436172085896e-05, "loss": 0.7005, "step": 780 }, { "epoch": 1.76, "learning_rate": 1.9989629581990038e-05, "loss": 0.7057, "step": 790 }, { "epoch": 1.79, "learning_rate": 1.9988790359456236e-05, "loss": 0.8062, "step": 800 }, { "epoch": 1.81, "learning_rate": 1.998791850722593e-05, "loss": 0.9057, "step": 810 }, { "epoch": 1.83, "learning_rate": 1.9987014028147124e-05, "loss": 0.7797, "step": 820 }, { "epoch": 1.85, "learning_rate": 1.9986076925174433e-05, "loss": 0.9398, "step": 830 }, { "epoch": 1.88, "learning_rate": 1.9985107201369024e-05, "loss": 0.6892, "step": 840 }, { "epoch": 1.9, "learning_rate": 1.998410485989862e-05, "loss": 0.6339, "step": 850 }, { "epoch": 1.92, "learning_rate": 1.9983069904037506e-05, "loss": 0.7377, "step": 860 }, { "epoch": 1.94, "learning_rate": 1.99820023371665e-05, "loss": 0.8047, "step": 870 }, { "epoch": 1.96, "learning_rate": 1.9980902162772952e-05, "loss": 0.6106, "step": 880 }, { "epoch": 1.99, "learning_rate": 1.9979769384450728e-05, "loss": 0.7189, "step": 890 }, { "epoch": 2.01, "learning_rate": 1.9978604005900187e-05, "loss": 0.7957, "step": 900 }, { "epoch": 2.03, "learning_rate": 1.9977406030928205e-05, "loss": 0.5467, "step": 910 }, { "epoch": 2.05, "learning_rate": 1.9976175463448113e-05, "loss": 0.7066, "step": 920 }, { "epoch": 2.08, "learning_rate": 1.997491230747973e-05, "loss": 0.5595, "step": 930 }, { "epoch": 2.1, "learning_rate": 1.9973616567149316e-05, "loss": 0.4314, "step": 940 }, { "epoch": 2.12, "learning_rate": 1.9972288246689576e-05, "loss": 0.3828, "step": 950 }, { "epoch": 2.14, "learning_rate": 1.9970927350439647e-05, "loss": 0.7207, "step": 960 }, { "epoch": 2.17, "learning_rate": 1.9969533882845076e-05, "loss": 0.5498, "step": 970 }, { "epoch": 2.19, "learning_rate": 1.9968107848457804e-05, "loss": 0.5375, "step": 980 }, { "epoch": 2.21, "learning_rate": 1.9966649251936164e-05, "loss": 0.6203, "step": 990 }, { "epoch": 2.23, "learning_rate": 1.996515809804485e-05, "loss": 0.6041, "step": 1000 }, { "epoch": 2.25, "learning_rate": 1.9963634391654912e-05, "loss": 0.5675, "step": 1010 }, { "epoch": 2.28, "learning_rate": 1.996207813774374e-05, "loss": 0.5832, "step": 1020 }, { "epoch": 2.3, "learning_rate": 1.9960489341395043e-05, "loss": 0.5791, "step": 1030 }, { "epoch": 2.32, "learning_rate": 1.9958868007798828e-05, "loss": 0.6242, "step": 1040 }, { "epoch": 2.34, "learning_rate": 1.9957214142251392e-05, "loss": 0.4392, "step": 1050 }, { "epoch": 2.37, "learning_rate": 1.9955527750155315e-05, "loss": 0.6139, "step": 1060 }, { "epoch": 2.39, "learning_rate": 1.995380883701941e-05, "loss": 0.401, "step": 1070 }, { "epoch": 2.41, "learning_rate": 1.995205740845874e-05, "loss": 0.5512, "step": 1080 }, { "epoch": 2.43, "learning_rate": 1.9950273470194566e-05, "loss": 0.603, "step": 1090 }, { "epoch": 2.46, "learning_rate": 1.9948457028054364e-05, "loss": 0.5771, "step": 1100 }, { "epoch": 2.48, "learning_rate": 1.994660808797178e-05, "loss": 0.4462, "step": 1110 }, { "epoch": 2.5, "learning_rate": 1.9944726655986618e-05, "loss": 0.576, "step": 1120 }, { "epoch": 2.52, "learning_rate": 1.9942812738244827e-05, "loss": 0.5742, "step": 1130 }, { "epoch": 2.54, "learning_rate": 1.9940866340998464e-05, "loss": 0.4595, "step": 1140 }, { "epoch": 2.57, "learning_rate": 1.99388874706057e-05, "loss": 0.4734, "step": 1150 }, { "epoch": 2.59, "learning_rate": 1.9936876133530766e-05, "loss": 0.5546, "step": 1160 }, { "epoch": 2.61, "learning_rate": 1.9934832336343968e-05, "loss": 0.4874, "step": 1170 }, { "epoch": 2.63, "learning_rate": 1.993275608572163e-05, "loss": 0.4997, "step": 1180 }, { "epoch": 2.66, "learning_rate": 1.99306473884461e-05, "loss": 0.5466, "step": 1190 }, { "epoch": 2.68, "learning_rate": 1.992850625140572e-05, "loss": 0.5512, "step": 1200 }, { "epoch": 2.7, "learning_rate": 1.992633268159479e-05, "loss": 0.3668, "step": 1210 }, { "epoch": 2.72, "learning_rate": 1.992412668611356e-05, "loss": 0.4967, "step": 1220 }, { "epoch": 2.75, "learning_rate": 1.992188827216821e-05, "loss": 0.4352, "step": 1230 }, { "epoch": 2.77, "learning_rate": 1.9919617447070808e-05, "loss": 0.5797, "step": 1240 }, { "epoch": 2.79, "learning_rate": 1.991731421823931e-05, "loss": 0.3668, "step": 1250 }, { "epoch": 2.81, "learning_rate": 1.9914978593197507e-05, "loss": 0.3524, "step": 1260 }, { "epoch": 2.83, "learning_rate": 1.991261057957503e-05, "loss": 0.7638, "step": 1270 }, { "epoch": 2.86, "learning_rate": 1.9910210185107306e-05, "loss": 0.5214, "step": 1280 }, { "epoch": 2.88, "learning_rate": 1.9907777417635537e-05, "loss": 0.5143, "step": 1290 }, { "epoch": 2.9, "learning_rate": 1.9905312285106682e-05, "loss": 0.5393, "step": 1300 }, { "epoch": 2.92, "learning_rate": 1.9902814795573416e-05, "loss": 0.7084, "step": 1310 }, { "epoch": 2.95, "learning_rate": 1.9900284957194115e-05, "loss": 0.4321, "step": 1320 }, { "epoch": 2.97, "learning_rate": 1.9897722778232836e-05, "loss": 0.6334, "step": 1330 }, { "epoch": 2.99, "learning_rate": 1.9895128267059268e-05, "loss": 0.7473, "step": 1340 }, { "epoch": 3.01, "learning_rate": 1.9892501432148728e-05, "loss": 0.4093, "step": 1350 }, { "epoch": 3.04, "learning_rate": 1.988984228208211e-05, "loss": 0.3285, "step": 1360 }, { "epoch": 3.06, "learning_rate": 1.988715082554588e-05, "loss": 0.5238, "step": 1370 }, { "epoch": 3.08, "learning_rate": 1.988442707133204e-05, "loss": 0.4093, "step": 1380 }, { "epoch": 3.1, "learning_rate": 1.988167102833808e-05, "loss": 0.3202, "step": 1390 }, { "epoch": 3.12, "learning_rate": 1.9878882705566988e-05, "loss": 0.3607, "step": 1400 }, { "epoch": 3.15, "learning_rate": 1.9876062112127176e-05, "loss": 0.3255, "step": 1410 }, { "epoch": 3.17, "learning_rate": 1.9873209257232494e-05, "loss": 0.3821, "step": 1420 }, { "epoch": 3.19, "learning_rate": 1.987032415020216e-05, "loss": 0.43, "step": 1430 }, { "epoch": 3.21, "learning_rate": 1.986740680046076e-05, "loss": 0.3271, "step": 1440 }, { "epoch": 3.24, "learning_rate": 1.9864457217538198e-05, "loss": 0.5113, "step": 1450 }, { "epoch": 3.26, "learning_rate": 1.986147541106967e-05, "loss": 0.3202, "step": 1460 }, { "epoch": 3.28, "learning_rate": 1.9858461390795648e-05, "loss": 0.3178, "step": 1470 }, { "epoch": 3.3, "learning_rate": 1.985541516656182e-05, "loss": 0.4218, "step": 1480 }, { "epoch": 3.33, "learning_rate": 1.985233674831908e-05, "loss": 0.3942, "step": 1490 }, { "epoch": 3.35, "learning_rate": 1.984922614612348e-05, "loss": 0.4392, "step": 1500 }, { "epoch": 3.37, "learning_rate": 1.9846083370136217e-05, "loss": 0.4372, "step": 1510 }, { "epoch": 3.39, "learning_rate": 1.984290843062358e-05, "loss": 0.3706, "step": 1520 }, { "epoch": 3.42, "learning_rate": 1.9839701337956922e-05, "loss": 0.4427, "step": 1530 }, { "epoch": 3.44, "learning_rate": 1.9836462102612643e-05, "loss": 0.3347, "step": 1540 }, { "epoch": 3.46, "learning_rate": 1.9833190735172117e-05, "loss": 0.2392, "step": 1550 }, { "epoch": 3.48, "learning_rate": 1.982988724632171e-05, "loss": 0.353, "step": 1560 }, { "epoch": 3.5, "learning_rate": 1.9826551646852696e-05, "loss": 0.4637, "step": 1570 }, { "epoch": 3.53, "learning_rate": 1.982318394766124e-05, "loss": 0.3287, "step": 1580 }, { "epoch": 3.55, "learning_rate": 1.9819784159748394e-05, "loss": 0.382, "step": 1590 }, { "epoch": 3.57, "learning_rate": 1.9816352294219995e-05, "loss": 0.3413, "step": 1600 }, { "epoch": 3.59, "learning_rate": 1.981288836228669e-05, "loss": 0.3233, "step": 1610 }, { "epoch": 3.62, "learning_rate": 1.9809392375263865e-05, "loss": 0.2846, "step": 1620 }, { "epoch": 3.64, "learning_rate": 1.9805864344571625e-05, "loss": 0.3512, "step": 1630 }, { "epoch": 3.66, "learning_rate": 1.980230428173474e-05, "loss": 0.4272, "step": 1640 }, { "epoch": 3.68, "learning_rate": 1.979871219838263e-05, "loss": 0.3992, "step": 1650 }, { "epoch": 3.71, "learning_rate": 1.9795088106249306e-05, "loss": 0.407, "step": 1660 }, { "epoch": 3.73, "learning_rate": 1.9791432017173338e-05, "loss": 0.4494, "step": 1670 }, { "epoch": 3.75, "learning_rate": 1.978774394309782e-05, "loss": 0.2791, "step": 1680 }, { "epoch": 3.77, "learning_rate": 1.9784023896070336e-05, "loss": 0.4276, "step": 1690 }, { "epoch": 3.79, "learning_rate": 1.9780271888242904e-05, "loss": 0.3395, "step": 1700 }, { "epoch": 3.82, "learning_rate": 1.9776487931871958e-05, "loss": 0.3548, "step": 1710 }, { "epoch": 3.84, "learning_rate": 1.9772672039318278e-05, "loss": 0.3003, "step": 1720 }, { "epoch": 3.86, "learning_rate": 1.976882422304699e-05, "loss": 0.415, "step": 1730 }, { "epoch": 3.88, "learning_rate": 1.976494449562748e-05, "loss": 0.3812, "step": 1740 }, { "epoch": 3.91, "learning_rate": 1.9761032869733397e-05, "loss": 0.361, "step": 1750 }, { "epoch": 3.93, "learning_rate": 1.9757089358142573e-05, "loss": 0.4472, "step": 1760 }, { "epoch": 3.95, "learning_rate": 1.9753113973737016e-05, "loss": 0.3828, "step": 1770 }, { "epoch": 3.97, "learning_rate": 1.9749106729502833e-05, "loss": 0.2568, "step": 1780 }, { "epoch": 4.0, "learning_rate": 1.974506763853022e-05, "loss": 0.4671, "step": 1790 }, { "epoch": 4.02, "learning_rate": 1.9740996714013394e-05, "loss": 0.4138, "step": 1800 }, { "epoch": 4.04, "learning_rate": 1.9736893969250567e-05, "loss": 0.2093, "step": 1810 }, { "epoch": 4.06, "learning_rate": 1.97327594176439e-05, "loss": 0.1923, "step": 1820 }, { "epoch": 4.08, "learning_rate": 1.972859307269944e-05, "loss": 0.2213, "step": 1830 }, { "epoch": 4.11, "learning_rate": 1.9724394948027102e-05, "loss": 0.2253, "step": 1840 }, { "epoch": 4.13, "learning_rate": 1.9720165057340616e-05, "loss": 0.3045, "step": 1850 }, { "epoch": 4.15, "learning_rate": 1.971590341445747e-05, "loss": 0.2749, "step": 1860 }, { "epoch": 4.17, "learning_rate": 1.9711610033298888e-05, "loss": 0.2329, "step": 1870 }, { "epoch": 4.2, "learning_rate": 1.9707284927889757e-05, "loss": 0.2785, "step": 1880 }, { "epoch": 4.22, "learning_rate": 1.970292811235861e-05, "loss": 0.2412, "step": 1890 }, { "epoch": 4.24, "learning_rate": 1.969853960093755e-05, "loss": 0.2403, "step": 1900 }, { "epoch": 4.26, "learning_rate": 1.969411940796223e-05, "loss": 0.2864, "step": 1910 }, { "epoch": 4.29, "learning_rate": 1.9689667547871788e-05, "loss": 0.2575, "step": 1920 }, { "epoch": 4.31, "learning_rate": 1.9685184035208814e-05, "loss": 0.2621, "step": 1930 }, { "epoch": 4.33, "learning_rate": 1.968066888461929e-05, "loss": 0.2937, "step": 1940 }, { "epoch": 4.35, "learning_rate": 1.9676122110852547e-05, "loss": 0.2597, "step": 1950 }, { "epoch": 4.38, "learning_rate": 1.9671543728761226e-05, "loss": 0.2299, "step": 1960 }, { "epoch": 4.4, "learning_rate": 1.9666933753301203e-05, "loss": 0.3193, "step": 1970 }, { "epoch": 4.42, "learning_rate": 1.9662292199531575e-05, "loss": 0.2214, "step": 1980 }, { "epoch": 4.44, "learning_rate": 1.9657619082614588e-05, "loss": 0.2271, "step": 1990 }, { "epoch": 4.46, "learning_rate": 1.965291441781559e-05, "loss": 0.2829, "step": 2000 }, { "epoch": 4.49, "learning_rate": 1.964817822050299e-05, "loss": 0.3148, "step": 2010 }, { "epoch": 4.51, "learning_rate": 1.9643410506148196e-05, "loss": 0.2828, "step": 2020 }, { "epoch": 4.53, "learning_rate": 1.9638611290325576e-05, "loss": 0.2846, "step": 2030 }, { "epoch": 4.55, "learning_rate": 1.96337805887124e-05, "loss": 0.2779, "step": 2040 }, { "epoch": 4.58, "learning_rate": 1.9628918417088785e-05, "loss": 0.2877, "step": 2050 }, { "epoch": 4.6, "learning_rate": 1.9624024791337657e-05, "loss": 0.3089, "step": 2060 }, { "epoch": 4.62, "learning_rate": 1.9619099727444686e-05, "loss": 0.3351, "step": 2070 }, { "epoch": 4.64, "learning_rate": 1.961414324149824e-05, "loss": 0.2175, "step": 2080 }, { "epoch": 4.67, "learning_rate": 1.9609155349689338e-05, "loss": 0.3142, "step": 2090 }, { "epoch": 4.69, "learning_rate": 1.9604136068311577e-05, "loss": 0.2378, "step": 2100 }, { "epoch": 4.71, "learning_rate": 1.95990854137611e-05, "loss": 0.3042, "step": 2110 }, { "epoch": 4.73, "learning_rate": 1.959400340253654e-05, "loss": 0.2508, "step": 2120 }, { "epoch": 4.75, "learning_rate": 1.9588890051238945e-05, "loss": 0.1733, "step": 2130 }, { "epoch": 4.78, "learning_rate": 1.9583745376571756e-05, "loss": 0.2595, "step": 2140 }, { "epoch": 4.8, "learning_rate": 1.9578569395340727e-05, "loss": 0.3652, "step": 2150 }, { "epoch": 4.82, "learning_rate": 1.9573362124453884e-05, "loss": 0.3227, "step": 2160 }, { "epoch": 4.84, "learning_rate": 1.9568123580921453e-05, "loss": 0.2818, "step": 2170 }, { "epoch": 4.87, "learning_rate": 1.9562853781855836e-05, "loss": 0.3277, "step": 2180 }, { "epoch": 4.89, "learning_rate": 1.9557552744471517e-05, "loss": 0.2576, "step": 2190 }, { "epoch": 4.91, "learning_rate": 1.9552220486085033e-05, "loss": 0.3362, "step": 2200 }, { "epoch": 4.93, "learning_rate": 1.954685702411491e-05, "loss": 0.3035, "step": 2210 }, { "epoch": 4.96, "learning_rate": 1.9541462376081594e-05, "loss": 0.2588, "step": 2220 }, { "epoch": 4.98, "learning_rate": 1.953603655960742e-05, "loss": 0.2766, "step": 2230 }, { "epoch": 5.0, "learning_rate": 1.9530579592416522e-05, "loss": 0.2879, "step": 2240 }, { "epoch": 5.02, "learning_rate": 1.9525091492334813e-05, "loss": 0.1825, "step": 2250 }, { "epoch": 5.04, "learning_rate": 1.951957227728988e-05, "loss": 0.2012, "step": 2260 }, { "epoch": 5.07, "learning_rate": 1.9514021965310972e-05, "loss": 0.2014, "step": 2270 }, { "epoch": 5.09, "learning_rate": 1.950844057452891e-05, "loss": 0.1917, "step": 2280 }, { "epoch": 5.11, "learning_rate": 1.9502828123176042e-05, "loss": 0.1853, "step": 2290 }, { "epoch": 5.13, "learning_rate": 1.9497184629586176e-05, "loss": 0.1726, "step": 2300 }, { "epoch": 5.16, "learning_rate": 1.949151011219453e-05, "loss": 0.1738, "step": 2310 }, { "epoch": 5.18, "learning_rate": 1.9485804589537655e-05, "loss": 0.2221, "step": 2320 }, { "epoch": 5.2, "learning_rate": 1.9480068080253393e-05, "loss": 0.152, "step": 2330 }, { "epoch": 5.22, "learning_rate": 1.9474300603080805e-05, "loss": 0.1959, "step": 2340 }, { "epoch": 5.25, "learning_rate": 1.9468502176860117e-05, "loss": 0.221, "step": 2350 }, { "epoch": 5.27, "learning_rate": 1.9462672820532643e-05, "loss": 0.2443, "step": 2360 }, { "epoch": 5.29, "learning_rate": 1.9456812553140744e-05, "loss": 0.269, "step": 2370 }, { "epoch": 5.31, "learning_rate": 1.945092139382776e-05, "loss": 0.2092, "step": 2380 }, { "epoch": 5.33, "learning_rate": 1.944499936183793e-05, "loss": 0.2421, "step": 2390 }, { "epoch": 5.36, "learning_rate": 1.9439046476516356e-05, "loss": 0.2037, "step": 2400 }, { "epoch": 5.38, "learning_rate": 1.9433062757308914e-05, "loss": 0.3727, "step": 2410 }, { "epoch": 5.4, "learning_rate": 1.9427048223762212e-05, "loss": 0.1812, "step": 2420 }, { "epoch": 5.42, "learning_rate": 1.9421002895523515e-05, "loss": 0.1861, "step": 2430 }, { "epoch": 5.45, "learning_rate": 1.941492679234068e-05, "loss": 0.1987, "step": 2440 }, { "epoch": 5.47, "learning_rate": 1.9408819934062098e-05, "loss": 0.2456, "step": 2450 }, { "epoch": 5.49, "learning_rate": 1.9402682340636625e-05, "loss": 0.1986, "step": 2460 }, { "epoch": 5.51, "learning_rate": 1.9396514032113514e-05, "loss": 0.1954, "step": 2470 }, { "epoch": 5.54, "learning_rate": 1.9390315028642355e-05, "loss": 0.2107, "step": 2480 }, { "epoch": 5.56, "learning_rate": 1.9384085350473016e-05, "loss": 0.2205, "step": 2490 }, { "epoch": 5.58, "learning_rate": 1.9377825017955548e-05, "loss": 0.1677, "step": 2500 }, { "epoch": 5.6, "learning_rate": 1.9371534051540158e-05, "loss": 0.1757, "step": 2510 }, { "epoch": 5.62, "learning_rate": 1.9365212471777113e-05, "loss": 0.1849, "step": 2520 }, { "epoch": 5.65, "learning_rate": 1.935886029931668e-05, "loss": 0.2591, "step": 2530 }, { "epoch": 5.67, "learning_rate": 1.9352477554909067e-05, "loss": 0.2387, "step": 2540 }, { "epoch": 5.69, "learning_rate": 1.934606425940435e-05, "loss": 0.1754, "step": 2550 }, { "epoch": 5.71, "learning_rate": 1.93396204337524e-05, "loss": 0.1956, "step": 2560 }, { "epoch": 5.74, "learning_rate": 1.9333146099002826e-05, "loss": 0.199, "step": 2570 }, { "epoch": 5.76, "learning_rate": 1.932664127630488e-05, "loss": 0.195, "step": 2580 }, { "epoch": 5.78, "learning_rate": 1.9320105986907433e-05, "loss": 0.27, "step": 2590 }, { "epoch": 5.8, "learning_rate": 1.931354025215886e-05, "loss": 0.2301, "step": 2600 }, { "epoch": 5.83, "learning_rate": 1.9306944093507e-05, "loss": 0.2423, "step": 2610 }, { "epoch": 5.85, "learning_rate": 1.930031753249907e-05, "loss": 0.2396, "step": 2620 }, { "epoch": 5.87, "learning_rate": 1.9293660590781603e-05, "loss": 0.209, "step": 2630 }, { "epoch": 5.89, "learning_rate": 1.928697329010037e-05, "loss": 0.2518, "step": 2640 }, { "epoch": 5.92, "learning_rate": 1.9280255652300326e-05, "loss": 0.2025, "step": 2650 }, { "epoch": 5.94, "learning_rate": 1.9273507699325513e-05, "loss": 0.1896, "step": 2660 }, { "epoch": 5.96, "learning_rate": 1.9266729453219008e-05, "loss": 0.2762, "step": 2670 }, { "epoch": 5.98, "learning_rate": 1.925992093612284e-05, "loss": 0.2312, "step": 2680 }, { "epoch": 6.0, "learning_rate": 1.925308217027792e-05, "loss": 0.2104, "step": 2690 }, { "epoch": 6.03, "learning_rate": 1.924621317802399e-05, "loss": 0.1653, "step": 2700 }, { "epoch": 6.05, "learning_rate": 1.9239313981799507e-05, "loss": 0.2042, "step": 2710 }, { "epoch": 6.07, "learning_rate": 1.92323846041416e-05, "loss": 0.1439, "step": 2720 }, { "epoch": 6.09, "learning_rate": 1.9225425067685995e-05, "loss": 0.1415, "step": 2730 }, { "epoch": 6.12, "learning_rate": 1.9218435395166933e-05, "loss": 0.1458, "step": 2740 }, { "epoch": 6.14, "learning_rate": 1.9211415609417097e-05, "loss": 0.1255, "step": 2750 }, { "epoch": 6.16, "learning_rate": 1.920436573336754e-05, "loss": 0.1407, "step": 2760 }, { "epoch": 6.18, "learning_rate": 1.919728579004761e-05, "loss": 0.1808, "step": 2770 }, { "epoch": 6.21, "learning_rate": 1.919017580258487e-05, "loss": 0.1939, "step": 2780 }, { "epoch": 6.23, "learning_rate": 1.918303579420503e-05, "loss": 0.1663, "step": 2790 }, { "epoch": 6.25, "learning_rate": 1.917586578823186e-05, "loss": 0.2042, "step": 2800 }, { "epoch": 6.27, "learning_rate": 1.916866580808714e-05, "loss": 0.2018, "step": 2810 }, { "epoch": 6.29, "learning_rate": 1.9161435877290538e-05, "loss": 0.1778, "step": 2820 }, { "epoch": 6.32, "learning_rate": 1.915417601945958e-05, "loss": 0.1583, "step": 2830 }, { "epoch": 6.34, "learning_rate": 1.9146886258309548e-05, "loss": 0.1483, "step": 2840 }, { "epoch": 6.36, "learning_rate": 1.9139566617653395e-05, "loss": 0.2115, "step": 2850 }, { "epoch": 6.38, "learning_rate": 1.9132217121401698e-05, "loss": 0.1656, "step": 2860 }, { "epoch": 6.41, "learning_rate": 1.912483779356255e-05, "loss": 0.1889, "step": 2870 }, { "epoch": 6.43, "learning_rate": 1.9117428658241498e-05, "loss": 0.1656, "step": 2880 }, { "epoch": 6.45, "learning_rate": 1.9109989739641446e-05, "loss": 0.1844, "step": 2890 }, { "epoch": 6.47, "learning_rate": 1.9102521062062615e-05, "loss": 0.1918, "step": 2900 }, { "epoch": 6.5, "learning_rate": 1.909502264990241e-05, "loss": 0.1731, "step": 2910 }, { "epoch": 6.52, "learning_rate": 1.9087494527655383e-05, "loss": 0.1824, "step": 2920 }, { "epoch": 6.54, "learning_rate": 1.9079936719913138e-05, "loss": 0.1369, "step": 2930 }, { "epoch": 6.56, "learning_rate": 1.9072349251364238e-05, "loss": 0.2788, "step": 2940 }, { "epoch": 6.58, "learning_rate": 1.906473214679416e-05, "loss": 0.1722, "step": 2950 }, { "epoch": 6.61, "learning_rate": 1.9057085431085163e-05, "loss": 0.1454, "step": 2960 }, { "epoch": 6.63, "learning_rate": 1.904940912921626e-05, "loss": 0.1644, "step": 2970 }, { "epoch": 6.65, "learning_rate": 1.9041703266263095e-05, "loss": 0.2158, "step": 2980 }, { "epoch": 6.67, "learning_rate": 1.9033967867397883e-05, "loss": 0.1916, "step": 2990 }, { "epoch": 6.7, "learning_rate": 1.902620295788932e-05, "loss": 0.2604, "step": 3000 }, { "epoch": 6.72, "learning_rate": 1.9018408563102505e-05, "loss": 0.2261, "step": 3010 }, { "epoch": 6.74, "learning_rate": 1.901058470849885e-05, "loss": 0.1846, "step": 3020 }, { "epoch": 6.76, "learning_rate": 1.900273141963601e-05, "loss": 0.2215, "step": 3030 }, { "epoch": 6.79, "learning_rate": 1.899484872216778e-05, "loss": 0.1609, "step": 3040 }, { "epoch": 6.81, "learning_rate": 1.8986936641844025e-05, "loss": 0.1814, "step": 3050 }, { "epoch": 6.83, "learning_rate": 1.8978995204510605e-05, "loss": 0.1731, "step": 3060 }, { "epoch": 6.85, "learning_rate": 1.897102443610926e-05, "loss": 0.1599, "step": 3070 }, { "epoch": 6.88, "learning_rate": 1.8963024362677557e-05, "loss": 0.1508, "step": 3080 }, { "epoch": 6.9, "learning_rate": 1.895499501034878e-05, "loss": 0.1586, "step": 3090 }, { "epoch": 6.92, "learning_rate": 1.8946936405351877e-05, "loss": 0.1836, "step": 3100 }, { "epoch": 6.94, "learning_rate": 1.893884857401133e-05, "loss": 0.1639, "step": 3110 }, { "epoch": 6.96, "learning_rate": 1.8930731542747108e-05, "loss": 0.1322, "step": 3120 }, { "epoch": 6.99, "learning_rate": 1.8922585338074556e-05, "loss": 0.1665, "step": 3130 }, { "epoch": 7.01, "learning_rate": 1.8914409986604327e-05, "loss": 0.2296, "step": 3140 }, { "epoch": 7.03, "learning_rate": 1.8906205515042272e-05, "loss": 0.166, "step": 3150 }, { "epoch": 7.05, "learning_rate": 1.8897971950189385e-05, "loss": 0.1786, "step": 3160 }, { "epoch": 7.08, "learning_rate": 1.888970931894169e-05, "loss": 0.1708, "step": 3170 }, { "epoch": 7.1, "learning_rate": 1.888141764829015e-05, "loss": 0.1554, "step": 3180 }, { "epoch": 7.12, "learning_rate": 1.8873096965320597e-05, "loss": 0.1067, "step": 3190 }, { "epoch": 7.14, "learning_rate": 1.886474729721364e-05, "loss": 0.1548, "step": 3200 }, { "epoch": 7.17, "learning_rate": 1.8856368671244565e-05, "loss": 0.1609, "step": 3210 }, { "epoch": 7.19, "learning_rate": 1.8847961114783254e-05, "loss": 0.1485, "step": 3220 }, { "epoch": 7.21, "learning_rate": 1.88395246552941e-05, "loss": 0.1499, "step": 3230 }, { "epoch": 7.23, "learning_rate": 1.8831059320335902e-05, "loss": 0.1625, "step": 3240 }, { "epoch": 7.25, "learning_rate": 1.882256513756179e-05, "loss": 0.2173, "step": 3250 }, { "epoch": 7.28, "learning_rate": 1.881404213471913e-05, "loss": 0.1799, "step": 3260 }, { "epoch": 7.3, "learning_rate": 1.8805490339649428e-05, "loss": 0.1554, "step": 3270 }, { "epoch": 7.32, "learning_rate": 1.879690978028825e-05, "loss": 0.1544, "step": 3280 }, { "epoch": 7.34, "learning_rate": 1.8788300484665118e-05, "loss": 0.1347, "step": 3290 }, { "epoch": 7.37, "learning_rate": 1.877966248090343e-05, "loss": 0.1699, "step": 3300 }, { "epoch": 7.39, "learning_rate": 1.8770995797220356e-05, "loss": 0.1382, "step": 3310 }, { "epoch": 7.41, "learning_rate": 1.8762300461926766e-05, "loss": 0.1356, "step": 3320 }, { "epoch": 7.43, "learning_rate": 1.8753576503427107e-05, "loss": 0.134, "step": 3330 }, { "epoch": 7.46, "learning_rate": 1.874482395021934e-05, "loss": 0.1597, "step": 3340 }, { "epoch": 7.48, "learning_rate": 1.8736042830894828e-05, "loss": 0.1383, "step": 3350 }, { "epoch": 7.5, "learning_rate": 1.8727233174138254e-05, "loss": 0.214, "step": 3360 }, { "epoch": 7.52, "learning_rate": 1.871839500872752e-05, "loss": 0.1322, "step": 3370 }, { "epoch": 7.54, "learning_rate": 1.8709528363533653e-05, "loss": 0.1664, "step": 3380 }, { "epoch": 7.57, "learning_rate": 1.8700633267520715e-05, "loss": 0.1222, "step": 3390 }, { "epoch": 7.59, "learning_rate": 1.8691709749745705e-05, "loss": 0.1477, "step": 3400 }, { "epoch": 7.61, "learning_rate": 1.8682757839358472e-05, "loss": 0.1433, "step": 3410 }, { "epoch": 7.63, "learning_rate": 1.86737775656016e-05, "loss": 0.114, "step": 3420 }, { "epoch": 7.66, "learning_rate": 1.866476895781034e-05, "loss": 0.16, "step": 3430 }, { "epoch": 7.68, "learning_rate": 1.8655732045412488e-05, "loss": 0.143, "step": 3440 }, { "epoch": 7.7, "learning_rate": 1.8646666857928314e-05, "loss": 0.1586, "step": 3450 }, { "epoch": 7.72, "learning_rate": 1.8637573424970435e-05, "loss": 0.2475, "step": 3460 }, { "epoch": 7.75, "learning_rate": 1.862845177624375e-05, "loss": 0.1663, "step": 3470 }, { "epoch": 7.77, "learning_rate": 1.8619301941545323e-05, "loss": 0.1411, "step": 3480 }, { "epoch": 7.79, "learning_rate": 1.8610123950764288e-05, "loss": 0.156, "step": 3490 }, { "epoch": 7.81, "learning_rate": 1.8600917833881765e-05, "loss": 0.2059, "step": 3500 }, { "epoch": 7.83, "learning_rate": 1.8591683620970737e-05, "loss": 0.1535, "step": 3510 }, { "epoch": 7.86, "learning_rate": 1.858242134219598e-05, "loss": 0.1605, "step": 3520 }, { "epoch": 7.88, "learning_rate": 1.8573131027813945e-05, "loss": 0.153, "step": 3530 }, { "epoch": 7.9, "learning_rate": 1.856381270817266e-05, "loss": 0.1622, "step": 3540 }, { "epoch": 7.92, "learning_rate": 1.8554466413711644e-05, "loss": 0.1757, "step": 3550 }, { "epoch": 7.95, "learning_rate": 1.8545092174961795e-05, "loss": 0.1776, "step": 3560 }, { "epoch": 7.97, "learning_rate": 1.85356900225453e-05, "loss": 0.1494, "step": 3570 }, { "epoch": 7.99, "learning_rate": 1.852625998717552e-05, "loss": 0.138, "step": 3580 }, { "epoch": 8.01, "learning_rate": 1.8516802099656907e-05, "loss": 0.1342, "step": 3590 }, { "epoch": 8.04, "learning_rate": 1.8507316390884894e-05, "loss": 0.1856, "step": 3600 }, { "epoch": 8.06, "learning_rate": 1.84978028918458e-05, "loss": 0.1542, "step": 3610 }, { "epoch": 8.08, "learning_rate": 1.848826163361671e-05, "loss": 0.132, "step": 3620 }, { "epoch": 8.1, "learning_rate": 1.8478692647365402e-05, "loss": 0.1367, "step": 3630 }, { "epoch": 8.12, "learning_rate": 1.846909596435023e-05, "loss": 0.1024, "step": 3640 }, { "epoch": 8.15, "learning_rate": 1.845947161592002e-05, "loss": 0.1216, "step": 3650 }, { "epoch": 8.17, "learning_rate": 1.844981963351397e-05, "loss": 0.1546, "step": 3660 }, { "epoch": 8.19, "learning_rate": 1.8440140048661547e-05, "loss": 0.1421, "step": 3670 }, { "epoch": 8.21, "learning_rate": 1.843043289298239e-05, "loss": 0.1191, "step": 3680 }, { "epoch": 8.24, "learning_rate": 1.8420698198186197e-05, "loss": 0.1476, "step": 3690 }, { "epoch": 8.26, "learning_rate": 1.841093599607263e-05, "loss": 0.1406, "step": 3700 }, { "epoch": 8.28, "learning_rate": 1.8401146318531204e-05, "loss": 0.1636, "step": 3710 }, { "epoch": 8.3, "learning_rate": 1.8391329197541186e-05, "loss": 0.1358, "step": 3720 }, { "epoch": 8.33, "learning_rate": 1.838148466517149e-05, "loss": 0.1187, "step": 3730 }, { "epoch": 8.35, "learning_rate": 1.8371612753580583e-05, "loss": 0.1813, "step": 3740 }, { "epoch": 8.37, "learning_rate": 1.8361713495016354e-05, "loss": 0.2404, "step": 3750 }, { "epoch": 8.39, "learning_rate": 1.8351786921816037e-05, "loss": 0.1502, "step": 3760 }, { "epoch": 8.42, "learning_rate": 1.8341833066406083e-05, "loss": 0.1166, "step": 3770 }, { "epoch": 8.44, "learning_rate": 1.8331851961302075e-05, "loss": 0.1509, "step": 3780 }, { "epoch": 8.46, "learning_rate": 1.83218436391086e-05, "loss": 0.1222, "step": 3790 }, { "epoch": 8.48, "learning_rate": 1.8311808132519157e-05, "loss": 0.1202, "step": 3800 }, { "epoch": 8.5, "learning_rate": 1.830174547431605e-05, "loss": 0.1354, "step": 3810 }, { "epoch": 8.53, "learning_rate": 1.8291655697370276e-05, "loss": 0.1861, "step": 3820 }, { "epoch": 8.55, "learning_rate": 1.8281538834641416e-05, "loss": 0.1477, "step": 3830 }, { "epoch": 8.57, "learning_rate": 1.8271394919177528e-05, "loss": 0.1162, "step": 3840 }, { "epoch": 8.59, "learning_rate": 1.8261223984115052e-05, "loss": 0.1183, "step": 3850 }, { "epoch": 8.62, "learning_rate": 1.8251026062678673e-05, "loss": 0.168, "step": 3860 }, { "epoch": 8.64, "learning_rate": 1.8240801188181257e-05, "loss": 0.1765, "step": 3870 }, { "epoch": 8.66, "learning_rate": 1.823054939402369e-05, "loss": 0.1573, "step": 3880 }, { "epoch": 8.68, "learning_rate": 1.8220270713694803e-05, "loss": 0.1615, "step": 3890 }, { "epoch": 8.71, "learning_rate": 1.8209965180771262e-05, "loss": 0.1164, "step": 3900 }, { "epoch": 8.73, "learning_rate": 1.8199632828917445e-05, "loss": 0.1042, "step": 3910 }, { "epoch": 8.75, "learning_rate": 1.8189273691885336e-05, "loss": 0.1322, "step": 3920 }, { "epoch": 8.77, "learning_rate": 1.8178887803514415e-05, "loss": 0.1346, "step": 3930 }, { "epoch": 8.79, "learning_rate": 1.8168475197731553e-05, "loss": 0.1488, "step": 3940 }, { "epoch": 8.82, "learning_rate": 1.81580359085509e-05, "loss": 0.15, "step": 3950 }, { "epoch": 8.84, "learning_rate": 1.814756997007376e-05, "loss": 0.1608, "step": 3960 }, { "epoch": 8.86, "learning_rate": 1.8137077416488496e-05, "loss": 0.129, "step": 3970 }, { "epoch": 8.88, "learning_rate": 1.8126558282070417e-05, "loss": 0.1404, "step": 3980 }, { "epoch": 8.91, "learning_rate": 1.8116012601181655e-05, "loss": 0.1542, "step": 3990 }, { "epoch": 8.93, "learning_rate": 1.810544040827107e-05, "loss": 0.1611, "step": 4000 }, { "epoch": 8.95, "learning_rate": 1.8094841737874108e-05, "loss": 0.1089, "step": 4010 }, { "epoch": 8.97, "learning_rate": 1.8084216624612726e-05, "loss": 0.1659, "step": 4020 }, { "epoch": 9.0, "learning_rate": 1.8073565103195254e-05, "loss": 0.2411, "step": 4030 }, { "epoch": 9.02, "learning_rate": 1.8062887208416282e-05, "loss": 0.1263, "step": 4040 }, { "epoch": 9.04, "learning_rate": 1.8052182975156557e-05, "loss": 0.1005, "step": 4050 }, { "epoch": 9.06, "learning_rate": 1.8041452438382873e-05, "loss": 0.1692, "step": 4060 }, { "epoch": 9.08, "learning_rate": 1.8030695633147926e-05, "loss": 0.0977, "step": 4070 }, { "epoch": 9.11, "learning_rate": 1.801991259459024e-05, "loss": 0.0963, "step": 4080 }, { "epoch": 9.13, "learning_rate": 1.8009103357934024e-05, "loss": 0.1449, "step": 4090 }, { "epoch": 9.15, "learning_rate": 1.7998267958489076e-05, "loss": 0.1491, "step": 4100 }, { "epoch": 9.17, "learning_rate": 1.7987406431650653e-05, "loss": 0.0934, "step": 4110 }, { "epoch": 9.2, "learning_rate": 1.797651881289935e-05, "loss": 0.1384, "step": 4120 }, { "epoch": 9.22, "learning_rate": 1.7965605137801015e-05, "loss": 0.167, "step": 4130 }, { "epoch": 9.24, "learning_rate": 1.79546654420066e-05, "loss": 0.1378, "step": 4140 }, { "epoch": 9.26, "learning_rate": 1.7943699761252057e-05, "loss": 0.1286, "step": 4150 }, { "epoch": 9.29, "learning_rate": 1.7932708131358222e-05, "loss": 0.1483, "step": 4160 }, { "epoch": 9.31, "learning_rate": 1.7921690588230698e-05, "loss": 0.1236, "step": 4170 }, { "epoch": 9.33, "learning_rate": 1.7910647167859744e-05, "loss": 0.1181, "step": 4180 }, { "epoch": 9.35, "learning_rate": 1.7899577906320135e-05, "loss": 0.1946, "step": 4190 }, { "epoch": 9.38, "learning_rate": 1.7888482839771074e-05, "loss": 0.1184, "step": 4200 }, { "epoch": 9.4, "learning_rate": 1.787736200445606e-05, "loss": 0.1348, "step": 4210 }, { "epoch": 9.42, "learning_rate": 1.786621543670275e-05, "loss": 0.1342, "step": 4220 }, { "epoch": 9.44, "learning_rate": 1.7855043172922883e-05, "loss": 0.1261, "step": 4230 }, { "epoch": 9.46, "learning_rate": 1.7843845249612122e-05, "loss": 0.1277, "step": 4240 }, { "epoch": 9.49, "learning_rate": 1.7832621703349956e-05, "loss": 0.1325, "step": 4250 }, { "epoch": 9.51, "learning_rate": 1.7821372570799574e-05, "loss": 0.0981, "step": 4260 }, { "epoch": 9.53, "learning_rate": 1.781009788870775e-05, "loss": 0.137, "step": 4270 }, { "epoch": 9.55, "learning_rate": 1.779879769390471e-05, "loss": 0.103, "step": 4280 }, { "epoch": 9.58, "learning_rate": 1.7787472023304023e-05, "loss": 0.1506, "step": 4290 }, { "epoch": 9.6, "learning_rate": 1.7776120913902487e-05, "loss": 0.134, "step": 4300 }, { "epoch": 9.62, "learning_rate": 1.7764744402779992e-05, "loss": 0.1274, "step": 4310 }, { "epoch": 9.64, "learning_rate": 1.77533425270994e-05, "loss": 0.1171, "step": 4320 }, { "epoch": 9.67, "learning_rate": 1.7741915324106445e-05, "loss": 0.128, "step": 4330 }, { "epoch": 9.69, "learning_rate": 1.7730462831129584e-05, "loss": 0.1936, "step": 4340 }, { "epoch": 9.71, "learning_rate": 1.771898508557989e-05, "loss": 0.1356, "step": 4350 }, { "epoch": 9.73, "learning_rate": 1.7707482124950923e-05, "loss": 0.1126, "step": 4360 }, { "epoch": 9.75, "learning_rate": 1.7695953986818625e-05, "loss": 0.1356, "step": 4370 }, { "epoch": 9.78, "learning_rate": 1.7684400708841165e-05, "loss": 0.135, "step": 4380 }, { "epoch": 9.8, "learning_rate": 1.7672822328758852e-05, "loss": 0.0998, "step": 4390 }, { "epoch": 9.82, "learning_rate": 1.7661218884393977e-05, "loss": 0.1443, "step": 4400 }, { "epoch": 9.84, "learning_rate": 1.764959041365073e-05, "loss": 0.1485, "step": 4410 }, { "epoch": 9.87, "learning_rate": 1.7637936954515026e-05, "loss": 0.1455, "step": 4420 }, { "epoch": 9.89, "learning_rate": 1.7626258545054425e-05, "loss": 0.1356, "step": 4430 }, { "epoch": 9.91, "learning_rate": 1.7614555223417987e-05, "loss": 0.1497, "step": 4440 }, { "epoch": 9.93, "learning_rate": 1.7602827027836153e-05, "loss": 0.1463, "step": 4450 }, { "epoch": 9.96, "learning_rate": 1.7591073996620607e-05, "loss": 0.1413, "step": 4460 }, { "epoch": 9.98, "learning_rate": 1.757929616816418e-05, "loss": 0.1611, "step": 4470 }, { "epoch": 10.0, "learning_rate": 1.756749358094069e-05, "loss": 0.1405, "step": 4480 }, { "epoch": 10.02, "learning_rate": 1.755566627350484e-05, "loss": 0.0849, "step": 4490 }, { "epoch": 10.04, "learning_rate": 1.754381428449209e-05, "loss": 0.1303, "step": 4500 }, { "epoch": 10.07, "learning_rate": 1.7531937652618515e-05, "loss": 0.126, "step": 4510 }, { "epoch": 10.09, "learning_rate": 1.7520036416680687e-05, "loss": 0.0996, "step": 4520 }, { "epoch": 10.11, "learning_rate": 1.7508110615555573e-05, "loss": 0.0939, "step": 4530 }, { "epoch": 10.13, "learning_rate": 1.749616028820036e-05, "loss": 0.133, "step": 4540 }, { "epoch": 10.16, "learning_rate": 1.748418547365236e-05, "loss": 0.1169, "step": 4550 }, { "epoch": 10.18, "learning_rate": 1.7472186211028884e-05, "loss": 0.1122, "step": 4560 }, { "epoch": 10.2, "learning_rate": 1.7460162539527104e-05, "loss": 0.1198, "step": 4570 }, { "epoch": 10.22, "learning_rate": 1.7448114498423915e-05, "loss": 0.0999, "step": 4580 }, { "epoch": 10.25, "learning_rate": 1.743604212707583e-05, "loss": 0.1395, "step": 4590 }, { "epoch": 10.27, "learning_rate": 1.7423945464918835e-05, "loss": 0.099, "step": 4600 }, { "epoch": 10.29, "learning_rate": 1.741182455146827e-05, "loss": 0.165, "step": 4610 }, { "epoch": 10.31, "learning_rate": 1.739967942631869e-05, "loss": 0.0973, "step": 4620 }, { "epoch": 10.33, "learning_rate": 1.738751012914375e-05, "loss": 0.1153, "step": 4630 }, { "epoch": 10.36, "learning_rate": 1.7375316699696042e-05, "loss": 0.1615, "step": 4640 }, { "epoch": 10.38, "learning_rate": 1.736309917780702e-05, "loss": 0.1608, "step": 4650 }, { "epoch": 10.4, "learning_rate": 1.7350857603386816e-05, "loss": 0.1321, "step": 4660 }, { "epoch": 10.42, "learning_rate": 1.733859201642415e-05, "loss": 0.1102, "step": 4670 }, { "epoch": 10.45, "learning_rate": 1.732630245698617e-05, "loss": 0.1674, "step": 4680 }, { "epoch": 10.47, "learning_rate": 1.7313988965218337e-05, "loss": 0.155, "step": 4690 }, { "epoch": 10.49, "learning_rate": 1.730165158134429e-05, "loss": 0.1256, "step": 4700 }, { "epoch": 10.51, "learning_rate": 1.7289290345665713e-05, "loss": 0.1126, "step": 4710 }, { "epoch": 10.54, "learning_rate": 1.7276905298562208e-05, "loss": 0.1321, "step": 4720 }, { "epoch": 10.56, "learning_rate": 1.7264496480491165e-05, "loss": 0.1292, "step": 4730 }, { "epoch": 10.58, "learning_rate": 1.7252063931987607e-05, "loss": 0.0945, "step": 4740 }, { "epoch": 10.6, "learning_rate": 1.7239607693664103e-05, "loss": 0.1223, "step": 4750 }, { "epoch": 10.62, "learning_rate": 1.7227127806210578e-05, "loss": 0.1241, "step": 4760 }, { "epoch": 10.65, "learning_rate": 1.7214624310394236e-05, "loss": 0.1155, "step": 4770 }, { "epoch": 10.67, "learning_rate": 1.7202097247059383e-05, "loss": 0.1248, "step": 4780 }, { "epoch": 10.69, "learning_rate": 1.7189546657127315e-05, "loss": 0.1301, "step": 4790 }, { "epoch": 10.71, "learning_rate": 1.717697258159619e-05, "loss": 0.1341, "step": 4800 }, { "epoch": 10.74, "learning_rate": 1.7164375061540877e-05, "loss": 0.1007, "step": 4810 }, { "epoch": 10.76, "learning_rate": 1.715175413811283e-05, "loss": 0.1041, "step": 4820 }, { "epoch": 10.78, "learning_rate": 1.7139109852539954e-05, "loss": 0.1271, "step": 4830 }, { "epoch": 10.8, "learning_rate": 1.712644224612647e-05, "loss": 0.1184, "step": 4840 }, { "epoch": 10.83, "learning_rate": 1.7113751360252777e-05, "loss": 0.1254, "step": 4850 }, { "epoch": 10.85, "learning_rate": 1.7101037236375324e-05, "loss": 0.1574, "step": 4860 }, { "epoch": 10.87, "learning_rate": 1.708829991602647e-05, "loss": 0.1189, "step": 4870 }, { "epoch": 10.89, "learning_rate": 1.707553944081434e-05, "loss": 0.1134, "step": 4880 }, { "epoch": 10.92, "learning_rate": 1.7062755852422705e-05, "loss": 0.1508, "step": 4890 }, { "epoch": 10.94, "learning_rate": 1.7049949192610845e-05, "loss": 0.1292, "step": 4900 }, { "epoch": 10.96, "learning_rate": 1.7037119503213385e-05, "loss": 0.0917, "step": 4910 }, { "epoch": 10.98, "learning_rate": 1.7024266826140194e-05, "loss": 0.1032, "step": 4920 }, { "epoch": 11.0, "learning_rate": 1.701139120337624e-05, "loss": 0.0984, "step": 4930 }, { "epoch": 11.03, "learning_rate": 1.699849267698143e-05, "loss": 0.1162, "step": 4940 }, { "epoch": 11.05, "learning_rate": 1.698557128909049e-05, "loss": 0.088, "step": 4950 }, { "epoch": 11.07, "learning_rate": 1.6972627081912848e-05, "loss": 0.0844, "step": 4960 }, { "epoch": 11.09, "learning_rate": 1.695966009773244e-05, "loss": 0.153, "step": 4970 }, { "epoch": 11.12, "learning_rate": 1.6946670378907635e-05, "loss": 0.1279, "step": 4980 }, { "epoch": 11.14, "learning_rate": 1.6933657967871056e-05, "loss": 0.1098, "step": 4990 }, { "epoch": 11.16, "learning_rate": 1.6920622907129452e-05, "loss": 0.0935, "step": 5000 }, { "epoch": 11.18, "learning_rate": 1.690756523926356e-05, "loss": 0.1088, "step": 5010 }, { "epoch": 11.21, "learning_rate": 1.6894485006927972e-05, "loss": 0.1226, "step": 5020 }, { "epoch": 11.23, "learning_rate": 1.688138225285098e-05, "loss": 0.1202, "step": 5030 }, { "epoch": 11.25, "learning_rate": 1.6868257019834464e-05, "loss": 0.0983, "step": 5040 }, { "epoch": 11.27, "learning_rate": 1.685510935075371e-05, "loss": 0.1028, "step": 5050 }, { "epoch": 11.29, "learning_rate": 1.684193928855731e-05, "loss": 0.0997, "step": 5060 }, { "epoch": 11.32, "learning_rate": 1.682874687626701e-05, "loss": 0.1018, "step": 5070 }, { "epoch": 11.34, "learning_rate": 1.6815532156977553e-05, "loss": 0.1135, "step": 5080 }, { "epoch": 11.36, "learning_rate": 1.6802295173856558e-05, "loss": 0.1053, "step": 5090 }, { "epoch": 11.38, "learning_rate": 1.678903597014437e-05, "loss": 0.1099, "step": 5100 }, { "epoch": 11.41, "learning_rate": 1.6775754589153913e-05, "loss": 0.1319, "step": 5110 }, { "epoch": 11.43, "learning_rate": 1.676245107427058e-05, "loss": 0.11, "step": 5120 }, { "epoch": 11.45, "learning_rate": 1.6749125468952033e-05, "loss": 0.1343, "step": 5130 }, { "epoch": 11.47, "learning_rate": 1.673577781672812e-05, "loss": 0.1054, "step": 5140 }, { "epoch": 11.5, "learning_rate": 1.672240816120071e-05, "loss": 0.1285, "step": 5150 }, { "epoch": 11.52, "learning_rate": 1.670901654604353e-05, "loss": 0.1402, "step": 5160 }, { "epoch": 11.54, "learning_rate": 1.669560301500205e-05, "loss": 0.1584, "step": 5170 }, { "epoch": 11.56, "learning_rate": 1.668216761189334e-05, "loss": 0.1066, "step": 5180 }, { "epoch": 11.58, "learning_rate": 1.6668710380605902e-05, "loss": 0.1021, "step": 5190 }, { "epoch": 11.61, "learning_rate": 1.6655231365099556e-05, "loss": 0.0997, "step": 5200 }, { "epoch": 11.63, "learning_rate": 1.6641730609405276e-05, "loss": 0.1108, "step": 5210 }, { "epoch": 11.65, "learning_rate": 1.6628208157625055e-05, "loss": 0.1394, "step": 5220 }, { "epoch": 11.67, "learning_rate": 1.6614664053931757e-05, "loss": 0.133, "step": 5230 }, { "epoch": 11.7, "learning_rate": 1.6601098342568978e-05, "loss": 0.1151, "step": 5240 }, { "epoch": 11.72, "learning_rate": 1.65875110678509e-05, "loss": 0.0964, "step": 5250 }, { "epoch": 11.74, "learning_rate": 1.6573902274162135e-05, "loss": 0.103, "step": 5260 }, { "epoch": 11.76, "learning_rate": 1.6560272005957604e-05, "loss": 0.1951, "step": 5270 }, { "epoch": 11.79, "learning_rate": 1.6546620307762364e-05, "loss": 0.1511, "step": 5280 }, { "epoch": 11.81, "learning_rate": 1.6532947224171482e-05, "loss": 0.1219, "step": 5290 }, { "epoch": 11.83, "learning_rate": 1.6519252799849887e-05, "loss": 0.1096, "step": 5300 }, { "epoch": 11.85, "learning_rate": 1.650553707953221e-05, "loss": 0.0845, "step": 5310 }, { "epoch": 11.88, "learning_rate": 1.6491800108022657e-05, "loss": 0.1231, "step": 5320 }, { "epoch": 11.9, "learning_rate": 1.6478041930194848e-05, "loss": 0.1064, "step": 5330 }, { "epoch": 11.92, "learning_rate": 1.6464262590991683e-05, "loss": 0.1247, "step": 5340 }, { "epoch": 11.94, "learning_rate": 1.6450462135425187e-05, "loss": 0.1105, "step": 5350 }, { "epoch": 11.96, "learning_rate": 1.6436640608576354e-05, "loss": 0.1168, "step": 5360 }, { "epoch": 11.99, "learning_rate": 1.642279805559502e-05, "loss": 0.1313, "step": 5370 }, { "epoch": 12.01, "learning_rate": 1.6408934521699706e-05, "loss": 0.1124, "step": 5380 }, { "epoch": 12.03, "learning_rate": 1.639505005217747e-05, "loss": 0.0774, "step": 5390 }, { "epoch": 12.05, "learning_rate": 1.6381144692383754e-05, "loss": 0.1125, "step": 5400 }, { "epoch": 12.08, "learning_rate": 1.636721848774224e-05, "loss": 0.0943, "step": 5410 }, { "epoch": 12.1, "learning_rate": 1.635327148374471e-05, "loss": 0.0996, "step": 5420 }, { "epoch": 12.12, "learning_rate": 1.633930372595088e-05, "loss": 0.0839, "step": 5430 }, { "epoch": 12.14, "learning_rate": 1.6325315259988275e-05, "loss": 0.1085, "step": 5440 }, { "epoch": 12.17, "learning_rate": 1.631130613155205e-05, "loss": 0.0751, "step": 5450 }, { "epoch": 12.19, "learning_rate": 1.6297276386404872e-05, "loss": 0.0946, "step": 5460 }, { "epoch": 12.21, "learning_rate": 1.628322607037674e-05, "loss": 0.0975, "step": 5470 }, { "epoch": 12.23, "learning_rate": 1.626915522936486e-05, "loss": 0.1105, "step": 5480 }, { "epoch": 12.25, "learning_rate": 1.6255063909333486e-05, "loss": 0.0876, "step": 5490 }, { "epoch": 12.28, "learning_rate": 1.6240952156313762e-05, "loss": 0.1014, "step": 5500 }, { "epoch": 12.3, "learning_rate": 1.622682001640359e-05, "loss": 0.087, "step": 5510 }, { "epoch": 12.32, "learning_rate": 1.6212667535767456e-05, "loss": 0.1555, "step": 5520 }, { "epoch": 12.34, "learning_rate": 1.6198494760636303e-05, "loss": 0.0878, "step": 5530 }, { "epoch": 12.37, "learning_rate": 1.618430173730736e-05, "loss": 0.1181, "step": 5540 }, { "epoch": 12.39, "learning_rate": 1.617008851214401e-05, "loss": 0.1031, "step": 5550 }, { "epoch": 12.41, "learning_rate": 1.6155855131575614e-05, "loss": 0.0998, "step": 5560 }, { "epoch": 12.43, "learning_rate": 1.6141601642097382e-05, "loss": 0.0926, "step": 5570 }, { "epoch": 12.46, "learning_rate": 1.6127328090270213e-05, "loss": 0.0829, "step": 5580 }, { "epoch": 12.48, "learning_rate": 1.611303452272053e-05, "loss": 0.1246, "step": 5590 }, { "epoch": 12.5, "learning_rate": 1.609872098614017e-05, "loss": 0.08, "step": 5600 }, { "epoch": 12.52, "learning_rate": 1.608438752728616e-05, "loss": 0.1063, "step": 5610 }, { "epoch": 12.54, "learning_rate": 1.6070034192980638e-05, "loss": 0.175, "step": 5620 }, { "epoch": 12.57, "learning_rate": 1.6055661030110655e-05, "loss": 0.1021, "step": 5630 }, { "epoch": 12.59, "learning_rate": 1.6041268085628042e-05, "loss": 0.1085, "step": 5640 }, { "epoch": 12.61, "learning_rate": 1.602685540654924e-05, "loss": 0.1007, "step": 5650 }, { "epoch": 12.63, "learning_rate": 1.6012423039955153e-05, "loss": 0.107, "step": 5660 }, { "epoch": 12.66, "learning_rate": 1.5997971032991007e-05, "loss": 0.1331, "step": 5670 }, { "epoch": 12.68, "learning_rate": 1.5983499432866187e-05, "loss": 0.1165, "step": 5680 }, { "epoch": 12.7, "learning_rate": 1.596900828685407e-05, "loss": 0.1051, "step": 5690 }, { "epoch": 12.72, "learning_rate": 1.5954497642291897e-05, "loss": 0.1058, "step": 5700 }, { "epoch": 12.75, "learning_rate": 1.593996754658059e-05, "loss": 0.1249, "step": 5710 }, { "epoch": 12.77, "learning_rate": 1.5925418047184615e-05, "loss": 0.1124, "step": 5720 }, { "epoch": 12.79, "learning_rate": 1.591084919163183e-05, "loss": 0.0974, "step": 5730 }, { "epoch": 12.81, "learning_rate": 1.589626102751331e-05, "loss": 0.1393, "step": 5740 }, { "epoch": 12.83, "learning_rate": 1.588165360248321e-05, "loss": 0.0852, "step": 5750 }, { "epoch": 12.86, "learning_rate": 1.5867026964258614e-05, "loss": 0.1135, "step": 5760 }, { "epoch": 12.88, "learning_rate": 1.5852381160619343e-05, "loss": 0.1014, "step": 5770 }, { "epoch": 12.9, "learning_rate": 1.5837716239407855e-05, "loss": 0.0875, "step": 5780 }, { "epoch": 12.92, "learning_rate": 1.582303224852903e-05, "loss": 0.1782, "step": 5790 }, { "epoch": 12.95, "learning_rate": 1.580832923595006e-05, "loss": 0.1207, "step": 5800 }, { "epoch": 12.97, "learning_rate": 1.5793607249700268e-05, "loss": 0.1528, "step": 5810 }, { "epoch": 12.99, "learning_rate": 1.5778866337870952e-05, "loss": 0.1701, "step": 5820 }, { "epoch": 13.01, "learning_rate": 1.5764106548615244e-05, "loss": 0.0959, "step": 5830 }, { "epoch": 13.04, "learning_rate": 1.5749327930147932e-05, "loss": 0.0841, "step": 5840 }, { "epoch": 13.06, "learning_rate": 1.573453053074532e-05, "loss": 0.1153, "step": 5850 }, { "epoch": 13.08, "learning_rate": 1.571971439874505e-05, "loss": 0.101, "step": 5860 }, { "epoch": 13.1, "learning_rate": 1.570487958254597e-05, "loss": 0.0999, "step": 5870 }, { "epoch": 13.12, "learning_rate": 1.569002613060796e-05, "loss": 0.1071, "step": 5880 }, { "epoch": 13.15, "learning_rate": 1.5675154091451765e-05, "loss": 0.101, "step": 5890 }, { "epoch": 13.17, "learning_rate": 1.566026351365886e-05, "loss": 0.107, "step": 5900 }, { "epoch": 13.19, "learning_rate": 1.5645354445871274e-05, "loss": 0.0975, "step": 5910 }, { "epoch": 13.21, "learning_rate": 1.5630426936791433e-05, "loss": 0.0876, "step": 5920 }, { "epoch": 13.24, "learning_rate": 1.5615481035182013e-05, "loss": 0.1238, "step": 5930 }, { "epoch": 13.26, "learning_rate": 1.5600516789865767e-05, "loss": 0.1028, "step": 5940 }, { "epoch": 13.28, "learning_rate": 1.5585534249725362e-05, "loss": 0.0883, "step": 5950 }, { "epoch": 13.3, "learning_rate": 1.5570533463703233e-05, "loss": 0.0945, "step": 5960 }, { "epoch": 13.33, "learning_rate": 1.555551448080143e-05, "loss": 0.1148, "step": 5970 }, { "epoch": 13.35, "learning_rate": 1.5540477350081423e-05, "loss": 0.1207, "step": 5980 }, { "epoch": 13.37, "learning_rate": 1.5525422120663986e-05, "loss": 0.0874, "step": 5990 }, { "epoch": 13.39, "learning_rate": 1.5510348841728997e-05, "loss": 0.0927, "step": 6000 }, { "epoch": 13.42, "learning_rate": 1.5495257562515308e-05, "loss": 0.0971, "step": 6010 }, { "epoch": 13.44, "learning_rate": 1.5480148332320562e-05, "loss": 0.0997, "step": 6020 }, { "epoch": 13.46, "learning_rate": 1.5465021200501046e-05, "loss": 0.0915, "step": 6030 }, { "epoch": 13.48, "learning_rate": 1.5449876216471525e-05, "loss": 0.0958, "step": 6040 }, { "epoch": 13.5, "learning_rate": 1.5434713429705078e-05, "loss": 0.1452, "step": 6050 }, { "epoch": 13.53, "learning_rate": 1.5419532889732943e-05, "loss": 0.0934, "step": 6060 }, { "epoch": 13.55, "learning_rate": 1.540433464614435e-05, "loss": 0.0974, "step": 6070 }, { "epoch": 13.57, "learning_rate": 1.5389118748586357e-05, "loss": 0.1125, "step": 6080 }, { "epoch": 13.59, "learning_rate": 1.537388524676369e-05, "loss": 0.1034, "step": 6090 }, { "epoch": 13.62, "learning_rate": 1.5358634190438592e-05, "loss": 0.1037, "step": 6100 }, { "epoch": 13.64, "learning_rate": 1.5343365629430638e-05, "loss": 0.0932, "step": 6110 }, { "epoch": 13.66, "learning_rate": 1.5328079613616592e-05, "loss": 0.0992, "step": 6120 }, { "epoch": 13.68, "learning_rate": 1.531277619293023e-05, "loss": 0.105, "step": 6130 }, { "epoch": 13.71, "learning_rate": 1.5297455417362194e-05, "loss": 0.0902, "step": 6140 }, { "epoch": 13.73, "learning_rate": 1.52821173369598e-05, "loss": 0.1047, "step": 6150 }, { "epoch": 13.75, "learning_rate": 1.526676200182691e-05, "loss": 0.0969, "step": 6160 }, { "epoch": 13.77, "learning_rate": 1.5251389462123748e-05, "loss": 0.0982, "step": 6170 }, { "epoch": 13.79, "learning_rate": 1.5235999768066729e-05, "loss": 0.0928, "step": 6180 }, { "epoch": 13.82, "learning_rate": 1.5220592969928313e-05, "loss": 0.1082, "step": 6190 }, { "epoch": 13.84, "learning_rate": 1.5205169118036831e-05, "loss": 0.082, "step": 6200 }, { "epoch": 13.86, "learning_rate": 1.5189728262776325e-05, "loss": 0.0939, "step": 6210 }, { "epoch": 13.88, "learning_rate": 1.5174270454586375e-05, "loss": 0.1805, "step": 6220 }, { "epoch": 13.91, "learning_rate": 1.5158795743961942e-05, "loss": 0.1409, "step": 6230 }, { "epoch": 13.93, "learning_rate": 1.5143304181453204e-05, "loss": 0.1409, "step": 6240 }, { "epoch": 13.95, "learning_rate": 1.5127795817665389e-05, "loss": 0.1206, "step": 6250 }, { "epoch": 13.97, "learning_rate": 1.5112270703258602e-05, "loss": 0.1379, "step": 6260 }, { "epoch": 14.0, "learning_rate": 1.5096728888947669e-05, "loss": 0.0794, "step": 6270 }, { "epoch": 14.02, "learning_rate": 1.508117042550197e-05, "loss": 0.1078, "step": 6280 }, { "epoch": 14.04, "learning_rate": 1.5065595363745272e-05, "loss": 0.0941, "step": 6290 }, { "epoch": 14.06, "learning_rate": 1.505000375455556e-05, "loss": 0.0893, "step": 6300 }, { "epoch": 14.08, "learning_rate": 1.503439564886487e-05, "loss": 0.1003, "step": 6310 }, { "epoch": 14.11, "learning_rate": 1.501877109765914e-05, "loss": 0.0851, "step": 6320 }, { "epoch": 14.13, "learning_rate": 1.5003130151978012e-05, "loss": 0.0819, "step": 6330 }, { "epoch": 14.15, "learning_rate": 1.4987472862914697e-05, "loss": 0.0819, "step": 6340 }, { "epoch": 14.17, "learning_rate": 1.4971799281615782e-05, "loss": 0.0852, "step": 6350 }, { "epoch": 14.2, "learning_rate": 1.4956109459281083e-05, "loss": 0.0748, "step": 6360 }, { "epoch": 14.22, "learning_rate": 1.4940403447163467e-05, "loss": 0.1047, "step": 6370 }, { "epoch": 14.24, "learning_rate": 1.4924681296568689e-05, "loss": 0.0865, "step": 6380 }, { "epoch": 14.26, "learning_rate": 1.4908943058855213e-05, "loss": 0.1038, "step": 6390 }, { "epoch": 14.29, "learning_rate": 1.4893188785434067e-05, "loss": 0.1166, "step": 6400 }, { "epoch": 14.31, "learning_rate": 1.4877418527768654e-05, "loss": 0.1093, "step": 6410 }, { "epoch": 14.33, "learning_rate": 1.4861632337374596e-05, "loss": 0.0827, "step": 6420 }, { "epoch": 14.35, "learning_rate": 1.4845830265819552e-05, "loss": 0.1074, "step": 6430 }, { "epoch": 14.38, "learning_rate": 1.483001236472307e-05, "loss": 0.0929, "step": 6440 }, { "epoch": 14.4, "learning_rate": 1.4814178685756405e-05, "loss": 0.083, "step": 6450 }, { "epoch": 14.42, "learning_rate": 1.4798329280642345e-05, "loss": 0.0842, "step": 6460 }, { "epoch": 14.44, "learning_rate": 1.4782464201155057e-05, "loss": 0.0988, "step": 6470 }, { "epoch": 14.46, "learning_rate": 1.476658349911991e-05, "loss": 0.0776, "step": 6480 }, { "epoch": 14.49, "learning_rate": 1.4750687226413305e-05, "loss": 0.0863, "step": 6490 }, { "epoch": 14.51, "learning_rate": 1.4734775434962504e-05, "loss": 0.0995, "step": 6500 }, { "epoch": 14.53, "learning_rate": 1.471884817674546e-05, "loss": 0.0917, "step": 6510 }, { "epoch": 14.55, "learning_rate": 1.4702905503790668e-05, "loss": 0.2156, "step": 6520 }, { "epoch": 14.58, "learning_rate": 1.4686947468176955e-05, "loss": 0.084, "step": 6530 }, { "epoch": 14.6, "learning_rate": 1.467097412203334e-05, "loss": 0.1124, "step": 6540 }, { "epoch": 14.62, "learning_rate": 1.4654985517538864e-05, "loss": 0.1854, "step": 6550 }, { "epoch": 14.64, "learning_rate": 1.4638981706922401e-05, "loss": 0.0905, "step": 6560 }, { "epoch": 14.67, "learning_rate": 1.4622962742462503e-05, "loss": 0.0947, "step": 6570 }, { "epoch": 14.69, "learning_rate": 1.4606928676487223e-05, "loss": 0.1043, "step": 6580 }, { "epoch": 14.71, "learning_rate": 1.459087956137394e-05, "loss": 0.1081, "step": 6590 }, { "epoch": 14.73, "learning_rate": 1.4574815449549209e-05, "loss": 0.0771, "step": 6600 }, { "epoch": 14.75, "learning_rate": 1.4558736393488553e-05, "loss": 0.1048, "step": 6610 }, { "epoch": 14.78, "learning_rate": 1.4542642445716326e-05, "loss": 0.1048, "step": 6620 }, { "epoch": 14.8, "learning_rate": 1.4526533658805517e-05, "loss": 0.0987, "step": 6630 }, { "epoch": 14.82, "learning_rate": 1.4510410085377606e-05, "loss": 0.1066, "step": 6640 }, { "epoch": 14.84, "learning_rate": 1.4494271778102358e-05, "loss": 0.0792, "step": 6650 }, { "epoch": 14.87, "learning_rate": 1.4478118789697675e-05, "loss": 0.0925, "step": 6660 }, { "epoch": 14.89, "learning_rate": 1.4461951172929419e-05, "loss": 0.1078, "step": 6670 }, { "epoch": 14.91, "learning_rate": 1.4445768980611233e-05, "loss": 0.0942, "step": 6680 }, { "epoch": 14.93, "learning_rate": 1.4429572265604375e-05, "loss": 0.0829, "step": 6690 }, { "epoch": 14.96, "learning_rate": 1.4413361080817545e-05, "loss": 0.1201, "step": 6700 }, { "epoch": 14.98, "learning_rate": 1.4397135479206705e-05, "loss": 0.1201, "step": 6710 }, { "epoch": 15.0, "learning_rate": 1.4380895513774922e-05, "loss": 0.0985, "step": 6720 }, { "epoch": 15.02, "learning_rate": 1.436464123757217e-05, "loss": 0.068, "step": 6730 }, { "epoch": 15.04, "learning_rate": 1.4348372703695184e-05, "loss": 0.0923, "step": 6740 }, { "epoch": 15.07, "learning_rate": 1.4332089965287266e-05, "loss": 0.0808, "step": 6750 }, { "epoch": 15.09, "learning_rate": 1.431579307553812e-05, "loss": 0.079, "step": 6760 }, { "epoch": 15.11, "learning_rate": 1.429948208768368e-05, "loss": 0.074, "step": 6770 }, { "epoch": 15.13, "learning_rate": 1.4283157055005928e-05, "loss": 0.0812, "step": 6780 }, { "epoch": 15.16, "learning_rate": 1.4266818030832732e-05, "loss": 0.0744, "step": 6790 }, { "epoch": 15.18, "learning_rate": 1.4250465068537664e-05, "loss": 0.0786, "step": 6800 }, { "epoch": 15.2, "learning_rate": 1.4234098221539818e-05, "loss": 0.08, "step": 6810 }, { "epoch": 15.22, "learning_rate": 1.4217717543303657e-05, "loss": 0.1014, "step": 6820 }, { "epoch": 15.25, "learning_rate": 1.4201323087338816e-05, "loss": 0.0751, "step": 6830 }, { "epoch": 15.27, "learning_rate": 1.4184914907199942e-05, "loss": 0.0937, "step": 6840 }, { "epoch": 15.29, "learning_rate": 1.4168493056486512e-05, "loss": 0.0852, "step": 6850 }, { "epoch": 15.31, "learning_rate": 1.4152057588842657e-05, "loss": 0.0896, "step": 6860 }, { "epoch": 15.33, "learning_rate": 1.4135608557956992e-05, "loss": 0.1104, "step": 6870 }, { "epoch": 15.36, "learning_rate": 1.4119146017562441e-05, "loss": 0.0914, "step": 6880 }, { "epoch": 15.38, "learning_rate": 1.4102670021436059e-05, "loss": 0.075, "step": 6890 }, { "epoch": 15.4, "learning_rate": 1.4086180623398842e-05, "loss": 0.0746, "step": 6900 }, { "epoch": 15.42, "learning_rate": 1.4069677877315587e-05, "loss": 0.1284, "step": 6910 }, { "epoch": 15.45, "learning_rate": 1.4053161837094675e-05, "loss": 0.0999, "step": 6920 }, { "epoch": 15.47, "learning_rate": 1.4036632556687927e-05, "loss": 0.094, "step": 6930 }, { "epoch": 15.49, "learning_rate": 1.4020090090090408e-05, "loss": 0.1239, "step": 6940 }, { "epoch": 15.51, "learning_rate": 1.4003534491340259e-05, "loss": 0.1013, "step": 6950 }, { "epoch": 15.54, "learning_rate": 1.3986965814518521e-05, "loss": 0.1134, "step": 6960 }, { "epoch": 15.56, "learning_rate": 1.3970384113748951e-05, "loss": 0.0859, "step": 6970 }, { "epoch": 15.58, "learning_rate": 1.3953789443197857e-05, "loss": 0.1392, "step": 6980 }, { "epoch": 15.6, "learning_rate": 1.3937181857073912e-05, "loss": 0.0904, "step": 6990 }, { "epoch": 15.62, "learning_rate": 1.3920561409627974e-05, "loss": 0.0792, "step": 7000 }, { "epoch": 15.65, "learning_rate": 1.3903928155152926e-05, "loss": 0.0778, "step": 7010 }, { "epoch": 15.67, "learning_rate": 1.3887282147983472e-05, "loss": 0.0995, "step": 7020 }, { "epoch": 15.69, "learning_rate": 1.3870623442495987e-05, "loss": 0.1112, "step": 7030 }, { "epoch": 15.71, "learning_rate": 1.3853952093108323e-05, "loss": 0.088, "step": 7040 }, { "epoch": 15.74, "learning_rate": 1.3837268154279628e-05, "loss": 0.0833, "step": 7050 }, { "epoch": 15.76, "learning_rate": 1.3820571680510187e-05, "loss": 0.1009, "step": 7060 }, { "epoch": 15.78, "learning_rate": 1.3803862726341224e-05, "loss": 0.1223, "step": 7070 }, { "epoch": 15.8, "learning_rate": 1.3787141346354733e-05, "loss": 0.1115, "step": 7080 }, { "epoch": 15.83, "learning_rate": 1.3770407595173301e-05, "loss": 0.0905, "step": 7090 }, { "epoch": 15.85, "learning_rate": 1.375366152745992e-05, "loss": 0.0843, "step": 7100 }, { "epoch": 15.87, "learning_rate": 1.373690319791783e-05, "loss": 0.0878, "step": 7110 }, { "epoch": 15.89, "learning_rate": 1.3720132661290311e-05, "loss": 0.1105, "step": 7120 }, { "epoch": 15.92, "learning_rate": 1.3703349972360527e-05, "loss": 0.0942, "step": 7130 }, { "epoch": 15.94, "learning_rate": 1.3686555185951334e-05, "loss": 0.1013, "step": 7140 }, { "epoch": 15.96, "learning_rate": 1.3669748356925112e-05, "loss": 0.1016, "step": 7150 }, { "epoch": 15.98, "learning_rate": 1.3652929540183578e-05, "loss": 0.0893, "step": 7160 }, { "epoch": 16.0, "learning_rate": 1.3636098790667605e-05, "loss": 0.0836, "step": 7170 }, { "epoch": 16.03, "learning_rate": 1.3619256163357046e-05, "loss": 0.0896, "step": 7180 }, { "epoch": 16.05, "learning_rate": 1.3602401713270566e-05, "loss": 0.0786, "step": 7190 }, { "epoch": 16.07, "learning_rate": 1.3585535495465432e-05, "loss": 0.0731, "step": 7200 }, { "epoch": 16.09, "learning_rate": 1.3568657565037365e-05, "loss": 0.1016, "step": 7210 }, { "epoch": 16.12, "learning_rate": 1.3551767977120341e-05, "loss": 0.0826, "step": 7220 }, { "epoch": 16.14, "learning_rate": 1.353486678688642e-05, "loss": 0.0733, "step": 7230 }, { "epoch": 16.16, "learning_rate": 1.351795404954556e-05, "loss": 0.0677, "step": 7240 }, { "epoch": 16.18, "learning_rate": 1.3501029820345446e-05, "loss": 0.0948, "step": 7250 }, { "epoch": 16.21, "learning_rate": 1.3484094154571286e-05, "loss": 0.1281, "step": 7260 }, { "epoch": 16.23, "learning_rate": 1.3467147107545668e-05, "loss": 0.0812, "step": 7270 }, { "epoch": 16.25, "learning_rate": 1.3450188734628344e-05, "loss": 0.0653, "step": 7280 }, { "epoch": 16.27, "learning_rate": 1.3433219091216069e-05, "loss": 0.0809, "step": 7290 }, { "epoch": 16.29, "learning_rate": 1.3416238232742414e-05, "loss": 0.0919, "step": 7300 }, { "epoch": 16.32, "learning_rate": 1.3399246214677583e-05, "loss": 0.0706, "step": 7310 }, { "epoch": 16.34, "learning_rate": 1.338224309252824e-05, "loss": 0.0947, "step": 7320 }, { "epoch": 16.36, "learning_rate": 1.3365228921837314e-05, "loss": 0.0592, "step": 7330 }, { "epoch": 16.38, "learning_rate": 1.3348203758183831e-05, "loss": 0.0839, "step": 7340 }, { "epoch": 16.41, "learning_rate": 1.3331167657182726e-05, "loss": 0.0852, "step": 7350 }, { "epoch": 16.43, "learning_rate": 1.3314120674484663e-05, "loss": 0.0893, "step": 7360 }, { "epoch": 16.45, "learning_rate": 1.3297062865775851e-05, "loss": 0.0943, "step": 7370 }, { "epoch": 16.47, "learning_rate": 1.327999428677786e-05, "loss": 0.0839, "step": 7380 }, { "epoch": 16.5, "learning_rate": 1.3262914993247454e-05, "loss": 0.0985, "step": 7390 }, { "epoch": 16.52, "learning_rate": 1.324582504097638e-05, "loss": 0.0892, "step": 7400 }, { "epoch": 16.54, "learning_rate": 1.3228724485791225e-05, "loss": 0.0724, "step": 7410 }, { "epoch": 16.56, "learning_rate": 1.321161338355319e-05, "loss": 0.0732, "step": 7420 }, { "epoch": 16.58, "learning_rate": 1.3194491790157947e-05, "loss": 0.072, "step": 7430 }, { "epoch": 16.61, "learning_rate": 1.3177359761535427e-05, "loss": 0.1024, "step": 7440 }, { "epoch": 16.63, "learning_rate": 1.3160217353649652e-05, "loss": 0.0986, "step": 7450 }, { "epoch": 16.65, "learning_rate": 1.3143064622498551e-05, "loss": 0.0729, "step": 7460 }, { "epoch": 16.67, "learning_rate": 1.312590162411378e-05, "loss": 0.0892, "step": 7470 }, { "epoch": 16.7, "learning_rate": 1.310872841456052e-05, "loss": 0.0931, "step": 7480 }, { "epoch": 16.72, "learning_rate": 1.3091545049937322e-05, "loss": 0.128, "step": 7490 }, { "epoch": 16.74, "learning_rate": 1.3074351586375906e-05, "loss": 0.1005, "step": 7500 }, { "epoch": 16.76, "learning_rate": 1.305714808004098e-05, "loss": 0.0697, "step": 7510 }, { "epoch": 16.79, "learning_rate": 1.3039934587130056e-05, "loss": 0.1304, "step": 7520 }, { "epoch": 16.81, "learning_rate": 1.3022711163873272e-05, "loss": 0.078, "step": 7530 }, { "epoch": 16.83, "learning_rate": 1.3005477866533202e-05, "loss": 0.0796, "step": 7540 }, { "epoch": 16.85, "learning_rate": 1.2988234751404683e-05, "loss": 0.08, "step": 7550 }, { "epoch": 16.88, "learning_rate": 1.2970981874814613e-05, "loss": 0.0796, "step": 7560 }, { "epoch": 16.9, "learning_rate": 1.2953719293121775e-05, "loss": 0.1008, "step": 7570 }, { "epoch": 16.92, "learning_rate": 1.2936447062716668e-05, "loss": 0.1491, "step": 7580 }, { "epoch": 16.94, "learning_rate": 1.2919165240021303e-05, "loss": 0.0754, "step": 7590 }, { "epoch": 16.96, "learning_rate": 1.2901873881489021e-05, "loss": 0.0843, "step": 7600 }, { "epoch": 16.99, "learning_rate": 1.288457304360432e-05, "loss": 0.0734, "step": 7610 }, { "epoch": 17.01, "learning_rate": 1.2867262782882662e-05, "loss": 0.0672, "step": 7620 }, { "epoch": 17.03, "learning_rate": 1.2849943155870284e-05, "loss": 0.0677, "step": 7630 }, { "epoch": 17.05, "learning_rate": 1.2832614219144027e-05, "loss": 0.1288, "step": 7640 }, { "epoch": 17.08, "learning_rate": 1.2815276029311138e-05, "loss": 0.0643, "step": 7650 }, { "epoch": 17.1, "learning_rate": 1.2797928643009097e-05, "loss": 0.0823, "step": 7660 }, { "epoch": 17.12, "learning_rate": 1.2780572116905418e-05, "loss": 0.071, "step": 7670 }, { "epoch": 17.14, "learning_rate": 1.276320650769748e-05, "loss": 0.0754, "step": 7680 }, { "epoch": 17.17, "learning_rate": 1.2745831872112318e-05, "loss": 0.0638, "step": 7690 }, { "epoch": 17.19, "learning_rate": 1.2728448266906468e-05, "loss": 0.0794, "step": 7700 }, { "epoch": 17.21, "learning_rate": 1.2711055748865765e-05, "loss": 0.0865, "step": 7710 }, { "epoch": 17.23, "learning_rate": 1.2693654374805148e-05, "loss": 0.0854, "step": 7720 }, { "epoch": 17.25, "learning_rate": 1.2676244201568498e-05, "loss": 0.0831, "step": 7730 }, { "epoch": 17.28, "learning_rate": 1.2658825286028428e-05, "loss": 0.0749, "step": 7740 }, { "epoch": 17.3, "learning_rate": 1.2641397685086124e-05, "loss": 0.0809, "step": 7750 }, { "epoch": 17.32, "learning_rate": 1.2623961455671125e-05, "loss": 0.071, "step": 7760 }, { "epoch": 17.34, "learning_rate": 1.2606516654741172e-05, "loss": 0.0777, "step": 7770 }, { "epoch": 17.37, "learning_rate": 1.2589063339281995e-05, "loss": 0.1106, "step": 7780 }, { "epoch": 17.39, "learning_rate": 1.257160156630715e-05, "loss": 0.0739, "step": 7790 }, { "epoch": 17.41, "learning_rate": 1.2554131392857812e-05, "loss": 0.0669, "step": 7800 }, { "epoch": 17.43, "learning_rate": 1.253665287600259e-05, "loss": 0.0927, "step": 7810 }, { "epoch": 17.46, "learning_rate": 1.2519166072837368e-05, "loss": 0.0701, "step": 7820 }, { "epoch": 17.48, "learning_rate": 1.250167104048508e-05, "loss": 0.076, "step": 7830 }, { "epoch": 17.5, "learning_rate": 1.248416783609555e-05, "loss": 0.087, "step": 7840 }, { "epoch": 17.52, "learning_rate": 1.2466656516845293e-05, "loss": 0.0743, "step": 7850 }, { "epoch": 17.54, "learning_rate": 1.244913713993734e-05, "loss": 0.0949, "step": 7860 }, { "epoch": 17.57, "learning_rate": 1.2431609762601036e-05, "loss": 0.0738, "step": 7870 }, { "epoch": 17.59, "learning_rate": 1.241407444209186e-05, "loss": 0.0623, "step": 7880 }, { "epoch": 17.61, "learning_rate": 1.2396531235691245e-05, "loss": 0.0812, "step": 7890 }, { "epoch": 17.63, "learning_rate": 1.2378980200706376e-05, "loss": 0.0899, "step": 7900 }, { "epoch": 17.66, "learning_rate": 1.236142139447002e-05, "loss": 0.089, "step": 7910 }, { "epoch": 17.68, "learning_rate": 1.2343854874340324e-05, "loss": 0.0817, "step": 7920 }, { "epoch": 17.7, "learning_rate": 1.2326280697700632e-05, "loss": 0.1199, "step": 7930 }, { "epoch": 17.72, "learning_rate": 1.2308698921959306e-05, "loss": 0.0904, "step": 7940 }, { "epoch": 17.75, "learning_rate": 1.2291109604549525e-05, "loss": 0.0667, "step": 7950 }, { "epoch": 17.77, "learning_rate": 1.2273512802929107e-05, "loss": 0.1035, "step": 7960 }, { "epoch": 17.79, "learning_rate": 1.2255908574580311e-05, "loss": 0.0842, "step": 7970 }, { "epoch": 17.81, "learning_rate": 1.2238296977009672e-05, "loss": 0.0892, "step": 7980 }, { "epoch": 17.83, "learning_rate": 1.2220678067747785e-05, "loss": 0.0846, "step": 7990 }, { "epoch": 17.86, "learning_rate": 1.2203051904349128e-05, "loss": 0.0829, "step": 8000 }, { "epoch": 17.88, "learning_rate": 1.2185418544391885e-05, "loss": 0.0822, "step": 8010 }, { "epoch": 17.9, "learning_rate": 1.2167778045477743e-05, "loss": 0.088, "step": 8020 }, { "epoch": 17.92, "learning_rate": 1.215013046523171e-05, "loss": 0.0973, "step": 8030 }, { "epoch": 17.95, "learning_rate": 1.2132475861301928e-05, "loss": 0.0811, "step": 8040 }, { "epoch": 17.97, "learning_rate": 1.2114814291359476e-05, "loss": 0.0749, "step": 8050 }, { "epoch": 17.99, "learning_rate": 1.20971458130982e-05, "loss": 0.0894, "step": 8060 }, { "epoch": 18.01, "learning_rate": 1.20794704842345e-05, "loss": 0.0807, "step": 8070 }, { "epoch": 18.04, "learning_rate": 1.2061788362507168e-05, "loss": 0.0713, "step": 8080 }, { "epoch": 18.06, "learning_rate": 1.204409950567717e-05, "loss": 0.0682, "step": 8090 }, { "epoch": 18.08, "learning_rate": 1.2026403971527487e-05, "loss": 0.0653, "step": 8100 }, { "epoch": 18.1, "learning_rate": 1.2008701817862906e-05, "loss": 0.0656, "step": 8110 }, { "epoch": 18.12, "learning_rate": 1.1990993102509838e-05, "loss": 0.0801, "step": 8120 }, { "epoch": 18.15, "learning_rate": 1.1973277883316128e-05, "loss": 0.0697, "step": 8130 }, { "epoch": 18.17, "learning_rate": 1.1955556218150872e-05, "loss": 0.0742, "step": 8140 }, { "epoch": 18.19, "learning_rate": 1.1937828164904216e-05, "loss": 0.076, "step": 8150 }, { "epoch": 18.21, "learning_rate": 1.1920093781487175e-05, "loss": 0.1006, "step": 8160 }, { "epoch": 18.24, "learning_rate": 1.1902353125831441e-05, "loss": 0.1085, "step": 8170 }, { "epoch": 18.26, "learning_rate": 1.1884606255889203e-05, "loss": 0.096, "step": 8180 }, { "epoch": 18.28, "learning_rate": 1.1866853229632942e-05, "loss": 0.0735, "step": 8190 }, { "epoch": 18.3, "learning_rate": 1.1849094105055248e-05, "loss": 0.0591, "step": 8200 }, { "epoch": 18.33, "learning_rate": 1.1831328940168638e-05, "loss": 0.1013, "step": 8210 }, { "epoch": 18.35, "learning_rate": 1.181355779300536e-05, "loss": 0.0744, "step": 8220 }, { "epoch": 18.37, "learning_rate": 1.1795780721617199e-05, "loss": 0.0704, "step": 8230 }, { "epoch": 18.39, "learning_rate": 1.1777997784075294e-05, "loss": 0.0693, "step": 8240 }, { "epoch": 18.42, "learning_rate": 1.176020903846995e-05, "loss": 0.067, "step": 8250 }, { "epoch": 18.44, "learning_rate": 1.1742414542910444e-05, "loss": 0.0709, "step": 8260 }, { "epoch": 18.46, "learning_rate": 1.1724614355524832e-05, "loss": 0.0848, "step": 8270 }, { "epoch": 18.48, "learning_rate": 1.1706808534459768e-05, "loss": 0.0898, "step": 8280 }, { "epoch": 18.5, "learning_rate": 1.16889971378803e-05, "loss": 0.0763, "step": 8290 }, { "epoch": 18.53, "learning_rate": 1.1671180223969705e-05, "loss": 0.0886, "step": 8300 }, { "epoch": 18.55, "learning_rate": 1.1653357850929268e-05, "loss": 0.0829, "step": 8310 }, { "epoch": 18.57, "learning_rate": 1.1635530076978115e-05, "loss": 0.0736, "step": 8320 }, { "epoch": 18.59, "learning_rate": 1.161769696035301e-05, "loss": 0.0867, "step": 8330 }, { "epoch": 18.62, "learning_rate": 1.1599858559308175e-05, "loss": 0.088, "step": 8340 }, { "epoch": 18.64, "learning_rate": 1.158201493211509e-05, "loss": 0.0732, "step": 8350 }, { "epoch": 18.66, "learning_rate": 1.156416613706231e-05, "loss": 0.0675, "step": 8360 }, { "epoch": 18.68, "learning_rate": 1.1546312232455266e-05, "loss": 0.0729, "step": 8370 }, { "epoch": 18.71, "learning_rate": 1.152845327661609e-05, "loss": 0.0635, "step": 8380 }, { "epoch": 18.73, "learning_rate": 1.1510589327883406e-05, "loss": 0.0776, "step": 8390 }, { "epoch": 18.75, "learning_rate": 1.1492720444612148e-05, "loss": 0.0625, "step": 8400 }, { "epoch": 18.77, "learning_rate": 1.1474846685173374e-05, "loss": 0.0653, "step": 8410 }, { "epoch": 18.79, "learning_rate": 1.1456968107954066e-05, "loss": 0.0592, "step": 8420 }, { "epoch": 18.82, "learning_rate": 1.143908477135695e-05, "loss": 0.0811, "step": 8430 }, { "epoch": 18.84, "learning_rate": 1.1421196733800291e-05, "loss": 0.0989, "step": 8440 }, { "epoch": 18.86, "learning_rate": 1.1403304053717719e-05, "loss": 0.0673, "step": 8450 }, { "epoch": 18.88, "learning_rate": 1.138540678955802e-05, "loss": 0.0691, "step": 8460 }, { "epoch": 18.91, "learning_rate": 1.1367504999784963e-05, "loss": 0.0656, "step": 8470 }, { "epoch": 18.93, "learning_rate": 1.1349598742877097e-05, "loss": 0.0964, "step": 8480 }, { "epoch": 18.95, "learning_rate": 1.1331688077327563e-05, "loss": 0.1015, "step": 8490 }, { "epoch": 18.97, "learning_rate": 1.1313773061643905e-05, "loss": 0.1236, "step": 8500 }, { "epoch": 19.0, "learning_rate": 1.1295853754347876e-05, "loss": 0.0762, "step": 8510 }, { "epoch": 19.02, "learning_rate": 1.1277930213975249e-05, "loss": 0.0847, "step": 8520 }, { "epoch": 19.04, "learning_rate": 1.1260002499075617e-05, "loss": 0.0643, "step": 8530 }, { "epoch": 19.06, "learning_rate": 1.1242070668212227e-05, "loss": 0.0858, "step": 8540 }, { "epoch": 19.08, "learning_rate": 1.1224134779961758e-05, "loss": 0.0586, "step": 8550 }, { "epoch": 19.11, "learning_rate": 1.1206194892914142e-05, "loss": 0.0868, "step": 8560 }, { "epoch": 19.13, "learning_rate": 1.1188251065672382e-05, "loss": 0.0629, "step": 8570 }, { "epoch": 19.15, "learning_rate": 1.117030335685235e-05, "loss": 0.0594, "step": 8580 }, { "epoch": 19.17, "learning_rate": 1.1152351825082588e-05, "loss": 0.0558, "step": 8590 }, { "epoch": 19.2, "learning_rate": 1.1134396529004143e-05, "loss": 0.0621, "step": 8600 }, { "epoch": 19.22, "learning_rate": 1.1116437527270343e-05, "loss": 0.0687, "step": 8610 }, { "epoch": 19.24, "learning_rate": 1.109847487854663e-05, "loss": 0.0739, "step": 8620 }, { "epoch": 19.26, "learning_rate": 1.1080508641510357e-05, "loss": 0.069, "step": 8630 }, { "epoch": 19.29, "learning_rate": 1.1062538874850597e-05, "loss": 0.078, "step": 8640 }, { "epoch": 19.31, "learning_rate": 1.1044565637267957e-05, "loss": 0.0656, "step": 8650 }, { "epoch": 19.33, "learning_rate": 1.1026588987474379e-05, "loss": 0.1075, "step": 8660 }, { "epoch": 19.35, "learning_rate": 1.100860898419295e-05, "loss": 0.0751, "step": 8670 }, { "epoch": 19.38, "learning_rate": 1.0990625686157714e-05, "loss": 0.0578, "step": 8680 }, { "epoch": 19.4, "learning_rate": 1.097263915211348e-05, "loss": 0.0664, "step": 8690 }, { "epoch": 19.42, "learning_rate": 1.0954649440815625e-05, "loss": 0.0714, "step": 8700 }, { "epoch": 19.44, "learning_rate": 1.0936656611029901e-05, "loss": 0.0731, "step": 8710 }, { "epoch": 19.46, "learning_rate": 1.091866072153226e-05, "loss": 0.0647, "step": 8720 }, { "epoch": 19.49, "learning_rate": 1.090066183110863e-05, "loss": 0.0902, "step": 8730 }, { "epoch": 19.51, "learning_rate": 1.0882659998554759e-05, "loss": 0.0809, "step": 8740 }, { "epoch": 19.53, "learning_rate": 1.0864655282675997e-05, "loss": 0.0769, "step": 8750 }, { "epoch": 19.55, "learning_rate": 1.0846647742287116e-05, "loss": 0.0708, "step": 8760 }, { "epoch": 19.58, "learning_rate": 1.0828637436212111e-05, "loss": 0.0723, "step": 8770 }, { "epoch": 19.6, "learning_rate": 1.0810624423284012e-05, "loss": 0.0747, "step": 8780 }, { "epoch": 19.62, "learning_rate": 1.07926087623447e-05, "loss": 0.0634, "step": 8790 }, { "epoch": 19.64, "learning_rate": 1.0774590512244694e-05, "loss": 0.0708, "step": 8800 }, { "epoch": 19.67, "learning_rate": 1.0756569731842978e-05, "loss": 0.0698, "step": 8810 }, { "epoch": 19.69, "learning_rate": 1.07385464800068e-05, "loss": 0.0605, "step": 8820 }, { "epoch": 19.71, "learning_rate": 1.0720520815611476e-05, "loss": 0.0744, "step": 8830 }, { "epoch": 19.73, "learning_rate": 1.0702492797540214e-05, "loss": 0.0819, "step": 8840 }, { "epoch": 19.75, "learning_rate": 1.06844624846839e-05, "loss": 0.0727, "step": 8850 }, { "epoch": 19.78, "learning_rate": 1.0666429935940925e-05, "loss": 0.0738, "step": 8860 }, { "epoch": 19.8, "learning_rate": 1.0648395210216975e-05, "loss": 0.0805, "step": 8870 }, { "epoch": 19.82, "learning_rate": 1.0630358366424856e-05, "loss": 0.0687, "step": 8880 }, { "epoch": 19.84, "learning_rate": 1.0612319463484286e-05, "loss": 0.0622, "step": 8890 }, { "epoch": 19.87, "learning_rate": 1.0594278560321713e-05, "loss": 0.0894, "step": 8900 }, { "epoch": 19.89, "learning_rate": 1.0576235715870119e-05, "loss": 0.0682, "step": 8910 }, { "epoch": 19.91, "learning_rate": 1.0558190989068822e-05, "loss": 0.0812, "step": 8920 }, { "epoch": 19.93, "learning_rate": 1.0540144438863302e-05, "loss": 0.132, "step": 8930 }, { "epoch": 19.96, "learning_rate": 1.052209612420498e-05, "loss": 0.0985, "step": 8940 }, { "epoch": 19.98, "learning_rate": 1.050404610405105e-05, "loss": 0.0716, "step": 8950 }, { "epoch": 20.0, "learning_rate": 1.0485994437364278e-05, "loss": 0.0676, "step": 8960 }, { "epoch": 20.02, "learning_rate": 1.0467941183112801e-05, "loss": 0.0585, "step": 8970 }, { "epoch": 20.04, "learning_rate": 1.0449886400269952e-05, "loss": 0.0711, "step": 8980 }, { "epoch": 20.07, "learning_rate": 1.0431830147814049e-05, "loss": 0.0558, "step": 8990 }, { "epoch": 20.09, "learning_rate": 1.0413772484728211e-05, "loss": 0.0473, "step": 9000 }, { "epoch": 20.11, "learning_rate": 1.0395713470000173e-05, "loss": 0.058, "step": 9010 }, { "epoch": 20.13, "learning_rate": 1.0377653162622076e-05, "loss": 0.0631, "step": 9020 }, { "epoch": 20.16, "learning_rate": 1.0359591621590292e-05, "loss": 0.065, "step": 9030 }, { "epoch": 20.18, "learning_rate": 1.034152890590521e-05, "loss": 0.0676, "step": 9040 }, { "epoch": 20.2, "learning_rate": 1.0323465074571078e-05, "loss": 0.0845, "step": 9050 }, { "epoch": 20.22, "learning_rate": 1.0305400186595764e-05, "loss": 0.0723, "step": 9060 }, { "epoch": 20.25, "learning_rate": 1.0287334300990602e-05, "loss": 0.105, "step": 9070 }, { "epoch": 20.27, "learning_rate": 1.026926747677018e-05, "loss": 0.0714, "step": 9080 }, { "epoch": 20.29, "learning_rate": 1.025119977295216e-05, "loss": 0.0684, "step": 9090 }, { "epoch": 20.31, "learning_rate": 1.0233131248557067e-05, "loss": 0.0696, "step": 9100 }, { "epoch": 20.33, "learning_rate": 1.0215061962608111e-05, "loss": 0.0572, "step": 9110 }, { "epoch": 20.36, "learning_rate": 1.0196991974130986e-05, "loss": 0.0968, "step": 9120 }, { "epoch": 20.38, "learning_rate": 1.017892134215369e-05, "loss": 0.0655, "step": 9130 }, { "epoch": 20.4, "learning_rate": 1.0160850125706314e-05, "loss": 0.0611, "step": 9140 }, { "epoch": 20.42, "learning_rate": 1.0142778383820861e-05, "loss": 0.0839, "step": 9150 }, { "epoch": 20.45, "learning_rate": 1.0124706175531054e-05, "loss": 0.0655, "step": 9160 }, { "epoch": 20.47, "learning_rate": 1.0106633559872135e-05, "loss": 0.0528, "step": 9170 }, { "epoch": 20.49, "learning_rate": 1.0088560595880676e-05, "loss": 0.0706, "step": 9180 }, { "epoch": 20.51, "learning_rate": 1.0070487342594392e-05, "loss": 0.0676, "step": 9190 }, { "epoch": 20.54, "learning_rate": 1.005241385905194e-05, "loss": 0.0712, "step": 9200 }, { "epoch": 20.56, "learning_rate": 1.0034340204292728e-05, "loss": 0.0657, "step": 9210 }, { "epoch": 20.58, "learning_rate": 1.0016266437356727e-05, "loss": 0.0713, "step": 9220 }, { "epoch": 20.6, "learning_rate": 9.998192617284271e-06, "loss": 0.065, "step": 9230 }, { "epoch": 20.62, "learning_rate": 9.980118803115867e-06, "loss": 0.0734, "step": 9240 }, { "epoch": 20.65, "learning_rate": 9.962045053892004e-06, "loss": 0.061, "step": 9250 }, { "epoch": 20.67, "learning_rate": 9.94397142865296e-06, "loss": 0.0727, "step": 9260 }, { "epoch": 20.69, "learning_rate": 9.925897986438613e-06, "loss": 0.0702, "step": 9270 }, { "epoch": 20.71, "learning_rate": 9.907824786288226e-06, "loss": 0.0979, "step": 9280 }, { "epoch": 20.74, "learning_rate": 9.889751887240296e-06, "loss": 0.0534, "step": 9290 }, { "epoch": 20.76, "learning_rate": 9.87167934833231e-06, "loss": 0.0634, "step": 9300 }, { "epoch": 20.78, "learning_rate": 9.853607228600602e-06, "loss": 0.0566, "step": 9310 }, { "epoch": 20.8, "learning_rate": 9.835535587080118e-06, "loss": 0.0774, "step": 9320 }, { "epoch": 20.83, "learning_rate": 9.817464482804257e-06, "loss": 0.0737, "step": 9330 }, { "epoch": 20.85, "learning_rate": 9.799393974804651e-06, "loss": 0.0621, "step": 9340 }, { "epoch": 20.87, "learning_rate": 9.781324122110993e-06, "loss": 0.0648, "step": 9350 }, { "epoch": 20.89, "learning_rate": 9.763254983750829e-06, "loss": 0.0753, "step": 9360 }, { "epoch": 20.92, "learning_rate": 9.745186618749373e-06, "loss": 0.0769, "step": 9370 }, { "epoch": 20.94, "learning_rate": 9.727119086129321e-06, "loss": 0.0764, "step": 9380 }, { "epoch": 20.96, "learning_rate": 9.709052444910636e-06, "loss": 0.0763, "step": 9390 }, { "epoch": 20.98, "learning_rate": 9.690986754110378e-06, "loss": 0.0742, "step": 9400 }, { "epoch": 21.0, "learning_rate": 9.6729220727425e-06, "loss": 0.0998, "step": 9410 }, { "epoch": 21.03, "learning_rate": 9.654858459817663e-06, "loss": 0.0583, "step": 9420 }, { "epoch": 21.05, "learning_rate": 9.636795974343023e-06, "loss": 0.0642, "step": 9430 }, { "epoch": 21.07, "learning_rate": 9.61873467532207e-06, "loss": 0.0636, "step": 9440 }, { "epoch": 21.09, "learning_rate": 9.600674621754406e-06, "loss": 0.046, "step": 9450 }, { "epoch": 21.12, "learning_rate": 9.582615872635578e-06, "loss": 0.0551, "step": 9460 }, { "epoch": 21.14, "learning_rate": 9.564558486956853e-06, "loss": 0.0534, "step": 9470 }, { "epoch": 21.16, "learning_rate": 9.546502523705057e-06, "loss": 0.0825, "step": 9480 }, { "epoch": 21.18, "learning_rate": 9.528448041862375e-06, "loss": 0.0592, "step": 9490 }, { "epoch": 21.21, "learning_rate": 9.510395100406136e-06, "loss": 0.0598, "step": 9500 }, { "epoch": 21.23, "learning_rate": 9.492343758308651e-06, "loss": 0.0615, "step": 9510 }, { "epoch": 21.25, "learning_rate": 9.474294074536996e-06, "loss": 0.067, "step": 9520 }, { "epoch": 21.27, "learning_rate": 9.456246108052844e-06, "loss": 0.0526, "step": 9530 }, { "epoch": 21.29, "learning_rate": 9.438199917812241e-06, "loss": 0.0577, "step": 9540 }, { "epoch": 21.32, "learning_rate": 9.420155562765443e-06, "loss": 0.0627, "step": 9550 }, { "epoch": 21.34, "learning_rate": 9.402113101856705e-06, "loss": 0.0575, "step": 9560 }, { "epoch": 21.36, "learning_rate": 9.384072594024103e-06, "loss": 0.07, "step": 9570 }, { "epoch": 21.38, "learning_rate": 9.366034098199317e-06, "loss": 0.0578, "step": 9580 }, { "epoch": 21.41, "learning_rate": 9.347997673307473e-06, "loss": 0.0543, "step": 9590 }, { "epoch": 21.43, "learning_rate": 9.329963378266919e-06, "loss": 0.0553, "step": 9600 }, { "epoch": 21.45, "learning_rate": 9.31193127198905e-06, "loss": 0.0581, "step": 9610 }, { "epoch": 21.47, "learning_rate": 9.293901413378116e-06, "loss": 0.0494, "step": 9620 }, { "epoch": 21.5, "learning_rate": 9.275873861331012e-06, "loss": 0.0688, "step": 9630 }, { "epoch": 21.52, "learning_rate": 9.257848674737112e-06, "loss": 0.0548, "step": 9640 }, { "epoch": 21.54, "learning_rate": 9.239825912478054e-06, "loss": 0.0744, "step": 9650 }, { "epoch": 21.56, "learning_rate": 9.221805633427564e-06, "loss": 0.0701, "step": 9660 }, { "epoch": 21.58, "learning_rate": 9.203787896451246e-06, "loss": 0.0495, "step": 9670 }, { "epoch": 21.61, "learning_rate": 9.185772760406408e-06, "loss": 0.0528, "step": 9680 }, { "epoch": 21.63, "learning_rate": 9.167760284141859e-06, "loss": 0.0656, "step": 9690 }, { "epoch": 21.65, "learning_rate": 9.149750526497725e-06, "loss": 0.0549, "step": 9700 }, { "epoch": 21.67, "learning_rate": 9.131743546305235e-06, "loss": 0.0596, "step": 9710 }, { "epoch": 21.7, "learning_rate": 9.113739402386566e-06, "loss": 0.0602, "step": 9720 }, { "epoch": 21.72, "learning_rate": 9.095738153554624e-06, "loss": 0.0816, "step": 9730 }, { "epoch": 21.74, "learning_rate": 9.077739858612843e-06, "loss": 0.0752, "step": 9740 }, { "epoch": 21.76, "learning_rate": 9.059744576355027e-06, "loss": 0.0666, "step": 9750 }, { "epoch": 21.79, "learning_rate": 9.041752365565125e-06, "loss": 0.0584, "step": 9760 }, { "epoch": 21.81, "learning_rate": 9.023763285017065e-06, "loss": 0.063, "step": 9770 }, { "epoch": 21.83, "learning_rate": 9.005777393474534e-06, "loss": 0.0545, "step": 9780 }, { "epoch": 21.85, "learning_rate": 8.987794749690819e-06, "loss": 0.0552, "step": 9790 }, { "epoch": 21.88, "learning_rate": 8.969815412408583e-06, "loss": 0.0589, "step": 9800 }, { "epoch": 21.9, "learning_rate": 8.951839440359701e-06, "loss": 0.0698, "step": 9810 }, { "epoch": 21.92, "learning_rate": 8.93386689226504e-06, "loss": 0.0911, "step": 9820 }, { "epoch": 21.94, "learning_rate": 8.915897826834295e-06, "loss": 0.0779, "step": 9830 }, { "epoch": 21.96, "learning_rate": 8.89793230276578e-06, "loss": 0.0606, "step": 9840 }, { "epoch": 21.99, "learning_rate": 8.879970378746238e-06, "loss": 0.0787, "step": 9850 }, { "epoch": 22.01, "learning_rate": 8.862012113450662e-06, "loss": 0.0574, "step": 9860 }, { "epoch": 22.03, "learning_rate": 8.844057565542074e-06, "loss": 0.0581, "step": 9870 }, { "epoch": 22.05, "learning_rate": 8.826106793671376e-06, "loss": 0.0437, "step": 9880 }, { "epoch": 22.08, "learning_rate": 8.808159856477115e-06, "loss": 0.069, "step": 9890 }, { "epoch": 22.1, "learning_rate": 8.790216812585327e-06, "loss": 0.046, "step": 9900 }, { "epoch": 22.12, "learning_rate": 8.772277720609312e-06, "loss": 0.0825, "step": 9910 }, { "epoch": 22.14, "learning_rate": 8.754342639149486e-06, "loss": 0.0659, "step": 9920 }, { "epoch": 22.17, "learning_rate": 8.736411626793139e-06, "loss": 0.0649, "step": 9930 }, { "epoch": 22.19, "learning_rate": 8.718484742114285e-06, "loss": 0.0563, "step": 9940 }, { "epoch": 22.21, "learning_rate": 8.700562043673448e-06, "loss": 0.043, "step": 9950 }, { "epoch": 22.23, "learning_rate": 8.682643590017474e-06, "loss": 0.0482, "step": 9960 }, { "epoch": 22.25, "learning_rate": 8.664729439679354e-06, "loss": 0.0436, "step": 9970 }, { "epoch": 22.28, "learning_rate": 8.646819651178008e-06, "loss": 0.0536, "step": 9980 }, { "epoch": 22.3, "learning_rate": 8.628914283018119e-06, "loss": 0.0615, "step": 9990 }, { "epoch": 22.32, "learning_rate": 8.61101339368992e-06, "loss": 0.0461, "step": 10000 }, { "epoch": 22.34, "learning_rate": 8.593117041669024e-06, "loss": 0.0605, "step": 10010 }, { "epoch": 22.37, "learning_rate": 8.57522528541621e-06, "loss": 0.07, "step": 10020 }, { "epoch": 22.39, "learning_rate": 8.55733818337726e-06, "loss": 0.059, "step": 10030 }, { "epoch": 22.41, "learning_rate": 8.539455793982737e-06, "loss": 0.047, "step": 10040 }, { "epoch": 22.43, "learning_rate": 8.521578175647823e-06, "loss": 0.0859, "step": 10050 }, { "epoch": 22.46, "learning_rate": 8.503705386772098e-06, "loss": 0.0596, "step": 10060 }, { "epoch": 22.48, "learning_rate": 8.485837485739384e-06, "loss": 0.0516, "step": 10070 }, { "epoch": 22.5, "learning_rate": 8.467974530917524e-06, "loss": 0.0495, "step": 10080 }, { "epoch": 22.52, "learning_rate": 8.450116580658208e-06, "loss": 0.0562, "step": 10090 }, { "epoch": 22.54, "learning_rate": 8.432263693296783e-06, "loss": 0.0549, "step": 10100 }, { "epoch": 22.57, "learning_rate": 8.414415927152042e-06, "loss": 0.0671, "step": 10110 }, { "epoch": 22.59, "learning_rate": 8.396573340526069e-06, "loss": 0.0678, "step": 10120 }, { "epoch": 22.61, "learning_rate": 8.37873599170401e-06, "loss": 0.0668, "step": 10130 }, { "epoch": 22.63, "learning_rate": 8.360903938953914e-06, "loss": 0.0544, "step": 10140 }, { "epoch": 22.66, "learning_rate": 8.343077240526522e-06, "loss": 0.0684, "step": 10150 }, { "epoch": 22.68, "learning_rate": 8.325255954655093e-06, "loss": 0.0971, "step": 10160 }, { "epoch": 22.7, "learning_rate": 8.307440139555192e-06, "loss": 0.0519, "step": 10170 }, { "epoch": 22.72, "learning_rate": 8.289629853424526e-06, "loss": 0.057, "step": 10180 }, { "epoch": 22.75, "learning_rate": 8.271825154442732e-06, "loss": 0.0626, "step": 10190 }, { "epoch": 22.77, "learning_rate": 8.2540261007712e-06, "loss": 0.0477, "step": 10200 }, { "epoch": 22.79, "learning_rate": 8.236232750552881e-06, "loss": 0.0616, "step": 10210 }, { "epoch": 22.81, "learning_rate": 8.218445161912088e-06, "loss": 0.0492, "step": 10220 }, { "epoch": 22.83, "learning_rate": 8.20066339295432e-06, "loss": 0.0723, "step": 10230 }, { "epoch": 22.86, "learning_rate": 8.182887501766059e-06, "loss": 0.0562, "step": 10240 }, { "epoch": 22.88, "learning_rate": 8.165117546414595e-06, "loss": 0.0536, "step": 10250 }, { "epoch": 22.9, "learning_rate": 8.147353584947818e-06, "loss": 0.0571, "step": 10260 }, { "epoch": 22.92, "learning_rate": 8.129595675394045e-06, "loss": 0.0767, "step": 10270 }, { "epoch": 22.95, "learning_rate": 8.11184387576182e-06, "loss": 0.0885, "step": 10280 }, { "epoch": 22.97, "learning_rate": 8.094098244039734e-06, "loss": 0.0684, "step": 10290 }, { "epoch": 22.99, "learning_rate": 8.076358838196216e-06, "loss": 0.0601, "step": 10300 }, { "epoch": 23.01, "learning_rate": 8.058625716179375e-06, "loss": 0.0388, "step": 10310 }, { "epoch": 23.04, "learning_rate": 8.04089893591678e-06, "loss": 0.0422, "step": 10320 }, { "epoch": 23.06, "learning_rate": 8.023178555315291e-06, "loss": 0.0459, "step": 10330 }, { "epoch": 23.08, "learning_rate": 8.005464632260862e-06, "loss": 0.0329, "step": 10340 }, { "epoch": 23.1, "learning_rate": 7.987757224618346e-06, "loss": 0.051, "step": 10350 }, { "epoch": 23.12, "learning_rate": 7.970056390231323e-06, "loss": 0.0545, "step": 10360 }, { "epoch": 23.15, "learning_rate": 7.952362186921889e-06, "loss": 0.0463, "step": 10370 }, { "epoch": 23.17, "learning_rate": 7.934674672490488e-06, "loss": 0.0614, "step": 10380 }, { "epoch": 23.19, "learning_rate": 7.916993904715708e-06, "loss": 0.0491, "step": 10390 }, { "epoch": 23.21, "learning_rate": 7.899319941354107e-06, "loss": 0.0441, "step": 10400 }, { "epoch": 23.24, "learning_rate": 7.881652840140001e-06, "loss": 0.0505, "step": 10410 }, { "epoch": 23.26, "learning_rate": 7.863992658785302e-06, "loss": 0.0517, "step": 10420 }, { "epoch": 23.28, "learning_rate": 7.846339454979312e-06, "loss": 0.0604, "step": 10430 }, { "epoch": 23.3, "learning_rate": 7.828693286388542e-06, "loss": 0.0441, "step": 10440 }, { "epoch": 23.33, "learning_rate": 7.811054210656526e-06, "loss": 0.0535, "step": 10450 }, { "epoch": 23.35, "learning_rate": 7.793422285403614e-06, "loss": 0.0534, "step": 10460 }, { "epoch": 23.37, "learning_rate": 7.775797568226816e-06, "loss": 0.0566, "step": 10470 }, { "epoch": 23.39, "learning_rate": 7.758180116699578e-06, "loss": 0.085, "step": 10480 }, { "epoch": 23.42, "learning_rate": 7.74056998837163e-06, "loss": 0.0563, "step": 10490 }, { "epoch": 23.44, "learning_rate": 7.722967240768761e-06, "loss": 0.0613, "step": 10500 }, { "epoch": 23.46, "learning_rate": 7.705371931392668e-06, "loss": 0.0413, "step": 10510 }, { "epoch": 23.48, "learning_rate": 7.687784117720736e-06, "loss": 0.0538, "step": 10520 }, { "epoch": 23.5, "learning_rate": 7.670203857205877e-06, "loss": 0.0576, "step": 10530 }, { "epoch": 23.53, "learning_rate": 7.652631207276311e-06, "loss": 0.0467, "step": 10540 }, { "epoch": 23.55, "learning_rate": 7.635066225335417e-06, "loss": 0.0404, "step": 10550 }, { "epoch": 23.57, "learning_rate": 7.617508968761519e-06, "loss": 0.0625, "step": 10560 }, { "epoch": 23.59, "learning_rate": 7.599959494907695e-06, "loss": 0.0598, "step": 10570 }, { "epoch": 23.62, "learning_rate": 7.582417861101614e-06, "loss": 0.0449, "step": 10580 }, { "epoch": 23.64, "learning_rate": 7.564884124645325e-06, "loss": 0.072, "step": 10590 }, { "epoch": 23.66, "learning_rate": 7.547358342815089e-06, "loss": 0.0607, "step": 10600 }, { "epoch": 23.68, "learning_rate": 7.5298405728611645e-06, "loss": 0.0687, "step": 10610 }, { "epoch": 23.71, "learning_rate": 7.512330872007659e-06, "loss": 0.0524, "step": 10620 }, { "epoch": 23.73, "learning_rate": 7.494829297452306e-06, "loss": 0.048, "step": 10630 }, { "epoch": 23.75, "learning_rate": 7.4773359063663045e-06, "loss": 0.0569, "step": 10640 }, { "epoch": 23.77, "learning_rate": 7.459850755894108e-06, "loss": 0.0585, "step": 10650 }, { "epoch": 23.79, "learning_rate": 7.442373903153266e-06, "loss": 0.0859, "step": 10660 }, { "epoch": 23.82, "learning_rate": 7.424905405234209e-06, "loss": 0.0498, "step": 10670 }, { "epoch": 23.84, "learning_rate": 7.407445319200083e-06, "loss": 0.0523, "step": 10680 }, { "epoch": 23.86, "learning_rate": 7.38999370208656e-06, "loss": 0.0711, "step": 10690 }, { "epoch": 23.88, "learning_rate": 7.37255061090163e-06, "loss": 0.0398, "step": 10700 }, { "epoch": 23.91, "learning_rate": 7.355116102625451e-06, "loss": 0.0765, "step": 10710 }, { "epoch": 23.93, "learning_rate": 7.337690234210132e-06, "loss": 0.0547, "step": 10720 }, { "epoch": 23.95, "learning_rate": 7.320273062579568e-06, "loss": 0.055, "step": 10730 }, { "epoch": 23.97, "learning_rate": 7.3028646446292295e-06, "loss": 0.0524, "step": 10740 }, { "epoch": 24.0, "learning_rate": 7.28546503722601e-06, "loss": 0.0479, "step": 10750 }, { "epoch": 24.02, "learning_rate": 7.268074297208008e-06, "loss": 0.0418, "step": 10760 }, { "epoch": 24.04, "learning_rate": 7.250692481384366e-06, "loss": 0.0531, "step": 10770 }, { "epoch": 24.06, "learning_rate": 7.233319646535067e-06, "loss": 0.0401, "step": 10780 }, { "epoch": 24.08, "learning_rate": 7.21595584941076e-06, "loss": 0.0646, "step": 10790 }, { "epoch": 24.11, "learning_rate": 7.198601146732573e-06, "loss": 0.0477, "step": 10800 }, { "epoch": 24.13, "learning_rate": 7.181255595191919e-06, "loss": 0.0331, "step": 10810 }, { "epoch": 24.15, "learning_rate": 7.1639192514503265e-06, "loss": 0.0463, "step": 10820 }, { "epoch": 24.17, "learning_rate": 7.146592172139234e-06, "loss": 0.0472, "step": 10830 }, { "epoch": 24.2, "learning_rate": 7.129274413859832e-06, "loss": 0.0462, "step": 10840 }, { "epoch": 24.22, "learning_rate": 7.111966033182845e-06, "loss": 0.0404, "step": 10850 }, { "epoch": 24.24, "learning_rate": 7.094667086648381e-06, "loss": 0.0602, "step": 10860 }, { "epoch": 24.26, "learning_rate": 7.077377630765716e-06, "loss": 0.044, "step": 10870 }, { "epoch": 24.29, "learning_rate": 7.060097722013137e-06, "loss": 0.0413, "step": 10880 }, { "epoch": 24.31, "learning_rate": 7.042827416837728e-06, "loss": 0.0369, "step": 10890 }, { "epoch": 24.33, "learning_rate": 7.025566771655219e-06, "loss": 0.055, "step": 10900 }, { "epoch": 24.35, "learning_rate": 7.00831584284977e-06, "loss": 0.0454, "step": 10910 }, { "epoch": 24.38, "learning_rate": 6.991074686773809e-06, "loss": 0.0475, "step": 10920 }, { "epoch": 24.4, "learning_rate": 6.973843359747845e-06, "loss": 0.0545, "step": 10930 }, { "epoch": 24.42, "learning_rate": 6.95662191806026e-06, "loss": 0.0459, "step": 10940 }, { "epoch": 24.44, "learning_rate": 6.939410417967168e-06, "loss": 0.053, "step": 10950 }, { "epoch": 24.46, "learning_rate": 6.922208915692186e-06, "loss": 0.0417, "step": 10960 }, { "epoch": 24.49, "learning_rate": 6.905017467426291e-06, "loss": 0.0568, "step": 10970 }, { "epoch": 24.51, "learning_rate": 6.887836129327602e-06, "loss": 0.076, "step": 10980 }, { "epoch": 24.53, "learning_rate": 6.870664957521225e-06, "loss": 0.0478, "step": 10990 }, { "epoch": 24.55, "learning_rate": 6.85350400809904e-06, "loss": 0.0418, "step": 11000 } ], "max_steps": 17920, "num_train_epochs": 40, "total_flos": 6.444053730584166e+16, "trial_name": null, "trial_params": null }