{ "best_metric": null, "best_model_checkpoint": null, "epoch": 4.999259295837243, "eval_steps": 7000.0, "global_step": 25310, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 1.975503753457132e-07, "loss": 7.9807, "step": 10 }, { "epoch": 0.0, "learning_rate": 3.951007506914264e-07, "loss": 7.3025, "step": 20 }, { "epoch": 0.01, "learning_rate": 5.926511260371394e-07, "loss": 6.442, "step": 30 }, { "epoch": 0.01, "learning_rate": 7.902015013828528e-07, "loss": 4.9928, "step": 40 }, { "epoch": 0.01, "learning_rate": 9.877518767285659e-07, "loss": 3.0368, "step": 50 }, { "epoch": 0.01, "learning_rate": 1.1853022520742789e-06, "loss": 1.5885, "step": 60 }, { "epoch": 0.01, "learning_rate": 1.382852627419992e-06, "loss": 1.118, "step": 70 }, { "epoch": 0.02, "learning_rate": 1.5804030027657055e-06, "loss": 1.0564, "step": 80 }, { "epoch": 0.02, "learning_rate": 1.7779533781114185e-06, "loss": 1.1117, "step": 90 }, { "epoch": 0.02, "learning_rate": 1.9755037534571318e-06, "loss": 1.1398, "step": 100 }, { "epoch": 0.02, "learning_rate": 2.173054128802845e-06, "loss": 0.9834, "step": 110 }, { "epoch": 0.02, "learning_rate": 2.3706045041485578e-06, "loss": 1.0677, "step": 120 }, { "epoch": 0.03, "learning_rate": 2.5681548794942714e-06, "loss": 0.9412, "step": 130 }, { "epoch": 0.03, "learning_rate": 2.765705254839984e-06, "loss": 0.9347, "step": 140 }, { "epoch": 0.03, "learning_rate": 2.9632556301856974e-06, "loss": 0.9574, "step": 150 }, { "epoch": 0.03, "learning_rate": 3.160806005531411e-06, "loss": 0.9889, "step": 160 }, { "epoch": 0.03, "learning_rate": 3.3583563808771234e-06, "loss": 0.9102, "step": 170 }, { "epoch": 0.04, "learning_rate": 3.555906756222837e-06, "loss": 0.9191, "step": 180 }, { "epoch": 0.04, "learning_rate": 3.7534571315685503e-06, "loss": 0.9766, "step": 190 }, { "epoch": 0.04, "learning_rate": 3.9510075069142635e-06, "loss": 0.9135, "step": 200 }, { "epoch": 0.04, "learning_rate": 4.148557882259976e-06, "loss": 0.8757, "step": 210 }, { "epoch": 0.04, "learning_rate": 4.34610825760569e-06, "loss": 0.9067, "step": 220 }, { "epoch": 0.05, "learning_rate": 4.543658632951403e-06, "loss": 0.9219, "step": 230 }, { "epoch": 0.05, "learning_rate": 4.7412090082971155e-06, "loss": 0.8996, "step": 240 }, { "epoch": 0.05, "learning_rate": 4.938759383642829e-06, "loss": 0.9467, "step": 250 }, { "epoch": 0.05, "learning_rate": 5.136309758988543e-06, "loss": 0.8877, "step": 260 }, { "epoch": 0.05, "learning_rate": 5.333860134334256e-06, "loss": 0.8398, "step": 270 }, { "epoch": 0.06, "learning_rate": 5.531410509679968e-06, "loss": 0.8764, "step": 280 }, { "epoch": 0.06, "learning_rate": 5.728960885025682e-06, "loss": 0.8809, "step": 290 }, { "epoch": 0.06, "learning_rate": 5.926511260371395e-06, "loss": 0.8699, "step": 300 }, { "epoch": 0.06, "learning_rate": 6.124061635717108e-06, "loss": 0.8377, "step": 310 }, { "epoch": 0.06, "learning_rate": 6.321612011062822e-06, "loss": 0.8943, "step": 320 }, { "epoch": 0.07, "learning_rate": 6.519162386408534e-06, "loss": 0.8393, "step": 330 }, { "epoch": 0.07, "learning_rate": 6.716712761754247e-06, "loss": 0.8192, "step": 340 }, { "epoch": 0.07, "learning_rate": 6.9142631370999605e-06, "loss": 0.903, "step": 350 }, { "epoch": 0.07, "learning_rate": 7.111813512445674e-06, "loss": 0.8917, "step": 360 }, { "epoch": 0.07, "learning_rate": 7.309363887791387e-06, "loss": 0.916, "step": 370 }, { "epoch": 0.08, "learning_rate": 7.506914263137101e-06, "loss": 0.864, "step": 380 }, { "epoch": 0.08, "learning_rate": 7.704464638482814e-06, "loss": 0.7809, "step": 390 }, { "epoch": 0.08, "learning_rate": 7.902015013828527e-06, "loss": 0.885, "step": 400 }, { "epoch": 0.08, "learning_rate": 8.09956538917424e-06, "loss": 0.8183, "step": 410 }, { "epoch": 0.08, "learning_rate": 8.297115764519953e-06, "loss": 0.7897, "step": 420 }, { "epoch": 0.08, "learning_rate": 8.494666139865665e-06, "loss": 0.8315, "step": 430 }, { "epoch": 0.09, "learning_rate": 8.69221651521138e-06, "loss": 0.8266, "step": 440 }, { "epoch": 0.09, "learning_rate": 8.889766890557093e-06, "loss": 0.8755, "step": 450 }, { "epoch": 0.09, "learning_rate": 9.087317265902805e-06, "loss": 0.8103, "step": 460 }, { "epoch": 0.09, "learning_rate": 9.28486764124852e-06, "loss": 0.8226, "step": 470 }, { "epoch": 0.09, "learning_rate": 9.482418016594231e-06, "loss": 0.8174, "step": 480 }, { "epoch": 0.1, "learning_rate": 9.679968391939946e-06, "loss": 0.8336, "step": 490 }, { "epoch": 0.1, "learning_rate": 9.877518767285658e-06, "loss": 0.8214, "step": 500 }, { "epoch": 0.1, "learning_rate": 1.0075069142631371e-05, "loss": 0.8534, "step": 510 }, { "epoch": 0.1, "learning_rate": 1.0272619517977086e-05, "loss": 0.8148, "step": 520 }, { "epoch": 0.1, "learning_rate": 1.0470169893322798e-05, "loss": 0.8032, "step": 530 }, { "epoch": 0.11, "learning_rate": 1.0667720268668511e-05, "loss": 0.7561, "step": 540 }, { "epoch": 0.11, "learning_rate": 1.0865270644014224e-05, "loss": 0.8219, "step": 550 }, { "epoch": 0.11, "learning_rate": 1.1062821019359937e-05, "loss": 0.8108, "step": 560 }, { "epoch": 0.11, "learning_rate": 1.126037139470565e-05, "loss": 0.7673, "step": 570 }, { "epoch": 0.11, "learning_rate": 1.1457921770051364e-05, "loss": 0.7731, "step": 580 }, { "epoch": 0.12, "learning_rate": 1.1655472145397077e-05, "loss": 0.8135, "step": 590 }, { "epoch": 0.12, "learning_rate": 1.185302252074279e-05, "loss": 0.8525, "step": 600 }, { "epoch": 0.12, "learning_rate": 1.2050572896088504e-05, "loss": 0.8473, "step": 610 }, { "epoch": 0.12, "learning_rate": 1.2248123271434215e-05, "loss": 0.7937, "step": 620 }, { "epoch": 0.12, "learning_rate": 1.244567364677993e-05, "loss": 0.7924, "step": 630 }, { "epoch": 0.13, "learning_rate": 1.2643224022125644e-05, "loss": 0.7778, "step": 640 }, { "epoch": 0.13, "learning_rate": 1.2840774397471355e-05, "loss": 0.8282, "step": 650 }, { "epoch": 0.13, "learning_rate": 1.3038324772817068e-05, "loss": 0.811, "step": 660 }, { "epoch": 0.13, "learning_rate": 1.3235875148162783e-05, "loss": 0.7167, "step": 670 }, { "epoch": 0.13, "learning_rate": 1.3433425523508494e-05, "loss": 0.8321, "step": 680 }, { "epoch": 0.14, "learning_rate": 1.363097589885421e-05, "loss": 0.7477, "step": 690 }, { "epoch": 0.14, "learning_rate": 1.3828526274199921e-05, "loss": 0.8232, "step": 700 }, { "epoch": 0.14, "learning_rate": 1.4026076649545635e-05, "loss": 0.799, "step": 710 }, { "epoch": 0.14, "learning_rate": 1.4223627024891348e-05, "loss": 0.8554, "step": 720 }, { "epoch": 0.14, "learning_rate": 1.442117740023706e-05, "loss": 0.778, "step": 730 }, { "epoch": 0.15, "learning_rate": 1.4618727775582774e-05, "loss": 0.8337, "step": 740 }, { "epoch": 0.15, "learning_rate": 1.4816278150928487e-05, "loss": 0.7793, "step": 750 }, { "epoch": 0.15, "learning_rate": 1.5013828526274201e-05, "loss": 0.7812, "step": 760 }, { "epoch": 0.15, "learning_rate": 1.5211378901619914e-05, "loss": 0.7837, "step": 770 }, { "epoch": 0.15, "learning_rate": 1.540892927696563e-05, "loss": 0.7714, "step": 780 }, { "epoch": 0.16, "learning_rate": 1.560647965231134e-05, "loss": 0.74, "step": 790 }, { "epoch": 0.16, "learning_rate": 1.5804030027657054e-05, "loss": 0.7725, "step": 800 }, { "epoch": 0.16, "learning_rate": 1.6001580403002767e-05, "loss": 0.7618, "step": 810 }, { "epoch": 0.16, "learning_rate": 1.619913077834848e-05, "loss": 0.767, "step": 820 }, { "epoch": 0.16, "learning_rate": 1.6396681153694192e-05, "loss": 0.8257, "step": 830 }, { "epoch": 0.17, "learning_rate": 1.6594231529039905e-05, "loss": 0.8116, "step": 840 }, { "epoch": 0.17, "learning_rate": 1.679178190438562e-05, "loss": 0.7649, "step": 850 }, { "epoch": 0.17, "learning_rate": 1.698933227973133e-05, "loss": 0.7616, "step": 860 }, { "epoch": 0.17, "learning_rate": 1.7186882655077047e-05, "loss": 0.7219, "step": 870 }, { "epoch": 0.17, "learning_rate": 1.738443303042276e-05, "loss": 0.7532, "step": 880 }, { "epoch": 0.18, "learning_rate": 1.758198340576847e-05, "loss": 0.7122, "step": 890 }, { "epoch": 0.18, "learning_rate": 1.7779533781114185e-05, "loss": 0.8242, "step": 900 }, { "epoch": 0.18, "learning_rate": 1.7977084156459898e-05, "loss": 0.7554, "step": 910 }, { "epoch": 0.18, "learning_rate": 1.817463453180561e-05, "loss": 0.7588, "step": 920 }, { "epoch": 0.18, "learning_rate": 1.8372184907151324e-05, "loss": 0.7404, "step": 930 }, { "epoch": 0.19, "learning_rate": 1.856973528249704e-05, "loss": 0.7907, "step": 940 }, { "epoch": 0.19, "learning_rate": 1.876728565784275e-05, "loss": 0.7775, "step": 950 }, { "epoch": 0.19, "learning_rate": 1.8964836033188462e-05, "loss": 0.7544, "step": 960 }, { "epoch": 0.19, "learning_rate": 1.9162386408534178e-05, "loss": 0.8053, "step": 970 }, { "epoch": 0.19, "learning_rate": 1.935993678387989e-05, "loss": 0.7226, "step": 980 }, { "epoch": 0.2, "learning_rate": 1.9557487159225604e-05, "loss": 0.7482, "step": 990 }, { "epoch": 0.2, "learning_rate": 1.9755037534571317e-05, "loss": 0.7171, "step": 1000 }, { "epoch": 0.2, "learning_rate": 1.995258790991703e-05, "loss": 0.7421, "step": 1010 }, { "epoch": 0.2, "learning_rate": 2.0150138285262742e-05, "loss": 0.7356, "step": 1020 }, { "epoch": 0.2, "learning_rate": 2.0347688660608455e-05, "loss": 0.713, "step": 1030 }, { "epoch": 0.21, "learning_rate": 2.054523903595417e-05, "loss": 0.7705, "step": 1040 }, { "epoch": 0.21, "learning_rate": 2.074278941129988e-05, "loss": 0.7501, "step": 1050 }, { "epoch": 0.21, "learning_rate": 2.0940339786645597e-05, "loss": 0.7572, "step": 1060 }, { "epoch": 0.21, "learning_rate": 2.113789016199131e-05, "loss": 0.7382, "step": 1070 }, { "epoch": 0.21, "learning_rate": 2.1335440537337022e-05, "loss": 0.7596, "step": 1080 }, { "epoch": 0.22, "learning_rate": 2.1532990912682735e-05, "loss": 0.7063, "step": 1090 }, { "epoch": 0.22, "learning_rate": 2.1730541288028448e-05, "loss": 0.7239, "step": 1100 }, { "epoch": 0.22, "learning_rate": 2.192809166337416e-05, "loss": 0.7648, "step": 1110 }, { "epoch": 0.22, "learning_rate": 2.2125642038719874e-05, "loss": 0.7539, "step": 1120 }, { "epoch": 0.22, "learning_rate": 2.232319241406559e-05, "loss": 0.7498, "step": 1130 }, { "epoch": 0.23, "learning_rate": 2.25207427894113e-05, "loss": 0.7647, "step": 1140 }, { "epoch": 0.23, "learning_rate": 2.2718293164757015e-05, "loss": 0.7657, "step": 1150 }, { "epoch": 0.23, "learning_rate": 2.2915843540102728e-05, "loss": 0.7451, "step": 1160 }, { "epoch": 0.23, "learning_rate": 2.3113393915448438e-05, "loss": 0.7528, "step": 1170 }, { "epoch": 0.23, "learning_rate": 2.3310944290794154e-05, "loss": 0.7548, "step": 1180 }, { "epoch": 0.24, "learning_rate": 2.3508494666139867e-05, "loss": 0.7616, "step": 1190 }, { "epoch": 0.24, "learning_rate": 2.370604504148558e-05, "loss": 0.7654, "step": 1200 }, { "epoch": 0.24, "learning_rate": 2.3903595416831292e-05, "loss": 0.7681, "step": 1210 }, { "epoch": 0.24, "learning_rate": 2.410114579217701e-05, "loss": 0.7302, "step": 1220 }, { "epoch": 0.24, "learning_rate": 2.4298696167522718e-05, "loss": 0.7516, "step": 1230 }, { "epoch": 0.24, "learning_rate": 2.449624654286843e-05, "loss": 0.7331, "step": 1240 }, { "epoch": 0.25, "learning_rate": 2.4693796918214147e-05, "loss": 0.7709, "step": 1250 }, { "epoch": 0.25, "learning_rate": 2.489134729355986e-05, "loss": 0.696, "step": 1260 }, { "epoch": 0.25, "learning_rate": 2.508889766890557e-05, "loss": 0.7537, "step": 1270 }, { "epoch": 0.25, "learning_rate": 2.528644804425129e-05, "loss": 0.7789, "step": 1280 }, { "epoch": 0.25, "learning_rate": 2.5483998419596998e-05, "loss": 0.7277, "step": 1290 }, { "epoch": 0.26, "learning_rate": 2.568154879494271e-05, "loss": 0.6968, "step": 1300 }, { "epoch": 0.26, "learning_rate": 2.5879099170288423e-05, "loss": 0.7329, "step": 1310 }, { "epoch": 0.26, "learning_rate": 2.6076649545634136e-05, "loss": 0.7501, "step": 1320 }, { "epoch": 0.26, "learning_rate": 2.6274199920979852e-05, "loss": 0.7224, "step": 1330 }, { "epoch": 0.26, "learning_rate": 2.6471750296325565e-05, "loss": 0.7276, "step": 1340 }, { "epoch": 0.27, "learning_rate": 2.6669300671671278e-05, "loss": 0.6827, "step": 1350 }, { "epoch": 0.27, "learning_rate": 2.6866851047016987e-05, "loss": 0.7762, "step": 1360 }, { "epoch": 0.27, "learning_rate": 2.7064401422362707e-05, "loss": 0.7124, "step": 1370 }, { "epoch": 0.27, "learning_rate": 2.726195179770842e-05, "loss": 0.6812, "step": 1380 }, { "epoch": 0.27, "learning_rate": 2.745950217305413e-05, "loss": 0.6922, "step": 1390 }, { "epoch": 0.28, "learning_rate": 2.7657052548399842e-05, "loss": 0.7216, "step": 1400 }, { "epoch": 0.28, "learning_rate": 2.7854602923745555e-05, "loss": 0.6899, "step": 1410 }, { "epoch": 0.28, "learning_rate": 2.805215329909127e-05, "loss": 0.7433, "step": 1420 }, { "epoch": 0.28, "learning_rate": 2.8249703674436984e-05, "loss": 0.7103, "step": 1430 }, { "epoch": 0.28, "learning_rate": 2.8447254049782697e-05, "loss": 0.7348, "step": 1440 }, { "epoch": 0.29, "learning_rate": 2.8644804425128406e-05, "loss": 0.8088, "step": 1450 }, { "epoch": 0.29, "learning_rate": 2.884235480047412e-05, "loss": 0.7432, "step": 1460 }, { "epoch": 0.29, "learning_rate": 2.903990517581984e-05, "loss": 0.7381, "step": 1470 }, { "epoch": 0.29, "learning_rate": 2.9237455551165548e-05, "loss": 0.7195, "step": 1480 }, { "epoch": 0.29, "learning_rate": 2.943500592651126e-05, "loss": 0.7038, "step": 1490 }, { "epoch": 0.3, "learning_rate": 2.9632556301856973e-05, "loss": 0.6643, "step": 1500 }, { "epoch": 0.3, "learning_rate": 2.983010667720269e-05, "loss": 0.7159, "step": 1510 }, { "epoch": 0.3, "learning_rate": 3.0027657052548402e-05, "loss": 0.6977, "step": 1520 }, { "epoch": 0.3, "learning_rate": 3.0225207427894115e-05, "loss": 0.713, "step": 1530 }, { "epoch": 0.3, "learning_rate": 3.0422757803239828e-05, "loss": 0.6713, "step": 1540 }, { "epoch": 0.31, "learning_rate": 3.062030817858554e-05, "loss": 0.6948, "step": 1550 }, { "epoch": 0.31, "learning_rate": 3.081785855393126e-05, "loss": 0.6964, "step": 1560 }, { "epoch": 0.31, "learning_rate": 3.101540892927697e-05, "loss": 0.7344, "step": 1570 }, { "epoch": 0.31, "learning_rate": 3.121295930462268e-05, "loss": 0.6921, "step": 1580 }, { "epoch": 0.31, "learning_rate": 3.141050967996839e-05, "loss": 0.7013, "step": 1590 }, { "epoch": 0.32, "learning_rate": 3.160806005531411e-05, "loss": 0.6525, "step": 1600 }, { "epoch": 0.32, "learning_rate": 3.180561043065982e-05, "loss": 0.6412, "step": 1610 }, { "epoch": 0.32, "learning_rate": 3.2003160806005534e-05, "loss": 0.6697, "step": 1620 }, { "epoch": 0.32, "learning_rate": 3.2200711181351246e-05, "loss": 0.7215, "step": 1630 }, { "epoch": 0.32, "learning_rate": 3.239826155669696e-05, "loss": 0.6918, "step": 1640 }, { "epoch": 0.33, "learning_rate": 3.259581193204267e-05, "loss": 0.7667, "step": 1650 }, { "epoch": 0.33, "learning_rate": 3.2793362307388385e-05, "loss": 0.69, "step": 1660 }, { "epoch": 0.33, "learning_rate": 3.29909126827341e-05, "loss": 0.6899, "step": 1670 }, { "epoch": 0.33, "learning_rate": 3.318846305807981e-05, "loss": 0.663, "step": 1680 }, { "epoch": 0.33, "learning_rate": 3.338601343342552e-05, "loss": 0.7416, "step": 1690 }, { "epoch": 0.34, "learning_rate": 3.358356380877124e-05, "loss": 0.689, "step": 1700 }, { "epoch": 0.34, "learning_rate": 3.378111418411695e-05, "loss": 0.6992, "step": 1710 }, { "epoch": 0.34, "learning_rate": 3.397866455946266e-05, "loss": 0.6602, "step": 1720 }, { "epoch": 0.34, "learning_rate": 3.4176214934808374e-05, "loss": 0.6961, "step": 1730 }, { "epoch": 0.34, "learning_rate": 3.4373765310154094e-05, "loss": 0.7106, "step": 1740 }, { "epoch": 0.35, "learning_rate": 3.457131568549981e-05, "loss": 0.6957, "step": 1750 }, { "epoch": 0.35, "learning_rate": 3.476886606084552e-05, "loss": 0.6981, "step": 1760 }, { "epoch": 0.35, "learning_rate": 3.496641643619123e-05, "loss": 0.7585, "step": 1770 }, { "epoch": 0.35, "learning_rate": 3.516396681153694e-05, "loss": 0.6903, "step": 1780 }, { "epoch": 0.35, "learning_rate": 3.536151718688266e-05, "loss": 0.6592, "step": 1790 }, { "epoch": 0.36, "learning_rate": 3.555906756222837e-05, "loss": 0.732, "step": 1800 }, { "epoch": 0.36, "learning_rate": 3.5756617937574084e-05, "loss": 0.703, "step": 1810 }, { "epoch": 0.36, "learning_rate": 3.5954168312919796e-05, "loss": 0.7397, "step": 1820 }, { "epoch": 0.36, "learning_rate": 3.615171868826551e-05, "loss": 0.6665, "step": 1830 }, { "epoch": 0.36, "learning_rate": 3.634926906361122e-05, "loss": 0.7053, "step": 1840 }, { "epoch": 0.37, "learning_rate": 3.6546819438956935e-05, "loss": 0.7008, "step": 1850 }, { "epoch": 0.37, "learning_rate": 3.674436981430265e-05, "loss": 0.7181, "step": 1860 }, { "epoch": 0.37, "learning_rate": 3.694192018964836e-05, "loss": 0.7223, "step": 1870 }, { "epoch": 0.37, "learning_rate": 3.713947056499408e-05, "loss": 0.7027, "step": 1880 }, { "epoch": 0.37, "learning_rate": 3.733702094033979e-05, "loss": 0.6727, "step": 1890 }, { "epoch": 0.38, "learning_rate": 3.75345713156855e-05, "loss": 0.6756, "step": 1900 }, { "epoch": 0.38, "learning_rate": 3.773212169103121e-05, "loss": 0.7386, "step": 1910 }, { "epoch": 0.38, "learning_rate": 3.7929672066376924e-05, "loss": 0.6824, "step": 1920 }, { "epoch": 0.38, "learning_rate": 3.8127222441722644e-05, "loss": 0.7292, "step": 1930 }, { "epoch": 0.38, "learning_rate": 3.8324772817068357e-05, "loss": 0.716, "step": 1940 }, { "epoch": 0.39, "learning_rate": 3.852232319241407e-05, "loss": 0.657, "step": 1950 }, { "epoch": 0.39, "learning_rate": 3.871987356775978e-05, "loss": 0.6604, "step": 1960 }, { "epoch": 0.39, "learning_rate": 3.891742394310549e-05, "loss": 0.679, "step": 1970 }, { "epoch": 0.39, "learning_rate": 3.911497431845121e-05, "loss": 0.6932, "step": 1980 }, { "epoch": 0.39, "learning_rate": 3.931252469379692e-05, "loss": 0.7374, "step": 1990 }, { "epoch": 0.4, "learning_rate": 3.951007506914263e-05, "loss": 0.7186, "step": 2000 }, { "epoch": 0.4, "learning_rate": 3.9707625444488346e-05, "loss": 0.7086, "step": 2010 }, { "epoch": 0.4, "learning_rate": 3.990517581983406e-05, "loss": 0.6731, "step": 2020 }, { "epoch": 0.4, "learning_rate": 4.010272619517977e-05, "loss": 0.69, "step": 2030 }, { "epoch": 0.4, "learning_rate": 4.0300276570525485e-05, "loss": 0.6567, "step": 2040 }, { "epoch": 0.4, "learning_rate": 4.04978269458712e-05, "loss": 0.6974, "step": 2050 }, { "epoch": 0.41, "learning_rate": 4.069537732121691e-05, "loss": 0.7603, "step": 2060 }, { "epoch": 0.41, "learning_rate": 4.089292769656263e-05, "loss": 0.7012, "step": 2070 }, { "epoch": 0.41, "learning_rate": 4.109047807190834e-05, "loss": 0.7585, "step": 2080 }, { "epoch": 0.41, "learning_rate": 4.128802844725405e-05, "loss": 0.7086, "step": 2090 }, { "epoch": 0.41, "learning_rate": 4.148557882259976e-05, "loss": 0.7345, "step": 2100 }, { "epoch": 0.42, "learning_rate": 4.168312919794548e-05, "loss": 0.6888, "step": 2110 }, { "epoch": 0.42, "learning_rate": 4.1880679573291194e-05, "loss": 0.6637, "step": 2120 }, { "epoch": 0.42, "learning_rate": 4.2078229948636906e-05, "loss": 0.6936, "step": 2130 }, { "epoch": 0.42, "learning_rate": 4.227578032398262e-05, "loss": 0.6844, "step": 2140 }, { "epoch": 0.42, "learning_rate": 4.2473330699328325e-05, "loss": 0.709, "step": 2150 }, { "epoch": 0.43, "learning_rate": 4.2670881074674045e-05, "loss": 0.6244, "step": 2160 }, { "epoch": 0.43, "learning_rate": 4.286843145001976e-05, "loss": 0.6936, "step": 2170 }, { "epoch": 0.43, "learning_rate": 4.306598182536547e-05, "loss": 0.7073, "step": 2180 }, { "epoch": 0.43, "learning_rate": 4.326353220071118e-05, "loss": 0.7202, "step": 2190 }, { "epoch": 0.43, "learning_rate": 4.3461082576056896e-05, "loss": 0.6506, "step": 2200 }, { "epoch": 0.44, "learning_rate": 4.365863295140261e-05, "loss": 0.6339, "step": 2210 }, { "epoch": 0.44, "learning_rate": 4.385618332674832e-05, "loss": 0.6635, "step": 2220 }, { "epoch": 0.44, "learning_rate": 4.4053733702094034e-05, "loss": 0.6674, "step": 2230 }, { "epoch": 0.44, "learning_rate": 4.425128407743975e-05, "loss": 0.6857, "step": 2240 }, { "epoch": 0.44, "learning_rate": 4.444883445278547e-05, "loss": 0.7223, "step": 2250 }, { "epoch": 0.45, "learning_rate": 4.464638482813118e-05, "loss": 0.6716, "step": 2260 }, { "epoch": 0.45, "learning_rate": 4.4843935203476886e-05, "loss": 0.7094, "step": 2270 }, { "epoch": 0.45, "learning_rate": 4.50414855788226e-05, "loss": 0.6477, "step": 2280 }, { "epoch": 0.45, "learning_rate": 4.523903595416831e-05, "loss": 0.6409, "step": 2290 }, { "epoch": 0.45, "learning_rate": 4.543658632951403e-05, "loss": 0.6682, "step": 2300 }, { "epoch": 0.46, "learning_rate": 4.5634136704859744e-05, "loss": 0.6727, "step": 2310 }, { "epoch": 0.46, "learning_rate": 4.5831687080205456e-05, "loss": 0.6348, "step": 2320 }, { "epoch": 0.46, "learning_rate": 4.602923745555117e-05, "loss": 0.683, "step": 2330 }, { "epoch": 0.46, "learning_rate": 4.6226787830896875e-05, "loss": 0.5872, "step": 2340 }, { "epoch": 0.46, "learning_rate": 4.6424338206242595e-05, "loss": 0.7156, "step": 2350 }, { "epoch": 0.47, "learning_rate": 4.662188858158831e-05, "loss": 0.6441, "step": 2360 }, { "epoch": 0.47, "learning_rate": 4.681943895693402e-05, "loss": 0.647, "step": 2370 }, { "epoch": 0.47, "learning_rate": 4.701698933227973e-05, "loss": 0.7303, "step": 2380 }, { "epoch": 0.47, "learning_rate": 4.7214539707625446e-05, "loss": 0.6304, "step": 2390 }, { "epoch": 0.47, "learning_rate": 4.741209008297116e-05, "loss": 0.6577, "step": 2400 }, { "epoch": 0.48, "learning_rate": 4.760964045831687e-05, "loss": 0.6941, "step": 2410 }, { "epoch": 0.48, "learning_rate": 4.7807190833662584e-05, "loss": 0.6309, "step": 2420 }, { "epoch": 0.48, "learning_rate": 4.80047412090083e-05, "loss": 0.6584, "step": 2430 }, { "epoch": 0.48, "learning_rate": 4.820229158435402e-05, "loss": 0.6387, "step": 2440 }, { "epoch": 0.48, "learning_rate": 4.839984195969973e-05, "loss": 0.6602, "step": 2450 }, { "epoch": 0.49, "learning_rate": 4.8597392335045435e-05, "loss": 0.7043, "step": 2460 }, { "epoch": 0.49, "learning_rate": 4.879494271039115e-05, "loss": 0.711, "step": 2470 }, { "epoch": 0.49, "learning_rate": 4.899249308573686e-05, "loss": 0.6393, "step": 2480 }, { "epoch": 0.49, "learning_rate": 4.919004346108258e-05, "loss": 0.7015, "step": 2490 }, { "epoch": 0.49, "learning_rate": 4.9387593836428293e-05, "loss": 0.6153, "step": 2500 }, { "epoch": 0.5, "learning_rate": 4.9585144211774006e-05, "loss": 0.6787, "step": 2510 }, { "epoch": 0.5, "learning_rate": 4.978269458711972e-05, "loss": 0.6316, "step": 2520 }, { "epoch": 0.5, "learning_rate": 4.998024496246543e-05, "loss": 0.6494, "step": 2530 }, { "epoch": 0.5, "learning_rate": 5.017779533781114e-05, "loss": 0.6925, "step": 2540 }, { "epoch": 0.5, "learning_rate": 5.0375345713156864e-05, "loss": 0.7249, "step": 2550 }, { "epoch": 0.51, "learning_rate": 5.057289608850258e-05, "loss": 0.6799, "step": 2560 }, { "epoch": 0.51, "learning_rate": 5.077044646384829e-05, "loss": 0.7094, "step": 2570 }, { "epoch": 0.51, "learning_rate": 5.0967996839193996e-05, "loss": 0.7007, "step": 2580 }, { "epoch": 0.51, "learning_rate": 5.116554721453971e-05, "loss": 0.7456, "step": 2590 }, { "epoch": 0.51, "learning_rate": 5.136309758988542e-05, "loss": 0.6767, "step": 2600 }, { "epoch": 0.52, "learning_rate": 5.1560647965231134e-05, "loss": 0.7366, "step": 2610 }, { "epoch": 0.52, "learning_rate": 5.175819834057685e-05, "loss": 0.6579, "step": 2620 }, { "epoch": 0.52, "learning_rate": 5.195574871592256e-05, "loss": 0.6985, "step": 2630 }, { "epoch": 0.52, "learning_rate": 5.215329909126827e-05, "loss": 0.7202, "step": 2640 }, { "epoch": 0.52, "learning_rate": 5.235084946661399e-05, "loss": 0.6233, "step": 2650 }, { "epoch": 0.53, "learning_rate": 5.2548399841959705e-05, "loss": 0.6271, "step": 2660 }, { "epoch": 0.53, "learning_rate": 5.274595021730542e-05, "loss": 0.697, "step": 2670 }, { "epoch": 0.53, "learning_rate": 5.294350059265113e-05, "loss": 0.6596, "step": 2680 }, { "epoch": 0.53, "learning_rate": 5.314105096799684e-05, "loss": 0.6474, "step": 2690 }, { "epoch": 0.53, "learning_rate": 5.3338601343342556e-05, "loss": 0.6291, "step": 2700 }, { "epoch": 0.54, "learning_rate": 5.353615171868827e-05, "loss": 0.634, "step": 2710 }, { "epoch": 0.54, "learning_rate": 5.3733702094033975e-05, "loss": 0.67, "step": 2720 }, { "epoch": 0.54, "learning_rate": 5.393125246937969e-05, "loss": 0.6784, "step": 2730 }, { "epoch": 0.54, "learning_rate": 5.4128802844725414e-05, "loss": 0.6826, "step": 2740 }, { "epoch": 0.54, "learning_rate": 5.432635322007113e-05, "loss": 0.6485, "step": 2750 }, { "epoch": 0.55, "learning_rate": 5.452390359541684e-05, "loss": 0.6364, "step": 2760 }, { "epoch": 0.55, "learning_rate": 5.4721453970762546e-05, "loss": 0.6196, "step": 2770 }, { "epoch": 0.55, "learning_rate": 5.491900434610826e-05, "loss": 0.6812, "step": 2780 }, { "epoch": 0.55, "learning_rate": 5.511655472145397e-05, "loss": 0.6404, "step": 2790 }, { "epoch": 0.55, "learning_rate": 5.5314105096799684e-05, "loss": 0.6725, "step": 2800 }, { "epoch": 0.56, "learning_rate": 5.55116554721454e-05, "loss": 0.6811, "step": 2810 }, { "epoch": 0.56, "learning_rate": 5.570920584749111e-05, "loss": 0.6282, "step": 2820 }, { "epoch": 0.56, "learning_rate": 5.590675622283683e-05, "loss": 0.6505, "step": 2830 }, { "epoch": 0.56, "learning_rate": 5.610430659818254e-05, "loss": 0.6352, "step": 2840 }, { "epoch": 0.56, "learning_rate": 5.6301856973528255e-05, "loss": 0.6836, "step": 2850 }, { "epoch": 0.56, "learning_rate": 5.649940734887397e-05, "loss": 0.5995, "step": 2860 }, { "epoch": 0.57, "learning_rate": 5.669695772421968e-05, "loss": 0.6773, "step": 2870 }, { "epoch": 0.57, "learning_rate": 5.689450809956539e-05, "loss": 0.6272, "step": 2880 }, { "epoch": 0.57, "learning_rate": 5.7092058474911106e-05, "loss": 0.681, "step": 2890 }, { "epoch": 0.57, "learning_rate": 5.728960885025681e-05, "loss": 0.6307, "step": 2900 }, { "epoch": 0.57, "learning_rate": 5.7487159225602525e-05, "loss": 0.6643, "step": 2910 }, { "epoch": 0.58, "learning_rate": 5.768470960094824e-05, "loss": 0.6184, "step": 2920 }, { "epoch": 0.58, "learning_rate": 5.7882259976293964e-05, "loss": 0.6572, "step": 2930 }, { "epoch": 0.58, "learning_rate": 5.807981035163968e-05, "loss": 0.6898, "step": 2940 }, { "epoch": 0.58, "learning_rate": 5.827736072698538e-05, "loss": 0.6442, "step": 2950 }, { "epoch": 0.58, "learning_rate": 5.8474911102331095e-05, "loss": 0.6782, "step": 2960 }, { "epoch": 0.59, "learning_rate": 5.867246147767681e-05, "loss": 0.6749, "step": 2970 }, { "epoch": 0.59, "learning_rate": 5.887001185302252e-05, "loss": 0.6217, "step": 2980 }, { "epoch": 0.59, "learning_rate": 5.9067562228368234e-05, "loss": 0.6644, "step": 2990 }, { "epoch": 0.59, "learning_rate": 5.926511260371395e-05, "loss": 0.6527, "step": 3000 }, { "epoch": 0.59, "learning_rate": 5.946266297905966e-05, "loss": 0.6961, "step": 3010 }, { "epoch": 0.6, "learning_rate": 5.966021335440538e-05, "loss": 0.6514, "step": 3020 }, { "epoch": 0.6, "learning_rate": 5.985776372975109e-05, "loss": 0.6066, "step": 3030 }, { "epoch": 0.6, "learning_rate": 6.0055314105096805e-05, "loss": 0.673, "step": 3040 }, { "epoch": 0.6, "learning_rate": 6.025286448044252e-05, "loss": 0.6685, "step": 3050 }, { "epoch": 0.6, "learning_rate": 6.045041485578823e-05, "loss": 0.6274, "step": 3060 }, { "epoch": 0.61, "learning_rate": 6.064796523113394e-05, "loss": 0.7091, "step": 3070 }, { "epoch": 0.61, "learning_rate": 6.0845515606479656e-05, "loss": 0.6791, "step": 3080 }, { "epoch": 0.61, "learning_rate": 6.104306598182537e-05, "loss": 0.6348, "step": 3090 }, { "epoch": 0.61, "learning_rate": 6.124061635717107e-05, "loss": 0.6043, "step": 3100 }, { "epoch": 0.61, "learning_rate": 6.14381667325168e-05, "loss": 0.6991, "step": 3110 }, { "epoch": 0.62, "learning_rate": 6.163571710786251e-05, "loss": 0.6613, "step": 3120 }, { "epoch": 0.62, "learning_rate": 6.183326748320822e-05, "loss": 0.6498, "step": 3130 }, { "epoch": 0.62, "learning_rate": 6.203081785855394e-05, "loss": 0.6819, "step": 3140 }, { "epoch": 0.62, "learning_rate": 6.222836823389965e-05, "loss": 0.6316, "step": 3150 }, { "epoch": 0.62, "learning_rate": 6.242591860924536e-05, "loss": 0.6672, "step": 3160 }, { "epoch": 0.63, "learning_rate": 6.262346898459107e-05, "loss": 0.6602, "step": 3170 }, { "epoch": 0.63, "learning_rate": 6.282101935993678e-05, "loss": 0.6463, "step": 3180 }, { "epoch": 0.63, "learning_rate": 6.30185697352825e-05, "loss": 0.614, "step": 3190 }, { "epoch": 0.63, "learning_rate": 6.321612011062822e-05, "loss": 0.6689, "step": 3200 }, { "epoch": 0.63, "learning_rate": 6.341367048597394e-05, "loss": 0.6224, "step": 3210 }, { "epoch": 0.64, "learning_rate": 6.361122086131964e-05, "loss": 0.5965, "step": 3220 }, { "epoch": 0.64, "learning_rate": 6.380877123666536e-05, "loss": 0.6219, "step": 3230 }, { "epoch": 0.64, "learning_rate": 6.400632161201107e-05, "loss": 0.6675, "step": 3240 }, { "epoch": 0.64, "learning_rate": 6.420387198735677e-05, "loss": 0.7074, "step": 3250 }, { "epoch": 0.64, "learning_rate": 6.440142236270249e-05, "loss": 0.6304, "step": 3260 }, { "epoch": 0.65, "learning_rate": 6.45989727380482e-05, "loss": 0.6729, "step": 3270 }, { "epoch": 0.65, "learning_rate": 6.479652311339392e-05, "loss": 0.6177, "step": 3280 }, { "epoch": 0.65, "learning_rate": 6.499407348873962e-05, "loss": 0.6709, "step": 3290 }, { "epoch": 0.65, "learning_rate": 6.519162386408534e-05, "loss": 0.7026, "step": 3300 }, { "epoch": 0.65, "learning_rate": 6.538917423943106e-05, "loss": 0.6838, "step": 3310 }, { "epoch": 0.66, "learning_rate": 6.558672461477677e-05, "loss": 0.6294, "step": 3320 }, { "epoch": 0.66, "learning_rate": 6.578427499012249e-05, "loss": 0.6447, "step": 3330 }, { "epoch": 0.66, "learning_rate": 6.59818253654682e-05, "loss": 0.592, "step": 3340 }, { "epoch": 0.66, "learning_rate": 6.617937574081391e-05, "loss": 0.6698, "step": 3350 }, { "epoch": 0.66, "learning_rate": 6.637692611615962e-05, "loss": 0.6528, "step": 3360 }, { "epoch": 0.67, "learning_rate": 6.657447649150533e-05, "loss": 0.6679, "step": 3370 }, { "epoch": 0.67, "learning_rate": 6.677202686685105e-05, "loss": 0.6379, "step": 3380 }, { "epoch": 0.67, "learning_rate": 6.696957724219677e-05, "loss": 0.6213, "step": 3390 }, { "epoch": 0.67, "learning_rate": 6.716712761754249e-05, "loss": 0.6696, "step": 3400 }, { "epoch": 0.67, "learning_rate": 6.736467799288819e-05, "loss": 0.6174, "step": 3410 }, { "epoch": 0.68, "learning_rate": 6.75622283682339e-05, "loss": 0.6211, "step": 3420 }, { "epoch": 0.68, "learning_rate": 6.775977874357962e-05, "loss": 0.6016, "step": 3430 }, { "epoch": 0.68, "learning_rate": 6.795732911892532e-05, "loss": 0.6407, "step": 3440 }, { "epoch": 0.68, "learning_rate": 6.815487949427104e-05, "loss": 0.6349, "step": 3450 }, { "epoch": 0.68, "learning_rate": 6.835242986961675e-05, "loss": 0.6478, "step": 3460 }, { "epoch": 0.69, "learning_rate": 6.854998024496247e-05, "loss": 0.6733, "step": 3470 }, { "epoch": 0.69, "learning_rate": 6.874753062030819e-05, "loss": 0.6037, "step": 3480 }, { "epoch": 0.69, "learning_rate": 6.89450809956539e-05, "loss": 0.6621, "step": 3490 }, { "epoch": 0.69, "learning_rate": 6.914263137099961e-05, "loss": 0.6122, "step": 3500 }, { "epoch": 0.69, "learning_rate": 6.934018174634532e-05, "loss": 0.594, "step": 3510 }, { "epoch": 0.7, "learning_rate": 6.953773212169104e-05, "loss": 0.6445, "step": 3520 }, { "epoch": 0.7, "learning_rate": 6.973528249703675e-05, "loss": 0.6363, "step": 3530 }, { "epoch": 0.7, "learning_rate": 6.993283287238246e-05, "loss": 0.635, "step": 3540 }, { "epoch": 0.7, "learning_rate": 7.013038324772817e-05, "loss": 0.6506, "step": 3550 }, { "epoch": 0.7, "learning_rate": 7.032793362307388e-05, "loss": 0.6546, "step": 3560 }, { "epoch": 0.71, "learning_rate": 7.052548399841961e-05, "loss": 0.6486, "step": 3570 }, { "epoch": 0.71, "learning_rate": 7.072303437376532e-05, "loss": 0.6543, "step": 3580 }, { "epoch": 0.71, "learning_rate": 7.092058474911104e-05, "loss": 0.6799, "step": 3590 }, { "epoch": 0.71, "learning_rate": 7.111813512445674e-05, "loss": 0.665, "step": 3600 }, { "epoch": 0.71, "learning_rate": 7.131568549980245e-05, "loss": 0.6728, "step": 3610 }, { "epoch": 0.72, "learning_rate": 7.151323587514817e-05, "loss": 0.6574, "step": 3620 }, { "epoch": 0.72, "learning_rate": 7.171078625049387e-05, "loss": 0.6618, "step": 3630 }, { "epoch": 0.72, "learning_rate": 7.190833662583959e-05, "loss": 0.5915, "step": 3640 }, { "epoch": 0.72, "learning_rate": 7.21058870011853e-05, "loss": 0.5632, "step": 3650 }, { "epoch": 0.72, "learning_rate": 7.230343737653102e-05, "loss": 0.6802, "step": 3660 }, { "epoch": 0.72, "learning_rate": 7.250098775187674e-05, "loss": 0.6749, "step": 3670 }, { "epoch": 0.73, "learning_rate": 7.269853812722244e-05, "loss": 0.665, "step": 3680 }, { "epoch": 0.73, "learning_rate": 7.289608850256816e-05, "loss": 0.7057, "step": 3690 }, { "epoch": 0.73, "learning_rate": 7.309363887791387e-05, "loss": 0.6284, "step": 3700 }, { "epoch": 0.73, "learning_rate": 7.329118925325959e-05, "loss": 0.6445, "step": 3710 }, { "epoch": 0.73, "learning_rate": 7.34887396286053e-05, "loss": 0.6352, "step": 3720 }, { "epoch": 0.74, "learning_rate": 7.368629000395101e-05, "loss": 0.6989, "step": 3730 }, { "epoch": 0.74, "learning_rate": 7.388384037929672e-05, "loss": 0.6579, "step": 3740 }, { "epoch": 0.74, "learning_rate": 7.408139075464243e-05, "loss": 0.6215, "step": 3750 }, { "epoch": 0.74, "learning_rate": 7.427894112998816e-05, "loss": 0.5917, "step": 3760 }, { "epoch": 0.74, "learning_rate": 7.447649150533387e-05, "loss": 0.6685, "step": 3770 }, { "epoch": 0.75, "learning_rate": 7.467404188067959e-05, "loss": 0.6314, "step": 3780 }, { "epoch": 0.75, "learning_rate": 7.487159225602529e-05, "loss": 0.6184, "step": 3790 }, { "epoch": 0.75, "learning_rate": 7.5069142631371e-05, "loss": 0.6256, "step": 3800 }, { "epoch": 0.75, "learning_rate": 7.526669300671672e-05, "loss": 0.6055, "step": 3810 }, { "epoch": 0.75, "learning_rate": 7.546424338206242e-05, "loss": 0.6527, "step": 3820 }, { "epoch": 0.76, "learning_rate": 7.566179375740814e-05, "loss": 0.6632, "step": 3830 }, { "epoch": 0.76, "learning_rate": 7.585934413275385e-05, "loss": 0.5831, "step": 3840 }, { "epoch": 0.76, "learning_rate": 7.605689450809957e-05, "loss": 0.6703, "step": 3850 }, { "epoch": 0.76, "learning_rate": 7.625444488344529e-05, "loss": 0.6218, "step": 3860 }, { "epoch": 0.76, "learning_rate": 7.6451995258791e-05, "loss": 0.6795, "step": 3870 }, { "epoch": 0.77, "learning_rate": 7.664954563413671e-05, "loss": 0.6779, "step": 3880 }, { "epoch": 0.77, "learning_rate": 7.684709600948242e-05, "loss": 0.6174, "step": 3890 }, { "epoch": 0.77, "learning_rate": 7.704464638482814e-05, "loss": 0.6547, "step": 3900 }, { "epoch": 0.77, "learning_rate": 7.724219676017384e-05, "loss": 0.6003, "step": 3910 }, { "epoch": 0.77, "learning_rate": 7.743974713551956e-05, "loss": 0.6768, "step": 3920 }, { "epoch": 0.78, "learning_rate": 7.763729751086527e-05, "loss": 0.6117, "step": 3930 }, { "epoch": 0.78, "learning_rate": 7.783484788621098e-05, "loss": 0.618, "step": 3940 }, { "epoch": 0.78, "learning_rate": 7.803239826155671e-05, "loss": 0.6749, "step": 3950 }, { "epoch": 0.78, "learning_rate": 7.822994863690242e-05, "loss": 0.6358, "step": 3960 }, { "epoch": 0.78, "learning_rate": 7.842749901224814e-05, "loss": 0.6954, "step": 3970 }, { "epoch": 0.79, "learning_rate": 7.862504938759384e-05, "loss": 0.5934, "step": 3980 }, { "epoch": 0.79, "learning_rate": 7.882259976293955e-05, "loss": 0.6526, "step": 3990 }, { "epoch": 0.79, "learning_rate": 7.902015013828527e-05, "loss": 0.6171, "step": 4000 }, { "epoch": 0.79, "learning_rate": 7.921770051363097e-05, "loss": 0.6219, "step": 4010 }, { "epoch": 0.79, "learning_rate": 7.941525088897669e-05, "loss": 0.6421, "step": 4020 }, { "epoch": 0.8, "learning_rate": 7.96128012643224e-05, "loss": 0.6531, "step": 4030 }, { "epoch": 0.8, "learning_rate": 7.981035163966812e-05, "loss": 0.5801, "step": 4040 }, { "epoch": 0.8, "learning_rate": 8.000790201501384e-05, "loss": 0.63, "step": 4050 }, { "epoch": 0.8, "learning_rate": 8.020545239035954e-05, "loss": 0.6303, "step": 4060 }, { "epoch": 0.8, "learning_rate": 8.040300276570526e-05, "loss": 0.612, "step": 4070 }, { "epoch": 0.81, "learning_rate": 8.060055314105097e-05, "loss": 0.6031, "step": 4080 }, { "epoch": 0.81, "learning_rate": 8.079810351639669e-05, "loss": 0.6569, "step": 4090 }, { "epoch": 0.81, "learning_rate": 8.09956538917424e-05, "loss": 0.7017, "step": 4100 }, { "epoch": 0.81, "learning_rate": 8.119320426708811e-05, "loss": 0.6456, "step": 4110 }, { "epoch": 0.81, "learning_rate": 8.139075464243382e-05, "loss": 0.6752, "step": 4120 }, { "epoch": 0.82, "learning_rate": 8.158830501777954e-05, "loss": 0.5988, "step": 4130 }, { "epoch": 0.82, "learning_rate": 8.178585539312526e-05, "loss": 0.6224, "step": 4140 }, { "epoch": 0.82, "learning_rate": 8.198340576847097e-05, "loss": 0.6238, "step": 4150 }, { "epoch": 0.82, "learning_rate": 8.218095614381669e-05, "loss": 0.6953, "step": 4160 }, { "epoch": 0.82, "learning_rate": 8.237850651916239e-05, "loss": 0.675, "step": 4170 }, { "epoch": 0.83, "learning_rate": 8.25760568945081e-05, "loss": 0.6091, "step": 4180 }, { "epoch": 0.83, "learning_rate": 8.277360726985382e-05, "loss": 0.6527, "step": 4190 }, { "epoch": 0.83, "learning_rate": 8.297115764519952e-05, "loss": 0.6183, "step": 4200 }, { "epoch": 0.83, "learning_rate": 8.316870802054524e-05, "loss": 0.6101, "step": 4210 }, { "epoch": 0.83, "learning_rate": 8.336625839589096e-05, "loss": 0.6639, "step": 4220 }, { "epoch": 0.84, "learning_rate": 8.356380877123667e-05, "loss": 0.6226, "step": 4230 }, { "epoch": 0.84, "learning_rate": 8.376135914658239e-05, "loss": 0.6441, "step": 4240 }, { "epoch": 0.84, "learning_rate": 8.39589095219281e-05, "loss": 0.5824, "step": 4250 }, { "epoch": 0.84, "learning_rate": 8.415645989727381e-05, "loss": 0.6669, "step": 4260 }, { "epoch": 0.84, "learning_rate": 8.435401027261952e-05, "loss": 0.6821, "step": 4270 }, { "epoch": 0.85, "learning_rate": 8.455156064796524e-05, "loss": 0.6243, "step": 4280 }, { "epoch": 0.85, "learning_rate": 8.474911102331094e-05, "loss": 0.5964, "step": 4290 }, { "epoch": 0.85, "learning_rate": 8.494666139865665e-05, "loss": 0.5887, "step": 4300 }, { "epoch": 0.85, "learning_rate": 8.514421177400237e-05, "loss": 0.6496, "step": 4310 }, { "epoch": 0.85, "learning_rate": 8.534176214934809e-05, "loss": 0.6563, "step": 4320 }, { "epoch": 0.86, "learning_rate": 8.553931252469381e-05, "loss": 0.6282, "step": 4330 }, { "epoch": 0.86, "learning_rate": 8.573686290003952e-05, "loss": 0.6189, "step": 4340 }, { "epoch": 0.86, "learning_rate": 8.593441327538523e-05, "loss": 0.6376, "step": 4350 }, { "epoch": 0.86, "learning_rate": 8.613196365073094e-05, "loss": 0.5814, "step": 4360 }, { "epoch": 0.86, "learning_rate": 8.632951402607665e-05, "loss": 0.5461, "step": 4370 }, { "epoch": 0.87, "learning_rate": 8.652706440142237e-05, "loss": 0.6651, "step": 4380 }, { "epoch": 0.87, "learning_rate": 8.672461477676807e-05, "loss": 0.6614, "step": 4390 }, { "epoch": 0.87, "learning_rate": 8.692216515211379e-05, "loss": 0.604, "step": 4400 }, { "epoch": 0.87, "learning_rate": 8.711971552745951e-05, "loss": 0.6378, "step": 4410 }, { "epoch": 0.87, "learning_rate": 8.731726590280522e-05, "loss": 0.6211, "step": 4420 }, { "epoch": 0.88, "learning_rate": 8.751481627815094e-05, "loss": 0.5943, "step": 4430 }, { "epoch": 0.88, "learning_rate": 8.771236665349664e-05, "loss": 0.6493, "step": 4440 }, { "epoch": 0.88, "learning_rate": 8.790991702884236e-05, "loss": 0.643, "step": 4450 }, { "epoch": 0.88, "learning_rate": 8.810746740418807e-05, "loss": 0.6311, "step": 4460 }, { "epoch": 0.88, "learning_rate": 8.830501777953379e-05, "loss": 0.6606, "step": 4470 }, { "epoch": 0.88, "learning_rate": 8.85025681548795e-05, "loss": 0.615, "step": 4480 }, { "epoch": 0.89, "learning_rate": 8.87001185302252e-05, "loss": 0.6847, "step": 4490 }, { "epoch": 0.89, "learning_rate": 8.889766890557093e-05, "loss": 0.6207, "step": 4500 }, { "epoch": 0.89, "learning_rate": 8.909521928091664e-05, "loss": 0.6094, "step": 4510 }, { "epoch": 0.89, "learning_rate": 8.929276965626236e-05, "loss": 0.6294, "step": 4520 }, { "epoch": 0.89, "learning_rate": 8.949032003160807e-05, "loss": 0.644, "step": 4530 }, { "epoch": 0.9, "learning_rate": 8.968787040695377e-05, "loss": 0.6265, "step": 4540 }, { "epoch": 0.9, "learning_rate": 8.988542078229949e-05, "loss": 0.6348, "step": 4550 }, { "epoch": 0.9, "learning_rate": 9.00829711576452e-05, "loss": 0.5913, "step": 4560 }, { "epoch": 0.9, "learning_rate": 9.028052153299092e-05, "loss": 0.6377, "step": 4570 }, { "epoch": 0.9, "learning_rate": 9.047807190833662e-05, "loss": 0.5929, "step": 4580 }, { "epoch": 0.91, "learning_rate": 9.067562228368236e-05, "loss": 0.6708, "step": 4590 }, { "epoch": 0.91, "learning_rate": 9.087317265902806e-05, "loss": 0.6314, "step": 4600 }, { "epoch": 0.91, "learning_rate": 9.107072303437377e-05, "loss": 0.6853, "step": 4610 }, { "epoch": 0.91, "learning_rate": 9.126827340971949e-05, "loss": 0.6196, "step": 4620 }, { "epoch": 0.91, "learning_rate": 9.146582378506519e-05, "loss": 0.5957, "step": 4630 }, { "epoch": 0.92, "learning_rate": 9.166337416041091e-05, "loss": 0.5991, "step": 4640 }, { "epoch": 0.92, "learning_rate": 9.186092453575662e-05, "loss": 0.6602, "step": 4650 }, { "epoch": 0.92, "learning_rate": 9.205847491110234e-05, "loss": 0.6852, "step": 4660 }, { "epoch": 0.92, "learning_rate": 9.225602528644804e-05, "loss": 0.5723, "step": 4670 }, { "epoch": 0.92, "learning_rate": 9.245357566179375e-05, "loss": 0.6218, "step": 4680 }, { "epoch": 0.93, "learning_rate": 9.265112603713948e-05, "loss": 0.625, "step": 4690 }, { "epoch": 0.93, "learning_rate": 9.284867641248519e-05, "loss": 0.6035, "step": 4700 }, { "epoch": 0.93, "learning_rate": 9.304622678783091e-05, "loss": 0.6119, "step": 4710 }, { "epoch": 0.93, "learning_rate": 9.324377716317662e-05, "loss": 0.6284, "step": 4720 }, { "epoch": 0.93, "learning_rate": 9.344132753852232e-05, "loss": 0.5956, "step": 4730 }, { "epoch": 0.94, "learning_rate": 9.363887791386804e-05, "loss": 0.5773, "step": 4740 }, { "epoch": 0.94, "learning_rate": 9.383642828921375e-05, "loss": 0.6735, "step": 4750 }, { "epoch": 0.94, "learning_rate": 9.403397866455947e-05, "loss": 0.6019, "step": 4760 }, { "epoch": 0.94, "learning_rate": 9.423152903990517e-05, "loss": 0.6271, "step": 4770 }, { "epoch": 0.94, "learning_rate": 9.442907941525089e-05, "loss": 0.6201, "step": 4780 }, { "epoch": 0.95, "learning_rate": 9.462662979059661e-05, "loss": 0.6904, "step": 4790 }, { "epoch": 0.95, "learning_rate": 9.482418016594232e-05, "loss": 0.6405, "step": 4800 }, { "epoch": 0.95, "learning_rate": 9.502173054128804e-05, "loss": 0.5189, "step": 4810 }, { "epoch": 0.95, "learning_rate": 9.521928091663374e-05, "loss": 0.623, "step": 4820 }, { "epoch": 0.95, "learning_rate": 9.541683129197946e-05, "loss": 0.6278, "step": 4830 }, { "epoch": 0.96, "learning_rate": 9.561438166732517e-05, "loss": 0.6264, "step": 4840 }, { "epoch": 0.96, "learning_rate": 9.581193204267089e-05, "loss": 0.6407, "step": 4850 }, { "epoch": 0.96, "learning_rate": 9.60094824180166e-05, "loss": 0.6766, "step": 4860 }, { "epoch": 0.96, "learning_rate": 9.620703279336231e-05, "loss": 0.6492, "step": 4870 }, { "epoch": 0.96, "learning_rate": 9.640458316870803e-05, "loss": 0.6087, "step": 4880 }, { "epoch": 0.97, "learning_rate": 9.660213354405374e-05, "loss": 0.5838, "step": 4890 }, { "epoch": 0.97, "learning_rate": 9.679968391939946e-05, "loss": 0.6389, "step": 4900 }, { "epoch": 0.97, "learning_rate": 9.699723429474516e-05, "loss": 0.664, "step": 4910 }, { "epoch": 0.97, "learning_rate": 9.719478467009087e-05, "loss": 0.6164, "step": 4920 }, { "epoch": 0.97, "learning_rate": 9.739233504543659e-05, "loss": 0.6112, "step": 4930 }, { "epoch": 0.98, "learning_rate": 9.75898854207823e-05, "loss": 0.5928, "step": 4940 }, { "epoch": 0.98, "learning_rate": 9.778743579612802e-05, "loss": 0.6131, "step": 4950 }, { "epoch": 0.98, "learning_rate": 9.798498617147372e-05, "loss": 0.6221, "step": 4960 }, { "epoch": 0.98, "learning_rate": 9.818253654681944e-05, "loss": 0.6302, "step": 4970 }, { "epoch": 0.98, "learning_rate": 9.838008692216516e-05, "loss": 0.6185, "step": 4980 }, { "epoch": 0.99, "learning_rate": 9.857763729751087e-05, "loss": 0.6397, "step": 4990 }, { "epoch": 0.99, "learning_rate": 9.877518767285659e-05, "loss": 0.6381, "step": 5000 }, { "epoch": 0.99, "learning_rate": 9.897273804820229e-05, "loss": 0.6274, "step": 5010 }, { "epoch": 0.99, "learning_rate": 9.917028842354801e-05, "loss": 0.5895, "step": 5020 }, { "epoch": 0.99, "learning_rate": 9.936783879889372e-05, "loss": 0.6603, "step": 5030 }, { "epoch": 1.0, "learning_rate": 9.956538917423944e-05, "loss": 0.5812, "step": 5040 }, { "epoch": 1.0, "learning_rate": 9.976293954958514e-05, "loss": 0.6215, "step": 5050 }, { "epoch": 1.0, "learning_rate": 9.996048992493086e-05, "loss": 0.5478, "step": 5060 }, { "epoch": 1.0, "learning_rate": 9.996048992493086e-05, "loss": 0.5893, "step": 5070 }, { "epoch": 1.0, "learning_rate": 9.991110233109443e-05, "loss": 0.6172, "step": 5080 }, { "epoch": 1.01, "learning_rate": 9.986171473725801e-05, "loss": 0.609, "step": 5090 }, { "epoch": 1.01, "learning_rate": 9.981232714342158e-05, "loss": 0.5751, "step": 5100 }, { "epoch": 1.01, "learning_rate": 9.976293954958514e-05, "loss": 0.6062, "step": 5110 }, { "epoch": 1.01, "learning_rate": 9.971355195574872e-05, "loss": 0.5886, "step": 5120 }, { "epoch": 1.01, "learning_rate": 9.966416436191229e-05, "loss": 0.5286, "step": 5130 }, { "epoch": 1.02, "learning_rate": 9.961477676807587e-05, "loss": 0.611, "step": 5140 }, { "epoch": 1.02, "learning_rate": 9.956538917423944e-05, "loss": 0.5796, "step": 5150 }, { "epoch": 1.02, "learning_rate": 9.9516001580403e-05, "loss": 0.5771, "step": 5160 }, { "epoch": 1.02, "learning_rate": 9.946661398656658e-05, "loss": 0.5545, "step": 5170 }, { "epoch": 1.02, "learning_rate": 9.941722639273015e-05, "loss": 0.5793, "step": 5180 }, { "epoch": 1.03, "learning_rate": 9.936783879889372e-05, "loss": 0.6067, "step": 5190 }, { "epoch": 1.03, "learning_rate": 9.93184512050573e-05, "loss": 0.5685, "step": 5200 }, { "epoch": 1.03, "learning_rate": 9.926906361122087e-05, "loss": 0.5841, "step": 5210 }, { "epoch": 1.03, "learning_rate": 9.921967601738443e-05, "loss": 0.5445, "step": 5220 }, { "epoch": 1.03, "learning_rate": 9.917028842354801e-05, "loss": 0.5709, "step": 5230 }, { "epoch": 1.04, "learning_rate": 9.912090082971158e-05, "loss": 0.607, "step": 5240 }, { "epoch": 1.04, "learning_rate": 9.907151323587516e-05, "loss": 0.5702, "step": 5250 }, { "epoch": 1.04, "learning_rate": 9.902212564203873e-05, "loss": 0.5739, "step": 5260 }, { "epoch": 1.04, "learning_rate": 9.897273804820229e-05, "loss": 0.6375, "step": 5270 }, { "epoch": 1.04, "learning_rate": 9.892335045436587e-05, "loss": 0.6164, "step": 5280 }, { "epoch": 1.04, "learning_rate": 9.887396286052944e-05, "loss": 0.6189, "step": 5290 }, { "epoch": 1.05, "learning_rate": 9.8824575266693e-05, "loss": 0.5755, "step": 5300 }, { "epoch": 1.05, "learning_rate": 9.877518767285659e-05, "loss": 0.5973, "step": 5310 }, { "epoch": 1.05, "learning_rate": 9.872580007902015e-05, "loss": 0.6061, "step": 5320 }, { "epoch": 1.05, "learning_rate": 9.867641248518372e-05, "loss": 0.5847, "step": 5330 }, { "epoch": 1.05, "learning_rate": 9.86270248913473e-05, "loss": 0.5368, "step": 5340 }, { "epoch": 1.06, "learning_rate": 9.857763729751087e-05, "loss": 0.5204, "step": 5350 }, { "epoch": 1.06, "learning_rate": 9.852824970367445e-05, "loss": 0.5737, "step": 5360 }, { "epoch": 1.06, "learning_rate": 9.847886210983801e-05, "loss": 0.5824, "step": 5370 }, { "epoch": 1.06, "learning_rate": 9.842947451600158e-05, "loss": 0.5648, "step": 5380 }, { "epoch": 1.06, "learning_rate": 9.838008692216516e-05, "loss": 0.5426, "step": 5390 }, { "epoch": 1.07, "learning_rate": 9.833069932832873e-05, "loss": 0.6269, "step": 5400 }, { "epoch": 1.07, "learning_rate": 9.82813117344923e-05, "loss": 0.5373, "step": 5410 }, { "epoch": 1.07, "learning_rate": 9.823192414065587e-05, "loss": 0.5895, "step": 5420 }, { "epoch": 1.07, "learning_rate": 9.818253654681944e-05, "loss": 0.5642, "step": 5430 }, { "epoch": 1.07, "learning_rate": 9.813314895298301e-05, "loss": 0.5318, "step": 5440 }, { "epoch": 1.08, "learning_rate": 9.808376135914659e-05, "loss": 0.6078, "step": 5450 }, { "epoch": 1.08, "learning_rate": 9.803437376531016e-05, "loss": 0.5631, "step": 5460 }, { "epoch": 1.08, "learning_rate": 9.798498617147372e-05, "loss": 0.6024, "step": 5470 }, { "epoch": 1.08, "learning_rate": 9.79355985776373e-05, "loss": 0.6337, "step": 5480 }, { "epoch": 1.08, "learning_rate": 9.788621098380087e-05, "loss": 0.5616, "step": 5490 }, { "epoch": 1.09, "learning_rate": 9.783682338996445e-05, "loss": 0.5906, "step": 5500 }, { "epoch": 1.09, "learning_rate": 9.778743579612802e-05, "loss": 0.6046, "step": 5510 }, { "epoch": 1.09, "learning_rate": 9.773804820229158e-05, "loss": 0.5795, "step": 5520 }, { "epoch": 1.09, "learning_rate": 9.768866060845516e-05, "loss": 0.6262, "step": 5530 }, { "epoch": 1.09, "learning_rate": 9.763927301461874e-05, "loss": 0.6531, "step": 5540 }, { "epoch": 1.1, "learning_rate": 9.75898854207823e-05, "loss": 0.6423, "step": 5550 }, { "epoch": 1.1, "learning_rate": 9.754049782694588e-05, "loss": 0.6074, "step": 5560 }, { "epoch": 1.1, "learning_rate": 9.749111023310946e-05, "loss": 0.5798, "step": 5570 }, { "epoch": 1.1, "learning_rate": 9.744172263927301e-05, "loss": 0.5806, "step": 5580 }, { "epoch": 1.1, "learning_rate": 9.739233504543659e-05, "loss": 0.5655, "step": 5590 }, { "epoch": 1.11, "learning_rate": 9.734294745160016e-05, "loss": 0.523, "step": 5600 }, { "epoch": 1.11, "learning_rate": 9.729355985776374e-05, "loss": 0.566, "step": 5610 }, { "epoch": 1.11, "learning_rate": 9.72441722639273e-05, "loss": 0.5482, "step": 5620 }, { "epoch": 1.11, "learning_rate": 9.719478467009087e-05, "loss": 0.5899, "step": 5630 }, { "epoch": 1.11, "learning_rate": 9.714539707625445e-05, "loss": 0.5508, "step": 5640 }, { "epoch": 1.12, "learning_rate": 9.709600948241803e-05, "loss": 0.5866, "step": 5650 }, { "epoch": 1.12, "learning_rate": 9.704662188858158e-05, "loss": 0.5878, "step": 5660 }, { "epoch": 1.12, "learning_rate": 9.699723429474516e-05, "loss": 0.5563, "step": 5670 }, { "epoch": 1.12, "learning_rate": 9.694784670090875e-05, "loss": 0.6075, "step": 5680 }, { "epoch": 1.12, "learning_rate": 9.68984591070723e-05, "loss": 0.6032, "step": 5690 }, { "epoch": 1.13, "learning_rate": 9.684907151323588e-05, "loss": 0.5908, "step": 5700 }, { "epoch": 1.13, "learning_rate": 9.679968391939946e-05, "loss": 0.5863, "step": 5710 }, { "epoch": 1.13, "learning_rate": 9.675029632556303e-05, "loss": 0.5858, "step": 5720 }, { "epoch": 1.13, "learning_rate": 9.670090873172659e-05, "loss": 0.5929, "step": 5730 }, { "epoch": 1.13, "learning_rate": 9.665152113789017e-05, "loss": 0.5891, "step": 5740 }, { "epoch": 1.14, "learning_rate": 9.660213354405374e-05, "loss": 0.618, "step": 5750 }, { "epoch": 1.14, "learning_rate": 9.65527459502173e-05, "loss": 0.6104, "step": 5760 }, { "epoch": 1.14, "learning_rate": 9.650335835638089e-05, "loss": 0.588, "step": 5770 }, { "epoch": 1.14, "learning_rate": 9.645397076254445e-05, "loss": 0.6054, "step": 5780 }, { "epoch": 1.14, "learning_rate": 9.640458316870803e-05, "loss": 0.5846, "step": 5790 }, { "epoch": 1.15, "learning_rate": 9.635519557487159e-05, "loss": 0.6161, "step": 5800 }, { "epoch": 1.15, "learning_rate": 9.630580798103517e-05, "loss": 0.5995, "step": 5810 }, { "epoch": 1.15, "learning_rate": 9.625642038719875e-05, "loss": 0.5678, "step": 5820 }, { "epoch": 1.15, "learning_rate": 9.620703279336231e-05, "loss": 0.6255, "step": 5830 }, { "epoch": 1.15, "learning_rate": 9.615764519952588e-05, "loss": 0.5903, "step": 5840 }, { "epoch": 1.16, "learning_rate": 9.610825760568946e-05, "loss": 0.5584, "step": 5850 }, { "epoch": 1.16, "learning_rate": 9.605887001185303e-05, "loss": 0.5297, "step": 5860 }, { "epoch": 1.16, "learning_rate": 9.60094824180166e-05, "loss": 0.5777, "step": 5870 }, { "epoch": 1.16, "learning_rate": 9.596009482418017e-05, "loss": 0.6187, "step": 5880 }, { "epoch": 1.16, "learning_rate": 9.591070723034374e-05, "loss": 0.5935, "step": 5890 }, { "epoch": 1.17, "learning_rate": 9.586131963650732e-05, "loss": 0.5897, "step": 5900 }, { "epoch": 1.17, "learning_rate": 9.581193204267089e-05, "loss": 0.5581, "step": 5910 }, { "epoch": 1.17, "learning_rate": 9.576254444883445e-05, "loss": 0.6178, "step": 5920 }, { "epoch": 1.17, "learning_rate": 9.571315685499804e-05, "loss": 0.5879, "step": 5930 }, { "epoch": 1.17, "learning_rate": 9.56637692611616e-05, "loss": 0.6004, "step": 5940 }, { "epoch": 1.18, "learning_rate": 9.561438166732517e-05, "loss": 0.5726, "step": 5950 }, { "epoch": 1.18, "learning_rate": 9.556499407348875e-05, "loss": 0.5794, "step": 5960 }, { "epoch": 1.18, "learning_rate": 9.551560647965232e-05, "loss": 0.586, "step": 5970 }, { "epoch": 1.18, "learning_rate": 9.546621888581588e-05, "loss": 0.5863, "step": 5980 }, { "epoch": 1.18, "learning_rate": 9.541683129197946e-05, "loss": 0.6062, "step": 5990 }, { "epoch": 1.19, "learning_rate": 9.536744369814303e-05, "loss": 0.6103, "step": 6000 }, { "epoch": 1.19, "learning_rate": 9.531805610430661e-05, "loss": 0.6179, "step": 6010 }, { "epoch": 1.19, "learning_rate": 9.526866851047018e-05, "loss": 0.5564, "step": 6020 }, { "epoch": 1.19, "learning_rate": 9.521928091663374e-05, "loss": 0.5664, "step": 6030 }, { "epoch": 1.19, "learning_rate": 9.516989332279732e-05, "loss": 0.5681, "step": 6040 }, { "epoch": 1.2, "learning_rate": 9.512050572896089e-05, "loss": 0.5789, "step": 6050 }, { "epoch": 1.2, "learning_rate": 9.507111813512446e-05, "loss": 0.59, "step": 6060 }, { "epoch": 1.2, "learning_rate": 9.502173054128804e-05, "loss": 0.5384, "step": 6070 }, { "epoch": 1.2, "learning_rate": 9.49723429474516e-05, "loss": 0.5677, "step": 6080 }, { "epoch": 1.2, "learning_rate": 9.492295535361517e-05, "loss": 0.6119, "step": 6090 }, { "epoch": 1.2, "learning_rate": 9.487356775977875e-05, "loss": 0.5626, "step": 6100 }, { "epoch": 1.21, "learning_rate": 9.482418016594232e-05, "loss": 0.5925, "step": 6110 }, { "epoch": 1.21, "learning_rate": 9.477479257210588e-05, "loss": 0.5993, "step": 6120 }, { "epoch": 1.21, "learning_rate": 9.472540497826946e-05, "loss": 0.5264, "step": 6130 }, { "epoch": 1.21, "learning_rate": 9.467601738443303e-05, "loss": 0.5545, "step": 6140 }, { "epoch": 1.21, "learning_rate": 9.462662979059661e-05, "loss": 0.5822, "step": 6150 }, { "epoch": 1.22, "learning_rate": 9.457724219676018e-05, "loss": 0.5915, "step": 6160 }, { "epoch": 1.22, "learning_rate": 9.452785460292374e-05, "loss": 0.5689, "step": 6170 }, { "epoch": 1.22, "learning_rate": 9.447846700908733e-05, "loss": 0.5721, "step": 6180 }, { "epoch": 1.22, "learning_rate": 9.442907941525089e-05, "loss": 0.5692, "step": 6190 }, { "epoch": 1.22, "learning_rate": 9.437969182141446e-05, "loss": 0.5578, "step": 6200 }, { "epoch": 1.23, "learning_rate": 9.433030422757804e-05, "loss": 0.5678, "step": 6210 }, { "epoch": 1.23, "learning_rate": 9.42809166337416e-05, "loss": 0.5738, "step": 6220 }, { "epoch": 1.23, "learning_rate": 9.423152903990517e-05, "loss": 0.5563, "step": 6230 }, { "epoch": 1.23, "learning_rate": 9.418214144606875e-05, "loss": 0.5703, "step": 6240 }, { "epoch": 1.23, "learning_rate": 9.413275385223232e-05, "loss": 0.5612, "step": 6250 }, { "epoch": 1.24, "learning_rate": 9.40833662583959e-05, "loss": 0.5578, "step": 6260 }, { "epoch": 1.24, "learning_rate": 9.403397866455947e-05, "loss": 0.609, "step": 6270 }, { "epoch": 1.24, "learning_rate": 9.398459107072303e-05, "loss": 0.5784, "step": 6280 }, { "epoch": 1.24, "learning_rate": 9.393520347688661e-05, "loss": 0.5322, "step": 6290 }, { "epoch": 1.24, "learning_rate": 9.38858158830502e-05, "loss": 0.5788, "step": 6300 }, { "epoch": 1.25, "learning_rate": 9.383642828921375e-05, "loss": 0.5519, "step": 6310 }, { "epoch": 1.25, "learning_rate": 9.378704069537733e-05, "loss": 0.5735, "step": 6320 }, { "epoch": 1.25, "learning_rate": 9.373765310154091e-05, "loss": 0.5542, "step": 6330 }, { "epoch": 1.25, "learning_rate": 9.368826550770446e-05, "loss": 0.5588, "step": 6340 }, { "epoch": 1.25, "learning_rate": 9.363887791386804e-05, "loss": 0.6296, "step": 6350 }, { "epoch": 1.26, "learning_rate": 9.358949032003162e-05, "loss": 0.592, "step": 6360 }, { "epoch": 1.26, "learning_rate": 9.354010272619519e-05, "loss": 0.54, "step": 6370 }, { "epoch": 1.26, "learning_rate": 9.349071513235875e-05, "loss": 0.5527, "step": 6380 }, { "epoch": 1.26, "learning_rate": 9.344132753852232e-05, "loss": 0.5999, "step": 6390 }, { "epoch": 1.26, "learning_rate": 9.33919399446859e-05, "loss": 0.5323, "step": 6400 }, { "epoch": 1.27, "learning_rate": 9.334255235084947e-05, "loss": 0.5933, "step": 6410 }, { "epoch": 1.27, "learning_rate": 9.329316475701303e-05, "loss": 0.5564, "step": 6420 }, { "epoch": 1.27, "learning_rate": 9.324377716317662e-05, "loss": 0.6013, "step": 6430 }, { "epoch": 1.27, "learning_rate": 9.31943895693402e-05, "loss": 0.5646, "step": 6440 }, { "epoch": 1.27, "learning_rate": 9.314500197550375e-05, "loss": 0.5512, "step": 6450 }, { "epoch": 1.28, "learning_rate": 9.309561438166733e-05, "loss": 0.5529, "step": 6460 }, { "epoch": 1.28, "learning_rate": 9.304622678783091e-05, "loss": 0.6227, "step": 6470 }, { "epoch": 1.28, "learning_rate": 9.299683919399448e-05, "loss": 0.5761, "step": 6480 }, { "epoch": 1.28, "learning_rate": 9.294745160015804e-05, "loss": 0.5036, "step": 6490 }, { "epoch": 1.28, "learning_rate": 9.289806400632162e-05, "loss": 0.5148, "step": 6500 }, { "epoch": 1.29, "learning_rate": 9.284867641248519e-05, "loss": 0.5503, "step": 6510 }, { "epoch": 1.29, "learning_rate": 9.279928881864876e-05, "loss": 0.5882, "step": 6520 }, { "epoch": 1.29, "learning_rate": 9.274990122481234e-05, "loss": 0.5803, "step": 6530 }, { "epoch": 1.29, "learning_rate": 9.27005136309759e-05, "loss": 0.5741, "step": 6540 }, { "epoch": 1.29, "learning_rate": 9.265112603713948e-05, "loss": 0.6181, "step": 6550 }, { "epoch": 1.3, "learning_rate": 9.260173844330304e-05, "loss": 0.594, "step": 6560 }, { "epoch": 1.3, "learning_rate": 9.255235084946662e-05, "loss": 0.5033, "step": 6570 }, { "epoch": 1.3, "learning_rate": 9.25029632556302e-05, "loss": 0.6282, "step": 6580 }, { "epoch": 1.3, "learning_rate": 9.245357566179375e-05, "loss": 0.5376, "step": 6590 }, { "epoch": 1.3, "learning_rate": 9.240418806795733e-05, "loss": 0.5705, "step": 6600 }, { "epoch": 1.31, "learning_rate": 9.235480047412091e-05, "loss": 0.5539, "step": 6610 }, { "epoch": 1.31, "learning_rate": 9.230541288028448e-05, "loss": 0.5966, "step": 6620 }, { "epoch": 1.31, "learning_rate": 9.225602528644804e-05, "loss": 0.5605, "step": 6630 }, { "epoch": 1.31, "learning_rate": 9.220663769261162e-05, "loss": 0.6484, "step": 6640 }, { "epoch": 1.31, "learning_rate": 9.215725009877519e-05, "loss": 0.5461, "step": 6650 }, { "epoch": 1.32, "learning_rate": 9.210786250493877e-05, "loss": 0.6082, "step": 6660 }, { "epoch": 1.32, "learning_rate": 9.205847491110234e-05, "loss": 0.5517, "step": 6670 }, { "epoch": 1.32, "learning_rate": 9.20090873172659e-05, "loss": 0.5512, "step": 6680 }, { "epoch": 1.32, "learning_rate": 9.195969972342949e-05, "loss": 0.4835, "step": 6690 }, { "epoch": 1.32, "learning_rate": 9.191031212959305e-05, "loss": 0.5864, "step": 6700 }, { "epoch": 1.33, "learning_rate": 9.186092453575662e-05, "loss": 0.5513, "step": 6710 }, { "epoch": 1.33, "learning_rate": 9.18115369419202e-05, "loss": 0.5527, "step": 6720 }, { "epoch": 1.33, "learning_rate": 9.176214934808377e-05, "loss": 0.5251, "step": 6730 }, { "epoch": 1.33, "learning_rate": 9.171276175424733e-05, "loss": 0.5467, "step": 6740 }, { "epoch": 1.33, "learning_rate": 9.166337416041091e-05, "loss": 0.5539, "step": 6750 }, { "epoch": 1.34, "learning_rate": 9.161398656657448e-05, "loss": 0.5269, "step": 6760 }, { "epoch": 1.34, "learning_rate": 9.156459897273805e-05, "loss": 0.5198, "step": 6770 }, { "epoch": 1.34, "learning_rate": 9.151521137890163e-05, "loss": 0.5678, "step": 6780 }, { "epoch": 1.34, "learning_rate": 9.146582378506519e-05, "loss": 0.6333, "step": 6790 }, { "epoch": 1.34, "learning_rate": 9.141643619122877e-05, "loss": 0.5445, "step": 6800 }, { "epoch": 1.35, "learning_rate": 9.136704859739234e-05, "loss": 0.5594, "step": 6810 }, { "epoch": 1.35, "learning_rate": 9.131766100355591e-05, "loss": 0.5557, "step": 6820 }, { "epoch": 1.35, "learning_rate": 9.126827340971949e-05, "loss": 0.5797, "step": 6830 }, { "epoch": 1.35, "learning_rate": 9.121888581588305e-05, "loss": 0.5981, "step": 6840 }, { "epoch": 1.35, "learning_rate": 9.116949822204662e-05, "loss": 0.535, "step": 6850 }, { "epoch": 1.35, "learning_rate": 9.11201106282102e-05, "loss": 0.5275, "step": 6860 }, { "epoch": 1.36, "learning_rate": 9.107072303437377e-05, "loss": 0.5531, "step": 6870 }, { "epoch": 1.36, "learning_rate": 9.102133544053733e-05, "loss": 0.5669, "step": 6880 }, { "epoch": 1.36, "learning_rate": 9.097194784670091e-05, "loss": 0.5635, "step": 6890 }, { "epoch": 1.36, "learning_rate": 9.092256025286448e-05, "loss": 0.5911, "step": 6900 }, { "epoch": 1.36, "learning_rate": 9.087317265902806e-05, "loss": 0.5518, "step": 6910 }, { "epoch": 1.37, "learning_rate": 9.082378506519163e-05, "loss": 0.6044, "step": 6920 }, { "epoch": 1.37, "learning_rate": 9.07743974713552e-05, "loss": 0.5872, "step": 6930 }, { "epoch": 1.37, "learning_rate": 9.072500987751878e-05, "loss": 0.5214, "step": 6940 }, { "epoch": 1.37, "learning_rate": 9.067562228368236e-05, "loss": 0.5623, "step": 6950 }, { "epoch": 1.37, "learning_rate": 9.062623468984591e-05, "loss": 0.5604, "step": 6960 }, { "epoch": 1.38, "learning_rate": 9.057684709600949e-05, "loss": 0.5896, "step": 6970 }, { "epoch": 1.38, "learning_rate": 9.052745950217306e-05, "loss": 0.5493, "step": 6980 }, { "epoch": 1.38, "learning_rate": 9.047807190833662e-05, "loss": 0.4864, "step": 6990 }, { "epoch": 1.38, "learning_rate": 9.04286843145002e-05, "loss": 0.5621, "step": 7000 }, { "epoch": 1.38, "learning_rate": 9.037929672066377e-05, "loss": 0.6065, "step": 7010 }, { "epoch": 1.39, "learning_rate": 9.032990912682735e-05, "loss": 0.577, "step": 7020 }, { "epoch": 1.39, "learning_rate": 9.028052153299092e-05, "loss": 0.5552, "step": 7030 }, { "epoch": 1.39, "learning_rate": 9.023113393915448e-05, "loss": 0.5673, "step": 7040 }, { "epoch": 1.39, "learning_rate": 9.018174634531806e-05, "loss": 0.5608, "step": 7050 }, { "epoch": 1.39, "learning_rate": 9.013235875148163e-05, "loss": 0.5397, "step": 7060 }, { "epoch": 1.4, "learning_rate": 9.00829711576452e-05, "loss": 0.5458, "step": 7070 }, { "epoch": 1.4, "learning_rate": 9.003358356380878e-05, "loss": 0.5561, "step": 7080 }, { "epoch": 1.4, "learning_rate": 8.998419596997236e-05, "loss": 0.519, "step": 7090 }, { "epoch": 1.4, "learning_rate": 8.993480837613591e-05, "loss": 0.5473, "step": 7100 }, { "epoch": 1.4, "learning_rate": 8.988542078229949e-05, "loss": 0.5303, "step": 7110 }, { "epoch": 1.41, "learning_rate": 8.983603318846307e-05, "loss": 0.5713, "step": 7120 }, { "epoch": 1.41, "learning_rate": 8.978664559462664e-05, "loss": 0.5767, "step": 7130 }, { "epoch": 1.41, "learning_rate": 8.97372580007902e-05, "loss": 0.531, "step": 7140 }, { "epoch": 1.41, "learning_rate": 8.968787040695377e-05, "loss": 0.5751, "step": 7150 }, { "epoch": 1.41, "learning_rate": 8.963848281311735e-05, "loss": 0.5907, "step": 7160 }, { "epoch": 1.42, "learning_rate": 8.958909521928092e-05, "loss": 0.5272, "step": 7170 }, { "epoch": 1.42, "learning_rate": 8.953970762544448e-05, "loss": 0.5602, "step": 7180 }, { "epoch": 1.42, "learning_rate": 8.949032003160807e-05, "loss": 0.5272, "step": 7190 }, { "epoch": 1.42, "learning_rate": 8.944093243777165e-05, "loss": 0.5467, "step": 7200 }, { "epoch": 1.42, "learning_rate": 8.93915448439352e-05, "loss": 0.5632, "step": 7210 }, { "epoch": 1.43, "learning_rate": 8.934215725009878e-05, "loss": 0.5478, "step": 7220 }, { "epoch": 1.43, "learning_rate": 8.929276965626236e-05, "loss": 0.5539, "step": 7230 }, { "epoch": 1.43, "learning_rate": 8.924338206242591e-05, "loss": 0.58, "step": 7240 }, { "epoch": 1.43, "learning_rate": 8.919399446858949e-05, "loss": 0.5757, "step": 7250 }, { "epoch": 1.43, "learning_rate": 8.914460687475307e-05, "loss": 0.5549, "step": 7260 }, { "epoch": 1.44, "learning_rate": 8.909521928091664e-05, "loss": 0.55, "step": 7270 }, { "epoch": 1.44, "learning_rate": 8.90458316870802e-05, "loss": 0.5813, "step": 7280 }, { "epoch": 1.44, "learning_rate": 8.899644409324379e-05, "loss": 0.53, "step": 7290 }, { "epoch": 1.44, "learning_rate": 8.894705649940735e-05, "loss": 0.5474, "step": 7300 }, { "epoch": 1.44, "learning_rate": 8.889766890557093e-05, "loss": 0.5329, "step": 7310 }, { "epoch": 1.45, "learning_rate": 8.88482813117345e-05, "loss": 0.5353, "step": 7320 }, { "epoch": 1.45, "learning_rate": 8.879889371789807e-05, "loss": 0.5732, "step": 7330 }, { "epoch": 1.45, "learning_rate": 8.874950612406165e-05, "loss": 0.5627, "step": 7340 }, { "epoch": 1.45, "learning_rate": 8.87001185302252e-05, "loss": 0.5384, "step": 7350 }, { "epoch": 1.45, "learning_rate": 8.865073093638878e-05, "loss": 0.5707, "step": 7360 }, { "epoch": 1.46, "learning_rate": 8.860134334255236e-05, "loss": 0.5267, "step": 7370 }, { "epoch": 1.46, "learning_rate": 8.855195574871593e-05, "loss": 0.5762, "step": 7380 }, { "epoch": 1.46, "learning_rate": 8.85025681548795e-05, "loss": 0.5558, "step": 7390 }, { "epoch": 1.46, "learning_rate": 8.845318056104307e-05, "loss": 0.5497, "step": 7400 }, { "epoch": 1.46, "learning_rate": 8.840379296720664e-05, "loss": 0.536, "step": 7410 }, { "epoch": 1.47, "learning_rate": 8.835440537337021e-05, "loss": 0.5529, "step": 7420 }, { "epoch": 1.47, "learning_rate": 8.830501777953379e-05, "loss": 0.5797, "step": 7430 }, { "epoch": 1.47, "learning_rate": 8.825563018569736e-05, "loss": 0.5439, "step": 7440 }, { "epoch": 1.47, "learning_rate": 8.820624259186094e-05, "loss": 0.5486, "step": 7450 }, { "epoch": 1.47, "learning_rate": 8.81568549980245e-05, "loss": 0.5299, "step": 7460 }, { "epoch": 1.48, "learning_rate": 8.810746740418807e-05, "loss": 0.4768, "step": 7470 }, { "epoch": 1.48, "learning_rate": 8.805807981035165e-05, "loss": 0.5329, "step": 7480 }, { "epoch": 1.48, "learning_rate": 8.800869221651522e-05, "loss": 0.6029, "step": 7490 }, { "epoch": 1.48, "learning_rate": 8.795930462267878e-05, "loss": 0.5658, "step": 7500 }, { "epoch": 1.48, "learning_rate": 8.790991702884236e-05, "loss": 0.5596, "step": 7510 }, { "epoch": 1.49, "learning_rate": 8.786052943500593e-05, "loss": 0.5864, "step": 7520 }, { "epoch": 1.49, "learning_rate": 8.78111418411695e-05, "loss": 0.5159, "step": 7530 }, { "epoch": 1.49, "learning_rate": 8.776175424733308e-05, "loss": 0.5958, "step": 7540 }, { "epoch": 1.49, "learning_rate": 8.771236665349664e-05, "loss": 0.5756, "step": 7550 }, { "epoch": 1.49, "learning_rate": 8.766297905966022e-05, "loss": 0.5216, "step": 7560 }, { "epoch": 1.5, "learning_rate": 8.761359146582379e-05, "loss": 0.5333, "step": 7570 }, { "epoch": 1.5, "learning_rate": 8.756420387198736e-05, "loss": 0.5592, "step": 7580 }, { "epoch": 1.5, "learning_rate": 8.751481627815094e-05, "loss": 0.5026, "step": 7590 }, { "epoch": 1.5, "learning_rate": 8.74654286843145e-05, "loss": 0.519, "step": 7600 }, { "epoch": 1.5, "learning_rate": 8.741604109047807e-05, "loss": 0.5751, "step": 7610 }, { "epoch": 1.51, "learning_rate": 8.736665349664165e-05, "loss": 0.526, "step": 7620 }, { "epoch": 1.51, "learning_rate": 8.731726590280522e-05, "loss": 0.5635, "step": 7630 }, { "epoch": 1.51, "learning_rate": 8.726787830896878e-05, "loss": 0.55, "step": 7640 }, { "epoch": 1.51, "learning_rate": 8.721849071513236e-05, "loss": 0.5643, "step": 7650 }, { "epoch": 1.51, "learning_rate": 8.716910312129593e-05, "loss": 0.567, "step": 7660 }, { "epoch": 1.51, "learning_rate": 8.711971552745951e-05, "loss": 0.5279, "step": 7670 }, { "epoch": 1.52, "learning_rate": 8.707032793362308e-05, "loss": 0.6123, "step": 7680 }, { "epoch": 1.52, "learning_rate": 8.702094033978665e-05, "loss": 0.5593, "step": 7690 }, { "epoch": 1.52, "learning_rate": 8.697155274595023e-05, "loss": 0.5988, "step": 7700 }, { "epoch": 1.52, "learning_rate": 8.692216515211379e-05, "loss": 0.5519, "step": 7710 }, { "epoch": 1.52, "learning_rate": 8.687277755827736e-05, "loss": 0.5786, "step": 7720 }, { "epoch": 1.53, "learning_rate": 8.682338996444094e-05, "loss": 0.5308, "step": 7730 }, { "epoch": 1.53, "learning_rate": 8.67740023706045e-05, "loss": 0.5661, "step": 7740 }, { "epoch": 1.53, "learning_rate": 8.672461477676807e-05, "loss": 0.551, "step": 7750 }, { "epoch": 1.53, "learning_rate": 8.667522718293165e-05, "loss": 0.5475, "step": 7760 }, { "epoch": 1.53, "learning_rate": 8.662583958909522e-05, "loss": 0.5375, "step": 7770 }, { "epoch": 1.54, "learning_rate": 8.65764519952588e-05, "loss": 0.4967, "step": 7780 }, { "epoch": 1.54, "learning_rate": 8.652706440142237e-05, "loss": 0.5559, "step": 7790 }, { "epoch": 1.54, "learning_rate": 8.647767680758593e-05, "loss": 0.5681, "step": 7800 }, { "epoch": 1.54, "learning_rate": 8.642828921374951e-05, "loss": 0.5456, "step": 7810 }, { "epoch": 1.54, "learning_rate": 8.637890161991308e-05, "loss": 0.516, "step": 7820 }, { "epoch": 1.55, "learning_rate": 8.632951402607665e-05, "loss": 0.5706, "step": 7830 }, { "epoch": 1.55, "learning_rate": 8.628012643224023e-05, "loss": 0.541, "step": 7840 }, { "epoch": 1.55, "learning_rate": 8.623073883840381e-05, "loss": 0.5563, "step": 7850 }, { "epoch": 1.55, "learning_rate": 8.618135124456736e-05, "loss": 0.5303, "step": 7860 }, { "epoch": 1.55, "learning_rate": 8.613196365073094e-05, "loss": 0.588, "step": 7870 }, { "epoch": 1.56, "learning_rate": 8.608257605689452e-05, "loss": 0.5393, "step": 7880 }, { "epoch": 1.56, "learning_rate": 8.603318846305807e-05, "loss": 0.5165, "step": 7890 }, { "epoch": 1.56, "learning_rate": 8.598380086922165e-05, "loss": 0.5388, "step": 7900 }, { "epoch": 1.56, "learning_rate": 8.593441327538523e-05, "loss": 0.5849, "step": 7910 }, { "epoch": 1.56, "learning_rate": 8.58850256815488e-05, "loss": 0.5098, "step": 7920 }, { "epoch": 1.57, "learning_rate": 8.583563808771237e-05, "loss": 0.5349, "step": 7930 }, { "epoch": 1.57, "learning_rate": 8.578625049387594e-05, "loss": 0.5354, "step": 7940 }, { "epoch": 1.57, "learning_rate": 8.573686290003952e-05, "loss": 0.4985, "step": 7950 }, { "epoch": 1.57, "learning_rate": 8.56874753062031e-05, "loss": 0.5504, "step": 7960 }, { "epoch": 1.57, "learning_rate": 8.563808771236665e-05, "loss": 0.5041, "step": 7970 }, { "epoch": 1.58, "learning_rate": 8.558870011853023e-05, "loss": 0.5079, "step": 7980 }, { "epoch": 1.58, "learning_rate": 8.553931252469381e-05, "loss": 0.5435, "step": 7990 }, { "epoch": 1.58, "learning_rate": 8.548992493085736e-05, "loss": 0.5538, "step": 8000 }, { "epoch": 1.58, "learning_rate": 8.544053733702094e-05, "loss": 0.5413, "step": 8010 }, { "epoch": 1.58, "learning_rate": 8.539114974318452e-05, "loss": 0.5474, "step": 8020 }, { "epoch": 1.59, "learning_rate": 8.534176214934809e-05, "loss": 0.5741, "step": 8030 }, { "epoch": 1.59, "learning_rate": 8.529237455551166e-05, "loss": 0.4931, "step": 8040 }, { "epoch": 1.59, "learning_rate": 8.524298696167524e-05, "loss": 0.5258, "step": 8050 }, { "epoch": 1.59, "learning_rate": 8.51935993678388e-05, "loss": 0.522, "step": 8060 }, { "epoch": 1.59, "learning_rate": 8.514421177400237e-05, "loss": 0.5691, "step": 8070 }, { "epoch": 1.6, "learning_rate": 8.509482418016595e-05, "loss": 0.5119, "step": 8080 }, { "epoch": 1.6, "learning_rate": 8.504543658632952e-05, "loss": 0.5037, "step": 8090 }, { "epoch": 1.6, "learning_rate": 8.49960489924931e-05, "loss": 0.5156, "step": 8100 }, { "epoch": 1.6, "learning_rate": 8.494666139865665e-05, "loss": 0.5428, "step": 8110 }, { "epoch": 1.6, "learning_rate": 8.489727380482023e-05, "loss": 0.5575, "step": 8120 }, { "epoch": 1.61, "learning_rate": 8.484788621098381e-05, "loss": 0.5455, "step": 8130 }, { "epoch": 1.61, "learning_rate": 8.479849861714738e-05, "loss": 0.544, "step": 8140 }, { "epoch": 1.61, "learning_rate": 8.474911102331094e-05, "loss": 0.514, "step": 8150 }, { "epoch": 1.61, "learning_rate": 8.469972342947452e-05, "loss": 0.5407, "step": 8160 }, { "epoch": 1.61, "learning_rate": 8.465033583563809e-05, "loss": 0.5199, "step": 8170 }, { "epoch": 1.62, "learning_rate": 8.460094824180166e-05, "loss": 0.5594, "step": 8180 }, { "epoch": 1.62, "learning_rate": 8.455156064796524e-05, "loss": 0.5609, "step": 8190 }, { "epoch": 1.62, "learning_rate": 8.45021730541288e-05, "loss": 0.5069, "step": 8200 }, { "epoch": 1.62, "learning_rate": 8.445278546029239e-05, "loss": 0.5429, "step": 8210 }, { "epoch": 1.62, "learning_rate": 8.440339786645595e-05, "loss": 0.5732, "step": 8220 }, { "epoch": 1.63, "learning_rate": 8.435401027261952e-05, "loss": 0.5299, "step": 8230 }, { "epoch": 1.63, "learning_rate": 8.43046226787831e-05, "loss": 0.502, "step": 8240 }, { "epoch": 1.63, "learning_rate": 8.425523508494667e-05, "loss": 0.5781, "step": 8250 }, { "epoch": 1.63, "learning_rate": 8.420584749111023e-05, "loss": 0.531, "step": 8260 }, { "epoch": 1.63, "learning_rate": 8.415645989727381e-05, "loss": 0.5025, "step": 8270 }, { "epoch": 1.64, "learning_rate": 8.410707230343738e-05, "loss": 0.5567, "step": 8280 }, { "epoch": 1.64, "learning_rate": 8.405768470960095e-05, "loss": 0.5461, "step": 8290 }, { "epoch": 1.64, "learning_rate": 8.400829711576453e-05, "loss": 0.5654, "step": 8300 }, { "epoch": 1.64, "learning_rate": 8.39589095219281e-05, "loss": 0.5052, "step": 8310 }, { "epoch": 1.64, "learning_rate": 8.390952192809167e-05, "loss": 0.534, "step": 8320 }, { "epoch": 1.65, "learning_rate": 8.386013433425524e-05, "loss": 0.5098, "step": 8330 }, { "epoch": 1.65, "learning_rate": 8.381074674041881e-05, "loss": 0.5172, "step": 8340 }, { "epoch": 1.65, "learning_rate": 8.376135914658239e-05, "loss": 0.5299, "step": 8350 }, { "epoch": 1.65, "learning_rate": 8.371197155274595e-05, "loss": 0.5552, "step": 8360 }, { "epoch": 1.65, "learning_rate": 8.366258395890952e-05, "loss": 0.5383, "step": 8370 }, { "epoch": 1.66, "learning_rate": 8.36131963650731e-05, "loss": 0.5857, "step": 8380 }, { "epoch": 1.66, "learning_rate": 8.356380877123667e-05, "loss": 0.5113, "step": 8390 }, { "epoch": 1.66, "learning_rate": 8.351442117740023e-05, "loss": 0.5048, "step": 8400 }, { "epoch": 1.66, "learning_rate": 8.346503358356381e-05, "loss": 0.5455, "step": 8410 }, { "epoch": 1.66, "learning_rate": 8.341564598972738e-05, "loss": 0.5871, "step": 8420 }, { "epoch": 1.67, "learning_rate": 8.336625839589096e-05, "loss": 0.5267, "step": 8430 }, { "epoch": 1.67, "learning_rate": 8.331687080205453e-05, "loss": 0.5342, "step": 8440 }, { "epoch": 1.67, "learning_rate": 8.32674832082181e-05, "loss": 0.5157, "step": 8450 }, { "epoch": 1.67, "learning_rate": 8.321809561438168e-05, "loss": 0.589, "step": 8460 }, { "epoch": 1.67, "learning_rate": 8.316870802054524e-05, "loss": 0.5377, "step": 8470 }, { "epoch": 1.67, "learning_rate": 8.311932042670881e-05, "loss": 0.5775, "step": 8480 }, { "epoch": 1.68, "learning_rate": 8.306993283287239e-05, "loss": 0.561, "step": 8490 }, { "epoch": 1.68, "learning_rate": 8.302054523903597e-05, "loss": 0.5283, "step": 8500 }, { "epoch": 1.68, "learning_rate": 8.297115764519952e-05, "loss": 0.5261, "step": 8510 }, { "epoch": 1.68, "learning_rate": 8.29217700513631e-05, "loss": 0.5529, "step": 8520 }, { "epoch": 1.68, "learning_rate": 8.287238245752667e-05, "loss": 0.5438, "step": 8530 }, { "epoch": 1.69, "learning_rate": 8.282299486369024e-05, "loss": 0.5124, "step": 8540 }, { "epoch": 1.69, "learning_rate": 8.277360726985382e-05, "loss": 0.5415, "step": 8550 }, { "epoch": 1.69, "learning_rate": 8.272421967601738e-05, "loss": 0.5833, "step": 8560 }, { "epoch": 1.69, "learning_rate": 8.267483208218096e-05, "loss": 0.5628, "step": 8570 }, { "epoch": 1.69, "learning_rate": 8.262544448834453e-05, "loss": 0.5185, "step": 8580 }, { "epoch": 1.7, "learning_rate": 8.25760568945081e-05, "loss": 0.5689, "step": 8590 }, { "epoch": 1.7, "learning_rate": 8.252666930067168e-05, "loss": 0.5576, "step": 8600 }, { "epoch": 1.7, "learning_rate": 8.247728170683526e-05, "loss": 0.4908, "step": 8610 }, { "epoch": 1.7, "learning_rate": 8.242789411299881e-05, "loss": 0.565, "step": 8620 }, { "epoch": 1.7, "learning_rate": 8.237850651916239e-05, "loss": 0.5648, "step": 8630 }, { "epoch": 1.71, "learning_rate": 8.232911892532597e-05, "loss": 0.595, "step": 8640 }, { "epoch": 1.71, "learning_rate": 8.227973133148952e-05, "loss": 0.5363, "step": 8650 }, { "epoch": 1.71, "learning_rate": 8.22303437376531e-05, "loss": 0.5395, "step": 8660 }, { "epoch": 1.71, "learning_rate": 8.218095614381669e-05, "loss": 0.5673, "step": 8670 }, { "epoch": 1.71, "learning_rate": 8.213156854998025e-05, "loss": 0.5638, "step": 8680 }, { "epoch": 1.72, "learning_rate": 8.208218095614382e-05, "loss": 0.5137, "step": 8690 }, { "epoch": 1.72, "learning_rate": 8.203279336230739e-05, "loss": 0.5278, "step": 8700 }, { "epoch": 1.72, "learning_rate": 8.198340576847097e-05, "loss": 0.5497, "step": 8710 }, { "epoch": 1.72, "learning_rate": 8.193401817463453e-05, "loss": 0.4916, "step": 8720 }, { "epoch": 1.72, "learning_rate": 8.18846305807981e-05, "loss": 0.5153, "step": 8730 }, { "epoch": 1.73, "learning_rate": 8.183524298696168e-05, "loss": 0.5019, "step": 8740 }, { "epoch": 1.73, "learning_rate": 8.178585539312526e-05, "loss": 0.5613, "step": 8750 }, { "epoch": 1.73, "learning_rate": 8.173646779928881e-05, "loss": 0.5541, "step": 8760 }, { "epoch": 1.73, "learning_rate": 8.168708020545239e-05, "loss": 0.4776, "step": 8770 }, { "epoch": 1.73, "learning_rate": 8.163769261161597e-05, "loss": 0.5519, "step": 8780 }, { "epoch": 1.74, "learning_rate": 8.158830501777954e-05, "loss": 0.5948, "step": 8790 }, { "epoch": 1.74, "learning_rate": 8.15389174239431e-05, "loss": 0.542, "step": 8800 }, { "epoch": 1.74, "learning_rate": 8.148952983010669e-05, "loss": 0.5083, "step": 8810 }, { "epoch": 1.74, "learning_rate": 8.144014223627025e-05, "loss": 0.5197, "step": 8820 }, { "epoch": 1.74, "learning_rate": 8.139075464243382e-05, "loss": 0.5461, "step": 8830 }, { "epoch": 1.75, "learning_rate": 8.13413670485974e-05, "loss": 0.5376, "step": 8840 }, { "epoch": 1.75, "learning_rate": 8.129197945476097e-05, "loss": 0.5457, "step": 8850 }, { "epoch": 1.75, "learning_rate": 8.124259186092455e-05, "loss": 0.558, "step": 8860 }, { "epoch": 1.75, "learning_rate": 8.119320426708811e-05, "loss": 0.5685, "step": 8870 }, { "epoch": 1.75, "learning_rate": 8.114381667325168e-05, "loss": 0.5052, "step": 8880 }, { "epoch": 1.76, "learning_rate": 8.109442907941526e-05, "loss": 0.5487, "step": 8890 }, { "epoch": 1.76, "learning_rate": 8.104504148557883e-05, "loss": 0.5546, "step": 8900 }, { "epoch": 1.76, "learning_rate": 8.09956538917424e-05, "loss": 0.5393, "step": 8910 }, { "epoch": 1.76, "learning_rate": 8.094626629790597e-05, "loss": 0.5066, "step": 8920 }, { "epoch": 1.76, "learning_rate": 8.089687870406954e-05, "loss": 0.5824, "step": 8930 }, { "epoch": 1.77, "learning_rate": 8.084749111023311e-05, "loss": 0.5131, "step": 8940 }, { "epoch": 1.77, "learning_rate": 8.079810351639669e-05, "loss": 0.5413, "step": 8950 }, { "epoch": 1.77, "learning_rate": 8.074871592256026e-05, "loss": 0.5316, "step": 8960 }, { "epoch": 1.77, "learning_rate": 8.069932832872384e-05, "loss": 0.5096, "step": 8970 }, { "epoch": 1.77, "learning_rate": 8.06499407348874e-05, "loss": 0.4838, "step": 8980 }, { "epoch": 1.78, "learning_rate": 8.060055314105097e-05, "loss": 0.5129, "step": 8990 }, { "epoch": 1.78, "learning_rate": 8.055116554721455e-05, "loss": 0.5302, "step": 9000 }, { "epoch": 1.78, "learning_rate": 8.050177795337812e-05, "loss": 0.5163, "step": 9010 }, { "epoch": 1.78, "learning_rate": 8.045239035954168e-05, "loss": 0.5598, "step": 9020 }, { "epoch": 1.78, "learning_rate": 8.040300276570526e-05, "loss": 0.5391, "step": 9030 }, { "epoch": 1.79, "learning_rate": 8.035361517186883e-05, "loss": 0.5095, "step": 9040 }, { "epoch": 1.79, "learning_rate": 8.03042275780324e-05, "loss": 0.5442, "step": 9050 }, { "epoch": 1.79, "learning_rate": 8.025483998419598e-05, "loss": 0.5027, "step": 9060 }, { "epoch": 1.79, "learning_rate": 8.020545239035954e-05, "loss": 0.5685, "step": 9070 }, { "epoch": 1.79, "learning_rate": 8.015606479652312e-05, "loss": 0.5163, "step": 9080 }, { "epoch": 1.8, "learning_rate": 8.010667720268669e-05, "loss": 0.5429, "step": 9090 }, { "epoch": 1.8, "learning_rate": 8.005728960885026e-05, "loss": 0.5705, "step": 9100 }, { "epoch": 1.8, "learning_rate": 8.000790201501384e-05, "loss": 0.5038, "step": 9110 }, { "epoch": 1.8, "learning_rate": 7.99585144211774e-05, "loss": 0.5625, "step": 9120 }, { "epoch": 1.8, "learning_rate": 7.990912682734097e-05, "loss": 0.5201, "step": 9130 }, { "epoch": 1.81, "learning_rate": 7.985973923350455e-05, "loss": 0.535, "step": 9140 }, { "epoch": 1.81, "learning_rate": 7.981035163966812e-05, "loss": 0.4948, "step": 9150 }, { "epoch": 1.81, "learning_rate": 7.976096404583168e-05, "loss": 0.5307, "step": 9160 }, { "epoch": 1.81, "learning_rate": 7.971157645199526e-05, "loss": 0.5289, "step": 9170 }, { "epoch": 1.81, "learning_rate": 7.966218885815883e-05, "loss": 0.5167, "step": 9180 }, { "epoch": 1.82, "learning_rate": 7.96128012643224e-05, "loss": 0.5574, "step": 9190 }, { "epoch": 1.82, "learning_rate": 7.956341367048598e-05, "loss": 0.5457, "step": 9200 }, { "epoch": 1.82, "learning_rate": 7.951402607664955e-05, "loss": 0.5446, "step": 9210 }, { "epoch": 1.82, "learning_rate": 7.946463848281313e-05, "loss": 0.5077, "step": 9220 }, { "epoch": 1.82, "learning_rate": 7.941525088897669e-05, "loss": 0.5293, "step": 9230 }, { "epoch": 1.83, "learning_rate": 7.936586329514026e-05, "loss": 0.5282, "step": 9240 }, { "epoch": 1.83, "learning_rate": 7.931647570130384e-05, "loss": 0.5269, "step": 9250 }, { "epoch": 1.83, "learning_rate": 7.926708810746742e-05, "loss": 0.5453, "step": 9260 }, { "epoch": 1.83, "learning_rate": 7.921770051363097e-05, "loss": 0.5281, "step": 9270 }, { "epoch": 1.83, "learning_rate": 7.916831291979455e-05, "loss": 0.5279, "step": 9280 }, { "epoch": 1.83, "learning_rate": 7.911892532595812e-05, "loss": 0.5144, "step": 9290 }, { "epoch": 1.84, "learning_rate": 7.906953773212169e-05, "loss": 0.5459, "step": 9300 }, { "epoch": 1.84, "learning_rate": 7.902015013828527e-05, "loss": 0.5746, "step": 9310 }, { "epoch": 1.84, "learning_rate": 7.897076254444883e-05, "loss": 0.5081, "step": 9320 }, { "epoch": 1.84, "learning_rate": 7.892137495061241e-05, "loss": 0.513, "step": 9330 }, { "epoch": 1.84, "learning_rate": 7.887198735677598e-05, "loss": 0.5399, "step": 9340 }, { "epoch": 1.85, "learning_rate": 7.882259976293955e-05, "loss": 0.5265, "step": 9350 }, { "epoch": 1.85, "learning_rate": 7.877321216910313e-05, "loss": 0.52, "step": 9360 }, { "epoch": 1.85, "learning_rate": 7.87238245752667e-05, "loss": 0.494, "step": 9370 }, { "epoch": 1.85, "learning_rate": 7.867443698143026e-05, "loss": 0.5251, "step": 9380 }, { "epoch": 1.85, "learning_rate": 7.862504938759384e-05, "loss": 0.5215, "step": 9390 }, { "epoch": 1.86, "learning_rate": 7.857566179375742e-05, "loss": 0.5235, "step": 9400 }, { "epoch": 1.86, "learning_rate": 7.852627419992097e-05, "loss": 0.5492, "step": 9410 }, { "epoch": 1.86, "learning_rate": 7.847688660608455e-05, "loss": 0.4368, "step": 9420 }, { "epoch": 1.86, "learning_rate": 7.842749901224814e-05, "loss": 0.4861, "step": 9430 }, { "epoch": 1.86, "learning_rate": 7.83781114184117e-05, "loss": 0.5302, "step": 9440 }, { "epoch": 1.87, "learning_rate": 7.832872382457527e-05, "loss": 0.5117, "step": 9450 }, { "epoch": 1.87, "learning_rate": 7.827933623073885e-05, "loss": 0.5522, "step": 9460 }, { "epoch": 1.87, "learning_rate": 7.822994863690242e-05, "loss": 0.5352, "step": 9470 }, { "epoch": 1.87, "learning_rate": 7.818056104306598e-05, "loss": 0.5258, "step": 9480 }, { "epoch": 1.87, "learning_rate": 7.813117344922955e-05, "loss": 0.5125, "step": 9490 }, { "epoch": 1.88, "learning_rate": 7.808178585539313e-05, "loss": 0.511, "step": 9500 }, { "epoch": 1.88, "learning_rate": 7.803239826155671e-05, "loss": 0.5078, "step": 9510 }, { "epoch": 1.88, "learning_rate": 7.798301066772026e-05, "loss": 0.5273, "step": 9520 }, { "epoch": 1.88, "learning_rate": 7.793362307388384e-05, "loss": 0.5147, "step": 9530 }, { "epoch": 1.88, "learning_rate": 7.788423548004742e-05, "loss": 0.5201, "step": 9540 }, { "epoch": 1.89, "learning_rate": 7.783484788621098e-05, "loss": 0.5182, "step": 9550 }, { "epoch": 1.89, "learning_rate": 7.778546029237456e-05, "loss": 0.4878, "step": 9560 }, { "epoch": 1.89, "learning_rate": 7.773607269853814e-05, "loss": 0.5009, "step": 9570 }, { "epoch": 1.89, "learning_rate": 7.76866851047017e-05, "loss": 0.5436, "step": 9580 }, { "epoch": 1.89, "learning_rate": 7.763729751086527e-05, "loss": 0.5333, "step": 9590 }, { "epoch": 1.9, "learning_rate": 7.758790991702885e-05, "loss": 0.5228, "step": 9600 }, { "epoch": 1.9, "learning_rate": 7.753852232319242e-05, "loss": 0.5475, "step": 9610 }, { "epoch": 1.9, "learning_rate": 7.7489134729356e-05, "loss": 0.5059, "step": 9620 }, { "epoch": 1.9, "learning_rate": 7.743974713551956e-05, "loss": 0.5058, "step": 9630 }, { "epoch": 1.9, "learning_rate": 7.739035954168313e-05, "loss": 0.4555, "step": 9640 }, { "epoch": 1.91, "learning_rate": 7.734097194784671e-05, "loss": 0.5018, "step": 9650 }, { "epoch": 1.91, "learning_rate": 7.729158435401026e-05, "loss": 0.5356, "step": 9660 }, { "epoch": 1.91, "learning_rate": 7.724219676017384e-05, "loss": 0.5103, "step": 9670 }, { "epoch": 1.91, "learning_rate": 7.719280916633743e-05, "loss": 0.5163, "step": 9680 }, { "epoch": 1.91, "learning_rate": 7.714342157250099e-05, "loss": 0.5261, "step": 9690 }, { "epoch": 1.92, "learning_rate": 7.709403397866456e-05, "loss": 0.5344, "step": 9700 }, { "epoch": 1.92, "learning_rate": 7.704464638482814e-05, "loss": 0.5286, "step": 9710 }, { "epoch": 1.92, "learning_rate": 7.69952587909917e-05, "loss": 0.5077, "step": 9720 }, { "epoch": 1.92, "learning_rate": 7.694587119715529e-05, "loss": 0.5161, "step": 9730 }, { "epoch": 1.92, "learning_rate": 7.689648360331885e-05, "loss": 0.4923, "step": 9740 }, { "epoch": 1.93, "learning_rate": 7.684709600948242e-05, "loss": 0.4912, "step": 9750 }, { "epoch": 1.93, "learning_rate": 7.6797708415646e-05, "loss": 0.5102, "step": 9760 }, { "epoch": 1.93, "learning_rate": 7.674832082180957e-05, "loss": 0.4956, "step": 9770 }, { "epoch": 1.93, "learning_rate": 7.669893322797313e-05, "loss": 0.5357, "step": 9780 }, { "epoch": 1.93, "learning_rate": 7.664954563413671e-05, "loss": 0.5348, "step": 9790 }, { "epoch": 1.94, "learning_rate": 7.660015804030028e-05, "loss": 0.4979, "step": 9800 }, { "epoch": 1.94, "learning_rate": 7.655077044646385e-05, "loss": 0.5565, "step": 9810 }, { "epoch": 1.94, "learning_rate": 7.650138285262743e-05, "loss": 0.4912, "step": 9820 }, { "epoch": 1.94, "learning_rate": 7.6451995258791e-05, "loss": 0.4824, "step": 9830 }, { "epoch": 1.94, "learning_rate": 7.640260766495456e-05, "loss": 0.4981, "step": 9840 }, { "epoch": 1.95, "learning_rate": 7.635322007111814e-05, "loss": 0.5504, "step": 9850 }, { "epoch": 1.95, "learning_rate": 7.630383247728171e-05, "loss": 0.5166, "step": 9860 }, { "epoch": 1.95, "learning_rate": 7.625444488344529e-05, "loss": 0.4982, "step": 9870 }, { "epoch": 1.95, "learning_rate": 7.620505728960885e-05, "loss": 0.554, "step": 9880 }, { "epoch": 1.95, "learning_rate": 7.615566969577242e-05, "loss": 0.539, "step": 9890 }, { "epoch": 1.96, "learning_rate": 7.6106282101936e-05, "loss": 0.5196, "step": 9900 }, { "epoch": 1.96, "learning_rate": 7.605689450809957e-05, "loss": 0.5404, "step": 9910 }, { "epoch": 1.96, "learning_rate": 7.600750691426313e-05, "loss": 0.5312, "step": 9920 }, { "epoch": 1.96, "learning_rate": 7.595811932042672e-05, "loss": 0.5174, "step": 9930 }, { "epoch": 1.96, "learning_rate": 7.590873172659028e-05, "loss": 0.4971, "step": 9940 }, { "epoch": 1.97, "learning_rate": 7.585934413275385e-05, "loss": 0.4996, "step": 9950 }, { "epoch": 1.97, "learning_rate": 7.580995653891743e-05, "loss": 0.5504, "step": 9960 }, { "epoch": 1.97, "learning_rate": 7.5760568945081e-05, "loss": 0.5405, "step": 9970 }, { "epoch": 1.97, "learning_rate": 7.571118135124458e-05, "loss": 0.5649, "step": 9980 }, { "epoch": 1.97, "learning_rate": 7.566179375740814e-05, "loss": 0.5453, "step": 9990 }, { "epoch": 1.98, "learning_rate": 7.561240616357171e-05, "loss": 0.536, "step": 10000 }, { "epoch": 1.98, "learning_rate": 7.556301856973529e-05, "loss": 0.5153, "step": 10010 }, { "epoch": 1.98, "learning_rate": 7.551363097589886e-05, "loss": 0.4975, "step": 10020 }, { "epoch": 1.98, "learning_rate": 7.546424338206242e-05, "loss": 0.5167, "step": 10030 }, { "epoch": 1.98, "learning_rate": 7.5414855788226e-05, "loss": 0.5005, "step": 10040 }, { "epoch": 1.99, "learning_rate": 7.536546819438958e-05, "loss": 0.4912, "step": 10050 }, { "epoch": 1.99, "learning_rate": 7.531608060055314e-05, "loss": 0.4992, "step": 10060 }, { "epoch": 1.99, "learning_rate": 7.526669300671672e-05, "loss": 0.507, "step": 10070 }, { "epoch": 1.99, "learning_rate": 7.521730541288028e-05, "loss": 0.4761, "step": 10080 }, { "epoch": 1.99, "learning_rate": 7.516791781904386e-05, "loss": 0.5454, "step": 10090 }, { "epoch": 1.99, "learning_rate": 7.511853022520743e-05, "loss": 0.5022, "step": 10100 }, { "epoch": 2.0, "learning_rate": 7.5069142631371e-05, "loss": 0.5319, "step": 10110 }, { "epoch": 2.0, "learning_rate": 7.501975503753458e-05, "loss": 0.512, "step": 10120 }, { "epoch": 2.0, "learning_rate": 7.497036744369814e-05, "loss": 0.4747, "step": 10130 }, { "epoch": 2.0, "learning_rate": 7.492097984986171e-05, "loss": 0.4652, "step": 10140 }, { "epoch": 2.0, "learning_rate": 7.487159225602529e-05, "loss": 0.408, "step": 10150 }, { "epoch": 2.01, "learning_rate": 7.482220466218887e-05, "loss": 0.4311, "step": 10160 }, { "epoch": 2.01, "learning_rate": 7.477281706835242e-05, "loss": 0.451, "step": 10170 }, { "epoch": 2.01, "learning_rate": 7.4723429474516e-05, "loss": 0.4021, "step": 10180 }, { "epoch": 2.01, "learning_rate": 7.467404188067959e-05, "loss": 0.45, "step": 10190 }, { "epoch": 2.01, "learning_rate": 7.462465428684314e-05, "loss": 0.446, "step": 10200 }, { "epoch": 2.02, "learning_rate": 7.457526669300672e-05, "loss": 0.4205, "step": 10210 }, { "epoch": 2.02, "learning_rate": 7.45258790991703e-05, "loss": 0.435, "step": 10220 }, { "epoch": 2.02, "learning_rate": 7.447649150533387e-05, "loss": 0.4292, "step": 10230 }, { "epoch": 2.02, "learning_rate": 7.442710391149743e-05, "loss": 0.4087, "step": 10240 }, { "epoch": 2.02, "learning_rate": 7.4377716317661e-05, "loss": 0.4601, "step": 10250 }, { "epoch": 2.03, "learning_rate": 7.432832872382458e-05, "loss": 0.4399, "step": 10260 }, { "epoch": 2.03, "learning_rate": 7.427894112998816e-05, "loss": 0.4307, "step": 10270 }, { "epoch": 2.03, "learning_rate": 7.422955353615171e-05, "loss": 0.4409, "step": 10280 }, { "epoch": 2.03, "learning_rate": 7.418016594231529e-05, "loss": 0.4337, "step": 10290 }, { "epoch": 2.03, "learning_rate": 7.413077834847887e-05, "loss": 0.4195, "step": 10300 }, { "epoch": 2.04, "learning_rate": 7.408139075464243e-05, "loss": 0.4394, "step": 10310 }, { "epoch": 2.04, "learning_rate": 7.403200316080601e-05, "loss": 0.4069, "step": 10320 }, { "epoch": 2.04, "learning_rate": 7.398261556696959e-05, "loss": 0.5025, "step": 10330 }, { "epoch": 2.04, "learning_rate": 7.393322797313315e-05, "loss": 0.4302, "step": 10340 }, { "epoch": 2.04, "learning_rate": 7.388384037929672e-05, "loss": 0.3836, "step": 10350 }, { "epoch": 2.05, "learning_rate": 7.38344527854603e-05, "loss": 0.4198, "step": 10360 }, { "epoch": 2.05, "learning_rate": 7.378506519162387e-05, "loss": 0.4316, "step": 10370 }, { "epoch": 2.05, "learning_rate": 7.373567759778745e-05, "loss": 0.421, "step": 10380 }, { "epoch": 2.05, "learning_rate": 7.368629000395101e-05, "loss": 0.4232, "step": 10390 }, { "epoch": 2.05, "learning_rate": 7.363690241011458e-05, "loss": 0.462, "step": 10400 }, { "epoch": 2.06, "learning_rate": 7.358751481627816e-05, "loss": 0.4554, "step": 10410 }, { "epoch": 2.06, "learning_rate": 7.353812722244173e-05, "loss": 0.4388, "step": 10420 }, { "epoch": 2.06, "learning_rate": 7.34887396286053e-05, "loss": 0.4013, "step": 10430 }, { "epoch": 2.06, "learning_rate": 7.343935203476888e-05, "loss": 0.4185, "step": 10440 }, { "epoch": 2.06, "learning_rate": 7.338996444093244e-05, "loss": 0.4563, "step": 10450 }, { "epoch": 2.07, "learning_rate": 7.334057684709601e-05, "loss": 0.3997, "step": 10460 }, { "epoch": 2.07, "learning_rate": 7.329118925325959e-05, "loss": 0.4497, "step": 10470 }, { "epoch": 2.07, "learning_rate": 7.324180165942316e-05, "loss": 0.4241, "step": 10480 }, { "epoch": 2.07, "learning_rate": 7.319241406558672e-05, "loss": 0.4438, "step": 10490 }, { "epoch": 2.07, "learning_rate": 7.31430264717503e-05, "loss": 0.429, "step": 10500 }, { "epoch": 2.08, "learning_rate": 7.309363887791387e-05, "loss": 0.4455, "step": 10510 }, { "epoch": 2.08, "learning_rate": 7.304425128407745e-05, "loss": 0.4267, "step": 10520 }, { "epoch": 2.08, "learning_rate": 7.299486369024102e-05, "loss": 0.4512, "step": 10530 }, { "epoch": 2.08, "learning_rate": 7.294547609640458e-05, "loss": 0.4394, "step": 10540 }, { "epoch": 2.08, "learning_rate": 7.289608850256816e-05, "loss": 0.4488, "step": 10550 }, { "epoch": 2.09, "learning_rate": 7.284670090873173e-05, "loss": 0.4088, "step": 10560 }, { "epoch": 2.09, "learning_rate": 7.27973133148953e-05, "loss": 0.4847, "step": 10570 }, { "epoch": 2.09, "learning_rate": 7.274792572105888e-05, "loss": 0.4287, "step": 10580 }, { "epoch": 2.09, "learning_rate": 7.269853812722244e-05, "loss": 0.4414, "step": 10590 }, { "epoch": 2.09, "learning_rate": 7.264915053338601e-05, "loss": 0.4162, "step": 10600 }, { "epoch": 2.1, "learning_rate": 7.259976293954959e-05, "loss": 0.4359, "step": 10610 }, { "epoch": 2.1, "learning_rate": 7.255037534571316e-05, "loss": 0.4468, "step": 10620 }, { "epoch": 2.1, "learning_rate": 7.250098775187674e-05, "loss": 0.4561, "step": 10630 }, { "epoch": 2.1, "learning_rate": 7.24516001580403e-05, "loss": 0.4619, "step": 10640 }, { "epoch": 2.1, "learning_rate": 7.240221256420387e-05, "loss": 0.4395, "step": 10650 }, { "epoch": 2.11, "learning_rate": 7.235282497036745e-05, "loss": 0.4446, "step": 10660 }, { "epoch": 2.11, "learning_rate": 7.230343737653102e-05, "loss": 0.4841, "step": 10670 }, { "epoch": 2.11, "learning_rate": 7.225404978269458e-05, "loss": 0.3839, "step": 10680 }, { "epoch": 2.11, "learning_rate": 7.220466218885817e-05, "loss": 0.4462, "step": 10690 }, { "epoch": 2.11, "learning_rate": 7.215527459502173e-05, "loss": 0.4722, "step": 10700 }, { "epoch": 2.12, "learning_rate": 7.21058870011853e-05, "loss": 0.4182, "step": 10710 }, { "epoch": 2.12, "learning_rate": 7.205649940734888e-05, "loss": 0.431, "step": 10720 }, { "epoch": 2.12, "learning_rate": 7.200711181351245e-05, "loss": 0.4057, "step": 10730 }, { "epoch": 2.12, "learning_rate": 7.195772421967603e-05, "loss": 0.4348, "step": 10740 }, { "epoch": 2.12, "learning_rate": 7.190833662583959e-05, "loss": 0.4301, "step": 10750 }, { "epoch": 2.13, "learning_rate": 7.185894903200316e-05, "loss": 0.4799, "step": 10760 }, { "epoch": 2.13, "learning_rate": 7.180956143816674e-05, "loss": 0.4091, "step": 10770 }, { "epoch": 2.13, "learning_rate": 7.17601738443303e-05, "loss": 0.4327, "step": 10780 }, { "epoch": 2.13, "learning_rate": 7.171078625049387e-05, "loss": 0.4473, "step": 10790 }, { "epoch": 2.13, "learning_rate": 7.166139865665745e-05, "loss": 0.4408, "step": 10800 }, { "epoch": 2.14, "learning_rate": 7.161201106282103e-05, "loss": 0.4369, "step": 10810 }, { "epoch": 2.14, "learning_rate": 7.156262346898459e-05, "loss": 0.4663, "step": 10820 }, { "epoch": 2.14, "learning_rate": 7.151323587514817e-05, "loss": 0.4839, "step": 10830 }, { "epoch": 2.14, "learning_rate": 7.146384828131173e-05, "loss": 0.436, "step": 10840 }, { "epoch": 2.14, "learning_rate": 7.14144606874753e-05, "loss": 0.4519, "step": 10850 }, { "epoch": 2.15, "learning_rate": 7.136507309363888e-05, "loss": 0.443, "step": 10860 }, { "epoch": 2.15, "learning_rate": 7.131568549980245e-05, "loss": 0.4536, "step": 10870 }, { "epoch": 2.15, "learning_rate": 7.126629790596603e-05, "loss": 0.4421, "step": 10880 }, { "epoch": 2.15, "learning_rate": 7.12169103121296e-05, "loss": 0.4217, "step": 10890 }, { "epoch": 2.15, "learning_rate": 7.116752271829316e-05, "loss": 0.4341, "step": 10900 }, { "epoch": 2.15, "learning_rate": 7.111813512445674e-05, "loss": 0.462, "step": 10910 }, { "epoch": 2.16, "learning_rate": 7.106874753062032e-05, "loss": 0.4451, "step": 10920 }, { "epoch": 2.16, "learning_rate": 7.101935993678387e-05, "loss": 0.4213, "step": 10930 }, { "epoch": 2.16, "learning_rate": 7.096997234294746e-05, "loss": 0.4463, "step": 10940 }, { "epoch": 2.16, "learning_rate": 7.092058474911104e-05, "loss": 0.4522, "step": 10950 }, { "epoch": 2.16, "learning_rate": 7.087119715527459e-05, "loss": 0.4331, "step": 10960 }, { "epoch": 2.17, "learning_rate": 7.082180956143817e-05, "loss": 0.4318, "step": 10970 }, { "epoch": 2.17, "learning_rate": 7.077242196760175e-05, "loss": 0.4153, "step": 10980 }, { "epoch": 2.17, "learning_rate": 7.072303437376532e-05, "loss": 0.4129, "step": 10990 }, { "epoch": 2.17, "learning_rate": 7.067364677992888e-05, "loss": 0.4668, "step": 11000 }, { "epoch": 2.17, "learning_rate": 7.062425918609246e-05, "loss": 0.429, "step": 11010 }, { "epoch": 2.18, "learning_rate": 7.057487159225603e-05, "loss": 0.4205, "step": 11020 }, { "epoch": 2.18, "learning_rate": 7.052548399841961e-05, "loss": 0.4671, "step": 11030 }, { "epoch": 2.18, "learning_rate": 7.047609640458316e-05, "loss": 0.4439, "step": 11040 }, { "epoch": 2.18, "learning_rate": 7.042670881074674e-05, "loss": 0.4121, "step": 11050 }, { "epoch": 2.18, "learning_rate": 7.037732121691032e-05, "loss": 0.4212, "step": 11060 }, { "epoch": 2.19, "learning_rate": 7.032793362307388e-05, "loss": 0.4569, "step": 11070 }, { "epoch": 2.19, "learning_rate": 7.027854602923746e-05, "loss": 0.4484, "step": 11080 }, { "epoch": 2.19, "learning_rate": 7.022915843540104e-05, "loss": 0.4246, "step": 11090 }, { "epoch": 2.19, "learning_rate": 7.01797708415646e-05, "loss": 0.4445, "step": 11100 }, { "epoch": 2.19, "learning_rate": 7.013038324772817e-05, "loss": 0.4236, "step": 11110 }, { "epoch": 2.2, "learning_rate": 7.008099565389175e-05, "loss": 0.4249, "step": 11120 }, { "epoch": 2.2, "learning_rate": 7.003160806005532e-05, "loss": 0.4506, "step": 11130 }, { "epoch": 2.2, "learning_rate": 6.998222046621888e-05, "loss": 0.4407, "step": 11140 }, { "epoch": 2.2, "learning_rate": 6.993283287238246e-05, "loss": 0.4506, "step": 11150 }, { "epoch": 2.2, "learning_rate": 6.988344527854603e-05, "loss": 0.4303, "step": 11160 }, { "epoch": 2.21, "learning_rate": 6.983405768470961e-05, "loss": 0.4384, "step": 11170 }, { "epoch": 2.21, "learning_rate": 6.978467009087318e-05, "loss": 0.4633, "step": 11180 }, { "epoch": 2.21, "learning_rate": 6.973528249703675e-05, "loss": 0.4578, "step": 11190 }, { "epoch": 2.21, "learning_rate": 6.968589490320033e-05, "loss": 0.4353, "step": 11200 }, { "epoch": 2.21, "learning_rate": 6.963650730936389e-05, "loss": 0.4524, "step": 11210 }, { "epoch": 2.22, "learning_rate": 6.958711971552746e-05, "loss": 0.4377, "step": 11220 }, { "epoch": 2.22, "learning_rate": 6.953773212169104e-05, "loss": 0.4522, "step": 11230 }, { "epoch": 2.22, "learning_rate": 6.94883445278546e-05, "loss": 0.4306, "step": 11240 }, { "epoch": 2.22, "learning_rate": 6.943895693401817e-05, "loss": 0.4247, "step": 11250 }, { "epoch": 2.22, "learning_rate": 6.938956934018175e-05, "loss": 0.4427, "step": 11260 }, { "epoch": 2.23, "learning_rate": 6.934018174634532e-05, "loss": 0.4287, "step": 11270 }, { "epoch": 2.23, "learning_rate": 6.92907941525089e-05, "loss": 0.4011, "step": 11280 }, { "epoch": 2.23, "learning_rate": 6.924140655867247e-05, "loss": 0.4387, "step": 11290 }, { "epoch": 2.23, "learning_rate": 6.919201896483603e-05, "loss": 0.4872, "step": 11300 }, { "epoch": 2.23, "learning_rate": 6.914263137099961e-05, "loss": 0.418, "step": 11310 }, { "epoch": 2.24, "learning_rate": 6.909324377716318e-05, "loss": 0.4239, "step": 11320 }, { "epoch": 2.24, "learning_rate": 6.904385618332675e-05, "loss": 0.437, "step": 11330 }, { "epoch": 2.24, "learning_rate": 6.899446858949033e-05, "loss": 0.391, "step": 11340 }, { "epoch": 2.24, "learning_rate": 6.89450809956539e-05, "loss": 0.4657, "step": 11350 }, { "epoch": 2.24, "learning_rate": 6.889569340181746e-05, "loss": 0.4634, "step": 11360 }, { "epoch": 2.25, "learning_rate": 6.884630580798104e-05, "loss": 0.4405, "step": 11370 }, { "epoch": 2.25, "learning_rate": 6.879691821414461e-05, "loss": 0.4326, "step": 11380 }, { "epoch": 2.25, "learning_rate": 6.874753062030819e-05, "loss": 0.4539, "step": 11390 }, { "epoch": 2.25, "learning_rate": 6.869814302647175e-05, "loss": 0.443, "step": 11400 }, { "epoch": 2.25, "learning_rate": 6.864875543263532e-05, "loss": 0.4269, "step": 11410 }, { "epoch": 2.26, "learning_rate": 6.85993678387989e-05, "loss": 0.4318, "step": 11420 }, { "epoch": 2.26, "learning_rate": 6.854998024496247e-05, "loss": 0.4316, "step": 11430 }, { "epoch": 2.26, "learning_rate": 6.850059265112604e-05, "loss": 0.4196, "step": 11440 }, { "epoch": 2.26, "learning_rate": 6.845120505728962e-05, "loss": 0.438, "step": 11450 }, { "epoch": 2.26, "learning_rate": 6.840181746345318e-05, "loss": 0.493, "step": 11460 }, { "epoch": 2.27, "learning_rate": 6.835242986961675e-05, "loss": 0.4801, "step": 11470 }, { "epoch": 2.27, "learning_rate": 6.830304227578033e-05, "loss": 0.4111, "step": 11480 }, { "epoch": 2.27, "learning_rate": 6.82536546819439e-05, "loss": 0.3861, "step": 11490 }, { "epoch": 2.27, "learning_rate": 6.820426708810746e-05, "loss": 0.4511, "step": 11500 }, { "epoch": 2.27, "learning_rate": 6.815487949427104e-05, "loss": 0.436, "step": 11510 }, { "epoch": 2.28, "learning_rate": 6.810549190043461e-05, "loss": 0.4236, "step": 11520 }, { "epoch": 2.28, "learning_rate": 6.805610430659819e-05, "loss": 0.4738, "step": 11530 }, { "epoch": 2.28, "learning_rate": 6.800671671276176e-05, "loss": 0.4166, "step": 11540 }, { "epoch": 2.28, "learning_rate": 6.795732911892532e-05, "loss": 0.4309, "step": 11550 }, { "epoch": 2.28, "learning_rate": 6.79079415250889e-05, "loss": 0.4716, "step": 11560 }, { "epoch": 2.29, "learning_rate": 6.785855393125248e-05, "loss": 0.4377, "step": 11570 }, { "epoch": 2.29, "learning_rate": 6.780916633741604e-05, "loss": 0.4413, "step": 11580 }, { "epoch": 2.29, "learning_rate": 6.775977874357962e-05, "loss": 0.4279, "step": 11590 }, { "epoch": 2.29, "learning_rate": 6.77103911497432e-05, "loss": 0.4277, "step": 11600 }, { "epoch": 2.29, "learning_rate": 6.766100355590675e-05, "loss": 0.4062, "step": 11610 }, { "epoch": 2.3, "learning_rate": 6.761161596207033e-05, "loss": 0.4405, "step": 11620 }, { "epoch": 2.3, "learning_rate": 6.75622283682339e-05, "loss": 0.428, "step": 11630 }, { "epoch": 2.3, "learning_rate": 6.751284077439748e-05, "loss": 0.4285, "step": 11640 }, { "epoch": 2.3, "learning_rate": 6.746345318056104e-05, "loss": 0.4557, "step": 11650 }, { "epoch": 2.3, "learning_rate": 6.741406558672461e-05, "loss": 0.4266, "step": 11660 }, { "epoch": 2.31, "learning_rate": 6.736467799288819e-05, "loss": 0.3776, "step": 11670 }, { "epoch": 2.31, "learning_rate": 6.731529039905177e-05, "loss": 0.4296, "step": 11680 }, { "epoch": 2.31, "learning_rate": 6.726590280521533e-05, "loss": 0.4286, "step": 11690 }, { "epoch": 2.31, "learning_rate": 6.72165152113789e-05, "loss": 0.4275, "step": 11700 }, { "epoch": 2.31, "learning_rate": 6.716712761754249e-05, "loss": 0.4664, "step": 11710 }, { "epoch": 2.31, "learning_rate": 6.711774002370604e-05, "loss": 0.4078, "step": 11720 }, { "epoch": 2.32, "learning_rate": 6.706835242986962e-05, "loss": 0.4596, "step": 11730 }, { "epoch": 2.32, "learning_rate": 6.70189648360332e-05, "loss": 0.4449, "step": 11740 }, { "epoch": 2.32, "learning_rate": 6.696957724219677e-05, "loss": 0.3848, "step": 11750 }, { "epoch": 2.32, "learning_rate": 6.692018964836033e-05, "loss": 0.4489, "step": 11760 }, { "epoch": 2.32, "learning_rate": 6.687080205452391e-05, "loss": 0.4493, "step": 11770 }, { "epoch": 2.33, "learning_rate": 6.682141446068748e-05, "loss": 0.4462, "step": 11780 }, { "epoch": 2.33, "learning_rate": 6.677202686685105e-05, "loss": 0.441, "step": 11790 }, { "epoch": 2.33, "learning_rate": 6.672263927301463e-05, "loss": 0.4282, "step": 11800 }, { "epoch": 2.33, "learning_rate": 6.66732516791782e-05, "loss": 0.4058, "step": 11810 }, { "epoch": 2.33, "learning_rate": 6.662386408534177e-05, "loss": 0.4412, "step": 11820 }, { "epoch": 2.34, "learning_rate": 6.657447649150533e-05, "loss": 0.4441, "step": 11830 }, { "epoch": 2.34, "learning_rate": 6.652508889766891e-05, "loss": 0.3896, "step": 11840 }, { "epoch": 2.34, "learning_rate": 6.647570130383249e-05, "loss": 0.4064, "step": 11850 }, { "epoch": 2.34, "learning_rate": 6.642631370999605e-05, "loss": 0.4218, "step": 11860 }, { "epoch": 2.34, "learning_rate": 6.637692611615962e-05, "loss": 0.4407, "step": 11870 }, { "epoch": 2.35, "learning_rate": 6.63275385223232e-05, "loss": 0.4624, "step": 11880 }, { "epoch": 2.35, "learning_rate": 6.627815092848677e-05, "loss": 0.4041, "step": 11890 }, { "epoch": 2.35, "learning_rate": 6.622876333465033e-05, "loss": 0.4273, "step": 11900 }, { "epoch": 2.35, "learning_rate": 6.617937574081391e-05, "loss": 0.4042, "step": 11910 }, { "epoch": 2.35, "learning_rate": 6.612998814697748e-05, "loss": 0.4119, "step": 11920 }, { "epoch": 2.36, "learning_rate": 6.608060055314106e-05, "loss": 0.4397, "step": 11930 }, { "epoch": 2.36, "learning_rate": 6.603121295930463e-05, "loss": 0.4398, "step": 11940 }, { "epoch": 2.36, "learning_rate": 6.59818253654682e-05, "loss": 0.4322, "step": 11950 }, { "epoch": 2.36, "learning_rate": 6.593243777163178e-05, "loss": 0.4379, "step": 11960 }, { "epoch": 2.36, "learning_rate": 6.588305017779534e-05, "loss": 0.4268, "step": 11970 }, { "epoch": 2.37, "learning_rate": 6.583366258395891e-05, "loss": 0.4101, "step": 11980 }, { "epoch": 2.37, "learning_rate": 6.578427499012249e-05, "loss": 0.4564, "step": 11990 }, { "epoch": 2.37, "learning_rate": 6.573488739628606e-05, "loss": 0.3988, "step": 12000 }, { "epoch": 2.37, "learning_rate": 6.568549980244962e-05, "loss": 0.4567, "step": 12010 }, { "epoch": 2.37, "learning_rate": 6.56361122086132e-05, "loss": 0.4723, "step": 12020 }, { "epoch": 2.38, "learning_rate": 6.558672461477677e-05, "loss": 0.4304, "step": 12030 }, { "epoch": 2.38, "learning_rate": 6.553733702094035e-05, "loss": 0.431, "step": 12040 }, { "epoch": 2.38, "learning_rate": 6.548794942710392e-05, "loss": 0.4345, "step": 12050 }, { "epoch": 2.38, "learning_rate": 6.543856183326748e-05, "loss": 0.4082, "step": 12060 }, { "epoch": 2.38, "learning_rate": 6.538917423943106e-05, "loss": 0.4282, "step": 12070 }, { "epoch": 2.39, "learning_rate": 6.533978664559463e-05, "loss": 0.4181, "step": 12080 }, { "epoch": 2.39, "learning_rate": 6.52903990517582e-05, "loss": 0.452, "step": 12090 }, { "epoch": 2.39, "learning_rate": 6.524101145792178e-05, "loss": 0.4253, "step": 12100 }, { "epoch": 2.39, "learning_rate": 6.519162386408534e-05, "loss": 0.4242, "step": 12110 }, { "epoch": 2.39, "learning_rate": 6.514223627024891e-05, "loss": 0.4548, "step": 12120 }, { "epoch": 2.4, "learning_rate": 6.509284867641249e-05, "loss": 0.4542, "step": 12130 }, { "epoch": 2.4, "learning_rate": 6.504346108257606e-05, "loss": 0.4085, "step": 12140 }, { "epoch": 2.4, "learning_rate": 6.499407348873962e-05, "loss": 0.4062, "step": 12150 }, { "epoch": 2.4, "learning_rate": 6.49446858949032e-05, "loss": 0.416, "step": 12160 }, { "epoch": 2.4, "learning_rate": 6.489529830106677e-05, "loss": 0.4605, "step": 12170 }, { "epoch": 2.41, "learning_rate": 6.484591070723035e-05, "loss": 0.4359, "step": 12180 }, { "epoch": 2.41, "learning_rate": 6.479652311339392e-05, "loss": 0.4457, "step": 12190 }, { "epoch": 2.41, "learning_rate": 6.474713551955749e-05, "loss": 0.456, "step": 12200 }, { "epoch": 2.41, "learning_rate": 6.469774792572107e-05, "loss": 0.4304, "step": 12210 }, { "epoch": 2.41, "learning_rate": 6.464836033188463e-05, "loss": 0.4564, "step": 12220 }, { "epoch": 2.42, "learning_rate": 6.45989727380482e-05, "loss": 0.4614, "step": 12230 }, { "epoch": 2.42, "learning_rate": 6.454958514421178e-05, "loss": 0.439, "step": 12240 }, { "epoch": 2.42, "learning_rate": 6.450019755037535e-05, "loss": 0.4148, "step": 12250 }, { "epoch": 2.42, "learning_rate": 6.445080995653891e-05, "loss": 0.4512, "step": 12260 }, { "epoch": 2.42, "learning_rate": 6.440142236270249e-05, "loss": 0.4143, "step": 12270 }, { "epoch": 2.43, "learning_rate": 6.435203476886606e-05, "loss": 0.4421, "step": 12280 }, { "epoch": 2.43, "learning_rate": 6.430264717502964e-05, "loss": 0.4673, "step": 12290 }, { "epoch": 2.43, "learning_rate": 6.42532595811932e-05, "loss": 0.4568, "step": 12300 }, { "epoch": 2.43, "learning_rate": 6.420387198735677e-05, "loss": 0.4564, "step": 12310 }, { "epoch": 2.43, "learning_rate": 6.415448439352035e-05, "loss": 0.4454, "step": 12320 }, { "epoch": 2.44, "learning_rate": 6.410509679968393e-05, "loss": 0.4466, "step": 12330 }, { "epoch": 2.44, "learning_rate": 6.405570920584749e-05, "loss": 0.4037, "step": 12340 }, { "epoch": 2.44, "learning_rate": 6.400632161201107e-05, "loss": 0.4273, "step": 12350 }, { "epoch": 2.44, "learning_rate": 6.395693401817465e-05, "loss": 0.4395, "step": 12360 }, { "epoch": 2.44, "learning_rate": 6.39075464243382e-05, "loss": 0.4137, "step": 12370 }, { "epoch": 2.45, "learning_rate": 6.385815883050178e-05, "loss": 0.4224, "step": 12380 }, { "epoch": 2.45, "learning_rate": 6.380877123666536e-05, "loss": 0.432, "step": 12390 }, { "epoch": 2.45, "learning_rate": 6.375938364282893e-05, "loss": 0.4386, "step": 12400 }, { "epoch": 2.45, "learning_rate": 6.37099960489925e-05, "loss": 0.4359, "step": 12410 }, { "epoch": 2.45, "learning_rate": 6.366060845515606e-05, "loss": 0.4532, "step": 12420 }, { "epoch": 2.46, "learning_rate": 6.361122086131964e-05, "loss": 0.4196, "step": 12430 }, { "epoch": 2.46, "learning_rate": 6.356183326748321e-05, "loss": 0.4427, "step": 12440 }, { "epoch": 2.46, "learning_rate": 6.351244567364678e-05, "loss": 0.4148, "step": 12450 }, { "epoch": 2.46, "learning_rate": 6.346305807981036e-05, "loss": 0.4365, "step": 12460 }, { "epoch": 2.46, "learning_rate": 6.341367048597394e-05, "loss": 0.462, "step": 12470 }, { "epoch": 2.47, "learning_rate": 6.336428289213749e-05, "loss": 0.423, "step": 12480 }, { "epoch": 2.47, "learning_rate": 6.331489529830107e-05, "loss": 0.4387, "step": 12490 }, { "epoch": 2.47, "learning_rate": 6.326550770446465e-05, "loss": 0.4262, "step": 12500 }, { "epoch": 2.47, "learning_rate": 6.321612011062822e-05, "loss": 0.4392, "step": 12510 }, { "epoch": 2.47, "learning_rate": 6.316673251679178e-05, "loss": 0.4169, "step": 12520 }, { "epoch": 2.47, "learning_rate": 6.311734492295536e-05, "loss": 0.4089, "step": 12530 }, { "epoch": 2.48, "learning_rate": 6.306795732911893e-05, "loss": 0.4084, "step": 12540 }, { "epoch": 2.48, "learning_rate": 6.30185697352825e-05, "loss": 0.4025, "step": 12550 }, { "epoch": 2.48, "learning_rate": 6.296918214144608e-05, "loss": 0.4114, "step": 12560 }, { "epoch": 2.48, "learning_rate": 6.291979454760964e-05, "loss": 0.4322, "step": 12570 }, { "epoch": 2.48, "learning_rate": 6.287040695377322e-05, "loss": 0.4735, "step": 12580 }, { "epoch": 2.49, "learning_rate": 6.282101935993678e-05, "loss": 0.4246, "step": 12590 }, { "epoch": 2.49, "learning_rate": 6.277163176610036e-05, "loss": 0.4018, "step": 12600 }, { "epoch": 2.49, "learning_rate": 6.272224417226394e-05, "loss": 0.4872, "step": 12610 }, { "epoch": 2.49, "learning_rate": 6.267285657842749e-05, "loss": 0.4206, "step": 12620 }, { "epoch": 2.49, "learning_rate": 6.262346898459107e-05, "loss": 0.4216, "step": 12630 }, { "epoch": 2.5, "learning_rate": 6.257408139075465e-05, "loss": 0.3996, "step": 12640 }, { "epoch": 2.5, "learning_rate": 6.252469379691822e-05, "loss": 0.4295, "step": 12650 }, { "epoch": 2.5, "learning_rate": 6.247530620308178e-05, "loss": 0.4356, "step": 12660 }, { "epoch": 2.5, "learning_rate": 6.242591860924536e-05, "loss": 0.4098, "step": 12670 }, { "epoch": 2.5, "learning_rate": 6.237653101540893e-05, "loss": 0.4221, "step": 12680 }, { "epoch": 2.51, "learning_rate": 6.232714342157251e-05, "loss": 0.4482, "step": 12690 }, { "epoch": 2.51, "learning_rate": 6.227775582773608e-05, "loss": 0.4459, "step": 12700 }, { "epoch": 2.51, "learning_rate": 6.222836823389965e-05, "loss": 0.4125, "step": 12710 }, { "epoch": 2.51, "learning_rate": 6.217898064006323e-05, "loss": 0.4238, "step": 12720 }, { "epoch": 2.51, "learning_rate": 6.212959304622679e-05, "loss": 0.4127, "step": 12730 }, { "epoch": 2.52, "learning_rate": 6.208020545239036e-05, "loss": 0.4254, "step": 12740 }, { "epoch": 2.52, "learning_rate": 6.203081785855394e-05, "loss": 0.4443, "step": 12750 }, { "epoch": 2.52, "learning_rate": 6.19814302647175e-05, "loss": 0.4644, "step": 12760 }, { "epoch": 2.52, "learning_rate": 6.193204267088107e-05, "loss": 0.4019, "step": 12770 }, { "epoch": 2.52, "learning_rate": 6.188265507704465e-05, "loss": 0.4329, "step": 12780 }, { "epoch": 2.53, "learning_rate": 6.183326748320822e-05, "loss": 0.3864, "step": 12790 }, { "epoch": 2.53, "learning_rate": 6.178387988937179e-05, "loss": 0.3482, "step": 12800 }, { "epoch": 2.53, "learning_rate": 6.173449229553537e-05, "loss": 0.4138, "step": 12810 }, { "epoch": 2.53, "learning_rate": 6.168510470169893e-05, "loss": 0.4379, "step": 12820 }, { "epoch": 2.53, "learning_rate": 6.163571710786251e-05, "loss": 0.4259, "step": 12830 }, { "epoch": 2.54, "learning_rate": 6.158632951402608e-05, "loss": 0.413, "step": 12840 }, { "epoch": 2.54, "learning_rate": 6.153694192018965e-05, "loss": 0.4407, "step": 12850 }, { "epoch": 2.54, "learning_rate": 6.148755432635323e-05, "loss": 0.4385, "step": 12860 }, { "epoch": 2.54, "learning_rate": 6.14381667325168e-05, "loss": 0.4296, "step": 12870 }, { "epoch": 2.54, "learning_rate": 6.138877913868036e-05, "loss": 0.429, "step": 12880 }, { "epoch": 2.55, "learning_rate": 6.133939154484394e-05, "loss": 0.4411, "step": 12890 }, { "epoch": 2.55, "learning_rate": 6.129000395100751e-05, "loss": 0.4304, "step": 12900 }, { "epoch": 2.55, "learning_rate": 6.124061635717107e-05, "loss": 0.4279, "step": 12910 }, { "epoch": 2.55, "learning_rate": 6.119122876333465e-05, "loss": 0.4188, "step": 12920 }, { "epoch": 2.55, "learning_rate": 6.114184116949822e-05, "loss": 0.3988, "step": 12930 }, { "epoch": 2.56, "learning_rate": 6.10924535756618e-05, "loss": 0.4214, "step": 12940 }, { "epoch": 2.56, "learning_rate": 6.104306598182537e-05, "loss": 0.4093, "step": 12950 }, { "epoch": 2.56, "learning_rate": 6.099367838798894e-05, "loss": 0.4118, "step": 12960 }, { "epoch": 2.56, "learning_rate": 6.0944290794152516e-05, "loss": 0.4042, "step": 12970 }, { "epoch": 2.56, "learning_rate": 6.0894903200316076e-05, "loss": 0.4027, "step": 12980 }, { "epoch": 2.57, "learning_rate": 6.0845515606479656e-05, "loss": 0.4342, "step": 12990 }, { "epoch": 2.57, "learning_rate": 6.079612801264323e-05, "loss": 0.4169, "step": 13000 }, { "epoch": 2.57, "learning_rate": 6.07467404188068e-05, "loss": 0.4564, "step": 13010 }, { "epoch": 2.57, "learning_rate": 6.069735282497036e-05, "loss": 0.4336, "step": 13020 }, { "epoch": 2.57, "learning_rate": 6.064796523113394e-05, "loss": 0.428, "step": 13030 }, { "epoch": 2.58, "learning_rate": 6.0598577637297517e-05, "loss": 0.3839, "step": 13040 }, { "epoch": 2.58, "learning_rate": 6.054919004346109e-05, "loss": 0.4209, "step": 13050 }, { "epoch": 2.58, "learning_rate": 6.049980244962466e-05, "loss": 0.4252, "step": 13060 }, { "epoch": 2.58, "learning_rate": 6.045041485578823e-05, "loss": 0.4015, "step": 13070 }, { "epoch": 2.58, "learning_rate": 6.0401027261951804e-05, "loss": 0.4486, "step": 13080 }, { "epoch": 2.59, "learning_rate": 6.035163966811537e-05, "loss": 0.4282, "step": 13090 }, { "epoch": 2.59, "learning_rate": 6.0302252074278944e-05, "loss": 0.4691, "step": 13100 }, { "epoch": 2.59, "learning_rate": 6.025286448044252e-05, "loss": 0.415, "step": 13110 }, { "epoch": 2.59, "learning_rate": 6.020347688660609e-05, "loss": 0.3866, "step": 13120 }, { "epoch": 2.59, "learning_rate": 6.015408929276966e-05, "loss": 0.3815, "step": 13130 }, { "epoch": 2.6, "learning_rate": 6.010470169893323e-05, "loss": 0.3996, "step": 13140 }, { "epoch": 2.6, "learning_rate": 6.0055314105096805e-05, "loss": 0.4019, "step": 13150 }, { "epoch": 2.6, "learning_rate": 6.000592651126038e-05, "loss": 0.4649, "step": 13160 }, { "epoch": 2.6, "learning_rate": 5.9956538917423945e-05, "loss": 0.4253, "step": 13170 }, { "epoch": 2.6, "learning_rate": 5.990715132358752e-05, "loss": 0.4123, "step": 13180 }, { "epoch": 2.61, "learning_rate": 5.985776372975109e-05, "loss": 0.3989, "step": 13190 }, { "epoch": 2.61, "learning_rate": 5.980837613591466e-05, "loss": 0.4311, "step": 13200 }, { "epoch": 2.61, "learning_rate": 5.975898854207823e-05, "loss": 0.4318, "step": 13210 }, { "epoch": 2.61, "learning_rate": 5.9709600948241806e-05, "loss": 0.4704, "step": 13220 }, { "epoch": 2.61, "learning_rate": 5.966021335440538e-05, "loss": 0.4169, "step": 13230 }, { "epoch": 2.62, "learning_rate": 5.9610825760568946e-05, "loss": 0.4656, "step": 13240 }, { "epoch": 2.62, "learning_rate": 5.956143816673252e-05, "loss": 0.4127, "step": 13250 }, { "epoch": 2.62, "learning_rate": 5.951205057289609e-05, "loss": 0.422, "step": 13260 }, { "epoch": 2.62, "learning_rate": 5.946266297905966e-05, "loss": 0.4139, "step": 13270 }, { "epoch": 2.62, "learning_rate": 5.941327538522323e-05, "loss": 0.4202, "step": 13280 }, { "epoch": 2.63, "learning_rate": 5.9363887791386806e-05, "loss": 0.441, "step": 13290 }, { "epoch": 2.63, "learning_rate": 5.931450019755038e-05, "loss": 0.4289, "step": 13300 }, { "epoch": 2.63, "learning_rate": 5.926511260371395e-05, "loss": 0.4173, "step": 13310 }, { "epoch": 2.63, "learning_rate": 5.921572500987752e-05, "loss": 0.4481, "step": 13320 }, { "epoch": 2.63, "learning_rate": 5.9166337416041094e-05, "loss": 0.3799, "step": 13330 }, { "epoch": 2.63, "learning_rate": 5.911694982220467e-05, "loss": 0.3955, "step": 13340 }, { "epoch": 2.64, "learning_rate": 5.9067562228368234e-05, "loss": 0.4377, "step": 13350 }, { "epoch": 2.64, "learning_rate": 5.901817463453181e-05, "loss": 0.3976, "step": 13360 }, { "epoch": 2.64, "learning_rate": 5.896878704069538e-05, "loss": 0.425, "step": 13370 }, { "epoch": 2.64, "learning_rate": 5.891939944685895e-05, "loss": 0.4147, "step": 13380 }, { "epoch": 2.64, "learning_rate": 5.887001185302252e-05, "loss": 0.4448, "step": 13390 }, { "epoch": 2.65, "learning_rate": 5.8820624259186095e-05, "loss": 0.4123, "step": 13400 }, { "epoch": 2.65, "learning_rate": 5.8771236665349675e-05, "loss": 0.4351, "step": 13410 }, { "epoch": 2.65, "learning_rate": 5.8721849071513235e-05, "loss": 0.3927, "step": 13420 }, { "epoch": 2.65, "learning_rate": 5.867246147767681e-05, "loss": 0.4087, "step": 13430 }, { "epoch": 2.65, "learning_rate": 5.862307388384038e-05, "loss": 0.437, "step": 13440 }, { "epoch": 2.66, "learning_rate": 5.857368629000395e-05, "loss": 0.4434, "step": 13450 }, { "epoch": 2.66, "learning_rate": 5.852429869616752e-05, "loss": 0.452, "step": 13460 }, { "epoch": 2.66, "learning_rate": 5.8474911102331095e-05, "loss": 0.4226, "step": 13470 }, { "epoch": 2.66, "learning_rate": 5.8425523508494676e-05, "loss": 0.4446, "step": 13480 }, { "epoch": 2.66, "learning_rate": 5.8376135914658236e-05, "loss": 0.4026, "step": 13490 }, { "epoch": 2.67, "learning_rate": 5.832674832082181e-05, "loss": 0.4248, "step": 13500 }, { "epoch": 2.67, "learning_rate": 5.827736072698538e-05, "loss": 0.4419, "step": 13510 }, { "epoch": 2.67, "learning_rate": 5.822797313314896e-05, "loss": 0.4275, "step": 13520 }, { "epoch": 2.67, "learning_rate": 5.817858553931252e-05, "loss": 0.4187, "step": 13530 }, { "epoch": 2.67, "learning_rate": 5.8129197945476096e-05, "loss": 0.4089, "step": 13540 }, { "epoch": 2.68, "learning_rate": 5.807981035163968e-05, "loss": 0.4586, "step": 13550 }, { "epoch": 2.68, "learning_rate": 5.803042275780324e-05, "loss": 0.4524, "step": 13560 }, { "epoch": 2.68, "learning_rate": 5.798103516396681e-05, "loss": 0.4284, "step": 13570 }, { "epoch": 2.68, "learning_rate": 5.793164757013039e-05, "loss": 0.4006, "step": 13580 }, { "epoch": 2.68, "learning_rate": 5.7882259976293964e-05, "loss": 0.3958, "step": 13590 }, { "epoch": 2.69, "learning_rate": 5.7832872382457524e-05, "loss": 0.4378, "step": 13600 }, { "epoch": 2.69, "learning_rate": 5.77834847886211e-05, "loss": 0.4391, "step": 13610 }, { "epoch": 2.69, "learning_rate": 5.773409719478468e-05, "loss": 0.4126, "step": 13620 }, { "epoch": 2.69, "learning_rate": 5.768470960094824e-05, "loss": 0.441, "step": 13630 }, { "epoch": 2.69, "learning_rate": 5.763532200711181e-05, "loss": 0.3825, "step": 13640 }, { "epoch": 2.7, "learning_rate": 5.758593441327539e-05, "loss": 0.409, "step": 13650 }, { "epoch": 2.7, "learning_rate": 5.7536546819438965e-05, "loss": 0.4022, "step": 13660 }, { "epoch": 2.7, "learning_rate": 5.7487159225602525e-05, "loss": 0.3988, "step": 13670 }, { "epoch": 2.7, "learning_rate": 5.7437771631766105e-05, "loss": 0.4181, "step": 13680 }, { "epoch": 2.7, "learning_rate": 5.738838403792968e-05, "loss": 0.4183, "step": 13690 }, { "epoch": 2.71, "learning_rate": 5.733899644409325e-05, "loss": 0.4215, "step": 13700 }, { "epoch": 2.71, "learning_rate": 5.728960885025681e-05, "loss": 0.4224, "step": 13710 }, { "epoch": 2.71, "learning_rate": 5.724022125642039e-05, "loss": 0.4139, "step": 13720 }, { "epoch": 2.71, "learning_rate": 5.7190833662583966e-05, "loss": 0.4011, "step": 13730 }, { "epoch": 2.71, "learning_rate": 5.7141446068747526e-05, "loss": 0.4197, "step": 13740 }, { "epoch": 2.72, "learning_rate": 5.7092058474911106e-05, "loss": 0.4358, "step": 13750 }, { "epoch": 2.72, "learning_rate": 5.704267088107468e-05, "loss": 0.4496, "step": 13760 }, { "epoch": 2.72, "learning_rate": 5.699328328723825e-05, "loss": 0.4151, "step": 13770 }, { "epoch": 2.72, "learning_rate": 5.694389569340182e-05, "loss": 0.4136, "step": 13780 }, { "epoch": 2.72, "learning_rate": 5.689450809956539e-05, "loss": 0.4124, "step": 13790 }, { "epoch": 2.73, "learning_rate": 5.684512050572897e-05, "loss": 0.3851, "step": 13800 }, { "epoch": 2.73, "learning_rate": 5.679573291189254e-05, "loss": 0.4061, "step": 13810 }, { "epoch": 2.73, "learning_rate": 5.674634531805611e-05, "loss": 0.3556, "step": 13820 }, { "epoch": 2.73, "learning_rate": 5.669695772421968e-05, "loss": 0.4147, "step": 13830 }, { "epoch": 2.73, "learning_rate": 5.6647570130383254e-05, "loss": 0.429, "step": 13840 }, { "epoch": 2.74, "learning_rate": 5.659818253654682e-05, "loss": 0.411, "step": 13850 }, { "epoch": 2.74, "learning_rate": 5.6548794942710394e-05, "loss": 0.4436, "step": 13860 }, { "epoch": 2.74, "learning_rate": 5.649940734887397e-05, "loss": 0.4155, "step": 13870 }, { "epoch": 2.74, "learning_rate": 5.645001975503754e-05, "loss": 0.4051, "step": 13880 }, { "epoch": 2.74, "learning_rate": 5.640063216120111e-05, "loss": 0.391, "step": 13890 }, { "epoch": 2.75, "learning_rate": 5.635124456736468e-05, "loss": 0.3718, "step": 13900 }, { "epoch": 2.75, "learning_rate": 5.6301856973528255e-05, "loss": 0.4108, "step": 13910 }, { "epoch": 2.75, "learning_rate": 5.625246937969182e-05, "loss": 0.3909, "step": 13920 }, { "epoch": 2.75, "learning_rate": 5.6203081785855395e-05, "loss": 0.3991, "step": 13930 }, { "epoch": 2.75, "learning_rate": 5.615369419201897e-05, "loss": 0.4356, "step": 13940 }, { "epoch": 2.76, "learning_rate": 5.610430659818254e-05, "loss": 0.3949, "step": 13950 }, { "epoch": 2.76, "learning_rate": 5.605491900434611e-05, "loss": 0.3956, "step": 13960 }, { "epoch": 2.76, "learning_rate": 5.600553141050968e-05, "loss": 0.4562, "step": 13970 }, { "epoch": 2.76, "learning_rate": 5.5956143816673256e-05, "loss": 0.385, "step": 13980 }, { "epoch": 2.76, "learning_rate": 5.590675622283683e-05, "loss": 0.3809, "step": 13990 }, { "epoch": 2.77, "learning_rate": 5.5857368629000396e-05, "loss": 0.4266, "step": 14000 }, { "epoch": 2.77, "learning_rate": 5.580798103516397e-05, "loss": 0.4042, "step": 14010 }, { "epoch": 2.77, "learning_rate": 5.575859344132754e-05, "loss": 0.3825, "step": 14020 }, { "epoch": 2.77, "learning_rate": 5.570920584749111e-05, "loss": 0.4056, "step": 14030 }, { "epoch": 2.77, "learning_rate": 5.565981825365468e-05, "loss": 0.4277, "step": 14040 }, { "epoch": 2.78, "learning_rate": 5.5610430659818257e-05, "loss": 0.3771, "step": 14050 }, { "epoch": 2.78, "learning_rate": 5.556104306598183e-05, "loss": 0.4125, "step": 14060 }, { "epoch": 2.78, "learning_rate": 5.55116554721454e-05, "loss": 0.4124, "step": 14070 }, { "epoch": 2.78, "learning_rate": 5.546226787830897e-05, "loss": 0.4217, "step": 14080 }, { "epoch": 2.78, "learning_rate": 5.5412880284472544e-05, "loss": 0.4127, "step": 14090 }, { "epoch": 2.79, "learning_rate": 5.536349269063611e-05, "loss": 0.4032, "step": 14100 }, { "epoch": 2.79, "learning_rate": 5.5314105096799684e-05, "loss": 0.4126, "step": 14110 }, { "epoch": 2.79, "learning_rate": 5.526471750296326e-05, "loss": 0.3737, "step": 14120 }, { "epoch": 2.79, "learning_rate": 5.521532990912683e-05, "loss": 0.413, "step": 14130 }, { "epoch": 2.79, "learning_rate": 5.51659423152904e-05, "loss": 0.416, "step": 14140 }, { "epoch": 2.79, "learning_rate": 5.511655472145397e-05, "loss": 0.4142, "step": 14150 }, { "epoch": 2.8, "learning_rate": 5.5067167127617545e-05, "loss": 0.4011, "step": 14160 }, { "epoch": 2.8, "learning_rate": 5.5017779533781125e-05, "loss": 0.4222, "step": 14170 }, { "epoch": 2.8, "learning_rate": 5.4968391939944685e-05, "loss": 0.3955, "step": 14180 }, { "epoch": 2.8, "learning_rate": 5.491900434610826e-05, "loss": 0.4008, "step": 14190 }, { "epoch": 2.8, "learning_rate": 5.486961675227183e-05, "loss": 0.3979, "step": 14200 }, { "epoch": 2.81, "learning_rate": 5.48202291584354e-05, "loss": 0.4192, "step": 14210 }, { "epoch": 2.81, "learning_rate": 5.477084156459897e-05, "loss": 0.4114, "step": 14220 }, { "epoch": 2.81, "learning_rate": 5.4721453970762546e-05, "loss": 0.4553, "step": 14230 }, { "epoch": 2.81, "learning_rate": 5.4672066376926126e-05, "loss": 0.4052, "step": 14240 }, { "epoch": 2.81, "learning_rate": 5.4622678783089686e-05, "loss": 0.4359, "step": 14250 }, { "epoch": 2.82, "learning_rate": 5.457329118925326e-05, "loss": 0.4314, "step": 14260 }, { "epoch": 2.82, "learning_rate": 5.452390359541684e-05, "loss": 0.4177, "step": 14270 }, { "epoch": 2.82, "learning_rate": 5.44745160015804e-05, "loss": 0.4357, "step": 14280 }, { "epoch": 2.82, "learning_rate": 5.442512840774397e-05, "loss": 0.3843, "step": 14290 }, { "epoch": 2.82, "learning_rate": 5.4375740813907547e-05, "loss": 0.3932, "step": 14300 }, { "epoch": 2.83, "learning_rate": 5.432635322007113e-05, "loss": 0.4185, "step": 14310 }, { "epoch": 2.83, "learning_rate": 5.427696562623469e-05, "loss": 0.4298, "step": 14320 }, { "epoch": 2.83, "learning_rate": 5.422757803239826e-05, "loss": 0.4111, "step": 14330 }, { "epoch": 2.83, "learning_rate": 5.417819043856184e-05, "loss": 0.4448, "step": 14340 }, { "epoch": 2.83, "learning_rate": 5.4128802844725414e-05, "loss": 0.4313, "step": 14350 }, { "epoch": 2.84, "learning_rate": 5.4079415250888974e-05, "loss": 0.4421, "step": 14360 }, { "epoch": 2.84, "learning_rate": 5.4030027657052554e-05, "loss": 0.4264, "step": 14370 }, { "epoch": 2.84, "learning_rate": 5.398064006321613e-05, "loss": 0.4453, "step": 14380 }, { "epoch": 2.84, "learning_rate": 5.393125246937969e-05, "loss": 0.3988, "step": 14390 }, { "epoch": 2.84, "learning_rate": 5.388186487554326e-05, "loss": 0.4341, "step": 14400 }, { "epoch": 2.85, "learning_rate": 5.383247728170684e-05, "loss": 0.4488, "step": 14410 }, { "epoch": 2.85, "learning_rate": 5.3783089687870415e-05, "loss": 0.3876, "step": 14420 }, { "epoch": 2.85, "learning_rate": 5.3733702094033975e-05, "loss": 0.4479, "step": 14430 }, { "epoch": 2.85, "learning_rate": 5.3684314500197555e-05, "loss": 0.427, "step": 14440 }, { "epoch": 2.85, "learning_rate": 5.363492690636113e-05, "loss": 0.4068, "step": 14450 }, { "epoch": 2.86, "learning_rate": 5.35855393125247e-05, "loss": 0.4129, "step": 14460 }, { "epoch": 2.86, "learning_rate": 5.353615171868827e-05, "loss": 0.4137, "step": 14470 }, { "epoch": 2.86, "learning_rate": 5.348676412485184e-05, "loss": 0.4154, "step": 14480 }, { "epoch": 2.86, "learning_rate": 5.3437376531015416e-05, "loss": 0.3544, "step": 14490 }, { "epoch": 2.86, "learning_rate": 5.3387988937178976e-05, "loss": 0.4694, "step": 14500 }, { "epoch": 2.87, "learning_rate": 5.3338601343342556e-05, "loss": 0.4371, "step": 14510 }, { "epoch": 2.87, "learning_rate": 5.328921374950613e-05, "loss": 0.4016, "step": 14520 }, { "epoch": 2.87, "learning_rate": 5.32398261556697e-05, "loss": 0.4011, "step": 14530 }, { "epoch": 2.87, "learning_rate": 5.319043856183327e-05, "loss": 0.4195, "step": 14540 }, { "epoch": 2.87, "learning_rate": 5.314105096799684e-05, "loss": 0.3984, "step": 14550 }, { "epoch": 2.88, "learning_rate": 5.309166337416042e-05, "loss": 0.3978, "step": 14560 }, { "epoch": 2.88, "learning_rate": 5.304227578032398e-05, "loss": 0.4238, "step": 14570 }, { "epoch": 2.88, "learning_rate": 5.299288818648756e-05, "loss": 0.3793, "step": 14580 }, { "epoch": 2.88, "learning_rate": 5.294350059265113e-05, "loss": 0.4108, "step": 14590 }, { "epoch": 2.88, "learning_rate": 5.2894112998814704e-05, "loss": 0.4094, "step": 14600 }, { "epoch": 2.89, "learning_rate": 5.284472540497827e-05, "loss": 0.3884, "step": 14610 }, { "epoch": 2.89, "learning_rate": 5.2795337811141844e-05, "loss": 0.4426, "step": 14620 }, { "epoch": 2.89, "learning_rate": 5.274595021730542e-05, "loss": 0.3882, "step": 14630 }, { "epoch": 2.89, "learning_rate": 5.269656262346899e-05, "loss": 0.4311, "step": 14640 }, { "epoch": 2.89, "learning_rate": 5.264717502963256e-05, "loss": 0.4436, "step": 14650 }, { "epoch": 2.9, "learning_rate": 5.259778743579613e-05, "loss": 0.4572, "step": 14660 }, { "epoch": 2.9, "learning_rate": 5.2548399841959705e-05, "loss": 0.4099, "step": 14670 }, { "epoch": 2.9, "learning_rate": 5.249901224812327e-05, "loss": 0.3856, "step": 14680 }, { "epoch": 2.9, "learning_rate": 5.2449624654286845e-05, "loss": 0.4118, "step": 14690 }, { "epoch": 2.9, "learning_rate": 5.240023706045042e-05, "loss": 0.4544, "step": 14700 }, { "epoch": 2.91, "learning_rate": 5.235084946661399e-05, "loss": 0.4207, "step": 14710 }, { "epoch": 2.91, "learning_rate": 5.230146187277756e-05, "loss": 0.4185, "step": 14720 }, { "epoch": 2.91, "learning_rate": 5.225207427894113e-05, "loss": 0.4108, "step": 14730 }, { "epoch": 2.91, "learning_rate": 5.2202686685104706e-05, "loss": 0.3965, "step": 14740 }, { "epoch": 2.91, "learning_rate": 5.215329909126827e-05, "loss": 0.4176, "step": 14750 }, { "epoch": 2.92, "learning_rate": 5.2103911497431846e-05, "loss": 0.405, "step": 14760 }, { "epoch": 2.92, "learning_rate": 5.205452390359542e-05, "loss": 0.3756, "step": 14770 }, { "epoch": 2.92, "learning_rate": 5.200513630975899e-05, "loss": 0.3911, "step": 14780 }, { "epoch": 2.92, "learning_rate": 5.195574871592256e-05, "loss": 0.3586, "step": 14790 }, { "epoch": 2.92, "learning_rate": 5.190636112208613e-05, "loss": 0.4117, "step": 14800 }, { "epoch": 2.93, "learning_rate": 5.185697352824971e-05, "loss": 0.3841, "step": 14810 }, { "epoch": 2.93, "learning_rate": 5.180758593441328e-05, "loss": 0.4003, "step": 14820 }, { "epoch": 2.93, "learning_rate": 5.175819834057685e-05, "loss": 0.3773, "step": 14830 }, { "epoch": 2.93, "learning_rate": 5.170881074674042e-05, "loss": 0.4073, "step": 14840 }, { "epoch": 2.93, "learning_rate": 5.1659423152903994e-05, "loss": 0.408, "step": 14850 }, { "epoch": 2.94, "learning_rate": 5.161003555906756e-05, "loss": 0.4132, "step": 14860 }, { "epoch": 2.94, "learning_rate": 5.1560647965231134e-05, "loss": 0.394, "step": 14870 }, { "epoch": 2.94, "learning_rate": 5.151126037139471e-05, "loss": 0.4014, "step": 14880 }, { "epoch": 2.94, "learning_rate": 5.146187277755828e-05, "loss": 0.4092, "step": 14890 }, { "epoch": 2.94, "learning_rate": 5.141248518372185e-05, "loss": 0.3545, "step": 14900 }, { "epoch": 2.95, "learning_rate": 5.136309758988542e-05, "loss": 0.409, "step": 14910 }, { "epoch": 2.95, "learning_rate": 5.1313709996048995e-05, "loss": 0.4186, "step": 14920 }, { "epoch": 2.95, "learning_rate": 5.126432240221256e-05, "loss": 0.375, "step": 14930 }, { "epoch": 2.95, "learning_rate": 5.1214934808376135e-05, "loss": 0.4, "step": 14940 }, { "epoch": 2.95, "learning_rate": 5.116554721453971e-05, "loss": 0.4022, "step": 14950 }, { "epoch": 2.95, "learning_rate": 5.111615962070329e-05, "loss": 0.3847, "step": 14960 }, { "epoch": 2.96, "learning_rate": 5.106677202686685e-05, "loss": 0.3834, "step": 14970 }, { "epoch": 2.96, "learning_rate": 5.101738443303042e-05, "loss": 0.4265, "step": 14980 }, { "epoch": 2.96, "learning_rate": 5.0967996839193996e-05, "loss": 0.4009, "step": 14990 }, { "epoch": 2.96, "learning_rate": 5.0918609245357576e-05, "loss": 0.444, "step": 15000 }, { "epoch": 2.96, "learning_rate": 5.0869221651521136e-05, "loss": 0.3884, "step": 15010 }, { "epoch": 2.97, "learning_rate": 5.081983405768471e-05, "loss": 0.4156, "step": 15020 }, { "epoch": 2.97, "learning_rate": 5.077044646384829e-05, "loss": 0.4297, "step": 15030 }, { "epoch": 2.97, "learning_rate": 5.072105887001185e-05, "loss": 0.382, "step": 15040 }, { "epoch": 2.97, "learning_rate": 5.067167127617542e-05, "loss": 0.4082, "step": 15050 }, { "epoch": 2.97, "learning_rate": 5.0622283682339003e-05, "loss": 0.3724, "step": 15060 }, { "epoch": 2.98, "learning_rate": 5.057289608850258e-05, "loss": 0.4088, "step": 15070 }, { "epoch": 2.98, "learning_rate": 5.052350849466614e-05, "loss": 0.383, "step": 15080 }, { "epoch": 2.98, "learning_rate": 5.047412090082971e-05, "loss": 0.3427, "step": 15090 }, { "epoch": 2.98, "learning_rate": 5.042473330699329e-05, "loss": 0.3816, "step": 15100 }, { "epoch": 2.98, "learning_rate": 5.0375345713156864e-05, "loss": 0.3817, "step": 15110 }, { "epoch": 2.99, "learning_rate": 5.0325958119320424e-05, "loss": 0.3915, "step": 15120 }, { "epoch": 2.99, "learning_rate": 5.0276570525484004e-05, "loss": 0.4578, "step": 15130 }, { "epoch": 2.99, "learning_rate": 5.022718293164758e-05, "loss": 0.3954, "step": 15140 }, { "epoch": 2.99, "learning_rate": 5.017779533781114e-05, "loss": 0.4024, "step": 15150 }, { "epoch": 2.99, "learning_rate": 5.012840774397471e-05, "loss": 0.3851, "step": 15160 }, { "epoch": 3.0, "learning_rate": 5.007902015013829e-05, "loss": 0.3971, "step": 15170 }, { "epoch": 3.0, "learning_rate": 5.0029632556301865e-05, "loss": 0.4095, "step": 15180 }, { "epoch": 3.0, "learning_rate": 4.998024496246543e-05, "loss": 0.4094, "step": 15190 }, { "epoch": 3.0, "learning_rate": 4.9930857368629005e-05, "loss": 0.3296, "step": 15200 }, { "epoch": 3.0, "learning_rate": 4.988146977479257e-05, "loss": 0.2881, "step": 15210 }, { "epoch": 3.01, "learning_rate": 4.9832082180956146e-05, "loss": 0.3069, "step": 15220 }, { "epoch": 3.01, "learning_rate": 4.978269458711972e-05, "loss": 0.2996, "step": 15230 }, { "epoch": 3.01, "learning_rate": 4.973330699328329e-05, "loss": 0.2881, "step": 15240 }, { "epoch": 3.01, "learning_rate": 4.968391939944686e-05, "loss": 0.2905, "step": 15250 }, { "epoch": 3.01, "learning_rate": 4.963453180561043e-05, "loss": 0.3136, "step": 15260 }, { "epoch": 3.02, "learning_rate": 4.9585144211774006e-05, "loss": 0.3475, "step": 15270 }, { "epoch": 3.02, "learning_rate": 4.953575661793758e-05, "loss": 0.2878, "step": 15280 }, { "epoch": 3.02, "learning_rate": 4.9486369024101146e-05, "loss": 0.321, "step": 15290 }, { "epoch": 3.02, "learning_rate": 4.943698143026472e-05, "loss": 0.3034, "step": 15300 }, { "epoch": 3.02, "learning_rate": 4.9387593836428293e-05, "loss": 0.3253, "step": 15310 }, { "epoch": 3.03, "learning_rate": 4.933820624259186e-05, "loss": 0.2929, "step": 15320 }, { "epoch": 3.03, "learning_rate": 4.9288818648755434e-05, "loss": 0.2879, "step": 15330 }, { "epoch": 3.03, "learning_rate": 4.923943105491901e-05, "loss": 0.3079, "step": 15340 }, { "epoch": 3.03, "learning_rate": 4.919004346108258e-05, "loss": 0.3179, "step": 15350 }, { "epoch": 3.03, "learning_rate": 4.914065586724615e-05, "loss": 0.3148, "step": 15360 }, { "epoch": 3.04, "learning_rate": 4.909126827340972e-05, "loss": 0.2789, "step": 15370 }, { "epoch": 3.04, "learning_rate": 4.9041880679573294e-05, "loss": 0.3012, "step": 15380 }, { "epoch": 3.04, "learning_rate": 4.899249308573686e-05, "loss": 0.3162, "step": 15390 }, { "epoch": 3.04, "learning_rate": 4.8943105491900435e-05, "loss": 0.2971, "step": 15400 }, { "epoch": 3.04, "learning_rate": 4.889371789806401e-05, "loss": 0.3091, "step": 15410 }, { "epoch": 3.05, "learning_rate": 4.884433030422758e-05, "loss": 0.2959, "step": 15420 }, { "epoch": 3.05, "learning_rate": 4.879494271039115e-05, "loss": 0.3013, "step": 15430 }, { "epoch": 3.05, "learning_rate": 4.874555511655473e-05, "loss": 0.2888, "step": 15440 }, { "epoch": 3.05, "learning_rate": 4.8696167522718295e-05, "loss": 0.2972, "step": 15450 }, { "epoch": 3.05, "learning_rate": 4.864677992888187e-05, "loss": 0.3125, "step": 15460 }, { "epoch": 3.06, "learning_rate": 4.8597392335045435e-05, "loss": 0.2928, "step": 15470 }, { "epoch": 3.06, "learning_rate": 4.8548004741209016e-05, "loss": 0.3053, "step": 15480 }, { "epoch": 3.06, "learning_rate": 4.849861714737258e-05, "loss": 0.3214, "step": 15490 }, { "epoch": 3.06, "learning_rate": 4.844922955353615e-05, "loss": 0.3178, "step": 15500 }, { "epoch": 3.06, "learning_rate": 4.839984195969973e-05, "loss": 0.2788, "step": 15510 }, { "epoch": 3.07, "learning_rate": 4.8350454365863296e-05, "loss": 0.3009, "step": 15520 }, { "epoch": 3.07, "learning_rate": 4.830106677202687e-05, "loss": 0.2969, "step": 15530 }, { "epoch": 3.07, "learning_rate": 4.825167917819044e-05, "loss": 0.3172, "step": 15540 }, { "epoch": 3.07, "learning_rate": 4.820229158435402e-05, "loss": 0.2936, "step": 15550 }, { "epoch": 3.07, "learning_rate": 4.815290399051758e-05, "loss": 0.3046, "step": 15560 }, { "epoch": 3.08, "learning_rate": 4.810351639668116e-05, "loss": 0.3451, "step": 15570 }, { "epoch": 3.08, "learning_rate": 4.805412880284473e-05, "loss": 0.2999, "step": 15580 }, { "epoch": 3.08, "learning_rate": 4.80047412090083e-05, "loss": 0.3081, "step": 15590 }, { "epoch": 3.08, "learning_rate": 4.795535361517187e-05, "loss": 0.3065, "step": 15600 }, { "epoch": 3.08, "learning_rate": 4.7905966021335444e-05, "loss": 0.2653, "step": 15610 }, { "epoch": 3.09, "learning_rate": 4.785657842749902e-05, "loss": 0.2908, "step": 15620 }, { "epoch": 3.09, "learning_rate": 4.7807190833662584e-05, "loss": 0.3376, "step": 15630 }, { "epoch": 3.09, "learning_rate": 4.775780323982616e-05, "loss": 0.3216, "step": 15640 }, { "epoch": 3.09, "learning_rate": 4.770841564598973e-05, "loss": 0.3099, "step": 15650 }, { "epoch": 3.09, "learning_rate": 4.7659028052153305e-05, "loss": 0.3028, "step": 15660 }, { "epoch": 3.1, "learning_rate": 4.760964045831687e-05, "loss": 0.302, "step": 15670 }, { "epoch": 3.1, "learning_rate": 4.7560252864480445e-05, "loss": 0.3128, "step": 15680 }, { "epoch": 3.1, "learning_rate": 4.751086527064402e-05, "loss": 0.3148, "step": 15690 }, { "epoch": 3.1, "learning_rate": 4.7461477676807585e-05, "loss": 0.3171, "step": 15700 }, { "epoch": 3.1, "learning_rate": 4.741209008297116e-05, "loss": 0.2928, "step": 15710 }, { "epoch": 3.11, "learning_rate": 4.736270248913473e-05, "loss": 0.3024, "step": 15720 }, { "epoch": 3.11, "learning_rate": 4.7313314895298306e-05, "loss": 0.2786, "step": 15730 }, { "epoch": 3.11, "learning_rate": 4.726392730146187e-05, "loss": 0.3194, "step": 15740 }, { "epoch": 3.11, "learning_rate": 4.7214539707625446e-05, "loss": 0.2984, "step": 15750 }, { "epoch": 3.11, "learning_rate": 4.716515211378902e-05, "loss": 0.3068, "step": 15760 }, { "epoch": 3.11, "learning_rate": 4.7115764519952586e-05, "loss": 0.3203, "step": 15770 }, { "epoch": 3.12, "learning_rate": 4.706637692611616e-05, "loss": 0.3129, "step": 15780 }, { "epoch": 3.12, "learning_rate": 4.701698933227973e-05, "loss": 0.2999, "step": 15790 }, { "epoch": 3.12, "learning_rate": 4.6967601738443307e-05, "loss": 0.3214, "step": 15800 }, { "epoch": 3.12, "learning_rate": 4.691821414460687e-05, "loss": 0.3347, "step": 15810 }, { "epoch": 3.12, "learning_rate": 4.6868826550770454e-05, "loss": 0.3023, "step": 15820 }, { "epoch": 3.13, "learning_rate": 4.681943895693402e-05, "loss": 0.3318, "step": 15830 }, { "epoch": 3.13, "learning_rate": 4.6770051363097594e-05, "loss": 0.3104, "step": 15840 }, { "epoch": 3.13, "learning_rate": 4.672066376926116e-05, "loss": 0.311, "step": 15850 }, { "epoch": 3.13, "learning_rate": 4.6671276175424734e-05, "loss": 0.2847, "step": 15860 }, { "epoch": 3.13, "learning_rate": 4.662188858158831e-05, "loss": 0.2851, "step": 15870 }, { "epoch": 3.14, "learning_rate": 4.6572500987751874e-05, "loss": 0.3261, "step": 15880 }, { "epoch": 3.14, "learning_rate": 4.6523113393915455e-05, "loss": 0.293, "step": 15890 }, { "epoch": 3.14, "learning_rate": 4.647372580007902e-05, "loss": 0.3195, "step": 15900 }, { "epoch": 3.14, "learning_rate": 4.6424338206242595e-05, "loss": 0.3181, "step": 15910 }, { "epoch": 3.14, "learning_rate": 4.637495061240617e-05, "loss": 0.2906, "step": 15920 }, { "epoch": 3.15, "learning_rate": 4.632556301856974e-05, "loss": 0.3198, "step": 15930 }, { "epoch": 3.15, "learning_rate": 4.627617542473331e-05, "loss": 0.2868, "step": 15940 }, { "epoch": 3.15, "learning_rate": 4.6226787830896875e-05, "loss": 0.3022, "step": 15950 }, { "epoch": 3.15, "learning_rate": 4.6177400237060455e-05, "loss": 0.3057, "step": 15960 }, { "epoch": 3.15, "learning_rate": 4.612801264322402e-05, "loss": 0.3087, "step": 15970 }, { "epoch": 3.16, "learning_rate": 4.6078625049387596e-05, "loss": 0.3242, "step": 15980 }, { "epoch": 3.16, "learning_rate": 4.602923745555117e-05, "loss": 0.3266, "step": 15990 }, { "epoch": 3.16, "learning_rate": 4.597984986171474e-05, "loss": 0.3334, "step": 16000 }, { "epoch": 3.16, "learning_rate": 4.593046226787831e-05, "loss": 0.3177, "step": 16010 }, { "epoch": 3.16, "learning_rate": 4.588107467404188e-05, "loss": 0.2657, "step": 16020 }, { "epoch": 3.17, "learning_rate": 4.5831687080205456e-05, "loss": 0.2846, "step": 16030 }, { "epoch": 3.17, "learning_rate": 4.578229948636902e-05, "loss": 0.3635, "step": 16040 }, { "epoch": 3.17, "learning_rate": 4.5732911892532597e-05, "loss": 0.2876, "step": 16050 }, { "epoch": 3.17, "learning_rate": 4.568352429869617e-05, "loss": 0.3235, "step": 16060 }, { "epoch": 3.17, "learning_rate": 4.5634136704859744e-05, "loss": 0.3077, "step": 16070 }, { "epoch": 3.18, "learning_rate": 4.558474911102331e-05, "loss": 0.3249, "step": 16080 }, { "epoch": 3.18, "learning_rate": 4.5535361517186884e-05, "loss": 0.3016, "step": 16090 }, { "epoch": 3.18, "learning_rate": 4.548597392335046e-05, "loss": 0.2902, "step": 16100 }, { "epoch": 3.18, "learning_rate": 4.543658632951403e-05, "loss": 0.3181, "step": 16110 }, { "epoch": 3.18, "learning_rate": 4.53871987356776e-05, "loss": 0.3215, "step": 16120 }, { "epoch": 3.19, "learning_rate": 4.533781114184118e-05, "loss": 0.2952, "step": 16130 }, { "epoch": 3.19, "learning_rate": 4.5288423548004744e-05, "loss": 0.3118, "step": 16140 }, { "epoch": 3.19, "learning_rate": 4.523903595416831e-05, "loss": 0.3244, "step": 16150 }, { "epoch": 3.19, "learning_rate": 4.5189648360331885e-05, "loss": 0.3102, "step": 16160 }, { "epoch": 3.19, "learning_rate": 4.514026076649546e-05, "loss": 0.3261, "step": 16170 }, { "epoch": 3.2, "learning_rate": 4.509087317265903e-05, "loss": 0.309, "step": 16180 }, { "epoch": 3.2, "learning_rate": 4.50414855788226e-05, "loss": 0.3049, "step": 16190 }, { "epoch": 3.2, "learning_rate": 4.499209798498618e-05, "loss": 0.2866, "step": 16200 }, { "epoch": 3.2, "learning_rate": 4.4942710391149745e-05, "loss": 0.3178, "step": 16210 }, { "epoch": 3.2, "learning_rate": 4.489332279731332e-05, "loss": 0.3131, "step": 16220 }, { "epoch": 3.21, "learning_rate": 4.4843935203476886e-05, "loss": 0.2704, "step": 16230 }, { "epoch": 3.21, "learning_rate": 4.479454760964046e-05, "loss": 0.3098, "step": 16240 }, { "epoch": 3.21, "learning_rate": 4.474516001580403e-05, "loss": 0.3088, "step": 16250 }, { "epoch": 3.21, "learning_rate": 4.46957724219676e-05, "loss": 0.3023, "step": 16260 }, { "epoch": 3.21, "learning_rate": 4.465132358751482e-05, "loss": 0.3216, "step": 16270 }, { "epoch": 3.22, "learning_rate": 4.460193599367839e-05, "loss": 0.3198, "step": 16280 }, { "epoch": 3.22, "learning_rate": 4.4552548399841966e-05, "loss": 0.2943, "step": 16290 }, { "epoch": 3.22, "learning_rate": 4.450316080600553e-05, "loss": 0.3127, "step": 16300 }, { "epoch": 3.22, "learning_rate": 4.4453773212169106e-05, "loss": 0.2966, "step": 16310 }, { "epoch": 3.22, "learning_rate": 4.440438561833268e-05, "loss": 0.2986, "step": 16320 }, { "epoch": 3.23, "learning_rate": 4.4354998024496246e-05, "loss": 0.3048, "step": 16330 }, { "epoch": 3.23, "learning_rate": 4.430561043065982e-05, "loss": 0.3139, "step": 16340 }, { "epoch": 3.23, "learning_rate": 4.425622283682339e-05, "loss": 0.2993, "step": 16350 }, { "epoch": 3.23, "learning_rate": 4.420683524298697e-05, "loss": 0.3145, "step": 16360 }, { "epoch": 3.23, "learning_rate": 4.4157447649150533e-05, "loss": 0.3179, "step": 16370 }, { "epoch": 3.24, "learning_rate": 4.410806005531411e-05, "loss": 0.2933, "step": 16380 }, { "epoch": 3.24, "learning_rate": 4.405867246147768e-05, "loss": 0.2915, "step": 16390 }, { "epoch": 3.24, "learning_rate": 4.400928486764125e-05, "loss": 0.2924, "step": 16400 }, { "epoch": 3.24, "learning_rate": 4.395989727380482e-05, "loss": 0.3115, "step": 16410 }, { "epoch": 3.24, "learning_rate": 4.3910509679968394e-05, "loss": 0.3096, "step": 16420 }, { "epoch": 3.25, "learning_rate": 4.386112208613197e-05, "loss": 0.3064, "step": 16430 }, { "epoch": 3.25, "learning_rate": 4.3811734492295534e-05, "loss": 0.2926, "step": 16440 }, { "epoch": 3.25, "learning_rate": 4.376234689845911e-05, "loss": 0.3072, "step": 16450 }, { "epoch": 3.25, "learning_rate": 4.371295930462268e-05, "loss": 0.2842, "step": 16460 }, { "epoch": 3.25, "learning_rate": 4.3663571710786255e-05, "loss": 0.3054, "step": 16470 }, { "epoch": 3.26, "learning_rate": 4.361418411694982e-05, "loss": 0.3039, "step": 16480 }, { "epoch": 3.26, "learning_rate": 4.3564796523113395e-05, "loss": 0.3171, "step": 16490 }, { "epoch": 3.26, "learning_rate": 4.351540892927697e-05, "loss": 0.3206, "step": 16500 }, { "epoch": 3.26, "learning_rate": 4.3466021335440535e-05, "loss": 0.3238, "step": 16510 }, { "epoch": 3.26, "learning_rate": 4.3416633741604116e-05, "loss": 0.302, "step": 16520 }, { "epoch": 3.27, "learning_rate": 4.336724614776768e-05, "loss": 0.2937, "step": 16530 }, { "epoch": 3.27, "learning_rate": 4.3317858553931256e-05, "loss": 0.2842, "step": 16540 }, { "epoch": 3.27, "learning_rate": 4.326847096009482e-05, "loss": 0.2942, "step": 16550 }, { "epoch": 3.27, "learning_rate": 4.32190833662584e-05, "loss": 0.316, "step": 16560 }, { "epoch": 3.27, "learning_rate": 4.316969577242197e-05, "loss": 0.3259, "step": 16570 }, { "epoch": 3.27, "learning_rate": 4.3120308178585536e-05, "loss": 0.3187, "step": 16580 }, { "epoch": 3.28, "learning_rate": 4.3070920584749116e-05, "loss": 0.3233, "step": 16590 }, { "epoch": 3.28, "learning_rate": 4.302153299091268e-05, "loss": 0.32, "step": 16600 }, { "epoch": 3.28, "learning_rate": 4.297214539707626e-05, "loss": 0.3153, "step": 16610 }, { "epoch": 3.28, "learning_rate": 4.292275780323983e-05, "loss": 0.2888, "step": 16620 }, { "epoch": 3.28, "learning_rate": 4.2873370209403404e-05, "loss": 0.2856, "step": 16630 }, { "epoch": 3.29, "learning_rate": 4.282398261556697e-05, "loss": 0.2774, "step": 16640 }, { "epoch": 3.29, "learning_rate": 4.2774595021730544e-05, "loss": 0.3354, "step": 16650 }, { "epoch": 3.29, "learning_rate": 4.272520742789412e-05, "loss": 0.2878, "step": 16660 }, { "epoch": 3.29, "learning_rate": 4.2675819834057684e-05, "loss": 0.2815, "step": 16670 }, { "epoch": 3.29, "learning_rate": 4.262643224022126e-05, "loss": 0.2946, "step": 16680 }, { "epoch": 3.3, "learning_rate": 4.257704464638483e-05, "loss": 0.2923, "step": 16690 }, { "epoch": 3.3, "learning_rate": 4.2527657052548405e-05, "loss": 0.3077, "step": 16700 }, { "epoch": 3.3, "learning_rate": 4.247826945871197e-05, "loss": 0.3122, "step": 16710 }, { "epoch": 3.3, "learning_rate": 4.2428881864875545e-05, "loss": 0.3078, "step": 16720 }, { "epoch": 3.3, "learning_rate": 4.237949427103912e-05, "loss": 0.2745, "step": 16730 }, { "epoch": 3.31, "learning_rate": 4.233010667720269e-05, "loss": 0.2914, "step": 16740 }, { "epoch": 3.31, "learning_rate": 4.228071908336626e-05, "loss": 0.2961, "step": 16750 }, { "epoch": 3.31, "learning_rate": 4.223133148952983e-05, "loss": 0.348, "step": 16760 }, { "epoch": 3.31, "learning_rate": 4.2181943895693406e-05, "loss": 0.299, "step": 16770 }, { "epoch": 3.31, "learning_rate": 4.213255630185697e-05, "loss": 0.2921, "step": 16780 }, { "epoch": 3.32, "learning_rate": 4.2083168708020546e-05, "loss": 0.3079, "step": 16790 }, { "epoch": 3.32, "learning_rate": 4.203378111418412e-05, "loss": 0.3188, "step": 16800 }, { "epoch": 3.32, "learning_rate": 4.198439352034769e-05, "loss": 0.3113, "step": 16810 }, { "epoch": 3.32, "learning_rate": 4.193500592651126e-05, "loss": 0.2778, "step": 16820 }, { "epoch": 3.32, "learning_rate": 4.188561833267484e-05, "loss": 0.3, "step": 16830 }, { "epoch": 3.33, "learning_rate": 4.1836230738838406e-05, "loss": 0.2995, "step": 16840 }, { "epoch": 3.33, "learning_rate": 4.178684314500197e-05, "loss": 0.3162, "step": 16850 }, { "epoch": 3.33, "learning_rate": 4.173745555116555e-05, "loss": 0.3278, "step": 16860 }, { "epoch": 3.33, "learning_rate": 4.168806795732912e-05, "loss": 0.2983, "step": 16870 }, { "epoch": 3.33, "learning_rate": 4.1638680363492694e-05, "loss": 0.2986, "step": 16880 }, { "epoch": 3.34, "learning_rate": 4.158929276965626e-05, "loss": 0.3173, "step": 16890 }, { "epoch": 3.34, "learning_rate": 4.153990517581984e-05, "loss": 0.3156, "step": 16900 }, { "epoch": 3.34, "learning_rate": 4.149051758198341e-05, "loss": 0.344, "step": 16910 }, { "epoch": 3.34, "learning_rate": 4.144112998814698e-05, "loss": 0.315, "step": 16920 }, { "epoch": 3.34, "learning_rate": 4.139174239431055e-05, "loss": 0.3198, "step": 16930 }, { "epoch": 3.35, "learning_rate": 4.134235480047412e-05, "loss": 0.3204, "step": 16940 }, { "epoch": 3.35, "learning_rate": 4.1292967206637695e-05, "loss": 0.3095, "step": 16950 }, { "epoch": 3.35, "learning_rate": 4.124357961280126e-05, "loss": 0.3087, "step": 16960 }, { "epoch": 3.35, "learning_rate": 4.119419201896484e-05, "loss": 0.2919, "step": 16970 }, { "epoch": 3.35, "learning_rate": 4.114480442512841e-05, "loss": 0.3063, "step": 16980 }, { "epoch": 3.36, "learning_rate": 4.109541683129198e-05, "loss": 0.3064, "step": 16990 }, { "epoch": 3.36, "learning_rate": 4.1046029237455555e-05, "loss": 0.3244, "step": 17000 }, { "epoch": 3.36, "learning_rate": 4.099664164361913e-05, "loss": 0.312, "step": 17010 }, { "epoch": 3.36, "learning_rate": 4.0947254049782695e-05, "loss": 0.2897, "step": 17020 }, { "epoch": 3.36, "learning_rate": 4.089786645594627e-05, "loss": 0.3097, "step": 17030 }, { "epoch": 3.37, "learning_rate": 4.084847886210984e-05, "loss": 0.3064, "step": 17040 }, { "epoch": 3.37, "learning_rate": 4.079909126827341e-05, "loss": 0.2972, "step": 17050 }, { "epoch": 3.37, "learning_rate": 4.074970367443698e-05, "loss": 0.3276, "step": 17060 }, { "epoch": 3.37, "learning_rate": 4.0700316080600556e-05, "loss": 0.3026, "step": 17070 }, { "epoch": 3.37, "learning_rate": 4.065092848676413e-05, "loss": 0.3203, "step": 17080 }, { "epoch": 3.38, "learning_rate": 4.0601540892927696e-05, "loss": 0.2626, "step": 17090 }, { "epoch": 3.38, "learning_rate": 4.055215329909127e-05, "loss": 0.3163, "step": 17100 }, { "epoch": 3.38, "learning_rate": 4.050276570525484e-05, "loss": 0.3096, "step": 17110 }, { "epoch": 3.38, "learning_rate": 4.045337811141842e-05, "loss": 0.3043, "step": 17120 }, { "epoch": 3.38, "learning_rate": 4.0403990517581984e-05, "loss": 0.2992, "step": 17130 }, { "epoch": 3.39, "learning_rate": 4.035460292374556e-05, "loss": 0.2828, "step": 17140 }, { "epoch": 3.39, "learning_rate": 4.030521532990913e-05, "loss": 0.2931, "step": 17150 }, { "epoch": 3.39, "learning_rate": 4.02558277360727e-05, "loss": 0.3316, "step": 17160 }, { "epoch": 3.39, "learning_rate": 4.020644014223627e-05, "loss": 0.3247, "step": 17170 }, { "epoch": 3.39, "learning_rate": 4.0157052548399844e-05, "loss": 0.3047, "step": 17180 }, { "epoch": 3.4, "learning_rate": 4.010766495456342e-05, "loss": 0.2743, "step": 17190 }, { "epoch": 3.4, "learning_rate": 4.0058277360726985e-05, "loss": 0.3335, "step": 17200 }, { "epoch": 3.4, "learning_rate": 4.0008889766890565e-05, "loss": 0.2842, "step": 17210 }, { "epoch": 3.4, "learning_rate": 3.995950217305413e-05, "loss": 0.3128, "step": 17220 }, { "epoch": 3.4, "learning_rate": 3.99101145792177e-05, "loss": 0.2957, "step": 17230 }, { "epoch": 3.41, "learning_rate": 3.986072698538127e-05, "loss": 0.2878, "step": 17240 }, { "epoch": 3.41, "learning_rate": 3.9811339391544845e-05, "loss": 0.279, "step": 17250 }, { "epoch": 3.41, "learning_rate": 3.976195179770842e-05, "loss": 0.2989, "step": 17260 }, { "epoch": 3.41, "learning_rate": 3.9712564203871985e-05, "loss": 0.3143, "step": 17270 }, { "epoch": 3.41, "learning_rate": 3.9663176610035566e-05, "loss": 0.3124, "step": 17280 }, { "epoch": 3.42, "learning_rate": 3.961378901619913e-05, "loss": 0.2871, "step": 17290 }, { "epoch": 3.42, "learning_rate": 3.9564401422362706e-05, "loss": 0.2909, "step": 17300 }, { "epoch": 3.42, "learning_rate": 3.951501382852628e-05, "loss": 0.337, "step": 17310 }, { "epoch": 3.42, "learning_rate": 3.9465626234689846e-05, "loss": 0.2757, "step": 17320 }, { "epoch": 3.42, "learning_rate": 3.941623864085342e-05, "loss": 0.3027, "step": 17330 }, { "epoch": 3.43, "learning_rate": 3.9366851047016986e-05, "loss": 0.3198, "step": 17340 }, { "epoch": 3.43, "learning_rate": 3.931746345318057e-05, "loss": 0.2902, "step": 17350 }, { "epoch": 3.43, "learning_rate": 3.926807585934413e-05, "loss": 0.2846, "step": 17360 }, { "epoch": 3.43, "learning_rate": 3.921868826550771e-05, "loss": 0.2988, "step": 17370 }, { "epoch": 3.43, "learning_rate": 3.916930067167128e-05, "loss": 0.2779, "step": 17380 }, { "epoch": 3.43, "learning_rate": 3.9119913077834854e-05, "loss": 0.3375, "step": 17390 }, { "epoch": 3.44, "learning_rate": 3.907052548399842e-05, "loss": 0.271, "step": 17400 }, { "epoch": 3.44, "learning_rate": 3.902113789016199e-05, "loss": 0.2879, "step": 17410 }, { "epoch": 3.44, "learning_rate": 3.897175029632557e-05, "loss": 0.3018, "step": 17420 }, { "epoch": 3.44, "learning_rate": 3.8922362702489134e-05, "loss": 0.3029, "step": 17430 }, { "epoch": 3.44, "learning_rate": 3.887297510865271e-05, "loss": 0.323, "step": 17440 }, { "epoch": 3.45, "learning_rate": 3.882358751481628e-05, "loss": 0.3053, "step": 17450 }, { "epoch": 3.45, "learning_rate": 3.8774199920979855e-05, "loss": 0.304, "step": 17460 }, { "epoch": 3.45, "learning_rate": 3.872481232714342e-05, "loss": 0.3027, "step": 17470 }, { "epoch": 3.45, "learning_rate": 3.8675424733306995e-05, "loss": 0.2652, "step": 17480 }, { "epoch": 3.45, "learning_rate": 3.862603713947057e-05, "loss": 0.3122, "step": 17490 }, { "epoch": 3.46, "learning_rate": 3.8576649545634135e-05, "loss": 0.3255, "step": 17500 }, { "epoch": 3.46, "learning_rate": 3.852726195179771e-05, "loss": 0.2935, "step": 17510 }, { "epoch": 3.46, "learning_rate": 3.847787435796128e-05, "loss": 0.3082, "step": 17520 }, { "epoch": 3.46, "learning_rate": 3.8428486764124856e-05, "loss": 0.2994, "step": 17530 }, { "epoch": 3.46, "learning_rate": 3.837909917028842e-05, "loss": 0.2871, "step": 17540 }, { "epoch": 3.47, "learning_rate": 3.8329711576451996e-05, "loss": 0.2793, "step": 17550 }, { "epoch": 3.47, "learning_rate": 3.828032398261557e-05, "loss": 0.322, "step": 17560 }, { "epoch": 3.47, "learning_rate": 3.823093638877914e-05, "loss": 0.3007, "step": 17570 }, { "epoch": 3.47, "learning_rate": 3.818154879494271e-05, "loss": 0.3015, "step": 17580 }, { "epoch": 3.47, "learning_rate": 3.813216120110628e-05, "loss": 0.292, "step": 17590 }, { "epoch": 3.48, "learning_rate": 3.8082773607269857e-05, "loss": 0.3037, "step": 17600 }, { "epoch": 3.48, "learning_rate": 3.803338601343342e-05, "loss": 0.2884, "step": 17610 }, { "epoch": 3.48, "learning_rate": 3.7983998419597e-05, "loss": 0.3092, "step": 17620 }, { "epoch": 3.48, "learning_rate": 3.793461082576057e-05, "loss": 0.3096, "step": 17630 }, { "epoch": 3.48, "learning_rate": 3.7885223231924144e-05, "loss": 0.3167, "step": 17640 }, { "epoch": 3.49, "learning_rate": 3.783583563808771e-05, "loss": 0.2856, "step": 17650 }, { "epoch": 3.49, "learning_rate": 3.778644804425129e-05, "loss": 0.3014, "step": 17660 }, { "epoch": 3.49, "learning_rate": 3.773706045041486e-05, "loss": 0.3484, "step": 17670 }, { "epoch": 3.49, "learning_rate": 3.768767285657843e-05, "loss": 0.2797, "step": 17680 }, { "epoch": 3.49, "learning_rate": 3.7638285262742004e-05, "loss": 0.2989, "step": 17690 }, { "epoch": 3.5, "learning_rate": 3.758889766890557e-05, "loss": 0.3149, "step": 17700 }, { "epoch": 3.5, "learning_rate": 3.7539510075069145e-05, "loss": 0.3028, "step": 17710 }, { "epoch": 3.5, "learning_rate": 3.749012248123271e-05, "loss": 0.2997, "step": 17720 }, { "epoch": 3.5, "learning_rate": 3.744073488739629e-05, "loss": 0.311, "step": 17730 }, { "epoch": 3.5, "learning_rate": 3.739134729355986e-05, "loss": 0.2868, "step": 17740 }, { "epoch": 3.51, "learning_rate": 3.734195969972343e-05, "loss": 0.2973, "step": 17750 }, { "epoch": 3.51, "learning_rate": 3.7292572105887005e-05, "loss": 0.2864, "step": 17760 }, { "epoch": 3.51, "learning_rate": 3.724318451205058e-05, "loss": 0.3013, "step": 17770 }, { "epoch": 3.51, "learning_rate": 3.7193796918214146e-05, "loss": 0.3212, "step": 17780 }, { "epoch": 3.51, "learning_rate": 3.714440932437772e-05, "loss": 0.2902, "step": 17790 }, { "epoch": 3.52, "learning_rate": 3.709502173054129e-05, "loss": 0.2872, "step": 17800 }, { "epoch": 3.52, "learning_rate": 3.704563413670486e-05, "loss": 0.3161, "step": 17810 }, { "epoch": 3.52, "learning_rate": 3.699624654286843e-05, "loss": 0.309, "step": 17820 }, { "epoch": 3.52, "learning_rate": 3.6946858949032006e-05, "loss": 0.2949, "step": 17830 }, { "epoch": 3.52, "learning_rate": 3.689747135519558e-05, "loss": 0.2855, "step": 17840 }, { "epoch": 3.53, "learning_rate": 3.6848083761359147e-05, "loss": 0.3054, "step": 17850 }, { "epoch": 3.53, "learning_rate": 3.679869616752272e-05, "loss": 0.2998, "step": 17860 }, { "epoch": 3.53, "learning_rate": 3.6749308573686294e-05, "loss": 0.2755, "step": 17870 }, { "epoch": 3.53, "learning_rate": 3.669992097984986e-05, "loss": 0.2883, "step": 17880 }, { "epoch": 3.53, "learning_rate": 3.6650533386013434e-05, "loss": 0.314, "step": 17890 }, { "epoch": 3.54, "learning_rate": 3.660114579217701e-05, "loss": 0.329, "step": 17900 }, { "epoch": 3.54, "learning_rate": 3.655175819834058e-05, "loss": 0.2971, "step": 17910 }, { "epoch": 3.54, "learning_rate": 3.650237060450415e-05, "loss": 0.2966, "step": 17920 }, { "epoch": 3.54, "learning_rate": 3.645298301066772e-05, "loss": 0.2967, "step": 17930 }, { "epoch": 3.54, "learning_rate": 3.6403595416831294e-05, "loss": 0.2906, "step": 17940 }, { "epoch": 3.55, "learning_rate": 3.635420782299487e-05, "loss": 0.2784, "step": 17950 }, { "epoch": 3.55, "learning_rate": 3.6304820229158435e-05, "loss": 0.3072, "step": 17960 }, { "epoch": 3.55, "learning_rate": 3.625543263532201e-05, "loss": 0.3132, "step": 17970 }, { "epoch": 3.55, "learning_rate": 3.620604504148558e-05, "loss": 0.2892, "step": 17980 }, { "epoch": 3.55, "learning_rate": 3.615665744764915e-05, "loss": 0.3263, "step": 17990 }, { "epoch": 3.56, "learning_rate": 3.610726985381272e-05, "loss": 0.3042, "step": 18000 }, { "epoch": 3.56, "learning_rate": 3.6057882259976295e-05, "loss": 0.2969, "step": 18010 }, { "epoch": 3.56, "learning_rate": 3.600849466613987e-05, "loss": 0.3205, "step": 18020 }, { "epoch": 3.56, "learning_rate": 3.5959107072303436e-05, "loss": 0.27, "step": 18030 }, { "epoch": 3.56, "learning_rate": 3.5909719478467016e-05, "loss": 0.2854, "step": 18040 }, { "epoch": 3.57, "learning_rate": 3.586033188463058e-05, "loss": 0.3176, "step": 18050 }, { "epoch": 3.57, "learning_rate": 3.581094429079415e-05, "loss": 0.3057, "step": 18060 }, { "epoch": 3.57, "learning_rate": 3.576155669695773e-05, "loss": 0.3008, "step": 18070 }, { "epoch": 3.57, "learning_rate": 3.5712169103121296e-05, "loss": 0.315, "step": 18080 }, { "epoch": 3.57, "learning_rate": 3.566278150928487e-05, "loss": 0.3071, "step": 18090 }, { "epoch": 3.58, "learning_rate": 3.5613393915448436e-05, "loss": 0.2843, "step": 18100 }, { "epoch": 3.58, "learning_rate": 3.556400632161202e-05, "loss": 0.2941, "step": 18110 }, { "epoch": 3.58, "learning_rate": 3.5514618727775583e-05, "loss": 0.2894, "step": 18120 }, { "epoch": 3.58, "learning_rate": 3.546523113393916e-05, "loss": 0.2947, "step": 18130 }, { "epoch": 3.58, "learning_rate": 3.541584354010273e-05, "loss": 0.307, "step": 18140 }, { "epoch": 3.59, "learning_rate": 3.53664559462663e-05, "loss": 0.2845, "step": 18150 }, { "epoch": 3.59, "learning_rate": 3.531706835242987e-05, "loss": 0.3046, "step": 18160 }, { "epoch": 3.59, "learning_rate": 3.5267680758593444e-05, "loss": 0.3094, "step": 18170 }, { "epoch": 3.59, "learning_rate": 3.521829316475702e-05, "loss": 0.2841, "step": 18180 }, { "epoch": 3.59, "learning_rate": 3.5168905570920584e-05, "loss": 0.2924, "step": 18190 }, { "epoch": 3.59, "learning_rate": 3.511951797708416e-05, "loss": 0.3113, "step": 18200 }, { "epoch": 3.6, "learning_rate": 3.507013038324773e-05, "loss": 0.3007, "step": 18210 }, { "epoch": 3.6, "learning_rate": 3.5020742789411305e-05, "loss": 0.3239, "step": 18220 }, { "epoch": 3.6, "learning_rate": 3.497135519557487e-05, "loss": 0.2743, "step": 18230 }, { "epoch": 3.6, "learning_rate": 3.4921967601738445e-05, "loss": 0.2828, "step": 18240 }, { "epoch": 3.6, "learning_rate": 3.487258000790202e-05, "loss": 0.2861, "step": 18250 }, { "epoch": 3.61, "learning_rate": 3.4823192414065585e-05, "loss": 0.2954, "step": 18260 }, { "epoch": 3.61, "learning_rate": 3.477380482022916e-05, "loss": 0.2797, "step": 18270 }, { "epoch": 3.61, "learning_rate": 3.472441722639273e-05, "loss": 0.2784, "step": 18280 }, { "epoch": 3.61, "learning_rate": 3.4675029632556306e-05, "loss": 0.2882, "step": 18290 }, { "epoch": 3.61, "learning_rate": 3.462564203871987e-05, "loss": 0.3058, "step": 18300 }, { "epoch": 3.62, "learning_rate": 3.4576254444883446e-05, "loss": 0.3179, "step": 18310 }, { "epoch": 3.62, "learning_rate": 3.452686685104702e-05, "loss": 0.2869, "step": 18320 }, { "epoch": 3.62, "learning_rate": 3.4477479257210586e-05, "loss": 0.3129, "step": 18330 }, { "epoch": 3.62, "learning_rate": 3.442809166337416e-05, "loss": 0.3025, "step": 18340 }, { "epoch": 3.62, "learning_rate": 3.437870406953773e-05, "loss": 0.3527, "step": 18350 }, { "epoch": 3.63, "learning_rate": 3.432931647570131e-05, "loss": 0.2966, "step": 18360 }, { "epoch": 3.63, "learning_rate": 3.4279928881864873e-05, "loss": 0.2657, "step": 18370 }, { "epoch": 3.63, "learning_rate": 3.4230541288028454e-05, "loss": 0.3055, "step": 18380 }, { "epoch": 3.63, "learning_rate": 3.418115369419202e-05, "loss": 0.3219, "step": 18390 }, { "epoch": 3.63, "learning_rate": 3.4131766100355594e-05, "loss": 0.3077, "step": 18400 }, { "epoch": 3.64, "learning_rate": 3.408237850651916e-05, "loss": 0.2738, "step": 18410 }, { "epoch": 3.64, "learning_rate": 3.403299091268274e-05, "loss": 0.3153, "step": 18420 }, { "epoch": 3.64, "learning_rate": 3.398360331884631e-05, "loss": 0.2917, "step": 18430 }, { "epoch": 3.64, "learning_rate": 3.3934215725009874e-05, "loss": 0.3054, "step": 18440 }, { "epoch": 3.64, "learning_rate": 3.3884828131173455e-05, "loss": 0.2964, "step": 18450 }, { "epoch": 3.65, "learning_rate": 3.383544053733702e-05, "loss": 0.3137, "step": 18460 }, { "epoch": 3.65, "learning_rate": 3.3786052943500595e-05, "loss": 0.3207, "step": 18470 }, { "epoch": 3.65, "learning_rate": 3.373666534966416e-05, "loss": 0.3139, "step": 18480 }, { "epoch": 3.65, "learning_rate": 3.368727775582774e-05, "loss": 0.2947, "step": 18490 }, { "epoch": 3.65, "learning_rate": 3.363789016199131e-05, "loss": 0.3013, "step": 18500 }, { "epoch": 3.66, "learning_rate": 3.358850256815488e-05, "loss": 0.3065, "step": 18510 }, { "epoch": 3.66, "learning_rate": 3.3539114974318456e-05, "loss": 0.2882, "step": 18520 }, { "epoch": 3.66, "learning_rate": 3.348972738048202e-05, "loss": 0.3229, "step": 18530 }, { "epoch": 3.66, "learning_rate": 3.3440339786645596e-05, "loss": 0.2909, "step": 18540 }, { "epoch": 3.66, "learning_rate": 3.339095219280917e-05, "loss": 0.3113, "step": 18550 }, { "epoch": 3.67, "learning_rate": 3.334156459897274e-05, "loss": 0.2907, "step": 18560 }, { "epoch": 3.67, "learning_rate": 3.329217700513631e-05, "loss": 0.311, "step": 18570 }, { "epoch": 3.67, "learning_rate": 3.324278941129988e-05, "loss": 0.2985, "step": 18580 }, { "epoch": 3.67, "learning_rate": 3.3193401817463456e-05, "loss": 0.3131, "step": 18590 }, { "epoch": 3.67, "learning_rate": 3.314401422362703e-05, "loss": 0.2932, "step": 18600 }, { "epoch": 3.68, "learning_rate": 3.30946266297906e-05, "loss": 0.3201, "step": 18610 }, { "epoch": 3.68, "learning_rate": 3.304523903595417e-05, "loss": 0.2925, "step": 18620 }, { "epoch": 3.68, "learning_rate": 3.2995851442117744e-05, "loss": 0.3028, "step": 18630 }, { "epoch": 3.68, "learning_rate": 3.294646384828131e-05, "loss": 0.2915, "step": 18640 }, { "epoch": 3.68, "learning_rate": 3.2897076254444884e-05, "loss": 0.2979, "step": 18650 }, { "epoch": 3.69, "learning_rate": 3.284768866060846e-05, "loss": 0.3002, "step": 18660 }, { "epoch": 3.69, "learning_rate": 3.279830106677203e-05, "loss": 0.291, "step": 18670 }, { "epoch": 3.69, "learning_rate": 3.27489134729356e-05, "loss": 0.2957, "step": 18680 }, { "epoch": 3.69, "learning_rate": 3.269952587909917e-05, "loss": 0.3264, "step": 18690 }, { "epoch": 3.69, "learning_rate": 3.2650138285262745e-05, "loss": 0.2988, "step": 18700 }, { "epoch": 3.7, "learning_rate": 3.260075069142631e-05, "loss": 0.2877, "step": 18710 }, { "epoch": 3.7, "learning_rate": 3.2551363097589885e-05, "loss": 0.2711, "step": 18720 }, { "epoch": 3.7, "learning_rate": 3.250197550375346e-05, "loss": 0.3154, "step": 18730 }, { "epoch": 3.7, "learning_rate": 3.245258790991703e-05, "loss": 0.2821, "step": 18740 }, { "epoch": 3.7, "learning_rate": 3.24032003160806e-05, "loss": 0.3037, "step": 18750 }, { "epoch": 3.71, "learning_rate": 3.235381272224418e-05, "loss": 0.2927, "step": 18760 }, { "epoch": 3.71, "learning_rate": 3.2304425128407745e-05, "loss": 0.3069, "step": 18770 }, { "epoch": 3.71, "learning_rate": 3.225503753457132e-05, "loss": 0.2811, "step": 18780 }, { "epoch": 3.71, "learning_rate": 3.2205649940734886e-05, "loss": 0.2717, "step": 18790 }, { "epoch": 3.71, "learning_rate": 3.215626234689846e-05, "loss": 0.2793, "step": 18800 }, { "epoch": 3.72, "learning_rate": 3.210687475306203e-05, "loss": 0.3108, "step": 18810 }, { "epoch": 3.72, "learning_rate": 3.20574871592256e-05, "loss": 0.2932, "step": 18820 }, { "epoch": 3.72, "learning_rate": 3.200809956538918e-05, "loss": 0.3209, "step": 18830 }, { "epoch": 3.72, "learning_rate": 3.1958711971552746e-05, "loss": 0.3143, "step": 18840 }, { "epoch": 3.72, "learning_rate": 3.190932437771632e-05, "loss": 0.2956, "step": 18850 }, { "epoch": 3.73, "learning_rate": 3.1859936783879893e-05, "loss": 0.278, "step": 18860 }, { "epoch": 3.73, "learning_rate": 3.181054919004347e-05, "loss": 0.2769, "step": 18870 }, { "epoch": 3.73, "learning_rate": 3.1761161596207034e-05, "loss": 0.299, "step": 18880 }, { "epoch": 3.73, "learning_rate": 3.17117740023706e-05, "loss": 0.3062, "step": 18890 }, { "epoch": 3.73, "learning_rate": 3.166238640853418e-05, "loss": 0.2961, "step": 18900 }, { "epoch": 3.74, "learning_rate": 3.161299881469775e-05, "loss": 0.291, "step": 18910 }, { "epoch": 3.74, "learning_rate": 3.156361122086132e-05, "loss": 0.2619, "step": 18920 }, { "epoch": 3.74, "learning_rate": 3.1514223627024894e-05, "loss": 0.2959, "step": 18930 }, { "epoch": 3.74, "learning_rate": 3.146483603318847e-05, "loss": 0.2728, "step": 18940 }, { "epoch": 3.74, "learning_rate": 3.1415448439352035e-05, "loss": 0.3019, "step": 18950 }, { "epoch": 3.75, "learning_rate": 3.136606084551561e-05, "loss": 0.3124, "step": 18960 }, { "epoch": 3.75, "learning_rate": 3.131667325167918e-05, "loss": 0.2799, "step": 18970 }, { "epoch": 3.75, "learning_rate": 3.126728565784275e-05, "loss": 0.2747, "step": 18980 }, { "epoch": 3.75, "learning_rate": 3.121789806400632e-05, "loss": 0.3079, "step": 18990 }, { "epoch": 3.75, "learning_rate": 3.1168510470169895e-05, "loss": 0.2851, "step": 19000 }, { "epoch": 3.75, "learning_rate": 3.111912287633347e-05, "loss": 0.3185, "step": 19010 }, { "epoch": 3.76, "learning_rate": 3.1069735282497035e-05, "loss": 0.2465, "step": 19020 }, { "epoch": 3.76, "learning_rate": 3.102034768866061e-05, "loss": 0.2879, "step": 19030 }, { "epoch": 3.76, "learning_rate": 3.097589885420782e-05, "loss": 0.305, "step": 19040 }, { "epoch": 3.76, "learning_rate": 3.0926511260371395e-05, "loss": 0.3262, "step": 19050 }, { "epoch": 3.76, "learning_rate": 3.087712366653497e-05, "loss": 0.3165, "step": 19060 }, { "epoch": 3.77, "learning_rate": 3.0827736072698535e-05, "loss": 0.2926, "step": 19070 }, { "epoch": 3.77, "learning_rate": 3.0778348478862116e-05, "loss": 0.3106, "step": 19080 }, { "epoch": 3.77, "learning_rate": 3.072896088502568e-05, "loss": 0.2656, "step": 19090 }, { "epoch": 3.77, "learning_rate": 3.0679573291189256e-05, "loss": 0.2805, "step": 19100 }, { "epoch": 3.77, "learning_rate": 3.063018569735282e-05, "loss": 0.2925, "step": 19110 }, { "epoch": 3.78, "learning_rate": 3.05807981035164e-05, "loss": 0.3071, "step": 19120 }, { "epoch": 3.78, "learning_rate": 3.053141050967997e-05, "loss": 0.2697, "step": 19130 }, { "epoch": 3.78, "learning_rate": 3.048202291584354e-05, "loss": 0.275, "step": 19140 }, { "epoch": 3.78, "learning_rate": 3.0432635322007113e-05, "loss": 0.2938, "step": 19150 }, { "epoch": 3.78, "learning_rate": 3.0383247728170683e-05, "loss": 0.2778, "step": 19160 }, { "epoch": 3.79, "learning_rate": 3.0333860134334257e-05, "loss": 0.3018, "step": 19170 }, { "epoch": 3.79, "learning_rate": 3.0284472540497827e-05, "loss": 0.2811, "step": 19180 }, { "epoch": 3.79, "learning_rate": 3.02350849466614e-05, "loss": 0.3074, "step": 19190 }, { "epoch": 3.79, "learning_rate": 3.018569735282497e-05, "loss": 0.277, "step": 19200 }, { "epoch": 3.79, "learning_rate": 3.0136309758988544e-05, "loss": 0.2726, "step": 19210 }, { "epoch": 3.8, "learning_rate": 3.0086922165152114e-05, "loss": 0.3137, "step": 19220 }, { "epoch": 3.8, "learning_rate": 3.003753457131569e-05, "loss": 0.2914, "step": 19230 }, { "epoch": 3.8, "learning_rate": 2.9988146977479258e-05, "loss": 0.2926, "step": 19240 }, { "epoch": 3.8, "learning_rate": 2.9938759383642828e-05, "loss": 0.2604, "step": 19250 }, { "epoch": 3.8, "learning_rate": 2.98893717898064e-05, "loss": 0.3051, "step": 19260 }, { "epoch": 3.81, "learning_rate": 2.983998419596997e-05, "loss": 0.2819, "step": 19270 }, { "epoch": 3.81, "learning_rate": 2.979059660213355e-05, "loss": 0.2634, "step": 19280 }, { "epoch": 3.81, "learning_rate": 2.9741209008297115e-05, "loss": 0.3182, "step": 19290 }, { "epoch": 3.81, "learning_rate": 2.9691821414460692e-05, "loss": 0.2994, "step": 19300 }, { "epoch": 3.81, "learning_rate": 2.964243382062426e-05, "loss": 0.2832, "step": 19310 }, { "epoch": 3.82, "learning_rate": 2.9593046226787836e-05, "loss": 0.2882, "step": 19320 }, { "epoch": 3.82, "learning_rate": 2.9543658632951406e-05, "loss": 0.2945, "step": 19330 }, { "epoch": 3.82, "learning_rate": 2.9494271039114972e-05, "loss": 0.2791, "step": 19340 }, { "epoch": 3.82, "learning_rate": 2.944488344527855e-05, "loss": 0.2878, "step": 19350 }, { "epoch": 3.82, "learning_rate": 2.9395495851442116e-05, "loss": 0.2927, "step": 19360 }, { "epoch": 3.83, "learning_rate": 2.9346108257605693e-05, "loss": 0.26, "step": 19370 }, { "epoch": 3.83, "learning_rate": 2.9296720663769263e-05, "loss": 0.3125, "step": 19380 }, { "epoch": 3.83, "learning_rate": 2.9247333069932836e-05, "loss": 0.2838, "step": 19390 }, { "epoch": 3.83, "learning_rate": 2.9197945476096407e-05, "loss": 0.286, "step": 19400 }, { "epoch": 3.83, "learning_rate": 2.914855788225998e-05, "loss": 0.2613, "step": 19410 }, { "epoch": 3.84, "learning_rate": 2.909917028842355e-05, "loss": 0.2983, "step": 19420 }, { "epoch": 3.84, "learning_rate": 2.904978269458712e-05, "loss": 0.2832, "step": 19430 }, { "epoch": 3.84, "learning_rate": 2.9000395100750694e-05, "loss": 0.2839, "step": 19440 }, { "epoch": 3.84, "learning_rate": 2.8951007506914264e-05, "loss": 0.291, "step": 19450 }, { "epoch": 3.84, "learning_rate": 2.8901619913077837e-05, "loss": 0.2975, "step": 19460 }, { "epoch": 3.85, "learning_rate": 2.8852232319241407e-05, "loss": 0.2812, "step": 19470 }, { "epoch": 3.85, "learning_rate": 2.880284472540498e-05, "loss": 0.2656, "step": 19480 }, { "epoch": 3.85, "learning_rate": 2.875345713156855e-05, "loss": 0.2952, "step": 19490 }, { "epoch": 3.85, "learning_rate": 2.8704069537732125e-05, "loss": 0.2937, "step": 19500 }, { "epoch": 3.85, "learning_rate": 2.8654681943895695e-05, "loss": 0.2863, "step": 19510 }, { "epoch": 3.86, "learning_rate": 2.8605294350059265e-05, "loss": 0.2747, "step": 19520 }, { "epoch": 3.86, "learning_rate": 2.8555906756222838e-05, "loss": 0.3162, "step": 19530 }, { "epoch": 3.86, "learning_rate": 2.850651916238641e-05, "loss": 0.2862, "step": 19540 }, { "epoch": 3.86, "learning_rate": 2.8457131568549982e-05, "loss": 0.3329, "step": 19550 }, { "epoch": 3.86, "learning_rate": 2.8407743974713552e-05, "loss": 0.2911, "step": 19560 }, { "epoch": 3.87, "learning_rate": 2.8358356380877125e-05, "loss": 0.2677, "step": 19570 }, { "epoch": 3.87, "learning_rate": 2.8308968787040696e-05, "loss": 0.2785, "step": 19580 }, { "epoch": 3.87, "learning_rate": 2.8259581193204272e-05, "loss": 0.2791, "step": 19590 }, { "epoch": 3.87, "learning_rate": 2.821019359936784e-05, "loss": 0.3019, "step": 19600 }, { "epoch": 3.87, "learning_rate": 2.816080600553141e-05, "loss": 0.3028, "step": 19610 }, { "epoch": 3.88, "learning_rate": 2.8111418411694983e-05, "loss": 0.291, "step": 19620 }, { "epoch": 3.88, "learning_rate": 2.8062030817858553e-05, "loss": 0.2861, "step": 19630 }, { "epoch": 3.88, "learning_rate": 2.801264322402213e-05, "loss": 0.3183, "step": 19640 }, { "epoch": 3.88, "learning_rate": 2.7963255630185696e-05, "loss": 0.2645, "step": 19650 }, { "epoch": 3.88, "learning_rate": 2.7913868036349273e-05, "loss": 0.2863, "step": 19660 }, { "epoch": 3.89, "learning_rate": 2.786448044251284e-05, "loss": 0.302, "step": 19670 }, { "epoch": 3.89, "learning_rate": 2.7815092848676417e-05, "loss": 0.3091, "step": 19680 }, { "epoch": 3.89, "learning_rate": 2.7765705254839987e-05, "loss": 0.3034, "step": 19690 }, { "epoch": 3.89, "learning_rate": 2.7716317661003554e-05, "loss": 0.2969, "step": 19700 }, { "epoch": 3.89, "learning_rate": 2.766693006716713e-05, "loss": 0.2975, "step": 19710 }, { "epoch": 3.9, "learning_rate": 2.7617542473330697e-05, "loss": 0.2903, "step": 19720 }, { "epoch": 3.9, "learning_rate": 2.7568154879494274e-05, "loss": 0.2698, "step": 19730 }, { "epoch": 3.9, "learning_rate": 2.751876728565784e-05, "loss": 0.2794, "step": 19740 }, { "epoch": 3.9, "learning_rate": 2.7469379691821418e-05, "loss": 0.2711, "step": 19750 }, { "epoch": 3.9, "learning_rate": 2.7419992097984988e-05, "loss": 0.2747, "step": 19760 }, { "epoch": 3.9, "learning_rate": 2.737060450414856e-05, "loss": 0.3038, "step": 19770 }, { "epoch": 3.91, "learning_rate": 2.732121691031213e-05, "loss": 0.29, "step": 19780 }, { "epoch": 3.91, "learning_rate": 2.72718293164757e-05, "loss": 0.2584, "step": 19790 }, { "epoch": 3.91, "learning_rate": 2.7222441722639275e-05, "loss": 0.2765, "step": 19800 }, { "epoch": 3.91, "learning_rate": 2.7173054128802845e-05, "loss": 0.3226, "step": 19810 }, { "epoch": 3.91, "learning_rate": 2.712366653496642e-05, "loss": 0.2937, "step": 19820 }, { "epoch": 3.92, "learning_rate": 2.707427894112999e-05, "loss": 0.334, "step": 19830 }, { "epoch": 3.92, "learning_rate": 2.7024891347293562e-05, "loss": 0.3204, "step": 19840 }, { "epoch": 3.92, "learning_rate": 2.6975503753457133e-05, "loss": 0.2844, "step": 19850 }, { "epoch": 3.92, "learning_rate": 2.6926116159620706e-05, "loss": 0.2686, "step": 19860 }, { "epoch": 3.92, "learning_rate": 2.6876728565784276e-05, "loss": 0.2849, "step": 19870 }, { "epoch": 3.93, "learning_rate": 2.6827340971947846e-05, "loss": 0.2724, "step": 19880 }, { "epoch": 3.93, "learning_rate": 2.677795337811142e-05, "loss": 0.275, "step": 19890 }, { "epoch": 3.93, "learning_rate": 2.672856578427499e-05, "loss": 0.2885, "step": 19900 }, { "epoch": 3.93, "learning_rate": 2.6679178190438563e-05, "loss": 0.2967, "step": 19910 }, { "epoch": 3.93, "learning_rate": 2.6629790596602133e-05, "loss": 0.2844, "step": 19920 }, { "epoch": 3.94, "learning_rate": 2.6580403002765707e-05, "loss": 0.3013, "step": 19930 }, { "epoch": 3.94, "learning_rate": 2.6531015408929277e-05, "loss": 0.2939, "step": 19940 }, { "epoch": 3.94, "learning_rate": 2.648162781509285e-05, "loss": 0.274, "step": 19950 }, { "epoch": 3.94, "learning_rate": 2.643224022125642e-05, "loss": 0.3016, "step": 19960 }, { "epoch": 3.94, "learning_rate": 2.6382852627419998e-05, "loss": 0.2717, "step": 19970 }, { "epoch": 3.95, "learning_rate": 2.6333465033583564e-05, "loss": 0.3, "step": 19980 }, { "epoch": 3.95, "learning_rate": 2.6284077439747134e-05, "loss": 0.2556, "step": 19990 }, { "epoch": 3.95, "learning_rate": 2.6234689845910708e-05, "loss": 0.2874, "step": 20000 }, { "epoch": 3.95, "learning_rate": 2.6185302252074278e-05, "loss": 0.2634, "step": 20010 }, { "epoch": 3.95, "learning_rate": 2.6135914658237855e-05, "loss": 0.3132, "step": 20020 }, { "epoch": 3.96, "learning_rate": 2.608652706440142e-05, "loss": 0.2753, "step": 20030 }, { "epoch": 3.96, "learning_rate": 2.6037139470565e-05, "loss": 0.2937, "step": 20040 }, { "epoch": 3.96, "learning_rate": 2.5987751876728565e-05, "loss": 0.2915, "step": 20050 }, { "epoch": 3.96, "learning_rate": 2.5938364282892142e-05, "loss": 0.2849, "step": 20060 }, { "epoch": 3.96, "learning_rate": 2.5888976689055712e-05, "loss": 0.2791, "step": 20070 }, { "epoch": 3.97, "learning_rate": 2.583958909521928e-05, "loss": 0.3088, "step": 20080 }, { "epoch": 3.97, "learning_rate": 2.5790201501382856e-05, "loss": 0.2957, "step": 20090 }, { "epoch": 3.97, "learning_rate": 2.5740813907546422e-05, "loss": 0.3183, "step": 20100 }, { "epoch": 3.97, "learning_rate": 2.569142631371e-05, "loss": 0.2607, "step": 20110 }, { "epoch": 3.97, "learning_rate": 2.564203871987357e-05, "loss": 0.2954, "step": 20120 }, { "epoch": 3.98, "learning_rate": 2.5592651126037143e-05, "loss": 0.2764, "step": 20130 }, { "epoch": 3.98, "learning_rate": 2.5543263532200713e-05, "loss": 0.3018, "step": 20140 }, { "epoch": 3.98, "learning_rate": 2.5493875938364287e-05, "loss": 0.2829, "step": 20150 }, { "epoch": 3.98, "learning_rate": 2.5444488344527857e-05, "loss": 0.3127, "step": 20160 }, { "epoch": 3.98, "learning_rate": 2.5395100750691427e-05, "loss": 0.2915, "step": 20170 }, { "epoch": 3.99, "learning_rate": 2.5345713156855e-05, "loss": 0.29, "step": 20180 }, { "epoch": 3.99, "learning_rate": 2.529632556301857e-05, "loss": 0.318, "step": 20190 }, { "epoch": 3.99, "learning_rate": 2.5246937969182144e-05, "loss": 0.2786, "step": 20200 }, { "epoch": 3.99, "learning_rate": 2.5197550375345714e-05, "loss": 0.2842, "step": 20210 }, { "epoch": 3.99, "learning_rate": 2.5148162781509287e-05, "loss": 0.2954, "step": 20220 }, { "epoch": 4.0, "learning_rate": 2.5098775187672858e-05, "loss": 0.2969, "step": 20230 }, { "epoch": 4.0, "learning_rate": 2.504938759383643e-05, "loss": 0.2977, "step": 20240 }, { "epoch": 4.0, "learning_rate": 2.5e-05, "loss": 0.2796, "step": 20250 }, { "epoch": 4.0, "learning_rate": 2.4950612406163575e-05, "loss": 0.1823, "step": 20260 }, { "epoch": 4.0, "learning_rate": 2.4901224812327145e-05, "loss": 0.185, "step": 20270 }, { "epoch": 4.01, "learning_rate": 2.4851837218490718e-05, "loss": 0.1916, "step": 20280 }, { "epoch": 4.01, "learning_rate": 2.480244962465429e-05, "loss": 0.1983, "step": 20290 }, { "epoch": 4.01, "learning_rate": 2.475306203081786e-05, "loss": 0.2032, "step": 20300 }, { "epoch": 4.01, "learning_rate": 2.4703674436981432e-05, "loss": 0.187, "step": 20310 }, { "epoch": 4.01, "learning_rate": 2.4654286843145002e-05, "loss": 0.1894, "step": 20320 }, { "epoch": 4.02, "learning_rate": 2.4604899249308576e-05, "loss": 0.1981, "step": 20330 }, { "epoch": 4.02, "learning_rate": 2.4555511655472146e-05, "loss": 0.1942, "step": 20340 }, { "epoch": 4.02, "learning_rate": 2.450612406163572e-05, "loss": 0.1887, "step": 20350 }, { "epoch": 4.02, "learning_rate": 2.445673646779929e-05, "loss": 0.1938, "step": 20360 }, { "epoch": 4.02, "learning_rate": 2.4407348873962863e-05, "loss": 0.1857, "step": 20370 }, { "epoch": 4.03, "learning_rate": 2.4357961280126433e-05, "loss": 0.1913, "step": 20380 }, { "epoch": 4.03, "learning_rate": 2.4308573686290003e-05, "loss": 0.1993, "step": 20390 }, { "epoch": 4.03, "learning_rate": 2.4259186092453577e-05, "loss": 0.1861, "step": 20400 }, { "epoch": 4.03, "learning_rate": 2.4209798498617147e-05, "loss": 0.1944, "step": 20410 }, { "epoch": 4.03, "learning_rate": 2.416041090478072e-05, "loss": 0.1867, "step": 20420 }, { "epoch": 4.04, "learning_rate": 2.411102331094429e-05, "loss": 0.1835, "step": 20430 }, { "epoch": 4.04, "learning_rate": 2.4061635717107864e-05, "loss": 0.1877, "step": 20440 }, { "epoch": 4.04, "learning_rate": 2.4012248123271437e-05, "loss": 0.1851, "step": 20450 }, { "epoch": 4.04, "learning_rate": 2.3962860529435007e-05, "loss": 0.1979, "step": 20460 }, { "epoch": 4.04, "learning_rate": 2.391347293559858e-05, "loss": 0.1892, "step": 20470 }, { "epoch": 4.05, "learning_rate": 2.386408534176215e-05, "loss": 0.1964, "step": 20480 }, { "epoch": 4.05, "learning_rate": 2.381469774792572e-05, "loss": 0.1947, "step": 20490 }, { "epoch": 4.05, "learning_rate": 2.3765310154089295e-05, "loss": 0.1952, "step": 20500 }, { "epoch": 4.05, "learning_rate": 2.3715922560252865e-05, "loss": 0.187, "step": 20510 }, { "epoch": 4.05, "learning_rate": 2.3666534966416438e-05, "loss": 0.1989, "step": 20520 }, { "epoch": 4.06, "learning_rate": 2.3617147372580008e-05, "loss": 0.1806, "step": 20530 }, { "epoch": 4.06, "learning_rate": 2.3567759778743582e-05, "loss": 0.1715, "step": 20540 }, { "epoch": 4.06, "learning_rate": 2.3518372184907152e-05, "loss": 0.1911, "step": 20550 }, { "epoch": 4.06, "learning_rate": 2.3468984591070725e-05, "loss": 0.1751, "step": 20560 }, { "epoch": 4.06, "learning_rate": 2.3419596997234295e-05, "loss": 0.1958, "step": 20570 }, { "epoch": 4.06, "learning_rate": 2.3370209403397866e-05, "loss": 0.1886, "step": 20580 }, { "epoch": 4.07, "learning_rate": 2.332082180956144e-05, "loss": 0.2096, "step": 20590 }, { "epoch": 4.07, "learning_rate": 2.327143421572501e-05, "loss": 0.1853, "step": 20600 }, { "epoch": 4.07, "learning_rate": 2.3222046621888583e-05, "loss": 0.1999, "step": 20610 }, { "epoch": 4.07, "learning_rate": 2.3172659028052153e-05, "loss": 0.1729, "step": 20620 }, { "epoch": 4.07, "learning_rate": 2.3123271434215726e-05, "loss": 0.1689, "step": 20630 }, { "epoch": 4.08, "learning_rate": 2.30738838403793e-05, "loss": 0.2011, "step": 20640 }, { "epoch": 4.08, "learning_rate": 2.302449624654287e-05, "loss": 0.1817, "step": 20650 }, { "epoch": 4.08, "learning_rate": 2.2975108652706443e-05, "loss": 0.1762, "step": 20660 }, { "epoch": 4.08, "learning_rate": 2.292572105887001e-05, "loss": 0.1958, "step": 20670 }, { "epoch": 4.08, "learning_rate": 2.2876333465033584e-05, "loss": 0.188, "step": 20680 }, { "epoch": 4.09, "learning_rate": 2.2826945871197157e-05, "loss": 0.189, "step": 20690 }, { "epoch": 4.09, "learning_rate": 2.2777558277360727e-05, "loss": 0.1698, "step": 20700 }, { "epoch": 4.09, "learning_rate": 2.27281706835243e-05, "loss": 0.1896, "step": 20710 }, { "epoch": 4.09, "learning_rate": 2.267878308968787e-05, "loss": 0.1801, "step": 20720 }, { "epoch": 4.09, "learning_rate": 2.2629395495851444e-05, "loss": 0.1773, "step": 20730 }, { "epoch": 4.1, "learning_rate": 2.2580007902015014e-05, "loss": 0.1919, "step": 20740 }, { "epoch": 4.1, "learning_rate": 2.2530620308178588e-05, "loss": 0.1873, "step": 20750 }, { "epoch": 4.1, "learning_rate": 2.2481232714342158e-05, "loss": 0.1888, "step": 20760 }, { "epoch": 4.1, "learning_rate": 2.2431845120505728e-05, "loss": 0.1849, "step": 20770 }, { "epoch": 4.1, "learning_rate": 2.23824575266693e-05, "loss": 0.194, "step": 20780 }, { "epoch": 4.11, "learning_rate": 2.2333069932832872e-05, "loss": 0.1901, "step": 20790 }, { "epoch": 4.11, "learning_rate": 2.2283682338996445e-05, "loss": 0.1836, "step": 20800 }, { "epoch": 4.11, "learning_rate": 2.2234294745160015e-05, "loss": 0.1926, "step": 20810 }, { "epoch": 4.11, "learning_rate": 2.218490715132359e-05, "loss": 0.1726, "step": 20820 }, { "epoch": 4.11, "learning_rate": 2.2135519557487162e-05, "loss": 0.1776, "step": 20830 }, { "epoch": 4.12, "learning_rate": 2.2086131963650732e-05, "loss": 0.1832, "step": 20840 }, { "epoch": 4.12, "learning_rate": 2.2036744369814306e-05, "loss": 0.1795, "step": 20850 }, { "epoch": 4.12, "learning_rate": 2.1987356775977873e-05, "loss": 0.1785, "step": 20860 }, { "epoch": 4.12, "learning_rate": 2.1937969182141446e-05, "loss": 0.1771, "step": 20870 }, { "epoch": 4.12, "learning_rate": 2.188858158830502e-05, "loss": 0.1837, "step": 20880 }, { "epoch": 4.13, "learning_rate": 2.183919399446859e-05, "loss": 0.1812, "step": 20890 }, { "epoch": 4.13, "learning_rate": 2.1789806400632163e-05, "loss": 0.1909, "step": 20900 }, { "epoch": 4.13, "learning_rate": 2.1740418806795733e-05, "loss": 0.1689, "step": 20910 }, { "epoch": 4.13, "learning_rate": 2.1691031212959307e-05, "loss": 0.1983, "step": 20920 }, { "epoch": 4.13, "learning_rate": 2.1641643619122877e-05, "loss": 0.1856, "step": 20930 }, { "epoch": 4.14, "learning_rate": 2.159225602528645e-05, "loss": 0.1838, "step": 20940 }, { "epoch": 4.14, "learning_rate": 2.154286843145002e-05, "loss": 0.1851, "step": 20950 }, { "epoch": 4.14, "learning_rate": 2.149348083761359e-05, "loss": 0.1917, "step": 20960 }, { "epoch": 4.14, "learning_rate": 2.1444093243777164e-05, "loss": 0.1975, "step": 20970 }, { "epoch": 4.14, "learning_rate": 2.1394705649940734e-05, "loss": 0.2196, "step": 20980 }, { "epoch": 4.15, "learning_rate": 2.1345318056104308e-05, "loss": 0.1942, "step": 20990 }, { "epoch": 4.15, "learning_rate": 2.129593046226788e-05, "loss": 0.2145, "step": 21000 }, { "epoch": 4.15, "learning_rate": 2.124654286843145e-05, "loss": 0.1884, "step": 21010 }, { "epoch": 4.15, "learning_rate": 2.1197155274595025e-05, "loss": 0.1789, "step": 21020 }, { "epoch": 4.15, "learning_rate": 2.1147767680758595e-05, "loss": 0.1989, "step": 21030 }, { "epoch": 4.16, "learning_rate": 2.1098380086922165e-05, "loss": 0.1898, "step": 21040 }, { "epoch": 4.16, "learning_rate": 2.1048992493085735e-05, "loss": 0.1995, "step": 21050 }, { "epoch": 4.16, "learning_rate": 2.099960489924931e-05, "loss": 0.1841, "step": 21060 }, { "epoch": 4.16, "learning_rate": 2.0950217305412882e-05, "loss": 0.1972, "step": 21070 }, { "epoch": 4.16, "learning_rate": 2.0900829711576452e-05, "loss": 0.1879, "step": 21080 }, { "epoch": 4.17, "learning_rate": 2.0851442117740026e-05, "loss": 0.1769, "step": 21090 }, { "epoch": 4.17, "learning_rate": 2.0802054523903596e-05, "loss": 0.1949, "step": 21100 }, { "epoch": 4.17, "learning_rate": 2.075266693006717e-05, "loss": 0.1905, "step": 21110 }, { "epoch": 4.17, "learning_rate": 2.070327933623074e-05, "loss": 0.1904, "step": 21120 }, { "epoch": 4.17, "learning_rate": 2.0653891742394313e-05, "loss": 0.1832, "step": 21130 }, { "epoch": 4.18, "learning_rate": 2.0604504148557883e-05, "loss": 0.182, "step": 21140 }, { "epoch": 4.18, "learning_rate": 2.0555116554721453e-05, "loss": 0.1911, "step": 21150 }, { "epoch": 4.18, "learning_rate": 2.0505728960885027e-05, "loss": 0.1768, "step": 21160 }, { "epoch": 4.18, "learning_rate": 2.0456341367048597e-05, "loss": 0.1878, "step": 21170 }, { "epoch": 4.18, "learning_rate": 2.040695377321217e-05, "loss": 0.2035, "step": 21180 }, { "epoch": 4.19, "learning_rate": 2.0357566179375744e-05, "loss": 0.1999, "step": 21190 }, { "epoch": 4.19, "learning_rate": 2.0308178585539314e-05, "loss": 0.1813, "step": 21200 }, { "epoch": 4.19, "learning_rate": 2.0258790991702887e-05, "loss": 0.1783, "step": 21210 }, { "epoch": 4.19, "learning_rate": 2.0209403397866457e-05, "loss": 0.1919, "step": 21220 }, { "epoch": 4.19, "learning_rate": 2.0160015804030028e-05, "loss": 0.1782, "step": 21230 }, { "epoch": 4.2, "learning_rate": 2.01106282101936e-05, "loss": 0.1945, "step": 21240 }, { "epoch": 4.2, "learning_rate": 2.006124061635717e-05, "loss": 0.2023, "step": 21250 }, { "epoch": 4.2, "learning_rate": 2.0011853022520745e-05, "loss": 0.1889, "step": 21260 }, { "epoch": 4.2, "learning_rate": 1.9962465428684315e-05, "loss": 0.167, "step": 21270 }, { "epoch": 4.2, "learning_rate": 1.9913077834847888e-05, "loss": 0.1846, "step": 21280 }, { "epoch": 4.21, "learning_rate": 1.986369024101146e-05, "loss": 0.1909, "step": 21290 }, { "epoch": 4.21, "learning_rate": 1.9814302647175032e-05, "loss": 0.1865, "step": 21300 }, { "epoch": 4.21, "learning_rate": 1.9764915053338602e-05, "loss": 0.1895, "step": 21310 }, { "epoch": 4.21, "learning_rate": 1.9715527459502172e-05, "loss": 0.1956, "step": 21320 }, { "epoch": 4.21, "learning_rate": 1.9666139865665746e-05, "loss": 0.1756, "step": 21330 }, { "epoch": 4.22, "learning_rate": 1.9616752271829316e-05, "loss": 0.179, "step": 21340 }, { "epoch": 4.22, "learning_rate": 1.956736467799289e-05, "loss": 0.1948, "step": 21350 }, { "epoch": 4.22, "learning_rate": 1.951797708415646e-05, "loss": 0.1967, "step": 21360 }, { "epoch": 4.22, "learning_rate": 1.9473528249703675e-05, "loss": 0.182, "step": 21370 }, { "epoch": 4.22, "learning_rate": 1.942414065586725e-05, "loss": 0.1942, "step": 21380 }, { "epoch": 4.22, "learning_rate": 1.937475306203082e-05, "loss": 0.2012, "step": 21390 }, { "epoch": 4.23, "learning_rate": 1.932536546819439e-05, "loss": 0.178, "step": 21400 }, { "epoch": 4.23, "learning_rate": 1.9275977874357963e-05, "loss": 0.1869, "step": 21410 }, { "epoch": 4.23, "learning_rate": 1.9226590280521533e-05, "loss": 0.1978, "step": 21420 }, { "epoch": 4.23, "learning_rate": 1.9177202686685106e-05, "loss": 0.1859, "step": 21430 }, { "epoch": 4.23, "learning_rate": 1.9127815092848676e-05, "loss": 0.182, "step": 21440 }, { "epoch": 4.24, "learning_rate": 1.907842749901225e-05, "loss": 0.1926, "step": 21450 }, { "epoch": 4.24, "learning_rate": 1.9029039905175823e-05, "loss": 0.1916, "step": 21460 }, { "epoch": 4.24, "learning_rate": 1.8979652311339393e-05, "loss": 0.1681, "step": 21470 }, { "epoch": 4.24, "learning_rate": 1.8930264717502964e-05, "loss": 0.2204, "step": 21480 }, { "epoch": 4.24, "learning_rate": 1.8880877123666534e-05, "loss": 0.1922, "step": 21490 }, { "epoch": 4.25, "learning_rate": 1.8831489529830107e-05, "loss": 0.182, "step": 21500 }, { "epoch": 4.25, "learning_rate": 1.8782101935993677e-05, "loss": 0.1931, "step": 21510 }, { "epoch": 4.25, "learning_rate": 1.873271434215725e-05, "loss": 0.1923, "step": 21520 }, { "epoch": 4.25, "learning_rate": 1.8683326748320824e-05, "loss": 0.1803, "step": 21530 }, { "epoch": 4.25, "learning_rate": 1.8633939154484394e-05, "loss": 0.2077, "step": 21540 }, { "epoch": 4.26, "learning_rate": 1.8584551560647968e-05, "loss": 0.1874, "step": 21550 }, { "epoch": 4.26, "learning_rate": 1.8535163966811538e-05, "loss": 0.1831, "step": 21560 }, { "epoch": 4.26, "learning_rate": 1.8485776372975108e-05, "loss": 0.1912, "step": 21570 }, { "epoch": 4.26, "learning_rate": 1.843638877913868e-05, "loss": 0.1894, "step": 21580 }, { "epoch": 4.26, "learning_rate": 1.838700118530225e-05, "loss": 0.1834, "step": 21590 }, { "epoch": 4.27, "learning_rate": 1.8337613591465825e-05, "loss": 0.1994, "step": 21600 }, { "epoch": 4.27, "learning_rate": 1.8288225997629395e-05, "loss": 0.1862, "step": 21610 }, { "epoch": 4.27, "learning_rate": 1.823883840379297e-05, "loss": 0.183, "step": 21620 }, { "epoch": 4.27, "learning_rate": 1.818945080995654e-05, "loss": 0.1888, "step": 21630 }, { "epoch": 4.27, "learning_rate": 1.8140063216120112e-05, "loss": 0.1912, "step": 21640 }, { "epoch": 4.28, "learning_rate": 1.8090675622283686e-05, "loss": 0.1781, "step": 21650 }, { "epoch": 4.28, "learning_rate": 1.8041288028447256e-05, "loss": 0.188, "step": 21660 }, { "epoch": 4.28, "learning_rate": 1.7991900434610826e-05, "loss": 0.1811, "step": 21670 }, { "epoch": 4.28, "learning_rate": 1.7942512840774396e-05, "loss": 0.1772, "step": 21680 }, { "epoch": 4.28, "learning_rate": 1.789312524693797e-05, "loss": 0.191, "step": 21690 }, { "epoch": 4.29, "learning_rate": 1.7843737653101543e-05, "loss": 0.1824, "step": 21700 }, { "epoch": 4.29, "learning_rate": 1.7794350059265113e-05, "loss": 0.1767, "step": 21710 }, { "epoch": 4.29, "learning_rate": 1.7744962465428687e-05, "loss": 0.1884, "step": 21720 }, { "epoch": 4.29, "learning_rate": 1.7695574871592257e-05, "loss": 0.2063, "step": 21730 }, { "epoch": 4.29, "learning_rate": 1.764618727775583e-05, "loss": 0.1772, "step": 21740 }, { "epoch": 4.3, "learning_rate": 1.75967996839194e-05, "loss": 0.1933, "step": 21750 }, { "epoch": 4.3, "learning_rate": 1.754741209008297e-05, "loss": 0.189, "step": 21760 }, { "epoch": 4.3, "learning_rate": 1.7498024496246544e-05, "loss": 0.1847, "step": 21770 }, { "epoch": 4.3, "learning_rate": 1.7448636902410114e-05, "loss": 0.1963, "step": 21780 }, { "epoch": 4.3, "learning_rate": 1.7399249308573688e-05, "loss": 0.1652, "step": 21790 }, { "epoch": 4.31, "learning_rate": 1.7349861714737258e-05, "loss": 0.171, "step": 21800 }, { "epoch": 4.31, "learning_rate": 1.730047412090083e-05, "loss": 0.1915, "step": 21810 }, { "epoch": 4.31, "learning_rate": 1.72510865270644e-05, "loss": 0.1782, "step": 21820 }, { "epoch": 4.31, "learning_rate": 1.7201698933227975e-05, "loss": 0.1753, "step": 21830 }, { "epoch": 4.31, "learning_rate": 1.715231133939155e-05, "loss": 0.1956, "step": 21840 }, { "epoch": 4.32, "learning_rate": 1.7102923745555115e-05, "loss": 0.1777, "step": 21850 }, { "epoch": 4.32, "learning_rate": 1.705353615171869e-05, "loss": 0.1903, "step": 21860 }, { "epoch": 4.32, "learning_rate": 1.700414855788226e-05, "loss": 0.1769, "step": 21870 }, { "epoch": 4.32, "learning_rate": 1.6954760964045832e-05, "loss": 0.1854, "step": 21880 }, { "epoch": 4.32, "learning_rate": 1.6905373370209406e-05, "loss": 0.2025, "step": 21890 }, { "epoch": 4.33, "learning_rate": 1.6855985776372976e-05, "loss": 0.1985, "step": 21900 }, { "epoch": 4.33, "learning_rate": 1.680659818253655e-05, "loss": 0.1809, "step": 21910 }, { "epoch": 4.33, "learning_rate": 1.675721058870012e-05, "loss": 0.1913, "step": 21920 }, { "epoch": 4.33, "learning_rate": 1.6707822994863693e-05, "loss": 0.1748, "step": 21930 }, { "epoch": 4.33, "learning_rate": 1.6658435401027263e-05, "loss": 0.1939, "step": 21940 }, { "epoch": 4.34, "learning_rate": 1.6609047807190833e-05, "loss": 0.1721, "step": 21950 }, { "epoch": 4.34, "learning_rate": 1.6559660213354407e-05, "loss": 0.1758, "step": 21960 }, { "epoch": 4.34, "learning_rate": 1.6510272619517977e-05, "loss": 0.185, "step": 21970 }, { "epoch": 4.34, "learning_rate": 1.646088502568155e-05, "loss": 0.1805, "step": 21980 }, { "epoch": 4.34, "learning_rate": 1.641149743184512e-05, "loss": 0.2059, "step": 21990 }, { "epoch": 4.35, "learning_rate": 1.6362109838008694e-05, "loss": 0.1904, "step": 22000 }, { "epoch": 4.35, "learning_rate": 1.6312722244172264e-05, "loss": 0.1811, "step": 22010 }, { "epoch": 4.35, "learning_rate": 1.6263334650335837e-05, "loss": 0.1944, "step": 22020 }, { "epoch": 4.35, "learning_rate": 1.621394705649941e-05, "loss": 0.1856, "step": 22030 }, { "epoch": 4.35, "learning_rate": 1.6169498222046624e-05, "loss": 0.1895, "step": 22040 }, { "epoch": 4.36, "learning_rate": 1.6120110628210194e-05, "loss": 0.1796, "step": 22050 }, { "epoch": 4.36, "learning_rate": 1.6070723034373767e-05, "loss": 0.1707, "step": 22060 }, { "epoch": 4.36, "learning_rate": 1.6021335440537337e-05, "loss": 0.187, "step": 22070 }, { "epoch": 4.36, "learning_rate": 1.597194784670091e-05, "loss": 0.2011, "step": 22080 }, { "epoch": 4.36, "learning_rate": 1.592256025286448e-05, "loss": 0.1804, "step": 22090 }, { "epoch": 4.37, "learning_rate": 1.587317265902805e-05, "loss": 0.1848, "step": 22100 }, { "epoch": 4.37, "learning_rate": 1.5823785065191625e-05, "loss": 0.203, "step": 22110 }, { "epoch": 4.37, "learning_rate": 1.5774397471355195e-05, "loss": 0.1819, "step": 22120 }, { "epoch": 4.37, "learning_rate": 1.5725009877518768e-05, "loss": 0.1612, "step": 22130 }, { "epoch": 4.37, "learning_rate": 1.567562228368234e-05, "loss": 0.174, "step": 22140 }, { "epoch": 4.38, "learning_rate": 1.5626234689845912e-05, "loss": 0.1847, "step": 22150 }, { "epoch": 4.38, "learning_rate": 1.5576847096009485e-05, "loss": 0.1849, "step": 22160 }, { "epoch": 4.38, "learning_rate": 1.5527459502173055e-05, "loss": 0.174, "step": 22170 }, { "epoch": 4.38, "learning_rate": 1.547807190833663e-05, "loss": 0.1802, "step": 22180 }, { "epoch": 4.38, "learning_rate": 1.5428684314500196e-05, "loss": 0.1907, "step": 22190 }, { "epoch": 4.38, "learning_rate": 1.537929672066377e-05, "loss": 0.192, "step": 22200 }, { "epoch": 4.39, "learning_rate": 1.532990912682734e-05, "loss": 0.1872, "step": 22210 }, { "epoch": 4.39, "learning_rate": 1.5280521532990913e-05, "loss": 0.1816, "step": 22220 }, { "epoch": 4.39, "learning_rate": 1.5231133939154485e-05, "loss": 0.201, "step": 22230 }, { "epoch": 4.39, "learning_rate": 1.5181746345318056e-05, "loss": 0.1647, "step": 22240 }, { "epoch": 4.39, "learning_rate": 1.5132358751481628e-05, "loss": 0.1927, "step": 22250 }, { "epoch": 4.4, "learning_rate": 1.5082971157645202e-05, "loss": 0.2083, "step": 22260 }, { "epoch": 4.4, "learning_rate": 1.5033583563808773e-05, "loss": 0.1713, "step": 22270 }, { "epoch": 4.4, "learning_rate": 1.4984195969972345e-05, "loss": 0.1744, "step": 22280 }, { "epoch": 4.4, "learning_rate": 1.4934808376135914e-05, "loss": 0.201, "step": 22290 }, { "epoch": 4.4, "learning_rate": 1.4885420782299485e-05, "loss": 0.1935, "step": 22300 }, { "epoch": 4.41, "learning_rate": 1.4836033188463059e-05, "loss": 0.1887, "step": 22310 }, { "epoch": 4.41, "learning_rate": 1.478664559462663e-05, "loss": 0.189, "step": 22320 }, { "epoch": 4.41, "learning_rate": 1.4737258000790203e-05, "loss": 0.175, "step": 22330 }, { "epoch": 4.41, "learning_rate": 1.4687870406953774e-05, "loss": 0.2061, "step": 22340 }, { "epoch": 4.41, "learning_rate": 1.4638482813117346e-05, "loss": 0.1866, "step": 22350 }, { "epoch": 4.42, "learning_rate": 1.4589095219280918e-05, "loss": 0.1853, "step": 22360 }, { "epoch": 4.42, "learning_rate": 1.453970762544449e-05, "loss": 0.1842, "step": 22370 }, { "epoch": 4.42, "learning_rate": 1.449032003160806e-05, "loss": 0.1909, "step": 22380 }, { "epoch": 4.42, "learning_rate": 1.4440932437771632e-05, "loss": 0.1953, "step": 22390 }, { "epoch": 4.42, "learning_rate": 1.4391544843935203e-05, "loss": 0.1903, "step": 22400 }, { "epoch": 4.43, "learning_rate": 1.4342157250098775e-05, "loss": 0.1858, "step": 22410 }, { "epoch": 4.43, "learning_rate": 1.4292769656262347e-05, "loss": 0.2018, "step": 22420 }, { "epoch": 4.43, "learning_rate": 1.4243382062425919e-05, "loss": 0.1787, "step": 22430 }, { "epoch": 4.43, "learning_rate": 1.4193994468589492e-05, "loss": 0.1821, "step": 22440 }, { "epoch": 4.43, "learning_rate": 1.4144606874753064e-05, "loss": 0.1813, "step": 22450 }, { "epoch": 4.44, "learning_rate": 1.4095219280916636e-05, "loss": 0.1895, "step": 22460 }, { "epoch": 4.44, "learning_rate": 1.4045831687080204e-05, "loss": 0.1827, "step": 22470 }, { "epoch": 4.44, "learning_rate": 1.3996444093243776e-05, "loss": 0.2002, "step": 22480 }, { "epoch": 4.44, "learning_rate": 1.3947056499407348e-05, "loss": 0.1798, "step": 22490 }, { "epoch": 4.44, "learning_rate": 1.3897668905570921e-05, "loss": 0.1854, "step": 22500 }, { "epoch": 4.45, "learning_rate": 1.3848281311734493e-05, "loss": 0.183, "step": 22510 }, { "epoch": 4.45, "learning_rate": 1.3798893717898065e-05, "loss": 0.1823, "step": 22520 }, { "epoch": 4.45, "learning_rate": 1.3749506124061637e-05, "loss": 0.1892, "step": 22530 }, { "epoch": 4.45, "learning_rate": 1.3700118530225209e-05, "loss": 0.1873, "step": 22540 }, { "epoch": 4.45, "learning_rate": 1.365073093638878e-05, "loss": 0.187, "step": 22550 }, { "epoch": 4.46, "learning_rate": 1.3601343342552352e-05, "loss": 0.1782, "step": 22560 }, { "epoch": 4.46, "learning_rate": 1.3551955748715922e-05, "loss": 0.1758, "step": 22570 }, { "epoch": 4.46, "learning_rate": 1.3502568154879494e-05, "loss": 0.1842, "step": 22580 }, { "epoch": 4.46, "learning_rate": 1.3453180561043066e-05, "loss": 0.18, "step": 22590 }, { "epoch": 4.46, "learning_rate": 1.3403792967206638e-05, "loss": 0.1935, "step": 22600 }, { "epoch": 4.47, "learning_rate": 1.335440537337021e-05, "loss": 0.1938, "step": 22610 }, { "epoch": 4.47, "learning_rate": 1.3305017779533781e-05, "loss": 0.2138, "step": 22620 }, { "epoch": 4.47, "learning_rate": 1.3255630185697355e-05, "loss": 0.1856, "step": 22630 }, { "epoch": 4.47, "learning_rate": 1.3206242591860927e-05, "loss": 0.1783, "step": 22640 }, { "epoch": 4.47, "learning_rate": 1.3156854998024499e-05, "loss": 0.1851, "step": 22650 }, { "epoch": 4.48, "learning_rate": 1.3107467404188067e-05, "loss": 0.1783, "step": 22660 }, { "epoch": 4.48, "learning_rate": 1.3058079810351639e-05, "loss": 0.1683, "step": 22670 }, { "epoch": 4.48, "learning_rate": 1.3008692216515212e-05, "loss": 0.1865, "step": 22680 }, { "epoch": 4.48, "learning_rate": 1.2959304622678784e-05, "loss": 0.1842, "step": 22690 }, { "epoch": 4.48, "learning_rate": 1.2909917028842356e-05, "loss": 0.1835, "step": 22700 }, { "epoch": 4.49, "learning_rate": 1.2860529435005928e-05, "loss": 0.1733, "step": 22710 }, { "epoch": 4.49, "learning_rate": 1.28111418411695e-05, "loss": 0.1885, "step": 22720 }, { "epoch": 4.49, "learning_rate": 1.2761754247333071e-05, "loss": 0.1931, "step": 22730 }, { "epoch": 4.49, "learning_rate": 1.2712366653496643e-05, "loss": 0.1758, "step": 22740 }, { "epoch": 4.49, "learning_rate": 1.2662979059660213e-05, "loss": 0.18, "step": 22750 }, { "epoch": 4.5, "learning_rate": 1.2613591465823785e-05, "loss": 0.1717, "step": 22760 }, { "epoch": 4.5, "learning_rate": 1.2564203871987357e-05, "loss": 0.1922, "step": 22770 }, { "epoch": 4.5, "learning_rate": 1.2514816278150929e-05, "loss": 0.1882, "step": 22780 }, { "epoch": 4.5, "learning_rate": 1.24654286843145e-05, "loss": 0.1946, "step": 22790 }, { "epoch": 4.5, "learning_rate": 1.2416041090478072e-05, "loss": 0.1801, "step": 22800 }, { "epoch": 4.51, "learning_rate": 1.2366653496641644e-05, "loss": 0.1655, "step": 22810 }, { "epoch": 4.51, "learning_rate": 1.2317265902805216e-05, "loss": 0.1955, "step": 22820 }, { "epoch": 4.51, "learning_rate": 1.2267878308968788e-05, "loss": 0.1937, "step": 22830 }, { "epoch": 4.51, "learning_rate": 1.221849071513236e-05, "loss": 0.1864, "step": 22840 }, { "epoch": 4.51, "learning_rate": 1.2169103121295931e-05, "loss": 0.1585, "step": 22850 }, { "epoch": 4.52, "learning_rate": 1.2119715527459503e-05, "loss": 0.1897, "step": 22860 }, { "epoch": 4.52, "learning_rate": 1.2070327933623075e-05, "loss": 0.1858, "step": 22870 }, { "epoch": 4.52, "learning_rate": 1.2020940339786647e-05, "loss": 0.16, "step": 22880 }, { "epoch": 4.52, "learning_rate": 1.1971552745950218e-05, "loss": 0.1782, "step": 22890 }, { "epoch": 4.52, "learning_rate": 1.192216515211379e-05, "loss": 0.2001, "step": 22900 }, { "epoch": 4.53, "learning_rate": 1.187277755827736e-05, "loss": 0.1794, "step": 22910 }, { "epoch": 4.53, "learning_rate": 1.1823389964440932e-05, "loss": 0.1668, "step": 22920 }, { "epoch": 4.53, "learning_rate": 1.1774002370604504e-05, "loss": 0.1832, "step": 22930 }, { "epoch": 4.53, "learning_rate": 1.1724614776768077e-05, "loss": 0.1803, "step": 22940 }, { "epoch": 4.53, "learning_rate": 1.167522718293165e-05, "loss": 0.184, "step": 22950 }, { "epoch": 4.54, "learning_rate": 1.162583958909522e-05, "loss": 0.1686, "step": 22960 }, { "epoch": 4.54, "learning_rate": 1.1576451995258791e-05, "loss": 0.2012, "step": 22970 }, { "epoch": 4.54, "learning_rate": 1.1527064401422363e-05, "loss": 0.1874, "step": 22980 }, { "epoch": 4.54, "learning_rate": 1.1477676807585935e-05, "loss": 0.1701, "step": 22990 }, { "epoch": 4.54, "learning_rate": 1.1428289213749508e-05, "loss": 0.186, "step": 23000 }, { "epoch": 4.54, "learning_rate": 1.1378901619913078e-05, "loss": 0.1987, "step": 23010 }, { "epoch": 4.55, "learning_rate": 1.132951402607665e-05, "loss": 0.1953, "step": 23020 }, { "epoch": 4.55, "learning_rate": 1.1280126432240222e-05, "loss": 0.1809, "step": 23030 }, { "epoch": 4.55, "learning_rate": 1.1230738838403794e-05, "loss": 0.1726, "step": 23040 }, { "epoch": 4.55, "learning_rate": 1.1181351244567364e-05, "loss": 0.1969, "step": 23050 }, { "epoch": 4.55, "learning_rate": 1.1131963650730937e-05, "loss": 0.1842, "step": 23060 }, { "epoch": 4.56, "learning_rate": 1.1082576056894509e-05, "loss": 0.1874, "step": 23070 }, { "epoch": 4.56, "learning_rate": 1.1033188463058081e-05, "loss": 0.1826, "step": 23080 }, { "epoch": 4.56, "learning_rate": 1.0983800869221653e-05, "loss": 0.1864, "step": 23090 }, { "epoch": 4.56, "learning_rate": 1.0934413275385223e-05, "loss": 0.1967, "step": 23100 }, { "epoch": 4.56, "learning_rate": 1.0885025681548795e-05, "loss": 0.2023, "step": 23110 }, { "epoch": 4.57, "learning_rate": 1.0835638087712368e-05, "loss": 0.1886, "step": 23120 }, { "epoch": 4.57, "learning_rate": 1.078625049387594e-05, "loss": 0.1717, "step": 23130 }, { "epoch": 4.57, "learning_rate": 1.073686290003951e-05, "loss": 0.1783, "step": 23140 }, { "epoch": 4.57, "learning_rate": 1.0687475306203082e-05, "loss": 0.1727, "step": 23150 }, { "epoch": 4.57, "learning_rate": 1.0638087712366654e-05, "loss": 0.2023, "step": 23160 }, { "epoch": 4.58, "learning_rate": 1.0588700118530225e-05, "loss": 0.187, "step": 23170 }, { "epoch": 4.58, "learning_rate": 1.0539312524693797e-05, "loss": 0.1782, "step": 23180 }, { "epoch": 4.58, "learning_rate": 1.0489924930857369e-05, "loss": 0.1801, "step": 23190 }, { "epoch": 4.58, "learning_rate": 1.044053733702094e-05, "loss": 0.1813, "step": 23200 }, { "epoch": 4.58, "learning_rate": 1.0391149743184513e-05, "loss": 0.1837, "step": 23210 }, { "epoch": 4.59, "learning_rate": 1.0341762149348084e-05, "loss": 0.1751, "step": 23220 }, { "epoch": 4.59, "learning_rate": 1.0292374555511656e-05, "loss": 0.1772, "step": 23230 }, { "epoch": 4.59, "learning_rate": 1.0242986961675228e-05, "loss": 0.1939, "step": 23240 }, { "epoch": 4.59, "learning_rate": 1.01935993678388e-05, "loss": 0.1821, "step": 23250 }, { "epoch": 4.59, "learning_rate": 1.0144211774002372e-05, "loss": 0.1724, "step": 23260 }, { "epoch": 4.6, "learning_rate": 1.0094824180165943e-05, "loss": 0.2009, "step": 23270 }, { "epoch": 4.6, "learning_rate": 1.0045436586329514e-05, "loss": 0.1665, "step": 23280 }, { "epoch": 4.6, "learning_rate": 9.996048992493085e-06, "loss": 0.1721, "step": 23290 }, { "epoch": 4.6, "learning_rate": 9.946661398656657e-06, "loss": 0.1952, "step": 23300 }, { "epoch": 4.6, "learning_rate": 9.89727380482023e-06, "loss": 0.181, "step": 23310 }, { "epoch": 4.61, "learning_rate": 9.847886210983802e-06, "loss": 0.187, "step": 23320 }, { "epoch": 4.61, "learning_rate": 9.798498617147373e-06, "loss": 0.1726, "step": 23330 }, { "epoch": 4.61, "learning_rate": 9.749111023310944e-06, "loss": 0.1819, "step": 23340 }, { "epoch": 4.61, "learning_rate": 9.699723429474516e-06, "loss": 0.1892, "step": 23350 }, { "epoch": 4.61, "learning_rate": 9.650335835638088e-06, "loss": 0.1619, "step": 23360 }, { "epoch": 4.62, "learning_rate": 9.60094824180166e-06, "loss": 0.1696, "step": 23370 }, { "epoch": 4.62, "learning_rate": 9.551560647965232e-06, "loss": 0.1852, "step": 23380 }, { "epoch": 4.62, "learning_rate": 9.502173054128803e-06, "loss": 0.1964, "step": 23390 }, { "epoch": 4.62, "learning_rate": 9.452785460292375e-06, "loss": 0.1814, "step": 23400 }, { "epoch": 4.62, "learning_rate": 9.403397866455947e-06, "loss": 0.1591, "step": 23410 }, { "epoch": 4.63, "learning_rate": 9.354010272619517e-06, "loss": 0.1793, "step": 23420 }, { "epoch": 4.63, "learning_rate": 9.30462267878309e-06, "loss": 0.1918, "step": 23430 }, { "epoch": 4.63, "learning_rate": 9.255235084946662e-06, "loss": 0.1817, "step": 23440 }, { "epoch": 4.63, "learning_rate": 9.205847491110234e-06, "loss": 0.174, "step": 23450 }, { "epoch": 4.63, "learning_rate": 9.156459897273806e-06, "loss": 0.1709, "step": 23460 }, { "epoch": 4.64, "learning_rate": 9.107072303437376e-06, "loss": 0.1943, "step": 23470 }, { "epoch": 4.64, "learning_rate": 9.057684709600948e-06, "loss": 0.1875, "step": 23480 }, { "epoch": 4.64, "learning_rate": 9.00829711576452e-06, "loss": 0.1808, "step": 23490 }, { "epoch": 4.64, "learning_rate": 8.958909521928093e-06, "loss": 0.1641, "step": 23500 }, { "epoch": 4.64, "learning_rate": 8.909521928091665e-06, "loss": 0.1885, "step": 23510 }, { "epoch": 4.65, "learning_rate": 8.860134334255235e-06, "loss": 0.171, "step": 23520 }, { "epoch": 4.65, "learning_rate": 8.810746740418807e-06, "loss": 0.1791, "step": 23530 }, { "epoch": 4.65, "learning_rate": 8.761359146582379e-06, "loss": 0.1745, "step": 23540 }, { "epoch": 4.65, "learning_rate": 8.71197155274595e-06, "loss": 0.1751, "step": 23550 }, { "epoch": 4.65, "learning_rate": 8.662583958909522e-06, "loss": 0.1736, "step": 23560 }, { "epoch": 4.66, "learning_rate": 8.613196365073094e-06, "loss": 0.169, "step": 23570 }, { "epoch": 4.66, "learning_rate": 8.563808771236666e-06, "loss": 0.1835, "step": 23580 }, { "epoch": 4.66, "learning_rate": 8.514421177400238e-06, "loss": 0.1683, "step": 23590 }, { "epoch": 4.66, "learning_rate": 8.46503358356381e-06, "loss": 0.1647, "step": 23600 }, { "epoch": 4.66, "learning_rate": 8.41564598972738e-06, "loss": 0.179, "step": 23610 }, { "epoch": 4.67, "learning_rate": 8.366258395890953e-06, "loss": 0.165, "step": 23620 }, { "epoch": 4.67, "learning_rate": 8.316870802054525e-06, "loss": 0.1883, "step": 23630 }, { "epoch": 4.67, "learning_rate": 8.267483208218097e-06, "loss": 0.1778, "step": 23640 }, { "epoch": 4.67, "learning_rate": 8.218095614381667e-06, "loss": 0.1779, "step": 23650 }, { "epoch": 4.67, "learning_rate": 8.168708020545239e-06, "loss": 0.1888, "step": 23660 }, { "epoch": 4.68, "learning_rate": 8.11932042670881e-06, "loss": 0.1793, "step": 23670 }, { "epoch": 4.68, "learning_rate": 8.069932832872382e-06, "loss": 0.1763, "step": 23680 }, { "epoch": 4.68, "learning_rate": 8.020545239035956e-06, "loss": 0.1645, "step": 23690 }, { "epoch": 4.68, "learning_rate": 7.971157645199526e-06, "loss": 0.1984, "step": 23700 }, { "epoch": 4.68, "learning_rate": 7.921770051363098e-06, "loss": 0.193, "step": 23710 }, { "epoch": 4.69, "learning_rate": 7.87238245752667e-06, "loss": 0.1711, "step": 23720 }, { "epoch": 4.69, "learning_rate": 7.822994863690241e-06, "loss": 0.1749, "step": 23730 }, { "epoch": 4.69, "learning_rate": 7.773607269853813e-06, "loss": 0.1704, "step": 23740 }, { "epoch": 4.69, "learning_rate": 7.724219676017385e-06, "loss": 0.1783, "step": 23750 }, { "epoch": 4.69, "learning_rate": 7.674832082180957e-06, "loss": 0.1939, "step": 23760 }, { "epoch": 4.7, "learning_rate": 7.6254444883445284e-06, "loss": 0.1808, "step": 23770 }, { "epoch": 4.7, "learning_rate": 7.5760568945081e-06, "loss": 0.1899, "step": 23780 }, { "epoch": 4.7, "learning_rate": 7.526669300671671e-06, "loss": 0.1881, "step": 23790 }, { "epoch": 4.7, "learning_rate": 7.477281706835243e-06, "loss": 0.1889, "step": 23800 }, { "epoch": 4.7, "learning_rate": 7.427894112998815e-06, "loss": 0.1776, "step": 23810 }, { "epoch": 4.7, "learning_rate": 7.378506519162387e-06, "loss": 0.199, "step": 23820 }, { "epoch": 4.71, "learning_rate": 7.329118925325959e-06, "loss": 0.1817, "step": 23830 }, { "epoch": 4.71, "learning_rate": 7.279731331489529e-06, "loss": 0.1637, "step": 23840 }, { "epoch": 4.71, "learning_rate": 7.230343737653102e-06, "loss": 0.1759, "step": 23850 }, { "epoch": 4.71, "learning_rate": 7.180956143816674e-06, "loss": 0.2028, "step": 23860 }, { "epoch": 4.71, "learning_rate": 7.131568549980246e-06, "loss": 0.1766, "step": 23870 }, { "epoch": 4.72, "learning_rate": 7.082180956143817e-06, "loss": 0.1764, "step": 23880 }, { "epoch": 4.72, "learning_rate": 7.032793362307388e-06, "loss": 0.1877, "step": 23890 }, { "epoch": 4.72, "learning_rate": 6.98340576847096e-06, "loss": 0.186, "step": 23900 }, { "epoch": 4.72, "learning_rate": 6.934018174634532e-06, "loss": 0.1876, "step": 23910 }, { "epoch": 4.72, "learning_rate": 6.884630580798105e-06, "loss": 0.1898, "step": 23920 }, { "epoch": 4.73, "learning_rate": 6.835242986961675e-06, "loss": 0.1916, "step": 23930 }, { "epoch": 4.73, "learning_rate": 6.7858553931252465e-06, "loss": 0.1765, "step": 23940 }, { "epoch": 4.73, "learning_rate": 6.736467799288819e-06, "loss": 0.1874, "step": 23950 }, { "epoch": 4.73, "learning_rate": 6.687080205452391e-06, "loss": 0.1708, "step": 23960 }, { "epoch": 4.73, "learning_rate": 6.637692611615963e-06, "loss": 0.1719, "step": 23970 }, { "epoch": 4.74, "learning_rate": 6.588305017779534e-06, "loss": 0.177, "step": 23980 }, { "epoch": 4.74, "learning_rate": 6.5389174239431055e-06, "loss": 0.1736, "step": 23990 }, { "epoch": 4.74, "learning_rate": 6.489529830106677e-06, "loss": 0.176, "step": 24000 }, { "epoch": 4.74, "learning_rate": 6.44014223627025e-06, "loss": 0.1806, "step": 24010 }, { "epoch": 4.74, "learning_rate": 6.390754642433822e-06, "loss": 0.1728, "step": 24020 }, { "epoch": 4.75, "learning_rate": 6.341367048597392e-06, "loss": 0.1809, "step": 24030 }, { "epoch": 4.75, "learning_rate": 6.2919794547609645e-06, "loss": 0.194, "step": 24040 }, { "epoch": 4.75, "learning_rate": 6.242591860924536e-06, "loss": 0.1807, "step": 24050 }, { "epoch": 4.75, "learning_rate": 6.193204267088107e-06, "loss": 0.1785, "step": 24060 }, { "epoch": 4.75, "learning_rate": 6.14381667325168e-06, "loss": 0.1976, "step": 24070 }, { "epoch": 4.76, "learning_rate": 6.094429079415251e-06, "loss": 0.1682, "step": 24080 }, { "epoch": 4.76, "learning_rate": 6.045041485578823e-06, "loss": 0.1743, "step": 24090 }, { "epoch": 4.76, "learning_rate": 5.9956538917423945e-06, "loss": 0.1702, "step": 24100 }, { "epoch": 4.76, "learning_rate": 5.946266297905966e-06, "loss": 0.1859, "step": 24110 }, { "epoch": 4.76, "learning_rate": 5.896878704069538e-06, "loss": 0.1801, "step": 24120 }, { "epoch": 4.77, "learning_rate": 5.84749111023311e-06, "loss": 0.1765, "step": 24130 }, { "epoch": 4.77, "learning_rate": 5.798103516396682e-06, "loss": 0.1992, "step": 24140 }, { "epoch": 4.77, "learning_rate": 5.748715922560253e-06, "loss": 0.1815, "step": 24150 }, { "epoch": 4.77, "learning_rate": 5.6993283287238244e-06, "loss": 0.1852, "step": 24160 }, { "epoch": 4.77, "learning_rate": 5.649940734887397e-06, "loss": 0.1684, "step": 24170 }, { "epoch": 4.78, "learning_rate": 5.600553141050968e-06, "loss": 0.1999, "step": 24180 }, { "epoch": 4.78, "learning_rate": 5.55116554721454e-06, "loss": 0.1686, "step": 24190 }, { "epoch": 4.78, "learning_rate": 5.501777953378112e-06, "loss": 0.1724, "step": 24200 }, { "epoch": 4.78, "learning_rate": 5.4523903595416835e-06, "loss": 0.1715, "step": 24210 }, { "epoch": 4.78, "learning_rate": 5.403002765705254e-06, "loss": 0.1776, "step": 24220 }, { "epoch": 4.79, "learning_rate": 5.353615171868827e-06, "loss": 0.1864, "step": 24230 }, { "epoch": 4.79, "learning_rate": 5.304227578032399e-06, "loss": 0.1757, "step": 24240 }, { "epoch": 4.79, "learning_rate": 5.25483998419597e-06, "loss": 0.1817, "step": 24250 }, { "epoch": 4.79, "learning_rate": 5.2054523903595425e-06, "loss": 0.1491, "step": 24260 }, { "epoch": 4.79, "learning_rate": 5.156064796523113e-06, "loss": 0.1888, "step": 24270 }, { "epoch": 4.8, "learning_rate": 5.106677202686685e-06, "loss": 0.1802, "step": 24280 }, { "epoch": 4.8, "learning_rate": 5.057289608850257e-06, "loss": 0.1772, "step": 24290 }, { "epoch": 4.8, "learning_rate": 5.007902015013829e-06, "loss": 0.1772, "step": 24300 }, { "epoch": 4.8, "learning_rate": 4.958514421177401e-06, "loss": 0.1703, "step": 24310 }, { "epoch": 4.8, "learning_rate": 4.909126827340972e-06, "loss": 0.1652, "step": 24320 }, { "epoch": 4.81, "learning_rate": 4.859739233504544e-06, "loss": 0.1872, "step": 24330 }, { "epoch": 4.81, "learning_rate": 4.810351639668115e-06, "loss": 0.1929, "step": 24340 }, { "epoch": 4.81, "learning_rate": 4.760964045831688e-06, "loss": 0.1801, "step": 24350 }, { "epoch": 4.81, "learning_rate": 4.711576451995259e-06, "loss": 0.173, "step": 24360 }, { "epoch": 4.81, "learning_rate": 4.662188858158831e-06, "loss": 0.1828, "step": 24370 }, { "epoch": 4.82, "learning_rate": 4.612801264322402e-06, "loss": 0.1913, "step": 24380 }, { "epoch": 4.82, "learning_rate": 4.563413670485974e-06, "loss": 0.1825, "step": 24390 }, { "epoch": 4.82, "learning_rate": 4.514026076649546e-06, "loss": 0.1883, "step": 24400 }, { "epoch": 4.82, "learning_rate": 4.464638482813118e-06, "loss": 0.179, "step": 24410 }, { "epoch": 4.82, "learning_rate": 4.41525088897669e-06, "loss": 0.1648, "step": 24420 }, { "epoch": 4.83, "learning_rate": 4.3658632951402605e-06, "loss": 0.1854, "step": 24430 }, { "epoch": 4.83, "learning_rate": 4.316475701303832e-06, "loss": 0.1969, "step": 24440 }, { "epoch": 4.83, "learning_rate": 4.267088107467404e-06, "loss": 0.1736, "step": 24450 }, { "epoch": 4.83, "learning_rate": 4.217700513630976e-06, "loss": 0.186, "step": 24460 }, { "epoch": 4.83, "learning_rate": 4.168312919794548e-06, "loss": 0.1938, "step": 24470 }, { "epoch": 4.84, "learning_rate": 4.1189253259581196e-06, "loss": 0.1908, "step": 24480 }, { "epoch": 4.84, "learning_rate": 4.069537732121691e-06, "loss": 0.1877, "step": 24490 }, { "epoch": 4.84, "learning_rate": 4.020150138285262e-06, "loss": 0.1735, "step": 24500 }, { "epoch": 4.84, "learning_rate": 3.970762544448835e-06, "loss": 0.1693, "step": 24510 }, { "epoch": 4.84, "learning_rate": 3.921374950612406e-06, "loss": 0.1646, "step": 24520 }, { "epoch": 4.85, "learning_rate": 3.871987356775978e-06, "loss": 0.1776, "step": 24530 }, { "epoch": 4.85, "learning_rate": 3.82259976293955e-06, "loss": 0.1729, "step": 24540 }, { "epoch": 4.85, "learning_rate": 3.7732121691031213e-06, "loss": 0.1582, "step": 24550 }, { "epoch": 4.85, "learning_rate": 3.7238245752666935e-06, "loss": 0.1801, "step": 24560 }, { "epoch": 4.85, "learning_rate": 3.6744369814302645e-06, "loss": 0.1799, "step": 24570 }, { "epoch": 4.86, "learning_rate": 3.6250493875938367e-06, "loss": 0.1843, "step": 24580 }, { "epoch": 4.86, "learning_rate": 3.575661793757408e-06, "loss": 0.1702, "step": 24590 }, { "epoch": 4.86, "learning_rate": 3.52627419992098e-06, "loss": 0.1884, "step": 24600 }, { "epoch": 4.86, "learning_rate": 3.476886606084552e-06, "loss": 0.179, "step": 24610 }, { "epoch": 4.86, "learning_rate": 3.4274990122481235e-06, "loss": 0.2046, "step": 24620 }, { "epoch": 4.86, "learning_rate": 3.3781114184116953e-06, "loss": 0.179, "step": 24630 }, { "epoch": 4.87, "learning_rate": 3.3287238245752667e-06, "loss": 0.1849, "step": 24640 }, { "epoch": 4.87, "learning_rate": 3.2793362307388385e-06, "loss": 0.1817, "step": 24650 }, { "epoch": 4.87, "learning_rate": 3.22994863690241e-06, "loss": 0.1776, "step": 24660 }, { "epoch": 4.87, "learning_rate": 3.180561043065982e-06, "loss": 0.1689, "step": 24670 }, { "epoch": 4.87, "learning_rate": 3.131173449229554e-06, "loss": 0.1681, "step": 24680 }, { "epoch": 4.88, "learning_rate": 3.0817858553931253e-06, "loss": 0.1817, "step": 24690 }, { "epoch": 4.88, "learning_rate": 3.032398261556697e-06, "loss": 0.19, "step": 24700 }, { "epoch": 4.88, "learning_rate": 2.983010667720269e-06, "loss": 0.1689, "step": 24710 }, { "epoch": 4.88, "learning_rate": 2.9336230738838407e-06, "loss": 0.1728, "step": 24720 }, { "epoch": 4.88, "learning_rate": 2.8842354800474125e-06, "loss": 0.1836, "step": 24730 }, { "epoch": 4.89, "learning_rate": 2.834847886210984e-06, "loss": 0.1775, "step": 24740 }, { "epoch": 4.89, "learning_rate": 2.7854602923745556e-06, "loss": 0.1994, "step": 24750 }, { "epoch": 4.89, "learning_rate": 2.7360726985381275e-06, "loss": 0.1708, "step": 24760 }, { "epoch": 4.89, "learning_rate": 2.686685104701699e-06, "loss": 0.1955, "step": 24770 }, { "epoch": 4.89, "learning_rate": 2.6372975108652706e-06, "loss": 0.1842, "step": 24780 }, { "epoch": 4.9, "learning_rate": 2.5879099170288424e-06, "loss": 0.1728, "step": 24790 }, { "epoch": 4.9, "learning_rate": 2.5385223231924142e-06, "loss": 0.1729, "step": 24800 }, { "epoch": 4.9, "learning_rate": 2.489134729355986e-06, "loss": 0.1741, "step": 24810 }, { "epoch": 4.9, "learning_rate": 2.4397471355195574e-06, "loss": 0.1715, "step": 24820 }, { "epoch": 4.9, "learning_rate": 2.3903595416831292e-06, "loss": 0.1774, "step": 24830 }, { "epoch": 4.91, "learning_rate": 2.340971947846701e-06, "loss": 0.1748, "step": 24840 }, { "epoch": 4.91, "learning_rate": 2.2915843540102724e-06, "loss": 0.1779, "step": 24850 }, { "epoch": 4.91, "learning_rate": 2.2421967601738446e-06, "loss": 0.1804, "step": 24860 }, { "epoch": 4.91, "learning_rate": 2.1928091663374164e-06, "loss": 0.1857, "step": 24870 }, { "epoch": 4.91, "learning_rate": 2.143421572500988e-06, "loss": 0.1768, "step": 24880 }, { "epoch": 4.92, "learning_rate": 2.0940339786645596e-06, "loss": 0.1559, "step": 24890 }, { "epoch": 4.92, "learning_rate": 2.0446463848281314e-06, "loss": 0.1702, "step": 24900 }, { "epoch": 4.92, "learning_rate": 1.9952587909917028e-06, "loss": 0.1816, "step": 24910 }, { "epoch": 4.92, "learning_rate": 1.9458711971552746e-06, "loss": 0.1655, "step": 24920 }, { "epoch": 4.92, "learning_rate": 1.8964836033188466e-06, "loss": 0.1879, "step": 24930 }, { "epoch": 4.93, "learning_rate": 1.8470960094824182e-06, "loss": 0.1613, "step": 24940 }, { "epoch": 4.93, "learning_rate": 1.79770841564599e-06, "loss": 0.188, "step": 24950 }, { "epoch": 4.93, "learning_rate": 1.7483208218095616e-06, "loss": 0.1798, "step": 24960 }, { "epoch": 4.93, "learning_rate": 1.6989332279731332e-06, "loss": 0.1982, "step": 24970 }, { "epoch": 4.93, "learning_rate": 1.649545634136705e-06, "loss": 0.1847, "step": 24980 }, { "epoch": 4.94, "learning_rate": 1.6001580403002766e-06, "loss": 0.1808, "step": 24990 }, { "epoch": 4.94, "learning_rate": 1.5507704464638484e-06, "loss": 0.1737, "step": 25000 }, { "epoch": 4.94, "learning_rate": 1.50138285262742e-06, "loss": 0.1682, "step": 25010 }, { "epoch": 4.94, "learning_rate": 1.4519952587909917e-06, "loss": 0.1799, "step": 25020 }, { "epoch": 4.94, "learning_rate": 1.4026076649545635e-06, "loss": 0.1776, "step": 25030 }, { "epoch": 4.95, "learning_rate": 1.3532200711181351e-06, "loss": 0.181, "step": 25040 }, { "epoch": 4.95, "learning_rate": 1.303832477281707e-06, "loss": 0.1956, "step": 25050 }, { "epoch": 4.95, "learning_rate": 1.2544448834452787e-06, "loss": 0.1748, "step": 25060 }, { "epoch": 4.95, "learning_rate": 1.2050572896088503e-06, "loss": 0.1677, "step": 25070 }, { "epoch": 4.95, "learning_rate": 1.155669695772422e-06, "loss": 0.1825, "step": 25080 }, { "epoch": 4.96, "learning_rate": 1.1062821019359937e-06, "loss": 0.1809, "step": 25090 }, { "epoch": 4.96, "learning_rate": 1.0568945080995655e-06, "loss": 0.1748, "step": 25100 }, { "epoch": 4.96, "learning_rate": 1.0075069142631371e-06, "loss": 0.1755, "step": 25110 }, { "epoch": 4.96, "learning_rate": 9.581193204267087e-07, "loss": 0.1801, "step": 25120 }, { "epoch": 4.96, "learning_rate": 9.087317265902806e-07, "loss": 0.171, "step": 25130 }, { "epoch": 4.97, "learning_rate": 8.593441327538523e-07, "loss": 0.1663, "step": 25140 }, { "epoch": 4.97, "learning_rate": 8.09956538917424e-07, "loss": 0.1921, "step": 25150 }, { "epoch": 4.97, "learning_rate": 7.605689450809957e-07, "loss": 0.1915, "step": 25160 }, { "epoch": 4.97, "learning_rate": 7.111813512445674e-07, "loss": 0.1828, "step": 25170 }, { "epoch": 4.97, "learning_rate": 6.617937574081391e-07, "loss": 0.1678, "step": 25180 }, { "epoch": 4.98, "learning_rate": 6.124061635717108e-07, "loss": 0.1939, "step": 25190 }, { "epoch": 4.98, "learning_rate": 5.630185697352825e-07, "loss": 0.187, "step": 25200 }, { "epoch": 4.98, "learning_rate": 5.136309758988542e-07, "loss": 0.1907, "step": 25210 }, { "epoch": 4.98, "learning_rate": 4.642433820624259e-07, "loss": 0.1704, "step": 25220 }, { "epoch": 4.98, "learning_rate": 4.1485578822599767e-07, "loss": 0.165, "step": 25230 }, { "epoch": 4.99, "learning_rate": 3.6546819438956937e-07, "loss": 0.1813, "step": 25240 }, { "epoch": 4.99, "learning_rate": 3.1608060055314106e-07, "loss": 0.1613, "step": 25250 }, { "epoch": 4.99, "learning_rate": 2.6669300671671276e-07, "loss": 0.1905, "step": 25260 }, { "epoch": 4.99, "learning_rate": 2.173054128802845e-07, "loss": 0.1787, "step": 25270 }, { "epoch": 4.99, "learning_rate": 1.6791781904385618e-07, "loss": 0.1793, "step": 25280 }, { "epoch": 5.0, "learning_rate": 1.1853022520742791e-07, "loss": 0.1861, "step": 25290 }, { "epoch": 5.0, "learning_rate": 6.914263137099961e-08, "loss": 0.1564, "step": 25300 }, { "epoch": 5.0, "learning_rate": 1.9755037534571316e-08, "loss": 0.1724, "step": 25310 }, { "epoch": 5.0, "step": 25310, "total_flos": 6.581451072932413e+17, "train_loss": 0.44226374260195156, "train_runtime": 20787.5605, "train_samples_per_second": 38.967, "train_steps_per_second": 1.218 } ], "logging_steps": 10, "max_steps": 25310, "num_train_epochs": 5, "save_steps": 7000, "total_flos": 6.581451072932413e+17, "trial_name": null, "trial_params": null }