{ "best_metric": null, "best_model_checkpoint": null, "epoch": 2.565335898669232, "global_step": 20000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 2.8e-07, "loss": 12.4665, "step": 10 }, { "epoch": 0.0, "learning_rate": 6.800000000000001e-07, "loss": 9.5435, "step": 20 }, { "epoch": 0.0, "learning_rate": 1.08e-06, "loss": 5.8596, "step": 30 }, { "epoch": 0.01, "learning_rate": 1.48e-06, "loss": 3.3176, "step": 40 }, { "epoch": 0.01, "learning_rate": 1.8800000000000002e-06, "loss": 0.8566, "step": 50 }, { "epoch": 0.01, "learning_rate": 2.28e-06, "loss": 0.5592, "step": 60 }, { "epoch": 0.01, "learning_rate": 2.68e-06, "loss": 0.4696, "step": 70 }, { "epoch": 0.01, "learning_rate": 3.08e-06, "loss": 0.4379, "step": 80 }, { "epoch": 0.01, "learning_rate": 3.48e-06, "loss": 0.4189, "step": 90 }, { "epoch": 0.01, "learning_rate": 3.88e-06, "loss": 0.2835, "step": 100 }, { "epoch": 0.01, "learning_rate": 4.2800000000000005e-06, "loss": 0.1801, "step": 110 }, { "epoch": 0.02, "learning_rate": 4.680000000000001e-06, "loss": 0.1381, "step": 120 }, { "epoch": 0.02, "learning_rate": 5.0800000000000005e-06, "loss": 0.1705, "step": 130 }, { "epoch": 0.02, "learning_rate": 5.480000000000001e-06, "loss": 0.1384, "step": 140 }, { "epoch": 0.02, "learning_rate": 5.8800000000000005e-06, "loss": 0.1373, "step": 150 }, { "epoch": 0.02, "learning_rate": 6.280000000000001e-06, "loss": 0.0988, "step": 160 }, { "epoch": 0.02, "learning_rate": 6.680000000000001e-06, "loss": 0.1076, "step": 170 }, { "epoch": 0.02, "learning_rate": 7.08e-06, "loss": 0.0752, "step": 180 }, { "epoch": 0.02, "learning_rate": 7.48e-06, "loss": 0.0867, "step": 190 }, { "epoch": 0.03, "learning_rate": 7.88e-06, "loss": 0.1092, "step": 200 }, { "epoch": 0.03, "learning_rate": 8.28e-06, "loss": 0.0701, "step": 210 }, { "epoch": 0.03, "learning_rate": 8.68e-06, "loss": 0.0718, "step": 220 }, { "epoch": 0.03, "learning_rate": 9.080000000000001e-06, "loss": 0.0563, "step": 230 }, { "epoch": 0.03, "learning_rate": 9.48e-06, "loss": 0.0384, "step": 240 }, { "epoch": 0.03, "learning_rate": 9.88e-06, "loss": 0.0629, "step": 250 }, { "epoch": 0.03, "learning_rate": 1.0280000000000002e-05, "loss": 0.0472, "step": 260 }, { "epoch": 0.03, "learning_rate": 1.0680000000000001e-05, "loss": 0.0386, "step": 270 }, { "epoch": 0.04, "learning_rate": 1.1080000000000002e-05, "loss": 0.0325, "step": 280 }, { "epoch": 0.04, "learning_rate": 1.148e-05, "loss": 0.0522, "step": 290 }, { "epoch": 0.04, "learning_rate": 1.188e-05, "loss": 0.0358, "step": 300 }, { "epoch": 0.04, "learning_rate": 1.2280000000000001e-05, "loss": 0.0431, "step": 310 }, { "epoch": 0.04, "learning_rate": 1.268e-05, "loss": 0.0459, "step": 320 }, { "epoch": 0.04, "learning_rate": 1.3080000000000002e-05, "loss": 0.0377, "step": 330 }, { "epoch": 0.04, "learning_rate": 1.3480000000000001e-05, "loss": 0.0424, "step": 340 }, { "epoch": 0.04, "learning_rate": 1.3880000000000001e-05, "loss": 0.0324, "step": 350 }, { "epoch": 0.05, "learning_rate": 1.428e-05, "loss": 0.0428, "step": 360 }, { "epoch": 0.05, "learning_rate": 1.4680000000000002e-05, "loss": 0.0245, "step": 370 }, { "epoch": 0.05, "learning_rate": 1.5080000000000001e-05, "loss": 0.0281, "step": 380 }, { "epoch": 0.05, "learning_rate": 1.548e-05, "loss": 0.0344, "step": 390 }, { "epoch": 0.05, "learning_rate": 1.588e-05, "loss": 0.0239, "step": 400 }, { "epoch": 0.05, "learning_rate": 1.628e-05, "loss": 0.043, "step": 410 }, { "epoch": 0.05, "learning_rate": 1.668e-05, "loss": 0.0309, "step": 420 }, { "epoch": 0.06, "learning_rate": 1.7080000000000002e-05, "loss": 0.0297, "step": 430 }, { "epoch": 0.06, "learning_rate": 1.7480000000000002e-05, "loss": 0.0232, "step": 440 }, { "epoch": 0.06, "learning_rate": 1.788e-05, "loss": 0.0271, "step": 450 }, { "epoch": 0.06, "learning_rate": 1.828e-05, "loss": 0.0238, "step": 460 }, { "epoch": 0.06, "learning_rate": 1.8680000000000004e-05, "loss": 0.0214, "step": 470 }, { "epoch": 0.06, "learning_rate": 1.908e-05, "loss": 0.0225, "step": 480 }, { "epoch": 0.06, "learning_rate": 1.948e-05, "loss": 0.0301, "step": 490 }, { "epoch": 0.06, "learning_rate": 1.9880000000000003e-05, "loss": 0.0216, "step": 500 }, { "epoch": 0.07, "learning_rate": 1.9993883257602237e-05, "loss": 0.0304, "step": 510 }, { "epoch": 0.07, "learning_rate": 1.9985145054176863e-05, "loss": 0.0233, "step": 520 }, { "epoch": 0.07, "learning_rate": 1.9976406850751487e-05, "loss": 0.0313, "step": 530 }, { "epoch": 0.07, "learning_rate": 1.996766864732611e-05, "loss": 0.0169, "step": 540 }, { "epoch": 0.07, "learning_rate": 1.9958930443900733e-05, "loss": 0.0206, "step": 550 }, { "epoch": 0.07, "learning_rate": 1.995019224047536e-05, "loss": 0.0295, "step": 560 }, { "epoch": 0.07, "learning_rate": 1.9941454037049983e-05, "loss": 0.0238, "step": 570 }, { "epoch": 0.07, "learning_rate": 1.9932715833624606e-05, "loss": 0.0218, "step": 580 }, { "epoch": 0.08, "learning_rate": 1.9923977630199233e-05, "loss": 0.027, "step": 590 }, { "epoch": 0.08, "learning_rate": 1.991523942677386e-05, "loss": 0.026, "step": 600 }, { "epoch": 0.08, "learning_rate": 1.9906501223348483e-05, "loss": 0.0235, "step": 610 }, { "epoch": 0.08, "learning_rate": 1.9897763019923106e-05, "loss": 0.0207, "step": 620 }, { "epoch": 0.08, "learning_rate": 1.988902481649773e-05, "loss": 0.0213, "step": 630 }, { "epoch": 0.08, "learning_rate": 1.9880286613072356e-05, "loss": 0.0194, "step": 640 }, { "epoch": 0.08, "learning_rate": 1.987154840964698e-05, "loss": 0.0266, "step": 650 }, { "epoch": 0.08, "learning_rate": 1.9862810206221602e-05, "loss": 0.0225, "step": 660 }, { "epoch": 0.09, "learning_rate": 1.985407200279623e-05, "loss": 0.0295, "step": 670 }, { "epoch": 0.09, "learning_rate": 1.9845333799370852e-05, "loss": 0.0261, "step": 680 }, { "epoch": 0.09, "learning_rate": 1.9836595595945475e-05, "loss": 0.0207, "step": 690 }, { "epoch": 0.09, "learning_rate": 1.98278573925201e-05, "loss": 0.0241, "step": 700 }, { "epoch": 0.09, "learning_rate": 1.9819119189094725e-05, "loss": 0.0229, "step": 710 }, { "epoch": 0.09, "learning_rate": 1.981038098566935e-05, "loss": 0.0209, "step": 720 }, { "epoch": 0.09, "learning_rate": 1.980164278224397e-05, "loss": 0.0227, "step": 730 }, { "epoch": 0.09, "learning_rate": 1.9792904578818598e-05, "loss": 0.0154, "step": 740 }, { "epoch": 0.1, "learning_rate": 1.978416637539322e-05, "loss": 0.0177, "step": 750 }, { "epoch": 0.1, "learning_rate": 1.9775428171967845e-05, "loss": 0.0194, "step": 760 }, { "epoch": 0.1, "learning_rate": 1.976668996854247e-05, "loss": 0.0261, "step": 770 }, { "epoch": 0.1, "learning_rate": 1.9757951765117094e-05, "loss": 0.0177, "step": 780 }, { "epoch": 0.1, "learning_rate": 1.9749213561691718e-05, "loss": 0.0181, "step": 790 }, { "epoch": 0.1, "learning_rate": 1.974047535826634e-05, "loss": 0.0243, "step": 800 }, { "epoch": 0.1, "learning_rate": 1.9731737154840968e-05, "loss": 0.0265, "step": 810 }, { "epoch": 0.11, "learning_rate": 1.972299895141559e-05, "loss": 0.0251, "step": 820 }, { "epoch": 0.11, "learning_rate": 1.9714260747990214e-05, "loss": 0.0219, "step": 830 }, { "epoch": 0.11, "learning_rate": 1.970552254456484e-05, "loss": 0.0158, "step": 840 }, { "epoch": 0.11, "learning_rate": 1.9696784341139464e-05, "loss": 0.0225, "step": 850 }, { "epoch": 0.11, "learning_rate": 1.9688046137714087e-05, "loss": 0.0217, "step": 860 }, { "epoch": 0.11, "learning_rate": 1.967930793428871e-05, "loss": 0.0162, "step": 870 }, { "epoch": 0.11, "learning_rate": 1.9670569730863337e-05, "loss": 0.0147, "step": 880 }, { "epoch": 0.11, "learning_rate": 1.966183152743796e-05, "loss": 0.0164, "step": 890 }, { "epoch": 0.12, "learning_rate": 1.9653093324012583e-05, "loss": 0.0235, "step": 900 }, { "epoch": 0.12, "learning_rate": 1.964435512058721e-05, "loss": 0.0184, "step": 910 }, { "epoch": 0.12, "learning_rate": 1.9635616917161833e-05, "loss": 0.0213, "step": 920 }, { "epoch": 0.12, "learning_rate": 1.9626878713736456e-05, "loss": 0.0258, "step": 930 }, { "epoch": 0.12, "learning_rate": 1.961814051031108e-05, "loss": 0.0161, "step": 940 }, { "epoch": 0.12, "learning_rate": 1.9609402306885706e-05, "loss": 0.0193, "step": 950 }, { "epoch": 0.12, "learning_rate": 1.960066410346033e-05, "loss": 0.0273, "step": 960 }, { "epoch": 0.12, "learning_rate": 1.9591925900034953e-05, "loss": 0.017, "step": 970 }, { "epoch": 0.13, "learning_rate": 1.958318769660958e-05, "loss": 0.022, "step": 980 }, { "epoch": 0.13, "learning_rate": 1.9574449493184203e-05, "loss": 0.0136, "step": 990 }, { "epoch": 0.13, "learning_rate": 1.9565711289758826e-05, "loss": 0.0162, "step": 1000 }, { "epoch": 0.13, "learning_rate": 1.955697308633345e-05, "loss": 0.0206, "step": 1010 }, { "epoch": 0.13, "learning_rate": 1.9548234882908076e-05, "loss": 0.0191, "step": 1020 }, { "epoch": 0.13, "learning_rate": 1.95394966794827e-05, "loss": 0.0215, "step": 1030 }, { "epoch": 0.13, "learning_rate": 1.9530758476057322e-05, "loss": 0.0187, "step": 1040 }, { "epoch": 0.13, "learning_rate": 1.952202027263195e-05, "loss": 0.0219, "step": 1050 }, { "epoch": 0.14, "learning_rate": 1.9513282069206572e-05, "loss": 0.0148, "step": 1060 }, { "epoch": 0.14, "learning_rate": 1.95045438657812e-05, "loss": 0.0181, "step": 1070 }, { "epoch": 0.14, "learning_rate": 1.9495805662355822e-05, "loss": 0.0175, "step": 1080 }, { "epoch": 0.14, "learning_rate": 1.9487067458930445e-05, "loss": 0.0203, "step": 1090 }, { "epoch": 0.14, "learning_rate": 1.947832925550507e-05, "loss": 0.0153, "step": 1100 }, { "epoch": 0.14, "learning_rate": 1.9469591052079695e-05, "loss": 0.0143, "step": 1110 }, { "epoch": 0.14, "learning_rate": 1.9460852848654318e-05, "loss": 0.014, "step": 1120 }, { "epoch": 0.14, "learning_rate": 1.9452114645228945e-05, "loss": 0.0158, "step": 1130 }, { "epoch": 0.15, "learning_rate": 1.9443376441803568e-05, "loss": 0.0147, "step": 1140 }, { "epoch": 0.15, "learning_rate": 1.943463823837819e-05, "loss": 0.0211, "step": 1150 }, { "epoch": 0.15, "learning_rate": 1.9425900034952814e-05, "loss": 0.018, "step": 1160 }, { "epoch": 0.15, "learning_rate": 1.941716183152744e-05, "loss": 0.0192, "step": 1170 }, { "epoch": 0.15, "learning_rate": 1.9408423628102064e-05, "loss": 0.0196, "step": 1180 }, { "epoch": 0.15, "learning_rate": 1.9399685424676687e-05, "loss": 0.0184, "step": 1190 }, { "epoch": 0.15, "learning_rate": 1.9390947221251314e-05, "loss": 0.016, "step": 1200 }, { "epoch": 0.16, "learning_rate": 1.9382209017825937e-05, "loss": 0.0227, "step": 1210 }, { "epoch": 0.16, "learning_rate": 1.937347081440056e-05, "loss": 0.024, "step": 1220 }, { "epoch": 0.16, "learning_rate": 1.9364732610975184e-05, "loss": 0.0209, "step": 1230 }, { "epoch": 0.16, "learning_rate": 1.935599440754981e-05, "loss": 0.0164, "step": 1240 }, { "epoch": 0.16, "learning_rate": 1.9347256204124434e-05, "loss": 0.0187, "step": 1250 }, { "epoch": 0.16, "learning_rate": 1.9338518000699057e-05, "loss": 0.0153, "step": 1260 }, { "epoch": 0.16, "learning_rate": 1.9329779797273683e-05, "loss": 0.017, "step": 1270 }, { "epoch": 0.16, "learning_rate": 1.9321041593848307e-05, "loss": 0.0131, "step": 1280 }, { "epoch": 0.17, "learning_rate": 1.931230339042293e-05, "loss": 0.0158, "step": 1290 }, { "epoch": 0.17, "learning_rate": 1.9303565186997556e-05, "loss": 0.016, "step": 1300 }, { "epoch": 0.17, "learning_rate": 1.929482698357218e-05, "loss": 0.0133, "step": 1310 }, { "epoch": 0.17, "learning_rate": 1.9286088780146803e-05, "loss": 0.0225, "step": 1320 }, { "epoch": 0.17, "learning_rate": 1.9277350576721426e-05, "loss": 0.0158, "step": 1330 }, { "epoch": 0.17, "learning_rate": 1.9268612373296053e-05, "loss": 0.0226, "step": 1340 }, { "epoch": 0.17, "learning_rate": 1.9259874169870676e-05, "loss": 0.0211, "step": 1350 }, { "epoch": 0.17, "learning_rate": 1.92511359664453e-05, "loss": 0.0142, "step": 1360 }, { "epoch": 0.18, "learning_rate": 1.9242397763019926e-05, "loss": 0.0197, "step": 1370 }, { "epoch": 0.18, "learning_rate": 1.923365955959455e-05, "loss": 0.0183, "step": 1380 }, { "epoch": 0.18, "learning_rate": 1.9224921356169172e-05, "loss": 0.0178, "step": 1390 }, { "epoch": 0.18, "learning_rate": 1.9216183152743796e-05, "loss": 0.0167, "step": 1400 }, { "epoch": 0.18, "learning_rate": 1.9207444949318422e-05, "loss": 0.0244, "step": 1410 }, { "epoch": 0.18, "learning_rate": 1.9198706745893045e-05, "loss": 0.0166, "step": 1420 }, { "epoch": 0.18, "learning_rate": 1.918996854246767e-05, "loss": 0.0141, "step": 1430 }, { "epoch": 0.18, "learning_rate": 1.9181230339042295e-05, "loss": 0.0146, "step": 1440 }, { "epoch": 0.19, "learning_rate": 1.917249213561692e-05, "loss": 0.0186, "step": 1450 }, { "epoch": 0.19, "learning_rate": 1.916375393219154e-05, "loss": 0.021, "step": 1460 }, { "epoch": 0.19, "learning_rate": 1.9155015728766165e-05, "loss": 0.0187, "step": 1470 }, { "epoch": 0.19, "learning_rate": 1.914627752534079e-05, "loss": 0.0161, "step": 1480 }, { "epoch": 0.19, "learning_rate": 1.9137539321915415e-05, "loss": 0.0148, "step": 1490 }, { "epoch": 0.19, "learning_rate": 1.9128801118490038e-05, "loss": 0.016, "step": 1500 }, { "epoch": 0.19, "learning_rate": 1.9120062915064665e-05, "loss": 0.0149, "step": 1510 }, { "epoch": 0.19, "learning_rate": 1.9111324711639288e-05, "loss": 0.015, "step": 1520 }, { "epoch": 0.2, "learning_rate": 1.910258650821391e-05, "loss": 0.0187, "step": 1530 }, { "epoch": 0.2, "learning_rate": 1.9093848304788538e-05, "loss": 0.019, "step": 1540 }, { "epoch": 0.2, "learning_rate": 1.908511010136316e-05, "loss": 0.0208, "step": 1550 }, { "epoch": 0.2, "learning_rate": 1.9076371897937787e-05, "loss": 0.018, "step": 1560 }, { "epoch": 0.2, "learning_rate": 1.906763369451241e-05, "loss": 0.023, "step": 1570 }, { "epoch": 0.2, "learning_rate": 1.9058895491087034e-05, "loss": 0.0169, "step": 1580 }, { "epoch": 0.2, "learning_rate": 1.905015728766166e-05, "loss": 0.0204, "step": 1590 }, { "epoch": 0.21, "learning_rate": 1.9041419084236284e-05, "loss": 0.0175, "step": 1600 }, { "epoch": 0.21, "learning_rate": 1.9032680880810907e-05, "loss": 0.016, "step": 1610 }, { "epoch": 0.21, "learning_rate": 1.902394267738553e-05, "loss": 0.0132, "step": 1620 }, { "epoch": 0.21, "learning_rate": 1.9015204473960157e-05, "loss": 0.0167, "step": 1630 }, { "epoch": 0.21, "learning_rate": 1.900646627053478e-05, "loss": 0.0146, "step": 1640 }, { "epoch": 0.21, "learning_rate": 1.8997728067109403e-05, "loss": 0.0153, "step": 1650 }, { "epoch": 0.21, "learning_rate": 1.898898986368403e-05, "loss": 0.017, "step": 1660 }, { "epoch": 0.21, "learning_rate": 1.8980251660258653e-05, "loss": 0.0149, "step": 1670 }, { "epoch": 0.22, "learning_rate": 1.8971513456833276e-05, "loss": 0.0209, "step": 1680 }, { "epoch": 0.22, "learning_rate": 1.89627752534079e-05, "loss": 0.0155, "step": 1690 }, { "epoch": 0.22, "learning_rate": 1.8954037049982526e-05, "loss": 0.0183, "step": 1700 }, { "epoch": 0.22, "learning_rate": 1.894529884655715e-05, "loss": 0.0159, "step": 1710 }, { "epoch": 0.22, "learning_rate": 1.8936560643131773e-05, "loss": 0.0172, "step": 1720 }, { "epoch": 0.22, "learning_rate": 1.89278224397064e-05, "loss": 0.0141, "step": 1730 }, { "epoch": 0.22, "learning_rate": 1.8919084236281022e-05, "loss": 0.0174, "step": 1740 }, { "epoch": 0.22, "learning_rate": 1.8910346032855646e-05, "loss": 0.0207, "step": 1750 }, { "epoch": 0.23, "learning_rate": 1.8901607829430272e-05, "loss": 0.016, "step": 1760 }, { "epoch": 0.23, "learning_rate": 1.8892869626004896e-05, "loss": 0.0206, "step": 1770 }, { "epoch": 0.23, "learning_rate": 1.888413142257952e-05, "loss": 0.0174, "step": 1780 }, { "epoch": 0.23, "learning_rate": 1.8875393219154142e-05, "loss": 0.0152, "step": 1790 }, { "epoch": 0.23, "learning_rate": 1.886665501572877e-05, "loss": 0.0218, "step": 1800 }, { "epoch": 0.23, "learning_rate": 1.8857916812303392e-05, "loss": 0.0186, "step": 1810 }, { "epoch": 0.23, "learning_rate": 1.8849178608878015e-05, "loss": 0.0193, "step": 1820 }, { "epoch": 0.23, "learning_rate": 1.8840440405452642e-05, "loss": 0.0171, "step": 1830 }, { "epoch": 0.24, "learning_rate": 1.8831702202027265e-05, "loss": 0.0164, "step": 1840 }, { "epoch": 0.24, "learning_rate": 1.8822963998601888e-05, "loss": 0.0219, "step": 1850 }, { "epoch": 0.24, "learning_rate": 1.881422579517651e-05, "loss": 0.02, "step": 1860 }, { "epoch": 0.24, "learning_rate": 1.8805487591751138e-05, "loss": 0.017, "step": 1870 }, { "epoch": 0.24, "learning_rate": 1.879674938832576e-05, "loss": 0.0235, "step": 1880 }, { "epoch": 0.24, "learning_rate": 1.8788011184900384e-05, "loss": 0.0135, "step": 1890 }, { "epoch": 0.24, "learning_rate": 1.877927298147501e-05, "loss": 0.0179, "step": 1900 }, { "epoch": 0.24, "learning_rate": 1.8770534778049634e-05, "loss": 0.0207, "step": 1910 }, { "epoch": 0.25, "learning_rate": 1.8761796574624258e-05, "loss": 0.0131, "step": 1920 }, { "epoch": 0.25, "learning_rate": 1.875305837119888e-05, "loss": 0.0204, "step": 1930 }, { "epoch": 0.25, "learning_rate": 1.8744320167773507e-05, "loss": 0.0173, "step": 1940 }, { "epoch": 0.25, "learning_rate": 1.873558196434813e-05, "loss": 0.0218, "step": 1950 }, { "epoch": 0.25, "learning_rate": 1.8726843760922754e-05, "loss": 0.0122, "step": 1960 }, { "epoch": 0.25, "learning_rate": 1.871810555749738e-05, "loss": 0.0151, "step": 1970 }, { "epoch": 0.25, "learning_rate": 1.8709367354072004e-05, "loss": 0.02, "step": 1980 }, { "epoch": 0.26, "learning_rate": 1.8700629150646627e-05, "loss": 0.0182, "step": 1990 }, { "epoch": 0.26, "learning_rate": 1.869189094722125e-05, "loss": 0.0172, "step": 2000 }, { "epoch": 0.26, "learning_rate": 1.8683152743795877e-05, "loss": 0.0177, "step": 2010 }, { "epoch": 0.26, "learning_rate": 1.8674414540370503e-05, "loss": 0.0179, "step": 2020 }, { "epoch": 0.26, "learning_rate": 1.8665676336945127e-05, "loss": 0.0139, "step": 2030 }, { "epoch": 0.26, "learning_rate": 1.865693813351975e-05, "loss": 0.0163, "step": 2040 }, { "epoch": 0.26, "learning_rate": 1.8648199930094376e-05, "loss": 0.0121, "step": 2050 }, { "epoch": 0.26, "learning_rate": 1.8639461726669e-05, "loss": 0.0187, "step": 2060 }, { "epoch": 0.27, "learning_rate": 1.8630723523243623e-05, "loss": 0.0173, "step": 2070 }, { "epoch": 0.27, "learning_rate": 1.8621985319818246e-05, "loss": 0.0173, "step": 2080 }, { "epoch": 0.27, "learning_rate": 1.8613247116392873e-05, "loss": 0.0137, "step": 2090 }, { "epoch": 0.27, "learning_rate": 1.8604508912967496e-05, "loss": 0.0158, "step": 2100 }, { "epoch": 0.27, "learning_rate": 1.859577070954212e-05, "loss": 0.0113, "step": 2110 }, { "epoch": 0.27, "learning_rate": 1.8587032506116746e-05, "loss": 0.018, "step": 2120 }, { "epoch": 0.27, "learning_rate": 1.857829430269137e-05, "loss": 0.0177, "step": 2130 }, { "epoch": 0.27, "learning_rate": 1.8569556099265992e-05, "loss": 0.0182, "step": 2140 }, { "epoch": 0.28, "learning_rate": 1.8560817895840615e-05, "loss": 0.0126, "step": 2150 }, { "epoch": 0.28, "learning_rate": 1.8552079692415242e-05, "loss": 0.0151, "step": 2160 }, { "epoch": 0.28, "learning_rate": 1.8543341488989865e-05, "loss": 0.0174, "step": 2170 }, { "epoch": 0.28, "learning_rate": 1.853460328556449e-05, "loss": 0.0139, "step": 2180 }, { "epoch": 0.28, "learning_rate": 1.8525865082139115e-05, "loss": 0.0229, "step": 2190 }, { "epoch": 0.28, "learning_rate": 1.851712687871374e-05, "loss": 0.015, "step": 2200 }, { "epoch": 0.28, "learning_rate": 1.850838867528836e-05, "loss": 0.019, "step": 2210 }, { "epoch": 0.28, "learning_rate": 1.8499650471862988e-05, "loss": 0.0139, "step": 2220 }, { "epoch": 0.29, "learning_rate": 1.849091226843761e-05, "loss": 0.0174, "step": 2230 }, { "epoch": 0.29, "learning_rate": 1.8482174065012235e-05, "loss": 0.0204, "step": 2240 }, { "epoch": 0.29, "learning_rate": 1.8473435861586858e-05, "loss": 0.0134, "step": 2250 }, { "epoch": 0.29, "learning_rate": 1.8464697658161484e-05, "loss": 0.0189, "step": 2260 }, { "epoch": 0.29, "learning_rate": 1.8455959454736108e-05, "loss": 0.0097, "step": 2270 }, { "epoch": 0.29, "learning_rate": 1.844722125131073e-05, "loss": 0.0181, "step": 2280 }, { "epoch": 0.29, "learning_rate": 1.8438483047885358e-05, "loss": 0.0163, "step": 2290 }, { "epoch": 0.3, "learning_rate": 1.842974484445998e-05, "loss": 0.0175, "step": 2300 }, { "epoch": 0.3, "learning_rate": 1.8421006641034604e-05, "loss": 0.0125, "step": 2310 }, { "epoch": 0.3, "learning_rate": 1.8412268437609227e-05, "loss": 0.0129, "step": 2320 }, { "epoch": 0.3, "learning_rate": 1.8403530234183854e-05, "loss": 0.0159, "step": 2330 }, { "epoch": 0.3, "learning_rate": 1.8394792030758477e-05, "loss": 0.0193, "step": 2340 }, { "epoch": 0.3, "learning_rate": 1.83860538273331e-05, "loss": 0.0167, "step": 2350 }, { "epoch": 0.3, "learning_rate": 1.8377315623907727e-05, "loss": 0.0171, "step": 2360 }, { "epoch": 0.3, "learning_rate": 1.836857742048235e-05, "loss": 0.0168, "step": 2370 }, { "epoch": 0.31, "learning_rate": 1.8359839217056973e-05, "loss": 0.013, "step": 2380 }, { "epoch": 0.31, "learning_rate": 1.8351101013631597e-05, "loss": 0.0154, "step": 2390 }, { "epoch": 0.31, "learning_rate": 1.8342362810206223e-05, "loss": 0.0134, "step": 2400 }, { "epoch": 0.31, "learning_rate": 1.8333624606780846e-05, "loss": 0.0152, "step": 2410 }, { "epoch": 0.31, "learning_rate": 1.832488640335547e-05, "loss": 0.0163, "step": 2420 }, { "epoch": 0.31, "learning_rate": 1.8316148199930096e-05, "loss": 0.0131, "step": 2430 }, { "epoch": 0.31, "learning_rate": 1.830740999650472e-05, "loss": 0.0147, "step": 2440 }, { "epoch": 0.31, "learning_rate": 1.8298671793079343e-05, "loss": 0.0181, "step": 2450 }, { "epoch": 0.32, "learning_rate": 1.8289933589653966e-05, "loss": 0.0154, "step": 2460 }, { "epoch": 0.32, "learning_rate": 1.8281195386228593e-05, "loss": 0.0118, "step": 2470 }, { "epoch": 0.32, "learning_rate": 1.827245718280322e-05, "loss": 0.0189, "step": 2480 }, { "epoch": 0.32, "learning_rate": 1.8263718979377842e-05, "loss": 0.0136, "step": 2490 }, { "epoch": 0.32, "learning_rate": 1.8254980775952466e-05, "loss": 0.0163, "step": 2500 }, { "epoch": 0.32, "learning_rate": 1.8246242572527092e-05, "loss": 0.0135, "step": 2510 }, { "epoch": 0.32, "learning_rate": 1.8237504369101715e-05, "loss": 0.0121, "step": 2520 }, { "epoch": 0.32, "learning_rate": 1.822876616567634e-05, "loss": 0.0162, "step": 2530 }, { "epoch": 0.33, "learning_rate": 1.8220027962250962e-05, "loss": 0.0156, "step": 2540 }, { "epoch": 0.33, "learning_rate": 1.821128975882559e-05, "loss": 0.017, "step": 2550 }, { "epoch": 0.33, "learning_rate": 1.8202551555400212e-05, "loss": 0.016, "step": 2560 }, { "epoch": 0.33, "learning_rate": 1.8193813351974835e-05, "loss": 0.0138, "step": 2570 }, { "epoch": 0.33, "learning_rate": 1.818507514854946e-05, "loss": 0.0158, "step": 2580 }, { "epoch": 0.33, "learning_rate": 1.8176336945124085e-05, "loss": 0.0137, "step": 2590 }, { "epoch": 0.33, "learning_rate": 1.8167598741698708e-05, "loss": 0.0186, "step": 2600 }, { "epoch": 0.33, "learning_rate": 1.815886053827333e-05, "loss": 0.0158, "step": 2610 }, { "epoch": 0.34, "learning_rate": 1.8150122334847958e-05, "loss": 0.0187, "step": 2620 }, { "epoch": 0.34, "learning_rate": 1.814138413142258e-05, "loss": 0.0185, "step": 2630 }, { "epoch": 0.34, "learning_rate": 1.8132645927997204e-05, "loss": 0.0139, "step": 2640 }, { "epoch": 0.34, "learning_rate": 1.812390772457183e-05, "loss": 0.0209, "step": 2650 }, { "epoch": 0.34, "learning_rate": 1.8115169521146454e-05, "loss": 0.0161, "step": 2660 }, { "epoch": 0.34, "learning_rate": 1.8106431317721077e-05, "loss": 0.018, "step": 2670 }, { "epoch": 0.34, "learning_rate": 1.8097693114295704e-05, "loss": 0.0177, "step": 2680 }, { "epoch": 0.35, "learning_rate": 1.8088954910870327e-05, "loss": 0.0144, "step": 2690 }, { "epoch": 0.35, "learning_rate": 1.808021670744495e-05, "loss": 0.0229, "step": 2700 }, { "epoch": 0.35, "learning_rate": 1.8071478504019574e-05, "loss": 0.0129, "step": 2710 }, { "epoch": 0.35, "learning_rate": 1.80627403005942e-05, "loss": 0.0177, "step": 2720 }, { "epoch": 0.35, "learning_rate": 1.8054002097168824e-05, "loss": 0.0221, "step": 2730 }, { "epoch": 0.35, "learning_rate": 1.8045263893743447e-05, "loss": 0.0167, "step": 2740 }, { "epoch": 0.35, "learning_rate": 1.8036525690318073e-05, "loss": 0.0146, "step": 2750 }, { "epoch": 0.35, "learning_rate": 1.8027787486892697e-05, "loss": 0.0173, "step": 2760 }, { "epoch": 0.36, "learning_rate": 1.801904928346732e-05, "loss": 0.0169, "step": 2770 }, { "epoch": 0.36, "learning_rate": 1.8010311080041943e-05, "loss": 0.0148, "step": 2780 }, { "epoch": 0.36, "learning_rate": 1.800157287661657e-05, "loss": 0.015, "step": 2790 }, { "epoch": 0.36, "learning_rate": 1.7992834673191193e-05, "loss": 0.0181, "step": 2800 }, { "epoch": 0.36, "learning_rate": 1.7984096469765816e-05, "loss": 0.015, "step": 2810 }, { "epoch": 0.36, "learning_rate": 1.7975358266340443e-05, "loss": 0.017, "step": 2820 }, { "epoch": 0.36, "learning_rate": 1.7966620062915066e-05, "loss": 0.0203, "step": 2830 }, { "epoch": 0.36, "learning_rate": 1.795788185948969e-05, "loss": 0.0155, "step": 2840 }, { "epoch": 0.37, "learning_rate": 1.7949143656064312e-05, "loss": 0.013, "step": 2850 }, { "epoch": 0.37, "learning_rate": 1.794040545263894e-05, "loss": 0.0192, "step": 2860 }, { "epoch": 0.37, "learning_rate": 1.7931667249213562e-05, "loss": 0.0151, "step": 2870 }, { "epoch": 0.37, "learning_rate": 1.7922929045788186e-05, "loss": 0.0189, "step": 2880 }, { "epoch": 0.37, "learning_rate": 1.7914190842362812e-05, "loss": 0.0138, "step": 2890 }, { "epoch": 0.37, "learning_rate": 1.7905452638937435e-05, "loss": 0.0202, "step": 2900 }, { "epoch": 0.37, "learning_rate": 1.789671443551206e-05, "loss": 0.0186, "step": 2910 }, { "epoch": 0.37, "learning_rate": 1.7887976232086682e-05, "loss": 0.0139, "step": 2920 }, { "epoch": 0.38, "learning_rate": 1.787923802866131e-05, "loss": 0.0231, "step": 2930 }, { "epoch": 0.38, "learning_rate": 1.787049982523593e-05, "loss": 0.0216, "step": 2940 }, { "epoch": 0.38, "learning_rate": 1.7861761621810558e-05, "loss": 0.0158, "step": 2950 }, { "epoch": 0.38, "learning_rate": 1.785302341838518e-05, "loss": 0.0164, "step": 2960 }, { "epoch": 0.38, "learning_rate": 1.7844285214959808e-05, "loss": 0.0157, "step": 2970 }, { "epoch": 0.38, "learning_rate": 1.783554701153443e-05, "loss": 0.0174, "step": 2980 }, { "epoch": 0.38, "learning_rate": 1.7826808808109055e-05, "loss": 0.0161, "step": 2990 }, { "epoch": 0.38, "learning_rate": 1.7818070604683678e-05, "loss": 0.0171, "step": 3000 }, { "epoch": 0.39, "learning_rate": 1.7809332401258304e-05, "loss": 0.0119, "step": 3010 }, { "epoch": 0.39, "learning_rate": 1.7800594197832928e-05, "loss": 0.0127, "step": 3020 }, { "epoch": 0.39, "learning_rate": 1.779185599440755e-05, "loss": 0.0193, "step": 3030 }, { "epoch": 0.39, "learning_rate": 1.7783117790982177e-05, "loss": 0.0208, "step": 3040 }, { "epoch": 0.39, "learning_rate": 1.77743795875568e-05, "loss": 0.0173, "step": 3050 }, { "epoch": 0.39, "learning_rate": 1.7765641384131424e-05, "loss": 0.0151, "step": 3060 }, { "epoch": 0.39, "learning_rate": 1.7756903180706047e-05, "loss": 0.0169, "step": 3070 }, { "epoch": 0.4, "learning_rate": 1.7748164977280674e-05, "loss": 0.0257, "step": 3080 }, { "epoch": 0.4, "learning_rate": 1.7739426773855297e-05, "loss": 0.0158, "step": 3090 }, { "epoch": 0.4, "learning_rate": 1.773068857042992e-05, "loss": 0.0193, "step": 3100 }, { "epoch": 0.4, "learning_rate": 1.7721950367004547e-05, "loss": 0.0219, "step": 3110 }, { "epoch": 0.4, "learning_rate": 1.771321216357917e-05, "loss": 0.0188, "step": 3120 }, { "epoch": 0.4, "learning_rate": 1.7704473960153793e-05, "loss": 0.0141, "step": 3130 }, { "epoch": 0.4, "learning_rate": 1.769573575672842e-05, "loss": 0.0182, "step": 3140 }, { "epoch": 0.4, "learning_rate": 1.7686997553303043e-05, "loss": 0.0177, "step": 3150 }, { "epoch": 0.41, "learning_rate": 1.7678259349877666e-05, "loss": 0.014, "step": 3160 }, { "epoch": 0.41, "learning_rate": 1.766952114645229e-05, "loss": 0.02, "step": 3170 }, { "epoch": 0.41, "learning_rate": 1.7660782943026916e-05, "loss": 0.0159, "step": 3180 }, { "epoch": 0.41, "learning_rate": 1.765204473960154e-05, "loss": 0.0147, "step": 3190 }, { "epoch": 0.41, "learning_rate": 1.7643306536176163e-05, "loss": 0.014, "step": 3200 }, { "epoch": 0.41, "learning_rate": 1.763456833275079e-05, "loss": 0.015, "step": 3210 }, { "epoch": 0.41, "learning_rate": 1.7625830129325412e-05, "loss": 0.0185, "step": 3220 }, { "epoch": 0.41, "learning_rate": 1.7617091925900036e-05, "loss": 0.0137, "step": 3230 }, { "epoch": 0.42, "learning_rate": 1.760835372247466e-05, "loss": 0.0185, "step": 3240 }, { "epoch": 0.42, "learning_rate": 1.7599615519049286e-05, "loss": 0.0209, "step": 3250 }, { "epoch": 0.42, "learning_rate": 1.759087731562391e-05, "loss": 0.0211, "step": 3260 }, { "epoch": 0.42, "learning_rate": 1.7582139112198532e-05, "loss": 0.0132, "step": 3270 }, { "epoch": 0.42, "learning_rate": 1.757340090877316e-05, "loss": 0.0163, "step": 3280 }, { "epoch": 0.42, "learning_rate": 1.7564662705347782e-05, "loss": 0.0143, "step": 3290 }, { "epoch": 0.42, "learning_rate": 1.7555924501922405e-05, "loss": 0.0166, "step": 3300 }, { "epoch": 0.42, "learning_rate": 1.7547186298497028e-05, "loss": 0.0196, "step": 3310 }, { "epoch": 0.43, "learning_rate": 1.7538448095071655e-05, "loss": 0.0181, "step": 3320 }, { "epoch": 0.43, "learning_rate": 1.7529709891646278e-05, "loss": 0.0209, "step": 3330 }, { "epoch": 0.43, "learning_rate": 1.75209716882209e-05, "loss": 0.0155, "step": 3340 }, { "epoch": 0.43, "learning_rate": 1.7512233484795528e-05, "loss": 0.0148, "step": 3350 }, { "epoch": 0.43, "learning_rate": 1.750349528137015e-05, "loss": 0.0191, "step": 3360 }, { "epoch": 0.43, "learning_rate": 1.7494757077944774e-05, "loss": 0.0143, "step": 3370 }, { "epoch": 0.43, "learning_rate": 1.7486018874519398e-05, "loss": 0.0185, "step": 3380 }, { "epoch": 0.43, "learning_rate": 1.7477280671094024e-05, "loss": 0.014, "step": 3390 }, { "epoch": 0.44, "learning_rate": 1.7468542467668648e-05, "loss": 0.0162, "step": 3400 }, { "epoch": 0.44, "learning_rate": 1.745980426424327e-05, "loss": 0.0192, "step": 3410 }, { "epoch": 0.44, "learning_rate": 1.7451066060817897e-05, "loss": 0.0163, "step": 3420 }, { "epoch": 0.44, "learning_rate": 1.7442327857392524e-05, "loss": 0.0136, "step": 3430 }, { "epoch": 0.44, "learning_rate": 1.7433589653967147e-05, "loss": 0.0183, "step": 3440 }, { "epoch": 0.44, "learning_rate": 1.742485145054177e-05, "loss": 0.0169, "step": 3450 }, { "epoch": 0.44, "learning_rate": 1.7416113247116394e-05, "loss": 0.0146, "step": 3460 }, { "epoch": 0.45, "learning_rate": 1.740737504369102e-05, "loss": 0.0191, "step": 3470 }, { "epoch": 0.45, "learning_rate": 1.7398636840265643e-05, "loss": 0.0138, "step": 3480 }, { "epoch": 0.45, "learning_rate": 1.7389898636840267e-05, "loss": 0.0188, "step": 3490 }, { "epoch": 0.45, "learning_rate": 1.7381160433414893e-05, "loss": 0.0153, "step": 3500 }, { "epoch": 0.45, "learning_rate": 1.7372422229989517e-05, "loss": 0.017, "step": 3510 }, { "epoch": 0.45, "learning_rate": 1.736368402656414e-05, "loss": 0.0169, "step": 3520 }, { "epoch": 0.45, "learning_rate": 1.7354945823138763e-05, "loss": 0.0149, "step": 3530 }, { "epoch": 0.45, "learning_rate": 1.734620761971339e-05, "loss": 0.0127, "step": 3540 }, { "epoch": 0.46, "learning_rate": 1.7337469416288013e-05, "loss": 0.0116, "step": 3550 }, { "epoch": 0.46, "learning_rate": 1.7328731212862636e-05, "loss": 0.0153, "step": 3560 }, { "epoch": 0.46, "learning_rate": 1.7319993009437263e-05, "loss": 0.0155, "step": 3570 }, { "epoch": 0.46, "learning_rate": 1.7311254806011886e-05, "loss": 0.019, "step": 3580 }, { "epoch": 0.46, "learning_rate": 1.730251660258651e-05, "loss": 0.0183, "step": 3590 }, { "epoch": 0.46, "learning_rate": 1.7293778399161136e-05, "loss": 0.013, "step": 3600 }, { "epoch": 0.46, "learning_rate": 1.728504019573576e-05, "loss": 0.0203, "step": 3610 }, { "epoch": 0.46, "learning_rate": 1.7276301992310382e-05, "loss": 0.014, "step": 3620 }, { "epoch": 0.47, "learning_rate": 1.7267563788885005e-05, "loss": 0.0125, "step": 3630 }, { "epoch": 0.47, "learning_rate": 1.7258825585459632e-05, "loss": 0.0154, "step": 3640 }, { "epoch": 0.47, "learning_rate": 1.7250087382034255e-05, "loss": 0.0118, "step": 3650 }, { "epoch": 0.47, "learning_rate": 1.724134917860888e-05, "loss": 0.0143, "step": 3660 }, { "epoch": 0.47, "learning_rate": 1.7232610975183505e-05, "loss": 0.0121, "step": 3670 }, { "epoch": 0.47, "learning_rate": 1.722387277175813e-05, "loss": 0.0076, "step": 3680 }, { "epoch": 0.47, "learning_rate": 1.721513456833275e-05, "loss": 0.0185, "step": 3690 }, { "epoch": 0.47, "learning_rate": 1.7206396364907375e-05, "loss": 0.0189, "step": 3700 }, { "epoch": 0.48, "learning_rate": 1.7197658161482e-05, "loss": 0.0171, "step": 3710 }, { "epoch": 0.48, "learning_rate": 1.7188919958056625e-05, "loss": 0.0141, "step": 3720 }, { "epoch": 0.48, "learning_rate": 1.7180181754631248e-05, "loss": 0.0159, "step": 3730 }, { "epoch": 0.48, "learning_rate": 1.7171443551205874e-05, "loss": 0.0157, "step": 3740 }, { "epoch": 0.48, "learning_rate": 1.7162705347780498e-05, "loss": 0.0182, "step": 3750 }, { "epoch": 0.48, "learning_rate": 1.715396714435512e-05, "loss": 0.0142, "step": 3760 }, { "epoch": 0.48, "learning_rate": 1.7145228940929744e-05, "loss": 0.0152, "step": 3770 }, { "epoch": 0.48, "learning_rate": 1.713649073750437e-05, "loss": 0.0192, "step": 3780 }, { "epoch": 0.49, "learning_rate": 1.7127752534078994e-05, "loss": 0.0198, "step": 3790 }, { "epoch": 0.49, "learning_rate": 1.7119014330653617e-05, "loss": 0.0099, "step": 3800 }, { "epoch": 0.49, "learning_rate": 1.7110276127228244e-05, "loss": 0.0195, "step": 3810 }, { "epoch": 0.49, "learning_rate": 1.7101537923802867e-05, "loss": 0.0119, "step": 3820 }, { "epoch": 0.49, "learning_rate": 1.709279972037749e-05, "loss": 0.0169, "step": 3830 }, { "epoch": 0.49, "learning_rate": 1.7084061516952114e-05, "loss": 0.0135, "step": 3840 }, { "epoch": 0.49, "learning_rate": 1.707532331352674e-05, "loss": 0.0101, "step": 3850 }, { "epoch": 0.5, "learning_rate": 1.7066585110101363e-05, "loss": 0.0178, "step": 3860 }, { "epoch": 0.5, "learning_rate": 1.7057846906675987e-05, "loss": 0.0144, "step": 3870 }, { "epoch": 0.5, "learning_rate": 1.7049108703250613e-05, "loss": 0.0136, "step": 3880 }, { "epoch": 0.5, "learning_rate": 1.704037049982524e-05, "loss": 0.0191, "step": 3890 }, { "epoch": 0.5, "learning_rate": 1.7031632296399863e-05, "loss": 0.0175, "step": 3900 }, { "epoch": 0.5, "learning_rate": 1.7022894092974486e-05, "loss": 0.0153, "step": 3910 }, { "epoch": 0.5, "learning_rate": 1.701415588954911e-05, "loss": 0.0134, "step": 3920 }, { "epoch": 0.5, "learning_rate": 1.7005417686123736e-05, "loss": 0.0164, "step": 3930 }, { "epoch": 0.51, "learning_rate": 1.699667948269836e-05, "loss": 0.0204, "step": 3940 }, { "epoch": 0.51, "learning_rate": 1.6987941279272983e-05, "loss": 0.0144, "step": 3950 }, { "epoch": 0.51, "learning_rate": 1.697920307584761e-05, "loss": 0.0169, "step": 3960 }, { "epoch": 0.51, "learning_rate": 1.6970464872422232e-05, "loss": 0.0167, "step": 3970 }, { "epoch": 0.51, "learning_rate": 1.6961726668996856e-05, "loss": 0.0152, "step": 3980 }, { "epoch": 0.51, "learning_rate": 1.695298846557148e-05, "loss": 0.0175, "step": 3990 }, { "epoch": 0.51, "learning_rate": 1.6944250262146105e-05, "loss": 0.0192, "step": 4000 }, { "epoch": 0.51, "learning_rate": 1.693551205872073e-05, "loss": 0.0192, "step": 4010 }, { "epoch": 0.52, "learning_rate": 1.6926773855295352e-05, "loss": 0.016, "step": 4020 }, { "epoch": 0.52, "learning_rate": 1.691803565186998e-05, "loss": 0.0162, "step": 4030 }, { "epoch": 0.52, "learning_rate": 1.6909297448444602e-05, "loss": 0.013, "step": 4040 }, { "epoch": 0.52, "learning_rate": 1.6900559245019225e-05, "loss": 0.0172, "step": 4050 }, { "epoch": 0.52, "learning_rate": 1.689182104159385e-05, "loss": 0.0149, "step": 4060 }, { "epoch": 0.52, "learning_rate": 1.6883082838168475e-05, "loss": 0.0182, "step": 4070 }, { "epoch": 0.52, "learning_rate": 1.6874344634743098e-05, "loss": 0.0191, "step": 4080 }, { "epoch": 0.52, "learning_rate": 1.686560643131772e-05, "loss": 0.0164, "step": 4090 }, { "epoch": 0.53, "learning_rate": 1.6856868227892348e-05, "loss": 0.017, "step": 4100 }, { "epoch": 0.53, "learning_rate": 1.684813002446697e-05, "loss": 0.0139, "step": 4110 }, { "epoch": 0.53, "learning_rate": 1.6839391821041594e-05, "loss": 0.0198, "step": 4120 }, { "epoch": 0.53, "learning_rate": 1.683065361761622e-05, "loss": 0.0113, "step": 4130 }, { "epoch": 0.53, "learning_rate": 1.6821915414190844e-05, "loss": 0.0153, "step": 4140 }, { "epoch": 0.53, "learning_rate": 1.6813177210765467e-05, "loss": 0.0157, "step": 4150 }, { "epoch": 0.53, "learning_rate": 1.680443900734009e-05, "loss": 0.0204, "step": 4160 }, { "epoch": 0.53, "learning_rate": 1.6795700803914717e-05, "loss": 0.0138, "step": 4170 }, { "epoch": 0.54, "learning_rate": 1.678696260048934e-05, "loss": 0.0151, "step": 4180 }, { "epoch": 0.54, "learning_rate": 1.6778224397063964e-05, "loss": 0.0138, "step": 4190 }, { "epoch": 0.54, "learning_rate": 1.676948619363859e-05, "loss": 0.0147, "step": 4200 }, { "epoch": 0.54, "learning_rate": 1.6760747990213214e-05, "loss": 0.0122, "step": 4210 }, { "epoch": 0.54, "learning_rate": 1.6752009786787837e-05, "loss": 0.0161, "step": 4220 }, { "epoch": 0.54, "learning_rate": 1.674327158336246e-05, "loss": 0.0158, "step": 4230 }, { "epoch": 0.54, "learning_rate": 1.6734533379937087e-05, "loss": 0.0176, "step": 4240 }, { "epoch": 0.55, "learning_rate": 1.672579517651171e-05, "loss": 0.0167, "step": 4250 }, { "epoch": 0.55, "learning_rate": 1.6717056973086333e-05, "loss": 0.0166, "step": 4260 }, { "epoch": 0.55, "learning_rate": 1.670831876966096e-05, "loss": 0.0138, "step": 4270 }, { "epoch": 0.55, "learning_rate": 1.6699580566235583e-05, "loss": 0.0156, "step": 4280 }, { "epoch": 0.55, "learning_rate": 1.6690842362810206e-05, "loss": 0.0123, "step": 4290 }, { "epoch": 0.55, "learning_rate": 1.668210415938483e-05, "loss": 0.0133, "step": 4300 }, { "epoch": 0.55, "learning_rate": 1.6673365955959456e-05, "loss": 0.0198, "step": 4310 }, { "epoch": 0.55, "learning_rate": 1.666462775253408e-05, "loss": 0.0169, "step": 4320 }, { "epoch": 0.56, "learning_rate": 1.6655889549108702e-05, "loss": 0.0165, "step": 4330 }, { "epoch": 0.56, "learning_rate": 1.664715134568333e-05, "loss": 0.0156, "step": 4340 }, { "epoch": 0.56, "learning_rate": 1.6638413142257956e-05, "loss": 0.0176, "step": 4350 }, { "epoch": 0.56, "learning_rate": 1.662967493883258e-05, "loss": 0.0165, "step": 4360 }, { "epoch": 0.56, "learning_rate": 1.6620936735407202e-05, "loss": 0.0177, "step": 4370 }, { "epoch": 0.56, "learning_rate": 1.6612198531981825e-05, "loss": 0.0133, "step": 4380 }, { "epoch": 0.56, "learning_rate": 1.6603460328556452e-05, "loss": 0.017, "step": 4390 }, { "epoch": 0.56, "learning_rate": 1.6594722125131075e-05, "loss": 0.0101, "step": 4400 }, { "epoch": 0.57, "learning_rate": 1.65859839217057e-05, "loss": 0.0158, "step": 4410 }, { "epoch": 0.57, "learning_rate": 1.6577245718280325e-05, "loss": 0.0173, "step": 4420 }, { "epoch": 0.57, "learning_rate": 1.6568507514854948e-05, "loss": 0.0204, "step": 4430 }, { "epoch": 0.57, "learning_rate": 1.655976931142957e-05, "loss": 0.0128, "step": 4440 }, { "epoch": 0.57, "learning_rate": 1.6551031108004195e-05, "loss": 0.0195, "step": 4450 }, { "epoch": 0.57, "learning_rate": 1.654229290457882e-05, "loss": 0.0142, "step": 4460 }, { "epoch": 0.57, "learning_rate": 1.6533554701153445e-05, "loss": 0.0132, "step": 4470 }, { "epoch": 0.57, "learning_rate": 1.6524816497728068e-05, "loss": 0.0168, "step": 4480 }, { "epoch": 0.58, "learning_rate": 1.6516078294302694e-05, "loss": 0.0142, "step": 4490 }, { "epoch": 0.58, "learning_rate": 1.6507340090877318e-05, "loss": 0.0157, "step": 4500 }, { "epoch": 0.58, "learning_rate": 1.649860188745194e-05, "loss": 0.0162, "step": 4510 }, { "epoch": 0.58, "learning_rate": 1.6489863684026567e-05, "loss": 0.0176, "step": 4520 }, { "epoch": 0.58, "learning_rate": 1.648112548060119e-05, "loss": 0.0169, "step": 4530 }, { "epoch": 0.58, "learning_rate": 1.6472387277175814e-05, "loss": 0.0146, "step": 4540 }, { "epoch": 0.58, "learning_rate": 1.6463649073750437e-05, "loss": 0.0159, "step": 4550 }, { "epoch": 0.58, "learning_rate": 1.6454910870325064e-05, "loss": 0.0227, "step": 4560 }, { "epoch": 0.59, "learning_rate": 1.6446172666899687e-05, "loss": 0.0159, "step": 4570 }, { "epoch": 0.59, "learning_rate": 1.643743446347431e-05, "loss": 0.0152, "step": 4580 }, { "epoch": 0.59, "learning_rate": 1.6428696260048937e-05, "loss": 0.0145, "step": 4590 }, { "epoch": 0.59, "learning_rate": 1.641995805662356e-05, "loss": 0.0146, "step": 4600 }, { "epoch": 0.59, "learning_rate": 1.6411219853198183e-05, "loss": 0.0139, "step": 4610 }, { "epoch": 0.59, "learning_rate": 1.6402481649772807e-05, "loss": 0.0127, "step": 4620 }, { "epoch": 0.59, "learning_rate": 1.6393743446347433e-05, "loss": 0.0163, "step": 4630 }, { "epoch": 0.6, "learning_rate": 1.6385005242922056e-05, "loss": 0.0174, "step": 4640 }, { "epoch": 0.6, "learning_rate": 1.637626703949668e-05, "loss": 0.0159, "step": 4650 }, { "epoch": 0.6, "learning_rate": 1.6367528836071306e-05, "loss": 0.0118, "step": 4660 }, { "epoch": 0.6, "learning_rate": 1.635879063264593e-05, "loss": 0.0163, "step": 4670 }, { "epoch": 0.6, "learning_rate": 1.6350052429220553e-05, "loss": 0.017, "step": 4680 }, { "epoch": 0.6, "learning_rate": 1.6341314225795176e-05, "loss": 0.0123, "step": 4690 }, { "epoch": 0.6, "learning_rate": 1.6332576022369802e-05, "loss": 0.0155, "step": 4700 }, { "epoch": 0.6, "learning_rate": 1.6323837818944426e-05, "loss": 0.015, "step": 4710 }, { "epoch": 0.61, "learning_rate": 1.631509961551905e-05, "loss": 0.0117, "step": 4720 }, { "epoch": 0.61, "learning_rate": 1.6306361412093676e-05, "loss": 0.0156, "step": 4730 }, { "epoch": 0.61, "learning_rate": 1.62976232086683e-05, "loss": 0.0147, "step": 4740 }, { "epoch": 0.61, "learning_rate": 1.6288885005242922e-05, "loss": 0.0145, "step": 4750 }, { "epoch": 0.61, "learning_rate": 1.6280146801817545e-05, "loss": 0.0125, "step": 4760 }, { "epoch": 0.61, "learning_rate": 1.6271408598392172e-05, "loss": 0.0146, "step": 4770 }, { "epoch": 0.61, "learning_rate": 1.6262670394966795e-05, "loss": 0.0085, "step": 4780 }, { "epoch": 0.61, "learning_rate": 1.6253932191541418e-05, "loss": 0.019, "step": 4790 }, { "epoch": 0.62, "learning_rate": 1.6245193988116045e-05, "loss": 0.0162, "step": 4800 }, { "epoch": 0.62, "learning_rate": 1.6236455784690668e-05, "loss": 0.0144, "step": 4810 }, { "epoch": 0.62, "learning_rate": 1.6227717581265295e-05, "loss": 0.0164, "step": 4820 }, { "epoch": 0.62, "learning_rate": 1.6218979377839918e-05, "loss": 0.0162, "step": 4830 }, { "epoch": 0.62, "learning_rate": 1.621024117441454e-05, "loss": 0.0135, "step": 4840 }, { "epoch": 0.62, "learning_rate": 1.6201502970989168e-05, "loss": 0.0115, "step": 4850 }, { "epoch": 0.62, "learning_rate": 1.619276476756379e-05, "loss": 0.0142, "step": 4860 }, { "epoch": 0.62, "learning_rate": 1.6184026564138414e-05, "loss": 0.0143, "step": 4870 }, { "epoch": 0.63, "learning_rate": 1.617528836071304e-05, "loss": 0.0149, "step": 4880 }, { "epoch": 0.63, "learning_rate": 1.6166550157287664e-05, "loss": 0.0147, "step": 4890 }, { "epoch": 0.63, "learning_rate": 1.6157811953862287e-05, "loss": 0.0117, "step": 4900 }, { "epoch": 0.63, "learning_rate": 1.614907375043691e-05, "loss": 0.0142, "step": 4910 }, { "epoch": 0.63, "learning_rate": 1.6140335547011537e-05, "loss": 0.0155, "step": 4920 }, { "epoch": 0.63, "learning_rate": 1.613159734358616e-05, "loss": 0.0144, "step": 4930 }, { "epoch": 0.63, "learning_rate": 1.6122859140160784e-05, "loss": 0.0096, "step": 4940 }, { "epoch": 0.63, "learning_rate": 1.611412093673541e-05, "loss": 0.0197, "step": 4950 }, { "epoch": 0.64, "learning_rate": 1.6105382733310033e-05, "loss": 0.0171, "step": 4960 }, { "epoch": 0.64, "learning_rate": 1.6096644529884657e-05, "loss": 0.0149, "step": 4970 }, { "epoch": 0.64, "learning_rate": 1.6087906326459283e-05, "loss": 0.0127, "step": 4980 }, { "epoch": 0.64, "learning_rate": 1.6079168123033907e-05, "loss": 0.0123, "step": 4990 }, { "epoch": 0.64, "learning_rate": 1.607042991960853e-05, "loss": 0.019, "step": 5000 }, { "epoch": 0.64, "learning_rate": 1.6061691716183153e-05, "loss": 0.0126, "step": 5010 }, { "epoch": 0.64, "learning_rate": 1.605295351275778e-05, "loss": 0.0157, "step": 5020 }, { "epoch": 0.65, "learning_rate": 1.6044215309332403e-05, "loss": 0.0139, "step": 5030 }, { "epoch": 0.65, "learning_rate": 1.6035477105907026e-05, "loss": 0.0096, "step": 5040 }, { "epoch": 0.65, "learning_rate": 1.6026738902481653e-05, "loss": 0.0185, "step": 5050 }, { "epoch": 0.65, "learning_rate": 1.6018000699056276e-05, "loss": 0.0113, "step": 5060 }, { "epoch": 0.65, "learning_rate": 1.60092624956309e-05, "loss": 0.0174, "step": 5070 }, { "epoch": 0.65, "learning_rate": 1.6000524292205522e-05, "loss": 0.0202, "step": 5080 }, { "epoch": 0.65, "learning_rate": 1.599178608878015e-05, "loss": 0.0191, "step": 5090 }, { "epoch": 0.65, "learning_rate": 1.5983047885354772e-05, "loss": 0.0185, "step": 5100 }, { "epoch": 0.66, "learning_rate": 1.5974309681929395e-05, "loss": 0.0152, "step": 5110 }, { "epoch": 0.66, "learning_rate": 1.5965571478504022e-05, "loss": 0.0162, "step": 5120 }, { "epoch": 0.66, "learning_rate": 1.5956833275078645e-05, "loss": 0.0166, "step": 5130 }, { "epoch": 0.66, "learning_rate": 1.594809507165327e-05, "loss": 0.0127, "step": 5140 }, { "epoch": 0.66, "learning_rate": 1.5939356868227892e-05, "loss": 0.0135, "step": 5150 }, { "epoch": 0.66, "learning_rate": 1.593061866480252e-05, "loss": 0.0158, "step": 5160 }, { "epoch": 0.66, "learning_rate": 1.592188046137714e-05, "loss": 0.0185, "step": 5170 }, { "epoch": 0.66, "learning_rate": 1.5913142257951765e-05, "loss": 0.0134, "step": 5180 }, { "epoch": 0.67, "learning_rate": 1.590440405452639e-05, "loss": 0.0179, "step": 5190 }, { "epoch": 0.67, "learning_rate": 1.5895665851101015e-05, "loss": 0.0172, "step": 5200 }, { "epoch": 0.67, "learning_rate": 1.5886927647675638e-05, "loss": 0.0141, "step": 5210 }, { "epoch": 0.67, "learning_rate": 1.587818944425026e-05, "loss": 0.0171, "step": 5220 }, { "epoch": 0.67, "learning_rate": 1.5869451240824888e-05, "loss": 0.0131, "step": 5230 }, { "epoch": 0.67, "learning_rate": 1.586071303739951e-05, "loss": 0.017, "step": 5240 }, { "epoch": 0.67, "learning_rate": 1.5851974833974134e-05, "loss": 0.0137, "step": 5250 }, { "epoch": 0.67, "learning_rate": 1.584323663054876e-05, "loss": 0.0114, "step": 5260 }, { "epoch": 0.68, "learning_rate": 1.5834498427123384e-05, "loss": 0.0138, "step": 5270 }, { "epoch": 0.68, "learning_rate": 1.5825760223698007e-05, "loss": 0.0213, "step": 5280 }, { "epoch": 0.68, "learning_rate": 1.5817022020272634e-05, "loss": 0.0113, "step": 5290 }, { "epoch": 0.68, "learning_rate": 1.5808283816847257e-05, "loss": 0.016, "step": 5300 }, { "epoch": 0.68, "learning_rate": 1.5799545613421884e-05, "loss": 0.0191, "step": 5310 }, { "epoch": 0.68, "learning_rate": 1.5790807409996507e-05, "loss": 0.0135, "step": 5320 }, { "epoch": 0.68, "learning_rate": 1.578206920657113e-05, "loss": 0.0144, "step": 5330 }, { "epoch": 0.68, "learning_rate": 1.5773331003145757e-05, "loss": 0.0146, "step": 5340 }, { "epoch": 0.69, "learning_rate": 1.576459279972038e-05, "loss": 0.0181, "step": 5350 }, { "epoch": 0.69, "learning_rate": 1.5755854596295003e-05, "loss": 0.0147, "step": 5360 }, { "epoch": 0.69, "learning_rate": 1.5747116392869626e-05, "loss": 0.016, "step": 5370 }, { "epoch": 0.69, "learning_rate": 1.5738378189444253e-05, "loss": 0.0155, "step": 5380 }, { "epoch": 0.69, "learning_rate": 1.5729639986018876e-05, "loss": 0.0146, "step": 5390 }, { "epoch": 0.69, "learning_rate": 1.57209017825935e-05, "loss": 0.0114, "step": 5400 }, { "epoch": 0.69, "learning_rate": 1.5712163579168126e-05, "loss": 0.0177, "step": 5410 }, { "epoch": 0.7, "learning_rate": 1.570342537574275e-05, "loss": 0.0163, "step": 5420 }, { "epoch": 0.7, "learning_rate": 1.5694687172317373e-05, "loss": 0.0148, "step": 5430 }, { "epoch": 0.7, "learning_rate": 1.5685948968891996e-05, "loss": 0.0139, "step": 5440 }, { "epoch": 0.7, "learning_rate": 1.5677210765466622e-05, "loss": 0.0105, "step": 5450 }, { "epoch": 0.7, "learning_rate": 1.5668472562041246e-05, "loss": 0.0114, "step": 5460 }, { "epoch": 0.7, "learning_rate": 1.565973435861587e-05, "loss": 0.0118, "step": 5470 }, { "epoch": 0.7, "learning_rate": 1.5650996155190495e-05, "loss": 0.0182, "step": 5480 }, { "epoch": 0.7, "learning_rate": 1.564225795176512e-05, "loss": 0.0154, "step": 5490 }, { "epoch": 0.71, "learning_rate": 1.5633519748339742e-05, "loss": 0.0155, "step": 5500 }, { "epoch": 0.71, "learning_rate": 1.562478154491437e-05, "loss": 0.0175, "step": 5510 }, { "epoch": 0.71, "learning_rate": 1.5616043341488992e-05, "loss": 0.0113, "step": 5520 }, { "epoch": 0.71, "learning_rate": 1.5607305138063615e-05, "loss": 0.0153, "step": 5530 }, { "epoch": 0.71, "learning_rate": 1.5598566934638238e-05, "loss": 0.0101, "step": 5540 }, { "epoch": 0.71, "learning_rate": 1.5589828731212865e-05, "loss": 0.0161, "step": 5550 }, { "epoch": 0.71, "learning_rate": 1.5581090527787488e-05, "loss": 0.0135, "step": 5560 }, { "epoch": 0.71, "learning_rate": 1.557235232436211e-05, "loss": 0.0134, "step": 5570 }, { "epoch": 0.72, "learning_rate": 1.5563614120936738e-05, "loss": 0.0139, "step": 5580 }, { "epoch": 0.72, "learning_rate": 1.555487591751136e-05, "loss": 0.0135, "step": 5590 }, { "epoch": 0.72, "learning_rate": 1.5546137714085984e-05, "loss": 0.0112, "step": 5600 }, { "epoch": 0.72, "learning_rate": 1.5537399510660608e-05, "loss": 0.0107, "step": 5610 }, { "epoch": 0.72, "learning_rate": 1.5528661307235234e-05, "loss": 0.0152, "step": 5620 }, { "epoch": 0.72, "learning_rate": 1.5519923103809857e-05, "loss": 0.0183, "step": 5630 }, { "epoch": 0.72, "learning_rate": 1.551118490038448e-05, "loss": 0.0164, "step": 5640 }, { "epoch": 0.72, "learning_rate": 1.5502446696959107e-05, "loss": 0.0181, "step": 5650 }, { "epoch": 0.73, "learning_rate": 1.549370849353373e-05, "loss": 0.0135, "step": 5660 }, { "epoch": 0.73, "learning_rate": 1.5484970290108354e-05, "loss": 0.0148, "step": 5670 }, { "epoch": 0.73, "learning_rate": 1.5476232086682977e-05, "loss": 0.0162, "step": 5680 }, { "epoch": 0.73, "learning_rate": 1.5467493883257604e-05, "loss": 0.0102, "step": 5690 }, { "epoch": 0.73, "learning_rate": 1.5458755679832227e-05, "loss": 0.0154, "step": 5700 }, { "epoch": 0.73, "learning_rate": 1.545001747640685e-05, "loss": 0.0138, "step": 5710 }, { "epoch": 0.73, "learning_rate": 1.5441279272981477e-05, "loss": 0.0181, "step": 5720 }, { "epoch": 0.73, "learning_rate": 1.54325410695561e-05, "loss": 0.0131, "step": 5730 }, { "epoch": 0.74, "learning_rate": 1.5423802866130723e-05, "loss": 0.0189, "step": 5740 }, { "epoch": 0.74, "learning_rate": 1.5415064662705346e-05, "loss": 0.0134, "step": 5750 }, { "epoch": 0.74, "learning_rate": 1.5406326459279973e-05, "loss": 0.0147, "step": 5760 }, { "epoch": 0.74, "learning_rate": 1.53975882558546e-05, "loss": 0.0174, "step": 5770 }, { "epoch": 0.74, "learning_rate": 1.5388850052429223e-05, "loss": 0.0104, "step": 5780 }, { "epoch": 0.74, "learning_rate": 1.5380111849003846e-05, "loss": 0.0147, "step": 5790 }, { "epoch": 0.74, "learning_rate": 1.5371373645578473e-05, "loss": 0.0147, "step": 5800 }, { "epoch": 0.75, "learning_rate": 1.5362635442153096e-05, "loss": 0.0136, "step": 5810 }, { "epoch": 0.75, "learning_rate": 1.535389723872772e-05, "loss": 0.0146, "step": 5820 }, { "epoch": 0.75, "learning_rate": 1.5345159035302342e-05, "loss": 0.0149, "step": 5830 }, { "epoch": 0.75, "learning_rate": 1.533642083187697e-05, "loss": 0.0128, "step": 5840 }, { "epoch": 0.75, "learning_rate": 1.5327682628451592e-05, "loss": 0.0116, "step": 5850 }, { "epoch": 0.75, "learning_rate": 1.5318944425026215e-05, "loss": 0.0147, "step": 5860 }, { "epoch": 0.75, "learning_rate": 1.5310206221600842e-05, "loss": 0.0142, "step": 5870 }, { "epoch": 0.75, "learning_rate": 1.5301468018175465e-05, "loss": 0.0149, "step": 5880 }, { "epoch": 0.76, "learning_rate": 1.529272981475009e-05, "loss": 0.0171, "step": 5890 }, { "epoch": 0.76, "learning_rate": 1.528399161132471e-05, "loss": 0.0181, "step": 5900 }, { "epoch": 0.76, "learning_rate": 1.5275253407899338e-05, "loss": 0.0192, "step": 5910 }, { "epoch": 0.76, "learning_rate": 1.526651520447396e-05, "loss": 0.0113, "step": 5920 }, { "epoch": 0.76, "learning_rate": 1.5257777001048586e-05, "loss": 0.0155, "step": 5930 }, { "epoch": 0.76, "learning_rate": 1.524903879762321e-05, "loss": 0.0152, "step": 5940 }, { "epoch": 0.76, "learning_rate": 1.5240300594197835e-05, "loss": 0.0164, "step": 5950 }, { "epoch": 0.76, "learning_rate": 1.5231562390772458e-05, "loss": 0.0085, "step": 5960 }, { "epoch": 0.77, "learning_rate": 1.5222824187347083e-05, "loss": 0.0195, "step": 5970 }, { "epoch": 0.77, "learning_rate": 1.5214085983921708e-05, "loss": 0.0156, "step": 5980 }, { "epoch": 0.77, "learning_rate": 1.520534778049633e-05, "loss": 0.0179, "step": 5990 }, { "epoch": 0.77, "learning_rate": 1.5196609577070956e-05, "loss": 0.01, "step": 6000 }, { "epoch": 0.77, "learning_rate": 1.5187871373645579e-05, "loss": 0.0156, "step": 6010 }, { "epoch": 0.77, "learning_rate": 1.5179133170220204e-05, "loss": 0.0148, "step": 6020 }, { "epoch": 0.77, "learning_rate": 1.5170394966794829e-05, "loss": 0.0187, "step": 6030 }, { "epoch": 0.77, "learning_rate": 1.5161656763369452e-05, "loss": 0.0147, "step": 6040 }, { "epoch": 0.78, "learning_rate": 1.5152918559944077e-05, "loss": 0.0134, "step": 6050 }, { "epoch": 0.78, "learning_rate": 1.51441803565187e-05, "loss": 0.0158, "step": 6060 }, { "epoch": 0.78, "learning_rate": 1.5135442153093325e-05, "loss": 0.0125, "step": 6070 }, { "epoch": 0.78, "learning_rate": 1.5126703949667948e-05, "loss": 0.0166, "step": 6080 }, { "epoch": 0.78, "learning_rate": 1.5117965746242573e-05, "loss": 0.0153, "step": 6090 }, { "epoch": 0.78, "learning_rate": 1.5109227542817198e-05, "loss": 0.0151, "step": 6100 }, { "epoch": 0.78, "learning_rate": 1.5100489339391821e-05, "loss": 0.0154, "step": 6110 }, { "epoch": 0.78, "learning_rate": 1.5091751135966446e-05, "loss": 0.0127, "step": 6120 }, { "epoch": 0.79, "learning_rate": 1.508301293254107e-05, "loss": 0.0134, "step": 6130 }, { "epoch": 0.79, "learning_rate": 1.5074274729115694e-05, "loss": 0.0173, "step": 6140 }, { "epoch": 0.79, "learning_rate": 1.5065536525690318e-05, "loss": 0.0122, "step": 6150 }, { "epoch": 0.79, "learning_rate": 1.5056798322264943e-05, "loss": 0.0153, "step": 6160 }, { "epoch": 0.79, "learning_rate": 1.5048060118839568e-05, "loss": 0.0133, "step": 6170 }, { "epoch": 0.79, "learning_rate": 1.503932191541419e-05, "loss": 0.011, "step": 6180 }, { "epoch": 0.79, "learning_rate": 1.5030583711988816e-05, "loss": 0.0117, "step": 6190 }, { "epoch": 0.8, "learning_rate": 1.5021845508563439e-05, "loss": 0.017, "step": 6200 }, { "epoch": 0.8, "learning_rate": 1.5013107305138064e-05, "loss": 0.0214, "step": 6210 }, { "epoch": 0.8, "learning_rate": 1.5004369101712689e-05, "loss": 0.019, "step": 6220 }, { "epoch": 0.8, "learning_rate": 1.4995630898287314e-05, "loss": 0.0151, "step": 6230 }, { "epoch": 0.8, "learning_rate": 1.4986892694861939e-05, "loss": 0.0114, "step": 6240 }, { "epoch": 0.8, "learning_rate": 1.4978154491436564e-05, "loss": 0.0162, "step": 6250 }, { "epoch": 0.8, "learning_rate": 1.4969416288011187e-05, "loss": 0.0189, "step": 6260 }, { "epoch": 0.8, "learning_rate": 1.4960678084585812e-05, "loss": 0.0144, "step": 6270 }, { "epoch": 0.81, "learning_rate": 1.4951939881160435e-05, "loss": 0.0114, "step": 6280 }, { "epoch": 0.81, "learning_rate": 1.494320167773506e-05, "loss": 0.0142, "step": 6290 }, { "epoch": 0.81, "learning_rate": 1.4934463474309685e-05, "loss": 0.0129, "step": 6300 }, { "epoch": 0.81, "learning_rate": 1.4925725270884308e-05, "loss": 0.0178, "step": 6310 }, { "epoch": 0.81, "learning_rate": 1.4916987067458933e-05, "loss": 0.0113, "step": 6320 }, { "epoch": 0.81, "learning_rate": 1.4908248864033556e-05, "loss": 0.0175, "step": 6330 }, { "epoch": 0.81, "learning_rate": 1.4899510660608181e-05, "loss": 0.0168, "step": 6340 }, { "epoch": 0.81, "learning_rate": 1.4890772457182804e-05, "loss": 0.0113, "step": 6350 }, { "epoch": 0.82, "learning_rate": 1.488203425375743e-05, "loss": 0.0146, "step": 6360 }, { "epoch": 0.82, "learning_rate": 1.4873296050332054e-05, "loss": 0.0145, "step": 6370 }, { "epoch": 0.82, "learning_rate": 1.4864557846906677e-05, "loss": 0.0149, "step": 6380 }, { "epoch": 0.82, "learning_rate": 1.4855819643481302e-05, "loss": 0.0129, "step": 6390 }, { "epoch": 0.82, "learning_rate": 1.4847081440055925e-05, "loss": 0.0165, "step": 6400 }, { "epoch": 0.82, "learning_rate": 1.483834323663055e-05, "loss": 0.0113, "step": 6410 }, { "epoch": 0.82, "learning_rate": 1.4829605033205174e-05, "loss": 0.0176, "step": 6420 }, { "epoch": 0.82, "learning_rate": 1.4820866829779799e-05, "loss": 0.0157, "step": 6430 }, { "epoch": 0.83, "learning_rate": 1.4812128626354423e-05, "loss": 0.0135, "step": 6440 }, { "epoch": 0.83, "learning_rate": 1.4803390422929047e-05, "loss": 0.0138, "step": 6450 }, { "epoch": 0.83, "learning_rate": 1.4794652219503672e-05, "loss": 0.0124, "step": 6460 }, { "epoch": 0.83, "learning_rate": 1.4785914016078295e-05, "loss": 0.0125, "step": 6470 }, { "epoch": 0.83, "learning_rate": 1.477717581265292e-05, "loss": 0.0131, "step": 6480 }, { "epoch": 0.83, "learning_rate": 1.4768437609227545e-05, "loss": 0.0159, "step": 6490 }, { "epoch": 0.83, "learning_rate": 1.4759699405802168e-05, "loss": 0.0146, "step": 6500 }, { "epoch": 0.84, "learning_rate": 1.4750961202376793e-05, "loss": 0.0135, "step": 6510 }, { "epoch": 0.84, "learning_rate": 1.4742222998951416e-05, "loss": 0.0129, "step": 6520 }, { "epoch": 0.84, "learning_rate": 1.4733484795526041e-05, "loss": 0.0177, "step": 6530 }, { "epoch": 0.84, "learning_rate": 1.4724746592100664e-05, "loss": 0.0167, "step": 6540 }, { "epoch": 0.84, "learning_rate": 1.4716008388675289e-05, "loss": 0.0156, "step": 6550 }, { "epoch": 0.84, "learning_rate": 1.4707270185249914e-05, "loss": 0.0189, "step": 6560 }, { "epoch": 0.84, "learning_rate": 1.4698531981824537e-05, "loss": 0.0101, "step": 6570 }, { "epoch": 0.84, "learning_rate": 1.4689793778399162e-05, "loss": 0.0161, "step": 6580 }, { "epoch": 0.85, "learning_rate": 1.4681055574973785e-05, "loss": 0.0165, "step": 6590 }, { "epoch": 0.85, "learning_rate": 1.467231737154841e-05, "loss": 0.0149, "step": 6600 }, { "epoch": 0.85, "learning_rate": 1.4663579168123034e-05, "loss": 0.0179, "step": 6610 }, { "epoch": 0.85, "learning_rate": 1.4654840964697658e-05, "loss": 0.0133, "step": 6620 }, { "epoch": 0.85, "learning_rate": 1.4646102761272283e-05, "loss": 0.0154, "step": 6630 }, { "epoch": 0.85, "learning_rate": 1.4637364557846907e-05, "loss": 0.0175, "step": 6640 }, { "epoch": 0.85, "learning_rate": 1.4628626354421532e-05, "loss": 0.0143, "step": 6650 }, { "epoch": 0.85, "learning_rate": 1.4619888150996155e-05, "loss": 0.0149, "step": 6660 }, { "epoch": 0.86, "learning_rate": 1.461114994757078e-05, "loss": 0.0133, "step": 6670 }, { "epoch": 0.86, "learning_rate": 1.4602411744145405e-05, "loss": 0.0149, "step": 6680 }, { "epoch": 0.86, "learning_rate": 1.4593673540720028e-05, "loss": 0.0145, "step": 6690 }, { "epoch": 0.86, "learning_rate": 1.4584935337294654e-05, "loss": 0.0112, "step": 6700 }, { "epoch": 0.86, "learning_rate": 1.457619713386928e-05, "loss": 0.0103, "step": 6710 }, { "epoch": 0.86, "learning_rate": 1.4567458930443903e-05, "loss": 0.0125, "step": 6720 }, { "epoch": 0.86, "learning_rate": 1.4558720727018528e-05, "loss": 0.0195, "step": 6730 }, { "epoch": 0.86, "learning_rate": 1.454998252359315e-05, "loss": 0.011, "step": 6740 }, { "epoch": 0.87, "learning_rate": 1.4541244320167776e-05, "loss": 0.0171, "step": 6750 }, { "epoch": 0.87, "learning_rate": 1.4532506116742399e-05, "loss": 0.0123, "step": 6760 }, { "epoch": 0.87, "learning_rate": 1.4523767913317024e-05, "loss": 0.0216, "step": 6770 }, { "epoch": 0.87, "learning_rate": 1.4515029709891649e-05, "loss": 0.0175, "step": 6780 }, { "epoch": 0.87, "learning_rate": 1.4506291506466272e-05, "loss": 0.0142, "step": 6790 }, { "epoch": 0.87, "learning_rate": 1.4497553303040897e-05, "loss": 0.0145, "step": 6800 }, { "epoch": 0.87, "learning_rate": 1.448881509961552e-05, "loss": 0.0158, "step": 6810 }, { "epoch": 0.87, "learning_rate": 1.4480076896190145e-05, "loss": 0.0181, "step": 6820 }, { "epoch": 0.88, "learning_rate": 1.447133869276477e-05, "loss": 0.0192, "step": 6830 }, { "epoch": 0.88, "learning_rate": 1.4462600489339393e-05, "loss": 0.0192, "step": 6840 }, { "epoch": 0.88, "learning_rate": 1.4453862285914018e-05, "loss": 0.0123, "step": 6850 }, { "epoch": 0.88, "learning_rate": 1.4445124082488641e-05, "loss": 0.0172, "step": 6860 }, { "epoch": 0.88, "learning_rate": 1.4436385879063266e-05, "loss": 0.0194, "step": 6870 }, { "epoch": 0.88, "learning_rate": 1.442764767563789e-05, "loss": 0.0159, "step": 6880 }, { "epoch": 0.88, "learning_rate": 1.4418909472212514e-05, "loss": 0.011, "step": 6890 }, { "epoch": 0.89, "learning_rate": 1.441017126878714e-05, "loss": 0.0131, "step": 6900 }, { "epoch": 0.89, "learning_rate": 1.4401433065361763e-05, "loss": 0.0111, "step": 6910 }, { "epoch": 0.89, "learning_rate": 1.4392694861936387e-05, "loss": 0.0148, "step": 6920 }, { "epoch": 0.89, "learning_rate": 1.438395665851101e-05, "loss": 0.0193, "step": 6930 }, { "epoch": 0.89, "learning_rate": 1.4375218455085636e-05, "loss": 0.0175, "step": 6940 }, { "epoch": 0.89, "learning_rate": 1.436648025166026e-05, "loss": 0.0165, "step": 6950 }, { "epoch": 0.89, "learning_rate": 1.4357742048234884e-05, "loss": 0.0186, "step": 6960 }, { "epoch": 0.89, "learning_rate": 1.4349003844809509e-05, "loss": 0.0155, "step": 6970 }, { "epoch": 0.9, "learning_rate": 1.4340265641384132e-05, "loss": 0.0196, "step": 6980 }, { "epoch": 0.9, "learning_rate": 1.4331527437958757e-05, "loss": 0.0219, "step": 6990 }, { "epoch": 0.9, "learning_rate": 1.432278923453338e-05, "loss": 0.0146, "step": 7000 }, { "epoch": 0.9, "learning_rate": 1.4314051031108005e-05, "loss": 0.0134, "step": 7010 }, { "epoch": 0.9, "learning_rate": 1.430531282768263e-05, "loss": 0.0142, "step": 7020 }, { "epoch": 0.9, "learning_rate": 1.4296574624257253e-05, "loss": 0.0196, "step": 7030 }, { "epoch": 0.9, "learning_rate": 1.4287836420831878e-05, "loss": 0.0142, "step": 7040 }, { "epoch": 0.9, "learning_rate": 1.4279098217406501e-05, "loss": 0.0146, "step": 7050 }, { "epoch": 0.91, "learning_rate": 1.4270360013981126e-05, "loss": 0.0174, "step": 7060 }, { "epoch": 0.91, "learning_rate": 1.426162181055575e-05, "loss": 0.0132, "step": 7070 }, { "epoch": 0.91, "learning_rate": 1.4252883607130374e-05, "loss": 0.0159, "step": 7080 }, { "epoch": 0.91, "learning_rate": 1.4244145403705e-05, "loss": 0.015, "step": 7090 }, { "epoch": 0.91, "learning_rate": 1.4235407200279622e-05, "loss": 0.0137, "step": 7100 }, { "epoch": 0.91, "learning_rate": 1.4226668996854247e-05, "loss": 0.014, "step": 7110 }, { "epoch": 0.91, "learning_rate": 1.421793079342887e-05, "loss": 0.0169, "step": 7120 }, { "epoch": 0.91, "learning_rate": 1.4209192590003496e-05, "loss": 0.0148, "step": 7130 }, { "epoch": 0.92, "learning_rate": 1.420045438657812e-05, "loss": 0.0133, "step": 7140 }, { "epoch": 0.92, "learning_rate": 1.4191716183152744e-05, "loss": 0.0139, "step": 7150 }, { "epoch": 0.92, "learning_rate": 1.4182977979727369e-05, "loss": 0.0068, "step": 7160 }, { "epoch": 0.92, "learning_rate": 1.4174239776301995e-05, "loss": 0.0169, "step": 7170 }, { "epoch": 0.92, "learning_rate": 1.4165501572876618e-05, "loss": 0.0119, "step": 7180 }, { "epoch": 0.92, "learning_rate": 1.4156763369451243e-05, "loss": 0.0174, "step": 7190 }, { "epoch": 0.92, "learning_rate": 1.4148025166025867e-05, "loss": 0.0109, "step": 7200 }, { "epoch": 0.92, "learning_rate": 1.4139286962600492e-05, "loss": 0.0133, "step": 7210 }, { "epoch": 0.93, "learning_rate": 1.4130548759175115e-05, "loss": 0.0151, "step": 7220 }, { "epoch": 0.93, "learning_rate": 1.412181055574974e-05, "loss": 0.0164, "step": 7230 }, { "epoch": 0.93, "learning_rate": 1.4113072352324365e-05, "loss": 0.0124, "step": 7240 }, { "epoch": 0.93, "learning_rate": 1.4104334148898988e-05, "loss": 0.0173, "step": 7250 }, { "epoch": 0.93, "learning_rate": 1.4095595945473613e-05, "loss": 0.0132, "step": 7260 }, { "epoch": 0.93, "learning_rate": 1.4086857742048236e-05, "loss": 0.0099, "step": 7270 }, { "epoch": 0.93, "learning_rate": 1.4078119538622861e-05, "loss": 0.0136, "step": 7280 }, { "epoch": 0.94, "learning_rate": 1.4069381335197486e-05, "loss": 0.014, "step": 7290 }, { "epoch": 0.94, "learning_rate": 1.4060643131772109e-05, "loss": 0.0125, "step": 7300 }, { "epoch": 0.94, "learning_rate": 1.4051904928346734e-05, "loss": 0.0152, "step": 7310 }, { "epoch": 0.94, "learning_rate": 1.4043166724921357e-05, "loss": 0.0189, "step": 7320 }, { "epoch": 0.94, "learning_rate": 1.4034428521495982e-05, "loss": 0.0224, "step": 7330 }, { "epoch": 0.94, "learning_rate": 1.4025690318070605e-05, "loss": 0.0149, "step": 7340 }, { "epoch": 0.94, "learning_rate": 1.401695211464523e-05, "loss": 0.0162, "step": 7350 }, { "epoch": 0.94, "learning_rate": 1.4008213911219855e-05, "loss": 0.0165, "step": 7360 }, { "epoch": 0.95, "learning_rate": 1.3999475707794478e-05, "loss": 0.0144, "step": 7370 }, { "epoch": 0.95, "learning_rate": 1.3990737504369103e-05, "loss": 0.0138, "step": 7380 }, { "epoch": 0.95, "learning_rate": 1.3981999300943727e-05, "loss": 0.0137, "step": 7390 }, { "epoch": 0.95, "learning_rate": 1.3973261097518351e-05, "loss": 0.0136, "step": 7400 }, { "epoch": 0.95, "learning_rate": 1.3964522894092976e-05, "loss": 0.0147, "step": 7410 }, { "epoch": 0.95, "learning_rate": 1.39557846906676e-05, "loss": 0.0133, "step": 7420 }, { "epoch": 0.95, "learning_rate": 1.3947046487242225e-05, "loss": 0.0123, "step": 7430 }, { "epoch": 0.95, "learning_rate": 1.3938308283816848e-05, "loss": 0.0138, "step": 7440 }, { "epoch": 0.96, "learning_rate": 1.3929570080391473e-05, "loss": 0.0131, "step": 7450 }, { "epoch": 0.96, "learning_rate": 1.3920831876966096e-05, "loss": 0.0149, "step": 7460 }, { "epoch": 0.96, "learning_rate": 1.391209367354072e-05, "loss": 0.0135, "step": 7470 }, { "epoch": 0.96, "learning_rate": 1.3903355470115346e-05, "loss": 0.0172, "step": 7480 }, { "epoch": 0.96, "learning_rate": 1.3894617266689969e-05, "loss": 0.0137, "step": 7490 }, { "epoch": 0.96, "learning_rate": 1.3885879063264594e-05, "loss": 0.0157, "step": 7500 }, { "epoch": 0.96, "learning_rate": 1.3877140859839217e-05, "loss": 0.0145, "step": 7510 }, { "epoch": 0.96, "learning_rate": 1.3868402656413842e-05, "loss": 0.0109, "step": 7520 }, { "epoch": 0.97, "learning_rate": 1.3859664452988465e-05, "loss": 0.0129, "step": 7530 }, { "epoch": 0.97, "learning_rate": 1.385092624956309e-05, "loss": 0.0147, "step": 7540 }, { "epoch": 0.97, "learning_rate": 1.3842188046137715e-05, "loss": 0.0159, "step": 7550 }, { "epoch": 0.97, "learning_rate": 1.3833449842712338e-05, "loss": 0.0142, "step": 7560 }, { "epoch": 0.97, "learning_rate": 1.3824711639286963e-05, "loss": 0.0143, "step": 7570 }, { "epoch": 0.97, "learning_rate": 1.3815973435861586e-05, "loss": 0.0132, "step": 7580 }, { "epoch": 0.97, "learning_rate": 1.3807235232436211e-05, "loss": 0.0149, "step": 7590 }, { "epoch": 0.97, "learning_rate": 1.3798497029010836e-05, "loss": 0.0161, "step": 7600 }, { "epoch": 0.98, "learning_rate": 1.378975882558546e-05, "loss": 0.0118, "step": 7610 }, { "epoch": 0.98, "learning_rate": 1.3781020622160084e-05, "loss": 0.0102, "step": 7620 }, { "epoch": 0.98, "learning_rate": 1.3772282418734708e-05, "loss": 0.0131, "step": 7630 }, { "epoch": 0.98, "learning_rate": 1.3763544215309334e-05, "loss": 0.0137, "step": 7640 }, { "epoch": 0.98, "learning_rate": 1.375480601188396e-05, "loss": 0.0142, "step": 7650 }, { "epoch": 0.98, "learning_rate": 1.3746067808458582e-05, "loss": 0.011, "step": 7660 }, { "epoch": 0.98, "learning_rate": 1.3737329605033207e-05, "loss": 0.0156, "step": 7670 }, { "epoch": 0.99, "learning_rate": 1.372859140160783e-05, "loss": 0.0132, "step": 7680 }, { "epoch": 0.99, "learning_rate": 1.3719853198182456e-05, "loss": 0.0141, "step": 7690 }, { "epoch": 0.99, "learning_rate": 1.371111499475708e-05, "loss": 0.0125, "step": 7700 }, { "epoch": 0.99, "learning_rate": 1.3702376791331704e-05, "loss": 0.0175, "step": 7710 }, { "epoch": 0.99, "learning_rate": 1.3693638587906329e-05, "loss": 0.0151, "step": 7720 }, { "epoch": 0.99, "learning_rate": 1.3684900384480952e-05, "loss": 0.0132, "step": 7730 }, { "epoch": 0.99, "learning_rate": 1.3676162181055577e-05, "loss": 0.0135, "step": 7740 }, { "epoch": 0.99, "learning_rate": 1.3667423977630202e-05, "loss": 0.0145, "step": 7750 }, { "epoch": 1.0, "learning_rate": 1.3658685774204825e-05, "loss": 0.0088, "step": 7760 }, { "epoch": 1.0, "learning_rate": 1.364994757077945e-05, "loss": 0.0135, "step": 7770 }, { "epoch": 1.0, "learning_rate": 1.3641209367354073e-05, "loss": 0.0137, "step": 7780 }, { "epoch": 1.0, "learning_rate": 1.3632471163928698e-05, "loss": 0.0132, "step": 7790 }, { "epoch": 1.0, "eval_loss": 0.013936151750385761, "eval_runtime": 55.2292, "eval_samples_per_second": 250.954, "eval_steps_per_second": 31.378, "step": 7796 }, { "epoch": 1.0, "learning_rate": 1.3623732960503321e-05, "loss": 0.0141, "step": 7800 }, { "epoch": 1.0, "learning_rate": 1.3614994757077946e-05, "loss": 0.0128, "step": 7810 }, { "epoch": 1.0, "learning_rate": 1.3606256553652571e-05, "loss": 0.0144, "step": 7820 }, { "epoch": 1.0, "learning_rate": 1.3597518350227194e-05, "loss": 0.016, "step": 7830 }, { "epoch": 1.01, "learning_rate": 1.358878014680182e-05, "loss": 0.0157, "step": 7840 }, { "epoch": 1.01, "learning_rate": 1.3580041943376442e-05, "loss": 0.0138, "step": 7850 }, { "epoch": 1.01, "learning_rate": 1.3571303739951067e-05, "loss": 0.0132, "step": 7860 }, { "epoch": 1.01, "learning_rate": 1.3562565536525692e-05, "loss": 0.0202, "step": 7870 }, { "epoch": 1.01, "learning_rate": 1.3553827333100315e-05, "loss": 0.0187, "step": 7880 }, { "epoch": 1.01, "learning_rate": 1.354508912967494e-05, "loss": 0.0141, "step": 7890 }, { "epoch": 1.01, "learning_rate": 1.3536350926249564e-05, "loss": 0.0123, "step": 7900 }, { "epoch": 1.01, "learning_rate": 1.3527612722824189e-05, "loss": 0.0117, "step": 7910 }, { "epoch": 1.02, "learning_rate": 1.3518874519398812e-05, "loss": 0.014, "step": 7920 }, { "epoch": 1.02, "learning_rate": 1.3510136315973437e-05, "loss": 0.0122, "step": 7930 }, { "epoch": 1.02, "learning_rate": 1.3501398112548062e-05, "loss": 0.0112, "step": 7940 }, { "epoch": 1.02, "learning_rate": 1.3492659909122685e-05, "loss": 0.0149, "step": 7950 }, { "epoch": 1.02, "learning_rate": 1.348392170569731e-05, "loss": 0.0136, "step": 7960 }, { "epoch": 1.02, "learning_rate": 1.3475183502271933e-05, "loss": 0.0142, "step": 7970 }, { "epoch": 1.02, "learning_rate": 1.3466445298846558e-05, "loss": 0.016, "step": 7980 }, { "epoch": 1.02, "learning_rate": 1.3457707095421181e-05, "loss": 0.0126, "step": 7990 }, { "epoch": 1.03, "learning_rate": 1.3448968891995806e-05, "loss": 0.0098, "step": 8000 }, { "epoch": 1.03, "learning_rate": 1.3440230688570431e-05, "loss": 0.0157, "step": 8010 }, { "epoch": 1.03, "learning_rate": 1.3431492485145054e-05, "loss": 0.0124, "step": 8020 }, { "epoch": 1.03, "learning_rate": 1.3422754281719679e-05, "loss": 0.014, "step": 8030 }, { "epoch": 1.03, "learning_rate": 1.3414016078294302e-05, "loss": 0.0151, "step": 8040 }, { "epoch": 1.03, "learning_rate": 1.3405277874868927e-05, "loss": 0.0108, "step": 8050 }, { "epoch": 1.03, "learning_rate": 1.3396539671443552e-05, "loss": 0.0131, "step": 8060 }, { "epoch": 1.04, "learning_rate": 1.3387801468018175e-05, "loss": 0.0168, "step": 8070 }, { "epoch": 1.04, "learning_rate": 1.33790632645928e-05, "loss": 0.009, "step": 8080 }, { "epoch": 1.04, "learning_rate": 1.3370325061167424e-05, "loss": 0.0168, "step": 8090 }, { "epoch": 1.04, "learning_rate": 1.3361586857742048e-05, "loss": 0.018, "step": 8100 }, { "epoch": 1.04, "learning_rate": 1.3352848654316675e-05, "loss": 0.0142, "step": 8110 }, { "epoch": 1.04, "learning_rate": 1.3344110450891298e-05, "loss": 0.0181, "step": 8120 }, { "epoch": 1.04, "learning_rate": 1.3335372247465923e-05, "loss": 0.0153, "step": 8130 }, { "epoch": 1.04, "learning_rate": 1.3326634044040546e-05, "loss": 0.0142, "step": 8140 }, { "epoch": 1.05, "learning_rate": 1.3317895840615171e-05, "loss": 0.0127, "step": 8150 }, { "epoch": 1.05, "learning_rate": 1.3309157637189796e-05, "loss": 0.0096, "step": 8160 }, { "epoch": 1.05, "learning_rate": 1.330041943376442e-05, "loss": 0.0125, "step": 8170 }, { "epoch": 1.05, "learning_rate": 1.3291681230339044e-05, "loss": 0.0138, "step": 8180 }, { "epoch": 1.05, "learning_rate": 1.3282943026913668e-05, "loss": 0.015, "step": 8190 }, { "epoch": 1.05, "learning_rate": 1.3274204823488293e-05, "loss": 0.0153, "step": 8200 }, { "epoch": 1.05, "learning_rate": 1.3265466620062918e-05, "loss": 0.0144, "step": 8210 }, { "epoch": 1.05, "learning_rate": 1.325672841663754e-05, "loss": 0.0126, "step": 8220 }, { "epoch": 1.06, "learning_rate": 1.3247990213212166e-05, "loss": 0.0136, "step": 8230 }, { "epoch": 1.06, "learning_rate": 1.3239252009786789e-05, "loss": 0.0126, "step": 8240 }, { "epoch": 1.06, "learning_rate": 1.3230513806361414e-05, "loss": 0.012, "step": 8250 }, { "epoch": 1.06, "learning_rate": 1.3221775602936037e-05, "loss": 0.0122, "step": 8260 }, { "epoch": 1.06, "learning_rate": 1.3213037399510662e-05, "loss": 0.0101, "step": 8270 }, { "epoch": 1.06, "learning_rate": 1.3204299196085287e-05, "loss": 0.0132, "step": 8280 }, { "epoch": 1.06, "learning_rate": 1.319556099265991e-05, "loss": 0.0156, "step": 8290 }, { "epoch": 1.06, "learning_rate": 1.3186822789234535e-05, "loss": 0.0169, "step": 8300 }, { "epoch": 1.07, "learning_rate": 1.3178084585809158e-05, "loss": 0.015, "step": 8310 }, { "epoch": 1.07, "learning_rate": 1.3169346382383783e-05, "loss": 0.0157, "step": 8320 }, { "epoch": 1.07, "learning_rate": 1.3160608178958408e-05, "loss": 0.016, "step": 8330 }, { "epoch": 1.07, "learning_rate": 1.3151869975533031e-05, "loss": 0.017, "step": 8340 }, { "epoch": 1.07, "learning_rate": 1.3143131772107656e-05, "loss": 0.0117, "step": 8350 }, { "epoch": 1.07, "learning_rate": 1.313439356868228e-05, "loss": 0.0106, "step": 8360 }, { "epoch": 1.07, "learning_rate": 1.3125655365256904e-05, "loss": 0.0114, "step": 8370 }, { "epoch": 1.07, "learning_rate": 1.3116917161831528e-05, "loss": 0.0128, "step": 8380 }, { "epoch": 1.08, "learning_rate": 1.3108178958406153e-05, "loss": 0.0137, "step": 8390 }, { "epoch": 1.08, "learning_rate": 1.3099440754980777e-05, "loss": 0.0116, "step": 8400 }, { "epoch": 1.08, "learning_rate": 1.30907025515554e-05, "loss": 0.0159, "step": 8410 }, { "epoch": 1.08, "learning_rate": 1.3081964348130026e-05, "loss": 0.0123, "step": 8420 }, { "epoch": 1.08, "learning_rate": 1.3073226144704649e-05, "loss": 0.0161, "step": 8430 }, { "epoch": 1.08, "learning_rate": 1.3064487941279274e-05, "loss": 0.0154, "step": 8440 }, { "epoch": 1.08, "learning_rate": 1.3055749737853897e-05, "loss": 0.0126, "step": 8450 }, { "epoch": 1.09, "learning_rate": 1.3047011534428522e-05, "loss": 0.0149, "step": 8460 }, { "epoch": 1.09, "learning_rate": 1.3038273331003147e-05, "loss": 0.0093, "step": 8470 }, { "epoch": 1.09, "learning_rate": 1.302953512757777e-05, "loss": 0.0147, "step": 8480 }, { "epoch": 1.09, "learning_rate": 1.3020796924152395e-05, "loss": 0.0164, "step": 8490 }, { "epoch": 1.09, "learning_rate": 1.3012058720727018e-05, "loss": 0.0116, "step": 8500 }, { "epoch": 1.09, "learning_rate": 1.3003320517301643e-05, "loss": 0.0175, "step": 8510 }, { "epoch": 1.09, "learning_rate": 1.2994582313876268e-05, "loss": 0.0149, "step": 8520 }, { "epoch": 1.09, "learning_rate": 1.2985844110450891e-05, "loss": 0.0148, "step": 8530 }, { "epoch": 1.1, "learning_rate": 1.2977105907025516e-05, "loss": 0.0173, "step": 8540 }, { "epoch": 1.1, "learning_rate": 1.296836770360014e-05, "loss": 0.016, "step": 8550 }, { "epoch": 1.1, "learning_rate": 1.2959629500174764e-05, "loss": 0.0201, "step": 8560 }, { "epoch": 1.1, "learning_rate": 1.2950891296749388e-05, "loss": 0.0196, "step": 8570 }, { "epoch": 1.1, "learning_rate": 1.2942153093324014e-05, "loss": 0.0142, "step": 8580 }, { "epoch": 1.1, "learning_rate": 1.2933414889898639e-05, "loss": 0.0124, "step": 8590 }, { "epoch": 1.1, "learning_rate": 1.2924676686473262e-05, "loss": 0.0167, "step": 8600 }, { "epoch": 1.1, "learning_rate": 1.2915938483047887e-05, "loss": 0.0139, "step": 8610 }, { "epoch": 1.11, "learning_rate": 1.2907200279622512e-05, "loss": 0.0125, "step": 8620 }, { "epoch": 1.11, "learning_rate": 1.2898462076197135e-05, "loss": 0.0175, "step": 8630 }, { "epoch": 1.11, "learning_rate": 1.288972387277176e-05, "loss": 0.0181, "step": 8640 }, { "epoch": 1.11, "learning_rate": 1.2880985669346384e-05, "loss": 0.01, "step": 8650 }, { "epoch": 1.11, "learning_rate": 1.2872247465921008e-05, "loss": 0.0115, "step": 8660 }, { "epoch": 1.11, "learning_rate": 1.2863509262495633e-05, "loss": 0.0118, "step": 8670 }, { "epoch": 1.11, "learning_rate": 1.2854771059070257e-05, "loss": 0.0137, "step": 8680 }, { "epoch": 1.11, "learning_rate": 1.2846032855644882e-05, "loss": 0.0121, "step": 8690 }, { "epoch": 1.12, "learning_rate": 1.2837294652219505e-05, "loss": 0.0151, "step": 8700 }, { "epoch": 1.12, "learning_rate": 1.282855644879413e-05, "loss": 0.0187, "step": 8710 }, { "epoch": 1.12, "learning_rate": 1.2819818245368753e-05, "loss": 0.0106, "step": 8720 }, { "epoch": 1.12, "learning_rate": 1.2811080041943378e-05, "loss": 0.0119, "step": 8730 }, { "epoch": 1.12, "learning_rate": 1.2802341838518003e-05, "loss": 0.0134, "step": 8740 }, { "epoch": 1.12, "learning_rate": 1.2793603635092626e-05, "loss": 0.0145, "step": 8750 }, { "epoch": 1.12, "learning_rate": 1.2784865431667251e-05, "loss": 0.0115, "step": 8760 }, { "epoch": 1.12, "learning_rate": 1.2776127228241874e-05, "loss": 0.0155, "step": 8770 }, { "epoch": 1.13, "learning_rate": 1.2767389024816499e-05, "loss": 0.0123, "step": 8780 }, { "epoch": 1.13, "learning_rate": 1.2758650821391122e-05, "loss": 0.0121, "step": 8790 }, { "epoch": 1.13, "learning_rate": 1.2749912617965747e-05, "loss": 0.0198, "step": 8800 }, { "epoch": 1.13, "learning_rate": 1.2741174414540372e-05, "loss": 0.0177, "step": 8810 }, { "epoch": 1.13, "learning_rate": 1.2732436211114995e-05, "loss": 0.0122, "step": 8820 }, { "epoch": 1.13, "learning_rate": 1.272369800768962e-05, "loss": 0.0139, "step": 8830 }, { "epoch": 1.13, "learning_rate": 1.2714959804264243e-05, "loss": 0.0177, "step": 8840 }, { "epoch": 1.14, "learning_rate": 1.2706221600838868e-05, "loss": 0.01, "step": 8850 }, { "epoch": 1.14, "learning_rate": 1.2697483397413493e-05, "loss": 0.0114, "step": 8860 }, { "epoch": 1.14, "learning_rate": 1.2688745193988117e-05, "loss": 0.0122, "step": 8870 }, { "epoch": 1.14, "learning_rate": 1.2680006990562741e-05, "loss": 0.0195, "step": 8880 }, { "epoch": 1.14, "learning_rate": 1.2671268787137365e-05, "loss": 0.0176, "step": 8890 }, { "epoch": 1.14, "learning_rate": 1.266253058371199e-05, "loss": 0.0126, "step": 8900 }, { "epoch": 1.14, "learning_rate": 1.2653792380286613e-05, "loss": 0.0172, "step": 8910 }, { "epoch": 1.14, "learning_rate": 1.2645054176861238e-05, "loss": 0.0118, "step": 8920 }, { "epoch": 1.15, "learning_rate": 1.2636315973435863e-05, "loss": 0.0133, "step": 8930 }, { "epoch": 1.15, "learning_rate": 1.2627577770010486e-05, "loss": 0.0126, "step": 8940 }, { "epoch": 1.15, "learning_rate": 1.261883956658511e-05, "loss": 0.0204, "step": 8950 }, { "epoch": 1.15, "learning_rate": 1.2610101363159734e-05, "loss": 0.0128, "step": 8960 }, { "epoch": 1.15, "learning_rate": 1.2601363159734359e-05, "loss": 0.0139, "step": 8970 }, { "epoch": 1.15, "learning_rate": 1.2592624956308984e-05, "loss": 0.0124, "step": 8980 }, { "epoch": 1.15, "learning_rate": 1.2583886752883607e-05, "loss": 0.0163, "step": 8990 }, { "epoch": 1.15, "learning_rate": 1.2575148549458232e-05, "loss": 0.0135, "step": 9000 }, { "epoch": 1.16, "learning_rate": 1.2566410346032855e-05, "loss": 0.0135, "step": 9010 }, { "epoch": 1.16, "learning_rate": 1.255767214260748e-05, "loss": 0.0135, "step": 9020 }, { "epoch": 1.16, "learning_rate": 1.2548933939182103e-05, "loss": 0.0099, "step": 9030 }, { "epoch": 1.16, "learning_rate": 1.254019573575673e-05, "loss": 0.0145, "step": 9040 }, { "epoch": 1.16, "learning_rate": 1.2531457532331355e-05, "loss": 0.0118, "step": 9050 }, { "epoch": 1.16, "learning_rate": 1.2522719328905978e-05, "loss": 0.0122, "step": 9060 }, { "epoch": 1.16, "learning_rate": 1.2513981125480603e-05, "loss": 0.0138, "step": 9070 }, { "epoch": 1.16, "learning_rate": 1.2505242922055228e-05, "loss": 0.012, "step": 9080 }, { "epoch": 1.17, "learning_rate": 1.2496504718629851e-05, "loss": 0.0116, "step": 9090 }, { "epoch": 1.17, "learning_rate": 1.2487766515204476e-05, "loss": 0.0147, "step": 9100 }, { "epoch": 1.17, "learning_rate": 1.24790283117791e-05, "loss": 0.0089, "step": 9110 }, { "epoch": 1.17, "learning_rate": 1.2470290108353724e-05, "loss": 0.0095, "step": 9120 }, { "epoch": 1.17, "learning_rate": 1.246155190492835e-05, "loss": 0.0114, "step": 9130 }, { "epoch": 1.17, "learning_rate": 1.2452813701502972e-05, "loss": 0.0165, "step": 9140 }, { "epoch": 1.17, "learning_rate": 1.2444075498077597e-05, "loss": 0.016, "step": 9150 }, { "epoch": 1.17, "learning_rate": 1.243533729465222e-05, "loss": 0.0189, "step": 9160 }, { "epoch": 1.18, "learning_rate": 1.2426599091226846e-05, "loss": 0.013, "step": 9170 }, { "epoch": 1.18, "learning_rate": 1.2417860887801469e-05, "loss": 0.0145, "step": 9180 }, { "epoch": 1.18, "learning_rate": 1.2409122684376094e-05, "loss": 0.014, "step": 9190 }, { "epoch": 1.18, "learning_rate": 1.2400384480950719e-05, "loss": 0.0196, "step": 9200 }, { "epoch": 1.18, "learning_rate": 1.2391646277525342e-05, "loss": 0.0143, "step": 9210 }, { "epoch": 1.18, "learning_rate": 1.2382908074099967e-05, "loss": 0.0127, "step": 9220 }, { "epoch": 1.18, "learning_rate": 1.237416987067459e-05, "loss": 0.0128, "step": 9230 }, { "epoch": 1.19, "learning_rate": 1.2365431667249215e-05, "loss": 0.0112, "step": 9240 }, { "epoch": 1.19, "learning_rate": 1.2356693463823838e-05, "loss": 0.0149, "step": 9250 }, { "epoch": 1.19, "learning_rate": 1.2347955260398463e-05, "loss": 0.0122, "step": 9260 }, { "epoch": 1.19, "learning_rate": 1.2339217056973088e-05, "loss": 0.012, "step": 9270 }, { "epoch": 1.19, "learning_rate": 1.2330478853547711e-05, "loss": 0.016, "step": 9280 }, { "epoch": 1.19, "learning_rate": 1.2321740650122336e-05, "loss": 0.0128, "step": 9290 }, { "epoch": 1.19, "learning_rate": 1.231300244669696e-05, "loss": 0.017, "step": 9300 }, { "epoch": 1.19, "learning_rate": 1.2304264243271584e-05, "loss": 0.0085, "step": 9310 }, { "epoch": 1.2, "learning_rate": 1.229552603984621e-05, "loss": 0.0128, "step": 9320 }, { "epoch": 1.2, "learning_rate": 1.2286787836420832e-05, "loss": 0.0171, "step": 9330 }, { "epoch": 1.2, "learning_rate": 1.2278049632995457e-05, "loss": 0.0137, "step": 9340 }, { "epoch": 1.2, "learning_rate": 1.226931142957008e-05, "loss": 0.0101, "step": 9350 }, { "epoch": 1.2, "learning_rate": 1.2260573226144705e-05, "loss": 0.0135, "step": 9360 }, { "epoch": 1.2, "learning_rate": 1.2251835022719329e-05, "loss": 0.0138, "step": 9370 }, { "epoch": 1.2, "learning_rate": 1.2243096819293954e-05, "loss": 0.0129, "step": 9380 }, { "epoch": 1.2, "learning_rate": 1.2234358615868579e-05, "loss": 0.017, "step": 9390 }, { "epoch": 1.21, "learning_rate": 1.2225620412443202e-05, "loss": 0.0112, "step": 9400 }, { "epoch": 1.21, "learning_rate": 1.2216882209017827e-05, "loss": 0.0121, "step": 9410 }, { "epoch": 1.21, "learning_rate": 1.220814400559245e-05, "loss": 0.0161, "step": 9420 }, { "epoch": 1.21, "learning_rate": 1.2199405802167075e-05, "loss": 0.0115, "step": 9430 }, { "epoch": 1.21, "learning_rate": 1.21906675987417e-05, "loss": 0.0152, "step": 9440 }, { "epoch": 1.21, "learning_rate": 1.2181929395316323e-05, "loss": 0.0153, "step": 9450 }, { "epoch": 1.21, "learning_rate": 1.2173191191890948e-05, "loss": 0.0132, "step": 9460 }, { "epoch": 1.21, "learning_rate": 1.2164452988465571e-05, "loss": 0.0113, "step": 9470 }, { "epoch": 1.22, "learning_rate": 1.2155714785040196e-05, "loss": 0.0143, "step": 9480 }, { "epoch": 1.22, "learning_rate": 1.214697658161482e-05, "loss": 0.0134, "step": 9490 }, { "epoch": 1.22, "learning_rate": 1.2138238378189444e-05, "loss": 0.0147, "step": 9500 }, { "epoch": 1.22, "learning_rate": 1.212950017476407e-05, "loss": 0.015, "step": 9510 }, { "epoch": 1.22, "learning_rate": 1.2120761971338694e-05, "loss": 0.0117, "step": 9520 }, { "epoch": 1.22, "learning_rate": 1.2112023767913319e-05, "loss": 0.0206, "step": 9530 }, { "epoch": 1.22, "learning_rate": 1.2103285564487944e-05, "loss": 0.0111, "step": 9540 }, { "epoch": 1.22, "learning_rate": 1.2094547361062567e-05, "loss": 0.0152, "step": 9550 }, { "epoch": 1.23, "learning_rate": 1.2085809157637192e-05, "loss": 0.012, "step": 9560 }, { "epoch": 1.23, "learning_rate": 1.2077070954211815e-05, "loss": 0.0177, "step": 9570 }, { "epoch": 1.23, "learning_rate": 1.206833275078644e-05, "loss": 0.0134, "step": 9580 }, { "epoch": 1.23, "learning_rate": 1.2059594547361065e-05, "loss": 0.0126, "step": 9590 }, { "epoch": 1.23, "learning_rate": 1.2050856343935688e-05, "loss": 0.0142, "step": 9600 }, { "epoch": 1.23, "learning_rate": 1.2042118140510313e-05, "loss": 0.0172, "step": 9610 }, { "epoch": 1.23, "learning_rate": 1.2033379937084936e-05, "loss": 0.0138, "step": 9620 }, { "epoch": 1.24, "learning_rate": 1.2024641733659561e-05, "loss": 0.016, "step": 9630 }, { "epoch": 1.24, "learning_rate": 1.2015903530234185e-05, "loss": 0.0123, "step": 9640 }, { "epoch": 1.24, "learning_rate": 1.200716532680881e-05, "loss": 0.0169, "step": 9650 }, { "epoch": 1.24, "learning_rate": 1.1998427123383434e-05, "loss": 0.0134, "step": 9660 }, { "epoch": 1.24, "learning_rate": 1.1989688919958058e-05, "loss": 0.0127, "step": 9670 }, { "epoch": 1.24, "learning_rate": 1.1980950716532683e-05, "loss": 0.0106, "step": 9680 }, { "epoch": 1.24, "learning_rate": 1.1972212513107306e-05, "loss": 0.0137, "step": 9690 }, { "epoch": 1.24, "learning_rate": 1.196347430968193e-05, "loss": 0.0127, "step": 9700 }, { "epoch": 1.25, "learning_rate": 1.1954736106256554e-05, "loss": 0.0156, "step": 9710 }, { "epoch": 1.25, "learning_rate": 1.1945997902831179e-05, "loss": 0.0123, "step": 9720 }, { "epoch": 1.25, "learning_rate": 1.1937259699405804e-05, "loss": 0.0141, "step": 9730 }, { "epoch": 1.25, "learning_rate": 1.1928521495980427e-05, "loss": 0.0152, "step": 9740 }, { "epoch": 1.25, "learning_rate": 1.1919783292555052e-05, "loss": 0.013, "step": 9750 }, { "epoch": 1.25, "learning_rate": 1.1911045089129675e-05, "loss": 0.0137, "step": 9760 }, { "epoch": 1.25, "learning_rate": 1.19023068857043e-05, "loss": 0.0128, "step": 9770 }, { "epoch": 1.25, "learning_rate": 1.1893568682278925e-05, "loss": 0.0136, "step": 9780 }, { "epoch": 1.26, "learning_rate": 1.1884830478853548e-05, "loss": 0.0152, "step": 9790 }, { "epoch": 1.26, "learning_rate": 1.1876092275428173e-05, "loss": 0.0139, "step": 9800 }, { "epoch": 1.26, "learning_rate": 1.1867354072002796e-05, "loss": 0.014, "step": 9810 }, { "epoch": 1.26, "learning_rate": 1.1858615868577421e-05, "loss": 0.0088, "step": 9820 }, { "epoch": 1.26, "learning_rate": 1.1849877665152045e-05, "loss": 0.0093, "step": 9830 }, { "epoch": 1.26, "learning_rate": 1.184113946172667e-05, "loss": 0.0195, "step": 9840 }, { "epoch": 1.26, "learning_rate": 1.1832401258301294e-05, "loss": 0.0155, "step": 9850 }, { "epoch": 1.26, "learning_rate": 1.1823663054875918e-05, "loss": 0.0139, "step": 9860 }, { "epoch": 1.27, "learning_rate": 1.1814924851450543e-05, "loss": 0.0128, "step": 9870 }, { "epoch": 1.27, "learning_rate": 1.1806186648025166e-05, "loss": 0.015, "step": 9880 }, { "epoch": 1.27, "learning_rate": 1.179744844459979e-05, "loss": 0.0152, "step": 9890 }, { "epoch": 1.27, "learning_rate": 1.1788710241174416e-05, "loss": 0.0122, "step": 9900 }, { "epoch": 1.27, "learning_rate": 1.1779972037749039e-05, "loss": 0.0204, "step": 9910 }, { "epoch": 1.27, "learning_rate": 1.1771233834323664e-05, "loss": 0.0143, "step": 9920 }, { "epoch": 1.27, "learning_rate": 1.1762495630898287e-05, "loss": 0.0135, "step": 9930 }, { "epoch": 1.27, "learning_rate": 1.1753757427472912e-05, "loss": 0.0115, "step": 9940 }, { "epoch": 1.28, "learning_rate": 1.1745019224047535e-05, "loss": 0.0169, "step": 9950 }, { "epoch": 1.28, "learning_rate": 1.173628102062216e-05, "loss": 0.0146, "step": 9960 }, { "epoch": 1.28, "learning_rate": 1.1727542817196785e-05, "loss": 0.0159, "step": 9970 }, { "epoch": 1.28, "learning_rate": 1.171880461377141e-05, "loss": 0.0131, "step": 9980 }, { "epoch": 1.28, "learning_rate": 1.1710066410346035e-05, "loss": 0.0162, "step": 9990 }, { "epoch": 1.28, "learning_rate": 1.170132820692066e-05, "loss": 0.0143, "step": 10000 }, { "epoch": 1.28, "learning_rate": 1.1692590003495283e-05, "loss": 0.016, "step": 10010 }, { "epoch": 1.29, "learning_rate": 1.1683851800069908e-05, "loss": 0.0146, "step": 10020 }, { "epoch": 1.29, "learning_rate": 1.1675113596644531e-05, "loss": 0.0148, "step": 10030 }, { "epoch": 1.29, "learning_rate": 1.1666375393219156e-05, "loss": 0.0131, "step": 10040 }, { "epoch": 1.29, "learning_rate": 1.1657637189793781e-05, "loss": 0.0155, "step": 10050 }, { "epoch": 1.29, "learning_rate": 1.1648898986368404e-05, "loss": 0.0156, "step": 10060 }, { "epoch": 1.29, "learning_rate": 1.1640160782943029e-05, "loss": 0.0116, "step": 10070 }, { "epoch": 1.29, "learning_rate": 1.1631422579517652e-05, "loss": 0.0131, "step": 10080 }, { "epoch": 1.29, "learning_rate": 1.1622684376092277e-05, "loss": 0.0136, "step": 10090 }, { "epoch": 1.3, "learning_rate": 1.16139461726669e-05, "loss": 0.0102, "step": 10100 }, { "epoch": 1.3, "learning_rate": 1.1605207969241525e-05, "loss": 0.0113, "step": 10110 }, { "epoch": 1.3, "learning_rate": 1.159646976581615e-05, "loss": 0.0148, "step": 10120 }, { "epoch": 1.3, "learning_rate": 1.1587731562390774e-05, "loss": 0.013, "step": 10130 }, { "epoch": 1.3, "learning_rate": 1.1578993358965398e-05, "loss": 0.0123, "step": 10140 }, { "epoch": 1.3, "learning_rate": 1.1570255155540022e-05, "loss": 0.015, "step": 10150 }, { "epoch": 1.3, "learning_rate": 1.1561516952114647e-05, "loss": 0.0102, "step": 10160 }, { "epoch": 1.3, "learning_rate": 1.155277874868927e-05, "loss": 0.0132, "step": 10170 }, { "epoch": 1.31, "learning_rate": 1.1544040545263895e-05, "loss": 0.0147, "step": 10180 }, { "epoch": 1.31, "learning_rate": 1.153530234183852e-05, "loss": 0.0159, "step": 10190 }, { "epoch": 1.31, "learning_rate": 1.1526564138413143e-05, "loss": 0.0134, "step": 10200 }, { "epoch": 1.31, "learning_rate": 1.1517825934987768e-05, "loss": 0.0134, "step": 10210 }, { "epoch": 1.31, "learning_rate": 1.1509087731562391e-05, "loss": 0.012, "step": 10220 }, { "epoch": 1.31, "learning_rate": 1.1500349528137016e-05, "loss": 0.018, "step": 10230 }, { "epoch": 1.31, "learning_rate": 1.1491611324711641e-05, "loss": 0.017, "step": 10240 }, { "epoch": 1.31, "learning_rate": 1.1482873121286264e-05, "loss": 0.0151, "step": 10250 }, { "epoch": 1.32, "learning_rate": 1.1474134917860889e-05, "loss": 0.0145, "step": 10260 }, { "epoch": 1.32, "learning_rate": 1.1465396714435512e-05, "loss": 0.008, "step": 10270 }, { "epoch": 1.32, "learning_rate": 1.1456658511010137e-05, "loss": 0.012, "step": 10280 }, { "epoch": 1.32, "learning_rate": 1.144792030758476e-05, "loss": 0.0138, "step": 10290 }, { "epoch": 1.32, "learning_rate": 1.1439182104159385e-05, "loss": 0.0145, "step": 10300 }, { "epoch": 1.32, "learning_rate": 1.143044390073401e-05, "loss": 0.0134, "step": 10310 }, { "epoch": 1.32, "learning_rate": 1.1421705697308633e-05, "loss": 0.0097, "step": 10320 }, { "epoch": 1.32, "learning_rate": 1.1412967493883258e-05, "loss": 0.0101, "step": 10330 }, { "epoch": 1.33, "learning_rate": 1.1404229290457882e-05, "loss": 0.0127, "step": 10340 }, { "epoch": 1.33, "learning_rate": 1.1395491087032507e-05, "loss": 0.0138, "step": 10350 }, { "epoch": 1.33, "learning_rate": 1.138675288360713e-05, "loss": 0.0104, "step": 10360 }, { "epoch": 1.33, "learning_rate": 1.1378014680181755e-05, "loss": 0.0122, "step": 10370 }, { "epoch": 1.33, "learning_rate": 1.136927647675638e-05, "loss": 0.0175, "step": 10380 }, { "epoch": 1.33, "learning_rate": 1.1360538273331003e-05, "loss": 0.0173, "step": 10390 }, { "epoch": 1.33, "learning_rate": 1.1351800069905628e-05, "loss": 0.0135, "step": 10400 }, { "epoch": 1.34, "learning_rate": 1.1343061866480251e-05, "loss": 0.0151, "step": 10410 }, { "epoch": 1.34, "learning_rate": 1.1334323663054876e-05, "loss": 0.012, "step": 10420 }, { "epoch": 1.34, "learning_rate": 1.13255854596295e-05, "loss": 0.0137, "step": 10430 }, { "epoch": 1.34, "learning_rate": 1.1316847256204124e-05, "loss": 0.0142, "step": 10440 }, { "epoch": 1.34, "learning_rate": 1.130810905277875e-05, "loss": 0.0161, "step": 10450 }, { "epoch": 1.34, "learning_rate": 1.1299370849353376e-05, "loss": 0.0139, "step": 10460 }, { "epoch": 1.34, "learning_rate": 1.1290632645927999e-05, "loss": 0.0121, "step": 10470 }, { "epoch": 1.34, "learning_rate": 1.1281894442502624e-05, "loss": 0.0146, "step": 10480 }, { "epoch": 1.35, "learning_rate": 1.1273156239077247e-05, "loss": 0.0116, "step": 10490 }, { "epoch": 1.35, "learning_rate": 1.1264418035651872e-05, "loss": 0.0168, "step": 10500 }, { "epoch": 1.35, "learning_rate": 1.1255679832226497e-05, "loss": 0.0155, "step": 10510 }, { "epoch": 1.35, "learning_rate": 1.124694162880112e-05, "loss": 0.0159, "step": 10520 }, { "epoch": 1.35, "learning_rate": 1.1238203425375745e-05, "loss": 0.0134, "step": 10530 }, { "epoch": 1.35, "learning_rate": 1.1229465221950368e-05, "loss": 0.0132, "step": 10540 }, { "epoch": 1.35, "learning_rate": 1.1220727018524993e-05, "loss": 0.0166, "step": 10550 }, { "epoch": 1.35, "learning_rate": 1.1211988815099616e-05, "loss": 0.0122, "step": 10560 }, { "epoch": 1.36, "learning_rate": 1.1203250611674241e-05, "loss": 0.017, "step": 10570 }, { "epoch": 1.36, "learning_rate": 1.1194512408248866e-05, "loss": 0.0137, "step": 10580 }, { "epoch": 1.36, "learning_rate": 1.118577420482349e-05, "loss": 0.008, "step": 10590 }, { "epoch": 1.36, "learning_rate": 1.1177036001398114e-05, "loss": 0.0157, "step": 10600 }, { "epoch": 1.36, "learning_rate": 1.1168297797972738e-05, "loss": 0.0123, "step": 10610 }, { "epoch": 1.36, "learning_rate": 1.1159559594547362e-05, "loss": 0.0125, "step": 10620 }, { "epoch": 1.36, "learning_rate": 1.1150821391121986e-05, "loss": 0.0145, "step": 10630 }, { "epoch": 1.36, "learning_rate": 1.114208318769661e-05, "loss": 0.0141, "step": 10640 }, { "epoch": 1.37, "learning_rate": 1.1133344984271236e-05, "loss": 0.0116, "step": 10650 }, { "epoch": 1.37, "learning_rate": 1.1124606780845859e-05, "loss": 0.0147, "step": 10660 }, { "epoch": 1.37, "learning_rate": 1.1115868577420484e-05, "loss": 0.0139, "step": 10670 }, { "epoch": 1.37, "learning_rate": 1.1107130373995107e-05, "loss": 0.016, "step": 10680 }, { "epoch": 1.37, "learning_rate": 1.1098392170569732e-05, "loss": 0.0168, "step": 10690 }, { "epoch": 1.37, "learning_rate": 1.1089653967144357e-05, "loss": 0.0152, "step": 10700 }, { "epoch": 1.37, "learning_rate": 1.108091576371898e-05, "loss": 0.0134, "step": 10710 }, { "epoch": 1.38, "learning_rate": 1.1072177560293605e-05, "loss": 0.0153, "step": 10720 }, { "epoch": 1.38, "learning_rate": 1.1063439356868228e-05, "loss": 0.012, "step": 10730 }, { "epoch": 1.38, "learning_rate": 1.1054701153442853e-05, "loss": 0.0153, "step": 10740 }, { "epoch": 1.38, "learning_rate": 1.1045962950017476e-05, "loss": 0.0134, "step": 10750 }, { "epoch": 1.38, "learning_rate": 1.1037224746592101e-05, "loss": 0.0124, "step": 10760 }, { "epoch": 1.38, "learning_rate": 1.1028486543166726e-05, "loss": 0.013, "step": 10770 }, { "epoch": 1.38, "learning_rate": 1.101974833974135e-05, "loss": 0.0165, "step": 10780 }, { "epoch": 1.38, "learning_rate": 1.1011010136315974e-05, "loss": 0.0171, "step": 10790 }, { "epoch": 1.39, "learning_rate": 1.1002271932890597e-05, "loss": 0.016, "step": 10800 }, { "epoch": 1.39, "learning_rate": 1.0993533729465222e-05, "loss": 0.0118, "step": 10810 }, { "epoch": 1.39, "learning_rate": 1.0984795526039846e-05, "loss": 0.0206, "step": 10820 }, { "epoch": 1.39, "learning_rate": 1.097605732261447e-05, "loss": 0.0144, "step": 10830 }, { "epoch": 1.39, "learning_rate": 1.0967319119189095e-05, "loss": 0.0128, "step": 10840 }, { "epoch": 1.39, "learning_rate": 1.0958580915763719e-05, "loss": 0.0151, "step": 10850 }, { "epoch": 1.39, "learning_rate": 1.0949842712338344e-05, "loss": 0.0136, "step": 10860 }, { "epoch": 1.39, "learning_rate": 1.0941104508912967e-05, "loss": 0.0181, "step": 10870 }, { "epoch": 1.4, "learning_rate": 1.0932366305487592e-05, "loss": 0.0122, "step": 10880 }, { "epoch": 1.4, "learning_rate": 1.0923628102062217e-05, "loss": 0.016, "step": 10890 }, { "epoch": 1.4, "learning_rate": 1.091488989863684e-05, "loss": 0.019, "step": 10900 }, { "epoch": 1.4, "learning_rate": 1.0906151695211465e-05, "loss": 0.0163, "step": 10910 }, { "epoch": 1.4, "learning_rate": 1.0897413491786091e-05, "loss": 0.0155, "step": 10920 }, { "epoch": 1.4, "learning_rate": 1.0888675288360715e-05, "loss": 0.0137, "step": 10930 }, { "epoch": 1.4, "learning_rate": 1.087993708493534e-05, "loss": 0.0181, "step": 10940 }, { "epoch": 1.4, "learning_rate": 1.0871198881509963e-05, "loss": 0.0127, "step": 10950 }, { "epoch": 1.41, "learning_rate": 1.0862460678084588e-05, "loss": 0.0108, "step": 10960 }, { "epoch": 1.41, "learning_rate": 1.0853722474659213e-05, "loss": 0.0157, "step": 10970 }, { "epoch": 1.41, "learning_rate": 1.0844984271233836e-05, "loss": 0.0135, "step": 10980 }, { "epoch": 1.41, "learning_rate": 1.083624606780846e-05, "loss": 0.0151, "step": 10990 }, { "epoch": 1.41, "learning_rate": 1.0827507864383084e-05, "loss": 0.0135, "step": 11000 }, { "epoch": 1.41, "learning_rate": 1.0818769660957709e-05, "loss": 0.0161, "step": 11010 }, { "epoch": 1.41, "learning_rate": 1.0810031457532332e-05, "loss": 0.0194, "step": 11020 }, { "epoch": 1.41, "learning_rate": 1.0801293254106957e-05, "loss": 0.0159, "step": 11030 }, { "epoch": 1.42, "learning_rate": 1.0792555050681582e-05, "loss": 0.0172, "step": 11040 }, { "epoch": 1.42, "learning_rate": 1.0783816847256205e-05, "loss": 0.0104, "step": 11050 }, { "epoch": 1.42, "learning_rate": 1.077507864383083e-05, "loss": 0.0125, "step": 11060 }, { "epoch": 1.42, "learning_rate": 1.0766340440405453e-05, "loss": 0.0135, "step": 11070 }, { "epoch": 1.42, "learning_rate": 1.0757602236980078e-05, "loss": 0.0109, "step": 11080 }, { "epoch": 1.42, "learning_rate": 1.0748864033554702e-05, "loss": 0.0113, "step": 11090 }, { "epoch": 1.42, "learning_rate": 1.0740125830129326e-05, "loss": 0.0154, "step": 11100 }, { "epoch": 1.43, "learning_rate": 1.0731387626703951e-05, "loss": 0.0126, "step": 11110 }, { "epoch": 1.43, "learning_rate": 1.0722649423278575e-05, "loss": 0.0155, "step": 11120 }, { "epoch": 1.43, "learning_rate": 1.07139112198532e-05, "loss": 0.0117, "step": 11130 }, { "epoch": 1.43, "learning_rate": 1.0705173016427823e-05, "loss": 0.0184, "step": 11140 }, { "epoch": 1.43, "learning_rate": 1.0696434813002448e-05, "loss": 0.0134, "step": 11150 }, { "epoch": 1.43, "learning_rate": 1.0687696609577073e-05, "loss": 0.0107, "step": 11160 }, { "epoch": 1.43, "learning_rate": 1.0678958406151696e-05, "loss": 0.0194, "step": 11170 }, { "epoch": 1.43, "learning_rate": 1.067022020272632e-05, "loss": 0.0186, "step": 11180 }, { "epoch": 1.44, "learning_rate": 1.0661481999300944e-05, "loss": 0.0127, "step": 11190 }, { "epoch": 1.44, "learning_rate": 1.0652743795875569e-05, "loss": 0.0101, "step": 11200 }, { "epoch": 1.44, "learning_rate": 1.0644005592450192e-05, "loss": 0.0133, "step": 11210 }, { "epoch": 1.44, "learning_rate": 1.0635267389024817e-05, "loss": 0.0169, "step": 11220 }, { "epoch": 1.44, "learning_rate": 1.0626529185599442e-05, "loss": 0.0138, "step": 11230 }, { "epoch": 1.44, "learning_rate": 1.0617790982174065e-05, "loss": 0.0139, "step": 11240 }, { "epoch": 1.44, "learning_rate": 1.060905277874869e-05, "loss": 0.0154, "step": 11250 }, { "epoch": 1.44, "learning_rate": 1.0600314575323313e-05, "loss": 0.0132, "step": 11260 }, { "epoch": 1.45, "learning_rate": 1.0591576371897938e-05, "loss": 0.0148, "step": 11270 }, { "epoch": 1.45, "learning_rate": 1.0582838168472561e-05, "loss": 0.0138, "step": 11280 }, { "epoch": 1.45, "learning_rate": 1.0574099965047186e-05, "loss": 0.0143, "step": 11290 }, { "epoch": 1.45, "learning_rate": 1.0565361761621811e-05, "loss": 0.0154, "step": 11300 }, { "epoch": 1.45, "learning_rate": 1.0556623558196435e-05, "loss": 0.0151, "step": 11310 }, { "epoch": 1.45, "learning_rate": 1.054788535477106e-05, "loss": 0.0141, "step": 11320 }, { "epoch": 1.45, "learning_rate": 1.0539147151345683e-05, "loss": 0.0117, "step": 11330 }, { "epoch": 1.45, "learning_rate": 1.0530408947920308e-05, "loss": 0.0142, "step": 11340 }, { "epoch": 1.46, "learning_rate": 1.0521670744494933e-05, "loss": 0.0126, "step": 11350 }, { "epoch": 1.46, "learning_rate": 1.0512932541069556e-05, "loss": 0.0129, "step": 11360 }, { "epoch": 1.46, "learning_rate": 1.050419433764418e-05, "loss": 0.0155, "step": 11370 }, { "epoch": 1.46, "learning_rate": 1.0495456134218804e-05, "loss": 0.0125, "step": 11380 }, { "epoch": 1.46, "learning_rate": 1.048671793079343e-05, "loss": 0.0131, "step": 11390 }, { "epoch": 1.46, "learning_rate": 1.0477979727368055e-05, "loss": 0.0142, "step": 11400 }, { "epoch": 1.46, "learning_rate": 1.0469241523942679e-05, "loss": 0.0173, "step": 11410 }, { "epoch": 1.46, "learning_rate": 1.0460503320517304e-05, "loss": 0.0174, "step": 11420 }, { "epoch": 1.47, "learning_rate": 1.0451765117091927e-05, "loss": 0.017, "step": 11430 }, { "epoch": 1.47, "learning_rate": 1.0443026913666552e-05, "loss": 0.013, "step": 11440 }, { "epoch": 1.47, "learning_rate": 1.0434288710241177e-05, "loss": 0.0141, "step": 11450 }, { "epoch": 1.47, "learning_rate": 1.04255505068158e-05, "loss": 0.013, "step": 11460 }, { "epoch": 1.47, "learning_rate": 1.0416812303390425e-05, "loss": 0.0173, "step": 11470 }, { "epoch": 1.47, "learning_rate": 1.0408074099965048e-05, "loss": 0.0096, "step": 11480 }, { "epoch": 1.47, "learning_rate": 1.0399335896539673e-05, "loss": 0.0078, "step": 11490 }, { "epoch": 1.48, "learning_rate": 1.0390597693114298e-05, "loss": 0.0102, "step": 11500 }, { "epoch": 1.48, "learning_rate": 1.0381859489688921e-05, "loss": 0.0104, "step": 11510 }, { "epoch": 1.48, "learning_rate": 1.0373121286263546e-05, "loss": 0.014, "step": 11520 }, { "epoch": 1.48, "learning_rate": 1.036438308283817e-05, "loss": 0.0177, "step": 11530 }, { "epoch": 1.48, "learning_rate": 1.0355644879412794e-05, "loss": 0.0109, "step": 11540 }, { "epoch": 1.48, "learning_rate": 1.0346906675987417e-05, "loss": 0.0145, "step": 11550 }, { "epoch": 1.48, "learning_rate": 1.0338168472562042e-05, "loss": 0.0139, "step": 11560 }, { "epoch": 1.48, "learning_rate": 1.0329430269136667e-05, "loss": 0.0169, "step": 11570 }, { "epoch": 1.49, "learning_rate": 1.032069206571129e-05, "loss": 0.0154, "step": 11580 }, { "epoch": 1.49, "learning_rate": 1.0311953862285915e-05, "loss": 0.0128, "step": 11590 }, { "epoch": 1.49, "learning_rate": 1.0303215658860539e-05, "loss": 0.0114, "step": 11600 }, { "epoch": 1.49, "learning_rate": 1.0294477455435164e-05, "loss": 0.012, "step": 11610 }, { "epoch": 1.49, "learning_rate": 1.0285739252009788e-05, "loss": 0.0147, "step": 11620 }, { "epoch": 1.49, "learning_rate": 1.0277001048584412e-05, "loss": 0.0156, "step": 11630 }, { "epoch": 1.49, "learning_rate": 1.0268262845159037e-05, "loss": 0.0144, "step": 11640 }, { "epoch": 1.49, "learning_rate": 1.025952464173366e-05, "loss": 0.0158, "step": 11650 }, { "epoch": 1.5, "learning_rate": 1.0250786438308285e-05, "loss": 0.0151, "step": 11660 }, { "epoch": 1.5, "learning_rate": 1.0242048234882908e-05, "loss": 0.016, "step": 11670 }, { "epoch": 1.5, "learning_rate": 1.0233310031457533e-05, "loss": 0.0145, "step": 11680 }, { "epoch": 1.5, "learning_rate": 1.0224571828032158e-05, "loss": 0.018, "step": 11690 }, { "epoch": 1.5, "learning_rate": 1.0215833624606781e-05, "loss": 0.0112, "step": 11700 }, { "epoch": 1.5, "learning_rate": 1.0207095421181406e-05, "loss": 0.0125, "step": 11710 }, { "epoch": 1.5, "learning_rate": 1.019835721775603e-05, "loss": 0.0122, "step": 11720 }, { "epoch": 1.5, "learning_rate": 1.0189619014330654e-05, "loss": 0.0157, "step": 11730 }, { "epoch": 1.51, "learning_rate": 1.0180880810905277e-05, "loss": 0.0146, "step": 11740 }, { "epoch": 1.51, "learning_rate": 1.0172142607479902e-05, "loss": 0.0123, "step": 11750 }, { "epoch": 1.51, "learning_rate": 1.0163404404054527e-05, "loss": 0.0125, "step": 11760 }, { "epoch": 1.51, "learning_rate": 1.015466620062915e-05, "loss": 0.015, "step": 11770 }, { "epoch": 1.51, "learning_rate": 1.0145927997203775e-05, "loss": 0.0164, "step": 11780 }, { "epoch": 1.51, "learning_rate": 1.0137189793778399e-05, "loss": 0.014, "step": 11790 }, { "epoch": 1.51, "learning_rate": 1.0128451590353023e-05, "loss": 0.0176, "step": 11800 }, { "epoch": 1.51, "learning_rate": 1.0119713386927648e-05, "loss": 0.0133, "step": 11810 }, { "epoch": 1.52, "learning_rate": 1.0110975183502272e-05, "loss": 0.014, "step": 11820 }, { "epoch": 1.52, "learning_rate": 1.0102236980076897e-05, "loss": 0.0132, "step": 11830 }, { "epoch": 1.52, "learning_rate": 1.009349877665152e-05, "loss": 0.0163, "step": 11840 }, { "epoch": 1.52, "learning_rate": 1.0084760573226145e-05, "loss": 0.0133, "step": 11850 }, { "epoch": 1.52, "learning_rate": 1.0076022369800771e-05, "loss": 0.0168, "step": 11860 }, { "epoch": 1.52, "learning_rate": 1.0067284166375395e-05, "loss": 0.0139, "step": 11870 }, { "epoch": 1.52, "learning_rate": 1.005854596295002e-05, "loss": 0.0163, "step": 11880 }, { "epoch": 1.53, "learning_rate": 1.0049807759524643e-05, "loss": 0.0143, "step": 11890 }, { "epoch": 1.53, "learning_rate": 1.0041069556099268e-05, "loss": 0.0118, "step": 11900 }, { "epoch": 1.53, "learning_rate": 1.0032331352673893e-05, "loss": 0.0131, "step": 11910 }, { "epoch": 1.53, "learning_rate": 1.0023593149248516e-05, "loss": 0.0145, "step": 11920 }, { "epoch": 1.53, "learning_rate": 1.001485494582314e-05, "loss": 0.0135, "step": 11930 }, { "epoch": 1.53, "learning_rate": 1.0006116742397764e-05, "loss": 0.0135, "step": 11940 }, { "epoch": 1.53, "learning_rate": 9.997378538972387e-06, "loss": 0.0146, "step": 11950 }, { "epoch": 1.53, "learning_rate": 9.988640335547012e-06, "loss": 0.0135, "step": 11960 }, { "epoch": 1.54, "learning_rate": 9.979902132121635e-06, "loss": 0.0136, "step": 11970 }, { "epoch": 1.54, "learning_rate": 9.971163928696262e-06, "loss": 0.0128, "step": 11980 }, { "epoch": 1.54, "learning_rate": 9.962425725270885e-06, "loss": 0.0134, "step": 11990 }, { "epoch": 1.54, "learning_rate": 9.95368752184551e-06, "loss": 0.0118, "step": 12000 }, { "epoch": 1.54, "learning_rate": 9.944949318420133e-06, "loss": 0.0131, "step": 12010 }, { "epoch": 1.54, "learning_rate": 9.936211114994758e-06, "loss": 0.0129, "step": 12020 }, { "epoch": 1.54, "learning_rate": 9.927472911569383e-06, "loss": 0.0113, "step": 12030 }, { "epoch": 1.54, "learning_rate": 9.918734708144006e-06, "loss": 0.0156, "step": 12040 }, { "epoch": 1.55, "learning_rate": 9.909996504718631e-06, "loss": 0.0166, "step": 12050 }, { "epoch": 1.55, "learning_rate": 9.901258301293254e-06, "loss": 0.0119, "step": 12060 }, { "epoch": 1.55, "learning_rate": 9.89252009786788e-06, "loss": 0.0142, "step": 12070 }, { "epoch": 1.55, "learning_rate": 9.883781894442504e-06, "loss": 0.0158, "step": 12080 }, { "epoch": 1.55, "learning_rate": 9.875043691017128e-06, "loss": 0.01, "step": 12090 }, { "epoch": 1.55, "learning_rate": 9.866305487591752e-06, "loss": 0.0184, "step": 12100 }, { "epoch": 1.55, "learning_rate": 9.857567284166376e-06, "loss": 0.0153, "step": 12110 }, { "epoch": 1.55, "learning_rate": 9.848829080741e-06, "loss": 0.0168, "step": 12120 }, { "epoch": 1.56, "learning_rate": 9.840090877315624e-06, "loss": 0.013, "step": 12130 }, { "epoch": 1.56, "learning_rate": 9.831352673890249e-06, "loss": 0.0125, "step": 12140 }, { "epoch": 1.56, "learning_rate": 9.822614470464874e-06, "loss": 0.0182, "step": 12150 }, { "epoch": 1.56, "learning_rate": 9.813876267039497e-06, "loss": 0.0134, "step": 12160 }, { "epoch": 1.56, "learning_rate": 9.805138063614122e-06, "loss": 0.0159, "step": 12170 }, { "epoch": 1.56, "learning_rate": 9.796399860188745e-06, "loss": 0.0145, "step": 12180 }, { "epoch": 1.56, "learning_rate": 9.78766165676337e-06, "loss": 0.0154, "step": 12190 }, { "epoch": 1.56, "learning_rate": 9.778923453337993e-06, "loss": 0.0134, "step": 12200 }, { "epoch": 1.57, "learning_rate": 9.77018524991262e-06, "loss": 0.0144, "step": 12210 }, { "epoch": 1.57, "learning_rate": 9.761447046487243e-06, "loss": 0.0155, "step": 12220 }, { "epoch": 1.57, "learning_rate": 9.752708843061868e-06, "loss": 0.0164, "step": 12230 }, { "epoch": 1.57, "learning_rate": 9.743970639636491e-06, "loss": 0.0153, "step": 12240 }, { "epoch": 1.57, "learning_rate": 9.735232436211116e-06, "loss": 0.0073, "step": 12250 }, { "epoch": 1.57, "learning_rate": 9.726494232785741e-06, "loss": 0.0126, "step": 12260 }, { "epoch": 1.57, "learning_rate": 9.717756029360364e-06, "loss": 0.0111, "step": 12270 }, { "epoch": 1.58, "learning_rate": 9.70901782593499e-06, "loss": 0.0115, "step": 12280 }, { "epoch": 1.58, "learning_rate": 9.700279622509612e-06, "loss": 0.0124, "step": 12290 }, { "epoch": 1.58, "learning_rate": 9.691541419084237e-06, "loss": 0.0148, "step": 12300 }, { "epoch": 1.58, "learning_rate": 9.682803215658862e-06, "loss": 0.0167, "step": 12310 }, { "epoch": 1.58, "learning_rate": 9.674065012233485e-06, "loss": 0.0151, "step": 12320 }, { "epoch": 1.58, "learning_rate": 9.66532680880811e-06, "loss": 0.0137, "step": 12330 }, { "epoch": 1.58, "learning_rate": 9.656588605382734e-06, "loss": 0.014, "step": 12340 }, { "epoch": 1.58, "learning_rate": 9.647850401957359e-06, "loss": 0.0172, "step": 12350 }, { "epoch": 1.59, "learning_rate": 9.639112198531982e-06, "loss": 0.013, "step": 12360 }, { "epoch": 1.59, "learning_rate": 9.630373995106607e-06, "loss": 0.0121, "step": 12370 }, { "epoch": 1.59, "learning_rate": 9.621635791681232e-06, "loss": 0.0116, "step": 12380 }, { "epoch": 1.59, "learning_rate": 9.612897588255855e-06, "loss": 0.0135, "step": 12390 }, { "epoch": 1.59, "learning_rate": 9.60415938483048e-06, "loss": 0.0178, "step": 12400 }, { "epoch": 1.59, "learning_rate": 9.595421181405103e-06, "loss": 0.0137, "step": 12410 }, { "epoch": 1.59, "learning_rate": 9.586682977979728e-06, "loss": 0.0145, "step": 12420 }, { "epoch": 1.59, "learning_rate": 9.577944774554351e-06, "loss": 0.0151, "step": 12430 }, { "epoch": 1.6, "learning_rate": 9.569206571128978e-06, "loss": 0.0124, "step": 12440 }, { "epoch": 1.6, "learning_rate": 9.560468367703601e-06, "loss": 0.012, "step": 12450 }, { "epoch": 1.6, "learning_rate": 9.551730164278226e-06, "loss": 0.0151, "step": 12460 }, { "epoch": 1.6, "learning_rate": 9.542991960852849e-06, "loss": 0.0187, "step": 12470 }, { "epoch": 1.6, "learning_rate": 9.534253757427474e-06, "loss": 0.011, "step": 12480 }, { "epoch": 1.6, "learning_rate": 9.525515554002099e-06, "loss": 0.018, "step": 12490 }, { "epoch": 1.6, "learning_rate": 9.516777350576722e-06, "loss": 0.0127, "step": 12500 }, { "epoch": 1.6, "learning_rate": 9.508039147151347e-06, "loss": 0.0116, "step": 12510 }, { "epoch": 1.61, "learning_rate": 9.49930094372597e-06, "loss": 0.0092, "step": 12520 }, { "epoch": 1.61, "learning_rate": 9.490562740300595e-06, "loss": 0.0147, "step": 12530 }, { "epoch": 1.61, "learning_rate": 9.48182453687522e-06, "loss": 0.0167, "step": 12540 }, { "epoch": 1.61, "learning_rate": 9.473086333449843e-06, "loss": 0.0103, "step": 12550 }, { "epoch": 1.61, "learning_rate": 9.464348130024468e-06, "loss": 0.0137, "step": 12560 }, { "epoch": 1.61, "learning_rate": 9.455609926599092e-06, "loss": 0.0124, "step": 12570 }, { "epoch": 1.61, "learning_rate": 9.446871723173716e-06, "loss": 0.0165, "step": 12580 }, { "epoch": 1.61, "learning_rate": 9.43813351974834e-06, "loss": 0.013, "step": 12590 }, { "epoch": 1.62, "learning_rate": 9.429395316322965e-06, "loss": 0.0146, "step": 12600 }, { "epoch": 1.62, "learning_rate": 9.42065711289759e-06, "loss": 0.0127, "step": 12610 }, { "epoch": 1.62, "learning_rate": 9.411918909472213e-06, "loss": 0.015, "step": 12620 }, { "epoch": 1.62, "learning_rate": 9.403180706046838e-06, "loss": 0.0122, "step": 12630 }, { "epoch": 1.62, "learning_rate": 9.394442502621461e-06, "loss": 0.0126, "step": 12640 }, { "epoch": 1.62, "learning_rate": 9.385704299196086e-06, "loss": 0.0134, "step": 12650 }, { "epoch": 1.62, "learning_rate": 9.376966095770709e-06, "loss": 0.0125, "step": 12660 }, { "epoch": 1.63, "learning_rate": 9.368227892345334e-06, "loss": 0.0134, "step": 12670 }, { "epoch": 1.63, "learning_rate": 9.359489688919959e-06, "loss": 0.0103, "step": 12680 }, { "epoch": 1.63, "learning_rate": 9.350751485494584e-06, "loss": 0.0113, "step": 12690 }, { "epoch": 1.63, "learning_rate": 9.342013282069207e-06, "loss": 0.0157, "step": 12700 }, { "epoch": 1.63, "learning_rate": 9.333275078643832e-06, "loss": 0.0139, "step": 12710 }, { "epoch": 1.63, "learning_rate": 9.324536875218457e-06, "loss": 0.0119, "step": 12720 }, { "epoch": 1.63, "learning_rate": 9.31579867179308e-06, "loss": 0.0168, "step": 12730 }, { "epoch": 1.63, "learning_rate": 9.307060468367705e-06, "loss": 0.0143, "step": 12740 }, { "epoch": 1.64, "learning_rate": 9.298322264942328e-06, "loss": 0.0115, "step": 12750 }, { "epoch": 1.64, "learning_rate": 9.289584061516953e-06, "loss": 0.0155, "step": 12760 }, { "epoch": 1.64, "learning_rate": 9.280845858091576e-06, "loss": 0.014, "step": 12770 }, { "epoch": 1.64, "learning_rate": 9.272107654666201e-06, "loss": 0.013, "step": 12780 }, { "epoch": 1.64, "learning_rate": 9.263369451240826e-06, "loss": 0.0147, "step": 12790 }, { "epoch": 1.64, "learning_rate": 9.25463124781545e-06, "loss": 0.0133, "step": 12800 }, { "epoch": 1.64, "learning_rate": 9.245893044390074e-06, "loss": 0.0156, "step": 12810 }, { "epoch": 1.64, "learning_rate": 9.237154840964698e-06, "loss": 0.0131, "step": 12820 }, { "epoch": 1.65, "learning_rate": 9.228416637539323e-06, "loss": 0.0137, "step": 12830 }, { "epoch": 1.65, "learning_rate": 9.219678434113947e-06, "loss": 0.0146, "step": 12840 }, { "epoch": 1.65, "learning_rate": 9.21094023068857e-06, "loss": 0.0106, "step": 12850 }, { "epoch": 1.65, "learning_rate": 9.202202027263196e-06, "loss": 0.0139, "step": 12860 }, { "epoch": 1.65, "learning_rate": 9.193463823837819e-06, "loss": 0.0131, "step": 12870 }, { "epoch": 1.65, "learning_rate": 9.184725620412444e-06, "loss": 0.015, "step": 12880 }, { "epoch": 1.65, "learning_rate": 9.175987416987067e-06, "loss": 0.0166, "step": 12890 }, { "epoch": 1.65, "learning_rate": 9.167249213561692e-06, "loss": 0.0151, "step": 12900 }, { "epoch": 1.66, "learning_rate": 9.158511010136317e-06, "loss": 0.0127, "step": 12910 }, { "epoch": 1.66, "learning_rate": 9.149772806710942e-06, "loss": 0.0127, "step": 12920 }, { "epoch": 1.66, "learning_rate": 9.141034603285565e-06, "loss": 0.0162, "step": 12930 }, { "epoch": 1.66, "learning_rate": 9.13229639986019e-06, "loss": 0.0174, "step": 12940 }, { "epoch": 1.66, "learning_rate": 9.123558196434815e-06, "loss": 0.0093, "step": 12950 }, { "epoch": 1.66, "learning_rate": 9.114819993009438e-06, "loss": 0.0106, "step": 12960 }, { "epoch": 1.66, "learning_rate": 9.106081789584063e-06, "loss": 0.013, "step": 12970 }, { "epoch": 1.66, "learning_rate": 9.097343586158686e-06, "loss": 0.0088, "step": 12980 }, { "epoch": 1.67, "learning_rate": 9.088605382733311e-06, "loss": 0.0129, "step": 12990 }, { "epoch": 1.67, "learning_rate": 9.079867179307934e-06, "loss": 0.0123, "step": 13000 }, { "epoch": 1.67, "learning_rate": 9.07112897588256e-06, "loss": 0.0139, "step": 13010 }, { "epoch": 1.67, "learning_rate": 9.062390772457184e-06, "loss": 0.0091, "step": 13020 }, { "epoch": 1.67, "learning_rate": 9.053652569031807e-06, "loss": 0.0123, "step": 13030 }, { "epoch": 1.67, "learning_rate": 9.044914365606432e-06, "loss": 0.0133, "step": 13040 }, { "epoch": 1.67, "learning_rate": 9.036176162181056e-06, "loss": 0.0152, "step": 13050 }, { "epoch": 1.68, "learning_rate": 9.02743795875568e-06, "loss": 0.0128, "step": 13060 }, { "epoch": 1.68, "learning_rate": 9.018699755330305e-06, "loss": 0.0095, "step": 13070 }, { "epoch": 1.68, "learning_rate": 9.009961551904929e-06, "loss": 0.0137, "step": 13080 }, { "epoch": 1.68, "learning_rate": 9.001223348479554e-06, "loss": 0.0114, "step": 13090 }, { "epoch": 1.68, "learning_rate": 8.992485145054177e-06, "loss": 0.0148, "step": 13100 }, { "epoch": 1.68, "learning_rate": 8.983746941628802e-06, "loss": 0.0131, "step": 13110 }, { "epoch": 1.68, "learning_rate": 8.975008738203425e-06, "loss": 0.0112, "step": 13120 }, { "epoch": 1.68, "learning_rate": 8.96627053477805e-06, "loss": 0.0135, "step": 13130 }, { "epoch": 1.69, "learning_rate": 8.957532331352675e-06, "loss": 0.0109, "step": 13140 }, { "epoch": 1.69, "learning_rate": 8.9487941279273e-06, "loss": 0.0154, "step": 13150 }, { "epoch": 1.69, "learning_rate": 8.940055924501923e-06, "loss": 0.0136, "step": 13160 }, { "epoch": 1.69, "learning_rate": 8.931317721076548e-06, "loss": 0.0141, "step": 13170 }, { "epoch": 1.69, "learning_rate": 8.922579517651173e-06, "loss": 0.0159, "step": 13180 }, { "epoch": 1.69, "learning_rate": 8.913841314225796e-06, "loss": 0.0149, "step": 13190 }, { "epoch": 1.69, "learning_rate": 8.905103110800421e-06, "loss": 0.0164, "step": 13200 }, { "epoch": 1.69, "learning_rate": 8.896364907375044e-06, "loss": 0.0159, "step": 13210 }, { "epoch": 1.7, "learning_rate": 8.887626703949669e-06, "loss": 0.0112, "step": 13220 }, { "epoch": 1.7, "learning_rate": 8.878888500524292e-06, "loss": 0.0161, "step": 13230 }, { "epoch": 1.7, "learning_rate": 8.870150297098917e-06, "loss": 0.0126, "step": 13240 }, { "epoch": 1.7, "learning_rate": 8.861412093673542e-06, "loss": 0.0127, "step": 13250 }, { "epoch": 1.7, "learning_rate": 8.852673890248165e-06, "loss": 0.0147, "step": 13260 }, { "epoch": 1.7, "learning_rate": 8.84393568682279e-06, "loss": 0.0122, "step": 13270 }, { "epoch": 1.7, "learning_rate": 8.835197483397413e-06, "loss": 0.0107, "step": 13280 }, { "epoch": 1.7, "learning_rate": 8.826459279972038e-06, "loss": 0.0169, "step": 13290 }, { "epoch": 1.71, "learning_rate": 8.817721076546663e-06, "loss": 0.0123, "step": 13300 }, { "epoch": 1.71, "learning_rate": 8.808982873121287e-06, "loss": 0.0111, "step": 13310 }, { "epoch": 1.71, "learning_rate": 8.800244669695911e-06, "loss": 0.0148, "step": 13320 }, { "epoch": 1.71, "learning_rate": 8.791506466270535e-06, "loss": 0.0144, "step": 13330 }, { "epoch": 1.71, "learning_rate": 8.78276826284516e-06, "loss": 0.0172, "step": 13340 }, { "epoch": 1.71, "learning_rate": 8.774030059419783e-06, "loss": 0.0118, "step": 13350 }, { "epoch": 1.71, "learning_rate": 8.765291855994408e-06, "loss": 0.0114, "step": 13360 }, { "epoch": 1.71, "learning_rate": 8.756553652569033e-06, "loss": 0.0206, "step": 13370 }, { "epoch": 1.72, "learning_rate": 8.747815449143658e-06, "loss": 0.015, "step": 13380 }, { "epoch": 1.72, "learning_rate": 8.73907724571828e-06, "loss": 0.0148, "step": 13390 }, { "epoch": 1.72, "learning_rate": 8.730339042292906e-06, "loss": 0.0155, "step": 13400 }, { "epoch": 1.72, "learning_rate": 8.72160083886753e-06, "loss": 0.0105, "step": 13410 }, { "epoch": 1.72, "learning_rate": 8.712862635442154e-06, "loss": 0.0138, "step": 13420 }, { "epoch": 1.72, "learning_rate": 8.704124432016779e-06, "loss": 0.0162, "step": 13430 }, { "epoch": 1.72, "learning_rate": 8.695386228591402e-06, "loss": 0.0155, "step": 13440 }, { "epoch": 1.73, "learning_rate": 8.686648025166027e-06, "loss": 0.0161, "step": 13450 }, { "epoch": 1.73, "learning_rate": 8.67790982174065e-06, "loss": 0.0152, "step": 13460 }, { "epoch": 1.73, "learning_rate": 8.669171618315275e-06, "loss": 0.0147, "step": 13470 }, { "epoch": 1.73, "learning_rate": 8.6604334148899e-06, "loss": 0.0106, "step": 13480 }, { "epoch": 1.73, "learning_rate": 8.651695211464523e-06, "loss": 0.0145, "step": 13490 }, { "epoch": 1.73, "learning_rate": 8.642957008039148e-06, "loss": 0.0131, "step": 13500 }, { "epoch": 1.73, "learning_rate": 8.634218804613771e-06, "loss": 0.0144, "step": 13510 }, { "epoch": 1.73, "learning_rate": 8.625480601188396e-06, "loss": 0.0104, "step": 13520 }, { "epoch": 1.74, "learning_rate": 8.616742397763021e-06, "loss": 0.0139, "step": 13530 }, { "epoch": 1.74, "learning_rate": 8.608004194337644e-06, "loss": 0.0121, "step": 13540 }, { "epoch": 1.74, "learning_rate": 8.59926599091227e-06, "loss": 0.0137, "step": 13550 }, { "epoch": 1.74, "learning_rate": 8.590527787486893e-06, "loss": 0.0116, "step": 13560 }, { "epoch": 1.74, "learning_rate": 8.581789584061518e-06, "loss": 0.0128, "step": 13570 }, { "epoch": 1.74, "learning_rate": 8.57305138063614e-06, "loss": 0.0148, "step": 13580 }, { "epoch": 1.74, "learning_rate": 8.564313177210766e-06, "loss": 0.0153, "step": 13590 }, { "epoch": 1.74, "learning_rate": 8.55557497378539e-06, "loss": 0.0132, "step": 13600 }, { "epoch": 1.75, "learning_rate": 8.546836770360014e-06, "loss": 0.0126, "step": 13610 }, { "epoch": 1.75, "learning_rate": 8.538098566934639e-06, "loss": 0.0185, "step": 13620 }, { "epoch": 1.75, "learning_rate": 8.529360363509264e-06, "loss": 0.0145, "step": 13630 }, { "epoch": 1.75, "learning_rate": 8.520622160083889e-06, "loss": 0.0197, "step": 13640 }, { "epoch": 1.75, "learning_rate": 8.511883956658512e-06, "loss": 0.0164, "step": 13650 }, { "epoch": 1.75, "learning_rate": 8.503145753233137e-06, "loss": 0.0174, "step": 13660 }, { "epoch": 1.75, "learning_rate": 8.49440754980776e-06, "loss": 0.0141, "step": 13670 }, { "epoch": 1.75, "learning_rate": 8.485669346382385e-06, "loss": 0.0134, "step": 13680 }, { "epoch": 1.76, "learning_rate": 8.476931142957008e-06, "loss": 0.0149, "step": 13690 }, { "epoch": 1.76, "learning_rate": 8.468192939531633e-06, "loss": 0.0104, "step": 13700 }, { "epoch": 1.76, "learning_rate": 8.459454736106258e-06, "loss": 0.0146, "step": 13710 }, { "epoch": 1.76, "learning_rate": 8.450716532680881e-06, "loss": 0.0125, "step": 13720 }, { "epoch": 1.76, "learning_rate": 8.441978329255506e-06, "loss": 0.0134, "step": 13730 }, { "epoch": 1.76, "learning_rate": 8.43324012583013e-06, "loss": 0.0163, "step": 13740 }, { "epoch": 1.76, "learning_rate": 8.424501922404754e-06, "loss": 0.0167, "step": 13750 }, { "epoch": 1.76, "learning_rate": 8.41576371897938e-06, "loss": 0.0145, "step": 13760 }, { "epoch": 1.77, "learning_rate": 8.407025515554002e-06, "loss": 0.0123, "step": 13770 }, { "epoch": 1.77, "learning_rate": 8.398287312128627e-06, "loss": 0.0134, "step": 13780 }, { "epoch": 1.77, "learning_rate": 8.38954910870325e-06, "loss": 0.0133, "step": 13790 }, { "epoch": 1.77, "learning_rate": 8.380810905277875e-06, "loss": 0.0154, "step": 13800 }, { "epoch": 1.77, "learning_rate": 8.372072701852499e-06, "loss": 0.0131, "step": 13810 }, { "epoch": 1.77, "learning_rate": 8.363334498427124e-06, "loss": 0.0137, "step": 13820 }, { "epoch": 1.77, "learning_rate": 8.354596295001749e-06, "loss": 0.012, "step": 13830 }, { "epoch": 1.78, "learning_rate": 8.345858091576372e-06, "loss": 0.0125, "step": 13840 }, { "epoch": 1.78, "learning_rate": 8.337119888150997e-06, "loss": 0.0172, "step": 13850 }, { "epoch": 1.78, "learning_rate": 8.328381684725622e-06, "loss": 0.0089, "step": 13860 }, { "epoch": 1.78, "learning_rate": 8.319643481300247e-06, "loss": 0.0122, "step": 13870 }, { "epoch": 1.78, "learning_rate": 8.31090527787487e-06, "loss": 0.0172, "step": 13880 }, { "epoch": 1.78, "learning_rate": 8.302167074449495e-06, "loss": 0.0138, "step": 13890 }, { "epoch": 1.78, "learning_rate": 8.293428871024118e-06, "loss": 0.0129, "step": 13900 }, { "epoch": 1.78, "learning_rate": 8.284690667598743e-06, "loss": 0.0093, "step": 13910 }, { "epoch": 1.79, "learning_rate": 8.275952464173366e-06, "loss": 0.0122, "step": 13920 }, { "epoch": 1.79, "learning_rate": 8.267214260747991e-06, "loss": 0.0133, "step": 13930 }, { "epoch": 1.79, "learning_rate": 8.258476057322616e-06, "loss": 0.0163, "step": 13940 }, { "epoch": 1.79, "learning_rate": 8.249737853897239e-06, "loss": 0.0135, "step": 13950 }, { "epoch": 1.79, "learning_rate": 8.240999650471864e-06, "loss": 0.0132, "step": 13960 }, { "epoch": 1.79, "learning_rate": 8.232261447046487e-06, "loss": 0.0174, "step": 13970 }, { "epoch": 1.79, "learning_rate": 8.223523243621112e-06, "loss": 0.0115, "step": 13980 }, { "epoch": 1.79, "learning_rate": 8.214785040195737e-06, "loss": 0.0163, "step": 13990 }, { "epoch": 1.8, "learning_rate": 8.20604683677036e-06, "loss": 0.0138, "step": 14000 }, { "epoch": 1.8, "learning_rate": 8.197308633344985e-06, "loss": 0.0181, "step": 14010 }, { "epoch": 1.8, "learning_rate": 8.188570429919608e-06, "loss": 0.0177, "step": 14020 }, { "epoch": 1.8, "learning_rate": 8.179832226494233e-06, "loss": 0.0111, "step": 14030 }, { "epoch": 1.8, "learning_rate": 8.171094023068857e-06, "loss": 0.0113, "step": 14040 }, { "epoch": 1.8, "learning_rate": 8.162355819643482e-06, "loss": 0.0128, "step": 14050 }, { "epoch": 1.8, "learning_rate": 8.153617616218106e-06, "loss": 0.0135, "step": 14060 }, { "epoch": 1.8, "learning_rate": 8.14487941279273e-06, "loss": 0.0134, "step": 14070 }, { "epoch": 1.81, "learning_rate": 8.136141209367355e-06, "loss": 0.0114, "step": 14080 }, { "epoch": 1.81, "learning_rate": 8.12740300594198e-06, "loss": 0.0099, "step": 14090 }, { "epoch": 1.81, "learning_rate": 8.118664802516604e-06, "loss": 0.015, "step": 14100 }, { "epoch": 1.81, "learning_rate": 8.109926599091228e-06, "loss": 0.0163, "step": 14110 }, { "epoch": 1.81, "learning_rate": 8.101188395665853e-06, "loss": 0.013, "step": 14120 }, { "epoch": 1.81, "learning_rate": 8.092450192240476e-06, "loss": 0.0143, "step": 14130 }, { "epoch": 1.81, "learning_rate": 8.0837119888151e-06, "loss": 0.0148, "step": 14140 }, { "epoch": 1.81, "learning_rate": 8.074973785389724e-06, "loss": 0.0159, "step": 14150 }, { "epoch": 1.82, "learning_rate": 8.066235581964349e-06, "loss": 0.0136, "step": 14160 }, { "epoch": 1.82, "learning_rate": 8.057497378538974e-06, "loss": 0.0123, "step": 14170 }, { "epoch": 1.82, "learning_rate": 8.048759175113597e-06, "loss": 0.0111, "step": 14180 }, { "epoch": 1.82, "learning_rate": 8.040020971688222e-06, "loss": 0.0118, "step": 14190 }, { "epoch": 1.82, "learning_rate": 8.031282768262845e-06, "loss": 0.0121, "step": 14200 }, { "epoch": 1.82, "learning_rate": 8.02254456483747e-06, "loss": 0.0127, "step": 14210 }, { "epoch": 1.82, "learning_rate": 8.013806361412095e-06, "loss": 0.0127, "step": 14220 }, { "epoch": 1.83, "learning_rate": 8.005068157986718e-06, "loss": 0.0113, "step": 14230 }, { "epoch": 1.83, "learning_rate": 7.996329954561343e-06, "loss": 0.0104, "step": 14240 }, { "epoch": 1.83, "learning_rate": 7.987591751135966e-06, "loss": 0.012, "step": 14250 }, { "epoch": 1.83, "learning_rate": 7.978853547710591e-06, "loss": 0.0185, "step": 14260 }, { "epoch": 1.83, "learning_rate": 7.970115344285215e-06, "loss": 0.0139, "step": 14270 }, { "epoch": 1.83, "learning_rate": 7.96137714085984e-06, "loss": 0.0127, "step": 14280 }, { "epoch": 1.83, "learning_rate": 7.952638937434464e-06, "loss": 0.0119, "step": 14290 }, { "epoch": 1.83, "learning_rate": 7.943900734009088e-06, "loss": 0.0132, "step": 14300 }, { "epoch": 1.84, "learning_rate": 7.935162530583713e-06, "loss": 0.0121, "step": 14310 }, { "epoch": 1.84, "learning_rate": 7.926424327158337e-06, "loss": 0.0142, "step": 14320 }, { "epoch": 1.84, "learning_rate": 7.917686123732962e-06, "loss": 0.0129, "step": 14330 }, { "epoch": 1.84, "learning_rate": 7.908947920307586e-06, "loss": 0.0156, "step": 14340 }, { "epoch": 1.84, "learning_rate": 7.90020971688221e-06, "loss": 0.0102, "step": 14350 }, { "epoch": 1.84, "learning_rate": 7.891471513456834e-06, "loss": 0.012, "step": 14360 }, { "epoch": 1.84, "learning_rate": 7.882733310031459e-06, "loss": 0.0125, "step": 14370 }, { "epoch": 1.84, "learning_rate": 7.873995106606082e-06, "loss": 0.0114, "step": 14380 }, { "epoch": 1.85, "learning_rate": 7.865256903180707e-06, "loss": 0.0128, "step": 14390 }, { "epoch": 1.85, "learning_rate": 7.856518699755332e-06, "loss": 0.0151, "step": 14400 }, { "epoch": 1.85, "learning_rate": 7.847780496329955e-06, "loss": 0.0144, "step": 14410 }, { "epoch": 1.85, "learning_rate": 7.83904229290458e-06, "loss": 0.0132, "step": 14420 }, { "epoch": 1.85, "learning_rate": 7.830304089479203e-06, "loss": 0.0138, "step": 14430 }, { "epoch": 1.85, "learning_rate": 7.821565886053828e-06, "loss": 0.0108, "step": 14440 }, { "epoch": 1.85, "learning_rate": 7.812827682628453e-06, "loss": 0.016, "step": 14450 }, { "epoch": 1.85, "learning_rate": 7.804089479203076e-06, "loss": 0.0081, "step": 14460 }, { "epoch": 1.86, "learning_rate": 7.795351275777701e-06, "loss": 0.0143, "step": 14470 }, { "epoch": 1.86, "learning_rate": 7.786613072352324e-06, "loss": 0.0178, "step": 14480 }, { "epoch": 1.86, "learning_rate": 7.77787486892695e-06, "loss": 0.0136, "step": 14490 }, { "epoch": 1.86, "learning_rate": 7.769136665501572e-06, "loss": 0.0129, "step": 14500 }, { "epoch": 1.86, "learning_rate": 7.760398462076197e-06, "loss": 0.015, "step": 14510 }, { "epoch": 1.86, "learning_rate": 7.751660258650822e-06, "loss": 0.0113, "step": 14520 }, { "epoch": 1.86, "learning_rate": 7.742922055225446e-06, "loss": 0.013, "step": 14530 }, { "epoch": 1.86, "learning_rate": 7.73418385180007e-06, "loss": 0.0162, "step": 14540 }, { "epoch": 1.87, "learning_rate": 7.725445648374694e-06, "loss": 0.0135, "step": 14550 }, { "epoch": 1.87, "learning_rate": 7.71670744494932e-06, "loss": 0.015, "step": 14560 }, { "epoch": 1.87, "learning_rate": 7.707969241523944e-06, "loss": 0.0162, "step": 14570 }, { "epoch": 1.87, "learning_rate": 7.699231038098568e-06, "loss": 0.0146, "step": 14580 }, { "epoch": 1.87, "learning_rate": 7.690492834673192e-06, "loss": 0.0089, "step": 14590 }, { "epoch": 1.87, "learning_rate": 7.681754631247817e-06, "loss": 0.0143, "step": 14600 }, { "epoch": 1.87, "learning_rate": 7.67301642782244e-06, "loss": 0.0167, "step": 14610 }, { "epoch": 1.88, "learning_rate": 7.664278224397065e-06, "loss": 0.013, "step": 14620 }, { "epoch": 1.88, "learning_rate": 7.65554002097169e-06, "loss": 0.0123, "step": 14630 }, { "epoch": 1.88, "learning_rate": 7.646801817546313e-06, "loss": 0.0174, "step": 14640 }, { "epoch": 1.88, "learning_rate": 7.638063614120938e-06, "loss": 0.0098, "step": 14650 }, { "epoch": 1.88, "learning_rate": 7.629325410695562e-06, "loss": 0.0132, "step": 14660 }, { "epoch": 1.88, "learning_rate": 7.620587207270186e-06, "loss": 0.011, "step": 14670 }, { "epoch": 1.88, "learning_rate": 7.61184900384481e-06, "loss": 0.0158, "step": 14680 }, { "epoch": 1.88, "learning_rate": 7.603110800419434e-06, "loss": 0.0145, "step": 14690 }, { "epoch": 1.89, "learning_rate": 7.594372596994058e-06, "loss": 0.0145, "step": 14700 }, { "epoch": 1.89, "learning_rate": 7.585634393568683e-06, "loss": 0.0114, "step": 14710 }, { "epoch": 1.89, "learning_rate": 7.576896190143307e-06, "loss": 0.0153, "step": 14720 }, { "epoch": 1.89, "learning_rate": 7.568157986717931e-06, "loss": 0.0131, "step": 14730 }, { "epoch": 1.89, "learning_rate": 7.559419783292555e-06, "loss": 0.015, "step": 14740 }, { "epoch": 1.89, "learning_rate": 7.550681579867179e-06, "loss": 0.0133, "step": 14750 }, { "epoch": 1.89, "learning_rate": 7.5419433764418035e-06, "loss": 0.0127, "step": 14760 }, { "epoch": 1.89, "learning_rate": 7.533205173016428e-06, "loss": 0.0135, "step": 14770 }, { "epoch": 1.9, "learning_rate": 7.5244669695910524e-06, "loss": 0.0123, "step": 14780 }, { "epoch": 1.9, "learning_rate": 7.515728766165677e-06, "loss": 0.0137, "step": 14790 }, { "epoch": 1.9, "learning_rate": 7.5069905627403014e-06, "loss": 0.0149, "step": 14800 }, { "epoch": 1.9, "learning_rate": 7.4982523593149255e-06, "loss": 0.0137, "step": 14810 }, { "epoch": 1.9, "learning_rate": 7.4895141558895504e-06, "loss": 0.0104, "step": 14820 }, { "epoch": 1.9, "learning_rate": 7.4807759524641745e-06, "loss": 0.0158, "step": 14830 }, { "epoch": 1.9, "learning_rate": 7.472037749038799e-06, "loss": 0.0117, "step": 14840 }, { "epoch": 1.9, "learning_rate": 7.463299545613423e-06, "loss": 0.0138, "step": 14850 }, { "epoch": 1.91, "learning_rate": 7.454561342188047e-06, "loss": 0.0106, "step": 14860 }, { "epoch": 1.91, "learning_rate": 7.445823138762671e-06, "loss": 0.014, "step": 14870 }, { "epoch": 1.91, "learning_rate": 7.437084935337296e-06, "loss": 0.0139, "step": 14880 }, { "epoch": 1.91, "learning_rate": 7.42834673191192e-06, "loss": 0.0128, "step": 14890 }, { "epoch": 1.91, "learning_rate": 7.419608528486544e-06, "loss": 0.0131, "step": 14900 }, { "epoch": 1.91, "learning_rate": 7.410870325061168e-06, "loss": 0.0152, "step": 14910 }, { "epoch": 1.91, "learning_rate": 7.402132121635792e-06, "loss": 0.0181, "step": 14920 }, { "epoch": 1.92, "learning_rate": 7.393393918210416e-06, "loss": 0.018, "step": 14930 }, { "epoch": 1.92, "learning_rate": 7.384655714785041e-06, "loss": 0.0116, "step": 14940 }, { "epoch": 1.92, "learning_rate": 7.375917511359665e-06, "loss": 0.0145, "step": 14950 }, { "epoch": 1.92, "learning_rate": 7.367179307934289e-06, "loss": 0.0162, "step": 14960 }, { "epoch": 1.92, "learning_rate": 7.358441104508913e-06, "loss": 0.0107, "step": 14970 }, { "epoch": 1.92, "learning_rate": 7.349702901083537e-06, "loss": 0.0115, "step": 14980 }, { "epoch": 1.92, "learning_rate": 7.340964697658161e-06, "loss": 0.0127, "step": 14990 }, { "epoch": 1.92, "learning_rate": 7.3322264942327855e-06, "loss": 0.0094, "step": 15000 }, { "epoch": 1.93, "learning_rate": 7.32348829080741e-06, "loss": 0.0096, "step": 15010 }, { "epoch": 1.93, "learning_rate": 7.314750087382035e-06, "loss": 0.0175, "step": 15020 }, { "epoch": 1.93, "learning_rate": 7.306011883956659e-06, "loss": 0.0128, "step": 15030 }, { "epoch": 1.93, "learning_rate": 7.2972736805312834e-06, "loss": 0.0113, "step": 15040 }, { "epoch": 1.93, "learning_rate": 7.288535477105908e-06, "loss": 0.0144, "step": 15050 }, { "epoch": 1.93, "learning_rate": 7.2797972736805324e-06, "loss": 0.0158, "step": 15060 }, { "epoch": 1.93, "learning_rate": 7.2710590702551565e-06, "loss": 0.016, "step": 15070 }, { "epoch": 1.93, "learning_rate": 7.262320866829781e-06, "loss": 0.0117, "step": 15080 }, { "epoch": 1.94, "learning_rate": 7.253582663404405e-06, "loss": 0.0129, "step": 15090 }, { "epoch": 1.94, "learning_rate": 7.244844459979029e-06, "loss": 0.0105, "step": 15100 }, { "epoch": 1.94, "learning_rate": 7.236106256553654e-06, "loss": 0.0128, "step": 15110 }, { "epoch": 1.94, "learning_rate": 7.227368053128278e-06, "loss": 0.014, "step": 15120 }, { "epoch": 1.94, "learning_rate": 7.218629849702902e-06, "loss": 0.0138, "step": 15130 }, { "epoch": 1.94, "learning_rate": 7.209891646277526e-06, "loss": 0.0125, "step": 15140 }, { "epoch": 1.94, "learning_rate": 7.20115344285215e-06, "loss": 0.0136, "step": 15150 }, { "epoch": 1.94, "learning_rate": 7.192415239426774e-06, "loss": 0.0175, "step": 15160 }, { "epoch": 1.95, "learning_rate": 7.183677036001399e-06, "loss": 0.0153, "step": 15170 }, { "epoch": 1.95, "learning_rate": 7.174938832576023e-06, "loss": 0.0151, "step": 15180 }, { "epoch": 1.95, "learning_rate": 7.166200629150647e-06, "loss": 0.0134, "step": 15190 }, { "epoch": 1.95, "learning_rate": 7.157462425725271e-06, "loss": 0.0155, "step": 15200 }, { "epoch": 1.95, "learning_rate": 7.148724222299895e-06, "loss": 0.0151, "step": 15210 }, { "epoch": 1.95, "learning_rate": 7.139986018874519e-06, "loss": 0.0092, "step": 15220 }, { "epoch": 1.95, "learning_rate": 7.131247815449143e-06, "loss": 0.0129, "step": 15230 }, { "epoch": 1.95, "learning_rate": 7.122509612023768e-06, "loss": 0.013, "step": 15240 }, { "epoch": 1.96, "learning_rate": 7.113771408598392e-06, "loss": 0.0152, "step": 15250 }, { "epoch": 1.96, "learning_rate": 7.105033205173017e-06, "loss": 0.0123, "step": 15260 }, { "epoch": 1.96, "learning_rate": 7.096295001747641e-06, "loss": 0.0121, "step": 15270 }, { "epoch": 1.96, "learning_rate": 7.087556798322266e-06, "loss": 0.016, "step": 15280 }, { "epoch": 1.96, "learning_rate": 7.07881859489689e-06, "loss": 0.014, "step": 15290 }, { "epoch": 1.96, "learning_rate": 7.0700803914715144e-06, "loss": 0.0138, "step": 15300 }, { "epoch": 1.96, "learning_rate": 7.0613421880461385e-06, "loss": 0.012, "step": 15310 }, { "epoch": 1.97, "learning_rate": 7.052603984620763e-06, "loss": 0.0137, "step": 15320 }, { "epoch": 1.97, "learning_rate": 7.043865781195387e-06, "loss": 0.0114, "step": 15330 }, { "epoch": 1.97, "learning_rate": 7.035127577770012e-06, "loss": 0.0124, "step": 15340 }, { "epoch": 1.97, "learning_rate": 7.026389374344636e-06, "loss": 0.0093, "step": 15350 }, { "epoch": 1.97, "learning_rate": 7.01765117091926e-06, "loss": 0.0165, "step": 15360 }, { "epoch": 1.97, "learning_rate": 7.008912967493884e-06, "loss": 0.0149, "step": 15370 }, { "epoch": 1.97, "learning_rate": 7.000174764068508e-06, "loss": 0.0106, "step": 15380 }, { "epoch": 1.97, "learning_rate": 6.991436560643132e-06, "loss": 0.0101, "step": 15390 }, { "epoch": 1.98, "learning_rate": 6.982698357217757e-06, "loss": 0.012, "step": 15400 }, { "epoch": 1.98, "learning_rate": 6.973960153792381e-06, "loss": 0.0086, "step": 15410 }, { "epoch": 1.98, "learning_rate": 6.965221950367005e-06, "loss": 0.0161, "step": 15420 }, { "epoch": 1.98, "learning_rate": 6.956483746941629e-06, "loss": 0.0135, "step": 15430 }, { "epoch": 1.98, "learning_rate": 6.947745543516253e-06, "loss": 0.0094, "step": 15440 }, { "epoch": 1.98, "learning_rate": 6.939007340090877e-06, "loss": 0.0117, "step": 15450 }, { "epoch": 1.98, "learning_rate": 6.930269136665501e-06, "loss": 0.0151, "step": 15460 }, { "epoch": 1.98, "learning_rate": 6.921530933240126e-06, "loss": 0.0147, "step": 15470 }, { "epoch": 1.99, "learning_rate": 6.91279272981475e-06, "loss": 0.015, "step": 15480 }, { "epoch": 1.99, "learning_rate": 6.904054526389375e-06, "loss": 0.0135, "step": 15490 }, { "epoch": 1.99, "learning_rate": 6.895316322963999e-06, "loss": 0.0134, "step": 15500 }, { "epoch": 1.99, "learning_rate": 6.886578119538624e-06, "loss": 0.0131, "step": 15510 }, { "epoch": 1.99, "learning_rate": 6.877839916113248e-06, "loss": 0.0134, "step": 15520 }, { "epoch": 1.99, "learning_rate": 6.869101712687872e-06, "loss": 0.0111, "step": 15530 }, { "epoch": 1.99, "learning_rate": 6.8603635092624964e-06, "loss": 0.0155, "step": 15540 }, { "epoch": 1.99, "learning_rate": 6.8516253058371205e-06, "loss": 0.0108, "step": 15550 }, { "epoch": 2.0, "learning_rate": 6.842887102411745e-06, "loss": 0.0116, "step": 15560 }, { "epoch": 2.0, "learning_rate": 6.8341488989863695e-06, "loss": 0.0094, "step": 15570 }, { "epoch": 2.0, "learning_rate": 6.825410695560994e-06, "loss": 0.01, "step": 15580 }, { "epoch": 2.0, "learning_rate": 6.816672492135618e-06, "loss": 0.0115, "step": 15590 }, { "epoch": 2.0, "eval_loss": 0.013224462978541851, "eval_runtime": 56.1598, "eval_samples_per_second": 246.796, "eval_steps_per_second": 30.858, "step": 15592 }, { "epoch": 2.0, "learning_rate": 6.807934288710242e-06, "loss": 0.0142, "step": 15600 }, { "epoch": 2.0, "learning_rate": 6.799196085284866e-06, "loss": 0.0145, "step": 15610 }, { "epoch": 2.0, "learning_rate": 6.79045788185949e-06, "loss": 0.0143, "step": 15620 }, { "epoch": 2.0, "learning_rate": 6.781719678434115e-06, "loss": 0.0153, "step": 15630 }, { "epoch": 2.01, "learning_rate": 6.772981475008739e-06, "loss": 0.0156, "step": 15640 }, { "epoch": 2.01, "learning_rate": 6.764243271583363e-06, "loss": 0.0107, "step": 15650 }, { "epoch": 2.01, "learning_rate": 6.755505068157987e-06, "loss": 0.0147, "step": 15660 }, { "epoch": 2.01, "learning_rate": 6.746766864732611e-06, "loss": 0.0109, "step": 15670 }, { "epoch": 2.01, "learning_rate": 6.738028661307235e-06, "loss": 0.0147, "step": 15680 }, { "epoch": 2.01, "learning_rate": 6.729290457881859e-06, "loss": 0.0113, "step": 15690 }, { "epoch": 2.01, "learning_rate": 6.720552254456484e-06, "loss": 0.0112, "step": 15700 }, { "epoch": 2.02, "learning_rate": 6.711814051031108e-06, "loss": 0.0129, "step": 15710 }, { "epoch": 2.02, "learning_rate": 6.703075847605732e-06, "loss": 0.0102, "step": 15720 }, { "epoch": 2.02, "learning_rate": 6.694337644180357e-06, "loss": 0.0127, "step": 15730 }, { "epoch": 2.02, "learning_rate": 6.685599440754982e-06, "loss": 0.0109, "step": 15740 }, { "epoch": 2.02, "learning_rate": 6.676861237329606e-06, "loss": 0.0118, "step": 15750 }, { "epoch": 2.02, "learning_rate": 6.66812303390423e-06, "loss": 0.014, "step": 15760 }, { "epoch": 2.02, "learning_rate": 6.659384830478854e-06, "loss": 0.0118, "step": 15770 }, { "epoch": 2.02, "learning_rate": 6.6506466270534784e-06, "loss": 0.0118, "step": 15780 }, { "epoch": 2.03, "learning_rate": 6.6419084236281025e-06, "loss": 0.0145, "step": 15790 }, { "epoch": 2.03, "learning_rate": 6.6331702202027274e-06, "loss": 0.0126, "step": 15800 }, { "epoch": 2.03, "learning_rate": 6.6244320167773515e-06, "loss": 0.0127, "step": 15810 }, { "epoch": 2.03, "learning_rate": 6.615693813351976e-06, "loss": 0.0162, "step": 15820 }, { "epoch": 2.03, "learning_rate": 6.6069556099266e-06, "loss": 0.0102, "step": 15830 }, { "epoch": 2.03, "learning_rate": 6.598217406501224e-06, "loss": 0.0111, "step": 15840 }, { "epoch": 2.03, "learning_rate": 6.589479203075848e-06, "loss": 0.015, "step": 15850 }, { "epoch": 2.03, "learning_rate": 6.580740999650473e-06, "loss": 0.0146, "step": 15860 }, { "epoch": 2.04, "learning_rate": 6.572002796225097e-06, "loss": 0.0131, "step": 15870 }, { "epoch": 2.04, "learning_rate": 6.563264592799721e-06, "loss": 0.0148, "step": 15880 }, { "epoch": 2.04, "learning_rate": 6.554526389374345e-06, "loss": 0.0113, "step": 15890 }, { "epoch": 2.04, "learning_rate": 6.545788185948969e-06, "loss": 0.0156, "step": 15900 }, { "epoch": 2.04, "learning_rate": 6.537049982523593e-06, "loss": 0.01, "step": 15910 }, { "epoch": 2.04, "learning_rate": 6.528311779098217e-06, "loss": 0.009, "step": 15920 }, { "epoch": 2.04, "learning_rate": 6.519573575672842e-06, "loss": 0.0167, "step": 15930 }, { "epoch": 2.04, "learning_rate": 6.510835372247466e-06, "loss": 0.0121, "step": 15940 }, { "epoch": 2.05, "learning_rate": 6.50209716882209e-06, "loss": 0.0126, "step": 15950 }, { "epoch": 2.05, "learning_rate": 6.493358965396715e-06, "loss": 0.0141, "step": 15960 }, { "epoch": 2.05, "learning_rate": 6.48462076197134e-06, "loss": 0.0131, "step": 15970 }, { "epoch": 2.05, "learning_rate": 6.475882558545964e-06, "loss": 0.012, "step": 15980 }, { "epoch": 2.05, "learning_rate": 6.467144355120588e-06, "loss": 0.015, "step": 15990 }, { "epoch": 2.05, "learning_rate": 6.458406151695212e-06, "loss": 0.018, "step": 16000 }, { "epoch": 2.05, "learning_rate": 6.449667948269836e-06, "loss": 0.0117, "step": 16010 }, { "epoch": 2.05, "learning_rate": 6.4409297448444604e-06, "loss": 0.0148, "step": 16020 }, { "epoch": 2.06, "learning_rate": 6.432191541419085e-06, "loss": 0.0109, "step": 16030 }, { "epoch": 2.06, "learning_rate": 6.4234533379937094e-06, "loss": 0.0133, "step": 16040 }, { "epoch": 2.06, "learning_rate": 6.4147151345683335e-06, "loss": 0.0136, "step": 16050 }, { "epoch": 2.06, "learning_rate": 6.405976931142958e-06, "loss": 0.0162, "step": 16060 }, { "epoch": 2.06, "learning_rate": 6.397238727717582e-06, "loss": 0.0151, "step": 16070 }, { "epoch": 2.06, "learning_rate": 6.388500524292206e-06, "loss": 0.0117, "step": 16080 }, { "epoch": 2.06, "learning_rate": 6.379762320866831e-06, "loss": 0.0119, "step": 16090 }, { "epoch": 2.07, "learning_rate": 6.371024117441455e-06, "loss": 0.0113, "step": 16100 }, { "epoch": 2.07, "learning_rate": 6.362285914016079e-06, "loss": 0.012, "step": 16110 }, { "epoch": 2.07, "learning_rate": 6.353547710590703e-06, "loss": 0.0142, "step": 16120 }, { "epoch": 2.07, "learning_rate": 6.344809507165327e-06, "loss": 0.0115, "step": 16130 }, { "epoch": 2.07, "learning_rate": 6.336071303739951e-06, "loss": 0.0145, "step": 16140 }, { "epoch": 2.07, "learning_rate": 6.327333100314575e-06, "loss": 0.0141, "step": 16150 }, { "epoch": 2.07, "learning_rate": 6.3185948968892e-06, "loss": 0.014, "step": 16160 }, { "epoch": 2.07, "learning_rate": 6.309856693463824e-06, "loss": 0.0126, "step": 16170 }, { "epoch": 2.08, "learning_rate": 6.301118490038448e-06, "loss": 0.0132, "step": 16180 }, { "epoch": 2.08, "learning_rate": 6.292380286613072e-06, "loss": 0.0105, "step": 16190 }, { "epoch": 2.08, "learning_rate": 6.283642083187698e-06, "loss": 0.0168, "step": 16200 }, { "epoch": 2.08, "learning_rate": 6.274903879762322e-06, "loss": 0.0162, "step": 16210 }, { "epoch": 2.08, "learning_rate": 6.266165676336946e-06, "loss": 0.0165, "step": 16220 }, { "epoch": 2.08, "learning_rate": 6.25742747291157e-06, "loss": 0.0111, "step": 16230 }, { "epoch": 2.08, "learning_rate": 6.248689269486194e-06, "loss": 0.0119, "step": 16240 }, { "epoch": 2.08, "learning_rate": 6.239951066060818e-06, "loss": 0.0182, "step": 16250 }, { "epoch": 2.09, "learning_rate": 6.231212862635443e-06, "loss": 0.0091, "step": 16260 }, { "epoch": 2.09, "learning_rate": 6.222474659210067e-06, "loss": 0.0139, "step": 16270 }, { "epoch": 2.09, "learning_rate": 6.2137364557846914e-06, "loss": 0.0105, "step": 16280 }, { "epoch": 2.09, "learning_rate": 6.2049982523593155e-06, "loss": 0.0153, "step": 16290 }, { "epoch": 2.09, "learning_rate": 6.19626004893394e-06, "loss": 0.01, "step": 16300 }, { "epoch": 2.09, "learning_rate": 6.187521845508564e-06, "loss": 0.0126, "step": 16310 }, { "epoch": 2.09, "learning_rate": 6.178783642083188e-06, "loss": 0.0146, "step": 16320 }, { "epoch": 2.09, "learning_rate": 6.170045438657813e-06, "loss": 0.011, "step": 16330 }, { "epoch": 2.1, "learning_rate": 6.161307235232437e-06, "loss": 0.01, "step": 16340 }, { "epoch": 2.1, "learning_rate": 6.152569031807061e-06, "loss": 0.0159, "step": 16350 }, { "epoch": 2.1, "learning_rate": 6.143830828381685e-06, "loss": 0.0137, "step": 16360 }, { "epoch": 2.1, "learning_rate": 6.135092624956309e-06, "loss": 0.0148, "step": 16370 }, { "epoch": 2.1, "learning_rate": 6.126354421530933e-06, "loss": 0.0136, "step": 16380 }, { "epoch": 2.1, "learning_rate": 6.117616218105558e-06, "loss": 0.0126, "step": 16390 }, { "epoch": 2.1, "learning_rate": 6.108878014680182e-06, "loss": 0.0132, "step": 16400 }, { "epoch": 2.1, "learning_rate": 6.100139811254806e-06, "loss": 0.0133, "step": 16410 }, { "epoch": 2.11, "learning_rate": 6.09140160782943e-06, "loss": 0.0133, "step": 16420 }, { "epoch": 2.11, "learning_rate": 6.082663404404056e-06, "loss": 0.0119, "step": 16430 }, { "epoch": 2.11, "learning_rate": 6.07392520097868e-06, "loss": 0.0138, "step": 16440 }, { "epoch": 2.11, "learning_rate": 6.065186997553304e-06, "loss": 0.0109, "step": 16450 }, { "epoch": 2.11, "learning_rate": 6.056448794127928e-06, "loss": 0.0135, "step": 16460 }, { "epoch": 2.11, "learning_rate": 6.047710590702552e-06, "loss": 0.0127, "step": 16470 }, { "epoch": 2.11, "learning_rate": 6.038972387277176e-06, "loss": 0.0151, "step": 16480 }, { "epoch": 2.12, "learning_rate": 6.030234183851801e-06, "loss": 0.0131, "step": 16490 }, { "epoch": 2.12, "learning_rate": 6.021495980426425e-06, "loss": 0.0099, "step": 16500 }, { "epoch": 2.12, "learning_rate": 6.012757777001049e-06, "loss": 0.0128, "step": 16510 }, { "epoch": 2.12, "learning_rate": 6.0040195735756734e-06, "loss": 0.0106, "step": 16520 }, { "epoch": 2.12, "learning_rate": 5.9952813701502975e-06, "loss": 0.0165, "step": 16530 }, { "epoch": 2.12, "learning_rate": 5.986543166724922e-06, "loss": 0.0137, "step": 16540 }, { "epoch": 2.12, "learning_rate": 5.977804963299546e-06, "loss": 0.0134, "step": 16550 }, { "epoch": 2.12, "learning_rate": 5.9690667598741706e-06, "loss": 0.0127, "step": 16560 }, { "epoch": 2.13, "learning_rate": 5.960328556448795e-06, "loss": 0.0128, "step": 16570 }, { "epoch": 2.13, "learning_rate": 5.951590353023419e-06, "loss": 0.0154, "step": 16580 }, { "epoch": 2.13, "learning_rate": 5.942852149598043e-06, "loss": 0.0115, "step": 16590 }, { "epoch": 2.13, "learning_rate": 5.934113946172667e-06, "loss": 0.0143, "step": 16600 }, { "epoch": 2.13, "learning_rate": 5.925375742747291e-06, "loss": 0.0127, "step": 16610 }, { "epoch": 2.13, "learning_rate": 5.916637539321916e-06, "loss": 0.0146, "step": 16620 }, { "epoch": 2.13, "learning_rate": 5.90789933589654e-06, "loss": 0.0144, "step": 16630 }, { "epoch": 2.13, "learning_rate": 5.899161132471164e-06, "loss": 0.015, "step": 16640 }, { "epoch": 2.14, "learning_rate": 5.890422929045788e-06, "loss": 0.0103, "step": 16650 }, { "epoch": 2.14, "learning_rate": 5.881684725620412e-06, "loss": 0.0115, "step": 16660 }, { "epoch": 2.14, "learning_rate": 5.872946522195038e-06, "loss": 0.0171, "step": 16670 }, { "epoch": 2.14, "learning_rate": 5.864208318769662e-06, "loss": 0.0123, "step": 16680 }, { "epoch": 2.14, "learning_rate": 5.855470115344286e-06, "loss": 0.0124, "step": 16690 }, { "epoch": 2.14, "learning_rate": 5.84673191191891e-06, "loss": 0.0121, "step": 16700 }, { "epoch": 2.14, "learning_rate": 5.837993708493534e-06, "loss": 0.0114, "step": 16710 }, { "epoch": 2.14, "learning_rate": 5.829255505068159e-06, "loss": 0.0148, "step": 16720 }, { "epoch": 2.15, "learning_rate": 5.820517301642783e-06, "loss": 0.0102, "step": 16730 }, { "epoch": 2.15, "learning_rate": 5.811779098217407e-06, "loss": 0.0126, "step": 16740 }, { "epoch": 2.15, "learning_rate": 5.803040894792031e-06, "loss": 0.0119, "step": 16750 }, { "epoch": 2.15, "learning_rate": 5.7943026913666554e-06, "loss": 0.0126, "step": 16760 }, { "epoch": 2.15, "learning_rate": 5.7855644879412795e-06, "loss": 0.0112, "step": 16770 }, { "epoch": 2.15, "learning_rate": 5.776826284515904e-06, "loss": 0.0156, "step": 16780 }, { "epoch": 2.15, "learning_rate": 5.7680880810905285e-06, "loss": 0.0154, "step": 16790 }, { "epoch": 2.15, "learning_rate": 5.7593498776651526e-06, "loss": 0.0137, "step": 16800 }, { "epoch": 2.16, "learning_rate": 5.750611674239777e-06, "loss": 0.0086, "step": 16810 }, { "epoch": 2.16, "learning_rate": 5.741873470814401e-06, "loss": 0.0114, "step": 16820 }, { "epoch": 2.16, "learning_rate": 5.733135267389025e-06, "loss": 0.0134, "step": 16830 }, { "epoch": 2.16, "learning_rate": 5.724397063963649e-06, "loss": 0.0107, "step": 16840 }, { "epoch": 2.16, "learning_rate": 5.715658860538274e-06, "loss": 0.0177, "step": 16850 }, { "epoch": 2.16, "learning_rate": 5.706920657112898e-06, "loss": 0.0134, "step": 16860 }, { "epoch": 2.16, "learning_rate": 5.698182453687522e-06, "loss": 0.0179, "step": 16870 }, { "epoch": 2.17, "learning_rate": 5.689444250262146e-06, "loss": 0.0159, "step": 16880 }, { "epoch": 2.17, "learning_rate": 5.68070604683677e-06, "loss": 0.0132, "step": 16890 }, { "epoch": 2.17, "learning_rate": 5.671967843411396e-06, "loss": 0.012, "step": 16900 }, { "epoch": 2.17, "learning_rate": 5.66322963998602e-06, "loss": 0.0158, "step": 16910 }, { "epoch": 2.17, "learning_rate": 5.654491436560644e-06, "loss": 0.0124, "step": 16920 }, { "epoch": 2.17, "learning_rate": 5.645753233135268e-06, "loss": 0.0143, "step": 16930 }, { "epoch": 2.17, "learning_rate": 5.637015029709892e-06, "loss": 0.0108, "step": 16940 }, { "epoch": 2.17, "learning_rate": 5.628276826284517e-06, "loss": 0.0128, "step": 16950 }, { "epoch": 2.18, "learning_rate": 5.619538622859141e-06, "loss": 0.0115, "step": 16960 }, { "epoch": 2.18, "learning_rate": 5.610800419433765e-06, "loss": 0.0166, "step": 16970 }, { "epoch": 2.18, "learning_rate": 5.602062216008389e-06, "loss": 0.0146, "step": 16980 }, { "epoch": 2.18, "learning_rate": 5.593324012583013e-06, "loss": 0.0106, "step": 16990 }, { "epoch": 2.18, "learning_rate": 5.5845858091576374e-06, "loss": 0.0134, "step": 17000 }, { "epoch": 2.18, "learning_rate": 5.5758476057322615e-06, "loss": 0.0141, "step": 17010 }, { "epoch": 2.18, "learning_rate": 5.5671094023068864e-06, "loss": 0.01, "step": 17020 }, { "epoch": 2.18, "learning_rate": 5.5583711988815105e-06, "loss": 0.0164, "step": 17030 }, { "epoch": 2.19, "learning_rate": 5.549632995456135e-06, "loss": 0.0121, "step": 17040 }, { "epoch": 2.19, "learning_rate": 5.540894792030759e-06, "loss": 0.0133, "step": 17050 }, { "epoch": 2.19, "learning_rate": 5.532156588605383e-06, "loss": 0.0128, "step": 17060 }, { "epoch": 2.19, "learning_rate": 5.523418385180007e-06, "loss": 0.0126, "step": 17070 }, { "epoch": 2.19, "learning_rate": 5.514680181754632e-06, "loss": 0.0133, "step": 17080 }, { "epoch": 2.19, "learning_rate": 5.505941978329256e-06, "loss": 0.0144, "step": 17090 }, { "epoch": 2.19, "learning_rate": 5.49720377490388e-06, "loss": 0.0155, "step": 17100 }, { "epoch": 2.19, "learning_rate": 5.488465571478504e-06, "loss": 0.0105, "step": 17110 }, { "epoch": 2.2, "learning_rate": 5.479727368053128e-06, "loss": 0.0106, "step": 17120 }, { "epoch": 2.2, "learning_rate": 5.470989164627754e-06, "loss": 0.0137, "step": 17130 }, { "epoch": 2.2, "learning_rate": 5.462250961202378e-06, "loss": 0.0129, "step": 17140 }, { "epoch": 2.2, "learning_rate": 5.453512757777002e-06, "loss": 0.013, "step": 17150 }, { "epoch": 2.2, "learning_rate": 5.444774554351626e-06, "loss": 0.016, "step": 17160 }, { "epoch": 2.2, "learning_rate": 5.43603635092625e-06, "loss": 0.009, "step": 17170 }, { "epoch": 2.2, "learning_rate": 5.427298147500875e-06, "loss": 0.0156, "step": 17180 }, { "epoch": 2.2, "learning_rate": 5.418559944075499e-06, "loss": 0.0135, "step": 17190 }, { "epoch": 2.21, "learning_rate": 5.409821740650123e-06, "loss": 0.0138, "step": 17200 }, { "epoch": 2.21, "learning_rate": 5.401083537224747e-06, "loss": 0.014, "step": 17210 }, { "epoch": 2.21, "learning_rate": 5.392345333799371e-06, "loss": 0.0156, "step": 17220 }, { "epoch": 2.21, "learning_rate": 5.383607130373995e-06, "loss": 0.0149, "step": 17230 }, { "epoch": 2.21, "learning_rate": 5.3748689269486194e-06, "loss": 0.0119, "step": 17240 }, { "epoch": 2.21, "learning_rate": 5.366130723523244e-06, "loss": 0.018, "step": 17250 }, { "epoch": 2.21, "learning_rate": 5.3573925200978684e-06, "loss": 0.0152, "step": 17260 }, { "epoch": 2.22, "learning_rate": 5.3486543166724925e-06, "loss": 0.0112, "step": 17270 }, { "epoch": 2.22, "learning_rate": 5.339916113247117e-06, "loss": 0.0149, "step": 17280 }, { "epoch": 2.22, "learning_rate": 5.331177909821741e-06, "loss": 0.0172, "step": 17290 }, { "epoch": 2.22, "learning_rate": 5.322439706396365e-06, "loss": 0.0107, "step": 17300 }, { "epoch": 2.22, "learning_rate": 5.31370150297099e-06, "loss": 0.0162, "step": 17310 }, { "epoch": 2.22, "learning_rate": 5.304963299545614e-06, "loss": 0.0148, "step": 17320 }, { "epoch": 2.22, "learning_rate": 5.296225096120238e-06, "loss": 0.0124, "step": 17330 }, { "epoch": 2.22, "learning_rate": 5.287486892694862e-06, "loss": 0.0122, "step": 17340 }, { "epoch": 2.23, "learning_rate": 5.278748689269486e-06, "loss": 0.0127, "step": 17350 }, { "epoch": 2.23, "learning_rate": 5.27001048584411e-06, "loss": 0.0101, "step": 17360 }, { "epoch": 2.23, "learning_rate": 5.261272282418736e-06, "loss": 0.0134, "step": 17370 }, { "epoch": 2.23, "learning_rate": 5.25253407899336e-06, "loss": 0.0114, "step": 17380 }, { "epoch": 2.23, "learning_rate": 5.243795875567984e-06, "loss": 0.0147, "step": 17390 }, { "epoch": 2.23, "learning_rate": 5.235057672142608e-06, "loss": 0.0124, "step": 17400 }, { "epoch": 2.23, "learning_rate": 5.226319468717233e-06, "loss": 0.0113, "step": 17410 }, { "epoch": 2.23, "learning_rate": 5.217581265291857e-06, "loss": 0.0102, "step": 17420 }, { "epoch": 2.24, "learning_rate": 5.208843061866481e-06, "loss": 0.0166, "step": 17430 }, { "epoch": 2.24, "learning_rate": 5.200104858441105e-06, "loss": 0.0148, "step": 17440 }, { "epoch": 2.24, "learning_rate": 5.191366655015729e-06, "loss": 0.011, "step": 17450 }, { "epoch": 2.24, "learning_rate": 5.182628451590353e-06, "loss": 0.0103, "step": 17460 }, { "epoch": 2.24, "learning_rate": 5.173890248164977e-06, "loss": 0.0107, "step": 17470 }, { "epoch": 2.24, "learning_rate": 5.165152044739602e-06, "loss": 0.0122, "step": 17480 }, { "epoch": 2.24, "learning_rate": 5.156413841314226e-06, "loss": 0.0109, "step": 17490 }, { "epoch": 2.24, "learning_rate": 5.1476756378888504e-06, "loss": 0.0164, "step": 17500 }, { "epoch": 2.25, "learning_rate": 5.1389374344634745e-06, "loss": 0.0131, "step": 17510 }, { "epoch": 2.25, "learning_rate": 5.130199231038099e-06, "loss": 0.0118, "step": 17520 }, { "epoch": 2.25, "learning_rate": 5.121461027612723e-06, "loss": 0.0137, "step": 17530 }, { "epoch": 2.25, "learning_rate": 5.1127228241873476e-06, "loss": 0.012, "step": 17540 }, { "epoch": 2.25, "learning_rate": 5.103984620761972e-06, "loss": 0.0145, "step": 17550 }, { "epoch": 2.25, "learning_rate": 5.095246417336596e-06, "loss": 0.0138, "step": 17560 }, { "epoch": 2.25, "learning_rate": 5.08650821391122e-06, "loss": 0.0162, "step": 17570 }, { "epoch": 2.25, "learning_rate": 5.077770010485844e-06, "loss": 0.0147, "step": 17580 }, { "epoch": 2.26, "learning_rate": 5.069031807060468e-06, "loss": 0.0132, "step": 17590 }, { "epoch": 2.26, "learning_rate": 5.060293603635094e-06, "loss": 0.0119, "step": 17600 }, { "epoch": 2.26, "learning_rate": 5.051555400209718e-06, "loss": 0.0138, "step": 17610 }, { "epoch": 2.26, "learning_rate": 5.042817196784342e-06, "loss": 0.0153, "step": 17620 }, { "epoch": 2.26, "learning_rate": 5.034078993358966e-06, "loss": 0.0161, "step": 17630 }, { "epoch": 2.26, "learning_rate": 5.02534078993359e-06, "loss": 0.0103, "step": 17640 }, { "epoch": 2.26, "learning_rate": 5.016602586508215e-06, "loss": 0.0132, "step": 17650 }, { "epoch": 2.27, "learning_rate": 5.007864383082839e-06, "loss": 0.0102, "step": 17660 }, { "epoch": 2.27, "learning_rate": 4.999126179657463e-06, "loss": 0.0108, "step": 17670 }, { "epoch": 2.27, "learning_rate": 4.990387976232087e-06, "loss": 0.0135, "step": 17680 }, { "epoch": 2.27, "learning_rate": 4.981649772806711e-06, "loss": 0.0117, "step": 17690 }, { "epoch": 2.27, "learning_rate": 4.972911569381335e-06, "loss": 0.014, "step": 17700 }, { "epoch": 2.27, "learning_rate": 4.96417336595596e-06, "loss": 0.0077, "step": 17710 }, { "epoch": 2.27, "learning_rate": 4.955435162530584e-06, "loss": 0.0132, "step": 17720 }, { "epoch": 2.27, "learning_rate": 4.946696959105208e-06, "loss": 0.0099, "step": 17730 }, { "epoch": 2.28, "learning_rate": 4.9379587556798324e-06, "loss": 0.0131, "step": 17740 }, { "epoch": 2.28, "learning_rate": 4.9292205522544565e-06, "loss": 0.0139, "step": 17750 }, { "epoch": 2.28, "learning_rate": 4.920482348829081e-06, "loss": 0.0126, "step": 17760 }, { "epoch": 2.28, "learning_rate": 4.9117441454037055e-06, "loss": 0.0112, "step": 17770 }, { "epoch": 2.28, "learning_rate": 4.9030059419783296e-06, "loss": 0.0128, "step": 17780 }, { "epoch": 2.28, "learning_rate": 4.8942677385529545e-06, "loss": 0.0151, "step": 17790 }, { "epoch": 2.28, "learning_rate": 4.8855295351275786e-06, "loss": 0.0126, "step": 17800 }, { "epoch": 2.28, "learning_rate": 4.876791331702203e-06, "loss": 0.0106, "step": 17810 }, { "epoch": 2.29, "learning_rate": 4.868053128276827e-06, "loss": 0.0175, "step": 17820 }, { "epoch": 2.29, "learning_rate": 4.859314924851451e-06, "loss": 0.011, "step": 17830 }, { "epoch": 2.29, "learning_rate": 4.850576721426075e-06, "loss": 0.0118, "step": 17840 }, { "epoch": 2.29, "learning_rate": 4.8418385180007e-06, "loss": 0.0133, "step": 17850 }, { "epoch": 2.29, "learning_rate": 4.833100314575324e-06, "loss": 0.014, "step": 17860 }, { "epoch": 2.29, "learning_rate": 4.824362111149948e-06, "loss": 0.0154, "step": 17870 }, { "epoch": 2.29, "learning_rate": 4.815623907724572e-06, "loss": 0.0159, "step": 17880 }, { "epoch": 2.29, "learning_rate": 4.806885704299196e-06, "loss": 0.0149, "step": 17890 }, { "epoch": 2.3, "learning_rate": 4.798147500873821e-06, "loss": 0.0103, "step": 17900 }, { "epoch": 2.3, "learning_rate": 4.789409297448445e-06, "loss": 0.0128, "step": 17910 }, { "epoch": 2.3, "learning_rate": 4.780671094023069e-06, "loss": 0.0114, "step": 17920 }, { "epoch": 2.3, "learning_rate": 4.771932890597693e-06, "loss": 0.0148, "step": 17930 }, { "epoch": 2.3, "learning_rate": 4.763194687172318e-06, "loss": 0.0129, "step": 17940 }, { "epoch": 2.3, "learning_rate": 4.754456483746942e-06, "loss": 0.0117, "step": 17950 }, { "epoch": 2.3, "learning_rate": 4.745718280321566e-06, "loss": 0.0116, "step": 17960 }, { "epoch": 2.3, "learning_rate": 4.73698007689619e-06, "loss": 0.0129, "step": 17970 }, { "epoch": 2.31, "learning_rate": 4.7282418734708144e-06, "loss": 0.0124, "step": 17980 }, { "epoch": 2.31, "learning_rate": 4.7195036700454385e-06, "loss": 0.01, "step": 17990 }, { "epoch": 2.31, "learning_rate": 4.7107654666200634e-06, "loss": 0.0131, "step": 18000 }, { "epoch": 2.31, "learning_rate": 4.7020272631946875e-06, "loss": 0.0106, "step": 18010 }, { "epoch": 2.31, "learning_rate": 4.693289059769312e-06, "loss": 0.0104, "step": 18020 }, { "epoch": 2.31, "learning_rate": 4.6845508563439365e-06, "loss": 0.0122, "step": 18030 }, { "epoch": 2.31, "learning_rate": 4.6758126529185606e-06, "loss": 0.0135, "step": 18040 }, { "epoch": 2.32, "learning_rate": 4.667074449493185e-06, "loss": 0.012, "step": 18050 }, { "epoch": 2.32, "learning_rate": 4.658336246067809e-06, "loss": 0.0138, "step": 18060 }, { "epoch": 2.32, "learning_rate": 4.649598042642433e-06, "loss": 0.0084, "step": 18070 }, { "epoch": 2.32, "learning_rate": 4.640859839217058e-06, "loss": 0.0098, "step": 18080 }, { "epoch": 2.32, "learning_rate": 4.632121635791682e-06, "loss": 0.0122, "step": 18090 }, { "epoch": 2.32, "learning_rate": 4.623383432366306e-06, "loss": 0.0131, "step": 18100 }, { "epoch": 2.32, "learning_rate": 4.61464522894093e-06, "loss": 0.0122, "step": 18110 }, { "epoch": 2.32, "learning_rate": 4.605907025515554e-06, "loss": 0.0123, "step": 18120 }, { "epoch": 2.33, "learning_rate": 4.597168822090179e-06, "loss": 0.0142, "step": 18130 }, { "epoch": 2.33, "learning_rate": 4.588430618664803e-06, "loss": 0.0121, "step": 18140 }, { "epoch": 2.33, "learning_rate": 4.579692415239427e-06, "loss": 0.0121, "step": 18150 }, { "epoch": 2.33, "learning_rate": 4.570954211814051e-06, "loss": 0.014, "step": 18160 }, { "epoch": 2.33, "learning_rate": 4.562216008388676e-06, "loss": 0.0122, "step": 18170 }, { "epoch": 2.33, "learning_rate": 4.5534778049633e-06, "loss": 0.0141, "step": 18180 }, { "epoch": 2.33, "learning_rate": 4.544739601537924e-06, "loss": 0.0137, "step": 18190 }, { "epoch": 2.33, "learning_rate": 4.536001398112548e-06, "loss": 0.0175, "step": 18200 }, { "epoch": 2.34, "learning_rate": 4.527263194687172e-06, "loss": 0.014, "step": 18210 }, { "epoch": 2.34, "learning_rate": 4.5185249912617964e-06, "loss": 0.0141, "step": 18220 }, { "epoch": 2.34, "learning_rate": 4.509786787836421e-06, "loss": 0.0088, "step": 18230 }, { "epoch": 2.34, "learning_rate": 4.5010485844110454e-06, "loss": 0.0138, "step": 18240 }, { "epoch": 2.34, "learning_rate": 4.49231038098567e-06, "loss": 0.0123, "step": 18250 }, { "epoch": 2.34, "learning_rate": 4.483572177560294e-06, "loss": 0.011, "step": 18260 }, { "epoch": 2.34, "learning_rate": 4.4748339741349185e-06, "loss": 0.0173, "step": 18270 }, { "epoch": 2.34, "learning_rate": 4.4660957707095426e-06, "loss": 0.0115, "step": 18280 }, { "epoch": 2.35, "learning_rate": 4.457357567284167e-06, "loss": 0.0086, "step": 18290 }, { "epoch": 2.35, "learning_rate": 4.448619363858791e-06, "loss": 0.0116, "step": 18300 }, { "epoch": 2.35, "learning_rate": 4.439881160433416e-06, "loss": 0.012, "step": 18310 }, { "epoch": 2.35, "learning_rate": 4.43114295700804e-06, "loss": 0.0115, "step": 18320 }, { "epoch": 2.35, "learning_rate": 4.422404753582664e-06, "loss": 0.0131, "step": 18330 }, { "epoch": 2.35, "learning_rate": 4.413666550157288e-06, "loss": 0.0145, "step": 18340 }, { "epoch": 2.35, "learning_rate": 4.404928346731912e-06, "loss": 0.0132, "step": 18350 }, { "epoch": 2.35, "learning_rate": 4.396190143306536e-06, "loss": 0.0156, "step": 18360 }, { "epoch": 2.36, "learning_rate": 4.387451939881161e-06, "loss": 0.0143, "step": 18370 }, { "epoch": 2.36, "learning_rate": 4.378713736455785e-06, "loss": 0.0164, "step": 18380 }, { "epoch": 2.36, "learning_rate": 4.369975533030409e-06, "loss": 0.0184, "step": 18390 }, { "epoch": 2.36, "learning_rate": 4.361237329605034e-06, "loss": 0.0109, "step": 18400 }, { "epoch": 2.36, "learning_rate": 4.352499126179658e-06, "loss": 0.0166, "step": 18410 }, { "epoch": 2.36, "learning_rate": 4.343760922754282e-06, "loss": 0.0097, "step": 18420 }, { "epoch": 2.36, "learning_rate": 4.335022719328906e-06, "loss": 0.0135, "step": 18430 }, { "epoch": 2.37, "learning_rate": 4.32628451590353e-06, "loss": 0.0153, "step": 18440 }, { "epoch": 2.37, "learning_rate": 4.317546312478154e-06, "loss": 0.0118, "step": 18450 }, { "epoch": 2.37, "learning_rate": 4.308808109052779e-06, "loss": 0.0139, "step": 18460 }, { "epoch": 2.37, "learning_rate": 4.300069905627403e-06, "loss": 0.0139, "step": 18470 }, { "epoch": 2.37, "learning_rate": 4.291331702202028e-06, "loss": 0.0085, "step": 18480 }, { "epoch": 2.37, "learning_rate": 4.282593498776652e-06, "loss": 0.0136, "step": 18490 }, { "epoch": 2.37, "learning_rate": 4.2738552953512764e-06, "loss": 0.0139, "step": 18500 }, { "epoch": 2.37, "learning_rate": 4.2651170919259005e-06, "loss": 0.0159, "step": 18510 }, { "epoch": 2.38, "learning_rate": 4.2563788885005246e-06, "loss": 0.0145, "step": 18520 }, { "epoch": 2.38, "learning_rate": 4.247640685075149e-06, "loss": 0.0107, "step": 18530 }, { "epoch": 2.38, "learning_rate": 4.238902481649773e-06, "loss": 0.0157, "step": 18540 }, { "epoch": 2.38, "learning_rate": 4.230164278224398e-06, "loss": 0.0083, "step": 18550 }, { "epoch": 2.38, "learning_rate": 4.221426074799022e-06, "loss": 0.016, "step": 18560 }, { "epoch": 2.38, "learning_rate": 4.212687871373646e-06, "loss": 0.0145, "step": 18570 }, { "epoch": 2.38, "learning_rate": 4.20394966794827e-06, "loss": 0.0139, "step": 18580 }, { "epoch": 2.38, "learning_rate": 4.195211464522894e-06, "loss": 0.0126, "step": 18590 }, { "epoch": 2.39, "learning_rate": 4.186473261097519e-06, "loss": 0.0106, "step": 18600 }, { "epoch": 2.39, "learning_rate": 4.177735057672143e-06, "loss": 0.0134, "step": 18610 }, { "epoch": 2.39, "learning_rate": 4.168996854246767e-06, "loss": 0.0105, "step": 18620 }, { "epoch": 2.39, "learning_rate": 4.160258650821392e-06, "loss": 0.0201, "step": 18630 }, { "epoch": 2.39, "learning_rate": 4.151520447396016e-06, "loss": 0.0169, "step": 18640 }, { "epoch": 2.39, "learning_rate": 4.14278224397064e-06, "loss": 0.0167, "step": 18650 }, { "epoch": 2.39, "learning_rate": 4.134044040545264e-06, "loss": 0.0133, "step": 18660 }, { "epoch": 2.39, "learning_rate": 4.125305837119888e-06, "loss": 0.0158, "step": 18670 }, { "epoch": 2.4, "learning_rate": 4.116567633694512e-06, "loss": 0.0128, "step": 18680 }, { "epoch": 2.4, "learning_rate": 4.107829430269137e-06, "loss": 0.0149, "step": 18690 }, { "epoch": 2.4, "learning_rate": 4.099091226843761e-06, "loss": 0.0132, "step": 18700 }, { "epoch": 2.4, "learning_rate": 4.090353023418385e-06, "loss": 0.016, "step": 18710 }, { "epoch": 2.4, "learning_rate": 4.08161481999301e-06, "loss": 0.0125, "step": 18720 }, { "epoch": 2.4, "learning_rate": 4.072876616567634e-06, "loss": 0.0137, "step": 18730 }, { "epoch": 2.4, "learning_rate": 4.0641384131422584e-06, "loss": 0.0099, "step": 18740 }, { "epoch": 2.41, "learning_rate": 4.0554002097168825e-06, "loss": 0.0156, "step": 18750 }, { "epoch": 2.41, "learning_rate": 4.0466620062915066e-06, "loss": 0.0125, "step": 18760 }, { "epoch": 2.41, "learning_rate": 4.037923802866131e-06, "loss": 0.0106, "step": 18770 }, { "epoch": 2.41, "learning_rate": 4.0291855994407556e-06, "loss": 0.0126, "step": 18780 }, { "epoch": 2.41, "learning_rate": 4.02044739601538e-06, "loss": 0.0158, "step": 18790 }, { "epoch": 2.41, "learning_rate": 4.011709192590004e-06, "loss": 0.0173, "step": 18800 }, { "epoch": 2.41, "learning_rate": 4.002970989164628e-06, "loss": 0.0099, "step": 18810 }, { "epoch": 2.41, "learning_rate": 3.994232785739252e-06, "loss": 0.0153, "step": 18820 }, { "epoch": 2.42, "learning_rate": 3.985494582313877e-06, "loss": 0.0146, "step": 18830 }, { "epoch": 2.42, "learning_rate": 3.976756378888501e-06, "loss": 0.0148, "step": 18840 }, { "epoch": 2.42, "learning_rate": 3.968018175463125e-06, "loss": 0.0117, "step": 18850 }, { "epoch": 2.42, "learning_rate": 3.95927997203775e-06, "loss": 0.0129, "step": 18860 }, { "epoch": 2.42, "learning_rate": 3.950541768612374e-06, "loss": 0.0115, "step": 18870 }, { "epoch": 2.42, "learning_rate": 3.941803565186998e-06, "loss": 0.0122, "step": 18880 }, { "epoch": 2.42, "learning_rate": 3.933065361761622e-06, "loss": 0.0143, "step": 18890 }, { "epoch": 2.42, "learning_rate": 3.924327158336246e-06, "loss": 0.0106, "step": 18900 }, { "epoch": 2.43, "learning_rate": 3.91558895491087e-06, "loss": 0.0121, "step": 18910 }, { "epoch": 2.43, "learning_rate": 3.906850751485495e-06, "loss": 0.007, "step": 18920 }, { "epoch": 2.43, "learning_rate": 3.898112548060119e-06, "loss": 0.0124, "step": 18930 }, { "epoch": 2.43, "learning_rate": 3.889374344634743e-06, "loss": 0.0106, "step": 18940 }, { "epoch": 2.43, "learning_rate": 3.880636141209368e-06, "loss": 0.0167, "step": 18950 }, { "epoch": 2.43, "learning_rate": 3.871897937783992e-06, "loss": 0.013, "step": 18960 }, { "epoch": 2.43, "learning_rate": 3.863159734358616e-06, "loss": 0.0115, "step": 18970 }, { "epoch": 2.43, "learning_rate": 3.8544215309332404e-06, "loss": 0.0094, "step": 18980 }, { "epoch": 2.44, "learning_rate": 3.8456833275078645e-06, "loss": 0.0131, "step": 18990 }, { "epoch": 2.44, "learning_rate": 3.8369451240824886e-06, "loss": 0.0124, "step": 19000 }, { "epoch": 2.44, "learning_rate": 3.8282069206571135e-06, "loss": 0.0122, "step": 19010 }, { "epoch": 2.44, "learning_rate": 3.8194687172317376e-06, "loss": 0.012, "step": 19020 }, { "epoch": 2.44, "learning_rate": 3.8107305138063616e-06, "loss": 0.0121, "step": 19030 }, { "epoch": 2.44, "learning_rate": 3.8019923103809857e-06, "loss": 0.0128, "step": 19040 }, { "epoch": 2.44, "learning_rate": 3.79325410695561e-06, "loss": 0.0145, "step": 19050 }, { "epoch": 2.44, "learning_rate": 3.7845159035302343e-06, "loss": 0.0116, "step": 19060 }, { "epoch": 2.45, "learning_rate": 3.775777700104859e-06, "loss": 0.0132, "step": 19070 }, { "epoch": 2.45, "learning_rate": 3.7670394966794833e-06, "loss": 0.0114, "step": 19080 }, { "epoch": 2.45, "learning_rate": 3.7583012932541074e-06, "loss": 0.0163, "step": 19090 }, { "epoch": 2.45, "learning_rate": 3.7495630898287314e-06, "loss": 0.0121, "step": 19100 }, { "epoch": 2.45, "learning_rate": 3.740824886403356e-06, "loss": 0.0133, "step": 19110 }, { "epoch": 2.45, "learning_rate": 3.73208668297798e-06, "loss": 0.0132, "step": 19120 }, { "epoch": 2.45, "learning_rate": 3.723348479552604e-06, "loss": 0.0139, "step": 19130 }, { "epoch": 2.46, "learning_rate": 3.7146102761272286e-06, "loss": 0.015, "step": 19140 }, { "epoch": 2.46, "learning_rate": 3.7058720727018526e-06, "loss": 0.0121, "step": 19150 }, { "epoch": 2.46, "learning_rate": 3.6971338692764767e-06, "loss": 0.0114, "step": 19160 }, { "epoch": 2.46, "learning_rate": 3.688395665851101e-06, "loss": 0.0135, "step": 19170 }, { "epoch": 2.46, "learning_rate": 3.6796574624257253e-06, "loss": 0.0121, "step": 19180 }, { "epoch": 2.46, "learning_rate": 3.67091925900035e-06, "loss": 0.0147, "step": 19190 }, { "epoch": 2.46, "learning_rate": 3.6621810555749743e-06, "loss": 0.0134, "step": 19200 }, { "epoch": 2.46, "learning_rate": 3.6534428521495984e-06, "loss": 0.0125, "step": 19210 }, { "epoch": 2.47, "learning_rate": 3.644704648724223e-06, "loss": 0.0136, "step": 19220 }, { "epoch": 2.47, "learning_rate": 3.635966445298847e-06, "loss": 0.0102, "step": 19230 }, { "epoch": 2.47, "learning_rate": 3.627228241873471e-06, "loss": 0.011, "step": 19240 }, { "epoch": 2.47, "learning_rate": 3.6184900384480955e-06, "loss": 0.0134, "step": 19250 }, { "epoch": 2.47, "learning_rate": 3.6097518350227196e-06, "loss": 0.0138, "step": 19260 }, { "epoch": 2.47, "learning_rate": 3.6010136315973436e-06, "loss": 0.0109, "step": 19270 }, { "epoch": 2.47, "learning_rate": 3.592275428171968e-06, "loss": 0.0104, "step": 19280 }, { "epoch": 2.47, "learning_rate": 3.583537224746592e-06, "loss": 0.0118, "step": 19290 }, { "epoch": 2.48, "learning_rate": 3.574799021321217e-06, "loss": 0.0114, "step": 19300 }, { "epoch": 2.48, "learning_rate": 3.566060817895841e-06, "loss": 0.014, "step": 19310 }, { "epoch": 2.48, "learning_rate": 3.5573226144704653e-06, "loss": 0.015, "step": 19320 }, { "epoch": 2.48, "learning_rate": 3.5485844110450894e-06, "loss": 0.0139, "step": 19330 }, { "epoch": 2.48, "learning_rate": 3.539846207619714e-06, "loss": 0.0141, "step": 19340 }, { "epoch": 2.48, "learning_rate": 3.531108004194338e-06, "loss": 0.0124, "step": 19350 }, { "epoch": 2.48, "learning_rate": 3.522369800768962e-06, "loss": 0.0128, "step": 19360 }, { "epoch": 2.48, "learning_rate": 3.5136315973435865e-06, "loss": 0.0167, "step": 19370 }, { "epoch": 2.49, "learning_rate": 3.5048933939182106e-06, "loss": 0.0107, "step": 19380 }, { "epoch": 2.49, "learning_rate": 3.4961551904928346e-06, "loss": 0.0135, "step": 19390 }, { "epoch": 2.49, "learning_rate": 3.487416987067459e-06, "loss": 0.0122, "step": 19400 }, { "epoch": 2.49, "learning_rate": 3.4786787836420832e-06, "loss": 0.0102, "step": 19410 }, { "epoch": 2.49, "learning_rate": 3.469940580216708e-06, "loss": 0.0133, "step": 19420 }, { "epoch": 2.49, "learning_rate": 3.461202376791332e-06, "loss": 0.0109, "step": 19430 }, { "epoch": 2.49, "learning_rate": 3.4524641733659563e-06, "loss": 0.0154, "step": 19440 }, { "epoch": 2.49, "learning_rate": 3.4437259699405808e-06, "loss": 0.0132, "step": 19450 }, { "epoch": 2.5, "learning_rate": 3.434987766515205e-06, "loss": 0.0139, "step": 19460 }, { "epoch": 2.5, "learning_rate": 3.426249563089829e-06, "loss": 0.0171, "step": 19470 }, { "epoch": 2.5, "learning_rate": 3.4175113596644534e-06, "loss": 0.0107, "step": 19480 }, { "epoch": 2.5, "learning_rate": 3.4087731562390775e-06, "loss": 0.0156, "step": 19490 }, { "epoch": 2.5, "learning_rate": 3.4000349528137016e-06, "loss": 0.0124, "step": 19500 }, { "epoch": 2.5, "learning_rate": 3.391296749388326e-06, "loss": 0.0138, "step": 19510 }, { "epoch": 2.5, "learning_rate": 3.38255854596295e-06, "loss": 0.0171, "step": 19520 }, { "epoch": 2.51, "learning_rate": 3.3738203425375742e-06, "loss": 0.0102, "step": 19530 }, { "epoch": 2.51, "learning_rate": 3.365082139112199e-06, "loss": 0.0119, "step": 19540 }, { "epoch": 2.51, "learning_rate": 3.356343935686823e-06, "loss": 0.0126, "step": 19550 }, { "epoch": 2.51, "learning_rate": 3.3476057322614473e-06, "loss": 0.0139, "step": 19560 }, { "epoch": 2.51, "learning_rate": 3.3388675288360718e-06, "loss": 0.0163, "step": 19570 }, { "epoch": 2.51, "learning_rate": 3.330129325410696e-06, "loss": 0.0132, "step": 19580 }, { "epoch": 2.51, "learning_rate": 3.32139112198532e-06, "loss": 0.0147, "step": 19590 }, { "epoch": 2.51, "learning_rate": 3.3126529185599444e-06, "loss": 0.017, "step": 19600 }, { "epoch": 2.52, "learning_rate": 3.3039147151345685e-06, "loss": 0.0138, "step": 19610 }, { "epoch": 2.52, "learning_rate": 3.2951765117091926e-06, "loss": 0.013, "step": 19620 }, { "epoch": 2.52, "learning_rate": 3.286438308283817e-06, "loss": 0.0154, "step": 19630 }, { "epoch": 2.52, "learning_rate": 3.277700104858441e-06, "loss": 0.0115, "step": 19640 }, { "epoch": 2.52, "learning_rate": 3.2689619014330652e-06, "loss": 0.0095, "step": 19650 }, { "epoch": 2.52, "learning_rate": 3.26022369800769e-06, "loss": 0.0098, "step": 19660 }, { "epoch": 2.52, "learning_rate": 3.251485494582314e-06, "loss": 0.0148, "step": 19670 }, { "epoch": 2.52, "learning_rate": 3.2427472911569387e-06, "loss": 0.012, "step": 19680 }, { "epoch": 2.53, "learning_rate": 3.2340090877315628e-06, "loss": 0.0145, "step": 19690 }, { "epoch": 2.53, "learning_rate": 3.225270884306187e-06, "loss": 0.0111, "step": 19700 }, { "epoch": 2.53, "learning_rate": 3.2165326808808113e-06, "loss": 0.013, "step": 19710 }, { "epoch": 2.53, "learning_rate": 3.2077944774554354e-06, "loss": 0.0168, "step": 19720 }, { "epoch": 2.53, "learning_rate": 3.1990562740300595e-06, "loss": 0.0132, "step": 19730 }, { "epoch": 2.53, "learning_rate": 3.190318070604684e-06, "loss": 0.0147, "step": 19740 }, { "epoch": 2.53, "learning_rate": 3.181579867179308e-06, "loss": 0.0117, "step": 19750 }, { "epoch": 2.53, "learning_rate": 3.172841663753932e-06, "loss": 0.0132, "step": 19760 }, { "epoch": 2.54, "learning_rate": 3.164103460328557e-06, "loss": 0.0135, "step": 19770 }, { "epoch": 2.54, "learning_rate": 3.155365256903181e-06, "loss": 0.0177, "step": 19780 }, { "epoch": 2.54, "learning_rate": 3.146627053477805e-06, "loss": 0.0118, "step": 19790 }, { "epoch": 2.54, "learning_rate": 3.1378888500524297e-06, "loss": 0.0142, "step": 19800 }, { "epoch": 2.54, "learning_rate": 3.1291506466270538e-06, "loss": 0.0126, "step": 19810 }, { "epoch": 2.54, "learning_rate": 3.120412443201678e-06, "loss": 0.0152, "step": 19820 }, { "epoch": 2.54, "learning_rate": 3.1116742397763023e-06, "loss": 0.0124, "step": 19830 }, { "epoch": 2.54, "learning_rate": 3.1029360363509264e-06, "loss": 0.0134, "step": 19840 }, { "epoch": 2.55, "learning_rate": 3.0941978329255505e-06, "loss": 0.0108, "step": 19850 }, { "epoch": 2.55, "learning_rate": 3.085459629500175e-06, "loss": 0.0158, "step": 19860 }, { "epoch": 2.55, "learning_rate": 3.076721426074799e-06, "loss": 0.0093, "step": 19870 }, { "epoch": 2.55, "learning_rate": 3.067983222649423e-06, "loss": 0.0118, "step": 19880 }, { "epoch": 2.55, "learning_rate": 3.059245019224048e-06, "loss": 0.0118, "step": 19890 }, { "epoch": 2.55, "learning_rate": 3.050506815798672e-06, "loss": 0.0106, "step": 19900 }, { "epoch": 2.55, "learning_rate": 3.0417686123732966e-06, "loss": 0.0124, "step": 19910 }, { "epoch": 2.56, "learning_rate": 3.0330304089479207e-06, "loss": 0.0144, "step": 19920 }, { "epoch": 2.56, "learning_rate": 3.0242922055225448e-06, "loss": 0.014, "step": 19930 }, { "epoch": 2.56, "learning_rate": 3.0155540020971693e-06, "loss": 0.0149, "step": 19940 }, { "epoch": 2.56, "learning_rate": 3.0068157986717933e-06, "loss": 0.0145, "step": 19950 }, { "epoch": 2.56, "learning_rate": 2.9980775952464174e-06, "loss": 0.0132, "step": 19960 }, { "epoch": 2.56, "learning_rate": 2.989339391821042e-06, "loss": 0.0135, "step": 19970 }, { "epoch": 2.56, "learning_rate": 2.980601188395666e-06, "loss": 0.013, "step": 19980 }, { "epoch": 2.56, "learning_rate": 2.97186298497029e-06, "loss": 0.014, "step": 19990 }, { "epoch": 2.57, "learning_rate": 2.963124781544914e-06, "loss": 0.0144, "step": 20000 } ], "max_steps": 23388, "num_train_epochs": 3, "total_flos": 7.429605556224e+16, "trial_name": null, "trial_params": null }