{ "best_metric": null, "best_model_checkpoint": null, "epoch": 1.0316723408645414, "eval_steps": 500, "global_step": 10000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 3.333333333333334e-06, "loss": 2.6106, "step": 10 }, { "epoch": 0.0, "learning_rate": 4.336766652213271e-06, "loss": 1.5727, "step": 20 }, { "epoch": 0.0, "learning_rate": 4.923737515732209e-06, "loss": 1.5652, "step": 30 }, { "epoch": 0.0, "learning_rate": 5.3401999710932085e-06, "loss": 1.3798, "step": 40 }, { "epoch": 0.01, "learning_rate": 5.6632333477867295e-06, "loss": 1.2357, "step": 50 }, { "epoch": 0.01, "learning_rate": 5.927170834612146e-06, "loss": 1.1619, "step": 60 }, { "epoch": 0.01, "learning_rate": 6.150326800047524e-06, "loss": 1.2551, "step": 70 }, { "epoch": 0.01, "learning_rate": 6.343633289973146e-06, "loss": 1.2001, "step": 80 }, { "epoch": 0.01, "learning_rate": 6.514141698131084e-06, "loss": 1.1448, "step": 90 }, { "epoch": 0.01, "learning_rate": 6.666666666666668e-06, "loss": 1.1931, "step": 100 }, { "epoch": 0.01, "learning_rate": 6.804642283860752e-06, "loss": 1.0584, "step": 110 }, { "epoch": 0.01, "learning_rate": 6.930604153492083e-06, "loss": 1.0047, "step": 120 }, { "epoch": 0.01, "learning_rate": 7.04647784102279e-06, "loss": 1.1146, "step": 130 }, { "epoch": 0.01, "learning_rate": 7.153760118927461e-06, "loss": 1.0603, "step": 140 }, { "epoch": 0.02, "learning_rate": 7.253637530185604e-06, "loss": 1.0357, "step": 150 }, { "epoch": 0.02, "learning_rate": 7.347066608853083e-06, "loss": 1.1379, "step": 160 }, { "epoch": 0.02, "learning_rate": 7.434829737927582e-06, "loss": 1.0961, "step": 170 }, { "epoch": 0.02, "learning_rate": 7.517575017011021e-06, "loss": 0.9989, "step": 180 }, { "epoch": 0.02, "learning_rate": 7.5958453365094305e-06, "loss": 1.0777, "step": 190 }, { "epoch": 0.02, "learning_rate": 7.670099985546604e-06, "loss": 0.9513, "step": 200 }, { "epoch": 0.02, "learning_rate": 7.740730982446398e-06, "loss": 1.0395, "step": 210 }, { "epoch": 0.02, "learning_rate": 7.808075602740689e-06, "loss": 1.0065, "step": 220 }, { "epoch": 0.02, "learning_rate": 7.872426120058644e-06, "loss": 1.0587, "step": 230 }, { "epoch": 0.02, "learning_rate": 7.93403747237202e-06, "loss": 0.9835, "step": 240 }, { "epoch": 0.03, "learning_rate": 7.993133362240125e-06, "loss": 1.0397, "step": 250 }, { "epoch": 0.03, "learning_rate": 8.049911159902726e-06, "loss": 0.9604, "step": 260 }, { "epoch": 0.03, "learning_rate": 8.10454588052996e-06, "loss": 1.062, "step": 270 }, { "epoch": 0.03, "learning_rate": 8.157193437807398e-06, "loss": 1.0354, "step": 280 }, { "epoch": 0.03, "learning_rate": 8.207993326329854e-06, "loss": 1.0079, "step": 290 }, { "epoch": 0.03, "learning_rate": 8.257070849065542e-06, "loss": 1.025, "step": 300 }, { "epoch": 0.03, "learning_rate": 8.304538979447577e-06, "loss": 1.031, "step": 310 }, { "epoch": 0.03, "learning_rate": 8.35049992773302e-06, "loss": 0.9556, "step": 320 }, { "epoch": 0.03, "learning_rate": 8.395046466259626e-06, "loss": 0.953, "step": 330 }, { "epoch": 0.04, "learning_rate": 8.438263056807519e-06, "loss": 1.0261, "step": 340 }, { "epoch": 0.04, "learning_rate": 8.480226814500919e-06, "loss": 0.9867, "step": 350 }, { "epoch": 0.04, "learning_rate": 8.52100833589096e-06, "loss": 1.0307, "step": 360 }, { "epoch": 0.04, "learning_rate": 8.560672413556651e-06, "loss": 1.0861, "step": 370 }, { "epoch": 0.04, "learning_rate": 8.599278655389369e-06, "loss": 0.9894, "step": 380 }, { "epoch": 0.04, "learning_rate": 8.636882023421665e-06, "loss": 0.977, "step": 390 }, { "epoch": 0.04, "learning_rate": 8.673533304426543e-06, "loss": 0.9932, "step": 400 }, { "epoch": 0.04, "learning_rate": 8.70927952239912e-06, "loss": 0.9188, "step": 410 }, { "epoch": 0.04, "learning_rate": 8.744164301326335e-06, "loss": 0.9795, "step": 420 }, { "epoch": 0.04, "learning_rate": 8.77822818526529e-06, "loss": 1.0477, "step": 430 }, { "epoch": 0.05, "learning_rate": 8.811508921620624e-06, "loss": 1.0468, "step": 440 }, { "epoch": 0.05, "learning_rate": 8.84404171258448e-06, "loss": 0.9682, "step": 450 }, { "epoch": 0.05, "learning_rate": 8.875859438938582e-06, "loss": 1.0267, "step": 460 }, { "epoch": 0.05, "learning_rate": 8.906992859785726e-06, "loss": 0.9496, "step": 470 }, { "epoch": 0.05, "learning_rate": 8.937470791251958e-06, "loss": 0.9669, "step": 480 }, { "epoch": 0.05, "learning_rate": 8.967320266761713e-06, "loss": 0.9814, "step": 490 }, { "epoch": 0.05, "learning_rate": 8.996566681120064e-06, "loss": 0.9928, "step": 500 }, { "epoch": 0.05, "learning_rate": 9.025233920326456e-06, "loss": 0.958, "step": 510 }, { "epoch": 0.05, "learning_rate": 9.053344478782665e-06, "loss": 0.9652, "step": 520 }, { "epoch": 0.05, "learning_rate": 9.080919565335966e-06, "loss": 0.9688, "step": 530 }, { "epoch": 0.06, "learning_rate": 9.107979199409897e-06, "loss": 1.0448, "step": 540 }, { "epoch": 0.06, "learning_rate": 9.134542298314147e-06, "loss": 0.9465, "step": 550 }, { "epoch": 0.06, "learning_rate": 9.160626756687335e-06, "loss": 1.0627, "step": 560 }, { "epoch": 0.06, "learning_rate": 9.186249518908307e-06, "loss": 0.9456, "step": 570 }, { "epoch": 0.06, "learning_rate": 9.211426645209793e-06, "loss": 0.9926, "step": 580 }, { "epoch": 0.06, "learning_rate": 9.23617337214048e-06, "loss": 0.9588, "step": 590 }, { "epoch": 0.06, "learning_rate": 9.26050416794548e-06, "loss": 1.0404, "step": 600 }, { "epoch": 0.06, "learning_rate": 9.284432783369225e-06, "loss": 0.9622, "step": 610 }, { "epoch": 0.06, "learning_rate": 9.307972298327514e-06, "loss": 0.9708, "step": 620 }, { "epoch": 0.06, "learning_rate": 9.331135164845275e-06, "loss": 0.9856, "step": 630 }, { "epoch": 0.07, "learning_rate": 9.353933246612959e-06, "loss": 0.9355, "step": 640 }, { "epoch": 0.07, "learning_rate": 9.376377855476186e-06, "loss": 0.8562, "step": 650 }, { "epoch": 0.07, "learning_rate": 9.398479785139564e-06, "loss": 1.0288, "step": 660 }, { "epoch": 0.07, "learning_rate": 9.420249342336088e-06, "loss": 0.9901, "step": 670 }, { "epoch": 0.07, "learning_rate": 9.441696375687454e-06, "loss": 0.9852, "step": 680 }, { "epoch": 0.07, "learning_rate": 9.46283030245752e-06, "loss": 1.0026, "step": 690 }, { "epoch": 0.07, "learning_rate": 9.483660133380856e-06, "loss": 0.9461, "step": 700 }, { "epoch": 0.07, "learning_rate": 9.504194495730252e-06, "loss": 1.0419, "step": 710 }, { "epoch": 0.07, "learning_rate": 9.524441654770897e-06, "loss": 1.0165, "step": 720 }, { "epoch": 0.08, "learning_rate": 9.544409533734855e-06, "loss": 0.9352, "step": 730 }, { "epoch": 0.08, "learning_rate": 9.564105732436588e-06, "loss": 0.9799, "step": 740 }, { "epoch": 0.08, "learning_rate": 9.583537544639e-06, "loss": 0.9588, "step": 750 }, { "epoch": 0.08, "learning_rate": 9.602711974269306e-06, "loss": 0.9504, "step": 760 }, { "epoch": 0.08, "learning_rate": 9.62163575057494e-06, "loss": 0.9078, "step": 770 }, { "epoch": 0.08, "learning_rate": 9.640315342301602e-06, "loss": 0.8803, "step": 780 }, { "epoch": 0.08, "learning_rate": 9.65875697096814e-06, "loss": 0.9084, "step": 790 }, { "epoch": 0.08, "learning_rate": 9.67696662330648e-06, "loss": 0.9443, "step": 800 }, { "epoch": 0.08, "learning_rate": 9.694950062928833e-06, "loss": 0.8999, "step": 810 }, { "epoch": 0.08, "learning_rate": 9.712712841279056e-06, "loss": 0.9956, "step": 820 }, { "epoch": 0.09, "learning_rate": 9.730260307920247e-06, "loss": 0.9747, "step": 830 }, { "epoch": 0.09, "learning_rate": 9.747597620206274e-06, "loss": 0.9841, "step": 840 }, { "epoch": 0.09, "learning_rate": 9.764729752380979e-06, "loss": 0.9923, "step": 850 }, { "epoch": 0.09, "learning_rate": 9.781661504145226e-06, "loss": 1.0748, "step": 860 }, { "epoch": 0.09, "learning_rate": 9.79839750872873e-06, "loss": 0.9848, "step": 870 }, { "epoch": 0.09, "learning_rate": 9.814942240500561e-06, "loss": 0.9379, "step": 880 }, { "epoch": 0.09, "learning_rate": 9.83130002214971e-06, "loss": 0.9742, "step": 890 }, { "epoch": 0.09, "learning_rate": 9.847475031464418e-06, "loss": 0.9195, "step": 900 }, { "epoch": 0.09, "learning_rate": 9.863471307736979e-06, "loss": 0.9626, "step": 910 }, { "epoch": 0.09, "learning_rate": 9.87929275781852e-06, "loss": 0.8797, "step": 920 }, { "epoch": 0.1, "learning_rate": 9.894943161846451e-06, "loss": 0.9836, "step": 930 }, { "epoch": 0.1, "learning_rate": 9.910426178665664e-06, "loss": 1.0288, "step": 940 }, { "epoch": 0.1, "learning_rate": 9.925745350962828e-06, "loss": 0.8993, "step": 950 }, { "epoch": 0.1, "learning_rate": 9.940904110131896e-06, "loss": 1.0167, "step": 960 }, { "epoch": 0.1, "learning_rate": 9.955905780887485e-06, "loss": 0.9826, "step": 970 }, { "epoch": 0.1, "learning_rate": 9.97075358564165e-06, "loss": 0.923, "step": 980 }, { "epoch": 0.1, "learning_rate": 9.985450648658501e-06, "loss": 0.9491, "step": 990 }, { "epoch": 0.1, "learning_rate": 1e-05, "loss": 0.9143, "step": 1000 }, { "epoch": 0.1, "learning_rate": 9.99510497117372e-06, "loss": 0.9344, "step": 1010 }, { "epoch": 0.11, "learning_rate": 9.98966605025563e-06, "loss": 0.9759, "step": 1020 }, { "epoch": 0.11, "learning_rate": 9.98422712933754e-06, "loss": 0.8452, "step": 1030 }, { "epoch": 0.11, "learning_rate": 9.97878820841945e-06, "loss": 0.9419, "step": 1040 }, { "epoch": 0.11, "learning_rate": 9.97334928750136e-06, "loss": 0.9672, "step": 1050 }, { "epoch": 0.11, "learning_rate": 9.967910366583272e-06, "loss": 0.9575, "step": 1060 }, { "epoch": 0.11, "learning_rate": 9.962471445665181e-06, "loss": 1.0185, "step": 1070 }, { "epoch": 0.11, "learning_rate": 9.957032524747091e-06, "loss": 0.9757, "step": 1080 }, { "epoch": 0.11, "learning_rate": 9.951593603829e-06, "loss": 1.0028, "step": 1090 }, { "epoch": 0.11, "learning_rate": 9.94615468291091e-06, "loss": 1.016, "step": 1100 }, { "epoch": 0.11, "learning_rate": 9.940715761992822e-06, "loss": 0.9078, "step": 1110 }, { "epoch": 0.12, "learning_rate": 9.935276841074732e-06, "loss": 0.8926, "step": 1120 }, { "epoch": 0.12, "learning_rate": 9.929837920156641e-06, "loss": 0.944, "step": 1130 }, { "epoch": 0.12, "learning_rate": 9.924398999238551e-06, "loss": 1.0196, "step": 1140 }, { "epoch": 0.12, "learning_rate": 9.918960078320463e-06, "loss": 0.9221, "step": 1150 }, { "epoch": 0.12, "learning_rate": 9.913521157402372e-06, "loss": 0.9515, "step": 1160 }, { "epoch": 0.12, "learning_rate": 9.908082236484282e-06, "loss": 0.9711, "step": 1170 }, { "epoch": 0.12, "learning_rate": 9.902643315566192e-06, "loss": 0.909, "step": 1180 }, { "epoch": 0.12, "learning_rate": 9.897204394648102e-06, "loss": 0.9462, "step": 1190 }, { "epoch": 0.12, "learning_rate": 9.891765473730013e-06, "loss": 0.8828, "step": 1200 }, { "epoch": 0.12, "learning_rate": 9.886326552811923e-06, "loss": 1.0132, "step": 1210 }, { "epoch": 0.13, "learning_rate": 9.880887631893833e-06, "loss": 0.9295, "step": 1220 }, { "epoch": 0.13, "learning_rate": 9.875448710975744e-06, "loss": 0.932, "step": 1230 }, { "epoch": 0.13, "learning_rate": 9.870009790057654e-06, "loss": 0.9089, "step": 1240 }, { "epoch": 0.13, "learning_rate": 9.864570869139564e-06, "loss": 1.0115, "step": 1250 }, { "epoch": 0.13, "learning_rate": 9.859131948221473e-06, "loss": 0.9132, "step": 1260 }, { "epoch": 0.13, "learning_rate": 9.853693027303383e-06, "loss": 1.0324, "step": 1270 }, { "epoch": 0.13, "learning_rate": 9.848254106385295e-06, "loss": 1.0514, "step": 1280 }, { "epoch": 0.13, "learning_rate": 9.842815185467204e-06, "loss": 1.005, "step": 1290 }, { "epoch": 0.13, "learning_rate": 9.837376264549114e-06, "loss": 0.974, "step": 1300 }, { "epoch": 0.14, "learning_rate": 9.831937343631024e-06, "loss": 0.9395, "step": 1310 }, { "epoch": 0.14, "learning_rate": 9.826498422712935e-06, "loss": 0.8957, "step": 1320 }, { "epoch": 0.14, "learning_rate": 9.821059501794845e-06, "loss": 0.9005, "step": 1330 }, { "epoch": 0.14, "learning_rate": 9.815620580876755e-06, "loss": 0.9167, "step": 1340 }, { "epoch": 0.14, "learning_rate": 9.810181659958664e-06, "loss": 0.9572, "step": 1350 }, { "epoch": 0.14, "learning_rate": 9.804742739040574e-06, "loss": 0.8927, "step": 1360 }, { "epoch": 0.14, "learning_rate": 9.799303818122486e-06, "loss": 0.9553, "step": 1370 }, { "epoch": 0.14, "learning_rate": 9.793864897204395e-06, "loss": 0.9278, "step": 1380 }, { "epoch": 0.14, "learning_rate": 9.788425976286305e-06, "loss": 0.9593, "step": 1390 }, { "epoch": 0.14, "learning_rate": 9.782987055368217e-06, "loss": 0.9144, "step": 1400 }, { "epoch": 0.15, "learning_rate": 9.777548134450126e-06, "loss": 0.9605, "step": 1410 }, { "epoch": 0.15, "learning_rate": 9.772109213532036e-06, "loss": 1.0299, "step": 1420 }, { "epoch": 0.15, "learning_rate": 9.766670292613948e-06, "loss": 0.9275, "step": 1430 }, { "epoch": 0.15, "learning_rate": 9.761231371695856e-06, "loss": 0.8546, "step": 1440 }, { "epoch": 0.15, "learning_rate": 9.755792450777767e-06, "loss": 0.9156, "step": 1450 }, { "epoch": 0.15, "learning_rate": 9.750353529859677e-06, "loss": 1.0148, "step": 1460 }, { "epoch": 0.15, "learning_rate": 9.744914608941587e-06, "loss": 0.9446, "step": 1470 }, { "epoch": 0.15, "learning_rate": 9.739475688023496e-06, "loss": 0.9648, "step": 1480 }, { "epoch": 0.15, "learning_rate": 9.734036767105408e-06, "loss": 0.8864, "step": 1490 }, { "epoch": 0.15, "learning_rate": 9.728597846187317e-06, "loss": 1.0421, "step": 1500 }, { "epoch": 0.16, "learning_rate": 9.723158925269227e-06, "loss": 1.0442, "step": 1510 }, { "epoch": 0.16, "learning_rate": 9.717720004351137e-06, "loss": 0.9484, "step": 1520 }, { "epoch": 0.16, "learning_rate": 9.712281083433047e-06, "loss": 0.8694, "step": 1530 }, { "epoch": 0.16, "learning_rate": 9.706842162514958e-06, "loss": 0.9045, "step": 1540 }, { "epoch": 0.16, "learning_rate": 9.701403241596868e-06, "loss": 1.0326, "step": 1550 }, { "epoch": 0.16, "learning_rate": 9.695964320678778e-06, "loss": 0.9055, "step": 1560 }, { "epoch": 0.16, "learning_rate": 9.690525399760689e-06, "loss": 1.0101, "step": 1570 }, { "epoch": 0.16, "learning_rate": 9.685086478842599e-06, "loss": 0.9309, "step": 1580 }, { "epoch": 0.16, "learning_rate": 9.679647557924509e-06, "loss": 0.9804, "step": 1590 }, { "epoch": 0.17, "learning_rate": 9.67420863700642e-06, "loss": 0.8933, "step": 1600 }, { "epoch": 0.17, "learning_rate": 9.668769716088328e-06, "loss": 0.951, "step": 1610 }, { "epoch": 0.17, "learning_rate": 9.66333079517024e-06, "loss": 0.865, "step": 1620 }, { "epoch": 0.17, "learning_rate": 9.65789187425215e-06, "loss": 0.9339, "step": 1630 }, { "epoch": 0.17, "learning_rate": 9.652452953334059e-06, "loss": 0.9365, "step": 1640 }, { "epoch": 0.17, "learning_rate": 9.64701403241597e-06, "loss": 0.9776, "step": 1650 }, { "epoch": 0.17, "learning_rate": 9.64157511149788e-06, "loss": 0.9163, "step": 1660 }, { "epoch": 0.17, "learning_rate": 9.63613619057979e-06, "loss": 0.9185, "step": 1670 }, { "epoch": 0.17, "learning_rate": 9.6306972696617e-06, "loss": 0.8722, "step": 1680 }, { "epoch": 0.17, "learning_rate": 9.62525834874361e-06, "loss": 0.9687, "step": 1690 }, { "epoch": 0.18, "learning_rate": 9.61981942782552e-06, "loss": 0.957, "step": 1700 }, { "epoch": 0.18, "learning_rate": 9.61438050690743e-06, "loss": 1.0049, "step": 1710 }, { "epoch": 0.18, "learning_rate": 9.60894158598934e-06, "loss": 0.9988, "step": 1720 }, { "epoch": 0.18, "learning_rate": 9.60350266507125e-06, "loss": 0.9559, "step": 1730 }, { "epoch": 0.18, "learning_rate": 9.598063744153162e-06, "loss": 0.9382, "step": 1740 }, { "epoch": 0.18, "learning_rate": 9.59262482323507e-06, "loss": 0.9768, "step": 1750 }, { "epoch": 0.18, "learning_rate": 9.587185902316981e-06, "loss": 0.9117, "step": 1760 }, { "epoch": 0.18, "learning_rate": 9.581746981398893e-06, "loss": 0.9207, "step": 1770 }, { "epoch": 0.18, "learning_rate": 9.5763080604808e-06, "loss": 0.9364, "step": 1780 }, { "epoch": 0.18, "learning_rate": 9.570869139562712e-06, "loss": 0.9356, "step": 1790 }, { "epoch": 0.19, "learning_rate": 9.565430218644622e-06, "loss": 0.8421, "step": 1800 }, { "epoch": 0.19, "learning_rate": 9.559991297726532e-06, "loss": 0.8875, "step": 1810 }, { "epoch": 0.19, "learning_rate": 9.554552376808443e-06, "loss": 0.9186, "step": 1820 }, { "epoch": 0.19, "learning_rate": 9.549113455890353e-06, "loss": 0.9512, "step": 1830 }, { "epoch": 0.19, "learning_rate": 9.543674534972262e-06, "loss": 0.9518, "step": 1840 }, { "epoch": 0.19, "learning_rate": 9.538235614054172e-06, "loss": 0.8742, "step": 1850 }, { "epoch": 0.19, "learning_rate": 9.532796693136082e-06, "loss": 0.8865, "step": 1860 }, { "epoch": 0.19, "learning_rate": 9.527357772217992e-06, "loss": 0.9434, "step": 1870 }, { "epoch": 0.19, "learning_rate": 9.521918851299903e-06, "loss": 0.9112, "step": 1880 }, { "epoch": 0.19, "learning_rate": 9.516479930381813e-06, "loss": 1.021, "step": 1890 }, { "epoch": 0.2, "learning_rate": 9.511041009463723e-06, "loss": 0.9226, "step": 1900 }, { "epoch": 0.2, "learning_rate": 9.505602088545634e-06, "loss": 1.1388, "step": 1910 }, { "epoch": 0.2, "learning_rate": 9.500163167627542e-06, "loss": 0.9428, "step": 1920 }, { "epoch": 0.2, "learning_rate": 9.494724246709454e-06, "loss": 0.9463, "step": 1930 }, { "epoch": 0.2, "learning_rate": 9.489285325791363e-06, "loss": 1.0233, "step": 1940 }, { "epoch": 0.2, "learning_rate": 9.483846404873273e-06, "loss": 0.9855, "step": 1950 }, { "epoch": 0.2, "learning_rate": 9.478407483955185e-06, "loss": 0.9213, "step": 1960 }, { "epoch": 0.2, "learning_rate": 9.472968563037094e-06, "loss": 0.8818, "step": 1970 }, { "epoch": 0.2, "learning_rate": 9.467529642119004e-06, "loss": 0.8945, "step": 1980 }, { "epoch": 0.21, "learning_rate": 9.462090721200915e-06, "loss": 0.8794, "step": 1990 }, { "epoch": 0.21, "learning_rate": 9.456651800282825e-06, "loss": 0.8888, "step": 2000 }, { "epoch": 0.21, "learning_rate": 9.451212879364735e-06, "loss": 1.0205, "step": 2010 }, { "epoch": 0.21, "learning_rate": 9.445773958446645e-06, "loss": 0.8877, "step": 2020 }, { "epoch": 0.21, "learning_rate": 9.440335037528554e-06, "loss": 0.8376, "step": 2030 }, { "epoch": 0.21, "learning_rate": 9.434896116610466e-06, "loss": 0.9704, "step": 2040 }, { "epoch": 0.21, "learning_rate": 9.429457195692376e-06, "loss": 0.9746, "step": 2050 }, { "epoch": 0.21, "learning_rate": 9.424018274774285e-06, "loss": 1.0305, "step": 2060 }, { "epoch": 0.21, "learning_rate": 9.418579353856195e-06, "loss": 0.8811, "step": 2070 }, { "epoch": 0.21, "learning_rate": 9.413140432938107e-06, "loss": 0.9581, "step": 2080 }, { "epoch": 0.22, "learning_rate": 9.407701512020015e-06, "loss": 1.0344, "step": 2090 }, { "epoch": 0.22, "learning_rate": 9.402262591101926e-06, "loss": 0.8794, "step": 2100 }, { "epoch": 0.22, "learning_rate": 9.396823670183836e-06, "loss": 0.8707, "step": 2110 }, { "epoch": 0.22, "learning_rate": 9.391384749265746e-06, "loss": 0.8849, "step": 2120 }, { "epoch": 0.22, "learning_rate": 9.385945828347657e-06, "loss": 0.9847, "step": 2130 }, { "epoch": 0.22, "learning_rate": 9.380506907429567e-06, "loss": 0.9208, "step": 2140 }, { "epoch": 0.22, "learning_rate": 9.375067986511477e-06, "loss": 0.9634, "step": 2150 }, { "epoch": 0.22, "learning_rate": 9.369629065593388e-06, "loss": 0.8081, "step": 2160 }, { "epoch": 0.22, "learning_rate": 9.364190144675298e-06, "loss": 0.9122, "step": 2170 }, { "epoch": 0.22, "learning_rate": 9.358751223757207e-06, "loss": 0.8631, "step": 2180 }, { "epoch": 0.23, "learning_rate": 9.353312302839117e-06, "loss": 0.929, "step": 2190 }, { "epoch": 0.23, "learning_rate": 9.347873381921027e-06, "loss": 0.9087, "step": 2200 }, { "epoch": 0.23, "learning_rate": 9.342434461002938e-06, "loss": 0.8679, "step": 2210 }, { "epoch": 0.23, "learning_rate": 9.336995540084848e-06, "loss": 0.9182, "step": 2220 }, { "epoch": 0.23, "learning_rate": 9.331556619166758e-06, "loss": 0.9705, "step": 2230 }, { "epoch": 0.23, "learning_rate": 9.326117698248668e-06, "loss": 0.9558, "step": 2240 }, { "epoch": 0.23, "learning_rate": 9.320678777330579e-06, "loss": 0.8742, "step": 2250 }, { "epoch": 0.23, "learning_rate": 9.315239856412489e-06, "loss": 0.9211, "step": 2260 }, { "epoch": 0.23, "learning_rate": 9.309800935494399e-06, "loss": 0.8997, "step": 2270 }, { "epoch": 0.24, "learning_rate": 9.304362014576308e-06, "loss": 0.9156, "step": 2280 }, { "epoch": 0.24, "learning_rate": 9.298923093658218e-06, "loss": 0.9096, "step": 2290 }, { "epoch": 0.24, "learning_rate": 9.29348417274013e-06, "loss": 0.9163, "step": 2300 }, { "epoch": 0.24, "learning_rate": 9.28804525182204e-06, "loss": 0.9584, "step": 2310 }, { "epoch": 0.24, "learning_rate": 9.282606330903949e-06, "loss": 0.811, "step": 2320 }, { "epoch": 0.24, "learning_rate": 9.27716740998586e-06, "loss": 0.9817, "step": 2330 }, { "epoch": 0.24, "learning_rate": 9.27172848906777e-06, "loss": 0.9837, "step": 2340 }, { "epoch": 0.24, "learning_rate": 9.26628956814968e-06, "loss": 1.0754, "step": 2350 }, { "epoch": 0.24, "learning_rate": 9.26085064723159e-06, "loss": 0.8921, "step": 2360 }, { "epoch": 0.24, "learning_rate": 9.2554117263135e-06, "loss": 0.9506, "step": 2370 }, { "epoch": 0.25, "learning_rate": 9.249972805395411e-06, "loss": 0.9563, "step": 2380 }, { "epoch": 0.25, "learning_rate": 9.24453388447732e-06, "loss": 0.9381, "step": 2390 }, { "epoch": 0.25, "learning_rate": 9.23909496355923e-06, "loss": 0.8863, "step": 2400 }, { "epoch": 0.25, "learning_rate": 9.23365604264114e-06, "loss": 0.9708, "step": 2410 }, { "epoch": 0.25, "learning_rate": 9.228217121723052e-06, "loss": 0.9005, "step": 2420 }, { "epoch": 0.25, "learning_rate": 9.222778200804961e-06, "loss": 0.9038, "step": 2430 }, { "epoch": 0.25, "learning_rate": 9.217339279886871e-06, "loss": 0.8887, "step": 2440 }, { "epoch": 0.25, "learning_rate": 9.211900358968781e-06, "loss": 0.9044, "step": 2450 }, { "epoch": 0.25, "learning_rate": 9.20646143805069e-06, "loss": 0.9485, "step": 2460 }, { "epoch": 0.25, "learning_rate": 9.201022517132602e-06, "loss": 0.9431, "step": 2470 }, { "epoch": 0.26, "learning_rate": 9.195583596214512e-06, "loss": 0.8901, "step": 2480 }, { "epoch": 0.26, "learning_rate": 9.190144675296422e-06, "loss": 0.9431, "step": 2490 }, { "epoch": 0.26, "learning_rate": 9.184705754378333e-06, "loss": 0.9076, "step": 2500 }, { "epoch": 0.26, "learning_rate": 9.179266833460243e-06, "loss": 0.92, "step": 2510 }, { "epoch": 0.26, "learning_rate": 9.173827912542153e-06, "loss": 0.8699, "step": 2520 }, { "epoch": 0.26, "learning_rate": 9.168388991624062e-06, "loss": 0.9598, "step": 2530 }, { "epoch": 0.26, "learning_rate": 9.162950070705972e-06, "loss": 0.9327, "step": 2540 }, { "epoch": 0.26, "learning_rate": 9.157511149787883e-06, "loss": 0.9215, "step": 2550 }, { "epoch": 0.26, "learning_rate": 9.152072228869793e-06, "loss": 1.0092, "step": 2560 }, { "epoch": 0.27, "learning_rate": 9.146633307951703e-06, "loss": 0.9127, "step": 2570 }, { "epoch": 0.27, "learning_rate": 9.141194387033613e-06, "loss": 1.0034, "step": 2580 }, { "epoch": 0.27, "learning_rate": 9.135755466115524e-06, "loss": 0.9249, "step": 2590 }, { "epoch": 0.27, "learning_rate": 9.130316545197434e-06, "loss": 0.8519, "step": 2600 }, { "epoch": 0.27, "learning_rate": 9.124877624279344e-06, "loss": 0.9484, "step": 2610 }, { "epoch": 0.27, "learning_rate": 9.119438703361253e-06, "loss": 0.9517, "step": 2620 }, { "epoch": 0.27, "learning_rate": 9.113999782443163e-06, "loss": 0.8678, "step": 2630 }, { "epoch": 0.27, "learning_rate": 9.108560861525075e-06, "loss": 0.9806, "step": 2640 }, { "epoch": 0.27, "learning_rate": 9.103121940606984e-06, "loss": 0.8237, "step": 2650 }, { "epoch": 0.27, "learning_rate": 9.097683019688894e-06, "loss": 0.9503, "step": 2660 }, { "epoch": 0.28, "learning_rate": 9.092244098770806e-06, "loss": 0.9539, "step": 2670 }, { "epoch": 0.28, "learning_rate": 9.086805177852714e-06, "loss": 0.9866, "step": 2680 }, { "epoch": 0.28, "learning_rate": 9.081366256934625e-06, "loss": 0.9679, "step": 2690 }, { "epoch": 0.28, "learning_rate": 9.075927336016535e-06, "loss": 0.9127, "step": 2700 }, { "epoch": 0.28, "learning_rate": 9.070488415098445e-06, "loss": 0.9168, "step": 2710 }, { "epoch": 0.28, "learning_rate": 9.065049494180356e-06, "loss": 0.8988, "step": 2720 }, { "epoch": 0.28, "learning_rate": 9.059610573262266e-06, "loss": 0.8839, "step": 2730 }, { "epoch": 0.28, "learning_rate": 9.054171652344175e-06, "loss": 0.9461, "step": 2740 }, { "epoch": 0.28, "learning_rate": 9.048732731426085e-06, "loss": 0.8859, "step": 2750 }, { "epoch": 0.28, "learning_rate": 9.043293810507997e-06, "loss": 0.92, "step": 2760 }, { "epoch": 0.29, "learning_rate": 9.037854889589906e-06, "loss": 0.9274, "step": 2770 }, { "epoch": 0.29, "learning_rate": 9.032415968671816e-06, "loss": 0.8567, "step": 2780 }, { "epoch": 0.29, "learning_rate": 9.026977047753726e-06, "loss": 0.8631, "step": 2790 }, { "epoch": 0.29, "learning_rate": 9.021538126835636e-06, "loss": 0.8292, "step": 2800 }, { "epoch": 0.29, "learning_rate": 9.016099205917547e-06, "loss": 0.9457, "step": 2810 }, { "epoch": 0.29, "learning_rate": 9.010660284999457e-06, "loss": 1.0156, "step": 2820 }, { "epoch": 0.29, "learning_rate": 9.005221364081367e-06, "loss": 0.9607, "step": 2830 }, { "epoch": 0.29, "learning_rate": 8.999782443163278e-06, "loss": 0.982, "step": 2840 }, { "epoch": 0.29, "learning_rate": 8.994343522245186e-06, "loss": 0.9278, "step": 2850 }, { "epoch": 0.3, "learning_rate": 8.988904601327098e-06, "loss": 0.9351, "step": 2860 }, { "epoch": 0.3, "learning_rate": 8.983465680409007e-06, "loss": 0.8853, "step": 2870 }, { "epoch": 0.3, "learning_rate": 8.978026759490917e-06, "loss": 0.8922, "step": 2880 }, { "epoch": 0.3, "learning_rate": 8.972587838572828e-06, "loss": 0.8916, "step": 2890 }, { "epoch": 0.3, "learning_rate": 8.967148917654738e-06, "loss": 0.8931, "step": 2900 }, { "epoch": 0.3, "learning_rate": 8.961709996736648e-06, "loss": 0.9384, "step": 2910 }, { "epoch": 0.3, "learning_rate": 8.956271075818558e-06, "loss": 0.9872, "step": 2920 }, { "epoch": 0.3, "learning_rate": 8.95083215490047e-06, "loss": 0.9648, "step": 2930 }, { "epoch": 0.3, "learning_rate": 8.945393233982379e-06, "loss": 0.9628, "step": 2940 }, { "epoch": 0.3, "learning_rate": 8.939954313064289e-06, "loss": 0.9195, "step": 2950 }, { "epoch": 0.31, "learning_rate": 8.934515392146198e-06, "loss": 0.9834, "step": 2960 }, { "epoch": 0.31, "learning_rate": 8.929076471228108e-06, "loss": 0.9282, "step": 2970 }, { "epoch": 0.31, "learning_rate": 8.92363755031002e-06, "loss": 0.9218, "step": 2980 }, { "epoch": 0.31, "learning_rate": 8.91819862939193e-06, "loss": 0.9737, "step": 2990 }, { "epoch": 0.31, "learning_rate": 8.912759708473839e-06, "loss": 0.8811, "step": 3000 }, { "epoch": 0.31, "learning_rate": 8.90732078755575e-06, "loss": 0.9512, "step": 3010 }, { "epoch": 0.31, "learning_rate": 8.901881866637659e-06, "loss": 0.8873, "step": 3020 }, { "epoch": 0.31, "learning_rate": 8.89644294571957e-06, "loss": 0.9111, "step": 3030 }, { "epoch": 0.31, "learning_rate": 8.89100402480148e-06, "loss": 0.9662, "step": 3040 }, { "epoch": 0.31, "learning_rate": 8.88556510388339e-06, "loss": 0.8619, "step": 3050 }, { "epoch": 0.32, "learning_rate": 8.880126182965301e-06, "loss": 0.9541, "step": 3060 }, { "epoch": 0.32, "learning_rate": 8.87468726204721e-06, "loss": 1.0127, "step": 3070 }, { "epoch": 0.32, "learning_rate": 8.86924834112912e-06, "loss": 0.9135, "step": 3080 }, { "epoch": 0.32, "learning_rate": 8.86380942021103e-06, "loss": 0.9706, "step": 3090 }, { "epoch": 0.32, "learning_rate": 8.858370499292942e-06, "loss": 0.8487, "step": 3100 }, { "epoch": 0.32, "learning_rate": 8.852931578374851e-06, "loss": 0.8969, "step": 3110 }, { "epoch": 0.32, "learning_rate": 8.847492657456761e-06, "loss": 0.8835, "step": 3120 }, { "epoch": 0.32, "learning_rate": 8.842053736538671e-06, "loss": 0.973, "step": 3130 }, { "epoch": 0.32, "learning_rate": 8.83661481562058e-06, "loss": 0.9039, "step": 3140 }, { "epoch": 0.32, "learning_rate": 8.831175894702492e-06, "loss": 0.8958, "step": 3150 }, { "epoch": 0.33, "learning_rate": 8.825736973784402e-06, "loss": 0.9725, "step": 3160 }, { "epoch": 0.33, "learning_rate": 8.820298052866312e-06, "loss": 0.9183, "step": 3170 }, { "epoch": 0.33, "learning_rate": 8.814859131948223e-06, "loss": 0.9092, "step": 3180 }, { "epoch": 0.33, "learning_rate": 8.809420211030131e-06, "loss": 0.8689, "step": 3190 }, { "epoch": 0.33, "learning_rate": 8.803981290112043e-06, "loss": 0.8239, "step": 3200 }, { "epoch": 0.33, "learning_rate": 8.798542369193952e-06, "loss": 0.912, "step": 3210 }, { "epoch": 0.33, "learning_rate": 8.793103448275862e-06, "loss": 0.9324, "step": 3220 }, { "epoch": 0.33, "learning_rate": 8.787664527357774e-06, "loss": 0.9406, "step": 3230 }, { "epoch": 0.33, "learning_rate": 8.782225606439683e-06, "loss": 0.8657, "step": 3240 }, { "epoch": 0.34, "learning_rate": 8.776786685521593e-06, "loss": 0.9102, "step": 3250 }, { "epoch": 0.34, "learning_rate": 8.771347764603504e-06, "loss": 0.9147, "step": 3260 }, { "epoch": 0.34, "learning_rate": 8.765908843685414e-06, "loss": 0.9209, "step": 3270 }, { "epoch": 0.34, "learning_rate": 8.760469922767324e-06, "loss": 0.8654, "step": 3280 }, { "epoch": 0.34, "learning_rate": 8.755031001849234e-06, "loss": 0.9643, "step": 3290 }, { "epoch": 0.34, "learning_rate": 8.749592080931143e-06, "loss": 0.8726, "step": 3300 }, { "epoch": 0.34, "learning_rate": 8.744153160013053e-06, "loss": 0.8887, "step": 3310 }, { "epoch": 0.34, "learning_rate": 8.738714239094965e-06, "loss": 0.9306, "step": 3320 }, { "epoch": 0.34, "learning_rate": 8.733275318176874e-06, "loss": 0.8317, "step": 3330 }, { "epoch": 0.34, "learning_rate": 8.727836397258784e-06, "loss": 0.9198, "step": 3340 }, { "epoch": 0.35, "learning_rate": 8.722397476340696e-06, "loss": 0.8511, "step": 3350 }, { "epoch": 0.35, "learning_rate": 8.716958555422604e-06, "loss": 0.9394, "step": 3360 }, { "epoch": 0.35, "learning_rate": 8.711519634504515e-06, "loss": 0.9014, "step": 3370 }, { "epoch": 0.35, "learning_rate": 8.706080713586425e-06, "loss": 0.8943, "step": 3380 }, { "epoch": 0.35, "learning_rate": 8.700641792668335e-06, "loss": 0.9321, "step": 3390 }, { "epoch": 0.35, "learning_rate": 8.695202871750246e-06, "loss": 0.9983, "step": 3400 }, { "epoch": 0.35, "learning_rate": 8.689763950832156e-06, "loss": 0.8932, "step": 3410 }, { "epoch": 0.35, "learning_rate": 8.684325029914066e-06, "loss": 0.8859, "step": 3420 }, { "epoch": 0.35, "learning_rate": 8.678886108995977e-06, "loss": 0.9741, "step": 3430 }, { "epoch": 0.35, "learning_rate": 8.673447188077887e-06, "loss": 0.9678, "step": 3440 }, { "epoch": 0.36, "learning_rate": 8.668008267159796e-06, "loss": 0.8863, "step": 3450 }, { "epoch": 0.36, "learning_rate": 8.662569346241706e-06, "loss": 0.893, "step": 3460 }, { "epoch": 0.36, "learning_rate": 8.657130425323616e-06, "loss": 0.9949, "step": 3470 }, { "epoch": 0.36, "learning_rate": 8.651691504405527e-06, "loss": 0.9327, "step": 3480 }, { "epoch": 0.36, "learning_rate": 8.646252583487437e-06, "loss": 0.9409, "step": 3490 }, { "epoch": 0.36, "learning_rate": 8.640813662569347e-06, "loss": 0.8782, "step": 3500 }, { "epoch": 0.36, "learning_rate": 8.635374741651257e-06, "loss": 0.9952, "step": 3510 }, { "epoch": 0.36, "learning_rate": 8.629935820733168e-06, "loss": 0.9087, "step": 3520 }, { "epoch": 0.36, "learning_rate": 8.624496899815076e-06, "loss": 0.871, "step": 3530 }, { "epoch": 0.37, "learning_rate": 8.619057978896988e-06, "loss": 0.8489, "step": 3540 }, { "epoch": 0.37, "learning_rate": 8.613619057978897e-06, "loss": 0.8665, "step": 3550 }, { "epoch": 0.37, "learning_rate": 8.608180137060807e-06, "loss": 0.9827, "step": 3560 }, { "epoch": 0.37, "learning_rate": 8.602741216142719e-06, "loss": 0.9041, "step": 3570 }, { "epoch": 0.37, "learning_rate": 8.597302295224628e-06, "loss": 0.8981, "step": 3580 }, { "epoch": 0.37, "learning_rate": 8.591863374306538e-06, "loss": 0.9071, "step": 3590 }, { "epoch": 0.37, "learning_rate": 8.58642445338845e-06, "loss": 0.9052, "step": 3600 }, { "epoch": 0.37, "learning_rate": 8.58098553247036e-06, "loss": 0.8012, "step": 3610 }, { "epoch": 0.37, "learning_rate": 8.575546611552269e-06, "loss": 0.9814, "step": 3620 }, { "epoch": 0.37, "learning_rate": 8.570107690634179e-06, "loss": 0.9647, "step": 3630 }, { "epoch": 0.38, "learning_rate": 8.564668769716088e-06, "loss": 1.0055, "step": 3640 }, { "epoch": 0.38, "learning_rate": 8.559229848798e-06, "loss": 0.9685, "step": 3650 }, { "epoch": 0.38, "learning_rate": 8.55379092787991e-06, "loss": 0.9363, "step": 3660 }, { "epoch": 0.38, "learning_rate": 8.54835200696182e-06, "loss": 0.9785, "step": 3670 }, { "epoch": 0.38, "learning_rate": 8.542913086043729e-06, "loss": 0.9085, "step": 3680 }, { "epoch": 0.38, "learning_rate": 8.53747416512564e-06, "loss": 0.9522, "step": 3690 }, { "epoch": 0.38, "learning_rate": 8.53203524420755e-06, "loss": 0.9065, "step": 3700 }, { "epoch": 0.38, "learning_rate": 8.52659632328946e-06, "loss": 0.882, "step": 3710 }, { "epoch": 0.38, "learning_rate": 8.52115740237137e-06, "loss": 0.9243, "step": 3720 }, { "epoch": 0.38, "learning_rate": 8.51571848145328e-06, "loss": 0.9088, "step": 3730 }, { "epoch": 0.39, "learning_rate": 8.510279560535191e-06, "loss": 0.9072, "step": 3740 }, { "epoch": 0.39, "learning_rate": 8.5048406396171e-06, "loss": 1.017, "step": 3750 }, { "epoch": 0.39, "learning_rate": 8.49940171869901e-06, "loss": 0.9069, "step": 3760 }, { "epoch": 0.39, "learning_rate": 8.493962797780922e-06, "loss": 0.905, "step": 3770 }, { "epoch": 0.39, "learning_rate": 8.48852387686283e-06, "loss": 0.8551, "step": 3780 }, { "epoch": 0.39, "learning_rate": 8.483084955944741e-06, "loss": 0.8838, "step": 3790 }, { "epoch": 0.39, "learning_rate": 8.477646035026651e-06, "loss": 0.9411, "step": 3800 }, { "epoch": 0.39, "learning_rate": 8.472207114108561e-06, "loss": 0.889, "step": 3810 }, { "epoch": 0.39, "learning_rate": 8.466768193190472e-06, "loss": 0.9699, "step": 3820 }, { "epoch": 0.4, "learning_rate": 8.461329272272382e-06, "loss": 0.9363, "step": 3830 }, { "epoch": 0.4, "learning_rate": 8.455890351354292e-06, "loss": 0.92, "step": 3840 }, { "epoch": 0.4, "learning_rate": 8.450451430436202e-06, "loss": 0.9821, "step": 3850 }, { "epoch": 0.4, "learning_rate": 8.445012509518113e-06, "loss": 1.0008, "step": 3860 }, { "epoch": 0.4, "learning_rate": 8.439573588600023e-06, "loss": 0.7779, "step": 3870 }, { "epoch": 0.4, "learning_rate": 8.434134667681933e-06, "loss": 0.9229, "step": 3880 }, { "epoch": 0.4, "learning_rate": 8.428695746763842e-06, "loss": 0.8386, "step": 3890 }, { "epoch": 0.4, "learning_rate": 8.423256825845752e-06, "loss": 0.9353, "step": 3900 }, { "epoch": 0.4, "learning_rate": 8.417817904927664e-06, "loss": 0.9313, "step": 3910 }, { "epoch": 0.4, "learning_rate": 8.412378984009573e-06, "loss": 0.9342, "step": 3920 }, { "epoch": 0.41, "learning_rate": 8.406940063091483e-06, "loss": 0.9216, "step": 3930 }, { "epoch": 0.41, "learning_rate": 8.401501142173394e-06, "loss": 0.8671, "step": 3940 }, { "epoch": 0.41, "learning_rate": 8.396062221255303e-06, "loss": 0.9725, "step": 3950 }, { "epoch": 0.41, "learning_rate": 8.390623300337214e-06, "loss": 0.8487, "step": 3960 }, { "epoch": 0.41, "learning_rate": 8.385184379419124e-06, "loss": 0.7941, "step": 3970 }, { "epoch": 0.41, "learning_rate": 8.379745458501033e-06, "loss": 0.9374, "step": 3980 }, { "epoch": 0.41, "learning_rate": 8.374306537582945e-06, "loss": 0.9293, "step": 3990 }, { "epoch": 0.41, "learning_rate": 8.368867616664855e-06, "loss": 0.8974, "step": 4000 }, { "epoch": 0.41, "learning_rate": 8.363428695746764e-06, "loss": 0.8938, "step": 4010 }, { "epoch": 0.41, "learning_rate": 8.357989774828674e-06, "loss": 0.8779, "step": 4020 }, { "epoch": 0.42, "learning_rate": 8.352550853910586e-06, "loss": 0.8734, "step": 4030 }, { "epoch": 0.42, "learning_rate": 8.347111932992495e-06, "loss": 0.882, "step": 4040 }, { "epoch": 0.42, "learning_rate": 8.341673012074405e-06, "loss": 0.88, "step": 4050 }, { "epoch": 0.42, "learning_rate": 8.336234091156315e-06, "loss": 0.9591, "step": 4060 }, { "epoch": 0.42, "learning_rate": 8.330795170238225e-06, "loss": 0.9314, "step": 4070 }, { "epoch": 0.42, "learning_rate": 8.325356249320136e-06, "loss": 0.871, "step": 4080 }, { "epoch": 0.42, "learning_rate": 8.319917328402046e-06, "loss": 0.8621, "step": 4090 }, { "epoch": 0.42, "learning_rate": 8.314478407483956e-06, "loss": 0.8413, "step": 4100 }, { "epoch": 0.42, "learning_rate": 8.309039486565867e-06, "loss": 0.8655, "step": 4110 }, { "epoch": 0.43, "learning_rate": 8.303600565647775e-06, "loss": 0.9458, "step": 4120 }, { "epoch": 0.43, "learning_rate": 8.298161644729686e-06, "loss": 1.0266, "step": 4130 }, { "epoch": 0.43, "learning_rate": 8.292722723811596e-06, "loss": 1.0139, "step": 4140 }, { "epoch": 0.43, "learning_rate": 8.287283802893506e-06, "loss": 0.9082, "step": 4150 }, { "epoch": 0.43, "learning_rate": 8.281844881975417e-06, "loss": 0.8839, "step": 4160 }, { "epoch": 0.43, "learning_rate": 8.276405961057327e-06, "loss": 0.8856, "step": 4170 }, { "epoch": 0.43, "learning_rate": 8.270967040139237e-06, "loss": 0.9612, "step": 4180 }, { "epoch": 0.43, "learning_rate": 8.265528119221147e-06, "loss": 0.8982, "step": 4190 }, { "epoch": 0.43, "learning_rate": 8.260089198303058e-06, "loss": 0.8794, "step": 4200 }, { "epoch": 0.43, "learning_rate": 8.254650277384968e-06, "loss": 0.9062, "step": 4210 }, { "epoch": 0.44, "learning_rate": 8.249211356466878e-06, "loss": 1.0483, "step": 4220 }, { "epoch": 0.44, "learning_rate": 8.243772435548787e-06, "loss": 0.9079, "step": 4230 }, { "epoch": 0.44, "learning_rate": 8.238333514630697e-06, "loss": 0.9127, "step": 4240 }, { "epoch": 0.44, "learning_rate": 8.232894593712609e-06, "loss": 0.9079, "step": 4250 }, { "epoch": 0.44, "learning_rate": 8.227455672794518e-06, "loss": 0.9584, "step": 4260 }, { "epoch": 0.44, "learning_rate": 8.222016751876428e-06, "loss": 0.8584, "step": 4270 }, { "epoch": 0.44, "learning_rate": 8.21657783095834e-06, "loss": 0.9107, "step": 4280 }, { "epoch": 0.44, "learning_rate": 8.211138910040248e-06, "loss": 0.9358, "step": 4290 }, { "epoch": 0.44, "learning_rate": 8.205699989122159e-06, "loss": 0.9942, "step": 4300 }, { "epoch": 0.44, "learning_rate": 8.200261068204069e-06, "loss": 0.9147, "step": 4310 }, { "epoch": 0.45, "learning_rate": 8.194822147285979e-06, "loss": 0.8893, "step": 4320 }, { "epoch": 0.45, "learning_rate": 8.18938322636789e-06, "loss": 0.8602, "step": 4330 }, { "epoch": 0.45, "learning_rate": 8.1839443054498e-06, "loss": 0.8962, "step": 4340 }, { "epoch": 0.45, "learning_rate": 8.17850538453171e-06, "loss": 0.8843, "step": 4350 }, { "epoch": 0.45, "learning_rate": 8.17306646361362e-06, "loss": 0.8907, "step": 4360 }, { "epoch": 0.45, "learning_rate": 8.16762754269553e-06, "loss": 0.8685, "step": 4370 }, { "epoch": 0.45, "learning_rate": 8.16218862177744e-06, "loss": 0.8499, "step": 4380 }, { "epoch": 0.45, "learning_rate": 8.15674970085935e-06, "loss": 0.9896, "step": 4390 }, { "epoch": 0.45, "learning_rate": 8.15131077994126e-06, "loss": 0.9117, "step": 4400 }, { "epoch": 0.45, "learning_rate": 8.14587185902317e-06, "loss": 0.8675, "step": 4410 }, { "epoch": 0.46, "learning_rate": 8.140432938105081e-06, "loss": 0.8865, "step": 4420 }, { "epoch": 0.46, "learning_rate": 8.13499401718699e-06, "loss": 0.8735, "step": 4430 }, { "epoch": 0.46, "learning_rate": 8.1295550962689e-06, "loss": 0.8923, "step": 4440 }, { "epoch": 0.46, "learning_rate": 8.124116175350812e-06, "loss": 0.8156, "step": 4450 }, { "epoch": 0.46, "learning_rate": 8.11867725443272e-06, "loss": 0.9159, "step": 4460 }, { "epoch": 0.46, "learning_rate": 8.113238333514632e-06, "loss": 1.0465, "step": 4470 }, { "epoch": 0.46, "learning_rate": 8.107799412596541e-06, "loss": 0.9716, "step": 4480 }, { "epoch": 0.46, "learning_rate": 8.102360491678451e-06, "loss": 0.859, "step": 4490 }, { "epoch": 0.46, "learning_rate": 8.096921570760362e-06, "loss": 0.8693, "step": 4500 }, { "epoch": 0.47, "learning_rate": 8.091482649842272e-06, "loss": 0.8514, "step": 4510 }, { "epoch": 0.47, "learning_rate": 8.086043728924182e-06, "loss": 0.9315, "step": 4520 }, { "epoch": 0.47, "learning_rate": 8.080604808006092e-06, "loss": 1.0026, "step": 4530 }, { "epoch": 0.47, "learning_rate": 8.075165887088003e-06, "loss": 1.0319, "step": 4540 }, { "epoch": 0.47, "learning_rate": 8.069726966169913e-06, "loss": 0.8855, "step": 4550 }, { "epoch": 0.47, "learning_rate": 8.064288045251823e-06, "loss": 0.9006, "step": 4560 }, { "epoch": 0.47, "learning_rate": 8.058849124333732e-06, "loss": 1.0628, "step": 4570 }, { "epoch": 0.47, "learning_rate": 8.053410203415642e-06, "loss": 0.9277, "step": 4580 }, { "epoch": 0.47, "learning_rate": 8.047971282497554e-06, "loss": 0.9067, "step": 4590 }, { "epoch": 0.47, "learning_rate": 8.042532361579463e-06, "loss": 0.8537, "step": 4600 }, { "epoch": 0.48, "learning_rate": 8.037093440661373e-06, "loss": 0.9993, "step": 4610 }, { "epoch": 0.48, "learning_rate": 8.031654519743285e-06, "loss": 1.0009, "step": 4620 }, { "epoch": 0.48, "learning_rate": 8.026215598825193e-06, "loss": 0.832, "step": 4630 }, { "epoch": 0.48, "learning_rate": 8.020776677907104e-06, "loss": 0.9099, "step": 4640 }, { "epoch": 0.48, "learning_rate": 8.015337756989014e-06, "loss": 1.0442, "step": 4650 }, { "epoch": 0.48, "learning_rate": 8.009898836070924e-06, "loss": 0.9313, "step": 4660 }, { "epoch": 0.48, "learning_rate": 8.004459915152835e-06, "loss": 0.8611, "step": 4670 }, { "epoch": 0.48, "learning_rate": 7.999020994234745e-06, "loss": 0.9372, "step": 4680 }, { "epoch": 0.48, "learning_rate": 7.993582073316654e-06, "loss": 0.8676, "step": 4690 }, { "epoch": 0.48, "learning_rate": 7.988143152398566e-06, "loss": 0.8742, "step": 4700 }, { "epoch": 0.49, "learning_rate": 7.982704231480474e-06, "loss": 0.9689, "step": 4710 }, { "epoch": 0.49, "learning_rate": 7.977265310562385e-06, "loss": 0.8652, "step": 4720 }, { "epoch": 0.49, "learning_rate": 7.971826389644295e-06, "loss": 0.9451, "step": 4730 }, { "epoch": 0.49, "learning_rate": 7.966387468726205e-06, "loss": 0.9227, "step": 4740 }, { "epoch": 0.49, "learning_rate": 7.960948547808115e-06, "loss": 0.9032, "step": 4750 }, { "epoch": 0.49, "learning_rate": 7.955509626890026e-06, "loss": 0.8749, "step": 4760 }, { "epoch": 0.49, "learning_rate": 7.950070705971936e-06, "loss": 0.8649, "step": 4770 }, { "epoch": 0.49, "learning_rate": 7.944631785053846e-06, "loss": 0.9872, "step": 4780 }, { "epoch": 0.49, "learning_rate": 7.939192864135757e-06, "loss": 0.9091, "step": 4790 }, { "epoch": 0.5, "learning_rate": 7.933753943217665e-06, "loss": 0.8575, "step": 4800 }, { "epoch": 0.5, "learning_rate": 7.928315022299577e-06, "loss": 0.9621, "step": 4810 }, { "epoch": 0.5, "learning_rate": 7.922876101381486e-06, "loss": 0.872, "step": 4820 }, { "epoch": 0.5, "learning_rate": 7.917437180463396e-06, "loss": 1.0035, "step": 4830 }, { "epoch": 0.5, "learning_rate": 7.911998259545307e-06, "loss": 0.91, "step": 4840 }, { "epoch": 0.5, "learning_rate": 7.906559338627217e-06, "loss": 0.8972, "step": 4850 }, { "epoch": 0.5, "learning_rate": 7.901120417709127e-06, "loss": 0.8824, "step": 4860 }, { "epoch": 0.5, "learning_rate": 7.895681496791038e-06, "loss": 0.8744, "step": 4870 }, { "epoch": 0.5, "learning_rate": 7.890242575872946e-06, "loss": 0.8643, "step": 4880 }, { "epoch": 0.5, "learning_rate": 7.884803654954858e-06, "loss": 0.9442, "step": 4890 }, { "epoch": 0.51, "learning_rate": 7.879364734036768e-06, "loss": 1.0287, "step": 4900 }, { "epoch": 0.51, "learning_rate": 7.873925813118677e-06, "loss": 0.9974, "step": 4910 }, { "epoch": 0.51, "learning_rate": 7.868486892200589e-06, "loss": 0.9217, "step": 4920 }, { "epoch": 0.51, "learning_rate": 7.863047971282499e-06, "loss": 0.9269, "step": 4930 }, { "epoch": 0.51, "learning_rate": 7.857609050364408e-06, "loss": 0.9418, "step": 4940 }, { "epoch": 0.51, "learning_rate": 7.852170129446318e-06, "loss": 0.9268, "step": 4950 }, { "epoch": 0.51, "learning_rate": 7.84673120852823e-06, "loss": 1.019, "step": 4960 }, { "epoch": 0.51, "learning_rate": 7.841292287610138e-06, "loss": 0.8528, "step": 4970 }, { "epoch": 0.51, "learning_rate": 7.835853366692049e-06, "loss": 1.0123, "step": 4980 }, { "epoch": 0.51, "learning_rate": 7.830414445773959e-06, "loss": 0.8738, "step": 4990 }, { "epoch": 0.52, "learning_rate": 7.824975524855869e-06, "loss": 0.9835, "step": 5000 }, { "epoch": 0.52, "learning_rate": 7.81953660393778e-06, "loss": 0.8421, "step": 5010 }, { "epoch": 0.52, "learning_rate": 7.81409768301969e-06, "loss": 0.8968, "step": 5020 }, { "epoch": 0.52, "learning_rate": 7.8086587621016e-06, "loss": 0.8893, "step": 5030 }, { "epoch": 0.52, "learning_rate": 7.803219841183511e-06, "loss": 0.9422, "step": 5040 }, { "epoch": 0.52, "learning_rate": 7.797780920265419e-06, "loss": 0.9306, "step": 5050 }, { "epoch": 0.52, "learning_rate": 7.79234199934733e-06, "loss": 0.8557, "step": 5060 }, { "epoch": 0.52, "learning_rate": 7.78690307842924e-06, "loss": 0.9081, "step": 5070 }, { "epoch": 0.52, "learning_rate": 7.78146415751115e-06, "loss": 0.8734, "step": 5080 }, { "epoch": 0.53, "learning_rate": 7.776025236593061e-06, "loss": 0.8906, "step": 5090 }, { "epoch": 0.53, "learning_rate": 7.770586315674971e-06, "loss": 0.8805, "step": 5100 }, { "epoch": 0.53, "learning_rate": 7.765147394756881e-06, "loss": 0.9945, "step": 5110 }, { "epoch": 0.53, "learning_rate": 7.75970847383879e-06, "loss": 0.8724, "step": 5120 }, { "epoch": 0.53, "learning_rate": 7.754269552920702e-06, "loss": 0.9495, "step": 5130 }, { "epoch": 0.53, "learning_rate": 7.74883063200261e-06, "loss": 0.9403, "step": 5140 }, { "epoch": 0.53, "learning_rate": 7.743391711084522e-06, "loss": 0.9006, "step": 5150 }, { "epoch": 0.53, "learning_rate": 7.737952790166431e-06, "loss": 0.8878, "step": 5160 }, { "epoch": 0.53, "learning_rate": 7.732513869248341e-06, "loss": 0.9871, "step": 5170 }, { "epoch": 0.53, "learning_rate": 7.727074948330252e-06, "loss": 0.8615, "step": 5180 }, { "epoch": 0.54, "learning_rate": 7.721636027412162e-06, "loss": 0.9406, "step": 5190 }, { "epoch": 0.54, "learning_rate": 7.716197106494072e-06, "loss": 0.8913, "step": 5200 }, { "epoch": 0.54, "learning_rate": 7.710758185575983e-06, "loss": 0.966, "step": 5210 }, { "epoch": 0.54, "learning_rate": 7.705319264657891e-06, "loss": 1.0135, "step": 5220 }, { "epoch": 0.54, "learning_rate": 7.699880343739803e-06, "loss": 0.8437, "step": 5230 }, { "epoch": 0.54, "learning_rate": 7.694441422821713e-06, "loss": 1.0414, "step": 5240 }, { "epoch": 0.54, "learning_rate": 7.689002501903622e-06, "loss": 0.9284, "step": 5250 }, { "epoch": 0.54, "learning_rate": 7.683563580985534e-06, "loss": 0.8383, "step": 5260 }, { "epoch": 0.54, "learning_rate": 7.678124660067444e-06, "loss": 0.8854, "step": 5270 }, { "epoch": 0.54, "learning_rate": 7.672685739149353e-06, "loss": 0.8536, "step": 5280 }, { "epoch": 0.55, "learning_rate": 7.667246818231263e-06, "loss": 0.873, "step": 5290 }, { "epoch": 0.55, "learning_rate": 7.661807897313175e-06, "loss": 0.9999, "step": 5300 }, { "epoch": 0.55, "learning_rate": 7.656368976395084e-06, "loss": 0.9058, "step": 5310 }, { "epoch": 0.55, "learning_rate": 7.650930055476994e-06, "loss": 0.8779, "step": 5320 }, { "epoch": 0.55, "learning_rate": 7.645491134558904e-06, "loss": 0.8563, "step": 5330 }, { "epoch": 0.55, "learning_rate": 7.640052213640814e-06, "loss": 0.9523, "step": 5340 }, { "epoch": 0.55, "learning_rate": 7.634613292722725e-06, "loss": 0.9108, "step": 5350 }, { "epoch": 0.55, "learning_rate": 7.629174371804634e-06, "loss": 0.9046, "step": 5360 }, { "epoch": 0.55, "learning_rate": 7.6237354508865445e-06, "loss": 0.9188, "step": 5370 }, { "epoch": 0.56, "learning_rate": 7.618296529968455e-06, "loss": 0.9603, "step": 5380 }, { "epoch": 0.56, "learning_rate": 7.612857609050365e-06, "loss": 0.9303, "step": 5390 }, { "epoch": 0.56, "learning_rate": 7.6074186881322754e-06, "loss": 0.9168, "step": 5400 }, { "epoch": 0.56, "learning_rate": 7.601979767214185e-06, "loss": 0.9629, "step": 5410 }, { "epoch": 0.56, "learning_rate": 7.596540846296096e-06, "loss": 0.923, "step": 5420 }, { "epoch": 0.56, "learning_rate": 7.591101925378006e-06, "loss": 0.916, "step": 5430 }, { "epoch": 0.56, "learning_rate": 7.585663004459915e-06, "loss": 0.8986, "step": 5440 }, { "epoch": 0.56, "learning_rate": 7.580224083541826e-06, "loss": 0.9505, "step": 5450 }, { "epoch": 0.56, "learning_rate": 7.574785162623736e-06, "loss": 0.8884, "step": 5460 }, { "epoch": 0.56, "learning_rate": 7.569346241705646e-06, "loss": 0.9566, "step": 5470 }, { "epoch": 0.57, "learning_rate": 7.563907320787557e-06, "loss": 0.861, "step": 5480 }, { "epoch": 0.57, "learning_rate": 7.5584683998694666e-06, "loss": 0.9182, "step": 5490 }, { "epoch": 0.57, "learning_rate": 7.553029478951377e-06, "loss": 0.903, "step": 5500 }, { "epoch": 0.57, "learning_rate": 7.547590558033286e-06, "loss": 0.9609, "step": 5510 }, { "epoch": 0.57, "learning_rate": 7.542151637115197e-06, "loss": 0.9164, "step": 5520 }, { "epoch": 0.57, "learning_rate": 7.536712716197107e-06, "loss": 0.93, "step": 5530 }, { "epoch": 0.57, "learning_rate": 7.531273795279017e-06, "loss": 0.9707, "step": 5540 }, { "epoch": 0.57, "learning_rate": 7.525834874360928e-06, "loss": 0.9872, "step": 5550 }, { "epoch": 0.57, "learning_rate": 7.520395953442837e-06, "loss": 0.9352, "step": 5560 }, { "epoch": 0.57, "learning_rate": 7.514957032524748e-06, "loss": 0.9391, "step": 5570 }, { "epoch": 0.58, "learning_rate": 7.509518111606658e-06, "loss": 0.8276, "step": 5580 }, { "epoch": 0.58, "learning_rate": 7.504079190688568e-06, "loss": 0.9967, "step": 5590 }, { "epoch": 0.58, "learning_rate": 7.498640269770479e-06, "loss": 0.8528, "step": 5600 }, { "epoch": 0.58, "learning_rate": 7.493201348852388e-06, "loss": 0.8907, "step": 5610 }, { "epoch": 0.58, "learning_rate": 7.487762427934298e-06, "loss": 0.9797, "step": 5620 }, { "epoch": 0.58, "learning_rate": 7.482323507016208e-06, "loss": 0.8878, "step": 5630 }, { "epoch": 0.58, "learning_rate": 7.476884586098119e-06, "loss": 0.8941, "step": 5640 }, { "epoch": 0.58, "learning_rate": 7.471445665180029e-06, "loss": 0.9322, "step": 5650 }, { "epoch": 0.58, "learning_rate": 7.466006744261939e-06, "loss": 0.8805, "step": 5660 }, { "epoch": 0.58, "learning_rate": 7.46056782334385e-06, "loss": 0.8254, "step": 5670 }, { "epoch": 0.59, "learning_rate": 7.455128902425759e-06, "loss": 0.9692, "step": 5680 }, { "epoch": 0.59, "learning_rate": 7.449689981507669e-06, "loss": 0.8475, "step": 5690 }, { "epoch": 0.59, "learning_rate": 7.44425106058958e-06, "loss": 0.9227, "step": 5700 }, { "epoch": 0.59, "learning_rate": 7.4388121396714895e-06, "loss": 0.9216, "step": 5710 }, { "epoch": 0.59, "learning_rate": 7.4333732187534e-06, "loss": 0.9737, "step": 5720 }, { "epoch": 0.59, "learning_rate": 7.42793429783531e-06, "loss": 0.9357, "step": 5730 }, { "epoch": 0.59, "learning_rate": 7.4224953769172205e-06, "loss": 0.8771, "step": 5740 }, { "epoch": 0.59, "learning_rate": 7.417056455999131e-06, "loss": 0.9442, "step": 5750 }, { "epoch": 0.59, "learning_rate": 7.411617535081041e-06, "loss": 0.8341, "step": 5760 }, { "epoch": 0.6, "learning_rate": 7.406178614162951e-06, "loss": 0.8047, "step": 5770 }, { "epoch": 0.6, "learning_rate": 7.40073969324486e-06, "loss": 0.8865, "step": 5780 }, { "epoch": 0.6, "learning_rate": 7.395300772326771e-06, "loss": 0.9274, "step": 5790 }, { "epoch": 0.6, "learning_rate": 7.389861851408681e-06, "loss": 0.9028, "step": 5800 }, { "epoch": 0.6, "learning_rate": 7.384422930490591e-06, "loss": 0.8728, "step": 5810 }, { "epoch": 0.6, "learning_rate": 7.378984009572502e-06, "loss": 0.8905, "step": 5820 }, { "epoch": 0.6, "learning_rate": 7.373545088654412e-06, "loss": 0.9257, "step": 5830 }, { "epoch": 0.6, "learning_rate": 7.368106167736322e-06, "loss": 0.9113, "step": 5840 }, { "epoch": 0.6, "learning_rate": 7.362667246818231e-06, "loss": 0.8313, "step": 5850 }, { "epoch": 0.6, "learning_rate": 7.357228325900142e-06, "loss": 0.9787, "step": 5860 }, { "epoch": 0.61, "learning_rate": 7.351789404982052e-06, "loss": 0.8983, "step": 5870 }, { "epoch": 0.61, "learning_rate": 7.346350484063962e-06, "loss": 0.9152, "step": 5880 }, { "epoch": 0.61, "learning_rate": 7.340911563145873e-06, "loss": 0.9679, "step": 5890 }, { "epoch": 0.61, "learning_rate": 7.335472642227782e-06, "loss": 0.9205, "step": 5900 }, { "epoch": 0.61, "learning_rate": 7.330033721309693e-06, "loss": 0.8937, "step": 5910 }, { "epoch": 0.61, "learning_rate": 7.3245948003916036e-06, "loss": 0.8767, "step": 5920 }, { "epoch": 0.61, "learning_rate": 7.319155879473513e-06, "loss": 0.9649, "step": 5930 }, { "epoch": 0.61, "learning_rate": 7.313716958555424e-06, "loss": 0.9091, "step": 5940 }, { "epoch": 0.61, "learning_rate": 7.308278037637333e-06, "loss": 0.8537, "step": 5950 }, { "epoch": 0.61, "learning_rate": 7.302839116719243e-06, "loss": 0.8862, "step": 5960 }, { "epoch": 0.62, "learning_rate": 7.297400195801153e-06, "loss": 0.9046, "step": 5970 }, { "epoch": 0.62, "learning_rate": 7.291961274883064e-06, "loss": 0.9745, "step": 5980 }, { "epoch": 0.62, "learning_rate": 7.286522353964974e-06, "loss": 0.9326, "step": 5990 }, { "epoch": 0.62, "learning_rate": 7.281083433046884e-06, "loss": 0.8505, "step": 6000 }, { "epoch": 0.62, "learning_rate": 7.275644512128795e-06, "loss": 0.8275, "step": 6010 }, { "epoch": 0.62, "learning_rate": 7.270205591210704e-06, "loss": 0.9607, "step": 6020 }, { "epoch": 0.62, "learning_rate": 7.264766670292614e-06, "loss": 0.9749, "step": 6030 }, { "epoch": 0.62, "learning_rate": 7.259327749374525e-06, "loss": 0.9795, "step": 6040 }, { "epoch": 0.62, "learning_rate": 7.2538888284564345e-06, "loss": 0.8252, "step": 6050 }, { "epoch": 0.63, "learning_rate": 7.248449907538345e-06, "loss": 0.8637, "step": 6060 }, { "epoch": 0.63, "learning_rate": 7.243010986620255e-06, "loss": 0.9193, "step": 6070 }, { "epoch": 0.63, "learning_rate": 7.2375720657021655e-06, "loss": 0.9216, "step": 6080 }, { "epoch": 0.63, "learning_rate": 7.232133144784076e-06, "loss": 0.8906, "step": 6090 }, { "epoch": 0.63, "learning_rate": 7.226694223865985e-06, "loss": 0.8998, "step": 6100 }, { "epoch": 0.63, "learning_rate": 7.2212553029478964e-06, "loss": 0.9018, "step": 6110 }, { "epoch": 0.63, "learning_rate": 7.215816382029805e-06, "loss": 0.8381, "step": 6120 }, { "epoch": 0.63, "learning_rate": 7.210377461111716e-06, "loss": 0.9336, "step": 6130 }, { "epoch": 0.63, "learning_rate": 7.2049385401936265e-06, "loss": 0.9667, "step": 6140 }, { "epoch": 0.63, "learning_rate": 7.199499619275536e-06, "loss": 0.9724, "step": 6150 }, { "epoch": 0.64, "learning_rate": 7.194060698357447e-06, "loss": 0.8622, "step": 6160 }, { "epoch": 0.64, "learning_rate": 7.188621777439357e-06, "loss": 0.8128, "step": 6170 }, { "epoch": 0.64, "learning_rate": 7.183182856521267e-06, "loss": 0.8464, "step": 6180 }, { "epoch": 0.64, "learning_rate": 7.177743935603176e-06, "loss": 0.9024, "step": 6190 }, { "epoch": 0.64, "learning_rate": 7.172305014685087e-06, "loss": 0.9076, "step": 6200 }, { "epoch": 0.64, "learning_rate": 7.166866093766997e-06, "loss": 0.8348, "step": 6210 }, { "epoch": 0.64, "learning_rate": 7.161427172848907e-06, "loss": 0.9037, "step": 6220 }, { "epoch": 0.64, "learning_rate": 7.155988251930818e-06, "loss": 0.9079, "step": 6230 }, { "epoch": 0.64, "learning_rate": 7.150549331012727e-06, "loss": 1.0029, "step": 6240 }, { "epoch": 0.64, "learning_rate": 7.145110410094638e-06, "loss": 0.8917, "step": 6250 }, { "epoch": 0.65, "learning_rate": 7.139671489176549e-06, "loss": 1.0127, "step": 6260 }, { "epoch": 0.65, "learning_rate": 7.1342325682584575e-06, "loss": 0.8671, "step": 6270 }, { "epoch": 0.65, "learning_rate": 7.128793647340368e-06, "loss": 0.869, "step": 6280 }, { "epoch": 0.65, "learning_rate": 7.123354726422278e-06, "loss": 0.9672, "step": 6290 }, { "epoch": 0.65, "learning_rate": 7.1179158055041884e-06, "loss": 0.8272, "step": 6300 }, { "epoch": 0.65, "learning_rate": 7.112476884586099e-06, "loss": 0.9054, "step": 6310 }, { "epoch": 0.65, "learning_rate": 7.107037963668009e-06, "loss": 0.9043, "step": 6320 }, { "epoch": 0.65, "learning_rate": 7.101599042749919e-06, "loss": 0.9264, "step": 6330 }, { "epoch": 0.65, "learning_rate": 7.096160121831829e-06, "loss": 0.9263, "step": 6340 }, { "epoch": 0.66, "learning_rate": 7.09072120091374e-06, "loss": 0.9282, "step": 6350 }, { "epoch": 0.66, "learning_rate": 7.08528227999565e-06, "loss": 0.8974, "step": 6360 }, { "epoch": 0.66, "learning_rate": 7.079843359077559e-06, "loss": 0.8773, "step": 6370 }, { "epoch": 0.66, "learning_rate": 7.07440443815947e-06, "loss": 0.8831, "step": 6380 }, { "epoch": 0.66, "learning_rate": 7.0689655172413796e-06, "loss": 0.9244, "step": 6390 }, { "epoch": 0.66, "learning_rate": 7.06352659632329e-06, "loss": 0.8432, "step": 6400 }, { "epoch": 0.66, "learning_rate": 7.0580876754052e-06, "loss": 0.8935, "step": 6410 }, { "epoch": 0.66, "learning_rate": 7.0526487544871105e-06, "loss": 0.9204, "step": 6420 }, { "epoch": 0.66, "learning_rate": 7.047209833569021e-06, "loss": 0.9142, "step": 6430 }, { "epoch": 0.66, "learning_rate": 7.04177091265093e-06, "loss": 0.8909, "step": 6440 }, { "epoch": 0.67, "learning_rate": 7.036331991732841e-06, "loss": 0.9273, "step": 6450 }, { "epoch": 0.67, "learning_rate": 7.03089307081475e-06, "loss": 0.9322, "step": 6460 }, { "epoch": 0.67, "learning_rate": 7.025454149896661e-06, "loss": 0.8865, "step": 6470 }, { "epoch": 0.67, "learning_rate": 7.0200152289785715e-06, "loss": 0.8738, "step": 6480 }, { "epoch": 0.67, "learning_rate": 7.014576308060481e-06, "loss": 0.8553, "step": 6490 }, { "epoch": 0.67, "learning_rate": 7.009137387142392e-06, "loss": 0.8481, "step": 6500 }, { "epoch": 0.67, "learning_rate": 7.003698466224302e-06, "loss": 0.8587, "step": 6510 }, { "epoch": 0.67, "learning_rate": 6.998259545306212e-06, "loss": 0.8784, "step": 6520 }, { "epoch": 0.67, "learning_rate": 6.992820624388123e-06, "loss": 0.885, "step": 6530 }, { "epoch": 0.67, "learning_rate": 6.987381703470032e-06, "loss": 0.7942, "step": 6540 }, { "epoch": 0.68, "learning_rate": 6.981942782551942e-06, "loss": 0.9477, "step": 6550 }, { "epoch": 0.68, "learning_rate": 6.976503861633852e-06, "loss": 0.8859, "step": 6560 }, { "epoch": 0.68, "learning_rate": 6.971064940715763e-06, "loss": 0.9291, "step": 6570 }, { "epoch": 0.68, "learning_rate": 6.965626019797672e-06, "loss": 0.8977, "step": 6580 }, { "epoch": 0.68, "learning_rate": 6.960187098879583e-06, "loss": 0.9804, "step": 6590 }, { "epoch": 0.68, "learning_rate": 6.954748177961494e-06, "loss": 0.9152, "step": 6600 }, { "epoch": 0.68, "learning_rate": 6.9493092570434025e-06, "loss": 0.8102, "step": 6610 }, { "epoch": 0.68, "learning_rate": 6.943870336125313e-06, "loss": 0.9699, "step": 6620 }, { "epoch": 0.68, "learning_rate": 6.938431415207223e-06, "loss": 0.9176, "step": 6630 }, { "epoch": 0.69, "learning_rate": 6.9329924942891335e-06, "loss": 0.9258, "step": 6640 }, { "epoch": 0.69, "learning_rate": 6.927553573371044e-06, "loss": 0.9005, "step": 6650 }, { "epoch": 0.69, "learning_rate": 6.922114652452954e-06, "loss": 0.907, "step": 6660 }, { "epoch": 0.69, "learning_rate": 6.916675731534864e-06, "loss": 0.9998, "step": 6670 }, { "epoch": 0.69, "learning_rate": 6.911236810616774e-06, "loss": 0.9125, "step": 6680 }, { "epoch": 0.69, "learning_rate": 6.905797889698685e-06, "loss": 0.8303, "step": 6690 }, { "epoch": 0.69, "learning_rate": 6.900358968780595e-06, "loss": 0.9333, "step": 6700 }, { "epoch": 0.69, "learning_rate": 6.894920047862504e-06, "loss": 0.9343, "step": 6710 }, { "epoch": 0.69, "learning_rate": 6.889481126944415e-06, "loss": 0.9409, "step": 6720 }, { "epoch": 0.69, "learning_rate": 6.884042206026325e-06, "loss": 1.0103, "step": 6730 }, { "epoch": 0.7, "learning_rate": 6.878603285108235e-06, "loss": 0.8811, "step": 6740 }, { "epoch": 0.7, "learning_rate": 6.873164364190146e-06, "loss": 0.9705, "step": 6750 }, { "epoch": 0.7, "learning_rate": 6.8677254432720555e-06, "loss": 0.9507, "step": 6760 }, { "epoch": 0.7, "learning_rate": 6.862286522353966e-06, "loss": 0.9089, "step": 6770 }, { "epoch": 0.7, "learning_rate": 6.856847601435875e-06, "loss": 0.8332, "step": 6780 }, { "epoch": 0.7, "learning_rate": 6.851408680517786e-06, "loss": 0.8344, "step": 6790 }, { "epoch": 0.7, "learning_rate": 6.845969759599695e-06, "loss": 0.9324, "step": 6800 }, { "epoch": 0.7, "learning_rate": 6.840530838681606e-06, "loss": 0.8638, "step": 6810 }, { "epoch": 0.7, "learning_rate": 6.8350919177635166e-06, "loss": 0.9077, "step": 6820 }, { "epoch": 0.7, "learning_rate": 6.829652996845426e-06, "loss": 0.9208, "step": 6830 }, { "epoch": 0.71, "learning_rate": 6.824214075927337e-06, "loss": 0.8446, "step": 6840 }, { "epoch": 0.71, "learning_rate": 6.818775155009247e-06, "loss": 0.9981, "step": 6850 }, { "epoch": 0.71, "learning_rate": 6.813336234091157e-06, "loss": 0.8762, "step": 6860 }, { "epoch": 0.71, "learning_rate": 6.807897313173068e-06, "loss": 0.9507, "step": 6870 }, { "epoch": 0.71, "learning_rate": 6.802458392254977e-06, "loss": 0.837, "step": 6880 }, { "epoch": 0.71, "learning_rate": 6.797019471336887e-06, "loss": 0.8824, "step": 6890 }, { "epoch": 0.71, "learning_rate": 6.791580550418797e-06, "loss": 0.9189, "step": 6900 }, { "epoch": 0.71, "learning_rate": 6.786141629500708e-06, "loss": 0.8384, "step": 6910 }, { "epoch": 0.71, "learning_rate": 6.780702708582618e-06, "loss": 0.9678, "step": 6920 }, { "epoch": 0.71, "learning_rate": 6.775263787664528e-06, "loss": 0.9269, "step": 6930 }, { "epoch": 0.72, "learning_rate": 6.769824866746439e-06, "loss": 0.8444, "step": 6940 }, { "epoch": 0.72, "learning_rate": 6.7643859458283475e-06, "loss": 0.8508, "step": 6950 }, { "epoch": 0.72, "learning_rate": 6.758947024910258e-06, "loss": 0.9537, "step": 6960 }, { "epoch": 0.72, "learning_rate": 6.753508103992169e-06, "loss": 0.9155, "step": 6970 }, { "epoch": 0.72, "learning_rate": 6.7480691830740785e-06, "loss": 0.879, "step": 6980 }, { "epoch": 0.72, "learning_rate": 6.742630262155989e-06, "loss": 1.0105, "step": 6990 }, { "epoch": 0.72, "learning_rate": 6.737191341237899e-06, "loss": 0.9191, "step": 7000 }, { "epoch": 0.72, "learning_rate": 6.731752420319809e-06, "loss": 0.8184, "step": 7010 }, { "epoch": 0.72, "learning_rate": 6.726313499401718e-06, "loss": 0.8852, "step": 7020 }, { "epoch": 0.73, "learning_rate": 6.720874578483629e-06, "loss": 0.8979, "step": 7030 }, { "epoch": 0.73, "learning_rate": 6.71543565756554e-06, "loss": 0.8606, "step": 7040 }, { "epoch": 0.73, "learning_rate": 6.709996736647449e-06, "loss": 0.8855, "step": 7050 }, { "epoch": 0.73, "learning_rate": 6.70455781572936e-06, "loss": 0.8519, "step": 7060 }, { "epoch": 0.73, "learning_rate": 6.69911889481127e-06, "loss": 0.9772, "step": 7070 }, { "epoch": 0.73, "learning_rate": 6.69367997389318e-06, "loss": 0.9361, "step": 7080 }, { "epoch": 0.73, "learning_rate": 6.688241052975091e-06, "loss": 0.9194, "step": 7090 }, { "epoch": 0.73, "learning_rate": 6.6828021320570006e-06, "loss": 0.9202, "step": 7100 }, { "epoch": 0.73, "learning_rate": 6.677363211138911e-06, "loss": 0.9361, "step": 7110 }, { "epoch": 0.73, "learning_rate": 6.67192429022082e-06, "loss": 0.8968, "step": 7120 }, { "epoch": 0.74, "learning_rate": 6.666485369302731e-06, "loss": 0.8848, "step": 7130 }, { "epoch": 0.74, "learning_rate": 6.661046448384641e-06, "loss": 0.9004, "step": 7140 }, { "epoch": 0.74, "learning_rate": 6.655607527466551e-06, "loss": 0.8883, "step": 7150 }, { "epoch": 0.74, "learning_rate": 6.650168606548462e-06, "loss": 0.8807, "step": 7160 }, { "epoch": 0.74, "learning_rate": 6.644729685630371e-06, "loss": 0.8439, "step": 7170 }, { "epoch": 0.74, "learning_rate": 6.639290764712282e-06, "loss": 0.8581, "step": 7180 }, { "epoch": 0.74, "learning_rate": 6.633851843794191e-06, "loss": 0.9348, "step": 7190 }, { "epoch": 0.74, "learning_rate": 6.6284129228761014e-06, "loss": 0.9294, "step": 7200 }, { "epoch": 0.74, "learning_rate": 6.622974001958012e-06, "loss": 0.8992, "step": 7210 }, { "epoch": 0.74, "learning_rate": 6.617535081039922e-06, "loss": 0.8276, "step": 7220 }, { "epoch": 0.75, "learning_rate": 6.612096160121832e-06, "loss": 0.8895, "step": 7230 }, { "epoch": 0.75, "learning_rate": 6.606657239203742e-06, "loss": 0.858, "step": 7240 }, { "epoch": 0.75, "learning_rate": 6.601218318285653e-06, "loss": 0.8357, "step": 7250 }, { "epoch": 0.75, "learning_rate": 6.595779397367563e-06, "loss": 0.8649, "step": 7260 }, { "epoch": 0.75, "learning_rate": 6.590340476449473e-06, "loss": 0.8936, "step": 7270 }, { "epoch": 0.75, "learning_rate": 6.584901555531384e-06, "loss": 0.9575, "step": 7280 }, { "epoch": 0.75, "learning_rate": 6.5794626346132926e-06, "loss": 0.8958, "step": 7290 }, { "epoch": 0.75, "learning_rate": 6.574023713695203e-06, "loss": 0.9774, "step": 7300 }, { "epoch": 0.75, "learning_rate": 6.568584792777114e-06, "loss": 1.009, "step": 7310 }, { "epoch": 0.76, "learning_rate": 6.5631458718590235e-06, "loss": 0.8427, "step": 7320 }, { "epoch": 0.76, "learning_rate": 6.557706950940934e-06, "loss": 0.8912, "step": 7330 }, { "epoch": 0.76, "learning_rate": 6.552268030022844e-06, "loss": 0.9034, "step": 7340 }, { "epoch": 0.76, "learning_rate": 6.5468291091047544e-06, "loss": 0.9126, "step": 7350 }, { "epoch": 0.76, "learning_rate": 6.541390188186665e-06, "loss": 0.7849, "step": 7360 }, { "epoch": 0.76, "learning_rate": 6.535951267268574e-06, "loss": 0.8782, "step": 7370 }, { "epoch": 0.76, "learning_rate": 6.5305123463504845e-06, "loss": 1.0593, "step": 7380 }, { "epoch": 0.76, "learning_rate": 6.525073425432394e-06, "loss": 0.8501, "step": 7390 }, { "epoch": 0.76, "learning_rate": 6.519634504514305e-06, "loss": 0.8333, "step": 7400 }, { "epoch": 0.76, "learning_rate": 6.514195583596215e-06, "loss": 0.7898, "step": 7410 }, { "epoch": 0.77, "learning_rate": 6.508756662678125e-06, "loss": 0.8407, "step": 7420 }, { "epoch": 0.77, "learning_rate": 6.503317741760036e-06, "loss": 0.866, "step": 7430 }, { "epoch": 0.77, "learning_rate": 6.4978788208419456e-06, "loss": 0.876, "step": 7440 }, { "epoch": 0.77, "learning_rate": 6.492439899923856e-06, "loss": 0.9264, "step": 7450 }, { "epoch": 0.77, "learning_rate": 6.487000979005765e-06, "loss": 0.8972, "step": 7460 }, { "epoch": 0.77, "learning_rate": 6.481562058087676e-06, "loss": 0.8706, "step": 7470 }, { "epoch": 0.77, "learning_rate": 6.476123137169586e-06, "loss": 0.9409, "step": 7480 }, { "epoch": 0.77, "learning_rate": 6.470684216251496e-06, "loss": 0.937, "step": 7490 }, { "epoch": 0.77, "learning_rate": 6.465245295333407e-06, "loss": 0.8528, "step": 7500 }, { "epoch": 0.77, "learning_rate": 6.459806374415316e-06, "loss": 1.0074, "step": 7510 }, { "epoch": 0.78, "learning_rate": 6.454367453497227e-06, "loss": 0.8906, "step": 7520 }, { "epoch": 0.78, "learning_rate": 6.4489285325791375e-06, "loss": 0.8985, "step": 7530 }, { "epoch": 0.78, "learning_rate": 6.4434896116610464e-06, "loss": 0.8803, "step": 7540 }, { "epoch": 0.78, "learning_rate": 6.438050690742957e-06, "loss": 0.879, "step": 7550 }, { "epoch": 0.78, "learning_rate": 6.432611769824867e-06, "loss": 0.7659, "step": 7560 }, { "epoch": 0.78, "learning_rate": 6.427172848906777e-06, "loss": 0.9401, "step": 7570 }, { "epoch": 0.78, "learning_rate": 6.421733927988688e-06, "loss": 0.7697, "step": 7580 }, { "epoch": 0.78, "learning_rate": 6.416295007070598e-06, "loss": 0.901, "step": 7590 }, { "epoch": 0.78, "learning_rate": 6.410856086152508e-06, "loss": 0.9372, "step": 7600 }, { "epoch": 0.79, "learning_rate": 6.405417165234418e-06, "loss": 0.9334, "step": 7610 }, { "epoch": 0.79, "learning_rate": 6.399978244316329e-06, "loss": 0.8452, "step": 7620 }, { "epoch": 0.79, "learning_rate": 6.394539323398238e-06, "loss": 0.855, "step": 7630 }, { "epoch": 0.79, "learning_rate": 6.389100402480148e-06, "loss": 0.8771, "step": 7640 }, { "epoch": 0.79, "learning_rate": 6.383661481562059e-06, "loss": 0.8508, "step": 7650 }, { "epoch": 0.79, "learning_rate": 6.3782225606439685e-06, "loss": 0.9772, "step": 7660 }, { "epoch": 0.79, "learning_rate": 6.372783639725879e-06, "loss": 0.9182, "step": 7670 }, { "epoch": 0.79, "learning_rate": 6.367344718807789e-06, "loss": 0.8586, "step": 7680 }, { "epoch": 0.79, "learning_rate": 6.3619057978896995e-06, "loss": 0.9894, "step": 7690 }, { "epoch": 0.79, "learning_rate": 6.35646687697161e-06, "loss": 0.9508, "step": 7700 }, { "epoch": 0.8, "learning_rate": 6.351027956053519e-06, "loss": 0.9216, "step": 7710 }, { "epoch": 0.8, "learning_rate": 6.3455890351354296e-06, "loss": 0.9088, "step": 7720 }, { "epoch": 0.8, "learning_rate": 6.340150114217339e-06, "loss": 0.8797, "step": 7730 }, { "epoch": 0.8, "learning_rate": 6.33471119329925e-06, "loss": 0.9232, "step": 7740 }, { "epoch": 0.8, "learning_rate": 6.3292722723811605e-06, "loss": 0.8929, "step": 7750 }, { "epoch": 0.8, "learning_rate": 6.32383335146307e-06, "loss": 0.9063, "step": 7760 }, { "epoch": 0.8, "learning_rate": 6.318394430544981e-06, "loss": 0.8574, "step": 7770 }, { "epoch": 0.8, "learning_rate": 6.312955509626891e-06, "loss": 0.8947, "step": 7780 }, { "epoch": 0.8, "learning_rate": 6.307516588708801e-06, "loss": 0.8571, "step": 7790 }, { "epoch": 0.8, "learning_rate": 6.30207766779071e-06, "loss": 0.9294, "step": 7800 }, { "epoch": 0.81, "learning_rate": 6.296638746872621e-06, "loss": 0.9437, "step": 7810 }, { "epoch": 0.81, "learning_rate": 6.291199825954531e-06, "loss": 0.8518, "step": 7820 }, { "epoch": 0.81, "learning_rate": 6.285760905036441e-06, "loss": 0.8881, "step": 7830 }, { "epoch": 0.81, "learning_rate": 6.280321984118352e-06, "loss": 0.8676, "step": 7840 }, { "epoch": 0.81, "learning_rate": 6.274883063200261e-06, "loss": 0.983, "step": 7850 }, { "epoch": 0.81, "learning_rate": 6.269444142282172e-06, "loss": 0.8399, "step": 7860 }, { "epoch": 0.81, "learning_rate": 6.2640052213640826e-06, "loss": 0.8508, "step": 7870 }, { "epoch": 0.81, "learning_rate": 6.2585663004459915e-06, "loss": 0.9102, "step": 7880 }, { "epoch": 0.81, "learning_rate": 6.253127379527902e-06, "loss": 0.909, "step": 7890 }, { "epoch": 0.82, "learning_rate": 6.247688458609812e-06, "loss": 0.9332, "step": 7900 }, { "epoch": 0.82, "learning_rate": 6.242249537691722e-06, "loss": 0.9419, "step": 7910 }, { "epoch": 0.82, "learning_rate": 6.236810616773633e-06, "loss": 0.9858, "step": 7920 }, { "epoch": 0.82, "learning_rate": 6.231371695855543e-06, "loss": 0.9269, "step": 7930 }, { "epoch": 0.82, "learning_rate": 6.225932774937453e-06, "loss": 0.9736, "step": 7940 }, { "epoch": 0.82, "learning_rate": 6.220493854019362e-06, "loss": 0.8927, "step": 7950 }, { "epoch": 0.82, "learning_rate": 6.215054933101274e-06, "loss": 0.8486, "step": 7960 }, { "epoch": 0.82, "learning_rate": 6.209616012183184e-06, "loss": 0.879, "step": 7970 }, { "epoch": 0.82, "learning_rate": 6.204177091265093e-06, "loss": 0.8754, "step": 7980 }, { "epoch": 0.82, "learning_rate": 6.198738170347004e-06, "loss": 0.9566, "step": 7990 }, { "epoch": 0.83, "learning_rate": 6.1932992494289135e-06, "loss": 0.8437, "step": 8000 }, { "epoch": 0.83, "learning_rate": 6.187860328510824e-06, "loss": 0.8976, "step": 8010 }, { "epoch": 0.83, "learning_rate": 6.182421407592734e-06, "loss": 0.8005, "step": 8020 }, { "epoch": 0.83, "learning_rate": 6.1769824866746445e-06, "loss": 0.873, "step": 8030 }, { "epoch": 0.83, "learning_rate": 6.171543565756555e-06, "loss": 0.878, "step": 8040 }, { "epoch": 0.83, "learning_rate": 6.166104644838464e-06, "loss": 0.8515, "step": 8050 }, { "epoch": 0.83, "learning_rate": 6.160665723920375e-06, "loss": 0.8934, "step": 8060 }, { "epoch": 0.83, "learning_rate": 6.155226803002284e-06, "loss": 0.8985, "step": 8070 }, { "epoch": 0.83, "learning_rate": 6.149787882084195e-06, "loss": 0.8936, "step": 8080 }, { "epoch": 0.83, "learning_rate": 6.1443489611661055e-06, "loss": 0.9007, "step": 8090 }, { "epoch": 0.84, "learning_rate": 6.138910040248015e-06, "loss": 0.8972, "step": 8100 }, { "epoch": 0.84, "learning_rate": 6.133471119329926e-06, "loss": 0.8786, "step": 8110 }, { "epoch": 0.84, "learning_rate": 6.128032198411835e-06, "loss": 0.9224, "step": 8120 }, { "epoch": 0.84, "learning_rate": 6.122593277493745e-06, "loss": 0.805, "step": 8130 }, { "epoch": 0.84, "learning_rate": 6.117154356575656e-06, "loss": 0.8075, "step": 8140 }, { "epoch": 0.84, "learning_rate": 6.111715435657566e-06, "loss": 0.9063, "step": 8150 }, { "epoch": 0.84, "learning_rate": 6.106276514739476e-06, "loss": 0.8818, "step": 8160 }, { "epoch": 0.84, "learning_rate": 6.100837593821386e-06, "loss": 0.9576, "step": 8170 }, { "epoch": 0.84, "learning_rate": 6.095398672903297e-06, "loss": 0.8727, "step": 8180 }, { "epoch": 0.84, "learning_rate": 6.089959751985207e-06, "loss": 0.8449, "step": 8190 }, { "epoch": 0.85, "learning_rate": 6.084520831067117e-06, "loss": 0.913, "step": 8200 }, { "epoch": 0.85, "learning_rate": 6.079081910149028e-06, "loss": 0.8746, "step": 8210 }, { "epoch": 0.85, "learning_rate": 6.0736429892309365e-06, "loss": 0.8998, "step": 8220 }, { "epoch": 0.85, "learning_rate": 6.068204068312847e-06, "loss": 0.853, "step": 8230 }, { "epoch": 0.85, "learning_rate": 6.062765147394757e-06, "loss": 0.8889, "step": 8240 }, { "epoch": 0.85, "learning_rate": 6.0573262264766674e-06, "loss": 0.8896, "step": 8250 }, { "epoch": 0.85, "learning_rate": 6.051887305558578e-06, "loss": 0.8968, "step": 8260 }, { "epoch": 0.85, "learning_rate": 6.046448384640488e-06, "loss": 0.9676, "step": 8270 }, { "epoch": 0.85, "learning_rate": 6.041009463722398e-06, "loss": 0.8711, "step": 8280 }, { "epoch": 0.86, "learning_rate": 6.035570542804307e-06, "loss": 0.9059, "step": 8290 }, { "epoch": 0.86, "learning_rate": 6.030131621886218e-06, "loss": 0.9587, "step": 8300 }, { "epoch": 0.86, "learning_rate": 6.0246927009681285e-06, "loss": 0.8959, "step": 8310 }, { "epoch": 0.86, "learning_rate": 6.019253780050038e-06, "loss": 0.8412, "step": 8320 }, { "epoch": 0.86, "learning_rate": 6.013814859131949e-06, "loss": 0.9255, "step": 8330 }, { "epoch": 0.86, "learning_rate": 6.0083759382138586e-06, "loss": 0.8607, "step": 8340 }, { "epoch": 0.86, "learning_rate": 6.002937017295769e-06, "loss": 0.8295, "step": 8350 }, { "epoch": 0.86, "learning_rate": 5.99749809637768e-06, "loss": 0.8786, "step": 8360 }, { "epoch": 0.86, "learning_rate": 5.9920591754595895e-06, "loss": 0.9297, "step": 8370 }, { "epoch": 0.86, "learning_rate": 5.9866202545415e-06, "loss": 0.9778, "step": 8380 }, { "epoch": 0.87, "learning_rate": 5.981181333623409e-06, "loss": 0.8765, "step": 8390 }, { "epoch": 0.87, "learning_rate": 5.97574241270532e-06, "loss": 0.8672, "step": 8400 }, { "epoch": 0.87, "learning_rate": 5.97030349178723e-06, "loss": 0.8195, "step": 8410 }, { "epoch": 0.87, "learning_rate": 5.96486457086914e-06, "loss": 0.8696, "step": 8420 }, { "epoch": 0.87, "learning_rate": 5.9594256499510505e-06, "loss": 0.8869, "step": 8430 }, { "epoch": 0.87, "learning_rate": 5.95398672903296e-06, "loss": 0.976, "step": 8440 }, { "epoch": 0.87, "learning_rate": 5.948547808114871e-06, "loss": 0.895, "step": 8450 }, { "epoch": 0.87, "learning_rate": 5.94310888719678e-06, "loss": 0.7592, "step": 8460 }, { "epoch": 0.87, "learning_rate": 5.93766996627869e-06, "loss": 0.9033, "step": 8470 }, { "epoch": 0.87, "learning_rate": 5.932231045360601e-06, "loss": 0.8141, "step": 8480 }, { "epoch": 0.88, "learning_rate": 5.926792124442511e-06, "loss": 0.8407, "step": 8490 }, { "epoch": 0.88, "learning_rate": 5.921353203524421e-06, "loss": 0.8771, "step": 8500 }, { "epoch": 0.88, "learning_rate": 5.915914282606331e-06, "loss": 0.8943, "step": 8510 }, { "epoch": 0.88, "learning_rate": 5.910475361688242e-06, "loss": 0.8942, "step": 8520 }, { "epoch": 0.88, "learning_rate": 5.905036440770152e-06, "loss": 0.9232, "step": 8530 }, { "epoch": 0.88, "learning_rate": 5.899597519852062e-06, "loss": 0.9772, "step": 8540 }, { "epoch": 0.88, "learning_rate": 5.894158598933973e-06, "loss": 0.8998, "step": 8550 }, { "epoch": 0.88, "learning_rate": 5.8887196780158815e-06, "loss": 1.0183, "step": 8560 }, { "epoch": 0.88, "learning_rate": 5.883280757097792e-06, "loss": 0.7739, "step": 8570 }, { "epoch": 0.89, "learning_rate": 5.877841836179703e-06, "loss": 0.8312, "step": 8580 }, { "epoch": 0.89, "learning_rate": 5.8724029152616125e-06, "loss": 0.8573, "step": 8590 }, { "epoch": 0.89, "learning_rate": 5.866963994343523e-06, "loss": 0.9189, "step": 8600 }, { "epoch": 0.89, "learning_rate": 5.861525073425433e-06, "loss": 0.9484, "step": 8610 }, { "epoch": 0.89, "learning_rate": 5.856086152507343e-06, "loss": 0.8612, "step": 8620 }, { "epoch": 0.89, "learning_rate": 5.850647231589252e-06, "loss": 0.9482, "step": 8630 }, { "epoch": 0.89, "learning_rate": 5.845208310671163e-06, "loss": 0.9045, "step": 8640 }, { "epoch": 0.89, "learning_rate": 5.8397693897530735e-06, "loss": 0.8658, "step": 8650 }, { "epoch": 0.89, "learning_rate": 5.834330468834983e-06, "loss": 0.8749, "step": 8660 }, { "epoch": 0.89, "learning_rate": 5.828891547916894e-06, "loss": 0.8687, "step": 8670 }, { "epoch": 0.9, "learning_rate": 5.823452626998804e-06, "loss": 0.8892, "step": 8680 }, { "epoch": 0.9, "learning_rate": 5.818013706080714e-06, "loss": 1.0057, "step": 8690 }, { "epoch": 0.9, "learning_rate": 5.812574785162625e-06, "loss": 0.9176, "step": 8700 }, { "epoch": 0.9, "learning_rate": 5.8071358642445345e-06, "loss": 0.8726, "step": 8710 }, { "epoch": 0.9, "learning_rate": 5.801696943326445e-06, "loss": 0.9006, "step": 8720 }, { "epoch": 0.9, "learning_rate": 5.796258022408354e-06, "loss": 0.8329, "step": 8730 }, { "epoch": 0.9, "learning_rate": 5.790819101490265e-06, "loss": 0.8358, "step": 8740 }, { "epoch": 0.9, "learning_rate": 5.785380180572175e-06, "loss": 0.8403, "step": 8750 }, { "epoch": 0.9, "learning_rate": 5.779941259654085e-06, "loss": 0.8753, "step": 8760 }, { "epoch": 0.9, "learning_rate": 5.7745023387359956e-06, "loss": 0.9886, "step": 8770 }, { "epoch": 0.91, "learning_rate": 5.769063417817905e-06, "loss": 0.8052, "step": 8780 }, { "epoch": 0.91, "learning_rate": 5.763624496899816e-06, "loss": 0.8716, "step": 8790 }, { "epoch": 0.91, "learning_rate": 5.7581855759817265e-06, "loss": 0.9073, "step": 8800 }, { "epoch": 0.91, "learning_rate": 5.752746655063635e-06, "loss": 0.8511, "step": 8810 }, { "epoch": 0.91, "learning_rate": 5.747307734145546e-06, "loss": 0.9136, "step": 8820 }, { "epoch": 0.91, "learning_rate": 5.741868813227456e-06, "loss": 0.8524, "step": 8830 }, { "epoch": 0.91, "learning_rate": 5.736429892309366e-06, "loss": 0.8528, "step": 8840 }, { "epoch": 0.91, "learning_rate": 5.730990971391276e-06, "loss": 0.8041, "step": 8850 }, { "epoch": 0.91, "learning_rate": 5.725552050473187e-06, "loss": 0.8565, "step": 8860 }, { "epoch": 0.92, "learning_rate": 5.720113129555097e-06, "loss": 0.9044, "step": 8870 }, { "epoch": 0.92, "learning_rate": 5.714674208637007e-06, "loss": 0.9229, "step": 8880 }, { "epoch": 0.92, "learning_rate": 5.709235287718918e-06, "loss": 0.9337, "step": 8890 }, { "epoch": 0.92, "learning_rate": 5.7037963668008265e-06, "loss": 0.8455, "step": 8900 }, { "epoch": 0.92, "learning_rate": 5.698357445882737e-06, "loss": 0.8935, "step": 8910 }, { "epoch": 0.92, "learning_rate": 5.692918524964648e-06, "loss": 0.8312, "step": 8920 }, { "epoch": 0.92, "learning_rate": 5.6874796040465575e-06, "loss": 0.9648, "step": 8930 }, { "epoch": 0.92, "learning_rate": 5.682040683128468e-06, "loss": 0.9205, "step": 8940 }, { "epoch": 0.92, "learning_rate": 5.676601762210378e-06, "loss": 0.8634, "step": 8950 }, { "epoch": 0.92, "learning_rate": 5.671162841292288e-06, "loss": 0.892, "step": 8960 }, { "epoch": 0.93, "learning_rate": 5.665723920374199e-06, "loss": 0.857, "step": 8970 }, { "epoch": 0.93, "learning_rate": 5.660284999456108e-06, "loss": 1.0037, "step": 8980 }, { "epoch": 0.93, "learning_rate": 5.6548460785380185e-06, "loss": 0.9842, "step": 8990 }, { "epoch": 0.93, "learning_rate": 5.649407157619928e-06, "loss": 0.9369, "step": 9000 }, { "epoch": 0.93, "learning_rate": 5.643968236701839e-06, "loss": 0.8786, "step": 9010 }, { "epoch": 0.93, "learning_rate": 5.6385293157837495e-06, "loss": 0.8541, "step": 9020 }, { "epoch": 0.93, "learning_rate": 5.633090394865659e-06, "loss": 0.8276, "step": 9030 }, { "epoch": 0.93, "learning_rate": 5.62765147394757e-06, "loss": 0.9506, "step": 9040 }, { "epoch": 0.93, "learning_rate": 5.622212553029479e-06, "loss": 0.778, "step": 9050 }, { "epoch": 0.93, "learning_rate": 5.616773632111389e-06, "loss": 0.9741, "step": 9060 }, { "epoch": 0.94, "learning_rate": 5.611334711193299e-06, "loss": 0.9217, "step": 9070 }, { "epoch": 0.94, "learning_rate": 5.60589579027521e-06, "loss": 0.8475, "step": 9080 }, { "epoch": 0.94, "learning_rate": 5.60045686935712e-06, "loss": 0.8861, "step": 9090 }, { "epoch": 0.94, "learning_rate": 5.59501794843903e-06, "loss": 0.8665, "step": 9100 }, { "epoch": 0.94, "learning_rate": 5.589579027520941e-06, "loss": 0.8996, "step": 9110 }, { "epoch": 0.94, "learning_rate": 5.58414010660285e-06, "loss": 0.871, "step": 9120 }, { "epoch": 0.94, "learning_rate": 5.578701185684761e-06, "loss": 0.9497, "step": 9130 }, { "epoch": 0.94, "learning_rate": 5.5732622647666715e-06, "loss": 0.9039, "step": 9140 }, { "epoch": 0.94, "learning_rate": 5.5678233438485804e-06, "loss": 0.899, "step": 9150 }, { "epoch": 0.95, "learning_rate": 5.562384422930491e-06, "loss": 0.8931, "step": 9160 }, { "epoch": 0.95, "learning_rate": 5.556945502012401e-06, "loss": 0.7711, "step": 9170 }, { "epoch": 0.95, "learning_rate": 5.551506581094311e-06, "loss": 0.7955, "step": 9180 }, { "epoch": 0.95, "learning_rate": 5.546067660176222e-06, "loss": 0.8669, "step": 9190 }, { "epoch": 0.95, "learning_rate": 5.540628739258132e-06, "loss": 0.9031, "step": 9200 }, { "epoch": 0.95, "learning_rate": 5.535189818340042e-06, "loss": 0.8662, "step": 9210 }, { "epoch": 0.95, "learning_rate": 5.529750897421951e-06, "loss": 0.938, "step": 9220 }, { "epoch": 0.95, "learning_rate": 5.524311976503862e-06, "loss": 0.9369, "step": 9230 }, { "epoch": 0.95, "learning_rate": 5.5188730555857716e-06, "loss": 0.8771, "step": 9240 }, { "epoch": 0.95, "learning_rate": 5.513434134667682e-06, "loss": 0.8914, "step": 9250 }, { "epoch": 0.96, "learning_rate": 5.507995213749593e-06, "loss": 0.836, "step": 9260 }, { "epoch": 0.96, "learning_rate": 5.5025562928315025e-06, "loss": 0.9128, "step": 9270 }, { "epoch": 0.96, "learning_rate": 5.497117371913413e-06, "loss": 0.9087, "step": 9280 }, { "epoch": 0.96, "learning_rate": 5.491678450995323e-06, "loss": 0.9938, "step": 9290 }, { "epoch": 0.96, "learning_rate": 5.4862395300772334e-06, "loss": 0.9122, "step": 9300 }, { "epoch": 0.96, "learning_rate": 5.480800609159144e-06, "loss": 0.8879, "step": 9310 }, { "epoch": 0.96, "learning_rate": 5.475361688241053e-06, "loss": 0.904, "step": 9320 }, { "epoch": 0.96, "learning_rate": 5.4699227673229635e-06, "loss": 0.89, "step": 9330 }, { "epoch": 0.96, "learning_rate": 5.464483846404873e-06, "loss": 0.8769, "step": 9340 }, { "epoch": 0.96, "learning_rate": 5.459044925486784e-06, "loss": 0.9539, "step": 9350 }, { "epoch": 0.97, "learning_rate": 5.4536060045686945e-06, "loss": 0.9041, "step": 9360 }, { "epoch": 0.97, "learning_rate": 5.448167083650604e-06, "loss": 0.8246, "step": 9370 }, { "epoch": 0.97, "learning_rate": 5.442728162732515e-06, "loss": 0.7293, "step": 9380 }, { "epoch": 0.97, "learning_rate": 5.437289241814424e-06, "loss": 0.8496, "step": 9390 }, { "epoch": 0.97, "learning_rate": 5.431850320896334e-06, "loss": 0.9075, "step": 9400 }, { "epoch": 0.97, "learning_rate": 5.426411399978245e-06, "loss": 0.8667, "step": 9410 }, { "epoch": 0.97, "learning_rate": 5.420972479060155e-06, "loss": 0.9143, "step": 9420 }, { "epoch": 0.97, "learning_rate": 5.415533558142065e-06, "loss": 0.9056, "step": 9430 }, { "epoch": 0.97, "learning_rate": 5.410094637223975e-06, "loss": 0.9242, "step": 9440 }, { "epoch": 0.97, "learning_rate": 5.404655716305886e-06, "loss": 0.8402, "step": 9450 }, { "epoch": 0.98, "learning_rate": 5.399216795387795e-06, "loss": 0.9269, "step": 9460 }, { "epoch": 0.98, "learning_rate": 5.393777874469706e-06, "loss": 0.9192, "step": 9470 }, { "epoch": 0.98, "learning_rate": 5.3883389535516165e-06, "loss": 1.007, "step": 9480 }, { "epoch": 0.98, "learning_rate": 5.3829000326335254e-06, "loss": 0.8806, "step": 9490 }, { "epoch": 0.98, "learning_rate": 5.377461111715436e-06, "loss": 0.88, "step": 9500 }, { "epoch": 0.98, "learning_rate": 5.372022190797346e-06, "loss": 0.8204, "step": 9510 }, { "epoch": 0.98, "learning_rate": 5.366583269879256e-06, "loss": 0.9112, "step": 9520 }, { "epoch": 0.98, "learning_rate": 5.361144348961167e-06, "loss": 0.8132, "step": 9530 }, { "epoch": 0.98, "learning_rate": 5.355705428043077e-06, "loss": 0.8327, "step": 9540 }, { "epoch": 0.99, "learning_rate": 5.350266507124987e-06, "loss": 0.9499, "step": 9550 }, { "epoch": 0.99, "learning_rate": 5.344827586206896e-06, "loss": 0.7893, "step": 9560 }, { "epoch": 0.99, "learning_rate": 5.339388665288807e-06, "loss": 0.8466, "step": 9570 }, { "epoch": 0.99, "learning_rate": 5.333949744370717e-06, "loss": 0.9118, "step": 9580 }, { "epoch": 0.99, "learning_rate": 5.328510823452627e-06, "loss": 0.904, "step": 9590 }, { "epoch": 0.99, "learning_rate": 5.323071902534538e-06, "loss": 0.8745, "step": 9600 }, { "epoch": 0.99, "learning_rate": 5.3176329816164475e-06, "loss": 0.8835, "step": 9610 }, { "epoch": 0.99, "learning_rate": 5.312194060698358e-06, "loss": 0.9116, "step": 9620 }, { "epoch": 0.99, "learning_rate": 5.306755139780269e-06, "loss": 0.809, "step": 9630 }, { "epoch": 0.99, "learning_rate": 5.3013162188621785e-06, "loss": 0.8624, "step": 9640 }, { "epoch": 1.0, "learning_rate": 5.295877297944089e-06, "loss": 0.92, "step": 9650 }, { "epoch": 1.0, "learning_rate": 5.290438377025998e-06, "loss": 0.8814, "step": 9660 }, { "epoch": 1.0, "learning_rate": 5.2849994561079086e-06, "loss": 0.9462, "step": 9670 }, { "epoch": 1.0, "learning_rate": 5.279560535189818e-06, "loss": 1.0481, "step": 9680 }, { "epoch": 1.0, "learning_rate": 5.274121614271729e-06, "loss": 0.8671, "step": 9690 }, { "epoch": 1.0, "learning_rate": 5.2686826933536395e-06, "loss": 0.731, "step": 9700 }, { "epoch": 1.0, "learning_rate": 5.263243772435549e-06, "loss": 0.6246, "step": 9710 }, { "epoch": 1.0, "learning_rate": 5.25780485151746e-06, "loss": 0.6453, "step": 9720 }, { "epoch": 1.0, "learning_rate": 5.252365930599369e-06, "loss": 0.5982, "step": 9730 }, { "epoch": 1.0, "learning_rate": 5.246927009681279e-06, "loss": 0.6612, "step": 9740 }, { "epoch": 1.01, "learning_rate": 5.24148808876319e-06, "loss": 0.6167, "step": 9750 }, { "epoch": 1.01, "learning_rate": 5.2360491678451e-06, "loss": 0.6322, "step": 9760 }, { "epoch": 1.01, "learning_rate": 5.23061024692701e-06, "loss": 0.6477, "step": 9770 }, { "epoch": 1.01, "learning_rate": 5.22517132600892e-06, "loss": 0.6291, "step": 9780 }, { "epoch": 1.01, "learning_rate": 5.219732405090831e-06, "loss": 0.6146, "step": 9790 }, { "epoch": 1.01, "learning_rate": 5.214293484172741e-06, "loss": 0.5723, "step": 9800 }, { "epoch": 1.01, "learning_rate": 5.208854563254651e-06, "loss": 0.597, "step": 9810 }, { "epoch": 1.01, "learning_rate": 5.2034156423365616e-06, "loss": 0.656, "step": 9820 }, { "epoch": 1.01, "learning_rate": 5.1979767214184705e-06, "loss": 0.5783, "step": 9830 }, { "epoch": 1.02, "learning_rate": 5.192537800500381e-06, "loss": 0.6593, "step": 9840 }, { "epoch": 1.02, "learning_rate": 5.187098879582291e-06, "loss": 0.6825, "step": 9850 }, { "epoch": 1.02, "learning_rate": 5.181659958664201e-06, "loss": 0.6604, "step": 9860 }, { "epoch": 1.02, "learning_rate": 5.176221037746112e-06, "loss": 0.6274, "step": 9870 }, { "epoch": 1.02, "learning_rate": 5.170782116828022e-06, "loss": 0.651, "step": 9880 }, { "epoch": 1.02, "learning_rate": 5.165343195909932e-06, "loss": 0.6247, "step": 9890 }, { "epoch": 1.02, "learning_rate": 5.159904274991841e-06, "loss": 0.5922, "step": 9900 }, { "epoch": 1.02, "learning_rate": 5.154465354073752e-06, "loss": 0.5712, "step": 9910 }, { "epoch": 1.02, "learning_rate": 5.1490264331556624e-06, "loss": 0.6071, "step": 9920 }, { "epoch": 1.02, "learning_rate": 5.143587512237572e-06, "loss": 0.6497, "step": 9930 }, { "epoch": 1.03, "learning_rate": 5.138148591319483e-06, "loss": 0.609, "step": 9940 }, { "epoch": 1.03, "learning_rate": 5.1327096704013925e-06, "loss": 0.6134, "step": 9950 }, { "epoch": 1.03, "learning_rate": 5.127270749483303e-06, "loss": 0.6429, "step": 9960 }, { "epoch": 1.03, "learning_rate": 5.121831828565214e-06, "loss": 0.5364, "step": 9970 }, { "epoch": 1.03, "learning_rate": 5.116392907647123e-06, "loss": 0.6338, "step": 9980 }, { "epoch": 1.03, "learning_rate": 5.110953986729033e-06, "loss": 0.5744, "step": 9990 }, { "epoch": 1.03, "learning_rate": 5.105515065810943e-06, "loss": 0.625, "step": 10000 } ], "logging_steps": 10, "max_steps": 19386, "num_train_epochs": 2, "save_steps": 500, "total_flos": 1749207062806528.0, "trial_name": null, "trial_params": null }